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_set_cold(inode) set_file(inode, FADVISE_COLD_BIT) 641 #define file_clear_cold(inode) clear_file(inode, FADVISE_COLD_BIT) 642 643 #define file_wrong_pino(inode) is_file(inode, FADVISE_LOST_PINO_BIT) 644 #define file_lost_pino(inode) set_file(inode, FADVISE_LOST_PINO_BIT) 645 #define file_got_pino(inode) clear_file(inode, FADVISE_LOST_PINO_BIT) 646 647 #define file_is_encrypt(inode) is_file(inode, FADVISE_ENCRYPT_BIT) 648 #define file_set_encrypt(inode) set_file(inode, FADVISE_ENCRYPT_BIT) 649 650 #define file_enc_name(inode) is_file(inode, FADVISE_ENC_NAME_BIT) 651 #define file_set_enc_name(inode) set_file(inode, FADVISE_ENC_NAME_BIT) 652 653 #define file_keep_isize(inode) is_file(inode, FADVISE_KEEP_SIZE_BIT) 654 #define file_set_keep_isize(inode) set_file(inode, FADVISE_KEEP_SIZE_BIT) 655 656 #define file_is_hot(inode) is_file(inode, FADVISE_HOT_BIT) 657 #define file_set_hot(inode) set_file(inode, FADVISE_HOT_BIT) 658 #define file_clear_hot(inode) clear_file(inode, FADVISE_HOT_BIT) 659 660 #define file_is_verity(inode) is_file(inode, FADVISE_VERITY_BIT) 661 #define file_set_verity(inode) set_file(inode, FADVISE_VERITY_BIT) 662 663 #define DEF_DIR_LEVEL 0 664 665 enum { 666 GC_FAILURE_PIN, 667 GC_FAILURE_ATOMIC, 668 MAX_GC_FAILURE 669 }; 670 671 /* used for f2fs_inode_info->flags */ 672 enum { 673 FI_NEW_INODE, /* indicate newly allocated inode */ 674 FI_DIRTY_INODE, /* indicate inode is dirty or not */ 675 FI_AUTO_RECOVER, /* indicate inode is recoverable */ 676 FI_DIRTY_DIR, /* indicate directory has dirty pages */ 677 FI_INC_LINK, /* need to increment i_nlink */ 678 FI_ACL_MODE, /* indicate acl mode */ 679 FI_NO_ALLOC, /* should not allocate any blocks */ 680 FI_FREE_NID, /* free allocated nide */ 681 FI_NO_EXTENT, /* not to use the extent cache */ 682 FI_INLINE_XATTR, /* used for inline xattr */ 683 FI_INLINE_DATA, /* used for inline data*/ 684 FI_INLINE_DENTRY, /* used for inline dentry */ 685 FI_APPEND_WRITE, /* inode has appended data */ 686 FI_UPDATE_WRITE, /* inode has in-place-update data */ 687 FI_NEED_IPU, /* used for ipu per file */ 688 FI_ATOMIC_FILE, /* indicate atomic file */ 689 FI_ATOMIC_COMMIT, /* indicate the state of atomical committing */ 690 FI_VOLATILE_FILE, /* indicate volatile file */ 691 FI_FIRST_BLOCK_WRITTEN, /* indicate #0 data block was written */ 692 FI_DROP_CACHE, /* drop dirty page cache */ 693 FI_DATA_EXIST, /* indicate data exists */ 694 FI_INLINE_DOTS, /* indicate inline dot dentries */ 695 FI_DO_DEFRAG, /* indicate defragment is running */ 696 FI_DIRTY_FILE, /* indicate regular/symlink has dirty pages */ 697 FI_NO_PREALLOC, /* indicate skipped preallocated blocks */ 698 FI_HOT_DATA, /* indicate file is hot */ 699 FI_EXTRA_ATTR, /* indicate file has extra attribute */ 700 FI_PROJ_INHERIT, /* indicate file inherits projectid */ 701 FI_PIN_FILE, /* indicate file should not be gced */ 702 FI_ATOMIC_REVOKE_REQUEST, /* request to drop atomic data */ 703 FI_VERITY_IN_PROGRESS, /* building fs-verity Merkle tree */ 704 FI_COMPRESSED_FILE, /* indicate file's data can be compressed */ 705 FI_COMPRESS_CORRUPT, /* indicate compressed cluster is corrupted */ 706 FI_MMAP_FILE, /* indicate file was mmapped */ 707 FI_ENABLE_COMPRESS, /* enable compression in "user" compression mode */ 708 FI_MAX, /* max flag, never be used */ 709 }; 710 711 struct f2fs_inode_info { 712 struct inode vfs_inode; /* serve a vfs inode */ 713 unsigned long i_flags; /* keep an inode flags for ioctl */ 714 unsigned char i_advise; /* use to give file attribute hints */ 715 unsigned char i_dir_level; /* use for dentry level for large dir */ 716 unsigned int i_current_depth; /* only for directory depth */ 717 /* for gc failure statistic */ 718 unsigned int i_gc_failures[MAX_GC_FAILURE]; 719 unsigned int i_pino; /* parent inode number */ 720 umode_t i_acl_mode; /* keep file acl mode temporarily */ 721 722 /* Use below internally in f2fs*/ 723 unsigned long flags[BITS_TO_LONGS(FI_MAX)]; /* use to pass per-file flags */ 724 struct rw_semaphore i_sem; /* protect fi info */ 725 atomic_t dirty_pages; /* # of dirty pages */ 726 f2fs_hash_t chash; /* hash value of given file name */ 727 unsigned int clevel; /* maximum level of given file name */ 728 struct task_struct *task; /* lookup and create consistency */ 729 struct task_struct *cp_task; /* separate cp/wb IO stats*/ 730 nid_t i_xattr_nid; /* node id that contains xattrs */ 731 loff_t last_disk_size; /* lastly written file size */ 732 spinlock_t i_size_lock; /* protect last_disk_size */ 733 734 #ifdef CONFIG_QUOTA 735 struct dquot *i_dquot[MAXQUOTAS]; 736 737 /* quota space reservation, managed internally by quota code */ 738 qsize_t i_reserved_quota; 739 #endif 740 struct list_head dirty_list; /* dirty list for dirs and files */ 741 struct list_head gdirty_list; /* linked in global dirty list */ 742 struct list_head inmem_ilist; /* list for inmem inodes */ 743 struct list_head inmem_pages; /* inmemory pages managed by f2fs */ 744 struct task_struct *inmem_task; /* store inmemory task */ 745 struct mutex inmem_lock; /* lock for inmemory pages */ 746 struct extent_tree *extent_tree; /* cached extent_tree entry */ 747 748 /* avoid racing between foreground op and gc */ 749 struct rw_semaphore i_gc_rwsem[2]; 750 struct rw_semaphore i_mmap_sem; 751 struct rw_semaphore i_xattr_sem; /* avoid racing between reading and changing EAs */ 752 753 int i_extra_isize; /* size of extra space located in i_addr */ 754 kprojid_t i_projid; /* id for project quota */ 755 int i_inline_xattr_size; /* inline xattr size */ 756 struct timespec64 i_crtime; /* inode creation time */ 757 struct timespec64 i_disk_time[4];/* inode disk times */ 758 759 /* for file compress */ 760 atomic_t i_compr_blocks; /* # of compressed blocks */ 761 unsigned char i_compress_algorithm; /* algorithm type */ 762 unsigned char i_log_cluster_size; /* log of cluster size */ 763 unsigned char i_compress_level; /* compress level (lz4hc,zstd) */ 764 unsigned short i_compress_flag; /* compress flag */ 765 unsigned int i_cluster_size; /* cluster size */ 766 }; 767 768 static inline void get_extent_info(struct extent_info *ext, 769 struct f2fs_extent *i_ext) 770 { 771 ext->fofs = le32_to_cpu(i_ext->fofs); 772 ext->blk = le32_to_cpu(i_ext->blk); 773 ext->len = le32_to_cpu(i_ext->len); 774 } 775 776 static inline void set_raw_extent(struct extent_info *ext, 777 struct f2fs_extent *i_ext) 778 { 779 i_ext->fofs = cpu_to_le32(ext->fofs); 780 i_ext->blk = cpu_to_le32(ext->blk); 781 i_ext->len = cpu_to_le32(ext->len); 782 } 783 784 static inline void set_extent_info(struct extent_info *ei, unsigned int fofs, 785 u32 blk, unsigned int len) 786 { 787 ei->fofs = fofs; 788 ei->blk = blk; 789 ei->len = len; 790 } 791 792 static inline bool __is_discard_mergeable(struct discard_info *back, 793 struct discard_info *front, unsigned int max_len) 794 { 795 return (back->lstart + back->len == front->lstart) && 796 (back->len + front->len <= max_len); 797 } 798 799 static inline bool __is_discard_back_mergeable(struct discard_info *cur, 800 struct discard_info *back, unsigned int max_len) 801 { 802 return __is_discard_mergeable(back, cur, max_len); 803 } 804 805 static inline bool __is_discard_front_mergeable(struct discard_info *cur, 806 struct discard_info *front, unsigned int max_len) 807 { 808 return __is_discard_mergeable(cur, front, max_len); 809 } 810 811 static inline bool __is_extent_mergeable(struct extent_info *back, 812 struct extent_info *front) 813 { 814 return (back->fofs + back->len == front->fofs && 815 back->blk + back->len == front->blk); 816 } 817 818 static inline bool __is_back_mergeable(struct extent_info *cur, 819 struct extent_info *back) 820 { 821 return __is_extent_mergeable(back, cur); 822 } 823 824 static inline bool __is_front_mergeable(struct extent_info *cur, 825 struct extent_info *front) 826 { 827 return __is_extent_mergeable(cur, front); 828 } 829 830 extern void f2fs_mark_inode_dirty_sync(struct inode *inode, bool sync); 831 static inline void __try_update_largest_extent(struct extent_tree *et, 832 struct extent_node *en) 833 { 834 if (en->ei.len > et->largest.len) { 835 et->largest = en->ei; 836 et->largest_updated = true; 837 } 838 } 839 840 /* 841 * For free nid management 842 */ 843 enum nid_state { 844 FREE_NID, /* newly added to free nid list */ 845 PREALLOC_NID, /* it is preallocated */ 846 MAX_NID_STATE, 847 }; 848 849 enum nat_state { 850 TOTAL_NAT, 851 DIRTY_NAT, 852 RECLAIMABLE_NAT, 853 MAX_NAT_STATE, 854 }; 855 856 struct f2fs_nm_info { 857 block_t nat_blkaddr; /* base disk address of NAT */ 858 nid_t max_nid; /* maximum possible node ids */ 859 nid_t available_nids; /* # of available node ids */ 860 nid_t next_scan_nid; /* the next nid to be scanned */ 861 unsigned int ram_thresh; /* control the memory footprint */ 862 unsigned int ra_nid_pages; /* # of nid pages to be readaheaded */ 863 unsigned int dirty_nats_ratio; /* control dirty nats ratio threshold */ 864 865 /* NAT cache management */ 866 struct radix_tree_root nat_root;/* root of the nat entry cache */ 867 struct radix_tree_root nat_set_root;/* root of the nat set cache */ 868 struct rw_semaphore nat_tree_lock; /* protect nat_tree_lock */ 869 struct list_head nat_entries; /* cached nat entry list (clean) */ 870 spinlock_t nat_list_lock; /* protect clean nat entry list */ 871 unsigned int nat_cnt[MAX_NAT_STATE]; /* the # of cached nat entries */ 872 unsigned int nat_blocks; /* # of nat blocks */ 873 874 /* free node ids management */ 875 struct radix_tree_root free_nid_root;/* root of the free_nid cache */ 876 struct list_head free_nid_list; /* list for free nids excluding preallocated nids */ 877 unsigned int nid_cnt[MAX_NID_STATE]; /* the number of free node id */ 878 spinlock_t nid_list_lock; /* protect nid lists ops */ 879 struct mutex build_lock; /* lock for build free nids */ 880 unsigned char **free_nid_bitmap; 881 unsigned char *nat_block_bitmap; 882 unsigned short *free_nid_count; /* free nid count of NAT block */ 883 884 /* for checkpoint */ 885 char *nat_bitmap; /* NAT bitmap pointer */ 886 887 unsigned int nat_bits_blocks; /* # of nat bits blocks */ 888 unsigned char *nat_bits; /* NAT bits blocks */ 889 unsigned char *full_nat_bits; /* full NAT pages */ 890 unsigned char *empty_nat_bits; /* empty NAT pages */ 891 #ifdef CONFIG_F2FS_CHECK_FS 892 char *nat_bitmap_mir; /* NAT bitmap mirror */ 893 #endif 894 int bitmap_size; /* bitmap size */ 895 }; 896 897 /* 898 * this structure is used as one of function parameters. 899 * all the information are dedicated to a given direct node block determined 900 * by the data offset in a file. 901 */ 902 struct dnode_of_data { 903 struct inode *inode; /* vfs inode pointer */ 904 struct page *inode_page; /* its inode page, NULL is possible */ 905 struct page *node_page; /* cached direct node page */ 906 nid_t nid; /* node id of the direct node block */ 907 unsigned int ofs_in_node; /* data offset in the node page */ 908 bool inode_page_locked; /* inode page is locked or not */ 909 bool node_changed; /* is node block changed */ 910 char cur_level; /* level of hole node page */ 911 char max_level; /* level of current page located */ 912 block_t data_blkaddr; /* block address of the node block */ 913 }; 914 915 static inline void set_new_dnode(struct dnode_of_data *dn, struct inode *inode, 916 struct page *ipage, struct page *npage, nid_t nid) 917 { 918 memset(dn, 0, sizeof(*dn)); 919 dn->inode = inode; 920 dn->inode_page = ipage; 921 dn->node_page = npage; 922 dn->nid = nid; 923 } 924 925 /* 926 * For SIT manager 927 * 928 * By default, there are 6 active log areas across the whole main area. 929 * When considering hot and cold data separation to reduce cleaning overhead, 930 * we split 3 for data logs and 3 for node logs as hot, warm, and cold types, 931 * respectively. 932 * In the current design, you should not change the numbers intentionally. 933 * Instead, as a mount option such as active_logs=x, you can use 2, 4, and 6 934 * logs individually according to the underlying devices. (default: 6) 935 * Just in case, on-disk layout covers maximum 16 logs that consist of 8 for 936 * data and 8 for node logs. 937 */ 938 #define NR_CURSEG_DATA_TYPE (3) 939 #define NR_CURSEG_NODE_TYPE (3) 940 #define NR_CURSEG_INMEM_TYPE (2) 941 #define NR_CURSEG_PERSIST_TYPE (NR_CURSEG_DATA_TYPE + NR_CURSEG_NODE_TYPE) 942 #define NR_CURSEG_TYPE (NR_CURSEG_INMEM_TYPE + NR_CURSEG_PERSIST_TYPE) 943 944 enum { 945 CURSEG_HOT_DATA = 0, /* directory entry blocks */ 946 CURSEG_WARM_DATA, /* data blocks */ 947 CURSEG_COLD_DATA, /* multimedia or GCed data blocks */ 948 CURSEG_HOT_NODE, /* direct node blocks of directory files */ 949 CURSEG_WARM_NODE, /* direct node blocks of normal files */ 950 CURSEG_COLD_NODE, /* indirect node blocks */ 951 NR_PERSISTENT_LOG, /* number of persistent log */ 952 CURSEG_COLD_DATA_PINNED = NR_PERSISTENT_LOG, 953 /* pinned file that needs consecutive block address */ 954 CURSEG_ALL_DATA_ATGC, /* SSR alloctor in hot/warm/cold data area */ 955 NO_CHECK_TYPE, /* number of persistent & inmem log */ 956 }; 957 958 struct flush_cmd { 959 struct completion wait; 960 struct llist_node llnode; 961 nid_t ino; 962 int ret; 963 }; 964 965 struct flush_cmd_control { 966 struct task_struct *f2fs_issue_flush; /* flush thread */ 967 wait_queue_head_t flush_wait_queue; /* waiting queue for wake-up */ 968 atomic_t issued_flush; /* # of issued flushes */ 969 atomic_t queued_flush; /* # of queued flushes */ 970 struct llist_head issue_list; /* list for command issue */ 971 struct llist_node *dispatch_list; /* list for command dispatch */ 972 }; 973 974 struct f2fs_sm_info { 975 struct sit_info *sit_info; /* whole segment information */ 976 struct free_segmap_info *free_info; /* free segment information */ 977 struct dirty_seglist_info *dirty_info; /* dirty segment information */ 978 struct curseg_info *curseg_array; /* active segment information */ 979 980 struct rw_semaphore curseg_lock; /* for preventing curseg change */ 981 982 block_t seg0_blkaddr; /* block address of 0'th segment */ 983 block_t main_blkaddr; /* start block address of main area */ 984 block_t ssa_blkaddr; /* start block address of SSA area */ 985 986 unsigned int segment_count; /* total # of segments */ 987 unsigned int main_segments; /* # of segments in main area */ 988 unsigned int reserved_segments; /* # of reserved segments */ 989 unsigned int ovp_segments; /* # of overprovision segments */ 990 991 /* a threshold to reclaim prefree segments */ 992 unsigned int rec_prefree_segments; 993 994 /* for batched trimming */ 995 unsigned int trim_sections; /* # of sections to trim */ 996 997 struct list_head sit_entry_set; /* sit entry set list */ 998 999 unsigned int ipu_policy; /* in-place-update policy */ 1000 unsigned int min_ipu_util; /* in-place-update threshold */ 1001 unsigned int min_fsync_blocks; /* threshold for fsync */ 1002 unsigned int min_seq_blocks; /* threshold for sequential blocks */ 1003 unsigned int min_hot_blocks; /* threshold for hot block allocation */ 1004 unsigned int min_ssr_sections; /* threshold to trigger SSR allocation */ 1005 1006 /* for flush command control */ 1007 struct flush_cmd_control *fcc_info; 1008 1009 /* for discard command control */ 1010 struct discard_cmd_control *dcc_info; 1011 }; 1012 1013 /* 1014 * For superblock 1015 */ 1016 /* 1017 * COUNT_TYPE for monitoring 1018 * 1019 * f2fs monitors the number of several block types such as on-writeback, 1020 * dirty dentry blocks, dirty node blocks, and dirty meta blocks. 1021 */ 1022 #define WB_DATA_TYPE(p) (__is_cp_guaranteed(p) ? F2FS_WB_CP_DATA : F2FS_WB_DATA) 1023 enum count_type { 1024 F2FS_DIRTY_DENTS, 1025 F2FS_DIRTY_DATA, 1026 F2FS_DIRTY_QDATA, 1027 F2FS_DIRTY_NODES, 1028 F2FS_DIRTY_META, 1029 F2FS_INMEM_PAGES, 1030 F2FS_DIRTY_IMETA, 1031 F2FS_WB_CP_DATA, 1032 F2FS_WB_DATA, 1033 F2FS_RD_DATA, 1034 F2FS_RD_NODE, 1035 F2FS_RD_META, 1036 F2FS_DIO_WRITE, 1037 F2FS_DIO_READ, 1038 NR_COUNT_TYPE, 1039 }; 1040 1041 /* 1042 * The below are the page types of bios used in submit_bio(). 1043 * The available types are: 1044 * DATA User data pages. It operates as async mode. 1045 * NODE Node pages. It operates as async mode. 1046 * META FS metadata pages such as SIT, NAT, CP. 1047 * NR_PAGE_TYPE The number of page types. 1048 * META_FLUSH Make sure the previous pages are written 1049 * with waiting the bio's completion 1050 * ... Only can be used with META. 1051 */ 1052 #define PAGE_TYPE_OF_BIO(type) ((type) > META ? META : (type)) 1053 enum page_type { 1054 DATA, 1055 NODE, 1056 META, 1057 NR_PAGE_TYPE, 1058 META_FLUSH, 1059 INMEM, /* the below types are used by tracepoints only. */ 1060 INMEM_DROP, 1061 INMEM_INVALIDATE, 1062 INMEM_REVOKE, 1063 IPU, 1064 OPU, 1065 }; 1066 1067 enum temp_type { 1068 HOT = 0, /* must be zero for meta bio */ 1069 WARM, 1070 COLD, 1071 NR_TEMP_TYPE, 1072 }; 1073 1074 enum need_lock_type { 1075 LOCK_REQ = 0, 1076 LOCK_DONE, 1077 LOCK_RETRY, 1078 }; 1079 1080 enum cp_reason_type { 1081 CP_NO_NEEDED, 1082 CP_NON_REGULAR, 1083 CP_COMPRESSED, 1084 CP_HARDLINK, 1085 CP_SB_NEED_CP, 1086 CP_WRONG_PINO, 1087 CP_NO_SPC_ROLL, 1088 CP_NODE_NEED_CP, 1089 CP_FASTBOOT_MODE, 1090 CP_SPEC_LOG_NUM, 1091 CP_RECOVER_DIR, 1092 }; 1093 1094 enum iostat_type { 1095 /* WRITE IO */ 1096 APP_DIRECT_IO, /* app direct write IOs */ 1097 APP_BUFFERED_IO, /* app buffered write IOs */ 1098 APP_WRITE_IO, /* app write IOs */ 1099 APP_MAPPED_IO, /* app mapped IOs */ 1100 FS_DATA_IO, /* data IOs from kworker/fsync/reclaimer */ 1101 FS_NODE_IO, /* node IOs from kworker/fsync/reclaimer */ 1102 FS_META_IO, /* meta IOs from kworker/reclaimer */ 1103 FS_GC_DATA_IO, /* data IOs from forground gc */ 1104 FS_GC_NODE_IO, /* node IOs from forground gc */ 1105 FS_CP_DATA_IO, /* data IOs from checkpoint */ 1106 FS_CP_NODE_IO, /* node IOs from checkpoint */ 1107 FS_CP_META_IO, /* meta IOs from checkpoint */ 1108 1109 /* READ IO */ 1110 APP_DIRECT_READ_IO, /* app direct read IOs */ 1111 APP_BUFFERED_READ_IO, /* app buffered read IOs */ 1112 APP_READ_IO, /* app read IOs */ 1113 APP_MAPPED_READ_IO, /* app mapped read IOs */ 1114 FS_DATA_READ_IO, /* data read IOs */ 1115 FS_GDATA_READ_IO, /* data read IOs from background gc */ 1116 FS_CDATA_READ_IO, /* compressed data read IOs */ 1117 FS_NODE_READ_IO, /* node read IOs */ 1118 FS_META_READ_IO, /* meta read IOs */ 1119 1120 /* other */ 1121 FS_DISCARD, /* discard */ 1122 NR_IO_TYPE, 1123 }; 1124 1125 struct f2fs_io_info { 1126 struct f2fs_sb_info *sbi; /* f2fs_sb_info pointer */ 1127 nid_t ino; /* inode number */ 1128 enum page_type type; /* contains DATA/NODE/META/META_FLUSH */ 1129 enum temp_type temp; /* contains HOT/WARM/COLD */ 1130 int op; /* contains REQ_OP_ */ 1131 int op_flags; /* req_flag_bits */ 1132 block_t new_blkaddr; /* new block address to be written */ 1133 block_t old_blkaddr; /* old block address before Cow */ 1134 struct page *page; /* page to be written */ 1135 struct page *encrypted_page; /* encrypted page */ 1136 struct page *compressed_page; /* compressed page */ 1137 struct list_head list; /* serialize IOs */ 1138 bool submitted; /* indicate IO submission */ 1139 int need_lock; /* indicate we need to lock cp_rwsem */ 1140 bool in_list; /* indicate fio is in io_list */ 1141 bool is_por; /* indicate IO is from recovery or not */ 1142 bool retry; /* need to reallocate block address */ 1143 int compr_blocks; /* # of compressed block addresses */ 1144 bool encrypted; /* indicate file is encrypted */ 1145 enum iostat_type io_type; /* io type */ 1146 struct writeback_control *io_wbc; /* writeback control */ 1147 struct bio **bio; /* bio for ipu */ 1148 sector_t *last_block; /* last block number in bio */ 1149 unsigned char version; /* version of the node */ 1150 }; 1151 1152 struct bio_entry { 1153 struct bio *bio; 1154 struct list_head list; 1155 }; 1156 1157 #define is_read_io(rw) ((rw) == READ) 1158 struct f2fs_bio_info { 1159 struct f2fs_sb_info *sbi; /* f2fs superblock */ 1160 struct bio *bio; /* bios to merge */ 1161 sector_t last_block_in_bio; /* last block number */ 1162 struct f2fs_io_info fio; /* store buffered io info. */ 1163 struct rw_semaphore io_rwsem; /* blocking op for bio */ 1164 spinlock_t io_lock; /* serialize DATA/NODE IOs */ 1165 struct list_head io_list; /* track fios */ 1166 struct list_head bio_list; /* bio entry list head */ 1167 struct rw_semaphore bio_list_lock; /* lock to protect bio entry list */ 1168 }; 1169 1170 #define FDEV(i) (sbi->devs[i]) 1171 #define RDEV(i) (raw_super->devs[i]) 1172 struct f2fs_dev_info { 1173 struct block_device *bdev; 1174 char path[MAX_PATH_LEN]; 1175 unsigned int total_segments; 1176 block_t start_blk; 1177 block_t end_blk; 1178 #ifdef CONFIG_BLK_DEV_ZONED 1179 unsigned int nr_blkz; /* Total number of zones */ 1180 unsigned long *blkz_seq; /* Bitmap indicating sequential zones */ 1181 block_t *zone_capacity_blocks; /* Array of zone capacity in blks */ 1182 #endif 1183 }; 1184 1185 enum inode_type { 1186 DIR_INODE, /* for dirty dir inode */ 1187 FILE_INODE, /* for dirty regular/symlink inode */ 1188 DIRTY_META, /* for all dirtied inode metadata */ 1189 ATOMIC_FILE, /* for all atomic files */ 1190 NR_INODE_TYPE, 1191 }; 1192 1193 /* for inner inode cache management */ 1194 struct inode_management { 1195 struct radix_tree_root ino_root; /* ino entry array */ 1196 spinlock_t ino_lock; /* for ino entry lock */ 1197 struct list_head ino_list; /* inode list head */ 1198 unsigned long ino_num; /* number of entries */ 1199 }; 1200 1201 /* for GC_AT */ 1202 struct atgc_management { 1203 bool atgc_enabled; /* ATGC is enabled or not */ 1204 struct rb_root_cached root; /* root of victim rb-tree */ 1205 struct list_head victim_list; /* linked with all victim entries */ 1206 unsigned int victim_count; /* victim count in rb-tree */ 1207 unsigned int candidate_ratio; /* candidate ratio */ 1208 unsigned int max_candidate_count; /* max candidate count */ 1209 unsigned int age_weight; /* age weight, vblock_weight = 100 - age_weight */ 1210 unsigned long long age_threshold; /* age threshold */ 1211 }; 1212 1213 /* For s_flag in struct f2fs_sb_info */ 1214 enum { 1215 SBI_IS_DIRTY, /* dirty flag for checkpoint */ 1216 SBI_IS_CLOSE, /* specify unmounting */ 1217 SBI_NEED_FSCK, /* need fsck.f2fs to fix */ 1218 SBI_POR_DOING, /* recovery is doing or not */ 1219 SBI_NEED_SB_WRITE, /* need to recover superblock */ 1220 SBI_NEED_CP, /* need to checkpoint */ 1221 SBI_IS_SHUTDOWN, /* shutdown by ioctl */ 1222 SBI_IS_RECOVERED, /* recovered orphan/data */ 1223 SBI_CP_DISABLED, /* CP was disabled last mount */ 1224 SBI_CP_DISABLED_QUICK, /* CP was disabled quickly */ 1225 SBI_QUOTA_NEED_FLUSH, /* need to flush quota info in CP */ 1226 SBI_QUOTA_SKIP_FLUSH, /* skip flushing quota in current CP */ 1227 SBI_QUOTA_NEED_REPAIR, /* quota file may be corrupted */ 1228 SBI_IS_RESIZEFS, /* resizefs is in process */ 1229 }; 1230 1231 enum { 1232 CP_TIME, 1233 REQ_TIME, 1234 DISCARD_TIME, 1235 GC_TIME, 1236 DISABLE_TIME, 1237 UMOUNT_DISCARD_TIMEOUT, 1238 MAX_TIME, 1239 }; 1240 1241 enum { 1242 GC_NORMAL, 1243 GC_IDLE_CB, 1244 GC_IDLE_GREEDY, 1245 GC_IDLE_AT, 1246 GC_URGENT_HIGH, 1247 GC_URGENT_LOW, 1248 }; 1249 1250 enum { 1251 BGGC_MODE_ON, /* background gc is on */ 1252 BGGC_MODE_OFF, /* background gc is off */ 1253 BGGC_MODE_SYNC, /* 1254 * background gc is on, migrating blocks 1255 * like foreground gc 1256 */ 1257 }; 1258 1259 enum { 1260 FS_MODE_ADAPTIVE, /* use both lfs/ssr allocation */ 1261 FS_MODE_LFS, /* use lfs allocation only */ 1262 }; 1263 1264 enum { 1265 WHINT_MODE_OFF, /* not pass down write hints */ 1266 WHINT_MODE_USER, /* try to pass down hints given by users */ 1267 WHINT_MODE_FS, /* pass down hints with F2FS policy */ 1268 }; 1269 1270 enum { 1271 ALLOC_MODE_DEFAULT, /* stay default */ 1272 ALLOC_MODE_REUSE, /* reuse segments as much as possible */ 1273 }; 1274 1275 enum fsync_mode { 1276 FSYNC_MODE_POSIX, /* fsync follows posix semantics */ 1277 FSYNC_MODE_STRICT, /* fsync behaves in line with ext4 */ 1278 FSYNC_MODE_NOBARRIER, /* fsync behaves nobarrier based on posix */ 1279 }; 1280 1281 enum { 1282 COMPR_MODE_FS, /* 1283 * automatically compress compression 1284 * enabled files 1285 */ 1286 COMPR_MODE_USER, /* 1287 * automatical compression is disabled. 1288 * user can control the file compression 1289 * using ioctls 1290 */ 1291 }; 1292 1293 /* 1294 * this value is set in page as a private data which indicate that 1295 * the page is atomically written, and it is in inmem_pages list. 1296 */ 1297 #define ATOMIC_WRITTEN_PAGE ((unsigned long)-1) 1298 #define DUMMY_WRITTEN_PAGE ((unsigned long)-2) 1299 1300 #define IS_ATOMIC_WRITTEN_PAGE(page) \ 1301 (page_private(page) == ATOMIC_WRITTEN_PAGE) 1302 #define IS_DUMMY_WRITTEN_PAGE(page) \ 1303 (page_private(page) == DUMMY_WRITTEN_PAGE) 1304 1305 #ifdef CONFIG_F2FS_IO_TRACE 1306 #define IS_IO_TRACED_PAGE(page) \ 1307 (page_private(page) > 0 && \ 1308 page_private(page) < (unsigned long)PID_MAX_LIMIT) 1309 #else 1310 #define IS_IO_TRACED_PAGE(page) (0) 1311 #endif 1312 1313 /* For compression */ 1314 enum compress_algorithm_type { 1315 COMPRESS_LZO, 1316 COMPRESS_LZ4, 1317 COMPRESS_ZSTD, 1318 COMPRESS_LZORLE, 1319 COMPRESS_MAX, 1320 }; 1321 1322 enum compress_flag { 1323 COMPRESS_CHKSUM, 1324 COMPRESS_MAX_FLAG, 1325 }; 1326 1327 #define COMPRESS_DATA_RESERVED_SIZE 4 1328 struct compress_data { 1329 __le32 clen; /* compressed data size */ 1330 __le32 chksum; /* compressed data chksum */ 1331 __le32 reserved[COMPRESS_DATA_RESERVED_SIZE]; /* reserved */ 1332 u8 cdata[]; /* compressed data */ 1333 }; 1334 1335 #define COMPRESS_HEADER_SIZE (sizeof(struct compress_data)) 1336 1337 #define F2FS_COMPRESSED_PAGE_MAGIC 0xF5F2C000 1338 1339 #define COMPRESS_LEVEL_OFFSET 8 1340 1341 /* compress context */ 1342 struct compress_ctx { 1343 struct inode *inode; /* inode the context belong to */ 1344 pgoff_t cluster_idx; /* cluster index number */ 1345 unsigned int cluster_size; /* page count in cluster */ 1346 unsigned int log_cluster_size; /* log of cluster size */ 1347 struct page **rpages; /* pages store raw data in cluster */ 1348 unsigned int nr_rpages; /* total page number in rpages */ 1349 struct page **cpages; /* pages store compressed data in cluster */ 1350 unsigned int nr_cpages; /* total page number in cpages */ 1351 void *rbuf; /* virtual mapped address on rpages */ 1352 struct compress_data *cbuf; /* virtual mapped address on cpages */ 1353 size_t rlen; /* valid data length in rbuf */ 1354 size_t clen; /* valid data length in cbuf */ 1355 void *private; /* payload buffer for specified compression algorithm */ 1356 void *private2; /* extra payload buffer */ 1357 }; 1358 1359 /* compress context for write IO path */ 1360 struct compress_io_ctx { 1361 u32 magic; /* magic number to indicate page is compressed */ 1362 struct inode *inode; /* inode the context belong to */ 1363 struct page **rpages; /* pages store raw data in cluster */ 1364 unsigned int nr_rpages; /* total page number in rpages */ 1365 atomic_t pending_pages; /* in-flight compressed page count */ 1366 }; 1367 1368 /* Context for decompressing one cluster on the read IO path */ 1369 struct decompress_io_ctx { 1370 u32 magic; /* magic number to indicate page is compressed */ 1371 struct inode *inode; /* inode the context belong to */ 1372 pgoff_t cluster_idx; /* cluster index number */ 1373 unsigned int cluster_size; /* page count in cluster */ 1374 unsigned int log_cluster_size; /* log of cluster size */ 1375 struct page **rpages; /* pages store raw data in cluster */ 1376 unsigned int nr_rpages; /* total page number in rpages */ 1377 struct page **cpages; /* pages store compressed data in cluster */ 1378 unsigned int nr_cpages; /* total page number in cpages */ 1379 struct page **tpages; /* temp pages to pad holes in cluster */ 1380 void *rbuf; /* virtual mapped address on rpages */ 1381 struct compress_data *cbuf; /* virtual mapped address on cpages */ 1382 size_t rlen; /* valid data length in rbuf */ 1383 size_t clen; /* valid data length in cbuf */ 1384 1385 /* 1386 * The number of compressed pages remaining to be read in this cluster. 1387 * This is initially nr_cpages. It is decremented by 1 each time a page 1388 * has been read (or failed to be read). When it reaches 0, the cluster 1389 * is decompressed (or an error is reported). 1390 * 1391 * If an error occurs before all the pages have been submitted for I/O, 1392 * then this will never reach 0. In this case the I/O submitter is 1393 * responsible for calling f2fs_decompress_end_io() instead. 1394 */ 1395 atomic_t remaining_pages; 1396 1397 /* 1398 * Number of references to this decompress_io_ctx. 1399 * 1400 * One reference is held for I/O completion. This reference is dropped 1401 * after the pagecache pages are updated and unlocked -- either after 1402 * decompression (and verity if enabled), or after an error. 1403 * 1404 * In addition, each compressed page holds a reference while it is in a 1405 * bio. These references are necessary prevent compressed pages from 1406 * being freed while they are still in a bio. 1407 */ 1408 refcount_t refcnt; 1409 1410 bool failed; /* IO error occurred before decompression? */ 1411 bool need_verity; /* need fs-verity verification after decompression? */ 1412 void *private; /* payload buffer for specified decompression algorithm */ 1413 void *private2; /* extra payload buffer */ 1414 struct work_struct verity_work; /* work to verify the decompressed pages */ 1415 }; 1416 1417 #define NULL_CLUSTER ((unsigned int)(~0)) 1418 #define MIN_COMPRESS_LOG_SIZE 2 1419 #define MAX_COMPRESS_LOG_SIZE 8 1420 #define MAX_COMPRESS_WINDOW_SIZE(log_size) ((PAGE_SIZE) << (log_size)) 1421 1422 struct f2fs_sb_info { 1423 struct super_block *sb; /* pointer to VFS super block */ 1424 struct proc_dir_entry *s_proc; /* proc entry */ 1425 struct f2fs_super_block *raw_super; /* raw super block pointer */ 1426 struct rw_semaphore sb_lock; /* lock for raw super block */ 1427 int valid_super_block; /* valid super block no */ 1428 unsigned long s_flag; /* flags for sbi */ 1429 struct mutex writepages; /* mutex for writepages() */ 1430 1431 #ifdef CONFIG_BLK_DEV_ZONED 1432 unsigned int blocks_per_blkz; /* F2FS blocks per zone */ 1433 unsigned int log_blocks_per_blkz; /* log2 F2FS blocks per zone */ 1434 #endif 1435 1436 /* for node-related operations */ 1437 struct f2fs_nm_info *nm_info; /* node manager */ 1438 struct inode *node_inode; /* cache node blocks */ 1439 1440 /* for segment-related operations */ 1441 struct f2fs_sm_info *sm_info; /* segment manager */ 1442 1443 /* for bio operations */ 1444 struct f2fs_bio_info *write_io[NR_PAGE_TYPE]; /* for write bios */ 1445 /* keep migration IO order for LFS mode */ 1446 struct rw_semaphore io_order_lock; 1447 mempool_t *write_io_dummy; /* Dummy pages */ 1448 1449 /* for checkpoint */ 1450 struct f2fs_checkpoint *ckpt; /* raw checkpoint pointer */ 1451 int cur_cp_pack; /* remain current cp pack */ 1452 spinlock_t cp_lock; /* for flag in ckpt */ 1453 struct inode *meta_inode; /* cache meta blocks */ 1454 struct rw_semaphore cp_global_sem; /* checkpoint procedure lock */ 1455 struct rw_semaphore cp_rwsem; /* blocking FS operations */ 1456 struct rw_semaphore node_write; /* locking node writes */ 1457 struct rw_semaphore node_change; /* locking node change */ 1458 wait_queue_head_t cp_wait; 1459 unsigned long last_time[MAX_TIME]; /* to store time in jiffies */ 1460 long interval_time[MAX_TIME]; /* to store thresholds */ 1461 struct ckpt_req_control cprc_info; /* for checkpoint request control */ 1462 1463 struct inode_management im[MAX_INO_ENTRY]; /* manage inode cache */ 1464 1465 spinlock_t fsync_node_lock; /* for node entry lock */ 1466 struct list_head fsync_node_list; /* node list head */ 1467 unsigned int fsync_seg_id; /* sequence id */ 1468 unsigned int fsync_node_num; /* number of node entries */ 1469 1470 /* for orphan inode, use 0'th array */ 1471 unsigned int max_orphans; /* max orphan inodes */ 1472 1473 /* for inode management */ 1474 struct list_head inode_list[NR_INODE_TYPE]; /* dirty inode list */ 1475 spinlock_t inode_lock[NR_INODE_TYPE]; /* for dirty inode list lock */ 1476 struct mutex flush_lock; /* for flush exclusion */ 1477 1478 /* for extent tree cache */ 1479 struct radix_tree_root extent_tree_root;/* cache extent cache entries */ 1480 struct mutex extent_tree_lock; /* locking extent radix tree */ 1481 struct list_head extent_list; /* lru list for shrinker */ 1482 spinlock_t extent_lock; /* locking extent lru list */ 1483 atomic_t total_ext_tree; /* extent tree count */ 1484 struct list_head zombie_list; /* extent zombie tree list */ 1485 atomic_t total_zombie_tree; /* extent zombie tree count */ 1486 atomic_t total_ext_node; /* extent info count */ 1487 1488 /* basic filesystem units */ 1489 unsigned int log_sectors_per_block; /* log2 sectors per block */ 1490 unsigned int log_blocksize; /* log2 block size */ 1491 unsigned int blocksize; /* block size */ 1492 unsigned int root_ino_num; /* root inode number*/ 1493 unsigned int node_ino_num; /* node inode number*/ 1494 unsigned int meta_ino_num; /* meta inode number*/ 1495 unsigned int log_blocks_per_seg; /* log2 blocks per segment */ 1496 unsigned int blocks_per_seg; /* blocks per segment */ 1497 unsigned int segs_per_sec; /* segments per section */ 1498 unsigned int secs_per_zone; /* sections per zone */ 1499 unsigned int total_sections; /* total section count */ 1500 unsigned int total_node_count; /* total node block count */ 1501 unsigned int total_valid_node_count; /* valid node block count */ 1502 int dir_level; /* directory level */ 1503 int readdir_ra; /* readahead inode in readdir */ 1504 u64 max_io_bytes; /* max io bytes to merge IOs */ 1505 1506 block_t user_block_count; /* # of user blocks */ 1507 block_t total_valid_block_count; /* # of valid blocks */ 1508 block_t discard_blks; /* discard command candidats */ 1509 block_t last_valid_block_count; /* for recovery */ 1510 block_t reserved_blocks; /* configurable reserved blocks */ 1511 block_t current_reserved_blocks; /* current reserved blocks */ 1512 1513 /* Additional tracking for no checkpoint mode */ 1514 block_t unusable_block_count; /* # of blocks saved by last cp */ 1515 1516 unsigned int nquota_files; /* # of quota sysfile */ 1517 struct rw_semaphore quota_sem; /* blocking cp for flags */ 1518 1519 /* # of pages, see count_type */ 1520 atomic_t nr_pages[NR_COUNT_TYPE]; 1521 /* # of allocated blocks */ 1522 struct percpu_counter alloc_valid_block_count; 1523 1524 /* writeback control */ 1525 atomic_t wb_sync_req[META]; /* count # of WB_SYNC threads */ 1526 1527 /* valid inode count */ 1528 struct percpu_counter total_valid_inode_count; 1529 1530 struct f2fs_mount_info mount_opt; /* mount options */ 1531 1532 /* for cleaning operations */ 1533 struct rw_semaphore gc_lock; /* 1534 * semaphore for GC, avoid 1535 * race between GC and GC or CP 1536 */ 1537 struct f2fs_gc_kthread *gc_thread; /* GC thread */ 1538 struct atgc_management am; /* atgc management */ 1539 unsigned int cur_victim_sec; /* current victim section num */ 1540 unsigned int gc_mode; /* current GC state */ 1541 unsigned int next_victim_seg[2]; /* next segment in victim section */ 1542 1543 /* for skip statistic */ 1544 unsigned int atomic_files; /* # of opened atomic file */ 1545 unsigned long long skipped_atomic_files[2]; /* FG_GC and BG_GC */ 1546 unsigned long long skipped_gc_rwsem; /* FG_GC only */ 1547 1548 /* threshold for gc trials on pinned files */ 1549 u64 gc_pin_file_threshold; 1550 struct rw_semaphore pin_sem; 1551 1552 /* maximum # of trials to find a victim segment for SSR and GC */ 1553 unsigned int max_victim_search; 1554 /* migration granularity of garbage collection, unit: segment */ 1555 unsigned int migration_granularity; 1556 1557 /* 1558 * for stat information. 1559 * one is for the LFS mode, and the other is for the SSR mode. 1560 */ 1561 #ifdef CONFIG_F2FS_STAT_FS 1562 struct f2fs_stat_info *stat_info; /* FS status information */ 1563 atomic_t meta_count[META_MAX]; /* # of meta blocks */ 1564 unsigned int segment_count[2]; /* # of allocated segments */ 1565 unsigned int block_count[2]; /* # of allocated blocks */ 1566 atomic_t inplace_count; /* # of inplace update */ 1567 atomic64_t total_hit_ext; /* # of lookup extent cache */ 1568 atomic64_t read_hit_rbtree; /* # of hit rbtree extent node */ 1569 atomic64_t read_hit_largest; /* # of hit largest extent node */ 1570 atomic64_t read_hit_cached; /* # of hit cached extent node */ 1571 atomic_t inline_xattr; /* # of inline_xattr inodes */ 1572 atomic_t inline_inode; /* # of inline_data inodes */ 1573 atomic_t inline_dir; /* # of inline_dentry inodes */ 1574 atomic_t compr_inode; /* # of compressed inodes */ 1575 atomic64_t compr_blocks; /* # of compressed blocks */ 1576 atomic_t vw_cnt; /* # of volatile writes */ 1577 atomic_t max_aw_cnt; /* max # of atomic writes */ 1578 atomic_t max_vw_cnt; /* max # of volatile writes */ 1579 unsigned int io_skip_bggc; /* skip background gc for in-flight IO */ 1580 unsigned int other_skip_bggc; /* skip background gc for other reasons */ 1581 unsigned int ndirty_inode[NR_INODE_TYPE]; /* # of dirty inodes */ 1582 #endif 1583 spinlock_t stat_lock; /* lock for stat operations */ 1584 1585 /* For app/fs IO statistics */ 1586 spinlock_t iostat_lock; 1587 unsigned long long rw_iostat[NR_IO_TYPE]; 1588 unsigned long long prev_rw_iostat[NR_IO_TYPE]; 1589 bool iostat_enable; 1590 unsigned long iostat_next_period; 1591 unsigned int iostat_period_ms; 1592 1593 /* to attach REQ_META|REQ_FUA flags */ 1594 unsigned int data_io_flag; 1595 unsigned int node_io_flag; 1596 1597 /* For sysfs suppport */ 1598 struct kobject s_kobj; /* /sys/fs/f2fs/<devname> */ 1599 struct completion s_kobj_unregister; 1600 1601 struct kobject s_stat_kobj; /* /sys/fs/f2fs/<devname>/stat */ 1602 struct completion s_stat_kobj_unregister; 1603 1604 /* For shrinker support */ 1605 struct list_head s_list; 1606 int s_ndevs; /* number of devices */ 1607 struct f2fs_dev_info *devs; /* for device list */ 1608 unsigned int dirty_device; /* for checkpoint data flush */ 1609 spinlock_t dev_lock; /* protect dirty_device */ 1610 struct mutex umount_mutex; 1611 unsigned int shrinker_run_no; 1612 1613 /* For write statistics */ 1614 u64 sectors_written_start; 1615 u64 kbytes_written; 1616 1617 /* Reference to checksum algorithm driver via cryptoapi */ 1618 struct crypto_shash *s_chksum_driver; 1619 1620 /* Precomputed FS UUID checksum for seeding other checksums */ 1621 __u32 s_chksum_seed; 1622 1623 struct workqueue_struct *post_read_wq; /* post read workqueue */ 1624 1625 struct kmem_cache *inline_xattr_slab; /* inline xattr entry */ 1626 unsigned int inline_xattr_slab_size; /* default inline xattr slab size */ 1627 1628 #ifdef CONFIG_F2FS_FS_COMPRESSION 1629 struct kmem_cache *page_array_slab; /* page array entry */ 1630 unsigned int page_array_slab_size; /* default page array slab size */ 1631 #endif 1632 }; 1633 1634 struct f2fs_private_dio { 1635 struct inode *inode; 1636 void *orig_private; 1637 bio_end_io_t *orig_end_io; 1638 bool write; 1639 }; 1640 1641 #ifdef CONFIG_F2FS_FAULT_INJECTION 1642 #define f2fs_show_injection_info(sbi, type) \ 1643 printk_ratelimited("%sF2FS-fs (%s) : inject %s in %s of %pS\n", \ 1644 KERN_INFO, sbi->sb->s_id, \ 1645 f2fs_fault_name[type], \ 1646 __func__, __builtin_return_address(0)) 1647 static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type) 1648 { 1649 struct f2fs_fault_info *ffi = &F2FS_OPTION(sbi).fault_info; 1650 1651 if (!ffi->inject_rate) 1652 return false; 1653 1654 if (!IS_FAULT_SET(ffi, type)) 1655 return false; 1656 1657 atomic_inc(&ffi->inject_ops); 1658 if (atomic_read(&ffi->inject_ops) >= ffi->inject_rate) { 1659 atomic_set(&ffi->inject_ops, 0); 1660 return true; 1661 } 1662 return false; 1663 } 1664 #else 1665 #define f2fs_show_injection_info(sbi, type) do { } while (0) 1666 static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type) 1667 { 1668 return false; 1669 } 1670 #endif 1671 1672 /* 1673 * Test if the mounted volume is a multi-device volume. 1674 * - For a single regular disk volume, sbi->s_ndevs is 0. 1675 * - For a single zoned disk volume, sbi->s_ndevs is 1. 1676 * - For a multi-device volume, sbi->s_ndevs is always 2 or more. 1677 */ 1678 static inline bool f2fs_is_multi_device(struct f2fs_sb_info *sbi) 1679 { 1680 return sbi->s_ndevs > 1; 1681 } 1682 1683 static inline void f2fs_update_time(struct f2fs_sb_info *sbi, int type) 1684 { 1685 unsigned long now = jiffies; 1686 1687 sbi->last_time[type] = now; 1688 1689 /* DISCARD_TIME and GC_TIME are based on REQ_TIME */ 1690 if (type == REQ_TIME) { 1691 sbi->last_time[DISCARD_TIME] = now; 1692 sbi->last_time[GC_TIME] = now; 1693 } 1694 } 1695 1696 static inline bool f2fs_time_over(struct f2fs_sb_info *sbi, int type) 1697 { 1698 unsigned long interval = sbi->interval_time[type] * HZ; 1699 1700 return time_after(jiffies, sbi->last_time[type] + interval); 1701 } 1702 1703 static inline unsigned int f2fs_time_to_wait(struct f2fs_sb_info *sbi, 1704 int type) 1705 { 1706 unsigned long interval = sbi->interval_time[type] * HZ; 1707 unsigned int wait_ms = 0; 1708 long delta; 1709 1710 delta = (sbi->last_time[type] + interval) - jiffies; 1711 if (delta > 0) 1712 wait_ms = jiffies_to_msecs(delta); 1713 1714 return wait_ms; 1715 } 1716 1717 /* 1718 * Inline functions 1719 */ 1720 static inline u32 __f2fs_crc32(struct f2fs_sb_info *sbi, u32 crc, 1721 const void *address, unsigned int length) 1722 { 1723 struct { 1724 struct shash_desc shash; 1725 char ctx[4]; 1726 } desc; 1727 int err; 1728 1729 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver) != sizeof(desc.ctx)); 1730 1731 desc.shash.tfm = sbi->s_chksum_driver; 1732 *(u32 *)desc.ctx = crc; 1733 1734 err = crypto_shash_update(&desc.shash, address, length); 1735 BUG_ON(err); 1736 1737 return *(u32 *)desc.ctx; 1738 } 1739 1740 static inline u32 f2fs_crc32(struct f2fs_sb_info *sbi, const void *address, 1741 unsigned int length) 1742 { 1743 return __f2fs_crc32(sbi, F2FS_SUPER_MAGIC, address, length); 1744 } 1745 1746 static inline bool f2fs_crc_valid(struct f2fs_sb_info *sbi, __u32 blk_crc, 1747 void *buf, size_t buf_size) 1748 { 1749 return f2fs_crc32(sbi, buf, buf_size) == blk_crc; 1750 } 1751 1752 static inline u32 f2fs_chksum(struct f2fs_sb_info *sbi, u32 crc, 1753 const void *address, unsigned int length) 1754 { 1755 return __f2fs_crc32(sbi, crc, address, length); 1756 } 1757 1758 static inline struct f2fs_inode_info *F2FS_I(struct inode *inode) 1759 { 1760 return container_of(inode, struct f2fs_inode_info, vfs_inode); 1761 } 1762 1763 static inline struct f2fs_sb_info *F2FS_SB(struct super_block *sb) 1764 { 1765 return sb->s_fs_info; 1766 } 1767 1768 static inline struct f2fs_sb_info *F2FS_I_SB(struct inode *inode) 1769 { 1770 return F2FS_SB(inode->i_sb); 1771 } 1772 1773 static inline struct f2fs_sb_info *F2FS_M_SB(struct address_space *mapping) 1774 { 1775 return F2FS_I_SB(mapping->host); 1776 } 1777 1778 static inline struct f2fs_sb_info *F2FS_P_SB(struct page *page) 1779 { 1780 return F2FS_M_SB(page_file_mapping(page)); 1781 } 1782 1783 static inline struct f2fs_super_block *F2FS_RAW_SUPER(struct f2fs_sb_info *sbi) 1784 { 1785 return (struct f2fs_super_block *)(sbi->raw_super); 1786 } 1787 1788 static inline struct f2fs_checkpoint *F2FS_CKPT(struct f2fs_sb_info *sbi) 1789 { 1790 return (struct f2fs_checkpoint *)(sbi->ckpt); 1791 } 1792 1793 static inline struct f2fs_node *F2FS_NODE(struct page *page) 1794 { 1795 return (struct f2fs_node *)page_address(page); 1796 } 1797 1798 static inline struct f2fs_inode *F2FS_INODE(struct page *page) 1799 { 1800 return &((struct f2fs_node *)page_address(page))->i; 1801 } 1802 1803 static inline struct f2fs_nm_info *NM_I(struct f2fs_sb_info *sbi) 1804 { 1805 return (struct f2fs_nm_info *)(sbi->nm_info); 1806 } 1807 1808 static inline struct f2fs_sm_info *SM_I(struct f2fs_sb_info *sbi) 1809 { 1810 return (struct f2fs_sm_info *)(sbi->sm_info); 1811 } 1812 1813 static inline struct sit_info *SIT_I(struct f2fs_sb_info *sbi) 1814 { 1815 return (struct sit_info *)(SM_I(sbi)->sit_info); 1816 } 1817 1818 static inline struct free_segmap_info *FREE_I(struct f2fs_sb_info *sbi) 1819 { 1820 return (struct free_segmap_info *)(SM_I(sbi)->free_info); 1821 } 1822 1823 static inline struct dirty_seglist_info *DIRTY_I(struct f2fs_sb_info *sbi) 1824 { 1825 return (struct dirty_seglist_info *)(SM_I(sbi)->dirty_info); 1826 } 1827 1828 static inline struct address_space *META_MAPPING(struct f2fs_sb_info *sbi) 1829 { 1830 return sbi->meta_inode->i_mapping; 1831 } 1832 1833 static inline struct address_space *NODE_MAPPING(struct f2fs_sb_info *sbi) 1834 { 1835 return sbi->node_inode->i_mapping; 1836 } 1837 1838 static inline bool is_sbi_flag_set(struct f2fs_sb_info *sbi, unsigned int type) 1839 { 1840 return test_bit(type, &sbi->s_flag); 1841 } 1842 1843 static inline void set_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type) 1844 { 1845 set_bit(type, &sbi->s_flag); 1846 } 1847 1848 static inline void clear_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type) 1849 { 1850 clear_bit(type, &sbi->s_flag); 1851 } 1852 1853 static inline unsigned long long cur_cp_version(struct f2fs_checkpoint *cp) 1854 { 1855 return le64_to_cpu(cp->checkpoint_ver); 1856 } 1857 1858 static inline unsigned long f2fs_qf_ino(struct super_block *sb, int type) 1859 { 1860 if (type < F2FS_MAX_QUOTAS) 1861 return le32_to_cpu(F2FS_SB(sb)->raw_super->qf_ino[type]); 1862 return 0; 1863 } 1864 1865 static inline __u64 cur_cp_crc(struct f2fs_checkpoint *cp) 1866 { 1867 size_t crc_offset = le32_to_cpu(cp->checksum_offset); 1868 return le32_to_cpu(*((__le32 *)((unsigned char *)cp + crc_offset))); 1869 } 1870 1871 static inline bool __is_set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 1872 { 1873 unsigned int ckpt_flags = le32_to_cpu(cp->ckpt_flags); 1874 1875 return ckpt_flags & f; 1876 } 1877 1878 static inline bool is_set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 1879 { 1880 return __is_set_ckpt_flags(F2FS_CKPT(sbi), f); 1881 } 1882 1883 static inline void __set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 1884 { 1885 unsigned int ckpt_flags; 1886 1887 ckpt_flags = le32_to_cpu(cp->ckpt_flags); 1888 ckpt_flags |= f; 1889 cp->ckpt_flags = cpu_to_le32(ckpt_flags); 1890 } 1891 1892 static inline void set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 1893 { 1894 unsigned long flags; 1895 1896 spin_lock_irqsave(&sbi->cp_lock, flags); 1897 __set_ckpt_flags(F2FS_CKPT(sbi), f); 1898 spin_unlock_irqrestore(&sbi->cp_lock, flags); 1899 } 1900 1901 static inline void __clear_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 1902 { 1903 unsigned int ckpt_flags; 1904 1905 ckpt_flags = le32_to_cpu(cp->ckpt_flags); 1906 ckpt_flags &= (~f); 1907 cp->ckpt_flags = cpu_to_le32(ckpt_flags); 1908 } 1909 1910 static inline void clear_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 1911 { 1912 unsigned long flags; 1913 1914 spin_lock_irqsave(&sbi->cp_lock, flags); 1915 __clear_ckpt_flags(F2FS_CKPT(sbi), f); 1916 spin_unlock_irqrestore(&sbi->cp_lock, flags); 1917 } 1918 1919 static inline void disable_nat_bits(struct f2fs_sb_info *sbi, bool lock) 1920 { 1921 unsigned long flags; 1922 unsigned char *nat_bits; 1923 1924 /* 1925 * In order to re-enable nat_bits we need to call fsck.f2fs by 1926 * set_sbi_flag(sbi, SBI_NEED_FSCK). But it may give huge cost, 1927 * so let's rely on regular fsck or unclean shutdown. 1928 */ 1929 1930 if (lock) 1931 spin_lock_irqsave(&sbi->cp_lock, flags); 1932 __clear_ckpt_flags(F2FS_CKPT(sbi), CP_NAT_BITS_FLAG); 1933 nat_bits = NM_I(sbi)->nat_bits; 1934 NM_I(sbi)->nat_bits = NULL; 1935 if (lock) 1936 spin_unlock_irqrestore(&sbi->cp_lock, flags); 1937 1938 kvfree(nat_bits); 1939 } 1940 1941 static inline bool enabled_nat_bits(struct f2fs_sb_info *sbi, 1942 struct cp_control *cpc) 1943 { 1944 bool set = is_set_ckpt_flags(sbi, CP_NAT_BITS_FLAG); 1945 1946 return (cpc) ? (cpc->reason & CP_UMOUNT) && set : set; 1947 } 1948 1949 static inline void f2fs_lock_op(struct f2fs_sb_info *sbi) 1950 { 1951 down_read(&sbi->cp_rwsem); 1952 } 1953 1954 static inline int f2fs_trylock_op(struct f2fs_sb_info *sbi) 1955 { 1956 return down_read_trylock(&sbi->cp_rwsem); 1957 } 1958 1959 static inline void f2fs_unlock_op(struct f2fs_sb_info *sbi) 1960 { 1961 up_read(&sbi->cp_rwsem); 1962 } 1963 1964 static inline void f2fs_lock_all(struct f2fs_sb_info *sbi) 1965 { 1966 down_write(&sbi->cp_rwsem); 1967 } 1968 1969 static inline void f2fs_unlock_all(struct f2fs_sb_info *sbi) 1970 { 1971 up_write(&sbi->cp_rwsem); 1972 } 1973 1974 static inline int __get_cp_reason(struct f2fs_sb_info *sbi) 1975 { 1976 int reason = CP_SYNC; 1977 1978 if (test_opt(sbi, FASTBOOT)) 1979 reason = CP_FASTBOOT; 1980 if (is_sbi_flag_set(sbi, SBI_IS_CLOSE)) 1981 reason = CP_UMOUNT; 1982 return reason; 1983 } 1984 1985 static inline bool __remain_node_summaries(int reason) 1986 { 1987 return (reason & (CP_UMOUNT | CP_FASTBOOT)); 1988 } 1989 1990 static inline bool __exist_node_summaries(struct f2fs_sb_info *sbi) 1991 { 1992 return (is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG) || 1993 is_set_ckpt_flags(sbi, CP_FASTBOOT_FLAG)); 1994 } 1995 1996 /* 1997 * Check whether the inode has blocks or not 1998 */ 1999 static inline int F2FS_HAS_BLOCKS(struct inode *inode) 2000 { 2001 block_t xattr_block = F2FS_I(inode)->i_xattr_nid ? 1 : 0; 2002 2003 return (inode->i_blocks >> F2FS_LOG_SECTORS_PER_BLOCK) > xattr_block; 2004 } 2005 2006 static inline bool f2fs_has_xattr_block(unsigned int ofs) 2007 { 2008 return ofs == XATTR_NODE_OFFSET; 2009 } 2010 2011 static inline bool __allow_reserved_blocks(struct f2fs_sb_info *sbi, 2012 struct inode *inode, bool cap) 2013 { 2014 if (!inode) 2015 return true; 2016 if (!test_opt(sbi, RESERVE_ROOT)) 2017 return false; 2018 if (IS_NOQUOTA(inode)) 2019 return true; 2020 if (uid_eq(F2FS_OPTION(sbi).s_resuid, current_fsuid())) 2021 return true; 2022 if (!gid_eq(F2FS_OPTION(sbi).s_resgid, GLOBAL_ROOT_GID) && 2023 in_group_p(F2FS_OPTION(sbi).s_resgid)) 2024 return true; 2025 if (cap && capable(CAP_SYS_RESOURCE)) 2026 return true; 2027 return false; 2028 } 2029 2030 static inline void f2fs_i_blocks_write(struct inode *, block_t, bool, bool); 2031 static inline int inc_valid_block_count(struct f2fs_sb_info *sbi, 2032 struct inode *inode, blkcnt_t *count) 2033 { 2034 blkcnt_t diff = 0, release = 0; 2035 block_t avail_user_block_count; 2036 int ret; 2037 2038 ret = dquot_reserve_block(inode, *count); 2039 if (ret) 2040 return ret; 2041 2042 if (time_to_inject(sbi, FAULT_BLOCK)) { 2043 f2fs_show_injection_info(sbi, FAULT_BLOCK); 2044 release = *count; 2045 goto release_quota; 2046 } 2047 2048 /* 2049 * let's increase this in prior to actual block count change in order 2050 * for f2fs_sync_file to avoid data races when deciding checkpoint. 2051 */ 2052 percpu_counter_add(&sbi->alloc_valid_block_count, (*count)); 2053 2054 spin_lock(&sbi->stat_lock); 2055 sbi->total_valid_block_count += (block_t)(*count); 2056 avail_user_block_count = sbi->user_block_count - 2057 sbi->current_reserved_blocks; 2058 2059 if (!__allow_reserved_blocks(sbi, inode, true)) 2060 avail_user_block_count -= F2FS_OPTION(sbi).root_reserved_blocks; 2061 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) { 2062 if (avail_user_block_count > sbi->unusable_block_count) 2063 avail_user_block_count -= sbi->unusable_block_count; 2064 else 2065 avail_user_block_count = 0; 2066 } 2067 if (unlikely(sbi->total_valid_block_count > avail_user_block_count)) { 2068 diff = sbi->total_valid_block_count - avail_user_block_count; 2069 if (diff > *count) 2070 diff = *count; 2071 *count -= diff; 2072 release = diff; 2073 sbi->total_valid_block_count -= diff; 2074 if (!*count) { 2075 spin_unlock(&sbi->stat_lock); 2076 goto enospc; 2077 } 2078 } 2079 spin_unlock(&sbi->stat_lock); 2080 2081 if (unlikely(release)) { 2082 percpu_counter_sub(&sbi->alloc_valid_block_count, release); 2083 dquot_release_reservation_block(inode, release); 2084 } 2085 f2fs_i_blocks_write(inode, *count, true, true); 2086 return 0; 2087 2088 enospc: 2089 percpu_counter_sub(&sbi->alloc_valid_block_count, release); 2090 release_quota: 2091 dquot_release_reservation_block(inode, release); 2092 return -ENOSPC; 2093 } 2094 2095 __printf(2, 3) 2096 void f2fs_printk(struct f2fs_sb_info *sbi, const char *fmt, ...); 2097 2098 #define f2fs_err(sbi, fmt, ...) \ 2099 f2fs_printk(sbi, KERN_ERR fmt, ##__VA_ARGS__) 2100 #define f2fs_warn(sbi, fmt, ...) \ 2101 f2fs_printk(sbi, KERN_WARNING fmt, ##__VA_ARGS__) 2102 #define f2fs_notice(sbi, fmt, ...) \ 2103 f2fs_printk(sbi, KERN_NOTICE fmt, ##__VA_ARGS__) 2104 #define f2fs_info(sbi, fmt, ...) \ 2105 f2fs_printk(sbi, KERN_INFO fmt, ##__VA_ARGS__) 2106 #define f2fs_debug(sbi, fmt, ...) \ 2107 f2fs_printk(sbi, KERN_DEBUG fmt, ##__VA_ARGS__) 2108 2109 static inline void dec_valid_block_count(struct f2fs_sb_info *sbi, 2110 struct inode *inode, 2111 block_t count) 2112 { 2113 blkcnt_t sectors = count << F2FS_LOG_SECTORS_PER_BLOCK; 2114 2115 spin_lock(&sbi->stat_lock); 2116 f2fs_bug_on(sbi, sbi->total_valid_block_count < (block_t) count); 2117 sbi->total_valid_block_count -= (block_t)count; 2118 if (sbi->reserved_blocks && 2119 sbi->current_reserved_blocks < sbi->reserved_blocks) 2120 sbi->current_reserved_blocks = min(sbi->reserved_blocks, 2121 sbi->current_reserved_blocks + count); 2122 spin_unlock(&sbi->stat_lock); 2123 if (unlikely(inode->i_blocks < sectors)) { 2124 f2fs_warn(sbi, "Inconsistent i_blocks, ino:%lu, iblocks:%llu, sectors:%llu", 2125 inode->i_ino, 2126 (unsigned long long)inode->i_blocks, 2127 (unsigned long long)sectors); 2128 set_sbi_flag(sbi, SBI_NEED_FSCK); 2129 return; 2130 } 2131 f2fs_i_blocks_write(inode, count, false, true); 2132 } 2133 2134 static inline void inc_page_count(struct f2fs_sb_info *sbi, int count_type) 2135 { 2136 atomic_inc(&sbi->nr_pages[count_type]); 2137 2138 if (count_type == F2FS_DIRTY_DENTS || 2139 count_type == F2FS_DIRTY_NODES || 2140 count_type == F2FS_DIRTY_META || 2141 count_type == F2FS_DIRTY_QDATA || 2142 count_type == F2FS_DIRTY_IMETA) 2143 set_sbi_flag(sbi, SBI_IS_DIRTY); 2144 } 2145 2146 static inline void inode_inc_dirty_pages(struct inode *inode) 2147 { 2148 atomic_inc(&F2FS_I(inode)->dirty_pages); 2149 inc_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ? 2150 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA); 2151 if (IS_NOQUOTA(inode)) 2152 inc_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA); 2153 } 2154 2155 static inline void dec_page_count(struct f2fs_sb_info *sbi, int count_type) 2156 { 2157 atomic_dec(&sbi->nr_pages[count_type]); 2158 } 2159 2160 static inline void inode_dec_dirty_pages(struct inode *inode) 2161 { 2162 if (!S_ISDIR(inode->i_mode) && !S_ISREG(inode->i_mode) && 2163 !S_ISLNK(inode->i_mode)) 2164 return; 2165 2166 atomic_dec(&F2FS_I(inode)->dirty_pages); 2167 dec_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ? 2168 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA); 2169 if (IS_NOQUOTA(inode)) 2170 dec_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA); 2171 } 2172 2173 static inline s64 get_pages(struct f2fs_sb_info *sbi, int count_type) 2174 { 2175 return atomic_read(&sbi->nr_pages[count_type]); 2176 } 2177 2178 static inline int get_dirty_pages(struct inode *inode) 2179 { 2180 return atomic_read(&F2FS_I(inode)->dirty_pages); 2181 } 2182 2183 static inline int get_blocktype_secs(struct f2fs_sb_info *sbi, int block_type) 2184 { 2185 unsigned int pages_per_sec = sbi->segs_per_sec * sbi->blocks_per_seg; 2186 unsigned int segs = (get_pages(sbi, block_type) + pages_per_sec - 1) >> 2187 sbi->log_blocks_per_seg; 2188 2189 return segs / sbi->segs_per_sec; 2190 } 2191 2192 static inline block_t valid_user_blocks(struct f2fs_sb_info *sbi) 2193 { 2194 return sbi->total_valid_block_count; 2195 } 2196 2197 static inline block_t discard_blocks(struct f2fs_sb_info *sbi) 2198 { 2199 return sbi->discard_blks; 2200 } 2201 2202 static inline unsigned long __bitmap_size(struct f2fs_sb_info *sbi, int flag) 2203 { 2204 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi); 2205 2206 /* return NAT or SIT bitmap */ 2207 if (flag == NAT_BITMAP) 2208 return le32_to_cpu(ckpt->nat_ver_bitmap_bytesize); 2209 else if (flag == SIT_BITMAP) 2210 return le32_to_cpu(ckpt->sit_ver_bitmap_bytesize); 2211 2212 return 0; 2213 } 2214 2215 static inline block_t __cp_payload(struct f2fs_sb_info *sbi) 2216 { 2217 return le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_payload); 2218 } 2219 2220 static inline void *__bitmap_ptr(struct f2fs_sb_info *sbi, int flag) 2221 { 2222 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi); 2223 void *tmp_ptr = &ckpt->sit_nat_version_bitmap; 2224 int offset; 2225 2226 if (is_set_ckpt_flags(sbi, CP_LARGE_NAT_BITMAP_FLAG)) { 2227 offset = (flag == SIT_BITMAP) ? 2228 le32_to_cpu(ckpt->nat_ver_bitmap_bytesize) : 0; 2229 /* 2230 * if large_nat_bitmap feature is enabled, leave checksum 2231 * protection for all nat/sit bitmaps. 2232 */ 2233 return tmp_ptr + offset + sizeof(__le32); 2234 } 2235 2236 if (__cp_payload(sbi) > 0) { 2237 if (flag == NAT_BITMAP) 2238 return &ckpt->sit_nat_version_bitmap; 2239 else 2240 return (unsigned char *)ckpt + F2FS_BLKSIZE; 2241 } else { 2242 offset = (flag == NAT_BITMAP) ? 2243 le32_to_cpu(ckpt->sit_ver_bitmap_bytesize) : 0; 2244 return tmp_ptr + offset; 2245 } 2246 } 2247 2248 static inline block_t __start_cp_addr(struct f2fs_sb_info *sbi) 2249 { 2250 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr); 2251 2252 if (sbi->cur_cp_pack == 2) 2253 start_addr += sbi->blocks_per_seg; 2254 return start_addr; 2255 } 2256 2257 static inline block_t __start_cp_next_addr(struct f2fs_sb_info *sbi) 2258 { 2259 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr); 2260 2261 if (sbi->cur_cp_pack == 1) 2262 start_addr += sbi->blocks_per_seg; 2263 return start_addr; 2264 } 2265 2266 static inline void __set_cp_next_pack(struct f2fs_sb_info *sbi) 2267 { 2268 sbi->cur_cp_pack = (sbi->cur_cp_pack == 1) ? 2 : 1; 2269 } 2270 2271 static inline block_t __start_sum_addr(struct f2fs_sb_info *sbi) 2272 { 2273 return le32_to_cpu(F2FS_CKPT(sbi)->cp_pack_start_sum); 2274 } 2275 2276 static inline int inc_valid_node_count(struct f2fs_sb_info *sbi, 2277 struct inode *inode, bool is_inode) 2278 { 2279 block_t valid_block_count; 2280 unsigned int valid_node_count, user_block_count; 2281 int err; 2282 2283 if (is_inode) { 2284 if (inode) { 2285 err = dquot_alloc_inode(inode); 2286 if (err) 2287 return err; 2288 } 2289 } else { 2290 err = dquot_reserve_block(inode, 1); 2291 if (err) 2292 return err; 2293 } 2294 2295 if (time_to_inject(sbi, FAULT_BLOCK)) { 2296 f2fs_show_injection_info(sbi, FAULT_BLOCK); 2297 goto enospc; 2298 } 2299 2300 spin_lock(&sbi->stat_lock); 2301 2302 valid_block_count = sbi->total_valid_block_count + 2303 sbi->current_reserved_blocks + 1; 2304 2305 if (!__allow_reserved_blocks(sbi, inode, false)) 2306 valid_block_count += F2FS_OPTION(sbi).root_reserved_blocks; 2307 user_block_count = sbi->user_block_count; 2308 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) 2309 user_block_count -= sbi->unusable_block_count; 2310 2311 if (unlikely(valid_block_count > user_block_count)) { 2312 spin_unlock(&sbi->stat_lock); 2313 goto enospc; 2314 } 2315 2316 valid_node_count = sbi->total_valid_node_count + 1; 2317 if (unlikely(valid_node_count > sbi->total_node_count)) { 2318 spin_unlock(&sbi->stat_lock); 2319 goto enospc; 2320 } 2321 2322 sbi->total_valid_node_count++; 2323 sbi->total_valid_block_count++; 2324 spin_unlock(&sbi->stat_lock); 2325 2326 if (inode) { 2327 if (is_inode) 2328 f2fs_mark_inode_dirty_sync(inode, true); 2329 else 2330 f2fs_i_blocks_write(inode, 1, true, true); 2331 } 2332 2333 percpu_counter_inc(&sbi->alloc_valid_block_count); 2334 return 0; 2335 2336 enospc: 2337 if (is_inode) { 2338 if (inode) 2339 dquot_free_inode(inode); 2340 } else { 2341 dquot_release_reservation_block(inode, 1); 2342 } 2343 return -ENOSPC; 2344 } 2345 2346 static inline void dec_valid_node_count(struct f2fs_sb_info *sbi, 2347 struct inode *inode, bool is_inode) 2348 { 2349 spin_lock(&sbi->stat_lock); 2350 2351 f2fs_bug_on(sbi, !sbi->total_valid_block_count); 2352 f2fs_bug_on(sbi, !sbi->total_valid_node_count); 2353 2354 sbi->total_valid_node_count--; 2355 sbi->total_valid_block_count--; 2356 if (sbi->reserved_blocks && 2357 sbi->current_reserved_blocks < sbi->reserved_blocks) 2358 sbi->current_reserved_blocks++; 2359 2360 spin_unlock(&sbi->stat_lock); 2361 2362 if (is_inode) { 2363 dquot_free_inode(inode); 2364 } else { 2365 if (unlikely(inode->i_blocks == 0)) { 2366 f2fs_warn(sbi, "dec_valid_node_count: inconsistent i_blocks, ino:%lu, iblocks:%llu", 2367 inode->i_ino, 2368 (unsigned long long)inode->i_blocks); 2369 set_sbi_flag(sbi, SBI_NEED_FSCK); 2370 return; 2371 } 2372 f2fs_i_blocks_write(inode, 1, false, true); 2373 } 2374 } 2375 2376 static inline unsigned int valid_node_count(struct f2fs_sb_info *sbi) 2377 { 2378 return sbi->total_valid_node_count; 2379 } 2380 2381 static inline void inc_valid_inode_count(struct f2fs_sb_info *sbi) 2382 { 2383 percpu_counter_inc(&sbi->total_valid_inode_count); 2384 } 2385 2386 static inline void dec_valid_inode_count(struct f2fs_sb_info *sbi) 2387 { 2388 percpu_counter_dec(&sbi->total_valid_inode_count); 2389 } 2390 2391 static inline s64 valid_inode_count(struct f2fs_sb_info *sbi) 2392 { 2393 return percpu_counter_sum_positive(&sbi->total_valid_inode_count); 2394 } 2395 2396 static inline struct page *f2fs_grab_cache_page(struct address_space *mapping, 2397 pgoff_t index, bool for_write) 2398 { 2399 struct page *page; 2400 2401 if (IS_ENABLED(CONFIG_F2FS_FAULT_INJECTION)) { 2402 if (!for_write) 2403 page = find_get_page_flags(mapping, index, 2404 FGP_LOCK | FGP_ACCESSED); 2405 else 2406 page = find_lock_page(mapping, index); 2407 if (page) 2408 return page; 2409 2410 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_ALLOC)) { 2411 f2fs_show_injection_info(F2FS_M_SB(mapping), 2412 FAULT_PAGE_ALLOC); 2413 return NULL; 2414 } 2415 } 2416 2417 if (!for_write) 2418 return grab_cache_page(mapping, index); 2419 return grab_cache_page_write_begin(mapping, index, AOP_FLAG_NOFS); 2420 } 2421 2422 static inline struct page *f2fs_pagecache_get_page( 2423 struct address_space *mapping, pgoff_t index, 2424 int fgp_flags, gfp_t gfp_mask) 2425 { 2426 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_GET)) { 2427 f2fs_show_injection_info(F2FS_M_SB(mapping), FAULT_PAGE_GET); 2428 return NULL; 2429 } 2430 2431 return pagecache_get_page(mapping, index, fgp_flags, gfp_mask); 2432 } 2433 2434 static inline void f2fs_copy_page(struct page *src, struct page *dst) 2435 { 2436 char *src_kaddr = kmap(src); 2437 char *dst_kaddr = kmap(dst); 2438 2439 memcpy(dst_kaddr, src_kaddr, PAGE_SIZE); 2440 kunmap(dst); 2441 kunmap(src); 2442 } 2443 2444 static inline void f2fs_put_page(struct page *page, int unlock) 2445 { 2446 if (!page) 2447 return; 2448 2449 if (unlock) { 2450 f2fs_bug_on(F2FS_P_SB(page), !PageLocked(page)); 2451 unlock_page(page); 2452 } 2453 put_page(page); 2454 } 2455 2456 static inline void f2fs_put_dnode(struct dnode_of_data *dn) 2457 { 2458 if (dn->node_page) 2459 f2fs_put_page(dn->node_page, 1); 2460 if (dn->inode_page && dn->node_page != dn->inode_page) 2461 f2fs_put_page(dn->inode_page, 0); 2462 dn->node_page = NULL; 2463 dn->inode_page = NULL; 2464 } 2465 2466 static inline struct kmem_cache *f2fs_kmem_cache_create(const char *name, 2467 size_t size) 2468 { 2469 return kmem_cache_create(name, size, 0, SLAB_RECLAIM_ACCOUNT, NULL); 2470 } 2471 2472 static inline void *f2fs_kmem_cache_alloc(struct kmem_cache *cachep, 2473 gfp_t flags) 2474 { 2475 void *entry; 2476 2477 entry = kmem_cache_alloc(cachep, flags); 2478 if (!entry) 2479 entry = kmem_cache_alloc(cachep, flags | __GFP_NOFAIL); 2480 return entry; 2481 } 2482 2483 static inline bool is_inflight_io(struct f2fs_sb_info *sbi, int type) 2484 { 2485 if (get_pages(sbi, F2FS_RD_DATA) || get_pages(sbi, F2FS_RD_NODE) || 2486 get_pages(sbi, F2FS_RD_META) || get_pages(sbi, F2FS_WB_DATA) || 2487 get_pages(sbi, F2FS_WB_CP_DATA) || 2488 get_pages(sbi, F2FS_DIO_READ) || 2489 get_pages(sbi, F2FS_DIO_WRITE)) 2490 return true; 2491 2492 if (type != DISCARD_TIME && SM_I(sbi) && SM_I(sbi)->dcc_info && 2493 atomic_read(&SM_I(sbi)->dcc_info->queued_discard)) 2494 return true; 2495 2496 if (SM_I(sbi) && SM_I(sbi)->fcc_info && 2497 atomic_read(&SM_I(sbi)->fcc_info->queued_flush)) 2498 return true; 2499 return false; 2500 } 2501 2502 static inline bool is_idle(struct f2fs_sb_info *sbi, int type) 2503 { 2504 if (sbi->gc_mode == GC_URGENT_HIGH) 2505 return true; 2506 2507 if (is_inflight_io(sbi, type)) 2508 return false; 2509 2510 if (sbi->gc_mode == GC_URGENT_LOW && 2511 (type == DISCARD_TIME || type == GC_TIME)) 2512 return true; 2513 2514 return f2fs_time_over(sbi, type); 2515 } 2516 2517 static inline void f2fs_radix_tree_insert(struct radix_tree_root *root, 2518 unsigned long index, void *item) 2519 { 2520 while (radix_tree_insert(root, index, item)) 2521 cond_resched(); 2522 } 2523 2524 #define RAW_IS_INODE(p) ((p)->footer.nid == (p)->footer.ino) 2525 2526 static inline bool IS_INODE(struct page *page) 2527 { 2528 struct f2fs_node *p = F2FS_NODE(page); 2529 2530 return RAW_IS_INODE(p); 2531 } 2532 2533 static inline int offset_in_addr(struct f2fs_inode *i) 2534 { 2535 return (i->i_inline & F2FS_EXTRA_ATTR) ? 2536 (le16_to_cpu(i->i_extra_isize) / sizeof(__le32)) : 0; 2537 } 2538 2539 static inline __le32 *blkaddr_in_node(struct f2fs_node *node) 2540 { 2541 return RAW_IS_INODE(node) ? node->i.i_addr : node->dn.addr; 2542 } 2543 2544 static inline int f2fs_has_extra_attr(struct inode *inode); 2545 static inline block_t data_blkaddr(struct inode *inode, 2546 struct page *node_page, unsigned int offset) 2547 { 2548 struct f2fs_node *raw_node; 2549 __le32 *addr_array; 2550 int base = 0; 2551 bool is_inode = IS_INODE(node_page); 2552 2553 raw_node = F2FS_NODE(node_page); 2554 2555 if (is_inode) { 2556 if (!inode) 2557 /* from GC path only */ 2558 base = offset_in_addr(&raw_node->i); 2559 else if (f2fs_has_extra_attr(inode)) 2560 base = get_extra_isize(inode); 2561 } 2562 2563 addr_array = blkaddr_in_node(raw_node); 2564 return le32_to_cpu(addr_array[base + offset]); 2565 } 2566 2567 static inline block_t f2fs_data_blkaddr(struct dnode_of_data *dn) 2568 { 2569 return data_blkaddr(dn->inode, dn->node_page, dn->ofs_in_node); 2570 } 2571 2572 static inline int f2fs_test_bit(unsigned int nr, char *addr) 2573 { 2574 int mask; 2575 2576 addr += (nr >> 3); 2577 mask = 1 << (7 - (nr & 0x07)); 2578 return mask & *addr; 2579 } 2580 2581 static inline void f2fs_set_bit(unsigned int nr, char *addr) 2582 { 2583 int mask; 2584 2585 addr += (nr >> 3); 2586 mask = 1 << (7 - (nr & 0x07)); 2587 *addr |= mask; 2588 } 2589 2590 static inline void f2fs_clear_bit(unsigned int nr, char *addr) 2591 { 2592 int mask; 2593 2594 addr += (nr >> 3); 2595 mask = 1 << (7 - (nr & 0x07)); 2596 *addr &= ~mask; 2597 } 2598 2599 static inline int f2fs_test_and_set_bit(unsigned int nr, char *addr) 2600 { 2601 int mask; 2602 int ret; 2603 2604 addr += (nr >> 3); 2605 mask = 1 << (7 - (nr & 0x07)); 2606 ret = mask & *addr; 2607 *addr |= mask; 2608 return ret; 2609 } 2610 2611 static inline int f2fs_test_and_clear_bit(unsigned int nr, char *addr) 2612 { 2613 int mask; 2614 int ret; 2615 2616 addr += (nr >> 3); 2617 mask = 1 << (7 - (nr & 0x07)); 2618 ret = mask & *addr; 2619 *addr &= ~mask; 2620 return ret; 2621 } 2622 2623 static inline void f2fs_change_bit(unsigned int nr, char *addr) 2624 { 2625 int mask; 2626 2627 addr += (nr >> 3); 2628 mask = 1 << (7 - (nr & 0x07)); 2629 *addr ^= mask; 2630 } 2631 2632 /* 2633 * On-disk inode flags (f2fs_inode::i_flags) 2634 */ 2635 #define F2FS_COMPR_FL 0x00000004 /* Compress file */ 2636 #define F2FS_SYNC_FL 0x00000008 /* Synchronous updates */ 2637 #define F2FS_IMMUTABLE_FL 0x00000010 /* Immutable file */ 2638 #define F2FS_APPEND_FL 0x00000020 /* writes to file may only append */ 2639 #define F2FS_NODUMP_FL 0x00000040 /* do not dump file */ 2640 #define F2FS_NOATIME_FL 0x00000080 /* do not update atime */ 2641 #define F2FS_NOCOMP_FL 0x00000400 /* Don't compress */ 2642 #define F2FS_INDEX_FL 0x00001000 /* hash-indexed directory */ 2643 #define F2FS_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */ 2644 #define F2FS_PROJINHERIT_FL 0x20000000 /* Create with parents projid */ 2645 #define F2FS_CASEFOLD_FL 0x40000000 /* Casefolded file */ 2646 2647 /* Flags that should be inherited by new inodes from their parent. */ 2648 #define F2FS_FL_INHERITED (F2FS_SYNC_FL | F2FS_NODUMP_FL | F2FS_NOATIME_FL | \ 2649 F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \ 2650 F2FS_CASEFOLD_FL | F2FS_COMPR_FL | F2FS_NOCOMP_FL) 2651 2652 /* Flags that are appropriate for regular files (all but dir-specific ones). */ 2653 #define F2FS_REG_FLMASK (~(F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \ 2654 F2FS_CASEFOLD_FL)) 2655 2656 /* Flags that are appropriate for non-directories/regular files. */ 2657 #define F2FS_OTHER_FLMASK (F2FS_NODUMP_FL | F2FS_NOATIME_FL) 2658 2659 static inline __u32 f2fs_mask_flags(umode_t mode, __u32 flags) 2660 { 2661 if (S_ISDIR(mode)) 2662 return flags; 2663 else if (S_ISREG(mode)) 2664 return flags & F2FS_REG_FLMASK; 2665 else 2666 return flags & F2FS_OTHER_FLMASK; 2667 } 2668 2669 static inline void __mark_inode_dirty_flag(struct inode *inode, 2670 int flag, bool set) 2671 { 2672 switch (flag) { 2673 case FI_INLINE_XATTR: 2674 case FI_INLINE_DATA: 2675 case FI_INLINE_DENTRY: 2676 case FI_NEW_INODE: 2677 if (set) 2678 return; 2679 fallthrough; 2680 case FI_DATA_EXIST: 2681 case FI_INLINE_DOTS: 2682 case FI_PIN_FILE: 2683 f2fs_mark_inode_dirty_sync(inode, true); 2684 } 2685 } 2686 2687 static inline void set_inode_flag(struct inode *inode, int flag) 2688 { 2689 set_bit(flag, F2FS_I(inode)->flags); 2690 __mark_inode_dirty_flag(inode, flag, true); 2691 } 2692 2693 static inline int is_inode_flag_set(struct inode *inode, int flag) 2694 { 2695 return test_bit(flag, F2FS_I(inode)->flags); 2696 } 2697 2698 static inline void clear_inode_flag(struct inode *inode, int flag) 2699 { 2700 clear_bit(flag, F2FS_I(inode)->flags); 2701 __mark_inode_dirty_flag(inode, flag, false); 2702 } 2703 2704 static inline bool f2fs_verity_in_progress(struct inode *inode) 2705 { 2706 return IS_ENABLED(CONFIG_FS_VERITY) && 2707 is_inode_flag_set(inode, FI_VERITY_IN_PROGRESS); 2708 } 2709 2710 static inline void set_acl_inode(struct inode *inode, umode_t mode) 2711 { 2712 F2FS_I(inode)->i_acl_mode = mode; 2713 set_inode_flag(inode, FI_ACL_MODE); 2714 f2fs_mark_inode_dirty_sync(inode, false); 2715 } 2716 2717 static inline void f2fs_i_links_write(struct inode *inode, bool inc) 2718 { 2719 if (inc) 2720 inc_nlink(inode); 2721 else 2722 drop_nlink(inode); 2723 f2fs_mark_inode_dirty_sync(inode, true); 2724 } 2725 2726 static inline void f2fs_i_blocks_write(struct inode *inode, 2727 block_t diff, bool add, bool claim) 2728 { 2729 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE); 2730 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER); 2731 2732 /* add = 1, claim = 1 should be dquot_reserve_block in pair */ 2733 if (add) { 2734 if (claim) 2735 dquot_claim_block(inode, diff); 2736 else 2737 dquot_alloc_block_nofail(inode, diff); 2738 } else { 2739 dquot_free_block(inode, diff); 2740 } 2741 2742 f2fs_mark_inode_dirty_sync(inode, true); 2743 if (clean || recover) 2744 set_inode_flag(inode, FI_AUTO_RECOVER); 2745 } 2746 2747 static inline void f2fs_i_size_write(struct inode *inode, loff_t i_size) 2748 { 2749 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE); 2750 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER); 2751 2752 if (i_size_read(inode) == i_size) 2753 return; 2754 2755 i_size_write(inode, i_size); 2756 f2fs_mark_inode_dirty_sync(inode, true); 2757 if (clean || recover) 2758 set_inode_flag(inode, FI_AUTO_RECOVER); 2759 } 2760 2761 static inline void f2fs_i_depth_write(struct inode *inode, unsigned int depth) 2762 { 2763 F2FS_I(inode)->i_current_depth = depth; 2764 f2fs_mark_inode_dirty_sync(inode, true); 2765 } 2766 2767 static inline void f2fs_i_gc_failures_write(struct inode *inode, 2768 unsigned int count) 2769 { 2770 F2FS_I(inode)->i_gc_failures[GC_FAILURE_PIN] = count; 2771 f2fs_mark_inode_dirty_sync(inode, true); 2772 } 2773 2774 static inline void f2fs_i_xnid_write(struct inode *inode, nid_t xnid) 2775 { 2776 F2FS_I(inode)->i_xattr_nid = xnid; 2777 f2fs_mark_inode_dirty_sync(inode, true); 2778 } 2779 2780 static inline void f2fs_i_pino_write(struct inode *inode, nid_t pino) 2781 { 2782 F2FS_I(inode)->i_pino = pino; 2783 f2fs_mark_inode_dirty_sync(inode, true); 2784 } 2785 2786 static inline void get_inline_info(struct inode *inode, struct f2fs_inode *ri) 2787 { 2788 struct f2fs_inode_info *fi = F2FS_I(inode); 2789 2790 if (ri->i_inline & F2FS_INLINE_XATTR) 2791 set_bit(FI_INLINE_XATTR, fi->flags); 2792 if (ri->i_inline & F2FS_INLINE_DATA) 2793 set_bit(FI_INLINE_DATA, fi->flags); 2794 if (ri->i_inline & F2FS_INLINE_DENTRY) 2795 set_bit(FI_INLINE_DENTRY, fi->flags); 2796 if (ri->i_inline & F2FS_DATA_EXIST) 2797 set_bit(FI_DATA_EXIST, fi->flags); 2798 if (ri->i_inline & F2FS_INLINE_DOTS) 2799 set_bit(FI_INLINE_DOTS, fi->flags); 2800 if (ri->i_inline & F2FS_EXTRA_ATTR) 2801 set_bit(FI_EXTRA_ATTR, fi->flags); 2802 if (ri->i_inline & F2FS_PIN_FILE) 2803 set_bit(FI_PIN_FILE, fi->flags); 2804 } 2805 2806 static inline void set_raw_inline(struct inode *inode, struct f2fs_inode *ri) 2807 { 2808 ri->i_inline = 0; 2809 2810 if (is_inode_flag_set(inode, FI_INLINE_XATTR)) 2811 ri->i_inline |= F2FS_INLINE_XATTR; 2812 if (is_inode_flag_set(inode, FI_INLINE_DATA)) 2813 ri->i_inline |= F2FS_INLINE_DATA; 2814 if (is_inode_flag_set(inode, FI_INLINE_DENTRY)) 2815 ri->i_inline |= F2FS_INLINE_DENTRY; 2816 if (is_inode_flag_set(inode, FI_DATA_EXIST)) 2817 ri->i_inline |= F2FS_DATA_EXIST; 2818 if (is_inode_flag_set(inode, FI_INLINE_DOTS)) 2819 ri->i_inline |= F2FS_INLINE_DOTS; 2820 if (is_inode_flag_set(inode, FI_EXTRA_ATTR)) 2821 ri->i_inline |= F2FS_EXTRA_ATTR; 2822 if (is_inode_flag_set(inode, FI_PIN_FILE)) 2823 ri->i_inline |= F2FS_PIN_FILE; 2824 } 2825 2826 static inline int f2fs_has_extra_attr(struct inode *inode) 2827 { 2828 return is_inode_flag_set(inode, FI_EXTRA_ATTR); 2829 } 2830 2831 static inline int f2fs_has_inline_xattr(struct inode *inode) 2832 { 2833 return is_inode_flag_set(inode, FI_INLINE_XATTR); 2834 } 2835 2836 static inline int f2fs_compressed_file(struct inode *inode) 2837 { 2838 return S_ISREG(inode->i_mode) && 2839 is_inode_flag_set(inode, FI_COMPRESSED_FILE); 2840 } 2841 2842 static inline bool f2fs_need_compress_data(struct inode *inode) 2843 { 2844 int compress_mode = F2FS_OPTION(F2FS_I_SB(inode)).compress_mode; 2845 2846 if (!f2fs_compressed_file(inode)) 2847 return false; 2848 2849 if (compress_mode == COMPR_MODE_FS) 2850 return true; 2851 else if (compress_mode == COMPR_MODE_USER && 2852 is_inode_flag_set(inode, FI_ENABLE_COMPRESS)) 2853 return true; 2854 2855 return false; 2856 } 2857 2858 static inline unsigned int addrs_per_inode(struct inode *inode) 2859 { 2860 unsigned int addrs = CUR_ADDRS_PER_INODE(inode) - 2861 get_inline_xattr_addrs(inode); 2862 2863 if (!f2fs_compressed_file(inode)) 2864 return addrs; 2865 return ALIGN_DOWN(addrs, F2FS_I(inode)->i_cluster_size); 2866 } 2867 2868 static inline unsigned int addrs_per_block(struct inode *inode) 2869 { 2870 if (!f2fs_compressed_file(inode)) 2871 return DEF_ADDRS_PER_BLOCK; 2872 return ALIGN_DOWN(DEF_ADDRS_PER_BLOCK, F2FS_I(inode)->i_cluster_size); 2873 } 2874 2875 static inline void *inline_xattr_addr(struct inode *inode, struct page *page) 2876 { 2877 struct f2fs_inode *ri = F2FS_INODE(page); 2878 2879 return (void *)&(ri->i_addr[DEF_ADDRS_PER_INODE - 2880 get_inline_xattr_addrs(inode)]); 2881 } 2882 2883 static inline int inline_xattr_size(struct inode *inode) 2884 { 2885 if (f2fs_has_inline_xattr(inode)) 2886 return get_inline_xattr_addrs(inode) * sizeof(__le32); 2887 return 0; 2888 } 2889 2890 static inline int f2fs_has_inline_data(struct inode *inode) 2891 { 2892 return is_inode_flag_set(inode, FI_INLINE_DATA); 2893 } 2894 2895 static inline int f2fs_exist_data(struct inode *inode) 2896 { 2897 return is_inode_flag_set(inode, FI_DATA_EXIST); 2898 } 2899 2900 static inline int f2fs_has_inline_dots(struct inode *inode) 2901 { 2902 return is_inode_flag_set(inode, FI_INLINE_DOTS); 2903 } 2904 2905 static inline int f2fs_is_mmap_file(struct inode *inode) 2906 { 2907 return is_inode_flag_set(inode, FI_MMAP_FILE); 2908 } 2909 2910 static inline bool f2fs_is_pinned_file(struct inode *inode) 2911 { 2912 return is_inode_flag_set(inode, FI_PIN_FILE); 2913 } 2914 2915 static inline bool f2fs_is_atomic_file(struct inode *inode) 2916 { 2917 return is_inode_flag_set(inode, FI_ATOMIC_FILE); 2918 } 2919 2920 static inline bool f2fs_is_commit_atomic_write(struct inode *inode) 2921 { 2922 return is_inode_flag_set(inode, FI_ATOMIC_COMMIT); 2923 } 2924 2925 static inline bool f2fs_is_volatile_file(struct inode *inode) 2926 { 2927 return is_inode_flag_set(inode, FI_VOLATILE_FILE); 2928 } 2929 2930 static inline bool f2fs_is_first_block_written(struct inode *inode) 2931 { 2932 return is_inode_flag_set(inode, FI_FIRST_BLOCK_WRITTEN); 2933 } 2934 2935 static inline bool f2fs_is_drop_cache(struct inode *inode) 2936 { 2937 return is_inode_flag_set(inode, FI_DROP_CACHE); 2938 } 2939 2940 static inline void *inline_data_addr(struct inode *inode, struct page *page) 2941 { 2942 struct f2fs_inode *ri = F2FS_INODE(page); 2943 int extra_size = get_extra_isize(inode); 2944 2945 return (void *)&(ri->i_addr[extra_size + DEF_INLINE_RESERVED_SIZE]); 2946 } 2947 2948 static inline int f2fs_has_inline_dentry(struct inode *inode) 2949 { 2950 return is_inode_flag_set(inode, FI_INLINE_DENTRY); 2951 } 2952 2953 static inline int is_file(struct inode *inode, int type) 2954 { 2955 return F2FS_I(inode)->i_advise & type; 2956 } 2957 2958 static inline void set_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 void clear_file(struct inode *inode, int type) 2965 { 2966 F2FS_I(inode)->i_advise &= ~type; 2967 f2fs_mark_inode_dirty_sync(inode, true); 2968 } 2969 2970 static inline bool f2fs_is_time_consistent(struct inode *inode) 2971 { 2972 if (!timespec64_equal(F2FS_I(inode)->i_disk_time, &inode->i_atime)) 2973 return false; 2974 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 1, &inode->i_ctime)) 2975 return false; 2976 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 2, &inode->i_mtime)) 2977 return false; 2978 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 3, 2979 &F2FS_I(inode)->i_crtime)) 2980 return false; 2981 return true; 2982 } 2983 2984 static inline bool f2fs_skip_inode_update(struct inode *inode, int dsync) 2985 { 2986 bool ret; 2987 2988 if (dsync) { 2989 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 2990 2991 spin_lock(&sbi->inode_lock[DIRTY_META]); 2992 ret = list_empty(&F2FS_I(inode)->gdirty_list); 2993 spin_unlock(&sbi->inode_lock[DIRTY_META]); 2994 return ret; 2995 } 2996 if (!is_inode_flag_set(inode, FI_AUTO_RECOVER) || 2997 file_keep_isize(inode) || 2998 i_size_read(inode) & ~PAGE_MASK) 2999 return false; 3000 3001 if (!f2fs_is_time_consistent(inode)) 3002 return false; 3003 3004 spin_lock(&F2FS_I(inode)->i_size_lock); 3005 ret = F2FS_I(inode)->last_disk_size == i_size_read(inode); 3006 spin_unlock(&F2FS_I(inode)->i_size_lock); 3007 3008 return ret; 3009 } 3010 3011 static inline bool f2fs_readonly(struct super_block *sb) 3012 { 3013 return sb_rdonly(sb); 3014 } 3015 3016 static inline bool f2fs_cp_error(struct f2fs_sb_info *sbi) 3017 { 3018 return is_set_ckpt_flags(sbi, CP_ERROR_FLAG); 3019 } 3020 3021 static inline bool is_dot_dotdot(const u8 *name, size_t len) 3022 { 3023 if (len == 1 && name[0] == '.') 3024 return true; 3025 3026 if (len == 2 && name[0] == '.' && name[1] == '.') 3027 return true; 3028 3029 return false; 3030 } 3031 3032 static inline bool f2fs_may_extent_tree(struct inode *inode) 3033 { 3034 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 3035 3036 if (!test_opt(sbi, EXTENT_CACHE) || 3037 is_inode_flag_set(inode, FI_NO_EXTENT) || 3038 is_inode_flag_set(inode, FI_COMPRESSED_FILE)) 3039 return false; 3040 3041 /* 3042 * for recovered files during mount do not create extents 3043 * if shrinker is not registered. 3044 */ 3045 if (list_empty(&sbi->s_list)) 3046 return false; 3047 3048 return S_ISREG(inode->i_mode); 3049 } 3050 3051 static inline void *f2fs_kmalloc(struct f2fs_sb_info *sbi, 3052 size_t size, gfp_t flags) 3053 { 3054 if (time_to_inject(sbi, FAULT_KMALLOC)) { 3055 f2fs_show_injection_info(sbi, FAULT_KMALLOC); 3056 return NULL; 3057 } 3058 3059 return kmalloc(size, flags); 3060 } 3061 3062 static inline void *f2fs_kzalloc(struct f2fs_sb_info *sbi, 3063 size_t size, gfp_t flags) 3064 { 3065 return f2fs_kmalloc(sbi, size, flags | __GFP_ZERO); 3066 } 3067 3068 static inline void *f2fs_kvmalloc(struct f2fs_sb_info *sbi, 3069 size_t size, gfp_t flags) 3070 { 3071 if (time_to_inject(sbi, FAULT_KVMALLOC)) { 3072 f2fs_show_injection_info(sbi, FAULT_KVMALLOC); 3073 return NULL; 3074 } 3075 3076 return kvmalloc(size, flags); 3077 } 3078 3079 static inline void *f2fs_kvzalloc(struct f2fs_sb_info *sbi, 3080 size_t size, gfp_t flags) 3081 { 3082 return f2fs_kvmalloc(sbi, size, flags | __GFP_ZERO); 3083 } 3084 3085 static inline int get_extra_isize(struct inode *inode) 3086 { 3087 return F2FS_I(inode)->i_extra_isize / sizeof(__le32); 3088 } 3089 3090 static inline int get_inline_xattr_addrs(struct inode *inode) 3091 { 3092 return F2FS_I(inode)->i_inline_xattr_size; 3093 } 3094 3095 #define f2fs_get_inode_mode(i) \ 3096 ((is_inode_flag_set(i, FI_ACL_MODE)) ? \ 3097 (F2FS_I(i)->i_acl_mode) : ((i)->i_mode)) 3098 3099 #define F2FS_TOTAL_EXTRA_ATTR_SIZE \ 3100 (offsetof(struct f2fs_inode, i_extra_end) - \ 3101 offsetof(struct f2fs_inode, i_extra_isize)) \ 3102 3103 #define F2FS_OLD_ATTRIBUTE_SIZE (offsetof(struct f2fs_inode, i_addr)) 3104 #define F2FS_FITS_IN_INODE(f2fs_inode, extra_isize, field) \ 3105 ((offsetof(typeof(*(f2fs_inode)), field) + \ 3106 sizeof((f2fs_inode)->field)) \ 3107 <= (F2FS_OLD_ATTRIBUTE_SIZE + (extra_isize))) \ 3108 3109 #define DEFAULT_IOSTAT_PERIOD_MS 3000 3110 #define MIN_IOSTAT_PERIOD_MS 100 3111 /* maximum period of iostat tracing is 1 day */ 3112 #define MAX_IOSTAT_PERIOD_MS 8640000 3113 3114 static inline void f2fs_reset_iostat(struct f2fs_sb_info *sbi) 3115 { 3116 int i; 3117 3118 spin_lock(&sbi->iostat_lock); 3119 for (i = 0; i < NR_IO_TYPE; i++) { 3120 sbi->rw_iostat[i] = 0; 3121 sbi->prev_rw_iostat[i] = 0; 3122 } 3123 spin_unlock(&sbi->iostat_lock); 3124 } 3125 3126 extern void f2fs_record_iostat(struct f2fs_sb_info *sbi); 3127 3128 static inline void f2fs_update_iostat(struct f2fs_sb_info *sbi, 3129 enum iostat_type type, unsigned long long io_bytes) 3130 { 3131 if (!sbi->iostat_enable) 3132 return; 3133 spin_lock(&sbi->iostat_lock); 3134 sbi->rw_iostat[type] += io_bytes; 3135 3136 if (type == APP_WRITE_IO || type == APP_DIRECT_IO) 3137 sbi->rw_iostat[APP_BUFFERED_IO] = 3138 sbi->rw_iostat[APP_WRITE_IO] - 3139 sbi->rw_iostat[APP_DIRECT_IO]; 3140 3141 if (type == APP_READ_IO || type == APP_DIRECT_READ_IO) 3142 sbi->rw_iostat[APP_BUFFERED_READ_IO] = 3143 sbi->rw_iostat[APP_READ_IO] - 3144 sbi->rw_iostat[APP_DIRECT_READ_IO]; 3145 spin_unlock(&sbi->iostat_lock); 3146 3147 f2fs_record_iostat(sbi); 3148 } 3149 3150 #define __is_large_section(sbi) ((sbi)->segs_per_sec > 1) 3151 3152 #define __is_meta_io(fio) (PAGE_TYPE_OF_BIO((fio)->type) == META) 3153 3154 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi, 3155 block_t blkaddr, int type); 3156 static inline void verify_blkaddr(struct f2fs_sb_info *sbi, 3157 block_t blkaddr, int type) 3158 { 3159 if (!f2fs_is_valid_blkaddr(sbi, blkaddr, type)) { 3160 f2fs_err(sbi, "invalid blkaddr: %u, type: %d, run fsck to fix.", 3161 blkaddr, type); 3162 f2fs_bug_on(sbi, 1); 3163 } 3164 } 3165 3166 static inline bool __is_valid_data_blkaddr(block_t blkaddr) 3167 { 3168 if (blkaddr == NEW_ADDR || blkaddr == NULL_ADDR || 3169 blkaddr == COMPRESS_ADDR) 3170 return false; 3171 return true; 3172 } 3173 3174 static inline void f2fs_set_page_private(struct page *page, 3175 unsigned long data) 3176 { 3177 if (PagePrivate(page)) 3178 return; 3179 3180 attach_page_private(page, (void *)data); 3181 } 3182 3183 static inline void f2fs_clear_page_private(struct page *page) 3184 { 3185 detach_page_private(page); 3186 } 3187 3188 /* 3189 * file.c 3190 */ 3191 int f2fs_sync_file(struct file *file, loff_t start, loff_t end, int datasync); 3192 void f2fs_truncate_data_blocks(struct dnode_of_data *dn); 3193 int f2fs_do_truncate_blocks(struct inode *inode, u64 from, bool lock); 3194 int f2fs_truncate_blocks(struct inode *inode, u64 from, bool lock); 3195 int f2fs_truncate(struct inode *inode); 3196 int f2fs_getattr(struct user_namespace *mnt_userns, const struct path *path, 3197 struct kstat *stat, u32 request_mask, unsigned int flags); 3198 int f2fs_setattr(struct user_namespace *mnt_userns, struct dentry *dentry, 3199 struct iattr *attr); 3200 int f2fs_truncate_hole(struct inode *inode, pgoff_t pg_start, pgoff_t pg_end); 3201 void f2fs_truncate_data_blocks_range(struct dnode_of_data *dn, int count); 3202 int f2fs_precache_extents(struct inode *inode); 3203 long f2fs_ioctl(struct file *filp, unsigned int cmd, unsigned long arg); 3204 long f2fs_compat_ioctl(struct file *file, unsigned int cmd, unsigned long arg); 3205 int f2fs_transfer_project_quota(struct inode *inode, kprojid_t kprojid); 3206 int f2fs_pin_file_control(struct inode *inode, bool inc); 3207 3208 /* 3209 * inode.c 3210 */ 3211 void f2fs_set_inode_flags(struct inode *inode); 3212 bool f2fs_inode_chksum_verify(struct f2fs_sb_info *sbi, struct page *page); 3213 void f2fs_inode_chksum_set(struct f2fs_sb_info *sbi, struct page *page); 3214 struct inode *f2fs_iget(struct super_block *sb, unsigned long ino); 3215 struct inode *f2fs_iget_retry(struct super_block *sb, unsigned long ino); 3216 int f2fs_try_to_free_nats(struct f2fs_sb_info *sbi, int nr_shrink); 3217 void f2fs_update_inode(struct inode *inode, struct page *node_page); 3218 void f2fs_update_inode_page(struct inode *inode); 3219 int f2fs_write_inode(struct inode *inode, struct writeback_control *wbc); 3220 void f2fs_evict_inode(struct inode *inode); 3221 void f2fs_handle_failed_inode(struct inode *inode); 3222 3223 /* 3224 * namei.c 3225 */ 3226 int f2fs_update_extension_list(struct f2fs_sb_info *sbi, const char *name, 3227 bool hot, bool set); 3228 struct dentry *f2fs_get_parent(struct dentry *child); 3229 3230 /* 3231 * dir.c 3232 */ 3233 unsigned char f2fs_get_de_type(struct f2fs_dir_entry *de); 3234 int f2fs_init_casefolded_name(const struct inode *dir, 3235 struct f2fs_filename *fname); 3236 int f2fs_setup_filename(struct inode *dir, const struct qstr *iname, 3237 int lookup, struct f2fs_filename *fname); 3238 int f2fs_prepare_lookup(struct inode *dir, struct dentry *dentry, 3239 struct f2fs_filename *fname); 3240 void f2fs_free_filename(struct f2fs_filename *fname); 3241 struct f2fs_dir_entry *f2fs_find_target_dentry(const struct f2fs_dentry_ptr *d, 3242 const struct f2fs_filename *fname, int *max_slots); 3243 int f2fs_fill_dentries(struct dir_context *ctx, struct f2fs_dentry_ptr *d, 3244 unsigned int start_pos, struct fscrypt_str *fstr); 3245 void f2fs_do_make_empty_dir(struct inode *inode, struct inode *parent, 3246 struct f2fs_dentry_ptr *d); 3247 struct page *f2fs_init_inode_metadata(struct inode *inode, struct inode *dir, 3248 const struct f2fs_filename *fname, struct page *dpage); 3249 void f2fs_update_parent_metadata(struct inode *dir, struct inode *inode, 3250 unsigned int current_depth); 3251 int f2fs_room_for_filename(const void *bitmap, int slots, int max_slots); 3252 void f2fs_drop_nlink(struct inode *dir, struct inode *inode); 3253 struct f2fs_dir_entry *__f2fs_find_entry(struct inode *dir, 3254 const struct f2fs_filename *fname, 3255 struct page **res_page); 3256 struct f2fs_dir_entry *f2fs_find_entry(struct inode *dir, 3257 const struct qstr *child, struct page **res_page); 3258 struct f2fs_dir_entry *f2fs_parent_dir(struct inode *dir, struct page **p); 3259 ino_t f2fs_inode_by_name(struct inode *dir, const struct qstr *qstr, 3260 struct page **page); 3261 void f2fs_set_link(struct inode *dir, struct f2fs_dir_entry *de, 3262 struct page *page, struct inode *inode); 3263 bool f2fs_has_enough_room(struct inode *dir, struct page *ipage, 3264 const struct f2fs_filename *fname); 3265 void f2fs_update_dentry(nid_t ino, umode_t mode, struct f2fs_dentry_ptr *d, 3266 const struct fscrypt_str *name, f2fs_hash_t name_hash, 3267 unsigned int bit_pos); 3268 int f2fs_add_regular_entry(struct inode *dir, const struct f2fs_filename *fname, 3269 struct inode *inode, nid_t ino, umode_t mode); 3270 int f2fs_add_dentry(struct inode *dir, const struct f2fs_filename *fname, 3271 struct inode *inode, nid_t ino, umode_t mode); 3272 int f2fs_do_add_link(struct inode *dir, const struct qstr *name, 3273 struct inode *inode, nid_t ino, umode_t mode); 3274 void f2fs_delete_entry(struct f2fs_dir_entry *dentry, struct page *page, 3275 struct inode *dir, struct inode *inode); 3276 int f2fs_do_tmpfile(struct inode *inode, struct inode *dir); 3277 bool f2fs_empty_dir(struct inode *dir); 3278 3279 static inline int f2fs_add_link(struct dentry *dentry, struct inode *inode) 3280 { 3281 if (fscrypt_is_nokey_name(dentry)) 3282 return -ENOKEY; 3283 return f2fs_do_add_link(d_inode(dentry->d_parent), &dentry->d_name, 3284 inode, inode->i_ino, inode->i_mode); 3285 } 3286 3287 /* 3288 * super.c 3289 */ 3290 int f2fs_inode_dirtied(struct inode *inode, bool sync); 3291 void f2fs_inode_synced(struct inode *inode); 3292 int f2fs_enable_quota_files(struct f2fs_sb_info *sbi, bool rdonly); 3293 int f2fs_quota_sync(struct super_block *sb, int type); 3294 loff_t max_file_blocks(struct inode *inode); 3295 void f2fs_quota_off_umount(struct super_block *sb); 3296 int f2fs_commit_super(struct f2fs_sb_info *sbi, bool recover); 3297 int f2fs_sync_fs(struct super_block *sb, int sync); 3298 int f2fs_sanity_check_ckpt(struct f2fs_sb_info *sbi); 3299 3300 /* 3301 * hash.c 3302 */ 3303 void f2fs_hash_filename(const struct inode *dir, struct f2fs_filename *fname); 3304 3305 /* 3306 * node.c 3307 */ 3308 struct dnode_of_data; 3309 struct node_info; 3310 3311 int f2fs_check_nid_range(struct f2fs_sb_info *sbi, nid_t nid); 3312 bool f2fs_available_free_memory(struct f2fs_sb_info *sbi, int type); 3313 bool f2fs_in_warm_node_list(struct f2fs_sb_info *sbi, struct page *page); 3314 void f2fs_init_fsync_node_info(struct f2fs_sb_info *sbi); 3315 void f2fs_del_fsync_node_entry(struct f2fs_sb_info *sbi, struct page *page); 3316 void f2fs_reset_fsync_node_info(struct f2fs_sb_info *sbi); 3317 int f2fs_need_dentry_mark(struct f2fs_sb_info *sbi, nid_t nid); 3318 bool f2fs_is_checkpointed_node(struct f2fs_sb_info *sbi, nid_t nid); 3319 bool f2fs_need_inode_block_update(struct f2fs_sb_info *sbi, nid_t ino); 3320 int f2fs_get_node_info(struct f2fs_sb_info *sbi, nid_t nid, 3321 struct node_info *ni); 3322 pgoff_t f2fs_get_next_page_offset(struct dnode_of_data *dn, pgoff_t pgofs); 3323 int f2fs_get_dnode_of_data(struct dnode_of_data *dn, pgoff_t index, int mode); 3324 int f2fs_truncate_inode_blocks(struct inode *inode, pgoff_t from); 3325 int f2fs_truncate_xattr_node(struct inode *inode); 3326 int f2fs_wait_on_node_pages_writeback(struct f2fs_sb_info *sbi, 3327 unsigned int seq_id); 3328 int f2fs_remove_inode_page(struct inode *inode); 3329 struct page *f2fs_new_inode_page(struct inode *inode); 3330 struct page *f2fs_new_node_page(struct dnode_of_data *dn, unsigned int ofs); 3331 void f2fs_ra_node_page(struct f2fs_sb_info *sbi, nid_t nid); 3332 struct page *f2fs_get_node_page(struct f2fs_sb_info *sbi, pgoff_t nid); 3333 struct page *f2fs_get_node_page_ra(struct page *parent, int start); 3334 int f2fs_move_node_page(struct page *node_page, int gc_type); 3335 void f2fs_flush_inline_data(struct f2fs_sb_info *sbi); 3336 int f2fs_fsync_node_pages(struct f2fs_sb_info *sbi, struct inode *inode, 3337 struct writeback_control *wbc, bool atomic, 3338 unsigned int *seq_id); 3339 int f2fs_sync_node_pages(struct f2fs_sb_info *sbi, 3340 struct writeback_control *wbc, 3341 bool do_balance, enum iostat_type io_type); 3342 int f2fs_build_free_nids(struct f2fs_sb_info *sbi, bool sync, bool mount); 3343 bool f2fs_alloc_nid(struct f2fs_sb_info *sbi, nid_t *nid); 3344 void f2fs_alloc_nid_done(struct f2fs_sb_info *sbi, nid_t nid); 3345 void f2fs_alloc_nid_failed(struct f2fs_sb_info *sbi, nid_t nid); 3346 int f2fs_try_to_free_nids(struct f2fs_sb_info *sbi, int nr_shrink); 3347 int f2fs_recover_inline_xattr(struct inode *inode, struct page *page); 3348 int f2fs_recover_xattr_data(struct inode *inode, struct page *page); 3349 int f2fs_recover_inode_page(struct f2fs_sb_info *sbi, struct page *page); 3350 int f2fs_restore_node_summary(struct f2fs_sb_info *sbi, 3351 unsigned int segno, struct f2fs_summary_block *sum); 3352 int f2fs_flush_nat_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3353 int f2fs_build_node_manager(struct f2fs_sb_info *sbi); 3354 void f2fs_destroy_node_manager(struct f2fs_sb_info *sbi); 3355 int __init f2fs_create_node_manager_caches(void); 3356 void f2fs_destroy_node_manager_caches(void); 3357 3358 /* 3359 * segment.c 3360 */ 3361 bool f2fs_need_SSR(struct f2fs_sb_info *sbi); 3362 void f2fs_register_inmem_page(struct inode *inode, struct page *page); 3363 void f2fs_drop_inmem_pages_all(struct f2fs_sb_info *sbi, bool gc_failure); 3364 void f2fs_drop_inmem_pages(struct inode *inode); 3365 void f2fs_drop_inmem_page(struct inode *inode, struct page *page); 3366 int f2fs_commit_inmem_pages(struct inode *inode); 3367 void f2fs_balance_fs(struct f2fs_sb_info *sbi, bool need); 3368 void f2fs_balance_fs_bg(struct f2fs_sb_info *sbi, bool from_bg); 3369 int f2fs_issue_flush(struct f2fs_sb_info *sbi, nid_t ino); 3370 int f2fs_create_flush_cmd_control(struct f2fs_sb_info *sbi); 3371 int f2fs_flush_device_cache(struct f2fs_sb_info *sbi); 3372 void f2fs_destroy_flush_cmd_control(struct f2fs_sb_info *sbi, bool free); 3373 void f2fs_invalidate_blocks(struct f2fs_sb_info *sbi, block_t addr); 3374 bool f2fs_is_checkpointed_data(struct f2fs_sb_info *sbi, block_t blkaddr); 3375 void f2fs_drop_discard_cmd(struct f2fs_sb_info *sbi); 3376 void f2fs_stop_discard_thread(struct f2fs_sb_info *sbi); 3377 bool f2fs_issue_discard_timeout(struct f2fs_sb_info *sbi); 3378 void f2fs_clear_prefree_segments(struct f2fs_sb_info *sbi, 3379 struct cp_control *cpc); 3380 void f2fs_dirty_to_prefree(struct f2fs_sb_info *sbi); 3381 block_t f2fs_get_unusable_blocks(struct f2fs_sb_info *sbi); 3382 int f2fs_disable_cp_again(struct f2fs_sb_info *sbi, block_t unusable); 3383 void f2fs_release_discard_addrs(struct f2fs_sb_info *sbi); 3384 int f2fs_npages_for_summary_flush(struct f2fs_sb_info *sbi, bool for_ra); 3385 void f2fs_init_inmem_curseg(struct f2fs_sb_info *sbi); 3386 void f2fs_save_inmem_curseg(struct f2fs_sb_info *sbi); 3387 void f2fs_restore_inmem_curseg(struct f2fs_sb_info *sbi); 3388 void f2fs_get_new_segment(struct f2fs_sb_info *sbi, 3389 unsigned int *newseg, bool new_sec, int dir); 3390 void f2fs_allocate_segment_for_resize(struct f2fs_sb_info *sbi, int type, 3391 unsigned int start, unsigned int end); 3392 void f2fs_allocate_new_segment(struct f2fs_sb_info *sbi, int type); 3393 void f2fs_allocate_new_segments(struct f2fs_sb_info *sbi); 3394 int f2fs_trim_fs(struct f2fs_sb_info *sbi, struct fstrim_range *range); 3395 bool f2fs_exist_trim_candidates(struct f2fs_sb_info *sbi, 3396 struct cp_control *cpc); 3397 struct page *f2fs_get_sum_page(struct f2fs_sb_info *sbi, unsigned int segno); 3398 void f2fs_update_meta_page(struct f2fs_sb_info *sbi, void *src, 3399 block_t blk_addr); 3400 void f2fs_do_write_meta_page(struct f2fs_sb_info *sbi, struct page *page, 3401 enum iostat_type io_type); 3402 void f2fs_do_write_node_page(unsigned int nid, struct f2fs_io_info *fio); 3403 void f2fs_outplace_write_data(struct dnode_of_data *dn, 3404 struct f2fs_io_info *fio); 3405 int f2fs_inplace_write_data(struct f2fs_io_info *fio); 3406 void f2fs_do_replace_block(struct f2fs_sb_info *sbi, struct f2fs_summary *sum, 3407 block_t old_blkaddr, block_t new_blkaddr, 3408 bool recover_curseg, bool recover_newaddr, 3409 bool from_gc); 3410 void f2fs_replace_block(struct f2fs_sb_info *sbi, struct dnode_of_data *dn, 3411 block_t old_addr, block_t new_addr, 3412 unsigned char version, bool recover_curseg, 3413 bool recover_newaddr); 3414 void f2fs_allocate_data_block(struct f2fs_sb_info *sbi, struct page *page, 3415 block_t old_blkaddr, block_t *new_blkaddr, 3416 struct f2fs_summary *sum, int type, 3417 struct f2fs_io_info *fio); 3418 void f2fs_wait_on_page_writeback(struct page *page, 3419 enum page_type type, bool ordered, bool locked); 3420 void f2fs_wait_on_block_writeback(struct inode *inode, block_t blkaddr); 3421 void f2fs_wait_on_block_writeback_range(struct inode *inode, block_t blkaddr, 3422 block_t len); 3423 void f2fs_write_data_summaries(struct f2fs_sb_info *sbi, block_t start_blk); 3424 void f2fs_write_node_summaries(struct f2fs_sb_info *sbi, block_t start_blk); 3425 int f2fs_lookup_journal_in_cursum(struct f2fs_journal *journal, int type, 3426 unsigned int val, int alloc); 3427 void f2fs_flush_sit_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3428 int f2fs_fix_curseg_write_pointer(struct f2fs_sb_info *sbi); 3429 int f2fs_check_write_pointer(struct f2fs_sb_info *sbi); 3430 int f2fs_build_segment_manager(struct f2fs_sb_info *sbi); 3431 void f2fs_destroy_segment_manager(struct f2fs_sb_info *sbi); 3432 int __init f2fs_create_segment_manager_caches(void); 3433 void f2fs_destroy_segment_manager_caches(void); 3434 int f2fs_rw_hint_to_seg_type(enum rw_hint hint); 3435 enum rw_hint f2fs_io_type_to_rw_hint(struct f2fs_sb_info *sbi, 3436 enum page_type type, enum temp_type temp); 3437 unsigned int f2fs_usable_segs_in_sec(struct f2fs_sb_info *sbi, 3438 unsigned int segno); 3439 unsigned int f2fs_usable_blks_in_seg(struct f2fs_sb_info *sbi, 3440 unsigned int segno); 3441 3442 /* 3443 * checkpoint.c 3444 */ 3445 void f2fs_stop_checkpoint(struct f2fs_sb_info *sbi, bool end_io); 3446 struct page *f2fs_grab_meta_page(struct f2fs_sb_info *sbi, pgoff_t index); 3447 struct page *f2fs_get_meta_page(struct f2fs_sb_info *sbi, pgoff_t index); 3448 struct page *f2fs_get_meta_page_retry(struct f2fs_sb_info *sbi, pgoff_t index); 3449 struct page *f2fs_get_tmp_page(struct f2fs_sb_info *sbi, pgoff_t index); 3450 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi, 3451 block_t blkaddr, int type); 3452 int f2fs_ra_meta_pages(struct f2fs_sb_info *sbi, block_t start, int nrpages, 3453 int type, bool sync); 3454 void f2fs_ra_meta_pages_cond(struct f2fs_sb_info *sbi, pgoff_t index); 3455 long f2fs_sync_meta_pages(struct f2fs_sb_info *sbi, enum page_type type, 3456 long nr_to_write, enum iostat_type io_type); 3457 void f2fs_add_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type); 3458 void f2fs_remove_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type); 3459 void f2fs_release_ino_entry(struct f2fs_sb_info *sbi, bool all); 3460 bool f2fs_exist_written_data(struct f2fs_sb_info *sbi, nid_t ino, int mode); 3461 void f2fs_set_dirty_device(struct f2fs_sb_info *sbi, nid_t ino, 3462 unsigned int devidx, int type); 3463 bool f2fs_is_dirty_device(struct f2fs_sb_info *sbi, nid_t ino, 3464 unsigned int devidx, int type); 3465 int f2fs_sync_inode_meta(struct f2fs_sb_info *sbi); 3466 int f2fs_acquire_orphan_inode(struct f2fs_sb_info *sbi); 3467 void f2fs_release_orphan_inode(struct f2fs_sb_info *sbi); 3468 void f2fs_add_orphan_inode(struct inode *inode); 3469 void f2fs_remove_orphan_inode(struct f2fs_sb_info *sbi, nid_t ino); 3470 int f2fs_recover_orphan_inodes(struct f2fs_sb_info *sbi); 3471 int f2fs_get_valid_checkpoint(struct f2fs_sb_info *sbi); 3472 void f2fs_update_dirty_page(struct inode *inode, struct page *page); 3473 void f2fs_remove_dirty_inode(struct inode *inode); 3474 int f2fs_sync_dirty_inodes(struct f2fs_sb_info *sbi, enum inode_type type); 3475 void f2fs_wait_on_all_pages(struct f2fs_sb_info *sbi, int type); 3476 u64 f2fs_get_sectors_written(struct f2fs_sb_info *sbi); 3477 int f2fs_write_checkpoint(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3478 void f2fs_init_ino_entry_info(struct f2fs_sb_info *sbi); 3479 int __init f2fs_create_checkpoint_caches(void); 3480 void f2fs_destroy_checkpoint_caches(void); 3481 int f2fs_issue_checkpoint(struct f2fs_sb_info *sbi); 3482 int f2fs_start_ckpt_thread(struct f2fs_sb_info *sbi); 3483 void f2fs_stop_ckpt_thread(struct f2fs_sb_info *sbi); 3484 void f2fs_init_ckpt_req_control(struct f2fs_sb_info *sbi); 3485 3486 /* 3487 * data.c 3488 */ 3489 int __init f2fs_init_bioset(void); 3490 void f2fs_destroy_bioset(void); 3491 int f2fs_init_bio_entry_cache(void); 3492 void f2fs_destroy_bio_entry_cache(void); 3493 void f2fs_submit_bio(struct f2fs_sb_info *sbi, 3494 struct bio *bio, enum page_type type); 3495 void f2fs_submit_merged_write(struct f2fs_sb_info *sbi, enum page_type type); 3496 void f2fs_submit_merged_write_cond(struct f2fs_sb_info *sbi, 3497 struct inode *inode, struct page *page, 3498 nid_t ino, enum page_type type); 3499 void f2fs_submit_merged_ipu_write(struct f2fs_sb_info *sbi, 3500 struct bio **bio, struct page *page); 3501 void f2fs_flush_merged_writes(struct f2fs_sb_info *sbi); 3502 int f2fs_submit_page_bio(struct f2fs_io_info *fio); 3503 int f2fs_merge_page_bio(struct f2fs_io_info *fio); 3504 void f2fs_submit_page_write(struct f2fs_io_info *fio); 3505 struct block_device *f2fs_target_device(struct f2fs_sb_info *sbi, 3506 block_t blk_addr, struct bio *bio); 3507 int f2fs_target_device_index(struct f2fs_sb_info *sbi, block_t blkaddr); 3508 void f2fs_set_data_blkaddr(struct dnode_of_data *dn); 3509 void f2fs_update_data_blkaddr(struct dnode_of_data *dn, block_t blkaddr); 3510 int f2fs_reserve_new_blocks(struct dnode_of_data *dn, blkcnt_t count); 3511 int f2fs_reserve_new_block(struct dnode_of_data *dn); 3512 int f2fs_get_block(struct dnode_of_data *dn, pgoff_t index); 3513 int f2fs_preallocate_blocks(struct kiocb *iocb, struct iov_iter *from); 3514 int f2fs_reserve_block(struct dnode_of_data *dn, pgoff_t index); 3515 struct page *f2fs_get_read_data_page(struct inode *inode, pgoff_t index, 3516 int op_flags, bool for_write); 3517 struct page *f2fs_find_data_page(struct inode *inode, pgoff_t index); 3518 struct page *f2fs_get_lock_data_page(struct inode *inode, pgoff_t index, 3519 bool for_write); 3520 struct page *f2fs_get_new_data_page(struct inode *inode, 3521 struct page *ipage, pgoff_t index, bool new_i_size); 3522 int f2fs_do_write_data_page(struct f2fs_io_info *fio); 3523 void f2fs_do_map_lock(struct f2fs_sb_info *sbi, int flag, bool lock); 3524 int f2fs_map_blocks(struct inode *inode, struct f2fs_map_blocks *map, 3525 int create, int flag); 3526 int f2fs_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, 3527 u64 start, u64 len); 3528 int f2fs_encrypt_one_page(struct f2fs_io_info *fio); 3529 bool f2fs_should_update_inplace(struct inode *inode, struct f2fs_io_info *fio); 3530 bool f2fs_should_update_outplace(struct inode *inode, struct f2fs_io_info *fio); 3531 int f2fs_write_single_data_page(struct page *page, int *submitted, 3532 struct bio **bio, sector_t *last_block, 3533 struct writeback_control *wbc, 3534 enum iostat_type io_type, 3535 int compr_blocks, bool allow_balance); 3536 void f2fs_invalidate_page(struct page *page, unsigned int offset, 3537 unsigned int length); 3538 int f2fs_release_page(struct page *page, gfp_t wait); 3539 #ifdef CONFIG_MIGRATION 3540 int f2fs_migrate_page(struct address_space *mapping, struct page *newpage, 3541 struct page *page, enum migrate_mode mode); 3542 #endif 3543 bool f2fs_overwrite_io(struct inode *inode, loff_t pos, size_t len); 3544 void f2fs_clear_page_cache_dirty_tag(struct page *page); 3545 int f2fs_init_post_read_processing(void); 3546 void f2fs_destroy_post_read_processing(void); 3547 int f2fs_init_post_read_wq(struct f2fs_sb_info *sbi); 3548 void f2fs_destroy_post_read_wq(struct f2fs_sb_info *sbi); 3549 3550 /* 3551 * gc.c 3552 */ 3553 int f2fs_start_gc_thread(struct f2fs_sb_info *sbi); 3554 void f2fs_stop_gc_thread(struct f2fs_sb_info *sbi); 3555 block_t f2fs_start_bidx_of_node(unsigned int node_ofs, struct inode *inode); 3556 int f2fs_gc(struct f2fs_sb_info *sbi, bool sync, bool background, bool force, 3557 unsigned int segno); 3558 void f2fs_build_gc_manager(struct f2fs_sb_info *sbi); 3559 int f2fs_resize_fs(struct f2fs_sb_info *sbi, __u64 block_count); 3560 int __init f2fs_create_garbage_collection_cache(void); 3561 void f2fs_destroy_garbage_collection_cache(void); 3562 3563 /* 3564 * recovery.c 3565 */ 3566 int f2fs_recover_fsync_data(struct f2fs_sb_info *sbi, bool check_only); 3567 bool f2fs_space_for_roll_forward(struct f2fs_sb_info *sbi); 3568 3569 /* 3570 * debug.c 3571 */ 3572 #ifdef CONFIG_F2FS_STAT_FS 3573 struct f2fs_stat_info { 3574 struct list_head stat_list; 3575 struct f2fs_sb_info *sbi; 3576 int all_area_segs, sit_area_segs, nat_area_segs, ssa_area_segs; 3577 int main_area_segs, main_area_sections, main_area_zones; 3578 unsigned long long hit_largest, hit_cached, hit_rbtree; 3579 unsigned long long hit_total, total_ext; 3580 int ext_tree, zombie_tree, ext_node; 3581 int ndirty_node, ndirty_dent, ndirty_meta, ndirty_imeta; 3582 int ndirty_data, ndirty_qdata; 3583 int inmem_pages; 3584 unsigned int ndirty_dirs, ndirty_files, nquota_files, ndirty_all; 3585 int nats, dirty_nats, sits, dirty_sits; 3586 int free_nids, avail_nids, alloc_nids; 3587 int total_count, utilization; 3588 int bg_gc, nr_wb_cp_data, nr_wb_data; 3589 int nr_rd_data, nr_rd_node, nr_rd_meta; 3590 int nr_dio_read, nr_dio_write; 3591 unsigned int io_skip_bggc, other_skip_bggc; 3592 int nr_flushing, nr_flushed, flush_list_empty; 3593 int nr_discarding, nr_discarded; 3594 int nr_discard_cmd; 3595 unsigned int undiscard_blks; 3596 int nr_issued_ckpt, nr_total_ckpt, nr_queued_ckpt; 3597 unsigned int cur_ckpt_time, peak_ckpt_time; 3598 int inline_xattr, inline_inode, inline_dir, append, update, orphans; 3599 int compr_inode; 3600 unsigned long long compr_blocks; 3601 int aw_cnt, max_aw_cnt, vw_cnt, max_vw_cnt; 3602 unsigned int valid_count, valid_node_count, valid_inode_count, discard_blks; 3603 unsigned int bimodal, avg_vblocks; 3604 int util_free, util_valid, util_invalid; 3605 int rsvd_segs, overp_segs; 3606 int dirty_count, node_pages, meta_pages; 3607 int prefree_count, call_count, cp_count, bg_cp_count; 3608 int tot_segs, node_segs, data_segs, free_segs, free_secs; 3609 int bg_node_segs, bg_data_segs; 3610 int tot_blks, data_blks, node_blks; 3611 int bg_data_blks, bg_node_blks; 3612 unsigned long long skipped_atomic_files[2]; 3613 int curseg[NR_CURSEG_TYPE]; 3614 int cursec[NR_CURSEG_TYPE]; 3615 int curzone[NR_CURSEG_TYPE]; 3616 unsigned int dirty_seg[NR_CURSEG_TYPE]; 3617 unsigned int full_seg[NR_CURSEG_TYPE]; 3618 unsigned int valid_blks[NR_CURSEG_TYPE]; 3619 3620 unsigned int meta_count[META_MAX]; 3621 unsigned int segment_count[2]; 3622 unsigned int block_count[2]; 3623 unsigned int inplace_count; 3624 unsigned long long base_mem, cache_mem, page_mem; 3625 }; 3626 3627 static inline struct f2fs_stat_info *F2FS_STAT(struct f2fs_sb_info *sbi) 3628 { 3629 return (struct f2fs_stat_info *)sbi->stat_info; 3630 } 3631 3632 #define stat_inc_cp_count(si) ((si)->cp_count++) 3633 #define stat_inc_bg_cp_count(si) ((si)->bg_cp_count++) 3634 #define stat_inc_call_count(si) ((si)->call_count++) 3635 #define stat_inc_bggc_count(si) ((si)->bg_gc++) 3636 #define stat_io_skip_bggc_count(sbi) ((sbi)->io_skip_bggc++) 3637 #define stat_other_skip_bggc_count(sbi) ((sbi)->other_skip_bggc++) 3638 #define stat_inc_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]++) 3639 #define stat_dec_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]--) 3640 #define stat_inc_total_hit(sbi) (atomic64_inc(&(sbi)->total_hit_ext)) 3641 #define stat_inc_rbtree_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_rbtree)) 3642 #define stat_inc_largest_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_largest)) 3643 #define stat_inc_cached_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_cached)) 3644 #define stat_inc_inline_xattr(inode) \ 3645 do { \ 3646 if (f2fs_has_inline_xattr(inode)) \ 3647 (atomic_inc(&F2FS_I_SB(inode)->inline_xattr)); \ 3648 } while (0) 3649 #define stat_dec_inline_xattr(inode) \ 3650 do { \ 3651 if (f2fs_has_inline_xattr(inode)) \ 3652 (atomic_dec(&F2FS_I_SB(inode)->inline_xattr)); \ 3653 } while (0) 3654 #define stat_inc_inline_inode(inode) \ 3655 do { \ 3656 if (f2fs_has_inline_data(inode)) \ 3657 (atomic_inc(&F2FS_I_SB(inode)->inline_inode)); \ 3658 } while (0) 3659 #define stat_dec_inline_inode(inode) \ 3660 do { \ 3661 if (f2fs_has_inline_data(inode)) \ 3662 (atomic_dec(&F2FS_I_SB(inode)->inline_inode)); \ 3663 } while (0) 3664 #define stat_inc_inline_dir(inode) \ 3665 do { \ 3666 if (f2fs_has_inline_dentry(inode)) \ 3667 (atomic_inc(&F2FS_I_SB(inode)->inline_dir)); \ 3668 } while (0) 3669 #define stat_dec_inline_dir(inode) \ 3670 do { \ 3671 if (f2fs_has_inline_dentry(inode)) \ 3672 (atomic_dec(&F2FS_I_SB(inode)->inline_dir)); \ 3673 } while (0) 3674 #define stat_inc_compr_inode(inode) \ 3675 do { \ 3676 if (f2fs_compressed_file(inode)) \ 3677 (atomic_inc(&F2FS_I_SB(inode)->compr_inode)); \ 3678 } while (0) 3679 #define stat_dec_compr_inode(inode) \ 3680 do { \ 3681 if (f2fs_compressed_file(inode)) \ 3682 (atomic_dec(&F2FS_I_SB(inode)->compr_inode)); \ 3683 } while (0) 3684 #define stat_add_compr_blocks(inode, blocks) \ 3685 (atomic64_add(blocks, &F2FS_I_SB(inode)->compr_blocks)) 3686 #define stat_sub_compr_blocks(inode, blocks) \ 3687 (atomic64_sub(blocks, &F2FS_I_SB(inode)->compr_blocks)) 3688 #define stat_inc_meta_count(sbi, blkaddr) \ 3689 do { \ 3690 if (blkaddr < SIT_I(sbi)->sit_base_addr) \ 3691 atomic_inc(&(sbi)->meta_count[META_CP]); \ 3692 else if (blkaddr < NM_I(sbi)->nat_blkaddr) \ 3693 atomic_inc(&(sbi)->meta_count[META_SIT]); \ 3694 else if (blkaddr < SM_I(sbi)->ssa_blkaddr) \ 3695 atomic_inc(&(sbi)->meta_count[META_NAT]); \ 3696 else if (blkaddr < SM_I(sbi)->main_blkaddr) \ 3697 atomic_inc(&(sbi)->meta_count[META_SSA]); \ 3698 } while (0) 3699 #define stat_inc_seg_type(sbi, curseg) \ 3700 ((sbi)->segment_count[(curseg)->alloc_type]++) 3701 #define stat_inc_block_count(sbi, curseg) \ 3702 ((sbi)->block_count[(curseg)->alloc_type]++) 3703 #define stat_inc_inplace_blocks(sbi) \ 3704 (atomic_inc(&(sbi)->inplace_count)) 3705 #define stat_update_max_atomic_write(inode) \ 3706 do { \ 3707 int cur = F2FS_I_SB(inode)->atomic_files; \ 3708 int max = atomic_read(&F2FS_I_SB(inode)->max_aw_cnt); \ 3709 if (cur > max) \ 3710 atomic_set(&F2FS_I_SB(inode)->max_aw_cnt, cur); \ 3711 } while (0) 3712 #define stat_inc_volatile_write(inode) \ 3713 (atomic_inc(&F2FS_I_SB(inode)->vw_cnt)) 3714 #define stat_dec_volatile_write(inode) \ 3715 (atomic_dec(&F2FS_I_SB(inode)->vw_cnt)) 3716 #define stat_update_max_volatile_write(inode) \ 3717 do { \ 3718 int cur = atomic_read(&F2FS_I_SB(inode)->vw_cnt); \ 3719 int max = atomic_read(&F2FS_I_SB(inode)->max_vw_cnt); \ 3720 if (cur > max) \ 3721 atomic_set(&F2FS_I_SB(inode)->max_vw_cnt, cur); \ 3722 } while (0) 3723 #define stat_inc_seg_count(sbi, type, gc_type) \ 3724 do { \ 3725 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 3726 si->tot_segs++; \ 3727 if ((type) == SUM_TYPE_DATA) { \ 3728 si->data_segs++; \ 3729 si->bg_data_segs += (gc_type == BG_GC) ? 1 : 0; \ 3730 } else { \ 3731 si->node_segs++; \ 3732 si->bg_node_segs += (gc_type == BG_GC) ? 1 : 0; \ 3733 } \ 3734 } while (0) 3735 3736 #define stat_inc_tot_blk_count(si, blks) \ 3737 ((si)->tot_blks += (blks)) 3738 3739 #define stat_inc_data_blk_count(sbi, blks, gc_type) \ 3740 do { \ 3741 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 3742 stat_inc_tot_blk_count(si, blks); \ 3743 si->data_blks += (blks); \ 3744 si->bg_data_blks += ((gc_type) == BG_GC) ? (blks) : 0; \ 3745 } while (0) 3746 3747 #define stat_inc_node_blk_count(sbi, blks, gc_type) \ 3748 do { \ 3749 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 3750 stat_inc_tot_blk_count(si, blks); \ 3751 si->node_blks += (blks); \ 3752 si->bg_node_blks += ((gc_type) == BG_GC) ? (blks) : 0; \ 3753 } while (0) 3754 3755 int f2fs_build_stats(struct f2fs_sb_info *sbi); 3756 void f2fs_destroy_stats(struct f2fs_sb_info *sbi); 3757 void __init f2fs_create_root_stats(void); 3758 void f2fs_destroy_root_stats(void); 3759 void f2fs_update_sit_info(struct f2fs_sb_info *sbi); 3760 #else 3761 #define stat_inc_cp_count(si) do { } while (0) 3762 #define stat_inc_bg_cp_count(si) do { } while (0) 3763 #define stat_inc_call_count(si) do { } while (0) 3764 #define stat_inc_bggc_count(si) do { } while (0) 3765 #define stat_io_skip_bggc_count(sbi) do { } while (0) 3766 #define stat_other_skip_bggc_count(sbi) do { } while (0) 3767 #define stat_inc_dirty_inode(sbi, type) do { } while (0) 3768 #define stat_dec_dirty_inode(sbi, type) do { } while (0) 3769 #define stat_inc_total_hit(sbi) do { } while (0) 3770 #define stat_inc_rbtree_node_hit(sbi) do { } while (0) 3771 #define stat_inc_largest_node_hit(sbi) do { } while (0) 3772 #define stat_inc_cached_node_hit(sbi) do { } while (0) 3773 #define stat_inc_inline_xattr(inode) do { } while (0) 3774 #define stat_dec_inline_xattr(inode) do { } while (0) 3775 #define stat_inc_inline_inode(inode) do { } while (0) 3776 #define stat_dec_inline_inode(inode) do { } while (0) 3777 #define stat_inc_inline_dir(inode) do { } while (0) 3778 #define stat_dec_inline_dir(inode) do { } while (0) 3779 #define stat_inc_compr_inode(inode) do { } while (0) 3780 #define stat_dec_compr_inode(inode) do { } while (0) 3781 #define stat_add_compr_blocks(inode, blocks) do { } while (0) 3782 #define stat_sub_compr_blocks(inode, blocks) do { } while (0) 3783 #define stat_update_max_atomic_write(inode) do { } while (0) 3784 #define stat_inc_volatile_write(inode) do { } while (0) 3785 #define stat_dec_volatile_write(inode) do { } while (0) 3786 #define stat_update_max_volatile_write(inode) do { } while (0) 3787 #define stat_inc_meta_count(sbi, blkaddr) do { } while (0) 3788 #define stat_inc_seg_type(sbi, curseg) do { } while (0) 3789 #define stat_inc_block_count(sbi, curseg) do { } while (0) 3790 #define stat_inc_inplace_blocks(sbi) do { } while (0) 3791 #define stat_inc_seg_count(sbi, type, gc_type) do { } while (0) 3792 #define stat_inc_tot_blk_count(si, blks) do { } while (0) 3793 #define stat_inc_data_blk_count(sbi, blks, gc_type) do { } while (0) 3794 #define stat_inc_node_blk_count(sbi, blks, gc_type) do { } while (0) 3795 3796 static inline int f2fs_build_stats(struct f2fs_sb_info *sbi) { return 0; } 3797 static inline void f2fs_destroy_stats(struct f2fs_sb_info *sbi) { } 3798 static inline void __init f2fs_create_root_stats(void) { } 3799 static inline void f2fs_destroy_root_stats(void) { } 3800 static inline void f2fs_update_sit_info(struct f2fs_sb_info *sbi) {} 3801 #endif 3802 3803 extern const struct file_operations f2fs_dir_operations; 3804 extern const struct file_operations f2fs_file_operations; 3805 extern const struct inode_operations f2fs_file_inode_operations; 3806 extern const struct address_space_operations f2fs_dblock_aops; 3807 extern const struct address_space_operations f2fs_node_aops; 3808 extern const struct address_space_operations f2fs_meta_aops; 3809 extern const struct inode_operations f2fs_dir_inode_operations; 3810 extern const struct inode_operations f2fs_symlink_inode_operations; 3811 extern const struct inode_operations f2fs_encrypted_symlink_inode_operations; 3812 extern const struct inode_operations f2fs_special_inode_operations; 3813 extern struct kmem_cache *f2fs_inode_entry_slab; 3814 3815 /* 3816 * inline.c 3817 */ 3818 bool f2fs_may_inline_data(struct inode *inode); 3819 bool f2fs_may_inline_dentry(struct inode *inode); 3820 void f2fs_do_read_inline_data(struct page *page, struct page *ipage); 3821 void f2fs_truncate_inline_inode(struct inode *inode, 3822 struct page *ipage, u64 from); 3823 int f2fs_read_inline_data(struct inode *inode, struct page *page); 3824 int f2fs_convert_inline_page(struct dnode_of_data *dn, struct page *page); 3825 int f2fs_convert_inline_inode(struct inode *inode); 3826 int f2fs_try_convert_inline_dir(struct inode *dir, struct dentry *dentry); 3827 int f2fs_write_inline_data(struct inode *inode, struct page *page); 3828 int f2fs_recover_inline_data(struct inode *inode, struct page *npage); 3829 struct f2fs_dir_entry *f2fs_find_in_inline_dir(struct inode *dir, 3830 const struct f2fs_filename *fname, 3831 struct page **res_page); 3832 int f2fs_make_empty_inline_dir(struct inode *inode, struct inode *parent, 3833 struct page *ipage); 3834 int f2fs_add_inline_entry(struct inode *dir, const struct f2fs_filename *fname, 3835 struct inode *inode, nid_t ino, umode_t mode); 3836 void f2fs_delete_inline_entry(struct f2fs_dir_entry *dentry, 3837 struct page *page, struct inode *dir, 3838 struct inode *inode); 3839 bool f2fs_empty_inline_dir(struct inode *dir); 3840 int f2fs_read_inline_dir(struct file *file, struct dir_context *ctx, 3841 struct fscrypt_str *fstr); 3842 int f2fs_inline_data_fiemap(struct inode *inode, 3843 struct fiemap_extent_info *fieinfo, 3844 __u64 start, __u64 len); 3845 3846 /* 3847 * shrinker.c 3848 */ 3849 unsigned long f2fs_shrink_count(struct shrinker *shrink, 3850 struct shrink_control *sc); 3851 unsigned long f2fs_shrink_scan(struct shrinker *shrink, 3852 struct shrink_control *sc); 3853 void f2fs_join_shrinker(struct f2fs_sb_info *sbi); 3854 void f2fs_leave_shrinker(struct f2fs_sb_info *sbi); 3855 3856 /* 3857 * extent_cache.c 3858 */ 3859 struct rb_entry *f2fs_lookup_rb_tree(struct rb_root_cached *root, 3860 struct rb_entry *cached_re, unsigned int ofs); 3861 struct rb_node **f2fs_lookup_rb_tree_ext(struct f2fs_sb_info *sbi, 3862 struct rb_root_cached *root, 3863 struct rb_node **parent, 3864 unsigned long long key, bool *left_most); 3865 struct rb_node **f2fs_lookup_rb_tree_for_insert(struct f2fs_sb_info *sbi, 3866 struct rb_root_cached *root, 3867 struct rb_node **parent, 3868 unsigned int ofs, bool *leftmost); 3869 struct rb_entry *f2fs_lookup_rb_tree_ret(struct rb_root_cached *root, 3870 struct rb_entry *cached_re, unsigned int ofs, 3871 struct rb_entry **prev_entry, struct rb_entry **next_entry, 3872 struct rb_node ***insert_p, struct rb_node **insert_parent, 3873 bool force, bool *leftmost); 3874 bool f2fs_check_rb_tree_consistence(struct f2fs_sb_info *sbi, 3875 struct rb_root_cached *root, bool check_key); 3876 unsigned int f2fs_shrink_extent_tree(struct f2fs_sb_info *sbi, int nr_shrink); 3877 void f2fs_init_extent_tree(struct inode *inode, struct page *ipage); 3878 void f2fs_drop_extent_tree(struct inode *inode); 3879 unsigned int f2fs_destroy_extent_node(struct inode *inode); 3880 void f2fs_destroy_extent_tree(struct inode *inode); 3881 bool f2fs_lookup_extent_cache(struct inode *inode, pgoff_t pgofs, 3882 struct extent_info *ei); 3883 void f2fs_update_extent_cache(struct dnode_of_data *dn); 3884 void f2fs_update_extent_cache_range(struct dnode_of_data *dn, 3885 pgoff_t fofs, block_t blkaddr, unsigned int len); 3886 void f2fs_init_extent_cache_info(struct f2fs_sb_info *sbi); 3887 int __init f2fs_create_extent_cache(void); 3888 void f2fs_destroy_extent_cache(void); 3889 3890 /* 3891 * sysfs.c 3892 */ 3893 int __init f2fs_init_sysfs(void); 3894 void f2fs_exit_sysfs(void); 3895 int f2fs_register_sysfs(struct f2fs_sb_info *sbi); 3896 void f2fs_unregister_sysfs(struct f2fs_sb_info *sbi); 3897 3898 /* verity.c */ 3899 extern const struct fsverity_operations f2fs_verityops; 3900 3901 /* 3902 * crypto support 3903 */ 3904 static inline bool f2fs_encrypted_file(struct inode *inode) 3905 { 3906 return IS_ENCRYPTED(inode) && S_ISREG(inode->i_mode); 3907 } 3908 3909 static inline void f2fs_set_encrypted_inode(struct inode *inode) 3910 { 3911 #ifdef CONFIG_FS_ENCRYPTION 3912 file_set_encrypt(inode); 3913 f2fs_set_inode_flags(inode); 3914 #endif 3915 } 3916 3917 /* 3918 * Returns true if the reads of the inode's data need to undergo some 3919 * postprocessing step, like decryption or authenticity verification. 3920 */ 3921 static inline bool f2fs_post_read_required(struct inode *inode) 3922 { 3923 return f2fs_encrypted_file(inode) || fsverity_active(inode) || 3924 f2fs_compressed_file(inode); 3925 } 3926 3927 /* 3928 * compress.c 3929 */ 3930 #ifdef CONFIG_F2FS_FS_COMPRESSION 3931 bool f2fs_is_compressed_page(struct page *page); 3932 struct page *f2fs_compress_control_page(struct page *page); 3933 int f2fs_prepare_compress_overwrite(struct inode *inode, 3934 struct page **pagep, pgoff_t index, void **fsdata); 3935 bool f2fs_compress_write_end(struct inode *inode, void *fsdata, 3936 pgoff_t index, unsigned copied); 3937 int f2fs_truncate_partial_cluster(struct inode *inode, u64 from, bool lock); 3938 void f2fs_compress_write_end_io(struct bio *bio, struct page *page); 3939 bool f2fs_is_compress_backend_ready(struct inode *inode); 3940 int f2fs_init_compress_mempool(void); 3941 void f2fs_destroy_compress_mempool(void); 3942 void f2fs_end_read_compressed_page(struct page *page, bool failed); 3943 bool f2fs_cluster_is_empty(struct compress_ctx *cc); 3944 bool f2fs_cluster_can_merge_page(struct compress_ctx *cc, pgoff_t index); 3945 void f2fs_compress_ctx_add_page(struct compress_ctx *cc, struct page *page); 3946 int f2fs_write_multi_pages(struct compress_ctx *cc, 3947 int *submitted, 3948 struct writeback_control *wbc, 3949 enum iostat_type io_type); 3950 int f2fs_is_compressed_cluster(struct inode *inode, pgoff_t index); 3951 int f2fs_read_multi_pages(struct compress_ctx *cc, struct bio **bio_ret, 3952 unsigned nr_pages, sector_t *last_block_in_bio, 3953 bool is_readahead, bool for_write); 3954 struct decompress_io_ctx *f2fs_alloc_dic(struct compress_ctx *cc); 3955 void f2fs_decompress_end_io(struct decompress_io_ctx *dic, bool failed); 3956 void f2fs_put_page_dic(struct page *page); 3957 int f2fs_init_compress_ctx(struct compress_ctx *cc); 3958 void f2fs_destroy_compress_ctx(struct compress_ctx *cc); 3959 void f2fs_init_compress_info(struct f2fs_sb_info *sbi); 3960 int f2fs_init_page_array_cache(struct f2fs_sb_info *sbi); 3961 void f2fs_destroy_page_array_cache(struct f2fs_sb_info *sbi); 3962 int __init f2fs_init_compress_cache(void); 3963 void f2fs_destroy_compress_cache(void); 3964 #else 3965 static inline bool f2fs_is_compressed_page(struct page *page) { return false; } 3966 static inline bool f2fs_is_compress_backend_ready(struct inode *inode) 3967 { 3968 if (!f2fs_compressed_file(inode)) 3969 return true; 3970 /* not support compression */ 3971 return false; 3972 } 3973 static inline struct page *f2fs_compress_control_page(struct page *page) 3974 { 3975 WARN_ON_ONCE(1); 3976 return ERR_PTR(-EINVAL); 3977 } 3978 static inline int f2fs_init_compress_mempool(void) { return 0; } 3979 static inline void f2fs_destroy_compress_mempool(void) { } 3980 static inline void f2fs_end_read_compressed_page(struct page *page, bool failed) 3981 { 3982 WARN_ON_ONCE(1); 3983 } 3984 static inline void f2fs_put_page_dic(struct page *page) 3985 { 3986 WARN_ON_ONCE(1); 3987 } 3988 static inline int f2fs_init_page_array_cache(struct f2fs_sb_info *sbi) { return 0; } 3989 static inline void f2fs_destroy_page_array_cache(struct f2fs_sb_info *sbi) { } 3990 static inline int __init f2fs_init_compress_cache(void) { return 0; } 3991 static inline void f2fs_destroy_compress_cache(void) { } 3992 #endif 3993 3994 static inline void set_compress_context(struct inode *inode) 3995 { 3996 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 3997 3998 F2FS_I(inode)->i_compress_algorithm = 3999 F2FS_OPTION(sbi).compress_algorithm; 4000 F2FS_I(inode)->i_log_cluster_size = 4001 F2FS_OPTION(sbi).compress_log_size; 4002 F2FS_I(inode)->i_compress_flag = 4003 F2FS_OPTION(sbi).compress_chksum ? 4004 1 << COMPRESS_CHKSUM : 0; 4005 F2FS_I(inode)->i_cluster_size = 4006 1 << F2FS_I(inode)->i_log_cluster_size; 4007 if (F2FS_I(inode)->i_compress_algorithm == COMPRESS_LZ4 && 4008 F2FS_OPTION(sbi).compress_level) 4009 F2FS_I(inode)->i_compress_flag |= 4010 F2FS_OPTION(sbi).compress_level << 4011 COMPRESS_LEVEL_OFFSET; 4012 F2FS_I(inode)->i_flags |= F2FS_COMPR_FL; 4013 set_inode_flag(inode, FI_COMPRESSED_FILE); 4014 stat_inc_compr_inode(inode); 4015 f2fs_mark_inode_dirty_sync(inode, true); 4016 } 4017 4018 static inline bool f2fs_disable_compressed_file(struct inode *inode) 4019 { 4020 struct f2fs_inode_info *fi = F2FS_I(inode); 4021 4022 if (!f2fs_compressed_file(inode)) 4023 return true; 4024 if (S_ISREG(inode->i_mode) && 4025 (get_dirty_pages(inode) || atomic_read(&fi->i_compr_blocks))) 4026 return false; 4027 4028 fi->i_flags &= ~F2FS_COMPR_FL; 4029 stat_dec_compr_inode(inode); 4030 clear_inode_flag(inode, FI_COMPRESSED_FILE); 4031 f2fs_mark_inode_dirty_sync(inode, true); 4032 return true; 4033 } 4034 4035 #define F2FS_FEATURE_FUNCS(name, flagname) \ 4036 static inline int f2fs_sb_has_##name(struct f2fs_sb_info *sbi) \ 4037 { \ 4038 return F2FS_HAS_FEATURE(sbi, F2FS_FEATURE_##flagname); \ 4039 } 4040 4041 F2FS_FEATURE_FUNCS(encrypt, ENCRYPT); 4042 F2FS_FEATURE_FUNCS(blkzoned, BLKZONED); 4043 F2FS_FEATURE_FUNCS(extra_attr, EXTRA_ATTR); 4044 F2FS_FEATURE_FUNCS(project_quota, PRJQUOTA); 4045 F2FS_FEATURE_FUNCS(inode_chksum, INODE_CHKSUM); 4046 F2FS_FEATURE_FUNCS(flexible_inline_xattr, FLEXIBLE_INLINE_XATTR); 4047 F2FS_FEATURE_FUNCS(quota_ino, QUOTA_INO); 4048 F2FS_FEATURE_FUNCS(inode_crtime, INODE_CRTIME); 4049 F2FS_FEATURE_FUNCS(lost_found, LOST_FOUND); 4050 F2FS_FEATURE_FUNCS(verity, VERITY); 4051 F2FS_FEATURE_FUNCS(sb_chksum, SB_CHKSUM); 4052 F2FS_FEATURE_FUNCS(casefold, CASEFOLD); 4053 F2FS_FEATURE_FUNCS(compression, COMPRESSION); 4054 4055 #ifdef CONFIG_BLK_DEV_ZONED 4056 static inline bool f2fs_blkz_is_seq(struct f2fs_sb_info *sbi, int devi, 4057 block_t blkaddr) 4058 { 4059 unsigned int zno = blkaddr >> sbi->log_blocks_per_blkz; 4060 4061 return test_bit(zno, FDEV(devi).blkz_seq); 4062 } 4063 #endif 4064 4065 static inline bool f2fs_hw_should_discard(struct f2fs_sb_info *sbi) 4066 { 4067 return f2fs_sb_has_blkzoned(sbi); 4068 } 4069 4070 static inline bool f2fs_bdev_support_discard(struct block_device *bdev) 4071 { 4072 return blk_queue_discard(bdev_get_queue(bdev)) || 4073 bdev_is_zoned(bdev); 4074 } 4075 4076 static inline bool f2fs_hw_support_discard(struct f2fs_sb_info *sbi) 4077 { 4078 int i; 4079 4080 if (!f2fs_is_multi_device(sbi)) 4081 return f2fs_bdev_support_discard(sbi->sb->s_bdev); 4082 4083 for (i = 0; i < sbi->s_ndevs; i++) 4084 if (f2fs_bdev_support_discard(FDEV(i).bdev)) 4085 return true; 4086 return false; 4087 } 4088 4089 static inline bool f2fs_realtime_discard_enable(struct f2fs_sb_info *sbi) 4090 { 4091 return (test_opt(sbi, DISCARD) && f2fs_hw_support_discard(sbi)) || 4092 f2fs_hw_should_discard(sbi); 4093 } 4094 4095 static inline bool f2fs_hw_is_readonly(struct f2fs_sb_info *sbi) 4096 { 4097 int i; 4098 4099 if (!f2fs_is_multi_device(sbi)) 4100 return bdev_read_only(sbi->sb->s_bdev); 4101 4102 for (i = 0; i < sbi->s_ndevs; i++) 4103 if (bdev_read_only(FDEV(i).bdev)) 4104 return true; 4105 return false; 4106 } 4107 4108 static inline bool f2fs_lfs_mode(struct f2fs_sb_info *sbi) 4109 { 4110 return F2FS_OPTION(sbi).fs_mode == FS_MODE_LFS; 4111 } 4112 4113 static inline bool f2fs_may_compress(struct inode *inode) 4114 { 4115 if (IS_SWAPFILE(inode) || f2fs_is_pinned_file(inode) || 4116 f2fs_is_atomic_file(inode) || 4117 f2fs_is_volatile_file(inode)) 4118 return false; 4119 return S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode); 4120 } 4121 4122 static inline void f2fs_i_compr_blocks_update(struct inode *inode, 4123 u64 blocks, bool add) 4124 { 4125 int diff = F2FS_I(inode)->i_cluster_size - blocks; 4126 struct f2fs_inode_info *fi = F2FS_I(inode); 4127 4128 /* don't update i_compr_blocks if saved blocks were released */ 4129 if (!add && !atomic_read(&fi->i_compr_blocks)) 4130 return; 4131 4132 if (add) { 4133 atomic_add(diff, &fi->i_compr_blocks); 4134 stat_add_compr_blocks(inode, diff); 4135 } else { 4136 atomic_sub(diff, &fi->i_compr_blocks); 4137 stat_sub_compr_blocks(inode, diff); 4138 } 4139 f2fs_mark_inode_dirty_sync(inode, true); 4140 } 4141 4142 static inline int block_unaligned_IO(struct inode *inode, 4143 struct kiocb *iocb, struct iov_iter *iter) 4144 { 4145 unsigned int i_blkbits = READ_ONCE(inode->i_blkbits); 4146 unsigned int blocksize_mask = (1 << i_blkbits) - 1; 4147 loff_t offset = iocb->ki_pos; 4148 unsigned long align = offset | iov_iter_alignment(iter); 4149 4150 return align & blocksize_mask; 4151 } 4152 4153 static inline int allow_outplace_dio(struct inode *inode, 4154 struct kiocb *iocb, struct iov_iter *iter) 4155 { 4156 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 4157 int rw = iov_iter_rw(iter); 4158 4159 return (f2fs_lfs_mode(sbi) && (rw == WRITE) && 4160 !block_unaligned_IO(inode, iocb, iter)); 4161 } 4162 4163 static inline bool f2fs_force_buffered_io(struct inode *inode, 4164 struct kiocb *iocb, struct iov_iter *iter) 4165 { 4166 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 4167 int rw = iov_iter_rw(iter); 4168 4169 if (f2fs_post_read_required(inode)) 4170 return true; 4171 if (f2fs_is_multi_device(sbi)) 4172 return true; 4173 /* 4174 * for blkzoned device, fallback direct IO to buffered IO, so 4175 * all IOs can be serialized by log-structured write. 4176 */ 4177 if (f2fs_sb_has_blkzoned(sbi)) 4178 return true; 4179 if (f2fs_lfs_mode(sbi) && (rw == WRITE)) { 4180 if (block_unaligned_IO(inode, iocb, iter)) 4181 return true; 4182 if (F2FS_IO_ALIGNED(sbi)) 4183 return true; 4184 } 4185 if (is_sbi_flag_set(F2FS_I_SB(inode), SBI_CP_DISABLED)) 4186 return true; 4187 4188 return false; 4189 } 4190 4191 static inline bool f2fs_need_verity(const struct inode *inode, pgoff_t idx) 4192 { 4193 return fsverity_active(inode) && 4194 idx < DIV_ROUND_UP(inode->i_size, PAGE_SIZE); 4195 } 4196 4197 #ifdef CONFIG_F2FS_FAULT_INJECTION 4198 extern void f2fs_build_fault_attr(struct f2fs_sb_info *sbi, unsigned int rate, 4199 unsigned int type); 4200 #else 4201 #define f2fs_build_fault_attr(sbi, rate, type) do { } while (0) 4202 #endif 4203 4204 static inline bool is_journalled_quota(struct f2fs_sb_info *sbi) 4205 { 4206 #ifdef CONFIG_QUOTA 4207 if (f2fs_sb_has_quota_ino(sbi)) 4208 return true; 4209 if (F2FS_OPTION(sbi).s_qf_names[USRQUOTA] || 4210 F2FS_OPTION(sbi).s_qf_names[GRPQUOTA] || 4211 F2FS_OPTION(sbi).s_qf_names[PRJQUOTA]) 4212 return true; 4213 #endif 4214 return false; 4215 } 4216 4217 #define EFSBADCRC EBADMSG /* Bad CRC detected */ 4218 #define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */ 4219 4220 #endif /* _LINUX_F2FS_H */ 4221