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 /* Context for decompressing one cluster on the 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 1361 /* 1362 * The number of compressed pages remaining to be read in this cluster. 1363 * This is initially nr_cpages. It is decremented by 1 each time a page 1364 * has been read (or failed to be read). When it reaches 0, the cluster 1365 * is decompressed (or an error is reported). 1366 * 1367 * If an error occurs before all the pages have been submitted for I/O, 1368 * then this will never reach 0. In this case the I/O submitter is 1369 * responsible for calling f2fs_decompress_end_io() instead. 1370 */ 1371 atomic_t remaining_pages; 1372 1373 /* 1374 * Number of references to this decompress_io_ctx. 1375 * 1376 * One reference is held for I/O completion. This reference is dropped 1377 * after the pagecache pages are updated and unlocked -- either after 1378 * decompression (and verity if enabled), or after an error. 1379 * 1380 * In addition, each compressed page holds a reference while it is in a 1381 * bio. These references are necessary prevent compressed pages from 1382 * being freed while they are still in a bio. 1383 */ 1384 refcount_t refcnt; 1385 1386 bool failed; /* IO error occurred before decompression? */ 1387 bool need_verity; /* need fs-verity verification after decompression? */ 1388 void *private; /* payload buffer for specified decompression algorithm */ 1389 void *private2; /* extra payload buffer */ 1390 struct work_struct verity_work; /* work to verify the decompressed pages */ 1391 }; 1392 1393 #define NULL_CLUSTER ((unsigned int)(~0)) 1394 #define MIN_COMPRESS_LOG_SIZE 2 1395 #define MAX_COMPRESS_LOG_SIZE 8 1396 #define MAX_COMPRESS_WINDOW_SIZE(log_size) ((PAGE_SIZE) << (log_size)) 1397 1398 struct f2fs_sb_info { 1399 struct super_block *sb; /* pointer to VFS super block */ 1400 struct proc_dir_entry *s_proc; /* proc entry */ 1401 struct f2fs_super_block *raw_super; /* raw super block pointer */ 1402 struct rw_semaphore sb_lock; /* lock for raw super block */ 1403 int valid_super_block; /* valid super block no */ 1404 unsigned long s_flag; /* flags for sbi */ 1405 struct mutex writepages; /* mutex for writepages() */ 1406 1407 #ifdef CONFIG_BLK_DEV_ZONED 1408 unsigned int blocks_per_blkz; /* F2FS blocks per zone */ 1409 unsigned int log_blocks_per_blkz; /* log2 F2FS blocks per zone */ 1410 #endif 1411 1412 /* for node-related operations */ 1413 struct f2fs_nm_info *nm_info; /* node manager */ 1414 struct inode *node_inode; /* cache node blocks */ 1415 1416 /* for segment-related operations */ 1417 struct f2fs_sm_info *sm_info; /* segment manager */ 1418 1419 /* for bio operations */ 1420 struct f2fs_bio_info *write_io[NR_PAGE_TYPE]; /* for write bios */ 1421 /* keep migration IO order for LFS mode */ 1422 struct rw_semaphore io_order_lock; 1423 mempool_t *write_io_dummy; /* Dummy pages */ 1424 1425 /* for checkpoint */ 1426 struct f2fs_checkpoint *ckpt; /* raw checkpoint pointer */ 1427 int cur_cp_pack; /* remain current cp pack */ 1428 spinlock_t cp_lock; /* for flag in ckpt */ 1429 struct inode *meta_inode; /* cache meta blocks */ 1430 struct rw_semaphore cp_global_sem; /* checkpoint procedure lock */ 1431 struct rw_semaphore cp_rwsem; /* blocking FS operations */ 1432 struct rw_semaphore node_write; /* locking node writes */ 1433 struct rw_semaphore node_change; /* locking node change */ 1434 wait_queue_head_t cp_wait; 1435 unsigned long last_time[MAX_TIME]; /* to store time in jiffies */ 1436 long interval_time[MAX_TIME]; /* to store thresholds */ 1437 1438 struct inode_management im[MAX_INO_ENTRY]; /* manage inode cache */ 1439 1440 spinlock_t fsync_node_lock; /* for node entry lock */ 1441 struct list_head fsync_node_list; /* node list head */ 1442 unsigned int fsync_seg_id; /* sequence id */ 1443 unsigned int fsync_node_num; /* number of node entries */ 1444 1445 /* for orphan inode, use 0'th array */ 1446 unsigned int max_orphans; /* max orphan inodes */ 1447 1448 /* for inode management */ 1449 struct list_head inode_list[NR_INODE_TYPE]; /* dirty inode list */ 1450 spinlock_t inode_lock[NR_INODE_TYPE]; /* for dirty inode list lock */ 1451 struct mutex flush_lock; /* for flush exclusion */ 1452 1453 /* for extent tree cache */ 1454 struct radix_tree_root extent_tree_root;/* cache extent cache entries */ 1455 struct mutex extent_tree_lock; /* locking extent radix tree */ 1456 struct list_head extent_list; /* lru list for shrinker */ 1457 spinlock_t extent_lock; /* locking extent lru list */ 1458 atomic_t total_ext_tree; /* extent tree count */ 1459 struct list_head zombie_list; /* extent zombie tree list */ 1460 atomic_t total_zombie_tree; /* extent zombie tree count */ 1461 atomic_t total_ext_node; /* extent info count */ 1462 1463 /* basic filesystem units */ 1464 unsigned int log_sectors_per_block; /* log2 sectors per block */ 1465 unsigned int log_blocksize; /* log2 block size */ 1466 unsigned int blocksize; /* block size */ 1467 unsigned int root_ino_num; /* root inode number*/ 1468 unsigned int node_ino_num; /* node inode number*/ 1469 unsigned int meta_ino_num; /* meta inode number*/ 1470 unsigned int log_blocks_per_seg; /* log2 blocks per segment */ 1471 unsigned int blocks_per_seg; /* blocks per segment */ 1472 unsigned int segs_per_sec; /* segments per section */ 1473 unsigned int secs_per_zone; /* sections per zone */ 1474 unsigned int total_sections; /* total section count */ 1475 unsigned int total_node_count; /* total node block count */ 1476 unsigned int total_valid_node_count; /* valid node block count */ 1477 int dir_level; /* directory level */ 1478 int readdir_ra; /* readahead inode in readdir */ 1479 u64 max_io_bytes; /* max io bytes to merge IOs */ 1480 1481 block_t user_block_count; /* # of user blocks */ 1482 block_t total_valid_block_count; /* # of valid blocks */ 1483 block_t discard_blks; /* discard command candidats */ 1484 block_t last_valid_block_count; /* for recovery */ 1485 block_t reserved_blocks; /* configurable reserved blocks */ 1486 block_t current_reserved_blocks; /* current reserved blocks */ 1487 1488 /* Additional tracking for no checkpoint mode */ 1489 block_t unusable_block_count; /* # of blocks saved by last cp */ 1490 1491 unsigned int nquota_files; /* # of quota sysfile */ 1492 struct rw_semaphore quota_sem; /* blocking cp for flags */ 1493 1494 /* # of pages, see count_type */ 1495 atomic_t nr_pages[NR_COUNT_TYPE]; 1496 /* # of allocated blocks */ 1497 struct percpu_counter alloc_valid_block_count; 1498 1499 /* writeback control */ 1500 atomic_t wb_sync_req[META]; /* count # of WB_SYNC threads */ 1501 1502 /* valid inode count */ 1503 struct percpu_counter total_valid_inode_count; 1504 1505 struct f2fs_mount_info mount_opt; /* mount options */ 1506 1507 /* for cleaning operations */ 1508 struct rw_semaphore gc_lock; /* 1509 * semaphore for GC, avoid 1510 * race between GC and GC or CP 1511 */ 1512 struct f2fs_gc_kthread *gc_thread; /* GC thread */ 1513 struct atgc_management am; /* atgc management */ 1514 unsigned int cur_victim_sec; /* current victim section num */ 1515 unsigned int gc_mode; /* current GC state */ 1516 unsigned int next_victim_seg[2]; /* next segment in victim section */ 1517 1518 /* for skip statistic */ 1519 unsigned int atomic_files; /* # of opened atomic file */ 1520 unsigned long long skipped_atomic_files[2]; /* FG_GC and BG_GC */ 1521 unsigned long long skipped_gc_rwsem; /* FG_GC only */ 1522 1523 /* threshold for gc trials on pinned files */ 1524 u64 gc_pin_file_threshold; 1525 struct rw_semaphore pin_sem; 1526 1527 /* maximum # of trials to find a victim segment for SSR and GC */ 1528 unsigned int max_victim_search; 1529 /* migration granularity of garbage collection, unit: segment */ 1530 unsigned int migration_granularity; 1531 1532 /* 1533 * for stat information. 1534 * one is for the LFS mode, and the other is for the SSR mode. 1535 */ 1536 #ifdef CONFIG_F2FS_STAT_FS 1537 struct f2fs_stat_info *stat_info; /* FS status information */ 1538 atomic_t meta_count[META_MAX]; /* # of meta blocks */ 1539 unsigned int segment_count[2]; /* # of allocated segments */ 1540 unsigned int block_count[2]; /* # of allocated blocks */ 1541 atomic_t inplace_count; /* # of inplace update */ 1542 atomic64_t total_hit_ext; /* # of lookup extent cache */ 1543 atomic64_t read_hit_rbtree; /* # of hit rbtree extent node */ 1544 atomic64_t read_hit_largest; /* # of hit largest extent node */ 1545 atomic64_t read_hit_cached; /* # of hit cached extent node */ 1546 atomic_t inline_xattr; /* # of inline_xattr inodes */ 1547 atomic_t inline_inode; /* # of inline_data inodes */ 1548 atomic_t inline_dir; /* # of inline_dentry inodes */ 1549 atomic_t compr_inode; /* # of compressed inodes */ 1550 atomic64_t compr_blocks; /* # of compressed blocks */ 1551 atomic_t vw_cnt; /* # of volatile writes */ 1552 atomic_t max_aw_cnt; /* max # of atomic writes */ 1553 atomic_t max_vw_cnt; /* max # of volatile writes */ 1554 unsigned int io_skip_bggc; /* skip background gc for in-flight IO */ 1555 unsigned int other_skip_bggc; /* skip background gc for other reasons */ 1556 unsigned int ndirty_inode[NR_INODE_TYPE]; /* # of dirty inodes */ 1557 #endif 1558 spinlock_t stat_lock; /* lock for stat operations */ 1559 1560 /* For app/fs IO statistics */ 1561 spinlock_t iostat_lock; 1562 unsigned long long rw_iostat[NR_IO_TYPE]; 1563 unsigned long long prev_rw_iostat[NR_IO_TYPE]; 1564 bool iostat_enable; 1565 unsigned long iostat_next_period; 1566 unsigned int iostat_period_ms; 1567 1568 /* to attach REQ_META|REQ_FUA flags */ 1569 unsigned int data_io_flag; 1570 unsigned int node_io_flag; 1571 1572 /* For sysfs suppport */ 1573 struct kobject s_kobj; /* /sys/fs/f2fs/<devname> */ 1574 struct completion s_kobj_unregister; 1575 1576 struct kobject s_stat_kobj; /* /sys/fs/f2fs/<devname>/stat */ 1577 struct completion s_stat_kobj_unregister; 1578 1579 /* For shrinker support */ 1580 struct list_head s_list; 1581 int s_ndevs; /* number of devices */ 1582 struct f2fs_dev_info *devs; /* for device list */ 1583 unsigned int dirty_device; /* for checkpoint data flush */ 1584 spinlock_t dev_lock; /* protect dirty_device */ 1585 struct mutex umount_mutex; 1586 unsigned int shrinker_run_no; 1587 1588 /* For write statistics */ 1589 u64 sectors_written_start; 1590 u64 kbytes_written; 1591 1592 /* Reference to checksum algorithm driver via cryptoapi */ 1593 struct crypto_shash *s_chksum_driver; 1594 1595 /* Precomputed FS UUID checksum for seeding other checksums */ 1596 __u32 s_chksum_seed; 1597 1598 struct workqueue_struct *post_read_wq; /* post read workqueue */ 1599 1600 struct kmem_cache *inline_xattr_slab; /* inline xattr entry */ 1601 unsigned int inline_xattr_slab_size; /* default inline xattr slab size */ 1602 1603 #ifdef CONFIG_F2FS_FS_COMPRESSION 1604 struct kmem_cache *page_array_slab; /* page array entry */ 1605 unsigned int page_array_slab_size; /* default page array slab size */ 1606 #endif 1607 }; 1608 1609 struct f2fs_private_dio { 1610 struct inode *inode; 1611 void *orig_private; 1612 bio_end_io_t *orig_end_io; 1613 bool write; 1614 }; 1615 1616 #ifdef CONFIG_F2FS_FAULT_INJECTION 1617 #define f2fs_show_injection_info(sbi, type) \ 1618 printk_ratelimited("%sF2FS-fs (%s) : inject %s in %s of %pS\n", \ 1619 KERN_INFO, sbi->sb->s_id, \ 1620 f2fs_fault_name[type], \ 1621 __func__, __builtin_return_address(0)) 1622 static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type) 1623 { 1624 struct f2fs_fault_info *ffi = &F2FS_OPTION(sbi).fault_info; 1625 1626 if (!ffi->inject_rate) 1627 return false; 1628 1629 if (!IS_FAULT_SET(ffi, type)) 1630 return false; 1631 1632 atomic_inc(&ffi->inject_ops); 1633 if (atomic_read(&ffi->inject_ops) >= ffi->inject_rate) { 1634 atomic_set(&ffi->inject_ops, 0); 1635 return true; 1636 } 1637 return false; 1638 } 1639 #else 1640 #define f2fs_show_injection_info(sbi, type) do { } while (0) 1641 static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type) 1642 { 1643 return false; 1644 } 1645 #endif 1646 1647 /* 1648 * Test if the mounted volume is a multi-device volume. 1649 * - For a single regular disk volume, sbi->s_ndevs is 0. 1650 * - For a single zoned disk volume, sbi->s_ndevs is 1. 1651 * - For a multi-device volume, sbi->s_ndevs is always 2 or more. 1652 */ 1653 static inline bool f2fs_is_multi_device(struct f2fs_sb_info *sbi) 1654 { 1655 return sbi->s_ndevs > 1; 1656 } 1657 1658 static inline void f2fs_update_time(struct f2fs_sb_info *sbi, int type) 1659 { 1660 unsigned long now = jiffies; 1661 1662 sbi->last_time[type] = now; 1663 1664 /* DISCARD_TIME and GC_TIME are based on REQ_TIME */ 1665 if (type == REQ_TIME) { 1666 sbi->last_time[DISCARD_TIME] = now; 1667 sbi->last_time[GC_TIME] = now; 1668 } 1669 } 1670 1671 static inline bool f2fs_time_over(struct f2fs_sb_info *sbi, int type) 1672 { 1673 unsigned long interval = sbi->interval_time[type] * HZ; 1674 1675 return time_after(jiffies, sbi->last_time[type] + interval); 1676 } 1677 1678 static inline unsigned int f2fs_time_to_wait(struct f2fs_sb_info *sbi, 1679 int type) 1680 { 1681 unsigned long interval = sbi->interval_time[type] * HZ; 1682 unsigned int wait_ms = 0; 1683 long delta; 1684 1685 delta = (sbi->last_time[type] + interval) - jiffies; 1686 if (delta > 0) 1687 wait_ms = jiffies_to_msecs(delta); 1688 1689 return wait_ms; 1690 } 1691 1692 /* 1693 * Inline functions 1694 */ 1695 static inline u32 __f2fs_crc32(struct f2fs_sb_info *sbi, u32 crc, 1696 const void *address, unsigned int length) 1697 { 1698 struct { 1699 struct shash_desc shash; 1700 char ctx[4]; 1701 } desc; 1702 int err; 1703 1704 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver) != sizeof(desc.ctx)); 1705 1706 desc.shash.tfm = sbi->s_chksum_driver; 1707 *(u32 *)desc.ctx = crc; 1708 1709 err = crypto_shash_update(&desc.shash, address, length); 1710 BUG_ON(err); 1711 1712 return *(u32 *)desc.ctx; 1713 } 1714 1715 static inline u32 f2fs_crc32(struct f2fs_sb_info *sbi, const void *address, 1716 unsigned int length) 1717 { 1718 return __f2fs_crc32(sbi, F2FS_SUPER_MAGIC, address, length); 1719 } 1720 1721 static inline bool f2fs_crc_valid(struct f2fs_sb_info *sbi, __u32 blk_crc, 1722 void *buf, size_t buf_size) 1723 { 1724 return f2fs_crc32(sbi, buf, buf_size) == blk_crc; 1725 } 1726 1727 static inline u32 f2fs_chksum(struct f2fs_sb_info *sbi, u32 crc, 1728 const void *address, unsigned int length) 1729 { 1730 return __f2fs_crc32(sbi, crc, address, length); 1731 } 1732 1733 static inline struct f2fs_inode_info *F2FS_I(struct inode *inode) 1734 { 1735 return container_of(inode, struct f2fs_inode_info, vfs_inode); 1736 } 1737 1738 static inline struct f2fs_sb_info *F2FS_SB(struct super_block *sb) 1739 { 1740 return sb->s_fs_info; 1741 } 1742 1743 static inline struct f2fs_sb_info *F2FS_I_SB(struct inode *inode) 1744 { 1745 return F2FS_SB(inode->i_sb); 1746 } 1747 1748 static inline struct f2fs_sb_info *F2FS_M_SB(struct address_space *mapping) 1749 { 1750 return F2FS_I_SB(mapping->host); 1751 } 1752 1753 static inline struct f2fs_sb_info *F2FS_P_SB(struct page *page) 1754 { 1755 return F2FS_M_SB(page_file_mapping(page)); 1756 } 1757 1758 static inline struct f2fs_super_block *F2FS_RAW_SUPER(struct f2fs_sb_info *sbi) 1759 { 1760 return (struct f2fs_super_block *)(sbi->raw_super); 1761 } 1762 1763 static inline struct f2fs_checkpoint *F2FS_CKPT(struct f2fs_sb_info *sbi) 1764 { 1765 return (struct f2fs_checkpoint *)(sbi->ckpt); 1766 } 1767 1768 static inline struct f2fs_node *F2FS_NODE(struct page *page) 1769 { 1770 return (struct f2fs_node *)page_address(page); 1771 } 1772 1773 static inline struct f2fs_inode *F2FS_INODE(struct page *page) 1774 { 1775 return &((struct f2fs_node *)page_address(page))->i; 1776 } 1777 1778 static inline struct f2fs_nm_info *NM_I(struct f2fs_sb_info *sbi) 1779 { 1780 return (struct f2fs_nm_info *)(sbi->nm_info); 1781 } 1782 1783 static inline struct f2fs_sm_info *SM_I(struct f2fs_sb_info *sbi) 1784 { 1785 return (struct f2fs_sm_info *)(sbi->sm_info); 1786 } 1787 1788 static inline struct sit_info *SIT_I(struct f2fs_sb_info *sbi) 1789 { 1790 return (struct sit_info *)(SM_I(sbi)->sit_info); 1791 } 1792 1793 static inline struct free_segmap_info *FREE_I(struct f2fs_sb_info *sbi) 1794 { 1795 return (struct free_segmap_info *)(SM_I(sbi)->free_info); 1796 } 1797 1798 static inline struct dirty_seglist_info *DIRTY_I(struct f2fs_sb_info *sbi) 1799 { 1800 return (struct dirty_seglist_info *)(SM_I(sbi)->dirty_info); 1801 } 1802 1803 static inline struct address_space *META_MAPPING(struct f2fs_sb_info *sbi) 1804 { 1805 return sbi->meta_inode->i_mapping; 1806 } 1807 1808 static inline struct address_space *NODE_MAPPING(struct f2fs_sb_info *sbi) 1809 { 1810 return sbi->node_inode->i_mapping; 1811 } 1812 1813 static inline bool is_sbi_flag_set(struct f2fs_sb_info *sbi, unsigned int type) 1814 { 1815 return test_bit(type, &sbi->s_flag); 1816 } 1817 1818 static inline void set_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type) 1819 { 1820 set_bit(type, &sbi->s_flag); 1821 } 1822 1823 static inline void clear_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type) 1824 { 1825 clear_bit(type, &sbi->s_flag); 1826 } 1827 1828 static inline unsigned long long cur_cp_version(struct f2fs_checkpoint *cp) 1829 { 1830 return le64_to_cpu(cp->checkpoint_ver); 1831 } 1832 1833 static inline unsigned long f2fs_qf_ino(struct super_block *sb, int type) 1834 { 1835 if (type < F2FS_MAX_QUOTAS) 1836 return le32_to_cpu(F2FS_SB(sb)->raw_super->qf_ino[type]); 1837 return 0; 1838 } 1839 1840 static inline __u64 cur_cp_crc(struct f2fs_checkpoint *cp) 1841 { 1842 size_t crc_offset = le32_to_cpu(cp->checksum_offset); 1843 return le32_to_cpu(*((__le32 *)((unsigned char *)cp + crc_offset))); 1844 } 1845 1846 static inline bool __is_set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 1847 { 1848 unsigned int ckpt_flags = le32_to_cpu(cp->ckpt_flags); 1849 1850 return ckpt_flags & f; 1851 } 1852 1853 static inline bool is_set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 1854 { 1855 return __is_set_ckpt_flags(F2FS_CKPT(sbi), f); 1856 } 1857 1858 static inline void __set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 1859 { 1860 unsigned int ckpt_flags; 1861 1862 ckpt_flags = le32_to_cpu(cp->ckpt_flags); 1863 ckpt_flags |= f; 1864 cp->ckpt_flags = cpu_to_le32(ckpt_flags); 1865 } 1866 1867 static inline void set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 1868 { 1869 unsigned long flags; 1870 1871 spin_lock_irqsave(&sbi->cp_lock, flags); 1872 __set_ckpt_flags(F2FS_CKPT(sbi), f); 1873 spin_unlock_irqrestore(&sbi->cp_lock, flags); 1874 } 1875 1876 static inline void __clear_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 1877 { 1878 unsigned int ckpt_flags; 1879 1880 ckpt_flags = le32_to_cpu(cp->ckpt_flags); 1881 ckpt_flags &= (~f); 1882 cp->ckpt_flags = cpu_to_le32(ckpt_flags); 1883 } 1884 1885 static inline void clear_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 1886 { 1887 unsigned long flags; 1888 1889 spin_lock_irqsave(&sbi->cp_lock, flags); 1890 __clear_ckpt_flags(F2FS_CKPT(sbi), f); 1891 spin_unlock_irqrestore(&sbi->cp_lock, flags); 1892 } 1893 1894 static inline void disable_nat_bits(struct f2fs_sb_info *sbi, bool lock) 1895 { 1896 unsigned long flags; 1897 unsigned char *nat_bits; 1898 1899 /* 1900 * In order to re-enable nat_bits we need to call fsck.f2fs by 1901 * set_sbi_flag(sbi, SBI_NEED_FSCK). But it may give huge cost, 1902 * so let's rely on regular fsck or unclean shutdown. 1903 */ 1904 1905 if (lock) 1906 spin_lock_irqsave(&sbi->cp_lock, flags); 1907 __clear_ckpt_flags(F2FS_CKPT(sbi), CP_NAT_BITS_FLAG); 1908 nat_bits = NM_I(sbi)->nat_bits; 1909 NM_I(sbi)->nat_bits = NULL; 1910 if (lock) 1911 spin_unlock_irqrestore(&sbi->cp_lock, flags); 1912 1913 kvfree(nat_bits); 1914 } 1915 1916 static inline bool enabled_nat_bits(struct f2fs_sb_info *sbi, 1917 struct cp_control *cpc) 1918 { 1919 bool set = is_set_ckpt_flags(sbi, CP_NAT_BITS_FLAG); 1920 1921 return (cpc) ? (cpc->reason & CP_UMOUNT) && set : set; 1922 } 1923 1924 static inline void f2fs_lock_op(struct f2fs_sb_info *sbi) 1925 { 1926 down_read(&sbi->cp_rwsem); 1927 } 1928 1929 static inline int f2fs_trylock_op(struct f2fs_sb_info *sbi) 1930 { 1931 return down_read_trylock(&sbi->cp_rwsem); 1932 } 1933 1934 static inline void f2fs_unlock_op(struct f2fs_sb_info *sbi) 1935 { 1936 up_read(&sbi->cp_rwsem); 1937 } 1938 1939 static inline void f2fs_lock_all(struct f2fs_sb_info *sbi) 1940 { 1941 down_write(&sbi->cp_rwsem); 1942 } 1943 1944 static inline void f2fs_unlock_all(struct f2fs_sb_info *sbi) 1945 { 1946 up_write(&sbi->cp_rwsem); 1947 } 1948 1949 static inline int __get_cp_reason(struct f2fs_sb_info *sbi) 1950 { 1951 int reason = CP_SYNC; 1952 1953 if (test_opt(sbi, FASTBOOT)) 1954 reason = CP_FASTBOOT; 1955 if (is_sbi_flag_set(sbi, SBI_IS_CLOSE)) 1956 reason = CP_UMOUNT; 1957 return reason; 1958 } 1959 1960 static inline bool __remain_node_summaries(int reason) 1961 { 1962 return (reason & (CP_UMOUNT | CP_FASTBOOT)); 1963 } 1964 1965 static inline bool __exist_node_summaries(struct f2fs_sb_info *sbi) 1966 { 1967 return (is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG) || 1968 is_set_ckpt_flags(sbi, CP_FASTBOOT_FLAG)); 1969 } 1970 1971 /* 1972 * Check whether the inode has blocks or not 1973 */ 1974 static inline int F2FS_HAS_BLOCKS(struct inode *inode) 1975 { 1976 block_t xattr_block = F2FS_I(inode)->i_xattr_nid ? 1 : 0; 1977 1978 return (inode->i_blocks >> F2FS_LOG_SECTORS_PER_BLOCK) > xattr_block; 1979 } 1980 1981 static inline bool f2fs_has_xattr_block(unsigned int ofs) 1982 { 1983 return ofs == XATTR_NODE_OFFSET; 1984 } 1985 1986 static inline bool __allow_reserved_blocks(struct f2fs_sb_info *sbi, 1987 struct inode *inode, bool cap) 1988 { 1989 if (!inode) 1990 return true; 1991 if (!test_opt(sbi, RESERVE_ROOT)) 1992 return false; 1993 if (IS_NOQUOTA(inode)) 1994 return true; 1995 if (uid_eq(F2FS_OPTION(sbi).s_resuid, current_fsuid())) 1996 return true; 1997 if (!gid_eq(F2FS_OPTION(sbi).s_resgid, GLOBAL_ROOT_GID) && 1998 in_group_p(F2FS_OPTION(sbi).s_resgid)) 1999 return true; 2000 if (cap && capable(CAP_SYS_RESOURCE)) 2001 return true; 2002 return false; 2003 } 2004 2005 static inline void f2fs_i_blocks_write(struct inode *, block_t, bool, bool); 2006 static inline int inc_valid_block_count(struct f2fs_sb_info *sbi, 2007 struct inode *inode, blkcnt_t *count) 2008 { 2009 blkcnt_t diff = 0, release = 0; 2010 block_t avail_user_block_count; 2011 int ret; 2012 2013 ret = dquot_reserve_block(inode, *count); 2014 if (ret) 2015 return ret; 2016 2017 if (time_to_inject(sbi, FAULT_BLOCK)) { 2018 f2fs_show_injection_info(sbi, FAULT_BLOCK); 2019 release = *count; 2020 goto release_quota; 2021 } 2022 2023 /* 2024 * let's increase this in prior to actual block count change in order 2025 * for f2fs_sync_file to avoid data races when deciding checkpoint. 2026 */ 2027 percpu_counter_add(&sbi->alloc_valid_block_count, (*count)); 2028 2029 spin_lock(&sbi->stat_lock); 2030 sbi->total_valid_block_count += (block_t)(*count); 2031 avail_user_block_count = sbi->user_block_count - 2032 sbi->current_reserved_blocks; 2033 2034 if (!__allow_reserved_blocks(sbi, inode, true)) 2035 avail_user_block_count -= F2FS_OPTION(sbi).root_reserved_blocks; 2036 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) { 2037 if (avail_user_block_count > sbi->unusable_block_count) 2038 avail_user_block_count -= sbi->unusable_block_count; 2039 else 2040 avail_user_block_count = 0; 2041 } 2042 if (unlikely(sbi->total_valid_block_count > avail_user_block_count)) { 2043 diff = sbi->total_valid_block_count - avail_user_block_count; 2044 if (diff > *count) 2045 diff = *count; 2046 *count -= diff; 2047 release = diff; 2048 sbi->total_valid_block_count -= diff; 2049 if (!*count) { 2050 spin_unlock(&sbi->stat_lock); 2051 goto enospc; 2052 } 2053 } 2054 spin_unlock(&sbi->stat_lock); 2055 2056 if (unlikely(release)) { 2057 percpu_counter_sub(&sbi->alloc_valid_block_count, release); 2058 dquot_release_reservation_block(inode, release); 2059 } 2060 f2fs_i_blocks_write(inode, *count, true, true); 2061 return 0; 2062 2063 enospc: 2064 percpu_counter_sub(&sbi->alloc_valid_block_count, release); 2065 release_quota: 2066 dquot_release_reservation_block(inode, release); 2067 return -ENOSPC; 2068 } 2069 2070 __printf(2, 3) 2071 void f2fs_printk(struct f2fs_sb_info *sbi, const char *fmt, ...); 2072 2073 #define f2fs_err(sbi, fmt, ...) \ 2074 f2fs_printk(sbi, KERN_ERR fmt, ##__VA_ARGS__) 2075 #define f2fs_warn(sbi, fmt, ...) \ 2076 f2fs_printk(sbi, KERN_WARNING fmt, ##__VA_ARGS__) 2077 #define f2fs_notice(sbi, fmt, ...) \ 2078 f2fs_printk(sbi, KERN_NOTICE fmt, ##__VA_ARGS__) 2079 #define f2fs_info(sbi, fmt, ...) \ 2080 f2fs_printk(sbi, KERN_INFO fmt, ##__VA_ARGS__) 2081 #define f2fs_debug(sbi, fmt, ...) \ 2082 f2fs_printk(sbi, KERN_DEBUG fmt, ##__VA_ARGS__) 2083 2084 static inline void dec_valid_block_count(struct f2fs_sb_info *sbi, 2085 struct inode *inode, 2086 block_t count) 2087 { 2088 blkcnt_t sectors = count << F2FS_LOG_SECTORS_PER_BLOCK; 2089 2090 spin_lock(&sbi->stat_lock); 2091 f2fs_bug_on(sbi, sbi->total_valid_block_count < (block_t) count); 2092 sbi->total_valid_block_count -= (block_t)count; 2093 if (sbi->reserved_blocks && 2094 sbi->current_reserved_blocks < sbi->reserved_blocks) 2095 sbi->current_reserved_blocks = min(sbi->reserved_blocks, 2096 sbi->current_reserved_blocks + count); 2097 spin_unlock(&sbi->stat_lock); 2098 if (unlikely(inode->i_blocks < sectors)) { 2099 f2fs_warn(sbi, "Inconsistent i_blocks, ino:%lu, iblocks:%llu, sectors:%llu", 2100 inode->i_ino, 2101 (unsigned long long)inode->i_blocks, 2102 (unsigned long long)sectors); 2103 set_sbi_flag(sbi, SBI_NEED_FSCK); 2104 return; 2105 } 2106 f2fs_i_blocks_write(inode, count, false, true); 2107 } 2108 2109 static inline void inc_page_count(struct f2fs_sb_info *sbi, int count_type) 2110 { 2111 atomic_inc(&sbi->nr_pages[count_type]); 2112 2113 if (count_type == F2FS_DIRTY_DENTS || 2114 count_type == F2FS_DIRTY_NODES || 2115 count_type == F2FS_DIRTY_META || 2116 count_type == F2FS_DIRTY_QDATA || 2117 count_type == F2FS_DIRTY_IMETA) 2118 set_sbi_flag(sbi, SBI_IS_DIRTY); 2119 } 2120 2121 static inline void inode_inc_dirty_pages(struct inode *inode) 2122 { 2123 atomic_inc(&F2FS_I(inode)->dirty_pages); 2124 inc_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ? 2125 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA); 2126 if (IS_NOQUOTA(inode)) 2127 inc_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA); 2128 } 2129 2130 static inline void dec_page_count(struct f2fs_sb_info *sbi, int count_type) 2131 { 2132 atomic_dec(&sbi->nr_pages[count_type]); 2133 } 2134 2135 static inline void inode_dec_dirty_pages(struct inode *inode) 2136 { 2137 if (!S_ISDIR(inode->i_mode) && !S_ISREG(inode->i_mode) && 2138 !S_ISLNK(inode->i_mode)) 2139 return; 2140 2141 atomic_dec(&F2FS_I(inode)->dirty_pages); 2142 dec_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ? 2143 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA); 2144 if (IS_NOQUOTA(inode)) 2145 dec_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA); 2146 } 2147 2148 static inline s64 get_pages(struct f2fs_sb_info *sbi, int count_type) 2149 { 2150 return atomic_read(&sbi->nr_pages[count_type]); 2151 } 2152 2153 static inline int get_dirty_pages(struct inode *inode) 2154 { 2155 return atomic_read(&F2FS_I(inode)->dirty_pages); 2156 } 2157 2158 static inline int get_blocktype_secs(struct f2fs_sb_info *sbi, int block_type) 2159 { 2160 unsigned int pages_per_sec = sbi->segs_per_sec * sbi->blocks_per_seg; 2161 unsigned int segs = (get_pages(sbi, block_type) + pages_per_sec - 1) >> 2162 sbi->log_blocks_per_seg; 2163 2164 return segs / sbi->segs_per_sec; 2165 } 2166 2167 static inline block_t valid_user_blocks(struct f2fs_sb_info *sbi) 2168 { 2169 return sbi->total_valid_block_count; 2170 } 2171 2172 static inline block_t discard_blocks(struct f2fs_sb_info *sbi) 2173 { 2174 return sbi->discard_blks; 2175 } 2176 2177 static inline unsigned long __bitmap_size(struct f2fs_sb_info *sbi, int flag) 2178 { 2179 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi); 2180 2181 /* return NAT or SIT bitmap */ 2182 if (flag == NAT_BITMAP) 2183 return le32_to_cpu(ckpt->nat_ver_bitmap_bytesize); 2184 else if (flag == SIT_BITMAP) 2185 return le32_to_cpu(ckpt->sit_ver_bitmap_bytesize); 2186 2187 return 0; 2188 } 2189 2190 static inline block_t __cp_payload(struct f2fs_sb_info *sbi) 2191 { 2192 return le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_payload); 2193 } 2194 2195 static inline void *__bitmap_ptr(struct f2fs_sb_info *sbi, int flag) 2196 { 2197 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi); 2198 int offset; 2199 2200 if (is_set_ckpt_flags(sbi, CP_LARGE_NAT_BITMAP_FLAG)) { 2201 offset = (flag == SIT_BITMAP) ? 2202 le32_to_cpu(ckpt->nat_ver_bitmap_bytesize) : 0; 2203 /* 2204 * if large_nat_bitmap feature is enabled, leave checksum 2205 * protection for all nat/sit bitmaps. 2206 */ 2207 return &ckpt->sit_nat_version_bitmap + offset + sizeof(__le32); 2208 } 2209 2210 if (__cp_payload(sbi) > 0) { 2211 if (flag == NAT_BITMAP) 2212 return &ckpt->sit_nat_version_bitmap; 2213 else 2214 return (unsigned char *)ckpt + F2FS_BLKSIZE; 2215 } else { 2216 offset = (flag == NAT_BITMAP) ? 2217 le32_to_cpu(ckpt->sit_ver_bitmap_bytesize) : 0; 2218 return &ckpt->sit_nat_version_bitmap + offset; 2219 } 2220 } 2221 2222 static inline block_t __start_cp_addr(struct f2fs_sb_info *sbi) 2223 { 2224 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr); 2225 2226 if (sbi->cur_cp_pack == 2) 2227 start_addr += sbi->blocks_per_seg; 2228 return start_addr; 2229 } 2230 2231 static inline block_t __start_cp_next_addr(struct f2fs_sb_info *sbi) 2232 { 2233 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr); 2234 2235 if (sbi->cur_cp_pack == 1) 2236 start_addr += sbi->blocks_per_seg; 2237 return start_addr; 2238 } 2239 2240 static inline void __set_cp_next_pack(struct f2fs_sb_info *sbi) 2241 { 2242 sbi->cur_cp_pack = (sbi->cur_cp_pack == 1) ? 2 : 1; 2243 } 2244 2245 static inline block_t __start_sum_addr(struct f2fs_sb_info *sbi) 2246 { 2247 return le32_to_cpu(F2FS_CKPT(sbi)->cp_pack_start_sum); 2248 } 2249 2250 static inline int inc_valid_node_count(struct f2fs_sb_info *sbi, 2251 struct inode *inode, bool is_inode) 2252 { 2253 block_t valid_block_count; 2254 unsigned int valid_node_count, user_block_count; 2255 int err; 2256 2257 if (is_inode) { 2258 if (inode) { 2259 err = dquot_alloc_inode(inode); 2260 if (err) 2261 return err; 2262 } 2263 } else { 2264 err = dquot_reserve_block(inode, 1); 2265 if (err) 2266 return err; 2267 } 2268 2269 if (time_to_inject(sbi, FAULT_BLOCK)) { 2270 f2fs_show_injection_info(sbi, FAULT_BLOCK); 2271 goto enospc; 2272 } 2273 2274 spin_lock(&sbi->stat_lock); 2275 2276 valid_block_count = sbi->total_valid_block_count + 2277 sbi->current_reserved_blocks + 1; 2278 2279 if (!__allow_reserved_blocks(sbi, inode, false)) 2280 valid_block_count += F2FS_OPTION(sbi).root_reserved_blocks; 2281 user_block_count = sbi->user_block_count; 2282 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) 2283 user_block_count -= sbi->unusable_block_count; 2284 2285 if (unlikely(valid_block_count > user_block_count)) { 2286 spin_unlock(&sbi->stat_lock); 2287 goto enospc; 2288 } 2289 2290 valid_node_count = sbi->total_valid_node_count + 1; 2291 if (unlikely(valid_node_count > sbi->total_node_count)) { 2292 spin_unlock(&sbi->stat_lock); 2293 goto enospc; 2294 } 2295 2296 sbi->total_valid_node_count++; 2297 sbi->total_valid_block_count++; 2298 spin_unlock(&sbi->stat_lock); 2299 2300 if (inode) { 2301 if (is_inode) 2302 f2fs_mark_inode_dirty_sync(inode, true); 2303 else 2304 f2fs_i_blocks_write(inode, 1, true, true); 2305 } 2306 2307 percpu_counter_inc(&sbi->alloc_valid_block_count); 2308 return 0; 2309 2310 enospc: 2311 if (is_inode) { 2312 if (inode) 2313 dquot_free_inode(inode); 2314 } else { 2315 dquot_release_reservation_block(inode, 1); 2316 } 2317 return -ENOSPC; 2318 } 2319 2320 static inline void dec_valid_node_count(struct f2fs_sb_info *sbi, 2321 struct inode *inode, bool is_inode) 2322 { 2323 spin_lock(&sbi->stat_lock); 2324 2325 f2fs_bug_on(sbi, !sbi->total_valid_block_count); 2326 f2fs_bug_on(sbi, !sbi->total_valid_node_count); 2327 2328 sbi->total_valid_node_count--; 2329 sbi->total_valid_block_count--; 2330 if (sbi->reserved_blocks && 2331 sbi->current_reserved_blocks < sbi->reserved_blocks) 2332 sbi->current_reserved_blocks++; 2333 2334 spin_unlock(&sbi->stat_lock); 2335 2336 if (is_inode) { 2337 dquot_free_inode(inode); 2338 } else { 2339 if (unlikely(inode->i_blocks == 0)) { 2340 f2fs_warn(sbi, "dec_valid_node_count: inconsistent i_blocks, ino:%lu, iblocks:%llu", 2341 inode->i_ino, 2342 (unsigned long long)inode->i_blocks); 2343 set_sbi_flag(sbi, SBI_NEED_FSCK); 2344 return; 2345 } 2346 f2fs_i_blocks_write(inode, 1, false, true); 2347 } 2348 } 2349 2350 static inline unsigned int valid_node_count(struct f2fs_sb_info *sbi) 2351 { 2352 return sbi->total_valid_node_count; 2353 } 2354 2355 static inline void inc_valid_inode_count(struct f2fs_sb_info *sbi) 2356 { 2357 percpu_counter_inc(&sbi->total_valid_inode_count); 2358 } 2359 2360 static inline void dec_valid_inode_count(struct f2fs_sb_info *sbi) 2361 { 2362 percpu_counter_dec(&sbi->total_valid_inode_count); 2363 } 2364 2365 static inline s64 valid_inode_count(struct f2fs_sb_info *sbi) 2366 { 2367 return percpu_counter_sum_positive(&sbi->total_valid_inode_count); 2368 } 2369 2370 static inline struct page *f2fs_grab_cache_page(struct address_space *mapping, 2371 pgoff_t index, bool for_write) 2372 { 2373 struct page *page; 2374 2375 if (IS_ENABLED(CONFIG_F2FS_FAULT_INJECTION)) { 2376 if (!for_write) 2377 page = find_get_page_flags(mapping, index, 2378 FGP_LOCK | FGP_ACCESSED); 2379 else 2380 page = find_lock_page(mapping, index); 2381 if (page) 2382 return page; 2383 2384 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_ALLOC)) { 2385 f2fs_show_injection_info(F2FS_M_SB(mapping), 2386 FAULT_PAGE_ALLOC); 2387 return NULL; 2388 } 2389 } 2390 2391 if (!for_write) 2392 return grab_cache_page(mapping, index); 2393 return grab_cache_page_write_begin(mapping, index, AOP_FLAG_NOFS); 2394 } 2395 2396 static inline struct page *f2fs_pagecache_get_page( 2397 struct address_space *mapping, pgoff_t index, 2398 int fgp_flags, gfp_t gfp_mask) 2399 { 2400 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_GET)) { 2401 f2fs_show_injection_info(F2FS_M_SB(mapping), FAULT_PAGE_GET); 2402 return NULL; 2403 } 2404 2405 return pagecache_get_page(mapping, index, fgp_flags, gfp_mask); 2406 } 2407 2408 static inline void f2fs_copy_page(struct page *src, struct page *dst) 2409 { 2410 char *src_kaddr = kmap(src); 2411 char *dst_kaddr = kmap(dst); 2412 2413 memcpy(dst_kaddr, src_kaddr, PAGE_SIZE); 2414 kunmap(dst); 2415 kunmap(src); 2416 } 2417 2418 static inline void f2fs_put_page(struct page *page, int unlock) 2419 { 2420 if (!page) 2421 return; 2422 2423 if (unlock) { 2424 f2fs_bug_on(F2FS_P_SB(page), !PageLocked(page)); 2425 unlock_page(page); 2426 } 2427 put_page(page); 2428 } 2429 2430 static inline void f2fs_put_dnode(struct dnode_of_data *dn) 2431 { 2432 if (dn->node_page) 2433 f2fs_put_page(dn->node_page, 1); 2434 if (dn->inode_page && dn->node_page != dn->inode_page) 2435 f2fs_put_page(dn->inode_page, 0); 2436 dn->node_page = NULL; 2437 dn->inode_page = NULL; 2438 } 2439 2440 static inline struct kmem_cache *f2fs_kmem_cache_create(const char *name, 2441 size_t size) 2442 { 2443 return kmem_cache_create(name, size, 0, SLAB_RECLAIM_ACCOUNT, NULL); 2444 } 2445 2446 static inline void *f2fs_kmem_cache_alloc(struct kmem_cache *cachep, 2447 gfp_t flags) 2448 { 2449 void *entry; 2450 2451 entry = kmem_cache_alloc(cachep, flags); 2452 if (!entry) 2453 entry = kmem_cache_alloc(cachep, flags | __GFP_NOFAIL); 2454 return entry; 2455 } 2456 2457 static inline bool is_inflight_io(struct f2fs_sb_info *sbi, int type) 2458 { 2459 if (get_pages(sbi, F2FS_RD_DATA) || get_pages(sbi, F2FS_RD_NODE) || 2460 get_pages(sbi, F2FS_RD_META) || get_pages(sbi, F2FS_WB_DATA) || 2461 get_pages(sbi, F2FS_WB_CP_DATA) || 2462 get_pages(sbi, F2FS_DIO_READ) || 2463 get_pages(sbi, F2FS_DIO_WRITE)) 2464 return true; 2465 2466 if (type != DISCARD_TIME && SM_I(sbi) && SM_I(sbi)->dcc_info && 2467 atomic_read(&SM_I(sbi)->dcc_info->queued_discard)) 2468 return true; 2469 2470 if (SM_I(sbi) && SM_I(sbi)->fcc_info && 2471 atomic_read(&SM_I(sbi)->fcc_info->queued_flush)) 2472 return true; 2473 return false; 2474 } 2475 2476 static inline bool is_idle(struct f2fs_sb_info *sbi, int type) 2477 { 2478 if (sbi->gc_mode == GC_URGENT_HIGH) 2479 return true; 2480 2481 if (is_inflight_io(sbi, type)) 2482 return false; 2483 2484 if (sbi->gc_mode == GC_URGENT_LOW && 2485 (type == DISCARD_TIME || type == GC_TIME)) 2486 return true; 2487 2488 return f2fs_time_over(sbi, type); 2489 } 2490 2491 static inline void f2fs_radix_tree_insert(struct radix_tree_root *root, 2492 unsigned long index, void *item) 2493 { 2494 while (radix_tree_insert(root, index, item)) 2495 cond_resched(); 2496 } 2497 2498 #define RAW_IS_INODE(p) ((p)->footer.nid == (p)->footer.ino) 2499 2500 static inline bool IS_INODE(struct page *page) 2501 { 2502 struct f2fs_node *p = F2FS_NODE(page); 2503 2504 return RAW_IS_INODE(p); 2505 } 2506 2507 static inline int offset_in_addr(struct f2fs_inode *i) 2508 { 2509 return (i->i_inline & F2FS_EXTRA_ATTR) ? 2510 (le16_to_cpu(i->i_extra_isize) / sizeof(__le32)) : 0; 2511 } 2512 2513 static inline __le32 *blkaddr_in_node(struct f2fs_node *node) 2514 { 2515 return RAW_IS_INODE(node) ? node->i.i_addr : node->dn.addr; 2516 } 2517 2518 static inline int f2fs_has_extra_attr(struct inode *inode); 2519 static inline block_t data_blkaddr(struct inode *inode, 2520 struct page *node_page, unsigned int offset) 2521 { 2522 struct f2fs_node *raw_node; 2523 __le32 *addr_array; 2524 int base = 0; 2525 bool is_inode = IS_INODE(node_page); 2526 2527 raw_node = F2FS_NODE(node_page); 2528 2529 if (is_inode) { 2530 if (!inode) 2531 /* from GC path only */ 2532 base = offset_in_addr(&raw_node->i); 2533 else if (f2fs_has_extra_attr(inode)) 2534 base = get_extra_isize(inode); 2535 } 2536 2537 addr_array = blkaddr_in_node(raw_node); 2538 return le32_to_cpu(addr_array[base + offset]); 2539 } 2540 2541 static inline block_t f2fs_data_blkaddr(struct dnode_of_data *dn) 2542 { 2543 return data_blkaddr(dn->inode, dn->node_page, dn->ofs_in_node); 2544 } 2545 2546 static inline int f2fs_test_bit(unsigned int nr, char *addr) 2547 { 2548 int mask; 2549 2550 addr += (nr >> 3); 2551 mask = 1 << (7 - (nr & 0x07)); 2552 return mask & *addr; 2553 } 2554 2555 static inline void f2fs_set_bit(unsigned int nr, char *addr) 2556 { 2557 int mask; 2558 2559 addr += (nr >> 3); 2560 mask = 1 << (7 - (nr & 0x07)); 2561 *addr |= mask; 2562 } 2563 2564 static inline void f2fs_clear_bit(unsigned int nr, char *addr) 2565 { 2566 int mask; 2567 2568 addr += (nr >> 3); 2569 mask = 1 << (7 - (nr & 0x07)); 2570 *addr &= ~mask; 2571 } 2572 2573 static inline int f2fs_test_and_set_bit(unsigned int nr, char *addr) 2574 { 2575 int mask; 2576 int ret; 2577 2578 addr += (nr >> 3); 2579 mask = 1 << (7 - (nr & 0x07)); 2580 ret = mask & *addr; 2581 *addr |= mask; 2582 return ret; 2583 } 2584 2585 static inline int f2fs_test_and_clear_bit(unsigned int nr, char *addr) 2586 { 2587 int mask; 2588 int ret; 2589 2590 addr += (nr >> 3); 2591 mask = 1 << (7 - (nr & 0x07)); 2592 ret = mask & *addr; 2593 *addr &= ~mask; 2594 return ret; 2595 } 2596 2597 static inline void f2fs_change_bit(unsigned int nr, char *addr) 2598 { 2599 int mask; 2600 2601 addr += (nr >> 3); 2602 mask = 1 << (7 - (nr & 0x07)); 2603 *addr ^= mask; 2604 } 2605 2606 /* 2607 * On-disk inode flags (f2fs_inode::i_flags) 2608 */ 2609 #define F2FS_COMPR_FL 0x00000004 /* Compress file */ 2610 #define F2FS_SYNC_FL 0x00000008 /* Synchronous updates */ 2611 #define F2FS_IMMUTABLE_FL 0x00000010 /* Immutable file */ 2612 #define F2FS_APPEND_FL 0x00000020 /* writes to file may only append */ 2613 #define F2FS_NODUMP_FL 0x00000040 /* do not dump file */ 2614 #define F2FS_NOATIME_FL 0x00000080 /* do not update atime */ 2615 #define F2FS_NOCOMP_FL 0x00000400 /* Don't compress */ 2616 #define F2FS_INDEX_FL 0x00001000 /* hash-indexed directory */ 2617 #define F2FS_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */ 2618 #define F2FS_PROJINHERIT_FL 0x20000000 /* Create with parents projid */ 2619 #define F2FS_CASEFOLD_FL 0x40000000 /* Casefolded file */ 2620 2621 /* Flags that should be inherited by new inodes from their parent. */ 2622 #define F2FS_FL_INHERITED (F2FS_SYNC_FL | F2FS_NODUMP_FL | F2FS_NOATIME_FL | \ 2623 F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \ 2624 F2FS_CASEFOLD_FL | F2FS_COMPR_FL | F2FS_NOCOMP_FL) 2625 2626 /* Flags that are appropriate for regular files (all but dir-specific ones). */ 2627 #define F2FS_REG_FLMASK (~(F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \ 2628 F2FS_CASEFOLD_FL)) 2629 2630 /* Flags that are appropriate for non-directories/regular files. */ 2631 #define F2FS_OTHER_FLMASK (F2FS_NODUMP_FL | F2FS_NOATIME_FL) 2632 2633 static inline __u32 f2fs_mask_flags(umode_t mode, __u32 flags) 2634 { 2635 if (S_ISDIR(mode)) 2636 return flags; 2637 else if (S_ISREG(mode)) 2638 return flags & F2FS_REG_FLMASK; 2639 else 2640 return flags & F2FS_OTHER_FLMASK; 2641 } 2642 2643 static inline void __mark_inode_dirty_flag(struct inode *inode, 2644 int flag, bool set) 2645 { 2646 switch (flag) { 2647 case FI_INLINE_XATTR: 2648 case FI_INLINE_DATA: 2649 case FI_INLINE_DENTRY: 2650 case FI_NEW_INODE: 2651 if (set) 2652 return; 2653 fallthrough; 2654 case FI_DATA_EXIST: 2655 case FI_INLINE_DOTS: 2656 case FI_PIN_FILE: 2657 f2fs_mark_inode_dirty_sync(inode, true); 2658 } 2659 } 2660 2661 static inline void set_inode_flag(struct inode *inode, int flag) 2662 { 2663 set_bit(flag, F2FS_I(inode)->flags); 2664 __mark_inode_dirty_flag(inode, flag, true); 2665 } 2666 2667 static inline int is_inode_flag_set(struct inode *inode, int flag) 2668 { 2669 return test_bit(flag, F2FS_I(inode)->flags); 2670 } 2671 2672 static inline void clear_inode_flag(struct inode *inode, int flag) 2673 { 2674 clear_bit(flag, F2FS_I(inode)->flags); 2675 __mark_inode_dirty_flag(inode, flag, false); 2676 } 2677 2678 static inline bool f2fs_verity_in_progress(struct inode *inode) 2679 { 2680 return IS_ENABLED(CONFIG_FS_VERITY) && 2681 is_inode_flag_set(inode, FI_VERITY_IN_PROGRESS); 2682 } 2683 2684 static inline void set_acl_inode(struct inode *inode, umode_t mode) 2685 { 2686 F2FS_I(inode)->i_acl_mode = mode; 2687 set_inode_flag(inode, FI_ACL_MODE); 2688 f2fs_mark_inode_dirty_sync(inode, false); 2689 } 2690 2691 static inline void f2fs_i_links_write(struct inode *inode, bool inc) 2692 { 2693 if (inc) 2694 inc_nlink(inode); 2695 else 2696 drop_nlink(inode); 2697 f2fs_mark_inode_dirty_sync(inode, true); 2698 } 2699 2700 static inline void f2fs_i_blocks_write(struct inode *inode, 2701 block_t diff, bool add, bool claim) 2702 { 2703 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE); 2704 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER); 2705 2706 /* add = 1, claim = 1 should be dquot_reserve_block in pair */ 2707 if (add) { 2708 if (claim) 2709 dquot_claim_block(inode, diff); 2710 else 2711 dquot_alloc_block_nofail(inode, diff); 2712 } else { 2713 dquot_free_block(inode, diff); 2714 } 2715 2716 f2fs_mark_inode_dirty_sync(inode, true); 2717 if (clean || recover) 2718 set_inode_flag(inode, FI_AUTO_RECOVER); 2719 } 2720 2721 static inline void f2fs_i_size_write(struct inode *inode, loff_t i_size) 2722 { 2723 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE); 2724 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER); 2725 2726 if (i_size_read(inode) == i_size) 2727 return; 2728 2729 i_size_write(inode, i_size); 2730 f2fs_mark_inode_dirty_sync(inode, true); 2731 if (clean || recover) 2732 set_inode_flag(inode, FI_AUTO_RECOVER); 2733 } 2734 2735 static inline void f2fs_i_depth_write(struct inode *inode, unsigned int depth) 2736 { 2737 F2FS_I(inode)->i_current_depth = depth; 2738 f2fs_mark_inode_dirty_sync(inode, true); 2739 } 2740 2741 static inline void f2fs_i_gc_failures_write(struct inode *inode, 2742 unsigned int count) 2743 { 2744 F2FS_I(inode)->i_gc_failures[GC_FAILURE_PIN] = count; 2745 f2fs_mark_inode_dirty_sync(inode, true); 2746 } 2747 2748 static inline void f2fs_i_xnid_write(struct inode *inode, nid_t xnid) 2749 { 2750 F2FS_I(inode)->i_xattr_nid = xnid; 2751 f2fs_mark_inode_dirty_sync(inode, true); 2752 } 2753 2754 static inline void f2fs_i_pino_write(struct inode *inode, nid_t pino) 2755 { 2756 F2FS_I(inode)->i_pino = pino; 2757 f2fs_mark_inode_dirty_sync(inode, true); 2758 } 2759 2760 static inline void get_inline_info(struct inode *inode, struct f2fs_inode *ri) 2761 { 2762 struct f2fs_inode_info *fi = F2FS_I(inode); 2763 2764 if (ri->i_inline & F2FS_INLINE_XATTR) 2765 set_bit(FI_INLINE_XATTR, fi->flags); 2766 if (ri->i_inline & F2FS_INLINE_DATA) 2767 set_bit(FI_INLINE_DATA, fi->flags); 2768 if (ri->i_inline & F2FS_INLINE_DENTRY) 2769 set_bit(FI_INLINE_DENTRY, fi->flags); 2770 if (ri->i_inline & F2FS_DATA_EXIST) 2771 set_bit(FI_DATA_EXIST, fi->flags); 2772 if (ri->i_inline & F2FS_INLINE_DOTS) 2773 set_bit(FI_INLINE_DOTS, fi->flags); 2774 if (ri->i_inline & F2FS_EXTRA_ATTR) 2775 set_bit(FI_EXTRA_ATTR, fi->flags); 2776 if (ri->i_inline & F2FS_PIN_FILE) 2777 set_bit(FI_PIN_FILE, fi->flags); 2778 } 2779 2780 static inline void set_raw_inline(struct inode *inode, struct f2fs_inode *ri) 2781 { 2782 ri->i_inline = 0; 2783 2784 if (is_inode_flag_set(inode, FI_INLINE_XATTR)) 2785 ri->i_inline |= F2FS_INLINE_XATTR; 2786 if (is_inode_flag_set(inode, FI_INLINE_DATA)) 2787 ri->i_inline |= F2FS_INLINE_DATA; 2788 if (is_inode_flag_set(inode, FI_INLINE_DENTRY)) 2789 ri->i_inline |= F2FS_INLINE_DENTRY; 2790 if (is_inode_flag_set(inode, FI_DATA_EXIST)) 2791 ri->i_inline |= F2FS_DATA_EXIST; 2792 if (is_inode_flag_set(inode, FI_INLINE_DOTS)) 2793 ri->i_inline |= F2FS_INLINE_DOTS; 2794 if (is_inode_flag_set(inode, FI_EXTRA_ATTR)) 2795 ri->i_inline |= F2FS_EXTRA_ATTR; 2796 if (is_inode_flag_set(inode, FI_PIN_FILE)) 2797 ri->i_inline |= F2FS_PIN_FILE; 2798 } 2799 2800 static inline int f2fs_has_extra_attr(struct inode *inode) 2801 { 2802 return is_inode_flag_set(inode, FI_EXTRA_ATTR); 2803 } 2804 2805 static inline int f2fs_has_inline_xattr(struct inode *inode) 2806 { 2807 return is_inode_flag_set(inode, FI_INLINE_XATTR); 2808 } 2809 2810 static inline int f2fs_compressed_file(struct inode *inode) 2811 { 2812 return S_ISREG(inode->i_mode) && 2813 is_inode_flag_set(inode, FI_COMPRESSED_FILE); 2814 } 2815 2816 static inline bool f2fs_need_compress_data(struct inode *inode) 2817 { 2818 int compress_mode = F2FS_OPTION(F2FS_I_SB(inode)).compress_mode; 2819 2820 if (!f2fs_compressed_file(inode)) 2821 return false; 2822 2823 if (compress_mode == COMPR_MODE_FS) 2824 return true; 2825 else if (compress_mode == COMPR_MODE_USER && 2826 is_inode_flag_set(inode, FI_ENABLE_COMPRESS)) 2827 return true; 2828 2829 return false; 2830 } 2831 2832 static inline unsigned int addrs_per_inode(struct inode *inode) 2833 { 2834 unsigned int addrs = CUR_ADDRS_PER_INODE(inode) - 2835 get_inline_xattr_addrs(inode); 2836 2837 if (!f2fs_compressed_file(inode)) 2838 return addrs; 2839 return ALIGN_DOWN(addrs, F2FS_I(inode)->i_cluster_size); 2840 } 2841 2842 static inline unsigned int addrs_per_block(struct inode *inode) 2843 { 2844 if (!f2fs_compressed_file(inode)) 2845 return DEF_ADDRS_PER_BLOCK; 2846 return ALIGN_DOWN(DEF_ADDRS_PER_BLOCK, F2FS_I(inode)->i_cluster_size); 2847 } 2848 2849 static inline void *inline_xattr_addr(struct inode *inode, struct page *page) 2850 { 2851 struct f2fs_inode *ri = F2FS_INODE(page); 2852 2853 return (void *)&(ri->i_addr[DEF_ADDRS_PER_INODE - 2854 get_inline_xattr_addrs(inode)]); 2855 } 2856 2857 static inline int inline_xattr_size(struct inode *inode) 2858 { 2859 if (f2fs_has_inline_xattr(inode)) 2860 return get_inline_xattr_addrs(inode) * sizeof(__le32); 2861 return 0; 2862 } 2863 2864 static inline int f2fs_has_inline_data(struct inode *inode) 2865 { 2866 return is_inode_flag_set(inode, FI_INLINE_DATA); 2867 } 2868 2869 static inline int f2fs_exist_data(struct inode *inode) 2870 { 2871 return is_inode_flag_set(inode, FI_DATA_EXIST); 2872 } 2873 2874 static inline int f2fs_has_inline_dots(struct inode *inode) 2875 { 2876 return is_inode_flag_set(inode, FI_INLINE_DOTS); 2877 } 2878 2879 static inline int f2fs_is_mmap_file(struct inode *inode) 2880 { 2881 return is_inode_flag_set(inode, FI_MMAP_FILE); 2882 } 2883 2884 static inline bool f2fs_is_pinned_file(struct inode *inode) 2885 { 2886 return is_inode_flag_set(inode, FI_PIN_FILE); 2887 } 2888 2889 static inline bool f2fs_is_atomic_file(struct inode *inode) 2890 { 2891 return is_inode_flag_set(inode, FI_ATOMIC_FILE); 2892 } 2893 2894 static inline bool f2fs_is_commit_atomic_write(struct inode *inode) 2895 { 2896 return is_inode_flag_set(inode, FI_ATOMIC_COMMIT); 2897 } 2898 2899 static inline bool f2fs_is_volatile_file(struct inode *inode) 2900 { 2901 return is_inode_flag_set(inode, FI_VOLATILE_FILE); 2902 } 2903 2904 static inline bool f2fs_is_first_block_written(struct inode *inode) 2905 { 2906 return is_inode_flag_set(inode, FI_FIRST_BLOCK_WRITTEN); 2907 } 2908 2909 static inline bool f2fs_is_drop_cache(struct inode *inode) 2910 { 2911 return is_inode_flag_set(inode, FI_DROP_CACHE); 2912 } 2913 2914 static inline void *inline_data_addr(struct inode *inode, struct page *page) 2915 { 2916 struct f2fs_inode *ri = F2FS_INODE(page); 2917 int extra_size = get_extra_isize(inode); 2918 2919 return (void *)&(ri->i_addr[extra_size + DEF_INLINE_RESERVED_SIZE]); 2920 } 2921 2922 static inline int f2fs_has_inline_dentry(struct inode *inode) 2923 { 2924 return is_inode_flag_set(inode, FI_INLINE_DENTRY); 2925 } 2926 2927 static inline int is_file(struct inode *inode, int type) 2928 { 2929 return F2FS_I(inode)->i_advise & type; 2930 } 2931 2932 static inline void set_file(struct inode *inode, int type) 2933 { 2934 F2FS_I(inode)->i_advise |= type; 2935 f2fs_mark_inode_dirty_sync(inode, true); 2936 } 2937 2938 static inline void clear_file(struct inode *inode, int type) 2939 { 2940 F2FS_I(inode)->i_advise &= ~type; 2941 f2fs_mark_inode_dirty_sync(inode, true); 2942 } 2943 2944 static inline bool f2fs_is_time_consistent(struct inode *inode) 2945 { 2946 if (!timespec64_equal(F2FS_I(inode)->i_disk_time, &inode->i_atime)) 2947 return false; 2948 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 1, &inode->i_ctime)) 2949 return false; 2950 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 2, &inode->i_mtime)) 2951 return false; 2952 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 3, 2953 &F2FS_I(inode)->i_crtime)) 2954 return false; 2955 return true; 2956 } 2957 2958 static inline bool f2fs_skip_inode_update(struct inode *inode, int dsync) 2959 { 2960 bool ret; 2961 2962 if (dsync) { 2963 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 2964 2965 spin_lock(&sbi->inode_lock[DIRTY_META]); 2966 ret = list_empty(&F2FS_I(inode)->gdirty_list); 2967 spin_unlock(&sbi->inode_lock[DIRTY_META]); 2968 return ret; 2969 } 2970 if (!is_inode_flag_set(inode, FI_AUTO_RECOVER) || 2971 file_keep_isize(inode) || 2972 i_size_read(inode) & ~PAGE_MASK) 2973 return false; 2974 2975 if (!f2fs_is_time_consistent(inode)) 2976 return false; 2977 2978 spin_lock(&F2FS_I(inode)->i_size_lock); 2979 ret = F2FS_I(inode)->last_disk_size == i_size_read(inode); 2980 spin_unlock(&F2FS_I(inode)->i_size_lock); 2981 2982 return ret; 2983 } 2984 2985 static inline bool f2fs_readonly(struct super_block *sb) 2986 { 2987 return sb_rdonly(sb); 2988 } 2989 2990 static inline bool f2fs_cp_error(struct f2fs_sb_info *sbi) 2991 { 2992 return is_set_ckpt_flags(sbi, CP_ERROR_FLAG); 2993 } 2994 2995 static inline bool is_dot_dotdot(const u8 *name, size_t len) 2996 { 2997 if (len == 1 && name[0] == '.') 2998 return true; 2999 3000 if (len == 2 && name[0] == '.' && name[1] == '.') 3001 return true; 3002 3003 return false; 3004 } 3005 3006 static inline bool f2fs_may_extent_tree(struct inode *inode) 3007 { 3008 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 3009 3010 if (!test_opt(sbi, EXTENT_CACHE) || 3011 is_inode_flag_set(inode, FI_NO_EXTENT) || 3012 is_inode_flag_set(inode, FI_COMPRESSED_FILE)) 3013 return false; 3014 3015 /* 3016 * for recovered files during mount do not create extents 3017 * if shrinker is not registered. 3018 */ 3019 if (list_empty(&sbi->s_list)) 3020 return false; 3021 3022 return S_ISREG(inode->i_mode); 3023 } 3024 3025 static inline void *f2fs_kmalloc(struct f2fs_sb_info *sbi, 3026 size_t size, gfp_t flags) 3027 { 3028 if (time_to_inject(sbi, FAULT_KMALLOC)) { 3029 f2fs_show_injection_info(sbi, FAULT_KMALLOC); 3030 return NULL; 3031 } 3032 3033 return kmalloc(size, flags); 3034 } 3035 3036 static inline void *f2fs_kzalloc(struct f2fs_sb_info *sbi, 3037 size_t size, gfp_t flags) 3038 { 3039 return f2fs_kmalloc(sbi, size, flags | __GFP_ZERO); 3040 } 3041 3042 static inline void *f2fs_kvmalloc(struct f2fs_sb_info *sbi, 3043 size_t size, gfp_t flags) 3044 { 3045 if (time_to_inject(sbi, FAULT_KVMALLOC)) { 3046 f2fs_show_injection_info(sbi, FAULT_KVMALLOC); 3047 return NULL; 3048 } 3049 3050 return kvmalloc(size, flags); 3051 } 3052 3053 static inline void *f2fs_kvzalloc(struct f2fs_sb_info *sbi, 3054 size_t size, gfp_t flags) 3055 { 3056 return f2fs_kvmalloc(sbi, size, flags | __GFP_ZERO); 3057 } 3058 3059 static inline int get_extra_isize(struct inode *inode) 3060 { 3061 return F2FS_I(inode)->i_extra_isize / sizeof(__le32); 3062 } 3063 3064 static inline int get_inline_xattr_addrs(struct inode *inode) 3065 { 3066 return F2FS_I(inode)->i_inline_xattr_size; 3067 } 3068 3069 #define f2fs_get_inode_mode(i) \ 3070 ((is_inode_flag_set(i, FI_ACL_MODE)) ? \ 3071 (F2FS_I(i)->i_acl_mode) : ((i)->i_mode)) 3072 3073 #define F2FS_TOTAL_EXTRA_ATTR_SIZE \ 3074 (offsetof(struct f2fs_inode, i_extra_end) - \ 3075 offsetof(struct f2fs_inode, i_extra_isize)) \ 3076 3077 #define F2FS_OLD_ATTRIBUTE_SIZE (offsetof(struct f2fs_inode, i_addr)) 3078 #define F2FS_FITS_IN_INODE(f2fs_inode, extra_isize, field) \ 3079 ((offsetof(typeof(*(f2fs_inode)), field) + \ 3080 sizeof((f2fs_inode)->field)) \ 3081 <= (F2FS_OLD_ATTRIBUTE_SIZE + (extra_isize))) \ 3082 3083 #define DEFAULT_IOSTAT_PERIOD_MS 3000 3084 #define MIN_IOSTAT_PERIOD_MS 100 3085 /* maximum period of iostat tracing is 1 day */ 3086 #define MAX_IOSTAT_PERIOD_MS 8640000 3087 3088 static inline void f2fs_reset_iostat(struct f2fs_sb_info *sbi) 3089 { 3090 int i; 3091 3092 spin_lock(&sbi->iostat_lock); 3093 for (i = 0; i < NR_IO_TYPE; i++) { 3094 sbi->rw_iostat[i] = 0; 3095 sbi->prev_rw_iostat[i] = 0; 3096 } 3097 spin_unlock(&sbi->iostat_lock); 3098 } 3099 3100 extern void f2fs_record_iostat(struct f2fs_sb_info *sbi); 3101 3102 static inline void f2fs_update_iostat(struct f2fs_sb_info *sbi, 3103 enum iostat_type type, unsigned long long io_bytes) 3104 { 3105 if (!sbi->iostat_enable) 3106 return; 3107 spin_lock(&sbi->iostat_lock); 3108 sbi->rw_iostat[type] += io_bytes; 3109 3110 if (type == APP_WRITE_IO || type == APP_DIRECT_IO) 3111 sbi->rw_iostat[APP_BUFFERED_IO] = 3112 sbi->rw_iostat[APP_WRITE_IO] - 3113 sbi->rw_iostat[APP_DIRECT_IO]; 3114 3115 if (type == APP_READ_IO || type == APP_DIRECT_READ_IO) 3116 sbi->rw_iostat[APP_BUFFERED_READ_IO] = 3117 sbi->rw_iostat[APP_READ_IO] - 3118 sbi->rw_iostat[APP_DIRECT_READ_IO]; 3119 spin_unlock(&sbi->iostat_lock); 3120 3121 f2fs_record_iostat(sbi); 3122 } 3123 3124 #define __is_large_section(sbi) ((sbi)->segs_per_sec > 1) 3125 3126 #define __is_meta_io(fio) (PAGE_TYPE_OF_BIO((fio)->type) == META) 3127 3128 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi, 3129 block_t blkaddr, int type); 3130 static inline void verify_blkaddr(struct f2fs_sb_info *sbi, 3131 block_t blkaddr, int type) 3132 { 3133 if (!f2fs_is_valid_blkaddr(sbi, blkaddr, type)) { 3134 f2fs_err(sbi, "invalid blkaddr: %u, type: %d, run fsck to fix.", 3135 blkaddr, type); 3136 f2fs_bug_on(sbi, 1); 3137 } 3138 } 3139 3140 static inline bool __is_valid_data_blkaddr(block_t blkaddr) 3141 { 3142 if (blkaddr == NEW_ADDR || blkaddr == NULL_ADDR || 3143 blkaddr == COMPRESS_ADDR) 3144 return false; 3145 return true; 3146 } 3147 3148 static inline void f2fs_set_page_private(struct page *page, 3149 unsigned long data) 3150 { 3151 if (PagePrivate(page)) 3152 return; 3153 3154 attach_page_private(page, (void *)data); 3155 } 3156 3157 static inline void f2fs_clear_page_private(struct page *page) 3158 { 3159 detach_page_private(page); 3160 } 3161 3162 /* 3163 * file.c 3164 */ 3165 int f2fs_sync_file(struct file *file, loff_t start, loff_t end, int datasync); 3166 void f2fs_truncate_data_blocks(struct dnode_of_data *dn); 3167 int f2fs_do_truncate_blocks(struct inode *inode, u64 from, bool lock); 3168 int f2fs_truncate_blocks(struct inode *inode, u64 from, bool lock); 3169 int f2fs_truncate(struct inode *inode); 3170 int f2fs_getattr(const struct path *path, struct kstat *stat, 3171 u32 request_mask, unsigned int flags); 3172 int f2fs_setattr(struct dentry *dentry, struct iattr *attr); 3173 int f2fs_truncate_hole(struct inode *inode, pgoff_t pg_start, pgoff_t pg_end); 3174 void f2fs_truncate_data_blocks_range(struct dnode_of_data *dn, int count); 3175 int f2fs_precache_extents(struct inode *inode); 3176 long f2fs_ioctl(struct file *filp, unsigned int cmd, unsigned long arg); 3177 long f2fs_compat_ioctl(struct file *file, unsigned int cmd, unsigned long arg); 3178 int f2fs_transfer_project_quota(struct inode *inode, kprojid_t kprojid); 3179 int f2fs_pin_file_control(struct inode *inode, bool inc); 3180 3181 /* 3182 * inode.c 3183 */ 3184 void f2fs_set_inode_flags(struct inode *inode); 3185 bool f2fs_inode_chksum_verify(struct f2fs_sb_info *sbi, struct page *page); 3186 void f2fs_inode_chksum_set(struct f2fs_sb_info *sbi, struct page *page); 3187 struct inode *f2fs_iget(struct super_block *sb, unsigned long ino); 3188 struct inode *f2fs_iget_retry(struct super_block *sb, unsigned long ino); 3189 int f2fs_try_to_free_nats(struct f2fs_sb_info *sbi, int nr_shrink); 3190 void f2fs_update_inode(struct inode *inode, struct page *node_page); 3191 void f2fs_update_inode_page(struct inode *inode); 3192 int f2fs_write_inode(struct inode *inode, struct writeback_control *wbc); 3193 void f2fs_evict_inode(struct inode *inode); 3194 void f2fs_handle_failed_inode(struct inode *inode); 3195 3196 /* 3197 * namei.c 3198 */ 3199 int f2fs_update_extension_list(struct f2fs_sb_info *sbi, const char *name, 3200 bool hot, bool set); 3201 struct dentry *f2fs_get_parent(struct dentry *child); 3202 3203 /* 3204 * dir.c 3205 */ 3206 unsigned char f2fs_get_de_type(struct f2fs_dir_entry *de); 3207 int f2fs_init_casefolded_name(const struct inode *dir, 3208 struct f2fs_filename *fname); 3209 int f2fs_setup_filename(struct inode *dir, const struct qstr *iname, 3210 int lookup, struct f2fs_filename *fname); 3211 int f2fs_prepare_lookup(struct inode *dir, struct dentry *dentry, 3212 struct f2fs_filename *fname); 3213 void f2fs_free_filename(struct f2fs_filename *fname); 3214 struct f2fs_dir_entry *f2fs_find_target_dentry(const struct f2fs_dentry_ptr *d, 3215 const struct f2fs_filename *fname, int *max_slots); 3216 int f2fs_fill_dentries(struct dir_context *ctx, struct f2fs_dentry_ptr *d, 3217 unsigned int start_pos, struct fscrypt_str *fstr); 3218 void f2fs_do_make_empty_dir(struct inode *inode, struct inode *parent, 3219 struct f2fs_dentry_ptr *d); 3220 struct page *f2fs_init_inode_metadata(struct inode *inode, struct inode *dir, 3221 const struct f2fs_filename *fname, struct page *dpage); 3222 void f2fs_update_parent_metadata(struct inode *dir, struct inode *inode, 3223 unsigned int current_depth); 3224 int f2fs_room_for_filename(const void *bitmap, int slots, int max_slots); 3225 void f2fs_drop_nlink(struct inode *dir, struct inode *inode); 3226 struct f2fs_dir_entry *__f2fs_find_entry(struct inode *dir, 3227 const struct f2fs_filename *fname, 3228 struct page **res_page); 3229 struct f2fs_dir_entry *f2fs_find_entry(struct inode *dir, 3230 const struct qstr *child, struct page **res_page); 3231 struct f2fs_dir_entry *f2fs_parent_dir(struct inode *dir, struct page **p); 3232 ino_t f2fs_inode_by_name(struct inode *dir, const struct qstr *qstr, 3233 struct page **page); 3234 void f2fs_set_link(struct inode *dir, struct f2fs_dir_entry *de, 3235 struct page *page, struct inode *inode); 3236 bool f2fs_has_enough_room(struct inode *dir, struct page *ipage, 3237 const struct f2fs_filename *fname); 3238 void f2fs_update_dentry(nid_t ino, umode_t mode, struct f2fs_dentry_ptr *d, 3239 const struct fscrypt_str *name, f2fs_hash_t name_hash, 3240 unsigned int bit_pos); 3241 int f2fs_add_regular_entry(struct inode *dir, const struct f2fs_filename *fname, 3242 struct inode *inode, nid_t ino, umode_t mode); 3243 int f2fs_add_dentry(struct inode *dir, const struct f2fs_filename *fname, 3244 struct inode *inode, nid_t ino, umode_t mode); 3245 int f2fs_do_add_link(struct inode *dir, const struct qstr *name, 3246 struct inode *inode, nid_t ino, umode_t mode); 3247 void f2fs_delete_entry(struct f2fs_dir_entry *dentry, struct page *page, 3248 struct inode *dir, struct inode *inode); 3249 int f2fs_do_tmpfile(struct inode *inode, struct inode *dir); 3250 bool f2fs_empty_dir(struct inode *dir); 3251 3252 static inline int f2fs_add_link(struct dentry *dentry, struct inode *inode) 3253 { 3254 if (fscrypt_is_nokey_name(dentry)) 3255 return -ENOKEY; 3256 return f2fs_do_add_link(d_inode(dentry->d_parent), &dentry->d_name, 3257 inode, inode->i_ino, inode->i_mode); 3258 } 3259 3260 /* 3261 * super.c 3262 */ 3263 int f2fs_inode_dirtied(struct inode *inode, bool sync); 3264 void f2fs_inode_synced(struct inode *inode); 3265 int f2fs_enable_quota_files(struct f2fs_sb_info *sbi, bool rdonly); 3266 int f2fs_quota_sync(struct super_block *sb, int type); 3267 loff_t max_file_blocks(struct inode *inode); 3268 void f2fs_quota_off_umount(struct super_block *sb); 3269 int f2fs_commit_super(struct f2fs_sb_info *sbi, bool recover); 3270 int f2fs_sync_fs(struct super_block *sb, int sync); 3271 int f2fs_sanity_check_ckpt(struct f2fs_sb_info *sbi); 3272 3273 /* 3274 * hash.c 3275 */ 3276 void f2fs_hash_filename(const struct inode *dir, struct f2fs_filename *fname); 3277 3278 /* 3279 * node.c 3280 */ 3281 struct dnode_of_data; 3282 struct node_info; 3283 3284 int f2fs_check_nid_range(struct f2fs_sb_info *sbi, nid_t nid); 3285 bool f2fs_available_free_memory(struct f2fs_sb_info *sbi, int type); 3286 bool f2fs_in_warm_node_list(struct f2fs_sb_info *sbi, struct page *page); 3287 void f2fs_init_fsync_node_info(struct f2fs_sb_info *sbi); 3288 void f2fs_del_fsync_node_entry(struct f2fs_sb_info *sbi, struct page *page); 3289 void f2fs_reset_fsync_node_info(struct f2fs_sb_info *sbi); 3290 int f2fs_need_dentry_mark(struct f2fs_sb_info *sbi, nid_t nid); 3291 bool f2fs_is_checkpointed_node(struct f2fs_sb_info *sbi, nid_t nid); 3292 bool f2fs_need_inode_block_update(struct f2fs_sb_info *sbi, nid_t ino); 3293 int f2fs_get_node_info(struct f2fs_sb_info *sbi, nid_t nid, 3294 struct node_info *ni); 3295 pgoff_t f2fs_get_next_page_offset(struct dnode_of_data *dn, pgoff_t pgofs); 3296 int f2fs_get_dnode_of_data(struct dnode_of_data *dn, pgoff_t index, int mode); 3297 int f2fs_truncate_inode_blocks(struct inode *inode, pgoff_t from); 3298 int f2fs_truncate_xattr_node(struct inode *inode); 3299 int f2fs_wait_on_node_pages_writeback(struct f2fs_sb_info *sbi, 3300 unsigned int seq_id); 3301 int f2fs_remove_inode_page(struct inode *inode); 3302 struct page *f2fs_new_inode_page(struct inode *inode); 3303 struct page *f2fs_new_node_page(struct dnode_of_data *dn, unsigned int ofs); 3304 void f2fs_ra_node_page(struct f2fs_sb_info *sbi, nid_t nid); 3305 struct page *f2fs_get_node_page(struct f2fs_sb_info *sbi, pgoff_t nid); 3306 struct page *f2fs_get_node_page_ra(struct page *parent, int start); 3307 int f2fs_move_node_page(struct page *node_page, int gc_type); 3308 void f2fs_flush_inline_data(struct f2fs_sb_info *sbi); 3309 int f2fs_fsync_node_pages(struct f2fs_sb_info *sbi, struct inode *inode, 3310 struct writeback_control *wbc, bool atomic, 3311 unsigned int *seq_id); 3312 int f2fs_sync_node_pages(struct f2fs_sb_info *sbi, 3313 struct writeback_control *wbc, 3314 bool do_balance, enum iostat_type io_type); 3315 int f2fs_build_free_nids(struct f2fs_sb_info *sbi, bool sync, bool mount); 3316 bool f2fs_alloc_nid(struct f2fs_sb_info *sbi, nid_t *nid); 3317 void f2fs_alloc_nid_done(struct f2fs_sb_info *sbi, nid_t nid); 3318 void f2fs_alloc_nid_failed(struct f2fs_sb_info *sbi, nid_t nid); 3319 int f2fs_try_to_free_nids(struct f2fs_sb_info *sbi, int nr_shrink); 3320 int f2fs_recover_inline_xattr(struct inode *inode, struct page *page); 3321 int f2fs_recover_xattr_data(struct inode *inode, struct page *page); 3322 int f2fs_recover_inode_page(struct f2fs_sb_info *sbi, struct page *page); 3323 int f2fs_restore_node_summary(struct f2fs_sb_info *sbi, 3324 unsigned int segno, struct f2fs_summary_block *sum); 3325 int f2fs_flush_nat_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3326 int f2fs_build_node_manager(struct f2fs_sb_info *sbi); 3327 void f2fs_destroy_node_manager(struct f2fs_sb_info *sbi); 3328 int __init f2fs_create_node_manager_caches(void); 3329 void f2fs_destroy_node_manager_caches(void); 3330 3331 /* 3332 * segment.c 3333 */ 3334 bool f2fs_need_SSR(struct f2fs_sb_info *sbi); 3335 void f2fs_register_inmem_page(struct inode *inode, struct page *page); 3336 void f2fs_drop_inmem_pages_all(struct f2fs_sb_info *sbi, bool gc_failure); 3337 void f2fs_drop_inmem_pages(struct inode *inode); 3338 void f2fs_drop_inmem_page(struct inode *inode, struct page *page); 3339 int f2fs_commit_inmem_pages(struct inode *inode); 3340 void f2fs_balance_fs(struct f2fs_sb_info *sbi, bool need); 3341 void f2fs_balance_fs_bg(struct f2fs_sb_info *sbi, bool from_bg); 3342 int f2fs_issue_flush(struct f2fs_sb_info *sbi, nid_t ino); 3343 int f2fs_create_flush_cmd_control(struct f2fs_sb_info *sbi); 3344 int f2fs_flush_device_cache(struct f2fs_sb_info *sbi); 3345 void f2fs_destroy_flush_cmd_control(struct f2fs_sb_info *sbi, bool free); 3346 void f2fs_invalidate_blocks(struct f2fs_sb_info *sbi, block_t addr); 3347 bool f2fs_is_checkpointed_data(struct f2fs_sb_info *sbi, block_t blkaddr); 3348 void f2fs_drop_discard_cmd(struct f2fs_sb_info *sbi); 3349 void f2fs_stop_discard_thread(struct f2fs_sb_info *sbi); 3350 bool f2fs_issue_discard_timeout(struct f2fs_sb_info *sbi); 3351 void f2fs_clear_prefree_segments(struct f2fs_sb_info *sbi, 3352 struct cp_control *cpc); 3353 void f2fs_dirty_to_prefree(struct f2fs_sb_info *sbi); 3354 block_t f2fs_get_unusable_blocks(struct f2fs_sb_info *sbi); 3355 int f2fs_disable_cp_again(struct f2fs_sb_info *sbi, block_t unusable); 3356 void f2fs_release_discard_addrs(struct f2fs_sb_info *sbi); 3357 int f2fs_npages_for_summary_flush(struct f2fs_sb_info *sbi, bool for_ra); 3358 void f2fs_init_inmem_curseg(struct f2fs_sb_info *sbi); 3359 void f2fs_save_inmem_curseg(struct f2fs_sb_info *sbi); 3360 void f2fs_restore_inmem_curseg(struct f2fs_sb_info *sbi); 3361 void f2fs_get_new_segment(struct f2fs_sb_info *sbi, 3362 unsigned int *newseg, bool new_sec, int dir); 3363 void f2fs_allocate_segment_for_resize(struct f2fs_sb_info *sbi, int type, 3364 unsigned int start, unsigned int end); 3365 void f2fs_allocate_new_segment(struct f2fs_sb_info *sbi, int type); 3366 void f2fs_allocate_new_segments(struct f2fs_sb_info *sbi); 3367 int f2fs_trim_fs(struct f2fs_sb_info *sbi, struct fstrim_range *range); 3368 bool f2fs_exist_trim_candidates(struct f2fs_sb_info *sbi, 3369 struct cp_control *cpc); 3370 struct page *f2fs_get_sum_page(struct f2fs_sb_info *sbi, unsigned int segno); 3371 void f2fs_update_meta_page(struct f2fs_sb_info *sbi, void *src, 3372 block_t blk_addr); 3373 void f2fs_do_write_meta_page(struct f2fs_sb_info *sbi, struct page *page, 3374 enum iostat_type io_type); 3375 void f2fs_do_write_node_page(unsigned int nid, struct f2fs_io_info *fio); 3376 void f2fs_outplace_write_data(struct dnode_of_data *dn, 3377 struct f2fs_io_info *fio); 3378 int f2fs_inplace_write_data(struct f2fs_io_info *fio); 3379 void f2fs_do_replace_block(struct f2fs_sb_info *sbi, struct f2fs_summary *sum, 3380 block_t old_blkaddr, block_t new_blkaddr, 3381 bool recover_curseg, bool recover_newaddr, 3382 bool from_gc); 3383 void f2fs_replace_block(struct f2fs_sb_info *sbi, struct dnode_of_data *dn, 3384 block_t old_addr, block_t new_addr, 3385 unsigned char version, bool recover_curseg, 3386 bool recover_newaddr); 3387 void f2fs_allocate_data_block(struct f2fs_sb_info *sbi, struct page *page, 3388 block_t old_blkaddr, block_t *new_blkaddr, 3389 struct f2fs_summary *sum, int type, 3390 struct f2fs_io_info *fio); 3391 void f2fs_wait_on_page_writeback(struct page *page, 3392 enum page_type type, bool ordered, bool locked); 3393 void f2fs_wait_on_block_writeback(struct inode *inode, block_t blkaddr); 3394 void f2fs_wait_on_block_writeback_range(struct inode *inode, block_t blkaddr, 3395 block_t len); 3396 void f2fs_write_data_summaries(struct f2fs_sb_info *sbi, block_t start_blk); 3397 void f2fs_write_node_summaries(struct f2fs_sb_info *sbi, block_t start_blk); 3398 int f2fs_lookup_journal_in_cursum(struct f2fs_journal *journal, int type, 3399 unsigned int val, int alloc); 3400 void f2fs_flush_sit_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3401 int f2fs_fix_curseg_write_pointer(struct f2fs_sb_info *sbi); 3402 int f2fs_check_write_pointer(struct f2fs_sb_info *sbi); 3403 int f2fs_build_segment_manager(struct f2fs_sb_info *sbi); 3404 void f2fs_destroy_segment_manager(struct f2fs_sb_info *sbi); 3405 int __init f2fs_create_segment_manager_caches(void); 3406 void f2fs_destroy_segment_manager_caches(void); 3407 int f2fs_rw_hint_to_seg_type(enum rw_hint hint); 3408 enum rw_hint f2fs_io_type_to_rw_hint(struct f2fs_sb_info *sbi, 3409 enum page_type type, enum temp_type temp); 3410 unsigned int f2fs_usable_segs_in_sec(struct f2fs_sb_info *sbi, 3411 unsigned int segno); 3412 unsigned int f2fs_usable_blks_in_seg(struct f2fs_sb_info *sbi, 3413 unsigned int segno); 3414 3415 /* 3416 * checkpoint.c 3417 */ 3418 void f2fs_stop_checkpoint(struct f2fs_sb_info *sbi, bool end_io); 3419 struct page *f2fs_grab_meta_page(struct f2fs_sb_info *sbi, pgoff_t index); 3420 struct page *f2fs_get_meta_page(struct f2fs_sb_info *sbi, pgoff_t index); 3421 struct page *f2fs_get_meta_page_retry(struct f2fs_sb_info *sbi, pgoff_t index); 3422 struct page *f2fs_get_tmp_page(struct f2fs_sb_info *sbi, pgoff_t index); 3423 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi, 3424 block_t blkaddr, int type); 3425 int f2fs_ra_meta_pages(struct f2fs_sb_info *sbi, block_t start, int nrpages, 3426 int type, bool sync); 3427 void f2fs_ra_meta_pages_cond(struct f2fs_sb_info *sbi, pgoff_t index); 3428 long f2fs_sync_meta_pages(struct f2fs_sb_info *sbi, enum page_type type, 3429 long nr_to_write, enum iostat_type io_type); 3430 void f2fs_add_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type); 3431 void f2fs_remove_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type); 3432 void f2fs_release_ino_entry(struct f2fs_sb_info *sbi, bool all); 3433 bool f2fs_exist_written_data(struct f2fs_sb_info *sbi, nid_t ino, int mode); 3434 void f2fs_set_dirty_device(struct f2fs_sb_info *sbi, nid_t ino, 3435 unsigned int devidx, int type); 3436 bool f2fs_is_dirty_device(struct f2fs_sb_info *sbi, nid_t ino, 3437 unsigned int devidx, int type); 3438 int f2fs_sync_inode_meta(struct f2fs_sb_info *sbi); 3439 int f2fs_acquire_orphan_inode(struct f2fs_sb_info *sbi); 3440 void f2fs_release_orphan_inode(struct f2fs_sb_info *sbi); 3441 void f2fs_add_orphan_inode(struct inode *inode); 3442 void f2fs_remove_orphan_inode(struct f2fs_sb_info *sbi, nid_t ino); 3443 int f2fs_recover_orphan_inodes(struct f2fs_sb_info *sbi); 3444 int f2fs_get_valid_checkpoint(struct f2fs_sb_info *sbi); 3445 void f2fs_update_dirty_page(struct inode *inode, struct page *page); 3446 void f2fs_remove_dirty_inode(struct inode *inode); 3447 int f2fs_sync_dirty_inodes(struct f2fs_sb_info *sbi, enum inode_type type); 3448 void f2fs_wait_on_all_pages(struct f2fs_sb_info *sbi, int type); 3449 u64 f2fs_get_sectors_written(struct f2fs_sb_info *sbi); 3450 int f2fs_write_checkpoint(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3451 void f2fs_init_ino_entry_info(struct f2fs_sb_info *sbi); 3452 int __init f2fs_create_checkpoint_caches(void); 3453 void f2fs_destroy_checkpoint_caches(void); 3454 3455 /* 3456 * data.c 3457 */ 3458 int __init f2fs_init_bioset(void); 3459 void f2fs_destroy_bioset(void); 3460 struct bio *f2fs_bio_alloc(struct f2fs_sb_info *sbi, int npages, bool noio); 3461 int f2fs_init_bio_entry_cache(void); 3462 void f2fs_destroy_bio_entry_cache(void); 3463 void f2fs_submit_bio(struct f2fs_sb_info *sbi, 3464 struct bio *bio, enum page_type type); 3465 void f2fs_submit_merged_write(struct f2fs_sb_info *sbi, enum page_type type); 3466 void f2fs_submit_merged_write_cond(struct f2fs_sb_info *sbi, 3467 struct inode *inode, struct page *page, 3468 nid_t ino, enum page_type type); 3469 void f2fs_submit_merged_ipu_write(struct f2fs_sb_info *sbi, 3470 struct bio **bio, struct page *page); 3471 void f2fs_flush_merged_writes(struct f2fs_sb_info *sbi); 3472 int f2fs_submit_page_bio(struct f2fs_io_info *fio); 3473 int f2fs_merge_page_bio(struct f2fs_io_info *fio); 3474 void f2fs_submit_page_write(struct f2fs_io_info *fio); 3475 struct block_device *f2fs_target_device(struct f2fs_sb_info *sbi, 3476 block_t blk_addr, struct bio *bio); 3477 int f2fs_target_device_index(struct f2fs_sb_info *sbi, block_t blkaddr); 3478 void f2fs_set_data_blkaddr(struct dnode_of_data *dn); 3479 void f2fs_update_data_blkaddr(struct dnode_of_data *dn, block_t blkaddr); 3480 int f2fs_reserve_new_blocks(struct dnode_of_data *dn, blkcnt_t count); 3481 int f2fs_reserve_new_block(struct dnode_of_data *dn); 3482 int f2fs_get_block(struct dnode_of_data *dn, pgoff_t index); 3483 int f2fs_preallocate_blocks(struct kiocb *iocb, struct iov_iter *from); 3484 int f2fs_reserve_block(struct dnode_of_data *dn, pgoff_t index); 3485 struct page *f2fs_get_read_data_page(struct inode *inode, pgoff_t index, 3486 int op_flags, bool for_write); 3487 struct page *f2fs_find_data_page(struct inode *inode, pgoff_t index); 3488 struct page *f2fs_get_lock_data_page(struct inode *inode, pgoff_t index, 3489 bool for_write); 3490 struct page *f2fs_get_new_data_page(struct inode *inode, 3491 struct page *ipage, pgoff_t index, bool new_i_size); 3492 int f2fs_do_write_data_page(struct f2fs_io_info *fio); 3493 void f2fs_do_map_lock(struct f2fs_sb_info *sbi, int flag, bool lock); 3494 int f2fs_map_blocks(struct inode *inode, struct f2fs_map_blocks *map, 3495 int create, int flag); 3496 int f2fs_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, 3497 u64 start, u64 len); 3498 int f2fs_encrypt_one_page(struct f2fs_io_info *fio); 3499 bool f2fs_should_update_inplace(struct inode *inode, struct f2fs_io_info *fio); 3500 bool f2fs_should_update_outplace(struct inode *inode, struct f2fs_io_info *fio); 3501 int f2fs_write_single_data_page(struct page *page, int *submitted, 3502 struct bio **bio, sector_t *last_block, 3503 struct writeback_control *wbc, 3504 enum iostat_type io_type, 3505 int compr_blocks, bool allow_balance); 3506 void f2fs_invalidate_page(struct page *page, unsigned int offset, 3507 unsigned int length); 3508 int f2fs_release_page(struct page *page, gfp_t wait); 3509 #ifdef CONFIG_MIGRATION 3510 int f2fs_migrate_page(struct address_space *mapping, struct page *newpage, 3511 struct page *page, enum migrate_mode mode); 3512 #endif 3513 bool f2fs_overwrite_io(struct inode *inode, loff_t pos, size_t len); 3514 void f2fs_clear_page_cache_dirty_tag(struct page *page); 3515 int f2fs_init_post_read_processing(void); 3516 void f2fs_destroy_post_read_processing(void); 3517 int f2fs_init_post_read_wq(struct f2fs_sb_info *sbi); 3518 void f2fs_destroy_post_read_wq(struct f2fs_sb_info *sbi); 3519 3520 /* 3521 * gc.c 3522 */ 3523 int f2fs_start_gc_thread(struct f2fs_sb_info *sbi); 3524 void f2fs_stop_gc_thread(struct f2fs_sb_info *sbi); 3525 block_t f2fs_start_bidx_of_node(unsigned int node_ofs, struct inode *inode); 3526 int f2fs_gc(struct f2fs_sb_info *sbi, bool sync, bool background, 3527 unsigned int segno); 3528 void f2fs_build_gc_manager(struct f2fs_sb_info *sbi); 3529 int f2fs_resize_fs(struct f2fs_sb_info *sbi, __u64 block_count); 3530 int __init f2fs_create_garbage_collection_cache(void); 3531 void f2fs_destroy_garbage_collection_cache(void); 3532 3533 /* 3534 * recovery.c 3535 */ 3536 int f2fs_recover_fsync_data(struct f2fs_sb_info *sbi, bool check_only); 3537 bool f2fs_space_for_roll_forward(struct f2fs_sb_info *sbi); 3538 3539 /* 3540 * debug.c 3541 */ 3542 #ifdef CONFIG_F2FS_STAT_FS 3543 struct f2fs_stat_info { 3544 struct list_head stat_list; 3545 struct f2fs_sb_info *sbi; 3546 int all_area_segs, sit_area_segs, nat_area_segs, ssa_area_segs; 3547 int main_area_segs, main_area_sections, main_area_zones; 3548 unsigned long long hit_largest, hit_cached, hit_rbtree; 3549 unsigned long long hit_total, total_ext; 3550 int ext_tree, zombie_tree, ext_node; 3551 int ndirty_node, ndirty_dent, ndirty_meta, ndirty_imeta; 3552 int ndirty_data, ndirty_qdata; 3553 int inmem_pages; 3554 unsigned int ndirty_dirs, ndirty_files, nquota_files, ndirty_all; 3555 int nats, dirty_nats, sits, dirty_sits; 3556 int free_nids, avail_nids, alloc_nids; 3557 int total_count, utilization; 3558 int bg_gc, nr_wb_cp_data, nr_wb_data; 3559 int nr_rd_data, nr_rd_node, nr_rd_meta; 3560 int nr_dio_read, nr_dio_write; 3561 unsigned int io_skip_bggc, other_skip_bggc; 3562 int nr_flushing, nr_flushed, flush_list_empty; 3563 int nr_discarding, nr_discarded; 3564 int nr_discard_cmd; 3565 unsigned int undiscard_blks; 3566 int inline_xattr, inline_inode, inline_dir, append, update, orphans; 3567 int compr_inode; 3568 unsigned long long compr_blocks; 3569 int aw_cnt, max_aw_cnt, vw_cnt, max_vw_cnt; 3570 unsigned int valid_count, valid_node_count, valid_inode_count, discard_blks; 3571 unsigned int bimodal, avg_vblocks; 3572 int util_free, util_valid, util_invalid; 3573 int rsvd_segs, overp_segs; 3574 int dirty_count, node_pages, meta_pages; 3575 int prefree_count, call_count, cp_count, bg_cp_count; 3576 int tot_segs, node_segs, data_segs, free_segs, free_secs; 3577 int bg_node_segs, bg_data_segs; 3578 int tot_blks, data_blks, node_blks; 3579 int bg_data_blks, bg_node_blks; 3580 unsigned long long skipped_atomic_files[2]; 3581 int curseg[NR_CURSEG_TYPE]; 3582 int cursec[NR_CURSEG_TYPE]; 3583 int curzone[NR_CURSEG_TYPE]; 3584 unsigned int dirty_seg[NR_CURSEG_TYPE]; 3585 unsigned int full_seg[NR_CURSEG_TYPE]; 3586 unsigned int valid_blks[NR_CURSEG_TYPE]; 3587 3588 unsigned int meta_count[META_MAX]; 3589 unsigned int segment_count[2]; 3590 unsigned int block_count[2]; 3591 unsigned int inplace_count; 3592 unsigned long long base_mem, cache_mem, page_mem; 3593 }; 3594 3595 static inline struct f2fs_stat_info *F2FS_STAT(struct f2fs_sb_info *sbi) 3596 { 3597 return (struct f2fs_stat_info *)sbi->stat_info; 3598 } 3599 3600 #define stat_inc_cp_count(si) ((si)->cp_count++) 3601 #define stat_inc_bg_cp_count(si) ((si)->bg_cp_count++) 3602 #define stat_inc_call_count(si) ((si)->call_count++) 3603 #define stat_inc_bggc_count(si) ((si)->bg_gc++) 3604 #define stat_io_skip_bggc_count(sbi) ((sbi)->io_skip_bggc++) 3605 #define stat_other_skip_bggc_count(sbi) ((sbi)->other_skip_bggc++) 3606 #define stat_inc_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]++) 3607 #define stat_dec_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]--) 3608 #define stat_inc_total_hit(sbi) (atomic64_inc(&(sbi)->total_hit_ext)) 3609 #define stat_inc_rbtree_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_rbtree)) 3610 #define stat_inc_largest_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_largest)) 3611 #define stat_inc_cached_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_cached)) 3612 #define stat_inc_inline_xattr(inode) \ 3613 do { \ 3614 if (f2fs_has_inline_xattr(inode)) \ 3615 (atomic_inc(&F2FS_I_SB(inode)->inline_xattr)); \ 3616 } while (0) 3617 #define stat_dec_inline_xattr(inode) \ 3618 do { \ 3619 if (f2fs_has_inline_xattr(inode)) \ 3620 (atomic_dec(&F2FS_I_SB(inode)->inline_xattr)); \ 3621 } while (0) 3622 #define stat_inc_inline_inode(inode) \ 3623 do { \ 3624 if (f2fs_has_inline_data(inode)) \ 3625 (atomic_inc(&F2FS_I_SB(inode)->inline_inode)); \ 3626 } while (0) 3627 #define stat_dec_inline_inode(inode) \ 3628 do { \ 3629 if (f2fs_has_inline_data(inode)) \ 3630 (atomic_dec(&F2FS_I_SB(inode)->inline_inode)); \ 3631 } while (0) 3632 #define stat_inc_inline_dir(inode) \ 3633 do { \ 3634 if (f2fs_has_inline_dentry(inode)) \ 3635 (atomic_inc(&F2FS_I_SB(inode)->inline_dir)); \ 3636 } while (0) 3637 #define stat_dec_inline_dir(inode) \ 3638 do { \ 3639 if (f2fs_has_inline_dentry(inode)) \ 3640 (atomic_dec(&F2FS_I_SB(inode)->inline_dir)); \ 3641 } while (0) 3642 #define stat_inc_compr_inode(inode) \ 3643 do { \ 3644 if (f2fs_compressed_file(inode)) \ 3645 (atomic_inc(&F2FS_I_SB(inode)->compr_inode)); \ 3646 } while (0) 3647 #define stat_dec_compr_inode(inode) \ 3648 do { \ 3649 if (f2fs_compressed_file(inode)) \ 3650 (atomic_dec(&F2FS_I_SB(inode)->compr_inode)); \ 3651 } while (0) 3652 #define stat_add_compr_blocks(inode, blocks) \ 3653 (atomic64_add(blocks, &F2FS_I_SB(inode)->compr_blocks)) 3654 #define stat_sub_compr_blocks(inode, blocks) \ 3655 (atomic64_sub(blocks, &F2FS_I_SB(inode)->compr_blocks)) 3656 #define stat_inc_meta_count(sbi, blkaddr) \ 3657 do { \ 3658 if (blkaddr < SIT_I(sbi)->sit_base_addr) \ 3659 atomic_inc(&(sbi)->meta_count[META_CP]); \ 3660 else if (blkaddr < NM_I(sbi)->nat_blkaddr) \ 3661 atomic_inc(&(sbi)->meta_count[META_SIT]); \ 3662 else if (blkaddr < SM_I(sbi)->ssa_blkaddr) \ 3663 atomic_inc(&(sbi)->meta_count[META_NAT]); \ 3664 else if (blkaddr < SM_I(sbi)->main_blkaddr) \ 3665 atomic_inc(&(sbi)->meta_count[META_SSA]); \ 3666 } while (0) 3667 #define stat_inc_seg_type(sbi, curseg) \ 3668 ((sbi)->segment_count[(curseg)->alloc_type]++) 3669 #define stat_inc_block_count(sbi, curseg) \ 3670 ((sbi)->block_count[(curseg)->alloc_type]++) 3671 #define stat_inc_inplace_blocks(sbi) \ 3672 (atomic_inc(&(sbi)->inplace_count)) 3673 #define stat_update_max_atomic_write(inode) \ 3674 do { \ 3675 int cur = F2FS_I_SB(inode)->atomic_files; \ 3676 int max = atomic_read(&F2FS_I_SB(inode)->max_aw_cnt); \ 3677 if (cur > max) \ 3678 atomic_set(&F2FS_I_SB(inode)->max_aw_cnt, cur); \ 3679 } while (0) 3680 #define stat_inc_volatile_write(inode) \ 3681 (atomic_inc(&F2FS_I_SB(inode)->vw_cnt)) 3682 #define stat_dec_volatile_write(inode) \ 3683 (atomic_dec(&F2FS_I_SB(inode)->vw_cnt)) 3684 #define stat_update_max_volatile_write(inode) \ 3685 do { \ 3686 int cur = atomic_read(&F2FS_I_SB(inode)->vw_cnt); \ 3687 int max = atomic_read(&F2FS_I_SB(inode)->max_vw_cnt); \ 3688 if (cur > max) \ 3689 atomic_set(&F2FS_I_SB(inode)->max_vw_cnt, cur); \ 3690 } while (0) 3691 #define stat_inc_seg_count(sbi, type, gc_type) \ 3692 do { \ 3693 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 3694 si->tot_segs++; \ 3695 if ((type) == SUM_TYPE_DATA) { \ 3696 si->data_segs++; \ 3697 si->bg_data_segs += (gc_type == BG_GC) ? 1 : 0; \ 3698 } else { \ 3699 si->node_segs++; \ 3700 si->bg_node_segs += (gc_type == BG_GC) ? 1 : 0; \ 3701 } \ 3702 } while (0) 3703 3704 #define stat_inc_tot_blk_count(si, blks) \ 3705 ((si)->tot_blks += (blks)) 3706 3707 #define stat_inc_data_blk_count(sbi, blks, gc_type) \ 3708 do { \ 3709 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 3710 stat_inc_tot_blk_count(si, blks); \ 3711 si->data_blks += (blks); \ 3712 si->bg_data_blks += ((gc_type) == BG_GC) ? (blks) : 0; \ 3713 } while (0) 3714 3715 #define stat_inc_node_blk_count(sbi, blks, gc_type) \ 3716 do { \ 3717 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 3718 stat_inc_tot_blk_count(si, blks); \ 3719 si->node_blks += (blks); \ 3720 si->bg_node_blks += ((gc_type) == BG_GC) ? (blks) : 0; \ 3721 } while (0) 3722 3723 int f2fs_build_stats(struct f2fs_sb_info *sbi); 3724 void f2fs_destroy_stats(struct f2fs_sb_info *sbi); 3725 void __init f2fs_create_root_stats(void); 3726 void f2fs_destroy_root_stats(void); 3727 void f2fs_update_sit_info(struct f2fs_sb_info *sbi); 3728 #else 3729 #define stat_inc_cp_count(si) do { } while (0) 3730 #define stat_inc_bg_cp_count(si) do { } while (0) 3731 #define stat_inc_call_count(si) do { } while (0) 3732 #define stat_inc_bggc_count(si) do { } while (0) 3733 #define stat_io_skip_bggc_count(sbi) do { } while (0) 3734 #define stat_other_skip_bggc_count(sbi) do { } while (0) 3735 #define stat_inc_dirty_inode(sbi, type) do { } while (0) 3736 #define stat_dec_dirty_inode(sbi, type) do { } while (0) 3737 #define stat_inc_total_hit(sbi) do { } while (0) 3738 #define stat_inc_rbtree_node_hit(sbi) do { } while (0) 3739 #define stat_inc_largest_node_hit(sbi) do { } while (0) 3740 #define stat_inc_cached_node_hit(sbi) do { } while (0) 3741 #define stat_inc_inline_xattr(inode) do { } while (0) 3742 #define stat_dec_inline_xattr(inode) do { } while (0) 3743 #define stat_inc_inline_inode(inode) do { } while (0) 3744 #define stat_dec_inline_inode(inode) do { } while (0) 3745 #define stat_inc_inline_dir(inode) do { } while (0) 3746 #define stat_dec_inline_dir(inode) do { } while (0) 3747 #define stat_inc_compr_inode(inode) do { } while (0) 3748 #define stat_dec_compr_inode(inode) do { } while (0) 3749 #define stat_add_compr_blocks(inode, blocks) do { } while (0) 3750 #define stat_sub_compr_blocks(inode, blocks) do { } while (0) 3751 #define stat_update_max_atomic_write(inode) do { } while (0) 3752 #define stat_inc_volatile_write(inode) do { } while (0) 3753 #define stat_dec_volatile_write(inode) do { } while (0) 3754 #define stat_update_max_volatile_write(inode) do { } while (0) 3755 #define stat_inc_meta_count(sbi, blkaddr) do { } while (0) 3756 #define stat_inc_seg_type(sbi, curseg) do { } while (0) 3757 #define stat_inc_block_count(sbi, curseg) do { } while (0) 3758 #define stat_inc_inplace_blocks(sbi) do { } while (0) 3759 #define stat_inc_seg_count(sbi, type, gc_type) do { } while (0) 3760 #define stat_inc_tot_blk_count(si, blks) do { } while (0) 3761 #define stat_inc_data_blk_count(sbi, blks, gc_type) do { } while (0) 3762 #define stat_inc_node_blk_count(sbi, blks, gc_type) do { } while (0) 3763 3764 static inline int f2fs_build_stats(struct f2fs_sb_info *sbi) { return 0; } 3765 static inline void f2fs_destroy_stats(struct f2fs_sb_info *sbi) { } 3766 static inline void __init f2fs_create_root_stats(void) { } 3767 static inline void f2fs_destroy_root_stats(void) { } 3768 static inline void f2fs_update_sit_info(struct f2fs_sb_info *sbi) {} 3769 #endif 3770 3771 extern const struct file_operations f2fs_dir_operations; 3772 extern const struct file_operations f2fs_file_operations; 3773 extern const struct inode_operations f2fs_file_inode_operations; 3774 extern const struct address_space_operations f2fs_dblock_aops; 3775 extern const struct address_space_operations f2fs_node_aops; 3776 extern const struct address_space_operations f2fs_meta_aops; 3777 extern const struct inode_operations f2fs_dir_inode_operations; 3778 extern const struct inode_operations f2fs_symlink_inode_operations; 3779 extern const struct inode_operations f2fs_encrypted_symlink_inode_operations; 3780 extern const struct inode_operations f2fs_special_inode_operations; 3781 extern struct kmem_cache *f2fs_inode_entry_slab; 3782 3783 /* 3784 * inline.c 3785 */ 3786 bool f2fs_may_inline_data(struct inode *inode); 3787 bool f2fs_may_inline_dentry(struct inode *inode); 3788 void f2fs_do_read_inline_data(struct page *page, struct page *ipage); 3789 void f2fs_truncate_inline_inode(struct inode *inode, 3790 struct page *ipage, u64 from); 3791 int f2fs_read_inline_data(struct inode *inode, struct page *page); 3792 int f2fs_convert_inline_page(struct dnode_of_data *dn, struct page *page); 3793 int f2fs_convert_inline_inode(struct inode *inode); 3794 int f2fs_try_convert_inline_dir(struct inode *dir, struct dentry *dentry); 3795 int f2fs_write_inline_data(struct inode *inode, struct page *page); 3796 int f2fs_recover_inline_data(struct inode *inode, struct page *npage); 3797 struct f2fs_dir_entry *f2fs_find_in_inline_dir(struct inode *dir, 3798 const struct f2fs_filename *fname, 3799 struct page **res_page); 3800 int f2fs_make_empty_inline_dir(struct inode *inode, struct inode *parent, 3801 struct page *ipage); 3802 int f2fs_add_inline_entry(struct inode *dir, const struct f2fs_filename *fname, 3803 struct inode *inode, nid_t ino, umode_t mode); 3804 void f2fs_delete_inline_entry(struct f2fs_dir_entry *dentry, 3805 struct page *page, struct inode *dir, 3806 struct inode *inode); 3807 bool f2fs_empty_inline_dir(struct inode *dir); 3808 int f2fs_read_inline_dir(struct file *file, struct dir_context *ctx, 3809 struct fscrypt_str *fstr); 3810 int f2fs_inline_data_fiemap(struct inode *inode, 3811 struct fiemap_extent_info *fieinfo, 3812 __u64 start, __u64 len); 3813 3814 /* 3815 * shrinker.c 3816 */ 3817 unsigned long f2fs_shrink_count(struct shrinker *shrink, 3818 struct shrink_control *sc); 3819 unsigned long f2fs_shrink_scan(struct shrinker *shrink, 3820 struct shrink_control *sc); 3821 void f2fs_join_shrinker(struct f2fs_sb_info *sbi); 3822 void f2fs_leave_shrinker(struct f2fs_sb_info *sbi); 3823 3824 /* 3825 * extent_cache.c 3826 */ 3827 struct rb_entry *f2fs_lookup_rb_tree(struct rb_root_cached *root, 3828 struct rb_entry *cached_re, unsigned int ofs); 3829 struct rb_node **f2fs_lookup_rb_tree_ext(struct f2fs_sb_info *sbi, 3830 struct rb_root_cached *root, 3831 struct rb_node **parent, 3832 unsigned long long key, bool *left_most); 3833 struct rb_node **f2fs_lookup_rb_tree_for_insert(struct f2fs_sb_info *sbi, 3834 struct rb_root_cached *root, 3835 struct rb_node **parent, 3836 unsigned int ofs, bool *leftmost); 3837 struct rb_entry *f2fs_lookup_rb_tree_ret(struct rb_root_cached *root, 3838 struct rb_entry *cached_re, unsigned int ofs, 3839 struct rb_entry **prev_entry, struct rb_entry **next_entry, 3840 struct rb_node ***insert_p, struct rb_node **insert_parent, 3841 bool force, bool *leftmost); 3842 bool f2fs_check_rb_tree_consistence(struct f2fs_sb_info *sbi, 3843 struct rb_root_cached *root, bool check_key); 3844 unsigned int f2fs_shrink_extent_tree(struct f2fs_sb_info *sbi, int nr_shrink); 3845 void f2fs_init_extent_tree(struct inode *inode, struct page *ipage); 3846 void f2fs_drop_extent_tree(struct inode *inode); 3847 unsigned int f2fs_destroy_extent_node(struct inode *inode); 3848 void f2fs_destroy_extent_tree(struct inode *inode); 3849 bool f2fs_lookup_extent_cache(struct inode *inode, pgoff_t pgofs, 3850 struct extent_info *ei); 3851 void f2fs_update_extent_cache(struct dnode_of_data *dn); 3852 void f2fs_update_extent_cache_range(struct dnode_of_data *dn, 3853 pgoff_t fofs, block_t blkaddr, unsigned int len); 3854 void f2fs_init_extent_cache_info(struct f2fs_sb_info *sbi); 3855 int __init f2fs_create_extent_cache(void); 3856 void f2fs_destroy_extent_cache(void); 3857 3858 /* 3859 * sysfs.c 3860 */ 3861 int __init f2fs_init_sysfs(void); 3862 void f2fs_exit_sysfs(void); 3863 int f2fs_register_sysfs(struct f2fs_sb_info *sbi); 3864 void f2fs_unregister_sysfs(struct f2fs_sb_info *sbi); 3865 3866 /* verity.c */ 3867 extern const struct fsverity_operations f2fs_verityops; 3868 3869 /* 3870 * crypto support 3871 */ 3872 static inline bool f2fs_encrypted_file(struct inode *inode) 3873 { 3874 return IS_ENCRYPTED(inode) && S_ISREG(inode->i_mode); 3875 } 3876 3877 static inline void f2fs_set_encrypted_inode(struct inode *inode) 3878 { 3879 #ifdef CONFIG_FS_ENCRYPTION 3880 file_set_encrypt(inode); 3881 f2fs_set_inode_flags(inode); 3882 #endif 3883 } 3884 3885 /* 3886 * Returns true if the reads of the inode's data need to undergo some 3887 * postprocessing step, like decryption or authenticity verification. 3888 */ 3889 static inline bool f2fs_post_read_required(struct inode *inode) 3890 { 3891 return f2fs_encrypted_file(inode) || fsverity_active(inode) || 3892 f2fs_compressed_file(inode); 3893 } 3894 3895 /* 3896 * compress.c 3897 */ 3898 #ifdef CONFIG_F2FS_FS_COMPRESSION 3899 bool f2fs_is_compressed_page(struct page *page); 3900 struct page *f2fs_compress_control_page(struct page *page); 3901 int f2fs_prepare_compress_overwrite(struct inode *inode, 3902 struct page **pagep, pgoff_t index, void **fsdata); 3903 bool f2fs_compress_write_end(struct inode *inode, void *fsdata, 3904 pgoff_t index, unsigned copied); 3905 int f2fs_truncate_partial_cluster(struct inode *inode, u64 from, bool lock); 3906 void f2fs_compress_write_end_io(struct bio *bio, struct page *page); 3907 bool f2fs_is_compress_backend_ready(struct inode *inode); 3908 int f2fs_init_compress_mempool(void); 3909 void f2fs_destroy_compress_mempool(void); 3910 void f2fs_end_read_compressed_page(struct page *page, bool failed); 3911 bool f2fs_cluster_is_empty(struct compress_ctx *cc); 3912 bool f2fs_cluster_can_merge_page(struct compress_ctx *cc, pgoff_t index); 3913 void f2fs_compress_ctx_add_page(struct compress_ctx *cc, struct page *page); 3914 int f2fs_write_multi_pages(struct compress_ctx *cc, 3915 int *submitted, 3916 struct writeback_control *wbc, 3917 enum iostat_type io_type); 3918 int f2fs_is_compressed_cluster(struct inode *inode, pgoff_t index); 3919 int f2fs_read_multi_pages(struct compress_ctx *cc, struct bio **bio_ret, 3920 unsigned nr_pages, sector_t *last_block_in_bio, 3921 bool is_readahead, bool for_write); 3922 struct decompress_io_ctx *f2fs_alloc_dic(struct compress_ctx *cc); 3923 void f2fs_decompress_end_io(struct decompress_io_ctx *dic, bool failed); 3924 void f2fs_put_page_dic(struct page *page); 3925 int f2fs_init_compress_ctx(struct compress_ctx *cc); 3926 void f2fs_destroy_compress_ctx(struct compress_ctx *cc); 3927 void f2fs_init_compress_info(struct f2fs_sb_info *sbi); 3928 int f2fs_init_page_array_cache(struct f2fs_sb_info *sbi); 3929 void f2fs_destroy_page_array_cache(struct f2fs_sb_info *sbi); 3930 int __init f2fs_init_compress_cache(void); 3931 void f2fs_destroy_compress_cache(void); 3932 #else 3933 static inline bool f2fs_is_compressed_page(struct page *page) { return false; } 3934 static inline bool f2fs_is_compress_backend_ready(struct inode *inode) 3935 { 3936 if (!f2fs_compressed_file(inode)) 3937 return true; 3938 /* not support compression */ 3939 return false; 3940 } 3941 static inline struct page *f2fs_compress_control_page(struct page *page) 3942 { 3943 WARN_ON_ONCE(1); 3944 return ERR_PTR(-EINVAL); 3945 } 3946 static inline int f2fs_init_compress_mempool(void) { return 0; } 3947 static inline void f2fs_destroy_compress_mempool(void) { } 3948 static inline void f2fs_end_read_compressed_page(struct page *page, bool failed) 3949 { 3950 WARN_ON_ONCE(1); 3951 } 3952 static inline void f2fs_put_page_dic(struct page *page) 3953 { 3954 WARN_ON_ONCE(1); 3955 } 3956 static inline int f2fs_init_page_array_cache(struct f2fs_sb_info *sbi) { return 0; } 3957 static inline void f2fs_destroy_page_array_cache(struct f2fs_sb_info *sbi) { } 3958 static inline int __init f2fs_init_compress_cache(void) { return 0; } 3959 static inline void f2fs_destroy_compress_cache(void) { } 3960 #endif 3961 3962 static inline void set_compress_context(struct inode *inode) 3963 { 3964 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 3965 3966 F2FS_I(inode)->i_compress_algorithm = 3967 F2FS_OPTION(sbi).compress_algorithm; 3968 F2FS_I(inode)->i_log_cluster_size = 3969 F2FS_OPTION(sbi).compress_log_size; 3970 F2FS_I(inode)->i_compress_flag = 3971 F2FS_OPTION(sbi).compress_chksum ? 3972 1 << COMPRESS_CHKSUM : 0; 3973 F2FS_I(inode)->i_cluster_size = 3974 1 << F2FS_I(inode)->i_log_cluster_size; 3975 if (F2FS_I(inode)->i_compress_algorithm == COMPRESS_LZ4 && 3976 F2FS_OPTION(sbi).compress_level) 3977 F2FS_I(inode)->i_compress_flag |= 3978 F2FS_OPTION(sbi).compress_level << 3979 COMPRESS_LEVEL_OFFSET; 3980 F2FS_I(inode)->i_flags |= F2FS_COMPR_FL; 3981 set_inode_flag(inode, FI_COMPRESSED_FILE); 3982 stat_inc_compr_inode(inode); 3983 f2fs_mark_inode_dirty_sync(inode, true); 3984 } 3985 3986 static inline bool f2fs_disable_compressed_file(struct inode *inode) 3987 { 3988 struct f2fs_inode_info *fi = F2FS_I(inode); 3989 3990 if (!f2fs_compressed_file(inode)) 3991 return true; 3992 if (S_ISREG(inode->i_mode) && 3993 (get_dirty_pages(inode) || atomic_read(&fi->i_compr_blocks))) 3994 return false; 3995 3996 fi->i_flags &= ~F2FS_COMPR_FL; 3997 stat_dec_compr_inode(inode); 3998 clear_inode_flag(inode, FI_COMPRESSED_FILE); 3999 f2fs_mark_inode_dirty_sync(inode, true); 4000 return true; 4001 } 4002 4003 #define F2FS_FEATURE_FUNCS(name, flagname) \ 4004 static inline int f2fs_sb_has_##name(struct f2fs_sb_info *sbi) \ 4005 { \ 4006 return F2FS_HAS_FEATURE(sbi, F2FS_FEATURE_##flagname); \ 4007 } 4008 4009 F2FS_FEATURE_FUNCS(encrypt, ENCRYPT); 4010 F2FS_FEATURE_FUNCS(blkzoned, BLKZONED); 4011 F2FS_FEATURE_FUNCS(extra_attr, EXTRA_ATTR); 4012 F2FS_FEATURE_FUNCS(project_quota, PRJQUOTA); 4013 F2FS_FEATURE_FUNCS(inode_chksum, INODE_CHKSUM); 4014 F2FS_FEATURE_FUNCS(flexible_inline_xattr, FLEXIBLE_INLINE_XATTR); 4015 F2FS_FEATURE_FUNCS(quota_ino, QUOTA_INO); 4016 F2FS_FEATURE_FUNCS(inode_crtime, INODE_CRTIME); 4017 F2FS_FEATURE_FUNCS(lost_found, LOST_FOUND); 4018 F2FS_FEATURE_FUNCS(verity, VERITY); 4019 F2FS_FEATURE_FUNCS(sb_chksum, SB_CHKSUM); 4020 F2FS_FEATURE_FUNCS(casefold, CASEFOLD); 4021 F2FS_FEATURE_FUNCS(compression, COMPRESSION); 4022 4023 #ifdef CONFIG_BLK_DEV_ZONED 4024 static inline bool f2fs_blkz_is_seq(struct f2fs_sb_info *sbi, int devi, 4025 block_t blkaddr) 4026 { 4027 unsigned int zno = blkaddr >> sbi->log_blocks_per_blkz; 4028 4029 return test_bit(zno, FDEV(devi).blkz_seq); 4030 } 4031 #endif 4032 4033 static inline bool f2fs_hw_should_discard(struct f2fs_sb_info *sbi) 4034 { 4035 return f2fs_sb_has_blkzoned(sbi); 4036 } 4037 4038 static inline bool f2fs_bdev_support_discard(struct block_device *bdev) 4039 { 4040 return blk_queue_discard(bdev_get_queue(bdev)) || 4041 bdev_is_zoned(bdev); 4042 } 4043 4044 static inline bool f2fs_hw_support_discard(struct f2fs_sb_info *sbi) 4045 { 4046 int i; 4047 4048 if (!f2fs_is_multi_device(sbi)) 4049 return f2fs_bdev_support_discard(sbi->sb->s_bdev); 4050 4051 for (i = 0; i < sbi->s_ndevs; i++) 4052 if (f2fs_bdev_support_discard(FDEV(i).bdev)) 4053 return true; 4054 return false; 4055 } 4056 4057 static inline bool f2fs_realtime_discard_enable(struct f2fs_sb_info *sbi) 4058 { 4059 return (test_opt(sbi, DISCARD) && f2fs_hw_support_discard(sbi)) || 4060 f2fs_hw_should_discard(sbi); 4061 } 4062 4063 static inline bool f2fs_hw_is_readonly(struct f2fs_sb_info *sbi) 4064 { 4065 int i; 4066 4067 if (!f2fs_is_multi_device(sbi)) 4068 return bdev_read_only(sbi->sb->s_bdev); 4069 4070 for (i = 0; i < sbi->s_ndevs; i++) 4071 if (bdev_read_only(FDEV(i).bdev)) 4072 return true; 4073 return false; 4074 } 4075 4076 static inline bool f2fs_lfs_mode(struct f2fs_sb_info *sbi) 4077 { 4078 return F2FS_OPTION(sbi).fs_mode == FS_MODE_LFS; 4079 } 4080 4081 static inline bool f2fs_may_compress(struct inode *inode) 4082 { 4083 if (IS_SWAPFILE(inode) || f2fs_is_pinned_file(inode) || 4084 f2fs_is_atomic_file(inode) || 4085 f2fs_is_volatile_file(inode)) 4086 return false; 4087 return S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode); 4088 } 4089 4090 static inline void f2fs_i_compr_blocks_update(struct inode *inode, 4091 u64 blocks, bool add) 4092 { 4093 int diff = F2FS_I(inode)->i_cluster_size - blocks; 4094 struct f2fs_inode_info *fi = F2FS_I(inode); 4095 4096 /* don't update i_compr_blocks if saved blocks were released */ 4097 if (!add && !atomic_read(&fi->i_compr_blocks)) 4098 return; 4099 4100 if (add) { 4101 atomic_add(diff, &fi->i_compr_blocks); 4102 stat_add_compr_blocks(inode, diff); 4103 } else { 4104 atomic_sub(diff, &fi->i_compr_blocks); 4105 stat_sub_compr_blocks(inode, diff); 4106 } 4107 f2fs_mark_inode_dirty_sync(inode, true); 4108 } 4109 4110 static inline int block_unaligned_IO(struct inode *inode, 4111 struct kiocb *iocb, struct iov_iter *iter) 4112 { 4113 unsigned int i_blkbits = READ_ONCE(inode->i_blkbits); 4114 unsigned int blocksize_mask = (1 << i_blkbits) - 1; 4115 loff_t offset = iocb->ki_pos; 4116 unsigned long align = offset | iov_iter_alignment(iter); 4117 4118 return align & blocksize_mask; 4119 } 4120 4121 static inline int allow_outplace_dio(struct inode *inode, 4122 struct kiocb *iocb, struct iov_iter *iter) 4123 { 4124 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 4125 int rw = iov_iter_rw(iter); 4126 4127 return (f2fs_lfs_mode(sbi) && (rw == WRITE) && 4128 !block_unaligned_IO(inode, iocb, iter)); 4129 } 4130 4131 static inline bool f2fs_force_buffered_io(struct inode *inode, 4132 struct kiocb *iocb, struct iov_iter *iter) 4133 { 4134 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 4135 int rw = iov_iter_rw(iter); 4136 4137 if (f2fs_post_read_required(inode)) 4138 return true; 4139 if (f2fs_is_multi_device(sbi)) 4140 return true; 4141 /* 4142 * for blkzoned device, fallback direct IO to buffered IO, so 4143 * all IOs can be serialized by log-structured write. 4144 */ 4145 if (f2fs_sb_has_blkzoned(sbi)) 4146 return true; 4147 if (f2fs_lfs_mode(sbi) && (rw == WRITE)) { 4148 if (block_unaligned_IO(inode, iocb, iter)) 4149 return true; 4150 if (F2FS_IO_ALIGNED(sbi)) 4151 return true; 4152 } 4153 if (is_sbi_flag_set(F2FS_I_SB(inode), SBI_CP_DISABLED) && 4154 !IS_SWAPFILE(inode)) 4155 return true; 4156 4157 return false; 4158 } 4159 4160 static inline bool f2fs_need_verity(const struct inode *inode, pgoff_t idx) 4161 { 4162 return fsverity_active(inode) && 4163 idx < DIV_ROUND_UP(inode->i_size, PAGE_SIZE); 4164 } 4165 4166 #ifdef CONFIG_F2FS_FAULT_INJECTION 4167 extern void f2fs_build_fault_attr(struct f2fs_sb_info *sbi, unsigned int rate, 4168 unsigned int type); 4169 #else 4170 #define f2fs_build_fault_attr(sbi, rate, type) do { } while (0) 4171 #endif 4172 4173 static inline bool is_journalled_quota(struct f2fs_sb_info *sbi) 4174 { 4175 #ifdef CONFIG_QUOTA 4176 if (f2fs_sb_has_quota_ino(sbi)) 4177 return true; 4178 if (F2FS_OPTION(sbi).s_qf_names[USRQUOTA] || 4179 F2FS_OPTION(sbi).s_qf_names[GRPQUOTA] || 4180 F2FS_OPTION(sbi).s_qf_names[PRJQUOTA]) 4181 return true; 4182 #endif 4183 return false; 4184 } 4185 4186 #define EFSBADCRC EBADMSG /* Bad CRC detected */ 4187 #define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */ 4188 4189 #endif /* _LINUX_F2FS_H */ 4190