1 /* SPDX-License-Identifier: GPL-2.0 */ 2 /* 3 * fs/f2fs/f2fs.h 4 * 5 * Copyright (c) 2012 Samsung Electronics Co., Ltd. 6 * http://www.samsung.com/ 7 */ 8 #ifndef _LINUX_F2FS_H 9 #define _LINUX_F2FS_H 10 11 #include <linux/uio.h> 12 #include <linux/types.h> 13 #include <linux/page-flags.h> 14 #include <linux/buffer_head.h> 15 #include <linux/slab.h> 16 #include <linux/crc32.h> 17 #include <linux/magic.h> 18 #include <linux/kobject.h> 19 #include <linux/sched.h> 20 #include <linux/cred.h> 21 #include <linux/vmalloc.h> 22 #include <linux/bio.h> 23 #include <linux/blkdev.h> 24 #include <linux/quotaops.h> 25 #include <linux/part_stat.h> 26 #include <crypto/hash.h> 27 28 #include <linux/fscrypt.h> 29 #include <linux/fsverity.h> 30 31 #ifdef CONFIG_F2FS_CHECK_FS 32 #define f2fs_bug_on(sbi, condition) BUG_ON(condition) 33 #else 34 #define f2fs_bug_on(sbi, condition) \ 35 do { \ 36 if (WARN_ON(condition)) \ 37 set_sbi_flag(sbi, SBI_NEED_FSCK); \ 38 } while (0) 39 #endif 40 41 enum { 42 FAULT_KMALLOC, 43 FAULT_KVMALLOC, 44 FAULT_PAGE_ALLOC, 45 FAULT_PAGE_GET, 46 FAULT_ALLOC_BIO, 47 FAULT_ALLOC_NID, 48 FAULT_ORPHAN, 49 FAULT_BLOCK, 50 FAULT_DIR_DEPTH, 51 FAULT_EVICT_INODE, 52 FAULT_TRUNCATE, 53 FAULT_READ_IO, 54 FAULT_CHECKPOINT, 55 FAULT_DISCARD, 56 FAULT_WRITE_IO, 57 FAULT_MAX, 58 }; 59 60 #ifdef CONFIG_F2FS_FAULT_INJECTION 61 #define F2FS_ALL_FAULT_TYPE ((1 << FAULT_MAX) - 1) 62 63 struct f2fs_fault_info { 64 atomic_t inject_ops; 65 unsigned int inject_rate; 66 unsigned int inject_type; 67 }; 68 69 extern const char *f2fs_fault_name[FAULT_MAX]; 70 #define IS_FAULT_SET(fi, type) ((fi)->inject_type & (1 << (type))) 71 #endif 72 73 /* 74 * For mount options 75 */ 76 #define F2FS_MOUNT_DISABLE_ROLL_FORWARD 0x00000002 77 #define F2FS_MOUNT_DISCARD 0x00000004 78 #define F2FS_MOUNT_NOHEAP 0x00000008 79 #define F2FS_MOUNT_XATTR_USER 0x00000010 80 #define F2FS_MOUNT_POSIX_ACL 0x00000020 81 #define F2FS_MOUNT_DISABLE_EXT_IDENTIFY 0x00000040 82 #define F2FS_MOUNT_INLINE_XATTR 0x00000080 83 #define F2FS_MOUNT_INLINE_DATA 0x00000100 84 #define F2FS_MOUNT_INLINE_DENTRY 0x00000200 85 #define F2FS_MOUNT_FLUSH_MERGE 0x00000400 86 #define F2FS_MOUNT_NOBARRIER 0x00000800 87 #define F2FS_MOUNT_FASTBOOT 0x00001000 88 #define F2FS_MOUNT_EXTENT_CACHE 0x00002000 89 #define F2FS_MOUNT_DATA_FLUSH 0x00008000 90 #define F2FS_MOUNT_FAULT_INJECTION 0x00010000 91 #define F2FS_MOUNT_USRQUOTA 0x00080000 92 #define F2FS_MOUNT_GRPQUOTA 0x00100000 93 #define F2FS_MOUNT_PRJQUOTA 0x00200000 94 #define F2FS_MOUNT_QUOTA 0x00400000 95 #define F2FS_MOUNT_INLINE_XATTR_SIZE 0x00800000 96 #define F2FS_MOUNT_RESERVE_ROOT 0x01000000 97 #define F2FS_MOUNT_DISABLE_CHECKPOINT 0x02000000 98 #define F2FS_MOUNT_NORECOVERY 0x04000000 99 #define F2FS_MOUNT_ATGC 0x08000000 100 101 #define F2FS_OPTION(sbi) ((sbi)->mount_opt) 102 #define clear_opt(sbi, option) (F2FS_OPTION(sbi).opt &= ~F2FS_MOUNT_##option) 103 #define set_opt(sbi, option) (F2FS_OPTION(sbi).opt |= F2FS_MOUNT_##option) 104 #define test_opt(sbi, option) (F2FS_OPTION(sbi).opt & F2FS_MOUNT_##option) 105 106 #define ver_after(a, b) (typecheck(unsigned long long, a) && \ 107 typecheck(unsigned long long, b) && \ 108 ((long long)((a) - (b)) > 0)) 109 110 typedef u32 block_t; /* 111 * should not change u32, since it is the on-disk block 112 * address format, __le32. 113 */ 114 typedef u32 nid_t; 115 116 #define COMPRESS_EXT_NUM 16 117 118 struct f2fs_mount_info { 119 unsigned int opt; 120 int write_io_size_bits; /* Write IO size bits */ 121 block_t root_reserved_blocks; /* root reserved blocks */ 122 kuid_t s_resuid; /* reserved blocks for uid */ 123 kgid_t s_resgid; /* reserved blocks for gid */ 124 int active_logs; /* # of active logs */ 125 int inline_xattr_size; /* inline xattr size */ 126 #ifdef CONFIG_F2FS_FAULT_INJECTION 127 struct f2fs_fault_info fault_info; /* For fault injection */ 128 #endif 129 #ifdef CONFIG_QUOTA 130 /* Names of quota files with journalled quota */ 131 char *s_qf_names[MAXQUOTAS]; 132 int s_jquota_fmt; /* Format of quota to use */ 133 #endif 134 /* For which write hints are passed down to block layer */ 135 int whint_mode; 136 int alloc_mode; /* segment allocation policy */ 137 int fsync_mode; /* fsync policy */ 138 int fs_mode; /* fs mode: LFS or ADAPTIVE */ 139 int bggc_mode; /* bggc mode: off, on or sync */ 140 struct fscrypt_dummy_policy dummy_enc_policy; /* test dummy encryption */ 141 block_t unusable_cap_perc; /* percentage for cap */ 142 block_t unusable_cap; /* Amount of space allowed to be 143 * unusable when disabling checkpoint 144 */ 145 146 /* For compression */ 147 unsigned char compress_algorithm; /* algorithm type */ 148 unsigned char compress_log_size; /* cluster log size */ 149 unsigned char compress_level; /* compress level */ 150 bool compress_chksum; /* compressed data chksum */ 151 unsigned char compress_ext_cnt; /* extension count */ 152 int compress_mode; /* compression mode */ 153 unsigned char extensions[COMPRESS_EXT_NUM][F2FS_EXTENSION_LEN]; /* extensions */ 154 }; 155 156 #define F2FS_FEATURE_ENCRYPT 0x0001 157 #define F2FS_FEATURE_BLKZONED 0x0002 158 #define F2FS_FEATURE_ATOMIC_WRITE 0x0004 159 #define F2FS_FEATURE_EXTRA_ATTR 0x0008 160 #define F2FS_FEATURE_PRJQUOTA 0x0010 161 #define F2FS_FEATURE_INODE_CHKSUM 0x0020 162 #define F2FS_FEATURE_FLEXIBLE_INLINE_XATTR 0x0040 163 #define F2FS_FEATURE_QUOTA_INO 0x0080 164 #define F2FS_FEATURE_INODE_CRTIME 0x0100 165 #define F2FS_FEATURE_LOST_FOUND 0x0200 166 #define F2FS_FEATURE_VERITY 0x0400 167 #define F2FS_FEATURE_SB_CHKSUM 0x0800 168 #define F2FS_FEATURE_CASEFOLD 0x1000 169 #define F2FS_FEATURE_COMPRESSION 0x2000 170 171 #define __F2FS_HAS_FEATURE(raw_super, mask) \ 172 ((raw_super->feature & cpu_to_le32(mask)) != 0) 173 #define F2FS_HAS_FEATURE(sbi, mask) __F2FS_HAS_FEATURE(sbi->raw_super, mask) 174 #define F2FS_SET_FEATURE(sbi, mask) \ 175 (sbi->raw_super->feature |= cpu_to_le32(mask)) 176 #define F2FS_CLEAR_FEATURE(sbi, mask) \ 177 (sbi->raw_super->feature &= ~cpu_to_le32(mask)) 178 179 /* 180 * Default values for user and/or group using reserved blocks 181 */ 182 #define F2FS_DEF_RESUID 0 183 #define F2FS_DEF_RESGID 0 184 185 /* 186 * For checkpoint manager 187 */ 188 enum { 189 NAT_BITMAP, 190 SIT_BITMAP 191 }; 192 193 #define CP_UMOUNT 0x00000001 194 #define CP_FASTBOOT 0x00000002 195 #define CP_SYNC 0x00000004 196 #define CP_RECOVERY 0x00000008 197 #define CP_DISCARD 0x00000010 198 #define CP_TRIMMED 0x00000020 199 #define CP_PAUSE 0x00000040 200 #define CP_RESIZE 0x00000080 201 202 #define MAX_DISCARD_BLOCKS(sbi) BLKS_PER_SEC(sbi) 203 #define DEF_MAX_DISCARD_REQUEST 8 /* issue 8 discards per round */ 204 #define DEF_MIN_DISCARD_ISSUE_TIME 50 /* 50 ms, if exists */ 205 #define DEF_MID_DISCARD_ISSUE_TIME 500 /* 500 ms, if device busy */ 206 #define DEF_MAX_DISCARD_ISSUE_TIME 60000 /* 60 s, if no candidates */ 207 #define DEF_DISCARD_URGENT_UTIL 80 /* do more discard over 80% */ 208 #define DEF_CP_INTERVAL 60 /* 60 secs */ 209 #define DEF_IDLE_INTERVAL 5 /* 5 secs */ 210 #define DEF_DISABLE_INTERVAL 5 /* 5 secs */ 211 #define DEF_DISABLE_QUICK_INTERVAL 1 /* 1 secs */ 212 #define DEF_UMOUNT_DISCARD_TIMEOUT 5 /* 5 secs */ 213 214 struct cp_control { 215 int reason; 216 __u64 trim_start; 217 __u64 trim_end; 218 __u64 trim_minlen; 219 }; 220 221 /* 222 * indicate meta/data type 223 */ 224 enum { 225 META_CP, 226 META_NAT, 227 META_SIT, 228 META_SSA, 229 META_MAX, 230 META_POR, 231 DATA_GENERIC, /* check range only */ 232 DATA_GENERIC_ENHANCE, /* strong check on range and segment bitmap */ 233 DATA_GENERIC_ENHANCE_READ, /* 234 * strong check on range and segment 235 * bitmap but no warning due to race 236 * condition of read on truncated area 237 * by extent_cache 238 */ 239 META_GENERIC, 240 }; 241 242 /* for the list of ino */ 243 enum { 244 ORPHAN_INO, /* for orphan ino list */ 245 APPEND_INO, /* for append ino list */ 246 UPDATE_INO, /* for update ino list */ 247 TRANS_DIR_INO, /* for trasactions dir ino list */ 248 FLUSH_INO, /* for multiple device flushing */ 249 MAX_INO_ENTRY, /* max. list */ 250 }; 251 252 struct ino_entry { 253 struct list_head list; /* list head */ 254 nid_t ino; /* inode number */ 255 unsigned int dirty_device; /* dirty device bitmap */ 256 }; 257 258 /* for the list of inodes to be GCed */ 259 struct inode_entry { 260 struct list_head list; /* list head */ 261 struct inode *inode; /* vfs inode pointer */ 262 }; 263 264 struct fsync_node_entry { 265 struct list_head list; /* list head */ 266 struct page *page; /* warm node page pointer */ 267 unsigned int seq_id; /* sequence id */ 268 }; 269 270 /* for the bitmap indicate blocks to be discarded */ 271 struct discard_entry { 272 struct list_head list; /* list head */ 273 block_t start_blkaddr; /* start blockaddr of current segment */ 274 unsigned char discard_map[SIT_VBLOCK_MAP_SIZE]; /* segment discard bitmap */ 275 }; 276 277 /* default discard granularity of inner discard thread, unit: block count */ 278 #define DEFAULT_DISCARD_GRANULARITY 16 279 280 /* max discard pend list number */ 281 #define MAX_PLIST_NUM 512 282 #define plist_idx(blk_num) ((blk_num) >= MAX_PLIST_NUM ? \ 283 (MAX_PLIST_NUM - 1) : ((blk_num) - 1)) 284 285 enum { 286 D_PREP, /* initial */ 287 D_PARTIAL, /* partially submitted */ 288 D_SUBMIT, /* all submitted */ 289 D_DONE, /* finished */ 290 }; 291 292 struct discard_info { 293 block_t lstart; /* logical start address */ 294 block_t len; /* length */ 295 block_t start; /* actual start address in dev */ 296 }; 297 298 struct discard_cmd { 299 struct rb_node rb_node; /* rb node located in rb-tree */ 300 union { 301 struct { 302 block_t lstart; /* logical start address */ 303 block_t len; /* length */ 304 block_t start; /* actual start address in dev */ 305 }; 306 struct discard_info di; /* discard info */ 307 308 }; 309 struct list_head list; /* command list */ 310 struct completion wait; /* compleation */ 311 struct block_device *bdev; /* bdev */ 312 unsigned short ref; /* reference count */ 313 unsigned char state; /* state */ 314 unsigned char queued; /* queued discard */ 315 int error; /* bio error */ 316 spinlock_t lock; /* for state/bio_ref updating */ 317 unsigned short bio_ref; /* bio reference count */ 318 }; 319 320 enum { 321 DPOLICY_BG, 322 DPOLICY_FORCE, 323 DPOLICY_FSTRIM, 324 DPOLICY_UMOUNT, 325 MAX_DPOLICY, 326 }; 327 328 struct discard_policy { 329 int type; /* type of discard */ 330 unsigned int min_interval; /* used for candidates exist */ 331 unsigned int mid_interval; /* used for device busy */ 332 unsigned int max_interval; /* used for candidates not exist */ 333 unsigned int max_requests; /* # of discards issued per round */ 334 unsigned int io_aware_gran; /* minimum granularity discard not be aware of I/O */ 335 bool io_aware; /* issue discard in idle time */ 336 bool sync; /* submit discard with REQ_SYNC flag */ 337 bool ordered; /* issue discard by lba order */ 338 bool timeout; /* discard timeout for put_super */ 339 unsigned int granularity; /* discard granularity */ 340 }; 341 342 struct discard_cmd_control { 343 struct task_struct *f2fs_issue_discard; /* discard thread */ 344 struct list_head entry_list; /* 4KB discard entry list */ 345 struct list_head pend_list[MAX_PLIST_NUM];/* store pending entries */ 346 struct list_head wait_list; /* store on-flushing entries */ 347 struct list_head fstrim_list; /* in-flight discard from fstrim */ 348 wait_queue_head_t discard_wait_queue; /* waiting queue for wake-up */ 349 unsigned int discard_wake; /* to wake up discard thread */ 350 struct mutex cmd_lock; 351 unsigned int nr_discards; /* # of discards in the list */ 352 unsigned int max_discards; /* max. discards to be issued */ 353 unsigned int discard_granularity; /* discard granularity */ 354 unsigned int undiscard_blks; /* # of undiscard blocks */ 355 unsigned int next_pos; /* next discard position */ 356 atomic_t issued_discard; /* # of issued discard */ 357 atomic_t queued_discard; /* # of queued discard */ 358 atomic_t discard_cmd_cnt; /* # of cached cmd count */ 359 struct rb_root_cached root; /* root of discard rb-tree */ 360 bool rbtree_check; /* config for consistence check */ 361 }; 362 363 /* for the list of fsync inodes, used only during recovery */ 364 struct fsync_inode_entry { 365 struct list_head list; /* list head */ 366 struct inode *inode; /* vfs inode pointer */ 367 block_t blkaddr; /* block address locating the last fsync */ 368 block_t last_dentry; /* block address locating the last dentry */ 369 }; 370 371 #define nats_in_cursum(jnl) (le16_to_cpu((jnl)->n_nats)) 372 #define sits_in_cursum(jnl) (le16_to_cpu((jnl)->n_sits)) 373 374 #define nat_in_journal(jnl, i) ((jnl)->nat_j.entries[i].ne) 375 #define nid_in_journal(jnl, i) ((jnl)->nat_j.entries[i].nid) 376 #define sit_in_journal(jnl, i) ((jnl)->sit_j.entries[i].se) 377 #define segno_in_journal(jnl, i) ((jnl)->sit_j.entries[i].segno) 378 379 #define MAX_NAT_JENTRIES(jnl) (NAT_JOURNAL_ENTRIES - nats_in_cursum(jnl)) 380 #define MAX_SIT_JENTRIES(jnl) (SIT_JOURNAL_ENTRIES - sits_in_cursum(jnl)) 381 382 static inline int update_nats_in_cursum(struct f2fs_journal *journal, int i) 383 { 384 int before = nats_in_cursum(journal); 385 386 journal->n_nats = cpu_to_le16(before + i); 387 return before; 388 } 389 390 static inline int update_sits_in_cursum(struct f2fs_journal *journal, int i) 391 { 392 int before = sits_in_cursum(journal); 393 394 journal->n_sits = cpu_to_le16(before + i); 395 return before; 396 } 397 398 static inline bool __has_cursum_space(struct f2fs_journal *journal, 399 int size, int type) 400 { 401 if (type == NAT_JOURNAL) 402 return size <= MAX_NAT_JENTRIES(journal); 403 return size <= MAX_SIT_JENTRIES(journal); 404 } 405 406 /* for inline stuff */ 407 #define DEF_INLINE_RESERVED_SIZE 1 408 static inline int get_extra_isize(struct inode *inode); 409 static inline int get_inline_xattr_addrs(struct inode *inode); 410 #define MAX_INLINE_DATA(inode) (sizeof(__le32) * \ 411 (CUR_ADDRS_PER_INODE(inode) - \ 412 get_inline_xattr_addrs(inode) - \ 413 DEF_INLINE_RESERVED_SIZE)) 414 415 /* for inline dir */ 416 #define NR_INLINE_DENTRY(inode) (MAX_INLINE_DATA(inode) * BITS_PER_BYTE / \ 417 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \ 418 BITS_PER_BYTE + 1)) 419 #define INLINE_DENTRY_BITMAP_SIZE(inode) \ 420 DIV_ROUND_UP(NR_INLINE_DENTRY(inode), BITS_PER_BYTE) 421 #define INLINE_RESERVED_SIZE(inode) (MAX_INLINE_DATA(inode) - \ 422 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \ 423 NR_INLINE_DENTRY(inode) + \ 424 INLINE_DENTRY_BITMAP_SIZE(inode))) 425 426 /* 427 * For INODE and NODE manager 428 */ 429 /* for directory operations */ 430 431 struct f2fs_filename { 432 /* 433 * The filename the user specified. This is NULL for some 434 * filesystem-internal operations, e.g. converting an inline directory 435 * to a non-inline one, or roll-forward recovering an encrypted dentry. 436 */ 437 const struct qstr *usr_fname; 438 439 /* 440 * The on-disk filename. For encrypted directories, this is encrypted. 441 * This may be NULL for lookups in an encrypted dir without the key. 442 */ 443 struct fscrypt_str disk_name; 444 445 /* The dirhash of this filename */ 446 f2fs_hash_t hash; 447 448 #ifdef CONFIG_FS_ENCRYPTION 449 /* 450 * For lookups in encrypted directories: either the buffer backing 451 * disk_name, or a buffer that holds the decoded no-key name. 452 */ 453 struct fscrypt_str crypto_buf; 454 #endif 455 #ifdef CONFIG_UNICODE 456 /* 457 * For casefolded directories: the casefolded name, but it's left NULL 458 * if the original name is not valid Unicode, if the directory is both 459 * casefolded and encrypted and its encryption key is unavailable, or if 460 * the filesystem is doing an internal operation where usr_fname is also 461 * NULL. In all these cases we fall back to treating the name as an 462 * opaque byte sequence. 463 */ 464 struct fscrypt_str cf_name; 465 #endif 466 }; 467 468 struct f2fs_dentry_ptr { 469 struct inode *inode; 470 void *bitmap; 471 struct f2fs_dir_entry *dentry; 472 __u8 (*filename)[F2FS_SLOT_LEN]; 473 int max; 474 int nr_bitmap; 475 }; 476 477 static inline void make_dentry_ptr_block(struct inode *inode, 478 struct f2fs_dentry_ptr *d, struct f2fs_dentry_block *t) 479 { 480 d->inode = inode; 481 d->max = NR_DENTRY_IN_BLOCK; 482 d->nr_bitmap = SIZE_OF_DENTRY_BITMAP; 483 d->bitmap = t->dentry_bitmap; 484 d->dentry = t->dentry; 485 d->filename = t->filename; 486 } 487 488 static inline void make_dentry_ptr_inline(struct inode *inode, 489 struct f2fs_dentry_ptr *d, void *t) 490 { 491 int entry_cnt = NR_INLINE_DENTRY(inode); 492 int bitmap_size = INLINE_DENTRY_BITMAP_SIZE(inode); 493 int reserved_size = INLINE_RESERVED_SIZE(inode); 494 495 d->inode = inode; 496 d->max = entry_cnt; 497 d->nr_bitmap = bitmap_size; 498 d->bitmap = t; 499 d->dentry = t + bitmap_size + reserved_size; 500 d->filename = t + bitmap_size + reserved_size + 501 SIZE_OF_DIR_ENTRY * entry_cnt; 502 } 503 504 /* 505 * XATTR_NODE_OFFSET stores xattrs to one node block per file keeping -1 506 * as its node offset to distinguish from index node blocks. 507 * But some bits are used to mark the node block. 508 */ 509 #define XATTR_NODE_OFFSET ((((unsigned int)-1) << OFFSET_BIT_SHIFT) \ 510 >> OFFSET_BIT_SHIFT) 511 enum { 512 ALLOC_NODE, /* allocate a new node page if needed */ 513 LOOKUP_NODE, /* look up a node without readahead */ 514 LOOKUP_NODE_RA, /* 515 * look up a node with readahead called 516 * by get_data_block. 517 */ 518 }; 519 520 #define DEFAULT_RETRY_IO_COUNT 8 /* maximum retry read IO count */ 521 522 /* congestion wait timeout value, default: 20ms */ 523 #define DEFAULT_IO_TIMEOUT (msecs_to_jiffies(20)) 524 525 /* maximum retry quota flush count */ 526 #define DEFAULT_RETRY_QUOTA_FLUSH_COUNT 8 527 528 #define F2FS_LINK_MAX 0xffffffff /* maximum link count per file */ 529 530 #define MAX_DIR_RA_PAGES 4 /* maximum ra pages of dir */ 531 532 /* for in-memory extent cache entry */ 533 #define F2FS_MIN_EXTENT_LEN 64 /* minimum extent length */ 534 535 /* number of extent info in extent cache we try to shrink */ 536 #define EXTENT_CACHE_SHRINK_NUMBER 128 537 538 struct rb_entry { 539 struct rb_node rb_node; /* rb node located in rb-tree */ 540 union { 541 struct { 542 unsigned int ofs; /* start offset of the entry */ 543 unsigned int len; /* length of the entry */ 544 }; 545 unsigned long long key; /* 64-bits key */ 546 } __packed; 547 }; 548 549 struct extent_info { 550 unsigned int fofs; /* start offset in a file */ 551 unsigned int len; /* length of the extent */ 552 u32 blk; /* start block address of the extent */ 553 }; 554 555 struct extent_node { 556 struct rb_node rb_node; /* rb node located in rb-tree */ 557 struct extent_info ei; /* extent info */ 558 struct list_head list; /* node in global extent list of sbi */ 559 struct extent_tree *et; /* extent tree pointer */ 560 }; 561 562 struct extent_tree { 563 nid_t ino; /* inode number */ 564 struct rb_root_cached root; /* root of extent info rb-tree */ 565 struct extent_node *cached_en; /* recently accessed extent node */ 566 struct extent_info largest; /* largested extent info */ 567 struct list_head list; /* to be used by sbi->zombie_list */ 568 rwlock_t lock; /* protect extent info rb-tree */ 569 atomic_t node_cnt; /* # of extent node in rb-tree*/ 570 bool largest_updated; /* largest extent updated */ 571 }; 572 573 /* 574 * This structure is taken from ext4_map_blocks. 575 * 576 * Note that, however, f2fs uses NEW and MAPPED flags for f2fs_map_blocks(). 577 */ 578 #define F2FS_MAP_NEW (1 << BH_New) 579 #define F2FS_MAP_MAPPED (1 << BH_Mapped) 580 #define F2FS_MAP_UNWRITTEN (1 << BH_Unwritten) 581 #define F2FS_MAP_FLAGS (F2FS_MAP_NEW | F2FS_MAP_MAPPED |\ 582 F2FS_MAP_UNWRITTEN) 583 584 struct f2fs_map_blocks { 585 block_t m_pblk; 586 block_t m_lblk; 587 unsigned int m_len; 588 unsigned int m_flags; 589 pgoff_t *m_next_pgofs; /* point next possible non-hole pgofs */ 590 pgoff_t *m_next_extent; /* point to next possible extent */ 591 int m_seg_type; 592 bool m_may_create; /* indicate it is from write path */ 593 }; 594 595 /* for flag in get_data_block */ 596 enum { 597 F2FS_GET_BLOCK_DEFAULT, 598 F2FS_GET_BLOCK_FIEMAP, 599 F2FS_GET_BLOCK_BMAP, 600 F2FS_GET_BLOCK_DIO, 601 F2FS_GET_BLOCK_PRE_DIO, 602 F2FS_GET_BLOCK_PRE_AIO, 603 F2FS_GET_BLOCK_PRECACHE, 604 }; 605 606 /* 607 * i_advise uses FADVISE_XXX_BIT. We can add additional hints later. 608 */ 609 #define FADVISE_COLD_BIT 0x01 610 #define FADVISE_LOST_PINO_BIT 0x02 611 #define FADVISE_ENCRYPT_BIT 0x04 612 #define FADVISE_ENC_NAME_BIT 0x08 613 #define FADVISE_KEEP_SIZE_BIT 0x10 614 #define FADVISE_HOT_BIT 0x20 615 #define FADVISE_VERITY_BIT 0x40 616 617 #define FADVISE_MODIFIABLE_BITS (FADVISE_COLD_BIT | FADVISE_HOT_BIT) 618 619 #define file_is_cold(inode) is_file(inode, FADVISE_COLD_BIT) 620 #define file_wrong_pino(inode) is_file(inode, FADVISE_LOST_PINO_BIT) 621 #define file_set_cold(inode) set_file(inode, FADVISE_COLD_BIT) 622 #define file_lost_pino(inode) set_file(inode, FADVISE_LOST_PINO_BIT) 623 #define file_clear_cold(inode) clear_file(inode, FADVISE_COLD_BIT) 624 #define file_got_pino(inode) clear_file(inode, FADVISE_LOST_PINO_BIT) 625 #define file_is_encrypt(inode) is_file(inode, FADVISE_ENCRYPT_BIT) 626 #define file_set_encrypt(inode) set_file(inode, FADVISE_ENCRYPT_BIT) 627 #define file_clear_encrypt(inode) clear_file(inode, FADVISE_ENCRYPT_BIT) 628 #define file_enc_name(inode) is_file(inode, FADVISE_ENC_NAME_BIT) 629 #define file_set_enc_name(inode) set_file(inode, FADVISE_ENC_NAME_BIT) 630 #define file_keep_isize(inode) is_file(inode, FADVISE_KEEP_SIZE_BIT) 631 #define file_set_keep_isize(inode) set_file(inode, FADVISE_KEEP_SIZE_BIT) 632 #define file_is_hot(inode) is_file(inode, FADVISE_HOT_BIT) 633 #define file_set_hot(inode) set_file(inode, FADVISE_HOT_BIT) 634 #define file_clear_hot(inode) clear_file(inode, FADVISE_HOT_BIT) 635 #define file_is_verity(inode) is_file(inode, FADVISE_VERITY_BIT) 636 #define file_set_verity(inode) set_file(inode, FADVISE_VERITY_BIT) 637 638 #define DEF_DIR_LEVEL 0 639 640 enum { 641 GC_FAILURE_PIN, 642 GC_FAILURE_ATOMIC, 643 MAX_GC_FAILURE 644 }; 645 646 /* used for f2fs_inode_info->flags */ 647 enum { 648 FI_NEW_INODE, /* indicate newly allocated inode */ 649 FI_DIRTY_INODE, /* indicate inode is dirty or not */ 650 FI_AUTO_RECOVER, /* indicate inode is recoverable */ 651 FI_DIRTY_DIR, /* indicate directory has dirty pages */ 652 FI_INC_LINK, /* need to increment i_nlink */ 653 FI_ACL_MODE, /* indicate acl mode */ 654 FI_NO_ALLOC, /* should not allocate any blocks */ 655 FI_FREE_NID, /* free allocated nide */ 656 FI_NO_EXTENT, /* not to use the extent cache */ 657 FI_INLINE_XATTR, /* used for inline xattr */ 658 FI_INLINE_DATA, /* used for inline data*/ 659 FI_INLINE_DENTRY, /* used for inline dentry */ 660 FI_APPEND_WRITE, /* inode has appended data */ 661 FI_UPDATE_WRITE, /* inode has in-place-update data */ 662 FI_NEED_IPU, /* used for ipu per file */ 663 FI_ATOMIC_FILE, /* indicate atomic file */ 664 FI_ATOMIC_COMMIT, /* indicate the state of atomical committing */ 665 FI_VOLATILE_FILE, /* indicate volatile file */ 666 FI_FIRST_BLOCK_WRITTEN, /* indicate #0 data block was written */ 667 FI_DROP_CACHE, /* drop dirty page cache */ 668 FI_DATA_EXIST, /* indicate data exists */ 669 FI_INLINE_DOTS, /* indicate inline dot dentries */ 670 FI_DO_DEFRAG, /* indicate defragment is running */ 671 FI_DIRTY_FILE, /* indicate regular/symlink has dirty pages */ 672 FI_NO_PREALLOC, /* indicate skipped preallocated blocks */ 673 FI_HOT_DATA, /* indicate file is hot */ 674 FI_EXTRA_ATTR, /* indicate file has extra attribute */ 675 FI_PROJ_INHERIT, /* indicate file inherits projectid */ 676 FI_PIN_FILE, /* indicate file should not be gced */ 677 FI_ATOMIC_REVOKE_REQUEST, /* request to drop atomic data */ 678 FI_VERITY_IN_PROGRESS, /* building fs-verity Merkle tree */ 679 FI_COMPRESSED_FILE, /* indicate file's data can be compressed */ 680 FI_COMPRESS_CORRUPT, /* indicate compressed cluster is corrupted */ 681 FI_MMAP_FILE, /* indicate file was mmapped */ 682 FI_ENABLE_COMPRESS, /* enable compression in "user" compression mode */ 683 FI_MAX, /* max flag, never be used */ 684 }; 685 686 struct f2fs_inode_info { 687 struct inode vfs_inode; /* serve a vfs inode */ 688 unsigned long i_flags; /* keep an inode flags for ioctl */ 689 unsigned char i_advise; /* use to give file attribute hints */ 690 unsigned char i_dir_level; /* use for dentry level for large dir */ 691 unsigned int i_current_depth; /* only for directory depth */ 692 /* for gc failure statistic */ 693 unsigned int i_gc_failures[MAX_GC_FAILURE]; 694 unsigned int i_pino; /* parent inode number */ 695 umode_t i_acl_mode; /* keep file acl mode temporarily */ 696 697 /* Use below internally in f2fs*/ 698 unsigned long flags[BITS_TO_LONGS(FI_MAX)]; /* use to pass per-file flags */ 699 struct rw_semaphore i_sem; /* protect fi info */ 700 atomic_t dirty_pages; /* # of dirty pages */ 701 f2fs_hash_t chash; /* hash value of given file name */ 702 unsigned int clevel; /* maximum level of given file name */ 703 struct task_struct *task; /* lookup and create consistency */ 704 struct task_struct *cp_task; /* separate cp/wb IO stats*/ 705 nid_t i_xattr_nid; /* node id that contains xattrs */ 706 loff_t last_disk_size; /* lastly written file size */ 707 spinlock_t i_size_lock; /* protect last_disk_size */ 708 709 #ifdef CONFIG_QUOTA 710 struct dquot *i_dquot[MAXQUOTAS]; 711 712 /* quota space reservation, managed internally by quota code */ 713 qsize_t i_reserved_quota; 714 #endif 715 struct list_head dirty_list; /* dirty list for dirs and files */ 716 struct list_head gdirty_list; /* linked in global dirty list */ 717 struct list_head inmem_ilist; /* list for inmem inodes */ 718 struct list_head inmem_pages; /* inmemory pages managed by f2fs */ 719 struct task_struct *inmem_task; /* store inmemory task */ 720 struct mutex inmem_lock; /* lock for inmemory pages */ 721 pgoff_t ra_offset; /* ongoing readahead offset */ 722 struct extent_tree *extent_tree; /* cached extent_tree entry */ 723 724 /* avoid racing between foreground op and gc */ 725 struct rw_semaphore i_gc_rwsem[2]; 726 struct rw_semaphore i_mmap_sem; 727 struct rw_semaphore i_xattr_sem; /* avoid racing between reading and changing EAs */ 728 729 int i_extra_isize; /* size of extra space located in i_addr */ 730 kprojid_t i_projid; /* id for project quota */ 731 int i_inline_xattr_size; /* inline xattr size */ 732 struct timespec64 i_crtime; /* inode creation time */ 733 struct timespec64 i_disk_time[4];/* inode disk times */ 734 735 /* for file compress */ 736 atomic_t i_compr_blocks; /* # of compressed blocks */ 737 unsigned char i_compress_algorithm; /* algorithm type */ 738 unsigned char i_log_cluster_size; /* log of cluster size */ 739 unsigned char i_compress_level; /* compress level (lz4hc,zstd) */ 740 unsigned short i_compress_flag; /* compress flag */ 741 unsigned int i_cluster_size; /* cluster size */ 742 }; 743 744 static inline void get_extent_info(struct extent_info *ext, 745 struct f2fs_extent *i_ext) 746 { 747 ext->fofs = le32_to_cpu(i_ext->fofs); 748 ext->blk = le32_to_cpu(i_ext->blk); 749 ext->len = le32_to_cpu(i_ext->len); 750 } 751 752 static inline void set_raw_extent(struct extent_info *ext, 753 struct f2fs_extent *i_ext) 754 { 755 i_ext->fofs = cpu_to_le32(ext->fofs); 756 i_ext->blk = cpu_to_le32(ext->blk); 757 i_ext->len = cpu_to_le32(ext->len); 758 } 759 760 static inline void set_extent_info(struct extent_info *ei, unsigned int fofs, 761 u32 blk, unsigned int len) 762 { 763 ei->fofs = fofs; 764 ei->blk = blk; 765 ei->len = len; 766 } 767 768 static inline bool __is_discard_mergeable(struct discard_info *back, 769 struct discard_info *front, unsigned int max_len) 770 { 771 return (back->lstart + back->len == front->lstart) && 772 (back->len + front->len <= max_len); 773 } 774 775 static inline bool __is_discard_back_mergeable(struct discard_info *cur, 776 struct discard_info *back, unsigned int max_len) 777 { 778 return __is_discard_mergeable(back, cur, max_len); 779 } 780 781 static inline bool __is_discard_front_mergeable(struct discard_info *cur, 782 struct discard_info *front, unsigned int max_len) 783 { 784 return __is_discard_mergeable(cur, front, max_len); 785 } 786 787 static inline bool __is_extent_mergeable(struct extent_info *back, 788 struct extent_info *front) 789 { 790 return (back->fofs + back->len == front->fofs && 791 back->blk + back->len == front->blk); 792 } 793 794 static inline bool __is_back_mergeable(struct extent_info *cur, 795 struct extent_info *back) 796 { 797 return __is_extent_mergeable(back, cur); 798 } 799 800 static inline bool __is_front_mergeable(struct extent_info *cur, 801 struct extent_info *front) 802 { 803 return __is_extent_mergeable(cur, front); 804 } 805 806 extern void f2fs_mark_inode_dirty_sync(struct inode *inode, bool sync); 807 static inline void __try_update_largest_extent(struct extent_tree *et, 808 struct extent_node *en) 809 { 810 if (en->ei.len > et->largest.len) { 811 et->largest = en->ei; 812 et->largest_updated = true; 813 } 814 } 815 816 /* 817 * For free nid management 818 */ 819 enum nid_state { 820 FREE_NID, /* newly added to free nid list */ 821 PREALLOC_NID, /* it is preallocated */ 822 MAX_NID_STATE, 823 }; 824 825 enum nat_state { 826 TOTAL_NAT, 827 DIRTY_NAT, 828 RECLAIMABLE_NAT, 829 MAX_NAT_STATE, 830 }; 831 832 struct f2fs_nm_info { 833 block_t nat_blkaddr; /* base disk address of NAT */ 834 nid_t max_nid; /* maximum possible node ids */ 835 nid_t available_nids; /* # of available node ids */ 836 nid_t next_scan_nid; /* the next nid to be scanned */ 837 unsigned int ram_thresh; /* control the memory footprint */ 838 unsigned int ra_nid_pages; /* # of nid pages to be readaheaded */ 839 unsigned int dirty_nats_ratio; /* control dirty nats ratio threshold */ 840 841 /* NAT cache management */ 842 struct radix_tree_root nat_root;/* root of the nat entry cache */ 843 struct radix_tree_root nat_set_root;/* root of the nat set cache */ 844 struct rw_semaphore nat_tree_lock; /* protect nat_tree_lock */ 845 struct list_head nat_entries; /* cached nat entry list (clean) */ 846 spinlock_t nat_list_lock; /* protect clean nat entry list */ 847 unsigned int nat_cnt[MAX_NAT_STATE]; /* the # of cached nat entries */ 848 unsigned int nat_blocks; /* # of nat blocks */ 849 850 /* free node ids management */ 851 struct radix_tree_root free_nid_root;/* root of the free_nid cache */ 852 struct list_head free_nid_list; /* list for free nids excluding preallocated nids */ 853 unsigned int nid_cnt[MAX_NID_STATE]; /* the number of free node id */ 854 spinlock_t nid_list_lock; /* protect nid lists ops */ 855 struct mutex build_lock; /* lock for build free nids */ 856 unsigned char **free_nid_bitmap; 857 unsigned char *nat_block_bitmap; 858 unsigned short *free_nid_count; /* free nid count of NAT block */ 859 860 /* for checkpoint */ 861 char *nat_bitmap; /* NAT bitmap pointer */ 862 863 unsigned int nat_bits_blocks; /* # of nat bits blocks */ 864 unsigned char *nat_bits; /* NAT bits blocks */ 865 unsigned char *full_nat_bits; /* full NAT pages */ 866 unsigned char *empty_nat_bits; /* empty NAT pages */ 867 #ifdef CONFIG_F2FS_CHECK_FS 868 char *nat_bitmap_mir; /* NAT bitmap mirror */ 869 #endif 870 int bitmap_size; /* bitmap size */ 871 }; 872 873 /* 874 * this structure is used as one of function parameters. 875 * all the information are dedicated to a given direct node block determined 876 * by the data offset in a file. 877 */ 878 struct dnode_of_data { 879 struct inode *inode; /* vfs inode pointer */ 880 struct page *inode_page; /* its inode page, NULL is possible */ 881 struct page *node_page; /* cached direct node page */ 882 nid_t nid; /* node id of the direct node block */ 883 unsigned int ofs_in_node; /* data offset in the node page */ 884 bool inode_page_locked; /* inode page is locked or not */ 885 bool node_changed; /* is node block changed */ 886 char cur_level; /* level of hole node page */ 887 char max_level; /* level of current page located */ 888 block_t data_blkaddr; /* block address of the node block */ 889 }; 890 891 static inline void set_new_dnode(struct dnode_of_data *dn, struct inode *inode, 892 struct page *ipage, struct page *npage, nid_t nid) 893 { 894 memset(dn, 0, sizeof(*dn)); 895 dn->inode = inode; 896 dn->inode_page = ipage; 897 dn->node_page = npage; 898 dn->nid = nid; 899 } 900 901 /* 902 * For SIT manager 903 * 904 * By default, there are 6 active log areas across the whole main area. 905 * When considering hot and cold data separation to reduce cleaning overhead, 906 * we split 3 for data logs and 3 for node logs as hot, warm, and cold types, 907 * respectively. 908 * In the current design, you should not change the numbers intentionally. 909 * Instead, as a mount option such as active_logs=x, you can use 2, 4, and 6 910 * logs individually according to the underlying devices. (default: 6) 911 * Just in case, on-disk layout covers maximum 16 logs that consist of 8 for 912 * data and 8 for node logs. 913 */ 914 #define NR_CURSEG_DATA_TYPE (3) 915 #define NR_CURSEG_NODE_TYPE (3) 916 #define NR_CURSEG_INMEM_TYPE (2) 917 #define NR_CURSEG_PERSIST_TYPE (NR_CURSEG_DATA_TYPE + NR_CURSEG_NODE_TYPE) 918 #define NR_CURSEG_TYPE (NR_CURSEG_INMEM_TYPE + NR_CURSEG_PERSIST_TYPE) 919 920 enum { 921 CURSEG_HOT_DATA = 0, /* directory entry blocks */ 922 CURSEG_WARM_DATA, /* data blocks */ 923 CURSEG_COLD_DATA, /* multimedia or GCed data blocks */ 924 CURSEG_HOT_NODE, /* direct node blocks of directory files */ 925 CURSEG_WARM_NODE, /* direct node blocks of normal files */ 926 CURSEG_COLD_NODE, /* indirect node blocks */ 927 NR_PERSISTENT_LOG, /* number of persistent log */ 928 CURSEG_COLD_DATA_PINNED = NR_PERSISTENT_LOG, 929 /* pinned file that needs consecutive block address */ 930 CURSEG_ALL_DATA_ATGC, /* SSR alloctor in hot/warm/cold data area */ 931 NO_CHECK_TYPE, /* number of persistent & inmem log */ 932 }; 933 934 struct flush_cmd { 935 struct completion wait; 936 struct llist_node llnode; 937 nid_t ino; 938 int ret; 939 }; 940 941 struct flush_cmd_control { 942 struct task_struct *f2fs_issue_flush; /* flush thread */ 943 wait_queue_head_t flush_wait_queue; /* waiting queue for wake-up */ 944 atomic_t issued_flush; /* # of issued flushes */ 945 atomic_t queued_flush; /* # of queued flushes */ 946 struct llist_head issue_list; /* list for command issue */ 947 struct llist_node *dispatch_list; /* list for command dispatch */ 948 }; 949 950 struct f2fs_sm_info { 951 struct sit_info *sit_info; /* whole segment information */ 952 struct free_segmap_info *free_info; /* free segment information */ 953 struct dirty_seglist_info *dirty_info; /* dirty segment information */ 954 struct curseg_info *curseg_array; /* active segment information */ 955 956 struct rw_semaphore curseg_lock; /* for preventing curseg change */ 957 958 block_t seg0_blkaddr; /* block address of 0'th segment */ 959 block_t main_blkaddr; /* start block address of main area */ 960 block_t ssa_blkaddr; /* start block address of SSA area */ 961 962 unsigned int segment_count; /* total # of segments */ 963 unsigned int main_segments; /* # of segments in main area */ 964 unsigned int reserved_segments; /* # of reserved segments */ 965 unsigned int ovp_segments; /* # of overprovision segments */ 966 967 /* a threshold to reclaim prefree segments */ 968 unsigned int rec_prefree_segments; 969 970 /* for batched trimming */ 971 unsigned int trim_sections; /* # of sections to trim */ 972 973 struct list_head sit_entry_set; /* sit entry set list */ 974 975 unsigned int ipu_policy; /* in-place-update policy */ 976 unsigned int min_ipu_util; /* in-place-update threshold */ 977 unsigned int min_fsync_blocks; /* threshold for fsync */ 978 unsigned int min_seq_blocks; /* threshold for sequential blocks */ 979 unsigned int min_hot_blocks; /* threshold for hot block allocation */ 980 unsigned int min_ssr_sections; /* threshold to trigger SSR allocation */ 981 982 /* for flush command control */ 983 struct flush_cmd_control *fcc_info; 984 985 /* for discard command control */ 986 struct discard_cmd_control *dcc_info; 987 }; 988 989 /* 990 * For superblock 991 */ 992 /* 993 * COUNT_TYPE for monitoring 994 * 995 * f2fs monitors the number of several block types such as on-writeback, 996 * dirty dentry blocks, dirty node blocks, and dirty meta blocks. 997 */ 998 #define WB_DATA_TYPE(p) (__is_cp_guaranteed(p) ? F2FS_WB_CP_DATA : F2FS_WB_DATA) 999 enum count_type { 1000 F2FS_DIRTY_DENTS, 1001 F2FS_DIRTY_DATA, 1002 F2FS_DIRTY_QDATA, 1003 F2FS_DIRTY_NODES, 1004 F2FS_DIRTY_META, 1005 F2FS_INMEM_PAGES, 1006 F2FS_DIRTY_IMETA, 1007 F2FS_WB_CP_DATA, 1008 F2FS_WB_DATA, 1009 F2FS_RD_DATA, 1010 F2FS_RD_NODE, 1011 F2FS_RD_META, 1012 F2FS_DIO_WRITE, 1013 F2FS_DIO_READ, 1014 NR_COUNT_TYPE, 1015 }; 1016 1017 /* 1018 * The below are the page types of bios used in submit_bio(). 1019 * The available types are: 1020 * DATA User data pages. It operates as async mode. 1021 * NODE Node pages. It operates as async mode. 1022 * META FS metadata pages such as SIT, NAT, CP. 1023 * NR_PAGE_TYPE The number of page types. 1024 * META_FLUSH Make sure the previous pages are written 1025 * with waiting the bio's completion 1026 * ... Only can be used with META. 1027 */ 1028 #define PAGE_TYPE_OF_BIO(type) ((type) > META ? META : (type)) 1029 enum page_type { 1030 DATA, 1031 NODE, 1032 META, 1033 NR_PAGE_TYPE, 1034 META_FLUSH, 1035 INMEM, /* the below types are used by tracepoints only. */ 1036 INMEM_DROP, 1037 INMEM_INVALIDATE, 1038 INMEM_REVOKE, 1039 IPU, 1040 OPU, 1041 }; 1042 1043 enum temp_type { 1044 HOT = 0, /* must be zero for meta bio */ 1045 WARM, 1046 COLD, 1047 NR_TEMP_TYPE, 1048 }; 1049 1050 enum need_lock_type { 1051 LOCK_REQ = 0, 1052 LOCK_DONE, 1053 LOCK_RETRY, 1054 }; 1055 1056 enum cp_reason_type { 1057 CP_NO_NEEDED, 1058 CP_NON_REGULAR, 1059 CP_COMPRESSED, 1060 CP_HARDLINK, 1061 CP_SB_NEED_CP, 1062 CP_WRONG_PINO, 1063 CP_NO_SPC_ROLL, 1064 CP_NODE_NEED_CP, 1065 CP_FASTBOOT_MODE, 1066 CP_SPEC_LOG_NUM, 1067 CP_RECOVER_DIR, 1068 }; 1069 1070 enum iostat_type { 1071 /* WRITE IO */ 1072 APP_DIRECT_IO, /* app direct write IOs */ 1073 APP_BUFFERED_IO, /* app buffered write IOs */ 1074 APP_WRITE_IO, /* app write IOs */ 1075 APP_MAPPED_IO, /* app mapped IOs */ 1076 FS_DATA_IO, /* data IOs from kworker/fsync/reclaimer */ 1077 FS_NODE_IO, /* node IOs from kworker/fsync/reclaimer */ 1078 FS_META_IO, /* meta IOs from kworker/reclaimer */ 1079 FS_GC_DATA_IO, /* data IOs from forground gc */ 1080 FS_GC_NODE_IO, /* node IOs from forground gc */ 1081 FS_CP_DATA_IO, /* data IOs from checkpoint */ 1082 FS_CP_NODE_IO, /* node IOs from checkpoint */ 1083 FS_CP_META_IO, /* meta IOs from checkpoint */ 1084 1085 /* READ IO */ 1086 APP_DIRECT_READ_IO, /* app direct read IOs */ 1087 APP_BUFFERED_READ_IO, /* app buffered read IOs */ 1088 APP_READ_IO, /* app read IOs */ 1089 APP_MAPPED_READ_IO, /* app mapped read IOs */ 1090 FS_DATA_READ_IO, /* data read IOs */ 1091 FS_GDATA_READ_IO, /* data read IOs from background gc */ 1092 FS_CDATA_READ_IO, /* compressed data read IOs */ 1093 FS_NODE_READ_IO, /* node read IOs */ 1094 FS_META_READ_IO, /* meta read IOs */ 1095 1096 /* other */ 1097 FS_DISCARD, /* discard */ 1098 NR_IO_TYPE, 1099 }; 1100 1101 struct f2fs_io_info { 1102 struct f2fs_sb_info *sbi; /* f2fs_sb_info pointer */ 1103 nid_t ino; /* inode number */ 1104 enum page_type type; /* contains DATA/NODE/META/META_FLUSH */ 1105 enum temp_type temp; /* contains HOT/WARM/COLD */ 1106 int op; /* contains REQ_OP_ */ 1107 int op_flags; /* req_flag_bits */ 1108 block_t new_blkaddr; /* new block address to be written */ 1109 block_t old_blkaddr; /* old block address before Cow */ 1110 struct page *page; /* page to be written */ 1111 struct page *encrypted_page; /* encrypted page */ 1112 struct page *compressed_page; /* compressed page */ 1113 struct list_head list; /* serialize IOs */ 1114 bool submitted; /* indicate IO submission */ 1115 int need_lock; /* indicate we need to lock cp_rwsem */ 1116 bool in_list; /* indicate fio is in io_list */ 1117 bool is_por; /* indicate IO is from recovery or not */ 1118 bool retry; /* need to reallocate block address */ 1119 int compr_blocks; /* # of compressed block addresses */ 1120 bool encrypted; /* indicate file is encrypted */ 1121 enum iostat_type io_type; /* io type */ 1122 struct writeback_control *io_wbc; /* writeback control */ 1123 struct bio **bio; /* bio for ipu */ 1124 sector_t *last_block; /* last block number in bio */ 1125 unsigned char version; /* version of the node */ 1126 }; 1127 1128 struct bio_entry { 1129 struct bio *bio; 1130 struct list_head list; 1131 }; 1132 1133 #define is_read_io(rw) ((rw) == READ) 1134 struct f2fs_bio_info { 1135 struct f2fs_sb_info *sbi; /* f2fs superblock */ 1136 struct bio *bio; /* bios to merge */ 1137 sector_t last_block_in_bio; /* last block number */ 1138 struct f2fs_io_info fio; /* store buffered io info. */ 1139 struct rw_semaphore io_rwsem; /* blocking op for bio */ 1140 spinlock_t io_lock; /* serialize DATA/NODE IOs */ 1141 struct list_head io_list; /* track fios */ 1142 struct list_head bio_list; /* bio entry list head */ 1143 struct rw_semaphore bio_list_lock; /* lock to protect bio entry list */ 1144 }; 1145 1146 #define FDEV(i) (sbi->devs[i]) 1147 #define RDEV(i) (raw_super->devs[i]) 1148 struct f2fs_dev_info { 1149 struct block_device *bdev; 1150 char path[MAX_PATH_LEN]; 1151 unsigned int total_segments; 1152 block_t start_blk; 1153 block_t end_blk; 1154 #ifdef CONFIG_BLK_DEV_ZONED 1155 unsigned int nr_blkz; /* Total number of zones */ 1156 unsigned long *blkz_seq; /* Bitmap indicating sequential zones */ 1157 block_t *zone_capacity_blocks; /* Array of zone capacity in blks */ 1158 #endif 1159 }; 1160 1161 enum inode_type { 1162 DIR_INODE, /* for dirty dir inode */ 1163 FILE_INODE, /* for dirty regular/symlink inode */ 1164 DIRTY_META, /* for all dirtied inode metadata */ 1165 ATOMIC_FILE, /* for all atomic files */ 1166 NR_INODE_TYPE, 1167 }; 1168 1169 /* for inner inode cache management */ 1170 struct inode_management { 1171 struct radix_tree_root ino_root; /* ino entry array */ 1172 spinlock_t ino_lock; /* for ino entry lock */ 1173 struct list_head ino_list; /* inode list head */ 1174 unsigned long ino_num; /* number of entries */ 1175 }; 1176 1177 /* for GC_AT */ 1178 struct atgc_management { 1179 bool atgc_enabled; /* ATGC is enabled or not */ 1180 struct rb_root_cached root; /* root of victim rb-tree */ 1181 struct list_head victim_list; /* linked with all victim entries */ 1182 unsigned int victim_count; /* victim count in rb-tree */ 1183 unsigned int candidate_ratio; /* candidate ratio */ 1184 unsigned int max_candidate_count; /* max candidate count */ 1185 unsigned int age_weight; /* age weight, vblock_weight = 100 - age_weight */ 1186 unsigned long long age_threshold; /* age threshold */ 1187 }; 1188 1189 /* For s_flag in struct f2fs_sb_info */ 1190 enum { 1191 SBI_IS_DIRTY, /* dirty flag for checkpoint */ 1192 SBI_IS_CLOSE, /* specify unmounting */ 1193 SBI_NEED_FSCK, /* need fsck.f2fs to fix */ 1194 SBI_POR_DOING, /* recovery is doing or not */ 1195 SBI_NEED_SB_WRITE, /* need to recover superblock */ 1196 SBI_NEED_CP, /* need to checkpoint */ 1197 SBI_IS_SHUTDOWN, /* shutdown by ioctl */ 1198 SBI_IS_RECOVERED, /* recovered orphan/data */ 1199 SBI_CP_DISABLED, /* CP was disabled last mount */ 1200 SBI_CP_DISABLED_QUICK, /* CP was disabled quickly */ 1201 SBI_QUOTA_NEED_FLUSH, /* need to flush quota info in CP */ 1202 SBI_QUOTA_SKIP_FLUSH, /* skip flushing quota in current CP */ 1203 SBI_QUOTA_NEED_REPAIR, /* quota file may be corrupted */ 1204 SBI_IS_RESIZEFS, /* resizefs is in process */ 1205 }; 1206 1207 enum { 1208 CP_TIME, 1209 REQ_TIME, 1210 DISCARD_TIME, 1211 GC_TIME, 1212 DISABLE_TIME, 1213 UMOUNT_DISCARD_TIMEOUT, 1214 MAX_TIME, 1215 }; 1216 1217 enum { 1218 GC_NORMAL, 1219 GC_IDLE_CB, 1220 GC_IDLE_GREEDY, 1221 GC_IDLE_AT, 1222 GC_URGENT_HIGH, 1223 GC_URGENT_LOW, 1224 }; 1225 1226 enum { 1227 BGGC_MODE_ON, /* background gc is on */ 1228 BGGC_MODE_OFF, /* background gc is off */ 1229 BGGC_MODE_SYNC, /* 1230 * background gc is on, migrating blocks 1231 * like foreground gc 1232 */ 1233 }; 1234 1235 enum { 1236 FS_MODE_ADAPTIVE, /* use both lfs/ssr allocation */ 1237 FS_MODE_LFS, /* use lfs allocation only */ 1238 }; 1239 1240 enum { 1241 WHINT_MODE_OFF, /* not pass down write hints */ 1242 WHINT_MODE_USER, /* try to pass down hints given by users */ 1243 WHINT_MODE_FS, /* pass down hints with F2FS policy */ 1244 }; 1245 1246 enum { 1247 ALLOC_MODE_DEFAULT, /* stay default */ 1248 ALLOC_MODE_REUSE, /* reuse segments as much as possible */ 1249 }; 1250 1251 enum fsync_mode { 1252 FSYNC_MODE_POSIX, /* fsync follows posix semantics */ 1253 FSYNC_MODE_STRICT, /* fsync behaves in line with ext4 */ 1254 FSYNC_MODE_NOBARRIER, /* fsync behaves nobarrier based on posix */ 1255 }; 1256 1257 enum { 1258 COMPR_MODE_FS, /* 1259 * automatically compress compression 1260 * enabled files 1261 */ 1262 COMPR_MODE_USER, /* 1263 * automatical compression is disabled. 1264 * user can control the file compression 1265 * using ioctls 1266 */ 1267 }; 1268 1269 /* 1270 * this value is set in page as a private data which indicate that 1271 * the page is atomically written, and it is in inmem_pages list. 1272 */ 1273 #define ATOMIC_WRITTEN_PAGE ((unsigned long)-1) 1274 #define DUMMY_WRITTEN_PAGE ((unsigned long)-2) 1275 1276 #define IS_ATOMIC_WRITTEN_PAGE(page) \ 1277 (page_private(page) == ATOMIC_WRITTEN_PAGE) 1278 #define IS_DUMMY_WRITTEN_PAGE(page) \ 1279 (page_private(page) == DUMMY_WRITTEN_PAGE) 1280 1281 #ifdef CONFIG_F2FS_IO_TRACE 1282 #define IS_IO_TRACED_PAGE(page) \ 1283 (page_private(page) > 0 && \ 1284 page_private(page) < (unsigned long)PID_MAX_LIMIT) 1285 #else 1286 #define IS_IO_TRACED_PAGE(page) (0) 1287 #endif 1288 1289 /* For compression */ 1290 enum compress_algorithm_type { 1291 COMPRESS_LZO, 1292 COMPRESS_LZ4, 1293 COMPRESS_ZSTD, 1294 COMPRESS_LZORLE, 1295 COMPRESS_MAX, 1296 }; 1297 1298 enum compress_flag { 1299 COMPRESS_CHKSUM, 1300 COMPRESS_MAX_FLAG, 1301 }; 1302 1303 #define COMPRESS_DATA_RESERVED_SIZE 4 1304 struct compress_data { 1305 __le32 clen; /* compressed data size */ 1306 __le32 chksum; /* compressed data chksum */ 1307 __le32 reserved[COMPRESS_DATA_RESERVED_SIZE]; /* reserved */ 1308 u8 cdata[]; /* compressed data */ 1309 }; 1310 1311 #define COMPRESS_HEADER_SIZE (sizeof(struct compress_data)) 1312 1313 #define F2FS_COMPRESSED_PAGE_MAGIC 0xF5F2C000 1314 1315 #define COMPRESS_LEVEL_OFFSET 8 1316 1317 /* compress context */ 1318 struct compress_ctx { 1319 struct inode *inode; /* inode the context belong to */ 1320 pgoff_t cluster_idx; /* cluster index number */ 1321 unsigned int cluster_size; /* page count in cluster */ 1322 unsigned int log_cluster_size; /* log of cluster size */ 1323 struct page **rpages; /* pages store raw data in cluster */ 1324 unsigned int nr_rpages; /* total page number in rpages */ 1325 struct page **cpages; /* pages store compressed data in cluster */ 1326 unsigned int nr_cpages; /* total page number in cpages */ 1327 void *rbuf; /* virtual mapped address on rpages */ 1328 struct compress_data *cbuf; /* virtual mapped address on cpages */ 1329 size_t rlen; /* valid data length in rbuf */ 1330 size_t clen; /* valid data length in cbuf */ 1331 void *private; /* payload buffer for specified compression algorithm */ 1332 void *private2; /* extra payload buffer */ 1333 }; 1334 1335 /* compress context for write IO path */ 1336 struct compress_io_ctx { 1337 u32 magic; /* magic number to indicate page is compressed */ 1338 struct inode *inode; /* inode the context belong to */ 1339 struct page **rpages; /* pages store raw data in cluster */ 1340 unsigned int nr_rpages; /* total page number in rpages */ 1341 atomic_t pending_pages; /* in-flight compressed page count */ 1342 }; 1343 1344 /* decompress io context for read IO path */ 1345 struct decompress_io_ctx { 1346 u32 magic; /* magic number to indicate page is compressed */ 1347 struct inode *inode; /* inode the context belong to */ 1348 pgoff_t cluster_idx; /* cluster index number */ 1349 unsigned int cluster_size; /* page count in cluster */ 1350 unsigned int log_cluster_size; /* log of cluster size */ 1351 struct page **rpages; /* pages store raw data in cluster */ 1352 unsigned int nr_rpages; /* total page number in rpages */ 1353 struct page **cpages; /* pages store compressed data in cluster */ 1354 unsigned int nr_cpages; /* total page number in cpages */ 1355 struct page **tpages; /* temp pages to pad holes in cluster */ 1356 void *rbuf; /* virtual mapped address on rpages */ 1357 struct compress_data *cbuf; /* virtual mapped address on cpages */ 1358 size_t rlen; /* valid data length in rbuf */ 1359 size_t clen; /* valid data length in cbuf */ 1360 atomic_t pending_pages; /* in-flight compressed page count */ 1361 atomic_t verity_pages; /* in-flight page count for verity */ 1362 bool failed; /* indicate IO error during decompression */ 1363 void *private; /* payload buffer for specified decompression algorithm */ 1364 void *private2; /* extra payload buffer */ 1365 }; 1366 1367 #define NULL_CLUSTER ((unsigned int)(~0)) 1368 #define MIN_COMPRESS_LOG_SIZE 2 1369 #define MAX_COMPRESS_LOG_SIZE 8 1370 #define MAX_COMPRESS_WINDOW_SIZE(log_size) ((PAGE_SIZE) << (log_size)) 1371 1372 struct f2fs_sb_info { 1373 struct super_block *sb; /* pointer to VFS super block */ 1374 struct proc_dir_entry *s_proc; /* proc entry */ 1375 struct f2fs_super_block *raw_super; /* raw super block pointer */ 1376 struct rw_semaphore sb_lock; /* lock for raw super block */ 1377 int valid_super_block; /* valid super block no */ 1378 unsigned long s_flag; /* flags for sbi */ 1379 struct mutex writepages; /* mutex for writepages() */ 1380 1381 #ifdef CONFIG_BLK_DEV_ZONED 1382 unsigned int blocks_per_blkz; /* F2FS blocks per zone */ 1383 unsigned int log_blocks_per_blkz; /* log2 F2FS blocks per zone */ 1384 #endif 1385 1386 /* for node-related operations */ 1387 struct f2fs_nm_info *nm_info; /* node manager */ 1388 struct inode *node_inode; /* cache node blocks */ 1389 1390 /* for segment-related operations */ 1391 struct f2fs_sm_info *sm_info; /* segment manager */ 1392 1393 /* for bio operations */ 1394 struct f2fs_bio_info *write_io[NR_PAGE_TYPE]; /* for write bios */ 1395 /* keep migration IO order for LFS mode */ 1396 struct rw_semaphore io_order_lock; 1397 mempool_t *write_io_dummy; /* Dummy pages */ 1398 1399 /* for checkpoint */ 1400 struct f2fs_checkpoint *ckpt; /* raw checkpoint pointer */ 1401 int cur_cp_pack; /* remain current cp pack */ 1402 spinlock_t cp_lock; /* for flag in ckpt */ 1403 struct inode *meta_inode; /* cache meta blocks */ 1404 struct rw_semaphore cp_global_sem; /* checkpoint procedure lock */ 1405 struct rw_semaphore cp_rwsem; /* blocking FS operations */ 1406 struct rw_semaphore node_write; /* locking node writes */ 1407 struct rw_semaphore node_change; /* locking node change */ 1408 wait_queue_head_t cp_wait; 1409 unsigned long last_time[MAX_TIME]; /* to store time in jiffies */ 1410 long interval_time[MAX_TIME]; /* to store thresholds */ 1411 1412 struct inode_management im[MAX_INO_ENTRY]; /* manage inode cache */ 1413 1414 spinlock_t fsync_node_lock; /* for node entry lock */ 1415 struct list_head fsync_node_list; /* node list head */ 1416 unsigned int fsync_seg_id; /* sequence id */ 1417 unsigned int fsync_node_num; /* number of node entries */ 1418 1419 /* for orphan inode, use 0'th array */ 1420 unsigned int max_orphans; /* max orphan inodes */ 1421 1422 /* for inode management */ 1423 struct list_head inode_list[NR_INODE_TYPE]; /* dirty inode list */ 1424 spinlock_t inode_lock[NR_INODE_TYPE]; /* for dirty inode list lock */ 1425 struct mutex flush_lock; /* for flush exclusion */ 1426 1427 /* for extent tree cache */ 1428 struct radix_tree_root extent_tree_root;/* cache extent cache entries */ 1429 struct mutex extent_tree_lock; /* locking extent radix tree */ 1430 struct list_head extent_list; /* lru list for shrinker */ 1431 spinlock_t extent_lock; /* locking extent lru list */ 1432 atomic_t total_ext_tree; /* extent tree count */ 1433 struct list_head zombie_list; /* extent zombie tree list */ 1434 atomic_t total_zombie_tree; /* extent zombie tree count */ 1435 atomic_t total_ext_node; /* extent info count */ 1436 1437 /* basic filesystem units */ 1438 unsigned int log_sectors_per_block; /* log2 sectors per block */ 1439 unsigned int log_blocksize; /* log2 block size */ 1440 unsigned int blocksize; /* block size */ 1441 unsigned int root_ino_num; /* root inode number*/ 1442 unsigned int node_ino_num; /* node inode number*/ 1443 unsigned int meta_ino_num; /* meta inode number*/ 1444 unsigned int log_blocks_per_seg; /* log2 blocks per segment */ 1445 unsigned int blocks_per_seg; /* blocks per segment */ 1446 unsigned int segs_per_sec; /* segments per section */ 1447 unsigned int secs_per_zone; /* sections per zone */ 1448 unsigned int total_sections; /* total section count */ 1449 unsigned int total_node_count; /* total node block count */ 1450 unsigned int total_valid_node_count; /* valid node block count */ 1451 loff_t max_file_blocks; /* max block index of file */ 1452 int dir_level; /* directory level */ 1453 int readdir_ra; /* readahead inode in readdir */ 1454 u64 max_io_bytes; /* max io bytes to merge IOs */ 1455 1456 block_t user_block_count; /* # of user blocks */ 1457 block_t total_valid_block_count; /* # of valid blocks */ 1458 block_t discard_blks; /* discard command candidats */ 1459 block_t last_valid_block_count; /* for recovery */ 1460 block_t reserved_blocks; /* configurable reserved blocks */ 1461 block_t current_reserved_blocks; /* current reserved blocks */ 1462 1463 /* Additional tracking for no checkpoint mode */ 1464 block_t unusable_block_count; /* # of blocks saved by last cp */ 1465 1466 unsigned int nquota_files; /* # of quota sysfile */ 1467 struct rw_semaphore quota_sem; /* blocking cp for flags */ 1468 1469 /* # of pages, see count_type */ 1470 atomic_t nr_pages[NR_COUNT_TYPE]; 1471 /* # of allocated blocks */ 1472 struct percpu_counter alloc_valid_block_count; 1473 1474 /* writeback control */ 1475 atomic_t wb_sync_req[META]; /* count # of WB_SYNC threads */ 1476 1477 /* valid inode count */ 1478 struct percpu_counter total_valid_inode_count; 1479 1480 struct f2fs_mount_info mount_opt; /* mount options */ 1481 1482 /* for cleaning operations */ 1483 struct rw_semaphore gc_lock; /* 1484 * semaphore for GC, avoid 1485 * race between GC and GC or CP 1486 */ 1487 struct f2fs_gc_kthread *gc_thread; /* GC thread */ 1488 struct atgc_management am; /* atgc management */ 1489 unsigned int cur_victim_sec; /* current victim section num */ 1490 unsigned int gc_mode; /* current GC state */ 1491 unsigned int next_victim_seg[2]; /* next segment in victim section */ 1492 1493 /* for skip statistic */ 1494 unsigned int atomic_files; /* # of opened atomic file */ 1495 unsigned long long skipped_atomic_files[2]; /* FG_GC and BG_GC */ 1496 unsigned long long skipped_gc_rwsem; /* FG_GC only */ 1497 1498 /* threshold for gc trials on pinned files */ 1499 u64 gc_pin_file_threshold; 1500 struct rw_semaphore pin_sem; 1501 1502 /* maximum # of trials to find a victim segment for SSR and GC */ 1503 unsigned int max_victim_search; 1504 /* migration granularity of garbage collection, unit: segment */ 1505 unsigned int migration_granularity; 1506 1507 /* 1508 * for stat information. 1509 * one is for the LFS mode, and the other is for the SSR mode. 1510 */ 1511 #ifdef CONFIG_F2FS_STAT_FS 1512 struct f2fs_stat_info *stat_info; /* FS status information */ 1513 atomic_t meta_count[META_MAX]; /* # of meta blocks */ 1514 unsigned int segment_count[2]; /* # of allocated segments */ 1515 unsigned int block_count[2]; /* # of allocated blocks */ 1516 atomic_t inplace_count; /* # of inplace update */ 1517 atomic64_t total_hit_ext; /* # of lookup extent cache */ 1518 atomic64_t read_hit_rbtree; /* # of hit rbtree extent node */ 1519 atomic64_t read_hit_largest; /* # of hit largest extent node */ 1520 atomic64_t read_hit_cached; /* # of hit cached extent node */ 1521 atomic_t inline_xattr; /* # of inline_xattr inodes */ 1522 atomic_t inline_inode; /* # of inline_data inodes */ 1523 atomic_t inline_dir; /* # of inline_dentry inodes */ 1524 atomic_t compr_inode; /* # of compressed inodes */ 1525 atomic64_t compr_blocks; /* # of compressed blocks */ 1526 atomic_t vw_cnt; /* # of volatile writes */ 1527 atomic_t max_aw_cnt; /* max # of atomic writes */ 1528 atomic_t max_vw_cnt; /* max # of volatile writes */ 1529 unsigned int io_skip_bggc; /* skip background gc for in-flight IO */ 1530 unsigned int other_skip_bggc; /* skip background gc for other reasons */ 1531 unsigned int ndirty_inode[NR_INODE_TYPE]; /* # of dirty inodes */ 1532 #endif 1533 spinlock_t stat_lock; /* lock for stat operations */ 1534 1535 /* For app/fs IO statistics */ 1536 spinlock_t iostat_lock; 1537 unsigned long long rw_iostat[NR_IO_TYPE]; 1538 unsigned long long prev_rw_iostat[NR_IO_TYPE]; 1539 bool iostat_enable; 1540 unsigned long iostat_next_period; 1541 unsigned int iostat_period_ms; 1542 1543 /* to attach REQ_META|REQ_FUA flags */ 1544 unsigned int data_io_flag; 1545 unsigned int node_io_flag; 1546 1547 /* For sysfs suppport */ 1548 struct kobject s_kobj; /* /sys/fs/f2fs/<devname> */ 1549 struct completion s_kobj_unregister; 1550 1551 struct kobject s_stat_kobj; /* /sys/fs/f2fs/<devname>/stat */ 1552 struct completion s_stat_kobj_unregister; 1553 1554 /* For shrinker support */ 1555 struct list_head s_list; 1556 int s_ndevs; /* number of devices */ 1557 struct f2fs_dev_info *devs; /* for device list */ 1558 unsigned int dirty_device; /* for checkpoint data flush */ 1559 spinlock_t dev_lock; /* protect dirty_device */ 1560 struct mutex umount_mutex; 1561 unsigned int shrinker_run_no; 1562 1563 /* For write statistics */ 1564 u64 sectors_written_start; 1565 u64 kbytes_written; 1566 1567 /* Reference to checksum algorithm driver via cryptoapi */ 1568 struct crypto_shash *s_chksum_driver; 1569 1570 /* Precomputed FS UUID checksum for seeding other checksums */ 1571 __u32 s_chksum_seed; 1572 1573 struct workqueue_struct *post_read_wq; /* post read workqueue */ 1574 1575 struct kmem_cache *inline_xattr_slab; /* inline xattr entry */ 1576 unsigned int inline_xattr_slab_size; /* default inline xattr slab size */ 1577 1578 #ifdef CONFIG_F2FS_FS_COMPRESSION 1579 struct kmem_cache *page_array_slab; /* page array entry */ 1580 unsigned int page_array_slab_size; /* default page array slab size */ 1581 #endif 1582 }; 1583 1584 struct f2fs_private_dio { 1585 struct inode *inode; 1586 void *orig_private; 1587 bio_end_io_t *orig_end_io; 1588 bool write; 1589 }; 1590 1591 #ifdef CONFIG_F2FS_FAULT_INJECTION 1592 #define f2fs_show_injection_info(sbi, type) \ 1593 printk_ratelimited("%sF2FS-fs (%s) : inject %s in %s of %pS\n", \ 1594 KERN_INFO, sbi->sb->s_id, \ 1595 f2fs_fault_name[type], \ 1596 __func__, __builtin_return_address(0)) 1597 static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type) 1598 { 1599 struct f2fs_fault_info *ffi = &F2FS_OPTION(sbi).fault_info; 1600 1601 if (!ffi->inject_rate) 1602 return false; 1603 1604 if (!IS_FAULT_SET(ffi, type)) 1605 return false; 1606 1607 atomic_inc(&ffi->inject_ops); 1608 if (atomic_read(&ffi->inject_ops) >= ffi->inject_rate) { 1609 atomic_set(&ffi->inject_ops, 0); 1610 return true; 1611 } 1612 return false; 1613 } 1614 #else 1615 #define f2fs_show_injection_info(sbi, type) do { } while (0) 1616 static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type) 1617 { 1618 return false; 1619 } 1620 #endif 1621 1622 /* 1623 * Test if the mounted volume is a multi-device volume. 1624 * - For a single regular disk volume, sbi->s_ndevs is 0. 1625 * - For a single zoned disk volume, sbi->s_ndevs is 1. 1626 * - For a multi-device volume, sbi->s_ndevs is always 2 or more. 1627 */ 1628 static inline bool f2fs_is_multi_device(struct f2fs_sb_info *sbi) 1629 { 1630 return sbi->s_ndevs > 1; 1631 } 1632 1633 static inline void f2fs_update_time(struct f2fs_sb_info *sbi, int type) 1634 { 1635 unsigned long now = jiffies; 1636 1637 sbi->last_time[type] = now; 1638 1639 /* DISCARD_TIME and GC_TIME are based on REQ_TIME */ 1640 if (type == REQ_TIME) { 1641 sbi->last_time[DISCARD_TIME] = now; 1642 sbi->last_time[GC_TIME] = now; 1643 } 1644 } 1645 1646 static inline bool f2fs_time_over(struct f2fs_sb_info *sbi, int type) 1647 { 1648 unsigned long interval = sbi->interval_time[type] * HZ; 1649 1650 return time_after(jiffies, sbi->last_time[type] + interval); 1651 } 1652 1653 static inline unsigned int f2fs_time_to_wait(struct f2fs_sb_info *sbi, 1654 int type) 1655 { 1656 unsigned long interval = sbi->interval_time[type] * HZ; 1657 unsigned int wait_ms = 0; 1658 long delta; 1659 1660 delta = (sbi->last_time[type] + interval) - jiffies; 1661 if (delta > 0) 1662 wait_ms = jiffies_to_msecs(delta); 1663 1664 return wait_ms; 1665 } 1666 1667 /* 1668 * Inline functions 1669 */ 1670 static inline u32 __f2fs_crc32(struct f2fs_sb_info *sbi, u32 crc, 1671 const void *address, unsigned int length) 1672 { 1673 struct { 1674 struct shash_desc shash; 1675 char ctx[4]; 1676 } desc; 1677 int err; 1678 1679 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver) != sizeof(desc.ctx)); 1680 1681 desc.shash.tfm = sbi->s_chksum_driver; 1682 *(u32 *)desc.ctx = crc; 1683 1684 err = crypto_shash_update(&desc.shash, address, length); 1685 BUG_ON(err); 1686 1687 return *(u32 *)desc.ctx; 1688 } 1689 1690 static inline u32 f2fs_crc32(struct f2fs_sb_info *sbi, const void *address, 1691 unsigned int length) 1692 { 1693 return __f2fs_crc32(sbi, F2FS_SUPER_MAGIC, address, length); 1694 } 1695 1696 static inline bool f2fs_crc_valid(struct f2fs_sb_info *sbi, __u32 blk_crc, 1697 void *buf, size_t buf_size) 1698 { 1699 return f2fs_crc32(sbi, buf, buf_size) == blk_crc; 1700 } 1701 1702 static inline u32 f2fs_chksum(struct f2fs_sb_info *sbi, u32 crc, 1703 const void *address, unsigned int length) 1704 { 1705 return __f2fs_crc32(sbi, crc, address, length); 1706 } 1707 1708 static inline struct f2fs_inode_info *F2FS_I(struct inode *inode) 1709 { 1710 return container_of(inode, struct f2fs_inode_info, vfs_inode); 1711 } 1712 1713 static inline struct f2fs_sb_info *F2FS_SB(struct super_block *sb) 1714 { 1715 return sb->s_fs_info; 1716 } 1717 1718 static inline struct f2fs_sb_info *F2FS_I_SB(struct inode *inode) 1719 { 1720 return F2FS_SB(inode->i_sb); 1721 } 1722 1723 static inline struct f2fs_sb_info *F2FS_M_SB(struct address_space *mapping) 1724 { 1725 return F2FS_I_SB(mapping->host); 1726 } 1727 1728 static inline struct f2fs_sb_info *F2FS_P_SB(struct page *page) 1729 { 1730 return F2FS_M_SB(page_file_mapping(page)); 1731 } 1732 1733 static inline struct f2fs_super_block *F2FS_RAW_SUPER(struct f2fs_sb_info *sbi) 1734 { 1735 return (struct f2fs_super_block *)(sbi->raw_super); 1736 } 1737 1738 static inline struct f2fs_checkpoint *F2FS_CKPT(struct f2fs_sb_info *sbi) 1739 { 1740 return (struct f2fs_checkpoint *)(sbi->ckpt); 1741 } 1742 1743 static inline struct f2fs_node *F2FS_NODE(struct page *page) 1744 { 1745 return (struct f2fs_node *)page_address(page); 1746 } 1747 1748 static inline struct f2fs_inode *F2FS_INODE(struct page *page) 1749 { 1750 return &((struct f2fs_node *)page_address(page))->i; 1751 } 1752 1753 static inline struct f2fs_nm_info *NM_I(struct f2fs_sb_info *sbi) 1754 { 1755 return (struct f2fs_nm_info *)(sbi->nm_info); 1756 } 1757 1758 static inline struct f2fs_sm_info *SM_I(struct f2fs_sb_info *sbi) 1759 { 1760 return (struct f2fs_sm_info *)(sbi->sm_info); 1761 } 1762 1763 static inline struct sit_info *SIT_I(struct f2fs_sb_info *sbi) 1764 { 1765 return (struct sit_info *)(SM_I(sbi)->sit_info); 1766 } 1767 1768 static inline struct free_segmap_info *FREE_I(struct f2fs_sb_info *sbi) 1769 { 1770 return (struct free_segmap_info *)(SM_I(sbi)->free_info); 1771 } 1772 1773 static inline struct dirty_seglist_info *DIRTY_I(struct f2fs_sb_info *sbi) 1774 { 1775 return (struct dirty_seglist_info *)(SM_I(sbi)->dirty_info); 1776 } 1777 1778 static inline struct address_space *META_MAPPING(struct f2fs_sb_info *sbi) 1779 { 1780 return sbi->meta_inode->i_mapping; 1781 } 1782 1783 static inline struct address_space *NODE_MAPPING(struct f2fs_sb_info *sbi) 1784 { 1785 return sbi->node_inode->i_mapping; 1786 } 1787 1788 static inline bool is_sbi_flag_set(struct f2fs_sb_info *sbi, unsigned int type) 1789 { 1790 return test_bit(type, &sbi->s_flag); 1791 } 1792 1793 static inline void set_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type) 1794 { 1795 set_bit(type, &sbi->s_flag); 1796 } 1797 1798 static inline void clear_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type) 1799 { 1800 clear_bit(type, &sbi->s_flag); 1801 } 1802 1803 static inline unsigned long long cur_cp_version(struct f2fs_checkpoint *cp) 1804 { 1805 return le64_to_cpu(cp->checkpoint_ver); 1806 } 1807 1808 static inline unsigned long f2fs_qf_ino(struct super_block *sb, int type) 1809 { 1810 if (type < F2FS_MAX_QUOTAS) 1811 return le32_to_cpu(F2FS_SB(sb)->raw_super->qf_ino[type]); 1812 return 0; 1813 } 1814 1815 static inline __u64 cur_cp_crc(struct f2fs_checkpoint *cp) 1816 { 1817 size_t crc_offset = le32_to_cpu(cp->checksum_offset); 1818 return le32_to_cpu(*((__le32 *)((unsigned char *)cp + crc_offset))); 1819 } 1820 1821 static inline bool __is_set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 1822 { 1823 unsigned int ckpt_flags = le32_to_cpu(cp->ckpt_flags); 1824 1825 return ckpt_flags & f; 1826 } 1827 1828 static inline bool is_set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 1829 { 1830 return __is_set_ckpt_flags(F2FS_CKPT(sbi), f); 1831 } 1832 1833 static inline void __set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 1834 { 1835 unsigned int ckpt_flags; 1836 1837 ckpt_flags = le32_to_cpu(cp->ckpt_flags); 1838 ckpt_flags |= f; 1839 cp->ckpt_flags = cpu_to_le32(ckpt_flags); 1840 } 1841 1842 static inline void set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 1843 { 1844 unsigned long flags; 1845 1846 spin_lock_irqsave(&sbi->cp_lock, flags); 1847 __set_ckpt_flags(F2FS_CKPT(sbi), f); 1848 spin_unlock_irqrestore(&sbi->cp_lock, flags); 1849 } 1850 1851 static inline void __clear_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 1852 { 1853 unsigned int ckpt_flags; 1854 1855 ckpt_flags = le32_to_cpu(cp->ckpt_flags); 1856 ckpt_flags &= (~f); 1857 cp->ckpt_flags = cpu_to_le32(ckpt_flags); 1858 } 1859 1860 static inline void clear_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 1861 { 1862 unsigned long flags; 1863 1864 spin_lock_irqsave(&sbi->cp_lock, flags); 1865 __clear_ckpt_flags(F2FS_CKPT(sbi), f); 1866 spin_unlock_irqrestore(&sbi->cp_lock, flags); 1867 } 1868 1869 static inline void disable_nat_bits(struct f2fs_sb_info *sbi, bool lock) 1870 { 1871 unsigned long flags; 1872 unsigned char *nat_bits; 1873 1874 /* 1875 * In order to re-enable nat_bits we need to call fsck.f2fs by 1876 * set_sbi_flag(sbi, SBI_NEED_FSCK). But it may give huge cost, 1877 * so let's rely on regular fsck or unclean shutdown. 1878 */ 1879 1880 if (lock) 1881 spin_lock_irqsave(&sbi->cp_lock, flags); 1882 __clear_ckpt_flags(F2FS_CKPT(sbi), CP_NAT_BITS_FLAG); 1883 nat_bits = NM_I(sbi)->nat_bits; 1884 NM_I(sbi)->nat_bits = NULL; 1885 if (lock) 1886 spin_unlock_irqrestore(&sbi->cp_lock, flags); 1887 1888 kvfree(nat_bits); 1889 } 1890 1891 static inline bool enabled_nat_bits(struct f2fs_sb_info *sbi, 1892 struct cp_control *cpc) 1893 { 1894 bool set = is_set_ckpt_flags(sbi, CP_NAT_BITS_FLAG); 1895 1896 return (cpc) ? (cpc->reason & CP_UMOUNT) && set : set; 1897 } 1898 1899 static inline void f2fs_lock_op(struct f2fs_sb_info *sbi) 1900 { 1901 down_read(&sbi->cp_rwsem); 1902 } 1903 1904 static inline int f2fs_trylock_op(struct f2fs_sb_info *sbi) 1905 { 1906 return down_read_trylock(&sbi->cp_rwsem); 1907 } 1908 1909 static inline void f2fs_unlock_op(struct f2fs_sb_info *sbi) 1910 { 1911 up_read(&sbi->cp_rwsem); 1912 } 1913 1914 static inline void f2fs_lock_all(struct f2fs_sb_info *sbi) 1915 { 1916 down_write(&sbi->cp_rwsem); 1917 } 1918 1919 static inline void f2fs_unlock_all(struct f2fs_sb_info *sbi) 1920 { 1921 up_write(&sbi->cp_rwsem); 1922 } 1923 1924 static inline int __get_cp_reason(struct f2fs_sb_info *sbi) 1925 { 1926 int reason = CP_SYNC; 1927 1928 if (test_opt(sbi, FASTBOOT)) 1929 reason = CP_FASTBOOT; 1930 if (is_sbi_flag_set(sbi, SBI_IS_CLOSE)) 1931 reason = CP_UMOUNT; 1932 return reason; 1933 } 1934 1935 static inline bool __remain_node_summaries(int reason) 1936 { 1937 return (reason & (CP_UMOUNT | CP_FASTBOOT)); 1938 } 1939 1940 static inline bool __exist_node_summaries(struct f2fs_sb_info *sbi) 1941 { 1942 return (is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG) || 1943 is_set_ckpt_flags(sbi, CP_FASTBOOT_FLAG)); 1944 } 1945 1946 /* 1947 * Check whether the inode has blocks or not 1948 */ 1949 static inline int F2FS_HAS_BLOCKS(struct inode *inode) 1950 { 1951 block_t xattr_block = F2FS_I(inode)->i_xattr_nid ? 1 : 0; 1952 1953 return (inode->i_blocks >> F2FS_LOG_SECTORS_PER_BLOCK) > xattr_block; 1954 } 1955 1956 static inline bool f2fs_has_xattr_block(unsigned int ofs) 1957 { 1958 return ofs == XATTR_NODE_OFFSET; 1959 } 1960 1961 static inline bool __allow_reserved_blocks(struct f2fs_sb_info *sbi, 1962 struct inode *inode, bool cap) 1963 { 1964 if (!inode) 1965 return true; 1966 if (!test_opt(sbi, RESERVE_ROOT)) 1967 return false; 1968 if (IS_NOQUOTA(inode)) 1969 return true; 1970 if (uid_eq(F2FS_OPTION(sbi).s_resuid, current_fsuid())) 1971 return true; 1972 if (!gid_eq(F2FS_OPTION(sbi).s_resgid, GLOBAL_ROOT_GID) && 1973 in_group_p(F2FS_OPTION(sbi).s_resgid)) 1974 return true; 1975 if (cap && capable(CAP_SYS_RESOURCE)) 1976 return true; 1977 return false; 1978 } 1979 1980 static inline void f2fs_i_blocks_write(struct inode *, block_t, bool, bool); 1981 static inline int inc_valid_block_count(struct f2fs_sb_info *sbi, 1982 struct inode *inode, blkcnt_t *count) 1983 { 1984 blkcnt_t diff = 0, release = 0; 1985 block_t avail_user_block_count; 1986 int ret; 1987 1988 ret = dquot_reserve_block(inode, *count); 1989 if (ret) 1990 return ret; 1991 1992 if (time_to_inject(sbi, FAULT_BLOCK)) { 1993 f2fs_show_injection_info(sbi, FAULT_BLOCK); 1994 release = *count; 1995 goto release_quota; 1996 } 1997 1998 /* 1999 * let's increase this in prior to actual block count change in order 2000 * for f2fs_sync_file to avoid data races when deciding checkpoint. 2001 */ 2002 percpu_counter_add(&sbi->alloc_valid_block_count, (*count)); 2003 2004 spin_lock(&sbi->stat_lock); 2005 sbi->total_valid_block_count += (block_t)(*count); 2006 avail_user_block_count = sbi->user_block_count - 2007 sbi->current_reserved_blocks; 2008 2009 if (!__allow_reserved_blocks(sbi, inode, true)) 2010 avail_user_block_count -= F2FS_OPTION(sbi).root_reserved_blocks; 2011 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) { 2012 if (avail_user_block_count > sbi->unusable_block_count) 2013 avail_user_block_count -= sbi->unusable_block_count; 2014 else 2015 avail_user_block_count = 0; 2016 } 2017 if (unlikely(sbi->total_valid_block_count > avail_user_block_count)) { 2018 diff = sbi->total_valid_block_count - avail_user_block_count; 2019 if (diff > *count) 2020 diff = *count; 2021 *count -= diff; 2022 release = diff; 2023 sbi->total_valid_block_count -= diff; 2024 if (!*count) { 2025 spin_unlock(&sbi->stat_lock); 2026 goto enospc; 2027 } 2028 } 2029 spin_unlock(&sbi->stat_lock); 2030 2031 if (unlikely(release)) { 2032 percpu_counter_sub(&sbi->alloc_valid_block_count, release); 2033 dquot_release_reservation_block(inode, release); 2034 } 2035 f2fs_i_blocks_write(inode, *count, true, true); 2036 return 0; 2037 2038 enospc: 2039 percpu_counter_sub(&sbi->alloc_valid_block_count, release); 2040 release_quota: 2041 dquot_release_reservation_block(inode, release); 2042 return -ENOSPC; 2043 } 2044 2045 __printf(2, 3) 2046 void f2fs_printk(struct f2fs_sb_info *sbi, const char *fmt, ...); 2047 2048 #define f2fs_err(sbi, fmt, ...) \ 2049 f2fs_printk(sbi, KERN_ERR fmt, ##__VA_ARGS__) 2050 #define f2fs_warn(sbi, fmt, ...) \ 2051 f2fs_printk(sbi, KERN_WARNING fmt, ##__VA_ARGS__) 2052 #define f2fs_notice(sbi, fmt, ...) \ 2053 f2fs_printk(sbi, KERN_NOTICE fmt, ##__VA_ARGS__) 2054 #define f2fs_info(sbi, fmt, ...) \ 2055 f2fs_printk(sbi, KERN_INFO fmt, ##__VA_ARGS__) 2056 #define f2fs_debug(sbi, fmt, ...) \ 2057 f2fs_printk(sbi, KERN_DEBUG fmt, ##__VA_ARGS__) 2058 2059 static inline void dec_valid_block_count(struct f2fs_sb_info *sbi, 2060 struct inode *inode, 2061 block_t count) 2062 { 2063 blkcnt_t sectors = count << F2FS_LOG_SECTORS_PER_BLOCK; 2064 2065 spin_lock(&sbi->stat_lock); 2066 f2fs_bug_on(sbi, sbi->total_valid_block_count < (block_t) count); 2067 sbi->total_valid_block_count -= (block_t)count; 2068 if (sbi->reserved_blocks && 2069 sbi->current_reserved_blocks < sbi->reserved_blocks) 2070 sbi->current_reserved_blocks = min(sbi->reserved_blocks, 2071 sbi->current_reserved_blocks + count); 2072 spin_unlock(&sbi->stat_lock); 2073 if (unlikely(inode->i_blocks < sectors)) { 2074 f2fs_warn(sbi, "Inconsistent i_blocks, ino:%lu, iblocks:%llu, sectors:%llu", 2075 inode->i_ino, 2076 (unsigned long long)inode->i_blocks, 2077 (unsigned long long)sectors); 2078 set_sbi_flag(sbi, SBI_NEED_FSCK); 2079 return; 2080 } 2081 f2fs_i_blocks_write(inode, count, false, true); 2082 } 2083 2084 static inline void inc_page_count(struct f2fs_sb_info *sbi, int count_type) 2085 { 2086 atomic_inc(&sbi->nr_pages[count_type]); 2087 2088 if (count_type == F2FS_DIRTY_DENTS || 2089 count_type == F2FS_DIRTY_NODES || 2090 count_type == F2FS_DIRTY_META || 2091 count_type == F2FS_DIRTY_QDATA || 2092 count_type == F2FS_DIRTY_IMETA) 2093 set_sbi_flag(sbi, SBI_IS_DIRTY); 2094 } 2095 2096 static inline void inode_inc_dirty_pages(struct inode *inode) 2097 { 2098 atomic_inc(&F2FS_I(inode)->dirty_pages); 2099 inc_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ? 2100 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA); 2101 if (IS_NOQUOTA(inode)) 2102 inc_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA); 2103 } 2104 2105 static inline void dec_page_count(struct f2fs_sb_info *sbi, int count_type) 2106 { 2107 atomic_dec(&sbi->nr_pages[count_type]); 2108 } 2109 2110 static inline void inode_dec_dirty_pages(struct inode *inode) 2111 { 2112 if (!S_ISDIR(inode->i_mode) && !S_ISREG(inode->i_mode) && 2113 !S_ISLNK(inode->i_mode)) 2114 return; 2115 2116 atomic_dec(&F2FS_I(inode)->dirty_pages); 2117 dec_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ? 2118 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA); 2119 if (IS_NOQUOTA(inode)) 2120 dec_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA); 2121 } 2122 2123 static inline s64 get_pages(struct f2fs_sb_info *sbi, int count_type) 2124 { 2125 return atomic_read(&sbi->nr_pages[count_type]); 2126 } 2127 2128 static inline int get_dirty_pages(struct inode *inode) 2129 { 2130 return atomic_read(&F2FS_I(inode)->dirty_pages); 2131 } 2132 2133 static inline int get_blocktype_secs(struct f2fs_sb_info *sbi, int block_type) 2134 { 2135 unsigned int pages_per_sec = sbi->segs_per_sec * sbi->blocks_per_seg; 2136 unsigned int segs = (get_pages(sbi, block_type) + pages_per_sec - 1) >> 2137 sbi->log_blocks_per_seg; 2138 2139 return segs / sbi->segs_per_sec; 2140 } 2141 2142 static inline block_t valid_user_blocks(struct f2fs_sb_info *sbi) 2143 { 2144 return sbi->total_valid_block_count; 2145 } 2146 2147 static inline block_t discard_blocks(struct f2fs_sb_info *sbi) 2148 { 2149 return sbi->discard_blks; 2150 } 2151 2152 static inline unsigned long __bitmap_size(struct f2fs_sb_info *sbi, int flag) 2153 { 2154 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi); 2155 2156 /* return NAT or SIT bitmap */ 2157 if (flag == NAT_BITMAP) 2158 return le32_to_cpu(ckpt->nat_ver_bitmap_bytesize); 2159 else if (flag == SIT_BITMAP) 2160 return le32_to_cpu(ckpt->sit_ver_bitmap_bytesize); 2161 2162 return 0; 2163 } 2164 2165 static inline block_t __cp_payload(struct f2fs_sb_info *sbi) 2166 { 2167 return le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_payload); 2168 } 2169 2170 static inline void *__bitmap_ptr(struct f2fs_sb_info *sbi, int flag) 2171 { 2172 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi); 2173 int offset; 2174 2175 if (is_set_ckpt_flags(sbi, CP_LARGE_NAT_BITMAP_FLAG)) { 2176 offset = (flag == SIT_BITMAP) ? 2177 le32_to_cpu(ckpt->nat_ver_bitmap_bytesize) : 0; 2178 /* 2179 * if large_nat_bitmap feature is enabled, leave checksum 2180 * protection for all nat/sit bitmaps. 2181 */ 2182 return &ckpt->sit_nat_version_bitmap + offset + sizeof(__le32); 2183 } 2184 2185 if (__cp_payload(sbi) > 0) { 2186 if (flag == NAT_BITMAP) 2187 return &ckpt->sit_nat_version_bitmap; 2188 else 2189 return (unsigned char *)ckpt + F2FS_BLKSIZE; 2190 } else { 2191 offset = (flag == NAT_BITMAP) ? 2192 le32_to_cpu(ckpt->sit_ver_bitmap_bytesize) : 0; 2193 return &ckpt->sit_nat_version_bitmap + offset; 2194 } 2195 } 2196 2197 static inline block_t __start_cp_addr(struct f2fs_sb_info *sbi) 2198 { 2199 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr); 2200 2201 if (sbi->cur_cp_pack == 2) 2202 start_addr += sbi->blocks_per_seg; 2203 return start_addr; 2204 } 2205 2206 static inline block_t __start_cp_next_addr(struct f2fs_sb_info *sbi) 2207 { 2208 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr); 2209 2210 if (sbi->cur_cp_pack == 1) 2211 start_addr += sbi->blocks_per_seg; 2212 return start_addr; 2213 } 2214 2215 static inline void __set_cp_next_pack(struct f2fs_sb_info *sbi) 2216 { 2217 sbi->cur_cp_pack = (sbi->cur_cp_pack == 1) ? 2 : 1; 2218 } 2219 2220 static inline block_t __start_sum_addr(struct f2fs_sb_info *sbi) 2221 { 2222 return le32_to_cpu(F2FS_CKPT(sbi)->cp_pack_start_sum); 2223 } 2224 2225 static inline int inc_valid_node_count(struct f2fs_sb_info *sbi, 2226 struct inode *inode, bool is_inode) 2227 { 2228 block_t valid_block_count; 2229 unsigned int valid_node_count, user_block_count; 2230 int err; 2231 2232 if (is_inode) { 2233 if (inode) { 2234 err = dquot_alloc_inode(inode); 2235 if (err) 2236 return err; 2237 } 2238 } else { 2239 err = dquot_reserve_block(inode, 1); 2240 if (err) 2241 return err; 2242 } 2243 2244 if (time_to_inject(sbi, FAULT_BLOCK)) { 2245 f2fs_show_injection_info(sbi, FAULT_BLOCK); 2246 goto enospc; 2247 } 2248 2249 spin_lock(&sbi->stat_lock); 2250 2251 valid_block_count = sbi->total_valid_block_count + 2252 sbi->current_reserved_blocks + 1; 2253 2254 if (!__allow_reserved_blocks(sbi, inode, false)) 2255 valid_block_count += F2FS_OPTION(sbi).root_reserved_blocks; 2256 user_block_count = sbi->user_block_count; 2257 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) 2258 user_block_count -= sbi->unusable_block_count; 2259 2260 if (unlikely(valid_block_count > user_block_count)) { 2261 spin_unlock(&sbi->stat_lock); 2262 goto enospc; 2263 } 2264 2265 valid_node_count = sbi->total_valid_node_count + 1; 2266 if (unlikely(valid_node_count > sbi->total_node_count)) { 2267 spin_unlock(&sbi->stat_lock); 2268 goto enospc; 2269 } 2270 2271 sbi->total_valid_node_count++; 2272 sbi->total_valid_block_count++; 2273 spin_unlock(&sbi->stat_lock); 2274 2275 if (inode) { 2276 if (is_inode) 2277 f2fs_mark_inode_dirty_sync(inode, true); 2278 else 2279 f2fs_i_blocks_write(inode, 1, true, true); 2280 } 2281 2282 percpu_counter_inc(&sbi->alloc_valid_block_count); 2283 return 0; 2284 2285 enospc: 2286 if (is_inode) { 2287 if (inode) 2288 dquot_free_inode(inode); 2289 } else { 2290 dquot_release_reservation_block(inode, 1); 2291 } 2292 return -ENOSPC; 2293 } 2294 2295 static inline void dec_valid_node_count(struct f2fs_sb_info *sbi, 2296 struct inode *inode, bool is_inode) 2297 { 2298 spin_lock(&sbi->stat_lock); 2299 2300 f2fs_bug_on(sbi, !sbi->total_valid_block_count); 2301 f2fs_bug_on(sbi, !sbi->total_valid_node_count); 2302 2303 sbi->total_valid_node_count--; 2304 sbi->total_valid_block_count--; 2305 if (sbi->reserved_blocks && 2306 sbi->current_reserved_blocks < sbi->reserved_blocks) 2307 sbi->current_reserved_blocks++; 2308 2309 spin_unlock(&sbi->stat_lock); 2310 2311 if (is_inode) { 2312 dquot_free_inode(inode); 2313 } else { 2314 if (unlikely(inode->i_blocks == 0)) { 2315 f2fs_warn(sbi, "dec_valid_node_count: inconsistent i_blocks, ino:%lu, iblocks:%llu", 2316 inode->i_ino, 2317 (unsigned long long)inode->i_blocks); 2318 set_sbi_flag(sbi, SBI_NEED_FSCK); 2319 return; 2320 } 2321 f2fs_i_blocks_write(inode, 1, false, true); 2322 } 2323 } 2324 2325 static inline unsigned int valid_node_count(struct f2fs_sb_info *sbi) 2326 { 2327 return sbi->total_valid_node_count; 2328 } 2329 2330 static inline void inc_valid_inode_count(struct f2fs_sb_info *sbi) 2331 { 2332 percpu_counter_inc(&sbi->total_valid_inode_count); 2333 } 2334 2335 static inline void dec_valid_inode_count(struct f2fs_sb_info *sbi) 2336 { 2337 percpu_counter_dec(&sbi->total_valid_inode_count); 2338 } 2339 2340 static inline s64 valid_inode_count(struct f2fs_sb_info *sbi) 2341 { 2342 return percpu_counter_sum_positive(&sbi->total_valid_inode_count); 2343 } 2344 2345 static inline struct page *f2fs_grab_cache_page(struct address_space *mapping, 2346 pgoff_t index, bool for_write) 2347 { 2348 struct page *page; 2349 2350 if (IS_ENABLED(CONFIG_F2FS_FAULT_INJECTION)) { 2351 if (!for_write) 2352 page = find_get_page_flags(mapping, index, 2353 FGP_LOCK | FGP_ACCESSED); 2354 else 2355 page = find_lock_page(mapping, index); 2356 if (page) 2357 return page; 2358 2359 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_ALLOC)) { 2360 f2fs_show_injection_info(F2FS_M_SB(mapping), 2361 FAULT_PAGE_ALLOC); 2362 return NULL; 2363 } 2364 } 2365 2366 if (!for_write) 2367 return grab_cache_page(mapping, index); 2368 return grab_cache_page_write_begin(mapping, index, AOP_FLAG_NOFS); 2369 } 2370 2371 static inline struct page *f2fs_pagecache_get_page( 2372 struct address_space *mapping, pgoff_t index, 2373 int fgp_flags, gfp_t gfp_mask) 2374 { 2375 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_GET)) { 2376 f2fs_show_injection_info(F2FS_M_SB(mapping), FAULT_PAGE_GET); 2377 return NULL; 2378 } 2379 2380 return pagecache_get_page(mapping, index, fgp_flags, gfp_mask); 2381 } 2382 2383 static inline void f2fs_copy_page(struct page *src, struct page *dst) 2384 { 2385 char *src_kaddr = kmap(src); 2386 char *dst_kaddr = kmap(dst); 2387 2388 memcpy(dst_kaddr, src_kaddr, PAGE_SIZE); 2389 kunmap(dst); 2390 kunmap(src); 2391 } 2392 2393 static inline void f2fs_put_page(struct page *page, int unlock) 2394 { 2395 if (!page) 2396 return; 2397 2398 if (unlock) { 2399 f2fs_bug_on(F2FS_P_SB(page), !PageLocked(page)); 2400 unlock_page(page); 2401 } 2402 put_page(page); 2403 } 2404 2405 static inline void f2fs_put_dnode(struct dnode_of_data *dn) 2406 { 2407 if (dn->node_page) 2408 f2fs_put_page(dn->node_page, 1); 2409 if (dn->inode_page && dn->node_page != dn->inode_page) 2410 f2fs_put_page(dn->inode_page, 0); 2411 dn->node_page = NULL; 2412 dn->inode_page = NULL; 2413 } 2414 2415 static inline struct kmem_cache *f2fs_kmem_cache_create(const char *name, 2416 size_t size) 2417 { 2418 return kmem_cache_create(name, size, 0, SLAB_RECLAIM_ACCOUNT, NULL); 2419 } 2420 2421 static inline void *f2fs_kmem_cache_alloc(struct kmem_cache *cachep, 2422 gfp_t flags) 2423 { 2424 void *entry; 2425 2426 entry = kmem_cache_alloc(cachep, flags); 2427 if (!entry) 2428 entry = kmem_cache_alloc(cachep, flags | __GFP_NOFAIL); 2429 return entry; 2430 } 2431 2432 static inline bool is_inflight_io(struct f2fs_sb_info *sbi, int type) 2433 { 2434 if (get_pages(sbi, F2FS_RD_DATA) || get_pages(sbi, F2FS_RD_NODE) || 2435 get_pages(sbi, F2FS_RD_META) || get_pages(sbi, F2FS_WB_DATA) || 2436 get_pages(sbi, F2FS_WB_CP_DATA) || 2437 get_pages(sbi, F2FS_DIO_READ) || 2438 get_pages(sbi, F2FS_DIO_WRITE)) 2439 return true; 2440 2441 if (type != DISCARD_TIME && SM_I(sbi) && SM_I(sbi)->dcc_info && 2442 atomic_read(&SM_I(sbi)->dcc_info->queued_discard)) 2443 return true; 2444 2445 if (SM_I(sbi) && SM_I(sbi)->fcc_info && 2446 atomic_read(&SM_I(sbi)->fcc_info->queued_flush)) 2447 return true; 2448 return false; 2449 } 2450 2451 static inline bool is_idle(struct f2fs_sb_info *sbi, int type) 2452 { 2453 if (sbi->gc_mode == GC_URGENT_HIGH) 2454 return true; 2455 2456 if (is_inflight_io(sbi, type)) 2457 return false; 2458 2459 if (sbi->gc_mode == GC_URGENT_LOW && 2460 (type == DISCARD_TIME || type == GC_TIME)) 2461 return true; 2462 2463 return f2fs_time_over(sbi, type); 2464 } 2465 2466 static inline void f2fs_radix_tree_insert(struct radix_tree_root *root, 2467 unsigned long index, void *item) 2468 { 2469 while (radix_tree_insert(root, index, item)) 2470 cond_resched(); 2471 } 2472 2473 #define RAW_IS_INODE(p) ((p)->footer.nid == (p)->footer.ino) 2474 2475 static inline bool IS_INODE(struct page *page) 2476 { 2477 struct f2fs_node *p = F2FS_NODE(page); 2478 2479 return RAW_IS_INODE(p); 2480 } 2481 2482 static inline int offset_in_addr(struct f2fs_inode *i) 2483 { 2484 return (i->i_inline & F2FS_EXTRA_ATTR) ? 2485 (le16_to_cpu(i->i_extra_isize) / sizeof(__le32)) : 0; 2486 } 2487 2488 static inline __le32 *blkaddr_in_node(struct f2fs_node *node) 2489 { 2490 return RAW_IS_INODE(node) ? node->i.i_addr : node->dn.addr; 2491 } 2492 2493 static inline int f2fs_has_extra_attr(struct inode *inode); 2494 static inline block_t data_blkaddr(struct inode *inode, 2495 struct page *node_page, unsigned int offset) 2496 { 2497 struct f2fs_node *raw_node; 2498 __le32 *addr_array; 2499 int base = 0; 2500 bool is_inode = IS_INODE(node_page); 2501 2502 raw_node = F2FS_NODE(node_page); 2503 2504 if (is_inode) { 2505 if (!inode) 2506 /* from GC path only */ 2507 base = offset_in_addr(&raw_node->i); 2508 else if (f2fs_has_extra_attr(inode)) 2509 base = get_extra_isize(inode); 2510 } 2511 2512 addr_array = blkaddr_in_node(raw_node); 2513 return le32_to_cpu(addr_array[base + offset]); 2514 } 2515 2516 static inline block_t f2fs_data_blkaddr(struct dnode_of_data *dn) 2517 { 2518 return data_blkaddr(dn->inode, dn->node_page, dn->ofs_in_node); 2519 } 2520 2521 static inline int f2fs_test_bit(unsigned int nr, char *addr) 2522 { 2523 int mask; 2524 2525 addr += (nr >> 3); 2526 mask = 1 << (7 - (nr & 0x07)); 2527 return mask & *addr; 2528 } 2529 2530 static inline void f2fs_set_bit(unsigned int nr, char *addr) 2531 { 2532 int mask; 2533 2534 addr += (nr >> 3); 2535 mask = 1 << (7 - (nr & 0x07)); 2536 *addr |= mask; 2537 } 2538 2539 static inline void f2fs_clear_bit(unsigned int nr, char *addr) 2540 { 2541 int mask; 2542 2543 addr += (nr >> 3); 2544 mask = 1 << (7 - (nr & 0x07)); 2545 *addr &= ~mask; 2546 } 2547 2548 static inline int f2fs_test_and_set_bit(unsigned int nr, char *addr) 2549 { 2550 int mask; 2551 int ret; 2552 2553 addr += (nr >> 3); 2554 mask = 1 << (7 - (nr & 0x07)); 2555 ret = mask & *addr; 2556 *addr |= mask; 2557 return ret; 2558 } 2559 2560 static inline int f2fs_test_and_clear_bit(unsigned int nr, char *addr) 2561 { 2562 int mask; 2563 int ret; 2564 2565 addr += (nr >> 3); 2566 mask = 1 << (7 - (nr & 0x07)); 2567 ret = mask & *addr; 2568 *addr &= ~mask; 2569 return ret; 2570 } 2571 2572 static inline void f2fs_change_bit(unsigned int nr, char *addr) 2573 { 2574 int mask; 2575 2576 addr += (nr >> 3); 2577 mask = 1 << (7 - (nr & 0x07)); 2578 *addr ^= mask; 2579 } 2580 2581 /* 2582 * On-disk inode flags (f2fs_inode::i_flags) 2583 */ 2584 #define F2FS_COMPR_FL 0x00000004 /* Compress file */ 2585 #define F2FS_SYNC_FL 0x00000008 /* Synchronous updates */ 2586 #define F2FS_IMMUTABLE_FL 0x00000010 /* Immutable file */ 2587 #define F2FS_APPEND_FL 0x00000020 /* writes to file may only append */ 2588 #define F2FS_NODUMP_FL 0x00000040 /* do not dump file */ 2589 #define F2FS_NOATIME_FL 0x00000080 /* do not update atime */ 2590 #define F2FS_NOCOMP_FL 0x00000400 /* Don't compress */ 2591 #define F2FS_INDEX_FL 0x00001000 /* hash-indexed directory */ 2592 #define F2FS_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */ 2593 #define F2FS_PROJINHERIT_FL 0x20000000 /* Create with parents projid */ 2594 #define F2FS_CASEFOLD_FL 0x40000000 /* Casefolded file */ 2595 2596 /* Flags that should be inherited by new inodes from their parent. */ 2597 #define F2FS_FL_INHERITED (F2FS_SYNC_FL | F2FS_NODUMP_FL | F2FS_NOATIME_FL | \ 2598 F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \ 2599 F2FS_CASEFOLD_FL | F2FS_COMPR_FL | F2FS_NOCOMP_FL) 2600 2601 /* Flags that are appropriate for regular files (all but dir-specific ones). */ 2602 #define F2FS_REG_FLMASK (~(F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \ 2603 F2FS_CASEFOLD_FL)) 2604 2605 /* Flags that are appropriate for non-directories/regular files. */ 2606 #define F2FS_OTHER_FLMASK (F2FS_NODUMP_FL | F2FS_NOATIME_FL) 2607 2608 static inline __u32 f2fs_mask_flags(umode_t mode, __u32 flags) 2609 { 2610 if (S_ISDIR(mode)) 2611 return flags; 2612 else if (S_ISREG(mode)) 2613 return flags & F2FS_REG_FLMASK; 2614 else 2615 return flags & F2FS_OTHER_FLMASK; 2616 } 2617 2618 static inline void __mark_inode_dirty_flag(struct inode *inode, 2619 int flag, bool set) 2620 { 2621 switch (flag) { 2622 case FI_INLINE_XATTR: 2623 case FI_INLINE_DATA: 2624 case FI_INLINE_DENTRY: 2625 case FI_NEW_INODE: 2626 if (set) 2627 return; 2628 fallthrough; 2629 case FI_DATA_EXIST: 2630 case FI_INLINE_DOTS: 2631 case FI_PIN_FILE: 2632 f2fs_mark_inode_dirty_sync(inode, true); 2633 } 2634 } 2635 2636 static inline void set_inode_flag(struct inode *inode, int flag) 2637 { 2638 set_bit(flag, F2FS_I(inode)->flags); 2639 __mark_inode_dirty_flag(inode, flag, true); 2640 } 2641 2642 static inline int is_inode_flag_set(struct inode *inode, int flag) 2643 { 2644 return test_bit(flag, F2FS_I(inode)->flags); 2645 } 2646 2647 static inline void clear_inode_flag(struct inode *inode, int flag) 2648 { 2649 clear_bit(flag, F2FS_I(inode)->flags); 2650 __mark_inode_dirty_flag(inode, flag, false); 2651 } 2652 2653 static inline bool f2fs_verity_in_progress(struct inode *inode) 2654 { 2655 return IS_ENABLED(CONFIG_FS_VERITY) && 2656 is_inode_flag_set(inode, FI_VERITY_IN_PROGRESS); 2657 } 2658 2659 static inline void set_acl_inode(struct inode *inode, umode_t mode) 2660 { 2661 F2FS_I(inode)->i_acl_mode = mode; 2662 set_inode_flag(inode, FI_ACL_MODE); 2663 f2fs_mark_inode_dirty_sync(inode, false); 2664 } 2665 2666 static inline void f2fs_i_links_write(struct inode *inode, bool inc) 2667 { 2668 if (inc) 2669 inc_nlink(inode); 2670 else 2671 drop_nlink(inode); 2672 f2fs_mark_inode_dirty_sync(inode, true); 2673 } 2674 2675 static inline void f2fs_i_blocks_write(struct inode *inode, 2676 block_t diff, bool add, bool claim) 2677 { 2678 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE); 2679 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER); 2680 2681 /* add = 1, claim = 1 should be dquot_reserve_block in pair */ 2682 if (add) { 2683 if (claim) 2684 dquot_claim_block(inode, diff); 2685 else 2686 dquot_alloc_block_nofail(inode, diff); 2687 } else { 2688 dquot_free_block(inode, diff); 2689 } 2690 2691 f2fs_mark_inode_dirty_sync(inode, true); 2692 if (clean || recover) 2693 set_inode_flag(inode, FI_AUTO_RECOVER); 2694 } 2695 2696 static inline void f2fs_i_size_write(struct inode *inode, loff_t i_size) 2697 { 2698 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE); 2699 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER); 2700 2701 if (i_size_read(inode) == i_size) 2702 return; 2703 2704 i_size_write(inode, i_size); 2705 f2fs_mark_inode_dirty_sync(inode, true); 2706 if (clean || recover) 2707 set_inode_flag(inode, FI_AUTO_RECOVER); 2708 } 2709 2710 static inline void f2fs_i_depth_write(struct inode *inode, unsigned int depth) 2711 { 2712 F2FS_I(inode)->i_current_depth = depth; 2713 f2fs_mark_inode_dirty_sync(inode, true); 2714 } 2715 2716 static inline void f2fs_i_gc_failures_write(struct inode *inode, 2717 unsigned int count) 2718 { 2719 F2FS_I(inode)->i_gc_failures[GC_FAILURE_PIN] = count; 2720 f2fs_mark_inode_dirty_sync(inode, true); 2721 } 2722 2723 static inline void f2fs_i_xnid_write(struct inode *inode, nid_t xnid) 2724 { 2725 F2FS_I(inode)->i_xattr_nid = xnid; 2726 f2fs_mark_inode_dirty_sync(inode, true); 2727 } 2728 2729 static inline void f2fs_i_pino_write(struct inode *inode, nid_t pino) 2730 { 2731 F2FS_I(inode)->i_pino = pino; 2732 f2fs_mark_inode_dirty_sync(inode, true); 2733 } 2734 2735 static inline void get_inline_info(struct inode *inode, struct f2fs_inode *ri) 2736 { 2737 struct f2fs_inode_info *fi = F2FS_I(inode); 2738 2739 if (ri->i_inline & F2FS_INLINE_XATTR) 2740 set_bit(FI_INLINE_XATTR, fi->flags); 2741 if (ri->i_inline & F2FS_INLINE_DATA) 2742 set_bit(FI_INLINE_DATA, fi->flags); 2743 if (ri->i_inline & F2FS_INLINE_DENTRY) 2744 set_bit(FI_INLINE_DENTRY, fi->flags); 2745 if (ri->i_inline & F2FS_DATA_EXIST) 2746 set_bit(FI_DATA_EXIST, fi->flags); 2747 if (ri->i_inline & F2FS_INLINE_DOTS) 2748 set_bit(FI_INLINE_DOTS, fi->flags); 2749 if (ri->i_inline & F2FS_EXTRA_ATTR) 2750 set_bit(FI_EXTRA_ATTR, fi->flags); 2751 if (ri->i_inline & F2FS_PIN_FILE) 2752 set_bit(FI_PIN_FILE, fi->flags); 2753 } 2754 2755 static inline void set_raw_inline(struct inode *inode, struct f2fs_inode *ri) 2756 { 2757 ri->i_inline = 0; 2758 2759 if (is_inode_flag_set(inode, FI_INLINE_XATTR)) 2760 ri->i_inline |= F2FS_INLINE_XATTR; 2761 if (is_inode_flag_set(inode, FI_INLINE_DATA)) 2762 ri->i_inline |= F2FS_INLINE_DATA; 2763 if (is_inode_flag_set(inode, FI_INLINE_DENTRY)) 2764 ri->i_inline |= F2FS_INLINE_DENTRY; 2765 if (is_inode_flag_set(inode, FI_DATA_EXIST)) 2766 ri->i_inline |= F2FS_DATA_EXIST; 2767 if (is_inode_flag_set(inode, FI_INLINE_DOTS)) 2768 ri->i_inline |= F2FS_INLINE_DOTS; 2769 if (is_inode_flag_set(inode, FI_EXTRA_ATTR)) 2770 ri->i_inline |= F2FS_EXTRA_ATTR; 2771 if (is_inode_flag_set(inode, FI_PIN_FILE)) 2772 ri->i_inline |= F2FS_PIN_FILE; 2773 } 2774 2775 static inline int f2fs_has_extra_attr(struct inode *inode) 2776 { 2777 return is_inode_flag_set(inode, FI_EXTRA_ATTR); 2778 } 2779 2780 static inline int f2fs_has_inline_xattr(struct inode *inode) 2781 { 2782 return is_inode_flag_set(inode, FI_INLINE_XATTR); 2783 } 2784 2785 static inline int f2fs_compressed_file(struct inode *inode) 2786 { 2787 return S_ISREG(inode->i_mode) && 2788 is_inode_flag_set(inode, FI_COMPRESSED_FILE); 2789 } 2790 2791 static inline bool f2fs_need_compress_data(struct inode *inode) 2792 { 2793 int compress_mode = F2FS_OPTION(F2FS_I_SB(inode)).compress_mode; 2794 2795 if (!f2fs_compressed_file(inode)) 2796 return false; 2797 2798 if (compress_mode == COMPR_MODE_FS) 2799 return true; 2800 else if (compress_mode == COMPR_MODE_USER && 2801 is_inode_flag_set(inode, FI_ENABLE_COMPRESS)) 2802 return true; 2803 2804 return false; 2805 } 2806 2807 static inline unsigned int addrs_per_inode(struct inode *inode) 2808 { 2809 unsigned int addrs = CUR_ADDRS_PER_INODE(inode) - 2810 get_inline_xattr_addrs(inode); 2811 2812 if (!f2fs_compressed_file(inode)) 2813 return addrs; 2814 return ALIGN_DOWN(addrs, F2FS_I(inode)->i_cluster_size); 2815 } 2816 2817 static inline unsigned int addrs_per_block(struct inode *inode) 2818 { 2819 if (!f2fs_compressed_file(inode)) 2820 return DEF_ADDRS_PER_BLOCK; 2821 return ALIGN_DOWN(DEF_ADDRS_PER_BLOCK, F2FS_I(inode)->i_cluster_size); 2822 } 2823 2824 static inline void *inline_xattr_addr(struct inode *inode, struct page *page) 2825 { 2826 struct f2fs_inode *ri = F2FS_INODE(page); 2827 2828 return (void *)&(ri->i_addr[DEF_ADDRS_PER_INODE - 2829 get_inline_xattr_addrs(inode)]); 2830 } 2831 2832 static inline int inline_xattr_size(struct inode *inode) 2833 { 2834 if (f2fs_has_inline_xattr(inode)) 2835 return get_inline_xattr_addrs(inode) * sizeof(__le32); 2836 return 0; 2837 } 2838 2839 static inline int f2fs_has_inline_data(struct inode *inode) 2840 { 2841 return is_inode_flag_set(inode, FI_INLINE_DATA); 2842 } 2843 2844 static inline int f2fs_exist_data(struct inode *inode) 2845 { 2846 return is_inode_flag_set(inode, FI_DATA_EXIST); 2847 } 2848 2849 static inline int f2fs_has_inline_dots(struct inode *inode) 2850 { 2851 return is_inode_flag_set(inode, FI_INLINE_DOTS); 2852 } 2853 2854 static inline int f2fs_is_mmap_file(struct inode *inode) 2855 { 2856 return is_inode_flag_set(inode, FI_MMAP_FILE); 2857 } 2858 2859 static inline bool f2fs_is_pinned_file(struct inode *inode) 2860 { 2861 return is_inode_flag_set(inode, FI_PIN_FILE); 2862 } 2863 2864 static inline bool f2fs_is_atomic_file(struct inode *inode) 2865 { 2866 return is_inode_flag_set(inode, FI_ATOMIC_FILE); 2867 } 2868 2869 static inline bool f2fs_is_commit_atomic_write(struct inode *inode) 2870 { 2871 return is_inode_flag_set(inode, FI_ATOMIC_COMMIT); 2872 } 2873 2874 static inline bool f2fs_is_volatile_file(struct inode *inode) 2875 { 2876 return is_inode_flag_set(inode, FI_VOLATILE_FILE); 2877 } 2878 2879 static inline bool f2fs_is_first_block_written(struct inode *inode) 2880 { 2881 return is_inode_flag_set(inode, FI_FIRST_BLOCK_WRITTEN); 2882 } 2883 2884 static inline bool f2fs_is_drop_cache(struct inode *inode) 2885 { 2886 return is_inode_flag_set(inode, FI_DROP_CACHE); 2887 } 2888 2889 static inline void *inline_data_addr(struct inode *inode, struct page *page) 2890 { 2891 struct f2fs_inode *ri = F2FS_INODE(page); 2892 int extra_size = get_extra_isize(inode); 2893 2894 return (void *)&(ri->i_addr[extra_size + DEF_INLINE_RESERVED_SIZE]); 2895 } 2896 2897 static inline int f2fs_has_inline_dentry(struct inode *inode) 2898 { 2899 return is_inode_flag_set(inode, FI_INLINE_DENTRY); 2900 } 2901 2902 static inline int is_file(struct inode *inode, int type) 2903 { 2904 return F2FS_I(inode)->i_advise & type; 2905 } 2906 2907 static inline void set_file(struct inode *inode, int type) 2908 { 2909 F2FS_I(inode)->i_advise |= type; 2910 f2fs_mark_inode_dirty_sync(inode, true); 2911 } 2912 2913 static inline void clear_file(struct inode *inode, int type) 2914 { 2915 F2FS_I(inode)->i_advise &= ~type; 2916 f2fs_mark_inode_dirty_sync(inode, true); 2917 } 2918 2919 static inline bool f2fs_is_time_consistent(struct inode *inode) 2920 { 2921 if (!timespec64_equal(F2FS_I(inode)->i_disk_time, &inode->i_atime)) 2922 return false; 2923 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 1, &inode->i_ctime)) 2924 return false; 2925 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 2, &inode->i_mtime)) 2926 return false; 2927 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 3, 2928 &F2FS_I(inode)->i_crtime)) 2929 return false; 2930 return true; 2931 } 2932 2933 static inline bool f2fs_skip_inode_update(struct inode *inode, int dsync) 2934 { 2935 bool ret; 2936 2937 if (dsync) { 2938 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 2939 2940 spin_lock(&sbi->inode_lock[DIRTY_META]); 2941 ret = list_empty(&F2FS_I(inode)->gdirty_list); 2942 spin_unlock(&sbi->inode_lock[DIRTY_META]); 2943 return ret; 2944 } 2945 if (!is_inode_flag_set(inode, FI_AUTO_RECOVER) || 2946 file_keep_isize(inode) || 2947 i_size_read(inode) & ~PAGE_MASK) 2948 return false; 2949 2950 if (!f2fs_is_time_consistent(inode)) 2951 return false; 2952 2953 spin_lock(&F2FS_I(inode)->i_size_lock); 2954 ret = F2FS_I(inode)->last_disk_size == i_size_read(inode); 2955 spin_unlock(&F2FS_I(inode)->i_size_lock); 2956 2957 return ret; 2958 } 2959 2960 static inline bool f2fs_readonly(struct super_block *sb) 2961 { 2962 return sb_rdonly(sb); 2963 } 2964 2965 static inline bool f2fs_cp_error(struct f2fs_sb_info *sbi) 2966 { 2967 return is_set_ckpt_flags(sbi, CP_ERROR_FLAG); 2968 } 2969 2970 static inline bool is_dot_dotdot(const u8 *name, size_t len) 2971 { 2972 if (len == 1 && name[0] == '.') 2973 return true; 2974 2975 if (len == 2 && name[0] == '.' && name[1] == '.') 2976 return true; 2977 2978 return false; 2979 } 2980 2981 static inline bool f2fs_may_extent_tree(struct inode *inode) 2982 { 2983 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 2984 2985 if (!test_opt(sbi, EXTENT_CACHE) || 2986 is_inode_flag_set(inode, FI_NO_EXTENT) || 2987 is_inode_flag_set(inode, FI_COMPRESSED_FILE)) 2988 return false; 2989 2990 /* 2991 * for recovered files during mount do not create extents 2992 * if shrinker is not registered. 2993 */ 2994 if (list_empty(&sbi->s_list)) 2995 return false; 2996 2997 return S_ISREG(inode->i_mode); 2998 } 2999 3000 static inline void *f2fs_kmalloc(struct f2fs_sb_info *sbi, 3001 size_t size, gfp_t flags) 3002 { 3003 if (time_to_inject(sbi, FAULT_KMALLOC)) { 3004 f2fs_show_injection_info(sbi, FAULT_KMALLOC); 3005 return NULL; 3006 } 3007 3008 return kmalloc(size, flags); 3009 } 3010 3011 static inline void *f2fs_kzalloc(struct f2fs_sb_info *sbi, 3012 size_t size, gfp_t flags) 3013 { 3014 return f2fs_kmalloc(sbi, size, flags | __GFP_ZERO); 3015 } 3016 3017 static inline void *f2fs_kvmalloc(struct f2fs_sb_info *sbi, 3018 size_t size, gfp_t flags) 3019 { 3020 if (time_to_inject(sbi, FAULT_KVMALLOC)) { 3021 f2fs_show_injection_info(sbi, FAULT_KVMALLOC); 3022 return NULL; 3023 } 3024 3025 return kvmalloc(size, flags); 3026 } 3027 3028 static inline void *f2fs_kvzalloc(struct f2fs_sb_info *sbi, 3029 size_t size, gfp_t flags) 3030 { 3031 return f2fs_kvmalloc(sbi, size, flags | __GFP_ZERO); 3032 } 3033 3034 static inline int get_extra_isize(struct inode *inode) 3035 { 3036 return F2FS_I(inode)->i_extra_isize / sizeof(__le32); 3037 } 3038 3039 static inline int get_inline_xattr_addrs(struct inode *inode) 3040 { 3041 return F2FS_I(inode)->i_inline_xattr_size; 3042 } 3043 3044 #define f2fs_get_inode_mode(i) \ 3045 ((is_inode_flag_set(i, FI_ACL_MODE)) ? \ 3046 (F2FS_I(i)->i_acl_mode) : ((i)->i_mode)) 3047 3048 #define F2FS_TOTAL_EXTRA_ATTR_SIZE \ 3049 (offsetof(struct f2fs_inode, i_extra_end) - \ 3050 offsetof(struct f2fs_inode, i_extra_isize)) \ 3051 3052 #define F2FS_OLD_ATTRIBUTE_SIZE (offsetof(struct f2fs_inode, i_addr)) 3053 #define F2FS_FITS_IN_INODE(f2fs_inode, extra_isize, field) \ 3054 ((offsetof(typeof(*(f2fs_inode)), field) + \ 3055 sizeof((f2fs_inode)->field)) \ 3056 <= (F2FS_OLD_ATTRIBUTE_SIZE + (extra_isize))) \ 3057 3058 #define DEFAULT_IOSTAT_PERIOD_MS 3000 3059 #define MIN_IOSTAT_PERIOD_MS 100 3060 /* maximum period of iostat tracing is 1 day */ 3061 #define MAX_IOSTAT_PERIOD_MS 8640000 3062 3063 static inline void f2fs_reset_iostat(struct f2fs_sb_info *sbi) 3064 { 3065 int i; 3066 3067 spin_lock(&sbi->iostat_lock); 3068 for (i = 0; i < NR_IO_TYPE; i++) { 3069 sbi->rw_iostat[i] = 0; 3070 sbi->prev_rw_iostat[i] = 0; 3071 } 3072 spin_unlock(&sbi->iostat_lock); 3073 } 3074 3075 extern void f2fs_record_iostat(struct f2fs_sb_info *sbi); 3076 3077 static inline void f2fs_update_iostat(struct f2fs_sb_info *sbi, 3078 enum iostat_type type, unsigned long long io_bytes) 3079 { 3080 if (!sbi->iostat_enable) 3081 return; 3082 spin_lock(&sbi->iostat_lock); 3083 sbi->rw_iostat[type] += io_bytes; 3084 3085 if (type == APP_WRITE_IO || type == APP_DIRECT_IO) 3086 sbi->rw_iostat[APP_BUFFERED_IO] = 3087 sbi->rw_iostat[APP_WRITE_IO] - 3088 sbi->rw_iostat[APP_DIRECT_IO]; 3089 3090 if (type == APP_READ_IO || type == APP_DIRECT_READ_IO) 3091 sbi->rw_iostat[APP_BUFFERED_READ_IO] = 3092 sbi->rw_iostat[APP_READ_IO] - 3093 sbi->rw_iostat[APP_DIRECT_READ_IO]; 3094 spin_unlock(&sbi->iostat_lock); 3095 3096 f2fs_record_iostat(sbi); 3097 } 3098 3099 #define __is_large_section(sbi) ((sbi)->segs_per_sec > 1) 3100 3101 #define __is_meta_io(fio) (PAGE_TYPE_OF_BIO((fio)->type) == META) 3102 3103 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi, 3104 block_t blkaddr, int type); 3105 static inline void verify_blkaddr(struct f2fs_sb_info *sbi, 3106 block_t blkaddr, int type) 3107 { 3108 if (!f2fs_is_valid_blkaddr(sbi, blkaddr, type)) { 3109 f2fs_err(sbi, "invalid blkaddr: %u, type: %d, run fsck to fix.", 3110 blkaddr, type); 3111 f2fs_bug_on(sbi, 1); 3112 } 3113 } 3114 3115 static inline bool __is_valid_data_blkaddr(block_t blkaddr) 3116 { 3117 if (blkaddr == NEW_ADDR || blkaddr == NULL_ADDR || 3118 blkaddr == COMPRESS_ADDR) 3119 return false; 3120 return true; 3121 } 3122 3123 static inline void f2fs_set_page_private(struct page *page, 3124 unsigned long data) 3125 { 3126 if (PagePrivate(page)) 3127 return; 3128 3129 attach_page_private(page, (void *)data); 3130 } 3131 3132 static inline void f2fs_clear_page_private(struct page *page) 3133 { 3134 detach_page_private(page); 3135 } 3136 3137 /* 3138 * file.c 3139 */ 3140 int f2fs_sync_file(struct file *file, loff_t start, loff_t end, int datasync); 3141 void f2fs_truncate_data_blocks(struct dnode_of_data *dn); 3142 int f2fs_do_truncate_blocks(struct inode *inode, u64 from, bool lock); 3143 int f2fs_truncate_blocks(struct inode *inode, u64 from, bool lock); 3144 int f2fs_truncate(struct inode *inode); 3145 int f2fs_getattr(const struct path *path, struct kstat *stat, 3146 u32 request_mask, unsigned int flags); 3147 int f2fs_setattr(struct dentry *dentry, struct iattr *attr); 3148 int f2fs_truncate_hole(struct inode *inode, pgoff_t pg_start, pgoff_t pg_end); 3149 void f2fs_truncate_data_blocks_range(struct dnode_of_data *dn, int count); 3150 int f2fs_precache_extents(struct inode *inode); 3151 long f2fs_ioctl(struct file *filp, unsigned int cmd, unsigned long arg); 3152 long f2fs_compat_ioctl(struct file *file, unsigned int cmd, unsigned long arg); 3153 int f2fs_transfer_project_quota(struct inode *inode, kprojid_t kprojid); 3154 int f2fs_pin_file_control(struct inode *inode, bool inc); 3155 3156 /* 3157 * inode.c 3158 */ 3159 void f2fs_set_inode_flags(struct inode *inode); 3160 bool f2fs_inode_chksum_verify(struct f2fs_sb_info *sbi, struct page *page); 3161 void f2fs_inode_chksum_set(struct f2fs_sb_info *sbi, struct page *page); 3162 struct inode *f2fs_iget(struct super_block *sb, unsigned long ino); 3163 struct inode *f2fs_iget_retry(struct super_block *sb, unsigned long ino); 3164 int f2fs_try_to_free_nats(struct f2fs_sb_info *sbi, int nr_shrink); 3165 void f2fs_update_inode(struct inode *inode, struct page *node_page); 3166 void f2fs_update_inode_page(struct inode *inode); 3167 int f2fs_write_inode(struct inode *inode, struct writeback_control *wbc); 3168 void f2fs_evict_inode(struct inode *inode); 3169 void f2fs_handle_failed_inode(struct inode *inode); 3170 3171 /* 3172 * namei.c 3173 */ 3174 int f2fs_update_extension_list(struct f2fs_sb_info *sbi, const char *name, 3175 bool hot, bool set); 3176 struct dentry *f2fs_get_parent(struct dentry *child); 3177 3178 /* 3179 * dir.c 3180 */ 3181 unsigned char f2fs_get_de_type(struct f2fs_dir_entry *de); 3182 int f2fs_init_casefolded_name(const struct inode *dir, 3183 struct f2fs_filename *fname); 3184 int f2fs_setup_filename(struct inode *dir, const struct qstr *iname, 3185 int lookup, struct f2fs_filename *fname); 3186 int f2fs_prepare_lookup(struct inode *dir, struct dentry *dentry, 3187 struct f2fs_filename *fname); 3188 void f2fs_free_filename(struct f2fs_filename *fname); 3189 struct f2fs_dir_entry *f2fs_find_target_dentry(const struct f2fs_dentry_ptr *d, 3190 const struct f2fs_filename *fname, int *max_slots); 3191 int f2fs_fill_dentries(struct dir_context *ctx, struct f2fs_dentry_ptr *d, 3192 unsigned int start_pos, struct fscrypt_str *fstr); 3193 void f2fs_do_make_empty_dir(struct inode *inode, struct inode *parent, 3194 struct f2fs_dentry_ptr *d); 3195 struct page *f2fs_init_inode_metadata(struct inode *inode, struct inode *dir, 3196 const struct f2fs_filename *fname, struct page *dpage); 3197 void f2fs_update_parent_metadata(struct inode *dir, struct inode *inode, 3198 unsigned int current_depth); 3199 int f2fs_room_for_filename(const void *bitmap, int slots, int max_slots); 3200 void f2fs_drop_nlink(struct inode *dir, struct inode *inode); 3201 struct f2fs_dir_entry *__f2fs_find_entry(struct inode *dir, 3202 const struct f2fs_filename *fname, 3203 struct page **res_page); 3204 struct f2fs_dir_entry *f2fs_find_entry(struct inode *dir, 3205 const struct qstr *child, struct page **res_page); 3206 struct f2fs_dir_entry *f2fs_parent_dir(struct inode *dir, struct page **p); 3207 ino_t f2fs_inode_by_name(struct inode *dir, const struct qstr *qstr, 3208 struct page **page); 3209 void f2fs_set_link(struct inode *dir, struct f2fs_dir_entry *de, 3210 struct page *page, struct inode *inode); 3211 bool f2fs_has_enough_room(struct inode *dir, struct page *ipage, 3212 const struct f2fs_filename *fname); 3213 void f2fs_update_dentry(nid_t ino, umode_t mode, struct f2fs_dentry_ptr *d, 3214 const struct fscrypt_str *name, f2fs_hash_t name_hash, 3215 unsigned int bit_pos); 3216 int f2fs_add_regular_entry(struct inode *dir, const struct f2fs_filename *fname, 3217 struct inode *inode, nid_t ino, umode_t mode); 3218 int f2fs_add_dentry(struct inode *dir, const struct f2fs_filename *fname, 3219 struct inode *inode, nid_t ino, umode_t mode); 3220 int f2fs_do_add_link(struct inode *dir, const struct qstr *name, 3221 struct inode *inode, nid_t ino, umode_t mode); 3222 void f2fs_delete_entry(struct f2fs_dir_entry *dentry, struct page *page, 3223 struct inode *dir, struct inode *inode); 3224 int f2fs_do_tmpfile(struct inode *inode, struct inode *dir); 3225 bool f2fs_empty_dir(struct inode *dir); 3226 3227 static inline int f2fs_add_link(struct dentry *dentry, struct inode *inode) 3228 { 3229 if (fscrypt_is_nokey_name(dentry)) 3230 return -ENOKEY; 3231 return f2fs_do_add_link(d_inode(dentry->d_parent), &dentry->d_name, 3232 inode, inode->i_ino, inode->i_mode); 3233 } 3234 3235 /* 3236 * super.c 3237 */ 3238 int f2fs_inode_dirtied(struct inode *inode, bool sync); 3239 void f2fs_inode_synced(struct inode *inode); 3240 int f2fs_enable_quota_files(struct f2fs_sb_info *sbi, bool rdonly); 3241 int f2fs_quota_sync(struct super_block *sb, int type); 3242 void f2fs_quota_off_umount(struct super_block *sb); 3243 int f2fs_commit_super(struct f2fs_sb_info *sbi, bool recover); 3244 int f2fs_sync_fs(struct super_block *sb, int sync); 3245 int f2fs_sanity_check_ckpt(struct f2fs_sb_info *sbi); 3246 3247 /* 3248 * hash.c 3249 */ 3250 void f2fs_hash_filename(const struct inode *dir, struct f2fs_filename *fname); 3251 3252 /* 3253 * node.c 3254 */ 3255 struct dnode_of_data; 3256 struct node_info; 3257 3258 int f2fs_check_nid_range(struct f2fs_sb_info *sbi, nid_t nid); 3259 bool f2fs_available_free_memory(struct f2fs_sb_info *sbi, int type); 3260 bool f2fs_in_warm_node_list(struct f2fs_sb_info *sbi, struct page *page); 3261 void f2fs_init_fsync_node_info(struct f2fs_sb_info *sbi); 3262 void f2fs_del_fsync_node_entry(struct f2fs_sb_info *sbi, struct page *page); 3263 void f2fs_reset_fsync_node_info(struct f2fs_sb_info *sbi); 3264 int f2fs_need_dentry_mark(struct f2fs_sb_info *sbi, nid_t nid); 3265 bool f2fs_is_checkpointed_node(struct f2fs_sb_info *sbi, nid_t nid); 3266 bool f2fs_need_inode_block_update(struct f2fs_sb_info *sbi, nid_t ino); 3267 int f2fs_get_node_info(struct f2fs_sb_info *sbi, nid_t nid, 3268 struct node_info *ni); 3269 pgoff_t f2fs_get_next_page_offset(struct dnode_of_data *dn, pgoff_t pgofs); 3270 int f2fs_get_dnode_of_data(struct dnode_of_data *dn, pgoff_t index, int mode); 3271 int f2fs_truncate_inode_blocks(struct inode *inode, pgoff_t from); 3272 int f2fs_truncate_xattr_node(struct inode *inode); 3273 int f2fs_wait_on_node_pages_writeback(struct f2fs_sb_info *sbi, 3274 unsigned int seq_id); 3275 int f2fs_remove_inode_page(struct inode *inode); 3276 struct page *f2fs_new_inode_page(struct inode *inode); 3277 struct page *f2fs_new_node_page(struct dnode_of_data *dn, unsigned int ofs); 3278 void f2fs_ra_node_page(struct f2fs_sb_info *sbi, nid_t nid); 3279 struct page *f2fs_get_node_page(struct f2fs_sb_info *sbi, pgoff_t nid); 3280 struct page *f2fs_get_node_page_ra(struct page *parent, int start); 3281 int f2fs_move_node_page(struct page *node_page, int gc_type); 3282 void f2fs_flush_inline_data(struct f2fs_sb_info *sbi); 3283 int f2fs_fsync_node_pages(struct f2fs_sb_info *sbi, struct inode *inode, 3284 struct writeback_control *wbc, bool atomic, 3285 unsigned int *seq_id); 3286 int f2fs_sync_node_pages(struct f2fs_sb_info *sbi, 3287 struct writeback_control *wbc, 3288 bool do_balance, enum iostat_type io_type); 3289 int f2fs_build_free_nids(struct f2fs_sb_info *sbi, bool sync, bool mount); 3290 bool f2fs_alloc_nid(struct f2fs_sb_info *sbi, nid_t *nid); 3291 void f2fs_alloc_nid_done(struct f2fs_sb_info *sbi, nid_t nid); 3292 void f2fs_alloc_nid_failed(struct f2fs_sb_info *sbi, nid_t nid); 3293 int f2fs_try_to_free_nids(struct f2fs_sb_info *sbi, int nr_shrink); 3294 int f2fs_recover_inline_xattr(struct inode *inode, struct page *page); 3295 int f2fs_recover_xattr_data(struct inode *inode, struct page *page); 3296 int f2fs_recover_inode_page(struct f2fs_sb_info *sbi, struct page *page); 3297 int f2fs_restore_node_summary(struct f2fs_sb_info *sbi, 3298 unsigned int segno, struct f2fs_summary_block *sum); 3299 int f2fs_flush_nat_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3300 int f2fs_build_node_manager(struct f2fs_sb_info *sbi); 3301 void f2fs_destroy_node_manager(struct f2fs_sb_info *sbi); 3302 int __init f2fs_create_node_manager_caches(void); 3303 void f2fs_destroy_node_manager_caches(void); 3304 3305 /* 3306 * segment.c 3307 */ 3308 bool f2fs_need_SSR(struct f2fs_sb_info *sbi); 3309 void f2fs_register_inmem_page(struct inode *inode, struct page *page); 3310 void f2fs_drop_inmem_pages_all(struct f2fs_sb_info *sbi, bool gc_failure); 3311 void f2fs_drop_inmem_pages(struct inode *inode); 3312 void f2fs_drop_inmem_page(struct inode *inode, struct page *page); 3313 int f2fs_commit_inmem_pages(struct inode *inode); 3314 void f2fs_balance_fs(struct f2fs_sb_info *sbi, bool need); 3315 void f2fs_balance_fs_bg(struct f2fs_sb_info *sbi, bool from_bg); 3316 int f2fs_issue_flush(struct f2fs_sb_info *sbi, nid_t ino); 3317 int f2fs_create_flush_cmd_control(struct f2fs_sb_info *sbi); 3318 int f2fs_flush_device_cache(struct f2fs_sb_info *sbi); 3319 void f2fs_destroy_flush_cmd_control(struct f2fs_sb_info *sbi, bool free); 3320 void f2fs_invalidate_blocks(struct f2fs_sb_info *sbi, block_t addr); 3321 bool f2fs_is_checkpointed_data(struct f2fs_sb_info *sbi, block_t blkaddr); 3322 void f2fs_drop_discard_cmd(struct f2fs_sb_info *sbi); 3323 void f2fs_stop_discard_thread(struct f2fs_sb_info *sbi); 3324 bool f2fs_issue_discard_timeout(struct f2fs_sb_info *sbi); 3325 void f2fs_clear_prefree_segments(struct f2fs_sb_info *sbi, 3326 struct cp_control *cpc); 3327 void f2fs_dirty_to_prefree(struct f2fs_sb_info *sbi); 3328 block_t f2fs_get_unusable_blocks(struct f2fs_sb_info *sbi); 3329 int f2fs_disable_cp_again(struct f2fs_sb_info *sbi, block_t unusable); 3330 void f2fs_release_discard_addrs(struct f2fs_sb_info *sbi); 3331 int f2fs_npages_for_summary_flush(struct f2fs_sb_info *sbi, bool for_ra); 3332 void f2fs_init_inmem_curseg(struct f2fs_sb_info *sbi); 3333 void f2fs_save_inmem_curseg(struct f2fs_sb_info *sbi); 3334 void f2fs_restore_inmem_curseg(struct f2fs_sb_info *sbi); 3335 void f2fs_get_new_segment(struct f2fs_sb_info *sbi, 3336 unsigned int *newseg, bool new_sec, int dir); 3337 void f2fs_allocate_segment_for_resize(struct f2fs_sb_info *sbi, int type, 3338 unsigned int start, unsigned int end); 3339 void f2fs_allocate_new_segment(struct f2fs_sb_info *sbi, int type); 3340 void f2fs_allocate_new_segments(struct f2fs_sb_info *sbi); 3341 int f2fs_trim_fs(struct f2fs_sb_info *sbi, struct fstrim_range *range); 3342 bool f2fs_exist_trim_candidates(struct f2fs_sb_info *sbi, 3343 struct cp_control *cpc); 3344 struct page *f2fs_get_sum_page(struct f2fs_sb_info *sbi, unsigned int segno); 3345 void f2fs_update_meta_page(struct f2fs_sb_info *sbi, void *src, 3346 block_t blk_addr); 3347 void f2fs_do_write_meta_page(struct f2fs_sb_info *sbi, struct page *page, 3348 enum iostat_type io_type); 3349 void f2fs_do_write_node_page(unsigned int nid, struct f2fs_io_info *fio); 3350 void f2fs_outplace_write_data(struct dnode_of_data *dn, 3351 struct f2fs_io_info *fio); 3352 int f2fs_inplace_write_data(struct f2fs_io_info *fio); 3353 void f2fs_do_replace_block(struct f2fs_sb_info *sbi, struct f2fs_summary *sum, 3354 block_t old_blkaddr, block_t new_blkaddr, 3355 bool recover_curseg, bool recover_newaddr, 3356 bool from_gc); 3357 void f2fs_replace_block(struct f2fs_sb_info *sbi, struct dnode_of_data *dn, 3358 block_t old_addr, block_t new_addr, 3359 unsigned char version, bool recover_curseg, 3360 bool recover_newaddr); 3361 void f2fs_allocate_data_block(struct f2fs_sb_info *sbi, struct page *page, 3362 block_t old_blkaddr, block_t *new_blkaddr, 3363 struct f2fs_summary *sum, int type, 3364 struct f2fs_io_info *fio); 3365 void f2fs_wait_on_page_writeback(struct page *page, 3366 enum page_type type, bool ordered, bool locked); 3367 void f2fs_wait_on_block_writeback(struct inode *inode, block_t blkaddr); 3368 void f2fs_wait_on_block_writeback_range(struct inode *inode, block_t blkaddr, 3369 block_t len); 3370 void f2fs_write_data_summaries(struct f2fs_sb_info *sbi, block_t start_blk); 3371 void f2fs_write_node_summaries(struct f2fs_sb_info *sbi, block_t start_blk); 3372 int f2fs_lookup_journal_in_cursum(struct f2fs_journal *journal, int type, 3373 unsigned int val, int alloc); 3374 void f2fs_flush_sit_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3375 int f2fs_fix_curseg_write_pointer(struct f2fs_sb_info *sbi); 3376 int f2fs_check_write_pointer(struct f2fs_sb_info *sbi); 3377 int f2fs_build_segment_manager(struct f2fs_sb_info *sbi); 3378 void f2fs_destroy_segment_manager(struct f2fs_sb_info *sbi); 3379 int __init f2fs_create_segment_manager_caches(void); 3380 void f2fs_destroy_segment_manager_caches(void); 3381 int f2fs_rw_hint_to_seg_type(enum rw_hint hint); 3382 enum rw_hint f2fs_io_type_to_rw_hint(struct f2fs_sb_info *sbi, 3383 enum page_type type, enum temp_type temp); 3384 unsigned int f2fs_usable_segs_in_sec(struct f2fs_sb_info *sbi, 3385 unsigned int segno); 3386 unsigned int f2fs_usable_blks_in_seg(struct f2fs_sb_info *sbi, 3387 unsigned int segno); 3388 3389 /* 3390 * checkpoint.c 3391 */ 3392 void f2fs_stop_checkpoint(struct f2fs_sb_info *sbi, bool end_io); 3393 struct page *f2fs_grab_meta_page(struct f2fs_sb_info *sbi, pgoff_t index); 3394 struct page *f2fs_get_meta_page(struct f2fs_sb_info *sbi, pgoff_t index); 3395 struct page *f2fs_get_meta_page_retry(struct f2fs_sb_info *sbi, pgoff_t index); 3396 struct page *f2fs_get_tmp_page(struct f2fs_sb_info *sbi, pgoff_t index); 3397 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi, 3398 block_t blkaddr, int type); 3399 int f2fs_ra_meta_pages(struct f2fs_sb_info *sbi, block_t start, int nrpages, 3400 int type, bool sync); 3401 void f2fs_ra_meta_pages_cond(struct f2fs_sb_info *sbi, pgoff_t index); 3402 long f2fs_sync_meta_pages(struct f2fs_sb_info *sbi, enum page_type type, 3403 long nr_to_write, enum iostat_type io_type); 3404 void f2fs_add_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type); 3405 void f2fs_remove_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type); 3406 void f2fs_release_ino_entry(struct f2fs_sb_info *sbi, bool all); 3407 bool f2fs_exist_written_data(struct f2fs_sb_info *sbi, nid_t ino, int mode); 3408 void f2fs_set_dirty_device(struct f2fs_sb_info *sbi, nid_t ino, 3409 unsigned int devidx, int type); 3410 bool f2fs_is_dirty_device(struct f2fs_sb_info *sbi, nid_t ino, 3411 unsigned int devidx, int type); 3412 int f2fs_sync_inode_meta(struct f2fs_sb_info *sbi); 3413 int f2fs_acquire_orphan_inode(struct f2fs_sb_info *sbi); 3414 void f2fs_release_orphan_inode(struct f2fs_sb_info *sbi); 3415 void f2fs_add_orphan_inode(struct inode *inode); 3416 void f2fs_remove_orphan_inode(struct f2fs_sb_info *sbi, nid_t ino); 3417 int f2fs_recover_orphan_inodes(struct f2fs_sb_info *sbi); 3418 int f2fs_get_valid_checkpoint(struct f2fs_sb_info *sbi); 3419 void f2fs_update_dirty_page(struct inode *inode, struct page *page); 3420 void f2fs_remove_dirty_inode(struct inode *inode); 3421 int f2fs_sync_dirty_inodes(struct f2fs_sb_info *sbi, enum inode_type type); 3422 void f2fs_wait_on_all_pages(struct f2fs_sb_info *sbi, int type); 3423 u64 f2fs_get_sectors_written(struct f2fs_sb_info *sbi); 3424 int f2fs_write_checkpoint(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3425 void f2fs_init_ino_entry_info(struct f2fs_sb_info *sbi); 3426 int __init f2fs_create_checkpoint_caches(void); 3427 void f2fs_destroy_checkpoint_caches(void); 3428 3429 /* 3430 * data.c 3431 */ 3432 int __init f2fs_init_bioset(void); 3433 void f2fs_destroy_bioset(void); 3434 struct bio *f2fs_bio_alloc(struct f2fs_sb_info *sbi, int npages, bool noio); 3435 int f2fs_init_bio_entry_cache(void); 3436 void f2fs_destroy_bio_entry_cache(void); 3437 void f2fs_submit_bio(struct f2fs_sb_info *sbi, 3438 struct bio *bio, enum page_type type); 3439 void f2fs_submit_merged_write(struct f2fs_sb_info *sbi, enum page_type type); 3440 void f2fs_submit_merged_write_cond(struct f2fs_sb_info *sbi, 3441 struct inode *inode, struct page *page, 3442 nid_t ino, enum page_type type); 3443 void f2fs_submit_merged_ipu_write(struct f2fs_sb_info *sbi, 3444 struct bio **bio, struct page *page); 3445 void f2fs_flush_merged_writes(struct f2fs_sb_info *sbi); 3446 int f2fs_submit_page_bio(struct f2fs_io_info *fio); 3447 int f2fs_merge_page_bio(struct f2fs_io_info *fio); 3448 void f2fs_submit_page_write(struct f2fs_io_info *fio); 3449 struct block_device *f2fs_target_device(struct f2fs_sb_info *sbi, 3450 block_t blk_addr, struct bio *bio); 3451 int f2fs_target_device_index(struct f2fs_sb_info *sbi, block_t blkaddr); 3452 void f2fs_set_data_blkaddr(struct dnode_of_data *dn); 3453 void f2fs_update_data_blkaddr(struct dnode_of_data *dn, block_t blkaddr); 3454 int f2fs_reserve_new_blocks(struct dnode_of_data *dn, blkcnt_t count); 3455 int f2fs_reserve_new_block(struct dnode_of_data *dn); 3456 int f2fs_get_block(struct dnode_of_data *dn, pgoff_t index); 3457 int f2fs_preallocate_blocks(struct kiocb *iocb, struct iov_iter *from); 3458 int f2fs_reserve_block(struct dnode_of_data *dn, pgoff_t index); 3459 struct page *f2fs_get_read_data_page(struct inode *inode, pgoff_t index, 3460 int op_flags, bool for_write); 3461 struct page *f2fs_find_data_page(struct inode *inode, pgoff_t index); 3462 struct page *f2fs_get_lock_data_page(struct inode *inode, pgoff_t index, 3463 bool for_write); 3464 struct page *f2fs_get_new_data_page(struct inode *inode, 3465 struct page *ipage, pgoff_t index, bool new_i_size); 3466 int f2fs_do_write_data_page(struct f2fs_io_info *fio); 3467 void f2fs_do_map_lock(struct f2fs_sb_info *sbi, int flag, bool lock); 3468 int f2fs_map_blocks(struct inode *inode, struct f2fs_map_blocks *map, 3469 int create, int flag); 3470 int f2fs_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, 3471 u64 start, u64 len); 3472 int f2fs_encrypt_one_page(struct f2fs_io_info *fio); 3473 bool f2fs_should_update_inplace(struct inode *inode, struct f2fs_io_info *fio); 3474 bool f2fs_should_update_outplace(struct inode *inode, struct f2fs_io_info *fio); 3475 int f2fs_write_single_data_page(struct page *page, int *submitted, 3476 struct bio **bio, sector_t *last_block, 3477 struct writeback_control *wbc, 3478 enum iostat_type io_type, 3479 int compr_blocks); 3480 void f2fs_invalidate_page(struct page *page, unsigned int offset, 3481 unsigned int length); 3482 int f2fs_release_page(struct page *page, gfp_t wait); 3483 #ifdef CONFIG_MIGRATION 3484 int f2fs_migrate_page(struct address_space *mapping, struct page *newpage, 3485 struct page *page, enum migrate_mode mode); 3486 #endif 3487 bool f2fs_overwrite_io(struct inode *inode, loff_t pos, size_t len); 3488 void f2fs_clear_page_cache_dirty_tag(struct page *page); 3489 int f2fs_init_post_read_processing(void); 3490 void f2fs_destroy_post_read_processing(void); 3491 int f2fs_init_post_read_wq(struct f2fs_sb_info *sbi); 3492 void f2fs_destroy_post_read_wq(struct f2fs_sb_info *sbi); 3493 3494 /* 3495 * gc.c 3496 */ 3497 int f2fs_start_gc_thread(struct f2fs_sb_info *sbi); 3498 void f2fs_stop_gc_thread(struct f2fs_sb_info *sbi); 3499 block_t f2fs_start_bidx_of_node(unsigned int node_ofs, struct inode *inode); 3500 int f2fs_gc(struct f2fs_sb_info *sbi, bool sync, bool background, 3501 unsigned int segno); 3502 void f2fs_build_gc_manager(struct f2fs_sb_info *sbi); 3503 int f2fs_resize_fs(struct f2fs_sb_info *sbi, __u64 block_count); 3504 int __init f2fs_create_garbage_collection_cache(void); 3505 void f2fs_destroy_garbage_collection_cache(void); 3506 3507 /* 3508 * recovery.c 3509 */ 3510 int f2fs_recover_fsync_data(struct f2fs_sb_info *sbi, bool check_only); 3511 bool f2fs_space_for_roll_forward(struct f2fs_sb_info *sbi); 3512 3513 /* 3514 * debug.c 3515 */ 3516 #ifdef CONFIG_F2FS_STAT_FS 3517 struct f2fs_stat_info { 3518 struct list_head stat_list; 3519 struct f2fs_sb_info *sbi; 3520 int all_area_segs, sit_area_segs, nat_area_segs, ssa_area_segs; 3521 int main_area_segs, main_area_sections, main_area_zones; 3522 unsigned long long hit_largest, hit_cached, hit_rbtree; 3523 unsigned long long hit_total, total_ext; 3524 int ext_tree, zombie_tree, ext_node; 3525 int ndirty_node, ndirty_dent, ndirty_meta, ndirty_imeta; 3526 int ndirty_data, ndirty_qdata; 3527 int inmem_pages; 3528 unsigned int ndirty_dirs, ndirty_files, nquota_files, ndirty_all; 3529 int nats, dirty_nats, sits, dirty_sits; 3530 int free_nids, avail_nids, alloc_nids; 3531 int total_count, utilization; 3532 int bg_gc, nr_wb_cp_data, nr_wb_data; 3533 int nr_rd_data, nr_rd_node, nr_rd_meta; 3534 int nr_dio_read, nr_dio_write; 3535 unsigned int io_skip_bggc, other_skip_bggc; 3536 int nr_flushing, nr_flushed, flush_list_empty; 3537 int nr_discarding, nr_discarded; 3538 int nr_discard_cmd; 3539 unsigned int undiscard_blks; 3540 int inline_xattr, inline_inode, inline_dir, append, update, orphans; 3541 int compr_inode; 3542 unsigned long long compr_blocks; 3543 int aw_cnt, max_aw_cnt, vw_cnt, max_vw_cnt; 3544 unsigned int valid_count, valid_node_count, valid_inode_count, discard_blks; 3545 unsigned int bimodal, avg_vblocks; 3546 int util_free, util_valid, util_invalid; 3547 int rsvd_segs, overp_segs; 3548 int dirty_count, node_pages, meta_pages; 3549 int prefree_count, call_count, cp_count, bg_cp_count; 3550 int tot_segs, node_segs, data_segs, free_segs, free_secs; 3551 int bg_node_segs, bg_data_segs; 3552 int tot_blks, data_blks, node_blks; 3553 int bg_data_blks, bg_node_blks; 3554 unsigned long long skipped_atomic_files[2]; 3555 int curseg[NR_CURSEG_TYPE]; 3556 int cursec[NR_CURSEG_TYPE]; 3557 int curzone[NR_CURSEG_TYPE]; 3558 unsigned int dirty_seg[NR_CURSEG_TYPE]; 3559 unsigned int full_seg[NR_CURSEG_TYPE]; 3560 unsigned int valid_blks[NR_CURSEG_TYPE]; 3561 3562 unsigned int meta_count[META_MAX]; 3563 unsigned int segment_count[2]; 3564 unsigned int block_count[2]; 3565 unsigned int inplace_count; 3566 unsigned long long base_mem, cache_mem, page_mem; 3567 }; 3568 3569 static inline struct f2fs_stat_info *F2FS_STAT(struct f2fs_sb_info *sbi) 3570 { 3571 return (struct f2fs_stat_info *)sbi->stat_info; 3572 } 3573 3574 #define stat_inc_cp_count(si) ((si)->cp_count++) 3575 #define stat_inc_bg_cp_count(si) ((si)->bg_cp_count++) 3576 #define stat_inc_call_count(si) ((si)->call_count++) 3577 #define stat_inc_bggc_count(si) ((si)->bg_gc++) 3578 #define stat_io_skip_bggc_count(sbi) ((sbi)->io_skip_bggc++) 3579 #define stat_other_skip_bggc_count(sbi) ((sbi)->other_skip_bggc++) 3580 #define stat_inc_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]++) 3581 #define stat_dec_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]--) 3582 #define stat_inc_total_hit(sbi) (atomic64_inc(&(sbi)->total_hit_ext)) 3583 #define stat_inc_rbtree_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_rbtree)) 3584 #define stat_inc_largest_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_largest)) 3585 #define stat_inc_cached_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_cached)) 3586 #define stat_inc_inline_xattr(inode) \ 3587 do { \ 3588 if (f2fs_has_inline_xattr(inode)) \ 3589 (atomic_inc(&F2FS_I_SB(inode)->inline_xattr)); \ 3590 } while (0) 3591 #define stat_dec_inline_xattr(inode) \ 3592 do { \ 3593 if (f2fs_has_inline_xattr(inode)) \ 3594 (atomic_dec(&F2FS_I_SB(inode)->inline_xattr)); \ 3595 } while (0) 3596 #define stat_inc_inline_inode(inode) \ 3597 do { \ 3598 if (f2fs_has_inline_data(inode)) \ 3599 (atomic_inc(&F2FS_I_SB(inode)->inline_inode)); \ 3600 } while (0) 3601 #define stat_dec_inline_inode(inode) \ 3602 do { \ 3603 if (f2fs_has_inline_data(inode)) \ 3604 (atomic_dec(&F2FS_I_SB(inode)->inline_inode)); \ 3605 } while (0) 3606 #define stat_inc_inline_dir(inode) \ 3607 do { \ 3608 if (f2fs_has_inline_dentry(inode)) \ 3609 (atomic_inc(&F2FS_I_SB(inode)->inline_dir)); \ 3610 } while (0) 3611 #define stat_dec_inline_dir(inode) \ 3612 do { \ 3613 if (f2fs_has_inline_dentry(inode)) \ 3614 (atomic_dec(&F2FS_I_SB(inode)->inline_dir)); \ 3615 } while (0) 3616 #define stat_inc_compr_inode(inode) \ 3617 do { \ 3618 if (f2fs_compressed_file(inode)) \ 3619 (atomic_inc(&F2FS_I_SB(inode)->compr_inode)); \ 3620 } while (0) 3621 #define stat_dec_compr_inode(inode) \ 3622 do { \ 3623 if (f2fs_compressed_file(inode)) \ 3624 (atomic_dec(&F2FS_I_SB(inode)->compr_inode)); \ 3625 } while (0) 3626 #define stat_add_compr_blocks(inode, blocks) \ 3627 (atomic64_add(blocks, &F2FS_I_SB(inode)->compr_blocks)) 3628 #define stat_sub_compr_blocks(inode, blocks) \ 3629 (atomic64_sub(blocks, &F2FS_I_SB(inode)->compr_blocks)) 3630 #define stat_inc_meta_count(sbi, blkaddr) \ 3631 do { \ 3632 if (blkaddr < SIT_I(sbi)->sit_base_addr) \ 3633 atomic_inc(&(sbi)->meta_count[META_CP]); \ 3634 else if (blkaddr < NM_I(sbi)->nat_blkaddr) \ 3635 atomic_inc(&(sbi)->meta_count[META_SIT]); \ 3636 else if (blkaddr < SM_I(sbi)->ssa_blkaddr) \ 3637 atomic_inc(&(sbi)->meta_count[META_NAT]); \ 3638 else if (blkaddr < SM_I(sbi)->main_blkaddr) \ 3639 atomic_inc(&(sbi)->meta_count[META_SSA]); \ 3640 } while (0) 3641 #define stat_inc_seg_type(sbi, curseg) \ 3642 ((sbi)->segment_count[(curseg)->alloc_type]++) 3643 #define stat_inc_block_count(sbi, curseg) \ 3644 ((sbi)->block_count[(curseg)->alloc_type]++) 3645 #define stat_inc_inplace_blocks(sbi) \ 3646 (atomic_inc(&(sbi)->inplace_count)) 3647 #define stat_update_max_atomic_write(inode) \ 3648 do { \ 3649 int cur = F2FS_I_SB(inode)->atomic_files; \ 3650 int max = atomic_read(&F2FS_I_SB(inode)->max_aw_cnt); \ 3651 if (cur > max) \ 3652 atomic_set(&F2FS_I_SB(inode)->max_aw_cnt, cur); \ 3653 } while (0) 3654 #define stat_inc_volatile_write(inode) \ 3655 (atomic_inc(&F2FS_I_SB(inode)->vw_cnt)) 3656 #define stat_dec_volatile_write(inode) \ 3657 (atomic_dec(&F2FS_I_SB(inode)->vw_cnt)) 3658 #define stat_update_max_volatile_write(inode) \ 3659 do { \ 3660 int cur = atomic_read(&F2FS_I_SB(inode)->vw_cnt); \ 3661 int max = atomic_read(&F2FS_I_SB(inode)->max_vw_cnt); \ 3662 if (cur > max) \ 3663 atomic_set(&F2FS_I_SB(inode)->max_vw_cnt, cur); \ 3664 } while (0) 3665 #define stat_inc_seg_count(sbi, type, gc_type) \ 3666 do { \ 3667 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 3668 si->tot_segs++; \ 3669 if ((type) == SUM_TYPE_DATA) { \ 3670 si->data_segs++; \ 3671 si->bg_data_segs += (gc_type == BG_GC) ? 1 : 0; \ 3672 } else { \ 3673 si->node_segs++; \ 3674 si->bg_node_segs += (gc_type == BG_GC) ? 1 : 0; \ 3675 } \ 3676 } while (0) 3677 3678 #define stat_inc_tot_blk_count(si, blks) \ 3679 ((si)->tot_blks += (blks)) 3680 3681 #define stat_inc_data_blk_count(sbi, blks, gc_type) \ 3682 do { \ 3683 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 3684 stat_inc_tot_blk_count(si, blks); \ 3685 si->data_blks += (blks); \ 3686 si->bg_data_blks += ((gc_type) == BG_GC) ? (blks) : 0; \ 3687 } while (0) 3688 3689 #define stat_inc_node_blk_count(sbi, blks, gc_type) \ 3690 do { \ 3691 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 3692 stat_inc_tot_blk_count(si, blks); \ 3693 si->node_blks += (blks); \ 3694 si->bg_node_blks += ((gc_type) == BG_GC) ? (blks) : 0; \ 3695 } while (0) 3696 3697 int f2fs_build_stats(struct f2fs_sb_info *sbi); 3698 void f2fs_destroy_stats(struct f2fs_sb_info *sbi); 3699 void __init f2fs_create_root_stats(void); 3700 void f2fs_destroy_root_stats(void); 3701 void f2fs_update_sit_info(struct f2fs_sb_info *sbi); 3702 #else 3703 #define stat_inc_cp_count(si) do { } while (0) 3704 #define stat_inc_bg_cp_count(si) do { } while (0) 3705 #define stat_inc_call_count(si) do { } while (0) 3706 #define stat_inc_bggc_count(si) do { } while (0) 3707 #define stat_io_skip_bggc_count(sbi) do { } while (0) 3708 #define stat_other_skip_bggc_count(sbi) do { } while (0) 3709 #define stat_inc_dirty_inode(sbi, type) do { } while (0) 3710 #define stat_dec_dirty_inode(sbi, type) do { } while (0) 3711 #define stat_inc_total_hit(sbi) do { } while (0) 3712 #define stat_inc_rbtree_node_hit(sbi) do { } while (0) 3713 #define stat_inc_largest_node_hit(sbi) do { } while (0) 3714 #define stat_inc_cached_node_hit(sbi) do { } while (0) 3715 #define stat_inc_inline_xattr(inode) do { } while (0) 3716 #define stat_dec_inline_xattr(inode) do { } while (0) 3717 #define stat_inc_inline_inode(inode) do { } while (0) 3718 #define stat_dec_inline_inode(inode) do { } while (0) 3719 #define stat_inc_inline_dir(inode) do { } while (0) 3720 #define stat_dec_inline_dir(inode) do { } while (0) 3721 #define stat_inc_compr_inode(inode) do { } while (0) 3722 #define stat_dec_compr_inode(inode) do { } while (0) 3723 #define stat_add_compr_blocks(inode, blocks) do { } while (0) 3724 #define stat_sub_compr_blocks(inode, blocks) do { } while (0) 3725 #define stat_inc_atomic_write(inode) do { } while (0) 3726 #define stat_dec_atomic_write(inode) do { } while (0) 3727 #define stat_update_max_atomic_write(inode) do { } while (0) 3728 #define stat_inc_volatile_write(inode) do { } while (0) 3729 #define stat_dec_volatile_write(inode) do { } while (0) 3730 #define stat_update_max_volatile_write(inode) do { } while (0) 3731 #define stat_inc_meta_count(sbi, blkaddr) do { } while (0) 3732 #define stat_inc_seg_type(sbi, curseg) do { } while (0) 3733 #define stat_inc_block_count(sbi, curseg) do { } while (0) 3734 #define stat_inc_inplace_blocks(sbi) do { } while (0) 3735 #define stat_inc_seg_count(sbi, type, gc_type) do { } while (0) 3736 #define stat_inc_tot_blk_count(si, blks) do { } while (0) 3737 #define stat_inc_data_blk_count(sbi, blks, gc_type) do { } while (0) 3738 #define stat_inc_node_blk_count(sbi, blks, gc_type) do { } while (0) 3739 3740 static inline int f2fs_build_stats(struct f2fs_sb_info *sbi) { return 0; } 3741 static inline void f2fs_destroy_stats(struct f2fs_sb_info *sbi) { } 3742 static inline void __init f2fs_create_root_stats(void) { } 3743 static inline void f2fs_destroy_root_stats(void) { } 3744 static inline void f2fs_update_sit_info(struct f2fs_sb_info *sbi) {} 3745 #endif 3746 3747 extern const struct file_operations f2fs_dir_operations; 3748 extern const struct file_operations f2fs_file_operations; 3749 extern const struct inode_operations f2fs_file_inode_operations; 3750 extern const struct address_space_operations f2fs_dblock_aops; 3751 extern const struct address_space_operations f2fs_node_aops; 3752 extern const struct address_space_operations f2fs_meta_aops; 3753 extern const struct inode_operations f2fs_dir_inode_operations; 3754 extern const struct inode_operations f2fs_symlink_inode_operations; 3755 extern const struct inode_operations f2fs_encrypted_symlink_inode_operations; 3756 extern const struct inode_operations f2fs_special_inode_operations; 3757 extern struct kmem_cache *f2fs_inode_entry_slab; 3758 3759 /* 3760 * inline.c 3761 */ 3762 bool f2fs_may_inline_data(struct inode *inode); 3763 bool f2fs_may_inline_dentry(struct inode *inode); 3764 void f2fs_do_read_inline_data(struct page *page, struct page *ipage); 3765 void f2fs_truncate_inline_inode(struct inode *inode, 3766 struct page *ipage, u64 from); 3767 int f2fs_read_inline_data(struct inode *inode, struct page *page); 3768 int f2fs_convert_inline_page(struct dnode_of_data *dn, struct page *page); 3769 int f2fs_convert_inline_inode(struct inode *inode); 3770 int f2fs_try_convert_inline_dir(struct inode *dir, struct dentry *dentry); 3771 int f2fs_write_inline_data(struct inode *inode, struct page *page); 3772 int f2fs_recover_inline_data(struct inode *inode, struct page *npage); 3773 struct f2fs_dir_entry *f2fs_find_in_inline_dir(struct inode *dir, 3774 const struct f2fs_filename *fname, 3775 struct page **res_page); 3776 int f2fs_make_empty_inline_dir(struct inode *inode, struct inode *parent, 3777 struct page *ipage); 3778 int f2fs_add_inline_entry(struct inode *dir, const struct f2fs_filename *fname, 3779 struct inode *inode, nid_t ino, umode_t mode); 3780 void f2fs_delete_inline_entry(struct f2fs_dir_entry *dentry, 3781 struct page *page, struct inode *dir, 3782 struct inode *inode); 3783 bool f2fs_empty_inline_dir(struct inode *dir); 3784 int f2fs_read_inline_dir(struct file *file, struct dir_context *ctx, 3785 struct fscrypt_str *fstr); 3786 int f2fs_inline_data_fiemap(struct inode *inode, 3787 struct fiemap_extent_info *fieinfo, 3788 __u64 start, __u64 len); 3789 3790 /* 3791 * shrinker.c 3792 */ 3793 unsigned long f2fs_shrink_count(struct shrinker *shrink, 3794 struct shrink_control *sc); 3795 unsigned long f2fs_shrink_scan(struct shrinker *shrink, 3796 struct shrink_control *sc); 3797 void f2fs_join_shrinker(struct f2fs_sb_info *sbi); 3798 void f2fs_leave_shrinker(struct f2fs_sb_info *sbi); 3799 3800 /* 3801 * extent_cache.c 3802 */ 3803 struct rb_entry *f2fs_lookup_rb_tree(struct rb_root_cached *root, 3804 struct rb_entry *cached_re, unsigned int ofs); 3805 struct rb_node **f2fs_lookup_rb_tree_ext(struct f2fs_sb_info *sbi, 3806 struct rb_root_cached *root, 3807 struct rb_node **parent, 3808 unsigned long long key, bool *left_most); 3809 struct rb_node **f2fs_lookup_rb_tree_for_insert(struct f2fs_sb_info *sbi, 3810 struct rb_root_cached *root, 3811 struct rb_node **parent, 3812 unsigned int ofs, bool *leftmost); 3813 struct rb_entry *f2fs_lookup_rb_tree_ret(struct rb_root_cached *root, 3814 struct rb_entry *cached_re, unsigned int ofs, 3815 struct rb_entry **prev_entry, struct rb_entry **next_entry, 3816 struct rb_node ***insert_p, struct rb_node **insert_parent, 3817 bool force, bool *leftmost); 3818 bool f2fs_check_rb_tree_consistence(struct f2fs_sb_info *sbi, 3819 struct rb_root_cached *root, bool check_key); 3820 unsigned int f2fs_shrink_extent_tree(struct f2fs_sb_info *sbi, int nr_shrink); 3821 void f2fs_init_extent_tree(struct inode *inode, struct page *ipage); 3822 void f2fs_drop_extent_tree(struct inode *inode); 3823 unsigned int f2fs_destroy_extent_node(struct inode *inode); 3824 void f2fs_destroy_extent_tree(struct inode *inode); 3825 bool f2fs_lookup_extent_cache(struct inode *inode, pgoff_t pgofs, 3826 struct extent_info *ei); 3827 void f2fs_update_extent_cache(struct dnode_of_data *dn); 3828 void f2fs_update_extent_cache_range(struct dnode_of_data *dn, 3829 pgoff_t fofs, block_t blkaddr, unsigned int len); 3830 void f2fs_init_extent_cache_info(struct f2fs_sb_info *sbi); 3831 int __init f2fs_create_extent_cache(void); 3832 void f2fs_destroy_extent_cache(void); 3833 3834 /* 3835 * sysfs.c 3836 */ 3837 int __init f2fs_init_sysfs(void); 3838 void f2fs_exit_sysfs(void); 3839 int f2fs_register_sysfs(struct f2fs_sb_info *sbi); 3840 void f2fs_unregister_sysfs(struct f2fs_sb_info *sbi); 3841 3842 /* verity.c */ 3843 extern const struct fsverity_operations f2fs_verityops; 3844 3845 /* 3846 * crypto support 3847 */ 3848 static inline bool f2fs_encrypted_file(struct inode *inode) 3849 { 3850 return IS_ENCRYPTED(inode) && S_ISREG(inode->i_mode); 3851 } 3852 3853 static inline void f2fs_set_encrypted_inode(struct inode *inode) 3854 { 3855 #ifdef CONFIG_FS_ENCRYPTION 3856 file_set_encrypt(inode); 3857 f2fs_set_inode_flags(inode); 3858 #endif 3859 } 3860 3861 /* 3862 * Returns true if the reads of the inode's data need to undergo some 3863 * postprocessing step, like decryption or authenticity verification. 3864 */ 3865 static inline bool f2fs_post_read_required(struct inode *inode) 3866 { 3867 return f2fs_encrypted_file(inode) || fsverity_active(inode) || 3868 f2fs_compressed_file(inode); 3869 } 3870 3871 /* 3872 * compress.c 3873 */ 3874 #ifdef CONFIG_F2FS_FS_COMPRESSION 3875 bool f2fs_is_compressed_page(struct page *page); 3876 struct page *f2fs_compress_control_page(struct page *page); 3877 int f2fs_prepare_compress_overwrite(struct inode *inode, 3878 struct page **pagep, pgoff_t index, void **fsdata); 3879 bool f2fs_compress_write_end(struct inode *inode, void *fsdata, 3880 pgoff_t index, unsigned copied); 3881 int f2fs_truncate_partial_cluster(struct inode *inode, u64 from, bool lock); 3882 void f2fs_compress_write_end_io(struct bio *bio, struct page *page); 3883 bool f2fs_is_compress_backend_ready(struct inode *inode); 3884 int f2fs_init_compress_mempool(void); 3885 void f2fs_destroy_compress_mempool(void); 3886 void f2fs_decompress_pages(struct bio *bio, struct page *page, bool verity); 3887 bool f2fs_cluster_is_empty(struct compress_ctx *cc); 3888 bool f2fs_cluster_can_merge_page(struct compress_ctx *cc, pgoff_t index); 3889 void f2fs_compress_ctx_add_page(struct compress_ctx *cc, struct page *page); 3890 int f2fs_write_multi_pages(struct compress_ctx *cc, 3891 int *submitted, 3892 struct writeback_control *wbc, 3893 enum iostat_type io_type); 3894 int f2fs_is_compressed_cluster(struct inode *inode, pgoff_t index); 3895 int f2fs_read_multi_pages(struct compress_ctx *cc, struct bio **bio_ret, 3896 unsigned nr_pages, sector_t *last_block_in_bio, 3897 bool is_readahead, bool for_write); 3898 struct decompress_io_ctx *f2fs_alloc_dic(struct compress_ctx *cc); 3899 void f2fs_free_dic(struct decompress_io_ctx *dic); 3900 void f2fs_decompress_end_io(struct page **rpages, 3901 unsigned int cluster_size, bool err, bool verity); 3902 int f2fs_init_compress_ctx(struct compress_ctx *cc); 3903 void f2fs_destroy_compress_ctx(struct compress_ctx *cc); 3904 void f2fs_init_compress_info(struct f2fs_sb_info *sbi); 3905 int f2fs_init_page_array_cache(struct f2fs_sb_info *sbi); 3906 void f2fs_destroy_page_array_cache(struct f2fs_sb_info *sbi); 3907 int __init f2fs_init_compress_cache(void); 3908 void f2fs_destroy_compress_cache(void); 3909 #else 3910 static inline bool f2fs_is_compressed_page(struct page *page) { return false; } 3911 static inline bool f2fs_is_compress_backend_ready(struct inode *inode) 3912 { 3913 if (!f2fs_compressed_file(inode)) 3914 return true; 3915 /* not support compression */ 3916 return false; 3917 } 3918 static inline struct page *f2fs_compress_control_page(struct page *page) 3919 { 3920 WARN_ON_ONCE(1); 3921 return ERR_PTR(-EINVAL); 3922 } 3923 static inline int f2fs_init_compress_mempool(void) { return 0; } 3924 static inline void f2fs_destroy_compress_mempool(void) { } 3925 static inline int f2fs_init_page_array_cache(struct f2fs_sb_info *sbi) { return 0; } 3926 static inline void f2fs_destroy_page_array_cache(struct f2fs_sb_info *sbi) { } 3927 static inline int __init f2fs_init_compress_cache(void) { return 0; } 3928 static inline void f2fs_destroy_compress_cache(void) { } 3929 #endif 3930 3931 static inline void set_compress_context(struct inode *inode) 3932 { 3933 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 3934 3935 F2FS_I(inode)->i_compress_algorithm = 3936 F2FS_OPTION(sbi).compress_algorithm; 3937 F2FS_I(inode)->i_log_cluster_size = 3938 F2FS_OPTION(sbi).compress_log_size; 3939 F2FS_I(inode)->i_compress_flag = 3940 F2FS_OPTION(sbi).compress_chksum ? 3941 1 << COMPRESS_CHKSUM : 0; 3942 F2FS_I(inode)->i_cluster_size = 3943 1 << F2FS_I(inode)->i_log_cluster_size; 3944 if (F2FS_I(inode)->i_compress_algorithm == COMPRESS_LZ4 && 3945 F2FS_OPTION(sbi).compress_level) 3946 F2FS_I(inode)->i_compress_flag |= 3947 F2FS_OPTION(sbi).compress_level << 3948 COMPRESS_LEVEL_OFFSET; 3949 F2FS_I(inode)->i_flags |= F2FS_COMPR_FL; 3950 set_inode_flag(inode, FI_COMPRESSED_FILE); 3951 stat_inc_compr_inode(inode); 3952 f2fs_mark_inode_dirty_sync(inode, true); 3953 } 3954 3955 static inline bool f2fs_disable_compressed_file(struct inode *inode) 3956 { 3957 struct f2fs_inode_info *fi = F2FS_I(inode); 3958 3959 if (!f2fs_compressed_file(inode)) 3960 return true; 3961 if (S_ISREG(inode->i_mode) && 3962 (get_dirty_pages(inode) || atomic_read(&fi->i_compr_blocks))) 3963 return false; 3964 3965 fi->i_flags &= ~F2FS_COMPR_FL; 3966 stat_dec_compr_inode(inode); 3967 clear_inode_flag(inode, FI_COMPRESSED_FILE); 3968 f2fs_mark_inode_dirty_sync(inode, true); 3969 return true; 3970 } 3971 3972 #define F2FS_FEATURE_FUNCS(name, flagname) \ 3973 static inline int f2fs_sb_has_##name(struct f2fs_sb_info *sbi) \ 3974 { \ 3975 return F2FS_HAS_FEATURE(sbi, F2FS_FEATURE_##flagname); \ 3976 } 3977 3978 F2FS_FEATURE_FUNCS(encrypt, ENCRYPT); 3979 F2FS_FEATURE_FUNCS(blkzoned, BLKZONED); 3980 F2FS_FEATURE_FUNCS(extra_attr, EXTRA_ATTR); 3981 F2FS_FEATURE_FUNCS(project_quota, PRJQUOTA); 3982 F2FS_FEATURE_FUNCS(inode_chksum, INODE_CHKSUM); 3983 F2FS_FEATURE_FUNCS(flexible_inline_xattr, FLEXIBLE_INLINE_XATTR); 3984 F2FS_FEATURE_FUNCS(quota_ino, QUOTA_INO); 3985 F2FS_FEATURE_FUNCS(inode_crtime, INODE_CRTIME); 3986 F2FS_FEATURE_FUNCS(lost_found, LOST_FOUND); 3987 F2FS_FEATURE_FUNCS(verity, VERITY); 3988 F2FS_FEATURE_FUNCS(sb_chksum, SB_CHKSUM); 3989 F2FS_FEATURE_FUNCS(casefold, CASEFOLD); 3990 F2FS_FEATURE_FUNCS(compression, COMPRESSION); 3991 3992 #ifdef CONFIG_BLK_DEV_ZONED 3993 static inline bool f2fs_blkz_is_seq(struct f2fs_sb_info *sbi, int devi, 3994 block_t blkaddr) 3995 { 3996 unsigned int zno = blkaddr >> sbi->log_blocks_per_blkz; 3997 3998 return test_bit(zno, FDEV(devi).blkz_seq); 3999 } 4000 #endif 4001 4002 static inline bool f2fs_hw_should_discard(struct f2fs_sb_info *sbi) 4003 { 4004 return f2fs_sb_has_blkzoned(sbi); 4005 } 4006 4007 static inline bool f2fs_bdev_support_discard(struct block_device *bdev) 4008 { 4009 return blk_queue_discard(bdev_get_queue(bdev)) || 4010 bdev_is_zoned(bdev); 4011 } 4012 4013 static inline bool f2fs_hw_support_discard(struct f2fs_sb_info *sbi) 4014 { 4015 int i; 4016 4017 if (!f2fs_is_multi_device(sbi)) 4018 return f2fs_bdev_support_discard(sbi->sb->s_bdev); 4019 4020 for (i = 0; i < sbi->s_ndevs; i++) 4021 if (f2fs_bdev_support_discard(FDEV(i).bdev)) 4022 return true; 4023 return false; 4024 } 4025 4026 static inline bool f2fs_realtime_discard_enable(struct f2fs_sb_info *sbi) 4027 { 4028 return (test_opt(sbi, DISCARD) && f2fs_hw_support_discard(sbi)) || 4029 f2fs_hw_should_discard(sbi); 4030 } 4031 4032 static inline bool f2fs_hw_is_readonly(struct f2fs_sb_info *sbi) 4033 { 4034 int i; 4035 4036 if (!f2fs_is_multi_device(sbi)) 4037 return bdev_read_only(sbi->sb->s_bdev); 4038 4039 for (i = 0; i < sbi->s_ndevs; i++) 4040 if (bdev_read_only(FDEV(i).bdev)) 4041 return true; 4042 return false; 4043 } 4044 4045 static inline bool f2fs_lfs_mode(struct f2fs_sb_info *sbi) 4046 { 4047 return F2FS_OPTION(sbi).fs_mode == FS_MODE_LFS; 4048 } 4049 4050 static inline bool f2fs_may_compress(struct inode *inode) 4051 { 4052 if (IS_SWAPFILE(inode) || f2fs_is_pinned_file(inode) || 4053 f2fs_is_atomic_file(inode) || 4054 f2fs_is_volatile_file(inode)) 4055 return false; 4056 return S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode); 4057 } 4058 4059 static inline void f2fs_i_compr_blocks_update(struct inode *inode, 4060 u64 blocks, bool add) 4061 { 4062 int diff = F2FS_I(inode)->i_cluster_size - blocks; 4063 struct f2fs_inode_info *fi = F2FS_I(inode); 4064 4065 /* don't update i_compr_blocks if saved blocks were released */ 4066 if (!add && !atomic_read(&fi->i_compr_blocks)) 4067 return; 4068 4069 if (add) { 4070 atomic_add(diff, &fi->i_compr_blocks); 4071 stat_add_compr_blocks(inode, diff); 4072 } else { 4073 atomic_sub(diff, &fi->i_compr_blocks); 4074 stat_sub_compr_blocks(inode, diff); 4075 } 4076 f2fs_mark_inode_dirty_sync(inode, true); 4077 } 4078 4079 static inline int block_unaligned_IO(struct inode *inode, 4080 struct kiocb *iocb, struct iov_iter *iter) 4081 { 4082 unsigned int i_blkbits = READ_ONCE(inode->i_blkbits); 4083 unsigned int blocksize_mask = (1 << i_blkbits) - 1; 4084 loff_t offset = iocb->ki_pos; 4085 unsigned long align = offset | iov_iter_alignment(iter); 4086 4087 return align & blocksize_mask; 4088 } 4089 4090 static inline int allow_outplace_dio(struct inode *inode, 4091 struct kiocb *iocb, struct iov_iter *iter) 4092 { 4093 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 4094 int rw = iov_iter_rw(iter); 4095 4096 return (f2fs_lfs_mode(sbi) && (rw == WRITE) && 4097 !block_unaligned_IO(inode, iocb, iter)); 4098 } 4099 4100 static inline bool f2fs_force_buffered_io(struct inode *inode, 4101 struct kiocb *iocb, struct iov_iter *iter) 4102 { 4103 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 4104 int rw = iov_iter_rw(iter); 4105 4106 if (f2fs_post_read_required(inode)) 4107 return true; 4108 if (f2fs_is_multi_device(sbi)) 4109 return true; 4110 /* 4111 * for blkzoned device, fallback direct IO to buffered IO, so 4112 * all IOs can be serialized by log-structured write. 4113 */ 4114 if (f2fs_sb_has_blkzoned(sbi)) 4115 return true; 4116 if (f2fs_lfs_mode(sbi) && (rw == WRITE)) { 4117 if (block_unaligned_IO(inode, iocb, iter)) 4118 return true; 4119 if (F2FS_IO_ALIGNED(sbi)) 4120 return true; 4121 } 4122 if (is_sbi_flag_set(F2FS_I_SB(inode), SBI_CP_DISABLED) && 4123 !IS_SWAPFILE(inode)) 4124 return true; 4125 4126 return false; 4127 } 4128 4129 #ifdef CONFIG_F2FS_FAULT_INJECTION 4130 extern void f2fs_build_fault_attr(struct f2fs_sb_info *sbi, unsigned int rate, 4131 unsigned int type); 4132 #else 4133 #define f2fs_build_fault_attr(sbi, rate, type) do { } while (0) 4134 #endif 4135 4136 static inline bool is_journalled_quota(struct f2fs_sb_info *sbi) 4137 { 4138 #ifdef CONFIG_QUOTA 4139 if (f2fs_sb_has_quota_ino(sbi)) 4140 return true; 4141 if (F2FS_OPTION(sbi).s_qf_names[USRQUOTA] || 4142 F2FS_OPTION(sbi).s_qf_names[GRPQUOTA] || 4143 F2FS_OPTION(sbi).s_qf_names[PRJQUOTA]) 4144 return true; 4145 #endif 4146 return false; 4147 } 4148 4149 #define EFSBADCRC EBADMSG /* Bad CRC detected */ 4150 #define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */ 4151 4152 #endif /* _LINUX_F2FS_H */ 4153