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