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