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 wait_queue_head_t read_waiters; 136 }; 137 138 struct f2fs_mount_info { 139 unsigned int opt; 140 int write_io_size_bits; /* Write IO size bits */ 141 block_t root_reserved_blocks; /* root reserved blocks */ 142 kuid_t s_resuid; /* reserved blocks for uid */ 143 kgid_t s_resgid; /* reserved blocks for gid */ 144 int active_logs; /* # of active logs */ 145 int inline_xattr_size; /* inline xattr size */ 146 #ifdef CONFIG_F2FS_FAULT_INJECTION 147 struct f2fs_fault_info fault_info; /* For fault injection */ 148 #endif 149 #ifdef CONFIG_QUOTA 150 /* Names of quota files with journalled quota */ 151 char *s_qf_names[MAXQUOTAS]; 152 int s_jquota_fmt; /* Format of quota to use */ 153 #endif 154 /* For which write hints are passed down to block layer */ 155 int whint_mode; 156 int alloc_mode; /* segment allocation policy */ 157 int fsync_mode; /* fsync policy */ 158 int fs_mode; /* fs mode: LFS or ADAPTIVE */ 159 int bggc_mode; /* bggc mode: off, on or sync */ 160 int discard_unit; /* 161 * discard command's offset/size should 162 * be aligned to this unit: block, 163 * segment or section 164 */ 165 struct fscrypt_dummy_policy dummy_enc_policy; /* test dummy encryption */ 166 block_t unusable_cap_perc; /* percentage for cap */ 167 block_t unusable_cap; /* Amount of space allowed to be 168 * unusable when disabling checkpoint 169 */ 170 171 /* For compression */ 172 unsigned char compress_algorithm; /* algorithm type */ 173 unsigned char compress_log_size; /* cluster log size */ 174 unsigned char compress_level; /* compress level */ 175 bool compress_chksum; /* compressed data chksum */ 176 unsigned char compress_ext_cnt; /* extension count */ 177 unsigned char nocompress_ext_cnt; /* nocompress extension count */ 178 int compress_mode; /* compression mode */ 179 unsigned char extensions[COMPRESS_EXT_NUM][F2FS_EXTENSION_LEN]; /* extensions */ 180 unsigned char noextensions[COMPRESS_EXT_NUM][F2FS_EXTENSION_LEN]; /* extensions */ 181 }; 182 183 #define F2FS_FEATURE_ENCRYPT 0x0001 184 #define F2FS_FEATURE_BLKZONED 0x0002 185 #define F2FS_FEATURE_ATOMIC_WRITE 0x0004 186 #define F2FS_FEATURE_EXTRA_ATTR 0x0008 187 #define F2FS_FEATURE_PRJQUOTA 0x0010 188 #define F2FS_FEATURE_INODE_CHKSUM 0x0020 189 #define F2FS_FEATURE_FLEXIBLE_INLINE_XATTR 0x0040 190 #define F2FS_FEATURE_QUOTA_INO 0x0080 191 #define F2FS_FEATURE_INODE_CRTIME 0x0100 192 #define F2FS_FEATURE_LOST_FOUND 0x0200 193 #define F2FS_FEATURE_VERITY 0x0400 194 #define F2FS_FEATURE_SB_CHKSUM 0x0800 195 #define F2FS_FEATURE_CASEFOLD 0x1000 196 #define F2FS_FEATURE_COMPRESSION 0x2000 197 #define F2FS_FEATURE_RO 0x4000 198 199 #define __F2FS_HAS_FEATURE(raw_super, mask) \ 200 ((raw_super->feature & cpu_to_le32(mask)) != 0) 201 #define F2FS_HAS_FEATURE(sbi, mask) __F2FS_HAS_FEATURE(sbi->raw_super, mask) 202 #define F2FS_SET_FEATURE(sbi, mask) \ 203 (sbi->raw_super->feature |= cpu_to_le32(mask)) 204 #define F2FS_CLEAR_FEATURE(sbi, mask) \ 205 (sbi->raw_super->feature &= ~cpu_to_le32(mask)) 206 207 /* 208 * Default values for user and/or group using reserved blocks 209 */ 210 #define F2FS_DEF_RESUID 0 211 #define F2FS_DEF_RESGID 0 212 213 /* 214 * For checkpoint manager 215 */ 216 enum { 217 NAT_BITMAP, 218 SIT_BITMAP 219 }; 220 221 #define CP_UMOUNT 0x00000001 222 #define CP_FASTBOOT 0x00000002 223 #define CP_SYNC 0x00000004 224 #define CP_RECOVERY 0x00000008 225 #define CP_DISCARD 0x00000010 226 #define CP_TRIMMED 0x00000020 227 #define CP_PAUSE 0x00000040 228 #define CP_RESIZE 0x00000080 229 230 #define MAX_DISCARD_BLOCKS(sbi) BLKS_PER_SEC(sbi) 231 #define DEF_MAX_DISCARD_REQUEST 8 /* issue 8 discards per round */ 232 #define DEF_MIN_DISCARD_ISSUE_TIME 50 /* 50 ms, if exists */ 233 #define DEF_MID_DISCARD_ISSUE_TIME 500 /* 500 ms, if device busy */ 234 #define DEF_MAX_DISCARD_ISSUE_TIME 60000 /* 60 s, if no candidates */ 235 #define DEF_DISCARD_URGENT_UTIL 80 /* do more discard over 80% */ 236 #define DEF_CP_INTERVAL 60 /* 60 secs */ 237 #define DEF_IDLE_INTERVAL 5 /* 5 secs */ 238 #define DEF_DISABLE_INTERVAL 5 /* 5 secs */ 239 #define DEF_DISABLE_QUICK_INTERVAL 1 /* 1 secs */ 240 #define DEF_UMOUNT_DISCARD_TIMEOUT 5 /* 5 secs */ 241 242 struct cp_control { 243 int reason; 244 __u64 trim_start; 245 __u64 trim_end; 246 __u64 trim_minlen; 247 }; 248 249 /* 250 * indicate meta/data type 251 */ 252 enum { 253 META_CP, 254 META_NAT, 255 META_SIT, 256 META_SSA, 257 META_MAX, 258 META_POR, 259 DATA_GENERIC, /* check range only */ 260 DATA_GENERIC_ENHANCE, /* strong check on range and segment bitmap */ 261 DATA_GENERIC_ENHANCE_READ, /* 262 * strong check on range and segment 263 * bitmap but no warning due to race 264 * condition of read on truncated area 265 * by extent_cache 266 */ 267 META_GENERIC, 268 }; 269 270 /* for the list of ino */ 271 enum { 272 ORPHAN_INO, /* for orphan ino list */ 273 APPEND_INO, /* for append ino list */ 274 UPDATE_INO, /* for update ino list */ 275 TRANS_DIR_INO, /* for trasactions dir ino list */ 276 FLUSH_INO, /* for multiple device flushing */ 277 MAX_INO_ENTRY, /* max. list */ 278 }; 279 280 struct ino_entry { 281 struct list_head list; /* list head */ 282 nid_t ino; /* inode number */ 283 unsigned int dirty_device; /* dirty device bitmap */ 284 }; 285 286 /* for the list of inodes to be GCed */ 287 struct inode_entry { 288 struct list_head list; /* list head */ 289 struct inode *inode; /* vfs inode pointer */ 290 }; 291 292 struct fsync_node_entry { 293 struct list_head list; /* list head */ 294 struct page *page; /* warm node page pointer */ 295 unsigned int seq_id; /* sequence id */ 296 }; 297 298 struct ckpt_req { 299 struct completion wait; /* completion for checkpoint done */ 300 struct llist_node llnode; /* llist_node to be linked in wait queue */ 301 int ret; /* return code of checkpoint */ 302 ktime_t queue_time; /* request queued time */ 303 }; 304 305 struct ckpt_req_control { 306 struct task_struct *f2fs_issue_ckpt; /* checkpoint task */ 307 int ckpt_thread_ioprio; /* checkpoint merge thread ioprio */ 308 wait_queue_head_t ckpt_wait_queue; /* waiting queue for wake-up */ 309 atomic_t issued_ckpt; /* # of actually issued ckpts */ 310 atomic_t total_ckpt; /* # of total ckpts */ 311 atomic_t queued_ckpt; /* # of queued ckpts */ 312 struct llist_head issue_list; /* list for command issue */ 313 spinlock_t stat_lock; /* lock for below checkpoint time stats */ 314 unsigned int cur_time; /* cur wait time in msec for currently issued checkpoint */ 315 unsigned int peak_time; /* peak wait time in msec until now */ 316 }; 317 318 /* for the bitmap indicate blocks to be discarded */ 319 struct discard_entry { 320 struct list_head list; /* list head */ 321 block_t start_blkaddr; /* start blockaddr of current segment */ 322 unsigned char discard_map[SIT_VBLOCK_MAP_SIZE]; /* segment discard bitmap */ 323 }; 324 325 /* default discard granularity of inner discard thread, unit: block count */ 326 #define DEFAULT_DISCARD_GRANULARITY 16 327 328 /* max discard pend list number */ 329 #define MAX_PLIST_NUM 512 330 #define plist_idx(blk_num) ((blk_num) >= MAX_PLIST_NUM ? \ 331 (MAX_PLIST_NUM - 1) : ((blk_num) - 1)) 332 333 enum { 334 D_PREP, /* initial */ 335 D_PARTIAL, /* partially submitted */ 336 D_SUBMIT, /* all submitted */ 337 D_DONE, /* finished */ 338 }; 339 340 struct discard_info { 341 block_t lstart; /* logical start address */ 342 block_t len; /* length */ 343 block_t start; /* actual start address in dev */ 344 }; 345 346 struct discard_cmd { 347 struct rb_node rb_node; /* rb node located in rb-tree */ 348 union { 349 struct { 350 block_t lstart; /* logical start address */ 351 block_t len; /* length */ 352 block_t start; /* actual start address in dev */ 353 }; 354 struct discard_info di; /* discard info */ 355 356 }; 357 struct list_head list; /* command list */ 358 struct completion wait; /* compleation */ 359 struct block_device *bdev; /* bdev */ 360 unsigned short ref; /* reference count */ 361 unsigned char state; /* state */ 362 unsigned char queued; /* queued discard */ 363 int error; /* bio error */ 364 spinlock_t lock; /* for state/bio_ref updating */ 365 unsigned short bio_ref; /* bio reference count */ 366 }; 367 368 enum { 369 DPOLICY_BG, 370 DPOLICY_FORCE, 371 DPOLICY_FSTRIM, 372 DPOLICY_UMOUNT, 373 MAX_DPOLICY, 374 }; 375 376 struct discard_policy { 377 int type; /* type of discard */ 378 unsigned int min_interval; /* used for candidates exist */ 379 unsigned int mid_interval; /* used for device busy */ 380 unsigned int max_interval; /* used for candidates not exist */ 381 unsigned int max_requests; /* # of discards issued per round */ 382 unsigned int io_aware_gran; /* minimum granularity discard not be aware of I/O */ 383 bool io_aware; /* issue discard in idle time */ 384 bool sync; /* submit discard with REQ_SYNC flag */ 385 bool ordered; /* issue discard by lba order */ 386 bool timeout; /* discard timeout for put_super */ 387 unsigned int granularity; /* discard granularity */ 388 }; 389 390 struct discard_cmd_control { 391 struct task_struct *f2fs_issue_discard; /* discard thread */ 392 struct list_head entry_list; /* 4KB discard entry list */ 393 struct list_head pend_list[MAX_PLIST_NUM];/* store pending entries */ 394 struct list_head wait_list; /* store on-flushing entries */ 395 struct list_head fstrim_list; /* in-flight discard from fstrim */ 396 wait_queue_head_t discard_wait_queue; /* waiting queue for wake-up */ 397 unsigned int discard_wake; /* to wake up discard thread */ 398 struct mutex cmd_lock; 399 unsigned int nr_discards; /* # of discards in the list */ 400 unsigned int max_discards; /* max. discards to be issued */ 401 unsigned int max_discard_request; /* max. discard request per round */ 402 unsigned int min_discard_issue_time; /* min. interval between discard issue */ 403 unsigned int mid_discard_issue_time; /* mid. interval between discard issue */ 404 unsigned int max_discard_issue_time; /* max. interval between discard issue */ 405 unsigned int discard_granularity; /* discard granularity */ 406 unsigned int undiscard_blks; /* # of undiscard blocks */ 407 unsigned int next_pos; /* next discard position */ 408 atomic_t issued_discard; /* # of issued discard */ 409 atomic_t queued_discard; /* # of queued discard */ 410 atomic_t discard_cmd_cnt; /* # of cached cmd count */ 411 struct rb_root_cached root; /* root of discard rb-tree */ 412 bool rbtree_check; /* config for consistence check */ 413 }; 414 415 /* for the list of fsync inodes, used only during recovery */ 416 struct fsync_inode_entry { 417 struct list_head list; /* list head */ 418 struct inode *inode; /* vfs inode pointer */ 419 block_t blkaddr; /* block address locating the last fsync */ 420 block_t last_dentry; /* block address locating the last dentry */ 421 }; 422 423 #define nats_in_cursum(jnl) (le16_to_cpu((jnl)->n_nats)) 424 #define sits_in_cursum(jnl) (le16_to_cpu((jnl)->n_sits)) 425 426 #define nat_in_journal(jnl, i) ((jnl)->nat_j.entries[i].ne) 427 #define nid_in_journal(jnl, i) ((jnl)->nat_j.entries[i].nid) 428 #define sit_in_journal(jnl, i) ((jnl)->sit_j.entries[i].se) 429 #define segno_in_journal(jnl, i) ((jnl)->sit_j.entries[i].segno) 430 431 #define MAX_NAT_JENTRIES(jnl) (NAT_JOURNAL_ENTRIES - nats_in_cursum(jnl)) 432 #define MAX_SIT_JENTRIES(jnl) (SIT_JOURNAL_ENTRIES - sits_in_cursum(jnl)) 433 434 static inline int update_nats_in_cursum(struct f2fs_journal *journal, int i) 435 { 436 int before = nats_in_cursum(journal); 437 438 journal->n_nats = cpu_to_le16(before + i); 439 return before; 440 } 441 442 static inline int update_sits_in_cursum(struct f2fs_journal *journal, int i) 443 { 444 int before = sits_in_cursum(journal); 445 446 journal->n_sits = cpu_to_le16(before + i); 447 return before; 448 } 449 450 static inline bool __has_cursum_space(struct f2fs_journal *journal, 451 int size, int type) 452 { 453 if (type == NAT_JOURNAL) 454 return size <= MAX_NAT_JENTRIES(journal); 455 return size <= MAX_SIT_JENTRIES(journal); 456 } 457 458 /* for inline stuff */ 459 #define DEF_INLINE_RESERVED_SIZE 1 460 static inline int get_extra_isize(struct inode *inode); 461 static inline int get_inline_xattr_addrs(struct inode *inode); 462 #define MAX_INLINE_DATA(inode) (sizeof(__le32) * \ 463 (CUR_ADDRS_PER_INODE(inode) - \ 464 get_inline_xattr_addrs(inode) - \ 465 DEF_INLINE_RESERVED_SIZE)) 466 467 /* for inline dir */ 468 #define NR_INLINE_DENTRY(inode) (MAX_INLINE_DATA(inode) * BITS_PER_BYTE / \ 469 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \ 470 BITS_PER_BYTE + 1)) 471 #define INLINE_DENTRY_BITMAP_SIZE(inode) \ 472 DIV_ROUND_UP(NR_INLINE_DENTRY(inode), BITS_PER_BYTE) 473 #define INLINE_RESERVED_SIZE(inode) (MAX_INLINE_DATA(inode) - \ 474 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \ 475 NR_INLINE_DENTRY(inode) + \ 476 INLINE_DENTRY_BITMAP_SIZE(inode))) 477 478 /* 479 * For INODE and NODE manager 480 */ 481 /* for directory operations */ 482 483 struct f2fs_filename { 484 /* 485 * The filename the user specified. This is NULL for some 486 * filesystem-internal operations, e.g. converting an inline directory 487 * to a non-inline one, or roll-forward recovering an encrypted dentry. 488 */ 489 const struct qstr *usr_fname; 490 491 /* 492 * The on-disk filename. For encrypted directories, this is encrypted. 493 * This may be NULL for lookups in an encrypted dir without the key. 494 */ 495 struct fscrypt_str disk_name; 496 497 /* The dirhash of this filename */ 498 f2fs_hash_t hash; 499 500 #ifdef CONFIG_FS_ENCRYPTION 501 /* 502 * For lookups in encrypted directories: either the buffer backing 503 * disk_name, or a buffer that holds the decoded no-key name. 504 */ 505 struct fscrypt_str crypto_buf; 506 #endif 507 #ifdef CONFIG_UNICODE 508 /* 509 * For casefolded directories: the casefolded name, but it's left NULL 510 * if the original name is not valid Unicode, if the directory is both 511 * casefolded and encrypted and its encryption key is unavailable, or if 512 * the filesystem is doing an internal operation where usr_fname is also 513 * NULL. In all these cases we fall back to treating the name as an 514 * opaque byte sequence. 515 */ 516 struct fscrypt_str cf_name; 517 #endif 518 }; 519 520 struct f2fs_dentry_ptr { 521 struct inode *inode; 522 void *bitmap; 523 struct f2fs_dir_entry *dentry; 524 __u8 (*filename)[F2FS_SLOT_LEN]; 525 int max; 526 int nr_bitmap; 527 }; 528 529 static inline void make_dentry_ptr_block(struct inode *inode, 530 struct f2fs_dentry_ptr *d, struct f2fs_dentry_block *t) 531 { 532 d->inode = inode; 533 d->max = NR_DENTRY_IN_BLOCK; 534 d->nr_bitmap = SIZE_OF_DENTRY_BITMAP; 535 d->bitmap = t->dentry_bitmap; 536 d->dentry = t->dentry; 537 d->filename = t->filename; 538 } 539 540 static inline void make_dentry_ptr_inline(struct inode *inode, 541 struct f2fs_dentry_ptr *d, void *t) 542 { 543 int entry_cnt = NR_INLINE_DENTRY(inode); 544 int bitmap_size = INLINE_DENTRY_BITMAP_SIZE(inode); 545 int reserved_size = INLINE_RESERVED_SIZE(inode); 546 547 d->inode = inode; 548 d->max = entry_cnt; 549 d->nr_bitmap = bitmap_size; 550 d->bitmap = t; 551 d->dentry = t + bitmap_size + reserved_size; 552 d->filename = t + bitmap_size + reserved_size + 553 SIZE_OF_DIR_ENTRY * entry_cnt; 554 } 555 556 /* 557 * XATTR_NODE_OFFSET stores xattrs to one node block per file keeping -1 558 * as its node offset to distinguish from index node blocks. 559 * But some bits are used to mark the node block. 560 */ 561 #define XATTR_NODE_OFFSET ((((unsigned int)-1) << OFFSET_BIT_SHIFT) \ 562 >> OFFSET_BIT_SHIFT) 563 enum { 564 ALLOC_NODE, /* allocate a new node page if needed */ 565 LOOKUP_NODE, /* look up a node without readahead */ 566 LOOKUP_NODE_RA, /* 567 * look up a node with readahead called 568 * by get_data_block. 569 */ 570 }; 571 572 #define DEFAULT_RETRY_IO_COUNT 8 /* maximum retry read IO or flush count */ 573 574 /* congestion wait timeout value, default: 20ms */ 575 #define DEFAULT_IO_TIMEOUT (msecs_to_jiffies(20)) 576 577 /* maximum retry quota flush count */ 578 #define DEFAULT_RETRY_QUOTA_FLUSH_COUNT 8 579 580 /* maximum retry of EIO'ed meta page */ 581 #define MAX_RETRY_META_PAGE_EIO 100 582 583 #define F2FS_LINK_MAX 0xffffffff /* maximum link count per file */ 584 585 #define MAX_DIR_RA_PAGES 4 /* maximum ra pages of dir */ 586 587 /* dirty segments threshold for triggering CP */ 588 #define DEFAULT_DIRTY_THRESHOLD 4 589 590 /* for in-memory extent cache entry */ 591 #define F2FS_MIN_EXTENT_LEN 64 /* minimum extent length */ 592 593 /* number of extent info in extent cache we try to shrink */ 594 #define EXTENT_CACHE_SHRINK_NUMBER 128 595 596 #define RECOVERY_MAX_RA_BLOCKS BIO_MAX_VECS 597 #define RECOVERY_MIN_RA_BLOCKS 1 598 599 struct rb_entry { 600 struct rb_node rb_node; /* rb node located in rb-tree */ 601 union { 602 struct { 603 unsigned int ofs; /* start offset of the entry */ 604 unsigned int len; /* length of the entry */ 605 }; 606 unsigned long long key; /* 64-bits key */ 607 } __packed; 608 }; 609 610 struct extent_info { 611 unsigned int fofs; /* start offset in a file */ 612 unsigned int len; /* length of the extent */ 613 u32 blk; /* start block address of the extent */ 614 #ifdef CONFIG_F2FS_FS_COMPRESSION 615 unsigned int c_len; /* physical extent length of compressed blocks */ 616 #endif 617 }; 618 619 struct extent_node { 620 struct rb_node rb_node; /* rb node located in rb-tree */ 621 struct extent_info ei; /* extent info */ 622 struct list_head list; /* node in global extent list of sbi */ 623 struct extent_tree *et; /* extent tree pointer */ 624 }; 625 626 struct extent_tree { 627 nid_t ino; /* inode number */ 628 struct rb_root_cached root; /* root of extent info rb-tree */ 629 struct extent_node *cached_en; /* recently accessed extent node */ 630 struct extent_info largest; /* largested extent info */ 631 struct list_head list; /* to be used by sbi->zombie_list */ 632 rwlock_t lock; /* protect extent info rb-tree */ 633 atomic_t node_cnt; /* # of extent node in rb-tree*/ 634 bool largest_updated; /* largest extent updated */ 635 }; 636 637 /* 638 * This structure is taken from ext4_map_blocks. 639 * 640 * Note that, however, f2fs uses NEW and MAPPED flags for f2fs_map_blocks(). 641 */ 642 #define F2FS_MAP_NEW (1 << BH_New) 643 #define F2FS_MAP_MAPPED (1 << BH_Mapped) 644 #define F2FS_MAP_UNWRITTEN (1 << BH_Unwritten) 645 #define F2FS_MAP_FLAGS (F2FS_MAP_NEW | F2FS_MAP_MAPPED |\ 646 F2FS_MAP_UNWRITTEN) 647 648 struct f2fs_map_blocks { 649 struct block_device *m_bdev; /* for multi-device dio */ 650 block_t m_pblk; 651 block_t m_lblk; 652 unsigned int m_len; 653 unsigned int m_flags; 654 pgoff_t *m_next_pgofs; /* point next possible non-hole pgofs */ 655 pgoff_t *m_next_extent; /* point to next possible extent */ 656 int m_seg_type; 657 bool m_may_create; /* indicate it is from write path */ 658 bool m_multidev_dio; /* indicate it allows multi-device dio */ 659 }; 660 661 /* for flag in get_data_block */ 662 enum { 663 F2FS_GET_BLOCK_DEFAULT, 664 F2FS_GET_BLOCK_FIEMAP, 665 F2FS_GET_BLOCK_BMAP, 666 F2FS_GET_BLOCK_DIO, 667 F2FS_GET_BLOCK_PRE_DIO, 668 F2FS_GET_BLOCK_PRE_AIO, 669 F2FS_GET_BLOCK_PRECACHE, 670 }; 671 672 /* 673 * i_advise uses FADVISE_XXX_BIT. We can add additional hints later. 674 */ 675 #define FADVISE_COLD_BIT 0x01 676 #define FADVISE_LOST_PINO_BIT 0x02 677 #define FADVISE_ENCRYPT_BIT 0x04 678 #define FADVISE_ENC_NAME_BIT 0x08 679 #define FADVISE_KEEP_SIZE_BIT 0x10 680 #define FADVISE_HOT_BIT 0x20 681 #define FADVISE_VERITY_BIT 0x40 682 #define FADVISE_TRUNC_BIT 0x80 683 684 #define FADVISE_MODIFIABLE_BITS (FADVISE_COLD_BIT | FADVISE_HOT_BIT) 685 686 #define file_is_cold(inode) is_file(inode, FADVISE_COLD_BIT) 687 #define file_set_cold(inode) set_file(inode, FADVISE_COLD_BIT) 688 #define file_clear_cold(inode) clear_file(inode, FADVISE_COLD_BIT) 689 690 #define file_wrong_pino(inode) is_file(inode, FADVISE_LOST_PINO_BIT) 691 #define file_lost_pino(inode) set_file(inode, FADVISE_LOST_PINO_BIT) 692 #define file_got_pino(inode) clear_file(inode, FADVISE_LOST_PINO_BIT) 693 694 #define file_is_encrypt(inode) is_file(inode, FADVISE_ENCRYPT_BIT) 695 #define file_set_encrypt(inode) set_file(inode, FADVISE_ENCRYPT_BIT) 696 697 #define file_enc_name(inode) is_file(inode, FADVISE_ENC_NAME_BIT) 698 #define file_set_enc_name(inode) set_file(inode, FADVISE_ENC_NAME_BIT) 699 700 #define file_keep_isize(inode) is_file(inode, FADVISE_KEEP_SIZE_BIT) 701 #define file_set_keep_isize(inode) set_file(inode, FADVISE_KEEP_SIZE_BIT) 702 703 #define file_is_hot(inode) is_file(inode, FADVISE_HOT_BIT) 704 #define file_set_hot(inode) set_file(inode, FADVISE_HOT_BIT) 705 #define file_clear_hot(inode) clear_file(inode, FADVISE_HOT_BIT) 706 707 #define file_is_verity(inode) is_file(inode, FADVISE_VERITY_BIT) 708 #define file_set_verity(inode) set_file(inode, FADVISE_VERITY_BIT) 709 710 #define file_should_truncate(inode) is_file(inode, FADVISE_TRUNC_BIT) 711 #define file_need_truncate(inode) set_file(inode, FADVISE_TRUNC_BIT) 712 #define file_dont_truncate(inode) clear_file(inode, FADVISE_TRUNC_BIT) 713 714 #define DEF_DIR_LEVEL 0 715 716 enum { 717 GC_FAILURE_PIN, 718 GC_FAILURE_ATOMIC, 719 MAX_GC_FAILURE 720 }; 721 722 /* used for f2fs_inode_info->flags */ 723 enum { 724 FI_NEW_INODE, /* indicate newly allocated inode */ 725 FI_DIRTY_INODE, /* indicate inode is dirty or not */ 726 FI_AUTO_RECOVER, /* indicate inode is recoverable */ 727 FI_DIRTY_DIR, /* indicate directory has dirty pages */ 728 FI_INC_LINK, /* need to increment i_nlink */ 729 FI_ACL_MODE, /* indicate acl mode */ 730 FI_NO_ALLOC, /* should not allocate any blocks */ 731 FI_FREE_NID, /* free allocated nide */ 732 FI_NO_EXTENT, /* not to use the extent cache */ 733 FI_INLINE_XATTR, /* used for inline xattr */ 734 FI_INLINE_DATA, /* used for inline data*/ 735 FI_INLINE_DENTRY, /* used for inline dentry */ 736 FI_APPEND_WRITE, /* inode has appended data */ 737 FI_UPDATE_WRITE, /* inode has in-place-update data */ 738 FI_NEED_IPU, /* used for ipu per file */ 739 FI_ATOMIC_FILE, /* indicate atomic file */ 740 FI_ATOMIC_COMMIT, /* indicate the state of atomical committing */ 741 FI_VOLATILE_FILE, /* indicate volatile file */ 742 FI_FIRST_BLOCK_WRITTEN, /* indicate #0 data block was written */ 743 FI_DROP_CACHE, /* drop dirty page cache */ 744 FI_DATA_EXIST, /* indicate data exists */ 745 FI_INLINE_DOTS, /* indicate inline dot dentries */ 746 FI_SKIP_WRITES, /* should skip data page writeback */ 747 FI_OPU_WRITE, /* used for opu per file */ 748 FI_DIRTY_FILE, /* indicate regular/symlink has dirty pages */ 749 FI_PREALLOCATED_ALL, /* all blocks for write were preallocated */ 750 FI_HOT_DATA, /* indicate file is hot */ 751 FI_EXTRA_ATTR, /* indicate file has extra attribute */ 752 FI_PROJ_INHERIT, /* indicate file inherits projectid */ 753 FI_PIN_FILE, /* indicate file should not be gced */ 754 FI_ATOMIC_REVOKE_REQUEST, /* request to drop atomic data */ 755 FI_VERITY_IN_PROGRESS, /* building fs-verity Merkle tree */ 756 FI_COMPRESSED_FILE, /* indicate file's data can be compressed */ 757 FI_COMPRESS_CORRUPT, /* indicate compressed cluster is corrupted */ 758 FI_MMAP_FILE, /* indicate file was mmapped */ 759 FI_ENABLE_COMPRESS, /* enable compression in "user" compression mode */ 760 FI_COMPRESS_RELEASED, /* compressed blocks were released */ 761 FI_ALIGNED_WRITE, /* enable aligned write */ 762 FI_MAX, /* max flag, never be used */ 763 }; 764 765 struct f2fs_inode_info { 766 struct inode vfs_inode; /* serve a vfs inode */ 767 unsigned long i_flags; /* keep an inode flags for ioctl */ 768 unsigned char i_advise; /* use to give file attribute hints */ 769 unsigned char i_dir_level; /* use for dentry level for large dir */ 770 unsigned int i_current_depth; /* only for directory depth */ 771 /* for gc failure statistic */ 772 unsigned int i_gc_failures[MAX_GC_FAILURE]; 773 unsigned int i_pino; /* parent inode number */ 774 umode_t i_acl_mode; /* keep file acl mode temporarily */ 775 776 /* Use below internally in f2fs*/ 777 unsigned long flags[BITS_TO_LONGS(FI_MAX)]; /* use to pass per-file flags */ 778 struct f2fs_rwsem i_sem; /* protect fi info */ 779 atomic_t dirty_pages; /* # of dirty pages */ 780 f2fs_hash_t chash; /* hash value of given file name */ 781 unsigned int clevel; /* maximum level of given file name */ 782 struct task_struct *task; /* lookup and create consistency */ 783 struct task_struct *cp_task; /* separate cp/wb IO stats*/ 784 nid_t i_xattr_nid; /* node id that contains xattrs */ 785 loff_t last_disk_size; /* lastly written file size */ 786 spinlock_t i_size_lock; /* protect last_disk_size */ 787 788 #ifdef CONFIG_QUOTA 789 struct dquot *i_dquot[MAXQUOTAS]; 790 791 /* quota space reservation, managed internally by quota code */ 792 qsize_t i_reserved_quota; 793 #endif 794 struct list_head dirty_list; /* dirty list for dirs and files */ 795 struct list_head gdirty_list; /* linked in global dirty list */ 796 struct list_head inmem_ilist; /* list for inmem inodes */ 797 struct list_head inmem_pages; /* inmemory pages managed by f2fs */ 798 struct task_struct *inmem_task; /* store inmemory task */ 799 struct mutex inmem_lock; /* lock for inmemory pages */ 800 struct extent_tree *extent_tree; /* cached extent_tree entry */ 801 802 /* avoid racing between foreground op and gc */ 803 struct f2fs_rwsem i_gc_rwsem[2]; 804 struct f2fs_rwsem i_xattr_sem; /* avoid racing between reading and changing EAs */ 805 806 int i_extra_isize; /* size of extra space located in i_addr */ 807 kprojid_t i_projid; /* id for project quota */ 808 int i_inline_xattr_size; /* inline xattr size */ 809 struct timespec64 i_crtime; /* inode creation time */ 810 struct timespec64 i_disk_time[4];/* inode disk times */ 811 812 /* for file compress */ 813 atomic_t i_compr_blocks; /* # of compressed blocks */ 814 unsigned char i_compress_algorithm; /* algorithm type */ 815 unsigned char i_log_cluster_size; /* log of cluster size */ 816 unsigned char i_compress_level; /* compress level (lz4hc,zstd) */ 817 unsigned short i_compress_flag; /* compress flag */ 818 unsigned int i_cluster_size; /* cluster size */ 819 }; 820 821 static inline void get_extent_info(struct extent_info *ext, 822 struct f2fs_extent *i_ext) 823 { 824 ext->fofs = le32_to_cpu(i_ext->fofs); 825 ext->blk = le32_to_cpu(i_ext->blk); 826 ext->len = le32_to_cpu(i_ext->len); 827 } 828 829 static inline void set_raw_extent(struct extent_info *ext, 830 struct f2fs_extent *i_ext) 831 { 832 i_ext->fofs = cpu_to_le32(ext->fofs); 833 i_ext->blk = cpu_to_le32(ext->blk); 834 i_ext->len = cpu_to_le32(ext->len); 835 } 836 837 static inline void set_extent_info(struct extent_info *ei, unsigned int fofs, 838 u32 blk, unsigned int len) 839 { 840 ei->fofs = fofs; 841 ei->blk = blk; 842 ei->len = len; 843 #ifdef CONFIG_F2FS_FS_COMPRESSION 844 ei->c_len = 0; 845 #endif 846 } 847 848 static inline bool __is_discard_mergeable(struct discard_info *back, 849 struct discard_info *front, unsigned int max_len) 850 { 851 return (back->lstart + back->len == front->lstart) && 852 (back->len + front->len <= max_len); 853 } 854 855 static inline bool __is_discard_back_mergeable(struct discard_info *cur, 856 struct discard_info *back, unsigned int max_len) 857 { 858 return __is_discard_mergeable(back, cur, max_len); 859 } 860 861 static inline bool __is_discard_front_mergeable(struct discard_info *cur, 862 struct discard_info *front, unsigned int max_len) 863 { 864 return __is_discard_mergeable(cur, front, max_len); 865 } 866 867 static inline bool __is_extent_mergeable(struct extent_info *back, 868 struct extent_info *front) 869 { 870 #ifdef CONFIG_F2FS_FS_COMPRESSION 871 if (back->c_len && back->len != back->c_len) 872 return false; 873 if (front->c_len && front->len != front->c_len) 874 return false; 875 #endif 876 return (back->fofs + back->len == front->fofs && 877 back->blk + back->len == front->blk); 878 } 879 880 static inline bool __is_back_mergeable(struct extent_info *cur, 881 struct extent_info *back) 882 { 883 return __is_extent_mergeable(back, cur); 884 } 885 886 static inline bool __is_front_mergeable(struct extent_info *cur, 887 struct extent_info *front) 888 { 889 return __is_extent_mergeable(cur, front); 890 } 891 892 extern void f2fs_mark_inode_dirty_sync(struct inode *inode, bool sync); 893 static inline void __try_update_largest_extent(struct extent_tree *et, 894 struct extent_node *en) 895 { 896 if (en->ei.len > et->largest.len) { 897 et->largest = en->ei; 898 et->largest_updated = true; 899 } 900 } 901 902 /* 903 * For free nid management 904 */ 905 enum nid_state { 906 FREE_NID, /* newly added to free nid list */ 907 PREALLOC_NID, /* it is preallocated */ 908 MAX_NID_STATE, 909 }; 910 911 enum nat_state { 912 TOTAL_NAT, 913 DIRTY_NAT, 914 RECLAIMABLE_NAT, 915 MAX_NAT_STATE, 916 }; 917 918 struct f2fs_nm_info { 919 block_t nat_blkaddr; /* base disk address of NAT */ 920 nid_t max_nid; /* maximum possible node ids */ 921 nid_t available_nids; /* # of available node ids */ 922 nid_t next_scan_nid; /* the next nid to be scanned */ 923 nid_t max_rf_node_blocks; /* max # of nodes for recovery */ 924 unsigned int ram_thresh; /* control the memory footprint */ 925 unsigned int ra_nid_pages; /* # of nid pages to be readaheaded */ 926 unsigned int dirty_nats_ratio; /* control dirty nats ratio threshold */ 927 928 /* NAT cache management */ 929 struct radix_tree_root nat_root;/* root of the nat entry cache */ 930 struct radix_tree_root nat_set_root;/* root of the nat set cache */ 931 struct f2fs_rwsem nat_tree_lock; /* protect nat entry tree */ 932 struct list_head nat_entries; /* cached nat entry list (clean) */ 933 spinlock_t nat_list_lock; /* protect clean nat entry list */ 934 unsigned int nat_cnt[MAX_NAT_STATE]; /* the # of cached nat entries */ 935 unsigned int nat_blocks; /* # of nat blocks */ 936 937 /* free node ids management */ 938 struct radix_tree_root free_nid_root;/* root of the free_nid cache */ 939 struct list_head free_nid_list; /* list for free nids excluding preallocated nids */ 940 unsigned int nid_cnt[MAX_NID_STATE]; /* the number of free node id */ 941 spinlock_t nid_list_lock; /* protect nid lists ops */ 942 struct mutex build_lock; /* lock for build free nids */ 943 unsigned char **free_nid_bitmap; 944 unsigned char *nat_block_bitmap; 945 unsigned short *free_nid_count; /* free nid count of NAT block */ 946 947 /* for checkpoint */ 948 char *nat_bitmap; /* NAT bitmap pointer */ 949 950 unsigned int nat_bits_blocks; /* # of nat bits blocks */ 951 unsigned char *nat_bits; /* NAT bits blocks */ 952 unsigned char *full_nat_bits; /* full NAT pages */ 953 unsigned char *empty_nat_bits; /* empty NAT pages */ 954 #ifdef CONFIG_F2FS_CHECK_FS 955 char *nat_bitmap_mir; /* NAT bitmap mirror */ 956 #endif 957 int bitmap_size; /* bitmap size */ 958 }; 959 960 /* 961 * this structure is used as one of function parameters. 962 * all the information are dedicated to a given direct node block determined 963 * by the data offset in a file. 964 */ 965 struct dnode_of_data { 966 struct inode *inode; /* vfs inode pointer */ 967 struct page *inode_page; /* its inode page, NULL is possible */ 968 struct page *node_page; /* cached direct node page */ 969 nid_t nid; /* node id of the direct node block */ 970 unsigned int ofs_in_node; /* data offset in the node page */ 971 bool inode_page_locked; /* inode page is locked or not */ 972 bool node_changed; /* is node block changed */ 973 char cur_level; /* level of hole node page */ 974 char max_level; /* level of current page located */ 975 block_t data_blkaddr; /* block address of the node block */ 976 }; 977 978 static inline void set_new_dnode(struct dnode_of_data *dn, struct inode *inode, 979 struct page *ipage, struct page *npage, nid_t nid) 980 { 981 memset(dn, 0, sizeof(*dn)); 982 dn->inode = inode; 983 dn->inode_page = ipage; 984 dn->node_page = npage; 985 dn->nid = nid; 986 } 987 988 /* 989 * For SIT manager 990 * 991 * By default, there are 6 active log areas across the whole main area. 992 * When considering hot and cold data separation to reduce cleaning overhead, 993 * we split 3 for data logs and 3 for node logs as hot, warm, and cold types, 994 * respectively. 995 * In the current design, you should not change the numbers intentionally. 996 * Instead, as a mount option such as active_logs=x, you can use 2, 4, and 6 997 * logs individually according to the underlying devices. (default: 6) 998 * Just in case, on-disk layout covers maximum 16 logs that consist of 8 for 999 * data and 8 for node logs. 1000 */ 1001 #define NR_CURSEG_DATA_TYPE (3) 1002 #define NR_CURSEG_NODE_TYPE (3) 1003 #define NR_CURSEG_INMEM_TYPE (2) 1004 #define NR_CURSEG_RO_TYPE (2) 1005 #define NR_CURSEG_PERSIST_TYPE (NR_CURSEG_DATA_TYPE + NR_CURSEG_NODE_TYPE) 1006 #define NR_CURSEG_TYPE (NR_CURSEG_INMEM_TYPE + NR_CURSEG_PERSIST_TYPE) 1007 1008 enum { 1009 CURSEG_HOT_DATA = 0, /* directory entry blocks */ 1010 CURSEG_WARM_DATA, /* data blocks */ 1011 CURSEG_COLD_DATA, /* multimedia or GCed data blocks */ 1012 CURSEG_HOT_NODE, /* direct node blocks of directory files */ 1013 CURSEG_WARM_NODE, /* direct node blocks of normal files */ 1014 CURSEG_COLD_NODE, /* indirect node blocks */ 1015 NR_PERSISTENT_LOG, /* number of persistent log */ 1016 CURSEG_COLD_DATA_PINNED = NR_PERSISTENT_LOG, 1017 /* pinned file that needs consecutive block address */ 1018 CURSEG_ALL_DATA_ATGC, /* SSR alloctor in hot/warm/cold data area */ 1019 NO_CHECK_TYPE, /* number of persistent & inmem log */ 1020 }; 1021 1022 struct flush_cmd { 1023 struct completion wait; 1024 struct llist_node llnode; 1025 nid_t ino; 1026 int ret; 1027 }; 1028 1029 struct flush_cmd_control { 1030 struct task_struct *f2fs_issue_flush; /* flush thread */ 1031 wait_queue_head_t flush_wait_queue; /* waiting queue for wake-up */ 1032 atomic_t issued_flush; /* # of issued flushes */ 1033 atomic_t queued_flush; /* # of queued flushes */ 1034 struct llist_head issue_list; /* list for command issue */ 1035 struct llist_node *dispatch_list; /* list for command dispatch */ 1036 }; 1037 1038 struct f2fs_sm_info { 1039 struct sit_info *sit_info; /* whole segment information */ 1040 struct free_segmap_info *free_info; /* free segment information */ 1041 struct dirty_seglist_info *dirty_info; /* dirty segment information */ 1042 struct curseg_info *curseg_array; /* active segment information */ 1043 1044 struct f2fs_rwsem curseg_lock; /* for preventing curseg change */ 1045 1046 block_t seg0_blkaddr; /* block address of 0'th segment */ 1047 block_t main_blkaddr; /* start block address of main area */ 1048 block_t ssa_blkaddr; /* start block address of SSA area */ 1049 1050 unsigned int segment_count; /* total # of segments */ 1051 unsigned int main_segments; /* # of segments in main area */ 1052 unsigned int reserved_segments; /* # of reserved segments */ 1053 unsigned int additional_reserved_segments;/* reserved segs for IO align feature */ 1054 unsigned int ovp_segments; /* # of overprovision segments */ 1055 1056 /* a threshold to reclaim prefree segments */ 1057 unsigned int rec_prefree_segments; 1058 1059 /* for batched trimming */ 1060 unsigned int trim_sections; /* # of sections to trim */ 1061 1062 struct list_head sit_entry_set; /* sit entry set list */ 1063 1064 unsigned int ipu_policy; /* in-place-update policy */ 1065 unsigned int min_ipu_util; /* in-place-update threshold */ 1066 unsigned int min_fsync_blocks; /* threshold for fsync */ 1067 unsigned int min_seq_blocks; /* threshold for sequential blocks */ 1068 unsigned int min_hot_blocks; /* threshold for hot block allocation */ 1069 unsigned int min_ssr_sections; /* threshold to trigger SSR allocation */ 1070 1071 /* for flush command control */ 1072 struct flush_cmd_control *fcc_info; 1073 1074 /* for discard command control */ 1075 struct discard_cmd_control *dcc_info; 1076 }; 1077 1078 /* 1079 * For superblock 1080 */ 1081 /* 1082 * COUNT_TYPE for monitoring 1083 * 1084 * f2fs monitors the number of several block types such as on-writeback, 1085 * dirty dentry blocks, dirty node blocks, and dirty meta blocks. 1086 */ 1087 #define WB_DATA_TYPE(p) (__is_cp_guaranteed(p) ? F2FS_WB_CP_DATA : F2FS_WB_DATA) 1088 enum count_type { 1089 F2FS_DIRTY_DENTS, 1090 F2FS_DIRTY_DATA, 1091 F2FS_DIRTY_QDATA, 1092 F2FS_DIRTY_NODES, 1093 F2FS_DIRTY_META, 1094 F2FS_INMEM_PAGES, 1095 F2FS_DIRTY_IMETA, 1096 F2FS_WB_CP_DATA, 1097 F2FS_WB_DATA, 1098 F2FS_RD_DATA, 1099 F2FS_RD_NODE, 1100 F2FS_RD_META, 1101 F2FS_DIO_WRITE, 1102 F2FS_DIO_READ, 1103 NR_COUNT_TYPE, 1104 }; 1105 1106 /* 1107 * The below are the page types of bios used in submit_bio(). 1108 * The available types are: 1109 * DATA User data pages. It operates as async mode. 1110 * NODE Node pages. It operates as async mode. 1111 * META FS metadata pages such as SIT, NAT, CP. 1112 * NR_PAGE_TYPE The number of page types. 1113 * META_FLUSH Make sure the previous pages are written 1114 * with waiting the bio's completion 1115 * ... Only can be used with META. 1116 */ 1117 #define PAGE_TYPE_OF_BIO(type) ((type) > META ? META : (type)) 1118 enum page_type { 1119 DATA, 1120 NODE, 1121 META, 1122 NR_PAGE_TYPE, 1123 META_FLUSH, 1124 INMEM, /* the below types are used by tracepoints only. */ 1125 INMEM_DROP, 1126 INMEM_INVALIDATE, 1127 INMEM_REVOKE, 1128 IPU, 1129 OPU, 1130 }; 1131 1132 enum temp_type { 1133 HOT = 0, /* must be zero for meta bio */ 1134 WARM, 1135 COLD, 1136 NR_TEMP_TYPE, 1137 }; 1138 1139 enum need_lock_type { 1140 LOCK_REQ = 0, 1141 LOCK_DONE, 1142 LOCK_RETRY, 1143 }; 1144 1145 enum cp_reason_type { 1146 CP_NO_NEEDED, 1147 CP_NON_REGULAR, 1148 CP_COMPRESSED, 1149 CP_HARDLINK, 1150 CP_SB_NEED_CP, 1151 CP_WRONG_PINO, 1152 CP_NO_SPC_ROLL, 1153 CP_NODE_NEED_CP, 1154 CP_FASTBOOT_MODE, 1155 CP_SPEC_LOG_NUM, 1156 CP_RECOVER_DIR, 1157 }; 1158 1159 enum iostat_type { 1160 /* WRITE IO */ 1161 APP_DIRECT_IO, /* app direct write IOs */ 1162 APP_BUFFERED_IO, /* app buffered write IOs */ 1163 APP_WRITE_IO, /* app write IOs */ 1164 APP_MAPPED_IO, /* app mapped IOs */ 1165 FS_DATA_IO, /* data IOs from kworker/fsync/reclaimer */ 1166 FS_NODE_IO, /* node IOs from kworker/fsync/reclaimer */ 1167 FS_META_IO, /* meta IOs from kworker/reclaimer */ 1168 FS_GC_DATA_IO, /* data IOs from forground gc */ 1169 FS_GC_NODE_IO, /* node IOs from forground gc */ 1170 FS_CP_DATA_IO, /* data IOs from checkpoint */ 1171 FS_CP_NODE_IO, /* node IOs from checkpoint */ 1172 FS_CP_META_IO, /* meta IOs from checkpoint */ 1173 1174 /* READ IO */ 1175 APP_DIRECT_READ_IO, /* app direct read IOs */ 1176 APP_BUFFERED_READ_IO, /* app buffered read IOs */ 1177 APP_READ_IO, /* app read IOs */ 1178 APP_MAPPED_READ_IO, /* app mapped read IOs */ 1179 FS_DATA_READ_IO, /* data read IOs */ 1180 FS_GDATA_READ_IO, /* data read IOs from background gc */ 1181 FS_CDATA_READ_IO, /* compressed data read IOs */ 1182 FS_NODE_READ_IO, /* node read IOs */ 1183 FS_META_READ_IO, /* meta read IOs */ 1184 1185 /* other */ 1186 FS_DISCARD, /* discard */ 1187 NR_IO_TYPE, 1188 }; 1189 1190 struct f2fs_io_info { 1191 struct f2fs_sb_info *sbi; /* f2fs_sb_info pointer */ 1192 nid_t ino; /* inode number */ 1193 enum page_type type; /* contains DATA/NODE/META/META_FLUSH */ 1194 enum temp_type temp; /* contains HOT/WARM/COLD */ 1195 int op; /* contains REQ_OP_ */ 1196 int op_flags; /* req_flag_bits */ 1197 block_t new_blkaddr; /* new block address to be written */ 1198 block_t old_blkaddr; /* old block address before Cow */ 1199 struct page *page; /* page to be written */ 1200 struct page *encrypted_page; /* encrypted page */ 1201 struct page *compressed_page; /* compressed page */ 1202 struct list_head list; /* serialize IOs */ 1203 bool submitted; /* indicate IO submission */ 1204 int need_lock; /* indicate we need to lock cp_rwsem */ 1205 bool in_list; /* indicate fio is in io_list */ 1206 bool is_por; /* indicate IO is from recovery or not */ 1207 bool retry; /* need to reallocate block address */ 1208 int compr_blocks; /* # of compressed block addresses */ 1209 bool encrypted; /* indicate file is encrypted */ 1210 enum iostat_type io_type; /* io type */ 1211 struct writeback_control *io_wbc; /* writeback control */ 1212 struct bio **bio; /* bio for ipu */ 1213 sector_t *last_block; /* last block number in bio */ 1214 unsigned char version; /* version of the node */ 1215 }; 1216 1217 struct bio_entry { 1218 struct bio *bio; 1219 struct list_head list; 1220 }; 1221 1222 #define is_read_io(rw) ((rw) == READ) 1223 struct f2fs_bio_info { 1224 struct f2fs_sb_info *sbi; /* f2fs superblock */ 1225 struct bio *bio; /* bios to merge */ 1226 sector_t last_block_in_bio; /* last block number */ 1227 struct f2fs_io_info fio; /* store buffered io info. */ 1228 struct f2fs_rwsem io_rwsem; /* blocking op for bio */ 1229 spinlock_t io_lock; /* serialize DATA/NODE IOs */ 1230 struct list_head io_list; /* track fios */ 1231 struct list_head bio_list; /* bio entry list head */ 1232 struct f2fs_rwsem bio_list_lock; /* lock to protect bio entry list */ 1233 }; 1234 1235 #define FDEV(i) (sbi->devs[i]) 1236 #define RDEV(i) (raw_super->devs[i]) 1237 struct f2fs_dev_info { 1238 struct block_device *bdev; 1239 char path[MAX_PATH_LEN]; 1240 unsigned int total_segments; 1241 block_t start_blk; 1242 block_t end_blk; 1243 #ifdef CONFIG_BLK_DEV_ZONED 1244 unsigned int nr_blkz; /* Total number of zones */ 1245 unsigned long *blkz_seq; /* Bitmap indicating sequential zones */ 1246 block_t *zone_capacity_blocks; /* Array of zone capacity in blks */ 1247 #endif 1248 }; 1249 1250 enum inode_type { 1251 DIR_INODE, /* for dirty dir inode */ 1252 FILE_INODE, /* for dirty regular/symlink inode */ 1253 DIRTY_META, /* for all dirtied inode metadata */ 1254 ATOMIC_FILE, /* for all atomic files */ 1255 NR_INODE_TYPE, 1256 }; 1257 1258 /* for inner inode cache management */ 1259 struct inode_management { 1260 struct radix_tree_root ino_root; /* ino entry array */ 1261 spinlock_t ino_lock; /* for ino entry lock */ 1262 struct list_head ino_list; /* inode list head */ 1263 unsigned long ino_num; /* number of entries */ 1264 }; 1265 1266 /* for GC_AT */ 1267 struct atgc_management { 1268 bool atgc_enabled; /* ATGC is enabled or not */ 1269 struct rb_root_cached root; /* root of victim rb-tree */ 1270 struct list_head victim_list; /* linked with all victim entries */ 1271 unsigned int victim_count; /* victim count in rb-tree */ 1272 unsigned int candidate_ratio; /* candidate ratio */ 1273 unsigned int max_candidate_count; /* max candidate count */ 1274 unsigned int age_weight; /* age weight, vblock_weight = 100 - age_weight */ 1275 unsigned long long age_threshold; /* age threshold */ 1276 }; 1277 1278 /* For s_flag in struct f2fs_sb_info */ 1279 enum { 1280 SBI_IS_DIRTY, /* dirty flag for checkpoint */ 1281 SBI_IS_CLOSE, /* specify unmounting */ 1282 SBI_NEED_FSCK, /* need fsck.f2fs to fix */ 1283 SBI_POR_DOING, /* recovery is doing or not */ 1284 SBI_NEED_SB_WRITE, /* need to recover superblock */ 1285 SBI_NEED_CP, /* need to checkpoint */ 1286 SBI_IS_SHUTDOWN, /* shutdown by ioctl */ 1287 SBI_IS_RECOVERED, /* recovered orphan/data */ 1288 SBI_CP_DISABLED, /* CP was disabled last mount */ 1289 SBI_CP_DISABLED_QUICK, /* CP was disabled quickly */ 1290 SBI_QUOTA_NEED_FLUSH, /* need to flush quota info in CP */ 1291 SBI_QUOTA_SKIP_FLUSH, /* skip flushing quota in current CP */ 1292 SBI_QUOTA_NEED_REPAIR, /* quota file may be corrupted */ 1293 SBI_IS_RESIZEFS, /* resizefs is in process */ 1294 }; 1295 1296 enum { 1297 CP_TIME, 1298 REQ_TIME, 1299 DISCARD_TIME, 1300 GC_TIME, 1301 DISABLE_TIME, 1302 UMOUNT_DISCARD_TIMEOUT, 1303 MAX_TIME, 1304 }; 1305 1306 enum { 1307 GC_NORMAL, 1308 GC_IDLE_CB, 1309 GC_IDLE_GREEDY, 1310 GC_IDLE_AT, 1311 GC_URGENT_HIGH, 1312 GC_URGENT_LOW, 1313 MAX_GC_MODE, 1314 }; 1315 1316 enum { 1317 BGGC_MODE_ON, /* background gc is on */ 1318 BGGC_MODE_OFF, /* background gc is off */ 1319 BGGC_MODE_SYNC, /* 1320 * background gc is on, migrating blocks 1321 * like foreground gc 1322 */ 1323 }; 1324 1325 enum { 1326 FS_MODE_ADAPTIVE, /* use both lfs/ssr allocation */ 1327 FS_MODE_LFS, /* use lfs allocation only */ 1328 FS_MODE_FRAGMENT_SEG, /* segment fragmentation mode */ 1329 FS_MODE_FRAGMENT_BLK, /* block fragmentation mode */ 1330 }; 1331 1332 enum { 1333 WHINT_MODE_OFF, /* not pass down write hints */ 1334 WHINT_MODE_USER, /* try to pass down hints given by users */ 1335 WHINT_MODE_FS, /* pass down hints with F2FS policy */ 1336 }; 1337 1338 enum { 1339 ALLOC_MODE_DEFAULT, /* stay default */ 1340 ALLOC_MODE_REUSE, /* reuse segments as much as possible */ 1341 }; 1342 1343 enum fsync_mode { 1344 FSYNC_MODE_POSIX, /* fsync follows posix semantics */ 1345 FSYNC_MODE_STRICT, /* fsync behaves in line with ext4 */ 1346 FSYNC_MODE_NOBARRIER, /* fsync behaves nobarrier based on posix */ 1347 }; 1348 1349 enum { 1350 COMPR_MODE_FS, /* 1351 * automatically compress compression 1352 * enabled files 1353 */ 1354 COMPR_MODE_USER, /* 1355 * automatical compression is disabled. 1356 * user can control the file compression 1357 * using ioctls 1358 */ 1359 }; 1360 1361 enum { 1362 DISCARD_UNIT_BLOCK, /* basic discard unit is block */ 1363 DISCARD_UNIT_SEGMENT, /* basic discard unit is segment */ 1364 DISCARD_UNIT_SECTION, /* basic discard unit is section */ 1365 }; 1366 1367 static inline int f2fs_test_bit(unsigned int nr, char *addr); 1368 static inline void f2fs_set_bit(unsigned int nr, char *addr); 1369 static inline void f2fs_clear_bit(unsigned int nr, char *addr); 1370 1371 /* 1372 * Layout of f2fs page.private: 1373 * 1374 * Layout A: lowest bit should be 1 1375 * | bit0 = 1 | bit1 | bit2 | ... | bit MAX | private data .... | 1376 * bit 0 PAGE_PRIVATE_NOT_POINTER 1377 * bit 1 PAGE_PRIVATE_ATOMIC_WRITE 1378 * bit 2 PAGE_PRIVATE_DUMMY_WRITE 1379 * bit 3 PAGE_PRIVATE_ONGOING_MIGRATION 1380 * bit 4 PAGE_PRIVATE_INLINE_INODE 1381 * bit 5 PAGE_PRIVATE_REF_RESOURCE 1382 * bit 6- f2fs private data 1383 * 1384 * Layout B: lowest bit should be 0 1385 * page.private is a wrapped pointer. 1386 */ 1387 enum { 1388 PAGE_PRIVATE_NOT_POINTER, /* private contains non-pointer data */ 1389 PAGE_PRIVATE_ATOMIC_WRITE, /* data page from atomic write path */ 1390 PAGE_PRIVATE_DUMMY_WRITE, /* data page for padding aligned IO */ 1391 PAGE_PRIVATE_ONGOING_MIGRATION, /* data page which is on-going migrating */ 1392 PAGE_PRIVATE_INLINE_INODE, /* inode page contains inline data */ 1393 PAGE_PRIVATE_REF_RESOURCE, /* dirty page has referenced resources */ 1394 PAGE_PRIVATE_MAX 1395 }; 1396 1397 #define PAGE_PRIVATE_GET_FUNC(name, flagname) \ 1398 static inline bool page_private_##name(struct page *page) \ 1399 { \ 1400 return PagePrivate(page) && \ 1401 test_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)) && \ 1402 test_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \ 1403 } 1404 1405 #define PAGE_PRIVATE_SET_FUNC(name, flagname) \ 1406 static inline void set_page_private_##name(struct page *page) \ 1407 { \ 1408 if (!PagePrivate(page)) { \ 1409 get_page(page); \ 1410 SetPagePrivate(page); \ 1411 set_page_private(page, 0); \ 1412 } \ 1413 set_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)); \ 1414 set_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \ 1415 } 1416 1417 #define PAGE_PRIVATE_CLEAR_FUNC(name, flagname) \ 1418 static inline void clear_page_private_##name(struct page *page) \ 1419 { \ 1420 clear_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \ 1421 if (page_private(page) == 1 << PAGE_PRIVATE_NOT_POINTER) { \ 1422 set_page_private(page, 0); \ 1423 if (PagePrivate(page)) { \ 1424 ClearPagePrivate(page); \ 1425 put_page(page); \ 1426 }\ 1427 } \ 1428 } 1429 1430 PAGE_PRIVATE_GET_FUNC(nonpointer, NOT_POINTER); 1431 PAGE_PRIVATE_GET_FUNC(reference, REF_RESOURCE); 1432 PAGE_PRIVATE_GET_FUNC(inline, INLINE_INODE); 1433 PAGE_PRIVATE_GET_FUNC(gcing, ONGOING_MIGRATION); 1434 PAGE_PRIVATE_GET_FUNC(atomic, ATOMIC_WRITE); 1435 PAGE_PRIVATE_GET_FUNC(dummy, DUMMY_WRITE); 1436 1437 PAGE_PRIVATE_SET_FUNC(reference, REF_RESOURCE); 1438 PAGE_PRIVATE_SET_FUNC(inline, INLINE_INODE); 1439 PAGE_PRIVATE_SET_FUNC(gcing, ONGOING_MIGRATION); 1440 PAGE_PRIVATE_SET_FUNC(atomic, ATOMIC_WRITE); 1441 PAGE_PRIVATE_SET_FUNC(dummy, DUMMY_WRITE); 1442 1443 PAGE_PRIVATE_CLEAR_FUNC(reference, REF_RESOURCE); 1444 PAGE_PRIVATE_CLEAR_FUNC(inline, INLINE_INODE); 1445 PAGE_PRIVATE_CLEAR_FUNC(gcing, ONGOING_MIGRATION); 1446 PAGE_PRIVATE_CLEAR_FUNC(atomic, ATOMIC_WRITE); 1447 PAGE_PRIVATE_CLEAR_FUNC(dummy, DUMMY_WRITE); 1448 1449 static inline unsigned long get_page_private_data(struct page *page) 1450 { 1451 unsigned long data = page_private(page); 1452 1453 if (!test_bit(PAGE_PRIVATE_NOT_POINTER, &data)) 1454 return 0; 1455 return data >> PAGE_PRIVATE_MAX; 1456 } 1457 1458 static inline void set_page_private_data(struct page *page, unsigned long data) 1459 { 1460 if (!PagePrivate(page)) { 1461 get_page(page); 1462 SetPagePrivate(page); 1463 set_page_private(page, 0); 1464 } 1465 set_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)); 1466 page_private(page) |= data << PAGE_PRIVATE_MAX; 1467 } 1468 1469 static inline void clear_page_private_data(struct page *page) 1470 { 1471 page_private(page) &= (1 << PAGE_PRIVATE_MAX) - 1; 1472 if (page_private(page) == 1 << PAGE_PRIVATE_NOT_POINTER) { 1473 set_page_private(page, 0); 1474 if (PagePrivate(page)) { 1475 ClearPagePrivate(page); 1476 put_page(page); 1477 } 1478 } 1479 } 1480 1481 /* For compression */ 1482 enum compress_algorithm_type { 1483 COMPRESS_LZO, 1484 COMPRESS_LZ4, 1485 COMPRESS_ZSTD, 1486 COMPRESS_LZORLE, 1487 COMPRESS_MAX, 1488 }; 1489 1490 enum compress_flag { 1491 COMPRESS_CHKSUM, 1492 COMPRESS_MAX_FLAG, 1493 }; 1494 1495 #define COMPRESS_WATERMARK 20 1496 #define COMPRESS_PERCENT 20 1497 1498 #define COMPRESS_DATA_RESERVED_SIZE 4 1499 struct compress_data { 1500 __le32 clen; /* compressed data size */ 1501 __le32 chksum; /* compressed data chksum */ 1502 __le32 reserved[COMPRESS_DATA_RESERVED_SIZE]; /* reserved */ 1503 u8 cdata[]; /* compressed data */ 1504 }; 1505 1506 #define COMPRESS_HEADER_SIZE (sizeof(struct compress_data)) 1507 1508 #define F2FS_COMPRESSED_PAGE_MAGIC 0xF5F2C000 1509 1510 #define COMPRESS_LEVEL_OFFSET 8 1511 1512 /* compress context */ 1513 struct compress_ctx { 1514 struct inode *inode; /* inode the context belong to */ 1515 pgoff_t cluster_idx; /* cluster index number */ 1516 unsigned int cluster_size; /* page count in cluster */ 1517 unsigned int log_cluster_size; /* log of cluster size */ 1518 struct page **rpages; /* pages store raw data in cluster */ 1519 unsigned int nr_rpages; /* total page number in rpages */ 1520 struct page **cpages; /* pages store compressed data in cluster */ 1521 unsigned int nr_cpages; /* total page number in cpages */ 1522 unsigned int valid_nr_cpages; /* valid page number in cpages */ 1523 void *rbuf; /* virtual mapped address on rpages */ 1524 struct compress_data *cbuf; /* virtual mapped address on cpages */ 1525 size_t rlen; /* valid data length in rbuf */ 1526 size_t clen; /* valid data length in cbuf */ 1527 void *private; /* payload buffer for specified compression algorithm */ 1528 void *private2; /* extra payload buffer */ 1529 }; 1530 1531 /* compress context for write IO path */ 1532 struct compress_io_ctx { 1533 u32 magic; /* magic number to indicate page is compressed */ 1534 struct inode *inode; /* inode the context belong to */ 1535 struct page **rpages; /* pages store raw data in cluster */ 1536 unsigned int nr_rpages; /* total page number in rpages */ 1537 atomic_t pending_pages; /* in-flight compressed page count */ 1538 }; 1539 1540 /* Context for decompressing one cluster on the read IO path */ 1541 struct decompress_io_ctx { 1542 u32 magic; /* magic number to indicate page is compressed */ 1543 struct inode *inode; /* inode the context belong to */ 1544 pgoff_t cluster_idx; /* cluster index number */ 1545 unsigned int cluster_size; /* page count in cluster */ 1546 unsigned int log_cluster_size; /* log of cluster size */ 1547 struct page **rpages; /* pages store raw data in cluster */ 1548 unsigned int nr_rpages; /* total page number in rpages */ 1549 struct page **cpages; /* pages store compressed data in cluster */ 1550 unsigned int nr_cpages; /* total page number in cpages */ 1551 struct page **tpages; /* temp pages to pad holes in cluster */ 1552 void *rbuf; /* virtual mapped address on rpages */ 1553 struct compress_data *cbuf; /* virtual mapped address on cpages */ 1554 size_t rlen; /* valid data length in rbuf */ 1555 size_t clen; /* valid data length in cbuf */ 1556 1557 /* 1558 * The number of compressed pages remaining to be read in this cluster. 1559 * This is initially nr_cpages. It is decremented by 1 each time a page 1560 * has been read (or failed to be read). When it reaches 0, the cluster 1561 * is decompressed (or an error is reported). 1562 * 1563 * If an error occurs before all the pages have been submitted for I/O, 1564 * then this will never reach 0. In this case the I/O submitter is 1565 * responsible for calling f2fs_decompress_end_io() instead. 1566 */ 1567 atomic_t remaining_pages; 1568 1569 /* 1570 * Number of references to this decompress_io_ctx. 1571 * 1572 * One reference is held for I/O completion. This reference is dropped 1573 * after the pagecache pages are updated and unlocked -- either after 1574 * decompression (and verity if enabled), or after an error. 1575 * 1576 * In addition, each compressed page holds a reference while it is in a 1577 * bio. These references are necessary prevent compressed pages from 1578 * being freed while they are still in a bio. 1579 */ 1580 refcount_t refcnt; 1581 1582 bool failed; /* IO error occurred before decompression? */ 1583 bool need_verity; /* need fs-verity verification after decompression? */ 1584 void *private; /* payload buffer for specified decompression algorithm */ 1585 void *private2; /* extra payload buffer */ 1586 struct work_struct verity_work; /* work to verify the decompressed pages */ 1587 }; 1588 1589 #define NULL_CLUSTER ((unsigned int)(~0)) 1590 #define MIN_COMPRESS_LOG_SIZE 2 1591 #define MAX_COMPRESS_LOG_SIZE 8 1592 #define MAX_COMPRESS_WINDOW_SIZE(log_size) ((PAGE_SIZE) << (log_size)) 1593 1594 struct f2fs_sb_info { 1595 struct super_block *sb; /* pointer to VFS super block */ 1596 struct proc_dir_entry *s_proc; /* proc entry */ 1597 struct f2fs_super_block *raw_super; /* raw super block pointer */ 1598 struct f2fs_rwsem sb_lock; /* lock for raw super block */ 1599 int valid_super_block; /* valid super block no */ 1600 unsigned long s_flag; /* flags for sbi */ 1601 struct mutex writepages; /* mutex for writepages() */ 1602 1603 #ifdef CONFIG_BLK_DEV_ZONED 1604 unsigned int blocks_per_blkz; /* F2FS blocks per zone */ 1605 unsigned int log_blocks_per_blkz; /* log2 F2FS blocks per zone */ 1606 #endif 1607 1608 /* for node-related operations */ 1609 struct f2fs_nm_info *nm_info; /* node manager */ 1610 struct inode *node_inode; /* cache node blocks */ 1611 1612 /* for segment-related operations */ 1613 struct f2fs_sm_info *sm_info; /* segment manager */ 1614 1615 /* for bio operations */ 1616 struct f2fs_bio_info *write_io[NR_PAGE_TYPE]; /* for write bios */ 1617 /* keep migration IO order for LFS mode */ 1618 struct f2fs_rwsem io_order_lock; 1619 mempool_t *write_io_dummy; /* Dummy pages */ 1620 pgoff_t metapage_eio_ofs; /* EIO page offset */ 1621 int metapage_eio_cnt; /* EIO count */ 1622 1623 /* for checkpoint */ 1624 struct f2fs_checkpoint *ckpt; /* raw checkpoint pointer */ 1625 int cur_cp_pack; /* remain current cp pack */ 1626 spinlock_t cp_lock; /* for flag in ckpt */ 1627 struct inode *meta_inode; /* cache meta blocks */ 1628 struct f2fs_rwsem cp_global_sem; /* checkpoint procedure lock */ 1629 struct f2fs_rwsem cp_rwsem; /* blocking FS operations */ 1630 struct f2fs_rwsem node_write; /* locking node writes */ 1631 struct f2fs_rwsem node_change; /* locking node change */ 1632 wait_queue_head_t cp_wait; 1633 unsigned long last_time[MAX_TIME]; /* to store time in jiffies */ 1634 long interval_time[MAX_TIME]; /* to store thresholds */ 1635 struct ckpt_req_control cprc_info; /* for checkpoint request control */ 1636 1637 struct inode_management im[MAX_INO_ENTRY]; /* manage inode cache */ 1638 1639 spinlock_t fsync_node_lock; /* for node entry lock */ 1640 struct list_head fsync_node_list; /* node list head */ 1641 unsigned int fsync_seg_id; /* sequence id */ 1642 unsigned int fsync_node_num; /* number of node entries */ 1643 1644 /* for orphan inode, use 0'th array */ 1645 unsigned int max_orphans; /* max orphan inodes */ 1646 1647 /* for inode management */ 1648 struct list_head inode_list[NR_INODE_TYPE]; /* dirty inode list */ 1649 spinlock_t inode_lock[NR_INODE_TYPE]; /* for dirty inode list lock */ 1650 struct mutex flush_lock; /* for flush exclusion */ 1651 1652 /* for extent tree cache */ 1653 struct radix_tree_root extent_tree_root;/* cache extent cache entries */ 1654 struct mutex extent_tree_lock; /* locking extent radix tree */ 1655 struct list_head extent_list; /* lru list for shrinker */ 1656 spinlock_t extent_lock; /* locking extent lru list */ 1657 atomic_t total_ext_tree; /* extent tree count */ 1658 struct list_head zombie_list; /* extent zombie tree list */ 1659 atomic_t total_zombie_tree; /* extent zombie tree count */ 1660 atomic_t total_ext_node; /* extent info count */ 1661 1662 /* basic filesystem units */ 1663 unsigned int log_sectors_per_block; /* log2 sectors per block */ 1664 unsigned int log_blocksize; /* log2 block size */ 1665 unsigned int blocksize; /* block size */ 1666 unsigned int root_ino_num; /* root inode number*/ 1667 unsigned int node_ino_num; /* node inode number*/ 1668 unsigned int meta_ino_num; /* meta inode number*/ 1669 unsigned int log_blocks_per_seg; /* log2 blocks per segment */ 1670 unsigned int blocks_per_seg; /* blocks per segment */ 1671 unsigned int segs_per_sec; /* segments per section */ 1672 unsigned int secs_per_zone; /* sections per zone */ 1673 unsigned int total_sections; /* total section count */ 1674 unsigned int total_node_count; /* total node block count */ 1675 unsigned int total_valid_node_count; /* valid node block count */ 1676 int dir_level; /* directory level */ 1677 int readdir_ra; /* readahead inode in readdir */ 1678 u64 max_io_bytes; /* max io bytes to merge IOs */ 1679 1680 block_t user_block_count; /* # of user blocks */ 1681 block_t total_valid_block_count; /* # of valid blocks */ 1682 block_t discard_blks; /* discard command candidats */ 1683 block_t last_valid_block_count; /* for recovery */ 1684 block_t reserved_blocks; /* configurable reserved blocks */ 1685 block_t current_reserved_blocks; /* current reserved blocks */ 1686 1687 /* Additional tracking for no checkpoint mode */ 1688 block_t unusable_block_count; /* # of blocks saved by last cp */ 1689 1690 unsigned int nquota_files; /* # of quota sysfile */ 1691 struct f2fs_rwsem quota_sem; /* blocking cp for flags */ 1692 1693 /* # of pages, see count_type */ 1694 atomic_t nr_pages[NR_COUNT_TYPE]; 1695 /* # of allocated blocks */ 1696 struct percpu_counter alloc_valid_block_count; 1697 /* # of node block writes as roll forward recovery */ 1698 struct percpu_counter rf_node_block_count; 1699 1700 /* writeback control */ 1701 atomic_t wb_sync_req[META]; /* count # of WB_SYNC threads */ 1702 1703 /* valid inode count */ 1704 struct percpu_counter total_valid_inode_count; 1705 1706 struct f2fs_mount_info mount_opt; /* mount options */ 1707 1708 /* for cleaning operations */ 1709 struct f2fs_rwsem gc_lock; /* 1710 * semaphore for GC, avoid 1711 * race between GC and GC or CP 1712 */ 1713 struct f2fs_gc_kthread *gc_thread; /* GC thread */ 1714 struct atgc_management am; /* atgc management */ 1715 unsigned int cur_victim_sec; /* current victim section num */ 1716 unsigned int gc_mode; /* current GC state */ 1717 unsigned int next_victim_seg[2]; /* next segment in victim section */ 1718 spinlock_t gc_urgent_high_lock; 1719 bool gc_urgent_high_limited; /* indicates having limited trial count */ 1720 unsigned int gc_urgent_high_remaining; /* remaining trial count for GC_URGENT_HIGH */ 1721 1722 /* for skip statistic */ 1723 unsigned int atomic_files; /* # of opened atomic file */ 1724 unsigned long long skipped_atomic_files[2]; /* FG_GC and BG_GC */ 1725 unsigned long long skipped_gc_rwsem; /* FG_GC only */ 1726 1727 /* threshold for gc trials on pinned files */ 1728 u64 gc_pin_file_threshold; 1729 struct f2fs_rwsem pin_sem; 1730 1731 /* maximum # of trials to find a victim segment for SSR and GC */ 1732 unsigned int max_victim_search; 1733 /* migration granularity of garbage collection, unit: segment */ 1734 unsigned int migration_granularity; 1735 1736 /* 1737 * for stat information. 1738 * one is for the LFS mode, and the other is for the SSR mode. 1739 */ 1740 #ifdef CONFIG_F2FS_STAT_FS 1741 struct f2fs_stat_info *stat_info; /* FS status information */ 1742 atomic_t meta_count[META_MAX]; /* # of meta blocks */ 1743 unsigned int segment_count[2]; /* # of allocated segments */ 1744 unsigned int block_count[2]; /* # of allocated blocks */ 1745 atomic_t inplace_count; /* # of inplace update */ 1746 atomic64_t total_hit_ext; /* # of lookup extent cache */ 1747 atomic64_t read_hit_rbtree; /* # of hit rbtree extent node */ 1748 atomic64_t read_hit_largest; /* # of hit largest extent node */ 1749 atomic64_t read_hit_cached; /* # of hit cached extent node */ 1750 atomic_t inline_xattr; /* # of inline_xattr inodes */ 1751 atomic_t inline_inode; /* # of inline_data inodes */ 1752 atomic_t inline_dir; /* # of inline_dentry inodes */ 1753 atomic_t compr_inode; /* # of compressed inodes */ 1754 atomic64_t compr_blocks; /* # of compressed blocks */ 1755 atomic_t vw_cnt; /* # of volatile writes */ 1756 atomic_t max_aw_cnt; /* max # of atomic writes */ 1757 atomic_t max_vw_cnt; /* max # of volatile writes */ 1758 unsigned int io_skip_bggc; /* skip background gc for in-flight IO */ 1759 unsigned int other_skip_bggc; /* skip background gc for other reasons */ 1760 unsigned int ndirty_inode[NR_INODE_TYPE]; /* # of dirty inodes */ 1761 #endif 1762 spinlock_t stat_lock; /* lock for stat operations */ 1763 1764 /* to attach REQ_META|REQ_FUA flags */ 1765 unsigned int data_io_flag; 1766 unsigned int node_io_flag; 1767 1768 /* For sysfs suppport */ 1769 struct kobject s_kobj; /* /sys/fs/f2fs/<devname> */ 1770 struct completion s_kobj_unregister; 1771 1772 struct kobject s_stat_kobj; /* /sys/fs/f2fs/<devname>/stat */ 1773 struct completion s_stat_kobj_unregister; 1774 1775 struct kobject s_feature_list_kobj; /* /sys/fs/f2fs/<devname>/feature_list */ 1776 struct completion s_feature_list_kobj_unregister; 1777 1778 /* For shrinker support */ 1779 struct list_head s_list; 1780 struct mutex umount_mutex; 1781 unsigned int shrinker_run_no; 1782 1783 /* For multi devices */ 1784 int s_ndevs; /* number of devices */ 1785 struct f2fs_dev_info *devs; /* for device list */ 1786 unsigned int dirty_device; /* for checkpoint data flush */ 1787 spinlock_t dev_lock; /* protect dirty_device */ 1788 bool aligned_blksize; /* all devices has the same logical blksize */ 1789 1790 /* For write statistics */ 1791 u64 sectors_written_start; 1792 u64 kbytes_written; 1793 1794 /* Reference to checksum algorithm driver via cryptoapi */ 1795 struct crypto_shash *s_chksum_driver; 1796 1797 /* Precomputed FS UUID checksum for seeding other checksums */ 1798 __u32 s_chksum_seed; 1799 1800 struct workqueue_struct *post_read_wq; /* post read workqueue */ 1801 1802 struct kmem_cache *inline_xattr_slab; /* inline xattr entry */ 1803 unsigned int inline_xattr_slab_size; /* default inline xattr slab size */ 1804 1805 /* For reclaimed segs statistics per each GC mode */ 1806 unsigned int gc_segment_mode; /* GC state for reclaimed segments */ 1807 unsigned int gc_reclaimed_segs[MAX_GC_MODE]; /* Reclaimed segs for each mode */ 1808 1809 unsigned long seq_file_ra_mul; /* multiplier for ra_pages of seq. files in fadvise */ 1810 1811 int max_fragment_chunk; /* max chunk size for block fragmentation mode */ 1812 int max_fragment_hole; /* max hole size for block fragmentation mode */ 1813 1814 #ifdef CONFIG_F2FS_FS_COMPRESSION 1815 struct kmem_cache *page_array_slab; /* page array entry */ 1816 unsigned int page_array_slab_size; /* default page array slab size */ 1817 1818 /* For runtime compression statistics */ 1819 u64 compr_written_block; 1820 u64 compr_saved_block; 1821 u32 compr_new_inode; 1822 1823 /* For compressed block cache */ 1824 struct inode *compress_inode; /* cache compressed blocks */ 1825 unsigned int compress_percent; /* cache page percentage */ 1826 unsigned int compress_watermark; /* cache page watermark */ 1827 atomic_t compress_page_hit; /* cache hit count */ 1828 #endif 1829 1830 #ifdef CONFIG_F2FS_IOSTAT 1831 /* For app/fs IO statistics */ 1832 spinlock_t iostat_lock; 1833 unsigned long long rw_iostat[NR_IO_TYPE]; 1834 unsigned long long prev_rw_iostat[NR_IO_TYPE]; 1835 bool iostat_enable; 1836 unsigned long iostat_next_period; 1837 unsigned int iostat_period_ms; 1838 1839 /* For io latency related statistics info in one iostat period */ 1840 spinlock_t iostat_lat_lock; 1841 struct iostat_lat_info *iostat_io_lat; 1842 #endif 1843 }; 1844 1845 #ifdef CONFIG_F2FS_FAULT_INJECTION 1846 #define f2fs_show_injection_info(sbi, type) \ 1847 printk_ratelimited("%sF2FS-fs (%s) : inject %s in %s of %pS\n", \ 1848 KERN_INFO, sbi->sb->s_id, \ 1849 f2fs_fault_name[type], \ 1850 __func__, __builtin_return_address(0)) 1851 static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type) 1852 { 1853 struct f2fs_fault_info *ffi = &F2FS_OPTION(sbi).fault_info; 1854 1855 if (!ffi->inject_rate) 1856 return false; 1857 1858 if (!IS_FAULT_SET(ffi, type)) 1859 return false; 1860 1861 atomic_inc(&ffi->inject_ops); 1862 if (atomic_read(&ffi->inject_ops) >= ffi->inject_rate) { 1863 atomic_set(&ffi->inject_ops, 0); 1864 return true; 1865 } 1866 return false; 1867 } 1868 #else 1869 #define f2fs_show_injection_info(sbi, type) do { } while (0) 1870 static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type) 1871 { 1872 return false; 1873 } 1874 #endif 1875 1876 /* 1877 * Test if the mounted volume is a multi-device volume. 1878 * - For a single regular disk volume, sbi->s_ndevs is 0. 1879 * - For a single zoned disk volume, sbi->s_ndevs is 1. 1880 * - For a multi-device volume, sbi->s_ndevs is always 2 or more. 1881 */ 1882 static inline bool f2fs_is_multi_device(struct f2fs_sb_info *sbi) 1883 { 1884 return sbi->s_ndevs > 1; 1885 } 1886 1887 static inline void f2fs_update_time(struct f2fs_sb_info *sbi, int type) 1888 { 1889 unsigned long now = jiffies; 1890 1891 sbi->last_time[type] = now; 1892 1893 /* DISCARD_TIME and GC_TIME are based on REQ_TIME */ 1894 if (type == REQ_TIME) { 1895 sbi->last_time[DISCARD_TIME] = now; 1896 sbi->last_time[GC_TIME] = now; 1897 } 1898 } 1899 1900 static inline bool f2fs_time_over(struct f2fs_sb_info *sbi, int type) 1901 { 1902 unsigned long interval = sbi->interval_time[type] * HZ; 1903 1904 return time_after(jiffies, sbi->last_time[type] + interval); 1905 } 1906 1907 static inline unsigned int f2fs_time_to_wait(struct f2fs_sb_info *sbi, 1908 int type) 1909 { 1910 unsigned long interval = sbi->interval_time[type] * HZ; 1911 unsigned int wait_ms = 0; 1912 long delta; 1913 1914 delta = (sbi->last_time[type] + interval) - jiffies; 1915 if (delta > 0) 1916 wait_ms = jiffies_to_msecs(delta); 1917 1918 return wait_ms; 1919 } 1920 1921 /* 1922 * Inline functions 1923 */ 1924 static inline u32 __f2fs_crc32(struct f2fs_sb_info *sbi, u32 crc, 1925 const void *address, unsigned int length) 1926 { 1927 struct { 1928 struct shash_desc shash; 1929 char ctx[4]; 1930 } desc; 1931 int err; 1932 1933 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver) != sizeof(desc.ctx)); 1934 1935 desc.shash.tfm = sbi->s_chksum_driver; 1936 *(u32 *)desc.ctx = crc; 1937 1938 err = crypto_shash_update(&desc.shash, address, length); 1939 BUG_ON(err); 1940 1941 return *(u32 *)desc.ctx; 1942 } 1943 1944 static inline u32 f2fs_crc32(struct f2fs_sb_info *sbi, const void *address, 1945 unsigned int length) 1946 { 1947 return __f2fs_crc32(sbi, F2FS_SUPER_MAGIC, address, length); 1948 } 1949 1950 static inline bool f2fs_crc_valid(struct f2fs_sb_info *sbi, __u32 blk_crc, 1951 void *buf, size_t buf_size) 1952 { 1953 return f2fs_crc32(sbi, buf, buf_size) == blk_crc; 1954 } 1955 1956 static inline u32 f2fs_chksum(struct f2fs_sb_info *sbi, u32 crc, 1957 const void *address, unsigned int length) 1958 { 1959 return __f2fs_crc32(sbi, crc, address, length); 1960 } 1961 1962 static inline struct f2fs_inode_info *F2FS_I(struct inode *inode) 1963 { 1964 return container_of(inode, struct f2fs_inode_info, vfs_inode); 1965 } 1966 1967 static inline struct f2fs_sb_info *F2FS_SB(struct super_block *sb) 1968 { 1969 return sb->s_fs_info; 1970 } 1971 1972 static inline struct f2fs_sb_info *F2FS_I_SB(struct inode *inode) 1973 { 1974 return F2FS_SB(inode->i_sb); 1975 } 1976 1977 static inline struct f2fs_sb_info *F2FS_M_SB(struct address_space *mapping) 1978 { 1979 return F2FS_I_SB(mapping->host); 1980 } 1981 1982 static inline struct f2fs_sb_info *F2FS_P_SB(struct page *page) 1983 { 1984 return F2FS_M_SB(page_file_mapping(page)); 1985 } 1986 1987 static inline struct f2fs_super_block *F2FS_RAW_SUPER(struct f2fs_sb_info *sbi) 1988 { 1989 return (struct f2fs_super_block *)(sbi->raw_super); 1990 } 1991 1992 static inline struct f2fs_checkpoint *F2FS_CKPT(struct f2fs_sb_info *sbi) 1993 { 1994 return (struct f2fs_checkpoint *)(sbi->ckpt); 1995 } 1996 1997 static inline struct f2fs_node *F2FS_NODE(struct page *page) 1998 { 1999 return (struct f2fs_node *)page_address(page); 2000 } 2001 2002 static inline struct f2fs_inode *F2FS_INODE(struct page *page) 2003 { 2004 return &((struct f2fs_node *)page_address(page))->i; 2005 } 2006 2007 static inline struct f2fs_nm_info *NM_I(struct f2fs_sb_info *sbi) 2008 { 2009 return (struct f2fs_nm_info *)(sbi->nm_info); 2010 } 2011 2012 static inline struct f2fs_sm_info *SM_I(struct f2fs_sb_info *sbi) 2013 { 2014 return (struct f2fs_sm_info *)(sbi->sm_info); 2015 } 2016 2017 static inline struct sit_info *SIT_I(struct f2fs_sb_info *sbi) 2018 { 2019 return (struct sit_info *)(SM_I(sbi)->sit_info); 2020 } 2021 2022 static inline struct free_segmap_info *FREE_I(struct f2fs_sb_info *sbi) 2023 { 2024 return (struct free_segmap_info *)(SM_I(sbi)->free_info); 2025 } 2026 2027 static inline struct dirty_seglist_info *DIRTY_I(struct f2fs_sb_info *sbi) 2028 { 2029 return (struct dirty_seglist_info *)(SM_I(sbi)->dirty_info); 2030 } 2031 2032 static inline struct address_space *META_MAPPING(struct f2fs_sb_info *sbi) 2033 { 2034 return sbi->meta_inode->i_mapping; 2035 } 2036 2037 static inline struct address_space *NODE_MAPPING(struct f2fs_sb_info *sbi) 2038 { 2039 return sbi->node_inode->i_mapping; 2040 } 2041 2042 static inline bool is_sbi_flag_set(struct f2fs_sb_info *sbi, unsigned int type) 2043 { 2044 return test_bit(type, &sbi->s_flag); 2045 } 2046 2047 static inline void set_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type) 2048 { 2049 set_bit(type, &sbi->s_flag); 2050 } 2051 2052 static inline void clear_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type) 2053 { 2054 clear_bit(type, &sbi->s_flag); 2055 } 2056 2057 static inline unsigned long long cur_cp_version(struct f2fs_checkpoint *cp) 2058 { 2059 return le64_to_cpu(cp->checkpoint_ver); 2060 } 2061 2062 static inline unsigned long f2fs_qf_ino(struct super_block *sb, int type) 2063 { 2064 if (type < F2FS_MAX_QUOTAS) 2065 return le32_to_cpu(F2FS_SB(sb)->raw_super->qf_ino[type]); 2066 return 0; 2067 } 2068 2069 static inline __u64 cur_cp_crc(struct f2fs_checkpoint *cp) 2070 { 2071 size_t crc_offset = le32_to_cpu(cp->checksum_offset); 2072 return le32_to_cpu(*((__le32 *)((unsigned char *)cp + crc_offset))); 2073 } 2074 2075 static inline bool __is_set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 2076 { 2077 unsigned int ckpt_flags = le32_to_cpu(cp->ckpt_flags); 2078 2079 return ckpt_flags & f; 2080 } 2081 2082 static inline bool is_set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 2083 { 2084 return __is_set_ckpt_flags(F2FS_CKPT(sbi), f); 2085 } 2086 2087 static inline void __set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 2088 { 2089 unsigned int ckpt_flags; 2090 2091 ckpt_flags = le32_to_cpu(cp->ckpt_flags); 2092 ckpt_flags |= f; 2093 cp->ckpt_flags = cpu_to_le32(ckpt_flags); 2094 } 2095 2096 static inline void set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 2097 { 2098 unsigned long flags; 2099 2100 spin_lock_irqsave(&sbi->cp_lock, flags); 2101 __set_ckpt_flags(F2FS_CKPT(sbi), f); 2102 spin_unlock_irqrestore(&sbi->cp_lock, flags); 2103 } 2104 2105 static inline void __clear_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 2106 { 2107 unsigned int ckpt_flags; 2108 2109 ckpt_flags = le32_to_cpu(cp->ckpt_flags); 2110 ckpt_flags &= (~f); 2111 cp->ckpt_flags = cpu_to_le32(ckpt_flags); 2112 } 2113 2114 static inline void clear_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 2115 { 2116 unsigned long flags; 2117 2118 spin_lock_irqsave(&sbi->cp_lock, flags); 2119 __clear_ckpt_flags(F2FS_CKPT(sbi), f); 2120 spin_unlock_irqrestore(&sbi->cp_lock, flags); 2121 } 2122 2123 #define init_f2fs_rwsem(sem) \ 2124 do { \ 2125 static struct lock_class_key __key; \ 2126 \ 2127 __init_f2fs_rwsem((sem), #sem, &__key); \ 2128 } while (0) 2129 2130 static inline void __init_f2fs_rwsem(struct f2fs_rwsem *sem, 2131 const char *sem_name, struct lock_class_key *key) 2132 { 2133 __init_rwsem(&sem->internal_rwsem, sem_name, key); 2134 init_waitqueue_head(&sem->read_waiters); 2135 } 2136 2137 static inline int f2fs_rwsem_is_locked(struct f2fs_rwsem *sem) 2138 { 2139 return rwsem_is_locked(&sem->internal_rwsem); 2140 } 2141 2142 static inline int f2fs_rwsem_is_contended(struct f2fs_rwsem *sem) 2143 { 2144 return rwsem_is_contended(&sem->internal_rwsem); 2145 } 2146 2147 static inline void f2fs_down_read(struct f2fs_rwsem *sem) 2148 { 2149 wait_event(sem->read_waiters, down_read_trylock(&sem->internal_rwsem)); 2150 } 2151 2152 static inline int f2fs_down_read_trylock(struct f2fs_rwsem *sem) 2153 { 2154 return down_read_trylock(&sem->internal_rwsem); 2155 } 2156 2157 #ifdef CONFIG_DEBUG_LOCK_ALLOC 2158 static inline void f2fs_down_read_nested(struct f2fs_rwsem *sem, int subclass) 2159 { 2160 down_read_nested(&sem->internal_rwsem, subclass); 2161 } 2162 #else 2163 #define f2fs_down_read_nested(sem, subclass) f2fs_down_read(sem) 2164 #endif 2165 2166 static inline void f2fs_up_read(struct f2fs_rwsem *sem) 2167 { 2168 up_read(&sem->internal_rwsem); 2169 } 2170 2171 static inline void f2fs_down_write(struct f2fs_rwsem *sem) 2172 { 2173 down_write(&sem->internal_rwsem); 2174 } 2175 2176 static inline int f2fs_down_write_trylock(struct f2fs_rwsem *sem) 2177 { 2178 return down_write_trylock(&sem->internal_rwsem); 2179 } 2180 2181 static inline void f2fs_up_write(struct f2fs_rwsem *sem) 2182 { 2183 up_write(&sem->internal_rwsem); 2184 wake_up_all(&sem->read_waiters); 2185 } 2186 2187 static inline void f2fs_lock_op(struct f2fs_sb_info *sbi) 2188 { 2189 f2fs_down_read(&sbi->cp_rwsem); 2190 } 2191 2192 static inline int f2fs_trylock_op(struct f2fs_sb_info *sbi) 2193 { 2194 if (time_to_inject(sbi, FAULT_LOCK_OP)) { 2195 f2fs_show_injection_info(sbi, FAULT_LOCK_OP); 2196 return 0; 2197 } 2198 return f2fs_down_read_trylock(&sbi->cp_rwsem); 2199 } 2200 2201 static inline void f2fs_unlock_op(struct f2fs_sb_info *sbi) 2202 { 2203 f2fs_up_read(&sbi->cp_rwsem); 2204 } 2205 2206 static inline void f2fs_lock_all(struct f2fs_sb_info *sbi) 2207 { 2208 f2fs_down_write(&sbi->cp_rwsem); 2209 } 2210 2211 static inline void f2fs_unlock_all(struct f2fs_sb_info *sbi) 2212 { 2213 f2fs_up_write(&sbi->cp_rwsem); 2214 } 2215 2216 static inline int __get_cp_reason(struct f2fs_sb_info *sbi) 2217 { 2218 int reason = CP_SYNC; 2219 2220 if (test_opt(sbi, FASTBOOT)) 2221 reason = CP_FASTBOOT; 2222 if (is_sbi_flag_set(sbi, SBI_IS_CLOSE)) 2223 reason = CP_UMOUNT; 2224 return reason; 2225 } 2226 2227 static inline bool __remain_node_summaries(int reason) 2228 { 2229 return (reason & (CP_UMOUNT | CP_FASTBOOT)); 2230 } 2231 2232 static inline bool __exist_node_summaries(struct f2fs_sb_info *sbi) 2233 { 2234 return (is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG) || 2235 is_set_ckpt_flags(sbi, CP_FASTBOOT_FLAG)); 2236 } 2237 2238 /* 2239 * Check whether the inode has blocks or not 2240 */ 2241 static inline int F2FS_HAS_BLOCKS(struct inode *inode) 2242 { 2243 block_t xattr_block = F2FS_I(inode)->i_xattr_nid ? 1 : 0; 2244 2245 return (inode->i_blocks >> F2FS_LOG_SECTORS_PER_BLOCK) > xattr_block; 2246 } 2247 2248 static inline bool f2fs_has_xattr_block(unsigned int ofs) 2249 { 2250 return ofs == XATTR_NODE_OFFSET; 2251 } 2252 2253 static inline bool __allow_reserved_blocks(struct f2fs_sb_info *sbi, 2254 struct inode *inode, bool cap) 2255 { 2256 if (!inode) 2257 return true; 2258 if (!test_opt(sbi, RESERVE_ROOT)) 2259 return false; 2260 if (IS_NOQUOTA(inode)) 2261 return true; 2262 if (uid_eq(F2FS_OPTION(sbi).s_resuid, current_fsuid())) 2263 return true; 2264 if (!gid_eq(F2FS_OPTION(sbi).s_resgid, GLOBAL_ROOT_GID) && 2265 in_group_p(F2FS_OPTION(sbi).s_resgid)) 2266 return true; 2267 if (cap && capable(CAP_SYS_RESOURCE)) 2268 return true; 2269 return false; 2270 } 2271 2272 static inline void f2fs_i_blocks_write(struct inode *, block_t, bool, bool); 2273 static inline int inc_valid_block_count(struct f2fs_sb_info *sbi, 2274 struct inode *inode, blkcnt_t *count) 2275 { 2276 blkcnt_t diff = 0, release = 0; 2277 block_t avail_user_block_count; 2278 int ret; 2279 2280 ret = dquot_reserve_block(inode, *count); 2281 if (ret) 2282 return ret; 2283 2284 if (time_to_inject(sbi, FAULT_BLOCK)) { 2285 f2fs_show_injection_info(sbi, FAULT_BLOCK); 2286 release = *count; 2287 goto release_quota; 2288 } 2289 2290 /* 2291 * let's increase this in prior to actual block count change in order 2292 * for f2fs_sync_file to avoid data races when deciding checkpoint. 2293 */ 2294 percpu_counter_add(&sbi->alloc_valid_block_count, (*count)); 2295 2296 spin_lock(&sbi->stat_lock); 2297 sbi->total_valid_block_count += (block_t)(*count); 2298 avail_user_block_count = sbi->user_block_count - 2299 sbi->current_reserved_blocks; 2300 2301 if (!__allow_reserved_blocks(sbi, inode, true)) 2302 avail_user_block_count -= F2FS_OPTION(sbi).root_reserved_blocks; 2303 2304 if (F2FS_IO_ALIGNED(sbi)) 2305 avail_user_block_count -= sbi->blocks_per_seg * 2306 SM_I(sbi)->additional_reserved_segments; 2307 2308 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) { 2309 if (avail_user_block_count > sbi->unusable_block_count) 2310 avail_user_block_count -= sbi->unusable_block_count; 2311 else 2312 avail_user_block_count = 0; 2313 } 2314 if (unlikely(sbi->total_valid_block_count > avail_user_block_count)) { 2315 diff = sbi->total_valid_block_count - avail_user_block_count; 2316 if (diff > *count) 2317 diff = *count; 2318 *count -= diff; 2319 release = diff; 2320 sbi->total_valid_block_count -= diff; 2321 if (!*count) { 2322 spin_unlock(&sbi->stat_lock); 2323 goto enospc; 2324 } 2325 } 2326 spin_unlock(&sbi->stat_lock); 2327 2328 if (unlikely(release)) { 2329 percpu_counter_sub(&sbi->alloc_valid_block_count, release); 2330 dquot_release_reservation_block(inode, release); 2331 } 2332 f2fs_i_blocks_write(inode, *count, true, true); 2333 return 0; 2334 2335 enospc: 2336 percpu_counter_sub(&sbi->alloc_valid_block_count, release); 2337 release_quota: 2338 dquot_release_reservation_block(inode, release); 2339 return -ENOSPC; 2340 } 2341 2342 __printf(2, 3) 2343 void f2fs_printk(struct f2fs_sb_info *sbi, const char *fmt, ...); 2344 2345 #define f2fs_err(sbi, fmt, ...) \ 2346 f2fs_printk(sbi, KERN_ERR fmt, ##__VA_ARGS__) 2347 #define f2fs_warn(sbi, fmt, ...) \ 2348 f2fs_printk(sbi, KERN_WARNING fmt, ##__VA_ARGS__) 2349 #define f2fs_notice(sbi, fmt, ...) \ 2350 f2fs_printk(sbi, KERN_NOTICE fmt, ##__VA_ARGS__) 2351 #define f2fs_info(sbi, fmt, ...) \ 2352 f2fs_printk(sbi, KERN_INFO fmt, ##__VA_ARGS__) 2353 #define f2fs_debug(sbi, fmt, ...) \ 2354 f2fs_printk(sbi, KERN_DEBUG fmt, ##__VA_ARGS__) 2355 2356 static inline void dec_valid_block_count(struct f2fs_sb_info *sbi, 2357 struct inode *inode, 2358 block_t count) 2359 { 2360 blkcnt_t sectors = count << F2FS_LOG_SECTORS_PER_BLOCK; 2361 2362 spin_lock(&sbi->stat_lock); 2363 f2fs_bug_on(sbi, sbi->total_valid_block_count < (block_t) count); 2364 sbi->total_valid_block_count -= (block_t)count; 2365 if (sbi->reserved_blocks && 2366 sbi->current_reserved_blocks < sbi->reserved_blocks) 2367 sbi->current_reserved_blocks = min(sbi->reserved_blocks, 2368 sbi->current_reserved_blocks + count); 2369 spin_unlock(&sbi->stat_lock); 2370 if (unlikely(inode->i_blocks < sectors)) { 2371 f2fs_warn(sbi, "Inconsistent i_blocks, ino:%lu, iblocks:%llu, sectors:%llu", 2372 inode->i_ino, 2373 (unsigned long long)inode->i_blocks, 2374 (unsigned long long)sectors); 2375 set_sbi_flag(sbi, SBI_NEED_FSCK); 2376 return; 2377 } 2378 f2fs_i_blocks_write(inode, count, false, true); 2379 } 2380 2381 static inline void inc_page_count(struct f2fs_sb_info *sbi, int count_type) 2382 { 2383 atomic_inc(&sbi->nr_pages[count_type]); 2384 2385 if (count_type == F2FS_DIRTY_DENTS || 2386 count_type == F2FS_DIRTY_NODES || 2387 count_type == F2FS_DIRTY_META || 2388 count_type == F2FS_DIRTY_QDATA || 2389 count_type == F2FS_DIRTY_IMETA) 2390 set_sbi_flag(sbi, SBI_IS_DIRTY); 2391 } 2392 2393 static inline void inode_inc_dirty_pages(struct inode *inode) 2394 { 2395 atomic_inc(&F2FS_I(inode)->dirty_pages); 2396 inc_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ? 2397 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA); 2398 if (IS_NOQUOTA(inode)) 2399 inc_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA); 2400 } 2401 2402 static inline void dec_page_count(struct f2fs_sb_info *sbi, int count_type) 2403 { 2404 atomic_dec(&sbi->nr_pages[count_type]); 2405 } 2406 2407 static inline void inode_dec_dirty_pages(struct inode *inode) 2408 { 2409 if (!S_ISDIR(inode->i_mode) && !S_ISREG(inode->i_mode) && 2410 !S_ISLNK(inode->i_mode)) 2411 return; 2412 2413 atomic_dec(&F2FS_I(inode)->dirty_pages); 2414 dec_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ? 2415 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA); 2416 if (IS_NOQUOTA(inode)) 2417 dec_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA); 2418 } 2419 2420 static inline s64 get_pages(struct f2fs_sb_info *sbi, int count_type) 2421 { 2422 return atomic_read(&sbi->nr_pages[count_type]); 2423 } 2424 2425 static inline int get_dirty_pages(struct inode *inode) 2426 { 2427 return atomic_read(&F2FS_I(inode)->dirty_pages); 2428 } 2429 2430 static inline int get_blocktype_secs(struct f2fs_sb_info *sbi, int block_type) 2431 { 2432 unsigned int pages_per_sec = sbi->segs_per_sec * sbi->blocks_per_seg; 2433 unsigned int segs = (get_pages(sbi, block_type) + pages_per_sec - 1) >> 2434 sbi->log_blocks_per_seg; 2435 2436 return segs / sbi->segs_per_sec; 2437 } 2438 2439 static inline block_t valid_user_blocks(struct f2fs_sb_info *sbi) 2440 { 2441 return sbi->total_valid_block_count; 2442 } 2443 2444 static inline block_t discard_blocks(struct f2fs_sb_info *sbi) 2445 { 2446 return sbi->discard_blks; 2447 } 2448 2449 static inline unsigned long __bitmap_size(struct f2fs_sb_info *sbi, int flag) 2450 { 2451 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi); 2452 2453 /* return NAT or SIT bitmap */ 2454 if (flag == NAT_BITMAP) 2455 return le32_to_cpu(ckpt->nat_ver_bitmap_bytesize); 2456 else if (flag == SIT_BITMAP) 2457 return le32_to_cpu(ckpt->sit_ver_bitmap_bytesize); 2458 2459 return 0; 2460 } 2461 2462 static inline block_t __cp_payload(struct f2fs_sb_info *sbi) 2463 { 2464 return le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_payload); 2465 } 2466 2467 static inline void *__bitmap_ptr(struct f2fs_sb_info *sbi, int flag) 2468 { 2469 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi); 2470 void *tmp_ptr = &ckpt->sit_nat_version_bitmap; 2471 int offset; 2472 2473 if (is_set_ckpt_flags(sbi, CP_LARGE_NAT_BITMAP_FLAG)) { 2474 offset = (flag == SIT_BITMAP) ? 2475 le32_to_cpu(ckpt->nat_ver_bitmap_bytesize) : 0; 2476 /* 2477 * if large_nat_bitmap feature is enabled, leave checksum 2478 * protection for all nat/sit bitmaps. 2479 */ 2480 return tmp_ptr + offset + sizeof(__le32); 2481 } 2482 2483 if (__cp_payload(sbi) > 0) { 2484 if (flag == NAT_BITMAP) 2485 return &ckpt->sit_nat_version_bitmap; 2486 else 2487 return (unsigned char *)ckpt + F2FS_BLKSIZE; 2488 } else { 2489 offset = (flag == NAT_BITMAP) ? 2490 le32_to_cpu(ckpt->sit_ver_bitmap_bytesize) : 0; 2491 return tmp_ptr + offset; 2492 } 2493 } 2494 2495 static inline block_t __start_cp_addr(struct f2fs_sb_info *sbi) 2496 { 2497 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr); 2498 2499 if (sbi->cur_cp_pack == 2) 2500 start_addr += sbi->blocks_per_seg; 2501 return start_addr; 2502 } 2503 2504 static inline block_t __start_cp_next_addr(struct f2fs_sb_info *sbi) 2505 { 2506 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr); 2507 2508 if (sbi->cur_cp_pack == 1) 2509 start_addr += sbi->blocks_per_seg; 2510 return start_addr; 2511 } 2512 2513 static inline void __set_cp_next_pack(struct f2fs_sb_info *sbi) 2514 { 2515 sbi->cur_cp_pack = (sbi->cur_cp_pack == 1) ? 2 : 1; 2516 } 2517 2518 static inline block_t __start_sum_addr(struct f2fs_sb_info *sbi) 2519 { 2520 return le32_to_cpu(F2FS_CKPT(sbi)->cp_pack_start_sum); 2521 } 2522 2523 static inline int inc_valid_node_count(struct f2fs_sb_info *sbi, 2524 struct inode *inode, bool is_inode) 2525 { 2526 block_t valid_block_count; 2527 unsigned int valid_node_count, user_block_count; 2528 int err; 2529 2530 if (is_inode) { 2531 if (inode) { 2532 err = dquot_alloc_inode(inode); 2533 if (err) 2534 return err; 2535 } 2536 } else { 2537 err = dquot_reserve_block(inode, 1); 2538 if (err) 2539 return err; 2540 } 2541 2542 if (time_to_inject(sbi, FAULT_BLOCK)) { 2543 f2fs_show_injection_info(sbi, FAULT_BLOCK); 2544 goto enospc; 2545 } 2546 2547 spin_lock(&sbi->stat_lock); 2548 2549 valid_block_count = sbi->total_valid_block_count + 2550 sbi->current_reserved_blocks + 1; 2551 2552 if (!__allow_reserved_blocks(sbi, inode, false)) 2553 valid_block_count += F2FS_OPTION(sbi).root_reserved_blocks; 2554 2555 if (F2FS_IO_ALIGNED(sbi)) 2556 valid_block_count += sbi->blocks_per_seg * 2557 SM_I(sbi)->additional_reserved_segments; 2558 2559 user_block_count = sbi->user_block_count; 2560 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) 2561 user_block_count -= sbi->unusable_block_count; 2562 2563 if (unlikely(valid_block_count > user_block_count)) { 2564 spin_unlock(&sbi->stat_lock); 2565 goto enospc; 2566 } 2567 2568 valid_node_count = sbi->total_valid_node_count + 1; 2569 if (unlikely(valid_node_count > sbi->total_node_count)) { 2570 spin_unlock(&sbi->stat_lock); 2571 goto enospc; 2572 } 2573 2574 sbi->total_valid_node_count++; 2575 sbi->total_valid_block_count++; 2576 spin_unlock(&sbi->stat_lock); 2577 2578 if (inode) { 2579 if (is_inode) 2580 f2fs_mark_inode_dirty_sync(inode, true); 2581 else 2582 f2fs_i_blocks_write(inode, 1, true, true); 2583 } 2584 2585 percpu_counter_inc(&sbi->alloc_valid_block_count); 2586 return 0; 2587 2588 enospc: 2589 if (is_inode) { 2590 if (inode) 2591 dquot_free_inode(inode); 2592 } else { 2593 dquot_release_reservation_block(inode, 1); 2594 } 2595 return -ENOSPC; 2596 } 2597 2598 static inline void dec_valid_node_count(struct f2fs_sb_info *sbi, 2599 struct inode *inode, bool is_inode) 2600 { 2601 spin_lock(&sbi->stat_lock); 2602 2603 f2fs_bug_on(sbi, !sbi->total_valid_block_count); 2604 f2fs_bug_on(sbi, !sbi->total_valid_node_count); 2605 2606 sbi->total_valid_node_count--; 2607 sbi->total_valid_block_count--; 2608 if (sbi->reserved_blocks && 2609 sbi->current_reserved_blocks < sbi->reserved_blocks) 2610 sbi->current_reserved_blocks++; 2611 2612 spin_unlock(&sbi->stat_lock); 2613 2614 if (is_inode) { 2615 dquot_free_inode(inode); 2616 } else { 2617 if (unlikely(inode->i_blocks == 0)) { 2618 f2fs_warn(sbi, "dec_valid_node_count: inconsistent i_blocks, ino:%lu, iblocks:%llu", 2619 inode->i_ino, 2620 (unsigned long long)inode->i_blocks); 2621 set_sbi_flag(sbi, SBI_NEED_FSCK); 2622 return; 2623 } 2624 f2fs_i_blocks_write(inode, 1, false, true); 2625 } 2626 } 2627 2628 static inline unsigned int valid_node_count(struct f2fs_sb_info *sbi) 2629 { 2630 return sbi->total_valid_node_count; 2631 } 2632 2633 static inline void inc_valid_inode_count(struct f2fs_sb_info *sbi) 2634 { 2635 percpu_counter_inc(&sbi->total_valid_inode_count); 2636 } 2637 2638 static inline void dec_valid_inode_count(struct f2fs_sb_info *sbi) 2639 { 2640 percpu_counter_dec(&sbi->total_valid_inode_count); 2641 } 2642 2643 static inline s64 valid_inode_count(struct f2fs_sb_info *sbi) 2644 { 2645 return percpu_counter_sum_positive(&sbi->total_valid_inode_count); 2646 } 2647 2648 static inline struct page *f2fs_grab_cache_page(struct address_space *mapping, 2649 pgoff_t index, bool for_write) 2650 { 2651 struct page *page; 2652 2653 if (IS_ENABLED(CONFIG_F2FS_FAULT_INJECTION)) { 2654 if (!for_write) 2655 page = find_get_page_flags(mapping, index, 2656 FGP_LOCK | FGP_ACCESSED); 2657 else 2658 page = find_lock_page(mapping, index); 2659 if (page) 2660 return page; 2661 2662 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_ALLOC)) { 2663 f2fs_show_injection_info(F2FS_M_SB(mapping), 2664 FAULT_PAGE_ALLOC); 2665 return NULL; 2666 } 2667 } 2668 2669 if (!for_write) 2670 return grab_cache_page(mapping, index); 2671 return grab_cache_page_write_begin(mapping, index, AOP_FLAG_NOFS); 2672 } 2673 2674 static inline struct page *f2fs_pagecache_get_page( 2675 struct address_space *mapping, pgoff_t index, 2676 int fgp_flags, gfp_t gfp_mask) 2677 { 2678 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_GET)) { 2679 f2fs_show_injection_info(F2FS_M_SB(mapping), FAULT_PAGE_GET); 2680 return NULL; 2681 } 2682 2683 return pagecache_get_page(mapping, index, fgp_flags, gfp_mask); 2684 } 2685 2686 static inline void f2fs_copy_page(struct page *src, struct page *dst) 2687 { 2688 char *src_kaddr = kmap(src); 2689 char *dst_kaddr = kmap(dst); 2690 2691 memcpy(dst_kaddr, src_kaddr, PAGE_SIZE); 2692 kunmap(dst); 2693 kunmap(src); 2694 } 2695 2696 static inline void f2fs_put_page(struct page *page, int unlock) 2697 { 2698 if (!page) 2699 return; 2700 2701 if (unlock) { 2702 f2fs_bug_on(F2FS_P_SB(page), !PageLocked(page)); 2703 unlock_page(page); 2704 } 2705 put_page(page); 2706 } 2707 2708 static inline void f2fs_put_dnode(struct dnode_of_data *dn) 2709 { 2710 if (dn->node_page) 2711 f2fs_put_page(dn->node_page, 1); 2712 if (dn->inode_page && dn->node_page != dn->inode_page) 2713 f2fs_put_page(dn->inode_page, 0); 2714 dn->node_page = NULL; 2715 dn->inode_page = NULL; 2716 } 2717 2718 static inline struct kmem_cache *f2fs_kmem_cache_create(const char *name, 2719 size_t size) 2720 { 2721 return kmem_cache_create(name, size, 0, SLAB_RECLAIM_ACCOUNT, NULL); 2722 } 2723 2724 static inline void *f2fs_kmem_cache_alloc_nofail(struct kmem_cache *cachep, 2725 gfp_t flags) 2726 { 2727 void *entry; 2728 2729 entry = kmem_cache_alloc(cachep, flags); 2730 if (!entry) 2731 entry = kmem_cache_alloc(cachep, flags | __GFP_NOFAIL); 2732 return entry; 2733 } 2734 2735 static inline void *f2fs_kmem_cache_alloc(struct kmem_cache *cachep, 2736 gfp_t flags, bool nofail, struct f2fs_sb_info *sbi) 2737 { 2738 if (nofail) 2739 return f2fs_kmem_cache_alloc_nofail(cachep, flags); 2740 2741 if (time_to_inject(sbi, FAULT_SLAB_ALLOC)) { 2742 f2fs_show_injection_info(sbi, FAULT_SLAB_ALLOC); 2743 return NULL; 2744 } 2745 2746 return kmem_cache_alloc(cachep, flags); 2747 } 2748 2749 static inline bool is_inflight_io(struct f2fs_sb_info *sbi, int type) 2750 { 2751 if (get_pages(sbi, F2FS_RD_DATA) || get_pages(sbi, F2FS_RD_NODE) || 2752 get_pages(sbi, F2FS_RD_META) || get_pages(sbi, F2FS_WB_DATA) || 2753 get_pages(sbi, F2FS_WB_CP_DATA) || 2754 get_pages(sbi, F2FS_DIO_READ) || 2755 get_pages(sbi, F2FS_DIO_WRITE)) 2756 return true; 2757 2758 if (type != DISCARD_TIME && SM_I(sbi) && SM_I(sbi)->dcc_info && 2759 atomic_read(&SM_I(sbi)->dcc_info->queued_discard)) 2760 return true; 2761 2762 if (SM_I(sbi) && SM_I(sbi)->fcc_info && 2763 atomic_read(&SM_I(sbi)->fcc_info->queued_flush)) 2764 return true; 2765 return false; 2766 } 2767 2768 static inline bool is_idle(struct f2fs_sb_info *sbi, int type) 2769 { 2770 if (sbi->gc_mode == GC_URGENT_HIGH) 2771 return true; 2772 2773 if (is_inflight_io(sbi, type)) 2774 return false; 2775 2776 if (sbi->gc_mode == GC_URGENT_LOW && 2777 (type == DISCARD_TIME || type == GC_TIME)) 2778 return true; 2779 2780 return f2fs_time_over(sbi, type); 2781 } 2782 2783 static inline void f2fs_radix_tree_insert(struct radix_tree_root *root, 2784 unsigned long index, void *item) 2785 { 2786 while (radix_tree_insert(root, index, item)) 2787 cond_resched(); 2788 } 2789 2790 #define RAW_IS_INODE(p) ((p)->footer.nid == (p)->footer.ino) 2791 2792 static inline bool IS_INODE(struct page *page) 2793 { 2794 struct f2fs_node *p = F2FS_NODE(page); 2795 2796 return RAW_IS_INODE(p); 2797 } 2798 2799 static inline int offset_in_addr(struct f2fs_inode *i) 2800 { 2801 return (i->i_inline & F2FS_EXTRA_ATTR) ? 2802 (le16_to_cpu(i->i_extra_isize) / sizeof(__le32)) : 0; 2803 } 2804 2805 static inline __le32 *blkaddr_in_node(struct f2fs_node *node) 2806 { 2807 return RAW_IS_INODE(node) ? node->i.i_addr : node->dn.addr; 2808 } 2809 2810 static inline int f2fs_has_extra_attr(struct inode *inode); 2811 static inline block_t data_blkaddr(struct inode *inode, 2812 struct page *node_page, unsigned int offset) 2813 { 2814 struct f2fs_node *raw_node; 2815 __le32 *addr_array; 2816 int base = 0; 2817 bool is_inode = IS_INODE(node_page); 2818 2819 raw_node = F2FS_NODE(node_page); 2820 2821 if (is_inode) { 2822 if (!inode) 2823 /* from GC path only */ 2824 base = offset_in_addr(&raw_node->i); 2825 else if (f2fs_has_extra_attr(inode)) 2826 base = get_extra_isize(inode); 2827 } 2828 2829 addr_array = blkaddr_in_node(raw_node); 2830 return le32_to_cpu(addr_array[base + offset]); 2831 } 2832 2833 static inline block_t f2fs_data_blkaddr(struct dnode_of_data *dn) 2834 { 2835 return data_blkaddr(dn->inode, dn->node_page, dn->ofs_in_node); 2836 } 2837 2838 static inline int f2fs_test_bit(unsigned int nr, char *addr) 2839 { 2840 int mask; 2841 2842 addr += (nr >> 3); 2843 mask = 1 << (7 - (nr & 0x07)); 2844 return mask & *addr; 2845 } 2846 2847 static inline void f2fs_set_bit(unsigned int nr, char *addr) 2848 { 2849 int mask; 2850 2851 addr += (nr >> 3); 2852 mask = 1 << (7 - (nr & 0x07)); 2853 *addr |= mask; 2854 } 2855 2856 static inline void f2fs_clear_bit(unsigned int nr, char *addr) 2857 { 2858 int mask; 2859 2860 addr += (nr >> 3); 2861 mask = 1 << (7 - (nr & 0x07)); 2862 *addr &= ~mask; 2863 } 2864 2865 static inline int f2fs_test_and_set_bit(unsigned int nr, char *addr) 2866 { 2867 int mask; 2868 int ret; 2869 2870 addr += (nr >> 3); 2871 mask = 1 << (7 - (nr & 0x07)); 2872 ret = mask & *addr; 2873 *addr |= mask; 2874 return ret; 2875 } 2876 2877 static inline int f2fs_test_and_clear_bit(unsigned int nr, char *addr) 2878 { 2879 int mask; 2880 int ret; 2881 2882 addr += (nr >> 3); 2883 mask = 1 << (7 - (nr & 0x07)); 2884 ret = mask & *addr; 2885 *addr &= ~mask; 2886 return ret; 2887 } 2888 2889 static inline void f2fs_change_bit(unsigned int nr, char *addr) 2890 { 2891 int mask; 2892 2893 addr += (nr >> 3); 2894 mask = 1 << (7 - (nr & 0x07)); 2895 *addr ^= mask; 2896 } 2897 2898 /* 2899 * On-disk inode flags (f2fs_inode::i_flags) 2900 */ 2901 #define F2FS_COMPR_FL 0x00000004 /* Compress file */ 2902 #define F2FS_SYNC_FL 0x00000008 /* Synchronous updates */ 2903 #define F2FS_IMMUTABLE_FL 0x00000010 /* Immutable file */ 2904 #define F2FS_APPEND_FL 0x00000020 /* writes to file may only append */ 2905 #define F2FS_NODUMP_FL 0x00000040 /* do not dump file */ 2906 #define F2FS_NOATIME_FL 0x00000080 /* do not update atime */ 2907 #define F2FS_NOCOMP_FL 0x00000400 /* Don't compress */ 2908 #define F2FS_INDEX_FL 0x00001000 /* hash-indexed directory */ 2909 #define F2FS_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */ 2910 #define F2FS_PROJINHERIT_FL 0x20000000 /* Create with parents projid */ 2911 #define F2FS_CASEFOLD_FL 0x40000000 /* Casefolded file */ 2912 2913 /* Flags that should be inherited by new inodes from their parent. */ 2914 #define F2FS_FL_INHERITED (F2FS_SYNC_FL | F2FS_NODUMP_FL | F2FS_NOATIME_FL | \ 2915 F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \ 2916 F2FS_CASEFOLD_FL | F2FS_COMPR_FL | F2FS_NOCOMP_FL) 2917 2918 /* Flags that are appropriate for regular files (all but dir-specific ones). */ 2919 #define F2FS_REG_FLMASK (~(F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \ 2920 F2FS_CASEFOLD_FL)) 2921 2922 /* Flags that are appropriate for non-directories/regular files. */ 2923 #define F2FS_OTHER_FLMASK (F2FS_NODUMP_FL | F2FS_NOATIME_FL) 2924 2925 static inline __u32 f2fs_mask_flags(umode_t mode, __u32 flags) 2926 { 2927 if (S_ISDIR(mode)) 2928 return flags; 2929 else if (S_ISREG(mode)) 2930 return flags & F2FS_REG_FLMASK; 2931 else 2932 return flags & F2FS_OTHER_FLMASK; 2933 } 2934 2935 static inline void __mark_inode_dirty_flag(struct inode *inode, 2936 int flag, bool set) 2937 { 2938 switch (flag) { 2939 case FI_INLINE_XATTR: 2940 case FI_INLINE_DATA: 2941 case FI_INLINE_DENTRY: 2942 case FI_NEW_INODE: 2943 if (set) 2944 return; 2945 fallthrough; 2946 case FI_DATA_EXIST: 2947 case FI_INLINE_DOTS: 2948 case FI_PIN_FILE: 2949 case FI_COMPRESS_RELEASED: 2950 f2fs_mark_inode_dirty_sync(inode, true); 2951 } 2952 } 2953 2954 static inline void set_inode_flag(struct inode *inode, int flag) 2955 { 2956 set_bit(flag, F2FS_I(inode)->flags); 2957 __mark_inode_dirty_flag(inode, flag, true); 2958 } 2959 2960 static inline int is_inode_flag_set(struct inode *inode, int flag) 2961 { 2962 return test_bit(flag, F2FS_I(inode)->flags); 2963 } 2964 2965 static inline void clear_inode_flag(struct inode *inode, int flag) 2966 { 2967 clear_bit(flag, F2FS_I(inode)->flags); 2968 __mark_inode_dirty_flag(inode, flag, false); 2969 } 2970 2971 static inline bool f2fs_verity_in_progress(struct inode *inode) 2972 { 2973 return IS_ENABLED(CONFIG_FS_VERITY) && 2974 is_inode_flag_set(inode, FI_VERITY_IN_PROGRESS); 2975 } 2976 2977 static inline void set_acl_inode(struct inode *inode, umode_t mode) 2978 { 2979 F2FS_I(inode)->i_acl_mode = mode; 2980 set_inode_flag(inode, FI_ACL_MODE); 2981 f2fs_mark_inode_dirty_sync(inode, false); 2982 } 2983 2984 static inline void f2fs_i_links_write(struct inode *inode, bool inc) 2985 { 2986 if (inc) 2987 inc_nlink(inode); 2988 else 2989 drop_nlink(inode); 2990 f2fs_mark_inode_dirty_sync(inode, true); 2991 } 2992 2993 static inline void f2fs_i_blocks_write(struct inode *inode, 2994 block_t diff, bool add, bool claim) 2995 { 2996 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE); 2997 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER); 2998 2999 /* add = 1, claim = 1 should be dquot_reserve_block in pair */ 3000 if (add) { 3001 if (claim) 3002 dquot_claim_block(inode, diff); 3003 else 3004 dquot_alloc_block_nofail(inode, diff); 3005 } else { 3006 dquot_free_block(inode, diff); 3007 } 3008 3009 f2fs_mark_inode_dirty_sync(inode, true); 3010 if (clean || recover) 3011 set_inode_flag(inode, FI_AUTO_RECOVER); 3012 } 3013 3014 static inline void f2fs_i_size_write(struct inode *inode, loff_t i_size) 3015 { 3016 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE); 3017 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER); 3018 3019 if (i_size_read(inode) == i_size) 3020 return; 3021 3022 i_size_write(inode, i_size); 3023 f2fs_mark_inode_dirty_sync(inode, true); 3024 if (clean || recover) 3025 set_inode_flag(inode, FI_AUTO_RECOVER); 3026 } 3027 3028 static inline void f2fs_i_depth_write(struct inode *inode, unsigned int depth) 3029 { 3030 F2FS_I(inode)->i_current_depth = depth; 3031 f2fs_mark_inode_dirty_sync(inode, true); 3032 } 3033 3034 static inline void f2fs_i_gc_failures_write(struct inode *inode, 3035 unsigned int count) 3036 { 3037 F2FS_I(inode)->i_gc_failures[GC_FAILURE_PIN] = count; 3038 f2fs_mark_inode_dirty_sync(inode, true); 3039 } 3040 3041 static inline void f2fs_i_xnid_write(struct inode *inode, nid_t xnid) 3042 { 3043 F2FS_I(inode)->i_xattr_nid = xnid; 3044 f2fs_mark_inode_dirty_sync(inode, true); 3045 } 3046 3047 static inline void f2fs_i_pino_write(struct inode *inode, nid_t pino) 3048 { 3049 F2FS_I(inode)->i_pino = pino; 3050 f2fs_mark_inode_dirty_sync(inode, true); 3051 } 3052 3053 static inline void get_inline_info(struct inode *inode, struct f2fs_inode *ri) 3054 { 3055 struct f2fs_inode_info *fi = F2FS_I(inode); 3056 3057 if (ri->i_inline & F2FS_INLINE_XATTR) 3058 set_bit(FI_INLINE_XATTR, fi->flags); 3059 if (ri->i_inline & F2FS_INLINE_DATA) 3060 set_bit(FI_INLINE_DATA, fi->flags); 3061 if (ri->i_inline & F2FS_INLINE_DENTRY) 3062 set_bit(FI_INLINE_DENTRY, fi->flags); 3063 if (ri->i_inline & F2FS_DATA_EXIST) 3064 set_bit(FI_DATA_EXIST, fi->flags); 3065 if (ri->i_inline & F2FS_INLINE_DOTS) 3066 set_bit(FI_INLINE_DOTS, fi->flags); 3067 if (ri->i_inline & F2FS_EXTRA_ATTR) 3068 set_bit(FI_EXTRA_ATTR, fi->flags); 3069 if (ri->i_inline & F2FS_PIN_FILE) 3070 set_bit(FI_PIN_FILE, fi->flags); 3071 if (ri->i_inline & F2FS_COMPRESS_RELEASED) 3072 set_bit(FI_COMPRESS_RELEASED, fi->flags); 3073 } 3074 3075 static inline void set_raw_inline(struct inode *inode, struct f2fs_inode *ri) 3076 { 3077 ri->i_inline = 0; 3078 3079 if (is_inode_flag_set(inode, FI_INLINE_XATTR)) 3080 ri->i_inline |= F2FS_INLINE_XATTR; 3081 if (is_inode_flag_set(inode, FI_INLINE_DATA)) 3082 ri->i_inline |= F2FS_INLINE_DATA; 3083 if (is_inode_flag_set(inode, FI_INLINE_DENTRY)) 3084 ri->i_inline |= F2FS_INLINE_DENTRY; 3085 if (is_inode_flag_set(inode, FI_DATA_EXIST)) 3086 ri->i_inline |= F2FS_DATA_EXIST; 3087 if (is_inode_flag_set(inode, FI_INLINE_DOTS)) 3088 ri->i_inline |= F2FS_INLINE_DOTS; 3089 if (is_inode_flag_set(inode, FI_EXTRA_ATTR)) 3090 ri->i_inline |= F2FS_EXTRA_ATTR; 3091 if (is_inode_flag_set(inode, FI_PIN_FILE)) 3092 ri->i_inline |= F2FS_PIN_FILE; 3093 if (is_inode_flag_set(inode, FI_COMPRESS_RELEASED)) 3094 ri->i_inline |= F2FS_COMPRESS_RELEASED; 3095 } 3096 3097 static inline int f2fs_has_extra_attr(struct inode *inode) 3098 { 3099 return is_inode_flag_set(inode, FI_EXTRA_ATTR); 3100 } 3101 3102 static inline int f2fs_has_inline_xattr(struct inode *inode) 3103 { 3104 return is_inode_flag_set(inode, FI_INLINE_XATTR); 3105 } 3106 3107 static inline int f2fs_compressed_file(struct inode *inode) 3108 { 3109 return S_ISREG(inode->i_mode) && 3110 is_inode_flag_set(inode, FI_COMPRESSED_FILE); 3111 } 3112 3113 static inline bool f2fs_need_compress_data(struct inode *inode) 3114 { 3115 int compress_mode = F2FS_OPTION(F2FS_I_SB(inode)).compress_mode; 3116 3117 if (!f2fs_compressed_file(inode)) 3118 return false; 3119 3120 if (compress_mode == COMPR_MODE_FS) 3121 return true; 3122 else if (compress_mode == COMPR_MODE_USER && 3123 is_inode_flag_set(inode, FI_ENABLE_COMPRESS)) 3124 return true; 3125 3126 return false; 3127 } 3128 3129 static inline unsigned int addrs_per_inode(struct inode *inode) 3130 { 3131 unsigned int addrs = CUR_ADDRS_PER_INODE(inode) - 3132 get_inline_xattr_addrs(inode); 3133 3134 if (!f2fs_compressed_file(inode)) 3135 return addrs; 3136 return ALIGN_DOWN(addrs, F2FS_I(inode)->i_cluster_size); 3137 } 3138 3139 static inline unsigned int addrs_per_block(struct inode *inode) 3140 { 3141 if (!f2fs_compressed_file(inode)) 3142 return DEF_ADDRS_PER_BLOCK; 3143 return ALIGN_DOWN(DEF_ADDRS_PER_BLOCK, F2FS_I(inode)->i_cluster_size); 3144 } 3145 3146 static inline void *inline_xattr_addr(struct inode *inode, struct page *page) 3147 { 3148 struct f2fs_inode *ri = F2FS_INODE(page); 3149 3150 return (void *)&(ri->i_addr[DEF_ADDRS_PER_INODE - 3151 get_inline_xattr_addrs(inode)]); 3152 } 3153 3154 static inline int inline_xattr_size(struct inode *inode) 3155 { 3156 if (f2fs_has_inline_xattr(inode)) 3157 return get_inline_xattr_addrs(inode) * sizeof(__le32); 3158 return 0; 3159 } 3160 3161 static inline int f2fs_has_inline_data(struct inode *inode) 3162 { 3163 return is_inode_flag_set(inode, FI_INLINE_DATA); 3164 } 3165 3166 static inline int f2fs_exist_data(struct inode *inode) 3167 { 3168 return is_inode_flag_set(inode, FI_DATA_EXIST); 3169 } 3170 3171 static inline int f2fs_has_inline_dots(struct inode *inode) 3172 { 3173 return is_inode_flag_set(inode, FI_INLINE_DOTS); 3174 } 3175 3176 static inline int f2fs_is_mmap_file(struct inode *inode) 3177 { 3178 return is_inode_flag_set(inode, FI_MMAP_FILE); 3179 } 3180 3181 static inline bool f2fs_is_pinned_file(struct inode *inode) 3182 { 3183 return is_inode_flag_set(inode, FI_PIN_FILE); 3184 } 3185 3186 static inline bool f2fs_is_atomic_file(struct inode *inode) 3187 { 3188 return is_inode_flag_set(inode, FI_ATOMIC_FILE); 3189 } 3190 3191 static inline bool f2fs_is_commit_atomic_write(struct inode *inode) 3192 { 3193 return is_inode_flag_set(inode, FI_ATOMIC_COMMIT); 3194 } 3195 3196 static inline bool f2fs_is_volatile_file(struct inode *inode) 3197 { 3198 return is_inode_flag_set(inode, FI_VOLATILE_FILE); 3199 } 3200 3201 static inline bool f2fs_is_first_block_written(struct inode *inode) 3202 { 3203 return is_inode_flag_set(inode, FI_FIRST_BLOCK_WRITTEN); 3204 } 3205 3206 static inline bool f2fs_is_drop_cache(struct inode *inode) 3207 { 3208 return is_inode_flag_set(inode, FI_DROP_CACHE); 3209 } 3210 3211 static inline void *inline_data_addr(struct inode *inode, struct page *page) 3212 { 3213 struct f2fs_inode *ri = F2FS_INODE(page); 3214 int extra_size = get_extra_isize(inode); 3215 3216 return (void *)&(ri->i_addr[extra_size + DEF_INLINE_RESERVED_SIZE]); 3217 } 3218 3219 static inline int f2fs_has_inline_dentry(struct inode *inode) 3220 { 3221 return is_inode_flag_set(inode, FI_INLINE_DENTRY); 3222 } 3223 3224 static inline int is_file(struct inode *inode, int type) 3225 { 3226 return F2FS_I(inode)->i_advise & type; 3227 } 3228 3229 static inline void set_file(struct inode *inode, int type) 3230 { 3231 if (is_file(inode, type)) 3232 return; 3233 F2FS_I(inode)->i_advise |= type; 3234 f2fs_mark_inode_dirty_sync(inode, true); 3235 } 3236 3237 static inline void clear_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 bool f2fs_is_time_consistent(struct inode *inode) 3246 { 3247 if (!timespec64_equal(F2FS_I(inode)->i_disk_time, &inode->i_atime)) 3248 return false; 3249 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 1, &inode->i_ctime)) 3250 return false; 3251 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 2, &inode->i_mtime)) 3252 return false; 3253 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 3, 3254 &F2FS_I(inode)->i_crtime)) 3255 return false; 3256 return true; 3257 } 3258 3259 static inline bool f2fs_skip_inode_update(struct inode *inode, int dsync) 3260 { 3261 bool ret; 3262 3263 if (dsync) { 3264 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 3265 3266 spin_lock(&sbi->inode_lock[DIRTY_META]); 3267 ret = list_empty(&F2FS_I(inode)->gdirty_list); 3268 spin_unlock(&sbi->inode_lock[DIRTY_META]); 3269 return ret; 3270 } 3271 if (!is_inode_flag_set(inode, FI_AUTO_RECOVER) || 3272 file_keep_isize(inode) || 3273 i_size_read(inode) & ~PAGE_MASK) 3274 return false; 3275 3276 if (!f2fs_is_time_consistent(inode)) 3277 return false; 3278 3279 spin_lock(&F2FS_I(inode)->i_size_lock); 3280 ret = F2FS_I(inode)->last_disk_size == i_size_read(inode); 3281 spin_unlock(&F2FS_I(inode)->i_size_lock); 3282 3283 return ret; 3284 } 3285 3286 static inline bool f2fs_readonly(struct super_block *sb) 3287 { 3288 return sb_rdonly(sb); 3289 } 3290 3291 static inline bool f2fs_cp_error(struct f2fs_sb_info *sbi) 3292 { 3293 return is_set_ckpt_flags(sbi, CP_ERROR_FLAG); 3294 } 3295 3296 static inline bool is_dot_dotdot(const u8 *name, size_t len) 3297 { 3298 if (len == 1 && name[0] == '.') 3299 return true; 3300 3301 if (len == 2 && name[0] == '.' && name[1] == '.') 3302 return true; 3303 3304 return false; 3305 } 3306 3307 static inline void *f2fs_kmalloc(struct f2fs_sb_info *sbi, 3308 size_t size, gfp_t flags) 3309 { 3310 if (time_to_inject(sbi, FAULT_KMALLOC)) { 3311 f2fs_show_injection_info(sbi, FAULT_KMALLOC); 3312 return NULL; 3313 } 3314 3315 return kmalloc(size, flags); 3316 } 3317 3318 static inline void *f2fs_kzalloc(struct f2fs_sb_info *sbi, 3319 size_t size, gfp_t flags) 3320 { 3321 return f2fs_kmalloc(sbi, size, flags | __GFP_ZERO); 3322 } 3323 3324 static inline void *f2fs_kvmalloc(struct f2fs_sb_info *sbi, 3325 size_t size, gfp_t flags) 3326 { 3327 if (time_to_inject(sbi, FAULT_KVMALLOC)) { 3328 f2fs_show_injection_info(sbi, FAULT_KVMALLOC); 3329 return NULL; 3330 } 3331 3332 return kvmalloc(size, flags); 3333 } 3334 3335 static inline void *f2fs_kvzalloc(struct f2fs_sb_info *sbi, 3336 size_t size, gfp_t flags) 3337 { 3338 return f2fs_kvmalloc(sbi, size, flags | __GFP_ZERO); 3339 } 3340 3341 static inline int get_extra_isize(struct inode *inode) 3342 { 3343 return F2FS_I(inode)->i_extra_isize / sizeof(__le32); 3344 } 3345 3346 static inline int get_inline_xattr_addrs(struct inode *inode) 3347 { 3348 return F2FS_I(inode)->i_inline_xattr_size; 3349 } 3350 3351 #define f2fs_get_inode_mode(i) \ 3352 ((is_inode_flag_set(i, FI_ACL_MODE)) ? \ 3353 (F2FS_I(i)->i_acl_mode) : ((i)->i_mode)) 3354 3355 #define F2FS_TOTAL_EXTRA_ATTR_SIZE \ 3356 (offsetof(struct f2fs_inode, i_extra_end) - \ 3357 offsetof(struct f2fs_inode, i_extra_isize)) \ 3358 3359 #define F2FS_OLD_ATTRIBUTE_SIZE (offsetof(struct f2fs_inode, i_addr)) 3360 #define F2FS_FITS_IN_INODE(f2fs_inode, extra_isize, field) \ 3361 ((offsetof(typeof(*(f2fs_inode)), field) + \ 3362 sizeof((f2fs_inode)->field)) \ 3363 <= (F2FS_OLD_ATTRIBUTE_SIZE + (extra_isize))) \ 3364 3365 #define __is_large_section(sbi) ((sbi)->segs_per_sec > 1) 3366 3367 #define __is_meta_io(fio) (PAGE_TYPE_OF_BIO((fio)->type) == META) 3368 3369 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi, 3370 block_t blkaddr, int type); 3371 static inline void verify_blkaddr(struct f2fs_sb_info *sbi, 3372 block_t blkaddr, int type) 3373 { 3374 if (!f2fs_is_valid_blkaddr(sbi, blkaddr, type)) { 3375 f2fs_err(sbi, "invalid blkaddr: %u, type: %d, run fsck to fix.", 3376 blkaddr, type); 3377 f2fs_bug_on(sbi, 1); 3378 } 3379 } 3380 3381 static inline bool __is_valid_data_blkaddr(block_t blkaddr) 3382 { 3383 if (blkaddr == NEW_ADDR || blkaddr == NULL_ADDR || 3384 blkaddr == COMPRESS_ADDR) 3385 return false; 3386 return true; 3387 } 3388 3389 /* 3390 * file.c 3391 */ 3392 int f2fs_sync_file(struct file *file, loff_t start, loff_t end, int datasync); 3393 void f2fs_truncate_data_blocks(struct dnode_of_data *dn); 3394 int f2fs_do_truncate_blocks(struct inode *inode, u64 from, bool lock); 3395 int f2fs_truncate_blocks(struct inode *inode, u64 from, bool lock); 3396 int f2fs_truncate(struct inode *inode); 3397 int f2fs_getattr(struct user_namespace *mnt_userns, const struct path *path, 3398 struct kstat *stat, u32 request_mask, unsigned int flags); 3399 int f2fs_setattr(struct user_namespace *mnt_userns, struct dentry *dentry, 3400 struct iattr *attr); 3401 int f2fs_truncate_hole(struct inode *inode, pgoff_t pg_start, pgoff_t pg_end); 3402 void f2fs_truncate_data_blocks_range(struct dnode_of_data *dn, int count); 3403 int f2fs_precache_extents(struct inode *inode); 3404 int f2fs_fileattr_get(struct dentry *dentry, struct fileattr *fa); 3405 int f2fs_fileattr_set(struct user_namespace *mnt_userns, 3406 struct dentry *dentry, struct fileattr *fa); 3407 long f2fs_ioctl(struct file *filp, unsigned int cmd, unsigned long arg); 3408 long f2fs_compat_ioctl(struct file *file, unsigned int cmd, unsigned long arg); 3409 int f2fs_transfer_project_quota(struct inode *inode, kprojid_t kprojid); 3410 int f2fs_pin_file_control(struct inode *inode, bool inc); 3411 3412 /* 3413 * inode.c 3414 */ 3415 void f2fs_set_inode_flags(struct inode *inode); 3416 bool f2fs_inode_chksum_verify(struct f2fs_sb_info *sbi, struct page *page); 3417 void f2fs_inode_chksum_set(struct f2fs_sb_info *sbi, struct page *page); 3418 struct inode *f2fs_iget(struct super_block *sb, unsigned long ino); 3419 struct inode *f2fs_iget_retry(struct super_block *sb, unsigned long ino); 3420 int f2fs_try_to_free_nats(struct f2fs_sb_info *sbi, int nr_shrink); 3421 void f2fs_update_inode(struct inode *inode, struct page *node_page); 3422 void f2fs_update_inode_page(struct inode *inode); 3423 int f2fs_write_inode(struct inode *inode, struct writeback_control *wbc); 3424 void f2fs_evict_inode(struct inode *inode); 3425 void f2fs_handle_failed_inode(struct inode *inode); 3426 3427 /* 3428 * namei.c 3429 */ 3430 int f2fs_update_extension_list(struct f2fs_sb_info *sbi, const char *name, 3431 bool hot, bool set); 3432 struct dentry *f2fs_get_parent(struct dentry *child); 3433 3434 /* 3435 * dir.c 3436 */ 3437 unsigned char f2fs_get_de_type(struct f2fs_dir_entry *de); 3438 int f2fs_init_casefolded_name(const struct inode *dir, 3439 struct f2fs_filename *fname); 3440 int f2fs_setup_filename(struct inode *dir, const struct qstr *iname, 3441 int lookup, struct f2fs_filename *fname); 3442 int f2fs_prepare_lookup(struct inode *dir, struct dentry *dentry, 3443 struct f2fs_filename *fname); 3444 void f2fs_free_filename(struct f2fs_filename *fname); 3445 struct f2fs_dir_entry *f2fs_find_target_dentry(const struct f2fs_dentry_ptr *d, 3446 const struct f2fs_filename *fname, int *max_slots); 3447 int f2fs_fill_dentries(struct dir_context *ctx, struct f2fs_dentry_ptr *d, 3448 unsigned int start_pos, struct fscrypt_str *fstr); 3449 void f2fs_do_make_empty_dir(struct inode *inode, struct inode *parent, 3450 struct f2fs_dentry_ptr *d); 3451 struct page *f2fs_init_inode_metadata(struct inode *inode, struct inode *dir, 3452 const struct f2fs_filename *fname, struct page *dpage); 3453 void f2fs_update_parent_metadata(struct inode *dir, struct inode *inode, 3454 unsigned int current_depth); 3455 int f2fs_room_for_filename(const void *bitmap, int slots, int max_slots); 3456 void f2fs_drop_nlink(struct inode *dir, struct inode *inode); 3457 struct f2fs_dir_entry *__f2fs_find_entry(struct inode *dir, 3458 const struct f2fs_filename *fname, 3459 struct page **res_page); 3460 struct f2fs_dir_entry *f2fs_find_entry(struct inode *dir, 3461 const struct qstr *child, struct page **res_page); 3462 struct f2fs_dir_entry *f2fs_parent_dir(struct inode *dir, struct page **p); 3463 ino_t f2fs_inode_by_name(struct inode *dir, const struct qstr *qstr, 3464 struct page **page); 3465 void f2fs_set_link(struct inode *dir, struct f2fs_dir_entry *de, 3466 struct page *page, struct inode *inode); 3467 bool f2fs_has_enough_room(struct inode *dir, struct page *ipage, 3468 const struct f2fs_filename *fname); 3469 void f2fs_update_dentry(nid_t ino, umode_t mode, struct f2fs_dentry_ptr *d, 3470 const struct fscrypt_str *name, f2fs_hash_t name_hash, 3471 unsigned int bit_pos); 3472 int f2fs_add_regular_entry(struct inode *dir, const struct f2fs_filename *fname, 3473 struct inode *inode, nid_t ino, umode_t mode); 3474 int f2fs_add_dentry(struct inode *dir, const struct f2fs_filename *fname, 3475 struct inode *inode, nid_t ino, umode_t mode); 3476 int f2fs_do_add_link(struct inode *dir, const struct qstr *name, 3477 struct inode *inode, nid_t ino, umode_t mode); 3478 void f2fs_delete_entry(struct f2fs_dir_entry *dentry, struct page *page, 3479 struct inode *dir, struct inode *inode); 3480 int f2fs_do_tmpfile(struct inode *inode, struct inode *dir); 3481 bool f2fs_empty_dir(struct inode *dir); 3482 3483 static inline int f2fs_add_link(struct dentry *dentry, struct inode *inode) 3484 { 3485 if (fscrypt_is_nokey_name(dentry)) 3486 return -ENOKEY; 3487 return f2fs_do_add_link(d_inode(dentry->d_parent), &dentry->d_name, 3488 inode, inode->i_ino, inode->i_mode); 3489 } 3490 3491 /* 3492 * super.c 3493 */ 3494 int f2fs_inode_dirtied(struct inode *inode, bool sync); 3495 void f2fs_inode_synced(struct inode *inode); 3496 int f2fs_dquot_initialize(struct inode *inode); 3497 int f2fs_enable_quota_files(struct f2fs_sb_info *sbi, bool rdonly); 3498 int f2fs_quota_sync(struct super_block *sb, int type); 3499 loff_t max_file_blocks(struct inode *inode); 3500 void f2fs_quota_off_umount(struct super_block *sb); 3501 int f2fs_commit_super(struct f2fs_sb_info *sbi, bool recover); 3502 int f2fs_sync_fs(struct super_block *sb, int sync); 3503 int f2fs_sanity_check_ckpt(struct f2fs_sb_info *sbi); 3504 3505 /* 3506 * hash.c 3507 */ 3508 void f2fs_hash_filename(const struct inode *dir, struct f2fs_filename *fname); 3509 3510 /* 3511 * node.c 3512 */ 3513 struct node_info; 3514 3515 int f2fs_check_nid_range(struct f2fs_sb_info *sbi, nid_t nid); 3516 bool f2fs_available_free_memory(struct f2fs_sb_info *sbi, int type); 3517 bool f2fs_in_warm_node_list(struct f2fs_sb_info *sbi, struct page *page); 3518 void f2fs_init_fsync_node_info(struct f2fs_sb_info *sbi); 3519 void f2fs_del_fsync_node_entry(struct f2fs_sb_info *sbi, struct page *page); 3520 void f2fs_reset_fsync_node_info(struct f2fs_sb_info *sbi); 3521 int f2fs_need_dentry_mark(struct f2fs_sb_info *sbi, nid_t nid); 3522 bool f2fs_is_checkpointed_node(struct f2fs_sb_info *sbi, nid_t nid); 3523 bool f2fs_need_inode_block_update(struct f2fs_sb_info *sbi, nid_t ino); 3524 int f2fs_get_node_info(struct f2fs_sb_info *sbi, nid_t nid, 3525 struct node_info *ni, bool checkpoint_context); 3526 pgoff_t f2fs_get_next_page_offset(struct dnode_of_data *dn, pgoff_t pgofs); 3527 int f2fs_get_dnode_of_data(struct dnode_of_data *dn, pgoff_t index, int mode); 3528 int f2fs_truncate_inode_blocks(struct inode *inode, pgoff_t from); 3529 int f2fs_truncate_xattr_node(struct inode *inode); 3530 int f2fs_wait_on_node_pages_writeback(struct f2fs_sb_info *sbi, 3531 unsigned int seq_id); 3532 bool f2fs_nat_bitmap_enabled(struct f2fs_sb_info *sbi); 3533 int f2fs_remove_inode_page(struct inode *inode); 3534 struct page *f2fs_new_inode_page(struct inode *inode); 3535 struct page *f2fs_new_node_page(struct dnode_of_data *dn, unsigned int ofs); 3536 void f2fs_ra_node_page(struct f2fs_sb_info *sbi, nid_t nid); 3537 struct page *f2fs_get_node_page(struct f2fs_sb_info *sbi, pgoff_t nid); 3538 struct page *f2fs_get_node_page_ra(struct page *parent, int start); 3539 int f2fs_move_node_page(struct page *node_page, int gc_type); 3540 void f2fs_flush_inline_data(struct f2fs_sb_info *sbi); 3541 int f2fs_fsync_node_pages(struct f2fs_sb_info *sbi, struct inode *inode, 3542 struct writeback_control *wbc, bool atomic, 3543 unsigned int *seq_id); 3544 int f2fs_sync_node_pages(struct f2fs_sb_info *sbi, 3545 struct writeback_control *wbc, 3546 bool do_balance, enum iostat_type io_type); 3547 int f2fs_build_free_nids(struct f2fs_sb_info *sbi, bool sync, bool mount); 3548 bool f2fs_alloc_nid(struct f2fs_sb_info *sbi, nid_t *nid); 3549 void f2fs_alloc_nid_done(struct f2fs_sb_info *sbi, nid_t nid); 3550 void f2fs_alloc_nid_failed(struct f2fs_sb_info *sbi, nid_t nid); 3551 int f2fs_try_to_free_nids(struct f2fs_sb_info *sbi, int nr_shrink); 3552 int f2fs_recover_inline_xattr(struct inode *inode, struct page *page); 3553 int f2fs_recover_xattr_data(struct inode *inode, struct page *page); 3554 int f2fs_recover_inode_page(struct f2fs_sb_info *sbi, struct page *page); 3555 int f2fs_restore_node_summary(struct f2fs_sb_info *sbi, 3556 unsigned int segno, struct f2fs_summary_block *sum); 3557 void f2fs_enable_nat_bits(struct f2fs_sb_info *sbi); 3558 int f2fs_flush_nat_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3559 int f2fs_build_node_manager(struct f2fs_sb_info *sbi); 3560 void f2fs_destroy_node_manager(struct f2fs_sb_info *sbi); 3561 int __init f2fs_create_node_manager_caches(void); 3562 void f2fs_destroy_node_manager_caches(void); 3563 3564 /* 3565 * segment.c 3566 */ 3567 bool f2fs_need_SSR(struct f2fs_sb_info *sbi); 3568 void f2fs_register_inmem_page(struct inode *inode, struct page *page); 3569 void f2fs_drop_inmem_pages_all(struct f2fs_sb_info *sbi, bool gc_failure); 3570 void f2fs_drop_inmem_pages(struct inode *inode); 3571 void f2fs_drop_inmem_page(struct inode *inode, struct page *page); 3572 int f2fs_commit_inmem_pages(struct inode *inode); 3573 void f2fs_balance_fs(struct f2fs_sb_info *sbi, bool need); 3574 void f2fs_balance_fs_bg(struct f2fs_sb_info *sbi, bool from_bg); 3575 int f2fs_issue_flush(struct f2fs_sb_info *sbi, nid_t ino); 3576 int f2fs_create_flush_cmd_control(struct f2fs_sb_info *sbi); 3577 int f2fs_flush_device_cache(struct f2fs_sb_info *sbi); 3578 void f2fs_destroy_flush_cmd_control(struct f2fs_sb_info *sbi, bool free); 3579 void f2fs_invalidate_blocks(struct f2fs_sb_info *sbi, block_t addr); 3580 bool f2fs_is_checkpointed_data(struct f2fs_sb_info *sbi, block_t blkaddr); 3581 int f2fs_start_discard_thread(struct f2fs_sb_info *sbi); 3582 void f2fs_drop_discard_cmd(struct f2fs_sb_info *sbi); 3583 void f2fs_stop_discard_thread(struct f2fs_sb_info *sbi); 3584 bool f2fs_issue_discard_timeout(struct f2fs_sb_info *sbi); 3585 void f2fs_clear_prefree_segments(struct f2fs_sb_info *sbi, 3586 struct cp_control *cpc); 3587 void f2fs_dirty_to_prefree(struct f2fs_sb_info *sbi); 3588 block_t f2fs_get_unusable_blocks(struct f2fs_sb_info *sbi); 3589 int f2fs_disable_cp_again(struct f2fs_sb_info *sbi, block_t unusable); 3590 void f2fs_release_discard_addrs(struct f2fs_sb_info *sbi); 3591 int f2fs_npages_for_summary_flush(struct f2fs_sb_info *sbi, bool for_ra); 3592 bool f2fs_segment_has_free_slot(struct f2fs_sb_info *sbi, int segno); 3593 void f2fs_init_inmem_curseg(struct f2fs_sb_info *sbi); 3594 void f2fs_save_inmem_curseg(struct f2fs_sb_info *sbi); 3595 void f2fs_restore_inmem_curseg(struct f2fs_sb_info *sbi); 3596 void f2fs_get_new_segment(struct f2fs_sb_info *sbi, 3597 unsigned int *newseg, bool new_sec, int dir); 3598 void f2fs_allocate_segment_for_resize(struct f2fs_sb_info *sbi, int type, 3599 unsigned int start, unsigned int end); 3600 void f2fs_allocate_new_section(struct f2fs_sb_info *sbi, int type, bool force); 3601 void f2fs_allocate_new_segments(struct f2fs_sb_info *sbi); 3602 int f2fs_trim_fs(struct f2fs_sb_info *sbi, struct fstrim_range *range); 3603 bool f2fs_exist_trim_candidates(struct f2fs_sb_info *sbi, 3604 struct cp_control *cpc); 3605 struct page *f2fs_get_sum_page(struct f2fs_sb_info *sbi, unsigned int segno); 3606 void f2fs_update_meta_page(struct f2fs_sb_info *sbi, void *src, 3607 block_t blk_addr); 3608 void f2fs_do_write_meta_page(struct f2fs_sb_info *sbi, struct page *page, 3609 enum iostat_type io_type); 3610 void f2fs_do_write_node_page(unsigned int nid, struct f2fs_io_info *fio); 3611 void f2fs_outplace_write_data(struct dnode_of_data *dn, 3612 struct f2fs_io_info *fio); 3613 int f2fs_inplace_write_data(struct f2fs_io_info *fio); 3614 void f2fs_do_replace_block(struct f2fs_sb_info *sbi, struct f2fs_summary *sum, 3615 block_t old_blkaddr, block_t new_blkaddr, 3616 bool recover_curseg, bool recover_newaddr, 3617 bool from_gc); 3618 void f2fs_replace_block(struct f2fs_sb_info *sbi, struct dnode_of_data *dn, 3619 block_t old_addr, block_t new_addr, 3620 unsigned char version, bool recover_curseg, 3621 bool recover_newaddr); 3622 void f2fs_allocate_data_block(struct f2fs_sb_info *sbi, struct page *page, 3623 block_t old_blkaddr, block_t *new_blkaddr, 3624 struct f2fs_summary *sum, int type, 3625 struct f2fs_io_info *fio); 3626 void f2fs_update_device_state(struct f2fs_sb_info *sbi, nid_t ino, 3627 block_t blkaddr, unsigned int blkcnt); 3628 void f2fs_wait_on_page_writeback(struct page *page, 3629 enum page_type type, bool ordered, bool locked); 3630 void f2fs_wait_on_block_writeback(struct inode *inode, block_t blkaddr); 3631 void f2fs_wait_on_block_writeback_range(struct inode *inode, block_t blkaddr, 3632 block_t len); 3633 void f2fs_write_data_summaries(struct f2fs_sb_info *sbi, block_t start_blk); 3634 void f2fs_write_node_summaries(struct f2fs_sb_info *sbi, block_t start_blk); 3635 int f2fs_lookup_journal_in_cursum(struct f2fs_journal *journal, int type, 3636 unsigned int val, int alloc); 3637 void f2fs_flush_sit_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3638 int f2fs_fix_curseg_write_pointer(struct f2fs_sb_info *sbi); 3639 int f2fs_check_write_pointer(struct f2fs_sb_info *sbi); 3640 int f2fs_build_segment_manager(struct f2fs_sb_info *sbi); 3641 void f2fs_destroy_segment_manager(struct f2fs_sb_info *sbi); 3642 int __init f2fs_create_segment_manager_caches(void); 3643 void f2fs_destroy_segment_manager_caches(void); 3644 int f2fs_rw_hint_to_seg_type(enum rw_hint hint); 3645 enum rw_hint f2fs_io_type_to_rw_hint(struct f2fs_sb_info *sbi, 3646 enum page_type type, enum temp_type temp); 3647 unsigned int f2fs_usable_segs_in_sec(struct f2fs_sb_info *sbi, 3648 unsigned int segno); 3649 unsigned int f2fs_usable_blks_in_seg(struct f2fs_sb_info *sbi, 3650 unsigned int segno); 3651 3652 #define DEF_FRAGMENT_SIZE 4 3653 #define MIN_FRAGMENT_SIZE 1 3654 #define MAX_FRAGMENT_SIZE 512 3655 3656 static inline bool f2fs_need_rand_seg(struct f2fs_sb_info *sbi) 3657 { 3658 return F2FS_OPTION(sbi).fs_mode == FS_MODE_FRAGMENT_SEG || 3659 F2FS_OPTION(sbi).fs_mode == FS_MODE_FRAGMENT_BLK; 3660 } 3661 3662 /* 3663 * checkpoint.c 3664 */ 3665 void f2fs_stop_checkpoint(struct f2fs_sb_info *sbi, bool end_io); 3666 struct page *f2fs_grab_meta_page(struct f2fs_sb_info *sbi, pgoff_t index); 3667 struct page *f2fs_get_meta_page(struct f2fs_sb_info *sbi, pgoff_t index); 3668 struct page *f2fs_get_meta_page_retry(struct f2fs_sb_info *sbi, pgoff_t index); 3669 struct page *f2fs_get_tmp_page(struct f2fs_sb_info *sbi, pgoff_t index); 3670 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi, 3671 block_t blkaddr, int type); 3672 int f2fs_ra_meta_pages(struct f2fs_sb_info *sbi, block_t start, int nrpages, 3673 int type, bool sync); 3674 void f2fs_ra_meta_pages_cond(struct f2fs_sb_info *sbi, pgoff_t index, 3675 unsigned int ra_blocks); 3676 long f2fs_sync_meta_pages(struct f2fs_sb_info *sbi, enum page_type type, 3677 long nr_to_write, enum iostat_type io_type); 3678 void f2fs_add_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type); 3679 void f2fs_remove_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type); 3680 void f2fs_release_ino_entry(struct f2fs_sb_info *sbi, bool all); 3681 bool f2fs_exist_written_data(struct f2fs_sb_info *sbi, nid_t ino, int mode); 3682 void f2fs_set_dirty_device(struct f2fs_sb_info *sbi, nid_t ino, 3683 unsigned int devidx, int type); 3684 bool f2fs_is_dirty_device(struct f2fs_sb_info *sbi, nid_t ino, 3685 unsigned int devidx, int type); 3686 int f2fs_sync_inode_meta(struct f2fs_sb_info *sbi); 3687 int f2fs_acquire_orphan_inode(struct f2fs_sb_info *sbi); 3688 void f2fs_release_orphan_inode(struct f2fs_sb_info *sbi); 3689 void f2fs_add_orphan_inode(struct inode *inode); 3690 void f2fs_remove_orphan_inode(struct f2fs_sb_info *sbi, nid_t ino); 3691 int f2fs_recover_orphan_inodes(struct f2fs_sb_info *sbi); 3692 int f2fs_get_valid_checkpoint(struct f2fs_sb_info *sbi); 3693 void f2fs_update_dirty_page(struct inode *inode, struct page *page); 3694 void f2fs_remove_dirty_inode(struct inode *inode); 3695 int f2fs_sync_dirty_inodes(struct f2fs_sb_info *sbi, enum inode_type type); 3696 void f2fs_wait_on_all_pages(struct f2fs_sb_info *sbi, int type); 3697 u64 f2fs_get_sectors_written(struct f2fs_sb_info *sbi); 3698 int f2fs_write_checkpoint(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3699 void f2fs_init_ino_entry_info(struct f2fs_sb_info *sbi); 3700 int __init f2fs_create_checkpoint_caches(void); 3701 void f2fs_destroy_checkpoint_caches(void); 3702 int f2fs_issue_checkpoint(struct f2fs_sb_info *sbi); 3703 int f2fs_start_ckpt_thread(struct f2fs_sb_info *sbi); 3704 void f2fs_stop_ckpt_thread(struct f2fs_sb_info *sbi); 3705 void f2fs_init_ckpt_req_control(struct f2fs_sb_info *sbi); 3706 3707 /* 3708 * data.c 3709 */ 3710 int __init f2fs_init_bioset(void); 3711 void f2fs_destroy_bioset(void); 3712 int f2fs_init_bio_entry_cache(void); 3713 void f2fs_destroy_bio_entry_cache(void); 3714 void f2fs_submit_bio(struct f2fs_sb_info *sbi, 3715 struct bio *bio, enum page_type type); 3716 void f2fs_submit_merged_write(struct f2fs_sb_info *sbi, enum page_type type); 3717 void f2fs_submit_merged_write_cond(struct f2fs_sb_info *sbi, 3718 struct inode *inode, struct page *page, 3719 nid_t ino, enum page_type type); 3720 void f2fs_submit_merged_ipu_write(struct f2fs_sb_info *sbi, 3721 struct bio **bio, struct page *page); 3722 void f2fs_flush_merged_writes(struct f2fs_sb_info *sbi); 3723 int f2fs_submit_page_bio(struct f2fs_io_info *fio); 3724 int f2fs_merge_page_bio(struct f2fs_io_info *fio); 3725 void f2fs_submit_page_write(struct f2fs_io_info *fio); 3726 struct block_device *f2fs_target_device(struct f2fs_sb_info *sbi, 3727 block_t blk_addr, struct bio *bio); 3728 int f2fs_target_device_index(struct f2fs_sb_info *sbi, block_t blkaddr); 3729 void f2fs_set_data_blkaddr(struct dnode_of_data *dn); 3730 void f2fs_update_data_blkaddr(struct dnode_of_data *dn, block_t blkaddr); 3731 int f2fs_reserve_new_blocks(struct dnode_of_data *dn, blkcnt_t count); 3732 int f2fs_reserve_new_block(struct dnode_of_data *dn); 3733 int f2fs_get_block(struct dnode_of_data *dn, pgoff_t index); 3734 int f2fs_reserve_block(struct dnode_of_data *dn, pgoff_t index); 3735 struct page *f2fs_get_read_data_page(struct inode *inode, pgoff_t index, 3736 int op_flags, bool for_write); 3737 struct page *f2fs_find_data_page(struct inode *inode, pgoff_t index); 3738 struct page *f2fs_get_lock_data_page(struct inode *inode, pgoff_t index, 3739 bool for_write); 3740 struct page *f2fs_get_new_data_page(struct inode *inode, 3741 struct page *ipage, pgoff_t index, bool new_i_size); 3742 int f2fs_do_write_data_page(struct f2fs_io_info *fio); 3743 void f2fs_do_map_lock(struct f2fs_sb_info *sbi, int flag, bool lock); 3744 int f2fs_map_blocks(struct inode *inode, struct f2fs_map_blocks *map, 3745 int create, int flag); 3746 int f2fs_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, 3747 u64 start, u64 len); 3748 int f2fs_encrypt_one_page(struct f2fs_io_info *fio); 3749 bool f2fs_should_update_inplace(struct inode *inode, struct f2fs_io_info *fio); 3750 bool f2fs_should_update_outplace(struct inode *inode, struct f2fs_io_info *fio); 3751 int f2fs_write_single_data_page(struct page *page, int *submitted, 3752 struct bio **bio, sector_t *last_block, 3753 struct writeback_control *wbc, 3754 enum iostat_type io_type, 3755 int compr_blocks, bool allow_balance); 3756 void f2fs_write_failed(struct inode *inode, loff_t to); 3757 void f2fs_invalidate_page(struct page *page, unsigned int offset, 3758 unsigned int length); 3759 int f2fs_release_page(struct page *page, gfp_t wait); 3760 #ifdef CONFIG_MIGRATION 3761 int f2fs_migrate_page(struct address_space *mapping, struct page *newpage, 3762 struct page *page, enum migrate_mode mode); 3763 #endif 3764 bool f2fs_overwrite_io(struct inode *inode, loff_t pos, size_t len); 3765 void f2fs_clear_page_cache_dirty_tag(struct page *page); 3766 int f2fs_init_post_read_processing(void); 3767 void f2fs_destroy_post_read_processing(void); 3768 int f2fs_init_post_read_wq(struct f2fs_sb_info *sbi); 3769 void f2fs_destroy_post_read_wq(struct f2fs_sb_info *sbi); 3770 extern const struct iomap_ops f2fs_iomap_ops; 3771 3772 /* 3773 * gc.c 3774 */ 3775 int f2fs_start_gc_thread(struct f2fs_sb_info *sbi); 3776 void f2fs_stop_gc_thread(struct f2fs_sb_info *sbi); 3777 block_t f2fs_start_bidx_of_node(unsigned int node_ofs, struct inode *inode); 3778 int f2fs_gc(struct f2fs_sb_info *sbi, bool sync, bool background, bool force, 3779 unsigned int segno); 3780 void f2fs_build_gc_manager(struct f2fs_sb_info *sbi); 3781 int f2fs_resize_fs(struct f2fs_sb_info *sbi, __u64 block_count); 3782 int __init f2fs_create_garbage_collection_cache(void); 3783 void f2fs_destroy_garbage_collection_cache(void); 3784 3785 /* 3786 * recovery.c 3787 */ 3788 int f2fs_recover_fsync_data(struct f2fs_sb_info *sbi, bool check_only); 3789 bool f2fs_space_for_roll_forward(struct f2fs_sb_info *sbi); 3790 int __init f2fs_create_recovery_cache(void); 3791 void f2fs_destroy_recovery_cache(void); 3792 3793 /* 3794 * debug.c 3795 */ 3796 #ifdef CONFIG_F2FS_STAT_FS 3797 struct f2fs_stat_info { 3798 struct list_head stat_list; 3799 struct f2fs_sb_info *sbi; 3800 int all_area_segs, sit_area_segs, nat_area_segs, ssa_area_segs; 3801 int main_area_segs, main_area_sections, main_area_zones; 3802 unsigned long long hit_largest, hit_cached, hit_rbtree; 3803 unsigned long long hit_total, total_ext; 3804 int ext_tree, zombie_tree, ext_node; 3805 int ndirty_node, ndirty_dent, ndirty_meta, ndirty_imeta; 3806 int ndirty_data, ndirty_qdata; 3807 int inmem_pages; 3808 unsigned int ndirty_dirs, ndirty_files, nquota_files, ndirty_all; 3809 int nats, dirty_nats, sits, dirty_sits; 3810 int free_nids, avail_nids, alloc_nids; 3811 int total_count, utilization; 3812 int bg_gc, nr_wb_cp_data, nr_wb_data; 3813 int nr_rd_data, nr_rd_node, nr_rd_meta; 3814 int nr_dio_read, nr_dio_write; 3815 unsigned int io_skip_bggc, other_skip_bggc; 3816 int nr_flushing, nr_flushed, flush_list_empty; 3817 int nr_discarding, nr_discarded; 3818 int nr_discard_cmd; 3819 unsigned int undiscard_blks; 3820 int nr_issued_ckpt, nr_total_ckpt, nr_queued_ckpt; 3821 unsigned int cur_ckpt_time, peak_ckpt_time; 3822 int inline_xattr, inline_inode, inline_dir, append, update, orphans; 3823 int compr_inode; 3824 unsigned long long compr_blocks; 3825 int aw_cnt, max_aw_cnt, vw_cnt, max_vw_cnt; 3826 unsigned int valid_count, valid_node_count, valid_inode_count, discard_blks; 3827 unsigned int bimodal, avg_vblocks; 3828 int util_free, util_valid, util_invalid; 3829 int rsvd_segs, overp_segs; 3830 int dirty_count, node_pages, meta_pages, compress_pages; 3831 int compress_page_hit; 3832 int prefree_count, call_count, cp_count, bg_cp_count; 3833 int tot_segs, node_segs, data_segs, free_segs, free_secs; 3834 int bg_node_segs, bg_data_segs; 3835 int tot_blks, data_blks, node_blks; 3836 int bg_data_blks, bg_node_blks; 3837 unsigned long long skipped_atomic_files[2]; 3838 int curseg[NR_CURSEG_TYPE]; 3839 int cursec[NR_CURSEG_TYPE]; 3840 int curzone[NR_CURSEG_TYPE]; 3841 unsigned int dirty_seg[NR_CURSEG_TYPE]; 3842 unsigned int full_seg[NR_CURSEG_TYPE]; 3843 unsigned int valid_blks[NR_CURSEG_TYPE]; 3844 3845 unsigned int meta_count[META_MAX]; 3846 unsigned int segment_count[2]; 3847 unsigned int block_count[2]; 3848 unsigned int inplace_count; 3849 unsigned long long base_mem, cache_mem, page_mem; 3850 }; 3851 3852 static inline struct f2fs_stat_info *F2FS_STAT(struct f2fs_sb_info *sbi) 3853 { 3854 return (struct f2fs_stat_info *)sbi->stat_info; 3855 } 3856 3857 #define stat_inc_cp_count(si) ((si)->cp_count++) 3858 #define stat_inc_bg_cp_count(si) ((si)->bg_cp_count++) 3859 #define stat_inc_call_count(si) ((si)->call_count++) 3860 #define stat_inc_bggc_count(si) ((si)->bg_gc++) 3861 #define stat_io_skip_bggc_count(sbi) ((sbi)->io_skip_bggc++) 3862 #define stat_other_skip_bggc_count(sbi) ((sbi)->other_skip_bggc++) 3863 #define stat_inc_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]++) 3864 #define stat_dec_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]--) 3865 #define stat_inc_total_hit(sbi) (atomic64_inc(&(sbi)->total_hit_ext)) 3866 #define stat_inc_rbtree_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_rbtree)) 3867 #define stat_inc_largest_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_largest)) 3868 #define stat_inc_cached_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_cached)) 3869 #define stat_inc_inline_xattr(inode) \ 3870 do { \ 3871 if (f2fs_has_inline_xattr(inode)) \ 3872 (atomic_inc(&F2FS_I_SB(inode)->inline_xattr)); \ 3873 } while (0) 3874 #define stat_dec_inline_xattr(inode) \ 3875 do { \ 3876 if (f2fs_has_inline_xattr(inode)) \ 3877 (atomic_dec(&F2FS_I_SB(inode)->inline_xattr)); \ 3878 } while (0) 3879 #define stat_inc_inline_inode(inode) \ 3880 do { \ 3881 if (f2fs_has_inline_data(inode)) \ 3882 (atomic_inc(&F2FS_I_SB(inode)->inline_inode)); \ 3883 } while (0) 3884 #define stat_dec_inline_inode(inode) \ 3885 do { \ 3886 if (f2fs_has_inline_data(inode)) \ 3887 (atomic_dec(&F2FS_I_SB(inode)->inline_inode)); \ 3888 } while (0) 3889 #define stat_inc_inline_dir(inode) \ 3890 do { \ 3891 if (f2fs_has_inline_dentry(inode)) \ 3892 (atomic_inc(&F2FS_I_SB(inode)->inline_dir)); \ 3893 } while (0) 3894 #define stat_dec_inline_dir(inode) \ 3895 do { \ 3896 if (f2fs_has_inline_dentry(inode)) \ 3897 (atomic_dec(&F2FS_I_SB(inode)->inline_dir)); \ 3898 } while (0) 3899 #define stat_inc_compr_inode(inode) \ 3900 do { \ 3901 if (f2fs_compressed_file(inode)) \ 3902 (atomic_inc(&F2FS_I_SB(inode)->compr_inode)); \ 3903 } while (0) 3904 #define stat_dec_compr_inode(inode) \ 3905 do { \ 3906 if (f2fs_compressed_file(inode)) \ 3907 (atomic_dec(&F2FS_I_SB(inode)->compr_inode)); \ 3908 } while (0) 3909 #define stat_add_compr_blocks(inode, blocks) \ 3910 (atomic64_add(blocks, &F2FS_I_SB(inode)->compr_blocks)) 3911 #define stat_sub_compr_blocks(inode, blocks) \ 3912 (atomic64_sub(blocks, &F2FS_I_SB(inode)->compr_blocks)) 3913 #define stat_inc_meta_count(sbi, blkaddr) \ 3914 do { \ 3915 if (blkaddr < SIT_I(sbi)->sit_base_addr) \ 3916 atomic_inc(&(sbi)->meta_count[META_CP]); \ 3917 else if (blkaddr < NM_I(sbi)->nat_blkaddr) \ 3918 atomic_inc(&(sbi)->meta_count[META_SIT]); \ 3919 else if (blkaddr < SM_I(sbi)->ssa_blkaddr) \ 3920 atomic_inc(&(sbi)->meta_count[META_NAT]); \ 3921 else if (blkaddr < SM_I(sbi)->main_blkaddr) \ 3922 atomic_inc(&(sbi)->meta_count[META_SSA]); \ 3923 } while (0) 3924 #define stat_inc_seg_type(sbi, curseg) \ 3925 ((sbi)->segment_count[(curseg)->alloc_type]++) 3926 #define stat_inc_block_count(sbi, curseg) \ 3927 ((sbi)->block_count[(curseg)->alloc_type]++) 3928 #define stat_inc_inplace_blocks(sbi) \ 3929 (atomic_inc(&(sbi)->inplace_count)) 3930 #define stat_update_max_atomic_write(inode) \ 3931 do { \ 3932 int cur = F2FS_I_SB(inode)->atomic_files; \ 3933 int max = atomic_read(&F2FS_I_SB(inode)->max_aw_cnt); \ 3934 if (cur > max) \ 3935 atomic_set(&F2FS_I_SB(inode)->max_aw_cnt, cur); \ 3936 } while (0) 3937 #define stat_inc_volatile_write(inode) \ 3938 (atomic_inc(&F2FS_I_SB(inode)->vw_cnt)) 3939 #define stat_dec_volatile_write(inode) \ 3940 (atomic_dec(&F2FS_I_SB(inode)->vw_cnt)) 3941 #define stat_update_max_volatile_write(inode) \ 3942 do { \ 3943 int cur = atomic_read(&F2FS_I_SB(inode)->vw_cnt); \ 3944 int max = atomic_read(&F2FS_I_SB(inode)->max_vw_cnt); \ 3945 if (cur > max) \ 3946 atomic_set(&F2FS_I_SB(inode)->max_vw_cnt, cur); \ 3947 } while (0) 3948 #define stat_inc_seg_count(sbi, type, gc_type) \ 3949 do { \ 3950 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 3951 si->tot_segs++; \ 3952 if ((type) == SUM_TYPE_DATA) { \ 3953 si->data_segs++; \ 3954 si->bg_data_segs += (gc_type == BG_GC) ? 1 : 0; \ 3955 } else { \ 3956 si->node_segs++; \ 3957 si->bg_node_segs += (gc_type == BG_GC) ? 1 : 0; \ 3958 } \ 3959 } while (0) 3960 3961 #define stat_inc_tot_blk_count(si, blks) \ 3962 ((si)->tot_blks += (blks)) 3963 3964 #define stat_inc_data_blk_count(sbi, blks, gc_type) \ 3965 do { \ 3966 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 3967 stat_inc_tot_blk_count(si, blks); \ 3968 si->data_blks += (blks); \ 3969 si->bg_data_blks += ((gc_type) == BG_GC) ? (blks) : 0; \ 3970 } while (0) 3971 3972 #define stat_inc_node_blk_count(sbi, blks, gc_type) \ 3973 do { \ 3974 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 3975 stat_inc_tot_blk_count(si, blks); \ 3976 si->node_blks += (blks); \ 3977 si->bg_node_blks += ((gc_type) == BG_GC) ? (blks) : 0; \ 3978 } while (0) 3979 3980 int f2fs_build_stats(struct f2fs_sb_info *sbi); 3981 void f2fs_destroy_stats(struct f2fs_sb_info *sbi); 3982 void __init f2fs_create_root_stats(void); 3983 void f2fs_destroy_root_stats(void); 3984 void f2fs_update_sit_info(struct f2fs_sb_info *sbi); 3985 #else 3986 #define stat_inc_cp_count(si) do { } while (0) 3987 #define stat_inc_bg_cp_count(si) do { } while (0) 3988 #define stat_inc_call_count(si) do { } while (0) 3989 #define stat_inc_bggc_count(si) do { } while (0) 3990 #define stat_io_skip_bggc_count(sbi) do { } while (0) 3991 #define stat_other_skip_bggc_count(sbi) do { } while (0) 3992 #define stat_inc_dirty_inode(sbi, type) do { } while (0) 3993 #define stat_dec_dirty_inode(sbi, type) do { } while (0) 3994 #define stat_inc_total_hit(sbi) do { } while (0) 3995 #define stat_inc_rbtree_node_hit(sbi) do { } while (0) 3996 #define stat_inc_largest_node_hit(sbi) do { } while (0) 3997 #define stat_inc_cached_node_hit(sbi) do { } while (0) 3998 #define stat_inc_inline_xattr(inode) do { } while (0) 3999 #define stat_dec_inline_xattr(inode) do { } while (0) 4000 #define stat_inc_inline_inode(inode) do { } while (0) 4001 #define stat_dec_inline_inode(inode) do { } while (0) 4002 #define stat_inc_inline_dir(inode) do { } while (0) 4003 #define stat_dec_inline_dir(inode) do { } while (0) 4004 #define stat_inc_compr_inode(inode) do { } while (0) 4005 #define stat_dec_compr_inode(inode) do { } while (0) 4006 #define stat_add_compr_blocks(inode, blocks) do { } while (0) 4007 #define stat_sub_compr_blocks(inode, blocks) do { } while (0) 4008 #define stat_update_max_atomic_write(inode) do { } while (0) 4009 #define stat_inc_volatile_write(inode) do { } while (0) 4010 #define stat_dec_volatile_write(inode) do { } while (0) 4011 #define stat_update_max_volatile_write(inode) do { } while (0) 4012 #define stat_inc_meta_count(sbi, blkaddr) do { } while (0) 4013 #define stat_inc_seg_type(sbi, curseg) do { } while (0) 4014 #define stat_inc_block_count(sbi, curseg) do { } while (0) 4015 #define stat_inc_inplace_blocks(sbi) do { } while (0) 4016 #define stat_inc_seg_count(sbi, type, gc_type) do { } while (0) 4017 #define stat_inc_tot_blk_count(si, blks) do { } while (0) 4018 #define stat_inc_data_blk_count(sbi, blks, gc_type) do { } while (0) 4019 #define stat_inc_node_blk_count(sbi, blks, gc_type) do { } while (0) 4020 4021 static inline int f2fs_build_stats(struct f2fs_sb_info *sbi) { return 0; } 4022 static inline void f2fs_destroy_stats(struct f2fs_sb_info *sbi) { } 4023 static inline void __init f2fs_create_root_stats(void) { } 4024 static inline void f2fs_destroy_root_stats(void) { } 4025 static inline void f2fs_update_sit_info(struct f2fs_sb_info *sbi) {} 4026 #endif 4027 4028 extern const struct file_operations f2fs_dir_operations; 4029 extern const struct file_operations f2fs_file_operations; 4030 extern const struct inode_operations f2fs_file_inode_operations; 4031 extern const struct address_space_operations f2fs_dblock_aops; 4032 extern const struct address_space_operations f2fs_node_aops; 4033 extern const struct address_space_operations f2fs_meta_aops; 4034 extern const struct inode_operations f2fs_dir_inode_operations; 4035 extern const struct inode_operations f2fs_symlink_inode_operations; 4036 extern const struct inode_operations f2fs_encrypted_symlink_inode_operations; 4037 extern const struct inode_operations f2fs_special_inode_operations; 4038 extern struct kmem_cache *f2fs_inode_entry_slab; 4039 4040 /* 4041 * inline.c 4042 */ 4043 bool f2fs_may_inline_data(struct inode *inode); 4044 bool f2fs_may_inline_dentry(struct inode *inode); 4045 void f2fs_do_read_inline_data(struct page *page, struct page *ipage); 4046 void f2fs_truncate_inline_inode(struct inode *inode, 4047 struct page *ipage, u64 from); 4048 int f2fs_read_inline_data(struct inode *inode, struct page *page); 4049 int f2fs_convert_inline_page(struct dnode_of_data *dn, struct page *page); 4050 int f2fs_convert_inline_inode(struct inode *inode); 4051 int f2fs_try_convert_inline_dir(struct inode *dir, struct dentry *dentry); 4052 int f2fs_write_inline_data(struct inode *inode, struct page *page); 4053 int f2fs_recover_inline_data(struct inode *inode, struct page *npage); 4054 struct f2fs_dir_entry *f2fs_find_in_inline_dir(struct inode *dir, 4055 const struct f2fs_filename *fname, 4056 struct page **res_page); 4057 int f2fs_make_empty_inline_dir(struct inode *inode, struct inode *parent, 4058 struct page *ipage); 4059 int f2fs_add_inline_entry(struct inode *dir, const struct f2fs_filename *fname, 4060 struct inode *inode, nid_t ino, umode_t mode); 4061 void f2fs_delete_inline_entry(struct f2fs_dir_entry *dentry, 4062 struct page *page, struct inode *dir, 4063 struct inode *inode); 4064 bool f2fs_empty_inline_dir(struct inode *dir); 4065 int f2fs_read_inline_dir(struct file *file, struct dir_context *ctx, 4066 struct fscrypt_str *fstr); 4067 int f2fs_inline_data_fiemap(struct inode *inode, 4068 struct fiemap_extent_info *fieinfo, 4069 __u64 start, __u64 len); 4070 4071 /* 4072 * shrinker.c 4073 */ 4074 unsigned long f2fs_shrink_count(struct shrinker *shrink, 4075 struct shrink_control *sc); 4076 unsigned long f2fs_shrink_scan(struct shrinker *shrink, 4077 struct shrink_control *sc); 4078 void f2fs_join_shrinker(struct f2fs_sb_info *sbi); 4079 void f2fs_leave_shrinker(struct f2fs_sb_info *sbi); 4080 4081 /* 4082 * extent_cache.c 4083 */ 4084 struct rb_entry *f2fs_lookup_rb_tree(struct rb_root_cached *root, 4085 struct rb_entry *cached_re, unsigned int ofs); 4086 struct rb_node **f2fs_lookup_rb_tree_ext(struct f2fs_sb_info *sbi, 4087 struct rb_root_cached *root, 4088 struct rb_node **parent, 4089 unsigned long long key, bool *left_most); 4090 struct rb_node **f2fs_lookup_rb_tree_for_insert(struct f2fs_sb_info *sbi, 4091 struct rb_root_cached *root, 4092 struct rb_node **parent, 4093 unsigned int ofs, bool *leftmost); 4094 struct rb_entry *f2fs_lookup_rb_tree_ret(struct rb_root_cached *root, 4095 struct rb_entry *cached_re, unsigned int ofs, 4096 struct rb_entry **prev_entry, struct rb_entry **next_entry, 4097 struct rb_node ***insert_p, struct rb_node **insert_parent, 4098 bool force, bool *leftmost); 4099 bool f2fs_check_rb_tree_consistence(struct f2fs_sb_info *sbi, 4100 struct rb_root_cached *root, bool check_key); 4101 unsigned int f2fs_shrink_extent_tree(struct f2fs_sb_info *sbi, int nr_shrink); 4102 void f2fs_init_extent_tree(struct inode *inode, struct page *ipage); 4103 void f2fs_drop_extent_tree(struct inode *inode); 4104 unsigned int f2fs_destroy_extent_node(struct inode *inode); 4105 void f2fs_destroy_extent_tree(struct inode *inode); 4106 bool f2fs_lookup_extent_cache(struct inode *inode, pgoff_t pgofs, 4107 struct extent_info *ei); 4108 void f2fs_update_extent_cache(struct dnode_of_data *dn); 4109 void f2fs_update_extent_cache_range(struct dnode_of_data *dn, 4110 pgoff_t fofs, block_t blkaddr, unsigned int len); 4111 void f2fs_init_extent_cache_info(struct f2fs_sb_info *sbi); 4112 int __init f2fs_create_extent_cache(void); 4113 void f2fs_destroy_extent_cache(void); 4114 4115 /* 4116 * sysfs.c 4117 */ 4118 #define MIN_RA_MUL 2 4119 #define MAX_RA_MUL 256 4120 4121 int __init f2fs_init_sysfs(void); 4122 void f2fs_exit_sysfs(void); 4123 int f2fs_register_sysfs(struct f2fs_sb_info *sbi); 4124 void f2fs_unregister_sysfs(struct f2fs_sb_info *sbi); 4125 4126 /* verity.c */ 4127 extern const struct fsverity_operations f2fs_verityops; 4128 4129 /* 4130 * crypto support 4131 */ 4132 static inline bool f2fs_encrypted_file(struct inode *inode) 4133 { 4134 return IS_ENCRYPTED(inode) && S_ISREG(inode->i_mode); 4135 } 4136 4137 static inline void f2fs_set_encrypted_inode(struct inode *inode) 4138 { 4139 #ifdef CONFIG_FS_ENCRYPTION 4140 file_set_encrypt(inode); 4141 f2fs_set_inode_flags(inode); 4142 #endif 4143 } 4144 4145 /* 4146 * Returns true if the reads of the inode's data need to undergo some 4147 * postprocessing step, like decryption or authenticity verification. 4148 */ 4149 static inline bool f2fs_post_read_required(struct inode *inode) 4150 { 4151 return f2fs_encrypted_file(inode) || fsverity_active(inode) || 4152 f2fs_compressed_file(inode); 4153 } 4154 4155 /* 4156 * compress.c 4157 */ 4158 #ifdef CONFIG_F2FS_FS_COMPRESSION 4159 bool f2fs_is_compressed_page(struct page *page); 4160 struct page *f2fs_compress_control_page(struct page *page); 4161 int f2fs_prepare_compress_overwrite(struct inode *inode, 4162 struct page **pagep, pgoff_t index, void **fsdata); 4163 bool f2fs_compress_write_end(struct inode *inode, void *fsdata, 4164 pgoff_t index, unsigned copied); 4165 int f2fs_truncate_partial_cluster(struct inode *inode, u64 from, bool lock); 4166 void f2fs_compress_write_end_io(struct bio *bio, struct page *page); 4167 bool f2fs_is_compress_backend_ready(struct inode *inode); 4168 int f2fs_init_compress_mempool(void); 4169 void f2fs_destroy_compress_mempool(void); 4170 void f2fs_decompress_cluster(struct decompress_io_ctx *dic); 4171 void f2fs_end_read_compressed_page(struct page *page, bool failed, 4172 block_t blkaddr); 4173 bool f2fs_cluster_is_empty(struct compress_ctx *cc); 4174 bool f2fs_cluster_can_merge_page(struct compress_ctx *cc, pgoff_t index); 4175 bool f2fs_all_cluster_page_loaded(struct compress_ctx *cc, struct pagevec *pvec, 4176 int index, int nr_pages); 4177 bool f2fs_sanity_check_cluster(struct dnode_of_data *dn); 4178 void f2fs_compress_ctx_add_page(struct compress_ctx *cc, struct page *page); 4179 int f2fs_write_multi_pages(struct compress_ctx *cc, 4180 int *submitted, 4181 struct writeback_control *wbc, 4182 enum iostat_type io_type); 4183 int f2fs_is_compressed_cluster(struct inode *inode, pgoff_t index); 4184 void f2fs_update_extent_tree_range_compressed(struct inode *inode, 4185 pgoff_t fofs, block_t blkaddr, unsigned int llen, 4186 unsigned int c_len); 4187 int f2fs_read_multi_pages(struct compress_ctx *cc, struct bio **bio_ret, 4188 unsigned nr_pages, sector_t *last_block_in_bio, 4189 bool is_readahead, bool for_write); 4190 struct decompress_io_ctx *f2fs_alloc_dic(struct compress_ctx *cc); 4191 void f2fs_decompress_end_io(struct decompress_io_ctx *dic, bool failed); 4192 void f2fs_put_page_dic(struct page *page); 4193 unsigned int f2fs_cluster_blocks_are_contiguous(struct dnode_of_data *dn); 4194 int f2fs_init_compress_ctx(struct compress_ctx *cc); 4195 void f2fs_destroy_compress_ctx(struct compress_ctx *cc, bool reuse); 4196 void f2fs_init_compress_info(struct f2fs_sb_info *sbi); 4197 int f2fs_init_compress_inode(struct f2fs_sb_info *sbi); 4198 void f2fs_destroy_compress_inode(struct f2fs_sb_info *sbi); 4199 int f2fs_init_page_array_cache(struct f2fs_sb_info *sbi); 4200 void f2fs_destroy_page_array_cache(struct f2fs_sb_info *sbi); 4201 int __init f2fs_init_compress_cache(void); 4202 void f2fs_destroy_compress_cache(void); 4203 struct address_space *COMPRESS_MAPPING(struct f2fs_sb_info *sbi); 4204 void f2fs_invalidate_compress_page(struct f2fs_sb_info *sbi, block_t blkaddr); 4205 void f2fs_cache_compressed_page(struct f2fs_sb_info *sbi, struct page *page, 4206 nid_t ino, block_t blkaddr); 4207 bool f2fs_load_compressed_page(struct f2fs_sb_info *sbi, struct page *page, 4208 block_t blkaddr); 4209 void f2fs_invalidate_compress_pages(struct f2fs_sb_info *sbi, nid_t ino); 4210 #define inc_compr_inode_stat(inode) \ 4211 do { \ 4212 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); \ 4213 sbi->compr_new_inode++; \ 4214 } while (0) 4215 #define add_compr_block_stat(inode, blocks) \ 4216 do { \ 4217 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); \ 4218 int diff = F2FS_I(inode)->i_cluster_size - blocks; \ 4219 sbi->compr_written_block += blocks; \ 4220 sbi->compr_saved_block += diff; \ 4221 } while (0) 4222 #else 4223 static inline bool f2fs_is_compressed_page(struct page *page) { return false; } 4224 static inline bool f2fs_is_compress_backend_ready(struct inode *inode) 4225 { 4226 if (!f2fs_compressed_file(inode)) 4227 return true; 4228 /* not support compression */ 4229 return false; 4230 } 4231 static inline struct page *f2fs_compress_control_page(struct page *page) 4232 { 4233 WARN_ON_ONCE(1); 4234 return ERR_PTR(-EINVAL); 4235 } 4236 static inline int f2fs_init_compress_mempool(void) { return 0; } 4237 static inline void f2fs_destroy_compress_mempool(void) { } 4238 static inline void f2fs_decompress_cluster(struct decompress_io_ctx *dic) { } 4239 static inline void f2fs_end_read_compressed_page(struct page *page, 4240 bool failed, block_t blkaddr) 4241 { 4242 WARN_ON_ONCE(1); 4243 } 4244 static inline void f2fs_put_page_dic(struct page *page) 4245 { 4246 WARN_ON_ONCE(1); 4247 } 4248 static inline unsigned int f2fs_cluster_blocks_are_contiguous(struct dnode_of_data *dn) { return 0; } 4249 static inline bool f2fs_sanity_check_cluster(struct dnode_of_data *dn) { return false; } 4250 static inline int f2fs_init_compress_inode(struct f2fs_sb_info *sbi) { return 0; } 4251 static inline void f2fs_destroy_compress_inode(struct f2fs_sb_info *sbi) { } 4252 static inline int f2fs_init_page_array_cache(struct f2fs_sb_info *sbi) { return 0; } 4253 static inline void f2fs_destroy_page_array_cache(struct f2fs_sb_info *sbi) { } 4254 static inline int __init f2fs_init_compress_cache(void) { return 0; } 4255 static inline void f2fs_destroy_compress_cache(void) { } 4256 static inline void f2fs_invalidate_compress_page(struct f2fs_sb_info *sbi, 4257 block_t blkaddr) { } 4258 static inline void f2fs_cache_compressed_page(struct f2fs_sb_info *sbi, 4259 struct page *page, nid_t ino, block_t blkaddr) { } 4260 static inline bool f2fs_load_compressed_page(struct f2fs_sb_info *sbi, 4261 struct page *page, block_t blkaddr) { return false; } 4262 static inline void f2fs_invalidate_compress_pages(struct f2fs_sb_info *sbi, 4263 nid_t ino) { } 4264 #define inc_compr_inode_stat(inode) do { } while (0) 4265 static inline void f2fs_update_extent_tree_range_compressed(struct inode *inode, 4266 pgoff_t fofs, block_t blkaddr, unsigned int llen, 4267 unsigned int c_len) { } 4268 #endif 4269 4270 static inline void set_compress_context(struct inode *inode) 4271 { 4272 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 4273 4274 F2FS_I(inode)->i_compress_algorithm = 4275 F2FS_OPTION(sbi).compress_algorithm; 4276 F2FS_I(inode)->i_log_cluster_size = 4277 F2FS_OPTION(sbi).compress_log_size; 4278 F2FS_I(inode)->i_compress_flag = 4279 F2FS_OPTION(sbi).compress_chksum ? 4280 1 << COMPRESS_CHKSUM : 0; 4281 F2FS_I(inode)->i_cluster_size = 4282 1 << F2FS_I(inode)->i_log_cluster_size; 4283 if ((F2FS_I(inode)->i_compress_algorithm == COMPRESS_LZ4 || 4284 F2FS_I(inode)->i_compress_algorithm == COMPRESS_ZSTD) && 4285 F2FS_OPTION(sbi).compress_level) 4286 F2FS_I(inode)->i_compress_flag |= 4287 F2FS_OPTION(sbi).compress_level << 4288 COMPRESS_LEVEL_OFFSET; 4289 F2FS_I(inode)->i_flags |= F2FS_COMPR_FL; 4290 set_inode_flag(inode, FI_COMPRESSED_FILE); 4291 stat_inc_compr_inode(inode); 4292 inc_compr_inode_stat(inode); 4293 f2fs_mark_inode_dirty_sync(inode, true); 4294 } 4295 4296 static inline bool f2fs_disable_compressed_file(struct inode *inode) 4297 { 4298 struct f2fs_inode_info *fi = F2FS_I(inode); 4299 4300 if (!f2fs_compressed_file(inode)) 4301 return true; 4302 if (S_ISREG(inode->i_mode) && F2FS_HAS_BLOCKS(inode)) 4303 return false; 4304 4305 fi->i_flags &= ~F2FS_COMPR_FL; 4306 stat_dec_compr_inode(inode); 4307 clear_inode_flag(inode, FI_COMPRESSED_FILE); 4308 f2fs_mark_inode_dirty_sync(inode, true); 4309 return true; 4310 } 4311 4312 #define F2FS_FEATURE_FUNCS(name, flagname) \ 4313 static inline int f2fs_sb_has_##name(struct f2fs_sb_info *sbi) \ 4314 { \ 4315 return F2FS_HAS_FEATURE(sbi, F2FS_FEATURE_##flagname); \ 4316 } 4317 4318 F2FS_FEATURE_FUNCS(encrypt, ENCRYPT); 4319 F2FS_FEATURE_FUNCS(blkzoned, BLKZONED); 4320 F2FS_FEATURE_FUNCS(extra_attr, EXTRA_ATTR); 4321 F2FS_FEATURE_FUNCS(project_quota, PRJQUOTA); 4322 F2FS_FEATURE_FUNCS(inode_chksum, INODE_CHKSUM); 4323 F2FS_FEATURE_FUNCS(flexible_inline_xattr, FLEXIBLE_INLINE_XATTR); 4324 F2FS_FEATURE_FUNCS(quota_ino, QUOTA_INO); 4325 F2FS_FEATURE_FUNCS(inode_crtime, INODE_CRTIME); 4326 F2FS_FEATURE_FUNCS(lost_found, LOST_FOUND); 4327 F2FS_FEATURE_FUNCS(verity, VERITY); 4328 F2FS_FEATURE_FUNCS(sb_chksum, SB_CHKSUM); 4329 F2FS_FEATURE_FUNCS(casefold, CASEFOLD); 4330 F2FS_FEATURE_FUNCS(compression, COMPRESSION); 4331 F2FS_FEATURE_FUNCS(readonly, RO); 4332 4333 static inline bool f2fs_may_extent_tree(struct inode *inode) 4334 { 4335 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 4336 4337 if (!test_opt(sbi, EXTENT_CACHE) || 4338 is_inode_flag_set(inode, FI_NO_EXTENT) || 4339 (is_inode_flag_set(inode, FI_COMPRESSED_FILE) && 4340 !f2fs_sb_has_readonly(sbi))) 4341 return false; 4342 4343 /* 4344 * for recovered files during mount do not create extents 4345 * if shrinker is not registered. 4346 */ 4347 if (list_empty(&sbi->s_list)) 4348 return false; 4349 4350 return S_ISREG(inode->i_mode); 4351 } 4352 4353 #ifdef CONFIG_BLK_DEV_ZONED 4354 static inline bool f2fs_blkz_is_seq(struct f2fs_sb_info *sbi, int devi, 4355 block_t blkaddr) 4356 { 4357 unsigned int zno = blkaddr >> sbi->log_blocks_per_blkz; 4358 4359 return test_bit(zno, FDEV(devi).blkz_seq); 4360 } 4361 #endif 4362 4363 static inline bool f2fs_hw_should_discard(struct f2fs_sb_info *sbi) 4364 { 4365 return f2fs_sb_has_blkzoned(sbi); 4366 } 4367 4368 static inline bool f2fs_bdev_support_discard(struct block_device *bdev) 4369 { 4370 return blk_queue_discard(bdev_get_queue(bdev)) || 4371 bdev_is_zoned(bdev); 4372 } 4373 4374 static inline bool f2fs_hw_support_discard(struct f2fs_sb_info *sbi) 4375 { 4376 int i; 4377 4378 if (!f2fs_is_multi_device(sbi)) 4379 return f2fs_bdev_support_discard(sbi->sb->s_bdev); 4380 4381 for (i = 0; i < sbi->s_ndevs; i++) 4382 if (f2fs_bdev_support_discard(FDEV(i).bdev)) 4383 return true; 4384 return false; 4385 } 4386 4387 static inline bool f2fs_realtime_discard_enable(struct f2fs_sb_info *sbi) 4388 { 4389 return (test_opt(sbi, DISCARD) && f2fs_hw_support_discard(sbi)) || 4390 f2fs_hw_should_discard(sbi); 4391 } 4392 4393 static inline bool f2fs_hw_is_readonly(struct f2fs_sb_info *sbi) 4394 { 4395 int i; 4396 4397 if (!f2fs_is_multi_device(sbi)) 4398 return bdev_read_only(sbi->sb->s_bdev); 4399 4400 for (i = 0; i < sbi->s_ndevs; i++) 4401 if (bdev_read_only(FDEV(i).bdev)) 4402 return true; 4403 return false; 4404 } 4405 4406 static inline bool f2fs_lfs_mode(struct f2fs_sb_info *sbi) 4407 { 4408 return F2FS_OPTION(sbi).fs_mode == FS_MODE_LFS; 4409 } 4410 4411 static inline bool f2fs_may_compress(struct inode *inode) 4412 { 4413 if (IS_SWAPFILE(inode) || f2fs_is_pinned_file(inode) || 4414 f2fs_is_atomic_file(inode) || 4415 f2fs_is_volatile_file(inode)) 4416 return false; 4417 return S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode); 4418 } 4419 4420 static inline void f2fs_i_compr_blocks_update(struct inode *inode, 4421 u64 blocks, bool add) 4422 { 4423 int diff = F2FS_I(inode)->i_cluster_size - blocks; 4424 struct f2fs_inode_info *fi = F2FS_I(inode); 4425 4426 /* don't update i_compr_blocks if saved blocks were released */ 4427 if (!add && !atomic_read(&fi->i_compr_blocks)) 4428 return; 4429 4430 if (add) { 4431 atomic_add(diff, &fi->i_compr_blocks); 4432 stat_add_compr_blocks(inode, diff); 4433 } else { 4434 atomic_sub(diff, &fi->i_compr_blocks); 4435 stat_sub_compr_blocks(inode, diff); 4436 } 4437 f2fs_mark_inode_dirty_sync(inode, true); 4438 } 4439 4440 static inline int block_unaligned_IO(struct inode *inode, 4441 struct kiocb *iocb, struct iov_iter *iter) 4442 { 4443 unsigned int i_blkbits = READ_ONCE(inode->i_blkbits); 4444 unsigned int blocksize_mask = (1 << i_blkbits) - 1; 4445 loff_t offset = iocb->ki_pos; 4446 unsigned long align = offset | iov_iter_alignment(iter); 4447 4448 return align & blocksize_mask; 4449 } 4450 4451 static inline bool f2fs_allow_multi_device_dio(struct f2fs_sb_info *sbi, 4452 int flag) 4453 { 4454 if (!f2fs_is_multi_device(sbi)) 4455 return false; 4456 if (flag != F2FS_GET_BLOCK_DIO) 4457 return false; 4458 return sbi->aligned_blksize; 4459 } 4460 4461 static inline bool f2fs_force_buffered_io(struct inode *inode, 4462 struct kiocb *iocb, struct iov_iter *iter) 4463 { 4464 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 4465 int rw = iov_iter_rw(iter); 4466 4467 if (f2fs_post_read_required(inode)) 4468 return true; 4469 4470 /* disallow direct IO if any of devices has unaligned blksize */ 4471 if (f2fs_is_multi_device(sbi) && !sbi->aligned_blksize) 4472 return true; 4473 /* 4474 * for blkzoned device, fallback direct IO to buffered IO, so 4475 * all IOs can be serialized by log-structured write. 4476 */ 4477 if (f2fs_sb_has_blkzoned(sbi)) 4478 return true; 4479 if (f2fs_lfs_mode(sbi) && (rw == WRITE)) { 4480 if (block_unaligned_IO(inode, iocb, iter)) 4481 return true; 4482 if (F2FS_IO_ALIGNED(sbi)) 4483 return true; 4484 } 4485 if (is_sbi_flag_set(F2FS_I_SB(inode), SBI_CP_DISABLED)) 4486 return true; 4487 4488 return false; 4489 } 4490 4491 static inline bool f2fs_need_verity(const struct inode *inode, pgoff_t idx) 4492 { 4493 return fsverity_active(inode) && 4494 idx < DIV_ROUND_UP(inode->i_size, PAGE_SIZE); 4495 } 4496 4497 #ifdef CONFIG_F2FS_FAULT_INJECTION 4498 extern void f2fs_build_fault_attr(struct f2fs_sb_info *sbi, unsigned int rate, 4499 unsigned int type); 4500 #else 4501 #define f2fs_build_fault_attr(sbi, rate, type) do { } while (0) 4502 #endif 4503 4504 static inline bool is_journalled_quota(struct f2fs_sb_info *sbi) 4505 { 4506 #ifdef CONFIG_QUOTA 4507 if (f2fs_sb_has_quota_ino(sbi)) 4508 return true; 4509 if (F2FS_OPTION(sbi).s_qf_names[USRQUOTA] || 4510 F2FS_OPTION(sbi).s_qf_names[GRPQUOTA] || 4511 F2FS_OPTION(sbi).s_qf_names[PRJQUOTA]) 4512 return true; 4513 #endif 4514 return false; 4515 } 4516 4517 static inline bool f2fs_block_unit_discard(struct f2fs_sb_info *sbi) 4518 { 4519 return F2FS_OPTION(sbi).discard_unit == DISCARD_UNIT_BLOCK; 4520 } 4521 4522 #define EFSBADCRC EBADMSG /* Bad CRC detected */ 4523 #define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */ 4524 4525 #endif /* _LINUX_F2FS_H */ 4526