1 /* SPDX-License-Identifier: GPL-2.0 */ 2 /* 3 * fs/f2fs/f2fs.h 4 * 5 * Copyright (c) 2012 Samsung Electronics Co., Ltd. 6 * http://www.samsung.com/ 7 */ 8 #ifndef _LINUX_F2FS_H 9 #define _LINUX_F2FS_H 10 11 #include <linux/uio.h> 12 #include <linux/types.h> 13 #include <linux/page-flags.h> 14 #include <linux/buffer_head.h> 15 #include <linux/slab.h> 16 #include <linux/crc32.h> 17 #include <linux/magic.h> 18 #include <linux/kobject.h> 19 #include <linux/sched.h> 20 #include <linux/cred.h> 21 #include <linux/vmalloc.h> 22 #include <linux/bio.h> 23 #include <linux/blkdev.h> 24 #include <linux/quotaops.h> 25 #include <linux/part_stat.h> 26 #include <crypto/hash.h> 27 28 #include <linux/fscrypt.h> 29 #include <linux/fsverity.h> 30 31 struct pagevec; 32 33 #ifdef CONFIG_F2FS_CHECK_FS 34 #define f2fs_bug_on(sbi, condition) BUG_ON(condition) 35 #else 36 #define f2fs_bug_on(sbi, condition) \ 37 do { \ 38 if (WARN_ON(condition)) \ 39 set_sbi_flag(sbi, SBI_NEED_FSCK); \ 40 } while (0) 41 #endif 42 43 enum { 44 FAULT_KMALLOC, 45 FAULT_KVMALLOC, 46 FAULT_PAGE_ALLOC, 47 FAULT_PAGE_GET, 48 FAULT_ALLOC_BIO, /* it's obsolete due to bio_alloc() will never fail */ 49 FAULT_ALLOC_NID, 50 FAULT_ORPHAN, 51 FAULT_BLOCK, 52 FAULT_DIR_DEPTH, 53 FAULT_EVICT_INODE, 54 FAULT_TRUNCATE, 55 FAULT_READ_IO, 56 FAULT_CHECKPOINT, 57 FAULT_DISCARD, 58 FAULT_WRITE_IO, 59 FAULT_SLAB_ALLOC, 60 FAULT_DQUOT_INIT, 61 FAULT_LOCK_OP, 62 FAULT_MAX, 63 }; 64 65 #ifdef CONFIG_F2FS_FAULT_INJECTION 66 #define F2FS_ALL_FAULT_TYPE ((1 << FAULT_MAX) - 1) 67 68 struct f2fs_fault_info { 69 atomic_t inject_ops; 70 unsigned int inject_rate; 71 unsigned int inject_type; 72 }; 73 74 extern const char *f2fs_fault_name[FAULT_MAX]; 75 #define IS_FAULT_SET(fi, type) ((fi)->inject_type & (1 << (type))) 76 #endif 77 78 /* 79 * For mount options 80 */ 81 #define F2FS_MOUNT_DISABLE_ROLL_FORWARD 0x00000002 82 #define F2FS_MOUNT_DISCARD 0x00000004 83 #define F2FS_MOUNT_NOHEAP 0x00000008 84 #define F2FS_MOUNT_XATTR_USER 0x00000010 85 #define F2FS_MOUNT_POSIX_ACL 0x00000020 86 #define F2FS_MOUNT_DISABLE_EXT_IDENTIFY 0x00000040 87 #define F2FS_MOUNT_INLINE_XATTR 0x00000080 88 #define F2FS_MOUNT_INLINE_DATA 0x00000100 89 #define F2FS_MOUNT_INLINE_DENTRY 0x00000200 90 #define F2FS_MOUNT_FLUSH_MERGE 0x00000400 91 #define F2FS_MOUNT_NOBARRIER 0x00000800 92 #define F2FS_MOUNT_FASTBOOT 0x00001000 93 #define F2FS_MOUNT_EXTENT_CACHE 0x00002000 94 #define F2FS_MOUNT_DATA_FLUSH 0x00008000 95 #define F2FS_MOUNT_FAULT_INJECTION 0x00010000 96 #define F2FS_MOUNT_USRQUOTA 0x00080000 97 #define F2FS_MOUNT_GRPQUOTA 0x00100000 98 #define F2FS_MOUNT_PRJQUOTA 0x00200000 99 #define F2FS_MOUNT_QUOTA 0x00400000 100 #define F2FS_MOUNT_INLINE_XATTR_SIZE 0x00800000 101 #define F2FS_MOUNT_RESERVE_ROOT 0x01000000 102 #define F2FS_MOUNT_DISABLE_CHECKPOINT 0x02000000 103 #define F2FS_MOUNT_NORECOVERY 0x04000000 104 #define F2FS_MOUNT_ATGC 0x08000000 105 #define F2FS_MOUNT_MERGE_CHECKPOINT 0x10000000 106 #define F2FS_MOUNT_GC_MERGE 0x20000000 107 #define F2FS_MOUNT_COMPRESS_CACHE 0x40000000 108 109 #define F2FS_OPTION(sbi) ((sbi)->mount_opt) 110 #define clear_opt(sbi, option) (F2FS_OPTION(sbi).opt &= ~F2FS_MOUNT_##option) 111 #define set_opt(sbi, option) (F2FS_OPTION(sbi).opt |= F2FS_MOUNT_##option) 112 #define test_opt(sbi, option) (F2FS_OPTION(sbi).opt & F2FS_MOUNT_##option) 113 114 #define ver_after(a, b) (typecheck(unsigned long long, a) && \ 115 typecheck(unsigned long long, b) && \ 116 ((long long)((a) - (b)) > 0)) 117 118 typedef u32 block_t; /* 119 * should not change u32, since it is the on-disk block 120 * address format, __le32. 121 */ 122 typedef u32 nid_t; 123 124 #define COMPRESS_EXT_NUM 16 125 126 /* 127 * An implementation of an rwsem that is explicitly unfair to readers. This 128 * prevents priority inversion when a low-priority reader acquires the read lock 129 * while sleeping on the write lock but the write lock is needed by 130 * higher-priority clients. 131 */ 132 133 struct f2fs_rwsem { 134 struct rw_semaphore internal_rwsem; 135 #ifdef CONFIG_F2FS_UNFAIR_RWSEM 136 wait_queue_head_t read_waiters; 137 #endif 138 }; 139 140 struct f2fs_mount_info { 141 unsigned int opt; 142 int write_io_size_bits; /* Write IO size bits */ 143 block_t root_reserved_blocks; /* root reserved blocks */ 144 kuid_t s_resuid; /* reserved blocks for uid */ 145 kgid_t s_resgid; /* reserved blocks for gid */ 146 int active_logs; /* # of active logs */ 147 int inline_xattr_size; /* inline xattr size */ 148 #ifdef CONFIG_F2FS_FAULT_INJECTION 149 struct f2fs_fault_info fault_info; /* For fault injection */ 150 #endif 151 #ifdef CONFIG_QUOTA 152 /* Names of quota files with journalled quota */ 153 char *s_qf_names[MAXQUOTAS]; 154 int s_jquota_fmt; /* Format of quota to use */ 155 #endif 156 /* For which write hints are passed down to block layer */ 157 int whint_mode; 158 int alloc_mode; /* segment allocation policy */ 159 int fsync_mode; /* fsync policy */ 160 int fs_mode; /* fs mode: LFS or ADAPTIVE */ 161 int bggc_mode; /* bggc mode: off, on or sync */ 162 int discard_unit; /* 163 * discard command's offset/size should 164 * be aligned to this unit: block, 165 * segment or section 166 */ 167 struct fscrypt_dummy_policy dummy_enc_policy; /* test dummy encryption */ 168 block_t unusable_cap_perc; /* percentage for cap */ 169 block_t unusable_cap; /* Amount of space allowed to be 170 * unusable when disabling checkpoint 171 */ 172 173 /* For compression */ 174 unsigned char compress_algorithm; /* algorithm type */ 175 unsigned char compress_log_size; /* cluster log size */ 176 unsigned char compress_level; /* compress level */ 177 bool compress_chksum; /* compressed data chksum */ 178 unsigned char compress_ext_cnt; /* extension count */ 179 unsigned char nocompress_ext_cnt; /* nocompress extension count */ 180 int compress_mode; /* compression mode */ 181 unsigned char extensions[COMPRESS_EXT_NUM][F2FS_EXTENSION_LEN]; /* extensions */ 182 unsigned char noextensions[COMPRESS_EXT_NUM][F2FS_EXTENSION_LEN]; /* extensions */ 183 }; 184 185 #define F2FS_FEATURE_ENCRYPT 0x0001 186 #define F2FS_FEATURE_BLKZONED 0x0002 187 #define F2FS_FEATURE_ATOMIC_WRITE 0x0004 188 #define F2FS_FEATURE_EXTRA_ATTR 0x0008 189 #define F2FS_FEATURE_PRJQUOTA 0x0010 190 #define F2FS_FEATURE_INODE_CHKSUM 0x0020 191 #define F2FS_FEATURE_FLEXIBLE_INLINE_XATTR 0x0040 192 #define F2FS_FEATURE_QUOTA_INO 0x0080 193 #define F2FS_FEATURE_INODE_CRTIME 0x0100 194 #define F2FS_FEATURE_LOST_FOUND 0x0200 195 #define F2FS_FEATURE_VERITY 0x0400 196 #define F2FS_FEATURE_SB_CHKSUM 0x0800 197 #define F2FS_FEATURE_CASEFOLD 0x1000 198 #define F2FS_FEATURE_COMPRESSION 0x2000 199 #define F2FS_FEATURE_RO 0x4000 200 201 #define __F2FS_HAS_FEATURE(raw_super, mask) \ 202 ((raw_super->feature & cpu_to_le32(mask)) != 0) 203 #define F2FS_HAS_FEATURE(sbi, mask) __F2FS_HAS_FEATURE(sbi->raw_super, mask) 204 #define F2FS_SET_FEATURE(sbi, mask) \ 205 (sbi->raw_super->feature |= cpu_to_le32(mask)) 206 #define F2FS_CLEAR_FEATURE(sbi, mask) \ 207 (sbi->raw_super->feature &= ~cpu_to_le32(mask)) 208 209 /* 210 * Default values for user and/or group using reserved blocks 211 */ 212 #define F2FS_DEF_RESUID 0 213 #define F2FS_DEF_RESGID 0 214 215 /* 216 * For checkpoint manager 217 */ 218 enum { 219 NAT_BITMAP, 220 SIT_BITMAP 221 }; 222 223 #define CP_UMOUNT 0x00000001 224 #define CP_FASTBOOT 0x00000002 225 #define CP_SYNC 0x00000004 226 #define CP_RECOVERY 0x00000008 227 #define CP_DISCARD 0x00000010 228 #define CP_TRIMMED 0x00000020 229 #define CP_PAUSE 0x00000040 230 #define CP_RESIZE 0x00000080 231 232 #define MAX_DISCARD_BLOCKS(sbi) BLKS_PER_SEC(sbi) 233 #define DEF_MAX_DISCARD_REQUEST 8 /* issue 8 discards per round */ 234 #define DEF_MIN_DISCARD_ISSUE_TIME 50 /* 50 ms, if exists */ 235 #define DEF_MID_DISCARD_ISSUE_TIME 500 /* 500 ms, if device busy */ 236 #define DEF_MAX_DISCARD_ISSUE_TIME 60000 /* 60 s, if no candidates */ 237 #define DEF_DISCARD_URGENT_UTIL 80 /* do more discard over 80% */ 238 #define DEF_CP_INTERVAL 60 /* 60 secs */ 239 #define DEF_IDLE_INTERVAL 5 /* 5 secs */ 240 #define DEF_DISABLE_INTERVAL 5 /* 5 secs */ 241 #define DEF_DISABLE_QUICK_INTERVAL 1 /* 1 secs */ 242 #define DEF_UMOUNT_DISCARD_TIMEOUT 5 /* 5 secs */ 243 244 struct cp_control { 245 int reason; 246 __u64 trim_start; 247 __u64 trim_end; 248 __u64 trim_minlen; 249 }; 250 251 /* 252 * indicate meta/data type 253 */ 254 enum { 255 META_CP, 256 META_NAT, 257 META_SIT, 258 META_SSA, 259 META_MAX, 260 META_POR, 261 DATA_GENERIC, /* check range only */ 262 DATA_GENERIC_ENHANCE, /* strong check on range and segment bitmap */ 263 DATA_GENERIC_ENHANCE_READ, /* 264 * strong check on range and segment 265 * bitmap but no warning due to race 266 * condition of read on truncated area 267 * by extent_cache 268 */ 269 META_GENERIC, 270 }; 271 272 /* for the list of ino */ 273 enum { 274 ORPHAN_INO, /* for orphan ino list */ 275 APPEND_INO, /* for append ino list */ 276 UPDATE_INO, /* for update ino list */ 277 TRANS_DIR_INO, /* for trasactions dir ino list */ 278 FLUSH_INO, /* for multiple device flushing */ 279 MAX_INO_ENTRY, /* max. list */ 280 }; 281 282 struct ino_entry { 283 struct list_head list; /* list head */ 284 nid_t ino; /* inode number */ 285 unsigned int dirty_device; /* dirty device bitmap */ 286 }; 287 288 /* for the list of inodes to be GCed */ 289 struct inode_entry { 290 struct list_head list; /* list head */ 291 struct inode *inode; /* vfs inode pointer */ 292 }; 293 294 struct fsync_node_entry { 295 struct list_head list; /* list head */ 296 struct page *page; /* warm node page pointer */ 297 unsigned int seq_id; /* sequence id */ 298 }; 299 300 struct ckpt_req { 301 struct completion wait; /* completion for checkpoint done */ 302 struct llist_node llnode; /* llist_node to be linked in wait queue */ 303 int ret; /* return code of checkpoint */ 304 ktime_t queue_time; /* request queued time */ 305 }; 306 307 struct ckpt_req_control { 308 struct task_struct *f2fs_issue_ckpt; /* checkpoint task */ 309 int ckpt_thread_ioprio; /* checkpoint merge thread ioprio */ 310 wait_queue_head_t ckpt_wait_queue; /* waiting queue for wake-up */ 311 atomic_t issued_ckpt; /* # of actually issued ckpts */ 312 atomic_t total_ckpt; /* # of total ckpts */ 313 atomic_t queued_ckpt; /* # of queued ckpts */ 314 struct llist_head issue_list; /* list for command issue */ 315 spinlock_t stat_lock; /* lock for below checkpoint time stats */ 316 unsigned int cur_time; /* cur wait time in msec for currently issued checkpoint */ 317 unsigned int peak_time; /* peak wait time in msec until now */ 318 }; 319 320 /* for the bitmap indicate blocks to be discarded */ 321 struct discard_entry { 322 struct list_head list; /* list head */ 323 block_t start_blkaddr; /* start blockaddr of current segment */ 324 unsigned char discard_map[SIT_VBLOCK_MAP_SIZE]; /* segment discard bitmap */ 325 }; 326 327 /* default discard granularity of inner discard thread, unit: block count */ 328 #define DEFAULT_DISCARD_GRANULARITY 16 329 330 /* max discard pend list number */ 331 #define MAX_PLIST_NUM 512 332 #define plist_idx(blk_num) ((blk_num) >= MAX_PLIST_NUM ? \ 333 (MAX_PLIST_NUM - 1) : ((blk_num) - 1)) 334 335 enum { 336 D_PREP, /* initial */ 337 D_PARTIAL, /* partially submitted */ 338 D_SUBMIT, /* all submitted */ 339 D_DONE, /* finished */ 340 }; 341 342 struct discard_info { 343 block_t lstart; /* logical start address */ 344 block_t len; /* length */ 345 block_t start; /* actual start address in dev */ 346 }; 347 348 struct discard_cmd { 349 struct rb_node rb_node; /* rb node located in rb-tree */ 350 union { 351 struct { 352 block_t lstart; /* logical start address */ 353 block_t len; /* length */ 354 block_t start; /* actual start address in dev */ 355 }; 356 struct discard_info di; /* discard info */ 357 358 }; 359 struct list_head list; /* command list */ 360 struct completion wait; /* compleation */ 361 struct block_device *bdev; /* bdev */ 362 unsigned short ref; /* reference count */ 363 unsigned char state; /* state */ 364 unsigned char queued; /* queued discard */ 365 int error; /* bio error */ 366 spinlock_t lock; /* for state/bio_ref updating */ 367 unsigned short bio_ref; /* bio reference count */ 368 }; 369 370 enum { 371 DPOLICY_BG, 372 DPOLICY_FORCE, 373 DPOLICY_FSTRIM, 374 DPOLICY_UMOUNT, 375 MAX_DPOLICY, 376 }; 377 378 struct discard_policy { 379 int type; /* type of discard */ 380 unsigned int min_interval; /* used for candidates exist */ 381 unsigned int mid_interval; /* used for device busy */ 382 unsigned int max_interval; /* used for candidates not exist */ 383 unsigned int max_requests; /* # of discards issued per round */ 384 unsigned int io_aware_gran; /* minimum granularity discard not be aware of I/O */ 385 bool io_aware; /* issue discard in idle time */ 386 bool sync; /* submit discard with REQ_SYNC flag */ 387 bool ordered; /* issue discard by lba order */ 388 bool timeout; /* discard timeout for put_super */ 389 unsigned int granularity; /* discard granularity */ 390 }; 391 392 struct discard_cmd_control { 393 struct task_struct *f2fs_issue_discard; /* discard thread */ 394 struct list_head entry_list; /* 4KB discard entry list */ 395 struct list_head pend_list[MAX_PLIST_NUM];/* store pending entries */ 396 struct list_head wait_list; /* store on-flushing entries */ 397 struct list_head fstrim_list; /* in-flight discard from fstrim */ 398 wait_queue_head_t discard_wait_queue; /* waiting queue for wake-up */ 399 unsigned int discard_wake; /* to wake up discard thread */ 400 struct mutex cmd_lock; 401 unsigned int nr_discards; /* # of discards in the list */ 402 unsigned int max_discards; /* max. discards to be issued */ 403 unsigned int max_discard_request; /* max. discard request per round */ 404 unsigned int min_discard_issue_time; /* min. interval between discard issue */ 405 unsigned int mid_discard_issue_time; /* mid. interval between discard issue */ 406 unsigned int max_discard_issue_time; /* max. interval between discard issue */ 407 unsigned int discard_granularity; /* discard granularity */ 408 unsigned int undiscard_blks; /* # of undiscard blocks */ 409 unsigned int next_pos; /* next discard position */ 410 atomic_t issued_discard; /* # of issued discard */ 411 atomic_t queued_discard; /* # of queued discard */ 412 atomic_t discard_cmd_cnt; /* # of cached cmd count */ 413 struct rb_root_cached root; /* root of discard rb-tree */ 414 bool rbtree_check; /* config for consistence check */ 415 }; 416 417 /* for the list of fsync inodes, used only during recovery */ 418 struct fsync_inode_entry { 419 struct list_head list; /* list head */ 420 struct inode *inode; /* vfs inode pointer */ 421 block_t blkaddr; /* block address locating the last fsync */ 422 block_t last_dentry; /* block address locating the last dentry */ 423 }; 424 425 #define nats_in_cursum(jnl) (le16_to_cpu((jnl)->n_nats)) 426 #define sits_in_cursum(jnl) (le16_to_cpu((jnl)->n_sits)) 427 428 #define nat_in_journal(jnl, i) ((jnl)->nat_j.entries[i].ne) 429 #define nid_in_journal(jnl, i) ((jnl)->nat_j.entries[i].nid) 430 #define sit_in_journal(jnl, i) ((jnl)->sit_j.entries[i].se) 431 #define segno_in_journal(jnl, i) ((jnl)->sit_j.entries[i].segno) 432 433 #define MAX_NAT_JENTRIES(jnl) (NAT_JOURNAL_ENTRIES - nats_in_cursum(jnl)) 434 #define MAX_SIT_JENTRIES(jnl) (SIT_JOURNAL_ENTRIES - sits_in_cursum(jnl)) 435 436 static inline int update_nats_in_cursum(struct f2fs_journal *journal, int i) 437 { 438 int before = nats_in_cursum(journal); 439 440 journal->n_nats = cpu_to_le16(before + i); 441 return before; 442 } 443 444 static inline int update_sits_in_cursum(struct f2fs_journal *journal, int i) 445 { 446 int before = sits_in_cursum(journal); 447 448 journal->n_sits = cpu_to_le16(before + i); 449 return before; 450 } 451 452 static inline bool __has_cursum_space(struct f2fs_journal *journal, 453 int size, int type) 454 { 455 if (type == NAT_JOURNAL) 456 return size <= MAX_NAT_JENTRIES(journal); 457 return size <= MAX_SIT_JENTRIES(journal); 458 } 459 460 /* for inline stuff */ 461 #define DEF_INLINE_RESERVED_SIZE 1 462 static inline int get_extra_isize(struct inode *inode); 463 static inline int get_inline_xattr_addrs(struct inode *inode); 464 #define MAX_INLINE_DATA(inode) (sizeof(__le32) * \ 465 (CUR_ADDRS_PER_INODE(inode) - \ 466 get_inline_xattr_addrs(inode) - \ 467 DEF_INLINE_RESERVED_SIZE)) 468 469 /* for inline dir */ 470 #define NR_INLINE_DENTRY(inode) (MAX_INLINE_DATA(inode) * BITS_PER_BYTE / \ 471 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \ 472 BITS_PER_BYTE + 1)) 473 #define INLINE_DENTRY_BITMAP_SIZE(inode) \ 474 DIV_ROUND_UP(NR_INLINE_DENTRY(inode), BITS_PER_BYTE) 475 #define INLINE_RESERVED_SIZE(inode) (MAX_INLINE_DATA(inode) - \ 476 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \ 477 NR_INLINE_DENTRY(inode) + \ 478 INLINE_DENTRY_BITMAP_SIZE(inode))) 479 480 /* 481 * For INODE and NODE manager 482 */ 483 /* for directory operations */ 484 485 struct f2fs_filename { 486 /* 487 * The filename the user specified. This is NULL for some 488 * filesystem-internal operations, e.g. converting an inline directory 489 * to a non-inline one, or roll-forward recovering an encrypted dentry. 490 */ 491 const struct qstr *usr_fname; 492 493 /* 494 * The on-disk filename. For encrypted directories, this is encrypted. 495 * This may be NULL for lookups in an encrypted dir without the key. 496 */ 497 struct fscrypt_str disk_name; 498 499 /* The dirhash of this filename */ 500 f2fs_hash_t hash; 501 502 #ifdef CONFIG_FS_ENCRYPTION 503 /* 504 * For lookups in encrypted directories: either the buffer backing 505 * disk_name, or a buffer that holds the decoded no-key name. 506 */ 507 struct fscrypt_str crypto_buf; 508 #endif 509 #ifdef CONFIG_UNICODE 510 /* 511 * For casefolded directories: the casefolded name, but it's left NULL 512 * if the original name is not valid Unicode, if the directory is both 513 * casefolded and encrypted and its encryption key is unavailable, or if 514 * the filesystem is doing an internal operation where usr_fname is also 515 * NULL. In all these cases we fall back to treating the name as an 516 * opaque byte sequence. 517 */ 518 struct fscrypt_str cf_name; 519 #endif 520 }; 521 522 struct f2fs_dentry_ptr { 523 struct inode *inode; 524 void *bitmap; 525 struct f2fs_dir_entry *dentry; 526 __u8 (*filename)[F2FS_SLOT_LEN]; 527 int max; 528 int nr_bitmap; 529 }; 530 531 static inline void make_dentry_ptr_block(struct inode *inode, 532 struct f2fs_dentry_ptr *d, struct f2fs_dentry_block *t) 533 { 534 d->inode = inode; 535 d->max = NR_DENTRY_IN_BLOCK; 536 d->nr_bitmap = SIZE_OF_DENTRY_BITMAP; 537 d->bitmap = t->dentry_bitmap; 538 d->dentry = t->dentry; 539 d->filename = t->filename; 540 } 541 542 static inline void make_dentry_ptr_inline(struct inode *inode, 543 struct f2fs_dentry_ptr *d, void *t) 544 { 545 int entry_cnt = NR_INLINE_DENTRY(inode); 546 int bitmap_size = INLINE_DENTRY_BITMAP_SIZE(inode); 547 int reserved_size = INLINE_RESERVED_SIZE(inode); 548 549 d->inode = inode; 550 d->max = entry_cnt; 551 d->nr_bitmap = bitmap_size; 552 d->bitmap = t; 553 d->dentry = t + bitmap_size + reserved_size; 554 d->filename = t + bitmap_size + reserved_size + 555 SIZE_OF_DIR_ENTRY * entry_cnt; 556 } 557 558 /* 559 * XATTR_NODE_OFFSET stores xattrs to one node block per file keeping -1 560 * as its node offset to distinguish from index node blocks. 561 * But some bits are used to mark the node block. 562 */ 563 #define XATTR_NODE_OFFSET ((((unsigned int)-1) << OFFSET_BIT_SHIFT) \ 564 >> OFFSET_BIT_SHIFT) 565 enum { 566 ALLOC_NODE, /* allocate a new node page if needed */ 567 LOOKUP_NODE, /* look up a node without readahead */ 568 LOOKUP_NODE_RA, /* 569 * look up a node with readahead called 570 * by get_data_block. 571 */ 572 }; 573 574 #define DEFAULT_RETRY_IO_COUNT 8 /* maximum retry read IO or flush count */ 575 576 /* congestion wait timeout value, default: 20ms */ 577 #define DEFAULT_IO_TIMEOUT (msecs_to_jiffies(20)) 578 579 /* maximum retry quota flush count */ 580 #define DEFAULT_RETRY_QUOTA_FLUSH_COUNT 8 581 582 /* maximum retry of EIO'ed meta page */ 583 #define MAX_RETRY_META_PAGE_EIO 100 584 585 #define F2FS_LINK_MAX 0xffffffff /* maximum link count per file */ 586 587 #define MAX_DIR_RA_PAGES 4 /* maximum ra pages of dir */ 588 589 /* dirty segments threshold for triggering CP */ 590 #define DEFAULT_DIRTY_THRESHOLD 4 591 592 /* for in-memory extent cache entry */ 593 #define F2FS_MIN_EXTENT_LEN 64 /* minimum extent length */ 594 595 /* number of extent info in extent cache we try to shrink */ 596 #define EXTENT_CACHE_SHRINK_NUMBER 128 597 598 #define RECOVERY_MAX_RA_BLOCKS BIO_MAX_VECS 599 #define RECOVERY_MIN_RA_BLOCKS 1 600 601 struct rb_entry { 602 struct rb_node rb_node; /* rb node located in rb-tree */ 603 union { 604 struct { 605 unsigned int ofs; /* start offset of the entry */ 606 unsigned int len; /* length of the entry */ 607 }; 608 unsigned long long key; /* 64-bits key */ 609 } __packed; 610 }; 611 612 struct extent_info { 613 unsigned int fofs; /* start offset in a file */ 614 unsigned int len; /* length of the extent */ 615 u32 blk; /* start block address of the extent */ 616 #ifdef CONFIG_F2FS_FS_COMPRESSION 617 unsigned int c_len; /* physical extent length of compressed blocks */ 618 #endif 619 }; 620 621 struct extent_node { 622 struct rb_node rb_node; /* rb node located in rb-tree */ 623 struct extent_info ei; /* extent info */ 624 struct list_head list; /* node in global extent list of sbi */ 625 struct extent_tree *et; /* extent tree pointer */ 626 }; 627 628 struct extent_tree { 629 nid_t ino; /* inode number */ 630 struct rb_root_cached root; /* root of extent info rb-tree */ 631 struct extent_node *cached_en; /* recently accessed extent node */ 632 struct extent_info largest; /* largested extent info */ 633 struct list_head list; /* to be used by sbi->zombie_list */ 634 rwlock_t lock; /* protect extent info rb-tree */ 635 atomic_t node_cnt; /* # of extent node in rb-tree*/ 636 bool largest_updated; /* largest extent updated */ 637 }; 638 639 /* 640 * This structure is taken from ext4_map_blocks. 641 * 642 * Note that, however, f2fs uses NEW and MAPPED flags for f2fs_map_blocks(). 643 */ 644 #define F2FS_MAP_NEW (1 << BH_New) 645 #define F2FS_MAP_MAPPED (1 << BH_Mapped) 646 #define F2FS_MAP_UNWRITTEN (1 << BH_Unwritten) 647 #define F2FS_MAP_FLAGS (F2FS_MAP_NEW | F2FS_MAP_MAPPED |\ 648 F2FS_MAP_UNWRITTEN) 649 650 struct f2fs_map_blocks { 651 struct block_device *m_bdev; /* for multi-device dio */ 652 block_t m_pblk; 653 block_t m_lblk; 654 unsigned int m_len; 655 unsigned int m_flags; 656 pgoff_t *m_next_pgofs; /* point next possible non-hole pgofs */ 657 pgoff_t *m_next_extent; /* point to next possible extent */ 658 int m_seg_type; 659 bool m_may_create; /* indicate it is from write path */ 660 bool m_multidev_dio; /* indicate it allows multi-device dio */ 661 }; 662 663 /* for flag in get_data_block */ 664 enum { 665 F2FS_GET_BLOCK_DEFAULT, 666 F2FS_GET_BLOCK_FIEMAP, 667 F2FS_GET_BLOCK_BMAP, 668 F2FS_GET_BLOCK_DIO, 669 F2FS_GET_BLOCK_PRE_DIO, 670 F2FS_GET_BLOCK_PRE_AIO, 671 F2FS_GET_BLOCK_PRECACHE, 672 }; 673 674 /* 675 * i_advise uses FADVISE_XXX_BIT. We can add additional hints later. 676 */ 677 #define FADVISE_COLD_BIT 0x01 678 #define FADVISE_LOST_PINO_BIT 0x02 679 #define FADVISE_ENCRYPT_BIT 0x04 680 #define FADVISE_ENC_NAME_BIT 0x08 681 #define FADVISE_KEEP_SIZE_BIT 0x10 682 #define FADVISE_HOT_BIT 0x20 683 #define FADVISE_VERITY_BIT 0x40 684 #define FADVISE_TRUNC_BIT 0x80 685 686 #define FADVISE_MODIFIABLE_BITS (FADVISE_COLD_BIT | FADVISE_HOT_BIT) 687 688 #define file_is_cold(inode) is_file(inode, FADVISE_COLD_BIT) 689 #define file_set_cold(inode) set_file(inode, FADVISE_COLD_BIT) 690 #define file_clear_cold(inode) clear_file(inode, FADVISE_COLD_BIT) 691 692 #define file_wrong_pino(inode) is_file(inode, FADVISE_LOST_PINO_BIT) 693 #define file_lost_pino(inode) set_file(inode, FADVISE_LOST_PINO_BIT) 694 #define file_got_pino(inode) clear_file(inode, FADVISE_LOST_PINO_BIT) 695 696 #define file_is_encrypt(inode) is_file(inode, FADVISE_ENCRYPT_BIT) 697 #define file_set_encrypt(inode) set_file(inode, FADVISE_ENCRYPT_BIT) 698 699 #define file_enc_name(inode) is_file(inode, FADVISE_ENC_NAME_BIT) 700 #define file_set_enc_name(inode) set_file(inode, FADVISE_ENC_NAME_BIT) 701 702 #define file_keep_isize(inode) is_file(inode, FADVISE_KEEP_SIZE_BIT) 703 #define file_set_keep_isize(inode) set_file(inode, FADVISE_KEEP_SIZE_BIT) 704 705 #define file_is_hot(inode) is_file(inode, FADVISE_HOT_BIT) 706 #define file_set_hot(inode) set_file(inode, FADVISE_HOT_BIT) 707 #define file_clear_hot(inode) clear_file(inode, FADVISE_HOT_BIT) 708 709 #define file_is_verity(inode) is_file(inode, FADVISE_VERITY_BIT) 710 #define file_set_verity(inode) set_file(inode, FADVISE_VERITY_BIT) 711 712 #define file_should_truncate(inode) is_file(inode, FADVISE_TRUNC_BIT) 713 #define file_need_truncate(inode) set_file(inode, FADVISE_TRUNC_BIT) 714 #define file_dont_truncate(inode) clear_file(inode, FADVISE_TRUNC_BIT) 715 716 #define DEF_DIR_LEVEL 0 717 718 enum { 719 GC_FAILURE_PIN, 720 GC_FAILURE_ATOMIC, 721 MAX_GC_FAILURE 722 }; 723 724 /* used for f2fs_inode_info->flags */ 725 enum { 726 FI_NEW_INODE, /* indicate newly allocated inode */ 727 FI_DIRTY_INODE, /* indicate inode is dirty or not */ 728 FI_AUTO_RECOVER, /* indicate inode is recoverable */ 729 FI_DIRTY_DIR, /* indicate directory has dirty pages */ 730 FI_INC_LINK, /* need to increment i_nlink */ 731 FI_ACL_MODE, /* indicate acl mode */ 732 FI_NO_ALLOC, /* should not allocate any blocks */ 733 FI_FREE_NID, /* free allocated nide */ 734 FI_NO_EXTENT, /* not to use the extent cache */ 735 FI_INLINE_XATTR, /* used for inline xattr */ 736 FI_INLINE_DATA, /* used for inline data*/ 737 FI_INLINE_DENTRY, /* used for inline dentry */ 738 FI_APPEND_WRITE, /* inode has appended data */ 739 FI_UPDATE_WRITE, /* inode has in-place-update data */ 740 FI_NEED_IPU, /* used for ipu per file */ 741 FI_ATOMIC_FILE, /* indicate atomic file */ 742 FI_ATOMIC_COMMIT, /* indicate the state of atomical committing */ 743 FI_VOLATILE_FILE, /* indicate volatile file */ 744 FI_FIRST_BLOCK_WRITTEN, /* indicate #0 data block was written */ 745 FI_DROP_CACHE, /* drop dirty page cache */ 746 FI_DATA_EXIST, /* indicate data exists */ 747 FI_INLINE_DOTS, /* indicate inline dot dentries */ 748 FI_SKIP_WRITES, /* should skip data page writeback */ 749 FI_OPU_WRITE, /* used for opu per file */ 750 FI_DIRTY_FILE, /* indicate regular/symlink has dirty pages */ 751 FI_PREALLOCATED_ALL, /* all blocks for write were preallocated */ 752 FI_HOT_DATA, /* indicate file is hot */ 753 FI_EXTRA_ATTR, /* indicate file has extra attribute */ 754 FI_PROJ_INHERIT, /* indicate file inherits projectid */ 755 FI_PIN_FILE, /* indicate file should not be gced */ 756 FI_ATOMIC_REVOKE_REQUEST, /* request to drop atomic data */ 757 FI_VERITY_IN_PROGRESS, /* building fs-verity Merkle tree */ 758 FI_COMPRESSED_FILE, /* indicate file's data can be compressed */ 759 FI_COMPRESS_CORRUPT, /* indicate compressed cluster is corrupted */ 760 FI_MMAP_FILE, /* indicate file was mmapped */ 761 FI_ENABLE_COMPRESS, /* enable compression in "user" compression mode */ 762 FI_COMPRESS_RELEASED, /* compressed blocks were released */ 763 FI_ALIGNED_WRITE, /* enable aligned write */ 764 FI_MAX, /* max flag, never be used */ 765 }; 766 767 struct f2fs_inode_info { 768 struct inode vfs_inode; /* serve a vfs inode */ 769 unsigned long i_flags; /* keep an inode flags for ioctl */ 770 unsigned char i_advise; /* use to give file attribute hints */ 771 unsigned char i_dir_level; /* use for dentry level for large dir */ 772 unsigned int i_current_depth; /* only for directory depth */ 773 /* for gc failure statistic */ 774 unsigned int i_gc_failures[MAX_GC_FAILURE]; 775 unsigned int i_pino; /* parent inode number */ 776 umode_t i_acl_mode; /* keep file acl mode temporarily */ 777 778 /* Use below internally in f2fs*/ 779 unsigned long flags[BITS_TO_LONGS(FI_MAX)]; /* use to pass per-file flags */ 780 struct f2fs_rwsem i_sem; /* protect fi info */ 781 atomic_t dirty_pages; /* # of dirty pages */ 782 f2fs_hash_t chash; /* hash value of given file name */ 783 unsigned int clevel; /* maximum level of given file name */ 784 struct task_struct *task; /* lookup and create consistency */ 785 struct task_struct *cp_task; /* separate cp/wb IO stats*/ 786 nid_t i_xattr_nid; /* node id that contains xattrs */ 787 loff_t last_disk_size; /* lastly written file size */ 788 spinlock_t i_size_lock; /* protect last_disk_size */ 789 790 #ifdef CONFIG_QUOTA 791 struct dquot *i_dquot[MAXQUOTAS]; 792 793 /* quota space reservation, managed internally by quota code */ 794 qsize_t i_reserved_quota; 795 #endif 796 struct list_head dirty_list; /* dirty list for dirs and files */ 797 struct list_head gdirty_list; /* linked in global dirty list */ 798 struct list_head inmem_ilist; /* list for inmem inodes */ 799 struct list_head inmem_pages; /* inmemory pages managed by f2fs */ 800 struct task_struct *inmem_task; /* store inmemory task */ 801 struct mutex inmem_lock; /* lock for inmemory pages */ 802 struct extent_tree *extent_tree; /* cached extent_tree entry */ 803 804 /* avoid racing between foreground op and gc */ 805 struct f2fs_rwsem i_gc_rwsem[2]; 806 struct f2fs_rwsem i_xattr_sem; /* avoid racing between reading and changing EAs */ 807 808 int i_extra_isize; /* size of extra space located in i_addr */ 809 kprojid_t i_projid; /* id for project quota */ 810 int i_inline_xattr_size; /* inline xattr size */ 811 struct timespec64 i_crtime; /* inode creation time */ 812 struct timespec64 i_disk_time[4];/* inode disk times */ 813 814 /* for file compress */ 815 atomic_t i_compr_blocks; /* # of compressed blocks */ 816 unsigned char i_compress_algorithm; /* algorithm type */ 817 unsigned char i_log_cluster_size; /* log of cluster size */ 818 unsigned char i_compress_level; /* compress level (lz4hc,zstd) */ 819 unsigned short i_compress_flag; /* compress flag */ 820 unsigned int i_cluster_size; /* cluster size */ 821 }; 822 823 static inline void get_extent_info(struct extent_info *ext, 824 struct f2fs_extent *i_ext) 825 { 826 ext->fofs = le32_to_cpu(i_ext->fofs); 827 ext->blk = le32_to_cpu(i_ext->blk); 828 ext->len = le32_to_cpu(i_ext->len); 829 } 830 831 static inline void set_raw_extent(struct extent_info *ext, 832 struct f2fs_extent *i_ext) 833 { 834 i_ext->fofs = cpu_to_le32(ext->fofs); 835 i_ext->blk = cpu_to_le32(ext->blk); 836 i_ext->len = cpu_to_le32(ext->len); 837 } 838 839 static inline void set_extent_info(struct extent_info *ei, unsigned int fofs, 840 u32 blk, unsigned int len) 841 { 842 ei->fofs = fofs; 843 ei->blk = blk; 844 ei->len = len; 845 #ifdef CONFIG_F2FS_FS_COMPRESSION 846 ei->c_len = 0; 847 #endif 848 } 849 850 static inline bool __is_discard_mergeable(struct discard_info *back, 851 struct discard_info *front, unsigned int max_len) 852 { 853 return (back->lstart + back->len == front->lstart) && 854 (back->len + front->len <= max_len); 855 } 856 857 static inline bool __is_discard_back_mergeable(struct discard_info *cur, 858 struct discard_info *back, unsigned int max_len) 859 { 860 return __is_discard_mergeable(back, cur, max_len); 861 } 862 863 static inline bool __is_discard_front_mergeable(struct discard_info *cur, 864 struct discard_info *front, unsigned int max_len) 865 { 866 return __is_discard_mergeable(cur, front, max_len); 867 } 868 869 static inline bool __is_extent_mergeable(struct extent_info *back, 870 struct extent_info *front) 871 { 872 #ifdef CONFIG_F2FS_FS_COMPRESSION 873 if (back->c_len && back->len != back->c_len) 874 return false; 875 if (front->c_len && front->len != front->c_len) 876 return false; 877 #endif 878 return (back->fofs + back->len == front->fofs && 879 back->blk + back->len == front->blk); 880 } 881 882 static inline bool __is_back_mergeable(struct extent_info *cur, 883 struct extent_info *back) 884 { 885 return __is_extent_mergeable(back, cur); 886 } 887 888 static inline bool __is_front_mergeable(struct extent_info *cur, 889 struct extent_info *front) 890 { 891 return __is_extent_mergeable(cur, front); 892 } 893 894 extern void f2fs_mark_inode_dirty_sync(struct inode *inode, bool sync); 895 static inline void __try_update_largest_extent(struct extent_tree *et, 896 struct extent_node *en) 897 { 898 if (en->ei.len > et->largest.len) { 899 et->largest = en->ei; 900 et->largest_updated = true; 901 } 902 } 903 904 /* 905 * For free nid management 906 */ 907 enum nid_state { 908 FREE_NID, /* newly added to free nid list */ 909 PREALLOC_NID, /* it is preallocated */ 910 MAX_NID_STATE, 911 }; 912 913 enum nat_state { 914 TOTAL_NAT, 915 DIRTY_NAT, 916 RECLAIMABLE_NAT, 917 MAX_NAT_STATE, 918 }; 919 920 struct f2fs_nm_info { 921 block_t nat_blkaddr; /* base disk address of NAT */ 922 nid_t max_nid; /* maximum possible node ids */ 923 nid_t available_nids; /* # of available node ids */ 924 nid_t next_scan_nid; /* the next nid to be scanned */ 925 nid_t max_rf_node_blocks; /* max # of nodes for recovery */ 926 unsigned int ram_thresh; /* control the memory footprint */ 927 unsigned int ra_nid_pages; /* # of nid pages to be readaheaded */ 928 unsigned int dirty_nats_ratio; /* control dirty nats ratio threshold */ 929 930 /* NAT cache management */ 931 struct radix_tree_root nat_root;/* root of the nat entry cache */ 932 struct radix_tree_root nat_set_root;/* root of the nat set cache */ 933 struct f2fs_rwsem nat_tree_lock; /* protect nat entry tree */ 934 struct list_head nat_entries; /* cached nat entry list (clean) */ 935 spinlock_t nat_list_lock; /* protect clean nat entry list */ 936 unsigned int nat_cnt[MAX_NAT_STATE]; /* the # of cached nat entries */ 937 unsigned int nat_blocks; /* # of nat blocks */ 938 939 /* free node ids management */ 940 struct radix_tree_root free_nid_root;/* root of the free_nid cache */ 941 struct list_head free_nid_list; /* list for free nids excluding preallocated nids */ 942 unsigned int nid_cnt[MAX_NID_STATE]; /* the number of free node id */ 943 spinlock_t nid_list_lock; /* protect nid lists ops */ 944 struct mutex build_lock; /* lock for build free nids */ 945 unsigned char **free_nid_bitmap; 946 unsigned char *nat_block_bitmap; 947 unsigned short *free_nid_count; /* free nid count of NAT block */ 948 949 /* for checkpoint */ 950 char *nat_bitmap; /* NAT bitmap pointer */ 951 952 unsigned int nat_bits_blocks; /* # of nat bits blocks */ 953 unsigned char *nat_bits; /* NAT bits blocks */ 954 unsigned char *full_nat_bits; /* full NAT pages */ 955 unsigned char *empty_nat_bits; /* empty NAT pages */ 956 #ifdef CONFIG_F2FS_CHECK_FS 957 char *nat_bitmap_mir; /* NAT bitmap mirror */ 958 #endif 959 int bitmap_size; /* bitmap size */ 960 }; 961 962 /* 963 * this structure is used as one of function parameters. 964 * all the information are dedicated to a given direct node block determined 965 * by the data offset in a file. 966 */ 967 struct dnode_of_data { 968 struct inode *inode; /* vfs inode pointer */ 969 struct page *inode_page; /* its inode page, NULL is possible */ 970 struct page *node_page; /* cached direct node page */ 971 nid_t nid; /* node id of the direct node block */ 972 unsigned int ofs_in_node; /* data offset in the node page */ 973 bool inode_page_locked; /* inode page is locked or not */ 974 bool node_changed; /* is node block changed */ 975 char cur_level; /* level of hole node page */ 976 char max_level; /* level of current page located */ 977 block_t data_blkaddr; /* block address of the node block */ 978 }; 979 980 static inline void set_new_dnode(struct dnode_of_data *dn, struct inode *inode, 981 struct page *ipage, struct page *npage, nid_t nid) 982 { 983 memset(dn, 0, sizeof(*dn)); 984 dn->inode = inode; 985 dn->inode_page = ipage; 986 dn->node_page = npage; 987 dn->nid = nid; 988 } 989 990 /* 991 * For SIT manager 992 * 993 * By default, there are 6 active log areas across the whole main area. 994 * When considering hot and cold data separation to reduce cleaning overhead, 995 * we split 3 for data logs and 3 for node logs as hot, warm, and cold types, 996 * respectively. 997 * In the current design, you should not change the numbers intentionally. 998 * Instead, as a mount option such as active_logs=x, you can use 2, 4, and 6 999 * logs individually according to the underlying devices. (default: 6) 1000 * Just in case, on-disk layout covers maximum 16 logs that consist of 8 for 1001 * data and 8 for node logs. 1002 */ 1003 #define NR_CURSEG_DATA_TYPE (3) 1004 #define NR_CURSEG_NODE_TYPE (3) 1005 #define NR_CURSEG_INMEM_TYPE (2) 1006 #define NR_CURSEG_RO_TYPE (2) 1007 #define NR_CURSEG_PERSIST_TYPE (NR_CURSEG_DATA_TYPE + NR_CURSEG_NODE_TYPE) 1008 #define NR_CURSEG_TYPE (NR_CURSEG_INMEM_TYPE + NR_CURSEG_PERSIST_TYPE) 1009 1010 enum { 1011 CURSEG_HOT_DATA = 0, /* directory entry blocks */ 1012 CURSEG_WARM_DATA, /* data blocks */ 1013 CURSEG_COLD_DATA, /* multimedia or GCed data blocks */ 1014 CURSEG_HOT_NODE, /* direct node blocks of directory files */ 1015 CURSEG_WARM_NODE, /* direct node blocks of normal files */ 1016 CURSEG_COLD_NODE, /* indirect node blocks */ 1017 NR_PERSISTENT_LOG, /* number of persistent log */ 1018 CURSEG_COLD_DATA_PINNED = NR_PERSISTENT_LOG, 1019 /* pinned file that needs consecutive block address */ 1020 CURSEG_ALL_DATA_ATGC, /* SSR alloctor in hot/warm/cold data area */ 1021 NO_CHECK_TYPE, /* number of persistent & inmem log */ 1022 }; 1023 1024 struct flush_cmd { 1025 struct completion wait; 1026 struct llist_node llnode; 1027 nid_t ino; 1028 int ret; 1029 }; 1030 1031 struct flush_cmd_control { 1032 struct task_struct *f2fs_issue_flush; /* flush thread */ 1033 wait_queue_head_t flush_wait_queue; /* waiting queue for wake-up */ 1034 atomic_t issued_flush; /* # of issued flushes */ 1035 atomic_t queued_flush; /* # of queued flushes */ 1036 struct llist_head issue_list; /* list for command issue */ 1037 struct llist_node *dispatch_list; /* list for command dispatch */ 1038 }; 1039 1040 struct f2fs_sm_info { 1041 struct sit_info *sit_info; /* whole segment information */ 1042 struct free_segmap_info *free_info; /* free segment information */ 1043 struct dirty_seglist_info *dirty_info; /* dirty segment information */ 1044 struct curseg_info *curseg_array; /* active segment information */ 1045 1046 struct f2fs_rwsem curseg_lock; /* for preventing curseg change */ 1047 1048 block_t seg0_blkaddr; /* block address of 0'th segment */ 1049 block_t main_blkaddr; /* start block address of main area */ 1050 block_t ssa_blkaddr; /* start block address of SSA area */ 1051 1052 unsigned int segment_count; /* total # of segments */ 1053 unsigned int main_segments; /* # of segments in main area */ 1054 unsigned int reserved_segments; /* # of reserved segments */ 1055 unsigned int additional_reserved_segments;/* reserved segs for IO align feature */ 1056 unsigned int ovp_segments; /* # of overprovision segments */ 1057 1058 /* a threshold to reclaim prefree segments */ 1059 unsigned int rec_prefree_segments; 1060 1061 /* for batched trimming */ 1062 unsigned int trim_sections; /* # of sections to trim */ 1063 1064 struct list_head sit_entry_set; /* sit entry set list */ 1065 1066 unsigned int ipu_policy; /* in-place-update policy */ 1067 unsigned int min_ipu_util; /* in-place-update threshold */ 1068 unsigned int min_fsync_blocks; /* threshold for fsync */ 1069 unsigned int min_seq_blocks; /* threshold for sequential blocks */ 1070 unsigned int min_hot_blocks; /* threshold for hot block allocation */ 1071 unsigned int min_ssr_sections; /* threshold to trigger SSR allocation */ 1072 1073 /* for flush command control */ 1074 struct flush_cmd_control *fcc_info; 1075 1076 /* for discard command control */ 1077 struct discard_cmd_control *dcc_info; 1078 }; 1079 1080 /* 1081 * For superblock 1082 */ 1083 /* 1084 * COUNT_TYPE for monitoring 1085 * 1086 * f2fs monitors the number of several block types such as on-writeback, 1087 * dirty dentry blocks, dirty node blocks, and dirty meta blocks. 1088 */ 1089 #define WB_DATA_TYPE(p) (__is_cp_guaranteed(p) ? F2FS_WB_CP_DATA : F2FS_WB_DATA) 1090 enum count_type { 1091 F2FS_DIRTY_DENTS, 1092 F2FS_DIRTY_DATA, 1093 F2FS_DIRTY_QDATA, 1094 F2FS_DIRTY_NODES, 1095 F2FS_DIRTY_META, 1096 F2FS_INMEM_PAGES, 1097 F2FS_DIRTY_IMETA, 1098 F2FS_WB_CP_DATA, 1099 F2FS_WB_DATA, 1100 F2FS_RD_DATA, 1101 F2FS_RD_NODE, 1102 F2FS_RD_META, 1103 F2FS_DIO_WRITE, 1104 F2FS_DIO_READ, 1105 NR_COUNT_TYPE, 1106 }; 1107 1108 /* 1109 * The below are the page types of bios used in submit_bio(). 1110 * The available types are: 1111 * DATA User data pages. It operates as async mode. 1112 * NODE Node pages. It operates as async mode. 1113 * META FS metadata pages such as SIT, NAT, CP. 1114 * NR_PAGE_TYPE The number of page types. 1115 * META_FLUSH Make sure the previous pages are written 1116 * with waiting the bio's completion 1117 * ... Only can be used with META. 1118 */ 1119 #define PAGE_TYPE_OF_BIO(type) ((type) > META ? META : (type)) 1120 enum page_type { 1121 DATA, 1122 NODE, 1123 META, 1124 NR_PAGE_TYPE, 1125 META_FLUSH, 1126 INMEM, /* the below types are used by tracepoints only. */ 1127 INMEM_DROP, 1128 INMEM_INVALIDATE, 1129 INMEM_REVOKE, 1130 IPU, 1131 OPU, 1132 }; 1133 1134 enum temp_type { 1135 HOT = 0, /* must be zero for meta bio */ 1136 WARM, 1137 COLD, 1138 NR_TEMP_TYPE, 1139 }; 1140 1141 enum need_lock_type { 1142 LOCK_REQ = 0, 1143 LOCK_DONE, 1144 LOCK_RETRY, 1145 }; 1146 1147 enum cp_reason_type { 1148 CP_NO_NEEDED, 1149 CP_NON_REGULAR, 1150 CP_COMPRESSED, 1151 CP_HARDLINK, 1152 CP_SB_NEED_CP, 1153 CP_WRONG_PINO, 1154 CP_NO_SPC_ROLL, 1155 CP_NODE_NEED_CP, 1156 CP_FASTBOOT_MODE, 1157 CP_SPEC_LOG_NUM, 1158 CP_RECOVER_DIR, 1159 }; 1160 1161 enum iostat_type { 1162 /* WRITE IO */ 1163 APP_DIRECT_IO, /* app direct write IOs */ 1164 APP_BUFFERED_IO, /* app buffered write IOs */ 1165 APP_WRITE_IO, /* app write IOs */ 1166 APP_MAPPED_IO, /* app mapped IOs */ 1167 FS_DATA_IO, /* data IOs from kworker/fsync/reclaimer */ 1168 FS_NODE_IO, /* node IOs from kworker/fsync/reclaimer */ 1169 FS_META_IO, /* meta IOs from kworker/reclaimer */ 1170 FS_GC_DATA_IO, /* data IOs from forground gc */ 1171 FS_GC_NODE_IO, /* node IOs from forground gc */ 1172 FS_CP_DATA_IO, /* data IOs from checkpoint */ 1173 FS_CP_NODE_IO, /* node IOs from checkpoint */ 1174 FS_CP_META_IO, /* meta IOs from checkpoint */ 1175 1176 /* READ IO */ 1177 APP_DIRECT_READ_IO, /* app direct read IOs */ 1178 APP_BUFFERED_READ_IO, /* app buffered read IOs */ 1179 APP_READ_IO, /* app read IOs */ 1180 APP_MAPPED_READ_IO, /* app mapped read IOs */ 1181 FS_DATA_READ_IO, /* data read IOs */ 1182 FS_GDATA_READ_IO, /* data read IOs from background gc */ 1183 FS_CDATA_READ_IO, /* compressed data read IOs */ 1184 FS_NODE_READ_IO, /* node read IOs */ 1185 FS_META_READ_IO, /* meta read IOs */ 1186 1187 /* other */ 1188 FS_DISCARD, /* discard */ 1189 NR_IO_TYPE, 1190 }; 1191 1192 struct f2fs_io_info { 1193 struct f2fs_sb_info *sbi; /* f2fs_sb_info pointer */ 1194 nid_t ino; /* inode number */ 1195 enum page_type type; /* contains DATA/NODE/META/META_FLUSH */ 1196 enum temp_type temp; /* contains HOT/WARM/COLD */ 1197 int op; /* contains REQ_OP_ */ 1198 int op_flags; /* req_flag_bits */ 1199 block_t new_blkaddr; /* new block address to be written */ 1200 block_t old_blkaddr; /* old block address before Cow */ 1201 struct page *page; /* page to be written */ 1202 struct page *encrypted_page; /* encrypted page */ 1203 struct page *compressed_page; /* compressed page */ 1204 struct list_head list; /* serialize IOs */ 1205 bool submitted; /* indicate IO submission */ 1206 int need_lock; /* indicate we need to lock cp_rwsem */ 1207 bool in_list; /* indicate fio is in io_list */ 1208 bool is_por; /* indicate IO is from recovery or not */ 1209 bool retry; /* need to reallocate block address */ 1210 int compr_blocks; /* # of compressed block addresses */ 1211 bool encrypted; /* indicate file is encrypted */ 1212 enum iostat_type io_type; /* io type */ 1213 struct writeback_control *io_wbc; /* writeback control */ 1214 struct bio **bio; /* bio for ipu */ 1215 sector_t *last_block; /* last block number in bio */ 1216 unsigned char version; /* version of the node */ 1217 }; 1218 1219 struct bio_entry { 1220 struct bio *bio; 1221 struct list_head list; 1222 }; 1223 1224 #define is_read_io(rw) ((rw) == READ) 1225 struct f2fs_bio_info { 1226 struct f2fs_sb_info *sbi; /* f2fs superblock */ 1227 struct bio *bio; /* bios to merge */ 1228 sector_t last_block_in_bio; /* last block number */ 1229 struct f2fs_io_info fio; /* store buffered io info. */ 1230 struct f2fs_rwsem io_rwsem; /* blocking op for bio */ 1231 spinlock_t io_lock; /* serialize DATA/NODE IOs */ 1232 struct list_head io_list; /* track fios */ 1233 struct list_head bio_list; /* bio entry list head */ 1234 struct f2fs_rwsem bio_list_lock; /* lock to protect bio entry list */ 1235 }; 1236 1237 #define FDEV(i) (sbi->devs[i]) 1238 #define RDEV(i) (raw_super->devs[i]) 1239 struct f2fs_dev_info { 1240 struct block_device *bdev; 1241 char path[MAX_PATH_LEN]; 1242 unsigned int total_segments; 1243 block_t start_blk; 1244 block_t end_blk; 1245 #ifdef CONFIG_BLK_DEV_ZONED 1246 unsigned int nr_blkz; /* Total number of zones */ 1247 unsigned long *blkz_seq; /* Bitmap indicating sequential zones */ 1248 block_t *zone_capacity_blocks; /* Array of zone capacity in blks */ 1249 #endif 1250 }; 1251 1252 enum inode_type { 1253 DIR_INODE, /* for dirty dir inode */ 1254 FILE_INODE, /* for dirty regular/symlink inode */ 1255 DIRTY_META, /* for all dirtied inode metadata */ 1256 ATOMIC_FILE, /* for all atomic files */ 1257 NR_INODE_TYPE, 1258 }; 1259 1260 /* for inner inode cache management */ 1261 struct inode_management { 1262 struct radix_tree_root ino_root; /* ino entry array */ 1263 spinlock_t ino_lock; /* for ino entry lock */ 1264 struct list_head ino_list; /* inode list head */ 1265 unsigned long ino_num; /* number of entries */ 1266 }; 1267 1268 /* for GC_AT */ 1269 struct atgc_management { 1270 bool atgc_enabled; /* ATGC is enabled or not */ 1271 struct rb_root_cached root; /* root of victim rb-tree */ 1272 struct list_head victim_list; /* linked with all victim entries */ 1273 unsigned int victim_count; /* victim count in rb-tree */ 1274 unsigned int candidate_ratio; /* candidate ratio */ 1275 unsigned int max_candidate_count; /* max candidate count */ 1276 unsigned int age_weight; /* age weight, vblock_weight = 100 - age_weight */ 1277 unsigned long long age_threshold; /* age threshold */ 1278 }; 1279 1280 /* For s_flag in struct f2fs_sb_info */ 1281 enum { 1282 SBI_IS_DIRTY, /* dirty flag for checkpoint */ 1283 SBI_IS_CLOSE, /* specify unmounting */ 1284 SBI_NEED_FSCK, /* need fsck.f2fs to fix */ 1285 SBI_POR_DOING, /* recovery is doing or not */ 1286 SBI_NEED_SB_WRITE, /* need to recover superblock */ 1287 SBI_NEED_CP, /* need to checkpoint */ 1288 SBI_IS_SHUTDOWN, /* shutdown by ioctl */ 1289 SBI_IS_RECOVERED, /* recovered orphan/data */ 1290 SBI_CP_DISABLED, /* CP was disabled last mount */ 1291 SBI_CP_DISABLED_QUICK, /* CP was disabled quickly */ 1292 SBI_QUOTA_NEED_FLUSH, /* need to flush quota info in CP */ 1293 SBI_QUOTA_SKIP_FLUSH, /* skip flushing quota in current CP */ 1294 SBI_QUOTA_NEED_REPAIR, /* quota file may be corrupted */ 1295 SBI_IS_RESIZEFS, /* resizefs is in process */ 1296 SBI_IS_FREEZING, /* freezefs is in process */ 1297 }; 1298 1299 enum { 1300 CP_TIME, 1301 REQ_TIME, 1302 DISCARD_TIME, 1303 GC_TIME, 1304 DISABLE_TIME, 1305 UMOUNT_DISCARD_TIMEOUT, 1306 MAX_TIME, 1307 }; 1308 1309 enum { 1310 GC_NORMAL, 1311 GC_IDLE_CB, 1312 GC_IDLE_GREEDY, 1313 GC_IDLE_AT, 1314 GC_URGENT_HIGH, 1315 GC_URGENT_LOW, 1316 MAX_GC_MODE, 1317 }; 1318 1319 enum { 1320 BGGC_MODE_ON, /* background gc is on */ 1321 BGGC_MODE_OFF, /* background gc is off */ 1322 BGGC_MODE_SYNC, /* 1323 * background gc is on, migrating blocks 1324 * like foreground gc 1325 */ 1326 }; 1327 1328 enum { 1329 FS_MODE_ADAPTIVE, /* use both lfs/ssr allocation */ 1330 FS_MODE_LFS, /* use lfs allocation only */ 1331 FS_MODE_FRAGMENT_SEG, /* segment fragmentation mode */ 1332 FS_MODE_FRAGMENT_BLK, /* block fragmentation mode */ 1333 }; 1334 1335 enum { 1336 WHINT_MODE_OFF, /* not pass down write hints */ 1337 WHINT_MODE_USER, /* try to pass down hints given by users */ 1338 WHINT_MODE_FS, /* pass down hints with F2FS policy */ 1339 }; 1340 1341 enum { 1342 ALLOC_MODE_DEFAULT, /* stay default */ 1343 ALLOC_MODE_REUSE, /* reuse segments as much as possible */ 1344 }; 1345 1346 enum fsync_mode { 1347 FSYNC_MODE_POSIX, /* fsync follows posix semantics */ 1348 FSYNC_MODE_STRICT, /* fsync behaves in line with ext4 */ 1349 FSYNC_MODE_NOBARRIER, /* fsync behaves nobarrier based on posix */ 1350 }; 1351 1352 enum { 1353 COMPR_MODE_FS, /* 1354 * automatically compress compression 1355 * enabled files 1356 */ 1357 COMPR_MODE_USER, /* 1358 * automatical compression is disabled. 1359 * user can control the file compression 1360 * using ioctls 1361 */ 1362 }; 1363 1364 enum { 1365 DISCARD_UNIT_BLOCK, /* basic discard unit is block */ 1366 DISCARD_UNIT_SEGMENT, /* basic discard unit is segment */ 1367 DISCARD_UNIT_SECTION, /* basic discard unit is section */ 1368 }; 1369 1370 static inline int f2fs_test_bit(unsigned int nr, char *addr); 1371 static inline void f2fs_set_bit(unsigned int nr, char *addr); 1372 static inline void f2fs_clear_bit(unsigned int nr, char *addr); 1373 1374 /* 1375 * Layout of f2fs page.private: 1376 * 1377 * Layout A: lowest bit should be 1 1378 * | bit0 = 1 | bit1 | bit2 | ... | bit MAX | private data .... | 1379 * bit 0 PAGE_PRIVATE_NOT_POINTER 1380 * bit 1 PAGE_PRIVATE_ATOMIC_WRITE 1381 * bit 2 PAGE_PRIVATE_DUMMY_WRITE 1382 * bit 3 PAGE_PRIVATE_ONGOING_MIGRATION 1383 * bit 4 PAGE_PRIVATE_INLINE_INODE 1384 * bit 5 PAGE_PRIVATE_REF_RESOURCE 1385 * bit 6- f2fs private data 1386 * 1387 * Layout B: lowest bit should be 0 1388 * page.private is a wrapped pointer. 1389 */ 1390 enum { 1391 PAGE_PRIVATE_NOT_POINTER, /* private contains non-pointer data */ 1392 PAGE_PRIVATE_ATOMIC_WRITE, /* data page from atomic write path */ 1393 PAGE_PRIVATE_DUMMY_WRITE, /* data page for padding aligned IO */ 1394 PAGE_PRIVATE_ONGOING_MIGRATION, /* data page which is on-going migrating */ 1395 PAGE_PRIVATE_INLINE_INODE, /* inode page contains inline data */ 1396 PAGE_PRIVATE_REF_RESOURCE, /* dirty page has referenced resources */ 1397 PAGE_PRIVATE_MAX 1398 }; 1399 1400 #define PAGE_PRIVATE_GET_FUNC(name, flagname) \ 1401 static inline bool page_private_##name(struct page *page) \ 1402 { \ 1403 return PagePrivate(page) && \ 1404 test_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)) && \ 1405 test_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \ 1406 } 1407 1408 #define PAGE_PRIVATE_SET_FUNC(name, flagname) \ 1409 static inline void set_page_private_##name(struct page *page) \ 1410 { \ 1411 if (!PagePrivate(page)) { \ 1412 get_page(page); \ 1413 SetPagePrivate(page); \ 1414 set_page_private(page, 0); \ 1415 } \ 1416 set_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)); \ 1417 set_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \ 1418 } 1419 1420 #define PAGE_PRIVATE_CLEAR_FUNC(name, flagname) \ 1421 static inline void clear_page_private_##name(struct page *page) \ 1422 { \ 1423 clear_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \ 1424 if (page_private(page) == 1 << PAGE_PRIVATE_NOT_POINTER) { \ 1425 set_page_private(page, 0); \ 1426 if (PagePrivate(page)) { \ 1427 ClearPagePrivate(page); \ 1428 put_page(page); \ 1429 }\ 1430 } \ 1431 } 1432 1433 PAGE_PRIVATE_GET_FUNC(nonpointer, NOT_POINTER); 1434 PAGE_PRIVATE_GET_FUNC(reference, REF_RESOURCE); 1435 PAGE_PRIVATE_GET_FUNC(inline, INLINE_INODE); 1436 PAGE_PRIVATE_GET_FUNC(gcing, ONGOING_MIGRATION); 1437 PAGE_PRIVATE_GET_FUNC(atomic, ATOMIC_WRITE); 1438 PAGE_PRIVATE_GET_FUNC(dummy, DUMMY_WRITE); 1439 1440 PAGE_PRIVATE_SET_FUNC(reference, REF_RESOURCE); 1441 PAGE_PRIVATE_SET_FUNC(inline, INLINE_INODE); 1442 PAGE_PRIVATE_SET_FUNC(gcing, ONGOING_MIGRATION); 1443 PAGE_PRIVATE_SET_FUNC(atomic, ATOMIC_WRITE); 1444 PAGE_PRIVATE_SET_FUNC(dummy, DUMMY_WRITE); 1445 1446 PAGE_PRIVATE_CLEAR_FUNC(reference, REF_RESOURCE); 1447 PAGE_PRIVATE_CLEAR_FUNC(inline, INLINE_INODE); 1448 PAGE_PRIVATE_CLEAR_FUNC(gcing, ONGOING_MIGRATION); 1449 PAGE_PRIVATE_CLEAR_FUNC(atomic, ATOMIC_WRITE); 1450 PAGE_PRIVATE_CLEAR_FUNC(dummy, DUMMY_WRITE); 1451 1452 static inline unsigned long get_page_private_data(struct page *page) 1453 { 1454 unsigned long data = page_private(page); 1455 1456 if (!test_bit(PAGE_PRIVATE_NOT_POINTER, &data)) 1457 return 0; 1458 return data >> PAGE_PRIVATE_MAX; 1459 } 1460 1461 static inline void set_page_private_data(struct page *page, unsigned long data) 1462 { 1463 if (!PagePrivate(page)) { 1464 get_page(page); 1465 SetPagePrivate(page); 1466 set_page_private(page, 0); 1467 } 1468 set_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)); 1469 page_private(page) |= data << PAGE_PRIVATE_MAX; 1470 } 1471 1472 static inline void clear_page_private_data(struct page *page) 1473 { 1474 page_private(page) &= (1 << PAGE_PRIVATE_MAX) - 1; 1475 if (page_private(page) == 1 << PAGE_PRIVATE_NOT_POINTER) { 1476 set_page_private(page, 0); 1477 if (PagePrivate(page)) { 1478 ClearPagePrivate(page); 1479 put_page(page); 1480 } 1481 } 1482 } 1483 1484 /* For compression */ 1485 enum compress_algorithm_type { 1486 COMPRESS_LZO, 1487 COMPRESS_LZ4, 1488 COMPRESS_ZSTD, 1489 COMPRESS_LZORLE, 1490 COMPRESS_MAX, 1491 }; 1492 1493 enum compress_flag { 1494 COMPRESS_CHKSUM, 1495 COMPRESS_MAX_FLAG, 1496 }; 1497 1498 #define COMPRESS_WATERMARK 20 1499 #define COMPRESS_PERCENT 20 1500 1501 #define COMPRESS_DATA_RESERVED_SIZE 4 1502 struct compress_data { 1503 __le32 clen; /* compressed data size */ 1504 __le32 chksum; /* compressed data chksum */ 1505 __le32 reserved[COMPRESS_DATA_RESERVED_SIZE]; /* reserved */ 1506 u8 cdata[]; /* compressed data */ 1507 }; 1508 1509 #define COMPRESS_HEADER_SIZE (sizeof(struct compress_data)) 1510 1511 #define F2FS_COMPRESSED_PAGE_MAGIC 0xF5F2C000 1512 1513 #define COMPRESS_LEVEL_OFFSET 8 1514 1515 /* compress context */ 1516 struct compress_ctx { 1517 struct inode *inode; /* inode the context belong to */ 1518 pgoff_t cluster_idx; /* cluster index number */ 1519 unsigned int cluster_size; /* page count in cluster */ 1520 unsigned int log_cluster_size; /* log of cluster size */ 1521 struct page **rpages; /* pages store raw data in cluster */ 1522 unsigned int nr_rpages; /* total page number in rpages */ 1523 struct page **cpages; /* pages store compressed data in cluster */ 1524 unsigned int nr_cpages; /* total page number in cpages */ 1525 unsigned int valid_nr_cpages; /* valid page number in cpages */ 1526 void *rbuf; /* virtual mapped address on rpages */ 1527 struct compress_data *cbuf; /* virtual mapped address on cpages */ 1528 size_t rlen; /* valid data length in rbuf */ 1529 size_t clen; /* valid data length in cbuf */ 1530 void *private; /* payload buffer for specified compression algorithm */ 1531 void *private2; /* extra payload buffer */ 1532 }; 1533 1534 /* compress context for write IO path */ 1535 struct compress_io_ctx { 1536 u32 magic; /* magic number to indicate page is compressed */ 1537 struct inode *inode; /* inode the context belong to */ 1538 struct page **rpages; /* pages store raw data in cluster */ 1539 unsigned int nr_rpages; /* total page number in rpages */ 1540 atomic_t pending_pages; /* in-flight compressed page count */ 1541 }; 1542 1543 /* Context for decompressing one cluster on the read IO path */ 1544 struct decompress_io_ctx { 1545 u32 magic; /* magic number to indicate page is compressed */ 1546 struct inode *inode; /* inode the context belong to */ 1547 pgoff_t cluster_idx; /* cluster index number */ 1548 unsigned int cluster_size; /* page count in cluster */ 1549 unsigned int log_cluster_size; /* log of cluster size */ 1550 struct page **rpages; /* pages store raw data in cluster */ 1551 unsigned int nr_rpages; /* total page number in rpages */ 1552 struct page **cpages; /* pages store compressed data in cluster */ 1553 unsigned int nr_cpages; /* total page number in cpages */ 1554 struct page **tpages; /* temp pages to pad holes in cluster */ 1555 void *rbuf; /* virtual mapped address on rpages */ 1556 struct compress_data *cbuf; /* virtual mapped address on cpages */ 1557 size_t rlen; /* valid data length in rbuf */ 1558 size_t clen; /* valid data length in cbuf */ 1559 1560 /* 1561 * The number of compressed pages remaining to be read in this cluster. 1562 * This is initially nr_cpages. It is decremented by 1 each time a page 1563 * has been read (or failed to be read). When it reaches 0, the cluster 1564 * is decompressed (or an error is reported). 1565 * 1566 * If an error occurs before all the pages have been submitted for I/O, 1567 * then this will never reach 0. In this case the I/O submitter is 1568 * responsible for calling f2fs_decompress_end_io() instead. 1569 */ 1570 atomic_t remaining_pages; 1571 1572 /* 1573 * Number of references to this decompress_io_ctx. 1574 * 1575 * One reference is held for I/O completion. This reference is dropped 1576 * after the pagecache pages are updated and unlocked -- either after 1577 * decompression (and verity if enabled), or after an error. 1578 * 1579 * In addition, each compressed page holds a reference while it is in a 1580 * bio. These references are necessary prevent compressed pages from 1581 * being freed while they are still in a bio. 1582 */ 1583 refcount_t refcnt; 1584 1585 bool failed; /* IO error occurred before decompression? */ 1586 bool need_verity; /* need fs-verity verification after decompression? */ 1587 void *private; /* payload buffer for specified decompression algorithm */ 1588 void *private2; /* extra payload buffer */ 1589 struct work_struct verity_work; /* work to verify the decompressed pages */ 1590 }; 1591 1592 #define NULL_CLUSTER ((unsigned int)(~0)) 1593 #define MIN_COMPRESS_LOG_SIZE 2 1594 #define MAX_COMPRESS_LOG_SIZE 8 1595 #define MAX_COMPRESS_WINDOW_SIZE(log_size) ((PAGE_SIZE) << (log_size)) 1596 1597 struct f2fs_sb_info { 1598 struct super_block *sb; /* pointer to VFS super block */ 1599 struct proc_dir_entry *s_proc; /* proc entry */ 1600 struct f2fs_super_block *raw_super; /* raw super block pointer */ 1601 struct f2fs_rwsem sb_lock; /* lock for raw super block */ 1602 int valid_super_block; /* valid super block no */ 1603 unsigned long s_flag; /* flags for sbi */ 1604 struct mutex writepages; /* mutex for writepages() */ 1605 1606 #ifdef CONFIG_BLK_DEV_ZONED 1607 unsigned int blocks_per_blkz; /* F2FS blocks per zone */ 1608 unsigned int log_blocks_per_blkz; /* log2 F2FS blocks per zone */ 1609 #endif 1610 1611 /* for node-related operations */ 1612 struct f2fs_nm_info *nm_info; /* node manager */ 1613 struct inode *node_inode; /* cache node blocks */ 1614 1615 /* for segment-related operations */ 1616 struct f2fs_sm_info *sm_info; /* segment manager */ 1617 1618 /* for bio operations */ 1619 struct f2fs_bio_info *write_io[NR_PAGE_TYPE]; /* for write bios */ 1620 /* keep migration IO order for LFS mode */ 1621 struct f2fs_rwsem io_order_lock; 1622 mempool_t *write_io_dummy; /* Dummy pages */ 1623 pgoff_t metapage_eio_ofs; /* EIO page offset */ 1624 int metapage_eio_cnt; /* EIO count */ 1625 1626 /* for checkpoint */ 1627 struct f2fs_checkpoint *ckpt; /* raw checkpoint pointer */ 1628 int cur_cp_pack; /* remain current cp pack */ 1629 spinlock_t cp_lock; /* for flag in ckpt */ 1630 struct inode *meta_inode; /* cache meta blocks */ 1631 struct f2fs_rwsem cp_global_sem; /* checkpoint procedure lock */ 1632 struct f2fs_rwsem cp_rwsem; /* blocking FS operations */ 1633 struct f2fs_rwsem node_write; /* locking node writes */ 1634 struct f2fs_rwsem node_change; /* locking node change */ 1635 wait_queue_head_t cp_wait; 1636 unsigned long last_time[MAX_TIME]; /* to store time in jiffies */ 1637 long interval_time[MAX_TIME]; /* to store thresholds */ 1638 struct ckpt_req_control cprc_info; /* for checkpoint request control */ 1639 1640 struct inode_management im[MAX_INO_ENTRY]; /* manage inode cache */ 1641 1642 spinlock_t fsync_node_lock; /* for node entry lock */ 1643 struct list_head fsync_node_list; /* node list head */ 1644 unsigned int fsync_seg_id; /* sequence id */ 1645 unsigned int fsync_node_num; /* number of node entries */ 1646 1647 /* for orphan inode, use 0'th array */ 1648 unsigned int max_orphans; /* max orphan inodes */ 1649 1650 /* for inode management */ 1651 struct list_head inode_list[NR_INODE_TYPE]; /* dirty inode list */ 1652 spinlock_t inode_lock[NR_INODE_TYPE]; /* for dirty inode list lock */ 1653 struct mutex flush_lock; /* for flush exclusion */ 1654 1655 /* for extent tree cache */ 1656 struct radix_tree_root extent_tree_root;/* cache extent cache entries */ 1657 struct mutex extent_tree_lock; /* locking extent radix tree */ 1658 struct list_head extent_list; /* lru list for shrinker */ 1659 spinlock_t extent_lock; /* locking extent lru list */ 1660 atomic_t total_ext_tree; /* extent tree count */ 1661 struct list_head zombie_list; /* extent zombie tree list */ 1662 atomic_t total_zombie_tree; /* extent zombie tree count */ 1663 atomic_t total_ext_node; /* extent info count */ 1664 1665 /* basic filesystem units */ 1666 unsigned int log_sectors_per_block; /* log2 sectors per block */ 1667 unsigned int log_blocksize; /* log2 block size */ 1668 unsigned int blocksize; /* block size */ 1669 unsigned int root_ino_num; /* root inode number*/ 1670 unsigned int node_ino_num; /* node inode number*/ 1671 unsigned int meta_ino_num; /* meta inode number*/ 1672 unsigned int log_blocks_per_seg; /* log2 blocks per segment */ 1673 unsigned int blocks_per_seg; /* blocks per segment */ 1674 unsigned int segs_per_sec; /* segments per section */ 1675 unsigned int secs_per_zone; /* sections per zone */ 1676 unsigned int total_sections; /* total section count */ 1677 unsigned int total_node_count; /* total node block count */ 1678 unsigned int total_valid_node_count; /* valid node block count */ 1679 int dir_level; /* directory level */ 1680 int readdir_ra; /* readahead inode in readdir */ 1681 u64 max_io_bytes; /* max io bytes to merge IOs */ 1682 1683 block_t user_block_count; /* # of user blocks */ 1684 block_t total_valid_block_count; /* # of valid blocks */ 1685 block_t discard_blks; /* discard command candidats */ 1686 block_t last_valid_block_count; /* for recovery */ 1687 block_t reserved_blocks; /* configurable reserved blocks */ 1688 block_t current_reserved_blocks; /* current reserved blocks */ 1689 1690 /* Additional tracking for no checkpoint mode */ 1691 block_t unusable_block_count; /* # of blocks saved by last cp */ 1692 1693 unsigned int nquota_files; /* # of quota sysfile */ 1694 struct f2fs_rwsem quota_sem; /* blocking cp for flags */ 1695 1696 /* # of pages, see count_type */ 1697 atomic_t nr_pages[NR_COUNT_TYPE]; 1698 /* # of allocated blocks */ 1699 struct percpu_counter alloc_valid_block_count; 1700 /* # of node block writes as roll forward recovery */ 1701 struct percpu_counter rf_node_block_count; 1702 1703 /* writeback control */ 1704 atomic_t wb_sync_req[META]; /* count # of WB_SYNC threads */ 1705 1706 /* valid inode count */ 1707 struct percpu_counter total_valid_inode_count; 1708 1709 struct f2fs_mount_info mount_opt; /* mount options */ 1710 1711 /* for cleaning operations */ 1712 struct f2fs_rwsem gc_lock; /* 1713 * semaphore for GC, avoid 1714 * race between GC and GC or CP 1715 */ 1716 struct f2fs_gc_kthread *gc_thread; /* GC thread */ 1717 struct atgc_management am; /* atgc management */ 1718 unsigned int cur_victim_sec; /* current victim section num */ 1719 unsigned int gc_mode; /* current GC state */ 1720 unsigned int next_victim_seg[2]; /* next segment in victim section */ 1721 spinlock_t gc_urgent_high_lock; 1722 bool gc_urgent_high_limited; /* indicates having limited trial count */ 1723 unsigned int gc_urgent_high_remaining; /* remaining trial count for GC_URGENT_HIGH */ 1724 1725 /* for skip statistic */ 1726 unsigned int atomic_files; /* # of opened atomic file */ 1727 unsigned long long skipped_atomic_files[2]; /* FG_GC and BG_GC */ 1728 unsigned long long skipped_gc_rwsem; /* FG_GC only */ 1729 1730 /* threshold for gc trials on pinned files */ 1731 u64 gc_pin_file_threshold; 1732 struct f2fs_rwsem pin_sem; 1733 1734 /* maximum # of trials to find a victim segment for SSR and GC */ 1735 unsigned int max_victim_search; 1736 /* migration granularity of garbage collection, unit: segment */ 1737 unsigned int migration_granularity; 1738 1739 /* 1740 * for stat information. 1741 * one is for the LFS mode, and the other is for the SSR mode. 1742 */ 1743 #ifdef CONFIG_F2FS_STAT_FS 1744 struct f2fs_stat_info *stat_info; /* FS status information */ 1745 atomic_t meta_count[META_MAX]; /* # of meta blocks */ 1746 unsigned int segment_count[2]; /* # of allocated segments */ 1747 unsigned int block_count[2]; /* # of allocated blocks */ 1748 atomic_t inplace_count; /* # of inplace update */ 1749 atomic64_t total_hit_ext; /* # of lookup extent cache */ 1750 atomic64_t read_hit_rbtree; /* # of hit rbtree extent node */ 1751 atomic64_t read_hit_largest; /* # of hit largest extent node */ 1752 atomic64_t read_hit_cached; /* # of hit cached extent node */ 1753 atomic_t inline_xattr; /* # of inline_xattr inodes */ 1754 atomic_t inline_inode; /* # of inline_data inodes */ 1755 atomic_t inline_dir; /* # of inline_dentry inodes */ 1756 atomic_t compr_inode; /* # of compressed inodes */ 1757 atomic64_t compr_blocks; /* # of compressed blocks */ 1758 atomic_t vw_cnt; /* # of volatile writes */ 1759 atomic_t max_aw_cnt; /* max # of atomic writes */ 1760 atomic_t max_vw_cnt; /* max # of volatile writes */ 1761 unsigned int io_skip_bggc; /* skip background gc for in-flight IO */ 1762 unsigned int other_skip_bggc; /* skip background gc for other reasons */ 1763 unsigned int ndirty_inode[NR_INODE_TYPE]; /* # of dirty inodes */ 1764 #endif 1765 spinlock_t stat_lock; /* lock for stat operations */ 1766 1767 /* to attach REQ_META|REQ_FUA flags */ 1768 unsigned int data_io_flag; 1769 unsigned int node_io_flag; 1770 1771 /* For sysfs suppport */ 1772 struct kobject s_kobj; /* /sys/fs/f2fs/<devname> */ 1773 struct completion s_kobj_unregister; 1774 1775 struct kobject s_stat_kobj; /* /sys/fs/f2fs/<devname>/stat */ 1776 struct completion s_stat_kobj_unregister; 1777 1778 struct kobject s_feature_list_kobj; /* /sys/fs/f2fs/<devname>/feature_list */ 1779 struct completion s_feature_list_kobj_unregister; 1780 1781 /* For shrinker support */ 1782 struct list_head s_list; 1783 struct mutex umount_mutex; 1784 unsigned int shrinker_run_no; 1785 1786 /* For multi devices */ 1787 int s_ndevs; /* number of devices */ 1788 struct f2fs_dev_info *devs; /* for device list */ 1789 unsigned int dirty_device; /* for checkpoint data flush */ 1790 spinlock_t dev_lock; /* protect dirty_device */ 1791 bool aligned_blksize; /* all devices has the same logical blksize */ 1792 1793 /* For write statistics */ 1794 u64 sectors_written_start; 1795 u64 kbytes_written; 1796 1797 /* Reference to checksum algorithm driver via cryptoapi */ 1798 struct crypto_shash *s_chksum_driver; 1799 1800 /* Precomputed FS UUID checksum for seeding other checksums */ 1801 __u32 s_chksum_seed; 1802 1803 struct workqueue_struct *post_read_wq; /* post read workqueue */ 1804 1805 struct kmem_cache *inline_xattr_slab; /* inline xattr entry */ 1806 unsigned int inline_xattr_slab_size; /* default inline xattr slab size */ 1807 1808 /* For reclaimed segs statistics per each GC mode */ 1809 unsigned int gc_segment_mode; /* GC state for reclaimed segments */ 1810 unsigned int gc_reclaimed_segs[MAX_GC_MODE]; /* Reclaimed segs for each mode */ 1811 1812 unsigned long seq_file_ra_mul; /* multiplier for ra_pages of seq. files in fadvise */ 1813 1814 int max_fragment_chunk; /* max chunk size for block fragmentation mode */ 1815 int max_fragment_hole; /* max hole size for block fragmentation mode */ 1816 1817 #ifdef CONFIG_F2FS_FS_COMPRESSION 1818 struct kmem_cache *page_array_slab; /* page array entry */ 1819 unsigned int page_array_slab_size; /* default page array slab size */ 1820 1821 /* For runtime compression statistics */ 1822 u64 compr_written_block; 1823 u64 compr_saved_block; 1824 u32 compr_new_inode; 1825 1826 /* For compressed block cache */ 1827 struct inode *compress_inode; /* cache compressed blocks */ 1828 unsigned int compress_percent; /* cache page percentage */ 1829 unsigned int compress_watermark; /* cache page watermark */ 1830 atomic_t compress_page_hit; /* cache hit count */ 1831 #endif 1832 1833 #ifdef CONFIG_F2FS_IOSTAT 1834 /* For app/fs IO statistics */ 1835 spinlock_t iostat_lock; 1836 unsigned long long rw_iostat[NR_IO_TYPE]; 1837 unsigned long long prev_rw_iostat[NR_IO_TYPE]; 1838 bool iostat_enable; 1839 unsigned long iostat_next_period; 1840 unsigned int iostat_period_ms; 1841 1842 /* For io latency related statistics info in one iostat period */ 1843 spinlock_t iostat_lat_lock; 1844 struct iostat_lat_info *iostat_io_lat; 1845 #endif 1846 }; 1847 1848 #ifdef CONFIG_F2FS_FAULT_INJECTION 1849 #define f2fs_show_injection_info(sbi, type) \ 1850 printk_ratelimited("%sF2FS-fs (%s) : inject %s in %s of %pS\n", \ 1851 KERN_INFO, sbi->sb->s_id, \ 1852 f2fs_fault_name[type], \ 1853 __func__, __builtin_return_address(0)) 1854 static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type) 1855 { 1856 struct f2fs_fault_info *ffi = &F2FS_OPTION(sbi).fault_info; 1857 1858 if (!ffi->inject_rate) 1859 return false; 1860 1861 if (!IS_FAULT_SET(ffi, type)) 1862 return false; 1863 1864 atomic_inc(&ffi->inject_ops); 1865 if (atomic_read(&ffi->inject_ops) >= ffi->inject_rate) { 1866 atomic_set(&ffi->inject_ops, 0); 1867 return true; 1868 } 1869 return false; 1870 } 1871 #else 1872 #define f2fs_show_injection_info(sbi, type) do { } while (0) 1873 static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type) 1874 { 1875 return false; 1876 } 1877 #endif 1878 1879 /* 1880 * Test if the mounted volume is a multi-device volume. 1881 * - For a single regular disk volume, sbi->s_ndevs is 0. 1882 * - For a single zoned disk volume, sbi->s_ndevs is 1. 1883 * - For a multi-device volume, sbi->s_ndevs is always 2 or more. 1884 */ 1885 static inline bool f2fs_is_multi_device(struct f2fs_sb_info *sbi) 1886 { 1887 return sbi->s_ndevs > 1; 1888 } 1889 1890 static inline void f2fs_update_time(struct f2fs_sb_info *sbi, int type) 1891 { 1892 unsigned long now = jiffies; 1893 1894 sbi->last_time[type] = now; 1895 1896 /* DISCARD_TIME and GC_TIME are based on REQ_TIME */ 1897 if (type == REQ_TIME) { 1898 sbi->last_time[DISCARD_TIME] = now; 1899 sbi->last_time[GC_TIME] = now; 1900 } 1901 } 1902 1903 static inline bool f2fs_time_over(struct f2fs_sb_info *sbi, int type) 1904 { 1905 unsigned long interval = sbi->interval_time[type] * HZ; 1906 1907 return time_after(jiffies, sbi->last_time[type] + interval); 1908 } 1909 1910 static inline unsigned int f2fs_time_to_wait(struct f2fs_sb_info *sbi, 1911 int type) 1912 { 1913 unsigned long interval = sbi->interval_time[type] * HZ; 1914 unsigned int wait_ms = 0; 1915 long delta; 1916 1917 delta = (sbi->last_time[type] + interval) - jiffies; 1918 if (delta > 0) 1919 wait_ms = jiffies_to_msecs(delta); 1920 1921 return wait_ms; 1922 } 1923 1924 /* 1925 * Inline functions 1926 */ 1927 static inline u32 __f2fs_crc32(struct f2fs_sb_info *sbi, u32 crc, 1928 const void *address, unsigned int length) 1929 { 1930 struct { 1931 struct shash_desc shash; 1932 char ctx[4]; 1933 } desc; 1934 int err; 1935 1936 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver) != sizeof(desc.ctx)); 1937 1938 desc.shash.tfm = sbi->s_chksum_driver; 1939 *(u32 *)desc.ctx = crc; 1940 1941 err = crypto_shash_update(&desc.shash, address, length); 1942 BUG_ON(err); 1943 1944 return *(u32 *)desc.ctx; 1945 } 1946 1947 static inline u32 f2fs_crc32(struct f2fs_sb_info *sbi, const void *address, 1948 unsigned int length) 1949 { 1950 return __f2fs_crc32(sbi, F2FS_SUPER_MAGIC, address, length); 1951 } 1952 1953 static inline bool f2fs_crc_valid(struct f2fs_sb_info *sbi, __u32 blk_crc, 1954 void *buf, size_t buf_size) 1955 { 1956 return f2fs_crc32(sbi, buf, buf_size) == blk_crc; 1957 } 1958 1959 static inline u32 f2fs_chksum(struct f2fs_sb_info *sbi, u32 crc, 1960 const void *address, unsigned int length) 1961 { 1962 return __f2fs_crc32(sbi, crc, address, length); 1963 } 1964 1965 static inline struct f2fs_inode_info *F2FS_I(struct inode *inode) 1966 { 1967 return container_of(inode, struct f2fs_inode_info, vfs_inode); 1968 } 1969 1970 static inline struct f2fs_sb_info *F2FS_SB(struct super_block *sb) 1971 { 1972 return sb->s_fs_info; 1973 } 1974 1975 static inline struct f2fs_sb_info *F2FS_I_SB(struct inode *inode) 1976 { 1977 return F2FS_SB(inode->i_sb); 1978 } 1979 1980 static inline struct f2fs_sb_info *F2FS_M_SB(struct address_space *mapping) 1981 { 1982 return F2FS_I_SB(mapping->host); 1983 } 1984 1985 static inline struct f2fs_sb_info *F2FS_P_SB(struct page *page) 1986 { 1987 return F2FS_M_SB(page_file_mapping(page)); 1988 } 1989 1990 static inline struct f2fs_super_block *F2FS_RAW_SUPER(struct f2fs_sb_info *sbi) 1991 { 1992 return (struct f2fs_super_block *)(sbi->raw_super); 1993 } 1994 1995 static inline struct f2fs_checkpoint *F2FS_CKPT(struct f2fs_sb_info *sbi) 1996 { 1997 return (struct f2fs_checkpoint *)(sbi->ckpt); 1998 } 1999 2000 static inline struct f2fs_node *F2FS_NODE(struct page *page) 2001 { 2002 return (struct f2fs_node *)page_address(page); 2003 } 2004 2005 static inline struct f2fs_inode *F2FS_INODE(struct page *page) 2006 { 2007 return &((struct f2fs_node *)page_address(page))->i; 2008 } 2009 2010 static inline struct f2fs_nm_info *NM_I(struct f2fs_sb_info *sbi) 2011 { 2012 return (struct f2fs_nm_info *)(sbi->nm_info); 2013 } 2014 2015 static inline struct f2fs_sm_info *SM_I(struct f2fs_sb_info *sbi) 2016 { 2017 return (struct f2fs_sm_info *)(sbi->sm_info); 2018 } 2019 2020 static inline struct sit_info *SIT_I(struct f2fs_sb_info *sbi) 2021 { 2022 return (struct sit_info *)(SM_I(sbi)->sit_info); 2023 } 2024 2025 static inline struct free_segmap_info *FREE_I(struct f2fs_sb_info *sbi) 2026 { 2027 return (struct free_segmap_info *)(SM_I(sbi)->free_info); 2028 } 2029 2030 static inline struct dirty_seglist_info *DIRTY_I(struct f2fs_sb_info *sbi) 2031 { 2032 return (struct dirty_seglist_info *)(SM_I(sbi)->dirty_info); 2033 } 2034 2035 static inline struct address_space *META_MAPPING(struct f2fs_sb_info *sbi) 2036 { 2037 return sbi->meta_inode->i_mapping; 2038 } 2039 2040 static inline struct address_space *NODE_MAPPING(struct f2fs_sb_info *sbi) 2041 { 2042 return sbi->node_inode->i_mapping; 2043 } 2044 2045 static inline bool is_sbi_flag_set(struct f2fs_sb_info *sbi, unsigned int type) 2046 { 2047 return test_bit(type, &sbi->s_flag); 2048 } 2049 2050 static inline void set_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type) 2051 { 2052 set_bit(type, &sbi->s_flag); 2053 } 2054 2055 static inline void clear_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type) 2056 { 2057 clear_bit(type, &sbi->s_flag); 2058 } 2059 2060 static inline unsigned long long cur_cp_version(struct f2fs_checkpoint *cp) 2061 { 2062 return le64_to_cpu(cp->checkpoint_ver); 2063 } 2064 2065 static inline unsigned long f2fs_qf_ino(struct super_block *sb, int type) 2066 { 2067 if (type < F2FS_MAX_QUOTAS) 2068 return le32_to_cpu(F2FS_SB(sb)->raw_super->qf_ino[type]); 2069 return 0; 2070 } 2071 2072 static inline __u64 cur_cp_crc(struct f2fs_checkpoint *cp) 2073 { 2074 size_t crc_offset = le32_to_cpu(cp->checksum_offset); 2075 return le32_to_cpu(*((__le32 *)((unsigned char *)cp + crc_offset))); 2076 } 2077 2078 static inline bool __is_set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 2079 { 2080 unsigned int ckpt_flags = le32_to_cpu(cp->ckpt_flags); 2081 2082 return ckpt_flags & f; 2083 } 2084 2085 static inline bool is_set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 2086 { 2087 return __is_set_ckpt_flags(F2FS_CKPT(sbi), f); 2088 } 2089 2090 static inline void __set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 2091 { 2092 unsigned int ckpt_flags; 2093 2094 ckpt_flags = le32_to_cpu(cp->ckpt_flags); 2095 ckpt_flags |= f; 2096 cp->ckpt_flags = cpu_to_le32(ckpt_flags); 2097 } 2098 2099 static inline void set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 2100 { 2101 unsigned long flags; 2102 2103 spin_lock_irqsave(&sbi->cp_lock, flags); 2104 __set_ckpt_flags(F2FS_CKPT(sbi), f); 2105 spin_unlock_irqrestore(&sbi->cp_lock, flags); 2106 } 2107 2108 static inline void __clear_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 2109 { 2110 unsigned int ckpt_flags; 2111 2112 ckpt_flags = le32_to_cpu(cp->ckpt_flags); 2113 ckpt_flags &= (~f); 2114 cp->ckpt_flags = cpu_to_le32(ckpt_flags); 2115 } 2116 2117 static inline void clear_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 2118 { 2119 unsigned long flags; 2120 2121 spin_lock_irqsave(&sbi->cp_lock, flags); 2122 __clear_ckpt_flags(F2FS_CKPT(sbi), f); 2123 spin_unlock_irqrestore(&sbi->cp_lock, flags); 2124 } 2125 2126 #define init_f2fs_rwsem(sem) \ 2127 do { \ 2128 static struct lock_class_key __key; \ 2129 \ 2130 __init_f2fs_rwsem((sem), #sem, &__key); \ 2131 } while (0) 2132 2133 static inline void __init_f2fs_rwsem(struct f2fs_rwsem *sem, 2134 const char *sem_name, struct lock_class_key *key) 2135 { 2136 __init_rwsem(&sem->internal_rwsem, sem_name, key); 2137 #ifdef CONFIG_F2FS_UNFAIR_RWSEM 2138 init_waitqueue_head(&sem->read_waiters); 2139 #endif 2140 } 2141 2142 static inline int f2fs_rwsem_is_locked(struct f2fs_rwsem *sem) 2143 { 2144 return rwsem_is_locked(&sem->internal_rwsem); 2145 } 2146 2147 static inline int f2fs_rwsem_is_contended(struct f2fs_rwsem *sem) 2148 { 2149 return rwsem_is_contended(&sem->internal_rwsem); 2150 } 2151 2152 static inline void f2fs_down_read(struct f2fs_rwsem *sem) 2153 { 2154 #ifdef CONFIG_F2FS_UNFAIR_RWSEM 2155 wait_event(sem->read_waiters, down_read_trylock(&sem->internal_rwsem)); 2156 #else 2157 down_read(&sem->internal_rwsem); 2158 #endif 2159 } 2160 2161 static inline int f2fs_down_read_trylock(struct f2fs_rwsem *sem) 2162 { 2163 return down_read_trylock(&sem->internal_rwsem); 2164 } 2165 2166 #ifdef CONFIG_DEBUG_LOCK_ALLOC 2167 static inline void f2fs_down_read_nested(struct f2fs_rwsem *sem, int subclass) 2168 { 2169 down_read_nested(&sem->internal_rwsem, subclass); 2170 } 2171 #else 2172 #define f2fs_down_read_nested(sem, subclass) f2fs_down_read(sem) 2173 #endif 2174 2175 static inline void f2fs_up_read(struct f2fs_rwsem *sem) 2176 { 2177 up_read(&sem->internal_rwsem); 2178 } 2179 2180 static inline void f2fs_down_write(struct f2fs_rwsem *sem) 2181 { 2182 down_write(&sem->internal_rwsem); 2183 } 2184 2185 static inline int f2fs_down_write_trylock(struct f2fs_rwsem *sem) 2186 { 2187 return down_write_trylock(&sem->internal_rwsem); 2188 } 2189 2190 static inline void f2fs_up_write(struct f2fs_rwsem *sem) 2191 { 2192 up_write(&sem->internal_rwsem); 2193 #ifdef CONFIG_F2FS_UNFAIR_RWSEM 2194 wake_up_all(&sem->read_waiters); 2195 #endif 2196 } 2197 2198 static inline void f2fs_lock_op(struct f2fs_sb_info *sbi) 2199 { 2200 f2fs_down_read(&sbi->cp_rwsem); 2201 } 2202 2203 static inline int f2fs_trylock_op(struct f2fs_sb_info *sbi) 2204 { 2205 if (time_to_inject(sbi, FAULT_LOCK_OP)) { 2206 f2fs_show_injection_info(sbi, FAULT_LOCK_OP); 2207 return 0; 2208 } 2209 return f2fs_down_read_trylock(&sbi->cp_rwsem); 2210 } 2211 2212 static inline void f2fs_unlock_op(struct f2fs_sb_info *sbi) 2213 { 2214 f2fs_up_read(&sbi->cp_rwsem); 2215 } 2216 2217 static inline void f2fs_lock_all(struct f2fs_sb_info *sbi) 2218 { 2219 f2fs_down_write(&sbi->cp_rwsem); 2220 } 2221 2222 static inline void f2fs_unlock_all(struct f2fs_sb_info *sbi) 2223 { 2224 f2fs_up_write(&sbi->cp_rwsem); 2225 } 2226 2227 static inline int __get_cp_reason(struct f2fs_sb_info *sbi) 2228 { 2229 int reason = CP_SYNC; 2230 2231 if (test_opt(sbi, FASTBOOT)) 2232 reason = CP_FASTBOOT; 2233 if (is_sbi_flag_set(sbi, SBI_IS_CLOSE)) 2234 reason = CP_UMOUNT; 2235 return reason; 2236 } 2237 2238 static inline bool __remain_node_summaries(int reason) 2239 { 2240 return (reason & (CP_UMOUNT | CP_FASTBOOT)); 2241 } 2242 2243 static inline bool __exist_node_summaries(struct f2fs_sb_info *sbi) 2244 { 2245 return (is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG) || 2246 is_set_ckpt_flags(sbi, CP_FASTBOOT_FLAG)); 2247 } 2248 2249 /* 2250 * Check whether the inode has blocks or not 2251 */ 2252 static inline int F2FS_HAS_BLOCKS(struct inode *inode) 2253 { 2254 block_t xattr_block = F2FS_I(inode)->i_xattr_nid ? 1 : 0; 2255 2256 return (inode->i_blocks >> F2FS_LOG_SECTORS_PER_BLOCK) > xattr_block; 2257 } 2258 2259 static inline bool f2fs_has_xattr_block(unsigned int ofs) 2260 { 2261 return ofs == XATTR_NODE_OFFSET; 2262 } 2263 2264 static inline bool __allow_reserved_blocks(struct f2fs_sb_info *sbi, 2265 struct inode *inode, bool cap) 2266 { 2267 if (!inode) 2268 return true; 2269 if (!test_opt(sbi, RESERVE_ROOT)) 2270 return false; 2271 if (IS_NOQUOTA(inode)) 2272 return true; 2273 if (uid_eq(F2FS_OPTION(sbi).s_resuid, current_fsuid())) 2274 return true; 2275 if (!gid_eq(F2FS_OPTION(sbi).s_resgid, GLOBAL_ROOT_GID) && 2276 in_group_p(F2FS_OPTION(sbi).s_resgid)) 2277 return true; 2278 if (cap && capable(CAP_SYS_RESOURCE)) 2279 return true; 2280 return false; 2281 } 2282 2283 static inline void f2fs_i_blocks_write(struct inode *, block_t, bool, bool); 2284 static inline int inc_valid_block_count(struct f2fs_sb_info *sbi, 2285 struct inode *inode, blkcnt_t *count) 2286 { 2287 blkcnt_t diff = 0, release = 0; 2288 block_t avail_user_block_count; 2289 int ret; 2290 2291 ret = dquot_reserve_block(inode, *count); 2292 if (ret) 2293 return ret; 2294 2295 if (time_to_inject(sbi, FAULT_BLOCK)) { 2296 f2fs_show_injection_info(sbi, FAULT_BLOCK); 2297 release = *count; 2298 goto release_quota; 2299 } 2300 2301 /* 2302 * let's increase this in prior to actual block count change in order 2303 * for f2fs_sync_file to avoid data races when deciding checkpoint. 2304 */ 2305 percpu_counter_add(&sbi->alloc_valid_block_count, (*count)); 2306 2307 spin_lock(&sbi->stat_lock); 2308 sbi->total_valid_block_count += (block_t)(*count); 2309 avail_user_block_count = sbi->user_block_count - 2310 sbi->current_reserved_blocks; 2311 2312 if (!__allow_reserved_blocks(sbi, inode, true)) 2313 avail_user_block_count -= F2FS_OPTION(sbi).root_reserved_blocks; 2314 2315 if (F2FS_IO_ALIGNED(sbi)) 2316 avail_user_block_count -= sbi->blocks_per_seg * 2317 SM_I(sbi)->additional_reserved_segments; 2318 2319 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) { 2320 if (avail_user_block_count > sbi->unusable_block_count) 2321 avail_user_block_count -= sbi->unusable_block_count; 2322 else 2323 avail_user_block_count = 0; 2324 } 2325 if (unlikely(sbi->total_valid_block_count > avail_user_block_count)) { 2326 diff = sbi->total_valid_block_count - avail_user_block_count; 2327 if (diff > *count) 2328 diff = *count; 2329 *count -= diff; 2330 release = diff; 2331 sbi->total_valid_block_count -= diff; 2332 if (!*count) { 2333 spin_unlock(&sbi->stat_lock); 2334 goto enospc; 2335 } 2336 } 2337 spin_unlock(&sbi->stat_lock); 2338 2339 if (unlikely(release)) { 2340 percpu_counter_sub(&sbi->alloc_valid_block_count, release); 2341 dquot_release_reservation_block(inode, release); 2342 } 2343 f2fs_i_blocks_write(inode, *count, true, true); 2344 return 0; 2345 2346 enospc: 2347 percpu_counter_sub(&sbi->alloc_valid_block_count, release); 2348 release_quota: 2349 dquot_release_reservation_block(inode, release); 2350 return -ENOSPC; 2351 } 2352 2353 __printf(2, 3) 2354 void f2fs_printk(struct f2fs_sb_info *sbi, const char *fmt, ...); 2355 2356 #define f2fs_err(sbi, fmt, ...) \ 2357 f2fs_printk(sbi, KERN_ERR fmt, ##__VA_ARGS__) 2358 #define f2fs_warn(sbi, fmt, ...) \ 2359 f2fs_printk(sbi, KERN_WARNING fmt, ##__VA_ARGS__) 2360 #define f2fs_notice(sbi, fmt, ...) \ 2361 f2fs_printk(sbi, KERN_NOTICE fmt, ##__VA_ARGS__) 2362 #define f2fs_info(sbi, fmt, ...) \ 2363 f2fs_printk(sbi, KERN_INFO fmt, ##__VA_ARGS__) 2364 #define f2fs_debug(sbi, fmt, ...) \ 2365 f2fs_printk(sbi, KERN_DEBUG fmt, ##__VA_ARGS__) 2366 2367 static inline void dec_valid_block_count(struct f2fs_sb_info *sbi, 2368 struct inode *inode, 2369 block_t count) 2370 { 2371 blkcnt_t sectors = count << F2FS_LOG_SECTORS_PER_BLOCK; 2372 2373 spin_lock(&sbi->stat_lock); 2374 f2fs_bug_on(sbi, sbi->total_valid_block_count < (block_t) count); 2375 sbi->total_valid_block_count -= (block_t)count; 2376 if (sbi->reserved_blocks && 2377 sbi->current_reserved_blocks < sbi->reserved_blocks) 2378 sbi->current_reserved_blocks = min(sbi->reserved_blocks, 2379 sbi->current_reserved_blocks + count); 2380 spin_unlock(&sbi->stat_lock); 2381 if (unlikely(inode->i_blocks < sectors)) { 2382 f2fs_warn(sbi, "Inconsistent i_blocks, ino:%lu, iblocks:%llu, sectors:%llu", 2383 inode->i_ino, 2384 (unsigned long long)inode->i_blocks, 2385 (unsigned long long)sectors); 2386 set_sbi_flag(sbi, SBI_NEED_FSCK); 2387 return; 2388 } 2389 f2fs_i_blocks_write(inode, count, false, true); 2390 } 2391 2392 static inline void inc_page_count(struct f2fs_sb_info *sbi, int count_type) 2393 { 2394 atomic_inc(&sbi->nr_pages[count_type]); 2395 2396 if (count_type == F2FS_DIRTY_DENTS || 2397 count_type == F2FS_DIRTY_NODES || 2398 count_type == F2FS_DIRTY_META || 2399 count_type == F2FS_DIRTY_QDATA || 2400 count_type == F2FS_DIRTY_IMETA) 2401 set_sbi_flag(sbi, SBI_IS_DIRTY); 2402 } 2403 2404 static inline void inode_inc_dirty_pages(struct inode *inode) 2405 { 2406 atomic_inc(&F2FS_I(inode)->dirty_pages); 2407 inc_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ? 2408 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA); 2409 if (IS_NOQUOTA(inode)) 2410 inc_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA); 2411 } 2412 2413 static inline void dec_page_count(struct f2fs_sb_info *sbi, int count_type) 2414 { 2415 atomic_dec(&sbi->nr_pages[count_type]); 2416 } 2417 2418 static inline void inode_dec_dirty_pages(struct inode *inode) 2419 { 2420 if (!S_ISDIR(inode->i_mode) && !S_ISREG(inode->i_mode) && 2421 !S_ISLNK(inode->i_mode)) 2422 return; 2423 2424 atomic_dec(&F2FS_I(inode)->dirty_pages); 2425 dec_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ? 2426 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA); 2427 if (IS_NOQUOTA(inode)) 2428 dec_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA); 2429 } 2430 2431 static inline s64 get_pages(struct f2fs_sb_info *sbi, int count_type) 2432 { 2433 return atomic_read(&sbi->nr_pages[count_type]); 2434 } 2435 2436 static inline int get_dirty_pages(struct inode *inode) 2437 { 2438 return atomic_read(&F2FS_I(inode)->dirty_pages); 2439 } 2440 2441 static inline int get_blocktype_secs(struct f2fs_sb_info *sbi, int block_type) 2442 { 2443 unsigned int pages_per_sec = sbi->segs_per_sec * sbi->blocks_per_seg; 2444 unsigned int segs = (get_pages(sbi, block_type) + pages_per_sec - 1) >> 2445 sbi->log_blocks_per_seg; 2446 2447 return segs / sbi->segs_per_sec; 2448 } 2449 2450 static inline block_t valid_user_blocks(struct f2fs_sb_info *sbi) 2451 { 2452 return sbi->total_valid_block_count; 2453 } 2454 2455 static inline block_t discard_blocks(struct f2fs_sb_info *sbi) 2456 { 2457 return sbi->discard_blks; 2458 } 2459 2460 static inline unsigned long __bitmap_size(struct f2fs_sb_info *sbi, int flag) 2461 { 2462 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi); 2463 2464 /* return NAT or SIT bitmap */ 2465 if (flag == NAT_BITMAP) 2466 return le32_to_cpu(ckpt->nat_ver_bitmap_bytesize); 2467 else if (flag == SIT_BITMAP) 2468 return le32_to_cpu(ckpt->sit_ver_bitmap_bytesize); 2469 2470 return 0; 2471 } 2472 2473 static inline block_t __cp_payload(struct f2fs_sb_info *sbi) 2474 { 2475 return le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_payload); 2476 } 2477 2478 static inline void *__bitmap_ptr(struct f2fs_sb_info *sbi, int flag) 2479 { 2480 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi); 2481 void *tmp_ptr = &ckpt->sit_nat_version_bitmap; 2482 int offset; 2483 2484 if (is_set_ckpt_flags(sbi, CP_LARGE_NAT_BITMAP_FLAG)) { 2485 offset = (flag == SIT_BITMAP) ? 2486 le32_to_cpu(ckpt->nat_ver_bitmap_bytesize) : 0; 2487 /* 2488 * if large_nat_bitmap feature is enabled, leave checksum 2489 * protection for all nat/sit bitmaps. 2490 */ 2491 return tmp_ptr + offset + sizeof(__le32); 2492 } 2493 2494 if (__cp_payload(sbi) > 0) { 2495 if (flag == NAT_BITMAP) 2496 return &ckpt->sit_nat_version_bitmap; 2497 else 2498 return (unsigned char *)ckpt + F2FS_BLKSIZE; 2499 } else { 2500 offset = (flag == NAT_BITMAP) ? 2501 le32_to_cpu(ckpt->sit_ver_bitmap_bytesize) : 0; 2502 return tmp_ptr + offset; 2503 } 2504 } 2505 2506 static inline block_t __start_cp_addr(struct f2fs_sb_info *sbi) 2507 { 2508 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr); 2509 2510 if (sbi->cur_cp_pack == 2) 2511 start_addr += sbi->blocks_per_seg; 2512 return start_addr; 2513 } 2514 2515 static inline block_t __start_cp_next_addr(struct f2fs_sb_info *sbi) 2516 { 2517 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr); 2518 2519 if (sbi->cur_cp_pack == 1) 2520 start_addr += sbi->blocks_per_seg; 2521 return start_addr; 2522 } 2523 2524 static inline void __set_cp_next_pack(struct f2fs_sb_info *sbi) 2525 { 2526 sbi->cur_cp_pack = (sbi->cur_cp_pack == 1) ? 2 : 1; 2527 } 2528 2529 static inline block_t __start_sum_addr(struct f2fs_sb_info *sbi) 2530 { 2531 return le32_to_cpu(F2FS_CKPT(sbi)->cp_pack_start_sum); 2532 } 2533 2534 static inline int inc_valid_node_count(struct f2fs_sb_info *sbi, 2535 struct inode *inode, bool is_inode) 2536 { 2537 block_t valid_block_count; 2538 unsigned int valid_node_count, user_block_count; 2539 int err; 2540 2541 if (is_inode) { 2542 if (inode) { 2543 err = dquot_alloc_inode(inode); 2544 if (err) 2545 return err; 2546 } 2547 } else { 2548 err = dquot_reserve_block(inode, 1); 2549 if (err) 2550 return err; 2551 } 2552 2553 if (time_to_inject(sbi, FAULT_BLOCK)) { 2554 f2fs_show_injection_info(sbi, FAULT_BLOCK); 2555 goto enospc; 2556 } 2557 2558 spin_lock(&sbi->stat_lock); 2559 2560 valid_block_count = sbi->total_valid_block_count + 2561 sbi->current_reserved_blocks + 1; 2562 2563 if (!__allow_reserved_blocks(sbi, inode, false)) 2564 valid_block_count += F2FS_OPTION(sbi).root_reserved_blocks; 2565 2566 if (F2FS_IO_ALIGNED(sbi)) 2567 valid_block_count += sbi->blocks_per_seg * 2568 SM_I(sbi)->additional_reserved_segments; 2569 2570 user_block_count = sbi->user_block_count; 2571 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) 2572 user_block_count -= sbi->unusable_block_count; 2573 2574 if (unlikely(valid_block_count > user_block_count)) { 2575 spin_unlock(&sbi->stat_lock); 2576 goto enospc; 2577 } 2578 2579 valid_node_count = sbi->total_valid_node_count + 1; 2580 if (unlikely(valid_node_count > sbi->total_node_count)) { 2581 spin_unlock(&sbi->stat_lock); 2582 goto enospc; 2583 } 2584 2585 sbi->total_valid_node_count++; 2586 sbi->total_valid_block_count++; 2587 spin_unlock(&sbi->stat_lock); 2588 2589 if (inode) { 2590 if (is_inode) 2591 f2fs_mark_inode_dirty_sync(inode, true); 2592 else 2593 f2fs_i_blocks_write(inode, 1, true, true); 2594 } 2595 2596 percpu_counter_inc(&sbi->alloc_valid_block_count); 2597 return 0; 2598 2599 enospc: 2600 if (is_inode) { 2601 if (inode) 2602 dquot_free_inode(inode); 2603 } else { 2604 dquot_release_reservation_block(inode, 1); 2605 } 2606 return -ENOSPC; 2607 } 2608 2609 static inline void dec_valid_node_count(struct f2fs_sb_info *sbi, 2610 struct inode *inode, bool is_inode) 2611 { 2612 spin_lock(&sbi->stat_lock); 2613 2614 f2fs_bug_on(sbi, !sbi->total_valid_block_count); 2615 f2fs_bug_on(sbi, !sbi->total_valid_node_count); 2616 2617 sbi->total_valid_node_count--; 2618 sbi->total_valid_block_count--; 2619 if (sbi->reserved_blocks && 2620 sbi->current_reserved_blocks < sbi->reserved_blocks) 2621 sbi->current_reserved_blocks++; 2622 2623 spin_unlock(&sbi->stat_lock); 2624 2625 if (is_inode) { 2626 dquot_free_inode(inode); 2627 } else { 2628 if (unlikely(inode->i_blocks == 0)) { 2629 f2fs_warn(sbi, "dec_valid_node_count: inconsistent i_blocks, ino:%lu, iblocks:%llu", 2630 inode->i_ino, 2631 (unsigned long long)inode->i_blocks); 2632 set_sbi_flag(sbi, SBI_NEED_FSCK); 2633 return; 2634 } 2635 f2fs_i_blocks_write(inode, 1, false, true); 2636 } 2637 } 2638 2639 static inline unsigned int valid_node_count(struct f2fs_sb_info *sbi) 2640 { 2641 return sbi->total_valid_node_count; 2642 } 2643 2644 static inline void inc_valid_inode_count(struct f2fs_sb_info *sbi) 2645 { 2646 percpu_counter_inc(&sbi->total_valid_inode_count); 2647 } 2648 2649 static inline void dec_valid_inode_count(struct f2fs_sb_info *sbi) 2650 { 2651 percpu_counter_dec(&sbi->total_valid_inode_count); 2652 } 2653 2654 static inline s64 valid_inode_count(struct f2fs_sb_info *sbi) 2655 { 2656 return percpu_counter_sum_positive(&sbi->total_valid_inode_count); 2657 } 2658 2659 static inline struct page *f2fs_grab_cache_page(struct address_space *mapping, 2660 pgoff_t index, bool for_write) 2661 { 2662 struct page *page; 2663 2664 if (IS_ENABLED(CONFIG_F2FS_FAULT_INJECTION)) { 2665 if (!for_write) 2666 page = find_get_page_flags(mapping, index, 2667 FGP_LOCK | FGP_ACCESSED); 2668 else 2669 page = find_lock_page(mapping, index); 2670 if (page) 2671 return page; 2672 2673 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_ALLOC)) { 2674 f2fs_show_injection_info(F2FS_M_SB(mapping), 2675 FAULT_PAGE_ALLOC); 2676 return NULL; 2677 } 2678 } 2679 2680 if (!for_write) 2681 return grab_cache_page(mapping, index); 2682 return grab_cache_page_write_begin(mapping, index, AOP_FLAG_NOFS); 2683 } 2684 2685 static inline struct page *f2fs_pagecache_get_page( 2686 struct address_space *mapping, pgoff_t index, 2687 int fgp_flags, gfp_t gfp_mask) 2688 { 2689 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_GET)) { 2690 f2fs_show_injection_info(F2FS_M_SB(mapping), FAULT_PAGE_GET); 2691 return NULL; 2692 } 2693 2694 return pagecache_get_page(mapping, index, fgp_flags, gfp_mask); 2695 } 2696 2697 static inline void f2fs_copy_page(struct page *src, struct page *dst) 2698 { 2699 char *src_kaddr = kmap(src); 2700 char *dst_kaddr = kmap(dst); 2701 2702 memcpy(dst_kaddr, src_kaddr, PAGE_SIZE); 2703 kunmap(dst); 2704 kunmap(src); 2705 } 2706 2707 static inline void f2fs_put_page(struct page *page, int unlock) 2708 { 2709 if (!page) 2710 return; 2711 2712 if (unlock) { 2713 f2fs_bug_on(F2FS_P_SB(page), !PageLocked(page)); 2714 unlock_page(page); 2715 } 2716 put_page(page); 2717 } 2718 2719 static inline void f2fs_put_dnode(struct dnode_of_data *dn) 2720 { 2721 if (dn->node_page) 2722 f2fs_put_page(dn->node_page, 1); 2723 if (dn->inode_page && dn->node_page != dn->inode_page) 2724 f2fs_put_page(dn->inode_page, 0); 2725 dn->node_page = NULL; 2726 dn->inode_page = NULL; 2727 } 2728 2729 static inline struct kmem_cache *f2fs_kmem_cache_create(const char *name, 2730 size_t size) 2731 { 2732 return kmem_cache_create(name, size, 0, SLAB_RECLAIM_ACCOUNT, NULL); 2733 } 2734 2735 static inline void *f2fs_kmem_cache_alloc_nofail(struct kmem_cache *cachep, 2736 gfp_t flags) 2737 { 2738 void *entry; 2739 2740 entry = kmem_cache_alloc(cachep, flags); 2741 if (!entry) 2742 entry = kmem_cache_alloc(cachep, flags | __GFP_NOFAIL); 2743 return entry; 2744 } 2745 2746 static inline void *f2fs_kmem_cache_alloc(struct kmem_cache *cachep, 2747 gfp_t flags, bool nofail, struct f2fs_sb_info *sbi) 2748 { 2749 if (nofail) 2750 return f2fs_kmem_cache_alloc_nofail(cachep, flags); 2751 2752 if (time_to_inject(sbi, FAULT_SLAB_ALLOC)) { 2753 f2fs_show_injection_info(sbi, FAULT_SLAB_ALLOC); 2754 return NULL; 2755 } 2756 2757 return kmem_cache_alloc(cachep, flags); 2758 } 2759 2760 static inline bool is_inflight_io(struct f2fs_sb_info *sbi, int type) 2761 { 2762 if (get_pages(sbi, F2FS_RD_DATA) || get_pages(sbi, F2FS_RD_NODE) || 2763 get_pages(sbi, F2FS_RD_META) || get_pages(sbi, F2FS_WB_DATA) || 2764 get_pages(sbi, F2FS_WB_CP_DATA) || 2765 get_pages(sbi, F2FS_DIO_READ) || 2766 get_pages(sbi, F2FS_DIO_WRITE)) 2767 return true; 2768 2769 if (type != DISCARD_TIME && SM_I(sbi) && SM_I(sbi)->dcc_info && 2770 atomic_read(&SM_I(sbi)->dcc_info->queued_discard)) 2771 return true; 2772 2773 if (SM_I(sbi) && SM_I(sbi)->fcc_info && 2774 atomic_read(&SM_I(sbi)->fcc_info->queued_flush)) 2775 return true; 2776 return false; 2777 } 2778 2779 static inline bool is_idle(struct f2fs_sb_info *sbi, int type) 2780 { 2781 if (sbi->gc_mode == GC_URGENT_HIGH) 2782 return true; 2783 2784 if (is_inflight_io(sbi, type)) 2785 return false; 2786 2787 if (sbi->gc_mode == GC_URGENT_LOW && 2788 (type == DISCARD_TIME || type == GC_TIME)) 2789 return true; 2790 2791 return f2fs_time_over(sbi, type); 2792 } 2793 2794 static inline void f2fs_radix_tree_insert(struct radix_tree_root *root, 2795 unsigned long index, void *item) 2796 { 2797 while (radix_tree_insert(root, index, item)) 2798 cond_resched(); 2799 } 2800 2801 #define RAW_IS_INODE(p) ((p)->footer.nid == (p)->footer.ino) 2802 2803 static inline bool IS_INODE(struct page *page) 2804 { 2805 struct f2fs_node *p = F2FS_NODE(page); 2806 2807 return RAW_IS_INODE(p); 2808 } 2809 2810 static inline int offset_in_addr(struct f2fs_inode *i) 2811 { 2812 return (i->i_inline & F2FS_EXTRA_ATTR) ? 2813 (le16_to_cpu(i->i_extra_isize) / sizeof(__le32)) : 0; 2814 } 2815 2816 static inline __le32 *blkaddr_in_node(struct f2fs_node *node) 2817 { 2818 return RAW_IS_INODE(node) ? node->i.i_addr : node->dn.addr; 2819 } 2820 2821 static inline int f2fs_has_extra_attr(struct inode *inode); 2822 static inline block_t data_blkaddr(struct inode *inode, 2823 struct page *node_page, unsigned int offset) 2824 { 2825 struct f2fs_node *raw_node; 2826 __le32 *addr_array; 2827 int base = 0; 2828 bool is_inode = IS_INODE(node_page); 2829 2830 raw_node = F2FS_NODE(node_page); 2831 2832 if (is_inode) { 2833 if (!inode) 2834 /* from GC path only */ 2835 base = offset_in_addr(&raw_node->i); 2836 else if (f2fs_has_extra_attr(inode)) 2837 base = get_extra_isize(inode); 2838 } 2839 2840 addr_array = blkaddr_in_node(raw_node); 2841 return le32_to_cpu(addr_array[base + offset]); 2842 } 2843 2844 static inline block_t f2fs_data_blkaddr(struct dnode_of_data *dn) 2845 { 2846 return data_blkaddr(dn->inode, dn->node_page, dn->ofs_in_node); 2847 } 2848 2849 static inline int f2fs_test_bit(unsigned int nr, char *addr) 2850 { 2851 int mask; 2852 2853 addr += (nr >> 3); 2854 mask = 1 << (7 - (nr & 0x07)); 2855 return mask & *addr; 2856 } 2857 2858 static inline void f2fs_set_bit(unsigned int nr, char *addr) 2859 { 2860 int mask; 2861 2862 addr += (nr >> 3); 2863 mask = 1 << (7 - (nr & 0x07)); 2864 *addr |= mask; 2865 } 2866 2867 static inline void f2fs_clear_bit(unsigned int nr, char *addr) 2868 { 2869 int mask; 2870 2871 addr += (nr >> 3); 2872 mask = 1 << (7 - (nr & 0x07)); 2873 *addr &= ~mask; 2874 } 2875 2876 static inline int f2fs_test_and_set_bit(unsigned int nr, char *addr) 2877 { 2878 int mask; 2879 int ret; 2880 2881 addr += (nr >> 3); 2882 mask = 1 << (7 - (nr & 0x07)); 2883 ret = mask & *addr; 2884 *addr |= mask; 2885 return ret; 2886 } 2887 2888 static inline int f2fs_test_and_clear_bit(unsigned int nr, char *addr) 2889 { 2890 int mask; 2891 int ret; 2892 2893 addr += (nr >> 3); 2894 mask = 1 << (7 - (nr & 0x07)); 2895 ret = mask & *addr; 2896 *addr &= ~mask; 2897 return ret; 2898 } 2899 2900 static inline void f2fs_change_bit(unsigned int nr, char *addr) 2901 { 2902 int mask; 2903 2904 addr += (nr >> 3); 2905 mask = 1 << (7 - (nr & 0x07)); 2906 *addr ^= mask; 2907 } 2908 2909 /* 2910 * On-disk inode flags (f2fs_inode::i_flags) 2911 */ 2912 #define F2FS_COMPR_FL 0x00000004 /* Compress file */ 2913 #define F2FS_SYNC_FL 0x00000008 /* Synchronous updates */ 2914 #define F2FS_IMMUTABLE_FL 0x00000010 /* Immutable file */ 2915 #define F2FS_APPEND_FL 0x00000020 /* writes to file may only append */ 2916 #define F2FS_NODUMP_FL 0x00000040 /* do not dump file */ 2917 #define F2FS_NOATIME_FL 0x00000080 /* do not update atime */ 2918 #define F2FS_NOCOMP_FL 0x00000400 /* Don't compress */ 2919 #define F2FS_INDEX_FL 0x00001000 /* hash-indexed directory */ 2920 #define F2FS_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */ 2921 #define F2FS_PROJINHERIT_FL 0x20000000 /* Create with parents projid */ 2922 #define F2FS_CASEFOLD_FL 0x40000000 /* Casefolded file */ 2923 2924 /* Flags that should be inherited by new inodes from their parent. */ 2925 #define F2FS_FL_INHERITED (F2FS_SYNC_FL | F2FS_NODUMP_FL | F2FS_NOATIME_FL | \ 2926 F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \ 2927 F2FS_CASEFOLD_FL | F2FS_COMPR_FL | F2FS_NOCOMP_FL) 2928 2929 /* Flags that are appropriate for regular files (all but dir-specific ones). */ 2930 #define F2FS_REG_FLMASK (~(F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \ 2931 F2FS_CASEFOLD_FL)) 2932 2933 /* Flags that are appropriate for non-directories/regular files. */ 2934 #define F2FS_OTHER_FLMASK (F2FS_NODUMP_FL | F2FS_NOATIME_FL) 2935 2936 static inline __u32 f2fs_mask_flags(umode_t mode, __u32 flags) 2937 { 2938 if (S_ISDIR(mode)) 2939 return flags; 2940 else if (S_ISREG(mode)) 2941 return flags & F2FS_REG_FLMASK; 2942 else 2943 return flags & F2FS_OTHER_FLMASK; 2944 } 2945 2946 static inline void __mark_inode_dirty_flag(struct inode *inode, 2947 int flag, bool set) 2948 { 2949 switch (flag) { 2950 case FI_INLINE_XATTR: 2951 case FI_INLINE_DATA: 2952 case FI_INLINE_DENTRY: 2953 case FI_NEW_INODE: 2954 if (set) 2955 return; 2956 fallthrough; 2957 case FI_DATA_EXIST: 2958 case FI_INLINE_DOTS: 2959 case FI_PIN_FILE: 2960 case FI_COMPRESS_RELEASED: 2961 f2fs_mark_inode_dirty_sync(inode, true); 2962 } 2963 } 2964 2965 static inline void set_inode_flag(struct inode *inode, int flag) 2966 { 2967 set_bit(flag, F2FS_I(inode)->flags); 2968 __mark_inode_dirty_flag(inode, flag, true); 2969 } 2970 2971 static inline int is_inode_flag_set(struct inode *inode, int flag) 2972 { 2973 return test_bit(flag, F2FS_I(inode)->flags); 2974 } 2975 2976 static inline void clear_inode_flag(struct inode *inode, int flag) 2977 { 2978 clear_bit(flag, F2FS_I(inode)->flags); 2979 __mark_inode_dirty_flag(inode, flag, false); 2980 } 2981 2982 static inline bool f2fs_verity_in_progress(struct inode *inode) 2983 { 2984 return IS_ENABLED(CONFIG_FS_VERITY) && 2985 is_inode_flag_set(inode, FI_VERITY_IN_PROGRESS); 2986 } 2987 2988 static inline void set_acl_inode(struct inode *inode, umode_t mode) 2989 { 2990 F2FS_I(inode)->i_acl_mode = mode; 2991 set_inode_flag(inode, FI_ACL_MODE); 2992 f2fs_mark_inode_dirty_sync(inode, false); 2993 } 2994 2995 static inline void f2fs_i_links_write(struct inode *inode, bool inc) 2996 { 2997 if (inc) 2998 inc_nlink(inode); 2999 else 3000 drop_nlink(inode); 3001 f2fs_mark_inode_dirty_sync(inode, true); 3002 } 3003 3004 static inline void f2fs_i_blocks_write(struct inode *inode, 3005 block_t diff, bool add, bool claim) 3006 { 3007 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE); 3008 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER); 3009 3010 /* add = 1, claim = 1 should be dquot_reserve_block in pair */ 3011 if (add) { 3012 if (claim) 3013 dquot_claim_block(inode, diff); 3014 else 3015 dquot_alloc_block_nofail(inode, diff); 3016 } else { 3017 dquot_free_block(inode, diff); 3018 } 3019 3020 f2fs_mark_inode_dirty_sync(inode, true); 3021 if (clean || recover) 3022 set_inode_flag(inode, FI_AUTO_RECOVER); 3023 } 3024 3025 static inline void f2fs_i_size_write(struct inode *inode, loff_t i_size) 3026 { 3027 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE); 3028 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER); 3029 3030 if (i_size_read(inode) == i_size) 3031 return; 3032 3033 i_size_write(inode, i_size); 3034 f2fs_mark_inode_dirty_sync(inode, true); 3035 if (clean || recover) 3036 set_inode_flag(inode, FI_AUTO_RECOVER); 3037 } 3038 3039 static inline void f2fs_i_depth_write(struct inode *inode, unsigned int depth) 3040 { 3041 F2FS_I(inode)->i_current_depth = depth; 3042 f2fs_mark_inode_dirty_sync(inode, true); 3043 } 3044 3045 static inline void f2fs_i_gc_failures_write(struct inode *inode, 3046 unsigned int count) 3047 { 3048 F2FS_I(inode)->i_gc_failures[GC_FAILURE_PIN] = count; 3049 f2fs_mark_inode_dirty_sync(inode, true); 3050 } 3051 3052 static inline void f2fs_i_xnid_write(struct inode *inode, nid_t xnid) 3053 { 3054 F2FS_I(inode)->i_xattr_nid = xnid; 3055 f2fs_mark_inode_dirty_sync(inode, true); 3056 } 3057 3058 static inline void f2fs_i_pino_write(struct inode *inode, nid_t pino) 3059 { 3060 F2FS_I(inode)->i_pino = pino; 3061 f2fs_mark_inode_dirty_sync(inode, true); 3062 } 3063 3064 static inline void get_inline_info(struct inode *inode, struct f2fs_inode *ri) 3065 { 3066 struct f2fs_inode_info *fi = F2FS_I(inode); 3067 3068 if (ri->i_inline & F2FS_INLINE_XATTR) 3069 set_bit(FI_INLINE_XATTR, fi->flags); 3070 if (ri->i_inline & F2FS_INLINE_DATA) 3071 set_bit(FI_INLINE_DATA, fi->flags); 3072 if (ri->i_inline & F2FS_INLINE_DENTRY) 3073 set_bit(FI_INLINE_DENTRY, fi->flags); 3074 if (ri->i_inline & F2FS_DATA_EXIST) 3075 set_bit(FI_DATA_EXIST, fi->flags); 3076 if (ri->i_inline & F2FS_INLINE_DOTS) 3077 set_bit(FI_INLINE_DOTS, fi->flags); 3078 if (ri->i_inline & F2FS_EXTRA_ATTR) 3079 set_bit(FI_EXTRA_ATTR, fi->flags); 3080 if (ri->i_inline & F2FS_PIN_FILE) 3081 set_bit(FI_PIN_FILE, fi->flags); 3082 if (ri->i_inline & F2FS_COMPRESS_RELEASED) 3083 set_bit(FI_COMPRESS_RELEASED, fi->flags); 3084 } 3085 3086 static inline void set_raw_inline(struct inode *inode, struct f2fs_inode *ri) 3087 { 3088 ri->i_inline = 0; 3089 3090 if (is_inode_flag_set(inode, FI_INLINE_XATTR)) 3091 ri->i_inline |= F2FS_INLINE_XATTR; 3092 if (is_inode_flag_set(inode, FI_INLINE_DATA)) 3093 ri->i_inline |= F2FS_INLINE_DATA; 3094 if (is_inode_flag_set(inode, FI_INLINE_DENTRY)) 3095 ri->i_inline |= F2FS_INLINE_DENTRY; 3096 if (is_inode_flag_set(inode, FI_DATA_EXIST)) 3097 ri->i_inline |= F2FS_DATA_EXIST; 3098 if (is_inode_flag_set(inode, FI_INLINE_DOTS)) 3099 ri->i_inline |= F2FS_INLINE_DOTS; 3100 if (is_inode_flag_set(inode, FI_EXTRA_ATTR)) 3101 ri->i_inline |= F2FS_EXTRA_ATTR; 3102 if (is_inode_flag_set(inode, FI_PIN_FILE)) 3103 ri->i_inline |= F2FS_PIN_FILE; 3104 if (is_inode_flag_set(inode, FI_COMPRESS_RELEASED)) 3105 ri->i_inline |= F2FS_COMPRESS_RELEASED; 3106 } 3107 3108 static inline int f2fs_has_extra_attr(struct inode *inode) 3109 { 3110 return is_inode_flag_set(inode, FI_EXTRA_ATTR); 3111 } 3112 3113 static inline int f2fs_has_inline_xattr(struct inode *inode) 3114 { 3115 return is_inode_flag_set(inode, FI_INLINE_XATTR); 3116 } 3117 3118 static inline int f2fs_compressed_file(struct inode *inode) 3119 { 3120 return S_ISREG(inode->i_mode) && 3121 is_inode_flag_set(inode, FI_COMPRESSED_FILE); 3122 } 3123 3124 static inline bool f2fs_need_compress_data(struct inode *inode) 3125 { 3126 int compress_mode = F2FS_OPTION(F2FS_I_SB(inode)).compress_mode; 3127 3128 if (!f2fs_compressed_file(inode)) 3129 return false; 3130 3131 if (compress_mode == COMPR_MODE_FS) 3132 return true; 3133 else if (compress_mode == COMPR_MODE_USER && 3134 is_inode_flag_set(inode, FI_ENABLE_COMPRESS)) 3135 return true; 3136 3137 return false; 3138 } 3139 3140 static inline unsigned int addrs_per_inode(struct inode *inode) 3141 { 3142 unsigned int addrs = CUR_ADDRS_PER_INODE(inode) - 3143 get_inline_xattr_addrs(inode); 3144 3145 if (!f2fs_compressed_file(inode)) 3146 return addrs; 3147 return ALIGN_DOWN(addrs, F2FS_I(inode)->i_cluster_size); 3148 } 3149 3150 static inline unsigned int addrs_per_block(struct inode *inode) 3151 { 3152 if (!f2fs_compressed_file(inode)) 3153 return DEF_ADDRS_PER_BLOCK; 3154 return ALIGN_DOWN(DEF_ADDRS_PER_BLOCK, F2FS_I(inode)->i_cluster_size); 3155 } 3156 3157 static inline void *inline_xattr_addr(struct inode *inode, struct page *page) 3158 { 3159 struct f2fs_inode *ri = F2FS_INODE(page); 3160 3161 return (void *)&(ri->i_addr[DEF_ADDRS_PER_INODE - 3162 get_inline_xattr_addrs(inode)]); 3163 } 3164 3165 static inline int inline_xattr_size(struct inode *inode) 3166 { 3167 if (f2fs_has_inline_xattr(inode)) 3168 return get_inline_xattr_addrs(inode) * sizeof(__le32); 3169 return 0; 3170 } 3171 3172 static inline int f2fs_has_inline_data(struct inode *inode) 3173 { 3174 return is_inode_flag_set(inode, FI_INLINE_DATA); 3175 } 3176 3177 static inline int f2fs_exist_data(struct inode *inode) 3178 { 3179 return is_inode_flag_set(inode, FI_DATA_EXIST); 3180 } 3181 3182 static inline int f2fs_has_inline_dots(struct inode *inode) 3183 { 3184 return is_inode_flag_set(inode, FI_INLINE_DOTS); 3185 } 3186 3187 static inline int f2fs_is_mmap_file(struct inode *inode) 3188 { 3189 return is_inode_flag_set(inode, FI_MMAP_FILE); 3190 } 3191 3192 static inline bool f2fs_is_pinned_file(struct inode *inode) 3193 { 3194 return is_inode_flag_set(inode, FI_PIN_FILE); 3195 } 3196 3197 static inline bool f2fs_is_atomic_file(struct inode *inode) 3198 { 3199 return is_inode_flag_set(inode, FI_ATOMIC_FILE); 3200 } 3201 3202 static inline bool f2fs_is_commit_atomic_write(struct inode *inode) 3203 { 3204 return is_inode_flag_set(inode, FI_ATOMIC_COMMIT); 3205 } 3206 3207 static inline bool f2fs_is_volatile_file(struct inode *inode) 3208 { 3209 return is_inode_flag_set(inode, FI_VOLATILE_FILE); 3210 } 3211 3212 static inline bool f2fs_is_first_block_written(struct inode *inode) 3213 { 3214 return is_inode_flag_set(inode, FI_FIRST_BLOCK_WRITTEN); 3215 } 3216 3217 static inline bool f2fs_is_drop_cache(struct inode *inode) 3218 { 3219 return is_inode_flag_set(inode, FI_DROP_CACHE); 3220 } 3221 3222 static inline void *inline_data_addr(struct inode *inode, struct page *page) 3223 { 3224 struct f2fs_inode *ri = F2FS_INODE(page); 3225 int extra_size = get_extra_isize(inode); 3226 3227 return (void *)&(ri->i_addr[extra_size + DEF_INLINE_RESERVED_SIZE]); 3228 } 3229 3230 static inline int f2fs_has_inline_dentry(struct inode *inode) 3231 { 3232 return is_inode_flag_set(inode, FI_INLINE_DENTRY); 3233 } 3234 3235 static inline int is_file(struct inode *inode, int type) 3236 { 3237 return F2FS_I(inode)->i_advise & type; 3238 } 3239 3240 static inline void set_file(struct inode *inode, int type) 3241 { 3242 if (is_file(inode, type)) 3243 return; 3244 F2FS_I(inode)->i_advise |= type; 3245 f2fs_mark_inode_dirty_sync(inode, true); 3246 } 3247 3248 static inline void clear_file(struct inode *inode, int type) 3249 { 3250 if (!is_file(inode, type)) 3251 return; 3252 F2FS_I(inode)->i_advise &= ~type; 3253 f2fs_mark_inode_dirty_sync(inode, true); 3254 } 3255 3256 static inline bool f2fs_is_time_consistent(struct inode *inode) 3257 { 3258 if (!timespec64_equal(F2FS_I(inode)->i_disk_time, &inode->i_atime)) 3259 return false; 3260 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 1, &inode->i_ctime)) 3261 return false; 3262 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 2, &inode->i_mtime)) 3263 return false; 3264 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 3, 3265 &F2FS_I(inode)->i_crtime)) 3266 return false; 3267 return true; 3268 } 3269 3270 static inline bool f2fs_skip_inode_update(struct inode *inode, int dsync) 3271 { 3272 bool ret; 3273 3274 if (dsync) { 3275 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 3276 3277 spin_lock(&sbi->inode_lock[DIRTY_META]); 3278 ret = list_empty(&F2FS_I(inode)->gdirty_list); 3279 spin_unlock(&sbi->inode_lock[DIRTY_META]); 3280 return ret; 3281 } 3282 if (!is_inode_flag_set(inode, FI_AUTO_RECOVER) || 3283 file_keep_isize(inode) || 3284 i_size_read(inode) & ~PAGE_MASK) 3285 return false; 3286 3287 if (!f2fs_is_time_consistent(inode)) 3288 return false; 3289 3290 spin_lock(&F2FS_I(inode)->i_size_lock); 3291 ret = F2FS_I(inode)->last_disk_size == i_size_read(inode); 3292 spin_unlock(&F2FS_I(inode)->i_size_lock); 3293 3294 return ret; 3295 } 3296 3297 static inline bool f2fs_readonly(struct super_block *sb) 3298 { 3299 return sb_rdonly(sb); 3300 } 3301 3302 static inline bool f2fs_cp_error(struct f2fs_sb_info *sbi) 3303 { 3304 return is_set_ckpt_flags(sbi, CP_ERROR_FLAG); 3305 } 3306 3307 static inline bool is_dot_dotdot(const u8 *name, size_t len) 3308 { 3309 if (len == 1 && name[0] == '.') 3310 return true; 3311 3312 if (len == 2 && name[0] == '.' && name[1] == '.') 3313 return true; 3314 3315 return false; 3316 } 3317 3318 static inline void *f2fs_kmalloc(struct f2fs_sb_info *sbi, 3319 size_t size, gfp_t flags) 3320 { 3321 if (time_to_inject(sbi, FAULT_KMALLOC)) { 3322 f2fs_show_injection_info(sbi, FAULT_KMALLOC); 3323 return NULL; 3324 } 3325 3326 return kmalloc(size, flags); 3327 } 3328 3329 static inline void *f2fs_kzalloc(struct f2fs_sb_info *sbi, 3330 size_t size, gfp_t flags) 3331 { 3332 return f2fs_kmalloc(sbi, size, flags | __GFP_ZERO); 3333 } 3334 3335 static inline void *f2fs_kvmalloc(struct f2fs_sb_info *sbi, 3336 size_t size, gfp_t flags) 3337 { 3338 if (time_to_inject(sbi, FAULT_KVMALLOC)) { 3339 f2fs_show_injection_info(sbi, FAULT_KVMALLOC); 3340 return NULL; 3341 } 3342 3343 return kvmalloc(size, flags); 3344 } 3345 3346 static inline void *f2fs_kvzalloc(struct f2fs_sb_info *sbi, 3347 size_t size, gfp_t flags) 3348 { 3349 return f2fs_kvmalloc(sbi, size, flags | __GFP_ZERO); 3350 } 3351 3352 static inline int get_extra_isize(struct inode *inode) 3353 { 3354 return F2FS_I(inode)->i_extra_isize / sizeof(__le32); 3355 } 3356 3357 static inline int get_inline_xattr_addrs(struct inode *inode) 3358 { 3359 return F2FS_I(inode)->i_inline_xattr_size; 3360 } 3361 3362 #define f2fs_get_inode_mode(i) \ 3363 ((is_inode_flag_set(i, FI_ACL_MODE)) ? \ 3364 (F2FS_I(i)->i_acl_mode) : ((i)->i_mode)) 3365 3366 #define F2FS_TOTAL_EXTRA_ATTR_SIZE \ 3367 (offsetof(struct f2fs_inode, i_extra_end) - \ 3368 offsetof(struct f2fs_inode, i_extra_isize)) \ 3369 3370 #define F2FS_OLD_ATTRIBUTE_SIZE (offsetof(struct f2fs_inode, i_addr)) 3371 #define F2FS_FITS_IN_INODE(f2fs_inode, extra_isize, field) \ 3372 ((offsetof(typeof(*(f2fs_inode)), field) + \ 3373 sizeof((f2fs_inode)->field)) \ 3374 <= (F2FS_OLD_ATTRIBUTE_SIZE + (extra_isize))) \ 3375 3376 #define __is_large_section(sbi) ((sbi)->segs_per_sec > 1) 3377 3378 #define __is_meta_io(fio) (PAGE_TYPE_OF_BIO((fio)->type) == META) 3379 3380 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi, 3381 block_t blkaddr, int type); 3382 static inline void verify_blkaddr(struct f2fs_sb_info *sbi, 3383 block_t blkaddr, int type) 3384 { 3385 if (!f2fs_is_valid_blkaddr(sbi, blkaddr, type)) { 3386 f2fs_err(sbi, "invalid blkaddr: %u, type: %d, run fsck to fix.", 3387 blkaddr, type); 3388 f2fs_bug_on(sbi, 1); 3389 } 3390 } 3391 3392 static inline bool __is_valid_data_blkaddr(block_t blkaddr) 3393 { 3394 if (blkaddr == NEW_ADDR || blkaddr == NULL_ADDR || 3395 blkaddr == COMPRESS_ADDR) 3396 return false; 3397 return true; 3398 } 3399 3400 /* 3401 * file.c 3402 */ 3403 int f2fs_sync_file(struct file *file, loff_t start, loff_t end, int datasync); 3404 void f2fs_truncate_data_blocks(struct dnode_of_data *dn); 3405 int f2fs_do_truncate_blocks(struct inode *inode, u64 from, bool lock); 3406 int f2fs_truncate_blocks(struct inode *inode, u64 from, bool lock); 3407 int f2fs_truncate(struct inode *inode); 3408 int f2fs_getattr(struct user_namespace *mnt_userns, const struct path *path, 3409 struct kstat *stat, u32 request_mask, unsigned int flags); 3410 int f2fs_setattr(struct user_namespace *mnt_userns, struct dentry *dentry, 3411 struct iattr *attr); 3412 int f2fs_truncate_hole(struct inode *inode, pgoff_t pg_start, pgoff_t pg_end); 3413 void f2fs_truncate_data_blocks_range(struct dnode_of_data *dn, int count); 3414 int f2fs_precache_extents(struct inode *inode); 3415 int f2fs_fileattr_get(struct dentry *dentry, struct fileattr *fa); 3416 int f2fs_fileattr_set(struct user_namespace *mnt_userns, 3417 struct dentry *dentry, struct fileattr *fa); 3418 long f2fs_ioctl(struct file *filp, unsigned int cmd, unsigned long arg); 3419 long f2fs_compat_ioctl(struct file *file, unsigned int cmd, unsigned long arg); 3420 int f2fs_transfer_project_quota(struct inode *inode, kprojid_t kprojid); 3421 int f2fs_pin_file_control(struct inode *inode, bool inc); 3422 3423 /* 3424 * inode.c 3425 */ 3426 void f2fs_set_inode_flags(struct inode *inode); 3427 bool f2fs_inode_chksum_verify(struct f2fs_sb_info *sbi, struct page *page); 3428 void f2fs_inode_chksum_set(struct f2fs_sb_info *sbi, struct page *page); 3429 struct inode *f2fs_iget(struct super_block *sb, unsigned long ino); 3430 struct inode *f2fs_iget_retry(struct super_block *sb, unsigned long ino); 3431 int f2fs_try_to_free_nats(struct f2fs_sb_info *sbi, int nr_shrink); 3432 void f2fs_update_inode(struct inode *inode, struct page *node_page); 3433 void f2fs_update_inode_page(struct inode *inode); 3434 int f2fs_write_inode(struct inode *inode, struct writeback_control *wbc); 3435 void f2fs_evict_inode(struct inode *inode); 3436 void f2fs_handle_failed_inode(struct inode *inode); 3437 3438 /* 3439 * namei.c 3440 */ 3441 int f2fs_update_extension_list(struct f2fs_sb_info *sbi, const char *name, 3442 bool hot, bool set); 3443 struct dentry *f2fs_get_parent(struct dentry *child); 3444 3445 /* 3446 * dir.c 3447 */ 3448 unsigned char f2fs_get_de_type(struct f2fs_dir_entry *de); 3449 int f2fs_init_casefolded_name(const struct inode *dir, 3450 struct f2fs_filename *fname); 3451 int f2fs_setup_filename(struct inode *dir, const struct qstr *iname, 3452 int lookup, struct f2fs_filename *fname); 3453 int f2fs_prepare_lookup(struct inode *dir, struct dentry *dentry, 3454 struct f2fs_filename *fname); 3455 void f2fs_free_filename(struct f2fs_filename *fname); 3456 struct f2fs_dir_entry *f2fs_find_target_dentry(const struct f2fs_dentry_ptr *d, 3457 const struct f2fs_filename *fname, int *max_slots); 3458 int f2fs_fill_dentries(struct dir_context *ctx, struct f2fs_dentry_ptr *d, 3459 unsigned int start_pos, struct fscrypt_str *fstr); 3460 void f2fs_do_make_empty_dir(struct inode *inode, struct inode *parent, 3461 struct f2fs_dentry_ptr *d); 3462 struct page *f2fs_init_inode_metadata(struct inode *inode, struct inode *dir, 3463 const struct f2fs_filename *fname, struct page *dpage); 3464 void f2fs_update_parent_metadata(struct inode *dir, struct inode *inode, 3465 unsigned int current_depth); 3466 int f2fs_room_for_filename(const void *bitmap, int slots, int max_slots); 3467 void f2fs_drop_nlink(struct inode *dir, struct inode *inode); 3468 struct f2fs_dir_entry *__f2fs_find_entry(struct inode *dir, 3469 const struct f2fs_filename *fname, 3470 struct page **res_page); 3471 struct f2fs_dir_entry *f2fs_find_entry(struct inode *dir, 3472 const struct qstr *child, struct page **res_page); 3473 struct f2fs_dir_entry *f2fs_parent_dir(struct inode *dir, struct page **p); 3474 ino_t f2fs_inode_by_name(struct inode *dir, const struct qstr *qstr, 3475 struct page **page); 3476 void f2fs_set_link(struct inode *dir, struct f2fs_dir_entry *de, 3477 struct page *page, struct inode *inode); 3478 bool f2fs_has_enough_room(struct inode *dir, struct page *ipage, 3479 const struct f2fs_filename *fname); 3480 void f2fs_update_dentry(nid_t ino, umode_t mode, struct f2fs_dentry_ptr *d, 3481 const struct fscrypt_str *name, f2fs_hash_t name_hash, 3482 unsigned int bit_pos); 3483 int f2fs_add_regular_entry(struct inode *dir, const struct f2fs_filename *fname, 3484 struct inode *inode, nid_t ino, umode_t mode); 3485 int f2fs_add_dentry(struct inode *dir, const struct f2fs_filename *fname, 3486 struct inode *inode, nid_t ino, umode_t mode); 3487 int f2fs_do_add_link(struct inode *dir, const struct qstr *name, 3488 struct inode *inode, nid_t ino, umode_t mode); 3489 void f2fs_delete_entry(struct f2fs_dir_entry *dentry, struct page *page, 3490 struct inode *dir, struct inode *inode); 3491 int f2fs_do_tmpfile(struct inode *inode, struct inode *dir); 3492 bool f2fs_empty_dir(struct inode *dir); 3493 3494 static inline int f2fs_add_link(struct dentry *dentry, struct inode *inode) 3495 { 3496 if (fscrypt_is_nokey_name(dentry)) 3497 return -ENOKEY; 3498 return f2fs_do_add_link(d_inode(dentry->d_parent), &dentry->d_name, 3499 inode, inode->i_ino, inode->i_mode); 3500 } 3501 3502 /* 3503 * super.c 3504 */ 3505 int f2fs_inode_dirtied(struct inode *inode, bool sync); 3506 void f2fs_inode_synced(struct inode *inode); 3507 int f2fs_dquot_initialize(struct inode *inode); 3508 int f2fs_enable_quota_files(struct f2fs_sb_info *sbi, bool rdonly); 3509 int f2fs_quota_sync(struct super_block *sb, int type); 3510 loff_t max_file_blocks(struct inode *inode); 3511 void f2fs_quota_off_umount(struct super_block *sb); 3512 int f2fs_commit_super(struct f2fs_sb_info *sbi, bool recover); 3513 int f2fs_sync_fs(struct super_block *sb, int sync); 3514 int f2fs_sanity_check_ckpt(struct f2fs_sb_info *sbi); 3515 3516 /* 3517 * hash.c 3518 */ 3519 void f2fs_hash_filename(const struct inode *dir, struct f2fs_filename *fname); 3520 3521 /* 3522 * node.c 3523 */ 3524 struct node_info; 3525 3526 int f2fs_check_nid_range(struct f2fs_sb_info *sbi, nid_t nid); 3527 bool f2fs_available_free_memory(struct f2fs_sb_info *sbi, int type); 3528 bool f2fs_in_warm_node_list(struct f2fs_sb_info *sbi, struct page *page); 3529 void f2fs_init_fsync_node_info(struct f2fs_sb_info *sbi); 3530 void f2fs_del_fsync_node_entry(struct f2fs_sb_info *sbi, struct page *page); 3531 void f2fs_reset_fsync_node_info(struct f2fs_sb_info *sbi); 3532 int f2fs_need_dentry_mark(struct f2fs_sb_info *sbi, nid_t nid); 3533 bool f2fs_is_checkpointed_node(struct f2fs_sb_info *sbi, nid_t nid); 3534 bool f2fs_need_inode_block_update(struct f2fs_sb_info *sbi, nid_t ino); 3535 int f2fs_get_node_info(struct f2fs_sb_info *sbi, nid_t nid, 3536 struct node_info *ni, bool checkpoint_context); 3537 pgoff_t f2fs_get_next_page_offset(struct dnode_of_data *dn, pgoff_t pgofs); 3538 int f2fs_get_dnode_of_data(struct dnode_of_data *dn, pgoff_t index, int mode); 3539 int f2fs_truncate_inode_blocks(struct inode *inode, pgoff_t from); 3540 int f2fs_truncate_xattr_node(struct inode *inode); 3541 int f2fs_wait_on_node_pages_writeback(struct f2fs_sb_info *sbi, 3542 unsigned int seq_id); 3543 bool f2fs_nat_bitmap_enabled(struct f2fs_sb_info *sbi); 3544 int f2fs_remove_inode_page(struct inode *inode); 3545 struct page *f2fs_new_inode_page(struct inode *inode); 3546 struct page *f2fs_new_node_page(struct dnode_of_data *dn, unsigned int ofs); 3547 void f2fs_ra_node_page(struct f2fs_sb_info *sbi, nid_t nid); 3548 struct page *f2fs_get_node_page(struct f2fs_sb_info *sbi, pgoff_t nid); 3549 struct page *f2fs_get_node_page_ra(struct page *parent, int start); 3550 int f2fs_move_node_page(struct page *node_page, int gc_type); 3551 void f2fs_flush_inline_data(struct f2fs_sb_info *sbi); 3552 int f2fs_fsync_node_pages(struct f2fs_sb_info *sbi, struct inode *inode, 3553 struct writeback_control *wbc, bool atomic, 3554 unsigned int *seq_id); 3555 int f2fs_sync_node_pages(struct f2fs_sb_info *sbi, 3556 struct writeback_control *wbc, 3557 bool do_balance, enum iostat_type io_type); 3558 int f2fs_build_free_nids(struct f2fs_sb_info *sbi, bool sync, bool mount); 3559 bool f2fs_alloc_nid(struct f2fs_sb_info *sbi, nid_t *nid); 3560 void f2fs_alloc_nid_done(struct f2fs_sb_info *sbi, nid_t nid); 3561 void f2fs_alloc_nid_failed(struct f2fs_sb_info *sbi, nid_t nid); 3562 int f2fs_try_to_free_nids(struct f2fs_sb_info *sbi, int nr_shrink); 3563 int f2fs_recover_inline_xattr(struct inode *inode, struct page *page); 3564 int f2fs_recover_xattr_data(struct inode *inode, struct page *page); 3565 int f2fs_recover_inode_page(struct f2fs_sb_info *sbi, struct page *page); 3566 int f2fs_restore_node_summary(struct f2fs_sb_info *sbi, 3567 unsigned int segno, struct f2fs_summary_block *sum); 3568 void f2fs_enable_nat_bits(struct f2fs_sb_info *sbi); 3569 int f2fs_flush_nat_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3570 int f2fs_build_node_manager(struct f2fs_sb_info *sbi); 3571 void f2fs_destroy_node_manager(struct f2fs_sb_info *sbi); 3572 int __init f2fs_create_node_manager_caches(void); 3573 void f2fs_destroy_node_manager_caches(void); 3574 3575 /* 3576 * segment.c 3577 */ 3578 bool f2fs_need_SSR(struct f2fs_sb_info *sbi); 3579 void f2fs_register_inmem_page(struct inode *inode, struct page *page); 3580 void f2fs_drop_inmem_pages_all(struct f2fs_sb_info *sbi, bool gc_failure); 3581 void f2fs_drop_inmem_pages(struct inode *inode); 3582 void f2fs_drop_inmem_page(struct inode *inode, struct page *page); 3583 int f2fs_commit_inmem_pages(struct inode *inode); 3584 void f2fs_balance_fs(struct f2fs_sb_info *sbi, bool need); 3585 void f2fs_balance_fs_bg(struct f2fs_sb_info *sbi, bool from_bg); 3586 int f2fs_issue_flush(struct f2fs_sb_info *sbi, nid_t ino); 3587 int f2fs_create_flush_cmd_control(struct f2fs_sb_info *sbi); 3588 int f2fs_flush_device_cache(struct f2fs_sb_info *sbi); 3589 void f2fs_destroy_flush_cmd_control(struct f2fs_sb_info *sbi, bool free); 3590 void f2fs_invalidate_blocks(struct f2fs_sb_info *sbi, block_t addr); 3591 bool f2fs_is_checkpointed_data(struct f2fs_sb_info *sbi, block_t blkaddr); 3592 int f2fs_start_discard_thread(struct f2fs_sb_info *sbi); 3593 void f2fs_drop_discard_cmd(struct f2fs_sb_info *sbi); 3594 void f2fs_stop_discard_thread(struct f2fs_sb_info *sbi); 3595 bool f2fs_issue_discard_timeout(struct f2fs_sb_info *sbi); 3596 void f2fs_clear_prefree_segments(struct f2fs_sb_info *sbi, 3597 struct cp_control *cpc); 3598 void f2fs_dirty_to_prefree(struct f2fs_sb_info *sbi); 3599 block_t f2fs_get_unusable_blocks(struct f2fs_sb_info *sbi); 3600 int f2fs_disable_cp_again(struct f2fs_sb_info *sbi, block_t unusable); 3601 void f2fs_release_discard_addrs(struct f2fs_sb_info *sbi); 3602 int f2fs_npages_for_summary_flush(struct f2fs_sb_info *sbi, bool for_ra); 3603 bool f2fs_segment_has_free_slot(struct f2fs_sb_info *sbi, int segno); 3604 void f2fs_init_inmem_curseg(struct f2fs_sb_info *sbi); 3605 void f2fs_save_inmem_curseg(struct f2fs_sb_info *sbi); 3606 void f2fs_restore_inmem_curseg(struct f2fs_sb_info *sbi); 3607 void f2fs_get_new_segment(struct f2fs_sb_info *sbi, 3608 unsigned int *newseg, bool new_sec, int dir); 3609 void f2fs_allocate_segment_for_resize(struct f2fs_sb_info *sbi, int type, 3610 unsigned int start, unsigned int end); 3611 void f2fs_allocate_new_section(struct f2fs_sb_info *sbi, int type, bool force); 3612 void f2fs_allocate_new_segments(struct f2fs_sb_info *sbi); 3613 int f2fs_trim_fs(struct f2fs_sb_info *sbi, struct fstrim_range *range); 3614 bool f2fs_exist_trim_candidates(struct f2fs_sb_info *sbi, 3615 struct cp_control *cpc); 3616 struct page *f2fs_get_sum_page(struct f2fs_sb_info *sbi, unsigned int segno); 3617 void f2fs_update_meta_page(struct f2fs_sb_info *sbi, void *src, 3618 block_t blk_addr); 3619 void f2fs_do_write_meta_page(struct f2fs_sb_info *sbi, struct page *page, 3620 enum iostat_type io_type); 3621 void f2fs_do_write_node_page(unsigned int nid, struct f2fs_io_info *fio); 3622 void f2fs_outplace_write_data(struct dnode_of_data *dn, 3623 struct f2fs_io_info *fio); 3624 int f2fs_inplace_write_data(struct f2fs_io_info *fio); 3625 void f2fs_do_replace_block(struct f2fs_sb_info *sbi, struct f2fs_summary *sum, 3626 block_t old_blkaddr, block_t new_blkaddr, 3627 bool recover_curseg, bool recover_newaddr, 3628 bool from_gc); 3629 void f2fs_replace_block(struct f2fs_sb_info *sbi, struct dnode_of_data *dn, 3630 block_t old_addr, block_t new_addr, 3631 unsigned char version, bool recover_curseg, 3632 bool recover_newaddr); 3633 void f2fs_allocate_data_block(struct f2fs_sb_info *sbi, struct page *page, 3634 block_t old_blkaddr, block_t *new_blkaddr, 3635 struct f2fs_summary *sum, int type, 3636 struct f2fs_io_info *fio); 3637 void f2fs_update_device_state(struct f2fs_sb_info *sbi, nid_t ino, 3638 block_t blkaddr, unsigned int blkcnt); 3639 void f2fs_wait_on_page_writeback(struct page *page, 3640 enum page_type type, bool ordered, bool locked); 3641 void f2fs_wait_on_block_writeback(struct inode *inode, block_t blkaddr); 3642 void f2fs_wait_on_block_writeback_range(struct inode *inode, block_t blkaddr, 3643 block_t len); 3644 void f2fs_write_data_summaries(struct f2fs_sb_info *sbi, block_t start_blk); 3645 void f2fs_write_node_summaries(struct f2fs_sb_info *sbi, block_t start_blk); 3646 int f2fs_lookup_journal_in_cursum(struct f2fs_journal *journal, int type, 3647 unsigned int val, int alloc); 3648 void f2fs_flush_sit_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3649 int f2fs_fix_curseg_write_pointer(struct f2fs_sb_info *sbi); 3650 int f2fs_check_write_pointer(struct f2fs_sb_info *sbi); 3651 int f2fs_build_segment_manager(struct f2fs_sb_info *sbi); 3652 void f2fs_destroy_segment_manager(struct f2fs_sb_info *sbi); 3653 int __init f2fs_create_segment_manager_caches(void); 3654 void f2fs_destroy_segment_manager_caches(void); 3655 int f2fs_rw_hint_to_seg_type(enum rw_hint hint); 3656 enum rw_hint f2fs_io_type_to_rw_hint(struct f2fs_sb_info *sbi, 3657 enum page_type type, enum temp_type temp); 3658 unsigned int f2fs_usable_segs_in_sec(struct f2fs_sb_info *sbi, 3659 unsigned int segno); 3660 unsigned int f2fs_usable_blks_in_seg(struct f2fs_sb_info *sbi, 3661 unsigned int segno); 3662 3663 #define DEF_FRAGMENT_SIZE 4 3664 #define MIN_FRAGMENT_SIZE 1 3665 #define MAX_FRAGMENT_SIZE 512 3666 3667 static inline bool f2fs_need_rand_seg(struct f2fs_sb_info *sbi) 3668 { 3669 return F2FS_OPTION(sbi).fs_mode == FS_MODE_FRAGMENT_SEG || 3670 F2FS_OPTION(sbi).fs_mode == FS_MODE_FRAGMENT_BLK; 3671 } 3672 3673 /* 3674 * checkpoint.c 3675 */ 3676 void f2fs_stop_checkpoint(struct f2fs_sb_info *sbi, bool end_io); 3677 struct page *f2fs_grab_meta_page(struct f2fs_sb_info *sbi, pgoff_t index); 3678 struct page *f2fs_get_meta_page(struct f2fs_sb_info *sbi, pgoff_t index); 3679 struct page *f2fs_get_meta_page_retry(struct f2fs_sb_info *sbi, pgoff_t index); 3680 struct page *f2fs_get_tmp_page(struct f2fs_sb_info *sbi, pgoff_t index); 3681 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi, 3682 block_t blkaddr, int type); 3683 int f2fs_ra_meta_pages(struct f2fs_sb_info *sbi, block_t start, int nrpages, 3684 int type, bool sync); 3685 void f2fs_ra_meta_pages_cond(struct f2fs_sb_info *sbi, pgoff_t index, 3686 unsigned int ra_blocks); 3687 long f2fs_sync_meta_pages(struct f2fs_sb_info *sbi, enum page_type type, 3688 long nr_to_write, enum iostat_type io_type); 3689 void f2fs_add_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type); 3690 void f2fs_remove_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type); 3691 void f2fs_release_ino_entry(struct f2fs_sb_info *sbi, bool all); 3692 bool f2fs_exist_written_data(struct f2fs_sb_info *sbi, nid_t ino, int mode); 3693 void f2fs_set_dirty_device(struct f2fs_sb_info *sbi, nid_t ino, 3694 unsigned int devidx, int type); 3695 bool f2fs_is_dirty_device(struct f2fs_sb_info *sbi, nid_t ino, 3696 unsigned int devidx, int type); 3697 int f2fs_sync_inode_meta(struct f2fs_sb_info *sbi); 3698 int f2fs_acquire_orphan_inode(struct f2fs_sb_info *sbi); 3699 void f2fs_release_orphan_inode(struct f2fs_sb_info *sbi); 3700 void f2fs_add_orphan_inode(struct inode *inode); 3701 void f2fs_remove_orphan_inode(struct f2fs_sb_info *sbi, nid_t ino); 3702 int f2fs_recover_orphan_inodes(struct f2fs_sb_info *sbi); 3703 int f2fs_get_valid_checkpoint(struct f2fs_sb_info *sbi); 3704 void f2fs_update_dirty_page(struct inode *inode, struct page *page); 3705 void f2fs_remove_dirty_inode(struct inode *inode); 3706 int f2fs_sync_dirty_inodes(struct f2fs_sb_info *sbi, enum inode_type type); 3707 void f2fs_wait_on_all_pages(struct f2fs_sb_info *sbi, int type); 3708 u64 f2fs_get_sectors_written(struct f2fs_sb_info *sbi); 3709 int f2fs_write_checkpoint(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3710 void f2fs_init_ino_entry_info(struct f2fs_sb_info *sbi); 3711 int __init f2fs_create_checkpoint_caches(void); 3712 void f2fs_destroy_checkpoint_caches(void); 3713 int f2fs_issue_checkpoint(struct f2fs_sb_info *sbi); 3714 int f2fs_start_ckpt_thread(struct f2fs_sb_info *sbi); 3715 void f2fs_stop_ckpt_thread(struct f2fs_sb_info *sbi); 3716 void f2fs_init_ckpt_req_control(struct f2fs_sb_info *sbi); 3717 3718 /* 3719 * data.c 3720 */ 3721 int __init f2fs_init_bioset(void); 3722 void f2fs_destroy_bioset(void); 3723 int f2fs_init_bio_entry_cache(void); 3724 void f2fs_destroy_bio_entry_cache(void); 3725 void f2fs_submit_bio(struct f2fs_sb_info *sbi, 3726 struct bio *bio, enum page_type type); 3727 void f2fs_submit_merged_write(struct f2fs_sb_info *sbi, enum page_type type); 3728 void f2fs_submit_merged_write_cond(struct f2fs_sb_info *sbi, 3729 struct inode *inode, struct page *page, 3730 nid_t ino, enum page_type type); 3731 void f2fs_submit_merged_ipu_write(struct f2fs_sb_info *sbi, 3732 struct bio **bio, struct page *page); 3733 void f2fs_flush_merged_writes(struct f2fs_sb_info *sbi); 3734 int f2fs_submit_page_bio(struct f2fs_io_info *fio); 3735 int f2fs_merge_page_bio(struct f2fs_io_info *fio); 3736 void f2fs_submit_page_write(struct f2fs_io_info *fio); 3737 struct block_device *f2fs_target_device(struct f2fs_sb_info *sbi, 3738 block_t blk_addr, struct bio *bio); 3739 int f2fs_target_device_index(struct f2fs_sb_info *sbi, block_t blkaddr); 3740 void f2fs_set_data_blkaddr(struct dnode_of_data *dn); 3741 void f2fs_update_data_blkaddr(struct dnode_of_data *dn, block_t blkaddr); 3742 int f2fs_reserve_new_blocks(struct dnode_of_data *dn, blkcnt_t count); 3743 int f2fs_reserve_new_block(struct dnode_of_data *dn); 3744 int f2fs_get_block(struct dnode_of_data *dn, pgoff_t index); 3745 int f2fs_reserve_block(struct dnode_of_data *dn, pgoff_t index); 3746 struct page *f2fs_get_read_data_page(struct inode *inode, pgoff_t index, 3747 int op_flags, bool for_write); 3748 struct page *f2fs_find_data_page(struct inode *inode, pgoff_t index); 3749 struct page *f2fs_get_lock_data_page(struct inode *inode, pgoff_t index, 3750 bool for_write); 3751 struct page *f2fs_get_new_data_page(struct inode *inode, 3752 struct page *ipage, pgoff_t index, bool new_i_size); 3753 int f2fs_do_write_data_page(struct f2fs_io_info *fio); 3754 void f2fs_do_map_lock(struct f2fs_sb_info *sbi, int flag, bool lock); 3755 int f2fs_map_blocks(struct inode *inode, struct f2fs_map_blocks *map, 3756 int create, int flag); 3757 int f2fs_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, 3758 u64 start, u64 len); 3759 int f2fs_encrypt_one_page(struct f2fs_io_info *fio); 3760 bool f2fs_should_update_inplace(struct inode *inode, struct f2fs_io_info *fio); 3761 bool f2fs_should_update_outplace(struct inode *inode, struct f2fs_io_info *fio); 3762 int f2fs_write_single_data_page(struct page *page, int *submitted, 3763 struct bio **bio, sector_t *last_block, 3764 struct writeback_control *wbc, 3765 enum iostat_type io_type, 3766 int compr_blocks, bool allow_balance); 3767 void f2fs_write_failed(struct inode *inode, loff_t to); 3768 void f2fs_invalidate_page(struct page *page, unsigned int offset, 3769 unsigned int length); 3770 int f2fs_release_page(struct page *page, gfp_t wait); 3771 #ifdef CONFIG_MIGRATION 3772 int f2fs_migrate_page(struct address_space *mapping, struct page *newpage, 3773 struct page *page, enum migrate_mode mode); 3774 #endif 3775 bool f2fs_overwrite_io(struct inode *inode, loff_t pos, size_t len); 3776 void f2fs_clear_page_cache_dirty_tag(struct page *page); 3777 int f2fs_init_post_read_processing(void); 3778 void f2fs_destroy_post_read_processing(void); 3779 int f2fs_init_post_read_wq(struct f2fs_sb_info *sbi); 3780 void f2fs_destroy_post_read_wq(struct f2fs_sb_info *sbi); 3781 extern const struct iomap_ops f2fs_iomap_ops; 3782 3783 /* 3784 * gc.c 3785 */ 3786 int f2fs_start_gc_thread(struct f2fs_sb_info *sbi); 3787 void f2fs_stop_gc_thread(struct f2fs_sb_info *sbi); 3788 block_t f2fs_start_bidx_of_node(unsigned int node_ofs, struct inode *inode); 3789 int f2fs_gc(struct f2fs_sb_info *sbi, bool sync, bool background, bool force, 3790 unsigned int segno); 3791 void f2fs_build_gc_manager(struct f2fs_sb_info *sbi); 3792 int f2fs_resize_fs(struct f2fs_sb_info *sbi, __u64 block_count); 3793 int __init f2fs_create_garbage_collection_cache(void); 3794 void f2fs_destroy_garbage_collection_cache(void); 3795 3796 /* 3797 * recovery.c 3798 */ 3799 int f2fs_recover_fsync_data(struct f2fs_sb_info *sbi, bool check_only); 3800 bool f2fs_space_for_roll_forward(struct f2fs_sb_info *sbi); 3801 int __init f2fs_create_recovery_cache(void); 3802 void f2fs_destroy_recovery_cache(void); 3803 3804 /* 3805 * debug.c 3806 */ 3807 #ifdef CONFIG_F2FS_STAT_FS 3808 struct f2fs_stat_info { 3809 struct list_head stat_list; 3810 struct f2fs_sb_info *sbi; 3811 int all_area_segs, sit_area_segs, nat_area_segs, ssa_area_segs; 3812 int main_area_segs, main_area_sections, main_area_zones; 3813 unsigned long long hit_largest, hit_cached, hit_rbtree; 3814 unsigned long long hit_total, total_ext; 3815 int ext_tree, zombie_tree, ext_node; 3816 int ndirty_node, ndirty_dent, ndirty_meta, ndirty_imeta; 3817 int ndirty_data, ndirty_qdata; 3818 int inmem_pages; 3819 unsigned int ndirty_dirs, ndirty_files, nquota_files, ndirty_all; 3820 int nats, dirty_nats, sits, dirty_sits; 3821 int free_nids, avail_nids, alloc_nids; 3822 int total_count, utilization; 3823 int bg_gc, nr_wb_cp_data, nr_wb_data; 3824 int nr_rd_data, nr_rd_node, nr_rd_meta; 3825 int nr_dio_read, nr_dio_write; 3826 unsigned int io_skip_bggc, other_skip_bggc; 3827 int nr_flushing, nr_flushed, flush_list_empty; 3828 int nr_discarding, nr_discarded; 3829 int nr_discard_cmd; 3830 unsigned int undiscard_blks; 3831 int nr_issued_ckpt, nr_total_ckpt, nr_queued_ckpt; 3832 unsigned int cur_ckpt_time, peak_ckpt_time; 3833 int inline_xattr, inline_inode, inline_dir, append, update, orphans; 3834 int compr_inode; 3835 unsigned long long compr_blocks; 3836 int aw_cnt, max_aw_cnt, vw_cnt, max_vw_cnt; 3837 unsigned int valid_count, valid_node_count, valid_inode_count, discard_blks; 3838 unsigned int bimodal, avg_vblocks; 3839 int util_free, util_valid, util_invalid; 3840 int rsvd_segs, overp_segs; 3841 int dirty_count, node_pages, meta_pages, compress_pages; 3842 int compress_page_hit; 3843 int prefree_count, call_count, cp_count, bg_cp_count; 3844 int tot_segs, node_segs, data_segs, free_segs, free_secs; 3845 int bg_node_segs, bg_data_segs; 3846 int tot_blks, data_blks, node_blks; 3847 int bg_data_blks, bg_node_blks; 3848 unsigned long long skipped_atomic_files[2]; 3849 int curseg[NR_CURSEG_TYPE]; 3850 int cursec[NR_CURSEG_TYPE]; 3851 int curzone[NR_CURSEG_TYPE]; 3852 unsigned int dirty_seg[NR_CURSEG_TYPE]; 3853 unsigned int full_seg[NR_CURSEG_TYPE]; 3854 unsigned int valid_blks[NR_CURSEG_TYPE]; 3855 3856 unsigned int meta_count[META_MAX]; 3857 unsigned int segment_count[2]; 3858 unsigned int block_count[2]; 3859 unsigned int inplace_count; 3860 unsigned long long base_mem, cache_mem, page_mem; 3861 }; 3862 3863 static inline struct f2fs_stat_info *F2FS_STAT(struct f2fs_sb_info *sbi) 3864 { 3865 return (struct f2fs_stat_info *)sbi->stat_info; 3866 } 3867 3868 #define stat_inc_cp_count(si) ((si)->cp_count++) 3869 #define stat_inc_bg_cp_count(si) ((si)->bg_cp_count++) 3870 #define stat_inc_call_count(si) ((si)->call_count++) 3871 #define stat_inc_bggc_count(si) ((si)->bg_gc++) 3872 #define stat_io_skip_bggc_count(sbi) ((sbi)->io_skip_bggc++) 3873 #define stat_other_skip_bggc_count(sbi) ((sbi)->other_skip_bggc++) 3874 #define stat_inc_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]++) 3875 #define stat_dec_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]--) 3876 #define stat_inc_total_hit(sbi) (atomic64_inc(&(sbi)->total_hit_ext)) 3877 #define stat_inc_rbtree_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_rbtree)) 3878 #define stat_inc_largest_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_largest)) 3879 #define stat_inc_cached_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_cached)) 3880 #define stat_inc_inline_xattr(inode) \ 3881 do { \ 3882 if (f2fs_has_inline_xattr(inode)) \ 3883 (atomic_inc(&F2FS_I_SB(inode)->inline_xattr)); \ 3884 } while (0) 3885 #define stat_dec_inline_xattr(inode) \ 3886 do { \ 3887 if (f2fs_has_inline_xattr(inode)) \ 3888 (atomic_dec(&F2FS_I_SB(inode)->inline_xattr)); \ 3889 } while (0) 3890 #define stat_inc_inline_inode(inode) \ 3891 do { \ 3892 if (f2fs_has_inline_data(inode)) \ 3893 (atomic_inc(&F2FS_I_SB(inode)->inline_inode)); \ 3894 } while (0) 3895 #define stat_dec_inline_inode(inode) \ 3896 do { \ 3897 if (f2fs_has_inline_data(inode)) \ 3898 (atomic_dec(&F2FS_I_SB(inode)->inline_inode)); \ 3899 } while (0) 3900 #define stat_inc_inline_dir(inode) \ 3901 do { \ 3902 if (f2fs_has_inline_dentry(inode)) \ 3903 (atomic_inc(&F2FS_I_SB(inode)->inline_dir)); \ 3904 } while (0) 3905 #define stat_dec_inline_dir(inode) \ 3906 do { \ 3907 if (f2fs_has_inline_dentry(inode)) \ 3908 (atomic_dec(&F2FS_I_SB(inode)->inline_dir)); \ 3909 } while (0) 3910 #define stat_inc_compr_inode(inode) \ 3911 do { \ 3912 if (f2fs_compressed_file(inode)) \ 3913 (atomic_inc(&F2FS_I_SB(inode)->compr_inode)); \ 3914 } while (0) 3915 #define stat_dec_compr_inode(inode) \ 3916 do { \ 3917 if (f2fs_compressed_file(inode)) \ 3918 (atomic_dec(&F2FS_I_SB(inode)->compr_inode)); \ 3919 } while (0) 3920 #define stat_add_compr_blocks(inode, blocks) \ 3921 (atomic64_add(blocks, &F2FS_I_SB(inode)->compr_blocks)) 3922 #define stat_sub_compr_blocks(inode, blocks) \ 3923 (atomic64_sub(blocks, &F2FS_I_SB(inode)->compr_blocks)) 3924 #define stat_inc_meta_count(sbi, blkaddr) \ 3925 do { \ 3926 if (blkaddr < SIT_I(sbi)->sit_base_addr) \ 3927 atomic_inc(&(sbi)->meta_count[META_CP]); \ 3928 else if (blkaddr < NM_I(sbi)->nat_blkaddr) \ 3929 atomic_inc(&(sbi)->meta_count[META_SIT]); \ 3930 else if (blkaddr < SM_I(sbi)->ssa_blkaddr) \ 3931 atomic_inc(&(sbi)->meta_count[META_NAT]); \ 3932 else if (blkaddr < SM_I(sbi)->main_blkaddr) \ 3933 atomic_inc(&(sbi)->meta_count[META_SSA]); \ 3934 } while (0) 3935 #define stat_inc_seg_type(sbi, curseg) \ 3936 ((sbi)->segment_count[(curseg)->alloc_type]++) 3937 #define stat_inc_block_count(sbi, curseg) \ 3938 ((sbi)->block_count[(curseg)->alloc_type]++) 3939 #define stat_inc_inplace_blocks(sbi) \ 3940 (atomic_inc(&(sbi)->inplace_count)) 3941 #define stat_update_max_atomic_write(inode) \ 3942 do { \ 3943 int cur = F2FS_I_SB(inode)->atomic_files; \ 3944 int max = atomic_read(&F2FS_I_SB(inode)->max_aw_cnt); \ 3945 if (cur > max) \ 3946 atomic_set(&F2FS_I_SB(inode)->max_aw_cnt, cur); \ 3947 } while (0) 3948 #define stat_inc_volatile_write(inode) \ 3949 (atomic_inc(&F2FS_I_SB(inode)->vw_cnt)) 3950 #define stat_dec_volatile_write(inode) \ 3951 (atomic_dec(&F2FS_I_SB(inode)->vw_cnt)) 3952 #define stat_update_max_volatile_write(inode) \ 3953 do { \ 3954 int cur = atomic_read(&F2FS_I_SB(inode)->vw_cnt); \ 3955 int max = atomic_read(&F2FS_I_SB(inode)->max_vw_cnt); \ 3956 if (cur > max) \ 3957 atomic_set(&F2FS_I_SB(inode)->max_vw_cnt, cur); \ 3958 } while (0) 3959 #define stat_inc_seg_count(sbi, type, gc_type) \ 3960 do { \ 3961 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 3962 si->tot_segs++; \ 3963 if ((type) == SUM_TYPE_DATA) { \ 3964 si->data_segs++; \ 3965 si->bg_data_segs += (gc_type == BG_GC) ? 1 : 0; \ 3966 } else { \ 3967 si->node_segs++; \ 3968 si->bg_node_segs += (gc_type == BG_GC) ? 1 : 0; \ 3969 } \ 3970 } while (0) 3971 3972 #define stat_inc_tot_blk_count(si, blks) \ 3973 ((si)->tot_blks += (blks)) 3974 3975 #define stat_inc_data_blk_count(sbi, blks, gc_type) \ 3976 do { \ 3977 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 3978 stat_inc_tot_blk_count(si, blks); \ 3979 si->data_blks += (blks); \ 3980 si->bg_data_blks += ((gc_type) == BG_GC) ? (blks) : 0; \ 3981 } while (0) 3982 3983 #define stat_inc_node_blk_count(sbi, blks, gc_type) \ 3984 do { \ 3985 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 3986 stat_inc_tot_blk_count(si, blks); \ 3987 si->node_blks += (blks); \ 3988 si->bg_node_blks += ((gc_type) == BG_GC) ? (blks) : 0; \ 3989 } while (0) 3990 3991 int f2fs_build_stats(struct f2fs_sb_info *sbi); 3992 void f2fs_destroy_stats(struct f2fs_sb_info *sbi); 3993 void __init f2fs_create_root_stats(void); 3994 void f2fs_destroy_root_stats(void); 3995 void f2fs_update_sit_info(struct f2fs_sb_info *sbi); 3996 #else 3997 #define stat_inc_cp_count(si) do { } while (0) 3998 #define stat_inc_bg_cp_count(si) do { } while (0) 3999 #define stat_inc_call_count(si) do { } while (0) 4000 #define stat_inc_bggc_count(si) do { } while (0) 4001 #define stat_io_skip_bggc_count(sbi) do { } while (0) 4002 #define stat_other_skip_bggc_count(sbi) do { } while (0) 4003 #define stat_inc_dirty_inode(sbi, type) do { } while (0) 4004 #define stat_dec_dirty_inode(sbi, type) do { } while (0) 4005 #define stat_inc_total_hit(sbi) do { } while (0) 4006 #define stat_inc_rbtree_node_hit(sbi) do { } while (0) 4007 #define stat_inc_largest_node_hit(sbi) do { } while (0) 4008 #define stat_inc_cached_node_hit(sbi) do { } while (0) 4009 #define stat_inc_inline_xattr(inode) do { } while (0) 4010 #define stat_dec_inline_xattr(inode) do { } while (0) 4011 #define stat_inc_inline_inode(inode) do { } while (0) 4012 #define stat_dec_inline_inode(inode) do { } while (0) 4013 #define stat_inc_inline_dir(inode) do { } while (0) 4014 #define stat_dec_inline_dir(inode) do { } while (0) 4015 #define stat_inc_compr_inode(inode) do { } while (0) 4016 #define stat_dec_compr_inode(inode) do { } while (0) 4017 #define stat_add_compr_blocks(inode, blocks) do { } while (0) 4018 #define stat_sub_compr_blocks(inode, blocks) do { } while (0) 4019 #define stat_update_max_atomic_write(inode) do { } while (0) 4020 #define stat_inc_volatile_write(inode) do { } while (0) 4021 #define stat_dec_volatile_write(inode) do { } while (0) 4022 #define stat_update_max_volatile_write(inode) do { } while (0) 4023 #define stat_inc_meta_count(sbi, blkaddr) do { } while (0) 4024 #define stat_inc_seg_type(sbi, curseg) do { } while (0) 4025 #define stat_inc_block_count(sbi, curseg) do { } while (0) 4026 #define stat_inc_inplace_blocks(sbi) do { } while (0) 4027 #define stat_inc_seg_count(sbi, type, gc_type) do { } while (0) 4028 #define stat_inc_tot_blk_count(si, blks) do { } while (0) 4029 #define stat_inc_data_blk_count(sbi, blks, gc_type) do { } while (0) 4030 #define stat_inc_node_blk_count(sbi, blks, gc_type) do { } while (0) 4031 4032 static inline int f2fs_build_stats(struct f2fs_sb_info *sbi) { return 0; } 4033 static inline void f2fs_destroy_stats(struct f2fs_sb_info *sbi) { } 4034 static inline void __init f2fs_create_root_stats(void) { } 4035 static inline void f2fs_destroy_root_stats(void) { } 4036 static inline void f2fs_update_sit_info(struct f2fs_sb_info *sbi) {} 4037 #endif 4038 4039 extern const struct file_operations f2fs_dir_operations; 4040 extern const struct file_operations f2fs_file_operations; 4041 extern const struct inode_operations f2fs_file_inode_operations; 4042 extern const struct address_space_operations f2fs_dblock_aops; 4043 extern const struct address_space_operations f2fs_node_aops; 4044 extern const struct address_space_operations f2fs_meta_aops; 4045 extern const struct inode_operations f2fs_dir_inode_operations; 4046 extern const struct inode_operations f2fs_symlink_inode_operations; 4047 extern const struct inode_operations f2fs_encrypted_symlink_inode_operations; 4048 extern const struct inode_operations f2fs_special_inode_operations; 4049 extern struct kmem_cache *f2fs_inode_entry_slab; 4050 4051 /* 4052 * inline.c 4053 */ 4054 bool f2fs_may_inline_data(struct inode *inode); 4055 bool f2fs_may_inline_dentry(struct inode *inode); 4056 void f2fs_do_read_inline_data(struct page *page, struct page *ipage); 4057 void f2fs_truncate_inline_inode(struct inode *inode, 4058 struct page *ipage, u64 from); 4059 int f2fs_read_inline_data(struct inode *inode, struct page *page); 4060 int f2fs_convert_inline_page(struct dnode_of_data *dn, struct page *page); 4061 int f2fs_convert_inline_inode(struct inode *inode); 4062 int f2fs_try_convert_inline_dir(struct inode *dir, struct dentry *dentry); 4063 int f2fs_write_inline_data(struct inode *inode, struct page *page); 4064 int f2fs_recover_inline_data(struct inode *inode, struct page *npage); 4065 struct f2fs_dir_entry *f2fs_find_in_inline_dir(struct inode *dir, 4066 const struct f2fs_filename *fname, 4067 struct page **res_page); 4068 int f2fs_make_empty_inline_dir(struct inode *inode, struct inode *parent, 4069 struct page *ipage); 4070 int f2fs_add_inline_entry(struct inode *dir, const struct f2fs_filename *fname, 4071 struct inode *inode, nid_t ino, umode_t mode); 4072 void f2fs_delete_inline_entry(struct f2fs_dir_entry *dentry, 4073 struct page *page, struct inode *dir, 4074 struct inode *inode); 4075 bool f2fs_empty_inline_dir(struct inode *dir); 4076 int f2fs_read_inline_dir(struct file *file, struct dir_context *ctx, 4077 struct fscrypt_str *fstr); 4078 int f2fs_inline_data_fiemap(struct inode *inode, 4079 struct fiemap_extent_info *fieinfo, 4080 __u64 start, __u64 len); 4081 4082 /* 4083 * shrinker.c 4084 */ 4085 unsigned long f2fs_shrink_count(struct shrinker *shrink, 4086 struct shrink_control *sc); 4087 unsigned long f2fs_shrink_scan(struct shrinker *shrink, 4088 struct shrink_control *sc); 4089 void f2fs_join_shrinker(struct f2fs_sb_info *sbi); 4090 void f2fs_leave_shrinker(struct f2fs_sb_info *sbi); 4091 4092 /* 4093 * extent_cache.c 4094 */ 4095 struct rb_entry *f2fs_lookup_rb_tree(struct rb_root_cached *root, 4096 struct rb_entry *cached_re, unsigned int ofs); 4097 struct rb_node **f2fs_lookup_rb_tree_ext(struct f2fs_sb_info *sbi, 4098 struct rb_root_cached *root, 4099 struct rb_node **parent, 4100 unsigned long long key, bool *left_most); 4101 struct rb_node **f2fs_lookup_rb_tree_for_insert(struct f2fs_sb_info *sbi, 4102 struct rb_root_cached *root, 4103 struct rb_node **parent, 4104 unsigned int ofs, bool *leftmost); 4105 struct rb_entry *f2fs_lookup_rb_tree_ret(struct rb_root_cached *root, 4106 struct rb_entry *cached_re, unsigned int ofs, 4107 struct rb_entry **prev_entry, struct rb_entry **next_entry, 4108 struct rb_node ***insert_p, struct rb_node **insert_parent, 4109 bool force, bool *leftmost); 4110 bool f2fs_check_rb_tree_consistence(struct f2fs_sb_info *sbi, 4111 struct rb_root_cached *root, bool check_key); 4112 unsigned int f2fs_shrink_extent_tree(struct f2fs_sb_info *sbi, int nr_shrink); 4113 void f2fs_init_extent_tree(struct inode *inode, struct page *ipage); 4114 void f2fs_drop_extent_tree(struct inode *inode); 4115 unsigned int f2fs_destroy_extent_node(struct inode *inode); 4116 void f2fs_destroy_extent_tree(struct inode *inode); 4117 bool f2fs_lookup_extent_cache(struct inode *inode, pgoff_t pgofs, 4118 struct extent_info *ei); 4119 void f2fs_update_extent_cache(struct dnode_of_data *dn); 4120 void f2fs_update_extent_cache_range(struct dnode_of_data *dn, 4121 pgoff_t fofs, block_t blkaddr, unsigned int len); 4122 void f2fs_init_extent_cache_info(struct f2fs_sb_info *sbi); 4123 int __init f2fs_create_extent_cache(void); 4124 void f2fs_destroy_extent_cache(void); 4125 4126 /* 4127 * sysfs.c 4128 */ 4129 #define MIN_RA_MUL 2 4130 #define MAX_RA_MUL 256 4131 4132 int __init f2fs_init_sysfs(void); 4133 void f2fs_exit_sysfs(void); 4134 int f2fs_register_sysfs(struct f2fs_sb_info *sbi); 4135 void f2fs_unregister_sysfs(struct f2fs_sb_info *sbi); 4136 4137 /* verity.c */ 4138 extern const struct fsverity_operations f2fs_verityops; 4139 4140 /* 4141 * crypto support 4142 */ 4143 static inline bool f2fs_encrypted_file(struct inode *inode) 4144 { 4145 return IS_ENCRYPTED(inode) && S_ISREG(inode->i_mode); 4146 } 4147 4148 static inline void f2fs_set_encrypted_inode(struct inode *inode) 4149 { 4150 #ifdef CONFIG_FS_ENCRYPTION 4151 file_set_encrypt(inode); 4152 f2fs_set_inode_flags(inode); 4153 #endif 4154 } 4155 4156 /* 4157 * Returns true if the reads of the inode's data need to undergo some 4158 * postprocessing step, like decryption or authenticity verification. 4159 */ 4160 static inline bool f2fs_post_read_required(struct inode *inode) 4161 { 4162 return f2fs_encrypted_file(inode) || fsverity_active(inode) || 4163 f2fs_compressed_file(inode); 4164 } 4165 4166 /* 4167 * compress.c 4168 */ 4169 #ifdef CONFIG_F2FS_FS_COMPRESSION 4170 bool f2fs_is_compressed_page(struct page *page); 4171 struct page *f2fs_compress_control_page(struct page *page); 4172 int f2fs_prepare_compress_overwrite(struct inode *inode, 4173 struct page **pagep, pgoff_t index, void **fsdata); 4174 bool f2fs_compress_write_end(struct inode *inode, void *fsdata, 4175 pgoff_t index, unsigned copied); 4176 int f2fs_truncate_partial_cluster(struct inode *inode, u64 from, bool lock); 4177 void f2fs_compress_write_end_io(struct bio *bio, struct page *page); 4178 bool f2fs_is_compress_backend_ready(struct inode *inode); 4179 int f2fs_init_compress_mempool(void); 4180 void f2fs_destroy_compress_mempool(void); 4181 void f2fs_decompress_cluster(struct decompress_io_ctx *dic); 4182 void f2fs_end_read_compressed_page(struct page *page, bool failed, 4183 block_t blkaddr); 4184 bool f2fs_cluster_is_empty(struct compress_ctx *cc); 4185 bool f2fs_cluster_can_merge_page(struct compress_ctx *cc, pgoff_t index); 4186 bool f2fs_all_cluster_page_loaded(struct compress_ctx *cc, struct pagevec *pvec, 4187 int index, int nr_pages); 4188 bool f2fs_sanity_check_cluster(struct dnode_of_data *dn); 4189 void f2fs_compress_ctx_add_page(struct compress_ctx *cc, struct page *page); 4190 int f2fs_write_multi_pages(struct compress_ctx *cc, 4191 int *submitted, 4192 struct writeback_control *wbc, 4193 enum iostat_type io_type); 4194 int f2fs_is_compressed_cluster(struct inode *inode, pgoff_t index); 4195 void f2fs_update_extent_tree_range_compressed(struct inode *inode, 4196 pgoff_t fofs, block_t blkaddr, unsigned int llen, 4197 unsigned int c_len); 4198 int f2fs_read_multi_pages(struct compress_ctx *cc, struct bio **bio_ret, 4199 unsigned nr_pages, sector_t *last_block_in_bio, 4200 bool is_readahead, bool for_write); 4201 struct decompress_io_ctx *f2fs_alloc_dic(struct compress_ctx *cc); 4202 void f2fs_decompress_end_io(struct decompress_io_ctx *dic, bool failed); 4203 void f2fs_put_page_dic(struct page *page); 4204 unsigned int f2fs_cluster_blocks_are_contiguous(struct dnode_of_data *dn); 4205 int f2fs_init_compress_ctx(struct compress_ctx *cc); 4206 void f2fs_destroy_compress_ctx(struct compress_ctx *cc, bool reuse); 4207 void f2fs_init_compress_info(struct f2fs_sb_info *sbi); 4208 int f2fs_init_compress_inode(struct f2fs_sb_info *sbi); 4209 void f2fs_destroy_compress_inode(struct f2fs_sb_info *sbi); 4210 int f2fs_init_page_array_cache(struct f2fs_sb_info *sbi); 4211 void f2fs_destroy_page_array_cache(struct f2fs_sb_info *sbi); 4212 int __init f2fs_init_compress_cache(void); 4213 void f2fs_destroy_compress_cache(void); 4214 struct address_space *COMPRESS_MAPPING(struct f2fs_sb_info *sbi); 4215 void f2fs_invalidate_compress_page(struct f2fs_sb_info *sbi, block_t blkaddr); 4216 void f2fs_cache_compressed_page(struct f2fs_sb_info *sbi, struct page *page, 4217 nid_t ino, block_t blkaddr); 4218 bool f2fs_load_compressed_page(struct f2fs_sb_info *sbi, struct page *page, 4219 block_t blkaddr); 4220 void f2fs_invalidate_compress_pages(struct f2fs_sb_info *sbi, nid_t ino); 4221 #define inc_compr_inode_stat(inode) \ 4222 do { \ 4223 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); \ 4224 sbi->compr_new_inode++; \ 4225 } while (0) 4226 #define add_compr_block_stat(inode, blocks) \ 4227 do { \ 4228 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); \ 4229 int diff = F2FS_I(inode)->i_cluster_size - blocks; \ 4230 sbi->compr_written_block += blocks; \ 4231 sbi->compr_saved_block += diff; \ 4232 } while (0) 4233 #else 4234 static inline bool f2fs_is_compressed_page(struct page *page) { return false; } 4235 static inline bool f2fs_is_compress_backend_ready(struct inode *inode) 4236 { 4237 if (!f2fs_compressed_file(inode)) 4238 return true; 4239 /* not support compression */ 4240 return false; 4241 } 4242 static inline struct page *f2fs_compress_control_page(struct page *page) 4243 { 4244 WARN_ON_ONCE(1); 4245 return ERR_PTR(-EINVAL); 4246 } 4247 static inline int f2fs_init_compress_mempool(void) { return 0; } 4248 static inline void f2fs_destroy_compress_mempool(void) { } 4249 static inline void f2fs_decompress_cluster(struct decompress_io_ctx *dic) { } 4250 static inline void f2fs_end_read_compressed_page(struct page *page, 4251 bool failed, block_t blkaddr) 4252 { 4253 WARN_ON_ONCE(1); 4254 } 4255 static inline void f2fs_put_page_dic(struct page *page) 4256 { 4257 WARN_ON_ONCE(1); 4258 } 4259 static inline unsigned int f2fs_cluster_blocks_are_contiguous(struct dnode_of_data *dn) { return 0; } 4260 static inline bool f2fs_sanity_check_cluster(struct dnode_of_data *dn) { return false; } 4261 static inline int f2fs_init_compress_inode(struct f2fs_sb_info *sbi) { return 0; } 4262 static inline void f2fs_destroy_compress_inode(struct f2fs_sb_info *sbi) { } 4263 static inline int f2fs_init_page_array_cache(struct f2fs_sb_info *sbi) { return 0; } 4264 static inline void f2fs_destroy_page_array_cache(struct f2fs_sb_info *sbi) { } 4265 static inline int __init f2fs_init_compress_cache(void) { return 0; } 4266 static inline void f2fs_destroy_compress_cache(void) { } 4267 static inline void f2fs_invalidate_compress_page(struct f2fs_sb_info *sbi, 4268 block_t blkaddr) { } 4269 static inline void f2fs_cache_compressed_page(struct f2fs_sb_info *sbi, 4270 struct page *page, nid_t ino, block_t blkaddr) { } 4271 static inline bool f2fs_load_compressed_page(struct f2fs_sb_info *sbi, 4272 struct page *page, block_t blkaddr) { return false; } 4273 static inline void f2fs_invalidate_compress_pages(struct f2fs_sb_info *sbi, 4274 nid_t ino) { } 4275 #define inc_compr_inode_stat(inode) do { } while (0) 4276 static inline void f2fs_update_extent_tree_range_compressed(struct inode *inode, 4277 pgoff_t fofs, block_t blkaddr, unsigned int llen, 4278 unsigned int c_len) { } 4279 #endif 4280 4281 static inline void set_compress_context(struct inode *inode) 4282 { 4283 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 4284 4285 F2FS_I(inode)->i_compress_algorithm = 4286 F2FS_OPTION(sbi).compress_algorithm; 4287 F2FS_I(inode)->i_log_cluster_size = 4288 F2FS_OPTION(sbi).compress_log_size; 4289 F2FS_I(inode)->i_compress_flag = 4290 F2FS_OPTION(sbi).compress_chksum ? 4291 1 << COMPRESS_CHKSUM : 0; 4292 F2FS_I(inode)->i_cluster_size = 4293 1 << F2FS_I(inode)->i_log_cluster_size; 4294 if ((F2FS_I(inode)->i_compress_algorithm == COMPRESS_LZ4 || 4295 F2FS_I(inode)->i_compress_algorithm == COMPRESS_ZSTD) && 4296 F2FS_OPTION(sbi).compress_level) 4297 F2FS_I(inode)->i_compress_flag |= 4298 F2FS_OPTION(sbi).compress_level << 4299 COMPRESS_LEVEL_OFFSET; 4300 F2FS_I(inode)->i_flags |= F2FS_COMPR_FL; 4301 set_inode_flag(inode, FI_COMPRESSED_FILE); 4302 stat_inc_compr_inode(inode); 4303 inc_compr_inode_stat(inode); 4304 f2fs_mark_inode_dirty_sync(inode, true); 4305 } 4306 4307 static inline bool f2fs_disable_compressed_file(struct inode *inode) 4308 { 4309 struct f2fs_inode_info *fi = F2FS_I(inode); 4310 4311 if (!f2fs_compressed_file(inode)) 4312 return true; 4313 if (S_ISREG(inode->i_mode) && F2FS_HAS_BLOCKS(inode)) 4314 return false; 4315 4316 fi->i_flags &= ~F2FS_COMPR_FL; 4317 stat_dec_compr_inode(inode); 4318 clear_inode_flag(inode, FI_COMPRESSED_FILE); 4319 f2fs_mark_inode_dirty_sync(inode, true); 4320 return true; 4321 } 4322 4323 #define F2FS_FEATURE_FUNCS(name, flagname) \ 4324 static inline int f2fs_sb_has_##name(struct f2fs_sb_info *sbi) \ 4325 { \ 4326 return F2FS_HAS_FEATURE(sbi, F2FS_FEATURE_##flagname); \ 4327 } 4328 4329 F2FS_FEATURE_FUNCS(encrypt, ENCRYPT); 4330 F2FS_FEATURE_FUNCS(blkzoned, BLKZONED); 4331 F2FS_FEATURE_FUNCS(extra_attr, EXTRA_ATTR); 4332 F2FS_FEATURE_FUNCS(project_quota, PRJQUOTA); 4333 F2FS_FEATURE_FUNCS(inode_chksum, INODE_CHKSUM); 4334 F2FS_FEATURE_FUNCS(flexible_inline_xattr, FLEXIBLE_INLINE_XATTR); 4335 F2FS_FEATURE_FUNCS(quota_ino, QUOTA_INO); 4336 F2FS_FEATURE_FUNCS(inode_crtime, INODE_CRTIME); 4337 F2FS_FEATURE_FUNCS(lost_found, LOST_FOUND); 4338 F2FS_FEATURE_FUNCS(verity, VERITY); 4339 F2FS_FEATURE_FUNCS(sb_chksum, SB_CHKSUM); 4340 F2FS_FEATURE_FUNCS(casefold, CASEFOLD); 4341 F2FS_FEATURE_FUNCS(compression, COMPRESSION); 4342 F2FS_FEATURE_FUNCS(readonly, RO); 4343 4344 static inline bool f2fs_may_extent_tree(struct inode *inode) 4345 { 4346 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 4347 4348 if (!test_opt(sbi, EXTENT_CACHE) || 4349 is_inode_flag_set(inode, FI_NO_EXTENT) || 4350 (is_inode_flag_set(inode, FI_COMPRESSED_FILE) && 4351 !f2fs_sb_has_readonly(sbi))) 4352 return false; 4353 4354 /* 4355 * for recovered files during mount do not create extents 4356 * if shrinker is not registered. 4357 */ 4358 if (list_empty(&sbi->s_list)) 4359 return false; 4360 4361 return S_ISREG(inode->i_mode); 4362 } 4363 4364 #ifdef CONFIG_BLK_DEV_ZONED 4365 static inline bool f2fs_blkz_is_seq(struct f2fs_sb_info *sbi, int devi, 4366 block_t blkaddr) 4367 { 4368 unsigned int zno = blkaddr >> sbi->log_blocks_per_blkz; 4369 4370 return test_bit(zno, FDEV(devi).blkz_seq); 4371 } 4372 #endif 4373 4374 static inline bool f2fs_hw_should_discard(struct f2fs_sb_info *sbi) 4375 { 4376 return f2fs_sb_has_blkzoned(sbi); 4377 } 4378 4379 static inline bool f2fs_bdev_support_discard(struct block_device *bdev) 4380 { 4381 return blk_queue_discard(bdev_get_queue(bdev)) || 4382 bdev_is_zoned(bdev); 4383 } 4384 4385 static inline bool f2fs_hw_support_discard(struct f2fs_sb_info *sbi) 4386 { 4387 int i; 4388 4389 if (!f2fs_is_multi_device(sbi)) 4390 return f2fs_bdev_support_discard(sbi->sb->s_bdev); 4391 4392 for (i = 0; i < sbi->s_ndevs; i++) 4393 if (f2fs_bdev_support_discard(FDEV(i).bdev)) 4394 return true; 4395 return false; 4396 } 4397 4398 static inline bool f2fs_realtime_discard_enable(struct f2fs_sb_info *sbi) 4399 { 4400 return (test_opt(sbi, DISCARD) && f2fs_hw_support_discard(sbi)) || 4401 f2fs_hw_should_discard(sbi); 4402 } 4403 4404 static inline bool f2fs_hw_is_readonly(struct f2fs_sb_info *sbi) 4405 { 4406 int i; 4407 4408 if (!f2fs_is_multi_device(sbi)) 4409 return bdev_read_only(sbi->sb->s_bdev); 4410 4411 for (i = 0; i < sbi->s_ndevs; i++) 4412 if (bdev_read_only(FDEV(i).bdev)) 4413 return true; 4414 return false; 4415 } 4416 4417 static inline bool f2fs_lfs_mode(struct f2fs_sb_info *sbi) 4418 { 4419 return F2FS_OPTION(sbi).fs_mode == FS_MODE_LFS; 4420 } 4421 4422 static inline bool f2fs_may_compress(struct inode *inode) 4423 { 4424 if (IS_SWAPFILE(inode) || f2fs_is_pinned_file(inode) || 4425 f2fs_is_atomic_file(inode) || 4426 f2fs_is_volatile_file(inode)) 4427 return false; 4428 return S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode); 4429 } 4430 4431 static inline void f2fs_i_compr_blocks_update(struct inode *inode, 4432 u64 blocks, bool add) 4433 { 4434 int diff = F2FS_I(inode)->i_cluster_size - blocks; 4435 struct f2fs_inode_info *fi = F2FS_I(inode); 4436 4437 /* don't update i_compr_blocks if saved blocks were released */ 4438 if (!add && !atomic_read(&fi->i_compr_blocks)) 4439 return; 4440 4441 if (add) { 4442 atomic_add(diff, &fi->i_compr_blocks); 4443 stat_add_compr_blocks(inode, diff); 4444 } else { 4445 atomic_sub(diff, &fi->i_compr_blocks); 4446 stat_sub_compr_blocks(inode, diff); 4447 } 4448 f2fs_mark_inode_dirty_sync(inode, true); 4449 } 4450 4451 static inline int block_unaligned_IO(struct inode *inode, 4452 struct kiocb *iocb, struct iov_iter *iter) 4453 { 4454 unsigned int i_blkbits = READ_ONCE(inode->i_blkbits); 4455 unsigned int blocksize_mask = (1 << i_blkbits) - 1; 4456 loff_t offset = iocb->ki_pos; 4457 unsigned long align = offset | iov_iter_alignment(iter); 4458 4459 return align & blocksize_mask; 4460 } 4461 4462 static inline bool f2fs_allow_multi_device_dio(struct f2fs_sb_info *sbi, 4463 int flag) 4464 { 4465 if (!f2fs_is_multi_device(sbi)) 4466 return false; 4467 if (flag != F2FS_GET_BLOCK_DIO) 4468 return false; 4469 return sbi->aligned_blksize; 4470 } 4471 4472 static inline bool f2fs_force_buffered_io(struct inode *inode, 4473 struct kiocb *iocb, struct iov_iter *iter) 4474 { 4475 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 4476 int rw = iov_iter_rw(iter); 4477 4478 if (f2fs_post_read_required(inode)) 4479 return true; 4480 4481 /* disallow direct IO if any of devices has unaligned blksize */ 4482 if (f2fs_is_multi_device(sbi) && !sbi->aligned_blksize) 4483 return true; 4484 /* 4485 * for blkzoned device, fallback direct IO to buffered IO, so 4486 * all IOs can be serialized by log-structured write. 4487 */ 4488 if (f2fs_sb_has_blkzoned(sbi)) 4489 return true; 4490 if (f2fs_lfs_mode(sbi) && (rw == WRITE)) { 4491 if (block_unaligned_IO(inode, iocb, iter)) 4492 return true; 4493 if (F2FS_IO_ALIGNED(sbi)) 4494 return true; 4495 } 4496 if (is_sbi_flag_set(F2FS_I_SB(inode), SBI_CP_DISABLED)) 4497 return true; 4498 4499 return false; 4500 } 4501 4502 static inline bool f2fs_need_verity(const struct inode *inode, pgoff_t idx) 4503 { 4504 return fsverity_active(inode) && 4505 idx < DIV_ROUND_UP(inode->i_size, PAGE_SIZE); 4506 } 4507 4508 #ifdef CONFIG_F2FS_FAULT_INJECTION 4509 extern void f2fs_build_fault_attr(struct f2fs_sb_info *sbi, unsigned int rate, 4510 unsigned int type); 4511 #else 4512 #define f2fs_build_fault_attr(sbi, rate, type) do { } while (0) 4513 #endif 4514 4515 static inline bool is_journalled_quota(struct f2fs_sb_info *sbi) 4516 { 4517 #ifdef CONFIG_QUOTA 4518 if (f2fs_sb_has_quota_ino(sbi)) 4519 return true; 4520 if (F2FS_OPTION(sbi).s_qf_names[USRQUOTA] || 4521 F2FS_OPTION(sbi).s_qf_names[GRPQUOTA] || 4522 F2FS_OPTION(sbi).s_qf_names[PRJQUOTA]) 4523 return true; 4524 #endif 4525 return false; 4526 } 4527 4528 static inline bool f2fs_block_unit_discard(struct f2fs_sb_info *sbi) 4529 { 4530 return F2FS_OPTION(sbi).discard_unit == DISCARD_UNIT_BLOCK; 4531 } 4532 4533 #define EFSBADCRC EBADMSG /* Bad CRC detected */ 4534 #define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */ 4535 4536 #endif /* _LINUX_F2FS_H */ 4537