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/sched/mm.h> 22 #include <linux/vmalloc.h> 23 #include <linux/bio.h> 24 #include <linux/blkdev.h> 25 #include <linux/quotaops.h> 26 #include <linux/part_stat.h> 27 #include <crypto/hash.h> 28 29 #include <linux/fscrypt.h> 30 #include <linux/fsverity.h> 31 32 struct pagevec; 33 34 #ifdef CONFIG_F2FS_CHECK_FS 35 #define f2fs_bug_on(sbi, condition) BUG_ON(condition) 36 #else 37 #define f2fs_bug_on(sbi, condition) \ 38 do { \ 39 if (WARN_ON(condition)) \ 40 set_sbi_flag(sbi, SBI_NEED_FSCK); \ 41 } while (0) 42 #endif 43 44 enum { 45 FAULT_KMALLOC, 46 FAULT_KVMALLOC, 47 FAULT_PAGE_ALLOC, 48 FAULT_PAGE_GET, 49 FAULT_ALLOC_BIO, /* it's obsolete due to bio_alloc() will never fail */ 50 FAULT_ALLOC_NID, 51 FAULT_ORPHAN, 52 FAULT_BLOCK, 53 FAULT_DIR_DEPTH, 54 FAULT_EVICT_INODE, 55 FAULT_TRUNCATE, 56 FAULT_READ_IO, 57 FAULT_CHECKPOINT, 58 FAULT_DISCARD, 59 FAULT_WRITE_IO, 60 FAULT_SLAB_ALLOC, 61 FAULT_DQUOT_INIT, 62 FAULT_LOCK_OP, 63 FAULT_MAX, 64 }; 65 66 #ifdef CONFIG_F2FS_FAULT_INJECTION 67 #define F2FS_ALL_FAULT_TYPE ((1 << FAULT_MAX) - 1) 68 69 struct f2fs_fault_info { 70 atomic_t inject_ops; 71 unsigned int inject_rate; 72 unsigned int inject_type; 73 }; 74 75 extern const char *f2fs_fault_name[FAULT_MAX]; 76 #define IS_FAULT_SET(fi, type) ((fi)->inject_type & (1 << (type))) 77 #endif 78 79 /* 80 * For mount options 81 */ 82 #define F2FS_MOUNT_DISABLE_ROLL_FORWARD 0x00000002 83 #define F2FS_MOUNT_DISCARD 0x00000004 84 #define F2FS_MOUNT_NOHEAP 0x00000008 85 #define F2FS_MOUNT_XATTR_USER 0x00000010 86 #define F2FS_MOUNT_POSIX_ACL 0x00000020 87 #define F2FS_MOUNT_DISABLE_EXT_IDENTIFY 0x00000040 88 #define F2FS_MOUNT_INLINE_XATTR 0x00000080 89 #define F2FS_MOUNT_INLINE_DATA 0x00000100 90 #define F2FS_MOUNT_INLINE_DENTRY 0x00000200 91 #define F2FS_MOUNT_FLUSH_MERGE 0x00000400 92 #define F2FS_MOUNT_NOBARRIER 0x00000800 93 #define F2FS_MOUNT_FASTBOOT 0x00001000 94 #define F2FS_MOUNT_EXTENT_CACHE 0x00002000 95 #define F2FS_MOUNT_DATA_FLUSH 0x00008000 96 #define F2FS_MOUNT_FAULT_INJECTION 0x00010000 97 #define F2FS_MOUNT_USRQUOTA 0x00080000 98 #define F2FS_MOUNT_GRPQUOTA 0x00100000 99 #define F2FS_MOUNT_PRJQUOTA 0x00200000 100 #define F2FS_MOUNT_QUOTA 0x00400000 101 #define F2FS_MOUNT_INLINE_XATTR_SIZE 0x00800000 102 #define F2FS_MOUNT_RESERVE_ROOT 0x01000000 103 #define F2FS_MOUNT_DISABLE_CHECKPOINT 0x02000000 104 #define F2FS_MOUNT_NORECOVERY 0x04000000 105 #define F2FS_MOUNT_ATGC 0x08000000 106 #define F2FS_MOUNT_MERGE_CHECKPOINT 0x10000000 107 #define F2FS_MOUNT_GC_MERGE 0x20000000 108 #define F2FS_MOUNT_COMPRESS_CACHE 0x40000000 109 110 #define F2FS_OPTION(sbi) ((sbi)->mount_opt) 111 #define clear_opt(sbi, option) (F2FS_OPTION(sbi).opt &= ~F2FS_MOUNT_##option) 112 #define set_opt(sbi, option) (F2FS_OPTION(sbi).opt |= F2FS_MOUNT_##option) 113 #define test_opt(sbi, option) (F2FS_OPTION(sbi).opt & F2FS_MOUNT_##option) 114 115 #define ver_after(a, b) (typecheck(unsigned long long, a) && \ 116 typecheck(unsigned long long, b) && \ 117 ((long long)((a) - (b)) > 0)) 118 119 typedef u32 block_t; /* 120 * should not change u32, since it is the on-disk block 121 * address format, __le32. 122 */ 123 typedef u32 nid_t; 124 125 #define COMPRESS_EXT_NUM 16 126 127 /* 128 * An implementation of an rwsem that is explicitly unfair to readers. This 129 * prevents priority inversion when a low-priority reader acquires the read lock 130 * while sleeping on the write lock but the write lock is needed by 131 * higher-priority clients. 132 */ 133 134 struct f2fs_rwsem { 135 struct rw_semaphore internal_rwsem; 136 #ifdef CONFIG_F2FS_UNFAIR_RWSEM 137 wait_queue_head_t read_waiters; 138 #endif 139 }; 140 141 struct f2fs_mount_info { 142 unsigned int opt; 143 int write_io_size_bits; /* Write IO size bits */ 144 block_t root_reserved_blocks; /* root reserved blocks */ 145 kuid_t s_resuid; /* reserved blocks for uid */ 146 kgid_t s_resgid; /* reserved blocks for gid */ 147 int active_logs; /* # of active logs */ 148 int inline_xattr_size; /* inline xattr size */ 149 #ifdef CONFIG_F2FS_FAULT_INJECTION 150 struct f2fs_fault_info fault_info; /* For fault injection */ 151 #endif 152 #ifdef CONFIG_QUOTA 153 /* Names of quota files with journalled quota */ 154 char *s_qf_names[MAXQUOTAS]; 155 int s_jquota_fmt; /* Format of quota to use */ 156 #endif 157 /* For which write hints are passed down to block layer */ 158 int alloc_mode; /* segment allocation policy */ 159 int fsync_mode; /* fsync policy */ 160 int fs_mode; /* fs mode: LFS or ADAPTIVE */ 161 int bggc_mode; /* bggc mode: off, on or sync */ 162 int memory_mode; /* memory mode */ 163 int discard_unit; /* 164 * discard command's offset/size should 165 * be aligned to this unit: block, 166 * segment or section 167 */ 168 struct fscrypt_dummy_policy dummy_enc_policy; /* test dummy encryption */ 169 block_t unusable_cap_perc; /* percentage for cap */ 170 block_t unusable_cap; /* Amount of space allowed to be 171 * unusable when disabling checkpoint 172 */ 173 174 /* For compression */ 175 unsigned char compress_algorithm; /* algorithm type */ 176 unsigned char compress_log_size; /* cluster log size */ 177 unsigned char compress_level; /* compress level */ 178 bool compress_chksum; /* compressed data chksum */ 179 unsigned char compress_ext_cnt; /* extension count */ 180 unsigned char nocompress_ext_cnt; /* nocompress extension count */ 181 int compress_mode; /* compression mode */ 182 unsigned char extensions[COMPRESS_EXT_NUM][F2FS_EXTENSION_LEN]; /* extensions */ 183 unsigned char noextensions[COMPRESS_EXT_NUM][F2FS_EXTENSION_LEN]; /* extensions */ 184 }; 185 186 #define F2FS_FEATURE_ENCRYPT 0x0001 187 #define F2FS_FEATURE_BLKZONED 0x0002 188 #define F2FS_FEATURE_ATOMIC_WRITE 0x0004 189 #define F2FS_FEATURE_EXTRA_ATTR 0x0008 190 #define F2FS_FEATURE_PRJQUOTA 0x0010 191 #define F2FS_FEATURE_INODE_CHKSUM 0x0020 192 #define F2FS_FEATURE_FLEXIBLE_INLINE_XATTR 0x0040 193 #define F2FS_FEATURE_QUOTA_INO 0x0080 194 #define F2FS_FEATURE_INODE_CRTIME 0x0100 195 #define F2FS_FEATURE_LOST_FOUND 0x0200 196 #define F2FS_FEATURE_VERITY 0x0400 197 #define F2FS_FEATURE_SB_CHKSUM 0x0800 198 #define F2FS_FEATURE_CASEFOLD 0x1000 199 #define F2FS_FEATURE_COMPRESSION 0x2000 200 #define F2FS_FEATURE_RO 0x4000 201 202 #define __F2FS_HAS_FEATURE(raw_super, mask) \ 203 ((raw_super->feature & cpu_to_le32(mask)) != 0) 204 #define F2FS_HAS_FEATURE(sbi, mask) __F2FS_HAS_FEATURE(sbi->raw_super, mask) 205 #define F2FS_SET_FEATURE(sbi, mask) \ 206 (sbi->raw_super->feature |= cpu_to_le32(mask)) 207 #define F2FS_CLEAR_FEATURE(sbi, mask) \ 208 (sbi->raw_super->feature &= ~cpu_to_le32(mask)) 209 210 /* 211 * Default values for user and/or group using reserved blocks 212 */ 213 #define F2FS_DEF_RESUID 0 214 #define F2FS_DEF_RESGID 0 215 216 /* 217 * For checkpoint manager 218 */ 219 enum { 220 NAT_BITMAP, 221 SIT_BITMAP 222 }; 223 224 #define CP_UMOUNT 0x00000001 225 #define CP_FASTBOOT 0x00000002 226 #define CP_SYNC 0x00000004 227 #define CP_RECOVERY 0x00000008 228 #define CP_DISCARD 0x00000010 229 #define CP_TRIMMED 0x00000020 230 #define CP_PAUSE 0x00000040 231 #define CP_RESIZE 0x00000080 232 233 #define DEF_MAX_DISCARD_REQUEST 8 /* issue 8 discards per round */ 234 #define DEF_MIN_DISCARD_ISSUE_TIME 50 /* 50 ms, if exists */ 235 #define DEF_MID_DISCARD_ISSUE_TIME 500 /* 500 ms, if device busy */ 236 #define DEF_MAX_DISCARD_ISSUE_TIME 60000 /* 60 s, if no candidates */ 237 #define DEF_DISCARD_URGENT_UTIL 80 /* do more discard over 80% */ 238 #define DEF_CP_INTERVAL 60 /* 60 secs */ 239 #define DEF_IDLE_INTERVAL 5 /* 5 secs */ 240 #define DEF_DISABLE_INTERVAL 5 /* 5 secs */ 241 #define DEF_DISABLE_QUICK_INTERVAL 1 /* 1 secs */ 242 #define DEF_UMOUNT_DISCARD_TIMEOUT 5 /* 5 secs */ 243 244 struct cp_control { 245 int reason; 246 __u64 trim_start; 247 __u64 trim_end; 248 __u64 trim_minlen; 249 }; 250 251 /* 252 * indicate meta/data type 253 */ 254 enum { 255 META_CP, 256 META_NAT, 257 META_SIT, 258 META_SSA, 259 META_MAX, 260 META_POR, 261 DATA_GENERIC, /* check range only */ 262 DATA_GENERIC_ENHANCE, /* strong check on range and segment bitmap */ 263 DATA_GENERIC_ENHANCE_READ, /* 264 * strong check on range and segment 265 * bitmap but no warning due to race 266 * condition of read on truncated area 267 * by extent_cache 268 */ 269 META_GENERIC, 270 }; 271 272 /* for the list of ino */ 273 enum { 274 ORPHAN_INO, /* for orphan ino list */ 275 APPEND_INO, /* for append ino list */ 276 UPDATE_INO, /* for update ino list */ 277 TRANS_DIR_INO, /* for trasactions dir ino list */ 278 FLUSH_INO, /* for multiple device flushing */ 279 MAX_INO_ENTRY, /* max. list */ 280 }; 281 282 struct ino_entry { 283 struct list_head list; /* list head */ 284 nid_t ino; /* inode number */ 285 unsigned int dirty_device; /* dirty device bitmap */ 286 }; 287 288 /* for the list of inodes to be GCed */ 289 struct inode_entry { 290 struct list_head list; /* list head */ 291 struct inode *inode; /* vfs inode pointer */ 292 }; 293 294 struct fsync_node_entry { 295 struct list_head list; /* list head */ 296 struct page *page; /* warm node page pointer */ 297 unsigned int seq_id; /* sequence id */ 298 }; 299 300 struct ckpt_req { 301 struct completion wait; /* completion for checkpoint done */ 302 struct llist_node llnode; /* llist_node to be linked in wait queue */ 303 int ret; /* return code of checkpoint */ 304 ktime_t queue_time; /* request queued time */ 305 }; 306 307 struct ckpt_req_control { 308 struct task_struct *f2fs_issue_ckpt; /* checkpoint task */ 309 int ckpt_thread_ioprio; /* checkpoint merge thread ioprio */ 310 wait_queue_head_t ckpt_wait_queue; /* waiting queue for wake-up */ 311 atomic_t issued_ckpt; /* # of actually issued ckpts */ 312 atomic_t total_ckpt; /* # of total ckpts */ 313 atomic_t queued_ckpt; /* # of queued ckpts */ 314 struct llist_head issue_list; /* list for command issue */ 315 spinlock_t stat_lock; /* lock for below checkpoint time stats */ 316 unsigned int cur_time; /* cur wait time in msec for currently issued checkpoint */ 317 unsigned int peak_time; /* peak wait time in msec until now */ 318 }; 319 320 /* for the bitmap indicate blocks to be discarded */ 321 struct discard_entry { 322 struct list_head list; /* list head */ 323 block_t start_blkaddr; /* start blockaddr of current segment */ 324 unsigned char discard_map[SIT_VBLOCK_MAP_SIZE]; /* segment discard bitmap */ 325 }; 326 327 /* default discard granularity of inner discard thread, unit: block count */ 328 #define DEFAULT_DISCARD_GRANULARITY 16 329 330 /* max discard pend list number */ 331 #define MAX_PLIST_NUM 512 332 #define plist_idx(blk_num) ((blk_num) >= MAX_PLIST_NUM ? \ 333 (MAX_PLIST_NUM - 1) : ((blk_num) - 1)) 334 335 enum { 336 D_PREP, /* initial */ 337 D_PARTIAL, /* partially submitted */ 338 D_SUBMIT, /* all submitted */ 339 D_DONE, /* finished */ 340 }; 341 342 struct discard_info { 343 block_t lstart; /* logical start address */ 344 block_t len; /* length */ 345 block_t start; /* actual start address in dev */ 346 }; 347 348 struct discard_cmd { 349 struct rb_node rb_node; /* rb node located in rb-tree */ 350 union { 351 struct { 352 block_t lstart; /* logical start address */ 353 block_t len; /* length */ 354 block_t start; /* actual start address in dev */ 355 }; 356 struct discard_info di; /* discard info */ 357 358 }; 359 struct list_head list; /* command list */ 360 struct completion wait; /* compleation */ 361 struct block_device *bdev; /* bdev */ 362 unsigned short ref; /* reference count */ 363 unsigned char state; /* state */ 364 unsigned char queued; /* queued discard */ 365 int error; /* bio error */ 366 spinlock_t lock; /* for state/bio_ref updating */ 367 unsigned short bio_ref; /* bio reference count */ 368 }; 369 370 enum { 371 DPOLICY_BG, 372 DPOLICY_FORCE, 373 DPOLICY_FSTRIM, 374 DPOLICY_UMOUNT, 375 MAX_DPOLICY, 376 }; 377 378 struct discard_policy { 379 int type; /* type of discard */ 380 unsigned int min_interval; /* used for candidates exist */ 381 unsigned int mid_interval; /* used for device busy */ 382 unsigned int max_interval; /* used for candidates not exist */ 383 unsigned int max_requests; /* # of discards issued per round */ 384 unsigned int io_aware_gran; /* minimum granularity discard not be aware of I/O */ 385 bool io_aware; /* issue discard in idle time */ 386 bool sync; /* submit discard with REQ_SYNC flag */ 387 bool ordered; /* issue discard by lba order */ 388 bool timeout; /* discard timeout for put_super */ 389 unsigned int granularity; /* discard granularity */ 390 }; 391 392 struct discard_cmd_control { 393 struct task_struct *f2fs_issue_discard; /* discard thread */ 394 struct list_head entry_list; /* 4KB discard entry list */ 395 struct list_head pend_list[MAX_PLIST_NUM];/* store pending entries */ 396 struct list_head wait_list; /* store on-flushing entries */ 397 struct list_head fstrim_list; /* in-flight discard from fstrim */ 398 wait_queue_head_t discard_wait_queue; /* waiting queue for wake-up */ 399 unsigned int discard_wake; /* to wake up discard thread */ 400 struct mutex cmd_lock; 401 unsigned int nr_discards; /* # of discards in the list */ 402 unsigned int max_discards; /* max. discards to be issued */ 403 unsigned int max_discard_request; /* max. discard request per round */ 404 unsigned int min_discard_issue_time; /* min. interval between discard issue */ 405 unsigned int mid_discard_issue_time; /* mid. interval between discard issue */ 406 unsigned int max_discard_issue_time; /* max. interval between discard issue */ 407 unsigned int discard_granularity; /* discard granularity */ 408 unsigned int undiscard_blks; /* # of undiscard blocks */ 409 unsigned int next_pos; /* next discard position */ 410 atomic_t issued_discard; /* # of issued discard */ 411 atomic_t queued_discard; /* # of queued discard */ 412 atomic_t discard_cmd_cnt; /* # of cached cmd count */ 413 struct rb_root_cached root; /* root of discard rb-tree */ 414 bool rbtree_check; /* config for consistence check */ 415 }; 416 417 /* for the list of fsync inodes, used only during recovery */ 418 struct fsync_inode_entry { 419 struct list_head list; /* list head */ 420 struct inode *inode; /* vfs inode pointer */ 421 block_t blkaddr; /* block address locating the last fsync */ 422 block_t last_dentry; /* block address locating the last dentry */ 423 }; 424 425 #define nats_in_cursum(jnl) (le16_to_cpu((jnl)->n_nats)) 426 #define sits_in_cursum(jnl) (le16_to_cpu((jnl)->n_sits)) 427 428 #define nat_in_journal(jnl, i) ((jnl)->nat_j.entries[i].ne) 429 #define nid_in_journal(jnl, i) ((jnl)->nat_j.entries[i].nid) 430 #define sit_in_journal(jnl, i) ((jnl)->sit_j.entries[i].se) 431 #define segno_in_journal(jnl, i) ((jnl)->sit_j.entries[i].segno) 432 433 #define MAX_NAT_JENTRIES(jnl) (NAT_JOURNAL_ENTRIES - nats_in_cursum(jnl)) 434 #define MAX_SIT_JENTRIES(jnl) (SIT_JOURNAL_ENTRIES - sits_in_cursum(jnl)) 435 436 static inline int update_nats_in_cursum(struct f2fs_journal *journal, int i) 437 { 438 int before = nats_in_cursum(journal); 439 440 journal->n_nats = cpu_to_le16(before + i); 441 return before; 442 } 443 444 static inline int update_sits_in_cursum(struct f2fs_journal *journal, int i) 445 { 446 int before = sits_in_cursum(journal); 447 448 journal->n_sits = cpu_to_le16(before + i); 449 return before; 450 } 451 452 static inline bool __has_cursum_space(struct f2fs_journal *journal, 453 int size, int type) 454 { 455 if (type == NAT_JOURNAL) 456 return size <= MAX_NAT_JENTRIES(journal); 457 return size <= MAX_SIT_JENTRIES(journal); 458 } 459 460 /* for inline stuff */ 461 #define DEF_INLINE_RESERVED_SIZE 1 462 static inline int get_extra_isize(struct inode *inode); 463 static inline int get_inline_xattr_addrs(struct inode *inode); 464 #define MAX_INLINE_DATA(inode) (sizeof(__le32) * \ 465 (CUR_ADDRS_PER_INODE(inode) - \ 466 get_inline_xattr_addrs(inode) - \ 467 DEF_INLINE_RESERVED_SIZE)) 468 469 /* for inline dir */ 470 #define NR_INLINE_DENTRY(inode) (MAX_INLINE_DATA(inode) * BITS_PER_BYTE / \ 471 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \ 472 BITS_PER_BYTE + 1)) 473 #define INLINE_DENTRY_BITMAP_SIZE(inode) \ 474 DIV_ROUND_UP(NR_INLINE_DENTRY(inode), BITS_PER_BYTE) 475 #define INLINE_RESERVED_SIZE(inode) (MAX_INLINE_DATA(inode) - \ 476 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \ 477 NR_INLINE_DENTRY(inode) + \ 478 INLINE_DENTRY_BITMAP_SIZE(inode))) 479 480 /* 481 * For INODE and NODE manager 482 */ 483 /* for directory operations */ 484 485 struct f2fs_filename { 486 /* 487 * The filename the user specified. This is NULL for some 488 * filesystem-internal operations, e.g. converting an inline directory 489 * to a non-inline one, or roll-forward recovering an encrypted dentry. 490 */ 491 const struct qstr *usr_fname; 492 493 /* 494 * The on-disk filename. For encrypted directories, this is encrypted. 495 * This may be NULL for lookups in an encrypted dir without the key. 496 */ 497 struct fscrypt_str disk_name; 498 499 /* The dirhash of this filename */ 500 f2fs_hash_t hash; 501 502 #ifdef CONFIG_FS_ENCRYPTION 503 /* 504 * For lookups in encrypted directories: either the buffer backing 505 * disk_name, or a buffer that holds the decoded no-key name. 506 */ 507 struct fscrypt_str crypto_buf; 508 #endif 509 #if IS_ENABLED(CONFIG_UNICODE) 510 /* 511 * For casefolded directories: the casefolded name, but it's left NULL 512 * if the original name is not valid Unicode, if the original name is 513 * "." or "..", if the directory is both casefolded and encrypted and 514 * its encryption key is unavailable, or if the filesystem is doing an 515 * internal operation where usr_fname is also NULL. In all these cases 516 * we fall back to treating the name as an opaque byte sequence. 517 */ 518 struct fscrypt_str cf_name; 519 #endif 520 }; 521 522 struct f2fs_dentry_ptr { 523 struct inode *inode; 524 void *bitmap; 525 struct f2fs_dir_entry *dentry; 526 __u8 (*filename)[F2FS_SLOT_LEN]; 527 int max; 528 int nr_bitmap; 529 }; 530 531 static inline void make_dentry_ptr_block(struct inode *inode, 532 struct f2fs_dentry_ptr *d, struct f2fs_dentry_block *t) 533 { 534 d->inode = inode; 535 d->max = NR_DENTRY_IN_BLOCK; 536 d->nr_bitmap = SIZE_OF_DENTRY_BITMAP; 537 d->bitmap = t->dentry_bitmap; 538 d->dentry = t->dentry; 539 d->filename = t->filename; 540 } 541 542 static inline void make_dentry_ptr_inline(struct inode *inode, 543 struct f2fs_dentry_ptr *d, void *t) 544 { 545 int entry_cnt = NR_INLINE_DENTRY(inode); 546 int bitmap_size = INLINE_DENTRY_BITMAP_SIZE(inode); 547 int reserved_size = INLINE_RESERVED_SIZE(inode); 548 549 d->inode = inode; 550 d->max = entry_cnt; 551 d->nr_bitmap = bitmap_size; 552 d->bitmap = t; 553 d->dentry = t + bitmap_size + reserved_size; 554 d->filename = t + bitmap_size + reserved_size + 555 SIZE_OF_DIR_ENTRY * entry_cnt; 556 } 557 558 /* 559 * XATTR_NODE_OFFSET stores xattrs to one node block per file keeping -1 560 * as its node offset to distinguish from index node blocks. 561 * But some bits are used to mark the node block. 562 */ 563 #define XATTR_NODE_OFFSET ((((unsigned int)-1) << OFFSET_BIT_SHIFT) \ 564 >> OFFSET_BIT_SHIFT) 565 enum { 566 ALLOC_NODE, /* allocate a new node page if needed */ 567 LOOKUP_NODE, /* look up a node without readahead */ 568 LOOKUP_NODE_RA, /* 569 * look up a node with readahead called 570 * by get_data_block. 571 */ 572 }; 573 574 #define DEFAULT_RETRY_IO_COUNT 8 /* maximum retry read IO or flush count */ 575 576 /* congestion wait timeout value, default: 20ms */ 577 #define DEFAULT_IO_TIMEOUT (msecs_to_jiffies(20)) 578 579 /* maximum retry quota flush count */ 580 #define DEFAULT_RETRY_QUOTA_FLUSH_COUNT 8 581 582 /* maximum retry of EIO'ed page */ 583 #define MAX_RETRY_PAGE_EIO 100 584 585 #define F2FS_LINK_MAX 0xffffffff /* maximum link count per file */ 586 587 #define MAX_DIR_RA_PAGES 4 /* maximum ra pages of dir */ 588 589 /* dirty segments threshold for triggering CP */ 590 #define DEFAULT_DIRTY_THRESHOLD 4 591 592 /* for in-memory extent cache entry */ 593 #define F2FS_MIN_EXTENT_LEN 64 /* minimum extent length */ 594 595 /* number of extent info in extent cache we try to shrink */ 596 #define EXTENT_CACHE_SHRINK_NUMBER 128 597 598 #define RECOVERY_MAX_RA_BLOCKS BIO_MAX_VECS 599 #define RECOVERY_MIN_RA_BLOCKS 1 600 601 #define F2FS_ONSTACK_PAGES 16 /* nr of onstack pages */ 602 603 struct rb_entry { 604 struct rb_node rb_node; /* rb node located in rb-tree */ 605 union { 606 struct { 607 unsigned int ofs; /* start offset of the entry */ 608 unsigned int len; /* length of the entry */ 609 }; 610 unsigned long long key; /* 64-bits key */ 611 } __packed; 612 }; 613 614 struct extent_info { 615 unsigned int fofs; /* start offset in a file */ 616 unsigned int len; /* length of the extent */ 617 u32 blk; /* start block address of the extent */ 618 #ifdef CONFIG_F2FS_FS_COMPRESSION 619 unsigned int c_len; /* physical extent length of compressed blocks */ 620 #endif 621 }; 622 623 struct extent_node { 624 struct rb_node rb_node; /* rb node located in rb-tree */ 625 struct extent_info ei; /* extent info */ 626 struct list_head list; /* node in global extent list of sbi */ 627 struct extent_tree *et; /* extent tree pointer */ 628 }; 629 630 struct extent_tree { 631 nid_t ino; /* inode number */ 632 struct rb_root_cached root; /* root of extent info rb-tree */ 633 struct extent_node *cached_en; /* recently accessed extent node */ 634 struct extent_info largest; /* largested extent info */ 635 struct list_head list; /* to be used by sbi->zombie_list */ 636 rwlock_t lock; /* protect extent info rb-tree */ 637 atomic_t node_cnt; /* # of extent node in rb-tree*/ 638 bool largest_updated; /* largest extent updated */ 639 }; 640 641 /* 642 * This structure is taken from ext4_map_blocks. 643 * 644 * Note that, however, f2fs uses NEW and MAPPED flags for f2fs_map_blocks(). 645 */ 646 #define F2FS_MAP_NEW (1 << BH_New) 647 #define F2FS_MAP_MAPPED (1 << BH_Mapped) 648 #define F2FS_MAP_UNWRITTEN (1 << BH_Unwritten) 649 #define F2FS_MAP_FLAGS (F2FS_MAP_NEW | F2FS_MAP_MAPPED |\ 650 F2FS_MAP_UNWRITTEN) 651 652 struct f2fs_map_blocks { 653 struct block_device *m_bdev; /* for multi-device dio */ 654 block_t m_pblk; 655 block_t m_lblk; 656 unsigned int m_len; 657 unsigned int m_flags; 658 pgoff_t *m_next_pgofs; /* point next possible non-hole pgofs */ 659 pgoff_t *m_next_extent; /* point to next possible extent */ 660 int m_seg_type; 661 bool m_may_create; /* indicate it is from write path */ 662 bool m_multidev_dio; /* indicate it allows multi-device dio */ 663 }; 664 665 /* for flag in get_data_block */ 666 enum { 667 F2FS_GET_BLOCK_DEFAULT, 668 F2FS_GET_BLOCK_FIEMAP, 669 F2FS_GET_BLOCK_BMAP, 670 F2FS_GET_BLOCK_DIO, 671 F2FS_GET_BLOCK_PRE_DIO, 672 F2FS_GET_BLOCK_PRE_AIO, 673 F2FS_GET_BLOCK_PRECACHE, 674 }; 675 676 /* 677 * i_advise uses FADVISE_XXX_BIT. We can add additional hints later. 678 */ 679 #define FADVISE_COLD_BIT 0x01 680 #define FADVISE_LOST_PINO_BIT 0x02 681 #define FADVISE_ENCRYPT_BIT 0x04 682 #define FADVISE_ENC_NAME_BIT 0x08 683 #define FADVISE_KEEP_SIZE_BIT 0x10 684 #define FADVISE_HOT_BIT 0x20 685 #define FADVISE_VERITY_BIT 0x40 686 #define FADVISE_TRUNC_BIT 0x80 687 688 #define FADVISE_MODIFIABLE_BITS (FADVISE_COLD_BIT | FADVISE_HOT_BIT) 689 690 #define file_is_cold(inode) is_file(inode, FADVISE_COLD_BIT) 691 #define file_set_cold(inode) set_file(inode, FADVISE_COLD_BIT) 692 #define file_clear_cold(inode) clear_file(inode, FADVISE_COLD_BIT) 693 694 #define file_wrong_pino(inode) is_file(inode, FADVISE_LOST_PINO_BIT) 695 #define file_lost_pino(inode) set_file(inode, FADVISE_LOST_PINO_BIT) 696 #define file_got_pino(inode) clear_file(inode, FADVISE_LOST_PINO_BIT) 697 698 #define file_is_encrypt(inode) is_file(inode, FADVISE_ENCRYPT_BIT) 699 #define file_set_encrypt(inode) set_file(inode, FADVISE_ENCRYPT_BIT) 700 701 #define file_enc_name(inode) is_file(inode, FADVISE_ENC_NAME_BIT) 702 #define file_set_enc_name(inode) set_file(inode, FADVISE_ENC_NAME_BIT) 703 704 #define file_keep_isize(inode) is_file(inode, FADVISE_KEEP_SIZE_BIT) 705 #define file_set_keep_isize(inode) set_file(inode, FADVISE_KEEP_SIZE_BIT) 706 707 #define file_is_hot(inode) is_file(inode, FADVISE_HOT_BIT) 708 #define file_set_hot(inode) set_file(inode, FADVISE_HOT_BIT) 709 #define file_clear_hot(inode) clear_file(inode, FADVISE_HOT_BIT) 710 711 #define file_is_verity(inode) is_file(inode, FADVISE_VERITY_BIT) 712 #define file_set_verity(inode) set_file(inode, FADVISE_VERITY_BIT) 713 714 #define file_should_truncate(inode) is_file(inode, FADVISE_TRUNC_BIT) 715 #define file_need_truncate(inode) set_file(inode, FADVISE_TRUNC_BIT) 716 #define file_dont_truncate(inode) clear_file(inode, FADVISE_TRUNC_BIT) 717 718 #define DEF_DIR_LEVEL 0 719 720 enum { 721 GC_FAILURE_PIN, 722 MAX_GC_FAILURE 723 }; 724 725 /* used for f2fs_inode_info->flags */ 726 enum { 727 FI_NEW_INODE, /* indicate newly allocated inode */ 728 FI_DIRTY_INODE, /* indicate inode is dirty or not */ 729 FI_AUTO_RECOVER, /* indicate inode is recoverable */ 730 FI_DIRTY_DIR, /* indicate directory has dirty pages */ 731 FI_INC_LINK, /* need to increment i_nlink */ 732 FI_ACL_MODE, /* indicate acl mode */ 733 FI_NO_ALLOC, /* should not allocate any blocks */ 734 FI_FREE_NID, /* free allocated nide */ 735 FI_NO_EXTENT, /* not to use the extent cache */ 736 FI_INLINE_XATTR, /* used for inline xattr */ 737 FI_INLINE_DATA, /* used for inline data*/ 738 FI_INLINE_DENTRY, /* used for inline dentry */ 739 FI_APPEND_WRITE, /* inode has appended data */ 740 FI_UPDATE_WRITE, /* inode has in-place-update data */ 741 FI_NEED_IPU, /* used for ipu per file */ 742 FI_ATOMIC_FILE, /* indicate atomic file */ 743 FI_FIRST_BLOCK_WRITTEN, /* indicate #0 data block was written */ 744 FI_DROP_CACHE, /* drop dirty page cache */ 745 FI_DATA_EXIST, /* indicate data exists */ 746 FI_INLINE_DOTS, /* indicate inline dot dentries */ 747 FI_SKIP_WRITES, /* should skip data page writeback */ 748 FI_OPU_WRITE, /* used for opu per file */ 749 FI_DIRTY_FILE, /* indicate regular/symlink has dirty pages */ 750 FI_PREALLOCATED_ALL, /* all blocks for write were preallocated */ 751 FI_HOT_DATA, /* indicate file is hot */ 752 FI_EXTRA_ATTR, /* indicate file has extra attribute */ 753 FI_PROJ_INHERIT, /* indicate file inherits projectid */ 754 FI_PIN_FILE, /* indicate file should not be gced */ 755 FI_VERITY_IN_PROGRESS, /* building fs-verity Merkle tree */ 756 FI_COMPRESSED_FILE, /* indicate file's data can be compressed */ 757 FI_COMPRESS_CORRUPT, /* indicate compressed cluster is corrupted */ 758 FI_MMAP_FILE, /* indicate file was mmapped */ 759 FI_ENABLE_COMPRESS, /* enable compression in "user" compression mode */ 760 FI_COMPRESS_RELEASED, /* compressed blocks were released */ 761 FI_ALIGNED_WRITE, /* enable aligned write */ 762 FI_COW_FILE, /* indicate COW file */ 763 FI_MAX, /* max flag, never be used */ 764 }; 765 766 struct f2fs_inode_info { 767 struct inode vfs_inode; /* serve a vfs inode */ 768 unsigned long i_flags; /* keep an inode flags for ioctl */ 769 unsigned char i_advise; /* use to give file attribute hints */ 770 unsigned char i_dir_level; /* use for dentry level for large dir */ 771 unsigned int i_current_depth; /* only for directory depth */ 772 /* for gc failure statistic */ 773 unsigned int i_gc_failures[MAX_GC_FAILURE]; 774 unsigned int i_pino; /* parent inode number */ 775 umode_t i_acl_mode; /* keep file acl mode temporarily */ 776 777 /* Use below internally in f2fs*/ 778 unsigned long flags[BITS_TO_LONGS(FI_MAX)]; /* use to pass per-file flags */ 779 struct f2fs_rwsem i_sem; /* protect fi info */ 780 atomic_t dirty_pages; /* # of dirty pages */ 781 f2fs_hash_t chash; /* hash value of given file name */ 782 unsigned int clevel; /* maximum level of given file name */ 783 struct task_struct *task; /* lookup and create consistency */ 784 struct task_struct *cp_task; /* separate cp/wb IO stats*/ 785 nid_t i_xattr_nid; /* node id that contains xattrs */ 786 loff_t last_disk_size; /* lastly written file size */ 787 spinlock_t i_size_lock; /* protect last_disk_size */ 788 789 #ifdef CONFIG_QUOTA 790 struct dquot *i_dquot[MAXQUOTAS]; 791 792 /* quota space reservation, managed internally by quota code */ 793 qsize_t i_reserved_quota; 794 #endif 795 struct list_head dirty_list; /* dirty list for dirs and files */ 796 struct list_head gdirty_list; /* linked in global dirty list */ 797 struct task_struct *atomic_write_task; /* store atomic write task */ 798 struct extent_tree *extent_tree; /* cached extent_tree entry */ 799 struct inode *cow_inode; /* copy-on-write inode for atomic write */ 800 801 /* avoid racing between foreground op and gc */ 802 struct f2fs_rwsem i_gc_rwsem[2]; 803 struct f2fs_rwsem i_xattr_sem; /* avoid racing between reading and changing EAs */ 804 805 int i_extra_isize; /* size of extra space located in i_addr */ 806 kprojid_t i_projid; /* id for project quota */ 807 int i_inline_xattr_size; /* inline xattr size */ 808 struct timespec64 i_crtime; /* inode creation time */ 809 struct timespec64 i_disk_time[4];/* inode disk times */ 810 811 /* for file compress */ 812 atomic_t i_compr_blocks; /* # of compressed blocks */ 813 unsigned char i_compress_algorithm; /* algorithm type */ 814 unsigned char i_log_cluster_size; /* log of cluster size */ 815 unsigned char i_compress_level; /* compress level (lz4hc,zstd) */ 816 unsigned short i_compress_flag; /* compress flag */ 817 unsigned int i_cluster_size; /* cluster size */ 818 819 unsigned int atomic_write_cnt; 820 }; 821 822 static inline void get_extent_info(struct extent_info *ext, 823 struct f2fs_extent *i_ext) 824 { 825 ext->fofs = le32_to_cpu(i_ext->fofs); 826 ext->blk = le32_to_cpu(i_ext->blk); 827 ext->len = le32_to_cpu(i_ext->len); 828 } 829 830 static inline void set_raw_extent(struct extent_info *ext, 831 struct f2fs_extent *i_ext) 832 { 833 i_ext->fofs = cpu_to_le32(ext->fofs); 834 i_ext->blk = cpu_to_le32(ext->blk); 835 i_ext->len = cpu_to_le32(ext->len); 836 } 837 838 static inline void set_extent_info(struct extent_info *ei, unsigned int fofs, 839 u32 blk, unsigned int len) 840 { 841 ei->fofs = fofs; 842 ei->blk = blk; 843 ei->len = len; 844 #ifdef CONFIG_F2FS_FS_COMPRESSION 845 ei->c_len = 0; 846 #endif 847 } 848 849 static inline bool __is_discard_mergeable(struct discard_info *back, 850 struct discard_info *front, unsigned int max_len) 851 { 852 return (back->lstart + back->len == front->lstart) && 853 (back->len + front->len <= max_len); 854 } 855 856 static inline bool __is_discard_back_mergeable(struct discard_info *cur, 857 struct discard_info *back, unsigned int max_len) 858 { 859 return __is_discard_mergeable(back, cur, max_len); 860 } 861 862 static inline bool __is_discard_front_mergeable(struct discard_info *cur, 863 struct discard_info *front, unsigned int max_len) 864 { 865 return __is_discard_mergeable(cur, front, max_len); 866 } 867 868 static inline bool __is_extent_mergeable(struct extent_info *back, 869 struct extent_info *front) 870 { 871 #ifdef CONFIG_F2FS_FS_COMPRESSION 872 if (back->c_len && back->len != back->c_len) 873 return false; 874 if (front->c_len && front->len != front->c_len) 875 return false; 876 #endif 877 return (back->fofs + back->len == front->fofs && 878 back->blk + back->len == front->blk); 879 } 880 881 static inline bool __is_back_mergeable(struct extent_info *cur, 882 struct extent_info *back) 883 { 884 return __is_extent_mergeable(back, cur); 885 } 886 887 static inline bool __is_front_mergeable(struct extent_info *cur, 888 struct extent_info *front) 889 { 890 return __is_extent_mergeable(cur, front); 891 } 892 893 extern void f2fs_mark_inode_dirty_sync(struct inode *inode, bool sync); 894 static inline void __try_update_largest_extent(struct extent_tree *et, 895 struct extent_node *en) 896 { 897 if (en->ei.len > et->largest.len) { 898 et->largest = en->ei; 899 et->largest_updated = true; 900 } 901 } 902 903 /* 904 * For free nid management 905 */ 906 enum nid_state { 907 FREE_NID, /* newly added to free nid list */ 908 PREALLOC_NID, /* it is preallocated */ 909 MAX_NID_STATE, 910 }; 911 912 enum nat_state { 913 TOTAL_NAT, 914 DIRTY_NAT, 915 RECLAIMABLE_NAT, 916 MAX_NAT_STATE, 917 }; 918 919 struct f2fs_nm_info { 920 block_t nat_blkaddr; /* base disk address of NAT */ 921 nid_t max_nid; /* maximum possible node ids */ 922 nid_t available_nids; /* # of available node ids */ 923 nid_t next_scan_nid; /* the next nid to be scanned */ 924 nid_t max_rf_node_blocks; /* max # of nodes for recovery */ 925 unsigned int ram_thresh; /* control the memory footprint */ 926 unsigned int ra_nid_pages; /* # of nid pages to be readaheaded */ 927 unsigned int dirty_nats_ratio; /* control dirty nats ratio threshold */ 928 929 /* NAT cache management */ 930 struct radix_tree_root nat_root;/* root of the nat entry cache */ 931 struct radix_tree_root nat_set_root;/* root of the nat set cache */ 932 struct f2fs_rwsem nat_tree_lock; /* protect nat entry tree */ 933 struct list_head nat_entries; /* cached nat entry list (clean) */ 934 spinlock_t nat_list_lock; /* protect clean nat entry list */ 935 unsigned int nat_cnt[MAX_NAT_STATE]; /* the # of cached nat entries */ 936 unsigned int nat_blocks; /* # of nat blocks */ 937 938 /* free node ids management */ 939 struct radix_tree_root free_nid_root;/* root of the free_nid cache */ 940 struct list_head free_nid_list; /* list for free nids excluding preallocated nids */ 941 unsigned int nid_cnt[MAX_NID_STATE]; /* the number of free node id */ 942 spinlock_t nid_list_lock; /* protect nid lists ops */ 943 struct mutex build_lock; /* lock for build free nids */ 944 unsigned char **free_nid_bitmap; 945 unsigned char *nat_block_bitmap; 946 unsigned short *free_nid_count; /* free nid count of NAT block */ 947 948 /* for checkpoint */ 949 char *nat_bitmap; /* NAT bitmap pointer */ 950 951 unsigned int nat_bits_blocks; /* # of nat bits blocks */ 952 unsigned char *nat_bits; /* NAT bits blocks */ 953 unsigned char *full_nat_bits; /* full NAT pages */ 954 unsigned char *empty_nat_bits; /* empty NAT pages */ 955 #ifdef CONFIG_F2FS_CHECK_FS 956 char *nat_bitmap_mir; /* NAT bitmap mirror */ 957 #endif 958 int bitmap_size; /* bitmap size */ 959 }; 960 961 /* 962 * this structure is used as one of function parameters. 963 * all the information are dedicated to a given direct node block determined 964 * by the data offset in a file. 965 */ 966 struct dnode_of_data { 967 struct inode *inode; /* vfs inode pointer */ 968 struct page *inode_page; /* its inode page, NULL is possible */ 969 struct page *node_page; /* cached direct node page */ 970 nid_t nid; /* node id of the direct node block */ 971 unsigned int ofs_in_node; /* data offset in the node page */ 972 bool inode_page_locked; /* inode page is locked or not */ 973 bool node_changed; /* is node block changed */ 974 char cur_level; /* level of hole node page */ 975 char max_level; /* level of current page located */ 976 block_t data_blkaddr; /* block address of the node block */ 977 }; 978 979 static inline void set_new_dnode(struct dnode_of_data *dn, struct inode *inode, 980 struct page *ipage, struct page *npage, nid_t nid) 981 { 982 memset(dn, 0, sizeof(*dn)); 983 dn->inode = inode; 984 dn->inode_page = ipage; 985 dn->node_page = npage; 986 dn->nid = nid; 987 } 988 989 /* 990 * For SIT manager 991 * 992 * By default, there are 6 active log areas across the whole main area. 993 * When considering hot and cold data separation to reduce cleaning overhead, 994 * we split 3 for data logs and 3 for node logs as hot, warm, and cold types, 995 * respectively. 996 * In the current design, you should not change the numbers intentionally. 997 * Instead, as a mount option such as active_logs=x, you can use 2, 4, and 6 998 * logs individually according to the underlying devices. (default: 6) 999 * Just in case, on-disk layout covers maximum 16 logs that consist of 8 for 1000 * data and 8 for node logs. 1001 */ 1002 #define NR_CURSEG_DATA_TYPE (3) 1003 #define NR_CURSEG_NODE_TYPE (3) 1004 #define NR_CURSEG_INMEM_TYPE (2) 1005 #define NR_CURSEG_RO_TYPE (2) 1006 #define NR_CURSEG_PERSIST_TYPE (NR_CURSEG_DATA_TYPE + NR_CURSEG_NODE_TYPE) 1007 #define NR_CURSEG_TYPE (NR_CURSEG_INMEM_TYPE + NR_CURSEG_PERSIST_TYPE) 1008 1009 enum { 1010 CURSEG_HOT_DATA = 0, /* directory entry blocks */ 1011 CURSEG_WARM_DATA, /* data blocks */ 1012 CURSEG_COLD_DATA, /* multimedia or GCed data blocks */ 1013 CURSEG_HOT_NODE, /* direct node blocks of directory files */ 1014 CURSEG_WARM_NODE, /* direct node blocks of normal files */ 1015 CURSEG_COLD_NODE, /* indirect node blocks */ 1016 NR_PERSISTENT_LOG, /* number of persistent log */ 1017 CURSEG_COLD_DATA_PINNED = NR_PERSISTENT_LOG, 1018 /* pinned file that needs consecutive block address */ 1019 CURSEG_ALL_DATA_ATGC, /* SSR alloctor in hot/warm/cold data area */ 1020 NO_CHECK_TYPE, /* number of persistent & inmem log */ 1021 }; 1022 1023 struct flush_cmd { 1024 struct completion wait; 1025 struct llist_node llnode; 1026 nid_t ino; 1027 int ret; 1028 }; 1029 1030 struct flush_cmd_control { 1031 struct task_struct *f2fs_issue_flush; /* flush thread */ 1032 wait_queue_head_t flush_wait_queue; /* waiting queue for wake-up */ 1033 atomic_t issued_flush; /* # of issued flushes */ 1034 atomic_t queued_flush; /* # of queued flushes */ 1035 struct llist_head issue_list; /* list for command issue */ 1036 struct llist_node *dispatch_list; /* list for command dispatch */ 1037 }; 1038 1039 struct f2fs_sm_info { 1040 struct sit_info *sit_info; /* whole segment information */ 1041 struct free_segmap_info *free_info; /* free segment information */ 1042 struct dirty_seglist_info *dirty_info; /* dirty segment information */ 1043 struct curseg_info *curseg_array; /* active segment information */ 1044 1045 struct f2fs_rwsem curseg_lock; /* for preventing curseg change */ 1046 1047 block_t seg0_blkaddr; /* block address of 0'th segment */ 1048 block_t main_blkaddr; /* start block address of main area */ 1049 block_t ssa_blkaddr; /* start block address of SSA area */ 1050 1051 unsigned int segment_count; /* total # of segments */ 1052 unsigned int main_segments; /* # of segments in main area */ 1053 unsigned int reserved_segments; /* # of reserved segments */ 1054 unsigned int additional_reserved_segments;/* reserved segs for IO align feature */ 1055 unsigned int ovp_segments; /* # of overprovision segments */ 1056 1057 /* a threshold to reclaim prefree segments */ 1058 unsigned int rec_prefree_segments; 1059 1060 /* for batched trimming */ 1061 unsigned int trim_sections; /* # of sections to trim */ 1062 1063 struct list_head sit_entry_set; /* sit entry set list */ 1064 1065 unsigned int ipu_policy; /* in-place-update policy */ 1066 unsigned int min_ipu_util; /* in-place-update threshold */ 1067 unsigned int min_fsync_blocks; /* threshold for fsync */ 1068 unsigned int min_seq_blocks; /* threshold for sequential blocks */ 1069 unsigned int min_hot_blocks; /* threshold for hot block allocation */ 1070 unsigned int min_ssr_sections; /* threshold to trigger SSR allocation */ 1071 1072 /* for flush command control */ 1073 struct flush_cmd_control *fcc_info; 1074 1075 /* for discard command control */ 1076 struct discard_cmd_control *dcc_info; 1077 }; 1078 1079 /* 1080 * For superblock 1081 */ 1082 /* 1083 * COUNT_TYPE for monitoring 1084 * 1085 * f2fs monitors the number of several block types such as on-writeback, 1086 * dirty dentry blocks, dirty node blocks, and dirty meta blocks. 1087 */ 1088 #define WB_DATA_TYPE(p) (__is_cp_guaranteed(p) ? F2FS_WB_CP_DATA : F2FS_WB_DATA) 1089 enum count_type { 1090 F2FS_DIRTY_DENTS, 1091 F2FS_DIRTY_DATA, 1092 F2FS_DIRTY_QDATA, 1093 F2FS_DIRTY_NODES, 1094 F2FS_DIRTY_META, 1095 F2FS_DIRTY_IMETA, 1096 F2FS_WB_CP_DATA, 1097 F2FS_WB_DATA, 1098 F2FS_RD_DATA, 1099 F2FS_RD_NODE, 1100 F2FS_RD_META, 1101 F2FS_DIO_WRITE, 1102 F2FS_DIO_READ, 1103 NR_COUNT_TYPE, 1104 }; 1105 1106 /* 1107 * The below are the page types of bios used in submit_bio(). 1108 * The available types are: 1109 * DATA User data pages. It operates as async mode. 1110 * NODE Node pages. It operates as async mode. 1111 * META FS metadata pages such as SIT, NAT, CP. 1112 * NR_PAGE_TYPE The number of page types. 1113 * META_FLUSH Make sure the previous pages are written 1114 * with waiting the bio's completion 1115 * ... Only can be used with META. 1116 */ 1117 #define PAGE_TYPE_OF_BIO(type) ((type) > META ? META : (type)) 1118 enum page_type { 1119 DATA = 0, 1120 NODE = 1, /* should not change this */ 1121 META, 1122 NR_PAGE_TYPE, 1123 META_FLUSH, 1124 IPU, /* the below types are used by tracepoints only. */ 1125 OPU, 1126 }; 1127 1128 enum temp_type { 1129 HOT = 0, /* must be zero for meta bio */ 1130 WARM, 1131 COLD, 1132 NR_TEMP_TYPE, 1133 }; 1134 1135 enum need_lock_type { 1136 LOCK_REQ = 0, 1137 LOCK_DONE, 1138 LOCK_RETRY, 1139 }; 1140 1141 enum cp_reason_type { 1142 CP_NO_NEEDED, 1143 CP_NON_REGULAR, 1144 CP_COMPRESSED, 1145 CP_HARDLINK, 1146 CP_SB_NEED_CP, 1147 CP_WRONG_PINO, 1148 CP_NO_SPC_ROLL, 1149 CP_NODE_NEED_CP, 1150 CP_FASTBOOT_MODE, 1151 CP_SPEC_LOG_NUM, 1152 CP_RECOVER_DIR, 1153 }; 1154 1155 enum iostat_type { 1156 /* WRITE IO */ 1157 APP_DIRECT_IO, /* app direct write IOs */ 1158 APP_BUFFERED_IO, /* app buffered write IOs */ 1159 APP_WRITE_IO, /* app write IOs */ 1160 APP_MAPPED_IO, /* app mapped IOs */ 1161 FS_DATA_IO, /* data IOs from kworker/fsync/reclaimer */ 1162 FS_NODE_IO, /* node IOs from kworker/fsync/reclaimer */ 1163 FS_META_IO, /* meta IOs from kworker/reclaimer */ 1164 FS_GC_DATA_IO, /* data IOs from forground gc */ 1165 FS_GC_NODE_IO, /* node IOs from forground gc */ 1166 FS_CP_DATA_IO, /* data IOs from checkpoint */ 1167 FS_CP_NODE_IO, /* node IOs from checkpoint */ 1168 FS_CP_META_IO, /* meta IOs from checkpoint */ 1169 1170 /* READ IO */ 1171 APP_DIRECT_READ_IO, /* app direct read IOs */ 1172 APP_BUFFERED_READ_IO, /* app buffered read IOs */ 1173 APP_READ_IO, /* app read IOs */ 1174 APP_MAPPED_READ_IO, /* app mapped read IOs */ 1175 FS_DATA_READ_IO, /* data read IOs */ 1176 FS_GDATA_READ_IO, /* data read IOs from background gc */ 1177 FS_CDATA_READ_IO, /* compressed data read IOs */ 1178 FS_NODE_READ_IO, /* node read IOs */ 1179 FS_META_READ_IO, /* meta read IOs */ 1180 1181 /* other */ 1182 FS_DISCARD, /* discard */ 1183 NR_IO_TYPE, 1184 }; 1185 1186 struct f2fs_io_info { 1187 struct f2fs_sb_info *sbi; /* f2fs_sb_info pointer */ 1188 nid_t ino; /* inode number */ 1189 enum page_type type; /* contains DATA/NODE/META/META_FLUSH */ 1190 enum temp_type temp; /* contains HOT/WARM/COLD */ 1191 enum req_op op; /* contains REQ_OP_ */ 1192 blk_opf_t op_flags; /* req_flag_bits */ 1193 block_t new_blkaddr; /* new block address to be written */ 1194 block_t old_blkaddr; /* old block address before Cow */ 1195 struct page *page; /* page to be written */ 1196 struct page *encrypted_page; /* encrypted page */ 1197 struct page *compressed_page; /* compressed page */ 1198 struct list_head list; /* serialize IOs */ 1199 bool submitted; /* indicate IO submission */ 1200 int need_lock; /* indicate we need to lock cp_rwsem */ 1201 bool in_list; /* indicate fio is in io_list */ 1202 bool is_por; /* indicate IO is from recovery or not */ 1203 bool retry; /* need to reallocate block address */ 1204 int compr_blocks; /* # of compressed block addresses */ 1205 bool encrypted; /* indicate file is encrypted */ 1206 bool post_read; /* require post read */ 1207 enum iostat_type io_type; /* io type */ 1208 struct writeback_control *io_wbc; /* writeback control */ 1209 struct bio **bio; /* bio for ipu */ 1210 sector_t *last_block; /* last block number in bio */ 1211 unsigned char version; /* version of the node */ 1212 }; 1213 1214 struct bio_entry { 1215 struct bio *bio; 1216 struct list_head list; 1217 }; 1218 1219 #define is_read_io(rw) ((rw) == READ) 1220 struct f2fs_bio_info { 1221 struct f2fs_sb_info *sbi; /* f2fs superblock */ 1222 struct bio *bio; /* bios to merge */ 1223 sector_t last_block_in_bio; /* last block number */ 1224 struct f2fs_io_info fio; /* store buffered io info. */ 1225 struct f2fs_rwsem io_rwsem; /* blocking op for bio */ 1226 spinlock_t io_lock; /* serialize DATA/NODE IOs */ 1227 struct list_head io_list; /* track fios */ 1228 struct list_head bio_list; /* bio entry list head */ 1229 struct f2fs_rwsem bio_list_lock; /* lock to protect bio entry list */ 1230 }; 1231 1232 #define FDEV(i) (sbi->devs[i]) 1233 #define RDEV(i) (raw_super->devs[i]) 1234 struct f2fs_dev_info { 1235 struct block_device *bdev; 1236 char path[MAX_PATH_LEN]; 1237 unsigned int total_segments; 1238 block_t start_blk; 1239 block_t end_blk; 1240 #ifdef CONFIG_BLK_DEV_ZONED 1241 unsigned int nr_blkz; /* Total number of zones */ 1242 unsigned long *blkz_seq; /* Bitmap indicating sequential zones */ 1243 #endif 1244 }; 1245 1246 enum inode_type { 1247 DIR_INODE, /* for dirty dir inode */ 1248 FILE_INODE, /* for dirty regular/symlink inode */ 1249 DIRTY_META, /* for all dirtied inode metadata */ 1250 ATOMIC_FILE, /* for all atomic files */ 1251 NR_INODE_TYPE, 1252 }; 1253 1254 /* for inner inode cache management */ 1255 struct inode_management { 1256 struct radix_tree_root ino_root; /* ino entry array */ 1257 spinlock_t ino_lock; /* for ino entry lock */ 1258 struct list_head ino_list; /* inode list head */ 1259 unsigned long ino_num; /* number of entries */ 1260 }; 1261 1262 /* for GC_AT */ 1263 struct atgc_management { 1264 bool atgc_enabled; /* ATGC is enabled or not */ 1265 struct rb_root_cached root; /* root of victim rb-tree */ 1266 struct list_head victim_list; /* linked with all victim entries */ 1267 unsigned int victim_count; /* victim count in rb-tree */ 1268 unsigned int candidate_ratio; /* candidate ratio */ 1269 unsigned int max_candidate_count; /* max candidate count */ 1270 unsigned int age_weight; /* age weight, vblock_weight = 100 - age_weight */ 1271 unsigned long long age_threshold; /* age threshold */ 1272 }; 1273 1274 struct f2fs_gc_control { 1275 unsigned int victim_segno; /* target victim segment number */ 1276 int init_gc_type; /* FG_GC or BG_GC */ 1277 bool no_bg_gc; /* check the space and stop bg_gc */ 1278 bool should_migrate_blocks; /* should migrate blocks */ 1279 bool err_gc_skipped; /* return EAGAIN if GC skipped */ 1280 unsigned int nr_free_secs; /* # of free sections to do GC */ 1281 }; 1282 1283 /* For s_flag in struct f2fs_sb_info */ 1284 enum { 1285 SBI_IS_DIRTY, /* dirty flag for checkpoint */ 1286 SBI_IS_CLOSE, /* specify unmounting */ 1287 SBI_NEED_FSCK, /* need fsck.f2fs to fix */ 1288 SBI_POR_DOING, /* recovery is doing or not */ 1289 SBI_NEED_SB_WRITE, /* need to recover superblock */ 1290 SBI_NEED_CP, /* need to checkpoint */ 1291 SBI_IS_SHUTDOWN, /* shutdown by ioctl */ 1292 SBI_IS_RECOVERED, /* recovered orphan/data */ 1293 SBI_CP_DISABLED, /* CP was disabled last mount */ 1294 SBI_CP_DISABLED_QUICK, /* CP was disabled quickly */ 1295 SBI_QUOTA_NEED_FLUSH, /* need to flush quota info in CP */ 1296 SBI_QUOTA_SKIP_FLUSH, /* skip flushing quota in current CP */ 1297 SBI_QUOTA_NEED_REPAIR, /* quota file may be corrupted */ 1298 SBI_IS_RESIZEFS, /* resizefs is in process */ 1299 SBI_IS_FREEZING, /* freezefs is in process */ 1300 }; 1301 1302 enum { 1303 CP_TIME, 1304 REQ_TIME, 1305 DISCARD_TIME, 1306 GC_TIME, 1307 DISABLE_TIME, 1308 UMOUNT_DISCARD_TIMEOUT, 1309 MAX_TIME, 1310 }; 1311 1312 enum { 1313 GC_NORMAL, 1314 GC_IDLE_CB, 1315 GC_IDLE_GREEDY, 1316 GC_IDLE_AT, 1317 GC_URGENT_HIGH, 1318 GC_URGENT_LOW, 1319 GC_URGENT_MID, 1320 MAX_GC_MODE, 1321 }; 1322 1323 enum { 1324 BGGC_MODE_ON, /* background gc is on */ 1325 BGGC_MODE_OFF, /* background gc is off */ 1326 BGGC_MODE_SYNC, /* 1327 * background gc is on, migrating blocks 1328 * like foreground gc 1329 */ 1330 }; 1331 1332 enum { 1333 FS_MODE_ADAPTIVE, /* use both lfs/ssr allocation */ 1334 FS_MODE_LFS, /* use lfs allocation only */ 1335 FS_MODE_FRAGMENT_SEG, /* segment fragmentation mode */ 1336 FS_MODE_FRAGMENT_BLK, /* block fragmentation mode */ 1337 }; 1338 1339 enum { 1340 ALLOC_MODE_DEFAULT, /* stay default */ 1341 ALLOC_MODE_REUSE, /* reuse segments as much as possible */ 1342 }; 1343 1344 enum fsync_mode { 1345 FSYNC_MODE_POSIX, /* fsync follows posix semantics */ 1346 FSYNC_MODE_STRICT, /* fsync behaves in line with ext4 */ 1347 FSYNC_MODE_NOBARRIER, /* fsync behaves nobarrier based on posix */ 1348 }; 1349 1350 enum { 1351 COMPR_MODE_FS, /* 1352 * automatically compress compression 1353 * enabled files 1354 */ 1355 COMPR_MODE_USER, /* 1356 * automatical compression is disabled. 1357 * user can control the file compression 1358 * using ioctls 1359 */ 1360 }; 1361 1362 enum { 1363 DISCARD_UNIT_BLOCK, /* basic discard unit is block */ 1364 DISCARD_UNIT_SEGMENT, /* basic discard unit is segment */ 1365 DISCARD_UNIT_SECTION, /* basic discard unit is section */ 1366 }; 1367 1368 enum { 1369 MEMORY_MODE_NORMAL, /* memory mode for normal devices */ 1370 MEMORY_MODE_LOW, /* memory mode for low memry devices */ 1371 }; 1372 1373 1374 1375 static inline int f2fs_test_bit(unsigned int nr, char *addr); 1376 static inline void f2fs_set_bit(unsigned int nr, char *addr); 1377 static inline void f2fs_clear_bit(unsigned int nr, char *addr); 1378 1379 /* 1380 * Layout of f2fs page.private: 1381 * 1382 * Layout A: lowest bit should be 1 1383 * | bit0 = 1 | bit1 | bit2 | ... | bit MAX | private data .... | 1384 * bit 0 PAGE_PRIVATE_NOT_POINTER 1385 * bit 1 PAGE_PRIVATE_ATOMIC_WRITE 1386 * bit 2 PAGE_PRIVATE_DUMMY_WRITE 1387 * bit 3 PAGE_PRIVATE_ONGOING_MIGRATION 1388 * bit 4 PAGE_PRIVATE_INLINE_INODE 1389 * bit 5 PAGE_PRIVATE_REF_RESOURCE 1390 * bit 6- f2fs private data 1391 * 1392 * Layout B: lowest bit should be 0 1393 * page.private is a wrapped pointer. 1394 */ 1395 enum { 1396 PAGE_PRIVATE_NOT_POINTER, /* private contains non-pointer data */ 1397 PAGE_PRIVATE_ATOMIC_WRITE, /* data page from atomic write path */ 1398 PAGE_PRIVATE_DUMMY_WRITE, /* data page for padding aligned IO */ 1399 PAGE_PRIVATE_ONGOING_MIGRATION, /* data page which is on-going migrating */ 1400 PAGE_PRIVATE_INLINE_INODE, /* inode page contains inline data */ 1401 PAGE_PRIVATE_REF_RESOURCE, /* dirty page has referenced resources */ 1402 PAGE_PRIVATE_MAX 1403 }; 1404 1405 #define PAGE_PRIVATE_GET_FUNC(name, flagname) \ 1406 static inline bool page_private_##name(struct page *page) \ 1407 { \ 1408 return PagePrivate(page) && \ 1409 test_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)) && \ 1410 test_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \ 1411 } 1412 1413 #define PAGE_PRIVATE_SET_FUNC(name, flagname) \ 1414 static inline void set_page_private_##name(struct page *page) \ 1415 { \ 1416 if (!PagePrivate(page)) { \ 1417 get_page(page); \ 1418 SetPagePrivate(page); \ 1419 set_page_private(page, 0); \ 1420 } \ 1421 set_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)); \ 1422 set_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \ 1423 } 1424 1425 #define PAGE_PRIVATE_CLEAR_FUNC(name, flagname) \ 1426 static inline void clear_page_private_##name(struct page *page) \ 1427 { \ 1428 clear_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \ 1429 if (page_private(page) == 1 << PAGE_PRIVATE_NOT_POINTER) { \ 1430 set_page_private(page, 0); \ 1431 if (PagePrivate(page)) { \ 1432 ClearPagePrivate(page); \ 1433 put_page(page); \ 1434 }\ 1435 } \ 1436 } 1437 1438 PAGE_PRIVATE_GET_FUNC(nonpointer, NOT_POINTER); 1439 PAGE_PRIVATE_GET_FUNC(reference, REF_RESOURCE); 1440 PAGE_PRIVATE_GET_FUNC(inline, INLINE_INODE); 1441 PAGE_PRIVATE_GET_FUNC(gcing, ONGOING_MIGRATION); 1442 PAGE_PRIVATE_GET_FUNC(atomic, ATOMIC_WRITE); 1443 PAGE_PRIVATE_GET_FUNC(dummy, DUMMY_WRITE); 1444 1445 PAGE_PRIVATE_SET_FUNC(reference, REF_RESOURCE); 1446 PAGE_PRIVATE_SET_FUNC(inline, INLINE_INODE); 1447 PAGE_PRIVATE_SET_FUNC(gcing, ONGOING_MIGRATION); 1448 PAGE_PRIVATE_SET_FUNC(atomic, ATOMIC_WRITE); 1449 PAGE_PRIVATE_SET_FUNC(dummy, DUMMY_WRITE); 1450 1451 PAGE_PRIVATE_CLEAR_FUNC(reference, REF_RESOURCE); 1452 PAGE_PRIVATE_CLEAR_FUNC(inline, INLINE_INODE); 1453 PAGE_PRIVATE_CLEAR_FUNC(gcing, ONGOING_MIGRATION); 1454 PAGE_PRIVATE_CLEAR_FUNC(atomic, ATOMIC_WRITE); 1455 PAGE_PRIVATE_CLEAR_FUNC(dummy, DUMMY_WRITE); 1456 1457 static inline unsigned long get_page_private_data(struct page *page) 1458 { 1459 unsigned long data = page_private(page); 1460 1461 if (!test_bit(PAGE_PRIVATE_NOT_POINTER, &data)) 1462 return 0; 1463 return data >> PAGE_PRIVATE_MAX; 1464 } 1465 1466 static inline void set_page_private_data(struct page *page, unsigned long data) 1467 { 1468 if (!PagePrivate(page)) { 1469 get_page(page); 1470 SetPagePrivate(page); 1471 set_page_private(page, 0); 1472 } 1473 set_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)); 1474 page_private(page) |= data << PAGE_PRIVATE_MAX; 1475 } 1476 1477 static inline void clear_page_private_data(struct page *page) 1478 { 1479 page_private(page) &= (1 << PAGE_PRIVATE_MAX) - 1; 1480 if (page_private(page) == 1 << PAGE_PRIVATE_NOT_POINTER) { 1481 set_page_private(page, 0); 1482 if (PagePrivate(page)) { 1483 ClearPagePrivate(page); 1484 put_page(page); 1485 } 1486 } 1487 } 1488 1489 /* For compression */ 1490 enum compress_algorithm_type { 1491 COMPRESS_LZO, 1492 COMPRESS_LZ4, 1493 COMPRESS_ZSTD, 1494 COMPRESS_LZORLE, 1495 COMPRESS_MAX, 1496 }; 1497 1498 enum compress_flag { 1499 COMPRESS_CHKSUM, 1500 COMPRESS_MAX_FLAG, 1501 }; 1502 1503 #define COMPRESS_WATERMARK 20 1504 #define COMPRESS_PERCENT 20 1505 1506 #define COMPRESS_DATA_RESERVED_SIZE 4 1507 struct compress_data { 1508 __le32 clen; /* compressed data size */ 1509 __le32 chksum; /* compressed data chksum */ 1510 __le32 reserved[COMPRESS_DATA_RESERVED_SIZE]; /* reserved */ 1511 u8 cdata[]; /* compressed data */ 1512 }; 1513 1514 #define COMPRESS_HEADER_SIZE (sizeof(struct compress_data)) 1515 1516 #define F2FS_COMPRESSED_PAGE_MAGIC 0xF5F2C000 1517 1518 #define COMPRESS_LEVEL_OFFSET 8 1519 1520 /* compress context */ 1521 struct compress_ctx { 1522 struct inode *inode; /* inode the context belong to */ 1523 pgoff_t cluster_idx; /* cluster index number */ 1524 unsigned int cluster_size; /* page count in cluster */ 1525 unsigned int log_cluster_size; /* log of cluster size */ 1526 struct page **rpages; /* pages store raw data in cluster */ 1527 unsigned int nr_rpages; /* total page number in rpages */ 1528 struct page **cpages; /* pages store compressed data in cluster */ 1529 unsigned int nr_cpages; /* total page number in cpages */ 1530 unsigned int valid_nr_cpages; /* valid page number in cpages */ 1531 void *rbuf; /* virtual mapped address on rpages */ 1532 struct compress_data *cbuf; /* virtual mapped address on cpages */ 1533 size_t rlen; /* valid data length in rbuf */ 1534 size_t clen; /* valid data length in cbuf */ 1535 void *private; /* payload buffer for specified compression algorithm */ 1536 void *private2; /* extra payload buffer */ 1537 }; 1538 1539 /* compress context for write IO path */ 1540 struct compress_io_ctx { 1541 u32 magic; /* magic number to indicate page is compressed */ 1542 struct inode *inode; /* inode the context belong to */ 1543 struct page **rpages; /* pages store raw data in cluster */ 1544 unsigned int nr_rpages; /* total page number in rpages */ 1545 atomic_t pending_pages; /* in-flight compressed page count */ 1546 }; 1547 1548 /* Context for decompressing one cluster on the read IO path */ 1549 struct decompress_io_ctx { 1550 u32 magic; /* magic number to indicate page is compressed */ 1551 struct inode *inode; /* inode the context belong to */ 1552 pgoff_t cluster_idx; /* cluster index number */ 1553 unsigned int cluster_size; /* page count in cluster */ 1554 unsigned int log_cluster_size; /* log of cluster size */ 1555 struct page **rpages; /* pages store raw data in cluster */ 1556 unsigned int nr_rpages; /* total page number in rpages */ 1557 struct page **cpages; /* pages store compressed data in cluster */ 1558 unsigned int nr_cpages; /* total page number in cpages */ 1559 struct page **tpages; /* temp pages to pad holes in cluster */ 1560 void *rbuf; /* virtual mapped address on rpages */ 1561 struct compress_data *cbuf; /* virtual mapped address on cpages */ 1562 size_t rlen; /* valid data length in rbuf */ 1563 size_t clen; /* valid data length in cbuf */ 1564 1565 /* 1566 * The number of compressed pages remaining to be read in this cluster. 1567 * This is initially nr_cpages. It is decremented by 1 each time a page 1568 * has been read (or failed to be read). When it reaches 0, the cluster 1569 * is decompressed (or an error is reported). 1570 * 1571 * If an error occurs before all the pages have been submitted for I/O, 1572 * then this will never reach 0. In this case the I/O submitter is 1573 * responsible for calling f2fs_decompress_end_io() instead. 1574 */ 1575 atomic_t remaining_pages; 1576 1577 /* 1578 * Number of references to this decompress_io_ctx. 1579 * 1580 * One reference is held for I/O completion. This reference is dropped 1581 * after the pagecache pages are updated and unlocked -- either after 1582 * decompression (and verity if enabled), or after an error. 1583 * 1584 * In addition, each compressed page holds a reference while it is in a 1585 * bio. These references are necessary prevent compressed pages from 1586 * being freed while they are still in a bio. 1587 */ 1588 refcount_t refcnt; 1589 1590 bool failed; /* IO error occurred before decompression? */ 1591 bool need_verity; /* need fs-verity verification after decompression? */ 1592 void *private; /* payload buffer for specified decompression algorithm */ 1593 void *private2; /* extra payload buffer */ 1594 struct work_struct verity_work; /* work to verify the decompressed pages */ 1595 struct work_struct free_work; /* work for late free this structure itself */ 1596 }; 1597 1598 #define NULL_CLUSTER ((unsigned int)(~0)) 1599 #define MIN_COMPRESS_LOG_SIZE 2 1600 #define MAX_COMPRESS_LOG_SIZE 8 1601 #define MAX_COMPRESS_WINDOW_SIZE(log_size) ((PAGE_SIZE) << (log_size)) 1602 1603 struct f2fs_sb_info { 1604 struct super_block *sb; /* pointer to VFS super block */ 1605 struct proc_dir_entry *s_proc; /* proc entry */ 1606 struct f2fs_super_block *raw_super; /* raw super block pointer */ 1607 struct f2fs_rwsem sb_lock; /* lock for raw super block */ 1608 int valid_super_block; /* valid super block no */ 1609 unsigned long s_flag; /* flags for sbi */ 1610 struct mutex writepages; /* mutex for writepages() */ 1611 1612 #ifdef CONFIG_BLK_DEV_ZONED 1613 unsigned int blocks_per_blkz; /* F2FS blocks per zone */ 1614 unsigned int log_blocks_per_blkz; /* log2 F2FS blocks per zone */ 1615 #endif 1616 1617 /* for node-related operations */ 1618 struct f2fs_nm_info *nm_info; /* node manager */ 1619 struct inode *node_inode; /* cache node blocks */ 1620 1621 /* for segment-related operations */ 1622 struct f2fs_sm_info *sm_info; /* segment manager */ 1623 1624 /* for bio operations */ 1625 struct f2fs_bio_info *write_io[NR_PAGE_TYPE]; /* for write bios */ 1626 /* keep migration IO order for LFS mode */ 1627 struct f2fs_rwsem io_order_lock; 1628 mempool_t *write_io_dummy; /* Dummy pages */ 1629 pgoff_t page_eio_ofs[NR_PAGE_TYPE]; /* EIO page offset */ 1630 int page_eio_cnt[NR_PAGE_TYPE]; /* EIO count */ 1631 1632 /* for checkpoint */ 1633 struct f2fs_checkpoint *ckpt; /* raw checkpoint pointer */ 1634 int cur_cp_pack; /* remain current cp pack */ 1635 spinlock_t cp_lock; /* for flag in ckpt */ 1636 struct inode *meta_inode; /* cache meta blocks */ 1637 struct f2fs_rwsem cp_global_sem; /* checkpoint procedure lock */ 1638 struct f2fs_rwsem cp_rwsem; /* blocking FS operations */ 1639 struct f2fs_rwsem node_write; /* locking node writes */ 1640 struct f2fs_rwsem node_change; /* locking node change */ 1641 wait_queue_head_t cp_wait; 1642 unsigned long last_time[MAX_TIME]; /* to store time in jiffies */ 1643 long interval_time[MAX_TIME]; /* to store thresholds */ 1644 struct ckpt_req_control cprc_info; /* for checkpoint request control */ 1645 1646 struct inode_management im[MAX_INO_ENTRY]; /* manage inode cache */ 1647 1648 spinlock_t fsync_node_lock; /* for node entry lock */ 1649 struct list_head fsync_node_list; /* node list head */ 1650 unsigned int fsync_seg_id; /* sequence id */ 1651 unsigned int fsync_node_num; /* number of node entries */ 1652 1653 /* for orphan inode, use 0'th array */ 1654 unsigned int max_orphans; /* max orphan inodes */ 1655 1656 /* for inode management */ 1657 struct list_head inode_list[NR_INODE_TYPE]; /* dirty inode list */ 1658 spinlock_t inode_lock[NR_INODE_TYPE]; /* for dirty inode list lock */ 1659 struct mutex flush_lock; /* for flush exclusion */ 1660 1661 /* for extent tree cache */ 1662 struct radix_tree_root extent_tree_root;/* cache extent cache entries */ 1663 struct mutex extent_tree_lock; /* locking extent radix tree */ 1664 struct list_head extent_list; /* lru list for shrinker */ 1665 spinlock_t extent_lock; /* locking extent lru list */ 1666 atomic_t total_ext_tree; /* extent tree count */ 1667 struct list_head zombie_list; /* extent zombie tree list */ 1668 atomic_t total_zombie_tree; /* extent zombie tree count */ 1669 atomic_t total_ext_node; /* extent info count */ 1670 1671 /* basic filesystem units */ 1672 unsigned int log_sectors_per_block; /* log2 sectors per block */ 1673 unsigned int log_blocksize; /* log2 block size */ 1674 unsigned int blocksize; /* block size */ 1675 unsigned int root_ino_num; /* root inode number*/ 1676 unsigned int node_ino_num; /* node inode number*/ 1677 unsigned int meta_ino_num; /* meta inode number*/ 1678 unsigned int log_blocks_per_seg; /* log2 blocks per segment */ 1679 unsigned int blocks_per_seg; /* blocks per segment */ 1680 unsigned int unusable_blocks_per_sec; /* unusable blocks per section */ 1681 unsigned int segs_per_sec; /* segments per section */ 1682 unsigned int secs_per_zone; /* sections per zone */ 1683 unsigned int total_sections; /* total section count */ 1684 unsigned int total_node_count; /* total node block count */ 1685 unsigned int total_valid_node_count; /* valid node block count */ 1686 int dir_level; /* directory level */ 1687 int readdir_ra; /* readahead inode in readdir */ 1688 u64 max_io_bytes; /* max io bytes to merge IOs */ 1689 1690 block_t user_block_count; /* # of user blocks */ 1691 block_t total_valid_block_count; /* # of valid blocks */ 1692 block_t discard_blks; /* discard command candidats */ 1693 block_t last_valid_block_count; /* for recovery */ 1694 block_t reserved_blocks; /* configurable reserved blocks */ 1695 block_t current_reserved_blocks; /* current reserved blocks */ 1696 1697 /* Additional tracking for no checkpoint mode */ 1698 block_t unusable_block_count; /* # of blocks saved by last cp */ 1699 1700 unsigned int nquota_files; /* # of quota sysfile */ 1701 struct f2fs_rwsem quota_sem; /* blocking cp for flags */ 1702 1703 /* # of pages, see count_type */ 1704 atomic_t nr_pages[NR_COUNT_TYPE]; 1705 /* # of allocated blocks */ 1706 struct percpu_counter alloc_valid_block_count; 1707 /* # of node block writes as roll forward recovery */ 1708 struct percpu_counter rf_node_block_count; 1709 1710 /* writeback control */ 1711 atomic_t wb_sync_req[META]; /* count # of WB_SYNC threads */ 1712 1713 /* valid inode count */ 1714 struct percpu_counter total_valid_inode_count; 1715 1716 struct f2fs_mount_info mount_opt; /* mount options */ 1717 1718 /* for cleaning operations */ 1719 struct f2fs_rwsem gc_lock; /* 1720 * semaphore for GC, avoid 1721 * race between GC and GC or CP 1722 */ 1723 struct f2fs_gc_kthread *gc_thread; /* GC thread */ 1724 struct atgc_management am; /* atgc management */ 1725 unsigned int cur_victim_sec; /* current victim section num */ 1726 unsigned int gc_mode; /* current GC state */ 1727 unsigned int next_victim_seg[2]; /* next segment in victim section */ 1728 spinlock_t gc_urgent_high_lock; 1729 bool gc_urgent_high_limited; /* indicates having limited trial count */ 1730 unsigned int gc_urgent_high_remaining; /* remaining trial count for GC_URGENT_HIGH */ 1731 1732 /* for skip statistic */ 1733 unsigned int atomic_files; /* # of opened atomic file */ 1734 unsigned long long skipped_gc_rwsem; /* FG_GC only */ 1735 1736 /* threshold for gc trials on pinned files */ 1737 u64 gc_pin_file_threshold; 1738 struct f2fs_rwsem pin_sem; 1739 1740 /* maximum # of trials to find a victim segment for SSR and GC */ 1741 unsigned int max_victim_search; 1742 /* migration granularity of garbage collection, unit: segment */ 1743 unsigned int migration_granularity; 1744 1745 /* 1746 * for stat information. 1747 * one is for the LFS mode, and the other is for the SSR mode. 1748 */ 1749 #ifdef CONFIG_F2FS_STAT_FS 1750 struct f2fs_stat_info *stat_info; /* FS status information */ 1751 atomic_t meta_count[META_MAX]; /* # of meta blocks */ 1752 unsigned int segment_count[2]; /* # of allocated segments */ 1753 unsigned int block_count[2]; /* # of allocated blocks */ 1754 atomic_t inplace_count; /* # of inplace update */ 1755 atomic64_t total_hit_ext; /* # of lookup extent cache */ 1756 atomic64_t read_hit_rbtree; /* # of hit rbtree extent node */ 1757 atomic64_t read_hit_largest; /* # of hit largest extent node */ 1758 atomic64_t read_hit_cached; /* # of hit cached extent node */ 1759 atomic_t inline_xattr; /* # of inline_xattr inodes */ 1760 atomic_t inline_inode; /* # of inline_data inodes */ 1761 atomic_t inline_dir; /* # of inline_dentry inodes */ 1762 atomic_t compr_inode; /* # of compressed inodes */ 1763 atomic64_t compr_blocks; /* # of compressed blocks */ 1764 atomic_t max_aw_cnt; /* max # of atomic writes */ 1765 unsigned int io_skip_bggc; /* skip background gc for in-flight IO */ 1766 unsigned int other_skip_bggc; /* skip background gc for other reasons */ 1767 unsigned int ndirty_inode[NR_INODE_TYPE]; /* # of dirty inodes */ 1768 #endif 1769 spinlock_t stat_lock; /* lock for stat operations */ 1770 1771 /* to attach REQ_META|REQ_FUA flags */ 1772 unsigned int data_io_flag; 1773 unsigned int node_io_flag; 1774 1775 /* For sysfs support */ 1776 struct kobject s_kobj; /* /sys/fs/f2fs/<devname> */ 1777 struct completion s_kobj_unregister; 1778 1779 struct kobject s_stat_kobj; /* /sys/fs/f2fs/<devname>/stat */ 1780 struct completion s_stat_kobj_unregister; 1781 1782 struct kobject s_feature_list_kobj; /* /sys/fs/f2fs/<devname>/feature_list */ 1783 struct completion s_feature_list_kobj_unregister; 1784 1785 /* For shrinker support */ 1786 struct list_head s_list; 1787 struct mutex umount_mutex; 1788 unsigned int shrinker_run_no; 1789 1790 /* For multi devices */ 1791 int s_ndevs; /* number of devices */ 1792 struct f2fs_dev_info *devs; /* for device list */ 1793 unsigned int dirty_device; /* for checkpoint data flush */ 1794 spinlock_t dev_lock; /* protect dirty_device */ 1795 bool aligned_blksize; /* all devices has the same logical blksize */ 1796 1797 /* For write statistics */ 1798 u64 sectors_written_start; 1799 u64 kbytes_written; 1800 1801 /* Reference to checksum algorithm driver via cryptoapi */ 1802 struct crypto_shash *s_chksum_driver; 1803 1804 /* Precomputed FS UUID checksum for seeding other checksums */ 1805 __u32 s_chksum_seed; 1806 1807 struct workqueue_struct *post_read_wq; /* post read workqueue */ 1808 1809 struct kmem_cache *inline_xattr_slab; /* inline xattr entry */ 1810 unsigned int inline_xattr_slab_size; /* default inline xattr slab size */ 1811 1812 /* For reclaimed segs statistics per each GC mode */ 1813 unsigned int gc_segment_mode; /* GC state for reclaimed segments */ 1814 unsigned int gc_reclaimed_segs[MAX_GC_MODE]; /* Reclaimed segs for each mode */ 1815 1816 unsigned long seq_file_ra_mul; /* multiplier for ra_pages of seq. files in fadvise */ 1817 1818 int max_fragment_chunk; /* max chunk size for block fragmentation mode */ 1819 int max_fragment_hole; /* max hole size for block fragmentation mode */ 1820 1821 /* For atomic write statistics */ 1822 atomic64_t current_atomic_write; 1823 s64 peak_atomic_write; 1824 u64 committed_atomic_block; 1825 u64 revoked_atomic_block; 1826 1827 #ifdef CONFIG_F2FS_FS_COMPRESSION 1828 struct kmem_cache *page_array_slab; /* page array entry */ 1829 unsigned int page_array_slab_size; /* default page array slab size */ 1830 1831 /* For runtime compression statistics */ 1832 u64 compr_written_block; 1833 u64 compr_saved_block; 1834 u32 compr_new_inode; 1835 1836 /* For compressed block cache */ 1837 struct inode *compress_inode; /* cache compressed blocks */ 1838 unsigned int compress_percent; /* cache page percentage */ 1839 unsigned int compress_watermark; /* cache page watermark */ 1840 atomic_t compress_page_hit; /* cache hit count */ 1841 #endif 1842 1843 #ifdef CONFIG_F2FS_IOSTAT 1844 /* For app/fs IO statistics */ 1845 spinlock_t iostat_lock; 1846 unsigned long long rw_iostat[NR_IO_TYPE]; 1847 unsigned long long prev_rw_iostat[NR_IO_TYPE]; 1848 bool iostat_enable; 1849 unsigned long iostat_next_period; 1850 unsigned int iostat_period_ms; 1851 1852 /* For io latency related statistics info in one iostat period */ 1853 spinlock_t iostat_lat_lock; 1854 struct iostat_lat_info *iostat_io_lat; 1855 #endif 1856 }; 1857 1858 #ifdef CONFIG_F2FS_FAULT_INJECTION 1859 #define f2fs_show_injection_info(sbi, type) \ 1860 printk_ratelimited("%sF2FS-fs (%s) : inject %s in %s of %pS\n", \ 1861 KERN_INFO, sbi->sb->s_id, \ 1862 f2fs_fault_name[type], \ 1863 __func__, __builtin_return_address(0)) 1864 static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type) 1865 { 1866 struct f2fs_fault_info *ffi = &F2FS_OPTION(sbi).fault_info; 1867 1868 if (!ffi->inject_rate) 1869 return false; 1870 1871 if (!IS_FAULT_SET(ffi, type)) 1872 return false; 1873 1874 atomic_inc(&ffi->inject_ops); 1875 if (atomic_read(&ffi->inject_ops) >= ffi->inject_rate) { 1876 atomic_set(&ffi->inject_ops, 0); 1877 return true; 1878 } 1879 return false; 1880 } 1881 #else 1882 #define f2fs_show_injection_info(sbi, type) do { } while (0) 1883 static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type) 1884 { 1885 return false; 1886 } 1887 #endif 1888 1889 /* 1890 * Test if the mounted volume is a multi-device volume. 1891 * - For a single regular disk volume, sbi->s_ndevs is 0. 1892 * - For a single zoned disk volume, sbi->s_ndevs is 1. 1893 * - For a multi-device volume, sbi->s_ndevs is always 2 or more. 1894 */ 1895 static inline bool f2fs_is_multi_device(struct f2fs_sb_info *sbi) 1896 { 1897 return sbi->s_ndevs > 1; 1898 } 1899 1900 static inline void f2fs_update_time(struct f2fs_sb_info *sbi, int type) 1901 { 1902 unsigned long now = jiffies; 1903 1904 sbi->last_time[type] = now; 1905 1906 /* DISCARD_TIME and GC_TIME are based on REQ_TIME */ 1907 if (type == REQ_TIME) { 1908 sbi->last_time[DISCARD_TIME] = now; 1909 sbi->last_time[GC_TIME] = now; 1910 } 1911 } 1912 1913 static inline bool f2fs_time_over(struct f2fs_sb_info *sbi, int type) 1914 { 1915 unsigned long interval = sbi->interval_time[type] * HZ; 1916 1917 return time_after(jiffies, sbi->last_time[type] + interval); 1918 } 1919 1920 static inline unsigned int f2fs_time_to_wait(struct f2fs_sb_info *sbi, 1921 int type) 1922 { 1923 unsigned long interval = sbi->interval_time[type] * HZ; 1924 unsigned int wait_ms = 0; 1925 long delta; 1926 1927 delta = (sbi->last_time[type] + interval) - jiffies; 1928 if (delta > 0) 1929 wait_ms = jiffies_to_msecs(delta); 1930 1931 return wait_ms; 1932 } 1933 1934 /* 1935 * Inline functions 1936 */ 1937 static inline u32 __f2fs_crc32(struct f2fs_sb_info *sbi, u32 crc, 1938 const void *address, unsigned int length) 1939 { 1940 struct { 1941 struct shash_desc shash; 1942 char ctx[4]; 1943 } desc; 1944 int err; 1945 1946 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver) != sizeof(desc.ctx)); 1947 1948 desc.shash.tfm = sbi->s_chksum_driver; 1949 *(u32 *)desc.ctx = crc; 1950 1951 err = crypto_shash_update(&desc.shash, address, length); 1952 BUG_ON(err); 1953 1954 return *(u32 *)desc.ctx; 1955 } 1956 1957 static inline u32 f2fs_crc32(struct f2fs_sb_info *sbi, const void *address, 1958 unsigned int length) 1959 { 1960 return __f2fs_crc32(sbi, F2FS_SUPER_MAGIC, address, length); 1961 } 1962 1963 static inline bool f2fs_crc_valid(struct f2fs_sb_info *sbi, __u32 blk_crc, 1964 void *buf, size_t buf_size) 1965 { 1966 return f2fs_crc32(sbi, buf, buf_size) == blk_crc; 1967 } 1968 1969 static inline u32 f2fs_chksum(struct f2fs_sb_info *sbi, u32 crc, 1970 const void *address, unsigned int length) 1971 { 1972 return __f2fs_crc32(sbi, crc, address, length); 1973 } 1974 1975 static inline struct f2fs_inode_info *F2FS_I(struct inode *inode) 1976 { 1977 return container_of(inode, struct f2fs_inode_info, vfs_inode); 1978 } 1979 1980 static inline struct f2fs_sb_info *F2FS_SB(struct super_block *sb) 1981 { 1982 return sb->s_fs_info; 1983 } 1984 1985 static inline struct f2fs_sb_info *F2FS_I_SB(struct inode *inode) 1986 { 1987 return F2FS_SB(inode->i_sb); 1988 } 1989 1990 static inline struct f2fs_sb_info *F2FS_M_SB(struct address_space *mapping) 1991 { 1992 return F2FS_I_SB(mapping->host); 1993 } 1994 1995 static inline struct f2fs_sb_info *F2FS_P_SB(struct page *page) 1996 { 1997 return F2FS_M_SB(page_file_mapping(page)); 1998 } 1999 2000 static inline struct f2fs_super_block *F2FS_RAW_SUPER(struct f2fs_sb_info *sbi) 2001 { 2002 return (struct f2fs_super_block *)(sbi->raw_super); 2003 } 2004 2005 static inline struct f2fs_checkpoint *F2FS_CKPT(struct f2fs_sb_info *sbi) 2006 { 2007 return (struct f2fs_checkpoint *)(sbi->ckpt); 2008 } 2009 2010 static inline struct f2fs_node *F2FS_NODE(struct page *page) 2011 { 2012 return (struct f2fs_node *)page_address(page); 2013 } 2014 2015 static inline struct f2fs_inode *F2FS_INODE(struct page *page) 2016 { 2017 return &((struct f2fs_node *)page_address(page))->i; 2018 } 2019 2020 static inline struct f2fs_nm_info *NM_I(struct f2fs_sb_info *sbi) 2021 { 2022 return (struct f2fs_nm_info *)(sbi->nm_info); 2023 } 2024 2025 static inline struct f2fs_sm_info *SM_I(struct f2fs_sb_info *sbi) 2026 { 2027 return (struct f2fs_sm_info *)(sbi->sm_info); 2028 } 2029 2030 static inline struct sit_info *SIT_I(struct f2fs_sb_info *sbi) 2031 { 2032 return (struct sit_info *)(SM_I(sbi)->sit_info); 2033 } 2034 2035 static inline struct free_segmap_info *FREE_I(struct f2fs_sb_info *sbi) 2036 { 2037 return (struct free_segmap_info *)(SM_I(sbi)->free_info); 2038 } 2039 2040 static inline struct dirty_seglist_info *DIRTY_I(struct f2fs_sb_info *sbi) 2041 { 2042 return (struct dirty_seglist_info *)(SM_I(sbi)->dirty_info); 2043 } 2044 2045 static inline struct address_space *META_MAPPING(struct f2fs_sb_info *sbi) 2046 { 2047 return sbi->meta_inode->i_mapping; 2048 } 2049 2050 static inline struct address_space *NODE_MAPPING(struct f2fs_sb_info *sbi) 2051 { 2052 return sbi->node_inode->i_mapping; 2053 } 2054 2055 static inline bool is_sbi_flag_set(struct f2fs_sb_info *sbi, unsigned int type) 2056 { 2057 return test_bit(type, &sbi->s_flag); 2058 } 2059 2060 static inline void set_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type) 2061 { 2062 set_bit(type, &sbi->s_flag); 2063 } 2064 2065 static inline void clear_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type) 2066 { 2067 clear_bit(type, &sbi->s_flag); 2068 } 2069 2070 static inline unsigned long long cur_cp_version(struct f2fs_checkpoint *cp) 2071 { 2072 return le64_to_cpu(cp->checkpoint_ver); 2073 } 2074 2075 static inline unsigned long f2fs_qf_ino(struct super_block *sb, int type) 2076 { 2077 if (type < F2FS_MAX_QUOTAS) 2078 return le32_to_cpu(F2FS_SB(sb)->raw_super->qf_ino[type]); 2079 return 0; 2080 } 2081 2082 static inline __u64 cur_cp_crc(struct f2fs_checkpoint *cp) 2083 { 2084 size_t crc_offset = le32_to_cpu(cp->checksum_offset); 2085 return le32_to_cpu(*((__le32 *)((unsigned char *)cp + crc_offset))); 2086 } 2087 2088 static inline bool __is_set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 2089 { 2090 unsigned int ckpt_flags = le32_to_cpu(cp->ckpt_flags); 2091 2092 return ckpt_flags & f; 2093 } 2094 2095 static inline bool is_set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 2096 { 2097 return __is_set_ckpt_flags(F2FS_CKPT(sbi), f); 2098 } 2099 2100 static inline void __set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 2101 { 2102 unsigned int ckpt_flags; 2103 2104 ckpt_flags = le32_to_cpu(cp->ckpt_flags); 2105 ckpt_flags |= f; 2106 cp->ckpt_flags = cpu_to_le32(ckpt_flags); 2107 } 2108 2109 static inline void set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 2110 { 2111 unsigned long flags; 2112 2113 spin_lock_irqsave(&sbi->cp_lock, flags); 2114 __set_ckpt_flags(F2FS_CKPT(sbi), f); 2115 spin_unlock_irqrestore(&sbi->cp_lock, flags); 2116 } 2117 2118 static inline void __clear_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 2119 { 2120 unsigned int ckpt_flags; 2121 2122 ckpt_flags = le32_to_cpu(cp->ckpt_flags); 2123 ckpt_flags &= (~f); 2124 cp->ckpt_flags = cpu_to_le32(ckpt_flags); 2125 } 2126 2127 static inline void clear_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 2128 { 2129 unsigned long flags; 2130 2131 spin_lock_irqsave(&sbi->cp_lock, flags); 2132 __clear_ckpt_flags(F2FS_CKPT(sbi), f); 2133 spin_unlock_irqrestore(&sbi->cp_lock, flags); 2134 } 2135 2136 #define init_f2fs_rwsem(sem) \ 2137 do { \ 2138 static struct lock_class_key __key; \ 2139 \ 2140 __init_f2fs_rwsem((sem), #sem, &__key); \ 2141 } while (0) 2142 2143 static inline void __init_f2fs_rwsem(struct f2fs_rwsem *sem, 2144 const char *sem_name, struct lock_class_key *key) 2145 { 2146 __init_rwsem(&sem->internal_rwsem, sem_name, key); 2147 #ifdef CONFIG_F2FS_UNFAIR_RWSEM 2148 init_waitqueue_head(&sem->read_waiters); 2149 #endif 2150 } 2151 2152 static inline int f2fs_rwsem_is_locked(struct f2fs_rwsem *sem) 2153 { 2154 return rwsem_is_locked(&sem->internal_rwsem); 2155 } 2156 2157 static inline int f2fs_rwsem_is_contended(struct f2fs_rwsem *sem) 2158 { 2159 return rwsem_is_contended(&sem->internal_rwsem); 2160 } 2161 2162 static inline void f2fs_down_read(struct f2fs_rwsem *sem) 2163 { 2164 #ifdef CONFIG_F2FS_UNFAIR_RWSEM 2165 wait_event(sem->read_waiters, down_read_trylock(&sem->internal_rwsem)); 2166 #else 2167 down_read(&sem->internal_rwsem); 2168 #endif 2169 } 2170 2171 static inline int f2fs_down_read_trylock(struct f2fs_rwsem *sem) 2172 { 2173 return down_read_trylock(&sem->internal_rwsem); 2174 } 2175 2176 #ifdef CONFIG_DEBUG_LOCK_ALLOC 2177 static inline void f2fs_down_read_nested(struct f2fs_rwsem *sem, int subclass) 2178 { 2179 down_read_nested(&sem->internal_rwsem, subclass); 2180 } 2181 #else 2182 #define f2fs_down_read_nested(sem, subclass) f2fs_down_read(sem) 2183 #endif 2184 2185 static inline void f2fs_up_read(struct f2fs_rwsem *sem) 2186 { 2187 up_read(&sem->internal_rwsem); 2188 } 2189 2190 static inline void f2fs_down_write(struct f2fs_rwsem *sem) 2191 { 2192 down_write(&sem->internal_rwsem); 2193 } 2194 2195 static inline int f2fs_down_write_trylock(struct f2fs_rwsem *sem) 2196 { 2197 return down_write_trylock(&sem->internal_rwsem); 2198 } 2199 2200 static inline void f2fs_up_write(struct f2fs_rwsem *sem) 2201 { 2202 up_write(&sem->internal_rwsem); 2203 #ifdef CONFIG_F2FS_UNFAIR_RWSEM 2204 wake_up_all(&sem->read_waiters); 2205 #endif 2206 } 2207 2208 static inline void f2fs_lock_op(struct f2fs_sb_info *sbi) 2209 { 2210 f2fs_down_read(&sbi->cp_rwsem); 2211 } 2212 2213 static inline int f2fs_trylock_op(struct f2fs_sb_info *sbi) 2214 { 2215 if (time_to_inject(sbi, FAULT_LOCK_OP)) { 2216 f2fs_show_injection_info(sbi, FAULT_LOCK_OP); 2217 return 0; 2218 } 2219 return f2fs_down_read_trylock(&sbi->cp_rwsem); 2220 } 2221 2222 static inline void f2fs_unlock_op(struct f2fs_sb_info *sbi) 2223 { 2224 f2fs_up_read(&sbi->cp_rwsem); 2225 } 2226 2227 static inline void f2fs_lock_all(struct f2fs_sb_info *sbi) 2228 { 2229 f2fs_down_write(&sbi->cp_rwsem); 2230 } 2231 2232 static inline void f2fs_unlock_all(struct f2fs_sb_info *sbi) 2233 { 2234 f2fs_up_write(&sbi->cp_rwsem); 2235 } 2236 2237 static inline int __get_cp_reason(struct f2fs_sb_info *sbi) 2238 { 2239 int reason = CP_SYNC; 2240 2241 if (test_opt(sbi, FASTBOOT)) 2242 reason = CP_FASTBOOT; 2243 if (is_sbi_flag_set(sbi, SBI_IS_CLOSE)) 2244 reason = CP_UMOUNT; 2245 return reason; 2246 } 2247 2248 static inline bool __remain_node_summaries(int reason) 2249 { 2250 return (reason & (CP_UMOUNT | CP_FASTBOOT)); 2251 } 2252 2253 static inline bool __exist_node_summaries(struct f2fs_sb_info *sbi) 2254 { 2255 return (is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG) || 2256 is_set_ckpt_flags(sbi, CP_FASTBOOT_FLAG)); 2257 } 2258 2259 /* 2260 * Check whether the inode has blocks or not 2261 */ 2262 static inline int F2FS_HAS_BLOCKS(struct inode *inode) 2263 { 2264 block_t xattr_block = F2FS_I(inode)->i_xattr_nid ? 1 : 0; 2265 2266 return (inode->i_blocks >> F2FS_LOG_SECTORS_PER_BLOCK) > xattr_block; 2267 } 2268 2269 static inline bool f2fs_has_xattr_block(unsigned int ofs) 2270 { 2271 return ofs == XATTR_NODE_OFFSET; 2272 } 2273 2274 static inline bool __allow_reserved_blocks(struct f2fs_sb_info *sbi, 2275 struct inode *inode, bool cap) 2276 { 2277 if (!inode) 2278 return true; 2279 if (!test_opt(sbi, RESERVE_ROOT)) 2280 return false; 2281 if (IS_NOQUOTA(inode)) 2282 return true; 2283 if (uid_eq(F2FS_OPTION(sbi).s_resuid, current_fsuid())) 2284 return true; 2285 if (!gid_eq(F2FS_OPTION(sbi).s_resgid, GLOBAL_ROOT_GID) && 2286 in_group_p(F2FS_OPTION(sbi).s_resgid)) 2287 return true; 2288 if (cap && capable(CAP_SYS_RESOURCE)) 2289 return true; 2290 return false; 2291 } 2292 2293 static inline void f2fs_i_blocks_write(struct inode *, block_t, bool, bool); 2294 static inline int inc_valid_block_count(struct f2fs_sb_info *sbi, 2295 struct inode *inode, blkcnt_t *count) 2296 { 2297 blkcnt_t diff = 0, release = 0; 2298 block_t avail_user_block_count; 2299 int ret; 2300 2301 ret = dquot_reserve_block(inode, *count); 2302 if (ret) 2303 return ret; 2304 2305 if (time_to_inject(sbi, FAULT_BLOCK)) { 2306 f2fs_show_injection_info(sbi, FAULT_BLOCK); 2307 release = *count; 2308 goto release_quota; 2309 } 2310 2311 /* 2312 * let's increase this in prior to actual block count change in order 2313 * for f2fs_sync_file to avoid data races when deciding checkpoint. 2314 */ 2315 percpu_counter_add(&sbi->alloc_valid_block_count, (*count)); 2316 2317 spin_lock(&sbi->stat_lock); 2318 sbi->total_valid_block_count += (block_t)(*count); 2319 avail_user_block_count = sbi->user_block_count - 2320 sbi->current_reserved_blocks; 2321 2322 if (!__allow_reserved_blocks(sbi, inode, true)) 2323 avail_user_block_count -= F2FS_OPTION(sbi).root_reserved_blocks; 2324 2325 if (F2FS_IO_ALIGNED(sbi)) 2326 avail_user_block_count -= sbi->blocks_per_seg * 2327 SM_I(sbi)->additional_reserved_segments; 2328 2329 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) { 2330 if (avail_user_block_count > sbi->unusable_block_count) 2331 avail_user_block_count -= sbi->unusable_block_count; 2332 else 2333 avail_user_block_count = 0; 2334 } 2335 if (unlikely(sbi->total_valid_block_count > avail_user_block_count)) { 2336 diff = sbi->total_valid_block_count - avail_user_block_count; 2337 if (diff > *count) 2338 diff = *count; 2339 *count -= diff; 2340 release = diff; 2341 sbi->total_valid_block_count -= diff; 2342 if (!*count) { 2343 spin_unlock(&sbi->stat_lock); 2344 goto enospc; 2345 } 2346 } 2347 spin_unlock(&sbi->stat_lock); 2348 2349 if (unlikely(release)) { 2350 percpu_counter_sub(&sbi->alloc_valid_block_count, release); 2351 dquot_release_reservation_block(inode, release); 2352 } 2353 f2fs_i_blocks_write(inode, *count, true, true); 2354 return 0; 2355 2356 enospc: 2357 percpu_counter_sub(&sbi->alloc_valid_block_count, release); 2358 release_quota: 2359 dquot_release_reservation_block(inode, release); 2360 return -ENOSPC; 2361 } 2362 2363 __printf(2, 3) 2364 void f2fs_printk(struct f2fs_sb_info *sbi, const char *fmt, ...); 2365 2366 #define f2fs_err(sbi, fmt, ...) \ 2367 f2fs_printk(sbi, KERN_ERR fmt, ##__VA_ARGS__) 2368 #define f2fs_warn(sbi, fmt, ...) \ 2369 f2fs_printk(sbi, KERN_WARNING fmt, ##__VA_ARGS__) 2370 #define f2fs_notice(sbi, fmt, ...) \ 2371 f2fs_printk(sbi, KERN_NOTICE fmt, ##__VA_ARGS__) 2372 #define f2fs_info(sbi, fmt, ...) \ 2373 f2fs_printk(sbi, KERN_INFO fmt, ##__VA_ARGS__) 2374 #define f2fs_debug(sbi, fmt, ...) \ 2375 f2fs_printk(sbi, KERN_DEBUG fmt, ##__VA_ARGS__) 2376 2377 static inline void dec_valid_block_count(struct f2fs_sb_info *sbi, 2378 struct inode *inode, 2379 block_t count) 2380 { 2381 blkcnt_t sectors = count << F2FS_LOG_SECTORS_PER_BLOCK; 2382 2383 spin_lock(&sbi->stat_lock); 2384 f2fs_bug_on(sbi, sbi->total_valid_block_count < (block_t) count); 2385 sbi->total_valid_block_count -= (block_t)count; 2386 if (sbi->reserved_blocks && 2387 sbi->current_reserved_blocks < sbi->reserved_blocks) 2388 sbi->current_reserved_blocks = min(sbi->reserved_blocks, 2389 sbi->current_reserved_blocks + count); 2390 spin_unlock(&sbi->stat_lock); 2391 if (unlikely(inode->i_blocks < sectors)) { 2392 f2fs_warn(sbi, "Inconsistent i_blocks, ino:%lu, iblocks:%llu, sectors:%llu", 2393 inode->i_ino, 2394 (unsigned long long)inode->i_blocks, 2395 (unsigned long long)sectors); 2396 set_sbi_flag(sbi, SBI_NEED_FSCK); 2397 return; 2398 } 2399 f2fs_i_blocks_write(inode, count, false, true); 2400 } 2401 2402 static inline void inc_page_count(struct f2fs_sb_info *sbi, int count_type) 2403 { 2404 atomic_inc(&sbi->nr_pages[count_type]); 2405 2406 if (count_type == F2FS_DIRTY_DENTS || 2407 count_type == F2FS_DIRTY_NODES || 2408 count_type == F2FS_DIRTY_META || 2409 count_type == F2FS_DIRTY_QDATA || 2410 count_type == F2FS_DIRTY_IMETA) 2411 set_sbi_flag(sbi, SBI_IS_DIRTY); 2412 } 2413 2414 static inline void inode_inc_dirty_pages(struct inode *inode) 2415 { 2416 atomic_inc(&F2FS_I(inode)->dirty_pages); 2417 inc_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ? 2418 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA); 2419 if (IS_NOQUOTA(inode)) 2420 inc_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA); 2421 } 2422 2423 static inline void dec_page_count(struct f2fs_sb_info *sbi, int count_type) 2424 { 2425 atomic_dec(&sbi->nr_pages[count_type]); 2426 } 2427 2428 static inline void inode_dec_dirty_pages(struct inode *inode) 2429 { 2430 if (!S_ISDIR(inode->i_mode) && !S_ISREG(inode->i_mode) && 2431 !S_ISLNK(inode->i_mode)) 2432 return; 2433 2434 atomic_dec(&F2FS_I(inode)->dirty_pages); 2435 dec_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ? 2436 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA); 2437 if (IS_NOQUOTA(inode)) 2438 dec_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA); 2439 } 2440 2441 static inline void inc_atomic_write_cnt(struct inode *inode) 2442 { 2443 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 2444 struct f2fs_inode_info *fi = F2FS_I(inode); 2445 u64 current_write; 2446 2447 fi->atomic_write_cnt++; 2448 atomic64_inc(&sbi->current_atomic_write); 2449 current_write = atomic64_read(&sbi->current_atomic_write); 2450 if (current_write > sbi->peak_atomic_write) 2451 sbi->peak_atomic_write = current_write; 2452 } 2453 2454 static inline void release_atomic_write_cnt(struct inode *inode) 2455 { 2456 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 2457 struct f2fs_inode_info *fi = F2FS_I(inode); 2458 2459 atomic64_sub(fi->atomic_write_cnt, &sbi->current_atomic_write); 2460 fi->atomic_write_cnt = 0; 2461 } 2462 2463 static inline s64 get_pages(struct f2fs_sb_info *sbi, int count_type) 2464 { 2465 return atomic_read(&sbi->nr_pages[count_type]); 2466 } 2467 2468 static inline int get_dirty_pages(struct inode *inode) 2469 { 2470 return atomic_read(&F2FS_I(inode)->dirty_pages); 2471 } 2472 2473 static inline int get_blocktype_secs(struct f2fs_sb_info *sbi, int block_type) 2474 { 2475 unsigned int pages_per_sec = sbi->segs_per_sec * sbi->blocks_per_seg; 2476 unsigned int segs = (get_pages(sbi, block_type) + pages_per_sec - 1) >> 2477 sbi->log_blocks_per_seg; 2478 2479 return segs / sbi->segs_per_sec; 2480 } 2481 2482 static inline block_t valid_user_blocks(struct f2fs_sb_info *sbi) 2483 { 2484 return sbi->total_valid_block_count; 2485 } 2486 2487 static inline block_t discard_blocks(struct f2fs_sb_info *sbi) 2488 { 2489 return sbi->discard_blks; 2490 } 2491 2492 static inline unsigned long __bitmap_size(struct f2fs_sb_info *sbi, int flag) 2493 { 2494 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi); 2495 2496 /* return NAT or SIT bitmap */ 2497 if (flag == NAT_BITMAP) 2498 return le32_to_cpu(ckpt->nat_ver_bitmap_bytesize); 2499 else if (flag == SIT_BITMAP) 2500 return le32_to_cpu(ckpt->sit_ver_bitmap_bytesize); 2501 2502 return 0; 2503 } 2504 2505 static inline block_t __cp_payload(struct f2fs_sb_info *sbi) 2506 { 2507 return le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_payload); 2508 } 2509 2510 static inline void *__bitmap_ptr(struct f2fs_sb_info *sbi, int flag) 2511 { 2512 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi); 2513 void *tmp_ptr = &ckpt->sit_nat_version_bitmap; 2514 int offset; 2515 2516 if (is_set_ckpt_flags(sbi, CP_LARGE_NAT_BITMAP_FLAG)) { 2517 offset = (flag == SIT_BITMAP) ? 2518 le32_to_cpu(ckpt->nat_ver_bitmap_bytesize) : 0; 2519 /* 2520 * if large_nat_bitmap feature is enabled, leave checksum 2521 * protection for all nat/sit bitmaps. 2522 */ 2523 return tmp_ptr + offset + sizeof(__le32); 2524 } 2525 2526 if (__cp_payload(sbi) > 0) { 2527 if (flag == NAT_BITMAP) 2528 return &ckpt->sit_nat_version_bitmap; 2529 else 2530 return (unsigned char *)ckpt + F2FS_BLKSIZE; 2531 } else { 2532 offset = (flag == NAT_BITMAP) ? 2533 le32_to_cpu(ckpt->sit_ver_bitmap_bytesize) : 0; 2534 return tmp_ptr + offset; 2535 } 2536 } 2537 2538 static inline block_t __start_cp_addr(struct f2fs_sb_info *sbi) 2539 { 2540 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr); 2541 2542 if (sbi->cur_cp_pack == 2) 2543 start_addr += sbi->blocks_per_seg; 2544 return start_addr; 2545 } 2546 2547 static inline block_t __start_cp_next_addr(struct f2fs_sb_info *sbi) 2548 { 2549 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr); 2550 2551 if (sbi->cur_cp_pack == 1) 2552 start_addr += sbi->blocks_per_seg; 2553 return start_addr; 2554 } 2555 2556 static inline void __set_cp_next_pack(struct f2fs_sb_info *sbi) 2557 { 2558 sbi->cur_cp_pack = (sbi->cur_cp_pack == 1) ? 2 : 1; 2559 } 2560 2561 static inline block_t __start_sum_addr(struct f2fs_sb_info *sbi) 2562 { 2563 return le32_to_cpu(F2FS_CKPT(sbi)->cp_pack_start_sum); 2564 } 2565 2566 static inline int inc_valid_node_count(struct f2fs_sb_info *sbi, 2567 struct inode *inode, bool is_inode) 2568 { 2569 block_t valid_block_count; 2570 unsigned int valid_node_count, user_block_count; 2571 int err; 2572 2573 if (is_inode) { 2574 if (inode) { 2575 err = dquot_alloc_inode(inode); 2576 if (err) 2577 return err; 2578 } 2579 } else { 2580 err = dquot_reserve_block(inode, 1); 2581 if (err) 2582 return err; 2583 } 2584 2585 if (time_to_inject(sbi, FAULT_BLOCK)) { 2586 f2fs_show_injection_info(sbi, FAULT_BLOCK); 2587 goto enospc; 2588 } 2589 2590 spin_lock(&sbi->stat_lock); 2591 2592 valid_block_count = sbi->total_valid_block_count + 2593 sbi->current_reserved_blocks + 1; 2594 2595 if (!__allow_reserved_blocks(sbi, inode, false)) 2596 valid_block_count += F2FS_OPTION(sbi).root_reserved_blocks; 2597 2598 if (F2FS_IO_ALIGNED(sbi)) 2599 valid_block_count += sbi->blocks_per_seg * 2600 SM_I(sbi)->additional_reserved_segments; 2601 2602 user_block_count = sbi->user_block_count; 2603 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) 2604 user_block_count -= sbi->unusable_block_count; 2605 2606 if (unlikely(valid_block_count > user_block_count)) { 2607 spin_unlock(&sbi->stat_lock); 2608 goto enospc; 2609 } 2610 2611 valid_node_count = sbi->total_valid_node_count + 1; 2612 if (unlikely(valid_node_count > sbi->total_node_count)) { 2613 spin_unlock(&sbi->stat_lock); 2614 goto enospc; 2615 } 2616 2617 sbi->total_valid_node_count++; 2618 sbi->total_valid_block_count++; 2619 spin_unlock(&sbi->stat_lock); 2620 2621 if (inode) { 2622 if (is_inode) 2623 f2fs_mark_inode_dirty_sync(inode, true); 2624 else 2625 f2fs_i_blocks_write(inode, 1, true, true); 2626 } 2627 2628 percpu_counter_inc(&sbi->alloc_valid_block_count); 2629 return 0; 2630 2631 enospc: 2632 if (is_inode) { 2633 if (inode) 2634 dquot_free_inode(inode); 2635 } else { 2636 dquot_release_reservation_block(inode, 1); 2637 } 2638 return -ENOSPC; 2639 } 2640 2641 static inline void dec_valid_node_count(struct f2fs_sb_info *sbi, 2642 struct inode *inode, bool is_inode) 2643 { 2644 spin_lock(&sbi->stat_lock); 2645 2646 if (unlikely(!sbi->total_valid_block_count || 2647 !sbi->total_valid_node_count)) { 2648 f2fs_warn(sbi, "dec_valid_node_count: inconsistent block counts, total_valid_block:%u, total_valid_node:%u", 2649 sbi->total_valid_block_count, 2650 sbi->total_valid_node_count); 2651 set_sbi_flag(sbi, SBI_NEED_FSCK); 2652 } else { 2653 sbi->total_valid_block_count--; 2654 sbi->total_valid_node_count--; 2655 } 2656 2657 if (sbi->reserved_blocks && 2658 sbi->current_reserved_blocks < sbi->reserved_blocks) 2659 sbi->current_reserved_blocks++; 2660 2661 spin_unlock(&sbi->stat_lock); 2662 2663 if (is_inode) { 2664 dquot_free_inode(inode); 2665 } else { 2666 if (unlikely(inode->i_blocks == 0)) { 2667 f2fs_warn(sbi, "dec_valid_node_count: inconsistent i_blocks, ino:%lu, iblocks:%llu", 2668 inode->i_ino, 2669 (unsigned long long)inode->i_blocks); 2670 set_sbi_flag(sbi, SBI_NEED_FSCK); 2671 return; 2672 } 2673 f2fs_i_blocks_write(inode, 1, false, true); 2674 } 2675 } 2676 2677 static inline unsigned int valid_node_count(struct f2fs_sb_info *sbi) 2678 { 2679 return sbi->total_valid_node_count; 2680 } 2681 2682 static inline void inc_valid_inode_count(struct f2fs_sb_info *sbi) 2683 { 2684 percpu_counter_inc(&sbi->total_valid_inode_count); 2685 } 2686 2687 static inline void dec_valid_inode_count(struct f2fs_sb_info *sbi) 2688 { 2689 percpu_counter_dec(&sbi->total_valid_inode_count); 2690 } 2691 2692 static inline s64 valid_inode_count(struct f2fs_sb_info *sbi) 2693 { 2694 return percpu_counter_sum_positive(&sbi->total_valid_inode_count); 2695 } 2696 2697 static inline struct page *f2fs_grab_cache_page(struct address_space *mapping, 2698 pgoff_t index, bool for_write) 2699 { 2700 struct page *page; 2701 unsigned int flags; 2702 2703 if (IS_ENABLED(CONFIG_F2FS_FAULT_INJECTION)) { 2704 if (!for_write) 2705 page = find_get_page_flags(mapping, index, 2706 FGP_LOCK | FGP_ACCESSED); 2707 else 2708 page = find_lock_page(mapping, index); 2709 if (page) 2710 return page; 2711 2712 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_ALLOC)) { 2713 f2fs_show_injection_info(F2FS_M_SB(mapping), 2714 FAULT_PAGE_ALLOC); 2715 return NULL; 2716 } 2717 } 2718 2719 if (!for_write) 2720 return grab_cache_page(mapping, index); 2721 2722 flags = memalloc_nofs_save(); 2723 page = grab_cache_page_write_begin(mapping, index); 2724 memalloc_nofs_restore(flags); 2725 2726 return page; 2727 } 2728 2729 static inline struct page *f2fs_pagecache_get_page( 2730 struct address_space *mapping, pgoff_t index, 2731 int fgp_flags, gfp_t gfp_mask) 2732 { 2733 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_GET)) { 2734 f2fs_show_injection_info(F2FS_M_SB(mapping), FAULT_PAGE_GET); 2735 return NULL; 2736 } 2737 2738 return pagecache_get_page(mapping, index, fgp_flags, gfp_mask); 2739 } 2740 2741 static inline void f2fs_put_page(struct page *page, int unlock) 2742 { 2743 if (!page) 2744 return; 2745 2746 if (unlock) { 2747 f2fs_bug_on(F2FS_P_SB(page), !PageLocked(page)); 2748 unlock_page(page); 2749 } 2750 put_page(page); 2751 } 2752 2753 static inline void f2fs_put_dnode(struct dnode_of_data *dn) 2754 { 2755 if (dn->node_page) 2756 f2fs_put_page(dn->node_page, 1); 2757 if (dn->inode_page && dn->node_page != dn->inode_page) 2758 f2fs_put_page(dn->inode_page, 0); 2759 dn->node_page = NULL; 2760 dn->inode_page = NULL; 2761 } 2762 2763 static inline struct kmem_cache *f2fs_kmem_cache_create(const char *name, 2764 size_t size) 2765 { 2766 return kmem_cache_create(name, size, 0, SLAB_RECLAIM_ACCOUNT, NULL); 2767 } 2768 2769 static inline void *f2fs_kmem_cache_alloc_nofail(struct kmem_cache *cachep, 2770 gfp_t flags) 2771 { 2772 void *entry; 2773 2774 entry = kmem_cache_alloc(cachep, flags); 2775 if (!entry) 2776 entry = kmem_cache_alloc(cachep, flags | __GFP_NOFAIL); 2777 return entry; 2778 } 2779 2780 static inline void *f2fs_kmem_cache_alloc(struct kmem_cache *cachep, 2781 gfp_t flags, bool nofail, struct f2fs_sb_info *sbi) 2782 { 2783 if (nofail) 2784 return f2fs_kmem_cache_alloc_nofail(cachep, flags); 2785 2786 if (time_to_inject(sbi, FAULT_SLAB_ALLOC)) { 2787 f2fs_show_injection_info(sbi, FAULT_SLAB_ALLOC); 2788 return NULL; 2789 } 2790 2791 return kmem_cache_alloc(cachep, flags); 2792 } 2793 2794 static inline bool is_inflight_io(struct f2fs_sb_info *sbi, int type) 2795 { 2796 if (get_pages(sbi, F2FS_RD_DATA) || get_pages(sbi, F2FS_RD_NODE) || 2797 get_pages(sbi, F2FS_RD_META) || get_pages(sbi, F2FS_WB_DATA) || 2798 get_pages(sbi, F2FS_WB_CP_DATA) || 2799 get_pages(sbi, F2FS_DIO_READ) || 2800 get_pages(sbi, F2FS_DIO_WRITE)) 2801 return true; 2802 2803 if (type != DISCARD_TIME && SM_I(sbi) && SM_I(sbi)->dcc_info && 2804 atomic_read(&SM_I(sbi)->dcc_info->queued_discard)) 2805 return true; 2806 2807 if (SM_I(sbi) && SM_I(sbi)->fcc_info && 2808 atomic_read(&SM_I(sbi)->fcc_info->queued_flush)) 2809 return true; 2810 return false; 2811 } 2812 2813 static inline bool is_idle(struct f2fs_sb_info *sbi, int type) 2814 { 2815 if (sbi->gc_mode == GC_URGENT_HIGH) 2816 return true; 2817 2818 if (is_inflight_io(sbi, type)) 2819 return false; 2820 2821 if (sbi->gc_mode == GC_URGENT_MID) 2822 return true; 2823 2824 if (sbi->gc_mode == GC_URGENT_LOW && 2825 (type == DISCARD_TIME || type == GC_TIME)) 2826 return true; 2827 2828 return f2fs_time_over(sbi, type); 2829 } 2830 2831 static inline void f2fs_radix_tree_insert(struct radix_tree_root *root, 2832 unsigned long index, void *item) 2833 { 2834 while (radix_tree_insert(root, index, item)) 2835 cond_resched(); 2836 } 2837 2838 #define RAW_IS_INODE(p) ((p)->footer.nid == (p)->footer.ino) 2839 2840 static inline bool IS_INODE(struct page *page) 2841 { 2842 struct f2fs_node *p = F2FS_NODE(page); 2843 2844 return RAW_IS_INODE(p); 2845 } 2846 2847 static inline int offset_in_addr(struct f2fs_inode *i) 2848 { 2849 return (i->i_inline & F2FS_EXTRA_ATTR) ? 2850 (le16_to_cpu(i->i_extra_isize) / sizeof(__le32)) : 0; 2851 } 2852 2853 static inline __le32 *blkaddr_in_node(struct f2fs_node *node) 2854 { 2855 return RAW_IS_INODE(node) ? node->i.i_addr : node->dn.addr; 2856 } 2857 2858 static inline int f2fs_has_extra_attr(struct inode *inode); 2859 static inline block_t data_blkaddr(struct inode *inode, 2860 struct page *node_page, unsigned int offset) 2861 { 2862 struct f2fs_node *raw_node; 2863 __le32 *addr_array; 2864 int base = 0; 2865 bool is_inode = IS_INODE(node_page); 2866 2867 raw_node = F2FS_NODE(node_page); 2868 2869 if (is_inode) { 2870 if (!inode) 2871 /* from GC path only */ 2872 base = offset_in_addr(&raw_node->i); 2873 else if (f2fs_has_extra_attr(inode)) 2874 base = get_extra_isize(inode); 2875 } 2876 2877 addr_array = blkaddr_in_node(raw_node); 2878 return le32_to_cpu(addr_array[base + offset]); 2879 } 2880 2881 static inline block_t f2fs_data_blkaddr(struct dnode_of_data *dn) 2882 { 2883 return data_blkaddr(dn->inode, dn->node_page, dn->ofs_in_node); 2884 } 2885 2886 static inline int f2fs_test_bit(unsigned int nr, char *addr) 2887 { 2888 int mask; 2889 2890 addr += (nr >> 3); 2891 mask = 1 << (7 - (nr & 0x07)); 2892 return mask & *addr; 2893 } 2894 2895 static inline void f2fs_set_bit(unsigned int nr, char *addr) 2896 { 2897 int mask; 2898 2899 addr += (nr >> 3); 2900 mask = 1 << (7 - (nr & 0x07)); 2901 *addr |= mask; 2902 } 2903 2904 static inline void f2fs_clear_bit(unsigned int nr, char *addr) 2905 { 2906 int mask; 2907 2908 addr += (nr >> 3); 2909 mask = 1 << (7 - (nr & 0x07)); 2910 *addr &= ~mask; 2911 } 2912 2913 static inline int f2fs_test_and_set_bit(unsigned int nr, char *addr) 2914 { 2915 int mask; 2916 int ret; 2917 2918 addr += (nr >> 3); 2919 mask = 1 << (7 - (nr & 0x07)); 2920 ret = mask & *addr; 2921 *addr |= mask; 2922 return ret; 2923 } 2924 2925 static inline int f2fs_test_and_clear_bit(unsigned int nr, char *addr) 2926 { 2927 int mask; 2928 int ret; 2929 2930 addr += (nr >> 3); 2931 mask = 1 << (7 - (nr & 0x07)); 2932 ret = mask & *addr; 2933 *addr &= ~mask; 2934 return ret; 2935 } 2936 2937 static inline void f2fs_change_bit(unsigned int nr, char *addr) 2938 { 2939 int mask; 2940 2941 addr += (nr >> 3); 2942 mask = 1 << (7 - (nr & 0x07)); 2943 *addr ^= mask; 2944 } 2945 2946 /* 2947 * On-disk inode flags (f2fs_inode::i_flags) 2948 */ 2949 #define F2FS_COMPR_FL 0x00000004 /* Compress file */ 2950 #define F2FS_SYNC_FL 0x00000008 /* Synchronous updates */ 2951 #define F2FS_IMMUTABLE_FL 0x00000010 /* Immutable file */ 2952 #define F2FS_APPEND_FL 0x00000020 /* writes to file may only append */ 2953 #define F2FS_NODUMP_FL 0x00000040 /* do not dump file */ 2954 #define F2FS_NOATIME_FL 0x00000080 /* do not update atime */ 2955 #define F2FS_NOCOMP_FL 0x00000400 /* Don't compress */ 2956 #define F2FS_INDEX_FL 0x00001000 /* hash-indexed directory */ 2957 #define F2FS_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */ 2958 #define F2FS_PROJINHERIT_FL 0x20000000 /* Create with parents projid */ 2959 #define F2FS_CASEFOLD_FL 0x40000000 /* Casefolded file */ 2960 2961 /* Flags that should be inherited by new inodes from their parent. */ 2962 #define F2FS_FL_INHERITED (F2FS_SYNC_FL | F2FS_NODUMP_FL | F2FS_NOATIME_FL | \ 2963 F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \ 2964 F2FS_CASEFOLD_FL | F2FS_COMPR_FL | F2FS_NOCOMP_FL) 2965 2966 /* Flags that are appropriate for regular files (all but dir-specific ones). */ 2967 #define F2FS_REG_FLMASK (~(F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \ 2968 F2FS_CASEFOLD_FL)) 2969 2970 /* Flags that are appropriate for non-directories/regular files. */ 2971 #define F2FS_OTHER_FLMASK (F2FS_NODUMP_FL | F2FS_NOATIME_FL) 2972 2973 static inline __u32 f2fs_mask_flags(umode_t mode, __u32 flags) 2974 { 2975 if (S_ISDIR(mode)) 2976 return flags; 2977 else if (S_ISREG(mode)) 2978 return flags & F2FS_REG_FLMASK; 2979 else 2980 return flags & F2FS_OTHER_FLMASK; 2981 } 2982 2983 static inline void __mark_inode_dirty_flag(struct inode *inode, 2984 int flag, bool set) 2985 { 2986 switch (flag) { 2987 case FI_INLINE_XATTR: 2988 case FI_INLINE_DATA: 2989 case FI_INLINE_DENTRY: 2990 case FI_NEW_INODE: 2991 if (set) 2992 return; 2993 fallthrough; 2994 case FI_DATA_EXIST: 2995 case FI_INLINE_DOTS: 2996 case FI_PIN_FILE: 2997 case FI_COMPRESS_RELEASED: 2998 f2fs_mark_inode_dirty_sync(inode, true); 2999 } 3000 } 3001 3002 static inline void set_inode_flag(struct inode *inode, int flag) 3003 { 3004 set_bit(flag, F2FS_I(inode)->flags); 3005 __mark_inode_dirty_flag(inode, flag, true); 3006 } 3007 3008 static inline int is_inode_flag_set(struct inode *inode, int flag) 3009 { 3010 return test_bit(flag, F2FS_I(inode)->flags); 3011 } 3012 3013 static inline void clear_inode_flag(struct inode *inode, int flag) 3014 { 3015 clear_bit(flag, F2FS_I(inode)->flags); 3016 __mark_inode_dirty_flag(inode, flag, false); 3017 } 3018 3019 static inline bool f2fs_verity_in_progress(struct inode *inode) 3020 { 3021 return IS_ENABLED(CONFIG_FS_VERITY) && 3022 is_inode_flag_set(inode, FI_VERITY_IN_PROGRESS); 3023 } 3024 3025 static inline void set_acl_inode(struct inode *inode, umode_t mode) 3026 { 3027 F2FS_I(inode)->i_acl_mode = mode; 3028 set_inode_flag(inode, FI_ACL_MODE); 3029 f2fs_mark_inode_dirty_sync(inode, false); 3030 } 3031 3032 static inline void f2fs_i_links_write(struct inode *inode, bool inc) 3033 { 3034 if (inc) 3035 inc_nlink(inode); 3036 else 3037 drop_nlink(inode); 3038 f2fs_mark_inode_dirty_sync(inode, true); 3039 } 3040 3041 static inline void f2fs_i_blocks_write(struct inode *inode, 3042 block_t diff, bool add, bool claim) 3043 { 3044 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE); 3045 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER); 3046 3047 /* add = 1, claim = 1 should be dquot_reserve_block in pair */ 3048 if (add) { 3049 if (claim) 3050 dquot_claim_block(inode, diff); 3051 else 3052 dquot_alloc_block_nofail(inode, diff); 3053 } else { 3054 dquot_free_block(inode, diff); 3055 } 3056 3057 f2fs_mark_inode_dirty_sync(inode, true); 3058 if (clean || recover) 3059 set_inode_flag(inode, FI_AUTO_RECOVER); 3060 } 3061 3062 static inline void f2fs_i_size_write(struct inode *inode, loff_t i_size) 3063 { 3064 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE); 3065 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER); 3066 3067 if (i_size_read(inode) == i_size) 3068 return; 3069 3070 i_size_write(inode, i_size); 3071 f2fs_mark_inode_dirty_sync(inode, true); 3072 if (clean || recover) 3073 set_inode_flag(inode, FI_AUTO_RECOVER); 3074 } 3075 3076 static inline void f2fs_i_depth_write(struct inode *inode, unsigned int depth) 3077 { 3078 F2FS_I(inode)->i_current_depth = depth; 3079 f2fs_mark_inode_dirty_sync(inode, true); 3080 } 3081 3082 static inline void f2fs_i_gc_failures_write(struct inode *inode, 3083 unsigned int count) 3084 { 3085 F2FS_I(inode)->i_gc_failures[GC_FAILURE_PIN] = count; 3086 f2fs_mark_inode_dirty_sync(inode, true); 3087 } 3088 3089 static inline void f2fs_i_xnid_write(struct inode *inode, nid_t xnid) 3090 { 3091 F2FS_I(inode)->i_xattr_nid = xnid; 3092 f2fs_mark_inode_dirty_sync(inode, true); 3093 } 3094 3095 static inline void f2fs_i_pino_write(struct inode *inode, nid_t pino) 3096 { 3097 F2FS_I(inode)->i_pino = pino; 3098 f2fs_mark_inode_dirty_sync(inode, true); 3099 } 3100 3101 static inline void get_inline_info(struct inode *inode, struct f2fs_inode *ri) 3102 { 3103 struct f2fs_inode_info *fi = F2FS_I(inode); 3104 3105 if (ri->i_inline & F2FS_INLINE_XATTR) 3106 set_bit(FI_INLINE_XATTR, fi->flags); 3107 if (ri->i_inline & F2FS_INLINE_DATA) 3108 set_bit(FI_INLINE_DATA, fi->flags); 3109 if (ri->i_inline & F2FS_INLINE_DENTRY) 3110 set_bit(FI_INLINE_DENTRY, fi->flags); 3111 if (ri->i_inline & F2FS_DATA_EXIST) 3112 set_bit(FI_DATA_EXIST, fi->flags); 3113 if (ri->i_inline & F2FS_INLINE_DOTS) 3114 set_bit(FI_INLINE_DOTS, fi->flags); 3115 if (ri->i_inline & F2FS_EXTRA_ATTR) 3116 set_bit(FI_EXTRA_ATTR, fi->flags); 3117 if (ri->i_inline & F2FS_PIN_FILE) 3118 set_bit(FI_PIN_FILE, fi->flags); 3119 if (ri->i_inline & F2FS_COMPRESS_RELEASED) 3120 set_bit(FI_COMPRESS_RELEASED, fi->flags); 3121 } 3122 3123 static inline void set_raw_inline(struct inode *inode, struct f2fs_inode *ri) 3124 { 3125 ri->i_inline = 0; 3126 3127 if (is_inode_flag_set(inode, FI_INLINE_XATTR)) 3128 ri->i_inline |= F2FS_INLINE_XATTR; 3129 if (is_inode_flag_set(inode, FI_INLINE_DATA)) 3130 ri->i_inline |= F2FS_INLINE_DATA; 3131 if (is_inode_flag_set(inode, FI_INLINE_DENTRY)) 3132 ri->i_inline |= F2FS_INLINE_DENTRY; 3133 if (is_inode_flag_set(inode, FI_DATA_EXIST)) 3134 ri->i_inline |= F2FS_DATA_EXIST; 3135 if (is_inode_flag_set(inode, FI_INLINE_DOTS)) 3136 ri->i_inline |= F2FS_INLINE_DOTS; 3137 if (is_inode_flag_set(inode, FI_EXTRA_ATTR)) 3138 ri->i_inline |= F2FS_EXTRA_ATTR; 3139 if (is_inode_flag_set(inode, FI_PIN_FILE)) 3140 ri->i_inline |= F2FS_PIN_FILE; 3141 if (is_inode_flag_set(inode, FI_COMPRESS_RELEASED)) 3142 ri->i_inline |= F2FS_COMPRESS_RELEASED; 3143 } 3144 3145 static inline int f2fs_has_extra_attr(struct inode *inode) 3146 { 3147 return is_inode_flag_set(inode, FI_EXTRA_ATTR); 3148 } 3149 3150 static inline int f2fs_has_inline_xattr(struct inode *inode) 3151 { 3152 return is_inode_flag_set(inode, FI_INLINE_XATTR); 3153 } 3154 3155 static inline int f2fs_compressed_file(struct inode *inode) 3156 { 3157 return S_ISREG(inode->i_mode) && 3158 is_inode_flag_set(inode, FI_COMPRESSED_FILE); 3159 } 3160 3161 static inline bool f2fs_need_compress_data(struct inode *inode) 3162 { 3163 int compress_mode = F2FS_OPTION(F2FS_I_SB(inode)).compress_mode; 3164 3165 if (!f2fs_compressed_file(inode)) 3166 return false; 3167 3168 if (compress_mode == COMPR_MODE_FS) 3169 return true; 3170 else if (compress_mode == COMPR_MODE_USER && 3171 is_inode_flag_set(inode, FI_ENABLE_COMPRESS)) 3172 return true; 3173 3174 return false; 3175 } 3176 3177 static inline unsigned int addrs_per_inode(struct inode *inode) 3178 { 3179 unsigned int addrs = CUR_ADDRS_PER_INODE(inode) - 3180 get_inline_xattr_addrs(inode); 3181 3182 if (!f2fs_compressed_file(inode)) 3183 return addrs; 3184 return ALIGN_DOWN(addrs, F2FS_I(inode)->i_cluster_size); 3185 } 3186 3187 static inline unsigned int addrs_per_block(struct inode *inode) 3188 { 3189 if (!f2fs_compressed_file(inode)) 3190 return DEF_ADDRS_PER_BLOCK; 3191 return ALIGN_DOWN(DEF_ADDRS_PER_BLOCK, F2FS_I(inode)->i_cluster_size); 3192 } 3193 3194 static inline void *inline_xattr_addr(struct inode *inode, struct page *page) 3195 { 3196 struct f2fs_inode *ri = F2FS_INODE(page); 3197 3198 return (void *)&(ri->i_addr[DEF_ADDRS_PER_INODE - 3199 get_inline_xattr_addrs(inode)]); 3200 } 3201 3202 static inline int inline_xattr_size(struct inode *inode) 3203 { 3204 if (f2fs_has_inline_xattr(inode)) 3205 return get_inline_xattr_addrs(inode) * sizeof(__le32); 3206 return 0; 3207 } 3208 3209 /* 3210 * Notice: check inline_data flag without inode page lock is unsafe. 3211 * It could change at any time by f2fs_convert_inline_page(). 3212 */ 3213 static inline int f2fs_has_inline_data(struct inode *inode) 3214 { 3215 return is_inode_flag_set(inode, FI_INLINE_DATA); 3216 } 3217 3218 static inline int f2fs_exist_data(struct inode *inode) 3219 { 3220 return is_inode_flag_set(inode, FI_DATA_EXIST); 3221 } 3222 3223 static inline int f2fs_has_inline_dots(struct inode *inode) 3224 { 3225 return is_inode_flag_set(inode, FI_INLINE_DOTS); 3226 } 3227 3228 static inline int f2fs_is_mmap_file(struct inode *inode) 3229 { 3230 return is_inode_flag_set(inode, FI_MMAP_FILE); 3231 } 3232 3233 static inline bool f2fs_is_pinned_file(struct inode *inode) 3234 { 3235 return is_inode_flag_set(inode, FI_PIN_FILE); 3236 } 3237 3238 static inline bool f2fs_is_atomic_file(struct inode *inode) 3239 { 3240 return is_inode_flag_set(inode, FI_ATOMIC_FILE); 3241 } 3242 3243 static inline bool f2fs_is_cow_file(struct inode *inode) 3244 { 3245 return is_inode_flag_set(inode, FI_COW_FILE); 3246 } 3247 3248 static inline bool f2fs_is_first_block_written(struct inode *inode) 3249 { 3250 return is_inode_flag_set(inode, FI_FIRST_BLOCK_WRITTEN); 3251 } 3252 3253 static inline bool f2fs_is_drop_cache(struct inode *inode) 3254 { 3255 return is_inode_flag_set(inode, FI_DROP_CACHE); 3256 } 3257 3258 static inline void *inline_data_addr(struct inode *inode, struct page *page) 3259 { 3260 struct f2fs_inode *ri = F2FS_INODE(page); 3261 int extra_size = get_extra_isize(inode); 3262 3263 return (void *)&(ri->i_addr[extra_size + DEF_INLINE_RESERVED_SIZE]); 3264 } 3265 3266 static inline int f2fs_has_inline_dentry(struct inode *inode) 3267 { 3268 return is_inode_flag_set(inode, FI_INLINE_DENTRY); 3269 } 3270 3271 static inline int is_file(struct inode *inode, int type) 3272 { 3273 return F2FS_I(inode)->i_advise & type; 3274 } 3275 3276 static inline void set_file(struct inode *inode, int type) 3277 { 3278 if (is_file(inode, type)) 3279 return; 3280 F2FS_I(inode)->i_advise |= type; 3281 f2fs_mark_inode_dirty_sync(inode, true); 3282 } 3283 3284 static inline void clear_file(struct inode *inode, int type) 3285 { 3286 if (!is_file(inode, type)) 3287 return; 3288 F2FS_I(inode)->i_advise &= ~type; 3289 f2fs_mark_inode_dirty_sync(inode, true); 3290 } 3291 3292 static inline bool f2fs_is_time_consistent(struct inode *inode) 3293 { 3294 if (!timespec64_equal(F2FS_I(inode)->i_disk_time, &inode->i_atime)) 3295 return false; 3296 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 1, &inode->i_ctime)) 3297 return false; 3298 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 2, &inode->i_mtime)) 3299 return false; 3300 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 3, 3301 &F2FS_I(inode)->i_crtime)) 3302 return false; 3303 return true; 3304 } 3305 3306 static inline bool f2fs_skip_inode_update(struct inode *inode, int dsync) 3307 { 3308 bool ret; 3309 3310 if (dsync) { 3311 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 3312 3313 spin_lock(&sbi->inode_lock[DIRTY_META]); 3314 ret = list_empty(&F2FS_I(inode)->gdirty_list); 3315 spin_unlock(&sbi->inode_lock[DIRTY_META]); 3316 return ret; 3317 } 3318 if (!is_inode_flag_set(inode, FI_AUTO_RECOVER) || 3319 file_keep_isize(inode) || 3320 i_size_read(inode) & ~PAGE_MASK) 3321 return false; 3322 3323 if (!f2fs_is_time_consistent(inode)) 3324 return false; 3325 3326 spin_lock(&F2FS_I(inode)->i_size_lock); 3327 ret = F2FS_I(inode)->last_disk_size == i_size_read(inode); 3328 spin_unlock(&F2FS_I(inode)->i_size_lock); 3329 3330 return ret; 3331 } 3332 3333 static inline bool f2fs_readonly(struct super_block *sb) 3334 { 3335 return sb_rdonly(sb); 3336 } 3337 3338 static inline bool f2fs_cp_error(struct f2fs_sb_info *sbi) 3339 { 3340 return is_set_ckpt_flags(sbi, CP_ERROR_FLAG); 3341 } 3342 3343 static inline bool is_dot_dotdot(const u8 *name, size_t len) 3344 { 3345 if (len == 1 && name[0] == '.') 3346 return true; 3347 3348 if (len == 2 && name[0] == '.' && name[1] == '.') 3349 return true; 3350 3351 return false; 3352 } 3353 3354 static inline void *f2fs_kmalloc(struct f2fs_sb_info *sbi, 3355 size_t size, gfp_t flags) 3356 { 3357 if (time_to_inject(sbi, FAULT_KMALLOC)) { 3358 f2fs_show_injection_info(sbi, FAULT_KMALLOC); 3359 return NULL; 3360 } 3361 3362 return kmalloc(size, flags); 3363 } 3364 3365 static inline void *f2fs_kzalloc(struct f2fs_sb_info *sbi, 3366 size_t size, gfp_t flags) 3367 { 3368 return f2fs_kmalloc(sbi, size, flags | __GFP_ZERO); 3369 } 3370 3371 static inline void *f2fs_kvmalloc(struct f2fs_sb_info *sbi, 3372 size_t size, gfp_t flags) 3373 { 3374 if (time_to_inject(sbi, FAULT_KVMALLOC)) { 3375 f2fs_show_injection_info(sbi, FAULT_KVMALLOC); 3376 return NULL; 3377 } 3378 3379 return kvmalloc(size, flags); 3380 } 3381 3382 static inline void *f2fs_kvzalloc(struct f2fs_sb_info *sbi, 3383 size_t size, gfp_t flags) 3384 { 3385 return f2fs_kvmalloc(sbi, size, flags | __GFP_ZERO); 3386 } 3387 3388 static inline int get_extra_isize(struct inode *inode) 3389 { 3390 return F2FS_I(inode)->i_extra_isize / sizeof(__le32); 3391 } 3392 3393 static inline int get_inline_xattr_addrs(struct inode *inode) 3394 { 3395 return F2FS_I(inode)->i_inline_xattr_size; 3396 } 3397 3398 #define f2fs_get_inode_mode(i) \ 3399 ((is_inode_flag_set(i, FI_ACL_MODE)) ? \ 3400 (F2FS_I(i)->i_acl_mode) : ((i)->i_mode)) 3401 3402 #define F2FS_TOTAL_EXTRA_ATTR_SIZE \ 3403 (offsetof(struct f2fs_inode, i_extra_end) - \ 3404 offsetof(struct f2fs_inode, i_extra_isize)) \ 3405 3406 #define F2FS_OLD_ATTRIBUTE_SIZE (offsetof(struct f2fs_inode, i_addr)) 3407 #define F2FS_FITS_IN_INODE(f2fs_inode, extra_isize, field) \ 3408 ((offsetof(typeof(*(f2fs_inode)), field) + \ 3409 sizeof((f2fs_inode)->field)) \ 3410 <= (F2FS_OLD_ATTRIBUTE_SIZE + (extra_isize))) \ 3411 3412 #define __is_large_section(sbi) ((sbi)->segs_per_sec > 1) 3413 3414 #define __is_meta_io(fio) (PAGE_TYPE_OF_BIO((fio)->type) == META) 3415 3416 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi, 3417 block_t blkaddr, int type); 3418 static inline void verify_blkaddr(struct f2fs_sb_info *sbi, 3419 block_t blkaddr, int type) 3420 { 3421 if (!f2fs_is_valid_blkaddr(sbi, blkaddr, type)) { 3422 f2fs_err(sbi, "invalid blkaddr: %u, type: %d, run fsck to fix.", 3423 blkaddr, type); 3424 f2fs_bug_on(sbi, 1); 3425 } 3426 } 3427 3428 static inline bool __is_valid_data_blkaddr(block_t blkaddr) 3429 { 3430 if (blkaddr == NEW_ADDR || blkaddr == NULL_ADDR || 3431 blkaddr == COMPRESS_ADDR) 3432 return false; 3433 return true; 3434 } 3435 3436 /* 3437 * file.c 3438 */ 3439 int f2fs_sync_file(struct file *file, loff_t start, loff_t end, int datasync); 3440 void f2fs_truncate_data_blocks(struct dnode_of_data *dn); 3441 int f2fs_do_truncate_blocks(struct inode *inode, u64 from, bool lock); 3442 int f2fs_truncate_blocks(struct inode *inode, u64 from, bool lock); 3443 int f2fs_truncate(struct inode *inode); 3444 int f2fs_getattr(struct user_namespace *mnt_userns, const struct path *path, 3445 struct kstat *stat, u32 request_mask, unsigned int flags); 3446 int f2fs_setattr(struct user_namespace *mnt_userns, struct dentry *dentry, 3447 struct iattr *attr); 3448 int f2fs_truncate_hole(struct inode *inode, pgoff_t pg_start, pgoff_t pg_end); 3449 void f2fs_truncate_data_blocks_range(struct dnode_of_data *dn, int count); 3450 int f2fs_precache_extents(struct inode *inode); 3451 int f2fs_fileattr_get(struct dentry *dentry, struct fileattr *fa); 3452 int f2fs_fileattr_set(struct user_namespace *mnt_userns, 3453 struct dentry *dentry, struct fileattr *fa); 3454 long f2fs_ioctl(struct file *filp, unsigned int cmd, unsigned long arg); 3455 long f2fs_compat_ioctl(struct file *file, unsigned int cmd, unsigned long arg); 3456 int f2fs_transfer_project_quota(struct inode *inode, kprojid_t kprojid); 3457 int f2fs_pin_file_control(struct inode *inode, bool inc); 3458 3459 /* 3460 * inode.c 3461 */ 3462 void f2fs_set_inode_flags(struct inode *inode); 3463 bool f2fs_inode_chksum_verify(struct f2fs_sb_info *sbi, struct page *page); 3464 void f2fs_inode_chksum_set(struct f2fs_sb_info *sbi, struct page *page); 3465 struct inode *f2fs_iget(struct super_block *sb, unsigned long ino); 3466 struct inode *f2fs_iget_retry(struct super_block *sb, unsigned long ino); 3467 int f2fs_try_to_free_nats(struct f2fs_sb_info *sbi, int nr_shrink); 3468 void f2fs_update_inode(struct inode *inode, struct page *node_page); 3469 void f2fs_update_inode_page(struct inode *inode); 3470 int f2fs_write_inode(struct inode *inode, struct writeback_control *wbc); 3471 void f2fs_evict_inode(struct inode *inode); 3472 void f2fs_handle_failed_inode(struct inode *inode); 3473 3474 /* 3475 * namei.c 3476 */ 3477 int f2fs_update_extension_list(struct f2fs_sb_info *sbi, const char *name, 3478 bool hot, bool set); 3479 struct dentry *f2fs_get_parent(struct dentry *child); 3480 int f2fs_get_tmpfile(struct user_namespace *mnt_userns, struct inode *dir, 3481 struct inode **new_inode); 3482 3483 /* 3484 * dir.c 3485 */ 3486 unsigned char f2fs_get_de_type(struct f2fs_dir_entry *de); 3487 int f2fs_init_casefolded_name(const struct inode *dir, 3488 struct f2fs_filename *fname); 3489 int f2fs_setup_filename(struct inode *dir, const struct qstr *iname, 3490 int lookup, struct f2fs_filename *fname); 3491 int f2fs_prepare_lookup(struct inode *dir, struct dentry *dentry, 3492 struct f2fs_filename *fname); 3493 void f2fs_free_filename(struct f2fs_filename *fname); 3494 struct f2fs_dir_entry *f2fs_find_target_dentry(const struct f2fs_dentry_ptr *d, 3495 const struct f2fs_filename *fname, int *max_slots); 3496 int f2fs_fill_dentries(struct dir_context *ctx, struct f2fs_dentry_ptr *d, 3497 unsigned int start_pos, struct fscrypt_str *fstr); 3498 void f2fs_do_make_empty_dir(struct inode *inode, struct inode *parent, 3499 struct f2fs_dentry_ptr *d); 3500 struct page *f2fs_init_inode_metadata(struct inode *inode, struct inode *dir, 3501 const struct f2fs_filename *fname, struct page *dpage); 3502 void f2fs_update_parent_metadata(struct inode *dir, struct inode *inode, 3503 unsigned int current_depth); 3504 int f2fs_room_for_filename(const void *bitmap, int slots, int max_slots); 3505 void f2fs_drop_nlink(struct inode *dir, struct inode *inode); 3506 struct f2fs_dir_entry *__f2fs_find_entry(struct inode *dir, 3507 const struct f2fs_filename *fname, 3508 struct page **res_page); 3509 struct f2fs_dir_entry *f2fs_find_entry(struct inode *dir, 3510 const struct qstr *child, struct page **res_page); 3511 struct f2fs_dir_entry *f2fs_parent_dir(struct inode *dir, struct page **p); 3512 ino_t f2fs_inode_by_name(struct inode *dir, const struct qstr *qstr, 3513 struct page **page); 3514 void f2fs_set_link(struct inode *dir, struct f2fs_dir_entry *de, 3515 struct page *page, struct inode *inode); 3516 bool f2fs_has_enough_room(struct inode *dir, struct page *ipage, 3517 const struct f2fs_filename *fname); 3518 void f2fs_update_dentry(nid_t ino, umode_t mode, struct f2fs_dentry_ptr *d, 3519 const struct fscrypt_str *name, f2fs_hash_t name_hash, 3520 unsigned int bit_pos); 3521 int f2fs_add_regular_entry(struct inode *dir, const struct f2fs_filename *fname, 3522 struct inode *inode, nid_t ino, umode_t mode); 3523 int f2fs_add_dentry(struct inode *dir, const struct f2fs_filename *fname, 3524 struct inode *inode, nid_t ino, umode_t mode); 3525 int f2fs_do_add_link(struct inode *dir, const struct qstr *name, 3526 struct inode *inode, nid_t ino, umode_t mode); 3527 void f2fs_delete_entry(struct f2fs_dir_entry *dentry, struct page *page, 3528 struct inode *dir, struct inode *inode); 3529 int f2fs_do_tmpfile(struct inode *inode, struct inode *dir); 3530 bool f2fs_empty_dir(struct inode *dir); 3531 3532 static inline int f2fs_add_link(struct dentry *dentry, struct inode *inode) 3533 { 3534 if (fscrypt_is_nokey_name(dentry)) 3535 return -ENOKEY; 3536 return f2fs_do_add_link(d_inode(dentry->d_parent), &dentry->d_name, 3537 inode, inode->i_ino, inode->i_mode); 3538 } 3539 3540 /* 3541 * super.c 3542 */ 3543 int f2fs_inode_dirtied(struct inode *inode, bool sync); 3544 void f2fs_inode_synced(struct inode *inode); 3545 int f2fs_dquot_initialize(struct inode *inode); 3546 int f2fs_enable_quota_files(struct f2fs_sb_info *sbi, bool rdonly); 3547 int f2fs_quota_sync(struct super_block *sb, int type); 3548 loff_t max_file_blocks(struct inode *inode); 3549 void f2fs_quota_off_umount(struct super_block *sb); 3550 int f2fs_commit_super(struct f2fs_sb_info *sbi, bool recover); 3551 int f2fs_sync_fs(struct super_block *sb, int sync); 3552 int f2fs_sanity_check_ckpt(struct f2fs_sb_info *sbi); 3553 3554 /* 3555 * hash.c 3556 */ 3557 void f2fs_hash_filename(const struct inode *dir, struct f2fs_filename *fname); 3558 3559 /* 3560 * node.c 3561 */ 3562 struct node_info; 3563 3564 int f2fs_check_nid_range(struct f2fs_sb_info *sbi, nid_t nid); 3565 bool f2fs_available_free_memory(struct f2fs_sb_info *sbi, int type); 3566 bool f2fs_in_warm_node_list(struct f2fs_sb_info *sbi, struct page *page); 3567 void f2fs_init_fsync_node_info(struct f2fs_sb_info *sbi); 3568 void f2fs_del_fsync_node_entry(struct f2fs_sb_info *sbi, struct page *page); 3569 void f2fs_reset_fsync_node_info(struct f2fs_sb_info *sbi); 3570 int f2fs_need_dentry_mark(struct f2fs_sb_info *sbi, nid_t nid); 3571 bool f2fs_is_checkpointed_node(struct f2fs_sb_info *sbi, nid_t nid); 3572 bool f2fs_need_inode_block_update(struct f2fs_sb_info *sbi, nid_t ino); 3573 int f2fs_get_node_info(struct f2fs_sb_info *sbi, nid_t nid, 3574 struct node_info *ni, bool checkpoint_context); 3575 pgoff_t f2fs_get_next_page_offset(struct dnode_of_data *dn, pgoff_t pgofs); 3576 int f2fs_get_dnode_of_data(struct dnode_of_data *dn, pgoff_t index, int mode); 3577 int f2fs_truncate_inode_blocks(struct inode *inode, pgoff_t from); 3578 int f2fs_truncate_xattr_node(struct inode *inode); 3579 int f2fs_wait_on_node_pages_writeback(struct f2fs_sb_info *sbi, 3580 unsigned int seq_id); 3581 bool f2fs_nat_bitmap_enabled(struct f2fs_sb_info *sbi); 3582 int f2fs_remove_inode_page(struct inode *inode); 3583 struct page *f2fs_new_inode_page(struct inode *inode); 3584 struct page *f2fs_new_node_page(struct dnode_of_data *dn, unsigned int ofs); 3585 void f2fs_ra_node_page(struct f2fs_sb_info *sbi, nid_t nid); 3586 struct page *f2fs_get_node_page(struct f2fs_sb_info *sbi, pgoff_t nid); 3587 struct page *f2fs_get_node_page_ra(struct page *parent, int start); 3588 int f2fs_move_node_page(struct page *node_page, int gc_type); 3589 void f2fs_flush_inline_data(struct f2fs_sb_info *sbi); 3590 int f2fs_fsync_node_pages(struct f2fs_sb_info *sbi, struct inode *inode, 3591 struct writeback_control *wbc, bool atomic, 3592 unsigned int *seq_id); 3593 int f2fs_sync_node_pages(struct f2fs_sb_info *sbi, 3594 struct writeback_control *wbc, 3595 bool do_balance, enum iostat_type io_type); 3596 int f2fs_build_free_nids(struct f2fs_sb_info *sbi, bool sync, bool mount); 3597 bool f2fs_alloc_nid(struct f2fs_sb_info *sbi, nid_t *nid); 3598 void f2fs_alloc_nid_done(struct f2fs_sb_info *sbi, nid_t nid); 3599 void f2fs_alloc_nid_failed(struct f2fs_sb_info *sbi, nid_t nid); 3600 int f2fs_try_to_free_nids(struct f2fs_sb_info *sbi, int nr_shrink); 3601 int f2fs_recover_inline_xattr(struct inode *inode, struct page *page); 3602 int f2fs_recover_xattr_data(struct inode *inode, struct page *page); 3603 int f2fs_recover_inode_page(struct f2fs_sb_info *sbi, struct page *page); 3604 int f2fs_restore_node_summary(struct f2fs_sb_info *sbi, 3605 unsigned int segno, struct f2fs_summary_block *sum); 3606 void f2fs_enable_nat_bits(struct f2fs_sb_info *sbi); 3607 int f2fs_flush_nat_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3608 int f2fs_build_node_manager(struct f2fs_sb_info *sbi); 3609 void f2fs_destroy_node_manager(struct f2fs_sb_info *sbi); 3610 int __init f2fs_create_node_manager_caches(void); 3611 void f2fs_destroy_node_manager_caches(void); 3612 3613 /* 3614 * segment.c 3615 */ 3616 bool f2fs_need_SSR(struct f2fs_sb_info *sbi); 3617 int f2fs_commit_atomic_write(struct inode *inode); 3618 void f2fs_abort_atomic_write(struct inode *inode, bool clean); 3619 void f2fs_balance_fs(struct f2fs_sb_info *sbi, bool need); 3620 void f2fs_balance_fs_bg(struct f2fs_sb_info *sbi, bool from_bg); 3621 int f2fs_issue_flush(struct f2fs_sb_info *sbi, nid_t ino); 3622 int f2fs_create_flush_cmd_control(struct f2fs_sb_info *sbi); 3623 int f2fs_flush_device_cache(struct f2fs_sb_info *sbi); 3624 void f2fs_destroy_flush_cmd_control(struct f2fs_sb_info *sbi, bool free); 3625 void f2fs_invalidate_blocks(struct f2fs_sb_info *sbi, block_t addr); 3626 bool f2fs_is_checkpointed_data(struct f2fs_sb_info *sbi, block_t blkaddr); 3627 int f2fs_start_discard_thread(struct f2fs_sb_info *sbi); 3628 void f2fs_drop_discard_cmd(struct f2fs_sb_info *sbi); 3629 void f2fs_stop_discard_thread(struct f2fs_sb_info *sbi); 3630 bool f2fs_issue_discard_timeout(struct f2fs_sb_info *sbi); 3631 void f2fs_clear_prefree_segments(struct f2fs_sb_info *sbi, 3632 struct cp_control *cpc); 3633 void f2fs_dirty_to_prefree(struct f2fs_sb_info *sbi); 3634 block_t f2fs_get_unusable_blocks(struct f2fs_sb_info *sbi); 3635 int f2fs_disable_cp_again(struct f2fs_sb_info *sbi, block_t unusable); 3636 void f2fs_release_discard_addrs(struct f2fs_sb_info *sbi); 3637 int f2fs_npages_for_summary_flush(struct f2fs_sb_info *sbi, bool for_ra); 3638 bool f2fs_segment_has_free_slot(struct f2fs_sb_info *sbi, int segno); 3639 void f2fs_init_inmem_curseg(struct f2fs_sb_info *sbi); 3640 void f2fs_save_inmem_curseg(struct f2fs_sb_info *sbi); 3641 void f2fs_restore_inmem_curseg(struct f2fs_sb_info *sbi); 3642 void f2fs_get_new_segment(struct f2fs_sb_info *sbi, 3643 unsigned int *newseg, bool new_sec, int dir); 3644 void f2fs_allocate_segment_for_resize(struct f2fs_sb_info *sbi, int type, 3645 unsigned int start, unsigned int end); 3646 void f2fs_allocate_new_section(struct f2fs_sb_info *sbi, int type, bool force); 3647 void f2fs_allocate_new_segments(struct f2fs_sb_info *sbi); 3648 int f2fs_trim_fs(struct f2fs_sb_info *sbi, struct fstrim_range *range); 3649 bool f2fs_exist_trim_candidates(struct f2fs_sb_info *sbi, 3650 struct cp_control *cpc); 3651 struct page *f2fs_get_sum_page(struct f2fs_sb_info *sbi, unsigned int segno); 3652 void f2fs_update_meta_page(struct f2fs_sb_info *sbi, void *src, 3653 block_t blk_addr); 3654 void f2fs_do_write_meta_page(struct f2fs_sb_info *sbi, struct page *page, 3655 enum iostat_type io_type); 3656 void f2fs_do_write_node_page(unsigned int nid, struct f2fs_io_info *fio); 3657 void f2fs_outplace_write_data(struct dnode_of_data *dn, 3658 struct f2fs_io_info *fio); 3659 int f2fs_inplace_write_data(struct f2fs_io_info *fio); 3660 void f2fs_do_replace_block(struct f2fs_sb_info *sbi, struct f2fs_summary *sum, 3661 block_t old_blkaddr, block_t new_blkaddr, 3662 bool recover_curseg, bool recover_newaddr, 3663 bool from_gc); 3664 void f2fs_replace_block(struct f2fs_sb_info *sbi, struct dnode_of_data *dn, 3665 block_t old_addr, block_t new_addr, 3666 unsigned char version, bool recover_curseg, 3667 bool recover_newaddr); 3668 void f2fs_allocate_data_block(struct f2fs_sb_info *sbi, struct page *page, 3669 block_t old_blkaddr, block_t *new_blkaddr, 3670 struct f2fs_summary *sum, int type, 3671 struct f2fs_io_info *fio); 3672 void f2fs_update_device_state(struct f2fs_sb_info *sbi, nid_t ino, 3673 block_t blkaddr, unsigned int blkcnt); 3674 void f2fs_wait_on_page_writeback(struct page *page, 3675 enum page_type type, bool ordered, bool locked); 3676 void f2fs_wait_on_block_writeback(struct inode *inode, block_t blkaddr); 3677 void f2fs_wait_on_block_writeback_range(struct inode *inode, block_t blkaddr, 3678 block_t len); 3679 void f2fs_write_data_summaries(struct f2fs_sb_info *sbi, block_t start_blk); 3680 void f2fs_write_node_summaries(struct f2fs_sb_info *sbi, block_t start_blk); 3681 int f2fs_lookup_journal_in_cursum(struct f2fs_journal *journal, int type, 3682 unsigned int val, int alloc); 3683 void f2fs_flush_sit_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3684 int f2fs_fix_curseg_write_pointer(struct f2fs_sb_info *sbi); 3685 int f2fs_check_write_pointer(struct f2fs_sb_info *sbi); 3686 int f2fs_build_segment_manager(struct f2fs_sb_info *sbi); 3687 void f2fs_destroy_segment_manager(struct f2fs_sb_info *sbi); 3688 int __init f2fs_create_segment_manager_caches(void); 3689 void f2fs_destroy_segment_manager_caches(void); 3690 int f2fs_rw_hint_to_seg_type(enum rw_hint hint); 3691 unsigned int f2fs_usable_segs_in_sec(struct f2fs_sb_info *sbi, 3692 unsigned int segno); 3693 unsigned int f2fs_usable_blks_in_seg(struct f2fs_sb_info *sbi, 3694 unsigned int segno); 3695 3696 #define DEF_FRAGMENT_SIZE 4 3697 #define MIN_FRAGMENT_SIZE 1 3698 #define MAX_FRAGMENT_SIZE 512 3699 3700 static inline bool f2fs_need_rand_seg(struct f2fs_sb_info *sbi) 3701 { 3702 return F2FS_OPTION(sbi).fs_mode == FS_MODE_FRAGMENT_SEG || 3703 F2FS_OPTION(sbi).fs_mode == FS_MODE_FRAGMENT_BLK; 3704 } 3705 3706 /* 3707 * checkpoint.c 3708 */ 3709 void f2fs_stop_checkpoint(struct f2fs_sb_info *sbi, bool end_io); 3710 struct page *f2fs_grab_meta_page(struct f2fs_sb_info *sbi, pgoff_t index); 3711 struct page *f2fs_get_meta_page(struct f2fs_sb_info *sbi, pgoff_t index); 3712 struct page *f2fs_get_meta_page_retry(struct f2fs_sb_info *sbi, pgoff_t index); 3713 struct page *f2fs_get_tmp_page(struct f2fs_sb_info *sbi, pgoff_t index); 3714 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi, 3715 block_t blkaddr, int type); 3716 int f2fs_ra_meta_pages(struct f2fs_sb_info *sbi, block_t start, int nrpages, 3717 int type, bool sync); 3718 void f2fs_ra_meta_pages_cond(struct f2fs_sb_info *sbi, pgoff_t index, 3719 unsigned int ra_blocks); 3720 long f2fs_sync_meta_pages(struct f2fs_sb_info *sbi, enum page_type type, 3721 long nr_to_write, enum iostat_type io_type); 3722 void f2fs_add_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type); 3723 void f2fs_remove_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type); 3724 void f2fs_release_ino_entry(struct f2fs_sb_info *sbi, bool all); 3725 bool f2fs_exist_written_data(struct f2fs_sb_info *sbi, nid_t ino, int mode); 3726 void f2fs_set_dirty_device(struct f2fs_sb_info *sbi, nid_t ino, 3727 unsigned int devidx, int type); 3728 bool f2fs_is_dirty_device(struct f2fs_sb_info *sbi, nid_t ino, 3729 unsigned int devidx, int type); 3730 int f2fs_sync_inode_meta(struct f2fs_sb_info *sbi); 3731 int f2fs_acquire_orphan_inode(struct f2fs_sb_info *sbi); 3732 void f2fs_release_orphan_inode(struct f2fs_sb_info *sbi); 3733 void f2fs_add_orphan_inode(struct inode *inode); 3734 void f2fs_remove_orphan_inode(struct f2fs_sb_info *sbi, nid_t ino); 3735 int f2fs_recover_orphan_inodes(struct f2fs_sb_info *sbi); 3736 int f2fs_get_valid_checkpoint(struct f2fs_sb_info *sbi); 3737 void f2fs_update_dirty_folio(struct inode *inode, struct folio *folio); 3738 void f2fs_remove_dirty_inode(struct inode *inode); 3739 int f2fs_sync_dirty_inodes(struct f2fs_sb_info *sbi, enum inode_type type); 3740 void f2fs_wait_on_all_pages(struct f2fs_sb_info *sbi, int type); 3741 u64 f2fs_get_sectors_written(struct f2fs_sb_info *sbi); 3742 int f2fs_write_checkpoint(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3743 void f2fs_init_ino_entry_info(struct f2fs_sb_info *sbi); 3744 int __init f2fs_create_checkpoint_caches(void); 3745 void f2fs_destroy_checkpoint_caches(void); 3746 int f2fs_issue_checkpoint(struct f2fs_sb_info *sbi); 3747 int f2fs_start_ckpt_thread(struct f2fs_sb_info *sbi); 3748 void f2fs_stop_ckpt_thread(struct f2fs_sb_info *sbi); 3749 void f2fs_init_ckpt_req_control(struct f2fs_sb_info *sbi); 3750 3751 /* 3752 * data.c 3753 */ 3754 int __init f2fs_init_bioset(void); 3755 void f2fs_destroy_bioset(void); 3756 int f2fs_init_bio_entry_cache(void); 3757 void f2fs_destroy_bio_entry_cache(void); 3758 void f2fs_submit_bio(struct f2fs_sb_info *sbi, 3759 struct bio *bio, enum page_type type); 3760 int f2fs_init_write_merge_io(struct f2fs_sb_info *sbi); 3761 void f2fs_submit_merged_write(struct f2fs_sb_info *sbi, enum page_type type); 3762 void f2fs_submit_merged_write_cond(struct f2fs_sb_info *sbi, 3763 struct inode *inode, struct page *page, 3764 nid_t ino, enum page_type type); 3765 void f2fs_submit_merged_ipu_write(struct f2fs_sb_info *sbi, 3766 struct bio **bio, struct page *page); 3767 void f2fs_flush_merged_writes(struct f2fs_sb_info *sbi); 3768 int f2fs_submit_page_bio(struct f2fs_io_info *fio); 3769 int f2fs_merge_page_bio(struct f2fs_io_info *fio); 3770 void f2fs_submit_page_write(struct f2fs_io_info *fio); 3771 struct block_device *f2fs_target_device(struct f2fs_sb_info *sbi, 3772 block_t blk_addr, sector_t *sector); 3773 int f2fs_target_device_index(struct f2fs_sb_info *sbi, block_t blkaddr); 3774 void f2fs_set_data_blkaddr(struct dnode_of_data *dn); 3775 void f2fs_update_data_blkaddr(struct dnode_of_data *dn, block_t blkaddr); 3776 int f2fs_reserve_new_blocks(struct dnode_of_data *dn, blkcnt_t count); 3777 int f2fs_reserve_new_block(struct dnode_of_data *dn); 3778 int f2fs_get_block(struct dnode_of_data *dn, pgoff_t index); 3779 int f2fs_reserve_block(struct dnode_of_data *dn, pgoff_t index); 3780 struct page *f2fs_get_read_data_page(struct inode *inode, pgoff_t index, 3781 blk_opf_t op_flags, bool for_write); 3782 struct page *f2fs_find_data_page(struct inode *inode, pgoff_t index); 3783 struct page *f2fs_get_lock_data_page(struct inode *inode, pgoff_t index, 3784 bool for_write); 3785 struct page *f2fs_get_new_data_page(struct inode *inode, 3786 struct page *ipage, pgoff_t index, bool new_i_size); 3787 int f2fs_do_write_data_page(struct f2fs_io_info *fio); 3788 void f2fs_do_map_lock(struct f2fs_sb_info *sbi, int flag, bool lock); 3789 int f2fs_map_blocks(struct inode *inode, struct f2fs_map_blocks *map, 3790 int create, int flag); 3791 int f2fs_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, 3792 u64 start, u64 len); 3793 int f2fs_encrypt_one_page(struct f2fs_io_info *fio); 3794 bool f2fs_should_update_inplace(struct inode *inode, struct f2fs_io_info *fio); 3795 bool f2fs_should_update_outplace(struct inode *inode, struct f2fs_io_info *fio); 3796 int f2fs_write_single_data_page(struct page *page, int *submitted, 3797 struct bio **bio, sector_t *last_block, 3798 struct writeback_control *wbc, 3799 enum iostat_type io_type, 3800 int compr_blocks, bool allow_balance); 3801 void f2fs_write_failed(struct inode *inode, loff_t to); 3802 void f2fs_invalidate_folio(struct folio *folio, size_t offset, size_t length); 3803 bool f2fs_release_folio(struct folio *folio, gfp_t wait); 3804 bool f2fs_overwrite_io(struct inode *inode, loff_t pos, size_t len); 3805 void f2fs_clear_page_cache_dirty_tag(struct page *page); 3806 int f2fs_init_post_read_processing(void); 3807 void f2fs_destroy_post_read_processing(void); 3808 int f2fs_init_post_read_wq(struct f2fs_sb_info *sbi); 3809 void f2fs_destroy_post_read_wq(struct f2fs_sb_info *sbi); 3810 extern const struct iomap_ops f2fs_iomap_ops; 3811 3812 /* 3813 * gc.c 3814 */ 3815 int f2fs_start_gc_thread(struct f2fs_sb_info *sbi); 3816 void f2fs_stop_gc_thread(struct f2fs_sb_info *sbi); 3817 block_t f2fs_start_bidx_of_node(unsigned int node_ofs, struct inode *inode); 3818 int f2fs_gc(struct f2fs_sb_info *sbi, struct f2fs_gc_control *gc_control); 3819 void f2fs_build_gc_manager(struct f2fs_sb_info *sbi); 3820 int f2fs_resize_fs(struct f2fs_sb_info *sbi, __u64 block_count); 3821 int __init f2fs_create_garbage_collection_cache(void); 3822 void f2fs_destroy_garbage_collection_cache(void); 3823 3824 /* 3825 * recovery.c 3826 */ 3827 int f2fs_recover_fsync_data(struct f2fs_sb_info *sbi, bool check_only); 3828 bool f2fs_space_for_roll_forward(struct f2fs_sb_info *sbi); 3829 int __init f2fs_create_recovery_cache(void); 3830 void f2fs_destroy_recovery_cache(void); 3831 3832 /* 3833 * debug.c 3834 */ 3835 #ifdef CONFIG_F2FS_STAT_FS 3836 struct f2fs_stat_info { 3837 struct list_head stat_list; 3838 struct f2fs_sb_info *sbi; 3839 int all_area_segs, sit_area_segs, nat_area_segs, ssa_area_segs; 3840 int main_area_segs, main_area_sections, main_area_zones; 3841 unsigned long long hit_largest, hit_cached, hit_rbtree; 3842 unsigned long long hit_total, total_ext; 3843 int ext_tree, zombie_tree, ext_node; 3844 int ndirty_node, ndirty_dent, ndirty_meta, ndirty_imeta; 3845 int ndirty_data, ndirty_qdata; 3846 unsigned int ndirty_dirs, ndirty_files, nquota_files, ndirty_all; 3847 int nats, dirty_nats, sits, dirty_sits; 3848 int free_nids, avail_nids, alloc_nids; 3849 int total_count, utilization; 3850 int bg_gc, nr_wb_cp_data, nr_wb_data; 3851 int nr_rd_data, nr_rd_node, nr_rd_meta; 3852 int nr_dio_read, nr_dio_write; 3853 unsigned int io_skip_bggc, other_skip_bggc; 3854 int nr_flushing, nr_flushed, flush_list_empty; 3855 int nr_discarding, nr_discarded; 3856 int nr_discard_cmd; 3857 unsigned int undiscard_blks; 3858 int nr_issued_ckpt, nr_total_ckpt, nr_queued_ckpt; 3859 unsigned int cur_ckpt_time, peak_ckpt_time; 3860 int inline_xattr, inline_inode, inline_dir, append, update, orphans; 3861 int compr_inode; 3862 unsigned long long compr_blocks; 3863 int aw_cnt, max_aw_cnt; 3864 unsigned int valid_count, valid_node_count, valid_inode_count, discard_blks; 3865 unsigned int bimodal, avg_vblocks; 3866 int util_free, util_valid, util_invalid; 3867 int rsvd_segs, overp_segs; 3868 int dirty_count, node_pages, meta_pages, compress_pages; 3869 int compress_page_hit; 3870 int prefree_count, call_count, cp_count, bg_cp_count; 3871 int tot_segs, node_segs, data_segs, free_segs, free_secs; 3872 int bg_node_segs, bg_data_segs; 3873 int tot_blks, data_blks, node_blks; 3874 int bg_data_blks, bg_node_blks; 3875 int curseg[NR_CURSEG_TYPE]; 3876 int cursec[NR_CURSEG_TYPE]; 3877 int curzone[NR_CURSEG_TYPE]; 3878 unsigned int dirty_seg[NR_CURSEG_TYPE]; 3879 unsigned int full_seg[NR_CURSEG_TYPE]; 3880 unsigned int valid_blks[NR_CURSEG_TYPE]; 3881 3882 unsigned int meta_count[META_MAX]; 3883 unsigned int segment_count[2]; 3884 unsigned int block_count[2]; 3885 unsigned int inplace_count; 3886 unsigned long long base_mem, cache_mem, page_mem; 3887 }; 3888 3889 static inline struct f2fs_stat_info *F2FS_STAT(struct f2fs_sb_info *sbi) 3890 { 3891 return (struct f2fs_stat_info *)sbi->stat_info; 3892 } 3893 3894 #define stat_inc_cp_count(si) ((si)->cp_count++) 3895 #define stat_inc_bg_cp_count(si) ((si)->bg_cp_count++) 3896 #define stat_inc_call_count(si) ((si)->call_count++) 3897 #define stat_inc_bggc_count(si) ((si)->bg_gc++) 3898 #define stat_io_skip_bggc_count(sbi) ((sbi)->io_skip_bggc++) 3899 #define stat_other_skip_bggc_count(sbi) ((sbi)->other_skip_bggc++) 3900 #define stat_inc_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]++) 3901 #define stat_dec_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]--) 3902 #define stat_inc_total_hit(sbi) (atomic64_inc(&(sbi)->total_hit_ext)) 3903 #define stat_inc_rbtree_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_rbtree)) 3904 #define stat_inc_largest_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_largest)) 3905 #define stat_inc_cached_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_cached)) 3906 #define stat_inc_inline_xattr(inode) \ 3907 do { \ 3908 if (f2fs_has_inline_xattr(inode)) \ 3909 (atomic_inc(&F2FS_I_SB(inode)->inline_xattr)); \ 3910 } while (0) 3911 #define stat_dec_inline_xattr(inode) \ 3912 do { \ 3913 if (f2fs_has_inline_xattr(inode)) \ 3914 (atomic_dec(&F2FS_I_SB(inode)->inline_xattr)); \ 3915 } while (0) 3916 #define stat_inc_inline_inode(inode) \ 3917 do { \ 3918 if (f2fs_has_inline_data(inode)) \ 3919 (atomic_inc(&F2FS_I_SB(inode)->inline_inode)); \ 3920 } while (0) 3921 #define stat_dec_inline_inode(inode) \ 3922 do { \ 3923 if (f2fs_has_inline_data(inode)) \ 3924 (atomic_dec(&F2FS_I_SB(inode)->inline_inode)); \ 3925 } while (0) 3926 #define stat_inc_inline_dir(inode) \ 3927 do { \ 3928 if (f2fs_has_inline_dentry(inode)) \ 3929 (atomic_inc(&F2FS_I_SB(inode)->inline_dir)); \ 3930 } while (0) 3931 #define stat_dec_inline_dir(inode) \ 3932 do { \ 3933 if (f2fs_has_inline_dentry(inode)) \ 3934 (atomic_dec(&F2FS_I_SB(inode)->inline_dir)); \ 3935 } while (0) 3936 #define stat_inc_compr_inode(inode) \ 3937 do { \ 3938 if (f2fs_compressed_file(inode)) \ 3939 (atomic_inc(&F2FS_I_SB(inode)->compr_inode)); \ 3940 } while (0) 3941 #define stat_dec_compr_inode(inode) \ 3942 do { \ 3943 if (f2fs_compressed_file(inode)) \ 3944 (atomic_dec(&F2FS_I_SB(inode)->compr_inode)); \ 3945 } while (0) 3946 #define stat_add_compr_blocks(inode, blocks) \ 3947 (atomic64_add(blocks, &F2FS_I_SB(inode)->compr_blocks)) 3948 #define stat_sub_compr_blocks(inode, blocks) \ 3949 (atomic64_sub(blocks, &F2FS_I_SB(inode)->compr_blocks)) 3950 #define stat_inc_meta_count(sbi, blkaddr) \ 3951 do { \ 3952 if (blkaddr < SIT_I(sbi)->sit_base_addr) \ 3953 atomic_inc(&(sbi)->meta_count[META_CP]); \ 3954 else if (blkaddr < NM_I(sbi)->nat_blkaddr) \ 3955 atomic_inc(&(sbi)->meta_count[META_SIT]); \ 3956 else if (blkaddr < SM_I(sbi)->ssa_blkaddr) \ 3957 atomic_inc(&(sbi)->meta_count[META_NAT]); \ 3958 else if (blkaddr < SM_I(sbi)->main_blkaddr) \ 3959 atomic_inc(&(sbi)->meta_count[META_SSA]); \ 3960 } while (0) 3961 #define stat_inc_seg_type(sbi, curseg) \ 3962 ((sbi)->segment_count[(curseg)->alloc_type]++) 3963 #define stat_inc_block_count(sbi, curseg) \ 3964 ((sbi)->block_count[(curseg)->alloc_type]++) 3965 #define stat_inc_inplace_blocks(sbi) \ 3966 (atomic_inc(&(sbi)->inplace_count)) 3967 #define stat_update_max_atomic_write(inode) \ 3968 do { \ 3969 int cur = F2FS_I_SB(inode)->atomic_files; \ 3970 int max = atomic_read(&F2FS_I_SB(inode)->max_aw_cnt); \ 3971 if (cur > max) \ 3972 atomic_set(&F2FS_I_SB(inode)->max_aw_cnt, cur); \ 3973 } while (0) 3974 #define stat_inc_seg_count(sbi, type, gc_type) \ 3975 do { \ 3976 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 3977 si->tot_segs++; \ 3978 if ((type) == SUM_TYPE_DATA) { \ 3979 si->data_segs++; \ 3980 si->bg_data_segs += (gc_type == BG_GC) ? 1 : 0; \ 3981 } else { \ 3982 si->node_segs++; \ 3983 si->bg_node_segs += (gc_type == BG_GC) ? 1 : 0; \ 3984 } \ 3985 } while (0) 3986 3987 #define stat_inc_tot_blk_count(si, blks) \ 3988 ((si)->tot_blks += (blks)) 3989 3990 #define stat_inc_data_blk_count(sbi, blks, gc_type) \ 3991 do { \ 3992 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 3993 stat_inc_tot_blk_count(si, blks); \ 3994 si->data_blks += (blks); \ 3995 si->bg_data_blks += ((gc_type) == BG_GC) ? (blks) : 0; \ 3996 } while (0) 3997 3998 #define stat_inc_node_blk_count(sbi, blks, gc_type) \ 3999 do { \ 4000 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 4001 stat_inc_tot_blk_count(si, blks); \ 4002 si->node_blks += (blks); \ 4003 si->bg_node_blks += ((gc_type) == BG_GC) ? (blks) : 0; \ 4004 } while (0) 4005 4006 int f2fs_build_stats(struct f2fs_sb_info *sbi); 4007 void f2fs_destroy_stats(struct f2fs_sb_info *sbi); 4008 void __init f2fs_create_root_stats(void); 4009 void f2fs_destroy_root_stats(void); 4010 void f2fs_update_sit_info(struct f2fs_sb_info *sbi); 4011 #else 4012 #define stat_inc_cp_count(si) do { } while (0) 4013 #define stat_inc_bg_cp_count(si) do { } while (0) 4014 #define stat_inc_call_count(si) do { } while (0) 4015 #define stat_inc_bggc_count(si) do { } while (0) 4016 #define stat_io_skip_bggc_count(sbi) do { } while (0) 4017 #define stat_other_skip_bggc_count(sbi) do { } while (0) 4018 #define stat_inc_dirty_inode(sbi, type) do { } while (0) 4019 #define stat_dec_dirty_inode(sbi, type) do { } while (0) 4020 #define stat_inc_total_hit(sbi) do { } while (0) 4021 #define stat_inc_rbtree_node_hit(sbi) do { } while (0) 4022 #define stat_inc_largest_node_hit(sbi) do { } while (0) 4023 #define stat_inc_cached_node_hit(sbi) do { } while (0) 4024 #define stat_inc_inline_xattr(inode) do { } while (0) 4025 #define stat_dec_inline_xattr(inode) do { } while (0) 4026 #define stat_inc_inline_inode(inode) do { } while (0) 4027 #define stat_dec_inline_inode(inode) do { } while (0) 4028 #define stat_inc_inline_dir(inode) do { } while (0) 4029 #define stat_dec_inline_dir(inode) do { } while (0) 4030 #define stat_inc_compr_inode(inode) do { } while (0) 4031 #define stat_dec_compr_inode(inode) do { } while (0) 4032 #define stat_add_compr_blocks(inode, blocks) do { } while (0) 4033 #define stat_sub_compr_blocks(inode, blocks) do { } while (0) 4034 #define stat_update_max_atomic_write(inode) do { } while (0) 4035 #define stat_inc_meta_count(sbi, blkaddr) do { } while (0) 4036 #define stat_inc_seg_type(sbi, curseg) do { } while (0) 4037 #define stat_inc_block_count(sbi, curseg) do { } while (0) 4038 #define stat_inc_inplace_blocks(sbi) do { } while (0) 4039 #define stat_inc_seg_count(sbi, type, gc_type) do { } while (0) 4040 #define stat_inc_tot_blk_count(si, blks) do { } while (0) 4041 #define stat_inc_data_blk_count(sbi, blks, gc_type) do { } while (0) 4042 #define stat_inc_node_blk_count(sbi, blks, gc_type) do { } while (0) 4043 4044 static inline int f2fs_build_stats(struct f2fs_sb_info *sbi) { return 0; } 4045 static inline void f2fs_destroy_stats(struct f2fs_sb_info *sbi) { } 4046 static inline void __init f2fs_create_root_stats(void) { } 4047 static inline void f2fs_destroy_root_stats(void) { } 4048 static inline void f2fs_update_sit_info(struct f2fs_sb_info *sbi) {} 4049 #endif 4050 4051 extern const struct file_operations f2fs_dir_operations; 4052 extern const struct file_operations f2fs_file_operations; 4053 extern const struct inode_operations f2fs_file_inode_operations; 4054 extern const struct address_space_operations f2fs_dblock_aops; 4055 extern const struct address_space_operations f2fs_node_aops; 4056 extern const struct address_space_operations f2fs_meta_aops; 4057 extern const struct inode_operations f2fs_dir_inode_operations; 4058 extern const struct inode_operations f2fs_symlink_inode_operations; 4059 extern const struct inode_operations f2fs_encrypted_symlink_inode_operations; 4060 extern const struct inode_operations f2fs_special_inode_operations; 4061 extern struct kmem_cache *f2fs_inode_entry_slab; 4062 4063 /* 4064 * inline.c 4065 */ 4066 bool f2fs_may_inline_data(struct inode *inode); 4067 bool f2fs_sanity_check_inline_data(struct inode *inode); 4068 bool f2fs_may_inline_dentry(struct inode *inode); 4069 void f2fs_do_read_inline_data(struct page *page, struct page *ipage); 4070 void f2fs_truncate_inline_inode(struct inode *inode, 4071 struct page *ipage, u64 from); 4072 int f2fs_read_inline_data(struct inode *inode, struct page *page); 4073 int f2fs_convert_inline_page(struct dnode_of_data *dn, struct page *page); 4074 int f2fs_convert_inline_inode(struct inode *inode); 4075 int f2fs_try_convert_inline_dir(struct inode *dir, struct dentry *dentry); 4076 int f2fs_write_inline_data(struct inode *inode, struct page *page); 4077 int f2fs_recover_inline_data(struct inode *inode, struct page *npage); 4078 struct f2fs_dir_entry *f2fs_find_in_inline_dir(struct inode *dir, 4079 const struct f2fs_filename *fname, 4080 struct page **res_page); 4081 int f2fs_make_empty_inline_dir(struct inode *inode, struct inode *parent, 4082 struct page *ipage); 4083 int f2fs_add_inline_entry(struct inode *dir, const struct f2fs_filename *fname, 4084 struct inode *inode, nid_t ino, umode_t mode); 4085 void f2fs_delete_inline_entry(struct f2fs_dir_entry *dentry, 4086 struct page *page, struct inode *dir, 4087 struct inode *inode); 4088 bool f2fs_empty_inline_dir(struct inode *dir); 4089 int f2fs_read_inline_dir(struct file *file, struct dir_context *ctx, 4090 struct fscrypt_str *fstr); 4091 int f2fs_inline_data_fiemap(struct inode *inode, 4092 struct fiemap_extent_info *fieinfo, 4093 __u64 start, __u64 len); 4094 4095 /* 4096 * shrinker.c 4097 */ 4098 unsigned long f2fs_shrink_count(struct shrinker *shrink, 4099 struct shrink_control *sc); 4100 unsigned long f2fs_shrink_scan(struct shrinker *shrink, 4101 struct shrink_control *sc); 4102 void f2fs_join_shrinker(struct f2fs_sb_info *sbi); 4103 void f2fs_leave_shrinker(struct f2fs_sb_info *sbi); 4104 4105 /* 4106 * extent_cache.c 4107 */ 4108 struct rb_entry *f2fs_lookup_rb_tree(struct rb_root_cached *root, 4109 struct rb_entry *cached_re, unsigned int ofs); 4110 struct rb_node **f2fs_lookup_rb_tree_ext(struct f2fs_sb_info *sbi, 4111 struct rb_root_cached *root, 4112 struct rb_node **parent, 4113 unsigned long long key, bool *left_most); 4114 struct rb_node **f2fs_lookup_rb_tree_for_insert(struct f2fs_sb_info *sbi, 4115 struct rb_root_cached *root, 4116 struct rb_node **parent, 4117 unsigned int ofs, bool *leftmost); 4118 struct rb_entry *f2fs_lookup_rb_tree_ret(struct rb_root_cached *root, 4119 struct rb_entry *cached_re, unsigned int ofs, 4120 struct rb_entry **prev_entry, struct rb_entry **next_entry, 4121 struct rb_node ***insert_p, struct rb_node **insert_parent, 4122 bool force, bool *leftmost); 4123 bool f2fs_check_rb_tree_consistence(struct f2fs_sb_info *sbi, 4124 struct rb_root_cached *root, bool check_key); 4125 unsigned int f2fs_shrink_extent_tree(struct f2fs_sb_info *sbi, int nr_shrink); 4126 void f2fs_init_extent_tree(struct inode *inode, struct page *ipage); 4127 void f2fs_drop_extent_tree(struct inode *inode); 4128 unsigned int f2fs_destroy_extent_node(struct inode *inode); 4129 void f2fs_destroy_extent_tree(struct inode *inode); 4130 bool f2fs_lookup_extent_cache(struct inode *inode, pgoff_t pgofs, 4131 struct extent_info *ei); 4132 void f2fs_update_extent_cache(struct dnode_of_data *dn); 4133 void f2fs_update_extent_cache_range(struct dnode_of_data *dn, 4134 pgoff_t fofs, block_t blkaddr, unsigned int len); 4135 void f2fs_init_extent_cache_info(struct f2fs_sb_info *sbi); 4136 int __init f2fs_create_extent_cache(void); 4137 void f2fs_destroy_extent_cache(void); 4138 4139 /* 4140 * sysfs.c 4141 */ 4142 #define MIN_RA_MUL 2 4143 #define MAX_RA_MUL 256 4144 4145 int __init f2fs_init_sysfs(void); 4146 void f2fs_exit_sysfs(void); 4147 int f2fs_register_sysfs(struct f2fs_sb_info *sbi); 4148 void f2fs_unregister_sysfs(struct f2fs_sb_info *sbi); 4149 4150 /* verity.c */ 4151 extern const struct fsverity_operations f2fs_verityops; 4152 4153 /* 4154 * crypto support 4155 */ 4156 static inline bool f2fs_encrypted_file(struct inode *inode) 4157 { 4158 return IS_ENCRYPTED(inode) && S_ISREG(inode->i_mode); 4159 } 4160 4161 static inline void f2fs_set_encrypted_inode(struct inode *inode) 4162 { 4163 #ifdef CONFIG_FS_ENCRYPTION 4164 file_set_encrypt(inode); 4165 f2fs_set_inode_flags(inode); 4166 #endif 4167 } 4168 4169 /* 4170 * Returns true if the reads of the inode's data need to undergo some 4171 * postprocessing step, like decryption or authenticity verification. 4172 */ 4173 static inline bool f2fs_post_read_required(struct inode *inode) 4174 { 4175 return f2fs_encrypted_file(inode) || fsverity_active(inode) || 4176 f2fs_compressed_file(inode); 4177 } 4178 4179 /* 4180 * compress.c 4181 */ 4182 #ifdef CONFIG_F2FS_FS_COMPRESSION 4183 bool f2fs_is_compressed_page(struct page *page); 4184 struct page *f2fs_compress_control_page(struct page *page); 4185 int f2fs_prepare_compress_overwrite(struct inode *inode, 4186 struct page **pagep, pgoff_t index, void **fsdata); 4187 bool f2fs_compress_write_end(struct inode *inode, void *fsdata, 4188 pgoff_t index, unsigned copied); 4189 int f2fs_truncate_partial_cluster(struct inode *inode, u64 from, bool lock); 4190 void f2fs_compress_write_end_io(struct bio *bio, struct page *page); 4191 bool f2fs_is_compress_backend_ready(struct inode *inode); 4192 int f2fs_init_compress_mempool(void); 4193 void f2fs_destroy_compress_mempool(void); 4194 void f2fs_decompress_cluster(struct decompress_io_ctx *dic, bool in_task); 4195 void f2fs_end_read_compressed_page(struct page *page, bool failed, 4196 block_t blkaddr, bool in_task); 4197 bool f2fs_cluster_is_empty(struct compress_ctx *cc); 4198 bool f2fs_cluster_can_merge_page(struct compress_ctx *cc, pgoff_t index); 4199 bool f2fs_all_cluster_page_ready(struct compress_ctx *cc, struct page **pages, 4200 int index, int nr_pages, bool uptodate); 4201 bool f2fs_sanity_check_cluster(struct dnode_of_data *dn); 4202 void f2fs_compress_ctx_add_page(struct compress_ctx *cc, struct page *page); 4203 int f2fs_write_multi_pages(struct compress_ctx *cc, 4204 int *submitted, 4205 struct writeback_control *wbc, 4206 enum iostat_type io_type); 4207 int f2fs_is_compressed_cluster(struct inode *inode, pgoff_t index); 4208 void f2fs_update_extent_tree_range_compressed(struct inode *inode, 4209 pgoff_t fofs, block_t blkaddr, unsigned int llen, 4210 unsigned int c_len); 4211 int f2fs_read_multi_pages(struct compress_ctx *cc, struct bio **bio_ret, 4212 unsigned nr_pages, sector_t *last_block_in_bio, 4213 bool is_readahead, bool for_write); 4214 struct decompress_io_ctx *f2fs_alloc_dic(struct compress_ctx *cc); 4215 void f2fs_decompress_end_io(struct decompress_io_ctx *dic, bool failed, 4216 bool in_task); 4217 void f2fs_put_page_dic(struct page *page, bool in_task); 4218 unsigned int f2fs_cluster_blocks_are_contiguous(struct dnode_of_data *dn); 4219 int f2fs_init_compress_ctx(struct compress_ctx *cc); 4220 void f2fs_destroy_compress_ctx(struct compress_ctx *cc, bool reuse); 4221 void f2fs_init_compress_info(struct f2fs_sb_info *sbi); 4222 int f2fs_init_compress_inode(struct f2fs_sb_info *sbi); 4223 void f2fs_destroy_compress_inode(struct f2fs_sb_info *sbi); 4224 int f2fs_init_page_array_cache(struct f2fs_sb_info *sbi); 4225 void f2fs_destroy_page_array_cache(struct f2fs_sb_info *sbi); 4226 int __init f2fs_init_compress_cache(void); 4227 void f2fs_destroy_compress_cache(void); 4228 struct address_space *COMPRESS_MAPPING(struct f2fs_sb_info *sbi); 4229 void f2fs_invalidate_compress_page(struct f2fs_sb_info *sbi, block_t blkaddr); 4230 void f2fs_cache_compressed_page(struct f2fs_sb_info *sbi, struct page *page, 4231 nid_t ino, block_t blkaddr); 4232 bool f2fs_load_compressed_page(struct f2fs_sb_info *sbi, struct page *page, 4233 block_t blkaddr); 4234 void f2fs_invalidate_compress_pages(struct f2fs_sb_info *sbi, nid_t ino); 4235 #define inc_compr_inode_stat(inode) \ 4236 do { \ 4237 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); \ 4238 sbi->compr_new_inode++; \ 4239 } while (0) 4240 #define add_compr_block_stat(inode, blocks) \ 4241 do { \ 4242 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); \ 4243 int diff = F2FS_I(inode)->i_cluster_size - blocks; \ 4244 sbi->compr_written_block += blocks; \ 4245 sbi->compr_saved_block += diff; \ 4246 } while (0) 4247 #else 4248 static inline bool f2fs_is_compressed_page(struct page *page) { return false; } 4249 static inline bool f2fs_is_compress_backend_ready(struct inode *inode) 4250 { 4251 if (!f2fs_compressed_file(inode)) 4252 return true; 4253 /* not support compression */ 4254 return false; 4255 } 4256 static inline struct page *f2fs_compress_control_page(struct page *page) 4257 { 4258 WARN_ON_ONCE(1); 4259 return ERR_PTR(-EINVAL); 4260 } 4261 static inline int f2fs_init_compress_mempool(void) { return 0; } 4262 static inline void f2fs_destroy_compress_mempool(void) { } 4263 static inline void f2fs_decompress_cluster(struct decompress_io_ctx *dic, 4264 bool in_task) { } 4265 static inline void f2fs_end_read_compressed_page(struct page *page, 4266 bool failed, block_t blkaddr, bool in_task) 4267 { 4268 WARN_ON_ONCE(1); 4269 } 4270 static inline void f2fs_put_page_dic(struct page *page, bool in_task) 4271 { 4272 WARN_ON_ONCE(1); 4273 } 4274 static inline unsigned int f2fs_cluster_blocks_are_contiguous(struct dnode_of_data *dn) { return 0; } 4275 static inline bool f2fs_sanity_check_cluster(struct dnode_of_data *dn) { return false; } 4276 static inline int f2fs_init_compress_inode(struct f2fs_sb_info *sbi) { return 0; } 4277 static inline void f2fs_destroy_compress_inode(struct f2fs_sb_info *sbi) { } 4278 static inline int f2fs_init_page_array_cache(struct f2fs_sb_info *sbi) { return 0; } 4279 static inline void f2fs_destroy_page_array_cache(struct f2fs_sb_info *sbi) { } 4280 static inline int __init f2fs_init_compress_cache(void) { return 0; } 4281 static inline void f2fs_destroy_compress_cache(void) { } 4282 static inline void f2fs_invalidate_compress_page(struct f2fs_sb_info *sbi, 4283 block_t blkaddr) { } 4284 static inline void f2fs_cache_compressed_page(struct f2fs_sb_info *sbi, 4285 struct page *page, nid_t ino, block_t blkaddr) { } 4286 static inline bool f2fs_load_compressed_page(struct f2fs_sb_info *sbi, 4287 struct page *page, block_t blkaddr) { return false; } 4288 static inline void f2fs_invalidate_compress_pages(struct f2fs_sb_info *sbi, 4289 nid_t ino) { } 4290 #define inc_compr_inode_stat(inode) do { } while (0) 4291 static inline void f2fs_update_extent_tree_range_compressed(struct inode *inode, 4292 pgoff_t fofs, block_t blkaddr, unsigned int llen, 4293 unsigned int c_len) { } 4294 #endif 4295 4296 static inline int set_compress_context(struct inode *inode) 4297 { 4298 #ifdef CONFIG_F2FS_FS_COMPRESSION 4299 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 4300 4301 F2FS_I(inode)->i_compress_algorithm = 4302 F2FS_OPTION(sbi).compress_algorithm; 4303 F2FS_I(inode)->i_log_cluster_size = 4304 F2FS_OPTION(sbi).compress_log_size; 4305 F2FS_I(inode)->i_compress_flag = 4306 F2FS_OPTION(sbi).compress_chksum ? 4307 1 << COMPRESS_CHKSUM : 0; 4308 F2FS_I(inode)->i_cluster_size = 4309 1 << F2FS_I(inode)->i_log_cluster_size; 4310 if ((F2FS_I(inode)->i_compress_algorithm == COMPRESS_LZ4 || 4311 F2FS_I(inode)->i_compress_algorithm == COMPRESS_ZSTD) && 4312 F2FS_OPTION(sbi).compress_level) 4313 F2FS_I(inode)->i_compress_flag |= 4314 F2FS_OPTION(sbi).compress_level << 4315 COMPRESS_LEVEL_OFFSET; 4316 F2FS_I(inode)->i_flags |= F2FS_COMPR_FL; 4317 set_inode_flag(inode, FI_COMPRESSED_FILE); 4318 stat_inc_compr_inode(inode); 4319 inc_compr_inode_stat(inode); 4320 f2fs_mark_inode_dirty_sync(inode, true); 4321 return 0; 4322 #else 4323 return -EOPNOTSUPP; 4324 #endif 4325 } 4326 4327 static inline bool f2fs_disable_compressed_file(struct inode *inode) 4328 { 4329 struct f2fs_inode_info *fi = F2FS_I(inode); 4330 4331 if (!f2fs_compressed_file(inode)) 4332 return true; 4333 if (S_ISREG(inode->i_mode) && F2FS_HAS_BLOCKS(inode)) 4334 return false; 4335 4336 fi->i_flags &= ~F2FS_COMPR_FL; 4337 stat_dec_compr_inode(inode); 4338 clear_inode_flag(inode, FI_COMPRESSED_FILE); 4339 f2fs_mark_inode_dirty_sync(inode, true); 4340 return true; 4341 } 4342 4343 #define F2FS_FEATURE_FUNCS(name, flagname) \ 4344 static inline int f2fs_sb_has_##name(struct f2fs_sb_info *sbi) \ 4345 { \ 4346 return F2FS_HAS_FEATURE(sbi, F2FS_FEATURE_##flagname); \ 4347 } 4348 4349 F2FS_FEATURE_FUNCS(encrypt, ENCRYPT); 4350 F2FS_FEATURE_FUNCS(blkzoned, BLKZONED); 4351 F2FS_FEATURE_FUNCS(extra_attr, EXTRA_ATTR); 4352 F2FS_FEATURE_FUNCS(project_quota, PRJQUOTA); 4353 F2FS_FEATURE_FUNCS(inode_chksum, INODE_CHKSUM); 4354 F2FS_FEATURE_FUNCS(flexible_inline_xattr, FLEXIBLE_INLINE_XATTR); 4355 F2FS_FEATURE_FUNCS(quota_ino, QUOTA_INO); 4356 F2FS_FEATURE_FUNCS(inode_crtime, INODE_CRTIME); 4357 F2FS_FEATURE_FUNCS(lost_found, LOST_FOUND); 4358 F2FS_FEATURE_FUNCS(verity, VERITY); 4359 F2FS_FEATURE_FUNCS(sb_chksum, SB_CHKSUM); 4360 F2FS_FEATURE_FUNCS(casefold, CASEFOLD); 4361 F2FS_FEATURE_FUNCS(compression, COMPRESSION); 4362 F2FS_FEATURE_FUNCS(readonly, RO); 4363 4364 static inline bool f2fs_may_extent_tree(struct inode *inode) 4365 { 4366 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 4367 4368 if (!test_opt(sbi, EXTENT_CACHE) || 4369 is_inode_flag_set(inode, FI_NO_EXTENT) || 4370 (is_inode_flag_set(inode, FI_COMPRESSED_FILE) && 4371 !f2fs_sb_has_readonly(sbi))) 4372 return false; 4373 4374 /* 4375 * for recovered files during mount do not create extents 4376 * if shrinker is not registered. 4377 */ 4378 if (list_empty(&sbi->s_list)) 4379 return false; 4380 4381 return S_ISREG(inode->i_mode); 4382 } 4383 4384 #ifdef CONFIG_BLK_DEV_ZONED 4385 static inline bool f2fs_blkz_is_seq(struct f2fs_sb_info *sbi, int devi, 4386 block_t blkaddr) 4387 { 4388 unsigned int zno = blkaddr >> sbi->log_blocks_per_blkz; 4389 4390 return test_bit(zno, FDEV(devi).blkz_seq); 4391 } 4392 #endif 4393 4394 static inline bool f2fs_hw_should_discard(struct f2fs_sb_info *sbi) 4395 { 4396 return f2fs_sb_has_blkzoned(sbi); 4397 } 4398 4399 static inline bool f2fs_bdev_support_discard(struct block_device *bdev) 4400 { 4401 return bdev_max_discard_sectors(bdev) || bdev_is_zoned(bdev); 4402 } 4403 4404 static inline bool f2fs_hw_support_discard(struct f2fs_sb_info *sbi) 4405 { 4406 int i; 4407 4408 if (!f2fs_is_multi_device(sbi)) 4409 return f2fs_bdev_support_discard(sbi->sb->s_bdev); 4410 4411 for (i = 0; i < sbi->s_ndevs; i++) 4412 if (f2fs_bdev_support_discard(FDEV(i).bdev)) 4413 return true; 4414 return false; 4415 } 4416 4417 static inline bool f2fs_realtime_discard_enable(struct f2fs_sb_info *sbi) 4418 { 4419 return (test_opt(sbi, DISCARD) && f2fs_hw_support_discard(sbi)) || 4420 f2fs_hw_should_discard(sbi); 4421 } 4422 4423 static inline bool f2fs_hw_is_readonly(struct f2fs_sb_info *sbi) 4424 { 4425 int i; 4426 4427 if (!f2fs_is_multi_device(sbi)) 4428 return bdev_read_only(sbi->sb->s_bdev); 4429 4430 for (i = 0; i < sbi->s_ndevs; i++) 4431 if (bdev_read_only(FDEV(i).bdev)) 4432 return true; 4433 return false; 4434 } 4435 4436 static inline bool f2fs_lfs_mode(struct f2fs_sb_info *sbi) 4437 { 4438 return F2FS_OPTION(sbi).fs_mode == FS_MODE_LFS; 4439 } 4440 4441 static inline bool f2fs_low_mem_mode(struct f2fs_sb_info *sbi) 4442 { 4443 return F2FS_OPTION(sbi).memory_mode == MEMORY_MODE_LOW; 4444 } 4445 4446 static inline bool f2fs_may_compress(struct inode *inode) 4447 { 4448 if (IS_SWAPFILE(inode) || f2fs_is_pinned_file(inode) || 4449 f2fs_is_atomic_file(inode) || f2fs_has_inline_data(inode)) 4450 return false; 4451 return S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode); 4452 } 4453 4454 static inline void f2fs_i_compr_blocks_update(struct inode *inode, 4455 u64 blocks, bool add) 4456 { 4457 struct f2fs_inode_info *fi = F2FS_I(inode); 4458 int diff = fi->i_cluster_size - blocks; 4459 4460 /* don't update i_compr_blocks if saved blocks were released */ 4461 if (!add && !atomic_read(&fi->i_compr_blocks)) 4462 return; 4463 4464 if (add) { 4465 atomic_add(diff, &fi->i_compr_blocks); 4466 stat_add_compr_blocks(inode, diff); 4467 } else { 4468 atomic_sub(diff, &fi->i_compr_blocks); 4469 stat_sub_compr_blocks(inode, diff); 4470 } 4471 f2fs_mark_inode_dirty_sync(inode, true); 4472 } 4473 4474 static inline int block_unaligned_IO(struct inode *inode, 4475 struct kiocb *iocb, struct iov_iter *iter) 4476 { 4477 unsigned int i_blkbits = READ_ONCE(inode->i_blkbits); 4478 unsigned int blocksize_mask = (1 << i_blkbits) - 1; 4479 loff_t offset = iocb->ki_pos; 4480 unsigned long align = offset | iov_iter_alignment(iter); 4481 4482 return align & blocksize_mask; 4483 } 4484 4485 static inline bool f2fs_allow_multi_device_dio(struct f2fs_sb_info *sbi, 4486 int flag) 4487 { 4488 if (!f2fs_is_multi_device(sbi)) 4489 return false; 4490 if (flag != F2FS_GET_BLOCK_DIO) 4491 return false; 4492 return sbi->aligned_blksize; 4493 } 4494 4495 static inline bool f2fs_force_buffered_io(struct inode *inode, 4496 struct kiocb *iocb, struct iov_iter *iter) 4497 { 4498 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 4499 int rw = iov_iter_rw(iter); 4500 4501 if (!fscrypt_dio_supported(iocb, iter)) 4502 return true; 4503 if (fsverity_active(inode)) 4504 return true; 4505 if (f2fs_compressed_file(inode)) 4506 return true; 4507 4508 /* disallow direct IO if any of devices has unaligned blksize */ 4509 if (f2fs_is_multi_device(sbi) && !sbi->aligned_blksize) 4510 return true; 4511 4512 if (f2fs_lfs_mode(sbi) && (rw == WRITE)) { 4513 if (block_unaligned_IO(inode, iocb, iter)) 4514 return true; 4515 if (F2FS_IO_ALIGNED(sbi)) 4516 return true; 4517 } 4518 if (is_sbi_flag_set(F2FS_I_SB(inode), SBI_CP_DISABLED)) 4519 return true; 4520 4521 return false; 4522 } 4523 4524 static inline bool f2fs_need_verity(const struct inode *inode, pgoff_t idx) 4525 { 4526 return fsverity_active(inode) && 4527 idx < DIV_ROUND_UP(inode->i_size, PAGE_SIZE); 4528 } 4529 4530 #ifdef CONFIG_F2FS_FAULT_INJECTION 4531 extern void f2fs_build_fault_attr(struct f2fs_sb_info *sbi, unsigned int rate, 4532 unsigned int type); 4533 #else 4534 #define f2fs_build_fault_attr(sbi, rate, type) do { } while (0) 4535 #endif 4536 4537 static inline bool is_journalled_quota(struct f2fs_sb_info *sbi) 4538 { 4539 #ifdef CONFIG_QUOTA 4540 if (f2fs_sb_has_quota_ino(sbi)) 4541 return true; 4542 if (F2FS_OPTION(sbi).s_qf_names[USRQUOTA] || 4543 F2FS_OPTION(sbi).s_qf_names[GRPQUOTA] || 4544 F2FS_OPTION(sbi).s_qf_names[PRJQUOTA]) 4545 return true; 4546 #endif 4547 return false; 4548 } 4549 4550 static inline bool f2fs_block_unit_discard(struct f2fs_sb_info *sbi) 4551 { 4552 return F2FS_OPTION(sbi).discard_unit == DISCARD_UNIT_BLOCK; 4553 } 4554 4555 static inline void f2fs_io_schedule_timeout(long timeout) 4556 { 4557 set_current_state(TASK_UNINTERRUPTIBLE); 4558 io_schedule_timeout(timeout); 4559 } 4560 4561 static inline void f2fs_handle_page_eio(struct f2fs_sb_info *sbi, pgoff_t ofs, 4562 enum page_type type) 4563 { 4564 if (unlikely(f2fs_cp_error(sbi))) 4565 return; 4566 4567 if (ofs == sbi->page_eio_ofs[type]) { 4568 if (sbi->page_eio_cnt[type]++ == MAX_RETRY_PAGE_EIO) 4569 set_ckpt_flags(sbi, CP_ERROR_FLAG); 4570 } else { 4571 sbi->page_eio_ofs[type] = ofs; 4572 sbi->page_eio_cnt[type] = 0; 4573 } 4574 } 4575 4576 #define EFSBADCRC EBADMSG /* Bad CRC detected */ 4577 #define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */ 4578 4579 #endif /* _LINUX_F2FS_H */ 4580