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 compress_log_size; /* cluster log size */ 151 unsigned char compress_ext_cnt; /* extension count */ 152 unsigned char extensions[COMPRESS_EXT_NUM][F2FS_EXTENSION_LEN]; /* extensions */ 153 }; 154 155 #define F2FS_FEATURE_ENCRYPT 0x0001 156 #define F2FS_FEATURE_BLKZONED 0x0002 157 #define F2FS_FEATURE_ATOMIC_WRITE 0x0004 158 #define F2FS_FEATURE_EXTRA_ATTR 0x0008 159 #define F2FS_FEATURE_PRJQUOTA 0x0010 160 #define F2FS_FEATURE_INODE_CHKSUM 0x0020 161 #define F2FS_FEATURE_FLEXIBLE_INLINE_XATTR 0x0040 162 #define F2FS_FEATURE_QUOTA_INO 0x0080 163 #define F2FS_FEATURE_INODE_CRTIME 0x0100 164 #define F2FS_FEATURE_LOST_FOUND 0x0200 165 #define F2FS_FEATURE_VERITY 0x0400 166 #define F2FS_FEATURE_SB_CHKSUM 0x0800 167 #define F2FS_FEATURE_CASEFOLD 0x1000 168 #define F2FS_FEATURE_COMPRESSION 0x2000 169 170 #define __F2FS_HAS_FEATURE(raw_super, mask) \ 171 ((raw_super->feature & cpu_to_le32(mask)) != 0) 172 #define F2FS_HAS_FEATURE(sbi, mask) __F2FS_HAS_FEATURE(sbi->raw_super, mask) 173 #define F2FS_SET_FEATURE(sbi, mask) \ 174 (sbi->raw_super->feature |= cpu_to_le32(mask)) 175 #define F2FS_CLEAR_FEATURE(sbi, mask) \ 176 (sbi->raw_super->feature &= ~cpu_to_le32(mask)) 177 178 /* 179 * Default values for user and/or group using reserved blocks 180 */ 181 #define F2FS_DEF_RESUID 0 182 #define F2FS_DEF_RESGID 0 183 184 /* 185 * For checkpoint manager 186 */ 187 enum { 188 NAT_BITMAP, 189 SIT_BITMAP 190 }; 191 192 #define CP_UMOUNT 0x00000001 193 #define CP_FASTBOOT 0x00000002 194 #define CP_SYNC 0x00000004 195 #define CP_RECOVERY 0x00000008 196 #define CP_DISCARD 0x00000010 197 #define CP_TRIMMED 0x00000020 198 #define CP_PAUSE 0x00000040 199 #define CP_RESIZE 0x00000080 200 201 #define MAX_DISCARD_BLOCKS(sbi) BLKS_PER_SEC(sbi) 202 #define DEF_MAX_DISCARD_REQUEST 8 /* issue 8 discards per round */ 203 #define DEF_MIN_DISCARD_ISSUE_TIME 50 /* 50 ms, if exists */ 204 #define DEF_MID_DISCARD_ISSUE_TIME 500 /* 500 ms, if device busy */ 205 #define DEF_MAX_DISCARD_ISSUE_TIME 60000 /* 60 s, if no candidates */ 206 #define DEF_DISCARD_URGENT_UTIL 80 /* do more discard over 80% */ 207 #define DEF_CP_INTERVAL 60 /* 60 secs */ 208 #define DEF_IDLE_INTERVAL 5 /* 5 secs */ 209 #define DEF_DISABLE_INTERVAL 5 /* 5 secs */ 210 #define DEF_DISABLE_QUICK_INTERVAL 1 /* 1 secs */ 211 #define DEF_UMOUNT_DISCARD_TIMEOUT 5 /* 5 secs */ 212 213 struct cp_control { 214 int reason; 215 __u64 trim_start; 216 __u64 trim_end; 217 __u64 trim_minlen; 218 }; 219 220 /* 221 * indicate meta/data type 222 */ 223 enum { 224 META_CP, 225 META_NAT, 226 META_SIT, 227 META_SSA, 228 META_MAX, 229 META_POR, 230 DATA_GENERIC, /* check range only */ 231 DATA_GENERIC_ENHANCE, /* strong check on range and segment bitmap */ 232 DATA_GENERIC_ENHANCE_READ, /* 233 * strong check on range and segment 234 * bitmap but no warning due to race 235 * condition of read on truncated area 236 * by extent_cache 237 */ 238 META_GENERIC, 239 }; 240 241 /* for the list of ino */ 242 enum { 243 ORPHAN_INO, /* for orphan ino list */ 244 APPEND_INO, /* for append ino list */ 245 UPDATE_INO, /* for update ino list */ 246 TRANS_DIR_INO, /* for trasactions dir ino list */ 247 FLUSH_INO, /* for multiple device flushing */ 248 MAX_INO_ENTRY, /* max. list */ 249 }; 250 251 struct ino_entry { 252 struct list_head list; /* list head */ 253 nid_t ino; /* inode number */ 254 unsigned int dirty_device; /* dirty device bitmap */ 255 }; 256 257 /* for the list of inodes to be GCed */ 258 struct inode_entry { 259 struct list_head list; /* list head */ 260 struct inode *inode; /* vfs inode pointer */ 261 }; 262 263 struct fsync_node_entry { 264 struct list_head list; /* list head */ 265 struct page *page; /* warm node page pointer */ 266 unsigned int seq_id; /* sequence id */ 267 }; 268 269 /* for the bitmap indicate blocks to be discarded */ 270 struct discard_entry { 271 struct list_head list; /* list head */ 272 block_t start_blkaddr; /* start blockaddr of current segment */ 273 unsigned char discard_map[SIT_VBLOCK_MAP_SIZE]; /* segment discard bitmap */ 274 }; 275 276 /* default discard granularity of inner discard thread, unit: block count */ 277 #define DEFAULT_DISCARD_GRANULARITY 16 278 279 /* max discard pend list number */ 280 #define MAX_PLIST_NUM 512 281 #define plist_idx(blk_num) ((blk_num) >= MAX_PLIST_NUM ? \ 282 (MAX_PLIST_NUM - 1) : ((blk_num) - 1)) 283 284 enum { 285 D_PREP, /* initial */ 286 D_PARTIAL, /* partially submitted */ 287 D_SUBMIT, /* all submitted */ 288 D_DONE, /* finished */ 289 }; 290 291 struct discard_info { 292 block_t lstart; /* logical start address */ 293 block_t len; /* length */ 294 block_t start; /* actual start address in dev */ 295 }; 296 297 struct discard_cmd { 298 struct rb_node rb_node; /* rb node located in rb-tree */ 299 union { 300 struct { 301 block_t lstart; /* logical start address */ 302 block_t len; /* length */ 303 block_t start; /* actual start address in dev */ 304 }; 305 struct discard_info di; /* discard info */ 306 307 }; 308 struct list_head list; /* command list */ 309 struct completion wait; /* compleation */ 310 struct block_device *bdev; /* bdev */ 311 unsigned short ref; /* reference count */ 312 unsigned char state; /* state */ 313 unsigned char queued; /* queued discard */ 314 int error; /* bio error */ 315 spinlock_t lock; /* for state/bio_ref updating */ 316 unsigned short bio_ref; /* bio reference count */ 317 }; 318 319 enum { 320 DPOLICY_BG, 321 DPOLICY_FORCE, 322 DPOLICY_FSTRIM, 323 DPOLICY_UMOUNT, 324 MAX_DPOLICY, 325 }; 326 327 struct discard_policy { 328 int type; /* type of discard */ 329 unsigned int min_interval; /* used for candidates exist */ 330 unsigned int mid_interval; /* used for device busy */ 331 unsigned int max_interval; /* used for candidates not exist */ 332 unsigned int max_requests; /* # of discards issued per round */ 333 unsigned int io_aware_gran; /* minimum granularity discard not be aware of I/O */ 334 bool io_aware; /* issue discard in idle time */ 335 bool sync; /* submit discard with REQ_SYNC flag */ 336 bool ordered; /* issue discard by lba order */ 337 bool timeout; /* discard timeout for put_super */ 338 unsigned int granularity; /* discard granularity */ 339 }; 340 341 struct discard_cmd_control { 342 struct task_struct *f2fs_issue_discard; /* discard thread */ 343 struct list_head entry_list; /* 4KB discard entry list */ 344 struct list_head pend_list[MAX_PLIST_NUM];/* store pending entries */ 345 struct list_head wait_list; /* store on-flushing entries */ 346 struct list_head fstrim_list; /* in-flight discard from fstrim */ 347 wait_queue_head_t discard_wait_queue; /* waiting queue for wake-up */ 348 unsigned int discard_wake; /* to wake up discard thread */ 349 struct mutex cmd_lock; 350 unsigned int nr_discards; /* # of discards in the list */ 351 unsigned int max_discards; /* max. discards to be issued */ 352 unsigned int discard_granularity; /* discard granularity */ 353 unsigned int undiscard_blks; /* # of undiscard blocks */ 354 unsigned int next_pos; /* next discard position */ 355 atomic_t issued_discard; /* # of issued discard */ 356 atomic_t queued_discard; /* # of queued discard */ 357 atomic_t discard_cmd_cnt; /* # of cached cmd count */ 358 struct rb_root_cached root; /* root of discard rb-tree */ 359 bool rbtree_check; /* config for consistence check */ 360 }; 361 362 /* for the list of fsync inodes, used only during recovery */ 363 struct fsync_inode_entry { 364 struct list_head list; /* list head */ 365 struct inode *inode; /* vfs inode pointer */ 366 block_t blkaddr; /* block address locating the last fsync */ 367 block_t last_dentry; /* block address locating the last dentry */ 368 }; 369 370 #define nats_in_cursum(jnl) (le16_to_cpu((jnl)->n_nats)) 371 #define sits_in_cursum(jnl) (le16_to_cpu((jnl)->n_sits)) 372 373 #define nat_in_journal(jnl, i) ((jnl)->nat_j.entries[i].ne) 374 #define nid_in_journal(jnl, i) ((jnl)->nat_j.entries[i].nid) 375 #define sit_in_journal(jnl, i) ((jnl)->sit_j.entries[i].se) 376 #define segno_in_journal(jnl, i) ((jnl)->sit_j.entries[i].segno) 377 378 #define MAX_NAT_JENTRIES(jnl) (NAT_JOURNAL_ENTRIES - nats_in_cursum(jnl)) 379 #define MAX_SIT_JENTRIES(jnl) (SIT_JOURNAL_ENTRIES - sits_in_cursum(jnl)) 380 381 static inline int update_nats_in_cursum(struct f2fs_journal *journal, int i) 382 { 383 int before = nats_in_cursum(journal); 384 385 journal->n_nats = cpu_to_le16(before + i); 386 return before; 387 } 388 389 static inline int update_sits_in_cursum(struct f2fs_journal *journal, int i) 390 { 391 int before = sits_in_cursum(journal); 392 393 journal->n_sits = cpu_to_le16(before + i); 394 return before; 395 } 396 397 static inline bool __has_cursum_space(struct f2fs_journal *journal, 398 int size, int type) 399 { 400 if (type == NAT_JOURNAL) 401 return size <= MAX_NAT_JENTRIES(journal); 402 return size <= MAX_SIT_JENTRIES(journal); 403 } 404 405 /* for inline stuff */ 406 #define DEF_INLINE_RESERVED_SIZE 1 407 static inline int get_extra_isize(struct inode *inode); 408 static inline int get_inline_xattr_addrs(struct inode *inode); 409 #define MAX_INLINE_DATA(inode) (sizeof(__le32) * \ 410 (CUR_ADDRS_PER_INODE(inode) - \ 411 get_inline_xattr_addrs(inode) - \ 412 DEF_INLINE_RESERVED_SIZE)) 413 414 /* for inline dir */ 415 #define NR_INLINE_DENTRY(inode) (MAX_INLINE_DATA(inode) * BITS_PER_BYTE / \ 416 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \ 417 BITS_PER_BYTE + 1)) 418 #define INLINE_DENTRY_BITMAP_SIZE(inode) \ 419 DIV_ROUND_UP(NR_INLINE_DENTRY(inode), BITS_PER_BYTE) 420 #define INLINE_RESERVED_SIZE(inode) (MAX_INLINE_DATA(inode) - \ 421 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \ 422 NR_INLINE_DENTRY(inode) + \ 423 INLINE_DENTRY_BITMAP_SIZE(inode))) 424 425 /* 426 * For INODE and NODE manager 427 */ 428 /* for directory operations */ 429 430 struct f2fs_filename { 431 /* 432 * The filename the user specified. This is NULL for some 433 * filesystem-internal operations, e.g. converting an inline directory 434 * to a non-inline one, or roll-forward recovering an encrypted dentry. 435 */ 436 const struct qstr *usr_fname; 437 438 /* 439 * The on-disk filename. For encrypted directories, this is encrypted. 440 * This may be NULL for lookups in an encrypted dir without the key. 441 */ 442 struct fscrypt_str disk_name; 443 444 /* The dirhash of this filename */ 445 f2fs_hash_t hash; 446 447 #ifdef CONFIG_FS_ENCRYPTION 448 /* 449 * For lookups in encrypted directories: either the buffer backing 450 * disk_name, or a buffer that holds the decoded no-key name. 451 */ 452 struct fscrypt_str crypto_buf; 453 #endif 454 #ifdef CONFIG_UNICODE 455 /* 456 * For casefolded directories: the casefolded name, but it's left NULL 457 * if the original name is not valid Unicode or if the filesystem is 458 * doing an internal operation where usr_fname is also NULL. In these 459 * cases we fall back to treating the name as an opaque byte sequence. 460 */ 461 struct fscrypt_str cf_name; 462 #endif 463 }; 464 465 struct f2fs_dentry_ptr { 466 struct inode *inode; 467 void *bitmap; 468 struct f2fs_dir_entry *dentry; 469 __u8 (*filename)[F2FS_SLOT_LEN]; 470 int max; 471 int nr_bitmap; 472 }; 473 474 static inline void make_dentry_ptr_block(struct inode *inode, 475 struct f2fs_dentry_ptr *d, struct f2fs_dentry_block *t) 476 { 477 d->inode = inode; 478 d->max = NR_DENTRY_IN_BLOCK; 479 d->nr_bitmap = SIZE_OF_DENTRY_BITMAP; 480 d->bitmap = t->dentry_bitmap; 481 d->dentry = t->dentry; 482 d->filename = t->filename; 483 } 484 485 static inline void make_dentry_ptr_inline(struct inode *inode, 486 struct f2fs_dentry_ptr *d, void *t) 487 { 488 int entry_cnt = NR_INLINE_DENTRY(inode); 489 int bitmap_size = INLINE_DENTRY_BITMAP_SIZE(inode); 490 int reserved_size = INLINE_RESERVED_SIZE(inode); 491 492 d->inode = inode; 493 d->max = entry_cnt; 494 d->nr_bitmap = bitmap_size; 495 d->bitmap = t; 496 d->dentry = t + bitmap_size + reserved_size; 497 d->filename = t + bitmap_size + reserved_size + 498 SIZE_OF_DIR_ENTRY * entry_cnt; 499 } 500 501 /* 502 * XATTR_NODE_OFFSET stores xattrs to one node block per file keeping -1 503 * as its node offset to distinguish from index node blocks. 504 * But some bits are used to mark the node block. 505 */ 506 #define XATTR_NODE_OFFSET ((((unsigned int)-1) << OFFSET_BIT_SHIFT) \ 507 >> OFFSET_BIT_SHIFT) 508 enum { 509 ALLOC_NODE, /* allocate a new node page if needed */ 510 LOOKUP_NODE, /* look up a node without readahead */ 511 LOOKUP_NODE_RA, /* 512 * look up a node with readahead called 513 * by get_data_block. 514 */ 515 }; 516 517 #define DEFAULT_RETRY_IO_COUNT 8 /* maximum retry read IO count */ 518 519 /* congestion wait timeout value, default: 20ms */ 520 #define DEFAULT_IO_TIMEOUT (msecs_to_jiffies(20)) 521 522 /* maximum retry quota flush count */ 523 #define DEFAULT_RETRY_QUOTA_FLUSH_COUNT 8 524 525 #define F2FS_LINK_MAX 0xffffffff /* maximum link count per file */ 526 527 #define MAX_DIR_RA_PAGES 4 /* maximum ra pages of dir */ 528 529 /* for in-memory extent cache entry */ 530 #define F2FS_MIN_EXTENT_LEN 64 /* minimum extent length */ 531 532 /* number of extent info in extent cache we try to shrink */ 533 #define EXTENT_CACHE_SHRINK_NUMBER 128 534 535 struct rb_entry { 536 struct rb_node rb_node; /* rb node located in rb-tree */ 537 union { 538 struct { 539 unsigned int ofs; /* start offset of the entry */ 540 unsigned int len; /* length of the entry */ 541 }; 542 unsigned long long key; /* 64-bits key */ 543 } __packed; 544 }; 545 546 struct extent_info { 547 unsigned int fofs; /* start offset in a file */ 548 unsigned int len; /* length of the extent */ 549 u32 blk; /* start block address of the extent */ 550 }; 551 552 struct extent_node { 553 struct rb_node rb_node; /* rb node located in rb-tree */ 554 struct extent_info ei; /* extent info */ 555 struct list_head list; /* node in global extent list of sbi */ 556 struct extent_tree *et; /* extent tree pointer */ 557 }; 558 559 struct extent_tree { 560 nid_t ino; /* inode number */ 561 struct rb_root_cached root; /* root of extent info rb-tree */ 562 struct extent_node *cached_en; /* recently accessed extent node */ 563 struct extent_info largest; /* largested extent info */ 564 struct list_head list; /* to be used by sbi->zombie_list */ 565 rwlock_t lock; /* protect extent info rb-tree */ 566 atomic_t node_cnt; /* # of extent node in rb-tree*/ 567 bool largest_updated; /* largest extent updated */ 568 }; 569 570 /* 571 * This structure is taken from ext4_map_blocks. 572 * 573 * Note that, however, f2fs uses NEW and MAPPED flags for f2fs_map_blocks(). 574 */ 575 #define F2FS_MAP_NEW (1 << BH_New) 576 #define F2FS_MAP_MAPPED (1 << BH_Mapped) 577 #define F2FS_MAP_UNWRITTEN (1 << BH_Unwritten) 578 #define F2FS_MAP_FLAGS (F2FS_MAP_NEW | F2FS_MAP_MAPPED |\ 579 F2FS_MAP_UNWRITTEN) 580 581 struct f2fs_map_blocks { 582 block_t m_pblk; 583 block_t m_lblk; 584 unsigned int m_len; 585 unsigned int m_flags; 586 pgoff_t *m_next_pgofs; /* point next possible non-hole pgofs */ 587 pgoff_t *m_next_extent; /* point to next possible extent */ 588 int m_seg_type; 589 bool m_may_create; /* indicate it is from write path */ 590 }; 591 592 /* for flag in get_data_block */ 593 enum { 594 F2FS_GET_BLOCK_DEFAULT, 595 F2FS_GET_BLOCK_FIEMAP, 596 F2FS_GET_BLOCK_BMAP, 597 F2FS_GET_BLOCK_DIO, 598 F2FS_GET_BLOCK_PRE_DIO, 599 F2FS_GET_BLOCK_PRE_AIO, 600 F2FS_GET_BLOCK_PRECACHE, 601 }; 602 603 /* 604 * i_advise uses FADVISE_XXX_BIT. We can add additional hints later. 605 */ 606 #define FADVISE_COLD_BIT 0x01 607 #define FADVISE_LOST_PINO_BIT 0x02 608 #define FADVISE_ENCRYPT_BIT 0x04 609 #define FADVISE_ENC_NAME_BIT 0x08 610 #define FADVISE_KEEP_SIZE_BIT 0x10 611 #define FADVISE_HOT_BIT 0x20 612 #define FADVISE_VERITY_BIT 0x40 613 614 #define FADVISE_MODIFIABLE_BITS (FADVISE_COLD_BIT | FADVISE_HOT_BIT) 615 616 #define file_is_cold(inode) is_file(inode, FADVISE_COLD_BIT) 617 #define file_wrong_pino(inode) is_file(inode, FADVISE_LOST_PINO_BIT) 618 #define file_set_cold(inode) set_file(inode, FADVISE_COLD_BIT) 619 #define file_lost_pino(inode) set_file(inode, FADVISE_LOST_PINO_BIT) 620 #define file_clear_cold(inode) clear_file(inode, FADVISE_COLD_BIT) 621 #define file_got_pino(inode) clear_file(inode, FADVISE_LOST_PINO_BIT) 622 #define file_is_encrypt(inode) is_file(inode, FADVISE_ENCRYPT_BIT) 623 #define file_set_encrypt(inode) set_file(inode, FADVISE_ENCRYPT_BIT) 624 #define file_clear_encrypt(inode) clear_file(inode, FADVISE_ENCRYPT_BIT) 625 #define file_enc_name(inode) is_file(inode, FADVISE_ENC_NAME_BIT) 626 #define file_set_enc_name(inode) set_file(inode, FADVISE_ENC_NAME_BIT) 627 #define file_keep_isize(inode) is_file(inode, FADVISE_KEEP_SIZE_BIT) 628 #define file_set_keep_isize(inode) set_file(inode, FADVISE_KEEP_SIZE_BIT) 629 #define file_is_hot(inode) is_file(inode, FADVISE_HOT_BIT) 630 #define file_set_hot(inode) set_file(inode, FADVISE_HOT_BIT) 631 #define file_clear_hot(inode) clear_file(inode, FADVISE_HOT_BIT) 632 #define file_is_verity(inode) is_file(inode, FADVISE_VERITY_BIT) 633 #define file_set_verity(inode) set_file(inode, FADVISE_VERITY_BIT) 634 635 #define DEF_DIR_LEVEL 0 636 637 enum { 638 GC_FAILURE_PIN, 639 GC_FAILURE_ATOMIC, 640 MAX_GC_FAILURE 641 }; 642 643 /* used for f2fs_inode_info->flags */ 644 enum { 645 FI_NEW_INODE, /* indicate newly allocated inode */ 646 FI_DIRTY_INODE, /* indicate inode is dirty or not */ 647 FI_AUTO_RECOVER, /* indicate inode is recoverable */ 648 FI_DIRTY_DIR, /* indicate directory has dirty pages */ 649 FI_INC_LINK, /* need to increment i_nlink */ 650 FI_ACL_MODE, /* indicate acl mode */ 651 FI_NO_ALLOC, /* should not allocate any blocks */ 652 FI_FREE_NID, /* free allocated nide */ 653 FI_NO_EXTENT, /* not to use the extent cache */ 654 FI_INLINE_XATTR, /* used for inline xattr */ 655 FI_INLINE_DATA, /* used for inline data*/ 656 FI_INLINE_DENTRY, /* used for inline dentry */ 657 FI_APPEND_WRITE, /* inode has appended data */ 658 FI_UPDATE_WRITE, /* inode has in-place-update data */ 659 FI_NEED_IPU, /* used for ipu per file */ 660 FI_ATOMIC_FILE, /* indicate atomic file */ 661 FI_ATOMIC_COMMIT, /* indicate the state of atomical committing */ 662 FI_VOLATILE_FILE, /* indicate volatile file */ 663 FI_FIRST_BLOCK_WRITTEN, /* indicate #0 data block was written */ 664 FI_DROP_CACHE, /* drop dirty page cache */ 665 FI_DATA_EXIST, /* indicate data exists */ 666 FI_INLINE_DOTS, /* indicate inline dot dentries */ 667 FI_DO_DEFRAG, /* indicate defragment is running */ 668 FI_DIRTY_FILE, /* indicate regular/symlink has dirty pages */ 669 FI_NO_PREALLOC, /* indicate skipped preallocated blocks */ 670 FI_HOT_DATA, /* indicate file is hot */ 671 FI_EXTRA_ATTR, /* indicate file has extra attribute */ 672 FI_PROJ_INHERIT, /* indicate file inherits projectid */ 673 FI_PIN_FILE, /* indicate file should not be gced */ 674 FI_ATOMIC_REVOKE_REQUEST, /* request to drop atomic data */ 675 FI_VERITY_IN_PROGRESS, /* building fs-verity Merkle tree */ 676 FI_COMPRESSED_FILE, /* indicate file's data can be compressed */ 677 FI_MMAP_FILE, /* indicate file was mmapped */ 678 FI_MAX, /* max flag, never be used */ 679 }; 680 681 struct f2fs_inode_info { 682 struct inode vfs_inode; /* serve a vfs inode */ 683 unsigned long i_flags; /* keep an inode flags for ioctl */ 684 unsigned char i_advise; /* use to give file attribute hints */ 685 unsigned char i_dir_level; /* use for dentry level for large dir */ 686 unsigned int i_current_depth; /* only for directory depth */ 687 /* for gc failure statistic */ 688 unsigned int i_gc_failures[MAX_GC_FAILURE]; 689 unsigned int i_pino; /* parent inode number */ 690 umode_t i_acl_mode; /* keep file acl mode temporarily */ 691 692 /* Use below internally in f2fs*/ 693 unsigned long flags[BITS_TO_LONGS(FI_MAX)]; /* use to pass per-file flags */ 694 struct rw_semaphore i_sem; /* protect fi info */ 695 atomic_t dirty_pages; /* # of dirty pages */ 696 f2fs_hash_t chash; /* hash value of given file name */ 697 unsigned int clevel; /* maximum level of given file name */ 698 struct task_struct *task; /* lookup and create consistency */ 699 struct task_struct *cp_task; /* separate cp/wb IO stats*/ 700 nid_t i_xattr_nid; /* node id that contains xattrs */ 701 loff_t last_disk_size; /* lastly written file size */ 702 spinlock_t i_size_lock; /* protect last_disk_size */ 703 704 #ifdef CONFIG_QUOTA 705 struct dquot *i_dquot[MAXQUOTAS]; 706 707 /* quota space reservation, managed internally by quota code */ 708 qsize_t i_reserved_quota; 709 #endif 710 struct list_head dirty_list; /* dirty list for dirs and files */ 711 struct list_head gdirty_list; /* linked in global dirty list */ 712 struct list_head inmem_ilist; /* list for inmem inodes */ 713 struct list_head inmem_pages; /* inmemory pages managed by f2fs */ 714 struct task_struct *inmem_task; /* store inmemory task */ 715 struct mutex inmem_lock; /* lock for inmemory pages */ 716 pgoff_t ra_offset; /* ongoing readahead offset */ 717 struct extent_tree *extent_tree; /* cached extent_tree entry */ 718 719 /* avoid racing between foreground op and gc */ 720 struct rw_semaphore i_gc_rwsem[2]; 721 struct rw_semaphore i_mmap_sem; 722 struct rw_semaphore i_xattr_sem; /* avoid racing between reading and changing EAs */ 723 724 int i_extra_isize; /* size of extra space located in i_addr */ 725 kprojid_t i_projid; /* id for project quota */ 726 int i_inline_xattr_size; /* inline xattr size */ 727 struct timespec64 i_crtime; /* inode creation time */ 728 struct timespec64 i_disk_time[4];/* inode disk times */ 729 730 /* for file compress */ 731 atomic_t i_compr_blocks; /* # of compressed blocks */ 732 unsigned char i_compress_algorithm; /* algorithm type */ 733 unsigned char i_log_cluster_size; /* log of cluster size */ 734 unsigned int i_cluster_size; /* cluster size */ 735 }; 736 737 static inline void get_extent_info(struct extent_info *ext, 738 struct f2fs_extent *i_ext) 739 { 740 ext->fofs = le32_to_cpu(i_ext->fofs); 741 ext->blk = le32_to_cpu(i_ext->blk); 742 ext->len = le32_to_cpu(i_ext->len); 743 } 744 745 static inline void set_raw_extent(struct extent_info *ext, 746 struct f2fs_extent *i_ext) 747 { 748 i_ext->fofs = cpu_to_le32(ext->fofs); 749 i_ext->blk = cpu_to_le32(ext->blk); 750 i_ext->len = cpu_to_le32(ext->len); 751 } 752 753 static inline void set_extent_info(struct extent_info *ei, unsigned int fofs, 754 u32 blk, unsigned int len) 755 { 756 ei->fofs = fofs; 757 ei->blk = blk; 758 ei->len = len; 759 } 760 761 static inline bool __is_discard_mergeable(struct discard_info *back, 762 struct discard_info *front, unsigned int max_len) 763 { 764 return (back->lstart + back->len == front->lstart) && 765 (back->len + front->len <= max_len); 766 } 767 768 static inline bool __is_discard_back_mergeable(struct discard_info *cur, 769 struct discard_info *back, unsigned int max_len) 770 { 771 return __is_discard_mergeable(back, cur, max_len); 772 } 773 774 static inline bool __is_discard_front_mergeable(struct discard_info *cur, 775 struct discard_info *front, unsigned int max_len) 776 { 777 return __is_discard_mergeable(cur, front, max_len); 778 } 779 780 static inline bool __is_extent_mergeable(struct extent_info *back, 781 struct extent_info *front) 782 { 783 return (back->fofs + back->len == front->fofs && 784 back->blk + back->len == front->blk); 785 } 786 787 static inline bool __is_back_mergeable(struct extent_info *cur, 788 struct extent_info *back) 789 { 790 return __is_extent_mergeable(back, cur); 791 } 792 793 static inline bool __is_front_mergeable(struct extent_info *cur, 794 struct extent_info *front) 795 { 796 return __is_extent_mergeable(cur, front); 797 } 798 799 extern void f2fs_mark_inode_dirty_sync(struct inode *inode, bool sync); 800 static inline void __try_update_largest_extent(struct extent_tree *et, 801 struct extent_node *en) 802 { 803 if (en->ei.len > et->largest.len) { 804 et->largest = en->ei; 805 et->largest_updated = true; 806 } 807 } 808 809 /* 810 * For free nid management 811 */ 812 enum nid_state { 813 FREE_NID, /* newly added to free nid list */ 814 PREALLOC_NID, /* it is preallocated */ 815 MAX_NID_STATE, 816 }; 817 818 struct f2fs_nm_info { 819 block_t nat_blkaddr; /* base disk address of NAT */ 820 nid_t max_nid; /* maximum possible node ids */ 821 nid_t available_nids; /* # of available node ids */ 822 nid_t next_scan_nid; /* the next nid to be scanned */ 823 unsigned int ram_thresh; /* control the memory footprint */ 824 unsigned int ra_nid_pages; /* # of nid pages to be readaheaded */ 825 unsigned int dirty_nats_ratio; /* control dirty nats ratio threshold */ 826 827 /* NAT cache management */ 828 struct radix_tree_root nat_root;/* root of the nat entry cache */ 829 struct radix_tree_root nat_set_root;/* root of the nat set cache */ 830 struct rw_semaphore nat_tree_lock; /* protect nat_tree_lock */ 831 struct list_head nat_entries; /* cached nat entry list (clean) */ 832 spinlock_t nat_list_lock; /* protect clean nat entry list */ 833 unsigned int nat_cnt; /* the # of cached nat entries */ 834 unsigned int dirty_nat_cnt; /* total num of nat entries in set */ 835 unsigned int nat_blocks; /* # of nat blocks */ 836 837 /* free node ids management */ 838 struct radix_tree_root free_nid_root;/* root of the free_nid cache */ 839 struct list_head free_nid_list; /* list for free nids excluding preallocated nids */ 840 unsigned int nid_cnt[MAX_NID_STATE]; /* the number of free node id */ 841 spinlock_t nid_list_lock; /* protect nid lists ops */ 842 struct mutex build_lock; /* lock for build free nids */ 843 unsigned char **free_nid_bitmap; 844 unsigned char *nat_block_bitmap; 845 unsigned short *free_nid_count; /* free nid count of NAT block */ 846 847 /* for checkpoint */ 848 char *nat_bitmap; /* NAT bitmap pointer */ 849 850 unsigned int nat_bits_blocks; /* # of nat bits blocks */ 851 unsigned char *nat_bits; /* NAT bits blocks */ 852 unsigned char *full_nat_bits; /* full NAT pages */ 853 unsigned char *empty_nat_bits; /* empty NAT pages */ 854 #ifdef CONFIG_F2FS_CHECK_FS 855 char *nat_bitmap_mir; /* NAT bitmap mirror */ 856 #endif 857 int bitmap_size; /* bitmap size */ 858 }; 859 860 /* 861 * this structure is used as one of function parameters. 862 * all the information are dedicated to a given direct node block determined 863 * by the data offset in a file. 864 */ 865 struct dnode_of_data { 866 struct inode *inode; /* vfs inode pointer */ 867 struct page *inode_page; /* its inode page, NULL is possible */ 868 struct page *node_page; /* cached direct node page */ 869 nid_t nid; /* node id of the direct node block */ 870 unsigned int ofs_in_node; /* data offset in the node page */ 871 bool inode_page_locked; /* inode page is locked or not */ 872 bool node_changed; /* is node block changed */ 873 char cur_level; /* level of hole node page */ 874 char max_level; /* level of current page located */ 875 block_t data_blkaddr; /* block address of the node block */ 876 }; 877 878 static inline void set_new_dnode(struct dnode_of_data *dn, struct inode *inode, 879 struct page *ipage, struct page *npage, nid_t nid) 880 { 881 memset(dn, 0, sizeof(*dn)); 882 dn->inode = inode; 883 dn->inode_page = ipage; 884 dn->node_page = npage; 885 dn->nid = nid; 886 } 887 888 /* 889 * For SIT manager 890 * 891 * By default, there are 6 active log areas across the whole main area. 892 * When considering hot and cold data separation to reduce cleaning overhead, 893 * we split 3 for data logs and 3 for node logs as hot, warm, and cold types, 894 * respectively. 895 * In the current design, you should not change the numbers intentionally. 896 * Instead, as a mount option such as active_logs=x, you can use 2, 4, and 6 897 * logs individually according to the underlying devices. (default: 6) 898 * Just in case, on-disk layout covers maximum 16 logs that consist of 8 for 899 * data and 8 for node logs. 900 */ 901 #define NR_CURSEG_DATA_TYPE (3) 902 #define NR_CURSEG_NODE_TYPE (3) 903 #define NR_CURSEG_INMEM_TYPE (2) 904 #define NR_CURSEG_PERSIST_TYPE (NR_CURSEG_DATA_TYPE + NR_CURSEG_NODE_TYPE) 905 #define NR_CURSEG_TYPE (NR_CURSEG_INMEM_TYPE + NR_CURSEG_PERSIST_TYPE) 906 907 enum { 908 CURSEG_HOT_DATA = 0, /* directory entry blocks */ 909 CURSEG_WARM_DATA, /* data blocks */ 910 CURSEG_COLD_DATA, /* multimedia or GCed data blocks */ 911 CURSEG_HOT_NODE, /* direct node blocks of directory files */ 912 CURSEG_WARM_NODE, /* direct node blocks of normal files */ 913 CURSEG_COLD_NODE, /* indirect node blocks */ 914 NR_PERSISTENT_LOG, /* number of persistent log */ 915 CURSEG_COLD_DATA_PINNED = NR_PERSISTENT_LOG, 916 /* pinned file that needs consecutive block address */ 917 CURSEG_ALL_DATA_ATGC, /* SSR alloctor in hot/warm/cold data area */ 918 NO_CHECK_TYPE, /* number of persistent & inmem log */ 919 }; 920 921 struct flush_cmd { 922 struct completion wait; 923 struct llist_node llnode; 924 nid_t ino; 925 int ret; 926 }; 927 928 struct flush_cmd_control { 929 struct task_struct *f2fs_issue_flush; /* flush thread */ 930 wait_queue_head_t flush_wait_queue; /* waiting queue for wake-up */ 931 atomic_t issued_flush; /* # of issued flushes */ 932 atomic_t queued_flush; /* # of queued flushes */ 933 struct llist_head issue_list; /* list for command issue */ 934 struct llist_node *dispatch_list; /* list for command dispatch */ 935 }; 936 937 struct f2fs_sm_info { 938 struct sit_info *sit_info; /* whole segment information */ 939 struct free_segmap_info *free_info; /* free segment information */ 940 struct dirty_seglist_info *dirty_info; /* dirty segment information */ 941 struct curseg_info *curseg_array; /* active segment information */ 942 943 struct rw_semaphore curseg_lock; /* for preventing curseg change */ 944 945 block_t seg0_blkaddr; /* block address of 0'th segment */ 946 block_t main_blkaddr; /* start block address of main area */ 947 block_t ssa_blkaddr; /* start block address of SSA area */ 948 949 unsigned int segment_count; /* total # of segments */ 950 unsigned int main_segments; /* # of segments in main area */ 951 unsigned int reserved_segments; /* # of reserved segments */ 952 unsigned int ovp_segments; /* # of overprovision segments */ 953 954 /* a threshold to reclaim prefree segments */ 955 unsigned int rec_prefree_segments; 956 957 /* for batched trimming */ 958 unsigned int trim_sections; /* # of sections to trim */ 959 960 struct list_head sit_entry_set; /* sit entry set list */ 961 962 unsigned int ipu_policy; /* in-place-update policy */ 963 unsigned int min_ipu_util; /* in-place-update threshold */ 964 unsigned int min_fsync_blocks; /* threshold for fsync */ 965 unsigned int min_seq_blocks; /* threshold for sequential blocks */ 966 unsigned int min_hot_blocks; /* threshold for hot block allocation */ 967 unsigned int min_ssr_sections; /* threshold to trigger SSR allocation */ 968 969 /* for flush command control */ 970 struct flush_cmd_control *fcc_info; 971 972 /* for discard command control */ 973 struct discard_cmd_control *dcc_info; 974 }; 975 976 /* 977 * For superblock 978 */ 979 /* 980 * COUNT_TYPE for monitoring 981 * 982 * f2fs monitors the number of several block types such as on-writeback, 983 * dirty dentry blocks, dirty node blocks, and dirty meta blocks. 984 */ 985 #define WB_DATA_TYPE(p) (__is_cp_guaranteed(p) ? F2FS_WB_CP_DATA : F2FS_WB_DATA) 986 enum count_type { 987 F2FS_DIRTY_DENTS, 988 F2FS_DIRTY_DATA, 989 F2FS_DIRTY_QDATA, 990 F2FS_DIRTY_NODES, 991 F2FS_DIRTY_META, 992 F2FS_INMEM_PAGES, 993 F2FS_DIRTY_IMETA, 994 F2FS_WB_CP_DATA, 995 F2FS_WB_DATA, 996 F2FS_RD_DATA, 997 F2FS_RD_NODE, 998 F2FS_RD_META, 999 F2FS_DIO_WRITE, 1000 F2FS_DIO_READ, 1001 NR_COUNT_TYPE, 1002 }; 1003 1004 /* 1005 * The below are the page types of bios used in submit_bio(). 1006 * The available types are: 1007 * DATA User data pages. It operates as async mode. 1008 * NODE Node pages. It operates as async mode. 1009 * META FS metadata pages such as SIT, NAT, CP. 1010 * NR_PAGE_TYPE The number of page types. 1011 * META_FLUSH Make sure the previous pages are written 1012 * with waiting the bio's completion 1013 * ... Only can be used with META. 1014 */ 1015 #define PAGE_TYPE_OF_BIO(type) ((type) > META ? META : (type)) 1016 enum page_type { 1017 DATA, 1018 NODE, 1019 META, 1020 NR_PAGE_TYPE, 1021 META_FLUSH, 1022 INMEM, /* the below types are used by tracepoints only. */ 1023 INMEM_DROP, 1024 INMEM_INVALIDATE, 1025 INMEM_REVOKE, 1026 IPU, 1027 OPU, 1028 }; 1029 1030 enum temp_type { 1031 HOT = 0, /* must be zero for meta bio */ 1032 WARM, 1033 COLD, 1034 NR_TEMP_TYPE, 1035 }; 1036 1037 enum need_lock_type { 1038 LOCK_REQ = 0, 1039 LOCK_DONE, 1040 LOCK_RETRY, 1041 }; 1042 1043 enum cp_reason_type { 1044 CP_NO_NEEDED, 1045 CP_NON_REGULAR, 1046 CP_COMPRESSED, 1047 CP_HARDLINK, 1048 CP_SB_NEED_CP, 1049 CP_WRONG_PINO, 1050 CP_NO_SPC_ROLL, 1051 CP_NODE_NEED_CP, 1052 CP_FASTBOOT_MODE, 1053 CP_SPEC_LOG_NUM, 1054 CP_RECOVER_DIR, 1055 }; 1056 1057 enum iostat_type { 1058 /* WRITE IO */ 1059 APP_DIRECT_IO, /* app direct write IOs */ 1060 APP_BUFFERED_IO, /* app buffered write IOs */ 1061 APP_WRITE_IO, /* app write IOs */ 1062 APP_MAPPED_IO, /* app mapped IOs */ 1063 FS_DATA_IO, /* data IOs from kworker/fsync/reclaimer */ 1064 FS_NODE_IO, /* node IOs from kworker/fsync/reclaimer */ 1065 FS_META_IO, /* meta IOs from kworker/reclaimer */ 1066 FS_GC_DATA_IO, /* data IOs from forground gc */ 1067 FS_GC_NODE_IO, /* node IOs from forground gc */ 1068 FS_CP_DATA_IO, /* data IOs from checkpoint */ 1069 FS_CP_NODE_IO, /* node IOs from checkpoint */ 1070 FS_CP_META_IO, /* meta IOs from checkpoint */ 1071 1072 /* READ IO */ 1073 APP_DIRECT_READ_IO, /* app direct read IOs */ 1074 APP_BUFFERED_READ_IO, /* app buffered read IOs */ 1075 APP_READ_IO, /* app read IOs */ 1076 APP_MAPPED_READ_IO, /* app mapped read IOs */ 1077 FS_DATA_READ_IO, /* data read IOs */ 1078 FS_GDATA_READ_IO, /* data read IOs from background gc */ 1079 FS_CDATA_READ_IO, /* compressed data read IOs */ 1080 FS_NODE_READ_IO, /* node read IOs */ 1081 FS_META_READ_IO, /* meta read IOs */ 1082 1083 /* other */ 1084 FS_DISCARD, /* discard */ 1085 NR_IO_TYPE, 1086 }; 1087 1088 struct f2fs_io_info { 1089 struct f2fs_sb_info *sbi; /* f2fs_sb_info pointer */ 1090 nid_t ino; /* inode number */ 1091 enum page_type type; /* contains DATA/NODE/META/META_FLUSH */ 1092 enum temp_type temp; /* contains HOT/WARM/COLD */ 1093 int op; /* contains REQ_OP_ */ 1094 int op_flags; /* req_flag_bits */ 1095 block_t new_blkaddr; /* new block address to be written */ 1096 block_t old_blkaddr; /* old block address before Cow */ 1097 struct page *page; /* page to be written */ 1098 struct page *encrypted_page; /* encrypted page */ 1099 struct page *compressed_page; /* compressed page */ 1100 struct list_head list; /* serialize IOs */ 1101 bool submitted; /* indicate IO submission */ 1102 int need_lock; /* indicate we need to lock cp_rwsem */ 1103 bool in_list; /* indicate fio is in io_list */ 1104 bool is_por; /* indicate IO is from recovery or not */ 1105 bool retry; /* need to reallocate block address */ 1106 int compr_blocks; /* # of compressed block addresses */ 1107 bool encrypted; /* indicate file is encrypted */ 1108 enum iostat_type io_type; /* io type */ 1109 struct writeback_control *io_wbc; /* writeback control */ 1110 struct bio **bio; /* bio for ipu */ 1111 sector_t *last_block; /* last block number in bio */ 1112 unsigned char version; /* version of the node */ 1113 }; 1114 1115 struct bio_entry { 1116 struct bio *bio; 1117 struct list_head list; 1118 }; 1119 1120 #define is_read_io(rw) ((rw) == READ) 1121 struct f2fs_bio_info { 1122 struct f2fs_sb_info *sbi; /* f2fs superblock */ 1123 struct bio *bio; /* bios to merge */ 1124 sector_t last_block_in_bio; /* last block number */ 1125 struct f2fs_io_info fio; /* store buffered io info. */ 1126 struct rw_semaphore io_rwsem; /* blocking op for bio */ 1127 spinlock_t io_lock; /* serialize DATA/NODE IOs */ 1128 struct list_head io_list; /* track fios */ 1129 struct list_head bio_list; /* bio entry list head */ 1130 struct rw_semaphore bio_list_lock; /* lock to protect bio entry list */ 1131 }; 1132 1133 #define FDEV(i) (sbi->devs[i]) 1134 #define RDEV(i) (raw_super->devs[i]) 1135 struct f2fs_dev_info { 1136 struct block_device *bdev; 1137 char path[MAX_PATH_LEN]; 1138 unsigned int total_segments; 1139 block_t start_blk; 1140 block_t end_blk; 1141 #ifdef CONFIG_BLK_DEV_ZONED 1142 unsigned int nr_blkz; /* Total number of zones */ 1143 unsigned long *blkz_seq; /* Bitmap indicating sequential zones */ 1144 block_t *zone_capacity_blocks; /* Array of zone capacity in blks */ 1145 #endif 1146 }; 1147 1148 enum inode_type { 1149 DIR_INODE, /* for dirty dir inode */ 1150 FILE_INODE, /* for dirty regular/symlink inode */ 1151 DIRTY_META, /* for all dirtied inode metadata */ 1152 ATOMIC_FILE, /* for all atomic files */ 1153 NR_INODE_TYPE, 1154 }; 1155 1156 /* for inner inode cache management */ 1157 struct inode_management { 1158 struct radix_tree_root ino_root; /* ino entry array */ 1159 spinlock_t ino_lock; /* for ino entry lock */ 1160 struct list_head ino_list; /* inode list head */ 1161 unsigned long ino_num; /* number of entries */ 1162 }; 1163 1164 /* for GC_AT */ 1165 struct atgc_management { 1166 bool atgc_enabled; /* ATGC is enabled or not */ 1167 struct rb_root_cached root; /* root of victim rb-tree */ 1168 struct list_head victim_list; /* linked with all victim entries */ 1169 unsigned int victim_count; /* victim count in rb-tree */ 1170 unsigned int candidate_ratio; /* candidate ratio */ 1171 unsigned int max_candidate_count; /* max candidate count */ 1172 unsigned int age_weight; /* age weight, vblock_weight = 100 - age_weight */ 1173 unsigned long long age_threshold; /* age threshold */ 1174 }; 1175 1176 /* For s_flag in struct f2fs_sb_info */ 1177 enum { 1178 SBI_IS_DIRTY, /* dirty flag for checkpoint */ 1179 SBI_IS_CLOSE, /* specify unmounting */ 1180 SBI_NEED_FSCK, /* need fsck.f2fs to fix */ 1181 SBI_POR_DOING, /* recovery is doing or not */ 1182 SBI_NEED_SB_WRITE, /* need to recover superblock */ 1183 SBI_NEED_CP, /* need to checkpoint */ 1184 SBI_IS_SHUTDOWN, /* shutdown by ioctl */ 1185 SBI_IS_RECOVERED, /* recovered orphan/data */ 1186 SBI_CP_DISABLED, /* CP was disabled last mount */ 1187 SBI_CP_DISABLED_QUICK, /* CP was disabled quickly */ 1188 SBI_QUOTA_NEED_FLUSH, /* need to flush quota info in CP */ 1189 SBI_QUOTA_SKIP_FLUSH, /* skip flushing quota in current CP */ 1190 SBI_QUOTA_NEED_REPAIR, /* quota file may be corrupted */ 1191 SBI_IS_RESIZEFS, /* resizefs is in process */ 1192 }; 1193 1194 enum { 1195 CP_TIME, 1196 REQ_TIME, 1197 DISCARD_TIME, 1198 GC_TIME, 1199 DISABLE_TIME, 1200 UMOUNT_DISCARD_TIMEOUT, 1201 MAX_TIME, 1202 }; 1203 1204 enum { 1205 GC_NORMAL, 1206 GC_IDLE_CB, 1207 GC_IDLE_GREEDY, 1208 GC_IDLE_AT, 1209 GC_URGENT_HIGH, 1210 GC_URGENT_LOW, 1211 }; 1212 1213 enum { 1214 BGGC_MODE_ON, /* background gc is on */ 1215 BGGC_MODE_OFF, /* background gc is off */ 1216 BGGC_MODE_SYNC, /* 1217 * background gc is on, migrating blocks 1218 * like foreground gc 1219 */ 1220 }; 1221 1222 enum { 1223 FS_MODE_ADAPTIVE, /* use both lfs/ssr allocation */ 1224 FS_MODE_LFS, /* use lfs allocation only */ 1225 }; 1226 1227 enum { 1228 WHINT_MODE_OFF, /* not pass down write hints */ 1229 WHINT_MODE_USER, /* try to pass down hints given by users */ 1230 WHINT_MODE_FS, /* pass down hints with F2FS policy */ 1231 }; 1232 1233 enum { 1234 ALLOC_MODE_DEFAULT, /* stay default */ 1235 ALLOC_MODE_REUSE, /* reuse segments as much as possible */ 1236 }; 1237 1238 enum fsync_mode { 1239 FSYNC_MODE_POSIX, /* fsync follows posix semantics */ 1240 FSYNC_MODE_STRICT, /* fsync behaves in line with ext4 */ 1241 FSYNC_MODE_NOBARRIER, /* fsync behaves nobarrier based on posix */ 1242 }; 1243 1244 /* 1245 * this value is set in page as a private data which indicate that 1246 * the page is atomically written, and it is in inmem_pages list. 1247 */ 1248 #define ATOMIC_WRITTEN_PAGE ((unsigned long)-1) 1249 #define DUMMY_WRITTEN_PAGE ((unsigned long)-2) 1250 1251 #define IS_ATOMIC_WRITTEN_PAGE(page) \ 1252 (page_private(page) == ATOMIC_WRITTEN_PAGE) 1253 #define IS_DUMMY_WRITTEN_PAGE(page) \ 1254 (page_private(page) == DUMMY_WRITTEN_PAGE) 1255 1256 #ifdef CONFIG_F2FS_IO_TRACE 1257 #define IS_IO_TRACED_PAGE(page) \ 1258 (page_private(page) > 0 && \ 1259 page_private(page) < (unsigned long)PID_MAX_LIMIT) 1260 #else 1261 #define IS_IO_TRACED_PAGE(page) (0) 1262 #endif 1263 1264 /* For compression */ 1265 enum compress_algorithm_type { 1266 COMPRESS_LZO, 1267 COMPRESS_LZ4, 1268 COMPRESS_ZSTD, 1269 COMPRESS_LZORLE, 1270 COMPRESS_MAX, 1271 }; 1272 1273 #define COMPRESS_DATA_RESERVED_SIZE 5 1274 struct compress_data { 1275 __le32 clen; /* compressed data size */ 1276 __le32 reserved[COMPRESS_DATA_RESERVED_SIZE]; /* reserved */ 1277 u8 cdata[]; /* compressed data */ 1278 }; 1279 1280 #define COMPRESS_HEADER_SIZE (sizeof(struct compress_data)) 1281 1282 #define F2FS_COMPRESSED_PAGE_MAGIC 0xF5F2C000 1283 1284 /* compress context */ 1285 struct compress_ctx { 1286 struct inode *inode; /* inode the context belong to */ 1287 pgoff_t cluster_idx; /* cluster index number */ 1288 unsigned int cluster_size; /* page count in cluster */ 1289 unsigned int log_cluster_size; /* log of cluster size */ 1290 struct page **rpages; /* pages store raw data in cluster */ 1291 unsigned int nr_rpages; /* total page number in rpages */ 1292 struct page **cpages; /* pages store compressed data in cluster */ 1293 unsigned int nr_cpages; /* total page number in cpages */ 1294 void *rbuf; /* virtual mapped address on rpages */ 1295 struct compress_data *cbuf; /* virtual mapped address on cpages */ 1296 size_t rlen; /* valid data length in rbuf */ 1297 size_t clen; /* valid data length in cbuf */ 1298 void *private; /* payload buffer for specified compression algorithm */ 1299 void *private2; /* extra payload buffer */ 1300 }; 1301 1302 /* compress context for write IO path */ 1303 struct compress_io_ctx { 1304 u32 magic; /* magic number to indicate page is compressed */ 1305 struct inode *inode; /* inode the context belong to */ 1306 struct page **rpages; /* pages store raw data in cluster */ 1307 unsigned int nr_rpages; /* total page number in rpages */ 1308 atomic_t pending_pages; /* in-flight compressed page count */ 1309 }; 1310 1311 /* decompress io context for read IO path */ 1312 struct decompress_io_ctx { 1313 u32 magic; /* magic number to indicate page is compressed */ 1314 struct inode *inode; /* inode the context belong to */ 1315 pgoff_t cluster_idx; /* cluster index number */ 1316 unsigned int cluster_size; /* page count in cluster */ 1317 unsigned int log_cluster_size; /* log of cluster size */ 1318 struct page **rpages; /* pages store raw data in cluster */ 1319 unsigned int nr_rpages; /* total page number in rpages */ 1320 struct page **cpages; /* pages store compressed data in cluster */ 1321 unsigned int nr_cpages; /* total page number in cpages */ 1322 struct page **tpages; /* temp pages to pad holes in cluster */ 1323 void *rbuf; /* virtual mapped address on rpages */ 1324 struct compress_data *cbuf; /* virtual mapped address on cpages */ 1325 size_t rlen; /* valid data length in rbuf */ 1326 size_t clen; /* valid data length in cbuf */ 1327 atomic_t pending_pages; /* in-flight compressed page count */ 1328 bool failed; /* indicate IO error during decompression */ 1329 void *private; /* payload buffer for specified decompression algorithm */ 1330 void *private2; /* extra payload buffer */ 1331 }; 1332 1333 #define NULL_CLUSTER ((unsigned int)(~0)) 1334 #define MIN_COMPRESS_LOG_SIZE 2 1335 #define MAX_COMPRESS_LOG_SIZE 8 1336 #define MAX_COMPRESS_WINDOW_SIZE(log_size) ((PAGE_SIZE) << (log_size)) 1337 1338 struct f2fs_sb_info { 1339 struct super_block *sb; /* pointer to VFS super block */ 1340 struct proc_dir_entry *s_proc; /* proc entry */ 1341 struct f2fs_super_block *raw_super; /* raw super block pointer */ 1342 struct rw_semaphore sb_lock; /* lock for raw super block */ 1343 int valid_super_block; /* valid super block no */ 1344 unsigned long s_flag; /* flags for sbi */ 1345 struct mutex writepages; /* mutex for writepages() */ 1346 1347 #ifdef CONFIG_BLK_DEV_ZONED 1348 unsigned int blocks_per_blkz; /* F2FS blocks per zone */ 1349 unsigned int log_blocks_per_blkz; /* log2 F2FS blocks per zone */ 1350 #endif 1351 1352 /* for node-related operations */ 1353 struct f2fs_nm_info *nm_info; /* node manager */ 1354 struct inode *node_inode; /* cache node blocks */ 1355 1356 /* for segment-related operations */ 1357 struct f2fs_sm_info *sm_info; /* segment manager */ 1358 1359 /* for bio operations */ 1360 struct f2fs_bio_info *write_io[NR_PAGE_TYPE]; /* for write bios */ 1361 /* keep migration IO order for LFS mode */ 1362 struct rw_semaphore io_order_lock; 1363 mempool_t *write_io_dummy; /* Dummy pages */ 1364 1365 /* for checkpoint */ 1366 struct f2fs_checkpoint *ckpt; /* raw checkpoint pointer */ 1367 int cur_cp_pack; /* remain current cp pack */ 1368 spinlock_t cp_lock; /* for flag in ckpt */ 1369 struct inode *meta_inode; /* cache meta blocks */ 1370 struct mutex cp_mutex; /* checkpoint procedure lock */ 1371 struct rw_semaphore cp_rwsem; /* blocking FS operations */ 1372 struct rw_semaphore node_write; /* locking node writes */ 1373 struct rw_semaphore node_change; /* locking node change */ 1374 wait_queue_head_t cp_wait; 1375 unsigned long last_time[MAX_TIME]; /* to store time in jiffies */ 1376 long interval_time[MAX_TIME]; /* to store thresholds */ 1377 1378 struct inode_management im[MAX_INO_ENTRY]; /* manage inode cache */ 1379 1380 spinlock_t fsync_node_lock; /* for node entry lock */ 1381 struct list_head fsync_node_list; /* node list head */ 1382 unsigned int fsync_seg_id; /* sequence id */ 1383 unsigned int fsync_node_num; /* number of node entries */ 1384 1385 /* for orphan inode, use 0'th array */ 1386 unsigned int max_orphans; /* max orphan inodes */ 1387 1388 /* for inode management */ 1389 struct list_head inode_list[NR_INODE_TYPE]; /* dirty inode list */ 1390 spinlock_t inode_lock[NR_INODE_TYPE]; /* for dirty inode list lock */ 1391 struct mutex flush_lock; /* for flush exclusion */ 1392 1393 /* for extent tree cache */ 1394 struct radix_tree_root extent_tree_root;/* cache extent cache entries */ 1395 struct mutex extent_tree_lock; /* locking extent radix tree */ 1396 struct list_head extent_list; /* lru list for shrinker */ 1397 spinlock_t extent_lock; /* locking extent lru list */ 1398 atomic_t total_ext_tree; /* extent tree count */ 1399 struct list_head zombie_list; /* extent zombie tree list */ 1400 atomic_t total_zombie_tree; /* extent zombie tree count */ 1401 atomic_t total_ext_node; /* extent info count */ 1402 1403 /* basic filesystem units */ 1404 unsigned int log_sectors_per_block; /* log2 sectors per block */ 1405 unsigned int log_blocksize; /* log2 block size */ 1406 unsigned int blocksize; /* block size */ 1407 unsigned int root_ino_num; /* root inode number*/ 1408 unsigned int node_ino_num; /* node inode number*/ 1409 unsigned int meta_ino_num; /* meta inode number*/ 1410 unsigned int log_blocks_per_seg; /* log2 blocks per segment */ 1411 unsigned int blocks_per_seg; /* blocks per segment */ 1412 unsigned int segs_per_sec; /* segments per section */ 1413 unsigned int secs_per_zone; /* sections per zone */ 1414 unsigned int total_sections; /* total section count */ 1415 unsigned int total_node_count; /* total node block count */ 1416 unsigned int total_valid_node_count; /* valid node block count */ 1417 loff_t max_file_blocks; /* max block index of file */ 1418 int dir_level; /* directory level */ 1419 int readdir_ra; /* readahead inode in readdir */ 1420 1421 block_t user_block_count; /* # of user blocks */ 1422 block_t total_valid_block_count; /* # of valid blocks */ 1423 block_t discard_blks; /* discard command candidats */ 1424 block_t last_valid_block_count; /* for recovery */ 1425 block_t reserved_blocks; /* configurable reserved blocks */ 1426 block_t current_reserved_blocks; /* current reserved blocks */ 1427 1428 /* Additional tracking for no checkpoint mode */ 1429 block_t unusable_block_count; /* # of blocks saved by last cp */ 1430 1431 unsigned int nquota_files; /* # of quota sysfile */ 1432 struct rw_semaphore quota_sem; /* blocking cp for flags */ 1433 1434 /* # of pages, see count_type */ 1435 atomic_t nr_pages[NR_COUNT_TYPE]; 1436 /* # of allocated blocks */ 1437 struct percpu_counter alloc_valid_block_count; 1438 1439 /* writeback control */ 1440 atomic_t wb_sync_req[META]; /* count # of WB_SYNC threads */ 1441 1442 /* valid inode count */ 1443 struct percpu_counter total_valid_inode_count; 1444 1445 struct f2fs_mount_info mount_opt; /* mount options */ 1446 1447 /* for cleaning operations */ 1448 struct rw_semaphore gc_lock; /* 1449 * semaphore for GC, avoid 1450 * race between GC and GC or CP 1451 */ 1452 struct f2fs_gc_kthread *gc_thread; /* GC thread */ 1453 struct atgc_management am; /* atgc management */ 1454 unsigned int cur_victim_sec; /* current victim section num */ 1455 unsigned int gc_mode; /* current GC state */ 1456 unsigned int next_victim_seg[2]; /* next segment in victim section */ 1457 1458 /* for skip statistic */ 1459 unsigned int atomic_files; /* # of opened atomic file */ 1460 unsigned long long skipped_atomic_files[2]; /* FG_GC and BG_GC */ 1461 unsigned long long skipped_gc_rwsem; /* FG_GC only */ 1462 1463 /* threshold for gc trials on pinned files */ 1464 u64 gc_pin_file_threshold; 1465 struct rw_semaphore pin_sem; 1466 1467 /* maximum # of trials to find a victim segment for SSR and GC */ 1468 unsigned int max_victim_search; 1469 /* migration granularity of garbage collection, unit: segment */ 1470 unsigned int migration_granularity; 1471 1472 /* 1473 * for stat information. 1474 * one is for the LFS mode, and the other is for the SSR mode. 1475 */ 1476 #ifdef CONFIG_F2FS_STAT_FS 1477 struct f2fs_stat_info *stat_info; /* FS status information */ 1478 atomic_t meta_count[META_MAX]; /* # of meta blocks */ 1479 unsigned int segment_count[2]; /* # of allocated segments */ 1480 unsigned int block_count[2]; /* # of allocated blocks */ 1481 atomic_t inplace_count; /* # of inplace update */ 1482 atomic64_t total_hit_ext; /* # of lookup extent cache */ 1483 atomic64_t read_hit_rbtree; /* # of hit rbtree extent node */ 1484 atomic64_t read_hit_largest; /* # of hit largest extent node */ 1485 atomic64_t read_hit_cached; /* # of hit cached extent node */ 1486 atomic_t inline_xattr; /* # of inline_xattr inodes */ 1487 atomic_t inline_inode; /* # of inline_data inodes */ 1488 atomic_t inline_dir; /* # of inline_dentry inodes */ 1489 atomic_t compr_inode; /* # of compressed inodes */ 1490 atomic64_t compr_blocks; /* # of compressed blocks */ 1491 atomic_t vw_cnt; /* # of volatile writes */ 1492 atomic_t max_aw_cnt; /* max # of atomic writes */ 1493 atomic_t max_vw_cnt; /* max # of volatile writes */ 1494 unsigned int io_skip_bggc; /* skip background gc for in-flight IO */ 1495 unsigned int other_skip_bggc; /* skip background gc for other reasons */ 1496 unsigned int ndirty_inode[NR_INODE_TYPE]; /* # of dirty inodes */ 1497 #endif 1498 spinlock_t stat_lock; /* lock for stat operations */ 1499 1500 /* For app/fs IO statistics */ 1501 spinlock_t iostat_lock; 1502 unsigned long long rw_iostat[NR_IO_TYPE]; 1503 unsigned long long prev_rw_iostat[NR_IO_TYPE]; 1504 bool iostat_enable; 1505 unsigned long iostat_next_period; 1506 unsigned int iostat_period_ms; 1507 1508 /* to attach REQ_META|REQ_FUA flags */ 1509 unsigned int data_io_flag; 1510 unsigned int node_io_flag; 1511 1512 /* For sysfs suppport */ 1513 struct kobject s_kobj; 1514 struct completion s_kobj_unregister; 1515 1516 /* For shrinker support */ 1517 struct list_head s_list; 1518 int s_ndevs; /* number of devices */ 1519 struct f2fs_dev_info *devs; /* for device list */ 1520 unsigned int dirty_device; /* for checkpoint data flush */ 1521 spinlock_t dev_lock; /* protect dirty_device */ 1522 struct mutex umount_mutex; 1523 unsigned int shrinker_run_no; 1524 1525 /* For write statistics */ 1526 u64 sectors_written_start; 1527 u64 kbytes_written; 1528 1529 /* Reference to checksum algorithm driver via cryptoapi */ 1530 struct crypto_shash *s_chksum_driver; 1531 1532 /* Precomputed FS UUID checksum for seeding other checksums */ 1533 __u32 s_chksum_seed; 1534 1535 struct workqueue_struct *post_read_wq; /* post read workqueue */ 1536 1537 struct kmem_cache *inline_xattr_slab; /* inline xattr entry */ 1538 unsigned int inline_xattr_slab_size; /* default inline xattr slab size */ 1539 1540 #ifdef CONFIG_F2FS_FS_COMPRESSION 1541 struct kmem_cache *page_array_slab; /* page array entry */ 1542 unsigned int page_array_slab_size; /* default page array slab size */ 1543 #endif 1544 }; 1545 1546 struct f2fs_private_dio { 1547 struct inode *inode; 1548 void *orig_private; 1549 bio_end_io_t *orig_end_io; 1550 bool write; 1551 }; 1552 1553 #ifdef CONFIG_F2FS_FAULT_INJECTION 1554 #define f2fs_show_injection_info(sbi, type) \ 1555 printk_ratelimited("%sF2FS-fs (%s) : inject %s in %s of %pS\n", \ 1556 KERN_INFO, sbi->sb->s_id, \ 1557 f2fs_fault_name[type], \ 1558 __func__, __builtin_return_address(0)) 1559 static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type) 1560 { 1561 struct f2fs_fault_info *ffi = &F2FS_OPTION(sbi).fault_info; 1562 1563 if (!ffi->inject_rate) 1564 return false; 1565 1566 if (!IS_FAULT_SET(ffi, type)) 1567 return false; 1568 1569 atomic_inc(&ffi->inject_ops); 1570 if (atomic_read(&ffi->inject_ops) >= ffi->inject_rate) { 1571 atomic_set(&ffi->inject_ops, 0); 1572 return true; 1573 } 1574 return false; 1575 } 1576 #else 1577 #define f2fs_show_injection_info(sbi, type) do { } while (0) 1578 static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type) 1579 { 1580 return false; 1581 } 1582 #endif 1583 1584 /* 1585 * Test if the mounted volume is a multi-device volume. 1586 * - For a single regular disk volume, sbi->s_ndevs is 0. 1587 * - For a single zoned disk volume, sbi->s_ndevs is 1. 1588 * - For a multi-device volume, sbi->s_ndevs is always 2 or more. 1589 */ 1590 static inline bool f2fs_is_multi_device(struct f2fs_sb_info *sbi) 1591 { 1592 return sbi->s_ndevs > 1; 1593 } 1594 1595 /* For write statistics. Suppose sector size is 512 bytes, 1596 * and the return value is in kbytes. s is of struct f2fs_sb_info. 1597 */ 1598 #define BD_PART_WRITTEN(s) \ 1599 (((u64)part_stat_read((s)->sb->s_bdev->bd_part, sectors[STAT_WRITE]) - \ 1600 (s)->sectors_written_start) >> 1) 1601 1602 static inline void f2fs_update_time(struct f2fs_sb_info *sbi, int type) 1603 { 1604 unsigned long now = jiffies; 1605 1606 sbi->last_time[type] = now; 1607 1608 /* DISCARD_TIME and GC_TIME are based on REQ_TIME */ 1609 if (type == REQ_TIME) { 1610 sbi->last_time[DISCARD_TIME] = now; 1611 sbi->last_time[GC_TIME] = now; 1612 } 1613 } 1614 1615 static inline bool f2fs_time_over(struct f2fs_sb_info *sbi, int type) 1616 { 1617 unsigned long interval = sbi->interval_time[type] * HZ; 1618 1619 return time_after(jiffies, sbi->last_time[type] + interval); 1620 } 1621 1622 static inline unsigned int f2fs_time_to_wait(struct f2fs_sb_info *sbi, 1623 int type) 1624 { 1625 unsigned long interval = sbi->interval_time[type] * HZ; 1626 unsigned int wait_ms = 0; 1627 long delta; 1628 1629 delta = (sbi->last_time[type] + interval) - jiffies; 1630 if (delta > 0) 1631 wait_ms = jiffies_to_msecs(delta); 1632 1633 return wait_ms; 1634 } 1635 1636 /* 1637 * Inline functions 1638 */ 1639 static inline u32 __f2fs_crc32(struct f2fs_sb_info *sbi, u32 crc, 1640 const void *address, unsigned int length) 1641 { 1642 struct { 1643 struct shash_desc shash; 1644 char ctx[4]; 1645 } desc; 1646 int err; 1647 1648 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver) != sizeof(desc.ctx)); 1649 1650 desc.shash.tfm = sbi->s_chksum_driver; 1651 *(u32 *)desc.ctx = crc; 1652 1653 err = crypto_shash_update(&desc.shash, address, length); 1654 BUG_ON(err); 1655 1656 return *(u32 *)desc.ctx; 1657 } 1658 1659 static inline u32 f2fs_crc32(struct f2fs_sb_info *sbi, const void *address, 1660 unsigned int length) 1661 { 1662 return __f2fs_crc32(sbi, F2FS_SUPER_MAGIC, address, length); 1663 } 1664 1665 static inline bool f2fs_crc_valid(struct f2fs_sb_info *sbi, __u32 blk_crc, 1666 void *buf, size_t buf_size) 1667 { 1668 return f2fs_crc32(sbi, buf, buf_size) == blk_crc; 1669 } 1670 1671 static inline u32 f2fs_chksum(struct f2fs_sb_info *sbi, u32 crc, 1672 const void *address, unsigned int length) 1673 { 1674 return __f2fs_crc32(sbi, crc, address, length); 1675 } 1676 1677 static inline struct f2fs_inode_info *F2FS_I(struct inode *inode) 1678 { 1679 return container_of(inode, struct f2fs_inode_info, vfs_inode); 1680 } 1681 1682 static inline struct f2fs_sb_info *F2FS_SB(struct super_block *sb) 1683 { 1684 return sb->s_fs_info; 1685 } 1686 1687 static inline struct f2fs_sb_info *F2FS_I_SB(struct inode *inode) 1688 { 1689 return F2FS_SB(inode->i_sb); 1690 } 1691 1692 static inline struct f2fs_sb_info *F2FS_M_SB(struct address_space *mapping) 1693 { 1694 return F2FS_I_SB(mapping->host); 1695 } 1696 1697 static inline struct f2fs_sb_info *F2FS_P_SB(struct page *page) 1698 { 1699 return F2FS_M_SB(page_file_mapping(page)); 1700 } 1701 1702 static inline struct f2fs_super_block *F2FS_RAW_SUPER(struct f2fs_sb_info *sbi) 1703 { 1704 return (struct f2fs_super_block *)(sbi->raw_super); 1705 } 1706 1707 static inline struct f2fs_checkpoint *F2FS_CKPT(struct f2fs_sb_info *sbi) 1708 { 1709 return (struct f2fs_checkpoint *)(sbi->ckpt); 1710 } 1711 1712 static inline struct f2fs_node *F2FS_NODE(struct page *page) 1713 { 1714 return (struct f2fs_node *)page_address(page); 1715 } 1716 1717 static inline struct f2fs_inode *F2FS_INODE(struct page *page) 1718 { 1719 return &((struct f2fs_node *)page_address(page))->i; 1720 } 1721 1722 static inline struct f2fs_nm_info *NM_I(struct f2fs_sb_info *sbi) 1723 { 1724 return (struct f2fs_nm_info *)(sbi->nm_info); 1725 } 1726 1727 static inline struct f2fs_sm_info *SM_I(struct f2fs_sb_info *sbi) 1728 { 1729 return (struct f2fs_sm_info *)(sbi->sm_info); 1730 } 1731 1732 static inline struct sit_info *SIT_I(struct f2fs_sb_info *sbi) 1733 { 1734 return (struct sit_info *)(SM_I(sbi)->sit_info); 1735 } 1736 1737 static inline struct free_segmap_info *FREE_I(struct f2fs_sb_info *sbi) 1738 { 1739 return (struct free_segmap_info *)(SM_I(sbi)->free_info); 1740 } 1741 1742 static inline struct dirty_seglist_info *DIRTY_I(struct f2fs_sb_info *sbi) 1743 { 1744 return (struct dirty_seglist_info *)(SM_I(sbi)->dirty_info); 1745 } 1746 1747 static inline struct address_space *META_MAPPING(struct f2fs_sb_info *sbi) 1748 { 1749 return sbi->meta_inode->i_mapping; 1750 } 1751 1752 static inline struct address_space *NODE_MAPPING(struct f2fs_sb_info *sbi) 1753 { 1754 return sbi->node_inode->i_mapping; 1755 } 1756 1757 static inline bool is_sbi_flag_set(struct f2fs_sb_info *sbi, unsigned int type) 1758 { 1759 return test_bit(type, &sbi->s_flag); 1760 } 1761 1762 static inline void set_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type) 1763 { 1764 set_bit(type, &sbi->s_flag); 1765 } 1766 1767 static inline void clear_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type) 1768 { 1769 clear_bit(type, &sbi->s_flag); 1770 } 1771 1772 static inline unsigned long long cur_cp_version(struct f2fs_checkpoint *cp) 1773 { 1774 return le64_to_cpu(cp->checkpoint_ver); 1775 } 1776 1777 static inline unsigned long f2fs_qf_ino(struct super_block *sb, int type) 1778 { 1779 if (type < F2FS_MAX_QUOTAS) 1780 return le32_to_cpu(F2FS_SB(sb)->raw_super->qf_ino[type]); 1781 return 0; 1782 } 1783 1784 static inline __u64 cur_cp_crc(struct f2fs_checkpoint *cp) 1785 { 1786 size_t crc_offset = le32_to_cpu(cp->checksum_offset); 1787 return le32_to_cpu(*((__le32 *)((unsigned char *)cp + crc_offset))); 1788 } 1789 1790 static inline bool __is_set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 1791 { 1792 unsigned int ckpt_flags = le32_to_cpu(cp->ckpt_flags); 1793 1794 return ckpt_flags & f; 1795 } 1796 1797 static inline bool is_set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 1798 { 1799 return __is_set_ckpt_flags(F2FS_CKPT(sbi), f); 1800 } 1801 1802 static inline void __set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 1803 { 1804 unsigned int ckpt_flags; 1805 1806 ckpt_flags = le32_to_cpu(cp->ckpt_flags); 1807 ckpt_flags |= f; 1808 cp->ckpt_flags = cpu_to_le32(ckpt_flags); 1809 } 1810 1811 static inline void set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 1812 { 1813 unsigned long flags; 1814 1815 spin_lock_irqsave(&sbi->cp_lock, flags); 1816 __set_ckpt_flags(F2FS_CKPT(sbi), f); 1817 spin_unlock_irqrestore(&sbi->cp_lock, flags); 1818 } 1819 1820 static inline void __clear_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 1821 { 1822 unsigned int ckpt_flags; 1823 1824 ckpt_flags = le32_to_cpu(cp->ckpt_flags); 1825 ckpt_flags &= (~f); 1826 cp->ckpt_flags = cpu_to_le32(ckpt_flags); 1827 } 1828 1829 static inline void clear_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 1830 { 1831 unsigned long flags; 1832 1833 spin_lock_irqsave(&sbi->cp_lock, flags); 1834 __clear_ckpt_flags(F2FS_CKPT(sbi), f); 1835 spin_unlock_irqrestore(&sbi->cp_lock, flags); 1836 } 1837 1838 static inline void disable_nat_bits(struct f2fs_sb_info *sbi, bool lock) 1839 { 1840 unsigned long flags; 1841 unsigned char *nat_bits; 1842 1843 /* 1844 * In order to re-enable nat_bits we need to call fsck.f2fs by 1845 * set_sbi_flag(sbi, SBI_NEED_FSCK). But it may give huge cost, 1846 * so let's rely on regular fsck or unclean shutdown. 1847 */ 1848 1849 if (lock) 1850 spin_lock_irqsave(&sbi->cp_lock, flags); 1851 __clear_ckpt_flags(F2FS_CKPT(sbi), CP_NAT_BITS_FLAG); 1852 nat_bits = NM_I(sbi)->nat_bits; 1853 NM_I(sbi)->nat_bits = NULL; 1854 if (lock) 1855 spin_unlock_irqrestore(&sbi->cp_lock, flags); 1856 1857 kvfree(nat_bits); 1858 } 1859 1860 static inline bool enabled_nat_bits(struct f2fs_sb_info *sbi, 1861 struct cp_control *cpc) 1862 { 1863 bool set = is_set_ckpt_flags(sbi, CP_NAT_BITS_FLAG); 1864 1865 return (cpc) ? (cpc->reason & CP_UMOUNT) && set : set; 1866 } 1867 1868 static inline void f2fs_lock_op(struct f2fs_sb_info *sbi) 1869 { 1870 down_read(&sbi->cp_rwsem); 1871 } 1872 1873 static inline int f2fs_trylock_op(struct f2fs_sb_info *sbi) 1874 { 1875 return down_read_trylock(&sbi->cp_rwsem); 1876 } 1877 1878 static inline void f2fs_unlock_op(struct f2fs_sb_info *sbi) 1879 { 1880 up_read(&sbi->cp_rwsem); 1881 } 1882 1883 static inline void f2fs_lock_all(struct f2fs_sb_info *sbi) 1884 { 1885 down_write(&sbi->cp_rwsem); 1886 } 1887 1888 static inline void f2fs_unlock_all(struct f2fs_sb_info *sbi) 1889 { 1890 up_write(&sbi->cp_rwsem); 1891 } 1892 1893 static inline int __get_cp_reason(struct f2fs_sb_info *sbi) 1894 { 1895 int reason = CP_SYNC; 1896 1897 if (test_opt(sbi, FASTBOOT)) 1898 reason = CP_FASTBOOT; 1899 if (is_sbi_flag_set(sbi, SBI_IS_CLOSE)) 1900 reason = CP_UMOUNT; 1901 return reason; 1902 } 1903 1904 static inline bool __remain_node_summaries(int reason) 1905 { 1906 return (reason & (CP_UMOUNT | CP_FASTBOOT)); 1907 } 1908 1909 static inline bool __exist_node_summaries(struct f2fs_sb_info *sbi) 1910 { 1911 return (is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG) || 1912 is_set_ckpt_flags(sbi, CP_FASTBOOT_FLAG)); 1913 } 1914 1915 /* 1916 * Check whether the inode has blocks or not 1917 */ 1918 static inline int F2FS_HAS_BLOCKS(struct inode *inode) 1919 { 1920 block_t xattr_block = F2FS_I(inode)->i_xattr_nid ? 1 : 0; 1921 1922 return (inode->i_blocks >> F2FS_LOG_SECTORS_PER_BLOCK) > xattr_block; 1923 } 1924 1925 static inline bool f2fs_has_xattr_block(unsigned int ofs) 1926 { 1927 return ofs == XATTR_NODE_OFFSET; 1928 } 1929 1930 static inline bool __allow_reserved_blocks(struct f2fs_sb_info *sbi, 1931 struct inode *inode, bool cap) 1932 { 1933 if (!inode) 1934 return true; 1935 if (!test_opt(sbi, RESERVE_ROOT)) 1936 return false; 1937 if (IS_NOQUOTA(inode)) 1938 return true; 1939 if (uid_eq(F2FS_OPTION(sbi).s_resuid, current_fsuid())) 1940 return true; 1941 if (!gid_eq(F2FS_OPTION(sbi).s_resgid, GLOBAL_ROOT_GID) && 1942 in_group_p(F2FS_OPTION(sbi).s_resgid)) 1943 return true; 1944 if (cap && capable(CAP_SYS_RESOURCE)) 1945 return true; 1946 return false; 1947 } 1948 1949 static inline void f2fs_i_blocks_write(struct inode *, block_t, bool, bool); 1950 static inline int inc_valid_block_count(struct f2fs_sb_info *sbi, 1951 struct inode *inode, blkcnt_t *count) 1952 { 1953 blkcnt_t diff = 0, release = 0; 1954 block_t avail_user_block_count; 1955 int ret; 1956 1957 ret = dquot_reserve_block(inode, *count); 1958 if (ret) 1959 return ret; 1960 1961 if (time_to_inject(sbi, FAULT_BLOCK)) { 1962 f2fs_show_injection_info(sbi, FAULT_BLOCK); 1963 release = *count; 1964 goto release_quota; 1965 } 1966 1967 /* 1968 * let's increase this in prior to actual block count change in order 1969 * for f2fs_sync_file to avoid data races when deciding checkpoint. 1970 */ 1971 percpu_counter_add(&sbi->alloc_valid_block_count, (*count)); 1972 1973 spin_lock(&sbi->stat_lock); 1974 sbi->total_valid_block_count += (block_t)(*count); 1975 avail_user_block_count = sbi->user_block_count - 1976 sbi->current_reserved_blocks; 1977 1978 if (!__allow_reserved_blocks(sbi, inode, true)) 1979 avail_user_block_count -= F2FS_OPTION(sbi).root_reserved_blocks; 1980 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) { 1981 if (avail_user_block_count > sbi->unusable_block_count) 1982 avail_user_block_count -= sbi->unusable_block_count; 1983 else 1984 avail_user_block_count = 0; 1985 } 1986 if (unlikely(sbi->total_valid_block_count > avail_user_block_count)) { 1987 diff = sbi->total_valid_block_count - avail_user_block_count; 1988 if (diff > *count) 1989 diff = *count; 1990 *count -= diff; 1991 release = diff; 1992 sbi->total_valid_block_count -= diff; 1993 if (!*count) { 1994 spin_unlock(&sbi->stat_lock); 1995 goto enospc; 1996 } 1997 } 1998 spin_unlock(&sbi->stat_lock); 1999 2000 if (unlikely(release)) { 2001 percpu_counter_sub(&sbi->alloc_valid_block_count, release); 2002 dquot_release_reservation_block(inode, release); 2003 } 2004 f2fs_i_blocks_write(inode, *count, true, true); 2005 return 0; 2006 2007 enospc: 2008 percpu_counter_sub(&sbi->alloc_valid_block_count, release); 2009 release_quota: 2010 dquot_release_reservation_block(inode, release); 2011 return -ENOSPC; 2012 } 2013 2014 __printf(2, 3) 2015 void f2fs_printk(struct f2fs_sb_info *sbi, const char *fmt, ...); 2016 2017 #define f2fs_err(sbi, fmt, ...) \ 2018 f2fs_printk(sbi, KERN_ERR fmt, ##__VA_ARGS__) 2019 #define f2fs_warn(sbi, fmt, ...) \ 2020 f2fs_printk(sbi, KERN_WARNING fmt, ##__VA_ARGS__) 2021 #define f2fs_notice(sbi, fmt, ...) \ 2022 f2fs_printk(sbi, KERN_NOTICE fmt, ##__VA_ARGS__) 2023 #define f2fs_info(sbi, fmt, ...) \ 2024 f2fs_printk(sbi, KERN_INFO fmt, ##__VA_ARGS__) 2025 #define f2fs_debug(sbi, fmt, ...) \ 2026 f2fs_printk(sbi, KERN_DEBUG fmt, ##__VA_ARGS__) 2027 2028 static inline void dec_valid_block_count(struct f2fs_sb_info *sbi, 2029 struct inode *inode, 2030 block_t count) 2031 { 2032 blkcnt_t sectors = count << F2FS_LOG_SECTORS_PER_BLOCK; 2033 2034 spin_lock(&sbi->stat_lock); 2035 f2fs_bug_on(sbi, sbi->total_valid_block_count < (block_t) count); 2036 sbi->total_valid_block_count -= (block_t)count; 2037 if (sbi->reserved_blocks && 2038 sbi->current_reserved_blocks < sbi->reserved_blocks) 2039 sbi->current_reserved_blocks = min(sbi->reserved_blocks, 2040 sbi->current_reserved_blocks + count); 2041 spin_unlock(&sbi->stat_lock); 2042 if (unlikely(inode->i_blocks < sectors)) { 2043 f2fs_warn(sbi, "Inconsistent i_blocks, ino:%lu, iblocks:%llu, sectors:%llu", 2044 inode->i_ino, 2045 (unsigned long long)inode->i_blocks, 2046 (unsigned long long)sectors); 2047 set_sbi_flag(sbi, SBI_NEED_FSCK); 2048 return; 2049 } 2050 f2fs_i_blocks_write(inode, count, false, true); 2051 } 2052 2053 static inline void inc_page_count(struct f2fs_sb_info *sbi, int count_type) 2054 { 2055 atomic_inc(&sbi->nr_pages[count_type]); 2056 2057 if (count_type == F2FS_DIRTY_DENTS || 2058 count_type == F2FS_DIRTY_NODES || 2059 count_type == F2FS_DIRTY_META || 2060 count_type == F2FS_DIRTY_QDATA || 2061 count_type == F2FS_DIRTY_IMETA) 2062 set_sbi_flag(sbi, SBI_IS_DIRTY); 2063 } 2064 2065 static inline void inode_inc_dirty_pages(struct inode *inode) 2066 { 2067 atomic_inc(&F2FS_I(inode)->dirty_pages); 2068 inc_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ? 2069 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA); 2070 if (IS_NOQUOTA(inode)) 2071 inc_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA); 2072 } 2073 2074 static inline void dec_page_count(struct f2fs_sb_info *sbi, int count_type) 2075 { 2076 atomic_dec(&sbi->nr_pages[count_type]); 2077 } 2078 2079 static inline void inode_dec_dirty_pages(struct inode *inode) 2080 { 2081 if (!S_ISDIR(inode->i_mode) && !S_ISREG(inode->i_mode) && 2082 !S_ISLNK(inode->i_mode)) 2083 return; 2084 2085 atomic_dec(&F2FS_I(inode)->dirty_pages); 2086 dec_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ? 2087 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA); 2088 if (IS_NOQUOTA(inode)) 2089 dec_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA); 2090 } 2091 2092 static inline s64 get_pages(struct f2fs_sb_info *sbi, int count_type) 2093 { 2094 return atomic_read(&sbi->nr_pages[count_type]); 2095 } 2096 2097 static inline int get_dirty_pages(struct inode *inode) 2098 { 2099 return atomic_read(&F2FS_I(inode)->dirty_pages); 2100 } 2101 2102 static inline int get_blocktype_secs(struct f2fs_sb_info *sbi, int block_type) 2103 { 2104 unsigned int pages_per_sec = sbi->segs_per_sec * sbi->blocks_per_seg; 2105 unsigned int segs = (get_pages(sbi, block_type) + pages_per_sec - 1) >> 2106 sbi->log_blocks_per_seg; 2107 2108 return segs / sbi->segs_per_sec; 2109 } 2110 2111 static inline block_t valid_user_blocks(struct f2fs_sb_info *sbi) 2112 { 2113 return sbi->total_valid_block_count; 2114 } 2115 2116 static inline block_t discard_blocks(struct f2fs_sb_info *sbi) 2117 { 2118 return sbi->discard_blks; 2119 } 2120 2121 static inline unsigned long __bitmap_size(struct f2fs_sb_info *sbi, int flag) 2122 { 2123 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi); 2124 2125 /* return NAT or SIT bitmap */ 2126 if (flag == NAT_BITMAP) 2127 return le32_to_cpu(ckpt->nat_ver_bitmap_bytesize); 2128 else if (flag == SIT_BITMAP) 2129 return le32_to_cpu(ckpt->sit_ver_bitmap_bytesize); 2130 2131 return 0; 2132 } 2133 2134 static inline block_t __cp_payload(struct f2fs_sb_info *sbi) 2135 { 2136 return le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_payload); 2137 } 2138 2139 static inline void *__bitmap_ptr(struct f2fs_sb_info *sbi, int flag) 2140 { 2141 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi); 2142 int offset; 2143 2144 if (is_set_ckpt_flags(sbi, CP_LARGE_NAT_BITMAP_FLAG)) { 2145 offset = (flag == SIT_BITMAP) ? 2146 le32_to_cpu(ckpt->nat_ver_bitmap_bytesize) : 0; 2147 /* 2148 * if large_nat_bitmap feature is enabled, leave checksum 2149 * protection for all nat/sit bitmaps. 2150 */ 2151 return &ckpt->sit_nat_version_bitmap + offset + sizeof(__le32); 2152 } 2153 2154 if (__cp_payload(sbi) > 0) { 2155 if (flag == NAT_BITMAP) 2156 return &ckpt->sit_nat_version_bitmap; 2157 else 2158 return (unsigned char *)ckpt + F2FS_BLKSIZE; 2159 } else { 2160 offset = (flag == NAT_BITMAP) ? 2161 le32_to_cpu(ckpt->sit_ver_bitmap_bytesize) : 0; 2162 return &ckpt->sit_nat_version_bitmap + offset; 2163 } 2164 } 2165 2166 static inline block_t __start_cp_addr(struct f2fs_sb_info *sbi) 2167 { 2168 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr); 2169 2170 if (sbi->cur_cp_pack == 2) 2171 start_addr += sbi->blocks_per_seg; 2172 return start_addr; 2173 } 2174 2175 static inline block_t __start_cp_next_addr(struct f2fs_sb_info *sbi) 2176 { 2177 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr); 2178 2179 if (sbi->cur_cp_pack == 1) 2180 start_addr += sbi->blocks_per_seg; 2181 return start_addr; 2182 } 2183 2184 static inline void __set_cp_next_pack(struct f2fs_sb_info *sbi) 2185 { 2186 sbi->cur_cp_pack = (sbi->cur_cp_pack == 1) ? 2 : 1; 2187 } 2188 2189 static inline block_t __start_sum_addr(struct f2fs_sb_info *sbi) 2190 { 2191 return le32_to_cpu(F2FS_CKPT(sbi)->cp_pack_start_sum); 2192 } 2193 2194 static inline int inc_valid_node_count(struct f2fs_sb_info *sbi, 2195 struct inode *inode, bool is_inode) 2196 { 2197 block_t valid_block_count; 2198 unsigned int valid_node_count, user_block_count; 2199 int err; 2200 2201 if (is_inode) { 2202 if (inode) { 2203 err = dquot_alloc_inode(inode); 2204 if (err) 2205 return err; 2206 } 2207 } else { 2208 err = dquot_reserve_block(inode, 1); 2209 if (err) 2210 return err; 2211 } 2212 2213 if (time_to_inject(sbi, FAULT_BLOCK)) { 2214 f2fs_show_injection_info(sbi, FAULT_BLOCK); 2215 goto enospc; 2216 } 2217 2218 spin_lock(&sbi->stat_lock); 2219 2220 valid_block_count = sbi->total_valid_block_count + 2221 sbi->current_reserved_blocks + 1; 2222 2223 if (!__allow_reserved_blocks(sbi, inode, false)) 2224 valid_block_count += F2FS_OPTION(sbi).root_reserved_blocks; 2225 user_block_count = sbi->user_block_count; 2226 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) 2227 user_block_count -= sbi->unusable_block_count; 2228 2229 if (unlikely(valid_block_count > user_block_count)) { 2230 spin_unlock(&sbi->stat_lock); 2231 goto enospc; 2232 } 2233 2234 valid_node_count = sbi->total_valid_node_count + 1; 2235 if (unlikely(valid_node_count > sbi->total_node_count)) { 2236 spin_unlock(&sbi->stat_lock); 2237 goto enospc; 2238 } 2239 2240 sbi->total_valid_node_count++; 2241 sbi->total_valid_block_count++; 2242 spin_unlock(&sbi->stat_lock); 2243 2244 if (inode) { 2245 if (is_inode) 2246 f2fs_mark_inode_dirty_sync(inode, true); 2247 else 2248 f2fs_i_blocks_write(inode, 1, true, true); 2249 } 2250 2251 percpu_counter_inc(&sbi->alloc_valid_block_count); 2252 return 0; 2253 2254 enospc: 2255 if (is_inode) { 2256 if (inode) 2257 dquot_free_inode(inode); 2258 } else { 2259 dquot_release_reservation_block(inode, 1); 2260 } 2261 return -ENOSPC; 2262 } 2263 2264 static inline void dec_valid_node_count(struct f2fs_sb_info *sbi, 2265 struct inode *inode, bool is_inode) 2266 { 2267 spin_lock(&sbi->stat_lock); 2268 2269 f2fs_bug_on(sbi, !sbi->total_valid_block_count); 2270 f2fs_bug_on(sbi, !sbi->total_valid_node_count); 2271 2272 sbi->total_valid_node_count--; 2273 sbi->total_valid_block_count--; 2274 if (sbi->reserved_blocks && 2275 sbi->current_reserved_blocks < sbi->reserved_blocks) 2276 sbi->current_reserved_blocks++; 2277 2278 spin_unlock(&sbi->stat_lock); 2279 2280 if (is_inode) { 2281 dquot_free_inode(inode); 2282 } else { 2283 if (unlikely(inode->i_blocks == 0)) { 2284 f2fs_warn(sbi, "dec_valid_node_count: inconsistent i_blocks, ino:%lu, iblocks:%llu", 2285 inode->i_ino, 2286 (unsigned long long)inode->i_blocks); 2287 set_sbi_flag(sbi, SBI_NEED_FSCK); 2288 return; 2289 } 2290 f2fs_i_blocks_write(inode, 1, false, true); 2291 } 2292 } 2293 2294 static inline unsigned int valid_node_count(struct f2fs_sb_info *sbi) 2295 { 2296 return sbi->total_valid_node_count; 2297 } 2298 2299 static inline void inc_valid_inode_count(struct f2fs_sb_info *sbi) 2300 { 2301 percpu_counter_inc(&sbi->total_valid_inode_count); 2302 } 2303 2304 static inline void dec_valid_inode_count(struct f2fs_sb_info *sbi) 2305 { 2306 percpu_counter_dec(&sbi->total_valid_inode_count); 2307 } 2308 2309 static inline s64 valid_inode_count(struct f2fs_sb_info *sbi) 2310 { 2311 return percpu_counter_sum_positive(&sbi->total_valid_inode_count); 2312 } 2313 2314 static inline struct page *f2fs_grab_cache_page(struct address_space *mapping, 2315 pgoff_t index, bool for_write) 2316 { 2317 struct page *page; 2318 2319 if (IS_ENABLED(CONFIG_F2FS_FAULT_INJECTION)) { 2320 if (!for_write) 2321 page = find_get_page_flags(mapping, index, 2322 FGP_LOCK | FGP_ACCESSED); 2323 else 2324 page = find_lock_page(mapping, index); 2325 if (page) 2326 return page; 2327 2328 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_ALLOC)) { 2329 f2fs_show_injection_info(F2FS_M_SB(mapping), 2330 FAULT_PAGE_ALLOC); 2331 return NULL; 2332 } 2333 } 2334 2335 if (!for_write) 2336 return grab_cache_page(mapping, index); 2337 return grab_cache_page_write_begin(mapping, index, AOP_FLAG_NOFS); 2338 } 2339 2340 static inline struct page *f2fs_pagecache_get_page( 2341 struct address_space *mapping, pgoff_t index, 2342 int fgp_flags, gfp_t gfp_mask) 2343 { 2344 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_GET)) { 2345 f2fs_show_injection_info(F2FS_M_SB(mapping), FAULT_PAGE_GET); 2346 return NULL; 2347 } 2348 2349 return pagecache_get_page(mapping, index, fgp_flags, gfp_mask); 2350 } 2351 2352 static inline void f2fs_copy_page(struct page *src, struct page *dst) 2353 { 2354 char *src_kaddr = kmap(src); 2355 char *dst_kaddr = kmap(dst); 2356 2357 memcpy(dst_kaddr, src_kaddr, PAGE_SIZE); 2358 kunmap(dst); 2359 kunmap(src); 2360 } 2361 2362 static inline void f2fs_put_page(struct page *page, int unlock) 2363 { 2364 if (!page) 2365 return; 2366 2367 if (unlock) { 2368 f2fs_bug_on(F2FS_P_SB(page), !PageLocked(page)); 2369 unlock_page(page); 2370 } 2371 put_page(page); 2372 } 2373 2374 static inline void f2fs_put_dnode(struct dnode_of_data *dn) 2375 { 2376 if (dn->node_page) 2377 f2fs_put_page(dn->node_page, 1); 2378 if (dn->inode_page && dn->node_page != dn->inode_page) 2379 f2fs_put_page(dn->inode_page, 0); 2380 dn->node_page = NULL; 2381 dn->inode_page = NULL; 2382 } 2383 2384 static inline struct kmem_cache *f2fs_kmem_cache_create(const char *name, 2385 size_t size) 2386 { 2387 return kmem_cache_create(name, size, 0, SLAB_RECLAIM_ACCOUNT, NULL); 2388 } 2389 2390 static inline void *f2fs_kmem_cache_alloc(struct kmem_cache *cachep, 2391 gfp_t flags) 2392 { 2393 void *entry; 2394 2395 entry = kmem_cache_alloc(cachep, flags); 2396 if (!entry) 2397 entry = kmem_cache_alloc(cachep, flags | __GFP_NOFAIL); 2398 return entry; 2399 } 2400 2401 static inline bool is_idle(struct f2fs_sb_info *sbi, int type) 2402 { 2403 if (sbi->gc_mode == GC_URGENT_HIGH) 2404 return true; 2405 2406 if (get_pages(sbi, F2FS_RD_DATA) || get_pages(sbi, F2FS_RD_NODE) || 2407 get_pages(sbi, F2FS_RD_META) || get_pages(sbi, F2FS_WB_DATA) || 2408 get_pages(sbi, F2FS_WB_CP_DATA) || 2409 get_pages(sbi, F2FS_DIO_READ) || 2410 get_pages(sbi, F2FS_DIO_WRITE)) 2411 return false; 2412 2413 if (type != DISCARD_TIME && SM_I(sbi) && SM_I(sbi)->dcc_info && 2414 atomic_read(&SM_I(sbi)->dcc_info->queued_discard)) 2415 return false; 2416 2417 if (SM_I(sbi) && SM_I(sbi)->fcc_info && 2418 atomic_read(&SM_I(sbi)->fcc_info->queued_flush)) 2419 return false; 2420 2421 if (sbi->gc_mode == GC_URGENT_LOW && 2422 (type == DISCARD_TIME || type == GC_TIME)) 2423 return true; 2424 2425 return f2fs_time_over(sbi, type); 2426 } 2427 2428 static inline void f2fs_radix_tree_insert(struct radix_tree_root *root, 2429 unsigned long index, void *item) 2430 { 2431 while (radix_tree_insert(root, index, item)) 2432 cond_resched(); 2433 } 2434 2435 #define RAW_IS_INODE(p) ((p)->footer.nid == (p)->footer.ino) 2436 2437 static inline bool IS_INODE(struct page *page) 2438 { 2439 struct f2fs_node *p = F2FS_NODE(page); 2440 2441 return RAW_IS_INODE(p); 2442 } 2443 2444 static inline int offset_in_addr(struct f2fs_inode *i) 2445 { 2446 return (i->i_inline & F2FS_EXTRA_ATTR) ? 2447 (le16_to_cpu(i->i_extra_isize) / sizeof(__le32)) : 0; 2448 } 2449 2450 static inline __le32 *blkaddr_in_node(struct f2fs_node *node) 2451 { 2452 return RAW_IS_INODE(node) ? node->i.i_addr : node->dn.addr; 2453 } 2454 2455 static inline int f2fs_has_extra_attr(struct inode *inode); 2456 static inline block_t data_blkaddr(struct inode *inode, 2457 struct page *node_page, unsigned int offset) 2458 { 2459 struct f2fs_node *raw_node; 2460 __le32 *addr_array; 2461 int base = 0; 2462 bool is_inode = IS_INODE(node_page); 2463 2464 raw_node = F2FS_NODE(node_page); 2465 2466 if (is_inode) { 2467 if (!inode) 2468 /* from GC path only */ 2469 base = offset_in_addr(&raw_node->i); 2470 else if (f2fs_has_extra_attr(inode)) 2471 base = get_extra_isize(inode); 2472 } 2473 2474 addr_array = blkaddr_in_node(raw_node); 2475 return le32_to_cpu(addr_array[base + offset]); 2476 } 2477 2478 static inline block_t f2fs_data_blkaddr(struct dnode_of_data *dn) 2479 { 2480 return data_blkaddr(dn->inode, dn->node_page, dn->ofs_in_node); 2481 } 2482 2483 static inline int f2fs_test_bit(unsigned int nr, char *addr) 2484 { 2485 int mask; 2486 2487 addr += (nr >> 3); 2488 mask = 1 << (7 - (nr & 0x07)); 2489 return mask & *addr; 2490 } 2491 2492 static inline void f2fs_set_bit(unsigned int nr, char *addr) 2493 { 2494 int mask; 2495 2496 addr += (nr >> 3); 2497 mask = 1 << (7 - (nr & 0x07)); 2498 *addr |= mask; 2499 } 2500 2501 static inline void f2fs_clear_bit(unsigned int nr, char *addr) 2502 { 2503 int mask; 2504 2505 addr += (nr >> 3); 2506 mask = 1 << (7 - (nr & 0x07)); 2507 *addr &= ~mask; 2508 } 2509 2510 static inline int f2fs_test_and_set_bit(unsigned int nr, char *addr) 2511 { 2512 int mask; 2513 int ret; 2514 2515 addr += (nr >> 3); 2516 mask = 1 << (7 - (nr & 0x07)); 2517 ret = mask & *addr; 2518 *addr |= mask; 2519 return ret; 2520 } 2521 2522 static inline int f2fs_test_and_clear_bit(unsigned int nr, char *addr) 2523 { 2524 int mask; 2525 int ret; 2526 2527 addr += (nr >> 3); 2528 mask = 1 << (7 - (nr & 0x07)); 2529 ret = mask & *addr; 2530 *addr &= ~mask; 2531 return ret; 2532 } 2533 2534 static inline void f2fs_change_bit(unsigned int nr, char *addr) 2535 { 2536 int mask; 2537 2538 addr += (nr >> 3); 2539 mask = 1 << (7 - (nr & 0x07)); 2540 *addr ^= mask; 2541 } 2542 2543 /* 2544 * On-disk inode flags (f2fs_inode::i_flags) 2545 */ 2546 #define F2FS_COMPR_FL 0x00000004 /* Compress file */ 2547 #define F2FS_SYNC_FL 0x00000008 /* Synchronous updates */ 2548 #define F2FS_IMMUTABLE_FL 0x00000010 /* Immutable file */ 2549 #define F2FS_APPEND_FL 0x00000020 /* writes to file may only append */ 2550 #define F2FS_NODUMP_FL 0x00000040 /* do not dump file */ 2551 #define F2FS_NOATIME_FL 0x00000080 /* do not update atime */ 2552 #define F2FS_NOCOMP_FL 0x00000400 /* Don't compress */ 2553 #define F2FS_INDEX_FL 0x00001000 /* hash-indexed directory */ 2554 #define F2FS_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */ 2555 #define F2FS_PROJINHERIT_FL 0x20000000 /* Create with parents projid */ 2556 #define F2FS_CASEFOLD_FL 0x40000000 /* Casefolded file */ 2557 2558 /* Flags that should be inherited by new inodes from their parent. */ 2559 #define F2FS_FL_INHERITED (F2FS_SYNC_FL | F2FS_NODUMP_FL | F2FS_NOATIME_FL | \ 2560 F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \ 2561 F2FS_CASEFOLD_FL | F2FS_COMPR_FL | F2FS_NOCOMP_FL) 2562 2563 /* Flags that are appropriate for regular files (all but dir-specific ones). */ 2564 #define F2FS_REG_FLMASK (~(F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \ 2565 F2FS_CASEFOLD_FL)) 2566 2567 /* Flags that are appropriate for non-directories/regular files. */ 2568 #define F2FS_OTHER_FLMASK (F2FS_NODUMP_FL | F2FS_NOATIME_FL) 2569 2570 static inline __u32 f2fs_mask_flags(umode_t mode, __u32 flags) 2571 { 2572 if (S_ISDIR(mode)) 2573 return flags; 2574 else if (S_ISREG(mode)) 2575 return flags & F2FS_REG_FLMASK; 2576 else 2577 return flags & F2FS_OTHER_FLMASK; 2578 } 2579 2580 static inline void __mark_inode_dirty_flag(struct inode *inode, 2581 int flag, bool set) 2582 { 2583 switch (flag) { 2584 case FI_INLINE_XATTR: 2585 case FI_INLINE_DATA: 2586 case FI_INLINE_DENTRY: 2587 case FI_NEW_INODE: 2588 if (set) 2589 return; 2590 fallthrough; 2591 case FI_DATA_EXIST: 2592 case FI_INLINE_DOTS: 2593 case FI_PIN_FILE: 2594 f2fs_mark_inode_dirty_sync(inode, true); 2595 } 2596 } 2597 2598 static inline void set_inode_flag(struct inode *inode, int flag) 2599 { 2600 set_bit(flag, F2FS_I(inode)->flags); 2601 __mark_inode_dirty_flag(inode, flag, true); 2602 } 2603 2604 static inline int is_inode_flag_set(struct inode *inode, int flag) 2605 { 2606 return test_bit(flag, F2FS_I(inode)->flags); 2607 } 2608 2609 static inline void clear_inode_flag(struct inode *inode, int flag) 2610 { 2611 clear_bit(flag, F2FS_I(inode)->flags); 2612 __mark_inode_dirty_flag(inode, flag, false); 2613 } 2614 2615 static inline bool f2fs_verity_in_progress(struct inode *inode) 2616 { 2617 return IS_ENABLED(CONFIG_FS_VERITY) && 2618 is_inode_flag_set(inode, FI_VERITY_IN_PROGRESS); 2619 } 2620 2621 static inline void set_acl_inode(struct inode *inode, umode_t mode) 2622 { 2623 F2FS_I(inode)->i_acl_mode = mode; 2624 set_inode_flag(inode, FI_ACL_MODE); 2625 f2fs_mark_inode_dirty_sync(inode, false); 2626 } 2627 2628 static inline void f2fs_i_links_write(struct inode *inode, bool inc) 2629 { 2630 if (inc) 2631 inc_nlink(inode); 2632 else 2633 drop_nlink(inode); 2634 f2fs_mark_inode_dirty_sync(inode, true); 2635 } 2636 2637 static inline void f2fs_i_blocks_write(struct inode *inode, 2638 block_t diff, bool add, bool claim) 2639 { 2640 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE); 2641 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER); 2642 2643 /* add = 1, claim = 1 should be dquot_reserve_block in pair */ 2644 if (add) { 2645 if (claim) 2646 dquot_claim_block(inode, diff); 2647 else 2648 dquot_alloc_block_nofail(inode, diff); 2649 } else { 2650 dquot_free_block(inode, diff); 2651 } 2652 2653 f2fs_mark_inode_dirty_sync(inode, true); 2654 if (clean || recover) 2655 set_inode_flag(inode, FI_AUTO_RECOVER); 2656 } 2657 2658 static inline void f2fs_i_size_write(struct inode *inode, loff_t i_size) 2659 { 2660 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE); 2661 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER); 2662 2663 if (i_size_read(inode) == i_size) 2664 return; 2665 2666 i_size_write(inode, i_size); 2667 f2fs_mark_inode_dirty_sync(inode, true); 2668 if (clean || recover) 2669 set_inode_flag(inode, FI_AUTO_RECOVER); 2670 } 2671 2672 static inline void f2fs_i_depth_write(struct inode *inode, unsigned int depth) 2673 { 2674 F2FS_I(inode)->i_current_depth = depth; 2675 f2fs_mark_inode_dirty_sync(inode, true); 2676 } 2677 2678 static inline void f2fs_i_gc_failures_write(struct inode *inode, 2679 unsigned int count) 2680 { 2681 F2FS_I(inode)->i_gc_failures[GC_FAILURE_PIN] = count; 2682 f2fs_mark_inode_dirty_sync(inode, true); 2683 } 2684 2685 static inline void f2fs_i_xnid_write(struct inode *inode, nid_t xnid) 2686 { 2687 F2FS_I(inode)->i_xattr_nid = xnid; 2688 f2fs_mark_inode_dirty_sync(inode, true); 2689 } 2690 2691 static inline void f2fs_i_pino_write(struct inode *inode, nid_t pino) 2692 { 2693 F2FS_I(inode)->i_pino = pino; 2694 f2fs_mark_inode_dirty_sync(inode, true); 2695 } 2696 2697 static inline void get_inline_info(struct inode *inode, struct f2fs_inode *ri) 2698 { 2699 struct f2fs_inode_info *fi = F2FS_I(inode); 2700 2701 if (ri->i_inline & F2FS_INLINE_XATTR) 2702 set_bit(FI_INLINE_XATTR, fi->flags); 2703 if (ri->i_inline & F2FS_INLINE_DATA) 2704 set_bit(FI_INLINE_DATA, fi->flags); 2705 if (ri->i_inline & F2FS_INLINE_DENTRY) 2706 set_bit(FI_INLINE_DENTRY, fi->flags); 2707 if (ri->i_inline & F2FS_DATA_EXIST) 2708 set_bit(FI_DATA_EXIST, fi->flags); 2709 if (ri->i_inline & F2FS_INLINE_DOTS) 2710 set_bit(FI_INLINE_DOTS, fi->flags); 2711 if (ri->i_inline & F2FS_EXTRA_ATTR) 2712 set_bit(FI_EXTRA_ATTR, fi->flags); 2713 if (ri->i_inline & F2FS_PIN_FILE) 2714 set_bit(FI_PIN_FILE, fi->flags); 2715 } 2716 2717 static inline void set_raw_inline(struct inode *inode, struct f2fs_inode *ri) 2718 { 2719 ri->i_inline = 0; 2720 2721 if (is_inode_flag_set(inode, FI_INLINE_XATTR)) 2722 ri->i_inline |= F2FS_INLINE_XATTR; 2723 if (is_inode_flag_set(inode, FI_INLINE_DATA)) 2724 ri->i_inline |= F2FS_INLINE_DATA; 2725 if (is_inode_flag_set(inode, FI_INLINE_DENTRY)) 2726 ri->i_inline |= F2FS_INLINE_DENTRY; 2727 if (is_inode_flag_set(inode, FI_DATA_EXIST)) 2728 ri->i_inline |= F2FS_DATA_EXIST; 2729 if (is_inode_flag_set(inode, FI_INLINE_DOTS)) 2730 ri->i_inline |= F2FS_INLINE_DOTS; 2731 if (is_inode_flag_set(inode, FI_EXTRA_ATTR)) 2732 ri->i_inline |= F2FS_EXTRA_ATTR; 2733 if (is_inode_flag_set(inode, FI_PIN_FILE)) 2734 ri->i_inline |= F2FS_PIN_FILE; 2735 } 2736 2737 static inline int f2fs_has_extra_attr(struct inode *inode) 2738 { 2739 return is_inode_flag_set(inode, FI_EXTRA_ATTR); 2740 } 2741 2742 static inline int f2fs_has_inline_xattr(struct inode *inode) 2743 { 2744 return is_inode_flag_set(inode, FI_INLINE_XATTR); 2745 } 2746 2747 static inline int f2fs_compressed_file(struct inode *inode) 2748 { 2749 return S_ISREG(inode->i_mode) && 2750 is_inode_flag_set(inode, FI_COMPRESSED_FILE); 2751 } 2752 2753 static inline unsigned int addrs_per_inode(struct inode *inode) 2754 { 2755 unsigned int addrs = CUR_ADDRS_PER_INODE(inode) - 2756 get_inline_xattr_addrs(inode); 2757 2758 if (!f2fs_compressed_file(inode)) 2759 return addrs; 2760 return ALIGN_DOWN(addrs, F2FS_I(inode)->i_cluster_size); 2761 } 2762 2763 static inline unsigned int addrs_per_block(struct inode *inode) 2764 { 2765 if (!f2fs_compressed_file(inode)) 2766 return DEF_ADDRS_PER_BLOCK; 2767 return ALIGN_DOWN(DEF_ADDRS_PER_BLOCK, F2FS_I(inode)->i_cluster_size); 2768 } 2769 2770 static inline void *inline_xattr_addr(struct inode *inode, struct page *page) 2771 { 2772 struct f2fs_inode *ri = F2FS_INODE(page); 2773 2774 return (void *)&(ri->i_addr[DEF_ADDRS_PER_INODE - 2775 get_inline_xattr_addrs(inode)]); 2776 } 2777 2778 static inline int inline_xattr_size(struct inode *inode) 2779 { 2780 if (f2fs_has_inline_xattr(inode)) 2781 return get_inline_xattr_addrs(inode) * sizeof(__le32); 2782 return 0; 2783 } 2784 2785 static inline int f2fs_has_inline_data(struct inode *inode) 2786 { 2787 return is_inode_flag_set(inode, FI_INLINE_DATA); 2788 } 2789 2790 static inline int f2fs_exist_data(struct inode *inode) 2791 { 2792 return is_inode_flag_set(inode, FI_DATA_EXIST); 2793 } 2794 2795 static inline int f2fs_has_inline_dots(struct inode *inode) 2796 { 2797 return is_inode_flag_set(inode, FI_INLINE_DOTS); 2798 } 2799 2800 static inline int f2fs_is_mmap_file(struct inode *inode) 2801 { 2802 return is_inode_flag_set(inode, FI_MMAP_FILE); 2803 } 2804 2805 static inline bool f2fs_is_pinned_file(struct inode *inode) 2806 { 2807 return is_inode_flag_set(inode, FI_PIN_FILE); 2808 } 2809 2810 static inline bool f2fs_is_atomic_file(struct inode *inode) 2811 { 2812 return is_inode_flag_set(inode, FI_ATOMIC_FILE); 2813 } 2814 2815 static inline bool f2fs_is_commit_atomic_write(struct inode *inode) 2816 { 2817 return is_inode_flag_set(inode, FI_ATOMIC_COMMIT); 2818 } 2819 2820 static inline bool f2fs_is_volatile_file(struct inode *inode) 2821 { 2822 return is_inode_flag_set(inode, FI_VOLATILE_FILE); 2823 } 2824 2825 static inline bool f2fs_is_first_block_written(struct inode *inode) 2826 { 2827 return is_inode_flag_set(inode, FI_FIRST_BLOCK_WRITTEN); 2828 } 2829 2830 static inline bool f2fs_is_drop_cache(struct inode *inode) 2831 { 2832 return is_inode_flag_set(inode, FI_DROP_CACHE); 2833 } 2834 2835 static inline void *inline_data_addr(struct inode *inode, struct page *page) 2836 { 2837 struct f2fs_inode *ri = F2FS_INODE(page); 2838 int extra_size = get_extra_isize(inode); 2839 2840 return (void *)&(ri->i_addr[extra_size + DEF_INLINE_RESERVED_SIZE]); 2841 } 2842 2843 static inline int f2fs_has_inline_dentry(struct inode *inode) 2844 { 2845 return is_inode_flag_set(inode, FI_INLINE_DENTRY); 2846 } 2847 2848 static inline int is_file(struct inode *inode, int type) 2849 { 2850 return F2FS_I(inode)->i_advise & type; 2851 } 2852 2853 static inline void set_file(struct inode *inode, int type) 2854 { 2855 F2FS_I(inode)->i_advise |= type; 2856 f2fs_mark_inode_dirty_sync(inode, true); 2857 } 2858 2859 static inline void clear_file(struct inode *inode, int type) 2860 { 2861 F2FS_I(inode)->i_advise &= ~type; 2862 f2fs_mark_inode_dirty_sync(inode, true); 2863 } 2864 2865 static inline bool f2fs_is_time_consistent(struct inode *inode) 2866 { 2867 if (!timespec64_equal(F2FS_I(inode)->i_disk_time, &inode->i_atime)) 2868 return false; 2869 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 1, &inode->i_ctime)) 2870 return false; 2871 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 2, &inode->i_mtime)) 2872 return false; 2873 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 3, 2874 &F2FS_I(inode)->i_crtime)) 2875 return false; 2876 return true; 2877 } 2878 2879 static inline bool f2fs_skip_inode_update(struct inode *inode, int dsync) 2880 { 2881 bool ret; 2882 2883 if (dsync) { 2884 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 2885 2886 spin_lock(&sbi->inode_lock[DIRTY_META]); 2887 ret = list_empty(&F2FS_I(inode)->gdirty_list); 2888 spin_unlock(&sbi->inode_lock[DIRTY_META]); 2889 return ret; 2890 } 2891 if (!is_inode_flag_set(inode, FI_AUTO_RECOVER) || 2892 file_keep_isize(inode) || 2893 i_size_read(inode) & ~PAGE_MASK) 2894 return false; 2895 2896 if (!f2fs_is_time_consistent(inode)) 2897 return false; 2898 2899 spin_lock(&F2FS_I(inode)->i_size_lock); 2900 ret = F2FS_I(inode)->last_disk_size == i_size_read(inode); 2901 spin_unlock(&F2FS_I(inode)->i_size_lock); 2902 2903 return ret; 2904 } 2905 2906 static inline bool f2fs_readonly(struct super_block *sb) 2907 { 2908 return sb_rdonly(sb); 2909 } 2910 2911 static inline bool f2fs_cp_error(struct f2fs_sb_info *sbi) 2912 { 2913 return is_set_ckpt_flags(sbi, CP_ERROR_FLAG); 2914 } 2915 2916 static inline bool is_dot_dotdot(const u8 *name, size_t len) 2917 { 2918 if (len == 1 && name[0] == '.') 2919 return true; 2920 2921 if (len == 2 && name[0] == '.' && name[1] == '.') 2922 return true; 2923 2924 return false; 2925 } 2926 2927 static inline bool f2fs_may_extent_tree(struct inode *inode) 2928 { 2929 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 2930 2931 if (!test_opt(sbi, EXTENT_CACHE) || 2932 is_inode_flag_set(inode, FI_NO_EXTENT) || 2933 is_inode_flag_set(inode, FI_COMPRESSED_FILE)) 2934 return false; 2935 2936 /* 2937 * for recovered files during mount do not create extents 2938 * if shrinker is not registered. 2939 */ 2940 if (list_empty(&sbi->s_list)) 2941 return false; 2942 2943 return S_ISREG(inode->i_mode); 2944 } 2945 2946 static inline void *f2fs_kmalloc(struct f2fs_sb_info *sbi, 2947 size_t size, gfp_t flags) 2948 { 2949 if (time_to_inject(sbi, FAULT_KMALLOC)) { 2950 f2fs_show_injection_info(sbi, FAULT_KMALLOC); 2951 return NULL; 2952 } 2953 2954 return kmalloc(size, flags); 2955 } 2956 2957 static inline void *f2fs_kzalloc(struct f2fs_sb_info *sbi, 2958 size_t size, gfp_t flags) 2959 { 2960 return f2fs_kmalloc(sbi, size, flags | __GFP_ZERO); 2961 } 2962 2963 static inline void *f2fs_kvmalloc(struct f2fs_sb_info *sbi, 2964 size_t size, gfp_t flags) 2965 { 2966 if (time_to_inject(sbi, FAULT_KVMALLOC)) { 2967 f2fs_show_injection_info(sbi, FAULT_KVMALLOC); 2968 return NULL; 2969 } 2970 2971 return kvmalloc(size, flags); 2972 } 2973 2974 static inline void *f2fs_kvzalloc(struct f2fs_sb_info *sbi, 2975 size_t size, gfp_t flags) 2976 { 2977 return f2fs_kvmalloc(sbi, size, flags | __GFP_ZERO); 2978 } 2979 2980 static inline int get_extra_isize(struct inode *inode) 2981 { 2982 return F2FS_I(inode)->i_extra_isize / sizeof(__le32); 2983 } 2984 2985 static inline int get_inline_xattr_addrs(struct inode *inode) 2986 { 2987 return F2FS_I(inode)->i_inline_xattr_size; 2988 } 2989 2990 #define f2fs_get_inode_mode(i) \ 2991 ((is_inode_flag_set(i, FI_ACL_MODE)) ? \ 2992 (F2FS_I(i)->i_acl_mode) : ((i)->i_mode)) 2993 2994 #define F2FS_TOTAL_EXTRA_ATTR_SIZE \ 2995 (offsetof(struct f2fs_inode, i_extra_end) - \ 2996 offsetof(struct f2fs_inode, i_extra_isize)) \ 2997 2998 #define F2FS_OLD_ATTRIBUTE_SIZE (offsetof(struct f2fs_inode, i_addr)) 2999 #define F2FS_FITS_IN_INODE(f2fs_inode, extra_isize, field) \ 3000 ((offsetof(typeof(*(f2fs_inode)), field) + \ 3001 sizeof((f2fs_inode)->field)) \ 3002 <= (F2FS_OLD_ATTRIBUTE_SIZE + (extra_isize))) \ 3003 3004 #define DEFAULT_IOSTAT_PERIOD_MS 3000 3005 #define MIN_IOSTAT_PERIOD_MS 100 3006 /* maximum period of iostat tracing is 1 day */ 3007 #define MAX_IOSTAT_PERIOD_MS 8640000 3008 3009 static inline void f2fs_reset_iostat(struct f2fs_sb_info *sbi) 3010 { 3011 int i; 3012 3013 spin_lock(&sbi->iostat_lock); 3014 for (i = 0; i < NR_IO_TYPE; i++) { 3015 sbi->rw_iostat[i] = 0; 3016 sbi->prev_rw_iostat[i] = 0; 3017 } 3018 spin_unlock(&sbi->iostat_lock); 3019 } 3020 3021 extern void f2fs_record_iostat(struct f2fs_sb_info *sbi); 3022 3023 static inline void f2fs_update_iostat(struct f2fs_sb_info *sbi, 3024 enum iostat_type type, unsigned long long io_bytes) 3025 { 3026 if (!sbi->iostat_enable) 3027 return; 3028 spin_lock(&sbi->iostat_lock); 3029 sbi->rw_iostat[type] += io_bytes; 3030 3031 if (type == APP_WRITE_IO || type == APP_DIRECT_IO) 3032 sbi->rw_iostat[APP_BUFFERED_IO] = 3033 sbi->rw_iostat[APP_WRITE_IO] - 3034 sbi->rw_iostat[APP_DIRECT_IO]; 3035 3036 if (type == APP_READ_IO || type == APP_DIRECT_READ_IO) 3037 sbi->rw_iostat[APP_BUFFERED_READ_IO] = 3038 sbi->rw_iostat[APP_READ_IO] - 3039 sbi->rw_iostat[APP_DIRECT_READ_IO]; 3040 spin_unlock(&sbi->iostat_lock); 3041 3042 f2fs_record_iostat(sbi); 3043 } 3044 3045 #define __is_large_section(sbi) ((sbi)->segs_per_sec > 1) 3046 3047 #define __is_meta_io(fio) (PAGE_TYPE_OF_BIO((fio)->type) == META) 3048 3049 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi, 3050 block_t blkaddr, int type); 3051 static inline void verify_blkaddr(struct f2fs_sb_info *sbi, 3052 block_t blkaddr, int type) 3053 { 3054 if (!f2fs_is_valid_blkaddr(sbi, blkaddr, type)) { 3055 f2fs_err(sbi, "invalid blkaddr: %u, type: %d, run fsck to fix.", 3056 blkaddr, type); 3057 f2fs_bug_on(sbi, 1); 3058 } 3059 } 3060 3061 static inline bool __is_valid_data_blkaddr(block_t blkaddr) 3062 { 3063 if (blkaddr == NEW_ADDR || blkaddr == NULL_ADDR || 3064 blkaddr == COMPRESS_ADDR) 3065 return false; 3066 return true; 3067 } 3068 3069 static inline void f2fs_set_page_private(struct page *page, 3070 unsigned long data) 3071 { 3072 if (PagePrivate(page)) 3073 return; 3074 3075 attach_page_private(page, (void *)data); 3076 } 3077 3078 static inline void f2fs_clear_page_private(struct page *page) 3079 { 3080 detach_page_private(page); 3081 } 3082 3083 /* 3084 * file.c 3085 */ 3086 int f2fs_sync_file(struct file *file, loff_t start, loff_t end, int datasync); 3087 void f2fs_truncate_data_blocks(struct dnode_of_data *dn); 3088 int f2fs_do_truncate_blocks(struct inode *inode, u64 from, bool lock); 3089 int f2fs_truncate_blocks(struct inode *inode, u64 from, bool lock); 3090 int f2fs_truncate(struct inode *inode); 3091 int f2fs_getattr(const struct path *path, struct kstat *stat, 3092 u32 request_mask, unsigned int flags); 3093 int f2fs_setattr(struct dentry *dentry, struct iattr *attr); 3094 int f2fs_truncate_hole(struct inode *inode, pgoff_t pg_start, pgoff_t pg_end); 3095 void f2fs_truncate_data_blocks_range(struct dnode_of_data *dn, int count); 3096 int f2fs_precache_extents(struct inode *inode); 3097 long f2fs_ioctl(struct file *filp, unsigned int cmd, unsigned long arg); 3098 long f2fs_compat_ioctl(struct file *file, unsigned int cmd, unsigned long arg); 3099 int f2fs_transfer_project_quota(struct inode *inode, kprojid_t kprojid); 3100 int f2fs_pin_file_control(struct inode *inode, bool inc); 3101 3102 /* 3103 * inode.c 3104 */ 3105 void f2fs_set_inode_flags(struct inode *inode); 3106 bool f2fs_inode_chksum_verify(struct f2fs_sb_info *sbi, struct page *page); 3107 void f2fs_inode_chksum_set(struct f2fs_sb_info *sbi, struct page *page); 3108 struct inode *f2fs_iget(struct super_block *sb, unsigned long ino); 3109 struct inode *f2fs_iget_retry(struct super_block *sb, unsigned long ino); 3110 int f2fs_try_to_free_nats(struct f2fs_sb_info *sbi, int nr_shrink); 3111 void f2fs_update_inode(struct inode *inode, struct page *node_page); 3112 void f2fs_update_inode_page(struct inode *inode); 3113 int f2fs_write_inode(struct inode *inode, struct writeback_control *wbc); 3114 void f2fs_evict_inode(struct inode *inode); 3115 void f2fs_handle_failed_inode(struct inode *inode); 3116 3117 /* 3118 * namei.c 3119 */ 3120 int f2fs_update_extension_list(struct f2fs_sb_info *sbi, const char *name, 3121 bool hot, bool set); 3122 struct dentry *f2fs_get_parent(struct dentry *child); 3123 3124 /* 3125 * dir.c 3126 */ 3127 unsigned char f2fs_get_de_type(struct f2fs_dir_entry *de); 3128 int f2fs_init_casefolded_name(const struct inode *dir, 3129 struct f2fs_filename *fname); 3130 int f2fs_setup_filename(struct inode *dir, const struct qstr *iname, 3131 int lookup, struct f2fs_filename *fname); 3132 int f2fs_prepare_lookup(struct inode *dir, struct dentry *dentry, 3133 struct f2fs_filename *fname); 3134 void f2fs_free_filename(struct f2fs_filename *fname); 3135 struct f2fs_dir_entry *f2fs_find_target_dentry(const struct f2fs_dentry_ptr *d, 3136 const struct f2fs_filename *fname, int *max_slots); 3137 int f2fs_fill_dentries(struct dir_context *ctx, struct f2fs_dentry_ptr *d, 3138 unsigned int start_pos, struct fscrypt_str *fstr); 3139 void f2fs_do_make_empty_dir(struct inode *inode, struct inode *parent, 3140 struct f2fs_dentry_ptr *d); 3141 struct page *f2fs_init_inode_metadata(struct inode *inode, struct inode *dir, 3142 const struct f2fs_filename *fname, struct page *dpage); 3143 void f2fs_update_parent_metadata(struct inode *dir, struct inode *inode, 3144 unsigned int current_depth); 3145 int f2fs_room_for_filename(const void *bitmap, int slots, int max_slots); 3146 void f2fs_drop_nlink(struct inode *dir, struct inode *inode); 3147 struct f2fs_dir_entry *__f2fs_find_entry(struct inode *dir, 3148 const struct f2fs_filename *fname, 3149 struct page **res_page); 3150 struct f2fs_dir_entry *f2fs_find_entry(struct inode *dir, 3151 const struct qstr *child, struct page **res_page); 3152 struct f2fs_dir_entry *f2fs_parent_dir(struct inode *dir, struct page **p); 3153 ino_t f2fs_inode_by_name(struct inode *dir, const struct qstr *qstr, 3154 struct page **page); 3155 void f2fs_set_link(struct inode *dir, struct f2fs_dir_entry *de, 3156 struct page *page, struct inode *inode); 3157 bool f2fs_has_enough_room(struct inode *dir, struct page *ipage, 3158 const struct f2fs_filename *fname); 3159 void f2fs_update_dentry(nid_t ino, umode_t mode, struct f2fs_dentry_ptr *d, 3160 const struct fscrypt_str *name, f2fs_hash_t name_hash, 3161 unsigned int bit_pos); 3162 int f2fs_add_regular_entry(struct inode *dir, const struct f2fs_filename *fname, 3163 struct inode *inode, nid_t ino, umode_t mode); 3164 int f2fs_add_dentry(struct inode *dir, const struct f2fs_filename *fname, 3165 struct inode *inode, nid_t ino, umode_t mode); 3166 int f2fs_do_add_link(struct inode *dir, const struct qstr *name, 3167 struct inode *inode, nid_t ino, umode_t mode); 3168 void f2fs_delete_entry(struct f2fs_dir_entry *dentry, struct page *page, 3169 struct inode *dir, struct inode *inode); 3170 int f2fs_do_tmpfile(struct inode *inode, struct inode *dir); 3171 bool f2fs_empty_dir(struct inode *dir); 3172 3173 static inline int f2fs_add_link(struct dentry *dentry, struct inode *inode) 3174 { 3175 return f2fs_do_add_link(d_inode(dentry->d_parent), &dentry->d_name, 3176 inode, inode->i_ino, inode->i_mode); 3177 } 3178 3179 /* 3180 * super.c 3181 */ 3182 int f2fs_inode_dirtied(struct inode *inode, bool sync); 3183 void f2fs_inode_synced(struct inode *inode); 3184 int f2fs_enable_quota_files(struct f2fs_sb_info *sbi, bool rdonly); 3185 int f2fs_quota_sync(struct super_block *sb, int type); 3186 void f2fs_quota_off_umount(struct super_block *sb); 3187 int f2fs_commit_super(struct f2fs_sb_info *sbi, bool recover); 3188 int f2fs_sync_fs(struct super_block *sb, int sync); 3189 int f2fs_sanity_check_ckpt(struct f2fs_sb_info *sbi); 3190 3191 /* 3192 * hash.c 3193 */ 3194 void f2fs_hash_filename(const struct inode *dir, struct f2fs_filename *fname); 3195 3196 /* 3197 * node.c 3198 */ 3199 struct dnode_of_data; 3200 struct node_info; 3201 3202 int f2fs_check_nid_range(struct f2fs_sb_info *sbi, nid_t nid); 3203 bool f2fs_available_free_memory(struct f2fs_sb_info *sbi, int type); 3204 bool f2fs_in_warm_node_list(struct f2fs_sb_info *sbi, struct page *page); 3205 void f2fs_init_fsync_node_info(struct f2fs_sb_info *sbi); 3206 void f2fs_del_fsync_node_entry(struct f2fs_sb_info *sbi, struct page *page); 3207 void f2fs_reset_fsync_node_info(struct f2fs_sb_info *sbi); 3208 int f2fs_need_dentry_mark(struct f2fs_sb_info *sbi, nid_t nid); 3209 bool f2fs_is_checkpointed_node(struct f2fs_sb_info *sbi, nid_t nid); 3210 bool f2fs_need_inode_block_update(struct f2fs_sb_info *sbi, nid_t ino); 3211 int f2fs_get_node_info(struct f2fs_sb_info *sbi, nid_t nid, 3212 struct node_info *ni); 3213 pgoff_t f2fs_get_next_page_offset(struct dnode_of_data *dn, pgoff_t pgofs); 3214 int f2fs_get_dnode_of_data(struct dnode_of_data *dn, pgoff_t index, int mode); 3215 int f2fs_truncate_inode_blocks(struct inode *inode, pgoff_t from); 3216 int f2fs_truncate_xattr_node(struct inode *inode); 3217 int f2fs_wait_on_node_pages_writeback(struct f2fs_sb_info *sbi, 3218 unsigned int seq_id); 3219 int f2fs_remove_inode_page(struct inode *inode); 3220 struct page *f2fs_new_inode_page(struct inode *inode); 3221 struct page *f2fs_new_node_page(struct dnode_of_data *dn, unsigned int ofs); 3222 void f2fs_ra_node_page(struct f2fs_sb_info *sbi, nid_t nid); 3223 struct page *f2fs_get_node_page(struct f2fs_sb_info *sbi, pgoff_t nid); 3224 struct page *f2fs_get_node_page_ra(struct page *parent, int start); 3225 int f2fs_move_node_page(struct page *node_page, int gc_type); 3226 void f2fs_flush_inline_data(struct f2fs_sb_info *sbi); 3227 int f2fs_fsync_node_pages(struct f2fs_sb_info *sbi, struct inode *inode, 3228 struct writeback_control *wbc, bool atomic, 3229 unsigned int *seq_id); 3230 int f2fs_sync_node_pages(struct f2fs_sb_info *sbi, 3231 struct writeback_control *wbc, 3232 bool do_balance, enum iostat_type io_type); 3233 int f2fs_build_free_nids(struct f2fs_sb_info *sbi, bool sync, bool mount); 3234 bool f2fs_alloc_nid(struct f2fs_sb_info *sbi, nid_t *nid); 3235 void f2fs_alloc_nid_done(struct f2fs_sb_info *sbi, nid_t nid); 3236 void f2fs_alloc_nid_failed(struct f2fs_sb_info *sbi, nid_t nid); 3237 int f2fs_try_to_free_nids(struct f2fs_sb_info *sbi, int nr_shrink); 3238 int f2fs_recover_inline_xattr(struct inode *inode, struct page *page); 3239 int f2fs_recover_xattr_data(struct inode *inode, struct page *page); 3240 int f2fs_recover_inode_page(struct f2fs_sb_info *sbi, struct page *page); 3241 int f2fs_restore_node_summary(struct f2fs_sb_info *sbi, 3242 unsigned int segno, struct f2fs_summary_block *sum); 3243 int f2fs_flush_nat_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3244 int f2fs_build_node_manager(struct f2fs_sb_info *sbi); 3245 void f2fs_destroy_node_manager(struct f2fs_sb_info *sbi); 3246 int __init f2fs_create_node_manager_caches(void); 3247 void f2fs_destroy_node_manager_caches(void); 3248 3249 /* 3250 * segment.c 3251 */ 3252 bool f2fs_need_SSR(struct f2fs_sb_info *sbi); 3253 void f2fs_register_inmem_page(struct inode *inode, struct page *page); 3254 void f2fs_drop_inmem_pages_all(struct f2fs_sb_info *sbi, bool gc_failure); 3255 void f2fs_drop_inmem_pages(struct inode *inode); 3256 void f2fs_drop_inmem_page(struct inode *inode, struct page *page); 3257 int f2fs_commit_inmem_pages(struct inode *inode); 3258 void f2fs_balance_fs(struct f2fs_sb_info *sbi, bool need); 3259 void f2fs_balance_fs_bg(struct f2fs_sb_info *sbi, bool from_bg); 3260 int f2fs_issue_flush(struct f2fs_sb_info *sbi, nid_t ino); 3261 int f2fs_create_flush_cmd_control(struct f2fs_sb_info *sbi); 3262 int f2fs_flush_device_cache(struct f2fs_sb_info *sbi); 3263 void f2fs_destroy_flush_cmd_control(struct f2fs_sb_info *sbi, bool free); 3264 void f2fs_invalidate_blocks(struct f2fs_sb_info *sbi, block_t addr); 3265 bool f2fs_is_checkpointed_data(struct f2fs_sb_info *sbi, block_t blkaddr); 3266 void f2fs_drop_discard_cmd(struct f2fs_sb_info *sbi); 3267 void f2fs_stop_discard_thread(struct f2fs_sb_info *sbi); 3268 bool f2fs_issue_discard_timeout(struct f2fs_sb_info *sbi); 3269 void f2fs_clear_prefree_segments(struct f2fs_sb_info *sbi, 3270 struct cp_control *cpc); 3271 void f2fs_dirty_to_prefree(struct f2fs_sb_info *sbi); 3272 block_t f2fs_get_unusable_blocks(struct f2fs_sb_info *sbi); 3273 int f2fs_disable_cp_again(struct f2fs_sb_info *sbi, block_t unusable); 3274 void f2fs_release_discard_addrs(struct f2fs_sb_info *sbi); 3275 int f2fs_npages_for_summary_flush(struct f2fs_sb_info *sbi, bool for_ra); 3276 void f2fs_init_inmem_curseg(struct f2fs_sb_info *sbi); 3277 void f2fs_save_inmem_curseg(struct f2fs_sb_info *sbi); 3278 void f2fs_restore_inmem_curseg(struct f2fs_sb_info *sbi); 3279 void f2fs_get_new_segment(struct f2fs_sb_info *sbi, 3280 unsigned int *newseg, bool new_sec, int dir); 3281 void f2fs_allocate_segment_for_resize(struct f2fs_sb_info *sbi, int type, 3282 unsigned int start, unsigned int end); 3283 void f2fs_allocate_new_segment(struct f2fs_sb_info *sbi, int type); 3284 void f2fs_allocate_new_segments(struct f2fs_sb_info *sbi); 3285 int f2fs_trim_fs(struct f2fs_sb_info *sbi, struct fstrim_range *range); 3286 bool f2fs_exist_trim_candidates(struct f2fs_sb_info *sbi, 3287 struct cp_control *cpc); 3288 struct page *f2fs_get_sum_page(struct f2fs_sb_info *sbi, unsigned int segno); 3289 void f2fs_update_meta_page(struct f2fs_sb_info *sbi, void *src, 3290 block_t blk_addr); 3291 void f2fs_do_write_meta_page(struct f2fs_sb_info *sbi, struct page *page, 3292 enum iostat_type io_type); 3293 void f2fs_do_write_node_page(unsigned int nid, struct f2fs_io_info *fio); 3294 void f2fs_outplace_write_data(struct dnode_of_data *dn, 3295 struct f2fs_io_info *fio); 3296 int f2fs_inplace_write_data(struct f2fs_io_info *fio); 3297 void f2fs_do_replace_block(struct f2fs_sb_info *sbi, struct f2fs_summary *sum, 3298 block_t old_blkaddr, block_t new_blkaddr, 3299 bool recover_curseg, bool recover_newaddr, 3300 bool from_gc); 3301 void f2fs_replace_block(struct f2fs_sb_info *sbi, struct dnode_of_data *dn, 3302 block_t old_addr, block_t new_addr, 3303 unsigned char version, bool recover_curseg, 3304 bool recover_newaddr); 3305 void f2fs_allocate_data_block(struct f2fs_sb_info *sbi, struct page *page, 3306 block_t old_blkaddr, block_t *new_blkaddr, 3307 struct f2fs_summary *sum, int type, 3308 struct f2fs_io_info *fio); 3309 void f2fs_wait_on_page_writeback(struct page *page, 3310 enum page_type type, bool ordered, bool locked); 3311 void f2fs_wait_on_block_writeback(struct inode *inode, block_t blkaddr); 3312 void f2fs_wait_on_block_writeback_range(struct inode *inode, block_t blkaddr, 3313 block_t len); 3314 void f2fs_write_data_summaries(struct f2fs_sb_info *sbi, block_t start_blk); 3315 void f2fs_write_node_summaries(struct f2fs_sb_info *sbi, block_t start_blk); 3316 int f2fs_lookup_journal_in_cursum(struct f2fs_journal *journal, int type, 3317 unsigned int val, int alloc); 3318 void f2fs_flush_sit_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3319 int f2fs_fix_curseg_write_pointer(struct f2fs_sb_info *sbi); 3320 int f2fs_check_write_pointer(struct f2fs_sb_info *sbi); 3321 int f2fs_build_segment_manager(struct f2fs_sb_info *sbi); 3322 void f2fs_destroy_segment_manager(struct f2fs_sb_info *sbi); 3323 int __init f2fs_create_segment_manager_caches(void); 3324 void f2fs_destroy_segment_manager_caches(void); 3325 int f2fs_rw_hint_to_seg_type(enum rw_hint hint); 3326 enum rw_hint f2fs_io_type_to_rw_hint(struct f2fs_sb_info *sbi, 3327 enum page_type type, enum temp_type temp); 3328 unsigned int f2fs_usable_segs_in_sec(struct f2fs_sb_info *sbi, 3329 unsigned int segno); 3330 unsigned int f2fs_usable_blks_in_seg(struct f2fs_sb_info *sbi, 3331 unsigned int segno); 3332 3333 /* 3334 * checkpoint.c 3335 */ 3336 void f2fs_stop_checkpoint(struct f2fs_sb_info *sbi, bool end_io); 3337 struct page *f2fs_grab_meta_page(struct f2fs_sb_info *sbi, pgoff_t index); 3338 struct page *f2fs_get_meta_page(struct f2fs_sb_info *sbi, pgoff_t index); 3339 struct page *f2fs_get_meta_page_retry(struct f2fs_sb_info *sbi, pgoff_t index); 3340 struct page *f2fs_get_tmp_page(struct f2fs_sb_info *sbi, pgoff_t index); 3341 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi, 3342 block_t blkaddr, int type); 3343 int f2fs_ra_meta_pages(struct f2fs_sb_info *sbi, block_t start, int nrpages, 3344 int type, bool sync); 3345 void f2fs_ra_meta_pages_cond(struct f2fs_sb_info *sbi, pgoff_t index); 3346 long f2fs_sync_meta_pages(struct f2fs_sb_info *sbi, enum page_type type, 3347 long nr_to_write, enum iostat_type io_type); 3348 void f2fs_add_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type); 3349 void f2fs_remove_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type); 3350 void f2fs_release_ino_entry(struct f2fs_sb_info *sbi, bool all); 3351 bool f2fs_exist_written_data(struct f2fs_sb_info *sbi, nid_t ino, int mode); 3352 void f2fs_set_dirty_device(struct f2fs_sb_info *sbi, nid_t ino, 3353 unsigned int devidx, int type); 3354 bool f2fs_is_dirty_device(struct f2fs_sb_info *sbi, nid_t ino, 3355 unsigned int devidx, int type); 3356 int f2fs_sync_inode_meta(struct f2fs_sb_info *sbi); 3357 int f2fs_acquire_orphan_inode(struct f2fs_sb_info *sbi); 3358 void f2fs_release_orphan_inode(struct f2fs_sb_info *sbi); 3359 void f2fs_add_orphan_inode(struct inode *inode); 3360 void f2fs_remove_orphan_inode(struct f2fs_sb_info *sbi, nid_t ino); 3361 int f2fs_recover_orphan_inodes(struct f2fs_sb_info *sbi); 3362 int f2fs_get_valid_checkpoint(struct f2fs_sb_info *sbi); 3363 void f2fs_update_dirty_page(struct inode *inode, struct page *page); 3364 void f2fs_remove_dirty_inode(struct inode *inode); 3365 int f2fs_sync_dirty_inodes(struct f2fs_sb_info *sbi, enum inode_type type); 3366 void f2fs_wait_on_all_pages(struct f2fs_sb_info *sbi, int type); 3367 int f2fs_write_checkpoint(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3368 void f2fs_init_ino_entry_info(struct f2fs_sb_info *sbi); 3369 int __init f2fs_create_checkpoint_caches(void); 3370 void f2fs_destroy_checkpoint_caches(void); 3371 3372 /* 3373 * data.c 3374 */ 3375 int __init f2fs_init_bioset(void); 3376 void f2fs_destroy_bioset(void); 3377 struct bio *f2fs_bio_alloc(struct f2fs_sb_info *sbi, int npages, bool noio); 3378 int f2fs_init_bio_entry_cache(void); 3379 void f2fs_destroy_bio_entry_cache(void); 3380 void f2fs_submit_bio(struct f2fs_sb_info *sbi, 3381 struct bio *bio, enum page_type type); 3382 void f2fs_submit_merged_write(struct f2fs_sb_info *sbi, enum page_type type); 3383 void f2fs_submit_merged_write_cond(struct f2fs_sb_info *sbi, 3384 struct inode *inode, struct page *page, 3385 nid_t ino, enum page_type type); 3386 void f2fs_submit_merged_ipu_write(struct f2fs_sb_info *sbi, 3387 struct bio **bio, struct page *page); 3388 void f2fs_flush_merged_writes(struct f2fs_sb_info *sbi); 3389 int f2fs_submit_page_bio(struct f2fs_io_info *fio); 3390 int f2fs_merge_page_bio(struct f2fs_io_info *fio); 3391 void f2fs_submit_page_write(struct f2fs_io_info *fio); 3392 struct block_device *f2fs_target_device(struct f2fs_sb_info *sbi, 3393 block_t blk_addr, struct bio *bio); 3394 int f2fs_target_device_index(struct f2fs_sb_info *sbi, block_t blkaddr); 3395 void f2fs_set_data_blkaddr(struct dnode_of_data *dn); 3396 void f2fs_update_data_blkaddr(struct dnode_of_data *dn, block_t blkaddr); 3397 int f2fs_reserve_new_blocks(struct dnode_of_data *dn, blkcnt_t count); 3398 int f2fs_reserve_new_block(struct dnode_of_data *dn); 3399 int f2fs_get_block(struct dnode_of_data *dn, pgoff_t index); 3400 int f2fs_preallocate_blocks(struct kiocb *iocb, struct iov_iter *from); 3401 int f2fs_reserve_block(struct dnode_of_data *dn, pgoff_t index); 3402 struct page *f2fs_get_read_data_page(struct inode *inode, pgoff_t index, 3403 int op_flags, bool for_write); 3404 struct page *f2fs_find_data_page(struct inode *inode, pgoff_t index); 3405 struct page *f2fs_get_lock_data_page(struct inode *inode, pgoff_t index, 3406 bool for_write); 3407 struct page *f2fs_get_new_data_page(struct inode *inode, 3408 struct page *ipage, pgoff_t index, bool new_i_size); 3409 int f2fs_do_write_data_page(struct f2fs_io_info *fio); 3410 void f2fs_do_map_lock(struct f2fs_sb_info *sbi, int flag, bool lock); 3411 int f2fs_map_blocks(struct inode *inode, struct f2fs_map_blocks *map, 3412 int create, int flag); 3413 int f2fs_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, 3414 u64 start, u64 len); 3415 int f2fs_encrypt_one_page(struct f2fs_io_info *fio); 3416 bool f2fs_should_update_inplace(struct inode *inode, struct f2fs_io_info *fio); 3417 bool f2fs_should_update_outplace(struct inode *inode, struct f2fs_io_info *fio); 3418 int f2fs_write_single_data_page(struct page *page, int *submitted, 3419 struct bio **bio, sector_t *last_block, 3420 struct writeback_control *wbc, 3421 enum iostat_type io_type, 3422 int compr_blocks); 3423 void f2fs_invalidate_page(struct page *page, unsigned int offset, 3424 unsigned int length); 3425 int f2fs_release_page(struct page *page, gfp_t wait); 3426 #ifdef CONFIG_MIGRATION 3427 int f2fs_migrate_page(struct address_space *mapping, struct page *newpage, 3428 struct page *page, enum migrate_mode mode); 3429 #endif 3430 bool f2fs_overwrite_io(struct inode *inode, loff_t pos, size_t len); 3431 void f2fs_clear_page_cache_dirty_tag(struct page *page); 3432 int f2fs_init_post_read_processing(void); 3433 void f2fs_destroy_post_read_processing(void); 3434 int f2fs_init_post_read_wq(struct f2fs_sb_info *sbi); 3435 void f2fs_destroy_post_read_wq(struct f2fs_sb_info *sbi); 3436 3437 /* 3438 * gc.c 3439 */ 3440 int f2fs_start_gc_thread(struct f2fs_sb_info *sbi); 3441 void f2fs_stop_gc_thread(struct f2fs_sb_info *sbi); 3442 block_t f2fs_start_bidx_of_node(unsigned int node_ofs, struct inode *inode); 3443 int f2fs_gc(struct f2fs_sb_info *sbi, bool sync, bool background, 3444 unsigned int segno); 3445 void f2fs_build_gc_manager(struct f2fs_sb_info *sbi); 3446 int f2fs_resize_fs(struct f2fs_sb_info *sbi, __u64 block_count); 3447 int __init f2fs_create_garbage_collection_cache(void); 3448 void f2fs_destroy_garbage_collection_cache(void); 3449 3450 /* 3451 * recovery.c 3452 */ 3453 int f2fs_recover_fsync_data(struct f2fs_sb_info *sbi, bool check_only); 3454 bool f2fs_space_for_roll_forward(struct f2fs_sb_info *sbi); 3455 3456 /* 3457 * debug.c 3458 */ 3459 #ifdef CONFIG_F2FS_STAT_FS 3460 struct f2fs_stat_info { 3461 struct list_head stat_list; 3462 struct f2fs_sb_info *sbi; 3463 int all_area_segs, sit_area_segs, nat_area_segs, ssa_area_segs; 3464 int main_area_segs, main_area_sections, main_area_zones; 3465 unsigned long long hit_largest, hit_cached, hit_rbtree; 3466 unsigned long long hit_total, total_ext; 3467 int ext_tree, zombie_tree, ext_node; 3468 int ndirty_node, ndirty_dent, ndirty_meta, ndirty_imeta; 3469 int ndirty_data, ndirty_qdata; 3470 int inmem_pages; 3471 unsigned int ndirty_dirs, ndirty_files, nquota_files, ndirty_all; 3472 int nats, dirty_nats, sits, dirty_sits; 3473 int free_nids, avail_nids, alloc_nids; 3474 int total_count, utilization; 3475 int bg_gc, nr_wb_cp_data, nr_wb_data; 3476 int nr_rd_data, nr_rd_node, nr_rd_meta; 3477 int nr_dio_read, nr_dio_write; 3478 unsigned int io_skip_bggc, other_skip_bggc; 3479 int nr_flushing, nr_flushed, flush_list_empty; 3480 int nr_discarding, nr_discarded; 3481 int nr_discard_cmd; 3482 unsigned int undiscard_blks; 3483 int inline_xattr, inline_inode, inline_dir, append, update, orphans; 3484 int compr_inode; 3485 unsigned long long compr_blocks; 3486 int aw_cnt, max_aw_cnt, vw_cnt, max_vw_cnt; 3487 unsigned int valid_count, valid_node_count, valid_inode_count, discard_blks; 3488 unsigned int bimodal, avg_vblocks; 3489 int util_free, util_valid, util_invalid; 3490 int rsvd_segs, overp_segs; 3491 int dirty_count, node_pages, meta_pages; 3492 int prefree_count, call_count, cp_count, bg_cp_count; 3493 int tot_segs, node_segs, data_segs, free_segs, free_secs; 3494 int bg_node_segs, bg_data_segs; 3495 int tot_blks, data_blks, node_blks; 3496 int bg_data_blks, bg_node_blks; 3497 unsigned long long skipped_atomic_files[2]; 3498 int curseg[NR_CURSEG_TYPE]; 3499 int cursec[NR_CURSEG_TYPE]; 3500 int curzone[NR_CURSEG_TYPE]; 3501 unsigned int dirty_seg[NR_CURSEG_TYPE]; 3502 unsigned int full_seg[NR_CURSEG_TYPE]; 3503 unsigned int valid_blks[NR_CURSEG_TYPE]; 3504 3505 unsigned int meta_count[META_MAX]; 3506 unsigned int segment_count[2]; 3507 unsigned int block_count[2]; 3508 unsigned int inplace_count; 3509 unsigned long long base_mem, cache_mem, page_mem; 3510 }; 3511 3512 static inline struct f2fs_stat_info *F2FS_STAT(struct f2fs_sb_info *sbi) 3513 { 3514 return (struct f2fs_stat_info *)sbi->stat_info; 3515 } 3516 3517 #define stat_inc_cp_count(si) ((si)->cp_count++) 3518 #define stat_inc_bg_cp_count(si) ((si)->bg_cp_count++) 3519 #define stat_inc_call_count(si) ((si)->call_count++) 3520 #define stat_inc_bggc_count(si) ((si)->bg_gc++) 3521 #define stat_io_skip_bggc_count(sbi) ((sbi)->io_skip_bggc++) 3522 #define stat_other_skip_bggc_count(sbi) ((sbi)->other_skip_bggc++) 3523 #define stat_inc_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]++) 3524 #define stat_dec_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]--) 3525 #define stat_inc_total_hit(sbi) (atomic64_inc(&(sbi)->total_hit_ext)) 3526 #define stat_inc_rbtree_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_rbtree)) 3527 #define stat_inc_largest_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_largest)) 3528 #define stat_inc_cached_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_cached)) 3529 #define stat_inc_inline_xattr(inode) \ 3530 do { \ 3531 if (f2fs_has_inline_xattr(inode)) \ 3532 (atomic_inc(&F2FS_I_SB(inode)->inline_xattr)); \ 3533 } while (0) 3534 #define stat_dec_inline_xattr(inode) \ 3535 do { \ 3536 if (f2fs_has_inline_xattr(inode)) \ 3537 (atomic_dec(&F2FS_I_SB(inode)->inline_xattr)); \ 3538 } while (0) 3539 #define stat_inc_inline_inode(inode) \ 3540 do { \ 3541 if (f2fs_has_inline_data(inode)) \ 3542 (atomic_inc(&F2FS_I_SB(inode)->inline_inode)); \ 3543 } while (0) 3544 #define stat_dec_inline_inode(inode) \ 3545 do { \ 3546 if (f2fs_has_inline_data(inode)) \ 3547 (atomic_dec(&F2FS_I_SB(inode)->inline_inode)); \ 3548 } while (0) 3549 #define stat_inc_inline_dir(inode) \ 3550 do { \ 3551 if (f2fs_has_inline_dentry(inode)) \ 3552 (atomic_inc(&F2FS_I_SB(inode)->inline_dir)); \ 3553 } while (0) 3554 #define stat_dec_inline_dir(inode) \ 3555 do { \ 3556 if (f2fs_has_inline_dentry(inode)) \ 3557 (atomic_dec(&F2FS_I_SB(inode)->inline_dir)); \ 3558 } while (0) 3559 #define stat_inc_compr_inode(inode) \ 3560 do { \ 3561 if (f2fs_compressed_file(inode)) \ 3562 (atomic_inc(&F2FS_I_SB(inode)->compr_inode)); \ 3563 } while (0) 3564 #define stat_dec_compr_inode(inode) \ 3565 do { \ 3566 if (f2fs_compressed_file(inode)) \ 3567 (atomic_dec(&F2FS_I_SB(inode)->compr_inode)); \ 3568 } while (0) 3569 #define stat_add_compr_blocks(inode, blocks) \ 3570 (atomic64_add(blocks, &F2FS_I_SB(inode)->compr_blocks)) 3571 #define stat_sub_compr_blocks(inode, blocks) \ 3572 (atomic64_sub(blocks, &F2FS_I_SB(inode)->compr_blocks)) 3573 #define stat_inc_meta_count(sbi, blkaddr) \ 3574 do { \ 3575 if (blkaddr < SIT_I(sbi)->sit_base_addr) \ 3576 atomic_inc(&(sbi)->meta_count[META_CP]); \ 3577 else if (blkaddr < NM_I(sbi)->nat_blkaddr) \ 3578 atomic_inc(&(sbi)->meta_count[META_SIT]); \ 3579 else if (blkaddr < SM_I(sbi)->ssa_blkaddr) \ 3580 atomic_inc(&(sbi)->meta_count[META_NAT]); \ 3581 else if (blkaddr < SM_I(sbi)->main_blkaddr) \ 3582 atomic_inc(&(sbi)->meta_count[META_SSA]); \ 3583 } while (0) 3584 #define stat_inc_seg_type(sbi, curseg) \ 3585 ((sbi)->segment_count[(curseg)->alloc_type]++) 3586 #define stat_inc_block_count(sbi, curseg) \ 3587 ((sbi)->block_count[(curseg)->alloc_type]++) 3588 #define stat_inc_inplace_blocks(sbi) \ 3589 (atomic_inc(&(sbi)->inplace_count)) 3590 #define stat_update_max_atomic_write(inode) \ 3591 do { \ 3592 int cur = F2FS_I_SB(inode)->atomic_files; \ 3593 int max = atomic_read(&F2FS_I_SB(inode)->max_aw_cnt); \ 3594 if (cur > max) \ 3595 atomic_set(&F2FS_I_SB(inode)->max_aw_cnt, cur); \ 3596 } while (0) 3597 #define stat_inc_volatile_write(inode) \ 3598 (atomic_inc(&F2FS_I_SB(inode)->vw_cnt)) 3599 #define stat_dec_volatile_write(inode) \ 3600 (atomic_dec(&F2FS_I_SB(inode)->vw_cnt)) 3601 #define stat_update_max_volatile_write(inode) \ 3602 do { \ 3603 int cur = atomic_read(&F2FS_I_SB(inode)->vw_cnt); \ 3604 int max = atomic_read(&F2FS_I_SB(inode)->max_vw_cnt); \ 3605 if (cur > max) \ 3606 atomic_set(&F2FS_I_SB(inode)->max_vw_cnt, cur); \ 3607 } while (0) 3608 #define stat_inc_seg_count(sbi, type, gc_type) \ 3609 do { \ 3610 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 3611 si->tot_segs++; \ 3612 if ((type) == SUM_TYPE_DATA) { \ 3613 si->data_segs++; \ 3614 si->bg_data_segs += (gc_type == BG_GC) ? 1 : 0; \ 3615 } else { \ 3616 si->node_segs++; \ 3617 si->bg_node_segs += (gc_type == BG_GC) ? 1 : 0; \ 3618 } \ 3619 } while (0) 3620 3621 #define stat_inc_tot_blk_count(si, blks) \ 3622 ((si)->tot_blks += (blks)) 3623 3624 #define stat_inc_data_blk_count(sbi, blks, gc_type) \ 3625 do { \ 3626 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 3627 stat_inc_tot_blk_count(si, blks); \ 3628 si->data_blks += (blks); \ 3629 si->bg_data_blks += ((gc_type) == BG_GC) ? (blks) : 0; \ 3630 } while (0) 3631 3632 #define stat_inc_node_blk_count(sbi, blks, gc_type) \ 3633 do { \ 3634 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 3635 stat_inc_tot_blk_count(si, blks); \ 3636 si->node_blks += (blks); \ 3637 si->bg_node_blks += ((gc_type) == BG_GC) ? (blks) : 0; \ 3638 } while (0) 3639 3640 int f2fs_build_stats(struct f2fs_sb_info *sbi); 3641 void f2fs_destroy_stats(struct f2fs_sb_info *sbi); 3642 void __init f2fs_create_root_stats(void); 3643 void f2fs_destroy_root_stats(void); 3644 void f2fs_update_sit_info(struct f2fs_sb_info *sbi); 3645 #else 3646 #define stat_inc_cp_count(si) do { } while (0) 3647 #define stat_inc_bg_cp_count(si) do { } while (0) 3648 #define stat_inc_call_count(si) do { } while (0) 3649 #define stat_inc_bggc_count(si) do { } while (0) 3650 #define stat_io_skip_bggc_count(sbi) do { } while (0) 3651 #define stat_other_skip_bggc_count(sbi) do { } while (0) 3652 #define stat_inc_dirty_inode(sbi, type) do { } while (0) 3653 #define stat_dec_dirty_inode(sbi, type) do { } while (0) 3654 #define stat_inc_total_hit(sbi) do { } while (0) 3655 #define stat_inc_rbtree_node_hit(sbi) do { } while (0) 3656 #define stat_inc_largest_node_hit(sbi) do { } while (0) 3657 #define stat_inc_cached_node_hit(sbi) do { } while (0) 3658 #define stat_inc_inline_xattr(inode) do { } while (0) 3659 #define stat_dec_inline_xattr(inode) do { } while (0) 3660 #define stat_inc_inline_inode(inode) do { } while (0) 3661 #define stat_dec_inline_inode(inode) do { } while (0) 3662 #define stat_inc_inline_dir(inode) do { } while (0) 3663 #define stat_dec_inline_dir(inode) do { } while (0) 3664 #define stat_inc_compr_inode(inode) do { } while (0) 3665 #define stat_dec_compr_inode(inode) do { } while (0) 3666 #define stat_add_compr_blocks(inode, blocks) do { } while (0) 3667 #define stat_sub_compr_blocks(inode, blocks) do { } while (0) 3668 #define stat_inc_atomic_write(inode) do { } while (0) 3669 #define stat_dec_atomic_write(inode) do { } while (0) 3670 #define stat_update_max_atomic_write(inode) do { } while (0) 3671 #define stat_inc_volatile_write(inode) do { } while (0) 3672 #define stat_dec_volatile_write(inode) do { } while (0) 3673 #define stat_update_max_volatile_write(inode) do { } while (0) 3674 #define stat_inc_meta_count(sbi, blkaddr) do { } while (0) 3675 #define stat_inc_seg_type(sbi, curseg) do { } while (0) 3676 #define stat_inc_block_count(sbi, curseg) do { } while (0) 3677 #define stat_inc_inplace_blocks(sbi) do { } while (0) 3678 #define stat_inc_seg_count(sbi, type, gc_type) do { } while (0) 3679 #define stat_inc_tot_blk_count(si, blks) do { } while (0) 3680 #define stat_inc_data_blk_count(sbi, blks, gc_type) do { } while (0) 3681 #define stat_inc_node_blk_count(sbi, blks, gc_type) do { } while (0) 3682 3683 static inline int f2fs_build_stats(struct f2fs_sb_info *sbi) { return 0; } 3684 static inline void f2fs_destroy_stats(struct f2fs_sb_info *sbi) { } 3685 static inline void __init f2fs_create_root_stats(void) { } 3686 static inline void f2fs_destroy_root_stats(void) { } 3687 static inline void f2fs_update_sit_info(struct f2fs_sb_info *sbi) {} 3688 #endif 3689 3690 extern const struct file_operations f2fs_dir_operations; 3691 #ifdef CONFIG_UNICODE 3692 extern const struct dentry_operations f2fs_dentry_ops; 3693 #endif 3694 extern const struct file_operations f2fs_file_operations; 3695 extern const struct inode_operations f2fs_file_inode_operations; 3696 extern const struct address_space_operations f2fs_dblock_aops; 3697 extern const struct address_space_operations f2fs_node_aops; 3698 extern const struct address_space_operations f2fs_meta_aops; 3699 extern const struct inode_operations f2fs_dir_inode_operations; 3700 extern const struct inode_operations f2fs_symlink_inode_operations; 3701 extern const struct inode_operations f2fs_encrypted_symlink_inode_operations; 3702 extern const struct inode_operations f2fs_special_inode_operations; 3703 extern struct kmem_cache *f2fs_inode_entry_slab; 3704 3705 /* 3706 * inline.c 3707 */ 3708 bool f2fs_may_inline_data(struct inode *inode); 3709 bool f2fs_may_inline_dentry(struct inode *inode); 3710 void f2fs_do_read_inline_data(struct page *page, struct page *ipage); 3711 void f2fs_truncate_inline_inode(struct inode *inode, 3712 struct page *ipage, u64 from); 3713 int f2fs_read_inline_data(struct inode *inode, struct page *page); 3714 int f2fs_convert_inline_page(struct dnode_of_data *dn, struct page *page); 3715 int f2fs_convert_inline_inode(struct inode *inode); 3716 int f2fs_try_convert_inline_dir(struct inode *dir, struct dentry *dentry); 3717 int f2fs_write_inline_data(struct inode *inode, struct page *page); 3718 int f2fs_recover_inline_data(struct inode *inode, struct page *npage); 3719 struct f2fs_dir_entry *f2fs_find_in_inline_dir(struct inode *dir, 3720 const struct f2fs_filename *fname, 3721 struct page **res_page); 3722 int f2fs_make_empty_inline_dir(struct inode *inode, struct inode *parent, 3723 struct page *ipage); 3724 int f2fs_add_inline_entry(struct inode *dir, const struct f2fs_filename *fname, 3725 struct inode *inode, nid_t ino, umode_t mode); 3726 void f2fs_delete_inline_entry(struct f2fs_dir_entry *dentry, 3727 struct page *page, struct inode *dir, 3728 struct inode *inode); 3729 bool f2fs_empty_inline_dir(struct inode *dir); 3730 int f2fs_read_inline_dir(struct file *file, struct dir_context *ctx, 3731 struct fscrypt_str *fstr); 3732 int f2fs_inline_data_fiemap(struct inode *inode, 3733 struct fiemap_extent_info *fieinfo, 3734 __u64 start, __u64 len); 3735 3736 /* 3737 * shrinker.c 3738 */ 3739 unsigned long f2fs_shrink_count(struct shrinker *shrink, 3740 struct shrink_control *sc); 3741 unsigned long f2fs_shrink_scan(struct shrinker *shrink, 3742 struct shrink_control *sc); 3743 void f2fs_join_shrinker(struct f2fs_sb_info *sbi); 3744 void f2fs_leave_shrinker(struct f2fs_sb_info *sbi); 3745 3746 /* 3747 * extent_cache.c 3748 */ 3749 struct rb_entry *f2fs_lookup_rb_tree(struct rb_root_cached *root, 3750 struct rb_entry *cached_re, unsigned int ofs); 3751 struct rb_node **f2fs_lookup_rb_tree_ext(struct f2fs_sb_info *sbi, 3752 struct rb_root_cached *root, 3753 struct rb_node **parent, 3754 unsigned long long key, bool *left_most); 3755 struct rb_node **f2fs_lookup_rb_tree_for_insert(struct f2fs_sb_info *sbi, 3756 struct rb_root_cached *root, 3757 struct rb_node **parent, 3758 unsigned int ofs, bool *leftmost); 3759 struct rb_entry *f2fs_lookup_rb_tree_ret(struct rb_root_cached *root, 3760 struct rb_entry *cached_re, unsigned int ofs, 3761 struct rb_entry **prev_entry, struct rb_entry **next_entry, 3762 struct rb_node ***insert_p, struct rb_node **insert_parent, 3763 bool force, bool *leftmost); 3764 bool f2fs_check_rb_tree_consistence(struct f2fs_sb_info *sbi, 3765 struct rb_root_cached *root, bool check_key); 3766 unsigned int f2fs_shrink_extent_tree(struct f2fs_sb_info *sbi, int nr_shrink); 3767 void f2fs_init_extent_tree(struct inode *inode, struct page *ipage); 3768 void f2fs_drop_extent_tree(struct inode *inode); 3769 unsigned int f2fs_destroy_extent_node(struct inode *inode); 3770 void f2fs_destroy_extent_tree(struct inode *inode); 3771 bool f2fs_lookup_extent_cache(struct inode *inode, pgoff_t pgofs, 3772 struct extent_info *ei); 3773 void f2fs_update_extent_cache(struct dnode_of_data *dn); 3774 void f2fs_update_extent_cache_range(struct dnode_of_data *dn, 3775 pgoff_t fofs, block_t blkaddr, unsigned int len); 3776 void f2fs_init_extent_cache_info(struct f2fs_sb_info *sbi); 3777 int __init f2fs_create_extent_cache(void); 3778 void f2fs_destroy_extent_cache(void); 3779 3780 /* 3781 * sysfs.c 3782 */ 3783 int __init f2fs_init_sysfs(void); 3784 void f2fs_exit_sysfs(void); 3785 int f2fs_register_sysfs(struct f2fs_sb_info *sbi); 3786 void f2fs_unregister_sysfs(struct f2fs_sb_info *sbi); 3787 3788 /* verity.c */ 3789 extern const struct fsverity_operations f2fs_verityops; 3790 3791 /* 3792 * crypto support 3793 */ 3794 static inline bool f2fs_encrypted_file(struct inode *inode) 3795 { 3796 return IS_ENCRYPTED(inode) && S_ISREG(inode->i_mode); 3797 } 3798 3799 static inline void f2fs_set_encrypted_inode(struct inode *inode) 3800 { 3801 #ifdef CONFIG_FS_ENCRYPTION 3802 file_set_encrypt(inode); 3803 f2fs_set_inode_flags(inode); 3804 #endif 3805 } 3806 3807 /* 3808 * Returns true if the reads of the inode's data need to undergo some 3809 * postprocessing step, like decryption or authenticity verification. 3810 */ 3811 static inline bool f2fs_post_read_required(struct inode *inode) 3812 { 3813 return f2fs_encrypted_file(inode) || fsverity_active(inode) || 3814 f2fs_compressed_file(inode); 3815 } 3816 3817 /* 3818 * compress.c 3819 */ 3820 #ifdef CONFIG_F2FS_FS_COMPRESSION 3821 bool f2fs_is_compressed_page(struct page *page); 3822 struct page *f2fs_compress_control_page(struct page *page); 3823 int f2fs_prepare_compress_overwrite(struct inode *inode, 3824 struct page **pagep, pgoff_t index, void **fsdata); 3825 bool f2fs_compress_write_end(struct inode *inode, void *fsdata, 3826 pgoff_t index, unsigned copied); 3827 int f2fs_truncate_partial_cluster(struct inode *inode, u64 from, bool lock); 3828 void f2fs_compress_write_end_io(struct bio *bio, struct page *page); 3829 bool f2fs_is_compress_backend_ready(struct inode *inode); 3830 int f2fs_init_compress_mempool(void); 3831 void f2fs_destroy_compress_mempool(void); 3832 void f2fs_decompress_pages(struct bio *bio, struct page *page, bool verity); 3833 bool f2fs_cluster_is_empty(struct compress_ctx *cc); 3834 bool f2fs_cluster_can_merge_page(struct compress_ctx *cc, pgoff_t index); 3835 void f2fs_compress_ctx_add_page(struct compress_ctx *cc, struct page *page); 3836 int f2fs_write_multi_pages(struct compress_ctx *cc, 3837 int *submitted, 3838 struct writeback_control *wbc, 3839 enum iostat_type io_type); 3840 int f2fs_is_compressed_cluster(struct inode *inode, pgoff_t index); 3841 int f2fs_read_multi_pages(struct compress_ctx *cc, struct bio **bio_ret, 3842 unsigned nr_pages, sector_t *last_block_in_bio, 3843 bool is_readahead, bool for_write); 3844 struct decompress_io_ctx *f2fs_alloc_dic(struct compress_ctx *cc); 3845 void f2fs_free_dic(struct decompress_io_ctx *dic); 3846 void f2fs_decompress_end_io(struct page **rpages, 3847 unsigned int cluster_size, bool err, bool verity); 3848 int f2fs_init_compress_ctx(struct compress_ctx *cc); 3849 void f2fs_destroy_compress_ctx(struct compress_ctx *cc); 3850 void f2fs_init_compress_info(struct f2fs_sb_info *sbi); 3851 int f2fs_init_page_array_cache(struct f2fs_sb_info *sbi); 3852 void f2fs_destroy_page_array_cache(struct f2fs_sb_info *sbi); 3853 int __init f2fs_init_compress_cache(void); 3854 void f2fs_destroy_compress_cache(void); 3855 #else 3856 static inline bool f2fs_is_compressed_page(struct page *page) { return false; } 3857 static inline bool f2fs_is_compress_backend_ready(struct inode *inode) 3858 { 3859 if (!f2fs_compressed_file(inode)) 3860 return true; 3861 /* not support compression */ 3862 return false; 3863 } 3864 static inline struct page *f2fs_compress_control_page(struct page *page) 3865 { 3866 WARN_ON_ONCE(1); 3867 return ERR_PTR(-EINVAL); 3868 } 3869 static inline int f2fs_init_compress_mempool(void) { return 0; } 3870 static inline void f2fs_destroy_compress_mempool(void) { } 3871 static inline int f2fs_init_page_array_cache(struct f2fs_sb_info *sbi) { return 0; } 3872 static inline void f2fs_destroy_page_array_cache(struct f2fs_sb_info *sbi) { } 3873 static inline int __init f2fs_init_compress_cache(void) { return 0; } 3874 static inline void f2fs_destroy_compress_cache(void) { } 3875 #endif 3876 3877 static inline void set_compress_context(struct inode *inode) 3878 { 3879 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 3880 3881 F2FS_I(inode)->i_compress_algorithm = 3882 F2FS_OPTION(sbi).compress_algorithm; 3883 F2FS_I(inode)->i_log_cluster_size = 3884 F2FS_OPTION(sbi).compress_log_size; 3885 F2FS_I(inode)->i_cluster_size = 3886 1 << F2FS_I(inode)->i_log_cluster_size; 3887 F2FS_I(inode)->i_flags |= F2FS_COMPR_FL; 3888 set_inode_flag(inode, FI_COMPRESSED_FILE); 3889 stat_inc_compr_inode(inode); 3890 f2fs_mark_inode_dirty_sync(inode, true); 3891 } 3892 3893 static inline bool f2fs_disable_compressed_file(struct inode *inode) 3894 { 3895 struct f2fs_inode_info *fi = F2FS_I(inode); 3896 3897 if (!f2fs_compressed_file(inode)) 3898 return true; 3899 if (S_ISREG(inode->i_mode) && 3900 (get_dirty_pages(inode) || atomic_read(&fi->i_compr_blocks))) 3901 return false; 3902 3903 fi->i_flags &= ~F2FS_COMPR_FL; 3904 stat_dec_compr_inode(inode); 3905 clear_inode_flag(inode, FI_COMPRESSED_FILE); 3906 f2fs_mark_inode_dirty_sync(inode, true); 3907 return true; 3908 } 3909 3910 #define F2FS_FEATURE_FUNCS(name, flagname) \ 3911 static inline int f2fs_sb_has_##name(struct f2fs_sb_info *sbi) \ 3912 { \ 3913 return F2FS_HAS_FEATURE(sbi, F2FS_FEATURE_##flagname); \ 3914 } 3915 3916 F2FS_FEATURE_FUNCS(encrypt, ENCRYPT); 3917 F2FS_FEATURE_FUNCS(blkzoned, BLKZONED); 3918 F2FS_FEATURE_FUNCS(extra_attr, EXTRA_ATTR); 3919 F2FS_FEATURE_FUNCS(project_quota, PRJQUOTA); 3920 F2FS_FEATURE_FUNCS(inode_chksum, INODE_CHKSUM); 3921 F2FS_FEATURE_FUNCS(flexible_inline_xattr, FLEXIBLE_INLINE_XATTR); 3922 F2FS_FEATURE_FUNCS(quota_ino, QUOTA_INO); 3923 F2FS_FEATURE_FUNCS(inode_crtime, INODE_CRTIME); 3924 F2FS_FEATURE_FUNCS(lost_found, LOST_FOUND); 3925 F2FS_FEATURE_FUNCS(verity, VERITY); 3926 F2FS_FEATURE_FUNCS(sb_chksum, SB_CHKSUM); 3927 F2FS_FEATURE_FUNCS(casefold, CASEFOLD); 3928 F2FS_FEATURE_FUNCS(compression, COMPRESSION); 3929 3930 #ifdef CONFIG_BLK_DEV_ZONED 3931 static inline bool f2fs_blkz_is_seq(struct f2fs_sb_info *sbi, int devi, 3932 block_t blkaddr) 3933 { 3934 unsigned int zno = blkaddr >> sbi->log_blocks_per_blkz; 3935 3936 return test_bit(zno, FDEV(devi).blkz_seq); 3937 } 3938 #endif 3939 3940 static inline bool f2fs_hw_should_discard(struct f2fs_sb_info *sbi) 3941 { 3942 return f2fs_sb_has_blkzoned(sbi); 3943 } 3944 3945 static inline bool f2fs_bdev_support_discard(struct block_device *bdev) 3946 { 3947 return blk_queue_discard(bdev_get_queue(bdev)) || 3948 bdev_is_zoned(bdev); 3949 } 3950 3951 static inline bool f2fs_hw_support_discard(struct f2fs_sb_info *sbi) 3952 { 3953 int i; 3954 3955 if (!f2fs_is_multi_device(sbi)) 3956 return f2fs_bdev_support_discard(sbi->sb->s_bdev); 3957 3958 for (i = 0; i < sbi->s_ndevs; i++) 3959 if (f2fs_bdev_support_discard(FDEV(i).bdev)) 3960 return true; 3961 return false; 3962 } 3963 3964 static inline bool f2fs_realtime_discard_enable(struct f2fs_sb_info *sbi) 3965 { 3966 return (test_opt(sbi, DISCARD) && f2fs_hw_support_discard(sbi)) || 3967 f2fs_hw_should_discard(sbi); 3968 } 3969 3970 static inline bool f2fs_hw_is_readonly(struct f2fs_sb_info *sbi) 3971 { 3972 int i; 3973 3974 if (!f2fs_is_multi_device(sbi)) 3975 return bdev_read_only(sbi->sb->s_bdev); 3976 3977 for (i = 0; i < sbi->s_ndevs; i++) 3978 if (bdev_read_only(FDEV(i).bdev)) 3979 return true; 3980 return false; 3981 } 3982 3983 static inline bool f2fs_lfs_mode(struct f2fs_sb_info *sbi) 3984 { 3985 return F2FS_OPTION(sbi).fs_mode == FS_MODE_LFS; 3986 } 3987 3988 static inline bool f2fs_may_compress(struct inode *inode) 3989 { 3990 if (IS_SWAPFILE(inode) || f2fs_is_pinned_file(inode) || 3991 f2fs_is_atomic_file(inode) || 3992 f2fs_is_volatile_file(inode)) 3993 return false; 3994 return S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode); 3995 } 3996 3997 static inline void f2fs_i_compr_blocks_update(struct inode *inode, 3998 u64 blocks, bool add) 3999 { 4000 int diff = F2FS_I(inode)->i_cluster_size - blocks; 4001 struct f2fs_inode_info *fi = F2FS_I(inode); 4002 4003 /* don't update i_compr_blocks if saved blocks were released */ 4004 if (!add && !atomic_read(&fi->i_compr_blocks)) 4005 return; 4006 4007 if (add) { 4008 atomic_add(diff, &fi->i_compr_blocks); 4009 stat_add_compr_blocks(inode, diff); 4010 } else { 4011 atomic_sub(diff, &fi->i_compr_blocks); 4012 stat_sub_compr_blocks(inode, diff); 4013 } 4014 f2fs_mark_inode_dirty_sync(inode, true); 4015 } 4016 4017 static inline int block_unaligned_IO(struct inode *inode, 4018 struct kiocb *iocb, struct iov_iter *iter) 4019 { 4020 unsigned int i_blkbits = READ_ONCE(inode->i_blkbits); 4021 unsigned int blocksize_mask = (1 << i_blkbits) - 1; 4022 loff_t offset = iocb->ki_pos; 4023 unsigned long align = offset | iov_iter_alignment(iter); 4024 4025 return align & blocksize_mask; 4026 } 4027 4028 static inline int allow_outplace_dio(struct inode *inode, 4029 struct kiocb *iocb, struct iov_iter *iter) 4030 { 4031 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 4032 int rw = iov_iter_rw(iter); 4033 4034 return (f2fs_lfs_mode(sbi) && (rw == WRITE) && 4035 !block_unaligned_IO(inode, iocb, iter)); 4036 } 4037 4038 static inline bool f2fs_force_buffered_io(struct inode *inode, 4039 struct kiocb *iocb, struct iov_iter *iter) 4040 { 4041 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 4042 int rw = iov_iter_rw(iter); 4043 4044 if (f2fs_post_read_required(inode)) 4045 return true; 4046 if (f2fs_is_multi_device(sbi)) 4047 return true; 4048 /* 4049 * for blkzoned device, fallback direct IO to buffered IO, so 4050 * all IOs can be serialized by log-structured write. 4051 */ 4052 if (f2fs_sb_has_blkzoned(sbi)) 4053 return true; 4054 if (f2fs_lfs_mode(sbi) && (rw == WRITE)) { 4055 if (block_unaligned_IO(inode, iocb, iter)) 4056 return true; 4057 if (F2FS_IO_ALIGNED(sbi)) 4058 return true; 4059 } 4060 if (is_sbi_flag_set(F2FS_I_SB(inode), SBI_CP_DISABLED) && 4061 !IS_SWAPFILE(inode)) 4062 return true; 4063 4064 return false; 4065 } 4066 4067 #ifdef CONFIG_F2FS_FAULT_INJECTION 4068 extern void f2fs_build_fault_attr(struct f2fs_sb_info *sbi, unsigned int rate, 4069 unsigned int type); 4070 #else 4071 #define f2fs_build_fault_attr(sbi, rate, type) do { } while (0) 4072 #endif 4073 4074 static inline bool is_journalled_quota(struct f2fs_sb_info *sbi) 4075 { 4076 #ifdef CONFIG_QUOTA 4077 if (f2fs_sb_has_quota_ino(sbi)) 4078 return true; 4079 if (F2FS_OPTION(sbi).s_qf_names[USRQUOTA] || 4080 F2FS_OPTION(sbi).s_qf_names[GRPQUOTA] || 4081 F2FS_OPTION(sbi).s_qf_names[PRJQUOTA]) 4082 return true; 4083 #endif 4084 return false; 4085 } 4086 4087 #define EFSBADCRC EBADMSG /* Bad CRC detected */ 4088 #define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */ 4089 4090 #endif /* _LINUX_F2FS_H */ 4091