1 /* SPDX-License-Identifier: GPL-2.0 */ 2 /* 3 * fs/f2fs/f2fs.h 4 * 5 * Copyright (c) 2012 Samsung Electronics Co., Ltd. 6 * http://www.samsung.com/ 7 */ 8 #ifndef _LINUX_F2FS_H 9 #define _LINUX_F2FS_H 10 11 #include <linux/uio.h> 12 #include <linux/types.h> 13 #include <linux/page-flags.h> 14 #include <linux/buffer_head.h> 15 #include <linux/slab.h> 16 #include <linux/crc32.h> 17 #include <linux/magic.h> 18 #include <linux/kobject.h> 19 #include <linux/sched.h> 20 #include <linux/cred.h> 21 #include <linux/vmalloc.h> 22 #include <linux/bio.h> 23 #include <linux/blkdev.h> 24 #include <linux/quotaops.h> 25 #include <linux/part_stat.h> 26 #include <crypto/hash.h> 27 28 #include <linux/fscrypt.h> 29 #include <linux/fsverity.h> 30 31 #ifdef CONFIG_F2FS_CHECK_FS 32 #define f2fs_bug_on(sbi, condition) BUG_ON(condition) 33 #else 34 #define f2fs_bug_on(sbi, condition) \ 35 do { \ 36 if (WARN_ON(condition)) \ 37 set_sbi_flag(sbi, SBI_NEED_FSCK); \ 38 } while (0) 39 #endif 40 41 enum { 42 FAULT_KMALLOC, 43 FAULT_KVMALLOC, 44 FAULT_PAGE_ALLOC, 45 FAULT_PAGE_GET, 46 FAULT_ALLOC_NID, 47 FAULT_ORPHAN, 48 FAULT_BLOCK, 49 FAULT_DIR_DEPTH, 50 FAULT_EVICT_INODE, 51 FAULT_TRUNCATE, 52 FAULT_READ_IO, 53 FAULT_CHECKPOINT, 54 FAULT_DISCARD, 55 FAULT_WRITE_IO, 56 FAULT_MAX, 57 }; 58 59 #ifdef CONFIG_F2FS_FAULT_INJECTION 60 #define F2FS_ALL_FAULT_TYPE ((1 << FAULT_MAX) - 1) 61 62 struct f2fs_fault_info { 63 atomic_t inject_ops; 64 unsigned int inject_rate; 65 unsigned int inject_type; 66 }; 67 68 extern const char *f2fs_fault_name[FAULT_MAX]; 69 #define IS_FAULT_SET(fi, type) ((fi)->inject_type & (1 << (type))) 70 #endif 71 72 /* 73 * For mount options 74 */ 75 #define F2FS_MOUNT_DISABLE_ROLL_FORWARD 0x00000002 76 #define F2FS_MOUNT_DISCARD 0x00000004 77 #define F2FS_MOUNT_NOHEAP 0x00000008 78 #define F2FS_MOUNT_XATTR_USER 0x00000010 79 #define F2FS_MOUNT_POSIX_ACL 0x00000020 80 #define F2FS_MOUNT_DISABLE_EXT_IDENTIFY 0x00000040 81 #define F2FS_MOUNT_INLINE_XATTR 0x00000080 82 #define F2FS_MOUNT_INLINE_DATA 0x00000100 83 #define F2FS_MOUNT_INLINE_DENTRY 0x00000200 84 #define F2FS_MOUNT_FLUSH_MERGE 0x00000400 85 #define F2FS_MOUNT_NOBARRIER 0x00000800 86 #define F2FS_MOUNT_FASTBOOT 0x00001000 87 #define F2FS_MOUNT_EXTENT_CACHE 0x00002000 88 #define F2FS_MOUNT_DATA_FLUSH 0x00008000 89 #define F2FS_MOUNT_FAULT_INJECTION 0x00010000 90 #define F2FS_MOUNT_USRQUOTA 0x00080000 91 #define F2FS_MOUNT_GRPQUOTA 0x00100000 92 #define F2FS_MOUNT_PRJQUOTA 0x00200000 93 #define F2FS_MOUNT_QUOTA 0x00400000 94 #define F2FS_MOUNT_INLINE_XATTR_SIZE 0x00800000 95 #define F2FS_MOUNT_RESERVE_ROOT 0x01000000 96 #define F2FS_MOUNT_DISABLE_CHECKPOINT 0x02000000 97 #define F2FS_MOUNT_NORECOVERY 0x04000000 98 #define F2FS_MOUNT_ATGC 0x08000000 99 #define F2FS_MOUNT_MERGE_CHECKPOINT 0x10000000 100 #define F2FS_MOUNT_GC_MERGE 0x20000000 101 102 #define F2FS_OPTION(sbi) ((sbi)->mount_opt) 103 #define clear_opt(sbi, option) (F2FS_OPTION(sbi).opt &= ~F2FS_MOUNT_##option) 104 #define set_opt(sbi, option) (F2FS_OPTION(sbi).opt |= F2FS_MOUNT_##option) 105 #define test_opt(sbi, option) (F2FS_OPTION(sbi).opt & F2FS_MOUNT_##option) 106 107 #define ver_after(a, b) (typecheck(unsigned long long, a) && \ 108 typecheck(unsigned long long, b) && \ 109 ((long long)((a) - (b)) > 0)) 110 111 typedef u32 block_t; /* 112 * should not change u32, since it is the on-disk block 113 * address format, __le32. 114 */ 115 typedef u32 nid_t; 116 117 #define COMPRESS_EXT_NUM 16 118 119 struct f2fs_mount_info { 120 unsigned int opt; 121 int write_io_size_bits; /* Write IO size bits */ 122 block_t root_reserved_blocks; /* root reserved blocks */ 123 kuid_t s_resuid; /* reserved blocks for uid */ 124 kgid_t s_resgid; /* reserved blocks for gid */ 125 int active_logs; /* # of active logs */ 126 int inline_xattr_size; /* inline xattr size */ 127 #ifdef CONFIG_F2FS_FAULT_INJECTION 128 struct f2fs_fault_info fault_info; /* For fault injection */ 129 #endif 130 #ifdef CONFIG_QUOTA 131 /* Names of quota files with journalled quota */ 132 char *s_qf_names[MAXQUOTAS]; 133 int s_jquota_fmt; /* Format of quota to use */ 134 #endif 135 /* For which write hints are passed down to block layer */ 136 int whint_mode; 137 int alloc_mode; /* segment allocation policy */ 138 int fsync_mode; /* fsync policy */ 139 int fs_mode; /* fs mode: LFS or ADAPTIVE */ 140 int bggc_mode; /* bggc mode: off, on or sync */ 141 struct fscrypt_dummy_policy dummy_enc_policy; /* test dummy encryption */ 142 block_t unusable_cap_perc; /* percentage for cap */ 143 block_t unusable_cap; /* Amount of space allowed to be 144 * unusable when disabling checkpoint 145 */ 146 147 /* For compression */ 148 unsigned char compress_algorithm; /* algorithm type */ 149 unsigned char compress_log_size; /* cluster log size */ 150 unsigned char compress_level; /* compress level */ 151 bool compress_chksum; /* compressed data chksum */ 152 unsigned char compress_ext_cnt; /* extension count */ 153 int compress_mode; /* compression mode */ 154 unsigned char extensions[COMPRESS_EXT_NUM][F2FS_EXTENSION_LEN]; /* extensions */ 155 }; 156 157 #define F2FS_FEATURE_ENCRYPT 0x0001 158 #define F2FS_FEATURE_BLKZONED 0x0002 159 #define F2FS_FEATURE_ATOMIC_WRITE 0x0004 160 #define F2FS_FEATURE_EXTRA_ATTR 0x0008 161 #define F2FS_FEATURE_PRJQUOTA 0x0010 162 #define F2FS_FEATURE_INODE_CHKSUM 0x0020 163 #define F2FS_FEATURE_FLEXIBLE_INLINE_XATTR 0x0040 164 #define F2FS_FEATURE_QUOTA_INO 0x0080 165 #define F2FS_FEATURE_INODE_CRTIME 0x0100 166 #define F2FS_FEATURE_LOST_FOUND 0x0200 167 #define F2FS_FEATURE_VERITY 0x0400 168 #define F2FS_FEATURE_SB_CHKSUM 0x0800 169 #define F2FS_FEATURE_CASEFOLD 0x1000 170 #define F2FS_FEATURE_COMPRESSION 0x2000 171 #define F2FS_FEATURE_RO 0x4000 172 173 #define __F2FS_HAS_FEATURE(raw_super, mask) \ 174 ((raw_super->feature & cpu_to_le32(mask)) != 0) 175 #define F2FS_HAS_FEATURE(sbi, mask) __F2FS_HAS_FEATURE(sbi->raw_super, mask) 176 #define F2FS_SET_FEATURE(sbi, mask) \ 177 (sbi->raw_super->feature |= cpu_to_le32(mask)) 178 #define F2FS_CLEAR_FEATURE(sbi, mask) \ 179 (sbi->raw_super->feature &= ~cpu_to_le32(mask)) 180 181 /* 182 * Default values for user and/or group using reserved blocks 183 */ 184 #define F2FS_DEF_RESUID 0 185 #define F2FS_DEF_RESGID 0 186 187 /* 188 * For checkpoint manager 189 */ 190 enum { 191 NAT_BITMAP, 192 SIT_BITMAP 193 }; 194 195 #define CP_UMOUNT 0x00000001 196 #define CP_FASTBOOT 0x00000002 197 #define CP_SYNC 0x00000004 198 #define CP_RECOVERY 0x00000008 199 #define CP_DISCARD 0x00000010 200 #define CP_TRIMMED 0x00000020 201 #define CP_PAUSE 0x00000040 202 #define CP_RESIZE 0x00000080 203 204 #define MAX_DISCARD_BLOCKS(sbi) BLKS_PER_SEC(sbi) 205 #define DEF_MAX_DISCARD_REQUEST 8 /* issue 8 discards per round */ 206 #define DEF_MIN_DISCARD_ISSUE_TIME 50 /* 50 ms, if exists */ 207 #define DEF_MID_DISCARD_ISSUE_TIME 500 /* 500 ms, if device busy */ 208 #define DEF_MAX_DISCARD_ISSUE_TIME 60000 /* 60 s, if no candidates */ 209 #define DEF_DISCARD_URGENT_UTIL 80 /* do more discard over 80% */ 210 #define DEF_CP_INTERVAL 60 /* 60 secs */ 211 #define DEF_IDLE_INTERVAL 5 /* 5 secs */ 212 #define DEF_DISABLE_INTERVAL 5 /* 5 secs */ 213 #define DEF_DISABLE_QUICK_INTERVAL 1 /* 1 secs */ 214 #define DEF_UMOUNT_DISCARD_TIMEOUT 5 /* 5 secs */ 215 216 struct cp_control { 217 int reason; 218 __u64 trim_start; 219 __u64 trim_end; 220 __u64 trim_minlen; 221 }; 222 223 /* 224 * indicate meta/data type 225 */ 226 enum { 227 META_CP, 228 META_NAT, 229 META_SIT, 230 META_SSA, 231 META_MAX, 232 META_POR, 233 DATA_GENERIC, /* check range only */ 234 DATA_GENERIC_ENHANCE, /* strong check on range and segment bitmap */ 235 DATA_GENERIC_ENHANCE_READ, /* 236 * strong check on range and segment 237 * bitmap but no warning due to race 238 * condition of read on truncated area 239 * by extent_cache 240 */ 241 META_GENERIC, 242 }; 243 244 /* for the list of ino */ 245 enum { 246 ORPHAN_INO, /* for orphan ino list */ 247 APPEND_INO, /* for append ino list */ 248 UPDATE_INO, /* for update ino list */ 249 TRANS_DIR_INO, /* for trasactions dir ino list */ 250 FLUSH_INO, /* for multiple device flushing */ 251 MAX_INO_ENTRY, /* max. list */ 252 }; 253 254 struct ino_entry { 255 struct list_head list; /* list head */ 256 nid_t ino; /* inode number */ 257 unsigned int dirty_device; /* dirty device bitmap */ 258 }; 259 260 /* for the list of inodes to be GCed */ 261 struct inode_entry { 262 struct list_head list; /* list head */ 263 struct inode *inode; /* vfs inode pointer */ 264 }; 265 266 struct fsync_node_entry { 267 struct list_head list; /* list head */ 268 struct page *page; /* warm node page pointer */ 269 unsigned int seq_id; /* sequence id */ 270 }; 271 272 struct ckpt_req { 273 struct completion wait; /* completion for checkpoint done */ 274 struct llist_node llnode; /* llist_node to be linked in wait queue */ 275 int ret; /* return code of checkpoint */ 276 ktime_t queue_time; /* request queued time */ 277 }; 278 279 struct ckpt_req_control { 280 struct task_struct *f2fs_issue_ckpt; /* checkpoint task */ 281 int ckpt_thread_ioprio; /* checkpoint merge thread ioprio */ 282 wait_queue_head_t ckpt_wait_queue; /* waiting queue for wake-up */ 283 atomic_t issued_ckpt; /* # of actually issued ckpts */ 284 atomic_t total_ckpt; /* # of total ckpts */ 285 atomic_t queued_ckpt; /* # of queued ckpts */ 286 struct llist_head issue_list; /* list for command issue */ 287 spinlock_t stat_lock; /* lock for below checkpoint time stats */ 288 unsigned int cur_time; /* cur wait time in msec for currently issued checkpoint */ 289 unsigned int peak_time; /* peak wait time in msec until now */ 290 }; 291 292 /* for the bitmap indicate blocks to be discarded */ 293 struct discard_entry { 294 struct list_head list; /* list head */ 295 block_t start_blkaddr; /* start blockaddr of current segment */ 296 unsigned char discard_map[SIT_VBLOCK_MAP_SIZE]; /* segment discard bitmap */ 297 }; 298 299 /* default discard granularity of inner discard thread, unit: block count */ 300 #define DEFAULT_DISCARD_GRANULARITY 16 301 302 /* max discard pend list number */ 303 #define MAX_PLIST_NUM 512 304 #define plist_idx(blk_num) ((blk_num) >= MAX_PLIST_NUM ? \ 305 (MAX_PLIST_NUM - 1) : ((blk_num) - 1)) 306 307 enum { 308 D_PREP, /* initial */ 309 D_PARTIAL, /* partially submitted */ 310 D_SUBMIT, /* all submitted */ 311 D_DONE, /* finished */ 312 }; 313 314 struct discard_info { 315 block_t lstart; /* logical start address */ 316 block_t len; /* length */ 317 block_t start; /* actual start address in dev */ 318 }; 319 320 struct discard_cmd { 321 struct rb_node rb_node; /* rb node located in rb-tree */ 322 union { 323 struct { 324 block_t lstart; /* logical start address */ 325 block_t len; /* length */ 326 block_t start; /* actual start address in dev */ 327 }; 328 struct discard_info di; /* discard info */ 329 330 }; 331 struct list_head list; /* command list */ 332 struct completion wait; /* compleation */ 333 struct block_device *bdev; /* bdev */ 334 unsigned short ref; /* reference count */ 335 unsigned char state; /* state */ 336 unsigned char queued; /* queued discard */ 337 int error; /* bio error */ 338 spinlock_t lock; /* for state/bio_ref updating */ 339 unsigned short bio_ref; /* bio reference count */ 340 }; 341 342 enum { 343 DPOLICY_BG, 344 DPOLICY_FORCE, 345 DPOLICY_FSTRIM, 346 DPOLICY_UMOUNT, 347 MAX_DPOLICY, 348 }; 349 350 struct discard_policy { 351 int type; /* type of discard */ 352 unsigned int min_interval; /* used for candidates exist */ 353 unsigned int mid_interval; /* used for device busy */ 354 unsigned int max_interval; /* used for candidates not exist */ 355 unsigned int max_requests; /* # of discards issued per round */ 356 unsigned int io_aware_gran; /* minimum granularity discard not be aware of I/O */ 357 bool io_aware; /* issue discard in idle time */ 358 bool sync; /* submit discard with REQ_SYNC flag */ 359 bool ordered; /* issue discard by lba order */ 360 bool timeout; /* discard timeout for put_super */ 361 unsigned int granularity; /* discard granularity */ 362 }; 363 364 struct discard_cmd_control { 365 struct task_struct *f2fs_issue_discard; /* discard thread */ 366 struct list_head entry_list; /* 4KB discard entry list */ 367 struct list_head pend_list[MAX_PLIST_NUM];/* store pending entries */ 368 struct list_head wait_list; /* store on-flushing entries */ 369 struct list_head fstrim_list; /* in-flight discard from fstrim */ 370 wait_queue_head_t discard_wait_queue; /* waiting queue for wake-up */ 371 unsigned int discard_wake; /* to wake up discard thread */ 372 struct mutex cmd_lock; 373 unsigned int nr_discards; /* # of discards in the list */ 374 unsigned int max_discards; /* max. discards to be issued */ 375 unsigned int discard_granularity; /* discard granularity */ 376 unsigned int undiscard_blks; /* # of undiscard blocks */ 377 unsigned int next_pos; /* next discard position */ 378 atomic_t issued_discard; /* # of issued discard */ 379 atomic_t queued_discard; /* # of queued discard */ 380 atomic_t discard_cmd_cnt; /* # of cached cmd count */ 381 struct rb_root_cached root; /* root of discard rb-tree */ 382 bool rbtree_check; /* config for consistence check */ 383 }; 384 385 /* for the list of fsync inodes, used only during recovery */ 386 struct fsync_inode_entry { 387 struct list_head list; /* list head */ 388 struct inode *inode; /* vfs inode pointer */ 389 block_t blkaddr; /* block address locating the last fsync */ 390 block_t last_dentry; /* block address locating the last dentry */ 391 }; 392 393 #define nats_in_cursum(jnl) (le16_to_cpu((jnl)->n_nats)) 394 #define sits_in_cursum(jnl) (le16_to_cpu((jnl)->n_sits)) 395 396 #define nat_in_journal(jnl, i) ((jnl)->nat_j.entries[i].ne) 397 #define nid_in_journal(jnl, i) ((jnl)->nat_j.entries[i].nid) 398 #define sit_in_journal(jnl, i) ((jnl)->sit_j.entries[i].se) 399 #define segno_in_journal(jnl, i) ((jnl)->sit_j.entries[i].segno) 400 401 #define MAX_NAT_JENTRIES(jnl) (NAT_JOURNAL_ENTRIES - nats_in_cursum(jnl)) 402 #define MAX_SIT_JENTRIES(jnl) (SIT_JOURNAL_ENTRIES - sits_in_cursum(jnl)) 403 404 static inline int update_nats_in_cursum(struct f2fs_journal *journal, int i) 405 { 406 int before = nats_in_cursum(journal); 407 408 journal->n_nats = cpu_to_le16(before + i); 409 return before; 410 } 411 412 static inline int update_sits_in_cursum(struct f2fs_journal *journal, int i) 413 { 414 int before = sits_in_cursum(journal); 415 416 journal->n_sits = cpu_to_le16(before + i); 417 return before; 418 } 419 420 static inline bool __has_cursum_space(struct f2fs_journal *journal, 421 int size, int type) 422 { 423 if (type == NAT_JOURNAL) 424 return size <= MAX_NAT_JENTRIES(journal); 425 return size <= MAX_SIT_JENTRIES(journal); 426 } 427 428 /* for inline stuff */ 429 #define DEF_INLINE_RESERVED_SIZE 1 430 static inline int get_extra_isize(struct inode *inode); 431 static inline int get_inline_xattr_addrs(struct inode *inode); 432 #define MAX_INLINE_DATA(inode) (sizeof(__le32) * \ 433 (CUR_ADDRS_PER_INODE(inode) - \ 434 get_inline_xattr_addrs(inode) - \ 435 DEF_INLINE_RESERVED_SIZE)) 436 437 /* for inline dir */ 438 #define NR_INLINE_DENTRY(inode) (MAX_INLINE_DATA(inode) * BITS_PER_BYTE / \ 439 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \ 440 BITS_PER_BYTE + 1)) 441 #define INLINE_DENTRY_BITMAP_SIZE(inode) \ 442 DIV_ROUND_UP(NR_INLINE_DENTRY(inode), BITS_PER_BYTE) 443 #define INLINE_RESERVED_SIZE(inode) (MAX_INLINE_DATA(inode) - \ 444 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \ 445 NR_INLINE_DENTRY(inode) + \ 446 INLINE_DENTRY_BITMAP_SIZE(inode))) 447 448 /* 449 * For INODE and NODE manager 450 */ 451 /* for directory operations */ 452 453 struct f2fs_filename { 454 /* 455 * The filename the user specified. This is NULL for some 456 * filesystem-internal operations, e.g. converting an inline directory 457 * to a non-inline one, or roll-forward recovering an encrypted dentry. 458 */ 459 const struct qstr *usr_fname; 460 461 /* 462 * The on-disk filename. For encrypted directories, this is encrypted. 463 * This may be NULL for lookups in an encrypted dir without the key. 464 */ 465 struct fscrypt_str disk_name; 466 467 /* The dirhash of this filename */ 468 f2fs_hash_t hash; 469 470 #ifdef CONFIG_FS_ENCRYPTION 471 /* 472 * For lookups in encrypted directories: either the buffer backing 473 * disk_name, or a buffer that holds the decoded no-key name. 474 */ 475 struct fscrypt_str crypto_buf; 476 #endif 477 #ifdef CONFIG_UNICODE 478 /* 479 * For casefolded directories: the casefolded name, but it's left NULL 480 * if the original name is not valid Unicode, if the directory is both 481 * casefolded and encrypted and its encryption key is unavailable, or if 482 * the filesystem is doing an internal operation where usr_fname is also 483 * NULL. In all these cases we fall back to treating the name as an 484 * opaque byte sequence. 485 */ 486 struct fscrypt_str cf_name; 487 #endif 488 }; 489 490 struct f2fs_dentry_ptr { 491 struct inode *inode; 492 void *bitmap; 493 struct f2fs_dir_entry *dentry; 494 __u8 (*filename)[F2FS_SLOT_LEN]; 495 int max; 496 int nr_bitmap; 497 }; 498 499 static inline void make_dentry_ptr_block(struct inode *inode, 500 struct f2fs_dentry_ptr *d, struct f2fs_dentry_block *t) 501 { 502 d->inode = inode; 503 d->max = NR_DENTRY_IN_BLOCK; 504 d->nr_bitmap = SIZE_OF_DENTRY_BITMAP; 505 d->bitmap = t->dentry_bitmap; 506 d->dentry = t->dentry; 507 d->filename = t->filename; 508 } 509 510 static inline void make_dentry_ptr_inline(struct inode *inode, 511 struct f2fs_dentry_ptr *d, void *t) 512 { 513 int entry_cnt = NR_INLINE_DENTRY(inode); 514 int bitmap_size = INLINE_DENTRY_BITMAP_SIZE(inode); 515 int reserved_size = INLINE_RESERVED_SIZE(inode); 516 517 d->inode = inode; 518 d->max = entry_cnt; 519 d->nr_bitmap = bitmap_size; 520 d->bitmap = t; 521 d->dentry = t + bitmap_size + reserved_size; 522 d->filename = t + bitmap_size + reserved_size + 523 SIZE_OF_DIR_ENTRY * entry_cnt; 524 } 525 526 /* 527 * XATTR_NODE_OFFSET stores xattrs to one node block per file keeping -1 528 * as its node offset to distinguish from index node blocks. 529 * But some bits are used to mark the node block. 530 */ 531 #define XATTR_NODE_OFFSET ((((unsigned int)-1) << OFFSET_BIT_SHIFT) \ 532 >> OFFSET_BIT_SHIFT) 533 enum { 534 ALLOC_NODE, /* allocate a new node page if needed */ 535 LOOKUP_NODE, /* look up a node without readahead */ 536 LOOKUP_NODE_RA, /* 537 * look up a node with readahead called 538 * by get_data_block. 539 */ 540 }; 541 542 #define DEFAULT_RETRY_IO_COUNT 8 /* maximum retry read IO count */ 543 544 /* congestion wait timeout value, default: 20ms */ 545 #define DEFAULT_IO_TIMEOUT (msecs_to_jiffies(20)) 546 547 /* maximum retry quota flush count */ 548 #define DEFAULT_RETRY_QUOTA_FLUSH_COUNT 8 549 550 #define F2FS_LINK_MAX 0xffffffff /* maximum link count per file */ 551 552 #define MAX_DIR_RA_PAGES 4 /* maximum ra pages of dir */ 553 554 /* for in-memory extent cache entry */ 555 #define F2FS_MIN_EXTENT_LEN 64 /* minimum extent length */ 556 557 /* number of extent info in extent cache we try to shrink */ 558 #define EXTENT_CACHE_SHRINK_NUMBER 128 559 560 struct rb_entry { 561 struct rb_node rb_node; /* rb node located in rb-tree */ 562 union { 563 struct { 564 unsigned int ofs; /* start offset of the entry */ 565 unsigned int len; /* length of the entry */ 566 }; 567 unsigned long long key; /* 64-bits key */ 568 } __packed; 569 }; 570 571 struct extent_info { 572 unsigned int fofs; /* start offset in a file */ 573 unsigned int len; /* length of the extent */ 574 u32 blk; /* start block address of the extent */ 575 }; 576 577 struct extent_node { 578 struct rb_node rb_node; /* rb node located in rb-tree */ 579 struct extent_info ei; /* extent info */ 580 struct list_head list; /* node in global extent list of sbi */ 581 struct extent_tree *et; /* extent tree pointer */ 582 }; 583 584 struct extent_tree { 585 nid_t ino; /* inode number */ 586 struct rb_root_cached root; /* root of extent info rb-tree */ 587 struct extent_node *cached_en; /* recently accessed extent node */ 588 struct extent_info largest; /* largested extent info */ 589 struct list_head list; /* to be used by sbi->zombie_list */ 590 rwlock_t lock; /* protect extent info rb-tree */ 591 atomic_t node_cnt; /* # of extent node in rb-tree*/ 592 bool largest_updated; /* largest extent updated */ 593 }; 594 595 /* 596 * This structure is taken from ext4_map_blocks. 597 * 598 * Note that, however, f2fs uses NEW and MAPPED flags for f2fs_map_blocks(). 599 */ 600 #define F2FS_MAP_NEW (1 << BH_New) 601 #define F2FS_MAP_MAPPED (1 << BH_Mapped) 602 #define F2FS_MAP_UNWRITTEN (1 << BH_Unwritten) 603 #define F2FS_MAP_FLAGS (F2FS_MAP_NEW | F2FS_MAP_MAPPED |\ 604 F2FS_MAP_UNWRITTEN) 605 606 struct f2fs_map_blocks { 607 block_t m_pblk; 608 block_t m_lblk; 609 unsigned int m_len; 610 unsigned int m_flags; 611 pgoff_t *m_next_pgofs; /* point next possible non-hole pgofs */ 612 pgoff_t *m_next_extent; /* point to next possible extent */ 613 int m_seg_type; 614 bool m_may_create; /* indicate it is from write path */ 615 }; 616 617 /* for flag in get_data_block */ 618 enum { 619 F2FS_GET_BLOCK_DEFAULT, 620 F2FS_GET_BLOCK_FIEMAP, 621 F2FS_GET_BLOCK_BMAP, 622 F2FS_GET_BLOCK_DIO, 623 F2FS_GET_BLOCK_PRE_DIO, 624 F2FS_GET_BLOCK_PRE_AIO, 625 F2FS_GET_BLOCK_PRECACHE, 626 }; 627 628 /* 629 * i_advise uses FADVISE_XXX_BIT. We can add additional hints later. 630 */ 631 #define FADVISE_COLD_BIT 0x01 632 #define FADVISE_LOST_PINO_BIT 0x02 633 #define FADVISE_ENCRYPT_BIT 0x04 634 #define FADVISE_ENC_NAME_BIT 0x08 635 #define FADVISE_KEEP_SIZE_BIT 0x10 636 #define FADVISE_HOT_BIT 0x20 637 #define FADVISE_VERITY_BIT 0x40 638 639 #define FADVISE_MODIFIABLE_BITS (FADVISE_COLD_BIT | FADVISE_HOT_BIT) 640 641 #define file_is_cold(inode) is_file(inode, FADVISE_COLD_BIT) 642 #define file_set_cold(inode) set_file(inode, FADVISE_COLD_BIT) 643 #define file_clear_cold(inode) clear_file(inode, FADVISE_COLD_BIT) 644 645 #define file_wrong_pino(inode) is_file(inode, FADVISE_LOST_PINO_BIT) 646 #define file_lost_pino(inode) set_file(inode, FADVISE_LOST_PINO_BIT) 647 #define file_got_pino(inode) clear_file(inode, FADVISE_LOST_PINO_BIT) 648 649 #define file_is_encrypt(inode) is_file(inode, FADVISE_ENCRYPT_BIT) 650 #define file_set_encrypt(inode) set_file(inode, FADVISE_ENCRYPT_BIT) 651 652 #define file_enc_name(inode) is_file(inode, FADVISE_ENC_NAME_BIT) 653 #define file_set_enc_name(inode) set_file(inode, FADVISE_ENC_NAME_BIT) 654 655 #define file_keep_isize(inode) is_file(inode, FADVISE_KEEP_SIZE_BIT) 656 #define file_set_keep_isize(inode) set_file(inode, FADVISE_KEEP_SIZE_BIT) 657 658 #define file_is_hot(inode) is_file(inode, FADVISE_HOT_BIT) 659 #define file_set_hot(inode) set_file(inode, FADVISE_HOT_BIT) 660 #define file_clear_hot(inode) clear_file(inode, FADVISE_HOT_BIT) 661 662 #define file_is_verity(inode) is_file(inode, FADVISE_VERITY_BIT) 663 #define file_set_verity(inode) set_file(inode, FADVISE_VERITY_BIT) 664 665 #define DEF_DIR_LEVEL 0 666 667 enum { 668 GC_FAILURE_PIN, 669 GC_FAILURE_ATOMIC, 670 MAX_GC_FAILURE 671 }; 672 673 /* used for f2fs_inode_info->flags */ 674 enum { 675 FI_NEW_INODE, /* indicate newly allocated inode */ 676 FI_DIRTY_INODE, /* indicate inode is dirty or not */ 677 FI_AUTO_RECOVER, /* indicate inode is recoverable */ 678 FI_DIRTY_DIR, /* indicate directory has dirty pages */ 679 FI_INC_LINK, /* need to increment i_nlink */ 680 FI_ACL_MODE, /* indicate acl mode */ 681 FI_NO_ALLOC, /* should not allocate any blocks */ 682 FI_FREE_NID, /* free allocated nide */ 683 FI_NO_EXTENT, /* not to use the extent cache */ 684 FI_INLINE_XATTR, /* used for inline xattr */ 685 FI_INLINE_DATA, /* used for inline data*/ 686 FI_INLINE_DENTRY, /* used for inline dentry */ 687 FI_APPEND_WRITE, /* inode has appended data */ 688 FI_UPDATE_WRITE, /* inode has in-place-update data */ 689 FI_NEED_IPU, /* used for ipu per file */ 690 FI_ATOMIC_FILE, /* indicate atomic file */ 691 FI_ATOMIC_COMMIT, /* indicate the state of atomical committing */ 692 FI_VOLATILE_FILE, /* indicate volatile file */ 693 FI_FIRST_BLOCK_WRITTEN, /* indicate #0 data block was written */ 694 FI_DROP_CACHE, /* drop dirty page cache */ 695 FI_DATA_EXIST, /* indicate data exists */ 696 FI_INLINE_DOTS, /* indicate inline dot dentries */ 697 FI_DO_DEFRAG, /* indicate defragment is running */ 698 FI_DIRTY_FILE, /* indicate regular/symlink has dirty pages */ 699 FI_NO_PREALLOC, /* indicate skipped preallocated blocks */ 700 FI_HOT_DATA, /* indicate file is hot */ 701 FI_EXTRA_ATTR, /* indicate file has extra attribute */ 702 FI_PROJ_INHERIT, /* indicate file inherits projectid */ 703 FI_PIN_FILE, /* indicate file should not be gced */ 704 FI_ATOMIC_REVOKE_REQUEST, /* request to drop atomic data */ 705 FI_VERITY_IN_PROGRESS, /* building fs-verity Merkle tree */ 706 FI_COMPRESSED_FILE, /* indicate file's data can be compressed */ 707 FI_COMPRESS_CORRUPT, /* indicate compressed cluster is corrupted */ 708 FI_MMAP_FILE, /* indicate file was mmapped */ 709 FI_ENABLE_COMPRESS, /* enable compression in "user" compression mode */ 710 FI_COMPRESS_RELEASED, /* compressed blocks were released */ 711 FI_MAX, /* max flag, never be used */ 712 }; 713 714 struct f2fs_inode_info { 715 struct inode vfs_inode; /* serve a vfs inode */ 716 unsigned long i_flags; /* keep an inode flags for ioctl */ 717 unsigned char i_advise; /* use to give file attribute hints */ 718 unsigned char i_dir_level; /* use for dentry level for large dir */ 719 unsigned int i_current_depth; /* only for directory depth */ 720 /* for gc failure statistic */ 721 unsigned int i_gc_failures[MAX_GC_FAILURE]; 722 unsigned int i_pino; /* parent inode number */ 723 umode_t i_acl_mode; /* keep file acl mode temporarily */ 724 725 /* Use below internally in f2fs*/ 726 unsigned long flags[BITS_TO_LONGS(FI_MAX)]; /* use to pass per-file flags */ 727 struct rw_semaphore i_sem; /* protect fi info */ 728 atomic_t dirty_pages; /* # of dirty pages */ 729 f2fs_hash_t chash; /* hash value of given file name */ 730 unsigned int clevel; /* maximum level of given file name */ 731 struct task_struct *task; /* lookup and create consistency */ 732 struct task_struct *cp_task; /* separate cp/wb IO stats*/ 733 nid_t i_xattr_nid; /* node id that contains xattrs */ 734 loff_t last_disk_size; /* lastly written file size */ 735 spinlock_t i_size_lock; /* protect last_disk_size */ 736 737 #ifdef CONFIG_QUOTA 738 struct dquot *i_dquot[MAXQUOTAS]; 739 740 /* quota space reservation, managed internally by quota code */ 741 qsize_t i_reserved_quota; 742 #endif 743 struct list_head dirty_list; /* dirty list for dirs and files */ 744 struct list_head gdirty_list; /* linked in global dirty list */ 745 struct list_head inmem_ilist; /* list for inmem inodes */ 746 struct list_head inmem_pages; /* inmemory pages managed by f2fs */ 747 struct task_struct *inmem_task; /* store inmemory task */ 748 struct mutex inmem_lock; /* lock for inmemory pages */ 749 struct extent_tree *extent_tree; /* cached extent_tree entry */ 750 751 /* avoid racing between foreground op and gc */ 752 struct rw_semaphore i_gc_rwsem[2]; 753 struct rw_semaphore i_mmap_sem; 754 struct rw_semaphore i_xattr_sem; /* avoid racing between reading and changing EAs */ 755 756 int i_extra_isize; /* size of extra space located in i_addr */ 757 kprojid_t i_projid; /* id for project quota */ 758 int i_inline_xattr_size; /* inline xattr size */ 759 struct timespec64 i_crtime; /* inode creation time */ 760 struct timespec64 i_disk_time[4];/* inode disk times */ 761 762 /* for file compress */ 763 atomic_t i_compr_blocks; /* # of compressed blocks */ 764 unsigned char i_compress_algorithm; /* algorithm type */ 765 unsigned char i_log_cluster_size; /* log of cluster size */ 766 unsigned char i_compress_level; /* compress level (lz4hc,zstd) */ 767 unsigned short i_compress_flag; /* compress flag */ 768 unsigned int i_cluster_size; /* cluster size */ 769 }; 770 771 static inline void get_extent_info(struct extent_info *ext, 772 struct f2fs_extent *i_ext) 773 { 774 ext->fofs = le32_to_cpu(i_ext->fofs); 775 ext->blk = le32_to_cpu(i_ext->blk); 776 ext->len = le32_to_cpu(i_ext->len); 777 } 778 779 static inline void set_raw_extent(struct extent_info *ext, 780 struct f2fs_extent *i_ext) 781 { 782 i_ext->fofs = cpu_to_le32(ext->fofs); 783 i_ext->blk = cpu_to_le32(ext->blk); 784 i_ext->len = cpu_to_le32(ext->len); 785 } 786 787 static inline void set_extent_info(struct extent_info *ei, unsigned int fofs, 788 u32 blk, unsigned int len) 789 { 790 ei->fofs = fofs; 791 ei->blk = blk; 792 ei->len = len; 793 } 794 795 static inline bool __is_discard_mergeable(struct discard_info *back, 796 struct discard_info *front, unsigned int max_len) 797 { 798 return (back->lstart + back->len == front->lstart) && 799 (back->len + front->len <= max_len); 800 } 801 802 static inline bool __is_discard_back_mergeable(struct discard_info *cur, 803 struct discard_info *back, unsigned int max_len) 804 { 805 return __is_discard_mergeable(back, cur, max_len); 806 } 807 808 static inline bool __is_discard_front_mergeable(struct discard_info *cur, 809 struct discard_info *front, unsigned int max_len) 810 { 811 return __is_discard_mergeable(cur, front, max_len); 812 } 813 814 static inline bool __is_extent_mergeable(struct extent_info *back, 815 struct extent_info *front) 816 { 817 return (back->fofs + back->len == front->fofs && 818 back->blk + back->len == front->blk); 819 } 820 821 static inline bool __is_back_mergeable(struct extent_info *cur, 822 struct extent_info *back) 823 { 824 return __is_extent_mergeable(back, cur); 825 } 826 827 static inline bool __is_front_mergeable(struct extent_info *cur, 828 struct extent_info *front) 829 { 830 return __is_extent_mergeable(cur, front); 831 } 832 833 extern void f2fs_mark_inode_dirty_sync(struct inode *inode, bool sync); 834 static inline void __try_update_largest_extent(struct extent_tree *et, 835 struct extent_node *en) 836 { 837 if (en->ei.len > et->largest.len) { 838 et->largest = en->ei; 839 et->largest_updated = true; 840 } 841 } 842 843 /* 844 * For free nid management 845 */ 846 enum nid_state { 847 FREE_NID, /* newly added to free nid list */ 848 PREALLOC_NID, /* it is preallocated */ 849 MAX_NID_STATE, 850 }; 851 852 enum nat_state { 853 TOTAL_NAT, 854 DIRTY_NAT, 855 RECLAIMABLE_NAT, 856 MAX_NAT_STATE, 857 }; 858 859 struct f2fs_nm_info { 860 block_t nat_blkaddr; /* base disk address of NAT */ 861 nid_t max_nid; /* maximum possible node ids */ 862 nid_t available_nids; /* # of available node ids */ 863 nid_t next_scan_nid; /* the next nid to be scanned */ 864 unsigned int ram_thresh; /* control the memory footprint */ 865 unsigned int ra_nid_pages; /* # of nid pages to be readaheaded */ 866 unsigned int dirty_nats_ratio; /* control dirty nats ratio threshold */ 867 868 /* NAT cache management */ 869 struct radix_tree_root nat_root;/* root of the nat entry cache */ 870 struct radix_tree_root nat_set_root;/* root of the nat set cache */ 871 struct rw_semaphore nat_tree_lock; /* protect nat entry tree */ 872 struct list_head nat_entries; /* cached nat entry list (clean) */ 873 spinlock_t nat_list_lock; /* protect clean nat entry list */ 874 unsigned int nat_cnt[MAX_NAT_STATE]; /* the # of cached nat entries */ 875 unsigned int nat_blocks; /* # of nat blocks */ 876 877 /* free node ids management */ 878 struct radix_tree_root free_nid_root;/* root of the free_nid cache */ 879 struct list_head free_nid_list; /* list for free nids excluding preallocated nids */ 880 unsigned int nid_cnt[MAX_NID_STATE]; /* the number of free node id */ 881 spinlock_t nid_list_lock; /* protect nid lists ops */ 882 struct mutex build_lock; /* lock for build free nids */ 883 unsigned char **free_nid_bitmap; 884 unsigned char *nat_block_bitmap; 885 unsigned short *free_nid_count; /* free nid count of NAT block */ 886 887 /* for checkpoint */ 888 char *nat_bitmap; /* NAT bitmap pointer */ 889 890 unsigned int nat_bits_blocks; /* # of nat bits blocks */ 891 unsigned char *nat_bits; /* NAT bits blocks */ 892 unsigned char *full_nat_bits; /* full NAT pages */ 893 unsigned char *empty_nat_bits; /* empty NAT pages */ 894 #ifdef CONFIG_F2FS_CHECK_FS 895 char *nat_bitmap_mir; /* NAT bitmap mirror */ 896 #endif 897 int bitmap_size; /* bitmap size */ 898 }; 899 900 /* 901 * this structure is used as one of function parameters. 902 * all the information are dedicated to a given direct node block determined 903 * by the data offset in a file. 904 */ 905 struct dnode_of_data { 906 struct inode *inode; /* vfs inode pointer */ 907 struct page *inode_page; /* its inode page, NULL is possible */ 908 struct page *node_page; /* cached direct node page */ 909 nid_t nid; /* node id of the direct node block */ 910 unsigned int ofs_in_node; /* data offset in the node page */ 911 bool inode_page_locked; /* inode page is locked or not */ 912 bool node_changed; /* is node block changed */ 913 char cur_level; /* level of hole node page */ 914 char max_level; /* level of current page located */ 915 block_t data_blkaddr; /* block address of the node block */ 916 }; 917 918 static inline void set_new_dnode(struct dnode_of_data *dn, struct inode *inode, 919 struct page *ipage, struct page *npage, nid_t nid) 920 { 921 memset(dn, 0, sizeof(*dn)); 922 dn->inode = inode; 923 dn->inode_page = ipage; 924 dn->node_page = npage; 925 dn->nid = nid; 926 } 927 928 /* 929 * For SIT manager 930 * 931 * By default, there are 6 active log areas across the whole main area. 932 * When considering hot and cold data separation to reduce cleaning overhead, 933 * we split 3 for data logs and 3 for node logs as hot, warm, and cold types, 934 * respectively. 935 * In the current design, you should not change the numbers intentionally. 936 * Instead, as a mount option such as active_logs=x, you can use 2, 4, and 6 937 * logs individually according to the underlying devices. (default: 6) 938 * Just in case, on-disk layout covers maximum 16 logs that consist of 8 for 939 * data and 8 for node logs. 940 */ 941 #define NR_CURSEG_DATA_TYPE (3) 942 #define NR_CURSEG_NODE_TYPE (3) 943 #define NR_CURSEG_INMEM_TYPE (2) 944 #define NR_CURSEG_RO_TYPE (2) 945 #define NR_CURSEG_PERSIST_TYPE (NR_CURSEG_DATA_TYPE + NR_CURSEG_NODE_TYPE) 946 #define NR_CURSEG_TYPE (NR_CURSEG_INMEM_TYPE + NR_CURSEG_PERSIST_TYPE) 947 948 enum { 949 CURSEG_HOT_DATA = 0, /* directory entry blocks */ 950 CURSEG_WARM_DATA, /* data blocks */ 951 CURSEG_COLD_DATA, /* multimedia or GCed data blocks */ 952 CURSEG_HOT_NODE, /* direct node blocks of directory files */ 953 CURSEG_WARM_NODE, /* direct node blocks of normal files */ 954 CURSEG_COLD_NODE, /* indirect node blocks */ 955 NR_PERSISTENT_LOG, /* number of persistent log */ 956 CURSEG_COLD_DATA_PINNED = NR_PERSISTENT_LOG, 957 /* pinned file that needs consecutive block address */ 958 CURSEG_ALL_DATA_ATGC, /* SSR alloctor in hot/warm/cold data area */ 959 NO_CHECK_TYPE, /* number of persistent & inmem log */ 960 }; 961 962 struct flush_cmd { 963 struct completion wait; 964 struct llist_node llnode; 965 nid_t ino; 966 int ret; 967 }; 968 969 struct flush_cmd_control { 970 struct task_struct *f2fs_issue_flush; /* flush thread */ 971 wait_queue_head_t flush_wait_queue; /* waiting queue for wake-up */ 972 atomic_t issued_flush; /* # of issued flushes */ 973 atomic_t queued_flush; /* # of queued flushes */ 974 struct llist_head issue_list; /* list for command issue */ 975 struct llist_node *dispatch_list; /* list for command dispatch */ 976 }; 977 978 struct f2fs_sm_info { 979 struct sit_info *sit_info; /* whole segment information */ 980 struct free_segmap_info *free_info; /* free segment information */ 981 struct dirty_seglist_info *dirty_info; /* dirty segment information */ 982 struct curseg_info *curseg_array; /* active segment information */ 983 984 struct rw_semaphore curseg_lock; /* for preventing curseg change */ 985 986 block_t seg0_blkaddr; /* block address of 0'th segment */ 987 block_t main_blkaddr; /* start block address of main area */ 988 block_t ssa_blkaddr; /* start block address of SSA area */ 989 990 unsigned int segment_count; /* total # of segments */ 991 unsigned int main_segments; /* # of segments in main area */ 992 unsigned int reserved_segments; /* # of reserved segments */ 993 unsigned int ovp_segments; /* # of overprovision segments */ 994 995 /* a threshold to reclaim prefree segments */ 996 unsigned int rec_prefree_segments; 997 998 /* for batched trimming */ 999 unsigned int trim_sections; /* # of sections to trim */ 1000 1001 struct list_head sit_entry_set; /* sit entry set list */ 1002 1003 unsigned int ipu_policy; /* in-place-update policy */ 1004 unsigned int min_ipu_util; /* in-place-update threshold */ 1005 unsigned int min_fsync_blocks; /* threshold for fsync */ 1006 unsigned int min_seq_blocks; /* threshold for sequential blocks */ 1007 unsigned int min_hot_blocks; /* threshold for hot block allocation */ 1008 unsigned int min_ssr_sections; /* threshold to trigger SSR allocation */ 1009 1010 /* for flush command control */ 1011 struct flush_cmd_control *fcc_info; 1012 1013 /* for discard command control */ 1014 struct discard_cmd_control *dcc_info; 1015 }; 1016 1017 /* 1018 * For superblock 1019 */ 1020 /* 1021 * COUNT_TYPE for monitoring 1022 * 1023 * f2fs monitors the number of several block types such as on-writeback, 1024 * dirty dentry blocks, dirty node blocks, and dirty meta blocks. 1025 */ 1026 #define WB_DATA_TYPE(p) (__is_cp_guaranteed(p) ? F2FS_WB_CP_DATA : F2FS_WB_DATA) 1027 enum count_type { 1028 F2FS_DIRTY_DENTS, 1029 F2FS_DIRTY_DATA, 1030 F2FS_DIRTY_QDATA, 1031 F2FS_DIRTY_NODES, 1032 F2FS_DIRTY_META, 1033 F2FS_INMEM_PAGES, 1034 F2FS_DIRTY_IMETA, 1035 F2FS_WB_CP_DATA, 1036 F2FS_WB_DATA, 1037 F2FS_RD_DATA, 1038 F2FS_RD_NODE, 1039 F2FS_RD_META, 1040 F2FS_DIO_WRITE, 1041 F2FS_DIO_READ, 1042 NR_COUNT_TYPE, 1043 }; 1044 1045 /* 1046 * The below are the page types of bios used in submit_bio(). 1047 * The available types are: 1048 * DATA User data pages. It operates as async mode. 1049 * NODE Node pages. It operates as async mode. 1050 * META FS metadata pages such as SIT, NAT, CP. 1051 * NR_PAGE_TYPE The number of page types. 1052 * META_FLUSH Make sure the previous pages are written 1053 * with waiting the bio's completion 1054 * ... Only can be used with META. 1055 */ 1056 #define PAGE_TYPE_OF_BIO(type) ((type) > META ? META : (type)) 1057 enum page_type { 1058 DATA, 1059 NODE, 1060 META, 1061 NR_PAGE_TYPE, 1062 META_FLUSH, 1063 INMEM, /* the below types are used by tracepoints only. */ 1064 INMEM_DROP, 1065 INMEM_INVALIDATE, 1066 INMEM_REVOKE, 1067 IPU, 1068 OPU, 1069 }; 1070 1071 enum temp_type { 1072 HOT = 0, /* must be zero for meta bio */ 1073 WARM, 1074 COLD, 1075 NR_TEMP_TYPE, 1076 }; 1077 1078 enum need_lock_type { 1079 LOCK_REQ = 0, 1080 LOCK_DONE, 1081 LOCK_RETRY, 1082 }; 1083 1084 enum cp_reason_type { 1085 CP_NO_NEEDED, 1086 CP_NON_REGULAR, 1087 CP_COMPRESSED, 1088 CP_HARDLINK, 1089 CP_SB_NEED_CP, 1090 CP_WRONG_PINO, 1091 CP_NO_SPC_ROLL, 1092 CP_NODE_NEED_CP, 1093 CP_FASTBOOT_MODE, 1094 CP_SPEC_LOG_NUM, 1095 CP_RECOVER_DIR, 1096 }; 1097 1098 enum iostat_type { 1099 /* WRITE IO */ 1100 APP_DIRECT_IO, /* app direct write IOs */ 1101 APP_BUFFERED_IO, /* app buffered write IOs */ 1102 APP_WRITE_IO, /* app write IOs */ 1103 APP_MAPPED_IO, /* app mapped IOs */ 1104 FS_DATA_IO, /* data IOs from kworker/fsync/reclaimer */ 1105 FS_NODE_IO, /* node IOs from kworker/fsync/reclaimer */ 1106 FS_META_IO, /* meta IOs from kworker/reclaimer */ 1107 FS_GC_DATA_IO, /* data IOs from forground gc */ 1108 FS_GC_NODE_IO, /* node IOs from forground gc */ 1109 FS_CP_DATA_IO, /* data IOs from checkpoint */ 1110 FS_CP_NODE_IO, /* node IOs from checkpoint */ 1111 FS_CP_META_IO, /* meta IOs from checkpoint */ 1112 1113 /* READ IO */ 1114 APP_DIRECT_READ_IO, /* app direct read IOs */ 1115 APP_BUFFERED_READ_IO, /* app buffered read IOs */ 1116 APP_READ_IO, /* app read IOs */ 1117 APP_MAPPED_READ_IO, /* app mapped read IOs */ 1118 FS_DATA_READ_IO, /* data read IOs */ 1119 FS_GDATA_READ_IO, /* data read IOs from background gc */ 1120 FS_CDATA_READ_IO, /* compressed data read IOs */ 1121 FS_NODE_READ_IO, /* node read IOs */ 1122 FS_META_READ_IO, /* meta read IOs */ 1123 1124 /* other */ 1125 FS_DISCARD, /* discard */ 1126 NR_IO_TYPE, 1127 }; 1128 1129 struct f2fs_io_info { 1130 struct f2fs_sb_info *sbi; /* f2fs_sb_info pointer */ 1131 nid_t ino; /* inode number */ 1132 enum page_type type; /* contains DATA/NODE/META/META_FLUSH */ 1133 enum temp_type temp; /* contains HOT/WARM/COLD */ 1134 int op; /* contains REQ_OP_ */ 1135 int op_flags; /* req_flag_bits */ 1136 block_t new_blkaddr; /* new block address to be written */ 1137 block_t old_blkaddr; /* old block address before Cow */ 1138 struct page *page; /* page to be written */ 1139 struct page *encrypted_page; /* encrypted page */ 1140 struct page *compressed_page; /* compressed page */ 1141 struct list_head list; /* serialize IOs */ 1142 bool submitted; /* indicate IO submission */ 1143 int need_lock; /* indicate we need to lock cp_rwsem */ 1144 bool in_list; /* indicate fio is in io_list */ 1145 bool is_por; /* indicate IO is from recovery or not */ 1146 bool retry; /* need to reallocate block address */ 1147 int compr_blocks; /* # of compressed block addresses */ 1148 bool encrypted; /* indicate file is encrypted */ 1149 enum iostat_type io_type; /* io type */ 1150 struct writeback_control *io_wbc; /* writeback control */ 1151 struct bio **bio; /* bio for ipu */ 1152 sector_t *last_block; /* last block number in bio */ 1153 unsigned char version; /* version of the node */ 1154 }; 1155 1156 struct bio_entry { 1157 struct bio *bio; 1158 struct list_head list; 1159 }; 1160 1161 #define is_read_io(rw) ((rw) == READ) 1162 struct f2fs_bio_info { 1163 struct f2fs_sb_info *sbi; /* f2fs superblock */ 1164 struct bio *bio; /* bios to merge */ 1165 sector_t last_block_in_bio; /* last block number */ 1166 struct f2fs_io_info fio; /* store buffered io info. */ 1167 struct rw_semaphore io_rwsem; /* blocking op for bio */ 1168 spinlock_t io_lock; /* serialize DATA/NODE IOs */ 1169 struct list_head io_list; /* track fios */ 1170 struct list_head bio_list; /* bio entry list head */ 1171 struct rw_semaphore bio_list_lock; /* lock to protect bio entry list */ 1172 }; 1173 1174 #define FDEV(i) (sbi->devs[i]) 1175 #define RDEV(i) (raw_super->devs[i]) 1176 struct f2fs_dev_info { 1177 struct block_device *bdev; 1178 char path[MAX_PATH_LEN]; 1179 unsigned int total_segments; 1180 block_t start_blk; 1181 block_t end_blk; 1182 #ifdef CONFIG_BLK_DEV_ZONED 1183 unsigned int nr_blkz; /* Total number of zones */ 1184 unsigned long *blkz_seq; /* Bitmap indicating sequential zones */ 1185 block_t *zone_capacity_blocks; /* Array of zone capacity in blks */ 1186 #endif 1187 }; 1188 1189 enum inode_type { 1190 DIR_INODE, /* for dirty dir inode */ 1191 FILE_INODE, /* for dirty regular/symlink inode */ 1192 DIRTY_META, /* for all dirtied inode metadata */ 1193 ATOMIC_FILE, /* for all atomic files */ 1194 NR_INODE_TYPE, 1195 }; 1196 1197 /* for inner inode cache management */ 1198 struct inode_management { 1199 struct radix_tree_root ino_root; /* ino entry array */ 1200 spinlock_t ino_lock; /* for ino entry lock */ 1201 struct list_head ino_list; /* inode list head */ 1202 unsigned long ino_num; /* number of entries */ 1203 }; 1204 1205 /* for GC_AT */ 1206 struct atgc_management { 1207 bool atgc_enabled; /* ATGC is enabled or not */ 1208 struct rb_root_cached root; /* root of victim rb-tree */ 1209 struct list_head victim_list; /* linked with all victim entries */ 1210 unsigned int victim_count; /* victim count in rb-tree */ 1211 unsigned int candidate_ratio; /* candidate ratio */ 1212 unsigned int max_candidate_count; /* max candidate count */ 1213 unsigned int age_weight; /* age weight, vblock_weight = 100 - age_weight */ 1214 unsigned long long age_threshold; /* age threshold */ 1215 }; 1216 1217 /* For s_flag in struct f2fs_sb_info */ 1218 enum { 1219 SBI_IS_DIRTY, /* dirty flag for checkpoint */ 1220 SBI_IS_CLOSE, /* specify unmounting */ 1221 SBI_NEED_FSCK, /* need fsck.f2fs to fix */ 1222 SBI_POR_DOING, /* recovery is doing or not */ 1223 SBI_NEED_SB_WRITE, /* need to recover superblock */ 1224 SBI_NEED_CP, /* need to checkpoint */ 1225 SBI_IS_SHUTDOWN, /* shutdown by ioctl */ 1226 SBI_IS_RECOVERED, /* recovered orphan/data */ 1227 SBI_CP_DISABLED, /* CP was disabled last mount */ 1228 SBI_CP_DISABLED_QUICK, /* CP was disabled quickly */ 1229 SBI_QUOTA_NEED_FLUSH, /* need to flush quota info in CP */ 1230 SBI_QUOTA_SKIP_FLUSH, /* skip flushing quota in current CP */ 1231 SBI_QUOTA_NEED_REPAIR, /* quota file may be corrupted */ 1232 SBI_IS_RESIZEFS, /* resizefs is in process */ 1233 }; 1234 1235 enum { 1236 CP_TIME, 1237 REQ_TIME, 1238 DISCARD_TIME, 1239 GC_TIME, 1240 DISABLE_TIME, 1241 UMOUNT_DISCARD_TIMEOUT, 1242 MAX_TIME, 1243 }; 1244 1245 enum { 1246 GC_NORMAL, 1247 GC_IDLE_CB, 1248 GC_IDLE_GREEDY, 1249 GC_IDLE_AT, 1250 GC_URGENT_HIGH, 1251 GC_URGENT_LOW, 1252 }; 1253 1254 enum { 1255 BGGC_MODE_ON, /* background gc is on */ 1256 BGGC_MODE_OFF, /* background gc is off */ 1257 BGGC_MODE_SYNC, /* 1258 * background gc is on, migrating blocks 1259 * like foreground gc 1260 */ 1261 }; 1262 1263 enum { 1264 FS_MODE_ADAPTIVE, /* use both lfs/ssr allocation */ 1265 FS_MODE_LFS, /* use lfs allocation only */ 1266 }; 1267 1268 enum { 1269 WHINT_MODE_OFF, /* not pass down write hints */ 1270 WHINT_MODE_USER, /* try to pass down hints given by users */ 1271 WHINT_MODE_FS, /* pass down hints with F2FS policy */ 1272 }; 1273 1274 enum { 1275 ALLOC_MODE_DEFAULT, /* stay default */ 1276 ALLOC_MODE_REUSE, /* reuse segments as much as possible */ 1277 }; 1278 1279 enum fsync_mode { 1280 FSYNC_MODE_POSIX, /* fsync follows posix semantics */ 1281 FSYNC_MODE_STRICT, /* fsync behaves in line with ext4 */ 1282 FSYNC_MODE_NOBARRIER, /* fsync behaves nobarrier based on posix */ 1283 }; 1284 1285 enum { 1286 COMPR_MODE_FS, /* 1287 * automatically compress compression 1288 * enabled files 1289 */ 1290 COMPR_MODE_USER, /* 1291 * automatical compression is disabled. 1292 * user can control the file compression 1293 * using ioctls 1294 */ 1295 }; 1296 1297 static inline int f2fs_test_bit(unsigned int nr, char *addr); 1298 static inline void f2fs_set_bit(unsigned int nr, char *addr); 1299 static inline void f2fs_clear_bit(unsigned int nr, char *addr); 1300 1301 /* 1302 * Layout of f2fs page.private: 1303 * 1304 * Layout A: lowest bit should be 1 1305 * | bit0 = 1 | bit1 | bit2 | ... | bit MAX | private data .... | 1306 * bit 0 PAGE_PRIVATE_NOT_POINTER 1307 * bit 1 PAGE_PRIVATE_ATOMIC_WRITE 1308 * bit 2 PAGE_PRIVATE_DUMMY_WRITE 1309 * bit 3 PAGE_PRIVATE_ONGOING_MIGRATION 1310 * bit 4 PAGE_PRIVATE_INLINE_INODE 1311 * bit 5 PAGE_PRIVATE_REF_RESOURCE 1312 * bit 6- f2fs private data 1313 * 1314 * Layout B: lowest bit should be 0 1315 * page.private is a wrapped pointer. 1316 */ 1317 enum { 1318 PAGE_PRIVATE_NOT_POINTER, /* private contains non-pointer data */ 1319 PAGE_PRIVATE_ATOMIC_WRITE, /* data page from atomic write path */ 1320 PAGE_PRIVATE_DUMMY_WRITE, /* data page for padding aligned IO */ 1321 PAGE_PRIVATE_ONGOING_MIGRATION, /* data page which is on-going migrating */ 1322 PAGE_PRIVATE_INLINE_INODE, /* inode page contains inline data */ 1323 PAGE_PRIVATE_REF_RESOURCE, /* dirty page has referenced resources */ 1324 PAGE_PRIVATE_MAX 1325 }; 1326 1327 #define PAGE_PRIVATE_GET_FUNC(name, flagname) \ 1328 static inline bool page_private_##name(struct page *page) \ 1329 { \ 1330 return test_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)) && \ 1331 test_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \ 1332 } 1333 1334 #define PAGE_PRIVATE_SET_FUNC(name, flagname) \ 1335 static inline void set_page_private_##name(struct page *page) \ 1336 { \ 1337 if (!PagePrivate(page)) { \ 1338 get_page(page); \ 1339 SetPagePrivate(page); \ 1340 } \ 1341 set_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)); \ 1342 set_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \ 1343 } 1344 1345 #define PAGE_PRIVATE_CLEAR_FUNC(name, flagname) \ 1346 static inline void clear_page_private_##name(struct page *page) \ 1347 { \ 1348 clear_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \ 1349 if (page_private(page) == 1 << PAGE_PRIVATE_NOT_POINTER) { \ 1350 set_page_private(page, 0); \ 1351 if (PagePrivate(page)) { \ 1352 ClearPagePrivate(page); \ 1353 put_page(page); \ 1354 }\ 1355 } \ 1356 } 1357 1358 PAGE_PRIVATE_GET_FUNC(nonpointer, NOT_POINTER); 1359 PAGE_PRIVATE_GET_FUNC(reference, REF_RESOURCE); 1360 PAGE_PRIVATE_GET_FUNC(inline, INLINE_INODE); 1361 PAGE_PRIVATE_GET_FUNC(gcing, ONGOING_MIGRATION); 1362 PAGE_PRIVATE_GET_FUNC(atomic, ATOMIC_WRITE); 1363 PAGE_PRIVATE_GET_FUNC(dummy, DUMMY_WRITE); 1364 1365 PAGE_PRIVATE_SET_FUNC(reference, REF_RESOURCE); 1366 PAGE_PRIVATE_SET_FUNC(inline, INLINE_INODE); 1367 PAGE_PRIVATE_SET_FUNC(gcing, ONGOING_MIGRATION); 1368 PAGE_PRIVATE_SET_FUNC(atomic, ATOMIC_WRITE); 1369 PAGE_PRIVATE_SET_FUNC(dummy, DUMMY_WRITE); 1370 1371 PAGE_PRIVATE_CLEAR_FUNC(reference, REF_RESOURCE); 1372 PAGE_PRIVATE_CLEAR_FUNC(inline, INLINE_INODE); 1373 PAGE_PRIVATE_CLEAR_FUNC(gcing, ONGOING_MIGRATION); 1374 PAGE_PRIVATE_CLEAR_FUNC(atomic, ATOMIC_WRITE); 1375 PAGE_PRIVATE_CLEAR_FUNC(dummy, DUMMY_WRITE); 1376 1377 /* For compression */ 1378 enum compress_algorithm_type { 1379 COMPRESS_LZO, 1380 COMPRESS_LZ4, 1381 COMPRESS_ZSTD, 1382 COMPRESS_LZORLE, 1383 COMPRESS_MAX, 1384 }; 1385 1386 enum compress_flag { 1387 COMPRESS_CHKSUM, 1388 COMPRESS_MAX_FLAG, 1389 }; 1390 1391 #define COMPRESS_DATA_RESERVED_SIZE 4 1392 struct compress_data { 1393 __le32 clen; /* compressed data size */ 1394 __le32 chksum; /* compressed data chksum */ 1395 __le32 reserved[COMPRESS_DATA_RESERVED_SIZE]; /* reserved */ 1396 u8 cdata[]; /* compressed data */ 1397 }; 1398 1399 #define COMPRESS_HEADER_SIZE (sizeof(struct compress_data)) 1400 1401 #define F2FS_COMPRESSED_PAGE_MAGIC 0xF5F2C000 1402 1403 #define COMPRESS_LEVEL_OFFSET 8 1404 1405 /* compress context */ 1406 struct compress_ctx { 1407 struct inode *inode; /* inode the context belong to */ 1408 pgoff_t cluster_idx; /* cluster index number */ 1409 unsigned int cluster_size; /* page count in cluster */ 1410 unsigned int log_cluster_size; /* log of cluster size */ 1411 struct page **rpages; /* pages store raw data in cluster */ 1412 unsigned int nr_rpages; /* total page number in rpages */ 1413 struct page **cpages; /* pages store compressed data in cluster */ 1414 unsigned int nr_cpages; /* total page number in cpages */ 1415 void *rbuf; /* virtual mapped address on rpages */ 1416 struct compress_data *cbuf; /* virtual mapped address on cpages */ 1417 size_t rlen; /* valid data length in rbuf */ 1418 size_t clen; /* valid data length in cbuf */ 1419 void *private; /* payload buffer for specified compression algorithm */ 1420 void *private2; /* extra payload buffer */ 1421 }; 1422 1423 /* compress context for write IO path */ 1424 struct compress_io_ctx { 1425 u32 magic; /* magic number to indicate page is compressed */ 1426 struct inode *inode; /* inode the context belong to */ 1427 struct page **rpages; /* pages store raw data in cluster */ 1428 unsigned int nr_rpages; /* total page number in rpages */ 1429 atomic_t pending_pages; /* in-flight compressed page count */ 1430 }; 1431 1432 /* Context for decompressing one cluster on the read IO path */ 1433 struct decompress_io_ctx { 1434 u32 magic; /* magic number to indicate page is compressed */ 1435 struct inode *inode; /* inode the context belong to */ 1436 pgoff_t cluster_idx; /* cluster index number */ 1437 unsigned int cluster_size; /* page count in cluster */ 1438 unsigned int log_cluster_size; /* log of cluster size */ 1439 struct page **rpages; /* pages store raw data in cluster */ 1440 unsigned int nr_rpages; /* total page number in rpages */ 1441 struct page **cpages; /* pages store compressed data in cluster */ 1442 unsigned int nr_cpages; /* total page number in cpages */ 1443 struct page **tpages; /* temp pages to pad holes in cluster */ 1444 void *rbuf; /* virtual mapped address on rpages */ 1445 struct compress_data *cbuf; /* virtual mapped address on cpages */ 1446 size_t rlen; /* valid data length in rbuf */ 1447 size_t clen; /* valid data length in cbuf */ 1448 1449 /* 1450 * The number of compressed pages remaining to be read in this cluster. 1451 * This is initially nr_cpages. It is decremented by 1 each time a page 1452 * has been read (or failed to be read). When it reaches 0, the cluster 1453 * is decompressed (or an error is reported). 1454 * 1455 * If an error occurs before all the pages have been submitted for I/O, 1456 * then this will never reach 0. In this case the I/O submitter is 1457 * responsible for calling f2fs_decompress_end_io() instead. 1458 */ 1459 atomic_t remaining_pages; 1460 1461 /* 1462 * Number of references to this decompress_io_ctx. 1463 * 1464 * One reference is held for I/O completion. This reference is dropped 1465 * after the pagecache pages are updated and unlocked -- either after 1466 * decompression (and verity if enabled), or after an error. 1467 * 1468 * In addition, each compressed page holds a reference while it is in a 1469 * bio. These references are necessary prevent compressed pages from 1470 * being freed while they are still in a bio. 1471 */ 1472 refcount_t refcnt; 1473 1474 bool failed; /* IO error occurred before decompression? */ 1475 bool need_verity; /* need fs-verity verification after decompression? */ 1476 void *private; /* payload buffer for specified decompression algorithm */ 1477 void *private2; /* extra payload buffer */ 1478 struct work_struct verity_work; /* work to verify the decompressed pages */ 1479 }; 1480 1481 #define NULL_CLUSTER ((unsigned int)(~0)) 1482 #define MIN_COMPRESS_LOG_SIZE 2 1483 #define MAX_COMPRESS_LOG_SIZE 8 1484 #define MAX_COMPRESS_WINDOW_SIZE(log_size) ((PAGE_SIZE) << (log_size)) 1485 1486 struct f2fs_sb_info { 1487 struct super_block *sb; /* pointer to VFS super block */ 1488 struct proc_dir_entry *s_proc; /* proc entry */ 1489 struct f2fs_super_block *raw_super; /* raw super block pointer */ 1490 struct rw_semaphore sb_lock; /* lock for raw super block */ 1491 int valid_super_block; /* valid super block no */ 1492 unsigned long s_flag; /* flags for sbi */ 1493 struct mutex writepages; /* mutex for writepages() */ 1494 1495 #ifdef CONFIG_BLK_DEV_ZONED 1496 unsigned int blocks_per_blkz; /* F2FS blocks per zone */ 1497 unsigned int log_blocks_per_blkz; /* log2 F2FS blocks per zone */ 1498 #endif 1499 1500 /* for node-related operations */ 1501 struct f2fs_nm_info *nm_info; /* node manager */ 1502 struct inode *node_inode; /* cache node blocks */ 1503 1504 /* for segment-related operations */ 1505 struct f2fs_sm_info *sm_info; /* segment manager */ 1506 1507 /* for bio operations */ 1508 struct f2fs_bio_info *write_io[NR_PAGE_TYPE]; /* for write bios */ 1509 /* keep migration IO order for LFS mode */ 1510 struct rw_semaphore io_order_lock; 1511 mempool_t *write_io_dummy; /* Dummy pages */ 1512 1513 /* for checkpoint */ 1514 struct f2fs_checkpoint *ckpt; /* raw checkpoint pointer */ 1515 int cur_cp_pack; /* remain current cp pack */ 1516 spinlock_t cp_lock; /* for flag in ckpt */ 1517 struct inode *meta_inode; /* cache meta blocks */ 1518 struct rw_semaphore cp_global_sem; /* checkpoint procedure lock */ 1519 struct rw_semaphore cp_rwsem; /* blocking FS operations */ 1520 struct rw_semaphore node_write; /* locking node writes */ 1521 struct rw_semaphore node_change; /* locking node change */ 1522 wait_queue_head_t cp_wait; 1523 unsigned long last_time[MAX_TIME]; /* to store time in jiffies */ 1524 long interval_time[MAX_TIME]; /* to store thresholds */ 1525 struct ckpt_req_control cprc_info; /* for checkpoint request control */ 1526 1527 struct inode_management im[MAX_INO_ENTRY]; /* manage inode cache */ 1528 1529 spinlock_t fsync_node_lock; /* for node entry lock */ 1530 struct list_head fsync_node_list; /* node list head */ 1531 unsigned int fsync_seg_id; /* sequence id */ 1532 unsigned int fsync_node_num; /* number of node entries */ 1533 1534 /* for orphan inode, use 0'th array */ 1535 unsigned int max_orphans; /* max orphan inodes */ 1536 1537 /* for inode management */ 1538 struct list_head inode_list[NR_INODE_TYPE]; /* dirty inode list */ 1539 spinlock_t inode_lock[NR_INODE_TYPE]; /* for dirty inode list lock */ 1540 struct mutex flush_lock; /* for flush exclusion */ 1541 1542 /* for extent tree cache */ 1543 struct radix_tree_root extent_tree_root;/* cache extent cache entries */ 1544 struct mutex extent_tree_lock; /* locking extent radix tree */ 1545 struct list_head extent_list; /* lru list for shrinker */ 1546 spinlock_t extent_lock; /* locking extent lru list */ 1547 atomic_t total_ext_tree; /* extent tree count */ 1548 struct list_head zombie_list; /* extent zombie tree list */ 1549 atomic_t total_zombie_tree; /* extent zombie tree count */ 1550 atomic_t total_ext_node; /* extent info count */ 1551 1552 /* basic filesystem units */ 1553 unsigned int log_sectors_per_block; /* log2 sectors per block */ 1554 unsigned int log_blocksize; /* log2 block size */ 1555 unsigned int blocksize; /* block size */ 1556 unsigned int root_ino_num; /* root inode number*/ 1557 unsigned int node_ino_num; /* node inode number*/ 1558 unsigned int meta_ino_num; /* meta inode number*/ 1559 unsigned int log_blocks_per_seg; /* log2 blocks per segment */ 1560 unsigned int blocks_per_seg; /* blocks per segment */ 1561 unsigned int segs_per_sec; /* segments per section */ 1562 unsigned int secs_per_zone; /* sections per zone */ 1563 unsigned int total_sections; /* total section count */ 1564 unsigned int total_node_count; /* total node block count */ 1565 unsigned int total_valid_node_count; /* valid node block count */ 1566 int dir_level; /* directory level */ 1567 int readdir_ra; /* readahead inode in readdir */ 1568 u64 max_io_bytes; /* max io bytes to merge IOs */ 1569 1570 block_t user_block_count; /* # of user blocks */ 1571 block_t total_valid_block_count; /* # of valid blocks */ 1572 block_t discard_blks; /* discard command candidats */ 1573 block_t last_valid_block_count; /* for recovery */ 1574 block_t reserved_blocks; /* configurable reserved blocks */ 1575 block_t current_reserved_blocks; /* current reserved blocks */ 1576 1577 /* Additional tracking for no checkpoint mode */ 1578 block_t unusable_block_count; /* # of blocks saved by last cp */ 1579 1580 unsigned int nquota_files; /* # of quota sysfile */ 1581 struct rw_semaphore quota_sem; /* blocking cp for flags */ 1582 1583 /* # of pages, see count_type */ 1584 atomic_t nr_pages[NR_COUNT_TYPE]; 1585 /* # of allocated blocks */ 1586 struct percpu_counter alloc_valid_block_count; 1587 1588 /* writeback control */ 1589 atomic_t wb_sync_req[META]; /* count # of WB_SYNC threads */ 1590 1591 /* valid inode count */ 1592 struct percpu_counter total_valid_inode_count; 1593 1594 struct f2fs_mount_info mount_opt; /* mount options */ 1595 1596 /* for cleaning operations */ 1597 struct rw_semaphore gc_lock; /* 1598 * semaphore for GC, avoid 1599 * race between GC and GC or CP 1600 */ 1601 struct f2fs_gc_kthread *gc_thread; /* GC thread */ 1602 struct atgc_management am; /* atgc management */ 1603 unsigned int cur_victim_sec; /* current victim section num */ 1604 unsigned int gc_mode; /* current GC state */ 1605 unsigned int next_victim_seg[2]; /* next segment in victim section */ 1606 1607 /* for skip statistic */ 1608 unsigned int atomic_files; /* # of opened atomic file */ 1609 unsigned long long skipped_atomic_files[2]; /* FG_GC and BG_GC */ 1610 unsigned long long skipped_gc_rwsem; /* FG_GC only */ 1611 1612 /* threshold for gc trials on pinned files */ 1613 u64 gc_pin_file_threshold; 1614 struct rw_semaphore pin_sem; 1615 1616 /* maximum # of trials to find a victim segment for SSR and GC */ 1617 unsigned int max_victim_search; 1618 /* migration granularity of garbage collection, unit: segment */ 1619 unsigned int migration_granularity; 1620 1621 /* 1622 * for stat information. 1623 * one is for the LFS mode, and the other is for the SSR mode. 1624 */ 1625 #ifdef CONFIG_F2FS_STAT_FS 1626 struct f2fs_stat_info *stat_info; /* FS status information */ 1627 atomic_t meta_count[META_MAX]; /* # of meta blocks */ 1628 unsigned int segment_count[2]; /* # of allocated segments */ 1629 unsigned int block_count[2]; /* # of allocated blocks */ 1630 atomic_t inplace_count; /* # of inplace update */ 1631 atomic64_t total_hit_ext; /* # of lookup extent cache */ 1632 atomic64_t read_hit_rbtree; /* # of hit rbtree extent node */ 1633 atomic64_t read_hit_largest; /* # of hit largest extent node */ 1634 atomic64_t read_hit_cached; /* # of hit cached extent node */ 1635 atomic_t inline_xattr; /* # of inline_xattr inodes */ 1636 atomic_t inline_inode; /* # of inline_data inodes */ 1637 atomic_t inline_dir; /* # of inline_dentry inodes */ 1638 atomic_t compr_inode; /* # of compressed inodes */ 1639 atomic64_t compr_blocks; /* # of compressed blocks */ 1640 atomic_t vw_cnt; /* # of volatile writes */ 1641 atomic_t max_aw_cnt; /* max # of atomic writes */ 1642 atomic_t max_vw_cnt; /* max # of volatile writes */ 1643 unsigned int io_skip_bggc; /* skip background gc for in-flight IO */ 1644 unsigned int other_skip_bggc; /* skip background gc for other reasons */ 1645 unsigned int ndirty_inode[NR_INODE_TYPE]; /* # of dirty inodes */ 1646 #endif 1647 spinlock_t stat_lock; /* lock for stat operations */ 1648 1649 /* For app/fs IO statistics */ 1650 spinlock_t iostat_lock; 1651 unsigned long long rw_iostat[NR_IO_TYPE]; 1652 unsigned long long prev_rw_iostat[NR_IO_TYPE]; 1653 bool iostat_enable; 1654 unsigned long iostat_next_period; 1655 unsigned int iostat_period_ms; 1656 1657 /* to attach REQ_META|REQ_FUA flags */ 1658 unsigned int data_io_flag; 1659 unsigned int node_io_flag; 1660 1661 /* For sysfs suppport */ 1662 struct kobject s_kobj; /* /sys/fs/f2fs/<devname> */ 1663 struct completion s_kobj_unregister; 1664 1665 struct kobject s_stat_kobj; /* /sys/fs/f2fs/<devname>/stat */ 1666 struct completion s_stat_kobj_unregister; 1667 1668 /* For shrinker support */ 1669 struct list_head s_list; 1670 int s_ndevs; /* number of devices */ 1671 struct f2fs_dev_info *devs; /* for device list */ 1672 unsigned int dirty_device; /* for checkpoint data flush */ 1673 spinlock_t dev_lock; /* protect dirty_device */ 1674 struct mutex umount_mutex; 1675 unsigned int shrinker_run_no; 1676 1677 /* For write statistics */ 1678 u64 sectors_written_start; 1679 u64 kbytes_written; 1680 1681 /* Reference to checksum algorithm driver via cryptoapi */ 1682 struct crypto_shash *s_chksum_driver; 1683 1684 /* Precomputed FS UUID checksum for seeding other checksums */ 1685 __u32 s_chksum_seed; 1686 1687 struct workqueue_struct *post_read_wq; /* post read workqueue */ 1688 1689 struct kmem_cache *inline_xattr_slab; /* inline xattr entry */ 1690 unsigned int inline_xattr_slab_size; /* default inline xattr slab size */ 1691 1692 #ifdef CONFIG_F2FS_FS_COMPRESSION 1693 struct kmem_cache *page_array_slab; /* page array entry */ 1694 unsigned int page_array_slab_size; /* default page array slab size */ 1695 1696 /* For runtime compression statistics */ 1697 u64 compr_written_block; 1698 u64 compr_saved_block; 1699 u32 compr_new_inode; 1700 #endif 1701 }; 1702 1703 struct f2fs_private_dio { 1704 struct inode *inode; 1705 void *orig_private; 1706 bio_end_io_t *orig_end_io; 1707 bool write; 1708 }; 1709 1710 #ifdef CONFIG_F2FS_FAULT_INJECTION 1711 #define f2fs_show_injection_info(sbi, type) \ 1712 printk_ratelimited("%sF2FS-fs (%s) : inject %s in %s of %pS\n", \ 1713 KERN_INFO, sbi->sb->s_id, \ 1714 f2fs_fault_name[type], \ 1715 __func__, __builtin_return_address(0)) 1716 static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type) 1717 { 1718 struct f2fs_fault_info *ffi = &F2FS_OPTION(sbi).fault_info; 1719 1720 if (!ffi->inject_rate) 1721 return false; 1722 1723 if (!IS_FAULT_SET(ffi, type)) 1724 return false; 1725 1726 atomic_inc(&ffi->inject_ops); 1727 if (atomic_read(&ffi->inject_ops) >= ffi->inject_rate) { 1728 atomic_set(&ffi->inject_ops, 0); 1729 return true; 1730 } 1731 return false; 1732 } 1733 #else 1734 #define f2fs_show_injection_info(sbi, type) do { } while (0) 1735 static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type) 1736 { 1737 return false; 1738 } 1739 #endif 1740 1741 /* 1742 * Test if the mounted volume is a multi-device volume. 1743 * - For a single regular disk volume, sbi->s_ndevs is 0. 1744 * - For a single zoned disk volume, sbi->s_ndevs is 1. 1745 * - For a multi-device volume, sbi->s_ndevs is always 2 or more. 1746 */ 1747 static inline bool f2fs_is_multi_device(struct f2fs_sb_info *sbi) 1748 { 1749 return sbi->s_ndevs > 1; 1750 } 1751 1752 static inline void f2fs_update_time(struct f2fs_sb_info *sbi, int type) 1753 { 1754 unsigned long now = jiffies; 1755 1756 sbi->last_time[type] = now; 1757 1758 /* DISCARD_TIME and GC_TIME are based on REQ_TIME */ 1759 if (type == REQ_TIME) { 1760 sbi->last_time[DISCARD_TIME] = now; 1761 sbi->last_time[GC_TIME] = now; 1762 } 1763 } 1764 1765 static inline bool f2fs_time_over(struct f2fs_sb_info *sbi, int type) 1766 { 1767 unsigned long interval = sbi->interval_time[type] * HZ; 1768 1769 return time_after(jiffies, sbi->last_time[type] + interval); 1770 } 1771 1772 static inline unsigned int f2fs_time_to_wait(struct f2fs_sb_info *sbi, 1773 int type) 1774 { 1775 unsigned long interval = sbi->interval_time[type] * HZ; 1776 unsigned int wait_ms = 0; 1777 long delta; 1778 1779 delta = (sbi->last_time[type] + interval) - jiffies; 1780 if (delta > 0) 1781 wait_ms = jiffies_to_msecs(delta); 1782 1783 return wait_ms; 1784 } 1785 1786 /* 1787 * Inline functions 1788 */ 1789 static inline u32 __f2fs_crc32(struct f2fs_sb_info *sbi, u32 crc, 1790 const void *address, unsigned int length) 1791 { 1792 struct { 1793 struct shash_desc shash; 1794 char ctx[4]; 1795 } desc; 1796 int err; 1797 1798 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver) != sizeof(desc.ctx)); 1799 1800 desc.shash.tfm = sbi->s_chksum_driver; 1801 *(u32 *)desc.ctx = crc; 1802 1803 err = crypto_shash_update(&desc.shash, address, length); 1804 BUG_ON(err); 1805 1806 return *(u32 *)desc.ctx; 1807 } 1808 1809 static inline u32 f2fs_crc32(struct f2fs_sb_info *sbi, const void *address, 1810 unsigned int length) 1811 { 1812 return __f2fs_crc32(sbi, F2FS_SUPER_MAGIC, address, length); 1813 } 1814 1815 static inline bool f2fs_crc_valid(struct f2fs_sb_info *sbi, __u32 blk_crc, 1816 void *buf, size_t buf_size) 1817 { 1818 return f2fs_crc32(sbi, buf, buf_size) == blk_crc; 1819 } 1820 1821 static inline u32 f2fs_chksum(struct f2fs_sb_info *sbi, u32 crc, 1822 const void *address, unsigned int length) 1823 { 1824 return __f2fs_crc32(sbi, crc, address, length); 1825 } 1826 1827 static inline struct f2fs_inode_info *F2FS_I(struct inode *inode) 1828 { 1829 return container_of(inode, struct f2fs_inode_info, vfs_inode); 1830 } 1831 1832 static inline struct f2fs_sb_info *F2FS_SB(struct super_block *sb) 1833 { 1834 return sb->s_fs_info; 1835 } 1836 1837 static inline struct f2fs_sb_info *F2FS_I_SB(struct inode *inode) 1838 { 1839 return F2FS_SB(inode->i_sb); 1840 } 1841 1842 static inline struct f2fs_sb_info *F2FS_M_SB(struct address_space *mapping) 1843 { 1844 return F2FS_I_SB(mapping->host); 1845 } 1846 1847 static inline struct f2fs_sb_info *F2FS_P_SB(struct page *page) 1848 { 1849 return F2FS_M_SB(page_file_mapping(page)); 1850 } 1851 1852 static inline struct f2fs_super_block *F2FS_RAW_SUPER(struct f2fs_sb_info *sbi) 1853 { 1854 return (struct f2fs_super_block *)(sbi->raw_super); 1855 } 1856 1857 static inline struct f2fs_checkpoint *F2FS_CKPT(struct f2fs_sb_info *sbi) 1858 { 1859 return (struct f2fs_checkpoint *)(sbi->ckpt); 1860 } 1861 1862 static inline struct f2fs_node *F2FS_NODE(struct page *page) 1863 { 1864 return (struct f2fs_node *)page_address(page); 1865 } 1866 1867 static inline struct f2fs_inode *F2FS_INODE(struct page *page) 1868 { 1869 return &((struct f2fs_node *)page_address(page))->i; 1870 } 1871 1872 static inline struct f2fs_nm_info *NM_I(struct f2fs_sb_info *sbi) 1873 { 1874 return (struct f2fs_nm_info *)(sbi->nm_info); 1875 } 1876 1877 static inline struct f2fs_sm_info *SM_I(struct f2fs_sb_info *sbi) 1878 { 1879 return (struct f2fs_sm_info *)(sbi->sm_info); 1880 } 1881 1882 static inline struct sit_info *SIT_I(struct f2fs_sb_info *sbi) 1883 { 1884 return (struct sit_info *)(SM_I(sbi)->sit_info); 1885 } 1886 1887 static inline struct free_segmap_info *FREE_I(struct f2fs_sb_info *sbi) 1888 { 1889 return (struct free_segmap_info *)(SM_I(sbi)->free_info); 1890 } 1891 1892 static inline struct dirty_seglist_info *DIRTY_I(struct f2fs_sb_info *sbi) 1893 { 1894 return (struct dirty_seglist_info *)(SM_I(sbi)->dirty_info); 1895 } 1896 1897 static inline struct address_space *META_MAPPING(struct f2fs_sb_info *sbi) 1898 { 1899 return sbi->meta_inode->i_mapping; 1900 } 1901 1902 static inline struct address_space *NODE_MAPPING(struct f2fs_sb_info *sbi) 1903 { 1904 return sbi->node_inode->i_mapping; 1905 } 1906 1907 static inline bool is_sbi_flag_set(struct f2fs_sb_info *sbi, unsigned int type) 1908 { 1909 return test_bit(type, &sbi->s_flag); 1910 } 1911 1912 static inline void set_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type) 1913 { 1914 set_bit(type, &sbi->s_flag); 1915 } 1916 1917 static inline void clear_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type) 1918 { 1919 clear_bit(type, &sbi->s_flag); 1920 } 1921 1922 static inline unsigned long long cur_cp_version(struct f2fs_checkpoint *cp) 1923 { 1924 return le64_to_cpu(cp->checkpoint_ver); 1925 } 1926 1927 static inline unsigned long f2fs_qf_ino(struct super_block *sb, int type) 1928 { 1929 if (type < F2FS_MAX_QUOTAS) 1930 return le32_to_cpu(F2FS_SB(sb)->raw_super->qf_ino[type]); 1931 return 0; 1932 } 1933 1934 static inline __u64 cur_cp_crc(struct f2fs_checkpoint *cp) 1935 { 1936 size_t crc_offset = le32_to_cpu(cp->checksum_offset); 1937 return le32_to_cpu(*((__le32 *)((unsigned char *)cp + crc_offset))); 1938 } 1939 1940 static inline bool __is_set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 1941 { 1942 unsigned int ckpt_flags = le32_to_cpu(cp->ckpt_flags); 1943 1944 return ckpt_flags & f; 1945 } 1946 1947 static inline bool is_set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 1948 { 1949 return __is_set_ckpt_flags(F2FS_CKPT(sbi), f); 1950 } 1951 1952 static inline void __set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 1953 { 1954 unsigned int ckpt_flags; 1955 1956 ckpt_flags = le32_to_cpu(cp->ckpt_flags); 1957 ckpt_flags |= f; 1958 cp->ckpt_flags = cpu_to_le32(ckpt_flags); 1959 } 1960 1961 static inline void set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 1962 { 1963 unsigned long flags; 1964 1965 spin_lock_irqsave(&sbi->cp_lock, flags); 1966 __set_ckpt_flags(F2FS_CKPT(sbi), f); 1967 spin_unlock_irqrestore(&sbi->cp_lock, flags); 1968 } 1969 1970 static inline void __clear_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 1971 { 1972 unsigned int ckpt_flags; 1973 1974 ckpt_flags = le32_to_cpu(cp->ckpt_flags); 1975 ckpt_flags &= (~f); 1976 cp->ckpt_flags = cpu_to_le32(ckpt_flags); 1977 } 1978 1979 static inline void clear_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 1980 { 1981 unsigned long flags; 1982 1983 spin_lock_irqsave(&sbi->cp_lock, flags); 1984 __clear_ckpt_flags(F2FS_CKPT(sbi), f); 1985 spin_unlock_irqrestore(&sbi->cp_lock, flags); 1986 } 1987 1988 static inline void disable_nat_bits(struct f2fs_sb_info *sbi, bool lock) 1989 { 1990 unsigned long flags; 1991 unsigned char *nat_bits; 1992 1993 /* 1994 * In order to re-enable nat_bits we need to call fsck.f2fs by 1995 * set_sbi_flag(sbi, SBI_NEED_FSCK). But it may give huge cost, 1996 * so let's rely on regular fsck or unclean shutdown. 1997 */ 1998 1999 if (lock) 2000 spin_lock_irqsave(&sbi->cp_lock, flags); 2001 __clear_ckpt_flags(F2FS_CKPT(sbi), CP_NAT_BITS_FLAG); 2002 nat_bits = NM_I(sbi)->nat_bits; 2003 NM_I(sbi)->nat_bits = NULL; 2004 if (lock) 2005 spin_unlock_irqrestore(&sbi->cp_lock, flags); 2006 2007 kvfree(nat_bits); 2008 } 2009 2010 static inline bool enabled_nat_bits(struct f2fs_sb_info *sbi, 2011 struct cp_control *cpc) 2012 { 2013 bool set = is_set_ckpt_flags(sbi, CP_NAT_BITS_FLAG); 2014 2015 return (cpc) ? (cpc->reason & CP_UMOUNT) && set : set; 2016 } 2017 2018 static inline void f2fs_lock_op(struct f2fs_sb_info *sbi) 2019 { 2020 down_read(&sbi->cp_rwsem); 2021 } 2022 2023 static inline int f2fs_trylock_op(struct f2fs_sb_info *sbi) 2024 { 2025 return down_read_trylock(&sbi->cp_rwsem); 2026 } 2027 2028 static inline void f2fs_unlock_op(struct f2fs_sb_info *sbi) 2029 { 2030 up_read(&sbi->cp_rwsem); 2031 } 2032 2033 static inline void f2fs_lock_all(struct f2fs_sb_info *sbi) 2034 { 2035 down_write(&sbi->cp_rwsem); 2036 } 2037 2038 static inline void f2fs_unlock_all(struct f2fs_sb_info *sbi) 2039 { 2040 up_write(&sbi->cp_rwsem); 2041 } 2042 2043 static inline int __get_cp_reason(struct f2fs_sb_info *sbi) 2044 { 2045 int reason = CP_SYNC; 2046 2047 if (test_opt(sbi, FASTBOOT)) 2048 reason = CP_FASTBOOT; 2049 if (is_sbi_flag_set(sbi, SBI_IS_CLOSE)) 2050 reason = CP_UMOUNT; 2051 return reason; 2052 } 2053 2054 static inline bool __remain_node_summaries(int reason) 2055 { 2056 return (reason & (CP_UMOUNT | CP_FASTBOOT)); 2057 } 2058 2059 static inline bool __exist_node_summaries(struct f2fs_sb_info *sbi) 2060 { 2061 return (is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG) || 2062 is_set_ckpt_flags(sbi, CP_FASTBOOT_FLAG)); 2063 } 2064 2065 /* 2066 * Check whether the inode has blocks or not 2067 */ 2068 static inline int F2FS_HAS_BLOCKS(struct inode *inode) 2069 { 2070 block_t xattr_block = F2FS_I(inode)->i_xattr_nid ? 1 : 0; 2071 2072 return (inode->i_blocks >> F2FS_LOG_SECTORS_PER_BLOCK) > xattr_block; 2073 } 2074 2075 static inline bool f2fs_has_xattr_block(unsigned int ofs) 2076 { 2077 return ofs == XATTR_NODE_OFFSET; 2078 } 2079 2080 static inline bool __allow_reserved_blocks(struct f2fs_sb_info *sbi, 2081 struct inode *inode, bool cap) 2082 { 2083 if (!inode) 2084 return true; 2085 if (!test_opt(sbi, RESERVE_ROOT)) 2086 return false; 2087 if (IS_NOQUOTA(inode)) 2088 return true; 2089 if (uid_eq(F2FS_OPTION(sbi).s_resuid, current_fsuid())) 2090 return true; 2091 if (!gid_eq(F2FS_OPTION(sbi).s_resgid, GLOBAL_ROOT_GID) && 2092 in_group_p(F2FS_OPTION(sbi).s_resgid)) 2093 return true; 2094 if (cap && capable(CAP_SYS_RESOURCE)) 2095 return true; 2096 return false; 2097 } 2098 2099 static inline void f2fs_i_blocks_write(struct inode *, block_t, bool, bool); 2100 static inline int inc_valid_block_count(struct f2fs_sb_info *sbi, 2101 struct inode *inode, blkcnt_t *count) 2102 { 2103 blkcnt_t diff = 0, release = 0; 2104 block_t avail_user_block_count; 2105 int ret; 2106 2107 ret = dquot_reserve_block(inode, *count); 2108 if (ret) 2109 return ret; 2110 2111 if (time_to_inject(sbi, FAULT_BLOCK)) { 2112 f2fs_show_injection_info(sbi, FAULT_BLOCK); 2113 release = *count; 2114 goto release_quota; 2115 } 2116 2117 /* 2118 * let's increase this in prior to actual block count change in order 2119 * for f2fs_sync_file to avoid data races when deciding checkpoint. 2120 */ 2121 percpu_counter_add(&sbi->alloc_valid_block_count, (*count)); 2122 2123 spin_lock(&sbi->stat_lock); 2124 sbi->total_valid_block_count += (block_t)(*count); 2125 avail_user_block_count = sbi->user_block_count - 2126 sbi->current_reserved_blocks; 2127 2128 if (!__allow_reserved_blocks(sbi, inode, true)) 2129 avail_user_block_count -= F2FS_OPTION(sbi).root_reserved_blocks; 2130 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) { 2131 if (avail_user_block_count > sbi->unusable_block_count) 2132 avail_user_block_count -= sbi->unusable_block_count; 2133 else 2134 avail_user_block_count = 0; 2135 } 2136 if (unlikely(sbi->total_valid_block_count > avail_user_block_count)) { 2137 diff = sbi->total_valid_block_count - avail_user_block_count; 2138 if (diff > *count) 2139 diff = *count; 2140 *count -= diff; 2141 release = diff; 2142 sbi->total_valid_block_count -= diff; 2143 if (!*count) { 2144 spin_unlock(&sbi->stat_lock); 2145 goto enospc; 2146 } 2147 } 2148 spin_unlock(&sbi->stat_lock); 2149 2150 if (unlikely(release)) { 2151 percpu_counter_sub(&sbi->alloc_valid_block_count, release); 2152 dquot_release_reservation_block(inode, release); 2153 } 2154 f2fs_i_blocks_write(inode, *count, true, true); 2155 return 0; 2156 2157 enospc: 2158 percpu_counter_sub(&sbi->alloc_valid_block_count, release); 2159 release_quota: 2160 dquot_release_reservation_block(inode, release); 2161 return -ENOSPC; 2162 } 2163 2164 __printf(2, 3) 2165 void f2fs_printk(struct f2fs_sb_info *sbi, const char *fmt, ...); 2166 2167 #define f2fs_err(sbi, fmt, ...) \ 2168 f2fs_printk(sbi, KERN_ERR fmt, ##__VA_ARGS__) 2169 #define f2fs_warn(sbi, fmt, ...) \ 2170 f2fs_printk(sbi, KERN_WARNING fmt, ##__VA_ARGS__) 2171 #define f2fs_notice(sbi, fmt, ...) \ 2172 f2fs_printk(sbi, KERN_NOTICE fmt, ##__VA_ARGS__) 2173 #define f2fs_info(sbi, fmt, ...) \ 2174 f2fs_printk(sbi, KERN_INFO fmt, ##__VA_ARGS__) 2175 #define f2fs_debug(sbi, fmt, ...) \ 2176 f2fs_printk(sbi, KERN_DEBUG fmt, ##__VA_ARGS__) 2177 2178 static inline void dec_valid_block_count(struct f2fs_sb_info *sbi, 2179 struct inode *inode, 2180 block_t count) 2181 { 2182 blkcnt_t sectors = count << F2FS_LOG_SECTORS_PER_BLOCK; 2183 2184 spin_lock(&sbi->stat_lock); 2185 f2fs_bug_on(sbi, sbi->total_valid_block_count < (block_t) count); 2186 sbi->total_valid_block_count -= (block_t)count; 2187 if (sbi->reserved_blocks && 2188 sbi->current_reserved_blocks < sbi->reserved_blocks) 2189 sbi->current_reserved_blocks = min(sbi->reserved_blocks, 2190 sbi->current_reserved_blocks + count); 2191 spin_unlock(&sbi->stat_lock); 2192 if (unlikely(inode->i_blocks < sectors)) { 2193 f2fs_warn(sbi, "Inconsistent i_blocks, ino:%lu, iblocks:%llu, sectors:%llu", 2194 inode->i_ino, 2195 (unsigned long long)inode->i_blocks, 2196 (unsigned long long)sectors); 2197 set_sbi_flag(sbi, SBI_NEED_FSCK); 2198 return; 2199 } 2200 f2fs_i_blocks_write(inode, count, false, true); 2201 } 2202 2203 static inline void inc_page_count(struct f2fs_sb_info *sbi, int count_type) 2204 { 2205 atomic_inc(&sbi->nr_pages[count_type]); 2206 2207 if (count_type == F2FS_DIRTY_DENTS || 2208 count_type == F2FS_DIRTY_NODES || 2209 count_type == F2FS_DIRTY_META || 2210 count_type == F2FS_DIRTY_QDATA || 2211 count_type == F2FS_DIRTY_IMETA) 2212 set_sbi_flag(sbi, SBI_IS_DIRTY); 2213 } 2214 2215 static inline void inode_inc_dirty_pages(struct inode *inode) 2216 { 2217 atomic_inc(&F2FS_I(inode)->dirty_pages); 2218 inc_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ? 2219 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA); 2220 if (IS_NOQUOTA(inode)) 2221 inc_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA); 2222 } 2223 2224 static inline void dec_page_count(struct f2fs_sb_info *sbi, int count_type) 2225 { 2226 atomic_dec(&sbi->nr_pages[count_type]); 2227 } 2228 2229 static inline void inode_dec_dirty_pages(struct inode *inode) 2230 { 2231 if (!S_ISDIR(inode->i_mode) && !S_ISREG(inode->i_mode) && 2232 !S_ISLNK(inode->i_mode)) 2233 return; 2234 2235 atomic_dec(&F2FS_I(inode)->dirty_pages); 2236 dec_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ? 2237 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA); 2238 if (IS_NOQUOTA(inode)) 2239 dec_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA); 2240 } 2241 2242 static inline s64 get_pages(struct f2fs_sb_info *sbi, int count_type) 2243 { 2244 return atomic_read(&sbi->nr_pages[count_type]); 2245 } 2246 2247 static inline int get_dirty_pages(struct inode *inode) 2248 { 2249 return atomic_read(&F2FS_I(inode)->dirty_pages); 2250 } 2251 2252 static inline int get_blocktype_secs(struct f2fs_sb_info *sbi, int block_type) 2253 { 2254 unsigned int pages_per_sec = sbi->segs_per_sec * sbi->blocks_per_seg; 2255 unsigned int segs = (get_pages(sbi, block_type) + pages_per_sec - 1) >> 2256 sbi->log_blocks_per_seg; 2257 2258 return segs / sbi->segs_per_sec; 2259 } 2260 2261 static inline block_t valid_user_blocks(struct f2fs_sb_info *sbi) 2262 { 2263 return sbi->total_valid_block_count; 2264 } 2265 2266 static inline block_t discard_blocks(struct f2fs_sb_info *sbi) 2267 { 2268 return sbi->discard_blks; 2269 } 2270 2271 static inline unsigned long __bitmap_size(struct f2fs_sb_info *sbi, int flag) 2272 { 2273 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi); 2274 2275 /* return NAT or SIT bitmap */ 2276 if (flag == NAT_BITMAP) 2277 return le32_to_cpu(ckpt->nat_ver_bitmap_bytesize); 2278 else if (flag == SIT_BITMAP) 2279 return le32_to_cpu(ckpt->sit_ver_bitmap_bytesize); 2280 2281 return 0; 2282 } 2283 2284 static inline block_t __cp_payload(struct f2fs_sb_info *sbi) 2285 { 2286 return le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_payload); 2287 } 2288 2289 static inline void *__bitmap_ptr(struct f2fs_sb_info *sbi, int flag) 2290 { 2291 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi); 2292 void *tmp_ptr = &ckpt->sit_nat_version_bitmap; 2293 int offset; 2294 2295 if (is_set_ckpt_flags(sbi, CP_LARGE_NAT_BITMAP_FLAG)) { 2296 offset = (flag == SIT_BITMAP) ? 2297 le32_to_cpu(ckpt->nat_ver_bitmap_bytesize) : 0; 2298 /* 2299 * if large_nat_bitmap feature is enabled, leave checksum 2300 * protection for all nat/sit bitmaps. 2301 */ 2302 return tmp_ptr + offset + sizeof(__le32); 2303 } 2304 2305 if (__cp_payload(sbi) > 0) { 2306 if (flag == NAT_BITMAP) 2307 return &ckpt->sit_nat_version_bitmap; 2308 else 2309 return (unsigned char *)ckpt + F2FS_BLKSIZE; 2310 } else { 2311 offset = (flag == NAT_BITMAP) ? 2312 le32_to_cpu(ckpt->sit_ver_bitmap_bytesize) : 0; 2313 return tmp_ptr + offset; 2314 } 2315 } 2316 2317 static inline block_t __start_cp_addr(struct f2fs_sb_info *sbi) 2318 { 2319 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr); 2320 2321 if (sbi->cur_cp_pack == 2) 2322 start_addr += sbi->blocks_per_seg; 2323 return start_addr; 2324 } 2325 2326 static inline block_t __start_cp_next_addr(struct f2fs_sb_info *sbi) 2327 { 2328 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr); 2329 2330 if (sbi->cur_cp_pack == 1) 2331 start_addr += sbi->blocks_per_seg; 2332 return start_addr; 2333 } 2334 2335 static inline void __set_cp_next_pack(struct f2fs_sb_info *sbi) 2336 { 2337 sbi->cur_cp_pack = (sbi->cur_cp_pack == 1) ? 2 : 1; 2338 } 2339 2340 static inline block_t __start_sum_addr(struct f2fs_sb_info *sbi) 2341 { 2342 return le32_to_cpu(F2FS_CKPT(sbi)->cp_pack_start_sum); 2343 } 2344 2345 static inline int inc_valid_node_count(struct f2fs_sb_info *sbi, 2346 struct inode *inode, bool is_inode) 2347 { 2348 block_t valid_block_count; 2349 unsigned int valid_node_count, user_block_count; 2350 int err; 2351 2352 if (is_inode) { 2353 if (inode) { 2354 err = dquot_alloc_inode(inode); 2355 if (err) 2356 return err; 2357 } 2358 } else { 2359 err = dquot_reserve_block(inode, 1); 2360 if (err) 2361 return err; 2362 } 2363 2364 if (time_to_inject(sbi, FAULT_BLOCK)) { 2365 f2fs_show_injection_info(sbi, FAULT_BLOCK); 2366 goto enospc; 2367 } 2368 2369 spin_lock(&sbi->stat_lock); 2370 2371 valid_block_count = sbi->total_valid_block_count + 2372 sbi->current_reserved_blocks + 1; 2373 2374 if (!__allow_reserved_blocks(sbi, inode, false)) 2375 valid_block_count += F2FS_OPTION(sbi).root_reserved_blocks; 2376 user_block_count = sbi->user_block_count; 2377 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) 2378 user_block_count -= sbi->unusable_block_count; 2379 2380 if (unlikely(valid_block_count > user_block_count)) { 2381 spin_unlock(&sbi->stat_lock); 2382 goto enospc; 2383 } 2384 2385 valid_node_count = sbi->total_valid_node_count + 1; 2386 if (unlikely(valid_node_count > sbi->total_node_count)) { 2387 spin_unlock(&sbi->stat_lock); 2388 goto enospc; 2389 } 2390 2391 sbi->total_valid_node_count++; 2392 sbi->total_valid_block_count++; 2393 spin_unlock(&sbi->stat_lock); 2394 2395 if (inode) { 2396 if (is_inode) 2397 f2fs_mark_inode_dirty_sync(inode, true); 2398 else 2399 f2fs_i_blocks_write(inode, 1, true, true); 2400 } 2401 2402 percpu_counter_inc(&sbi->alloc_valid_block_count); 2403 return 0; 2404 2405 enospc: 2406 if (is_inode) { 2407 if (inode) 2408 dquot_free_inode(inode); 2409 } else { 2410 dquot_release_reservation_block(inode, 1); 2411 } 2412 return -ENOSPC; 2413 } 2414 2415 static inline void dec_valid_node_count(struct f2fs_sb_info *sbi, 2416 struct inode *inode, bool is_inode) 2417 { 2418 spin_lock(&sbi->stat_lock); 2419 2420 f2fs_bug_on(sbi, !sbi->total_valid_block_count); 2421 f2fs_bug_on(sbi, !sbi->total_valid_node_count); 2422 2423 sbi->total_valid_node_count--; 2424 sbi->total_valid_block_count--; 2425 if (sbi->reserved_blocks && 2426 sbi->current_reserved_blocks < sbi->reserved_blocks) 2427 sbi->current_reserved_blocks++; 2428 2429 spin_unlock(&sbi->stat_lock); 2430 2431 if (is_inode) { 2432 dquot_free_inode(inode); 2433 } else { 2434 if (unlikely(inode->i_blocks == 0)) { 2435 f2fs_warn(sbi, "dec_valid_node_count: inconsistent i_blocks, ino:%lu, iblocks:%llu", 2436 inode->i_ino, 2437 (unsigned long long)inode->i_blocks); 2438 set_sbi_flag(sbi, SBI_NEED_FSCK); 2439 return; 2440 } 2441 f2fs_i_blocks_write(inode, 1, false, true); 2442 } 2443 } 2444 2445 static inline unsigned int valid_node_count(struct f2fs_sb_info *sbi) 2446 { 2447 return sbi->total_valid_node_count; 2448 } 2449 2450 static inline void inc_valid_inode_count(struct f2fs_sb_info *sbi) 2451 { 2452 percpu_counter_inc(&sbi->total_valid_inode_count); 2453 } 2454 2455 static inline void dec_valid_inode_count(struct f2fs_sb_info *sbi) 2456 { 2457 percpu_counter_dec(&sbi->total_valid_inode_count); 2458 } 2459 2460 static inline s64 valid_inode_count(struct f2fs_sb_info *sbi) 2461 { 2462 return percpu_counter_sum_positive(&sbi->total_valid_inode_count); 2463 } 2464 2465 static inline struct page *f2fs_grab_cache_page(struct address_space *mapping, 2466 pgoff_t index, bool for_write) 2467 { 2468 struct page *page; 2469 2470 if (IS_ENABLED(CONFIG_F2FS_FAULT_INJECTION)) { 2471 if (!for_write) 2472 page = find_get_page_flags(mapping, index, 2473 FGP_LOCK | FGP_ACCESSED); 2474 else 2475 page = find_lock_page(mapping, index); 2476 if (page) 2477 return page; 2478 2479 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_ALLOC)) { 2480 f2fs_show_injection_info(F2FS_M_SB(mapping), 2481 FAULT_PAGE_ALLOC); 2482 return NULL; 2483 } 2484 } 2485 2486 if (!for_write) 2487 return grab_cache_page(mapping, index); 2488 return grab_cache_page_write_begin(mapping, index, AOP_FLAG_NOFS); 2489 } 2490 2491 static inline struct page *f2fs_pagecache_get_page( 2492 struct address_space *mapping, pgoff_t index, 2493 int fgp_flags, gfp_t gfp_mask) 2494 { 2495 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_GET)) { 2496 f2fs_show_injection_info(F2FS_M_SB(mapping), FAULT_PAGE_GET); 2497 return NULL; 2498 } 2499 2500 return pagecache_get_page(mapping, index, fgp_flags, gfp_mask); 2501 } 2502 2503 static inline void f2fs_copy_page(struct page *src, struct page *dst) 2504 { 2505 char *src_kaddr = kmap(src); 2506 char *dst_kaddr = kmap(dst); 2507 2508 memcpy(dst_kaddr, src_kaddr, PAGE_SIZE); 2509 kunmap(dst); 2510 kunmap(src); 2511 } 2512 2513 static inline void f2fs_put_page(struct page *page, int unlock) 2514 { 2515 if (!page) 2516 return; 2517 2518 if (unlock) { 2519 f2fs_bug_on(F2FS_P_SB(page), !PageLocked(page)); 2520 unlock_page(page); 2521 } 2522 put_page(page); 2523 } 2524 2525 static inline void f2fs_put_dnode(struct dnode_of_data *dn) 2526 { 2527 if (dn->node_page) 2528 f2fs_put_page(dn->node_page, 1); 2529 if (dn->inode_page && dn->node_page != dn->inode_page) 2530 f2fs_put_page(dn->inode_page, 0); 2531 dn->node_page = NULL; 2532 dn->inode_page = NULL; 2533 } 2534 2535 static inline struct kmem_cache *f2fs_kmem_cache_create(const char *name, 2536 size_t size) 2537 { 2538 return kmem_cache_create(name, size, 0, SLAB_RECLAIM_ACCOUNT, NULL); 2539 } 2540 2541 static inline void *f2fs_kmem_cache_alloc(struct kmem_cache *cachep, 2542 gfp_t flags) 2543 { 2544 void *entry; 2545 2546 entry = kmem_cache_alloc(cachep, flags); 2547 if (!entry) 2548 entry = kmem_cache_alloc(cachep, flags | __GFP_NOFAIL); 2549 return entry; 2550 } 2551 2552 static inline bool is_inflight_io(struct f2fs_sb_info *sbi, int type) 2553 { 2554 if (get_pages(sbi, F2FS_RD_DATA) || get_pages(sbi, F2FS_RD_NODE) || 2555 get_pages(sbi, F2FS_RD_META) || get_pages(sbi, F2FS_WB_DATA) || 2556 get_pages(sbi, F2FS_WB_CP_DATA) || 2557 get_pages(sbi, F2FS_DIO_READ) || 2558 get_pages(sbi, F2FS_DIO_WRITE)) 2559 return true; 2560 2561 if (type != DISCARD_TIME && SM_I(sbi) && SM_I(sbi)->dcc_info && 2562 atomic_read(&SM_I(sbi)->dcc_info->queued_discard)) 2563 return true; 2564 2565 if (SM_I(sbi) && SM_I(sbi)->fcc_info && 2566 atomic_read(&SM_I(sbi)->fcc_info->queued_flush)) 2567 return true; 2568 return false; 2569 } 2570 2571 static inline bool is_idle(struct f2fs_sb_info *sbi, int type) 2572 { 2573 if (sbi->gc_mode == GC_URGENT_HIGH) 2574 return true; 2575 2576 if (is_inflight_io(sbi, type)) 2577 return false; 2578 2579 if (sbi->gc_mode == GC_URGENT_LOW && 2580 (type == DISCARD_TIME || type == GC_TIME)) 2581 return true; 2582 2583 return f2fs_time_over(sbi, type); 2584 } 2585 2586 static inline void f2fs_radix_tree_insert(struct radix_tree_root *root, 2587 unsigned long index, void *item) 2588 { 2589 while (radix_tree_insert(root, index, item)) 2590 cond_resched(); 2591 } 2592 2593 #define RAW_IS_INODE(p) ((p)->footer.nid == (p)->footer.ino) 2594 2595 static inline bool IS_INODE(struct page *page) 2596 { 2597 struct f2fs_node *p = F2FS_NODE(page); 2598 2599 return RAW_IS_INODE(p); 2600 } 2601 2602 static inline int offset_in_addr(struct f2fs_inode *i) 2603 { 2604 return (i->i_inline & F2FS_EXTRA_ATTR) ? 2605 (le16_to_cpu(i->i_extra_isize) / sizeof(__le32)) : 0; 2606 } 2607 2608 static inline __le32 *blkaddr_in_node(struct f2fs_node *node) 2609 { 2610 return RAW_IS_INODE(node) ? node->i.i_addr : node->dn.addr; 2611 } 2612 2613 static inline int f2fs_has_extra_attr(struct inode *inode); 2614 static inline block_t data_blkaddr(struct inode *inode, 2615 struct page *node_page, unsigned int offset) 2616 { 2617 struct f2fs_node *raw_node; 2618 __le32 *addr_array; 2619 int base = 0; 2620 bool is_inode = IS_INODE(node_page); 2621 2622 raw_node = F2FS_NODE(node_page); 2623 2624 if (is_inode) { 2625 if (!inode) 2626 /* from GC path only */ 2627 base = offset_in_addr(&raw_node->i); 2628 else if (f2fs_has_extra_attr(inode)) 2629 base = get_extra_isize(inode); 2630 } 2631 2632 addr_array = blkaddr_in_node(raw_node); 2633 return le32_to_cpu(addr_array[base + offset]); 2634 } 2635 2636 static inline block_t f2fs_data_blkaddr(struct dnode_of_data *dn) 2637 { 2638 return data_blkaddr(dn->inode, dn->node_page, dn->ofs_in_node); 2639 } 2640 2641 static inline int f2fs_test_bit(unsigned int nr, char *addr) 2642 { 2643 int mask; 2644 2645 addr += (nr >> 3); 2646 mask = 1 << (7 - (nr & 0x07)); 2647 return mask & *addr; 2648 } 2649 2650 static inline void f2fs_set_bit(unsigned int nr, char *addr) 2651 { 2652 int mask; 2653 2654 addr += (nr >> 3); 2655 mask = 1 << (7 - (nr & 0x07)); 2656 *addr |= mask; 2657 } 2658 2659 static inline void f2fs_clear_bit(unsigned int nr, char *addr) 2660 { 2661 int mask; 2662 2663 addr += (nr >> 3); 2664 mask = 1 << (7 - (nr & 0x07)); 2665 *addr &= ~mask; 2666 } 2667 2668 static inline int f2fs_test_and_set_bit(unsigned int nr, char *addr) 2669 { 2670 int mask; 2671 int ret; 2672 2673 addr += (nr >> 3); 2674 mask = 1 << (7 - (nr & 0x07)); 2675 ret = mask & *addr; 2676 *addr |= mask; 2677 return ret; 2678 } 2679 2680 static inline int f2fs_test_and_clear_bit(unsigned int nr, char *addr) 2681 { 2682 int mask; 2683 int ret; 2684 2685 addr += (nr >> 3); 2686 mask = 1 << (7 - (nr & 0x07)); 2687 ret = mask & *addr; 2688 *addr &= ~mask; 2689 return ret; 2690 } 2691 2692 static inline void f2fs_change_bit(unsigned int nr, char *addr) 2693 { 2694 int mask; 2695 2696 addr += (nr >> 3); 2697 mask = 1 << (7 - (nr & 0x07)); 2698 *addr ^= mask; 2699 } 2700 2701 /* 2702 * On-disk inode flags (f2fs_inode::i_flags) 2703 */ 2704 #define F2FS_COMPR_FL 0x00000004 /* Compress file */ 2705 #define F2FS_SYNC_FL 0x00000008 /* Synchronous updates */ 2706 #define F2FS_IMMUTABLE_FL 0x00000010 /* Immutable file */ 2707 #define F2FS_APPEND_FL 0x00000020 /* writes to file may only append */ 2708 #define F2FS_NODUMP_FL 0x00000040 /* do not dump file */ 2709 #define F2FS_NOATIME_FL 0x00000080 /* do not update atime */ 2710 #define F2FS_NOCOMP_FL 0x00000400 /* Don't compress */ 2711 #define F2FS_INDEX_FL 0x00001000 /* hash-indexed directory */ 2712 #define F2FS_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */ 2713 #define F2FS_PROJINHERIT_FL 0x20000000 /* Create with parents projid */ 2714 #define F2FS_CASEFOLD_FL 0x40000000 /* Casefolded file */ 2715 2716 /* Flags that should be inherited by new inodes from their parent. */ 2717 #define F2FS_FL_INHERITED (F2FS_SYNC_FL | F2FS_NODUMP_FL | F2FS_NOATIME_FL | \ 2718 F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \ 2719 F2FS_CASEFOLD_FL | F2FS_COMPR_FL | F2FS_NOCOMP_FL) 2720 2721 /* Flags that are appropriate for regular files (all but dir-specific ones). */ 2722 #define F2FS_REG_FLMASK (~(F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \ 2723 F2FS_CASEFOLD_FL)) 2724 2725 /* Flags that are appropriate for non-directories/regular files. */ 2726 #define F2FS_OTHER_FLMASK (F2FS_NODUMP_FL | F2FS_NOATIME_FL) 2727 2728 static inline __u32 f2fs_mask_flags(umode_t mode, __u32 flags) 2729 { 2730 if (S_ISDIR(mode)) 2731 return flags; 2732 else if (S_ISREG(mode)) 2733 return flags & F2FS_REG_FLMASK; 2734 else 2735 return flags & F2FS_OTHER_FLMASK; 2736 } 2737 2738 static inline void __mark_inode_dirty_flag(struct inode *inode, 2739 int flag, bool set) 2740 { 2741 switch (flag) { 2742 case FI_INLINE_XATTR: 2743 case FI_INLINE_DATA: 2744 case FI_INLINE_DENTRY: 2745 case FI_NEW_INODE: 2746 if (set) 2747 return; 2748 fallthrough; 2749 case FI_DATA_EXIST: 2750 case FI_INLINE_DOTS: 2751 case FI_PIN_FILE: 2752 case FI_COMPRESS_RELEASED: 2753 f2fs_mark_inode_dirty_sync(inode, true); 2754 } 2755 } 2756 2757 static inline void set_inode_flag(struct inode *inode, int flag) 2758 { 2759 set_bit(flag, F2FS_I(inode)->flags); 2760 __mark_inode_dirty_flag(inode, flag, true); 2761 } 2762 2763 static inline int is_inode_flag_set(struct inode *inode, int flag) 2764 { 2765 return test_bit(flag, F2FS_I(inode)->flags); 2766 } 2767 2768 static inline void clear_inode_flag(struct inode *inode, int flag) 2769 { 2770 clear_bit(flag, F2FS_I(inode)->flags); 2771 __mark_inode_dirty_flag(inode, flag, false); 2772 } 2773 2774 static inline bool f2fs_verity_in_progress(struct inode *inode) 2775 { 2776 return IS_ENABLED(CONFIG_FS_VERITY) && 2777 is_inode_flag_set(inode, FI_VERITY_IN_PROGRESS); 2778 } 2779 2780 static inline void set_acl_inode(struct inode *inode, umode_t mode) 2781 { 2782 F2FS_I(inode)->i_acl_mode = mode; 2783 set_inode_flag(inode, FI_ACL_MODE); 2784 f2fs_mark_inode_dirty_sync(inode, false); 2785 } 2786 2787 static inline void f2fs_i_links_write(struct inode *inode, bool inc) 2788 { 2789 if (inc) 2790 inc_nlink(inode); 2791 else 2792 drop_nlink(inode); 2793 f2fs_mark_inode_dirty_sync(inode, true); 2794 } 2795 2796 static inline void f2fs_i_blocks_write(struct inode *inode, 2797 block_t diff, bool add, bool claim) 2798 { 2799 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE); 2800 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER); 2801 2802 /* add = 1, claim = 1 should be dquot_reserve_block in pair */ 2803 if (add) { 2804 if (claim) 2805 dquot_claim_block(inode, diff); 2806 else 2807 dquot_alloc_block_nofail(inode, diff); 2808 } else { 2809 dquot_free_block(inode, diff); 2810 } 2811 2812 f2fs_mark_inode_dirty_sync(inode, true); 2813 if (clean || recover) 2814 set_inode_flag(inode, FI_AUTO_RECOVER); 2815 } 2816 2817 static inline void f2fs_i_size_write(struct inode *inode, loff_t i_size) 2818 { 2819 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE); 2820 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER); 2821 2822 if (i_size_read(inode) == i_size) 2823 return; 2824 2825 i_size_write(inode, i_size); 2826 f2fs_mark_inode_dirty_sync(inode, true); 2827 if (clean || recover) 2828 set_inode_flag(inode, FI_AUTO_RECOVER); 2829 } 2830 2831 static inline void f2fs_i_depth_write(struct inode *inode, unsigned int depth) 2832 { 2833 F2FS_I(inode)->i_current_depth = depth; 2834 f2fs_mark_inode_dirty_sync(inode, true); 2835 } 2836 2837 static inline void f2fs_i_gc_failures_write(struct inode *inode, 2838 unsigned int count) 2839 { 2840 F2FS_I(inode)->i_gc_failures[GC_FAILURE_PIN] = count; 2841 f2fs_mark_inode_dirty_sync(inode, true); 2842 } 2843 2844 static inline void f2fs_i_xnid_write(struct inode *inode, nid_t xnid) 2845 { 2846 F2FS_I(inode)->i_xattr_nid = xnid; 2847 f2fs_mark_inode_dirty_sync(inode, true); 2848 } 2849 2850 static inline void f2fs_i_pino_write(struct inode *inode, nid_t pino) 2851 { 2852 F2FS_I(inode)->i_pino = pino; 2853 f2fs_mark_inode_dirty_sync(inode, true); 2854 } 2855 2856 static inline void get_inline_info(struct inode *inode, struct f2fs_inode *ri) 2857 { 2858 struct f2fs_inode_info *fi = F2FS_I(inode); 2859 2860 if (ri->i_inline & F2FS_INLINE_XATTR) 2861 set_bit(FI_INLINE_XATTR, fi->flags); 2862 if (ri->i_inline & F2FS_INLINE_DATA) 2863 set_bit(FI_INLINE_DATA, fi->flags); 2864 if (ri->i_inline & F2FS_INLINE_DENTRY) 2865 set_bit(FI_INLINE_DENTRY, fi->flags); 2866 if (ri->i_inline & F2FS_DATA_EXIST) 2867 set_bit(FI_DATA_EXIST, fi->flags); 2868 if (ri->i_inline & F2FS_INLINE_DOTS) 2869 set_bit(FI_INLINE_DOTS, fi->flags); 2870 if (ri->i_inline & F2FS_EXTRA_ATTR) 2871 set_bit(FI_EXTRA_ATTR, fi->flags); 2872 if (ri->i_inline & F2FS_PIN_FILE) 2873 set_bit(FI_PIN_FILE, fi->flags); 2874 if (ri->i_inline & F2FS_COMPRESS_RELEASED) 2875 set_bit(FI_COMPRESS_RELEASED, fi->flags); 2876 } 2877 2878 static inline void set_raw_inline(struct inode *inode, struct f2fs_inode *ri) 2879 { 2880 ri->i_inline = 0; 2881 2882 if (is_inode_flag_set(inode, FI_INLINE_XATTR)) 2883 ri->i_inline |= F2FS_INLINE_XATTR; 2884 if (is_inode_flag_set(inode, FI_INLINE_DATA)) 2885 ri->i_inline |= F2FS_INLINE_DATA; 2886 if (is_inode_flag_set(inode, FI_INLINE_DENTRY)) 2887 ri->i_inline |= F2FS_INLINE_DENTRY; 2888 if (is_inode_flag_set(inode, FI_DATA_EXIST)) 2889 ri->i_inline |= F2FS_DATA_EXIST; 2890 if (is_inode_flag_set(inode, FI_INLINE_DOTS)) 2891 ri->i_inline |= F2FS_INLINE_DOTS; 2892 if (is_inode_flag_set(inode, FI_EXTRA_ATTR)) 2893 ri->i_inline |= F2FS_EXTRA_ATTR; 2894 if (is_inode_flag_set(inode, FI_PIN_FILE)) 2895 ri->i_inline |= F2FS_PIN_FILE; 2896 if (is_inode_flag_set(inode, FI_COMPRESS_RELEASED)) 2897 ri->i_inline |= F2FS_COMPRESS_RELEASED; 2898 } 2899 2900 static inline int f2fs_has_extra_attr(struct inode *inode) 2901 { 2902 return is_inode_flag_set(inode, FI_EXTRA_ATTR); 2903 } 2904 2905 static inline int f2fs_has_inline_xattr(struct inode *inode) 2906 { 2907 return is_inode_flag_set(inode, FI_INLINE_XATTR); 2908 } 2909 2910 static inline int f2fs_compressed_file(struct inode *inode) 2911 { 2912 return S_ISREG(inode->i_mode) && 2913 is_inode_flag_set(inode, FI_COMPRESSED_FILE); 2914 } 2915 2916 static inline bool f2fs_need_compress_data(struct inode *inode) 2917 { 2918 int compress_mode = F2FS_OPTION(F2FS_I_SB(inode)).compress_mode; 2919 2920 if (!f2fs_compressed_file(inode)) 2921 return false; 2922 2923 if (compress_mode == COMPR_MODE_FS) 2924 return true; 2925 else if (compress_mode == COMPR_MODE_USER && 2926 is_inode_flag_set(inode, FI_ENABLE_COMPRESS)) 2927 return true; 2928 2929 return false; 2930 } 2931 2932 static inline unsigned int addrs_per_inode(struct inode *inode) 2933 { 2934 unsigned int addrs = CUR_ADDRS_PER_INODE(inode) - 2935 get_inline_xattr_addrs(inode); 2936 2937 if (!f2fs_compressed_file(inode)) 2938 return addrs; 2939 return ALIGN_DOWN(addrs, F2FS_I(inode)->i_cluster_size); 2940 } 2941 2942 static inline unsigned int addrs_per_block(struct inode *inode) 2943 { 2944 if (!f2fs_compressed_file(inode)) 2945 return DEF_ADDRS_PER_BLOCK; 2946 return ALIGN_DOWN(DEF_ADDRS_PER_BLOCK, F2FS_I(inode)->i_cluster_size); 2947 } 2948 2949 static inline void *inline_xattr_addr(struct inode *inode, struct page *page) 2950 { 2951 struct f2fs_inode *ri = F2FS_INODE(page); 2952 2953 return (void *)&(ri->i_addr[DEF_ADDRS_PER_INODE - 2954 get_inline_xattr_addrs(inode)]); 2955 } 2956 2957 static inline int inline_xattr_size(struct inode *inode) 2958 { 2959 if (f2fs_has_inline_xattr(inode)) 2960 return get_inline_xattr_addrs(inode) * sizeof(__le32); 2961 return 0; 2962 } 2963 2964 static inline int f2fs_has_inline_data(struct inode *inode) 2965 { 2966 return is_inode_flag_set(inode, FI_INLINE_DATA); 2967 } 2968 2969 static inline int f2fs_exist_data(struct inode *inode) 2970 { 2971 return is_inode_flag_set(inode, FI_DATA_EXIST); 2972 } 2973 2974 static inline int f2fs_has_inline_dots(struct inode *inode) 2975 { 2976 return is_inode_flag_set(inode, FI_INLINE_DOTS); 2977 } 2978 2979 static inline int f2fs_is_mmap_file(struct inode *inode) 2980 { 2981 return is_inode_flag_set(inode, FI_MMAP_FILE); 2982 } 2983 2984 static inline bool f2fs_is_pinned_file(struct inode *inode) 2985 { 2986 return is_inode_flag_set(inode, FI_PIN_FILE); 2987 } 2988 2989 static inline bool f2fs_is_atomic_file(struct inode *inode) 2990 { 2991 return is_inode_flag_set(inode, FI_ATOMIC_FILE); 2992 } 2993 2994 static inline bool f2fs_is_commit_atomic_write(struct inode *inode) 2995 { 2996 return is_inode_flag_set(inode, FI_ATOMIC_COMMIT); 2997 } 2998 2999 static inline bool f2fs_is_volatile_file(struct inode *inode) 3000 { 3001 return is_inode_flag_set(inode, FI_VOLATILE_FILE); 3002 } 3003 3004 static inline bool f2fs_is_first_block_written(struct inode *inode) 3005 { 3006 return is_inode_flag_set(inode, FI_FIRST_BLOCK_WRITTEN); 3007 } 3008 3009 static inline bool f2fs_is_drop_cache(struct inode *inode) 3010 { 3011 return is_inode_flag_set(inode, FI_DROP_CACHE); 3012 } 3013 3014 static inline void *inline_data_addr(struct inode *inode, struct page *page) 3015 { 3016 struct f2fs_inode *ri = F2FS_INODE(page); 3017 int extra_size = get_extra_isize(inode); 3018 3019 return (void *)&(ri->i_addr[extra_size + DEF_INLINE_RESERVED_SIZE]); 3020 } 3021 3022 static inline int f2fs_has_inline_dentry(struct inode *inode) 3023 { 3024 return is_inode_flag_set(inode, FI_INLINE_DENTRY); 3025 } 3026 3027 static inline int is_file(struct inode *inode, int type) 3028 { 3029 return F2FS_I(inode)->i_advise & type; 3030 } 3031 3032 static inline void set_file(struct inode *inode, int type) 3033 { 3034 F2FS_I(inode)->i_advise |= type; 3035 f2fs_mark_inode_dirty_sync(inode, true); 3036 } 3037 3038 static inline void clear_file(struct inode *inode, int type) 3039 { 3040 F2FS_I(inode)->i_advise &= ~type; 3041 f2fs_mark_inode_dirty_sync(inode, true); 3042 } 3043 3044 static inline bool f2fs_is_time_consistent(struct inode *inode) 3045 { 3046 if (!timespec64_equal(F2FS_I(inode)->i_disk_time, &inode->i_atime)) 3047 return false; 3048 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 1, &inode->i_ctime)) 3049 return false; 3050 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 2, &inode->i_mtime)) 3051 return false; 3052 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 3, 3053 &F2FS_I(inode)->i_crtime)) 3054 return false; 3055 return true; 3056 } 3057 3058 static inline bool f2fs_skip_inode_update(struct inode *inode, int dsync) 3059 { 3060 bool ret; 3061 3062 if (dsync) { 3063 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 3064 3065 spin_lock(&sbi->inode_lock[DIRTY_META]); 3066 ret = list_empty(&F2FS_I(inode)->gdirty_list); 3067 spin_unlock(&sbi->inode_lock[DIRTY_META]); 3068 return ret; 3069 } 3070 if (!is_inode_flag_set(inode, FI_AUTO_RECOVER) || 3071 file_keep_isize(inode) || 3072 i_size_read(inode) & ~PAGE_MASK) 3073 return false; 3074 3075 if (!f2fs_is_time_consistent(inode)) 3076 return false; 3077 3078 spin_lock(&F2FS_I(inode)->i_size_lock); 3079 ret = F2FS_I(inode)->last_disk_size == i_size_read(inode); 3080 spin_unlock(&F2FS_I(inode)->i_size_lock); 3081 3082 return ret; 3083 } 3084 3085 static inline bool f2fs_readonly(struct super_block *sb) 3086 { 3087 return sb_rdonly(sb); 3088 } 3089 3090 static inline bool f2fs_cp_error(struct f2fs_sb_info *sbi) 3091 { 3092 return is_set_ckpt_flags(sbi, CP_ERROR_FLAG); 3093 } 3094 3095 static inline bool is_dot_dotdot(const u8 *name, size_t len) 3096 { 3097 if (len == 1 && name[0] == '.') 3098 return true; 3099 3100 if (len == 2 && name[0] == '.' && name[1] == '.') 3101 return true; 3102 3103 return false; 3104 } 3105 3106 static inline bool f2fs_may_extent_tree(struct inode *inode) 3107 { 3108 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 3109 3110 if (!test_opt(sbi, EXTENT_CACHE) || 3111 is_inode_flag_set(inode, FI_NO_EXTENT) || 3112 is_inode_flag_set(inode, FI_COMPRESSED_FILE)) 3113 return false; 3114 3115 /* 3116 * for recovered files during mount do not create extents 3117 * if shrinker is not registered. 3118 */ 3119 if (list_empty(&sbi->s_list)) 3120 return false; 3121 3122 return S_ISREG(inode->i_mode); 3123 } 3124 3125 static inline void *f2fs_kmalloc(struct f2fs_sb_info *sbi, 3126 size_t size, gfp_t flags) 3127 { 3128 if (time_to_inject(sbi, FAULT_KMALLOC)) { 3129 f2fs_show_injection_info(sbi, FAULT_KMALLOC); 3130 return NULL; 3131 } 3132 3133 return kmalloc(size, flags); 3134 } 3135 3136 static inline void *f2fs_kzalloc(struct f2fs_sb_info *sbi, 3137 size_t size, gfp_t flags) 3138 { 3139 return f2fs_kmalloc(sbi, size, flags | __GFP_ZERO); 3140 } 3141 3142 static inline void *f2fs_kvmalloc(struct f2fs_sb_info *sbi, 3143 size_t size, gfp_t flags) 3144 { 3145 if (time_to_inject(sbi, FAULT_KVMALLOC)) { 3146 f2fs_show_injection_info(sbi, FAULT_KVMALLOC); 3147 return NULL; 3148 } 3149 3150 return kvmalloc(size, flags); 3151 } 3152 3153 static inline void *f2fs_kvzalloc(struct f2fs_sb_info *sbi, 3154 size_t size, gfp_t flags) 3155 { 3156 return f2fs_kvmalloc(sbi, size, flags | __GFP_ZERO); 3157 } 3158 3159 static inline int get_extra_isize(struct inode *inode) 3160 { 3161 return F2FS_I(inode)->i_extra_isize / sizeof(__le32); 3162 } 3163 3164 static inline int get_inline_xattr_addrs(struct inode *inode) 3165 { 3166 return F2FS_I(inode)->i_inline_xattr_size; 3167 } 3168 3169 #define f2fs_get_inode_mode(i) \ 3170 ((is_inode_flag_set(i, FI_ACL_MODE)) ? \ 3171 (F2FS_I(i)->i_acl_mode) : ((i)->i_mode)) 3172 3173 #define F2FS_TOTAL_EXTRA_ATTR_SIZE \ 3174 (offsetof(struct f2fs_inode, i_extra_end) - \ 3175 offsetof(struct f2fs_inode, i_extra_isize)) \ 3176 3177 #define F2FS_OLD_ATTRIBUTE_SIZE (offsetof(struct f2fs_inode, i_addr)) 3178 #define F2FS_FITS_IN_INODE(f2fs_inode, extra_isize, field) \ 3179 ((offsetof(typeof(*(f2fs_inode)), field) + \ 3180 sizeof((f2fs_inode)->field)) \ 3181 <= (F2FS_OLD_ATTRIBUTE_SIZE + (extra_isize))) \ 3182 3183 #define DEFAULT_IOSTAT_PERIOD_MS 3000 3184 #define MIN_IOSTAT_PERIOD_MS 100 3185 /* maximum period of iostat tracing is 1 day */ 3186 #define MAX_IOSTAT_PERIOD_MS 8640000 3187 3188 static inline void f2fs_reset_iostat(struct f2fs_sb_info *sbi) 3189 { 3190 int i; 3191 3192 spin_lock(&sbi->iostat_lock); 3193 for (i = 0; i < NR_IO_TYPE; i++) { 3194 sbi->rw_iostat[i] = 0; 3195 sbi->prev_rw_iostat[i] = 0; 3196 } 3197 spin_unlock(&sbi->iostat_lock); 3198 } 3199 3200 extern void f2fs_record_iostat(struct f2fs_sb_info *sbi); 3201 3202 static inline void f2fs_update_iostat(struct f2fs_sb_info *sbi, 3203 enum iostat_type type, unsigned long long io_bytes) 3204 { 3205 if (!sbi->iostat_enable) 3206 return; 3207 spin_lock(&sbi->iostat_lock); 3208 sbi->rw_iostat[type] += io_bytes; 3209 3210 if (type == APP_WRITE_IO || type == APP_DIRECT_IO) 3211 sbi->rw_iostat[APP_BUFFERED_IO] = 3212 sbi->rw_iostat[APP_WRITE_IO] - 3213 sbi->rw_iostat[APP_DIRECT_IO]; 3214 3215 if (type == APP_READ_IO || type == APP_DIRECT_READ_IO) 3216 sbi->rw_iostat[APP_BUFFERED_READ_IO] = 3217 sbi->rw_iostat[APP_READ_IO] - 3218 sbi->rw_iostat[APP_DIRECT_READ_IO]; 3219 spin_unlock(&sbi->iostat_lock); 3220 3221 f2fs_record_iostat(sbi); 3222 } 3223 3224 #define __is_large_section(sbi) ((sbi)->segs_per_sec > 1) 3225 3226 #define __is_meta_io(fio) (PAGE_TYPE_OF_BIO((fio)->type) == META) 3227 3228 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi, 3229 block_t blkaddr, int type); 3230 static inline void verify_blkaddr(struct f2fs_sb_info *sbi, 3231 block_t blkaddr, int type) 3232 { 3233 if (!f2fs_is_valid_blkaddr(sbi, blkaddr, type)) { 3234 f2fs_err(sbi, "invalid blkaddr: %u, type: %d, run fsck to fix.", 3235 blkaddr, type); 3236 f2fs_bug_on(sbi, 1); 3237 } 3238 } 3239 3240 static inline bool __is_valid_data_blkaddr(block_t blkaddr) 3241 { 3242 if (blkaddr == NEW_ADDR || blkaddr == NULL_ADDR || 3243 blkaddr == COMPRESS_ADDR) 3244 return false; 3245 return true; 3246 } 3247 3248 /* 3249 * file.c 3250 */ 3251 int f2fs_sync_file(struct file *file, loff_t start, loff_t end, int datasync); 3252 void f2fs_truncate_data_blocks(struct dnode_of_data *dn); 3253 int f2fs_do_truncate_blocks(struct inode *inode, u64 from, bool lock); 3254 int f2fs_truncate_blocks(struct inode *inode, u64 from, bool lock); 3255 int f2fs_truncate(struct inode *inode); 3256 int f2fs_getattr(struct user_namespace *mnt_userns, const struct path *path, 3257 struct kstat *stat, u32 request_mask, unsigned int flags); 3258 int f2fs_setattr(struct user_namespace *mnt_userns, struct dentry *dentry, 3259 struct iattr *attr); 3260 int f2fs_truncate_hole(struct inode *inode, pgoff_t pg_start, pgoff_t pg_end); 3261 void f2fs_truncate_data_blocks_range(struct dnode_of_data *dn, int count); 3262 int f2fs_precache_extents(struct inode *inode); 3263 int f2fs_fileattr_get(struct dentry *dentry, struct fileattr *fa); 3264 int f2fs_fileattr_set(struct user_namespace *mnt_userns, 3265 struct dentry *dentry, struct fileattr *fa); 3266 long f2fs_ioctl(struct file *filp, unsigned int cmd, unsigned long arg); 3267 long f2fs_compat_ioctl(struct file *file, unsigned int cmd, unsigned long arg); 3268 int f2fs_transfer_project_quota(struct inode *inode, kprojid_t kprojid); 3269 int f2fs_pin_file_control(struct inode *inode, bool inc); 3270 3271 /* 3272 * inode.c 3273 */ 3274 void f2fs_set_inode_flags(struct inode *inode); 3275 bool f2fs_inode_chksum_verify(struct f2fs_sb_info *sbi, struct page *page); 3276 void f2fs_inode_chksum_set(struct f2fs_sb_info *sbi, struct page *page); 3277 struct inode *f2fs_iget(struct super_block *sb, unsigned long ino); 3278 struct inode *f2fs_iget_retry(struct super_block *sb, unsigned long ino); 3279 int f2fs_try_to_free_nats(struct f2fs_sb_info *sbi, int nr_shrink); 3280 void f2fs_update_inode(struct inode *inode, struct page *node_page); 3281 void f2fs_update_inode_page(struct inode *inode); 3282 int f2fs_write_inode(struct inode *inode, struct writeback_control *wbc); 3283 void f2fs_evict_inode(struct inode *inode); 3284 void f2fs_handle_failed_inode(struct inode *inode); 3285 3286 /* 3287 * namei.c 3288 */ 3289 int f2fs_update_extension_list(struct f2fs_sb_info *sbi, const char *name, 3290 bool hot, bool set); 3291 struct dentry *f2fs_get_parent(struct dentry *child); 3292 3293 /* 3294 * dir.c 3295 */ 3296 unsigned char f2fs_get_de_type(struct f2fs_dir_entry *de); 3297 int f2fs_init_casefolded_name(const struct inode *dir, 3298 struct f2fs_filename *fname); 3299 int f2fs_setup_filename(struct inode *dir, const struct qstr *iname, 3300 int lookup, struct f2fs_filename *fname); 3301 int f2fs_prepare_lookup(struct inode *dir, struct dentry *dentry, 3302 struct f2fs_filename *fname); 3303 void f2fs_free_filename(struct f2fs_filename *fname); 3304 struct f2fs_dir_entry *f2fs_find_target_dentry(const struct f2fs_dentry_ptr *d, 3305 const struct f2fs_filename *fname, int *max_slots); 3306 int f2fs_fill_dentries(struct dir_context *ctx, struct f2fs_dentry_ptr *d, 3307 unsigned int start_pos, struct fscrypt_str *fstr); 3308 void f2fs_do_make_empty_dir(struct inode *inode, struct inode *parent, 3309 struct f2fs_dentry_ptr *d); 3310 struct page *f2fs_init_inode_metadata(struct inode *inode, struct inode *dir, 3311 const struct f2fs_filename *fname, struct page *dpage); 3312 void f2fs_update_parent_metadata(struct inode *dir, struct inode *inode, 3313 unsigned int current_depth); 3314 int f2fs_room_for_filename(const void *bitmap, int slots, int max_slots); 3315 void f2fs_drop_nlink(struct inode *dir, struct inode *inode); 3316 struct f2fs_dir_entry *__f2fs_find_entry(struct inode *dir, 3317 const struct f2fs_filename *fname, 3318 struct page **res_page); 3319 struct f2fs_dir_entry *f2fs_find_entry(struct inode *dir, 3320 const struct qstr *child, struct page **res_page); 3321 struct f2fs_dir_entry *f2fs_parent_dir(struct inode *dir, struct page **p); 3322 ino_t f2fs_inode_by_name(struct inode *dir, const struct qstr *qstr, 3323 struct page **page); 3324 void f2fs_set_link(struct inode *dir, struct f2fs_dir_entry *de, 3325 struct page *page, struct inode *inode); 3326 bool f2fs_has_enough_room(struct inode *dir, struct page *ipage, 3327 const struct f2fs_filename *fname); 3328 void f2fs_update_dentry(nid_t ino, umode_t mode, struct f2fs_dentry_ptr *d, 3329 const struct fscrypt_str *name, f2fs_hash_t name_hash, 3330 unsigned int bit_pos); 3331 int f2fs_add_regular_entry(struct inode *dir, const struct f2fs_filename *fname, 3332 struct inode *inode, nid_t ino, umode_t mode); 3333 int f2fs_add_dentry(struct inode *dir, const struct f2fs_filename *fname, 3334 struct inode *inode, nid_t ino, umode_t mode); 3335 int f2fs_do_add_link(struct inode *dir, const struct qstr *name, 3336 struct inode *inode, nid_t ino, umode_t mode); 3337 void f2fs_delete_entry(struct f2fs_dir_entry *dentry, struct page *page, 3338 struct inode *dir, struct inode *inode); 3339 int f2fs_do_tmpfile(struct inode *inode, struct inode *dir); 3340 bool f2fs_empty_dir(struct inode *dir); 3341 3342 static inline int f2fs_add_link(struct dentry *dentry, struct inode *inode) 3343 { 3344 if (fscrypt_is_nokey_name(dentry)) 3345 return -ENOKEY; 3346 return f2fs_do_add_link(d_inode(dentry->d_parent), &dentry->d_name, 3347 inode, inode->i_ino, inode->i_mode); 3348 } 3349 3350 /* 3351 * super.c 3352 */ 3353 int f2fs_inode_dirtied(struct inode *inode, bool sync); 3354 void f2fs_inode_synced(struct inode *inode); 3355 int f2fs_enable_quota_files(struct f2fs_sb_info *sbi, bool rdonly); 3356 int f2fs_quota_sync(struct super_block *sb, int type); 3357 loff_t max_file_blocks(struct inode *inode); 3358 void f2fs_quota_off_umount(struct super_block *sb); 3359 int f2fs_commit_super(struct f2fs_sb_info *sbi, bool recover); 3360 int f2fs_sync_fs(struct super_block *sb, int sync); 3361 int f2fs_sanity_check_ckpt(struct f2fs_sb_info *sbi); 3362 3363 /* 3364 * hash.c 3365 */ 3366 void f2fs_hash_filename(const struct inode *dir, struct f2fs_filename *fname); 3367 3368 /* 3369 * node.c 3370 */ 3371 struct node_info; 3372 3373 int f2fs_check_nid_range(struct f2fs_sb_info *sbi, nid_t nid); 3374 bool f2fs_available_free_memory(struct f2fs_sb_info *sbi, int type); 3375 bool f2fs_in_warm_node_list(struct f2fs_sb_info *sbi, struct page *page); 3376 void f2fs_init_fsync_node_info(struct f2fs_sb_info *sbi); 3377 void f2fs_del_fsync_node_entry(struct f2fs_sb_info *sbi, struct page *page); 3378 void f2fs_reset_fsync_node_info(struct f2fs_sb_info *sbi); 3379 int f2fs_need_dentry_mark(struct f2fs_sb_info *sbi, nid_t nid); 3380 bool f2fs_is_checkpointed_node(struct f2fs_sb_info *sbi, nid_t nid); 3381 bool f2fs_need_inode_block_update(struct f2fs_sb_info *sbi, nid_t ino); 3382 int f2fs_get_node_info(struct f2fs_sb_info *sbi, nid_t nid, 3383 struct node_info *ni); 3384 pgoff_t f2fs_get_next_page_offset(struct dnode_of_data *dn, pgoff_t pgofs); 3385 int f2fs_get_dnode_of_data(struct dnode_of_data *dn, pgoff_t index, int mode); 3386 int f2fs_truncate_inode_blocks(struct inode *inode, pgoff_t from); 3387 int f2fs_truncate_xattr_node(struct inode *inode); 3388 int f2fs_wait_on_node_pages_writeback(struct f2fs_sb_info *sbi, 3389 unsigned int seq_id); 3390 int f2fs_remove_inode_page(struct inode *inode); 3391 struct page *f2fs_new_inode_page(struct inode *inode); 3392 struct page *f2fs_new_node_page(struct dnode_of_data *dn, unsigned int ofs); 3393 void f2fs_ra_node_page(struct f2fs_sb_info *sbi, nid_t nid); 3394 struct page *f2fs_get_node_page(struct f2fs_sb_info *sbi, pgoff_t nid); 3395 struct page *f2fs_get_node_page_ra(struct page *parent, int start); 3396 int f2fs_move_node_page(struct page *node_page, int gc_type); 3397 void f2fs_flush_inline_data(struct f2fs_sb_info *sbi); 3398 int f2fs_fsync_node_pages(struct f2fs_sb_info *sbi, struct inode *inode, 3399 struct writeback_control *wbc, bool atomic, 3400 unsigned int *seq_id); 3401 int f2fs_sync_node_pages(struct f2fs_sb_info *sbi, 3402 struct writeback_control *wbc, 3403 bool do_balance, enum iostat_type io_type); 3404 int f2fs_build_free_nids(struct f2fs_sb_info *sbi, bool sync, bool mount); 3405 bool f2fs_alloc_nid(struct f2fs_sb_info *sbi, nid_t *nid); 3406 void f2fs_alloc_nid_done(struct f2fs_sb_info *sbi, nid_t nid); 3407 void f2fs_alloc_nid_failed(struct f2fs_sb_info *sbi, nid_t nid); 3408 int f2fs_try_to_free_nids(struct f2fs_sb_info *sbi, int nr_shrink); 3409 int f2fs_recover_inline_xattr(struct inode *inode, struct page *page); 3410 int f2fs_recover_xattr_data(struct inode *inode, struct page *page); 3411 int f2fs_recover_inode_page(struct f2fs_sb_info *sbi, struct page *page); 3412 int f2fs_restore_node_summary(struct f2fs_sb_info *sbi, 3413 unsigned int segno, struct f2fs_summary_block *sum); 3414 int f2fs_flush_nat_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3415 int f2fs_build_node_manager(struct f2fs_sb_info *sbi); 3416 void f2fs_destroy_node_manager(struct f2fs_sb_info *sbi); 3417 int __init f2fs_create_node_manager_caches(void); 3418 void f2fs_destroy_node_manager_caches(void); 3419 3420 /* 3421 * segment.c 3422 */ 3423 bool f2fs_need_SSR(struct f2fs_sb_info *sbi); 3424 void f2fs_register_inmem_page(struct inode *inode, struct page *page); 3425 void f2fs_drop_inmem_pages_all(struct f2fs_sb_info *sbi, bool gc_failure); 3426 void f2fs_drop_inmem_pages(struct inode *inode); 3427 void f2fs_drop_inmem_page(struct inode *inode, struct page *page); 3428 int f2fs_commit_inmem_pages(struct inode *inode); 3429 void f2fs_balance_fs(struct f2fs_sb_info *sbi, bool need); 3430 void f2fs_balance_fs_bg(struct f2fs_sb_info *sbi, bool from_bg); 3431 int f2fs_issue_flush(struct f2fs_sb_info *sbi, nid_t ino); 3432 int f2fs_create_flush_cmd_control(struct f2fs_sb_info *sbi); 3433 int f2fs_flush_device_cache(struct f2fs_sb_info *sbi); 3434 void f2fs_destroy_flush_cmd_control(struct f2fs_sb_info *sbi, bool free); 3435 void f2fs_invalidate_blocks(struct f2fs_sb_info *sbi, block_t addr); 3436 bool f2fs_is_checkpointed_data(struct f2fs_sb_info *sbi, block_t blkaddr); 3437 void f2fs_drop_discard_cmd(struct f2fs_sb_info *sbi); 3438 void f2fs_stop_discard_thread(struct f2fs_sb_info *sbi); 3439 bool f2fs_issue_discard_timeout(struct f2fs_sb_info *sbi); 3440 void f2fs_clear_prefree_segments(struct f2fs_sb_info *sbi, 3441 struct cp_control *cpc); 3442 void f2fs_dirty_to_prefree(struct f2fs_sb_info *sbi); 3443 block_t f2fs_get_unusable_blocks(struct f2fs_sb_info *sbi); 3444 int f2fs_disable_cp_again(struct f2fs_sb_info *sbi, block_t unusable); 3445 void f2fs_release_discard_addrs(struct f2fs_sb_info *sbi); 3446 int f2fs_npages_for_summary_flush(struct f2fs_sb_info *sbi, bool for_ra); 3447 bool f2fs_segment_has_free_slot(struct f2fs_sb_info *sbi, int segno); 3448 void f2fs_init_inmem_curseg(struct f2fs_sb_info *sbi); 3449 void f2fs_save_inmem_curseg(struct f2fs_sb_info *sbi); 3450 void f2fs_restore_inmem_curseg(struct f2fs_sb_info *sbi); 3451 void f2fs_get_new_segment(struct f2fs_sb_info *sbi, 3452 unsigned int *newseg, bool new_sec, int dir); 3453 void f2fs_allocate_segment_for_resize(struct f2fs_sb_info *sbi, int type, 3454 unsigned int start, unsigned int end); 3455 void f2fs_allocate_new_section(struct f2fs_sb_info *sbi, int type, bool force); 3456 void f2fs_allocate_new_segments(struct f2fs_sb_info *sbi); 3457 int f2fs_trim_fs(struct f2fs_sb_info *sbi, struct fstrim_range *range); 3458 bool f2fs_exist_trim_candidates(struct f2fs_sb_info *sbi, 3459 struct cp_control *cpc); 3460 struct page *f2fs_get_sum_page(struct f2fs_sb_info *sbi, unsigned int segno); 3461 void f2fs_update_meta_page(struct f2fs_sb_info *sbi, void *src, 3462 block_t blk_addr); 3463 void f2fs_do_write_meta_page(struct f2fs_sb_info *sbi, struct page *page, 3464 enum iostat_type io_type); 3465 void f2fs_do_write_node_page(unsigned int nid, struct f2fs_io_info *fio); 3466 void f2fs_outplace_write_data(struct dnode_of_data *dn, 3467 struct f2fs_io_info *fio); 3468 int f2fs_inplace_write_data(struct f2fs_io_info *fio); 3469 void f2fs_do_replace_block(struct f2fs_sb_info *sbi, struct f2fs_summary *sum, 3470 block_t old_blkaddr, block_t new_blkaddr, 3471 bool recover_curseg, bool recover_newaddr, 3472 bool from_gc); 3473 void f2fs_replace_block(struct f2fs_sb_info *sbi, struct dnode_of_data *dn, 3474 block_t old_addr, block_t new_addr, 3475 unsigned char version, bool recover_curseg, 3476 bool recover_newaddr); 3477 void f2fs_allocate_data_block(struct f2fs_sb_info *sbi, struct page *page, 3478 block_t old_blkaddr, block_t *new_blkaddr, 3479 struct f2fs_summary *sum, int type, 3480 struct f2fs_io_info *fio); 3481 void f2fs_wait_on_page_writeback(struct page *page, 3482 enum page_type type, bool ordered, bool locked); 3483 void f2fs_wait_on_block_writeback(struct inode *inode, block_t blkaddr); 3484 void f2fs_wait_on_block_writeback_range(struct inode *inode, block_t blkaddr, 3485 block_t len); 3486 void f2fs_write_data_summaries(struct f2fs_sb_info *sbi, block_t start_blk); 3487 void f2fs_write_node_summaries(struct f2fs_sb_info *sbi, block_t start_blk); 3488 int f2fs_lookup_journal_in_cursum(struct f2fs_journal *journal, int type, 3489 unsigned int val, int alloc); 3490 void f2fs_flush_sit_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3491 int f2fs_fix_curseg_write_pointer(struct f2fs_sb_info *sbi); 3492 int f2fs_check_write_pointer(struct f2fs_sb_info *sbi); 3493 int f2fs_build_segment_manager(struct f2fs_sb_info *sbi); 3494 void f2fs_destroy_segment_manager(struct f2fs_sb_info *sbi); 3495 int __init f2fs_create_segment_manager_caches(void); 3496 void f2fs_destroy_segment_manager_caches(void); 3497 int f2fs_rw_hint_to_seg_type(enum rw_hint hint); 3498 enum rw_hint f2fs_io_type_to_rw_hint(struct f2fs_sb_info *sbi, 3499 enum page_type type, enum temp_type temp); 3500 unsigned int f2fs_usable_segs_in_sec(struct f2fs_sb_info *sbi, 3501 unsigned int segno); 3502 unsigned int f2fs_usable_blks_in_seg(struct f2fs_sb_info *sbi, 3503 unsigned int segno); 3504 3505 /* 3506 * checkpoint.c 3507 */ 3508 void f2fs_stop_checkpoint(struct f2fs_sb_info *sbi, bool end_io); 3509 struct page *f2fs_grab_meta_page(struct f2fs_sb_info *sbi, pgoff_t index); 3510 struct page *f2fs_get_meta_page(struct f2fs_sb_info *sbi, pgoff_t index); 3511 struct page *f2fs_get_meta_page_retry(struct f2fs_sb_info *sbi, pgoff_t index); 3512 struct page *f2fs_get_tmp_page(struct f2fs_sb_info *sbi, pgoff_t index); 3513 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi, 3514 block_t blkaddr, int type); 3515 int f2fs_ra_meta_pages(struct f2fs_sb_info *sbi, block_t start, int nrpages, 3516 int type, bool sync); 3517 void f2fs_ra_meta_pages_cond(struct f2fs_sb_info *sbi, pgoff_t index); 3518 long f2fs_sync_meta_pages(struct f2fs_sb_info *sbi, enum page_type type, 3519 long nr_to_write, enum iostat_type io_type); 3520 void f2fs_add_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type); 3521 void f2fs_remove_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type); 3522 void f2fs_release_ino_entry(struct f2fs_sb_info *sbi, bool all); 3523 bool f2fs_exist_written_data(struct f2fs_sb_info *sbi, nid_t ino, int mode); 3524 void f2fs_set_dirty_device(struct f2fs_sb_info *sbi, nid_t ino, 3525 unsigned int devidx, int type); 3526 bool f2fs_is_dirty_device(struct f2fs_sb_info *sbi, nid_t ino, 3527 unsigned int devidx, int type); 3528 int f2fs_sync_inode_meta(struct f2fs_sb_info *sbi); 3529 int f2fs_acquire_orphan_inode(struct f2fs_sb_info *sbi); 3530 void f2fs_release_orphan_inode(struct f2fs_sb_info *sbi); 3531 void f2fs_add_orphan_inode(struct inode *inode); 3532 void f2fs_remove_orphan_inode(struct f2fs_sb_info *sbi, nid_t ino); 3533 int f2fs_recover_orphan_inodes(struct f2fs_sb_info *sbi); 3534 int f2fs_get_valid_checkpoint(struct f2fs_sb_info *sbi); 3535 void f2fs_update_dirty_page(struct inode *inode, struct page *page); 3536 void f2fs_remove_dirty_inode(struct inode *inode); 3537 int f2fs_sync_dirty_inodes(struct f2fs_sb_info *sbi, enum inode_type type); 3538 void f2fs_wait_on_all_pages(struct f2fs_sb_info *sbi, int type); 3539 u64 f2fs_get_sectors_written(struct f2fs_sb_info *sbi); 3540 int f2fs_write_checkpoint(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3541 void f2fs_init_ino_entry_info(struct f2fs_sb_info *sbi); 3542 int __init f2fs_create_checkpoint_caches(void); 3543 void f2fs_destroy_checkpoint_caches(void); 3544 int f2fs_issue_checkpoint(struct f2fs_sb_info *sbi); 3545 int f2fs_start_ckpt_thread(struct f2fs_sb_info *sbi); 3546 void f2fs_stop_ckpt_thread(struct f2fs_sb_info *sbi); 3547 void f2fs_init_ckpt_req_control(struct f2fs_sb_info *sbi); 3548 3549 /* 3550 * data.c 3551 */ 3552 int __init f2fs_init_bioset(void); 3553 void f2fs_destroy_bioset(void); 3554 int f2fs_init_bio_entry_cache(void); 3555 void f2fs_destroy_bio_entry_cache(void); 3556 void f2fs_submit_bio(struct f2fs_sb_info *sbi, 3557 struct bio *bio, enum page_type type); 3558 void f2fs_submit_merged_write(struct f2fs_sb_info *sbi, enum page_type type); 3559 void f2fs_submit_merged_write_cond(struct f2fs_sb_info *sbi, 3560 struct inode *inode, struct page *page, 3561 nid_t ino, enum page_type type); 3562 void f2fs_submit_merged_ipu_write(struct f2fs_sb_info *sbi, 3563 struct bio **bio, struct page *page); 3564 void f2fs_flush_merged_writes(struct f2fs_sb_info *sbi); 3565 int f2fs_submit_page_bio(struct f2fs_io_info *fio); 3566 int f2fs_merge_page_bio(struct f2fs_io_info *fio); 3567 void f2fs_submit_page_write(struct f2fs_io_info *fio); 3568 struct block_device *f2fs_target_device(struct f2fs_sb_info *sbi, 3569 block_t blk_addr, struct bio *bio); 3570 int f2fs_target_device_index(struct f2fs_sb_info *sbi, block_t blkaddr); 3571 void f2fs_set_data_blkaddr(struct dnode_of_data *dn); 3572 void f2fs_update_data_blkaddr(struct dnode_of_data *dn, block_t blkaddr); 3573 int f2fs_reserve_new_blocks(struct dnode_of_data *dn, blkcnt_t count); 3574 int f2fs_reserve_new_block(struct dnode_of_data *dn); 3575 int f2fs_get_block(struct dnode_of_data *dn, pgoff_t index); 3576 int f2fs_preallocate_blocks(struct kiocb *iocb, struct iov_iter *from); 3577 int f2fs_reserve_block(struct dnode_of_data *dn, pgoff_t index); 3578 struct page *f2fs_get_read_data_page(struct inode *inode, pgoff_t index, 3579 int op_flags, bool for_write); 3580 struct page *f2fs_find_data_page(struct inode *inode, pgoff_t index); 3581 struct page *f2fs_get_lock_data_page(struct inode *inode, pgoff_t index, 3582 bool for_write); 3583 struct page *f2fs_get_new_data_page(struct inode *inode, 3584 struct page *ipage, pgoff_t index, bool new_i_size); 3585 int f2fs_do_write_data_page(struct f2fs_io_info *fio); 3586 void f2fs_do_map_lock(struct f2fs_sb_info *sbi, int flag, bool lock); 3587 int f2fs_map_blocks(struct inode *inode, struct f2fs_map_blocks *map, 3588 int create, int flag); 3589 int f2fs_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, 3590 u64 start, u64 len); 3591 int f2fs_encrypt_one_page(struct f2fs_io_info *fio); 3592 bool f2fs_should_update_inplace(struct inode *inode, struct f2fs_io_info *fio); 3593 bool f2fs_should_update_outplace(struct inode *inode, struct f2fs_io_info *fio); 3594 int f2fs_write_single_data_page(struct page *page, int *submitted, 3595 struct bio **bio, sector_t *last_block, 3596 struct writeback_control *wbc, 3597 enum iostat_type io_type, 3598 int compr_blocks, bool allow_balance); 3599 void f2fs_invalidate_page(struct page *page, unsigned int offset, 3600 unsigned int length); 3601 int f2fs_release_page(struct page *page, gfp_t wait); 3602 #ifdef CONFIG_MIGRATION 3603 int f2fs_migrate_page(struct address_space *mapping, struct page *newpage, 3604 struct page *page, enum migrate_mode mode); 3605 #endif 3606 bool f2fs_overwrite_io(struct inode *inode, loff_t pos, size_t len); 3607 void f2fs_clear_page_cache_dirty_tag(struct page *page); 3608 int f2fs_init_post_read_processing(void); 3609 void f2fs_destroy_post_read_processing(void); 3610 int f2fs_init_post_read_wq(struct f2fs_sb_info *sbi); 3611 void f2fs_destroy_post_read_wq(struct f2fs_sb_info *sbi); 3612 3613 /* 3614 * gc.c 3615 */ 3616 int f2fs_start_gc_thread(struct f2fs_sb_info *sbi); 3617 void f2fs_stop_gc_thread(struct f2fs_sb_info *sbi); 3618 block_t f2fs_start_bidx_of_node(unsigned int node_ofs, struct inode *inode); 3619 int f2fs_gc(struct f2fs_sb_info *sbi, bool sync, bool background, bool force, 3620 unsigned int segno); 3621 void f2fs_build_gc_manager(struct f2fs_sb_info *sbi); 3622 int f2fs_resize_fs(struct f2fs_sb_info *sbi, __u64 block_count); 3623 int __init f2fs_create_garbage_collection_cache(void); 3624 void f2fs_destroy_garbage_collection_cache(void); 3625 3626 /* 3627 * recovery.c 3628 */ 3629 int f2fs_recover_fsync_data(struct f2fs_sb_info *sbi, bool check_only); 3630 bool f2fs_space_for_roll_forward(struct f2fs_sb_info *sbi); 3631 int __init f2fs_create_recovery_cache(void); 3632 void f2fs_destroy_recovery_cache(void); 3633 3634 /* 3635 * debug.c 3636 */ 3637 #ifdef CONFIG_F2FS_STAT_FS 3638 struct f2fs_stat_info { 3639 struct list_head stat_list; 3640 struct f2fs_sb_info *sbi; 3641 int all_area_segs, sit_area_segs, nat_area_segs, ssa_area_segs; 3642 int main_area_segs, main_area_sections, main_area_zones; 3643 unsigned long long hit_largest, hit_cached, hit_rbtree; 3644 unsigned long long hit_total, total_ext; 3645 int ext_tree, zombie_tree, ext_node; 3646 int ndirty_node, ndirty_dent, ndirty_meta, ndirty_imeta; 3647 int ndirty_data, ndirty_qdata; 3648 int inmem_pages; 3649 unsigned int ndirty_dirs, ndirty_files, nquota_files, ndirty_all; 3650 int nats, dirty_nats, sits, dirty_sits; 3651 int free_nids, avail_nids, alloc_nids; 3652 int total_count, utilization; 3653 int bg_gc, nr_wb_cp_data, nr_wb_data; 3654 int nr_rd_data, nr_rd_node, nr_rd_meta; 3655 int nr_dio_read, nr_dio_write; 3656 unsigned int io_skip_bggc, other_skip_bggc; 3657 int nr_flushing, nr_flushed, flush_list_empty; 3658 int nr_discarding, nr_discarded; 3659 int nr_discard_cmd; 3660 unsigned int undiscard_blks; 3661 int nr_issued_ckpt, nr_total_ckpt, nr_queued_ckpt; 3662 unsigned int cur_ckpt_time, peak_ckpt_time; 3663 int inline_xattr, inline_inode, inline_dir, append, update, orphans; 3664 int compr_inode; 3665 unsigned long long compr_blocks; 3666 int aw_cnt, max_aw_cnt, vw_cnt, max_vw_cnt; 3667 unsigned int valid_count, valid_node_count, valid_inode_count, discard_blks; 3668 unsigned int bimodal, avg_vblocks; 3669 int util_free, util_valid, util_invalid; 3670 int rsvd_segs, overp_segs; 3671 int dirty_count, node_pages, meta_pages; 3672 int prefree_count, call_count, cp_count, bg_cp_count; 3673 int tot_segs, node_segs, data_segs, free_segs, free_secs; 3674 int bg_node_segs, bg_data_segs; 3675 int tot_blks, data_blks, node_blks; 3676 int bg_data_blks, bg_node_blks; 3677 unsigned long long skipped_atomic_files[2]; 3678 int curseg[NR_CURSEG_TYPE]; 3679 int cursec[NR_CURSEG_TYPE]; 3680 int curzone[NR_CURSEG_TYPE]; 3681 unsigned int dirty_seg[NR_CURSEG_TYPE]; 3682 unsigned int full_seg[NR_CURSEG_TYPE]; 3683 unsigned int valid_blks[NR_CURSEG_TYPE]; 3684 3685 unsigned int meta_count[META_MAX]; 3686 unsigned int segment_count[2]; 3687 unsigned int block_count[2]; 3688 unsigned int inplace_count; 3689 unsigned long long base_mem, cache_mem, page_mem; 3690 }; 3691 3692 static inline struct f2fs_stat_info *F2FS_STAT(struct f2fs_sb_info *sbi) 3693 { 3694 return (struct f2fs_stat_info *)sbi->stat_info; 3695 } 3696 3697 #define stat_inc_cp_count(si) ((si)->cp_count++) 3698 #define stat_inc_bg_cp_count(si) ((si)->bg_cp_count++) 3699 #define stat_inc_call_count(si) ((si)->call_count++) 3700 #define stat_inc_bggc_count(si) ((si)->bg_gc++) 3701 #define stat_io_skip_bggc_count(sbi) ((sbi)->io_skip_bggc++) 3702 #define stat_other_skip_bggc_count(sbi) ((sbi)->other_skip_bggc++) 3703 #define stat_inc_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]++) 3704 #define stat_dec_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]--) 3705 #define stat_inc_total_hit(sbi) (atomic64_inc(&(sbi)->total_hit_ext)) 3706 #define stat_inc_rbtree_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_rbtree)) 3707 #define stat_inc_largest_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_largest)) 3708 #define stat_inc_cached_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_cached)) 3709 #define stat_inc_inline_xattr(inode) \ 3710 do { \ 3711 if (f2fs_has_inline_xattr(inode)) \ 3712 (atomic_inc(&F2FS_I_SB(inode)->inline_xattr)); \ 3713 } while (0) 3714 #define stat_dec_inline_xattr(inode) \ 3715 do { \ 3716 if (f2fs_has_inline_xattr(inode)) \ 3717 (atomic_dec(&F2FS_I_SB(inode)->inline_xattr)); \ 3718 } while (0) 3719 #define stat_inc_inline_inode(inode) \ 3720 do { \ 3721 if (f2fs_has_inline_data(inode)) \ 3722 (atomic_inc(&F2FS_I_SB(inode)->inline_inode)); \ 3723 } while (0) 3724 #define stat_dec_inline_inode(inode) \ 3725 do { \ 3726 if (f2fs_has_inline_data(inode)) \ 3727 (atomic_dec(&F2FS_I_SB(inode)->inline_inode)); \ 3728 } while (0) 3729 #define stat_inc_inline_dir(inode) \ 3730 do { \ 3731 if (f2fs_has_inline_dentry(inode)) \ 3732 (atomic_inc(&F2FS_I_SB(inode)->inline_dir)); \ 3733 } while (0) 3734 #define stat_dec_inline_dir(inode) \ 3735 do { \ 3736 if (f2fs_has_inline_dentry(inode)) \ 3737 (atomic_dec(&F2FS_I_SB(inode)->inline_dir)); \ 3738 } while (0) 3739 #define stat_inc_compr_inode(inode) \ 3740 do { \ 3741 if (f2fs_compressed_file(inode)) \ 3742 (atomic_inc(&F2FS_I_SB(inode)->compr_inode)); \ 3743 } while (0) 3744 #define stat_dec_compr_inode(inode) \ 3745 do { \ 3746 if (f2fs_compressed_file(inode)) \ 3747 (atomic_dec(&F2FS_I_SB(inode)->compr_inode)); \ 3748 } while (0) 3749 #define stat_add_compr_blocks(inode, blocks) \ 3750 (atomic64_add(blocks, &F2FS_I_SB(inode)->compr_blocks)) 3751 #define stat_sub_compr_blocks(inode, blocks) \ 3752 (atomic64_sub(blocks, &F2FS_I_SB(inode)->compr_blocks)) 3753 #define stat_inc_meta_count(sbi, blkaddr) \ 3754 do { \ 3755 if (blkaddr < SIT_I(sbi)->sit_base_addr) \ 3756 atomic_inc(&(sbi)->meta_count[META_CP]); \ 3757 else if (blkaddr < NM_I(sbi)->nat_blkaddr) \ 3758 atomic_inc(&(sbi)->meta_count[META_SIT]); \ 3759 else if (blkaddr < SM_I(sbi)->ssa_blkaddr) \ 3760 atomic_inc(&(sbi)->meta_count[META_NAT]); \ 3761 else if (blkaddr < SM_I(sbi)->main_blkaddr) \ 3762 atomic_inc(&(sbi)->meta_count[META_SSA]); \ 3763 } while (0) 3764 #define stat_inc_seg_type(sbi, curseg) \ 3765 ((sbi)->segment_count[(curseg)->alloc_type]++) 3766 #define stat_inc_block_count(sbi, curseg) \ 3767 ((sbi)->block_count[(curseg)->alloc_type]++) 3768 #define stat_inc_inplace_blocks(sbi) \ 3769 (atomic_inc(&(sbi)->inplace_count)) 3770 #define stat_update_max_atomic_write(inode) \ 3771 do { \ 3772 int cur = F2FS_I_SB(inode)->atomic_files; \ 3773 int max = atomic_read(&F2FS_I_SB(inode)->max_aw_cnt); \ 3774 if (cur > max) \ 3775 atomic_set(&F2FS_I_SB(inode)->max_aw_cnt, cur); \ 3776 } while (0) 3777 #define stat_inc_volatile_write(inode) \ 3778 (atomic_inc(&F2FS_I_SB(inode)->vw_cnt)) 3779 #define stat_dec_volatile_write(inode) \ 3780 (atomic_dec(&F2FS_I_SB(inode)->vw_cnt)) 3781 #define stat_update_max_volatile_write(inode) \ 3782 do { \ 3783 int cur = atomic_read(&F2FS_I_SB(inode)->vw_cnt); \ 3784 int max = atomic_read(&F2FS_I_SB(inode)->max_vw_cnt); \ 3785 if (cur > max) \ 3786 atomic_set(&F2FS_I_SB(inode)->max_vw_cnt, cur); \ 3787 } while (0) 3788 #define stat_inc_seg_count(sbi, type, gc_type) \ 3789 do { \ 3790 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 3791 si->tot_segs++; \ 3792 if ((type) == SUM_TYPE_DATA) { \ 3793 si->data_segs++; \ 3794 si->bg_data_segs += (gc_type == BG_GC) ? 1 : 0; \ 3795 } else { \ 3796 si->node_segs++; \ 3797 si->bg_node_segs += (gc_type == BG_GC) ? 1 : 0; \ 3798 } \ 3799 } while (0) 3800 3801 #define stat_inc_tot_blk_count(si, blks) \ 3802 ((si)->tot_blks += (blks)) 3803 3804 #define stat_inc_data_blk_count(sbi, blks, gc_type) \ 3805 do { \ 3806 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 3807 stat_inc_tot_blk_count(si, blks); \ 3808 si->data_blks += (blks); \ 3809 si->bg_data_blks += ((gc_type) == BG_GC) ? (blks) : 0; \ 3810 } while (0) 3811 3812 #define stat_inc_node_blk_count(sbi, blks, gc_type) \ 3813 do { \ 3814 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 3815 stat_inc_tot_blk_count(si, blks); \ 3816 si->node_blks += (blks); \ 3817 si->bg_node_blks += ((gc_type) == BG_GC) ? (blks) : 0; \ 3818 } while (0) 3819 3820 int f2fs_build_stats(struct f2fs_sb_info *sbi); 3821 void f2fs_destroy_stats(struct f2fs_sb_info *sbi); 3822 void __init f2fs_create_root_stats(void); 3823 void f2fs_destroy_root_stats(void); 3824 void f2fs_update_sit_info(struct f2fs_sb_info *sbi); 3825 #else 3826 #define stat_inc_cp_count(si) do { } while (0) 3827 #define stat_inc_bg_cp_count(si) do { } while (0) 3828 #define stat_inc_call_count(si) do { } while (0) 3829 #define stat_inc_bggc_count(si) do { } while (0) 3830 #define stat_io_skip_bggc_count(sbi) do { } while (0) 3831 #define stat_other_skip_bggc_count(sbi) do { } while (0) 3832 #define stat_inc_dirty_inode(sbi, type) do { } while (0) 3833 #define stat_dec_dirty_inode(sbi, type) do { } while (0) 3834 #define stat_inc_total_hit(sbi) do { } while (0) 3835 #define stat_inc_rbtree_node_hit(sbi) do { } while (0) 3836 #define stat_inc_largest_node_hit(sbi) do { } while (0) 3837 #define stat_inc_cached_node_hit(sbi) do { } while (0) 3838 #define stat_inc_inline_xattr(inode) do { } while (0) 3839 #define stat_dec_inline_xattr(inode) do { } while (0) 3840 #define stat_inc_inline_inode(inode) do { } while (0) 3841 #define stat_dec_inline_inode(inode) do { } while (0) 3842 #define stat_inc_inline_dir(inode) do { } while (0) 3843 #define stat_dec_inline_dir(inode) do { } while (0) 3844 #define stat_inc_compr_inode(inode) do { } while (0) 3845 #define stat_dec_compr_inode(inode) do { } while (0) 3846 #define stat_add_compr_blocks(inode, blocks) do { } while (0) 3847 #define stat_sub_compr_blocks(inode, blocks) do { } while (0) 3848 #define stat_update_max_atomic_write(inode) do { } while (0) 3849 #define stat_inc_volatile_write(inode) do { } while (0) 3850 #define stat_dec_volatile_write(inode) do { } while (0) 3851 #define stat_update_max_volatile_write(inode) do { } while (0) 3852 #define stat_inc_meta_count(sbi, blkaddr) do { } while (0) 3853 #define stat_inc_seg_type(sbi, curseg) do { } while (0) 3854 #define stat_inc_block_count(sbi, curseg) do { } while (0) 3855 #define stat_inc_inplace_blocks(sbi) do { } while (0) 3856 #define stat_inc_seg_count(sbi, type, gc_type) do { } while (0) 3857 #define stat_inc_tot_blk_count(si, blks) do { } while (0) 3858 #define stat_inc_data_blk_count(sbi, blks, gc_type) do { } while (0) 3859 #define stat_inc_node_blk_count(sbi, blks, gc_type) do { } while (0) 3860 3861 static inline int f2fs_build_stats(struct f2fs_sb_info *sbi) { return 0; } 3862 static inline void f2fs_destroy_stats(struct f2fs_sb_info *sbi) { } 3863 static inline void __init f2fs_create_root_stats(void) { } 3864 static inline void f2fs_destroy_root_stats(void) { } 3865 static inline void f2fs_update_sit_info(struct f2fs_sb_info *sbi) {} 3866 #endif 3867 3868 extern const struct file_operations f2fs_dir_operations; 3869 extern const struct file_operations f2fs_file_operations; 3870 extern const struct inode_operations f2fs_file_inode_operations; 3871 extern const struct address_space_operations f2fs_dblock_aops; 3872 extern const struct address_space_operations f2fs_node_aops; 3873 extern const struct address_space_operations f2fs_meta_aops; 3874 extern const struct inode_operations f2fs_dir_inode_operations; 3875 extern const struct inode_operations f2fs_symlink_inode_operations; 3876 extern const struct inode_operations f2fs_encrypted_symlink_inode_operations; 3877 extern const struct inode_operations f2fs_special_inode_operations; 3878 extern struct kmem_cache *f2fs_inode_entry_slab; 3879 3880 /* 3881 * inline.c 3882 */ 3883 bool f2fs_may_inline_data(struct inode *inode); 3884 bool f2fs_may_inline_dentry(struct inode *inode); 3885 void f2fs_do_read_inline_data(struct page *page, struct page *ipage); 3886 void f2fs_truncate_inline_inode(struct inode *inode, 3887 struct page *ipage, u64 from); 3888 int f2fs_read_inline_data(struct inode *inode, struct page *page); 3889 int f2fs_convert_inline_page(struct dnode_of_data *dn, struct page *page); 3890 int f2fs_convert_inline_inode(struct inode *inode); 3891 int f2fs_try_convert_inline_dir(struct inode *dir, struct dentry *dentry); 3892 int f2fs_write_inline_data(struct inode *inode, struct page *page); 3893 int f2fs_recover_inline_data(struct inode *inode, struct page *npage); 3894 struct f2fs_dir_entry *f2fs_find_in_inline_dir(struct inode *dir, 3895 const struct f2fs_filename *fname, 3896 struct page **res_page); 3897 int f2fs_make_empty_inline_dir(struct inode *inode, struct inode *parent, 3898 struct page *ipage); 3899 int f2fs_add_inline_entry(struct inode *dir, const struct f2fs_filename *fname, 3900 struct inode *inode, nid_t ino, umode_t mode); 3901 void f2fs_delete_inline_entry(struct f2fs_dir_entry *dentry, 3902 struct page *page, struct inode *dir, 3903 struct inode *inode); 3904 bool f2fs_empty_inline_dir(struct inode *dir); 3905 int f2fs_read_inline_dir(struct file *file, struct dir_context *ctx, 3906 struct fscrypt_str *fstr); 3907 int f2fs_inline_data_fiemap(struct inode *inode, 3908 struct fiemap_extent_info *fieinfo, 3909 __u64 start, __u64 len); 3910 3911 /* 3912 * shrinker.c 3913 */ 3914 unsigned long f2fs_shrink_count(struct shrinker *shrink, 3915 struct shrink_control *sc); 3916 unsigned long f2fs_shrink_scan(struct shrinker *shrink, 3917 struct shrink_control *sc); 3918 void f2fs_join_shrinker(struct f2fs_sb_info *sbi); 3919 void f2fs_leave_shrinker(struct f2fs_sb_info *sbi); 3920 3921 /* 3922 * extent_cache.c 3923 */ 3924 struct rb_entry *f2fs_lookup_rb_tree(struct rb_root_cached *root, 3925 struct rb_entry *cached_re, unsigned int ofs); 3926 struct rb_node **f2fs_lookup_rb_tree_ext(struct f2fs_sb_info *sbi, 3927 struct rb_root_cached *root, 3928 struct rb_node **parent, 3929 unsigned long long key, bool *left_most); 3930 struct rb_node **f2fs_lookup_rb_tree_for_insert(struct f2fs_sb_info *sbi, 3931 struct rb_root_cached *root, 3932 struct rb_node **parent, 3933 unsigned int ofs, bool *leftmost); 3934 struct rb_entry *f2fs_lookup_rb_tree_ret(struct rb_root_cached *root, 3935 struct rb_entry *cached_re, unsigned int ofs, 3936 struct rb_entry **prev_entry, struct rb_entry **next_entry, 3937 struct rb_node ***insert_p, struct rb_node **insert_parent, 3938 bool force, bool *leftmost); 3939 bool f2fs_check_rb_tree_consistence(struct f2fs_sb_info *sbi, 3940 struct rb_root_cached *root, bool check_key); 3941 unsigned int f2fs_shrink_extent_tree(struct f2fs_sb_info *sbi, int nr_shrink); 3942 void f2fs_init_extent_tree(struct inode *inode, struct page *ipage); 3943 void f2fs_drop_extent_tree(struct inode *inode); 3944 unsigned int f2fs_destroy_extent_node(struct inode *inode); 3945 void f2fs_destroy_extent_tree(struct inode *inode); 3946 bool f2fs_lookup_extent_cache(struct inode *inode, pgoff_t pgofs, 3947 struct extent_info *ei); 3948 void f2fs_update_extent_cache(struct dnode_of_data *dn); 3949 void f2fs_update_extent_cache_range(struct dnode_of_data *dn, 3950 pgoff_t fofs, block_t blkaddr, unsigned int len); 3951 void f2fs_init_extent_cache_info(struct f2fs_sb_info *sbi); 3952 int __init f2fs_create_extent_cache(void); 3953 void f2fs_destroy_extent_cache(void); 3954 3955 /* 3956 * sysfs.c 3957 */ 3958 int __init f2fs_init_sysfs(void); 3959 void f2fs_exit_sysfs(void); 3960 int f2fs_register_sysfs(struct f2fs_sb_info *sbi); 3961 void f2fs_unregister_sysfs(struct f2fs_sb_info *sbi); 3962 3963 /* verity.c */ 3964 extern const struct fsverity_operations f2fs_verityops; 3965 3966 /* 3967 * crypto support 3968 */ 3969 static inline bool f2fs_encrypted_file(struct inode *inode) 3970 { 3971 return IS_ENCRYPTED(inode) && S_ISREG(inode->i_mode); 3972 } 3973 3974 static inline void f2fs_set_encrypted_inode(struct inode *inode) 3975 { 3976 #ifdef CONFIG_FS_ENCRYPTION 3977 file_set_encrypt(inode); 3978 f2fs_set_inode_flags(inode); 3979 #endif 3980 } 3981 3982 /* 3983 * Returns true if the reads of the inode's data need to undergo some 3984 * postprocessing step, like decryption or authenticity verification. 3985 */ 3986 static inline bool f2fs_post_read_required(struct inode *inode) 3987 { 3988 return f2fs_encrypted_file(inode) || fsverity_active(inode) || 3989 f2fs_compressed_file(inode); 3990 } 3991 3992 /* 3993 * compress.c 3994 */ 3995 #ifdef CONFIG_F2FS_FS_COMPRESSION 3996 bool f2fs_is_compressed_page(struct page *page); 3997 struct page *f2fs_compress_control_page(struct page *page); 3998 int f2fs_prepare_compress_overwrite(struct inode *inode, 3999 struct page **pagep, pgoff_t index, void **fsdata); 4000 bool f2fs_compress_write_end(struct inode *inode, void *fsdata, 4001 pgoff_t index, unsigned copied); 4002 int f2fs_truncate_partial_cluster(struct inode *inode, u64 from, bool lock); 4003 void f2fs_compress_write_end_io(struct bio *bio, struct page *page); 4004 bool f2fs_is_compress_backend_ready(struct inode *inode); 4005 int f2fs_init_compress_mempool(void); 4006 void f2fs_destroy_compress_mempool(void); 4007 void f2fs_end_read_compressed_page(struct page *page, bool failed); 4008 bool f2fs_cluster_is_empty(struct compress_ctx *cc); 4009 bool f2fs_cluster_can_merge_page(struct compress_ctx *cc, pgoff_t index); 4010 void f2fs_compress_ctx_add_page(struct compress_ctx *cc, struct page *page); 4011 int f2fs_write_multi_pages(struct compress_ctx *cc, 4012 int *submitted, 4013 struct writeback_control *wbc, 4014 enum iostat_type io_type); 4015 int f2fs_is_compressed_cluster(struct inode *inode, pgoff_t index); 4016 int f2fs_read_multi_pages(struct compress_ctx *cc, struct bio **bio_ret, 4017 unsigned nr_pages, sector_t *last_block_in_bio, 4018 bool is_readahead, bool for_write); 4019 struct decompress_io_ctx *f2fs_alloc_dic(struct compress_ctx *cc); 4020 void f2fs_decompress_end_io(struct decompress_io_ctx *dic, bool failed); 4021 void f2fs_put_page_dic(struct page *page); 4022 int f2fs_init_compress_ctx(struct compress_ctx *cc); 4023 void f2fs_destroy_compress_ctx(struct compress_ctx *cc, bool reuse); 4024 void f2fs_init_compress_info(struct f2fs_sb_info *sbi); 4025 int f2fs_init_page_array_cache(struct f2fs_sb_info *sbi); 4026 void f2fs_destroy_page_array_cache(struct f2fs_sb_info *sbi); 4027 int __init f2fs_init_compress_cache(void); 4028 void f2fs_destroy_compress_cache(void); 4029 #define inc_compr_inode_stat(inode) \ 4030 do { \ 4031 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); \ 4032 sbi->compr_new_inode++; \ 4033 } while (0) 4034 #define add_compr_block_stat(inode, blocks) \ 4035 do { \ 4036 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); \ 4037 int diff = F2FS_I(inode)->i_cluster_size - blocks; \ 4038 sbi->compr_written_block += blocks; \ 4039 sbi->compr_saved_block += diff; \ 4040 } while (0) 4041 #else 4042 static inline bool f2fs_is_compressed_page(struct page *page) { return false; } 4043 static inline bool f2fs_is_compress_backend_ready(struct inode *inode) 4044 { 4045 if (!f2fs_compressed_file(inode)) 4046 return true; 4047 /* not support compression */ 4048 return false; 4049 } 4050 static inline struct page *f2fs_compress_control_page(struct page *page) 4051 { 4052 WARN_ON_ONCE(1); 4053 return ERR_PTR(-EINVAL); 4054 } 4055 static inline int f2fs_init_compress_mempool(void) { return 0; } 4056 static inline void f2fs_destroy_compress_mempool(void) { } 4057 static inline void f2fs_end_read_compressed_page(struct page *page, bool failed) 4058 { 4059 WARN_ON_ONCE(1); 4060 } 4061 static inline void f2fs_put_page_dic(struct page *page) 4062 { 4063 WARN_ON_ONCE(1); 4064 } 4065 static inline int f2fs_init_page_array_cache(struct f2fs_sb_info *sbi) { return 0; } 4066 static inline void f2fs_destroy_page_array_cache(struct f2fs_sb_info *sbi) { } 4067 static inline int __init f2fs_init_compress_cache(void) { return 0; } 4068 static inline void f2fs_destroy_compress_cache(void) { } 4069 #define inc_compr_inode_stat(inode) do { } while (0) 4070 #endif 4071 4072 static inline void set_compress_context(struct inode *inode) 4073 { 4074 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 4075 4076 F2FS_I(inode)->i_compress_algorithm = 4077 F2FS_OPTION(sbi).compress_algorithm; 4078 F2FS_I(inode)->i_log_cluster_size = 4079 F2FS_OPTION(sbi).compress_log_size; 4080 F2FS_I(inode)->i_compress_flag = 4081 F2FS_OPTION(sbi).compress_chksum ? 4082 1 << COMPRESS_CHKSUM : 0; 4083 F2FS_I(inode)->i_cluster_size = 4084 1 << F2FS_I(inode)->i_log_cluster_size; 4085 if (F2FS_I(inode)->i_compress_algorithm == COMPRESS_LZ4 && 4086 F2FS_OPTION(sbi).compress_level) 4087 F2FS_I(inode)->i_compress_flag |= 4088 F2FS_OPTION(sbi).compress_level << 4089 COMPRESS_LEVEL_OFFSET; 4090 F2FS_I(inode)->i_flags |= F2FS_COMPR_FL; 4091 set_inode_flag(inode, FI_COMPRESSED_FILE); 4092 stat_inc_compr_inode(inode); 4093 inc_compr_inode_stat(inode); 4094 f2fs_mark_inode_dirty_sync(inode, true); 4095 } 4096 4097 static inline bool f2fs_disable_compressed_file(struct inode *inode) 4098 { 4099 struct f2fs_inode_info *fi = F2FS_I(inode); 4100 4101 if (!f2fs_compressed_file(inode)) 4102 return true; 4103 if (S_ISREG(inode->i_mode) && 4104 (get_dirty_pages(inode) || atomic_read(&fi->i_compr_blocks))) 4105 return false; 4106 4107 fi->i_flags &= ~F2FS_COMPR_FL; 4108 stat_dec_compr_inode(inode); 4109 clear_inode_flag(inode, FI_COMPRESSED_FILE); 4110 f2fs_mark_inode_dirty_sync(inode, true); 4111 return true; 4112 } 4113 4114 #define F2FS_FEATURE_FUNCS(name, flagname) \ 4115 static inline int f2fs_sb_has_##name(struct f2fs_sb_info *sbi) \ 4116 { \ 4117 return F2FS_HAS_FEATURE(sbi, F2FS_FEATURE_##flagname); \ 4118 } 4119 4120 F2FS_FEATURE_FUNCS(encrypt, ENCRYPT); 4121 F2FS_FEATURE_FUNCS(blkzoned, BLKZONED); 4122 F2FS_FEATURE_FUNCS(extra_attr, EXTRA_ATTR); 4123 F2FS_FEATURE_FUNCS(project_quota, PRJQUOTA); 4124 F2FS_FEATURE_FUNCS(inode_chksum, INODE_CHKSUM); 4125 F2FS_FEATURE_FUNCS(flexible_inline_xattr, FLEXIBLE_INLINE_XATTR); 4126 F2FS_FEATURE_FUNCS(quota_ino, QUOTA_INO); 4127 F2FS_FEATURE_FUNCS(inode_crtime, INODE_CRTIME); 4128 F2FS_FEATURE_FUNCS(lost_found, LOST_FOUND); 4129 F2FS_FEATURE_FUNCS(verity, VERITY); 4130 F2FS_FEATURE_FUNCS(sb_chksum, SB_CHKSUM); 4131 F2FS_FEATURE_FUNCS(casefold, CASEFOLD); 4132 F2FS_FEATURE_FUNCS(compression, COMPRESSION); 4133 F2FS_FEATURE_FUNCS(readonly, RO); 4134 4135 #ifdef CONFIG_BLK_DEV_ZONED 4136 static inline bool f2fs_blkz_is_seq(struct f2fs_sb_info *sbi, int devi, 4137 block_t blkaddr) 4138 { 4139 unsigned int zno = blkaddr >> sbi->log_blocks_per_blkz; 4140 4141 return test_bit(zno, FDEV(devi).blkz_seq); 4142 } 4143 #endif 4144 4145 static inline bool f2fs_hw_should_discard(struct f2fs_sb_info *sbi) 4146 { 4147 return f2fs_sb_has_blkzoned(sbi); 4148 } 4149 4150 static inline bool f2fs_bdev_support_discard(struct block_device *bdev) 4151 { 4152 return blk_queue_discard(bdev_get_queue(bdev)) || 4153 bdev_is_zoned(bdev); 4154 } 4155 4156 static inline bool f2fs_hw_support_discard(struct f2fs_sb_info *sbi) 4157 { 4158 int i; 4159 4160 if (!f2fs_is_multi_device(sbi)) 4161 return f2fs_bdev_support_discard(sbi->sb->s_bdev); 4162 4163 for (i = 0; i < sbi->s_ndevs; i++) 4164 if (f2fs_bdev_support_discard(FDEV(i).bdev)) 4165 return true; 4166 return false; 4167 } 4168 4169 static inline bool f2fs_realtime_discard_enable(struct f2fs_sb_info *sbi) 4170 { 4171 return (test_opt(sbi, DISCARD) && f2fs_hw_support_discard(sbi)) || 4172 f2fs_hw_should_discard(sbi); 4173 } 4174 4175 static inline bool f2fs_hw_is_readonly(struct f2fs_sb_info *sbi) 4176 { 4177 int i; 4178 4179 if (!f2fs_is_multi_device(sbi)) 4180 return bdev_read_only(sbi->sb->s_bdev); 4181 4182 for (i = 0; i < sbi->s_ndevs; i++) 4183 if (bdev_read_only(FDEV(i).bdev)) 4184 return true; 4185 return false; 4186 } 4187 4188 static inline bool f2fs_lfs_mode(struct f2fs_sb_info *sbi) 4189 { 4190 return F2FS_OPTION(sbi).fs_mode == FS_MODE_LFS; 4191 } 4192 4193 static inline bool f2fs_may_compress(struct inode *inode) 4194 { 4195 if (IS_SWAPFILE(inode) || f2fs_is_pinned_file(inode) || 4196 f2fs_is_atomic_file(inode) || 4197 f2fs_is_volatile_file(inode)) 4198 return false; 4199 return S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode); 4200 } 4201 4202 static inline void f2fs_i_compr_blocks_update(struct inode *inode, 4203 u64 blocks, bool add) 4204 { 4205 int diff = F2FS_I(inode)->i_cluster_size - blocks; 4206 struct f2fs_inode_info *fi = F2FS_I(inode); 4207 4208 /* don't update i_compr_blocks if saved blocks were released */ 4209 if (!add && !atomic_read(&fi->i_compr_blocks)) 4210 return; 4211 4212 if (add) { 4213 atomic_add(diff, &fi->i_compr_blocks); 4214 stat_add_compr_blocks(inode, diff); 4215 } else { 4216 atomic_sub(diff, &fi->i_compr_blocks); 4217 stat_sub_compr_blocks(inode, diff); 4218 } 4219 f2fs_mark_inode_dirty_sync(inode, true); 4220 } 4221 4222 static inline int block_unaligned_IO(struct inode *inode, 4223 struct kiocb *iocb, struct iov_iter *iter) 4224 { 4225 unsigned int i_blkbits = READ_ONCE(inode->i_blkbits); 4226 unsigned int blocksize_mask = (1 << i_blkbits) - 1; 4227 loff_t offset = iocb->ki_pos; 4228 unsigned long align = offset | iov_iter_alignment(iter); 4229 4230 return align & blocksize_mask; 4231 } 4232 4233 static inline int allow_outplace_dio(struct inode *inode, 4234 struct kiocb *iocb, struct iov_iter *iter) 4235 { 4236 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 4237 int rw = iov_iter_rw(iter); 4238 4239 return (f2fs_lfs_mode(sbi) && (rw == WRITE) && 4240 !block_unaligned_IO(inode, iocb, iter)); 4241 } 4242 4243 static inline bool f2fs_force_buffered_io(struct inode *inode, 4244 struct kiocb *iocb, struct iov_iter *iter) 4245 { 4246 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 4247 int rw = iov_iter_rw(iter); 4248 4249 if (f2fs_post_read_required(inode)) 4250 return true; 4251 if (f2fs_is_multi_device(sbi)) 4252 return true; 4253 /* 4254 * for blkzoned device, fallback direct IO to buffered IO, so 4255 * all IOs can be serialized by log-structured write. 4256 */ 4257 if (f2fs_sb_has_blkzoned(sbi)) 4258 return true; 4259 if (f2fs_lfs_mode(sbi) && (rw == WRITE)) { 4260 if (block_unaligned_IO(inode, iocb, iter)) 4261 return true; 4262 if (F2FS_IO_ALIGNED(sbi)) 4263 return true; 4264 } 4265 if (is_sbi_flag_set(F2FS_I_SB(inode), SBI_CP_DISABLED)) 4266 return true; 4267 4268 return false; 4269 } 4270 4271 static inline bool f2fs_need_verity(const struct inode *inode, pgoff_t idx) 4272 { 4273 return fsverity_active(inode) && 4274 idx < DIV_ROUND_UP(inode->i_size, PAGE_SIZE); 4275 } 4276 4277 #ifdef CONFIG_F2FS_FAULT_INJECTION 4278 extern void f2fs_build_fault_attr(struct f2fs_sb_info *sbi, unsigned int rate, 4279 unsigned int type); 4280 #else 4281 #define f2fs_build_fault_attr(sbi, rate, type) do { } while (0) 4282 #endif 4283 4284 static inline bool is_journalled_quota(struct f2fs_sb_info *sbi) 4285 { 4286 #ifdef CONFIG_QUOTA 4287 if (f2fs_sb_has_quota_ino(sbi)) 4288 return true; 4289 if (F2FS_OPTION(sbi).s_qf_names[USRQUOTA] || 4290 F2FS_OPTION(sbi).s_qf_names[GRPQUOTA] || 4291 F2FS_OPTION(sbi).s_qf_names[PRJQUOTA]) 4292 return true; 4293 #endif 4294 return false; 4295 } 4296 4297 #define EFSBADCRC EBADMSG /* Bad CRC detected */ 4298 #define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */ 4299 4300 #endif /* _LINUX_F2FS_H */ 4301