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 (unlikely(condition)) { \ 37 WARN_ON(1); \ 38 set_sbi_flag(sbi, SBI_NEED_FSCK); \ 39 } \ 40 } while (0) 41 #endif 42 43 enum { 44 FAULT_KMALLOC, 45 FAULT_KVMALLOC, 46 FAULT_PAGE_ALLOC, 47 FAULT_PAGE_GET, 48 FAULT_ALLOC_BIO, 49 FAULT_ALLOC_NID, 50 FAULT_ORPHAN, 51 FAULT_BLOCK, 52 FAULT_DIR_DEPTH, 53 FAULT_EVICT_INODE, 54 FAULT_TRUNCATE, 55 FAULT_READ_IO, 56 FAULT_CHECKPOINT, 57 FAULT_DISCARD, 58 FAULT_WRITE_IO, 59 FAULT_MAX, 60 }; 61 62 #ifdef CONFIG_F2FS_FAULT_INJECTION 63 #define F2FS_ALL_FAULT_TYPE ((1 << FAULT_MAX) - 1) 64 65 struct f2fs_fault_info { 66 atomic_t inject_ops; 67 unsigned int inject_rate; 68 unsigned int inject_type; 69 }; 70 71 extern const char *f2fs_fault_name[FAULT_MAX]; 72 #define IS_FAULT_SET(fi, type) ((fi)->inject_type & (1 << (type))) 73 #endif 74 75 /* 76 * For mount options 77 */ 78 #define F2FS_MOUNT_DISABLE_ROLL_FORWARD 0x00000002 79 #define F2FS_MOUNT_DISCARD 0x00000004 80 #define F2FS_MOUNT_NOHEAP 0x00000008 81 #define F2FS_MOUNT_XATTR_USER 0x00000010 82 #define F2FS_MOUNT_POSIX_ACL 0x00000020 83 #define F2FS_MOUNT_DISABLE_EXT_IDENTIFY 0x00000040 84 #define F2FS_MOUNT_INLINE_XATTR 0x00000080 85 #define F2FS_MOUNT_INLINE_DATA 0x00000100 86 #define F2FS_MOUNT_INLINE_DENTRY 0x00000200 87 #define F2FS_MOUNT_FLUSH_MERGE 0x00000400 88 #define F2FS_MOUNT_NOBARRIER 0x00000800 89 #define F2FS_MOUNT_FASTBOOT 0x00001000 90 #define F2FS_MOUNT_EXTENT_CACHE 0x00002000 91 #define F2FS_MOUNT_DATA_FLUSH 0x00008000 92 #define F2FS_MOUNT_FAULT_INJECTION 0x00010000 93 #define F2FS_MOUNT_USRQUOTA 0x00080000 94 #define F2FS_MOUNT_GRPQUOTA 0x00100000 95 #define F2FS_MOUNT_PRJQUOTA 0x00200000 96 #define F2FS_MOUNT_QUOTA 0x00400000 97 #define F2FS_MOUNT_INLINE_XATTR_SIZE 0x00800000 98 #define F2FS_MOUNT_RESERVE_ROOT 0x01000000 99 #define F2FS_MOUNT_DISABLE_CHECKPOINT 0x02000000 100 #define F2FS_MOUNT_NORECOVERY 0x04000000 101 #define F2FS_MOUNT_ATGC 0x08000000 102 103 #define F2FS_OPTION(sbi) ((sbi)->mount_opt) 104 #define clear_opt(sbi, option) (F2FS_OPTION(sbi).opt &= ~F2FS_MOUNT_##option) 105 #define set_opt(sbi, option) (F2FS_OPTION(sbi).opt |= F2FS_MOUNT_##option) 106 #define test_opt(sbi, option) (F2FS_OPTION(sbi).opt & F2FS_MOUNT_##option) 107 108 #define ver_after(a, b) (typecheck(unsigned long long, a) && \ 109 typecheck(unsigned long long, b) && \ 110 ((long long)((a) - (b)) > 0)) 111 112 typedef u32 block_t; /* 113 * should not change u32, since it is the on-disk block 114 * address format, __le32. 115 */ 116 typedef u32 nid_t; 117 118 #define COMPRESS_EXT_NUM 16 119 120 struct f2fs_mount_info { 121 unsigned int opt; 122 int write_io_size_bits; /* Write IO size bits */ 123 block_t root_reserved_blocks; /* root reserved blocks */ 124 kuid_t s_resuid; /* reserved blocks for uid */ 125 kgid_t s_resgid; /* reserved blocks for gid */ 126 int active_logs; /* # of active logs */ 127 int inline_xattr_size; /* inline xattr size */ 128 #ifdef CONFIG_F2FS_FAULT_INJECTION 129 struct f2fs_fault_info fault_info; /* For fault injection */ 130 #endif 131 #ifdef CONFIG_QUOTA 132 /* Names of quota files with journalled quota */ 133 char *s_qf_names[MAXQUOTAS]; 134 int s_jquota_fmt; /* Format of quota to use */ 135 #endif 136 /* For which write hints are passed down to block layer */ 137 int whint_mode; 138 int alloc_mode; /* segment allocation policy */ 139 int fsync_mode; /* fsync policy */ 140 int fs_mode; /* fs mode: LFS or ADAPTIVE */ 141 int bggc_mode; /* bggc mode: off, on or sync */ 142 struct fscrypt_dummy_policy dummy_enc_policy; /* test dummy encryption */ 143 block_t unusable_cap_perc; /* percentage for cap */ 144 block_t unusable_cap; /* Amount of space allowed to be 145 * unusable when disabling checkpoint 146 */ 147 148 /* For compression */ 149 unsigned char compress_algorithm; /* algorithm type */ 150 unsigned char compress_log_size; /* cluster log size */ 151 bool compress_chksum; /* compressed data chksum */ 152 unsigned char compress_ext_cnt; /* extension count */ 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 /* for the bitmap indicate blocks to be discarded */ 271 struct discard_entry { 272 struct list_head list; /* list head */ 273 block_t start_blkaddr; /* start blockaddr of current segment */ 274 unsigned char discard_map[SIT_VBLOCK_MAP_SIZE]; /* segment discard bitmap */ 275 }; 276 277 /* default discard granularity of inner discard thread, unit: block count */ 278 #define DEFAULT_DISCARD_GRANULARITY 16 279 280 /* max discard pend list number */ 281 #define MAX_PLIST_NUM 512 282 #define plist_idx(blk_num) ((blk_num) >= MAX_PLIST_NUM ? \ 283 (MAX_PLIST_NUM - 1) : ((blk_num) - 1)) 284 285 enum { 286 D_PREP, /* initial */ 287 D_PARTIAL, /* partially submitted */ 288 D_SUBMIT, /* all submitted */ 289 D_DONE, /* finished */ 290 }; 291 292 struct discard_info { 293 block_t lstart; /* logical start address */ 294 block_t len; /* length */ 295 block_t start; /* actual start address in dev */ 296 }; 297 298 struct discard_cmd { 299 struct rb_node rb_node; /* rb node located in rb-tree */ 300 union { 301 struct { 302 block_t lstart; /* logical start address */ 303 block_t len; /* length */ 304 block_t start; /* actual start address in dev */ 305 }; 306 struct discard_info di; /* discard info */ 307 308 }; 309 struct list_head list; /* command list */ 310 struct completion wait; /* compleation */ 311 struct block_device *bdev; /* bdev */ 312 unsigned short ref; /* reference count */ 313 unsigned char state; /* state */ 314 unsigned char queued; /* queued discard */ 315 int error; /* bio error */ 316 spinlock_t lock; /* for state/bio_ref updating */ 317 unsigned short bio_ref; /* bio reference count */ 318 }; 319 320 enum { 321 DPOLICY_BG, 322 DPOLICY_FORCE, 323 DPOLICY_FSTRIM, 324 DPOLICY_UMOUNT, 325 MAX_DPOLICY, 326 }; 327 328 struct discard_policy { 329 int type; /* type of discard */ 330 unsigned int min_interval; /* used for candidates exist */ 331 unsigned int mid_interval; /* used for device busy */ 332 unsigned int max_interval; /* used for candidates not exist */ 333 unsigned int max_requests; /* # of discards issued per round */ 334 unsigned int io_aware_gran; /* minimum granularity discard not be aware of I/O */ 335 bool io_aware; /* issue discard in idle time */ 336 bool sync; /* submit discard with REQ_SYNC flag */ 337 bool ordered; /* issue discard by lba order */ 338 bool timeout; /* discard timeout for put_super */ 339 unsigned int granularity; /* discard granularity */ 340 }; 341 342 struct discard_cmd_control { 343 struct task_struct *f2fs_issue_discard; /* discard thread */ 344 struct list_head entry_list; /* 4KB discard entry list */ 345 struct list_head pend_list[MAX_PLIST_NUM];/* store pending entries */ 346 struct list_head wait_list; /* store on-flushing entries */ 347 struct list_head fstrim_list; /* in-flight discard from fstrim */ 348 wait_queue_head_t discard_wait_queue; /* waiting queue for wake-up */ 349 unsigned int discard_wake; /* to wake up discard thread */ 350 struct mutex cmd_lock; 351 unsigned int nr_discards; /* # of discards in the list */ 352 unsigned int max_discards; /* max. discards to be issued */ 353 unsigned int discard_granularity; /* discard granularity */ 354 unsigned int undiscard_blks; /* # of undiscard blocks */ 355 unsigned int next_pos; /* next discard position */ 356 atomic_t issued_discard; /* # of issued discard */ 357 atomic_t queued_discard; /* # of queued discard */ 358 atomic_t discard_cmd_cnt; /* # of cached cmd count */ 359 struct rb_root_cached root; /* root of discard rb-tree */ 360 bool rbtree_check; /* config for consistence check */ 361 }; 362 363 /* for the list of fsync inodes, used only during recovery */ 364 struct fsync_inode_entry { 365 struct list_head list; /* list head */ 366 struct inode *inode; /* vfs inode pointer */ 367 block_t blkaddr; /* block address locating the last fsync */ 368 block_t last_dentry; /* block address locating the last dentry */ 369 }; 370 371 #define nats_in_cursum(jnl) (le16_to_cpu((jnl)->n_nats)) 372 #define sits_in_cursum(jnl) (le16_to_cpu((jnl)->n_sits)) 373 374 #define nat_in_journal(jnl, i) ((jnl)->nat_j.entries[i].ne) 375 #define nid_in_journal(jnl, i) ((jnl)->nat_j.entries[i].nid) 376 #define sit_in_journal(jnl, i) ((jnl)->sit_j.entries[i].se) 377 #define segno_in_journal(jnl, i) ((jnl)->sit_j.entries[i].segno) 378 379 #define MAX_NAT_JENTRIES(jnl) (NAT_JOURNAL_ENTRIES - nats_in_cursum(jnl)) 380 #define MAX_SIT_JENTRIES(jnl) (SIT_JOURNAL_ENTRIES - sits_in_cursum(jnl)) 381 382 static inline int update_nats_in_cursum(struct f2fs_journal *journal, int i) 383 { 384 int before = nats_in_cursum(journal); 385 386 journal->n_nats = cpu_to_le16(before + i); 387 return before; 388 } 389 390 static inline int update_sits_in_cursum(struct f2fs_journal *journal, int i) 391 { 392 int before = sits_in_cursum(journal); 393 394 journal->n_sits = cpu_to_le16(before + i); 395 return before; 396 } 397 398 static inline bool __has_cursum_space(struct f2fs_journal *journal, 399 int size, int type) 400 { 401 if (type == NAT_JOURNAL) 402 return size <= MAX_NAT_JENTRIES(journal); 403 return size <= MAX_SIT_JENTRIES(journal); 404 } 405 406 /* for inline stuff */ 407 #define DEF_INLINE_RESERVED_SIZE 1 408 static inline int get_extra_isize(struct inode *inode); 409 static inline int get_inline_xattr_addrs(struct inode *inode); 410 #define MAX_INLINE_DATA(inode) (sizeof(__le32) * \ 411 (CUR_ADDRS_PER_INODE(inode) - \ 412 get_inline_xattr_addrs(inode) - \ 413 DEF_INLINE_RESERVED_SIZE)) 414 415 /* for inline dir */ 416 #define NR_INLINE_DENTRY(inode) (MAX_INLINE_DATA(inode) * BITS_PER_BYTE / \ 417 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \ 418 BITS_PER_BYTE + 1)) 419 #define INLINE_DENTRY_BITMAP_SIZE(inode) \ 420 DIV_ROUND_UP(NR_INLINE_DENTRY(inode), BITS_PER_BYTE) 421 #define INLINE_RESERVED_SIZE(inode) (MAX_INLINE_DATA(inode) - \ 422 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \ 423 NR_INLINE_DENTRY(inode) + \ 424 INLINE_DENTRY_BITMAP_SIZE(inode))) 425 426 /* 427 * For INODE and NODE manager 428 */ 429 /* for directory operations */ 430 431 struct f2fs_filename { 432 /* 433 * The filename the user specified. This is NULL for some 434 * filesystem-internal operations, e.g. converting an inline directory 435 * to a non-inline one, or roll-forward recovering an encrypted dentry. 436 */ 437 const struct qstr *usr_fname; 438 439 /* 440 * The on-disk filename. For encrypted directories, this is encrypted. 441 * This may be NULL for lookups in an encrypted dir without the key. 442 */ 443 struct fscrypt_str disk_name; 444 445 /* The dirhash of this filename */ 446 f2fs_hash_t hash; 447 448 #ifdef CONFIG_FS_ENCRYPTION 449 /* 450 * For lookups in encrypted directories: either the buffer backing 451 * disk_name, or a buffer that holds the decoded no-key name. 452 */ 453 struct fscrypt_str crypto_buf; 454 #endif 455 #ifdef CONFIG_UNICODE 456 /* 457 * For casefolded directories: the casefolded name, but it's left NULL 458 * if the original name is not valid Unicode, if the directory is both 459 * casefolded and encrypted and its encryption key is unavailable, or if 460 * the filesystem is doing an internal operation where usr_fname is also 461 * NULL. In all these cases we fall back to treating the name as an 462 * opaque byte sequence. 463 */ 464 struct fscrypt_str cf_name; 465 #endif 466 }; 467 468 struct f2fs_dentry_ptr { 469 struct inode *inode; 470 void *bitmap; 471 struct f2fs_dir_entry *dentry; 472 __u8 (*filename)[F2FS_SLOT_LEN]; 473 int max; 474 int nr_bitmap; 475 }; 476 477 static inline void make_dentry_ptr_block(struct inode *inode, 478 struct f2fs_dentry_ptr *d, struct f2fs_dentry_block *t) 479 { 480 d->inode = inode; 481 d->max = NR_DENTRY_IN_BLOCK; 482 d->nr_bitmap = SIZE_OF_DENTRY_BITMAP; 483 d->bitmap = t->dentry_bitmap; 484 d->dentry = t->dentry; 485 d->filename = t->filename; 486 } 487 488 static inline void make_dentry_ptr_inline(struct inode *inode, 489 struct f2fs_dentry_ptr *d, void *t) 490 { 491 int entry_cnt = NR_INLINE_DENTRY(inode); 492 int bitmap_size = INLINE_DENTRY_BITMAP_SIZE(inode); 493 int reserved_size = INLINE_RESERVED_SIZE(inode); 494 495 d->inode = inode; 496 d->max = entry_cnt; 497 d->nr_bitmap = bitmap_size; 498 d->bitmap = t; 499 d->dentry = t + bitmap_size + reserved_size; 500 d->filename = t + bitmap_size + reserved_size + 501 SIZE_OF_DIR_ENTRY * entry_cnt; 502 } 503 504 /* 505 * XATTR_NODE_OFFSET stores xattrs to one node block per file keeping -1 506 * as its node offset to distinguish from index node blocks. 507 * But some bits are used to mark the node block. 508 */ 509 #define XATTR_NODE_OFFSET ((((unsigned int)-1) << OFFSET_BIT_SHIFT) \ 510 >> OFFSET_BIT_SHIFT) 511 enum { 512 ALLOC_NODE, /* allocate a new node page if needed */ 513 LOOKUP_NODE, /* look up a node without readahead */ 514 LOOKUP_NODE_RA, /* 515 * look up a node with readahead called 516 * by get_data_block. 517 */ 518 }; 519 520 #define DEFAULT_RETRY_IO_COUNT 8 /* maximum retry read IO count */ 521 522 /* congestion wait timeout value, default: 20ms */ 523 #define DEFAULT_IO_TIMEOUT (msecs_to_jiffies(20)) 524 525 /* maximum retry quota flush count */ 526 #define DEFAULT_RETRY_QUOTA_FLUSH_COUNT 8 527 528 #define F2FS_LINK_MAX 0xffffffff /* maximum link count per file */ 529 530 #define MAX_DIR_RA_PAGES 4 /* maximum ra pages of dir */ 531 532 /* for in-memory extent cache entry */ 533 #define F2FS_MIN_EXTENT_LEN 64 /* minimum extent length */ 534 535 /* number of extent info in extent cache we try to shrink */ 536 #define EXTENT_CACHE_SHRINK_NUMBER 128 537 538 struct rb_entry { 539 struct rb_node rb_node; /* rb node located in rb-tree */ 540 union { 541 struct { 542 unsigned int ofs; /* start offset of the entry */ 543 unsigned int len; /* length of the entry */ 544 }; 545 unsigned long long key; /* 64-bits key */ 546 } __packed; 547 }; 548 549 struct extent_info { 550 unsigned int fofs; /* start offset in a file */ 551 unsigned int len; /* length of the extent */ 552 u32 blk; /* start block address of the extent */ 553 }; 554 555 struct extent_node { 556 struct rb_node rb_node; /* rb node located in rb-tree */ 557 struct extent_info ei; /* extent info */ 558 struct list_head list; /* node in global extent list of sbi */ 559 struct extent_tree *et; /* extent tree pointer */ 560 }; 561 562 struct extent_tree { 563 nid_t ino; /* inode number */ 564 struct rb_root_cached root; /* root of extent info rb-tree */ 565 struct extent_node *cached_en; /* recently accessed extent node */ 566 struct extent_info largest; /* largested extent info */ 567 struct list_head list; /* to be used by sbi->zombie_list */ 568 rwlock_t lock; /* protect extent info rb-tree */ 569 atomic_t node_cnt; /* # of extent node in rb-tree*/ 570 bool largest_updated; /* largest extent updated */ 571 }; 572 573 /* 574 * This structure is taken from ext4_map_blocks. 575 * 576 * Note that, however, f2fs uses NEW and MAPPED flags for f2fs_map_blocks(). 577 */ 578 #define F2FS_MAP_NEW (1 << BH_New) 579 #define F2FS_MAP_MAPPED (1 << BH_Mapped) 580 #define F2FS_MAP_UNWRITTEN (1 << BH_Unwritten) 581 #define F2FS_MAP_FLAGS (F2FS_MAP_NEW | F2FS_MAP_MAPPED |\ 582 F2FS_MAP_UNWRITTEN) 583 584 struct f2fs_map_blocks { 585 block_t m_pblk; 586 block_t m_lblk; 587 unsigned int m_len; 588 unsigned int m_flags; 589 pgoff_t *m_next_pgofs; /* point next possible non-hole pgofs */ 590 pgoff_t *m_next_extent; /* point to next possible extent */ 591 int m_seg_type; 592 bool m_may_create; /* indicate it is from write path */ 593 }; 594 595 /* for flag in get_data_block */ 596 enum { 597 F2FS_GET_BLOCK_DEFAULT, 598 F2FS_GET_BLOCK_FIEMAP, 599 F2FS_GET_BLOCK_BMAP, 600 F2FS_GET_BLOCK_DIO, 601 F2FS_GET_BLOCK_PRE_DIO, 602 F2FS_GET_BLOCK_PRE_AIO, 603 F2FS_GET_BLOCK_PRECACHE, 604 }; 605 606 /* 607 * i_advise uses FADVISE_XXX_BIT. We can add additional hints later. 608 */ 609 #define FADVISE_COLD_BIT 0x01 610 #define FADVISE_LOST_PINO_BIT 0x02 611 #define FADVISE_ENCRYPT_BIT 0x04 612 #define FADVISE_ENC_NAME_BIT 0x08 613 #define FADVISE_KEEP_SIZE_BIT 0x10 614 #define FADVISE_HOT_BIT 0x20 615 #define FADVISE_VERITY_BIT 0x40 616 617 #define FADVISE_MODIFIABLE_BITS (FADVISE_COLD_BIT | FADVISE_HOT_BIT) 618 619 #define file_is_cold(inode) is_file(inode, FADVISE_COLD_BIT) 620 #define file_wrong_pino(inode) is_file(inode, FADVISE_LOST_PINO_BIT) 621 #define file_set_cold(inode) set_file(inode, FADVISE_COLD_BIT) 622 #define file_lost_pino(inode) set_file(inode, FADVISE_LOST_PINO_BIT) 623 #define file_clear_cold(inode) clear_file(inode, FADVISE_COLD_BIT) 624 #define file_got_pino(inode) clear_file(inode, FADVISE_LOST_PINO_BIT) 625 #define file_is_encrypt(inode) is_file(inode, FADVISE_ENCRYPT_BIT) 626 #define file_set_encrypt(inode) set_file(inode, FADVISE_ENCRYPT_BIT) 627 #define file_clear_encrypt(inode) clear_file(inode, FADVISE_ENCRYPT_BIT) 628 #define file_enc_name(inode) is_file(inode, FADVISE_ENC_NAME_BIT) 629 #define file_set_enc_name(inode) set_file(inode, FADVISE_ENC_NAME_BIT) 630 #define file_keep_isize(inode) is_file(inode, FADVISE_KEEP_SIZE_BIT) 631 #define file_set_keep_isize(inode) set_file(inode, FADVISE_KEEP_SIZE_BIT) 632 #define file_is_hot(inode) is_file(inode, FADVISE_HOT_BIT) 633 #define file_set_hot(inode) set_file(inode, FADVISE_HOT_BIT) 634 #define file_clear_hot(inode) clear_file(inode, FADVISE_HOT_BIT) 635 #define file_is_verity(inode) is_file(inode, FADVISE_VERITY_BIT) 636 #define file_set_verity(inode) set_file(inode, FADVISE_VERITY_BIT) 637 638 #define DEF_DIR_LEVEL 0 639 640 enum { 641 GC_FAILURE_PIN, 642 GC_FAILURE_ATOMIC, 643 MAX_GC_FAILURE 644 }; 645 646 /* used for f2fs_inode_info->flags */ 647 enum { 648 FI_NEW_INODE, /* indicate newly allocated inode */ 649 FI_DIRTY_INODE, /* indicate inode is dirty or not */ 650 FI_AUTO_RECOVER, /* indicate inode is recoverable */ 651 FI_DIRTY_DIR, /* indicate directory has dirty pages */ 652 FI_INC_LINK, /* need to increment i_nlink */ 653 FI_ACL_MODE, /* indicate acl mode */ 654 FI_NO_ALLOC, /* should not allocate any blocks */ 655 FI_FREE_NID, /* free allocated nide */ 656 FI_NO_EXTENT, /* not to use the extent cache */ 657 FI_INLINE_XATTR, /* used for inline xattr */ 658 FI_INLINE_DATA, /* used for inline data*/ 659 FI_INLINE_DENTRY, /* used for inline dentry */ 660 FI_APPEND_WRITE, /* inode has appended data */ 661 FI_UPDATE_WRITE, /* inode has in-place-update data */ 662 FI_NEED_IPU, /* used for ipu per file */ 663 FI_ATOMIC_FILE, /* indicate atomic file */ 664 FI_ATOMIC_COMMIT, /* indicate the state of atomical committing */ 665 FI_VOLATILE_FILE, /* indicate volatile file */ 666 FI_FIRST_BLOCK_WRITTEN, /* indicate #0 data block was written */ 667 FI_DROP_CACHE, /* drop dirty page cache */ 668 FI_DATA_EXIST, /* indicate data exists */ 669 FI_INLINE_DOTS, /* indicate inline dot dentries */ 670 FI_DO_DEFRAG, /* indicate defragment is running */ 671 FI_DIRTY_FILE, /* indicate regular/symlink has dirty pages */ 672 FI_NO_PREALLOC, /* indicate skipped preallocated blocks */ 673 FI_HOT_DATA, /* indicate file is hot */ 674 FI_EXTRA_ATTR, /* indicate file has extra attribute */ 675 FI_PROJ_INHERIT, /* indicate file inherits projectid */ 676 FI_PIN_FILE, /* indicate file should not be gced */ 677 FI_ATOMIC_REVOKE_REQUEST, /* request to drop atomic data */ 678 FI_VERITY_IN_PROGRESS, /* building fs-verity Merkle tree */ 679 FI_COMPRESSED_FILE, /* indicate file's data can be compressed */ 680 FI_COMPRESS_CORRUPT, /* indicate compressed cluster is corrupted */ 681 FI_MMAP_FILE, /* indicate file was mmapped */ 682 FI_MAX, /* max flag, never be used */ 683 }; 684 685 struct f2fs_inode_info { 686 struct inode vfs_inode; /* serve a vfs inode */ 687 unsigned long i_flags; /* keep an inode flags for ioctl */ 688 unsigned char i_advise; /* use to give file attribute hints */ 689 unsigned char i_dir_level; /* use for dentry level for large dir */ 690 unsigned int i_current_depth; /* only for directory depth */ 691 /* for gc failure statistic */ 692 unsigned int i_gc_failures[MAX_GC_FAILURE]; 693 unsigned int i_pino; /* parent inode number */ 694 umode_t i_acl_mode; /* keep file acl mode temporarily */ 695 696 /* Use below internally in f2fs*/ 697 unsigned long flags[BITS_TO_LONGS(FI_MAX)]; /* use to pass per-file flags */ 698 struct rw_semaphore i_sem; /* protect fi info */ 699 atomic_t dirty_pages; /* # of dirty pages */ 700 f2fs_hash_t chash; /* hash value of given file name */ 701 unsigned int clevel; /* maximum level of given file name */ 702 struct task_struct *task; /* lookup and create consistency */ 703 struct task_struct *cp_task; /* separate cp/wb IO stats*/ 704 nid_t i_xattr_nid; /* node id that contains xattrs */ 705 loff_t last_disk_size; /* lastly written file size */ 706 spinlock_t i_size_lock; /* protect last_disk_size */ 707 708 #ifdef CONFIG_QUOTA 709 struct dquot *i_dquot[MAXQUOTAS]; 710 711 /* quota space reservation, managed internally by quota code */ 712 qsize_t i_reserved_quota; 713 #endif 714 struct list_head dirty_list; /* dirty list for dirs and files */ 715 struct list_head gdirty_list; /* linked in global dirty list */ 716 struct list_head inmem_ilist; /* list for inmem inodes */ 717 struct list_head inmem_pages; /* inmemory pages managed by f2fs */ 718 struct task_struct *inmem_task; /* store inmemory task */ 719 struct mutex inmem_lock; /* lock for inmemory pages */ 720 pgoff_t ra_offset; /* ongoing readahead offset */ 721 struct extent_tree *extent_tree; /* cached extent_tree entry */ 722 723 /* avoid racing between foreground op and gc */ 724 struct rw_semaphore i_gc_rwsem[2]; 725 struct rw_semaphore i_mmap_sem; 726 struct rw_semaphore i_xattr_sem; /* avoid racing between reading and changing EAs */ 727 728 int i_extra_isize; /* size of extra space located in i_addr */ 729 kprojid_t i_projid; /* id for project quota */ 730 int i_inline_xattr_size; /* inline xattr size */ 731 struct timespec64 i_crtime; /* inode creation time */ 732 struct timespec64 i_disk_time[4];/* inode disk times */ 733 734 /* for file compress */ 735 atomic_t i_compr_blocks; /* # of compressed blocks */ 736 unsigned char i_compress_algorithm; /* algorithm type */ 737 unsigned char i_log_cluster_size; /* log of cluster size */ 738 unsigned short i_compress_flag; /* compress flag */ 739 unsigned int i_cluster_size; /* cluster size */ 740 }; 741 742 static inline void get_extent_info(struct extent_info *ext, 743 struct f2fs_extent *i_ext) 744 { 745 ext->fofs = le32_to_cpu(i_ext->fofs); 746 ext->blk = le32_to_cpu(i_ext->blk); 747 ext->len = le32_to_cpu(i_ext->len); 748 } 749 750 static inline void set_raw_extent(struct extent_info *ext, 751 struct f2fs_extent *i_ext) 752 { 753 i_ext->fofs = cpu_to_le32(ext->fofs); 754 i_ext->blk = cpu_to_le32(ext->blk); 755 i_ext->len = cpu_to_le32(ext->len); 756 } 757 758 static inline void set_extent_info(struct extent_info *ei, unsigned int fofs, 759 u32 blk, unsigned int len) 760 { 761 ei->fofs = fofs; 762 ei->blk = blk; 763 ei->len = len; 764 } 765 766 static inline bool __is_discard_mergeable(struct discard_info *back, 767 struct discard_info *front, unsigned int max_len) 768 { 769 return (back->lstart + back->len == front->lstart) && 770 (back->len + front->len <= max_len); 771 } 772 773 static inline bool __is_discard_back_mergeable(struct discard_info *cur, 774 struct discard_info *back, unsigned int max_len) 775 { 776 return __is_discard_mergeable(back, cur, max_len); 777 } 778 779 static inline bool __is_discard_front_mergeable(struct discard_info *cur, 780 struct discard_info *front, unsigned int max_len) 781 { 782 return __is_discard_mergeable(cur, front, max_len); 783 } 784 785 static inline bool __is_extent_mergeable(struct extent_info *back, 786 struct extent_info *front) 787 { 788 return (back->fofs + back->len == front->fofs && 789 back->blk + back->len == front->blk); 790 } 791 792 static inline bool __is_back_mergeable(struct extent_info *cur, 793 struct extent_info *back) 794 { 795 return __is_extent_mergeable(back, cur); 796 } 797 798 static inline bool __is_front_mergeable(struct extent_info *cur, 799 struct extent_info *front) 800 { 801 return __is_extent_mergeable(cur, front); 802 } 803 804 extern void f2fs_mark_inode_dirty_sync(struct inode *inode, bool sync); 805 static inline void __try_update_largest_extent(struct extent_tree *et, 806 struct extent_node *en) 807 { 808 if (en->ei.len > et->largest.len) { 809 et->largest = en->ei; 810 et->largest_updated = true; 811 } 812 } 813 814 /* 815 * For free nid management 816 */ 817 enum nid_state { 818 FREE_NID, /* newly added to free nid list */ 819 PREALLOC_NID, /* it is preallocated */ 820 MAX_NID_STATE, 821 }; 822 823 struct f2fs_nm_info { 824 block_t nat_blkaddr; /* base disk address of NAT */ 825 nid_t max_nid; /* maximum possible node ids */ 826 nid_t available_nids; /* # of available node ids */ 827 nid_t next_scan_nid; /* the next nid to be scanned */ 828 unsigned int ram_thresh; /* control the memory footprint */ 829 unsigned int ra_nid_pages; /* # of nid pages to be readaheaded */ 830 unsigned int dirty_nats_ratio; /* control dirty nats ratio threshold */ 831 832 /* NAT cache management */ 833 struct radix_tree_root nat_root;/* root of the nat entry cache */ 834 struct radix_tree_root nat_set_root;/* root of the nat set cache */ 835 struct rw_semaphore nat_tree_lock; /* protect nat_tree_lock */ 836 struct list_head nat_entries; /* cached nat entry list (clean) */ 837 spinlock_t nat_list_lock; /* protect clean nat entry list */ 838 unsigned int nat_cnt; /* the # of cached nat entries */ 839 unsigned int dirty_nat_cnt; /* total num of nat entries in set */ 840 unsigned int nat_blocks; /* # of nat blocks */ 841 842 /* free node ids management */ 843 struct radix_tree_root free_nid_root;/* root of the free_nid cache */ 844 struct list_head free_nid_list; /* list for free nids excluding preallocated nids */ 845 unsigned int nid_cnt[MAX_NID_STATE]; /* the number of free node id */ 846 spinlock_t nid_list_lock; /* protect nid lists ops */ 847 struct mutex build_lock; /* lock for build free nids */ 848 unsigned char **free_nid_bitmap; 849 unsigned char *nat_block_bitmap; 850 unsigned short *free_nid_count; /* free nid count of NAT block */ 851 852 /* for checkpoint */ 853 char *nat_bitmap; /* NAT bitmap pointer */ 854 855 unsigned int nat_bits_blocks; /* # of nat bits blocks */ 856 unsigned char *nat_bits; /* NAT bits blocks */ 857 unsigned char *full_nat_bits; /* full NAT pages */ 858 unsigned char *empty_nat_bits; /* empty NAT pages */ 859 #ifdef CONFIG_F2FS_CHECK_FS 860 char *nat_bitmap_mir; /* NAT bitmap mirror */ 861 #endif 862 int bitmap_size; /* bitmap size */ 863 }; 864 865 /* 866 * this structure is used as one of function parameters. 867 * all the information are dedicated to a given direct node block determined 868 * by the data offset in a file. 869 */ 870 struct dnode_of_data { 871 struct inode *inode; /* vfs inode pointer */ 872 struct page *inode_page; /* its inode page, NULL is possible */ 873 struct page *node_page; /* cached direct node page */ 874 nid_t nid; /* node id of the direct node block */ 875 unsigned int ofs_in_node; /* data offset in the node page */ 876 bool inode_page_locked; /* inode page is locked or not */ 877 bool node_changed; /* is node block changed */ 878 char cur_level; /* level of hole node page */ 879 char max_level; /* level of current page located */ 880 block_t data_blkaddr; /* block address of the node block */ 881 }; 882 883 static inline void set_new_dnode(struct dnode_of_data *dn, struct inode *inode, 884 struct page *ipage, struct page *npage, nid_t nid) 885 { 886 memset(dn, 0, sizeof(*dn)); 887 dn->inode = inode; 888 dn->inode_page = ipage; 889 dn->node_page = npage; 890 dn->nid = nid; 891 } 892 893 /* 894 * For SIT manager 895 * 896 * By default, there are 6 active log areas across the whole main area. 897 * When considering hot and cold data separation to reduce cleaning overhead, 898 * we split 3 for data logs and 3 for node logs as hot, warm, and cold types, 899 * respectively. 900 * In the current design, you should not change the numbers intentionally. 901 * Instead, as a mount option such as active_logs=x, you can use 2, 4, and 6 902 * logs individually according to the underlying devices. (default: 6) 903 * Just in case, on-disk layout covers maximum 16 logs that consist of 8 for 904 * data and 8 for node logs. 905 */ 906 #define NR_CURSEG_DATA_TYPE (3) 907 #define NR_CURSEG_NODE_TYPE (3) 908 #define NR_CURSEG_INMEM_TYPE (2) 909 #define NR_CURSEG_PERSIST_TYPE (NR_CURSEG_DATA_TYPE + NR_CURSEG_NODE_TYPE) 910 #define NR_CURSEG_TYPE (NR_CURSEG_INMEM_TYPE + NR_CURSEG_PERSIST_TYPE) 911 912 enum { 913 CURSEG_HOT_DATA = 0, /* directory entry blocks */ 914 CURSEG_WARM_DATA, /* data blocks */ 915 CURSEG_COLD_DATA, /* multimedia or GCed data blocks */ 916 CURSEG_HOT_NODE, /* direct node blocks of directory files */ 917 CURSEG_WARM_NODE, /* direct node blocks of normal files */ 918 CURSEG_COLD_NODE, /* indirect node blocks */ 919 NR_PERSISTENT_LOG, /* number of persistent log */ 920 CURSEG_COLD_DATA_PINNED = NR_PERSISTENT_LOG, 921 /* pinned file that needs consecutive block address */ 922 CURSEG_ALL_DATA_ATGC, /* SSR alloctor in hot/warm/cold data area */ 923 NO_CHECK_TYPE, /* number of persistent & inmem log */ 924 }; 925 926 struct flush_cmd { 927 struct completion wait; 928 struct llist_node llnode; 929 nid_t ino; 930 int ret; 931 }; 932 933 struct flush_cmd_control { 934 struct task_struct *f2fs_issue_flush; /* flush thread */ 935 wait_queue_head_t flush_wait_queue; /* waiting queue for wake-up */ 936 atomic_t issued_flush; /* # of issued flushes */ 937 atomic_t queued_flush; /* # of queued flushes */ 938 struct llist_head issue_list; /* list for command issue */ 939 struct llist_node *dispatch_list; /* list for command dispatch */ 940 }; 941 942 struct f2fs_sm_info { 943 struct sit_info *sit_info; /* whole segment information */ 944 struct free_segmap_info *free_info; /* free segment information */ 945 struct dirty_seglist_info *dirty_info; /* dirty segment information */ 946 struct curseg_info *curseg_array; /* active segment information */ 947 948 struct rw_semaphore curseg_lock; /* for preventing curseg change */ 949 950 block_t seg0_blkaddr; /* block address of 0'th segment */ 951 block_t main_blkaddr; /* start block address of main area */ 952 block_t ssa_blkaddr; /* start block address of SSA area */ 953 954 unsigned int segment_count; /* total # of segments */ 955 unsigned int main_segments; /* # of segments in main area */ 956 unsigned int reserved_segments; /* # of reserved segments */ 957 unsigned int ovp_segments; /* # of overprovision segments */ 958 959 /* a threshold to reclaim prefree segments */ 960 unsigned int rec_prefree_segments; 961 962 /* for batched trimming */ 963 unsigned int trim_sections; /* # of sections to trim */ 964 965 struct list_head sit_entry_set; /* sit entry set list */ 966 967 unsigned int ipu_policy; /* in-place-update policy */ 968 unsigned int min_ipu_util; /* in-place-update threshold */ 969 unsigned int min_fsync_blocks; /* threshold for fsync */ 970 unsigned int min_seq_blocks; /* threshold for sequential blocks */ 971 unsigned int min_hot_blocks; /* threshold for hot block allocation */ 972 unsigned int min_ssr_sections; /* threshold to trigger SSR allocation */ 973 974 /* for flush command control */ 975 struct flush_cmd_control *fcc_info; 976 977 /* for discard command control */ 978 struct discard_cmd_control *dcc_info; 979 }; 980 981 /* 982 * For superblock 983 */ 984 /* 985 * COUNT_TYPE for monitoring 986 * 987 * f2fs monitors the number of several block types such as on-writeback, 988 * dirty dentry blocks, dirty node blocks, and dirty meta blocks. 989 */ 990 #define WB_DATA_TYPE(p) (__is_cp_guaranteed(p) ? F2FS_WB_CP_DATA : F2FS_WB_DATA) 991 enum count_type { 992 F2FS_DIRTY_DENTS, 993 F2FS_DIRTY_DATA, 994 F2FS_DIRTY_QDATA, 995 F2FS_DIRTY_NODES, 996 F2FS_DIRTY_META, 997 F2FS_INMEM_PAGES, 998 F2FS_DIRTY_IMETA, 999 F2FS_WB_CP_DATA, 1000 F2FS_WB_DATA, 1001 F2FS_RD_DATA, 1002 F2FS_RD_NODE, 1003 F2FS_RD_META, 1004 F2FS_DIO_WRITE, 1005 F2FS_DIO_READ, 1006 NR_COUNT_TYPE, 1007 }; 1008 1009 /* 1010 * The below are the page types of bios used in submit_bio(). 1011 * The available types are: 1012 * DATA User data pages. It operates as async mode. 1013 * NODE Node pages. It operates as async mode. 1014 * META FS metadata pages such as SIT, NAT, CP. 1015 * NR_PAGE_TYPE The number of page types. 1016 * META_FLUSH Make sure the previous pages are written 1017 * with waiting the bio's completion 1018 * ... Only can be used with META. 1019 */ 1020 #define PAGE_TYPE_OF_BIO(type) ((type) > META ? META : (type)) 1021 enum page_type { 1022 DATA, 1023 NODE, 1024 META, 1025 NR_PAGE_TYPE, 1026 META_FLUSH, 1027 INMEM, /* the below types are used by tracepoints only. */ 1028 INMEM_DROP, 1029 INMEM_INVALIDATE, 1030 INMEM_REVOKE, 1031 IPU, 1032 OPU, 1033 }; 1034 1035 enum temp_type { 1036 HOT = 0, /* must be zero for meta bio */ 1037 WARM, 1038 COLD, 1039 NR_TEMP_TYPE, 1040 }; 1041 1042 enum need_lock_type { 1043 LOCK_REQ = 0, 1044 LOCK_DONE, 1045 LOCK_RETRY, 1046 }; 1047 1048 enum cp_reason_type { 1049 CP_NO_NEEDED, 1050 CP_NON_REGULAR, 1051 CP_COMPRESSED, 1052 CP_HARDLINK, 1053 CP_SB_NEED_CP, 1054 CP_WRONG_PINO, 1055 CP_NO_SPC_ROLL, 1056 CP_NODE_NEED_CP, 1057 CP_FASTBOOT_MODE, 1058 CP_SPEC_LOG_NUM, 1059 CP_RECOVER_DIR, 1060 }; 1061 1062 enum iostat_type { 1063 /* WRITE IO */ 1064 APP_DIRECT_IO, /* app direct write IOs */ 1065 APP_BUFFERED_IO, /* app buffered write IOs */ 1066 APP_WRITE_IO, /* app write IOs */ 1067 APP_MAPPED_IO, /* app mapped IOs */ 1068 FS_DATA_IO, /* data IOs from kworker/fsync/reclaimer */ 1069 FS_NODE_IO, /* node IOs from kworker/fsync/reclaimer */ 1070 FS_META_IO, /* meta IOs from kworker/reclaimer */ 1071 FS_GC_DATA_IO, /* data IOs from forground gc */ 1072 FS_GC_NODE_IO, /* node IOs from forground gc */ 1073 FS_CP_DATA_IO, /* data IOs from checkpoint */ 1074 FS_CP_NODE_IO, /* node IOs from checkpoint */ 1075 FS_CP_META_IO, /* meta IOs from checkpoint */ 1076 1077 /* READ IO */ 1078 APP_DIRECT_READ_IO, /* app direct read IOs */ 1079 APP_BUFFERED_READ_IO, /* app buffered read IOs */ 1080 APP_READ_IO, /* app read IOs */ 1081 APP_MAPPED_READ_IO, /* app mapped read IOs */ 1082 FS_DATA_READ_IO, /* data read IOs */ 1083 FS_GDATA_READ_IO, /* data read IOs from background gc */ 1084 FS_CDATA_READ_IO, /* compressed data read IOs */ 1085 FS_NODE_READ_IO, /* node read IOs */ 1086 FS_META_READ_IO, /* meta read IOs */ 1087 1088 /* other */ 1089 FS_DISCARD, /* discard */ 1090 NR_IO_TYPE, 1091 }; 1092 1093 struct f2fs_io_info { 1094 struct f2fs_sb_info *sbi; /* f2fs_sb_info pointer */ 1095 nid_t ino; /* inode number */ 1096 enum page_type type; /* contains DATA/NODE/META/META_FLUSH */ 1097 enum temp_type temp; /* contains HOT/WARM/COLD */ 1098 int op; /* contains REQ_OP_ */ 1099 int op_flags; /* req_flag_bits */ 1100 block_t new_blkaddr; /* new block address to be written */ 1101 block_t old_blkaddr; /* old block address before Cow */ 1102 struct page *page; /* page to be written */ 1103 struct page *encrypted_page; /* encrypted page */ 1104 struct page *compressed_page; /* compressed page */ 1105 struct list_head list; /* serialize IOs */ 1106 bool submitted; /* indicate IO submission */ 1107 int need_lock; /* indicate we need to lock cp_rwsem */ 1108 bool in_list; /* indicate fio is in io_list */ 1109 bool is_por; /* indicate IO is from recovery or not */ 1110 bool retry; /* need to reallocate block address */ 1111 int compr_blocks; /* # of compressed block addresses */ 1112 bool encrypted; /* indicate file is encrypted */ 1113 enum iostat_type io_type; /* io type */ 1114 struct writeback_control *io_wbc; /* writeback control */ 1115 struct bio **bio; /* bio for ipu */ 1116 sector_t *last_block; /* last block number in bio */ 1117 unsigned char version; /* version of the node */ 1118 }; 1119 1120 struct bio_entry { 1121 struct bio *bio; 1122 struct list_head list; 1123 }; 1124 1125 #define is_read_io(rw) ((rw) == READ) 1126 struct f2fs_bio_info { 1127 struct f2fs_sb_info *sbi; /* f2fs superblock */ 1128 struct bio *bio; /* bios to merge */ 1129 sector_t last_block_in_bio; /* last block number */ 1130 struct f2fs_io_info fio; /* store buffered io info. */ 1131 struct rw_semaphore io_rwsem; /* blocking op for bio */ 1132 spinlock_t io_lock; /* serialize DATA/NODE IOs */ 1133 struct list_head io_list; /* track fios */ 1134 struct list_head bio_list; /* bio entry list head */ 1135 struct rw_semaphore bio_list_lock; /* lock to protect bio entry list */ 1136 }; 1137 1138 #define FDEV(i) (sbi->devs[i]) 1139 #define RDEV(i) (raw_super->devs[i]) 1140 struct f2fs_dev_info { 1141 struct block_device *bdev; 1142 char path[MAX_PATH_LEN]; 1143 unsigned int total_segments; 1144 block_t start_blk; 1145 block_t end_blk; 1146 #ifdef CONFIG_BLK_DEV_ZONED 1147 unsigned int nr_blkz; /* Total number of zones */ 1148 unsigned long *blkz_seq; /* Bitmap indicating sequential zones */ 1149 block_t *zone_capacity_blocks; /* Array of zone capacity in blks */ 1150 #endif 1151 }; 1152 1153 enum inode_type { 1154 DIR_INODE, /* for dirty dir inode */ 1155 FILE_INODE, /* for dirty regular/symlink inode */ 1156 DIRTY_META, /* for all dirtied inode metadata */ 1157 ATOMIC_FILE, /* for all atomic files */ 1158 NR_INODE_TYPE, 1159 }; 1160 1161 /* for inner inode cache management */ 1162 struct inode_management { 1163 struct radix_tree_root ino_root; /* ino entry array */ 1164 spinlock_t ino_lock; /* for ino entry lock */ 1165 struct list_head ino_list; /* inode list head */ 1166 unsigned long ino_num; /* number of entries */ 1167 }; 1168 1169 /* for GC_AT */ 1170 struct atgc_management { 1171 bool atgc_enabled; /* ATGC is enabled or not */ 1172 struct rb_root_cached root; /* root of victim rb-tree */ 1173 struct list_head victim_list; /* linked with all victim entries */ 1174 unsigned int victim_count; /* victim count in rb-tree */ 1175 unsigned int candidate_ratio; /* candidate ratio */ 1176 unsigned int max_candidate_count; /* max candidate count */ 1177 unsigned int age_weight; /* age weight, vblock_weight = 100 - age_weight */ 1178 unsigned long long age_threshold; /* age threshold */ 1179 }; 1180 1181 /* For s_flag in struct f2fs_sb_info */ 1182 enum { 1183 SBI_IS_DIRTY, /* dirty flag for checkpoint */ 1184 SBI_IS_CLOSE, /* specify unmounting */ 1185 SBI_NEED_FSCK, /* need fsck.f2fs to fix */ 1186 SBI_POR_DOING, /* recovery is doing or not */ 1187 SBI_NEED_SB_WRITE, /* need to recover superblock */ 1188 SBI_NEED_CP, /* need to checkpoint */ 1189 SBI_IS_SHUTDOWN, /* shutdown by ioctl */ 1190 SBI_IS_RECOVERED, /* recovered orphan/data */ 1191 SBI_CP_DISABLED, /* CP was disabled last mount */ 1192 SBI_CP_DISABLED_QUICK, /* CP was disabled quickly */ 1193 SBI_QUOTA_NEED_FLUSH, /* need to flush quota info in CP */ 1194 SBI_QUOTA_SKIP_FLUSH, /* skip flushing quota in current CP */ 1195 SBI_QUOTA_NEED_REPAIR, /* quota file may be corrupted */ 1196 SBI_IS_RESIZEFS, /* resizefs is in process */ 1197 }; 1198 1199 enum { 1200 CP_TIME, 1201 REQ_TIME, 1202 DISCARD_TIME, 1203 GC_TIME, 1204 DISABLE_TIME, 1205 UMOUNT_DISCARD_TIMEOUT, 1206 MAX_TIME, 1207 }; 1208 1209 enum { 1210 GC_NORMAL, 1211 GC_IDLE_CB, 1212 GC_IDLE_GREEDY, 1213 GC_IDLE_AT, 1214 GC_URGENT_HIGH, 1215 GC_URGENT_LOW, 1216 }; 1217 1218 enum { 1219 BGGC_MODE_ON, /* background gc is on */ 1220 BGGC_MODE_OFF, /* background gc is off */ 1221 BGGC_MODE_SYNC, /* 1222 * background gc is on, migrating blocks 1223 * like foreground gc 1224 */ 1225 }; 1226 1227 enum { 1228 FS_MODE_ADAPTIVE, /* use both lfs/ssr allocation */ 1229 FS_MODE_LFS, /* use lfs allocation only */ 1230 }; 1231 1232 enum { 1233 WHINT_MODE_OFF, /* not pass down write hints */ 1234 WHINT_MODE_USER, /* try to pass down hints given by users */ 1235 WHINT_MODE_FS, /* pass down hints with F2FS policy */ 1236 }; 1237 1238 enum { 1239 ALLOC_MODE_DEFAULT, /* stay default */ 1240 ALLOC_MODE_REUSE, /* reuse segments as much as possible */ 1241 }; 1242 1243 enum fsync_mode { 1244 FSYNC_MODE_POSIX, /* fsync follows posix semantics */ 1245 FSYNC_MODE_STRICT, /* fsync behaves in line with ext4 */ 1246 FSYNC_MODE_NOBARRIER, /* fsync behaves nobarrier based on posix */ 1247 }; 1248 1249 /* 1250 * this value is set in page as a private data which indicate that 1251 * the page is atomically written, and it is in inmem_pages list. 1252 */ 1253 #define ATOMIC_WRITTEN_PAGE ((unsigned long)-1) 1254 #define DUMMY_WRITTEN_PAGE ((unsigned long)-2) 1255 1256 #define IS_ATOMIC_WRITTEN_PAGE(page) \ 1257 (page_private(page) == ATOMIC_WRITTEN_PAGE) 1258 #define IS_DUMMY_WRITTEN_PAGE(page) \ 1259 (page_private(page) == DUMMY_WRITTEN_PAGE) 1260 1261 #ifdef CONFIG_F2FS_IO_TRACE 1262 #define IS_IO_TRACED_PAGE(page) \ 1263 (page_private(page) > 0 && \ 1264 page_private(page) < (unsigned long)PID_MAX_LIMIT) 1265 #else 1266 #define IS_IO_TRACED_PAGE(page) (0) 1267 #endif 1268 1269 /* For compression */ 1270 enum compress_algorithm_type { 1271 COMPRESS_LZO, 1272 COMPRESS_LZ4, 1273 COMPRESS_ZSTD, 1274 COMPRESS_LZORLE, 1275 COMPRESS_MAX, 1276 }; 1277 1278 enum compress_flag { 1279 COMPRESS_CHKSUM, 1280 COMPRESS_MAX_FLAG, 1281 }; 1282 1283 #define COMPRESS_DATA_RESERVED_SIZE 4 1284 struct compress_data { 1285 __le32 clen; /* compressed data size */ 1286 __le32 chksum; /* compressed data chksum */ 1287 __le32 reserved[COMPRESS_DATA_RESERVED_SIZE]; /* reserved */ 1288 u8 cdata[]; /* compressed data */ 1289 }; 1290 1291 #define COMPRESS_HEADER_SIZE (sizeof(struct compress_data)) 1292 1293 #define F2FS_COMPRESSED_PAGE_MAGIC 0xF5F2C000 1294 1295 /* compress context */ 1296 struct compress_ctx { 1297 struct inode *inode; /* inode the context belong to */ 1298 pgoff_t cluster_idx; /* cluster index number */ 1299 unsigned int cluster_size; /* page count in cluster */ 1300 unsigned int log_cluster_size; /* log of cluster size */ 1301 struct page **rpages; /* pages store raw data in cluster */ 1302 unsigned int nr_rpages; /* total page number in rpages */ 1303 struct page **cpages; /* pages store compressed data in cluster */ 1304 unsigned int nr_cpages; /* total page number in cpages */ 1305 void *rbuf; /* virtual mapped address on rpages */ 1306 struct compress_data *cbuf; /* virtual mapped address on cpages */ 1307 size_t rlen; /* valid data length in rbuf */ 1308 size_t clen; /* valid data length in cbuf */ 1309 void *private; /* payload buffer for specified compression algorithm */ 1310 void *private2; /* extra payload buffer */ 1311 }; 1312 1313 /* compress context for write IO path */ 1314 struct compress_io_ctx { 1315 u32 magic; /* magic number to indicate page is compressed */ 1316 struct inode *inode; /* inode the context belong to */ 1317 struct page **rpages; /* pages store raw data in cluster */ 1318 unsigned int nr_rpages; /* total page number in rpages */ 1319 atomic_t pending_pages; /* in-flight compressed page count */ 1320 }; 1321 1322 /* decompress io context for read IO path */ 1323 struct decompress_io_ctx { 1324 u32 magic; /* magic number to indicate page is compressed */ 1325 struct inode *inode; /* inode the context belong to */ 1326 pgoff_t cluster_idx; /* cluster index number */ 1327 unsigned int cluster_size; /* page count in cluster */ 1328 unsigned int log_cluster_size; /* log of cluster size */ 1329 struct page **rpages; /* pages store raw data in cluster */ 1330 unsigned int nr_rpages; /* total page number in rpages */ 1331 struct page **cpages; /* pages store compressed data in cluster */ 1332 unsigned int nr_cpages; /* total page number in cpages */ 1333 struct page **tpages; /* temp pages to pad holes in cluster */ 1334 void *rbuf; /* virtual mapped address on rpages */ 1335 struct compress_data *cbuf; /* virtual mapped address on cpages */ 1336 size_t rlen; /* valid data length in rbuf */ 1337 size_t clen; /* valid data length in cbuf */ 1338 atomic_t pending_pages; /* in-flight compressed page count */ 1339 bool failed; /* indicate IO error during decompression */ 1340 void *private; /* payload buffer for specified decompression algorithm */ 1341 void *private2; /* extra payload buffer */ 1342 }; 1343 1344 #define NULL_CLUSTER ((unsigned int)(~0)) 1345 #define MIN_COMPRESS_LOG_SIZE 2 1346 #define MAX_COMPRESS_LOG_SIZE 8 1347 #define MAX_COMPRESS_WINDOW_SIZE(log_size) ((PAGE_SIZE) << (log_size)) 1348 1349 struct f2fs_sb_info { 1350 struct super_block *sb; /* pointer to VFS super block */ 1351 struct proc_dir_entry *s_proc; /* proc entry */ 1352 struct f2fs_super_block *raw_super; /* raw super block pointer */ 1353 struct rw_semaphore sb_lock; /* lock for raw super block */ 1354 int valid_super_block; /* valid super block no */ 1355 unsigned long s_flag; /* flags for sbi */ 1356 struct mutex writepages; /* mutex for writepages() */ 1357 1358 #ifdef CONFIG_BLK_DEV_ZONED 1359 unsigned int blocks_per_blkz; /* F2FS blocks per zone */ 1360 unsigned int log_blocks_per_blkz; /* log2 F2FS blocks per zone */ 1361 #endif 1362 1363 /* for node-related operations */ 1364 struct f2fs_nm_info *nm_info; /* node manager */ 1365 struct inode *node_inode; /* cache node blocks */ 1366 1367 /* for segment-related operations */ 1368 struct f2fs_sm_info *sm_info; /* segment manager */ 1369 1370 /* for bio operations */ 1371 struct f2fs_bio_info *write_io[NR_PAGE_TYPE]; /* for write bios */ 1372 /* keep migration IO order for LFS mode */ 1373 struct rw_semaphore io_order_lock; 1374 mempool_t *write_io_dummy; /* Dummy pages */ 1375 1376 /* for checkpoint */ 1377 struct f2fs_checkpoint *ckpt; /* raw checkpoint pointer */ 1378 int cur_cp_pack; /* remain current cp pack */ 1379 spinlock_t cp_lock; /* for flag in ckpt */ 1380 struct inode *meta_inode; /* cache meta blocks */ 1381 struct rw_semaphore cp_global_sem; /* checkpoint procedure lock */ 1382 struct rw_semaphore cp_rwsem; /* blocking FS operations */ 1383 struct rw_semaphore node_write; /* locking node writes */ 1384 struct rw_semaphore node_change; /* locking node change */ 1385 wait_queue_head_t cp_wait; 1386 unsigned long last_time[MAX_TIME]; /* to store time in jiffies */ 1387 long interval_time[MAX_TIME]; /* to store thresholds */ 1388 1389 struct inode_management im[MAX_INO_ENTRY]; /* manage inode cache */ 1390 1391 spinlock_t fsync_node_lock; /* for node entry lock */ 1392 struct list_head fsync_node_list; /* node list head */ 1393 unsigned int fsync_seg_id; /* sequence id */ 1394 unsigned int fsync_node_num; /* number of node entries */ 1395 1396 /* for orphan inode, use 0'th array */ 1397 unsigned int max_orphans; /* max orphan inodes */ 1398 1399 /* for inode management */ 1400 struct list_head inode_list[NR_INODE_TYPE]; /* dirty inode list */ 1401 spinlock_t inode_lock[NR_INODE_TYPE]; /* for dirty inode list lock */ 1402 struct mutex flush_lock; /* for flush exclusion */ 1403 1404 /* for extent tree cache */ 1405 struct radix_tree_root extent_tree_root;/* cache extent cache entries */ 1406 struct mutex extent_tree_lock; /* locking extent radix tree */ 1407 struct list_head extent_list; /* lru list for shrinker */ 1408 spinlock_t extent_lock; /* locking extent lru list */ 1409 atomic_t total_ext_tree; /* extent tree count */ 1410 struct list_head zombie_list; /* extent zombie tree list */ 1411 atomic_t total_zombie_tree; /* extent zombie tree count */ 1412 atomic_t total_ext_node; /* extent info count */ 1413 1414 /* basic filesystem units */ 1415 unsigned int log_sectors_per_block; /* log2 sectors per block */ 1416 unsigned int log_blocksize; /* log2 block size */ 1417 unsigned int blocksize; /* block size */ 1418 unsigned int root_ino_num; /* root inode number*/ 1419 unsigned int node_ino_num; /* node inode number*/ 1420 unsigned int meta_ino_num; /* meta inode number*/ 1421 unsigned int log_blocks_per_seg; /* log2 blocks per segment */ 1422 unsigned int blocks_per_seg; /* blocks per segment */ 1423 unsigned int segs_per_sec; /* segments per section */ 1424 unsigned int secs_per_zone; /* sections per zone */ 1425 unsigned int total_sections; /* total section count */ 1426 unsigned int total_node_count; /* total node block count */ 1427 unsigned int total_valid_node_count; /* valid node block count */ 1428 loff_t max_file_blocks; /* max block index of file */ 1429 int dir_level; /* directory level */ 1430 int readdir_ra; /* readahead inode in readdir */ 1431 1432 block_t user_block_count; /* # of user blocks */ 1433 block_t total_valid_block_count; /* # of valid blocks */ 1434 block_t discard_blks; /* discard command candidats */ 1435 block_t last_valid_block_count; /* for recovery */ 1436 block_t reserved_blocks; /* configurable reserved blocks */ 1437 block_t current_reserved_blocks; /* current reserved blocks */ 1438 1439 /* Additional tracking for no checkpoint mode */ 1440 block_t unusable_block_count; /* # of blocks saved by last cp */ 1441 1442 unsigned int nquota_files; /* # of quota sysfile */ 1443 struct rw_semaphore quota_sem; /* blocking cp for flags */ 1444 1445 /* # of pages, see count_type */ 1446 atomic_t nr_pages[NR_COUNT_TYPE]; 1447 /* # of allocated blocks */ 1448 struct percpu_counter alloc_valid_block_count; 1449 1450 /* writeback control */ 1451 atomic_t wb_sync_req[META]; /* count # of WB_SYNC threads */ 1452 1453 /* valid inode count */ 1454 struct percpu_counter total_valid_inode_count; 1455 1456 struct f2fs_mount_info mount_opt; /* mount options */ 1457 1458 /* for cleaning operations */ 1459 struct rw_semaphore gc_lock; /* 1460 * semaphore for GC, avoid 1461 * race between GC and GC or CP 1462 */ 1463 struct f2fs_gc_kthread *gc_thread; /* GC thread */ 1464 struct atgc_management am; /* atgc management */ 1465 unsigned int cur_victim_sec; /* current victim section num */ 1466 unsigned int gc_mode; /* current GC state */ 1467 unsigned int next_victim_seg[2]; /* next segment in victim section */ 1468 1469 /* for skip statistic */ 1470 unsigned int atomic_files; /* # of opened atomic file */ 1471 unsigned long long skipped_atomic_files[2]; /* FG_GC and BG_GC */ 1472 unsigned long long skipped_gc_rwsem; /* FG_GC only */ 1473 1474 /* threshold for gc trials on pinned files */ 1475 u64 gc_pin_file_threshold; 1476 struct rw_semaphore pin_sem; 1477 1478 /* maximum # of trials to find a victim segment for SSR and GC */ 1479 unsigned int max_victim_search; 1480 /* migration granularity of garbage collection, unit: segment */ 1481 unsigned int migration_granularity; 1482 1483 /* 1484 * for stat information. 1485 * one is for the LFS mode, and the other is for the SSR mode. 1486 */ 1487 #ifdef CONFIG_F2FS_STAT_FS 1488 struct f2fs_stat_info *stat_info; /* FS status information */ 1489 atomic_t meta_count[META_MAX]; /* # of meta blocks */ 1490 unsigned int segment_count[2]; /* # of allocated segments */ 1491 unsigned int block_count[2]; /* # of allocated blocks */ 1492 atomic_t inplace_count; /* # of inplace update */ 1493 atomic64_t total_hit_ext; /* # of lookup extent cache */ 1494 atomic64_t read_hit_rbtree; /* # of hit rbtree extent node */ 1495 atomic64_t read_hit_largest; /* # of hit largest extent node */ 1496 atomic64_t read_hit_cached; /* # of hit cached extent node */ 1497 atomic_t inline_xattr; /* # of inline_xattr inodes */ 1498 atomic_t inline_inode; /* # of inline_data inodes */ 1499 atomic_t inline_dir; /* # of inline_dentry inodes */ 1500 atomic_t compr_inode; /* # of compressed inodes */ 1501 atomic64_t compr_blocks; /* # of compressed blocks */ 1502 atomic_t vw_cnt; /* # of volatile writes */ 1503 atomic_t max_aw_cnt; /* max # of atomic writes */ 1504 atomic_t max_vw_cnt; /* max # of volatile writes */ 1505 unsigned int io_skip_bggc; /* skip background gc for in-flight IO */ 1506 unsigned int other_skip_bggc; /* skip background gc for other reasons */ 1507 unsigned int ndirty_inode[NR_INODE_TYPE]; /* # of dirty inodes */ 1508 #endif 1509 spinlock_t stat_lock; /* lock for stat operations */ 1510 1511 /* For app/fs IO statistics */ 1512 spinlock_t iostat_lock; 1513 unsigned long long rw_iostat[NR_IO_TYPE]; 1514 unsigned long long prev_rw_iostat[NR_IO_TYPE]; 1515 bool iostat_enable; 1516 unsigned long iostat_next_period; 1517 unsigned int iostat_period_ms; 1518 1519 /* to attach REQ_META|REQ_FUA flags */ 1520 unsigned int data_io_flag; 1521 unsigned int node_io_flag; 1522 1523 /* For sysfs suppport */ 1524 struct kobject s_kobj; 1525 struct completion s_kobj_unregister; 1526 1527 /* For shrinker support */ 1528 struct list_head s_list; 1529 int s_ndevs; /* number of devices */ 1530 struct f2fs_dev_info *devs; /* for device list */ 1531 unsigned int dirty_device; /* for checkpoint data flush */ 1532 spinlock_t dev_lock; /* protect dirty_device */ 1533 struct mutex umount_mutex; 1534 unsigned int shrinker_run_no; 1535 1536 /* For write statistics */ 1537 u64 sectors_written_start; 1538 u64 kbytes_written; 1539 1540 /* Reference to checksum algorithm driver via cryptoapi */ 1541 struct crypto_shash *s_chksum_driver; 1542 1543 /* Precomputed FS UUID checksum for seeding other checksums */ 1544 __u32 s_chksum_seed; 1545 1546 struct workqueue_struct *post_read_wq; /* post read workqueue */ 1547 1548 struct kmem_cache *inline_xattr_slab; /* inline xattr entry */ 1549 unsigned int inline_xattr_slab_size; /* default inline xattr slab size */ 1550 1551 #ifdef CONFIG_F2FS_FS_COMPRESSION 1552 struct kmem_cache *page_array_slab; /* page array entry */ 1553 unsigned int page_array_slab_size; /* default page array slab size */ 1554 #endif 1555 }; 1556 1557 struct f2fs_private_dio { 1558 struct inode *inode; 1559 void *orig_private; 1560 bio_end_io_t *orig_end_io; 1561 bool write; 1562 }; 1563 1564 #ifdef CONFIG_F2FS_FAULT_INJECTION 1565 #define f2fs_show_injection_info(sbi, type) \ 1566 printk_ratelimited("%sF2FS-fs (%s) : inject %s in %s of %pS\n", \ 1567 KERN_INFO, sbi->sb->s_id, \ 1568 f2fs_fault_name[type], \ 1569 __func__, __builtin_return_address(0)) 1570 static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type) 1571 { 1572 struct f2fs_fault_info *ffi = &F2FS_OPTION(sbi).fault_info; 1573 1574 if (!ffi->inject_rate) 1575 return false; 1576 1577 if (!IS_FAULT_SET(ffi, type)) 1578 return false; 1579 1580 atomic_inc(&ffi->inject_ops); 1581 if (atomic_read(&ffi->inject_ops) >= ffi->inject_rate) { 1582 atomic_set(&ffi->inject_ops, 0); 1583 return true; 1584 } 1585 return false; 1586 } 1587 #else 1588 #define f2fs_show_injection_info(sbi, type) do { } while (0) 1589 static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type) 1590 { 1591 return false; 1592 } 1593 #endif 1594 1595 /* 1596 * Test if the mounted volume is a multi-device volume. 1597 * - For a single regular disk volume, sbi->s_ndevs is 0. 1598 * - For a single zoned disk volume, sbi->s_ndevs is 1. 1599 * - For a multi-device volume, sbi->s_ndevs is always 2 or more. 1600 */ 1601 static inline bool f2fs_is_multi_device(struct f2fs_sb_info *sbi) 1602 { 1603 return sbi->s_ndevs > 1; 1604 } 1605 1606 static inline void f2fs_update_time(struct f2fs_sb_info *sbi, int type) 1607 { 1608 unsigned long now = jiffies; 1609 1610 sbi->last_time[type] = now; 1611 1612 /* DISCARD_TIME and GC_TIME are based on REQ_TIME */ 1613 if (type == REQ_TIME) { 1614 sbi->last_time[DISCARD_TIME] = now; 1615 sbi->last_time[GC_TIME] = now; 1616 } 1617 } 1618 1619 static inline bool f2fs_time_over(struct f2fs_sb_info *sbi, int type) 1620 { 1621 unsigned long interval = sbi->interval_time[type] * HZ; 1622 1623 return time_after(jiffies, sbi->last_time[type] + interval); 1624 } 1625 1626 static inline unsigned int f2fs_time_to_wait(struct f2fs_sb_info *sbi, 1627 int type) 1628 { 1629 unsigned long interval = sbi->interval_time[type] * HZ; 1630 unsigned int wait_ms = 0; 1631 long delta; 1632 1633 delta = (sbi->last_time[type] + interval) - jiffies; 1634 if (delta > 0) 1635 wait_ms = jiffies_to_msecs(delta); 1636 1637 return wait_ms; 1638 } 1639 1640 /* 1641 * Inline functions 1642 */ 1643 static inline u32 __f2fs_crc32(struct f2fs_sb_info *sbi, u32 crc, 1644 const void *address, unsigned int length) 1645 { 1646 struct { 1647 struct shash_desc shash; 1648 char ctx[4]; 1649 } desc; 1650 int err; 1651 1652 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver) != sizeof(desc.ctx)); 1653 1654 desc.shash.tfm = sbi->s_chksum_driver; 1655 *(u32 *)desc.ctx = crc; 1656 1657 err = crypto_shash_update(&desc.shash, address, length); 1658 BUG_ON(err); 1659 1660 return *(u32 *)desc.ctx; 1661 } 1662 1663 static inline u32 f2fs_crc32(struct f2fs_sb_info *sbi, const void *address, 1664 unsigned int length) 1665 { 1666 return __f2fs_crc32(sbi, F2FS_SUPER_MAGIC, address, length); 1667 } 1668 1669 static inline bool f2fs_crc_valid(struct f2fs_sb_info *sbi, __u32 blk_crc, 1670 void *buf, size_t buf_size) 1671 { 1672 return f2fs_crc32(sbi, buf, buf_size) == blk_crc; 1673 } 1674 1675 static inline u32 f2fs_chksum(struct f2fs_sb_info *sbi, u32 crc, 1676 const void *address, unsigned int length) 1677 { 1678 return __f2fs_crc32(sbi, crc, address, length); 1679 } 1680 1681 static inline struct f2fs_inode_info *F2FS_I(struct inode *inode) 1682 { 1683 return container_of(inode, struct f2fs_inode_info, vfs_inode); 1684 } 1685 1686 static inline struct f2fs_sb_info *F2FS_SB(struct super_block *sb) 1687 { 1688 return sb->s_fs_info; 1689 } 1690 1691 static inline struct f2fs_sb_info *F2FS_I_SB(struct inode *inode) 1692 { 1693 return F2FS_SB(inode->i_sb); 1694 } 1695 1696 static inline struct f2fs_sb_info *F2FS_M_SB(struct address_space *mapping) 1697 { 1698 return F2FS_I_SB(mapping->host); 1699 } 1700 1701 static inline struct f2fs_sb_info *F2FS_P_SB(struct page *page) 1702 { 1703 return F2FS_M_SB(page_file_mapping(page)); 1704 } 1705 1706 static inline struct f2fs_super_block *F2FS_RAW_SUPER(struct f2fs_sb_info *sbi) 1707 { 1708 return (struct f2fs_super_block *)(sbi->raw_super); 1709 } 1710 1711 static inline struct f2fs_checkpoint *F2FS_CKPT(struct f2fs_sb_info *sbi) 1712 { 1713 return (struct f2fs_checkpoint *)(sbi->ckpt); 1714 } 1715 1716 static inline struct f2fs_node *F2FS_NODE(struct page *page) 1717 { 1718 return (struct f2fs_node *)page_address(page); 1719 } 1720 1721 static inline struct f2fs_inode *F2FS_INODE(struct page *page) 1722 { 1723 return &((struct f2fs_node *)page_address(page))->i; 1724 } 1725 1726 static inline struct f2fs_nm_info *NM_I(struct f2fs_sb_info *sbi) 1727 { 1728 return (struct f2fs_nm_info *)(sbi->nm_info); 1729 } 1730 1731 static inline struct f2fs_sm_info *SM_I(struct f2fs_sb_info *sbi) 1732 { 1733 return (struct f2fs_sm_info *)(sbi->sm_info); 1734 } 1735 1736 static inline struct sit_info *SIT_I(struct f2fs_sb_info *sbi) 1737 { 1738 return (struct sit_info *)(SM_I(sbi)->sit_info); 1739 } 1740 1741 static inline struct free_segmap_info *FREE_I(struct f2fs_sb_info *sbi) 1742 { 1743 return (struct free_segmap_info *)(SM_I(sbi)->free_info); 1744 } 1745 1746 static inline struct dirty_seglist_info *DIRTY_I(struct f2fs_sb_info *sbi) 1747 { 1748 return (struct dirty_seglist_info *)(SM_I(sbi)->dirty_info); 1749 } 1750 1751 static inline struct address_space *META_MAPPING(struct f2fs_sb_info *sbi) 1752 { 1753 return sbi->meta_inode->i_mapping; 1754 } 1755 1756 static inline struct address_space *NODE_MAPPING(struct f2fs_sb_info *sbi) 1757 { 1758 return sbi->node_inode->i_mapping; 1759 } 1760 1761 static inline bool is_sbi_flag_set(struct f2fs_sb_info *sbi, unsigned int type) 1762 { 1763 return test_bit(type, &sbi->s_flag); 1764 } 1765 1766 static inline void set_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type) 1767 { 1768 set_bit(type, &sbi->s_flag); 1769 } 1770 1771 static inline void clear_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type) 1772 { 1773 clear_bit(type, &sbi->s_flag); 1774 } 1775 1776 static inline unsigned long long cur_cp_version(struct f2fs_checkpoint *cp) 1777 { 1778 return le64_to_cpu(cp->checkpoint_ver); 1779 } 1780 1781 static inline unsigned long f2fs_qf_ino(struct super_block *sb, int type) 1782 { 1783 if (type < F2FS_MAX_QUOTAS) 1784 return le32_to_cpu(F2FS_SB(sb)->raw_super->qf_ino[type]); 1785 return 0; 1786 } 1787 1788 static inline __u64 cur_cp_crc(struct f2fs_checkpoint *cp) 1789 { 1790 size_t crc_offset = le32_to_cpu(cp->checksum_offset); 1791 return le32_to_cpu(*((__le32 *)((unsigned char *)cp + crc_offset))); 1792 } 1793 1794 static inline bool __is_set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 1795 { 1796 unsigned int ckpt_flags = le32_to_cpu(cp->ckpt_flags); 1797 1798 return ckpt_flags & f; 1799 } 1800 1801 static inline bool is_set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 1802 { 1803 return __is_set_ckpt_flags(F2FS_CKPT(sbi), f); 1804 } 1805 1806 static inline void __set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 1807 { 1808 unsigned int ckpt_flags; 1809 1810 ckpt_flags = le32_to_cpu(cp->ckpt_flags); 1811 ckpt_flags |= f; 1812 cp->ckpt_flags = cpu_to_le32(ckpt_flags); 1813 } 1814 1815 static inline void set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 1816 { 1817 unsigned long flags; 1818 1819 spin_lock_irqsave(&sbi->cp_lock, flags); 1820 __set_ckpt_flags(F2FS_CKPT(sbi), f); 1821 spin_unlock_irqrestore(&sbi->cp_lock, flags); 1822 } 1823 1824 static inline void __clear_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 1825 { 1826 unsigned int ckpt_flags; 1827 1828 ckpt_flags = le32_to_cpu(cp->ckpt_flags); 1829 ckpt_flags &= (~f); 1830 cp->ckpt_flags = cpu_to_le32(ckpt_flags); 1831 } 1832 1833 static inline void clear_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 1834 { 1835 unsigned long flags; 1836 1837 spin_lock_irqsave(&sbi->cp_lock, flags); 1838 __clear_ckpt_flags(F2FS_CKPT(sbi), f); 1839 spin_unlock_irqrestore(&sbi->cp_lock, flags); 1840 } 1841 1842 static inline void disable_nat_bits(struct f2fs_sb_info *sbi, bool lock) 1843 { 1844 unsigned long flags; 1845 unsigned char *nat_bits; 1846 1847 /* 1848 * In order to re-enable nat_bits we need to call fsck.f2fs by 1849 * set_sbi_flag(sbi, SBI_NEED_FSCK). But it may give huge cost, 1850 * so let's rely on regular fsck or unclean shutdown. 1851 */ 1852 1853 if (lock) 1854 spin_lock_irqsave(&sbi->cp_lock, flags); 1855 __clear_ckpt_flags(F2FS_CKPT(sbi), CP_NAT_BITS_FLAG); 1856 nat_bits = NM_I(sbi)->nat_bits; 1857 NM_I(sbi)->nat_bits = NULL; 1858 if (lock) 1859 spin_unlock_irqrestore(&sbi->cp_lock, flags); 1860 1861 kvfree(nat_bits); 1862 } 1863 1864 static inline bool enabled_nat_bits(struct f2fs_sb_info *sbi, 1865 struct cp_control *cpc) 1866 { 1867 bool set = is_set_ckpt_flags(sbi, CP_NAT_BITS_FLAG); 1868 1869 return (cpc) ? (cpc->reason & CP_UMOUNT) && set : set; 1870 } 1871 1872 static inline void f2fs_lock_op(struct f2fs_sb_info *sbi) 1873 { 1874 down_read(&sbi->cp_rwsem); 1875 } 1876 1877 static inline int f2fs_trylock_op(struct f2fs_sb_info *sbi) 1878 { 1879 return down_read_trylock(&sbi->cp_rwsem); 1880 } 1881 1882 static inline void f2fs_unlock_op(struct f2fs_sb_info *sbi) 1883 { 1884 up_read(&sbi->cp_rwsem); 1885 } 1886 1887 static inline void f2fs_lock_all(struct f2fs_sb_info *sbi) 1888 { 1889 down_write(&sbi->cp_rwsem); 1890 } 1891 1892 static inline void f2fs_unlock_all(struct f2fs_sb_info *sbi) 1893 { 1894 up_write(&sbi->cp_rwsem); 1895 } 1896 1897 static inline int __get_cp_reason(struct f2fs_sb_info *sbi) 1898 { 1899 int reason = CP_SYNC; 1900 1901 if (test_opt(sbi, FASTBOOT)) 1902 reason = CP_FASTBOOT; 1903 if (is_sbi_flag_set(sbi, SBI_IS_CLOSE)) 1904 reason = CP_UMOUNT; 1905 return reason; 1906 } 1907 1908 static inline bool __remain_node_summaries(int reason) 1909 { 1910 return (reason & (CP_UMOUNT | CP_FASTBOOT)); 1911 } 1912 1913 static inline bool __exist_node_summaries(struct f2fs_sb_info *sbi) 1914 { 1915 return (is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG) || 1916 is_set_ckpt_flags(sbi, CP_FASTBOOT_FLAG)); 1917 } 1918 1919 /* 1920 * Check whether the inode has blocks or not 1921 */ 1922 static inline int F2FS_HAS_BLOCKS(struct inode *inode) 1923 { 1924 block_t xattr_block = F2FS_I(inode)->i_xattr_nid ? 1 : 0; 1925 1926 return (inode->i_blocks >> F2FS_LOG_SECTORS_PER_BLOCK) > xattr_block; 1927 } 1928 1929 static inline bool f2fs_has_xattr_block(unsigned int ofs) 1930 { 1931 return ofs == XATTR_NODE_OFFSET; 1932 } 1933 1934 static inline bool __allow_reserved_blocks(struct f2fs_sb_info *sbi, 1935 struct inode *inode, bool cap) 1936 { 1937 if (!inode) 1938 return true; 1939 if (!test_opt(sbi, RESERVE_ROOT)) 1940 return false; 1941 if (IS_NOQUOTA(inode)) 1942 return true; 1943 if (uid_eq(F2FS_OPTION(sbi).s_resuid, current_fsuid())) 1944 return true; 1945 if (!gid_eq(F2FS_OPTION(sbi).s_resgid, GLOBAL_ROOT_GID) && 1946 in_group_p(F2FS_OPTION(sbi).s_resgid)) 1947 return true; 1948 if (cap && capable(CAP_SYS_RESOURCE)) 1949 return true; 1950 return false; 1951 } 1952 1953 static inline void f2fs_i_blocks_write(struct inode *, block_t, bool, bool); 1954 static inline int inc_valid_block_count(struct f2fs_sb_info *sbi, 1955 struct inode *inode, blkcnt_t *count) 1956 { 1957 blkcnt_t diff = 0, release = 0; 1958 block_t avail_user_block_count; 1959 int ret; 1960 1961 ret = dquot_reserve_block(inode, *count); 1962 if (ret) 1963 return ret; 1964 1965 if (time_to_inject(sbi, FAULT_BLOCK)) { 1966 f2fs_show_injection_info(sbi, FAULT_BLOCK); 1967 release = *count; 1968 goto release_quota; 1969 } 1970 1971 /* 1972 * let's increase this in prior to actual block count change in order 1973 * for f2fs_sync_file to avoid data races when deciding checkpoint. 1974 */ 1975 percpu_counter_add(&sbi->alloc_valid_block_count, (*count)); 1976 1977 spin_lock(&sbi->stat_lock); 1978 sbi->total_valid_block_count += (block_t)(*count); 1979 avail_user_block_count = sbi->user_block_count - 1980 sbi->current_reserved_blocks; 1981 1982 if (!__allow_reserved_blocks(sbi, inode, true)) 1983 avail_user_block_count -= F2FS_OPTION(sbi).root_reserved_blocks; 1984 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) { 1985 if (avail_user_block_count > sbi->unusable_block_count) 1986 avail_user_block_count -= sbi->unusable_block_count; 1987 else 1988 avail_user_block_count = 0; 1989 } 1990 if (unlikely(sbi->total_valid_block_count > avail_user_block_count)) { 1991 diff = sbi->total_valid_block_count - avail_user_block_count; 1992 if (diff > *count) 1993 diff = *count; 1994 *count -= diff; 1995 release = diff; 1996 sbi->total_valid_block_count -= diff; 1997 if (!*count) { 1998 spin_unlock(&sbi->stat_lock); 1999 goto enospc; 2000 } 2001 } 2002 spin_unlock(&sbi->stat_lock); 2003 2004 if (unlikely(release)) { 2005 percpu_counter_sub(&sbi->alloc_valid_block_count, release); 2006 dquot_release_reservation_block(inode, release); 2007 } 2008 f2fs_i_blocks_write(inode, *count, true, true); 2009 return 0; 2010 2011 enospc: 2012 percpu_counter_sub(&sbi->alloc_valid_block_count, release); 2013 release_quota: 2014 dquot_release_reservation_block(inode, release); 2015 return -ENOSPC; 2016 } 2017 2018 __printf(2, 3) 2019 void f2fs_printk(struct f2fs_sb_info *sbi, const char *fmt, ...); 2020 2021 #define f2fs_err(sbi, fmt, ...) \ 2022 f2fs_printk(sbi, KERN_ERR fmt, ##__VA_ARGS__) 2023 #define f2fs_warn(sbi, fmt, ...) \ 2024 f2fs_printk(sbi, KERN_WARNING fmt, ##__VA_ARGS__) 2025 #define f2fs_notice(sbi, fmt, ...) \ 2026 f2fs_printk(sbi, KERN_NOTICE fmt, ##__VA_ARGS__) 2027 #define f2fs_info(sbi, fmt, ...) \ 2028 f2fs_printk(sbi, KERN_INFO fmt, ##__VA_ARGS__) 2029 #define f2fs_debug(sbi, fmt, ...) \ 2030 f2fs_printk(sbi, KERN_DEBUG fmt, ##__VA_ARGS__) 2031 2032 static inline void dec_valid_block_count(struct f2fs_sb_info *sbi, 2033 struct inode *inode, 2034 block_t count) 2035 { 2036 blkcnt_t sectors = count << F2FS_LOG_SECTORS_PER_BLOCK; 2037 2038 spin_lock(&sbi->stat_lock); 2039 f2fs_bug_on(sbi, sbi->total_valid_block_count < (block_t) count); 2040 sbi->total_valid_block_count -= (block_t)count; 2041 if (sbi->reserved_blocks && 2042 sbi->current_reserved_blocks < sbi->reserved_blocks) 2043 sbi->current_reserved_blocks = min(sbi->reserved_blocks, 2044 sbi->current_reserved_blocks + count); 2045 spin_unlock(&sbi->stat_lock); 2046 if (unlikely(inode->i_blocks < sectors)) { 2047 f2fs_warn(sbi, "Inconsistent i_blocks, ino:%lu, iblocks:%llu, sectors:%llu", 2048 inode->i_ino, 2049 (unsigned long long)inode->i_blocks, 2050 (unsigned long long)sectors); 2051 set_sbi_flag(sbi, SBI_NEED_FSCK); 2052 return; 2053 } 2054 f2fs_i_blocks_write(inode, count, false, true); 2055 } 2056 2057 static inline void inc_page_count(struct f2fs_sb_info *sbi, int count_type) 2058 { 2059 atomic_inc(&sbi->nr_pages[count_type]); 2060 2061 if (count_type == F2FS_DIRTY_DENTS || 2062 count_type == F2FS_DIRTY_NODES || 2063 count_type == F2FS_DIRTY_META || 2064 count_type == F2FS_DIRTY_QDATA || 2065 count_type == F2FS_DIRTY_IMETA) 2066 set_sbi_flag(sbi, SBI_IS_DIRTY); 2067 } 2068 2069 static inline void inode_inc_dirty_pages(struct inode *inode) 2070 { 2071 atomic_inc(&F2FS_I(inode)->dirty_pages); 2072 inc_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ? 2073 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA); 2074 if (IS_NOQUOTA(inode)) 2075 inc_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA); 2076 } 2077 2078 static inline void dec_page_count(struct f2fs_sb_info *sbi, int count_type) 2079 { 2080 atomic_dec(&sbi->nr_pages[count_type]); 2081 } 2082 2083 static inline void inode_dec_dirty_pages(struct inode *inode) 2084 { 2085 if (!S_ISDIR(inode->i_mode) && !S_ISREG(inode->i_mode) && 2086 !S_ISLNK(inode->i_mode)) 2087 return; 2088 2089 atomic_dec(&F2FS_I(inode)->dirty_pages); 2090 dec_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ? 2091 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA); 2092 if (IS_NOQUOTA(inode)) 2093 dec_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA); 2094 } 2095 2096 static inline s64 get_pages(struct f2fs_sb_info *sbi, int count_type) 2097 { 2098 return atomic_read(&sbi->nr_pages[count_type]); 2099 } 2100 2101 static inline int get_dirty_pages(struct inode *inode) 2102 { 2103 return atomic_read(&F2FS_I(inode)->dirty_pages); 2104 } 2105 2106 static inline int get_blocktype_secs(struct f2fs_sb_info *sbi, int block_type) 2107 { 2108 unsigned int pages_per_sec = sbi->segs_per_sec * sbi->blocks_per_seg; 2109 unsigned int segs = (get_pages(sbi, block_type) + pages_per_sec - 1) >> 2110 sbi->log_blocks_per_seg; 2111 2112 return segs / sbi->segs_per_sec; 2113 } 2114 2115 static inline block_t valid_user_blocks(struct f2fs_sb_info *sbi) 2116 { 2117 return sbi->total_valid_block_count; 2118 } 2119 2120 static inline block_t discard_blocks(struct f2fs_sb_info *sbi) 2121 { 2122 return sbi->discard_blks; 2123 } 2124 2125 static inline unsigned long __bitmap_size(struct f2fs_sb_info *sbi, int flag) 2126 { 2127 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi); 2128 2129 /* return NAT or SIT bitmap */ 2130 if (flag == NAT_BITMAP) 2131 return le32_to_cpu(ckpt->nat_ver_bitmap_bytesize); 2132 else if (flag == SIT_BITMAP) 2133 return le32_to_cpu(ckpt->sit_ver_bitmap_bytesize); 2134 2135 return 0; 2136 } 2137 2138 static inline block_t __cp_payload(struct f2fs_sb_info *sbi) 2139 { 2140 return le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_payload); 2141 } 2142 2143 static inline void *__bitmap_ptr(struct f2fs_sb_info *sbi, int flag) 2144 { 2145 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi); 2146 int offset; 2147 2148 if (is_set_ckpt_flags(sbi, CP_LARGE_NAT_BITMAP_FLAG)) { 2149 offset = (flag == SIT_BITMAP) ? 2150 le32_to_cpu(ckpt->nat_ver_bitmap_bytesize) : 0; 2151 /* 2152 * if large_nat_bitmap feature is enabled, leave checksum 2153 * protection for all nat/sit bitmaps. 2154 */ 2155 return &ckpt->sit_nat_version_bitmap + offset + sizeof(__le32); 2156 } 2157 2158 if (__cp_payload(sbi) > 0) { 2159 if (flag == NAT_BITMAP) 2160 return &ckpt->sit_nat_version_bitmap; 2161 else 2162 return (unsigned char *)ckpt + F2FS_BLKSIZE; 2163 } else { 2164 offset = (flag == NAT_BITMAP) ? 2165 le32_to_cpu(ckpt->sit_ver_bitmap_bytesize) : 0; 2166 return &ckpt->sit_nat_version_bitmap + offset; 2167 } 2168 } 2169 2170 static inline block_t __start_cp_addr(struct f2fs_sb_info *sbi) 2171 { 2172 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr); 2173 2174 if (sbi->cur_cp_pack == 2) 2175 start_addr += sbi->blocks_per_seg; 2176 return start_addr; 2177 } 2178 2179 static inline block_t __start_cp_next_addr(struct f2fs_sb_info *sbi) 2180 { 2181 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr); 2182 2183 if (sbi->cur_cp_pack == 1) 2184 start_addr += sbi->blocks_per_seg; 2185 return start_addr; 2186 } 2187 2188 static inline void __set_cp_next_pack(struct f2fs_sb_info *sbi) 2189 { 2190 sbi->cur_cp_pack = (sbi->cur_cp_pack == 1) ? 2 : 1; 2191 } 2192 2193 static inline block_t __start_sum_addr(struct f2fs_sb_info *sbi) 2194 { 2195 return le32_to_cpu(F2FS_CKPT(sbi)->cp_pack_start_sum); 2196 } 2197 2198 static inline int inc_valid_node_count(struct f2fs_sb_info *sbi, 2199 struct inode *inode, bool is_inode) 2200 { 2201 block_t valid_block_count; 2202 unsigned int valid_node_count, user_block_count; 2203 int err; 2204 2205 if (is_inode) { 2206 if (inode) { 2207 err = dquot_alloc_inode(inode); 2208 if (err) 2209 return err; 2210 } 2211 } else { 2212 err = dquot_reserve_block(inode, 1); 2213 if (err) 2214 return err; 2215 } 2216 2217 if (time_to_inject(sbi, FAULT_BLOCK)) { 2218 f2fs_show_injection_info(sbi, FAULT_BLOCK); 2219 goto enospc; 2220 } 2221 2222 spin_lock(&sbi->stat_lock); 2223 2224 valid_block_count = sbi->total_valid_block_count + 2225 sbi->current_reserved_blocks + 1; 2226 2227 if (!__allow_reserved_blocks(sbi, inode, false)) 2228 valid_block_count += F2FS_OPTION(sbi).root_reserved_blocks; 2229 user_block_count = sbi->user_block_count; 2230 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) 2231 user_block_count -= sbi->unusable_block_count; 2232 2233 if (unlikely(valid_block_count > user_block_count)) { 2234 spin_unlock(&sbi->stat_lock); 2235 goto enospc; 2236 } 2237 2238 valid_node_count = sbi->total_valid_node_count + 1; 2239 if (unlikely(valid_node_count > sbi->total_node_count)) { 2240 spin_unlock(&sbi->stat_lock); 2241 goto enospc; 2242 } 2243 2244 sbi->total_valid_node_count++; 2245 sbi->total_valid_block_count++; 2246 spin_unlock(&sbi->stat_lock); 2247 2248 if (inode) { 2249 if (is_inode) 2250 f2fs_mark_inode_dirty_sync(inode, true); 2251 else 2252 f2fs_i_blocks_write(inode, 1, true, true); 2253 } 2254 2255 percpu_counter_inc(&sbi->alloc_valid_block_count); 2256 return 0; 2257 2258 enospc: 2259 if (is_inode) { 2260 if (inode) 2261 dquot_free_inode(inode); 2262 } else { 2263 dquot_release_reservation_block(inode, 1); 2264 } 2265 return -ENOSPC; 2266 } 2267 2268 static inline void dec_valid_node_count(struct f2fs_sb_info *sbi, 2269 struct inode *inode, bool is_inode) 2270 { 2271 spin_lock(&sbi->stat_lock); 2272 2273 f2fs_bug_on(sbi, !sbi->total_valid_block_count); 2274 f2fs_bug_on(sbi, !sbi->total_valid_node_count); 2275 2276 sbi->total_valid_node_count--; 2277 sbi->total_valid_block_count--; 2278 if (sbi->reserved_blocks && 2279 sbi->current_reserved_blocks < sbi->reserved_blocks) 2280 sbi->current_reserved_blocks++; 2281 2282 spin_unlock(&sbi->stat_lock); 2283 2284 if (is_inode) { 2285 dquot_free_inode(inode); 2286 } else { 2287 if (unlikely(inode->i_blocks == 0)) { 2288 f2fs_warn(sbi, "dec_valid_node_count: inconsistent i_blocks, ino:%lu, iblocks:%llu", 2289 inode->i_ino, 2290 (unsigned long long)inode->i_blocks); 2291 set_sbi_flag(sbi, SBI_NEED_FSCK); 2292 return; 2293 } 2294 f2fs_i_blocks_write(inode, 1, false, true); 2295 } 2296 } 2297 2298 static inline unsigned int valid_node_count(struct f2fs_sb_info *sbi) 2299 { 2300 return sbi->total_valid_node_count; 2301 } 2302 2303 static inline void inc_valid_inode_count(struct f2fs_sb_info *sbi) 2304 { 2305 percpu_counter_inc(&sbi->total_valid_inode_count); 2306 } 2307 2308 static inline void dec_valid_inode_count(struct f2fs_sb_info *sbi) 2309 { 2310 percpu_counter_dec(&sbi->total_valid_inode_count); 2311 } 2312 2313 static inline s64 valid_inode_count(struct f2fs_sb_info *sbi) 2314 { 2315 return percpu_counter_sum_positive(&sbi->total_valid_inode_count); 2316 } 2317 2318 static inline struct page *f2fs_grab_cache_page(struct address_space *mapping, 2319 pgoff_t index, bool for_write) 2320 { 2321 struct page *page; 2322 2323 if (IS_ENABLED(CONFIG_F2FS_FAULT_INJECTION)) { 2324 if (!for_write) 2325 page = find_get_page_flags(mapping, index, 2326 FGP_LOCK | FGP_ACCESSED); 2327 else 2328 page = find_lock_page(mapping, index); 2329 if (page) 2330 return page; 2331 2332 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_ALLOC)) { 2333 f2fs_show_injection_info(F2FS_M_SB(mapping), 2334 FAULT_PAGE_ALLOC); 2335 return NULL; 2336 } 2337 } 2338 2339 if (!for_write) 2340 return grab_cache_page(mapping, index); 2341 return grab_cache_page_write_begin(mapping, index, AOP_FLAG_NOFS); 2342 } 2343 2344 static inline struct page *f2fs_pagecache_get_page( 2345 struct address_space *mapping, pgoff_t index, 2346 int fgp_flags, gfp_t gfp_mask) 2347 { 2348 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_GET)) { 2349 f2fs_show_injection_info(F2FS_M_SB(mapping), FAULT_PAGE_GET); 2350 return NULL; 2351 } 2352 2353 return pagecache_get_page(mapping, index, fgp_flags, gfp_mask); 2354 } 2355 2356 static inline void f2fs_copy_page(struct page *src, struct page *dst) 2357 { 2358 char *src_kaddr = kmap(src); 2359 char *dst_kaddr = kmap(dst); 2360 2361 memcpy(dst_kaddr, src_kaddr, PAGE_SIZE); 2362 kunmap(dst); 2363 kunmap(src); 2364 } 2365 2366 static inline void f2fs_put_page(struct page *page, int unlock) 2367 { 2368 if (!page) 2369 return; 2370 2371 if (unlock) { 2372 f2fs_bug_on(F2FS_P_SB(page), !PageLocked(page)); 2373 unlock_page(page); 2374 } 2375 put_page(page); 2376 } 2377 2378 static inline void f2fs_put_dnode(struct dnode_of_data *dn) 2379 { 2380 if (dn->node_page) 2381 f2fs_put_page(dn->node_page, 1); 2382 if (dn->inode_page && dn->node_page != dn->inode_page) 2383 f2fs_put_page(dn->inode_page, 0); 2384 dn->node_page = NULL; 2385 dn->inode_page = NULL; 2386 } 2387 2388 static inline struct kmem_cache *f2fs_kmem_cache_create(const char *name, 2389 size_t size) 2390 { 2391 return kmem_cache_create(name, size, 0, SLAB_RECLAIM_ACCOUNT, NULL); 2392 } 2393 2394 static inline void *f2fs_kmem_cache_alloc(struct kmem_cache *cachep, 2395 gfp_t flags) 2396 { 2397 void *entry; 2398 2399 entry = kmem_cache_alloc(cachep, flags); 2400 if (!entry) 2401 entry = kmem_cache_alloc(cachep, flags | __GFP_NOFAIL); 2402 return entry; 2403 } 2404 2405 static inline bool is_inflight_io(struct f2fs_sb_info *sbi, int type) 2406 { 2407 if (get_pages(sbi, F2FS_RD_DATA) || get_pages(sbi, F2FS_RD_NODE) || 2408 get_pages(sbi, F2FS_RD_META) || get_pages(sbi, F2FS_WB_DATA) || 2409 get_pages(sbi, F2FS_WB_CP_DATA) || 2410 get_pages(sbi, F2FS_DIO_READ) || 2411 get_pages(sbi, F2FS_DIO_WRITE)) 2412 return true; 2413 2414 if (type != DISCARD_TIME && SM_I(sbi) && SM_I(sbi)->dcc_info && 2415 atomic_read(&SM_I(sbi)->dcc_info->queued_discard)) 2416 return true; 2417 2418 if (SM_I(sbi) && SM_I(sbi)->fcc_info && 2419 atomic_read(&SM_I(sbi)->fcc_info->queued_flush)) 2420 return true; 2421 return false; 2422 } 2423 2424 static inline bool is_idle(struct f2fs_sb_info *sbi, int type) 2425 { 2426 if (sbi->gc_mode == GC_URGENT_HIGH) 2427 return true; 2428 2429 if (is_inflight_io(sbi, type)) 2430 return false; 2431 2432 if (sbi->gc_mode == GC_URGENT_LOW && 2433 (type == DISCARD_TIME || type == GC_TIME)) 2434 return true; 2435 2436 return f2fs_time_over(sbi, type); 2437 } 2438 2439 static inline void f2fs_radix_tree_insert(struct radix_tree_root *root, 2440 unsigned long index, void *item) 2441 { 2442 while (radix_tree_insert(root, index, item)) 2443 cond_resched(); 2444 } 2445 2446 #define RAW_IS_INODE(p) ((p)->footer.nid == (p)->footer.ino) 2447 2448 static inline bool IS_INODE(struct page *page) 2449 { 2450 struct f2fs_node *p = F2FS_NODE(page); 2451 2452 return RAW_IS_INODE(p); 2453 } 2454 2455 static inline int offset_in_addr(struct f2fs_inode *i) 2456 { 2457 return (i->i_inline & F2FS_EXTRA_ATTR) ? 2458 (le16_to_cpu(i->i_extra_isize) / sizeof(__le32)) : 0; 2459 } 2460 2461 static inline __le32 *blkaddr_in_node(struct f2fs_node *node) 2462 { 2463 return RAW_IS_INODE(node) ? node->i.i_addr : node->dn.addr; 2464 } 2465 2466 static inline int f2fs_has_extra_attr(struct inode *inode); 2467 static inline block_t data_blkaddr(struct inode *inode, 2468 struct page *node_page, unsigned int offset) 2469 { 2470 struct f2fs_node *raw_node; 2471 __le32 *addr_array; 2472 int base = 0; 2473 bool is_inode = IS_INODE(node_page); 2474 2475 raw_node = F2FS_NODE(node_page); 2476 2477 if (is_inode) { 2478 if (!inode) 2479 /* from GC path only */ 2480 base = offset_in_addr(&raw_node->i); 2481 else if (f2fs_has_extra_attr(inode)) 2482 base = get_extra_isize(inode); 2483 } 2484 2485 addr_array = blkaddr_in_node(raw_node); 2486 return le32_to_cpu(addr_array[base + offset]); 2487 } 2488 2489 static inline block_t f2fs_data_blkaddr(struct dnode_of_data *dn) 2490 { 2491 return data_blkaddr(dn->inode, dn->node_page, dn->ofs_in_node); 2492 } 2493 2494 static inline int f2fs_test_bit(unsigned int nr, char *addr) 2495 { 2496 int mask; 2497 2498 addr += (nr >> 3); 2499 mask = 1 << (7 - (nr & 0x07)); 2500 return mask & *addr; 2501 } 2502 2503 static inline void f2fs_set_bit(unsigned int nr, char *addr) 2504 { 2505 int mask; 2506 2507 addr += (nr >> 3); 2508 mask = 1 << (7 - (nr & 0x07)); 2509 *addr |= mask; 2510 } 2511 2512 static inline void f2fs_clear_bit(unsigned int nr, char *addr) 2513 { 2514 int mask; 2515 2516 addr += (nr >> 3); 2517 mask = 1 << (7 - (nr & 0x07)); 2518 *addr &= ~mask; 2519 } 2520 2521 static inline int f2fs_test_and_set_bit(unsigned int nr, char *addr) 2522 { 2523 int mask; 2524 int ret; 2525 2526 addr += (nr >> 3); 2527 mask = 1 << (7 - (nr & 0x07)); 2528 ret = mask & *addr; 2529 *addr |= mask; 2530 return ret; 2531 } 2532 2533 static inline int f2fs_test_and_clear_bit(unsigned int nr, char *addr) 2534 { 2535 int mask; 2536 int ret; 2537 2538 addr += (nr >> 3); 2539 mask = 1 << (7 - (nr & 0x07)); 2540 ret = mask & *addr; 2541 *addr &= ~mask; 2542 return ret; 2543 } 2544 2545 static inline void f2fs_change_bit(unsigned int nr, char *addr) 2546 { 2547 int mask; 2548 2549 addr += (nr >> 3); 2550 mask = 1 << (7 - (nr & 0x07)); 2551 *addr ^= mask; 2552 } 2553 2554 /* 2555 * On-disk inode flags (f2fs_inode::i_flags) 2556 */ 2557 #define F2FS_COMPR_FL 0x00000004 /* Compress file */ 2558 #define F2FS_SYNC_FL 0x00000008 /* Synchronous updates */ 2559 #define F2FS_IMMUTABLE_FL 0x00000010 /* Immutable file */ 2560 #define F2FS_APPEND_FL 0x00000020 /* writes to file may only append */ 2561 #define F2FS_NODUMP_FL 0x00000040 /* do not dump file */ 2562 #define F2FS_NOATIME_FL 0x00000080 /* do not update atime */ 2563 #define F2FS_NOCOMP_FL 0x00000400 /* Don't compress */ 2564 #define F2FS_INDEX_FL 0x00001000 /* hash-indexed directory */ 2565 #define F2FS_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */ 2566 #define F2FS_PROJINHERIT_FL 0x20000000 /* Create with parents projid */ 2567 #define F2FS_CASEFOLD_FL 0x40000000 /* Casefolded file */ 2568 2569 /* Flags that should be inherited by new inodes from their parent. */ 2570 #define F2FS_FL_INHERITED (F2FS_SYNC_FL | F2FS_NODUMP_FL | F2FS_NOATIME_FL | \ 2571 F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \ 2572 F2FS_CASEFOLD_FL | F2FS_COMPR_FL | F2FS_NOCOMP_FL) 2573 2574 /* Flags that are appropriate for regular files (all but dir-specific ones). */ 2575 #define F2FS_REG_FLMASK (~(F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \ 2576 F2FS_CASEFOLD_FL)) 2577 2578 /* Flags that are appropriate for non-directories/regular files. */ 2579 #define F2FS_OTHER_FLMASK (F2FS_NODUMP_FL | F2FS_NOATIME_FL) 2580 2581 static inline __u32 f2fs_mask_flags(umode_t mode, __u32 flags) 2582 { 2583 if (S_ISDIR(mode)) 2584 return flags; 2585 else if (S_ISREG(mode)) 2586 return flags & F2FS_REG_FLMASK; 2587 else 2588 return flags & F2FS_OTHER_FLMASK; 2589 } 2590 2591 static inline void __mark_inode_dirty_flag(struct inode *inode, 2592 int flag, bool set) 2593 { 2594 switch (flag) { 2595 case FI_INLINE_XATTR: 2596 case FI_INLINE_DATA: 2597 case FI_INLINE_DENTRY: 2598 case FI_NEW_INODE: 2599 if (set) 2600 return; 2601 fallthrough; 2602 case FI_DATA_EXIST: 2603 case FI_INLINE_DOTS: 2604 case FI_PIN_FILE: 2605 f2fs_mark_inode_dirty_sync(inode, true); 2606 } 2607 } 2608 2609 static inline void set_inode_flag(struct inode *inode, int flag) 2610 { 2611 set_bit(flag, F2FS_I(inode)->flags); 2612 __mark_inode_dirty_flag(inode, flag, true); 2613 } 2614 2615 static inline int is_inode_flag_set(struct inode *inode, int flag) 2616 { 2617 return test_bit(flag, F2FS_I(inode)->flags); 2618 } 2619 2620 static inline void clear_inode_flag(struct inode *inode, int flag) 2621 { 2622 clear_bit(flag, F2FS_I(inode)->flags); 2623 __mark_inode_dirty_flag(inode, flag, false); 2624 } 2625 2626 static inline bool f2fs_verity_in_progress(struct inode *inode) 2627 { 2628 return IS_ENABLED(CONFIG_FS_VERITY) && 2629 is_inode_flag_set(inode, FI_VERITY_IN_PROGRESS); 2630 } 2631 2632 static inline void set_acl_inode(struct inode *inode, umode_t mode) 2633 { 2634 F2FS_I(inode)->i_acl_mode = mode; 2635 set_inode_flag(inode, FI_ACL_MODE); 2636 f2fs_mark_inode_dirty_sync(inode, false); 2637 } 2638 2639 static inline void f2fs_i_links_write(struct inode *inode, bool inc) 2640 { 2641 if (inc) 2642 inc_nlink(inode); 2643 else 2644 drop_nlink(inode); 2645 f2fs_mark_inode_dirty_sync(inode, true); 2646 } 2647 2648 static inline void f2fs_i_blocks_write(struct inode *inode, 2649 block_t diff, bool add, bool claim) 2650 { 2651 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE); 2652 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER); 2653 2654 /* add = 1, claim = 1 should be dquot_reserve_block in pair */ 2655 if (add) { 2656 if (claim) 2657 dquot_claim_block(inode, diff); 2658 else 2659 dquot_alloc_block_nofail(inode, diff); 2660 } else { 2661 dquot_free_block(inode, diff); 2662 } 2663 2664 f2fs_mark_inode_dirty_sync(inode, true); 2665 if (clean || recover) 2666 set_inode_flag(inode, FI_AUTO_RECOVER); 2667 } 2668 2669 static inline void f2fs_i_size_write(struct inode *inode, loff_t i_size) 2670 { 2671 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE); 2672 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER); 2673 2674 if (i_size_read(inode) == i_size) 2675 return; 2676 2677 i_size_write(inode, i_size); 2678 f2fs_mark_inode_dirty_sync(inode, true); 2679 if (clean || recover) 2680 set_inode_flag(inode, FI_AUTO_RECOVER); 2681 } 2682 2683 static inline void f2fs_i_depth_write(struct inode *inode, unsigned int depth) 2684 { 2685 F2FS_I(inode)->i_current_depth = depth; 2686 f2fs_mark_inode_dirty_sync(inode, true); 2687 } 2688 2689 static inline void f2fs_i_gc_failures_write(struct inode *inode, 2690 unsigned int count) 2691 { 2692 F2FS_I(inode)->i_gc_failures[GC_FAILURE_PIN] = count; 2693 f2fs_mark_inode_dirty_sync(inode, true); 2694 } 2695 2696 static inline void f2fs_i_xnid_write(struct inode *inode, nid_t xnid) 2697 { 2698 F2FS_I(inode)->i_xattr_nid = xnid; 2699 f2fs_mark_inode_dirty_sync(inode, true); 2700 } 2701 2702 static inline void f2fs_i_pino_write(struct inode *inode, nid_t pino) 2703 { 2704 F2FS_I(inode)->i_pino = pino; 2705 f2fs_mark_inode_dirty_sync(inode, true); 2706 } 2707 2708 static inline void get_inline_info(struct inode *inode, struct f2fs_inode *ri) 2709 { 2710 struct f2fs_inode_info *fi = F2FS_I(inode); 2711 2712 if (ri->i_inline & F2FS_INLINE_XATTR) 2713 set_bit(FI_INLINE_XATTR, fi->flags); 2714 if (ri->i_inline & F2FS_INLINE_DATA) 2715 set_bit(FI_INLINE_DATA, fi->flags); 2716 if (ri->i_inline & F2FS_INLINE_DENTRY) 2717 set_bit(FI_INLINE_DENTRY, fi->flags); 2718 if (ri->i_inline & F2FS_DATA_EXIST) 2719 set_bit(FI_DATA_EXIST, fi->flags); 2720 if (ri->i_inline & F2FS_INLINE_DOTS) 2721 set_bit(FI_INLINE_DOTS, fi->flags); 2722 if (ri->i_inline & F2FS_EXTRA_ATTR) 2723 set_bit(FI_EXTRA_ATTR, fi->flags); 2724 if (ri->i_inline & F2FS_PIN_FILE) 2725 set_bit(FI_PIN_FILE, fi->flags); 2726 } 2727 2728 static inline void set_raw_inline(struct inode *inode, struct f2fs_inode *ri) 2729 { 2730 ri->i_inline = 0; 2731 2732 if (is_inode_flag_set(inode, FI_INLINE_XATTR)) 2733 ri->i_inline |= F2FS_INLINE_XATTR; 2734 if (is_inode_flag_set(inode, FI_INLINE_DATA)) 2735 ri->i_inline |= F2FS_INLINE_DATA; 2736 if (is_inode_flag_set(inode, FI_INLINE_DENTRY)) 2737 ri->i_inline |= F2FS_INLINE_DENTRY; 2738 if (is_inode_flag_set(inode, FI_DATA_EXIST)) 2739 ri->i_inline |= F2FS_DATA_EXIST; 2740 if (is_inode_flag_set(inode, FI_INLINE_DOTS)) 2741 ri->i_inline |= F2FS_INLINE_DOTS; 2742 if (is_inode_flag_set(inode, FI_EXTRA_ATTR)) 2743 ri->i_inline |= F2FS_EXTRA_ATTR; 2744 if (is_inode_flag_set(inode, FI_PIN_FILE)) 2745 ri->i_inline |= F2FS_PIN_FILE; 2746 } 2747 2748 static inline int f2fs_has_extra_attr(struct inode *inode) 2749 { 2750 return is_inode_flag_set(inode, FI_EXTRA_ATTR); 2751 } 2752 2753 static inline int f2fs_has_inline_xattr(struct inode *inode) 2754 { 2755 return is_inode_flag_set(inode, FI_INLINE_XATTR); 2756 } 2757 2758 static inline int f2fs_compressed_file(struct inode *inode) 2759 { 2760 return S_ISREG(inode->i_mode) && 2761 is_inode_flag_set(inode, FI_COMPRESSED_FILE); 2762 } 2763 2764 static inline unsigned int addrs_per_inode(struct inode *inode) 2765 { 2766 unsigned int addrs = CUR_ADDRS_PER_INODE(inode) - 2767 get_inline_xattr_addrs(inode); 2768 2769 if (!f2fs_compressed_file(inode)) 2770 return addrs; 2771 return ALIGN_DOWN(addrs, F2FS_I(inode)->i_cluster_size); 2772 } 2773 2774 static inline unsigned int addrs_per_block(struct inode *inode) 2775 { 2776 if (!f2fs_compressed_file(inode)) 2777 return DEF_ADDRS_PER_BLOCK; 2778 return ALIGN_DOWN(DEF_ADDRS_PER_BLOCK, F2FS_I(inode)->i_cluster_size); 2779 } 2780 2781 static inline void *inline_xattr_addr(struct inode *inode, struct page *page) 2782 { 2783 struct f2fs_inode *ri = F2FS_INODE(page); 2784 2785 return (void *)&(ri->i_addr[DEF_ADDRS_PER_INODE - 2786 get_inline_xattr_addrs(inode)]); 2787 } 2788 2789 static inline int inline_xattr_size(struct inode *inode) 2790 { 2791 if (f2fs_has_inline_xattr(inode)) 2792 return get_inline_xattr_addrs(inode) * sizeof(__le32); 2793 return 0; 2794 } 2795 2796 static inline int f2fs_has_inline_data(struct inode *inode) 2797 { 2798 return is_inode_flag_set(inode, FI_INLINE_DATA); 2799 } 2800 2801 static inline int f2fs_exist_data(struct inode *inode) 2802 { 2803 return is_inode_flag_set(inode, FI_DATA_EXIST); 2804 } 2805 2806 static inline int f2fs_has_inline_dots(struct inode *inode) 2807 { 2808 return is_inode_flag_set(inode, FI_INLINE_DOTS); 2809 } 2810 2811 static inline int f2fs_is_mmap_file(struct inode *inode) 2812 { 2813 return is_inode_flag_set(inode, FI_MMAP_FILE); 2814 } 2815 2816 static inline bool f2fs_is_pinned_file(struct inode *inode) 2817 { 2818 return is_inode_flag_set(inode, FI_PIN_FILE); 2819 } 2820 2821 static inline bool f2fs_is_atomic_file(struct inode *inode) 2822 { 2823 return is_inode_flag_set(inode, FI_ATOMIC_FILE); 2824 } 2825 2826 static inline bool f2fs_is_commit_atomic_write(struct inode *inode) 2827 { 2828 return is_inode_flag_set(inode, FI_ATOMIC_COMMIT); 2829 } 2830 2831 static inline bool f2fs_is_volatile_file(struct inode *inode) 2832 { 2833 return is_inode_flag_set(inode, FI_VOLATILE_FILE); 2834 } 2835 2836 static inline bool f2fs_is_first_block_written(struct inode *inode) 2837 { 2838 return is_inode_flag_set(inode, FI_FIRST_BLOCK_WRITTEN); 2839 } 2840 2841 static inline bool f2fs_is_drop_cache(struct inode *inode) 2842 { 2843 return is_inode_flag_set(inode, FI_DROP_CACHE); 2844 } 2845 2846 static inline void *inline_data_addr(struct inode *inode, struct page *page) 2847 { 2848 struct f2fs_inode *ri = F2FS_INODE(page); 2849 int extra_size = get_extra_isize(inode); 2850 2851 return (void *)&(ri->i_addr[extra_size + DEF_INLINE_RESERVED_SIZE]); 2852 } 2853 2854 static inline int f2fs_has_inline_dentry(struct inode *inode) 2855 { 2856 return is_inode_flag_set(inode, FI_INLINE_DENTRY); 2857 } 2858 2859 static inline int is_file(struct inode *inode, int type) 2860 { 2861 return F2FS_I(inode)->i_advise & type; 2862 } 2863 2864 static inline void set_file(struct inode *inode, int type) 2865 { 2866 F2FS_I(inode)->i_advise |= type; 2867 f2fs_mark_inode_dirty_sync(inode, true); 2868 } 2869 2870 static inline void clear_file(struct inode *inode, int type) 2871 { 2872 F2FS_I(inode)->i_advise &= ~type; 2873 f2fs_mark_inode_dirty_sync(inode, true); 2874 } 2875 2876 static inline bool f2fs_is_time_consistent(struct inode *inode) 2877 { 2878 if (!timespec64_equal(F2FS_I(inode)->i_disk_time, &inode->i_atime)) 2879 return false; 2880 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 1, &inode->i_ctime)) 2881 return false; 2882 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 2, &inode->i_mtime)) 2883 return false; 2884 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 3, 2885 &F2FS_I(inode)->i_crtime)) 2886 return false; 2887 return true; 2888 } 2889 2890 static inline bool f2fs_skip_inode_update(struct inode *inode, int dsync) 2891 { 2892 bool ret; 2893 2894 if (dsync) { 2895 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 2896 2897 spin_lock(&sbi->inode_lock[DIRTY_META]); 2898 ret = list_empty(&F2FS_I(inode)->gdirty_list); 2899 spin_unlock(&sbi->inode_lock[DIRTY_META]); 2900 return ret; 2901 } 2902 if (!is_inode_flag_set(inode, FI_AUTO_RECOVER) || 2903 file_keep_isize(inode) || 2904 i_size_read(inode) & ~PAGE_MASK) 2905 return false; 2906 2907 if (!f2fs_is_time_consistent(inode)) 2908 return false; 2909 2910 spin_lock(&F2FS_I(inode)->i_size_lock); 2911 ret = F2FS_I(inode)->last_disk_size == i_size_read(inode); 2912 spin_unlock(&F2FS_I(inode)->i_size_lock); 2913 2914 return ret; 2915 } 2916 2917 static inline bool f2fs_readonly(struct super_block *sb) 2918 { 2919 return sb_rdonly(sb); 2920 } 2921 2922 static inline bool f2fs_cp_error(struct f2fs_sb_info *sbi) 2923 { 2924 return is_set_ckpt_flags(sbi, CP_ERROR_FLAG); 2925 } 2926 2927 static inline bool is_dot_dotdot(const u8 *name, size_t len) 2928 { 2929 if (len == 1 && name[0] == '.') 2930 return true; 2931 2932 if (len == 2 && name[0] == '.' && name[1] == '.') 2933 return true; 2934 2935 return false; 2936 } 2937 2938 static inline bool f2fs_may_extent_tree(struct inode *inode) 2939 { 2940 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 2941 2942 if (!test_opt(sbi, EXTENT_CACHE) || 2943 is_inode_flag_set(inode, FI_NO_EXTENT) || 2944 is_inode_flag_set(inode, FI_COMPRESSED_FILE)) 2945 return false; 2946 2947 /* 2948 * for recovered files during mount do not create extents 2949 * if shrinker is not registered. 2950 */ 2951 if (list_empty(&sbi->s_list)) 2952 return false; 2953 2954 return S_ISREG(inode->i_mode); 2955 } 2956 2957 static inline void *f2fs_kmalloc(struct f2fs_sb_info *sbi, 2958 size_t size, gfp_t flags) 2959 { 2960 if (time_to_inject(sbi, FAULT_KMALLOC)) { 2961 f2fs_show_injection_info(sbi, FAULT_KMALLOC); 2962 return NULL; 2963 } 2964 2965 return kmalloc(size, flags); 2966 } 2967 2968 static inline void *f2fs_kzalloc(struct f2fs_sb_info *sbi, 2969 size_t size, gfp_t flags) 2970 { 2971 return f2fs_kmalloc(sbi, size, flags | __GFP_ZERO); 2972 } 2973 2974 static inline void *f2fs_kvmalloc(struct f2fs_sb_info *sbi, 2975 size_t size, gfp_t flags) 2976 { 2977 if (time_to_inject(sbi, FAULT_KVMALLOC)) { 2978 f2fs_show_injection_info(sbi, FAULT_KVMALLOC); 2979 return NULL; 2980 } 2981 2982 return kvmalloc(size, flags); 2983 } 2984 2985 static inline void *f2fs_kvzalloc(struct f2fs_sb_info *sbi, 2986 size_t size, gfp_t flags) 2987 { 2988 return f2fs_kvmalloc(sbi, size, flags | __GFP_ZERO); 2989 } 2990 2991 static inline int get_extra_isize(struct inode *inode) 2992 { 2993 return F2FS_I(inode)->i_extra_isize / sizeof(__le32); 2994 } 2995 2996 static inline int get_inline_xattr_addrs(struct inode *inode) 2997 { 2998 return F2FS_I(inode)->i_inline_xattr_size; 2999 } 3000 3001 #define f2fs_get_inode_mode(i) \ 3002 ((is_inode_flag_set(i, FI_ACL_MODE)) ? \ 3003 (F2FS_I(i)->i_acl_mode) : ((i)->i_mode)) 3004 3005 #define F2FS_TOTAL_EXTRA_ATTR_SIZE \ 3006 (offsetof(struct f2fs_inode, i_extra_end) - \ 3007 offsetof(struct f2fs_inode, i_extra_isize)) \ 3008 3009 #define F2FS_OLD_ATTRIBUTE_SIZE (offsetof(struct f2fs_inode, i_addr)) 3010 #define F2FS_FITS_IN_INODE(f2fs_inode, extra_isize, field) \ 3011 ((offsetof(typeof(*(f2fs_inode)), field) + \ 3012 sizeof((f2fs_inode)->field)) \ 3013 <= (F2FS_OLD_ATTRIBUTE_SIZE + (extra_isize))) \ 3014 3015 #define DEFAULT_IOSTAT_PERIOD_MS 3000 3016 #define MIN_IOSTAT_PERIOD_MS 100 3017 /* maximum period of iostat tracing is 1 day */ 3018 #define MAX_IOSTAT_PERIOD_MS 8640000 3019 3020 static inline void f2fs_reset_iostat(struct f2fs_sb_info *sbi) 3021 { 3022 int i; 3023 3024 spin_lock(&sbi->iostat_lock); 3025 for (i = 0; i < NR_IO_TYPE; i++) { 3026 sbi->rw_iostat[i] = 0; 3027 sbi->prev_rw_iostat[i] = 0; 3028 } 3029 spin_unlock(&sbi->iostat_lock); 3030 } 3031 3032 extern void f2fs_record_iostat(struct f2fs_sb_info *sbi); 3033 3034 static inline void f2fs_update_iostat(struct f2fs_sb_info *sbi, 3035 enum iostat_type type, unsigned long long io_bytes) 3036 { 3037 if (!sbi->iostat_enable) 3038 return; 3039 spin_lock(&sbi->iostat_lock); 3040 sbi->rw_iostat[type] += io_bytes; 3041 3042 if (type == APP_WRITE_IO || type == APP_DIRECT_IO) 3043 sbi->rw_iostat[APP_BUFFERED_IO] = 3044 sbi->rw_iostat[APP_WRITE_IO] - 3045 sbi->rw_iostat[APP_DIRECT_IO]; 3046 3047 if (type == APP_READ_IO || type == APP_DIRECT_READ_IO) 3048 sbi->rw_iostat[APP_BUFFERED_READ_IO] = 3049 sbi->rw_iostat[APP_READ_IO] - 3050 sbi->rw_iostat[APP_DIRECT_READ_IO]; 3051 spin_unlock(&sbi->iostat_lock); 3052 3053 f2fs_record_iostat(sbi); 3054 } 3055 3056 #define __is_large_section(sbi) ((sbi)->segs_per_sec > 1) 3057 3058 #define __is_meta_io(fio) (PAGE_TYPE_OF_BIO((fio)->type) == META) 3059 3060 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi, 3061 block_t blkaddr, int type); 3062 static inline void verify_blkaddr(struct f2fs_sb_info *sbi, 3063 block_t blkaddr, int type) 3064 { 3065 if (!f2fs_is_valid_blkaddr(sbi, blkaddr, type)) { 3066 f2fs_err(sbi, "invalid blkaddr: %u, type: %d, run fsck to fix.", 3067 blkaddr, type); 3068 f2fs_bug_on(sbi, 1); 3069 } 3070 } 3071 3072 static inline bool __is_valid_data_blkaddr(block_t blkaddr) 3073 { 3074 if (blkaddr == NEW_ADDR || blkaddr == NULL_ADDR || 3075 blkaddr == COMPRESS_ADDR) 3076 return false; 3077 return true; 3078 } 3079 3080 static inline void f2fs_set_page_private(struct page *page, 3081 unsigned long data) 3082 { 3083 if (PagePrivate(page)) 3084 return; 3085 3086 attach_page_private(page, (void *)data); 3087 } 3088 3089 static inline void f2fs_clear_page_private(struct page *page) 3090 { 3091 detach_page_private(page); 3092 } 3093 3094 /* 3095 * file.c 3096 */ 3097 int f2fs_sync_file(struct file *file, loff_t start, loff_t end, int datasync); 3098 void f2fs_truncate_data_blocks(struct dnode_of_data *dn); 3099 int f2fs_do_truncate_blocks(struct inode *inode, u64 from, bool lock); 3100 int f2fs_truncate_blocks(struct inode *inode, u64 from, bool lock); 3101 int f2fs_truncate(struct inode *inode); 3102 int f2fs_getattr(const struct path *path, struct kstat *stat, 3103 u32 request_mask, unsigned int flags); 3104 int f2fs_setattr(struct dentry *dentry, struct iattr *attr); 3105 int f2fs_truncate_hole(struct inode *inode, pgoff_t pg_start, pgoff_t pg_end); 3106 void f2fs_truncate_data_blocks_range(struct dnode_of_data *dn, int count); 3107 int f2fs_precache_extents(struct inode *inode); 3108 long f2fs_ioctl(struct file *filp, unsigned int cmd, unsigned long arg); 3109 long f2fs_compat_ioctl(struct file *file, unsigned int cmd, unsigned long arg); 3110 int f2fs_transfer_project_quota(struct inode *inode, kprojid_t kprojid); 3111 int f2fs_pin_file_control(struct inode *inode, bool inc); 3112 3113 /* 3114 * inode.c 3115 */ 3116 void f2fs_set_inode_flags(struct inode *inode); 3117 bool f2fs_inode_chksum_verify(struct f2fs_sb_info *sbi, struct page *page); 3118 void f2fs_inode_chksum_set(struct f2fs_sb_info *sbi, struct page *page); 3119 struct inode *f2fs_iget(struct super_block *sb, unsigned long ino); 3120 struct inode *f2fs_iget_retry(struct super_block *sb, unsigned long ino); 3121 int f2fs_try_to_free_nats(struct f2fs_sb_info *sbi, int nr_shrink); 3122 void f2fs_update_inode(struct inode *inode, struct page *node_page); 3123 void f2fs_update_inode_page(struct inode *inode); 3124 int f2fs_write_inode(struct inode *inode, struct writeback_control *wbc); 3125 void f2fs_evict_inode(struct inode *inode); 3126 void f2fs_handle_failed_inode(struct inode *inode); 3127 3128 /* 3129 * namei.c 3130 */ 3131 int f2fs_update_extension_list(struct f2fs_sb_info *sbi, const char *name, 3132 bool hot, bool set); 3133 struct dentry *f2fs_get_parent(struct dentry *child); 3134 3135 /* 3136 * dir.c 3137 */ 3138 unsigned char f2fs_get_de_type(struct f2fs_dir_entry *de); 3139 int f2fs_init_casefolded_name(const struct inode *dir, 3140 struct f2fs_filename *fname); 3141 int f2fs_setup_filename(struct inode *dir, const struct qstr *iname, 3142 int lookup, struct f2fs_filename *fname); 3143 int f2fs_prepare_lookup(struct inode *dir, struct dentry *dentry, 3144 struct f2fs_filename *fname); 3145 void f2fs_free_filename(struct f2fs_filename *fname); 3146 struct f2fs_dir_entry *f2fs_find_target_dentry(const struct f2fs_dentry_ptr *d, 3147 const struct f2fs_filename *fname, int *max_slots); 3148 int f2fs_fill_dentries(struct dir_context *ctx, struct f2fs_dentry_ptr *d, 3149 unsigned int start_pos, struct fscrypt_str *fstr); 3150 void f2fs_do_make_empty_dir(struct inode *inode, struct inode *parent, 3151 struct f2fs_dentry_ptr *d); 3152 struct page *f2fs_init_inode_metadata(struct inode *inode, struct inode *dir, 3153 const struct f2fs_filename *fname, struct page *dpage); 3154 void f2fs_update_parent_metadata(struct inode *dir, struct inode *inode, 3155 unsigned int current_depth); 3156 int f2fs_room_for_filename(const void *bitmap, int slots, int max_slots); 3157 void f2fs_drop_nlink(struct inode *dir, struct inode *inode); 3158 struct f2fs_dir_entry *__f2fs_find_entry(struct inode *dir, 3159 const struct f2fs_filename *fname, 3160 struct page **res_page); 3161 struct f2fs_dir_entry *f2fs_find_entry(struct inode *dir, 3162 const struct qstr *child, struct page **res_page); 3163 struct f2fs_dir_entry *f2fs_parent_dir(struct inode *dir, struct page **p); 3164 ino_t f2fs_inode_by_name(struct inode *dir, const struct qstr *qstr, 3165 struct page **page); 3166 void f2fs_set_link(struct inode *dir, struct f2fs_dir_entry *de, 3167 struct page *page, struct inode *inode); 3168 bool f2fs_has_enough_room(struct inode *dir, struct page *ipage, 3169 const struct f2fs_filename *fname); 3170 void f2fs_update_dentry(nid_t ino, umode_t mode, struct f2fs_dentry_ptr *d, 3171 const struct fscrypt_str *name, f2fs_hash_t name_hash, 3172 unsigned int bit_pos); 3173 int f2fs_add_regular_entry(struct inode *dir, const struct f2fs_filename *fname, 3174 struct inode *inode, nid_t ino, umode_t mode); 3175 int f2fs_add_dentry(struct inode *dir, const struct f2fs_filename *fname, 3176 struct inode *inode, nid_t ino, umode_t mode); 3177 int f2fs_do_add_link(struct inode *dir, const struct qstr *name, 3178 struct inode *inode, nid_t ino, umode_t mode); 3179 void f2fs_delete_entry(struct f2fs_dir_entry *dentry, struct page *page, 3180 struct inode *dir, struct inode *inode); 3181 int f2fs_do_tmpfile(struct inode *inode, struct inode *dir); 3182 bool f2fs_empty_dir(struct inode *dir); 3183 3184 static inline int f2fs_add_link(struct dentry *dentry, struct inode *inode) 3185 { 3186 return f2fs_do_add_link(d_inode(dentry->d_parent), &dentry->d_name, 3187 inode, inode->i_ino, inode->i_mode); 3188 } 3189 3190 /* 3191 * super.c 3192 */ 3193 int f2fs_inode_dirtied(struct inode *inode, bool sync); 3194 void f2fs_inode_synced(struct inode *inode); 3195 int f2fs_enable_quota_files(struct f2fs_sb_info *sbi, bool rdonly); 3196 int f2fs_quota_sync(struct super_block *sb, int type); 3197 void f2fs_quota_off_umount(struct super_block *sb); 3198 int f2fs_commit_super(struct f2fs_sb_info *sbi, bool recover); 3199 int f2fs_sync_fs(struct super_block *sb, int sync); 3200 int f2fs_sanity_check_ckpt(struct f2fs_sb_info *sbi); 3201 3202 /* 3203 * hash.c 3204 */ 3205 void f2fs_hash_filename(const struct inode *dir, struct f2fs_filename *fname); 3206 3207 /* 3208 * node.c 3209 */ 3210 struct dnode_of_data; 3211 struct node_info; 3212 3213 int f2fs_check_nid_range(struct f2fs_sb_info *sbi, nid_t nid); 3214 bool f2fs_available_free_memory(struct f2fs_sb_info *sbi, int type); 3215 bool f2fs_in_warm_node_list(struct f2fs_sb_info *sbi, struct page *page); 3216 void f2fs_init_fsync_node_info(struct f2fs_sb_info *sbi); 3217 void f2fs_del_fsync_node_entry(struct f2fs_sb_info *sbi, struct page *page); 3218 void f2fs_reset_fsync_node_info(struct f2fs_sb_info *sbi); 3219 int f2fs_need_dentry_mark(struct f2fs_sb_info *sbi, nid_t nid); 3220 bool f2fs_is_checkpointed_node(struct f2fs_sb_info *sbi, nid_t nid); 3221 bool f2fs_need_inode_block_update(struct f2fs_sb_info *sbi, nid_t ino); 3222 int f2fs_get_node_info(struct f2fs_sb_info *sbi, nid_t nid, 3223 struct node_info *ni); 3224 pgoff_t f2fs_get_next_page_offset(struct dnode_of_data *dn, pgoff_t pgofs); 3225 int f2fs_get_dnode_of_data(struct dnode_of_data *dn, pgoff_t index, int mode); 3226 int f2fs_truncate_inode_blocks(struct inode *inode, pgoff_t from); 3227 int f2fs_truncate_xattr_node(struct inode *inode); 3228 int f2fs_wait_on_node_pages_writeback(struct f2fs_sb_info *sbi, 3229 unsigned int seq_id); 3230 int f2fs_remove_inode_page(struct inode *inode); 3231 struct page *f2fs_new_inode_page(struct inode *inode); 3232 struct page *f2fs_new_node_page(struct dnode_of_data *dn, unsigned int ofs); 3233 void f2fs_ra_node_page(struct f2fs_sb_info *sbi, nid_t nid); 3234 struct page *f2fs_get_node_page(struct f2fs_sb_info *sbi, pgoff_t nid); 3235 struct page *f2fs_get_node_page_ra(struct page *parent, int start); 3236 int f2fs_move_node_page(struct page *node_page, int gc_type); 3237 void f2fs_flush_inline_data(struct f2fs_sb_info *sbi); 3238 int f2fs_fsync_node_pages(struct f2fs_sb_info *sbi, struct inode *inode, 3239 struct writeback_control *wbc, bool atomic, 3240 unsigned int *seq_id); 3241 int f2fs_sync_node_pages(struct f2fs_sb_info *sbi, 3242 struct writeback_control *wbc, 3243 bool do_balance, enum iostat_type io_type); 3244 int f2fs_build_free_nids(struct f2fs_sb_info *sbi, bool sync, bool mount); 3245 bool f2fs_alloc_nid(struct f2fs_sb_info *sbi, nid_t *nid); 3246 void f2fs_alloc_nid_done(struct f2fs_sb_info *sbi, nid_t nid); 3247 void f2fs_alloc_nid_failed(struct f2fs_sb_info *sbi, nid_t nid); 3248 int f2fs_try_to_free_nids(struct f2fs_sb_info *sbi, int nr_shrink); 3249 int f2fs_recover_inline_xattr(struct inode *inode, struct page *page); 3250 int f2fs_recover_xattr_data(struct inode *inode, struct page *page); 3251 int f2fs_recover_inode_page(struct f2fs_sb_info *sbi, struct page *page); 3252 int f2fs_restore_node_summary(struct f2fs_sb_info *sbi, 3253 unsigned int segno, struct f2fs_summary_block *sum); 3254 int f2fs_flush_nat_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3255 int f2fs_build_node_manager(struct f2fs_sb_info *sbi); 3256 void f2fs_destroy_node_manager(struct f2fs_sb_info *sbi); 3257 int __init f2fs_create_node_manager_caches(void); 3258 void f2fs_destroy_node_manager_caches(void); 3259 3260 /* 3261 * segment.c 3262 */ 3263 bool f2fs_need_SSR(struct f2fs_sb_info *sbi); 3264 void f2fs_register_inmem_page(struct inode *inode, struct page *page); 3265 void f2fs_drop_inmem_pages_all(struct f2fs_sb_info *sbi, bool gc_failure); 3266 void f2fs_drop_inmem_pages(struct inode *inode); 3267 void f2fs_drop_inmem_page(struct inode *inode, struct page *page); 3268 int f2fs_commit_inmem_pages(struct inode *inode); 3269 void f2fs_balance_fs(struct f2fs_sb_info *sbi, bool need); 3270 void f2fs_balance_fs_bg(struct f2fs_sb_info *sbi, bool from_bg); 3271 int f2fs_issue_flush(struct f2fs_sb_info *sbi, nid_t ino); 3272 int f2fs_create_flush_cmd_control(struct f2fs_sb_info *sbi); 3273 int f2fs_flush_device_cache(struct f2fs_sb_info *sbi); 3274 void f2fs_destroy_flush_cmd_control(struct f2fs_sb_info *sbi, bool free); 3275 void f2fs_invalidate_blocks(struct f2fs_sb_info *sbi, block_t addr); 3276 bool f2fs_is_checkpointed_data(struct f2fs_sb_info *sbi, block_t blkaddr); 3277 void f2fs_drop_discard_cmd(struct f2fs_sb_info *sbi); 3278 void f2fs_stop_discard_thread(struct f2fs_sb_info *sbi); 3279 bool f2fs_issue_discard_timeout(struct f2fs_sb_info *sbi); 3280 void f2fs_clear_prefree_segments(struct f2fs_sb_info *sbi, 3281 struct cp_control *cpc); 3282 void f2fs_dirty_to_prefree(struct f2fs_sb_info *sbi); 3283 block_t f2fs_get_unusable_blocks(struct f2fs_sb_info *sbi); 3284 int f2fs_disable_cp_again(struct f2fs_sb_info *sbi, block_t unusable); 3285 void f2fs_release_discard_addrs(struct f2fs_sb_info *sbi); 3286 int f2fs_npages_for_summary_flush(struct f2fs_sb_info *sbi, bool for_ra); 3287 void f2fs_init_inmem_curseg(struct f2fs_sb_info *sbi); 3288 void f2fs_save_inmem_curseg(struct f2fs_sb_info *sbi); 3289 void f2fs_restore_inmem_curseg(struct f2fs_sb_info *sbi); 3290 void f2fs_get_new_segment(struct f2fs_sb_info *sbi, 3291 unsigned int *newseg, bool new_sec, int dir); 3292 void f2fs_allocate_segment_for_resize(struct f2fs_sb_info *sbi, int type, 3293 unsigned int start, unsigned int end); 3294 void f2fs_allocate_new_segment(struct f2fs_sb_info *sbi, int type); 3295 void f2fs_allocate_new_segments(struct f2fs_sb_info *sbi); 3296 int f2fs_trim_fs(struct f2fs_sb_info *sbi, struct fstrim_range *range); 3297 bool f2fs_exist_trim_candidates(struct f2fs_sb_info *sbi, 3298 struct cp_control *cpc); 3299 struct page *f2fs_get_sum_page(struct f2fs_sb_info *sbi, unsigned int segno); 3300 void f2fs_update_meta_page(struct f2fs_sb_info *sbi, void *src, 3301 block_t blk_addr); 3302 void f2fs_do_write_meta_page(struct f2fs_sb_info *sbi, struct page *page, 3303 enum iostat_type io_type); 3304 void f2fs_do_write_node_page(unsigned int nid, struct f2fs_io_info *fio); 3305 void f2fs_outplace_write_data(struct dnode_of_data *dn, 3306 struct f2fs_io_info *fio); 3307 int f2fs_inplace_write_data(struct f2fs_io_info *fio); 3308 void f2fs_do_replace_block(struct f2fs_sb_info *sbi, struct f2fs_summary *sum, 3309 block_t old_blkaddr, block_t new_blkaddr, 3310 bool recover_curseg, bool recover_newaddr, 3311 bool from_gc); 3312 void f2fs_replace_block(struct f2fs_sb_info *sbi, struct dnode_of_data *dn, 3313 block_t old_addr, block_t new_addr, 3314 unsigned char version, bool recover_curseg, 3315 bool recover_newaddr); 3316 void f2fs_allocate_data_block(struct f2fs_sb_info *sbi, struct page *page, 3317 block_t old_blkaddr, block_t *new_blkaddr, 3318 struct f2fs_summary *sum, int type, 3319 struct f2fs_io_info *fio); 3320 void f2fs_wait_on_page_writeback(struct page *page, 3321 enum page_type type, bool ordered, bool locked); 3322 void f2fs_wait_on_block_writeback(struct inode *inode, block_t blkaddr); 3323 void f2fs_wait_on_block_writeback_range(struct inode *inode, block_t blkaddr, 3324 block_t len); 3325 void f2fs_write_data_summaries(struct f2fs_sb_info *sbi, block_t start_blk); 3326 void f2fs_write_node_summaries(struct f2fs_sb_info *sbi, block_t start_blk); 3327 int f2fs_lookup_journal_in_cursum(struct f2fs_journal *journal, int type, 3328 unsigned int val, int alloc); 3329 void f2fs_flush_sit_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3330 int f2fs_fix_curseg_write_pointer(struct f2fs_sb_info *sbi); 3331 int f2fs_check_write_pointer(struct f2fs_sb_info *sbi); 3332 int f2fs_build_segment_manager(struct f2fs_sb_info *sbi); 3333 void f2fs_destroy_segment_manager(struct f2fs_sb_info *sbi); 3334 int __init f2fs_create_segment_manager_caches(void); 3335 void f2fs_destroy_segment_manager_caches(void); 3336 int f2fs_rw_hint_to_seg_type(enum rw_hint hint); 3337 enum rw_hint f2fs_io_type_to_rw_hint(struct f2fs_sb_info *sbi, 3338 enum page_type type, enum temp_type temp); 3339 unsigned int f2fs_usable_segs_in_sec(struct f2fs_sb_info *sbi, 3340 unsigned int segno); 3341 unsigned int f2fs_usable_blks_in_seg(struct f2fs_sb_info *sbi, 3342 unsigned int segno); 3343 3344 /* 3345 * checkpoint.c 3346 */ 3347 void f2fs_stop_checkpoint(struct f2fs_sb_info *sbi, bool end_io); 3348 struct page *f2fs_grab_meta_page(struct f2fs_sb_info *sbi, pgoff_t index); 3349 struct page *f2fs_get_meta_page(struct f2fs_sb_info *sbi, pgoff_t index); 3350 struct page *f2fs_get_meta_page_retry(struct f2fs_sb_info *sbi, pgoff_t index); 3351 struct page *f2fs_get_tmp_page(struct f2fs_sb_info *sbi, pgoff_t index); 3352 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi, 3353 block_t blkaddr, int type); 3354 int f2fs_ra_meta_pages(struct f2fs_sb_info *sbi, block_t start, int nrpages, 3355 int type, bool sync); 3356 void f2fs_ra_meta_pages_cond(struct f2fs_sb_info *sbi, pgoff_t index); 3357 long f2fs_sync_meta_pages(struct f2fs_sb_info *sbi, enum page_type type, 3358 long nr_to_write, enum iostat_type io_type); 3359 void f2fs_add_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type); 3360 void f2fs_remove_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type); 3361 void f2fs_release_ino_entry(struct f2fs_sb_info *sbi, bool all); 3362 bool f2fs_exist_written_data(struct f2fs_sb_info *sbi, nid_t ino, int mode); 3363 void f2fs_set_dirty_device(struct f2fs_sb_info *sbi, nid_t ino, 3364 unsigned int devidx, int type); 3365 bool f2fs_is_dirty_device(struct f2fs_sb_info *sbi, nid_t ino, 3366 unsigned int devidx, int type); 3367 int f2fs_sync_inode_meta(struct f2fs_sb_info *sbi); 3368 int f2fs_acquire_orphan_inode(struct f2fs_sb_info *sbi); 3369 void f2fs_release_orphan_inode(struct f2fs_sb_info *sbi); 3370 void f2fs_add_orphan_inode(struct inode *inode); 3371 void f2fs_remove_orphan_inode(struct f2fs_sb_info *sbi, nid_t ino); 3372 int f2fs_recover_orphan_inodes(struct f2fs_sb_info *sbi); 3373 int f2fs_get_valid_checkpoint(struct f2fs_sb_info *sbi); 3374 void f2fs_update_dirty_page(struct inode *inode, struct page *page); 3375 void f2fs_remove_dirty_inode(struct inode *inode); 3376 int f2fs_sync_dirty_inodes(struct f2fs_sb_info *sbi, enum inode_type type); 3377 void f2fs_wait_on_all_pages(struct f2fs_sb_info *sbi, int type); 3378 u64 f2fs_get_sectors_written(struct f2fs_sb_info *sbi); 3379 int f2fs_write_checkpoint(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3380 void f2fs_init_ino_entry_info(struct f2fs_sb_info *sbi); 3381 int __init f2fs_create_checkpoint_caches(void); 3382 void f2fs_destroy_checkpoint_caches(void); 3383 3384 /* 3385 * data.c 3386 */ 3387 int __init f2fs_init_bioset(void); 3388 void f2fs_destroy_bioset(void); 3389 struct bio *f2fs_bio_alloc(struct f2fs_sb_info *sbi, int npages, bool noio); 3390 int f2fs_init_bio_entry_cache(void); 3391 void f2fs_destroy_bio_entry_cache(void); 3392 void f2fs_submit_bio(struct f2fs_sb_info *sbi, 3393 struct bio *bio, enum page_type type); 3394 void f2fs_submit_merged_write(struct f2fs_sb_info *sbi, enum page_type type); 3395 void f2fs_submit_merged_write_cond(struct f2fs_sb_info *sbi, 3396 struct inode *inode, struct page *page, 3397 nid_t ino, enum page_type type); 3398 void f2fs_submit_merged_ipu_write(struct f2fs_sb_info *sbi, 3399 struct bio **bio, struct page *page); 3400 void f2fs_flush_merged_writes(struct f2fs_sb_info *sbi); 3401 int f2fs_submit_page_bio(struct f2fs_io_info *fio); 3402 int f2fs_merge_page_bio(struct f2fs_io_info *fio); 3403 void f2fs_submit_page_write(struct f2fs_io_info *fio); 3404 struct block_device *f2fs_target_device(struct f2fs_sb_info *sbi, 3405 block_t blk_addr, struct bio *bio); 3406 int f2fs_target_device_index(struct f2fs_sb_info *sbi, block_t blkaddr); 3407 void f2fs_set_data_blkaddr(struct dnode_of_data *dn); 3408 void f2fs_update_data_blkaddr(struct dnode_of_data *dn, block_t blkaddr); 3409 int f2fs_reserve_new_blocks(struct dnode_of_data *dn, blkcnt_t count); 3410 int f2fs_reserve_new_block(struct dnode_of_data *dn); 3411 int f2fs_get_block(struct dnode_of_data *dn, pgoff_t index); 3412 int f2fs_preallocate_blocks(struct kiocb *iocb, struct iov_iter *from); 3413 int f2fs_reserve_block(struct dnode_of_data *dn, pgoff_t index); 3414 struct page *f2fs_get_read_data_page(struct inode *inode, pgoff_t index, 3415 int op_flags, bool for_write); 3416 struct page *f2fs_find_data_page(struct inode *inode, pgoff_t index); 3417 struct page *f2fs_get_lock_data_page(struct inode *inode, pgoff_t index, 3418 bool for_write); 3419 struct page *f2fs_get_new_data_page(struct inode *inode, 3420 struct page *ipage, pgoff_t index, bool new_i_size); 3421 int f2fs_do_write_data_page(struct f2fs_io_info *fio); 3422 void f2fs_do_map_lock(struct f2fs_sb_info *sbi, int flag, bool lock); 3423 int f2fs_map_blocks(struct inode *inode, struct f2fs_map_blocks *map, 3424 int create, int flag); 3425 int f2fs_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, 3426 u64 start, u64 len); 3427 int f2fs_encrypt_one_page(struct f2fs_io_info *fio); 3428 bool f2fs_should_update_inplace(struct inode *inode, struct f2fs_io_info *fio); 3429 bool f2fs_should_update_outplace(struct inode *inode, struct f2fs_io_info *fio); 3430 int f2fs_write_single_data_page(struct page *page, int *submitted, 3431 struct bio **bio, sector_t *last_block, 3432 struct writeback_control *wbc, 3433 enum iostat_type io_type, 3434 int compr_blocks); 3435 void f2fs_invalidate_page(struct page *page, unsigned int offset, 3436 unsigned int length); 3437 int f2fs_release_page(struct page *page, gfp_t wait); 3438 #ifdef CONFIG_MIGRATION 3439 int f2fs_migrate_page(struct address_space *mapping, struct page *newpage, 3440 struct page *page, enum migrate_mode mode); 3441 #endif 3442 bool f2fs_overwrite_io(struct inode *inode, loff_t pos, size_t len); 3443 void f2fs_clear_page_cache_dirty_tag(struct page *page); 3444 int f2fs_init_post_read_processing(void); 3445 void f2fs_destroy_post_read_processing(void); 3446 int f2fs_init_post_read_wq(struct f2fs_sb_info *sbi); 3447 void f2fs_destroy_post_read_wq(struct f2fs_sb_info *sbi); 3448 3449 /* 3450 * gc.c 3451 */ 3452 int f2fs_start_gc_thread(struct f2fs_sb_info *sbi); 3453 void f2fs_stop_gc_thread(struct f2fs_sb_info *sbi); 3454 block_t f2fs_start_bidx_of_node(unsigned int node_ofs, struct inode *inode); 3455 int f2fs_gc(struct f2fs_sb_info *sbi, bool sync, bool background, 3456 unsigned int segno); 3457 void f2fs_build_gc_manager(struct f2fs_sb_info *sbi); 3458 int f2fs_resize_fs(struct f2fs_sb_info *sbi, __u64 block_count); 3459 int __init f2fs_create_garbage_collection_cache(void); 3460 void f2fs_destroy_garbage_collection_cache(void); 3461 3462 /* 3463 * recovery.c 3464 */ 3465 int f2fs_recover_fsync_data(struct f2fs_sb_info *sbi, bool check_only); 3466 bool f2fs_space_for_roll_forward(struct f2fs_sb_info *sbi); 3467 3468 /* 3469 * debug.c 3470 */ 3471 #ifdef CONFIG_F2FS_STAT_FS 3472 struct f2fs_stat_info { 3473 struct list_head stat_list; 3474 struct f2fs_sb_info *sbi; 3475 int all_area_segs, sit_area_segs, nat_area_segs, ssa_area_segs; 3476 int main_area_segs, main_area_sections, main_area_zones; 3477 unsigned long long hit_largest, hit_cached, hit_rbtree; 3478 unsigned long long hit_total, total_ext; 3479 int ext_tree, zombie_tree, ext_node; 3480 int ndirty_node, ndirty_dent, ndirty_meta, ndirty_imeta; 3481 int ndirty_data, ndirty_qdata; 3482 int inmem_pages; 3483 unsigned int ndirty_dirs, ndirty_files, nquota_files, ndirty_all; 3484 int nats, dirty_nats, sits, dirty_sits; 3485 int free_nids, avail_nids, alloc_nids; 3486 int total_count, utilization; 3487 int bg_gc, nr_wb_cp_data, nr_wb_data; 3488 int nr_rd_data, nr_rd_node, nr_rd_meta; 3489 int nr_dio_read, nr_dio_write; 3490 unsigned int io_skip_bggc, other_skip_bggc; 3491 int nr_flushing, nr_flushed, flush_list_empty; 3492 int nr_discarding, nr_discarded; 3493 int nr_discard_cmd; 3494 unsigned int undiscard_blks; 3495 int inline_xattr, inline_inode, inline_dir, append, update, orphans; 3496 int compr_inode; 3497 unsigned long long compr_blocks; 3498 int aw_cnt, max_aw_cnt, vw_cnt, max_vw_cnt; 3499 unsigned int valid_count, valid_node_count, valid_inode_count, discard_blks; 3500 unsigned int bimodal, avg_vblocks; 3501 int util_free, util_valid, util_invalid; 3502 int rsvd_segs, overp_segs; 3503 int dirty_count, node_pages, meta_pages; 3504 int prefree_count, call_count, cp_count, bg_cp_count; 3505 int tot_segs, node_segs, data_segs, free_segs, free_secs; 3506 int bg_node_segs, bg_data_segs; 3507 int tot_blks, data_blks, node_blks; 3508 int bg_data_blks, bg_node_blks; 3509 unsigned long long skipped_atomic_files[2]; 3510 int curseg[NR_CURSEG_TYPE]; 3511 int cursec[NR_CURSEG_TYPE]; 3512 int curzone[NR_CURSEG_TYPE]; 3513 unsigned int dirty_seg[NR_CURSEG_TYPE]; 3514 unsigned int full_seg[NR_CURSEG_TYPE]; 3515 unsigned int valid_blks[NR_CURSEG_TYPE]; 3516 3517 unsigned int meta_count[META_MAX]; 3518 unsigned int segment_count[2]; 3519 unsigned int block_count[2]; 3520 unsigned int inplace_count; 3521 unsigned long long base_mem, cache_mem, page_mem; 3522 }; 3523 3524 static inline struct f2fs_stat_info *F2FS_STAT(struct f2fs_sb_info *sbi) 3525 { 3526 return (struct f2fs_stat_info *)sbi->stat_info; 3527 } 3528 3529 #define stat_inc_cp_count(si) ((si)->cp_count++) 3530 #define stat_inc_bg_cp_count(si) ((si)->bg_cp_count++) 3531 #define stat_inc_call_count(si) ((si)->call_count++) 3532 #define stat_inc_bggc_count(si) ((si)->bg_gc++) 3533 #define stat_io_skip_bggc_count(sbi) ((sbi)->io_skip_bggc++) 3534 #define stat_other_skip_bggc_count(sbi) ((sbi)->other_skip_bggc++) 3535 #define stat_inc_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]++) 3536 #define stat_dec_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]--) 3537 #define stat_inc_total_hit(sbi) (atomic64_inc(&(sbi)->total_hit_ext)) 3538 #define stat_inc_rbtree_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_rbtree)) 3539 #define stat_inc_largest_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_largest)) 3540 #define stat_inc_cached_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_cached)) 3541 #define stat_inc_inline_xattr(inode) \ 3542 do { \ 3543 if (f2fs_has_inline_xattr(inode)) \ 3544 (atomic_inc(&F2FS_I_SB(inode)->inline_xattr)); \ 3545 } while (0) 3546 #define stat_dec_inline_xattr(inode) \ 3547 do { \ 3548 if (f2fs_has_inline_xattr(inode)) \ 3549 (atomic_dec(&F2FS_I_SB(inode)->inline_xattr)); \ 3550 } while (0) 3551 #define stat_inc_inline_inode(inode) \ 3552 do { \ 3553 if (f2fs_has_inline_data(inode)) \ 3554 (atomic_inc(&F2FS_I_SB(inode)->inline_inode)); \ 3555 } while (0) 3556 #define stat_dec_inline_inode(inode) \ 3557 do { \ 3558 if (f2fs_has_inline_data(inode)) \ 3559 (atomic_dec(&F2FS_I_SB(inode)->inline_inode)); \ 3560 } while (0) 3561 #define stat_inc_inline_dir(inode) \ 3562 do { \ 3563 if (f2fs_has_inline_dentry(inode)) \ 3564 (atomic_inc(&F2FS_I_SB(inode)->inline_dir)); \ 3565 } while (0) 3566 #define stat_dec_inline_dir(inode) \ 3567 do { \ 3568 if (f2fs_has_inline_dentry(inode)) \ 3569 (atomic_dec(&F2FS_I_SB(inode)->inline_dir)); \ 3570 } while (0) 3571 #define stat_inc_compr_inode(inode) \ 3572 do { \ 3573 if (f2fs_compressed_file(inode)) \ 3574 (atomic_inc(&F2FS_I_SB(inode)->compr_inode)); \ 3575 } while (0) 3576 #define stat_dec_compr_inode(inode) \ 3577 do { \ 3578 if (f2fs_compressed_file(inode)) \ 3579 (atomic_dec(&F2FS_I_SB(inode)->compr_inode)); \ 3580 } while (0) 3581 #define stat_add_compr_blocks(inode, blocks) \ 3582 (atomic64_add(blocks, &F2FS_I_SB(inode)->compr_blocks)) 3583 #define stat_sub_compr_blocks(inode, blocks) \ 3584 (atomic64_sub(blocks, &F2FS_I_SB(inode)->compr_blocks)) 3585 #define stat_inc_meta_count(sbi, blkaddr) \ 3586 do { \ 3587 if (blkaddr < SIT_I(sbi)->sit_base_addr) \ 3588 atomic_inc(&(sbi)->meta_count[META_CP]); \ 3589 else if (blkaddr < NM_I(sbi)->nat_blkaddr) \ 3590 atomic_inc(&(sbi)->meta_count[META_SIT]); \ 3591 else if (blkaddr < SM_I(sbi)->ssa_blkaddr) \ 3592 atomic_inc(&(sbi)->meta_count[META_NAT]); \ 3593 else if (blkaddr < SM_I(sbi)->main_blkaddr) \ 3594 atomic_inc(&(sbi)->meta_count[META_SSA]); \ 3595 } while (0) 3596 #define stat_inc_seg_type(sbi, curseg) \ 3597 ((sbi)->segment_count[(curseg)->alloc_type]++) 3598 #define stat_inc_block_count(sbi, curseg) \ 3599 ((sbi)->block_count[(curseg)->alloc_type]++) 3600 #define stat_inc_inplace_blocks(sbi) \ 3601 (atomic_inc(&(sbi)->inplace_count)) 3602 #define stat_update_max_atomic_write(inode) \ 3603 do { \ 3604 int cur = F2FS_I_SB(inode)->atomic_files; \ 3605 int max = atomic_read(&F2FS_I_SB(inode)->max_aw_cnt); \ 3606 if (cur > max) \ 3607 atomic_set(&F2FS_I_SB(inode)->max_aw_cnt, cur); \ 3608 } while (0) 3609 #define stat_inc_volatile_write(inode) \ 3610 (atomic_inc(&F2FS_I_SB(inode)->vw_cnt)) 3611 #define stat_dec_volatile_write(inode) \ 3612 (atomic_dec(&F2FS_I_SB(inode)->vw_cnt)) 3613 #define stat_update_max_volatile_write(inode) \ 3614 do { \ 3615 int cur = atomic_read(&F2FS_I_SB(inode)->vw_cnt); \ 3616 int max = atomic_read(&F2FS_I_SB(inode)->max_vw_cnt); \ 3617 if (cur > max) \ 3618 atomic_set(&F2FS_I_SB(inode)->max_vw_cnt, cur); \ 3619 } while (0) 3620 #define stat_inc_seg_count(sbi, type, gc_type) \ 3621 do { \ 3622 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 3623 si->tot_segs++; \ 3624 if ((type) == SUM_TYPE_DATA) { \ 3625 si->data_segs++; \ 3626 si->bg_data_segs += (gc_type == BG_GC) ? 1 : 0; \ 3627 } else { \ 3628 si->node_segs++; \ 3629 si->bg_node_segs += (gc_type == BG_GC) ? 1 : 0; \ 3630 } \ 3631 } while (0) 3632 3633 #define stat_inc_tot_blk_count(si, blks) \ 3634 ((si)->tot_blks += (blks)) 3635 3636 #define stat_inc_data_blk_count(sbi, blks, gc_type) \ 3637 do { \ 3638 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 3639 stat_inc_tot_blk_count(si, blks); \ 3640 si->data_blks += (blks); \ 3641 si->bg_data_blks += ((gc_type) == BG_GC) ? (blks) : 0; \ 3642 } while (0) 3643 3644 #define stat_inc_node_blk_count(sbi, blks, gc_type) \ 3645 do { \ 3646 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 3647 stat_inc_tot_blk_count(si, blks); \ 3648 si->node_blks += (blks); \ 3649 si->bg_node_blks += ((gc_type) == BG_GC) ? (blks) : 0; \ 3650 } while (0) 3651 3652 int f2fs_build_stats(struct f2fs_sb_info *sbi); 3653 void f2fs_destroy_stats(struct f2fs_sb_info *sbi); 3654 void __init f2fs_create_root_stats(void); 3655 void f2fs_destroy_root_stats(void); 3656 void f2fs_update_sit_info(struct f2fs_sb_info *sbi); 3657 #else 3658 #define stat_inc_cp_count(si) do { } while (0) 3659 #define stat_inc_bg_cp_count(si) do { } while (0) 3660 #define stat_inc_call_count(si) do { } while (0) 3661 #define stat_inc_bggc_count(si) do { } while (0) 3662 #define stat_io_skip_bggc_count(sbi) do { } while (0) 3663 #define stat_other_skip_bggc_count(sbi) do { } while (0) 3664 #define stat_inc_dirty_inode(sbi, type) do { } while (0) 3665 #define stat_dec_dirty_inode(sbi, type) do { } while (0) 3666 #define stat_inc_total_hit(sbi) do { } while (0) 3667 #define stat_inc_rbtree_node_hit(sbi) do { } while (0) 3668 #define stat_inc_largest_node_hit(sbi) do { } while (0) 3669 #define stat_inc_cached_node_hit(sbi) do { } while (0) 3670 #define stat_inc_inline_xattr(inode) do { } while (0) 3671 #define stat_dec_inline_xattr(inode) do { } while (0) 3672 #define stat_inc_inline_inode(inode) do { } while (0) 3673 #define stat_dec_inline_inode(inode) do { } while (0) 3674 #define stat_inc_inline_dir(inode) do { } while (0) 3675 #define stat_dec_inline_dir(inode) do { } while (0) 3676 #define stat_inc_compr_inode(inode) do { } while (0) 3677 #define stat_dec_compr_inode(inode) do { } while (0) 3678 #define stat_add_compr_blocks(inode, blocks) do { } while (0) 3679 #define stat_sub_compr_blocks(inode, blocks) do { } while (0) 3680 #define stat_inc_atomic_write(inode) do { } while (0) 3681 #define stat_dec_atomic_write(inode) do { } while (0) 3682 #define stat_update_max_atomic_write(inode) do { } while (0) 3683 #define stat_inc_volatile_write(inode) do { } while (0) 3684 #define stat_dec_volatile_write(inode) do { } while (0) 3685 #define stat_update_max_volatile_write(inode) do { } while (0) 3686 #define stat_inc_meta_count(sbi, blkaddr) do { } while (0) 3687 #define stat_inc_seg_type(sbi, curseg) do { } while (0) 3688 #define stat_inc_block_count(sbi, curseg) do { } while (0) 3689 #define stat_inc_inplace_blocks(sbi) do { } while (0) 3690 #define stat_inc_seg_count(sbi, type, gc_type) do { } while (0) 3691 #define stat_inc_tot_blk_count(si, blks) do { } while (0) 3692 #define stat_inc_data_blk_count(sbi, blks, gc_type) do { } while (0) 3693 #define stat_inc_node_blk_count(sbi, blks, gc_type) do { } while (0) 3694 3695 static inline int f2fs_build_stats(struct f2fs_sb_info *sbi) { return 0; } 3696 static inline void f2fs_destroy_stats(struct f2fs_sb_info *sbi) { } 3697 static inline void __init f2fs_create_root_stats(void) { } 3698 static inline void f2fs_destroy_root_stats(void) { } 3699 static inline void f2fs_update_sit_info(struct f2fs_sb_info *sbi) {} 3700 #endif 3701 3702 extern const struct file_operations f2fs_dir_operations; 3703 extern const struct file_operations f2fs_file_operations; 3704 extern const struct inode_operations f2fs_file_inode_operations; 3705 extern const struct address_space_operations f2fs_dblock_aops; 3706 extern const struct address_space_operations f2fs_node_aops; 3707 extern const struct address_space_operations f2fs_meta_aops; 3708 extern const struct inode_operations f2fs_dir_inode_operations; 3709 extern const struct inode_operations f2fs_symlink_inode_operations; 3710 extern const struct inode_operations f2fs_encrypted_symlink_inode_operations; 3711 extern const struct inode_operations f2fs_special_inode_operations; 3712 extern struct kmem_cache *f2fs_inode_entry_slab; 3713 3714 /* 3715 * inline.c 3716 */ 3717 bool f2fs_may_inline_data(struct inode *inode); 3718 bool f2fs_may_inline_dentry(struct inode *inode); 3719 void f2fs_do_read_inline_data(struct page *page, struct page *ipage); 3720 void f2fs_truncate_inline_inode(struct inode *inode, 3721 struct page *ipage, u64 from); 3722 int f2fs_read_inline_data(struct inode *inode, struct page *page); 3723 int f2fs_convert_inline_page(struct dnode_of_data *dn, struct page *page); 3724 int f2fs_convert_inline_inode(struct inode *inode); 3725 int f2fs_try_convert_inline_dir(struct inode *dir, struct dentry *dentry); 3726 int f2fs_write_inline_data(struct inode *inode, struct page *page); 3727 int f2fs_recover_inline_data(struct inode *inode, struct page *npage); 3728 struct f2fs_dir_entry *f2fs_find_in_inline_dir(struct inode *dir, 3729 const struct f2fs_filename *fname, 3730 struct page **res_page); 3731 int f2fs_make_empty_inline_dir(struct inode *inode, struct inode *parent, 3732 struct page *ipage); 3733 int f2fs_add_inline_entry(struct inode *dir, const struct f2fs_filename *fname, 3734 struct inode *inode, nid_t ino, umode_t mode); 3735 void f2fs_delete_inline_entry(struct f2fs_dir_entry *dentry, 3736 struct page *page, struct inode *dir, 3737 struct inode *inode); 3738 bool f2fs_empty_inline_dir(struct inode *dir); 3739 int f2fs_read_inline_dir(struct file *file, struct dir_context *ctx, 3740 struct fscrypt_str *fstr); 3741 int f2fs_inline_data_fiemap(struct inode *inode, 3742 struct fiemap_extent_info *fieinfo, 3743 __u64 start, __u64 len); 3744 3745 /* 3746 * shrinker.c 3747 */ 3748 unsigned long f2fs_shrink_count(struct shrinker *shrink, 3749 struct shrink_control *sc); 3750 unsigned long f2fs_shrink_scan(struct shrinker *shrink, 3751 struct shrink_control *sc); 3752 void f2fs_join_shrinker(struct f2fs_sb_info *sbi); 3753 void f2fs_leave_shrinker(struct f2fs_sb_info *sbi); 3754 3755 /* 3756 * extent_cache.c 3757 */ 3758 struct rb_entry *f2fs_lookup_rb_tree(struct rb_root_cached *root, 3759 struct rb_entry *cached_re, unsigned int ofs); 3760 struct rb_node **f2fs_lookup_rb_tree_ext(struct f2fs_sb_info *sbi, 3761 struct rb_root_cached *root, 3762 struct rb_node **parent, 3763 unsigned long long key, bool *left_most); 3764 struct rb_node **f2fs_lookup_rb_tree_for_insert(struct f2fs_sb_info *sbi, 3765 struct rb_root_cached *root, 3766 struct rb_node **parent, 3767 unsigned int ofs, bool *leftmost); 3768 struct rb_entry *f2fs_lookup_rb_tree_ret(struct rb_root_cached *root, 3769 struct rb_entry *cached_re, unsigned int ofs, 3770 struct rb_entry **prev_entry, struct rb_entry **next_entry, 3771 struct rb_node ***insert_p, struct rb_node **insert_parent, 3772 bool force, bool *leftmost); 3773 bool f2fs_check_rb_tree_consistence(struct f2fs_sb_info *sbi, 3774 struct rb_root_cached *root, bool check_key); 3775 unsigned int f2fs_shrink_extent_tree(struct f2fs_sb_info *sbi, int nr_shrink); 3776 void f2fs_init_extent_tree(struct inode *inode, struct page *ipage); 3777 void f2fs_drop_extent_tree(struct inode *inode); 3778 unsigned int f2fs_destroy_extent_node(struct inode *inode); 3779 void f2fs_destroy_extent_tree(struct inode *inode); 3780 bool f2fs_lookup_extent_cache(struct inode *inode, pgoff_t pgofs, 3781 struct extent_info *ei); 3782 void f2fs_update_extent_cache(struct dnode_of_data *dn); 3783 void f2fs_update_extent_cache_range(struct dnode_of_data *dn, 3784 pgoff_t fofs, block_t blkaddr, unsigned int len); 3785 void f2fs_init_extent_cache_info(struct f2fs_sb_info *sbi); 3786 int __init f2fs_create_extent_cache(void); 3787 void f2fs_destroy_extent_cache(void); 3788 3789 /* 3790 * sysfs.c 3791 */ 3792 int __init f2fs_init_sysfs(void); 3793 void f2fs_exit_sysfs(void); 3794 int f2fs_register_sysfs(struct f2fs_sb_info *sbi); 3795 void f2fs_unregister_sysfs(struct f2fs_sb_info *sbi); 3796 3797 /* verity.c */ 3798 extern const struct fsverity_operations f2fs_verityops; 3799 3800 /* 3801 * crypto support 3802 */ 3803 static inline bool f2fs_encrypted_file(struct inode *inode) 3804 { 3805 return IS_ENCRYPTED(inode) && S_ISREG(inode->i_mode); 3806 } 3807 3808 static inline void f2fs_set_encrypted_inode(struct inode *inode) 3809 { 3810 #ifdef CONFIG_FS_ENCRYPTION 3811 file_set_encrypt(inode); 3812 f2fs_set_inode_flags(inode); 3813 #endif 3814 } 3815 3816 /* 3817 * Returns true if the reads of the inode's data need to undergo some 3818 * postprocessing step, like decryption or authenticity verification. 3819 */ 3820 static inline bool f2fs_post_read_required(struct inode *inode) 3821 { 3822 return f2fs_encrypted_file(inode) || fsverity_active(inode) || 3823 f2fs_compressed_file(inode); 3824 } 3825 3826 /* 3827 * compress.c 3828 */ 3829 #ifdef CONFIG_F2FS_FS_COMPRESSION 3830 bool f2fs_is_compressed_page(struct page *page); 3831 struct page *f2fs_compress_control_page(struct page *page); 3832 int f2fs_prepare_compress_overwrite(struct inode *inode, 3833 struct page **pagep, pgoff_t index, void **fsdata); 3834 bool f2fs_compress_write_end(struct inode *inode, void *fsdata, 3835 pgoff_t index, unsigned copied); 3836 int f2fs_truncate_partial_cluster(struct inode *inode, u64 from, bool lock); 3837 void f2fs_compress_write_end_io(struct bio *bio, struct page *page); 3838 bool f2fs_is_compress_backend_ready(struct inode *inode); 3839 int f2fs_init_compress_mempool(void); 3840 void f2fs_destroy_compress_mempool(void); 3841 void f2fs_decompress_pages(struct bio *bio, struct page *page, bool verity); 3842 bool f2fs_cluster_is_empty(struct compress_ctx *cc); 3843 bool f2fs_cluster_can_merge_page(struct compress_ctx *cc, pgoff_t index); 3844 void f2fs_compress_ctx_add_page(struct compress_ctx *cc, struct page *page); 3845 int f2fs_write_multi_pages(struct compress_ctx *cc, 3846 int *submitted, 3847 struct writeback_control *wbc, 3848 enum iostat_type io_type); 3849 int f2fs_is_compressed_cluster(struct inode *inode, pgoff_t index); 3850 int f2fs_read_multi_pages(struct compress_ctx *cc, struct bio **bio_ret, 3851 unsigned nr_pages, sector_t *last_block_in_bio, 3852 bool is_readahead, bool for_write); 3853 struct decompress_io_ctx *f2fs_alloc_dic(struct compress_ctx *cc); 3854 void f2fs_free_dic(struct decompress_io_ctx *dic); 3855 void f2fs_decompress_end_io(struct page **rpages, 3856 unsigned int cluster_size, bool err, bool verity); 3857 int f2fs_init_compress_ctx(struct compress_ctx *cc); 3858 void f2fs_destroy_compress_ctx(struct compress_ctx *cc); 3859 void f2fs_init_compress_info(struct f2fs_sb_info *sbi); 3860 int f2fs_init_page_array_cache(struct f2fs_sb_info *sbi); 3861 void f2fs_destroy_page_array_cache(struct f2fs_sb_info *sbi); 3862 int __init f2fs_init_compress_cache(void); 3863 void f2fs_destroy_compress_cache(void); 3864 #else 3865 static inline bool f2fs_is_compressed_page(struct page *page) { return false; } 3866 static inline bool f2fs_is_compress_backend_ready(struct inode *inode) 3867 { 3868 if (!f2fs_compressed_file(inode)) 3869 return true; 3870 /* not support compression */ 3871 return false; 3872 } 3873 static inline struct page *f2fs_compress_control_page(struct page *page) 3874 { 3875 WARN_ON_ONCE(1); 3876 return ERR_PTR(-EINVAL); 3877 } 3878 static inline int f2fs_init_compress_mempool(void) { return 0; } 3879 static inline void f2fs_destroy_compress_mempool(void) { } 3880 static inline int f2fs_init_page_array_cache(struct f2fs_sb_info *sbi) { return 0; } 3881 static inline void f2fs_destroy_page_array_cache(struct f2fs_sb_info *sbi) { } 3882 static inline int __init f2fs_init_compress_cache(void) { return 0; } 3883 static inline void f2fs_destroy_compress_cache(void) { } 3884 #endif 3885 3886 static inline void set_compress_context(struct inode *inode) 3887 { 3888 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 3889 3890 F2FS_I(inode)->i_compress_algorithm = 3891 F2FS_OPTION(sbi).compress_algorithm; 3892 F2FS_I(inode)->i_log_cluster_size = 3893 F2FS_OPTION(sbi).compress_log_size; 3894 F2FS_I(inode)->i_compress_flag = 3895 F2FS_OPTION(sbi).compress_chksum ? 3896 1 << COMPRESS_CHKSUM : 0; 3897 F2FS_I(inode)->i_cluster_size = 3898 1 << F2FS_I(inode)->i_log_cluster_size; 3899 F2FS_I(inode)->i_flags |= F2FS_COMPR_FL; 3900 set_inode_flag(inode, FI_COMPRESSED_FILE); 3901 stat_inc_compr_inode(inode); 3902 f2fs_mark_inode_dirty_sync(inode, true); 3903 } 3904 3905 static inline bool f2fs_disable_compressed_file(struct inode *inode) 3906 { 3907 struct f2fs_inode_info *fi = F2FS_I(inode); 3908 3909 if (!f2fs_compressed_file(inode)) 3910 return true; 3911 if (S_ISREG(inode->i_mode) && 3912 (get_dirty_pages(inode) || atomic_read(&fi->i_compr_blocks))) 3913 return false; 3914 3915 fi->i_flags &= ~F2FS_COMPR_FL; 3916 stat_dec_compr_inode(inode); 3917 clear_inode_flag(inode, FI_COMPRESSED_FILE); 3918 f2fs_mark_inode_dirty_sync(inode, true); 3919 return true; 3920 } 3921 3922 #define F2FS_FEATURE_FUNCS(name, flagname) \ 3923 static inline int f2fs_sb_has_##name(struct f2fs_sb_info *sbi) \ 3924 { \ 3925 return F2FS_HAS_FEATURE(sbi, F2FS_FEATURE_##flagname); \ 3926 } 3927 3928 F2FS_FEATURE_FUNCS(encrypt, ENCRYPT); 3929 F2FS_FEATURE_FUNCS(blkzoned, BLKZONED); 3930 F2FS_FEATURE_FUNCS(extra_attr, EXTRA_ATTR); 3931 F2FS_FEATURE_FUNCS(project_quota, PRJQUOTA); 3932 F2FS_FEATURE_FUNCS(inode_chksum, INODE_CHKSUM); 3933 F2FS_FEATURE_FUNCS(flexible_inline_xattr, FLEXIBLE_INLINE_XATTR); 3934 F2FS_FEATURE_FUNCS(quota_ino, QUOTA_INO); 3935 F2FS_FEATURE_FUNCS(inode_crtime, INODE_CRTIME); 3936 F2FS_FEATURE_FUNCS(lost_found, LOST_FOUND); 3937 F2FS_FEATURE_FUNCS(verity, VERITY); 3938 F2FS_FEATURE_FUNCS(sb_chksum, SB_CHKSUM); 3939 F2FS_FEATURE_FUNCS(casefold, CASEFOLD); 3940 F2FS_FEATURE_FUNCS(compression, COMPRESSION); 3941 3942 #ifdef CONFIG_BLK_DEV_ZONED 3943 static inline bool f2fs_blkz_is_seq(struct f2fs_sb_info *sbi, int devi, 3944 block_t blkaddr) 3945 { 3946 unsigned int zno = blkaddr >> sbi->log_blocks_per_blkz; 3947 3948 return test_bit(zno, FDEV(devi).blkz_seq); 3949 } 3950 #endif 3951 3952 static inline bool f2fs_hw_should_discard(struct f2fs_sb_info *sbi) 3953 { 3954 return f2fs_sb_has_blkzoned(sbi); 3955 } 3956 3957 static inline bool f2fs_bdev_support_discard(struct block_device *bdev) 3958 { 3959 return blk_queue_discard(bdev_get_queue(bdev)) || 3960 bdev_is_zoned(bdev); 3961 } 3962 3963 static inline bool f2fs_hw_support_discard(struct f2fs_sb_info *sbi) 3964 { 3965 int i; 3966 3967 if (!f2fs_is_multi_device(sbi)) 3968 return f2fs_bdev_support_discard(sbi->sb->s_bdev); 3969 3970 for (i = 0; i < sbi->s_ndevs; i++) 3971 if (f2fs_bdev_support_discard(FDEV(i).bdev)) 3972 return true; 3973 return false; 3974 } 3975 3976 static inline bool f2fs_realtime_discard_enable(struct f2fs_sb_info *sbi) 3977 { 3978 return (test_opt(sbi, DISCARD) && f2fs_hw_support_discard(sbi)) || 3979 f2fs_hw_should_discard(sbi); 3980 } 3981 3982 static inline bool f2fs_hw_is_readonly(struct f2fs_sb_info *sbi) 3983 { 3984 int i; 3985 3986 if (!f2fs_is_multi_device(sbi)) 3987 return bdev_read_only(sbi->sb->s_bdev); 3988 3989 for (i = 0; i < sbi->s_ndevs; i++) 3990 if (bdev_read_only(FDEV(i).bdev)) 3991 return true; 3992 return false; 3993 } 3994 3995 static inline bool f2fs_lfs_mode(struct f2fs_sb_info *sbi) 3996 { 3997 return F2FS_OPTION(sbi).fs_mode == FS_MODE_LFS; 3998 } 3999 4000 static inline bool f2fs_may_compress(struct inode *inode) 4001 { 4002 if (IS_SWAPFILE(inode) || f2fs_is_pinned_file(inode) || 4003 f2fs_is_atomic_file(inode) || 4004 f2fs_is_volatile_file(inode)) 4005 return false; 4006 return S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode); 4007 } 4008 4009 static inline void f2fs_i_compr_blocks_update(struct inode *inode, 4010 u64 blocks, bool add) 4011 { 4012 int diff = F2FS_I(inode)->i_cluster_size - blocks; 4013 struct f2fs_inode_info *fi = F2FS_I(inode); 4014 4015 /* don't update i_compr_blocks if saved blocks were released */ 4016 if (!add && !atomic_read(&fi->i_compr_blocks)) 4017 return; 4018 4019 if (add) { 4020 atomic_add(diff, &fi->i_compr_blocks); 4021 stat_add_compr_blocks(inode, diff); 4022 } else { 4023 atomic_sub(diff, &fi->i_compr_blocks); 4024 stat_sub_compr_blocks(inode, diff); 4025 } 4026 f2fs_mark_inode_dirty_sync(inode, true); 4027 } 4028 4029 static inline int block_unaligned_IO(struct inode *inode, 4030 struct kiocb *iocb, struct iov_iter *iter) 4031 { 4032 unsigned int i_blkbits = READ_ONCE(inode->i_blkbits); 4033 unsigned int blocksize_mask = (1 << i_blkbits) - 1; 4034 loff_t offset = iocb->ki_pos; 4035 unsigned long align = offset | iov_iter_alignment(iter); 4036 4037 return align & blocksize_mask; 4038 } 4039 4040 static inline int allow_outplace_dio(struct inode *inode, 4041 struct kiocb *iocb, struct iov_iter *iter) 4042 { 4043 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 4044 int rw = iov_iter_rw(iter); 4045 4046 return (f2fs_lfs_mode(sbi) && (rw == WRITE) && 4047 !block_unaligned_IO(inode, iocb, iter)); 4048 } 4049 4050 static inline bool f2fs_force_buffered_io(struct inode *inode, 4051 struct kiocb *iocb, struct iov_iter *iter) 4052 { 4053 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 4054 int rw = iov_iter_rw(iter); 4055 4056 if (f2fs_post_read_required(inode)) 4057 return true; 4058 if (f2fs_is_multi_device(sbi)) 4059 return true; 4060 /* 4061 * for blkzoned device, fallback direct IO to buffered IO, so 4062 * all IOs can be serialized by log-structured write. 4063 */ 4064 if (f2fs_sb_has_blkzoned(sbi)) 4065 return true; 4066 if (f2fs_lfs_mode(sbi) && (rw == WRITE)) { 4067 if (block_unaligned_IO(inode, iocb, iter)) 4068 return true; 4069 if (F2FS_IO_ALIGNED(sbi)) 4070 return true; 4071 } 4072 if (is_sbi_flag_set(F2FS_I_SB(inode), SBI_CP_DISABLED) && 4073 !IS_SWAPFILE(inode)) 4074 return true; 4075 4076 return false; 4077 } 4078 4079 #ifdef CONFIG_F2FS_FAULT_INJECTION 4080 extern void f2fs_build_fault_attr(struct f2fs_sb_info *sbi, unsigned int rate, 4081 unsigned int type); 4082 #else 4083 #define f2fs_build_fault_attr(sbi, rate, type) do { } while (0) 4084 #endif 4085 4086 static inline bool is_journalled_quota(struct f2fs_sb_info *sbi) 4087 { 4088 #ifdef CONFIG_QUOTA 4089 if (f2fs_sb_has_quota_ino(sbi)) 4090 return true; 4091 if (F2FS_OPTION(sbi).s_qf_names[USRQUOTA] || 4092 F2FS_OPTION(sbi).s_qf_names[GRPQUOTA] || 4093 F2FS_OPTION(sbi).s_qf_names[PRJQUOTA]) 4094 return true; 4095 #endif 4096 return false; 4097 } 4098 4099 #define EFSBADCRC EBADMSG /* Bad CRC detected */ 4100 #define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */ 4101 4102 #endif /* _LINUX_F2FS_H */ 4103