1 /* SPDX-License-Identifier: GPL-2.0 */ 2 /* 3 * fs/f2fs/f2fs.h 4 * 5 * Copyright (c) 2012 Samsung Electronics Co., Ltd. 6 * http://www.samsung.com/ 7 */ 8 #ifndef _LINUX_F2FS_H 9 #define _LINUX_F2FS_H 10 11 #include <linux/uio.h> 12 #include <linux/types.h> 13 #include <linux/page-flags.h> 14 #include <linux/buffer_head.h> 15 #include <linux/slab.h> 16 #include <linux/crc32.h> 17 #include <linux/magic.h> 18 #include <linux/kobject.h> 19 #include <linux/sched.h> 20 #include <linux/cred.h> 21 #include <linux/vmalloc.h> 22 #include <linux/bio.h> 23 #include <linux/blkdev.h> 24 #include <linux/quotaops.h> 25 #include <linux/part_stat.h> 26 #include <crypto/hash.h> 27 28 #include <linux/fscrypt.h> 29 #include <linux/fsverity.h> 30 31 #ifdef CONFIG_F2FS_CHECK_FS 32 #define f2fs_bug_on(sbi, condition) BUG_ON(condition) 33 #else 34 #define f2fs_bug_on(sbi, condition) \ 35 do { \ 36 if (WARN_ON(condition)) \ 37 set_sbi_flag(sbi, SBI_NEED_FSCK); \ 38 } while (0) 39 #endif 40 41 enum { 42 FAULT_KMALLOC, 43 FAULT_KVMALLOC, 44 FAULT_PAGE_ALLOC, 45 FAULT_PAGE_GET, 46 FAULT_ALLOC_NID, 47 FAULT_ORPHAN, 48 FAULT_BLOCK, 49 FAULT_DIR_DEPTH, 50 FAULT_EVICT_INODE, 51 FAULT_TRUNCATE, 52 FAULT_READ_IO, 53 FAULT_CHECKPOINT, 54 FAULT_DISCARD, 55 FAULT_WRITE_IO, 56 FAULT_MAX, 57 }; 58 59 #ifdef CONFIG_F2FS_FAULT_INJECTION 60 #define F2FS_ALL_FAULT_TYPE ((1 << FAULT_MAX) - 1) 61 62 struct f2fs_fault_info { 63 atomic_t inject_ops; 64 unsigned int inject_rate; 65 unsigned int inject_type; 66 }; 67 68 extern const char *f2fs_fault_name[FAULT_MAX]; 69 #define IS_FAULT_SET(fi, type) ((fi)->inject_type & (1 << (type))) 70 #endif 71 72 /* 73 * For mount options 74 */ 75 #define F2FS_MOUNT_DISABLE_ROLL_FORWARD 0x00000002 76 #define F2FS_MOUNT_DISCARD 0x00000004 77 #define F2FS_MOUNT_NOHEAP 0x00000008 78 #define F2FS_MOUNT_XATTR_USER 0x00000010 79 #define F2FS_MOUNT_POSIX_ACL 0x00000020 80 #define F2FS_MOUNT_DISABLE_EXT_IDENTIFY 0x00000040 81 #define F2FS_MOUNT_INLINE_XATTR 0x00000080 82 #define F2FS_MOUNT_INLINE_DATA 0x00000100 83 #define F2FS_MOUNT_INLINE_DENTRY 0x00000200 84 #define F2FS_MOUNT_FLUSH_MERGE 0x00000400 85 #define F2FS_MOUNT_NOBARRIER 0x00000800 86 #define F2FS_MOUNT_FASTBOOT 0x00001000 87 #define F2FS_MOUNT_EXTENT_CACHE 0x00002000 88 #define F2FS_MOUNT_DATA_FLUSH 0x00008000 89 #define F2FS_MOUNT_FAULT_INJECTION 0x00010000 90 #define F2FS_MOUNT_USRQUOTA 0x00080000 91 #define F2FS_MOUNT_GRPQUOTA 0x00100000 92 #define F2FS_MOUNT_PRJQUOTA 0x00200000 93 #define F2FS_MOUNT_QUOTA 0x00400000 94 #define F2FS_MOUNT_INLINE_XATTR_SIZE 0x00800000 95 #define F2FS_MOUNT_RESERVE_ROOT 0x01000000 96 #define F2FS_MOUNT_DISABLE_CHECKPOINT 0x02000000 97 #define F2FS_MOUNT_NORECOVERY 0x04000000 98 #define F2FS_MOUNT_ATGC 0x08000000 99 100 #define F2FS_OPTION(sbi) ((sbi)->mount_opt) 101 #define clear_opt(sbi, option) (F2FS_OPTION(sbi).opt &= ~F2FS_MOUNT_##option) 102 #define set_opt(sbi, option) (F2FS_OPTION(sbi).opt |= F2FS_MOUNT_##option) 103 #define test_opt(sbi, option) (F2FS_OPTION(sbi).opt & F2FS_MOUNT_##option) 104 105 #define ver_after(a, b) (typecheck(unsigned long long, a) && \ 106 typecheck(unsigned long long, b) && \ 107 ((long long)((a) - (b)) > 0)) 108 109 typedef u32 block_t; /* 110 * should not change u32, since it is the on-disk block 111 * address format, __le32. 112 */ 113 typedef u32 nid_t; 114 115 #define COMPRESS_EXT_NUM 16 116 117 struct f2fs_mount_info { 118 unsigned int opt; 119 int write_io_size_bits; /* Write IO size bits */ 120 block_t root_reserved_blocks; /* root reserved blocks */ 121 kuid_t s_resuid; /* reserved blocks for uid */ 122 kgid_t s_resgid; /* reserved blocks for gid */ 123 int active_logs; /* # of active logs */ 124 int inline_xattr_size; /* inline xattr size */ 125 #ifdef CONFIG_F2FS_FAULT_INJECTION 126 struct f2fs_fault_info fault_info; /* For fault injection */ 127 #endif 128 #ifdef CONFIG_QUOTA 129 /* Names of quota files with journalled quota */ 130 char *s_qf_names[MAXQUOTAS]; 131 int s_jquota_fmt; /* Format of quota to use */ 132 #endif 133 /* For which write hints are passed down to block layer */ 134 int whint_mode; 135 int alloc_mode; /* segment allocation policy */ 136 int fsync_mode; /* fsync policy */ 137 int fs_mode; /* fs mode: LFS or ADAPTIVE */ 138 int bggc_mode; /* bggc mode: off, on or sync */ 139 struct fscrypt_dummy_policy dummy_enc_policy; /* test dummy encryption */ 140 block_t unusable_cap_perc; /* percentage for cap */ 141 block_t unusable_cap; /* Amount of space allowed to be 142 * unusable when disabling checkpoint 143 */ 144 145 /* For compression */ 146 unsigned char compress_algorithm; /* algorithm type */ 147 unsigned char compress_log_size; /* cluster log size */ 148 bool compress_chksum; /* compressed data chksum */ 149 unsigned char compress_ext_cnt; /* extension count */ 150 int compress_mode; /* compression mode */ 151 unsigned char extensions[COMPRESS_EXT_NUM][F2FS_EXTENSION_LEN]; /* extensions */ 152 }; 153 154 #define F2FS_FEATURE_ENCRYPT 0x0001 155 #define F2FS_FEATURE_BLKZONED 0x0002 156 #define F2FS_FEATURE_ATOMIC_WRITE 0x0004 157 #define F2FS_FEATURE_EXTRA_ATTR 0x0008 158 #define F2FS_FEATURE_PRJQUOTA 0x0010 159 #define F2FS_FEATURE_INODE_CHKSUM 0x0020 160 #define F2FS_FEATURE_FLEXIBLE_INLINE_XATTR 0x0040 161 #define F2FS_FEATURE_QUOTA_INO 0x0080 162 #define F2FS_FEATURE_INODE_CRTIME 0x0100 163 #define F2FS_FEATURE_LOST_FOUND 0x0200 164 #define F2FS_FEATURE_VERITY 0x0400 165 #define F2FS_FEATURE_SB_CHKSUM 0x0800 166 #define F2FS_FEATURE_CASEFOLD 0x1000 167 #define F2FS_FEATURE_COMPRESSION 0x2000 168 169 #define __F2FS_HAS_FEATURE(raw_super, mask) \ 170 ((raw_super->feature & cpu_to_le32(mask)) != 0) 171 #define F2FS_HAS_FEATURE(sbi, mask) __F2FS_HAS_FEATURE(sbi->raw_super, mask) 172 #define F2FS_SET_FEATURE(sbi, mask) \ 173 (sbi->raw_super->feature |= cpu_to_le32(mask)) 174 #define F2FS_CLEAR_FEATURE(sbi, mask) \ 175 (sbi->raw_super->feature &= ~cpu_to_le32(mask)) 176 177 /* 178 * Default values for user and/or group using reserved blocks 179 */ 180 #define F2FS_DEF_RESUID 0 181 #define F2FS_DEF_RESGID 0 182 183 /* 184 * For checkpoint manager 185 */ 186 enum { 187 NAT_BITMAP, 188 SIT_BITMAP 189 }; 190 191 #define CP_UMOUNT 0x00000001 192 #define CP_FASTBOOT 0x00000002 193 #define CP_SYNC 0x00000004 194 #define CP_RECOVERY 0x00000008 195 #define CP_DISCARD 0x00000010 196 #define CP_TRIMMED 0x00000020 197 #define CP_PAUSE 0x00000040 198 #define CP_RESIZE 0x00000080 199 200 #define MAX_DISCARD_BLOCKS(sbi) BLKS_PER_SEC(sbi) 201 #define DEF_MAX_DISCARD_REQUEST 8 /* issue 8 discards per round */ 202 #define DEF_MIN_DISCARD_ISSUE_TIME 50 /* 50 ms, if exists */ 203 #define DEF_MID_DISCARD_ISSUE_TIME 500 /* 500 ms, if device busy */ 204 #define DEF_MAX_DISCARD_ISSUE_TIME 60000 /* 60 s, if no candidates */ 205 #define DEF_DISCARD_URGENT_UTIL 80 /* do more discard over 80% */ 206 #define DEF_CP_INTERVAL 60 /* 60 secs */ 207 #define DEF_IDLE_INTERVAL 5 /* 5 secs */ 208 #define DEF_DISABLE_INTERVAL 5 /* 5 secs */ 209 #define DEF_DISABLE_QUICK_INTERVAL 1 /* 1 secs */ 210 #define DEF_UMOUNT_DISCARD_TIMEOUT 5 /* 5 secs */ 211 212 struct cp_control { 213 int reason; 214 __u64 trim_start; 215 __u64 trim_end; 216 __u64 trim_minlen; 217 }; 218 219 /* 220 * indicate meta/data type 221 */ 222 enum { 223 META_CP, 224 META_NAT, 225 META_SIT, 226 META_SSA, 227 META_MAX, 228 META_POR, 229 DATA_GENERIC, /* check range only */ 230 DATA_GENERIC_ENHANCE, /* strong check on range and segment bitmap */ 231 DATA_GENERIC_ENHANCE_READ, /* 232 * strong check on range and segment 233 * bitmap but no warning due to race 234 * condition of read on truncated area 235 * by extent_cache 236 */ 237 META_GENERIC, 238 }; 239 240 /* for the list of ino */ 241 enum { 242 ORPHAN_INO, /* for orphan ino list */ 243 APPEND_INO, /* for append ino list */ 244 UPDATE_INO, /* for update ino list */ 245 TRANS_DIR_INO, /* for trasactions dir ino list */ 246 FLUSH_INO, /* for multiple device flushing */ 247 MAX_INO_ENTRY, /* max. list */ 248 }; 249 250 struct ino_entry { 251 struct list_head list; /* list head */ 252 nid_t ino; /* inode number */ 253 unsigned int dirty_device; /* dirty device bitmap */ 254 }; 255 256 /* for the list of inodes to be GCed */ 257 struct inode_entry { 258 struct list_head list; /* list head */ 259 struct inode *inode; /* vfs inode pointer */ 260 }; 261 262 struct fsync_node_entry { 263 struct list_head list; /* list head */ 264 struct page *page; /* warm node page pointer */ 265 unsigned int seq_id; /* sequence id */ 266 }; 267 268 /* for the bitmap indicate blocks to be discarded */ 269 struct discard_entry { 270 struct list_head list; /* list head */ 271 block_t start_blkaddr; /* start blockaddr of current segment */ 272 unsigned char discard_map[SIT_VBLOCK_MAP_SIZE]; /* segment discard bitmap */ 273 }; 274 275 /* default discard granularity of inner discard thread, unit: block count */ 276 #define DEFAULT_DISCARD_GRANULARITY 16 277 278 /* max discard pend list number */ 279 #define MAX_PLIST_NUM 512 280 #define plist_idx(blk_num) ((blk_num) >= MAX_PLIST_NUM ? \ 281 (MAX_PLIST_NUM - 1) : ((blk_num) - 1)) 282 283 enum { 284 D_PREP, /* initial */ 285 D_PARTIAL, /* partially submitted */ 286 D_SUBMIT, /* all submitted */ 287 D_DONE, /* finished */ 288 }; 289 290 struct discard_info { 291 block_t lstart; /* logical start address */ 292 block_t len; /* length */ 293 block_t start; /* actual start address in dev */ 294 }; 295 296 struct discard_cmd { 297 struct rb_node rb_node; /* rb node located in rb-tree */ 298 union { 299 struct { 300 block_t lstart; /* logical start address */ 301 block_t len; /* length */ 302 block_t start; /* actual start address in dev */ 303 }; 304 struct discard_info di; /* discard info */ 305 306 }; 307 struct list_head list; /* command list */ 308 struct completion wait; /* compleation */ 309 struct block_device *bdev; /* bdev */ 310 unsigned short ref; /* reference count */ 311 unsigned char state; /* state */ 312 unsigned char queued; /* queued discard */ 313 int error; /* bio error */ 314 spinlock_t lock; /* for state/bio_ref updating */ 315 unsigned short bio_ref; /* bio reference count */ 316 }; 317 318 enum { 319 DPOLICY_BG, 320 DPOLICY_FORCE, 321 DPOLICY_FSTRIM, 322 DPOLICY_UMOUNT, 323 MAX_DPOLICY, 324 }; 325 326 struct discard_policy { 327 int type; /* type of discard */ 328 unsigned int min_interval; /* used for candidates exist */ 329 unsigned int mid_interval; /* used for device busy */ 330 unsigned int max_interval; /* used for candidates not exist */ 331 unsigned int max_requests; /* # of discards issued per round */ 332 unsigned int io_aware_gran; /* minimum granularity discard not be aware of I/O */ 333 bool io_aware; /* issue discard in idle time */ 334 bool sync; /* submit discard with REQ_SYNC flag */ 335 bool ordered; /* issue discard by lba order */ 336 bool timeout; /* discard timeout for put_super */ 337 unsigned int granularity; /* discard granularity */ 338 }; 339 340 struct discard_cmd_control { 341 struct task_struct *f2fs_issue_discard; /* discard thread */ 342 struct list_head entry_list; /* 4KB discard entry list */ 343 struct list_head pend_list[MAX_PLIST_NUM];/* store pending entries */ 344 struct list_head wait_list; /* store on-flushing entries */ 345 struct list_head fstrim_list; /* in-flight discard from fstrim */ 346 wait_queue_head_t discard_wait_queue; /* waiting queue for wake-up */ 347 unsigned int discard_wake; /* to wake up discard thread */ 348 struct mutex cmd_lock; 349 unsigned int nr_discards; /* # of discards in the list */ 350 unsigned int max_discards; /* max. discards to be issued */ 351 unsigned int discard_granularity; /* discard granularity */ 352 unsigned int undiscard_blks; /* # of undiscard blocks */ 353 unsigned int next_pos; /* next discard position */ 354 atomic_t issued_discard; /* # of issued discard */ 355 atomic_t queued_discard; /* # of queued discard */ 356 atomic_t discard_cmd_cnt; /* # of cached cmd count */ 357 struct rb_root_cached root; /* root of discard rb-tree */ 358 bool rbtree_check; /* config for consistence check */ 359 }; 360 361 /* for the list of fsync inodes, used only during recovery */ 362 struct fsync_inode_entry { 363 struct list_head list; /* list head */ 364 struct inode *inode; /* vfs inode pointer */ 365 block_t blkaddr; /* block address locating the last fsync */ 366 block_t last_dentry; /* block address locating the last dentry */ 367 }; 368 369 #define nats_in_cursum(jnl) (le16_to_cpu((jnl)->n_nats)) 370 #define sits_in_cursum(jnl) (le16_to_cpu((jnl)->n_sits)) 371 372 #define nat_in_journal(jnl, i) ((jnl)->nat_j.entries[i].ne) 373 #define nid_in_journal(jnl, i) ((jnl)->nat_j.entries[i].nid) 374 #define sit_in_journal(jnl, i) ((jnl)->sit_j.entries[i].se) 375 #define segno_in_journal(jnl, i) ((jnl)->sit_j.entries[i].segno) 376 377 #define MAX_NAT_JENTRIES(jnl) (NAT_JOURNAL_ENTRIES - nats_in_cursum(jnl)) 378 #define MAX_SIT_JENTRIES(jnl) (SIT_JOURNAL_ENTRIES - sits_in_cursum(jnl)) 379 380 static inline int update_nats_in_cursum(struct f2fs_journal *journal, int i) 381 { 382 int before = nats_in_cursum(journal); 383 384 journal->n_nats = cpu_to_le16(before + i); 385 return before; 386 } 387 388 static inline int update_sits_in_cursum(struct f2fs_journal *journal, int i) 389 { 390 int before = sits_in_cursum(journal); 391 392 journal->n_sits = cpu_to_le16(before + i); 393 return before; 394 } 395 396 static inline bool __has_cursum_space(struct f2fs_journal *journal, 397 int size, int type) 398 { 399 if (type == NAT_JOURNAL) 400 return size <= MAX_NAT_JENTRIES(journal); 401 return size <= MAX_SIT_JENTRIES(journal); 402 } 403 404 /* for inline stuff */ 405 #define DEF_INLINE_RESERVED_SIZE 1 406 static inline int get_extra_isize(struct inode *inode); 407 static inline int get_inline_xattr_addrs(struct inode *inode); 408 #define MAX_INLINE_DATA(inode) (sizeof(__le32) * \ 409 (CUR_ADDRS_PER_INODE(inode) - \ 410 get_inline_xattr_addrs(inode) - \ 411 DEF_INLINE_RESERVED_SIZE)) 412 413 /* for inline dir */ 414 #define NR_INLINE_DENTRY(inode) (MAX_INLINE_DATA(inode) * BITS_PER_BYTE / \ 415 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \ 416 BITS_PER_BYTE + 1)) 417 #define INLINE_DENTRY_BITMAP_SIZE(inode) \ 418 DIV_ROUND_UP(NR_INLINE_DENTRY(inode), BITS_PER_BYTE) 419 #define INLINE_RESERVED_SIZE(inode) (MAX_INLINE_DATA(inode) - \ 420 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \ 421 NR_INLINE_DENTRY(inode) + \ 422 INLINE_DENTRY_BITMAP_SIZE(inode))) 423 424 /* 425 * For INODE and NODE manager 426 */ 427 /* for directory operations */ 428 429 struct f2fs_filename { 430 /* 431 * The filename the user specified. This is NULL for some 432 * filesystem-internal operations, e.g. converting an inline directory 433 * to a non-inline one, or roll-forward recovering an encrypted dentry. 434 */ 435 const struct qstr *usr_fname; 436 437 /* 438 * The on-disk filename. For encrypted directories, this is encrypted. 439 * This may be NULL for lookups in an encrypted dir without the key. 440 */ 441 struct fscrypt_str disk_name; 442 443 /* The dirhash of this filename */ 444 f2fs_hash_t hash; 445 446 #ifdef CONFIG_FS_ENCRYPTION 447 /* 448 * For lookups in encrypted directories: either the buffer backing 449 * disk_name, or a buffer that holds the decoded no-key name. 450 */ 451 struct fscrypt_str crypto_buf; 452 #endif 453 #ifdef CONFIG_UNICODE 454 /* 455 * For casefolded directories: the casefolded name, but it's left NULL 456 * if the original name is not valid Unicode, if the directory is both 457 * casefolded and encrypted and its encryption key is unavailable, or if 458 * the filesystem is doing an internal operation where usr_fname is also 459 * NULL. In all these cases we fall back to treating the name as an 460 * opaque byte sequence. 461 */ 462 struct fscrypt_str cf_name; 463 #endif 464 }; 465 466 struct f2fs_dentry_ptr { 467 struct inode *inode; 468 void *bitmap; 469 struct f2fs_dir_entry *dentry; 470 __u8 (*filename)[F2FS_SLOT_LEN]; 471 int max; 472 int nr_bitmap; 473 }; 474 475 static inline void make_dentry_ptr_block(struct inode *inode, 476 struct f2fs_dentry_ptr *d, struct f2fs_dentry_block *t) 477 { 478 d->inode = inode; 479 d->max = NR_DENTRY_IN_BLOCK; 480 d->nr_bitmap = SIZE_OF_DENTRY_BITMAP; 481 d->bitmap = t->dentry_bitmap; 482 d->dentry = t->dentry; 483 d->filename = t->filename; 484 } 485 486 static inline void make_dentry_ptr_inline(struct inode *inode, 487 struct f2fs_dentry_ptr *d, void *t) 488 { 489 int entry_cnt = NR_INLINE_DENTRY(inode); 490 int bitmap_size = INLINE_DENTRY_BITMAP_SIZE(inode); 491 int reserved_size = INLINE_RESERVED_SIZE(inode); 492 493 d->inode = inode; 494 d->max = entry_cnt; 495 d->nr_bitmap = bitmap_size; 496 d->bitmap = t; 497 d->dentry = t + bitmap_size + reserved_size; 498 d->filename = t + bitmap_size + reserved_size + 499 SIZE_OF_DIR_ENTRY * entry_cnt; 500 } 501 502 /* 503 * XATTR_NODE_OFFSET stores xattrs to one node block per file keeping -1 504 * as its node offset to distinguish from index node blocks. 505 * But some bits are used to mark the node block. 506 */ 507 #define XATTR_NODE_OFFSET ((((unsigned int)-1) << OFFSET_BIT_SHIFT) \ 508 >> OFFSET_BIT_SHIFT) 509 enum { 510 ALLOC_NODE, /* allocate a new node page if needed */ 511 LOOKUP_NODE, /* look up a node without readahead */ 512 LOOKUP_NODE_RA, /* 513 * look up a node with readahead called 514 * by get_data_block. 515 */ 516 }; 517 518 #define DEFAULT_RETRY_IO_COUNT 8 /* maximum retry read IO count */ 519 520 /* congestion wait timeout value, default: 20ms */ 521 #define DEFAULT_IO_TIMEOUT (msecs_to_jiffies(20)) 522 523 /* maximum retry quota flush count */ 524 #define DEFAULT_RETRY_QUOTA_FLUSH_COUNT 8 525 526 #define F2FS_LINK_MAX 0xffffffff /* maximum link count per file */ 527 528 #define MAX_DIR_RA_PAGES 4 /* maximum ra pages of dir */ 529 530 /* for in-memory extent cache entry */ 531 #define F2FS_MIN_EXTENT_LEN 64 /* minimum extent length */ 532 533 /* number of extent info in extent cache we try to shrink */ 534 #define EXTENT_CACHE_SHRINK_NUMBER 128 535 536 struct rb_entry { 537 struct rb_node rb_node; /* rb node located in rb-tree */ 538 union { 539 struct { 540 unsigned int ofs; /* start offset of the entry */ 541 unsigned int len; /* length of the entry */ 542 }; 543 unsigned long long key; /* 64-bits key */ 544 } __packed; 545 }; 546 547 struct extent_info { 548 unsigned int fofs; /* start offset in a file */ 549 unsigned int len; /* length of the extent */ 550 u32 blk; /* start block address of the extent */ 551 }; 552 553 struct extent_node { 554 struct rb_node rb_node; /* rb node located in rb-tree */ 555 struct extent_info ei; /* extent info */ 556 struct list_head list; /* node in global extent list of sbi */ 557 struct extent_tree *et; /* extent tree pointer */ 558 }; 559 560 struct extent_tree { 561 nid_t ino; /* inode number */ 562 struct rb_root_cached root; /* root of extent info rb-tree */ 563 struct extent_node *cached_en; /* recently accessed extent node */ 564 struct extent_info largest; /* largested extent info */ 565 struct list_head list; /* to be used by sbi->zombie_list */ 566 rwlock_t lock; /* protect extent info rb-tree */ 567 atomic_t node_cnt; /* # of extent node in rb-tree*/ 568 bool largest_updated; /* largest extent updated */ 569 }; 570 571 /* 572 * This structure is taken from ext4_map_blocks. 573 * 574 * Note that, however, f2fs uses NEW and MAPPED flags for f2fs_map_blocks(). 575 */ 576 #define F2FS_MAP_NEW (1 << BH_New) 577 #define F2FS_MAP_MAPPED (1 << BH_Mapped) 578 #define F2FS_MAP_UNWRITTEN (1 << BH_Unwritten) 579 #define F2FS_MAP_FLAGS (F2FS_MAP_NEW | F2FS_MAP_MAPPED |\ 580 F2FS_MAP_UNWRITTEN) 581 582 struct f2fs_map_blocks { 583 block_t m_pblk; 584 block_t m_lblk; 585 unsigned int m_len; 586 unsigned int m_flags; 587 pgoff_t *m_next_pgofs; /* point next possible non-hole pgofs */ 588 pgoff_t *m_next_extent; /* point to next possible extent */ 589 int m_seg_type; 590 bool m_may_create; /* indicate it is from write path */ 591 }; 592 593 /* for flag in get_data_block */ 594 enum { 595 F2FS_GET_BLOCK_DEFAULT, 596 F2FS_GET_BLOCK_FIEMAP, 597 F2FS_GET_BLOCK_BMAP, 598 F2FS_GET_BLOCK_DIO, 599 F2FS_GET_BLOCK_PRE_DIO, 600 F2FS_GET_BLOCK_PRE_AIO, 601 F2FS_GET_BLOCK_PRECACHE, 602 }; 603 604 /* 605 * i_advise uses FADVISE_XXX_BIT. We can add additional hints later. 606 */ 607 #define FADVISE_COLD_BIT 0x01 608 #define FADVISE_LOST_PINO_BIT 0x02 609 #define FADVISE_ENCRYPT_BIT 0x04 610 #define FADVISE_ENC_NAME_BIT 0x08 611 #define FADVISE_KEEP_SIZE_BIT 0x10 612 #define FADVISE_HOT_BIT 0x20 613 #define FADVISE_VERITY_BIT 0x40 614 615 #define FADVISE_MODIFIABLE_BITS (FADVISE_COLD_BIT | FADVISE_HOT_BIT) 616 617 #define file_is_cold(inode) is_file(inode, FADVISE_COLD_BIT) 618 #define file_wrong_pino(inode) is_file(inode, FADVISE_LOST_PINO_BIT) 619 #define file_set_cold(inode) set_file(inode, FADVISE_COLD_BIT) 620 #define file_lost_pino(inode) set_file(inode, FADVISE_LOST_PINO_BIT) 621 #define file_clear_cold(inode) clear_file(inode, FADVISE_COLD_BIT) 622 #define file_got_pino(inode) clear_file(inode, FADVISE_LOST_PINO_BIT) 623 #define file_is_encrypt(inode) is_file(inode, FADVISE_ENCRYPT_BIT) 624 #define file_set_encrypt(inode) set_file(inode, FADVISE_ENCRYPT_BIT) 625 #define file_clear_encrypt(inode) clear_file(inode, FADVISE_ENCRYPT_BIT) 626 #define file_enc_name(inode) is_file(inode, FADVISE_ENC_NAME_BIT) 627 #define file_set_enc_name(inode) set_file(inode, FADVISE_ENC_NAME_BIT) 628 #define file_keep_isize(inode) is_file(inode, FADVISE_KEEP_SIZE_BIT) 629 #define file_set_keep_isize(inode) set_file(inode, FADVISE_KEEP_SIZE_BIT) 630 #define file_is_hot(inode) is_file(inode, FADVISE_HOT_BIT) 631 #define file_set_hot(inode) set_file(inode, FADVISE_HOT_BIT) 632 #define file_clear_hot(inode) clear_file(inode, FADVISE_HOT_BIT) 633 #define file_is_verity(inode) is_file(inode, FADVISE_VERITY_BIT) 634 #define file_set_verity(inode) set_file(inode, FADVISE_VERITY_BIT) 635 636 #define DEF_DIR_LEVEL 0 637 638 enum { 639 GC_FAILURE_PIN, 640 GC_FAILURE_ATOMIC, 641 MAX_GC_FAILURE 642 }; 643 644 /* used for f2fs_inode_info->flags */ 645 enum { 646 FI_NEW_INODE, /* indicate newly allocated inode */ 647 FI_DIRTY_INODE, /* indicate inode is dirty or not */ 648 FI_AUTO_RECOVER, /* indicate inode is recoverable */ 649 FI_DIRTY_DIR, /* indicate directory has dirty pages */ 650 FI_INC_LINK, /* need to increment i_nlink */ 651 FI_ACL_MODE, /* indicate acl mode */ 652 FI_NO_ALLOC, /* should not allocate any blocks */ 653 FI_FREE_NID, /* free allocated nide */ 654 FI_NO_EXTENT, /* not to use the extent cache */ 655 FI_INLINE_XATTR, /* used for inline xattr */ 656 FI_INLINE_DATA, /* used for inline data*/ 657 FI_INLINE_DENTRY, /* used for inline dentry */ 658 FI_APPEND_WRITE, /* inode has appended data */ 659 FI_UPDATE_WRITE, /* inode has in-place-update data */ 660 FI_NEED_IPU, /* used for ipu per file */ 661 FI_ATOMIC_FILE, /* indicate atomic file */ 662 FI_ATOMIC_COMMIT, /* indicate the state of atomical committing */ 663 FI_VOLATILE_FILE, /* indicate volatile file */ 664 FI_FIRST_BLOCK_WRITTEN, /* indicate #0 data block was written */ 665 FI_DROP_CACHE, /* drop dirty page cache */ 666 FI_DATA_EXIST, /* indicate data exists */ 667 FI_INLINE_DOTS, /* indicate inline dot dentries */ 668 FI_DO_DEFRAG, /* indicate defragment is running */ 669 FI_DIRTY_FILE, /* indicate regular/symlink has dirty pages */ 670 FI_NO_PREALLOC, /* indicate skipped preallocated blocks */ 671 FI_HOT_DATA, /* indicate file is hot */ 672 FI_EXTRA_ATTR, /* indicate file has extra attribute */ 673 FI_PROJ_INHERIT, /* indicate file inherits projectid */ 674 FI_PIN_FILE, /* indicate file should not be gced */ 675 FI_ATOMIC_REVOKE_REQUEST, /* request to drop atomic data */ 676 FI_VERITY_IN_PROGRESS, /* building fs-verity Merkle tree */ 677 FI_COMPRESSED_FILE, /* indicate file's data can be compressed */ 678 FI_COMPRESS_CORRUPT, /* indicate compressed cluster is corrupted */ 679 FI_MMAP_FILE, /* indicate file was mmapped */ 680 FI_ENABLE_COMPRESS, /* enable compression in "user" compression mode */ 681 FI_MAX, /* max flag, never be used */ 682 }; 683 684 struct f2fs_inode_info { 685 struct inode vfs_inode; /* serve a vfs inode */ 686 unsigned long i_flags; /* keep an inode flags for ioctl */ 687 unsigned char i_advise; /* use to give file attribute hints */ 688 unsigned char i_dir_level; /* use for dentry level for large dir */ 689 unsigned int i_current_depth; /* only for directory depth */ 690 /* for gc failure statistic */ 691 unsigned int i_gc_failures[MAX_GC_FAILURE]; 692 unsigned int i_pino; /* parent inode number */ 693 umode_t i_acl_mode; /* keep file acl mode temporarily */ 694 695 /* Use below internally in f2fs*/ 696 unsigned long flags[BITS_TO_LONGS(FI_MAX)]; /* use to pass per-file flags */ 697 struct rw_semaphore i_sem; /* protect fi info */ 698 atomic_t dirty_pages; /* # of dirty pages */ 699 f2fs_hash_t chash; /* hash value of given file name */ 700 unsigned int clevel; /* maximum level of given file name */ 701 struct task_struct *task; /* lookup and create consistency */ 702 struct task_struct *cp_task; /* separate cp/wb IO stats*/ 703 nid_t i_xattr_nid; /* node id that contains xattrs */ 704 loff_t last_disk_size; /* lastly written file size */ 705 spinlock_t i_size_lock; /* protect last_disk_size */ 706 707 #ifdef CONFIG_QUOTA 708 struct dquot *i_dquot[MAXQUOTAS]; 709 710 /* quota space reservation, managed internally by quota code */ 711 qsize_t i_reserved_quota; 712 #endif 713 struct list_head dirty_list; /* dirty list for dirs and files */ 714 struct list_head gdirty_list; /* linked in global dirty list */ 715 struct list_head inmem_ilist; /* list for inmem inodes */ 716 struct list_head inmem_pages; /* inmemory pages managed by f2fs */ 717 struct task_struct *inmem_task; /* store inmemory task */ 718 struct mutex inmem_lock; /* lock for inmemory pages */ 719 pgoff_t ra_offset; /* ongoing readahead offset */ 720 struct extent_tree *extent_tree; /* cached extent_tree entry */ 721 722 /* avoid racing between foreground op and gc */ 723 struct rw_semaphore i_gc_rwsem[2]; 724 struct rw_semaphore i_mmap_sem; 725 struct rw_semaphore i_xattr_sem; /* avoid racing between reading and changing EAs */ 726 727 int i_extra_isize; /* size of extra space located in i_addr */ 728 kprojid_t i_projid; /* id for project quota */ 729 int i_inline_xattr_size; /* inline xattr size */ 730 struct timespec64 i_crtime; /* inode creation time */ 731 struct timespec64 i_disk_time[4];/* inode disk times */ 732 733 /* for file compress */ 734 atomic_t i_compr_blocks; /* # of compressed blocks */ 735 unsigned char i_compress_algorithm; /* algorithm type */ 736 unsigned char i_log_cluster_size; /* log of cluster size */ 737 unsigned short i_compress_flag; /* compress flag */ 738 unsigned int i_cluster_size; /* cluster size */ 739 }; 740 741 static inline void get_extent_info(struct extent_info *ext, 742 struct f2fs_extent *i_ext) 743 { 744 ext->fofs = le32_to_cpu(i_ext->fofs); 745 ext->blk = le32_to_cpu(i_ext->blk); 746 ext->len = le32_to_cpu(i_ext->len); 747 } 748 749 static inline void set_raw_extent(struct extent_info *ext, 750 struct f2fs_extent *i_ext) 751 { 752 i_ext->fofs = cpu_to_le32(ext->fofs); 753 i_ext->blk = cpu_to_le32(ext->blk); 754 i_ext->len = cpu_to_le32(ext->len); 755 } 756 757 static inline void set_extent_info(struct extent_info *ei, unsigned int fofs, 758 u32 blk, unsigned int len) 759 { 760 ei->fofs = fofs; 761 ei->blk = blk; 762 ei->len = len; 763 } 764 765 static inline bool __is_discard_mergeable(struct discard_info *back, 766 struct discard_info *front, unsigned int max_len) 767 { 768 return (back->lstart + back->len == front->lstart) && 769 (back->len + front->len <= max_len); 770 } 771 772 static inline bool __is_discard_back_mergeable(struct discard_info *cur, 773 struct discard_info *back, unsigned int max_len) 774 { 775 return __is_discard_mergeable(back, cur, max_len); 776 } 777 778 static inline bool __is_discard_front_mergeable(struct discard_info *cur, 779 struct discard_info *front, unsigned int max_len) 780 { 781 return __is_discard_mergeable(cur, front, max_len); 782 } 783 784 static inline bool __is_extent_mergeable(struct extent_info *back, 785 struct extent_info *front) 786 { 787 return (back->fofs + back->len == front->fofs && 788 back->blk + back->len == front->blk); 789 } 790 791 static inline bool __is_back_mergeable(struct extent_info *cur, 792 struct extent_info *back) 793 { 794 return __is_extent_mergeable(back, cur); 795 } 796 797 static inline bool __is_front_mergeable(struct extent_info *cur, 798 struct extent_info *front) 799 { 800 return __is_extent_mergeable(cur, front); 801 } 802 803 extern void f2fs_mark_inode_dirty_sync(struct inode *inode, bool sync); 804 static inline void __try_update_largest_extent(struct extent_tree *et, 805 struct extent_node *en) 806 { 807 if (en->ei.len > et->largest.len) { 808 et->largest = en->ei; 809 et->largest_updated = true; 810 } 811 } 812 813 /* 814 * For free nid management 815 */ 816 enum nid_state { 817 FREE_NID, /* newly added to free nid list */ 818 PREALLOC_NID, /* it is preallocated */ 819 MAX_NID_STATE, 820 }; 821 822 enum nat_state { 823 TOTAL_NAT, 824 DIRTY_NAT, 825 RECLAIMABLE_NAT, 826 MAX_NAT_STATE, 827 }; 828 829 struct f2fs_nm_info { 830 block_t nat_blkaddr; /* base disk address of NAT */ 831 nid_t max_nid; /* maximum possible node ids */ 832 nid_t available_nids; /* # of available node ids */ 833 nid_t next_scan_nid; /* the next nid to be scanned */ 834 unsigned int ram_thresh; /* control the memory footprint */ 835 unsigned int ra_nid_pages; /* # of nid pages to be readaheaded */ 836 unsigned int dirty_nats_ratio; /* control dirty nats ratio threshold */ 837 838 /* NAT cache management */ 839 struct radix_tree_root nat_root;/* root of the nat entry cache */ 840 struct radix_tree_root nat_set_root;/* root of the nat set cache */ 841 struct rw_semaphore nat_tree_lock; /* protect nat_tree_lock */ 842 struct list_head nat_entries; /* cached nat entry list (clean) */ 843 spinlock_t nat_list_lock; /* protect clean nat entry list */ 844 unsigned int nat_cnt[MAX_NAT_STATE]; /* the # of cached nat entries */ 845 unsigned int nat_blocks; /* # of nat blocks */ 846 847 /* free node ids management */ 848 struct radix_tree_root free_nid_root;/* root of the free_nid cache */ 849 struct list_head free_nid_list; /* list for free nids excluding preallocated nids */ 850 unsigned int nid_cnt[MAX_NID_STATE]; /* the number of free node id */ 851 spinlock_t nid_list_lock; /* protect nid lists ops */ 852 struct mutex build_lock; /* lock for build free nids */ 853 unsigned char **free_nid_bitmap; 854 unsigned char *nat_block_bitmap; 855 unsigned short *free_nid_count; /* free nid count of NAT block */ 856 857 /* for checkpoint */ 858 char *nat_bitmap; /* NAT bitmap pointer */ 859 860 unsigned int nat_bits_blocks; /* # of nat bits blocks */ 861 unsigned char *nat_bits; /* NAT bits blocks */ 862 unsigned char *full_nat_bits; /* full NAT pages */ 863 unsigned char *empty_nat_bits; /* empty NAT pages */ 864 #ifdef CONFIG_F2FS_CHECK_FS 865 char *nat_bitmap_mir; /* NAT bitmap mirror */ 866 #endif 867 int bitmap_size; /* bitmap size */ 868 }; 869 870 /* 871 * this structure is used as one of function parameters. 872 * all the information are dedicated to a given direct node block determined 873 * by the data offset in a file. 874 */ 875 struct dnode_of_data { 876 struct inode *inode; /* vfs inode pointer */ 877 struct page *inode_page; /* its inode page, NULL is possible */ 878 struct page *node_page; /* cached direct node page */ 879 nid_t nid; /* node id of the direct node block */ 880 unsigned int ofs_in_node; /* data offset in the node page */ 881 bool inode_page_locked; /* inode page is locked or not */ 882 bool node_changed; /* is node block changed */ 883 char cur_level; /* level of hole node page */ 884 char max_level; /* level of current page located */ 885 block_t data_blkaddr; /* block address of the node block */ 886 }; 887 888 static inline void set_new_dnode(struct dnode_of_data *dn, struct inode *inode, 889 struct page *ipage, struct page *npage, nid_t nid) 890 { 891 memset(dn, 0, sizeof(*dn)); 892 dn->inode = inode; 893 dn->inode_page = ipage; 894 dn->node_page = npage; 895 dn->nid = nid; 896 } 897 898 /* 899 * For SIT manager 900 * 901 * By default, there are 6 active log areas across the whole main area. 902 * When considering hot and cold data separation to reduce cleaning overhead, 903 * we split 3 for data logs and 3 for node logs as hot, warm, and cold types, 904 * respectively. 905 * In the current design, you should not change the numbers intentionally. 906 * Instead, as a mount option such as active_logs=x, you can use 2, 4, and 6 907 * logs individually according to the underlying devices. (default: 6) 908 * Just in case, on-disk layout covers maximum 16 logs that consist of 8 for 909 * data and 8 for node logs. 910 */ 911 #define NR_CURSEG_DATA_TYPE (3) 912 #define NR_CURSEG_NODE_TYPE (3) 913 #define NR_CURSEG_INMEM_TYPE (2) 914 #define NR_CURSEG_PERSIST_TYPE (NR_CURSEG_DATA_TYPE + NR_CURSEG_NODE_TYPE) 915 #define NR_CURSEG_TYPE (NR_CURSEG_INMEM_TYPE + NR_CURSEG_PERSIST_TYPE) 916 917 enum { 918 CURSEG_HOT_DATA = 0, /* directory entry blocks */ 919 CURSEG_WARM_DATA, /* data blocks */ 920 CURSEG_COLD_DATA, /* multimedia or GCed data blocks */ 921 CURSEG_HOT_NODE, /* direct node blocks of directory files */ 922 CURSEG_WARM_NODE, /* direct node blocks of normal files */ 923 CURSEG_COLD_NODE, /* indirect node blocks */ 924 NR_PERSISTENT_LOG, /* number of persistent log */ 925 CURSEG_COLD_DATA_PINNED = NR_PERSISTENT_LOG, 926 /* pinned file that needs consecutive block address */ 927 CURSEG_ALL_DATA_ATGC, /* SSR alloctor in hot/warm/cold data area */ 928 NO_CHECK_TYPE, /* number of persistent & inmem log */ 929 }; 930 931 struct flush_cmd { 932 struct completion wait; 933 struct llist_node llnode; 934 nid_t ino; 935 int ret; 936 }; 937 938 struct flush_cmd_control { 939 struct task_struct *f2fs_issue_flush; /* flush thread */ 940 wait_queue_head_t flush_wait_queue; /* waiting queue for wake-up */ 941 atomic_t issued_flush; /* # of issued flushes */ 942 atomic_t queued_flush; /* # of queued flushes */ 943 struct llist_head issue_list; /* list for command issue */ 944 struct llist_node *dispatch_list; /* list for command dispatch */ 945 }; 946 947 struct f2fs_sm_info { 948 struct sit_info *sit_info; /* whole segment information */ 949 struct free_segmap_info *free_info; /* free segment information */ 950 struct dirty_seglist_info *dirty_info; /* dirty segment information */ 951 struct curseg_info *curseg_array; /* active segment information */ 952 953 struct rw_semaphore curseg_lock; /* for preventing curseg change */ 954 955 block_t seg0_blkaddr; /* block address of 0'th segment */ 956 block_t main_blkaddr; /* start block address of main area */ 957 block_t ssa_blkaddr; /* start block address of SSA area */ 958 959 unsigned int segment_count; /* total # of segments */ 960 unsigned int main_segments; /* # of segments in main area */ 961 unsigned int reserved_segments; /* # of reserved segments */ 962 unsigned int ovp_segments; /* # of overprovision segments */ 963 964 /* a threshold to reclaim prefree segments */ 965 unsigned int rec_prefree_segments; 966 967 /* for batched trimming */ 968 unsigned int trim_sections; /* # of sections to trim */ 969 970 struct list_head sit_entry_set; /* sit entry set list */ 971 972 unsigned int ipu_policy; /* in-place-update policy */ 973 unsigned int min_ipu_util; /* in-place-update threshold */ 974 unsigned int min_fsync_blocks; /* threshold for fsync */ 975 unsigned int min_seq_blocks; /* threshold for sequential blocks */ 976 unsigned int min_hot_blocks; /* threshold for hot block allocation */ 977 unsigned int min_ssr_sections; /* threshold to trigger SSR allocation */ 978 979 /* for flush command control */ 980 struct flush_cmd_control *fcc_info; 981 982 /* for discard command control */ 983 struct discard_cmd_control *dcc_info; 984 }; 985 986 /* 987 * For superblock 988 */ 989 /* 990 * COUNT_TYPE for monitoring 991 * 992 * f2fs monitors the number of several block types such as on-writeback, 993 * dirty dentry blocks, dirty node blocks, and dirty meta blocks. 994 */ 995 #define WB_DATA_TYPE(p) (__is_cp_guaranteed(p) ? F2FS_WB_CP_DATA : F2FS_WB_DATA) 996 enum count_type { 997 F2FS_DIRTY_DENTS, 998 F2FS_DIRTY_DATA, 999 F2FS_DIRTY_QDATA, 1000 F2FS_DIRTY_NODES, 1001 F2FS_DIRTY_META, 1002 F2FS_INMEM_PAGES, 1003 F2FS_DIRTY_IMETA, 1004 F2FS_WB_CP_DATA, 1005 F2FS_WB_DATA, 1006 F2FS_RD_DATA, 1007 F2FS_RD_NODE, 1008 F2FS_RD_META, 1009 F2FS_DIO_WRITE, 1010 F2FS_DIO_READ, 1011 NR_COUNT_TYPE, 1012 }; 1013 1014 /* 1015 * The below are the page types of bios used in submit_bio(). 1016 * The available types are: 1017 * DATA User data pages. It operates as async mode. 1018 * NODE Node pages. It operates as async mode. 1019 * META FS metadata pages such as SIT, NAT, CP. 1020 * NR_PAGE_TYPE The number of page types. 1021 * META_FLUSH Make sure the previous pages are written 1022 * with waiting the bio's completion 1023 * ... Only can be used with META. 1024 */ 1025 #define PAGE_TYPE_OF_BIO(type) ((type) > META ? META : (type)) 1026 enum page_type { 1027 DATA, 1028 NODE, 1029 META, 1030 NR_PAGE_TYPE, 1031 META_FLUSH, 1032 INMEM, /* the below types are used by tracepoints only. */ 1033 INMEM_DROP, 1034 INMEM_INVALIDATE, 1035 INMEM_REVOKE, 1036 IPU, 1037 OPU, 1038 }; 1039 1040 enum temp_type { 1041 HOT = 0, /* must be zero for meta bio */ 1042 WARM, 1043 COLD, 1044 NR_TEMP_TYPE, 1045 }; 1046 1047 enum need_lock_type { 1048 LOCK_REQ = 0, 1049 LOCK_DONE, 1050 LOCK_RETRY, 1051 }; 1052 1053 enum cp_reason_type { 1054 CP_NO_NEEDED, 1055 CP_NON_REGULAR, 1056 CP_COMPRESSED, 1057 CP_HARDLINK, 1058 CP_SB_NEED_CP, 1059 CP_WRONG_PINO, 1060 CP_NO_SPC_ROLL, 1061 CP_NODE_NEED_CP, 1062 CP_FASTBOOT_MODE, 1063 CP_SPEC_LOG_NUM, 1064 CP_RECOVER_DIR, 1065 }; 1066 1067 enum iostat_type { 1068 /* WRITE IO */ 1069 APP_DIRECT_IO, /* app direct write IOs */ 1070 APP_BUFFERED_IO, /* app buffered write IOs */ 1071 APP_WRITE_IO, /* app write IOs */ 1072 APP_MAPPED_IO, /* app mapped IOs */ 1073 FS_DATA_IO, /* data IOs from kworker/fsync/reclaimer */ 1074 FS_NODE_IO, /* node IOs from kworker/fsync/reclaimer */ 1075 FS_META_IO, /* meta IOs from kworker/reclaimer */ 1076 FS_GC_DATA_IO, /* data IOs from forground gc */ 1077 FS_GC_NODE_IO, /* node IOs from forground gc */ 1078 FS_CP_DATA_IO, /* data IOs from checkpoint */ 1079 FS_CP_NODE_IO, /* node IOs from checkpoint */ 1080 FS_CP_META_IO, /* meta IOs from checkpoint */ 1081 1082 /* READ IO */ 1083 APP_DIRECT_READ_IO, /* app direct read IOs */ 1084 APP_BUFFERED_READ_IO, /* app buffered read IOs */ 1085 APP_READ_IO, /* app read IOs */ 1086 APP_MAPPED_READ_IO, /* app mapped read IOs */ 1087 FS_DATA_READ_IO, /* data read IOs */ 1088 FS_GDATA_READ_IO, /* data read IOs from background gc */ 1089 FS_CDATA_READ_IO, /* compressed data read IOs */ 1090 FS_NODE_READ_IO, /* node read IOs */ 1091 FS_META_READ_IO, /* meta read IOs */ 1092 1093 /* other */ 1094 FS_DISCARD, /* discard */ 1095 NR_IO_TYPE, 1096 }; 1097 1098 struct f2fs_io_info { 1099 struct f2fs_sb_info *sbi; /* f2fs_sb_info pointer */ 1100 nid_t ino; /* inode number */ 1101 enum page_type type; /* contains DATA/NODE/META/META_FLUSH */ 1102 enum temp_type temp; /* contains HOT/WARM/COLD */ 1103 int op; /* contains REQ_OP_ */ 1104 int op_flags; /* req_flag_bits */ 1105 block_t new_blkaddr; /* new block address to be written */ 1106 block_t old_blkaddr; /* old block address before Cow */ 1107 struct page *page; /* page to be written */ 1108 struct page *encrypted_page; /* encrypted page */ 1109 struct page *compressed_page; /* compressed page */ 1110 struct list_head list; /* serialize IOs */ 1111 bool submitted; /* indicate IO submission */ 1112 int need_lock; /* indicate we need to lock cp_rwsem */ 1113 bool in_list; /* indicate fio is in io_list */ 1114 bool is_por; /* indicate IO is from recovery or not */ 1115 bool retry; /* need to reallocate block address */ 1116 int compr_blocks; /* # of compressed block addresses */ 1117 bool encrypted; /* indicate file is encrypted */ 1118 enum iostat_type io_type; /* io type */ 1119 struct writeback_control *io_wbc; /* writeback control */ 1120 struct bio **bio; /* bio for ipu */ 1121 sector_t *last_block; /* last block number in bio */ 1122 unsigned char version; /* version of the node */ 1123 }; 1124 1125 struct bio_entry { 1126 struct bio *bio; 1127 struct list_head list; 1128 }; 1129 1130 #define is_read_io(rw) ((rw) == READ) 1131 struct f2fs_bio_info { 1132 struct f2fs_sb_info *sbi; /* f2fs superblock */ 1133 struct bio *bio; /* bios to merge */ 1134 sector_t last_block_in_bio; /* last block number */ 1135 struct f2fs_io_info fio; /* store buffered io info. */ 1136 struct rw_semaphore io_rwsem; /* blocking op for bio */ 1137 spinlock_t io_lock; /* serialize DATA/NODE IOs */ 1138 struct list_head io_list; /* track fios */ 1139 struct list_head bio_list; /* bio entry list head */ 1140 struct rw_semaphore bio_list_lock; /* lock to protect bio entry list */ 1141 }; 1142 1143 #define FDEV(i) (sbi->devs[i]) 1144 #define RDEV(i) (raw_super->devs[i]) 1145 struct f2fs_dev_info { 1146 struct block_device *bdev; 1147 char path[MAX_PATH_LEN]; 1148 unsigned int total_segments; 1149 block_t start_blk; 1150 block_t end_blk; 1151 #ifdef CONFIG_BLK_DEV_ZONED 1152 unsigned int nr_blkz; /* Total number of zones */ 1153 unsigned long *blkz_seq; /* Bitmap indicating sequential zones */ 1154 block_t *zone_capacity_blocks; /* Array of zone capacity in blks */ 1155 #endif 1156 }; 1157 1158 enum inode_type { 1159 DIR_INODE, /* for dirty dir inode */ 1160 FILE_INODE, /* for dirty regular/symlink inode */ 1161 DIRTY_META, /* for all dirtied inode metadata */ 1162 ATOMIC_FILE, /* for all atomic files */ 1163 NR_INODE_TYPE, 1164 }; 1165 1166 /* for inner inode cache management */ 1167 struct inode_management { 1168 struct radix_tree_root ino_root; /* ino entry array */ 1169 spinlock_t ino_lock; /* for ino entry lock */ 1170 struct list_head ino_list; /* inode list head */ 1171 unsigned long ino_num; /* number of entries */ 1172 }; 1173 1174 /* for GC_AT */ 1175 struct atgc_management { 1176 bool atgc_enabled; /* ATGC is enabled or not */ 1177 struct rb_root_cached root; /* root of victim rb-tree */ 1178 struct list_head victim_list; /* linked with all victim entries */ 1179 unsigned int victim_count; /* victim count in rb-tree */ 1180 unsigned int candidate_ratio; /* candidate ratio */ 1181 unsigned int max_candidate_count; /* max candidate count */ 1182 unsigned int age_weight; /* age weight, vblock_weight = 100 - age_weight */ 1183 unsigned long long age_threshold; /* age threshold */ 1184 }; 1185 1186 /* For s_flag in struct f2fs_sb_info */ 1187 enum { 1188 SBI_IS_DIRTY, /* dirty flag for checkpoint */ 1189 SBI_IS_CLOSE, /* specify unmounting */ 1190 SBI_NEED_FSCK, /* need fsck.f2fs to fix */ 1191 SBI_POR_DOING, /* recovery is doing or not */ 1192 SBI_NEED_SB_WRITE, /* need to recover superblock */ 1193 SBI_NEED_CP, /* need to checkpoint */ 1194 SBI_IS_SHUTDOWN, /* shutdown by ioctl */ 1195 SBI_IS_RECOVERED, /* recovered orphan/data */ 1196 SBI_CP_DISABLED, /* CP was disabled last mount */ 1197 SBI_CP_DISABLED_QUICK, /* CP was disabled quickly */ 1198 SBI_QUOTA_NEED_FLUSH, /* need to flush quota info in CP */ 1199 SBI_QUOTA_SKIP_FLUSH, /* skip flushing quota in current CP */ 1200 SBI_QUOTA_NEED_REPAIR, /* quota file may be corrupted */ 1201 SBI_IS_RESIZEFS, /* resizefs is in process */ 1202 }; 1203 1204 enum { 1205 CP_TIME, 1206 REQ_TIME, 1207 DISCARD_TIME, 1208 GC_TIME, 1209 DISABLE_TIME, 1210 UMOUNT_DISCARD_TIMEOUT, 1211 MAX_TIME, 1212 }; 1213 1214 enum { 1215 GC_NORMAL, 1216 GC_IDLE_CB, 1217 GC_IDLE_GREEDY, 1218 GC_IDLE_AT, 1219 GC_URGENT_HIGH, 1220 GC_URGENT_LOW, 1221 }; 1222 1223 enum { 1224 BGGC_MODE_ON, /* background gc is on */ 1225 BGGC_MODE_OFF, /* background gc is off */ 1226 BGGC_MODE_SYNC, /* 1227 * background gc is on, migrating blocks 1228 * like foreground gc 1229 */ 1230 }; 1231 1232 enum { 1233 FS_MODE_ADAPTIVE, /* use both lfs/ssr allocation */ 1234 FS_MODE_LFS, /* use lfs allocation only */ 1235 }; 1236 1237 enum { 1238 WHINT_MODE_OFF, /* not pass down write hints */ 1239 WHINT_MODE_USER, /* try to pass down hints given by users */ 1240 WHINT_MODE_FS, /* pass down hints with F2FS policy */ 1241 }; 1242 1243 enum { 1244 ALLOC_MODE_DEFAULT, /* stay default */ 1245 ALLOC_MODE_REUSE, /* reuse segments as much as possible */ 1246 }; 1247 1248 enum fsync_mode { 1249 FSYNC_MODE_POSIX, /* fsync follows posix semantics */ 1250 FSYNC_MODE_STRICT, /* fsync behaves in line with ext4 */ 1251 FSYNC_MODE_NOBARRIER, /* fsync behaves nobarrier based on posix */ 1252 }; 1253 1254 enum { 1255 COMPR_MODE_FS, /* 1256 * automatically compress compression 1257 * enabled files 1258 */ 1259 COMPR_MODE_USER, /* 1260 * automatical compression is disabled. 1261 * user can control the file compression 1262 * using ioctls 1263 */ 1264 }; 1265 1266 /* 1267 * this value is set in page as a private data which indicate that 1268 * the page is atomically written, and it is in inmem_pages list. 1269 */ 1270 #define ATOMIC_WRITTEN_PAGE ((unsigned long)-1) 1271 #define DUMMY_WRITTEN_PAGE ((unsigned long)-2) 1272 1273 #define IS_ATOMIC_WRITTEN_PAGE(page) \ 1274 (page_private(page) == ATOMIC_WRITTEN_PAGE) 1275 #define IS_DUMMY_WRITTEN_PAGE(page) \ 1276 (page_private(page) == DUMMY_WRITTEN_PAGE) 1277 1278 #ifdef CONFIG_F2FS_IO_TRACE 1279 #define IS_IO_TRACED_PAGE(page) \ 1280 (page_private(page) > 0 && \ 1281 page_private(page) < (unsigned long)PID_MAX_LIMIT) 1282 #else 1283 #define IS_IO_TRACED_PAGE(page) (0) 1284 #endif 1285 1286 /* For compression */ 1287 enum compress_algorithm_type { 1288 COMPRESS_LZO, 1289 COMPRESS_LZ4, 1290 COMPRESS_ZSTD, 1291 COMPRESS_LZORLE, 1292 COMPRESS_MAX, 1293 }; 1294 1295 enum compress_flag { 1296 COMPRESS_CHKSUM, 1297 COMPRESS_MAX_FLAG, 1298 }; 1299 1300 #define COMPRESS_DATA_RESERVED_SIZE 4 1301 struct compress_data { 1302 __le32 clen; /* compressed data size */ 1303 __le32 chksum; /* compressed data chksum */ 1304 __le32 reserved[COMPRESS_DATA_RESERVED_SIZE]; /* reserved */ 1305 u8 cdata[]; /* compressed data */ 1306 }; 1307 1308 #define COMPRESS_HEADER_SIZE (sizeof(struct compress_data)) 1309 1310 #define F2FS_COMPRESSED_PAGE_MAGIC 0xF5F2C000 1311 1312 /* compress context */ 1313 struct compress_ctx { 1314 struct inode *inode; /* inode the context belong to */ 1315 pgoff_t cluster_idx; /* cluster index number */ 1316 unsigned int cluster_size; /* page count in cluster */ 1317 unsigned int log_cluster_size; /* log of cluster size */ 1318 struct page **rpages; /* pages store raw data in cluster */ 1319 unsigned int nr_rpages; /* total page number in rpages */ 1320 struct page **cpages; /* pages store compressed data in cluster */ 1321 unsigned int nr_cpages; /* total page number in cpages */ 1322 void *rbuf; /* virtual mapped address on rpages */ 1323 struct compress_data *cbuf; /* virtual mapped address on cpages */ 1324 size_t rlen; /* valid data length in rbuf */ 1325 size_t clen; /* valid data length in cbuf */ 1326 void *private; /* payload buffer for specified compression algorithm */ 1327 void *private2; /* extra payload buffer */ 1328 }; 1329 1330 /* compress context for write IO path */ 1331 struct compress_io_ctx { 1332 u32 magic; /* magic number to indicate page is compressed */ 1333 struct inode *inode; /* inode the context belong to */ 1334 struct page **rpages; /* pages store raw data in cluster */ 1335 unsigned int nr_rpages; /* total page number in rpages */ 1336 atomic_t pending_pages; /* in-flight compressed page count */ 1337 }; 1338 1339 /* decompress io context for read IO path */ 1340 struct decompress_io_ctx { 1341 u32 magic; /* magic number to indicate page is compressed */ 1342 struct inode *inode; /* inode the context belong to */ 1343 pgoff_t cluster_idx; /* cluster index number */ 1344 unsigned int cluster_size; /* page count in cluster */ 1345 unsigned int log_cluster_size; /* log of cluster size */ 1346 struct page **rpages; /* pages store raw data in cluster */ 1347 unsigned int nr_rpages; /* total page number in rpages */ 1348 struct page **cpages; /* pages store compressed data in cluster */ 1349 unsigned int nr_cpages; /* total page number in cpages */ 1350 struct page **tpages; /* temp pages to pad holes in cluster */ 1351 void *rbuf; /* virtual mapped address on rpages */ 1352 struct compress_data *cbuf; /* virtual mapped address on cpages */ 1353 size_t rlen; /* valid data length in rbuf */ 1354 size_t clen; /* valid data length in cbuf */ 1355 atomic_t pending_pages; /* in-flight compressed page count */ 1356 atomic_t verity_pages; /* in-flight page count for verity */ 1357 bool failed; /* indicate IO error during decompression */ 1358 void *private; /* payload buffer for specified decompression algorithm */ 1359 void *private2; /* extra payload buffer */ 1360 }; 1361 1362 #define NULL_CLUSTER ((unsigned int)(~0)) 1363 #define MIN_COMPRESS_LOG_SIZE 2 1364 #define MAX_COMPRESS_LOG_SIZE 8 1365 #define MAX_COMPRESS_WINDOW_SIZE(log_size) ((PAGE_SIZE) << (log_size)) 1366 1367 struct f2fs_sb_info { 1368 struct super_block *sb; /* pointer to VFS super block */ 1369 struct proc_dir_entry *s_proc; /* proc entry */ 1370 struct f2fs_super_block *raw_super; /* raw super block pointer */ 1371 struct rw_semaphore sb_lock; /* lock for raw super block */ 1372 int valid_super_block; /* valid super block no */ 1373 unsigned long s_flag; /* flags for sbi */ 1374 struct mutex writepages; /* mutex for writepages() */ 1375 1376 #ifdef CONFIG_BLK_DEV_ZONED 1377 unsigned int blocks_per_blkz; /* F2FS blocks per zone */ 1378 unsigned int log_blocks_per_blkz; /* log2 F2FS blocks per zone */ 1379 #endif 1380 1381 /* for node-related operations */ 1382 struct f2fs_nm_info *nm_info; /* node manager */ 1383 struct inode *node_inode; /* cache node blocks */ 1384 1385 /* for segment-related operations */ 1386 struct f2fs_sm_info *sm_info; /* segment manager */ 1387 1388 /* for bio operations */ 1389 struct f2fs_bio_info *write_io[NR_PAGE_TYPE]; /* for write bios */ 1390 /* keep migration IO order for LFS mode */ 1391 struct rw_semaphore io_order_lock; 1392 mempool_t *write_io_dummy; /* Dummy pages */ 1393 1394 /* for checkpoint */ 1395 struct f2fs_checkpoint *ckpt; /* raw checkpoint pointer */ 1396 int cur_cp_pack; /* remain current cp pack */ 1397 spinlock_t cp_lock; /* for flag in ckpt */ 1398 struct inode *meta_inode; /* cache meta blocks */ 1399 struct rw_semaphore cp_global_sem; /* checkpoint procedure lock */ 1400 struct rw_semaphore cp_rwsem; /* blocking FS operations */ 1401 struct rw_semaphore node_write; /* locking node writes */ 1402 struct rw_semaphore node_change; /* locking node change */ 1403 wait_queue_head_t cp_wait; 1404 unsigned long last_time[MAX_TIME]; /* to store time in jiffies */ 1405 long interval_time[MAX_TIME]; /* to store thresholds */ 1406 1407 struct inode_management im[MAX_INO_ENTRY]; /* manage inode cache */ 1408 1409 spinlock_t fsync_node_lock; /* for node entry lock */ 1410 struct list_head fsync_node_list; /* node list head */ 1411 unsigned int fsync_seg_id; /* sequence id */ 1412 unsigned int fsync_node_num; /* number of node entries */ 1413 1414 /* for orphan inode, use 0'th array */ 1415 unsigned int max_orphans; /* max orphan inodes */ 1416 1417 /* for inode management */ 1418 struct list_head inode_list[NR_INODE_TYPE]; /* dirty inode list */ 1419 spinlock_t inode_lock[NR_INODE_TYPE]; /* for dirty inode list lock */ 1420 struct mutex flush_lock; /* for flush exclusion */ 1421 1422 /* for extent tree cache */ 1423 struct radix_tree_root extent_tree_root;/* cache extent cache entries */ 1424 struct mutex extent_tree_lock; /* locking extent radix tree */ 1425 struct list_head extent_list; /* lru list for shrinker */ 1426 spinlock_t extent_lock; /* locking extent lru list */ 1427 atomic_t total_ext_tree; /* extent tree count */ 1428 struct list_head zombie_list; /* extent zombie tree list */ 1429 atomic_t total_zombie_tree; /* extent zombie tree count */ 1430 atomic_t total_ext_node; /* extent info count */ 1431 1432 /* basic filesystem units */ 1433 unsigned int log_sectors_per_block; /* log2 sectors per block */ 1434 unsigned int log_blocksize; /* log2 block size */ 1435 unsigned int blocksize; /* block size */ 1436 unsigned int root_ino_num; /* root inode number*/ 1437 unsigned int node_ino_num; /* node inode number*/ 1438 unsigned int meta_ino_num; /* meta inode number*/ 1439 unsigned int log_blocks_per_seg; /* log2 blocks per segment */ 1440 unsigned int blocks_per_seg; /* blocks per segment */ 1441 unsigned int segs_per_sec; /* segments per section */ 1442 unsigned int secs_per_zone; /* sections per zone */ 1443 unsigned int total_sections; /* total section count */ 1444 unsigned int total_node_count; /* total node block count */ 1445 unsigned int total_valid_node_count; /* valid node block count */ 1446 loff_t max_file_blocks; /* max block index of file */ 1447 int dir_level; /* directory level */ 1448 int readdir_ra; /* readahead inode in readdir */ 1449 u64 max_io_bytes; /* max io bytes to merge IOs */ 1450 1451 block_t user_block_count; /* # of user blocks */ 1452 block_t total_valid_block_count; /* # of valid blocks */ 1453 block_t discard_blks; /* discard command candidats */ 1454 block_t last_valid_block_count; /* for recovery */ 1455 block_t reserved_blocks; /* configurable reserved blocks */ 1456 block_t current_reserved_blocks; /* current reserved blocks */ 1457 1458 /* Additional tracking for no checkpoint mode */ 1459 block_t unusable_block_count; /* # of blocks saved by last cp */ 1460 1461 unsigned int nquota_files; /* # of quota sysfile */ 1462 struct rw_semaphore quota_sem; /* blocking cp for flags */ 1463 1464 /* # of pages, see count_type */ 1465 atomic_t nr_pages[NR_COUNT_TYPE]; 1466 /* # of allocated blocks */ 1467 struct percpu_counter alloc_valid_block_count; 1468 1469 /* writeback control */ 1470 atomic_t wb_sync_req[META]; /* count # of WB_SYNC threads */ 1471 1472 /* valid inode count */ 1473 struct percpu_counter total_valid_inode_count; 1474 1475 struct f2fs_mount_info mount_opt; /* mount options */ 1476 1477 /* for cleaning operations */ 1478 struct rw_semaphore gc_lock; /* 1479 * semaphore for GC, avoid 1480 * race between GC and GC or CP 1481 */ 1482 struct f2fs_gc_kthread *gc_thread; /* GC thread */ 1483 struct atgc_management am; /* atgc management */ 1484 unsigned int cur_victim_sec; /* current victim section num */ 1485 unsigned int gc_mode; /* current GC state */ 1486 unsigned int next_victim_seg[2]; /* next segment in victim section */ 1487 1488 /* for skip statistic */ 1489 unsigned int atomic_files; /* # of opened atomic file */ 1490 unsigned long long skipped_atomic_files[2]; /* FG_GC and BG_GC */ 1491 unsigned long long skipped_gc_rwsem; /* FG_GC only */ 1492 1493 /* threshold for gc trials on pinned files */ 1494 u64 gc_pin_file_threshold; 1495 struct rw_semaphore pin_sem; 1496 1497 /* maximum # of trials to find a victim segment for SSR and GC */ 1498 unsigned int max_victim_search; 1499 /* migration granularity of garbage collection, unit: segment */ 1500 unsigned int migration_granularity; 1501 1502 /* 1503 * for stat information. 1504 * one is for the LFS mode, and the other is for the SSR mode. 1505 */ 1506 #ifdef CONFIG_F2FS_STAT_FS 1507 struct f2fs_stat_info *stat_info; /* FS status information */ 1508 atomic_t meta_count[META_MAX]; /* # of meta blocks */ 1509 unsigned int segment_count[2]; /* # of allocated segments */ 1510 unsigned int block_count[2]; /* # of allocated blocks */ 1511 atomic_t inplace_count; /* # of inplace update */ 1512 atomic64_t total_hit_ext; /* # of lookup extent cache */ 1513 atomic64_t read_hit_rbtree; /* # of hit rbtree extent node */ 1514 atomic64_t read_hit_largest; /* # of hit largest extent node */ 1515 atomic64_t read_hit_cached; /* # of hit cached extent node */ 1516 atomic_t inline_xattr; /* # of inline_xattr inodes */ 1517 atomic_t inline_inode; /* # of inline_data inodes */ 1518 atomic_t inline_dir; /* # of inline_dentry inodes */ 1519 atomic_t compr_inode; /* # of compressed inodes */ 1520 atomic64_t compr_blocks; /* # of compressed blocks */ 1521 atomic_t vw_cnt; /* # of volatile writes */ 1522 atomic_t max_aw_cnt; /* max # of atomic writes */ 1523 atomic_t max_vw_cnt; /* max # of volatile writes */ 1524 unsigned int io_skip_bggc; /* skip background gc for in-flight IO */ 1525 unsigned int other_skip_bggc; /* skip background gc for other reasons */ 1526 unsigned int ndirty_inode[NR_INODE_TYPE]; /* # of dirty inodes */ 1527 #endif 1528 spinlock_t stat_lock; /* lock for stat operations */ 1529 1530 /* For app/fs IO statistics */ 1531 spinlock_t iostat_lock; 1532 unsigned long long rw_iostat[NR_IO_TYPE]; 1533 unsigned long long prev_rw_iostat[NR_IO_TYPE]; 1534 bool iostat_enable; 1535 unsigned long iostat_next_period; 1536 unsigned int iostat_period_ms; 1537 1538 /* to attach REQ_META|REQ_FUA flags */ 1539 unsigned int data_io_flag; 1540 unsigned int node_io_flag; 1541 1542 /* For sysfs suppport */ 1543 struct kobject s_kobj; 1544 struct completion s_kobj_unregister; 1545 1546 /* For shrinker support */ 1547 struct list_head s_list; 1548 int s_ndevs; /* number of devices */ 1549 struct f2fs_dev_info *devs; /* for device list */ 1550 unsigned int dirty_device; /* for checkpoint data flush */ 1551 spinlock_t dev_lock; /* protect dirty_device */ 1552 struct mutex umount_mutex; 1553 unsigned int shrinker_run_no; 1554 1555 /* For write statistics */ 1556 u64 sectors_written_start; 1557 u64 kbytes_written; 1558 1559 /* Reference to checksum algorithm driver via cryptoapi */ 1560 struct crypto_shash *s_chksum_driver; 1561 1562 /* Precomputed FS UUID checksum for seeding other checksums */ 1563 __u32 s_chksum_seed; 1564 1565 struct workqueue_struct *post_read_wq; /* post read workqueue */ 1566 1567 struct kmem_cache *inline_xattr_slab; /* inline xattr entry */ 1568 unsigned int inline_xattr_slab_size; /* default inline xattr slab size */ 1569 1570 #ifdef CONFIG_F2FS_FS_COMPRESSION 1571 struct kmem_cache *page_array_slab; /* page array entry */ 1572 unsigned int page_array_slab_size; /* default page array slab size */ 1573 #endif 1574 }; 1575 1576 struct f2fs_private_dio { 1577 struct inode *inode; 1578 void *orig_private; 1579 bio_end_io_t *orig_end_io; 1580 bool write; 1581 }; 1582 1583 #ifdef CONFIG_F2FS_FAULT_INJECTION 1584 #define f2fs_show_injection_info(sbi, type) \ 1585 printk_ratelimited("%sF2FS-fs (%s) : inject %s in %s of %pS\n", \ 1586 KERN_INFO, sbi->sb->s_id, \ 1587 f2fs_fault_name[type], \ 1588 __func__, __builtin_return_address(0)) 1589 static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type) 1590 { 1591 struct f2fs_fault_info *ffi = &F2FS_OPTION(sbi).fault_info; 1592 1593 if (!ffi->inject_rate) 1594 return false; 1595 1596 if (!IS_FAULT_SET(ffi, type)) 1597 return false; 1598 1599 atomic_inc(&ffi->inject_ops); 1600 if (atomic_read(&ffi->inject_ops) >= ffi->inject_rate) { 1601 atomic_set(&ffi->inject_ops, 0); 1602 return true; 1603 } 1604 return false; 1605 } 1606 #else 1607 #define f2fs_show_injection_info(sbi, type) do { } while (0) 1608 static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type) 1609 { 1610 return false; 1611 } 1612 #endif 1613 1614 /* 1615 * Test if the mounted volume is a multi-device volume. 1616 * - For a single regular disk volume, sbi->s_ndevs is 0. 1617 * - For a single zoned disk volume, sbi->s_ndevs is 1. 1618 * - For a multi-device volume, sbi->s_ndevs is always 2 or more. 1619 */ 1620 static inline bool f2fs_is_multi_device(struct f2fs_sb_info *sbi) 1621 { 1622 return sbi->s_ndevs > 1; 1623 } 1624 1625 static inline void f2fs_update_time(struct f2fs_sb_info *sbi, int type) 1626 { 1627 unsigned long now = jiffies; 1628 1629 sbi->last_time[type] = now; 1630 1631 /* DISCARD_TIME and GC_TIME are based on REQ_TIME */ 1632 if (type == REQ_TIME) { 1633 sbi->last_time[DISCARD_TIME] = now; 1634 sbi->last_time[GC_TIME] = now; 1635 } 1636 } 1637 1638 static inline bool f2fs_time_over(struct f2fs_sb_info *sbi, int type) 1639 { 1640 unsigned long interval = sbi->interval_time[type] * HZ; 1641 1642 return time_after(jiffies, sbi->last_time[type] + interval); 1643 } 1644 1645 static inline unsigned int f2fs_time_to_wait(struct f2fs_sb_info *sbi, 1646 int type) 1647 { 1648 unsigned long interval = sbi->interval_time[type] * HZ; 1649 unsigned int wait_ms = 0; 1650 long delta; 1651 1652 delta = (sbi->last_time[type] + interval) - jiffies; 1653 if (delta > 0) 1654 wait_ms = jiffies_to_msecs(delta); 1655 1656 return wait_ms; 1657 } 1658 1659 /* 1660 * Inline functions 1661 */ 1662 static inline u32 __f2fs_crc32(struct f2fs_sb_info *sbi, u32 crc, 1663 const void *address, unsigned int length) 1664 { 1665 struct { 1666 struct shash_desc shash; 1667 char ctx[4]; 1668 } desc; 1669 int err; 1670 1671 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver) != sizeof(desc.ctx)); 1672 1673 desc.shash.tfm = sbi->s_chksum_driver; 1674 *(u32 *)desc.ctx = crc; 1675 1676 err = crypto_shash_update(&desc.shash, address, length); 1677 BUG_ON(err); 1678 1679 return *(u32 *)desc.ctx; 1680 } 1681 1682 static inline u32 f2fs_crc32(struct f2fs_sb_info *sbi, const void *address, 1683 unsigned int length) 1684 { 1685 return __f2fs_crc32(sbi, F2FS_SUPER_MAGIC, address, length); 1686 } 1687 1688 static inline bool f2fs_crc_valid(struct f2fs_sb_info *sbi, __u32 blk_crc, 1689 void *buf, size_t buf_size) 1690 { 1691 return f2fs_crc32(sbi, buf, buf_size) == blk_crc; 1692 } 1693 1694 static inline u32 f2fs_chksum(struct f2fs_sb_info *sbi, u32 crc, 1695 const void *address, unsigned int length) 1696 { 1697 return __f2fs_crc32(sbi, crc, address, length); 1698 } 1699 1700 static inline struct f2fs_inode_info *F2FS_I(struct inode *inode) 1701 { 1702 return container_of(inode, struct f2fs_inode_info, vfs_inode); 1703 } 1704 1705 static inline struct f2fs_sb_info *F2FS_SB(struct super_block *sb) 1706 { 1707 return sb->s_fs_info; 1708 } 1709 1710 static inline struct f2fs_sb_info *F2FS_I_SB(struct inode *inode) 1711 { 1712 return F2FS_SB(inode->i_sb); 1713 } 1714 1715 static inline struct f2fs_sb_info *F2FS_M_SB(struct address_space *mapping) 1716 { 1717 return F2FS_I_SB(mapping->host); 1718 } 1719 1720 static inline struct f2fs_sb_info *F2FS_P_SB(struct page *page) 1721 { 1722 return F2FS_M_SB(page_file_mapping(page)); 1723 } 1724 1725 static inline struct f2fs_super_block *F2FS_RAW_SUPER(struct f2fs_sb_info *sbi) 1726 { 1727 return (struct f2fs_super_block *)(sbi->raw_super); 1728 } 1729 1730 static inline struct f2fs_checkpoint *F2FS_CKPT(struct f2fs_sb_info *sbi) 1731 { 1732 return (struct f2fs_checkpoint *)(sbi->ckpt); 1733 } 1734 1735 static inline struct f2fs_node *F2FS_NODE(struct page *page) 1736 { 1737 return (struct f2fs_node *)page_address(page); 1738 } 1739 1740 static inline struct f2fs_inode *F2FS_INODE(struct page *page) 1741 { 1742 return &((struct f2fs_node *)page_address(page))->i; 1743 } 1744 1745 static inline struct f2fs_nm_info *NM_I(struct f2fs_sb_info *sbi) 1746 { 1747 return (struct f2fs_nm_info *)(sbi->nm_info); 1748 } 1749 1750 static inline struct f2fs_sm_info *SM_I(struct f2fs_sb_info *sbi) 1751 { 1752 return (struct f2fs_sm_info *)(sbi->sm_info); 1753 } 1754 1755 static inline struct sit_info *SIT_I(struct f2fs_sb_info *sbi) 1756 { 1757 return (struct sit_info *)(SM_I(sbi)->sit_info); 1758 } 1759 1760 static inline struct free_segmap_info *FREE_I(struct f2fs_sb_info *sbi) 1761 { 1762 return (struct free_segmap_info *)(SM_I(sbi)->free_info); 1763 } 1764 1765 static inline struct dirty_seglist_info *DIRTY_I(struct f2fs_sb_info *sbi) 1766 { 1767 return (struct dirty_seglist_info *)(SM_I(sbi)->dirty_info); 1768 } 1769 1770 static inline struct address_space *META_MAPPING(struct f2fs_sb_info *sbi) 1771 { 1772 return sbi->meta_inode->i_mapping; 1773 } 1774 1775 static inline struct address_space *NODE_MAPPING(struct f2fs_sb_info *sbi) 1776 { 1777 return sbi->node_inode->i_mapping; 1778 } 1779 1780 static inline bool is_sbi_flag_set(struct f2fs_sb_info *sbi, unsigned int type) 1781 { 1782 return test_bit(type, &sbi->s_flag); 1783 } 1784 1785 static inline void set_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type) 1786 { 1787 set_bit(type, &sbi->s_flag); 1788 } 1789 1790 static inline void clear_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type) 1791 { 1792 clear_bit(type, &sbi->s_flag); 1793 } 1794 1795 static inline unsigned long long cur_cp_version(struct f2fs_checkpoint *cp) 1796 { 1797 return le64_to_cpu(cp->checkpoint_ver); 1798 } 1799 1800 static inline unsigned long f2fs_qf_ino(struct super_block *sb, int type) 1801 { 1802 if (type < F2FS_MAX_QUOTAS) 1803 return le32_to_cpu(F2FS_SB(sb)->raw_super->qf_ino[type]); 1804 return 0; 1805 } 1806 1807 static inline __u64 cur_cp_crc(struct f2fs_checkpoint *cp) 1808 { 1809 size_t crc_offset = le32_to_cpu(cp->checksum_offset); 1810 return le32_to_cpu(*((__le32 *)((unsigned char *)cp + crc_offset))); 1811 } 1812 1813 static inline bool __is_set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 1814 { 1815 unsigned int ckpt_flags = le32_to_cpu(cp->ckpt_flags); 1816 1817 return ckpt_flags & f; 1818 } 1819 1820 static inline bool is_set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 1821 { 1822 return __is_set_ckpt_flags(F2FS_CKPT(sbi), f); 1823 } 1824 1825 static inline void __set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 1826 { 1827 unsigned int ckpt_flags; 1828 1829 ckpt_flags = le32_to_cpu(cp->ckpt_flags); 1830 ckpt_flags |= f; 1831 cp->ckpt_flags = cpu_to_le32(ckpt_flags); 1832 } 1833 1834 static inline void set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 1835 { 1836 unsigned long flags; 1837 1838 spin_lock_irqsave(&sbi->cp_lock, flags); 1839 __set_ckpt_flags(F2FS_CKPT(sbi), f); 1840 spin_unlock_irqrestore(&sbi->cp_lock, flags); 1841 } 1842 1843 static inline void __clear_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 1844 { 1845 unsigned int ckpt_flags; 1846 1847 ckpt_flags = le32_to_cpu(cp->ckpt_flags); 1848 ckpt_flags &= (~f); 1849 cp->ckpt_flags = cpu_to_le32(ckpt_flags); 1850 } 1851 1852 static inline void clear_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 1853 { 1854 unsigned long flags; 1855 1856 spin_lock_irqsave(&sbi->cp_lock, flags); 1857 __clear_ckpt_flags(F2FS_CKPT(sbi), f); 1858 spin_unlock_irqrestore(&sbi->cp_lock, flags); 1859 } 1860 1861 static inline void disable_nat_bits(struct f2fs_sb_info *sbi, bool lock) 1862 { 1863 unsigned long flags; 1864 unsigned char *nat_bits; 1865 1866 /* 1867 * In order to re-enable nat_bits we need to call fsck.f2fs by 1868 * set_sbi_flag(sbi, SBI_NEED_FSCK). But it may give huge cost, 1869 * so let's rely on regular fsck or unclean shutdown. 1870 */ 1871 1872 if (lock) 1873 spin_lock_irqsave(&sbi->cp_lock, flags); 1874 __clear_ckpt_flags(F2FS_CKPT(sbi), CP_NAT_BITS_FLAG); 1875 nat_bits = NM_I(sbi)->nat_bits; 1876 NM_I(sbi)->nat_bits = NULL; 1877 if (lock) 1878 spin_unlock_irqrestore(&sbi->cp_lock, flags); 1879 1880 kvfree(nat_bits); 1881 } 1882 1883 static inline bool enabled_nat_bits(struct f2fs_sb_info *sbi, 1884 struct cp_control *cpc) 1885 { 1886 bool set = is_set_ckpt_flags(sbi, CP_NAT_BITS_FLAG); 1887 1888 return (cpc) ? (cpc->reason & CP_UMOUNT) && set : set; 1889 } 1890 1891 static inline void f2fs_lock_op(struct f2fs_sb_info *sbi) 1892 { 1893 down_read(&sbi->cp_rwsem); 1894 } 1895 1896 static inline int f2fs_trylock_op(struct f2fs_sb_info *sbi) 1897 { 1898 return down_read_trylock(&sbi->cp_rwsem); 1899 } 1900 1901 static inline void f2fs_unlock_op(struct f2fs_sb_info *sbi) 1902 { 1903 up_read(&sbi->cp_rwsem); 1904 } 1905 1906 static inline void f2fs_lock_all(struct f2fs_sb_info *sbi) 1907 { 1908 down_write(&sbi->cp_rwsem); 1909 } 1910 1911 static inline void f2fs_unlock_all(struct f2fs_sb_info *sbi) 1912 { 1913 up_write(&sbi->cp_rwsem); 1914 } 1915 1916 static inline int __get_cp_reason(struct f2fs_sb_info *sbi) 1917 { 1918 int reason = CP_SYNC; 1919 1920 if (test_opt(sbi, FASTBOOT)) 1921 reason = CP_FASTBOOT; 1922 if (is_sbi_flag_set(sbi, SBI_IS_CLOSE)) 1923 reason = CP_UMOUNT; 1924 return reason; 1925 } 1926 1927 static inline bool __remain_node_summaries(int reason) 1928 { 1929 return (reason & (CP_UMOUNT | CP_FASTBOOT)); 1930 } 1931 1932 static inline bool __exist_node_summaries(struct f2fs_sb_info *sbi) 1933 { 1934 return (is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG) || 1935 is_set_ckpt_flags(sbi, CP_FASTBOOT_FLAG)); 1936 } 1937 1938 /* 1939 * Check whether the inode has blocks or not 1940 */ 1941 static inline int F2FS_HAS_BLOCKS(struct inode *inode) 1942 { 1943 block_t xattr_block = F2FS_I(inode)->i_xattr_nid ? 1 : 0; 1944 1945 return (inode->i_blocks >> F2FS_LOG_SECTORS_PER_BLOCK) > xattr_block; 1946 } 1947 1948 static inline bool f2fs_has_xattr_block(unsigned int ofs) 1949 { 1950 return ofs == XATTR_NODE_OFFSET; 1951 } 1952 1953 static inline bool __allow_reserved_blocks(struct f2fs_sb_info *sbi, 1954 struct inode *inode, bool cap) 1955 { 1956 if (!inode) 1957 return true; 1958 if (!test_opt(sbi, RESERVE_ROOT)) 1959 return false; 1960 if (IS_NOQUOTA(inode)) 1961 return true; 1962 if (uid_eq(F2FS_OPTION(sbi).s_resuid, current_fsuid())) 1963 return true; 1964 if (!gid_eq(F2FS_OPTION(sbi).s_resgid, GLOBAL_ROOT_GID) && 1965 in_group_p(F2FS_OPTION(sbi).s_resgid)) 1966 return true; 1967 if (cap && capable(CAP_SYS_RESOURCE)) 1968 return true; 1969 return false; 1970 } 1971 1972 static inline void f2fs_i_blocks_write(struct inode *, block_t, bool, bool); 1973 static inline int inc_valid_block_count(struct f2fs_sb_info *sbi, 1974 struct inode *inode, blkcnt_t *count) 1975 { 1976 blkcnt_t diff = 0, release = 0; 1977 block_t avail_user_block_count; 1978 int ret; 1979 1980 ret = dquot_reserve_block(inode, *count); 1981 if (ret) 1982 return ret; 1983 1984 if (time_to_inject(sbi, FAULT_BLOCK)) { 1985 f2fs_show_injection_info(sbi, FAULT_BLOCK); 1986 release = *count; 1987 goto release_quota; 1988 } 1989 1990 /* 1991 * let's increase this in prior to actual block count change in order 1992 * for f2fs_sync_file to avoid data races when deciding checkpoint. 1993 */ 1994 percpu_counter_add(&sbi->alloc_valid_block_count, (*count)); 1995 1996 spin_lock(&sbi->stat_lock); 1997 sbi->total_valid_block_count += (block_t)(*count); 1998 avail_user_block_count = sbi->user_block_count - 1999 sbi->current_reserved_blocks; 2000 2001 if (!__allow_reserved_blocks(sbi, inode, true)) 2002 avail_user_block_count -= F2FS_OPTION(sbi).root_reserved_blocks; 2003 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) { 2004 if (avail_user_block_count > sbi->unusable_block_count) 2005 avail_user_block_count -= sbi->unusable_block_count; 2006 else 2007 avail_user_block_count = 0; 2008 } 2009 if (unlikely(sbi->total_valid_block_count > avail_user_block_count)) { 2010 diff = sbi->total_valid_block_count - avail_user_block_count; 2011 if (diff > *count) 2012 diff = *count; 2013 *count -= diff; 2014 release = diff; 2015 sbi->total_valid_block_count -= diff; 2016 if (!*count) { 2017 spin_unlock(&sbi->stat_lock); 2018 goto enospc; 2019 } 2020 } 2021 spin_unlock(&sbi->stat_lock); 2022 2023 if (unlikely(release)) { 2024 percpu_counter_sub(&sbi->alloc_valid_block_count, release); 2025 dquot_release_reservation_block(inode, release); 2026 } 2027 f2fs_i_blocks_write(inode, *count, true, true); 2028 return 0; 2029 2030 enospc: 2031 percpu_counter_sub(&sbi->alloc_valid_block_count, release); 2032 release_quota: 2033 dquot_release_reservation_block(inode, release); 2034 return -ENOSPC; 2035 } 2036 2037 __printf(2, 3) 2038 void f2fs_printk(struct f2fs_sb_info *sbi, const char *fmt, ...); 2039 2040 #define f2fs_err(sbi, fmt, ...) \ 2041 f2fs_printk(sbi, KERN_ERR fmt, ##__VA_ARGS__) 2042 #define f2fs_warn(sbi, fmt, ...) \ 2043 f2fs_printk(sbi, KERN_WARNING fmt, ##__VA_ARGS__) 2044 #define f2fs_notice(sbi, fmt, ...) \ 2045 f2fs_printk(sbi, KERN_NOTICE fmt, ##__VA_ARGS__) 2046 #define f2fs_info(sbi, fmt, ...) \ 2047 f2fs_printk(sbi, KERN_INFO fmt, ##__VA_ARGS__) 2048 #define f2fs_debug(sbi, fmt, ...) \ 2049 f2fs_printk(sbi, KERN_DEBUG fmt, ##__VA_ARGS__) 2050 2051 static inline void dec_valid_block_count(struct f2fs_sb_info *sbi, 2052 struct inode *inode, 2053 block_t count) 2054 { 2055 blkcnt_t sectors = count << F2FS_LOG_SECTORS_PER_BLOCK; 2056 2057 spin_lock(&sbi->stat_lock); 2058 f2fs_bug_on(sbi, sbi->total_valid_block_count < (block_t) count); 2059 sbi->total_valid_block_count -= (block_t)count; 2060 if (sbi->reserved_blocks && 2061 sbi->current_reserved_blocks < sbi->reserved_blocks) 2062 sbi->current_reserved_blocks = min(sbi->reserved_blocks, 2063 sbi->current_reserved_blocks + count); 2064 spin_unlock(&sbi->stat_lock); 2065 if (unlikely(inode->i_blocks < sectors)) { 2066 f2fs_warn(sbi, "Inconsistent i_blocks, ino:%lu, iblocks:%llu, sectors:%llu", 2067 inode->i_ino, 2068 (unsigned long long)inode->i_blocks, 2069 (unsigned long long)sectors); 2070 set_sbi_flag(sbi, SBI_NEED_FSCK); 2071 return; 2072 } 2073 f2fs_i_blocks_write(inode, count, false, true); 2074 } 2075 2076 static inline void inc_page_count(struct f2fs_sb_info *sbi, int count_type) 2077 { 2078 atomic_inc(&sbi->nr_pages[count_type]); 2079 2080 if (count_type == F2FS_DIRTY_DENTS || 2081 count_type == F2FS_DIRTY_NODES || 2082 count_type == F2FS_DIRTY_META || 2083 count_type == F2FS_DIRTY_QDATA || 2084 count_type == F2FS_DIRTY_IMETA) 2085 set_sbi_flag(sbi, SBI_IS_DIRTY); 2086 } 2087 2088 static inline void inode_inc_dirty_pages(struct inode *inode) 2089 { 2090 atomic_inc(&F2FS_I(inode)->dirty_pages); 2091 inc_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ? 2092 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA); 2093 if (IS_NOQUOTA(inode)) 2094 inc_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA); 2095 } 2096 2097 static inline void dec_page_count(struct f2fs_sb_info *sbi, int count_type) 2098 { 2099 atomic_dec(&sbi->nr_pages[count_type]); 2100 } 2101 2102 static inline void inode_dec_dirty_pages(struct inode *inode) 2103 { 2104 if (!S_ISDIR(inode->i_mode) && !S_ISREG(inode->i_mode) && 2105 !S_ISLNK(inode->i_mode)) 2106 return; 2107 2108 atomic_dec(&F2FS_I(inode)->dirty_pages); 2109 dec_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ? 2110 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA); 2111 if (IS_NOQUOTA(inode)) 2112 dec_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA); 2113 } 2114 2115 static inline s64 get_pages(struct f2fs_sb_info *sbi, int count_type) 2116 { 2117 return atomic_read(&sbi->nr_pages[count_type]); 2118 } 2119 2120 static inline int get_dirty_pages(struct inode *inode) 2121 { 2122 return atomic_read(&F2FS_I(inode)->dirty_pages); 2123 } 2124 2125 static inline int get_blocktype_secs(struct f2fs_sb_info *sbi, int block_type) 2126 { 2127 unsigned int pages_per_sec = sbi->segs_per_sec * sbi->blocks_per_seg; 2128 unsigned int segs = (get_pages(sbi, block_type) + pages_per_sec - 1) >> 2129 sbi->log_blocks_per_seg; 2130 2131 return segs / sbi->segs_per_sec; 2132 } 2133 2134 static inline block_t valid_user_blocks(struct f2fs_sb_info *sbi) 2135 { 2136 return sbi->total_valid_block_count; 2137 } 2138 2139 static inline block_t discard_blocks(struct f2fs_sb_info *sbi) 2140 { 2141 return sbi->discard_blks; 2142 } 2143 2144 static inline unsigned long __bitmap_size(struct f2fs_sb_info *sbi, int flag) 2145 { 2146 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi); 2147 2148 /* return NAT or SIT bitmap */ 2149 if (flag == NAT_BITMAP) 2150 return le32_to_cpu(ckpt->nat_ver_bitmap_bytesize); 2151 else if (flag == SIT_BITMAP) 2152 return le32_to_cpu(ckpt->sit_ver_bitmap_bytesize); 2153 2154 return 0; 2155 } 2156 2157 static inline block_t __cp_payload(struct f2fs_sb_info *sbi) 2158 { 2159 return le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_payload); 2160 } 2161 2162 static inline void *__bitmap_ptr(struct f2fs_sb_info *sbi, int flag) 2163 { 2164 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi); 2165 int offset; 2166 2167 if (is_set_ckpt_flags(sbi, CP_LARGE_NAT_BITMAP_FLAG)) { 2168 offset = (flag == SIT_BITMAP) ? 2169 le32_to_cpu(ckpt->nat_ver_bitmap_bytesize) : 0; 2170 /* 2171 * if large_nat_bitmap feature is enabled, leave checksum 2172 * protection for all nat/sit bitmaps. 2173 */ 2174 return &ckpt->sit_nat_version_bitmap + offset + sizeof(__le32); 2175 } 2176 2177 if (__cp_payload(sbi) > 0) { 2178 if (flag == NAT_BITMAP) 2179 return &ckpt->sit_nat_version_bitmap; 2180 else 2181 return (unsigned char *)ckpt + F2FS_BLKSIZE; 2182 } else { 2183 offset = (flag == NAT_BITMAP) ? 2184 le32_to_cpu(ckpt->sit_ver_bitmap_bytesize) : 0; 2185 return &ckpt->sit_nat_version_bitmap + offset; 2186 } 2187 } 2188 2189 static inline block_t __start_cp_addr(struct f2fs_sb_info *sbi) 2190 { 2191 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr); 2192 2193 if (sbi->cur_cp_pack == 2) 2194 start_addr += sbi->blocks_per_seg; 2195 return start_addr; 2196 } 2197 2198 static inline block_t __start_cp_next_addr(struct f2fs_sb_info *sbi) 2199 { 2200 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr); 2201 2202 if (sbi->cur_cp_pack == 1) 2203 start_addr += sbi->blocks_per_seg; 2204 return start_addr; 2205 } 2206 2207 static inline void __set_cp_next_pack(struct f2fs_sb_info *sbi) 2208 { 2209 sbi->cur_cp_pack = (sbi->cur_cp_pack == 1) ? 2 : 1; 2210 } 2211 2212 static inline block_t __start_sum_addr(struct f2fs_sb_info *sbi) 2213 { 2214 return le32_to_cpu(F2FS_CKPT(sbi)->cp_pack_start_sum); 2215 } 2216 2217 static inline int inc_valid_node_count(struct f2fs_sb_info *sbi, 2218 struct inode *inode, bool is_inode) 2219 { 2220 block_t valid_block_count; 2221 unsigned int valid_node_count, user_block_count; 2222 int err; 2223 2224 if (is_inode) { 2225 if (inode) { 2226 err = dquot_alloc_inode(inode); 2227 if (err) 2228 return err; 2229 } 2230 } else { 2231 err = dquot_reserve_block(inode, 1); 2232 if (err) 2233 return err; 2234 } 2235 2236 if (time_to_inject(sbi, FAULT_BLOCK)) { 2237 f2fs_show_injection_info(sbi, FAULT_BLOCK); 2238 goto enospc; 2239 } 2240 2241 spin_lock(&sbi->stat_lock); 2242 2243 valid_block_count = sbi->total_valid_block_count + 2244 sbi->current_reserved_blocks + 1; 2245 2246 if (!__allow_reserved_blocks(sbi, inode, false)) 2247 valid_block_count += F2FS_OPTION(sbi).root_reserved_blocks; 2248 user_block_count = sbi->user_block_count; 2249 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) 2250 user_block_count -= sbi->unusable_block_count; 2251 2252 if (unlikely(valid_block_count > user_block_count)) { 2253 spin_unlock(&sbi->stat_lock); 2254 goto enospc; 2255 } 2256 2257 valid_node_count = sbi->total_valid_node_count + 1; 2258 if (unlikely(valid_node_count > sbi->total_node_count)) { 2259 spin_unlock(&sbi->stat_lock); 2260 goto enospc; 2261 } 2262 2263 sbi->total_valid_node_count++; 2264 sbi->total_valid_block_count++; 2265 spin_unlock(&sbi->stat_lock); 2266 2267 if (inode) { 2268 if (is_inode) 2269 f2fs_mark_inode_dirty_sync(inode, true); 2270 else 2271 f2fs_i_blocks_write(inode, 1, true, true); 2272 } 2273 2274 percpu_counter_inc(&sbi->alloc_valid_block_count); 2275 return 0; 2276 2277 enospc: 2278 if (is_inode) { 2279 if (inode) 2280 dquot_free_inode(inode); 2281 } else { 2282 dquot_release_reservation_block(inode, 1); 2283 } 2284 return -ENOSPC; 2285 } 2286 2287 static inline void dec_valid_node_count(struct f2fs_sb_info *sbi, 2288 struct inode *inode, bool is_inode) 2289 { 2290 spin_lock(&sbi->stat_lock); 2291 2292 f2fs_bug_on(sbi, !sbi->total_valid_block_count); 2293 f2fs_bug_on(sbi, !sbi->total_valid_node_count); 2294 2295 sbi->total_valid_node_count--; 2296 sbi->total_valid_block_count--; 2297 if (sbi->reserved_blocks && 2298 sbi->current_reserved_blocks < sbi->reserved_blocks) 2299 sbi->current_reserved_blocks++; 2300 2301 spin_unlock(&sbi->stat_lock); 2302 2303 if (is_inode) { 2304 dquot_free_inode(inode); 2305 } else { 2306 if (unlikely(inode->i_blocks == 0)) { 2307 f2fs_warn(sbi, "dec_valid_node_count: inconsistent i_blocks, ino:%lu, iblocks:%llu", 2308 inode->i_ino, 2309 (unsigned long long)inode->i_blocks); 2310 set_sbi_flag(sbi, SBI_NEED_FSCK); 2311 return; 2312 } 2313 f2fs_i_blocks_write(inode, 1, false, true); 2314 } 2315 } 2316 2317 static inline unsigned int valid_node_count(struct f2fs_sb_info *sbi) 2318 { 2319 return sbi->total_valid_node_count; 2320 } 2321 2322 static inline void inc_valid_inode_count(struct f2fs_sb_info *sbi) 2323 { 2324 percpu_counter_inc(&sbi->total_valid_inode_count); 2325 } 2326 2327 static inline void dec_valid_inode_count(struct f2fs_sb_info *sbi) 2328 { 2329 percpu_counter_dec(&sbi->total_valid_inode_count); 2330 } 2331 2332 static inline s64 valid_inode_count(struct f2fs_sb_info *sbi) 2333 { 2334 return percpu_counter_sum_positive(&sbi->total_valid_inode_count); 2335 } 2336 2337 static inline struct page *f2fs_grab_cache_page(struct address_space *mapping, 2338 pgoff_t index, bool for_write) 2339 { 2340 struct page *page; 2341 2342 if (IS_ENABLED(CONFIG_F2FS_FAULT_INJECTION)) { 2343 if (!for_write) 2344 page = find_get_page_flags(mapping, index, 2345 FGP_LOCK | FGP_ACCESSED); 2346 else 2347 page = find_lock_page(mapping, index); 2348 if (page) 2349 return page; 2350 2351 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_ALLOC)) { 2352 f2fs_show_injection_info(F2FS_M_SB(mapping), 2353 FAULT_PAGE_ALLOC); 2354 return NULL; 2355 } 2356 } 2357 2358 if (!for_write) 2359 return grab_cache_page(mapping, index); 2360 return grab_cache_page_write_begin(mapping, index, AOP_FLAG_NOFS); 2361 } 2362 2363 static inline struct page *f2fs_pagecache_get_page( 2364 struct address_space *mapping, pgoff_t index, 2365 int fgp_flags, gfp_t gfp_mask) 2366 { 2367 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_GET)) { 2368 f2fs_show_injection_info(F2FS_M_SB(mapping), FAULT_PAGE_GET); 2369 return NULL; 2370 } 2371 2372 return pagecache_get_page(mapping, index, fgp_flags, gfp_mask); 2373 } 2374 2375 static inline void f2fs_copy_page(struct page *src, struct page *dst) 2376 { 2377 char *src_kaddr = kmap(src); 2378 char *dst_kaddr = kmap(dst); 2379 2380 memcpy(dst_kaddr, src_kaddr, PAGE_SIZE); 2381 kunmap(dst); 2382 kunmap(src); 2383 } 2384 2385 static inline void f2fs_put_page(struct page *page, int unlock) 2386 { 2387 if (!page) 2388 return; 2389 2390 if (unlock) { 2391 f2fs_bug_on(F2FS_P_SB(page), !PageLocked(page)); 2392 unlock_page(page); 2393 } 2394 put_page(page); 2395 } 2396 2397 static inline void f2fs_put_dnode(struct dnode_of_data *dn) 2398 { 2399 if (dn->node_page) 2400 f2fs_put_page(dn->node_page, 1); 2401 if (dn->inode_page && dn->node_page != dn->inode_page) 2402 f2fs_put_page(dn->inode_page, 0); 2403 dn->node_page = NULL; 2404 dn->inode_page = NULL; 2405 } 2406 2407 static inline struct kmem_cache *f2fs_kmem_cache_create(const char *name, 2408 size_t size) 2409 { 2410 return kmem_cache_create(name, size, 0, SLAB_RECLAIM_ACCOUNT, NULL); 2411 } 2412 2413 static inline void *f2fs_kmem_cache_alloc(struct kmem_cache *cachep, 2414 gfp_t flags) 2415 { 2416 void *entry; 2417 2418 entry = kmem_cache_alloc(cachep, flags); 2419 if (!entry) 2420 entry = kmem_cache_alloc(cachep, flags | __GFP_NOFAIL); 2421 return entry; 2422 } 2423 2424 static inline bool is_inflight_io(struct f2fs_sb_info *sbi, int type) 2425 { 2426 if (get_pages(sbi, F2FS_RD_DATA) || get_pages(sbi, F2FS_RD_NODE) || 2427 get_pages(sbi, F2FS_RD_META) || get_pages(sbi, F2FS_WB_DATA) || 2428 get_pages(sbi, F2FS_WB_CP_DATA) || 2429 get_pages(sbi, F2FS_DIO_READ) || 2430 get_pages(sbi, F2FS_DIO_WRITE)) 2431 return true; 2432 2433 if (type != DISCARD_TIME && SM_I(sbi) && SM_I(sbi)->dcc_info && 2434 atomic_read(&SM_I(sbi)->dcc_info->queued_discard)) 2435 return true; 2436 2437 if (SM_I(sbi) && SM_I(sbi)->fcc_info && 2438 atomic_read(&SM_I(sbi)->fcc_info->queued_flush)) 2439 return true; 2440 return false; 2441 } 2442 2443 static inline bool is_idle(struct f2fs_sb_info *sbi, int type) 2444 { 2445 if (sbi->gc_mode == GC_URGENT_HIGH) 2446 return true; 2447 2448 if (is_inflight_io(sbi, type)) 2449 return false; 2450 2451 if (sbi->gc_mode == GC_URGENT_LOW && 2452 (type == DISCARD_TIME || type == GC_TIME)) 2453 return true; 2454 2455 return f2fs_time_over(sbi, type); 2456 } 2457 2458 static inline void f2fs_radix_tree_insert(struct radix_tree_root *root, 2459 unsigned long index, void *item) 2460 { 2461 while (radix_tree_insert(root, index, item)) 2462 cond_resched(); 2463 } 2464 2465 #define RAW_IS_INODE(p) ((p)->footer.nid == (p)->footer.ino) 2466 2467 static inline bool IS_INODE(struct page *page) 2468 { 2469 struct f2fs_node *p = F2FS_NODE(page); 2470 2471 return RAW_IS_INODE(p); 2472 } 2473 2474 static inline int offset_in_addr(struct f2fs_inode *i) 2475 { 2476 return (i->i_inline & F2FS_EXTRA_ATTR) ? 2477 (le16_to_cpu(i->i_extra_isize) / sizeof(__le32)) : 0; 2478 } 2479 2480 static inline __le32 *blkaddr_in_node(struct f2fs_node *node) 2481 { 2482 return RAW_IS_INODE(node) ? node->i.i_addr : node->dn.addr; 2483 } 2484 2485 static inline int f2fs_has_extra_attr(struct inode *inode); 2486 static inline block_t data_blkaddr(struct inode *inode, 2487 struct page *node_page, unsigned int offset) 2488 { 2489 struct f2fs_node *raw_node; 2490 __le32 *addr_array; 2491 int base = 0; 2492 bool is_inode = IS_INODE(node_page); 2493 2494 raw_node = F2FS_NODE(node_page); 2495 2496 if (is_inode) { 2497 if (!inode) 2498 /* from GC path only */ 2499 base = offset_in_addr(&raw_node->i); 2500 else if (f2fs_has_extra_attr(inode)) 2501 base = get_extra_isize(inode); 2502 } 2503 2504 addr_array = blkaddr_in_node(raw_node); 2505 return le32_to_cpu(addr_array[base + offset]); 2506 } 2507 2508 static inline block_t f2fs_data_blkaddr(struct dnode_of_data *dn) 2509 { 2510 return data_blkaddr(dn->inode, dn->node_page, dn->ofs_in_node); 2511 } 2512 2513 static inline int f2fs_test_bit(unsigned int nr, char *addr) 2514 { 2515 int mask; 2516 2517 addr += (nr >> 3); 2518 mask = 1 << (7 - (nr & 0x07)); 2519 return mask & *addr; 2520 } 2521 2522 static inline void f2fs_set_bit(unsigned int nr, char *addr) 2523 { 2524 int mask; 2525 2526 addr += (nr >> 3); 2527 mask = 1 << (7 - (nr & 0x07)); 2528 *addr |= mask; 2529 } 2530 2531 static inline void f2fs_clear_bit(unsigned int nr, char *addr) 2532 { 2533 int mask; 2534 2535 addr += (nr >> 3); 2536 mask = 1 << (7 - (nr & 0x07)); 2537 *addr &= ~mask; 2538 } 2539 2540 static inline int f2fs_test_and_set_bit(unsigned int nr, char *addr) 2541 { 2542 int mask; 2543 int ret; 2544 2545 addr += (nr >> 3); 2546 mask = 1 << (7 - (nr & 0x07)); 2547 ret = mask & *addr; 2548 *addr |= mask; 2549 return ret; 2550 } 2551 2552 static inline int f2fs_test_and_clear_bit(unsigned int nr, char *addr) 2553 { 2554 int mask; 2555 int ret; 2556 2557 addr += (nr >> 3); 2558 mask = 1 << (7 - (nr & 0x07)); 2559 ret = mask & *addr; 2560 *addr &= ~mask; 2561 return ret; 2562 } 2563 2564 static inline void f2fs_change_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 /* 2574 * On-disk inode flags (f2fs_inode::i_flags) 2575 */ 2576 #define F2FS_COMPR_FL 0x00000004 /* Compress file */ 2577 #define F2FS_SYNC_FL 0x00000008 /* Synchronous updates */ 2578 #define F2FS_IMMUTABLE_FL 0x00000010 /* Immutable file */ 2579 #define F2FS_APPEND_FL 0x00000020 /* writes to file may only append */ 2580 #define F2FS_NODUMP_FL 0x00000040 /* do not dump file */ 2581 #define F2FS_NOATIME_FL 0x00000080 /* do not update atime */ 2582 #define F2FS_NOCOMP_FL 0x00000400 /* Don't compress */ 2583 #define F2FS_INDEX_FL 0x00001000 /* hash-indexed directory */ 2584 #define F2FS_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */ 2585 #define F2FS_PROJINHERIT_FL 0x20000000 /* Create with parents projid */ 2586 #define F2FS_CASEFOLD_FL 0x40000000 /* Casefolded file */ 2587 2588 /* Flags that should be inherited by new inodes from their parent. */ 2589 #define F2FS_FL_INHERITED (F2FS_SYNC_FL | F2FS_NODUMP_FL | F2FS_NOATIME_FL | \ 2590 F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \ 2591 F2FS_CASEFOLD_FL | F2FS_COMPR_FL | F2FS_NOCOMP_FL) 2592 2593 /* Flags that are appropriate for regular files (all but dir-specific ones). */ 2594 #define F2FS_REG_FLMASK (~(F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \ 2595 F2FS_CASEFOLD_FL)) 2596 2597 /* Flags that are appropriate for non-directories/regular files. */ 2598 #define F2FS_OTHER_FLMASK (F2FS_NODUMP_FL | F2FS_NOATIME_FL) 2599 2600 static inline __u32 f2fs_mask_flags(umode_t mode, __u32 flags) 2601 { 2602 if (S_ISDIR(mode)) 2603 return flags; 2604 else if (S_ISREG(mode)) 2605 return flags & F2FS_REG_FLMASK; 2606 else 2607 return flags & F2FS_OTHER_FLMASK; 2608 } 2609 2610 static inline void __mark_inode_dirty_flag(struct inode *inode, 2611 int flag, bool set) 2612 { 2613 switch (flag) { 2614 case FI_INLINE_XATTR: 2615 case FI_INLINE_DATA: 2616 case FI_INLINE_DENTRY: 2617 case FI_NEW_INODE: 2618 if (set) 2619 return; 2620 fallthrough; 2621 case FI_DATA_EXIST: 2622 case FI_INLINE_DOTS: 2623 case FI_PIN_FILE: 2624 f2fs_mark_inode_dirty_sync(inode, true); 2625 } 2626 } 2627 2628 static inline void set_inode_flag(struct inode *inode, int flag) 2629 { 2630 set_bit(flag, F2FS_I(inode)->flags); 2631 __mark_inode_dirty_flag(inode, flag, true); 2632 } 2633 2634 static inline int is_inode_flag_set(struct inode *inode, int flag) 2635 { 2636 return test_bit(flag, F2FS_I(inode)->flags); 2637 } 2638 2639 static inline void clear_inode_flag(struct inode *inode, int flag) 2640 { 2641 clear_bit(flag, F2FS_I(inode)->flags); 2642 __mark_inode_dirty_flag(inode, flag, false); 2643 } 2644 2645 static inline bool f2fs_verity_in_progress(struct inode *inode) 2646 { 2647 return IS_ENABLED(CONFIG_FS_VERITY) && 2648 is_inode_flag_set(inode, FI_VERITY_IN_PROGRESS); 2649 } 2650 2651 static inline void set_acl_inode(struct inode *inode, umode_t mode) 2652 { 2653 F2FS_I(inode)->i_acl_mode = mode; 2654 set_inode_flag(inode, FI_ACL_MODE); 2655 f2fs_mark_inode_dirty_sync(inode, false); 2656 } 2657 2658 static inline void f2fs_i_links_write(struct inode *inode, bool inc) 2659 { 2660 if (inc) 2661 inc_nlink(inode); 2662 else 2663 drop_nlink(inode); 2664 f2fs_mark_inode_dirty_sync(inode, true); 2665 } 2666 2667 static inline void f2fs_i_blocks_write(struct inode *inode, 2668 block_t diff, bool add, bool claim) 2669 { 2670 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE); 2671 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER); 2672 2673 /* add = 1, claim = 1 should be dquot_reserve_block in pair */ 2674 if (add) { 2675 if (claim) 2676 dquot_claim_block(inode, diff); 2677 else 2678 dquot_alloc_block_nofail(inode, diff); 2679 } else { 2680 dquot_free_block(inode, diff); 2681 } 2682 2683 f2fs_mark_inode_dirty_sync(inode, true); 2684 if (clean || recover) 2685 set_inode_flag(inode, FI_AUTO_RECOVER); 2686 } 2687 2688 static inline void f2fs_i_size_write(struct inode *inode, loff_t i_size) 2689 { 2690 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE); 2691 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER); 2692 2693 if (i_size_read(inode) == i_size) 2694 return; 2695 2696 i_size_write(inode, i_size); 2697 f2fs_mark_inode_dirty_sync(inode, true); 2698 if (clean || recover) 2699 set_inode_flag(inode, FI_AUTO_RECOVER); 2700 } 2701 2702 static inline void f2fs_i_depth_write(struct inode *inode, unsigned int depth) 2703 { 2704 F2FS_I(inode)->i_current_depth = depth; 2705 f2fs_mark_inode_dirty_sync(inode, true); 2706 } 2707 2708 static inline void f2fs_i_gc_failures_write(struct inode *inode, 2709 unsigned int count) 2710 { 2711 F2FS_I(inode)->i_gc_failures[GC_FAILURE_PIN] = count; 2712 f2fs_mark_inode_dirty_sync(inode, true); 2713 } 2714 2715 static inline void f2fs_i_xnid_write(struct inode *inode, nid_t xnid) 2716 { 2717 F2FS_I(inode)->i_xattr_nid = xnid; 2718 f2fs_mark_inode_dirty_sync(inode, true); 2719 } 2720 2721 static inline void f2fs_i_pino_write(struct inode *inode, nid_t pino) 2722 { 2723 F2FS_I(inode)->i_pino = pino; 2724 f2fs_mark_inode_dirty_sync(inode, true); 2725 } 2726 2727 static inline void get_inline_info(struct inode *inode, struct f2fs_inode *ri) 2728 { 2729 struct f2fs_inode_info *fi = F2FS_I(inode); 2730 2731 if (ri->i_inline & F2FS_INLINE_XATTR) 2732 set_bit(FI_INLINE_XATTR, fi->flags); 2733 if (ri->i_inline & F2FS_INLINE_DATA) 2734 set_bit(FI_INLINE_DATA, fi->flags); 2735 if (ri->i_inline & F2FS_INLINE_DENTRY) 2736 set_bit(FI_INLINE_DENTRY, fi->flags); 2737 if (ri->i_inline & F2FS_DATA_EXIST) 2738 set_bit(FI_DATA_EXIST, fi->flags); 2739 if (ri->i_inline & F2FS_INLINE_DOTS) 2740 set_bit(FI_INLINE_DOTS, fi->flags); 2741 if (ri->i_inline & F2FS_EXTRA_ATTR) 2742 set_bit(FI_EXTRA_ATTR, fi->flags); 2743 if (ri->i_inline & F2FS_PIN_FILE) 2744 set_bit(FI_PIN_FILE, fi->flags); 2745 } 2746 2747 static inline void set_raw_inline(struct inode *inode, struct f2fs_inode *ri) 2748 { 2749 ri->i_inline = 0; 2750 2751 if (is_inode_flag_set(inode, FI_INLINE_XATTR)) 2752 ri->i_inline |= F2FS_INLINE_XATTR; 2753 if (is_inode_flag_set(inode, FI_INLINE_DATA)) 2754 ri->i_inline |= F2FS_INLINE_DATA; 2755 if (is_inode_flag_set(inode, FI_INLINE_DENTRY)) 2756 ri->i_inline |= F2FS_INLINE_DENTRY; 2757 if (is_inode_flag_set(inode, FI_DATA_EXIST)) 2758 ri->i_inline |= F2FS_DATA_EXIST; 2759 if (is_inode_flag_set(inode, FI_INLINE_DOTS)) 2760 ri->i_inline |= F2FS_INLINE_DOTS; 2761 if (is_inode_flag_set(inode, FI_EXTRA_ATTR)) 2762 ri->i_inline |= F2FS_EXTRA_ATTR; 2763 if (is_inode_flag_set(inode, FI_PIN_FILE)) 2764 ri->i_inline |= F2FS_PIN_FILE; 2765 } 2766 2767 static inline int f2fs_has_extra_attr(struct inode *inode) 2768 { 2769 return is_inode_flag_set(inode, FI_EXTRA_ATTR); 2770 } 2771 2772 static inline int f2fs_has_inline_xattr(struct inode *inode) 2773 { 2774 return is_inode_flag_set(inode, FI_INLINE_XATTR); 2775 } 2776 2777 static inline int f2fs_compressed_file(struct inode *inode) 2778 { 2779 return S_ISREG(inode->i_mode) && 2780 is_inode_flag_set(inode, FI_COMPRESSED_FILE); 2781 } 2782 2783 static inline bool f2fs_need_compress_data(struct inode *inode) 2784 { 2785 int compress_mode = F2FS_OPTION(F2FS_I_SB(inode)).compress_mode; 2786 2787 if (!f2fs_compressed_file(inode)) 2788 return false; 2789 2790 if (compress_mode == COMPR_MODE_FS) 2791 return true; 2792 else if (compress_mode == COMPR_MODE_USER && 2793 is_inode_flag_set(inode, FI_ENABLE_COMPRESS)) 2794 return true; 2795 2796 return false; 2797 } 2798 2799 static inline unsigned int addrs_per_inode(struct inode *inode) 2800 { 2801 unsigned int addrs = CUR_ADDRS_PER_INODE(inode) - 2802 get_inline_xattr_addrs(inode); 2803 2804 if (!f2fs_compressed_file(inode)) 2805 return addrs; 2806 return ALIGN_DOWN(addrs, F2FS_I(inode)->i_cluster_size); 2807 } 2808 2809 static inline unsigned int addrs_per_block(struct inode *inode) 2810 { 2811 if (!f2fs_compressed_file(inode)) 2812 return DEF_ADDRS_PER_BLOCK; 2813 return ALIGN_DOWN(DEF_ADDRS_PER_BLOCK, F2FS_I(inode)->i_cluster_size); 2814 } 2815 2816 static inline void *inline_xattr_addr(struct inode *inode, struct page *page) 2817 { 2818 struct f2fs_inode *ri = F2FS_INODE(page); 2819 2820 return (void *)&(ri->i_addr[DEF_ADDRS_PER_INODE - 2821 get_inline_xattr_addrs(inode)]); 2822 } 2823 2824 static inline int inline_xattr_size(struct inode *inode) 2825 { 2826 if (f2fs_has_inline_xattr(inode)) 2827 return get_inline_xattr_addrs(inode) * sizeof(__le32); 2828 return 0; 2829 } 2830 2831 static inline int f2fs_has_inline_data(struct inode *inode) 2832 { 2833 return is_inode_flag_set(inode, FI_INLINE_DATA); 2834 } 2835 2836 static inline int f2fs_exist_data(struct inode *inode) 2837 { 2838 return is_inode_flag_set(inode, FI_DATA_EXIST); 2839 } 2840 2841 static inline int f2fs_has_inline_dots(struct inode *inode) 2842 { 2843 return is_inode_flag_set(inode, FI_INLINE_DOTS); 2844 } 2845 2846 static inline int f2fs_is_mmap_file(struct inode *inode) 2847 { 2848 return is_inode_flag_set(inode, FI_MMAP_FILE); 2849 } 2850 2851 static inline bool f2fs_is_pinned_file(struct inode *inode) 2852 { 2853 return is_inode_flag_set(inode, FI_PIN_FILE); 2854 } 2855 2856 static inline bool f2fs_is_atomic_file(struct inode *inode) 2857 { 2858 return is_inode_flag_set(inode, FI_ATOMIC_FILE); 2859 } 2860 2861 static inline bool f2fs_is_commit_atomic_write(struct inode *inode) 2862 { 2863 return is_inode_flag_set(inode, FI_ATOMIC_COMMIT); 2864 } 2865 2866 static inline bool f2fs_is_volatile_file(struct inode *inode) 2867 { 2868 return is_inode_flag_set(inode, FI_VOLATILE_FILE); 2869 } 2870 2871 static inline bool f2fs_is_first_block_written(struct inode *inode) 2872 { 2873 return is_inode_flag_set(inode, FI_FIRST_BLOCK_WRITTEN); 2874 } 2875 2876 static inline bool f2fs_is_drop_cache(struct inode *inode) 2877 { 2878 return is_inode_flag_set(inode, FI_DROP_CACHE); 2879 } 2880 2881 static inline void *inline_data_addr(struct inode *inode, struct page *page) 2882 { 2883 struct f2fs_inode *ri = F2FS_INODE(page); 2884 int extra_size = get_extra_isize(inode); 2885 2886 return (void *)&(ri->i_addr[extra_size + DEF_INLINE_RESERVED_SIZE]); 2887 } 2888 2889 static inline int f2fs_has_inline_dentry(struct inode *inode) 2890 { 2891 return is_inode_flag_set(inode, FI_INLINE_DENTRY); 2892 } 2893 2894 static inline int is_file(struct inode *inode, int type) 2895 { 2896 return F2FS_I(inode)->i_advise & type; 2897 } 2898 2899 static inline void set_file(struct inode *inode, int type) 2900 { 2901 F2FS_I(inode)->i_advise |= type; 2902 f2fs_mark_inode_dirty_sync(inode, true); 2903 } 2904 2905 static inline void clear_file(struct inode *inode, int type) 2906 { 2907 F2FS_I(inode)->i_advise &= ~type; 2908 f2fs_mark_inode_dirty_sync(inode, true); 2909 } 2910 2911 static inline bool f2fs_is_time_consistent(struct inode *inode) 2912 { 2913 if (!timespec64_equal(F2FS_I(inode)->i_disk_time, &inode->i_atime)) 2914 return false; 2915 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 1, &inode->i_ctime)) 2916 return false; 2917 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 2, &inode->i_mtime)) 2918 return false; 2919 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 3, 2920 &F2FS_I(inode)->i_crtime)) 2921 return false; 2922 return true; 2923 } 2924 2925 static inline bool f2fs_skip_inode_update(struct inode *inode, int dsync) 2926 { 2927 bool ret; 2928 2929 if (dsync) { 2930 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 2931 2932 spin_lock(&sbi->inode_lock[DIRTY_META]); 2933 ret = list_empty(&F2FS_I(inode)->gdirty_list); 2934 spin_unlock(&sbi->inode_lock[DIRTY_META]); 2935 return ret; 2936 } 2937 if (!is_inode_flag_set(inode, FI_AUTO_RECOVER) || 2938 file_keep_isize(inode) || 2939 i_size_read(inode) & ~PAGE_MASK) 2940 return false; 2941 2942 if (!f2fs_is_time_consistent(inode)) 2943 return false; 2944 2945 spin_lock(&F2FS_I(inode)->i_size_lock); 2946 ret = F2FS_I(inode)->last_disk_size == i_size_read(inode); 2947 spin_unlock(&F2FS_I(inode)->i_size_lock); 2948 2949 return ret; 2950 } 2951 2952 static inline bool f2fs_readonly(struct super_block *sb) 2953 { 2954 return sb_rdonly(sb); 2955 } 2956 2957 static inline bool f2fs_cp_error(struct f2fs_sb_info *sbi) 2958 { 2959 return is_set_ckpt_flags(sbi, CP_ERROR_FLAG); 2960 } 2961 2962 static inline bool is_dot_dotdot(const u8 *name, size_t len) 2963 { 2964 if (len == 1 && name[0] == '.') 2965 return true; 2966 2967 if (len == 2 && name[0] == '.' && name[1] == '.') 2968 return true; 2969 2970 return false; 2971 } 2972 2973 static inline bool f2fs_may_extent_tree(struct inode *inode) 2974 { 2975 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 2976 2977 if (!test_opt(sbi, EXTENT_CACHE) || 2978 is_inode_flag_set(inode, FI_NO_EXTENT) || 2979 is_inode_flag_set(inode, FI_COMPRESSED_FILE)) 2980 return false; 2981 2982 /* 2983 * for recovered files during mount do not create extents 2984 * if shrinker is not registered. 2985 */ 2986 if (list_empty(&sbi->s_list)) 2987 return false; 2988 2989 return S_ISREG(inode->i_mode); 2990 } 2991 2992 static inline void *f2fs_kmalloc(struct f2fs_sb_info *sbi, 2993 size_t size, gfp_t flags) 2994 { 2995 if (time_to_inject(sbi, FAULT_KMALLOC)) { 2996 f2fs_show_injection_info(sbi, FAULT_KMALLOC); 2997 return NULL; 2998 } 2999 3000 return kmalloc(size, flags); 3001 } 3002 3003 static inline void *f2fs_kzalloc(struct f2fs_sb_info *sbi, 3004 size_t size, gfp_t flags) 3005 { 3006 return f2fs_kmalloc(sbi, size, flags | __GFP_ZERO); 3007 } 3008 3009 static inline void *f2fs_kvmalloc(struct f2fs_sb_info *sbi, 3010 size_t size, gfp_t flags) 3011 { 3012 if (time_to_inject(sbi, FAULT_KVMALLOC)) { 3013 f2fs_show_injection_info(sbi, FAULT_KVMALLOC); 3014 return NULL; 3015 } 3016 3017 return kvmalloc(size, flags); 3018 } 3019 3020 static inline void *f2fs_kvzalloc(struct f2fs_sb_info *sbi, 3021 size_t size, gfp_t flags) 3022 { 3023 return f2fs_kvmalloc(sbi, size, flags | __GFP_ZERO); 3024 } 3025 3026 static inline int get_extra_isize(struct inode *inode) 3027 { 3028 return F2FS_I(inode)->i_extra_isize / sizeof(__le32); 3029 } 3030 3031 static inline int get_inline_xattr_addrs(struct inode *inode) 3032 { 3033 return F2FS_I(inode)->i_inline_xattr_size; 3034 } 3035 3036 #define f2fs_get_inode_mode(i) \ 3037 ((is_inode_flag_set(i, FI_ACL_MODE)) ? \ 3038 (F2FS_I(i)->i_acl_mode) : ((i)->i_mode)) 3039 3040 #define F2FS_TOTAL_EXTRA_ATTR_SIZE \ 3041 (offsetof(struct f2fs_inode, i_extra_end) - \ 3042 offsetof(struct f2fs_inode, i_extra_isize)) \ 3043 3044 #define F2FS_OLD_ATTRIBUTE_SIZE (offsetof(struct f2fs_inode, i_addr)) 3045 #define F2FS_FITS_IN_INODE(f2fs_inode, extra_isize, field) \ 3046 ((offsetof(typeof(*(f2fs_inode)), field) + \ 3047 sizeof((f2fs_inode)->field)) \ 3048 <= (F2FS_OLD_ATTRIBUTE_SIZE + (extra_isize))) \ 3049 3050 #define DEFAULT_IOSTAT_PERIOD_MS 3000 3051 #define MIN_IOSTAT_PERIOD_MS 100 3052 /* maximum period of iostat tracing is 1 day */ 3053 #define MAX_IOSTAT_PERIOD_MS 8640000 3054 3055 static inline void f2fs_reset_iostat(struct f2fs_sb_info *sbi) 3056 { 3057 int i; 3058 3059 spin_lock(&sbi->iostat_lock); 3060 for (i = 0; i < NR_IO_TYPE; i++) { 3061 sbi->rw_iostat[i] = 0; 3062 sbi->prev_rw_iostat[i] = 0; 3063 } 3064 spin_unlock(&sbi->iostat_lock); 3065 } 3066 3067 extern void f2fs_record_iostat(struct f2fs_sb_info *sbi); 3068 3069 static inline void f2fs_update_iostat(struct f2fs_sb_info *sbi, 3070 enum iostat_type type, unsigned long long io_bytes) 3071 { 3072 if (!sbi->iostat_enable) 3073 return; 3074 spin_lock(&sbi->iostat_lock); 3075 sbi->rw_iostat[type] += io_bytes; 3076 3077 if (type == APP_WRITE_IO || type == APP_DIRECT_IO) 3078 sbi->rw_iostat[APP_BUFFERED_IO] = 3079 sbi->rw_iostat[APP_WRITE_IO] - 3080 sbi->rw_iostat[APP_DIRECT_IO]; 3081 3082 if (type == APP_READ_IO || type == APP_DIRECT_READ_IO) 3083 sbi->rw_iostat[APP_BUFFERED_READ_IO] = 3084 sbi->rw_iostat[APP_READ_IO] - 3085 sbi->rw_iostat[APP_DIRECT_READ_IO]; 3086 spin_unlock(&sbi->iostat_lock); 3087 3088 f2fs_record_iostat(sbi); 3089 } 3090 3091 #define __is_large_section(sbi) ((sbi)->segs_per_sec > 1) 3092 3093 #define __is_meta_io(fio) (PAGE_TYPE_OF_BIO((fio)->type) == META) 3094 3095 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi, 3096 block_t blkaddr, int type); 3097 static inline void verify_blkaddr(struct f2fs_sb_info *sbi, 3098 block_t blkaddr, int type) 3099 { 3100 if (!f2fs_is_valid_blkaddr(sbi, blkaddr, type)) { 3101 f2fs_err(sbi, "invalid blkaddr: %u, type: %d, run fsck to fix.", 3102 blkaddr, type); 3103 f2fs_bug_on(sbi, 1); 3104 } 3105 } 3106 3107 static inline bool __is_valid_data_blkaddr(block_t blkaddr) 3108 { 3109 if (blkaddr == NEW_ADDR || blkaddr == NULL_ADDR || 3110 blkaddr == COMPRESS_ADDR) 3111 return false; 3112 return true; 3113 } 3114 3115 static inline void f2fs_set_page_private(struct page *page, 3116 unsigned long data) 3117 { 3118 if (PagePrivate(page)) 3119 return; 3120 3121 attach_page_private(page, (void *)data); 3122 } 3123 3124 static inline void f2fs_clear_page_private(struct page *page) 3125 { 3126 detach_page_private(page); 3127 } 3128 3129 /* 3130 * file.c 3131 */ 3132 int f2fs_sync_file(struct file *file, loff_t start, loff_t end, int datasync); 3133 void f2fs_truncate_data_blocks(struct dnode_of_data *dn); 3134 int f2fs_do_truncate_blocks(struct inode *inode, u64 from, bool lock); 3135 int f2fs_truncate_blocks(struct inode *inode, u64 from, bool lock); 3136 int f2fs_truncate(struct inode *inode); 3137 int f2fs_getattr(const struct path *path, struct kstat *stat, 3138 u32 request_mask, unsigned int flags); 3139 int f2fs_setattr(struct dentry *dentry, struct iattr *attr); 3140 int f2fs_truncate_hole(struct inode *inode, pgoff_t pg_start, pgoff_t pg_end); 3141 void f2fs_truncate_data_blocks_range(struct dnode_of_data *dn, int count); 3142 int f2fs_precache_extents(struct inode *inode); 3143 long f2fs_ioctl(struct file *filp, unsigned int cmd, unsigned long arg); 3144 long f2fs_compat_ioctl(struct file *file, unsigned int cmd, unsigned long arg); 3145 int f2fs_transfer_project_quota(struct inode *inode, kprojid_t kprojid); 3146 int f2fs_pin_file_control(struct inode *inode, bool inc); 3147 3148 /* 3149 * inode.c 3150 */ 3151 void f2fs_set_inode_flags(struct inode *inode); 3152 bool f2fs_inode_chksum_verify(struct f2fs_sb_info *sbi, struct page *page); 3153 void f2fs_inode_chksum_set(struct f2fs_sb_info *sbi, struct page *page); 3154 struct inode *f2fs_iget(struct super_block *sb, unsigned long ino); 3155 struct inode *f2fs_iget_retry(struct super_block *sb, unsigned long ino); 3156 int f2fs_try_to_free_nats(struct f2fs_sb_info *sbi, int nr_shrink); 3157 void f2fs_update_inode(struct inode *inode, struct page *node_page); 3158 void f2fs_update_inode_page(struct inode *inode); 3159 int f2fs_write_inode(struct inode *inode, struct writeback_control *wbc); 3160 void f2fs_evict_inode(struct inode *inode); 3161 void f2fs_handle_failed_inode(struct inode *inode); 3162 3163 /* 3164 * namei.c 3165 */ 3166 int f2fs_update_extension_list(struct f2fs_sb_info *sbi, const char *name, 3167 bool hot, bool set); 3168 struct dentry *f2fs_get_parent(struct dentry *child); 3169 3170 /* 3171 * dir.c 3172 */ 3173 unsigned char f2fs_get_de_type(struct f2fs_dir_entry *de); 3174 int f2fs_init_casefolded_name(const struct inode *dir, 3175 struct f2fs_filename *fname); 3176 int f2fs_setup_filename(struct inode *dir, const struct qstr *iname, 3177 int lookup, struct f2fs_filename *fname); 3178 int f2fs_prepare_lookup(struct inode *dir, struct dentry *dentry, 3179 struct f2fs_filename *fname); 3180 void f2fs_free_filename(struct f2fs_filename *fname); 3181 struct f2fs_dir_entry *f2fs_find_target_dentry(const struct f2fs_dentry_ptr *d, 3182 const struct f2fs_filename *fname, int *max_slots); 3183 int f2fs_fill_dentries(struct dir_context *ctx, struct f2fs_dentry_ptr *d, 3184 unsigned int start_pos, struct fscrypt_str *fstr); 3185 void f2fs_do_make_empty_dir(struct inode *inode, struct inode *parent, 3186 struct f2fs_dentry_ptr *d); 3187 struct page *f2fs_init_inode_metadata(struct inode *inode, struct inode *dir, 3188 const struct f2fs_filename *fname, struct page *dpage); 3189 void f2fs_update_parent_metadata(struct inode *dir, struct inode *inode, 3190 unsigned int current_depth); 3191 int f2fs_room_for_filename(const void *bitmap, int slots, int max_slots); 3192 void f2fs_drop_nlink(struct inode *dir, struct inode *inode); 3193 struct f2fs_dir_entry *__f2fs_find_entry(struct inode *dir, 3194 const struct f2fs_filename *fname, 3195 struct page **res_page); 3196 struct f2fs_dir_entry *f2fs_find_entry(struct inode *dir, 3197 const struct qstr *child, struct page **res_page); 3198 struct f2fs_dir_entry *f2fs_parent_dir(struct inode *dir, struct page **p); 3199 ino_t f2fs_inode_by_name(struct inode *dir, const struct qstr *qstr, 3200 struct page **page); 3201 void f2fs_set_link(struct inode *dir, struct f2fs_dir_entry *de, 3202 struct page *page, struct inode *inode); 3203 bool f2fs_has_enough_room(struct inode *dir, struct page *ipage, 3204 const struct f2fs_filename *fname); 3205 void f2fs_update_dentry(nid_t ino, umode_t mode, struct f2fs_dentry_ptr *d, 3206 const struct fscrypt_str *name, f2fs_hash_t name_hash, 3207 unsigned int bit_pos); 3208 int f2fs_add_regular_entry(struct inode *dir, const struct f2fs_filename *fname, 3209 struct inode *inode, nid_t ino, umode_t mode); 3210 int f2fs_add_dentry(struct inode *dir, const struct f2fs_filename *fname, 3211 struct inode *inode, nid_t ino, umode_t mode); 3212 int f2fs_do_add_link(struct inode *dir, const struct qstr *name, 3213 struct inode *inode, nid_t ino, umode_t mode); 3214 void f2fs_delete_entry(struct f2fs_dir_entry *dentry, struct page *page, 3215 struct inode *dir, struct inode *inode); 3216 int f2fs_do_tmpfile(struct inode *inode, struct inode *dir); 3217 bool f2fs_empty_dir(struct inode *dir); 3218 3219 static inline int f2fs_add_link(struct dentry *dentry, struct inode *inode) 3220 { 3221 if (fscrypt_is_nokey_name(dentry)) 3222 return -ENOKEY; 3223 return f2fs_do_add_link(d_inode(dentry->d_parent), &dentry->d_name, 3224 inode, inode->i_ino, inode->i_mode); 3225 } 3226 3227 /* 3228 * super.c 3229 */ 3230 int f2fs_inode_dirtied(struct inode *inode, bool sync); 3231 void f2fs_inode_synced(struct inode *inode); 3232 int f2fs_enable_quota_files(struct f2fs_sb_info *sbi, bool rdonly); 3233 int f2fs_quota_sync(struct super_block *sb, int type); 3234 void f2fs_quota_off_umount(struct super_block *sb); 3235 int f2fs_commit_super(struct f2fs_sb_info *sbi, bool recover); 3236 int f2fs_sync_fs(struct super_block *sb, int sync); 3237 int f2fs_sanity_check_ckpt(struct f2fs_sb_info *sbi); 3238 3239 /* 3240 * hash.c 3241 */ 3242 void f2fs_hash_filename(const struct inode *dir, struct f2fs_filename *fname); 3243 3244 /* 3245 * node.c 3246 */ 3247 struct dnode_of_data; 3248 struct node_info; 3249 3250 int f2fs_check_nid_range(struct f2fs_sb_info *sbi, nid_t nid); 3251 bool f2fs_available_free_memory(struct f2fs_sb_info *sbi, int type); 3252 bool f2fs_in_warm_node_list(struct f2fs_sb_info *sbi, struct page *page); 3253 void f2fs_init_fsync_node_info(struct f2fs_sb_info *sbi); 3254 void f2fs_del_fsync_node_entry(struct f2fs_sb_info *sbi, struct page *page); 3255 void f2fs_reset_fsync_node_info(struct f2fs_sb_info *sbi); 3256 int f2fs_need_dentry_mark(struct f2fs_sb_info *sbi, nid_t nid); 3257 bool f2fs_is_checkpointed_node(struct f2fs_sb_info *sbi, nid_t nid); 3258 bool f2fs_need_inode_block_update(struct f2fs_sb_info *sbi, nid_t ino); 3259 int f2fs_get_node_info(struct f2fs_sb_info *sbi, nid_t nid, 3260 struct node_info *ni); 3261 pgoff_t f2fs_get_next_page_offset(struct dnode_of_data *dn, pgoff_t pgofs); 3262 int f2fs_get_dnode_of_data(struct dnode_of_data *dn, pgoff_t index, int mode); 3263 int f2fs_truncate_inode_blocks(struct inode *inode, pgoff_t from); 3264 int f2fs_truncate_xattr_node(struct inode *inode); 3265 int f2fs_wait_on_node_pages_writeback(struct f2fs_sb_info *sbi, 3266 unsigned int seq_id); 3267 int f2fs_remove_inode_page(struct inode *inode); 3268 struct page *f2fs_new_inode_page(struct inode *inode); 3269 struct page *f2fs_new_node_page(struct dnode_of_data *dn, unsigned int ofs); 3270 void f2fs_ra_node_page(struct f2fs_sb_info *sbi, nid_t nid); 3271 struct page *f2fs_get_node_page(struct f2fs_sb_info *sbi, pgoff_t nid); 3272 struct page *f2fs_get_node_page_ra(struct page *parent, int start); 3273 int f2fs_move_node_page(struct page *node_page, int gc_type); 3274 void f2fs_flush_inline_data(struct f2fs_sb_info *sbi); 3275 int f2fs_fsync_node_pages(struct f2fs_sb_info *sbi, struct inode *inode, 3276 struct writeback_control *wbc, bool atomic, 3277 unsigned int *seq_id); 3278 int f2fs_sync_node_pages(struct f2fs_sb_info *sbi, 3279 struct writeback_control *wbc, 3280 bool do_balance, enum iostat_type io_type); 3281 int f2fs_build_free_nids(struct f2fs_sb_info *sbi, bool sync, bool mount); 3282 bool f2fs_alloc_nid(struct f2fs_sb_info *sbi, nid_t *nid); 3283 void f2fs_alloc_nid_done(struct f2fs_sb_info *sbi, nid_t nid); 3284 void f2fs_alloc_nid_failed(struct f2fs_sb_info *sbi, nid_t nid); 3285 int f2fs_try_to_free_nids(struct f2fs_sb_info *sbi, int nr_shrink); 3286 int f2fs_recover_inline_xattr(struct inode *inode, struct page *page); 3287 int f2fs_recover_xattr_data(struct inode *inode, struct page *page); 3288 int f2fs_recover_inode_page(struct f2fs_sb_info *sbi, struct page *page); 3289 int f2fs_restore_node_summary(struct f2fs_sb_info *sbi, 3290 unsigned int segno, struct f2fs_summary_block *sum); 3291 int f2fs_flush_nat_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3292 int f2fs_build_node_manager(struct f2fs_sb_info *sbi); 3293 void f2fs_destroy_node_manager(struct f2fs_sb_info *sbi); 3294 int __init f2fs_create_node_manager_caches(void); 3295 void f2fs_destroy_node_manager_caches(void); 3296 3297 /* 3298 * segment.c 3299 */ 3300 bool f2fs_need_SSR(struct f2fs_sb_info *sbi); 3301 void f2fs_register_inmem_page(struct inode *inode, struct page *page); 3302 void f2fs_drop_inmem_pages_all(struct f2fs_sb_info *sbi, bool gc_failure); 3303 void f2fs_drop_inmem_pages(struct inode *inode); 3304 void f2fs_drop_inmem_page(struct inode *inode, struct page *page); 3305 int f2fs_commit_inmem_pages(struct inode *inode); 3306 void f2fs_balance_fs(struct f2fs_sb_info *sbi, bool need); 3307 void f2fs_balance_fs_bg(struct f2fs_sb_info *sbi, bool from_bg); 3308 int f2fs_issue_flush(struct f2fs_sb_info *sbi, nid_t ino); 3309 int f2fs_create_flush_cmd_control(struct f2fs_sb_info *sbi); 3310 int f2fs_flush_device_cache(struct f2fs_sb_info *sbi); 3311 void f2fs_destroy_flush_cmd_control(struct f2fs_sb_info *sbi, bool free); 3312 void f2fs_invalidate_blocks(struct f2fs_sb_info *sbi, block_t addr); 3313 bool f2fs_is_checkpointed_data(struct f2fs_sb_info *sbi, block_t blkaddr); 3314 void f2fs_drop_discard_cmd(struct f2fs_sb_info *sbi); 3315 void f2fs_stop_discard_thread(struct f2fs_sb_info *sbi); 3316 bool f2fs_issue_discard_timeout(struct f2fs_sb_info *sbi); 3317 void f2fs_clear_prefree_segments(struct f2fs_sb_info *sbi, 3318 struct cp_control *cpc); 3319 void f2fs_dirty_to_prefree(struct f2fs_sb_info *sbi); 3320 block_t f2fs_get_unusable_blocks(struct f2fs_sb_info *sbi); 3321 int f2fs_disable_cp_again(struct f2fs_sb_info *sbi, block_t unusable); 3322 void f2fs_release_discard_addrs(struct f2fs_sb_info *sbi); 3323 int f2fs_npages_for_summary_flush(struct f2fs_sb_info *sbi, bool for_ra); 3324 void f2fs_init_inmem_curseg(struct f2fs_sb_info *sbi); 3325 void f2fs_save_inmem_curseg(struct f2fs_sb_info *sbi); 3326 void f2fs_restore_inmem_curseg(struct f2fs_sb_info *sbi); 3327 void f2fs_get_new_segment(struct f2fs_sb_info *sbi, 3328 unsigned int *newseg, bool new_sec, int dir); 3329 void f2fs_allocate_segment_for_resize(struct f2fs_sb_info *sbi, int type, 3330 unsigned int start, unsigned int end); 3331 void f2fs_allocate_new_segment(struct f2fs_sb_info *sbi, int type); 3332 void f2fs_allocate_new_segments(struct f2fs_sb_info *sbi); 3333 int f2fs_trim_fs(struct f2fs_sb_info *sbi, struct fstrim_range *range); 3334 bool f2fs_exist_trim_candidates(struct f2fs_sb_info *sbi, 3335 struct cp_control *cpc); 3336 struct page *f2fs_get_sum_page(struct f2fs_sb_info *sbi, unsigned int segno); 3337 void f2fs_update_meta_page(struct f2fs_sb_info *sbi, void *src, 3338 block_t blk_addr); 3339 void f2fs_do_write_meta_page(struct f2fs_sb_info *sbi, struct page *page, 3340 enum iostat_type io_type); 3341 void f2fs_do_write_node_page(unsigned int nid, struct f2fs_io_info *fio); 3342 void f2fs_outplace_write_data(struct dnode_of_data *dn, 3343 struct f2fs_io_info *fio); 3344 int f2fs_inplace_write_data(struct f2fs_io_info *fio); 3345 void f2fs_do_replace_block(struct f2fs_sb_info *sbi, struct f2fs_summary *sum, 3346 block_t old_blkaddr, block_t new_blkaddr, 3347 bool recover_curseg, bool recover_newaddr, 3348 bool from_gc); 3349 void f2fs_replace_block(struct f2fs_sb_info *sbi, struct dnode_of_data *dn, 3350 block_t old_addr, block_t new_addr, 3351 unsigned char version, bool recover_curseg, 3352 bool recover_newaddr); 3353 void f2fs_allocate_data_block(struct f2fs_sb_info *sbi, struct page *page, 3354 block_t old_blkaddr, block_t *new_blkaddr, 3355 struct f2fs_summary *sum, int type, 3356 struct f2fs_io_info *fio); 3357 void f2fs_wait_on_page_writeback(struct page *page, 3358 enum page_type type, bool ordered, bool locked); 3359 void f2fs_wait_on_block_writeback(struct inode *inode, block_t blkaddr); 3360 void f2fs_wait_on_block_writeback_range(struct inode *inode, block_t blkaddr, 3361 block_t len); 3362 void f2fs_write_data_summaries(struct f2fs_sb_info *sbi, block_t start_blk); 3363 void f2fs_write_node_summaries(struct f2fs_sb_info *sbi, block_t start_blk); 3364 int f2fs_lookup_journal_in_cursum(struct f2fs_journal *journal, int type, 3365 unsigned int val, int alloc); 3366 void f2fs_flush_sit_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3367 int f2fs_fix_curseg_write_pointer(struct f2fs_sb_info *sbi); 3368 int f2fs_check_write_pointer(struct f2fs_sb_info *sbi); 3369 int f2fs_build_segment_manager(struct f2fs_sb_info *sbi); 3370 void f2fs_destroy_segment_manager(struct f2fs_sb_info *sbi); 3371 int __init f2fs_create_segment_manager_caches(void); 3372 void f2fs_destroy_segment_manager_caches(void); 3373 int f2fs_rw_hint_to_seg_type(enum rw_hint hint); 3374 enum rw_hint f2fs_io_type_to_rw_hint(struct f2fs_sb_info *sbi, 3375 enum page_type type, enum temp_type temp); 3376 unsigned int f2fs_usable_segs_in_sec(struct f2fs_sb_info *sbi, 3377 unsigned int segno); 3378 unsigned int f2fs_usable_blks_in_seg(struct f2fs_sb_info *sbi, 3379 unsigned int segno); 3380 3381 /* 3382 * checkpoint.c 3383 */ 3384 void f2fs_stop_checkpoint(struct f2fs_sb_info *sbi, bool end_io); 3385 struct page *f2fs_grab_meta_page(struct f2fs_sb_info *sbi, pgoff_t index); 3386 struct page *f2fs_get_meta_page(struct f2fs_sb_info *sbi, pgoff_t index); 3387 struct page *f2fs_get_meta_page_retry(struct f2fs_sb_info *sbi, pgoff_t index); 3388 struct page *f2fs_get_tmp_page(struct f2fs_sb_info *sbi, pgoff_t index); 3389 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi, 3390 block_t blkaddr, int type); 3391 int f2fs_ra_meta_pages(struct f2fs_sb_info *sbi, block_t start, int nrpages, 3392 int type, bool sync); 3393 void f2fs_ra_meta_pages_cond(struct f2fs_sb_info *sbi, pgoff_t index); 3394 long f2fs_sync_meta_pages(struct f2fs_sb_info *sbi, enum page_type type, 3395 long nr_to_write, enum iostat_type io_type); 3396 void f2fs_add_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type); 3397 void f2fs_remove_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type); 3398 void f2fs_release_ino_entry(struct f2fs_sb_info *sbi, bool all); 3399 bool f2fs_exist_written_data(struct f2fs_sb_info *sbi, nid_t ino, int mode); 3400 void f2fs_set_dirty_device(struct f2fs_sb_info *sbi, nid_t ino, 3401 unsigned int devidx, int type); 3402 bool f2fs_is_dirty_device(struct f2fs_sb_info *sbi, nid_t ino, 3403 unsigned int devidx, int type); 3404 int f2fs_sync_inode_meta(struct f2fs_sb_info *sbi); 3405 int f2fs_acquire_orphan_inode(struct f2fs_sb_info *sbi); 3406 void f2fs_release_orphan_inode(struct f2fs_sb_info *sbi); 3407 void f2fs_add_orphan_inode(struct inode *inode); 3408 void f2fs_remove_orphan_inode(struct f2fs_sb_info *sbi, nid_t ino); 3409 int f2fs_recover_orphan_inodes(struct f2fs_sb_info *sbi); 3410 int f2fs_get_valid_checkpoint(struct f2fs_sb_info *sbi); 3411 void f2fs_update_dirty_page(struct inode *inode, struct page *page); 3412 void f2fs_remove_dirty_inode(struct inode *inode); 3413 int f2fs_sync_dirty_inodes(struct f2fs_sb_info *sbi, enum inode_type type); 3414 void f2fs_wait_on_all_pages(struct f2fs_sb_info *sbi, int type); 3415 u64 f2fs_get_sectors_written(struct f2fs_sb_info *sbi); 3416 int f2fs_write_checkpoint(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3417 void f2fs_init_ino_entry_info(struct f2fs_sb_info *sbi); 3418 int __init f2fs_create_checkpoint_caches(void); 3419 void f2fs_destroy_checkpoint_caches(void); 3420 3421 /* 3422 * data.c 3423 */ 3424 int __init f2fs_init_bioset(void); 3425 void f2fs_destroy_bioset(void); 3426 int f2fs_init_bio_entry_cache(void); 3427 void f2fs_destroy_bio_entry_cache(void); 3428 void f2fs_submit_bio(struct f2fs_sb_info *sbi, 3429 struct bio *bio, enum page_type type); 3430 void f2fs_submit_merged_write(struct f2fs_sb_info *sbi, enum page_type type); 3431 void f2fs_submit_merged_write_cond(struct f2fs_sb_info *sbi, 3432 struct inode *inode, struct page *page, 3433 nid_t ino, enum page_type type); 3434 void f2fs_submit_merged_ipu_write(struct f2fs_sb_info *sbi, 3435 struct bio **bio, struct page *page); 3436 void f2fs_flush_merged_writes(struct f2fs_sb_info *sbi); 3437 int f2fs_submit_page_bio(struct f2fs_io_info *fio); 3438 int f2fs_merge_page_bio(struct f2fs_io_info *fio); 3439 void f2fs_submit_page_write(struct f2fs_io_info *fio); 3440 struct block_device *f2fs_target_device(struct f2fs_sb_info *sbi, 3441 block_t blk_addr, struct bio *bio); 3442 int f2fs_target_device_index(struct f2fs_sb_info *sbi, block_t blkaddr); 3443 void f2fs_set_data_blkaddr(struct dnode_of_data *dn); 3444 void f2fs_update_data_blkaddr(struct dnode_of_data *dn, block_t blkaddr); 3445 int f2fs_reserve_new_blocks(struct dnode_of_data *dn, blkcnt_t count); 3446 int f2fs_reserve_new_block(struct dnode_of_data *dn); 3447 int f2fs_get_block(struct dnode_of_data *dn, pgoff_t index); 3448 int f2fs_preallocate_blocks(struct kiocb *iocb, struct iov_iter *from); 3449 int f2fs_reserve_block(struct dnode_of_data *dn, pgoff_t index); 3450 struct page *f2fs_get_read_data_page(struct inode *inode, pgoff_t index, 3451 int op_flags, bool for_write); 3452 struct page *f2fs_find_data_page(struct inode *inode, pgoff_t index); 3453 struct page *f2fs_get_lock_data_page(struct inode *inode, pgoff_t index, 3454 bool for_write); 3455 struct page *f2fs_get_new_data_page(struct inode *inode, 3456 struct page *ipage, pgoff_t index, bool new_i_size); 3457 int f2fs_do_write_data_page(struct f2fs_io_info *fio); 3458 void f2fs_do_map_lock(struct f2fs_sb_info *sbi, int flag, bool lock); 3459 int f2fs_map_blocks(struct inode *inode, struct f2fs_map_blocks *map, 3460 int create, int flag); 3461 int f2fs_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, 3462 u64 start, u64 len); 3463 int f2fs_encrypt_one_page(struct f2fs_io_info *fio); 3464 bool f2fs_should_update_inplace(struct inode *inode, struct f2fs_io_info *fio); 3465 bool f2fs_should_update_outplace(struct inode *inode, struct f2fs_io_info *fio); 3466 int f2fs_write_single_data_page(struct page *page, int *submitted, 3467 struct bio **bio, sector_t *last_block, 3468 struct writeback_control *wbc, 3469 enum iostat_type io_type, 3470 int compr_blocks); 3471 void f2fs_invalidate_page(struct page *page, unsigned int offset, 3472 unsigned int length); 3473 int f2fs_release_page(struct page *page, gfp_t wait); 3474 #ifdef CONFIG_MIGRATION 3475 int f2fs_migrate_page(struct address_space *mapping, struct page *newpage, 3476 struct page *page, enum migrate_mode mode); 3477 #endif 3478 bool f2fs_overwrite_io(struct inode *inode, loff_t pos, size_t len); 3479 void f2fs_clear_page_cache_dirty_tag(struct page *page); 3480 int f2fs_init_post_read_processing(void); 3481 void f2fs_destroy_post_read_processing(void); 3482 int f2fs_init_post_read_wq(struct f2fs_sb_info *sbi); 3483 void f2fs_destroy_post_read_wq(struct f2fs_sb_info *sbi); 3484 3485 /* 3486 * gc.c 3487 */ 3488 int f2fs_start_gc_thread(struct f2fs_sb_info *sbi); 3489 void f2fs_stop_gc_thread(struct f2fs_sb_info *sbi); 3490 block_t f2fs_start_bidx_of_node(unsigned int node_ofs, struct inode *inode); 3491 int f2fs_gc(struct f2fs_sb_info *sbi, bool sync, bool background, 3492 unsigned int segno); 3493 void f2fs_build_gc_manager(struct f2fs_sb_info *sbi); 3494 int f2fs_resize_fs(struct f2fs_sb_info *sbi, __u64 block_count); 3495 int __init f2fs_create_garbage_collection_cache(void); 3496 void f2fs_destroy_garbage_collection_cache(void); 3497 3498 /* 3499 * recovery.c 3500 */ 3501 int f2fs_recover_fsync_data(struct f2fs_sb_info *sbi, bool check_only); 3502 bool f2fs_space_for_roll_forward(struct f2fs_sb_info *sbi); 3503 3504 /* 3505 * debug.c 3506 */ 3507 #ifdef CONFIG_F2FS_STAT_FS 3508 struct f2fs_stat_info { 3509 struct list_head stat_list; 3510 struct f2fs_sb_info *sbi; 3511 int all_area_segs, sit_area_segs, nat_area_segs, ssa_area_segs; 3512 int main_area_segs, main_area_sections, main_area_zones; 3513 unsigned long long hit_largest, hit_cached, hit_rbtree; 3514 unsigned long long hit_total, total_ext; 3515 int ext_tree, zombie_tree, ext_node; 3516 int ndirty_node, ndirty_dent, ndirty_meta, ndirty_imeta; 3517 int ndirty_data, ndirty_qdata; 3518 int inmem_pages; 3519 unsigned int ndirty_dirs, ndirty_files, nquota_files, ndirty_all; 3520 int nats, dirty_nats, sits, dirty_sits; 3521 int free_nids, avail_nids, alloc_nids; 3522 int total_count, utilization; 3523 int bg_gc, nr_wb_cp_data, nr_wb_data; 3524 int nr_rd_data, nr_rd_node, nr_rd_meta; 3525 int nr_dio_read, nr_dio_write; 3526 unsigned int io_skip_bggc, other_skip_bggc; 3527 int nr_flushing, nr_flushed, flush_list_empty; 3528 int nr_discarding, nr_discarded; 3529 int nr_discard_cmd; 3530 unsigned int undiscard_blks; 3531 int inline_xattr, inline_inode, inline_dir, append, update, orphans; 3532 int compr_inode; 3533 unsigned long long compr_blocks; 3534 int aw_cnt, max_aw_cnt, vw_cnt, max_vw_cnt; 3535 unsigned int valid_count, valid_node_count, valid_inode_count, discard_blks; 3536 unsigned int bimodal, avg_vblocks; 3537 int util_free, util_valid, util_invalid; 3538 int rsvd_segs, overp_segs; 3539 int dirty_count, node_pages, meta_pages; 3540 int prefree_count, call_count, cp_count, bg_cp_count; 3541 int tot_segs, node_segs, data_segs, free_segs, free_secs; 3542 int bg_node_segs, bg_data_segs; 3543 int tot_blks, data_blks, node_blks; 3544 int bg_data_blks, bg_node_blks; 3545 unsigned long long skipped_atomic_files[2]; 3546 int curseg[NR_CURSEG_TYPE]; 3547 int cursec[NR_CURSEG_TYPE]; 3548 int curzone[NR_CURSEG_TYPE]; 3549 unsigned int dirty_seg[NR_CURSEG_TYPE]; 3550 unsigned int full_seg[NR_CURSEG_TYPE]; 3551 unsigned int valid_blks[NR_CURSEG_TYPE]; 3552 3553 unsigned int meta_count[META_MAX]; 3554 unsigned int segment_count[2]; 3555 unsigned int block_count[2]; 3556 unsigned int inplace_count; 3557 unsigned long long base_mem, cache_mem, page_mem; 3558 }; 3559 3560 static inline struct f2fs_stat_info *F2FS_STAT(struct f2fs_sb_info *sbi) 3561 { 3562 return (struct f2fs_stat_info *)sbi->stat_info; 3563 } 3564 3565 #define stat_inc_cp_count(si) ((si)->cp_count++) 3566 #define stat_inc_bg_cp_count(si) ((si)->bg_cp_count++) 3567 #define stat_inc_call_count(si) ((si)->call_count++) 3568 #define stat_inc_bggc_count(si) ((si)->bg_gc++) 3569 #define stat_io_skip_bggc_count(sbi) ((sbi)->io_skip_bggc++) 3570 #define stat_other_skip_bggc_count(sbi) ((sbi)->other_skip_bggc++) 3571 #define stat_inc_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]++) 3572 #define stat_dec_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]--) 3573 #define stat_inc_total_hit(sbi) (atomic64_inc(&(sbi)->total_hit_ext)) 3574 #define stat_inc_rbtree_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_rbtree)) 3575 #define stat_inc_largest_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_largest)) 3576 #define stat_inc_cached_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_cached)) 3577 #define stat_inc_inline_xattr(inode) \ 3578 do { \ 3579 if (f2fs_has_inline_xattr(inode)) \ 3580 (atomic_inc(&F2FS_I_SB(inode)->inline_xattr)); \ 3581 } while (0) 3582 #define stat_dec_inline_xattr(inode) \ 3583 do { \ 3584 if (f2fs_has_inline_xattr(inode)) \ 3585 (atomic_dec(&F2FS_I_SB(inode)->inline_xattr)); \ 3586 } while (0) 3587 #define stat_inc_inline_inode(inode) \ 3588 do { \ 3589 if (f2fs_has_inline_data(inode)) \ 3590 (atomic_inc(&F2FS_I_SB(inode)->inline_inode)); \ 3591 } while (0) 3592 #define stat_dec_inline_inode(inode) \ 3593 do { \ 3594 if (f2fs_has_inline_data(inode)) \ 3595 (atomic_dec(&F2FS_I_SB(inode)->inline_inode)); \ 3596 } while (0) 3597 #define stat_inc_inline_dir(inode) \ 3598 do { \ 3599 if (f2fs_has_inline_dentry(inode)) \ 3600 (atomic_inc(&F2FS_I_SB(inode)->inline_dir)); \ 3601 } while (0) 3602 #define stat_dec_inline_dir(inode) \ 3603 do { \ 3604 if (f2fs_has_inline_dentry(inode)) \ 3605 (atomic_dec(&F2FS_I_SB(inode)->inline_dir)); \ 3606 } while (0) 3607 #define stat_inc_compr_inode(inode) \ 3608 do { \ 3609 if (f2fs_compressed_file(inode)) \ 3610 (atomic_inc(&F2FS_I_SB(inode)->compr_inode)); \ 3611 } while (0) 3612 #define stat_dec_compr_inode(inode) \ 3613 do { \ 3614 if (f2fs_compressed_file(inode)) \ 3615 (atomic_dec(&F2FS_I_SB(inode)->compr_inode)); \ 3616 } while (0) 3617 #define stat_add_compr_blocks(inode, blocks) \ 3618 (atomic64_add(blocks, &F2FS_I_SB(inode)->compr_blocks)) 3619 #define stat_sub_compr_blocks(inode, blocks) \ 3620 (atomic64_sub(blocks, &F2FS_I_SB(inode)->compr_blocks)) 3621 #define stat_inc_meta_count(sbi, blkaddr) \ 3622 do { \ 3623 if (blkaddr < SIT_I(sbi)->sit_base_addr) \ 3624 atomic_inc(&(sbi)->meta_count[META_CP]); \ 3625 else if (blkaddr < NM_I(sbi)->nat_blkaddr) \ 3626 atomic_inc(&(sbi)->meta_count[META_SIT]); \ 3627 else if (blkaddr < SM_I(sbi)->ssa_blkaddr) \ 3628 atomic_inc(&(sbi)->meta_count[META_NAT]); \ 3629 else if (blkaddr < SM_I(sbi)->main_blkaddr) \ 3630 atomic_inc(&(sbi)->meta_count[META_SSA]); \ 3631 } while (0) 3632 #define stat_inc_seg_type(sbi, curseg) \ 3633 ((sbi)->segment_count[(curseg)->alloc_type]++) 3634 #define stat_inc_block_count(sbi, curseg) \ 3635 ((sbi)->block_count[(curseg)->alloc_type]++) 3636 #define stat_inc_inplace_blocks(sbi) \ 3637 (atomic_inc(&(sbi)->inplace_count)) 3638 #define stat_update_max_atomic_write(inode) \ 3639 do { \ 3640 int cur = F2FS_I_SB(inode)->atomic_files; \ 3641 int max = atomic_read(&F2FS_I_SB(inode)->max_aw_cnt); \ 3642 if (cur > max) \ 3643 atomic_set(&F2FS_I_SB(inode)->max_aw_cnt, cur); \ 3644 } while (0) 3645 #define stat_inc_volatile_write(inode) \ 3646 (atomic_inc(&F2FS_I_SB(inode)->vw_cnt)) 3647 #define stat_dec_volatile_write(inode) \ 3648 (atomic_dec(&F2FS_I_SB(inode)->vw_cnt)) 3649 #define stat_update_max_volatile_write(inode) \ 3650 do { \ 3651 int cur = atomic_read(&F2FS_I_SB(inode)->vw_cnt); \ 3652 int max = atomic_read(&F2FS_I_SB(inode)->max_vw_cnt); \ 3653 if (cur > max) \ 3654 atomic_set(&F2FS_I_SB(inode)->max_vw_cnt, cur); \ 3655 } while (0) 3656 #define stat_inc_seg_count(sbi, type, gc_type) \ 3657 do { \ 3658 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 3659 si->tot_segs++; \ 3660 if ((type) == SUM_TYPE_DATA) { \ 3661 si->data_segs++; \ 3662 si->bg_data_segs += (gc_type == BG_GC) ? 1 : 0; \ 3663 } else { \ 3664 si->node_segs++; \ 3665 si->bg_node_segs += (gc_type == BG_GC) ? 1 : 0; \ 3666 } \ 3667 } while (0) 3668 3669 #define stat_inc_tot_blk_count(si, blks) \ 3670 ((si)->tot_blks += (blks)) 3671 3672 #define stat_inc_data_blk_count(sbi, blks, gc_type) \ 3673 do { \ 3674 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 3675 stat_inc_tot_blk_count(si, blks); \ 3676 si->data_blks += (blks); \ 3677 si->bg_data_blks += ((gc_type) == BG_GC) ? (blks) : 0; \ 3678 } while (0) 3679 3680 #define stat_inc_node_blk_count(sbi, blks, gc_type) \ 3681 do { \ 3682 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 3683 stat_inc_tot_blk_count(si, blks); \ 3684 si->node_blks += (blks); \ 3685 si->bg_node_blks += ((gc_type) == BG_GC) ? (blks) : 0; \ 3686 } while (0) 3687 3688 int f2fs_build_stats(struct f2fs_sb_info *sbi); 3689 void f2fs_destroy_stats(struct f2fs_sb_info *sbi); 3690 void __init f2fs_create_root_stats(void); 3691 void f2fs_destroy_root_stats(void); 3692 void f2fs_update_sit_info(struct f2fs_sb_info *sbi); 3693 #else 3694 #define stat_inc_cp_count(si) do { } while (0) 3695 #define stat_inc_bg_cp_count(si) do { } while (0) 3696 #define stat_inc_call_count(si) do { } while (0) 3697 #define stat_inc_bggc_count(si) do { } while (0) 3698 #define stat_io_skip_bggc_count(sbi) do { } while (0) 3699 #define stat_other_skip_bggc_count(sbi) do { } while (0) 3700 #define stat_inc_dirty_inode(sbi, type) do { } while (0) 3701 #define stat_dec_dirty_inode(sbi, type) do { } while (0) 3702 #define stat_inc_total_hit(sbi) do { } while (0) 3703 #define stat_inc_rbtree_node_hit(sbi) do { } while (0) 3704 #define stat_inc_largest_node_hit(sbi) do { } while (0) 3705 #define stat_inc_cached_node_hit(sbi) do { } while (0) 3706 #define stat_inc_inline_xattr(inode) do { } while (0) 3707 #define stat_dec_inline_xattr(inode) do { } while (0) 3708 #define stat_inc_inline_inode(inode) do { } while (0) 3709 #define stat_dec_inline_inode(inode) do { } while (0) 3710 #define stat_inc_inline_dir(inode) do { } while (0) 3711 #define stat_dec_inline_dir(inode) do { } while (0) 3712 #define stat_inc_compr_inode(inode) do { } while (0) 3713 #define stat_dec_compr_inode(inode) do { } while (0) 3714 #define stat_add_compr_blocks(inode, blocks) do { } while (0) 3715 #define stat_sub_compr_blocks(inode, blocks) do { } while (0) 3716 #define stat_inc_atomic_write(inode) do { } while (0) 3717 #define stat_dec_atomic_write(inode) do { } while (0) 3718 #define stat_update_max_atomic_write(inode) do { } while (0) 3719 #define stat_inc_volatile_write(inode) do { } while (0) 3720 #define stat_dec_volatile_write(inode) do { } while (0) 3721 #define stat_update_max_volatile_write(inode) do { } while (0) 3722 #define stat_inc_meta_count(sbi, blkaddr) do { } while (0) 3723 #define stat_inc_seg_type(sbi, curseg) do { } while (0) 3724 #define stat_inc_block_count(sbi, curseg) do { } while (0) 3725 #define stat_inc_inplace_blocks(sbi) do { } while (0) 3726 #define stat_inc_seg_count(sbi, type, gc_type) do { } while (0) 3727 #define stat_inc_tot_blk_count(si, blks) do { } while (0) 3728 #define stat_inc_data_blk_count(sbi, blks, gc_type) do { } while (0) 3729 #define stat_inc_node_blk_count(sbi, blks, gc_type) do { } while (0) 3730 3731 static inline int f2fs_build_stats(struct f2fs_sb_info *sbi) { return 0; } 3732 static inline void f2fs_destroy_stats(struct f2fs_sb_info *sbi) { } 3733 static inline void __init f2fs_create_root_stats(void) { } 3734 static inline void f2fs_destroy_root_stats(void) { } 3735 static inline void f2fs_update_sit_info(struct f2fs_sb_info *sbi) {} 3736 #endif 3737 3738 extern const struct file_operations f2fs_dir_operations; 3739 extern const struct file_operations f2fs_file_operations; 3740 extern const struct inode_operations f2fs_file_inode_operations; 3741 extern const struct address_space_operations f2fs_dblock_aops; 3742 extern const struct address_space_operations f2fs_node_aops; 3743 extern const struct address_space_operations f2fs_meta_aops; 3744 extern const struct inode_operations f2fs_dir_inode_operations; 3745 extern const struct inode_operations f2fs_symlink_inode_operations; 3746 extern const struct inode_operations f2fs_encrypted_symlink_inode_operations; 3747 extern const struct inode_operations f2fs_special_inode_operations; 3748 extern struct kmem_cache *f2fs_inode_entry_slab; 3749 3750 /* 3751 * inline.c 3752 */ 3753 bool f2fs_may_inline_data(struct inode *inode); 3754 bool f2fs_may_inline_dentry(struct inode *inode); 3755 void f2fs_do_read_inline_data(struct page *page, struct page *ipage); 3756 void f2fs_truncate_inline_inode(struct inode *inode, 3757 struct page *ipage, u64 from); 3758 int f2fs_read_inline_data(struct inode *inode, struct page *page); 3759 int f2fs_convert_inline_page(struct dnode_of_data *dn, struct page *page); 3760 int f2fs_convert_inline_inode(struct inode *inode); 3761 int f2fs_try_convert_inline_dir(struct inode *dir, struct dentry *dentry); 3762 int f2fs_write_inline_data(struct inode *inode, struct page *page); 3763 int f2fs_recover_inline_data(struct inode *inode, struct page *npage); 3764 struct f2fs_dir_entry *f2fs_find_in_inline_dir(struct inode *dir, 3765 const struct f2fs_filename *fname, 3766 struct page **res_page); 3767 int f2fs_make_empty_inline_dir(struct inode *inode, struct inode *parent, 3768 struct page *ipage); 3769 int f2fs_add_inline_entry(struct inode *dir, const struct f2fs_filename *fname, 3770 struct inode *inode, nid_t ino, umode_t mode); 3771 void f2fs_delete_inline_entry(struct f2fs_dir_entry *dentry, 3772 struct page *page, struct inode *dir, 3773 struct inode *inode); 3774 bool f2fs_empty_inline_dir(struct inode *dir); 3775 int f2fs_read_inline_dir(struct file *file, struct dir_context *ctx, 3776 struct fscrypt_str *fstr); 3777 int f2fs_inline_data_fiemap(struct inode *inode, 3778 struct fiemap_extent_info *fieinfo, 3779 __u64 start, __u64 len); 3780 3781 /* 3782 * shrinker.c 3783 */ 3784 unsigned long f2fs_shrink_count(struct shrinker *shrink, 3785 struct shrink_control *sc); 3786 unsigned long f2fs_shrink_scan(struct shrinker *shrink, 3787 struct shrink_control *sc); 3788 void f2fs_join_shrinker(struct f2fs_sb_info *sbi); 3789 void f2fs_leave_shrinker(struct f2fs_sb_info *sbi); 3790 3791 /* 3792 * extent_cache.c 3793 */ 3794 struct rb_entry *f2fs_lookup_rb_tree(struct rb_root_cached *root, 3795 struct rb_entry *cached_re, unsigned int ofs); 3796 struct rb_node **f2fs_lookup_rb_tree_ext(struct f2fs_sb_info *sbi, 3797 struct rb_root_cached *root, 3798 struct rb_node **parent, 3799 unsigned long long key, bool *left_most); 3800 struct rb_node **f2fs_lookup_rb_tree_for_insert(struct f2fs_sb_info *sbi, 3801 struct rb_root_cached *root, 3802 struct rb_node **parent, 3803 unsigned int ofs, bool *leftmost); 3804 struct rb_entry *f2fs_lookup_rb_tree_ret(struct rb_root_cached *root, 3805 struct rb_entry *cached_re, unsigned int ofs, 3806 struct rb_entry **prev_entry, struct rb_entry **next_entry, 3807 struct rb_node ***insert_p, struct rb_node **insert_parent, 3808 bool force, bool *leftmost); 3809 bool f2fs_check_rb_tree_consistence(struct f2fs_sb_info *sbi, 3810 struct rb_root_cached *root, bool check_key); 3811 unsigned int f2fs_shrink_extent_tree(struct f2fs_sb_info *sbi, int nr_shrink); 3812 void f2fs_init_extent_tree(struct inode *inode, struct page *ipage); 3813 void f2fs_drop_extent_tree(struct inode *inode); 3814 unsigned int f2fs_destroy_extent_node(struct inode *inode); 3815 void f2fs_destroy_extent_tree(struct inode *inode); 3816 bool f2fs_lookup_extent_cache(struct inode *inode, pgoff_t pgofs, 3817 struct extent_info *ei); 3818 void f2fs_update_extent_cache(struct dnode_of_data *dn); 3819 void f2fs_update_extent_cache_range(struct dnode_of_data *dn, 3820 pgoff_t fofs, block_t blkaddr, unsigned int len); 3821 void f2fs_init_extent_cache_info(struct f2fs_sb_info *sbi); 3822 int __init f2fs_create_extent_cache(void); 3823 void f2fs_destroy_extent_cache(void); 3824 3825 /* 3826 * sysfs.c 3827 */ 3828 int __init f2fs_init_sysfs(void); 3829 void f2fs_exit_sysfs(void); 3830 int f2fs_register_sysfs(struct f2fs_sb_info *sbi); 3831 void f2fs_unregister_sysfs(struct f2fs_sb_info *sbi); 3832 3833 /* verity.c */ 3834 extern const struct fsverity_operations f2fs_verityops; 3835 3836 /* 3837 * crypto support 3838 */ 3839 static inline bool f2fs_encrypted_file(struct inode *inode) 3840 { 3841 return IS_ENCRYPTED(inode) && S_ISREG(inode->i_mode); 3842 } 3843 3844 static inline void f2fs_set_encrypted_inode(struct inode *inode) 3845 { 3846 #ifdef CONFIG_FS_ENCRYPTION 3847 file_set_encrypt(inode); 3848 f2fs_set_inode_flags(inode); 3849 #endif 3850 } 3851 3852 /* 3853 * Returns true if the reads of the inode's data need to undergo some 3854 * postprocessing step, like decryption or authenticity verification. 3855 */ 3856 static inline bool f2fs_post_read_required(struct inode *inode) 3857 { 3858 return f2fs_encrypted_file(inode) || fsverity_active(inode) || 3859 f2fs_compressed_file(inode); 3860 } 3861 3862 /* 3863 * compress.c 3864 */ 3865 #ifdef CONFIG_F2FS_FS_COMPRESSION 3866 bool f2fs_is_compressed_page(struct page *page); 3867 struct page *f2fs_compress_control_page(struct page *page); 3868 int f2fs_prepare_compress_overwrite(struct inode *inode, 3869 struct page **pagep, pgoff_t index, void **fsdata); 3870 bool f2fs_compress_write_end(struct inode *inode, void *fsdata, 3871 pgoff_t index, unsigned copied); 3872 int f2fs_truncate_partial_cluster(struct inode *inode, u64 from, bool lock); 3873 void f2fs_compress_write_end_io(struct bio *bio, struct page *page); 3874 bool f2fs_is_compress_backend_ready(struct inode *inode); 3875 int f2fs_init_compress_mempool(void); 3876 void f2fs_destroy_compress_mempool(void); 3877 void f2fs_decompress_pages(struct bio *bio, struct page *page, bool verity); 3878 bool f2fs_cluster_is_empty(struct compress_ctx *cc); 3879 bool f2fs_cluster_can_merge_page(struct compress_ctx *cc, pgoff_t index); 3880 void f2fs_compress_ctx_add_page(struct compress_ctx *cc, struct page *page); 3881 int f2fs_write_multi_pages(struct compress_ctx *cc, 3882 int *submitted, 3883 struct writeback_control *wbc, 3884 enum iostat_type io_type); 3885 int f2fs_is_compressed_cluster(struct inode *inode, pgoff_t index); 3886 int f2fs_read_multi_pages(struct compress_ctx *cc, struct bio **bio_ret, 3887 unsigned nr_pages, sector_t *last_block_in_bio, 3888 bool is_readahead, bool for_write); 3889 struct decompress_io_ctx *f2fs_alloc_dic(struct compress_ctx *cc); 3890 void f2fs_free_dic(struct decompress_io_ctx *dic); 3891 void f2fs_decompress_end_io(struct page **rpages, 3892 unsigned int cluster_size, bool err, bool verity); 3893 int f2fs_init_compress_ctx(struct compress_ctx *cc); 3894 void f2fs_destroy_compress_ctx(struct compress_ctx *cc); 3895 void f2fs_init_compress_info(struct f2fs_sb_info *sbi); 3896 int f2fs_init_page_array_cache(struct f2fs_sb_info *sbi); 3897 void f2fs_destroy_page_array_cache(struct f2fs_sb_info *sbi); 3898 int __init f2fs_init_compress_cache(void); 3899 void f2fs_destroy_compress_cache(void); 3900 #else 3901 static inline bool f2fs_is_compressed_page(struct page *page) { return false; } 3902 static inline bool f2fs_is_compress_backend_ready(struct inode *inode) 3903 { 3904 if (!f2fs_compressed_file(inode)) 3905 return true; 3906 /* not support compression */ 3907 return false; 3908 } 3909 static inline struct page *f2fs_compress_control_page(struct page *page) 3910 { 3911 WARN_ON_ONCE(1); 3912 return ERR_PTR(-EINVAL); 3913 } 3914 static inline int f2fs_init_compress_mempool(void) { return 0; } 3915 static inline void f2fs_destroy_compress_mempool(void) { } 3916 static inline int f2fs_init_page_array_cache(struct f2fs_sb_info *sbi) { return 0; } 3917 static inline void f2fs_destroy_page_array_cache(struct f2fs_sb_info *sbi) { } 3918 static inline int __init f2fs_init_compress_cache(void) { return 0; } 3919 static inline void f2fs_destroy_compress_cache(void) { } 3920 #endif 3921 3922 static inline void set_compress_context(struct inode *inode) 3923 { 3924 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 3925 3926 F2FS_I(inode)->i_compress_algorithm = 3927 F2FS_OPTION(sbi).compress_algorithm; 3928 F2FS_I(inode)->i_log_cluster_size = 3929 F2FS_OPTION(sbi).compress_log_size; 3930 F2FS_I(inode)->i_compress_flag = 3931 F2FS_OPTION(sbi).compress_chksum ? 3932 1 << COMPRESS_CHKSUM : 0; 3933 F2FS_I(inode)->i_cluster_size = 3934 1 << F2FS_I(inode)->i_log_cluster_size; 3935 F2FS_I(inode)->i_flags |= F2FS_COMPR_FL; 3936 set_inode_flag(inode, FI_COMPRESSED_FILE); 3937 stat_inc_compr_inode(inode); 3938 f2fs_mark_inode_dirty_sync(inode, true); 3939 } 3940 3941 static inline bool f2fs_disable_compressed_file(struct inode *inode) 3942 { 3943 struct f2fs_inode_info *fi = F2FS_I(inode); 3944 3945 if (!f2fs_compressed_file(inode)) 3946 return true; 3947 if (S_ISREG(inode->i_mode) && 3948 (get_dirty_pages(inode) || atomic_read(&fi->i_compr_blocks))) 3949 return false; 3950 3951 fi->i_flags &= ~F2FS_COMPR_FL; 3952 stat_dec_compr_inode(inode); 3953 clear_inode_flag(inode, FI_COMPRESSED_FILE); 3954 f2fs_mark_inode_dirty_sync(inode, true); 3955 return true; 3956 } 3957 3958 #define F2FS_FEATURE_FUNCS(name, flagname) \ 3959 static inline int f2fs_sb_has_##name(struct f2fs_sb_info *sbi) \ 3960 { \ 3961 return F2FS_HAS_FEATURE(sbi, F2FS_FEATURE_##flagname); \ 3962 } 3963 3964 F2FS_FEATURE_FUNCS(encrypt, ENCRYPT); 3965 F2FS_FEATURE_FUNCS(blkzoned, BLKZONED); 3966 F2FS_FEATURE_FUNCS(extra_attr, EXTRA_ATTR); 3967 F2FS_FEATURE_FUNCS(project_quota, PRJQUOTA); 3968 F2FS_FEATURE_FUNCS(inode_chksum, INODE_CHKSUM); 3969 F2FS_FEATURE_FUNCS(flexible_inline_xattr, FLEXIBLE_INLINE_XATTR); 3970 F2FS_FEATURE_FUNCS(quota_ino, QUOTA_INO); 3971 F2FS_FEATURE_FUNCS(inode_crtime, INODE_CRTIME); 3972 F2FS_FEATURE_FUNCS(lost_found, LOST_FOUND); 3973 F2FS_FEATURE_FUNCS(verity, VERITY); 3974 F2FS_FEATURE_FUNCS(sb_chksum, SB_CHKSUM); 3975 F2FS_FEATURE_FUNCS(casefold, CASEFOLD); 3976 F2FS_FEATURE_FUNCS(compression, COMPRESSION); 3977 3978 #ifdef CONFIG_BLK_DEV_ZONED 3979 static inline bool f2fs_blkz_is_seq(struct f2fs_sb_info *sbi, int devi, 3980 block_t blkaddr) 3981 { 3982 unsigned int zno = blkaddr >> sbi->log_blocks_per_blkz; 3983 3984 return test_bit(zno, FDEV(devi).blkz_seq); 3985 } 3986 #endif 3987 3988 static inline bool f2fs_hw_should_discard(struct f2fs_sb_info *sbi) 3989 { 3990 return f2fs_sb_has_blkzoned(sbi); 3991 } 3992 3993 static inline bool f2fs_bdev_support_discard(struct block_device *bdev) 3994 { 3995 return blk_queue_discard(bdev_get_queue(bdev)) || 3996 bdev_is_zoned(bdev); 3997 } 3998 3999 static inline bool f2fs_hw_support_discard(struct f2fs_sb_info *sbi) 4000 { 4001 int i; 4002 4003 if (!f2fs_is_multi_device(sbi)) 4004 return f2fs_bdev_support_discard(sbi->sb->s_bdev); 4005 4006 for (i = 0; i < sbi->s_ndevs; i++) 4007 if (f2fs_bdev_support_discard(FDEV(i).bdev)) 4008 return true; 4009 return false; 4010 } 4011 4012 static inline bool f2fs_realtime_discard_enable(struct f2fs_sb_info *sbi) 4013 { 4014 return (test_opt(sbi, DISCARD) && f2fs_hw_support_discard(sbi)) || 4015 f2fs_hw_should_discard(sbi); 4016 } 4017 4018 static inline bool f2fs_hw_is_readonly(struct f2fs_sb_info *sbi) 4019 { 4020 int i; 4021 4022 if (!f2fs_is_multi_device(sbi)) 4023 return bdev_read_only(sbi->sb->s_bdev); 4024 4025 for (i = 0; i < sbi->s_ndevs; i++) 4026 if (bdev_read_only(FDEV(i).bdev)) 4027 return true; 4028 return false; 4029 } 4030 4031 static inline bool f2fs_lfs_mode(struct f2fs_sb_info *sbi) 4032 { 4033 return F2FS_OPTION(sbi).fs_mode == FS_MODE_LFS; 4034 } 4035 4036 static inline bool f2fs_may_compress(struct inode *inode) 4037 { 4038 if (IS_SWAPFILE(inode) || f2fs_is_pinned_file(inode) || 4039 f2fs_is_atomic_file(inode) || 4040 f2fs_is_volatile_file(inode)) 4041 return false; 4042 return S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode); 4043 } 4044 4045 static inline void f2fs_i_compr_blocks_update(struct inode *inode, 4046 u64 blocks, bool add) 4047 { 4048 int diff = F2FS_I(inode)->i_cluster_size - blocks; 4049 struct f2fs_inode_info *fi = F2FS_I(inode); 4050 4051 /* don't update i_compr_blocks if saved blocks were released */ 4052 if (!add && !atomic_read(&fi->i_compr_blocks)) 4053 return; 4054 4055 if (add) { 4056 atomic_add(diff, &fi->i_compr_blocks); 4057 stat_add_compr_blocks(inode, diff); 4058 } else { 4059 atomic_sub(diff, &fi->i_compr_blocks); 4060 stat_sub_compr_blocks(inode, diff); 4061 } 4062 f2fs_mark_inode_dirty_sync(inode, true); 4063 } 4064 4065 static inline int block_unaligned_IO(struct inode *inode, 4066 struct kiocb *iocb, struct iov_iter *iter) 4067 { 4068 unsigned int i_blkbits = READ_ONCE(inode->i_blkbits); 4069 unsigned int blocksize_mask = (1 << i_blkbits) - 1; 4070 loff_t offset = iocb->ki_pos; 4071 unsigned long align = offset | iov_iter_alignment(iter); 4072 4073 return align & blocksize_mask; 4074 } 4075 4076 static inline int allow_outplace_dio(struct inode *inode, 4077 struct kiocb *iocb, struct iov_iter *iter) 4078 { 4079 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 4080 int rw = iov_iter_rw(iter); 4081 4082 return (f2fs_lfs_mode(sbi) && (rw == WRITE) && 4083 !block_unaligned_IO(inode, iocb, iter)); 4084 } 4085 4086 static inline bool f2fs_force_buffered_io(struct inode *inode, 4087 struct kiocb *iocb, struct iov_iter *iter) 4088 { 4089 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 4090 int rw = iov_iter_rw(iter); 4091 4092 if (f2fs_post_read_required(inode)) 4093 return true; 4094 if (f2fs_is_multi_device(sbi)) 4095 return true; 4096 /* 4097 * for blkzoned device, fallback direct IO to buffered IO, so 4098 * all IOs can be serialized by log-structured write. 4099 */ 4100 if (f2fs_sb_has_blkzoned(sbi)) 4101 return true; 4102 if (f2fs_lfs_mode(sbi) && (rw == WRITE)) { 4103 if (block_unaligned_IO(inode, iocb, iter)) 4104 return true; 4105 if (F2FS_IO_ALIGNED(sbi)) 4106 return true; 4107 } 4108 if (is_sbi_flag_set(F2FS_I_SB(inode), SBI_CP_DISABLED) && 4109 !IS_SWAPFILE(inode)) 4110 return true; 4111 4112 return false; 4113 } 4114 4115 #ifdef CONFIG_F2FS_FAULT_INJECTION 4116 extern void f2fs_build_fault_attr(struct f2fs_sb_info *sbi, unsigned int rate, 4117 unsigned int type); 4118 #else 4119 #define f2fs_build_fault_attr(sbi, rate, type) do { } while (0) 4120 #endif 4121 4122 static inline bool is_journalled_quota(struct f2fs_sb_info *sbi) 4123 { 4124 #ifdef CONFIG_QUOTA 4125 if (f2fs_sb_has_quota_ino(sbi)) 4126 return true; 4127 if (F2FS_OPTION(sbi).s_qf_names[USRQUOTA] || 4128 F2FS_OPTION(sbi).s_qf_names[GRPQUOTA] || 4129 F2FS_OPTION(sbi).s_qf_names[PRJQUOTA]) 4130 return true; 4131 #endif 4132 return false; 4133 } 4134 4135 #define EFSBADCRC EBADMSG /* Bad CRC detected */ 4136 #define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */ 4137 4138 #endif /* _LINUX_F2FS_H */ 4139