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 discard_unit; /* 163 * discard command's offset/size should 164 * be aligned to this unit: block, 165 * segment or section 166 */ 167 struct fscrypt_dummy_policy dummy_enc_policy; /* test dummy encryption */ 168 block_t unusable_cap_perc; /* percentage for cap */ 169 block_t unusable_cap; /* Amount of space allowed to be 170 * unusable when disabling checkpoint 171 */ 172 173 /* For compression */ 174 unsigned char compress_algorithm; /* algorithm type */ 175 unsigned char compress_log_size; /* cluster log size */ 176 unsigned char compress_level; /* compress level */ 177 bool compress_chksum; /* compressed data chksum */ 178 unsigned char compress_ext_cnt; /* extension count */ 179 unsigned char nocompress_ext_cnt; /* nocompress extension count */ 180 int compress_mode; /* compression mode */ 181 unsigned char extensions[COMPRESS_EXT_NUM][F2FS_EXTENSION_LEN]; /* extensions */ 182 unsigned char noextensions[COMPRESS_EXT_NUM][F2FS_EXTENSION_LEN]; /* extensions */ 183 }; 184 185 #define F2FS_FEATURE_ENCRYPT 0x0001 186 #define F2FS_FEATURE_BLKZONED 0x0002 187 #define F2FS_FEATURE_ATOMIC_WRITE 0x0004 188 #define F2FS_FEATURE_EXTRA_ATTR 0x0008 189 #define F2FS_FEATURE_PRJQUOTA 0x0010 190 #define F2FS_FEATURE_INODE_CHKSUM 0x0020 191 #define F2FS_FEATURE_FLEXIBLE_INLINE_XATTR 0x0040 192 #define F2FS_FEATURE_QUOTA_INO 0x0080 193 #define F2FS_FEATURE_INODE_CRTIME 0x0100 194 #define F2FS_FEATURE_LOST_FOUND 0x0200 195 #define F2FS_FEATURE_VERITY 0x0400 196 #define F2FS_FEATURE_SB_CHKSUM 0x0800 197 #define F2FS_FEATURE_CASEFOLD 0x1000 198 #define F2FS_FEATURE_COMPRESSION 0x2000 199 #define F2FS_FEATURE_RO 0x4000 200 201 #define __F2FS_HAS_FEATURE(raw_super, mask) \ 202 ((raw_super->feature & cpu_to_le32(mask)) != 0) 203 #define F2FS_HAS_FEATURE(sbi, mask) __F2FS_HAS_FEATURE(sbi->raw_super, mask) 204 #define F2FS_SET_FEATURE(sbi, mask) \ 205 (sbi->raw_super->feature |= cpu_to_le32(mask)) 206 #define F2FS_CLEAR_FEATURE(sbi, mask) \ 207 (sbi->raw_super->feature &= ~cpu_to_le32(mask)) 208 209 /* 210 * Default values for user and/or group using reserved blocks 211 */ 212 #define F2FS_DEF_RESUID 0 213 #define F2FS_DEF_RESGID 0 214 215 /* 216 * For checkpoint manager 217 */ 218 enum { 219 NAT_BITMAP, 220 SIT_BITMAP 221 }; 222 223 #define CP_UMOUNT 0x00000001 224 #define CP_FASTBOOT 0x00000002 225 #define CP_SYNC 0x00000004 226 #define CP_RECOVERY 0x00000008 227 #define CP_DISCARD 0x00000010 228 #define CP_TRIMMED 0x00000020 229 #define CP_PAUSE 0x00000040 230 #define CP_RESIZE 0x00000080 231 232 #define MAX_DISCARD_BLOCKS(sbi) BLKS_PER_SEC(sbi) 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 struct rb_entry { 602 struct rb_node rb_node; /* rb node located in rb-tree */ 603 union { 604 struct { 605 unsigned int ofs; /* start offset of the entry */ 606 unsigned int len; /* length of the entry */ 607 }; 608 unsigned long long key; /* 64-bits key */ 609 } __packed; 610 }; 611 612 struct extent_info { 613 unsigned int fofs; /* start offset in a file */ 614 unsigned int len; /* length of the extent */ 615 u32 blk; /* start block address of the extent */ 616 #ifdef CONFIG_F2FS_FS_COMPRESSION 617 unsigned int c_len; /* physical extent length of compressed blocks */ 618 #endif 619 }; 620 621 struct extent_node { 622 struct rb_node rb_node; /* rb node located in rb-tree */ 623 struct extent_info ei; /* extent info */ 624 struct list_head list; /* node in global extent list of sbi */ 625 struct extent_tree *et; /* extent tree pointer */ 626 }; 627 628 struct extent_tree { 629 nid_t ino; /* inode number */ 630 struct rb_root_cached root; /* root of extent info rb-tree */ 631 struct extent_node *cached_en; /* recently accessed extent node */ 632 struct extent_info largest; /* largested extent info */ 633 struct list_head list; /* to be used by sbi->zombie_list */ 634 rwlock_t lock; /* protect extent info rb-tree */ 635 atomic_t node_cnt; /* # of extent node in rb-tree*/ 636 bool largest_updated; /* largest extent updated */ 637 }; 638 639 /* 640 * This structure is taken from ext4_map_blocks. 641 * 642 * Note that, however, f2fs uses NEW and MAPPED flags for f2fs_map_blocks(). 643 */ 644 #define F2FS_MAP_NEW (1 << BH_New) 645 #define F2FS_MAP_MAPPED (1 << BH_Mapped) 646 #define F2FS_MAP_UNWRITTEN (1 << BH_Unwritten) 647 #define F2FS_MAP_FLAGS (F2FS_MAP_NEW | F2FS_MAP_MAPPED |\ 648 F2FS_MAP_UNWRITTEN) 649 650 struct f2fs_map_blocks { 651 struct block_device *m_bdev; /* for multi-device dio */ 652 block_t m_pblk; 653 block_t m_lblk; 654 unsigned int m_len; 655 unsigned int m_flags; 656 pgoff_t *m_next_pgofs; /* point next possible non-hole pgofs */ 657 pgoff_t *m_next_extent; /* point to next possible extent */ 658 int m_seg_type; 659 bool m_may_create; /* indicate it is from write path */ 660 bool m_multidev_dio; /* indicate it allows multi-device dio */ 661 }; 662 663 /* for flag in get_data_block */ 664 enum { 665 F2FS_GET_BLOCK_DEFAULT, 666 F2FS_GET_BLOCK_FIEMAP, 667 F2FS_GET_BLOCK_BMAP, 668 F2FS_GET_BLOCK_DIO, 669 F2FS_GET_BLOCK_PRE_DIO, 670 F2FS_GET_BLOCK_PRE_AIO, 671 F2FS_GET_BLOCK_PRECACHE, 672 }; 673 674 /* 675 * i_advise uses FADVISE_XXX_BIT. We can add additional hints later. 676 */ 677 #define FADVISE_COLD_BIT 0x01 678 #define FADVISE_LOST_PINO_BIT 0x02 679 #define FADVISE_ENCRYPT_BIT 0x04 680 #define FADVISE_ENC_NAME_BIT 0x08 681 #define FADVISE_KEEP_SIZE_BIT 0x10 682 #define FADVISE_HOT_BIT 0x20 683 #define FADVISE_VERITY_BIT 0x40 684 #define FADVISE_TRUNC_BIT 0x80 685 686 #define FADVISE_MODIFIABLE_BITS (FADVISE_COLD_BIT | FADVISE_HOT_BIT) 687 688 #define file_is_cold(inode) is_file(inode, FADVISE_COLD_BIT) 689 #define file_set_cold(inode) set_file(inode, FADVISE_COLD_BIT) 690 #define file_clear_cold(inode) clear_file(inode, FADVISE_COLD_BIT) 691 692 #define file_wrong_pino(inode) is_file(inode, FADVISE_LOST_PINO_BIT) 693 #define file_lost_pino(inode) set_file(inode, FADVISE_LOST_PINO_BIT) 694 #define file_got_pino(inode) clear_file(inode, FADVISE_LOST_PINO_BIT) 695 696 #define file_is_encrypt(inode) is_file(inode, FADVISE_ENCRYPT_BIT) 697 #define file_set_encrypt(inode) set_file(inode, FADVISE_ENCRYPT_BIT) 698 699 #define file_enc_name(inode) is_file(inode, FADVISE_ENC_NAME_BIT) 700 #define file_set_enc_name(inode) set_file(inode, FADVISE_ENC_NAME_BIT) 701 702 #define file_keep_isize(inode) is_file(inode, FADVISE_KEEP_SIZE_BIT) 703 #define file_set_keep_isize(inode) set_file(inode, FADVISE_KEEP_SIZE_BIT) 704 705 #define file_is_hot(inode) is_file(inode, FADVISE_HOT_BIT) 706 #define file_set_hot(inode) set_file(inode, FADVISE_HOT_BIT) 707 #define file_clear_hot(inode) clear_file(inode, FADVISE_HOT_BIT) 708 709 #define file_is_verity(inode) is_file(inode, FADVISE_VERITY_BIT) 710 #define file_set_verity(inode) set_file(inode, FADVISE_VERITY_BIT) 711 712 #define file_should_truncate(inode) is_file(inode, FADVISE_TRUNC_BIT) 713 #define file_need_truncate(inode) set_file(inode, FADVISE_TRUNC_BIT) 714 #define file_dont_truncate(inode) clear_file(inode, FADVISE_TRUNC_BIT) 715 716 #define DEF_DIR_LEVEL 0 717 718 enum { 719 GC_FAILURE_PIN, 720 MAX_GC_FAILURE 721 }; 722 723 /* used for f2fs_inode_info->flags */ 724 enum { 725 FI_NEW_INODE, /* indicate newly allocated inode */ 726 FI_DIRTY_INODE, /* indicate inode is dirty or not */ 727 FI_AUTO_RECOVER, /* indicate inode is recoverable */ 728 FI_DIRTY_DIR, /* indicate directory has dirty pages */ 729 FI_INC_LINK, /* need to increment i_nlink */ 730 FI_ACL_MODE, /* indicate acl mode */ 731 FI_NO_ALLOC, /* should not allocate any blocks */ 732 FI_FREE_NID, /* free allocated nide */ 733 FI_NO_EXTENT, /* not to use the extent cache */ 734 FI_INLINE_XATTR, /* used for inline xattr */ 735 FI_INLINE_DATA, /* used for inline data*/ 736 FI_INLINE_DENTRY, /* used for inline dentry */ 737 FI_APPEND_WRITE, /* inode has appended data */ 738 FI_UPDATE_WRITE, /* inode has in-place-update data */ 739 FI_NEED_IPU, /* used for ipu per file */ 740 FI_ATOMIC_FILE, /* indicate atomic file */ 741 FI_FIRST_BLOCK_WRITTEN, /* indicate #0 data block was written */ 742 FI_DROP_CACHE, /* drop dirty page cache */ 743 FI_DATA_EXIST, /* indicate data exists */ 744 FI_INLINE_DOTS, /* indicate inline dot dentries */ 745 FI_SKIP_WRITES, /* should skip data page writeback */ 746 FI_OPU_WRITE, /* used for opu per file */ 747 FI_DIRTY_FILE, /* indicate regular/symlink has dirty pages */ 748 FI_PREALLOCATED_ALL, /* all blocks for write were preallocated */ 749 FI_HOT_DATA, /* indicate file is hot */ 750 FI_EXTRA_ATTR, /* indicate file has extra attribute */ 751 FI_PROJ_INHERIT, /* indicate file inherits projectid */ 752 FI_PIN_FILE, /* indicate file should not be gced */ 753 FI_VERITY_IN_PROGRESS, /* building fs-verity Merkle tree */ 754 FI_COMPRESSED_FILE, /* indicate file's data can be compressed */ 755 FI_COMPRESS_CORRUPT, /* indicate compressed cluster is corrupted */ 756 FI_MMAP_FILE, /* indicate file was mmapped */ 757 FI_ENABLE_COMPRESS, /* enable compression in "user" compression mode */ 758 FI_COMPRESS_RELEASED, /* compressed blocks were released */ 759 FI_ALIGNED_WRITE, /* enable aligned write */ 760 FI_MAX, /* max flag, never be used */ 761 }; 762 763 struct f2fs_inode_info { 764 struct inode vfs_inode; /* serve a vfs inode */ 765 unsigned long i_flags; /* keep an inode flags for ioctl */ 766 unsigned char i_advise; /* use to give file attribute hints */ 767 unsigned char i_dir_level; /* use for dentry level for large dir */ 768 unsigned int i_current_depth; /* only for directory depth */ 769 /* for gc failure statistic */ 770 unsigned int i_gc_failures[MAX_GC_FAILURE]; 771 unsigned int i_pino; /* parent inode number */ 772 umode_t i_acl_mode; /* keep file acl mode temporarily */ 773 774 /* Use below internally in f2fs*/ 775 unsigned long flags[BITS_TO_LONGS(FI_MAX)]; /* use to pass per-file flags */ 776 struct f2fs_rwsem i_sem; /* protect fi info */ 777 atomic_t dirty_pages; /* # of dirty pages */ 778 f2fs_hash_t chash; /* hash value of given file name */ 779 unsigned int clevel; /* maximum level of given file name */ 780 struct task_struct *task; /* lookup and create consistency */ 781 struct task_struct *cp_task; /* separate cp/wb IO stats*/ 782 nid_t i_xattr_nid; /* node id that contains xattrs */ 783 loff_t last_disk_size; /* lastly written file size */ 784 spinlock_t i_size_lock; /* protect last_disk_size */ 785 786 #ifdef CONFIG_QUOTA 787 struct dquot *i_dquot[MAXQUOTAS]; 788 789 /* quota space reservation, managed internally by quota code */ 790 qsize_t i_reserved_quota; 791 #endif 792 struct list_head dirty_list; /* dirty list for dirs and files */ 793 struct list_head gdirty_list; /* linked in global dirty list */ 794 struct task_struct *atomic_write_task; /* store atomic write task */ 795 struct extent_tree *extent_tree; /* cached extent_tree entry */ 796 struct inode *cow_inode; /* copy-on-write inode for atomic write */ 797 798 /* avoid racing between foreground op and gc */ 799 struct f2fs_rwsem i_gc_rwsem[2]; 800 struct f2fs_rwsem i_xattr_sem; /* avoid racing between reading and changing EAs */ 801 802 int i_extra_isize; /* size of extra space located in i_addr */ 803 kprojid_t i_projid; /* id for project quota */ 804 int i_inline_xattr_size; /* inline xattr size */ 805 struct timespec64 i_crtime; /* inode creation time */ 806 struct timespec64 i_disk_time[4];/* inode disk times */ 807 808 /* for file compress */ 809 atomic_t i_compr_blocks; /* # of compressed blocks */ 810 unsigned char i_compress_algorithm; /* algorithm type */ 811 unsigned char i_log_cluster_size; /* log of cluster size */ 812 unsigned char i_compress_level; /* compress level (lz4hc,zstd) */ 813 unsigned short i_compress_flag; /* compress flag */ 814 unsigned int i_cluster_size; /* cluster size */ 815 }; 816 817 static inline void get_extent_info(struct extent_info *ext, 818 struct f2fs_extent *i_ext) 819 { 820 ext->fofs = le32_to_cpu(i_ext->fofs); 821 ext->blk = le32_to_cpu(i_ext->blk); 822 ext->len = le32_to_cpu(i_ext->len); 823 } 824 825 static inline void set_raw_extent(struct extent_info *ext, 826 struct f2fs_extent *i_ext) 827 { 828 i_ext->fofs = cpu_to_le32(ext->fofs); 829 i_ext->blk = cpu_to_le32(ext->blk); 830 i_ext->len = cpu_to_le32(ext->len); 831 } 832 833 static inline void set_extent_info(struct extent_info *ei, unsigned int fofs, 834 u32 blk, unsigned int len) 835 { 836 ei->fofs = fofs; 837 ei->blk = blk; 838 ei->len = len; 839 #ifdef CONFIG_F2FS_FS_COMPRESSION 840 ei->c_len = 0; 841 #endif 842 } 843 844 static inline bool __is_discard_mergeable(struct discard_info *back, 845 struct discard_info *front, unsigned int max_len) 846 { 847 return (back->lstart + back->len == front->lstart) && 848 (back->len + front->len <= max_len); 849 } 850 851 static inline bool __is_discard_back_mergeable(struct discard_info *cur, 852 struct discard_info *back, unsigned int max_len) 853 { 854 return __is_discard_mergeable(back, cur, max_len); 855 } 856 857 static inline bool __is_discard_front_mergeable(struct discard_info *cur, 858 struct discard_info *front, unsigned int max_len) 859 { 860 return __is_discard_mergeable(cur, front, max_len); 861 } 862 863 static inline bool __is_extent_mergeable(struct extent_info *back, 864 struct extent_info *front) 865 { 866 #ifdef CONFIG_F2FS_FS_COMPRESSION 867 if (back->c_len && back->len != back->c_len) 868 return false; 869 if (front->c_len && front->len != front->c_len) 870 return false; 871 #endif 872 return (back->fofs + back->len == front->fofs && 873 back->blk + back->len == front->blk); 874 } 875 876 static inline bool __is_back_mergeable(struct extent_info *cur, 877 struct extent_info *back) 878 { 879 return __is_extent_mergeable(back, cur); 880 } 881 882 static inline bool __is_front_mergeable(struct extent_info *cur, 883 struct extent_info *front) 884 { 885 return __is_extent_mergeable(cur, front); 886 } 887 888 extern void f2fs_mark_inode_dirty_sync(struct inode *inode, bool sync); 889 static inline void __try_update_largest_extent(struct extent_tree *et, 890 struct extent_node *en) 891 { 892 if (en->ei.len > et->largest.len) { 893 et->largest = en->ei; 894 et->largest_updated = true; 895 } 896 } 897 898 /* 899 * For free nid management 900 */ 901 enum nid_state { 902 FREE_NID, /* newly added to free nid list */ 903 PREALLOC_NID, /* it is preallocated */ 904 MAX_NID_STATE, 905 }; 906 907 enum nat_state { 908 TOTAL_NAT, 909 DIRTY_NAT, 910 RECLAIMABLE_NAT, 911 MAX_NAT_STATE, 912 }; 913 914 struct f2fs_nm_info { 915 block_t nat_blkaddr; /* base disk address of NAT */ 916 nid_t max_nid; /* maximum possible node ids */ 917 nid_t available_nids; /* # of available node ids */ 918 nid_t next_scan_nid; /* the next nid to be scanned */ 919 nid_t max_rf_node_blocks; /* max # of nodes for recovery */ 920 unsigned int ram_thresh; /* control the memory footprint */ 921 unsigned int ra_nid_pages; /* # of nid pages to be readaheaded */ 922 unsigned int dirty_nats_ratio; /* control dirty nats ratio threshold */ 923 924 /* NAT cache management */ 925 struct radix_tree_root nat_root;/* root of the nat entry cache */ 926 struct radix_tree_root nat_set_root;/* root of the nat set cache */ 927 struct f2fs_rwsem nat_tree_lock; /* protect nat entry tree */ 928 struct list_head nat_entries; /* cached nat entry list (clean) */ 929 spinlock_t nat_list_lock; /* protect clean nat entry list */ 930 unsigned int nat_cnt[MAX_NAT_STATE]; /* the # of cached nat entries */ 931 unsigned int nat_blocks; /* # of nat blocks */ 932 933 /* free node ids management */ 934 struct radix_tree_root free_nid_root;/* root of the free_nid cache */ 935 struct list_head free_nid_list; /* list for free nids excluding preallocated nids */ 936 unsigned int nid_cnt[MAX_NID_STATE]; /* the number of free node id */ 937 spinlock_t nid_list_lock; /* protect nid lists ops */ 938 struct mutex build_lock; /* lock for build free nids */ 939 unsigned char **free_nid_bitmap; 940 unsigned char *nat_block_bitmap; 941 unsigned short *free_nid_count; /* free nid count of NAT block */ 942 943 /* for checkpoint */ 944 char *nat_bitmap; /* NAT bitmap pointer */ 945 946 unsigned int nat_bits_blocks; /* # of nat bits blocks */ 947 unsigned char *nat_bits; /* NAT bits blocks */ 948 unsigned char *full_nat_bits; /* full NAT pages */ 949 unsigned char *empty_nat_bits; /* empty NAT pages */ 950 #ifdef CONFIG_F2FS_CHECK_FS 951 char *nat_bitmap_mir; /* NAT bitmap mirror */ 952 #endif 953 int bitmap_size; /* bitmap size */ 954 }; 955 956 /* 957 * this structure is used as one of function parameters. 958 * all the information are dedicated to a given direct node block determined 959 * by the data offset in a file. 960 */ 961 struct dnode_of_data { 962 struct inode *inode; /* vfs inode pointer */ 963 struct page *inode_page; /* its inode page, NULL is possible */ 964 struct page *node_page; /* cached direct node page */ 965 nid_t nid; /* node id of the direct node block */ 966 unsigned int ofs_in_node; /* data offset in the node page */ 967 bool inode_page_locked; /* inode page is locked or not */ 968 bool node_changed; /* is node block changed */ 969 char cur_level; /* level of hole node page */ 970 char max_level; /* level of current page located */ 971 block_t data_blkaddr; /* block address of the node block */ 972 }; 973 974 static inline void set_new_dnode(struct dnode_of_data *dn, struct inode *inode, 975 struct page *ipage, struct page *npage, nid_t nid) 976 { 977 memset(dn, 0, sizeof(*dn)); 978 dn->inode = inode; 979 dn->inode_page = ipage; 980 dn->node_page = npage; 981 dn->nid = nid; 982 } 983 984 /* 985 * For SIT manager 986 * 987 * By default, there are 6 active log areas across the whole main area. 988 * When considering hot and cold data separation to reduce cleaning overhead, 989 * we split 3 for data logs and 3 for node logs as hot, warm, and cold types, 990 * respectively. 991 * In the current design, you should not change the numbers intentionally. 992 * Instead, as a mount option such as active_logs=x, you can use 2, 4, and 6 993 * logs individually according to the underlying devices. (default: 6) 994 * Just in case, on-disk layout covers maximum 16 logs that consist of 8 for 995 * data and 8 for node logs. 996 */ 997 #define NR_CURSEG_DATA_TYPE (3) 998 #define NR_CURSEG_NODE_TYPE (3) 999 #define NR_CURSEG_INMEM_TYPE (2) 1000 #define NR_CURSEG_RO_TYPE (2) 1001 #define NR_CURSEG_PERSIST_TYPE (NR_CURSEG_DATA_TYPE + NR_CURSEG_NODE_TYPE) 1002 #define NR_CURSEG_TYPE (NR_CURSEG_INMEM_TYPE + NR_CURSEG_PERSIST_TYPE) 1003 1004 enum { 1005 CURSEG_HOT_DATA = 0, /* directory entry blocks */ 1006 CURSEG_WARM_DATA, /* data blocks */ 1007 CURSEG_COLD_DATA, /* multimedia or GCed data blocks */ 1008 CURSEG_HOT_NODE, /* direct node blocks of directory files */ 1009 CURSEG_WARM_NODE, /* direct node blocks of normal files */ 1010 CURSEG_COLD_NODE, /* indirect node blocks */ 1011 NR_PERSISTENT_LOG, /* number of persistent log */ 1012 CURSEG_COLD_DATA_PINNED = NR_PERSISTENT_LOG, 1013 /* pinned file that needs consecutive block address */ 1014 CURSEG_ALL_DATA_ATGC, /* SSR alloctor in hot/warm/cold data area */ 1015 NO_CHECK_TYPE, /* number of persistent & inmem log */ 1016 }; 1017 1018 struct flush_cmd { 1019 struct completion wait; 1020 struct llist_node llnode; 1021 nid_t ino; 1022 int ret; 1023 }; 1024 1025 struct flush_cmd_control { 1026 struct task_struct *f2fs_issue_flush; /* flush thread */ 1027 wait_queue_head_t flush_wait_queue; /* waiting queue for wake-up */ 1028 atomic_t issued_flush; /* # of issued flushes */ 1029 atomic_t queued_flush; /* # of queued flushes */ 1030 struct llist_head issue_list; /* list for command issue */ 1031 struct llist_node *dispatch_list; /* list for command dispatch */ 1032 }; 1033 1034 struct f2fs_sm_info { 1035 struct sit_info *sit_info; /* whole segment information */ 1036 struct free_segmap_info *free_info; /* free segment information */ 1037 struct dirty_seglist_info *dirty_info; /* dirty segment information */ 1038 struct curseg_info *curseg_array; /* active segment information */ 1039 1040 struct f2fs_rwsem curseg_lock; /* for preventing curseg change */ 1041 1042 block_t seg0_blkaddr; /* block address of 0'th segment */ 1043 block_t main_blkaddr; /* start block address of main area */ 1044 block_t ssa_blkaddr; /* start block address of SSA area */ 1045 1046 unsigned int segment_count; /* total # of segments */ 1047 unsigned int main_segments; /* # of segments in main area */ 1048 unsigned int reserved_segments; /* # of reserved segments */ 1049 unsigned int additional_reserved_segments;/* reserved segs for IO align feature */ 1050 unsigned int ovp_segments; /* # of overprovision segments */ 1051 1052 /* a threshold to reclaim prefree segments */ 1053 unsigned int rec_prefree_segments; 1054 1055 /* for batched trimming */ 1056 unsigned int trim_sections; /* # of sections to trim */ 1057 1058 struct list_head sit_entry_set; /* sit entry set list */ 1059 1060 unsigned int ipu_policy; /* in-place-update policy */ 1061 unsigned int min_ipu_util; /* in-place-update threshold */ 1062 unsigned int min_fsync_blocks; /* threshold for fsync */ 1063 unsigned int min_seq_blocks; /* threshold for sequential blocks */ 1064 unsigned int min_hot_blocks; /* threshold for hot block allocation */ 1065 unsigned int min_ssr_sections; /* threshold to trigger SSR allocation */ 1066 1067 /* for flush command control */ 1068 struct flush_cmd_control *fcc_info; 1069 1070 /* for discard command control */ 1071 struct discard_cmd_control *dcc_info; 1072 }; 1073 1074 /* 1075 * For superblock 1076 */ 1077 /* 1078 * COUNT_TYPE for monitoring 1079 * 1080 * f2fs monitors the number of several block types such as on-writeback, 1081 * dirty dentry blocks, dirty node blocks, and dirty meta blocks. 1082 */ 1083 #define WB_DATA_TYPE(p) (__is_cp_guaranteed(p) ? F2FS_WB_CP_DATA : F2FS_WB_DATA) 1084 enum count_type { 1085 F2FS_DIRTY_DENTS, 1086 F2FS_DIRTY_DATA, 1087 F2FS_DIRTY_QDATA, 1088 F2FS_DIRTY_NODES, 1089 F2FS_DIRTY_META, 1090 F2FS_DIRTY_IMETA, 1091 F2FS_WB_CP_DATA, 1092 F2FS_WB_DATA, 1093 F2FS_RD_DATA, 1094 F2FS_RD_NODE, 1095 F2FS_RD_META, 1096 F2FS_DIO_WRITE, 1097 F2FS_DIO_READ, 1098 NR_COUNT_TYPE, 1099 }; 1100 1101 /* 1102 * The below are the page types of bios used in submit_bio(). 1103 * The available types are: 1104 * DATA User data pages. It operates as async mode. 1105 * NODE Node pages. It operates as async mode. 1106 * META FS metadata pages such as SIT, NAT, CP. 1107 * NR_PAGE_TYPE The number of page types. 1108 * META_FLUSH Make sure the previous pages are written 1109 * with waiting the bio's completion 1110 * ... Only can be used with META. 1111 */ 1112 #define PAGE_TYPE_OF_BIO(type) ((type) > META ? META : (type)) 1113 enum page_type { 1114 DATA = 0, 1115 NODE = 1, /* should not change this */ 1116 META, 1117 NR_PAGE_TYPE, 1118 META_FLUSH, 1119 IPU, /* the below types are used by tracepoints only. */ 1120 OPU, 1121 }; 1122 1123 enum temp_type { 1124 HOT = 0, /* must be zero for meta bio */ 1125 WARM, 1126 COLD, 1127 NR_TEMP_TYPE, 1128 }; 1129 1130 enum need_lock_type { 1131 LOCK_REQ = 0, 1132 LOCK_DONE, 1133 LOCK_RETRY, 1134 }; 1135 1136 enum cp_reason_type { 1137 CP_NO_NEEDED, 1138 CP_NON_REGULAR, 1139 CP_COMPRESSED, 1140 CP_HARDLINK, 1141 CP_SB_NEED_CP, 1142 CP_WRONG_PINO, 1143 CP_NO_SPC_ROLL, 1144 CP_NODE_NEED_CP, 1145 CP_FASTBOOT_MODE, 1146 CP_SPEC_LOG_NUM, 1147 CP_RECOVER_DIR, 1148 }; 1149 1150 enum iostat_type { 1151 /* WRITE IO */ 1152 APP_DIRECT_IO, /* app direct write IOs */ 1153 APP_BUFFERED_IO, /* app buffered write IOs */ 1154 APP_WRITE_IO, /* app write IOs */ 1155 APP_MAPPED_IO, /* app mapped IOs */ 1156 FS_DATA_IO, /* data IOs from kworker/fsync/reclaimer */ 1157 FS_NODE_IO, /* node IOs from kworker/fsync/reclaimer */ 1158 FS_META_IO, /* meta IOs from kworker/reclaimer */ 1159 FS_GC_DATA_IO, /* data IOs from forground gc */ 1160 FS_GC_NODE_IO, /* node IOs from forground gc */ 1161 FS_CP_DATA_IO, /* data IOs from checkpoint */ 1162 FS_CP_NODE_IO, /* node IOs from checkpoint */ 1163 FS_CP_META_IO, /* meta IOs from checkpoint */ 1164 1165 /* READ IO */ 1166 APP_DIRECT_READ_IO, /* app direct read IOs */ 1167 APP_BUFFERED_READ_IO, /* app buffered read IOs */ 1168 APP_READ_IO, /* app read IOs */ 1169 APP_MAPPED_READ_IO, /* app mapped read IOs */ 1170 FS_DATA_READ_IO, /* data read IOs */ 1171 FS_GDATA_READ_IO, /* data read IOs from background gc */ 1172 FS_CDATA_READ_IO, /* compressed data read IOs */ 1173 FS_NODE_READ_IO, /* node read IOs */ 1174 FS_META_READ_IO, /* meta read IOs */ 1175 1176 /* other */ 1177 FS_DISCARD, /* discard */ 1178 NR_IO_TYPE, 1179 }; 1180 1181 struct f2fs_io_info { 1182 struct f2fs_sb_info *sbi; /* f2fs_sb_info pointer */ 1183 nid_t ino; /* inode number */ 1184 enum page_type type; /* contains DATA/NODE/META/META_FLUSH */ 1185 enum temp_type temp; /* contains HOT/WARM/COLD */ 1186 int op; /* contains REQ_OP_ */ 1187 int op_flags; /* req_flag_bits */ 1188 block_t new_blkaddr; /* new block address to be written */ 1189 block_t old_blkaddr; /* old block address before Cow */ 1190 struct page *page; /* page to be written */ 1191 struct page *encrypted_page; /* encrypted page */ 1192 struct page *compressed_page; /* compressed page */ 1193 struct list_head list; /* serialize IOs */ 1194 bool submitted; /* indicate IO submission */ 1195 int need_lock; /* indicate we need to lock cp_rwsem */ 1196 bool in_list; /* indicate fio is in io_list */ 1197 bool is_por; /* indicate IO is from recovery or not */ 1198 bool retry; /* need to reallocate block address */ 1199 int compr_blocks; /* # of compressed block addresses */ 1200 bool encrypted; /* indicate file is encrypted */ 1201 enum iostat_type io_type; /* io type */ 1202 struct writeback_control *io_wbc; /* writeback control */ 1203 struct bio **bio; /* bio for ipu */ 1204 sector_t *last_block; /* last block number in bio */ 1205 unsigned char version; /* version of the node */ 1206 }; 1207 1208 struct bio_entry { 1209 struct bio *bio; 1210 struct list_head list; 1211 }; 1212 1213 #define is_read_io(rw) ((rw) == READ) 1214 struct f2fs_bio_info { 1215 struct f2fs_sb_info *sbi; /* f2fs superblock */ 1216 struct bio *bio; /* bios to merge */ 1217 sector_t last_block_in_bio; /* last block number */ 1218 struct f2fs_io_info fio; /* store buffered io info. */ 1219 struct f2fs_rwsem io_rwsem; /* blocking op for bio */ 1220 spinlock_t io_lock; /* serialize DATA/NODE IOs */ 1221 struct list_head io_list; /* track fios */ 1222 struct list_head bio_list; /* bio entry list head */ 1223 struct f2fs_rwsem bio_list_lock; /* lock to protect bio entry list */ 1224 }; 1225 1226 #define FDEV(i) (sbi->devs[i]) 1227 #define RDEV(i) (raw_super->devs[i]) 1228 struct f2fs_dev_info { 1229 struct block_device *bdev; 1230 char path[MAX_PATH_LEN]; 1231 unsigned int total_segments; 1232 block_t start_blk; 1233 block_t end_blk; 1234 #ifdef CONFIG_BLK_DEV_ZONED 1235 unsigned int nr_blkz; /* Total number of zones */ 1236 unsigned long *blkz_seq; /* Bitmap indicating sequential zones */ 1237 block_t *zone_capacity_blocks; /* Array of zone capacity in blks */ 1238 #endif 1239 }; 1240 1241 enum inode_type { 1242 DIR_INODE, /* for dirty dir inode */ 1243 FILE_INODE, /* for dirty regular/symlink inode */ 1244 DIRTY_META, /* for all dirtied inode metadata */ 1245 ATOMIC_FILE, /* for all atomic files */ 1246 NR_INODE_TYPE, 1247 }; 1248 1249 /* for inner inode cache management */ 1250 struct inode_management { 1251 struct radix_tree_root ino_root; /* ino entry array */ 1252 spinlock_t ino_lock; /* for ino entry lock */ 1253 struct list_head ino_list; /* inode list head */ 1254 unsigned long ino_num; /* number of entries */ 1255 }; 1256 1257 /* for GC_AT */ 1258 struct atgc_management { 1259 bool atgc_enabled; /* ATGC is enabled or not */ 1260 struct rb_root_cached root; /* root of victim rb-tree */ 1261 struct list_head victim_list; /* linked with all victim entries */ 1262 unsigned int victim_count; /* victim count in rb-tree */ 1263 unsigned int candidate_ratio; /* candidate ratio */ 1264 unsigned int max_candidate_count; /* max candidate count */ 1265 unsigned int age_weight; /* age weight, vblock_weight = 100 - age_weight */ 1266 unsigned long long age_threshold; /* age threshold */ 1267 }; 1268 1269 struct f2fs_gc_control { 1270 unsigned int victim_segno; /* target victim segment number */ 1271 int init_gc_type; /* FG_GC or BG_GC */ 1272 bool no_bg_gc; /* check the space and stop bg_gc */ 1273 bool should_migrate_blocks; /* should migrate blocks */ 1274 bool err_gc_skipped; /* return EAGAIN if GC skipped */ 1275 unsigned int nr_free_secs; /* # of free sections to do GC */ 1276 }; 1277 1278 /* For s_flag in struct f2fs_sb_info */ 1279 enum { 1280 SBI_IS_DIRTY, /* dirty flag for checkpoint */ 1281 SBI_IS_CLOSE, /* specify unmounting */ 1282 SBI_NEED_FSCK, /* need fsck.f2fs to fix */ 1283 SBI_POR_DOING, /* recovery is doing or not */ 1284 SBI_NEED_SB_WRITE, /* need to recover superblock */ 1285 SBI_NEED_CP, /* need to checkpoint */ 1286 SBI_IS_SHUTDOWN, /* shutdown by ioctl */ 1287 SBI_IS_RECOVERED, /* recovered orphan/data */ 1288 SBI_CP_DISABLED, /* CP was disabled last mount */ 1289 SBI_CP_DISABLED_QUICK, /* CP was disabled quickly */ 1290 SBI_QUOTA_NEED_FLUSH, /* need to flush quota info in CP */ 1291 SBI_QUOTA_SKIP_FLUSH, /* skip flushing quota in current CP */ 1292 SBI_QUOTA_NEED_REPAIR, /* quota file may be corrupted */ 1293 SBI_IS_RESIZEFS, /* resizefs is in process */ 1294 SBI_IS_FREEZING, /* freezefs is in process */ 1295 }; 1296 1297 enum { 1298 CP_TIME, 1299 REQ_TIME, 1300 DISCARD_TIME, 1301 GC_TIME, 1302 DISABLE_TIME, 1303 UMOUNT_DISCARD_TIMEOUT, 1304 MAX_TIME, 1305 }; 1306 1307 enum { 1308 GC_NORMAL, 1309 GC_IDLE_CB, 1310 GC_IDLE_GREEDY, 1311 GC_IDLE_AT, 1312 GC_URGENT_HIGH, 1313 GC_URGENT_LOW, 1314 GC_URGENT_MID, 1315 MAX_GC_MODE, 1316 }; 1317 1318 enum { 1319 BGGC_MODE_ON, /* background gc is on */ 1320 BGGC_MODE_OFF, /* background gc is off */ 1321 BGGC_MODE_SYNC, /* 1322 * background gc is on, migrating blocks 1323 * like foreground gc 1324 */ 1325 }; 1326 1327 enum { 1328 FS_MODE_ADAPTIVE, /* use both lfs/ssr allocation */ 1329 FS_MODE_LFS, /* use lfs allocation only */ 1330 FS_MODE_FRAGMENT_SEG, /* segment fragmentation mode */ 1331 FS_MODE_FRAGMENT_BLK, /* block fragmentation mode */ 1332 }; 1333 1334 enum { 1335 ALLOC_MODE_DEFAULT, /* stay default */ 1336 ALLOC_MODE_REUSE, /* reuse segments as much as possible */ 1337 }; 1338 1339 enum fsync_mode { 1340 FSYNC_MODE_POSIX, /* fsync follows posix semantics */ 1341 FSYNC_MODE_STRICT, /* fsync behaves in line with ext4 */ 1342 FSYNC_MODE_NOBARRIER, /* fsync behaves nobarrier based on posix */ 1343 }; 1344 1345 enum { 1346 COMPR_MODE_FS, /* 1347 * automatically compress compression 1348 * enabled files 1349 */ 1350 COMPR_MODE_USER, /* 1351 * automatical compression is disabled. 1352 * user can control the file compression 1353 * using ioctls 1354 */ 1355 }; 1356 1357 enum { 1358 DISCARD_UNIT_BLOCK, /* basic discard unit is block */ 1359 DISCARD_UNIT_SEGMENT, /* basic discard unit is segment */ 1360 DISCARD_UNIT_SECTION, /* basic discard unit is section */ 1361 }; 1362 1363 static inline int f2fs_test_bit(unsigned int nr, char *addr); 1364 static inline void f2fs_set_bit(unsigned int nr, char *addr); 1365 static inline void f2fs_clear_bit(unsigned int nr, char *addr); 1366 1367 /* 1368 * Layout of f2fs page.private: 1369 * 1370 * Layout A: lowest bit should be 1 1371 * | bit0 = 1 | bit1 | bit2 | ... | bit MAX | private data .... | 1372 * bit 0 PAGE_PRIVATE_NOT_POINTER 1373 * bit 1 PAGE_PRIVATE_ATOMIC_WRITE 1374 * bit 2 PAGE_PRIVATE_DUMMY_WRITE 1375 * bit 3 PAGE_PRIVATE_ONGOING_MIGRATION 1376 * bit 4 PAGE_PRIVATE_INLINE_INODE 1377 * bit 5 PAGE_PRIVATE_REF_RESOURCE 1378 * bit 6- f2fs private data 1379 * 1380 * Layout B: lowest bit should be 0 1381 * page.private is a wrapped pointer. 1382 */ 1383 enum { 1384 PAGE_PRIVATE_NOT_POINTER, /* private contains non-pointer data */ 1385 PAGE_PRIVATE_ATOMIC_WRITE, /* data page from atomic write path */ 1386 PAGE_PRIVATE_DUMMY_WRITE, /* data page for padding aligned IO */ 1387 PAGE_PRIVATE_ONGOING_MIGRATION, /* data page which is on-going migrating */ 1388 PAGE_PRIVATE_INLINE_INODE, /* inode page contains inline data */ 1389 PAGE_PRIVATE_REF_RESOURCE, /* dirty page has referenced resources */ 1390 PAGE_PRIVATE_MAX 1391 }; 1392 1393 #define PAGE_PRIVATE_GET_FUNC(name, flagname) \ 1394 static inline bool page_private_##name(struct page *page) \ 1395 { \ 1396 return PagePrivate(page) && \ 1397 test_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)) && \ 1398 test_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \ 1399 } 1400 1401 #define PAGE_PRIVATE_SET_FUNC(name, flagname) \ 1402 static inline void set_page_private_##name(struct page *page) \ 1403 { \ 1404 if (!PagePrivate(page)) { \ 1405 get_page(page); \ 1406 SetPagePrivate(page); \ 1407 set_page_private(page, 0); \ 1408 } \ 1409 set_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)); \ 1410 set_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \ 1411 } 1412 1413 #define PAGE_PRIVATE_CLEAR_FUNC(name, flagname) \ 1414 static inline void clear_page_private_##name(struct page *page) \ 1415 { \ 1416 clear_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \ 1417 if (page_private(page) == 1 << PAGE_PRIVATE_NOT_POINTER) { \ 1418 set_page_private(page, 0); \ 1419 if (PagePrivate(page)) { \ 1420 ClearPagePrivate(page); \ 1421 put_page(page); \ 1422 }\ 1423 } \ 1424 } 1425 1426 PAGE_PRIVATE_GET_FUNC(nonpointer, NOT_POINTER); 1427 PAGE_PRIVATE_GET_FUNC(reference, REF_RESOURCE); 1428 PAGE_PRIVATE_GET_FUNC(inline, INLINE_INODE); 1429 PAGE_PRIVATE_GET_FUNC(gcing, ONGOING_MIGRATION); 1430 PAGE_PRIVATE_GET_FUNC(atomic, ATOMIC_WRITE); 1431 PAGE_PRIVATE_GET_FUNC(dummy, DUMMY_WRITE); 1432 1433 PAGE_PRIVATE_SET_FUNC(reference, REF_RESOURCE); 1434 PAGE_PRIVATE_SET_FUNC(inline, INLINE_INODE); 1435 PAGE_PRIVATE_SET_FUNC(gcing, ONGOING_MIGRATION); 1436 PAGE_PRIVATE_SET_FUNC(atomic, ATOMIC_WRITE); 1437 PAGE_PRIVATE_SET_FUNC(dummy, DUMMY_WRITE); 1438 1439 PAGE_PRIVATE_CLEAR_FUNC(reference, REF_RESOURCE); 1440 PAGE_PRIVATE_CLEAR_FUNC(inline, INLINE_INODE); 1441 PAGE_PRIVATE_CLEAR_FUNC(gcing, ONGOING_MIGRATION); 1442 PAGE_PRIVATE_CLEAR_FUNC(atomic, ATOMIC_WRITE); 1443 PAGE_PRIVATE_CLEAR_FUNC(dummy, DUMMY_WRITE); 1444 1445 static inline unsigned long get_page_private_data(struct page *page) 1446 { 1447 unsigned long data = page_private(page); 1448 1449 if (!test_bit(PAGE_PRIVATE_NOT_POINTER, &data)) 1450 return 0; 1451 return data >> PAGE_PRIVATE_MAX; 1452 } 1453 1454 static inline void set_page_private_data(struct page *page, unsigned long data) 1455 { 1456 if (!PagePrivate(page)) { 1457 get_page(page); 1458 SetPagePrivate(page); 1459 set_page_private(page, 0); 1460 } 1461 set_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)); 1462 page_private(page) |= data << PAGE_PRIVATE_MAX; 1463 } 1464 1465 static inline void clear_page_private_data(struct page *page) 1466 { 1467 page_private(page) &= (1 << PAGE_PRIVATE_MAX) - 1; 1468 if (page_private(page) == 1 << PAGE_PRIVATE_NOT_POINTER) { 1469 set_page_private(page, 0); 1470 if (PagePrivate(page)) { 1471 ClearPagePrivate(page); 1472 put_page(page); 1473 } 1474 } 1475 } 1476 1477 /* For compression */ 1478 enum compress_algorithm_type { 1479 COMPRESS_LZO, 1480 COMPRESS_LZ4, 1481 COMPRESS_ZSTD, 1482 COMPRESS_LZORLE, 1483 COMPRESS_MAX, 1484 }; 1485 1486 enum compress_flag { 1487 COMPRESS_CHKSUM, 1488 COMPRESS_MAX_FLAG, 1489 }; 1490 1491 #define COMPRESS_WATERMARK 20 1492 #define COMPRESS_PERCENT 20 1493 1494 #define COMPRESS_DATA_RESERVED_SIZE 4 1495 struct compress_data { 1496 __le32 clen; /* compressed data size */ 1497 __le32 chksum; /* compressed data chksum */ 1498 __le32 reserved[COMPRESS_DATA_RESERVED_SIZE]; /* reserved */ 1499 u8 cdata[]; /* compressed data */ 1500 }; 1501 1502 #define COMPRESS_HEADER_SIZE (sizeof(struct compress_data)) 1503 1504 #define F2FS_COMPRESSED_PAGE_MAGIC 0xF5F2C000 1505 1506 #define COMPRESS_LEVEL_OFFSET 8 1507 1508 /* compress context */ 1509 struct compress_ctx { 1510 struct inode *inode; /* inode the context belong to */ 1511 pgoff_t cluster_idx; /* cluster index number */ 1512 unsigned int cluster_size; /* page count in cluster */ 1513 unsigned int log_cluster_size; /* log of cluster size */ 1514 struct page **rpages; /* pages store raw data in cluster */ 1515 unsigned int nr_rpages; /* total page number in rpages */ 1516 struct page **cpages; /* pages store compressed data in cluster */ 1517 unsigned int nr_cpages; /* total page number in cpages */ 1518 unsigned int valid_nr_cpages; /* valid page number in cpages */ 1519 void *rbuf; /* virtual mapped address on rpages */ 1520 struct compress_data *cbuf; /* virtual mapped address on cpages */ 1521 size_t rlen; /* valid data length in rbuf */ 1522 size_t clen; /* valid data length in cbuf */ 1523 void *private; /* payload buffer for specified compression algorithm */ 1524 void *private2; /* extra payload buffer */ 1525 }; 1526 1527 /* compress context for write IO path */ 1528 struct compress_io_ctx { 1529 u32 magic; /* magic number to indicate page is compressed */ 1530 struct inode *inode; /* inode the context belong to */ 1531 struct page **rpages; /* pages store raw data in cluster */ 1532 unsigned int nr_rpages; /* total page number in rpages */ 1533 atomic_t pending_pages; /* in-flight compressed page count */ 1534 }; 1535 1536 /* Context for decompressing one cluster on the read IO path */ 1537 struct decompress_io_ctx { 1538 u32 magic; /* magic number to indicate page is compressed */ 1539 struct inode *inode; /* inode the context belong to */ 1540 pgoff_t cluster_idx; /* cluster index number */ 1541 unsigned int cluster_size; /* page count in cluster */ 1542 unsigned int log_cluster_size; /* log of cluster size */ 1543 struct page **rpages; /* pages store raw data in cluster */ 1544 unsigned int nr_rpages; /* total page number in rpages */ 1545 struct page **cpages; /* pages store compressed data in cluster */ 1546 unsigned int nr_cpages; /* total page number in cpages */ 1547 struct page **tpages; /* temp pages to pad holes in cluster */ 1548 void *rbuf; /* virtual mapped address on rpages */ 1549 struct compress_data *cbuf; /* virtual mapped address on cpages */ 1550 size_t rlen; /* valid data length in rbuf */ 1551 size_t clen; /* valid data length in cbuf */ 1552 1553 /* 1554 * The number of compressed pages remaining to be read in this cluster. 1555 * This is initially nr_cpages. It is decremented by 1 each time a page 1556 * has been read (or failed to be read). When it reaches 0, the cluster 1557 * is decompressed (or an error is reported). 1558 * 1559 * If an error occurs before all the pages have been submitted for I/O, 1560 * then this will never reach 0. In this case the I/O submitter is 1561 * responsible for calling f2fs_decompress_end_io() instead. 1562 */ 1563 atomic_t remaining_pages; 1564 1565 /* 1566 * Number of references to this decompress_io_ctx. 1567 * 1568 * One reference is held for I/O completion. This reference is dropped 1569 * after the pagecache pages are updated and unlocked -- either after 1570 * decompression (and verity if enabled), or after an error. 1571 * 1572 * In addition, each compressed page holds a reference while it is in a 1573 * bio. These references are necessary prevent compressed pages from 1574 * being freed while they are still in a bio. 1575 */ 1576 refcount_t refcnt; 1577 1578 bool failed; /* IO error occurred before decompression? */ 1579 bool need_verity; /* need fs-verity verification after decompression? */ 1580 void *private; /* payload buffer for specified decompression algorithm */ 1581 void *private2; /* extra payload buffer */ 1582 struct work_struct verity_work; /* work to verify the decompressed pages */ 1583 }; 1584 1585 #define NULL_CLUSTER ((unsigned int)(~0)) 1586 #define MIN_COMPRESS_LOG_SIZE 2 1587 #define MAX_COMPRESS_LOG_SIZE 8 1588 #define MAX_COMPRESS_WINDOW_SIZE(log_size) ((PAGE_SIZE) << (log_size)) 1589 1590 struct f2fs_sb_info { 1591 struct super_block *sb; /* pointer to VFS super block */ 1592 struct proc_dir_entry *s_proc; /* proc entry */ 1593 struct f2fs_super_block *raw_super; /* raw super block pointer */ 1594 struct f2fs_rwsem sb_lock; /* lock for raw super block */ 1595 int valid_super_block; /* valid super block no */ 1596 unsigned long s_flag; /* flags for sbi */ 1597 struct mutex writepages; /* mutex for writepages() */ 1598 1599 #ifdef CONFIG_BLK_DEV_ZONED 1600 unsigned int blocks_per_blkz; /* F2FS blocks per zone */ 1601 unsigned int log_blocks_per_blkz; /* log2 F2FS blocks per zone */ 1602 #endif 1603 1604 /* for node-related operations */ 1605 struct f2fs_nm_info *nm_info; /* node manager */ 1606 struct inode *node_inode; /* cache node blocks */ 1607 1608 /* for segment-related operations */ 1609 struct f2fs_sm_info *sm_info; /* segment manager */ 1610 1611 /* for bio operations */ 1612 struct f2fs_bio_info *write_io[NR_PAGE_TYPE]; /* for write bios */ 1613 /* keep migration IO order for LFS mode */ 1614 struct f2fs_rwsem io_order_lock; 1615 mempool_t *write_io_dummy; /* Dummy pages */ 1616 pgoff_t page_eio_ofs[NR_PAGE_TYPE]; /* EIO page offset */ 1617 int page_eio_cnt[NR_PAGE_TYPE]; /* EIO count */ 1618 1619 /* for checkpoint */ 1620 struct f2fs_checkpoint *ckpt; /* raw checkpoint pointer */ 1621 int cur_cp_pack; /* remain current cp pack */ 1622 spinlock_t cp_lock; /* for flag in ckpt */ 1623 struct inode *meta_inode; /* cache meta blocks */ 1624 struct f2fs_rwsem cp_global_sem; /* checkpoint procedure lock */ 1625 struct f2fs_rwsem cp_rwsem; /* blocking FS operations */ 1626 struct f2fs_rwsem node_write; /* locking node writes */ 1627 struct f2fs_rwsem node_change; /* locking node change */ 1628 wait_queue_head_t cp_wait; 1629 unsigned long last_time[MAX_TIME]; /* to store time in jiffies */ 1630 long interval_time[MAX_TIME]; /* to store thresholds */ 1631 struct ckpt_req_control cprc_info; /* for checkpoint request control */ 1632 1633 struct inode_management im[MAX_INO_ENTRY]; /* manage inode cache */ 1634 1635 spinlock_t fsync_node_lock; /* for node entry lock */ 1636 struct list_head fsync_node_list; /* node list head */ 1637 unsigned int fsync_seg_id; /* sequence id */ 1638 unsigned int fsync_node_num; /* number of node entries */ 1639 1640 /* for orphan inode, use 0'th array */ 1641 unsigned int max_orphans; /* max orphan inodes */ 1642 1643 /* for inode management */ 1644 struct list_head inode_list[NR_INODE_TYPE]; /* dirty inode list */ 1645 spinlock_t inode_lock[NR_INODE_TYPE]; /* for dirty inode list lock */ 1646 struct mutex flush_lock; /* for flush exclusion */ 1647 1648 /* for extent tree cache */ 1649 struct radix_tree_root extent_tree_root;/* cache extent cache entries */ 1650 struct mutex extent_tree_lock; /* locking extent radix tree */ 1651 struct list_head extent_list; /* lru list for shrinker */ 1652 spinlock_t extent_lock; /* locking extent lru list */ 1653 atomic_t total_ext_tree; /* extent tree count */ 1654 struct list_head zombie_list; /* extent zombie tree list */ 1655 atomic_t total_zombie_tree; /* extent zombie tree count */ 1656 atomic_t total_ext_node; /* extent info count */ 1657 1658 /* basic filesystem units */ 1659 unsigned int log_sectors_per_block; /* log2 sectors per block */ 1660 unsigned int log_blocksize; /* log2 block size */ 1661 unsigned int blocksize; /* block size */ 1662 unsigned int root_ino_num; /* root inode number*/ 1663 unsigned int node_ino_num; /* node inode number*/ 1664 unsigned int meta_ino_num; /* meta inode number*/ 1665 unsigned int log_blocks_per_seg; /* log2 blocks per segment */ 1666 unsigned int blocks_per_seg; /* blocks per segment */ 1667 unsigned int segs_per_sec; /* segments per section */ 1668 unsigned int secs_per_zone; /* sections per zone */ 1669 unsigned int total_sections; /* total section count */ 1670 unsigned int total_node_count; /* total node block count */ 1671 unsigned int total_valid_node_count; /* valid node block count */ 1672 int dir_level; /* directory level */ 1673 int readdir_ra; /* readahead inode in readdir */ 1674 u64 max_io_bytes; /* max io bytes to merge IOs */ 1675 1676 block_t user_block_count; /* # of user blocks */ 1677 block_t total_valid_block_count; /* # of valid blocks */ 1678 block_t discard_blks; /* discard command candidats */ 1679 block_t last_valid_block_count; /* for recovery */ 1680 block_t reserved_blocks; /* configurable reserved blocks */ 1681 block_t current_reserved_blocks; /* current reserved blocks */ 1682 1683 /* Additional tracking for no checkpoint mode */ 1684 block_t unusable_block_count; /* # of blocks saved by last cp */ 1685 1686 unsigned int nquota_files; /* # of quota sysfile */ 1687 struct f2fs_rwsem quota_sem; /* blocking cp for flags */ 1688 1689 /* # of pages, see count_type */ 1690 atomic_t nr_pages[NR_COUNT_TYPE]; 1691 /* # of allocated blocks */ 1692 struct percpu_counter alloc_valid_block_count; 1693 /* # of node block writes as roll forward recovery */ 1694 struct percpu_counter rf_node_block_count; 1695 1696 /* writeback control */ 1697 atomic_t wb_sync_req[META]; /* count # of WB_SYNC threads */ 1698 1699 /* valid inode count */ 1700 struct percpu_counter total_valid_inode_count; 1701 1702 struct f2fs_mount_info mount_opt; /* mount options */ 1703 1704 /* for cleaning operations */ 1705 struct f2fs_rwsem gc_lock; /* 1706 * semaphore for GC, avoid 1707 * race between GC and GC or CP 1708 */ 1709 struct f2fs_gc_kthread *gc_thread; /* GC thread */ 1710 struct atgc_management am; /* atgc management */ 1711 unsigned int cur_victim_sec; /* current victim section num */ 1712 unsigned int gc_mode; /* current GC state */ 1713 unsigned int next_victim_seg[2]; /* next segment in victim section */ 1714 spinlock_t gc_urgent_high_lock; 1715 bool gc_urgent_high_limited; /* indicates having limited trial count */ 1716 unsigned int gc_urgent_high_remaining; /* remaining trial count for GC_URGENT_HIGH */ 1717 1718 /* for skip statistic */ 1719 unsigned int atomic_files; /* # of opened atomic file */ 1720 unsigned long long skipped_gc_rwsem; /* FG_GC only */ 1721 1722 /* threshold for gc trials on pinned files */ 1723 u64 gc_pin_file_threshold; 1724 struct f2fs_rwsem pin_sem; 1725 1726 /* maximum # of trials to find a victim segment for SSR and GC */ 1727 unsigned int max_victim_search; 1728 /* migration granularity of garbage collection, unit: segment */ 1729 unsigned int migration_granularity; 1730 1731 /* 1732 * for stat information. 1733 * one is for the LFS mode, and the other is for the SSR mode. 1734 */ 1735 #ifdef CONFIG_F2FS_STAT_FS 1736 struct f2fs_stat_info *stat_info; /* FS status information */ 1737 atomic_t meta_count[META_MAX]; /* # of meta blocks */ 1738 unsigned int segment_count[2]; /* # of allocated segments */ 1739 unsigned int block_count[2]; /* # of allocated blocks */ 1740 atomic_t inplace_count; /* # of inplace update */ 1741 atomic64_t total_hit_ext; /* # of lookup extent cache */ 1742 atomic64_t read_hit_rbtree; /* # of hit rbtree extent node */ 1743 atomic64_t read_hit_largest; /* # of hit largest extent node */ 1744 atomic64_t read_hit_cached; /* # of hit cached extent node */ 1745 atomic_t inline_xattr; /* # of inline_xattr inodes */ 1746 atomic_t inline_inode; /* # of inline_data inodes */ 1747 atomic_t inline_dir; /* # of inline_dentry inodes */ 1748 atomic_t compr_inode; /* # of compressed inodes */ 1749 atomic64_t compr_blocks; /* # of compressed blocks */ 1750 atomic_t max_aw_cnt; /* max # of atomic writes */ 1751 unsigned int io_skip_bggc; /* skip background gc for in-flight IO */ 1752 unsigned int other_skip_bggc; /* skip background gc for other reasons */ 1753 unsigned int ndirty_inode[NR_INODE_TYPE]; /* # of dirty inodes */ 1754 #endif 1755 spinlock_t stat_lock; /* lock for stat operations */ 1756 1757 /* to attach REQ_META|REQ_FUA flags */ 1758 unsigned int data_io_flag; 1759 unsigned int node_io_flag; 1760 1761 /* For sysfs support */ 1762 struct kobject s_kobj; /* /sys/fs/f2fs/<devname> */ 1763 struct completion s_kobj_unregister; 1764 1765 struct kobject s_stat_kobj; /* /sys/fs/f2fs/<devname>/stat */ 1766 struct completion s_stat_kobj_unregister; 1767 1768 struct kobject s_feature_list_kobj; /* /sys/fs/f2fs/<devname>/feature_list */ 1769 struct completion s_feature_list_kobj_unregister; 1770 1771 /* For shrinker support */ 1772 struct list_head s_list; 1773 struct mutex umount_mutex; 1774 unsigned int shrinker_run_no; 1775 1776 /* For multi devices */ 1777 int s_ndevs; /* number of devices */ 1778 struct f2fs_dev_info *devs; /* for device list */ 1779 unsigned int dirty_device; /* for checkpoint data flush */ 1780 spinlock_t dev_lock; /* protect dirty_device */ 1781 bool aligned_blksize; /* all devices has the same logical blksize */ 1782 1783 /* For write statistics */ 1784 u64 sectors_written_start; 1785 u64 kbytes_written; 1786 1787 /* Reference to checksum algorithm driver via cryptoapi */ 1788 struct crypto_shash *s_chksum_driver; 1789 1790 /* Precomputed FS UUID checksum for seeding other checksums */ 1791 __u32 s_chksum_seed; 1792 1793 struct workqueue_struct *post_read_wq; /* post read workqueue */ 1794 1795 struct kmem_cache *inline_xattr_slab; /* inline xattr entry */ 1796 unsigned int inline_xattr_slab_size; /* default inline xattr slab size */ 1797 1798 /* For reclaimed segs statistics per each GC mode */ 1799 unsigned int gc_segment_mode; /* GC state for reclaimed segments */ 1800 unsigned int gc_reclaimed_segs[MAX_GC_MODE]; /* Reclaimed segs for each mode */ 1801 1802 unsigned long seq_file_ra_mul; /* multiplier for ra_pages of seq. files in fadvise */ 1803 1804 int max_fragment_chunk; /* max chunk size for block fragmentation mode */ 1805 int max_fragment_hole; /* max hole size for block fragmentation mode */ 1806 1807 #ifdef CONFIG_F2FS_FS_COMPRESSION 1808 struct kmem_cache *page_array_slab; /* page array entry */ 1809 unsigned int page_array_slab_size; /* default page array slab size */ 1810 1811 /* For runtime compression statistics */ 1812 u64 compr_written_block; 1813 u64 compr_saved_block; 1814 u32 compr_new_inode; 1815 1816 /* For compressed block cache */ 1817 struct inode *compress_inode; /* cache compressed blocks */ 1818 unsigned int compress_percent; /* cache page percentage */ 1819 unsigned int compress_watermark; /* cache page watermark */ 1820 atomic_t compress_page_hit; /* cache hit count */ 1821 #endif 1822 1823 #ifdef CONFIG_F2FS_IOSTAT 1824 /* For app/fs IO statistics */ 1825 spinlock_t iostat_lock; 1826 unsigned long long rw_iostat[NR_IO_TYPE]; 1827 unsigned long long prev_rw_iostat[NR_IO_TYPE]; 1828 bool iostat_enable; 1829 unsigned long iostat_next_period; 1830 unsigned int iostat_period_ms; 1831 1832 /* For io latency related statistics info in one iostat period */ 1833 spinlock_t iostat_lat_lock; 1834 struct iostat_lat_info *iostat_io_lat; 1835 #endif 1836 }; 1837 1838 #ifdef CONFIG_F2FS_FAULT_INJECTION 1839 #define f2fs_show_injection_info(sbi, type) \ 1840 printk_ratelimited("%sF2FS-fs (%s) : inject %s in %s of %pS\n", \ 1841 KERN_INFO, sbi->sb->s_id, \ 1842 f2fs_fault_name[type], \ 1843 __func__, __builtin_return_address(0)) 1844 static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type) 1845 { 1846 struct f2fs_fault_info *ffi = &F2FS_OPTION(sbi).fault_info; 1847 1848 if (!ffi->inject_rate) 1849 return false; 1850 1851 if (!IS_FAULT_SET(ffi, type)) 1852 return false; 1853 1854 atomic_inc(&ffi->inject_ops); 1855 if (atomic_read(&ffi->inject_ops) >= ffi->inject_rate) { 1856 atomic_set(&ffi->inject_ops, 0); 1857 return true; 1858 } 1859 return false; 1860 } 1861 #else 1862 #define f2fs_show_injection_info(sbi, type) do { } while (0) 1863 static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type) 1864 { 1865 return false; 1866 } 1867 #endif 1868 1869 /* 1870 * Test if the mounted volume is a multi-device volume. 1871 * - For a single regular disk volume, sbi->s_ndevs is 0. 1872 * - For a single zoned disk volume, sbi->s_ndevs is 1. 1873 * - For a multi-device volume, sbi->s_ndevs is always 2 or more. 1874 */ 1875 static inline bool f2fs_is_multi_device(struct f2fs_sb_info *sbi) 1876 { 1877 return sbi->s_ndevs > 1; 1878 } 1879 1880 static inline void f2fs_update_time(struct f2fs_sb_info *sbi, int type) 1881 { 1882 unsigned long now = jiffies; 1883 1884 sbi->last_time[type] = now; 1885 1886 /* DISCARD_TIME and GC_TIME are based on REQ_TIME */ 1887 if (type == REQ_TIME) { 1888 sbi->last_time[DISCARD_TIME] = now; 1889 sbi->last_time[GC_TIME] = now; 1890 } 1891 } 1892 1893 static inline bool f2fs_time_over(struct f2fs_sb_info *sbi, int type) 1894 { 1895 unsigned long interval = sbi->interval_time[type] * HZ; 1896 1897 return time_after(jiffies, sbi->last_time[type] + interval); 1898 } 1899 1900 static inline unsigned int f2fs_time_to_wait(struct f2fs_sb_info *sbi, 1901 int type) 1902 { 1903 unsigned long interval = sbi->interval_time[type] * HZ; 1904 unsigned int wait_ms = 0; 1905 long delta; 1906 1907 delta = (sbi->last_time[type] + interval) - jiffies; 1908 if (delta > 0) 1909 wait_ms = jiffies_to_msecs(delta); 1910 1911 return wait_ms; 1912 } 1913 1914 /* 1915 * Inline functions 1916 */ 1917 static inline u32 __f2fs_crc32(struct f2fs_sb_info *sbi, u32 crc, 1918 const void *address, unsigned int length) 1919 { 1920 struct { 1921 struct shash_desc shash; 1922 char ctx[4]; 1923 } desc; 1924 int err; 1925 1926 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver) != sizeof(desc.ctx)); 1927 1928 desc.shash.tfm = sbi->s_chksum_driver; 1929 *(u32 *)desc.ctx = crc; 1930 1931 err = crypto_shash_update(&desc.shash, address, length); 1932 BUG_ON(err); 1933 1934 return *(u32 *)desc.ctx; 1935 } 1936 1937 static inline u32 f2fs_crc32(struct f2fs_sb_info *sbi, const void *address, 1938 unsigned int length) 1939 { 1940 return __f2fs_crc32(sbi, F2FS_SUPER_MAGIC, address, length); 1941 } 1942 1943 static inline bool f2fs_crc_valid(struct f2fs_sb_info *sbi, __u32 blk_crc, 1944 void *buf, size_t buf_size) 1945 { 1946 return f2fs_crc32(sbi, buf, buf_size) == blk_crc; 1947 } 1948 1949 static inline u32 f2fs_chksum(struct f2fs_sb_info *sbi, u32 crc, 1950 const void *address, unsigned int length) 1951 { 1952 return __f2fs_crc32(sbi, crc, address, length); 1953 } 1954 1955 static inline struct f2fs_inode_info *F2FS_I(struct inode *inode) 1956 { 1957 return container_of(inode, struct f2fs_inode_info, vfs_inode); 1958 } 1959 1960 static inline struct f2fs_sb_info *F2FS_SB(struct super_block *sb) 1961 { 1962 return sb->s_fs_info; 1963 } 1964 1965 static inline struct f2fs_sb_info *F2FS_I_SB(struct inode *inode) 1966 { 1967 return F2FS_SB(inode->i_sb); 1968 } 1969 1970 static inline struct f2fs_sb_info *F2FS_M_SB(struct address_space *mapping) 1971 { 1972 return F2FS_I_SB(mapping->host); 1973 } 1974 1975 static inline struct f2fs_sb_info *F2FS_P_SB(struct page *page) 1976 { 1977 return F2FS_M_SB(page_file_mapping(page)); 1978 } 1979 1980 static inline struct f2fs_super_block *F2FS_RAW_SUPER(struct f2fs_sb_info *sbi) 1981 { 1982 return (struct f2fs_super_block *)(sbi->raw_super); 1983 } 1984 1985 static inline struct f2fs_checkpoint *F2FS_CKPT(struct f2fs_sb_info *sbi) 1986 { 1987 return (struct f2fs_checkpoint *)(sbi->ckpt); 1988 } 1989 1990 static inline struct f2fs_node *F2FS_NODE(struct page *page) 1991 { 1992 return (struct f2fs_node *)page_address(page); 1993 } 1994 1995 static inline struct f2fs_inode *F2FS_INODE(struct page *page) 1996 { 1997 return &((struct f2fs_node *)page_address(page))->i; 1998 } 1999 2000 static inline struct f2fs_nm_info *NM_I(struct f2fs_sb_info *sbi) 2001 { 2002 return (struct f2fs_nm_info *)(sbi->nm_info); 2003 } 2004 2005 static inline struct f2fs_sm_info *SM_I(struct f2fs_sb_info *sbi) 2006 { 2007 return (struct f2fs_sm_info *)(sbi->sm_info); 2008 } 2009 2010 static inline struct sit_info *SIT_I(struct f2fs_sb_info *sbi) 2011 { 2012 return (struct sit_info *)(SM_I(sbi)->sit_info); 2013 } 2014 2015 static inline struct free_segmap_info *FREE_I(struct f2fs_sb_info *sbi) 2016 { 2017 return (struct free_segmap_info *)(SM_I(sbi)->free_info); 2018 } 2019 2020 static inline struct dirty_seglist_info *DIRTY_I(struct f2fs_sb_info *sbi) 2021 { 2022 return (struct dirty_seglist_info *)(SM_I(sbi)->dirty_info); 2023 } 2024 2025 static inline struct address_space *META_MAPPING(struct f2fs_sb_info *sbi) 2026 { 2027 return sbi->meta_inode->i_mapping; 2028 } 2029 2030 static inline struct address_space *NODE_MAPPING(struct f2fs_sb_info *sbi) 2031 { 2032 return sbi->node_inode->i_mapping; 2033 } 2034 2035 static inline bool is_sbi_flag_set(struct f2fs_sb_info *sbi, unsigned int type) 2036 { 2037 return test_bit(type, &sbi->s_flag); 2038 } 2039 2040 static inline void set_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type) 2041 { 2042 set_bit(type, &sbi->s_flag); 2043 } 2044 2045 static inline void clear_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type) 2046 { 2047 clear_bit(type, &sbi->s_flag); 2048 } 2049 2050 static inline unsigned long long cur_cp_version(struct f2fs_checkpoint *cp) 2051 { 2052 return le64_to_cpu(cp->checkpoint_ver); 2053 } 2054 2055 static inline unsigned long f2fs_qf_ino(struct super_block *sb, int type) 2056 { 2057 if (type < F2FS_MAX_QUOTAS) 2058 return le32_to_cpu(F2FS_SB(sb)->raw_super->qf_ino[type]); 2059 return 0; 2060 } 2061 2062 static inline __u64 cur_cp_crc(struct f2fs_checkpoint *cp) 2063 { 2064 size_t crc_offset = le32_to_cpu(cp->checksum_offset); 2065 return le32_to_cpu(*((__le32 *)((unsigned char *)cp + crc_offset))); 2066 } 2067 2068 static inline bool __is_set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 2069 { 2070 unsigned int ckpt_flags = le32_to_cpu(cp->ckpt_flags); 2071 2072 return ckpt_flags & f; 2073 } 2074 2075 static inline bool is_set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 2076 { 2077 return __is_set_ckpt_flags(F2FS_CKPT(sbi), f); 2078 } 2079 2080 static inline void __set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 2081 { 2082 unsigned int ckpt_flags; 2083 2084 ckpt_flags = le32_to_cpu(cp->ckpt_flags); 2085 ckpt_flags |= f; 2086 cp->ckpt_flags = cpu_to_le32(ckpt_flags); 2087 } 2088 2089 static inline void set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 2090 { 2091 unsigned long flags; 2092 2093 spin_lock_irqsave(&sbi->cp_lock, flags); 2094 __set_ckpt_flags(F2FS_CKPT(sbi), f); 2095 spin_unlock_irqrestore(&sbi->cp_lock, flags); 2096 } 2097 2098 static inline void __clear_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 2099 { 2100 unsigned int ckpt_flags; 2101 2102 ckpt_flags = le32_to_cpu(cp->ckpt_flags); 2103 ckpt_flags &= (~f); 2104 cp->ckpt_flags = cpu_to_le32(ckpt_flags); 2105 } 2106 2107 static inline void clear_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 2108 { 2109 unsigned long flags; 2110 2111 spin_lock_irqsave(&sbi->cp_lock, flags); 2112 __clear_ckpt_flags(F2FS_CKPT(sbi), f); 2113 spin_unlock_irqrestore(&sbi->cp_lock, flags); 2114 } 2115 2116 #define init_f2fs_rwsem(sem) \ 2117 do { \ 2118 static struct lock_class_key __key; \ 2119 \ 2120 __init_f2fs_rwsem((sem), #sem, &__key); \ 2121 } while (0) 2122 2123 static inline void __init_f2fs_rwsem(struct f2fs_rwsem *sem, 2124 const char *sem_name, struct lock_class_key *key) 2125 { 2126 __init_rwsem(&sem->internal_rwsem, sem_name, key); 2127 #ifdef CONFIG_F2FS_UNFAIR_RWSEM 2128 init_waitqueue_head(&sem->read_waiters); 2129 #endif 2130 } 2131 2132 static inline int f2fs_rwsem_is_locked(struct f2fs_rwsem *sem) 2133 { 2134 return rwsem_is_locked(&sem->internal_rwsem); 2135 } 2136 2137 static inline int f2fs_rwsem_is_contended(struct f2fs_rwsem *sem) 2138 { 2139 return rwsem_is_contended(&sem->internal_rwsem); 2140 } 2141 2142 static inline void f2fs_down_read(struct f2fs_rwsem *sem) 2143 { 2144 #ifdef CONFIG_F2FS_UNFAIR_RWSEM 2145 wait_event(sem->read_waiters, down_read_trylock(&sem->internal_rwsem)); 2146 #else 2147 down_read(&sem->internal_rwsem); 2148 #endif 2149 } 2150 2151 static inline int f2fs_down_read_trylock(struct f2fs_rwsem *sem) 2152 { 2153 return down_read_trylock(&sem->internal_rwsem); 2154 } 2155 2156 #ifdef CONFIG_DEBUG_LOCK_ALLOC 2157 static inline void f2fs_down_read_nested(struct f2fs_rwsem *sem, int subclass) 2158 { 2159 down_read_nested(&sem->internal_rwsem, subclass); 2160 } 2161 #else 2162 #define f2fs_down_read_nested(sem, subclass) f2fs_down_read(sem) 2163 #endif 2164 2165 static inline void f2fs_up_read(struct f2fs_rwsem *sem) 2166 { 2167 up_read(&sem->internal_rwsem); 2168 } 2169 2170 static inline void f2fs_down_write(struct f2fs_rwsem *sem) 2171 { 2172 down_write(&sem->internal_rwsem); 2173 } 2174 2175 static inline int f2fs_down_write_trylock(struct f2fs_rwsem *sem) 2176 { 2177 return down_write_trylock(&sem->internal_rwsem); 2178 } 2179 2180 static inline void f2fs_up_write(struct f2fs_rwsem *sem) 2181 { 2182 up_write(&sem->internal_rwsem); 2183 #ifdef CONFIG_F2FS_UNFAIR_RWSEM 2184 wake_up_all(&sem->read_waiters); 2185 #endif 2186 } 2187 2188 static inline void f2fs_lock_op(struct f2fs_sb_info *sbi) 2189 { 2190 f2fs_down_read(&sbi->cp_rwsem); 2191 } 2192 2193 static inline int f2fs_trylock_op(struct f2fs_sb_info *sbi) 2194 { 2195 if (time_to_inject(sbi, FAULT_LOCK_OP)) { 2196 f2fs_show_injection_info(sbi, FAULT_LOCK_OP); 2197 return 0; 2198 } 2199 return f2fs_down_read_trylock(&sbi->cp_rwsem); 2200 } 2201 2202 static inline void f2fs_unlock_op(struct f2fs_sb_info *sbi) 2203 { 2204 f2fs_up_read(&sbi->cp_rwsem); 2205 } 2206 2207 static inline void f2fs_lock_all(struct f2fs_sb_info *sbi) 2208 { 2209 f2fs_down_write(&sbi->cp_rwsem); 2210 } 2211 2212 static inline void f2fs_unlock_all(struct f2fs_sb_info *sbi) 2213 { 2214 f2fs_up_write(&sbi->cp_rwsem); 2215 } 2216 2217 static inline int __get_cp_reason(struct f2fs_sb_info *sbi) 2218 { 2219 int reason = CP_SYNC; 2220 2221 if (test_opt(sbi, FASTBOOT)) 2222 reason = CP_FASTBOOT; 2223 if (is_sbi_flag_set(sbi, SBI_IS_CLOSE)) 2224 reason = CP_UMOUNT; 2225 return reason; 2226 } 2227 2228 static inline bool __remain_node_summaries(int reason) 2229 { 2230 return (reason & (CP_UMOUNT | CP_FASTBOOT)); 2231 } 2232 2233 static inline bool __exist_node_summaries(struct f2fs_sb_info *sbi) 2234 { 2235 return (is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG) || 2236 is_set_ckpt_flags(sbi, CP_FASTBOOT_FLAG)); 2237 } 2238 2239 /* 2240 * Check whether the inode has blocks or not 2241 */ 2242 static inline int F2FS_HAS_BLOCKS(struct inode *inode) 2243 { 2244 block_t xattr_block = F2FS_I(inode)->i_xattr_nid ? 1 : 0; 2245 2246 return (inode->i_blocks >> F2FS_LOG_SECTORS_PER_BLOCK) > xattr_block; 2247 } 2248 2249 static inline bool f2fs_has_xattr_block(unsigned int ofs) 2250 { 2251 return ofs == XATTR_NODE_OFFSET; 2252 } 2253 2254 static inline bool __allow_reserved_blocks(struct f2fs_sb_info *sbi, 2255 struct inode *inode, bool cap) 2256 { 2257 if (!inode) 2258 return true; 2259 if (!test_opt(sbi, RESERVE_ROOT)) 2260 return false; 2261 if (IS_NOQUOTA(inode)) 2262 return true; 2263 if (uid_eq(F2FS_OPTION(sbi).s_resuid, current_fsuid())) 2264 return true; 2265 if (!gid_eq(F2FS_OPTION(sbi).s_resgid, GLOBAL_ROOT_GID) && 2266 in_group_p(F2FS_OPTION(sbi).s_resgid)) 2267 return true; 2268 if (cap && capable(CAP_SYS_RESOURCE)) 2269 return true; 2270 return false; 2271 } 2272 2273 static inline void f2fs_i_blocks_write(struct inode *, block_t, bool, bool); 2274 static inline int inc_valid_block_count(struct f2fs_sb_info *sbi, 2275 struct inode *inode, blkcnt_t *count) 2276 { 2277 blkcnt_t diff = 0, release = 0; 2278 block_t avail_user_block_count; 2279 int ret; 2280 2281 ret = dquot_reserve_block(inode, *count); 2282 if (ret) 2283 return ret; 2284 2285 if (time_to_inject(sbi, FAULT_BLOCK)) { 2286 f2fs_show_injection_info(sbi, FAULT_BLOCK); 2287 release = *count; 2288 goto release_quota; 2289 } 2290 2291 /* 2292 * let's increase this in prior to actual block count change in order 2293 * for f2fs_sync_file to avoid data races when deciding checkpoint. 2294 */ 2295 percpu_counter_add(&sbi->alloc_valid_block_count, (*count)); 2296 2297 spin_lock(&sbi->stat_lock); 2298 sbi->total_valid_block_count += (block_t)(*count); 2299 avail_user_block_count = sbi->user_block_count - 2300 sbi->current_reserved_blocks; 2301 2302 if (!__allow_reserved_blocks(sbi, inode, true)) 2303 avail_user_block_count -= F2FS_OPTION(sbi).root_reserved_blocks; 2304 2305 if (F2FS_IO_ALIGNED(sbi)) 2306 avail_user_block_count -= sbi->blocks_per_seg * 2307 SM_I(sbi)->additional_reserved_segments; 2308 2309 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) { 2310 if (avail_user_block_count > sbi->unusable_block_count) 2311 avail_user_block_count -= sbi->unusable_block_count; 2312 else 2313 avail_user_block_count = 0; 2314 } 2315 if (unlikely(sbi->total_valid_block_count > avail_user_block_count)) { 2316 diff = sbi->total_valid_block_count - avail_user_block_count; 2317 if (diff > *count) 2318 diff = *count; 2319 *count -= diff; 2320 release = diff; 2321 sbi->total_valid_block_count -= diff; 2322 if (!*count) { 2323 spin_unlock(&sbi->stat_lock); 2324 goto enospc; 2325 } 2326 } 2327 spin_unlock(&sbi->stat_lock); 2328 2329 if (unlikely(release)) { 2330 percpu_counter_sub(&sbi->alloc_valid_block_count, release); 2331 dquot_release_reservation_block(inode, release); 2332 } 2333 f2fs_i_blocks_write(inode, *count, true, true); 2334 return 0; 2335 2336 enospc: 2337 percpu_counter_sub(&sbi->alloc_valid_block_count, release); 2338 release_quota: 2339 dquot_release_reservation_block(inode, release); 2340 return -ENOSPC; 2341 } 2342 2343 __printf(2, 3) 2344 void f2fs_printk(struct f2fs_sb_info *sbi, const char *fmt, ...); 2345 2346 #define f2fs_err(sbi, fmt, ...) \ 2347 f2fs_printk(sbi, KERN_ERR fmt, ##__VA_ARGS__) 2348 #define f2fs_warn(sbi, fmt, ...) \ 2349 f2fs_printk(sbi, KERN_WARNING fmt, ##__VA_ARGS__) 2350 #define f2fs_notice(sbi, fmt, ...) \ 2351 f2fs_printk(sbi, KERN_NOTICE fmt, ##__VA_ARGS__) 2352 #define f2fs_info(sbi, fmt, ...) \ 2353 f2fs_printk(sbi, KERN_INFO fmt, ##__VA_ARGS__) 2354 #define f2fs_debug(sbi, fmt, ...) \ 2355 f2fs_printk(sbi, KERN_DEBUG fmt, ##__VA_ARGS__) 2356 2357 static inline void dec_valid_block_count(struct f2fs_sb_info *sbi, 2358 struct inode *inode, 2359 block_t count) 2360 { 2361 blkcnt_t sectors = count << F2FS_LOG_SECTORS_PER_BLOCK; 2362 2363 spin_lock(&sbi->stat_lock); 2364 f2fs_bug_on(sbi, sbi->total_valid_block_count < (block_t) count); 2365 sbi->total_valid_block_count -= (block_t)count; 2366 if (sbi->reserved_blocks && 2367 sbi->current_reserved_blocks < sbi->reserved_blocks) 2368 sbi->current_reserved_blocks = min(sbi->reserved_blocks, 2369 sbi->current_reserved_blocks + count); 2370 spin_unlock(&sbi->stat_lock); 2371 if (unlikely(inode->i_blocks < sectors)) { 2372 f2fs_warn(sbi, "Inconsistent i_blocks, ino:%lu, iblocks:%llu, sectors:%llu", 2373 inode->i_ino, 2374 (unsigned long long)inode->i_blocks, 2375 (unsigned long long)sectors); 2376 set_sbi_flag(sbi, SBI_NEED_FSCK); 2377 return; 2378 } 2379 f2fs_i_blocks_write(inode, count, false, true); 2380 } 2381 2382 static inline void inc_page_count(struct f2fs_sb_info *sbi, int count_type) 2383 { 2384 atomic_inc(&sbi->nr_pages[count_type]); 2385 2386 if (count_type == F2FS_DIRTY_DENTS || 2387 count_type == F2FS_DIRTY_NODES || 2388 count_type == F2FS_DIRTY_META || 2389 count_type == F2FS_DIRTY_QDATA || 2390 count_type == F2FS_DIRTY_IMETA) 2391 set_sbi_flag(sbi, SBI_IS_DIRTY); 2392 } 2393 2394 static inline void inode_inc_dirty_pages(struct inode *inode) 2395 { 2396 atomic_inc(&F2FS_I(inode)->dirty_pages); 2397 inc_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ? 2398 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA); 2399 if (IS_NOQUOTA(inode)) 2400 inc_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA); 2401 } 2402 2403 static inline void dec_page_count(struct f2fs_sb_info *sbi, int count_type) 2404 { 2405 atomic_dec(&sbi->nr_pages[count_type]); 2406 } 2407 2408 static inline void inode_dec_dirty_pages(struct inode *inode) 2409 { 2410 if (!S_ISDIR(inode->i_mode) && !S_ISREG(inode->i_mode) && 2411 !S_ISLNK(inode->i_mode)) 2412 return; 2413 2414 atomic_dec(&F2FS_I(inode)->dirty_pages); 2415 dec_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ? 2416 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA); 2417 if (IS_NOQUOTA(inode)) 2418 dec_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA); 2419 } 2420 2421 static inline s64 get_pages(struct f2fs_sb_info *sbi, int count_type) 2422 { 2423 return atomic_read(&sbi->nr_pages[count_type]); 2424 } 2425 2426 static inline int get_dirty_pages(struct inode *inode) 2427 { 2428 return atomic_read(&F2FS_I(inode)->dirty_pages); 2429 } 2430 2431 static inline int get_blocktype_secs(struct f2fs_sb_info *sbi, int block_type) 2432 { 2433 unsigned int pages_per_sec = sbi->segs_per_sec * sbi->blocks_per_seg; 2434 unsigned int segs = (get_pages(sbi, block_type) + pages_per_sec - 1) >> 2435 sbi->log_blocks_per_seg; 2436 2437 return segs / sbi->segs_per_sec; 2438 } 2439 2440 static inline block_t valid_user_blocks(struct f2fs_sb_info *sbi) 2441 { 2442 return sbi->total_valid_block_count; 2443 } 2444 2445 static inline block_t discard_blocks(struct f2fs_sb_info *sbi) 2446 { 2447 return sbi->discard_blks; 2448 } 2449 2450 static inline unsigned long __bitmap_size(struct f2fs_sb_info *sbi, int flag) 2451 { 2452 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi); 2453 2454 /* return NAT or SIT bitmap */ 2455 if (flag == NAT_BITMAP) 2456 return le32_to_cpu(ckpt->nat_ver_bitmap_bytesize); 2457 else if (flag == SIT_BITMAP) 2458 return le32_to_cpu(ckpt->sit_ver_bitmap_bytesize); 2459 2460 return 0; 2461 } 2462 2463 static inline block_t __cp_payload(struct f2fs_sb_info *sbi) 2464 { 2465 return le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_payload); 2466 } 2467 2468 static inline void *__bitmap_ptr(struct f2fs_sb_info *sbi, int flag) 2469 { 2470 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi); 2471 void *tmp_ptr = &ckpt->sit_nat_version_bitmap; 2472 int offset; 2473 2474 if (is_set_ckpt_flags(sbi, CP_LARGE_NAT_BITMAP_FLAG)) { 2475 offset = (flag == SIT_BITMAP) ? 2476 le32_to_cpu(ckpt->nat_ver_bitmap_bytesize) : 0; 2477 /* 2478 * if large_nat_bitmap feature is enabled, leave checksum 2479 * protection for all nat/sit bitmaps. 2480 */ 2481 return tmp_ptr + offset + sizeof(__le32); 2482 } 2483 2484 if (__cp_payload(sbi) > 0) { 2485 if (flag == NAT_BITMAP) 2486 return &ckpt->sit_nat_version_bitmap; 2487 else 2488 return (unsigned char *)ckpt + F2FS_BLKSIZE; 2489 } else { 2490 offset = (flag == NAT_BITMAP) ? 2491 le32_to_cpu(ckpt->sit_ver_bitmap_bytesize) : 0; 2492 return tmp_ptr + offset; 2493 } 2494 } 2495 2496 static inline block_t __start_cp_addr(struct f2fs_sb_info *sbi) 2497 { 2498 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr); 2499 2500 if (sbi->cur_cp_pack == 2) 2501 start_addr += sbi->blocks_per_seg; 2502 return start_addr; 2503 } 2504 2505 static inline block_t __start_cp_next_addr(struct f2fs_sb_info *sbi) 2506 { 2507 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr); 2508 2509 if (sbi->cur_cp_pack == 1) 2510 start_addr += sbi->blocks_per_seg; 2511 return start_addr; 2512 } 2513 2514 static inline void __set_cp_next_pack(struct f2fs_sb_info *sbi) 2515 { 2516 sbi->cur_cp_pack = (sbi->cur_cp_pack == 1) ? 2 : 1; 2517 } 2518 2519 static inline block_t __start_sum_addr(struct f2fs_sb_info *sbi) 2520 { 2521 return le32_to_cpu(F2FS_CKPT(sbi)->cp_pack_start_sum); 2522 } 2523 2524 static inline int inc_valid_node_count(struct f2fs_sb_info *sbi, 2525 struct inode *inode, bool is_inode) 2526 { 2527 block_t valid_block_count; 2528 unsigned int valid_node_count, user_block_count; 2529 int err; 2530 2531 if (is_inode) { 2532 if (inode) { 2533 err = dquot_alloc_inode(inode); 2534 if (err) 2535 return err; 2536 } 2537 } else { 2538 err = dquot_reserve_block(inode, 1); 2539 if (err) 2540 return err; 2541 } 2542 2543 if (time_to_inject(sbi, FAULT_BLOCK)) { 2544 f2fs_show_injection_info(sbi, FAULT_BLOCK); 2545 goto enospc; 2546 } 2547 2548 spin_lock(&sbi->stat_lock); 2549 2550 valid_block_count = sbi->total_valid_block_count + 2551 sbi->current_reserved_blocks + 1; 2552 2553 if (!__allow_reserved_blocks(sbi, inode, false)) 2554 valid_block_count += F2FS_OPTION(sbi).root_reserved_blocks; 2555 2556 if (F2FS_IO_ALIGNED(sbi)) 2557 valid_block_count += sbi->blocks_per_seg * 2558 SM_I(sbi)->additional_reserved_segments; 2559 2560 user_block_count = sbi->user_block_count; 2561 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) 2562 user_block_count -= sbi->unusable_block_count; 2563 2564 if (unlikely(valid_block_count > user_block_count)) { 2565 spin_unlock(&sbi->stat_lock); 2566 goto enospc; 2567 } 2568 2569 valid_node_count = sbi->total_valid_node_count + 1; 2570 if (unlikely(valid_node_count > sbi->total_node_count)) { 2571 spin_unlock(&sbi->stat_lock); 2572 goto enospc; 2573 } 2574 2575 sbi->total_valid_node_count++; 2576 sbi->total_valid_block_count++; 2577 spin_unlock(&sbi->stat_lock); 2578 2579 if (inode) { 2580 if (is_inode) 2581 f2fs_mark_inode_dirty_sync(inode, true); 2582 else 2583 f2fs_i_blocks_write(inode, 1, true, true); 2584 } 2585 2586 percpu_counter_inc(&sbi->alloc_valid_block_count); 2587 return 0; 2588 2589 enospc: 2590 if (is_inode) { 2591 if (inode) 2592 dquot_free_inode(inode); 2593 } else { 2594 dquot_release_reservation_block(inode, 1); 2595 } 2596 return -ENOSPC; 2597 } 2598 2599 static inline void dec_valid_node_count(struct f2fs_sb_info *sbi, 2600 struct inode *inode, bool is_inode) 2601 { 2602 spin_lock(&sbi->stat_lock); 2603 2604 if (unlikely(!sbi->total_valid_block_count || 2605 !sbi->total_valid_node_count)) { 2606 f2fs_warn(sbi, "dec_valid_node_count: inconsistent block counts, total_valid_block:%u, total_valid_node:%u", 2607 sbi->total_valid_block_count, 2608 sbi->total_valid_node_count); 2609 set_sbi_flag(sbi, SBI_NEED_FSCK); 2610 } else { 2611 sbi->total_valid_block_count--; 2612 sbi->total_valid_node_count--; 2613 } 2614 2615 if (sbi->reserved_blocks && 2616 sbi->current_reserved_blocks < sbi->reserved_blocks) 2617 sbi->current_reserved_blocks++; 2618 2619 spin_unlock(&sbi->stat_lock); 2620 2621 if (is_inode) { 2622 dquot_free_inode(inode); 2623 } else { 2624 if (unlikely(inode->i_blocks == 0)) { 2625 f2fs_warn(sbi, "dec_valid_node_count: inconsistent i_blocks, ino:%lu, iblocks:%llu", 2626 inode->i_ino, 2627 (unsigned long long)inode->i_blocks); 2628 set_sbi_flag(sbi, SBI_NEED_FSCK); 2629 return; 2630 } 2631 f2fs_i_blocks_write(inode, 1, false, true); 2632 } 2633 } 2634 2635 static inline unsigned int valid_node_count(struct f2fs_sb_info *sbi) 2636 { 2637 return sbi->total_valid_node_count; 2638 } 2639 2640 static inline void inc_valid_inode_count(struct f2fs_sb_info *sbi) 2641 { 2642 percpu_counter_inc(&sbi->total_valid_inode_count); 2643 } 2644 2645 static inline void dec_valid_inode_count(struct f2fs_sb_info *sbi) 2646 { 2647 percpu_counter_dec(&sbi->total_valid_inode_count); 2648 } 2649 2650 static inline s64 valid_inode_count(struct f2fs_sb_info *sbi) 2651 { 2652 return percpu_counter_sum_positive(&sbi->total_valid_inode_count); 2653 } 2654 2655 static inline struct page *f2fs_grab_cache_page(struct address_space *mapping, 2656 pgoff_t index, bool for_write) 2657 { 2658 struct page *page; 2659 unsigned int flags; 2660 2661 if (IS_ENABLED(CONFIG_F2FS_FAULT_INJECTION)) { 2662 if (!for_write) 2663 page = find_get_page_flags(mapping, index, 2664 FGP_LOCK | FGP_ACCESSED); 2665 else 2666 page = find_lock_page(mapping, index); 2667 if (page) 2668 return page; 2669 2670 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_ALLOC)) { 2671 f2fs_show_injection_info(F2FS_M_SB(mapping), 2672 FAULT_PAGE_ALLOC); 2673 return NULL; 2674 } 2675 } 2676 2677 if (!for_write) 2678 return grab_cache_page(mapping, index); 2679 2680 flags = memalloc_nofs_save(); 2681 page = grab_cache_page_write_begin(mapping, index); 2682 memalloc_nofs_restore(flags); 2683 2684 return page; 2685 } 2686 2687 static inline struct page *f2fs_pagecache_get_page( 2688 struct address_space *mapping, pgoff_t index, 2689 int fgp_flags, gfp_t gfp_mask) 2690 { 2691 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_GET)) { 2692 f2fs_show_injection_info(F2FS_M_SB(mapping), FAULT_PAGE_GET); 2693 return NULL; 2694 } 2695 2696 return pagecache_get_page(mapping, index, fgp_flags, gfp_mask); 2697 } 2698 2699 static inline void f2fs_copy_page(struct page *src, struct page *dst) 2700 { 2701 char *src_kaddr = kmap(src); 2702 char *dst_kaddr = kmap(dst); 2703 2704 memcpy(dst_kaddr, src_kaddr, PAGE_SIZE); 2705 kunmap(dst); 2706 kunmap(src); 2707 } 2708 2709 static inline void f2fs_put_page(struct page *page, int unlock) 2710 { 2711 if (!page) 2712 return; 2713 2714 if (unlock) { 2715 f2fs_bug_on(F2FS_P_SB(page), !PageLocked(page)); 2716 unlock_page(page); 2717 } 2718 put_page(page); 2719 } 2720 2721 static inline void f2fs_put_dnode(struct dnode_of_data *dn) 2722 { 2723 if (dn->node_page) 2724 f2fs_put_page(dn->node_page, 1); 2725 if (dn->inode_page && dn->node_page != dn->inode_page) 2726 f2fs_put_page(dn->inode_page, 0); 2727 dn->node_page = NULL; 2728 dn->inode_page = NULL; 2729 } 2730 2731 static inline struct kmem_cache *f2fs_kmem_cache_create(const char *name, 2732 size_t size) 2733 { 2734 return kmem_cache_create(name, size, 0, SLAB_RECLAIM_ACCOUNT, NULL); 2735 } 2736 2737 static inline void *f2fs_kmem_cache_alloc_nofail(struct kmem_cache *cachep, 2738 gfp_t flags) 2739 { 2740 void *entry; 2741 2742 entry = kmem_cache_alloc(cachep, flags); 2743 if (!entry) 2744 entry = kmem_cache_alloc(cachep, flags | __GFP_NOFAIL); 2745 return entry; 2746 } 2747 2748 static inline void *f2fs_kmem_cache_alloc(struct kmem_cache *cachep, 2749 gfp_t flags, bool nofail, struct f2fs_sb_info *sbi) 2750 { 2751 if (nofail) 2752 return f2fs_kmem_cache_alloc_nofail(cachep, flags); 2753 2754 if (time_to_inject(sbi, FAULT_SLAB_ALLOC)) { 2755 f2fs_show_injection_info(sbi, FAULT_SLAB_ALLOC); 2756 return NULL; 2757 } 2758 2759 return kmem_cache_alloc(cachep, flags); 2760 } 2761 2762 static inline bool is_inflight_io(struct f2fs_sb_info *sbi, int type) 2763 { 2764 if (get_pages(sbi, F2FS_RD_DATA) || get_pages(sbi, F2FS_RD_NODE) || 2765 get_pages(sbi, F2FS_RD_META) || get_pages(sbi, F2FS_WB_DATA) || 2766 get_pages(sbi, F2FS_WB_CP_DATA) || 2767 get_pages(sbi, F2FS_DIO_READ) || 2768 get_pages(sbi, F2FS_DIO_WRITE)) 2769 return true; 2770 2771 if (type != DISCARD_TIME && SM_I(sbi) && SM_I(sbi)->dcc_info && 2772 atomic_read(&SM_I(sbi)->dcc_info->queued_discard)) 2773 return true; 2774 2775 if (SM_I(sbi) && SM_I(sbi)->fcc_info && 2776 atomic_read(&SM_I(sbi)->fcc_info->queued_flush)) 2777 return true; 2778 return false; 2779 } 2780 2781 static inline bool is_idle(struct f2fs_sb_info *sbi, int type) 2782 { 2783 if (sbi->gc_mode == GC_URGENT_HIGH) 2784 return true; 2785 2786 if (is_inflight_io(sbi, type)) 2787 return false; 2788 2789 if (sbi->gc_mode == GC_URGENT_MID) 2790 return true; 2791 2792 if (sbi->gc_mode == GC_URGENT_LOW && 2793 (type == DISCARD_TIME || type == GC_TIME)) 2794 return true; 2795 2796 return f2fs_time_over(sbi, type); 2797 } 2798 2799 static inline void f2fs_radix_tree_insert(struct radix_tree_root *root, 2800 unsigned long index, void *item) 2801 { 2802 while (radix_tree_insert(root, index, item)) 2803 cond_resched(); 2804 } 2805 2806 #define RAW_IS_INODE(p) ((p)->footer.nid == (p)->footer.ino) 2807 2808 static inline bool IS_INODE(struct page *page) 2809 { 2810 struct f2fs_node *p = F2FS_NODE(page); 2811 2812 return RAW_IS_INODE(p); 2813 } 2814 2815 static inline int offset_in_addr(struct f2fs_inode *i) 2816 { 2817 return (i->i_inline & F2FS_EXTRA_ATTR) ? 2818 (le16_to_cpu(i->i_extra_isize) / sizeof(__le32)) : 0; 2819 } 2820 2821 static inline __le32 *blkaddr_in_node(struct f2fs_node *node) 2822 { 2823 return RAW_IS_INODE(node) ? node->i.i_addr : node->dn.addr; 2824 } 2825 2826 static inline int f2fs_has_extra_attr(struct inode *inode); 2827 static inline block_t data_blkaddr(struct inode *inode, 2828 struct page *node_page, unsigned int offset) 2829 { 2830 struct f2fs_node *raw_node; 2831 __le32 *addr_array; 2832 int base = 0; 2833 bool is_inode = IS_INODE(node_page); 2834 2835 raw_node = F2FS_NODE(node_page); 2836 2837 if (is_inode) { 2838 if (!inode) 2839 /* from GC path only */ 2840 base = offset_in_addr(&raw_node->i); 2841 else if (f2fs_has_extra_attr(inode)) 2842 base = get_extra_isize(inode); 2843 } 2844 2845 addr_array = blkaddr_in_node(raw_node); 2846 return le32_to_cpu(addr_array[base + offset]); 2847 } 2848 2849 static inline block_t f2fs_data_blkaddr(struct dnode_of_data *dn) 2850 { 2851 return data_blkaddr(dn->inode, dn->node_page, dn->ofs_in_node); 2852 } 2853 2854 static inline int f2fs_test_bit(unsigned int nr, char *addr) 2855 { 2856 int mask; 2857 2858 addr += (nr >> 3); 2859 mask = 1 << (7 - (nr & 0x07)); 2860 return mask & *addr; 2861 } 2862 2863 static inline void f2fs_set_bit(unsigned int nr, char *addr) 2864 { 2865 int mask; 2866 2867 addr += (nr >> 3); 2868 mask = 1 << (7 - (nr & 0x07)); 2869 *addr |= mask; 2870 } 2871 2872 static inline void f2fs_clear_bit(unsigned int nr, char *addr) 2873 { 2874 int mask; 2875 2876 addr += (nr >> 3); 2877 mask = 1 << (7 - (nr & 0x07)); 2878 *addr &= ~mask; 2879 } 2880 2881 static inline int f2fs_test_and_set_bit(unsigned int nr, char *addr) 2882 { 2883 int mask; 2884 int ret; 2885 2886 addr += (nr >> 3); 2887 mask = 1 << (7 - (nr & 0x07)); 2888 ret = mask & *addr; 2889 *addr |= mask; 2890 return ret; 2891 } 2892 2893 static inline int f2fs_test_and_clear_bit(unsigned int nr, char *addr) 2894 { 2895 int mask; 2896 int ret; 2897 2898 addr += (nr >> 3); 2899 mask = 1 << (7 - (nr & 0x07)); 2900 ret = mask & *addr; 2901 *addr &= ~mask; 2902 return ret; 2903 } 2904 2905 static inline void f2fs_change_bit(unsigned int nr, char *addr) 2906 { 2907 int mask; 2908 2909 addr += (nr >> 3); 2910 mask = 1 << (7 - (nr & 0x07)); 2911 *addr ^= mask; 2912 } 2913 2914 /* 2915 * On-disk inode flags (f2fs_inode::i_flags) 2916 */ 2917 #define F2FS_COMPR_FL 0x00000004 /* Compress file */ 2918 #define F2FS_SYNC_FL 0x00000008 /* Synchronous updates */ 2919 #define F2FS_IMMUTABLE_FL 0x00000010 /* Immutable file */ 2920 #define F2FS_APPEND_FL 0x00000020 /* writes to file may only append */ 2921 #define F2FS_NODUMP_FL 0x00000040 /* do not dump file */ 2922 #define F2FS_NOATIME_FL 0x00000080 /* do not update atime */ 2923 #define F2FS_NOCOMP_FL 0x00000400 /* Don't compress */ 2924 #define F2FS_INDEX_FL 0x00001000 /* hash-indexed directory */ 2925 #define F2FS_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */ 2926 #define F2FS_PROJINHERIT_FL 0x20000000 /* Create with parents projid */ 2927 #define F2FS_CASEFOLD_FL 0x40000000 /* Casefolded file */ 2928 2929 /* Flags that should be inherited by new inodes from their parent. */ 2930 #define F2FS_FL_INHERITED (F2FS_SYNC_FL | F2FS_NODUMP_FL | F2FS_NOATIME_FL | \ 2931 F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \ 2932 F2FS_CASEFOLD_FL | F2FS_COMPR_FL | F2FS_NOCOMP_FL) 2933 2934 /* Flags that are appropriate for regular files (all but dir-specific ones). */ 2935 #define F2FS_REG_FLMASK (~(F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \ 2936 F2FS_CASEFOLD_FL)) 2937 2938 /* Flags that are appropriate for non-directories/regular files. */ 2939 #define F2FS_OTHER_FLMASK (F2FS_NODUMP_FL | F2FS_NOATIME_FL) 2940 2941 static inline __u32 f2fs_mask_flags(umode_t mode, __u32 flags) 2942 { 2943 if (S_ISDIR(mode)) 2944 return flags; 2945 else if (S_ISREG(mode)) 2946 return flags & F2FS_REG_FLMASK; 2947 else 2948 return flags & F2FS_OTHER_FLMASK; 2949 } 2950 2951 static inline void __mark_inode_dirty_flag(struct inode *inode, 2952 int flag, bool set) 2953 { 2954 switch (flag) { 2955 case FI_INLINE_XATTR: 2956 case FI_INLINE_DATA: 2957 case FI_INLINE_DENTRY: 2958 case FI_NEW_INODE: 2959 if (set) 2960 return; 2961 fallthrough; 2962 case FI_DATA_EXIST: 2963 case FI_INLINE_DOTS: 2964 case FI_PIN_FILE: 2965 case FI_COMPRESS_RELEASED: 2966 f2fs_mark_inode_dirty_sync(inode, true); 2967 } 2968 } 2969 2970 static inline void set_inode_flag(struct inode *inode, int flag) 2971 { 2972 set_bit(flag, F2FS_I(inode)->flags); 2973 __mark_inode_dirty_flag(inode, flag, true); 2974 } 2975 2976 static inline int is_inode_flag_set(struct inode *inode, int flag) 2977 { 2978 return test_bit(flag, F2FS_I(inode)->flags); 2979 } 2980 2981 static inline void clear_inode_flag(struct inode *inode, int flag) 2982 { 2983 clear_bit(flag, F2FS_I(inode)->flags); 2984 __mark_inode_dirty_flag(inode, flag, false); 2985 } 2986 2987 static inline bool f2fs_verity_in_progress(struct inode *inode) 2988 { 2989 return IS_ENABLED(CONFIG_FS_VERITY) && 2990 is_inode_flag_set(inode, FI_VERITY_IN_PROGRESS); 2991 } 2992 2993 static inline void set_acl_inode(struct inode *inode, umode_t mode) 2994 { 2995 F2FS_I(inode)->i_acl_mode = mode; 2996 set_inode_flag(inode, FI_ACL_MODE); 2997 f2fs_mark_inode_dirty_sync(inode, false); 2998 } 2999 3000 static inline void f2fs_i_links_write(struct inode *inode, bool inc) 3001 { 3002 if (inc) 3003 inc_nlink(inode); 3004 else 3005 drop_nlink(inode); 3006 f2fs_mark_inode_dirty_sync(inode, true); 3007 } 3008 3009 static inline void f2fs_i_blocks_write(struct inode *inode, 3010 block_t diff, bool add, bool claim) 3011 { 3012 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE); 3013 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER); 3014 3015 /* add = 1, claim = 1 should be dquot_reserve_block in pair */ 3016 if (add) { 3017 if (claim) 3018 dquot_claim_block(inode, diff); 3019 else 3020 dquot_alloc_block_nofail(inode, diff); 3021 } else { 3022 dquot_free_block(inode, diff); 3023 } 3024 3025 f2fs_mark_inode_dirty_sync(inode, true); 3026 if (clean || recover) 3027 set_inode_flag(inode, FI_AUTO_RECOVER); 3028 } 3029 3030 static inline void f2fs_i_size_write(struct inode *inode, loff_t i_size) 3031 { 3032 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE); 3033 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER); 3034 3035 if (i_size_read(inode) == i_size) 3036 return; 3037 3038 i_size_write(inode, i_size); 3039 f2fs_mark_inode_dirty_sync(inode, true); 3040 if (clean || recover) 3041 set_inode_flag(inode, FI_AUTO_RECOVER); 3042 } 3043 3044 static inline void f2fs_i_depth_write(struct inode *inode, unsigned int depth) 3045 { 3046 F2FS_I(inode)->i_current_depth = depth; 3047 f2fs_mark_inode_dirty_sync(inode, true); 3048 } 3049 3050 static inline void f2fs_i_gc_failures_write(struct inode *inode, 3051 unsigned int count) 3052 { 3053 F2FS_I(inode)->i_gc_failures[GC_FAILURE_PIN] = count; 3054 f2fs_mark_inode_dirty_sync(inode, true); 3055 } 3056 3057 static inline void f2fs_i_xnid_write(struct inode *inode, nid_t xnid) 3058 { 3059 F2FS_I(inode)->i_xattr_nid = xnid; 3060 f2fs_mark_inode_dirty_sync(inode, true); 3061 } 3062 3063 static inline void f2fs_i_pino_write(struct inode *inode, nid_t pino) 3064 { 3065 F2FS_I(inode)->i_pino = pino; 3066 f2fs_mark_inode_dirty_sync(inode, true); 3067 } 3068 3069 static inline void get_inline_info(struct inode *inode, struct f2fs_inode *ri) 3070 { 3071 struct f2fs_inode_info *fi = F2FS_I(inode); 3072 3073 if (ri->i_inline & F2FS_INLINE_XATTR) 3074 set_bit(FI_INLINE_XATTR, fi->flags); 3075 if (ri->i_inline & F2FS_INLINE_DATA) 3076 set_bit(FI_INLINE_DATA, fi->flags); 3077 if (ri->i_inline & F2FS_INLINE_DENTRY) 3078 set_bit(FI_INLINE_DENTRY, fi->flags); 3079 if (ri->i_inline & F2FS_DATA_EXIST) 3080 set_bit(FI_DATA_EXIST, fi->flags); 3081 if (ri->i_inline & F2FS_INLINE_DOTS) 3082 set_bit(FI_INLINE_DOTS, fi->flags); 3083 if (ri->i_inline & F2FS_EXTRA_ATTR) 3084 set_bit(FI_EXTRA_ATTR, fi->flags); 3085 if (ri->i_inline & F2FS_PIN_FILE) 3086 set_bit(FI_PIN_FILE, fi->flags); 3087 if (ri->i_inline & F2FS_COMPRESS_RELEASED) 3088 set_bit(FI_COMPRESS_RELEASED, fi->flags); 3089 } 3090 3091 static inline void set_raw_inline(struct inode *inode, struct f2fs_inode *ri) 3092 { 3093 ri->i_inline = 0; 3094 3095 if (is_inode_flag_set(inode, FI_INLINE_XATTR)) 3096 ri->i_inline |= F2FS_INLINE_XATTR; 3097 if (is_inode_flag_set(inode, FI_INLINE_DATA)) 3098 ri->i_inline |= F2FS_INLINE_DATA; 3099 if (is_inode_flag_set(inode, FI_INLINE_DENTRY)) 3100 ri->i_inline |= F2FS_INLINE_DENTRY; 3101 if (is_inode_flag_set(inode, FI_DATA_EXIST)) 3102 ri->i_inline |= F2FS_DATA_EXIST; 3103 if (is_inode_flag_set(inode, FI_INLINE_DOTS)) 3104 ri->i_inline |= F2FS_INLINE_DOTS; 3105 if (is_inode_flag_set(inode, FI_EXTRA_ATTR)) 3106 ri->i_inline |= F2FS_EXTRA_ATTR; 3107 if (is_inode_flag_set(inode, FI_PIN_FILE)) 3108 ri->i_inline |= F2FS_PIN_FILE; 3109 if (is_inode_flag_set(inode, FI_COMPRESS_RELEASED)) 3110 ri->i_inline |= F2FS_COMPRESS_RELEASED; 3111 } 3112 3113 static inline int f2fs_has_extra_attr(struct inode *inode) 3114 { 3115 return is_inode_flag_set(inode, FI_EXTRA_ATTR); 3116 } 3117 3118 static inline int f2fs_has_inline_xattr(struct inode *inode) 3119 { 3120 return is_inode_flag_set(inode, FI_INLINE_XATTR); 3121 } 3122 3123 static inline int f2fs_compressed_file(struct inode *inode) 3124 { 3125 return S_ISREG(inode->i_mode) && 3126 is_inode_flag_set(inode, FI_COMPRESSED_FILE); 3127 } 3128 3129 static inline bool f2fs_need_compress_data(struct inode *inode) 3130 { 3131 int compress_mode = F2FS_OPTION(F2FS_I_SB(inode)).compress_mode; 3132 3133 if (!f2fs_compressed_file(inode)) 3134 return false; 3135 3136 if (compress_mode == COMPR_MODE_FS) 3137 return true; 3138 else if (compress_mode == COMPR_MODE_USER && 3139 is_inode_flag_set(inode, FI_ENABLE_COMPRESS)) 3140 return true; 3141 3142 return false; 3143 } 3144 3145 static inline unsigned int addrs_per_inode(struct inode *inode) 3146 { 3147 unsigned int addrs = CUR_ADDRS_PER_INODE(inode) - 3148 get_inline_xattr_addrs(inode); 3149 3150 if (!f2fs_compressed_file(inode)) 3151 return addrs; 3152 return ALIGN_DOWN(addrs, F2FS_I(inode)->i_cluster_size); 3153 } 3154 3155 static inline unsigned int addrs_per_block(struct inode *inode) 3156 { 3157 if (!f2fs_compressed_file(inode)) 3158 return DEF_ADDRS_PER_BLOCK; 3159 return ALIGN_DOWN(DEF_ADDRS_PER_BLOCK, F2FS_I(inode)->i_cluster_size); 3160 } 3161 3162 static inline void *inline_xattr_addr(struct inode *inode, struct page *page) 3163 { 3164 struct f2fs_inode *ri = F2FS_INODE(page); 3165 3166 return (void *)&(ri->i_addr[DEF_ADDRS_PER_INODE - 3167 get_inline_xattr_addrs(inode)]); 3168 } 3169 3170 static inline int inline_xattr_size(struct inode *inode) 3171 { 3172 if (f2fs_has_inline_xattr(inode)) 3173 return get_inline_xattr_addrs(inode) * sizeof(__le32); 3174 return 0; 3175 } 3176 3177 /* 3178 * Notice: check inline_data flag without inode page lock is unsafe. 3179 * It could change at any time by f2fs_convert_inline_page(). 3180 */ 3181 static inline int f2fs_has_inline_data(struct inode *inode) 3182 { 3183 return is_inode_flag_set(inode, FI_INLINE_DATA); 3184 } 3185 3186 static inline int f2fs_exist_data(struct inode *inode) 3187 { 3188 return is_inode_flag_set(inode, FI_DATA_EXIST); 3189 } 3190 3191 static inline int f2fs_has_inline_dots(struct inode *inode) 3192 { 3193 return is_inode_flag_set(inode, FI_INLINE_DOTS); 3194 } 3195 3196 static inline int f2fs_is_mmap_file(struct inode *inode) 3197 { 3198 return is_inode_flag_set(inode, FI_MMAP_FILE); 3199 } 3200 3201 static inline bool f2fs_is_pinned_file(struct inode *inode) 3202 { 3203 return is_inode_flag_set(inode, FI_PIN_FILE); 3204 } 3205 3206 static inline bool f2fs_is_atomic_file(struct inode *inode) 3207 { 3208 return is_inode_flag_set(inode, FI_ATOMIC_FILE); 3209 } 3210 3211 static inline bool f2fs_is_first_block_written(struct inode *inode) 3212 { 3213 return is_inode_flag_set(inode, FI_FIRST_BLOCK_WRITTEN); 3214 } 3215 3216 static inline bool f2fs_is_drop_cache(struct inode *inode) 3217 { 3218 return is_inode_flag_set(inode, FI_DROP_CACHE); 3219 } 3220 3221 static inline void *inline_data_addr(struct inode *inode, struct page *page) 3222 { 3223 struct f2fs_inode *ri = F2FS_INODE(page); 3224 int extra_size = get_extra_isize(inode); 3225 3226 return (void *)&(ri->i_addr[extra_size + DEF_INLINE_RESERVED_SIZE]); 3227 } 3228 3229 static inline int f2fs_has_inline_dentry(struct inode *inode) 3230 { 3231 return is_inode_flag_set(inode, FI_INLINE_DENTRY); 3232 } 3233 3234 static inline int is_file(struct inode *inode, int type) 3235 { 3236 return F2FS_I(inode)->i_advise & type; 3237 } 3238 3239 static inline void set_file(struct inode *inode, int type) 3240 { 3241 if (is_file(inode, type)) 3242 return; 3243 F2FS_I(inode)->i_advise |= type; 3244 f2fs_mark_inode_dirty_sync(inode, true); 3245 } 3246 3247 static inline void clear_file(struct inode *inode, int type) 3248 { 3249 if (!is_file(inode, type)) 3250 return; 3251 F2FS_I(inode)->i_advise &= ~type; 3252 f2fs_mark_inode_dirty_sync(inode, true); 3253 } 3254 3255 static inline bool f2fs_is_time_consistent(struct inode *inode) 3256 { 3257 if (!timespec64_equal(F2FS_I(inode)->i_disk_time, &inode->i_atime)) 3258 return false; 3259 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 1, &inode->i_ctime)) 3260 return false; 3261 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 2, &inode->i_mtime)) 3262 return false; 3263 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 3, 3264 &F2FS_I(inode)->i_crtime)) 3265 return false; 3266 return true; 3267 } 3268 3269 static inline bool f2fs_skip_inode_update(struct inode *inode, int dsync) 3270 { 3271 bool ret; 3272 3273 if (dsync) { 3274 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 3275 3276 spin_lock(&sbi->inode_lock[DIRTY_META]); 3277 ret = list_empty(&F2FS_I(inode)->gdirty_list); 3278 spin_unlock(&sbi->inode_lock[DIRTY_META]); 3279 return ret; 3280 } 3281 if (!is_inode_flag_set(inode, FI_AUTO_RECOVER) || 3282 file_keep_isize(inode) || 3283 i_size_read(inode) & ~PAGE_MASK) 3284 return false; 3285 3286 if (!f2fs_is_time_consistent(inode)) 3287 return false; 3288 3289 spin_lock(&F2FS_I(inode)->i_size_lock); 3290 ret = F2FS_I(inode)->last_disk_size == i_size_read(inode); 3291 spin_unlock(&F2FS_I(inode)->i_size_lock); 3292 3293 return ret; 3294 } 3295 3296 static inline bool f2fs_readonly(struct super_block *sb) 3297 { 3298 return sb_rdonly(sb); 3299 } 3300 3301 static inline bool f2fs_cp_error(struct f2fs_sb_info *sbi) 3302 { 3303 return is_set_ckpt_flags(sbi, CP_ERROR_FLAG); 3304 } 3305 3306 static inline bool is_dot_dotdot(const u8 *name, size_t len) 3307 { 3308 if (len == 1 && name[0] == '.') 3309 return true; 3310 3311 if (len == 2 && name[0] == '.' && name[1] == '.') 3312 return true; 3313 3314 return false; 3315 } 3316 3317 static inline void *f2fs_kmalloc(struct f2fs_sb_info *sbi, 3318 size_t size, gfp_t flags) 3319 { 3320 if (time_to_inject(sbi, FAULT_KMALLOC)) { 3321 f2fs_show_injection_info(sbi, FAULT_KMALLOC); 3322 return NULL; 3323 } 3324 3325 return kmalloc(size, flags); 3326 } 3327 3328 static inline void *f2fs_kzalloc(struct f2fs_sb_info *sbi, 3329 size_t size, gfp_t flags) 3330 { 3331 return f2fs_kmalloc(sbi, size, flags | __GFP_ZERO); 3332 } 3333 3334 static inline void *f2fs_kvmalloc(struct f2fs_sb_info *sbi, 3335 size_t size, gfp_t flags) 3336 { 3337 if (time_to_inject(sbi, FAULT_KVMALLOC)) { 3338 f2fs_show_injection_info(sbi, FAULT_KVMALLOC); 3339 return NULL; 3340 } 3341 3342 return kvmalloc(size, flags); 3343 } 3344 3345 static inline void *f2fs_kvzalloc(struct f2fs_sb_info *sbi, 3346 size_t size, gfp_t flags) 3347 { 3348 return f2fs_kvmalloc(sbi, size, flags | __GFP_ZERO); 3349 } 3350 3351 static inline int get_extra_isize(struct inode *inode) 3352 { 3353 return F2FS_I(inode)->i_extra_isize / sizeof(__le32); 3354 } 3355 3356 static inline int get_inline_xattr_addrs(struct inode *inode) 3357 { 3358 return F2FS_I(inode)->i_inline_xattr_size; 3359 } 3360 3361 #define f2fs_get_inode_mode(i) \ 3362 ((is_inode_flag_set(i, FI_ACL_MODE)) ? \ 3363 (F2FS_I(i)->i_acl_mode) : ((i)->i_mode)) 3364 3365 #define F2FS_TOTAL_EXTRA_ATTR_SIZE \ 3366 (offsetof(struct f2fs_inode, i_extra_end) - \ 3367 offsetof(struct f2fs_inode, i_extra_isize)) \ 3368 3369 #define F2FS_OLD_ATTRIBUTE_SIZE (offsetof(struct f2fs_inode, i_addr)) 3370 #define F2FS_FITS_IN_INODE(f2fs_inode, extra_isize, field) \ 3371 ((offsetof(typeof(*(f2fs_inode)), field) + \ 3372 sizeof((f2fs_inode)->field)) \ 3373 <= (F2FS_OLD_ATTRIBUTE_SIZE + (extra_isize))) \ 3374 3375 #define __is_large_section(sbi) ((sbi)->segs_per_sec > 1) 3376 3377 #define __is_meta_io(fio) (PAGE_TYPE_OF_BIO((fio)->type) == META) 3378 3379 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi, 3380 block_t blkaddr, int type); 3381 static inline void verify_blkaddr(struct f2fs_sb_info *sbi, 3382 block_t blkaddr, int type) 3383 { 3384 if (!f2fs_is_valid_blkaddr(sbi, blkaddr, type)) { 3385 f2fs_err(sbi, "invalid blkaddr: %u, type: %d, run fsck to fix.", 3386 blkaddr, type); 3387 f2fs_bug_on(sbi, 1); 3388 } 3389 } 3390 3391 static inline bool __is_valid_data_blkaddr(block_t blkaddr) 3392 { 3393 if (blkaddr == NEW_ADDR || blkaddr == NULL_ADDR || 3394 blkaddr == COMPRESS_ADDR) 3395 return false; 3396 return true; 3397 } 3398 3399 /* 3400 * file.c 3401 */ 3402 int f2fs_sync_file(struct file *file, loff_t start, loff_t end, int datasync); 3403 void f2fs_truncate_data_blocks(struct dnode_of_data *dn); 3404 int f2fs_do_truncate_blocks(struct inode *inode, u64 from, bool lock); 3405 int f2fs_truncate_blocks(struct inode *inode, u64 from, bool lock); 3406 int f2fs_truncate(struct inode *inode); 3407 int f2fs_getattr(struct user_namespace *mnt_userns, const struct path *path, 3408 struct kstat *stat, u32 request_mask, unsigned int flags); 3409 int f2fs_setattr(struct user_namespace *mnt_userns, struct dentry *dentry, 3410 struct iattr *attr); 3411 int f2fs_truncate_hole(struct inode *inode, pgoff_t pg_start, pgoff_t pg_end); 3412 void f2fs_truncate_data_blocks_range(struct dnode_of_data *dn, int count); 3413 int f2fs_precache_extents(struct inode *inode); 3414 int f2fs_fileattr_get(struct dentry *dentry, struct fileattr *fa); 3415 int f2fs_fileattr_set(struct user_namespace *mnt_userns, 3416 struct dentry *dentry, struct fileattr *fa); 3417 long f2fs_ioctl(struct file *filp, unsigned int cmd, unsigned long arg); 3418 long f2fs_compat_ioctl(struct file *file, unsigned int cmd, unsigned long arg); 3419 int f2fs_transfer_project_quota(struct inode *inode, kprojid_t kprojid); 3420 int f2fs_pin_file_control(struct inode *inode, bool inc); 3421 3422 /* 3423 * inode.c 3424 */ 3425 void f2fs_set_inode_flags(struct inode *inode); 3426 bool f2fs_inode_chksum_verify(struct f2fs_sb_info *sbi, struct page *page); 3427 void f2fs_inode_chksum_set(struct f2fs_sb_info *sbi, struct page *page); 3428 struct inode *f2fs_iget(struct super_block *sb, unsigned long ino); 3429 struct inode *f2fs_iget_retry(struct super_block *sb, unsigned long ino); 3430 int f2fs_try_to_free_nats(struct f2fs_sb_info *sbi, int nr_shrink); 3431 void f2fs_update_inode(struct inode *inode, struct page *node_page); 3432 void f2fs_update_inode_page(struct inode *inode); 3433 int f2fs_write_inode(struct inode *inode, struct writeback_control *wbc); 3434 void f2fs_evict_inode(struct inode *inode); 3435 void f2fs_handle_failed_inode(struct inode *inode); 3436 3437 /* 3438 * namei.c 3439 */ 3440 int f2fs_update_extension_list(struct f2fs_sb_info *sbi, const char *name, 3441 bool hot, bool set); 3442 struct dentry *f2fs_get_parent(struct dentry *child); 3443 int f2fs_get_tmpfile(struct user_namespace *mnt_userns, struct inode *dir, 3444 struct inode **new_inode); 3445 3446 /* 3447 * dir.c 3448 */ 3449 unsigned char f2fs_get_de_type(struct f2fs_dir_entry *de); 3450 int f2fs_init_casefolded_name(const struct inode *dir, 3451 struct f2fs_filename *fname); 3452 int f2fs_setup_filename(struct inode *dir, const struct qstr *iname, 3453 int lookup, struct f2fs_filename *fname); 3454 int f2fs_prepare_lookup(struct inode *dir, struct dentry *dentry, 3455 struct f2fs_filename *fname); 3456 void f2fs_free_filename(struct f2fs_filename *fname); 3457 struct f2fs_dir_entry *f2fs_find_target_dentry(const struct f2fs_dentry_ptr *d, 3458 const struct f2fs_filename *fname, int *max_slots); 3459 int f2fs_fill_dentries(struct dir_context *ctx, struct f2fs_dentry_ptr *d, 3460 unsigned int start_pos, struct fscrypt_str *fstr); 3461 void f2fs_do_make_empty_dir(struct inode *inode, struct inode *parent, 3462 struct f2fs_dentry_ptr *d); 3463 struct page *f2fs_init_inode_metadata(struct inode *inode, struct inode *dir, 3464 const struct f2fs_filename *fname, struct page *dpage); 3465 void f2fs_update_parent_metadata(struct inode *dir, struct inode *inode, 3466 unsigned int current_depth); 3467 int f2fs_room_for_filename(const void *bitmap, int slots, int max_slots); 3468 void f2fs_drop_nlink(struct inode *dir, struct inode *inode); 3469 struct f2fs_dir_entry *__f2fs_find_entry(struct inode *dir, 3470 const struct f2fs_filename *fname, 3471 struct page **res_page); 3472 struct f2fs_dir_entry *f2fs_find_entry(struct inode *dir, 3473 const struct qstr *child, struct page **res_page); 3474 struct f2fs_dir_entry *f2fs_parent_dir(struct inode *dir, struct page **p); 3475 ino_t f2fs_inode_by_name(struct inode *dir, const struct qstr *qstr, 3476 struct page **page); 3477 void f2fs_set_link(struct inode *dir, struct f2fs_dir_entry *de, 3478 struct page *page, struct inode *inode); 3479 bool f2fs_has_enough_room(struct inode *dir, struct page *ipage, 3480 const struct f2fs_filename *fname); 3481 void f2fs_update_dentry(nid_t ino, umode_t mode, struct f2fs_dentry_ptr *d, 3482 const struct fscrypt_str *name, f2fs_hash_t name_hash, 3483 unsigned int bit_pos); 3484 int f2fs_add_regular_entry(struct inode *dir, const struct f2fs_filename *fname, 3485 struct inode *inode, nid_t ino, umode_t mode); 3486 int f2fs_add_dentry(struct inode *dir, const struct f2fs_filename *fname, 3487 struct inode *inode, nid_t ino, umode_t mode); 3488 int f2fs_do_add_link(struct inode *dir, const struct qstr *name, 3489 struct inode *inode, nid_t ino, umode_t mode); 3490 void f2fs_delete_entry(struct f2fs_dir_entry *dentry, struct page *page, 3491 struct inode *dir, struct inode *inode); 3492 int f2fs_do_tmpfile(struct inode *inode, struct inode *dir); 3493 bool f2fs_empty_dir(struct inode *dir); 3494 3495 static inline int f2fs_add_link(struct dentry *dentry, struct inode *inode) 3496 { 3497 if (fscrypt_is_nokey_name(dentry)) 3498 return -ENOKEY; 3499 return f2fs_do_add_link(d_inode(dentry->d_parent), &dentry->d_name, 3500 inode, inode->i_ino, inode->i_mode); 3501 } 3502 3503 /* 3504 * super.c 3505 */ 3506 int f2fs_inode_dirtied(struct inode *inode, bool sync); 3507 void f2fs_inode_synced(struct inode *inode); 3508 int f2fs_dquot_initialize(struct inode *inode); 3509 int f2fs_enable_quota_files(struct f2fs_sb_info *sbi, bool rdonly); 3510 int f2fs_quota_sync(struct super_block *sb, int type); 3511 loff_t max_file_blocks(struct inode *inode); 3512 void f2fs_quota_off_umount(struct super_block *sb); 3513 int f2fs_commit_super(struct f2fs_sb_info *sbi, bool recover); 3514 int f2fs_sync_fs(struct super_block *sb, int sync); 3515 int f2fs_sanity_check_ckpt(struct f2fs_sb_info *sbi); 3516 3517 /* 3518 * hash.c 3519 */ 3520 void f2fs_hash_filename(const struct inode *dir, struct f2fs_filename *fname); 3521 3522 /* 3523 * node.c 3524 */ 3525 struct node_info; 3526 3527 int f2fs_check_nid_range(struct f2fs_sb_info *sbi, nid_t nid); 3528 bool f2fs_available_free_memory(struct f2fs_sb_info *sbi, int type); 3529 bool f2fs_in_warm_node_list(struct f2fs_sb_info *sbi, struct page *page); 3530 void f2fs_init_fsync_node_info(struct f2fs_sb_info *sbi); 3531 void f2fs_del_fsync_node_entry(struct f2fs_sb_info *sbi, struct page *page); 3532 void f2fs_reset_fsync_node_info(struct f2fs_sb_info *sbi); 3533 int f2fs_need_dentry_mark(struct f2fs_sb_info *sbi, nid_t nid); 3534 bool f2fs_is_checkpointed_node(struct f2fs_sb_info *sbi, nid_t nid); 3535 bool f2fs_need_inode_block_update(struct f2fs_sb_info *sbi, nid_t ino); 3536 int f2fs_get_node_info(struct f2fs_sb_info *sbi, nid_t nid, 3537 struct node_info *ni, bool checkpoint_context); 3538 pgoff_t f2fs_get_next_page_offset(struct dnode_of_data *dn, pgoff_t pgofs); 3539 int f2fs_get_dnode_of_data(struct dnode_of_data *dn, pgoff_t index, int mode); 3540 int f2fs_truncate_inode_blocks(struct inode *inode, pgoff_t from); 3541 int f2fs_truncate_xattr_node(struct inode *inode); 3542 int f2fs_wait_on_node_pages_writeback(struct f2fs_sb_info *sbi, 3543 unsigned int seq_id); 3544 bool f2fs_nat_bitmap_enabled(struct f2fs_sb_info *sbi); 3545 int f2fs_remove_inode_page(struct inode *inode); 3546 struct page *f2fs_new_inode_page(struct inode *inode); 3547 struct page *f2fs_new_node_page(struct dnode_of_data *dn, unsigned int ofs); 3548 void f2fs_ra_node_page(struct f2fs_sb_info *sbi, nid_t nid); 3549 struct page *f2fs_get_node_page(struct f2fs_sb_info *sbi, pgoff_t nid); 3550 struct page *f2fs_get_node_page_ra(struct page *parent, int start); 3551 int f2fs_move_node_page(struct page *node_page, int gc_type); 3552 void f2fs_flush_inline_data(struct f2fs_sb_info *sbi); 3553 int f2fs_fsync_node_pages(struct f2fs_sb_info *sbi, struct inode *inode, 3554 struct writeback_control *wbc, bool atomic, 3555 unsigned int *seq_id); 3556 int f2fs_sync_node_pages(struct f2fs_sb_info *sbi, 3557 struct writeback_control *wbc, 3558 bool do_balance, enum iostat_type io_type); 3559 int f2fs_build_free_nids(struct f2fs_sb_info *sbi, bool sync, bool mount); 3560 bool f2fs_alloc_nid(struct f2fs_sb_info *sbi, nid_t *nid); 3561 void f2fs_alloc_nid_done(struct f2fs_sb_info *sbi, nid_t nid); 3562 void f2fs_alloc_nid_failed(struct f2fs_sb_info *sbi, nid_t nid); 3563 int f2fs_try_to_free_nids(struct f2fs_sb_info *sbi, int nr_shrink); 3564 int f2fs_recover_inline_xattr(struct inode *inode, struct page *page); 3565 int f2fs_recover_xattr_data(struct inode *inode, struct page *page); 3566 int f2fs_recover_inode_page(struct f2fs_sb_info *sbi, struct page *page); 3567 int f2fs_restore_node_summary(struct f2fs_sb_info *sbi, 3568 unsigned int segno, struct f2fs_summary_block *sum); 3569 void f2fs_enable_nat_bits(struct f2fs_sb_info *sbi); 3570 int f2fs_flush_nat_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3571 int f2fs_build_node_manager(struct f2fs_sb_info *sbi); 3572 void f2fs_destroy_node_manager(struct f2fs_sb_info *sbi); 3573 int __init f2fs_create_node_manager_caches(void); 3574 void f2fs_destroy_node_manager_caches(void); 3575 3576 /* 3577 * segment.c 3578 */ 3579 bool f2fs_need_SSR(struct f2fs_sb_info *sbi); 3580 int f2fs_commit_atomic_write(struct inode *inode); 3581 void f2fs_abort_atomic_write(struct inode *inode, bool clean); 3582 void f2fs_balance_fs(struct f2fs_sb_info *sbi, bool need); 3583 void f2fs_balance_fs_bg(struct f2fs_sb_info *sbi, bool from_bg); 3584 int f2fs_issue_flush(struct f2fs_sb_info *sbi, nid_t ino); 3585 int f2fs_create_flush_cmd_control(struct f2fs_sb_info *sbi); 3586 int f2fs_flush_device_cache(struct f2fs_sb_info *sbi); 3587 void f2fs_destroy_flush_cmd_control(struct f2fs_sb_info *sbi, bool free); 3588 void f2fs_invalidate_blocks(struct f2fs_sb_info *sbi, block_t addr); 3589 bool f2fs_is_checkpointed_data(struct f2fs_sb_info *sbi, block_t blkaddr); 3590 int f2fs_start_discard_thread(struct f2fs_sb_info *sbi); 3591 void f2fs_drop_discard_cmd(struct f2fs_sb_info *sbi); 3592 void f2fs_stop_discard_thread(struct f2fs_sb_info *sbi); 3593 bool f2fs_issue_discard_timeout(struct f2fs_sb_info *sbi); 3594 void f2fs_clear_prefree_segments(struct f2fs_sb_info *sbi, 3595 struct cp_control *cpc); 3596 void f2fs_dirty_to_prefree(struct f2fs_sb_info *sbi); 3597 block_t f2fs_get_unusable_blocks(struct f2fs_sb_info *sbi); 3598 int f2fs_disable_cp_again(struct f2fs_sb_info *sbi, block_t unusable); 3599 void f2fs_release_discard_addrs(struct f2fs_sb_info *sbi); 3600 int f2fs_npages_for_summary_flush(struct f2fs_sb_info *sbi, bool for_ra); 3601 bool f2fs_segment_has_free_slot(struct f2fs_sb_info *sbi, int segno); 3602 void f2fs_init_inmem_curseg(struct f2fs_sb_info *sbi); 3603 void f2fs_save_inmem_curseg(struct f2fs_sb_info *sbi); 3604 void f2fs_restore_inmem_curseg(struct f2fs_sb_info *sbi); 3605 void f2fs_get_new_segment(struct f2fs_sb_info *sbi, 3606 unsigned int *newseg, bool new_sec, int dir); 3607 void f2fs_allocate_segment_for_resize(struct f2fs_sb_info *sbi, int type, 3608 unsigned int start, unsigned int end); 3609 void f2fs_allocate_new_section(struct f2fs_sb_info *sbi, int type, bool force); 3610 void f2fs_allocate_new_segments(struct f2fs_sb_info *sbi); 3611 int f2fs_trim_fs(struct f2fs_sb_info *sbi, struct fstrim_range *range); 3612 bool f2fs_exist_trim_candidates(struct f2fs_sb_info *sbi, 3613 struct cp_control *cpc); 3614 struct page *f2fs_get_sum_page(struct f2fs_sb_info *sbi, unsigned int segno); 3615 void f2fs_update_meta_page(struct f2fs_sb_info *sbi, void *src, 3616 block_t blk_addr); 3617 void f2fs_do_write_meta_page(struct f2fs_sb_info *sbi, struct page *page, 3618 enum iostat_type io_type); 3619 void f2fs_do_write_node_page(unsigned int nid, struct f2fs_io_info *fio); 3620 void f2fs_outplace_write_data(struct dnode_of_data *dn, 3621 struct f2fs_io_info *fio); 3622 int f2fs_inplace_write_data(struct f2fs_io_info *fio); 3623 void f2fs_do_replace_block(struct f2fs_sb_info *sbi, struct f2fs_summary *sum, 3624 block_t old_blkaddr, block_t new_blkaddr, 3625 bool recover_curseg, bool recover_newaddr, 3626 bool from_gc); 3627 void f2fs_replace_block(struct f2fs_sb_info *sbi, struct dnode_of_data *dn, 3628 block_t old_addr, block_t new_addr, 3629 unsigned char version, bool recover_curseg, 3630 bool recover_newaddr); 3631 void f2fs_allocate_data_block(struct f2fs_sb_info *sbi, struct page *page, 3632 block_t old_blkaddr, block_t *new_blkaddr, 3633 struct f2fs_summary *sum, int type, 3634 struct f2fs_io_info *fio); 3635 void f2fs_update_device_state(struct f2fs_sb_info *sbi, nid_t ino, 3636 block_t blkaddr, unsigned int blkcnt); 3637 void f2fs_wait_on_page_writeback(struct page *page, 3638 enum page_type type, bool ordered, bool locked); 3639 void f2fs_wait_on_block_writeback(struct inode *inode, block_t blkaddr); 3640 void f2fs_wait_on_block_writeback_range(struct inode *inode, block_t blkaddr, 3641 block_t len); 3642 void f2fs_write_data_summaries(struct f2fs_sb_info *sbi, block_t start_blk); 3643 void f2fs_write_node_summaries(struct f2fs_sb_info *sbi, block_t start_blk); 3644 int f2fs_lookup_journal_in_cursum(struct f2fs_journal *journal, int type, 3645 unsigned int val, int alloc); 3646 void f2fs_flush_sit_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3647 int f2fs_fix_curseg_write_pointer(struct f2fs_sb_info *sbi); 3648 int f2fs_check_write_pointer(struct f2fs_sb_info *sbi); 3649 int f2fs_build_segment_manager(struct f2fs_sb_info *sbi); 3650 void f2fs_destroy_segment_manager(struct f2fs_sb_info *sbi); 3651 int __init f2fs_create_segment_manager_caches(void); 3652 void f2fs_destroy_segment_manager_caches(void); 3653 int f2fs_rw_hint_to_seg_type(enum rw_hint hint); 3654 unsigned int f2fs_usable_segs_in_sec(struct f2fs_sb_info *sbi, 3655 unsigned int segno); 3656 unsigned int f2fs_usable_blks_in_seg(struct f2fs_sb_info *sbi, 3657 unsigned int segno); 3658 3659 #define DEF_FRAGMENT_SIZE 4 3660 #define MIN_FRAGMENT_SIZE 1 3661 #define MAX_FRAGMENT_SIZE 512 3662 3663 static inline bool f2fs_need_rand_seg(struct f2fs_sb_info *sbi) 3664 { 3665 return F2FS_OPTION(sbi).fs_mode == FS_MODE_FRAGMENT_SEG || 3666 F2FS_OPTION(sbi).fs_mode == FS_MODE_FRAGMENT_BLK; 3667 } 3668 3669 /* 3670 * checkpoint.c 3671 */ 3672 void f2fs_stop_checkpoint(struct f2fs_sb_info *sbi, bool end_io); 3673 struct page *f2fs_grab_meta_page(struct f2fs_sb_info *sbi, pgoff_t index); 3674 struct page *f2fs_get_meta_page(struct f2fs_sb_info *sbi, pgoff_t index); 3675 struct page *f2fs_get_meta_page_retry(struct f2fs_sb_info *sbi, pgoff_t index); 3676 struct page *f2fs_get_tmp_page(struct f2fs_sb_info *sbi, pgoff_t index); 3677 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi, 3678 block_t blkaddr, int type); 3679 int f2fs_ra_meta_pages(struct f2fs_sb_info *sbi, block_t start, int nrpages, 3680 int type, bool sync); 3681 void f2fs_ra_meta_pages_cond(struct f2fs_sb_info *sbi, pgoff_t index, 3682 unsigned int ra_blocks); 3683 long f2fs_sync_meta_pages(struct f2fs_sb_info *sbi, enum page_type type, 3684 long nr_to_write, enum iostat_type io_type); 3685 void f2fs_add_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type); 3686 void f2fs_remove_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type); 3687 void f2fs_release_ino_entry(struct f2fs_sb_info *sbi, bool all); 3688 bool f2fs_exist_written_data(struct f2fs_sb_info *sbi, nid_t ino, int mode); 3689 void f2fs_set_dirty_device(struct f2fs_sb_info *sbi, nid_t ino, 3690 unsigned int devidx, int type); 3691 bool f2fs_is_dirty_device(struct f2fs_sb_info *sbi, nid_t ino, 3692 unsigned int devidx, int type); 3693 int f2fs_sync_inode_meta(struct f2fs_sb_info *sbi); 3694 int f2fs_acquire_orphan_inode(struct f2fs_sb_info *sbi); 3695 void f2fs_release_orphan_inode(struct f2fs_sb_info *sbi); 3696 void f2fs_add_orphan_inode(struct inode *inode); 3697 void f2fs_remove_orphan_inode(struct f2fs_sb_info *sbi, nid_t ino); 3698 int f2fs_recover_orphan_inodes(struct f2fs_sb_info *sbi); 3699 int f2fs_get_valid_checkpoint(struct f2fs_sb_info *sbi); 3700 void f2fs_update_dirty_folio(struct inode *inode, struct folio *folio); 3701 void f2fs_remove_dirty_inode(struct inode *inode); 3702 int f2fs_sync_dirty_inodes(struct f2fs_sb_info *sbi, enum inode_type type); 3703 void f2fs_wait_on_all_pages(struct f2fs_sb_info *sbi, int type); 3704 u64 f2fs_get_sectors_written(struct f2fs_sb_info *sbi); 3705 int f2fs_write_checkpoint(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3706 void f2fs_init_ino_entry_info(struct f2fs_sb_info *sbi); 3707 int __init f2fs_create_checkpoint_caches(void); 3708 void f2fs_destroy_checkpoint_caches(void); 3709 int f2fs_issue_checkpoint(struct f2fs_sb_info *sbi); 3710 int f2fs_start_ckpt_thread(struct f2fs_sb_info *sbi); 3711 void f2fs_stop_ckpt_thread(struct f2fs_sb_info *sbi); 3712 void f2fs_init_ckpt_req_control(struct f2fs_sb_info *sbi); 3713 3714 /* 3715 * data.c 3716 */ 3717 int __init f2fs_init_bioset(void); 3718 void f2fs_destroy_bioset(void); 3719 int f2fs_init_bio_entry_cache(void); 3720 void f2fs_destroy_bio_entry_cache(void); 3721 void f2fs_submit_bio(struct f2fs_sb_info *sbi, 3722 struct bio *bio, enum page_type type); 3723 int f2fs_init_write_merge_io(struct f2fs_sb_info *sbi); 3724 void f2fs_submit_merged_write(struct f2fs_sb_info *sbi, enum page_type type); 3725 void f2fs_submit_merged_write_cond(struct f2fs_sb_info *sbi, 3726 struct inode *inode, struct page *page, 3727 nid_t ino, enum page_type type); 3728 void f2fs_submit_merged_ipu_write(struct f2fs_sb_info *sbi, 3729 struct bio **bio, struct page *page); 3730 void f2fs_flush_merged_writes(struct f2fs_sb_info *sbi); 3731 int f2fs_submit_page_bio(struct f2fs_io_info *fio); 3732 int f2fs_merge_page_bio(struct f2fs_io_info *fio); 3733 void f2fs_submit_page_write(struct f2fs_io_info *fio); 3734 struct block_device *f2fs_target_device(struct f2fs_sb_info *sbi, 3735 block_t blk_addr, sector_t *sector); 3736 int f2fs_target_device_index(struct f2fs_sb_info *sbi, block_t blkaddr); 3737 void f2fs_set_data_blkaddr(struct dnode_of_data *dn); 3738 void f2fs_update_data_blkaddr(struct dnode_of_data *dn, block_t blkaddr); 3739 int f2fs_reserve_new_blocks(struct dnode_of_data *dn, blkcnt_t count); 3740 int f2fs_reserve_new_block(struct dnode_of_data *dn); 3741 int f2fs_get_block(struct dnode_of_data *dn, pgoff_t index); 3742 int f2fs_reserve_block(struct dnode_of_data *dn, pgoff_t index); 3743 struct page *f2fs_get_read_data_page(struct inode *inode, pgoff_t index, 3744 int op_flags, bool for_write); 3745 struct page *f2fs_find_data_page(struct inode *inode, pgoff_t index); 3746 struct page *f2fs_get_lock_data_page(struct inode *inode, pgoff_t index, 3747 bool for_write); 3748 struct page *f2fs_get_new_data_page(struct inode *inode, 3749 struct page *ipage, pgoff_t index, bool new_i_size); 3750 int f2fs_do_write_data_page(struct f2fs_io_info *fio); 3751 void f2fs_do_map_lock(struct f2fs_sb_info *sbi, int flag, bool lock); 3752 int f2fs_map_blocks(struct inode *inode, struct f2fs_map_blocks *map, 3753 int create, int flag); 3754 int f2fs_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, 3755 u64 start, u64 len); 3756 int f2fs_encrypt_one_page(struct f2fs_io_info *fio); 3757 bool f2fs_should_update_inplace(struct inode *inode, struct f2fs_io_info *fio); 3758 bool f2fs_should_update_outplace(struct inode *inode, struct f2fs_io_info *fio); 3759 int f2fs_write_single_data_page(struct page *page, int *submitted, 3760 struct bio **bio, sector_t *last_block, 3761 struct writeback_control *wbc, 3762 enum iostat_type io_type, 3763 int compr_blocks, bool allow_balance); 3764 void f2fs_write_failed(struct inode *inode, loff_t to); 3765 void f2fs_invalidate_folio(struct folio *folio, size_t offset, size_t length); 3766 bool f2fs_release_folio(struct folio *folio, gfp_t wait); 3767 #ifdef CONFIG_MIGRATION 3768 int f2fs_migrate_page(struct address_space *mapping, struct page *newpage, 3769 struct page *page, enum migrate_mode mode); 3770 #endif 3771 bool f2fs_overwrite_io(struct inode *inode, loff_t pos, size_t len); 3772 void f2fs_clear_page_cache_dirty_tag(struct page *page); 3773 int f2fs_init_post_read_processing(void); 3774 void f2fs_destroy_post_read_processing(void); 3775 int f2fs_init_post_read_wq(struct f2fs_sb_info *sbi); 3776 void f2fs_destroy_post_read_wq(struct f2fs_sb_info *sbi); 3777 extern const struct iomap_ops f2fs_iomap_ops; 3778 3779 /* 3780 * gc.c 3781 */ 3782 int f2fs_start_gc_thread(struct f2fs_sb_info *sbi); 3783 void f2fs_stop_gc_thread(struct f2fs_sb_info *sbi); 3784 block_t f2fs_start_bidx_of_node(unsigned int node_ofs, struct inode *inode); 3785 int f2fs_gc(struct f2fs_sb_info *sbi, struct f2fs_gc_control *gc_control); 3786 void f2fs_build_gc_manager(struct f2fs_sb_info *sbi); 3787 int f2fs_resize_fs(struct f2fs_sb_info *sbi, __u64 block_count); 3788 int __init f2fs_create_garbage_collection_cache(void); 3789 void f2fs_destroy_garbage_collection_cache(void); 3790 3791 /* 3792 * recovery.c 3793 */ 3794 int f2fs_recover_fsync_data(struct f2fs_sb_info *sbi, bool check_only); 3795 bool f2fs_space_for_roll_forward(struct f2fs_sb_info *sbi); 3796 int __init f2fs_create_recovery_cache(void); 3797 void f2fs_destroy_recovery_cache(void); 3798 3799 /* 3800 * debug.c 3801 */ 3802 #ifdef CONFIG_F2FS_STAT_FS 3803 struct f2fs_stat_info { 3804 struct list_head stat_list; 3805 struct f2fs_sb_info *sbi; 3806 int all_area_segs, sit_area_segs, nat_area_segs, ssa_area_segs; 3807 int main_area_segs, main_area_sections, main_area_zones; 3808 unsigned long long hit_largest, hit_cached, hit_rbtree; 3809 unsigned long long hit_total, total_ext; 3810 int ext_tree, zombie_tree, ext_node; 3811 int ndirty_node, ndirty_dent, ndirty_meta, ndirty_imeta; 3812 int ndirty_data, ndirty_qdata; 3813 unsigned int ndirty_dirs, ndirty_files, nquota_files, ndirty_all; 3814 int nats, dirty_nats, sits, dirty_sits; 3815 int free_nids, avail_nids, alloc_nids; 3816 int total_count, utilization; 3817 int bg_gc, nr_wb_cp_data, nr_wb_data; 3818 int nr_rd_data, nr_rd_node, nr_rd_meta; 3819 int nr_dio_read, nr_dio_write; 3820 unsigned int io_skip_bggc, other_skip_bggc; 3821 int nr_flushing, nr_flushed, flush_list_empty; 3822 int nr_discarding, nr_discarded; 3823 int nr_discard_cmd; 3824 unsigned int undiscard_blks; 3825 int nr_issued_ckpt, nr_total_ckpt, nr_queued_ckpt; 3826 unsigned int cur_ckpt_time, peak_ckpt_time; 3827 int inline_xattr, inline_inode, inline_dir, append, update, orphans; 3828 int compr_inode; 3829 unsigned long long compr_blocks; 3830 int aw_cnt, max_aw_cnt; 3831 unsigned int valid_count, valid_node_count, valid_inode_count, discard_blks; 3832 unsigned int bimodal, avg_vblocks; 3833 int util_free, util_valid, util_invalid; 3834 int rsvd_segs, overp_segs; 3835 int dirty_count, node_pages, meta_pages, compress_pages; 3836 int compress_page_hit; 3837 int prefree_count, call_count, cp_count, bg_cp_count; 3838 int tot_segs, node_segs, data_segs, free_segs, free_secs; 3839 int bg_node_segs, bg_data_segs; 3840 int tot_blks, data_blks, node_blks; 3841 int bg_data_blks, bg_node_blks; 3842 int curseg[NR_CURSEG_TYPE]; 3843 int cursec[NR_CURSEG_TYPE]; 3844 int curzone[NR_CURSEG_TYPE]; 3845 unsigned int dirty_seg[NR_CURSEG_TYPE]; 3846 unsigned int full_seg[NR_CURSEG_TYPE]; 3847 unsigned int valid_blks[NR_CURSEG_TYPE]; 3848 3849 unsigned int meta_count[META_MAX]; 3850 unsigned int segment_count[2]; 3851 unsigned int block_count[2]; 3852 unsigned int inplace_count; 3853 unsigned long long base_mem, cache_mem, page_mem; 3854 }; 3855 3856 static inline struct f2fs_stat_info *F2FS_STAT(struct f2fs_sb_info *sbi) 3857 { 3858 return (struct f2fs_stat_info *)sbi->stat_info; 3859 } 3860 3861 #define stat_inc_cp_count(si) ((si)->cp_count++) 3862 #define stat_inc_bg_cp_count(si) ((si)->bg_cp_count++) 3863 #define stat_inc_call_count(si) ((si)->call_count++) 3864 #define stat_inc_bggc_count(si) ((si)->bg_gc++) 3865 #define stat_io_skip_bggc_count(sbi) ((sbi)->io_skip_bggc++) 3866 #define stat_other_skip_bggc_count(sbi) ((sbi)->other_skip_bggc++) 3867 #define stat_inc_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]++) 3868 #define stat_dec_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]--) 3869 #define stat_inc_total_hit(sbi) (atomic64_inc(&(sbi)->total_hit_ext)) 3870 #define stat_inc_rbtree_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_rbtree)) 3871 #define stat_inc_largest_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_largest)) 3872 #define stat_inc_cached_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_cached)) 3873 #define stat_inc_inline_xattr(inode) \ 3874 do { \ 3875 if (f2fs_has_inline_xattr(inode)) \ 3876 (atomic_inc(&F2FS_I_SB(inode)->inline_xattr)); \ 3877 } while (0) 3878 #define stat_dec_inline_xattr(inode) \ 3879 do { \ 3880 if (f2fs_has_inline_xattr(inode)) \ 3881 (atomic_dec(&F2FS_I_SB(inode)->inline_xattr)); \ 3882 } while (0) 3883 #define stat_inc_inline_inode(inode) \ 3884 do { \ 3885 if (f2fs_has_inline_data(inode)) \ 3886 (atomic_inc(&F2FS_I_SB(inode)->inline_inode)); \ 3887 } while (0) 3888 #define stat_dec_inline_inode(inode) \ 3889 do { \ 3890 if (f2fs_has_inline_data(inode)) \ 3891 (atomic_dec(&F2FS_I_SB(inode)->inline_inode)); \ 3892 } while (0) 3893 #define stat_inc_inline_dir(inode) \ 3894 do { \ 3895 if (f2fs_has_inline_dentry(inode)) \ 3896 (atomic_inc(&F2FS_I_SB(inode)->inline_dir)); \ 3897 } while (0) 3898 #define stat_dec_inline_dir(inode) \ 3899 do { \ 3900 if (f2fs_has_inline_dentry(inode)) \ 3901 (atomic_dec(&F2FS_I_SB(inode)->inline_dir)); \ 3902 } while (0) 3903 #define stat_inc_compr_inode(inode) \ 3904 do { \ 3905 if (f2fs_compressed_file(inode)) \ 3906 (atomic_inc(&F2FS_I_SB(inode)->compr_inode)); \ 3907 } while (0) 3908 #define stat_dec_compr_inode(inode) \ 3909 do { \ 3910 if (f2fs_compressed_file(inode)) \ 3911 (atomic_dec(&F2FS_I_SB(inode)->compr_inode)); \ 3912 } while (0) 3913 #define stat_add_compr_blocks(inode, blocks) \ 3914 (atomic64_add(blocks, &F2FS_I_SB(inode)->compr_blocks)) 3915 #define stat_sub_compr_blocks(inode, blocks) \ 3916 (atomic64_sub(blocks, &F2FS_I_SB(inode)->compr_blocks)) 3917 #define stat_inc_meta_count(sbi, blkaddr) \ 3918 do { \ 3919 if (blkaddr < SIT_I(sbi)->sit_base_addr) \ 3920 atomic_inc(&(sbi)->meta_count[META_CP]); \ 3921 else if (blkaddr < NM_I(sbi)->nat_blkaddr) \ 3922 atomic_inc(&(sbi)->meta_count[META_SIT]); \ 3923 else if (blkaddr < SM_I(sbi)->ssa_blkaddr) \ 3924 atomic_inc(&(sbi)->meta_count[META_NAT]); \ 3925 else if (blkaddr < SM_I(sbi)->main_blkaddr) \ 3926 atomic_inc(&(sbi)->meta_count[META_SSA]); \ 3927 } while (0) 3928 #define stat_inc_seg_type(sbi, curseg) \ 3929 ((sbi)->segment_count[(curseg)->alloc_type]++) 3930 #define stat_inc_block_count(sbi, curseg) \ 3931 ((sbi)->block_count[(curseg)->alloc_type]++) 3932 #define stat_inc_inplace_blocks(sbi) \ 3933 (atomic_inc(&(sbi)->inplace_count)) 3934 #define stat_update_max_atomic_write(inode) \ 3935 do { \ 3936 int cur = F2FS_I_SB(inode)->atomic_files; \ 3937 int max = atomic_read(&F2FS_I_SB(inode)->max_aw_cnt); \ 3938 if (cur > max) \ 3939 atomic_set(&F2FS_I_SB(inode)->max_aw_cnt, cur); \ 3940 } while (0) 3941 #define stat_inc_seg_count(sbi, type, gc_type) \ 3942 do { \ 3943 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 3944 si->tot_segs++; \ 3945 if ((type) == SUM_TYPE_DATA) { \ 3946 si->data_segs++; \ 3947 si->bg_data_segs += (gc_type == BG_GC) ? 1 : 0; \ 3948 } else { \ 3949 si->node_segs++; \ 3950 si->bg_node_segs += (gc_type == BG_GC) ? 1 : 0; \ 3951 } \ 3952 } while (0) 3953 3954 #define stat_inc_tot_blk_count(si, blks) \ 3955 ((si)->tot_blks += (blks)) 3956 3957 #define stat_inc_data_blk_count(sbi, blks, gc_type) \ 3958 do { \ 3959 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 3960 stat_inc_tot_blk_count(si, blks); \ 3961 si->data_blks += (blks); \ 3962 si->bg_data_blks += ((gc_type) == BG_GC) ? (blks) : 0; \ 3963 } while (0) 3964 3965 #define stat_inc_node_blk_count(sbi, blks, gc_type) \ 3966 do { \ 3967 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 3968 stat_inc_tot_blk_count(si, blks); \ 3969 si->node_blks += (blks); \ 3970 si->bg_node_blks += ((gc_type) == BG_GC) ? (blks) : 0; \ 3971 } while (0) 3972 3973 int f2fs_build_stats(struct f2fs_sb_info *sbi); 3974 void f2fs_destroy_stats(struct f2fs_sb_info *sbi); 3975 void __init f2fs_create_root_stats(void); 3976 void f2fs_destroy_root_stats(void); 3977 void f2fs_update_sit_info(struct f2fs_sb_info *sbi); 3978 #else 3979 #define stat_inc_cp_count(si) do { } while (0) 3980 #define stat_inc_bg_cp_count(si) do { } while (0) 3981 #define stat_inc_call_count(si) do { } while (0) 3982 #define stat_inc_bggc_count(si) do { } while (0) 3983 #define stat_io_skip_bggc_count(sbi) do { } while (0) 3984 #define stat_other_skip_bggc_count(sbi) do { } while (0) 3985 #define stat_inc_dirty_inode(sbi, type) do { } while (0) 3986 #define stat_dec_dirty_inode(sbi, type) do { } while (0) 3987 #define stat_inc_total_hit(sbi) do { } while (0) 3988 #define stat_inc_rbtree_node_hit(sbi) do { } while (0) 3989 #define stat_inc_largest_node_hit(sbi) do { } while (0) 3990 #define stat_inc_cached_node_hit(sbi) do { } while (0) 3991 #define stat_inc_inline_xattr(inode) do { } while (0) 3992 #define stat_dec_inline_xattr(inode) do { } while (0) 3993 #define stat_inc_inline_inode(inode) do { } while (0) 3994 #define stat_dec_inline_inode(inode) do { } while (0) 3995 #define stat_inc_inline_dir(inode) do { } while (0) 3996 #define stat_dec_inline_dir(inode) do { } while (0) 3997 #define stat_inc_compr_inode(inode) do { } while (0) 3998 #define stat_dec_compr_inode(inode) do { } while (0) 3999 #define stat_add_compr_blocks(inode, blocks) do { } while (0) 4000 #define stat_sub_compr_blocks(inode, blocks) do { } while (0) 4001 #define stat_update_max_atomic_write(inode) do { } while (0) 4002 #define stat_inc_meta_count(sbi, blkaddr) do { } while (0) 4003 #define stat_inc_seg_type(sbi, curseg) do { } while (0) 4004 #define stat_inc_block_count(sbi, curseg) do { } while (0) 4005 #define stat_inc_inplace_blocks(sbi) do { } while (0) 4006 #define stat_inc_seg_count(sbi, type, gc_type) do { } while (0) 4007 #define stat_inc_tot_blk_count(si, blks) do { } while (0) 4008 #define stat_inc_data_blk_count(sbi, blks, gc_type) do { } while (0) 4009 #define stat_inc_node_blk_count(sbi, blks, gc_type) do { } while (0) 4010 4011 static inline int f2fs_build_stats(struct f2fs_sb_info *sbi) { return 0; } 4012 static inline void f2fs_destroy_stats(struct f2fs_sb_info *sbi) { } 4013 static inline void __init f2fs_create_root_stats(void) { } 4014 static inline void f2fs_destroy_root_stats(void) { } 4015 static inline void f2fs_update_sit_info(struct f2fs_sb_info *sbi) {} 4016 #endif 4017 4018 extern const struct file_operations f2fs_dir_operations; 4019 extern const struct file_operations f2fs_file_operations; 4020 extern const struct inode_operations f2fs_file_inode_operations; 4021 extern const struct address_space_operations f2fs_dblock_aops; 4022 extern const struct address_space_operations f2fs_node_aops; 4023 extern const struct address_space_operations f2fs_meta_aops; 4024 extern const struct inode_operations f2fs_dir_inode_operations; 4025 extern const struct inode_operations f2fs_symlink_inode_operations; 4026 extern const struct inode_operations f2fs_encrypted_symlink_inode_operations; 4027 extern const struct inode_operations f2fs_special_inode_operations; 4028 extern struct kmem_cache *f2fs_inode_entry_slab; 4029 4030 /* 4031 * inline.c 4032 */ 4033 bool f2fs_may_inline_data(struct inode *inode); 4034 bool f2fs_sanity_check_inline_data(struct inode *inode); 4035 bool f2fs_may_inline_dentry(struct inode *inode); 4036 void f2fs_do_read_inline_data(struct page *page, struct page *ipage); 4037 void f2fs_truncate_inline_inode(struct inode *inode, 4038 struct page *ipage, u64 from); 4039 int f2fs_read_inline_data(struct inode *inode, struct page *page); 4040 int f2fs_convert_inline_page(struct dnode_of_data *dn, struct page *page); 4041 int f2fs_convert_inline_inode(struct inode *inode); 4042 int f2fs_try_convert_inline_dir(struct inode *dir, struct dentry *dentry); 4043 int f2fs_write_inline_data(struct inode *inode, struct page *page); 4044 int f2fs_recover_inline_data(struct inode *inode, struct page *npage); 4045 struct f2fs_dir_entry *f2fs_find_in_inline_dir(struct inode *dir, 4046 const struct f2fs_filename *fname, 4047 struct page **res_page); 4048 int f2fs_make_empty_inline_dir(struct inode *inode, struct inode *parent, 4049 struct page *ipage); 4050 int f2fs_add_inline_entry(struct inode *dir, const struct f2fs_filename *fname, 4051 struct inode *inode, nid_t ino, umode_t mode); 4052 void f2fs_delete_inline_entry(struct f2fs_dir_entry *dentry, 4053 struct page *page, struct inode *dir, 4054 struct inode *inode); 4055 bool f2fs_empty_inline_dir(struct inode *dir); 4056 int f2fs_read_inline_dir(struct file *file, struct dir_context *ctx, 4057 struct fscrypt_str *fstr); 4058 int f2fs_inline_data_fiemap(struct inode *inode, 4059 struct fiemap_extent_info *fieinfo, 4060 __u64 start, __u64 len); 4061 4062 /* 4063 * shrinker.c 4064 */ 4065 unsigned long f2fs_shrink_count(struct shrinker *shrink, 4066 struct shrink_control *sc); 4067 unsigned long f2fs_shrink_scan(struct shrinker *shrink, 4068 struct shrink_control *sc); 4069 void f2fs_join_shrinker(struct f2fs_sb_info *sbi); 4070 void f2fs_leave_shrinker(struct f2fs_sb_info *sbi); 4071 4072 /* 4073 * extent_cache.c 4074 */ 4075 struct rb_entry *f2fs_lookup_rb_tree(struct rb_root_cached *root, 4076 struct rb_entry *cached_re, unsigned int ofs); 4077 struct rb_node **f2fs_lookup_rb_tree_ext(struct f2fs_sb_info *sbi, 4078 struct rb_root_cached *root, 4079 struct rb_node **parent, 4080 unsigned long long key, bool *left_most); 4081 struct rb_node **f2fs_lookup_rb_tree_for_insert(struct f2fs_sb_info *sbi, 4082 struct rb_root_cached *root, 4083 struct rb_node **parent, 4084 unsigned int ofs, bool *leftmost); 4085 struct rb_entry *f2fs_lookup_rb_tree_ret(struct rb_root_cached *root, 4086 struct rb_entry *cached_re, unsigned int ofs, 4087 struct rb_entry **prev_entry, struct rb_entry **next_entry, 4088 struct rb_node ***insert_p, struct rb_node **insert_parent, 4089 bool force, bool *leftmost); 4090 bool f2fs_check_rb_tree_consistence(struct f2fs_sb_info *sbi, 4091 struct rb_root_cached *root, bool check_key); 4092 unsigned int f2fs_shrink_extent_tree(struct f2fs_sb_info *sbi, int nr_shrink); 4093 void f2fs_init_extent_tree(struct inode *inode, struct page *ipage); 4094 void f2fs_drop_extent_tree(struct inode *inode); 4095 unsigned int f2fs_destroy_extent_node(struct inode *inode); 4096 void f2fs_destroy_extent_tree(struct inode *inode); 4097 bool f2fs_lookup_extent_cache(struct inode *inode, pgoff_t pgofs, 4098 struct extent_info *ei); 4099 void f2fs_update_extent_cache(struct dnode_of_data *dn); 4100 void f2fs_update_extent_cache_range(struct dnode_of_data *dn, 4101 pgoff_t fofs, block_t blkaddr, unsigned int len); 4102 void f2fs_init_extent_cache_info(struct f2fs_sb_info *sbi); 4103 int __init f2fs_create_extent_cache(void); 4104 void f2fs_destroy_extent_cache(void); 4105 4106 /* 4107 * sysfs.c 4108 */ 4109 #define MIN_RA_MUL 2 4110 #define MAX_RA_MUL 256 4111 4112 int __init f2fs_init_sysfs(void); 4113 void f2fs_exit_sysfs(void); 4114 int f2fs_register_sysfs(struct f2fs_sb_info *sbi); 4115 void f2fs_unregister_sysfs(struct f2fs_sb_info *sbi); 4116 4117 /* verity.c */ 4118 extern const struct fsverity_operations f2fs_verityops; 4119 4120 /* 4121 * crypto support 4122 */ 4123 static inline bool f2fs_encrypted_file(struct inode *inode) 4124 { 4125 return IS_ENCRYPTED(inode) && S_ISREG(inode->i_mode); 4126 } 4127 4128 static inline void f2fs_set_encrypted_inode(struct inode *inode) 4129 { 4130 #ifdef CONFIG_FS_ENCRYPTION 4131 file_set_encrypt(inode); 4132 f2fs_set_inode_flags(inode); 4133 #endif 4134 } 4135 4136 /* 4137 * Returns true if the reads of the inode's data need to undergo some 4138 * postprocessing step, like decryption or authenticity verification. 4139 */ 4140 static inline bool f2fs_post_read_required(struct inode *inode) 4141 { 4142 return f2fs_encrypted_file(inode) || fsverity_active(inode) || 4143 f2fs_compressed_file(inode); 4144 } 4145 4146 /* 4147 * compress.c 4148 */ 4149 #ifdef CONFIG_F2FS_FS_COMPRESSION 4150 bool f2fs_is_compressed_page(struct page *page); 4151 struct page *f2fs_compress_control_page(struct page *page); 4152 int f2fs_prepare_compress_overwrite(struct inode *inode, 4153 struct page **pagep, pgoff_t index, void **fsdata); 4154 bool f2fs_compress_write_end(struct inode *inode, void *fsdata, 4155 pgoff_t index, unsigned copied); 4156 int f2fs_truncate_partial_cluster(struct inode *inode, u64 from, bool lock); 4157 void f2fs_compress_write_end_io(struct bio *bio, struct page *page); 4158 bool f2fs_is_compress_backend_ready(struct inode *inode); 4159 int f2fs_init_compress_mempool(void); 4160 void f2fs_destroy_compress_mempool(void); 4161 void f2fs_decompress_cluster(struct decompress_io_ctx *dic); 4162 void f2fs_end_read_compressed_page(struct page *page, bool failed, 4163 block_t blkaddr); 4164 bool f2fs_cluster_is_empty(struct compress_ctx *cc); 4165 bool f2fs_cluster_can_merge_page(struct compress_ctx *cc, pgoff_t index); 4166 bool f2fs_all_cluster_page_loaded(struct compress_ctx *cc, struct pagevec *pvec, 4167 int index, int nr_pages); 4168 bool f2fs_sanity_check_cluster(struct dnode_of_data *dn); 4169 void f2fs_compress_ctx_add_page(struct compress_ctx *cc, struct page *page); 4170 int f2fs_write_multi_pages(struct compress_ctx *cc, 4171 int *submitted, 4172 struct writeback_control *wbc, 4173 enum iostat_type io_type); 4174 int f2fs_is_compressed_cluster(struct inode *inode, pgoff_t index); 4175 void f2fs_update_extent_tree_range_compressed(struct inode *inode, 4176 pgoff_t fofs, block_t blkaddr, unsigned int llen, 4177 unsigned int c_len); 4178 int f2fs_read_multi_pages(struct compress_ctx *cc, struct bio **bio_ret, 4179 unsigned nr_pages, sector_t *last_block_in_bio, 4180 bool is_readahead, bool for_write); 4181 struct decompress_io_ctx *f2fs_alloc_dic(struct compress_ctx *cc); 4182 void f2fs_decompress_end_io(struct decompress_io_ctx *dic, bool failed); 4183 void f2fs_put_page_dic(struct page *page); 4184 unsigned int f2fs_cluster_blocks_are_contiguous(struct dnode_of_data *dn); 4185 int f2fs_init_compress_ctx(struct compress_ctx *cc); 4186 void f2fs_destroy_compress_ctx(struct compress_ctx *cc, bool reuse); 4187 void f2fs_init_compress_info(struct f2fs_sb_info *sbi); 4188 int f2fs_init_compress_inode(struct f2fs_sb_info *sbi); 4189 void f2fs_destroy_compress_inode(struct f2fs_sb_info *sbi); 4190 int f2fs_init_page_array_cache(struct f2fs_sb_info *sbi); 4191 void f2fs_destroy_page_array_cache(struct f2fs_sb_info *sbi); 4192 int __init f2fs_init_compress_cache(void); 4193 void f2fs_destroy_compress_cache(void); 4194 struct address_space *COMPRESS_MAPPING(struct f2fs_sb_info *sbi); 4195 void f2fs_invalidate_compress_page(struct f2fs_sb_info *sbi, block_t blkaddr); 4196 void f2fs_cache_compressed_page(struct f2fs_sb_info *sbi, struct page *page, 4197 nid_t ino, block_t blkaddr); 4198 bool f2fs_load_compressed_page(struct f2fs_sb_info *sbi, struct page *page, 4199 block_t blkaddr); 4200 void f2fs_invalidate_compress_pages(struct f2fs_sb_info *sbi, nid_t ino); 4201 #define inc_compr_inode_stat(inode) \ 4202 do { \ 4203 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); \ 4204 sbi->compr_new_inode++; \ 4205 } while (0) 4206 #define add_compr_block_stat(inode, blocks) \ 4207 do { \ 4208 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); \ 4209 int diff = F2FS_I(inode)->i_cluster_size - blocks; \ 4210 sbi->compr_written_block += blocks; \ 4211 sbi->compr_saved_block += diff; \ 4212 } while (0) 4213 #else 4214 static inline bool f2fs_is_compressed_page(struct page *page) { return false; } 4215 static inline bool f2fs_is_compress_backend_ready(struct inode *inode) 4216 { 4217 if (!f2fs_compressed_file(inode)) 4218 return true; 4219 /* not support compression */ 4220 return false; 4221 } 4222 static inline struct page *f2fs_compress_control_page(struct page *page) 4223 { 4224 WARN_ON_ONCE(1); 4225 return ERR_PTR(-EINVAL); 4226 } 4227 static inline int f2fs_init_compress_mempool(void) { return 0; } 4228 static inline void f2fs_destroy_compress_mempool(void) { } 4229 static inline void f2fs_decompress_cluster(struct decompress_io_ctx *dic) { } 4230 static inline void f2fs_end_read_compressed_page(struct page *page, 4231 bool failed, block_t blkaddr) 4232 { 4233 WARN_ON_ONCE(1); 4234 } 4235 static inline void f2fs_put_page_dic(struct page *page) 4236 { 4237 WARN_ON_ONCE(1); 4238 } 4239 static inline unsigned int f2fs_cluster_blocks_are_contiguous(struct dnode_of_data *dn) { return 0; } 4240 static inline bool f2fs_sanity_check_cluster(struct dnode_of_data *dn) { return false; } 4241 static inline int f2fs_init_compress_inode(struct f2fs_sb_info *sbi) { return 0; } 4242 static inline void f2fs_destroy_compress_inode(struct f2fs_sb_info *sbi) { } 4243 static inline int f2fs_init_page_array_cache(struct f2fs_sb_info *sbi) { return 0; } 4244 static inline void f2fs_destroy_page_array_cache(struct f2fs_sb_info *sbi) { } 4245 static inline int __init f2fs_init_compress_cache(void) { return 0; } 4246 static inline void f2fs_destroy_compress_cache(void) { } 4247 static inline void f2fs_invalidate_compress_page(struct f2fs_sb_info *sbi, 4248 block_t blkaddr) { } 4249 static inline void f2fs_cache_compressed_page(struct f2fs_sb_info *sbi, 4250 struct page *page, nid_t ino, block_t blkaddr) { } 4251 static inline bool f2fs_load_compressed_page(struct f2fs_sb_info *sbi, 4252 struct page *page, block_t blkaddr) { return false; } 4253 static inline void f2fs_invalidate_compress_pages(struct f2fs_sb_info *sbi, 4254 nid_t ino) { } 4255 #define inc_compr_inode_stat(inode) do { } while (0) 4256 static inline void f2fs_update_extent_tree_range_compressed(struct inode *inode, 4257 pgoff_t fofs, block_t blkaddr, unsigned int llen, 4258 unsigned int c_len) { } 4259 #endif 4260 4261 static inline void set_compress_context(struct inode *inode) 4262 { 4263 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 4264 4265 F2FS_I(inode)->i_compress_algorithm = 4266 F2FS_OPTION(sbi).compress_algorithm; 4267 F2FS_I(inode)->i_log_cluster_size = 4268 F2FS_OPTION(sbi).compress_log_size; 4269 F2FS_I(inode)->i_compress_flag = 4270 F2FS_OPTION(sbi).compress_chksum ? 4271 1 << COMPRESS_CHKSUM : 0; 4272 F2FS_I(inode)->i_cluster_size = 4273 1 << F2FS_I(inode)->i_log_cluster_size; 4274 if ((F2FS_I(inode)->i_compress_algorithm == COMPRESS_LZ4 || 4275 F2FS_I(inode)->i_compress_algorithm == COMPRESS_ZSTD) && 4276 F2FS_OPTION(sbi).compress_level) 4277 F2FS_I(inode)->i_compress_flag |= 4278 F2FS_OPTION(sbi).compress_level << 4279 COMPRESS_LEVEL_OFFSET; 4280 F2FS_I(inode)->i_flags |= F2FS_COMPR_FL; 4281 set_inode_flag(inode, FI_COMPRESSED_FILE); 4282 stat_inc_compr_inode(inode); 4283 inc_compr_inode_stat(inode); 4284 f2fs_mark_inode_dirty_sync(inode, true); 4285 } 4286 4287 static inline bool f2fs_disable_compressed_file(struct inode *inode) 4288 { 4289 struct f2fs_inode_info *fi = F2FS_I(inode); 4290 4291 if (!f2fs_compressed_file(inode)) 4292 return true; 4293 if (S_ISREG(inode->i_mode) && F2FS_HAS_BLOCKS(inode)) 4294 return false; 4295 4296 fi->i_flags &= ~F2FS_COMPR_FL; 4297 stat_dec_compr_inode(inode); 4298 clear_inode_flag(inode, FI_COMPRESSED_FILE); 4299 f2fs_mark_inode_dirty_sync(inode, true); 4300 return true; 4301 } 4302 4303 #define F2FS_FEATURE_FUNCS(name, flagname) \ 4304 static inline int f2fs_sb_has_##name(struct f2fs_sb_info *sbi) \ 4305 { \ 4306 return F2FS_HAS_FEATURE(sbi, F2FS_FEATURE_##flagname); \ 4307 } 4308 4309 F2FS_FEATURE_FUNCS(encrypt, ENCRYPT); 4310 F2FS_FEATURE_FUNCS(blkzoned, BLKZONED); 4311 F2FS_FEATURE_FUNCS(extra_attr, EXTRA_ATTR); 4312 F2FS_FEATURE_FUNCS(project_quota, PRJQUOTA); 4313 F2FS_FEATURE_FUNCS(inode_chksum, INODE_CHKSUM); 4314 F2FS_FEATURE_FUNCS(flexible_inline_xattr, FLEXIBLE_INLINE_XATTR); 4315 F2FS_FEATURE_FUNCS(quota_ino, QUOTA_INO); 4316 F2FS_FEATURE_FUNCS(inode_crtime, INODE_CRTIME); 4317 F2FS_FEATURE_FUNCS(lost_found, LOST_FOUND); 4318 F2FS_FEATURE_FUNCS(verity, VERITY); 4319 F2FS_FEATURE_FUNCS(sb_chksum, SB_CHKSUM); 4320 F2FS_FEATURE_FUNCS(casefold, CASEFOLD); 4321 F2FS_FEATURE_FUNCS(compression, COMPRESSION); 4322 F2FS_FEATURE_FUNCS(readonly, RO); 4323 4324 static inline bool f2fs_may_extent_tree(struct inode *inode) 4325 { 4326 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 4327 4328 if (!test_opt(sbi, EXTENT_CACHE) || 4329 is_inode_flag_set(inode, FI_NO_EXTENT) || 4330 (is_inode_flag_set(inode, FI_COMPRESSED_FILE) && 4331 !f2fs_sb_has_readonly(sbi))) 4332 return false; 4333 4334 /* 4335 * for recovered files during mount do not create extents 4336 * if shrinker is not registered. 4337 */ 4338 if (list_empty(&sbi->s_list)) 4339 return false; 4340 4341 return S_ISREG(inode->i_mode); 4342 } 4343 4344 #ifdef CONFIG_BLK_DEV_ZONED 4345 static inline bool f2fs_blkz_is_seq(struct f2fs_sb_info *sbi, int devi, 4346 block_t blkaddr) 4347 { 4348 unsigned int zno = blkaddr >> sbi->log_blocks_per_blkz; 4349 4350 return test_bit(zno, FDEV(devi).blkz_seq); 4351 } 4352 #endif 4353 4354 static inline bool f2fs_hw_should_discard(struct f2fs_sb_info *sbi) 4355 { 4356 return f2fs_sb_has_blkzoned(sbi); 4357 } 4358 4359 static inline bool f2fs_bdev_support_discard(struct block_device *bdev) 4360 { 4361 return bdev_max_discard_sectors(bdev) || bdev_is_zoned(bdev); 4362 } 4363 4364 static inline bool f2fs_hw_support_discard(struct f2fs_sb_info *sbi) 4365 { 4366 int i; 4367 4368 if (!f2fs_is_multi_device(sbi)) 4369 return f2fs_bdev_support_discard(sbi->sb->s_bdev); 4370 4371 for (i = 0; i < sbi->s_ndevs; i++) 4372 if (f2fs_bdev_support_discard(FDEV(i).bdev)) 4373 return true; 4374 return false; 4375 } 4376 4377 static inline bool f2fs_realtime_discard_enable(struct f2fs_sb_info *sbi) 4378 { 4379 return (test_opt(sbi, DISCARD) && f2fs_hw_support_discard(sbi)) || 4380 f2fs_hw_should_discard(sbi); 4381 } 4382 4383 static inline bool f2fs_hw_is_readonly(struct f2fs_sb_info *sbi) 4384 { 4385 int i; 4386 4387 if (!f2fs_is_multi_device(sbi)) 4388 return bdev_read_only(sbi->sb->s_bdev); 4389 4390 for (i = 0; i < sbi->s_ndevs; i++) 4391 if (bdev_read_only(FDEV(i).bdev)) 4392 return true; 4393 return false; 4394 } 4395 4396 static inline bool f2fs_lfs_mode(struct f2fs_sb_info *sbi) 4397 { 4398 return F2FS_OPTION(sbi).fs_mode == FS_MODE_LFS; 4399 } 4400 4401 static inline bool f2fs_may_compress(struct inode *inode) 4402 { 4403 if (IS_SWAPFILE(inode) || f2fs_is_pinned_file(inode) || 4404 f2fs_is_atomic_file(inode)) 4405 return false; 4406 return S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode); 4407 } 4408 4409 static inline void f2fs_i_compr_blocks_update(struct inode *inode, 4410 u64 blocks, bool add) 4411 { 4412 struct f2fs_inode_info *fi = F2FS_I(inode); 4413 int diff = fi->i_cluster_size - blocks; 4414 4415 /* don't update i_compr_blocks if saved blocks were released */ 4416 if (!add && !atomic_read(&fi->i_compr_blocks)) 4417 return; 4418 4419 if (add) { 4420 atomic_add(diff, &fi->i_compr_blocks); 4421 stat_add_compr_blocks(inode, diff); 4422 } else { 4423 atomic_sub(diff, &fi->i_compr_blocks); 4424 stat_sub_compr_blocks(inode, diff); 4425 } 4426 f2fs_mark_inode_dirty_sync(inode, true); 4427 } 4428 4429 static inline int block_unaligned_IO(struct inode *inode, 4430 struct kiocb *iocb, struct iov_iter *iter) 4431 { 4432 unsigned int i_blkbits = READ_ONCE(inode->i_blkbits); 4433 unsigned int blocksize_mask = (1 << i_blkbits) - 1; 4434 loff_t offset = iocb->ki_pos; 4435 unsigned long align = offset | iov_iter_alignment(iter); 4436 4437 return align & blocksize_mask; 4438 } 4439 4440 static inline bool f2fs_allow_multi_device_dio(struct f2fs_sb_info *sbi, 4441 int flag) 4442 { 4443 if (!f2fs_is_multi_device(sbi)) 4444 return false; 4445 if (flag != F2FS_GET_BLOCK_DIO) 4446 return false; 4447 return sbi->aligned_blksize; 4448 } 4449 4450 static inline bool f2fs_force_buffered_io(struct inode *inode, 4451 struct kiocb *iocb, struct iov_iter *iter) 4452 { 4453 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 4454 int rw = iov_iter_rw(iter); 4455 4456 if (!fscrypt_dio_supported(iocb, iter)) 4457 return true; 4458 if (fsverity_active(inode)) 4459 return true; 4460 if (f2fs_compressed_file(inode)) 4461 return true; 4462 4463 /* disallow direct IO if any of devices has unaligned blksize */ 4464 if (f2fs_is_multi_device(sbi) && !sbi->aligned_blksize) 4465 return true; 4466 /* 4467 * for blkzoned device, fallback direct IO to buffered IO, so 4468 * all IOs can be serialized by log-structured write. 4469 */ 4470 if (f2fs_sb_has_blkzoned(sbi)) 4471 return true; 4472 if (f2fs_lfs_mode(sbi) && (rw == WRITE)) { 4473 if (block_unaligned_IO(inode, iocb, iter)) 4474 return true; 4475 if (F2FS_IO_ALIGNED(sbi)) 4476 return true; 4477 } 4478 if (is_sbi_flag_set(F2FS_I_SB(inode), SBI_CP_DISABLED)) 4479 return true; 4480 4481 return false; 4482 } 4483 4484 static inline bool f2fs_need_verity(const struct inode *inode, pgoff_t idx) 4485 { 4486 return fsverity_active(inode) && 4487 idx < DIV_ROUND_UP(inode->i_size, PAGE_SIZE); 4488 } 4489 4490 #ifdef CONFIG_F2FS_FAULT_INJECTION 4491 extern void f2fs_build_fault_attr(struct f2fs_sb_info *sbi, unsigned int rate, 4492 unsigned int type); 4493 #else 4494 #define f2fs_build_fault_attr(sbi, rate, type) do { } while (0) 4495 #endif 4496 4497 static inline bool is_journalled_quota(struct f2fs_sb_info *sbi) 4498 { 4499 #ifdef CONFIG_QUOTA 4500 if (f2fs_sb_has_quota_ino(sbi)) 4501 return true; 4502 if (F2FS_OPTION(sbi).s_qf_names[USRQUOTA] || 4503 F2FS_OPTION(sbi).s_qf_names[GRPQUOTA] || 4504 F2FS_OPTION(sbi).s_qf_names[PRJQUOTA]) 4505 return true; 4506 #endif 4507 return false; 4508 } 4509 4510 static inline bool f2fs_block_unit_discard(struct f2fs_sb_info *sbi) 4511 { 4512 return F2FS_OPTION(sbi).discard_unit == DISCARD_UNIT_BLOCK; 4513 } 4514 4515 static inline void f2fs_io_schedule_timeout(long timeout) 4516 { 4517 set_current_state(TASK_UNINTERRUPTIBLE); 4518 io_schedule_timeout(timeout); 4519 } 4520 4521 static inline void f2fs_handle_page_eio(struct f2fs_sb_info *sbi, pgoff_t ofs, 4522 enum page_type type) 4523 { 4524 if (unlikely(f2fs_cp_error(sbi))) 4525 return; 4526 4527 if (ofs == sbi->page_eio_ofs[type]) { 4528 if (sbi->page_eio_cnt[type]++ == MAX_RETRY_PAGE_EIO) 4529 set_ckpt_flags(sbi, CP_ERROR_FLAG); 4530 } else { 4531 sbi->page_eio_ofs[type] = ofs; 4532 sbi->page_eio_cnt[type] = 0; 4533 } 4534 } 4535 4536 #define EFSBADCRC EBADMSG /* Bad CRC detected */ 4537 #define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */ 4538 4539 #endif /* _LINUX_F2FS_H */ 4540