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 directory is both 512 * casefolded and encrypted and its encryption key is unavailable, or if 513 * the filesystem is doing an internal operation where usr_fname is also 514 * NULL. In all these cases we fall back to treating the name as an 515 * 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 meta page */ 582 #define MAX_RETRY_META_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 GC_FAILURE_ATOMIC, 720 MAX_GC_FAILURE 721 }; 722 723 /* used for f2fs_inode_info->flags */ 724 enum { 725 FI_NEW_INODE, /* indicate newly allocated inode */ 726 FI_DIRTY_INODE, /* indicate inode is dirty or not */ 727 FI_AUTO_RECOVER, /* indicate inode is recoverable */ 728 FI_DIRTY_DIR, /* indicate directory has dirty pages */ 729 FI_INC_LINK, /* need to increment i_nlink */ 730 FI_ACL_MODE, /* indicate acl mode */ 731 FI_NO_ALLOC, /* should not allocate any blocks */ 732 FI_FREE_NID, /* free allocated nide */ 733 FI_NO_EXTENT, /* not to use the extent cache */ 734 FI_INLINE_XATTR, /* used for inline xattr */ 735 FI_INLINE_DATA, /* used for inline data*/ 736 FI_INLINE_DENTRY, /* used for inline dentry */ 737 FI_APPEND_WRITE, /* inode has appended data */ 738 FI_UPDATE_WRITE, /* inode has in-place-update data */ 739 FI_NEED_IPU, /* used for ipu per file */ 740 FI_ATOMIC_FILE, /* indicate atomic file */ 741 FI_ATOMIC_COMMIT, /* indicate the state of atomical committing */ 742 FI_VOLATILE_FILE, /* indicate volatile file */ 743 FI_FIRST_BLOCK_WRITTEN, /* indicate #0 data block was written */ 744 FI_DROP_CACHE, /* drop dirty page cache */ 745 FI_DATA_EXIST, /* indicate data exists */ 746 FI_INLINE_DOTS, /* indicate inline dot dentries */ 747 FI_SKIP_WRITES, /* should skip data page writeback */ 748 FI_OPU_WRITE, /* used for opu per file */ 749 FI_DIRTY_FILE, /* indicate regular/symlink has dirty pages */ 750 FI_PREALLOCATED_ALL, /* all blocks for write were preallocated */ 751 FI_HOT_DATA, /* indicate file is hot */ 752 FI_EXTRA_ATTR, /* indicate file has extra attribute */ 753 FI_PROJ_INHERIT, /* indicate file inherits projectid */ 754 FI_PIN_FILE, /* indicate file should not be gced */ 755 FI_ATOMIC_REVOKE_REQUEST, /* request to drop atomic data */ 756 FI_VERITY_IN_PROGRESS, /* building fs-verity Merkle tree */ 757 FI_COMPRESSED_FILE, /* indicate file's data can be compressed */ 758 FI_COMPRESS_CORRUPT, /* indicate compressed cluster is corrupted */ 759 FI_MMAP_FILE, /* indicate file was mmapped */ 760 FI_ENABLE_COMPRESS, /* enable compression in "user" compression mode */ 761 FI_COMPRESS_RELEASED, /* compressed blocks were released */ 762 FI_ALIGNED_WRITE, /* enable aligned write */ 763 FI_MAX, /* max flag, never be used */ 764 }; 765 766 struct f2fs_inode_info { 767 struct inode vfs_inode; /* serve a vfs inode */ 768 unsigned long i_flags; /* keep an inode flags for ioctl */ 769 unsigned char i_advise; /* use to give file attribute hints */ 770 unsigned char i_dir_level; /* use for dentry level for large dir */ 771 unsigned int i_current_depth; /* only for directory depth */ 772 /* for gc failure statistic */ 773 unsigned int i_gc_failures[MAX_GC_FAILURE]; 774 unsigned int i_pino; /* parent inode number */ 775 umode_t i_acl_mode; /* keep file acl mode temporarily */ 776 777 /* Use below internally in f2fs*/ 778 unsigned long flags[BITS_TO_LONGS(FI_MAX)]; /* use to pass per-file flags */ 779 struct f2fs_rwsem i_sem; /* protect fi info */ 780 atomic_t dirty_pages; /* # of dirty pages */ 781 f2fs_hash_t chash; /* hash value of given file name */ 782 unsigned int clevel; /* maximum level of given file name */ 783 struct task_struct *task; /* lookup and create consistency */ 784 struct task_struct *cp_task; /* separate cp/wb IO stats*/ 785 nid_t i_xattr_nid; /* node id that contains xattrs */ 786 loff_t last_disk_size; /* lastly written file size */ 787 spinlock_t i_size_lock; /* protect last_disk_size */ 788 789 #ifdef CONFIG_QUOTA 790 struct dquot *i_dquot[MAXQUOTAS]; 791 792 /* quota space reservation, managed internally by quota code */ 793 qsize_t i_reserved_quota; 794 #endif 795 struct list_head dirty_list; /* dirty list for dirs and files */ 796 struct list_head gdirty_list; /* linked in global dirty list */ 797 struct list_head inmem_ilist; /* list for inmem inodes */ 798 struct list_head inmem_pages; /* inmemory pages managed by f2fs */ 799 struct task_struct *inmem_task; /* store inmemory task */ 800 struct mutex inmem_lock; /* lock for inmemory pages */ 801 struct extent_tree *extent_tree; /* cached extent_tree entry */ 802 803 /* avoid racing between foreground op and gc */ 804 struct f2fs_rwsem i_gc_rwsem[2]; 805 struct f2fs_rwsem i_xattr_sem; /* avoid racing between reading and changing EAs */ 806 807 int i_extra_isize; /* size of extra space located in i_addr */ 808 kprojid_t i_projid; /* id for project quota */ 809 int i_inline_xattr_size; /* inline xattr size */ 810 struct timespec64 i_crtime; /* inode creation time */ 811 struct timespec64 i_disk_time[4];/* inode disk times */ 812 813 /* for file compress */ 814 atomic_t i_compr_blocks; /* # of compressed blocks */ 815 unsigned char i_compress_algorithm; /* algorithm type */ 816 unsigned char i_log_cluster_size; /* log of cluster size */ 817 unsigned char i_compress_level; /* compress level (lz4hc,zstd) */ 818 unsigned short i_compress_flag; /* compress flag */ 819 unsigned int i_cluster_size; /* cluster size */ 820 }; 821 822 static inline void get_extent_info(struct extent_info *ext, 823 struct f2fs_extent *i_ext) 824 { 825 ext->fofs = le32_to_cpu(i_ext->fofs); 826 ext->blk = le32_to_cpu(i_ext->blk); 827 ext->len = le32_to_cpu(i_ext->len); 828 } 829 830 static inline void set_raw_extent(struct extent_info *ext, 831 struct f2fs_extent *i_ext) 832 { 833 i_ext->fofs = cpu_to_le32(ext->fofs); 834 i_ext->blk = cpu_to_le32(ext->blk); 835 i_ext->len = cpu_to_le32(ext->len); 836 } 837 838 static inline void set_extent_info(struct extent_info *ei, unsigned int fofs, 839 u32 blk, unsigned int len) 840 { 841 ei->fofs = fofs; 842 ei->blk = blk; 843 ei->len = len; 844 #ifdef CONFIG_F2FS_FS_COMPRESSION 845 ei->c_len = 0; 846 #endif 847 } 848 849 static inline bool __is_discard_mergeable(struct discard_info *back, 850 struct discard_info *front, unsigned int max_len) 851 { 852 return (back->lstart + back->len == front->lstart) && 853 (back->len + front->len <= max_len); 854 } 855 856 static inline bool __is_discard_back_mergeable(struct discard_info *cur, 857 struct discard_info *back, unsigned int max_len) 858 { 859 return __is_discard_mergeable(back, cur, max_len); 860 } 861 862 static inline bool __is_discard_front_mergeable(struct discard_info *cur, 863 struct discard_info *front, unsigned int max_len) 864 { 865 return __is_discard_mergeable(cur, front, max_len); 866 } 867 868 static inline bool __is_extent_mergeable(struct extent_info *back, 869 struct extent_info *front) 870 { 871 #ifdef CONFIG_F2FS_FS_COMPRESSION 872 if (back->c_len && back->len != back->c_len) 873 return false; 874 if (front->c_len && front->len != front->c_len) 875 return false; 876 #endif 877 return (back->fofs + back->len == front->fofs && 878 back->blk + back->len == front->blk); 879 } 880 881 static inline bool __is_back_mergeable(struct extent_info *cur, 882 struct extent_info *back) 883 { 884 return __is_extent_mergeable(back, cur); 885 } 886 887 static inline bool __is_front_mergeable(struct extent_info *cur, 888 struct extent_info *front) 889 { 890 return __is_extent_mergeable(cur, front); 891 } 892 893 extern void f2fs_mark_inode_dirty_sync(struct inode *inode, bool sync); 894 static inline void __try_update_largest_extent(struct extent_tree *et, 895 struct extent_node *en) 896 { 897 if (en->ei.len > et->largest.len) { 898 et->largest = en->ei; 899 et->largest_updated = true; 900 } 901 } 902 903 /* 904 * For free nid management 905 */ 906 enum nid_state { 907 FREE_NID, /* newly added to free nid list */ 908 PREALLOC_NID, /* it is preallocated */ 909 MAX_NID_STATE, 910 }; 911 912 enum nat_state { 913 TOTAL_NAT, 914 DIRTY_NAT, 915 RECLAIMABLE_NAT, 916 MAX_NAT_STATE, 917 }; 918 919 struct f2fs_nm_info { 920 block_t nat_blkaddr; /* base disk address of NAT */ 921 nid_t max_nid; /* maximum possible node ids */ 922 nid_t available_nids; /* # of available node ids */ 923 nid_t next_scan_nid; /* the next nid to be scanned */ 924 nid_t max_rf_node_blocks; /* max # of nodes for recovery */ 925 unsigned int ram_thresh; /* control the memory footprint */ 926 unsigned int ra_nid_pages; /* # of nid pages to be readaheaded */ 927 unsigned int dirty_nats_ratio; /* control dirty nats ratio threshold */ 928 929 /* NAT cache management */ 930 struct radix_tree_root nat_root;/* root of the nat entry cache */ 931 struct radix_tree_root nat_set_root;/* root of the nat set cache */ 932 struct f2fs_rwsem nat_tree_lock; /* protect nat entry tree */ 933 struct list_head nat_entries; /* cached nat entry list (clean) */ 934 spinlock_t nat_list_lock; /* protect clean nat entry list */ 935 unsigned int nat_cnt[MAX_NAT_STATE]; /* the # of cached nat entries */ 936 unsigned int nat_blocks; /* # of nat blocks */ 937 938 /* free node ids management */ 939 struct radix_tree_root free_nid_root;/* root of the free_nid cache */ 940 struct list_head free_nid_list; /* list for free nids excluding preallocated nids */ 941 unsigned int nid_cnt[MAX_NID_STATE]; /* the number of free node id */ 942 spinlock_t nid_list_lock; /* protect nid lists ops */ 943 struct mutex build_lock; /* lock for build free nids */ 944 unsigned char **free_nid_bitmap; 945 unsigned char *nat_block_bitmap; 946 unsigned short *free_nid_count; /* free nid count of NAT block */ 947 948 /* for checkpoint */ 949 char *nat_bitmap; /* NAT bitmap pointer */ 950 951 unsigned int nat_bits_blocks; /* # of nat bits blocks */ 952 unsigned char *nat_bits; /* NAT bits blocks */ 953 unsigned char *full_nat_bits; /* full NAT pages */ 954 unsigned char *empty_nat_bits; /* empty NAT pages */ 955 #ifdef CONFIG_F2FS_CHECK_FS 956 char *nat_bitmap_mir; /* NAT bitmap mirror */ 957 #endif 958 int bitmap_size; /* bitmap size */ 959 }; 960 961 /* 962 * this structure is used as one of function parameters. 963 * all the information are dedicated to a given direct node block determined 964 * by the data offset in a file. 965 */ 966 struct dnode_of_data { 967 struct inode *inode; /* vfs inode pointer */ 968 struct page *inode_page; /* its inode page, NULL is possible */ 969 struct page *node_page; /* cached direct node page */ 970 nid_t nid; /* node id of the direct node block */ 971 unsigned int ofs_in_node; /* data offset in the node page */ 972 bool inode_page_locked; /* inode page is locked or not */ 973 bool node_changed; /* is node block changed */ 974 char cur_level; /* level of hole node page */ 975 char max_level; /* level of current page located */ 976 block_t data_blkaddr; /* block address of the node block */ 977 }; 978 979 static inline void set_new_dnode(struct dnode_of_data *dn, struct inode *inode, 980 struct page *ipage, struct page *npage, nid_t nid) 981 { 982 memset(dn, 0, sizeof(*dn)); 983 dn->inode = inode; 984 dn->inode_page = ipage; 985 dn->node_page = npage; 986 dn->nid = nid; 987 } 988 989 /* 990 * For SIT manager 991 * 992 * By default, there are 6 active log areas across the whole main area. 993 * When considering hot and cold data separation to reduce cleaning overhead, 994 * we split 3 for data logs and 3 for node logs as hot, warm, and cold types, 995 * respectively. 996 * In the current design, you should not change the numbers intentionally. 997 * Instead, as a mount option such as active_logs=x, you can use 2, 4, and 6 998 * logs individually according to the underlying devices. (default: 6) 999 * Just in case, on-disk layout covers maximum 16 logs that consist of 8 for 1000 * data and 8 for node logs. 1001 */ 1002 #define NR_CURSEG_DATA_TYPE (3) 1003 #define NR_CURSEG_NODE_TYPE (3) 1004 #define NR_CURSEG_INMEM_TYPE (2) 1005 #define NR_CURSEG_RO_TYPE (2) 1006 #define NR_CURSEG_PERSIST_TYPE (NR_CURSEG_DATA_TYPE + NR_CURSEG_NODE_TYPE) 1007 #define NR_CURSEG_TYPE (NR_CURSEG_INMEM_TYPE + NR_CURSEG_PERSIST_TYPE) 1008 1009 enum { 1010 CURSEG_HOT_DATA = 0, /* directory entry blocks */ 1011 CURSEG_WARM_DATA, /* data blocks */ 1012 CURSEG_COLD_DATA, /* multimedia or GCed data blocks */ 1013 CURSEG_HOT_NODE, /* direct node blocks of directory files */ 1014 CURSEG_WARM_NODE, /* direct node blocks of normal files */ 1015 CURSEG_COLD_NODE, /* indirect node blocks */ 1016 NR_PERSISTENT_LOG, /* number of persistent log */ 1017 CURSEG_COLD_DATA_PINNED = NR_PERSISTENT_LOG, 1018 /* pinned file that needs consecutive block address */ 1019 CURSEG_ALL_DATA_ATGC, /* SSR alloctor in hot/warm/cold data area */ 1020 NO_CHECK_TYPE, /* number of persistent & inmem log */ 1021 }; 1022 1023 struct flush_cmd { 1024 struct completion wait; 1025 struct llist_node llnode; 1026 nid_t ino; 1027 int ret; 1028 }; 1029 1030 struct flush_cmd_control { 1031 struct task_struct *f2fs_issue_flush; /* flush thread */ 1032 wait_queue_head_t flush_wait_queue; /* waiting queue for wake-up */ 1033 atomic_t issued_flush; /* # of issued flushes */ 1034 atomic_t queued_flush; /* # of queued flushes */ 1035 struct llist_head issue_list; /* list for command issue */ 1036 struct llist_node *dispatch_list; /* list for command dispatch */ 1037 }; 1038 1039 struct f2fs_sm_info { 1040 struct sit_info *sit_info; /* whole segment information */ 1041 struct free_segmap_info *free_info; /* free segment information */ 1042 struct dirty_seglist_info *dirty_info; /* dirty segment information */ 1043 struct curseg_info *curseg_array; /* active segment information */ 1044 1045 struct f2fs_rwsem curseg_lock; /* for preventing curseg change */ 1046 1047 block_t seg0_blkaddr; /* block address of 0'th segment */ 1048 block_t main_blkaddr; /* start block address of main area */ 1049 block_t ssa_blkaddr; /* start block address of SSA area */ 1050 1051 unsigned int segment_count; /* total # of segments */ 1052 unsigned int main_segments; /* # of segments in main area */ 1053 unsigned int reserved_segments; /* # of reserved segments */ 1054 unsigned int additional_reserved_segments;/* reserved segs for IO align feature */ 1055 unsigned int ovp_segments; /* # of overprovision segments */ 1056 1057 /* a threshold to reclaim prefree segments */ 1058 unsigned int rec_prefree_segments; 1059 1060 /* for batched trimming */ 1061 unsigned int trim_sections; /* # of sections to trim */ 1062 1063 struct list_head sit_entry_set; /* sit entry set list */ 1064 1065 unsigned int ipu_policy; /* in-place-update policy */ 1066 unsigned int min_ipu_util; /* in-place-update threshold */ 1067 unsigned int min_fsync_blocks; /* threshold for fsync */ 1068 unsigned int min_seq_blocks; /* threshold for sequential blocks */ 1069 unsigned int min_hot_blocks; /* threshold for hot block allocation */ 1070 unsigned int min_ssr_sections; /* threshold to trigger SSR allocation */ 1071 1072 /* for flush command control */ 1073 struct flush_cmd_control *fcc_info; 1074 1075 /* for discard command control */ 1076 struct discard_cmd_control *dcc_info; 1077 }; 1078 1079 /* 1080 * For superblock 1081 */ 1082 /* 1083 * COUNT_TYPE for monitoring 1084 * 1085 * f2fs monitors the number of several block types such as on-writeback, 1086 * dirty dentry blocks, dirty node blocks, and dirty meta blocks. 1087 */ 1088 #define WB_DATA_TYPE(p) (__is_cp_guaranteed(p) ? F2FS_WB_CP_DATA : F2FS_WB_DATA) 1089 enum count_type { 1090 F2FS_DIRTY_DENTS, 1091 F2FS_DIRTY_DATA, 1092 F2FS_DIRTY_QDATA, 1093 F2FS_DIRTY_NODES, 1094 F2FS_DIRTY_META, 1095 F2FS_INMEM_PAGES, 1096 F2FS_DIRTY_IMETA, 1097 F2FS_WB_CP_DATA, 1098 F2FS_WB_DATA, 1099 F2FS_RD_DATA, 1100 F2FS_RD_NODE, 1101 F2FS_RD_META, 1102 F2FS_DIO_WRITE, 1103 F2FS_DIO_READ, 1104 NR_COUNT_TYPE, 1105 }; 1106 1107 /* 1108 * The below are the page types of bios used in submit_bio(). 1109 * The available types are: 1110 * DATA User data pages. It operates as async mode. 1111 * NODE Node pages. It operates as async mode. 1112 * META FS metadata pages such as SIT, NAT, CP. 1113 * NR_PAGE_TYPE The number of page types. 1114 * META_FLUSH Make sure the previous pages are written 1115 * with waiting the bio's completion 1116 * ... Only can be used with META. 1117 */ 1118 #define PAGE_TYPE_OF_BIO(type) ((type) > META ? META : (type)) 1119 enum page_type { 1120 DATA, 1121 NODE, 1122 META, 1123 NR_PAGE_TYPE, 1124 META_FLUSH, 1125 INMEM, /* the below types are used by tracepoints only. */ 1126 INMEM_DROP, 1127 INMEM_INVALIDATE, 1128 INMEM_REVOKE, 1129 IPU, 1130 OPU, 1131 }; 1132 1133 enum temp_type { 1134 HOT = 0, /* must be zero for meta bio */ 1135 WARM, 1136 COLD, 1137 NR_TEMP_TYPE, 1138 }; 1139 1140 enum need_lock_type { 1141 LOCK_REQ = 0, 1142 LOCK_DONE, 1143 LOCK_RETRY, 1144 }; 1145 1146 enum cp_reason_type { 1147 CP_NO_NEEDED, 1148 CP_NON_REGULAR, 1149 CP_COMPRESSED, 1150 CP_HARDLINK, 1151 CP_SB_NEED_CP, 1152 CP_WRONG_PINO, 1153 CP_NO_SPC_ROLL, 1154 CP_NODE_NEED_CP, 1155 CP_FASTBOOT_MODE, 1156 CP_SPEC_LOG_NUM, 1157 CP_RECOVER_DIR, 1158 }; 1159 1160 enum iostat_type { 1161 /* WRITE IO */ 1162 APP_DIRECT_IO, /* app direct write IOs */ 1163 APP_BUFFERED_IO, /* app buffered write IOs */ 1164 APP_WRITE_IO, /* app write IOs */ 1165 APP_MAPPED_IO, /* app mapped IOs */ 1166 FS_DATA_IO, /* data IOs from kworker/fsync/reclaimer */ 1167 FS_NODE_IO, /* node IOs from kworker/fsync/reclaimer */ 1168 FS_META_IO, /* meta IOs from kworker/reclaimer */ 1169 FS_GC_DATA_IO, /* data IOs from forground gc */ 1170 FS_GC_NODE_IO, /* node IOs from forground gc */ 1171 FS_CP_DATA_IO, /* data IOs from checkpoint */ 1172 FS_CP_NODE_IO, /* node IOs from checkpoint */ 1173 FS_CP_META_IO, /* meta IOs from checkpoint */ 1174 1175 /* READ IO */ 1176 APP_DIRECT_READ_IO, /* app direct read IOs */ 1177 APP_BUFFERED_READ_IO, /* app buffered read IOs */ 1178 APP_READ_IO, /* app read IOs */ 1179 APP_MAPPED_READ_IO, /* app mapped read IOs */ 1180 FS_DATA_READ_IO, /* data read IOs */ 1181 FS_GDATA_READ_IO, /* data read IOs from background gc */ 1182 FS_CDATA_READ_IO, /* compressed data read IOs */ 1183 FS_NODE_READ_IO, /* node read IOs */ 1184 FS_META_READ_IO, /* meta read IOs */ 1185 1186 /* other */ 1187 FS_DISCARD, /* discard */ 1188 NR_IO_TYPE, 1189 }; 1190 1191 struct f2fs_io_info { 1192 struct f2fs_sb_info *sbi; /* f2fs_sb_info pointer */ 1193 nid_t ino; /* inode number */ 1194 enum page_type type; /* contains DATA/NODE/META/META_FLUSH */ 1195 enum temp_type temp; /* contains HOT/WARM/COLD */ 1196 int op; /* contains REQ_OP_ */ 1197 int op_flags; /* req_flag_bits */ 1198 block_t new_blkaddr; /* new block address to be written */ 1199 block_t old_blkaddr; /* old block address before Cow */ 1200 struct page *page; /* page to be written */ 1201 struct page *encrypted_page; /* encrypted page */ 1202 struct page *compressed_page; /* compressed page */ 1203 struct list_head list; /* serialize IOs */ 1204 bool submitted; /* indicate IO submission */ 1205 int need_lock; /* indicate we need to lock cp_rwsem */ 1206 bool in_list; /* indicate fio is in io_list */ 1207 bool is_por; /* indicate IO is from recovery or not */ 1208 bool retry; /* need to reallocate block address */ 1209 int compr_blocks; /* # of compressed block addresses */ 1210 bool encrypted; /* indicate file is encrypted */ 1211 enum iostat_type io_type; /* io type */ 1212 struct writeback_control *io_wbc; /* writeback control */ 1213 struct bio **bio; /* bio for ipu */ 1214 sector_t *last_block; /* last block number in bio */ 1215 unsigned char version; /* version of the node */ 1216 }; 1217 1218 struct bio_entry { 1219 struct bio *bio; 1220 struct list_head list; 1221 }; 1222 1223 #define is_read_io(rw) ((rw) == READ) 1224 struct f2fs_bio_info { 1225 struct f2fs_sb_info *sbi; /* f2fs superblock */ 1226 struct bio *bio; /* bios to merge */ 1227 sector_t last_block_in_bio; /* last block number */ 1228 struct f2fs_io_info fio; /* store buffered io info. */ 1229 struct f2fs_rwsem io_rwsem; /* blocking op for bio */ 1230 spinlock_t io_lock; /* serialize DATA/NODE IOs */ 1231 struct list_head io_list; /* track fios */ 1232 struct list_head bio_list; /* bio entry list head */ 1233 struct f2fs_rwsem bio_list_lock; /* lock to protect bio entry list */ 1234 }; 1235 1236 #define FDEV(i) (sbi->devs[i]) 1237 #define RDEV(i) (raw_super->devs[i]) 1238 struct f2fs_dev_info { 1239 struct block_device *bdev; 1240 char path[MAX_PATH_LEN]; 1241 unsigned int total_segments; 1242 block_t start_blk; 1243 block_t end_blk; 1244 #ifdef CONFIG_BLK_DEV_ZONED 1245 unsigned int nr_blkz; /* Total number of zones */ 1246 unsigned long *blkz_seq; /* Bitmap indicating sequential zones */ 1247 block_t *zone_capacity_blocks; /* Array of zone capacity in blks */ 1248 #endif 1249 }; 1250 1251 enum inode_type { 1252 DIR_INODE, /* for dirty dir inode */ 1253 FILE_INODE, /* for dirty regular/symlink inode */ 1254 DIRTY_META, /* for all dirtied inode metadata */ 1255 ATOMIC_FILE, /* for all atomic files */ 1256 NR_INODE_TYPE, 1257 }; 1258 1259 /* for inner inode cache management */ 1260 struct inode_management { 1261 struct radix_tree_root ino_root; /* ino entry array */ 1262 spinlock_t ino_lock; /* for ino entry lock */ 1263 struct list_head ino_list; /* inode list head */ 1264 unsigned long ino_num; /* number of entries */ 1265 }; 1266 1267 /* for GC_AT */ 1268 struct atgc_management { 1269 bool atgc_enabled; /* ATGC is enabled or not */ 1270 struct rb_root_cached root; /* root of victim rb-tree */ 1271 struct list_head victim_list; /* linked with all victim entries */ 1272 unsigned int victim_count; /* victim count in rb-tree */ 1273 unsigned int candidate_ratio; /* candidate ratio */ 1274 unsigned int max_candidate_count; /* max candidate count */ 1275 unsigned int age_weight; /* age weight, vblock_weight = 100 - age_weight */ 1276 unsigned long long age_threshold; /* age threshold */ 1277 }; 1278 1279 /* For s_flag in struct f2fs_sb_info */ 1280 enum { 1281 SBI_IS_DIRTY, /* dirty flag for checkpoint */ 1282 SBI_IS_CLOSE, /* specify unmounting */ 1283 SBI_NEED_FSCK, /* need fsck.f2fs to fix */ 1284 SBI_POR_DOING, /* recovery is doing or not */ 1285 SBI_NEED_SB_WRITE, /* need to recover superblock */ 1286 SBI_NEED_CP, /* need to checkpoint */ 1287 SBI_IS_SHUTDOWN, /* shutdown by ioctl */ 1288 SBI_IS_RECOVERED, /* recovered orphan/data */ 1289 SBI_CP_DISABLED, /* CP was disabled last mount */ 1290 SBI_CP_DISABLED_QUICK, /* CP was disabled quickly */ 1291 SBI_QUOTA_NEED_FLUSH, /* need to flush quota info in CP */ 1292 SBI_QUOTA_SKIP_FLUSH, /* skip flushing quota in current CP */ 1293 SBI_QUOTA_NEED_REPAIR, /* quota file may be corrupted */ 1294 SBI_IS_RESIZEFS, /* resizefs is in process */ 1295 SBI_IS_FREEZING, /* freezefs is in process */ 1296 }; 1297 1298 enum { 1299 CP_TIME, 1300 REQ_TIME, 1301 DISCARD_TIME, 1302 GC_TIME, 1303 DISABLE_TIME, 1304 UMOUNT_DISCARD_TIMEOUT, 1305 MAX_TIME, 1306 }; 1307 1308 enum { 1309 GC_NORMAL, 1310 GC_IDLE_CB, 1311 GC_IDLE_GREEDY, 1312 GC_IDLE_AT, 1313 GC_URGENT_HIGH, 1314 GC_URGENT_LOW, 1315 GC_URGENT_MID, 1316 MAX_GC_MODE, 1317 }; 1318 1319 enum { 1320 BGGC_MODE_ON, /* background gc is on */ 1321 BGGC_MODE_OFF, /* background gc is off */ 1322 BGGC_MODE_SYNC, /* 1323 * background gc is on, migrating blocks 1324 * like foreground gc 1325 */ 1326 }; 1327 1328 enum { 1329 FS_MODE_ADAPTIVE, /* use both lfs/ssr allocation */ 1330 FS_MODE_LFS, /* use lfs allocation only */ 1331 FS_MODE_FRAGMENT_SEG, /* segment fragmentation mode */ 1332 FS_MODE_FRAGMENT_BLK, /* block fragmentation mode */ 1333 }; 1334 1335 enum { 1336 ALLOC_MODE_DEFAULT, /* stay default */ 1337 ALLOC_MODE_REUSE, /* reuse segments as much as possible */ 1338 }; 1339 1340 enum fsync_mode { 1341 FSYNC_MODE_POSIX, /* fsync follows posix semantics */ 1342 FSYNC_MODE_STRICT, /* fsync behaves in line with ext4 */ 1343 FSYNC_MODE_NOBARRIER, /* fsync behaves nobarrier based on posix */ 1344 }; 1345 1346 enum { 1347 COMPR_MODE_FS, /* 1348 * automatically compress compression 1349 * enabled files 1350 */ 1351 COMPR_MODE_USER, /* 1352 * automatical compression is disabled. 1353 * user can control the file compression 1354 * using ioctls 1355 */ 1356 }; 1357 1358 enum { 1359 DISCARD_UNIT_BLOCK, /* basic discard unit is block */ 1360 DISCARD_UNIT_SEGMENT, /* basic discard unit is segment */ 1361 DISCARD_UNIT_SECTION, /* basic discard unit is section */ 1362 }; 1363 1364 static inline int f2fs_test_bit(unsigned int nr, char *addr); 1365 static inline void f2fs_set_bit(unsigned int nr, char *addr); 1366 static inline void f2fs_clear_bit(unsigned int nr, char *addr); 1367 1368 /* 1369 * Layout of f2fs page.private: 1370 * 1371 * Layout A: lowest bit should be 1 1372 * | bit0 = 1 | bit1 | bit2 | ... | bit MAX | private data .... | 1373 * bit 0 PAGE_PRIVATE_NOT_POINTER 1374 * bit 1 PAGE_PRIVATE_ATOMIC_WRITE 1375 * bit 2 PAGE_PRIVATE_DUMMY_WRITE 1376 * bit 3 PAGE_PRIVATE_ONGOING_MIGRATION 1377 * bit 4 PAGE_PRIVATE_INLINE_INODE 1378 * bit 5 PAGE_PRIVATE_REF_RESOURCE 1379 * bit 6- f2fs private data 1380 * 1381 * Layout B: lowest bit should be 0 1382 * page.private is a wrapped pointer. 1383 */ 1384 enum { 1385 PAGE_PRIVATE_NOT_POINTER, /* private contains non-pointer data */ 1386 PAGE_PRIVATE_ATOMIC_WRITE, /* data page from atomic write path */ 1387 PAGE_PRIVATE_DUMMY_WRITE, /* data page for padding aligned IO */ 1388 PAGE_PRIVATE_ONGOING_MIGRATION, /* data page which is on-going migrating */ 1389 PAGE_PRIVATE_INLINE_INODE, /* inode page contains inline data */ 1390 PAGE_PRIVATE_REF_RESOURCE, /* dirty page has referenced resources */ 1391 PAGE_PRIVATE_MAX 1392 }; 1393 1394 #define PAGE_PRIVATE_GET_FUNC(name, flagname) \ 1395 static inline bool page_private_##name(struct page *page) \ 1396 { \ 1397 return PagePrivate(page) && \ 1398 test_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)) && \ 1399 test_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \ 1400 } 1401 1402 #define PAGE_PRIVATE_SET_FUNC(name, flagname) \ 1403 static inline void set_page_private_##name(struct page *page) \ 1404 { \ 1405 if (!PagePrivate(page)) { \ 1406 get_page(page); \ 1407 SetPagePrivate(page); \ 1408 set_page_private(page, 0); \ 1409 } \ 1410 set_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)); \ 1411 set_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \ 1412 } 1413 1414 #define PAGE_PRIVATE_CLEAR_FUNC(name, flagname) \ 1415 static inline void clear_page_private_##name(struct page *page) \ 1416 { \ 1417 clear_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \ 1418 if (page_private(page) == 1 << PAGE_PRIVATE_NOT_POINTER) { \ 1419 set_page_private(page, 0); \ 1420 if (PagePrivate(page)) { \ 1421 ClearPagePrivate(page); \ 1422 put_page(page); \ 1423 }\ 1424 } \ 1425 } 1426 1427 PAGE_PRIVATE_GET_FUNC(nonpointer, NOT_POINTER); 1428 PAGE_PRIVATE_GET_FUNC(reference, REF_RESOURCE); 1429 PAGE_PRIVATE_GET_FUNC(inline, INLINE_INODE); 1430 PAGE_PRIVATE_GET_FUNC(gcing, ONGOING_MIGRATION); 1431 PAGE_PRIVATE_GET_FUNC(atomic, ATOMIC_WRITE); 1432 PAGE_PRIVATE_GET_FUNC(dummy, DUMMY_WRITE); 1433 1434 PAGE_PRIVATE_SET_FUNC(reference, REF_RESOURCE); 1435 PAGE_PRIVATE_SET_FUNC(inline, INLINE_INODE); 1436 PAGE_PRIVATE_SET_FUNC(gcing, ONGOING_MIGRATION); 1437 PAGE_PRIVATE_SET_FUNC(atomic, ATOMIC_WRITE); 1438 PAGE_PRIVATE_SET_FUNC(dummy, DUMMY_WRITE); 1439 1440 PAGE_PRIVATE_CLEAR_FUNC(reference, REF_RESOURCE); 1441 PAGE_PRIVATE_CLEAR_FUNC(inline, INLINE_INODE); 1442 PAGE_PRIVATE_CLEAR_FUNC(gcing, ONGOING_MIGRATION); 1443 PAGE_PRIVATE_CLEAR_FUNC(atomic, ATOMIC_WRITE); 1444 PAGE_PRIVATE_CLEAR_FUNC(dummy, DUMMY_WRITE); 1445 1446 static inline unsigned long get_page_private_data(struct page *page) 1447 { 1448 unsigned long data = page_private(page); 1449 1450 if (!test_bit(PAGE_PRIVATE_NOT_POINTER, &data)) 1451 return 0; 1452 return data >> PAGE_PRIVATE_MAX; 1453 } 1454 1455 static inline void set_page_private_data(struct page *page, unsigned long data) 1456 { 1457 if (!PagePrivate(page)) { 1458 get_page(page); 1459 SetPagePrivate(page); 1460 set_page_private(page, 0); 1461 } 1462 set_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)); 1463 page_private(page) |= data << PAGE_PRIVATE_MAX; 1464 } 1465 1466 static inline void clear_page_private_data(struct page *page) 1467 { 1468 page_private(page) &= (1 << PAGE_PRIVATE_MAX) - 1; 1469 if (page_private(page) == 1 << PAGE_PRIVATE_NOT_POINTER) { 1470 set_page_private(page, 0); 1471 if (PagePrivate(page)) { 1472 ClearPagePrivate(page); 1473 put_page(page); 1474 } 1475 } 1476 } 1477 1478 /* For compression */ 1479 enum compress_algorithm_type { 1480 COMPRESS_LZO, 1481 COMPRESS_LZ4, 1482 COMPRESS_ZSTD, 1483 COMPRESS_LZORLE, 1484 COMPRESS_MAX, 1485 }; 1486 1487 enum compress_flag { 1488 COMPRESS_CHKSUM, 1489 COMPRESS_MAX_FLAG, 1490 }; 1491 1492 #define COMPRESS_WATERMARK 20 1493 #define COMPRESS_PERCENT 20 1494 1495 #define COMPRESS_DATA_RESERVED_SIZE 4 1496 struct compress_data { 1497 __le32 clen; /* compressed data size */ 1498 __le32 chksum; /* compressed data chksum */ 1499 __le32 reserved[COMPRESS_DATA_RESERVED_SIZE]; /* reserved */ 1500 u8 cdata[]; /* compressed data */ 1501 }; 1502 1503 #define COMPRESS_HEADER_SIZE (sizeof(struct compress_data)) 1504 1505 #define F2FS_COMPRESSED_PAGE_MAGIC 0xF5F2C000 1506 1507 #define COMPRESS_LEVEL_OFFSET 8 1508 1509 /* compress context */ 1510 struct compress_ctx { 1511 struct inode *inode; /* inode the context belong to */ 1512 pgoff_t cluster_idx; /* cluster index number */ 1513 unsigned int cluster_size; /* page count in cluster */ 1514 unsigned int log_cluster_size; /* log of cluster size */ 1515 struct page **rpages; /* pages store raw data in cluster */ 1516 unsigned int nr_rpages; /* total page number in rpages */ 1517 struct page **cpages; /* pages store compressed data in cluster */ 1518 unsigned int nr_cpages; /* total page number in cpages */ 1519 unsigned int valid_nr_cpages; /* valid page number in cpages */ 1520 void *rbuf; /* virtual mapped address on rpages */ 1521 struct compress_data *cbuf; /* virtual mapped address on cpages */ 1522 size_t rlen; /* valid data length in rbuf */ 1523 size_t clen; /* valid data length in cbuf */ 1524 void *private; /* payload buffer for specified compression algorithm */ 1525 void *private2; /* extra payload buffer */ 1526 }; 1527 1528 /* compress context for write IO path */ 1529 struct compress_io_ctx { 1530 u32 magic; /* magic number to indicate page is compressed */ 1531 struct inode *inode; /* inode the context belong to */ 1532 struct page **rpages; /* pages store raw data in cluster */ 1533 unsigned int nr_rpages; /* total page number in rpages */ 1534 atomic_t pending_pages; /* in-flight compressed page count */ 1535 }; 1536 1537 /* Context for decompressing one cluster on the read IO path */ 1538 struct decompress_io_ctx { 1539 u32 magic; /* magic number to indicate page is compressed */ 1540 struct inode *inode; /* inode the context belong to */ 1541 pgoff_t cluster_idx; /* cluster index number */ 1542 unsigned int cluster_size; /* page count in cluster */ 1543 unsigned int log_cluster_size; /* log of cluster size */ 1544 struct page **rpages; /* pages store raw data in cluster */ 1545 unsigned int nr_rpages; /* total page number in rpages */ 1546 struct page **cpages; /* pages store compressed data in cluster */ 1547 unsigned int nr_cpages; /* total page number in cpages */ 1548 struct page **tpages; /* temp pages to pad holes in cluster */ 1549 void *rbuf; /* virtual mapped address on rpages */ 1550 struct compress_data *cbuf; /* virtual mapped address on cpages */ 1551 size_t rlen; /* valid data length in rbuf */ 1552 size_t clen; /* valid data length in cbuf */ 1553 1554 /* 1555 * The number of compressed pages remaining to be read in this cluster. 1556 * This is initially nr_cpages. It is decremented by 1 each time a page 1557 * has been read (or failed to be read). When it reaches 0, the cluster 1558 * is decompressed (or an error is reported). 1559 * 1560 * If an error occurs before all the pages have been submitted for I/O, 1561 * then this will never reach 0. In this case the I/O submitter is 1562 * responsible for calling f2fs_decompress_end_io() instead. 1563 */ 1564 atomic_t remaining_pages; 1565 1566 /* 1567 * Number of references to this decompress_io_ctx. 1568 * 1569 * One reference is held for I/O completion. This reference is dropped 1570 * after the pagecache pages are updated and unlocked -- either after 1571 * decompression (and verity if enabled), or after an error. 1572 * 1573 * In addition, each compressed page holds a reference while it is in a 1574 * bio. These references are necessary prevent compressed pages from 1575 * being freed while they are still in a bio. 1576 */ 1577 refcount_t refcnt; 1578 1579 bool failed; /* IO error occurred before decompression? */ 1580 bool need_verity; /* need fs-verity verification after decompression? */ 1581 void *private; /* payload buffer for specified decompression algorithm */ 1582 void *private2; /* extra payload buffer */ 1583 struct work_struct verity_work; /* work to verify the decompressed pages */ 1584 }; 1585 1586 #define NULL_CLUSTER ((unsigned int)(~0)) 1587 #define MIN_COMPRESS_LOG_SIZE 2 1588 #define MAX_COMPRESS_LOG_SIZE 8 1589 #define MAX_COMPRESS_WINDOW_SIZE(log_size) ((PAGE_SIZE) << (log_size)) 1590 1591 struct f2fs_sb_info { 1592 struct super_block *sb; /* pointer to VFS super block */ 1593 struct proc_dir_entry *s_proc; /* proc entry */ 1594 struct f2fs_super_block *raw_super; /* raw super block pointer */ 1595 struct f2fs_rwsem sb_lock; /* lock for raw super block */ 1596 int valid_super_block; /* valid super block no */ 1597 unsigned long s_flag; /* flags for sbi */ 1598 struct mutex writepages; /* mutex for writepages() */ 1599 1600 #ifdef CONFIG_BLK_DEV_ZONED 1601 unsigned int blocks_per_blkz; /* F2FS blocks per zone */ 1602 unsigned int log_blocks_per_blkz; /* log2 F2FS blocks per zone */ 1603 #endif 1604 1605 /* for node-related operations */ 1606 struct f2fs_nm_info *nm_info; /* node manager */ 1607 struct inode *node_inode; /* cache node blocks */ 1608 1609 /* for segment-related operations */ 1610 struct f2fs_sm_info *sm_info; /* segment manager */ 1611 1612 /* for bio operations */ 1613 struct f2fs_bio_info *write_io[NR_PAGE_TYPE]; /* for write bios */ 1614 /* keep migration IO order for LFS mode */ 1615 struct f2fs_rwsem io_order_lock; 1616 mempool_t *write_io_dummy; /* Dummy pages */ 1617 pgoff_t metapage_eio_ofs; /* EIO page offset */ 1618 int metapage_eio_cnt; /* EIO count */ 1619 1620 /* for checkpoint */ 1621 struct f2fs_checkpoint *ckpt; /* raw checkpoint pointer */ 1622 int cur_cp_pack; /* remain current cp pack */ 1623 spinlock_t cp_lock; /* for flag in ckpt */ 1624 struct inode *meta_inode; /* cache meta blocks */ 1625 struct f2fs_rwsem cp_global_sem; /* checkpoint procedure lock */ 1626 struct f2fs_rwsem cp_rwsem; /* blocking FS operations */ 1627 struct f2fs_rwsem node_write; /* locking node writes */ 1628 struct f2fs_rwsem node_change; /* locking node change */ 1629 wait_queue_head_t cp_wait; 1630 unsigned long last_time[MAX_TIME]; /* to store time in jiffies */ 1631 long interval_time[MAX_TIME]; /* to store thresholds */ 1632 struct ckpt_req_control cprc_info; /* for checkpoint request control */ 1633 1634 struct inode_management im[MAX_INO_ENTRY]; /* manage inode cache */ 1635 1636 spinlock_t fsync_node_lock; /* for node entry lock */ 1637 struct list_head fsync_node_list; /* node list head */ 1638 unsigned int fsync_seg_id; /* sequence id */ 1639 unsigned int fsync_node_num; /* number of node entries */ 1640 1641 /* for orphan inode, use 0'th array */ 1642 unsigned int max_orphans; /* max orphan inodes */ 1643 1644 /* for inode management */ 1645 struct list_head inode_list[NR_INODE_TYPE]; /* dirty inode list */ 1646 spinlock_t inode_lock[NR_INODE_TYPE]; /* for dirty inode list lock */ 1647 struct mutex flush_lock; /* for flush exclusion */ 1648 1649 /* for extent tree cache */ 1650 struct radix_tree_root extent_tree_root;/* cache extent cache entries */ 1651 struct mutex extent_tree_lock; /* locking extent radix tree */ 1652 struct list_head extent_list; /* lru list for shrinker */ 1653 spinlock_t extent_lock; /* locking extent lru list */ 1654 atomic_t total_ext_tree; /* extent tree count */ 1655 struct list_head zombie_list; /* extent zombie tree list */ 1656 atomic_t total_zombie_tree; /* extent zombie tree count */ 1657 atomic_t total_ext_node; /* extent info count */ 1658 1659 /* basic filesystem units */ 1660 unsigned int log_sectors_per_block; /* log2 sectors per block */ 1661 unsigned int log_blocksize; /* log2 block size */ 1662 unsigned int blocksize; /* block size */ 1663 unsigned int root_ino_num; /* root inode number*/ 1664 unsigned int node_ino_num; /* node inode number*/ 1665 unsigned int meta_ino_num; /* meta inode number*/ 1666 unsigned int log_blocks_per_seg; /* log2 blocks per segment */ 1667 unsigned int blocks_per_seg; /* blocks per segment */ 1668 unsigned int segs_per_sec; /* segments per section */ 1669 unsigned int secs_per_zone; /* sections per zone */ 1670 unsigned int total_sections; /* total section count */ 1671 unsigned int total_node_count; /* total node block count */ 1672 unsigned int total_valid_node_count; /* valid node block count */ 1673 int dir_level; /* directory level */ 1674 int readdir_ra; /* readahead inode in readdir */ 1675 u64 max_io_bytes; /* max io bytes to merge IOs */ 1676 1677 block_t user_block_count; /* # of user blocks */ 1678 block_t total_valid_block_count; /* # of valid blocks */ 1679 block_t discard_blks; /* discard command candidats */ 1680 block_t last_valid_block_count; /* for recovery */ 1681 block_t reserved_blocks; /* configurable reserved blocks */ 1682 block_t current_reserved_blocks; /* current reserved blocks */ 1683 1684 /* Additional tracking for no checkpoint mode */ 1685 block_t unusable_block_count; /* # of blocks saved by last cp */ 1686 1687 unsigned int nquota_files; /* # of quota sysfile */ 1688 struct f2fs_rwsem quota_sem; /* blocking cp for flags */ 1689 1690 /* # of pages, see count_type */ 1691 atomic_t nr_pages[NR_COUNT_TYPE]; 1692 /* # of allocated blocks */ 1693 struct percpu_counter alloc_valid_block_count; 1694 /* # of node block writes as roll forward recovery */ 1695 struct percpu_counter rf_node_block_count; 1696 1697 /* writeback control */ 1698 atomic_t wb_sync_req[META]; /* count # of WB_SYNC threads */ 1699 1700 /* valid inode count */ 1701 struct percpu_counter total_valid_inode_count; 1702 1703 struct f2fs_mount_info mount_opt; /* mount options */ 1704 1705 /* for cleaning operations */ 1706 struct f2fs_rwsem gc_lock; /* 1707 * semaphore for GC, avoid 1708 * race between GC and GC or CP 1709 */ 1710 struct f2fs_gc_kthread *gc_thread; /* GC thread */ 1711 struct atgc_management am; /* atgc management */ 1712 unsigned int cur_victim_sec; /* current victim section num */ 1713 unsigned int gc_mode; /* current GC state */ 1714 unsigned int next_victim_seg[2]; /* next segment in victim section */ 1715 spinlock_t gc_urgent_high_lock; 1716 bool gc_urgent_high_limited; /* indicates having limited trial count */ 1717 unsigned int gc_urgent_high_remaining; /* remaining trial count for GC_URGENT_HIGH */ 1718 1719 /* for skip statistic */ 1720 unsigned int atomic_files; /* # of opened atomic file */ 1721 unsigned long long skipped_atomic_files[2]; /* FG_GC and BG_GC */ 1722 unsigned long long skipped_gc_rwsem; /* FG_GC only */ 1723 1724 /* threshold for gc trials on pinned files */ 1725 u64 gc_pin_file_threshold; 1726 struct f2fs_rwsem pin_sem; 1727 1728 /* maximum # of trials to find a victim segment for SSR and GC */ 1729 unsigned int max_victim_search; 1730 /* migration granularity of garbage collection, unit: segment */ 1731 unsigned int migration_granularity; 1732 1733 /* 1734 * for stat information. 1735 * one is for the LFS mode, and the other is for the SSR mode. 1736 */ 1737 #ifdef CONFIG_F2FS_STAT_FS 1738 struct f2fs_stat_info *stat_info; /* FS status information */ 1739 atomic_t meta_count[META_MAX]; /* # of meta blocks */ 1740 unsigned int segment_count[2]; /* # of allocated segments */ 1741 unsigned int block_count[2]; /* # of allocated blocks */ 1742 atomic_t inplace_count; /* # of inplace update */ 1743 atomic64_t total_hit_ext; /* # of lookup extent cache */ 1744 atomic64_t read_hit_rbtree; /* # of hit rbtree extent node */ 1745 atomic64_t read_hit_largest; /* # of hit largest extent node */ 1746 atomic64_t read_hit_cached; /* # of hit cached extent node */ 1747 atomic_t inline_xattr; /* # of inline_xattr inodes */ 1748 atomic_t inline_inode; /* # of inline_data inodes */ 1749 atomic_t inline_dir; /* # of inline_dentry inodes */ 1750 atomic_t compr_inode; /* # of compressed inodes */ 1751 atomic64_t compr_blocks; /* # of compressed blocks */ 1752 atomic_t vw_cnt; /* # of volatile writes */ 1753 atomic_t max_aw_cnt; /* max # of atomic writes */ 1754 atomic_t max_vw_cnt; /* max # of volatile writes */ 1755 unsigned int io_skip_bggc; /* skip background gc for in-flight IO */ 1756 unsigned int other_skip_bggc; /* skip background gc for other reasons */ 1757 unsigned int ndirty_inode[NR_INODE_TYPE]; /* # of dirty inodes */ 1758 #endif 1759 spinlock_t stat_lock; /* lock for stat operations */ 1760 1761 /* to attach REQ_META|REQ_FUA flags */ 1762 unsigned int data_io_flag; 1763 unsigned int node_io_flag; 1764 1765 /* For sysfs suppport */ 1766 struct kobject s_kobj; /* /sys/fs/f2fs/<devname> */ 1767 struct completion s_kobj_unregister; 1768 1769 struct kobject s_stat_kobj; /* /sys/fs/f2fs/<devname>/stat */ 1770 struct completion s_stat_kobj_unregister; 1771 1772 struct kobject s_feature_list_kobj; /* /sys/fs/f2fs/<devname>/feature_list */ 1773 struct completion s_feature_list_kobj_unregister; 1774 1775 /* For shrinker support */ 1776 struct list_head s_list; 1777 struct mutex umount_mutex; 1778 unsigned int shrinker_run_no; 1779 1780 /* For multi devices */ 1781 int s_ndevs; /* number of devices */ 1782 struct f2fs_dev_info *devs; /* for device list */ 1783 unsigned int dirty_device; /* for checkpoint data flush */ 1784 spinlock_t dev_lock; /* protect dirty_device */ 1785 bool aligned_blksize; /* all devices has the same logical blksize */ 1786 1787 /* For write statistics */ 1788 u64 sectors_written_start; 1789 u64 kbytes_written; 1790 1791 /* Reference to checksum algorithm driver via cryptoapi */ 1792 struct crypto_shash *s_chksum_driver; 1793 1794 /* Precomputed FS UUID checksum for seeding other checksums */ 1795 __u32 s_chksum_seed; 1796 1797 struct workqueue_struct *post_read_wq; /* post read workqueue */ 1798 1799 struct kmem_cache *inline_xattr_slab; /* inline xattr entry */ 1800 unsigned int inline_xattr_slab_size; /* default inline xattr slab size */ 1801 1802 /* For reclaimed segs statistics per each GC mode */ 1803 unsigned int gc_segment_mode; /* GC state for reclaimed segments */ 1804 unsigned int gc_reclaimed_segs[MAX_GC_MODE]; /* Reclaimed segs for each mode */ 1805 1806 unsigned long seq_file_ra_mul; /* multiplier for ra_pages of seq. files in fadvise */ 1807 1808 int max_fragment_chunk; /* max chunk size for block fragmentation mode */ 1809 int max_fragment_hole; /* max hole size for block fragmentation mode */ 1810 1811 #ifdef CONFIG_F2FS_FS_COMPRESSION 1812 struct kmem_cache *page_array_slab; /* page array entry */ 1813 unsigned int page_array_slab_size; /* default page array slab size */ 1814 1815 /* For runtime compression statistics */ 1816 u64 compr_written_block; 1817 u64 compr_saved_block; 1818 u32 compr_new_inode; 1819 1820 /* For compressed block cache */ 1821 struct inode *compress_inode; /* cache compressed blocks */ 1822 unsigned int compress_percent; /* cache page percentage */ 1823 unsigned int compress_watermark; /* cache page watermark */ 1824 atomic_t compress_page_hit; /* cache hit count */ 1825 #endif 1826 1827 #ifdef CONFIG_F2FS_IOSTAT 1828 /* For app/fs IO statistics */ 1829 spinlock_t iostat_lock; 1830 unsigned long long rw_iostat[NR_IO_TYPE]; 1831 unsigned long long prev_rw_iostat[NR_IO_TYPE]; 1832 bool iostat_enable; 1833 unsigned long iostat_next_period; 1834 unsigned int iostat_period_ms; 1835 1836 /* For io latency related statistics info in one iostat period */ 1837 spinlock_t iostat_lat_lock; 1838 struct iostat_lat_info *iostat_io_lat; 1839 #endif 1840 }; 1841 1842 #ifdef CONFIG_F2FS_FAULT_INJECTION 1843 #define f2fs_show_injection_info(sbi, type) \ 1844 printk_ratelimited("%sF2FS-fs (%s) : inject %s in %s of %pS\n", \ 1845 KERN_INFO, sbi->sb->s_id, \ 1846 f2fs_fault_name[type], \ 1847 __func__, __builtin_return_address(0)) 1848 static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type) 1849 { 1850 struct f2fs_fault_info *ffi = &F2FS_OPTION(sbi).fault_info; 1851 1852 if (!ffi->inject_rate) 1853 return false; 1854 1855 if (!IS_FAULT_SET(ffi, type)) 1856 return false; 1857 1858 atomic_inc(&ffi->inject_ops); 1859 if (atomic_read(&ffi->inject_ops) >= ffi->inject_rate) { 1860 atomic_set(&ffi->inject_ops, 0); 1861 return true; 1862 } 1863 return false; 1864 } 1865 #else 1866 #define f2fs_show_injection_info(sbi, type) do { } while (0) 1867 static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type) 1868 { 1869 return false; 1870 } 1871 #endif 1872 1873 /* 1874 * Test if the mounted volume is a multi-device volume. 1875 * - For a single regular disk volume, sbi->s_ndevs is 0. 1876 * - For a single zoned disk volume, sbi->s_ndevs is 1. 1877 * - For a multi-device volume, sbi->s_ndevs is always 2 or more. 1878 */ 1879 static inline bool f2fs_is_multi_device(struct f2fs_sb_info *sbi) 1880 { 1881 return sbi->s_ndevs > 1; 1882 } 1883 1884 static inline void f2fs_update_time(struct f2fs_sb_info *sbi, int type) 1885 { 1886 unsigned long now = jiffies; 1887 1888 sbi->last_time[type] = now; 1889 1890 /* DISCARD_TIME and GC_TIME are based on REQ_TIME */ 1891 if (type == REQ_TIME) { 1892 sbi->last_time[DISCARD_TIME] = now; 1893 sbi->last_time[GC_TIME] = now; 1894 } 1895 } 1896 1897 static inline bool f2fs_time_over(struct f2fs_sb_info *sbi, int type) 1898 { 1899 unsigned long interval = sbi->interval_time[type] * HZ; 1900 1901 return time_after(jiffies, sbi->last_time[type] + interval); 1902 } 1903 1904 static inline unsigned int f2fs_time_to_wait(struct f2fs_sb_info *sbi, 1905 int type) 1906 { 1907 unsigned long interval = sbi->interval_time[type] * HZ; 1908 unsigned int wait_ms = 0; 1909 long delta; 1910 1911 delta = (sbi->last_time[type] + interval) - jiffies; 1912 if (delta > 0) 1913 wait_ms = jiffies_to_msecs(delta); 1914 1915 return wait_ms; 1916 } 1917 1918 /* 1919 * Inline functions 1920 */ 1921 static inline u32 __f2fs_crc32(struct f2fs_sb_info *sbi, u32 crc, 1922 const void *address, unsigned int length) 1923 { 1924 struct { 1925 struct shash_desc shash; 1926 char ctx[4]; 1927 } desc; 1928 int err; 1929 1930 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver) != sizeof(desc.ctx)); 1931 1932 desc.shash.tfm = sbi->s_chksum_driver; 1933 *(u32 *)desc.ctx = crc; 1934 1935 err = crypto_shash_update(&desc.shash, address, length); 1936 BUG_ON(err); 1937 1938 return *(u32 *)desc.ctx; 1939 } 1940 1941 static inline u32 f2fs_crc32(struct f2fs_sb_info *sbi, const void *address, 1942 unsigned int length) 1943 { 1944 return __f2fs_crc32(sbi, F2FS_SUPER_MAGIC, address, length); 1945 } 1946 1947 static inline bool f2fs_crc_valid(struct f2fs_sb_info *sbi, __u32 blk_crc, 1948 void *buf, size_t buf_size) 1949 { 1950 return f2fs_crc32(sbi, buf, buf_size) == blk_crc; 1951 } 1952 1953 static inline u32 f2fs_chksum(struct f2fs_sb_info *sbi, u32 crc, 1954 const void *address, unsigned int length) 1955 { 1956 return __f2fs_crc32(sbi, crc, address, length); 1957 } 1958 1959 static inline struct f2fs_inode_info *F2FS_I(struct inode *inode) 1960 { 1961 return container_of(inode, struct f2fs_inode_info, vfs_inode); 1962 } 1963 1964 static inline struct f2fs_sb_info *F2FS_SB(struct super_block *sb) 1965 { 1966 return sb->s_fs_info; 1967 } 1968 1969 static inline struct f2fs_sb_info *F2FS_I_SB(struct inode *inode) 1970 { 1971 return F2FS_SB(inode->i_sb); 1972 } 1973 1974 static inline struct f2fs_sb_info *F2FS_M_SB(struct address_space *mapping) 1975 { 1976 return F2FS_I_SB(mapping->host); 1977 } 1978 1979 static inline struct f2fs_sb_info *F2FS_P_SB(struct page *page) 1980 { 1981 return F2FS_M_SB(page_file_mapping(page)); 1982 } 1983 1984 static inline struct f2fs_super_block *F2FS_RAW_SUPER(struct f2fs_sb_info *sbi) 1985 { 1986 return (struct f2fs_super_block *)(sbi->raw_super); 1987 } 1988 1989 static inline struct f2fs_checkpoint *F2FS_CKPT(struct f2fs_sb_info *sbi) 1990 { 1991 return (struct f2fs_checkpoint *)(sbi->ckpt); 1992 } 1993 1994 static inline struct f2fs_node *F2FS_NODE(struct page *page) 1995 { 1996 return (struct f2fs_node *)page_address(page); 1997 } 1998 1999 static inline struct f2fs_inode *F2FS_INODE(struct page *page) 2000 { 2001 return &((struct f2fs_node *)page_address(page))->i; 2002 } 2003 2004 static inline struct f2fs_nm_info *NM_I(struct f2fs_sb_info *sbi) 2005 { 2006 return (struct f2fs_nm_info *)(sbi->nm_info); 2007 } 2008 2009 static inline struct f2fs_sm_info *SM_I(struct f2fs_sb_info *sbi) 2010 { 2011 return (struct f2fs_sm_info *)(sbi->sm_info); 2012 } 2013 2014 static inline struct sit_info *SIT_I(struct f2fs_sb_info *sbi) 2015 { 2016 return (struct sit_info *)(SM_I(sbi)->sit_info); 2017 } 2018 2019 static inline struct free_segmap_info *FREE_I(struct f2fs_sb_info *sbi) 2020 { 2021 return (struct free_segmap_info *)(SM_I(sbi)->free_info); 2022 } 2023 2024 static inline struct dirty_seglist_info *DIRTY_I(struct f2fs_sb_info *sbi) 2025 { 2026 return (struct dirty_seglist_info *)(SM_I(sbi)->dirty_info); 2027 } 2028 2029 static inline struct address_space *META_MAPPING(struct f2fs_sb_info *sbi) 2030 { 2031 return sbi->meta_inode->i_mapping; 2032 } 2033 2034 static inline struct address_space *NODE_MAPPING(struct f2fs_sb_info *sbi) 2035 { 2036 return sbi->node_inode->i_mapping; 2037 } 2038 2039 static inline bool is_sbi_flag_set(struct f2fs_sb_info *sbi, unsigned int type) 2040 { 2041 return test_bit(type, &sbi->s_flag); 2042 } 2043 2044 static inline void set_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type) 2045 { 2046 set_bit(type, &sbi->s_flag); 2047 } 2048 2049 static inline void clear_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type) 2050 { 2051 clear_bit(type, &sbi->s_flag); 2052 } 2053 2054 static inline unsigned long long cur_cp_version(struct f2fs_checkpoint *cp) 2055 { 2056 return le64_to_cpu(cp->checkpoint_ver); 2057 } 2058 2059 static inline unsigned long f2fs_qf_ino(struct super_block *sb, int type) 2060 { 2061 if (type < F2FS_MAX_QUOTAS) 2062 return le32_to_cpu(F2FS_SB(sb)->raw_super->qf_ino[type]); 2063 return 0; 2064 } 2065 2066 static inline __u64 cur_cp_crc(struct f2fs_checkpoint *cp) 2067 { 2068 size_t crc_offset = le32_to_cpu(cp->checksum_offset); 2069 return le32_to_cpu(*((__le32 *)((unsigned char *)cp + crc_offset))); 2070 } 2071 2072 static inline bool __is_set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 2073 { 2074 unsigned int ckpt_flags = le32_to_cpu(cp->ckpt_flags); 2075 2076 return ckpt_flags & f; 2077 } 2078 2079 static inline bool is_set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 2080 { 2081 return __is_set_ckpt_flags(F2FS_CKPT(sbi), f); 2082 } 2083 2084 static inline void __set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 2085 { 2086 unsigned int ckpt_flags; 2087 2088 ckpt_flags = le32_to_cpu(cp->ckpt_flags); 2089 ckpt_flags |= f; 2090 cp->ckpt_flags = cpu_to_le32(ckpt_flags); 2091 } 2092 2093 static inline void set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 2094 { 2095 unsigned long flags; 2096 2097 spin_lock_irqsave(&sbi->cp_lock, flags); 2098 __set_ckpt_flags(F2FS_CKPT(sbi), f); 2099 spin_unlock_irqrestore(&sbi->cp_lock, flags); 2100 } 2101 2102 static inline void __clear_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 2103 { 2104 unsigned int ckpt_flags; 2105 2106 ckpt_flags = le32_to_cpu(cp->ckpt_flags); 2107 ckpt_flags &= (~f); 2108 cp->ckpt_flags = cpu_to_le32(ckpt_flags); 2109 } 2110 2111 static inline void clear_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 2112 { 2113 unsigned long flags; 2114 2115 spin_lock_irqsave(&sbi->cp_lock, flags); 2116 __clear_ckpt_flags(F2FS_CKPT(sbi), f); 2117 spin_unlock_irqrestore(&sbi->cp_lock, flags); 2118 } 2119 2120 #define init_f2fs_rwsem(sem) \ 2121 do { \ 2122 static struct lock_class_key __key; \ 2123 \ 2124 __init_f2fs_rwsem((sem), #sem, &__key); \ 2125 } while (0) 2126 2127 static inline void __init_f2fs_rwsem(struct f2fs_rwsem *sem, 2128 const char *sem_name, struct lock_class_key *key) 2129 { 2130 __init_rwsem(&sem->internal_rwsem, sem_name, key); 2131 #ifdef CONFIG_F2FS_UNFAIR_RWSEM 2132 init_waitqueue_head(&sem->read_waiters); 2133 #endif 2134 } 2135 2136 static inline int f2fs_rwsem_is_locked(struct f2fs_rwsem *sem) 2137 { 2138 return rwsem_is_locked(&sem->internal_rwsem); 2139 } 2140 2141 static inline int f2fs_rwsem_is_contended(struct f2fs_rwsem *sem) 2142 { 2143 return rwsem_is_contended(&sem->internal_rwsem); 2144 } 2145 2146 static inline void f2fs_down_read(struct f2fs_rwsem *sem) 2147 { 2148 #ifdef CONFIG_F2FS_UNFAIR_RWSEM 2149 wait_event(sem->read_waiters, down_read_trylock(&sem->internal_rwsem)); 2150 #else 2151 down_read(&sem->internal_rwsem); 2152 #endif 2153 } 2154 2155 static inline int f2fs_down_read_trylock(struct f2fs_rwsem *sem) 2156 { 2157 return down_read_trylock(&sem->internal_rwsem); 2158 } 2159 2160 #ifdef CONFIG_DEBUG_LOCK_ALLOC 2161 static inline void f2fs_down_read_nested(struct f2fs_rwsem *sem, int subclass) 2162 { 2163 down_read_nested(&sem->internal_rwsem, subclass); 2164 } 2165 #else 2166 #define f2fs_down_read_nested(sem, subclass) f2fs_down_read(sem) 2167 #endif 2168 2169 static inline void f2fs_up_read(struct f2fs_rwsem *sem) 2170 { 2171 up_read(&sem->internal_rwsem); 2172 } 2173 2174 static inline void f2fs_down_write(struct f2fs_rwsem *sem) 2175 { 2176 down_write(&sem->internal_rwsem); 2177 } 2178 2179 static inline int f2fs_down_write_trylock(struct f2fs_rwsem *sem) 2180 { 2181 return down_write_trylock(&sem->internal_rwsem); 2182 } 2183 2184 static inline void f2fs_up_write(struct f2fs_rwsem *sem) 2185 { 2186 up_write(&sem->internal_rwsem); 2187 #ifdef CONFIG_F2FS_UNFAIR_RWSEM 2188 wake_up_all(&sem->read_waiters); 2189 #endif 2190 } 2191 2192 static inline void f2fs_lock_op(struct f2fs_sb_info *sbi) 2193 { 2194 f2fs_down_read(&sbi->cp_rwsem); 2195 } 2196 2197 static inline int f2fs_trylock_op(struct f2fs_sb_info *sbi) 2198 { 2199 if (time_to_inject(sbi, FAULT_LOCK_OP)) { 2200 f2fs_show_injection_info(sbi, FAULT_LOCK_OP); 2201 return 0; 2202 } 2203 return f2fs_down_read_trylock(&sbi->cp_rwsem); 2204 } 2205 2206 static inline void f2fs_unlock_op(struct f2fs_sb_info *sbi) 2207 { 2208 f2fs_up_read(&sbi->cp_rwsem); 2209 } 2210 2211 static inline void f2fs_lock_all(struct f2fs_sb_info *sbi) 2212 { 2213 f2fs_down_write(&sbi->cp_rwsem); 2214 } 2215 2216 static inline void f2fs_unlock_all(struct f2fs_sb_info *sbi) 2217 { 2218 f2fs_up_write(&sbi->cp_rwsem); 2219 } 2220 2221 static inline int __get_cp_reason(struct f2fs_sb_info *sbi) 2222 { 2223 int reason = CP_SYNC; 2224 2225 if (test_opt(sbi, FASTBOOT)) 2226 reason = CP_FASTBOOT; 2227 if (is_sbi_flag_set(sbi, SBI_IS_CLOSE)) 2228 reason = CP_UMOUNT; 2229 return reason; 2230 } 2231 2232 static inline bool __remain_node_summaries(int reason) 2233 { 2234 return (reason & (CP_UMOUNT | CP_FASTBOOT)); 2235 } 2236 2237 static inline bool __exist_node_summaries(struct f2fs_sb_info *sbi) 2238 { 2239 return (is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG) || 2240 is_set_ckpt_flags(sbi, CP_FASTBOOT_FLAG)); 2241 } 2242 2243 /* 2244 * Check whether the inode has blocks or not 2245 */ 2246 static inline int F2FS_HAS_BLOCKS(struct inode *inode) 2247 { 2248 block_t xattr_block = F2FS_I(inode)->i_xattr_nid ? 1 : 0; 2249 2250 return (inode->i_blocks >> F2FS_LOG_SECTORS_PER_BLOCK) > xattr_block; 2251 } 2252 2253 static inline bool f2fs_has_xattr_block(unsigned int ofs) 2254 { 2255 return ofs == XATTR_NODE_OFFSET; 2256 } 2257 2258 static inline bool __allow_reserved_blocks(struct f2fs_sb_info *sbi, 2259 struct inode *inode, bool cap) 2260 { 2261 if (!inode) 2262 return true; 2263 if (!test_opt(sbi, RESERVE_ROOT)) 2264 return false; 2265 if (IS_NOQUOTA(inode)) 2266 return true; 2267 if (uid_eq(F2FS_OPTION(sbi).s_resuid, current_fsuid())) 2268 return true; 2269 if (!gid_eq(F2FS_OPTION(sbi).s_resgid, GLOBAL_ROOT_GID) && 2270 in_group_p(F2FS_OPTION(sbi).s_resgid)) 2271 return true; 2272 if (cap && capable(CAP_SYS_RESOURCE)) 2273 return true; 2274 return false; 2275 } 2276 2277 static inline void f2fs_i_blocks_write(struct inode *, block_t, bool, bool); 2278 static inline int inc_valid_block_count(struct f2fs_sb_info *sbi, 2279 struct inode *inode, blkcnt_t *count) 2280 { 2281 blkcnt_t diff = 0, release = 0; 2282 block_t avail_user_block_count; 2283 int ret; 2284 2285 ret = dquot_reserve_block(inode, *count); 2286 if (ret) 2287 return ret; 2288 2289 if (time_to_inject(sbi, FAULT_BLOCK)) { 2290 f2fs_show_injection_info(sbi, FAULT_BLOCK); 2291 release = *count; 2292 goto release_quota; 2293 } 2294 2295 /* 2296 * let's increase this in prior to actual block count change in order 2297 * for f2fs_sync_file to avoid data races when deciding checkpoint. 2298 */ 2299 percpu_counter_add(&sbi->alloc_valid_block_count, (*count)); 2300 2301 spin_lock(&sbi->stat_lock); 2302 sbi->total_valid_block_count += (block_t)(*count); 2303 avail_user_block_count = sbi->user_block_count - 2304 sbi->current_reserved_blocks; 2305 2306 if (!__allow_reserved_blocks(sbi, inode, true)) 2307 avail_user_block_count -= F2FS_OPTION(sbi).root_reserved_blocks; 2308 2309 if (F2FS_IO_ALIGNED(sbi)) 2310 avail_user_block_count -= sbi->blocks_per_seg * 2311 SM_I(sbi)->additional_reserved_segments; 2312 2313 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) { 2314 if (avail_user_block_count > sbi->unusable_block_count) 2315 avail_user_block_count -= sbi->unusable_block_count; 2316 else 2317 avail_user_block_count = 0; 2318 } 2319 if (unlikely(sbi->total_valid_block_count > avail_user_block_count)) { 2320 diff = sbi->total_valid_block_count - avail_user_block_count; 2321 if (diff > *count) 2322 diff = *count; 2323 *count -= diff; 2324 release = diff; 2325 sbi->total_valid_block_count -= diff; 2326 if (!*count) { 2327 spin_unlock(&sbi->stat_lock); 2328 goto enospc; 2329 } 2330 } 2331 spin_unlock(&sbi->stat_lock); 2332 2333 if (unlikely(release)) { 2334 percpu_counter_sub(&sbi->alloc_valid_block_count, release); 2335 dquot_release_reservation_block(inode, release); 2336 } 2337 f2fs_i_blocks_write(inode, *count, true, true); 2338 return 0; 2339 2340 enospc: 2341 percpu_counter_sub(&sbi->alloc_valid_block_count, release); 2342 release_quota: 2343 dquot_release_reservation_block(inode, release); 2344 return -ENOSPC; 2345 } 2346 2347 __printf(2, 3) 2348 void f2fs_printk(struct f2fs_sb_info *sbi, const char *fmt, ...); 2349 2350 #define f2fs_err(sbi, fmt, ...) \ 2351 f2fs_printk(sbi, KERN_ERR fmt, ##__VA_ARGS__) 2352 #define f2fs_warn(sbi, fmt, ...) \ 2353 f2fs_printk(sbi, KERN_WARNING fmt, ##__VA_ARGS__) 2354 #define f2fs_notice(sbi, fmt, ...) \ 2355 f2fs_printk(sbi, KERN_NOTICE fmt, ##__VA_ARGS__) 2356 #define f2fs_info(sbi, fmt, ...) \ 2357 f2fs_printk(sbi, KERN_INFO fmt, ##__VA_ARGS__) 2358 #define f2fs_debug(sbi, fmt, ...) \ 2359 f2fs_printk(sbi, KERN_DEBUG fmt, ##__VA_ARGS__) 2360 2361 static inline void dec_valid_block_count(struct f2fs_sb_info *sbi, 2362 struct inode *inode, 2363 block_t count) 2364 { 2365 blkcnt_t sectors = count << F2FS_LOG_SECTORS_PER_BLOCK; 2366 2367 spin_lock(&sbi->stat_lock); 2368 f2fs_bug_on(sbi, sbi->total_valid_block_count < (block_t) count); 2369 sbi->total_valid_block_count -= (block_t)count; 2370 if (sbi->reserved_blocks && 2371 sbi->current_reserved_blocks < sbi->reserved_blocks) 2372 sbi->current_reserved_blocks = min(sbi->reserved_blocks, 2373 sbi->current_reserved_blocks + count); 2374 spin_unlock(&sbi->stat_lock); 2375 if (unlikely(inode->i_blocks < sectors)) { 2376 f2fs_warn(sbi, "Inconsistent i_blocks, ino:%lu, iblocks:%llu, sectors:%llu", 2377 inode->i_ino, 2378 (unsigned long long)inode->i_blocks, 2379 (unsigned long long)sectors); 2380 set_sbi_flag(sbi, SBI_NEED_FSCK); 2381 return; 2382 } 2383 f2fs_i_blocks_write(inode, count, false, true); 2384 } 2385 2386 static inline void inc_page_count(struct f2fs_sb_info *sbi, int count_type) 2387 { 2388 atomic_inc(&sbi->nr_pages[count_type]); 2389 2390 if (count_type == F2FS_DIRTY_DENTS || 2391 count_type == F2FS_DIRTY_NODES || 2392 count_type == F2FS_DIRTY_META || 2393 count_type == F2FS_DIRTY_QDATA || 2394 count_type == F2FS_DIRTY_IMETA) 2395 set_sbi_flag(sbi, SBI_IS_DIRTY); 2396 } 2397 2398 static inline void inode_inc_dirty_pages(struct inode *inode) 2399 { 2400 atomic_inc(&F2FS_I(inode)->dirty_pages); 2401 inc_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ? 2402 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA); 2403 if (IS_NOQUOTA(inode)) 2404 inc_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA); 2405 } 2406 2407 static inline void dec_page_count(struct f2fs_sb_info *sbi, int count_type) 2408 { 2409 atomic_dec(&sbi->nr_pages[count_type]); 2410 } 2411 2412 static inline void inode_dec_dirty_pages(struct inode *inode) 2413 { 2414 if (!S_ISDIR(inode->i_mode) && !S_ISREG(inode->i_mode) && 2415 !S_ISLNK(inode->i_mode)) 2416 return; 2417 2418 atomic_dec(&F2FS_I(inode)->dirty_pages); 2419 dec_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ? 2420 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA); 2421 if (IS_NOQUOTA(inode)) 2422 dec_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA); 2423 } 2424 2425 static inline s64 get_pages(struct f2fs_sb_info *sbi, int count_type) 2426 { 2427 return atomic_read(&sbi->nr_pages[count_type]); 2428 } 2429 2430 static inline int get_dirty_pages(struct inode *inode) 2431 { 2432 return atomic_read(&F2FS_I(inode)->dirty_pages); 2433 } 2434 2435 static inline int get_blocktype_secs(struct f2fs_sb_info *sbi, int block_type) 2436 { 2437 unsigned int pages_per_sec = sbi->segs_per_sec * sbi->blocks_per_seg; 2438 unsigned int segs = (get_pages(sbi, block_type) + pages_per_sec - 1) >> 2439 sbi->log_blocks_per_seg; 2440 2441 return segs / sbi->segs_per_sec; 2442 } 2443 2444 static inline block_t valid_user_blocks(struct f2fs_sb_info *sbi) 2445 { 2446 return sbi->total_valid_block_count; 2447 } 2448 2449 static inline block_t discard_blocks(struct f2fs_sb_info *sbi) 2450 { 2451 return sbi->discard_blks; 2452 } 2453 2454 static inline unsigned long __bitmap_size(struct f2fs_sb_info *sbi, int flag) 2455 { 2456 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi); 2457 2458 /* return NAT or SIT bitmap */ 2459 if (flag == NAT_BITMAP) 2460 return le32_to_cpu(ckpt->nat_ver_bitmap_bytesize); 2461 else if (flag == SIT_BITMAP) 2462 return le32_to_cpu(ckpt->sit_ver_bitmap_bytesize); 2463 2464 return 0; 2465 } 2466 2467 static inline block_t __cp_payload(struct f2fs_sb_info *sbi) 2468 { 2469 return le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_payload); 2470 } 2471 2472 static inline void *__bitmap_ptr(struct f2fs_sb_info *sbi, int flag) 2473 { 2474 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi); 2475 void *tmp_ptr = &ckpt->sit_nat_version_bitmap; 2476 int offset; 2477 2478 if (is_set_ckpt_flags(sbi, CP_LARGE_NAT_BITMAP_FLAG)) { 2479 offset = (flag == SIT_BITMAP) ? 2480 le32_to_cpu(ckpt->nat_ver_bitmap_bytesize) : 0; 2481 /* 2482 * if large_nat_bitmap feature is enabled, leave checksum 2483 * protection for all nat/sit bitmaps. 2484 */ 2485 return tmp_ptr + offset + sizeof(__le32); 2486 } 2487 2488 if (__cp_payload(sbi) > 0) { 2489 if (flag == NAT_BITMAP) 2490 return &ckpt->sit_nat_version_bitmap; 2491 else 2492 return (unsigned char *)ckpt + F2FS_BLKSIZE; 2493 } else { 2494 offset = (flag == NAT_BITMAP) ? 2495 le32_to_cpu(ckpt->sit_ver_bitmap_bytesize) : 0; 2496 return tmp_ptr + offset; 2497 } 2498 } 2499 2500 static inline block_t __start_cp_addr(struct f2fs_sb_info *sbi) 2501 { 2502 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr); 2503 2504 if (sbi->cur_cp_pack == 2) 2505 start_addr += sbi->blocks_per_seg; 2506 return start_addr; 2507 } 2508 2509 static inline block_t __start_cp_next_addr(struct f2fs_sb_info *sbi) 2510 { 2511 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr); 2512 2513 if (sbi->cur_cp_pack == 1) 2514 start_addr += sbi->blocks_per_seg; 2515 return start_addr; 2516 } 2517 2518 static inline void __set_cp_next_pack(struct f2fs_sb_info *sbi) 2519 { 2520 sbi->cur_cp_pack = (sbi->cur_cp_pack == 1) ? 2 : 1; 2521 } 2522 2523 static inline block_t __start_sum_addr(struct f2fs_sb_info *sbi) 2524 { 2525 return le32_to_cpu(F2FS_CKPT(sbi)->cp_pack_start_sum); 2526 } 2527 2528 static inline int inc_valid_node_count(struct f2fs_sb_info *sbi, 2529 struct inode *inode, bool is_inode) 2530 { 2531 block_t valid_block_count; 2532 unsigned int valid_node_count, user_block_count; 2533 int err; 2534 2535 if (is_inode) { 2536 if (inode) { 2537 err = dquot_alloc_inode(inode); 2538 if (err) 2539 return err; 2540 } 2541 } else { 2542 err = dquot_reserve_block(inode, 1); 2543 if (err) 2544 return err; 2545 } 2546 2547 if (time_to_inject(sbi, FAULT_BLOCK)) { 2548 f2fs_show_injection_info(sbi, FAULT_BLOCK); 2549 goto enospc; 2550 } 2551 2552 spin_lock(&sbi->stat_lock); 2553 2554 valid_block_count = sbi->total_valid_block_count + 2555 sbi->current_reserved_blocks + 1; 2556 2557 if (!__allow_reserved_blocks(sbi, inode, false)) 2558 valid_block_count += F2FS_OPTION(sbi).root_reserved_blocks; 2559 2560 if (F2FS_IO_ALIGNED(sbi)) 2561 valid_block_count += sbi->blocks_per_seg * 2562 SM_I(sbi)->additional_reserved_segments; 2563 2564 user_block_count = sbi->user_block_count; 2565 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) 2566 user_block_count -= sbi->unusable_block_count; 2567 2568 if (unlikely(valid_block_count > user_block_count)) { 2569 spin_unlock(&sbi->stat_lock); 2570 goto enospc; 2571 } 2572 2573 valid_node_count = sbi->total_valid_node_count + 1; 2574 if (unlikely(valid_node_count > sbi->total_node_count)) { 2575 spin_unlock(&sbi->stat_lock); 2576 goto enospc; 2577 } 2578 2579 sbi->total_valid_node_count++; 2580 sbi->total_valid_block_count++; 2581 spin_unlock(&sbi->stat_lock); 2582 2583 if (inode) { 2584 if (is_inode) 2585 f2fs_mark_inode_dirty_sync(inode, true); 2586 else 2587 f2fs_i_blocks_write(inode, 1, true, true); 2588 } 2589 2590 percpu_counter_inc(&sbi->alloc_valid_block_count); 2591 return 0; 2592 2593 enospc: 2594 if (is_inode) { 2595 if (inode) 2596 dquot_free_inode(inode); 2597 } else { 2598 dquot_release_reservation_block(inode, 1); 2599 } 2600 return -ENOSPC; 2601 } 2602 2603 static inline void dec_valid_node_count(struct f2fs_sb_info *sbi, 2604 struct inode *inode, bool is_inode) 2605 { 2606 spin_lock(&sbi->stat_lock); 2607 2608 f2fs_bug_on(sbi, !sbi->total_valid_block_count); 2609 f2fs_bug_on(sbi, !sbi->total_valid_node_count); 2610 2611 sbi->total_valid_node_count--; 2612 sbi->total_valid_block_count--; 2613 if (sbi->reserved_blocks && 2614 sbi->current_reserved_blocks < sbi->reserved_blocks) 2615 sbi->current_reserved_blocks++; 2616 2617 spin_unlock(&sbi->stat_lock); 2618 2619 if (is_inode) { 2620 dquot_free_inode(inode); 2621 } else { 2622 if (unlikely(inode->i_blocks == 0)) { 2623 f2fs_warn(sbi, "dec_valid_node_count: inconsistent i_blocks, ino:%lu, iblocks:%llu", 2624 inode->i_ino, 2625 (unsigned long long)inode->i_blocks); 2626 set_sbi_flag(sbi, SBI_NEED_FSCK); 2627 return; 2628 } 2629 f2fs_i_blocks_write(inode, 1, false, true); 2630 } 2631 } 2632 2633 static inline unsigned int valid_node_count(struct f2fs_sb_info *sbi) 2634 { 2635 return sbi->total_valid_node_count; 2636 } 2637 2638 static inline void inc_valid_inode_count(struct f2fs_sb_info *sbi) 2639 { 2640 percpu_counter_inc(&sbi->total_valid_inode_count); 2641 } 2642 2643 static inline void dec_valid_inode_count(struct f2fs_sb_info *sbi) 2644 { 2645 percpu_counter_dec(&sbi->total_valid_inode_count); 2646 } 2647 2648 static inline s64 valid_inode_count(struct f2fs_sb_info *sbi) 2649 { 2650 return percpu_counter_sum_positive(&sbi->total_valid_inode_count); 2651 } 2652 2653 static inline struct page *f2fs_grab_cache_page(struct address_space *mapping, 2654 pgoff_t index, bool for_write) 2655 { 2656 struct page *page; 2657 2658 if (IS_ENABLED(CONFIG_F2FS_FAULT_INJECTION)) { 2659 if (!for_write) 2660 page = find_get_page_flags(mapping, index, 2661 FGP_LOCK | FGP_ACCESSED); 2662 else 2663 page = find_lock_page(mapping, index); 2664 if (page) 2665 return page; 2666 2667 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_ALLOC)) { 2668 f2fs_show_injection_info(F2FS_M_SB(mapping), 2669 FAULT_PAGE_ALLOC); 2670 return NULL; 2671 } 2672 } 2673 2674 if (!for_write) 2675 return grab_cache_page(mapping, index); 2676 return grab_cache_page_write_begin(mapping, index, AOP_FLAG_NOFS); 2677 } 2678 2679 static inline struct page *f2fs_pagecache_get_page( 2680 struct address_space *mapping, pgoff_t index, 2681 int fgp_flags, gfp_t gfp_mask) 2682 { 2683 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_GET)) { 2684 f2fs_show_injection_info(F2FS_M_SB(mapping), FAULT_PAGE_GET); 2685 return NULL; 2686 } 2687 2688 return pagecache_get_page(mapping, index, fgp_flags, gfp_mask); 2689 } 2690 2691 static inline void f2fs_copy_page(struct page *src, struct page *dst) 2692 { 2693 char *src_kaddr = kmap(src); 2694 char *dst_kaddr = kmap(dst); 2695 2696 memcpy(dst_kaddr, src_kaddr, PAGE_SIZE); 2697 kunmap(dst); 2698 kunmap(src); 2699 } 2700 2701 static inline void f2fs_put_page(struct page *page, int unlock) 2702 { 2703 if (!page) 2704 return; 2705 2706 if (unlock) { 2707 f2fs_bug_on(F2FS_P_SB(page), !PageLocked(page)); 2708 unlock_page(page); 2709 } 2710 put_page(page); 2711 } 2712 2713 static inline void f2fs_put_dnode(struct dnode_of_data *dn) 2714 { 2715 if (dn->node_page) 2716 f2fs_put_page(dn->node_page, 1); 2717 if (dn->inode_page && dn->node_page != dn->inode_page) 2718 f2fs_put_page(dn->inode_page, 0); 2719 dn->node_page = NULL; 2720 dn->inode_page = NULL; 2721 } 2722 2723 static inline struct kmem_cache *f2fs_kmem_cache_create(const char *name, 2724 size_t size) 2725 { 2726 return kmem_cache_create(name, size, 0, SLAB_RECLAIM_ACCOUNT, NULL); 2727 } 2728 2729 static inline void *f2fs_kmem_cache_alloc_nofail(struct kmem_cache *cachep, 2730 gfp_t flags) 2731 { 2732 void *entry; 2733 2734 entry = kmem_cache_alloc(cachep, flags); 2735 if (!entry) 2736 entry = kmem_cache_alloc(cachep, flags | __GFP_NOFAIL); 2737 return entry; 2738 } 2739 2740 static inline void *f2fs_kmem_cache_alloc(struct kmem_cache *cachep, 2741 gfp_t flags, bool nofail, struct f2fs_sb_info *sbi) 2742 { 2743 if (nofail) 2744 return f2fs_kmem_cache_alloc_nofail(cachep, flags); 2745 2746 if (time_to_inject(sbi, FAULT_SLAB_ALLOC)) { 2747 f2fs_show_injection_info(sbi, FAULT_SLAB_ALLOC); 2748 return NULL; 2749 } 2750 2751 return kmem_cache_alloc(cachep, flags); 2752 } 2753 2754 static inline bool is_inflight_io(struct f2fs_sb_info *sbi, int type) 2755 { 2756 if (get_pages(sbi, F2FS_RD_DATA) || get_pages(sbi, F2FS_RD_NODE) || 2757 get_pages(sbi, F2FS_RD_META) || get_pages(sbi, F2FS_WB_DATA) || 2758 get_pages(sbi, F2FS_WB_CP_DATA) || 2759 get_pages(sbi, F2FS_DIO_READ) || 2760 get_pages(sbi, F2FS_DIO_WRITE)) 2761 return true; 2762 2763 if (type != DISCARD_TIME && SM_I(sbi) && SM_I(sbi)->dcc_info && 2764 atomic_read(&SM_I(sbi)->dcc_info->queued_discard)) 2765 return true; 2766 2767 if (SM_I(sbi) && SM_I(sbi)->fcc_info && 2768 atomic_read(&SM_I(sbi)->fcc_info->queued_flush)) 2769 return true; 2770 return false; 2771 } 2772 2773 static inline bool is_idle(struct f2fs_sb_info *sbi, int type) 2774 { 2775 if (sbi->gc_mode == GC_URGENT_HIGH) 2776 return true; 2777 2778 if (is_inflight_io(sbi, type)) 2779 return false; 2780 2781 if (sbi->gc_mode == GC_URGENT_MID) 2782 return true; 2783 2784 if (sbi->gc_mode == GC_URGENT_LOW && 2785 (type == DISCARD_TIME || type == GC_TIME)) 2786 return true; 2787 2788 return f2fs_time_over(sbi, type); 2789 } 2790 2791 static inline void f2fs_radix_tree_insert(struct radix_tree_root *root, 2792 unsigned long index, void *item) 2793 { 2794 while (radix_tree_insert(root, index, item)) 2795 cond_resched(); 2796 } 2797 2798 #define RAW_IS_INODE(p) ((p)->footer.nid == (p)->footer.ino) 2799 2800 static inline bool IS_INODE(struct page *page) 2801 { 2802 struct f2fs_node *p = F2FS_NODE(page); 2803 2804 return RAW_IS_INODE(p); 2805 } 2806 2807 static inline int offset_in_addr(struct f2fs_inode *i) 2808 { 2809 return (i->i_inline & F2FS_EXTRA_ATTR) ? 2810 (le16_to_cpu(i->i_extra_isize) / sizeof(__le32)) : 0; 2811 } 2812 2813 static inline __le32 *blkaddr_in_node(struct f2fs_node *node) 2814 { 2815 return RAW_IS_INODE(node) ? node->i.i_addr : node->dn.addr; 2816 } 2817 2818 static inline int f2fs_has_extra_attr(struct inode *inode); 2819 static inline block_t data_blkaddr(struct inode *inode, 2820 struct page *node_page, unsigned int offset) 2821 { 2822 struct f2fs_node *raw_node; 2823 __le32 *addr_array; 2824 int base = 0; 2825 bool is_inode = IS_INODE(node_page); 2826 2827 raw_node = F2FS_NODE(node_page); 2828 2829 if (is_inode) { 2830 if (!inode) 2831 /* from GC path only */ 2832 base = offset_in_addr(&raw_node->i); 2833 else if (f2fs_has_extra_attr(inode)) 2834 base = get_extra_isize(inode); 2835 } 2836 2837 addr_array = blkaddr_in_node(raw_node); 2838 return le32_to_cpu(addr_array[base + offset]); 2839 } 2840 2841 static inline block_t f2fs_data_blkaddr(struct dnode_of_data *dn) 2842 { 2843 return data_blkaddr(dn->inode, dn->node_page, dn->ofs_in_node); 2844 } 2845 2846 static inline int f2fs_test_bit(unsigned int nr, char *addr) 2847 { 2848 int mask; 2849 2850 addr += (nr >> 3); 2851 mask = 1 << (7 - (nr & 0x07)); 2852 return mask & *addr; 2853 } 2854 2855 static inline void f2fs_set_bit(unsigned int nr, char *addr) 2856 { 2857 int mask; 2858 2859 addr += (nr >> 3); 2860 mask = 1 << (7 - (nr & 0x07)); 2861 *addr |= mask; 2862 } 2863 2864 static inline void f2fs_clear_bit(unsigned int nr, char *addr) 2865 { 2866 int mask; 2867 2868 addr += (nr >> 3); 2869 mask = 1 << (7 - (nr & 0x07)); 2870 *addr &= ~mask; 2871 } 2872 2873 static inline int f2fs_test_and_set_bit(unsigned int nr, char *addr) 2874 { 2875 int mask; 2876 int ret; 2877 2878 addr += (nr >> 3); 2879 mask = 1 << (7 - (nr & 0x07)); 2880 ret = mask & *addr; 2881 *addr |= mask; 2882 return ret; 2883 } 2884 2885 static inline int f2fs_test_and_clear_bit(unsigned int nr, char *addr) 2886 { 2887 int mask; 2888 int ret; 2889 2890 addr += (nr >> 3); 2891 mask = 1 << (7 - (nr & 0x07)); 2892 ret = mask & *addr; 2893 *addr &= ~mask; 2894 return ret; 2895 } 2896 2897 static inline void f2fs_change_bit(unsigned int nr, char *addr) 2898 { 2899 int mask; 2900 2901 addr += (nr >> 3); 2902 mask = 1 << (7 - (nr & 0x07)); 2903 *addr ^= mask; 2904 } 2905 2906 /* 2907 * On-disk inode flags (f2fs_inode::i_flags) 2908 */ 2909 #define F2FS_COMPR_FL 0x00000004 /* Compress file */ 2910 #define F2FS_SYNC_FL 0x00000008 /* Synchronous updates */ 2911 #define F2FS_IMMUTABLE_FL 0x00000010 /* Immutable file */ 2912 #define F2FS_APPEND_FL 0x00000020 /* writes to file may only append */ 2913 #define F2FS_NODUMP_FL 0x00000040 /* do not dump file */ 2914 #define F2FS_NOATIME_FL 0x00000080 /* do not update atime */ 2915 #define F2FS_NOCOMP_FL 0x00000400 /* Don't compress */ 2916 #define F2FS_INDEX_FL 0x00001000 /* hash-indexed directory */ 2917 #define F2FS_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */ 2918 #define F2FS_PROJINHERIT_FL 0x20000000 /* Create with parents projid */ 2919 #define F2FS_CASEFOLD_FL 0x40000000 /* Casefolded file */ 2920 2921 /* Flags that should be inherited by new inodes from their parent. */ 2922 #define F2FS_FL_INHERITED (F2FS_SYNC_FL | F2FS_NODUMP_FL | F2FS_NOATIME_FL | \ 2923 F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \ 2924 F2FS_CASEFOLD_FL | F2FS_COMPR_FL | F2FS_NOCOMP_FL) 2925 2926 /* Flags that are appropriate for regular files (all but dir-specific ones). */ 2927 #define F2FS_REG_FLMASK (~(F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \ 2928 F2FS_CASEFOLD_FL)) 2929 2930 /* Flags that are appropriate for non-directories/regular files. */ 2931 #define F2FS_OTHER_FLMASK (F2FS_NODUMP_FL | F2FS_NOATIME_FL) 2932 2933 static inline __u32 f2fs_mask_flags(umode_t mode, __u32 flags) 2934 { 2935 if (S_ISDIR(mode)) 2936 return flags; 2937 else if (S_ISREG(mode)) 2938 return flags & F2FS_REG_FLMASK; 2939 else 2940 return flags & F2FS_OTHER_FLMASK; 2941 } 2942 2943 static inline void __mark_inode_dirty_flag(struct inode *inode, 2944 int flag, bool set) 2945 { 2946 switch (flag) { 2947 case FI_INLINE_XATTR: 2948 case FI_INLINE_DATA: 2949 case FI_INLINE_DENTRY: 2950 case FI_NEW_INODE: 2951 if (set) 2952 return; 2953 fallthrough; 2954 case FI_DATA_EXIST: 2955 case FI_INLINE_DOTS: 2956 case FI_PIN_FILE: 2957 case FI_COMPRESS_RELEASED: 2958 f2fs_mark_inode_dirty_sync(inode, true); 2959 } 2960 } 2961 2962 static inline void set_inode_flag(struct inode *inode, int flag) 2963 { 2964 set_bit(flag, F2FS_I(inode)->flags); 2965 __mark_inode_dirty_flag(inode, flag, true); 2966 } 2967 2968 static inline int is_inode_flag_set(struct inode *inode, int flag) 2969 { 2970 return test_bit(flag, F2FS_I(inode)->flags); 2971 } 2972 2973 static inline void clear_inode_flag(struct inode *inode, int flag) 2974 { 2975 clear_bit(flag, F2FS_I(inode)->flags); 2976 __mark_inode_dirty_flag(inode, flag, false); 2977 } 2978 2979 static inline bool f2fs_verity_in_progress(struct inode *inode) 2980 { 2981 return IS_ENABLED(CONFIG_FS_VERITY) && 2982 is_inode_flag_set(inode, FI_VERITY_IN_PROGRESS); 2983 } 2984 2985 static inline void set_acl_inode(struct inode *inode, umode_t mode) 2986 { 2987 F2FS_I(inode)->i_acl_mode = mode; 2988 set_inode_flag(inode, FI_ACL_MODE); 2989 f2fs_mark_inode_dirty_sync(inode, false); 2990 } 2991 2992 static inline void f2fs_i_links_write(struct inode *inode, bool inc) 2993 { 2994 if (inc) 2995 inc_nlink(inode); 2996 else 2997 drop_nlink(inode); 2998 f2fs_mark_inode_dirty_sync(inode, true); 2999 } 3000 3001 static inline void f2fs_i_blocks_write(struct inode *inode, 3002 block_t diff, bool add, bool claim) 3003 { 3004 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE); 3005 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER); 3006 3007 /* add = 1, claim = 1 should be dquot_reserve_block in pair */ 3008 if (add) { 3009 if (claim) 3010 dquot_claim_block(inode, diff); 3011 else 3012 dquot_alloc_block_nofail(inode, diff); 3013 } else { 3014 dquot_free_block(inode, diff); 3015 } 3016 3017 f2fs_mark_inode_dirty_sync(inode, true); 3018 if (clean || recover) 3019 set_inode_flag(inode, FI_AUTO_RECOVER); 3020 } 3021 3022 static inline void f2fs_i_size_write(struct inode *inode, loff_t i_size) 3023 { 3024 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE); 3025 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER); 3026 3027 if (i_size_read(inode) == i_size) 3028 return; 3029 3030 i_size_write(inode, i_size); 3031 f2fs_mark_inode_dirty_sync(inode, true); 3032 if (clean || recover) 3033 set_inode_flag(inode, FI_AUTO_RECOVER); 3034 } 3035 3036 static inline void f2fs_i_depth_write(struct inode *inode, unsigned int depth) 3037 { 3038 F2FS_I(inode)->i_current_depth = depth; 3039 f2fs_mark_inode_dirty_sync(inode, true); 3040 } 3041 3042 static inline void f2fs_i_gc_failures_write(struct inode *inode, 3043 unsigned int count) 3044 { 3045 F2FS_I(inode)->i_gc_failures[GC_FAILURE_PIN] = count; 3046 f2fs_mark_inode_dirty_sync(inode, true); 3047 } 3048 3049 static inline void f2fs_i_xnid_write(struct inode *inode, nid_t xnid) 3050 { 3051 F2FS_I(inode)->i_xattr_nid = xnid; 3052 f2fs_mark_inode_dirty_sync(inode, true); 3053 } 3054 3055 static inline void f2fs_i_pino_write(struct inode *inode, nid_t pino) 3056 { 3057 F2FS_I(inode)->i_pino = pino; 3058 f2fs_mark_inode_dirty_sync(inode, true); 3059 } 3060 3061 static inline void get_inline_info(struct inode *inode, struct f2fs_inode *ri) 3062 { 3063 struct f2fs_inode_info *fi = F2FS_I(inode); 3064 3065 if (ri->i_inline & F2FS_INLINE_XATTR) 3066 set_bit(FI_INLINE_XATTR, fi->flags); 3067 if (ri->i_inline & F2FS_INLINE_DATA) 3068 set_bit(FI_INLINE_DATA, fi->flags); 3069 if (ri->i_inline & F2FS_INLINE_DENTRY) 3070 set_bit(FI_INLINE_DENTRY, fi->flags); 3071 if (ri->i_inline & F2FS_DATA_EXIST) 3072 set_bit(FI_DATA_EXIST, fi->flags); 3073 if (ri->i_inline & F2FS_INLINE_DOTS) 3074 set_bit(FI_INLINE_DOTS, fi->flags); 3075 if (ri->i_inline & F2FS_EXTRA_ATTR) 3076 set_bit(FI_EXTRA_ATTR, fi->flags); 3077 if (ri->i_inline & F2FS_PIN_FILE) 3078 set_bit(FI_PIN_FILE, fi->flags); 3079 if (ri->i_inline & F2FS_COMPRESS_RELEASED) 3080 set_bit(FI_COMPRESS_RELEASED, fi->flags); 3081 } 3082 3083 static inline void set_raw_inline(struct inode *inode, struct f2fs_inode *ri) 3084 { 3085 ri->i_inline = 0; 3086 3087 if (is_inode_flag_set(inode, FI_INLINE_XATTR)) 3088 ri->i_inline |= F2FS_INLINE_XATTR; 3089 if (is_inode_flag_set(inode, FI_INLINE_DATA)) 3090 ri->i_inline |= F2FS_INLINE_DATA; 3091 if (is_inode_flag_set(inode, FI_INLINE_DENTRY)) 3092 ri->i_inline |= F2FS_INLINE_DENTRY; 3093 if (is_inode_flag_set(inode, FI_DATA_EXIST)) 3094 ri->i_inline |= F2FS_DATA_EXIST; 3095 if (is_inode_flag_set(inode, FI_INLINE_DOTS)) 3096 ri->i_inline |= F2FS_INLINE_DOTS; 3097 if (is_inode_flag_set(inode, FI_EXTRA_ATTR)) 3098 ri->i_inline |= F2FS_EXTRA_ATTR; 3099 if (is_inode_flag_set(inode, FI_PIN_FILE)) 3100 ri->i_inline |= F2FS_PIN_FILE; 3101 if (is_inode_flag_set(inode, FI_COMPRESS_RELEASED)) 3102 ri->i_inline |= F2FS_COMPRESS_RELEASED; 3103 } 3104 3105 static inline int f2fs_has_extra_attr(struct inode *inode) 3106 { 3107 return is_inode_flag_set(inode, FI_EXTRA_ATTR); 3108 } 3109 3110 static inline int f2fs_has_inline_xattr(struct inode *inode) 3111 { 3112 return is_inode_flag_set(inode, FI_INLINE_XATTR); 3113 } 3114 3115 static inline int f2fs_compressed_file(struct inode *inode) 3116 { 3117 return S_ISREG(inode->i_mode) && 3118 is_inode_flag_set(inode, FI_COMPRESSED_FILE); 3119 } 3120 3121 static inline bool f2fs_need_compress_data(struct inode *inode) 3122 { 3123 int compress_mode = F2FS_OPTION(F2FS_I_SB(inode)).compress_mode; 3124 3125 if (!f2fs_compressed_file(inode)) 3126 return false; 3127 3128 if (compress_mode == COMPR_MODE_FS) 3129 return true; 3130 else if (compress_mode == COMPR_MODE_USER && 3131 is_inode_flag_set(inode, FI_ENABLE_COMPRESS)) 3132 return true; 3133 3134 return false; 3135 } 3136 3137 static inline unsigned int addrs_per_inode(struct inode *inode) 3138 { 3139 unsigned int addrs = CUR_ADDRS_PER_INODE(inode) - 3140 get_inline_xattr_addrs(inode); 3141 3142 if (!f2fs_compressed_file(inode)) 3143 return addrs; 3144 return ALIGN_DOWN(addrs, F2FS_I(inode)->i_cluster_size); 3145 } 3146 3147 static inline unsigned int addrs_per_block(struct inode *inode) 3148 { 3149 if (!f2fs_compressed_file(inode)) 3150 return DEF_ADDRS_PER_BLOCK; 3151 return ALIGN_DOWN(DEF_ADDRS_PER_BLOCK, F2FS_I(inode)->i_cluster_size); 3152 } 3153 3154 static inline void *inline_xattr_addr(struct inode *inode, struct page *page) 3155 { 3156 struct f2fs_inode *ri = F2FS_INODE(page); 3157 3158 return (void *)&(ri->i_addr[DEF_ADDRS_PER_INODE - 3159 get_inline_xattr_addrs(inode)]); 3160 } 3161 3162 static inline int inline_xattr_size(struct inode *inode) 3163 { 3164 if (f2fs_has_inline_xattr(inode)) 3165 return get_inline_xattr_addrs(inode) * sizeof(__le32); 3166 return 0; 3167 } 3168 3169 static inline int f2fs_has_inline_data(struct inode *inode) 3170 { 3171 return is_inode_flag_set(inode, FI_INLINE_DATA); 3172 } 3173 3174 static inline int f2fs_exist_data(struct inode *inode) 3175 { 3176 return is_inode_flag_set(inode, FI_DATA_EXIST); 3177 } 3178 3179 static inline int f2fs_has_inline_dots(struct inode *inode) 3180 { 3181 return is_inode_flag_set(inode, FI_INLINE_DOTS); 3182 } 3183 3184 static inline int f2fs_is_mmap_file(struct inode *inode) 3185 { 3186 return is_inode_flag_set(inode, FI_MMAP_FILE); 3187 } 3188 3189 static inline bool f2fs_is_pinned_file(struct inode *inode) 3190 { 3191 return is_inode_flag_set(inode, FI_PIN_FILE); 3192 } 3193 3194 static inline bool f2fs_is_atomic_file(struct inode *inode) 3195 { 3196 return is_inode_flag_set(inode, FI_ATOMIC_FILE); 3197 } 3198 3199 static inline bool f2fs_is_commit_atomic_write(struct inode *inode) 3200 { 3201 return is_inode_flag_set(inode, FI_ATOMIC_COMMIT); 3202 } 3203 3204 static inline bool f2fs_is_volatile_file(struct inode *inode) 3205 { 3206 return is_inode_flag_set(inode, FI_VOLATILE_FILE); 3207 } 3208 3209 static inline bool f2fs_is_first_block_written(struct inode *inode) 3210 { 3211 return is_inode_flag_set(inode, FI_FIRST_BLOCK_WRITTEN); 3212 } 3213 3214 static inline bool f2fs_is_drop_cache(struct inode *inode) 3215 { 3216 return is_inode_flag_set(inode, FI_DROP_CACHE); 3217 } 3218 3219 static inline void *inline_data_addr(struct inode *inode, struct page *page) 3220 { 3221 struct f2fs_inode *ri = F2FS_INODE(page); 3222 int extra_size = get_extra_isize(inode); 3223 3224 return (void *)&(ri->i_addr[extra_size + DEF_INLINE_RESERVED_SIZE]); 3225 } 3226 3227 static inline int f2fs_has_inline_dentry(struct inode *inode) 3228 { 3229 return is_inode_flag_set(inode, FI_INLINE_DENTRY); 3230 } 3231 3232 static inline int is_file(struct inode *inode, int type) 3233 { 3234 return F2FS_I(inode)->i_advise & type; 3235 } 3236 3237 static inline void set_file(struct inode *inode, int type) 3238 { 3239 if (is_file(inode, type)) 3240 return; 3241 F2FS_I(inode)->i_advise |= type; 3242 f2fs_mark_inode_dirty_sync(inode, true); 3243 } 3244 3245 static inline void clear_file(struct inode *inode, int type) 3246 { 3247 if (!is_file(inode, type)) 3248 return; 3249 F2FS_I(inode)->i_advise &= ~type; 3250 f2fs_mark_inode_dirty_sync(inode, true); 3251 } 3252 3253 static inline bool f2fs_is_time_consistent(struct inode *inode) 3254 { 3255 if (!timespec64_equal(F2FS_I(inode)->i_disk_time, &inode->i_atime)) 3256 return false; 3257 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 1, &inode->i_ctime)) 3258 return false; 3259 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 2, &inode->i_mtime)) 3260 return false; 3261 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 3, 3262 &F2FS_I(inode)->i_crtime)) 3263 return false; 3264 return true; 3265 } 3266 3267 static inline bool f2fs_skip_inode_update(struct inode *inode, int dsync) 3268 { 3269 bool ret; 3270 3271 if (dsync) { 3272 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 3273 3274 spin_lock(&sbi->inode_lock[DIRTY_META]); 3275 ret = list_empty(&F2FS_I(inode)->gdirty_list); 3276 spin_unlock(&sbi->inode_lock[DIRTY_META]); 3277 return ret; 3278 } 3279 if (!is_inode_flag_set(inode, FI_AUTO_RECOVER) || 3280 file_keep_isize(inode) || 3281 i_size_read(inode) & ~PAGE_MASK) 3282 return false; 3283 3284 if (!f2fs_is_time_consistent(inode)) 3285 return false; 3286 3287 spin_lock(&F2FS_I(inode)->i_size_lock); 3288 ret = F2FS_I(inode)->last_disk_size == i_size_read(inode); 3289 spin_unlock(&F2FS_I(inode)->i_size_lock); 3290 3291 return ret; 3292 } 3293 3294 static inline bool f2fs_readonly(struct super_block *sb) 3295 { 3296 return sb_rdonly(sb); 3297 } 3298 3299 static inline bool f2fs_cp_error(struct f2fs_sb_info *sbi) 3300 { 3301 return is_set_ckpt_flags(sbi, CP_ERROR_FLAG); 3302 } 3303 3304 static inline bool is_dot_dotdot(const u8 *name, size_t len) 3305 { 3306 if (len == 1 && name[0] == '.') 3307 return true; 3308 3309 if (len == 2 && name[0] == '.' && name[1] == '.') 3310 return true; 3311 3312 return false; 3313 } 3314 3315 static inline void *f2fs_kmalloc(struct f2fs_sb_info *sbi, 3316 size_t size, gfp_t flags) 3317 { 3318 if (time_to_inject(sbi, FAULT_KMALLOC)) { 3319 f2fs_show_injection_info(sbi, FAULT_KMALLOC); 3320 return NULL; 3321 } 3322 3323 return kmalloc(size, flags); 3324 } 3325 3326 static inline void *f2fs_kzalloc(struct f2fs_sb_info *sbi, 3327 size_t size, gfp_t flags) 3328 { 3329 return f2fs_kmalloc(sbi, size, flags | __GFP_ZERO); 3330 } 3331 3332 static inline void *f2fs_kvmalloc(struct f2fs_sb_info *sbi, 3333 size_t size, gfp_t flags) 3334 { 3335 if (time_to_inject(sbi, FAULT_KVMALLOC)) { 3336 f2fs_show_injection_info(sbi, FAULT_KVMALLOC); 3337 return NULL; 3338 } 3339 3340 return kvmalloc(size, flags); 3341 } 3342 3343 static inline void *f2fs_kvzalloc(struct f2fs_sb_info *sbi, 3344 size_t size, gfp_t flags) 3345 { 3346 return f2fs_kvmalloc(sbi, size, flags | __GFP_ZERO); 3347 } 3348 3349 static inline int get_extra_isize(struct inode *inode) 3350 { 3351 return F2FS_I(inode)->i_extra_isize / sizeof(__le32); 3352 } 3353 3354 static inline int get_inline_xattr_addrs(struct inode *inode) 3355 { 3356 return F2FS_I(inode)->i_inline_xattr_size; 3357 } 3358 3359 #define f2fs_get_inode_mode(i) \ 3360 ((is_inode_flag_set(i, FI_ACL_MODE)) ? \ 3361 (F2FS_I(i)->i_acl_mode) : ((i)->i_mode)) 3362 3363 #define F2FS_TOTAL_EXTRA_ATTR_SIZE \ 3364 (offsetof(struct f2fs_inode, i_extra_end) - \ 3365 offsetof(struct f2fs_inode, i_extra_isize)) \ 3366 3367 #define F2FS_OLD_ATTRIBUTE_SIZE (offsetof(struct f2fs_inode, i_addr)) 3368 #define F2FS_FITS_IN_INODE(f2fs_inode, extra_isize, field) \ 3369 ((offsetof(typeof(*(f2fs_inode)), field) + \ 3370 sizeof((f2fs_inode)->field)) \ 3371 <= (F2FS_OLD_ATTRIBUTE_SIZE + (extra_isize))) \ 3372 3373 #define __is_large_section(sbi) ((sbi)->segs_per_sec > 1) 3374 3375 #define __is_meta_io(fio) (PAGE_TYPE_OF_BIO((fio)->type) == META) 3376 3377 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi, 3378 block_t blkaddr, int type); 3379 static inline void verify_blkaddr(struct f2fs_sb_info *sbi, 3380 block_t blkaddr, int type) 3381 { 3382 if (!f2fs_is_valid_blkaddr(sbi, blkaddr, type)) { 3383 f2fs_err(sbi, "invalid blkaddr: %u, type: %d, run fsck to fix.", 3384 blkaddr, type); 3385 f2fs_bug_on(sbi, 1); 3386 } 3387 } 3388 3389 static inline bool __is_valid_data_blkaddr(block_t blkaddr) 3390 { 3391 if (blkaddr == NEW_ADDR || blkaddr == NULL_ADDR || 3392 blkaddr == COMPRESS_ADDR) 3393 return false; 3394 return true; 3395 } 3396 3397 /* 3398 * file.c 3399 */ 3400 int f2fs_sync_file(struct file *file, loff_t start, loff_t end, int datasync); 3401 void f2fs_truncate_data_blocks(struct dnode_of_data *dn); 3402 int f2fs_do_truncate_blocks(struct inode *inode, u64 from, bool lock); 3403 int f2fs_truncate_blocks(struct inode *inode, u64 from, bool lock); 3404 int f2fs_truncate(struct inode *inode); 3405 int f2fs_getattr(struct user_namespace *mnt_userns, const struct path *path, 3406 struct kstat *stat, u32 request_mask, unsigned int flags); 3407 int f2fs_setattr(struct user_namespace *mnt_userns, struct dentry *dentry, 3408 struct iattr *attr); 3409 int f2fs_truncate_hole(struct inode *inode, pgoff_t pg_start, pgoff_t pg_end); 3410 void f2fs_truncate_data_blocks_range(struct dnode_of_data *dn, int count); 3411 int f2fs_precache_extents(struct inode *inode); 3412 int f2fs_fileattr_get(struct dentry *dentry, struct fileattr *fa); 3413 int f2fs_fileattr_set(struct user_namespace *mnt_userns, 3414 struct dentry *dentry, struct fileattr *fa); 3415 long f2fs_ioctl(struct file *filp, unsigned int cmd, unsigned long arg); 3416 long f2fs_compat_ioctl(struct file *file, unsigned int cmd, unsigned long arg); 3417 int f2fs_transfer_project_quota(struct inode *inode, kprojid_t kprojid); 3418 int f2fs_pin_file_control(struct inode *inode, bool inc); 3419 3420 /* 3421 * inode.c 3422 */ 3423 void f2fs_set_inode_flags(struct inode *inode); 3424 bool f2fs_inode_chksum_verify(struct f2fs_sb_info *sbi, struct page *page); 3425 void f2fs_inode_chksum_set(struct f2fs_sb_info *sbi, struct page *page); 3426 struct inode *f2fs_iget(struct super_block *sb, unsigned long ino); 3427 struct inode *f2fs_iget_retry(struct super_block *sb, unsigned long ino); 3428 int f2fs_try_to_free_nats(struct f2fs_sb_info *sbi, int nr_shrink); 3429 void f2fs_update_inode(struct inode *inode, struct page *node_page); 3430 void f2fs_update_inode_page(struct inode *inode); 3431 int f2fs_write_inode(struct inode *inode, struct writeback_control *wbc); 3432 void f2fs_evict_inode(struct inode *inode); 3433 void f2fs_handle_failed_inode(struct inode *inode); 3434 3435 /* 3436 * namei.c 3437 */ 3438 int f2fs_update_extension_list(struct f2fs_sb_info *sbi, const char *name, 3439 bool hot, bool set); 3440 struct dentry *f2fs_get_parent(struct dentry *child); 3441 3442 /* 3443 * dir.c 3444 */ 3445 unsigned char f2fs_get_de_type(struct f2fs_dir_entry *de); 3446 int f2fs_init_casefolded_name(const struct inode *dir, 3447 struct f2fs_filename *fname); 3448 int f2fs_setup_filename(struct inode *dir, const struct qstr *iname, 3449 int lookup, struct f2fs_filename *fname); 3450 int f2fs_prepare_lookup(struct inode *dir, struct dentry *dentry, 3451 struct f2fs_filename *fname); 3452 void f2fs_free_filename(struct f2fs_filename *fname); 3453 struct f2fs_dir_entry *f2fs_find_target_dentry(const struct f2fs_dentry_ptr *d, 3454 const struct f2fs_filename *fname, int *max_slots); 3455 int f2fs_fill_dentries(struct dir_context *ctx, struct f2fs_dentry_ptr *d, 3456 unsigned int start_pos, struct fscrypt_str *fstr); 3457 void f2fs_do_make_empty_dir(struct inode *inode, struct inode *parent, 3458 struct f2fs_dentry_ptr *d); 3459 struct page *f2fs_init_inode_metadata(struct inode *inode, struct inode *dir, 3460 const struct f2fs_filename *fname, struct page *dpage); 3461 void f2fs_update_parent_metadata(struct inode *dir, struct inode *inode, 3462 unsigned int current_depth); 3463 int f2fs_room_for_filename(const void *bitmap, int slots, int max_slots); 3464 void f2fs_drop_nlink(struct inode *dir, struct inode *inode); 3465 struct f2fs_dir_entry *__f2fs_find_entry(struct inode *dir, 3466 const struct f2fs_filename *fname, 3467 struct page **res_page); 3468 struct f2fs_dir_entry *f2fs_find_entry(struct inode *dir, 3469 const struct qstr *child, struct page **res_page); 3470 struct f2fs_dir_entry *f2fs_parent_dir(struct inode *dir, struct page **p); 3471 ino_t f2fs_inode_by_name(struct inode *dir, const struct qstr *qstr, 3472 struct page **page); 3473 void f2fs_set_link(struct inode *dir, struct f2fs_dir_entry *de, 3474 struct page *page, struct inode *inode); 3475 bool f2fs_has_enough_room(struct inode *dir, struct page *ipage, 3476 const struct f2fs_filename *fname); 3477 void f2fs_update_dentry(nid_t ino, umode_t mode, struct f2fs_dentry_ptr *d, 3478 const struct fscrypt_str *name, f2fs_hash_t name_hash, 3479 unsigned int bit_pos); 3480 int f2fs_add_regular_entry(struct inode *dir, const struct f2fs_filename *fname, 3481 struct inode *inode, nid_t ino, umode_t mode); 3482 int f2fs_add_dentry(struct inode *dir, const struct f2fs_filename *fname, 3483 struct inode *inode, nid_t ino, umode_t mode); 3484 int f2fs_do_add_link(struct inode *dir, const struct qstr *name, 3485 struct inode *inode, nid_t ino, umode_t mode); 3486 void f2fs_delete_entry(struct f2fs_dir_entry *dentry, struct page *page, 3487 struct inode *dir, struct inode *inode); 3488 int f2fs_do_tmpfile(struct inode *inode, struct inode *dir); 3489 bool f2fs_empty_dir(struct inode *dir); 3490 3491 static inline int f2fs_add_link(struct dentry *dentry, struct inode *inode) 3492 { 3493 if (fscrypt_is_nokey_name(dentry)) 3494 return -ENOKEY; 3495 return f2fs_do_add_link(d_inode(dentry->d_parent), &dentry->d_name, 3496 inode, inode->i_ino, inode->i_mode); 3497 } 3498 3499 /* 3500 * super.c 3501 */ 3502 int f2fs_inode_dirtied(struct inode *inode, bool sync); 3503 void f2fs_inode_synced(struct inode *inode); 3504 int f2fs_dquot_initialize(struct inode *inode); 3505 int f2fs_enable_quota_files(struct f2fs_sb_info *sbi, bool rdonly); 3506 int f2fs_quota_sync(struct super_block *sb, int type); 3507 loff_t max_file_blocks(struct inode *inode); 3508 void f2fs_quota_off_umount(struct super_block *sb); 3509 int f2fs_commit_super(struct f2fs_sb_info *sbi, bool recover); 3510 int f2fs_sync_fs(struct super_block *sb, int sync); 3511 int f2fs_sanity_check_ckpt(struct f2fs_sb_info *sbi); 3512 3513 /* 3514 * hash.c 3515 */ 3516 void f2fs_hash_filename(const struct inode *dir, struct f2fs_filename *fname); 3517 3518 /* 3519 * node.c 3520 */ 3521 struct node_info; 3522 3523 int f2fs_check_nid_range(struct f2fs_sb_info *sbi, nid_t nid); 3524 bool f2fs_available_free_memory(struct f2fs_sb_info *sbi, int type); 3525 bool f2fs_in_warm_node_list(struct f2fs_sb_info *sbi, struct page *page); 3526 void f2fs_init_fsync_node_info(struct f2fs_sb_info *sbi); 3527 void f2fs_del_fsync_node_entry(struct f2fs_sb_info *sbi, struct page *page); 3528 void f2fs_reset_fsync_node_info(struct f2fs_sb_info *sbi); 3529 int f2fs_need_dentry_mark(struct f2fs_sb_info *sbi, nid_t nid); 3530 bool f2fs_is_checkpointed_node(struct f2fs_sb_info *sbi, nid_t nid); 3531 bool f2fs_need_inode_block_update(struct f2fs_sb_info *sbi, nid_t ino); 3532 int f2fs_get_node_info(struct f2fs_sb_info *sbi, nid_t nid, 3533 struct node_info *ni, bool checkpoint_context); 3534 pgoff_t f2fs_get_next_page_offset(struct dnode_of_data *dn, pgoff_t pgofs); 3535 int f2fs_get_dnode_of_data(struct dnode_of_data *dn, pgoff_t index, int mode); 3536 int f2fs_truncate_inode_blocks(struct inode *inode, pgoff_t from); 3537 int f2fs_truncate_xattr_node(struct inode *inode); 3538 int f2fs_wait_on_node_pages_writeback(struct f2fs_sb_info *sbi, 3539 unsigned int seq_id); 3540 bool f2fs_nat_bitmap_enabled(struct f2fs_sb_info *sbi); 3541 int f2fs_remove_inode_page(struct inode *inode); 3542 struct page *f2fs_new_inode_page(struct inode *inode); 3543 struct page *f2fs_new_node_page(struct dnode_of_data *dn, unsigned int ofs); 3544 void f2fs_ra_node_page(struct f2fs_sb_info *sbi, nid_t nid); 3545 struct page *f2fs_get_node_page(struct f2fs_sb_info *sbi, pgoff_t nid); 3546 struct page *f2fs_get_node_page_ra(struct page *parent, int start); 3547 int f2fs_move_node_page(struct page *node_page, int gc_type); 3548 void f2fs_flush_inline_data(struct f2fs_sb_info *sbi); 3549 int f2fs_fsync_node_pages(struct f2fs_sb_info *sbi, struct inode *inode, 3550 struct writeback_control *wbc, bool atomic, 3551 unsigned int *seq_id); 3552 int f2fs_sync_node_pages(struct f2fs_sb_info *sbi, 3553 struct writeback_control *wbc, 3554 bool do_balance, enum iostat_type io_type); 3555 int f2fs_build_free_nids(struct f2fs_sb_info *sbi, bool sync, bool mount); 3556 bool f2fs_alloc_nid(struct f2fs_sb_info *sbi, nid_t *nid); 3557 void f2fs_alloc_nid_done(struct f2fs_sb_info *sbi, nid_t nid); 3558 void f2fs_alloc_nid_failed(struct f2fs_sb_info *sbi, nid_t nid); 3559 int f2fs_try_to_free_nids(struct f2fs_sb_info *sbi, int nr_shrink); 3560 int f2fs_recover_inline_xattr(struct inode *inode, struct page *page); 3561 int f2fs_recover_xattr_data(struct inode *inode, struct page *page); 3562 int f2fs_recover_inode_page(struct f2fs_sb_info *sbi, struct page *page); 3563 int f2fs_restore_node_summary(struct f2fs_sb_info *sbi, 3564 unsigned int segno, struct f2fs_summary_block *sum); 3565 void f2fs_enable_nat_bits(struct f2fs_sb_info *sbi); 3566 int f2fs_flush_nat_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3567 int f2fs_build_node_manager(struct f2fs_sb_info *sbi); 3568 void f2fs_destroy_node_manager(struct f2fs_sb_info *sbi); 3569 int __init f2fs_create_node_manager_caches(void); 3570 void f2fs_destroy_node_manager_caches(void); 3571 3572 /* 3573 * segment.c 3574 */ 3575 bool f2fs_need_SSR(struct f2fs_sb_info *sbi); 3576 void f2fs_register_inmem_page(struct inode *inode, struct page *page); 3577 void f2fs_drop_inmem_pages_all(struct f2fs_sb_info *sbi, bool gc_failure); 3578 void f2fs_drop_inmem_pages(struct inode *inode); 3579 void f2fs_drop_inmem_page(struct inode *inode, struct page *page); 3580 int f2fs_commit_inmem_pages(struct inode *inode); 3581 void f2fs_balance_fs(struct f2fs_sb_info *sbi, bool need); 3582 void f2fs_balance_fs_bg(struct f2fs_sb_info *sbi, bool from_bg); 3583 int f2fs_issue_flush(struct f2fs_sb_info *sbi, nid_t ino); 3584 int f2fs_create_flush_cmd_control(struct f2fs_sb_info *sbi); 3585 int f2fs_flush_device_cache(struct f2fs_sb_info *sbi); 3586 void f2fs_destroy_flush_cmd_control(struct f2fs_sb_info *sbi, bool free); 3587 void f2fs_invalidate_blocks(struct f2fs_sb_info *sbi, block_t addr); 3588 bool f2fs_is_checkpointed_data(struct f2fs_sb_info *sbi, block_t blkaddr); 3589 int f2fs_start_discard_thread(struct f2fs_sb_info *sbi); 3590 void f2fs_drop_discard_cmd(struct f2fs_sb_info *sbi); 3591 void f2fs_stop_discard_thread(struct f2fs_sb_info *sbi); 3592 bool f2fs_issue_discard_timeout(struct f2fs_sb_info *sbi); 3593 void f2fs_clear_prefree_segments(struct f2fs_sb_info *sbi, 3594 struct cp_control *cpc); 3595 void f2fs_dirty_to_prefree(struct f2fs_sb_info *sbi); 3596 block_t f2fs_get_unusable_blocks(struct f2fs_sb_info *sbi); 3597 int f2fs_disable_cp_again(struct f2fs_sb_info *sbi, block_t unusable); 3598 void f2fs_release_discard_addrs(struct f2fs_sb_info *sbi); 3599 int f2fs_npages_for_summary_flush(struct f2fs_sb_info *sbi, bool for_ra); 3600 bool f2fs_segment_has_free_slot(struct f2fs_sb_info *sbi, int segno); 3601 void f2fs_init_inmem_curseg(struct f2fs_sb_info *sbi); 3602 void f2fs_save_inmem_curseg(struct f2fs_sb_info *sbi); 3603 void f2fs_restore_inmem_curseg(struct f2fs_sb_info *sbi); 3604 void f2fs_get_new_segment(struct f2fs_sb_info *sbi, 3605 unsigned int *newseg, bool new_sec, int dir); 3606 void f2fs_allocate_segment_for_resize(struct f2fs_sb_info *sbi, int type, 3607 unsigned int start, unsigned int end); 3608 void f2fs_allocate_new_section(struct f2fs_sb_info *sbi, int type, bool force); 3609 void f2fs_allocate_new_segments(struct f2fs_sb_info *sbi); 3610 int f2fs_trim_fs(struct f2fs_sb_info *sbi, struct fstrim_range *range); 3611 bool f2fs_exist_trim_candidates(struct f2fs_sb_info *sbi, 3612 struct cp_control *cpc); 3613 struct page *f2fs_get_sum_page(struct f2fs_sb_info *sbi, unsigned int segno); 3614 void f2fs_update_meta_page(struct f2fs_sb_info *sbi, void *src, 3615 block_t blk_addr); 3616 void f2fs_do_write_meta_page(struct f2fs_sb_info *sbi, struct page *page, 3617 enum iostat_type io_type); 3618 void f2fs_do_write_node_page(unsigned int nid, struct f2fs_io_info *fio); 3619 void f2fs_outplace_write_data(struct dnode_of_data *dn, 3620 struct f2fs_io_info *fio); 3621 int f2fs_inplace_write_data(struct f2fs_io_info *fio); 3622 void f2fs_do_replace_block(struct f2fs_sb_info *sbi, struct f2fs_summary *sum, 3623 block_t old_blkaddr, block_t new_blkaddr, 3624 bool recover_curseg, bool recover_newaddr, 3625 bool from_gc); 3626 void f2fs_replace_block(struct f2fs_sb_info *sbi, struct dnode_of_data *dn, 3627 block_t old_addr, block_t new_addr, 3628 unsigned char version, bool recover_curseg, 3629 bool recover_newaddr); 3630 void f2fs_allocate_data_block(struct f2fs_sb_info *sbi, struct page *page, 3631 block_t old_blkaddr, block_t *new_blkaddr, 3632 struct f2fs_summary *sum, int type, 3633 struct f2fs_io_info *fio); 3634 void f2fs_update_device_state(struct f2fs_sb_info *sbi, nid_t ino, 3635 block_t blkaddr, unsigned int blkcnt); 3636 void f2fs_wait_on_page_writeback(struct page *page, 3637 enum page_type type, bool ordered, bool locked); 3638 void f2fs_wait_on_block_writeback(struct inode *inode, block_t blkaddr); 3639 void f2fs_wait_on_block_writeback_range(struct inode *inode, block_t blkaddr, 3640 block_t len); 3641 void f2fs_write_data_summaries(struct f2fs_sb_info *sbi, block_t start_blk); 3642 void f2fs_write_node_summaries(struct f2fs_sb_info *sbi, block_t start_blk); 3643 int f2fs_lookup_journal_in_cursum(struct f2fs_journal *journal, int type, 3644 unsigned int val, int alloc); 3645 void f2fs_flush_sit_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3646 int f2fs_fix_curseg_write_pointer(struct f2fs_sb_info *sbi); 3647 int f2fs_check_write_pointer(struct f2fs_sb_info *sbi); 3648 int f2fs_build_segment_manager(struct f2fs_sb_info *sbi); 3649 void f2fs_destroy_segment_manager(struct f2fs_sb_info *sbi); 3650 int __init f2fs_create_segment_manager_caches(void); 3651 void f2fs_destroy_segment_manager_caches(void); 3652 int f2fs_rw_hint_to_seg_type(enum rw_hint hint); 3653 unsigned int f2fs_usable_segs_in_sec(struct f2fs_sb_info *sbi, 3654 unsigned int segno); 3655 unsigned int f2fs_usable_blks_in_seg(struct f2fs_sb_info *sbi, 3656 unsigned int segno); 3657 3658 #define DEF_FRAGMENT_SIZE 4 3659 #define MIN_FRAGMENT_SIZE 1 3660 #define MAX_FRAGMENT_SIZE 512 3661 3662 static inline bool f2fs_need_rand_seg(struct f2fs_sb_info *sbi) 3663 { 3664 return F2FS_OPTION(sbi).fs_mode == FS_MODE_FRAGMENT_SEG || 3665 F2FS_OPTION(sbi).fs_mode == FS_MODE_FRAGMENT_BLK; 3666 } 3667 3668 /* 3669 * checkpoint.c 3670 */ 3671 void f2fs_stop_checkpoint(struct f2fs_sb_info *sbi, bool end_io); 3672 struct page *f2fs_grab_meta_page(struct f2fs_sb_info *sbi, pgoff_t index); 3673 struct page *f2fs_get_meta_page(struct f2fs_sb_info *sbi, pgoff_t index); 3674 struct page *f2fs_get_meta_page_retry(struct f2fs_sb_info *sbi, pgoff_t index); 3675 struct page *f2fs_get_tmp_page(struct f2fs_sb_info *sbi, pgoff_t index); 3676 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi, 3677 block_t blkaddr, int type); 3678 int f2fs_ra_meta_pages(struct f2fs_sb_info *sbi, block_t start, int nrpages, 3679 int type, bool sync); 3680 void f2fs_ra_meta_pages_cond(struct f2fs_sb_info *sbi, pgoff_t index, 3681 unsigned int ra_blocks); 3682 long f2fs_sync_meta_pages(struct f2fs_sb_info *sbi, enum page_type type, 3683 long nr_to_write, enum iostat_type io_type); 3684 void f2fs_add_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type); 3685 void f2fs_remove_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type); 3686 void f2fs_release_ino_entry(struct f2fs_sb_info *sbi, bool all); 3687 bool f2fs_exist_written_data(struct f2fs_sb_info *sbi, nid_t ino, int mode); 3688 void f2fs_set_dirty_device(struct f2fs_sb_info *sbi, nid_t ino, 3689 unsigned int devidx, int type); 3690 bool f2fs_is_dirty_device(struct f2fs_sb_info *sbi, nid_t ino, 3691 unsigned int devidx, int type); 3692 int f2fs_sync_inode_meta(struct f2fs_sb_info *sbi); 3693 int f2fs_acquire_orphan_inode(struct f2fs_sb_info *sbi); 3694 void f2fs_release_orphan_inode(struct f2fs_sb_info *sbi); 3695 void f2fs_add_orphan_inode(struct inode *inode); 3696 void f2fs_remove_orphan_inode(struct f2fs_sb_info *sbi, nid_t ino); 3697 int f2fs_recover_orphan_inodes(struct f2fs_sb_info *sbi); 3698 int f2fs_get_valid_checkpoint(struct f2fs_sb_info *sbi); 3699 void f2fs_update_dirty_folio(struct inode *inode, struct folio *folio); 3700 void f2fs_remove_dirty_inode(struct inode *inode); 3701 int f2fs_sync_dirty_inodes(struct f2fs_sb_info *sbi, enum inode_type type); 3702 void f2fs_wait_on_all_pages(struct f2fs_sb_info *sbi, int type); 3703 u64 f2fs_get_sectors_written(struct f2fs_sb_info *sbi); 3704 int f2fs_write_checkpoint(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3705 void f2fs_init_ino_entry_info(struct f2fs_sb_info *sbi); 3706 int __init f2fs_create_checkpoint_caches(void); 3707 void f2fs_destroy_checkpoint_caches(void); 3708 int f2fs_issue_checkpoint(struct f2fs_sb_info *sbi); 3709 int f2fs_start_ckpt_thread(struct f2fs_sb_info *sbi); 3710 void f2fs_stop_ckpt_thread(struct f2fs_sb_info *sbi); 3711 void f2fs_init_ckpt_req_control(struct f2fs_sb_info *sbi); 3712 3713 /* 3714 * data.c 3715 */ 3716 int __init f2fs_init_bioset(void); 3717 void f2fs_destroy_bioset(void); 3718 int f2fs_init_bio_entry_cache(void); 3719 void f2fs_destroy_bio_entry_cache(void); 3720 void f2fs_submit_bio(struct f2fs_sb_info *sbi, 3721 struct bio *bio, enum page_type type); 3722 void f2fs_submit_merged_write(struct f2fs_sb_info *sbi, enum page_type type); 3723 void f2fs_submit_merged_write_cond(struct f2fs_sb_info *sbi, 3724 struct inode *inode, struct page *page, 3725 nid_t ino, enum page_type type); 3726 void f2fs_submit_merged_ipu_write(struct f2fs_sb_info *sbi, 3727 struct bio **bio, struct page *page); 3728 void f2fs_flush_merged_writes(struct f2fs_sb_info *sbi); 3729 int f2fs_submit_page_bio(struct f2fs_io_info *fio); 3730 int f2fs_merge_page_bio(struct f2fs_io_info *fio); 3731 void f2fs_submit_page_write(struct f2fs_io_info *fio); 3732 struct block_device *f2fs_target_device(struct f2fs_sb_info *sbi, 3733 block_t blk_addr, sector_t *sector); 3734 int f2fs_target_device_index(struct f2fs_sb_info *sbi, block_t blkaddr); 3735 void f2fs_set_data_blkaddr(struct dnode_of_data *dn); 3736 void f2fs_update_data_blkaddr(struct dnode_of_data *dn, block_t blkaddr); 3737 int f2fs_reserve_new_blocks(struct dnode_of_data *dn, blkcnt_t count); 3738 int f2fs_reserve_new_block(struct dnode_of_data *dn); 3739 int f2fs_get_block(struct dnode_of_data *dn, pgoff_t index); 3740 int f2fs_reserve_block(struct dnode_of_data *dn, pgoff_t index); 3741 struct page *f2fs_get_read_data_page(struct inode *inode, pgoff_t index, 3742 int op_flags, bool for_write); 3743 struct page *f2fs_find_data_page(struct inode *inode, pgoff_t index); 3744 struct page *f2fs_get_lock_data_page(struct inode *inode, pgoff_t index, 3745 bool for_write); 3746 struct page *f2fs_get_new_data_page(struct inode *inode, 3747 struct page *ipage, pgoff_t index, bool new_i_size); 3748 int f2fs_do_write_data_page(struct f2fs_io_info *fio); 3749 void f2fs_do_map_lock(struct f2fs_sb_info *sbi, int flag, bool lock); 3750 int f2fs_map_blocks(struct inode *inode, struct f2fs_map_blocks *map, 3751 int create, int flag); 3752 int f2fs_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, 3753 u64 start, u64 len); 3754 int f2fs_encrypt_one_page(struct f2fs_io_info *fio); 3755 bool f2fs_should_update_inplace(struct inode *inode, struct f2fs_io_info *fio); 3756 bool f2fs_should_update_outplace(struct inode *inode, struct f2fs_io_info *fio); 3757 int f2fs_write_single_data_page(struct page *page, int *submitted, 3758 struct bio **bio, sector_t *last_block, 3759 struct writeback_control *wbc, 3760 enum iostat_type io_type, 3761 int compr_blocks, bool allow_balance); 3762 void f2fs_write_failed(struct inode *inode, loff_t to); 3763 void f2fs_invalidate_folio(struct folio *folio, size_t offset, size_t length); 3764 int f2fs_release_page(struct page *page, gfp_t wait); 3765 #ifdef CONFIG_MIGRATION 3766 int f2fs_migrate_page(struct address_space *mapping, struct page *newpage, 3767 struct page *page, enum migrate_mode mode); 3768 #endif 3769 bool f2fs_overwrite_io(struct inode *inode, loff_t pos, size_t len); 3770 void f2fs_clear_page_cache_dirty_tag(struct page *page); 3771 int f2fs_init_post_read_processing(void); 3772 void f2fs_destroy_post_read_processing(void); 3773 int f2fs_init_post_read_wq(struct f2fs_sb_info *sbi); 3774 void f2fs_destroy_post_read_wq(struct f2fs_sb_info *sbi); 3775 extern const struct iomap_ops f2fs_iomap_ops; 3776 3777 /* 3778 * gc.c 3779 */ 3780 int f2fs_start_gc_thread(struct f2fs_sb_info *sbi); 3781 void f2fs_stop_gc_thread(struct f2fs_sb_info *sbi); 3782 block_t f2fs_start_bidx_of_node(unsigned int node_ofs, struct inode *inode); 3783 int f2fs_gc(struct f2fs_sb_info *sbi, bool sync, bool background, bool force, 3784 unsigned int segno); 3785 void f2fs_build_gc_manager(struct f2fs_sb_info *sbi); 3786 int f2fs_resize_fs(struct f2fs_sb_info *sbi, __u64 block_count); 3787 int __init f2fs_create_garbage_collection_cache(void); 3788 void f2fs_destroy_garbage_collection_cache(void); 3789 3790 /* 3791 * recovery.c 3792 */ 3793 int f2fs_recover_fsync_data(struct f2fs_sb_info *sbi, bool check_only); 3794 bool f2fs_space_for_roll_forward(struct f2fs_sb_info *sbi); 3795 int __init f2fs_create_recovery_cache(void); 3796 void f2fs_destroy_recovery_cache(void); 3797 3798 /* 3799 * debug.c 3800 */ 3801 #ifdef CONFIG_F2FS_STAT_FS 3802 struct f2fs_stat_info { 3803 struct list_head stat_list; 3804 struct f2fs_sb_info *sbi; 3805 int all_area_segs, sit_area_segs, nat_area_segs, ssa_area_segs; 3806 int main_area_segs, main_area_sections, main_area_zones; 3807 unsigned long long hit_largest, hit_cached, hit_rbtree; 3808 unsigned long long hit_total, total_ext; 3809 int ext_tree, zombie_tree, ext_node; 3810 int ndirty_node, ndirty_dent, ndirty_meta, ndirty_imeta; 3811 int ndirty_data, ndirty_qdata; 3812 int inmem_pages; 3813 unsigned int ndirty_dirs, ndirty_files, nquota_files, ndirty_all; 3814 int nats, dirty_nats, sits, dirty_sits; 3815 int free_nids, avail_nids, alloc_nids; 3816 int total_count, utilization; 3817 int bg_gc, nr_wb_cp_data, nr_wb_data; 3818 int nr_rd_data, nr_rd_node, nr_rd_meta; 3819 int nr_dio_read, nr_dio_write; 3820 unsigned int io_skip_bggc, other_skip_bggc; 3821 int nr_flushing, nr_flushed, flush_list_empty; 3822 int nr_discarding, nr_discarded; 3823 int nr_discard_cmd; 3824 unsigned int undiscard_blks; 3825 int nr_issued_ckpt, nr_total_ckpt, nr_queued_ckpt; 3826 unsigned int cur_ckpt_time, peak_ckpt_time; 3827 int inline_xattr, inline_inode, inline_dir, append, update, orphans; 3828 int compr_inode; 3829 unsigned long long compr_blocks; 3830 int aw_cnt, max_aw_cnt, vw_cnt, max_vw_cnt; 3831 unsigned int valid_count, valid_node_count, valid_inode_count, discard_blks; 3832 unsigned int bimodal, avg_vblocks; 3833 int util_free, util_valid, util_invalid; 3834 int rsvd_segs, overp_segs; 3835 int dirty_count, node_pages, meta_pages, compress_pages; 3836 int compress_page_hit; 3837 int prefree_count, call_count, cp_count, bg_cp_count; 3838 int tot_segs, node_segs, data_segs, free_segs, free_secs; 3839 int bg_node_segs, bg_data_segs; 3840 int tot_blks, data_blks, node_blks; 3841 int bg_data_blks, bg_node_blks; 3842 unsigned long long skipped_atomic_files[2]; 3843 int curseg[NR_CURSEG_TYPE]; 3844 int cursec[NR_CURSEG_TYPE]; 3845 int curzone[NR_CURSEG_TYPE]; 3846 unsigned int dirty_seg[NR_CURSEG_TYPE]; 3847 unsigned int full_seg[NR_CURSEG_TYPE]; 3848 unsigned int valid_blks[NR_CURSEG_TYPE]; 3849 3850 unsigned int meta_count[META_MAX]; 3851 unsigned int segment_count[2]; 3852 unsigned int block_count[2]; 3853 unsigned int inplace_count; 3854 unsigned long long base_mem, cache_mem, page_mem; 3855 }; 3856 3857 static inline struct f2fs_stat_info *F2FS_STAT(struct f2fs_sb_info *sbi) 3858 { 3859 return (struct f2fs_stat_info *)sbi->stat_info; 3860 } 3861 3862 #define stat_inc_cp_count(si) ((si)->cp_count++) 3863 #define stat_inc_bg_cp_count(si) ((si)->bg_cp_count++) 3864 #define stat_inc_call_count(si) ((si)->call_count++) 3865 #define stat_inc_bggc_count(si) ((si)->bg_gc++) 3866 #define stat_io_skip_bggc_count(sbi) ((sbi)->io_skip_bggc++) 3867 #define stat_other_skip_bggc_count(sbi) ((sbi)->other_skip_bggc++) 3868 #define stat_inc_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]++) 3869 #define stat_dec_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]--) 3870 #define stat_inc_total_hit(sbi) (atomic64_inc(&(sbi)->total_hit_ext)) 3871 #define stat_inc_rbtree_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_rbtree)) 3872 #define stat_inc_largest_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_largest)) 3873 #define stat_inc_cached_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_cached)) 3874 #define stat_inc_inline_xattr(inode) \ 3875 do { \ 3876 if (f2fs_has_inline_xattr(inode)) \ 3877 (atomic_inc(&F2FS_I_SB(inode)->inline_xattr)); \ 3878 } while (0) 3879 #define stat_dec_inline_xattr(inode) \ 3880 do { \ 3881 if (f2fs_has_inline_xattr(inode)) \ 3882 (atomic_dec(&F2FS_I_SB(inode)->inline_xattr)); \ 3883 } while (0) 3884 #define stat_inc_inline_inode(inode) \ 3885 do { \ 3886 if (f2fs_has_inline_data(inode)) \ 3887 (atomic_inc(&F2FS_I_SB(inode)->inline_inode)); \ 3888 } while (0) 3889 #define stat_dec_inline_inode(inode) \ 3890 do { \ 3891 if (f2fs_has_inline_data(inode)) \ 3892 (atomic_dec(&F2FS_I_SB(inode)->inline_inode)); \ 3893 } while (0) 3894 #define stat_inc_inline_dir(inode) \ 3895 do { \ 3896 if (f2fs_has_inline_dentry(inode)) \ 3897 (atomic_inc(&F2FS_I_SB(inode)->inline_dir)); \ 3898 } while (0) 3899 #define stat_dec_inline_dir(inode) \ 3900 do { \ 3901 if (f2fs_has_inline_dentry(inode)) \ 3902 (atomic_dec(&F2FS_I_SB(inode)->inline_dir)); \ 3903 } while (0) 3904 #define stat_inc_compr_inode(inode) \ 3905 do { \ 3906 if (f2fs_compressed_file(inode)) \ 3907 (atomic_inc(&F2FS_I_SB(inode)->compr_inode)); \ 3908 } while (0) 3909 #define stat_dec_compr_inode(inode) \ 3910 do { \ 3911 if (f2fs_compressed_file(inode)) \ 3912 (atomic_dec(&F2FS_I_SB(inode)->compr_inode)); \ 3913 } while (0) 3914 #define stat_add_compr_blocks(inode, blocks) \ 3915 (atomic64_add(blocks, &F2FS_I_SB(inode)->compr_blocks)) 3916 #define stat_sub_compr_blocks(inode, blocks) \ 3917 (atomic64_sub(blocks, &F2FS_I_SB(inode)->compr_blocks)) 3918 #define stat_inc_meta_count(sbi, blkaddr) \ 3919 do { \ 3920 if (blkaddr < SIT_I(sbi)->sit_base_addr) \ 3921 atomic_inc(&(sbi)->meta_count[META_CP]); \ 3922 else if (blkaddr < NM_I(sbi)->nat_blkaddr) \ 3923 atomic_inc(&(sbi)->meta_count[META_SIT]); \ 3924 else if (blkaddr < SM_I(sbi)->ssa_blkaddr) \ 3925 atomic_inc(&(sbi)->meta_count[META_NAT]); \ 3926 else if (blkaddr < SM_I(sbi)->main_blkaddr) \ 3927 atomic_inc(&(sbi)->meta_count[META_SSA]); \ 3928 } while (0) 3929 #define stat_inc_seg_type(sbi, curseg) \ 3930 ((sbi)->segment_count[(curseg)->alloc_type]++) 3931 #define stat_inc_block_count(sbi, curseg) \ 3932 ((sbi)->block_count[(curseg)->alloc_type]++) 3933 #define stat_inc_inplace_blocks(sbi) \ 3934 (atomic_inc(&(sbi)->inplace_count)) 3935 #define stat_update_max_atomic_write(inode) \ 3936 do { \ 3937 int cur = F2FS_I_SB(inode)->atomic_files; \ 3938 int max = atomic_read(&F2FS_I_SB(inode)->max_aw_cnt); \ 3939 if (cur > max) \ 3940 atomic_set(&F2FS_I_SB(inode)->max_aw_cnt, cur); \ 3941 } while (0) 3942 #define stat_inc_volatile_write(inode) \ 3943 (atomic_inc(&F2FS_I_SB(inode)->vw_cnt)) 3944 #define stat_dec_volatile_write(inode) \ 3945 (atomic_dec(&F2FS_I_SB(inode)->vw_cnt)) 3946 #define stat_update_max_volatile_write(inode) \ 3947 do { \ 3948 int cur = atomic_read(&F2FS_I_SB(inode)->vw_cnt); \ 3949 int max = atomic_read(&F2FS_I_SB(inode)->max_vw_cnt); \ 3950 if (cur > max) \ 3951 atomic_set(&F2FS_I_SB(inode)->max_vw_cnt, cur); \ 3952 } while (0) 3953 #define stat_inc_seg_count(sbi, type, gc_type) \ 3954 do { \ 3955 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 3956 si->tot_segs++; \ 3957 if ((type) == SUM_TYPE_DATA) { \ 3958 si->data_segs++; \ 3959 si->bg_data_segs += (gc_type == BG_GC) ? 1 : 0; \ 3960 } else { \ 3961 si->node_segs++; \ 3962 si->bg_node_segs += (gc_type == BG_GC) ? 1 : 0; \ 3963 } \ 3964 } while (0) 3965 3966 #define stat_inc_tot_blk_count(si, blks) \ 3967 ((si)->tot_blks += (blks)) 3968 3969 #define stat_inc_data_blk_count(sbi, blks, gc_type) \ 3970 do { \ 3971 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 3972 stat_inc_tot_blk_count(si, blks); \ 3973 si->data_blks += (blks); \ 3974 si->bg_data_blks += ((gc_type) == BG_GC) ? (blks) : 0; \ 3975 } while (0) 3976 3977 #define stat_inc_node_blk_count(sbi, blks, gc_type) \ 3978 do { \ 3979 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 3980 stat_inc_tot_blk_count(si, blks); \ 3981 si->node_blks += (blks); \ 3982 si->bg_node_blks += ((gc_type) == BG_GC) ? (blks) : 0; \ 3983 } while (0) 3984 3985 int f2fs_build_stats(struct f2fs_sb_info *sbi); 3986 void f2fs_destroy_stats(struct f2fs_sb_info *sbi); 3987 void __init f2fs_create_root_stats(void); 3988 void f2fs_destroy_root_stats(void); 3989 void f2fs_update_sit_info(struct f2fs_sb_info *sbi); 3990 #else 3991 #define stat_inc_cp_count(si) do { } while (0) 3992 #define stat_inc_bg_cp_count(si) do { } while (0) 3993 #define stat_inc_call_count(si) do { } while (0) 3994 #define stat_inc_bggc_count(si) do { } while (0) 3995 #define stat_io_skip_bggc_count(sbi) do { } while (0) 3996 #define stat_other_skip_bggc_count(sbi) do { } while (0) 3997 #define stat_inc_dirty_inode(sbi, type) do { } while (0) 3998 #define stat_dec_dirty_inode(sbi, type) do { } while (0) 3999 #define stat_inc_total_hit(sbi) do { } while (0) 4000 #define stat_inc_rbtree_node_hit(sbi) do { } while (0) 4001 #define stat_inc_largest_node_hit(sbi) do { } while (0) 4002 #define stat_inc_cached_node_hit(sbi) do { } while (0) 4003 #define stat_inc_inline_xattr(inode) do { } while (0) 4004 #define stat_dec_inline_xattr(inode) do { } while (0) 4005 #define stat_inc_inline_inode(inode) do { } while (0) 4006 #define stat_dec_inline_inode(inode) do { } while (0) 4007 #define stat_inc_inline_dir(inode) do { } while (0) 4008 #define stat_dec_inline_dir(inode) do { } while (0) 4009 #define stat_inc_compr_inode(inode) do { } while (0) 4010 #define stat_dec_compr_inode(inode) do { } while (0) 4011 #define stat_add_compr_blocks(inode, blocks) do { } while (0) 4012 #define stat_sub_compr_blocks(inode, blocks) do { } while (0) 4013 #define stat_update_max_atomic_write(inode) do { } while (0) 4014 #define stat_inc_volatile_write(inode) do { } while (0) 4015 #define stat_dec_volatile_write(inode) do { } while (0) 4016 #define stat_update_max_volatile_write(inode) do { } while (0) 4017 #define stat_inc_meta_count(sbi, blkaddr) do { } while (0) 4018 #define stat_inc_seg_type(sbi, curseg) do { } while (0) 4019 #define stat_inc_block_count(sbi, curseg) do { } while (0) 4020 #define stat_inc_inplace_blocks(sbi) do { } while (0) 4021 #define stat_inc_seg_count(sbi, type, gc_type) do { } while (0) 4022 #define stat_inc_tot_blk_count(si, blks) do { } while (0) 4023 #define stat_inc_data_blk_count(sbi, blks, gc_type) do { } while (0) 4024 #define stat_inc_node_blk_count(sbi, blks, gc_type) do { } while (0) 4025 4026 static inline int f2fs_build_stats(struct f2fs_sb_info *sbi) { return 0; } 4027 static inline void f2fs_destroy_stats(struct f2fs_sb_info *sbi) { } 4028 static inline void __init f2fs_create_root_stats(void) { } 4029 static inline void f2fs_destroy_root_stats(void) { } 4030 static inline void f2fs_update_sit_info(struct f2fs_sb_info *sbi) {} 4031 #endif 4032 4033 extern const struct file_operations f2fs_dir_operations; 4034 extern const struct file_operations f2fs_file_operations; 4035 extern const struct inode_operations f2fs_file_inode_operations; 4036 extern const struct address_space_operations f2fs_dblock_aops; 4037 extern const struct address_space_operations f2fs_node_aops; 4038 extern const struct address_space_operations f2fs_meta_aops; 4039 extern const struct inode_operations f2fs_dir_inode_operations; 4040 extern const struct inode_operations f2fs_symlink_inode_operations; 4041 extern const struct inode_operations f2fs_encrypted_symlink_inode_operations; 4042 extern const struct inode_operations f2fs_special_inode_operations; 4043 extern struct kmem_cache *f2fs_inode_entry_slab; 4044 4045 /* 4046 * inline.c 4047 */ 4048 bool f2fs_may_inline_data(struct inode *inode); 4049 bool f2fs_may_inline_dentry(struct inode *inode); 4050 void f2fs_do_read_inline_data(struct page *page, struct page *ipage); 4051 void f2fs_truncate_inline_inode(struct inode *inode, 4052 struct page *ipage, u64 from); 4053 int f2fs_read_inline_data(struct inode *inode, struct page *page); 4054 int f2fs_convert_inline_page(struct dnode_of_data *dn, struct page *page); 4055 int f2fs_convert_inline_inode(struct inode *inode); 4056 int f2fs_try_convert_inline_dir(struct inode *dir, struct dentry *dentry); 4057 int f2fs_write_inline_data(struct inode *inode, struct page *page); 4058 int f2fs_recover_inline_data(struct inode *inode, struct page *npage); 4059 struct f2fs_dir_entry *f2fs_find_in_inline_dir(struct inode *dir, 4060 const struct f2fs_filename *fname, 4061 struct page **res_page); 4062 int f2fs_make_empty_inline_dir(struct inode *inode, struct inode *parent, 4063 struct page *ipage); 4064 int f2fs_add_inline_entry(struct inode *dir, const struct f2fs_filename *fname, 4065 struct inode *inode, nid_t ino, umode_t mode); 4066 void f2fs_delete_inline_entry(struct f2fs_dir_entry *dentry, 4067 struct page *page, struct inode *dir, 4068 struct inode *inode); 4069 bool f2fs_empty_inline_dir(struct inode *dir); 4070 int f2fs_read_inline_dir(struct file *file, struct dir_context *ctx, 4071 struct fscrypt_str *fstr); 4072 int f2fs_inline_data_fiemap(struct inode *inode, 4073 struct fiemap_extent_info *fieinfo, 4074 __u64 start, __u64 len); 4075 4076 /* 4077 * shrinker.c 4078 */ 4079 unsigned long f2fs_shrink_count(struct shrinker *shrink, 4080 struct shrink_control *sc); 4081 unsigned long f2fs_shrink_scan(struct shrinker *shrink, 4082 struct shrink_control *sc); 4083 void f2fs_join_shrinker(struct f2fs_sb_info *sbi); 4084 void f2fs_leave_shrinker(struct f2fs_sb_info *sbi); 4085 4086 /* 4087 * extent_cache.c 4088 */ 4089 struct rb_entry *f2fs_lookup_rb_tree(struct rb_root_cached *root, 4090 struct rb_entry *cached_re, unsigned int ofs); 4091 struct rb_node **f2fs_lookup_rb_tree_ext(struct f2fs_sb_info *sbi, 4092 struct rb_root_cached *root, 4093 struct rb_node **parent, 4094 unsigned long long key, bool *left_most); 4095 struct rb_node **f2fs_lookup_rb_tree_for_insert(struct f2fs_sb_info *sbi, 4096 struct rb_root_cached *root, 4097 struct rb_node **parent, 4098 unsigned int ofs, bool *leftmost); 4099 struct rb_entry *f2fs_lookup_rb_tree_ret(struct rb_root_cached *root, 4100 struct rb_entry *cached_re, unsigned int ofs, 4101 struct rb_entry **prev_entry, struct rb_entry **next_entry, 4102 struct rb_node ***insert_p, struct rb_node **insert_parent, 4103 bool force, bool *leftmost); 4104 bool f2fs_check_rb_tree_consistence(struct f2fs_sb_info *sbi, 4105 struct rb_root_cached *root, bool check_key); 4106 unsigned int f2fs_shrink_extent_tree(struct f2fs_sb_info *sbi, int nr_shrink); 4107 void f2fs_init_extent_tree(struct inode *inode, struct page *ipage); 4108 void f2fs_drop_extent_tree(struct inode *inode); 4109 unsigned int f2fs_destroy_extent_node(struct inode *inode); 4110 void f2fs_destroy_extent_tree(struct inode *inode); 4111 bool f2fs_lookup_extent_cache(struct inode *inode, pgoff_t pgofs, 4112 struct extent_info *ei); 4113 void f2fs_update_extent_cache(struct dnode_of_data *dn); 4114 void f2fs_update_extent_cache_range(struct dnode_of_data *dn, 4115 pgoff_t fofs, block_t blkaddr, unsigned int len); 4116 void f2fs_init_extent_cache_info(struct f2fs_sb_info *sbi); 4117 int __init f2fs_create_extent_cache(void); 4118 void f2fs_destroy_extent_cache(void); 4119 4120 /* 4121 * sysfs.c 4122 */ 4123 #define MIN_RA_MUL 2 4124 #define MAX_RA_MUL 256 4125 4126 int __init f2fs_init_sysfs(void); 4127 void f2fs_exit_sysfs(void); 4128 int f2fs_register_sysfs(struct f2fs_sb_info *sbi); 4129 void f2fs_unregister_sysfs(struct f2fs_sb_info *sbi); 4130 4131 /* verity.c */ 4132 extern const struct fsverity_operations f2fs_verityops; 4133 4134 /* 4135 * crypto support 4136 */ 4137 static inline bool f2fs_encrypted_file(struct inode *inode) 4138 { 4139 return IS_ENCRYPTED(inode) && S_ISREG(inode->i_mode); 4140 } 4141 4142 static inline void f2fs_set_encrypted_inode(struct inode *inode) 4143 { 4144 #ifdef CONFIG_FS_ENCRYPTION 4145 file_set_encrypt(inode); 4146 f2fs_set_inode_flags(inode); 4147 #endif 4148 } 4149 4150 /* 4151 * Returns true if the reads of the inode's data need to undergo some 4152 * postprocessing step, like decryption or authenticity verification. 4153 */ 4154 static inline bool f2fs_post_read_required(struct inode *inode) 4155 { 4156 return f2fs_encrypted_file(inode) || fsverity_active(inode) || 4157 f2fs_compressed_file(inode); 4158 } 4159 4160 /* 4161 * compress.c 4162 */ 4163 #ifdef CONFIG_F2FS_FS_COMPRESSION 4164 bool f2fs_is_compressed_page(struct page *page); 4165 struct page *f2fs_compress_control_page(struct page *page); 4166 int f2fs_prepare_compress_overwrite(struct inode *inode, 4167 struct page **pagep, pgoff_t index, void **fsdata); 4168 bool f2fs_compress_write_end(struct inode *inode, void *fsdata, 4169 pgoff_t index, unsigned copied); 4170 int f2fs_truncate_partial_cluster(struct inode *inode, u64 from, bool lock); 4171 void f2fs_compress_write_end_io(struct bio *bio, struct page *page); 4172 bool f2fs_is_compress_backend_ready(struct inode *inode); 4173 int f2fs_init_compress_mempool(void); 4174 void f2fs_destroy_compress_mempool(void); 4175 void f2fs_decompress_cluster(struct decompress_io_ctx *dic); 4176 void f2fs_end_read_compressed_page(struct page *page, bool failed, 4177 block_t blkaddr); 4178 bool f2fs_cluster_is_empty(struct compress_ctx *cc); 4179 bool f2fs_cluster_can_merge_page(struct compress_ctx *cc, pgoff_t index); 4180 bool f2fs_all_cluster_page_loaded(struct compress_ctx *cc, struct pagevec *pvec, 4181 int index, int nr_pages); 4182 bool f2fs_sanity_check_cluster(struct dnode_of_data *dn); 4183 void f2fs_compress_ctx_add_page(struct compress_ctx *cc, struct page *page); 4184 int f2fs_write_multi_pages(struct compress_ctx *cc, 4185 int *submitted, 4186 struct writeback_control *wbc, 4187 enum iostat_type io_type); 4188 int f2fs_is_compressed_cluster(struct inode *inode, pgoff_t index); 4189 void f2fs_update_extent_tree_range_compressed(struct inode *inode, 4190 pgoff_t fofs, block_t blkaddr, unsigned int llen, 4191 unsigned int c_len); 4192 int f2fs_read_multi_pages(struct compress_ctx *cc, struct bio **bio_ret, 4193 unsigned nr_pages, sector_t *last_block_in_bio, 4194 bool is_readahead, bool for_write); 4195 struct decompress_io_ctx *f2fs_alloc_dic(struct compress_ctx *cc); 4196 void f2fs_decompress_end_io(struct decompress_io_ctx *dic, bool failed); 4197 void f2fs_put_page_dic(struct page *page); 4198 unsigned int f2fs_cluster_blocks_are_contiguous(struct dnode_of_data *dn); 4199 int f2fs_init_compress_ctx(struct compress_ctx *cc); 4200 void f2fs_destroy_compress_ctx(struct compress_ctx *cc, bool reuse); 4201 void f2fs_init_compress_info(struct f2fs_sb_info *sbi); 4202 int f2fs_init_compress_inode(struct f2fs_sb_info *sbi); 4203 void f2fs_destroy_compress_inode(struct f2fs_sb_info *sbi); 4204 int f2fs_init_page_array_cache(struct f2fs_sb_info *sbi); 4205 void f2fs_destroy_page_array_cache(struct f2fs_sb_info *sbi); 4206 int __init f2fs_init_compress_cache(void); 4207 void f2fs_destroy_compress_cache(void); 4208 struct address_space *COMPRESS_MAPPING(struct f2fs_sb_info *sbi); 4209 void f2fs_invalidate_compress_page(struct f2fs_sb_info *sbi, block_t blkaddr); 4210 void f2fs_cache_compressed_page(struct f2fs_sb_info *sbi, struct page *page, 4211 nid_t ino, block_t blkaddr); 4212 bool f2fs_load_compressed_page(struct f2fs_sb_info *sbi, struct page *page, 4213 block_t blkaddr); 4214 void f2fs_invalidate_compress_pages(struct f2fs_sb_info *sbi, nid_t ino); 4215 #define inc_compr_inode_stat(inode) \ 4216 do { \ 4217 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); \ 4218 sbi->compr_new_inode++; \ 4219 } while (0) 4220 #define add_compr_block_stat(inode, blocks) \ 4221 do { \ 4222 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); \ 4223 int diff = F2FS_I(inode)->i_cluster_size - blocks; \ 4224 sbi->compr_written_block += blocks; \ 4225 sbi->compr_saved_block += diff; \ 4226 } while (0) 4227 #else 4228 static inline bool f2fs_is_compressed_page(struct page *page) { return false; } 4229 static inline bool f2fs_is_compress_backend_ready(struct inode *inode) 4230 { 4231 if (!f2fs_compressed_file(inode)) 4232 return true; 4233 /* not support compression */ 4234 return false; 4235 } 4236 static inline struct page *f2fs_compress_control_page(struct page *page) 4237 { 4238 WARN_ON_ONCE(1); 4239 return ERR_PTR(-EINVAL); 4240 } 4241 static inline int f2fs_init_compress_mempool(void) { return 0; } 4242 static inline void f2fs_destroy_compress_mempool(void) { } 4243 static inline void f2fs_decompress_cluster(struct decompress_io_ctx *dic) { } 4244 static inline void f2fs_end_read_compressed_page(struct page *page, 4245 bool failed, block_t blkaddr) 4246 { 4247 WARN_ON_ONCE(1); 4248 } 4249 static inline void f2fs_put_page_dic(struct page *page) 4250 { 4251 WARN_ON_ONCE(1); 4252 } 4253 static inline unsigned int f2fs_cluster_blocks_are_contiguous(struct dnode_of_data *dn) { return 0; } 4254 static inline bool f2fs_sanity_check_cluster(struct dnode_of_data *dn) { return false; } 4255 static inline int f2fs_init_compress_inode(struct f2fs_sb_info *sbi) { return 0; } 4256 static inline void f2fs_destroy_compress_inode(struct f2fs_sb_info *sbi) { } 4257 static inline int f2fs_init_page_array_cache(struct f2fs_sb_info *sbi) { return 0; } 4258 static inline void f2fs_destroy_page_array_cache(struct f2fs_sb_info *sbi) { } 4259 static inline int __init f2fs_init_compress_cache(void) { return 0; } 4260 static inline void f2fs_destroy_compress_cache(void) { } 4261 static inline void f2fs_invalidate_compress_page(struct f2fs_sb_info *sbi, 4262 block_t blkaddr) { } 4263 static inline void f2fs_cache_compressed_page(struct f2fs_sb_info *sbi, 4264 struct page *page, nid_t ino, block_t blkaddr) { } 4265 static inline bool f2fs_load_compressed_page(struct f2fs_sb_info *sbi, 4266 struct page *page, block_t blkaddr) { return false; } 4267 static inline void f2fs_invalidate_compress_pages(struct f2fs_sb_info *sbi, 4268 nid_t ino) { } 4269 #define inc_compr_inode_stat(inode) do { } while (0) 4270 static inline void f2fs_update_extent_tree_range_compressed(struct inode *inode, 4271 pgoff_t fofs, block_t blkaddr, unsigned int llen, 4272 unsigned int c_len) { } 4273 #endif 4274 4275 static inline void set_compress_context(struct inode *inode) 4276 { 4277 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 4278 4279 F2FS_I(inode)->i_compress_algorithm = 4280 F2FS_OPTION(sbi).compress_algorithm; 4281 F2FS_I(inode)->i_log_cluster_size = 4282 F2FS_OPTION(sbi).compress_log_size; 4283 F2FS_I(inode)->i_compress_flag = 4284 F2FS_OPTION(sbi).compress_chksum ? 4285 1 << COMPRESS_CHKSUM : 0; 4286 F2FS_I(inode)->i_cluster_size = 4287 1 << F2FS_I(inode)->i_log_cluster_size; 4288 if ((F2FS_I(inode)->i_compress_algorithm == COMPRESS_LZ4 || 4289 F2FS_I(inode)->i_compress_algorithm == COMPRESS_ZSTD) && 4290 F2FS_OPTION(sbi).compress_level) 4291 F2FS_I(inode)->i_compress_flag |= 4292 F2FS_OPTION(sbi).compress_level << 4293 COMPRESS_LEVEL_OFFSET; 4294 F2FS_I(inode)->i_flags |= F2FS_COMPR_FL; 4295 set_inode_flag(inode, FI_COMPRESSED_FILE); 4296 stat_inc_compr_inode(inode); 4297 inc_compr_inode_stat(inode); 4298 f2fs_mark_inode_dirty_sync(inode, true); 4299 } 4300 4301 static inline bool f2fs_disable_compressed_file(struct inode *inode) 4302 { 4303 struct f2fs_inode_info *fi = F2FS_I(inode); 4304 4305 if (!f2fs_compressed_file(inode)) 4306 return true; 4307 if (S_ISREG(inode->i_mode) && F2FS_HAS_BLOCKS(inode)) 4308 return false; 4309 4310 fi->i_flags &= ~F2FS_COMPR_FL; 4311 stat_dec_compr_inode(inode); 4312 clear_inode_flag(inode, FI_COMPRESSED_FILE); 4313 f2fs_mark_inode_dirty_sync(inode, true); 4314 return true; 4315 } 4316 4317 #define F2FS_FEATURE_FUNCS(name, flagname) \ 4318 static inline int f2fs_sb_has_##name(struct f2fs_sb_info *sbi) \ 4319 { \ 4320 return F2FS_HAS_FEATURE(sbi, F2FS_FEATURE_##flagname); \ 4321 } 4322 4323 F2FS_FEATURE_FUNCS(encrypt, ENCRYPT); 4324 F2FS_FEATURE_FUNCS(blkzoned, BLKZONED); 4325 F2FS_FEATURE_FUNCS(extra_attr, EXTRA_ATTR); 4326 F2FS_FEATURE_FUNCS(project_quota, PRJQUOTA); 4327 F2FS_FEATURE_FUNCS(inode_chksum, INODE_CHKSUM); 4328 F2FS_FEATURE_FUNCS(flexible_inline_xattr, FLEXIBLE_INLINE_XATTR); 4329 F2FS_FEATURE_FUNCS(quota_ino, QUOTA_INO); 4330 F2FS_FEATURE_FUNCS(inode_crtime, INODE_CRTIME); 4331 F2FS_FEATURE_FUNCS(lost_found, LOST_FOUND); 4332 F2FS_FEATURE_FUNCS(verity, VERITY); 4333 F2FS_FEATURE_FUNCS(sb_chksum, SB_CHKSUM); 4334 F2FS_FEATURE_FUNCS(casefold, CASEFOLD); 4335 F2FS_FEATURE_FUNCS(compression, COMPRESSION); 4336 F2FS_FEATURE_FUNCS(readonly, RO); 4337 4338 static inline bool f2fs_may_extent_tree(struct inode *inode) 4339 { 4340 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 4341 4342 if (!test_opt(sbi, EXTENT_CACHE) || 4343 is_inode_flag_set(inode, FI_NO_EXTENT) || 4344 (is_inode_flag_set(inode, FI_COMPRESSED_FILE) && 4345 !f2fs_sb_has_readonly(sbi))) 4346 return false; 4347 4348 /* 4349 * for recovered files during mount do not create extents 4350 * if shrinker is not registered. 4351 */ 4352 if (list_empty(&sbi->s_list)) 4353 return false; 4354 4355 return S_ISREG(inode->i_mode); 4356 } 4357 4358 #ifdef CONFIG_BLK_DEV_ZONED 4359 static inline bool f2fs_blkz_is_seq(struct f2fs_sb_info *sbi, int devi, 4360 block_t blkaddr) 4361 { 4362 unsigned int zno = blkaddr >> sbi->log_blocks_per_blkz; 4363 4364 return test_bit(zno, FDEV(devi).blkz_seq); 4365 } 4366 #endif 4367 4368 static inline bool f2fs_hw_should_discard(struct f2fs_sb_info *sbi) 4369 { 4370 return f2fs_sb_has_blkzoned(sbi); 4371 } 4372 4373 static inline bool f2fs_bdev_support_discard(struct block_device *bdev) 4374 { 4375 return bdev_max_discard_sectors(bdev) || bdev_is_zoned(bdev); 4376 } 4377 4378 static inline bool f2fs_hw_support_discard(struct f2fs_sb_info *sbi) 4379 { 4380 int i; 4381 4382 if (!f2fs_is_multi_device(sbi)) 4383 return f2fs_bdev_support_discard(sbi->sb->s_bdev); 4384 4385 for (i = 0; i < sbi->s_ndevs; i++) 4386 if (f2fs_bdev_support_discard(FDEV(i).bdev)) 4387 return true; 4388 return false; 4389 } 4390 4391 static inline bool f2fs_realtime_discard_enable(struct f2fs_sb_info *sbi) 4392 { 4393 return (test_opt(sbi, DISCARD) && f2fs_hw_support_discard(sbi)) || 4394 f2fs_hw_should_discard(sbi); 4395 } 4396 4397 static inline bool f2fs_hw_is_readonly(struct f2fs_sb_info *sbi) 4398 { 4399 int i; 4400 4401 if (!f2fs_is_multi_device(sbi)) 4402 return bdev_read_only(sbi->sb->s_bdev); 4403 4404 for (i = 0; i < sbi->s_ndevs; i++) 4405 if (bdev_read_only(FDEV(i).bdev)) 4406 return true; 4407 return false; 4408 } 4409 4410 static inline bool f2fs_lfs_mode(struct f2fs_sb_info *sbi) 4411 { 4412 return F2FS_OPTION(sbi).fs_mode == FS_MODE_LFS; 4413 } 4414 4415 static inline bool f2fs_may_compress(struct inode *inode) 4416 { 4417 if (IS_SWAPFILE(inode) || f2fs_is_pinned_file(inode) || 4418 f2fs_is_atomic_file(inode) || 4419 f2fs_is_volatile_file(inode)) 4420 return false; 4421 return S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode); 4422 } 4423 4424 static inline void f2fs_i_compr_blocks_update(struct inode *inode, 4425 u64 blocks, bool add) 4426 { 4427 int diff = F2FS_I(inode)->i_cluster_size - blocks; 4428 struct f2fs_inode_info *fi = F2FS_I(inode); 4429 4430 /* don't update i_compr_blocks if saved blocks were released */ 4431 if (!add && !atomic_read(&fi->i_compr_blocks)) 4432 return; 4433 4434 if (add) { 4435 atomic_add(diff, &fi->i_compr_blocks); 4436 stat_add_compr_blocks(inode, diff); 4437 } else { 4438 atomic_sub(diff, &fi->i_compr_blocks); 4439 stat_sub_compr_blocks(inode, diff); 4440 } 4441 f2fs_mark_inode_dirty_sync(inode, true); 4442 } 4443 4444 static inline int block_unaligned_IO(struct inode *inode, 4445 struct kiocb *iocb, struct iov_iter *iter) 4446 { 4447 unsigned int i_blkbits = READ_ONCE(inode->i_blkbits); 4448 unsigned int blocksize_mask = (1 << i_blkbits) - 1; 4449 loff_t offset = iocb->ki_pos; 4450 unsigned long align = offset | iov_iter_alignment(iter); 4451 4452 return align & blocksize_mask; 4453 } 4454 4455 static inline bool f2fs_allow_multi_device_dio(struct f2fs_sb_info *sbi, 4456 int flag) 4457 { 4458 if (!f2fs_is_multi_device(sbi)) 4459 return false; 4460 if (flag != F2FS_GET_BLOCK_DIO) 4461 return false; 4462 return sbi->aligned_blksize; 4463 } 4464 4465 static inline bool f2fs_force_buffered_io(struct inode *inode, 4466 struct kiocb *iocb, struct iov_iter *iter) 4467 { 4468 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 4469 int rw = iov_iter_rw(iter); 4470 4471 if (!fscrypt_dio_supported(iocb, iter)) 4472 return true; 4473 if (fsverity_active(inode)) 4474 return true; 4475 if (f2fs_compressed_file(inode)) 4476 return true; 4477 4478 /* disallow direct IO if any of devices has unaligned blksize */ 4479 if (f2fs_is_multi_device(sbi) && !sbi->aligned_blksize) 4480 return true; 4481 /* 4482 * for blkzoned device, fallback direct IO to buffered IO, so 4483 * all IOs can be serialized by log-structured write. 4484 */ 4485 if (f2fs_sb_has_blkzoned(sbi)) 4486 return true; 4487 if (f2fs_lfs_mode(sbi) && (rw == WRITE)) { 4488 if (block_unaligned_IO(inode, iocb, iter)) 4489 return true; 4490 if (F2FS_IO_ALIGNED(sbi)) 4491 return true; 4492 } 4493 if (is_sbi_flag_set(F2FS_I_SB(inode), SBI_CP_DISABLED)) 4494 return true; 4495 4496 return false; 4497 } 4498 4499 static inline bool f2fs_need_verity(const struct inode *inode, pgoff_t idx) 4500 { 4501 return fsverity_active(inode) && 4502 idx < DIV_ROUND_UP(inode->i_size, PAGE_SIZE); 4503 } 4504 4505 #ifdef CONFIG_F2FS_FAULT_INJECTION 4506 extern void f2fs_build_fault_attr(struct f2fs_sb_info *sbi, unsigned int rate, 4507 unsigned int type); 4508 #else 4509 #define f2fs_build_fault_attr(sbi, rate, type) do { } while (0) 4510 #endif 4511 4512 static inline bool is_journalled_quota(struct f2fs_sb_info *sbi) 4513 { 4514 #ifdef CONFIG_QUOTA 4515 if (f2fs_sb_has_quota_ino(sbi)) 4516 return true; 4517 if (F2FS_OPTION(sbi).s_qf_names[USRQUOTA] || 4518 F2FS_OPTION(sbi).s_qf_names[GRPQUOTA] || 4519 F2FS_OPTION(sbi).s_qf_names[PRJQUOTA]) 4520 return true; 4521 #endif 4522 return false; 4523 } 4524 4525 static inline bool f2fs_block_unit_discard(struct f2fs_sb_info *sbi) 4526 { 4527 return F2FS_OPTION(sbi).discard_unit == DISCARD_UNIT_BLOCK; 4528 } 4529 4530 static inline void f2fs_io_schedule_timeout(long timeout) 4531 { 4532 set_current_state(TASK_UNINTERRUPTIBLE); 4533 io_schedule_timeout(timeout); 4534 } 4535 4536 #define EFSBADCRC EBADMSG /* Bad CRC detected */ 4537 #define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */ 4538 4539 #endif /* _LINUX_F2FS_H */ 4540