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 #ifdef CONFIG_F2FS_CHECK_FS 32 #define f2fs_bug_on(sbi, condition) BUG_ON(condition) 33 #else 34 #define f2fs_bug_on(sbi, condition) \ 35 do { \ 36 if (WARN_ON(condition)) \ 37 set_sbi_flag(sbi, SBI_NEED_FSCK); \ 38 } while (0) 39 #endif 40 41 enum { 42 FAULT_KMALLOC, 43 FAULT_KVMALLOC, 44 FAULT_PAGE_ALLOC, 45 FAULT_PAGE_GET, 46 FAULT_ALLOC_NID, 47 FAULT_ORPHAN, 48 FAULT_BLOCK, 49 FAULT_DIR_DEPTH, 50 FAULT_EVICT_INODE, 51 FAULT_TRUNCATE, 52 FAULT_READ_IO, 53 FAULT_CHECKPOINT, 54 FAULT_DISCARD, 55 FAULT_WRITE_IO, 56 FAULT_MAX, 57 }; 58 59 #ifdef CONFIG_F2FS_FAULT_INJECTION 60 #define F2FS_ALL_FAULT_TYPE ((1 << FAULT_MAX) - 1) 61 62 struct f2fs_fault_info { 63 atomic_t inject_ops; 64 unsigned int inject_rate; 65 unsigned int inject_type; 66 }; 67 68 extern const char *f2fs_fault_name[FAULT_MAX]; 69 #define IS_FAULT_SET(fi, type) ((fi)->inject_type & (1 << (type))) 70 #endif 71 72 /* 73 * For mount options 74 */ 75 #define F2FS_MOUNT_DISABLE_ROLL_FORWARD 0x00000002 76 #define F2FS_MOUNT_DISCARD 0x00000004 77 #define F2FS_MOUNT_NOHEAP 0x00000008 78 #define F2FS_MOUNT_XATTR_USER 0x00000010 79 #define F2FS_MOUNT_POSIX_ACL 0x00000020 80 #define F2FS_MOUNT_DISABLE_EXT_IDENTIFY 0x00000040 81 #define F2FS_MOUNT_INLINE_XATTR 0x00000080 82 #define F2FS_MOUNT_INLINE_DATA 0x00000100 83 #define F2FS_MOUNT_INLINE_DENTRY 0x00000200 84 #define F2FS_MOUNT_FLUSH_MERGE 0x00000400 85 #define F2FS_MOUNT_NOBARRIER 0x00000800 86 #define F2FS_MOUNT_FASTBOOT 0x00001000 87 #define F2FS_MOUNT_EXTENT_CACHE 0x00002000 88 #define F2FS_MOUNT_DATA_FLUSH 0x00008000 89 #define F2FS_MOUNT_FAULT_INJECTION 0x00010000 90 #define F2FS_MOUNT_USRQUOTA 0x00080000 91 #define F2FS_MOUNT_GRPQUOTA 0x00100000 92 #define F2FS_MOUNT_PRJQUOTA 0x00200000 93 #define F2FS_MOUNT_QUOTA 0x00400000 94 #define F2FS_MOUNT_INLINE_XATTR_SIZE 0x00800000 95 #define F2FS_MOUNT_RESERVE_ROOT 0x01000000 96 #define F2FS_MOUNT_DISABLE_CHECKPOINT 0x02000000 97 #define F2FS_MOUNT_NORECOVERY 0x04000000 98 #define F2FS_MOUNT_ATGC 0x08000000 99 #define F2FS_MOUNT_MERGE_CHECKPOINT 0x10000000 100 101 #define F2FS_OPTION(sbi) ((sbi)->mount_opt) 102 #define clear_opt(sbi, option) (F2FS_OPTION(sbi).opt &= ~F2FS_MOUNT_##option) 103 #define set_opt(sbi, option) (F2FS_OPTION(sbi).opt |= F2FS_MOUNT_##option) 104 #define test_opt(sbi, option) (F2FS_OPTION(sbi).opt & F2FS_MOUNT_##option) 105 106 #define ver_after(a, b) (typecheck(unsigned long long, a) && \ 107 typecheck(unsigned long long, b) && \ 108 ((long long)((a) - (b)) > 0)) 109 110 typedef u32 block_t; /* 111 * should not change u32, since it is the on-disk block 112 * address format, __le32. 113 */ 114 typedef u32 nid_t; 115 116 #define COMPRESS_EXT_NUM 16 117 118 struct f2fs_mount_info { 119 unsigned int opt; 120 int write_io_size_bits; /* Write IO size bits */ 121 block_t root_reserved_blocks; /* root reserved blocks */ 122 kuid_t s_resuid; /* reserved blocks for uid */ 123 kgid_t s_resgid; /* reserved blocks for gid */ 124 int active_logs; /* # of active logs */ 125 int inline_xattr_size; /* inline xattr size */ 126 #ifdef CONFIG_F2FS_FAULT_INJECTION 127 struct f2fs_fault_info fault_info; /* For fault injection */ 128 #endif 129 #ifdef CONFIG_QUOTA 130 /* Names of quota files with journalled quota */ 131 char *s_qf_names[MAXQUOTAS]; 132 int s_jquota_fmt; /* Format of quota to use */ 133 #endif 134 /* For which write hints are passed down to block layer */ 135 int whint_mode; 136 int alloc_mode; /* segment allocation policy */ 137 int fsync_mode; /* fsync policy */ 138 int fs_mode; /* fs mode: LFS or ADAPTIVE */ 139 int bggc_mode; /* bggc mode: off, on or sync */ 140 struct fscrypt_dummy_policy dummy_enc_policy; /* test dummy encryption */ 141 block_t unusable_cap_perc; /* percentage for cap */ 142 block_t unusable_cap; /* Amount of space allowed to be 143 * unusable when disabling checkpoint 144 */ 145 146 /* For compression */ 147 unsigned char compress_algorithm; /* algorithm type */ 148 unsigned char compress_log_size; /* cluster log size */ 149 unsigned char compress_level; /* compress level */ 150 bool compress_chksum; /* compressed data chksum */ 151 unsigned char compress_ext_cnt; /* extension count */ 152 int compress_mode; /* compression mode */ 153 unsigned char extensions[COMPRESS_EXT_NUM][F2FS_EXTENSION_LEN]; /* extensions */ 154 }; 155 156 #define F2FS_FEATURE_ENCRYPT 0x0001 157 #define F2FS_FEATURE_BLKZONED 0x0002 158 #define F2FS_FEATURE_ATOMIC_WRITE 0x0004 159 #define F2FS_FEATURE_EXTRA_ATTR 0x0008 160 #define F2FS_FEATURE_PRJQUOTA 0x0010 161 #define F2FS_FEATURE_INODE_CHKSUM 0x0020 162 #define F2FS_FEATURE_FLEXIBLE_INLINE_XATTR 0x0040 163 #define F2FS_FEATURE_QUOTA_INO 0x0080 164 #define F2FS_FEATURE_INODE_CRTIME 0x0100 165 #define F2FS_FEATURE_LOST_FOUND 0x0200 166 #define F2FS_FEATURE_VERITY 0x0400 167 #define F2FS_FEATURE_SB_CHKSUM 0x0800 168 #define F2FS_FEATURE_CASEFOLD 0x1000 169 #define F2FS_FEATURE_COMPRESSION 0x2000 170 171 #define __F2FS_HAS_FEATURE(raw_super, mask) \ 172 ((raw_super->feature & cpu_to_le32(mask)) != 0) 173 #define F2FS_HAS_FEATURE(sbi, mask) __F2FS_HAS_FEATURE(sbi->raw_super, mask) 174 #define F2FS_SET_FEATURE(sbi, mask) \ 175 (sbi->raw_super->feature |= cpu_to_le32(mask)) 176 #define F2FS_CLEAR_FEATURE(sbi, mask) \ 177 (sbi->raw_super->feature &= ~cpu_to_le32(mask)) 178 179 /* 180 * Default values for user and/or group using reserved blocks 181 */ 182 #define F2FS_DEF_RESUID 0 183 #define F2FS_DEF_RESGID 0 184 185 /* 186 * For checkpoint manager 187 */ 188 enum { 189 NAT_BITMAP, 190 SIT_BITMAP 191 }; 192 193 #define CP_UMOUNT 0x00000001 194 #define CP_FASTBOOT 0x00000002 195 #define CP_SYNC 0x00000004 196 #define CP_RECOVERY 0x00000008 197 #define CP_DISCARD 0x00000010 198 #define CP_TRIMMED 0x00000020 199 #define CP_PAUSE 0x00000040 200 #define CP_RESIZE 0x00000080 201 202 #define MAX_DISCARD_BLOCKS(sbi) BLKS_PER_SEC(sbi) 203 #define DEF_MAX_DISCARD_REQUEST 8 /* issue 8 discards per round */ 204 #define DEF_MIN_DISCARD_ISSUE_TIME 50 /* 50 ms, if exists */ 205 #define DEF_MID_DISCARD_ISSUE_TIME 500 /* 500 ms, if device busy */ 206 #define DEF_MAX_DISCARD_ISSUE_TIME 60000 /* 60 s, if no candidates */ 207 #define DEF_DISCARD_URGENT_UTIL 80 /* do more discard over 80% */ 208 #define DEF_CP_INTERVAL 60 /* 60 secs */ 209 #define DEF_IDLE_INTERVAL 5 /* 5 secs */ 210 #define DEF_DISABLE_INTERVAL 5 /* 5 secs */ 211 #define DEF_DISABLE_QUICK_INTERVAL 1 /* 1 secs */ 212 #define DEF_UMOUNT_DISCARD_TIMEOUT 5 /* 5 secs */ 213 214 struct cp_control { 215 int reason; 216 __u64 trim_start; 217 __u64 trim_end; 218 __u64 trim_minlen; 219 }; 220 221 /* 222 * indicate meta/data type 223 */ 224 enum { 225 META_CP, 226 META_NAT, 227 META_SIT, 228 META_SSA, 229 META_MAX, 230 META_POR, 231 DATA_GENERIC, /* check range only */ 232 DATA_GENERIC_ENHANCE, /* strong check on range and segment bitmap */ 233 DATA_GENERIC_ENHANCE_READ, /* 234 * strong check on range and segment 235 * bitmap but no warning due to race 236 * condition of read on truncated area 237 * by extent_cache 238 */ 239 META_GENERIC, 240 }; 241 242 /* for the list of ino */ 243 enum { 244 ORPHAN_INO, /* for orphan ino list */ 245 APPEND_INO, /* for append ino list */ 246 UPDATE_INO, /* for update ino list */ 247 TRANS_DIR_INO, /* for trasactions dir ino list */ 248 FLUSH_INO, /* for multiple device flushing */ 249 MAX_INO_ENTRY, /* max. list */ 250 }; 251 252 struct ino_entry { 253 struct list_head list; /* list head */ 254 nid_t ino; /* inode number */ 255 unsigned int dirty_device; /* dirty device bitmap */ 256 }; 257 258 /* for the list of inodes to be GCed */ 259 struct inode_entry { 260 struct list_head list; /* list head */ 261 struct inode *inode; /* vfs inode pointer */ 262 }; 263 264 struct fsync_node_entry { 265 struct list_head list; /* list head */ 266 struct page *page; /* warm node page pointer */ 267 unsigned int seq_id; /* sequence id */ 268 }; 269 270 struct ckpt_req { 271 struct completion wait; /* completion for checkpoint done */ 272 struct llist_node llnode; /* llist_node to be linked in wait queue */ 273 int ret; /* return code of checkpoint */ 274 ktime_t queue_time; /* request queued time */ 275 }; 276 277 struct ckpt_req_control { 278 struct task_struct *f2fs_issue_ckpt; /* checkpoint task */ 279 int ckpt_thread_ioprio; /* checkpoint merge thread ioprio */ 280 wait_queue_head_t ckpt_wait_queue; /* waiting queue for wake-up */ 281 atomic_t issued_ckpt; /* # of actually issued ckpts */ 282 atomic_t total_ckpt; /* # of total ckpts */ 283 atomic_t queued_ckpt; /* # of queued ckpts */ 284 struct llist_head issue_list; /* list for command issue */ 285 spinlock_t stat_lock; /* lock for below checkpoint time stats */ 286 unsigned int cur_time; /* cur wait time in msec for currently issued checkpoint */ 287 unsigned int peak_time; /* peak wait time in msec until now */ 288 }; 289 290 /* for the bitmap indicate blocks to be discarded */ 291 struct discard_entry { 292 struct list_head list; /* list head */ 293 block_t start_blkaddr; /* start blockaddr of current segment */ 294 unsigned char discard_map[SIT_VBLOCK_MAP_SIZE]; /* segment discard bitmap */ 295 }; 296 297 /* default discard granularity of inner discard thread, unit: block count */ 298 #define DEFAULT_DISCARD_GRANULARITY 16 299 300 /* max discard pend list number */ 301 #define MAX_PLIST_NUM 512 302 #define plist_idx(blk_num) ((blk_num) >= MAX_PLIST_NUM ? \ 303 (MAX_PLIST_NUM - 1) : ((blk_num) - 1)) 304 305 enum { 306 D_PREP, /* initial */ 307 D_PARTIAL, /* partially submitted */ 308 D_SUBMIT, /* all submitted */ 309 D_DONE, /* finished */ 310 }; 311 312 struct discard_info { 313 block_t lstart; /* logical start address */ 314 block_t len; /* length */ 315 block_t start; /* actual start address in dev */ 316 }; 317 318 struct discard_cmd { 319 struct rb_node rb_node; /* rb node located in rb-tree */ 320 union { 321 struct { 322 block_t lstart; /* logical start address */ 323 block_t len; /* length */ 324 block_t start; /* actual start address in dev */ 325 }; 326 struct discard_info di; /* discard info */ 327 328 }; 329 struct list_head list; /* command list */ 330 struct completion wait; /* compleation */ 331 struct block_device *bdev; /* bdev */ 332 unsigned short ref; /* reference count */ 333 unsigned char state; /* state */ 334 unsigned char queued; /* queued discard */ 335 int error; /* bio error */ 336 spinlock_t lock; /* for state/bio_ref updating */ 337 unsigned short bio_ref; /* bio reference count */ 338 }; 339 340 enum { 341 DPOLICY_BG, 342 DPOLICY_FORCE, 343 DPOLICY_FSTRIM, 344 DPOLICY_UMOUNT, 345 MAX_DPOLICY, 346 }; 347 348 struct discard_policy { 349 int type; /* type of discard */ 350 unsigned int min_interval; /* used for candidates exist */ 351 unsigned int mid_interval; /* used for device busy */ 352 unsigned int max_interval; /* used for candidates not exist */ 353 unsigned int max_requests; /* # of discards issued per round */ 354 unsigned int io_aware_gran; /* minimum granularity discard not be aware of I/O */ 355 bool io_aware; /* issue discard in idle time */ 356 bool sync; /* submit discard with REQ_SYNC flag */ 357 bool ordered; /* issue discard by lba order */ 358 bool timeout; /* discard timeout for put_super */ 359 unsigned int granularity; /* discard granularity */ 360 }; 361 362 struct discard_cmd_control { 363 struct task_struct *f2fs_issue_discard; /* discard thread */ 364 struct list_head entry_list; /* 4KB discard entry list */ 365 struct list_head pend_list[MAX_PLIST_NUM];/* store pending entries */ 366 struct list_head wait_list; /* store on-flushing entries */ 367 struct list_head fstrim_list; /* in-flight discard from fstrim */ 368 wait_queue_head_t discard_wait_queue; /* waiting queue for wake-up */ 369 unsigned int discard_wake; /* to wake up discard thread */ 370 struct mutex cmd_lock; 371 unsigned int nr_discards; /* # of discards in the list */ 372 unsigned int max_discards; /* max. discards to be issued */ 373 unsigned int discard_granularity; /* discard granularity */ 374 unsigned int undiscard_blks; /* # of undiscard blocks */ 375 unsigned int next_pos; /* next discard position */ 376 atomic_t issued_discard; /* # of issued discard */ 377 atomic_t queued_discard; /* # of queued discard */ 378 atomic_t discard_cmd_cnt; /* # of cached cmd count */ 379 struct rb_root_cached root; /* root of discard rb-tree */ 380 bool rbtree_check; /* config for consistence check */ 381 }; 382 383 /* for the list of fsync inodes, used only during recovery */ 384 struct fsync_inode_entry { 385 struct list_head list; /* list head */ 386 struct inode *inode; /* vfs inode pointer */ 387 block_t blkaddr; /* block address locating the last fsync */ 388 block_t last_dentry; /* block address locating the last dentry */ 389 }; 390 391 #define nats_in_cursum(jnl) (le16_to_cpu((jnl)->n_nats)) 392 #define sits_in_cursum(jnl) (le16_to_cpu((jnl)->n_sits)) 393 394 #define nat_in_journal(jnl, i) ((jnl)->nat_j.entries[i].ne) 395 #define nid_in_journal(jnl, i) ((jnl)->nat_j.entries[i].nid) 396 #define sit_in_journal(jnl, i) ((jnl)->sit_j.entries[i].se) 397 #define segno_in_journal(jnl, i) ((jnl)->sit_j.entries[i].segno) 398 399 #define MAX_NAT_JENTRIES(jnl) (NAT_JOURNAL_ENTRIES - nats_in_cursum(jnl)) 400 #define MAX_SIT_JENTRIES(jnl) (SIT_JOURNAL_ENTRIES - sits_in_cursum(jnl)) 401 402 static inline int update_nats_in_cursum(struct f2fs_journal *journal, int i) 403 { 404 int before = nats_in_cursum(journal); 405 406 journal->n_nats = cpu_to_le16(before + i); 407 return before; 408 } 409 410 static inline int update_sits_in_cursum(struct f2fs_journal *journal, int i) 411 { 412 int before = sits_in_cursum(journal); 413 414 journal->n_sits = cpu_to_le16(before + i); 415 return before; 416 } 417 418 static inline bool __has_cursum_space(struct f2fs_journal *journal, 419 int size, int type) 420 { 421 if (type == NAT_JOURNAL) 422 return size <= MAX_NAT_JENTRIES(journal); 423 return size <= MAX_SIT_JENTRIES(journal); 424 } 425 426 /* for inline stuff */ 427 #define DEF_INLINE_RESERVED_SIZE 1 428 static inline int get_extra_isize(struct inode *inode); 429 static inline int get_inline_xattr_addrs(struct inode *inode); 430 #define MAX_INLINE_DATA(inode) (sizeof(__le32) * \ 431 (CUR_ADDRS_PER_INODE(inode) - \ 432 get_inline_xattr_addrs(inode) - \ 433 DEF_INLINE_RESERVED_SIZE)) 434 435 /* for inline dir */ 436 #define NR_INLINE_DENTRY(inode) (MAX_INLINE_DATA(inode) * BITS_PER_BYTE / \ 437 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \ 438 BITS_PER_BYTE + 1)) 439 #define INLINE_DENTRY_BITMAP_SIZE(inode) \ 440 DIV_ROUND_UP(NR_INLINE_DENTRY(inode), BITS_PER_BYTE) 441 #define INLINE_RESERVED_SIZE(inode) (MAX_INLINE_DATA(inode) - \ 442 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \ 443 NR_INLINE_DENTRY(inode) + \ 444 INLINE_DENTRY_BITMAP_SIZE(inode))) 445 446 /* 447 * For INODE and NODE manager 448 */ 449 /* for directory operations */ 450 451 struct f2fs_filename { 452 /* 453 * The filename the user specified. This is NULL for some 454 * filesystem-internal operations, e.g. converting an inline directory 455 * to a non-inline one, or roll-forward recovering an encrypted dentry. 456 */ 457 const struct qstr *usr_fname; 458 459 /* 460 * The on-disk filename. For encrypted directories, this is encrypted. 461 * This may be NULL for lookups in an encrypted dir without the key. 462 */ 463 struct fscrypt_str disk_name; 464 465 /* The dirhash of this filename */ 466 f2fs_hash_t hash; 467 468 #ifdef CONFIG_FS_ENCRYPTION 469 /* 470 * For lookups in encrypted directories: either the buffer backing 471 * disk_name, or a buffer that holds the decoded no-key name. 472 */ 473 struct fscrypt_str crypto_buf; 474 #endif 475 #ifdef CONFIG_UNICODE 476 /* 477 * For casefolded directories: the casefolded name, but it's left NULL 478 * if the original name is not valid Unicode, if the directory is both 479 * casefolded and encrypted and its encryption key is unavailable, or if 480 * the filesystem is doing an internal operation where usr_fname is also 481 * NULL. In all these cases we fall back to treating the name as an 482 * opaque byte sequence. 483 */ 484 struct fscrypt_str cf_name; 485 #endif 486 }; 487 488 struct f2fs_dentry_ptr { 489 struct inode *inode; 490 void *bitmap; 491 struct f2fs_dir_entry *dentry; 492 __u8 (*filename)[F2FS_SLOT_LEN]; 493 int max; 494 int nr_bitmap; 495 }; 496 497 static inline void make_dentry_ptr_block(struct inode *inode, 498 struct f2fs_dentry_ptr *d, struct f2fs_dentry_block *t) 499 { 500 d->inode = inode; 501 d->max = NR_DENTRY_IN_BLOCK; 502 d->nr_bitmap = SIZE_OF_DENTRY_BITMAP; 503 d->bitmap = t->dentry_bitmap; 504 d->dentry = t->dentry; 505 d->filename = t->filename; 506 } 507 508 static inline void make_dentry_ptr_inline(struct inode *inode, 509 struct f2fs_dentry_ptr *d, void *t) 510 { 511 int entry_cnt = NR_INLINE_DENTRY(inode); 512 int bitmap_size = INLINE_DENTRY_BITMAP_SIZE(inode); 513 int reserved_size = INLINE_RESERVED_SIZE(inode); 514 515 d->inode = inode; 516 d->max = entry_cnt; 517 d->nr_bitmap = bitmap_size; 518 d->bitmap = t; 519 d->dentry = t + bitmap_size + reserved_size; 520 d->filename = t + bitmap_size + reserved_size + 521 SIZE_OF_DIR_ENTRY * entry_cnt; 522 } 523 524 /* 525 * XATTR_NODE_OFFSET stores xattrs to one node block per file keeping -1 526 * as its node offset to distinguish from index node blocks. 527 * But some bits are used to mark the node block. 528 */ 529 #define XATTR_NODE_OFFSET ((((unsigned int)-1) << OFFSET_BIT_SHIFT) \ 530 >> OFFSET_BIT_SHIFT) 531 enum { 532 ALLOC_NODE, /* allocate a new node page if needed */ 533 LOOKUP_NODE, /* look up a node without readahead */ 534 LOOKUP_NODE_RA, /* 535 * look up a node with readahead called 536 * by get_data_block. 537 */ 538 }; 539 540 #define DEFAULT_RETRY_IO_COUNT 8 /* maximum retry read IO count */ 541 542 /* congestion wait timeout value, default: 20ms */ 543 #define DEFAULT_IO_TIMEOUT (msecs_to_jiffies(20)) 544 545 /* maximum retry quota flush count */ 546 #define DEFAULT_RETRY_QUOTA_FLUSH_COUNT 8 547 548 #define F2FS_LINK_MAX 0xffffffff /* maximum link count per file */ 549 550 #define MAX_DIR_RA_PAGES 4 /* maximum ra pages of dir */ 551 552 /* for in-memory extent cache entry */ 553 #define F2FS_MIN_EXTENT_LEN 64 /* minimum extent length */ 554 555 /* number of extent info in extent cache we try to shrink */ 556 #define EXTENT_CACHE_SHRINK_NUMBER 128 557 558 struct rb_entry { 559 struct rb_node rb_node; /* rb node located in rb-tree */ 560 union { 561 struct { 562 unsigned int ofs; /* start offset of the entry */ 563 unsigned int len; /* length of the entry */ 564 }; 565 unsigned long long key; /* 64-bits key */ 566 } __packed; 567 }; 568 569 struct extent_info { 570 unsigned int fofs; /* start offset in a file */ 571 unsigned int len; /* length of the extent */ 572 u32 blk; /* start block address of the extent */ 573 }; 574 575 struct extent_node { 576 struct rb_node rb_node; /* rb node located in rb-tree */ 577 struct extent_info ei; /* extent info */ 578 struct list_head list; /* node in global extent list of sbi */ 579 struct extent_tree *et; /* extent tree pointer */ 580 }; 581 582 struct extent_tree { 583 nid_t ino; /* inode number */ 584 struct rb_root_cached root; /* root of extent info rb-tree */ 585 struct extent_node *cached_en; /* recently accessed extent node */ 586 struct extent_info largest; /* largested extent info */ 587 struct list_head list; /* to be used by sbi->zombie_list */ 588 rwlock_t lock; /* protect extent info rb-tree */ 589 atomic_t node_cnt; /* # of extent node in rb-tree*/ 590 bool largest_updated; /* largest extent updated */ 591 }; 592 593 /* 594 * This structure is taken from ext4_map_blocks. 595 * 596 * Note that, however, f2fs uses NEW and MAPPED flags for f2fs_map_blocks(). 597 */ 598 #define F2FS_MAP_NEW (1 << BH_New) 599 #define F2FS_MAP_MAPPED (1 << BH_Mapped) 600 #define F2FS_MAP_UNWRITTEN (1 << BH_Unwritten) 601 #define F2FS_MAP_FLAGS (F2FS_MAP_NEW | F2FS_MAP_MAPPED |\ 602 F2FS_MAP_UNWRITTEN) 603 604 struct f2fs_map_blocks { 605 block_t m_pblk; 606 block_t m_lblk; 607 unsigned int m_len; 608 unsigned int m_flags; 609 pgoff_t *m_next_pgofs; /* point next possible non-hole pgofs */ 610 pgoff_t *m_next_extent; /* point to next possible extent */ 611 int m_seg_type; 612 bool m_may_create; /* indicate it is from write path */ 613 }; 614 615 /* for flag in get_data_block */ 616 enum { 617 F2FS_GET_BLOCK_DEFAULT, 618 F2FS_GET_BLOCK_FIEMAP, 619 F2FS_GET_BLOCK_BMAP, 620 F2FS_GET_BLOCK_DIO, 621 F2FS_GET_BLOCK_PRE_DIO, 622 F2FS_GET_BLOCK_PRE_AIO, 623 F2FS_GET_BLOCK_PRECACHE, 624 }; 625 626 /* 627 * i_advise uses FADVISE_XXX_BIT. We can add additional hints later. 628 */ 629 #define FADVISE_COLD_BIT 0x01 630 #define FADVISE_LOST_PINO_BIT 0x02 631 #define FADVISE_ENCRYPT_BIT 0x04 632 #define FADVISE_ENC_NAME_BIT 0x08 633 #define FADVISE_KEEP_SIZE_BIT 0x10 634 #define FADVISE_HOT_BIT 0x20 635 #define FADVISE_VERITY_BIT 0x40 636 637 #define FADVISE_MODIFIABLE_BITS (FADVISE_COLD_BIT | FADVISE_HOT_BIT) 638 639 #define file_is_cold(inode) is_file(inode, FADVISE_COLD_BIT) 640 #define file_wrong_pino(inode) is_file(inode, FADVISE_LOST_PINO_BIT) 641 #define file_set_cold(inode) set_file(inode, FADVISE_COLD_BIT) 642 #define file_lost_pino(inode) set_file(inode, FADVISE_LOST_PINO_BIT) 643 #define file_clear_cold(inode) clear_file(inode, FADVISE_COLD_BIT) 644 #define file_got_pino(inode) clear_file(inode, FADVISE_LOST_PINO_BIT) 645 #define file_is_encrypt(inode) is_file(inode, FADVISE_ENCRYPT_BIT) 646 #define file_set_encrypt(inode) set_file(inode, FADVISE_ENCRYPT_BIT) 647 #define file_clear_encrypt(inode) clear_file(inode, FADVISE_ENCRYPT_BIT) 648 #define file_enc_name(inode) is_file(inode, FADVISE_ENC_NAME_BIT) 649 #define file_set_enc_name(inode) set_file(inode, FADVISE_ENC_NAME_BIT) 650 #define file_keep_isize(inode) is_file(inode, FADVISE_KEEP_SIZE_BIT) 651 #define file_set_keep_isize(inode) set_file(inode, FADVISE_KEEP_SIZE_BIT) 652 #define file_is_hot(inode) is_file(inode, FADVISE_HOT_BIT) 653 #define file_set_hot(inode) set_file(inode, FADVISE_HOT_BIT) 654 #define file_clear_hot(inode) clear_file(inode, FADVISE_HOT_BIT) 655 #define file_is_verity(inode) is_file(inode, FADVISE_VERITY_BIT) 656 #define file_set_verity(inode) set_file(inode, FADVISE_VERITY_BIT) 657 658 #define DEF_DIR_LEVEL 0 659 660 enum { 661 GC_FAILURE_PIN, 662 GC_FAILURE_ATOMIC, 663 MAX_GC_FAILURE 664 }; 665 666 /* used for f2fs_inode_info->flags */ 667 enum { 668 FI_NEW_INODE, /* indicate newly allocated inode */ 669 FI_DIRTY_INODE, /* indicate inode is dirty or not */ 670 FI_AUTO_RECOVER, /* indicate inode is recoverable */ 671 FI_DIRTY_DIR, /* indicate directory has dirty pages */ 672 FI_INC_LINK, /* need to increment i_nlink */ 673 FI_ACL_MODE, /* indicate acl mode */ 674 FI_NO_ALLOC, /* should not allocate any blocks */ 675 FI_FREE_NID, /* free allocated nide */ 676 FI_NO_EXTENT, /* not to use the extent cache */ 677 FI_INLINE_XATTR, /* used for inline xattr */ 678 FI_INLINE_DATA, /* used for inline data*/ 679 FI_INLINE_DENTRY, /* used for inline dentry */ 680 FI_APPEND_WRITE, /* inode has appended data */ 681 FI_UPDATE_WRITE, /* inode has in-place-update data */ 682 FI_NEED_IPU, /* used for ipu per file */ 683 FI_ATOMIC_FILE, /* indicate atomic file */ 684 FI_ATOMIC_COMMIT, /* indicate the state of atomical committing */ 685 FI_VOLATILE_FILE, /* indicate volatile file */ 686 FI_FIRST_BLOCK_WRITTEN, /* indicate #0 data block was written */ 687 FI_DROP_CACHE, /* drop dirty page cache */ 688 FI_DATA_EXIST, /* indicate data exists */ 689 FI_INLINE_DOTS, /* indicate inline dot dentries */ 690 FI_DO_DEFRAG, /* indicate defragment is running */ 691 FI_DIRTY_FILE, /* indicate regular/symlink has dirty pages */ 692 FI_NO_PREALLOC, /* indicate skipped preallocated blocks */ 693 FI_HOT_DATA, /* indicate file is hot */ 694 FI_EXTRA_ATTR, /* indicate file has extra attribute */ 695 FI_PROJ_INHERIT, /* indicate file inherits projectid */ 696 FI_PIN_FILE, /* indicate file should not be gced */ 697 FI_ATOMIC_REVOKE_REQUEST, /* request to drop atomic data */ 698 FI_VERITY_IN_PROGRESS, /* building fs-verity Merkle tree */ 699 FI_COMPRESSED_FILE, /* indicate file's data can be compressed */ 700 FI_COMPRESS_CORRUPT, /* indicate compressed cluster is corrupted */ 701 FI_MMAP_FILE, /* indicate file was mmapped */ 702 FI_ENABLE_COMPRESS, /* enable compression in "user" compression mode */ 703 FI_MAX, /* max flag, never be used */ 704 }; 705 706 struct f2fs_inode_info { 707 struct inode vfs_inode; /* serve a vfs inode */ 708 unsigned long i_flags; /* keep an inode flags for ioctl */ 709 unsigned char i_advise; /* use to give file attribute hints */ 710 unsigned char i_dir_level; /* use for dentry level for large dir */ 711 unsigned int i_current_depth; /* only for directory depth */ 712 /* for gc failure statistic */ 713 unsigned int i_gc_failures[MAX_GC_FAILURE]; 714 unsigned int i_pino; /* parent inode number */ 715 umode_t i_acl_mode; /* keep file acl mode temporarily */ 716 717 /* Use below internally in f2fs*/ 718 unsigned long flags[BITS_TO_LONGS(FI_MAX)]; /* use to pass per-file flags */ 719 struct rw_semaphore i_sem; /* protect fi info */ 720 atomic_t dirty_pages; /* # of dirty pages */ 721 f2fs_hash_t chash; /* hash value of given file name */ 722 unsigned int clevel; /* maximum level of given file name */ 723 struct task_struct *task; /* lookup and create consistency */ 724 struct task_struct *cp_task; /* separate cp/wb IO stats*/ 725 nid_t i_xattr_nid; /* node id that contains xattrs */ 726 loff_t last_disk_size; /* lastly written file size */ 727 spinlock_t i_size_lock; /* protect last_disk_size */ 728 729 #ifdef CONFIG_QUOTA 730 struct dquot *i_dquot[MAXQUOTAS]; 731 732 /* quota space reservation, managed internally by quota code */ 733 qsize_t i_reserved_quota; 734 #endif 735 struct list_head dirty_list; /* dirty list for dirs and files */ 736 struct list_head gdirty_list; /* linked in global dirty list */ 737 struct list_head inmem_ilist; /* list for inmem inodes */ 738 struct list_head inmem_pages; /* inmemory pages managed by f2fs */ 739 struct task_struct *inmem_task; /* store inmemory task */ 740 struct mutex inmem_lock; /* lock for inmemory pages */ 741 struct extent_tree *extent_tree; /* cached extent_tree entry */ 742 743 /* avoid racing between foreground op and gc */ 744 struct rw_semaphore i_gc_rwsem[2]; 745 struct rw_semaphore i_mmap_sem; 746 struct rw_semaphore i_xattr_sem; /* avoid racing between reading and changing EAs */ 747 748 int i_extra_isize; /* size of extra space located in i_addr */ 749 kprojid_t i_projid; /* id for project quota */ 750 int i_inline_xattr_size; /* inline xattr size */ 751 struct timespec64 i_crtime; /* inode creation time */ 752 struct timespec64 i_disk_time[4];/* inode disk times */ 753 754 /* for file compress */ 755 atomic_t i_compr_blocks; /* # of compressed blocks */ 756 unsigned char i_compress_algorithm; /* algorithm type */ 757 unsigned char i_log_cluster_size; /* log of cluster size */ 758 unsigned char i_compress_level; /* compress level (lz4hc,zstd) */ 759 unsigned short i_compress_flag; /* compress flag */ 760 unsigned int i_cluster_size; /* cluster size */ 761 }; 762 763 static inline void get_extent_info(struct extent_info *ext, 764 struct f2fs_extent *i_ext) 765 { 766 ext->fofs = le32_to_cpu(i_ext->fofs); 767 ext->blk = le32_to_cpu(i_ext->blk); 768 ext->len = le32_to_cpu(i_ext->len); 769 } 770 771 static inline void set_raw_extent(struct extent_info *ext, 772 struct f2fs_extent *i_ext) 773 { 774 i_ext->fofs = cpu_to_le32(ext->fofs); 775 i_ext->blk = cpu_to_le32(ext->blk); 776 i_ext->len = cpu_to_le32(ext->len); 777 } 778 779 static inline void set_extent_info(struct extent_info *ei, unsigned int fofs, 780 u32 blk, unsigned int len) 781 { 782 ei->fofs = fofs; 783 ei->blk = blk; 784 ei->len = len; 785 } 786 787 static inline bool __is_discard_mergeable(struct discard_info *back, 788 struct discard_info *front, unsigned int max_len) 789 { 790 return (back->lstart + back->len == front->lstart) && 791 (back->len + front->len <= max_len); 792 } 793 794 static inline bool __is_discard_back_mergeable(struct discard_info *cur, 795 struct discard_info *back, unsigned int max_len) 796 { 797 return __is_discard_mergeable(back, cur, max_len); 798 } 799 800 static inline bool __is_discard_front_mergeable(struct discard_info *cur, 801 struct discard_info *front, unsigned int max_len) 802 { 803 return __is_discard_mergeable(cur, front, max_len); 804 } 805 806 static inline bool __is_extent_mergeable(struct extent_info *back, 807 struct extent_info *front) 808 { 809 return (back->fofs + back->len == front->fofs && 810 back->blk + back->len == front->blk); 811 } 812 813 static inline bool __is_back_mergeable(struct extent_info *cur, 814 struct extent_info *back) 815 { 816 return __is_extent_mergeable(back, cur); 817 } 818 819 static inline bool __is_front_mergeable(struct extent_info *cur, 820 struct extent_info *front) 821 { 822 return __is_extent_mergeable(cur, front); 823 } 824 825 extern void f2fs_mark_inode_dirty_sync(struct inode *inode, bool sync); 826 static inline void __try_update_largest_extent(struct extent_tree *et, 827 struct extent_node *en) 828 { 829 if (en->ei.len > et->largest.len) { 830 et->largest = en->ei; 831 et->largest_updated = true; 832 } 833 } 834 835 /* 836 * For free nid management 837 */ 838 enum nid_state { 839 FREE_NID, /* newly added to free nid list */ 840 PREALLOC_NID, /* it is preallocated */ 841 MAX_NID_STATE, 842 }; 843 844 enum nat_state { 845 TOTAL_NAT, 846 DIRTY_NAT, 847 RECLAIMABLE_NAT, 848 MAX_NAT_STATE, 849 }; 850 851 struct f2fs_nm_info { 852 block_t nat_blkaddr; /* base disk address of NAT */ 853 nid_t max_nid; /* maximum possible node ids */ 854 nid_t available_nids; /* # of available node ids */ 855 nid_t next_scan_nid; /* the next nid to be scanned */ 856 unsigned int ram_thresh; /* control the memory footprint */ 857 unsigned int ra_nid_pages; /* # of nid pages to be readaheaded */ 858 unsigned int dirty_nats_ratio; /* control dirty nats ratio threshold */ 859 860 /* NAT cache management */ 861 struct radix_tree_root nat_root;/* root of the nat entry cache */ 862 struct radix_tree_root nat_set_root;/* root of the nat set cache */ 863 struct rw_semaphore nat_tree_lock; /* protect nat_tree_lock */ 864 struct list_head nat_entries; /* cached nat entry list (clean) */ 865 spinlock_t nat_list_lock; /* protect clean nat entry list */ 866 unsigned int nat_cnt[MAX_NAT_STATE]; /* the # of cached nat entries */ 867 unsigned int nat_blocks; /* # of nat blocks */ 868 869 /* free node ids management */ 870 struct radix_tree_root free_nid_root;/* root of the free_nid cache */ 871 struct list_head free_nid_list; /* list for free nids excluding preallocated nids */ 872 unsigned int nid_cnt[MAX_NID_STATE]; /* the number of free node id */ 873 spinlock_t nid_list_lock; /* protect nid lists ops */ 874 struct mutex build_lock; /* lock for build free nids */ 875 unsigned char **free_nid_bitmap; 876 unsigned char *nat_block_bitmap; 877 unsigned short *free_nid_count; /* free nid count of NAT block */ 878 879 /* for checkpoint */ 880 char *nat_bitmap; /* NAT bitmap pointer */ 881 882 unsigned int nat_bits_blocks; /* # of nat bits blocks */ 883 unsigned char *nat_bits; /* NAT bits blocks */ 884 unsigned char *full_nat_bits; /* full NAT pages */ 885 unsigned char *empty_nat_bits; /* empty NAT pages */ 886 #ifdef CONFIG_F2FS_CHECK_FS 887 char *nat_bitmap_mir; /* NAT bitmap mirror */ 888 #endif 889 int bitmap_size; /* bitmap size */ 890 }; 891 892 /* 893 * this structure is used as one of function parameters. 894 * all the information are dedicated to a given direct node block determined 895 * by the data offset in a file. 896 */ 897 struct dnode_of_data { 898 struct inode *inode; /* vfs inode pointer */ 899 struct page *inode_page; /* its inode page, NULL is possible */ 900 struct page *node_page; /* cached direct node page */ 901 nid_t nid; /* node id of the direct node block */ 902 unsigned int ofs_in_node; /* data offset in the node page */ 903 bool inode_page_locked; /* inode page is locked or not */ 904 bool node_changed; /* is node block changed */ 905 char cur_level; /* level of hole node page */ 906 char max_level; /* level of current page located */ 907 block_t data_blkaddr; /* block address of the node block */ 908 }; 909 910 static inline void set_new_dnode(struct dnode_of_data *dn, struct inode *inode, 911 struct page *ipage, struct page *npage, nid_t nid) 912 { 913 memset(dn, 0, sizeof(*dn)); 914 dn->inode = inode; 915 dn->inode_page = ipage; 916 dn->node_page = npage; 917 dn->nid = nid; 918 } 919 920 /* 921 * For SIT manager 922 * 923 * By default, there are 6 active log areas across the whole main area. 924 * When considering hot and cold data separation to reduce cleaning overhead, 925 * we split 3 for data logs and 3 for node logs as hot, warm, and cold types, 926 * respectively. 927 * In the current design, you should not change the numbers intentionally. 928 * Instead, as a mount option such as active_logs=x, you can use 2, 4, and 6 929 * logs individually according to the underlying devices. (default: 6) 930 * Just in case, on-disk layout covers maximum 16 logs that consist of 8 for 931 * data and 8 for node logs. 932 */ 933 #define NR_CURSEG_DATA_TYPE (3) 934 #define NR_CURSEG_NODE_TYPE (3) 935 #define NR_CURSEG_INMEM_TYPE (2) 936 #define NR_CURSEG_PERSIST_TYPE (NR_CURSEG_DATA_TYPE + NR_CURSEG_NODE_TYPE) 937 #define NR_CURSEG_TYPE (NR_CURSEG_INMEM_TYPE + NR_CURSEG_PERSIST_TYPE) 938 939 enum { 940 CURSEG_HOT_DATA = 0, /* directory entry blocks */ 941 CURSEG_WARM_DATA, /* data blocks */ 942 CURSEG_COLD_DATA, /* multimedia or GCed data blocks */ 943 CURSEG_HOT_NODE, /* direct node blocks of directory files */ 944 CURSEG_WARM_NODE, /* direct node blocks of normal files */ 945 CURSEG_COLD_NODE, /* indirect node blocks */ 946 NR_PERSISTENT_LOG, /* number of persistent log */ 947 CURSEG_COLD_DATA_PINNED = NR_PERSISTENT_LOG, 948 /* pinned file that needs consecutive block address */ 949 CURSEG_ALL_DATA_ATGC, /* SSR alloctor in hot/warm/cold data area */ 950 NO_CHECK_TYPE, /* number of persistent & inmem log */ 951 }; 952 953 struct flush_cmd { 954 struct completion wait; 955 struct llist_node llnode; 956 nid_t ino; 957 int ret; 958 }; 959 960 struct flush_cmd_control { 961 struct task_struct *f2fs_issue_flush; /* flush thread */ 962 wait_queue_head_t flush_wait_queue; /* waiting queue for wake-up */ 963 atomic_t issued_flush; /* # of issued flushes */ 964 atomic_t queued_flush; /* # of queued flushes */ 965 struct llist_head issue_list; /* list for command issue */ 966 struct llist_node *dispatch_list; /* list for command dispatch */ 967 }; 968 969 struct f2fs_sm_info { 970 struct sit_info *sit_info; /* whole segment information */ 971 struct free_segmap_info *free_info; /* free segment information */ 972 struct dirty_seglist_info *dirty_info; /* dirty segment information */ 973 struct curseg_info *curseg_array; /* active segment information */ 974 975 struct rw_semaphore curseg_lock; /* for preventing curseg change */ 976 977 block_t seg0_blkaddr; /* block address of 0'th segment */ 978 block_t main_blkaddr; /* start block address of main area */ 979 block_t ssa_blkaddr; /* start block address of SSA area */ 980 981 unsigned int segment_count; /* total # of segments */ 982 unsigned int main_segments; /* # of segments in main area */ 983 unsigned int reserved_segments; /* # of reserved segments */ 984 unsigned int ovp_segments; /* # of overprovision segments */ 985 986 /* a threshold to reclaim prefree segments */ 987 unsigned int rec_prefree_segments; 988 989 /* for batched trimming */ 990 unsigned int trim_sections; /* # of sections to trim */ 991 992 struct list_head sit_entry_set; /* sit entry set list */ 993 994 unsigned int ipu_policy; /* in-place-update policy */ 995 unsigned int min_ipu_util; /* in-place-update threshold */ 996 unsigned int min_fsync_blocks; /* threshold for fsync */ 997 unsigned int min_seq_blocks; /* threshold for sequential blocks */ 998 unsigned int min_hot_blocks; /* threshold for hot block allocation */ 999 unsigned int min_ssr_sections; /* threshold to trigger SSR allocation */ 1000 1001 /* for flush command control */ 1002 struct flush_cmd_control *fcc_info; 1003 1004 /* for discard command control */ 1005 struct discard_cmd_control *dcc_info; 1006 }; 1007 1008 /* 1009 * For superblock 1010 */ 1011 /* 1012 * COUNT_TYPE for monitoring 1013 * 1014 * f2fs monitors the number of several block types such as on-writeback, 1015 * dirty dentry blocks, dirty node blocks, and dirty meta blocks. 1016 */ 1017 #define WB_DATA_TYPE(p) (__is_cp_guaranteed(p) ? F2FS_WB_CP_DATA : F2FS_WB_DATA) 1018 enum count_type { 1019 F2FS_DIRTY_DENTS, 1020 F2FS_DIRTY_DATA, 1021 F2FS_DIRTY_QDATA, 1022 F2FS_DIRTY_NODES, 1023 F2FS_DIRTY_META, 1024 F2FS_INMEM_PAGES, 1025 F2FS_DIRTY_IMETA, 1026 F2FS_WB_CP_DATA, 1027 F2FS_WB_DATA, 1028 F2FS_RD_DATA, 1029 F2FS_RD_NODE, 1030 F2FS_RD_META, 1031 F2FS_DIO_WRITE, 1032 F2FS_DIO_READ, 1033 NR_COUNT_TYPE, 1034 }; 1035 1036 /* 1037 * The below are the page types of bios used in submit_bio(). 1038 * The available types are: 1039 * DATA User data pages. It operates as async mode. 1040 * NODE Node pages. It operates as async mode. 1041 * META FS metadata pages such as SIT, NAT, CP. 1042 * NR_PAGE_TYPE The number of page types. 1043 * META_FLUSH Make sure the previous pages are written 1044 * with waiting the bio's completion 1045 * ... Only can be used with META. 1046 */ 1047 #define PAGE_TYPE_OF_BIO(type) ((type) > META ? META : (type)) 1048 enum page_type { 1049 DATA, 1050 NODE, 1051 META, 1052 NR_PAGE_TYPE, 1053 META_FLUSH, 1054 INMEM, /* the below types are used by tracepoints only. */ 1055 INMEM_DROP, 1056 INMEM_INVALIDATE, 1057 INMEM_REVOKE, 1058 IPU, 1059 OPU, 1060 }; 1061 1062 enum temp_type { 1063 HOT = 0, /* must be zero for meta bio */ 1064 WARM, 1065 COLD, 1066 NR_TEMP_TYPE, 1067 }; 1068 1069 enum need_lock_type { 1070 LOCK_REQ = 0, 1071 LOCK_DONE, 1072 LOCK_RETRY, 1073 }; 1074 1075 enum cp_reason_type { 1076 CP_NO_NEEDED, 1077 CP_NON_REGULAR, 1078 CP_COMPRESSED, 1079 CP_HARDLINK, 1080 CP_SB_NEED_CP, 1081 CP_WRONG_PINO, 1082 CP_NO_SPC_ROLL, 1083 CP_NODE_NEED_CP, 1084 CP_FASTBOOT_MODE, 1085 CP_SPEC_LOG_NUM, 1086 CP_RECOVER_DIR, 1087 }; 1088 1089 enum iostat_type { 1090 /* WRITE IO */ 1091 APP_DIRECT_IO, /* app direct write IOs */ 1092 APP_BUFFERED_IO, /* app buffered write IOs */ 1093 APP_WRITE_IO, /* app write IOs */ 1094 APP_MAPPED_IO, /* app mapped IOs */ 1095 FS_DATA_IO, /* data IOs from kworker/fsync/reclaimer */ 1096 FS_NODE_IO, /* node IOs from kworker/fsync/reclaimer */ 1097 FS_META_IO, /* meta IOs from kworker/reclaimer */ 1098 FS_GC_DATA_IO, /* data IOs from forground gc */ 1099 FS_GC_NODE_IO, /* node IOs from forground gc */ 1100 FS_CP_DATA_IO, /* data IOs from checkpoint */ 1101 FS_CP_NODE_IO, /* node IOs from checkpoint */ 1102 FS_CP_META_IO, /* meta IOs from checkpoint */ 1103 1104 /* READ IO */ 1105 APP_DIRECT_READ_IO, /* app direct read IOs */ 1106 APP_BUFFERED_READ_IO, /* app buffered read IOs */ 1107 APP_READ_IO, /* app read IOs */ 1108 APP_MAPPED_READ_IO, /* app mapped read IOs */ 1109 FS_DATA_READ_IO, /* data read IOs */ 1110 FS_GDATA_READ_IO, /* data read IOs from background gc */ 1111 FS_CDATA_READ_IO, /* compressed data read IOs */ 1112 FS_NODE_READ_IO, /* node read IOs */ 1113 FS_META_READ_IO, /* meta read IOs */ 1114 1115 /* other */ 1116 FS_DISCARD, /* discard */ 1117 NR_IO_TYPE, 1118 }; 1119 1120 struct f2fs_io_info { 1121 struct f2fs_sb_info *sbi; /* f2fs_sb_info pointer */ 1122 nid_t ino; /* inode number */ 1123 enum page_type type; /* contains DATA/NODE/META/META_FLUSH */ 1124 enum temp_type temp; /* contains HOT/WARM/COLD */ 1125 int op; /* contains REQ_OP_ */ 1126 int op_flags; /* req_flag_bits */ 1127 block_t new_blkaddr; /* new block address to be written */ 1128 block_t old_blkaddr; /* old block address before Cow */ 1129 struct page *page; /* page to be written */ 1130 struct page *encrypted_page; /* encrypted page */ 1131 struct page *compressed_page; /* compressed page */ 1132 struct list_head list; /* serialize IOs */ 1133 bool submitted; /* indicate IO submission */ 1134 int need_lock; /* indicate we need to lock cp_rwsem */ 1135 bool in_list; /* indicate fio is in io_list */ 1136 bool is_por; /* indicate IO is from recovery or not */ 1137 bool retry; /* need to reallocate block address */ 1138 int compr_blocks; /* # of compressed block addresses */ 1139 bool encrypted; /* indicate file is encrypted */ 1140 enum iostat_type io_type; /* io type */ 1141 struct writeback_control *io_wbc; /* writeback control */ 1142 struct bio **bio; /* bio for ipu */ 1143 sector_t *last_block; /* last block number in bio */ 1144 unsigned char version; /* version of the node */ 1145 }; 1146 1147 struct bio_entry { 1148 struct bio *bio; 1149 struct list_head list; 1150 }; 1151 1152 #define is_read_io(rw) ((rw) == READ) 1153 struct f2fs_bio_info { 1154 struct f2fs_sb_info *sbi; /* f2fs superblock */ 1155 struct bio *bio; /* bios to merge */ 1156 sector_t last_block_in_bio; /* last block number */ 1157 struct f2fs_io_info fio; /* store buffered io info. */ 1158 struct rw_semaphore io_rwsem; /* blocking op for bio */ 1159 spinlock_t io_lock; /* serialize DATA/NODE IOs */ 1160 struct list_head io_list; /* track fios */ 1161 struct list_head bio_list; /* bio entry list head */ 1162 struct rw_semaphore bio_list_lock; /* lock to protect bio entry list */ 1163 }; 1164 1165 #define FDEV(i) (sbi->devs[i]) 1166 #define RDEV(i) (raw_super->devs[i]) 1167 struct f2fs_dev_info { 1168 struct block_device *bdev; 1169 char path[MAX_PATH_LEN]; 1170 unsigned int total_segments; 1171 block_t start_blk; 1172 block_t end_blk; 1173 #ifdef CONFIG_BLK_DEV_ZONED 1174 unsigned int nr_blkz; /* Total number of zones */ 1175 unsigned long *blkz_seq; /* Bitmap indicating sequential zones */ 1176 block_t *zone_capacity_blocks; /* Array of zone capacity in blks */ 1177 #endif 1178 }; 1179 1180 enum inode_type { 1181 DIR_INODE, /* for dirty dir inode */ 1182 FILE_INODE, /* for dirty regular/symlink inode */ 1183 DIRTY_META, /* for all dirtied inode metadata */ 1184 ATOMIC_FILE, /* for all atomic files */ 1185 NR_INODE_TYPE, 1186 }; 1187 1188 /* for inner inode cache management */ 1189 struct inode_management { 1190 struct radix_tree_root ino_root; /* ino entry array */ 1191 spinlock_t ino_lock; /* for ino entry lock */ 1192 struct list_head ino_list; /* inode list head */ 1193 unsigned long ino_num; /* number of entries */ 1194 }; 1195 1196 /* for GC_AT */ 1197 struct atgc_management { 1198 bool atgc_enabled; /* ATGC is enabled or not */ 1199 struct rb_root_cached root; /* root of victim rb-tree */ 1200 struct list_head victim_list; /* linked with all victim entries */ 1201 unsigned int victim_count; /* victim count in rb-tree */ 1202 unsigned int candidate_ratio; /* candidate ratio */ 1203 unsigned int max_candidate_count; /* max candidate count */ 1204 unsigned int age_weight; /* age weight, vblock_weight = 100 - age_weight */ 1205 unsigned long long age_threshold; /* age threshold */ 1206 }; 1207 1208 /* For s_flag in struct f2fs_sb_info */ 1209 enum { 1210 SBI_IS_DIRTY, /* dirty flag for checkpoint */ 1211 SBI_IS_CLOSE, /* specify unmounting */ 1212 SBI_NEED_FSCK, /* need fsck.f2fs to fix */ 1213 SBI_POR_DOING, /* recovery is doing or not */ 1214 SBI_NEED_SB_WRITE, /* need to recover superblock */ 1215 SBI_NEED_CP, /* need to checkpoint */ 1216 SBI_IS_SHUTDOWN, /* shutdown by ioctl */ 1217 SBI_IS_RECOVERED, /* recovered orphan/data */ 1218 SBI_CP_DISABLED, /* CP was disabled last mount */ 1219 SBI_CP_DISABLED_QUICK, /* CP was disabled quickly */ 1220 SBI_QUOTA_NEED_FLUSH, /* need to flush quota info in CP */ 1221 SBI_QUOTA_SKIP_FLUSH, /* skip flushing quota in current CP */ 1222 SBI_QUOTA_NEED_REPAIR, /* quota file may be corrupted */ 1223 SBI_IS_RESIZEFS, /* resizefs is in process */ 1224 }; 1225 1226 enum { 1227 CP_TIME, 1228 REQ_TIME, 1229 DISCARD_TIME, 1230 GC_TIME, 1231 DISABLE_TIME, 1232 UMOUNT_DISCARD_TIMEOUT, 1233 MAX_TIME, 1234 }; 1235 1236 enum { 1237 GC_NORMAL, 1238 GC_IDLE_CB, 1239 GC_IDLE_GREEDY, 1240 GC_IDLE_AT, 1241 GC_URGENT_HIGH, 1242 GC_URGENT_LOW, 1243 }; 1244 1245 enum { 1246 BGGC_MODE_ON, /* background gc is on */ 1247 BGGC_MODE_OFF, /* background gc is off */ 1248 BGGC_MODE_SYNC, /* 1249 * background gc is on, migrating blocks 1250 * like foreground gc 1251 */ 1252 }; 1253 1254 enum { 1255 FS_MODE_ADAPTIVE, /* use both lfs/ssr allocation */ 1256 FS_MODE_LFS, /* use lfs allocation only */ 1257 }; 1258 1259 enum { 1260 WHINT_MODE_OFF, /* not pass down write hints */ 1261 WHINT_MODE_USER, /* try to pass down hints given by users */ 1262 WHINT_MODE_FS, /* pass down hints with F2FS policy */ 1263 }; 1264 1265 enum { 1266 ALLOC_MODE_DEFAULT, /* stay default */ 1267 ALLOC_MODE_REUSE, /* reuse segments as much as possible */ 1268 }; 1269 1270 enum fsync_mode { 1271 FSYNC_MODE_POSIX, /* fsync follows posix semantics */ 1272 FSYNC_MODE_STRICT, /* fsync behaves in line with ext4 */ 1273 FSYNC_MODE_NOBARRIER, /* fsync behaves nobarrier based on posix */ 1274 }; 1275 1276 enum { 1277 COMPR_MODE_FS, /* 1278 * automatically compress compression 1279 * enabled files 1280 */ 1281 COMPR_MODE_USER, /* 1282 * automatical compression is disabled. 1283 * user can control the file compression 1284 * using ioctls 1285 */ 1286 }; 1287 1288 /* 1289 * this value is set in page as a private data which indicate that 1290 * the page is atomically written, and it is in inmem_pages list. 1291 */ 1292 #define ATOMIC_WRITTEN_PAGE ((unsigned long)-1) 1293 #define DUMMY_WRITTEN_PAGE ((unsigned long)-2) 1294 1295 #define IS_ATOMIC_WRITTEN_PAGE(page) \ 1296 (page_private(page) == ATOMIC_WRITTEN_PAGE) 1297 #define IS_DUMMY_WRITTEN_PAGE(page) \ 1298 (page_private(page) == DUMMY_WRITTEN_PAGE) 1299 1300 #ifdef CONFIG_F2FS_IO_TRACE 1301 #define IS_IO_TRACED_PAGE(page) \ 1302 (page_private(page) > 0 && \ 1303 page_private(page) < (unsigned long)PID_MAX_LIMIT) 1304 #else 1305 #define IS_IO_TRACED_PAGE(page) (0) 1306 #endif 1307 1308 /* For compression */ 1309 enum compress_algorithm_type { 1310 COMPRESS_LZO, 1311 COMPRESS_LZ4, 1312 COMPRESS_ZSTD, 1313 COMPRESS_LZORLE, 1314 COMPRESS_MAX, 1315 }; 1316 1317 enum compress_flag { 1318 COMPRESS_CHKSUM, 1319 COMPRESS_MAX_FLAG, 1320 }; 1321 1322 #define COMPRESS_DATA_RESERVED_SIZE 4 1323 struct compress_data { 1324 __le32 clen; /* compressed data size */ 1325 __le32 chksum; /* compressed data chksum */ 1326 __le32 reserved[COMPRESS_DATA_RESERVED_SIZE]; /* reserved */ 1327 u8 cdata[]; /* compressed data */ 1328 }; 1329 1330 #define COMPRESS_HEADER_SIZE (sizeof(struct compress_data)) 1331 1332 #define F2FS_COMPRESSED_PAGE_MAGIC 0xF5F2C000 1333 1334 #define COMPRESS_LEVEL_OFFSET 8 1335 1336 /* compress context */ 1337 struct compress_ctx { 1338 struct inode *inode; /* inode the context belong to */ 1339 pgoff_t cluster_idx; /* cluster index number */ 1340 unsigned int cluster_size; /* page count in cluster */ 1341 unsigned int log_cluster_size; /* log of cluster size */ 1342 struct page **rpages; /* pages store raw data in cluster */ 1343 unsigned int nr_rpages; /* total page number in rpages */ 1344 struct page **cpages; /* pages store compressed data in cluster */ 1345 unsigned int nr_cpages; /* total page number in cpages */ 1346 void *rbuf; /* virtual mapped address on rpages */ 1347 struct compress_data *cbuf; /* virtual mapped address on cpages */ 1348 size_t rlen; /* valid data length in rbuf */ 1349 size_t clen; /* valid data length in cbuf */ 1350 void *private; /* payload buffer for specified compression algorithm */ 1351 void *private2; /* extra payload buffer */ 1352 }; 1353 1354 /* compress context for write IO path */ 1355 struct compress_io_ctx { 1356 u32 magic; /* magic number to indicate page is compressed */ 1357 struct inode *inode; /* inode the context belong to */ 1358 struct page **rpages; /* pages store raw data in cluster */ 1359 unsigned int nr_rpages; /* total page number in rpages */ 1360 atomic_t pending_pages; /* in-flight compressed page count */ 1361 }; 1362 1363 /* Context for decompressing one cluster on the read IO path */ 1364 struct decompress_io_ctx { 1365 u32 magic; /* magic number to indicate page is compressed */ 1366 struct inode *inode; /* inode the context belong to */ 1367 pgoff_t cluster_idx; /* cluster index number */ 1368 unsigned int cluster_size; /* page count in cluster */ 1369 unsigned int log_cluster_size; /* log of cluster size */ 1370 struct page **rpages; /* pages store raw data in cluster */ 1371 unsigned int nr_rpages; /* total page number in rpages */ 1372 struct page **cpages; /* pages store compressed data in cluster */ 1373 unsigned int nr_cpages; /* total page number in cpages */ 1374 struct page **tpages; /* temp pages to pad holes in cluster */ 1375 void *rbuf; /* virtual mapped address on rpages */ 1376 struct compress_data *cbuf; /* virtual mapped address on cpages */ 1377 size_t rlen; /* valid data length in rbuf */ 1378 size_t clen; /* valid data length in cbuf */ 1379 1380 /* 1381 * The number of compressed pages remaining to be read in this cluster. 1382 * This is initially nr_cpages. It is decremented by 1 each time a page 1383 * has been read (or failed to be read). When it reaches 0, the cluster 1384 * is decompressed (or an error is reported). 1385 * 1386 * If an error occurs before all the pages have been submitted for I/O, 1387 * then this will never reach 0. In this case the I/O submitter is 1388 * responsible for calling f2fs_decompress_end_io() instead. 1389 */ 1390 atomic_t remaining_pages; 1391 1392 /* 1393 * Number of references to this decompress_io_ctx. 1394 * 1395 * One reference is held for I/O completion. This reference is dropped 1396 * after the pagecache pages are updated and unlocked -- either after 1397 * decompression (and verity if enabled), or after an error. 1398 * 1399 * In addition, each compressed page holds a reference while it is in a 1400 * bio. These references are necessary prevent compressed pages from 1401 * being freed while they are still in a bio. 1402 */ 1403 refcount_t refcnt; 1404 1405 bool failed; /* IO error occurred before decompression? */ 1406 bool need_verity; /* need fs-verity verification after decompression? */ 1407 void *private; /* payload buffer for specified decompression algorithm */ 1408 void *private2; /* extra payload buffer */ 1409 struct work_struct verity_work; /* work to verify the decompressed pages */ 1410 }; 1411 1412 #define NULL_CLUSTER ((unsigned int)(~0)) 1413 #define MIN_COMPRESS_LOG_SIZE 2 1414 #define MAX_COMPRESS_LOG_SIZE 8 1415 #define MAX_COMPRESS_WINDOW_SIZE(log_size) ((PAGE_SIZE) << (log_size)) 1416 1417 struct f2fs_sb_info { 1418 struct super_block *sb; /* pointer to VFS super block */ 1419 struct proc_dir_entry *s_proc; /* proc entry */ 1420 struct f2fs_super_block *raw_super; /* raw super block pointer */ 1421 struct rw_semaphore sb_lock; /* lock for raw super block */ 1422 int valid_super_block; /* valid super block no */ 1423 unsigned long s_flag; /* flags for sbi */ 1424 struct mutex writepages; /* mutex for writepages() */ 1425 1426 #ifdef CONFIG_BLK_DEV_ZONED 1427 unsigned int blocks_per_blkz; /* F2FS blocks per zone */ 1428 unsigned int log_blocks_per_blkz; /* log2 F2FS blocks per zone */ 1429 #endif 1430 1431 /* for node-related operations */ 1432 struct f2fs_nm_info *nm_info; /* node manager */ 1433 struct inode *node_inode; /* cache node blocks */ 1434 1435 /* for segment-related operations */ 1436 struct f2fs_sm_info *sm_info; /* segment manager */ 1437 1438 /* for bio operations */ 1439 struct f2fs_bio_info *write_io[NR_PAGE_TYPE]; /* for write bios */ 1440 /* keep migration IO order for LFS mode */ 1441 struct rw_semaphore io_order_lock; 1442 mempool_t *write_io_dummy; /* Dummy pages */ 1443 1444 /* for checkpoint */ 1445 struct f2fs_checkpoint *ckpt; /* raw checkpoint pointer */ 1446 int cur_cp_pack; /* remain current cp pack */ 1447 spinlock_t cp_lock; /* for flag in ckpt */ 1448 struct inode *meta_inode; /* cache meta blocks */ 1449 struct rw_semaphore cp_global_sem; /* checkpoint procedure lock */ 1450 struct rw_semaphore cp_rwsem; /* blocking FS operations */ 1451 struct rw_semaphore node_write; /* locking node writes */ 1452 struct rw_semaphore node_change; /* locking node change */ 1453 wait_queue_head_t cp_wait; 1454 unsigned long last_time[MAX_TIME]; /* to store time in jiffies */ 1455 long interval_time[MAX_TIME]; /* to store thresholds */ 1456 struct ckpt_req_control cprc_info; /* for checkpoint request control */ 1457 1458 struct inode_management im[MAX_INO_ENTRY]; /* manage inode cache */ 1459 1460 spinlock_t fsync_node_lock; /* for node entry lock */ 1461 struct list_head fsync_node_list; /* node list head */ 1462 unsigned int fsync_seg_id; /* sequence id */ 1463 unsigned int fsync_node_num; /* number of node entries */ 1464 1465 /* for orphan inode, use 0'th array */ 1466 unsigned int max_orphans; /* max orphan inodes */ 1467 1468 /* for inode management */ 1469 struct list_head inode_list[NR_INODE_TYPE]; /* dirty inode list */ 1470 spinlock_t inode_lock[NR_INODE_TYPE]; /* for dirty inode list lock */ 1471 struct mutex flush_lock; /* for flush exclusion */ 1472 1473 /* for extent tree cache */ 1474 struct radix_tree_root extent_tree_root;/* cache extent cache entries */ 1475 struct mutex extent_tree_lock; /* locking extent radix tree */ 1476 struct list_head extent_list; /* lru list for shrinker */ 1477 spinlock_t extent_lock; /* locking extent lru list */ 1478 atomic_t total_ext_tree; /* extent tree count */ 1479 struct list_head zombie_list; /* extent zombie tree list */ 1480 atomic_t total_zombie_tree; /* extent zombie tree count */ 1481 atomic_t total_ext_node; /* extent info count */ 1482 1483 /* basic filesystem units */ 1484 unsigned int log_sectors_per_block; /* log2 sectors per block */ 1485 unsigned int log_blocksize; /* log2 block size */ 1486 unsigned int blocksize; /* block size */ 1487 unsigned int root_ino_num; /* root inode number*/ 1488 unsigned int node_ino_num; /* node inode number*/ 1489 unsigned int meta_ino_num; /* meta inode number*/ 1490 unsigned int log_blocks_per_seg; /* log2 blocks per segment */ 1491 unsigned int blocks_per_seg; /* blocks per segment */ 1492 unsigned int segs_per_sec; /* segments per section */ 1493 unsigned int secs_per_zone; /* sections per zone */ 1494 unsigned int total_sections; /* total section count */ 1495 unsigned int total_node_count; /* total node block count */ 1496 unsigned int total_valid_node_count; /* valid node block count */ 1497 int dir_level; /* directory level */ 1498 int readdir_ra; /* readahead inode in readdir */ 1499 u64 max_io_bytes; /* max io bytes to merge IOs */ 1500 1501 block_t user_block_count; /* # of user blocks */ 1502 block_t total_valid_block_count; /* # of valid blocks */ 1503 block_t discard_blks; /* discard command candidats */ 1504 block_t last_valid_block_count; /* for recovery */ 1505 block_t reserved_blocks; /* configurable reserved blocks */ 1506 block_t current_reserved_blocks; /* current reserved blocks */ 1507 1508 /* Additional tracking for no checkpoint mode */ 1509 block_t unusable_block_count; /* # of blocks saved by last cp */ 1510 1511 unsigned int nquota_files; /* # of quota sysfile */ 1512 struct rw_semaphore quota_sem; /* blocking cp for flags */ 1513 1514 /* # of pages, see count_type */ 1515 atomic_t nr_pages[NR_COUNT_TYPE]; 1516 /* # of allocated blocks */ 1517 struct percpu_counter alloc_valid_block_count; 1518 1519 /* writeback control */ 1520 atomic_t wb_sync_req[META]; /* count # of WB_SYNC threads */ 1521 1522 /* valid inode count */ 1523 struct percpu_counter total_valid_inode_count; 1524 1525 struct f2fs_mount_info mount_opt; /* mount options */ 1526 1527 /* for cleaning operations */ 1528 struct rw_semaphore gc_lock; /* 1529 * semaphore for GC, avoid 1530 * race between GC and GC or CP 1531 */ 1532 struct f2fs_gc_kthread *gc_thread; /* GC thread */ 1533 struct atgc_management am; /* atgc management */ 1534 unsigned int cur_victim_sec; /* current victim section num */ 1535 unsigned int gc_mode; /* current GC state */ 1536 unsigned int next_victim_seg[2]; /* next segment in victim section */ 1537 1538 /* for skip statistic */ 1539 unsigned int atomic_files; /* # of opened atomic file */ 1540 unsigned long long skipped_atomic_files[2]; /* FG_GC and BG_GC */ 1541 unsigned long long skipped_gc_rwsem; /* FG_GC only */ 1542 1543 /* threshold for gc trials on pinned files */ 1544 u64 gc_pin_file_threshold; 1545 struct rw_semaphore pin_sem; 1546 1547 /* maximum # of trials to find a victim segment for SSR and GC */ 1548 unsigned int max_victim_search; 1549 /* migration granularity of garbage collection, unit: segment */ 1550 unsigned int migration_granularity; 1551 1552 /* 1553 * for stat information. 1554 * one is for the LFS mode, and the other is for the SSR mode. 1555 */ 1556 #ifdef CONFIG_F2FS_STAT_FS 1557 struct f2fs_stat_info *stat_info; /* FS status information */ 1558 atomic_t meta_count[META_MAX]; /* # of meta blocks */ 1559 unsigned int segment_count[2]; /* # of allocated segments */ 1560 unsigned int block_count[2]; /* # of allocated blocks */ 1561 atomic_t inplace_count; /* # of inplace update */ 1562 atomic64_t total_hit_ext; /* # of lookup extent cache */ 1563 atomic64_t read_hit_rbtree; /* # of hit rbtree extent node */ 1564 atomic64_t read_hit_largest; /* # of hit largest extent node */ 1565 atomic64_t read_hit_cached; /* # of hit cached extent node */ 1566 atomic_t inline_xattr; /* # of inline_xattr inodes */ 1567 atomic_t inline_inode; /* # of inline_data inodes */ 1568 atomic_t inline_dir; /* # of inline_dentry inodes */ 1569 atomic_t compr_inode; /* # of compressed inodes */ 1570 atomic64_t compr_blocks; /* # of compressed blocks */ 1571 atomic_t vw_cnt; /* # of volatile writes */ 1572 atomic_t max_aw_cnt; /* max # of atomic writes */ 1573 atomic_t max_vw_cnt; /* max # of volatile writes */ 1574 unsigned int io_skip_bggc; /* skip background gc for in-flight IO */ 1575 unsigned int other_skip_bggc; /* skip background gc for other reasons */ 1576 unsigned int ndirty_inode[NR_INODE_TYPE]; /* # of dirty inodes */ 1577 #endif 1578 spinlock_t stat_lock; /* lock for stat operations */ 1579 1580 /* For app/fs IO statistics */ 1581 spinlock_t iostat_lock; 1582 unsigned long long rw_iostat[NR_IO_TYPE]; 1583 unsigned long long prev_rw_iostat[NR_IO_TYPE]; 1584 bool iostat_enable; 1585 unsigned long iostat_next_period; 1586 unsigned int iostat_period_ms; 1587 1588 /* to attach REQ_META|REQ_FUA flags */ 1589 unsigned int data_io_flag; 1590 unsigned int node_io_flag; 1591 1592 /* For sysfs suppport */ 1593 struct kobject s_kobj; /* /sys/fs/f2fs/<devname> */ 1594 struct completion s_kobj_unregister; 1595 1596 struct kobject s_stat_kobj; /* /sys/fs/f2fs/<devname>/stat */ 1597 struct completion s_stat_kobj_unregister; 1598 1599 /* For shrinker support */ 1600 struct list_head s_list; 1601 int s_ndevs; /* number of devices */ 1602 struct f2fs_dev_info *devs; /* for device list */ 1603 unsigned int dirty_device; /* for checkpoint data flush */ 1604 spinlock_t dev_lock; /* protect dirty_device */ 1605 struct mutex umount_mutex; 1606 unsigned int shrinker_run_no; 1607 1608 /* For write statistics */ 1609 u64 sectors_written_start; 1610 u64 kbytes_written; 1611 1612 /* Reference to checksum algorithm driver via cryptoapi */ 1613 struct crypto_shash *s_chksum_driver; 1614 1615 /* Precomputed FS UUID checksum for seeding other checksums */ 1616 __u32 s_chksum_seed; 1617 1618 struct workqueue_struct *post_read_wq; /* post read workqueue */ 1619 1620 struct kmem_cache *inline_xattr_slab; /* inline xattr entry */ 1621 unsigned int inline_xattr_slab_size; /* default inline xattr slab size */ 1622 1623 #ifdef CONFIG_F2FS_FS_COMPRESSION 1624 struct kmem_cache *page_array_slab; /* page array entry */ 1625 unsigned int page_array_slab_size; /* default page array slab size */ 1626 #endif 1627 }; 1628 1629 struct f2fs_private_dio { 1630 struct inode *inode; 1631 void *orig_private; 1632 bio_end_io_t *orig_end_io; 1633 bool write; 1634 }; 1635 1636 #ifdef CONFIG_F2FS_FAULT_INJECTION 1637 #define f2fs_show_injection_info(sbi, type) \ 1638 printk_ratelimited("%sF2FS-fs (%s) : inject %s in %s of %pS\n", \ 1639 KERN_INFO, sbi->sb->s_id, \ 1640 f2fs_fault_name[type], \ 1641 __func__, __builtin_return_address(0)) 1642 static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type) 1643 { 1644 struct f2fs_fault_info *ffi = &F2FS_OPTION(sbi).fault_info; 1645 1646 if (!ffi->inject_rate) 1647 return false; 1648 1649 if (!IS_FAULT_SET(ffi, type)) 1650 return false; 1651 1652 atomic_inc(&ffi->inject_ops); 1653 if (atomic_read(&ffi->inject_ops) >= ffi->inject_rate) { 1654 atomic_set(&ffi->inject_ops, 0); 1655 return true; 1656 } 1657 return false; 1658 } 1659 #else 1660 #define f2fs_show_injection_info(sbi, type) do { } while (0) 1661 static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type) 1662 { 1663 return false; 1664 } 1665 #endif 1666 1667 /* 1668 * Test if the mounted volume is a multi-device volume. 1669 * - For a single regular disk volume, sbi->s_ndevs is 0. 1670 * - For a single zoned disk volume, sbi->s_ndevs is 1. 1671 * - For a multi-device volume, sbi->s_ndevs is always 2 or more. 1672 */ 1673 static inline bool f2fs_is_multi_device(struct f2fs_sb_info *sbi) 1674 { 1675 return sbi->s_ndevs > 1; 1676 } 1677 1678 static inline void f2fs_update_time(struct f2fs_sb_info *sbi, int type) 1679 { 1680 unsigned long now = jiffies; 1681 1682 sbi->last_time[type] = now; 1683 1684 /* DISCARD_TIME and GC_TIME are based on REQ_TIME */ 1685 if (type == REQ_TIME) { 1686 sbi->last_time[DISCARD_TIME] = now; 1687 sbi->last_time[GC_TIME] = now; 1688 } 1689 } 1690 1691 static inline bool f2fs_time_over(struct f2fs_sb_info *sbi, int type) 1692 { 1693 unsigned long interval = sbi->interval_time[type] * HZ; 1694 1695 return time_after(jiffies, sbi->last_time[type] + interval); 1696 } 1697 1698 static inline unsigned int f2fs_time_to_wait(struct f2fs_sb_info *sbi, 1699 int type) 1700 { 1701 unsigned long interval = sbi->interval_time[type] * HZ; 1702 unsigned int wait_ms = 0; 1703 long delta; 1704 1705 delta = (sbi->last_time[type] + interval) - jiffies; 1706 if (delta > 0) 1707 wait_ms = jiffies_to_msecs(delta); 1708 1709 return wait_ms; 1710 } 1711 1712 /* 1713 * Inline functions 1714 */ 1715 static inline u32 __f2fs_crc32(struct f2fs_sb_info *sbi, u32 crc, 1716 const void *address, unsigned int length) 1717 { 1718 struct { 1719 struct shash_desc shash; 1720 char ctx[4]; 1721 } desc; 1722 int err; 1723 1724 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver) != sizeof(desc.ctx)); 1725 1726 desc.shash.tfm = sbi->s_chksum_driver; 1727 *(u32 *)desc.ctx = crc; 1728 1729 err = crypto_shash_update(&desc.shash, address, length); 1730 BUG_ON(err); 1731 1732 return *(u32 *)desc.ctx; 1733 } 1734 1735 static inline u32 f2fs_crc32(struct f2fs_sb_info *sbi, const void *address, 1736 unsigned int length) 1737 { 1738 return __f2fs_crc32(sbi, F2FS_SUPER_MAGIC, address, length); 1739 } 1740 1741 static inline bool f2fs_crc_valid(struct f2fs_sb_info *sbi, __u32 blk_crc, 1742 void *buf, size_t buf_size) 1743 { 1744 return f2fs_crc32(sbi, buf, buf_size) == blk_crc; 1745 } 1746 1747 static inline u32 f2fs_chksum(struct f2fs_sb_info *sbi, u32 crc, 1748 const void *address, unsigned int length) 1749 { 1750 return __f2fs_crc32(sbi, crc, address, length); 1751 } 1752 1753 static inline struct f2fs_inode_info *F2FS_I(struct inode *inode) 1754 { 1755 return container_of(inode, struct f2fs_inode_info, vfs_inode); 1756 } 1757 1758 static inline struct f2fs_sb_info *F2FS_SB(struct super_block *sb) 1759 { 1760 return sb->s_fs_info; 1761 } 1762 1763 static inline struct f2fs_sb_info *F2FS_I_SB(struct inode *inode) 1764 { 1765 return F2FS_SB(inode->i_sb); 1766 } 1767 1768 static inline struct f2fs_sb_info *F2FS_M_SB(struct address_space *mapping) 1769 { 1770 return F2FS_I_SB(mapping->host); 1771 } 1772 1773 static inline struct f2fs_sb_info *F2FS_P_SB(struct page *page) 1774 { 1775 return F2FS_M_SB(page_file_mapping(page)); 1776 } 1777 1778 static inline struct f2fs_super_block *F2FS_RAW_SUPER(struct f2fs_sb_info *sbi) 1779 { 1780 return (struct f2fs_super_block *)(sbi->raw_super); 1781 } 1782 1783 static inline struct f2fs_checkpoint *F2FS_CKPT(struct f2fs_sb_info *sbi) 1784 { 1785 return (struct f2fs_checkpoint *)(sbi->ckpt); 1786 } 1787 1788 static inline struct f2fs_node *F2FS_NODE(struct page *page) 1789 { 1790 return (struct f2fs_node *)page_address(page); 1791 } 1792 1793 static inline struct f2fs_inode *F2FS_INODE(struct page *page) 1794 { 1795 return &((struct f2fs_node *)page_address(page))->i; 1796 } 1797 1798 static inline struct f2fs_nm_info *NM_I(struct f2fs_sb_info *sbi) 1799 { 1800 return (struct f2fs_nm_info *)(sbi->nm_info); 1801 } 1802 1803 static inline struct f2fs_sm_info *SM_I(struct f2fs_sb_info *sbi) 1804 { 1805 return (struct f2fs_sm_info *)(sbi->sm_info); 1806 } 1807 1808 static inline struct sit_info *SIT_I(struct f2fs_sb_info *sbi) 1809 { 1810 return (struct sit_info *)(SM_I(sbi)->sit_info); 1811 } 1812 1813 static inline struct free_segmap_info *FREE_I(struct f2fs_sb_info *sbi) 1814 { 1815 return (struct free_segmap_info *)(SM_I(sbi)->free_info); 1816 } 1817 1818 static inline struct dirty_seglist_info *DIRTY_I(struct f2fs_sb_info *sbi) 1819 { 1820 return (struct dirty_seglist_info *)(SM_I(sbi)->dirty_info); 1821 } 1822 1823 static inline struct address_space *META_MAPPING(struct f2fs_sb_info *sbi) 1824 { 1825 return sbi->meta_inode->i_mapping; 1826 } 1827 1828 static inline struct address_space *NODE_MAPPING(struct f2fs_sb_info *sbi) 1829 { 1830 return sbi->node_inode->i_mapping; 1831 } 1832 1833 static inline bool is_sbi_flag_set(struct f2fs_sb_info *sbi, unsigned int type) 1834 { 1835 return test_bit(type, &sbi->s_flag); 1836 } 1837 1838 static inline void set_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type) 1839 { 1840 set_bit(type, &sbi->s_flag); 1841 } 1842 1843 static inline void clear_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type) 1844 { 1845 clear_bit(type, &sbi->s_flag); 1846 } 1847 1848 static inline unsigned long long cur_cp_version(struct f2fs_checkpoint *cp) 1849 { 1850 return le64_to_cpu(cp->checkpoint_ver); 1851 } 1852 1853 static inline unsigned long f2fs_qf_ino(struct super_block *sb, int type) 1854 { 1855 if (type < F2FS_MAX_QUOTAS) 1856 return le32_to_cpu(F2FS_SB(sb)->raw_super->qf_ino[type]); 1857 return 0; 1858 } 1859 1860 static inline __u64 cur_cp_crc(struct f2fs_checkpoint *cp) 1861 { 1862 size_t crc_offset = le32_to_cpu(cp->checksum_offset); 1863 return le32_to_cpu(*((__le32 *)((unsigned char *)cp + crc_offset))); 1864 } 1865 1866 static inline bool __is_set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 1867 { 1868 unsigned int ckpt_flags = le32_to_cpu(cp->ckpt_flags); 1869 1870 return ckpt_flags & f; 1871 } 1872 1873 static inline bool is_set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 1874 { 1875 return __is_set_ckpt_flags(F2FS_CKPT(sbi), f); 1876 } 1877 1878 static inline void __set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 1879 { 1880 unsigned int ckpt_flags; 1881 1882 ckpt_flags = le32_to_cpu(cp->ckpt_flags); 1883 ckpt_flags |= f; 1884 cp->ckpt_flags = cpu_to_le32(ckpt_flags); 1885 } 1886 1887 static inline void set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 1888 { 1889 unsigned long flags; 1890 1891 spin_lock_irqsave(&sbi->cp_lock, flags); 1892 __set_ckpt_flags(F2FS_CKPT(sbi), f); 1893 spin_unlock_irqrestore(&sbi->cp_lock, flags); 1894 } 1895 1896 static inline void __clear_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 1897 { 1898 unsigned int ckpt_flags; 1899 1900 ckpt_flags = le32_to_cpu(cp->ckpt_flags); 1901 ckpt_flags &= (~f); 1902 cp->ckpt_flags = cpu_to_le32(ckpt_flags); 1903 } 1904 1905 static inline void clear_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 1906 { 1907 unsigned long flags; 1908 1909 spin_lock_irqsave(&sbi->cp_lock, flags); 1910 __clear_ckpt_flags(F2FS_CKPT(sbi), f); 1911 spin_unlock_irqrestore(&sbi->cp_lock, flags); 1912 } 1913 1914 static inline void disable_nat_bits(struct f2fs_sb_info *sbi, bool lock) 1915 { 1916 unsigned long flags; 1917 unsigned char *nat_bits; 1918 1919 /* 1920 * In order to re-enable nat_bits we need to call fsck.f2fs by 1921 * set_sbi_flag(sbi, SBI_NEED_FSCK). But it may give huge cost, 1922 * so let's rely on regular fsck or unclean shutdown. 1923 */ 1924 1925 if (lock) 1926 spin_lock_irqsave(&sbi->cp_lock, flags); 1927 __clear_ckpt_flags(F2FS_CKPT(sbi), CP_NAT_BITS_FLAG); 1928 nat_bits = NM_I(sbi)->nat_bits; 1929 NM_I(sbi)->nat_bits = NULL; 1930 if (lock) 1931 spin_unlock_irqrestore(&sbi->cp_lock, flags); 1932 1933 kvfree(nat_bits); 1934 } 1935 1936 static inline bool enabled_nat_bits(struct f2fs_sb_info *sbi, 1937 struct cp_control *cpc) 1938 { 1939 bool set = is_set_ckpt_flags(sbi, CP_NAT_BITS_FLAG); 1940 1941 return (cpc) ? (cpc->reason & CP_UMOUNT) && set : set; 1942 } 1943 1944 static inline void f2fs_lock_op(struct f2fs_sb_info *sbi) 1945 { 1946 down_read(&sbi->cp_rwsem); 1947 } 1948 1949 static inline int f2fs_trylock_op(struct f2fs_sb_info *sbi) 1950 { 1951 return down_read_trylock(&sbi->cp_rwsem); 1952 } 1953 1954 static inline void f2fs_unlock_op(struct f2fs_sb_info *sbi) 1955 { 1956 up_read(&sbi->cp_rwsem); 1957 } 1958 1959 static inline void f2fs_lock_all(struct f2fs_sb_info *sbi) 1960 { 1961 down_write(&sbi->cp_rwsem); 1962 } 1963 1964 static inline void f2fs_unlock_all(struct f2fs_sb_info *sbi) 1965 { 1966 up_write(&sbi->cp_rwsem); 1967 } 1968 1969 static inline int __get_cp_reason(struct f2fs_sb_info *sbi) 1970 { 1971 int reason = CP_SYNC; 1972 1973 if (test_opt(sbi, FASTBOOT)) 1974 reason = CP_FASTBOOT; 1975 if (is_sbi_flag_set(sbi, SBI_IS_CLOSE)) 1976 reason = CP_UMOUNT; 1977 return reason; 1978 } 1979 1980 static inline bool __remain_node_summaries(int reason) 1981 { 1982 return (reason & (CP_UMOUNT | CP_FASTBOOT)); 1983 } 1984 1985 static inline bool __exist_node_summaries(struct f2fs_sb_info *sbi) 1986 { 1987 return (is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG) || 1988 is_set_ckpt_flags(sbi, CP_FASTBOOT_FLAG)); 1989 } 1990 1991 /* 1992 * Check whether the inode has blocks or not 1993 */ 1994 static inline int F2FS_HAS_BLOCKS(struct inode *inode) 1995 { 1996 block_t xattr_block = F2FS_I(inode)->i_xattr_nid ? 1 : 0; 1997 1998 return (inode->i_blocks >> F2FS_LOG_SECTORS_PER_BLOCK) > xattr_block; 1999 } 2000 2001 static inline bool f2fs_has_xattr_block(unsigned int ofs) 2002 { 2003 return ofs == XATTR_NODE_OFFSET; 2004 } 2005 2006 static inline bool __allow_reserved_blocks(struct f2fs_sb_info *sbi, 2007 struct inode *inode, bool cap) 2008 { 2009 if (!inode) 2010 return true; 2011 if (!test_opt(sbi, RESERVE_ROOT)) 2012 return false; 2013 if (IS_NOQUOTA(inode)) 2014 return true; 2015 if (uid_eq(F2FS_OPTION(sbi).s_resuid, current_fsuid())) 2016 return true; 2017 if (!gid_eq(F2FS_OPTION(sbi).s_resgid, GLOBAL_ROOT_GID) && 2018 in_group_p(F2FS_OPTION(sbi).s_resgid)) 2019 return true; 2020 if (cap && capable(CAP_SYS_RESOURCE)) 2021 return true; 2022 return false; 2023 } 2024 2025 static inline void f2fs_i_blocks_write(struct inode *, block_t, bool, bool); 2026 static inline int inc_valid_block_count(struct f2fs_sb_info *sbi, 2027 struct inode *inode, blkcnt_t *count) 2028 { 2029 blkcnt_t diff = 0, release = 0; 2030 block_t avail_user_block_count; 2031 int ret; 2032 2033 ret = dquot_reserve_block(inode, *count); 2034 if (ret) 2035 return ret; 2036 2037 if (time_to_inject(sbi, FAULT_BLOCK)) { 2038 f2fs_show_injection_info(sbi, FAULT_BLOCK); 2039 release = *count; 2040 goto release_quota; 2041 } 2042 2043 /* 2044 * let's increase this in prior to actual block count change in order 2045 * for f2fs_sync_file to avoid data races when deciding checkpoint. 2046 */ 2047 percpu_counter_add(&sbi->alloc_valid_block_count, (*count)); 2048 2049 spin_lock(&sbi->stat_lock); 2050 sbi->total_valid_block_count += (block_t)(*count); 2051 avail_user_block_count = sbi->user_block_count - 2052 sbi->current_reserved_blocks; 2053 2054 if (!__allow_reserved_blocks(sbi, inode, true)) 2055 avail_user_block_count -= F2FS_OPTION(sbi).root_reserved_blocks; 2056 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) { 2057 if (avail_user_block_count > sbi->unusable_block_count) 2058 avail_user_block_count -= sbi->unusable_block_count; 2059 else 2060 avail_user_block_count = 0; 2061 } 2062 if (unlikely(sbi->total_valid_block_count > avail_user_block_count)) { 2063 diff = sbi->total_valid_block_count - avail_user_block_count; 2064 if (diff > *count) 2065 diff = *count; 2066 *count -= diff; 2067 release = diff; 2068 sbi->total_valid_block_count -= diff; 2069 if (!*count) { 2070 spin_unlock(&sbi->stat_lock); 2071 goto enospc; 2072 } 2073 } 2074 spin_unlock(&sbi->stat_lock); 2075 2076 if (unlikely(release)) { 2077 percpu_counter_sub(&sbi->alloc_valid_block_count, release); 2078 dquot_release_reservation_block(inode, release); 2079 } 2080 f2fs_i_blocks_write(inode, *count, true, true); 2081 return 0; 2082 2083 enospc: 2084 percpu_counter_sub(&sbi->alloc_valid_block_count, release); 2085 release_quota: 2086 dquot_release_reservation_block(inode, release); 2087 return -ENOSPC; 2088 } 2089 2090 __printf(2, 3) 2091 void f2fs_printk(struct f2fs_sb_info *sbi, const char *fmt, ...); 2092 2093 #define f2fs_err(sbi, fmt, ...) \ 2094 f2fs_printk(sbi, KERN_ERR fmt, ##__VA_ARGS__) 2095 #define f2fs_warn(sbi, fmt, ...) \ 2096 f2fs_printk(sbi, KERN_WARNING fmt, ##__VA_ARGS__) 2097 #define f2fs_notice(sbi, fmt, ...) \ 2098 f2fs_printk(sbi, KERN_NOTICE fmt, ##__VA_ARGS__) 2099 #define f2fs_info(sbi, fmt, ...) \ 2100 f2fs_printk(sbi, KERN_INFO fmt, ##__VA_ARGS__) 2101 #define f2fs_debug(sbi, fmt, ...) \ 2102 f2fs_printk(sbi, KERN_DEBUG fmt, ##__VA_ARGS__) 2103 2104 static inline void dec_valid_block_count(struct f2fs_sb_info *sbi, 2105 struct inode *inode, 2106 block_t count) 2107 { 2108 blkcnt_t sectors = count << F2FS_LOG_SECTORS_PER_BLOCK; 2109 2110 spin_lock(&sbi->stat_lock); 2111 f2fs_bug_on(sbi, sbi->total_valid_block_count < (block_t) count); 2112 sbi->total_valid_block_count -= (block_t)count; 2113 if (sbi->reserved_blocks && 2114 sbi->current_reserved_blocks < sbi->reserved_blocks) 2115 sbi->current_reserved_blocks = min(sbi->reserved_blocks, 2116 sbi->current_reserved_blocks + count); 2117 spin_unlock(&sbi->stat_lock); 2118 if (unlikely(inode->i_blocks < sectors)) { 2119 f2fs_warn(sbi, "Inconsistent i_blocks, ino:%lu, iblocks:%llu, sectors:%llu", 2120 inode->i_ino, 2121 (unsigned long long)inode->i_blocks, 2122 (unsigned long long)sectors); 2123 set_sbi_flag(sbi, SBI_NEED_FSCK); 2124 return; 2125 } 2126 f2fs_i_blocks_write(inode, count, false, true); 2127 } 2128 2129 static inline void inc_page_count(struct f2fs_sb_info *sbi, int count_type) 2130 { 2131 atomic_inc(&sbi->nr_pages[count_type]); 2132 2133 if (count_type == F2FS_DIRTY_DENTS || 2134 count_type == F2FS_DIRTY_NODES || 2135 count_type == F2FS_DIRTY_META || 2136 count_type == F2FS_DIRTY_QDATA || 2137 count_type == F2FS_DIRTY_IMETA) 2138 set_sbi_flag(sbi, SBI_IS_DIRTY); 2139 } 2140 2141 static inline void inode_inc_dirty_pages(struct inode *inode) 2142 { 2143 atomic_inc(&F2FS_I(inode)->dirty_pages); 2144 inc_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ? 2145 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA); 2146 if (IS_NOQUOTA(inode)) 2147 inc_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA); 2148 } 2149 2150 static inline void dec_page_count(struct f2fs_sb_info *sbi, int count_type) 2151 { 2152 atomic_dec(&sbi->nr_pages[count_type]); 2153 } 2154 2155 static inline void inode_dec_dirty_pages(struct inode *inode) 2156 { 2157 if (!S_ISDIR(inode->i_mode) && !S_ISREG(inode->i_mode) && 2158 !S_ISLNK(inode->i_mode)) 2159 return; 2160 2161 atomic_dec(&F2FS_I(inode)->dirty_pages); 2162 dec_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ? 2163 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA); 2164 if (IS_NOQUOTA(inode)) 2165 dec_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA); 2166 } 2167 2168 static inline s64 get_pages(struct f2fs_sb_info *sbi, int count_type) 2169 { 2170 return atomic_read(&sbi->nr_pages[count_type]); 2171 } 2172 2173 static inline int get_dirty_pages(struct inode *inode) 2174 { 2175 return atomic_read(&F2FS_I(inode)->dirty_pages); 2176 } 2177 2178 static inline int get_blocktype_secs(struct f2fs_sb_info *sbi, int block_type) 2179 { 2180 unsigned int pages_per_sec = sbi->segs_per_sec * sbi->blocks_per_seg; 2181 unsigned int segs = (get_pages(sbi, block_type) + pages_per_sec - 1) >> 2182 sbi->log_blocks_per_seg; 2183 2184 return segs / sbi->segs_per_sec; 2185 } 2186 2187 static inline block_t valid_user_blocks(struct f2fs_sb_info *sbi) 2188 { 2189 return sbi->total_valid_block_count; 2190 } 2191 2192 static inline block_t discard_blocks(struct f2fs_sb_info *sbi) 2193 { 2194 return sbi->discard_blks; 2195 } 2196 2197 static inline unsigned long __bitmap_size(struct f2fs_sb_info *sbi, int flag) 2198 { 2199 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi); 2200 2201 /* return NAT or SIT bitmap */ 2202 if (flag == NAT_BITMAP) 2203 return le32_to_cpu(ckpt->nat_ver_bitmap_bytesize); 2204 else if (flag == SIT_BITMAP) 2205 return le32_to_cpu(ckpt->sit_ver_bitmap_bytesize); 2206 2207 return 0; 2208 } 2209 2210 static inline block_t __cp_payload(struct f2fs_sb_info *sbi) 2211 { 2212 return le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_payload); 2213 } 2214 2215 static inline void *__bitmap_ptr(struct f2fs_sb_info *sbi, int flag) 2216 { 2217 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi); 2218 int offset; 2219 2220 if (is_set_ckpt_flags(sbi, CP_LARGE_NAT_BITMAP_FLAG)) { 2221 offset = (flag == SIT_BITMAP) ? 2222 le32_to_cpu(ckpt->nat_ver_bitmap_bytesize) : 0; 2223 /* 2224 * if large_nat_bitmap feature is enabled, leave checksum 2225 * protection for all nat/sit bitmaps. 2226 */ 2227 return &ckpt->sit_nat_version_bitmap + offset + sizeof(__le32); 2228 } 2229 2230 if (__cp_payload(sbi) > 0) { 2231 if (flag == NAT_BITMAP) 2232 return &ckpt->sit_nat_version_bitmap; 2233 else 2234 return (unsigned char *)ckpt + F2FS_BLKSIZE; 2235 } else { 2236 offset = (flag == NAT_BITMAP) ? 2237 le32_to_cpu(ckpt->sit_ver_bitmap_bytesize) : 0; 2238 return &ckpt->sit_nat_version_bitmap + offset; 2239 } 2240 } 2241 2242 static inline block_t __start_cp_addr(struct f2fs_sb_info *sbi) 2243 { 2244 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr); 2245 2246 if (sbi->cur_cp_pack == 2) 2247 start_addr += sbi->blocks_per_seg; 2248 return start_addr; 2249 } 2250 2251 static inline block_t __start_cp_next_addr(struct f2fs_sb_info *sbi) 2252 { 2253 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr); 2254 2255 if (sbi->cur_cp_pack == 1) 2256 start_addr += sbi->blocks_per_seg; 2257 return start_addr; 2258 } 2259 2260 static inline void __set_cp_next_pack(struct f2fs_sb_info *sbi) 2261 { 2262 sbi->cur_cp_pack = (sbi->cur_cp_pack == 1) ? 2 : 1; 2263 } 2264 2265 static inline block_t __start_sum_addr(struct f2fs_sb_info *sbi) 2266 { 2267 return le32_to_cpu(F2FS_CKPT(sbi)->cp_pack_start_sum); 2268 } 2269 2270 static inline int inc_valid_node_count(struct f2fs_sb_info *sbi, 2271 struct inode *inode, bool is_inode) 2272 { 2273 block_t valid_block_count; 2274 unsigned int valid_node_count, user_block_count; 2275 int err; 2276 2277 if (is_inode) { 2278 if (inode) { 2279 err = dquot_alloc_inode(inode); 2280 if (err) 2281 return err; 2282 } 2283 } else { 2284 err = dquot_reserve_block(inode, 1); 2285 if (err) 2286 return err; 2287 } 2288 2289 if (time_to_inject(sbi, FAULT_BLOCK)) { 2290 f2fs_show_injection_info(sbi, FAULT_BLOCK); 2291 goto enospc; 2292 } 2293 2294 spin_lock(&sbi->stat_lock); 2295 2296 valid_block_count = sbi->total_valid_block_count + 2297 sbi->current_reserved_blocks + 1; 2298 2299 if (!__allow_reserved_blocks(sbi, inode, false)) 2300 valid_block_count += F2FS_OPTION(sbi).root_reserved_blocks; 2301 user_block_count = sbi->user_block_count; 2302 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) 2303 user_block_count -= sbi->unusable_block_count; 2304 2305 if (unlikely(valid_block_count > user_block_count)) { 2306 spin_unlock(&sbi->stat_lock); 2307 goto enospc; 2308 } 2309 2310 valid_node_count = sbi->total_valid_node_count + 1; 2311 if (unlikely(valid_node_count > sbi->total_node_count)) { 2312 spin_unlock(&sbi->stat_lock); 2313 goto enospc; 2314 } 2315 2316 sbi->total_valid_node_count++; 2317 sbi->total_valid_block_count++; 2318 spin_unlock(&sbi->stat_lock); 2319 2320 if (inode) { 2321 if (is_inode) 2322 f2fs_mark_inode_dirty_sync(inode, true); 2323 else 2324 f2fs_i_blocks_write(inode, 1, true, true); 2325 } 2326 2327 percpu_counter_inc(&sbi->alloc_valid_block_count); 2328 return 0; 2329 2330 enospc: 2331 if (is_inode) { 2332 if (inode) 2333 dquot_free_inode(inode); 2334 } else { 2335 dquot_release_reservation_block(inode, 1); 2336 } 2337 return -ENOSPC; 2338 } 2339 2340 static inline void dec_valid_node_count(struct f2fs_sb_info *sbi, 2341 struct inode *inode, bool is_inode) 2342 { 2343 spin_lock(&sbi->stat_lock); 2344 2345 f2fs_bug_on(sbi, !sbi->total_valid_block_count); 2346 f2fs_bug_on(sbi, !sbi->total_valid_node_count); 2347 2348 sbi->total_valid_node_count--; 2349 sbi->total_valid_block_count--; 2350 if (sbi->reserved_blocks && 2351 sbi->current_reserved_blocks < sbi->reserved_blocks) 2352 sbi->current_reserved_blocks++; 2353 2354 spin_unlock(&sbi->stat_lock); 2355 2356 if (is_inode) { 2357 dquot_free_inode(inode); 2358 } else { 2359 if (unlikely(inode->i_blocks == 0)) { 2360 f2fs_warn(sbi, "dec_valid_node_count: inconsistent i_blocks, ino:%lu, iblocks:%llu", 2361 inode->i_ino, 2362 (unsigned long long)inode->i_blocks); 2363 set_sbi_flag(sbi, SBI_NEED_FSCK); 2364 return; 2365 } 2366 f2fs_i_blocks_write(inode, 1, false, true); 2367 } 2368 } 2369 2370 static inline unsigned int valid_node_count(struct f2fs_sb_info *sbi) 2371 { 2372 return sbi->total_valid_node_count; 2373 } 2374 2375 static inline void inc_valid_inode_count(struct f2fs_sb_info *sbi) 2376 { 2377 percpu_counter_inc(&sbi->total_valid_inode_count); 2378 } 2379 2380 static inline void dec_valid_inode_count(struct f2fs_sb_info *sbi) 2381 { 2382 percpu_counter_dec(&sbi->total_valid_inode_count); 2383 } 2384 2385 static inline s64 valid_inode_count(struct f2fs_sb_info *sbi) 2386 { 2387 return percpu_counter_sum_positive(&sbi->total_valid_inode_count); 2388 } 2389 2390 static inline struct page *f2fs_grab_cache_page(struct address_space *mapping, 2391 pgoff_t index, bool for_write) 2392 { 2393 struct page *page; 2394 2395 if (IS_ENABLED(CONFIG_F2FS_FAULT_INJECTION)) { 2396 if (!for_write) 2397 page = find_get_page_flags(mapping, index, 2398 FGP_LOCK | FGP_ACCESSED); 2399 else 2400 page = find_lock_page(mapping, index); 2401 if (page) 2402 return page; 2403 2404 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_ALLOC)) { 2405 f2fs_show_injection_info(F2FS_M_SB(mapping), 2406 FAULT_PAGE_ALLOC); 2407 return NULL; 2408 } 2409 } 2410 2411 if (!for_write) 2412 return grab_cache_page(mapping, index); 2413 return grab_cache_page_write_begin(mapping, index, AOP_FLAG_NOFS); 2414 } 2415 2416 static inline struct page *f2fs_pagecache_get_page( 2417 struct address_space *mapping, pgoff_t index, 2418 int fgp_flags, gfp_t gfp_mask) 2419 { 2420 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_GET)) { 2421 f2fs_show_injection_info(F2FS_M_SB(mapping), FAULT_PAGE_GET); 2422 return NULL; 2423 } 2424 2425 return pagecache_get_page(mapping, index, fgp_flags, gfp_mask); 2426 } 2427 2428 static inline void f2fs_copy_page(struct page *src, struct page *dst) 2429 { 2430 char *src_kaddr = kmap(src); 2431 char *dst_kaddr = kmap(dst); 2432 2433 memcpy(dst_kaddr, src_kaddr, PAGE_SIZE); 2434 kunmap(dst); 2435 kunmap(src); 2436 } 2437 2438 static inline void f2fs_put_page(struct page *page, int unlock) 2439 { 2440 if (!page) 2441 return; 2442 2443 if (unlock) { 2444 f2fs_bug_on(F2FS_P_SB(page), !PageLocked(page)); 2445 unlock_page(page); 2446 } 2447 put_page(page); 2448 } 2449 2450 static inline void f2fs_put_dnode(struct dnode_of_data *dn) 2451 { 2452 if (dn->node_page) 2453 f2fs_put_page(dn->node_page, 1); 2454 if (dn->inode_page && dn->node_page != dn->inode_page) 2455 f2fs_put_page(dn->inode_page, 0); 2456 dn->node_page = NULL; 2457 dn->inode_page = NULL; 2458 } 2459 2460 static inline struct kmem_cache *f2fs_kmem_cache_create(const char *name, 2461 size_t size) 2462 { 2463 return kmem_cache_create(name, size, 0, SLAB_RECLAIM_ACCOUNT, NULL); 2464 } 2465 2466 static inline void *f2fs_kmem_cache_alloc(struct kmem_cache *cachep, 2467 gfp_t flags) 2468 { 2469 void *entry; 2470 2471 entry = kmem_cache_alloc(cachep, flags); 2472 if (!entry) 2473 entry = kmem_cache_alloc(cachep, flags | __GFP_NOFAIL); 2474 return entry; 2475 } 2476 2477 static inline bool is_inflight_io(struct f2fs_sb_info *sbi, int type) 2478 { 2479 if (get_pages(sbi, F2FS_RD_DATA) || get_pages(sbi, F2FS_RD_NODE) || 2480 get_pages(sbi, F2FS_RD_META) || get_pages(sbi, F2FS_WB_DATA) || 2481 get_pages(sbi, F2FS_WB_CP_DATA) || 2482 get_pages(sbi, F2FS_DIO_READ) || 2483 get_pages(sbi, F2FS_DIO_WRITE)) 2484 return true; 2485 2486 if (type != DISCARD_TIME && SM_I(sbi) && SM_I(sbi)->dcc_info && 2487 atomic_read(&SM_I(sbi)->dcc_info->queued_discard)) 2488 return true; 2489 2490 if (SM_I(sbi) && SM_I(sbi)->fcc_info && 2491 atomic_read(&SM_I(sbi)->fcc_info->queued_flush)) 2492 return true; 2493 return false; 2494 } 2495 2496 static inline bool is_idle(struct f2fs_sb_info *sbi, int type) 2497 { 2498 if (sbi->gc_mode == GC_URGENT_HIGH) 2499 return true; 2500 2501 if (is_inflight_io(sbi, type)) 2502 return false; 2503 2504 if (sbi->gc_mode == GC_URGENT_LOW && 2505 (type == DISCARD_TIME || type == GC_TIME)) 2506 return true; 2507 2508 return f2fs_time_over(sbi, type); 2509 } 2510 2511 static inline void f2fs_radix_tree_insert(struct radix_tree_root *root, 2512 unsigned long index, void *item) 2513 { 2514 while (radix_tree_insert(root, index, item)) 2515 cond_resched(); 2516 } 2517 2518 #define RAW_IS_INODE(p) ((p)->footer.nid == (p)->footer.ino) 2519 2520 static inline bool IS_INODE(struct page *page) 2521 { 2522 struct f2fs_node *p = F2FS_NODE(page); 2523 2524 return RAW_IS_INODE(p); 2525 } 2526 2527 static inline int offset_in_addr(struct f2fs_inode *i) 2528 { 2529 return (i->i_inline & F2FS_EXTRA_ATTR) ? 2530 (le16_to_cpu(i->i_extra_isize) / sizeof(__le32)) : 0; 2531 } 2532 2533 static inline __le32 *blkaddr_in_node(struct f2fs_node *node) 2534 { 2535 return RAW_IS_INODE(node) ? node->i.i_addr : node->dn.addr; 2536 } 2537 2538 static inline int f2fs_has_extra_attr(struct inode *inode); 2539 static inline block_t data_blkaddr(struct inode *inode, 2540 struct page *node_page, unsigned int offset) 2541 { 2542 struct f2fs_node *raw_node; 2543 __le32 *addr_array; 2544 int base = 0; 2545 bool is_inode = IS_INODE(node_page); 2546 2547 raw_node = F2FS_NODE(node_page); 2548 2549 if (is_inode) { 2550 if (!inode) 2551 /* from GC path only */ 2552 base = offset_in_addr(&raw_node->i); 2553 else if (f2fs_has_extra_attr(inode)) 2554 base = get_extra_isize(inode); 2555 } 2556 2557 addr_array = blkaddr_in_node(raw_node); 2558 return le32_to_cpu(addr_array[base + offset]); 2559 } 2560 2561 static inline block_t f2fs_data_blkaddr(struct dnode_of_data *dn) 2562 { 2563 return data_blkaddr(dn->inode, dn->node_page, dn->ofs_in_node); 2564 } 2565 2566 static inline int f2fs_test_bit(unsigned int nr, char *addr) 2567 { 2568 int mask; 2569 2570 addr += (nr >> 3); 2571 mask = 1 << (7 - (nr & 0x07)); 2572 return mask & *addr; 2573 } 2574 2575 static inline void f2fs_set_bit(unsigned int nr, char *addr) 2576 { 2577 int mask; 2578 2579 addr += (nr >> 3); 2580 mask = 1 << (7 - (nr & 0x07)); 2581 *addr |= mask; 2582 } 2583 2584 static inline void f2fs_clear_bit(unsigned int nr, char *addr) 2585 { 2586 int mask; 2587 2588 addr += (nr >> 3); 2589 mask = 1 << (7 - (nr & 0x07)); 2590 *addr &= ~mask; 2591 } 2592 2593 static inline int f2fs_test_and_set_bit(unsigned int nr, char *addr) 2594 { 2595 int mask; 2596 int ret; 2597 2598 addr += (nr >> 3); 2599 mask = 1 << (7 - (nr & 0x07)); 2600 ret = mask & *addr; 2601 *addr |= mask; 2602 return ret; 2603 } 2604 2605 static inline int f2fs_test_and_clear_bit(unsigned int nr, char *addr) 2606 { 2607 int mask; 2608 int ret; 2609 2610 addr += (nr >> 3); 2611 mask = 1 << (7 - (nr & 0x07)); 2612 ret = mask & *addr; 2613 *addr &= ~mask; 2614 return ret; 2615 } 2616 2617 static inline void f2fs_change_bit(unsigned int nr, char *addr) 2618 { 2619 int mask; 2620 2621 addr += (nr >> 3); 2622 mask = 1 << (7 - (nr & 0x07)); 2623 *addr ^= mask; 2624 } 2625 2626 /* 2627 * On-disk inode flags (f2fs_inode::i_flags) 2628 */ 2629 #define F2FS_COMPR_FL 0x00000004 /* Compress file */ 2630 #define F2FS_SYNC_FL 0x00000008 /* Synchronous updates */ 2631 #define F2FS_IMMUTABLE_FL 0x00000010 /* Immutable file */ 2632 #define F2FS_APPEND_FL 0x00000020 /* writes to file may only append */ 2633 #define F2FS_NODUMP_FL 0x00000040 /* do not dump file */ 2634 #define F2FS_NOATIME_FL 0x00000080 /* do not update atime */ 2635 #define F2FS_NOCOMP_FL 0x00000400 /* Don't compress */ 2636 #define F2FS_INDEX_FL 0x00001000 /* hash-indexed directory */ 2637 #define F2FS_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */ 2638 #define F2FS_PROJINHERIT_FL 0x20000000 /* Create with parents projid */ 2639 #define F2FS_CASEFOLD_FL 0x40000000 /* Casefolded file */ 2640 2641 /* Flags that should be inherited by new inodes from their parent. */ 2642 #define F2FS_FL_INHERITED (F2FS_SYNC_FL | F2FS_NODUMP_FL | F2FS_NOATIME_FL | \ 2643 F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \ 2644 F2FS_CASEFOLD_FL | F2FS_COMPR_FL | F2FS_NOCOMP_FL) 2645 2646 /* Flags that are appropriate for regular files (all but dir-specific ones). */ 2647 #define F2FS_REG_FLMASK (~(F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \ 2648 F2FS_CASEFOLD_FL)) 2649 2650 /* Flags that are appropriate for non-directories/regular files. */ 2651 #define F2FS_OTHER_FLMASK (F2FS_NODUMP_FL | F2FS_NOATIME_FL) 2652 2653 static inline __u32 f2fs_mask_flags(umode_t mode, __u32 flags) 2654 { 2655 if (S_ISDIR(mode)) 2656 return flags; 2657 else if (S_ISREG(mode)) 2658 return flags & F2FS_REG_FLMASK; 2659 else 2660 return flags & F2FS_OTHER_FLMASK; 2661 } 2662 2663 static inline void __mark_inode_dirty_flag(struct inode *inode, 2664 int flag, bool set) 2665 { 2666 switch (flag) { 2667 case FI_INLINE_XATTR: 2668 case FI_INLINE_DATA: 2669 case FI_INLINE_DENTRY: 2670 case FI_NEW_INODE: 2671 if (set) 2672 return; 2673 fallthrough; 2674 case FI_DATA_EXIST: 2675 case FI_INLINE_DOTS: 2676 case FI_PIN_FILE: 2677 f2fs_mark_inode_dirty_sync(inode, true); 2678 } 2679 } 2680 2681 static inline void set_inode_flag(struct inode *inode, int flag) 2682 { 2683 set_bit(flag, F2FS_I(inode)->flags); 2684 __mark_inode_dirty_flag(inode, flag, true); 2685 } 2686 2687 static inline int is_inode_flag_set(struct inode *inode, int flag) 2688 { 2689 return test_bit(flag, F2FS_I(inode)->flags); 2690 } 2691 2692 static inline void clear_inode_flag(struct inode *inode, int flag) 2693 { 2694 clear_bit(flag, F2FS_I(inode)->flags); 2695 __mark_inode_dirty_flag(inode, flag, false); 2696 } 2697 2698 static inline bool f2fs_verity_in_progress(struct inode *inode) 2699 { 2700 return IS_ENABLED(CONFIG_FS_VERITY) && 2701 is_inode_flag_set(inode, FI_VERITY_IN_PROGRESS); 2702 } 2703 2704 static inline void set_acl_inode(struct inode *inode, umode_t mode) 2705 { 2706 F2FS_I(inode)->i_acl_mode = mode; 2707 set_inode_flag(inode, FI_ACL_MODE); 2708 f2fs_mark_inode_dirty_sync(inode, false); 2709 } 2710 2711 static inline void f2fs_i_links_write(struct inode *inode, bool inc) 2712 { 2713 if (inc) 2714 inc_nlink(inode); 2715 else 2716 drop_nlink(inode); 2717 f2fs_mark_inode_dirty_sync(inode, true); 2718 } 2719 2720 static inline void f2fs_i_blocks_write(struct inode *inode, 2721 block_t diff, bool add, bool claim) 2722 { 2723 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE); 2724 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER); 2725 2726 /* add = 1, claim = 1 should be dquot_reserve_block in pair */ 2727 if (add) { 2728 if (claim) 2729 dquot_claim_block(inode, diff); 2730 else 2731 dquot_alloc_block_nofail(inode, diff); 2732 } else { 2733 dquot_free_block(inode, diff); 2734 } 2735 2736 f2fs_mark_inode_dirty_sync(inode, true); 2737 if (clean || recover) 2738 set_inode_flag(inode, FI_AUTO_RECOVER); 2739 } 2740 2741 static inline void f2fs_i_size_write(struct inode *inode, loff_t i_size) 2742 { 2743 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE); 2744 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER); 2745 2746 if (i_size_read(inode) == i_size) 2747 return; 2748 2749 i_size_write(inode, i_size); 2750 f2fs_mark_inode_dirty_sync(inode, true); 2751 if (clean || recover) 2752 set_inode_flag(inode, FI_AUTO_RECOVER); 2753 } 2754 2755 static inline void f2fs_i_depth_write(struct inode *inode, unsigned int depth) 2756 { 2757 F2FS_I(inode)->i_current_depth = depth; 2758 f2fs_mark_inode_dirty_sync(inode, true); 2759 } 2760 2761 static inline void f2fs_i_gc_failures_write(struct inode *inode, 2762 unsigned int count) 2763 { 2764 F2FS_I(inode)->i_gc_failures[GC_FAILURE_PIN] = count; 2765 f2fs_mark_inode_dirty_sync(inode, true); 2766 } 2767 2768 static inline void f2fs_i_xnid_write(struct inode *inode, nid_t xnid) 2769 { 2770 F2FS_I(inode)->i_xattr_nid = xnid; 2771 f2fs_mark_inode_dirty_sync(inode, true); 2772 } 2773 2774 static inline void f2fs_i_pino_write(struct inode *inode, nid_t pino) 2775 { 2776 F2FS_I(inode)->i_pino = pino; 2777 f2fs_mark_inode_dirty_sync(inode, true); 2778 } 2779 2780 static inline void get_inline_info(struct inode *inode, struct f2fs_inode *ri) 2781 { 2782 struct f2fs_inode_info *fi = F2FS_I(inode); 2783 2784 if (ri->i_inline & F2FS_INLINE_XATTR) 2785 set_bit(FI_INLINE_XATTR, fi->flags); 2786 if (ri->i_inline & F2FS_INLINE_DATA) 2787 set_bit(FI_INLINE_DATA, fi->flags); 2788 if (ri->i_inline & F2FS_INLINE_DENTRY) 2789 set_bit(FI_INLINE_DENTRY, fi->flags); 2790 if (ri->i_inline & F2FS_DATA_EXIST) 2791 set_bit(FI_DATA_EXIST, fi->flags); 2792 if (ri->i_inline & F2FS_INLINE_DOTS) 2793 set_bit(FI_INLINE_DOTS, fi->flags); 2794 if (ri->i_inline & F2FS_EXTRA_ATTR) 2795 set_bit(FI_EXTRA_ATTR, fi->flags); 2796 if (ri->i_inline & F2FS_PIN_FILE) 2797 set_bit(FI_PIN_FILE, fi->flags); 2798 } 2799 2800 static inline void set_raw_inline(struct inode *inode, struct f2fs_inode *ri) 2801 { 2802 ri->i_inline = 0; 2803 2804 if (is_inode_flag_set(inode, FI_INLINE_XATTR)) 2805 ri->i_inline |= F2FS_INLINE_XATTR; 2806 if (is_inode_flag_set(inode, FI_INLINE_DATA)) 2807 ri->i_inline |= F2FS_INLINE_DATA; 2808 if (is_inode_flag_set(inode, FI_INLINE_DENTRY)) 2809 ri->i_inline |= F2FS_INLINE_DENTRY; 2810 if (is_inode_flag_set(inode, FI_DATA_EXIST)) 2811 ri->i_inline |= F2FS_DATA_EXIST; 2812 if (is_inode_flag_set(inode, FI_INLINE_DOTS)) 2813 ri->i_inline |= F2FS_INLINE_DOTS; 2814 if (is_inode_flag_set(inode, FI_EXTRA_ATTR)) 2815 ri->i_inline |= F2FS_EXTRA_ATTR; 2816 if (is_inode_flag_set(inode, FI_PIN_FILE)) 2817 ri->i_inline |= F2FS_PIN_FILE; 2818 } 2819 2820 static inline int f2fs_has_extra_attr(struct inode *inode) 2821 { 2822 return is_inode_flag_set(inode, FI_EXTRA_ATTR); 2823 } 2824 2825 static inline int f2fs_has_inline_xattr(struct inode *inode) 2826 { 2827 return is_inode_flag_set(inode, FI_INLINE_XATTR); 2828 } 2829 2830 static inline int f2fs_compressed_file(struct inode *inode) 2831 { 2832 return S_ISREG(inode->i_mode) && 2833 is_inode_flag_set(inode, FI_COMPRESSED_FILE); 2834 } 2835 2836 static inline bool f2fs_need_compress_data(struct inode *inode) 2837 { 2838 int compress_mode = F2FS_OPTION(F2FS_I_SB(inode)).compress_mode; 2839 2840 if (!f2fs_compressed_file(inode)) 2841 return false; 2842 2843 if (compress_mode == COMPR_MODE_FS) 2844 return true; 2845 else if (compress_mode == COMPR_MODE_USER && 2846 is_inode_flag_set(inode, FI_ENABLE_COMPRESS)) 2847 return true; 2848 2849 return false; 2850 } 2851 2852 static inline unsigned int addrs_per_inode(struct inode *inode) 2853 { 2854 unsigned int addrs = CUR_ADDRS_PER_INODE(inode) - 2855 get_inline_xattr_addrs(inode); 2856 2857 if (!f2fs_compressed_file(inode)) 2858 return addrs; 2859 return ALIGN_DOWN(addrs, F2FS_I(inode)->i_cluster_size); 2860 } 2861 2862 static inline unsigned int addrs_per_block(struct inode *inode) 2863 { 2864 if (!f2fs_compressed_file(inode)) 2865 return DEF_ADDRS_PER_BLOCK; 2866 return ALIGN_DOWN(DEF_ADDRS_PER_BLOCK, F2FS_I(inode)->i_cluster_size); 2867 } 2868 2869 static inline void *inline_xattr_addr(struct inode *inode, struct page *page) 2870 { 2871 struct f2fs_inode *ri = F2FS_INODE(page); 2872 2873 return (void *)&(ri->i_addr[DEF_ADDRS_PER_INODE - 2874 get_inline_xattr_addrs(inode)]); 2875 } 2876 2877 static inline int inline_xattr_size(struct inode *inode) 2878 { 2879 if (f2fs_has_inline_xattr(inode)) 2880 return get_inline_xattr_addrs(inode) * sizeof(__le32); 2881 return 0; 2882 } 2883 2884 static inline int f2fs_has_inline_data(struct inode *inode) 2885 { 2886 return is_inode_flag_set(inode, FI_INLINE_DATA); 2887 } 2888 2889 static inline int f2fs_exist_data(struct inode *inode) 2890 { 2891 return is_inode_flag_set(inode, FI_DATA_EXIST); 2892 } 2893 2894 static inline int f2fs_has_inline_dots(struct inode *inode) 2895 { 2896 return is_inode_flag_set(inode, FI_INLINE_DOTS); 2897 } 2898 2899 static inline int f2fs_is_mmap_file(struct inode *inode) 2900 { 2901 return is_inode_flag_set(inode, FI_MMAP_FILE); 2902 } 2903 2904 static inline bool f2fs_is_pinned_file(struct inode *inode) 2905 { 2906 return is_inode_flag_set(inode, FI_PIN_FILE); 2907 } 2908 2909 static inline bool f2fs_is_atomic_file(struct inode *inode) 2910 { 2911 return is_inode_flag_set(inode, FI_ATOMIC_FILE); 2912 } 2913 2914 static inline bool f2fs_is_commit_atomic_write(struct inode *inode) 2915 { 2916 return is_inode_flag_set(inode, FI_ATOMIC_COMMIT); 2917 } 2918 2919 static inline bool f2fs_is_volatile_file(struct inode *inode) 2920 { 2921 return is_inode_flag_set(inode, FI_VOLATILE_FILE); 2922 } 2923 2924 static inline bool f2fs_is_first_block_written(struct inode *inode) 2925 { 2926 return is_inode_flag_set(inode, FI_FIRST_BLOCK_WRITTEN); 2927 } 2928 2929 static inline bool f2fs_is_drop_cache(struct inode *inode) 2930 { 2931 return is_inode_flag_set(inode, FI_DROP_CACHE); 2932 } 2933 2934 static inline void *inline_data_addr(struct inode *inode, struct page *page) 2935 { 2936 struct f2fs_inode *ri = F2FS_INODE(page); 2937 int extra_size = get_extra_isize(inode); 2938 2939 return (void *)&(ri->i_addr[extra_size + DEF_INLINE_RESERVED_SIZE]); 2940 } 2941 2942 static inline int f2fs_has_inline_dentry(struct inode *inode) 2943 { 2944 return is_inode_flag_set(inode, FI_INLINE_DENTRY); 2945 } 2946 2947 static inline int is_file(struct inode *inode, int type) 2948 { 2949 return F2FS_I(inode)->i_advise & type; 2950 } 2951 2952 static inline void set_file(struct inode *inode, int type) 2953 { 2954 F2FS_I(inode)->i_advise |= type; 2955 f2fs_mark_inode_dirty_sync(inode, true); 2956 } 2957 2958 static inline void clear_file(struct inode *inode, int type) 2959 { 2960 F2FS_I(inode)->i_advise &= ~type; 2961 f2fs_mark_inode_dirty_sync(inode, true); 2962 } 2963 2964 static inline bool f2fs_is_time_consistent(struct inode *inode) 2965 { 2966 if (!timespec64_equal(F2FS_I(inode)->i_disk_time, &inode->i_atime)) 2967 return false; 2968 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 1, &inode->i_ctime)) 2969 return false; 2970 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 2, &inode->i_mtime)) 2971 return false; 2972 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 3, 2973 &F2FS_I(inode)->i_crtime)) 2974 return false; 2975 return true; 2976 } 2977 2978 static inline bool f2fs_skip_inode_update(struct inode *inode, int dsync) 2979 { 2980 bool ret; 2981 2982 if (dsync) { 2983 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 2984 2985 spin_lock(&sbi->inode_lock[DIRTY_META]); 2986 ret = list_empty(&F2FS_I(inode)->gdirty_list); 2987 spin_unlock(&sbi->inode_lock[DIRTY_META]); 2988 return ret; 2989 } 2990 if (!is_inode_flag_set(inode, FI_AUTO_RECOVER) || 2991 file_keep_isize(inode) || 2992 i_size_read(inode) & ~PAGE_MASK) 2993 return false; 2994 2995 if (!f2fs_is_time_consistent(inode)) 2996 return false; 2997 2998 spin_lock(&F2FS_I(inode)->i_size_lock); 2999 ret = F2FS_I(inode)->last_disk_size == i_size_read(inode); 3000 spin_unlock(&F2FS_I(inode)->i_size_lock); 3001 3002 return ret; 3003 } 3004 3005 static inline bool f2fs_readonly(struct super_block *sb) 3006 { 3007 return sb_rdonly(sb); 3008 } 3009 3010 static inline bool f2fs_cp_error(struct f2fs_sb_info *sbi) 3011 { 3012 return is_set_ckpt_flags(sbi, CP_ERROR_FLAG); 3013 } 3014 3015 static inline bool is_dot_dotdot(const u8 *name, size_t len) 3016 { 3017 if (len == 1 && name[0] == '.') 3018 return true; 3019 3020 if (len == 2 && name[0] == '.' && name[1] == '.') 3021 return true; 3022 3023 return false; 3024 } 3025 3026 static inline bool f2fs_may_extent_tree(struct inode *inode) 3027 { 3028 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 3029 3030 if (!test_opt(sbi, EXTENT_CACHE) || 3031 is_inode_flag_set(inode, FI_NO_EXTENT) || 3032 is_inode_flag_set(inode, FI_COMPRESSED_FILE)) 3033 return false; 3034 3035 /* 3036 * for recovered files during mount do not create extents 3037 * if shrinker is not registered. 3038 */ 3039 if (list_empty(&sbi->s_list)) 3040 return false; 3041 3042 return S_ISREG(inode->i_mode); 3043 } 3044 3045 static inline void *f2fs_kmalloc(struct f2fs_sb_info *sbi, 3046 size_t size, gfp_t flags) 3047 { 3048 if (time_to_inject(sbi, FAULT_KMALLOC)) { 3049 f2fs_show_injection_info(sbi, FAULT_KMALLOC); 3050 return NULL; 3051 } 3052 3053 return kmalloc(size, flags); 3054 } 3055 3056 static inline void *f2fs_kzalloc(struct f2fs_sb_info *sbi, 3057 size_t size, gfp_t flags) 3058 { 3059 return f2fs_kmalloc(sbi, size, flags | __GFP_ZERO); 3060 } 3061 3062 static inline void *f2fs_kvmalloc(struct f2fs_sb_info *sbi, 3063 size_t size, gfp_t flags) 3064 { 3065 if (time_to_inject(sbi, FAULT_KVMALLOC)) { 3066 f2fs_show_injection_info(sbi, FAULT_KVMALLOC); 3067 return NULL; 3068 } 3069 3070 return kvmalloc(size, flags); 3071 } 3072 3073 static inline void *f2fs_kvzalloc(struct f2fs_sb_info *sbi, 3074 size_t size, gfp_t flags) 3075 { 3076 return f2fs_kvmalloc(sbi, size, flags | __GFP_ZERO); 3077 } 3078 3079 static inline int get_extra_isize(struct inode *inode) 3080 { 3081 return F2FS_I(inode)->i_extra_isize / sizeof(__le32); 3082 } 3083 3084 static inline int get_inline_xattr_addrs(struct inode *inode) 3085 { 3086 return F2FS_I(inode)->i_inline_xattr_size; 3087 } 3088 3089 #define f2fs_get_inode_mode(i) \ 3090 ((is_inode_flag_set(i, FI_ACL_MODE)) ? \ 3091 (F2FS_I(i)->i_acl_mode) : ((i)->i_mode)) 3092 3093 #define F2FS_TOTAL_EXTRA_ATTR_SIZE \ 3094 (offsetof(struct f2fs_inode, i_extra_end) - \ 3095 offsetof(struct f2fs_inode, i_extra_isize)) \ 3096 3097 #define F2FS_OLD_ATTRIBUTE_SIZE (offsetof(struct f2fs_inode, i_addr)) 3098 #define F2FS_FITS_IN_INODE(f2fs_inode, extra_isize, field) \ 3099 ((offsetof(typeof(*(f2fs_inode)), field) + \ 3100 sizeof((f2fs_inode)->field)) \ 3101 <= (F2FS_OLD_ATTRIBUTE_SIZE + (extra_isize))) \ 3102 3103 #define DEFAULT_IOSTAT_PERIOD_MS 3000 3104 #define MIN_IOSTAT_PERIOD_MS 100 3105 /* maximum period of iostat tracing is 1 day */ 3106 #define MAX_IOSTAT_PERIOD_MS 8640000 3107 3108 static inline void f2fs_reset_iostat(struct f2fs_sb_info *sbi) 3109 { 3110 int i; 3111 3112 spin_lock(&sbi->iostat_lock); 3113 for (i = 0; i < NR_IO_TYPE; i++) { 3114 sbi->rw_iostat[i] = 0; 3115 sbi->prev_rw_iostat[i] = 0; 3116 } 3117 spin_unlock(&sbi->iostat_lock); 3118 } 3119 3120 extern void f2fs_record_iostat(struct f2fs_sb_info *sbi); 3121 3122 static inline void f2fs_update_iostat(struct f2fs_sb_info *sbi, 3123 enum iostat_type type, unsigned long long io_bytes) 3124 { 3125 if (!sbi->iostat_enable) 3126 return; 3127 spin_lock(&sbi->iostat_lock); 3128 sbi->rw_iostat[type] += io_bytes; 3129 3130 if (type == APP_WRITE_IO || type == APP_DIRECT_IO) 3131 sbi->rw_iostat[APP_BUFFERED_IO] = 3132 sbi->rw_iostat[APP_WRITE_IO] - 3133 sbi->rw_iostat[APP_DIRECT_IO]; 3134 3135 if (type == APP_READ_IO || type == APP_DIRECT_READ_IO) 3136 sbi->rw_iostat[APP_BUFFERED_READ_IO] = 3137 sbi->rw_iostat[APP_READ_IO] - 3138 sbi->rw_iostat[APP_DIRECT_READ_IO]; 3139 spin_unlock(&sbi->iostat_lock); 3140 3141 f2fs_record_iostat(sbi); 3142 } 3143 3144 #define __is_large_section(sbi) ((sbi)->segs_per_sec > 1) 3145 3146 #define __is_meta_io(fio) (PAGE_TYPE_OF_BIO((fio)->type) == META) 3147 3148 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi, 3149 block_t blkaddr, int type); 3150 static inline void verify_blkaddr(struct f2fs_sb_info *sbi, 3151 block_t blkaddr, int type) 3152 { 3153 if (!f2fs_is_valid_blkaddr(sbi, blkaddr, type)) { 3154 f2fs_err(sbi, "invalid blkaddr: %u, type: %d, run fsck to fix.", 3155 blkaddr, type); 3156 f2fs_bug_on(sbi, 1); 3157 } 3158 } 3159 3160 static inline bool __is_valid_data_blkaddr(block_t blkaddr) 3161 { 3162 if (blkaddr == NEW_ADDR || blkaddr == NULL_ADDR || 3163 blkaddr == COMPRESS_ADDR) 3164 return false; 3165 return true; 3166 } 3167 3168 static inline void f2fs_set_page_private(struct page *page, 3169 unsigned long data) 3170 { 3171 if (PagePrivate(page)) 3172 return; 3173 3174 attach_page_private(page, (void *)data); 3175 } 3176 3177 static inline void f2fs_clear_page_private(struct page *page) 3178 { 3179 detach_page_private(page); 3180 } 3181 3182 /* 3183 * file.c 3184 */ 3185 int f2fs_sync_file(struct file *file, loff_t start, loff_t end, int datasync); 3186 void f2fs_truncate_data_blocks(struct dnode_of_data *dn); 3187 int f2fs_do_truncate_blocks(struct inode *inode, u64 from, bool lock); 3188 int f2fs_truncate_blocks(struct inode *inode, u64 from, bool lock); 3189 int f2fs_truncate(struct inode *inode); 3190 int f2fs_getattr(struct user_namespace *mnt_userns, const struct path *path, 3191 struct kstat *stat, u32 request_mask, unsigned int flags); 3192 int f2fs_setattr(struct user_namespace *mnt_userns, struct dentry *dentry, 3193 struct iattr *attr); 3194 int f2fs_truncate_hole(struct inode *inode, pgoff_t pg_start, pgoff_t pg_end); 3195 void f2fs_truncate_data_blocks_range(struct dnode_of_data *dn, int count); 3196 int f2fs_precache_extents(struct inode *inode); 3197 long f2fs_ioctl(struct file *filp, unsigned int cmd, unsigned long arg); 3198 long f2fs_compat_ioctl(struct file *file, unsigned int cmd, unsigned long arg); 3199 int f2fs_transfer_project_quota(struct inode *inode, kprojid_t kprojid); 3200 int f2fs_pin_file_control(struct inode *inode, bool inc); 3201 3202 /* 3203 * inode.c 3204 */ 3205 void f2fs_set_inode_flags(struct inode *inode); 3206 bool f2fs_inode_chksum_verify(struct f2fs_sb_info *sbi, struct page *page); 3207 void f2fs_inode_chksum_set(struct f2fs_sb_info *sbi, struct page *page); 3208 struct inode *f2fs_iget(struct super_block *sb, unsigned long ino); 3209 struct inode *f2fs_iget_retry(struct super_block *sb, unsigned long ino); 3210 int f2fs_try_to_free_nats(struct f2fs_sb_info *sbi, int nr_shrink); 3211 void f2fs_update_inode(struct inode *inode, struct page *node_page); 3212 void f2fs_update_inode_page(struct inode *inode); 3213 int f2fs_write_inode(struct inode *inode, struct writeback_control *wbc); 3214 void f2fs_evict_inode(struct inode *inode); 3215 void f2fs_handle_failed_inode(struct inode *inode); 3216 3217 /* 3218 * namei.c 3219 */ 3220 int f2fs_update_extension_list(struct f2fs_sb_info *sbi, const char *name, 3221 bool hot, bool set); 3222 struct dentry *f2fs_get_parent(struct dentry *child); 3223 3224 /* 3225 * dir.c 3226 */ 3227 unsigned char f2fs_get_de_type(struct f2fs_dir_entry *de); 3228 int f2fs_init_casefolded_name(const struct inode *dir, 3229 struct f2fs_filename *fname); 3230 int f2fs_setup_filename(struct inode *dir, const struct qstr *iname, 3231 int lookup, struct f2fs_filename *fname); 3232 int f2fs_prepare_lookup(struct inode *dir, struct dentry *dentry, 3233 struct f2fs_filename *fname); 3234 void f2fs_free_filename(struct f2fs_filename *fname); 3235 struct f2fs_dir_entry *f2fs_find_target_dentry(const struct f2fs_dentry_ptr *d, 3236 const struct f2fs_filename *fname, int *max_slots); 3237 int f2fs_fill_dentries(struct dir_context *ctx, struct f2fs_dentry_ptr *d, 3238 unsigned int start_pos, struct fscrypt_str *fstr); 3239 void f2fs_do_make_empty_dir(struct inode *inode, struct inode *parent, 3240 struct f2fs_dentry_ptr *d); 3241 struct page *f2fs_init_inode_metadata(struct inode *inode, struct inode *dir, 3242 const struct f2fs_filename *fname, struct page *dpage); 3243 void f2fs_update_parent_metadata(struct inode *dir, struct inode *inode, 3244 unsigned int current_depth); 3245 int f2fs_room_for_filename(const void *bitmap, int slots, int max_slots); 3246 void f2fs_drop_nlink(struct inode *dir, struct inode *inode); 3247 struct f2fs_dir_entry *__f2fs_find_entry(struct inode *dir, 3248 const struct f2fs_filename *fname, 3249 struct page **res_page); 3250 struct f2fs_dir_entry *f2fs_find_entry(struct inode *dir, 3251 const struct qstr *child, struct page **res_page); 3252 struct f2fs_dir_entry *f2fs_parent_dir(struct inode *dir, struct page **p); 3253 ino_t f2fs_inode_by_name(struct inode *dir, const struct qstr *qstr, 3254 struct page **page); 3255 void f2fs_set_link(struct inode *dir, struct f2fs_dir_entry *de, 3256 struct page *page, struct inode *inode); 3257 bool f2fs_has_enough_room(struct inode *dir, struct page *ipage, 3258 const struct f2fs_filename *fname); 3259 void f2fs_update_dentry(nid_t ino, umode_t mode, struct f2fs_dentry_ptr *d, 3260 const struct fscrypt_str *name, f2fs_hash_t name_hash, 3261 unsigned int bit_pos); 3262 int f2fs_add_regular_entry(struct inode *dir, const struct f2fs_filename *fname, 3263 struct inode *inode, nid_t ino, umode_t mode); 3264 int f2fs_add_dentry(struct inode *dir, const struct f2fs_filename *fname, 3265 struct inode *inode, nid_t ino, umode_t mode); 3266 int f2fs_do_add_link(struct inode *dir, const struct qstr *name, 3267 struct inode *inode, nid_t ino, umode_t mode); 3268 void f2fs_delete_entry(struct f2fs_dir_entry *dentry, struct page *page, 3269 struct inode *dir, struct inode *inode); 3270 int f2fs_do_tmpfile(struct inode *inode, struct inode *dir); 3271 bool f2fs_empty_dir(struct inode *dir); 3272 3273 static inline int f2fs_add_link(struct dentry *dentry, struct inode *inode) 3274 { 3275 if (fscrypt_is_nokey_name(dentry)) 3276 return -ENOKEY; 3277 return f2fs_do_add_link(d_inode(dentry->d_parent), &dentry->d_name, 3278 inode, inode->i_ino, inode->i_mode); 3279 } 3280 3281 /* 3282 * super.c 3283 */ 3284 int f2fs_inode_dirtied(struct inode *inode, bool sync); 3285 void f2fs_inode_synced(struct inode *inode); 3286 int f2fs_enable_quota_files(struct f2fs_sb_info *sbi, bool rdonly); 3287 int f2fs_quota_sync(struct super_block *sb, int type); 3288 loff_t max_file_blocks(struct inode *inode); 3289 void f2fs_quota_off_umount(struct super_block *sb); 3290 int f2fs_commit_super(struct f2fs_sb_info *sbi, bool recover); 3291 int f2fs_sync_fs(struct super_block *sb, int sync); 3292 int f2fs_sanity_check_ckpt(struct f2fs_sb_info *sbi); 3293 3294 /* 3295 * hash.c 3296 */ 3297 void f2fs_hash_filename(const struct inode *dir, struct f2fs_filename *fname); 3298 3299 /* 3300 * node.c 3301 */ 3302 struct dnode_of_data; 3303 struct node_info; 3304 3305 int f2fs_check_nid_range(struct f2fs_sb_info *sbi, nid_t nid); 3306 bool f2fs_available_free_memory(struct f2fs_sb_info *sbi, int type); 3307 bool f2fs_in_warm_node_list(struct f2fs_sb_info *sbi, struct page *page); 3308 void f2fs_init_fsync_node_info(struct f2fs_sb_info *sbi); 3309 void f2fs_del_fsync_node_entry(struct f2fs_sb_info *sbi, struct page *page); 3310 void f2fs_reset_fsync_node_info(struct f2fs_sb_info *sbi); 3311 int f2fs_need_dentry_mark(struct f2fs_sb_info *sbi, nid_t nid); 3312 bool f2fs_is_checkpointed_node(struct f2fs_sb_info *sbi, nid_t nid); 3313 bool f2fs_need_inode_block_update(struct f2fs_sb_info *sbi, nid_t ino); 3314 int f2fs_get_node_info(struct f2fs_sb_info *sbi, nid_t nid, 3315 struct node_info *ni); 3316 pgoff_t f2fs_get_next_page_offset(struct dnode_of_data *dn, pgoff_t pgofs); 3317 int f2fs_get_dnode_of_data(struct dnode_of_data *dn, pgoff_t index, int mode); 3318 int f2fs_truncate_inode_blocks(struct inode *inode, pgoff_t from); 3319 int f2fs_truncate_xattr_node(struct inode *inode); 3320 int f2fs_wait_on_node_pages_writeback(struct f2fs_sb_info *sbi, 3321 unsigned int seq_id); 3322 int f2fs_remove_inode_page(struct inode *inode); 3323 struct page *f2fs_new_inode_page(struct inode *inode); 3324 struct page *f2fs_new_node_page(struct dnode_of_data *dn, unsigned int ofs); 3325 void f2fs_ra_node_page(struct f2fs_sb_info *sbi, nid_t nid); 3326 struct page *f2fs_get_node_page(struct f2fs_sb_info *sbi, pgoff_t nid); 3327 struct page *f2fs_get_node_page_ra(struct page *parent, int start); 3328 int f2fs_move_node_page(struct page *node_page, int gc_type); 3329 void f2fs_flush_inline_data(struct f2fs_sb_info *sbi); 3330 int f2fs_fsync_node_pages(struct f2fs_sb_info *sbi, struct inode *inode, 3331 struct writeback_control *wbc, bool atomic, 3332 unsigned int *seq_id); 3333 int f2fs_sync_node_pages(struct f2fs_sb_info *sbi, 3334 struct writeback_control *wbc, 3335 bool do_balance, enum iostat_type io_type); 3336 int f2fs_build_free_nids(struct f2fs_sb_info *sbi, bool sync, bool mount); 3337 bool f2fs_alloc_nid(struct f2fs_sb_info *sbi, nid_t *nid); 3338 void f2fs_alloc_nid_done(struct f2fs_sb_info *sbi, nid_t nid); 3339 void f2fs_alloc_nid_failed(struct f2fs_sb_info *sbi, nid_t nid); 3340 int f2fs_try_to_free_nids(struct f2fs_sb_info *sbi, int nr_shrink); 3341 int f2fs_recover_inline_xattr(struct inode *inode, struct page *page); 3342 int f2fs_recover_xattr_data(struct inode *inode, struct page *page); 3343 int f2fs_recover_inode_page(struct f2fs_sb_info *sbi, struct page *page); 3344 int f2fs_restore_node_summary(struct f2fs_sb_info *sbi, 3345 unsigned int segno, struct f2fs_summary_block *sum); 3346 int f2fs_flush_nat_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3347 int f2fs_build_node_manager(struct f2fs_sb_info *sbi); 3348 void f2fs_destroy_node_manager(struct f2fs_sb_info *sbi); 3349 int __init f2fs_create_node_manager_caches(void); 3350 void f2fs_destroy_node_manager_caches(void); 3351 3352 /* 3353 * segment.c 3354 */ 3355 bool f2fs_need_SSR(struct f2fs_sb_info *sbi); 3356 void f2fs_register_inmem_page(struct inode *inode, struct page *page); 3357 void f2fs_drop_inmem_pages_all(struct f2fs_sb_info *sbi, bool gc_failure); 3358 void f2fs_drop_inmem_pages(struct inode *inode); 3359 void f2fs_drop_inmem_page(struct inode *inode, struct page *page); 3360 int f2fs_commit_inmem_pages(struct inode *inode); 3361 void f2fs_balance_fs(struct f2fs_sb_info *sbi, bool need); 3362 void f2fs_balance_fs_bg(struct f2fs_sb_info *sbi, bool from_bg); 3363 int f2fs_issue_flush(struct f2fs_sb_info *sbi, nid_t ino); 3364 int f2fs_create_flush_cmd_control(struct f2fs_sb_info *sbi); 3365 int f2fs_flush_device_cache(struct f2fs_sb_info *sbi); 3366 void f2fs_destroy_flush_cmd_control(struct f2fs_sb_info *sbi, bool free); 3367 void f2fs_invalidate_blocks(struct f2fs_sb_info *sbi, block_t addr); 3368 bool f2fs_is_checkpointed_data(struct f2fs_sb_info *sbi, block_t blkaddr); 3369 void f2fs_drop_discard_cmd(struct f2fs_sb_info *sbi); 3370 void f2fs_stop_discard_thread(struct f2fs_sb_info *sbi); 3371 bool f2fs_issue_discard_timeout(struct f2fs_sb_info *sbi); 3372 void f2fs_clear_prefree_segments(struct f2fs_sb_info *sbi, 3373 struct cp_control *cpc); 3374 void f2fs_dirty_to_prefree(struct f2fs_sb_info *sbi); 3375 block_t f2fs_get_unusable_blocks(struct f2fs_sb_info *sbi); 3376 int f2fs_disable_cp_again(struct f2fs_sb_info *sbi, block_t unusable); 3377 void f2fs_release_discard_addrs(struct f2fs_sb_info *sbi); 3378 int f2fs_npages_for_summary_flush(struct f2fs_sb_info *sbi, bool for_ra); 3379 void f2fs_init_inmem_curseg(struct f2fs_sb_info *sbi); 3380 void f2fs_save_inmem_curseg(struct f2fs_sb_info *sbi); 3381 void f2fs_restore_inmem_curseg(struct f2fs_sb_info *sbi); 3382 void f2fs_get_new_segment(struct f2fs_sb_info *sbi, 3383 unsigned int *newseg, bool new_sec, int dir); 3384 void f2fs_allocate_segment_for_resize(struct f2fs_sb_info *sbi, int type, 3385 unsigned int start, unsigned int end); 3386 void f2fs_allocate_new_segment(struct f2fs_sb_info *sbi, int type); 3387 void f2fs_allocate_new_segments(struct f2fs_sb_info *sbi); 3388 int f2fs_trim_fs(struct f2fs_sb_info *sbi, struct fstrim_range *range); 3389 bool f2fs_exist_trim_candidates(struct f2fs_sb_info *sbi, 3390 struct cp_control *cpc); 3391 struct page *f2fs_get_sum_page(struct f2fs_sb_info *sbi, unsigned int segno); 3392 void f2fs_update_meta_page(struct f2fs_sb_info *sbi, void *src, 3393 block_t blk_addr); 3394 void f2fs_do_write_meta_page(struct f2fs_sb_info *sbi, struct page *page, 3395 enum iostat_type io_type); 3396 void f2fs_do_write_node_page(unsigned int nid, struct f2fs_io_info *fio); 3397 void f2fs_outplace_write_data(struct dnode_of_data *dn, 3398 struct f2fs_io_info *fio); 3399 int f2fs_inplace_write_data(struct f2fs_io_info *fio); 3400 void f2fs_do_replace_block(struct f2fs_sb_info *sbi, struct f2fs_summary *sum, 3401 block_t old_blkaddr, block_t new_blkaddr, 3402 bool recover_curseg, bool recover_newaddr, 3403 bool from_gc); 3404 void f2fs_replace_block(struct f2fs_sb_info *sbi, struct dnode_of_data *dn, 3405 block_t old_addr, block_t new_addr, 3406 unsigned char version, bool recover_curseg, 3407 bool recover_newaddr); 3408 void f2fs_allocate_data_block(struct f2fs_sb_info *sbi, struct page *page, 3409 block_t old_blkaddr, block_t *new_blkaddr, 3410 struct f2fs_summary *sum, int type, 3411 struct f2fs_io_info *fio); 3412 void f2fs_wait_on_page_writeback(struct page *page, 3413 enum page_type type, bool ordered, bool locked); 3414 void f2fs_wait_on_block_writeback(struct inode *inode, block_t blkaddr); 3415 void f2fs_wait_on_block_writeback_range(struct inode *inode, block_t blkaddr, 3416 block_t len); 3417 void f2fs_write_data_summaries(struct f2fs_sb_info *sbi, block_t start_blk); 3418 void f2fs_write_node_summaries(struct f2fs_sb_info *sbi, block_t start_blk); 3419 int f2fs_lookup_journal_in_cursum(struct f2fs_journal *journal, int type, 3420 unsigned int val, int alloc); 3421 void f2fs_flush_sit_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3422 int f2fs_fix_curseg_write_pointer(struct f2fs_sb_info *sbi); 3423 int f2fs_check_write_pointer(struct f2fs_sb_info *sbi); 3424 int f2fs_build_segment_manager(struct f2fs_sb_info *sbi); 3425 void f2fs_destroy_segment_manager(struct f2fs_sb_info *sbi); 3426 int __init f2fs_create_segment_manager_caches(void); 3427 void f2fs_destroy_segment_manager_caches(void); 3428 int f2fs_rw_hint_to_seg_type(enum rw_hint hint); 3429 enum rw_hint f2fs_io_type_to_rw_hint(struct f2fs_sb_info *sbi, 3430 enum page_type type, enum temp_type temp); 3431 unsigned int f2fs_usable_segs_in_sec(struct f2fs_sb_info *sbi, 3432 unsigned int segno); 3433 unsigned int f2fs_usable_blks_in_seg(struct f2fs_sb_info *sbi, 3434 unsigned int segno); 3435 3436 /* 3437 * checkpoint.c 3438 */ 3439 void f2fs_stop_checkpoint(struct f2fs_sb_info *sbi, bool end_io); 3440 struct page *f2fs_grab_meta_page(struct f2fs_sb_info *sbi, pgoff_t index); 3441 struct page *f2fs_get_meta_page(struct f2fs_sb_info *sbi, pgoff_t index); 3442 struct page *f2fs_get_meta_page_retry(struct f2fs_sb_info *sbi, pgoff_t index); 3443 struct page *f2fs_get_tmp_page(struct f2fs_sb_info *sbi, pgoff_t index); 3444 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi, 3445 block_t blkaddr, int type); 3446 int f2fs_ra_meta_pages(struct f2fs_sb_info *sbi, block_t start, int nrpages, 3447 int type, bool sync); 3448 void f2fs_ra_meta_pages_cond(struct f2fs_sb_info *sbi, pgoff_t index); 3449 long f2fs_sync_meta_pages(struct f2fs_sb_info *sbi, enum page_type type, 3450 long nr_to_write, enum iostat_type io_type); 3451 void f2fs_add_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type); 3452 void f2fs_remove_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type); 3453 void f2fs_release_ino_entry(struct f2fs_sb_info *sbi, bool all); 3454 bool f2fs_exist_written_data(struct f2fs_sb_info *sbi, nid_t ino, int mode); 3455 void f2fs_set_dirty_device(struct f2fs_sb_info *sbi, nid_t ino, 3456 unsigned int devidx, int type); 3457 bool f2fs_is_dirty_device(struct f2fs_sb_info *sbi, nid_t ino, 3458 unsigned int devidx, int type); 3459 int f2fs_sync_inode_meta(struct f2fs_sb_info *sbi); 3460 int f2fs_acquire_orphan_inode(struct f2fs_sb_info *sbi); 3461 void f2fs_release_orphan_inode(struct f2fs_sb_info *sbi); 3462 void f2fs_add_orphan_inode(struct inode *inode); 3463 void f2fs_remove_orphan_inode(struct f2fs_sb_info *sbi, nid_t ino); 3464 int f2fs_recover_orphan_inodes(struct f2fs_sb_info *sbi); 3465 int f2fs_get_valid_checkpoint(struct f2fs_sb_info *sbi); 3466 void f2fs_update_dirty_page(struct inode *inode, struct page *page); 3467 void f2fs_remove_dirty_inode(struct inode *inode); 3468 int f2fs_sync_dirty_inodes(struct f2fs_sb_info *sbi, enum inode_type type); 3469 void f2fs_wait_on_all_pages(struct f2fs_sb_info *sbi, int type); 3470 u64 f2fs_get_sectors_written(struct f2fs_sb_info *sbi); 3471 int f2fs_write_checkpoint(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3472 void f2fs_init_ino_entry_info(struct f2fs_sb_info *sbi); 3473 int __init f2fs_create_checkpoint_caches(void); 3474 void f2fs_destroy_checkpoint_caches(void); 3475 int f2fs_issue_checkpoint(struct f2fs_sb_info *sbi); 3476 int f2fs_start_ckpt_thread(struct f2fs_sb_info *sbi); 3477 void f2fs_stop_ckpt_thread(struct f2fs_sb_info *sbi); 3478 void f2fs_init_ckpt_req_control(struct f2fs_sb_info *sbi); 3479 3480 /* 3481 * data.c 3482 */ 3483 int __init f2fs_init_bioset(void); 3484 void f2fs_destroy_bioset(void); 3485 int f2fs_init_bio_entry_cache(void); 3486 void f2fs_destroy_bio_entry_cache(void); 3487 void f2fs_submit_bio(struct f2fs_sb_info *sbi, 3488 struct bio *bio, enum page_type type); 3489 void f2fs_submit_merged_write(struct f2fs_sb_info *sbi, enum page_type type); 3490 void f2fs_submit_merged_write_cond(struct f2fs_sb_info *sbi, 3491 struct inode *inode, struct page *page, 3492 nid_t ino, enum page_type type); 3493 void f2fs_submit_merged_ipu_write(struct f2fs_sb_info *sbi, 3494 struct bio **bio, struct page *page); 3495 void f2fs_flush_merged_writes(struct f2fs_sb_info *sbi); 3496 int f2fs_submit_page_bio(struct f2fs_io_info *fio); 3497 int f2fs_merge_page_bio(struct f2fs_io_info *fio); 3498 void f2fs_submit_page_write(struct f2fs_io_info *fio); 3499 struct block_device *f2fs_target_device(struct f2fs_sb_info *sbi, 3500 block_t blk_addr, struct bio *bio); 3501 int f2fs_target_device_index(struct f2fs_sb_info *sbi, block_t blkaddr); 3502 void f2fs_set_data_blkaddr(struct dnode_of_data *dn); 3503 void f2fs_update_data_blkaddr(struct dnode_of_data *dn, block_t blkaddr); 3504 int f2fs_reserve_new_blocks(struct dnode_of_data *dn, blkcnt_t count); 3505 int f2fs_reserve_new_block(struct dnode_of_data *dn); 3506 int f2fs_get_block(struct dnode_of_data *dn, pgoff_t index); 3507 int f2fs_preallocate_blocks(struct kiocb *iocb, struct iov_iter *from); 3508 int f2fs_reserve_block(struct dnode_of_data *dn, pgoff_t index); 3509 struct page *f2fs_get_read_data_page(struct inode *inode, pgoff_t index, 3510 int op_flags, bool for_write); 3511 struct page *f2fs_find_data_page(struct inode *inode, pgoff_t index); 3512 struct page *f2fs_get_lock_data_page(struct inode *inode, pgoff_t index, 3513 bool for_write); 3514 struct page *f2fs_get_new_data_page(struct inode *inode, 3515 struct page *ipage, pgoff_t index, bool new_i_size); 3516 int f2fs_do_write_data_page(struct f2fs_io_info *fio); 3517 void f2fs_do_map_lock(struct f2fs_sb_info *sbi, int flag, bool lock); 3518 int f2fs_map_blocks(struct inode *inode, struct f2fs_map_blocks *map, 3519 int create, int flag); 3520 int f2fs_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, 3521 u64 start, u64 len); 3522 int f2fs_encrypt_one_page(struct f2fs_io_info *fio); 3523 bool f2fs_should_update_inplace(struct inode *inode, struct f2fs_io_info *fio); 3524 bool f2fs_should_update_outplace(struct inode *inode, struct f2fs_io_info *fio); 3525 int f2fs_write_single_data_page(struct page *page, int *submitted, 3526 struct bio **bio, sector_t *last_block, 3527 struct writeback_control *wbc, 3528 enum iostat_type io_type, 3529 int compr_blocks, bool allow_balance); 3530 void f2fs_invalidate_page(struct page *page, unsigned int offset, 3531 unsigned int length); 3532 int f2fs_release_page(struct page *page, gfp_t wait); 3533 #ifdef CONFIG_MIGRATION 3534 int f2fs_migrate_page(struct address_space *mapping, struct page *newpage, 3535 struct page *page, enum migrate_mode mode); 3536 #endif 3537 bool f2fs_overwrite_io(struct inode *inode, loff_t pos, size_t len); 3538 void f2fs_clear_page_cache_dirty_tag(struct page *page); 3539 int f2fs_init_post_read_processing(void); 3540 void f2fs_destroy_post_read_processing(void); 3541 int f2fs_init_post_read_wq(struct f2fs_sb_info *sbi); 3542 void f2fs_destroy_post_read_wq(struct f2fs_sb_info *sbi); 3543 3544 /* 3545 * gc.c 3546 */ 3547 int f2fs_start_gc_thread(struct f2fs_sb_info *sbi); 3548 void f2fs_stop_gc_thread(struct f2fs_sb_info *sbi); 3549 block_t f2fs_start_bidx_of_node(unsigned int node_ofs, struct inode *inode); 3550 int f2fs_gc(struct f2fs_sb_info *sbi, bool sync, bool background, 3551 unsigned int segno); 3552 void f2fs_build_gc_manager(struct f2fs_sb_info *sbi); 3553 int f2fs_resize_fs(struct f2fs_sb_info *sbi, __u64 block_count); 3554 int __init f2fs_create_garbage_collection_cache(void); 3555 void f2fs_destroy_garbage_collection_cache(void); 3556 3557 /* 3558 * recovery.c 3559 */ 3560 int f2fs_recover_fsync_data(struct f2fs_sb_info *sbi, bool check_only); 3561 bool f2fs_space_for_roll_forward(struct f2fs_sb_info *sbi); 3562 3563 /* 3564 * debug.c 3565 */ 3566 #ifdef CONFIG_F2FS_STAT_FS 3567 struct f2fs_stat_info { 3568 struct list_head stat_list; 3569 struct f2fs_sb_info *sbi; 3570 int all_area_segs, sit_area_segs, nat_area_segs, ssa_area_segs; 3571 int main_area_segs, main_area_sections, main_area_zones; 3572 unsigned long long hit_largest, hit_cached, hit_rbtree; 3573 unsigned long long hit_total, total_ext; 3574 int ext_tree, zombie_tree, ext_node; 3575 int ndirty_node, ndirty_dent, ndirty_meta, ndirty_imeta; 3576 int ndirty_data, ndirty_qdata; 3577 int inmem_pages; 3578 unsigned int ndirty_dirs, ndirty_files, nquota_files, ndirty_all; 3579 int nats, dirty_nats, sits, dirty_sits; 3580 int free_nids, avail_nids, alloc_nids; 3581 int total_count, utilization; 3582 int bg_gc, nr_wb_cp_data, nr_wb_data; 3583 int nr_rd_data, nr_rd_node, nr_rd_meta; 3584 int nr_dio_read, nr_dio_write; 3585 unsigned int io_skip_bggc, other_skip_bggc; 3586 int nr_flushing, nr_flushed, flush_list_empty; 3587 int nr_discarding, nr_discarded; 3588 int nr_discard_cmd; 3589 unsigned int undiscard_blks; 3590 int nr_issued_ckpt, nr_total_ckpt, nr_queued_ckpt; 3591 unsigned int cur_ckpt_time, peak_ckpt_time; 3592 int inline_xattr, inline_inode, inline_dir, append, update, orphans; 3593 int compr_inode; 3594 unsigned long long compr_blocks; 3595 int aw_cnt, max_aw_cnt, vw_cnt, max_vw_cnt; 3596 unsigned int valid_count, valid_node_count, valid_inode_count, discard_blks; 3597 unsigned int bimodal, avg_vblocks; 3598 int util_free, util_valid, util_invalid; 3599 int rsvd_segs, overp_segs; 3600 int dirty_count, node_pages, meta_pages; 3601 int prefree_count, call_count, cp_count, bg_cp_count; 3602 int tot_segs, node_segs, data_segs, free_segs, free_secs; 3603 int bg_node_segs, bg_data_segs; 3604 int tot_blks, data_blks, node_blks; 3605 int bg_data_blks, bg_node_blks; 3606 unsigned long long skipped_atomic_files[2]; 3607 int curseg[NR_CURSEG_TYPE]; 3608 int cursec[NR_CURSEG_TYPE]; 3609 int curzone[NR_CURSEG_TYPE]; 3610 unsigned int dirty_seg[NR_CURSEG_TYPE]; 3611 unsigned int full_seg[NR_CURSEG_TYPE]; 3612 unsigned int valid_blks[NR_CURSEG_TYPE]; 3613 3614 unsigned int meta_count[META_MAX]; 3615 unsigned int segment_count[2]; 3616 unsigned int block_count[2]; 3617 unsigned int inplace_count; 3618 unsigned long long base_mem, cache_mem, page_mem; 3619 }; 3620 3621 static inline struct f2fs_stat_info *F2FS_STAT(struct f2fs_sb_info *sbi) 3622 { 3623 return (struct f2fs_stat_info *)sbi->stat_info; 3624 } 3625 3626 #define stat_inc_cp_count(si) ((si)->cp_count++) 3627 #define stat_inc_bg_cp_count(si) ((si)->bg_cp_count++) 3628 #define stat_inc_call_count(si) ((si)->call_count++) 3629 #define stat_inc_bggc_count(si) ((si)->bg_gc++) 3630 #define stat_io_skip_bggc_count(sbi) ((sbi)->io_skip_bggc++) 3631 #define stat_other_skip_bggc_count(sbi) ((sbi)->other_skip_bggc++) 3632 #define stat_inc_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]++) 3633 #define stat_dec_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]--) 3634 #define stat_inc_total_hit(sbi) (atomic64_inc(&(sbi)->total_hit_ext)) 3635 #define stat_inc_rbtree_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_rbtree)) 3636 #define stat_inc_largest_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_largest)) 3637 #define stat_inc_cached_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_cached)) 3638 #define stat_inc_inline_xattr(inode) \ 3639 do { \ 3640 if (f2fs_has_inline_xattr(inode)) \ 3641 (atomic_inc(&F2FS_I_SB(inode)->inline_xattr)); \ 3642 } while (0) 3643 #define stat_dec_inline_xattr(inode) \ 3644 do { \ 3645 if (f2fs_has_inline_xattr(inode)) \ 3646 (atomic_dec(&F2FS_I_SB(inode)->inline_xattr)); \ 3647 } while (0) 3648 #define stat_inc_inline_inode(inode) \ 3649 do { \ 3650 if (f2fs_has_inline_data(inode)) \ 3651 (atomic_inc(&F2FS_I_SB(inode)->inline_inode)); \ 3652 } while (0) 3653 #define stat_dec_inline_inode(inode) \ 3654 do { \ 3655 if (f2fs_has_inline_data(inode)) \ 3656 (atomic_dec(&F2FS_I_SB(inode)->inline_inode)); \ 3657 } while (0) 3658 #define stat_inc_inline_dir(inode) \ 3659 do { \ 3660 if (f2fs_has_inline_dentry(inode)) \ 3661 (atomic_inc(&F2FS_I_SB(inode)->inline_dir)); \ 3662 } while (0) 3663 #define stat_dec_inline_dir(inode) \ 3664 do { \ 3665 if (f2fs_has_inline_dentry(inode)) \ 3666 (atomic_dec(&F2FS_I_SB(inode)->inline_dir)); \ 3667 } while (0) 3668 #define stat_inc_compr_inode(inode) \ 3669 do { \ 3670 if (f2fs_compressed_file(inode)) \ 3671 (atomic_inc(&F2FS_I_SB(inode)->compr_inode)); \ 3672 } while (0) 3673 #define stat_dec_compr_inode(inode) \ 3674 do { \ 3675 if (f2fs_compressed_file(inode)) \ 3676 (atomic_dec(&F2FS_I_SB(inode)->compr_inode)); \ 3677 } while (0) 3678 #define stat_add_compr_blocks(inode, blocks) \ 3679 (atomic64_add(blocks, &F2FS_I_SB(inode)->compr_blocks)) 3680 #define stat_sub_compr_blocks(inode, blocks) \ 3681 (atomic64_sub(blocks, &F2FS_I_SB(inode)->compr_blocks)) 3682 #define stat_inc_meta_count(sbi, blkaddr) \ 3683 do { \ 3684 if (blkaddr < SIT_I(sbi)->sit_base_addr) \ 3685 atomic_inc(&(sbi)->meta_count[META_CP]); \ 3686 else if (blkaddr < NM_I(sbi)->nat_blkaddr) \ 3687 atomic_inc(&(sbi)->meta_count[META_SIT]); \ 3688 else if (blkaddr < SM_I(sbi)->ssa_blkaddr) \ 3689 atomic_inc(&(sbi)->meta_count[META_NAT]); \ 3690 else if (blkaddr < SM_I(sbi)->main_blkaddr) \ 3691 atomic_inc(&(sbi)->meta_count[META_SSA]); \ 3692 } while (0) 3693 #define stat_inc_seg_type(sbi, curseg) \ 3694 ((sbi)->segment_count[(curseg)->alloc_type]++) 3695 #define stat_inc_block_count(sbi, curseg) \ 3696 ((sbi)->block_count[(curseg)->alloc_type]++) 3697 #define stat_inc_inplace_blocks(sbi) \ 3698 (atomic_inc(&(sbi)->inplace_count)) 3699 #define stat_update_max_atomic_write(inode) \ 3700 do { \ 3701 int cur = F2FS_I_SB(inode)->atomic_files; \ 3702 int max = atomic_read(&F2FS_I_SB(inode)->max_aw_cnt); \ 3703 if (cur > max) \ 3704 atomic_set(&F2FS_I_SB(inode)->max_aw_cnt, cur); \ 3705 } while (0) 3706 #define stat_inc_volatile_write(inode) \ 3707 (atomic_inc(&F2FS_I_SB(inode)->vw_cnt)) 3708 #define stat_dec_volatile_write(inode) \ 3709 (atomic_dec(&F2FS_I_SB(inode)->vw_cnt)) 3710 #define stat_update_max_volatile_write(inode) \ 3711 do { \ 3712 int cur = atomic_read(&F2FS_I_SB(inode)->vw_cnt); \ 3713 int max = atomic_read(&F2FS_I_SB(inode)->max_vw_cnt); \ 3714 if (cur > max) \ 3715 atomic_set(&F2FS_I_SB(inode)->max_vw_cnt, cur); \ 3716 } while (0) 3717 #define stat_inc_seg_count(sbi, type, gc_type) \ 3718 do { \ 3719 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 3720 si->tot_segs++; \ 3721 if ((type) == SUM_TYPE_DATA) { \ 3722 si->data_segs++; \ 3723 si->bg_data_segs += (gc_type == BG_GC) ? 1 : 0; \ 3724 } else { \ 3725 si->node_segs++; \ 3726 si->bg_node_segs += (gc_type == BG_GC) ? 1 : 0; \ 3727 } \ 3728 } while (0) 3729 3730 #define stat_inc_tot_blk_count(si, blks) \ 3731 ((si)->tot_blks += (blks)) 3732 3733 #define stat_inc_data_blk_count(sbi, blks, gc_type) \ 3734 do { \ 3735 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 3736 stat_inc_tot_blk_count(si, blks); \ 3737 si->data_blks += (blks); \ 3738 si->bg_data_blks += ((gc_type) == BG_GC) ? (blks) : 0; \ 3739 } while (0) 3740 3741 #define stat_inc_node_blk_count(sbi, blks, gc_type) \ 3742 do { \ 3743 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 3744 stat_inc_tot_blk_count(si, blks); \ 3745 si->node_blks += (blks); \ 3746 si->bg_node_blks += ((gc_type) == BG_GC) ? (blks) : 0; \ 3747 } while (0) 3748 3749 int f2fs_build_stats(struct f2fs_sb_info *sbi); 3750 void f2fs_destroy_stats(struct f2fs_sb_info *sbi); 3751 void __init f2fs_create_root_stats(void); 3752 void f2fs_destroy_root_stats(void); 3753 void f2fs_update_sit_info(struct f2fs_sb_info *sbi); 3754 #else 3755 #define stat_inc_cp_count(si) do { } while (0) 3756 #define stat_inc_bg_cp_count(si) do { } while (0) 3757 #define stat_inc_call_count(si) do { } while (0) 3758 #define stat_inc_bggc_count(si) do { } while (0) 3759 #define stat_io_skip_bggc_count(sbi) do { } while (0) 3760 #define stat_other_skip_bggc_count(sbi) do { } while (0) 3761 #define stat_inc_dirty_inode(sbi, type) do { } while (0) 3762 #define stat_dec_dirty_inode(sbi, type) do { } while (0) 3763 #define stat_inc_total_hit(sbi) do { } while (0) 3764 #define stat_inc_rbtree_node_hit(sbi) do { } while (0) 3765 #define stat_inc_largest_node_hit(sbi) do { } while (0) 3766 #define stat_inc_cached_node_hit(sbi) do { } while (0) 3767 #define stat_inc_inline_xattr(inode) do { } while (0) 3768 #define stat_dec_inline_xattr(inode) do { } while (0) 3769 #define stat_inc_inline_inode(inode) do { } while (0) 3770 #define stat_dec_inline_inode(inode) do { } while (0) 3771 #define stat_inc_inline_dir(inode) do { } while (0) 3772 #define stat_dec_inline_dir(inode) do { } while (0) 3773 #define stat_inc_compr_inode(inode) do { } while (0) 3774 #define stat_dec_compr_inode(inode) do { } while (0) 3775 #define stat_add_compr_blocks(inode, blocks) do { } while (0) 3776 #define stat_sub_compr_blocks(inode, blocks) do { } while (0) 3777 #define stat_update_max_atomic_write(inode) do { } while (0) 3778 #define stat_inc_volatile_write(inode) do { } while (0) 3779 #define stat_dec_volatile_write(inode) do { } while (0) 3780 #define stat_update_max_volatile_write(inode) do { } while (0) 3781 #define stat_inc_meta_count(sbi, blkaddr) do { } while (0) 3782 #define stat_inc_seg_type(sbi, curseg) do { } while (0) 3783 #define stat_inc_block_count(sbi, curseg) do { } while (0) 3784 #define stat_inc_inplace_blocks(sbi) do { } while (0) 3785 #define stat_inc_seg_count(sbi, type, gc_type) do { } while (0) 3786 #define stat_inc_tot_blk_count(si, blks) do { } while (0) 3787 #define stat_inc_data_blk_count(sbi, blks, gc_type) do { } while (0) 3788 #define stat_inc_node_blk_count(sbi, blks, gc_type) do { } while (0) 3789 3790 static inline int f2fs_build_stats(struct f2fs_sb_info *sbi) { return 0; } 3791 static inline void f2fs_destroy_stats(struct f2fs_sb_info *sbi) { } 3792 static inline void __init f2fs_create_root_stats(void) { } 3793 static inline void f2fs_destroy_root_stats(void) { } 3794 static inline void f2fs_update_sit_info(struct f2fs_sb_info *sbi) {} 3795 #endif 3796 3797 extern const struct file_operations f2fs_dir_operations; 3798 extern const struct file_operations f2fs_file_operations; 3799 extern const struct inode_operations f2fs_file_inode_operations; 3800 extern const struct address_space_operations f2fs_dblock_aops; 3801 extern const struct address_space_operations f2fs_node_aops; 3802 extern const struct address_space_operations f2fs_meta_aops; 3803 extern const struct inode_operations f2fs_dir_inode_operations; 3804 extern const struct inode_operations f2fs_symlink_inode_operations; 3805 extern const struct inode_operations f2fs_encrypted_symlink_inode_operations; 3806 extern const struct inode_operations f2fs_special_inode_operations; 3807 extern struct kmem_cache *f2fs_inode_entry_slab; 3808 3809 /* 3810 * inline.c 3811 */ 3812 bool f2fs_may_inline_data(struct inode *inode); 3813 bool f2fs_may_inline_dentry(struct inode *inode); 3814 void f2fs_do_read_inline_data(struct page *page, struct page *ipage); 3815 void f2fs_truncate_inline_inode(struct inode *inode, 3816 struct page *ipage, u64 from); 3817 int f2fs_read_inline_data(struct inode *inode, struct page *page); 3818 int f2fs_convert_inline_page(struct dnode_of_data *dn, struct page *page); 3819 int f2fs_convert_inline_inode(struct inode *inode); 3820 int f2fs_try_convert_inline_dir(struct inode *dir, struct dentry *dentry); 3821 int f2fs_write_inline_data(struct inode *inode, struct page *page); 3822 int f2fs_recover_inline_data(struct inode *inode, struct page *npage); 3823 struct f2fs_dir_entry *f2fs_find_in_inline_dir(struct inode *dir, 3824 const struct f2fs_filename *fname, 3825 struct page **res_page); 3826 int f2fs_make_empty_inline_dir(struct inode *inode, struct inode *parent, 3827 struct page *ipage); 3828 int f2fs_add_inline_entry(struct inode *dir, const struct f2fs_filename *fname, 3829 struct inode *inode, nid_t ino, umode_t mode); 3830 void f2fs_delete_inline_entry(struct f2fs_dir_entry *dentry, 3831 struct page *page, struct inode *dir, 3832 struct inode *inode); 3833 bool f2fs_empty_inline_dir(struct inode *dir); 3834 int f2fs_read_inline_dir(struct file *file, struct dir_context *ctx, 3835 struct fscrypt_str *fstr); 3836 int f2fs_inline_data_fiemap(struct inode *inode, 3837 struct fiemap_extent_info *fieinfo, 3838 __u64 start, __u64 len); 3839 3840 /* 3841 * shrinker.c 3842 */ 3843 unsigned long f2fs_shrink_count(struct shrinker *shrink, 3844 struct shrink_control *sc); 3845 unsigned long f2fs_shrink_scan(struct shrinker *shrink, 3846 struct shrink_control *sc); 3847 void f2fs_join_shrinker(struct f2fs_sb_info *sbi); 3848 void f2fs_leave_shrinker(struct f2fs_sb_info *sbi); 3849 3850 /* 3851 * extent_cache.c 3852 */ 3853 struct rb_entry *f2fs_lookup_rb_tree(struct rb_root_cached *root, 3854 struct rb_entry *cached_re, unsigned int ofs); 3855 struct rb_node **f2fs_lookup_rb_tree_ext(struct f2fs_sb_info *sbi, 3856 struct rb_root_cached *root, 3857 struct rb_node **parent, 3858 unsigned long long key, bool *left_most); 3859 struct rb_node **f2fs_lookup_rb_tree_for_insert(struct f2fs_sb_info *sbi, 3860 struct rb_root_cached *root, 3861 struct rb_node **parent, 3862 unsigned int ofs, bool *leftmost); 3863 struct rb_entry *f2fs_lookup_rb_tree_ret(struct rb_root_cached *root, 3864 struct rb_entry *cached_re, unsigned int ofs, 3865 struct rb_entry **prev_entry, struct rb_entry **next_entry, 3866 struct rb_node ***insert_p, struct rb_node **insert_parent, 3867 bool force, bool *leftmost); 3868 bool f2fs_check_rb_tree_consistence(struct f2fs_sb_info *sbi, 3869 struct rb_root_cached *root, bool check_key); 3870 unsigned int f2fs_shrink_extent_tree(struct f2fs_sb_info *sbi, int nr_shrink); 3871 void f2fs_init_extent_tree(struct inode *inode, struct page *ipage); 3872 void f2fs_drop_extent_tree(struct inode *inode); 3873 unsigned int f2fs_destroy_extent_node(struct inode *inode); 3874 void f2fs_destroy_extent_tree(struct inode *inode); 3875 bool f2fs_lookup_extent_cache(struct inode *inode, pgoff_t pgofs, 3876 struct extent_info *ei); 3877 void f2fs_update_extent_cache(struct dnode_of_data *dn); 3878 void f2fs_update_extent_cache_range(struct dnode_of_data *dn, 3879 pgoff_t fofs, block_t blkaddr, unsigned int len); 3880 void f2fs_init_extent_cache_info(struct f2fs_sb_info *sbi); 3881 int __init f2fs_create_extent_cache(void); 3882 void f2fs_destroy_extent_cache(void); 3883 3884 /* 3885 * sysfs.c 3886 */ 3887 int __init f2fs_init_sysfs(void); 3888 void f2fs_exit_sysfs(void); 3889 int f2fs_register_sysfs(struct f2fs_sb_info *sbi); 3890 void f2fs_unregister_sysfs(struct f2fs_sb_info *sbi); 3891 3892 /* verity.c */ 3893 extern const struct fsverity_operations f2fs_verityops; 3894 3895 /* 3896 * crypto support 3897 */ 3898 static inline bool f2fs_encrypted_file(struct inode *inode) 3899 { 3900 return IS_ENCRYPTED(inode) && S_ISREG(inode->i_mode); 3901 } 3902 3903 static inline void f2fs_set_encrypted_inode(struct inode *inode) 3904 { 3905 #ifdef CONFIG_FS_ENCRYPTION 3906 file_set_encrypt(inode); 3907 f2fs_set_inode_flags(inode); 3908 #endif 3909 } 3910 3911 /* 3912 * Returns true if the reads of the inode's data need to undergo some 3913 * postprocessing step, like decryption or authenticity verification. 3914 */ 3915 static inline bool f2fs_post_read_required(struct inode *inode) 3916 { 3917 return f2fs_encrypted_file(inode) || fsverity_active(inode) || 3918 f2fs_compressed_file(inode); 3919 } 3920 3921 /* 3922 * compress.c 3923 */ 3924 #ifdef CONFIG_F2FS_FS_COMPRESSION 3925 bool f2fs_is_compressed_page(struct page *page); 3926 struct page *f2fs_compress_control_page(struct page *page); 3927 int f2fs_prepare_compress_overwrite(struct inode *inode, 3928 struct page **pagep, pgoff_t index, void **fsdata); 3929 bool f2fs_compress_write_end(struct inode *inode, void *fsdata, 3930 pgoff_t index, unsigned copied); 3931 int f2fs_truncate_partial_cluster(struct inode *inode, u64 from, bool lock); 3932 void f2fs_compress_write_end_io(struct bio *bio, struct page *page); 3933 bool f2fs_is_compress_backend_ready(struct inode *inode); 3934 int f2fs_init_compress_mempool(void); 3935 void f2fs_destroy_compress_mempool(void); 3936 void f2fs_end_read_compressed_page(struct page *page, bool failed); 3937 bool f2fs_cluster_is_empty(struct compress_ctx *cc); 3938 bool f2fs_cluster_can_merge_page(struct compress_ctx *cc, pgoff_t index); 3939 void f2fs_compress_ctx_add_page(struct compress_ctx *cc, struct page *page); 3940 int f2fs_write_multi_pages(struct compress_ctx *cc, 3941 int *submitted, 3942 struct writeback_control *wbc, 3943 enum iostat_type io_type); 3944 int f2fs_is_compressed_cluster(struct inode *inode, pgoff_t index); 3945 int f2fs_read_multi_pages(struct compress_ctx *cc, struct bio **bio_ret, 3946 unsigned nr_pages, sector_t *last_block_in_bio, 3947 bool is_readahead, bool for_write); 3948 struct decompress_io_ctx *f2fs_alloc_dic(struct compress_ctx *cc); 3949 void f2fs_decompress_end_io(struct decompress_io_ctx *dic, bool failed); 3950 void f2fs_put_page_dic(struct page *page); 3951 int f2fs_init_compress_ctx(struct compress_ctx *cc); 3952 void f2fs_destroy_compress_ctx(struct compress_ctx *cc); 3953 void f2fs_init_compress_info(struct f2fs_sb_info *sbi); 3954 int f2fs_init_page_array_cache(struct f2fs_sb_info *sbi); 3955 void f2fs_destroy_page_array_cache(struct f2fs_sb_info *sbi); 3956 int __init f2fs_init_compress_cache(void); 3957 void f2fs_destroy_compress_cache(void); 3958 #else 3959 static inline bool f2fs_is_compressed_page(struct page *page) { return false; } 3960 static inline bool f2fs_is_compress_backend_ready(struct inode *inode) 3961 { 3962 if (!f2fs_compressed_file(inode)) 3963 return true; 3964 /* not support compression */ 3965 return false; 3966 } 3967 static inline struct page *f2fs_compress_control_page(struct page *page) 3968 { 3969 WARN_ON_ONCE(1); 3970 return ERR_PTR(-EINVAL); 3971 } 3972 static inline int f2fs_init_compress_mempool(void) { return 0; } 3973 static inline void f2fs_destroy_compress_mempool(void) { } 3974 static inline void f2fs_end_read_compressed_page(struct page *page, bool failed) 3975 { 3976 WARN_ON_ONCE(1); 3977 } 3978 static inline void f2fs_put_page_dic(struct page *page) 3979 { 3980 WARN_ON_ONCE(1); 3981 } 3982 static inline int f2fs_init_page_array_cache(struct f2fs_sb_info *sbi) { return 0; } 3983 static inline void f2fs_destroy_page_array_cache(struct f2fs_sb_info *sbi) { } 3984 static inline int __init f2fs_init_compress_cache(void) { return 0; } 3985 static inline void f2fs_destroy_compress_cache(void) { } 3986 #endif 3987 3988 static inline void set_compress_context(struct inode *inode) 3989 { 3990 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 3991 3992 F2FS_I(inode)->i_compress_algorithm = 3993 F2FS_OPTION(sbi).compress_algorithm; 3994 F2FS_I(inode)->i_log_cluster_size = 3995 F2FS_OPTION(sbi).compress_log_size; 3996 F2FS_I(inode)->i_compress_flag = 3997 F2FS_OPTION(sbi).compress_chksum ? 3998 1 << COMPRESS_CHKSUM : 0; 3999 F2FS_I(inode)->i_cluster_size = 4000 1 << F2FS_I(inode)->i_log_cluster_size; 4001 if (F2FS_I(inode)->i_compress_algorithm == COMPRESS_LZ4 && 4002 F2FS_OPTION(sbi).compress_level) 4003 F2FS_I(inode)->i_compress_flag |= 4004 F2FS_OPTION(sbi).compress_level << 4005 COMPRESS_LEVEL_OFFSET; 4006 F2FS_I(inode)->i_flags |= F2FS_COMPR_FL; 4007 set_inode_flag(inode, FI_COMPRESSED_FILE); 4008 stat_inc_compr_inode(inode); 4009 f2fs_mark_inode_dirty_sync(inode, true); 4010 } 4011 4012 static inline bool f2fs_disable_compressed_file(struct inode *inode) 4013 { 4014 struct f2fs_inode_info *fi = F2FS_I(inode); 4015 4016 if (!f2fs_compressed_file(inode)) 4017 return true; 4018 if (S_ISREG(inode->i_mode) && 4019 (get_dirty_pages(inode) || atomic_read(&fi->i_compr_blocks))) 4020 return false; 4021 4022 fi->i_flags &= ~F2FS_COMPR_FL; 4023 stat_dec_compr_inode(inode); 4024 clear_inode_flag(inode, FI_COMPRESSED_FILE); 4025 f2fs_mark_inode_dirty_sync(inode, true); 4026 return true; 4027 } 4028 4029 #define F2FS_FEATURE_FUNCS(name, flagname) \ 4030 static inline int f2fs_sb_has_##name(struct f2fs_sb_info *sbi) \ 4031 { \ 4032 return F2FS_HAS_FEATURE(sbi, F2FS_FEATURE_##flagname); \ 4033 } 4034 4035 F2FS_FEATURE_FUNCS(encrypt, ENCRYPT); 4036 F2FS_FEATURE_FUNCS(blkzoned, BLKZONED); 4037 F2FS_FEATURE_FUNCS(extra_attr, EXTRA_ATTR); 4038 F2FS_FEATURE_FUNCS(project_quota, PRJQUOTA); 4039 F2FS_FEATURE_FUNCS(inode_chksum, INODE_CHKSUM); 4040 F2FS_FEATURE_FUNCS(flexible_inline_xattr, FLEXIBLE_INLINE_XATTR); 4041 F2FS_FEATURE_FUNCS(quota_ino, QUOTA_INO); 4042 F2FS_FEATURE_FUNCS(inode_crtime, INODE_CRTIME); 4043 F2FS_FEATURE_FUNCS(lost_found, LOST_FOUND); 4044 F2FS_FEATURE_FUNCS(verity, VERITY); 4045 F2FS_FEATURE_FUNCS(sb_chksum, SB_CHKSUM); 4046 F2FS_FEATURE_FUNCS(casefold, CASEFOLD); 4047 F2FS_FEATURE_FUNCS(compression, COMPRESSION); 4048 4049 #ifdef CONFIG_BLK_DEV_ZONED 4050 static inline bool f2fs_blkz_is_seq(struct f2fs_sb_info *sbi, int devi, 4051 block_t blkaddr) 4052 { 4053 unsigned int zno = blkaddr >> sbi->log_blocks_per_blkz; 4054 4055 return test_bit(zno, FDEV(devi).blkz_seq); 4056 } 4057 #endif 4058 4059 static inline bool f2fs_hw_should_discard(struct f2fs_sb_info *sbi) 4060 { 4061 return f2fs_sb_has_blkzoned(sbi); 4062 } 4063 4064 static inline bool f2fs_bdev_support_discard(struct block_device *bdev) 4065 { 4066 return blk_queue_discard(bdev_get_queue(bdev)) || 4067 bdev_is_zoned(bdev); 4068 } 4069 4070 static inline bool f2fs_hw_support_discard(struct f2fs_sb_info *sbi) 4071 { 4072 int i; 4073 4074 if (!f2fs_is_multi_device(sbi)) 4075 return f2fs_bdev_support_discard(sbi->sb->s_bdev); 4076 4077 for (i = 0; i < sbi->s_ndevs; i++) 4078 if (f2fs_bdev_support_discard(FDEV(i).bdev)) 4079 return true; 4080 return false; 4081 } 4082 4083 static inline bool f2fs_realtime_discard_enable(struct f2fs_sb_info *sbi) 4084 { 4085 return (test_opt(sbi, DISCARD) && f2fs_hw_support_discard(sbi)) || 4086 f2fs_hw_should_discard(sbi); 4087 } 4088 4089 static inline bool f2fs_hw_is_readonly(struct f2fs_sb_info *sbi) 4090 { 4091 int i; 4092 4093 if (!f2fs_is_multi_device(sbi)) 4094 return bdev_read_only(sbi->sb->s_bdev); 4095 4096 for (i = 0; i < sbi->s_ndevs; i++) 4097 if (bdev_read_only(FDEV(i).bdev)) 4098 return true; 4099 return false; 4100 } 4101 4102 static inline bool f2fs_lfs_mode(struct f2fs_sb_info *sbi) 4103 { 4104 return F2FS_OPTION(sbi).fs_mode == FS_MODE_LFS; 4105 } 4106 4107 static inline bool f2fs_may_compress(struct inode *inode) 4108 { 4109 if (IS_SWAPFILE(inode) || f2fs_is_pinned_file(inode) || 4110 f2fs_is_atomic_file(inode) || 4111 f2fs_is_volatile_file(inode)) 4112 return false; 4113 return S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode); 4114 } 4115 4116 static inline void f2fs_i_compr_blocks_update(struct inode *inode, 4117 u64 blocks, bool add) 4118 { 4119 int diff = F2FS_I(inode)->i_cluster_size - blocks; 4120 struct f2fs_inode_info *fi = F2FS_I(inode); 4121 4122 /* don't update i_compr_blocks if saved blocks were released */ 4123 if (!add && !atomic_read(&fi->i_compr_blocks)) 4124 return; 4125 4126 if (add) { 4127 atomic_add(diff, &fi->i_compr_blocks); 4128 stat_add_compr_blocks(inode, diff); 4129 } else { 4130 atomic_sub(diff, &fi->i_compr_blocks); 4131 stat_sub_compr_blocks(inode, diff); 4132 } 4133 f2fs_mark_inode_dirty_sync(inode, true); 4134 } 4135 4136 static inline int block_unaligned_IO(struct inode *inode, 4137 struct kiocb *iocb, struct iov_iter *iter) 4138 { 4139 unsigned int i_blkbits = READ_ONCE(inode->i_blkbits); 4140 unsigned int blocksize_mask = (1 << i_blkbits) - 1; 4141 loff_t offset = iocb->ki_pos; 4142 unsigned long align = offset | iov_iter_alignment(iter); 4143 4144 return align & blocksize_mask; 4145 } 4146 4147 static inline int allow_outplace_dio(struct inode *inode, 4148 struct kiocb *iocb, struct iov_iter *iter) 4149 { 4150 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 4151 int rw = iov_iter_rw(iter); 4152 4153 return (f2fs_lfs_mode(sbi) && (rw == WRITE) && 4154 !block_unaligned_IO(inode, iocb, iter)); 4155 } 4156 4157 static inline bool f2fs_force_buffered_io(struct inode *inode, 4158 struct kiocb *iocb, struct iov_iter *iter) 4159 { 4160 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 4161 int rw = iov_iter_rw(iter); 4162 4163 if (f2fs_post_read_required(inode)) 4164 return true; 4165 if (f2fs_is_multi_device(sbi)) 4166 return true; 4167 /* 4168 * for blkzoned device, fallback direct IO to buffered IO, so 4169 * all IOs can be serialized by log-structured write. 4170 */ 4171 if (f2fs_sb_has_blkzoned(sbi)) 4172 return true; 4173 if (f2fs_lfs_mode(sbi) && (rw == WRITE)) { 4174 if (block_unaligned_IO(inode, iocb, iter)) 4175 return true; 4176 if (F2FS_IO_ALIGNED(sbi)) 4177 return true; 4178 } 4179 if (is_sbi_flag_set(F2FS_I_SB(inode), SBI_CP_DISABLED) && 4180 !IS_SWAPFILE(inode)) 4181 return true; 4182 4183 return false; 4184 } 4185 4186 static inline bool f2fs_need_verity(const struct inode *inode, pgoff_t idx) 4187 { 4188 return fsverity_active(inode) && 4189 idx < DIV_ROUND_UP(inode->i_size, PAGE_SIZE); 4190 } 4191 4192 #ifdef CONFIG_F2FS_FAULT_INJECTION 4193 extern void f2fs_build_fault_attr(struct f2fs_sb_info *sbi, unsigned int rate, 4194 unsigned int type); 4195 #else 4196 #define f2fs_build_fault_attr(sbi, rate, type) do { } while (0) 4197 #endif 4198 4199 static inline bool is_journalled_quota(struct f2fs_sb_info *sbi) 4200 { 4201 #ifdef CONFIG_QUOTA 4202 if (f2fs_sb_has_quota_ino(sbi)) 4203 return true; 4204 if (F2FS_OPTION(sbi).s_qf_names[USRQUOTA] || 4205 F2FS_OPTION(sbi).s_qf_names[GRPQUOTA] || 4206 F2FS_OPTION(sbi).s_qf_names[PRJQUOTA]) 4207 return true; 4208 #endif 4209 return false; 4210 } 4211 4212 #define EFSBADCRC EBADMSG /* Bad CRC detected */ 4213 #define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */ 4214 4215 #endif /* _LINUX_F2FS_H */ 4216