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 struct pagevec; 32 33 #ifdef CONFIG_F2FS_CHECK_FS 34 #define f2fs_bug_on(sbi, condition) BUG_ON(condition) 35 #else 36 #define f2fs_bug_on(sbi, condition) \ 37 do { \ 38 if (WARN_ON(condition)) \ 39 set_sbi_flag(sbi, SBI_NEED_FSCK); \ 40 } while (0) 41 #endif 42 43 enum { 44 FAULT_KMALLOC, 45 FAULT_KVMALLOC, 46 FAULT_PAGE_ALLOC, 47 FAULT_PAGE_GET, 48 FAULT_ALLOC_BIO, /* it's obsolete due to bio_alloc() will never fail */ 49 FAULT_ALLOC_NID, 50 FAULT_ORPHAN, 51 FAULT_BLOCK, 52 FAULT_DIR_DEPTH, 53 FAULT_EVICT_INODE, 54 FAULT_TRUNCATE, 55 FAULT_READ_IO, 56 FAULT_CHECKPOINT, 57 FAULT_DISCARD, 58 FAULT_WRITE_IO, 59 FAULT_SLAB_ALLOC, 60 FAULT_DQUOT_INIT, 61 FAULT_LOCK_OP, 62 FAULT_MAX, 63 }; 64 65 #ifdef CONFIG_F2FS_FAULT_INJECTION 66 #define F2FS_ALL_FAULT_TYPE ((1 << FAULT_MAX) - 1) 67 68 struct f2fs_fault_info { 69 atomic_t inject_ops; 70 unsigned int inject_rate; 71 unsigned int inject_type; 72 }; 73 74 extern const char *f2fs_fault_name[FAULT_MAX]; 75 #define IS_FAULT_SET(fi, type) ((fi)->inject_type & (1 << (type))) 76 #endif 77 78 /* 79 * For mount options 80 */ 81 #define F2FS_MOUNT_DISABLE_ROLL_FORWARD 0x00000002 82 #define F2FS_MOUNT_DISCARD 0x00000004 83 #define F2FS_MOUNT_NOHEAP 0x00000008 84 #define F2FS_MOUNT_XATTR_USER 0x00000010 85 #define F2FS_MOUNT_POSIX_ACL 0x00000020 86 #define F2FS_MOUNT_DISABLE_EXT_IDENTIFY 0x00000040 87 #define F2FS_MOUNT_INLINE_XATTR 0x00000080 88 #define F2FS_MOUNT_INLINE_DATA 0x00000100 89 #define F2FS_MOUNT_INLINE_DENTRY 0x00000200 90 #define F2FS_MOUNT_FLUSH_MERGE 0x00000400 91 #define F2FS_MOUNT_NOBARRIER 0x00000800 92 #define F2FS_MOUNT_FASTBOOT 0x00001000 93 #define F2FS_MOUNT_EXTENT_CACHE 0x00002000 94 #define F2FS_MOUNT_DATA_FLUSH 0x00008000 95 #define F2FS_MOUNT_FAULT_INJECTION 0x00010000 96 #define F2FS_MOUNT_USRQUOTA 0x00080000 97 #define F2FS_MOUNT_GRPQUOTA 0x00100000 98 #define F2FS_MOUNT_PRJQUOTA 0x00200000 99 #define F2FS_MOUNT_QUOTA 0x00400000 100 #define F2FS_MOUNT_INLINE_XATTR_SIZE 0x00800000 101 #define F2FS_MOUNT_RESERVE_ROOT 0x01000000 102 #define F2FS_MOUNT_DISABLE_CHECKPOINT 0x02000000 103 #define F2FS_MOUNT_NORECOVERY 0x04000000 104 #define F2FS_MOUNT_ATGC 0x08000000 105 #define F2FS_MOUNT_MERGE_CHECKPOINT 0x10000000 106 #define F2FS_MOUNT_GC_MERGE 0x20000000 107 #define F2FS_MOUNT_COMPRESS_CACHE 0x40000000 108 109 #define F2FS_OPTION(sbi) ((sbi)->mount_opt) 110 #define clear_opt(sbi, option) (F2FS_OPTION(sbi).opt &= ~F2FS_MOUNT_##option) 111 #define set_opt(sbi, option) (F2FS_OPTION(sbi).opt |= F2FS_MOUNT_##option) 112 #define test_opt(sbi, option) (F2FS_OPTION(sbi).opt & F2FS_MOUNT_##option) 113 114 #define ver_after(a, b) (typecheck(unsigned long long, a) && \ 115 typecheck(unsigned long long, b) && \ 116 ((long long)((a) - (b)) > 0)) 117 118 typedef u32 block_t; /* 119 * should not change u32, since it is the on-disk block 120 * address format, __le32. 121 */ 122 typedef u32 nid_t; 123 124 #define COMPRESS_EXT_NUM 16 125 126 /* 127 * An implementation of an rwsem that is explicitly unfair to readers. This 128 * prevents priority inversion when a low-priority reader acquires the read lock 129 * while sleeping on the write lock but the write lock is needed by 130 * higher-priority clients. 131 */ 132 133 struct f2fs_rwsem { 134 struct rw_semaphore internal_rwsem; 135 wait_queue_head_t read_waiters; 136 }; 137 138 struct f2fs_mount_info { 139 unsigned int opt; 140 int write_io_size_bits; /* Write IO size bits */ 141 block_t root_reserved_blocks; /* root reserved blocks */ 142 kuid_t s_resuid; /* reserved blocks for uid */ 143 kgid_t s_resgid; /* reserved blocks for gid */ 144 int active_logs; /* # of active logs */ 145 int inline_xattr_size; /* inline xattr size */ 146 #ifdef CONFIG_F2FS_FAULT_INJECTION 147 struct f2fs_fault_info fault_info; /* For fault injection */ 148 #endif 149 #ifdef CONFIG_QUOTA 150 /* Names of quota files with journalled quota */ 151 char *s_qf_names[MAXQUOTAS]; 152 int s_jquota_fmt; /* Format of quota to use */ 153 #endif 154 /* For which write hints are passed down to block layer */ 155 int whint_mode; 156 int alloc_mode; /* segment allocation policy */ 157 int fsync_mode; /* fsync policy */ 158 int fs_mode; /* fs mode: LFS or ADAPTIVE */ 159 int bggc_mode; /* bggc mode: off, on or sync */ 160 int discard_unit; /* 161 * discard command's offset/size should 162 * be aligned to this unit: block, 163 * segment or section 164 */ 165 struct fscrypt_dummy_policy dummy_enc_policy; /* test dummy encryption */ 166 block_t unusable_cap_perc; /* percentage for cap */ 167 block_t unusable_cap; /* Amount of space allowed to be 168 * unusable when disabling checkpoint 169 */ 170 171 /* For compression */ 172 unsigned char compress_algorithm; /* algorithm type */ 173 unsigned char compress_log_size; /* cluster log size */ 174 unsigned char compress_level; /* compress level */ 175 bool compress_chksum; /* compressed data chksum */ 176 unsigned char compress_ext_cnt; /* extension count */ 177 unsigned char nocompress_ext_cnt; /* nocompress extension count */ 178 int compress_mode; /* compression mode */ 179 unsigned char extensions[COMPRESS_EXT_NUM][F2FS_EXTENSION_LEN]; /* extensions */ 180 unsigned char noextensions[COMPRESS_EXT_NUM][F2FS_EXTENSION_LEN]; /* extensions */ 181 }; 182 183 #define F2FS_FEATURE_ENCRYPT 0x0001 184 #define F2FS_FEATURE_BLKZONED 0x0002 185 #define F2FS_FEATURE_ATOMIC_WRITE 0x0004 186 #define F2FS_FEATURE_EXTRA_ATTR 0x0008 187 #define F2FS_FEATURE_PRJQUOTA 0x0010 188 #define F2FS_FEATURE_INODE_CHKSUM 0x0020 189 #define F2FS_FEATURE_FLEXIBLE_INLINE_XATTR 0x0040 190 #define F2FS_FEATURE_QUOTA_INO 0x0080 191 #define F2FS_FEATURE_INODE_CRTIME 0x0100 192 #define F2FS_FEATURE_LOST_FOUND 0x0200 193 #define F2FS_FEATURE_VERITY 0x0400 194 #define F2FS_FEATURE_SB_CHKSUM 0x0800 195 #define F2FS_FEATURE_CASEFOLD 0x1000 196 #define F2FS_FEATURE_COMPRESSION 0x2000 197 #define F2FS_FEATURE_RO 0x4000 198 199 #define __F2FS_HAS_FEATURE(raw_super, mask) \ 200 ((raw_super->feature & cpu_to_le32(mask)) != 0) 201 #define F2FS_HAS_FEATURE(sbi, mask) __F2FS_HAS_FEATURE(sbi->raw_super, mask) 202 #define F2FS_SET_FEATURE(sbi, mask) \ 203 (sbi->raw_super->feature |= cpu_to_le32(mask)) 204 #define F2FS_CLEAR_FEATURE(sbi, mask) \ 205 (sbi->raw_super->feature &= ~cpu_to_le32(mask)) 206 207 /* 208 * Default values for user and/or group using reserved blocks 209 */ 210 #define F2FS_DEF_RESUID 0 211 #define F2FS_DEF_RESGID 0 212 213 /* 214 * For checkpoint manager 215 */ 216 enum { 217 NAT_BITMAP, 218 SIT_BITMAP 219 }; 220 221 #define CP_UMOUNT 0x00000001 222 #define CP_FASTBOOT 0x00000002 223 #define CP_SYNC 0x00000004 224 #define CP_RECOVERY 0x00000008 225 #define CP_DISCARD 0x00000010 226 #define CP_TRIMMED 0x00000020 227 #define CP_PAUSE 0x00000040 228 #define CP_RESIZE 0x00000080 229 230 #define MAX_DISCARD_BLOCKS(sbi) BLKS_PER_SEC(sbi) 231 #define DEF_MAX_DISCARD_REQUEST 8 /* issue 8 discards per round */ 232 #define DEF_MIN_DISCARD_ISSUE_TIME 50 /* 50 ms, if exists */ 233 #define DEF_MID_DISCARD_ISSUE_TIME 500 /* 500 ms, if device busy */ 234 #define DEF_MAX_DISCARD_ISSUE_TIME 60000 /* 60 s, if no candidates */ 235 #define DEF_DISCARD_URGENT_UTIL 80 /* do more discard over 80% */ 236 #define DEF_CP_INTERVAL 60 /* 60 secs */ 237 #define DEF_IDLE_INTERVAL 5 /* 5 secs */ 238 #define DEF_DISABLE_INTERVAL 5 /* 5 secs */ 239 #define DEF_DISABLE_QUICK_INTERVAL 1 /* 1 secs */ 240 #define DEF_UMOUNT_DISCARD_TIMEOUT 5 /* 5 secs */ 241 242 struct cp_control { 243 int reason; 244 __u64 trim_start; 245 __u64 trim_end; 246 __u64 trim_minlen; 247 }; 248 249 /* 250 * indicate meta/data type 251 */ 252 enum { 253 META_CP, 254 META_NAT, 255 META_SIT, 256 META_SSA, 257 META_MAX, 258 META_POR, 259 DATA_GENERIC, /* check range only */ 260 DATA_GENERIC_ENHANCE, /* strong check on range and segment bitmap */ 261 DATA_GENERIC_ENHANCE_READ, /* 262 * strong check on range and segment 263 * bitmap but no warning due to race 264 * condition of read on truncated area 265 * by extent_cache 266 */ 267 META_GENERIC, 268 }; 269 270 /* for the list of ino */ 271 enum { 272 ORPHAN_INO, /* for orphan ino list */ 273 APPEND_INO, /* for append ino list */ 274 UPDATE_INO, /* for update ino list */ 275 TRANS_DIR_INO, /* for trasactions dir ino list */ 276 FLUSH_INO, /* for multiple device flushing */ 277 MAX_INO_ENTRY, /* max. list */ 278 }; 279 280 struct ino_entry { 281 struct list_head list; /* list head */ 282 nid_t ino; /* inode number */ 283 unsigned int dirty_device; /* dirty device bitmap */ 284 }; 285 286 /* for the list of inodes to be GCed */ 287 struct inode_entry { 288 struct list_head list; /* list head */ 289 struct inode *inode; /* vfs inode pointer */ 290 }; 291 292 struct fsync_node_entry { 293 struct list_head list; /* list head */ 294 struct page *page; /* warm node page pointer */ 295 unsigned int seq_id; /* sequence id */ 296 }; 297 298 struct ckpt_req { 299 struct completion wait; /* completion for checkpoint done */ 300 struct llist_node llnode; /* llist_node to be linked in wait queue */ 301 int ret; /* return code of checkpoint */ 302 ktime_t queue_time; /* request queued time */ 303 }; 304 305 struct ckpt_req_control { 306 struct task_struct *f2fs_issue_ckpt; /* checkpoint task */ 307 int ckpt_thread_ioprio; /* checkpoint merge thread ioprio */ 308 wait_queue_head_t ckpt_wait_queue; /* waiting queue for wake-up */ 309 atomic_t issued_ckpt; /* # of actually issued ckpts */ 310 atomic_t total_ckpt; /* # of total ckpts */ 311 atomic_t queued_ckpt; /* # of queued ckpts */ 312 struct llist_head issue_list; /* list for command issue */ 313 spinlock_t stat_lock; /* lock for below checkpoint time stats */ 314 unsigned int cur_time; /* cur wait time in msec for currently issued checkpoint */ 315 unsigned int peak_time; /* peak wait time in msec until now */ 316 }; 317 318 /* for the bitmap indicate blocks to be discarded */ 319 struct discard_entry { 320 struct list_head list; /* list head */ 321 block_t start_blkaddr; /* start blockaddr of current segment */ 322 unsigned char discard_map[SIT_VBLOCK_MAP_SIZE]; /* segment discard bitmap */ 323 }; 324 325 /* default discard granularity of inner discard thread, unit: block count */ 326 #define DEFAULT_DISCARD_GRANULARITY 16 327 328 /* max discard pend list number */ 329 #define MAX_PLIST_NUM 512 330 #define plist_idx(blk_num) ((blk_num) >= MAX_PLIST_NUM ? \ 331 (MAX_PLIST_NUM - 1) : ((blk_num) - 1)) 332 333 enum { 334 D_PREP, /* initial */ 335 D_PARTIAL, /* partially submitted */ 336 D_SUBMIT, /* all submitted */ 337 D_DONE, /* finished */ 338 }; 339 340 struct discard_info { 341 block_t lstart; /* logical start address */ 342 block_t len; /* length */ 343 block_t start; /* actual start address in dev */ 344 }; 345 346 struct discard_cmd { 347 struct rb_node rb_node; /* rb node located in rb-tree */ 348 union { 349 struct { 350 block_t lstart; /* logical start address */ 351 block_t len; /* length */ 352 block_t start; /* actual start address in dev */ 353 }; 354 struct discard_info di; /* discard info */ 355 356 }; 357 struct list_head list; /* command list */ 358 struct completion wait; /* compleation */ 359 struct block_device *bdev; /* bdev */ 360 unsigned short ref; /* reference count */ 361 unsigned char state; /* state */ 362 unsigned char queued; /* queued discard */ 363 int error; /* bio error */ 364 spinlock_t lock; /* for state/bio_ref updating */ 365 unsigned short bio_ref; /* bio reference count */ 366 }; 367 368 enum { 369 DPOLICY_BG, 370 DPOLICY_FORCE, 371 DPOLICY_FSTRIM, 372 DPOLICY_UMOUNT, 373 MAX_DPOLICY, 374 }; 375 376 struct discard_policy { 377 int type; /* type of discard */ 378 unsigned int min_interval; /* used for candidates exist */ 379 unsigned int mid_interval; /* used for device busy */ 380 unsigned int max_interval; /* used for candidates not exist */ 381 unsigned int max_requests; /* # of discards issued per round */ 382 unsigned int io_aware_gran; /* minimum granularity discard not be aware of I/O */ 383 bool io_aware; /* issue discard in idle time */ 384 bool sync; /* submit discard with REQ_SYNC flag */ 385 bool ordered; /* issue discard by lba order */ 386 bool timeout; /* discard timeout for put_super */ 387 unsigned int granularity; /* discard granularity */ 388 }; 389 390 struct discard_cmd_control { 391 struct task_struct *f2fs_issue_discard; /* discard thread */ 392 struct list_head entry_list; /* 4KB discard entry list */ 393 struct list_head pend_list[MAX_PLIST_NUM];/* store pending entries */ 394 struct list_head wait_list; /* store on-flushing entries */ 395 struct list_head fstrim_list; /* in-flight discard from fstrim */ 396 wait_queue_head_t discard_wait_queue; /* waiting queue for wake-up */ 397 unsigned int discard_wake; /* to wake up discard thread */ 398 struct mutex cmd_lock; 399 unsigned int nr_discards; /* # of discards in the list */ 400 unsigned int max_discards; /* max. discards to be issued */ 401 unsigned int max_discard_request; /* max. discard request per round */ 402 unsigned int min_discard_issue_time; /* min. interval between discard issue */ 403 unsigned int mid_discard_issue_time; /* mid. interval between discard issue */ 404 unsigned int max_discard_issue_time; /* max. interval between discard issue */ 405 unsigned int discard_granularity; /* discard granularity */ 406 unsigned int undiscard_blks; /* # of undiscard blocks */ 407 unsigned int next_pos; /* next discard position */ 408 atomic_t issued_discard; /* # of issued discard */ 409 atomic_t queued_discard; /* # of queued discard */ 410 atomic_t discard_cmd_cnt; /* # of cached cmd count */ 411 struct rb_root_cached root; /* root of discard rb-tree */ 412 bool rbtree_check; /* config for consistence check */ 413 }; 414 415 /* for the list of fsync inodes, used only during recovery */ 416 struct fsync_inode_entry { 417 struct list_head list; /* list head */ 418 struct inode *inode; /* vfs inode pointer */ 419 block_t blkaddr; /* block address locating the last fsync */ 420 block_t last_dentry; /* block address locating the last dentry */ 421 }; 422 423 #define nats_in_cursum(jnl) (le16_to_cpu((jnl)->n_nats)) 424 #define sits_in_cursum(jnl) (le16_to_cpu((jnl)->n_sits)) 425 426 #define nat_in_journal(jnl, i) ((jnl)->nat_j.entries[i].ne) 427 #define nid_in_journal(jnl, i) ((jnl)->nat_j.entries[i].nid) 428 #define sit_in_journal(jnl, i) ((jnl)->sit_j.entries[i].se) 429 #define segno_in_journal(jnl, i) ((jnl)->sit_j.entries[i].segno) 430 431 #define MAX_NAT_JENTRIES(jnl) (NAT_JOURNAL_ENTRIES - nats_in_cursum(jnl)) 432 #define MAX_SIT_JENTRIES(jnl) (SIT_JOURNAL_ENTRIES - sits_in_cursum(jnl)) 433 434 static inline int update_nats_in_cursum(struct f2fs_journal *journal, int i) 435 { 436 int before = nats_in_cursum(journal); 437 438 journal->n_nats = cpu_to_le16(before + i); 439 return before; 440 } 441 442 static inline int update_sits_in_cursum(struct f2fs_journal *journal, int i) 443 { 444 int before = sits_in_cursum(journal); 445 446 journal->n_sits = cpu_to_le16(before + i); 447 return before; 448 } 449 450 static inline bool __has_cursum_space(struct f2fs_journal *journal, 451 int size, int type) 452 { 453 if (type == NAT_JOURNAL) 454 return size <= MAX_NAT_JENTRIES(journal); 455 return size <= MAX_SIT_JENTRIES(journal); 456 } 457 458 /* for inline stuff */ 459 #define DEF_INLINE_RESERVED_SIZE 1 460 static inline int get_extra_isize(struct inode *inode); 461 static inline int get_inline_xattr_addrs(struct inode *inode); 462 #define MAX_INLINE_DATA(inode) (sizeof(__le32) * \ 463 (CUR_ADDRS_PER_INODE(inode) - \ 464 get_inline_xattr_addrs(inode) - \ 465 DEF_INLINE_RESERVED_SIZE)) 466 467 /* for inline dir */ 468 #define NR_INLINE_DENTRY(inode) (MAX_INLINE_DATA(inode) * BITS_PER_BYTE / \ 469 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \ 470 BITS_PER_BYTE + 1)) 471 #define INLINE_DENTRY_BITMAP_SIZE(inode) \ 472 DIV_ROUND_UP(NR_INLINE_DENTRY(inode), BITS_PER_BYTE) 473 #define INLINE_RESERVED_SIZE(inode) (MAX_INLINE_DATA(inode) - \ 474 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \ 475 NR_INLINE_DENTRY(inode) + \ 476 INLINE_DENTRY_BITMAP_SIZE(inode))) 477 478 /* 479 * For INODE and NODE manager 480 */ 481 /* for directory operations */ 482 483 struct f2fs_filename { 484 /* 485 * The filename the user specified. This is NULL for some 486 * filesystem-internal operations, e.g. converting an inline directory 487 * to a non-inline one, or roll-forward recovering an encrypted dentry. 488 */ 489 const struct qstr *usr_fname; 490 491 /* 492 * The on-disk filename. For encrypted directories, this is encrypted. 493 * This may be NULL for lookups in an encrypted dir without the key. 494 */ 495 struct fscrypt_str disk_name; 496 497 /* The dirhash of this filename */ 498 f2fs_hash_t hash; 499 500 #ifdef CONFIG_FS_ENCRYPTION 501 /* 502 * For lookups in encrypted directories: either the buffer backing 503 * disk_name, or a buffer that holds the decoded no-key name. 504 */ 505 struct fscrypt_str crypto_buf; 506 #endif 507 #ifdef CONFIG_UNICODE 508 /* 509 * For casefolded directories: the casefolded name, but it's left NULL 510 * if the original name is not valid Unicode, if the directory is both 511 * casefolded and encrypted and its encryption key is unavailable, or if 512 * the filesystem is doing an internal operation where usr_fname is also 513 * NULL. In all these cases we fall back to treating the name as an 514 * opaque byte sequence. 515 */ 516 struct fscrypt_str cf_name; 517 #endif 518 }; 519 520 struct f2fs_dentry_ptr { 521 struct inode *inode; 522 void *bitmap; 523 struct f2fs_dir_entry *dentry; 524 __u8 (*filename)[F2FS_SLOT_LEN]; 525 int max; 526 int nr_bitmap; 527 }; 528 529 static inline void make_dentry_ptr_block(struct inode *inode, 530 struct f2fs_dentry_ptr *d, struct f2fs_dentry_block *t) 531 { 532 d->inode = inode; 533 d->max = NR_DENTRY_IN_BLOCK; 534 d->nr_bitmap = SIZE_OF_DENTRY_BITMAP; 535 d->bitmap = t->dentry_bitmap; 536 d->dentry = t->dentry; 537 d->filename = t->filename; 538 } 539 540 static inline void make_dentry_ptr_inline(struct inode *inode, 541 struct f2fs_dentry_ptr *d, void *t) 542 { 543 int entry_cnt = NR_INLINE_DENTRY(inode); 544 int bitmap_size = INLINE_DENTRY_BITMAP_SIZE(inode); 545 int reserved_size = INLINE_RESERVED_SIZE(inode); 546 547 d->inode = inode; 548 d->max = entry_cnt; 549 d->nr_bitmap = bitmap_size; 550 d->bitmap = t; 551 d->dentry = t + bitmap_size + reserved_size; 552 d->filename = t + bitmap_size + reserved_size + 553 SIZE_OF_DIR_ENTRY * entry_cnt; 554 } 555 556 /* 557 * XATTR_NODE_OFFSET stores xattrs to one node block per file keeping -1 558 * as its node offset to distinguish from index node blocks. 559 * But some bits are used to mark the node block. 560 */ 561 #define XATTR_NODE_OFFSET ((((unsigned int)-1) << OFFSET_BIT_SHIFT) \ 562 >> OFFSET_BIT_SHIFT) 563 enum { 564 ALLOC_NODE, /* allocate a new node page if needed */ 565 LOOKUP_NODE, /* look up a node without readahead */ 566 LOOKUP_NODE_RA, /* 567 * look up a node with readahead called 568 * by get_data_block. 569 */ 570 }; 571 572 #define DEFAULT_RETRY_IO_COUNT 8 /* maximum retry read IO or flush count */ 573 574 /* congestion wait timeout value, default: 20ms */ 575 #define DEFAULT_IO_TIMEOUT (msecs_to_jiffies(20)) 576 577 /* maximum retry quota flush count */ 578 #define DEFAULT_RETRY_QUOTA_FLUSH_COUNT 8 579 580 #define F2FS_LINK_MAX 0xffffffff /* maximum link count per file */ 581 582 #define MAX_DIR_RA_PAGES 4 /* maximum ra pages of dir */ 583 584 /* dirty segments threshold for triggering CP */ 585 #define DEFAULT_DIRTY_THRESHOLD 4 586 587 /* for in-memory extent cache entry */ 588 #define F2FS_MIN_EXTENT_LEN 64 /* minimum extent length */ 589 590 /* number of extent info in extent cache we try to shrink */ 591 #define EXTENT_CACHE_SHRINK_NUMBER 128 592 593 struct rb_entry { 594 struct rb_node rb_node; /* rb node located in rb-tree */ 595 union { 596 struct { 597 unsigned int ofs; /* start offset of the entry */ 598 unsigned int len; /* length of the entry */ 599 }; 600 unsigned long long key; /* 64-bits key */ 601 } __packed; 602 }; 603 604 struct extent_info { 605 unsigned int fofs; /* start offset in a file */ 606 unsigned int len; /* length of the extent */ 607 u32 blk; /* start block address of the extent */ 608 #ifdef CONFIG_F2FS_FS_COMPRESSION 609 unsigned int c_len; /* physical extent length of compressed blocks */ 610 #endif 611 }; 612 613 struct extent_node { 614 struct rb_node rb_node; /* rb node located in rb-tree */ 615 struct extent_info ei; /* extent info */ 616 struct list_head list; /* node in global extent list of sbi */ 617 struct extent_tree *et; /* extent tree pointer */ 618 }; 619 620 struct extent_tree { 621 nid_t ino; /* inode number */ 622 struct rb_root_cached root; /* root of extent info rb-tree */ 623 struct extent_node *cached_en; /* recently accessed extent node */ 624 struct extent_info largest; /* largested extent info */ 625 struct list_head list; /* to be used by sbi->zombie_list */ 626 rwlock_t lock; /* protect extent info rb-tree */ 627 atomic_t node_cnt; /* # of extent node in rb-tree*/ 628 bool largest_updated; /* largest extent updated */ 629 }; 630 631 /* 632 * This structure is taken from ext4_map_blocks. 633 * 634 * Note that, however, f2fs uses NEW and MAPPED flags for f2fs_map_blocks(). 635 */ 636 #define F2FS_MAP_NEW (1 << BH_New) 637 #define F2FS_MAP_MAPPED (1 << BH_Mapped) 638 #define F2FS_MAP_UNWRITTEN (1 << BH_Unwritten) 639 #define F2FS_MAP_FLAGS (F2FS_MAP_NEW | F2FS_MAP_MAPPED |\ 640 F2FS_MAP_UNWRITTEN) 641 642 struct f2fs_map_blocks { 643 struct block_device *m_bdev; /* for multi-device dio */ 644 block_t m_pblk; 645 block_t m_lblk; 646 unsigned int m_len; 647 unsigned int m_flags; 648 pgoff_t *m_next_pgofs; /* point next possible non-hole pgofs */ 649 pgoff_t *m_next_extent; /* point to next possible extent */ 650 int m_seg_type; 651 bool m_may_create; /* indicate it is from write path */ 652 bool m_multidev_dio; /* indicate it allows multi-device dio */ 653 }; 654 655 /* for flag in get_data_block */ 656 enum { 657 F2FS_GET_BLOCK_DEFAULT, 658 F2FS_GET_BLOCK_FIEMAP, 659 F2FS_GET_BLOCK_BMAP, 660 F2FS_GET_BLOCK_DIO, 661 F2FS_GET_BLOCK_PRE_DIO, 662 F2FS_GET_BLOCK_PRE_AIO, 663 F2FS_GET_BLOCK_PRECACHE, 664 }; 665 666 /* 667 * i_advise uses FADVISE_XXX_BIT. We can add additional hints later. 668 */ 669 #define FADVISE_COLD_BIT 0x01 670 #define FADVISE_LOST_PINO_BIT 0x02 671 #define FADVISE_ENCRYPT_BIT 0x04 672 #define FADVISE_ENC_NAME_BIT 0x08 673 #define FADVISE_KEEP_SIZE_BIT 0x10 674 #define FADVISE_HOT_BIT 0x20 675 #define FADVISE_VERITY_BIT 0x40 676 #define FADVISE_TRUNC_BIT 0x80 677 678 #define FADVISE_MODIFIABLE_BITS (FADVISE_COLD_BIT | FADVISE_HOT_BIT) 679 680 #define file_is_cold(inode) is_file(inode, FADVISE_COLD_BIT) 681 #define file_set_cold(inode) set_file(inode, FADVISE_COLD_BIT) 682 #define file_clear_cold(inode) clear_file(inode, FADVISE_COLD_BIT) 683 684 #define file_wrong_pino(inode) is_file(inode, FADVISE_LOST_PINO_BIT) 685 #define file_lost_pino(inode) set_file(inode, FADVISE_LOST_PINO_BIT) 686 #define file_got_pino(inode) clear_file(inode, FADVISE_LOST_PINO_BIT) 687 688 #define file_is_encrypt(inode) is_file(inode, FADVISE_ENCRYPT_BIT) 689 #define file_set_encrypt(inode) set_file(inode, FADVISE_ENCRYPT_BIT) 690 691 #define file_enc_name(inode) is_file(inode, FADVISE_ENC_NAME_BIT) 692 #define file_set_enc_name(inode) set_file(inode, FADVISE_ENC_NAME_BIT) 693 694 #define file_keep_isize(inode) is_file(inode, FADVISE_KEEP_SIZE_BIT) 695 #define file_set_keep_isize(inode) set_file(inode, FADVISE_KEEP_SIZE_BIT) 696 697 #define file_is_hot(inode) is_file(inode, FADVISE_HOT_BIT) 698 #define file_set_hot(inode) set_file(inode, FADVISE_HOT_BIT) 699 #define file_clear_hot(inode) clear_file(inode, FADVISE_HOT_BIT) 700 701 #define file_is_verity(inode) is_file(inode, FADVISE_VERITY_BIT) 702 #define file_set_verity(inode) set_file(inode, FADVISE_VERITY_BIT) 703 704 #define file_should_truncate(inode) is_file(inode, FADVISE_TRUNC_BIT) 705 #define file_need_truncate(inode) set_file(inode, FADVISE_TRUNC_BIT) 706 #define file_dont_truncate(inode) clear_file(inode, FADVISE_TRUNC_BIT) 707 708 #define DEF_DIR_LEVEL 0 709 710 enum { 711 GC_FAILURE_PIN, 712 GC_FAILURE_ATOMIC, 713 MAX_GC_FAILURE 714 }; 715 716 /* used for f2fs_inode_info->flags */ 717 enum { 718 FI_NEW_INODE, /* indicate newly allocated inode */ 719 FI_DIRTY_INODE, /* indicate inode is dirty or not */ 720 FI_AUTO_RECOVER, /* indicate inode is recoverable */ 721 FI_DIRTY_DIR, /* indicate directory has dirty pages */ 722 FI_INC_LINK, /* need to increment i_nlink */ 723 FI_ACL_MODE, /* indicate acl mode */ 724 FI_NO_ALLOC, /* should not allocate any blocks */ 725 FI_FREE_NID, /* free allocated nide */ 726 FI_NO_EXTENT, /* not to use the extent cache */ 727 FI_INLINE_XATTR, /* used for inline xattr */ 728 FI_INLINE_DATA, /* used for inline data*/ 729 FI_INLINE_DENTRY, /* used for inline dentry */ 730 FI_APPEND_WRITE, /* inode has appended data */ 731 FI_UPDATE_WRITE, /* inode has in-place-update data */ 732 FI_NEED_IPU, /* used for ipu per file */ 733 FI_ATOMIC_FILE, /* indicate atomic file */ 734 FI_ATOMIC_COMMIT, /* indicate the state of atomical committing */ 735 FI_VOLATILE_FILE, /* indicate volatile file */ 736 FI_FIRST_BLOCK_WRITTEN, /* indicate #0 data block was written */ 737 FI_DROP_CACHE, /* drop dirty page cache */ 738 FI_DATA_EXIST, /* indicate data exists */ 739 FI_INLINE_DOTS, /* indicate inline dot dentries */ 740 FI_DO_DEFRAG, /* indicate defragment is running */ 741 FI_DIRTY_FILE, /* indicate regular/symlink has dirty pages */ 742 FI_PREALLOCATED_ALL, /* all blocks for write were preallocated */ 743 FI_HOT_DATA, /* indicate file is hot */ 744 FI_EXTRA_ATTR, /* indicate file has extra attribute */ 745 FI_PROJ_INHERIT, /* indicate file inherits projectid */ 746 FI_PIN_FILE, /* indicate file should not be gced */ 747 FI_ATOMIC_REVOKE_REQUEST, /* request to drop atomic data */ 748 FI_VERITY_IN_PROGRESS, /* building fs-verity Merkle tree */ 749 FI_COMPRESSED_FILE, /* indicate file's data can be compressed */ 750 FI_COMPRESS_CORRUPT, /* indicate compressed cluster is corrupted */ 751 FI_MMAP_FILE, /* indicate file was mmapped */ 752 FI_ENABLE_COMPRESS, /* enable compression in "user" compression mode */ 753 FI_COMPRESS_RELEASED, /* compressed blocks were released */ 754 FI_ALIGNED_WRITE, /* enable aligned write */ 755 FI_MAX, /* max flag, never be used */ 756 }; 757 758 struct f2fs_inode_info { 759 struct inode vfs_inode; /* serve a vfs inode */ 760 unsigned long i_flags; /* keep an inode flags for ioctl */ 761 unsigned char i_advise; /* use to give file attribute hints */ 762 unsigned char i_dir_level; /* use for dentry level for large dir */ 763 unsigned int i_current_depth; /* only for directory depth */ 764 /* for gc failure statistic */ 765 unsigned int i_gc_failures[MAX_GC_FAILURE]; 766 unsigned int i_pino; /* parent inode number */ 767 umode_t i_acl_mode; /* keep file acl mode temporarily */ 768 769 /* Use below internally in f2fs*/ 770 unsigned long flags[BITS_TO_LONGS(FI_MAX)]; /* use to pass per-file flags */ 771 struct f2fs_rwsem i_sem; /* protect fi info */ 772 atomic_t dirty_pages; /* # of dirty pages */ 773 f2fs_hash_t chash; /* hash value of given file name */ 774 unsigned int clevel; /* maximum level of given file name */ 775 struct task_struct *task; /* lookup and create consistency */ 776 struct task_struct *cp_task; /* separate cp/wb IO stats*/ 777 nid_t i_xattr_nid; /* node id that contains xattrs */ 778 loff_t last_disk_size; /* lastly written file size */ 779 spinlock_t i_size_lock; /* protect last_disk_size */ 780 781 #ifdef CONFIG_QUOTA 782 struct dquot *i_dquot[MAXQUOTAS]; 783 784 /* quota space reservation, managed internally by quota code */ 785 qsize_t i_reserved_quota; 786 #endif 787 struct list_head dirty_list; /* dirty list for dirs and files */ 788 struct list_head gdirty_list; /* linked in global dirty list */ 789 struct list_head inmem_ilist; /* list for inmem inodes */ 790 struct list_head inmem_pages; /* inmemory pages managed by f2fs */ 791 struct task_struct *inmem_task; /* store inmemory task */ 792 struct mutex inmem_lock; /* lock for inmemory pages */ 793 struct extent_tree *extent_tree; /* cached extent_tree entry */ 794 795 /* avoid racing between foreground op and gc */ 796 struct f2fs_rwsem i_gc_rwsem[2]; 797 struct f2fs_rwsem i_xattr_sem; /* avoid racing between reading and changing EAs */ 798 799 int i_extra_isize; /* size of extra space located in i_addr */ 800 kprojid_t i_projid; /* id for project quota */ 801 int i_inline_xattr_size; /* inline xattr size */ 802 struct timespec64 i_crtime; /* inode creation time */ 803 struct timespec64 i_disk_time[4];/* inode disk times */ 804 805 /* for file compress */ 806 atomic_t i_compr_blocks; /* # of compressed blocks */ 807 unsigned char i_compress_algorithm; /* algorithm type */ 808 unsigned char i_log_cluster_size; /* log of cluster size */ 809 unsigned char i_compress_level; /* compress level (lz4hc,zstd) */ 810 unsigned short i_compress_flag; /* compress flag */ 811 unsigned int i_cluster_size; /* cluster size */ 812 }; 813 814 static inline void get_extent_info(struct extent_info *ext, 815 struct f2fs_extent *i_ext) 816 { 817 ext->fofs = le32_to_cpu(i_ext->fofs); 818 ext->blk = le32_to_cpu(i_ext->blk); 819 ext->len = le32_to_cpu(i_ext->len); 820 } 821 822 static inline void set_raw_extent(struct extent_info *ext, 823 struct f2fs_extent *i_ext) 824 { 825 i_ext->fofs = cpu_to_le32(ext->fofs); 826 i_ext->blk = cpu_to_le32(ext->blk); 827 i_ext->len = cpu_to_le32(ext->len); 828 } 829 830 static inline void set_extent_info(struct extent_info *ei, unsigned int fofs, 831 u32 blk, unsigned int len) 832 { 833 ei->fofs = fofs; 834 ei->blk = blk; 835 ei->len = len; 836 #ifdef CONFIG_F2FS_FS_COMPRESSION 837 ei->c_len = 0; 838 #endif 839 } 840 841 static inline bool __is_discard_mergeable(struct discard_info *back, 842 struct discard_info *front, unsigned int max_len) 843 { 844 return (back->lstart + back->len == front->lstart) && 845 (back->len + front->len <= max_len); 846 } 847 848 static inline bool __is_discard_back_mergeable(struct discard_info *cur, 849 struct discard_info *back, unsigned int max_len) 850 { 851 return __is_discard_mergeable(back, cur, max_len); 852 } 853 854 static inline bool __is_discard_front_mergeable(struct discard_info *cur, 855 struct discard_info *front, unsigned int max_len) 856 { 857 return __is_discard_mergeable(cur, front, max_len); 858 } 859 860 static inline bool __is_extent_mergeable(struct extent_info *back, 861 struct extent_info *front) 862 { 863 #ifdef CONFIG_F2FS_FS_COMPRESSION 864 if (back->c_len && back->len != back->c_len) 865 return false; 866 if (front->c_len && front->len != front->c_len) 867 return false; 868 #endif 869 return (back->fofs + back->len == front->fofs && 870 back->blk + back->len == front->blk); 871 } 872 873 static inline bool __is_back_mergeable(struct extent_info *cur, 874 struct extent_info *back) 875 { 876 return __is_extent_mergeable(back, cur); 877 } 878 879 static inline bool __is_front_mergeable(struct extent_info *cur, 880 struct extent_info *front) 881 { 882 return __is_extent_mergeable(cur, front); 883 } 884 885 extern void f2fs_mark_inode_dirty_sync(struct inode *inode, bool sync); 886 static inline void __try_update_largest_extent(struct extent_tree *et, 887 struct extent_node *en) 888 { 889 if (en->ei.len > et->largest.len) { 890 et->largest = en->ei; 891 et->largest_updated = true; 892 } 893 } 894 895 /* 896 * For free nid management 897 */ 898 enum nid_state { 899 FREE_NID, /* newly added to free nid list */ 900 PREALLOC_NID, /* it is preallocated */ 901 MAX_NID_STATE, 902 }; 903 904 enum nat_state { 905 TOTAL_NAT, 906 DIRTY_NAT, 907 RECLAIMABLE_NAT, 908 MAX_NAT_STATE, 909 }; 910 911 struct f2fs_nm_info { 912 block_t nat_blkaddr; /* base disk address of NAT */ 913 nid_t max_nid; /* maximum possible node ids */ 914 nid_t available_nids; /* # of available node ids */ 915 nid_t next_scan_nid; /* the next nid to be scanned */ 916 unsigned int ram_thresh; /* control the memory footprint */ 917 unsigned int ra_nid_pages; /* # of nid pages to be readaheaded */ 918 unsigned int dirty_nats_ratio; /* control dirty nats ratio threshold */ 919 920 /* NAT cache management */ 921 struct radix_tree_root nat_root;/* root of the nat entry cache */ 922 struct radix_tree_root nat_set_root;/* root of the nat set cache */ 923 struct f2fs_rwsem nat_tree_lock; /* protect nat entry tree */ 924 struct list_head nat_entries; /* cached nat entry list (clean) */ 925 spinlock_t nat_list_lock; /* protect clean nat entry list */ 926 unsigned int nat_cnt[MAX_NAT_STATE]; /* the # of cached nat entries */ 927 unsigned int nat_blocks; /* # of nat blocks */ 928 929 /* free node ids management */ 930 struct radix_tree_root free_nid_root;/* root of the free_nid cache */ 931 struct list_head free_nid_list; /* list for free nids excluding preallocated nids */ 932 unsigned int nid_cnt[MAX_NID_STATE]; /* the number of free node id */ 933 spinlock_t nid_list_lock; /* protect nid lists ops */ 934 struct mutex build_lock; /* lock for build free nids */ 935 unsigned char **free_nid_bitmap; 936 unsigned char *nat_block_bitmap; 937 unsigned short *free_nid_count; /* free nid count of NAT block */ 938 939 /* for checkpoint */ 940 char *nat_bitmap; /* NAT bitmap pointer */ 941 942 unsigned int nat_bits_blocks; /* # of nat bits blocks */ 943 unsigned char *nat_bits; /* NAT bits blocks */ 944 unsigned char *full_nat_bits; /* full NAT pages */ 945 unsigned char *empty_nat_bits; /* empty NAT pages */ 946 #ifdef CONFIG_F2FS_CHECK_FS 947 char *nat_bitmap_mir; /* NAT bitmap mirror */ 948 #endif 949 int bitmap_size; /* bitmap size */ 950 }; 951 952 /* 953 * this structure is used as one of function parameters. 954 * all the information are dedicated to a given direct node block determined 955 * by the data offset in a file. 956 */ 957 struct dnode_of_data { 958 struct inode *inode; /* vfs inode pointer */ 959 struct page *inode_page; /* its inode page, NULL is possible */ 960 struct page *node_page; /* cached direct node page */ 961 nid_t nid; /* node id of the direct node block */ 962 unsigned int ofs_in_node; /* data offset in the node page */ 963 bool inode_page_locked; /* inode page is locked or not */ 964 bool node_changed; /* is node block changed */ 965 char cur_level; /* level of hole node page */ 966 char max_level; /* level of current page located */ 967 block_t data_blkaddr; /* block address of the node block */ 968 }; 969 970 static inline void set_new_dnode(struct dnode_of_data *dn, struct inode *inode, 971 struct page *ipage, struct page *npage, nid_t nid) 972 { 973 memset(dn, 0, sizeof(*dn)); 974 dn->inode = inode; 975 dn->inode_page = ipage; 976 dn->node_page = npage; 977 dn->nid = nid; 978 } 979 980 /* 981 * For SIT manager 982 * 983 * By default, there are 6 active log areas across the whole main area. 984 * When considering hot and cold data separation to reduce cleaning overhead, 985 * we split 3 for data logs and 3 for node logs as hot, warm, and cold types, 986 * respectively. 987 * In the current design, you should not change the numbers intentionally. 988 * Instead, as a mount option such as active_logs=x, you can use 2, 4, and 6 989 * logs individually according to the underlying devices. (default: 6) 990 * Just in case, on-disk layout covers maximum 16 logs that consist of 8 for 991 * data and 8 for node logs. 992 */ 993 #define NR_CURSEG_DATA_TYPE (3) 994 #define NR_CURSEG_NODE_TYPE (3) 995 #define NR_CURSEG_INMEM_TYPE (2) 996 #define NR_CURSEG_RO_TYPE (2) 997 #define NR_CURSEG_PERSIST_TYPE (NR_CURSEG_DATA_TYPE + NR_CURSEG_NODE_TYPE) 998 #define NR_CURSEG_TYPE (NR_CURSEG_INMEM_TYPE + NR_CURSEG_PERSIST_TYPE) 999 1000 enum { 1001 CURSEG_HOT_DATA = 0, /* directory entry blocks */ 1002 CURSEG_WARM_DATA, /* data blocks */ 1003 CURSEG_COLD_DATA, /* multimedia or GCed data blocks */ 1004 CURSEG_HOT_NODE, /* direct node blocks of directory files */ 1005 CURSEG_WARM_NODE, /* direct node blocks of normal files */ 1006 CURSEG_COLD_NODE, /* indirect node blocks */ 1007 NR_PERSISTENT_LOG, /* number of persistent log */ 1008 CURSEG_COLD_DATA_PINNED = NR_PERSISTENT_LOG, 1009 /* pinned file that needs consecutive block address */ 1010 CURSEG_ALL_DATA_ATGC, /* SSR alloctor in hot/warm/cold data area */ 1011 NO_CHECK_TYPE, /* number of persistent & inmem log */ 1012 }; 1013 1014 struct flush_cmd { 1015 struct completion wait; 1016 struct llist_node llnode; 1017 nid_t ino; 1018 int ret; 1019 }; 1020 1021 struct flush_cmd_control { 1022 struct task_struct *f2fs_issue_flush; /* flush thread */ 1023 wait_queue_head_t flush_wait_queue; /* waiting queue for wake-up */ 1024 atomic_t issued_flush; /* # of issued flushes */ 1025 atomic_t queued_flush; /* # of queued flushes */ 1026 struct llist_head issue_list; /* list for command issue */ 1027 struct llist_node *dispatch_list; /* list for command dispatch */ 1028 }; 1029 1030 struct f2fs_sm_info { 1031 struct sit_info *sit_info; /* whole segment information */ 1032 struct free_segmap_info *free_info; /* free segment information */ 1033 struct dirty_seglist_info *dirty_info; /* dirty segment information */ 1034 struct curseg_info *curseg_array; /* active segment information */ 1035 1036 struct f2fs_rwsem curseg_lock; /* for preventing curseg change */ 1037 1038 block_t seg0_blkaddr; /* block address of 0'th segment */ 1039 block_t main_blkaddr; /* start block address of main area */ 1040 block_t ssa_blkaddr; /* start block address of SSA area */ 1041 1042 unsigned int segment_count; /* total # of segments */ 1043 unsigned int main_segments; /* # of segments in main area */ 1044 unsigned int reserved_segments; /* # of reserved segments */ 1045 unsigned int additional_reserved_segments;/* reserved segs for IO align feature */ 1046 unsigned int ovp_segments; /* # of overprovision segments */ 1047 1048 /* a threshold to reclaim prefree segments */ 1049 unsigned int rec_prefree_segments; 1050 1051 /* for batched trimming */ 1052 unsigned int trim_sections; /* # of sections to trim */ 1053 1054 struct list_head sit_entry_set; /* sit entry set list */ 1055 1056 unsigned int ipu_policy; /* in-place-update policy */ 1057 unsigned int min_ipu_util; /* in-place-update threshold */ 1058 unsigned int min_fsync_blocks; /* threshold for fsync */ 1059 unsigned int min_seq_blocks; /* threshold for sequential blocks */ 1060 unsigned int min_hot_blocks; /* threshold for hot block allocation */ 1061 unsigned int min_ssr_sections; /* threshold to trigger SSR allocation */ 1062 1063 /* for flush command control */ 1064 struct flush_cmd_control *fcc_info; 1065 1066 /* for discard command control */ 1067 struct discard_cmd_control *dcc_info; 1068 }; 1069 1070 /* 1071 * For superblock 1072 */ 1073 /* 1074 * COUNT_TYPE for monitoring 1075 * 1076 * f2fs monitors the number of several block types such as on-writeback, 1077 * dirty dentry blocks, dirty node blocks, and dirty meta blocks. 1078 */ 1079 #define WB_DATA_TYPE(p) (__is_cp_guaranteed(p) ? F2FS_WB_CP_DATA : F2FS_WB_DATA) 1080 enum count_type { 1081 F2FS_DIRTY_DENTS, 1082 F2FS_DIRTY_DATA, 1083 F2FS_DIRTY_QDATA, 1084 F2FS_DIRTY_NODES, 1085 F2FS_DIRTY_META, 1086 F2FS_INMEM_PAGES, 1087 F2FS_DIRTY_IMETA, 1088 F2FS_WB_CP_DATA, 1089 F2FS_WB_DATA, 1090 F2FS_RD_DATA, 1091 F2FS_RD_NODE, 1092 F2FS_RD_META, 1093 F2FS_DIO_WRITE, 1094 F2FS_DIO_READ, 1095 NR_COUNT_TYPE, 1096 }; 1097 1098 /* 1099 * The below are the page types of bios used in submit_bio(). 1100 * The available types are: 1101 * DATA User data pages. It operates as async mode. 1102 * NODE Node pages. It operates as async mode. 1103 * META FS metadata pages such as SIT, NAT, CP. 1104 * NR_PAGE_TYPE The number of page types. 1105 * META_FLUSH Make sure the previous pages are written 1106 * with waiting the bio's completion 1107 * ... Only can be used with META. 1108 */ 1109 #define PAGE_TYPE_OF_BIO(type) ((type) > META ? META : (type)) 1110 enum page_type { 1111 DATA, 1112 NODE, 1113 META, 1114 NR_PAGE_TYPE, 1115 META_FLUSH, 1116 INMEM, /* the below types are used by tracepoints only. */ 1117 INMEM_DROP, 1118 INMEM_INVALIDATE, 1119 INMEM_REVOKE, 1120 IPU, 1121 OPU, 1122 }; 1123 1124 enum temp_type { 1125 HOT = 0, /* must be zero for meta bio */ 1126 WARM, 1127 COLD, 1128 NR_TEMP_TYPE, 1129 }; 1130 1131 enum need_lock_type { 1132 LOCK_REQ = 0, 1133 LOCK_DONE, 1134 LOCK_RETRY, 1135 }; 1136 1137 enum cp_reason_type { 1138 CP_NO_NEEDED, 1139 CP_NON_REGULAR, 1140 CP_COMPRESSED, 1141 CP_HARDLINK, 1142 CP_SB_NEED_CP, 1143 CP_WRONG_PINO, 1144 CP_NO_SPC_ROLL, 1145 CP_NODE_NEED_CP, 1146 CP_FASTBOOT_MODE, 1147 CP_SPEC_LOG_NUM, 1148 CP_RECOVER_DIR, 1149 }; 1150 1151 enum iostat_type { 1152 /* WRITE IO */ 1153 APP_DIRECT_IO, /* app direct write IOs */ 1154 APP_BUFFERED_IO, /* app buffered write IOs */ 1155 APP_WRITE_IO, /* app write IOs */ 1156 APP_MAPPED_IO, /* app mapped IOs */ 1157 FS_DATA_IO, /* data IOs from kworker/fsync/reclaimer */ 1158 FS_NODE_IO, /* node IOs from kworker/fsync/reclaimer */ 1159 FS_META_IO, /* meta IOs from kworker/reclaimer */ 1160 FS_GC_DATA_IO, /* data IOs from forground gc */ 1161 FS_GC_NODE_IO, /* node IOs from forground gc */ 1162 FS_CP_DATA_IO, /* data IOs from checkpoint */ 1163 FS_CP_NODE_IO, /* node IOs from checkpoint */ 1164 FS_CP_META_IO, /* meta IOs from checkpoint */ 1165 1166 /* READ IO */ 1167 APP_DIRECT_READ_IO, /* app direct read IOs */ 1168 APP_BUFFERED_READ_IO, /* app buffered read IOs */ 1169 APP_READ_IO, /* app read IOs */ 1170 APP_MAPPED_READ_IO, /* app mapped read IOs */ 1171 FS_DATA_READ_IO, /* data read IOs */ 1172 FS_GDATA_READ_IO, /* data read IOs from background gc */ 1173 FS_CDATA_READ_IO, /* compressed data read IOs */ 1174 FS_NODE_READ_IO, /* node read IOs */ 1175 FS_META_READ_IO, /* meta read IOs */ 1176 1177 /* other */ 1178 FS_DISCARD, /* discard */ 1179 NR_IO_TYPE, 1180 }; 1181 1182 struct f2fs_io_info { 1183 struct f2fs_sb_info *sbi; /* f2fs_sb_info pointer */ 1184 nid_t ino; /* inode number */ 1185 enum page_type type; /* contains DATA/NODE/META/META_FLUSH */ 1186 enum temp_type temp; /* contains HOT/WARM/COLD */ 1187 int op; /* contains REQ_OP_ */ 1188 int op_flags; /* req_flag_bits */ 1189 block_t new_blkaddr; /* new block address to be written */ 1190 block_t old_blkaddr; /* old block address before Cow */ 1191 struct page *page; /* page to be written */ 1192 struct page *encrypted_page; /* encrypted page */ 1193 struct page *compressed_page; /* compressed page */ 1194 struct list_head list; /* serialize IOs */ 1195 bool submitted; /* indicate IO submission */ 1196 int need_lock; /* indicate we need to lock cp_rwsem */ 1197 bool in_list; /* indicate fio is in io_list */ 1198 bool is_por; /* indicate IO is from recovery or not */ 1199 bool retry; /* need to reallocate block address */ 1200 int compr_blocks; /* # of compressed block addresses */ 1201 bool encrypted; /* indicate file is encrypted */ 1202 enum iostat_type io_type; /* io type */ 1203 struct writeback_control *io_wbc; /* writeback control */ 1204 struct bio **bio; /* bio for ipu */ 1205 sector_t *last_block; /* last block number in bio */ 1206 unsigned char version; /* version of the node */ 1207 }; 1208 1209 struct bio_entry { 1210 struct bio *bio; 1211 struct list_head list; 1212 }; 1213 1214 #define is_read_io(rw) ((rw) == READ) 1215 struct f2fs_bio_info { 1216 struct f2fs_sb_info *sbi; /* f2fs superblock */ 1217 struct bio *bio; /* bios to merge */ 1218 sector_t last_block_in_bio; /* last block number */ 1219 struct f2fs_io_info fio; /* store buffered io info. */ 1220 struct f2fs_rwsem io_rwsem; /* blocking op for bio */ 1221 spinlock_t io_lock; /* serialize DATA/NODE IOs */ 1222 struct list_head io_list; /* track fios */ 1223 struct list_head bio_list; /* bio entry list head */ 1224 struct f2fs_rwsem bio_list_lock; /* lock to protect bio entry list */ 1225 }; 1226 1227 #define FDEV(i) (sbi->devs[i]) 1228 #define RDEV(i) (raw_super->devs[i]) 1229 struct f2fs_dev_info { 1230 struct block_device *bdev; 1231 char path[MAX_PATH_LEN]; 1232 unsigned int total_segments; 1233 block_t start_blk; 1234 block_t end_blk; 1235 #ifdef CONFIG_BLK_DEV_ZONED 1236 unsigned int nr_blkz; /* Total number of zones */ 1237 unsigned long *blkz_seq; /* Bitmap indicating sequential zones */ 1238 block_t *zone_capacity_blocks; /* Array of zone capacity in blks */ 1239 #endif 1240 }; 1241 1242 enum inode_type { 1243 DIR_INODE, /* for dirty dir inode */ 1244 FILE_INODE, /* for dirty regular/symlink inode */ 1245 DIRTY_META, /* for all dirtied inode metadata */ 1246 ATOMIC_FILE, /* for all atomic files */ 1247 NR_INODE_TYPE, 1248 }; 1249 1250 /* for inner inode cache management */ 1251 struct inode_management { 1252 struct radix_tree_root ino_root; /* ino entry array */ 1253 spinlock_t ino_lock; /* for ino entry lock */ 1254 struct list_head ino_list; /* inode list head */ 1255 unsigned long ino_num; /* number of entries */ 1256 }; 1257 1258 /* for GC_AT */ 1259 struct atgc_management { 1260 bool atgc_enabled; /* ATGC is enabled or not */ 1261 struct rb_root_cached root; /* root of victim rb-tree */ 1262 struct list_head victim_list; /* linked with all victim entries */ 1263 unsigned int victim_count; /* victim count in rb-tree */ 1264 unsigned int candidate_ratio; /* candidate ratio */ 1265 unsigned int max_candidate_count; /* max candidate count */ 1266 unsigned int age_weight; /* age weight, vblock_weight = 100 - age_weight */ 1267 unsigned long long age_threshold; /* age threshold */ 1268 }; 1269 1270 /* For s_flag in struct f2fs_sb_info */ 1271 enum { 1272 SBI_IS_DIRTY, /* dirty flag for checkpoint */ 1273 SBI_IS_CLOSE, /* specify unmounting */ 1274 SBI_NEED_FSCK, /* need fsck.f2fs to fix */ 1275 SBI_POR_DOING, /* recovery is doing or not */ 1276 SBI_NEED_SB_WRITE, /* need to recover superblock */ 1277 SBI_NEED_CP, /* need to checkpoint */ 1278 SBI_IS_SHUTDOWN, /* shutdown by ioctl */ 1279 SBI_IS_RECOVERED, /* recovered orphan/data */ 1280 SBI_CP_DISABLED, /* CP was disabled last mount */ 1281 SBI_CP_DISABLED_QUICK, /* CP was disabled quickly */ 1282 SBI_QUOTA_NEED_FLUSH, /* need to flush quota info in CP */ 1283 SBI_QUOTA_SKIP_FLUSH, /* skip flushing quota in current CP */ 1284 SBI_QUOTA_NEED_REPAIR, /* quota file may be corrupted */ 1285 SBI_IS_RESIZEFS, /* resizefs is in process */ 1286 }; 1287 1288 enum { 1289 CP_TIME, 1290 REQ_TIME, 1291 DISCARD_TIME, 1292 GC_TIME, 1293 DISABLE_TIME, 1294 UMOUNT_DISCARD_TIMEOUT, 1295 MAX_TIME, 1296 }; 1297 1298 enum { 1299 GC_NORMAL, 1300 GC_IDLE_CB, 1301 GC_IDLE_GREEDY, 1302 GC_IDLE_AT, 1303 GC_URGENT_HIGH, 1304 GC_URGENT_LOW, 1305 MAX_GC_MODE, 1306 }; 1307 1308 enum { 1309 BGGC_MODE_ON, /* background gc is on */ 1310 BGGC_MODE_OFF, /* background gc is off */ 1311 BGGC_MODE_SYNC, /* 1312 * background gc is on, migrating blocks 1313 * like foreground gc 1314 */ 1315 }; 1316 1317 enum { 1318 FS_MODE_ADAPTIVE, /* use both lfs/ssr allocation */ 1319 FS_MODE_LFS, /* use lfs allocation only */ 1320 FS_MODE_FRAGMENT_SEG, /* segment fragmentation mode */ 1321 FS_MODE_FRAGMENT_BLK, /* block fragmentation mode */ 1322 }; 1323 1324 enum { 1325 WHINT_MODE_OFF, /* not pass down write hints */ 1326 WHINT_MODE_USER, /* try to pass down hints given by users */ 1327 WHINT_MODE_FS, /* pass down hints with F2FS policy */ 1328 }; 1329 1330 enum { 1331 ALLOC_MODE_DEFAULT, /* stay default */ 1332 ALLOC_MODE_REUSE, /* reuse segments as much as possible */ 1333 }; 1334 1335 enum fsync_mode { 1336 FSYNC_MODE_POSIX, /* fsync follows posix semantics */ 1337 FSYNC_MODE_STRICT, /* fsync behaves in line with ext4 */ 1338 FSYNC_MODE_NOBARRIER, /* fsync behaves nobarrier based on posix */ 1339 }; 1340 1341 enum { 1342 COMPR_MODE_FS, /* 1343 * automatically compress compression 1344 * enabled files 1345 */ 1346 COMPR_MODE_USER, /* 1347 * automatical compression is disabled. 1348 * user can control the file compression 1349 * using ioctls 1350 */ 1351 }; 1352 1353 enum { 1354 DISCARD_UNIT_BLOCK, /* basic discard unit is block */ 1355 DISCARD_UNIT_SEGMENT, /* basic discard unit is segment */ 1356 DISCARD_UNIT_SECTION, /* basic discard unit is section */ 1357 }; 1358 1359 static inline int f2fs_test_bit(unsigned int nr, char *addr); 1360 static inline void f2fs_set_bit(unsigned int nr, char *addr); 1361 static inline void f2fs_clear_bit(unsigned int nr, char *addr); 1362 1363 /* 1364 * Layout of f2fs page.private: 1365 * 1366 * Layout A: lowest bit should be 1 1367 * | bit0 = 1 | bit1 | bit2 | ... | bit MAX | private data .... | 1368 * bit 0 PAGE_PRIVATE_NOT_POINTER 1369 * bit 1 PAGE_PRIVATE_ATOMIC_WRITE 1370 * bit 2 PAGE_PRIVATE_DUMMY_WRITE 1371 * bit 3 PAGE_PRIVATE_ONGOING_MIGRATION 1372 * bit 4 PAGE_PRIVATE_INLINE_INODE 1373 * bit 5 PAGE_PRIVATE_REF_RESOURCE 1374 * bit 6- f2fs private data 1375 * 1376 * Layout B: lowest bit should be 0 1377 * page.private is a wrapped pointer. 1378 */ 1379 enum { 1380 PAGE_PRIVATE_NOT_POINTER, /* private contains non-pointer data */ 1381 PAGE_PRIVATE_ATOMIC_WRITE, /* data page from atomic write path */ 1382 PAGE_PRIVATE_DUMMY_WRITE, /* data page for padding aligned IO */ 1383 PAGE_PRIVATE_ONGOING_MIGRATION, /* data page which is on-going migrating */ 1384 PAGE_PRIVATE_INLINE_INODE, /* inode page contains inline data */ 1385 PAGE_PRIVATE_REF_RESOURCE, /* dirty page has referenced resources */ 1386 PAGE_PRIVATE_MAX 1387 }; 1388 1389 #define PAGE_PRIVATE_GET_FUNC(name, flagname) \ 1390 static inline bool page_private_##name(struct page *page) \ 1391 { \ 1392 return PagePrivate(page) && \ 1393 test_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)) && \ 1394 test_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \ 1395 } 1396 1397 #define PAGE_PRIVATE_SET_FUNC(name, flagname) \ 1398 static inline void set_page_private_##name(struct page *page) \ 1399 { \ 1400 if (!PagePrivate(page)) { \ 1401 get_page(page); \ 1402 SetPagePrivate(page); \ 1403 set_page_private(page, 0); \ 1404 } \ 1405 set_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)); \ 1406 set_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \ 1407 } 1408 1409 #define PAGE_PRIVATE_CLEAR_FUNC(name, flagname) \ 1410 static inline void clear_page_private_##name(struct page *page) \ 1411 { \ 1412 clear_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \ 1413 if (page_private(page) == 1 << PAGE_PRIVATE_NOT_POINTER) { \ 1414 set_page_private(page, 0); \ 1415 if (PagePrivate(page)) { \ 1416 ClearPagePrivate(page); \ 1417 put_page(page); \ 1418 }\ 1419 } \ 1420 } 1421 1422 PAGE_PRIVATE_GET_FUNC(nonpointer, NOT_POINTER); 1423 PAGE_PRIVATE_GET_FUNC(reference, REF_RESOURCE); 1424 PAGE_PRIVATE_GET_FUNC(inline, INLINE_INODE); 1425 PAGE_PRIVATE_GET_FUNC(gcing, ONGOING_MIGRATION); 1426 PAGE_PRIVATE_GET_FUNC(atomic, ATOMIC_WRITE); 1427 PAGE_PRIVATE_GET_FUNC(dummy, DUMMY_WRITE); 1428 1429 PAGE_PRIVATE_SET_FUNC(reference, REF_RESOURCE); 1430 PAGE_PRIVATE_SET_FUNC(inline, INLINE_INODE); 1431 PAGE_PRIVATE_SET_FUNC(gcing, ONGOING_MIGRATION); 1432 PAGE_PRIVATE_SET_FUNC(atomic, ATOMIC_WRITE); 1433 PAGE_PRIVATE_SET_FUNC(dummy, DUMMY_WRITE); 1434 1435 PAGE_PRIVATE_CLEAR_FUNC(reference, REF_RESOURCE); 1436 PAGE_PRIVATE_CLEAR_FUNC(inline, INLINE_INODE); 1437 PAGE_PRIVATE_CLEAR_FUNC(gcing, ONGOING_MIGRATION); 1438 PAGE_PRIVATE_CLEAR_FUNC(atomic, ATOMIC_WRITE); 1439 PAGE_PRIVATE_CLEAR_FUNC(dummy, DUMMY_WRITE); 1440 1441 static inline unsigned long get_page_private_data(struct page *page) 1442 { 1443 unsigned long data = page_private(page); 1444 1445 if (!test_bit(PAGE_PRIVATE_NOT_POINTER, &data)) 1446 return 0; 1447 return data >> PAGE_PRIVATE_MAX; 1448 } 1449 1450 static inline void set_page_private_data(struct page *page, unsigned long data) 1451 { 1452 if (!PagePrivate(page)) { 1453 get_page(page); 1454 SetPagePrivate(page); 1455 set_page_private(page, 0); 1456 } 1457 set_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)); 1458 page_private(page) |= data << PAGE_PRIVATE_MAX; 1459 } 1460 1461 static inline void clear_page_private_data(struct page *page) 1462 { 1463 page_private(page) &= (1 << PAGE_PRIVATE_MAX) - 1; 1464 if (page_private(page) == 1 << PAGE_PRIVATE_NOT_POINTER) { 1465 set_page_private(page, 0); 1466 if (PagePrivate(page)) { 1467 ClearPagePrivate(page); 1468 put_page(page); 1469 } 1470 } 1471 } 1472 1473 /* For compression */ 1474 enum compress_algorithm_type { 1475 COMPRESS_LZO, 1476 COMPRESS_LZ4, 1477 COMPRESS_ZSTD, 1478 COMPRESS_LZORLE, 1479 COMPRESS_MAX, 1480 }; 1481 1482 enum compress_flag { 1483 COMPRESS_CHKSUM, 1484 COMPRESS_MAX_FLAG, 1485 }; 1486 1487 #define COMPRESS_WATERMARK 20 1488 #define COMPRESS_PERCENT 20 1489 1490 #define COMPRESS_DATA_RESERVED_SIZE 4 1491 struct compress_data { 1492 __le32 clen; /* compressed data size */ 1493 __le32 chksum; /* compressed data chksum */ 1494 __le32 reserved[COMPRESS_DATA_RESERVED_SIZE]; /* reserved */ 1495 u8 cdata[]; /* compressed data */ 1496 }; 1497 1498 #define COMPRESS_HEADER_SIZE (sizeof(struct compress_data)) 1499 1500 #define F2FS_COMPRESSED_PAGE_MAGIC 0xF5F2C000 1501 1502 #define COMPRESS_LEVEL_OFFSET 8 1503 1504 /* compress context */ 1505 struct compress_ctx { 1506 struct inode *inode; /* inode the context belong to */ 1507 pgoff_t cluster_idx; /* cluster index number */ 1508 unsigned int cluster_size; /* page count in cluster */ 1509 unsigned int log_cluster_size; /* log of cluster size */ 1510 struct page **rpages; /* pages store raw data in cluster */ 1511 unsigned int nr_rpages; /* total page number in rpages */ 1512 struct page **cpages; /* pages store compressed data in cluster */ 1513 unsigned int nr_cpages; /* total page number in cpages */ 1514 unsigned int valid_nr_cpages; /* valid page number in cpages */ 1515 void *rbuf; /* virtual mapped address on rpages */ 1516 struct compress_data *cbuf; /* virtual mapped address on cpages */ 1517 size_t rlen; /* valid data length in rbuf */ 1518 size_t clen; /* valid data length in cbuf */ 1519 void *private; /* payload buffer for specified compression algorithm */ 1520 void *private2; /* extra payload buffer */ 1521 }; 1522 1523 /* compress context for write IO path */ 1524 struct compress_io_ctx { 1525 u32 magic; /* magic number to indicate page is compressed */ 1526 struct inode *inode; /* inode the context belong to */ 1527 struct page **rpages; /* pages store raw data in cluster */ 1528 unsigned int nr_rpages; /* total page number in rpages */ 1529 atomic_t pending_pages; /* in-flight compressed page count */ 1530 }; 1531 1532 /* Context for decompressing one cluster on the read IO path */ 1533 struct decompress_io_ctx { 1534 u32 magic; /* magic number to indicate page is compressed */ 1535 struct inode *inode; /* inode the context belong to */ 1536 pgoff_t cluster_idx; /* cluster index number */ 1537 unsigned int cluster_size; /* page count in cluster */ 1538 unsigned int log_cluster_size; /* log of cluster size */ 1539 struct page **rpages; /* pages store raw data in cluster */ 1540 unsigned int nr_rpages; /* total page number in rpages */ 1541 struct page **cpages; /* pages store compressed data in cluster */ 1542 unsigned int nr_cpages; /* total page number in cpages */ 1543 struct page **tpages; /* temp pages to pad holes in cluster */ 1544 void *rbuf; /* virtual mapped address on rpages */ 1545 struct compress_data *cbuf; /* virtual mapped address on cpages */ 1546 size_t rlen; /* valid data length in rbuf */ 1547 size_t clen; /* valid data length in cbuf */ 1548 1549 /* 1550 * The number of compressed pages remaining to be read in this cluster. 1551 * This is initially nr_cpages. It is decremented by 1 each time a page 1552 * has been read (or failed to be read). When it reaches 0, the cluster 1553 * is decompressed (or an error is reported). 1554 * 1555 * If an error occurs before all the pages have been submitted for I/O, 1556 * then this will never reach 0. In this case the I/O submitter is 1557 * responsible for calling f2fs_decompress_end_io() instead. 1558 */ 1559 atomic_t remaining_pages; 1560 1561 /* 1562 * Number of references to this decompress_io_ctx. 1563 * 1564 * One reference is held for I/O completion. This reference is dropped 1565 * after the pagecache pages are updated and unlocked -- either after 1566 * decompression (and verity if enabled), or after an error. 1567 * 1568 * In addition, each compressed page holds a reference while it is in a 1569 * bio. These references are necessary prevent compressed pages from 1570 * being freed while they are still in a bio. 1571 */ 1572 refcount_t refcnt; 1573 1574 bool failed; /* IO error occurred before decompression? */ 1575 bool need_verity; /* need fs-verity verification after decompression? */ 1576 void *private; /* payload buffer for specified decompression algorithm */ 1577 void *private2; /* extra payload buffer */ 1578 struct work_struct verity_work; /* work to verify the decompressed pages */ 1579 }; 1580 1581 #define NULL_CLUSTER ((unsigned int)(~0)) 1582 #define MIN_COMPRESS_LOG_SIZE 2 1583 #define MAX_COMPRESS_LOG_SIZE 8 1584 #define MAX_COMPRESS_WINDOW_SIZE(log_size) ((PAGE_SIZE) << (log_size)) 1585 1586 struct f2fs_sb_info { 1587 struct super_block *sb; /* pointer to VFS super block */ 1588 struct proc_dir_entry *s_proc; /* proc entry */ 1589 struct f2fs_super_block *raw_super; /* raw super block pointer */ 1590 struct f2fs_rwsem sb_lock; /* lock for raw super block */ 1591 int valid_super_block; /* valid super block no */ 1592 unsigned long s_flag; /* flags for sbi */ 1593 struct mutex writepages; /* mutex for writepages() */ 1594 1595 #ifdef CONFIG_BLK_DEV_ZONED 1596 unsigned int blocks_per_blkz; /* F2FS blocks per zone */ 1597 unsigned int log_blocks_per_blkz; /* log2 F2FS blocks per zone */ 1598 #endif 1599 1600 /* for node-related operations */ 1601 struct f2fs_nm_info *nm_info; /* node manager */ 1602 struct inode *node_inode; /* cache node blocks */ 1603 1604 /* for segment-related operations */ 1605 struct f2fs_sm_info *sm_info; /* segment manager */ 1606 1607 /* for bio operations */ 1608 struct f2fs_bio_info *write_io[NR_PAGE_TYPE]; /* for write bios */ 1609 /* keep migration IO order for LFS mode */ 1610 struct f2fs_rwsem io_order_lock; 1611 mempool_t *write_io_dummy; /* Dummy pages */ 1612 1613 /* for checkpoint */ 1614 struct f2fs_checkpoint *ckpt; /* raw checkpoint pointer */ 1615 int cur_cp_pack; /* remain current cp pack */ 1616 spinlock_t cp_lock; /* for flag in ckpt */ 1617 struct inode *meta_inode; /* cache meta blocks */ 1618 struct f2fs_rwsem cp_global_sem; /* checkpoint procedure lock */ 1619 struct f2fs_rwsem cp_rwsem; /* blocking FS operations */ 1620 struct f2fs_rwsem node_write; /* locking node writes */ 1621 struct f2fs_rwsem node_change; /* locking node change */ 1622 wait_queue_head_t cp_wait; 1623 unsigned long last_time[MAX_TIME]; /* to store time in jiffies */ 1624 long interval_time[MAX_TIME]; /* to store thresholds */ 1625 struct ckpt_req_control cprc_info; /* for checkpoint request control */ 1626 1627 struct inode_management im[MAX_INO_ENTRY]; /* manage inode cache */ 1628 1629 spinlock_t fsync_node_lock; /* for node entry lock */ 1630 struct list_head fsync_node_list; /* node list head */ 1631 unsigned int fsync_seg_id; /* sequence id */ 1632 unsigned int fsync_node_num; /* number of node entries */ 1633 1634 /* for orphan inode, use 0'th array */ 1635 unsigned int max_orphans; /* max orphan inodes */ 1636 1637 /* for inode management */ 1638 struct list_head inode_list[NR_INODE_TYPE]; /* dirty inode list */ 1639 spinlock_t inode_lock[NR_INODE_TYPE]; /* for dirty inode list lock */ 1640 struct mutex flush_lock; /* for flush exclusion */ 1641 1642 /* for extent tree cache */ 1643 struct radix_tree_root extent_tree_root;/* cache extent cache entries */ 1644 struct mutex extent_tree_lock; /* locking extent radix tree */ 1645 struct list_head extent_list; /* lru list for shrinker */ 1646 spinlock_t extent_lock; /* locking extent lru list */ 1647 atomic_t total_ext_tree; /* extent tree count */ 1648 struct list_head zombie_list; /* extent zombie tree list */ 1649 atomic_t total_zombie_tree; /* extent zombie tree count */ 1650 atomic_t total_ext_node; /* extent info count */ 1651 1652 /* basic filesystem units */ 1653 unsigned int log_sectors_per_block; /* log2 sectors per block */ 1654 unsigned int log_blocksize; /* log2 block size */ 1655 unsigned int blocksize; /* block size */ 1656 unsigned int root_ino_num; /* root inode number*/ 1657 unsigned int node_ino_num; /* node inode number*/ 1658 unsigned int meta_ino_num; /* meta inode number*/ 1659 unsigned int log_blocks_per_seg; /* log2 blocks per segment */ 1660 unsigned int blocks_per_seg; /* blocks per segment */ 1661 unsigned int segs_per_sec; /* segments per section */ 1662 unsigned int secs_per_zone; /* sections per zone */ 1663 unsigned int total_sections; /* total section count */ 1664 unsigned int total_node_count; /* total node block count */ 1665 unsigned int total_valid_node_count; /* valid node block count */ 1666 int dir_level; /* directory level */ 1667 int readdir_ra; /* readahead inode in readdir */ 1668 u64 max_io_bytes; /* max io bytes to merge IOs */ 1669 1670 block_t user_block_count; /* # of user blocks */ 1671 block_t total_valid_block_count; /* # of valid blocks */ 1672 block_t discard_blks; /* discard command candidats */ 1673 block_t last_valid_block_count; /* for recovery */ 1674 block_t reserved_blocks; /* configurable reserved blocks */ 1675 block_t current_reserved_blocks; /* current reserved blocks */ 1676 1677 /* Additional tracking for no checkpoint mode */ 1678 block_t unusable_block_count; /* # of blocks saved by last cp */ 1679 1680 unsigned int nquota_files; /* # of quota sysfile */ 1681 struct f2fs_rwsem quota_sem; /* blocking cp for flags */ 1682 1683 /* # of pages, see count_type */ 1684 atomic_t nr_pages[NR_COUNT_TYPE]; 1685 /* # of allocated blocks */ 1686 struct percpu_counter alloc_valid_block_count; 1687 1688 /* writeback control */ 1689 atomic_t wb_sync_req[META]; /* count # of WB_SYNC threads */ 1690 1691 /* valid inode count */ 1692 struct percpu_counter total_valid_inode_count; 1693 1694 struct f2fs_mount_info mount_opt; /* mount options */ 1695 1696 /* for cleaning operations */ 1697 struct f2fs_rwsem gc_lock; /* 1698 * semaphore for GC, avoid 1699 * race between GC and GC or CP 1700 */ 1701 struct f2fs_gc_kthread *gc_thread; /* GC thread */ 1702 struct atgc_management am; /* atgc management */ 1703 unsigned int cur_victim_sec; /* current victim section num */ 1704 unsigned int gc_mode; /* current GC state */ 1705 unsigned int next_victim_seg[2]; /* next segment in victim section */ 1706 spinlock_t gc_urgent_high_lock; 1707 bool gc_urgent_high_limited; /* indicates having limited trial count */ 1708 unsigned int gc_urgent_high_remaining; /* remaining trial count for GC_URGENT_HIGH */ 1709 1710 /* for skip statistic */ 1711 unsigned int atomic_files; /* # of opened atomic file */ 1712 unsigned long long skipped_atomic_files[2]; /* FG_GC and BG_GC */ 1713 unsigned long long skipped_gc_rwsem; /* FG_GC only */ 1714 1715 /* threshold for gc trials on pinned files */ 1716 u64 gc_pin_file_threshold; 1717 struct f2fs_rwsem pin_sem; 1718 1719 /* maximum # of trials to find a victim segment for SSR and GC */ 1720 unsigned int max_victim_search; 1721 /* migration granularity of garbage collection, unit: segment */ 1722 unsigned int migration_granularity; 1723 1724 /* 1725 * for stat information. 1726 * one is for the LFS mode, and the other is for the SSR mode. 1727 */ 1728 #ifdef CONFIG_F2FS_STAT_FS 1729 struct f2fs_stat_info *stat_info; /* FS status information */ 1730 atomic_t meta_count[META_MAX]; /* # of meta blocks */ 1731 unsigned int segment_count[2]; /* # of allocated segments */ 1732 unsigned int block_count[2]; /* # of allocated blocks */ 1733 atomic_t inplace_count; /* # of inplace update */ 1734 atomic64_t total_hit_ext; /* # of lookup extent cache */ 1735 atomic64_t read_hit_rbtree; /* # of hit rbtree extent node */ 1736 atomic64_t read_hit_largest; /* # of hit largest extent node */ 1737 atomic64_t read_hit_cached; /* # of hit cached extent node */ 1738 atomic_t inline_xattr; /* # of inline_xattr inodes */ 1739 atomic_t inline_inode; /* # of inline_data inodes */ 1740 atomic_t inline_dir; /* # of inline_dentry inodes */ 1741 atomic_t compr_inode; /* # of compressed inodes */ 1742 atomic64_t compr_blocks; /* # of compressed blocks */ 1743 atomic_t vw_cnt; /* # of volatile writes */ 1744 atomic_t max_aw_cnt; /* max # of atomic writes */ 1745 atomic_t max_vw_cnt; /* max # of volatile writes */ 1746 unsigned int io_skip_bggc; /* skip background gc for in-flight IO */ 1747 unsigned int other_skip_bggc; /* skip background gc for other reasons */ 1748 unsigned int ndirty_inode[NR_INODE_TYPE]; /* # of dirty inodes */ 1749 #endif 1750 spinlock_t stat_lock; /* lock for stat operations */ 1751 1752 /* to attach REQ_META|REQ_FUA flags */ 1753 unsigned int data_io_flag; 1754 unsigned int node_io_flag; 1755 1756 /* For sysfs suppport */ 1757 struct kobject s_kobj; /* /sys/fs/f2fs/<devname> */ 1758 struct completion s_kobj_unregister; 1759 1760 struct kobject s_stat_kobj; /* /sys/fs/f2fs/<devname>/stat */ 1761 struct completion s_stat_kobj_unregister; 1762 1763 struct kobject s_feature_list_kobj; /* /sys/fs/f2fs/<devname>/feature_list */ 1764 struct completion s_feature_list_kobj_unregister; 1765 1766 /* For shrinker support */ 1767 struct list_head s_list; 1768 struct mutex umount_mutex; 1769 unsigned int shrinker_run_no; 1770 1771 /* For multi devices */ 1772 int s_ndevs; /* number of devices */ 1773 struct f2fs_dev_info *devs; /* for device list */ 1774 unsigned int dirty_device; /* for checkpoint data flush */ 1775 spinlock_t dev_lock; /* protect dirty_device */ 1776 bool aligned_blksize; /* all devices has the same logical blksize */ 1777 1778 /* For write statistics */ 1779 u64 sectors_written_start; 1780 u64 kbytes_written; 1781 1782 /* Reference to checksum algorithm driver via cryptoapi */ 1783 struct crypto_shash *s_chksum_driver; 1784 1785 /* Precomputed FS UUID checksum for seeding other checksums */ 1786 __u32 s_chksum_seed; 1787 1788 struct workqueue_struct *post_read_wq; /* post read workqueue */ 1789 1790 struct kmem_cache *inline_xattr_slab; /* inline xattr entry */ 1791 unsigned int inline_xattr_slab_size; /* default inline xattr slab size */ 1792 1793 /* For reclaimed segs statistics per each GC mode */ 1794 unsigned int gc_segment_mode; /* GC state for reclaimed segments */ 1795 unsigned int gc_reclaimed_segs[MAX_GC_MODE]; /* Reclaimed segs for each mode */ 1796 1797 unsigned long seq_file_ra_mul; /* multiplier for ra_pages of seq. files in fadvise */ 1798 1799 int max_fragment_chunk; /* max chunk size for block fragmentation mode */ 1800 int max_fragment_hole; /* max hole size for block fragmentation mode */ 1801 1802 #ifdef CONFIG_F2FS_FS_COMPRESSION 1803 struct kmem_cache *page_array_slab; /* page array entry */ 1804 unsigned int page_array_slab_size; /* default page array slab size */ 1805 1806 /* For runtime compression statistics */ 1807 u64 compr_written_block; 1808 u64 compr_saved_block; 1809 u32 compr_new_inode; 1810 1811 /* For compressed block cache */ 1812 struct inode *compress_inode; /* cache compressed blocks */ 1813 unsigned int compress_percent; /* cache page percentage */ 1814 unsigned int compress_watermark; /* cache page watermark */ 1815 atomic_t compress_page_hit; /* cache hit count */ 1816 #endif 1817 1818 #ifdef CONFIG_F2FS_IOSTAT 1819 /* For app/fs IO statistics */ 1820 spinlock_t iostat_lock; 1821 unsigned long long rw_iostat[NR_IO_TYPE]; 1822 unsigned long long prev_rw_iostat[NR_IO_TYPE]; 1823 bool iostat_enable; 1824 unsigned long iostat_next_period; 1825 unsigned int iostat_period_ms; 1826 1827 /* For io latency related statistics info in one iostat period */ 1828 spinlock_t iostat_lat_lock; 1829 struct iostat_lat_info *iostat_io_lat; 1830 #endif 1831 }; 1832 1833 #ifdef CONFIG_F2FS_FAULT_INJECTION 1834 #define f2fs_show_injection_info(sbi, type) \ 1835 printk_ratelimited("%sF2FS-fs (%s) : inject %s in %s of %pS\n", \ 1836 KERN_INFO, sbi->sb->s_id, \ 1837 f2fs_fault_name[type], \ 1838 __func__, __builtin_return_address(0)) 1839 static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type) 1840 { 1841 struct f2fs_fault_info *ffi = &F2FS_OPTION(sbi).fault_info; 1842 1843 if (!ffi->inject_rate) 1844 return false; 1845 1846 if (!IS_FAULT_SET(ffi, type)) 1847 return false; 1848 1849 atomic_inc(&ffi->inject_ops); 1850 if (atomic_read(&ffi->inject_ops) >= ffi->inject_rate) { 1851 atomic_set(&ffi->inject_ops, 0); 1852 return true; 1853 } 1854 return false; 1855 } 1856 #else 1857 #define f2fs_show_injection_info(sbi, type) do { } while (0) 1858 static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type) 1859 { 1860 return false; 1861 } 1862 #endif 1863 1864 /* 1865 * Test if the mounted volume is a multi-device volume. 1866 * - For a single regular disk volume, sbi->s_ndevs is 0. 1867 * - For a single zoned disk volume, sbi->s_ndevs is 1. 1868 * - For a multi-device volume, sbi->s_ndevs is always 2 or more. 1869 */ 1870 static inline bool f2fs_is_multi_device(struct f2fs_sb_info *sbi) 1871 { 1872 return sbi->s_ndevs > 1; 1873 } 1874 1875 static inline void f2fs_update_time(struct f2fs_sb_info *sbi, int type) 1876 { 1877 unsigned long now = jiffies; 1878 1879 sbi->last_time[type] = now; 1880 1881 /* DISCARD_TIME and GC_TIME are based on REQ_TIME */ 1882 if (type == REQ_TIME) { 1883 sbi->last_time[DISCARD_TIME] = now; 1884 sbi->last_time[GC_TIME] = now; 1885 } 1886 } 1887 1888 static inline bool f2fs_time_over(struct f2fs_sb_info *sbi, int type) 1889 { 1890 unsigned long interval = sbi->interval_time[type] * HZ; 1891 1892 return time_after(jiffies, sbi->last_time[type] + interval); 1893 } 1894 1895 static inline unsigned int f2fs_time_to_wait(struct f2fs_sb_info *sbi, 1896 int type) 1897 { 1898 unsigned long interval = sbi->interval_time[type] * HZ; 1899 unsigned int wait_ms = 0; 1900 long delta; 1901 1902 delta = (sbi->last_time[type] + interval) - jiffies; 1903 if (delta > 0) 1904 wait_ms = jiffies_to_msecs(delta); 1905 1906 return wait_ms; 1907 } 1908 1909 /* 1910 * Inline functions 1911 */ 1912 static inline u32 __f2fs_crc32(struct f2fs_sb_info *sbi, u32 crc, 1913 const void *address, unsigned int length) 1914 { 1915 struct { 1916 struct shash_desc shash; 1917 char ctx[4]; 1918 } desc; 1919 int err; 1920 1921 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver) != sizeof(desc.ctx)); 1922 1923 desc.shash.tfm = sbi->s_chksum_driver; 1924 *(u32 *)desc.ctx = crc; 1925 1926 err = crypto_shash_update(&desc.shash, address, length); 1927 BUG_ON(err); 1928 1929 return *(u32 *)desc.ctx; 1930 } 1931 1932 static inline u32 f2fs_crc32(struct f2fs_sb_info *sbi, const void *address, 1933 unsigned int length) 1934 { 1935 return __f2fs_crc32(sbi, F2FS_SUPER_MAGIC, address, length); 1936 } 1937 1938 static inline bool f2fs_crc_valid(struct f2fs_sb_info *sbi, __u32 blk_crc, 1939 void *buf, size_t buf_size) 1940 { 1941 return f2fs_crc32(sbi, buf, buf_size) == blk_crc; 1942 } 1943 1944 static inline u32 f2fs_chksum(struct f2fs_sb_info *sbi, u32 crc, 1945 const void *address, unsigned int length) 1946 { 1947 return __f2fs_crc32(sbi, crc, address, length); 1948 } 1949 1950 static inline struct f2fs_inode_info *F2FS_I(struct inode *inode) 1951 { 1952 return container_of(inode, struct f2fs_inode_info, vfs_inode); 1953 } 1954 1955 static inline struct f2fs_sb_info *F2FS_SB(struct super_block *sb) 1956 { 1957 return sb->s_fs_info; 1958 } 1959 1960 static inline struct f2fs_sb_info *F2FS_I_SB(struct inode *inode) 1961 { 1962 return F2FS_SB(inode->i_sb); 1963 } 1964 1965 static inline struct f2fs_sb_info *F2FS_M_SB(struct address_space *mapping) 1966 { 1967 return F2FS_I_SB(mapping->host); 1968 } 1969 1970 static inline struct f2fs_sb_info *F2FS_P_SB(struct page *page) 1971 { 1972 return F2FS_M_SB(page_file_mapping(page)); 1973 } 1974 1975 static inline struct f2fs_super_block *F2FS_RAW_SUPER(struct f2fs_sb_info *sbi) 1976 { 1977 return (struct f2fs_super_block *)(sbi->raw_super); 1978 } 1979 1980 static inline struct f2fs_checkpoint *F2FS_CKPT(struct f2fs_sb_info *sbi) 1981 { 1982 return (struct f2fs_checkpoint *)(sbi->ckpt); 1983 } 1984 1985 static inline struct f2fs_node *F2FS_NODE(struct page *page) 1986 { 1987 return (struct f2fs_node *)page_address(page); 1988 } 1989 1990 static inline struct f2fs_inode *F2FS_INODE(struct page *page) 1991 { 1992 return &((struct f2fs_node *)page_address(page))->i; 1993 } 1994 1995 static inline struct f2fs_nm_info *NM_I(struct f2fs_sb_info *sbi) 1996 { 1997 return (struct f2fs_nm_info *)(sbi->nm_info); 1998 } 1999 2000 static inline struct f2fs_sm_info *SM_I(struct f2fs_sb_info *sbi) 2001 { 2002 return (struct f2fs_sm_info *)(sbi->sm_info); 2003 } 2004 2005 static inline struct sit_info *SIT_I(struct f2fs_sb_info *sbi) 2006 { 2007 return (struct sit_info *)(SM_I(sbi)->sit_info); 2008 } 2009 2010 static inline struct free_segmap_info *FREE_I(struct f2fs_sb_info *sbi) 2011 { 2012 return (struct free_segmap_info *)(SM_I(sbi)->free_info); 2013 } 2014 2015 static inline struct dirty_seglist_info *DIRTY_I(struct f2fs_sb_info *sbi) 2016 { 2017 return (struct dirty_seglist_info *)(SM_I(sbi)->dirty_info); 2018 } 2019 2020 static inline struct address_space *META_MAPPING(struct f2fs_sb_info *sbi) 2021 { 2022 return sbi->meta_inode->i_mapping; 2023 } 2024 2025 static inline struct address_space *NODE_MAPPING(struct f2fs_sb_info *sbi) 2026 { 2027 return sbi->node_inode->i_mapping; 2028 } 2029 2030 static inline bool is_sbi_flag_set(struct f2fs_sb_info *sbi, unsigned int type) 2031 { 2032 return test_bit(type, &sbi->s_flag); 2033 } 2034 2035 static inline void set_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type) 2036 { 2037 set_bit(type, &sbi->s_flag); 2038 } 2039 2040 static inline void clear_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type) 2041 { 2042 clear_bit(type, &sbi->s_flag); 2043 } 2044 2045 static inline unsigned long long cur_cp_version(struct f2fs_checkpoint *cp) 2046 { 2047 return le64_to_cpu(cp->checkpoint_ver); 2048 } 2049 2050 static inline unsigned long f2fs_qf_ino(struct super_block *sb, int type) 2051 { 2052 if (type < F2FS_MAX_QUOTAS) 2053 return le32_to_cpu(F2FS_SB(sb)->raw_super->qf_ino[type]); 2054 return 0; 2055 } 2056 2057 static inline __u64 cur_cp_crc(struct f2fs_checkpoint *cp) 2058 { 2059 size_t crc_offset = le32_to_cpu(cp->checksum_offset); 2060 return le32_to_cpu(*((__le32 *)((unsigned char *)cp + crc_offset))); 2061 } 2062 2063 static inline bool __is_set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 2064 { 2065 unsigned int ckpt_flags = le32_to_cpu(cp->ckpt_flags); 2066 2067 return ckpt_flags & f; 2068 } 2069 2070 static inline bool is_set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 2071 { 2072 return __is_set_ckpt_flags(F2FS_CKPT(sbi), f); 2073 } 2074 2075 static inline void __set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 2076 { 2077 unsigned int ckpt_flags; 2078 2079 ckpt_flags = le32_to_cpu(cp->ckpt_flags); 2080 ckpt_flags |= f; 2081 cp->ckpt_flags = cpu_to_le32(ckpt_flags); 2082 } 2083 2084 static inline void set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 2085 { 2086 unsigned long flags; 2087 2088 spin_lock_irqsave(&sbi->cp_lock, flags); 2089 __set_ckpt_flags(F2FS_CKPT(sbi), f); 2090 spin_unlock_irqrestore(&sbi->cp_lock, flags); 2091 } 2092 2093 static inline void __clear_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 2094 { 2095 unsigned int ckpt_flags; 2096 2097 ckpt_flags = le32_to_cpu(cp->ckpt_flags); 2098 ckpt_flags &= (~f); 2099 cp->ckpt_flags = cpu_to_le32(ckpt_flags); 2100 } 2101 2102 static inline void clear_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 2103 { 2104 unsigned long flags; 2105 2106 spin_lock_irqsave(&sbi->cp_lock, flags); 2107 __clear_ckpt_flags(F2FS_CKPT(sbi), f); 2108 spin_unlock_irqrestore(&sbi->cp_lock, flags); 2109 } 2110 2111 static inline void init_f2fs_rwsem(struct f2fs_rwsem *sem) 2112 { 2113 init_rwsem(&sem->internal_rwsem); 2114 init_waitqueue_head(&sem->read_waiters); 2115 } 2116 2117 static inline int f2fs_rwsem_is_locked(struct f2fs_rwsem *sem) 2118 { 2119 return rwsem_is_locked(&sem->internal_rwsem); 2120 } 2121 2122 static inline int f2fs_rwsem_is_contended(struct f2fs_rwsem *sem) 2123 { 2124 return rwsem_is_contended(&sem->internal_rwsem); 2125 } 2126 2127 static inline void f2fs_down_read(struct f2fs_rwsem *sem) 2128 { 2129 wait_event(sem->read_waiters, down_read_trylock(&sem->internal_rwsem)); 2130 } 2131 2132 static inline int f2fs_down_read_trylock(struct f2fs_rwsem *sem) 2133 { 2134 return down_read_trylock(&sem->internal_rwsem); 2135 } 2136 2137 #ifdef CONFIG_DEBUG_LOCK_ALLOC 2138 static inline void f2fs_down_read_nested(struct f2fs_rwsem *sem, int subclass) 2139 { 2140 down_read_nested(&sem->internal_rwsem, subclass); 2141 } 2142 #else 2143 #define f2fs_down_read_nested(sem, subclass) f2fs_down_read(sem) 2144 #endif 2145 2146 static inline void f2fs_up_read(struct f2fs_rwsem *sem) 2147 { 2148 up_read(&sem->internal_rwsem); 2149 } 2150 2151 static inline void f2fs_down_write(struct f2fs_rwsem *sem) 2152 { 2153 down_write(&sem->internal_rwsem); 2154 } 2155 2156 static inline int f2fs_down_write_trylock(struct f2fs_rwsem *sem) 2157 { 2158 return down_write_trylock(&sem->internal_rwsem); 2159 } 2160 2161 static inline void f2fs_up_write(struct f2fs_rwsem *sem) 2162 { 2163 up_write(&sem->internal_rwsem); 2164 wake_up_all(&sem->read_waiters); 2165 } 2166 2167 static inline void f2fs_lock_op(struct f2fs_sb_info *sbi) 2168 { 2169 f2fs_down_read(&sbi->cp_rwsem); 2170 } 2171 2172 static inline int f2fs_trylock_op(struct f2fs_sb_info *sbi) 2173 { 2174 if (time_to_inject(sbi, FAULT_LOCK_OP)) { 2175 f2fs_show_injection_info(sbi, FAULT_LOCK_OP); 2176 return 0; 2177 } 2178 return f2fs_down_read_trylock(&sbi->cp_rwsem); 2179 } 2180 2181 static inline void f2fs_unlock_op(struct f2fs_sb_info *sbi) 2182 { 2183 f2fs_up_read(&sbi->cp_rwsem); 2184 } 2185 2186 static inline void f2fs_lock_all(struct f2fs_sb_info *sbi) 2187 { 2188 f2fs_down_write(&sbi->cp_rwsem); 2189 } 2190 2191 static inline void f2fs_unlock_all(struct f2fs_sb_info *sbi) 2192 { 2193 f2fs_up_write(&sbi->cp_rwsem); 2194 } 2195 2196 static inline int __get_cp_reason(struct f2fs_sb_info *sbi) 2197 { 2198 int reason = CP_SYNC; 2199 2200 if (test_opt(sbi, FASTBOOT)) 2201 reason = CP_FASTBOOT; 2202 if (is_sbi_flag_set(sbi, SBI_IS_CLOSE)) 2203 reason = CP_UMOUNT; 2204 return reason; 2205 } 2206 2207 static inline bool __remain_node_summaries(int reason) 2208 { 2209 return (reason & (CP_UMOUNT | CP_FASTBOOT)); 2210 } 2211 2212 static inline bool __exist_node_summaries(struct f2fs_sb_info *sbi) 2213 { 2214 return (is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG) || 2215 is_set_ckpt_flags(sbi, CP_FASTBOOT_FLAG)); 2216 } 2217 2218 /* 2219 * Check whether the inode has blocks or not 2220 */ 2221 static inline int F2FS_HAS_BLOCKS(struct inode *inode) 2222 { 2223 block_t xattr_block = F2FS_I(inode)->i_xattr_nid ? 1 : 0; 2224 2225 return (inode->i_blocks >> F2FS_LOG_SECTORS_PER_BLOCK) > xattr_block; 2226 } 2227 2228 static inline bool f2fs_has_xattr_block(unsigned int ofs) 2229 { 2230 return ofs == XATTR_NODE_OFFSET; 2231 } 2232 2233 static inline bool __allow_reserved_blocks(struct f2fs_sb_info *sbi, 2234 struct inode *inode, bool cap) 2235 { 2236 if (!inode) 2237 return true; 2238 if (!test_opt(sbi, RESERVE_ROOT)) 2239 return false; 2240 if (IS_NOQUOTA(inode)) 2241 return true; 2242 if (uid_eq(F2FS_OPTION(sbi).s_resuid, current_fsuid())) 2243 return true; 2244 if (!gid_eq(F2FS_OPTION(sbi).s_resgid, GLOBAL_ROOT_GID) && 2245 in_group_p(F2FS_OPTION(sbi).s_resgid)) 2246 return true; 2247 if (cap && capable(CAP_SYS_RESOURCE)) 2248 return true; 2249 return false; 2250 } 2251 2252 static inline void f2fs_i_blocks_write(struct inode *, block_t, bool, bool); 2253 static inline int inc_valid_block_count(struct f2fs_sb_info *sbi, 2254 struct inode *inode, blkcnt_t *count) 2255 { 2256 blkcnt_t diff = 0, release = 0; 2257 block_t avail_user_block_count; 2258 int ret; 2259 2260 ret = dquot_reserve_block(inode, *count); 2261 if (ret) 2262 return ret; 2263 2264 if (time_to_inject(sbi, FAULT_BLOCK)) { 2265 f2fs_show_injection_info(sbi, FAULT_BLOCK); 2266 release = *count; 2267 goto release_quota; 2268 } 2269 2270 /* 2271 * let's increase this in prior to actual block count change in order 2272 * for f2fs_sync_file to avoid data races when deciding checkpoint. 2273 */ 2274 percpu_counter_add(&sbi->alloc_valid_block_count, (*count)); 2275 2276 spin_lock(&sbi->stat_lock); 2277 sbi->total_valid_block_count += (block_t)(*count); 2278 avail_user_block_count = sbi->user_block_count - 2279 sbi->current_reserved_blocks; 2280 2281 if (!__allow_reserved_blocks(sbi, inode, true)) 2282 avail_user_block_count -= F2FS_OPTION(sbi).root_reserved_blocks; 2283 2284 if (F2FS_IO_ALIGNED(sbi)) 2285 avail_user_block_count -= sbi->blocks_per_seg * 2286 SM_I(sbi)->additional_reserved_segments; 2287 2288 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) { 2289 if (avail_user_block_count > sbi->unusable_block_count) 2290 avail_user_block_count -= sbi->unusable_block_count; 2291 else 2292 avail_user_block_count = 0; 2293 } 2294 if (unlikely(sbi->total_valid_block_count > avail_user_block_count)) { 2295 diff = sbi->total_valid_block_count - avail_user_block_count; 2296 if (diff > *count) 2297 diff = *count; 2298 *count -= diff; 2299 release = diff; 2300 sbi->total_valid_block_count -= diff; 2301 if (!*count) { 2302 spin_unlock(&sbi->stat_lock); 2303 goto enospc; 2304 } 2305 } 2306 spin_unlock(&sbi->stat_lock); 2307 2308 if (unlikely(release)) { 2309 percpu_counter_sub(&sbi->alloc_valid_block_count, release); 2310 dquot_release_reservation_block(inode, release); 2311 } 2312 f2fs_i_blocks_write(inode, *count, true, true); 2313 return 0; 2314 2315 enospc: 2316 percpu_counter_sub(&sbi->alloc_valid_block_count, release); 2317 release_quota: 2318 dquot_release_reservation_block(inode, release); 2319 return -ENOSPC; 2320 } 2321 2322 __printf(2, 3) 2323 void f2fs_printk(struct f2fs_sb_info *sbi, const char *fmt, ...); 2324 2325 #define f2fs_err(sbi, fmt, ...) \ 2326 f2fs_printk(sbi, KERN_ERR fmt, ##__VA_ARGS__) 2327 #define f2fs_warn(sbi, fmt, ...) \ 2328 f2fs_printk(sbi, KERN_WARNING fmt, ##__VA_ARGS__) 2329 #define f2fs_notice(sbi, fmt, ...) \ 2330 f2fs_printk(sbi, KERN_NOTICE fmt, ##__VA_ARGS__) 2331 #define f2fs_info(sbi, fmt, ...) \ 2332 f2fs_printk(sbi, KERN_INFO fmt, ##__VA_ARGS__) 2333 #define f2fs_debug(sbi, fmt, ...) \ 2334 f2fs_printk(sbi, KERN_DEBUG fmt, ##__VA_ARGS__) 2335 2336 static inline void dec_valid_block_count(struct f2fs_sb_info *sbi, 2337 struct inode *inode, 2338 block_t count) 2339 { 2340 blkcnt_t sectors = count << F2FS_LOG_SECTORS_PER_BLOCK; 2341 2342 spin_lock(&sbi->stat_lock); 2343 f2fs_bug_on(sbi, sbi->total_valid_block_count < (block_t) count); 2344 sbi->total_valid_block_count -= (block_t)count; 2345 if (sbi->reserved_blocks && 2346 sbi->current_reserved_blocks < sbi->reserved_blocks) 2347 sbi->current_reserved_blocks = min(sbi->reserved_blocks, 2348 sbi->current_reserved_blocks + count); 2349 spin_unlock(&sbi->stat_lock); 2350 if (unlikely(inode->i_blocks < sectors)) { 2351 f2fs_warn(sbi, "Inconsistent i_blocks, ino:%lu, iblocks:%llu, sectors:%llu", 2352 inode->i_ino, 2353 (unsigned long long)inode->i_blocks, 2354 (unsigned long long)sectors); 2355 set_sbi_flag(sbi, SBI_NEED_FSCK); 2356 return; 2357 } 2358 f2fs_i_blocks_write(inode, count, false, true); 2359 } 2360 2361 static inline void inc_page_count(struct f2fs_sb_info *sbi, int count_type) 2362 { 2363 atomic_inc(&sbi->nr_pages[count_type]); 2364 2365 if (count_type == F2FS_DIRTY_DENTS || 2366 count_type == F2FS_DIRTY_NODES || 2367 count_type == F2FS_DIRTY_META || 2368 count_type == F2FS_DIRTY_QDATA || 2369 count_type == F2FS_DIRTY_IMETA) 2370 set_sbi_flag(sbi, SBI_IS_DIRTY); 2371 } 2372 2373 static inline void inode_inc_dirty_pages(struct inode *inode) 2374 { 2375 atomic_inc(&F2FS_I(inode)->dirty_pages); 2376 inc_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ? 2377 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA); 2378 if (IS_NOQUOTA(inode)) 2379 inc_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA); 2380 } 2381 2382 static inline void dec_page_count(struct f2fs_sb_info *sbi, int count_type) 2383 { 2384 atomic_dec(&sbi->nr_pages[count_type]); 2385 } 2386 2387 static inline void inode_dec_dirty_pages(struct inode *inode) 2388 { 2389 if (!S_ISDIR(inode->i_mode) && !S_ISREG(inode->i_mode) && 2390 !S_ISLNK(inode->i_mode)) 2391 return; 2392 2393 atomic_dec(&F2FS_I(inode)->dirty_pages); 2394 dec_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ? 2395 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA); 2396 if (IS_NOQUOTA(inode)) 2397 dec_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA); 2398 } 2399 2400 static inline s64 get_pages(struct f2fs_sb_info *sbi, int count_type) 2401 { 2402 return atomic_read(&sbi->nr_pages[count_type]); 2403 } 2404 2405 static inline int get_dirty_pages(struct inode *inode) 2406 { 2407 return atomic_read(&F2FS_I(inode)->dirty_pages); 2408 } 2409 2410 static inline int get_blocktype_secs(struct f2fs_sb_info *sbi, int block_type) 2411 { 2412 unsigned int pages_per_sec = sbi->segs_per_sec * sbi->blocks_per_seg; 2413 unsigned int segs = (get_pages(sbi, block_type) + pages_per_sec - 1) >> 2414 sbi->log_blocks_per_seg; 2415 2416 return segs / sbi->segs_per_sec; 2417 } 2418 2419 static inline block_t valid_user_blocks(struct f2fs_sb_info *sbi) 2420 { 2421 return sbi->total_valid_block_count; 2422 } 2423 2424 static inline block_t discard_blocks(struct f2fs_sb_info *sbi) 2425 { 2426 return sbi->discard_blks; 2427 } 2428 2429 static inline unsigned long __bitmap_size(struct f2fs_sb_info *sbi, int flag) 2430 { 2431 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi); 2432 2433 /* return NAT or SIT bitmap */ 2434 if (flag == NAT_BITMAP) 2435 return le32_to_cpu(ckpt->nat_ver_bitmap_bytesize); 2436 else if (flag == SIT_BITMAP) 2437 return le32_to_cpu(ckpt->sit_ver_bitmap_bytesize); 2438 2439 return 0; 2440 } 2441 2442 static inline block_t __cp_payload(struct f2fs_sb_info *sbi) 2443 { 2444 return le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_payload); 2445 } 2446 2447 static inline void *__bitmap_ptr(struct f2fs_sb_info *sbi, int flag) 2448 { 2449 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi); 2450 void *tmp_ptr = &ckpt->sit_nat_version_bitmap; 2451 int offset; 2452 2453 if (is_set_ckpt_flags(sbi, CP_LARGE_NAT_BITMAP_FLAG)) { 2454 offset = (flag == SIT_BITMAP) ? 2455 le32_to_cpu(ckpt->nat_ver_bitmap_bytesize) : 0; 2456 /* 2457 * if large_nat_bitmap feature is enabled, leave checksum 2458 * protection for all nat/sit bitmaps. 2459 */ 2460 return tmp_ptr + offset + sizeof(__le32); 2461 } 2462 2463 if (__cp_payload(sbi) > 0) { 2464 if (flag == NAT_BITMAP) 2465 return &ckpt->sit_nat_version_bitmap; 2466 else 2467 return (unsigned char *)ckpt + F2FS_BLKSIZE; 2468 } else { 2469 offset = (flag == NAT_BITMAP) ? 2470 le32_to_cpu(ckpt->sit_ver_bitmap_bytesize) : 0; 2471 return tmp_ptr + offset; 2472 } 2473 } 2474 2475 static inline block_t __start_cp_addr(struct f2fs_sb_info *sbi) 2476 { 2477 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr); 2478 2479 if (sbi->cur_cp_pack == 2) 2480 start_addr += sbi->blocks_per_seg; 2481 return start_addr; 2482 } 2483 2484 static inline block_t __start_cp_next_addr(struct f2fs_sb_info *sbi) 2485 { 2486 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr); 2487 2488 if (sbi->cur_cp_pack == 1) 2489 start_addr += sbi->blocks_per_seg; 2490 return start_addr; 2491 } 2492 2493 static inline void __set_cp_next_pack(struct f2fs_sb_info *sbi) 2494 { 2495 sbi->cur_cp_pack = (sbi->cur_cp_pack == 1) ? 2 : 1; 2496 } 2497 2498 static inline block_t __start_sum_addr(struct f2fs_sb_info *sbi) 2499 { 2500 return le32_to_cpu(F2FS_CKPT(sbi)->cp_pack_start_sum); 2501 } 2502 2503 static inline int inc_valid_node_count(struct f2fs_sb_info *sbi, 2504 struct inode *inode, bool is_inode) 2505 { 2506 block_t valid_block_count; 2507 unsigned int valid_node_count, user_block_count; 2508 int err; 2509 2510 if (is_inode) { 2511 if (inode) { 2512 err = dquot_alloc_inode(inode); 2513 if (err) 2514 return err; 2515 } 2516 } else { 2517 err = dquot_reserve_block(inode, 1); 2518 if (err) 2519 return err; 2520 } 2521 2522 if (time_to_inject(sbi, FAULT_BLOCK)) { 2523 f2fs_show_injection_info(sbi, FAULT_BLOCK); 2524 goto enospc; 2525 } 2526 2527 spin_lock(&sbi->stat_lock); 2528 2529 valid_block_count = sbi->total_valid_block_count + 2530 sbi->current_reserved_blocks + 1; 2531 2532 if (!__allow_reserved_blocks(sbi, inode, false)) 2533 valid_block_count += F2FS_OPTION(sbi).root_reserved_blocks; 2534 2535 if (F2FS_IO_ALIGNED(sbi)) 2536 valid_block_count += sbi->blocks_per_seg * 2537 SM_I(sbi)->additional_reserved_segments; 2538 2539 user_block_count = sbi->user_block_count; 2540 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) 2541 user_block_count -= sbi->unusable_block_count; 2542 2543 if (unlikely(valid_block_count > user_block_count)) { 2544 spin_unlock(&sbi->stat_lock); 2545 goto enospc; 2546 } 2547 2548 valid_node_count = sbi->total_valid_node_count + 1; 2549 if (unlikely(valid_node_count > sbi->total_node_count)) { 2550 spin_unlock(&sbi->stat_lock); 2551 goto enospc; 2552 } 2553 2554 sbi->total_valid_node_count++; 2555 sbi->total_valid_block_count++; 2556 spin_unlock(&sbi->stat_lock); 2557 2558 if (inode) { 2559 if (is_inode) 2560 f2fs_mark_inode_dirty_sync(inode, true); 2561 else 2562 f2fs_i_blocks_write(inode, 1, true, true); 2563 } 2564 2565 percpu_counter_inc(&sbi->alloc_valid_block_count); 2566 return 0; 2567 2568 enospc: 2569 if (is_inode) { 2570 if (inode) 2571 dquot_free_inode(inode); 2572 } else { 2573 dquot_release_reservation_block(inode, 1); 2574 } 2575 return -ENOSPC; 2576 } 2577 2578 static inline void dec_valid_node_count(struct f2fs_sb_info *sbi, 2579 struct inode *inode, bool is_inode) 2580 { 2581 spin_lock(&sbi->stat_lock); 2582 2583 f2fs_bug_on(sbi, !sbi->total_valid_block_count); 2584 f2fs_bug_on(sbi, !sbi->total_valid_node_count); 2585 2586 sbi->total_valid_node_count--; 2587 sbi->total_valid_block_count--; 2588 if (sbi->reserved_blocks && 2589 sbi->current_reserved_blocks < sbi->reserved_blocks) 2590 sbi->current_reserved_blocks++; 2591 2592 spin_unlock(&sbi->stat_lock); 2593 2594 if (is_inode) { 2595 dquot_free_inode(inode); 2596 } else { 2597 if (unlikely(inode->i_blocks == 0)) { 2598 f2fs_warn(sbi, "dec_valid_node_count: inconsistent i_blocks, ino:%lu, iblocks:%llu", 2599 inode->i_ino, 2600 (unsigned long long)inode->i_blocks); 2601 set_sbi_flag(sbi, SBI_NEED_FSCK); 2602 return; 2603 } 2604 f2fs_i_blocks_write(inode, 1, false, true); 2605 } 2606 } 2607 2608 static inline unsigned int valid_node_count(struct f2fs_sb_info *sbi) 2609 { 2610 return sbi->total_valid_node_count; 2611 } 2612 2613 static inline void inc_valid_inode_count(struct f2fs_sb_info *sbi) 2614 { 2615 percpu_counter_inc(&sbi->total_valid_inode_count); 2616 } 2617 2618 static inline void dec_valid_inode_count(struct f2fs_sb_info *sbi) 2619 { 2620 percpu_counter_dec(&sbi->total_valid_inode_count); 2621 } 2622 2623 static inline s64 valid_inode_count(struct f2fs_sb_info *sbi) 2624 { 2625 return percpu_counter_sum_positive(&sbi->total_valid_inode_count); 2626 } 2627 2628 static inline struct page *f2fs_grab_cache_page(struct address_space *mapping, 2629 pgoff_t index, bool for_write) 2630 { 2631 struct page *page; 2632 2633 if (IS_ENABLED(CONFIG_F2FS_FAULT_INJECTION)) { 2634 if (!for_write) 2635 page = find_get_page_flags(mapping, index, 2636 FGP_LOCK | FGP_ACCESSED); 2637 else 2638 page = find_lock_page(mapping, index); 2639 if (page) 2640 return page; 2641 2642 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_ALLOC)) { 2643 f2fs_show_injection_info(F2FS_M_SB(mapping), 2644 FAULT_PAGE_ALLOC); 2645 return NULL; 2646 } 2647 } 2648 2649 if (!for_write) 2650 return grab_cache_page(mapping, index); 2651 return grab_cache_page_write_begin(mapping, index, AOP_FLAG_NOFS); 2652 } 2653 2654 static inline struct page *f2fs_pagecache_get_page( 2655 struct address_space *mapping, pgoff_t index, 2656 int fgp_flags, gfp_t gfp_mask) 2657 { 2658 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_GET)) { 2659 f2fs_show_injection_info(F2FS_M_SB(mapping), FAULT_PAGE_GET); 2660 return NULL; 2661 } 2662 2663 return pagecache_get_page(mapping, index, fgp_flags, gfp_mask); 2664 } 2665 2666 static inline void f2fs_copy_page(struct page *src, struct page *dst) 2667 { 2668 char *src_kaddr = kmap(src); 2669 char *dst_kaddr = kmap(dst); 2670 2671 memcpy(dst_kaddr, src_kaddr, PAGE_SIZE); 2672 kunmap(dst); 2673 kunmap(src); 2674 } 2675 2676 static inline void f2fs_put_page(struct page *page, int unlock) 2677 { 2678 if (!page) 2679 return; 2680 2681 if (unlock) { 2682 f2fs_bug_on(F2FS_P_SB(page), !PageLocked(page)); 2683 unlock_page(page); 2684 } 2685 put_page(page); 2686 } 2687 2688 static inline void f2fs_put_dnode(struct dnode_of_data *dn) 2689 { 2690 if (dn->node_page) 2691 f2fs_put_page(dn->node_page, 1); 2692 if (dn->inode_page && dn->node_page != dn->inode_page) 2693 f2fs_put_page(dn->inode_page, 0); 2694 dn->node_page = NULL; 2695 dn->inode_page = NULL; 2696 } 2697 2698 static inline struct kmem_cache *f2fs_kmem_cache_create(const char *name, 2699 size_t size) 2700 { 2701 return kmem_cache_create(name, size, 0, SLAB_RECLAIM_ACCOUNT, NULL); 2702 } 2703 2704 static inline void *f2fs_kmem_cache_alloc_nofail(struct kmem_cache *cachep, 2705 gfp_t flags) 2706 { 2707 void *entry; 2708 2709 entry = kmem_cache_alloc(cachep, flags); 2710 if (!entry) 2711 entry = kmem_cache_alloc(cachep, flags | __GFP_NOFAIL); 2712 return entry; 2713 } 2714 2715 static inline void *f2fs_kmem_cache_alloc(struct kmem_cache *cachep, 2716 gfp_t flags, bool nofail, struct f2fs_sb_info *sbi) 2717 { 2718 if (nofail) 2719 return f2fs_kmem_cache_alloc_nofail(cachep, flags); 2720 2721 if (time_to_inject(sbi, FAULT_SLAB_ALLOC)) { 2722 f2fs_show_injection_info(sbi, FAULT_SLAB_ALLOC); 2723 return NULL; 2724 } 2725 2726 return kmem_cache_alloc(cachep, flags); 2727 } 2728 2729 static inline bool is_inflight_io(struct f2fs_sb_info *sbi, int type) 2730 { 2731 if (get_pages(sbi, F2FS_RD_DATA) || get_pages(sbi, F2FS_RD_NODE) || 2732 get_pages(sbi, F2FS_RD_META) || get_pages(sbi, F2FS_WB_DATA) || 2733 get_pages(sbi, F2FS_WB_CP_DATA) || 2734 get_pages(sbi, F2FS_DIO_READ) || 2735 get_pages(sbi, F2FS_DIO_WRITE)) 2736 return true; 2737 2738 if (type != DISCARD_TIME && SM_I(sbi) && SM_I(sbi)->dcc_info && 2739 atomic_read(&SM_I(sbi)->dcc_info->queued_discard)) 2740 return true; 2741 2742 if (SM_I(sbi) && SM_I(sbi)->fcc_info && 2743 atomic_read(&SM_I(sbi)->fcc_info->queued_flush)) 2744 return true; 2745 return false; 2746 } 2747 2748 static inline bool is_idle(struct f2fs_sb_info *sbi, int type) 2749 { 2750 if (sbi->gc_mode == GC_URGENT_HIGH) 2751 return true; 2752 2753 if (is_inflight_io(sbi, type)) 2754 return false; 2755 2756 if (sbi->gc_mode == GC_URGENT_LOW && 2757 (type == DISCARD_TIME || type == GC_TIME)) 2758 return true; 2759 2760 return f2fs_time_over(sbi, type); 2761 } 2762 2763 static inline void f2fs_radix_tree_insert(struct radix_tree_root *root, 2764 unsigned long index, void *item) 2765 { 2766 while (radix_tree_insert(root, index, item)) 2767 cond_resched(); 2768 } 2769 2770 #define RAW_IS_INODE(p) ((p)->footer.nid == (p)->footer.ino) 2771 2772 static inline bool IS_INODE(struct page *page) 2773 { 2774 struct f2fs_node *p = F2FS_NODE(page); 2775 2776 return RAW_IS_INODE(p); 2777 } 2778 2779 static inline int offset_in_addr(struct f2fs_inode *i) 2780 { 2781 return (i->i_inline & F2FS_EXTRA_ATTR) ? 2782 (le16_to_cpu(i->i_extra_isize) / sizeof(__le32)) : 0; 2783 } 2784 2785 static inline __le32 *blkaddr_in_node(struct f2fs_node *node) 2786 { 2787 return RAW_IS_INODE(node) ? node->i.i_addr : node->dn.addr; 2788 } 2789 2790 static inline int f2fs_has_extra_attr(struct inode *inode); 2791 static inline block_t data_blkaddr(struct inode *inode, 2792 struct page *node_page, unsigned int offset) 2793 { 2794 struct f2fs_node *raw_node; 2795 __le32 *addr_array; 2796 int base = 0; 2797 bool is_inode = IS_INODE(node_page); 2798 2799 raw_node = F2FS_NODE(node_page); 2800 2801 if (is_inode) { 2802 if (!inode) 2803 /* from GC path only */ 2804 base = offset_in_addr(&raw_node->i); 2805 else if (f2fs_has_extra_attr(inode)) 2806 base = get_extra_isize(inode); 2807 } 2808 2809 addr_array = blkaddr_in_node(raw_node); 2810 return le32_to_cpu(addr_array[base + offset]); 2811 } 2812 2813 static inline block_t f2fs_data_blkaddr(struct dnode_of_data *dn) 2814 { 2815 return data_blkaddr(dn->inode, dn->node_page, dn->ofs_in_node); 2816 } 2817 2818 static inline int f2fs_test_bit(unsigned int nr, char *addr) 2819 { 2820 int mask; 2821 2822 addr += (nr >> 3); 2823 mask = 1 << (7 - (nr & 0x07)); 2824 return mask & *addr; 2825 } 2826 2827 static inline void f2fs_set_bit(unsigned int nr, char *addr) 2828 { 2829 int mask; 2830 2831 addr += (nr >> 3); 2832 mask = 1 << (7 - (nr & 0x07)); 2833 *addr |= mask; 2834 } 2835 2836 static inline void f2fs_clear_bit(unsigned int nr, char *addr) 2837 { 2838 int mask; 2839 2840 addr += (nr >> 3); 2841 mask = 1 << (7 - (nr & 0x07)); 2842 *addr &= ~mask; 2843 } 2844 2845 static inline int f2fs_test_and_set_bit(unsigned int nr, char *addr) 2846 { 2847 int mask; 2848 int ret; 2849 2850 addr += (nr >> 3); 2851 mask = 1 << (7 - (nr & 0x07)); 2852 ret = mask & *addr; 2853 *addr |= mask; 2854 return ret; 2855 } 2856 2857 static inline int f2fs_test_and_clear_bit(unsigned int nr, char *addr) 2858 { 2859 int mask; 2860 int ret; 2861 2862 addr += (nr >> 3); 2863 mask = 1 << (7 - (nr & 0x07)); 2864 ret = mask & *addr; 2865 *addr &= ~mask; 2866 return ret; 2867 } 2868 2869 static inline void f2fs_change_bit(unsigned int nr, char *addr) 2870 { 2871 int mask; 2872 2873 addr += (nr >> 3); 2874 mask = 1 << (7 - (nr & 0x07)); 2875 *addr ^= mask; 2876 } 2877 2878 /* 2879 * On-disk inode flags (f2fs_inode::i_flags) 2880 */ 2881 #define F2FS_COMPR_FL 0x00000004 /* Compress file */ 2882 #define F2FS_SYNC_FL 0x00000008 /* Synchronous updates */ 2883 #define F2FS_IMMUTABLE_FL 0x00000010 /* Immutable file */ 2884 #define F2FS_APPEND_FL 0x00000020 /* writes to file may only append */ 2885 #define F2FS_NODUMP_FL 0x00000040 /* do not dump file */ 2886 #define F2FS_NOATIME_FL 0x00000080 /* do not update atime */ 2887 #define F2FS_NOCOMP_FL 0x00000400 /* Don't compress */ 2888 #define F2FS_INDEX_FL 0x00001000 /* hash-indexed directory */ 2889 #define F2FS_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */ 2890 #define F2FS_PROJINHERIT_FL 0x20000000 /* Create with parents projid */ 2891 #define F2FS_CASEFOLD_FL 0x40000000 /* Casefolded file */ 2892 2893 /* Flags that should be inherited by new inodes from their parent. */ 2894 #define F2FS_FL_INHERITED (F2FS_SYNC_FL | F2FS_NODUMP_FL | F2FS_NOATIME_FL | \ 2895 F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \ 2896 F2FS_CASEFOLD_FL | F2FS_COMPR_FL | F2FS_NOCOMP_FL) 2897 2898 /* Flags that are appropriate for regular files (all but dir-specific ones). */ 2899 #define F2FS_REG_FLMASK (~(F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \ 2900 F2FS_CASEFOLD_FL)) 2901 2902 /* Flags that are appropriate for non-directories/regular files. */ 2903 #define F2FS_OTHER_FLMASK (F2FS_NODUMP_FL | F2FS_NOATIME_FL) 2904 2905 static inline __u32 f2fs_mask_flags(umode_t mode, __u32 flags) 2906 { 2907 if (S_ISDIR(mode)) 2908 return flags; 2909 else if (S_ISREG(mode)) 2910 return flags & F2FS_REG_FLMASK; 2911 else 2912 return flags & F2FS_OTHER_FLMASK; 2913 } 2914 2915 static inline void __mark_inode_dirty_flag(struct inode *inode, 2916 int flag, bool set) 2917 { 2918 switch (flag) { 2919 case FI_INLINE_XATTR: 2920 case FI_INLINE_DATA: 2921 case FI_INLINE_DENTRY: 2922 case FI_NEW_INODE: 2923 if (set) 2924 return; 2925 fallthrough; 2926 case FI_DATA_EXIST: 2927 case FI_INLINE_DOTS: 2928 case FI_PIN_FILE: 2929 case FI_COMPRESS_RELEASED: 2930 f2fs_mark_inode_dirty_sync(inode, true); 2931 } 2932 } 2933 2934 static inline void set_inode_flag(struct inode *inode, int flag) 2935 { 2936 set_bit(flag, F2FS_I(inode)->flags); 2937 __mark_inode_dirty_flag(inode, flag, true); 2938 } 2939 2940 static inline int is_inode_flag_set(struct inode *inode, int flag) 2941 { 2942 return test_bit(flag, F2FS_I(inode)->flags); 2943 } 2944 2945 static inline void clear_inode_flag(struct inode *inode, int flag) 2946 { 2947 clear_bit(flag, F2FS_I(inode)->flags); 2948 __mark_inode_dirty_flag(inode, flag, false); 2949 } 2950 2951 static inline bool f2fs_verity_in_progress(struct inode *inode) 2952 { 2953 return IS_ENABLED(CONFIG_FS_VERITY) && 2954 is_inode_flag_set(inode, FI_VERITY_IN_PROGRESS); 2955 } 2956 2957 static inline void set_acl_inode(struct inode *inode, umode_t mode) 2958 { 2959 F2FS_I(inode)->i_acl_mode = mode; 2960 set_inode_flag(inode, FI_ACL_MODE); 2961 f2fs_mark_inode_dirty_sync(inode, false); 2962 } 2963 2964 static inline void f2fs_i_links_write(struct inode *inode, bool inc) 2965 { 2966 if (inc) 2967 inc_nlink(inode); 2968 else 2969 drop_nlink(inode); 2970 f2fs_mark_inode_dirty_sync(inode, true); 2971 } 2972 2973 static inline void f2fs_i_blocks_write(struct inode *inode, 2974 block_t diff, bool add, bool claim) 2975 { 2976 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE); 2977 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER); 2978 2979 /* add = 1, claim = 1 should be dquot_reserve_block in pair */ 2980 if (add) { 2981 if (claim) 2982 dquot_claim_block(inode, diff); 2983 else 2984 dquot_alloc_block_nofail(inode, diff); 2985 } else { 2986 dquot_free_block(inode, diff); 2987 } 2988 2989 f2fs_mark_inode_dirty_sync(inode, true); 2990 if (clean || recover) 2991 set_inode_flag(inode, FI_AUTO_RECOVER); 2992 } 2993 2994 static inline void f2fs_i_size_write(struct inode *inode, loff_t i_size) 2995 { 2996 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE); 2997 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER); 2998 2999 if (i_size_read(inode) == i_size) 3000 return; 3001 3002 i_size_write(inode, i_size); 3003 f2fs_mark_inode_dirty_sync(inode, true); 3004 if (clean || recover) 3005 set_inode_flag(inode, FI_AUTO_RECOVER); 3006 } 3007 3008 static inline void f2fs_i_depth_write(struct inode *inode, unsigned int depth) 3009 { 3010 F2FS_I(inode)->i_current_depth = depth; 3011 f2fs_mark_inode_dirty_sync(inode, true); 3012 } 3013 3014 static inline void f2fs_i_gc_failures_write(struct inode *inode, 3015 unsigned int count) 3016 { 3017 F2FS_I(inode)->i_gc_failures[GC_FAILURE_PIN] = count; 3018 f2fs_mark_inode_dirty_sync(inode, true); 3019 } 3020 3021 static inline void f2fs_i_xnid_write(struct inode *inode, nid_t xnid) 3022 { 3023 F2FS_I(inode)->i_xattr_nid = xnid; 3024 f2fs_mark_inode_dirty_sync(inode, true); 3025 } 3026 3027 static inline void f2fs_i_pino_write(struct inode *inode, nid_t pino) 3028 { 3029 F2FS_I(inode)->i_pino = pino; 3030 f2fs_mark_inode_dirty_sync(inode, true); 3031 } 3032 3033 static inline void get_inline_info(struct inode *inode, struct f2fs_inode *ri) 3034 { 3035 struct f2fs_inode_info *fi = F2FS_I(inode); 3036 3037 if (ri->i_inline & F2FS_INLINE_XATTR) 3038 set_bit(FI_INLINE_XATTR, fi->flags); 3039 if (ri->i_inline & F2FS_INLINE_DATA) 3040 set_bit(FI_INLINE_DATA, fi->flags); 3041 if (ri->i_inline & F2FS_INLINE_DENTRY) 3042 set_bit(FI_INLINE_DENTRY, fi->flags); 3043 if (ri->i_inline & F2FS_DATA_EXIST) 3044 set_bit(FI_DATA_EXIST, fi->flags); 3045 if (ri->i_inline & F2FS_INLINE_DOTS) 3046 set_bit(FI_INLINE_DOTS, fi->flags); 3047 if (ri->i_inline & F2FS_EXTRA_ATTR) 3048 set_bit(FI_EXTRA_ATTR, fi->flags); 3049 if (ri->i_inline & F2FS_PIN_FILE) 3050 set_bit(FI_PIN_FILE, fi->flags); 3051 if (ri->i_inline & F2FS_COMPRESS_RELEASED) 3052 set_bit(FI_COMPRESS_RELEASED, fi->flags); 3053 } 3054 3055 static inline void set_raw_inline(struct inode *inode, struct f2fs_inode *ri) 3056 { 3057 ri->i_inline = 0; 3058 3059 if (is_inode_flag_set(inode, FI_INLINE_XATTR)) 3060 ri->i_inline |= F2FS_INLINE_XATTR; 3061 if (is_inode_flag_set(inode, FI_INLINE_DATA)) 3062 ri->i_inline |= F2FS_INLINE_DATA; 3063 if (is_inode_flag_set(inode, FI_INLINE_DENTRY)) 3064 ri->i_inline |= F2FS_INLINE_DENTRY; 3065 if (is_inode_flag_set(inode, FI_DATA_EXIST)) 3066 ri->i_inline |= F2FS_DATA_EXIST; 3067 if (is_inode_flag_set(inode, FI_INLINE_DOTS)) 3068 ri->i_inline |= F2FS_INLINE_DOTS; 3069 if (is_inode_flag_set(inode, FI_EXTRA_ATTR)) 3070 ri->i_inline |= F2FS_EXTRA_ATTR; 3071 if (is_inode_flag_set(inode, FI_PIN_FILE)) 3072 ri->i_inline |= F2FS_PIN_FILE; 3073 if (is_inode_flag_set(inode, FI_COMPRESS_RELEASED)) 3074 ri->i_inline |= F2FS_COMPRESS_RELEASED; 3075 } 3076 3077 static inline int f2fs_has_extra_attr(struct inode *inode) 3078 { 3079 return is_inode_flag_set(inode, FI_EXTRA_ATTR); 3080 } 3081 3082 static inline int f2fs_has_inline_xattr(struct inode *inode) 3083 { 3084 return is_inode_flag_set(inode, FI_INLINE_XATTR); 3085 } 3086 3087 static inline int f2fs_compressed_file(struct inode *inode) 3088 { 3089 return S_ISREG(inode->i_mode) && 3090 is_inode_flag_set(inode, FI_COMPRESSED_FILE); 3091 } 3092 3093 static inline bool f2fs_need_compress_data(struct inode *inode) 3094 { 3095 int compress_mode = F2FS_OPTION(F2FS_I_SB(inode)).compress_mode; 3096 3097 if (!f2fs_compressed_file(inode)) 3098 return false; 3099 3100 if (compress_mode == COMPR_MODE_FS) 3101 return true; 3102 else if (compress_mode == COMPR_MODE_USER && 3103 is_inode_flag_set(inode, FI_ENABLE_COMPRESS)) 3104 return true; 3105 3106 return false; 3107 } 3108 3109 static inline unsigned int addrs_per_inode(struct inode *inode) 3110 { 3111 unsigned int addrs = CUR_ADDRS_PER_INODE(inode) - 3112 get_inline_xattr_addrs(inode); 3113 3114 if (!f2fs_compressed_file(inode)) 3115 return addrs; 3116 return ALIGN_DOWN(addrs, F2FS_I(inode)->i_cluster_size); 3117 } 3118 3119 static inline unsigned int addrs_per_block(struct inode *inode) 3120 { 3121 if (!f2fs_compressed_file(inode)) 3122 return DEF_ADDRS_PER_BLOCK; 3123 return ALIGN_DOWN(DEF_ADDRS_PER_BLOCK, F2FS_I(inode)->i_cluster_size); 3124 } 3125 3126 static inline void *inline_xattr_addr(struct inode *inode, struct page *page) 3127 { 3128 struct f2fs_inode *ri = F2FS_INODE(page); 3129 3130 return (void *)&(ri->i_addr[DEF_ADDRS_PER_INODE - 3131 get_inline_xattr_addrs(inode)]); 3132 } 3133 3134 static inline int inline_xattr_size(struct inode *inode) 3135 { 3136 if (f2fs_has_inline_xattr(inode)) 3137 return get_inline_xattr_addrs(inode) * sizeof(__le32); 3138 return 0; 3139 } 3140 3141 static inline int f2fs_has_inline_data(struct inode *inode) 3142 { 3143 return is_inode_flag_set(inode, FI_INLINE_DATA); 3144 } 3145 3146 static inline int f2fs_exist_data(struct inode *inode) 3147 { 3148 return is_inode_flag_set(inode, FI_DATA_EXIST); 3149 } 3150 3151 static inline int f2fs_has_inline_dots(struct inode *inode) 3152 { 3153 return is_inode_flag_set(inode, FI_INLINE_DOTS); 3154 } 3155 3156 static inline int f2fs_is_mmap_file(struct inode *inode) 3157 { 3158 return is_inode_flag_set(inode, FI_MMAP_FILE); 3159 } 3160 3161 static inline bool f2fs_is_pinned_file(struct inode *inode) 3162 { 3163 return is_inode_flag_set(inode, FI_PIN_FILE); 3164 } 3165 3166 static inline bool f2fs_is_atomic_file(struct inode *inode) 3167 { 3168 return is_inode_flag_set(inode, FI_ATOMIC_FILE); 3169 } 3170 3171 static inline bool f2fs_is_commit_atomic_write(struct inode *inode) 3172 { 3173 return is_inode_flag_set(inode, FI_ATOMIC_COMMIT); 3174 } 3175 3176 static inline bool f2fs_is_volatile_file(struct inode *inode) 3177 { 3178 return is_inode_flag_set(inode, FI_VOLATILE_FILE); 3179 } 3180 3181 static inline bool f2fs_is_first_block_written(struct inode *inode) 3182 { 3183 return is_inode_flag_set(inode, FI_FIRST_BLOCK_WRITTEN); 3184 } 3185 3186 static inline bool f2fs_is_drop_cache(struct inode *inode) 3187 { 3188 return is_inode_flag_set(inode, FI_DROP_CACHE); 3189 } 3190 3191 static inline void *inline_data_addr(struct inode *inode, struct page *page) 3192 { 3193 struct f2fs_inode *ri = F2FS_INODE(page); 3194 int extra_size = get_extra_isize(inode); 3195 3196 return (void *)&(ri->i_addr[extra_size + DEF_INLINE_RESERVED_SIZE]); 3197 } 3198 3199 static inline int f2fs_has_inline_dentry(struct inode *inode) 3200 { 3201 return is_inode_flag_set(inode, FI_INLINE_DENTRY); 3202 } 3203 3204 static inline int is_file(struct inode *inode, int type) 3205 { 3206 return F2FS_I(inode)->i_advise & type; 3207 } 3208 3209 static inline void set_file(struct inode *inode, int type) 3210 { 3211 if (is_file(inode, type)) 3212 return; 3213 F2FS_I(inode)->i_advise |= type; 3214 f2fs_mark_inode_dirty_sync(inode, true); 3215 } 3216 3217 static inline void clear_file(struct inode *inode, int type) 3218 { 3219 if (!is_file(inode, type)) 3220 return; 3221 F2FS_I(inode)->i_advise &= ~type; 3222 f2fs_mark_inode_dirty_sync(inode, true); 3223 } 3224 3225 static inline bool f2fs_is_time_consistent(struct inode *inode) 3226 { 3227 if (!timespec64_equal(F2FS_I(inode)->i_disk_time, &inode->i_atime)) 3228 return false; 3229 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 1, &inode->i_ctime)) 3230 return false; 3231 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 2, &inode->i_mtime)) 3232 return false; 3233 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 3, 3234 &F2FS_I(inode)->i_crtime)) 3235 return false; 3236 return true; 3237 } 3238 3239 static inline bool f2fs_skip_inode_update(struct inode *inode, int dsync) 3240 { 3241 bool ret; 3242 3243 if (dsync) { 3244 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 3245 3246 spin_lock(&sbi->inode_lock[DIRTY_META]); 3247 ret = list_empty(&F2FS_I(inode)->gdirty_list); 3248 spin_unlock(&sbi->inode_lock[DIRTY_META]); 3249 return ret; 3250 } 3251 if (!is_inode_flag_set(inode, FI_AUTO_RECOVER) || 3252 file_keep_isize(inode) || 3253 i_size_read(inode) & ~PAGE_MASK) 3254 return false; 3255 3256 if (!f2fs_is_time_consistent(inode)) 3257 return false; 3258 3259 spin_lock(&F2FS_I(inode)->i_size_lock); 3260 ret = F2FS_I(inode)->last_disk_size == i_size_read(inode); 3261 spin_unlock(&F2FS_I(inode)->i_size_lock); 3262 3263 return ret; 3264 } 3265 3266 static inline bool f2fs_readonly(struct super_block *sb) 3267 { 3268 return sb_rdonly(sb); 3269 } 3270 3271 static inline bool f2fs_cp_error(struct f2fs_sb_info *sbi) 3272 { 3273 return is_set_ckpt_flags(sbi, CP_ERROR_FLAG); 3274 } 3275 3276 static inline bool is_dot_dotdot(const u8 *name, size_t len) 3277 { 3278 if (len == 1 && name[0] == '.') 3279 return true; 3280 3281 if (len == 2 && name[0] == '.' && name[1] == '.') 3282 return true; 3283 3284 return false; 3285 } 3286 3287 static inline void *f2fs_kmalloc(struct f2fs_sb_info *sbi, 3288 size_t size, gfp_t flags) 3289 { 3290 if (time_to_inject(sbi, FAULT_KMALLOC)) { 3291 f2fs_show_injection_info(sbi, FAULT_KMALLOC); 3292 return NULL; 3293 } 3294 3295 return kmalloc(size, flags); 3296 } 3297 3298 static inline void *f2fs_kzalloc(struct f2fs_sb_info *sbi, 3299 size_t size, gfp_t flags) 3300 { 3301 return f2fs_kmalloc(sbi, size, flags | __GFP_ZERO); 3302 } 3303 3304 static inline void *f2fs_kvmalloc(struct f2fs_sb_info *sbi, 3305 size_t size, gfp_t flags) 3306 { 3307 if (time_to_inject(sbi, FAULT_KVMALLOC)) { 3308 f2fs_show_injection_info(sbi, FAULT_KVMALLOC); 3309 return NULL; 3310 } 3311 3312 return kvmalloc(size, flags); 3313 } 3314 3315 static inline void *f2fs_kvzalloc(struct f2fs_sb_info *sbi, 3316 size_t size, gfp_t flags) 3317 { 3318 return f2fs_kvmalloc(sbi, size, flags | __GFP_ZERO); 3319 } 3320 3321 static inline int get_extra_isize(struct inode *inode) 3322 { 3323 return F2FS_I(inode)->i_extra_isize / sizeof(__le32); 3324 } 3325 3326 static inline int get_inline_xattr_addrs(struct inode *inode) 3327 { 3328 return F2FS_I(inode)->i_inline_xattr_size; 3329 } 3330 3331 #define f2fs_get_inode_mode(i) \ 3332 ((is_inode_flag_set(i, FI_ACL_MODE)) ? \ 3333 (F2FS_I(i)->i_acl_mode) : ((i)->i_mode)) 3334 3335 #define F2FS_TOTAL_EXTRA_ATTR_SIZE \ 3336 (offsetof(struct f2fs_inode, i_extra_end) - \ 3337 offsetof(struct f2fs_inode, i_extra_isize)) \ 3338 3339 #define F2FS_OLD_ATTRIBUTE_SIZE (offsetof(struct f2fs_inode, i_addr)) 3340 #define F2FS_FITS_IN_INODE(f2fs_inode, extra_isize, field) \ 3341 ((offsetof(typeof(*(f2fs_inode)), field) + \ 3342 sizeof((f2fs_inode)->field)) \ 3343 <= (F2FS_OLD_ATTRIBUTE_SIZE + (extra_isize))) \ 3344 3345 #define __is_large_section(sbi) ((sbi)->segs_per_sec > 1) 3346 3347 #define __is_meta_io(fio) (PAGE_TYPE_OF_BIO((fio)->type) == META) 3348 3349 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi, 3350 block_t blkaddr, int type); 3351 static inline void verify_blkaddr(struct f2fs_sb_info *sbi, 3352 block_t blkaddr, int type) 3353 { 3354 if (!f2fs_is_valid_blkaddr(sbi, blkaddr, type)) { 3355 f2fs_err(sbi, "invalid blkaddr: %u, type: %d, run fsck to fix.", 3356 blkaddr, type); 3357 f2fs_bug_on(sbi, 1); 3358 } 3359 } 3360 3361 static inline bool __is_valid_data_blkaddr(block_t blkaddr) 3362 { 3363 if (blkaddr == NEW_ADDR || blkaddr == NULL_ADDR || 3364 blkaddr == COMPRESS_ADDR) 3365 return false; 3366 return true; 3367 } 3368 3369 /* 3370 * file.c 3371 */ 3372 int f2fs_sync_file(struct file *file, loff_t start, loff_t end, int datasync); 3373 void f2fs_truncate_data_blocks(struct dnode_of_data *dn); 3374 int f2fs_do_truncate_blocks(struct inode *inode, u64 from, bool lock); 3375 int f2fs_truncate_blocks(struct inode *inode, u64 from, bool lock); 3376 int f2fs_truncate(struct inode *inode); 3377 int f2fs_getattr(struct user_namespace *mnt_userns, const struct path *path, 3378 struct kstat *stat, u32 request_mask, unsigned int flags); 3379 int f2fs_setattr(struct user_namespace *mnt_userns, struct dentry *dentry, 3380 struct iattr *attr); 3381 int f2fs_truncate_hole(struct inode *inode, pgoff_t pg_start, pgoff_t pg_end); 3382 void f2fs_truncate_data_blocks_range(struct dnode_of_data *dn, int count); 3383 int f2fs_precache_extents(struct inode *inode); 3384 int f2fs_fileattr_get(struct dentry *dentry, struct fileattr *fa); 3385 int f2fs_fileattr_set(struct user_namespace *mnt_userns, 3386 struct dentry *dentry, struct fileattr *fa); 3387 long f2fs_ioctl(struct file *filp, unsigned int cmd, unsigned long arg); 3388 long f2fs_compat_ioctl(struct file *file, unsigned int cmd, unsigned long arg); 3389 int f2fs_transfer_project_quota(struct inode *inode, kprojid_t kprojid); 3390 int f2fs_pin_file_control(struct inode *inode, bool inc); 3391 3392 /* 3393 * inode.c 3394 */ 3395 void f2fs_set_inode_flags(struct inode *inode); 3396 bool f2fs_inode_chksum_verify(struct f2fs_sb_info *sbi, struct page *page); 3397 void f2fs_inode_chksum_set(struct f2fs_sb_info *sbi, struct page *page); 3398 struct inode *f2fs_iget(struct super_block *sb, unsigned long ino); 3399 struct inode *f2fs_iget_retry(struct super_block *sb, unsigned long ino); 3400 int f2fs_try_to_free_nats(struct f2fs_sb_info *sbi, int nr_shrink); 3401 void f2fs_update_inode(struct inode *inode, struct page *node_page); 3402 void f2fs_update_inode_page(struct inode *inode); 3403 int f2fs_write_inode(struct inode *inode, struct writeback_control *wbc); 3404 void f2fs_evict_inode(struct inode *inode); 3405 void f2fs_handle_failed_inode(struct inode *inode); 3406 3407 /* 3408 * namei.c 3409 */ 3410 int f2fs_update_extension_list(struct f2fs_sb_info *sbi, const char *name, 3411 bool hot, bool set); 3412 struct dentry *f2fs_get_parent(struct dentry *child); 3413 3414 /* 3415 * dir.c 3416 */ 3417 unsigned char f2fs_get_de_type(struct f2fs_dir_entry *de); 3418 int f2fs_init_casefolded_name(const struct inode *dir, 3419 struct f2fs_filename *fname); 3420 int f2fs_setup_filename(struct inode *dir, const struct qstr *iname, 3421 int lookup, struct f2fs_filename *fname); 3422 int f2fs_prepare_lookup(struct inode *dir, struct dentry *dentry, 3423 struct f2fs_filename *fname); 3424 void f2fs_free_filename(struct f2fs_filename *fname); 3425 struct f2fs_dir_entry *f2fs_find_target_dentry(const struct f2fs_dentry_ptr *d, 3426 const struct f2fs_filename *fname, int *max_slots); 3427 int f2fs_fill_dentries(struct dir_context *ctx, struct f2fs_dentry_ptr *d, 3428 unsigned int start_pos, struct fscrypt_str *fstr); 3429 void f2fs_do_make_empty_dir(struct inode *inode, struct inode *parent, 3430 struct f2fs_dentry_ptr *d); 3431 struct page *f2fs_init_inode_metadata(struct inode *inode, struct inode *dir, 3432 const struct f2fs_filename *fname, struct page *dpage); 3433 void f2fs_update_parent_metadata(struct inode *dir, struct inode *inode, 3434 unsigned int current_depth); 3435 int f2fs_room_for_filename(const void *bitmap, int slots, int max_slots); 3436 void f2fs_drop_nlink(struct inode *dir, struct inode *inode); 3437 struct f2fs_dir_entry *__f2fs_find_entry(struct inode *dir, 3438 const struct f2fs_filename *fname, 3439 struct page **res_page); 3440 struct f2fs_dir_entry *f2fs_find_entry(struct inode *dir, 3441 const struct qstr *child, struct page **res_page); 3442 struct f2fs_dir_entry *f2fs_parent_dir(struct inode *dir, struct page **p); 3443 ino_t f2fs_inode_by_name(struct inode *dir, const struct qstr *qstr, 3444 struct page **page); 3445 void f2fs_set_link(struct inode *dir, struct f2fs_dir_entry *de, 3446 struct page *page, struct inode *inode); 3447 bool f2fs_has_enough_room(struct inode *dir, struct page *ipage, 3448 const struct f2fs_filename *fname); 3449 void f2fs_update_dentry(nid_t ino, umode_t mode, struct f2fs_dentry_ptr *d, 3450 const struct fscrypt_str *name, f2fs_hash_t name_hash, 3451 unsigned int bit_pos); 3452 int f2fs_add_regular_entry(struct inode *dir, const struct f2fs_filename *fname, 3453 struct inode *inode, nid_t ino, umode_t mode); 3454 int f2fs_add_dentry(struct inode *dir, const struct f2fs_filename *fname, 3455 struct inode *inode, nid_t ino, umode_t mode); 3456 int f2fs_do_add_link(struct inode *dir, const struct qstr *name, 3457 struct inode *inode, nid_t ino, umode_t mode); 3458 void f2fs_delete_entry(struct f2fs_dir_entry *dentry, struct page *page, 3459 struct inode *dir, struct inode *inode); 3460 int f2fs_do_tmpfile(struct inode *inode, struct inode *dir); 3461 bool f2fs_empty_dir(struct inode *dir); 3462 3463 static inline int f2fs_add_link(struct dentry *dentry, struct inode *inode) 3464 { 3465 if (fscrypt_is_nokey_name(dentry)) 3466 return -ENOKEY; 3467 return f2fs_do_add_link(d_inode(dentry->d_parent), &dentry->d_name, 3468 inode, inode->i_ino, inode->i_mode); 3469 } 3470 3471 /* 3472 * super.c 3473 */ 3474 int f2fs_inode_dirtied(struct inode *inode, bool sync); 3475 void f2fs_inode_synced(struct inode *inode); 3476 int f2fs_dquot_initialize(struct inode *inode); 3477 int f2fs_enable_quota_files(struct f2fs_sb_info *sbi, bool rdonly); 3478 int f2fs_quota_sync(struct super_block *sb, int type); 3479 loff_t max_file_blocks(struct inode *inode); 3480 void f2fs_quota_off_umount(struct super_block *sb); 3481 int f2fs_commit_super(struct f2fs_sb_info *sbi, bool recover); 3482 int f2fs_sync_fs(struct super_block *sb, int sync); 3483 int f2fs_sanity_check_ckpt(struct f2fs_sb_info *sbi); 3484 3485 /* 3486 * hash.c 3487 */ 3488 void f2fs_hash_filename(const struct inode *dir, struct f2fs_filename *fname); 3489 3490 /* 3491 * node.c 3492 */ 3493 struct node_info; 3494 3495 int f2fs_check_nid_range(struct f2fs_sb_info *sbi, nid_t nid); 3496 bool f2fs_available_free_memory(struct f2fs_sb_info *sbi, int type); 3497 bool f2fs_in_warm_node_list(struct f2fs_sb_info *sbi, struct page *page); 3498 void f2fs_init_fsync_node_info(struct f2fs_sb_info *sbi); 3499 void f2fs_del_fsync_node_entry(struct f2fs_sb_info *sbi, struct page *page); 3500 void f2fs_reset_fsync_node_info(struct f2fs_sb_info *sbi); 3501 int f2fs_need_dentry_mark(struct f2fs_sb_info *sbi, nid_t nid); 3502 bool f2fs_is_checkpointed_node(struct f2fs_sb_info *sbi, nid_t nid); 3503 bool f2fs_need_inode_block_update(struct f2fs_sb_info *sbi, nid_t ino); 3504 int f2fs_get_node_info(struct f2fs_sb_info *sbi, nid_t nid, 3505 struct node_info *ni, bool checkpoint_context); 3506 pgoff_t f2fs_get_next_page_offset(struct dnode_of_data *dn, pgoff_t pgofs); 3507 int f2fs_get_dnode_of_data(struct dnode_of_data *dn, pgoff_t index, int mode); 3508 int f2fs_truncate_inode_blocks(struct inode *inode, pgoff_t from); 3509 int f2fs_truncate_xattr_node(struct inode *inode); 3510 int f2fs_wait_on_node_pages_writeback(struct f2fs_sb_info *sbi, 3511 unsigned int seq_id); 3512 bool f2fs_nat_bitmap_enabled(struct f2fs_sb_info *sbi); 3513 int f2fs_remove_inode_page(struct inode *inode); 3514 struct page *f2fs_new_inode_page(struct inode *inode); 3515 struct page *f2fs_new_node_page(struct dnode_of_data *dn, unsigned int ofs); 3516 void f2fs_ra_node_page(struct f2fs_sb_info *sbi, nid_t nid); 3517 struct page *f2fs_get_node_page(struct f2fs_sb_info *sbi, pgoff_t nid); 3518 struct page *f2fs_get_node_page_ra(struct page *parent, int start); 3519 int f2fs_move_node_page(struct page *node_page, int gc_type); 3520 void f2fs_flush_inline_data(struct f2fs_sb_info *sbi); 3521 int f2fs_fsync_node_pages(struct f2fs_sb_info *sbi, struct inode *inode, 3522 struct writeback_control *wbc, bool atomic, 3523 unsigned int *seq_id); 3524 int f2fs_sync_node_pages(struct f2fs_sb_info *sbi, 3525 struct writeback_control *wbc, 3526 bool do_balance, enum iostat_type io_type); 3527 int f2fs_build_free_nids(struct f2fs_sb_info *sbi, bool sync, bool mount); 3528 bool f2fs_alloc_nid(struct f2fs_sb_info *sbi, nid_t *nid); 3529 void f2fs_alloc_nid_done(struct f2fs_sb_info *sbi, nid_t nid); 3530 void f2fs_alloc_nid_failed(struct f2fs_sb_info *sbi, nid_t nid); 3531 int f2fs_try_to_free_nids(struct f2fs_sb_info *sbi, int nr_shrink); 3532 int f2fs_recover_inline_xattr(struct inode *inode, struct page *page); 3533 int f2fs_recover_xattr_data(struct inode *inode, struct page *page); 3534 int f2fs_recover_inode_page(struct f2fs_sb_info *sbi, struct page *page); 3535 int f2fs_restore_node_summary(struct f2fs_sb_info *sbi, 3536 unsigned int segno, struct f2fs_summary_block *sum); 3537 void f2fs_enable_nat_bits(struct f2fs_sb_info *sbi); 3538 int f2fs_flush_nat_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3539 int f2fs_build_node_manager(struct f2fs_sb_info *sbi); 3540 void f2fs_destroy_node_manager(struct f2fs_sb_info *sbi); 3541 int __init f2fs_create_node_manager_caches(void); 3542 void f2fs_destroy_node_manager_caches(void); 3543 3544 /* 3545 * segment.c 3546 */ 3547 bool f2fs_need_SSR(struct f2fs_sb_info *sbi); 3548 void f2fs_register_inmem_page(struct inode *inode, struct page *page); 3549 void f2fs_drop_inmem_pages_all(struct f2fs_sb_info *sbi, bool gc_failure); 3550 void f2fs_drop_inmem_pages(struct inode *inode); 3551 void f2fs_drop_inmem_page(struct inode *inode, struct page *page); 3552 int f2fs_commit_inmem_pages(struct inode *inode); 3553 void f2fs_balance_fs(struct f2fs_sb_info *sbi, bool need); 3554 void f2fs_balance_fs_bg(struct f2fs_sb_info *sbi, bool from_bg); 3555 int f2fs_issue_flush(struct f2fs_sb_info *sbi, nid_t ino); 3556 int f2fs_create_flush_cmd_control(struct f2fs_sb_info *sbi); 3557 int f2fs_flush_device_cache(struct f2fs_sb_info *sbi); 3558 void f2fs_destroy_flush_cmd_control(struct f2fs_sb_info *sbi, bool free); 3559 void f2fs_invalidate_blocks(struct f2fs_sb_info *sbi, block_t addr); 3560 bool f2fs_is_checkpointed_data(struct f2fs_sb_info *sbi, block_t blkaddr); 3561 int f2fs_start_discard_thread(struct f2fs_sb_info *sbi); 3562 void f2fs_drop_discard_cmd(struct f2fs_sb_info *sbi); 3563 void f2fs_stop_discard_thread(struct f2fs_sb_info *sbi); 3564 bool f2fs_issue_discard_timeout(struct f2fs_sb_info *sbi); 3565 void f2fs_clear_prefree_segments(struct f2fs_sb_info *sbi, 3566 struct cp_control *cpc); 3567 void f2fs_dirty_to_prefree(struct f2fs_sb_info *sbi); 3568 block_t f2fs_get_unusable_blocks(struct f2fs_sb_info *sbi); 3569 int f2fs_disable_cp_again(struct f2fs_sb_info *sbi, block_t unusable); 3570 void f2fs_release_discard_addrs(struct f2fs_sb_info *sbi); 3571 int f2fs_npages_for_summary_flush(struct f2fs_sb_info *sbi, bool for_ra); 3572 bool f2fs_segment_has_free_slot(struct f2fs_sb_info *sbi, int segno); 3573 void f2fs_init_inmem_curseg(struct f2fs_sb_info *sbi); 3574 void f2fs_save_inmem_curseg(struct f2fs_sb_info *sbi); 3575 void f2fs_restore_inmem_curseg(struct f2fs_sb_info *sbi); 3576 void f2fs_get_new_segment(struct f2fs_sb_info *sbi, 3577 unsigned int *newseg, bool new_sec, int dir); 3578 void f2fs_allocate_segment_for_resize(struct f2fs_sb_info *sbi, int type, 3579 unsigned int start, unsigned int end); 3580 void f2fs_allocate_new_section(struct f2fs_sb_info *sbi, int type, bool force); 3581 void f2fs_allocate_new_segments(struct f2fs_sb_info *sbi); 3582 int f2fs_trim_fs(struct f2fs_sb_info *sbi, struct fstrim_range *range); 3583 bool f2fs_exist_trim_candidates(struct f2fs_sb_info *sbi, 3584 struct cp_control *cpc); 3585 struct page *f2fs_get_sum_page(struct f2fs_sb_info *sbi, unsigned int segno); 3586 void f2fs_update_meta_page(struct f2fs_sb_info *sbi, void *src, 3587 block_t blk_addr); 3588 void f2fs_do_write_meta_page(struct f2fs_sb_info *sbi, struct page *page, 3589 enum iostat_type io_type); 3590 void f2fs_do_write_node_page(unsigned int nid, struct f2fs_io_info *fio); 3591 void f2fs_outplace_write_data(struct dnode_of_data *dn, 3592 struct f2fs_io_info *fio); 3593 int f2fs_inplace_write_data(struct f2fs_io_info *fio); 3594 void f2fs_do_replace_block(struct f2fs_sb_info *sbi, struct f2fs_summary *sum, 3595 block_t old_blkaddr, block_t new_blkaddr, 3596 bool recover_curseg, bool recover_newaddr, 3597 bool from_gc); 3598 void f2fs_replace_block(struct f2fs_sb_info *sbi, struct dnode_of_data *dn, 3599 block_t old_addr, block_t new_addr, 3600 unsigned char version, bool recover_curseg, 3601 bool recover_newaddr); 3602 void f2fs_allocate_data_block(struct f2fs_sb_info *sbi, struct page *page, 3603 block_t old_blkaddr, block_t *new_blkaddr, 3604 struct f2fs_summary *sum, int type, 3605 struct f2fs_io_info *fio); 3606 void f2fs_update_device_state(struct f2fs_sb_info *sbi, nid_t ino, 3607 block_t blkaddr, unsigned int blkcnt); 3608 void f2fs_wait_on_page_writeback(struct page *page, 3609 enum page_type type, bool ordered, bool locked); 3610 void f2fs_wait_on_block_writeback(struct inode *inode, block_t blkaddr); 3611 void f2fs_wait_on_block_writeback_range(struct inode *inode, block_t blkaddr, 3612 block_t len); 3613 void f2fs_write_data_summaries(struct f2fs_sb_info *sbi, block_t start_blk); 3614 void f2fs_write_node_summaries(struct f2fs_sb_info *sbi, block_t start_blk); 3615 int f2fs_lookup_journal_in_cursum(struct f2fs_journal *journal, int type, 3616 unsigned int val, int alloc); 3617 void f2fs_flush_sit_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3618 int f2fs_fix_curseg_write_pointer(struct f2fs_sb_info *sbi); 3619 int f2fs_check_write_pointer(struct f2fs_sb_info *sbi); 3620 int f2fs_build_segment_manager(struct f2fs_sb_info *sbi); 3621 void f2fs_destroy_segment_manager(struct f2fs_sb_info *sbi); 3622 int __init f2fs_create_segment_manager_caches(void); 3623 void f2fs_destroy_segment_manager_caches(void); 3624 int f2fs_rw_hint_to_seg_type(enum rw_hint hint); 3625 enum rw_hint f2fs_io_type_to_rw_hint(struct f2fs_sb_info *sbi, 3626 enum page_type type, enum temp_type temp); 3627 unsigned int f2fs_usable_segs_in_sec(struct f2fs_sb_info *sbi, 3628 unsigned int segno); 3629 unsigned int f2fs_usable_blks_in_seg(struct f2fs_sb_info *sbi, 3630 unsigned int segno); 3631 3632 #define DEF_FRAGMENT_SIZE 4 3633 #define MIN_FRAGMENT_SIZE 1 3634 #define MAX_FRAGMENT_SIZE 512 3635 3636 static inline bool f2fs_need_rand_seg(struct f2fs_sb_info *sbi) 3637 { 3638 return F2FS_OPTION(sbi).fs_mode == FS_MODE_FRAGMENT_SEG || 3639 F2FS_OPTION(sbi).fs_mode == FS_MODE_FRAGMENT_BLK; 3640 } 3641 3642 /* 3643 * checkpoint.c 3644 */ 3645 void f2fs_stop_checkpoint(struct f2fs_sb_info *sbi, bool end_io); 3646 struct page *f2fs_grab_meta_page(struct f2fs_sb_info *sbi, pgoff_t index); 3647 struct page *f2fs_get_meta_page(struct f2fs_sb_info *sbi, pgoff_t index); 3648 struct page *f2fs_get_meta_page_retry(struct f2fs_sb_info *sbi, pgoff_t index); 3649 struct page *f2fs_get_tmp_page(struct f2fs_sb_info *sbi, pgoff_t index); 3650 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi, 3651 block_t blkaddr, int type); 3652 int f2fs_ra_meta_pages(struct f2fs_sb_info *sbi, block_t start, int nrpages, 3653 int type, bool sync); 3654 void f2fs_ra_meta_pages_cond(struct f2fs_sb_info *sbi, pgoff_t index); 3655 long f2fs_sync_meta_pages(struct f2fs_sb_info *sbi, enum page_type type, 3656 long nr_to_write, enum iostat_type io_type); 3657 void f2fs_add_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type); 3658 void f2fs_remove_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type); 3659 void f2fs_release_ino_entry(struct f2fs_sb_info *sbi, bool all); 3660 bool f2fs_exist_written_data(struct f2fs_sb_info *sbi, nid_t ino, int mode); 3661 void f2fs_set_dirty_device(struct f2fs_sb_info *sbi, nid_t ino, 3662 unsigned int devidx, int type); 3663 bool f2fs_is_dirty_device(struct f2fs_sb_info *sbi, nid_t ino, 3664 unsigned int devidx, int type); 3665 int f2fs_sync_inode_meta(struct f2fs_sb_info *sbi); 3666 int f2fs_acquire_orphan_inode(struct f2fs_sb_info *sbi); 3667 void f2fs_release_orphan_inode(struct f2fs_sb_info *sbi); 3668 void f2fs_add_orphan_inode(struct inode *inode); 3669 void f2fs_remove_orphan_inode(struct f2fs_sb_info *sbi, nid_t ino); 3670 int f2fs_recover_orphan_inodes(struct f2fs_sb_info *sbi); 3671 int f2fs_get_valid_checkpoint(struct f2fs_sb_info *sbi); 3672 void f2fs_update_dirty_page(struct inode *inode, struct page *page); 3673 void f2fs_remove_dirty_inode(struct inode *inode); 3674 int f2fs_sync_dirty_inodes(struct f2fs_sb_info *sbi, enum inode_type type); 3675 void f2fs_wait_on_all_pages(struct f2fs_sb_info *sbi, int type); 3676 u64 f2fs_get_sectors_written(struct f2fs_sb_info *sbi); 3677 int f2fs_write_checkpoint(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3678 void f2fs_init_ino_entry_info(struct f2fs_sb_info *sbi); 3679 int __init f2fs_create_checkpoint_caches(void); 3680 void f2fs_destroy_checkpoint_caches(void); 3681 int f2fs_issue_checkpoint(struct f2fs_sb_info *sbi); 3682 int f2fs_start_ckpt_thread(struct f2fs_sb_info *sbi); 3683 void f2fs_stop_ckpt_thread(struct f2fs_sb_info *sbi); 3684 void f2fs_init_ckpt_req_control(struct f2fs_sb_info *sbi); 3685 3686 /* 3687 * data.c 3688 */ 3689 int __init f2fs_init_bioset(void); 3690 void f2fs_destroy_bioset(void); 3691 int f2fs_init_bio_entry_cache(void); 3692 void f2fs_destroy_bio_entry_cache(void); 3693 void f2fs_submit_bio(struct f2fs_sb_info *sbi, 3694 struct bio *bio, enum page_type type); 3695 void f2fs_submit_merged_write(struct f2fs_sb_info *sbi, enum page_type type); 3696 void f2fs_submit_merged_write_cond(struct f2fs_sb_info *sbi, 3697 struct inode *inode, struct page *page, 3698 nid_t ino, enum page_type type); 3699 void f2fs_submit_merged_ipu_write(struct f2fs_sb_info *sbi, 3700 struct bio **bio, struct page *page); 3701 void f2fs_flush_merged_writes(struct f2fs_sb_info *sbi); 3702 int f2fs_submit_page_bio(struct f2fs_io_info *fio); 3703 int f2fs_merge_page_bio(struct f2fs_io_info *fio); 3704 void f2fs_submit_page_write(struct f2fs_io_info *fio); 3705 struct block_device *f2fs_target_device(struct f2fs_sb_info *sbi, 3706 block_t blk_addr, struct bio *bio); 3707 int f2fs_target_device_index(struct f2fs_sb_info *sbi, block_t blkaddr); 3708 void f2fs_set_data_blkaddr(struct dnode_of_data *dn); 3709 void f2fs_update_data_blkaddr(struct dnode_of_data *dn, block_t blkaddr); 3710 int f2fs_reserve_new_blocks(struct dnode_of_data *dn, blkcnt_t count); 3711 int f2fs_reserve_new_block(struct dnode_of_data *dn); 3712 int f2fs_get_block(struct dnode_of_data *dn, pgoff_t index); 3713 int f2fs_reserve_block(struct dnode_of_data *dn, pgoff_t index); 3714 struct page *f2fs_get_read_data_page(struct inode *inode, pgoff_t index, 3715 int op_flags, bool for_write); 3716 struct page *f2fs_find_data_page(struct inode *inode, pgoff_t index); 3717 struct page *f2fs_get_lock_data_page(struct inode *inode, pgoff_t index, 3718 bool for_write); 3719 struct page *f2fs_get_new_data_page(struct inode *inode, 3720 struct page *ipage, pgoff_t index, bool new_i_size); 3721 int f2fs_do_write_data_page(struct f2fs_io_info *fio); 3722 void f2fs_do_map_lock(struct f2fs_sb_info *sbi, int flag, bool lock); 3723 int f2fs_map_blocks(struct inode *inode, struct f2fs_map_blocks *map, 3724 int create, int flag); 3725 int f2fs_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, 3726 u64 start, u64 len); 3727 int f2fs_encrypt_one_page(struct f2fs_io_info *fio); 3728 bool f2fs_should_update_inplace(struct inode *inode, struct f2fs_io_info *fio); 3729 bool f2fs_should_update_outplace(struct inode *inode, struct f2fs_io_info *fio); 3730 int f2fs_write_single_data_page(struct page *page, int *submitted, 3731 struct bio **bio, sector_t *last_block, 3732 struct writeback_control *wbc, 3733 enum iostat_type io_type, 3734 int compr_blocks, bool allow_balance); 3735 void f2fs_write_failed(struct inode *inode, loff_t to); 3736 void f2fs_invalidate_page(struct page *page, unsigned int offset, 3737 unsigned int length); 3738 int f2fs_release_page(struct page *page, gfp_t wait); 3739 #ifdef CONFIG_MIGRATION 3740 int f2fs_migrate_page(struct address_space *mapping, struct page *newpage, 3741 struct page *page, enum migrate_mode mode); 3742 #endif 3743 bool f2fs_overwrite_io(struct inode *inode, loff_t pos, size_t len); 3744 void f2fs_clear_page_cache_dirty_tag(struct page *page); 3745 int f2fs_init_post_read_processing(void); 3746 void f2fs_destroy_post_read_processing(void); 3747 int f2fs_init_post_read_wq(struct f2fs_sb_info *sbi); 3748 void f2fs_destroy_post_read_wq(struct f2fs_sb_info *sbi); 3749 extern const struct iomap_ops f2fs_iomap_ops; 3750 3751 /* 3752 * gc.c 3753 */ 3754 int f2fs_start_gc_thread(struct f2fs_sb_info *sbi); 3755 void f2fs_stop_gc_thread(struct f2fs_sb_info *sbi); 3756 block_t f2fs_start_bidx_of_node(unsigned int node_ofs, struct inode *inode); 3757 int f2fs_gc(struct f2fs_sb_info *sbi, bool sync, bool background, bool force, 3758 unsigned int segno); 3759 void f2fs_build_gc_manager(struct f2fs_sb_info *sbi); 3760 int f2fs_resize_fs(struct f2fs_sb_info *sbi, __u64 block_count); 3761 int __init f2fs_create_garbage_collection_cache(void); 3762 void f2fs_destroy_garbage_collection_cache(void); 3763 3764 /* 3765 * recovery.c 3766 */ 3767 int f2fs_recover_fsync_data(struct f2fs_sb_info *sbi, bool check_only); 3768 bool f2fs_space_for_roll_forward(struct f2fs_sb_info *sbi); 3769 int __init f2fs_create_recovery_cache(void); 3770 void f2fs_destroy_recovery_cache(void); 3771 3772 /* 3773 * debug.c 3774 */ 3775 #ifdef CONFIG_F2FS_STAT_FS 3776 struct f2fs_stat_info { 3777 struct list_head stat_list; 3778 struct f2fs_sb_info *sbi; 3779 int all_area_segs, sit_area_segs, nat_area_segs, ssa_area_segs; 3780 int main_area_segs, main_area_sections, main_area_zones; 3781 unsigned long long hit_largest, hit_cached, hit_rbtree; 3782 unsigned long long hit_total, total_ext; 3783 int ext_tree, zombie_tree, ext_node; 3784 int ndirty_node, ndirty_dent, ndirty_meta, ndirty_imeta; 3785 int ndirty_data, ndirty_qdata; 3786 int inmem_pages; 3787 unsigned int ndirty_dirs, ndirty_files, nquota_files, ndirty_all; 3788 int nats, dirty_nats, sits, dirty_sits; 3789 int free_nids, avail_nids, alloc_nids; 3790 int total_count, utilization; 3791 int bg_gc, nr_wb_cp_data, nr_wb_data; 3792 int nr_rd_data, nr_rd_node, nr_rd_meta; 3793 int nr_dio_read, nr_dio_write; 3794 unsigned int io_skip_bggc, other_skip_bggc; 3795 int nr_flushing, nr_flushed, flush_list_empty; 3796 int nr_discarding, nr_discarded; 3797 int nr_discard_cmd; 3798 unsigned int undiscard_blks; 3799 int nr_issued_ckpt, nr_total_ckpt, nr_queued_ckpt; 3800 unsigned int cur_ckpt_time, peak_ckpt_time; 3801 int inline_xattr, inline_inode, inline_dir, append, update, orphans; 3802 int compr_inode; 3803 unsigned long long compr_blocks; 3804 int aw_cnt, max_aw_cnt, vw_cnt, max_vw_cnt; 3805 unsigned int valid_count, valid_node_count, valid_inode_count, discard_blks; 3806 unsigned int bimodal, avg_vblocks; 3807 int util_free, util_valid, util_invalid; 3808 int rsvd_segs, overp_segs; 3809 int dirty_count, node_pages, meta_pages, compress_pages; 3810 int compress_page_hit; 3811 int prefree_count, call_count, cp_count, bg_cp_count; 3812 int tot_segs, node_segs, data_segs, free_segs, free_secs; 3813 int bg_node_segs, bg_data_segs; 3814 int tot_blks, data_blks, node_blks; 3815 int bg_data_blks, bg_node_blks; 3816 unsigned long long skipped_atomic_files[2]; 3817 int curseg[NR_CURSEG_TYPE]; 3818 int cursec[NR_CURSEG_TYPE]; 3819 int curzone[NR_CURSEG_TYPE]; 3820 unsigned int dirty_seg[NR_CURSEG_TYPE]; 3821 unsigned int full_seg[NR_CURSEG_TYPE]; 3822 unsigned int valid_blks[NR_CURSEG_TYPE]; 3823 3824 unsigned int meta_count[META_MAX]; 3825 unsigned int segment_count[2]; 3826 unsigned int block_count[2]; 3827 unsigned int inplace_count; 3828 unsigned long long base_mem, cache_mem, page_mem; 3829 }; 3830 3831 static inline struct f2fs_stat_info *F2FS_STAT(struct f2fs_sb_info *sbi) 3832 { 3833 return (struct f2fs_stat_info *)sbi->stat_info; 3834 } 3835 3836 #define stat_inc_cp_count(si) ((si)->cp_count++) 3837 #define stat_inc_bg_cp_count(si) ((si)->bg_cp_count++) 3838 #define stat_inc_call_count(si) ((si)->call_count++) 3839 #define stat_inc_bggc_count(si) ((si)->bg_gc++) 3840 #define stat_io_skip_bggc_count(sbi) ((sbi)->io_skip_bggc++) 3841 #define stat_other_skip_bggc_count(sbi) ((sbi)->other_skip_bggc++) 3842 #define stat_inc_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]++) 3843 #define stat_dec_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]--) 3844 #define stat_inc_total_hit(sbi) (atomic64_inc(&(sbi)->total_hit_ext)) 3845 #define stat_inc_rbtree_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_rbtree)) 3846 #define stat_inc_largest_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_largest)) 3847 #define stat_inc_cached_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_cached)) 3848 #define stat_inc_inline_xattr(inode) \ 3849 do { \ 3850 if (f2fs_has_inline_xattr(inode)) \ 3851 (atomic_inc(&F2FS_I_SB(inode)->inline_xattr)); \ 3852 } while (0) 3853 #define stat_dec_inline_xattr(inode) \ 3854 do { \ 3855 if (f2fs_has_inline_xattr(inode)) \ 3856 (atomic_dec(&F2FS_I_SB(inode)->inline_xattr)); \ 3857 } while (0) 3858 #define stat_inc_inline_inode(inode) \ 3859 do { \ 3860 if (f2fs_has_inline_data(inode)) \ 3861 (atomic_inc(&F2FS_I_SB(inode)->inline_inode)); \ 3862 } while (0) 3863 #define stat_dec_inline_inode(inode) \ 3864 do { \ 3865 if (f2fs_has_inline_data(inode)) \ 3866 (atomic_dec(&F2FS_I_SB(inode)->inline_inode)); \ 3867 } while (0) 3868 #define stat_inc_inline_dir(inode) \ 3869 do { \ 3870 if (f2fs_has_inline_dentry(inode)) \ 3871 (atomic_inc(&F2FS_I_SB(inode)->inline_dir)); \ 3872 } while (0) 3873 #define stat_dec_inline_dir(inode) \ 3874 do { \ 3875 if (f2fs_has_inline_dentry(inode)) \ 3876 (atomic_dec(&F2FS_I_SB(inode)->inline_dir)); \ 3877 } while (0) 3878 #define stat_inc_compr_inode(inode) \ 3879 do { \ 3880 if (f2fs_compressed_file(inode)) \ 3881 (atomic_inc(&F2FS_I_SB(inode)->compr_inode)); \ 3882 } while (0) 3883 #define stat_dec_compr_inode(inode) \ 3884 do { \ 3885 if (f2fs_compressed_file(inode)) \ 3886 (atomic_dec(&F2FS_I_SB(inode)->compr_inode)); \ 3887 } while (0) 3888 #define stat_add_compr_blocks(inode, blocks) \ 3889 (atomic64_add(blocks, &F2FS_I_SB(inode)->compr_blocks)) 3890 #define stat_sub_compr_blocks(inode, blocks) \ 3891 (atomic64_sub(blocks, &F2FS_I_SB(inode)->compr_blocks)) 3892 #define stat_inc_meta_count(sbi, blkaddr) \ 3893 do { \ 3894 if (blkaddr < SIT_I(sbi)->sit_base_addr) \ 3895 atomic_inc(&(sbi)->meta_count[META_CP]); \ 3896 else if (blkaddr < NM_I(sbi)->nat_blkaddr) \ 3897 atomic_inc(&(sbi)->meta_count[META_SIT]); \ 3898 else if (blkaddr < SM_I(sbi)->ssa_blkaddr) \ 3899 atomic_inc(&(sbi)->meta_count[META_NAT]); \ 3900 else if (blkaddr < SM_I(sbi)->main_blkaddr) \ 3901 atomic_inc(&(sbi)->meta_count[META_SSA]); \ 3902 } while (0) 3903 #define stat_inc_seg_type(sbi, curseg) \ 3904 ((sbi)->segment_count[(curseg)->alloc_type]++) 3905 #define stat_inc_block_count(sbi, curseg) \ 3906 ((sbi)->block_count[(curseg)->alloc_type]++) 3907 #define stat_inc_inplace_blocks(sbi) \ 3908 (atomic_inc(&(sbi)->inplace_count)) 3909 #define stat_update_max_atomic_write(inode) \ 3910 do { \ 3911 int cur = F2FS_I_SB(inode)->atomic_files; \ 3912 int max = atomic_read(&F2FS_I_SB(inode)->max_aw_cnt); \ 3913 if (cur > max) \ 3914 atomic_set(&F2FS_I_SB(inode)->max_aw_cnt, cur); \ 3915 } while (0) 3916 #define stat_inc_volatile_write(inode) \ 3917 (atomic_inc(&F2FS_I_SB(inode)->vw_cnt)) 3918 #define stat_dec_volatile_write(inode) \ 3919 (atomic_dec(&F2FS_I_SB(inode)->vw_cnt)) 3920 #define stat_update_max_volatile_write(inode) \ 3921 do { \ 3922 int cur = atomic_read(&F2FS_I_SB(inode)->vw_cnt); \ 3923 int max = atomic_read(&F2FS_I_SB(inode)->max_vw_cnt); \ 3924 if (cur > max) \ 3925 atomic_set(&F2FS_I_SB(inode)->max_vw_cnt, cur); \ 3926 } while (0) 3927 #define stat_inc_seg_count(sbi, type, gc_type) \ 3928 do { \ 3929 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 3930 si->tot_segs++; \ 3931 if ((type) == SUM_TYPE_DATA) { \ 3932 si->data_segs++; \ 3933 si->bg_data_segs += (gc_type == BG_GC) ? 1 : 0; \ 3934 } else { \ 3935 si->node_segs++; \ 3936 si->bg_node_segs += (gc_type == BG_GC) ? 1 : 0; \ 3937 } \ 3938 } while (0) 3939 3940 #define stat_inc_tot_blk_count(si, blks) \ 3941 ((si)->tot_blks += (blks)) 3942 3943 #define stat_inc_data_blk_count(sbi, blks, gc_type) \ 3944 do { \ 3945 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 3946 stat_inc_tot_blk_count(si, blks); \ 3947 si->data_blks += (blks); \ 3948 si->bg_data_blks += ((gc_type) == BG_GC) ? (blks) : 0; \ 3949 } while (0) 3950 3951 #define stat_inc_node_blk_count(sbi, blks, gc_type) \ 3952 do { \ 3953 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 3954 stat_inc_tot_blk_count(si, blks); \ 3955 si->node_blks += (blks); \ 3956 si->bg_node_blks += ((gc_type) == BG_GC) ? (blks) : 0; \ 3957 } while (0) 3958 3959 int f2fs_build_stats(struct f2fs_sb_info *sbi); 3960 void f2fs_destroy_stats(struct f2fs_sb_info *sbi); 3961 void __init f2fs_create_root_stats(void); 3962 void f2fs_destroy_root_stats(void); 3963 void f2fs_update_sit_info(struct f2fs_sb_info *sbi); 3964 #else 3965 #define stat_inc_cp_count(si) do { } while (0) 3966 #define stat_inc_bg_cp_count(si) do { } while (0) 3967 #define stat_inc_call_count(si) do { } while (0) 3968 #define stat_inc_bggc_count(si) do { } while (0) 3969 #define stat_io_skip_bggc_count(sbi) do { } while (0) 3970 #define stat_other_skip_bggc_count(sbi) do { } while (0) 3971 #define stat_inc_dirty_inode(sbi, type) do { } while (0) 3972 #define stat_dec_dirty_inode(sbi, type) do { } while (0) 3973 #define stat_inc_total_hit(sbi) do { } while (0) 3974 #define stat_inc_rbtree_node_hit(sbi) do { } while (0) 3975 #define stat_inc_largest_node_hit(sbi) do { } while (0) 3976 #define stat_inc_cached_node_hit(sbi) do { } while (0) 3977 #define stat_inc_inline_xattr(inode) do { } while (0) 3978 #define stat_dec_inline_xattr(inode) do { } while (0) 3979 #define stat_inc_inline_inode(inode) do { } while (0) 3980 #define stat_dec_inline_inode(inode) do { } while (0) 3981 #define stat_inc_inline_dir(inode) do { } while (0) 3982 #define stat_dec_inline_dir(inode) do { } while (0) 3983 #define stat_inc_compr_inode(inode) do { } while (0) 3984 #define stat_dec_compr_inode(inode) do { } while (0) 3985 #define stat_add_compr_blocks(inode, blocks) do { } while (0) 3986 #define stat_sub_compr_blocks(inode, blocks) do { } while (0) 3987 #define stat_update_max_atomic_write(inode) do { } while (0) 3988 #define stat_inc_volatile_write(inode) do { } while (0) 3989 #define stat_dec_volatile_write(inode) do { } while (0) 3990 #define stat_update_max_volatile_write(inode) do { } while (0) 3991 #define stat_inc_meta_count(sbi, blkaddr) do { } while (0) 3992 #define stat_inc_seg_type(sbi, curseg) do { } while (0) 3993 #define stat_inc_block_count(sbi, curseg) do { } while (0) 3994 #define stat_inc_inplace_blocks(sbi) do { } while (0) 3995 #define stat_inc_seg_count(sbi, type, gc_type) do { } while (0) 3996 #define stat_inc_tot_blk_count(si, blks) do { } while (0) 3997 #define stat_inc_data_blk_count(sbi, blks, gc_type) do { } while (0) 3998 #define stat_inc_node_blk_count(sbi, blks, gc_type) do { } while (0) 3999 4000 static inline int f2fs_build_stats(struct f2fs_sb_info *sbi) { return 0; } 4001 static inline void f2fs_destroy_stats(struct f2fs_sb_info *sbi) { } 4002 static inline void __init f2fs_create_root_stats(void) { } 4003 static inline void f2fs_destroy_root_stats(void) { } 4004 static inline void f2fs_update_sit_info(struct f2fs_sb_info *sbi) {} 4005 #endif 4006 4007 extern const struct file_operations f2fs_dir_operations; 4008 extern const struct file_operations f2fs_file_operations; 4009 extern const struct inode_operations f2fs_file_inode_operations; 4010 extern const struct address_space_operations f2fs_dblock_aops; 4011 extern const struct address_space_operations f2fs_node_aops; 4012 extern const struct address_space_operations f2fs_meta_aops; 4013 extern const struct inode_operations f2fs_dir_inode_operations; 4014 extern const struct inode_operations f2fs_symlink_inode_operations; 4015 extern const struct inode_operations f2fs_encrypted_symlink_inode_operations; 4016 extern const struct inode_operations f2fs_special_inode_operations; 4017 extern struct kmem_cache *f2fs_inode_entry_slab; 4018 4019 /* 4020 * inline.c 4021 */ 4022 bool f2fs_may_inline_data(struct inode *inode); 4023 bool f2fs_may_inline_dentry(struct inode *inode); 4024 void f2fs_do_read_inline_data(struct page *page, struct page *ipage); 4025 void f2fs_truncate_inline_inode(struct inode *inode, 4026 struct page *ipage, u64 from); 4027 int f2fs_read_inline_data(struct inode *inode, struct page *page); 4028 int f2fs_convert_inline_page(struct dnode_of_data *dn, struct page *page); 4029 int f2fs_convert_inline_inode(struct inode *inode); 4030 int f2fs_try_convert_inline_dir(struct inode *dir, struct dentry *dentry); 4031 int f2fs_write_inline_data(struct inode *inode, struct page *page); 4032 int f2fs_recover_inline_data(struct inode *inode, struct page *npage); 4033 struct f2fs_dir_entry *f2fs_find_in_inline_dir(struct inode *dir, 4034 const struct f2fs_filename *fname, 4035 struct page **res_page); 4036 int f2fs_make_empty_inline_dir(struct inode *inode, struct inode *parent, 4037 struct page *ipage); 4038 int f2fs_add_inline_entry(struct inode *dir, const struct f2fs_filename *fname, 4039 struct inode *inode, nid_t ino, umode_t mode); 4040 void f2fs_delete_inline_entry(struct f2fs_dir_entry *dentry, 4041 struct page *page, struct inode *dir, 4042 struct inode *inode); 4043 bool f2fs_empty_inline_dir(struct inode *dir); 4044 int f2fs_read_inline_dir(struct file *file, struct dir_context *ctx, 4045 struct fscrypt_str *fstr); 4046 int f2fs_inline_data_fiemap(struct inode *inode, 4047 struct fiemap_extent_info *fieinfo, 4048 __u64 start, __u64 len); 4049 4050 /* 4051 * shrinker.c 4052 */ 4053 unsigned long f2fs_shrink_count(struct shrinker *shrink, 4054 struct shrink_control *sc); 4055 unsigned long f2fs_shrink_scan(struct shrinker *shrink, 4056 struct shrink_control *sc); 4057 void f2fs_join_shrinker(struct f2fs_sb_info *sbi); 4058 void f2fs_leave_shrinker(struct f2fs_sb_info *sbi); 4059 4060 /* 4061 * extent_cache.c 4062 */ 4063 struct rb_entry *f2fs_lookup_rb_tree(struct rb_root_cached *root, 4064 struct rb_entry *cached_re, unsigned int ofs); 4065 struct rb_node **f2fs_lookup_rb_tree_ext(struct f2fs_sb_info *sbi, 4066 struct rb_root_cached *root, 4067 struct rb_node **parent, 4068 unsigned long long key, bool *left_most); 4069 struct rb_node **f2fs_lookup_rb_tree_for_insert(struct f2fs_sb_info *sbi, 4070 struct rb_root_cached *root, 4071 struct rb_node **parent, 4072 unsigned int ofs, bool *leftmost); 4073 struct rb_entry *f2fs_lookup_rb_tree_ret(struct rb_root_cached *root, 4074 struct rb_entry *cached_re, unsigned int ofs, 4075 struct rb_entry **prev_entry, struct rb_entry **next_entry, 4076 struct rb_node ***insert_p, struct rb_node **insert_parent, 4077 bool force, bool *leftmost); 4078 bool f2fs_check_rb_tree_consistence(struct f2fs_sb_info *sbi, 4079 struct rb_root_cached *root, bool check_key); 4080 unsigned int f2fs_shrink_extent_tree(struct f2fs_sb_info *sbi, int nr_shrink); 4081 void f2fs_init_extent_tree(struct inode *inode, struct page *ipage); 4082 void f2fs_drop_extent_tree(struct inode *inode); 4083 unsigned int f2fs_destroy_extent_node(struct inode *inode); 4084 void f2fs_destroy_extent_tree(struct inode *inode); 4085 bool f2fs_lookup_extent_cache(struct inode *inode, pgoff_t pgofs, 4086 struct extent_info *ei); 4087 void f2fs_update_extent_cache(struct dnode_of_data *dn); 4088 void f2fs_update_extent_cache_range(struct dnode_of_data *dn, 4089 pgoff_t fofs, block_t blkaddr, unsigned int len); 4090 void f2fs_init_extent_cache_info(struct f2fs_sb_info *sbi); 4091 int __init f2fs_create_extent_cache(void); 4092 void f2fs_destroy_extent_cache(void); 4093 4094 /* 4095 * sysfs.c 4096 */ 4097 #define MIN_RA_MUL 2 4098 #define MAX_RA_MUL 256 4099 4100 int __init f2fs_init_sysfs(void); 4101 void f2fs_exit_sysfs(void); 4102 int f2fs_register_sysfs(struct f2fs_sb_info *sbi); 4103 void f2fs_unregister_sysfs(struct f2fs_sb_info *sbi); 4104 4105 /* verity.c */ 4106 extern const struct fsverity_operations f2fs_verityops; 4107 4108 /* 4109 * crypto support 4110 */ 4111 static inline bool f2fs_encrypted_file(struct inode *inode) 4112 { 4113 return IS_ENCRYPTED(inode) && S_ISREG(inode->i_mode); 4114 } 4115 4116 static inline void f2fs_set_encrypted_inode(struct inode *inode) 4117 { 4118 #ifdef CONFIG_FS_ENCRYPTION 4119 file_set_encrypt(inode); 4120 f2fs_set_inode_flags(inode); 4121 #endif 4122 } 4123 4124 /* 4125 * Returns true if the reads of the inode's data need to undergo some 4126 * postprocessing step, like decryption or authenticity verification. 4127 */ 4128 static inline bool f2fs_post_read_required(struct inode *inode) 4129 { 4130 return f2fs_encrypted_file(inode) || fsverity_active(inode) || 4131 f2fs_compressed_file(inode); 4132 } 4133 4134 /* 4135 * compress.c 4136 */ 4137 #ifdef CONFIG_F2FS_FS_COMPRESSION 4138 bool f2fs_is_compressed_page(struct page *page); 4139 struct page *f2fs_compress_control_page(struct page *page); 4140 int f2fs_prepare_compress_overwrite(struct inode *inode, 4141 struct page **pagep, pgoff_t index, void **fsdata); 4142 bool f2fs_compress_write_end(struct inode *inode, void *fsdata, 4143 pgoff_t index, unsigned copied); 4144 int f2fs_truncate_partial_cluster(struct inode *inode, u64 from, bool lock); 4145 void f2fs_compress_write_end_io(struct bio *bio, struct page *page); 4146 bool f2fs_is_compress_backend_ready(struct inode *inode); 4147 int f2fs_init_compress_mempool(void); 4148 void f2fs_destroy_compress_mempool(void); 4149 void f2fs_decompress_cluster(struct decompress_io_ctx *dic); 4150 void f2fs_end_read_compressed_page(struct page *page, bool failed, 4151 block_t blkaddr); 4152 bool f2fs_cluster_is_empty(struct compress_ctx *cc); 4153 bool f2fs_cluster_can_merge_page(struct compress_ctx *cc, pgoff_t index); 4154 bool f2fs_all_cluster_page_loaded(struct compress_ctx *cc, struct pagevec *pvec, 4155 int index, int nr_pages); 4156 bool f2fs_sanity_check_cluster(struct dnode_of_data *dn); 4157 void f2fs_compress_ctx_add_page(struct compress_ctx *cc, struct page *page); 4158 int f2fs_write_multi_pages(struct compress_ctx *cc, 4159 int *submitted, 4160 struct writeback_control *wbc, 4161 enum iostat_type io_type); 4162 int f2fs_is_compressed_cluster(struct inode *inode, pgoff_t index); 4163 void f2fs_update_extent_tree_range_compressed(struct inode *inode, 4164 pgoff_t fofs, block_t blkaddr, unsigned int llen, 4165 unsigned int c_len); 4166 int f2fs_read_multi_pages(struct compress_ctx *cc, struct bio **bio_ret, 4167 unsigned nr_pages, sector_t *last_block_in_bio, 4168 bool is_readahead, bool for_write); 4169 struct decompress_io_ctx *f2fs_alloc_dic(struct compress_ctx *cc); 4170 void f2fs_decompress_end_io(struct decompress_io_ctx *dic, bool failed); 4171 void f2fs_put_page_dic(struct page *page); 4172 unsigned int f2fs_cluster_blocks_are_contiguous(struct dnode_of_data *dn); 4173 int f2fs_init_compress_ctx(struct compress_ctx *cc); 4174 void f2fs_destroy_compress_ctx(struct compress_ctx *cc, bool reuse); 4175 void f2fs_init_compress_info(struct f2fs_sb_info *sbi); 4176 int f2fs_init_compress_inode(struct f2fs_sb_info *sbi); 4177 void f2fs_destroy_compress_inode(struct f2fs_sb_info *sbi); 4178 int f2fs_init_page_array_cache(struct f2fs_sb_info *sbi); 4179 void f2fs_destroy_page_array_cache(struct f2fs_sb_info *sbi); 4180 int __init f2fs_init_compress_cache(void); 4181 void f2fs_destroy_compress_cache(void); 4182 struct address_space *COMPRESS_MAPPING(struct f2fs_sb_info *sbi); 4183 void f2fs_invalidate_compress_page(struct f2fs_sb_info *sbi, block_t blkaddr); 4184 void f2fs_cache_compressed_page(struct f2fs_sb_info *sbi, struct page *page, 4185 nid_t ino, block_t blkaddr); 4186 bool f2fs_load_compressed_page(struct f2fs_sb_info *sbi, struct page *page, 4187 block_t blkaddr); 4188 void f2fs_invalidate_compress_pages(struct f2fs_sb_info *sbi, nid_t ino); 4189 #define inc_compr_inode_stat(inode) \ 4190 do { \ 4191 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); \ 4192 sbi->compr_new_inode++; \ 4193 } while (0) 4194 #define add_compr_block_stat(inode, blocks) \ 4195 do { \ 4196 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); \ 4197 int diff = F2FS_I(inode)->i_cluster_size - blocks; \ 4198 sbi->compr_written_block += blocks; \ 4199 sbi->compr_saved_block += diff; \ 4200 } while (0) 4201 #else 4202 static inline bool f2fs_is_compressed_page(struct page *page) { return false; } 4203 static inline bool f2fs_is_compress_backend_ready(struct inode *inode) 4204 { 4205 if (!f2fs_compressed_file(inode)) 4206 return true; 4207 /* not support compression */ 4208 return false; 4209 } 4210 static inline struct page *f2fs_compress_control_page(struct page *page) 4211 { 4212 WARN_ON_ONCE(1); 4213 return ERR_PTR(-EINVAL); 4214 } 4215 static inline int f2fs_init_compress_mempool(void) { return 0; } 4216 static inline void f2fs_destroy_compress_mempool(void) { } 4217 static inline void f2fs_decompress_cluster(struct decompress_io_ctx *dic) { } 4218 static inline void f2fs_end_read_compressed_page(struct page *page, 4219 bool failed, block_t blkaddr) 4220 { 4221 WARN_ON_ONCE(1); 4222 } 4223 static inline void f2fs_put_page_dic(struct page *page) 4224 { 4225 WARN_ON_ONCE(1); 4226 } 4227 static inline unsigned int f2fs_cluster_blocks_are_contiguous(struct dnode_of_data *dn) { return 0; } 4228 static inline bool f2fs_sanity_check_cluster(struct dnode_of_data *dn) { return false; } 4229 static inline int f2fs_init_compress_inode(struct f2fs_sb_info *sbi) { return 0; } 4230 static inline void f2fs_destroy_compress_inode(struct f2fs_sb_info *sbi) { } 4231 static inline int f2fs_init_page_array_cache(struct f2fs_sb_info *sbi) { return 0; } 4232 static inline void f2fs_destroy_page_array_cache(struct f2fs_sb_info *sbi) { } 4233 static inline int __init f2fs_init_compress_cache(void) { return 0; } 4234 static inline void f2fs_destroy_compress_cache(void) { } 4235 static inline void f2fs_invalidate_compress_page(struct f2fs_sb_info *sbi, 4236 block_t blkaddr) { } 4237 static inline void f2fs_cache_compressed_page(struct f2fs_sb_info *sbi, 4238 struct page *page, nid_t ino, block_t blkaddr) { } 4239 static inline bool f2fs_load_compressed_page(struct f2fs_sb_info *sbi, 4240 struct page *page, block_t blkaddr) { return false; } 4241 static inline void f2fs_invalidate_compress_pages(struct f2fs_sb_info *sbi, 4242 nid_t ino) { } 4243 #define inc_compr_inode_stat(inode) do { } while (0) 4244 static inline void f2fs_update_extent_tree_range_compressed(struct inode *inode, 4245 pgoff_t fofs, block_t blkaddr, unsigned int llen, 4246 unsigned int c_len) { } 4247 #endif 4248 4249 static inline void set_compress_context(struct inode *inode) 4250 { 4251 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 4252 4253 F2FS_I(inode)->i_compress_algorithm = 4254 F2FS_OPTION(sbi).compress_algorithm; 4255 F2FS_I(inode)->i_log_cluster_size = 4256 F2FS_OPTION(sbi).compress_log_size; 4257 F2FS_I(inode)->i_compress_flag = 4258 F2FS_OPTION(sbi).compress_chksum ? 4259 1 << COMPRESS_CHKSUM : 0; 4260 F2FS_I(inode)->i_cluster_size = 4261 1 << F2FS_I(inode)->i_log_cluster_size; 4262 if ((F2FS_I(inode)->i_compress_algorithm == COMPRESS_LZ4 || 4263 F2FS_I(inode)->i_compress_algorithm == COMPRESS_ZSTD) && 4264 F2FS_OPTION(sbi).compress_level) 4265 F2FS_I(inode)->i_compress_flag |= 4266 F2FS_OPTION(sbi).compress_level << 4267 COMPRESS_LEVEL_OFFSET; 4268 F2FS_I(inode)->i_flags |= F2FS_COMPR_FL; 4269 set_inode_flag(inode, FI_COMPRESSED_FILE); 4270 stat_inc_compr_inode(inode); 4271 inc_compr_inode_stat(inode); 4272 f2fs_mark_inode_dirty_sync(inode, true); 4273 } 4274 4275 static inline bool f2fs_disable_compressed_file(struct inode *inode) 4276 { 4277 struct f2fs_inode_info *fi = F2FS_I(inode); 4278 4279 if (!f2fs_compressed_file(inode)) 4280 return true; 4281 if (S_ISREG(inode->i_mode) && F2FS_HAS_BLOCKS(inode)) 4282 return false; 4283 4284 fi->i_flags &= ~F2FS_COMPR_FL; 4285 stat_dec_compr_inode(inode); 4286 clear_inode_flag(inode, FI_COMPRESSED_FILE); 4287 f2fs_mark_inode_dirty_sync(inode, true); 4288 return true; 4289 } 4290 4291 #define F2FS_FEATURE_FUNCS(name, flagname) \ 4292 static inline int f2fs_sb_has_##name(struct f2fs_sb_info *sbi) \ 4293 { \ 4294 return F2FS_HAS_FEATURE(sbi, F2FS_FEATURE_##flagname); \ 4295 } 4296 4297 F2FS_FEATURE_FUNCS(encrypt, ENCRYPT); 4298 F2FS_FEATURE_FUNCS(blkzoned, BLKZONED); 4299 F2FS_FEATURE_FUNCS(extra_attr, EXTRA_ATTR); 4300 F2FS_FEATURE_FUNCS(project_quota, PRJQUOTA); 4301 F2FS_FEATURE_FUNCS(inode_chksum, INODE_CHKSUM); 4302 F2FS_FEATURE_FUNCS(flexible_inline_xattr, FLEXIBLE_INLINE_XATTR); 4303 F2FS_FEATURE_FUNCS(quota_ino, QUOTA_INO); 4304 F2FS_FEATURE_FUNCS(inode_crtime, INODE_CRTIME); 4305 F2FS_FEATURE_FUNCS(lost_found, LOST_FOUND); 4306 F2FS_FEATURE_FUNCS(verity, VERITY); 4307 F2FS_FEATURE_FUNCS(sb_chksum, SB_CHKSUM); 4308 F2FS_FEATURE_FUNCS(casefold, CASEFOLD); 4309 F2FS_FEATURE_FUNCS(compression, COMPRESSION); 4310 F2FS_FEATURE_FUNCS(readonly, RO); 4311 4312 static inline bool f2fs_may_extent_tree(struct inode *inode) 4313 { 4314 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 4315 4316 if (!test_opt(sbi, EXTENT_CACHE) || 4317 is_inode_flag_set(inode, FI_NO_EXTENT) || 4318 (is_inode_flag_set(inode, FI_COMPRESSED_FILE) && 4319 !f2fs_sb_has_readonly(sbi))) 4320 return false; 4321 4322 /* 4323 * for recovered files during mount do not create extents 4324 * if shrinker is not registered. 4325 */ 4326 if (list_empty(&sbi->s_list)) 4327 return false; 4328 4329 return S_ISREG(inode->i_mode); 4330 } 4331 4332 #ifdef CONFIG_BLK_DEV_ZONED 4333 static inline bool f2fs_blkz_is_seq(struct f2fs_sb_info *sbi, int devi, 4334 block_t blkaddr) 4335 { 4336 unsigned int zno = blkaddr >> sbi->log_blocks_per_blkz; 4337 4338 return test_bit(zno, FDEV(devi).blkz_seq); 4339 } 4340 #endif 4341 4342 static inline bool f2fs_hw_should_discard(struct f2fs_sb_info *sbi) 4343 { 4344 return f2fs_sb_has_blkzoned(sbi); 4345 } 4346 4347 static inline bool f2fs_bdev_support_discard(struct block_device *bdev) 4348 { 4349 return blk_queue_discard(bdev_get_queue(bdev)) || 4350 bdev_is_zoned(bdev); 4351 } 4352 4353 static inline bool f2fs_hw_support_discard(struct f2fs_sb_info *sbi) 4354 { 4355 int i; 4356 4357 if (!f2fs_is_multi_device(sbi)) 4358 return f2fs_bdev_support_discard(sbi->sb->s_bdev); 4359 4360 for (i = 0; i < sbi->s_ndevs; i++) 4361 if (f2fs_bdev_support_discard(FDEV(i).bdev)) 4362 return true; 4363 return false; 4364 } 4365 4366 static inline bool f2fs_realtime_discard_enable(struct f2fs_sb_info *sbi) 4367 { 4368 return (test_opt(sbi, DISCARD) && f2fs_hw_support_discard(sbi)) || 4369 f2fs_hw_should_discard(sbi); 4370 } 4371 4372 static inline bool f2fs_hw_is_readonly(struct f2fs_sb_info *sbi) 4373 { 4374 int i; 4375 4376 if (!f2fs_is_multi_device(sbi)) 4377 return bdev_read_only(sbi->sb->s_bdev); 4378 4379 for (i = 0; i < sbi->s_ndevs; i++) 4380 if (bdev_read_only(FDEV(i).bdev)) 4381 return true; 4382 return false; 4383 } 4384 4385 static inline bool f2fs_lfs_mode(struct f2fs_sb_info *sbi) 4386 { 4387 return F2FS_OPTION(sbi).fs_mode == FS_MODE_LFS; 4388 } 4389 4390 static inline bool f2fs_may_compress(struct inode *inode) 4391 { 4392 if (IS_SWAPFILE(inode) || f2fs_is_pinned_file(inode) || 4393 f2fs_is_atomic_file(inode) || 4394 f2fs_is_volatile_file(inode)) 4395 return false; 4396 return S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode); 4397 } 4398 4399 static inline void f2fs_i_compr_blocks_update(struct inode *inode, 4400 u64 blocks, bool add) 4401 { 4402 int diff = F2FS_I(inode)->i_cluster_size - blocks; 4403 struct f2fs_inode_info *fi = F2FS_I(inode); 4404 4405 /* don't update i_compr_blocks if saved blocks were released */ 4406 if (!add && !atomic_read(&fi->i_compr_blocks)) 4407 return; 4408 4409 if (add) { 4410 atomic_add(diff, &fi->i_compr_blocks); 4411 stat_add_compr_blocks(inode, diff); 4412 } else { 4413 atomic_sub(diff, &fi->i_compr_blocks); 4414 stat_sub_compr_blocks(inode, diff); 4415 } 4416 f2fs_mark_inode_dirty_sync(inode, true); 4417 } 4418 4419 static inline int block_unaligned_IO(struct inode *inode, 4420 struct kiocb *iocb, struct iov_iter *iter) 4421 { 4422 unsigned int i_blkbits = READ_ONCE(inode->i_blkbits); 4423 unsigned int blocksize_mask = (1 << i_blkbits) - 1; 4424 loff_t offset = iocb->ki_pos; 4425 unsigned long align = offset | iov_iter_alignment(iter); 4426 4427 return align & blocksize_mask; 4428 } 4429 4430 static inline bool f2fs_allow_multi_device_dio(struct f2fs_sb_info *sbi, 4431 int flag) 4432 { 4433 if (!f2fs_is_multi_device(sbi)) 4434 return false; 4435 if (flag != F2FS_GET_BLOCK_DIO) 4436 return false; 4437 return sbi->aligned_blksize; 4438 } 4439 4440 static inline bool f2fs_force_buffered_io(struct inode *inode, 4441 struct kiocb *iocb, struct iov_iter *iter) 4442 { 4443 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 4444 int rw = iov_iter_rw(iter); 4445 4446 if (f2fs_post_read_required(inode)) 4447 return true; 4448 4449 /* disallow direct IO if any of devices has unaligned blksize */ 4450 if (f2fs_is_multi_device(sbi) && !sbi->aligned_blksize) 4451 return true; 4452 /* 4453 * for blkzoned device, fallback direct IO to buffered IO, so 4454 * all IOs can be serialized by log-structured write. 4455 */ 4456 if (f2fs_sb_has_blkzoned(sbi)) 4457 return true; 4458 if (f2fs_lfs_mode(sbi) && (rw == WRITE)) { 4459 if (block_unaligned_IO(inode, iocb, iter)) 4460 return true; 4461 if (F2FS_IO_ALIGNED(sbi)) 4462 return true; 4463 } 4464 if (is_sbi_flag_set(F2FS_I_SB(inode), SBI_CP_DISABLED)) 4465 return true; 4466 4467 return false; 4468 } 4469 4470 static inline bool f2fs_need_verity(const struct inode *inode, pgoff_t idx) 4471 { 4472 return fsverity_active(inode) && 4473 idx < DIV_ROUND_UP(inode->i_size, PAGE_SIZE); 4474 } 4475 4476 #ifdef CONFIG_F2FS_FAULT_INJECTION 4477 extern void f2fs_build_fault_attr(struct f2fs_sb_info *sbi, unsigned int rate, 4478 unsigned int type); 4479 #else 4480 #define f2fs_build_fault_attr(sbi, rate, type) do { } while (0) 4481 #endif 4482 4483 static inline bool is_journalled_quota(struct f2fs_sb_info *sbi) 4484 { 4485 #ifdef CONFIG_QUOTA 4486 if (f2fs_sb_has_quota_ino(sbi)) 4487 return true; 4488 if (F2FS_OPTION(sbi).s_qf_names[USRQUOTA] || 4489 F2FS_OPTION(sbi).s_qf_names[GRPQUOTA] || 4490 F2FS_OPTION(sbi).s_qf_names[PRJQUOTA]) 4491 return true; 4492 #endif 4493 return false; 4494 } 4495 4496 static inline bool f2fs_block_unit_discard(struct f2fs_sb_info *sbi) 4497 { 4498 return F2FS_OPTION(sbi).discard_unit == DISCARD_UNIT_BLOCK; 4499 } 4500 4501 #define EFSBADCRC EBADMSG /* Bad CRC detected */ 4502 #define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */ 4503 4504 #endif /* _LINUX_F2FS_H */ 4505