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 #define F2FS_MOUNT_COMPRESS_CACHE 0x40000000 102 103 #define F2FS_OPTION(sbi) ((sbi)->mount_opt) 104 #define clear_opt(sbi, option) (F2FS_OPTION(sbi).opt &= ~F2FS_MOUNT_##option) 105 #define set_opt(sbi, option) (F2FS_OPTION(sbi).opt |= F2FS_MOUNT_##option) 106 #define test_opt(sbi, option) (F2FS_OPTION(sbi).opt & F2FS_MOUNT_##option) 107 108 #define ver_after(a, b) (typecheck(unsigned long long, a) && \ 109 typecheck(unsigned long long, b) && \ 110 ((long long)((a) - (b)) > 0)) 111 112 typedef u32 block_t; /* 113 * should not change u32, since it is the on-disk block 114 * address format, __le32. 115 */ 116 typedef u32 nid_t; 117 118 #define COMPRESS_EXT_NUM 16 119 120 struct f2fs_mount_info { 121 unsigned int opt; 122 int write_io_size_bits; /* Write IO size bits */ 123 block_t root_reserved_blocks; /* root reserved blocks */ 124 kuid_t s_resuid; /* reserved blocks for uid */ 125 kgid_t s_resgid; /* reserved blocks for gid */ 126 int active_logs; /* # of active logs */ 127 int inline_xattr_size; /* inline xattr size */ 128 #ifdef CONFIG_F2FS_FAULT_INJECTION 129 struct f2fs_fault_info fault_info; /* For fault injection */ 130 #endif 131 #ifdef CONFIG_QUOTA 132 /* Names of quota files with journalled quota */ 133 char *s_qf_names[MAXQUOTAS]; 134 int s_jquota_fmt; /* Format of quota to use */ 135 #endif 136 /* For which write hints are passed down to block layer */ 137 int whint_mode; 138 int alloc_mode; /* segment allocation policy */ 139 int fsync_mode; /* fsync policy */ 140 int fs_mode; /* fs mode: LFS or ADAPTIVE */ 141 int bggc_mode; /* bggc mode: off, on or sync */ 142 struct fscrypt_dummy_policy dummy_enc_policy; /* test dummy encryption */ 143 block_t unusable_cap_perc; /* percentage for cap */ 144 block_t unusable_cap; /* Amount of space allowed to be 145 * unusable when disabling checkpoint 146 */ 147 148 /* For compression */ 149 unsigned char compress_algorithm; /* algorithm type */ 150 unsigned char compress_log_size; /* cluster log size */ 151 unsigned char compress_level; /* compress level */ 152 bool compress_chksum; /* compressed data chksum */ 153 unsigned char compress_ext_cnt; /* extension count */ 154 int compress_mode; /* compression mode */ 155 unsigned char extensions[COMPRESS_EXT_NUM][F2FS_EXTENSION_LEN]; /* extensions */ 156 }; 157 158 #define F2FS_FEATURE_ENCRYPT 0x0001 159 #define F2FS_FEATURE_BLKZONED 0x0002 160 #define F2FS_FEATURE_ATOMIC_WRITE 0x0004 161 #define F2FS_FEATURE_EXTRA_ATTR 0x0008 162 #define F2FS_FEATURE_PRJQUOTA 0x0010 163 #define F2FS_FEATURE_INODE_CHKSUM 0x0020 164 #define F2FS_FEATURE_FLEXIBLE_INLINE_XATTR 0x0040 165 #define F2FS_FEATURE_QUOTA_INO 0x0080 166 #define F2FS_FEATURE_INODE_CRTIME 0x0100 167 #define F2FS_FEATURE_LOST_FOUND 0x0200 168 #define F2FS_FEATURE_VERITY 0x0400 169 #define F2FS_FEATURE_SB_CHKSUM 0x0800 170 #define F2FS_FEATURE_CASEFOLD 0x1000 171 #define F2FS_FEATURE_COMPRESSION 0x2000 172 #define F2FS_FEATURE_RO 0x4000 173 174 #define __F2FS_HAS_FEATURE(raw_super, mask) \ 175 ((raw_super->feature & cpu_to_le32(mask)) != 0) 176 #define F2FS_HAS_FEATURE(sbi, mask) __F2FS_HAS_FEATURE(sbi->raw_super, mask) 177 #define F2FS_SET_FEATURE(sbi, mask) \ 178 (sbi->raw_super->feature |= cpu_to_le32(mask)) 179 #define F2FS_CLEAR_FEATURE(sbi, mask) \ 180 (sbi->raw_super->feature &= ~cpu_to_le32(mask)) 181 182 /* 183 * Default values for user and/or group using reserved blocks 184 */ 185 #define F2FS_DEF_RESUID 0 186 #define F2FS_DEF_RESGID 0 187 188 /* 189 * For checkpoint manager 190 */ 191 enum { 192 NAT_BITMAP, 193 SIT_BITMAP 194 }; 195 196 #define CP_UMOUNT 0x00000001 197 #define CP_FASTBOOT 0x00000002 198 #define CP_SYNC 0x00000004 199 #define CP_RECOVERY 0x00000008 200 #define CP_DISCARD 0x00000010 201 #define CP_TRIMMED 0x00000020 202 #define CP_PAUSE 0x00000040 203 #define CP_RESIZE 0x00000080 204 205 #define MAX_DISCARD_BLOCKS(sbi) BLKS_PER_SEC(sbi) 206 #define DEF_MAX_DISCARD_REQUEST 8 /* issue 8 discards per round */ 207 #define DEF_MIN_DISCARD_ISSUE_TIME 50 /* 50 ms, if exists */ 208 #define DEF_MID_DISCARD_ISSUE_TIME 500 /* 500 ms, if device busy */ 209 #define DEF_MAX_DISCARD_ISSUE_TIME 60000 /* 60 s, if no candidates */ 210 #define DEF_DISCARD_URGENT_UTIL 80 /* do more discard over 80% */ 211 #define DEF_CP_INTERVAL 60 /* 60 secs */ 212 #define DEF_IDLE_INTERVAL 5 /* 5 secs */ 213 #define DEF_DISABLE_INTERVAL 5 /* 5 secs */ 214 #define DEF_DISABLE_QUICK_INTERVAL 1 /* 1 secs */ 215 #define DEF_UMOUNT_DISCARD_TIMEOUT 5 /* 5 secs */ 216 217 struct cp_control { 218 int reason; 219 __u64 trim_start; 220 __u64 trim_end; 221 __u64 trim_minlen; 222 }; 223 224 /* 225 * indicate meta/data type 226 */ 227 enum { 228 META_CP, 229 META_NAT, 230 META_SIT, 231 META_SSA, 232 META_MAX, 233 META_POR, 234 DATA_GENERIC, /* check range only */ 235 DATA_GENERIC_ENHANCE, /* strong check on range and segment bitmap */ 236 DATA_GENERIC_ENHANCE_READ, /* 237 * strong check on range and segment 238 * bitmap but no warning due to race 239 * condition of read on truncated area 240 * by extent_cache 241 */ 242 META_GENERIC, 243 }; 244 245 /* for the list of ino */ 246 enum { 247 ORPHAN_INO, /* for orphan ino list */ 248 APPEND_INO, /* for append ino list */ 249 UPDATE_INO, /* for update ino list */ 250 TRANS_DIR_INO, /* for trasactions dir ino list */ 251 FLUSH_INO, /* for multiple device flushing */ 252 MAX_INO_ENTRY, /* max. list */ 253 }; 254 255 struct ino_entry { 256 struct list_head list; /* list head */ 257 nid_t ino; /* inode number */ 258 unsigned int dirty_device; /* dirty device bitmap */ 259 }; 260 261 /* for the list of inodes to be GCed */ 262 struct inode_entry { 263 struct list_head list; /* list head */ 264 struct inode *inode; /* vfs inode pointer */ 265 }; 266 267 struct fsync_node_entry { 268 struct list_head list; /* list head */ 269 struct page *page; /* warm node page pointer */ 270 unsigned int seq_id; /* sequence id */ 271 }; 272 273 struct ckpt_req { 274 struct completion wait; /* completion for checkpoint done */ 275 struct llist_node llnode; /* llist_node to be linked in wait queue */ 276 int ret; /* return code of checkpoint */ 277 ktime_t queue_time; /* request queued time */ 278 }; 279 280 struct ckpt_req_control { 281 struct task_struct *f2fs_issue_ckpt; /* checkpoint task */ 282 int ckpt_thread_ioprio; /* checkpoint merge thread ioprio */ 283 wait_queue_head_t ckpt_wait_queue; /* waiting queue for wake-up */ 284 atomic_t issued_ckpt; /* # of actually issued ckpts */ 285 atomic_t total_ckpt; /* # of total ckpts */ 286 atomic_t queued_ckpt; /* # of queued ckpts */ 287 struct llist_head issue_list; /* list for command issue */ 288 spinlock_t stat_lock; /* lock for below checkpoint time stats */ 289 unsigned int cur_time; /* cur wait time in msec for currently issued checkpoint */ 290 unsigned int peak_time; /* peak wait time in msec until now */ 291 }; 292 293 /* for the bitmap indicate blocks to be discarded */ 294 struct discard_entry { 295 struct list_head list; /* list head */ 296 block_t start_blkaddr; /* start blockaddr of current segment */ 297 unsigned char discard_map[SIT_VBLOCK_MAP_SIZE]; /* segment discard bitmap */ 298 }; 299 300 /* default discard granularity of inner discard thread, unit: block count */ 301 #define DEFAULT_DISCARD_GRANULARITY 16 302 303 /* max discard pend list number */ 304 #define MAX_PLIST_NUM 512 305 #define plist_idx(blk_num) ((blk_num) >= MAX_PLIST_NUM ? \ 306 (MAX_PLIST_NUM - 1) : ((blk_num) - 1)) 307 308 enum { 309 D_PREP, /* initial */ 310 D_PARTIAL, /* partially submitted */ 311 D_SUBMIT, /* all submitted */ 312 D_DONE, /* finished */ 313 }; 314 315 struct discard_info { 316 block_t lstart; /* logical start address */ 317 block_t len; /* length */ 318 block_t start; /* actual start address in dev */ 319 }; 320 321 struct discard_cmd { 322 struct rb_node rb_node; /* rb node located in rb-tree */ 323 union { 324 struct { 325 block_t lstart; /* logical start address */ 326 block_t len; /* length */ 327 block_t start; /* actual start address in dev */ 328 }; 329 struct discard_info di; /* discard info */ 330 331 }; 332 struct list_head list; /* command list */ 333 struct completion wait; /* compleation */ 334 struct block_device *bdev; /* bdev */ 335 unsigned short ref; /* reference count */ 336 unsigned char state; /* state */ 337 unsigned char queued; /* queued discard */ 338 int error; /* bio error */ 339 spinlock_t lock; /* for state/bio_ref updating */ 340 unsigned short bio_ref; /* bio reference count */ 341 }; 342 343 enum { 344 DPOLICY_BG, 345 DPOLICY_FORCE, 346 DPOLICY_FSTRIM, 347 DPOLICY_UMOUNT, 348 MAX_DPOLICY, 349 }; 350 351 struct discard_policy { 352 int type; /* type of discard */ 353 unsigned int min_interval; /* used for candidates exist */ 354 unsigned int mid_interval; /* used for device busy */ 355 unsigned int max_interval; /* used for candidates not exist */ 356 unsigned int max_requests; /* # of discards issued per round */ 357 unsigned int io_aware_gran; /* minimum granularity discard not be aware of I/O */ 358 bool io_aware; /* issue discard in idle time */ 359 bool sync; /* submit discard with REQ_SYNC flag */ 360 bool ordered; /* issue discard by lba order */ 361 bool timeout; /* discard timeout for put_super */ 362 unsigned int granularity; /* discard granularity */ 363 }; 364 365 struct discard_cmd_control { 366 struct task_struct *f2fs_issue_discard; /* discard thread */ 367 struct list_head entry_list; /* 4KB discard entry list */ 368 struct list_head pend_list[MAX_PLIST_NUM];/* store pending entries */ 369 struct list_head wait_list; /* store on-flushing entries */ 370 struct list_head fstrim_list; /* in-flight discard from fstrim */ 371 wait_queue_head_t discard_wait_queue; /* waiting queue for wake-up */ 372 unsigned int discard_wake; /* to wake up discard thread */ 373 struct mutex cmd_lock; 374 unsigned int nr_discards; /* # of discards in the list */ 375 unsigned int max_discards; /* max. discards to be issued */ 376 unsigned int discard_granularity; /* discard granularity */ 377 unsigned int undiscard_blks; /* # of undiscard blocks */ 378 unsigned int next_pos; /* next discard position */ 379 atomic_t issued_discard; /* # of issued discard */ 380 atomic_t queued_discard; /* # of queued discard */ 381 atomic_t discard_cmd_cnt; /* # of cached cmd count */ 382 struct rb_root_cached root; /* root of discard rb-tree */ 383 bool rbtree_check; /* config for consistence check */ 384 }; 385 386 /* for the list of fsync inodes, used only during recovery */ 387 struct fsync_inode_entry { 388 struct list_head list; /* list head */ 389 struct inode *inode; /* vfs inode pointer */ 390 block_t blkaddr; /* block address locating the last fsync */ 391 block_t last_dentry; /* block address locating the last dentry */ 392 }; 393 394 #define nats_in_cursum(jnl) (le16_to_cpu((jnl)->n_nats)) 395 #define sits_in_cursum(jnl) (le16_to_cpu((jnl)->n_sits)) 396 397 #define nat_in_journal(jnl, i) ((jnl)->nat_j.entries[i].ne) 398 #define nid_in_journal(jnl, i) ((jnl)->nat_j.entries[i].nid) 399 #define sit_in_journal(jnl, i) ((jnl)->sit_j.entries[i].se) 400 #define segno_in_journal(jnl, i) ((jnl)->sit_j.entries[i].segno) 401 402 #define MAX_NAT_JENTRIES(jnl) (NAT_JOURNAL_ENTRIES - nats_in_cursum(jnl)) 403 #define MAX_SIT_JENTRIES(jnl) (SIT_JOURNAL_ENTRIES - sits_in_cursum(jnl)) 404 405 static inline int update_nats_in_cursum(struct f2fs_journal *journal, int i) 406 { 407 int before = nats_in_cursum(journal); 408 409 journal->n_nats = cpu_to_le16(before + i); 410 return before; 411 } 412 413 static inline int update_sits_in_cursum(struct f2fs_journal *journal, int i) 414 { 415 int before = sits_in_cursum(journal); 416 417 journal->n_sits = cpu_to_le16(before + i); 418 return before; 419 } 420 421 static inline bool __has_cursum_space(struct f2fs_journal *journal, 422 int size, int type) 423 { 424 if (type == NAT_JOURNAL) 425 return size <= MAX_NAT_JENTRIES(journal); 426 return size <= MAX_SIT_JENTRIES(journal); 427 } 428 429 /* for inline stuff */ 430 #define DEF_INLINE_RESERVED_SIZE 1 431 static inline int get_extra_isize(struct inode *inode); 432 static inline int get_inline_xattr_addrs(struct inode *inode); 433 #define MAX_INLINE_DATA(inode) (sizeof(__le32) * \ 434 (CUR_ADDRS_PER_INODE(inode) - \ 435 get_inline_xattr_addrs(inode) - \ 436 DEF_INLINE_RESERVED_SIZE)) 437 438 /* for inline dir */ 439 #define NR_INLINE_DENTRY(inode) (MAX_INLINE_DATA(inode) * BITS_PER_BYTE / \ 440 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \ 441 BITS_PER_BYTE + 1)) 442 #define INLINE_DENTRY_BITMAP_SIZE(inode) \ 443 DIV_ROUND_UP(NR_INLINE_DENTRY(inode), BITS_PER_BYTE) 444 #define INLINE_RESERVED_SIZE(inode) (MAX_INLINE_DATA(inode) - \ 445 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \ 446 NR_INLINE_DENTRY(inode) + \ 447 INLINE_DENTRY_BITMAP_SIZE(inode))) 448 449 /* 450 * For INODE and NODE manager 451 */ 452 /* for directory operations */ 453 454 struct f2fs_filename { 455 /* 456 * The filename the user specified. This is NULL for some 457 * filesystem-internal operations, e.g. converting an inline directory 458 * to a non-inline one, or roll-forward recovering an encrypted dentry. 459 */ 460 const struct qstr *usr_fname; 461 462 /* 463 * The on-disk filename. For encrypted directories, this is encrypted. 464 * This may be NULL for lookups in an encrypted dir without the key. 465 */ 466 struct fscrypt_str disk_name; 467 468 /* The dirhash of this filename */ 469 f2fs_hash_t hash; 470 471 #ifdef CONFIG_FS_ENCRYPTION 472 /* 473 * For lookups in encrypted directories: either the buffer backing 474 * disk_name, or a buffer that holds the decoded no-key name. 475 */ 476 struct fscrypt_str crypto_buf; 477 #endif 478 #ifdef CONFIG_UNICODE 479 /* 480 * For casefolded directories: the casefolded name, but it's left NULL 481 * if the original name is not valid Unicode, if the directory is both 482 * casefolded and encrypted and its encryption key is unavailable, or if 483 * the filesystem is doing an internal operation where usr_fname is also 484 * NULL. In all these cases we fall back to treating the name as an 485 * opaque byte sequence. 486 */ 487 struct fscrypt_str cf_name; 488 #endif 489 }; 490 491 struct f2fs_dentry_ptr { 492 struct inode *inode; 493 void *bitmap; 494 struct f2fs_dir_entry *dentry; 495 __u8 (*filename)[F2FS_SLOT_LEN]; 496 int max; 497 int nr_bitmap; 498 }; 499 500 static inline void make_dentry_ptr_block(struct inode *inode, 501 struct f2fs_dentry_ptr *d, struct f2fs_dentry_block *t) 502 { 503 d->inode = inode; 504 d->max = NR_DENTRY_IN_BLOCK; 505 d->nr_bitmap = SIZE_OF_DENTRY_BITMAP; 506 d->bitmap = t->dentry_bitmap; 507 d->dentry = t->dentry; 508 d->filename = t->filename; 509 } 510 511 static inline void make_dentry_ptr_inline(struct inode *inode, 512 struct f2fs_dentry_ptr *d, void *t) 513 { 514 int entry_cnt = NR_INLINE_DENTRY(inode); 515 int bitmap_size = INLINE_DENTRY_BITMAP_SIZE(inode); 516 int reserved_size = INLINE_RESERVED_SIZE(inode); 517 518 d->inode = inode; 519 d->max = entry_cnt; 520 d->nr_bitmap = bitmap_size; 521 d->bitmap = t; 522 d->dentry = t + bitmap_size + reserved_size; 523 d->filename = t + bitmap_size + reserved_size + 524 SIZE_OF_DIR_ENTRY * entry_cnt; 525 } 526 527 /* 528 * XATTR_NODE_OFFSET stores xattrs to one node block per file keeping -1 529 * as its node offset to distinguish from index node blocks. 530 * But some bits are used to mark the node block. 531 */ 532 #define XATTR_NODE_OFFSET ((((unsigned int)-1) << OFFSET_BIT_SHIFT) \ 533 >> OFFSET_BIT_SHIFT) 534 enum { 535 ALLOC_NODE, /* allocate a new node page if needed */ 536 LOOKUP_NODE, /* look up a node without readahead */ 537 LOOKUP_NODE_RA, /* 538 * look up a node with readahead called 539 * by get_data_block. 540 */ 541 }; 542 543 #define DEFAULT_RETRY_IO_COUNT 8 /* maximum retry read IO count */ 544 545 /* congestion wait timeout value, default: 20ms */ 546 #define DEFAULT_IO_TIMEOUT (msecs_to_jiffies(20)) 547 548 /* maximum retry quota flush count */ 549 #define DEFAULT_RETRY_QUOTA_FLUSH_COUNT 8 550 551 #define F2FS_LINK_MAX 0xffffffff /* maximum link count per file */ 552 553 #define MAX_DIR_RA_PAGES 4 /* maximum ra pages of dir */ 554 555 /* for in-memory extent cache entry */ 556 #define F2FS_MIN_EXTENT_LEN 64 /* minimum extent length */ 557 558 /* number of extent info in extent cache we try to shrink */ 559 #define EXTENT_CACHE_SHRINK_NUMBER 128 560 561 struct rb_entry { 562 struct rb_node rb_node; /* rb node located in rb-tree */ 563 union { 564 struct { 565 unsigned int ofs; /* start offset of the entry */ 566 unsigned int len; /* length of the entry */ 567 }; 568 unsigned long long key; /* 64-bits key */ 569 } __packed; 570 }; 571 572 struct extent_info { 573 unsigned int fofs; /* start offset in a file */ 574 unsigned int len; /* length of the extent */ 575 u32 blk; /* start block address of the extent */ 576 }; 577 578 struct extent_node { 579 struct rb_node rb_node; /* rb node located in rb-tree */ 580 struct extent_info ei; /* extent info */ 581 struct list_head list; /* node in global extent list of sbi */ 582 struct extent_tree *et; /* extent tree pointer */ 583 }; 584 585 struct extent_tree { 586 nid_t ino; /* inode number */ 587 struct rb_root_cached root; /* root of extent info rb-tree */ 588 struct extent_node *cached_en; /* recently accessed extent node */ 589 struct extent_info largest; /* largested extent info */ 590 struct list_head list; /* to be used by sbi->zombie_list */ 591 rwlock_t lock; /* protect extent info rb-tree */ 592 atomic_t node_cnt; /* # of extent node in rb-tree*/ 593 bool largest_updated; /* largest extent updated */ 594 }; 595 596 /* 597 * This structure is taken from ext4_map_blocks. 598 * 599 * Note that, however, f2fs uses NEW and MAPPED flags for f2fs_map_blocks(). 600 */ 601 #define F2FS_MAP_NEW (1 << BH_New) 602 #define F2FS_MAP_MAPPED (1 << BH_Mapped) 603 #define F2FS_MAP_UNWRITTEN (1 << BH_Unwritten) 604 #define F2FS_MAP_FLAGS (F2FS_MAP_NEW | F2FS_MAP_MAPPED |\ 605 F2FS_MAP_UNWRITTEN) 606 607 struct f2fs_map_blocks { 608 block_t m_pblk; 609 block_t m_lblk; 610 unsigned int m_len; 611 unsigned int m_flags; 612 pgoff_t *m_next_pgofs; /* point next possible non-hole pgofs */ 613 pgoff_t *m_next_extent; /* point to next possible extent */ 614 int m_seg_type; 615 bool m_may_create; /* indicate it is from write path */ 616 }; 617 618 /* for flag in get_data_block */ 619 enum { 620 F2FS_GET_BLOCK_DEFAULT, 621 F2FS_GET_BLOCK_FIEMAP, 622 F2FS_GET_BLOCK_BMAP, 623 F2FS_GET_BLOCK_DIO, 624 F2FS_GET_BLOCK_PRE_DIO, 625 F2FS_GET_BLOCK_PRE_AIO, 626 F2FS_GET_BLOCK_PRECACHE, 627 }; 628 629 /* 630 * i_advise uses FADVISE_XXX_BIT. We can add additional hints later. 631 */ 632 #define FADVISE_COLD_BIT 0x01 633 #define FADVISE_LOST_PINO_BIT 0x02 634 #define FADVISE_ENCRYPT_BIT 0x04 635 #define FADVISE_ENC_NAME_BIT 0x08 636 #define FADVISE_KEEP_SIZE_BIT 0x10 637 #define FADVISE_HOT_BIT 0x20 638 #define FADVISE_VERITY_BIT 0x40 639 640 #define FADVISE_MODIFIABLE_BITS (FADVISE_COLD_BIT | FADVISE_HOT_BIT) 641 642 #define file_is_cold(inode) is_file(inode, FADVISE_COLD_BIT) 643 #define file_set_cold(inode) set_file(inode, FADVISE_COLD_BIT) 644 #define file_clear_cold(inode) clear_file(inode, FADVISE_COLD_BIT) 645 646 #define file_wrong_pino(inode) is_file(inode, FADVISE_LOST_PINO_BIT) 647 #define file_lost_pino(inode) set_file(inode, FADVISE_LOST_PINO_BIT) 648 #define file_got_pino(inode) clear_file(inode, FADVISE_LOST_PINO_BIT) 649 650 #define file_is_encrypt(inode) is_file(inode, FADVISE_ENCRYPT_BIT) 651 #define file_set_encrypt(inode) set_file(inode, FADVISE_ENCRYPT_BIT) 652 653 #define file_enc_name(inode) is_file(inode, FADVISE_ENC_NAME_BIT) 654 #define file_set_enc_name(inode) set_file(inode, FADVISE_ENC_NAME_BIT) 655 656 #define file_keep_isize(inode) is_file(inode, FADVISE_KEEP_SIZE_BIT) 657 #define file_set_keep_isize(inode) set_file(inode, FADVISE_KEEP_SIZE_BIT) 658 659 #define file_is_hot(inode) is_file(inode, FADVISE_HOT_BIT) 660 #define file_set_hot(inode) set_file(inode, FADVISE_HOT_BIT) 661 #define file_clear_hot(inode) clear_file(inode, FADVISE_HOT_BIT) 662 663 #define file_is_verity(inode) is_file(inode, FADVISE_VERITY_BIT) 664 #define file_set_verity(inode) set_file(inode, FADVISE_VERITY_BIT) 665 666 #define DEF_DIR_LEVEL 0 667 668 enum { 669 GC_FAILURE_PIN, 670 GC_FAILURE_ATOMIC, 671 MAX_GC_FAILURE 672 }; 673 674 /* used for f2fs_inode_info->flags */ 675 enum { 676 FI_NEW_INODE, /* indicate newly allocated inode */ 677 FI_DIRTY_INODE, /* indicate inode is dirty or not */ 678 FI_AUTO_RECOVER, /* indicate inode is recoverable */ 679 FI_DIRTY_DIR, /* indicate directory has dirty pages */ 680 FI_INC_LINK, /* need to increment i_nlink */ 681 FI_ACL_MODE, /* indicate acl mode */ 682 FI_NO_ALLOC, /* should not allocate any blocks */ 683 FI_FREE_NID, /* free allocated nide */ 684 FI_NO_EXTENT, /* not to use the extent cache */ 685 FI_INLINE_XATTR, /* used for inline xattr */ 686 FI_INLINE_DATA, /* used for inline data*/ 687 FI_INLINE_DENTRY, /* used for inline dentry */ 688 FI_APPEND_WRITE, /* inode has appended data */ 689 FI_UPDATE_WRITE, /* inode has in-place-update data */ 690 FI_NEED_IPU, /* used for ipu per file */ 691 FI_ATOMIC_FILE, /* indicate atomic file */ 692 FI_ATOMIC_COMMIT, /* indicate the state of atomical committing */ 693 FI_VOLATILE_FILE, /* indicate volatile file */ 694 FI_FIRST_BLOCK_WRITTEN, /* indicate #0 data block was written */ 695 FI_DROP_CACHE, /* drop dirty page cache */ 696 FI_DATA_EXIST, /* indicate data exists */ 697 FI_INLINE_DOTS, /* indicate inline dot dentries */ 698 FI_DO_DEFRAG, /* indicate defragment is running */ 699 FI_DIRTY_FILE, /* indicate regular/symlink has dirty pages */ 700 FI_NO_PREALLOC, /* indicate skipped preallocated blocks */ 701 FI_HOT_DATA, /* indicate file is hot */ 702 FI_EXTRA_ATTR, /* indicate file has extra attribute */ 703 FI_PROJ_INHERIT, /* indicate file inherits projectid */ 704 FI_PIN_FILE, /* indicate file should not be gced */ 705 FI_ATOMIC_REVOKE_REQUEST, /* request to drop atomic data */ 706 FI_VERITY_IN_PROGRESS, /* building fs-verity Merkle tree */ 707 FI_COMPRESSED_FILE, /* indicate file's data can be compressed */ 708 FI_COMPRESS_CORRUPT, /* indicate compressed cluster is corrupted */ 709 FI_MMAP_FILE, /* indicate file was mmapped */ 710 FI_ENABLE_COMPRESS, /* enable compression in "user" compression mode */ 711 FI_COMPRESS_RELEASED, /* compressed blocks were released */ 712 FI_MAX, /* max flag, never be used */ 713 }; 714 715 struct f2fs_inode_info { 716 struct inode vfs_inode; /* serve a vfs inode */ 717 unsigned long i_flags; /* keep an inode flags for ioctl */ 718 unsigned char i_advise; /* use to give file attribute hints */ 719 unsigned char i_dir_level; /* use for dentry level for large dir */ 720 unsigned int i_current_depth; /* only for directory depth */ 721 /* for gc failure statistic */ 722 unsigned int i_gc_failures[MAX_GC_FAILURE]; 723 unsigned int i_pino; /* parent inode number */ 724 umode_t i_acl_mode; /* keep file acl mode temporarily */ 725 726 /* Use below internally in f2fs*/ 727 unsigned long flags[BITS_TO_LONGS(FI_MAX)]; /* use to pass per-file flags */ 728 struct rw_semaphore i_sem; /* protect fi info */ 729 atomic_t dirty_pages; /* # of dirty pages */ 730 f2fs_hash_t chash; /* hash value of given file name */ 731 unsigned int clevel; /* maximum level of given file name */ 732 struct task_struct *task; /* lookup and create consistency */ 733 struct task_struct *cp_task; /* separate cp/wb IO stats*/ 734 nid_t i_xattr_nid; /* node id that contains xattrs */ 735 loff_t last_disk_size; /* lastly written file size */ 736 spinlock_t i_size_lock; /* protect last_disk_size */ 737 738 #ifdef CONFIG_QUOTA 739 struct dquot *i_dquot[MAXQUOTAS]; 740 741 /* quota space reservation, managed internally by quota code */ 742 qsize_t i_reserved_quota; 743 #endif 744 struct list_head dirty_list; /* dirty list for dirs and files */ 745 struct list_head gdirty_list; /* linked in global dirty list */ 746 struct list_head inmem_ilist; /* list for inmem inodes */ 747 struct list_head inmem_pages; /* inmemory pages managed by f2fs */ 748 struct task_struct *inmem_task; /* store inmemory task */ 749 struct mutex inmem_lock; /* lock for inmemory pages */ 750 struct extent_tree *extent_tree; /* cached extent_tree entry */ 751 752 /* avoid racing between foreground op and gc */ 753 struct rw_semaphore i_gc_rwsem[2]; 754 struct rw_semaphore i_mmap_sem; 755 struct rw_semaphore i_xattr_sem; /* avoid racing between reading and changing EAs */ 756 757 int i_extra_isize; /* size of extra space located in i_addr */ 758 kprojid_t i_projid; /* id for project quota */ 759 int i_inline_xattr_size; /* inline xattr size */ 760 struct timespec64 i_crtime; /* inode creation time */ 761 struct timespec64 i_disk_time[4];/* inode disk times */ 762 763 /* for file compress */ 764 atomic_t i_compr_blocks; /* # of compressed blocks */ 765 unsigned char i_compress_algorithm; /* algorithm type */ 766 unsigned char i_log_cluster_size; /* log of cluster size */ 767 unsigned char i_compress_level; /* compress level (lz4hc,zstd) */ 768 unsigned short i_compress_flag; /* compress flag */ 769 unsigned int i_cluster_size; /* cluster size */ 770 }; 771 772 static inline void get_extent_info(struct extent_info *ext, 773 struct f2fs_extent *i_ext) 774 { 775 ext->fofs = le32_to_cpu(i_ext->fofs); 776 ext->blk = le32_to_cpu(i_ext->blk); 777 ext->len = le32_to_cpu(i_ext->len); 778 } 779 780 static inline void set_raw_extent(struct extent_info *ext, 781 struct f2fs_extent *i_ext) 782 { 783 i_ext->fofs = cpu_to_le32(ext->fofs); 784 i_ext->blk = cpu_to_le32(ext->blk); 785 i_ext->len = cpu_to_le32(ext->len); 786 } 787 788 static inline void set_extent_info(struct extent_info *ei, unsigned int fofs, 789 u32 blk, unsigned int len) 790 { 791 ei->fofs = fofs; 792 ei->blk = blk; 793 ei->len = len; 794 } 795 796 static inline bool __is_discard_mergeable(struct discard_info *back, 797 struct discard_info *front, unsigned int max_len) 798 { 799 return (back->lstart + back->len == front->lstart) && 800 (back->len + front->len <= max_len); 801 } 802 803 static inline bool __is_discard_back_mergeable(struct discard_info *cur, 804 struct discard_info *back, unsigned int max_len) 805 { 806 return __is_discard_mergeable(back, cur, max_len); 807 } 808 809 static inline bool __is_discard_front_mergeable(struct discard_info *cur, 810 struct discard_info *front, unsigned int max_len) 811 { 812 return __is_discard_mergeable(cur, front, max_len); 813 } 814 815 static inline bool __is_extent_mergeable(struct extent_info *back, 816 struct extent_info *front) 817 { 818 return (back->fofs + back->len == front->fofs && 819 back->blk + back->len == front->blk); 820 } 821 822 static inline bool __is_back_mergeable(struct extent_info *cur, 823 struct extent_info *back) 824 { 825 return __is_extent_mergeable(back, cur); 826 } 827 828 static inline bool __is_front_mergeable(struct extent_info *cur, 829 struct extent_info *front) 830 { 831 return __is_extent_mergeable(cur, front); 832 } 833 834 extern void f2fs_mark_inode_dirty_sync(struct inode *inode, bool sync); 835 static inline void __try_update_largest_extent(struct extent_tree *et, 836 struct extent_node *en) 837 { 838 if (en->ei.len > et->largest.len) { 839 et->largest = en->ei; 840 et->largest_updated = true; 841 } 842 } 843 844 /* 845 * For free nid management 846 */ 847 enum nid_state { 848 FREE_NID, /* newly added to free nid list */ 849 PREALLOC_NID, /* it is preallocated */ 850 MAX_NID_STATE, 851 }; 852 853 enum nat_state { 854 TOTAL_NAT, 855 DIRTY_NAT, 856 RECLAIMABLE_NAT, 857 MAX_NAT_STATE, 858 }; 859 860 struct f2fs_nm_info { 861 block_t nat_blkaddr; /* base disk address of NAT */ 862 nid_t max_nid; /* maximum possible node ids */ 863 nid_t available_nids; /* # of available node ids */ 864 nid_t next_scan_nid; /* the next nid to be scanned */ 865 unsigned int ram_thresh; /* control the memory footprint */ 866 unsigned int ra_nid_pages; /* # of nid pages to be readaheaded */ 867 unsigned int dirty_nats_ratio; /* control dirty nats ratio threshold */ 868 869 /* NAT cache management */ 870 struct radix_tree_root nat_root;/* root of the nat entry cache */ 871 struct radix_tree_root nat_set_root;/* root of the nat set cache */ 872 struct rw_semaphore nat_tree_lock; /* protect nat entry tree */ 873 struct list_head nat_entries; /* cached nat entry list (clean) */ 874 spinlock_t nat_list_lock; /* protect clean nat entry list */ 875 unsigned int nat_cnt[MAX_NAT_STATE]; /* the # of cached nat entries */ 876 unsigned int nat_blocks; /* # of nat blocks */ 877 878 /* free node ids management */ 879 struct radix_tree_root free_nid_root;/* root of the free_nid cache */ 880 struct list_head free_nid_list; /* list for free nids excluding preallocated nids */ 881 unsigned int nid_cnt[MAX_NID_STATE]; /* the number of free node id */ 882 spinlock_t nid_list_lock; /* protect nid lists ops */ 883 struct mutex build_lock; /* lock for build free nids */ 884 unsigned char **free_nid_bitmap; 885 unsigned char *nat_block_bitmap; 886 unsigned short *free_nid_count; /* free nid count of NAT block */ 887 888 /* for checkpoint */ 889 char *nat_bitmap; /* NAT bitmap pointer */ 890 891 unsigned int nat_bits_blocks; /* # of nat bits blocks */ 892 unsigned char *nat_bits; /* NAT bits blocks */ 893 unsigned char *full_nat_bits; /* full NAT pages */ 894 unsigned char *empty_nat_bits; /* empty NAT pages */ 895 #ifdef CONFIG_F2FS_CHECK_FS 896 char *nat_bitmap_mir; /* NAT bitmap mirror */ 897 #endif 898 int bitmap_size; /* bitmap size */ 899 }; 900 901 /* 902 * this structure is used as one of function parameters. 903 * all the information are dedicated to a given direct node block determined 904 * by the data offset in a file. 905 */ 906 struct dnode_of_data { 907 struct inode *inode; /* vfs inode pointer */ 908 struct page *inode_page; /* its inode page, NULL is possible */ 909 struct page *node_page; /* cached direct node page */ 910 nid_t nid; /* node id of the direct node block */ 911 unsigned int ofs_in_node; /* data offset in the node page */ 912 bool inode_page_locked; /* inode page is locked or not */ 913 bool node_changed; /* is node block changed */ 914 char cur_level; /* level of hole node page */ 915 char max_level; /* level of current page located */ 916 block_t data_blkaddr; /* block address of the node block */ 917 }; 918 919 static inline void set_new_dnode(struct dnode_of_data *dn, struct inode *inode, 920 struct page *ipage, struct page *npage, nid_t nid) 921 { 922 memset(dn, 0, sizeof(*dn)); 923 dn->inode = inode; 924 dn->inode_page = ipage; 925 dn->node_page = npage; 926 dn->nid = nid; 927 } 928 929 /* 930 * For SIT manager 931 * 932 * By default, there are 6 active log areas across the whole main area. 933 * When considering hot and cold data separation to reduce cleaning overhead, 934 * we split 3 for data logs and 3 for node logs as hot, warm, and cold types, 935 * respectively. 936 * In the current design, you should not change the numbers intentionally. 937 * Instead, as a mount option such as active_logs=x, you can use 2, 4, and 6 938 * logs individually according to the underlying devices. (default: 6) 939 * Just in case, on-disk layout covers maximum 16 logs that consist of 8 for 940 * data and 8 for node logs. 941 */ 942 #define NR_CURSEG_DATA_TYPE (3) 943 #define NR_CURSEG_NODE_TYPE (3) 944 #define NR_CURSEG_INMEM_TYPE (2) 945 #define NR_CURSEG_RO_TYPE (2) 946 #define NR_CURSEG_PERSIST_TYPE (NR_CURSEG_DATA_TYPE + NR_CURSEG_NODE_TYPE) 947 #define NR_CURSEG_TYPE (NR_CURSEG_INMEM_TYPE + NR_CURSEG_PERSIST_TYPE) 948 949 enum { 950 CURSEG_HOT_DATA = 0, /* directory entry blocks */ 951 CURSEG_WARM_DATA, /* data blocks */ 952 CURSEG_COLD_DATA, /* multimedia or GCed data blocks */ 953 CURSEG_HOT_NODE, /* direct node blocks of directory files */ 954 CURSEG_WARM_NODE, /* direct node blocks of normal files */ 955 CURSEG_COLD_NODE, /* indirect node blocks */ 956 NR_PERSISTENT_LOG, /* number of persistent log */ 957 CURSEG_COLD_DATA_PINNED = NR_PERSISTENT_LOG, 958 /* pinned file that needs consecutive block address */ 959 CURSEG_ALL_DATA_ATGC, /* SSR alloctor in hot/warm/cold data area */ 960 NO_CHECK_TYPE, /* number of persistent & inmem log */ 961 }; 962 963 struct flush_cmd { 964 struct completion wait; 965 struct llist_node llnode; 966 nid_t ino; 967 int ret; 968 }; 969 970 struct flush_cmd_control { 971 struct task_struct *f2fs_issue_flush; /* flush thread */ 972 wait_queue_head_t flush_wait_queue; /* waiting queue for wake-up */ 973 atomic_t issued_flush; /* # of issued flushes */ 974 atomic_t queued_flush; /* # of queued flushes */ 975 struct llist_head issue_list; /* list for command issue */ 976 struct llist_node *dispatch_list; /* list for command dispatch */ 977 }; 978 979 struct f2fs_sm_info { 980 struct sit_info *sit_info; /* whole segment information */ 981 struct free_segmap_info *free_info; /* free segment information */ 982 struct dirty_seglist_info *dirty_info; /* dirty segment information */ 983 struct curseg_info *curseg_array; /* active segment information */ 984 985 struct rw_semaphore curseg_lock; /* for preventing curseg change */ 986 987 block_t seg0_blkaddr; /* block address of 0'th segment */ 988 block_t main_blkaddr; /* start block address of main area */ 989 block_t ssa_blkaddr; /* start block address of SSA area */ 990 991 unsigned int segment_count; /* total # of segments */ 992 unsigned int main_segments; /* # of segments in main area */ 993 unsigned int reserved_segments; /* # of reserved segments */ 994 unsigned int ovp_segments; /* # of overprovision segments */ 995 996 /* a threshold to reclaim prefree segments */ 997 unsigned int rec_prefree_segments; 998 999 /* for batched trimming */ 1000 unsigned int trim_sections; /* # of sections to trim */ 1001 1002 struct list_head sit_entry_set; /* sit entry set list */ 1003 1004 unsigned int ipu_policy; /* in-place-update policy */ 1005 unsigned int min_ipu_util; /* in-place-update threshold */ 1006 unsigned int min_fsync_blocks; /* threshold for fsync */ 1007 unsigned int min_seq_blocks; /* threshold for sequential blocks */ 1008 unsigned int min_hot_blocks; /* threshold for hot block allocation */ 1009 unsigned int min_ssr_sections; /* threshold to trigger SSR allocation */ 1010 1011 /* for flush command control */ 1012 struct flush_cmd_control *fcc_info; 1013 1014 /* for discard command control */ 1015 struct discard_cmd_control *dcc_info; 1016 }; 1017 1018 /* 1019 * For superblock 1020 */ 1021 /* 1022 * COUNT_TYPE for monitoring 1023 * 1024 * f2fs monitors the number of several block types such as on-writeback, 1025 * dirty dentry blocks, dirty node blocks, and dirty meta blocks. 1026 */ 1027 #define WB_DATA_TYPE(p) (__is_cp_guaranteed(p) ? F2FS_WB_CP_DATA : F2FS_WB_DATA) 1028 enum count_type { 1029 F2FS_DIRTY_DENTS, 1030 F2FS_DIRTY_DATA, 1031 F2FS_DIRTY_QDATA, 1032 F2FS_DIRTY_NODES, 1033 F2FS_DIRTY_META, 1034 F2FS_INMEM_PAGES, 1035 F2FS_DIRTY_IMETA, 1036 F2FS_WB_CP_DATA, 1037 F2FS_WB_DATA, 1038 F2FS_RD_DATA, 1039 F2FS_RD_NODE, 1040 F2FS_RD_META, 1041 F2FS_DIO_WRITE, 1042 F2FS_DIO_READ, 1043 NR_COUNT_TYPE, 1044 }; 1045 1046 /* 1047 * The below are the page types of bios used in submit_bio(). 1048 * The available types are: 1049 * DATA User data pages. It operates as async mode. 1050 * NODE Node pages. It operates as async mode. 1051 * META FS metadata pages such as SIT, NAT, CP. 1052 * NR_PAGE_TYPE The number of page types. 1053 * META_FLUSH Make sure the previous pages are written 1054 * with waiting the bio's completion 1055 * ... Only can be used with META. 1056 */ 1057 #define PAGE_TYPE_OF_BIO(type) ((type) > META ? META : (type)) 1058 enum page_type { 1059 DATA, 1060 NODE, 1061 META, 1062 NR_PAGE_TYPE, 1063 META_FLUSH, 1064 INMEM, /* the below types are used by tracepoints only. */ 1065 INMEM_DROP, 1066 INMEM_INVALIDATE, 1067 INMEM_REVOKE, 1068 IPU, 1069 OPU, 1070 }; 1071 1072 enum temp_type { 1073 HOT = 0, /* must be zero for meta bio */ 1074 WARM, 1075 COLD, 1076 NR_TEMP_TYPE, 1077 }; 1078 1079 enum need_lock_type { 1080 LOCK_REQ = 0, 1081 LOCK_DONE, 1082 LOCK_RETRY, 1083 }; 1084 1085 enum cp_reason_type { 1086 CP_NO_NEEDED, 1087 CP_NON_REGULAR, 1088 CP_COMPRESSED, 1089 CP_HARDLINK, 1090 CP_SB_NEED_CP, 1091 CP_WRONG_PINO, 1092 CP_NO_SPC_ROLL, 1093 CP_NODE_NEED_CP, 1094 CP_FASTBOOT_MODE, 1095 CP_SPEC_LOG_NUM, 1096 CP_RECOVER_DIR, 1097 }; 1098 1099 enum iostat_type { 1100 /* WRITE IO */ 1101 APP_DIRECT_IO, /* app direct write IOs */ 1102 APP_BUFFERED_IO, /* app buffered write IOs */ 1103 APP_WRITE_IO, /* app write IOs */ 1104 APP_MAPPED_IO, /* app mapped IOs */ 1105 FS_DATA_IO, /* data IOs from kworker/fsync/reclaimer */ 1106 FS_NODE_IO, /* node IOs from kworker/fsync/reclaimer */ 1107 FS_META_IO, /* meta IOs from kworker/reclaimer */ 1108 FS_GC_DATA_IO, /* data IOs from forground gc */ 1109 FS_GC_NODE_IO, /* node IOs from forground gc */ 1110 FS_CP_DATA_IO, /* data IOs from checkpoint */ 1111 FS_CP_NODE_IO, /* node IOs from checkpoint */ 1112 FS_CP_META_IO, /* meta IOs from checkpoint */ 1113 1114 /* READ IO */ 1115 APP_DIRECT_READ_IO, /* app direct read IOs */ 1116 APP_BUFFERED_READ_IO, /* app buffered read IOs */ 1117 APP_READ_IO, /* app read IOs */ 1118 APP_MAPPED_READ_IO, /* app mapped read IOs */ 1119 FS_DATA_READ_IO, /* data read IOs */ 1120 FS_GDATA_READ_IO, /* data read IOs from background gc */ 1121 FS_CDATA_READ_IO, /* compressed data read IOs */ 1122 FS_NODE_READ_IO, /* node read IOs */ 1123 FS_META_READ_IO, /* meta read IOs */ 1124 1125 /* other */ 1126 FS_DISCARD, /* discard */ 1127 NR_IO_TYPE, 1128 }; 1129 1130 struct f2fs_io_info { 1131 struct f2fs_sb_info *sbi; /* f2fs_sb_info pointer */ 1132 nid_t ino; /* inode number */ 1133 enum page_type type; /* contains DATA/NODE/META/META_FLUSH */ 1134 enum temp_type temp; /* contains HOT/WARM/COLD */ 1135 int op; /* contains REQ_OP_ */ 1136 int op_flags; /* req_flag_bits */ 1137 block_t new_blkaddr; /* new block address to be written */ 1138 block_t old_blkaddr; /* old block address before Cow */ 1139 struct page *page; /* page to be written */ 1140 struct page *encrypted_page; /* encrypted page */ 1141 struct page *compressed_page; /* compressed page */ 1142 struct list_head list; /* serialize IOs */ 1143 bool submitted; /* indicate IO submission */ 1144 int need_lock; /* indicate we need to lock cp_rwsem */ 1145 bool in_list; /* indicate fio is in io_list */ 1146 bool is_por; /* indicate IO is from recovery or not */ 1147 bool retry; /* need to reallocate block address */ 1148 int compr_blocks; /* # of compressed block addresses */ 1149 bool encrypted; /* indicate file is encrypted */ 1150 enum iostat_type io_type; /* io type */ 1151 struct writeback_control *io_wbc; /* writeback control */ 1152 struct bio **bio; /* bio for ipu */ 1153 sector_t *last_block; /* last block number in bio */ 1154 unsigned char version; /* version of the node */ 1155 }; 1156 1157 struct bio_entry { 1158 struct bio *bio; 1159 struct list_head list; 1160 }; 1161 1162 #define is_read_io(rw) ((rw) == READ) 1163 struct f2fs_bio_info { 1164 struct f2fs_sb_info *sbi; /* f2fs superblock */ 1165 struct bio *bio; /* bios to merge */ 1166 sector_t last_block_in_bio; /* last block number */ 1167 struct f2fs_io_info fio; /* store buffered io info. */ 1168 struct rw_semaphore io_rwsem; /* blocking op for bio */ 1169 spinlock_t io_lock; /* serialize DATA/NODE IOs */ 1170 struct list_head io_list; /* track fios */ 1171 struct list_head bio_list; /* bio entry list head */ 1172 struct rw_semaphore bio_list_lock; /* lock to protect bio entry list */ 1173 }; 1174 1175 #define FDEV(i) (sbi->devs[i]) 1176 #define RDEV(i) (raw_super->devs[i]) 1177 struct f2fs_dev_info { 1178 struct block_device *bdev; 1179 char path[MAX_PATH_LEN]; 1180 unsigned int total_segments; 1181 block_t start_blk; 1182 block_t end_blk; 1183 #ifdef CONFIG_BLK_DEV_ZONED 1184 unsigned int nr_blkz; /* Total number of zones */ 1185 unsigned long *blkz_seq; /* Bitmap indicating sequential zones */ 1186 block_t *zone_capacity_blocks; /* Array of zone capacity in blks */ 1187 #endif 1188 }; 1189 1190 enum inode_type { 1191 DIR_INODE, /* for dirty dir inode */ 1192 FILE_INODE, /* for dirty regular/symlink inode */ 1193 DIRTY_META, /* for all dirtied inode metadata */ 1194 ATOMIC_FILE, /* for all atomic files */ 1195 NR_INODE_TYPE, 1196 }; 1197 1198 /* for inner inode cache management */ 1199 struct inode_management { 1200 struct radix_tree_root ino_root; /* ino entry array */ 1201 spinlock_t ino_lock; /* for ino entry lock */ 1202 struct list_head ino_list; /* inode list head */ 1203 unsigned long ino_num; /* number of entries */ 1204 }; 1205 1206 /* for GC_AT */ 1207 struct atgc_management { 1208 bool atgc_enabled; /* ATGC is enabled or not */ 1209 struct rb_root_cached root; /* root of victim rb-tree */ 1210 struct list_head victim_list; /* linked with all victim entries */ 1211 unsigned int victim_count; /* victim count in rb-tree */ 1212 unsigned int candidate_ratio; /* candidate ratio */ 1213 unsigned int max_candidate_count; /* max candidate count */ 1214 unsigned int age_weight; /* age weight, vblock_weight = 100 - age_weight */ 1215 unsigned long long age_threshold; /* age threshold */ 1216 }; 1217 1218 /* For s_flag in struct f2fs_sb_info */ 1219 enum { 1220 SBI_IS_DIRTY, /* dirty flag for checkpoint */ 1221 SBI_IS_CLOSE, /* specify unmounting */ 1222 SBI_NEED_FSCK, /* need fsck.f2fs to fix */ 1223 SBI_POR_DOING, /* recovery is doing or not */ 1224 SBI_NEED_SB_WRITE, /* need to recover superblock */ 1225 SBI_NEED_CP, /* need to checkpoint */ 1226 SBI_IS_SHUTDOWN, /* shutdown by ioctl */ 1227 SBI_IS_RECOVERED, /* recovered orphan/data */ 1228 SBI_CP_DISABLED, /* CP was disabled last mount */ 1229 SBI_CP_DISABLED_QUICK, /* CP was disabled quickly */ 1230 SBI_QUOTA_NEED_FLUSH, /* need to flush quota info in CP */ 1231 SBI_QUOTA_SKIP_FLUSH, /* skip flushing quota in current CP */ 1232 SBI_QUOTA_NEED_REPAIR, /* quota file may be corrupted */ 1233 SBI_IS_RESIZEFS, /* resizefs is in process */ 1234 }; 1235 1236 enum { 1237 CP_TIME, 1238 REQ_TIME, 1239 DISCARD_TIME, 1240 GC_TIME, 1241 DISABLE_TIME, 1242 UMOUNT_DISCARD_TIMEOUT, 1243 MAX_TIME, 1244 }; 1245 1246 enum { 1247 GC_NORMAL, 1248 GC_IDLE_CB, 1249 GC_IDLE_GREEDY, 1250 GC_IDLE_AT, 1251 GC_URGENT_HIGH, 1252 GC_URGENT_LOW, 1253 }; 1254 1255 enum { 1256 BGGC_MODE_ON, /* background gc is on */ 1257 BGGC_MODE_OFF, /* background gc is off */ 1258 BGGC_MODE_SYNC, /* 1259 * background gc is on, migrating blocks 1260 * like foreground gc 1261 */ 1262 }; 1263 1264 enum { 1265 FS_MODE_ADAPTIVE, /* use both lfs/ssr allocation */ 1266 FS_MODE_LFS, /* use lfs allocation only */ 1267 }; 1268 1269 enum { 1270 WHINT_MODE_OFF, /* not pass down write hints */ 1271 WHINT_MODE_USER, /* try to pass down hints given by users */ 1272 WHINT_MODE_FS, /* pass down hints with F2FS policy */ 1273 }; 1274 1275 enum { 1276 ALLOC_MODE_DEFAULT, /* stay default */ 1277 ALLOC_MODE_REUSE, /* reuse segments as much as possible */ 1278 }; 1279 1280 enum fsync_mode { 1281 FSYNC_MODE_POSIX, /* fsync follows posix semantics */ 1282 FSYNC_MODE_STRICT, /* fsync behaves in line with ext4 */ 1283 FSYNC_MODE_NOBARRIER, /* fsync behaves nobarrier based on posix */ 1284 }; 1285 1286 enum { 1287 COMPR_MODE_FS, /* 1288 * automatically compress compression 1289 * enabled files 1290 */ 1291 COMPR_MODE_USER, /* 1292 * automatical compression is disabled. 1293 * user can control the file compression 1294 * using ioctls 1295 */ 1296 }; 1297 1298 static inline int f2fs_test_bit(unsigned int nr, char *addr); 1299 static inline void f2fs_set_bit(unsigned int nr, char *addr); 1300 static inline void f2fs_clear_bit(unsigned int nr, char *addr); 1301 1302 /* 1303 * Layout of f2fs page.private: 1304 * 1305 * Layout A: lowest bit should be 1 1306 * | bit0 = 1 | bit1 | bit2 | ... | bit MAX | private data .... | 1307 * bit 0 PAGE_PRIVATE_NOT_POINTER 1308 * bit 1 PAGE_PRIVATE_ATOMIC_WRITE 1309 * bit 2 PAGE_PRIVATE_DUMMY_WRITE 1310 * bit 3 PAGE_PRIVATE_ONGOING_MIGRATION 1311 * bit 4 PAGE_PRIVATE_INLINE_INODE 1312 * bit 5 PAGE_PRIVATE_REF_RESOURCE 1313 * bit 6- f2fs private data 1314 * 1315 * Layout B: lowest bit should be 0 1316 * page.private is a wrapped pointer. 1317 */ 1318 enum { 1319 PAGE_PRIVATE_NOT_POINTER, /* private contains non-pointer data */ 1320 PAGE_PRIVATE_ATOMIC_WRITE, /* data page from atomic write path */ 1321 PAGE_PRIVATE_DUMMY_WRITE, /* data page for padding aligned IO */ 1322 PAGE_PRIVATE_ONGOING_MIGRATION, /* data page which is on-going migrating */ 1323 PAGE_PRIVATE_INLINE_INODE, /* inode page contains inline data */ 1324 PAGE_PRIVATE_REF_RESOURCE, /* dirty page has referenced resources */ 1325 PAGE_PRIVATE_MAX 1326 }; 1327 1328 #define PAGE_PRIVATE_GET_FUNC(name, flagname) \ 1329 static inline bool page_private_##name(struct page *page) \ 1330 { \ 1331 return test_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)) && \ 1332 test_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \ 1333 } 1334 1335 #define PAGE_PRIVATE_SET_FUNC(name, flagname) \ 1336 static inline void set_page_private_##name(struct page *page) \ 1337 { \ 1338 if (!PagePrivate(page)) { \ 1339 get_page(page); \ 1340 SetPagePrivate(page); \ 1341 } \ 1342 set_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)); \ 1343 set_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \ 1344 } 1345 1346 #define PAGE_PRIVATE_CLEAR_FUNC(name, flagname) \ 1347 static inline void clear_page_private_##name(struct page *page) \ 1348 { \ 1349 clear_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \ 1350 if (page_private(page) == 1 << PAGE_PRIVATE_NOT_POINTER) { \ 1351 set_page_private(page, 0); \ 1352 if (PagePrivate(page)) { \ 1353 ClearPagePrivate(page); \ 1354 put_page(page); \ 1355 }\ 1356 } \ 1357 } 1358 1359 PAGE_PRIVATE_GET_FUNC(nonpointer, NOT_POINTER); 1360 PAGE_PRIVATE_GET_FUNC(reference, REF_RESOURCE); 1361 PAGE_PRIVATE_GET_FUNC(inline, INLINE_INODE); 1362 PAGE_PRIVATE_GET_FUNC(gcing, ONGOING_MIGRATION); 1363 PAGE_PRIVATE_GET_FUNC(atomic, ATOMIC_WRITE); 1364 PAGE_PRIVATE_GET_FUNC(dummy, DUMMY_WRITE); 1365 1366 PAGE_PRIVATE_SET_FUNC(reference, REF_RESOURCE); 1367 PAGE_PRIVATE_SET_FUNC(inline, INLINE_INODE); 1368 PAGE_PRIVATE_SET_FUNC(gcing, ONGOING_MIGRATION); 1369 PAGE_PRIVATE_SET_FUNC(atomic, ATOMIC_WRITE); 1370 PAGE_PRIVATE_SET_FUNC(dummy, DUMMY_WRITE); 1371 1372 PAGE_PRIVATE_CLEAR_FUNC(reference, REF_RESOURCE); 1373 PAGE_PRIVATE_CLEAR_FUNC(inline, INLINE_INODE); 1374 PAGE_PRIVATE_CLEAR_FUNC(gcing, ONGOING_MIGRATION); 1375 PAGE_PRIVATE_CLEAR_FUNC(atomic, ATOMIC_WRITE); 1376 PAGE_PRIVATE_CLEAR_FUNC(dummy, DUMMY_WRITE); 1377 1378 static inline unsigned long get_page_private_data(struct page *page) 1379 { 1380 unsigned long data = page_private(page); 1381 1382 if (!test_bit(PAGE_PRIVATE_NOT_POINTER, &data)) 1383 return 0; 1384 return data >> PAGE_PRIVATE_MAX; 1385 } 1386 1387 static inline void set_page_private_data(struct page *page, unsigned long data) 1388 { 1389 if (!PagePrivate(page)) { 1390 get_page(page); 1391 SetPagePrivate(page); 1392 } 1393 set_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)); 1394 page_private(page) |= data << PAGE_PRIVATE_MAX; 1395 } 1396 1397 static inline void clear_page_private_data(struct page *page) 1398 { 1399 page_private(page) &= (1 << PAGE_PRIVATE_MAX) - 1; 1400 if (page_private(page) == 1 << PAGE_PRIVATE_NOT_POINTER) { 1401 set_page_private(page, 0); 1402 if (PagePrivate(page)) { 1403 ClearPagePrivate(page); 1404 put_page(page); 1405 } 1406 } 1407 } 1408 1409 /* For compression */ 1410 enum compress_algorithm_type { 1411 COMPRESS_LZO, 1412 COMPRESS_LZ4, 1413 COMPRESS_ZSTD, 1414 COMPRESS_LZORLE, 1415 COMPRESS_MAX, 1416 }; 1417 1418 enum compress_flag { 1419 COMPRESS_CHKSUM, 1420 COMPRESS_MAX_FLAG, 1421 }; 1422 1423 #define COMPRESS_WATERMARK 20 1424 #define COMPRESS_PERCENT 20 1425 1426 #define COMPRESS_DATA_RESERVED_SIZE 4 1427 struct compress_data { 1428 __le32 clen; /* compressed data size */ 1429 __le32 chksum; /* compressed data chksum */ 1430 __le32 reserved[COMPRESS_DATA_RESERVED_SIZE]; /* reserved */ 1431 u8 cdata[]; /* compressed data */ 1432 }; 1433 1434 #define COMPRESS_HEADER_SIZE (sizeof(struct compress_data)) 1435 1436 #define F2FS_COMPRESSED_PAGE_MAGIC 0xF5F2C000 1437 1438 #define COMPRESS_LEVEL_OFFSET 8 1439 1440 /* compress context */ 1441 struct compress_ctx { 1442 struct inode *inode; /* inode the context belong to */ 1443 pgoff_t cluster_idx; /* cluster index number */ 1444 unsigned int cluster_size; /* page count in cluster */ 1445 unsigned int log_cluster_size; /* log of cluster size */ 1446 struct page **rpages; /* pages store raw data in cluster */ 1447 unsigned int nr_rpages; /* total page number in rpages */ 1448 struct page **cpages; /* pages store compressed data in cluster */ 1449 unsigned int nr_cpages; /* total page number in cpages */ 1450 void *rbuf; /* virtual mapped address on rpages */ 1451 struct compress_data *cbuf; /* virtual mapped address on cpages */ 1452 size_t rlen; /* valid data length in rbuf */ 1453 size_t clen; /* valid data length in cbuf */ 1454 void *private; /* payload buffer for specified compression algorithm */ 1455 void *private2; /* extra payload buffer */ 1456 }; 1457 1458 /* compress context for write IO path */ 1459 struct compress_io_ctx { 1460 u32 magic; /* magic number to indicate page is compressed */ 1461 struct inode *inode; /* inode the context belong to */ 1462 struct page **rpages; /* pages store raw data in cluster */ 1463 unsigned int nr_rpages; /* total page number in rpages */ 1464 atomic_t pending_pages; /* in-flight compressed page count */ 1465 }; 1466 1467 /* Context for decompressing one cluster on the read IO path */ 1468 struct decompress_io_ctx { 1469 u32 magic; /* magic number to indicate page is compressed */ 1470 struct inode *inode; /* inode the context belong to */ 1471 pgoff_t cluster_idx; /* cluster index number */ 1472 unsigned int cluster_size; /* page count in cluster */ 1473 unsigned int log_cluster_size; /* log of cluster size */ 1474 struct page **rpages; /* pages store raw data in cluster */ 1475 unsigned int nr_rpages; /* total page number in rpages */ 1476 struct page **cpages; /* pages store compressed data in cluster */ 1477 unsigned int nr_cpages; /* total page number in cpages */ 1478 struct page **tpages; /* temp pages to pad holes in cluster */ 1479 void *rbuf; /* virtual mapped address on rpages */ 1480 struct compress_data *cbuf; /* virtual mapped address on cpages */ 1481 size_t rlen; /* valid data length in rbuf */ 1482 size_t clen; /* valid data length in cbuf */ 1483 1484 /* 1485 * The number of compressed pages remaining to be read in this cluster. 1486 * This is initially nr_cpages. It is decremented by 1 each time a page 1487 * has been read (or failed to be read). When it reaches 0, the cluster 1488 * is decompressed (or an error is reported). 1489 * 1490 * If an error occurs before all the pages have been submitted for I/O, 1491 * then this will never reach 0. In this case the I/O submitter is 1492 * responsible for calling f2fs_decompress_end_io() instead. 1493 */ 1494 atomic_t remaining_pages; 1495 1496 /* 1497 * Number of references to this decompress_io_ctx. 1498 * 1499 * One reference is held for I/O completion. This reference is dropped 1500 * after the pagecache pages are updated and unlocked -- either after 1501 * decompression (and verity if enabled), or after an error. 1502 * 1503 * In addition, each compressed page holds a reference while it is in a 1504 * bio. These references are necessary prevent compressed pages from 1505 * being freed while they are still in a bio. 1506 */ 1507 refcount_t refcnt; 1508 1509 bool failed; /* IO error occurred before decompression? */ 1510 bool need_verity; /* need fs-verity verification after decompression? */ 1511 void *private; /* payload buffer for specified decompression algorithm */ 1512 void *private2; /* extra payload buffer */ 1513 struct work_struct verity_work; /* work to verify the decompressed pages */ 1514 }; 1515 1516 #define NULL_CLUSTER ((unsigned int)(~0)) 1517 #define MIN_COMPRESS_LOG_SIZE 2 1518 #define MAX_COMPRESS_LOG_SIZE 8 1519 #define MAX_COMPRESS_WINDOW_SIZE(log_size) ((PAGE_SIZE) << (log_size)) 1520 1521 struct f2fs_sb_info { 1522 struct super_block *sb; /* pointer to VFS super block */ 1523 struct proc_dir_entry *s_proc; /* proc entry */ 1524 struct f2fs_super_block *raw_super; /* raw super block pointer */ 1525 struct rw_semaphore sb_lock; /* lock for raw super block */ 1526 int valid_super_block; /* valid super block no */ 1527 unsigned long s_flag; /* flags for sbi */ 1528 struct mutex writepages; /* mutex for writepages() */ 1529 1530 #ifdef CONFIG_BLK_DEV_ZONED 1531 unsigned int blocks_per_blkz; /* F2FS blocks per zone */ 1532 unsigned int log_blocks_per_blkz; /* log2 F2FS blocks per zone */ 1533 #endif 1534 1535 /* for node-related operations */ 1536 struct f2fs_nm_info *nm_info; /* node manager */ 1537 struct inode *node_inode; /* cache node blocks */ 1538 1539 /* for segment-related operations */ 1540 struct f2fs_sm_info *sm_info; /* segment manager */ 1541 1542 /* for bio operations */ 1543 struct f2fs_bio_info *write_io[NR_PAGE_TYPE]; /* for write bios */ 1544 /* keep migration IO order for LFS mode */ 1545 struct rw_semaphore io_order_lock; 1546 mempool_t *write_io_dummy; /* Dummy pages */ 1547 1548 /* for checkpoint */ 1549 struct f2fs_checkpoint *ckpt; /* raw checkpoint pointer */ 1550 int cur_cp_pack; /* remain current cp pack */ 1551 spinlock_t cp_lock; /* for flag in ckpt */ 1552 struct inode *meta_inode; /* cache meta blocks */ 1553 struct rw_semaphore cp_global_sem; /* checkpoint procedure lock */ 1554 struct rw_semaphore cp_rwsem; /* blocking FS operations */ 1555 struct rw_semaphore node_write; /* locking node writes */ 1556 struct rw_semaphore node_change; /* locking node change */ 1557 wait_queue_head_t cp_wait; 1558 unsigned long last_time[MAX_TIME]; /* to store time in jiffies */ 1559 long interval_time[MAX_TIME]; /* to store thresholds */ 1560 struct ckpt_req_control cprc_info; /* for checkpoint request control */ 1561 1562 struct inode_management im[MAX_INO_ENTRY]; /* manage inode cache */ 1563 1564 spinlock_t fsync_node_lock; /* for node entry lock */ 1565 struct list_head fsync_node_list; /* node list head */ 1566 unsigned int fsync_seg_id; /* sequence id */ 1567 unsigned int fsync_node_num; /* number of node entries */ 1568 1569 /* for orphan inode, use 0'th array */ 1570 unsigned int max_orphans; /* max orphan inodes */ 1571 1572 /* for inode management */ 1573 struct list_head inode_list[NR_INODE_TYPE]; /* dirty inode list */ 1574 spinlock_t inode_lock[NR_INODE_TYPE]; /* for dirty inode list lock */ 1575 struct mutex flush_lock; /* for flush exclusion */ 1576 1577 /* for extent tree cache */ 1578 struct radix_tree_root extent_tree_root;/* cache extent cache entries */ 1579 struct mutex extent_tree_lock; /* locking extent radix tree */ 1580 struct list_head extent_list; /* lru list for shrinker */ 1581 spinlock_t extent_lock; /* locking extent lru list */ 1582 atomic_t total_ext_tree; /* extent tree count */ 1583 struct list_head zombie_list; /* extent zombie tree list */ 1584 atomic_t total_zombie_tree; /* extent zombie tree count */ 1585 atomic_t total_ext_node; /* extent info count */ 1586 1587 /* basic filesystem units */ 1588 unsigned int log_sectors_per_block; /* log2 sectors per block */ 1589 unsigned int log_blocksize; /* log2 block size */ 1590 unsigned int blocksize; /* block size */ 1591 unsigned int root_ino_num; /* root inode number*/ 1592 unsigned int node_ino_num; /* node inode number*/ 1593 unsigned int meta_ino_num; /* meta inode number*/ 1594 unsigned int log_blocks_per_seg; /* log2 blocks per segment */ 1595 unsigned int blocks_per_seg; /* blocks per segment */ 1596 unsigned int segs_per_sec; /* segments per section */ 1597 unsigned int secs_per_zone; /* sections per zone */ 1598 unsigned int total_sections; /* total section count */ 1599 unsigned int total_node_count; /* total node block count */ 1600 unsigned int total_valid_node_count; /* valid node block count */ 1601 int dir_level; /* directory level */ 1602 int readdir_ra; /* readahead inode in readdir */ 1603 u64 max_io_bytes; /* max io bytes to merge IOs */ 1604 1605 block_t user_block_count; /* # of user blocks */ 1606 block_t total_valid_block_count; /* # of valid blocks */ 1607 block_t discard_blks; /* discard command candidats */ 1608 block_t last_valid_block_count; /* for recovery */ 1609 block_t reserved_blocks; /* configurable reserved blocks */ 1610 block_t current_reserved_blocks; /* current reserved blocks */ 1611 1612 /* Additional tracking for no checkpoint mode */ 1613 block_t unusable_block_count; /* # of blocks saved by last cp */ 1614 1615 unsigned int nquota_files; /* # of quota sysfile */ 1616 struct rw_semaphore quota_sem; /* blocking cp for flags */ 1617 1618 /* # of pages, see count_type */ 1619 atomic_t nr_pages[NR_COUNT_TYPE]; 1620 /* # of allocated blocks */ 1621 struct percpu_counter alloc_valid_block_count; 1622 1623 /* writeback control */ 1624 atomic_t wb_sync_req[META]; /* count # of WB_SYNC threads */ 1625 1626 /* valid inode count */ 1627 struct percpu_counter total_valid_inode_count; 1628 1629 struct f2fs_mount_info mount_opt; /* mount options */ 1630 1631 /* for cleaning operations */ 1632 struct rw_semaphore gc_lock; /* 1633 * semaphore for GC, avoid 1634 * race between GC and GC or CP 1635 */ 1636 struct f2fs_gc_kthread *gc_thread; /* GC thread */ 1637 struct atgc_management am; /* atgc management */ 1638 unsigned int cur_victim_sec; /* current victim section num */ 1639 unsigned int gc_mode; /* current GC state */ 1640 unsigned int next_victim_seg[2]; /* next segment in victim section */ 1641 1642 /* for skip statistic */ 1643 unsigned int atomic_files; /* # of opened atomic file */ 1644 unsigned long long skipped_atomic_files[2]; /* FG_GC and BG_GC */ 1645 unsigned long long skipped_gc_rwsem; /* FG_GC only */ 1646 1647 /* threshold for gc trials on pinned files */ 1648 u64 gc_pin_file_threshold; 1649 struct rw_semaphore pin_sem; 1650 1651 /* maximum # of trials to find a victim segment for SSR and GC */ 1652 unsigned int max_victim_search; 1653 /* migration granularity of garbage collection, unit: segment */ 1654 unsigned int migration_granularity; 1655 1656 /* 1657 * for stat information. 1658 * one is for the LFS mode, and the other is for the SSR mode. 1659 */ 1660 #ifdef CONFIG_F2FS_STAT_FS 1661 struct f2fs_stat_info *stat_info; /* FS status information */ 1662 atomic_t meta_count[META_MAX]; /* # of meta blocks */ 1663 unsigned int segment_count[2]; /* # of allocated segments */ 1664 unsigned int block_count[2]; /* # of allocated blocks */ 1665 atomic_t inplace_count; /* # of inplace update */ 1666 atomic64_t total_hit_ext; /* # of lookup extent cache */ 1667 atomic64_t read_hit_rbtree; /* # of hit rbtree extent node */ 1668 atomic64_t read_hit_largest; /* # of hit largest extent node */ 1669 atomic64_t read_hit_cached; /* # of hit cached extent node */ 1670 atomic_t inline_xattr; /* # of inline_xattr inodes */ 1671 atomic_t inline_inode; /* # of inline_data inodes */ 1672 atomic_t inline_dir; /* # of inline_dentry inodes */ 1673 atomic_t compr_inode; /* # of compressed inodes */ 1674 atomic64_t compr_blocks; /* # of compressed blocks */ 1675 atomic_t vw_cnt; /* # of volatile writes */ 1676 atomic_t max_aw_cnt; /* max # of atomic writes */ 1677 atomic_t max_vw_cnt; /* max # of volatile writes */ 1678 unsigned int io_skip_bggc; /* skip background gc for in-flight IO */ 1679 unsigned int other_skip_bggc; /* skip background gc for other reasons */ 1680 unsigned int ndirty_inode[NR_INODE_TYPE]; /* # of dirty inodes */ 1681 #endif 1682 spinlock_t stat_lock; /* lock for stat operations */ 1683 1684 /* For app/fs IO statistics */ 1685 spinlock_t iostat_lock; 1686 unsigned long long rw_iostat[NR_IO_TYPE]; 1687 unsigned long long prev_rw_iostat[NR_IO_TYPE]; 1688 bool iostat_enable; 1689 unsigned long iostat_next_period; 1690 unsigned int iostat_period_ms; 1691 1692 /* to attach REQ_META|REQ_FUA flags */ 1693 unsigned int data_io_flag; 1694 unsigned int node_io_flag; 1695 1696 /* For sysfs suppport */ 1697 struct kobject s_kobj; /* /sys/fs/f2fs/<devname> */ 1698 struct completion s_kobj_unregister; 1699 1700 struct kobject s_stat_kobj; /* /sys/fs/f2fs/<devname>/stat */ 1701 struct completion s_stat_kobj_unregister; 1702 1703 struct kobject s_feature_list_kobj; /* /sys/fs/f2fs/<devname>/feature_list */ 1704 struct completion s_feature_list_kobj_unregister; 1705 1706 /* For shrinker support */ 1707 struct list_head s_list; 1708 int s_ndevs; /* number of devices */ 1709 struct f2fs_dev_info *devs; /* for device list */ 1710 unsigned int dirty_device; /* for checkpoint data flush */ 1711 spinlock_t dev_lock; /* protect dirty_device */ 1712 struct mutex umount_mutex; 1713 unsigned int shrinker_run_no; 1714 1715 /* For write statistics */ 1716 u64 sectors_written_start; 1717 u64 kbytes_written; 1718 1719 /* Reference to checksum algorithm driver via cryptoapi */ 1720 struct crypto_shash *s_chksum_driver; 1721 1722 /* Precomputed FS UUID checksum for seeding other checksums */ 1723 __u32 s_chksum_seed; 1724 1725 struct workqueue_struct *post_read_wq; /* post read workqueue */ 1726 1727 struct kmem_cache *inline_xattr_slab; /* inline xattr entry */ 1728 unsigned int inline_xattr_slab_size; /* default inline xattr slab size */ 1729 1730 #ifdef CONFIG_F2FS_FS_COMPRESSION 1731 struct kmem_cache *page_array_slab; /* page array entry */ 1732 unsigned int page_array_slab_size; /* default page array slab size */ 1733 1734 /* For runtime compression statistics */ 1735 u64 compr_written_block; 1736 u64 compr_saved_block; 1737 u32 compr_new_inode; 1738 1739 /* For compressed block cache */ 1740 struct inode *compress_inode; /* cache compressed blocks */ 1741 unsigned int compress_percent; /* cache page percentage */ 1742 unsigned int compress_watermark; /* cache page watermark */ 1743 atomic_t compress_page_hit; /* cache hit count */ 1744 #endif 1745 }; 1746 1747 struct f2fs_private_dio { 1748 struct inode *inode; 1749 void *orig_private; 1750 bio_end_io_t *orig_end_io; 1751 bool write; 1752 }; 1753 1754 #ifdef CONFIG_F2FS_FAULT_INJECTION 1755 #define f2fs_show_injection_info(sbi, type) \ 1756 printk_ratelimited("%sF2FS-fs (%s) : inject %s in %s of %pS\n", \ 1757 KERN_INFO, sbi->sb->s_id, \ 1758 f2fs_fault_name[type], \ 1759 __func__, __builtin_return_address(0)) 1760 static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type) 1761 { 1762 struct f2fs_fault_info *ffi = &F2FS_OPTION(sbi).fault_info; 1763 1764 if (!ffi->inject_rate) 1765 return false; 1766 1767 if (!IS_FAULT_SET(ffi, type)) 1768 return false; 1769 1770 atomic_inc(&ffi->inject_ops); 1771 if (atomic_read(&ffi->inject_ops) >= ffi->inject_rate) { 1772 atomic_set(&ffi->inject_ops, 0); 1773 return true; 1774 } 1775 return false; 1776 } 1777 #else 1778 #define f2fs_show_injection_info(sbi, type) do { } while (0) 1779 static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type) 1780 { 1781 return false; 1782 } 1783 #endif 1784 1785 /* 1786 * Test if the mounted volume is a multi-device volume. 1787 * - For a single regular disk volume, sbi->s_ndevs is 0. 1788 * - For a single zoned disk volume, sbi->s_ndevs is 1. 1789 * - For a multi-device volume, sbi->s_ndevs is always 2 or more. 1790 */ 1791 static inline bool f2fs_is_multi_device(struct f2fs_sb_info *sbi) 1792 { 1793 return sbi->s_ndevs > 1; 1794 } 1795 1796 static inline void f2fs_update_time(struct f2fs_sb_info *sbi, int type) 1797 { 1798 unsigned long now = jiffies; 1799 1800 sbi->last_time[type] = now; 1801 1802 /* DISCARD_TIME and GC_TIME are based on REQ_TIME */ 1803 if (type == REQ_TIME) { 1804 sbi->last_time[DISCARD_TIME] = now; 1805 sbi->last_time[GC_TIME] = now; 1806 } 1807 } 1808 1809 static inline bool f2fs_time_over(struct f2fs_sb_info *sbi, int type) 1810 { 1811 unsigned long interval = sbi->interval_time[type] * HZ; 1812 1813 return time_after(jiffies, sbi->last_time[type] + interval); 1814 } 1815 1816 static inline unsigned int f2fs_time_to_wait(struct f2fs_sb_info *sbi, 1817 int type) 1818 { 1819 unsigned long interval = sbi->interval_time[type] * HZ; 1820 unsigned int wait_ms = 0; 1821 long delta; 1822 1823 delta = (sbi->last_time[type] + interval) - jiffies; 1824 if (delta > 0) 1825 wait_ms = jiffies_to_msecs(delta); 1826 1827 return wait_ms; 1828 } 1829 1830 /* 1831 * Inline functions 1832 */ 1833 static inline u32 __f2fs_crc32(struct f2fs_sb_info *sbi, u32 crc, 1834 const void *address, unsigned int length) 1835 { 1836 struct { 1837 struct shash_desc shash; 1838 char ctx[4]; 1839 } desc; 1840 int err; 1841 1842 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver) != sizeof(desc.ctx)); 1843 1844 desc.shash.tfm = sbi->s_chksum_driver; 1845 *(u32 *)desc.ctx = crc; 1846 1847 err = crypto_shash_update(&desc.shash, address, length); 1848 BUG_ON(err); 1849 1850 return *(u32 *)desc.ctx; 1851 } 1852 1853 static inline u32 f2fs_crc32(struct f2fs_sb_info *sbi, const void *address, 1854 unsigned int length) 1855 { 1856 return __f2fs_crc32(sbi, F2FS_SUPER_MAGIC, address, length); 1857 } 1858 1859 static inline bool f2fs_crc_valid(struct f2fs_sb_info *sbi, __u32 blk_crc, 1860 void *buf, size_t buf_size) 1861 { 1862 return f2fs_crc32(sbi, buf, buf_size) == blk_crc; 1863 } 1864 1865 static inline u32 f2fs_chksum(struct f2fs_sb_info *sbi, u32 crc, 1866 const void *address, unsigned int length) 1867 { 1868 return __f2fs_crc32(sbi, crc, address, length); 1869 } 1870 1871 static inline struct f2fs_inode_info *F2FS_I(struct inode *inode) 1872 { 1873 return container_of(inode, struct f2fs_inode_info, vfs_inode); 1874 } 1875 1876 static inline struct f2fs_sb_info *F2FS_SB(struct super_block *sb) 1877 { 1878 return sb->s_fs_info; 1879 } 1880 1881 static inline struct f2fs_sb_info *F2FS_I_SB(struct inode *inode) 1882 { 1883 return F2FS_SB(inode->i_sb); 1884 } 1885 1886 static inline struct f2fs_sb_info *F2FS_M_SB(struct address_space *mapping) 1887 { 1888 return F2FS_I_SB(mapping->host); 1889 } 1890 1891 static inline struct f2fs_sb_info *F2FS_P_SB(struct page *page) 1892 { 1893 return F2FS_M_SB(page_file_mapping(page)); 1894 } 1895 1896 static inline struct f2fs_super_block *F2FS_RAW_SUPER(struct f2fs_sb_info *sbi) 1897 { 1898 return (struct f2fs_super_block *)(sbi->raw_super); 1899 } 1900 1901 static inline struct f2fs_checkpoint *F2FS_CKPT(struct f2fs_sb_info *sbi) 1902 { 1903 return (struct f2fs_checkpoint *)(sbi->ckpt); 1904 } 1905 1906 static inline struct f2fs_node *F2FS_NODE(struct page *page) 1907 { 1908 return (struct f2fs_node *)page_address(page); 1909 } 1910 1911 static inline struct f2fs_inode *F2FS_INODE(struct page *page) 1912 { 1913 return &((struct f2fs_node *)page_address(page))->i; 1914 } 1915 1916 static inline struct f2fs_nm_info *NM_I(struct f2fs_sb_info *sbi) 1917 { 1918 return (struct f2fs_nm_info *)(sbi->nm_info); 1919 } 1920 1921 static inline struct f2fs_sm_info *SM_I(struct f2fs_sb_info *sbi) 1922 { 1923 return (struct f2fs_sm_info *)(sbi->sm_info); 1924 } 1925 1926 static inline struct sit_info *SIT_I(struct f2fs_sb_info *sbi) 1927 { 1928 return (struct sit_info *)(SM_I(sbi)->sit_info); 1929 } 1930 1931 static inline struct free_segmap_info *FREE_I(struct f2fs_sb_info *sbi) 1932 { 1933 return (struct free_segmap_info *)(SM_I(sbi)->free_info); 1934 } 1935 1936 static inline struct dirty_seglist_info *DIRTY_I(struct f2fs_sb_info *sbi) 1937 { 1938 return (struct dirty_seglist_info *)(SM_I(sbi)->dirty_info); 1939 } 1940 1941 static inline struct address_space *META_MAPPING(struct f2fs_sb_info *sbi) 1942 { 1943 return sbi->meta_inode->i_mapping; 1944 } 1945 1946 static inline struct address_space *NODE_MAPPING(struct f2fs_sb_info *sbi) 1947 { 1948 return sbi->node_inode->i_mapping; 1949 } 1950 1951 static inline bool is_sbi_flag_set(struct f2fs_sb_info *sbi, unsigned int type) 1952 { 1953 return test_bit(type, &sbi->s_flag); 1954 } 1955 1956 static inline void set_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type) 1957 { 1958 set_bit(type, &sbi->s_flag); 1959 } 1960 1961 static inline void clear_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type) 1962 { 1963 clear_bit(type, &sbi->s_flag); 1964 } 1965 1966 static inline unsigned long long cur_cp_version(struct f2fs_checkpoint *cp) 1967 { 1968 return le64_to_cpu(cp->checkpoint_ver); 1969 } 1970 1971 static inline unsigned long f2fs_qf_ino(struct super_block *sb, int type) 1972 { 1973 if (type < F2FS_MAX_QUOTAS) 1974 return le32_to_cpu(F2FS_SB(sb)->raw_super->qf_ino[type]); 1975 return 0; 1976 } 1977 1978 static inline __u64 cur_cp_crc(struct f2fs_checkpoint *cp) 1979 { 1980 size_t crc_offset = le32_to_cpu(cp->checksum_offset); 1981 return le32_to_cpu(*((__le32 *)((unsigned char *)cp + crc_offset))); 1982 } 1983 1984 static inline bool __is_set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 1985 { 1986 unsigned int ckpt_flags = le32_to_cpu(cp->ckpt_flags); 1987 1988 return ckpt_flags & f; 1989 } 1990 1991 static inline bool is_set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 1992 { 1993 return __is_set_ckpt_flags(F2FS_CKPT(sbi), f); 1994 } 1995 1996 static inline void __set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 1997 { 1998 unsigned int ckpt_flags; 1999 2000 ckpt_flags = le32_to_cpu(cp->ckpt_flags); 2001 ckpt_flags |= f; 2002 cp->ckpt_flags = cpu_to_le32(ckpt_flags); 2003 } 2004 2005 static inline void set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 2006 { 2007 unsigned long flags; 2008 2009 spin_lock_irqsave(&sbi->cp_lock, flags); 2010 __set_ckpt_flags(F2FS_CKPT(sbi), f); 2011 spin_unlock_irqrestore(&sbi->cp_lock, flags); 2012 } 2013 2014 static inline void __clear_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 2015 { 2016 unsigned int ckpt_flags; 2017 2018 ckpt_flags = le32_to_cpu(cp->ckpt_flags); 2019 ckpt_flags &= (~f); 2020 cp->ckpt_flags = cpu_to_le32(ckpt_flags); 2021 } 2022 2023 static inline void clear_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 2024 { 2025 unsigned long flags; 2026 2027 spin_lock_irqsave(&sbi->cp_lock, flags); 2028 __clear_ckpt_flags(F2FS_CKPT(sbi), f); 2029 spin_unlock_irqrestore(&sbi->cp_lock, flags); 2030 } 2031 2032 static inline void disable_nat_bits(struct f2fs_sb_info *sbi, bool lock) 2033 { 2034 unsigned long flags; 2035 unsigned char *nat_bits; 2036 2037 /* 2038 * In order to re-enable nat_bits we need to call fsck.f2fs by 2039 * set_sbi_flag(sbi, SBI_NEED_FSCK). But it may give huge cost, 2040 * so let's rely on regular fsck or unclean shutdown. 2041 */ 2042 2043 if (lock) 2044 spin_lock_irqsave(&sbi->cp_lock, flags); 2045 __clear_ckpt_flags(F2FS_CKPT(sbi), CP_NAT_BITS_FLAG); 2046 nat_bits = NM_I(sbi)->nat_bits; 2047 NM_I(sbi)->nat_bits = NULL; 2048 if (lock) 2049 spin_unlock_irqrestore(&sbi->cp_lock, flags); 2050 2051 kvfree(nat_bits); 2052 } 2053 2054 static inline bool enabled_nat_bits(struct f2fs_sb_info *sbi, 2055 struct cp_control *cpc) 2056 { 2057 bool set = is_set_ckpt_flags(sbi, CP_NAT_BITS_FLAG); 2058 2059 return (cpc) ? (cpc->reason & CP_UMOUNT) && set : set; 2060 } 2061 2062 static inline void f2fs_lock_op(struct f2fs_sb_info *sbi) 2063 { 2064 down_read(&sbi->cp_rwsem); 2065 } 2066 2067 static inline int f2fs_trylock_op(struct f2fs_sb_info *sbi) 2068 { 2069 return down_read_trylock(&sbi->cp_rwsem); 2070 } 2071 2072 static inline void f2fs_unlock_op(struct f2fs_sb_info *sbi) 2073 { 2074 up_read(&sbi->cp_rwsem); 2075 } 2076 2077 static inline void f2fs_lock_all(struct f2fs_sb_info *sbi) 2078 { 2079 down_write(&sbi->cp_rwsem); 2080 } 2081 2082 static inline void f2fs_unlock_all(struct f2fs_sb_info *sbi) 2083 { 2084 up_write(&sbi->cp_rwsem); 2085 } 2086 2087 static inline int __get_cp_reason(struct f2fs_sb_info *sbi) 2088 { 2089 int reason = CP_SYNC; 2090 2091 if (test_opt(sbi, FASTBOOT)) 2092 reason = CP_FASTBOOT; 2093 if (is_sbi_flag_set(sbi, SBI_IS_CLOSE)) 2094 reason = CP_UMOUNT; 2095 return reason; 2096 } 2097 2098 static inline bool __remain_node_summaries(int reason) 2099 { 2100 return (reason & (CP_UMOUNT | CP_FASTBOOT)); 2101 } 2102 2103 static inline bool __exist_node_summaries(struct f2fs_sb_info *sbi) 2104 { 2105 return (is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG) || 2106 is_set_ckpt_flags(sbi, CP_FASTBOOT_FLAG)); 2107 } 2108 2109 /* 2110 * Check whether the inode has blocks or not 2111 */ 2112 static inline int F2FS_HAS_BLOCKS(struct inode *inode) 2113 { 2114 block_t xattr_block = F2FS_I(inode)->i_xattr_nid ? 1 : 0; 2115 2116 return (inode->i_blocks >> F2FS_LOG_SECTORS_PER_BLOCK) > xattr_block; 2117 } 2118 2119 static inline bool f2fs_has_xattr_block(unsigned int ofs) 2120 { 2121 return ofs == XATTR_NODE_OFFSET; 2122 } 2123 2124 static inline bool __allow_reserved_blocks(struct f2fs_sb_info *sbi, 2125 struct inode *inode, bool cap) 2126 { 2127 if (!inode) 2128 return true; 2129 if (!test_opt(sbi, RESERVE_ROOT)) 2130 return false; 2131 if (IS_NOQUOTA(inode)) 2132 return true; 2133 if (uid_eq(F2FS_OPTION(sbi).s_resuid, current_fsuid())) 2134 return true; 2135 if (!gid_eq(F2FS_OPTION(sbi).s_resgid, GLOBAL_ROOT_GID) && 2136 in_group_p(F2FS_OPTION(sbi).s_resgid)) 2137 return true; 2138 if (cap && capable(CAP_SYS_RESOURCE)) 2139 return true; 2140 return false; 2141 } 2142 2143 static inline void f2fs_i_blocks_write(struct inode *, block_t, bool, bool); 2144 static inline int inc_valid_block_count(struct f2fs_sb_info *sbi, 2145 struct inode *inode, blkcnt_t *count) 2146 { 2147 blkcnt_t diff = 0, release = 0; 2148 block_t avail_user_block_count; 2149 int ret; 2150 2151 ret = dquot_reserve_block(inode, *count); 2152 if (ret) 2153 return ret; 2154 2155 if (time_to_inject(sbi, FAULT_BLOCK)) { 2156 f2fs_show_injection_info(sbi, FAULT_BLOCK); 2157 release = *count; 2158 goto release_quota; 2159 } 2160 2161 /* 2162 * let's increase this in prior to actual block count change in order 2163 * for f2fs_sync_file to avoid data races when deciding checkpoint. 2164 */ 2165 percpu_counter_add(&sbi->alloc_valid_block_count, (*count)); 2166 2167 spin_lock(&sbi->stat_lock); 2168 sbi->total_valid_block_count += (block_t)(*count); 2169 avail_user_block_count = sbi->user_block_count - 2170 sbi->current_reserved_blocks; 2171 2172 if (!__allow_reserved_blocks(sbi, inode, true)) 2173 avail_user_block_count -= F2FS_OPTION(sbi).root_reserved_blocks; 2174 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) { 2175 if (avail_user_block_count > sbi->unusable_block_count) 2176 avail_user_block_count -= sbi->unusable_block_count; 2177 else 2178 avail_user_block_count = 0; 2179 } 2180 if (unlikely(sbi->total_valid_block_count > avail_user_block_count)) { 2181 diff = sbi->total_valid_block_count - avail_user_block_count; 2182 if (diff > *count) 2183 diff = *count; 2184 *count -= diff; 2185 release = diff; 2186 sbi->total_valid_block_count -= diff; 2187 if (!*count) { 2188 spin_unlock(&sbi->stat_lock); 2189 goto enospc; 2190 } 2191 } 2192 spin_unlock(&sbi->stat_lock); 2193 2194 if (unlikely(release)) { 2195 percpu_counter_sub(&sbi->alloc_valid_block_count, release); 2196 dquot_release_reservation_block(inode, release); 2197 } 2198 f2fs_i_blocks_write(inode, *count, true, true); 2199 return 0; 2200 2201 enospc: 2202 percpu_counter_sub(&sbi->alloc_valid_block_count, release); 2203 release_quota: 2204 dquot_release_reservation_block(inode, release); 2205 return -ENOSPC; 2206 } 2207 2208 __printf(2, 3) 2209 void f2fs_printk(struct f2fs_sb_info *sbi, const char *fmt, ...); 2210 2211 #define f2fs_err(sbi, fmt, ...) \ 2212 f2fs_printk(sbi, KERN_ERR fmt, ##__VA_ARGS__) 2213 #define f2fs_warn(sbi, fmt, ...) \ 2214 f2fs_printk(sbi, KERN_WARNING fmt, ##__VA_ARGS__) 2215 #define f2fs_notice(sbi, fmt, ...) \ 2216 f2fs_printk(sbi, KERN_NOTICE fmt, ##__VA_ARGS__) 2217 #define f2fs_info(sbi, fmt, ...) \ 2218 f2fs_printk(sbi, KERN_INFO fmt, ##__VA_ARGS__) 2219 #define f2fs_debug(sbi, fmt, ...) \ 2220 f2fs_printk(sbi, KERN_DEBUG fmt, ##__VA_ARGS__) 2221 2222 static inline void dec_valid_block_count(struct f2fs_sb_info *sbi, 2223 struct inode *inode, 2224 block_t count) 2225 { 2226 blkcnt_t sectors = count << F2FS_LOG_SECTORS_PER_BLOCK; 2227 2228 spin_lock(&sbi->stat_lock); 2229 f2fs_bug_on(sbi, sbi->total_valid_block_count < (block_t) count); 2230 sbi->total_valid_block_count -= (block_t)count; 2231 if (sbi->reserved_blocks && 2232 sbi->current_reserved_blocks < sbi->reserved_blocks) 2233 sbi->current_reserved_blocks = min(sbi->reserved_blocks, 2234 sbi->current_reserved_blocks + count); 2235 spin_unlock(&sbi->stat_lock); 2236 if (unlikely(inode->i_blocks < sectors)) { 2237 f2fs_warn(sbi, "Inconsistent i_blocks, ino:%lu, iblocks:%llu, sectors:%llu", 2238 inode->i_ino, 2239 (unsigned long long)inode->i_blocks, 2240 (unsigned long long)sectors); 2241 set_sbi_flag(sbi, SBI_NEED_FSCK); 2242 return; 2243 } 2244 f2fs_i_blocks_write(inode, count, false, true); 2245 } 2246 2247 static inline void inc_page_count(struct f2fs_sb_info *sbi, int count_type) 2248 { 2249 atomic_inc(&sbi->nr_pages[count_type]); 2250 2251 if (count_type == F2FS_DIRTY_DENTS || 2252 count_type == F2FS_DIRTY_NODES || 2253 count_type == F2FS_DIRTY_META || 2254 count_type == F2FS_DIRTY_QDATA || 2255 count_type == F2FS_DIRTY_IMETA) 2256 set_sbi_flag(sbi, SBI_IS_DIRTY); 2257 } 2258 2259 static inline void inode_inc_dirty_pages(struct inode *inode) 2260 { 2261 atomic_inc(&F2FS_I(inode)->dirty_pages); 2262 inc_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ? 2263 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA); 2264 if (IS_NOQUOTA(inode)) 2265 inc_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA); 2266 } 2267 2268 static inline void dec_page_count(struct f2fs_sb_info *sbi, int count_type) 2269 { 2270 atomic_dec(&sbi->nr_pages[count_type]); 2271 } 2272 2273 static inline void inode_dec_dirty_pages(struct inode *inode) 2274 { 2275 if (!S_ISDIR(inode->i_mode) && !S_ISREG(inode->i_mode) && 2276 !S_ISLNK(inode->i_mode)) 2277 return; 2278 2279 atomic_dec(&F2FS_I(inode)->dirty_pages); 2280 dec_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ? 2281 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA); 2282 if (IS_NOQUOTA(inode)) 2283 dec_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA); 2284 } 2285 2286 static inline s64 get_pages(struct f2fs_sb_info *sbi, int count_type) 2287 { 2288 return atomic_read(&sbi->nr_pages[count_type]); 2289 } 2290 2291 static inline int get_dirty_pages(struct inode *inode) 2292 { 2293 return atomic_read(&F2FS_I(inode)->dirty_pages); 2294 } 2295 2296 static inline int get_blocktype_secs(struct f2fs_sb_info *sbi, int block_type) 2297 { 2298 unsigned int pages_per_sec = sbi->segs_per_sec * sbi->blocks_per_seg; 2299 unsigned int segs = (get_pages(sbi, block_type) + pages_per_sec - 1) >> 2300 sbi->log_blocks_per_seg; 2301 2302 return segs / sbi->segs_per_sec; 2303 } 2304 2305 static inline block_t valid_user_blocks(struct f2fs_sb_info *sbi) 2306 { 2307 return sbi->total_valid_block_count; 2308 } 2309 2310 static inline block_t discard_blocks(struct f2fs_sb_info *sbi) 2311 { 2312 return sbi->discard_blks; 2313 } 2314 2315 static inline unsigned long __bitmap_size(struct f2fs_sb_info *sbi, int flag) 2316 { 2317 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi); 2318 2319 /* return NAT or SIT bitmap */ 2320 if (flag == NAT_BITMAP) 2321 return le32_to_cpu(ckpt->nat_ver_bitmap_bytesize); 2322 else if (flag == SIT_BITMAP) 2323 return le32_to_cpu(ckpt->sit_ver_bitmap_bytesize); 2324 2325 return 0; 2326 } 2327 2328 static inline block_t __cp_payload(struct f2fs_sb_info *sbi) 2329 { 2330 return le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_payload); 2331 } 2332 2333 static inline void *__bitmap_ptr(struct f2fs_sb_info *sbi, int flag) 2334 { 2335 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi); 2336 void *tmp_ptr = &ckpt->sit_nat_version_bitmap; 2337 int offset; 2338 2339 if (is_set_ckpt_flags(sbi, CP_LARGE_NAT_BITMAP_FLAG)) { 2340 offset = (flag == SIT_BITMAP) ? 2341 le32_to_cpu(ckpt->nat_ver_bitmap_bytesize) : 0; 2342 /* 2343 * if large_nat_bitmap feature is enabled, leave checksum 2344 * protection for all nat/sit bitmaps. 2345 */ 2346 return tmp_ptr + offset + sizeof(__le32); 2347 } 2348 2349 if (__cp_payload(sbi) > 0) { 2350 if (flag == NAT_BITMAP) 2351 return &ckpt->sit_nat_version_bitmap; 2352 else 2353 return (unsigned char *)ckpt + F2FS_BLKSIZE; 2354 } else { 2355 offset = (flag == NAT_BITMAP) ? 2356 le32_to_cpu(ckpt->sit_ver_bitmap_bytesize) : 0; 2357 return tmp_ptr + offset; 2358 } 2359 } 2360 2361 static inline block_t __start_cp_addr(struct f2fs_sb_info *sbi) 2362 { 2363 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr); 2364 2365 if (sbi->cur_cp_pack == 2) 2366 start_addr += sbi->blocks_per_seg; 2367 return start_addr; 2368 } 2369 2370 static inline block_t __start_cp_next_addr(struct f2fs_sb_info *sbi) 2371 { 2372 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr); 2373 2374 if (sbi->cur_cp_pack == 1) 2375 start_addr += sbi->blocks_per_seg; 2376 return start_addr; 2377 } 2378 2379 static inline void __set_cp_next_pack(struct f2fs_sb_info *sbi) 2380 { 2381 sbi->cur_cp_pack = (sbi->cur_cp_pack == 1) ? 2 : 1; 2382 } 2383 2384 static inline block_t __start_sum_addr(struct f2fs_sb_info *sbi) 2385 { 2386 return le32_to_cpu(F2FS_CKPT(sbi)->cp_pack_start_sum); 2387 } 2388 2389 static inline int inc_valid_node_count(struct f2fs_sb_info *sbi, 2390 struct inode *inode, bool is_inode) 2391 { 2392 block_t valid_block_count; 2393 unsigned int valid_node_count, user_block_count; 2394 int err; 2395 2396 if (is_inode) { 2397 if (inode) { 2398 err = dquot_alloc_inode(inode); 2399 if (err) 2400 return err; 2401 } 2402 } else { 2403 err = dquot_reserve_block(inode, 1); 2404 if (err) 2405 return err; 2406 } 2407 2408 if (time_to_inject(sbi, FAULT_BLOCK)) { 2409 f2fs_show_injection_info(sbi, FAULT_BLOCK); 2410 goto enospc; 2411 } 2412 2413 spin_lock(&sbi->stat_lock); 2414 2415 valid_block_count = sbi->total_valid_block_count + 2416 sbi->current_reserved_blocks + 1; 2417 2418 if (!__allow_reserved_blocks(sbi, inode, false)) 2419 valid_block_count += F2FS_OPTION(sbi).root_reserved_blocks; 2420 user_block_count = sbi->user_block_count; 2421 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) 2422 user_block_count -= sbi->unusable_block_count; 2423 2424 if (unlikely(valid_block_count > user_block_count)) { 2425 spin_unlock(&sbi->stat_lock); 2426 goto enospc; 2427 } 2428 2429 valid_node_count = sbi->total_valid_node_count + 1; 2430 if (unlikely(valid_node_count > sbi->total_node_count)) { 2431 spin_unlock(&sbi->stat_lock); 2432 goto enospc; 2433 } 2434 2435 sbi->total_valid_node_count++; 2436 sbi->total_valid_block_count++; 2437 spin_unlock(&sbi->stat_lock); 2438 2439 if (inode) { 2440 if (is_inode) 2441 f2fs_mark_inode_dirty_sync(inode, true); 2442 else 2443 f2fs_i_blocks_write(inode, 1, true, true); 2444 } 2445 2446 percpu_counter_inc(&sbi->alloc_valid_block_count); 2447 return 0; 2448 2449 enospc: 2450 if (is_inode) { 2451 if (inode) 2452 dquot_free_inode(inode); 2453 } else { 2454 dquot_release_reservation_block(inode, 1); 2455 } 2456 return -ENOSPC; 2457 } 2458 2459 static inline void dec_valid_node_count(struct f2fs_sb_info *sbi, 2460 struct inode *inode, bool is_inode) 2461 { 2462 spin_lock(&sbi->stat_lock); 2463 2464 f2fs_bug_on(sbi, !sbi->total_valid_block_count); 2465 f2fs_bug_on(sbi, !sbi->total_valid_node_count); 2466 2467 sbi->total_valid_node_count--; 2468 sbi->total_valid_block_count--; 2469 if (sbi->reserved_blocks && 2470 sbi->current_reserved_blocks < sbi->reserved_blocks) 2471 sbi->current_reserved_blocks++; 2472 2473 spin_unlock(&sbi->stat_lock); 2474 2475 if (is_inode) { 2476 dquot_free_inode(inode); 2477 } else { 2478 if (unlikely(inode->i_blocks == 0)) { 2479 f2fs_warn(sbi, "dec_valid_node_count: inconsistent i_blocks, ino:%lu, iblocks:%llu", 2480 inode->i_ino, 2481 (unsigned long long)inode->i_blocks); 2482 set_sbi_flag(sbi, SBI_NEED_FSCK); 2483 return; 2484 } 2485 f2fs_i_blocks_write(inode, 1, false, true); 2486 } 2487 } 2488 2489 static inline unsigned int valid_node_count(struct f2fs_sb_info *sbi) 2490 { 2491 return sbi->total_valid_node_count; 2492 } 2493 2494 static inline void inc_valid_inode_count(struct f2fs_sb_info *sbi) 2495 { 2496 percpu_counter_inc(&sbi->total_valid_inode_count); 2497 } 2498 2499 static inline void dec_valid_inode_count(struct f2fs_sb_info *sbi) 2500 { 2501 percpu_counter_dec(&sbi->total_valid_inode_count); 2502 } 2503 2504 static inline s64 valid_inode_count(struct f2fs_sb_info *sbi) 2505 { 2506 return percpu_counter_sum_positive(&sbi->total_valid_inode_count); 2507 } 2508 2509 static inline struct page *f2fs_grab_cache_page(struct address_space *mapping, 2510 pgoff_t index, bool for_write) 2511 { 2512 struct page *page; 2513 2514 if (IS_ENABLED(CONFIG_F2FS_FAULT_INJECTION)) { 2515 if (!for_write) 2516 page = find_get_page_flags(mapping, index, 2517 FGP_LOCK | FGP_ACCESSED); 2518 else 2519 page = find_lock_page(mapping, index); 2520 if (page) 2521 return page; 2522 2523 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_ALLOC)) { 2524 f2fs_show_injection_info(F2FS_M_SB(mapping), 2525 FAULT_PAGE_ALLOC); 2526 return NULL; 2527 } 2528 } 2529 2530 if (!for_write) 2531 return grab_cache_page(mapping, index); 2532 return grab_cache_page_write_begin(mapping, index, AOP_FLAG_NOFS); 2533 } 2534 2535 static inline struct page *f2fs_pagecache_get_page( 2536 struct address_space *mapping, pgoff_t index, 2537 int fgp_flags, gfp_t gfp_mask) 2538 { 2539 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_GET)) { 2540 f2fs_show_injection_info(F2FS_M_SB(mapping), FAULT_PAGE_GET); 2541 return NULL; 2542 } 2543 2544 return pagecache_get_page(mapping, index, fgp_flags, gfp_mask); 2545 } 2546 2547 static inline void f2fs_copy_page(struct page *src, struct page *dst) 2548 { 2549 char *src_kaddr = kmap(src); 2550 char *dst_kaddr = kmap(dst); 2551 2552 memcpy(dst_kaddr, src_kaddr, PAGE_SIZE); 2553 kunmap(dst); 2554 kunmap(src); 2555 } 2556 2557 static inline void f2fs_put_page(struct page *page, int unlock) 2558 { 2559 if (!page) 2560 return; 2561 2562 if (unlock) { 2563 f2fs_bug_on(F2FS_P_SB(page), !PageLocked(page)); 2564 unlock_page(page); 2565 } 2566 put_page(page); 2567 } 2568 2569 static inline void f2fs_put_dnode(struct dnode_of_data *dn) 2570 { 2571 if (dn->node_page) 2572 f2fs_put_page(dn->node_page, 1); 2573 if (dn->inode_page && dn->node_page != dn->inode_page) 2574 f2fs_put_page(dn->inode_page, 0); 2575 dn->node_page = NULL; 2576 dn->inode_page = NULL; 2577 } 2578 2579 static inline struct kmem_cache *f2fs_kmem_cache_create(const char *name, 2580 size_t size) 2581 { 2582 return kmem_cache_create(name, size, 0, SLAB_RECLAIM_ACCOUNT, NULL); 2583 } 2584 2585 static inline void *f2fs_kmem_cache_alloc(struct kmem_cache *cachep, 2586 gfp_t flags) 2587 { 2588 void *entry; 2589 2590 entry = kmem_cache_alloc(cachep, flags); 2591 if (!entry) 2592 entry = kmem_cache_alloc(cachep, flags | __GFP_NOFAIL); 2593 return entry; 2594 } 2595 2596 static inline bool is_inflight_io(struct f2fs_sb_info *sbi, int type) 2597 { 2598 if (get_pages(sbi, F2FS_RD_DATA) || get_pages(sbi, F2FS_RD_NODE) || 2599 get_pages(sbi, F2FS_RD_META) || get_pages(sbi, F2FS_WB_DATA) || 2600 get_pages(sbi, F2FS_WB_CP_DATA) || 2601 get_pages(sbi, F2FS_DIO_READ) || 2602 get_pages(sbi, F2FS_DIO_WRITE)) 2603 return true; 2604 2605 if (type != DISCARD_TIME && SM_I(sbi) && SM_I(sbi)->dcc_info && 2606 atomic_read(&SM_I(sbi)->dcc_info->queued_discard)) 2607 return true; 2608 2609 if (SM_I(sbi) && SM_I(sbi)->fcc_info && 2610 atomic_read(&SM_I(sbi)->fcc_info->queued_flush)) 2611 return true; 2612 return false; 2613 } 2614 2615 static inline bool is_idle(struct f2fs_sb_info *sbi, int type) 2616 { 2617 if (sbi->gc_mode == GC_URGENT_HIGH) 2618 return true; 2619 2620 if (is_inflight_io(sbi, type)) 2621 return false; 2622 2623 if (sbi->gc_mode == GC_URGENT_LOW && 2624 (type == DISCARD_TIME || type == GC_TIME)) 2625 return true; 2626 2627 return f2fs_time_over(sbi, type); 2628 } 2629 2630 static inline void f2fs_radix_tree_insert(struct radix_tree_root *root, 2631 unsigned long index, void *item) 2632 { 2633 while (radix_tree_insert(root, index, item)) 2634 cond_resched(); 2635 } 2636 2637 #define RAW_IS_INODE(p) ((p)->footer.nid == (p)->footer.ino) 2638 2639 static inline bool IS_INODE(struct page *page) 2640 { 2641 struct f2fs_node *p = F2FS_NODE(page); 2642 2643 return RAW_IS_INODE(p); 2644 } 2645 2646 static inline int offset_in_addr(struct f2fs_inode *i) 2647 { 2648 return (i->i_inline & F2FS_EXTRA_ATTR) ? 2649 (le16_to_cpu(i->i_extra_isize) / sizeof(__le32)) : 0; 2650 } 2651 2652 static inline __le32 *blkaddr_in_node(struct f2fs_node *node) 2653 { 2654 return RAW_IS_INODE(node) ? node->i.i_addr : node->dn.addr; 2655 } 2656 2657 static inline int f2fs_has_extra_attr(struct inode *inode); 2658 static inline block_t data_blkaddr(struct inode *inode, 2659 struct page *node_page, unsigned int offset) 2660 { 2661 struct f2fs_node *raw_node; 2662 __le32 *addr_array; 2663 int base = 0; 2664 bool is_inode = IS_INODE(node_page); 2665 2666 raw_node = F2FS_NODE(node_page); 2667 2668 if (is_inode) { 2669 if (!inode) 2670 /* from GC path only */ 2671 base = offset_in_addr(&raw_node->i); 2672 else if (f2fs_has_extra_attr(inode)) 2673 base = get_extra_isize(inode); 2674 } 2675 2676 addr_array = blkaddr_in_node(raw_node); 2677 return le32_to_cpu(addr_array[base + offset]); 2678 } 2679 2680 static inline block_t f2fs_data_blkaddr(struct dnode_of_data *dn) 2681 { 2682 return data_blkaddr(dn->inode, dn->node_page, dn->ofs_in_node); 2683 } 2684 2685 static inline int f2fs_test_bit(unsigned int nr, char *addr) 2686 { 2687 int mask; 2688 2689 addr += (nr >> 3); 2690 mask = 1 << (7 - (nr & 0x07)); 2691 return mask & *addr; 2692 } 2693 2694 static inline void f2fs_set_bit(unsigned int nr, char *addr) 2695 { 2696 int mask; 2697 2698 addr += (nr >> 3); 2699 mask = 1 << (7 - (nr & 0x07)); 2700 *addr |= mask; 2701 } 2702 2703 static inline void f2fs_clear_bit(unsigned int nr, char *addr) 2704 { 2705 int mask; 2706 2707 addr += (nr >> 3); 2708 mask = 1 << (7 - (nr & 0x07)); 2709 *addr &= ~mask; 2710 } 2711 2712 static inline int f2fs_test_and_set_bit(unsigned int nr, char *addr) 2713 { 2714 int mask; 2715 int ret; 2716 2717 addr += (nr >> 3); 2718 mask = 1 << (7 - (nr & 0x07)); 2719 ret = mask & *addr; 2720 *addr |= mask; 2721 return ret; 2722 } 2723 2724 static inline int f2fs_test_and_clear_bit(unsigned int nr, char *addr) 2725 { 2726 int mask; 2727 int ret; 2728 2729 addr += (nr >> 3); 2730 mask = 1 << (7 - (nr & 0x07)); 2731 ret = mask & *addr; 2732 *addr &= ~mask; 2733 return ret; 2734 } 2735 2736 static inline void f2fs_change_bit(unsigned int nr, char *addr) 2737 { 2738 int mask; 2739 2740 addr += (nr >> 3); 2741 mask = 1 << (7 - (nr & 0x07)); 2742 *addr ^= mask; 2743 } 2744 2745 /* 2746 * On-disk inode flags (f2fs_inode::i_flags) 2747 */ 2748 #define F2FS_COMPR_FL 0x00000004 /* Compress file */ 2749 #define F2FS_SYNC_FL 0x00000008 /* Synchronous updates */ 2750 #define F2FS_IMMUTABLE_FL 0x00000010 /* Immutable file */ 2751 #define F2FS_APPEND_FL 0x00000020 /* writes to file may only append */ 2752 #define F2FS_NODUMP_FL 0x00000040 /* do not dump file */ 2753 #define F2FS_NOATIME_FL 0x00000080 /* do not update atime */ 2754 #define F2FS_NOCOMP_FL 0x00000400 /* Don't compress */ 2755 #define F2FS_INDEX_FL 0x00001000 /* hash-indexed directory */ 2756 #define F2FS_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */ 2757 #define F2FS_PROJINHERIT_FL 0x20000000 /* Create with parents projid */ 2758 #define F2FS_CASEFOLD_FL 0x40000000 /* Casefolded file */ 2759 2760 /* Flags that should be inherited by new inodes from their parent. */ 2761 #define F2FS_FL_INHERITED (F2FS_SYNC_FL | F2FS_NODUMP_FL | F2FS_NOATIME_FL | \ 2762 F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \ 2763 F2FS_CASEFOLD_FL | F2FS_COMPR_FL | F2FS_NOCOMP_FL) 2764 2765 /* Flags that are appropriate for regular files (all but dir-specific ones). */ 2766 #define F2FS_REG_FLMASK (~(F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \ 2767 F2FS_CASEFOLD_FL)) 2768 2769 /* Flags that are appropriate for non-directories/regular files. */ 2770 #define F2FS_OTHER_FLMASK (F2FS_NODUMP_FL | F2FS_NOATIME_FL) 2771 2772 static inline __u32 f2fs_mask_flags(umode_t mode, __u32 flags) 2773 { 2774 if (S_ISDIR(mode)) 2775 return flags; 2776 else if (S_ISREG(mode)) 2777 return flags & F2FS_REG_FLMASK; 2778 else 2779 return flags & F2FS_OTHER_FLMASK; 2780 } 2781 2782 static inline void __mark_inode_dirty_flag(struct inode *inode, 2783 int flag, bool set) 2784 { 2785 switch (flag) { 2786 case FI_INLINE_XATTR: 2787 case FI_INLINE_DATA: 2788 case FI_INLINE_DENTRY: 2789 case FI_NEW_INODE: 2790 if (set) 2791 return; 2792 fallthrough; 2793 case FI_DATA_EXIST: 2794 case FI_INLINE_DOTS: 2795 case FI_PIN_FILE: 2796 case FI_COMPRESS_RELEASED: 2797 f2fs_mark_inode_dirty_sync(inode, true); 2798 } 2799 } 2800 2801 static inline void set_inode_flag(struct inode *inode, int flag) 2802 { 2803 set_bit(flag, F2FS_I(inode)->flags); 2804 __mark_inode_dirty_flag(inode, flag, true); 2805 } 2806 2807 static inline int is_inode_flag_set(struct inode *inode, int flag) 2808 { 2809 return test_bit(flag, F2FS_I(inode)->flags); 2810 } 2811 2812 static inline void clear_inode_flag(struct inode *inode, int flag) 2813 { 2814 clear_bit(flag, F2FS_I(inode)->flags); 2815 __mark_inode_dirty_flag(inode, flag, false); 2816 } 2817 2818 static inline bool f2fs_verity_in_progress(struct inode *inode) 2819 { 2820 return IS_ENABLED(CONFIG_FS_VERITY) && 2821 is_inode_flag_set(inode, FI_VERITY_IN_PROGRESS); 2822 } 2823 2824 static inline void set_acl_inode(struct inode *inode, umode_t mode) 2825 { 2826 F2FS_I(inode)->i_acl_mode = mode; 2827 set_inode_flag(inode, FI_ACL_MODE); 2828 f2fs_mark_inode_dirty_sync(inode, false); 2829 } 2830 2831 static inline void f2fs_i_links_write(struct inode *inode, bool inc) 2832 { 2833 if (inc) 2834 inc_nlink(inode); 2835 else 2836 drop_nlink(inode); 2837 f2fs_mark_inode_dirty_sync(inode, true); 2838 } 2839 2840 static inline void f2fs_i_blocks_write(struct inode *inode, 2841 block_t diff, bool add, bool claim) 2842 { 2843 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE); 2844 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER); 2845 2846 /* add = 1, claim = 1 should be dquot_reserve_block in pair */ 2847 if (add) { 2848 if (claim) 2849 dquot_claim_block(inode, diff); 2850 else 2851 dquot_alloc_block_nofail(inode, diff); 2852 } else { 2853 dquot_free_block(inode, diff); 2854 } 2855 2856 f2fs_mark_inode_dirty_sync(inode, true); 2857 if (clean || recover) 2858 set_inode_flag(inode, FI_AUTO_RECOVER); 2859 } 2860 2861 static inline void f2fs_i_size_write(struct inode *inode, loff_t i_size) 2862 { 2863 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE); 2864 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER); 2865 2866 if (i_size_read(inode) == i_size) 2867 return; 2868 2869 i_size_write(inode, i_size); 2870 f2fs_mark_inode_dirty_sync(inode, true); 2871 if (clean || recover) 2872 set_inode_flag(inode, FI_AUTO_RECOVER); 2873 } 2874 2875 static inline void f2fs_i_depth_write(struct inode *inode, unsigned int depth) 2876 { 2877 F2FS_I(inode)->i_current_depth = depth; 2878 f2fs_mark_inode_dirty_sync(inode, true); 2879 } 2880 2881 static inline void f2fs_i_gc_failures_write(struct inode *inode, 2882 unsigned int count) 2883 { 2884 F2FS_I(inode)->i_gc_failures[GC_FAILURE_PIN] = count; 2885 f2fs_mark_inode_dirty_sync(inode, true); 2886 } 2887 2888 static inline void f2fs_i_xnid_write(struct inode *inode, nid_t xnid) 2889 { 2890 F2FS_I(inode)->i_xattr_nid = xnid; 2891 f2fs_mark_inode_dirty_sync(inode, true); 2892 } 2893 2894 static inline void f2fs_i_pino_write(struct inode *inode, nid_t pino) 2895 { 2896 F2FS_I(inode)->i_pino = pino; 2897 f2fs_mark_inode_dirty_sync(inode, true); 2898 } 2899 2900 static inline void get_inline_info(struct inode *inode, struct f2fs_inode *ri) 2901 { 2902 struct f2fs_inode_info *fi = F2FS_I(inode); 2903 2904 if (ri->i_inline & F2FS_INLINE_XATTR) 2905 set_bit(FI_INLINE_XATTR, fi->flags); 2906 if (ri->i_inline & F2FS_INLINE_DATA) 2907 set_bit(FI_INLINE_DATA, fi->flags); 2908 if (ri->i_inline & F2FS_INLINE_DENTRY) 2909 set_bit(FI_INLINE_DENTRY, fi->flags); 2910 if (ri->i_inline & F2FS_DATA_EXIST) 2911 set_bit(FI_DATA_EXIST, fi->flags); 2912 if (ri->i_inline & F2FS_INLINE_DOTS) 2913 set_bit(FI_INLINE_DOTS, fi->flags); 2914 if (ri->i_inline & F2FS_EXTRA_ATTR) 2915 set_bit(FI_EXTRA_ATTR, fi->flags); 2916 if (ri->i_inline & F2FS_PIN_FILE) 2917 set_bit(FI_PIN_FILE, fi->flags); 2918 if (ri->i_inline & F2FS_COMPRESS_RELEASED) 2919 set_bit(FI_COMPRESS_RELEASED, fi->flags); 2920 } 2921 2922 static inline void set_raw_inline(struct inode *inode, struct f2fs_inode *ri) 2923 { 2924 ri->i_inline = 0; 2925 2926 if (is_inode_flag_set(inode, FI_INLINE_XATTR)) 2927 ri->i_inline |= F2FS_INLINE_XATTR; 2928 if (is_inode_flag_set(inode, FI_INLINE_DATA)) 2929 ri->i_inline |= F2FS_INLINE_DATA; 2930 if (is_inode_flag_set(inode, FI_INLINE_DENTRY)) 2931 ri->i_inline |= F2FS_INLINE_DENTRY; 2932 if (is_inode_flag_set(inode, FI_DATA_EXIST)) 2933 ri->i_inline |= F2FS_DATA_EXIST; 2934 if (is_inode_flag_set(inode, FI_INLINE_DOTS)) 2935 ri->i_inline |= F2FS_INLINE_DOTS; 2936 if (is_inode_flag_set(inode, FI_EXTRA_ATTR)) 2937 ri->i_inline |= F2FS_EXTRA_ATTR; 2938 if (is_inode_flag_set(inode, FI_PIN_FILE)) 2939 ri->i_inline |= F2FS_PIN_FILE; 2940 if (is_inode_flag_set(inode, FI_COMPRESS_RELEASED)) 2941 ri->i_inline |= F2FS_COMPRESS_RELEASED; 2942 } 2943 2944 static inline int f2fs_has_extra_attr(struct inode *inode) 2945 { 2946 return is_inode_flag_set(inode, FI_EXTRA_ATTR); 2947 } 2948 2949 static inline int f2fs_has_inline_xattr(struct inode *inode) 2950 { 2951 return is_inode_flag_set(inode, FI_INLINE_XATTR); 2952 } 2953 2954 static inline int f2fs_compressed_file(struct inode *inode) 2955 { 2956 return S_ISREG(inode->i_mode) && 2957 is_inode_flag_set(inode, FI_COMPRESSED_FILE); 2958 } 2959 2960 static inline bool f2fs_need_compress_data(struct inode *inode) 2961 { 2962 int compress_mode = F2FS_OPTION(F2FS_I_SB(inode)).compress_mode; 2963 2964 if (!f2fs_compressed_file(inode)) 2965 return false; 2966 2967 if (compress_mode == COMPR_MODE_FS) 2968 return true; 2969 else if (compress_mode == COMPR_MODE_USER && 2970 is_inode_flag_set(inode, FI_ENABLE_COMPRESS)) 2971 return true; 2972 2973 return false; 2974 } 2975 2976 static inline unsigned int addrs_per_inode(struct inode *inode) 2977 { 2978 unsigned int addrs = CUR_ADDRS_PER_INODE(inode) - 2979 get_inline_xattr_addrs(inode); 2980 2981 if (!f2fs_compressed_file(inode)) 2982 return addrs; 2983 return ALIGN_DOWN(addrs, F2FS_I(inode)->i_cluster_size); 2984 } 2985 2986 static inline unsigned int addrs_per_block(struct inode *inode) 2987 { 2988 if (!f2fs_compressed_file(inode)) 2989 return DEF_ADDRS_PER_BLOCK; 2990 return ALIGN_DOWN(DEF_ADDRS_PER_BLOCK, F2FS_I(inode)->i_cluster_size); 2991 } 2992 2993 static inline void *inline_xattr_addr(struct inode *inode, struct page *page) 2994 { 2995 struct f2fs_inode *ri = F2FS_INODE(page); 2996 2997 return (void *)&(ri->i_addr[DEF_ADDRS_PER_INODE - 2998 get_inline_xattr_addrs(inode)]); 2999 } 3000 3001 static inline int inline_xattr_size(struct inode *inode) 3002 { 3003 if (f2fs_has_inline_xattr(inode)) 3004 return get_inline_xattr_addrs(inode) * sizeof(__le32); 3005 return 0; 3006 } 3007 3008 static inline int f2fs_has_inline_data(struct inode *inode) 3009 { 3010 return is_inode_flag_set(inode, FI_INLINE_DATA); 3011 } 3012 3013 static inline int f2fs_exist_data(struct inode *inode) 3014 { 3015 return is_inode_flag_set(inode, FI_DATA_EXIST); 3016 } 3017 3018 static inline int f2fs_has_inline_dots(struct inode *inode) 3019 { 3020 return is_inode_flag_set(inode, FI_INLINE_DOTS); 3021 } 3022 3023 static inline int f2fs_is_mmap_file(struct inode *inode) 3024 { 3025 return is_inode_flag_set(inode, FI_MMAP_FILE); 3026 } 3027 3028 static inline bool f2fs_is_pinned_file(struct inode *inode) 3029 { 3030 return is_inode_flag_set(inode, FI_PIN_FILE); 3031 } 3032 3033 static inline bool f2fs_is_atomic_file(struct inode *inode) 3034 { 3035 return is_inode_flag_set(inode, FI_ATOMIC_FILE); 3036 } 3037 3038 static inline bool f2fs_is_commit_atomic_write(struct inode *inode) 3039 { 3040 return is_inode_flag_set(inode, FI_ATOMIC_COMMIT); 3041 } 3042 3043 static inline bool f2fs_is_volatile_file(struct inode *inode) 3044 { 3045 return is_inode_flag_set(inode, FI_VOLATILE_FILE); 3046 } 3047 3048 static inline bool f2fs_is_first_block_written(struct inode *inode) 3049 { 3050 return is_inode_flag_set(inode, FI_FIRST_BLOCK_WRITTEN); 3051 } 3052 3053 static inline bool f2fs_is_drop_cache(struct inode *inode) 3054 { 3055 return is_inode_flag_set(inode, FI_DROP_CACHE); 3056 } 3057 3058 static inline void *inline_data_addr(struct inode *inode, struct page *page) 3059 { 3060 struct f2fs_inode *ri = F2FS_INODE(page); 3061 int extra_size = get_extra_isize(inode); 3062 3063 return (void *)&(ri->i_addr[extra_size + DEF_INLINE_RESERVED_SIZE]); 3064 } 3065 3066 static inline int f2fs_has_inline_dentry(struct inode *inode) 3067 { 3068 return is_inode_flag_set(inode, FI_INLINE_DENTRY); 3069 } 3070 3071 static inline int is_file(struct inode *inode, int type) 3072 { 3073 return F2FS_I(inode)->i_advise & type; 3074 } 3075 3076 static inline void set_file(struct inode *inode, int type) 3077 { 3078 F2FS_I(inode)->i_advise |= type; 3079 f2fs_mark_inode_dirty_sync(inode, true); 3080 } 3081 3082 static inline void clear_file(struct inode *inode, int type) 3083 { 3084 F2FS_I(inode)->i_advise &= ~type; 3085 f2fs_mark_inode_dirty_sync(inode, true); 3086 } 3087 3088 static inline bool f2fs_is_time_consistent(struct inode *inode) 3089 { 3090 if (!timespec64_equal(F2FS_I(inode)->i_disk_time, &inode->i_atime)) 3091 return false; 3092 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 1, &inode->i_ctime)) 3093 return false; 3094 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 2, &inode->i_mtime)) 3095 return false; 3096 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 3, 3097 &F2FS_I(inode)->i_crtime)) 3098 return false; 3099 return true; 3100 } 3101 3102 static inline bool f2fs_skip_inode_update(struct inode *inode, int dsync) 3103 { 3104 bool ret; 3105 3106 if (dsync) { 3107 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 3108 3109 spin_lock(&sbi->inode_lock[DIRTY_META]); 3110 ret = list_empty(&F2FS_I(inode)->gdirty_list); 3111 spin_unlock(&sbi->inode_lock[DIRTY_META]); 3112 return ret; 3113 } 3114 if (!is_inode_flag_set(inode, FI_AUTO_RECOVER) || 3115 file_keep_isize(inode) || 3116 i_size_read(inode) & ~PAGE_MASK) 3117 return false; 3118 3119 if (!f2fs_is_time_consistent(inode)) 3120 return false; 3121 3122 spin_lock(&F2FS_I(inode)->i_size_lock); 3123 ret = F2FS_I(inode)->last_disk_size == i_size_read(inode); 3124 spin_unlock(&F2FS_I(inode)->i_size_lock); 3125 3126 return ret; 3127 } 3128 3129 static inline bool f2fs_readonly(struct super_block *sb) 3130 { 3131 return sb_rdonly(sb); 3132 } 3133 3134 static inline bool f2fs_cp_error(struct f2fs_sb_info *sbi) 3135 { 3136 return is_set_ckpt_flags(sbi, CP_ERROR_FLAG); 3137 } 3138 3139 static inline bool is_dot_dotdot(const u8 *name, size_t len) 3140 { 3141 if (len == 1 && name[0] == '.') 3142 return true; 3143 3144 if (len == 2 && name[0] == '.' && name[1] == '.') 3145 return true; 3146 3147 return false; 3148 } 3149 3150 static inline bool f2fs_may_extent_tree(struct inode *inode) 3151 { 3152 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 3153 3154 if (!test_opt(sbi, EXTENT_CACHE) || 3155 is_inode_flag_set(inode, FI_NO_EXTENT) || 3156 is_inode_flag_set(inode, FI_COMPRESSED_FILE)) 3157 return false; 3158 3159 /* 3160 * for recovered files during mount do not create extents 3161 * if shrinker is not registered. 3162 */ 3163 if (list_empty(&sbi->s_list)) 3164 return false; 3165 3166 return S_ISREG(inode->i_mode); 3167 } 3168 3169 static inline void *f2fs_kmalloc(struct f2fs_sb_info *sbi, 3170 size_t size, gfp_t flags) 3171 { 3172 if (time_to_inject(sbi, FAULT_KMALLOC)) { 3173 f2fs_show_injection_info(sbi, FAULT_KMALLOC); 3174 return NULL; 3175 } 3176 3177 return kmalloc(size, flags); 3178 } 3179 3180 static inline void *f2fs_kzalloc(struct f2fs_sb_info *sbi, 3181 size_t size, gfp_t flags) 3182 { 3183 return f2fs_kmalloc(sbi, size, flags | __GFP_ZERO); 3184 } 3185 3186 static inline void *f2fs_kvmalloc(struct f2fs_sb_info *sbi, 3187 size_t size, gfp_t flags) 3188 { 3189 if (time_to_inject(sbi, FAULT_KVMALLOC)) { 3190 f2fs_show_injection_info(sbi, FAULT_KVMALLOC); 3191 return NULL; 3192 } 3193 3194 return kvmalloc(size, flags); 3195 } 3196 3197 static inline void *f2fs_kvzalloc(struct f2fs_sb_info *sbi, 3198 size_t size, gfp_t flags) 3199 { 3200 return f2fs_kvmalloc(sbi, size, flags | __GFP_ZERO); 3201 } 3202 3203 static inline int get_extra_isize(struct inode *inode) 3204 { 3205 return F2FS_I(inode)->i_extra_isize / sizeof(__le32); 3206 } 3207 3208 static inline int get_inline_xattr_addrs(struct inode *inode) 3209 { 3210 return F2FS_I(inode)->i_inline_xattr_size; 3211 } 3212 3213 #define f2fs_get_inode_mode(i) \ 3214 ((is_inode_flag_set(i, FI_ACL_MODE)) ? \ 3215 (F2FS_I(i)->i_acl_mode) : ((i)->i_mode)) 3216 3217 #define F2FS_TOTAL_EXTRA_ATTR_SIZE \ 3218 (offsetof(struct f2fs_inode, i_extra_end) - \ 3219 offsetof(struct f2fs_inode, i_extra_isize)) \ 3220 3221 #define F2FS_OLD_ATTRIBUTE_SIZE (offsetof(struct f2fs_inode, i_addr)) 3222 #define F2FS_FITS_IN_INODE(f2fs_inode, extra_isize, field) \ 3223 ((offsetof(typeof(*(f2fs_inode)), field) + \ 3224 sizeof((f2fs_inode)->field)) \ 3225 <= (F2FS_OLD_ATTRIBUTE_SIZE + (extra_isize))) \ 3226 3227 #define DEFAULT_IOSTAT_PERIOD_MS 3000 3228 #define MIN_IOSTAT_PERIOD_MS 100 3229 /* maximum period of iostat tracing is 1 day */ 3230 #define MAX_IOSTAT_PERIOD_MS 8640000 3231 3232 static inline void f2fs_reset_iostat(struct f2fs_sb_info *sbi) 3233 { 3234 int i; 3235 3236 spin_lock(&sbi->iostat_lock); 3237 for (i = 0; i < NR_IO_TYPE; i++) { 3238 sbi->rw_iostat[i] = 0; 3239 sbi->prev_rw_iostat[i] = 0; 3240 } 3241 spin_unlock(&sbi->iostat_lock); 3242 } 3243 3244 extern void f2fs_record_iostat(struct f2fs_sb_info *sbi); 3245 3246 static inline void f2fs_update_iostat(struct f2fs_sb_info *sbi, 3247 enum iostat_type type, unsigned long long io_bytes) 3248 { 3249 if (!sbi->iostat_enable) 3250 return; 3251 spin_lock(&sbi->iostat_lock); 3252 sbi->rw_iostat[type] += io_bytes; 3253 3254 if (type == APP_WRITE_IO || type == APP_DIRECT_IO) 3255 sbi->rw_iostat[APP_BUFFERED_IO] = 3256 sbi->rw_iostat[APP_WRITE_IO] - 3257 sbi->rw_iostat[APP_DIRECT_IO]; 3258 3259 if (type == APP_READ_IO || type == APP_DIRECT_READ_IO) 3260 sbi->rw_iostat[APP_BUFFERED_READ_IO] = 3261 sbi->rw_iostat[APP_READ_IO] - 3262 sbi->rw_iostat[APP_DIRECT_READ_IO]; 3263 spin_unlock(&sbi->iostat_lock); 3264 3265 f2fs_record_iostat(sbi); 3266 } 3267 3268 #define __is_large_section(sbi) ((sbi)->segs_per_sec > 1) 3269 3270 #define __is_meta_io(fio) (PAGE_TYPE_OF_BIO((fio)->type) == META) 3271 3272 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi, 3273 block_t blkaddr, int type); 3274 static inline void verify_blkaddr(struct f2fs_sb_info *sbi, 3275 block_t blkaddr, int type) 3276 { 3277 if (!f2fs_is_valid_blkaddr(sbi, blkaddr, type)) { 3278 f2fs_err(sbi, "invalid blkaddr: %u, type: %d, run fsck to fix.", 3279 blkaddr, type); 3280 f2fs_bug_on(sbi, 1); 3281 } 3282 } 3283 3284 static inline bool __is_valid_data_blkaddr(block_t blkaddr) 3285 { 3286 if (blkaddr == NEW_ADDR || blkaddr == NULL_ADDR || 3287 blkaddr == COMPRESS_ADDR) 3288 return false; 3289 return true; 3290 } 3291 3292 /* 3293 * file.c 3294 */ 3295 int f2fs_sync_file(struct file *file, loff_t start, loff_t end, int datasync); 3296 void f2fs_truncate_data_blocks(struct dnode_of_data *dn); 3297 int f2fs_do_truncate_blocks(struct inode *inode, u64 from, bool lock); 3298 int f2fs_truncate_blocks(struct inode *inode, u64 from, bool lock); 3299 int f2fs_truncate(struct inode *inode); 3300 int f2fs_getattr(struct user_namespace *mnt_userns, const struct path *path, 3301 struct kstat *stat, u32 request_mask, unsigned int flags); 3302 int f2fs_setattr(struct user_namespace *mnt_userns, struct dentry *dentry, 3303 struct iattr *attr); 3304 int f2fs_truncate_hole(struct inode *inode, pgoff_t pg_start, pgoff_t pg_end); 3305 void f2fs_truncate_data_blocks_range(struct dnode_of_data *dn, int count); 3306 int f2fs_precache_extents(struct inode *inode); 3307 int f2fs_fileattr_get(struct dentry *dentry, struct fileattr *fa); 3308 int f2fs_fileattr_set(struct user_namespace *mnt_userns, 3309 struct dentry *dentry, struct fileattr *fa); 3310 long f2fs_ioctl(struct file *filp, unsigned int cmd, unsigned long arg); 3311 long f2fs_compat_ioctl(struct file *file, unsigned int cmd, unsigned long arg); 3312 int f2fs_transfer_project_quota(struct inode *inode, kprojid_t kprojid); 3313 int f2fs_pin_file_control(struct inode *inode, bool inc); 3314 3315 /* 3316 * inode.c 3317 */ 3318 void f2fs_set_inode_flags(struct inode *inode); 3319 bool f2fs_inode_chksum_verify(struct f2fs_sb_info *sbi, struct page *page); 3320 void f2fs_inode_chksum_set(struct f2fs_sb_info *sbi, struct page *page); 3321 struct inode *f2fs_iget(struct super_block *sb, unsigned long ino); 3322 struct inode *f2fs_iget_retry(struct super_block *sb, unsigned long ino); 3323 int f2fs_try_to_free_nats(struct f2fs_sb_info *sbi, int nr_shrink); 3324 void f2fs_update_inode(struct inode *inode, struct page *node_page); 3325 void f2fs_update_inode_page(struct inode *inode); 3326 int f2fs_write_inode(struct inode *inode, struct writeback_control *wbc); 3327 void f2fs_evict_inode(struct inode *inode); 3328 void f2fs_handle_failed_inode(struct inode *inode); 3329 3330 /* 3331 * namei.c 3332 */ 3333 int f2fs_update_extension_list(struct f2fs_sb_info *sbi, const char *name, 3334 bool hot, bool set); 3335 struct dentry *f2fs_get_parent(struct dentry *child); 3336 3337 /* 3338 * dir.c 3339 */ 3340 unsigned char f2fs_get_de_type(struct f2fs_dir_entry *de); 3341 int f2fs_init_casefolded_name(const struct inode *dir, 3342 struct f2fs_filename *fname); 3343 int f2fs_setup_filename(struct inode *dir, const struct qstr *iname, 3344 int lookup, struct f2fs_filename *fname); 3345 int f2fs_prepare_lookup(struct inode *dir, struct dentry *dentry, 3346 struct f2fs_filename *fname); 3347 void f2fs_free_filename(struct f2fs_filename *fname); 3348 struct f2fs_dir_entry *f2fs_find_target_dentry(const struct f2fs_dentry_ptr *d, 3349 const struct f2fs_filename *fname, int *max_slots); 3350 int f2fs_fill_dentries(struct dir_context *ctx, struct f2fs_dentry_ptr *d, 3351 unsigned int start_pos, struct fscrypt_str *fstr); 3352 void f2fs_do_make_empty_dir(struct inode *inode, struct inode *parent, 3353 struct f2fs_dentry_ptr *d); 3354 struct page *f2fs_init_inode_metadata(struct inode *inode, struct inode *dir, 3355 const struct f2fs_filename *fname, struct page *dpage); 3356 void f2fs_update_parent_metadata(struct inode *dir, struct inode *inode, 3357 unsigned int current_depth); 3358 int f2fs_room_for_filename(const void *bitmap, int slots, int max_slots); 3359 void f2fs_drop_nlink(struct inode *dir, struct inode *inode); 3360 struct f2fs_dir_entry *__f2fs_find_entry(struct inode *dir, 3361 const struct f2fs_filename *fname, 3362 struct page **res_page); 3363 struct f2fs_dir_entry *f2fs_find_entry(struct inode *dir, 3364 const struct qstr *child, struct page **res_page); 3365 struct f2fs_dir_entry *f2fs_parent_dir(struct inode *dir, struct page **p); 3366 ino_t f2fs_inode_by_name(struct inode *dir, const struct qstr *qstr, 3367 struct page **page); 3368 void f2fs_set_link(struct inode *dir, struct f2fs_dir_entry *de, 3369 struct page *page, struct inode *inode); 3370 bool f2fs_has_enough_room(struct inode *dir, struct page *ipage, 3371 const struct f2fs_filename *fname); 3372 void f2fs_update_dentry(nid_t ino, umode_t mode, struct f2fs_dentry_ptr *d, 3373 const struct fscrypt_str *name, f2fs_hash_t name_hash, 3374 unsigned int bit_pos); 3375 int f2fs_add_regular_entry(struct inode *dir, const struct f2fs_filename *fname, 3376 struct inode *inode, nid_t ino, umode_t mode); 3377 int f2fs_add_dentry(struct inode *dir, const struct f2fs_filename *fname, 3378 struct inode *inode, nid_t ino, umode_t mode); 3379 int f2fs_do_add_link(struct inode *dir, const struct qstr *name, 3380 struct inode *inode, nid_t ino, umode_t mode); 3381 void f2fs_delete_entry(struct f2fs_dir_entry *dentry, struct page *page, 3382 struct inode *dir, struct inode *inode); 3383 int f2fs_do_tmpfile(struct inode *inode, struct inode *dir); 3384 bool f2fs_empty_dir(struct inode *dir); 3385 3386 static inline int f2fs_add_link(struct dentry *dentry, struct inode *inode) 3387 { 3388 if (fscrypt_is_nokey_name(dentry)) 3389 return -ENOKEY; 3390 return f2fs_do_add_link(d_inode(dentry->d_parent), &dentry->d_name, 3391 inode, inode->i_ino, inode->i_mode); 3392 } 3393 3394 /* 3395 * super.c 3396 */ 3397 int f2fs_inode_dirtied(struct inode *inode, bool sync); 3398 void f2fs_inode_synced(struct inode *inode); 3399 int f2fs_enable_quota_files(struct f2fs_sb_info *sbi, bool rdonly); 3400 int f2fs_quota_sync(struct super_block *sb, int type); 3401 loff_t max_file_blocks(struct inode *inode); 3402 void f2fs_quota_off_umount(struct super_block *sb); 3403 int f2fs_commit_super(struct f2fs_sb_info *sbi, bool recover); 3404 int f2fs_sync_fs(struct super_block *sb, int sync); 3405 int f2fs_sanity_check_ckpt(struct f2fs_sb_info *sbi); 3406 3407 /* 3408 * hash.c 3409 */ 3410 void f2fs_hash_filename(const struct inode *dir, struct f2fs_filename *fname); 3411 3412 /* 3413 * node.c 3414 */ 3415 struct node_info; 3416 3417 int f2fs_check_nid_range(struct f2fs_sb_info *sbi, nid_t nid); 3418 bool f2fs_available_free_memory(struct f2fs_sb_info *sbi, int type); 3419 bool f2fs_in_warm_node_list(struct f2fs_sb_info *sbi, struct page *page); 3420 void f2fs_init_fsync_node_info(struct f2fs_sb_info *sbi); 3421 void f2fs_del_fsync_node_entry(struct f2fs_sb_info *sbi, struct page *page); 3422 void f2fs_reset_fsync_node_info(struct f2fs_sb_info *sbi); 3423 int f2fs_need_dentry_mark(struct f2fs_sb_info *sbi, nid_t nid); 3424 bool f2fs_is_checkpointed_node(struct f2fs_sb_info *sbi, nid_t nid); 3425 bool f2fs_need_inode_block_update(struct f2fs_sb_info *sbi, nid_t ino); 3426 int f2fs_get_node_info(struct f2fs_sb_info *sbi, nid_t nid, 3427 struct node_info *ni); 3428 pgoff_t f2fs_get_next_page_offset(struct dnode_of_data *dn, pgoff_t pgofs); 3429 int f2fs_get_dnode_of_data(struct dnode_of_data *dn, pgoff_t index, int mode); 3430 int f2fs_truncate_inode_blocks(struct inode *inode, pgoff_t from); 3431 int f2fs_truncate_xattr_node(struct inode *inode); 3432 int f2fs_wait_on_node_pages_writeback(struct f2fs_sb_info *sbi, 3433 unsigned int seq_id); 3434 int f2fs_remove_inode_page(struct inode *inode); 3435 struct page *f2fs_new_inode_page(struct inode *inode); 3436 struct page *f2fs_new_node_page(struct dnode_of_data *dn, unsigned int ofs); 3437 void f2fs_ra_node_page(struct f2fs_sb_info *sbi, nid_t nid); 3438 struct page *f2fs_get_node_page(struct f2fs_sb_info *sbi, pgoff_t nid); 3439 struct page *f2fs_get_node_page_ra(struct page *parent, int start); 3440 int f2fs_move_node_page(struct page *node_page, int gc_type); 3441 void f2fs_flush_inline_data(struct f2fs_sb_info *sbi); 3442 int f2fs_fsync_node_pages(struct f2fs_sb_info *sbi, struct inode *inode, 3443 struct writeback_control *wbc, bool atomic, 3444 unsigned int *seq_id); 3445 int f2fs_sync_node_pages(struct f2fs_sb_info *sbi, 3446 struct writeback_control *wbc, 3447 bool do_balance, enum iostat_type io_type); 3448 int f2fs_build_free_nids(struct f2fs_sb_info *sbi, bool sync, bool mount); 3449 bool f2fs_alloc_nid(struct f2fs_sb_info *sbi, nid_t *nid); 3450 void f2fs_alloc_nid_done(struct f2fs_sb_info *sbi, nid_t nid); 3451 void f2fs_alloc_nid_failed(struct f2fs_sb_info *sbi, nid_t nid); 3452 int f2fs_try_to_free_nids(struct f2fs_sb_info *sbi, int nr_shrink); 3453 int f2fs_recover_inline_xattr(struct inode *inode, struct page *page); 3454 int f2fs_recover_xattr_data(struct inode *inode, struct page *page); 3455 int f2fs_recover_inode_page(struct f2fs_sb_info *sbi, struct page *page); 3456 int f2fs_restore_node_summary(struct f2fs_sb_info *sbi, 3457 unsigned int segno, struct f2fs_summary_block *sum); 3458 int f2fs_flush_nat_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3459 int f2fs_build_node_manager(struct f2fs_sb_info *sbi); 3460 void f2fs_destroy_node_manager(struct f2fs_sb_info *sbi); 3461 int __init f2fs_create_node_manager_caches(void); 3462 void f2fs_destroy_node_manager_caches(void); 3463 3464 /* 3465 * segment.c 3466 */ 3467 bool f2fs_need_SSR(struct f2fs_sb_info *sbi); 3468 void f2fs_register_inmem_page(struct inode *inode, struct page *page); 3469 void f2fs_drop_inmem_pages_all(struct f2fs_sb_info *sbi, bool gc_failure); 3470 void f2fs_drop_inmem_pages(struct inode *inode); 3471 void f2fs_drop_inmem_page(struct inode *inode, struct page *page); 3472 int f2fs_commit_inmem_pages(struct inode *inode); 3473 void f2fs_balance_fs(struct f2fs_sb_info *sbi, bool need); 3474 void f2fs_balance_fs_bg(struct f2fs_sb_info *sbi, bool from_bg); 3475 int f2fs_issue_flush(struct f2fs_sb_info *sbi, nid_t ino); 3476 int f2fs_create_flush_cmd_control(struct f2fs_sb_info *sbi); 3477 int f2fs_flush_device_cache(struct f2fs_sb_info *sbi); 3478 void f2fs_destroy_flush_cmd_control(struct f2fs_sb_info *sbi, bool free); 3479 void f2fs_invalidate_blocks(struct f2fs_sb_info *sbi, block_t addr); 3480 bool f2fs_is_checkpointed_data(struct f2fs_sb_info *sbi, block_t blkaddr); 3481 void f2fs_drop_discard_cmd(struct f2fs_sb_info *sbi); 3482 void f2fs_stop_discard_thread(struct f2fs_sb_info *sbi); 3483 bool f2fs_issue_discard_timeout(struct f2fs_sb_info *sbi); 3484 void f2fs_clear_prefree_segments(struct f2fs_sb_info *sbi, 3485 struct cp_control *cpc); 3486 void f2fs_dirty_to_prefree(struct f2fs_sb_info *sbi); 3487 block_t f2fs_get_unusable_blocks(struct f2fs_sb_info *sbi); 3488 int f2fs_disable_cp_again(struct f2fs_sb_info *sbi, block_t unusable); 3489 void f2fs_release_discard_addrs(struct f2fs_sb_info *sbi); 3490 int f2fs_npages_for_summary_flush(struct f2fs_sb_info *sbi, bool for_ra); 3491 bool f2fs_segment_has_free_slot(struct f2fs_sb_info *sbi, int segno); 3492 void f2fs_init_inmem_curseg(struct f2fs_sb_info *sbi); 3493 void f2fs_save_inmem_curseg(struct f2fs_sb_info *sbi); 3494 void f2fs_restore_inmem_curseg(struct f2fs_sb_info *sbi); 3495 void f2fs_get_new_segment(struct f2fs_sb_info *sbi, 3496 unsigned int *newseg, bool new_sec, int dir); 3497 void f2fs_allocate_segment_for_resize(struct f2fs_sb_info *sbi, int type, 3498 unsigned int start, unsigned int end); 3499 void f2fs_allocate_new_section(struct f2fs_sb_info *sbi, int type, bool force); 3500 void f2fs_allocate_new_segments(struct f2fs_sb_info *sbi); 3501 int f2fs_trim_fs(struct f2fs_sb_info *sbi, struct fstrim_range *range); 3502 bool f2fs_exist_trim_candidates(struct f2fs_sb_info *sbi, 3503 struct cp_control *cpc); 3504 struct page *f2fs_get_sum_page(struct f2fs_sb_info *sbi, unsigned int segno); 3505 void f2fs_update_meta_page(struct f2fs_sb_info *sbi, void *src, 3506 block_t blk_addr); 3507 void f2fs_do_write_meta_page(struct f2fs_sb_info *sbi, struct page *page, 3508 enum iostat_type io_type); 3509 void f2fs_do_write_node_page(unsigned int nid, struct f2fs_io_info *fio); 3510 void f2fs_outplace_write_data(struct dnode_of_data *dn, 3511 struct f2fs_io_info *fio); 3512 int f2fs_inplace_write_data(struct f2fs_io_info *fio); 3513 void f2fs_do_replace_block(struct f2fs_sb_info *sbi, struct f2fs_summary *sum, 3514 block_t old_blkaddr, block_t new_blkaddr, 3515 bool recover_curseg, bool recover_newaddr, 3516 bool from_gc); 3517 void f2fs_replace_block(struct f2fs_sb_info *sbi, struct dnode_of_data *dn, 3518 block_t old_addr, block_t new_addr, 3519 unsigned char version, bool recover_curseg, 3520 bool recover_newaddr); 3521 void f2fs_allocate_data_block(struct f2fs_sb_info *sbi, struct page *page, 3522 block_t old_blkaddr, block_t *new_blkaddr, 3523 struct f2fs_summary *sum, int type, 3524 struct f2fs_io_info *fio); 3525 void f2fs_wait_on_page_writeback(struct page *page, 3526 enum page_type type, bool ordered, bool locked); 3527 void f2fs_wait_on_block_writeback(struct inode *inode, block_t blkaddr); 3528 void f2fs_wait_on_block_writeback_range(struct inode *inode, block_t blkaddr, 3529 block_t len); 3530 void f2fs_write_data_summaries(struct f2fs_sb_info *sbi, block_t start_blk); 3531 void f2fs_write_node_summaries(struct f2fs_sb_info *sbi, block_t start_blk); 3532 int f2fs_lookup_journal_in_cursum(struct f2fs_journal *journal, int type, 3533 unsigned int val, int alloc); 3534 void f2fs_flush_sit_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3535 int f2fs_fix_curseg_write_pointer(struct f2fs_sb_info *sbi); 3536 int f2fs_check_write_pointer(struct f2fs_sb_info *sbi); 3537 int f2fs_build_segment_manager(struct f2fs_sb_info *sbi); 3538 void f2fs_destroy_segment_manager(struct f2fs_sb_info *sbi); 3539 int __init f2fs_create_segment_manager_caches(void); 3540 void f2fs_destroy_segment_manager_caches(void); 3541 int f2fs_rw_hint_to_seg_type(enum rw_hint hint); 3542 enum rw_hint f2fs_io_type_to_rw_hint(struct f2fs_sb_info *sbi, 3543 enum page_type type, enum temp_type temp); 3544 unsigned int f2fs_usable_segs_in_sec(struct f2fs_sb_info *sbi, 3545 unsigned int segno); 3546 unsigned int f2fs_usable_blks_in_seg(struct f2fs_sb_info *sbi, 3547 unsigned int segno); 3548 3549 /* 3550 * checkpoint.c 3551 */ 3552 void f2fs_stop_checkpoint(struct f2fs_sb_info *sbi, bool end_io); 3553 struct page *f2fs_grab_meta_page(struct f2fs_sb_info *sbi, pgoff_t index); 3554 struct page *f2fs_get_meta_page(struct f2fs_sb_info *sbi, pgoff_t index); 3555 struct page *f2fs_get_meta_page_retry(struct f2fs_sb_info *sbi, pgoff_t index); 3556 struct page *f2fs_get_tmp_page(struct f2fs_sb_info *sbi, pgoff_t index); 3557 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi, 3558 block_t blkaddr, int type); 3559 int f2fs_ra_meta_pages(struct f2fs_sb_info *sbi, block_t start, int nrpages, 3560 int type, bool sync); 3561 void f2fs_ra_meta_pages_cond(struct f2fs_sb_info *sbi, pgoff_t index); 3562 long f2fs_sync_meta_pages(struct f2fs_sb_info *sbi, enum page_type type, 3563 long nr_to_write, enum iostat_type io_type); 3564 void f2fs_add_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type); 3565 void f2fs_remove_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type); 3566 void f2fs_release_ino_entry(struct f2fs_sb_info *sbi, bool all); 3567 bool f2fs_exist_written_data(struct f2fs_sb_info *sbi, nid_t ino, int mode); 3568 void f2fs_set_dirty_device(struct f2fs_sb_info *sbi, nid_t ino, 3569 unsigned int devidx, int type); 3570 bool f2fs_is_dirty_device(struct f2fs_sb_info *sbi, nid_t ino, 3571 unsigned int devidx, int type); 3572 int f2fs_sync_inode_meta(struct f2fs_sb_info *sbi); 3573 int f2fs_acquire_orphan_inode(struct f2fs_sb_info *sbi); 3574 void f2fs_release_orphan_inode(struct f2fs_sb_info *sbi); 3575 void f2fs_add_orphan_inode(struct inode *inode); 3576 void f2fs_remove_orphan_inode(struct f2fs_sb_info *sbi, nid_t ino); 3577 int f2fs_recover_orphan_inodes(struct f2fs_sb_info *sbi); 3578 int f2fs_get_valid_checkpoint(struct f2fs_sb_info *sbi); 3579 void f2fs_update_dirty_page(struct inode *inode, struct page *page); 3580 void f2fs_remove_dirty_inode(struct inode *inode); 3581 int f2fs_sync_dirty_inodes(struct f2fs_sb_info *sbi, enum inode_type type); 3582 void f2fs_wait_on_all_pages(struct f2fs_sb_info *sbi, int type); 3583 u64 f2fs_get_sectors_written(struct f2fs_sb_info *sbi); 3584 int f2fs_write_checkpoint(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3585 void f2fs_init_ino_entry_info(struct f2fs_sb_info *sbi); 3586 int __init f2fs_create_checkpoint_caches(void); 3587 void f2fs_destroy_checkpoint_caches(void); 3588 int f2fs_issue_checkpoint(struct f2fs_sb_info *sbi); 3589 int f2fs_start_ckpt_thread(struct f2fs_sb_info *sbi); 3590 void f2fs_stop_ckpt_thread(struct f2fs_sb_info *sbi); 3591 void f2fs_init_ckpt_req_control(struct f2fs_sb_info *sbi); 3592 3593 /* 3594 * data.c 3595 */ 3596 int __init f2fs_init_bioset(void); 3597 void f2fs_destroy_bioset(void); 3598 int f2fs_init_bio_entry_cache(void); 3599 void f2fs_destroy_bio_entry_cache(void); 3600 void f2fs_submit_bio(struct f2fs_sb_info *sbi, 3601 struct bio *bio, enum page_type type); 3602 void f2fs_submit_merged_write(struct f2fs_sb_info *sbi, enum page_type type); 3603 void f2fs_submit_merged_write_cond(struct f2fs_sb_info *sbi, 3604 struct inode *inode, struct page *page, 3605 nid_t ino, enum page_type type); 3606 void f2fs_submit_merged_ipu_write(struct f2fs_sb_info *sbi, 3607 struct bio **bio, struct page *page); 3608 void f2fs_flush_merged_writes(struct f2fs_sb_info *sbi); 3609 int f2fs_submit_page_bio(struct f2fs_io_info *fio); 3610 int f2fs_merge_page_bio(struct f2fs_io_info *fio); 3611 void f2fs_submit_page_write(struct f2fs_io_info *fio); 3612 struct block_device *f2fs_target_device(struct f2fs_sb_info *sbi, 3613 block_t blk_addr, struct bio *bio); 3614 int f2fs_target_device_index(struct f2fs_sb_info *sbi, block_t blkaddr); 3615 void f2fs_set_data_blkaddr(struct dnode_of_data *dn); 3616 void f2fs_update_data_blkaddr(struct dnode_of_data *dn, block_t blkaddr); 3617 int f2fs_reserve_new_blocks(struct dnode_of_data *dn, blkcnt_t count); 3618 int f2fs_reserve_new_block(struct dnode_of_data *dn); 3619 int f2fs_get_block(struct dnode_of_data *dn, pgoff_t index); 3620 int f2fs_preallocate_blocks(struct kiocb *iocb, struct iov_iter *from); 3621 int f2fs_reserve_block(struct dnode_of_data *dn, pgoff_t index); 3622 struct page *f2fs_get_read_data_page(struct inode *inode, pgoff_t index, 3623 int op_flags, bool for_write); 3624 struct page *f2fs_find_data_page(struct inode *inode, pgoff_t index); 3625 struct page *f2fs_get_lock_data_page(struct inode *inode, pgoff_t index, 3626 bool for_write); 3627 struct page *f2fs_get_new_data_page(struct inode *inode, 3628 struct page *ipage, pgoff_t index, bool new_i_size); 3629 int f2fs_do_write_data_page(struct f2fs_io_info *fio); 3630 void f2fs_do_map_lock(struct f2fs_sb_info *sbi, int flag, bool lock); 3631 int f2fs_map_blocks(struct inode *inode, struct f2fs_map_blocks *map, 3632 int create, int flag); 3633 int f2fs_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, 3634 u64 start, u64 len); 3635 int f2fs_encrypt_one_page(struct f2fs_io_info *fio); 3636 bool f2fs_should_update_inplace(struct inode *inode, struct f2fs_io_info *fio); 3637 bool f2fs_should_update_outplace(struct inode *inode, struct f2fs_io_info *fio); 3638 int f2fs_write_single_data_page(struct page *page, int *submitted, 3639 struct bio **bio, sector_t *last_block, 3640 struct writeback_control *wbc, 3641 enum iostat_type io_type, 3642 int compr_blocks, bool allow_balance); 3643 void f2fs_invalidate_page(struct page *page, unsigned int offset, 3644 unsigned int length); 3645 int f2fs_release_page(struct page *page, gfp_t wait); 3646 #ifdef CONFIG_MIGRATION 3647 int f2fs_migrate_page(struct address_space *mapping, struct page *newpage, 3648 struct page *page, enum migrate_mode mode); 3649 #endif 3650 bool f2fs_overwrite_io(struct inode *inode, loff_t pos, size_t len); 3651 void f2fs_clear_page_cache_dirty_tag(struct page *page); 3652 int f2fs_init_post_read_processing(void); 3653 void f2fs_destroy_post_read_processing(void); 3654 int f2fs_init_post_read_wq(struct f2fs_sb_info *sbi); 3655 void f2fs_destroy_post_read_wq(struct f2fs_sb_info *sbi); 3656 3657 /* 3658 * gc.c 3659 */ 3660 int f2fs_start_gc_thread(struct f2fs_sb_info *sbi); 3661 void f2fs_stop_gc_thread(struct f2fs_sb_info *sbi); 3662 block_t f2fs_start_bidx_of_node(unsigned int node_ofs, struct inode *inode); 3663 int f2fs_gc(struct f2fs_sb_info *sbi, bool sync, bool background, bool force, 3664 unsigned int segno); 3665 void f2fs_build_gc_manager(struct f2fs_sb_info *sbi); 3666 int f2fs_resize_fs(struct f2fs_sb_info *sbi, __u64 block_count); 3667 int __init f2fs_create_garbage_collection_cache(void); 3668 void f2fs_destroy_garbage_collection_cache(void); 3669 3670 /* 3671 * recovery.c 3672 */ 3673 int f2fs_recover_fsync_data(struct f2fs_sb_info *sbi, bool check_only); 3674 bool f2fs_space_for_roll_forward(struct f2fs_sb_info *sbi); 3675 int __init f2fs_create_recovery_cache(void); 3676 void f2fs_destroy_recovery_cache(void); 3677 3678 /* 3679 * debug.c 3680 */ 3681 #ifdef CONFIG_F2FS_STAT_FS 3682 struct f2fs_stat_info { 3683 struct list_head stat_list; 3684 struct f2fs_sb_info *sbi; 3685 int all_area_segs, sit_area_segs, nat_area_segs, ssa_area_segs; 3686 int main_area_segs, main_area_sections, main_area_zones; 3687 unsigned long long hit_largest, hit_cached, hit_rbtree; 3688 unsigned long long hit_total, total_ext; 3689 int ext_tree, zombie_tree, ext_node; 3690 int ndirty_node, ndirty_dent, ndirty_meta, ndirty_imeta; 3691 int ndirty_data, ndirty_qdata; 3692 int inmem_pages; 3693 unsigned int ndirty_dirs, ndirty_files, nquota_files, ndirty_all; 3694 int nats, dirty_nats, sits, dirty_sits; 3695 int free_nids, avail_nids, alloc_nids; 3696 int total_count, utilization; 3697 int bg_gc, nr_wb_cp_data, nr_wb_data; 3698 int nr_rd_data, nr_rd_node, nr_rd_meta; 3699 int nr_dio_read, nr_dio_write; 3700 unsigned int io_skip_bggc, other_skip_bggc; 3701 int nr_flushing, nr_flushed, flush_list_empty; 3702 int nr_discarding, nr_discarded; 3703 int nr_discard_cmd; 3704 unsigned int undiscard_blks; 3705 int nr_issued_ckpt, nr_total_ckpt, nr_queued_ckpt; 3706 unsigned int cur_ckpt_time, peak_ckpt_time; 3707 int inline_xattr, inline_inode, inline_dir, append, update, orphans; 3708 int compr_inode; 3709 unsigned long long compr_blocks; 3710 int aw_cnt, max_aw_cnt, vw_cnt, max_vw_cnt; 3711 unsigned int valid_count, valid_node_count, valid_inode_count, discard_blks; 3712 unsigned int bimodal, avg_vblocks; 3713 int util_free, util_valid, util_invalid; 3714 int rsvd_segs, overp_segs; 3715 int dirty_count, node_pages, meta_pages, compress_pages; 3716 int compress_page_hit; 3717 int prefree_count, call_count, cp_count, bg_cp_count; 3718 int tot_segs, node_segs, data_segs, free_segs, free_secs; 3719 int bg_node_segs, bg_data_segs; 3720 int tot_blks, data_blks, node_blks; 3721 int bg_data_blks, bg_node_blks; 3722 unsigned long long skipped_atomic_files[2]; 3723 int curseg[NR_CURSEG_TYPE]; 3724 int cursec[NR_CURSEG_TYPE]; 3725 int curzone[NR_CURSEG_TYPE]; 3726 unsigned int dirty_seg[NR_CURSEG_TYPE]; 3727 unsigned int full_seg[NR_CURSEG_TYPE]; 3728 unsigned int valid_blks[NR_CURSEG_TYPE]; 3729 3730 unsigned int meta_count[META_MAX]; 3731 unsigned int segment_count[2]; 3732 unsigned int block_count[2]; 3733 unsigned int inplace_count; 3734 unsigned long long base_mem, cache_mem, page_mem; 3735 }; 3736 3737 static inline struct f2fs_stat_info *F2FS_STAT(struct f2fs_sb_info *sbi) 3738 { 3739 return (struct f2fs_stat_info *)sbi->stat_info; 3740 } 3741 3742 #define stat_inc_cp_count(si) ((si)->cp_count++) 3743 #define stat_inc_bg_cp_count(si) ((si)->bg_cp_count++) 3744 #define stat_inc_call_count(si) ((si)->call_count++) 3745 #define stat_inc_bggc_count(si) ((si)->bg_gc++) 3746 #define stat_io_skip_bggc_count(sbi) ((sbi)->io_skip_bggc++) 3747 #define stat_other_skip_bggc_count(sbi) ((sbi)->other_skip_bggc++) 3748 #define stat_inc_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]++) 3749 #define stat_dec_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]--) 3750 #define stat_inc_total_hit(sbi) (atomic64_inc(&(sbi)->total_hit_ext)) 3751 #define stat_inc_rbtree_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_rbtree)) 3752 #define stat_inc_largest_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_largest)) 3753 #define stat_inc_cached_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_cached)) 3754 #define stat_inc_inline_xattr(inode) \ 3755 do { \ 3756 if (f2fs_has_inline_xattr(inode)) \ 3757 (atomic_inc(&F2FS_I_SB(inode)->inline_xattr)); \ 3758 } while (0) 3759 #define stat_dec_inline_xattr(inode) \ 3760 do { \ 3761 if (f2fs_has_inline_xattr(inode)) \ 3762 (atomic_dec(&F2FS_I_SB(inode)->inline_xattr)); \ 3763 } while (0) 3764 #define stat_inc_inline_inode(inode) \ 3765 do { \ 3766 if (f2fs_has_inline_data(inode)) \ 3767 (atomic_inc(&F2FS_I_SB(inode)->inline_inode)); \ 3768 } while (0) 3769 #define stat_dec_inline_inode(inode) \ 3770 do { \ 3771 if (f2fs_has_inline_data(inode)) \ 3772 (atomic_dec(&F2FS_I_SB(inode)->inline_inode)); \ 3773 } while (0) 3774 #define stat_inc_inline_dir(inode) \ 3775 do { \ 3776 if (f2fs_has_inline_dentry(inode)) \ 3777 (atomic_inc(&F2FS_I_SB(inode)->inline_dir)); \ 3778 } while (0) 3779 #define stat_dec_inline_dir(inode) \ 3780 do { \ 3781 if (f2fs_has_inline_dentry(inode)) \ 3782 (atomic_dec(&F2FS_I_SB(inode)->inline_dir)); \ 3783 } while (0) 3784 #define stat_inc_compr_inode(inode) \ 3785 do { \ 3786 if (f2fs_compressed_file(inode)) \ 3787 (atomic_inc(&F2FS_I_SB(inode)->compr_inode)); \ 3788 } while (0) 3789 #define stat_dec_compr_inode(inode) \ 3790 do { \ 3791 if (f2fs_compressed_file(inode)) \ 3792 (atomic_dec(&F2FS_I_SB(inode)->compr_inode)); \ 3793 } while (0) 3794 #define stat_add_compr_blocks(inode, blocks) \ 3795 (atomic64_add(blocks, &F2FS_I_SB(inode)->compr_blocks)) 3796 #define stat_sub_compr_blocks(inode, blocks) \ 3797 (atomic64_sub(blocks, &F2FS_I_SB(inode)->compr_blocks)) 3798 #define stat_inc_meta_count(sbi, blkaddr) \ 3799 do { \ 3800 if (blkaddr < SIT_I(sbi)->sit_base_addr) \ 3801 atomic_inc(&(sbi)->meta_count[META_CP]); \ 3802 else if (blkaddr < NM_I(sbi)->nat_blkaddr) \ 3803 atomic_inc(&(sbi)->meta_count[META_SIT]); \ 3804 else if (blkaddr < SM_I(sbi)->ssa_blkaddr) \ 3805 atomic_inc(&(sbi)->meta_count[META_NAT]); \ 3806 else if (blkaddr < SM_I(sbi)->main_blkaddr) \ 3807 atomic_inc(&(sbi)->meta_count[META_SSA]); \ 3808 } while (0) 3809 #define stat_inc_seg_type(sbi, curseg) \ 3810 ((sbi)->segment_count[(curseg)->alloc_type]++) 3811 #define stat_inc_block_count(sbi, curseg) \ 3812 ((sbi)->block_count[(curseg)->alloc_type]++) 3813 #define stat_inc_inplace_blocks(sbi) \ 3814 (atomic_inc(&(sbi)->inplace_count)) 3815 #define stat_update_max_atomic_write(inode) \ 3816 do { \ 3817 int cur = F2FS_I_SB(inode)->atomic_files; \ 3818 int max = atomic_read(&F2FS_I_SB(inode)->max_aw_cnt); \ 3819 if (cur > max) \ 3820 atomic_set(&F2FS_I_SB(inode)->max_aw_cnt, cur); \ 3821 } while (0) 3822 #define stat_inc_volatile_write(inode) \ 3823 (atomic_inc(&F2FS_I_SB(inode)->vw_cnt)) 3824 #define stat_dec_volatile_write(inode) \ 3825 (atomic_dec(&F2FS_I_SB(inode)->vw_cnt)) 3826 #define stat_update_max_volatile_write(inode) \ 3827 do { \ 3828 int cur = atomic_read(&F2FS_I_SB(inode)->vw_cnt); \ 3829 int max = atomic_read(&F2FS_I_SB(inode)->max_vw_cnt); \ 3830 if (cur > max) \ 3831 atomic_set(&F2FS_I_SB(inode)->max_vw_cnt, cur); \ 3832 } while (0) 3833 #define stat_inc_seg_count(sbi, type, gc_type) \ 3834 do { \ 3835 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 3836 si->tot_segs++; \ 3837 if ((type) == SUM_TYPE_DATA) { \ 3838 si->data_segs++; \ 3839 si->bg_data_segs += (gc_type == BG_GC) ? 1 : 0; \ 3840 } else { \ 3841 si->node_segs++; \ 3842 si->bg_node_segs += (gc_type == BG_GC) ? 1 : 0; \ 3843 } \ 3844 } while (0) 3845 3846 #define stat_inc_tot_blk_count(si, blks) \ 3847 ((si)->tot_blks += (blks)) 3848 3849 #define stat_inc_data_blk_count(sbi, blks, gc_type) \ 3850 do { \ 3851 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 3852 stat_inc_tot_blk_count(si, blks); \ 3853 si->data_blks += (blks); \ 3854 si->bg_data_blks += ((gc_type) == BG_GC) ? (blks) : 0; \ 3855 } while (0) 3856 3857 #define stat_inc_node_blk_count(sbi, blks, gc_type) \ 3858 do { \ 3859 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 3860 stat_inc_tot_blk_count(si, blks); \ 3861 si->node_blks += (blks); \ 3862 si->bg_node_blks += ((gc_type) == BG_GC) ? (blks) : 0; \ 3863 } while (0) 3864 3865 int f2fs_build_stats(struct f2fs_sb_info *sbi); 3866 void f2fs_destroy_stats(struct f2fs_sb_info *sbi); 3867 void __init f2fs_create_root_stats(void); 3868 void f2fs_destroy_root_stats(void); 3869 void f2fs_update_sit_info(struct f2fs_sb_info *sbi); 3870 #else 3871 #define stat_inc_cp_count(si) do { } while (0) 3872 #define stat_inc_bg_cp_count(si) do { } while (0) 3873 #define stat_inc_call_count(si) do { } while (0) 3874 #define stat_inc_bggc_count(si) do { } while (0) 3875 #define stat_io_skip_bggc_count(sbi) do { } while (0) 3876 #define stat_other_skip_bggc_count(sbi) do { } while (0) 3877 #define stat_inc_dirty_inode(sbi, type) do { } while (0) 3878 #define stat_dec_dirty_inode(sbi, type) do { } while (0) 3879 #define stat_inc_total_hit(sbi) do { } while (0) 3880 #define stat_inc_rbtree_node_hit(sbi) do { } while (0) 3881 #define stat_inc_largest_node_hit(sbi) do { } while (0) 3882 #define stat_inc_cached_node_hit(sbi) do { } while (0) 3883 #define stat_inc_inline_xattr(inode) do { } while (0) 3884 #define stat_dec_inline_xattr(inode) do { } while (0) 3885 #define stat_inc_inline_inode(inode) do { } while (0) 3886 #define stat_dec_inline_inode(inode) do { } while (0) 3887 #define stat_inc_inline_dir(inode) do { } while (0) 3888 #define stat_dec_inline_dir(inode) do { } while (0) 3889 #define stat_inc_compr_inode(inode) do { } while (0) 3890 #define stat_dec_compr_inode(inode) do { } while (0) 3891 #define stat_add_compr_blocks(inode, blocks) do { } while (0) 3892 #define stat_sub_compr_blocks(inode, blocks) do { } while (0) 3893 #define stat_update_max_atomic_write(inode) do { } while (0) 3894 #define stat_inc_volatile_write(inode) do { } while (0) 3895 #define stat_dec_volatile_write(inode) do { } while (0) 3896 #define stat_update_max_volatile_write(inode) do { } while (0) 3897 #define stat_inc_meta_count(sbi, blkaddr) do { } while (0) 3898 #define stat_inc_seg_type(sbi, curseg) do { } while (0) 3899 #define stat_inc_block_count(sbi, curseg) do { } while (0) 3900 #define stat_inc_inplace_blocks(sbi) do { } while (0) 3901 #define stat_inc_seg_count(sbi, type, gc_type) do { } while (0) 3902 #define stat_inc_tot_blk_count(si, blks) do { } while (0) 3903 #define stat_inc_data_blk_count(sbi, blks, gc_type) do { } while (0) 3904 #define stat_inc_node_blk_count(sbi, blks, gc_type) do { } while (0) 3905 3906 static inline int f2fs_build_stats(struct f2fs_sb_info *sbi) { return 0; } 3907 static inline void f2fs_destroy_stats(struct f2fs_sb_info *sbi) { } 3908 static inline void __init f2fs_create_root_stats(void) { } 3909 static inline void f2fs_destroy_root_stats(void) { } 3910 static inline void f2fs_update_sit_info(struct f2fs_sb_info *sbi) {} 3911 #endif 3912 3913 extern const struct file_operations f2fs_dir_operations; 3914 extern const struct file_operations f2fs_file_operations; 3915 extern const struct inode_operations f2fs_file_inode_operations; 3916 extern const struct address_space_operations f2fs_dblock_aops; 3917 extern const struct address_space_operations f2fs_node_aops; 3918 extern const struct address_space_operations f2fs_meta_aops; 3919 extern const struct inode_operations f2fs_dir_inode_operations; 3920 extern const struct inode_operations f2fs_symlink_inode_operations; 3921 extern const struct inode_operations f2fs_encrypted_symlink_inode_operations; 3922 extern const struct inode_operations f2fs_special_inode_operations; 3923 extern struct kmem_cache *f2fs_inode_entry_slab; 3924 3925 /* 3926 * inline.c 3927 */ 3928 bool f2fs_may_inline_data(struct inode *inode); 3929 bool f2fs_may_inline_dentry(struct inode *inode); 3930 void f2fs_do_read_inline_data(struct page *page, struct page *ipage); 3931 void f2fs_truncate_inline_inode(struct inode *inode, 3932 struct page *ipage, u64 from); 3933 int f2fs_read_inline_data(struct inode *inode, struct page *page); 3934 int f2fs_convert_inline_page(struct dnode_of_data *dn, struct page *page); 3935 int f2fs_convert_inline_inode(struct inode *inode); 3936 int f2fs_try_convert_inline_dir(struct inode *dir, struct dentry *dentry); 3937 int f2fs_write_inline_data(struct inode *inode, struct page *page); 3938 int f2fs_recover_inline_data(struct inode *inode, struct page *npage); 3939 struct f2fs_dir_entry *f2fs_find_in_inline_dir(struct inode *dir, 3940 const struct f2fs_filename *fname, 3941 struct page **res_page); 3942 int f2fs_make_empty_inline_dir(struct inode *inode, struct inode *parent, 3943 struct page *ipage); 3944 int f2fs_add_inline_entry(struct inode *dir, const struct f2fs_filename *fname, 3945 struct inode *inode, nid_t ino, umode_t mode); 3946 void f2fs_delete_inline_entry(struct f2fs_dir_entry *dentry, 3947 struct page *page, struct inode *dir, 3948 struct inode *inode); 3949 bool f2fs_empty_inline_dir(struct inode *dir); 3950 int f2fs_read_inline_dir(struct file *file, struct dir_context *ctx, 3951 struct fscrypt_str *fstr); 3952 int f2fs_inline_data_fiemap(struct inode *inode, 3953 struct fiemap_extent_info *fieinfo, 3954 __u64 start, __u64 len); 3955 3956 /* 3957 * shrinker.c 3958 */ 3959 unsigned long f2fs_shrink_count(struct shrinker *shrink, 3960 struct shrink_control *sc); 3961 unsigned long f2fs_shrink_scan(struct shrinker *shrink, 3962 struct shrink_control *sc); 3963 void f2fs_join_shrinker(struct f2fs_sb_info *sbi); 3964 void f2fs_leave_shrinker(struct f2fs_sb_info *sbi); 3965 3966 /* 3967 * extent_cache.c 3968 */ 3969 struct rb_entry *f2fs_lookup_rb_tree(struct rb_root_cached *root, 3970 struct rb_entry *cached_re, unsigned int ofs); 3971 struct rb_node **f2fs_lookup_rb_tree_ext(struct f2fs_sb_info *sbi, 3972 struct rb_root_cached *root, 3973 struct rb_node **parent, 3974 unsigned long long key, bool *left_most); 3975 struct rb_node **f2fs_lookup_rb_tree_for_insert(struct f2fs_sb_info *sbi, 3976 struct rb_root_cached *root, 3977 struct rb_node **parent, 3978 unsigned int ofs, bool *leftmost); 3979 struct rb_entry *f2fs_lookup_rb_tree_ret(struct rb_root_cached *root, 3980 struct rb_entry *cached_re, unsigned int ofs, 3981 struct rb_entry **prev_entry, struct rb_entry **next_entry, 3982 struct rb_node ***insert_p, struct rb_node **insert_parent, 3983 bool force, bool *leftmost); 3984 bool f2fs_check_rb_tree_consistence(struct f2fs_sb_info *sbi, 3985 struct rb_root_cached *root, bool check_key); 3986 unsigned int f2fs_shrink_extent_tree(struct f2fs_sb_info *sbi, int nr_shrink); 3987 void f2fs_init_extent_tree(struct inode *inode, struct page *ipage); 3988 void f2fs_drop_extent_tree(struct inode *inode); 3989 unsigned int f2fs_destroy_extent_node(struct inode *inode); 3990 void f2fs_destroy_extent_tree(struct inode *inode); 3991 bool f2fs_lookup_extent_cache(struct inode *inode, pgoff_t pgofs, 3992 struct extent_info *ei); 3993 void f2fs_update_extent_cache(struct dnode_of_data *dn); 3994 void f2fs_update_extent_cache_range(struct dnode_of_data *dn, 3995 pgoff_t fofs, block_t blkaddr, unsigned int len); 3996 void f2fs_init_extent_cache_info(struct f2fs_sb_info *sbi); 3997 int __init f2fs_create_extent_cache(void); 3998 void f2fs_destroy_extent_cache(void); 3999 4000 /* 4001 * sysfs.c 4002 */ 4003 int __init f2fs_init_sysfs(void); 4004 void f2fs_exit_sysfs(void); 4005 int f2fs_register_sysfs(struct f2fs_sb_info *sbi); 4006 void f2fs_unregister_sysfs(struct f2fs_sb_info *sbi); 4007 4008 /* verity.c */ 4009 extern const struct fsverity_operations f2fs_verityops; 4010 4011 /* 4012 * crypto support 4013 */ 4014 static inline bool f2fs_encrypted_file(struct inode *inode) 4015 { 4016 return IS_ENCRYPTED(inode) && S_ISREG(inode->i_mode); 4017 } 4018 4019 static inline void f2fs_set_encrypted_inode(struct inode *inode) 4020 { 4021 #ifdef CONFIG_FS_ENCRYPTION 4022 file_set_encrypt(inode); 4023 f2fs_set_inode_flags(inode); 4024 #endif 4025 } 4026 4027 /* 4028 * Returns true if the reads of the inode's data need to undergo some 4029 * postprocessing step, like decryption or authenticity verification. 4030 */ 4031 static inline bool f2fs_post_read_required(struct inode *inode) 4032 { 4033 return f2fs_encrypted_file(inode) || fsverity_active(inode) || 4034 f2fs_compressed_file(inode); 4035 } 4036 4037 /* 4038 * compress.c 4039 */ 4040 #ifdef CONFIG_F2FS_FS_COMPRESSION 4041 bool f2fs_is_compressed_page(struct page *page); 4042 struct page *f2fs_compress_control_page(struct page *page); 4043 int f2fs_prepare_compress_overwrite(struct inode *inode, 4044 struct page **pagep, pgoff_t index, void **fsdata); 4045 bool f2fs_compress_write_end(struct inode *inode, void *fsdata, 4046 pgoff_t index, unsigned copied); 4047 int f2fs_truncate_partial_cluster(struct inode *inode, u64 from, bool lock); 4048 void f2fs_compress_write_end_io(struct bio *bio, struct page *page); 4049 bool f2fs_is_compress_backend_ready(struct inode *inode); 4050 int f2fs_init_compress_mempool(void); 4051 void f2fs_destroy_compress_mempool(void); 4052 void f2fs_decompress_cluster(struct decompress_io_ctx *dic); 4053 void f2fs_end_read_compressed_page(struct page *page, bool failed, 4054 block_t blkaddr); 4055 bool f2fs_cluster_is_empty(struct compress_ctx *cc); 4056 bool f2fs_cluster_can_merge_page(struct compress_ctx *cc, pgoff_t index); 4057 void f2fs_compress_ctx_add_page(struct compress_ctx *cc, struct page *page); 4058 int f2fs_write_multi_pages(struct compress_ctx *cc, 4059 int *submitted, 4060 struct writeback_control *wbc, 4061 enum iostat_type io_type); 4062 int f2fs_is_compressed_cluster(struct inode *inode, pgoff_t index); 4063 int f2fs_read_multi_pages(struct compress_ctx *cc, struct bio **bio_ret, 4064 unsigned nr_pages, sector_t *last_block_in_bio, 4065 bool is_readahead, bool for_write); 4066 struct decompress_io_ctx *f2fs_alloc_dic(struct compress_ctx *cc); 4067 void f2fs_decompress_end_io(struct decompress_io_ctx *dic, bool failed); 4068 void f2fs_put_page_dic(struct page *page); 4069 int f2fs_init_compress_ctx(struct compress_ctx *cc); 4070 void f2fs_destroy_compress_ctx(struct compress_ctx *cc, bool reuse); 4071 void f2fs_init_compress_info(struct f2fs_sb_info *sbi); 4072 int f2fs_init_compress_inode(struct f2fs_sb_info *sbi); 4073 void f2fs_destroy_compress_inode(struct f2fs_sb_info *sbi); 4074 int f2fs_init_page_array_cache(struct f2fs_sb_info *sbi); 4075 void f2fs_destroy_page_array_cache(struct f2fs_sb_info *sbi); 4076 int __init f2fs_init_compress_cache(void); 4077 void f2fs_destroy_compress_cache(void); 4078 struct address_space *COMPRESS_MAPPING(struct f2fs_sb_info *sbi); 4079 void f2fs_invalidate_compress_page(struct f2fs_sb_info *sbi, block_t blkaddr); 4080 void f2fs_cache_compressed_page(struct f2fs_sb_info *sbi, struct page *page, 4081 nid_t ino, block_t blkaddr); 4082 bool f2fs_load_compressed_page(struct f2fs_sb_info *sbi, struct page *page, 4083 block_t blkaddr); 4084 void f2fs_invalidate_compress_pages(struct f2fs_sb_info *sbi, nid_t ino); 4085 #define inc_compr_inode_stat(inode) \ 4086 do { \ 4087 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); \ 4088 sbi->compr_new_inode++; \ 4089 } while (0) 4090 #define add_compr_block_stat(inode, blocks) \ 4091 do { \ 4092 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); \ 4093 int diff = F2FS_I(inode)->i_cluster_size - blocks; \ 4094 sbi->compr_written_block += blocks; \ 4095 sbi->compr_saved_block += diff; \ 4096 } while (0) 4097 #else 4098 static inline bool f2fs_is_compressed_page(struct page *page) { return false; } 4099 static inline bool f2fs_is_compress_backend_ready(struct inode *inode) 4100 { 4101 if (!f2fs_compressed_file(inode)) 4102 return true; 4103 /* not support compression */ 4104 return false; 4105 } 4106 static inline struct page *f2fs_compress_control_page(struct page *page) 4107 { 4108 WARN_ON_ONCE(1); 4109 return ERR_PTR(-EINVAL); 4110 } 4111 static inline int f2fs_init_compress_mempool(void) { return 0; } 4112 static inline void f2fs_destroy_compress_mempool(void) { } 4113 static inline void f2fs_decompress_cluster(struct decompress_io_ctx *dic) { } 4114 static inline void f2fs_end_read_compressed_page(struct page *page, 4115 bool failed, block_t blkaddr) 4116 { 4117 WARN_ON_ONCE(1); 4118 } 4119 static inline void f2fs_put_page_dic(struct page *page) 4120 { 4121 WARN_ON_ONCE(1); 4122 } 4123 static inline int f2fs_init_compress_inode(struct f2fs_sb_info *sbi) { return 0; } 4124 static inline void f2fs_destroy_compress_inode(struct f2fs_sb_info *sbi) { } 4125 static inline int f2fs_init_page_array_cache(struct f2fs_sb_info *sbi) { return 0; } 4126 static inline void f2fs_destroy_page_array_cache(struct f2fs_sb_info *sbi) { } 4127 static inline int __init f2fs_init_compress_cache(void) { return 0; } 4128 static inline void f2fs_destroy_compress_cache(void) { } 4129 static inline void f2fs_invalidate_compress_page(struct f2fs_sb_info *sbi, 4130 block_t blkaddr) { } 4131 static inline void f2fs_cache_compressed_page(struct f2fs_sb_info *sbi, 4132 struct page *page, nid_t ino, block_t blkaddr) { } 4133 static inline bool f2fs_load_compressed_page(struct f2fs_sb_info *sbi, 4134 struct page *page, block_t blkaddr) { return false; } 4135 static inline void f2fs_invalidate_compress_pages(struct f2fs_sb_info *sbi, 4136 nid_t ino) { } 4137 #define inc_compr_inode_stat(inode) do { } while (0) 4138 #endif 4139 4140 static inline void set_compress_context(struct inode *inode) 4141 { 4142 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 4143 4144 F2FS_I(inode)->i_compress_algorithm = 4145 F2FS_OPTION(sbi).compress_algorithm; 4146 F2FS_I(inode)->i_log_cluster_size = 4147 F2FS_OPTION(sbi).compress_log_size; 4148 F2FS_I(inode)->i_compress_flag = 4149 F2FS_OPTION(sbi).compress_chksum ? 4150 1 << COMPRESS_CHKSUM : 0; 4151 F2FS_I(inode)->i_cluster_size = 4152 1 << F2FS_I(inode)->i_log_cluster_size; 4153 if (F2FS_I(inode)->i_compress_algorithm == COMPRESS_LZ4 && 4154 F2FS_OPTION(sbi).compress_level) 4155 F2FS_I(inode)->i_compress_flag |= 4156 F2FS_OPTION(sbi).compress_level << 4157 COMPRESS_LEVEL_OFFSET; 4158 F2FS_I(inode)->i_flags |= F2FS_COMPR_FL; 4159 set_inode_flag(inode, FI_COMPRESSED_FILE); 4160 stat_inc_compr_inode(inode); 4161 inc_compr_inode_stat(inode); 4162 f2fs_mark_inode_dirty_sync(inode, true); 4163 } 4164 4165 static inline bool f2fs_disable_compressed_file(struct inode *inode) 4166 { 4167 struct f2fs_inode_info *fi = F2FS_I(inode); 4168 4169 if (!f2fs_compressed_file(inode)) 4170 return true; 4171 if (S_ISREG(inode->i_mode) && 4172 (get_dirty_pages(inode) || atomic_read(&fi->i_compr_blocks))) 4173 return false; 4174 4175 fi->i_flags &= ~F2FS_COMPR_FL; 4176 stat_dec_compr_inode(inode); 4177 clear_inode_flag(inode, FI_COMPRESSED_FILE); 4178 f2fs_mark_inode_dirty_sync(inode, true); 4179 return true; 4180 } 4181 4182 #define F2FS_FEATURE_FUNCS(name, flagname) \ 4183 static inline int f2fs_sb_has_##name(struct f2fs_sb_info *sbi) \ 4184 { \ 4185 return F2FS_HAS_FEATURE(sbi, F2FS_FEATURE_##flagname); \ 4186 } 4187 4188 F2FS_FEATURE_FUNCS(encrypt, ENCRYPT); 4189 F2FS_FEATURE_FUNCS(blkzoned, BLKZONED); 4190 F2FS_FEATURE_FUNCS(extra_attr, EXTRA_ATTR); 4191 F2FS_FEATURE_FUNCS(project_quota, PRJQUOTA); 4192 F2FS_FEATURE_FUNCS(inode_chksum, INODE_CHKSUM); 4193 F2FS_FEATURE_FUNCS(flexible_inline_xattr, FLEXIBLE_INLINE_XATTR); 4194 F2FS_FEATURE_FUNCS(quota_ino, QUOTA_INO); 4195 F2FS_FEATURE_FUNCS(inode_crtime, INODE_CRTIME); 4196 F2FS_FEATURE_FUNCS(lost_found, LOST_FOUND); 4197 F2FS_FEATURE_FUNCS(verity, VERITY); 4198 F2FS_FEATURE_FUNCS(sb_chksum, SB_CHKSUM); 4199 F2FS_FEATURE_FUNCS(casefold, CASEFOLD); 4200 F2FS_FEATURE_FUNCS(compression, COMPRESSION); 4201 F2FS_FEATURE_FUNCS(readonly, RO); 4202 4203 #ifdef CONFIG_BLK_DEV_ZONED 4204 static inline bool f2fs_blkz_is_seq(struct f2fs_sb_info *sbi, int devi, 4205 block_t blkaddr) 4206 { 4207 unsigned int zno = blkaddr >> sbi->log_blocks_per_blkz; 4208 4209 return test_bit(zno, FDEV(devi).blkz_seq); 4210 } 4211 #endif 4212 4213 static inline bool f2fs_hw_should_discard(struct f2fs_sb_info *sbi) 4214 { 4215 return f2fs_sb_has_blkzoned(sbi); 4216 } 4217 4218 static inline bool f2fs_bdev_support_discard(struct block_device *bdev) 4219 { 4220 return blk_queue_discard(bdev_get_queue(bdev)) || 4221 bdev_is_zoned(bdev); 4222 } 4223 4224 static inline bool f2fs_hw_support_discard(struct f2fs_sb_info *sbi) 4225 { 4226 int i; 4227 4228 if (!f2fs_is_multi_device(sbi)) 4229 return f2fs_bdev_support_discard(sbi->sb->s_bdev); 4230 4231 for (i = 0; i < sbi->s_ndevs; i++) 4232 if (f2fs_bdev_support_discard(FDEV(i).bdev)) 4233 return true; 4234 return false; 4235 } 4236 4237 static inline bool f2fs_realtime_discard_enable(struct f2fs_sb_info *sbi) 4238 { 4239 return (test_opt(sbi, DISCARD) && f2fs_hw_support_discard(sbi)) || 4240 f2fs_hw_should_discard(sbi); 4241 } 4242 4243 static inline bool f2fs_hw_is_readonly(struct f2fs_sb_info *sbi) 4244 { 4245 int i; 4246 4247 if (!f2fs_is_multi_device(sbi)) 4248 return bdev_read_only(sbi->sb->s_bdev); 4249 4250 for (i = 0; i < sbi->s_ndevs; i++) 4251 if (bdev_read_only(FDEV(i).bdev)) 4252 return true; 4253 return false; 4254 } 4255 4256 static inline bool f2fs_lfs_mode(struct f2fs_sb_info *sbi) 4257 { 4258 return F2FS_OPTION(sbi).fs_mode == FS_MODE_LFS; 4259 } 4260 4261 static inline bool f2fs_may_compress(struct inode *inode) 4262 { 4263 if (IS_SWAPFILE(inode) || f2fs_is_pinned_file(inode) || 4264 f2fs_is_atomic_file(inode) || 4265 f2fs_is_volatile_file(inode)) 4266 return false; 4267 return S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode); 4268 } 4269 4270 static inline void f2fs_i_compr_blocks_update(struct inode *inode, 4271 u64 blocks, bool add) 4272 { 4273 int diff = F2FS_I(inode)->i_cluster_size - blocks; 4274 struct f2fs_inode_info *fi = F2FS_I(inode); 4275 4276 /* don't update i_compr_blocks if saved blocks were released */ 4277 if (!add && !atomic_read(&fi->i_compr_blocks)) 4278 return; 4279 4280 if (add) { 4281 atomic_add(diff, &fi->i_compr_blocks); 4282 stat_add_compr_blocks(inode, diff); 4283 } else { 4284 atomic_sub(diff, &fi->i_compr_blocks); 4285 stat_sub_compr_blocks(inode, diff); 4286 } 4287 f2fs_mark_inode_dirty_sync(inode, true); 4288 } 4289 4290 static inline int block_unaligned_IO(struct inode *inode, 4291 struct kiocb *iocb, struct iov_iter *iter) 4292 { 4293 unsigned int i_blkbits = READ_ONCE(inode->i_blkbits); 4294 unsigned int blocksize_mask = (1 << i_blkbits) - 1; 4295 loff_t offset = iocb->ki_pos; 4296 unsigned long align = offset | iov_iter_alignment(iter); 4297 4298 return align & blocksize_mask; 4299 } 4300 4301 static inline int allow_outplace_dio(struct inode *inode, 4302 struct kiocb *iocb, struct iov_iter *iter) 4303 { 4304 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 4305 int rw = iov_iter_rw(iter); 4306 4307 return (f2fs_lfs_mode(sbi) && (rw == WRITE) && 4308 !block_unaligned_IO(inode, iocb, iter)); 4309 } 4310 4311 static inline bool f2fs_force_buffered_io(struct inode *inode, 4312 struct kiocb *iocb, struct iov_iter *iter) 4313 { 4314 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 4315 int rw = iov_iter_rw(iter); 4316 4317 if (f2fs_post_read_required(inode)) 4318 return true; 4319 if (f2fs_is_multi_device(sbi)) 4320 return true; 4321 /* 4322 * for blkzoned device, fallback direct IO to buffered IO, so 4323 * all IOs can be serialized by log-structured write. 4324 */ 4325 if (f2fs_sb_has_blkzoned(sbi)) 4326 return true; 4327 if (f2fs_lfs_mode(sbi) && (rw == WRITE)) { 4328 if (block_unaligned_IO(inode, iocb, iter)) 4329 return true; 4330 if (F2FS_IO_ALIGNED(sbi)) 4331 return true; 4332 } 4333 if (is_sbi_flag_set(F2FS_I_SB(inode), SBI_CP_DISABLED)) 4334 return true; 4335 4336 return false; 4337 } 4338 4339 static inline bool f2fs_need_verity(const struct inode *inode, pgoff_t idx) 4340 { 4341 return fsverity_active(inode) && 4342 idx < DIV_ROUND_UP(inode->i_size, PAGE_SIZE); 4343 } 4344 4345 #ifdef CONFIG_F2FS_FAULT_INJECTION 4346 extern void f2fs_build_fault_attr(struct f2fs_sb_info *sbi, unsigned int rate, 4347 unsigned int type); 4348 #else 4349 #define f2fs_build_fault_attr(sbi, rate, type) do { } while (0) 4350 #endif 4351 4352 static inline bool is_journalled_quota(struct f2fs_sb_info *sbi) 4353 { 4354 #ifdef CONFIG_QUOTA 4355 if (f2fs_sb_has_quota_ino(sbi)) 4356 return true; 4357 if (F2FS_OPTION(sbi).s_qf_names[USRQUOTA] || 4358 F2FS_OPTION(sbi).s_qf_names[GRPQUOTA] || 4359 F2FS_OPTION(sbi).s_qf_names[PRJQUOTA]) 4360 return true; 4361 #endif 4362 return false; 4363 } 4364 4365 #define EFSBADCRC EBADMSG /* Bad CRC detected */ 4366 #define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */ 4367 4368 #endif /* _LINUX_F2FS_H */ 4369