1 /* SPDX-License-Identifier: GPL-2.0 */ 2 /* 3 * fs/f2fs/f2fs.h 4 * 5 * Copyright (c) 2012 Samsung Electronics Co., Ltd. 6 * http://www.samsung.com/ 7 */ 8 #ifndef _LINUX_F2FS_H 9 #define _LINUX_F2FS_H 10 11 #include <linux/uio.h> 12 #include <linux/types.h> 13 #include <linux/page-flags.h> 14 #include <linux/buffer_head.h> 15 #include <linux/slab.h> 16 #include <linux/crc32.h> 17 #include <linux/magic.h> 18 #include <linux/kobject.h> 19 #include <linux/sched.h> 20 #include <linux/cred.h> 21 #include <linux/vmalloc.h> 22 #include <linux/bio.h> 23 #include <linux/blkdev.h> 24 #include <linux/quotaops.h> 25 #include <linux/part_stat.h> 26 #include <crypto/hash.h> 27 28 #include <linux/fscrypt.h> 29 #include <linux/fsverity.h> 30 31 struct pagevec; 32 33 #ifdef CONFIG_F2FS_CHECK_FS 34 #define f2fs_bug_on(sbi, condition) BUG_ON(condition) 35 #else 36 #define f2fs_bug_on(sbi, condition) \ 37 do { \ 38 if (WARN_ON(condition)) \ 39 set_sbi_flag(sbi, SBI_NEED_FSCK); \ 40 } while (0) 41 #endif 42 43 enum { 44 FAULT_KMALLOC, 45 FAULT_KVMALLOC, 46 FAULT_PAGE_ALLOC, 47 FAULT_PAGE_GET, 48 FAULT_ALLOC_BIO, /* it's obsolete due to bio_alloc() will never fail */ 49 FAULT_ALLOC_NID, 50 FAULT_ORPHAN, 51 FAULT_BLOCK, 52 FAULT_DIR_DEPTH, 53 FAULT_EVICT_INODE, 54 FAULT_TRUNCATE, 55 FAULT_READ_IO, 56 FAULT_CHECKPOINT, 57 FAULT_DISCARD, 58 FAULT_WRITE_IO, 59 FAULT_SLAB_ALLOC, 60 FAULT_DQUOT_INIT, 61 FAULT_LOCK_OP, 62 FAULT_MAX, 63 }; 64 65 #ifdef CONFIG_F2FS_FAULT_INJECTION 66 #define F2FS_ALL_FAULT_TYPE ((1 << FAULT_MAX) - 1) 67 68 struct f2fs_fault_info { 69 atomic_t inject_ops; 70 unsigned int inject_rate; 71 unsigned int inject_type; 72 }; 73 74 extern const char *f2fs_fault_name[FAULT_MAX]; 75 #define IS_FAULT_SET(fi, type) ((fi)->inject_type & (1 << (type))) 76 #endif 77 78 /* 79 * For mount options 80 */ 81 #define F2FS_MOUNT_DISABLE_ROLL_FORWARD 0x00000002 82 #define F2FS_MOUNT_DISCARD 0x00000004 83 #define F2FS_MOUNT_NOHEAP 0x00000008 84 #define F2FS_MOUNT_XATTR_USER 0x00000010 85 #define F2FS_MOUNT_POSIX_ACL 0x00000020 86 #define F2FS_MOUNT_DISABLE_EXT_IDENTIFY 0x00000040 87 #define F2FS_MOUNT_INLINE_XATTR 0x00000080 88 #define F2FS_MOUNT_INLINE_DATA 0x00000100 89 #define F2FS_MOUNT_INLINE_DENTRY 0x00000200 90 #define F2FS_MOUNT_FLUSH_MERGE 0x00000400 91 #define F2FS_MOUNT_NOBARRIER 0x00000800 92 #define F2FS_MOUNT_FASTBOOT 0x00001000 93 #define F2FS_MOUNT_EXTENT_CACHE 0x00002000 94 #define F2FS_MOUNT_DATA_FLUSH 0x00008000 95 #define F2FS_MOUNT_FAULT_INJECTION 0x00010000 96 #define F2FS_MOUNT_USRQUOTA 0x00080000 97 #define F2FS_MOUNT_GRPQUOTA 0x00100000 98 #define F2FS_MOUNT_PRJQUOTA 0x00200000 99 #define F2FS_MOUNT_QUOTA 0x00400000 100 #define F2FS_MOUNT_INLINE_XATTR_SIZE 0x00800000 101 #define F2FS_MOUNT_RESERVE_ROOT 0x01000000 102 #define F2FS_MOUNT_DISABLE_CHECKPOINT 0x02000000 103 #define F2FS_MOUNT_NORECOVERY 0x04000000 104 #define F2FS_MOUNT_ATGC 0x08000000 105 #define F2FS_MOUNT_MERGE_CHECKPOINT 0x10000000 106 #define F2FS_MOUNT_GC_MERGE 0x20000000 107 #define F2FS_MOUNT_COMPRESS_CACHE 0x40000000 108 109 #define F2FS_OPTION(sbi) ((sbi)->mount_opt) 110 #define clear_opt(sbi, option) (F2FS_OPTION(sbi).opt &= ~F2FS_MOUNT_##option) 111 #define set_opt(sbi, option) (F2FS_OPTION(sbi).opt |= F2FS_MOUNT_##option) 112 #define test_opt(sbi, option) (F2FS_OPTION(sbi).opt & F2FS_MOUNT_##option) 113 114 #define ver_after(a, b) (typecheck(unsigned long long, a) && \ 115 typecheck(unsigned long long, b) && \ 116 ((long long)((a) - (b)) > 0)) 117 118 typedef u32 block_t; /* 119 * should not change u32, since it is the on-disk block 120 * address format, __le32. 121 */ 122 typedef u32 nid_t; 123 124 #define COMPRESS_EXT_NUM 16 125 126 /* 127 * An implementation of an rwsem that is explicitly unfair to readers. This 128 * prevents priority inversion when a low-priority reader acquires the read lock 129 * while sleeping on the write lock but the write lock is needed by 130 * higher-priority clients. 131 */ 132 133 struct f2fs_rwsem { 134 struct rw_semaphore internal_rwsem; 135 wait_queue_head_t read_waiters; 136 }; 137 138 struct f2fs_mount_info { 139 unsigned int opt; 140 int write_io_size_bits; /* Write IO size bits */ 141 block_t root_reserved_blocks; /* root reserved blocks */ 142 kuid_t s_resuid; /* reserved blocks for uid */ 143 kgid_t s_resgid; /* reserved blocks for gid */ 144 int active_logs; /* # of active logs */ 145 int inline_xattr_size; /* inline xattr size */ 146 #ifdef CONFIG_F2FS_FAULT_INJECTION 147 struct f2fs_fault_info fault_info; /* For fault injection */ 148 #endif 149 #ifdef CONFIG_QUOTA 150 /* Names of quota files with journalled quota */ 151 char *s_qf_names[MAXQUOTAS]; 152 int s_jquota_fmt; /* Format of quota to use */ 153 #endif 154 /* For which write hints are passed down to block layer */ 155 int whint_mode; 156 int alloc_mode; /* segment allocation policy */ 157 int fsync_mode; /* fsync policy */ 158 int fs_mode; /* fs mode: LFS or ADAPTIVE */ 159 int bggc_mode; /* bggc mode: off, on or sync */ 160 int discard_unit; /* 161 * discard command's offset/size should 162 * be aligned to this unit: block, 163 * segment or section 164 */ 165 struct fscrypt_dummy_policy dummy_enc_policy; /* test dummy encryption */ 166 block_t unusable_cap_perc; /* percentage for cap */ 167 block_t unusable_cap; /* Amount of space allowed to be 168 * unusable when disabling checkpoint 169 */ 170 171 /* For compression */ 172 unsigned char compress_algorithm; /* algorithm type */ 173 unsigned char compress_log_size; /* cluster log size */ 174 unsigned char compress_level; /* compress level */ 175 bool compress_chksum; /* compressed data chksum */ 176 unsigned char compress_ext_cnt; /* extension count */ 177 unsigned char nocompress_ext_cnt; /* nocompress extension count */ 178 int compress_mode; /* compression mode */ 179 unsigned char extensions[COMPRESS_EXT_NUM][F2FS_EXTENSION_LEN]; /* extensions */ 180 unsigned char noextensions[COMPRESS_EXT_NUM][F2FS_EXTENSION_LEN]; /* extensions */ 181 }; 182 183 #define F2FS_FEATURE_ENCRYPT 0x0001 184 #define F2FS_FEATURE_BLKZONED 0x0002 185 #define F2FS_FEATURE_ATOMIC_WRITE 0x0004 186 #define F2FS_FEATURE_EXTRA_ATTR 0x0008 187 #define F2FS_FEATURE_PRJQUOTA 0x0010 188 #define F2FS_FEATURE_INODE_CHKSUM 0x0020 189 #define F2FS_FEATURE_FLEXIBLE_INLINE_XATTR 0x0040 190 #define F2FS_FEATURE_QUOTA_INO 0x0080 191 #define F2FS_FEATURE_INODE_CRTIME 0x0100 192 #define F2FS_FEATURE_LOST_FOUND 0x0200 193 #define F2FS_FEATURE_VERITY 0x0400 194 #define F2FS_FEATURE_SB_CHKSUM 0x0800 195 #define F2FS_FEATURE_CASEFOLD 0x1000 196 #define F2FS_FEATURE_COMPRESSION 0x2000 197 #define F2FS_FEATURE_RO 0x4000 198 199 #define __F2FS_HAS_FEATURE(raw_super, mask) \ 200 ((raw_super->feature & cpu_to_le32(mask)) != 0) 201 #define F2FS_HAS_FEATURE(sbi, mask) __F2FS_HAS_FEATURE(sbi->raw_super, mask) 202 #define F2FS_SET_FEATURE(sbi, mask) \ 203 (sbi->raw_super->feature |= cpu_to_le32(mask)) 204 #define F2FS_CLEAR_FEATURE(sbi, mask) \ 205 (sbi->raw_super->feature &= ~cpu_to_le32(mask)) 206 207 /* 208 * Default values for user and/or group using reserved blocks 209 */ 210 #define F2FS_DEF_RESUID 0 211 #define F2FS_DEF_RESGID 0 212 213 /* 214 * For checkpoint manager 215 */ 216 enum { 217 NAT_BITMAP, 218 SIT_BITMAP 219 }; 220 221 #define CP_UMOUNT 0x00000001 222 #define CP_FASTBOOT 0x00000002 223 #define CP_SYNC 0x00000004 224 #define CP_RECOVERY 0x00000008 225 #define CP_DISCARD 0x00000010 226 #define CP_TRIMMED 0x00000020 227 #define CP_PAUSE 0x00000040 228 #define CP_RESIZE 0x00000080 229 230 #define MAX_DISCARD_BLOCKS(sbi) BLKS_PER_SEC(sbi) 231 #define DEF_MAX_DISCARD_REQUEST 8 /* issue 8 discards per round */ 232 #define DEF_MIN_DISCARD_ISSUE_TIME 50 /* 50 ms, if exists */ 233 #define DEF_MID_DISCARD_ISSUE_TIME 500 /* 500 ms, if device busy */ 234 #define DEF_MAX_DISCARD_ISSUE_TIME 60000 /* 60 s, if no candidates */ 235 #define DEF_DISCARD_URGENT_UTIL 80 /* do more discard over 80% */ 236 #define DEF_CP_INTERVAL 60 /* 60 secs */ 237 #define DEF_IDLE_INTERVAL 5 /* 5 secs */ 238 #define DEF_DISABLE_INTERVAL 5 /* 5 secs */ 239 #define DEF_DISABLE_QUICK_INTERVAL 1 /* 1 secs */ 240 #define DEF_UMOUNT_DISCARD_TIMEOUT 5 /* 5 secs */ 241 242 struct cp_control { 243 int reason; 244 __u64 trim_start; 245 __u64 trim_end; 246 __u64 trim_minlen; 247 }; 248 249 /* 250 * indicate meta/data type 251 */ 252 enum { 253 META_CP, 254 META_NAT, 255 META_SIT, 256 META_SSA, 257 META_MAX, 258 META_POR, 259 DATA_GENERIC, /* check range only */ 260 DATA_GENERIC_ENHANCE, /* strong check on range and segment bitmap */ 261 DATA_GENERIC_ENHANCE_READ, /* 262 * strong check on range and segment 263 * bitmap but no warning due to race 264 * condition of read on truncated area 265 * by extent_cache 266 */ 267 META_GENERIC, 268 }; 269 270 /* for the list of ino */ 271 enum { 272 ORPHAN_INO, /* for orphan ino list */ 273 APPEND_INO, /* for append ino list */ 274 UPDATE_INO, /* for update ino list */ 275 TRANS_DIR_INO, /* for trasactions dir ino list */ 276 FLUSH_INO, /* for multiple device flushing */ 277 MAX_INO_ENTRY, /* max. list */ 278 }; 279 280 struct ino_entry { 281 struct list_head list; /* list head */ 282 nid_t ino; /* inode number */ 283 unsigned int dirty_device; /* dirty device bitmap */ 284 }; 285 286 /* for the list of inodes to be GCed */ 287 struct inode_entry { 288 struct list_head list; /* list head */ 289 struct inode *inode; /* vfs inode pointer */ 290 }; 291 292 struct fsync_node_entry { 293 struct list_head list; /* list head */ 294 struct page *page; /* warm node page pointer */ 295 unsigned int seq_id; /* sequence id */ 296 }; 297 298 struct ckpt_req { 299 struct completion wait; /* completion for checkpoint done */ 300 struct llist_node llnode; /* llist_node to be linked in wait queue */ 301 int ret; /* return code of checkpoint */ 302 ktime_t queue_time; /* request queued time */ 303 }; 304 305 struct ckpt_req_control { 306 struct task_struct *f2fs_issue_ckpt; /* checkpoint task */ 307 int ckpt_thread_ioprio; /* checkpoint merge thread ioprio */ 308 wait_queue_head_t ckpt_wait_queue; /* waiting queue for wake-up */ 309 atomic_t issued_ckpt; /* # of actually issued ckpts */ 310 atomic_t total_ckpt; /* # of total ckpts */ 311 atomic_t queued_ckpt; /* # of queued ckpts */ 312 struct llist_head issue_list; /* list for command issue */ 313 spinlock_t stat_lock; /* lock for below checkpoint time stats */ 314 unsigned int cur_time; /* cur wait time in msec for currently issued checkpoint */ 315 unsigned int peak_time; /* peak wait time in msec until now */ 316 }; 317 318 /* for the bitmap indicate blocks to be discarded */ 319 struct discard_entry { 320 struct list_head list; /* list head */ 321 block_t start_blkaddr; /* start blockaddr of current segment */ 322 unsigned char discard_map[SIT_VBLOCK_MAP_SIZE]; /* segment discard bitmap */ 323 }; 324 325 /* default discard granularity of inner discard thread, unit: block count */ 326 #define DEFAULT_DISCARD_GRANULARITY 16 327 328 /* max discard pend list number */ 329 #define MAX_PLIST_NUM 512 330 #define plist_idx(blk_num) ((blk_num) >= MAX_PLIST_NUM ? \ 331 (MAX_PLIST_NUM - 1) : ((blk_num) - 1)) 332 333 enum { 334 D_PREP, /* initial */ 335 D_PARTIAL, /* partially submitted */ 336 D_SUBMIT, /* all submitted */ 337 D_DONE, /* finished */ 338 }; 339 340 struct discard_info { 341 block_t lstart; /* logical start address */ 342 block_t len; /* length */ 343 block_t start; /* actual start address in dev */ 344 }; 345 346 struct discard_cmd { 347 struct rb_node rb_node; /* rb node located in rb-tree */ 348 union { 349 struct { 350 block_t lstart; /* logical start address */ 351 block_t len; /* length */ 352 block_t start; /* actual start address in dev */ 353 }; 354 struct discard_info di; /* discard info */ 355 356 }; 357 struct list_head list; /* command list */ 358 struct completion wait; /* compleation */ 359 struct block_device *bdev; /* bdev */ 360 unsigned short ref; /* reference count */ 361 unsigned char state; /* state */ 362 unsigned char queued; /* queued discard */ 363 int error; /* bio error */ 364 spinlock_t lock; /* for state/bio_ref updating */ 365 unsigned short bio_ref; /* bio reference count */ 366 }; 367 368 enum { 369 DPOLICY_BG, 370 DPOLICY_FORCE, 371 DPOLICY_FSTRIM, 372 DPOLICY_UMOUNT, 373 MAX_DPOLICY, 374 }; 375 376 struct discard_policy { 377 int type; /* type of discard */ 378 unsigned int min_interval; /* used for candidates exist */ 379 unsigned int mid_interval; /* used for device busy */ 380 unsigned int max_interval; /* used for candidates not exist */ 381 unsigned int max_requests; /* # of discards issued per round */ 382 unsigned int io_aware_gran; /* minimum granularity discard not be aware of I/O */ 383 bool io_aware; /* issue discard in idle time */ 384 bool sync; /* submit discard with REQ_SYNC flag */ 385 bool ordered; /* issue discard by lba order */ 386 bool timeout; /* discard timeout for put_super */ 387 unsigned int granularity; /* discard granularity */ 388 }; 389 390 struct discard_cmd_control { 391 struct task_struct *f2fs_issue_discard; /* discard thread */ 392 struct list_head entry_list; /* 4KB discard entry list */ 393 struct list_head pend_list[MAX_PLIST_NUM];/* store pending entries */ 394 struct list_head wait_list; /* store on-flushing entries */ 395 struct list_head fstrim_list; /* in-flight discard from fstrim */ 396 wait_queue_head_t discard_wait_queue; /* waiting queue for wake-up */ 397 unsigned int discard_wake; /* to wake up discard thread */ 398 struct mutex cmd_lock; 399 unsigned int nr_discards; /* # of discards in the list */ 400 unsigned int max_discards; /* max. discards to be issued */ 401 unsigned int max_discard_request; /* max. discard request per round */ 402 unsigned int min_discard_issue_time; /* min. interval between discard issue */ 403 unsigned int mid_discard_issue_time; /* mid. interval between discard issue */ 404 unsigned int max_discard_issue_time; /* max. interval between discard issue */ 405 unsigned int discard_granularity; /* discard granularity */ 406 unsigned int undiscard_blks; /* # of undiscard blocks */ 407 unsigned int next_pos; /* next discard position */ 408 atomic_t issued_discard; /* # of issued discard */ 409 atomic_t queued_discard; /* # of queued discard */ 410 atomic_t discard_cmd_cnt; /* # of cached cmd count */ 411 struct rb_root_cached root; /* root of discard rb-tree */ 412 bool rbtree_check; /* config for consistence check */ 413 }; 414 415 /* for the list of fsync inodes, used only during recovery */ 416 struct fsync_inode_entry { 417 struct list_head list; /* list head */ 418 struct inode *inode; /* vfs inode pointer */ 419 block_t blkaddr; /* block address locating the last fsync */ 420 block_t last_dentry; /* block address locating the last dentry */ 421 }; 422 423 #define nats_in_cursum(jnl) (le16_to_cpu((jnl)->n_nats)) 424 #define sits_in_cursum(jnl) (le16_to_cpu((jnl)->n_sits)) 425 426 #define nat_in_journal(jnl, i) ((jnl)->nat_j.entries[i].ne) 427 #define nid_in_journal(jnl, i) ((jnl)->nat_j.entries[i].nid) 428 #define sit_in_journal(jnl, i) ((jnl)->sit_j.entries[i].se) 429 #define segno_in_journal(jnl, i) ((jnl)->sit_j.entries[i].segno) 430 431 #define MAX_NAT_JENTRIES(jnl) (NAT_JOURNAL_ENTRIES - nats_in_cursum(jnl)) 432 #define MAX_SIT_JENTRIES(jnl) (SIT_JOURNAL_ENTRIES - sits_in_cursum(jnl)) 433 434 static inline int update_nats_in_cursum(struct f2fs_journal *journal, int i) 435 { 436 int before = nats_in_cursum(journal); 437 438 journal->n_nats = cpu_to_le16(before + i); 439 return before; 440 } 441 442 static inline int update_sits_in_cursum(struct f2fs_journal *journal, int i) 443 { 444 int before = sits_in_cursum(journal); 445 446 journal->n_sits = cpu_to_le16(before + i); 447 return before; 448 } 449 450 static inline bool __has_cursum_space(struct f2fs_journal *journal, 451 int size, int type) 452 { 453 if (type == NAT_JOURNAL) 454 return size <= MAX_NAT_JENTRIES(journal); 455 return size <= MAX_SIT_JENTRIES(journal); 456 } 457 458 /* for inline stuff */ 459 #define DEF_INLINE_RESERVED_SIZE 1 460 static inline int get_extra_isize(struct inode *inode); 461 static inline int get_inline_xattr_addrs(struct inode *inode); 462 #define MAX_INLINE_DATA(inode) (sizeof(__le32) * \ 463 (CUR_ADDRS_PER_INODE(inode) - \ 464 get_inline_xattr_addrs(inode) - \ 465 DEF_INLINE_RESERVED_SIZE)) 466 467 /* for inline dir */ 468 #define NR_INLINE_DENTRY(inode) (MAX_INLINE_DATA(inode) * BITS_PER_BYTE / \ 469 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \ 470 BITS_PER_BYTE + 1)) 471 #define INLINE_DENTRY_BITMAP_SIZE(inode) \ 472 DIV_ROUND_UP(NR_INLINE_DENTRY(inode), BITS_PER_BYTE) 473 #define INLINE_RESERVED_SIZE(inode) (MAX_INLINE_DATA(inode) - \ 474 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \ 475 NR_INLINE_DENTRY(inode) + \ 476 INLINE_DENTRY_BITMAP_SIZE(inode))) 477 478 /* 479 * For INODE and NODE manager 480 */ 481 /* for directory operations */ 482 483 struct f2fs_filename { 484 /* 485 * The filename the user specified. This is NULL for some 486 * filesystem-internal operations, e.g. converting an inline directory 487 * to a non-inline one, or roll-forward recovering an encrypted dentry. 488 */ 489 const struct qstr *usr_fname; 490 491 /* 492 * The on-disk filename. For encrypted directories, this is encrypted. 493 * This may be NULL for lookups in an encrypted dir without the key. 494 */ 495 struct fscrypt_str disk_name; 496 497 /* The dirhash of this filename */ 498 f2fs_hash_t hash; 499 500 #ifdef CONFIG_FS_ENCRYPTION 501 /* 502 * For lookups in encrypted directories: either the buffer backing 503 * disk_name, or a buffer that holds the decoded no-key name. 504 */ 505 struct fscrypt_str crypto_buf; 506 #endif 507 #ifdef CONFIG_UNICODE 508 /* 509 * For casefolded directories: the casefolded name, but it's left NULL 510 * if the original name is not valid Unicode, if the directory is both 511 * casefolded and encrypted and its encryption key is unavailable, or if 512 * the filesystem is doing an internal operation where usr_fname is also 513 * NULL. In all these cases we fall back to treating the name as an 514 * opaque byte sequence. 515 */ 516 struct fscrypt_str cf_name; 517 #endif 518 }; 519 520 struct f2fs_dentry_ptr { 521 struct inode *inode; 522 void *bitmap; 523 struct f2fs_dir_entry *dentry; 524 __u8 (*filename)[F2FS_SLOT_LEN]; 525 int max; 526 int nr_bitmap; 527 }; 528 529 static inline void make_dentry_ptr_block(struct inode *inode, 530 struct f2fs_dentry_ptr *d, struct f2fs_dentry_block *t) 531 { 532 d->inode = inode; 533 d->max = NR_DENTRY_IN_BLOCK; 534 d->nr_bitmap = SIZE_OF_DENTRY_BITMAP; 535 d->bitmap = t->dentry_bitmap; 536 d->dentry = t->dentry; 537 d->filename = t->filename; 538 } 539 540 static inline void make_dentry_ptr_inline(struct inode *inode, 541 struct f2fs_dentry_ptr *d, void *t) 542 { 543 int entry_cnt = NR_INLINE_DENTRY(inode); 544 int bitmap_size = INLINE_DENTRY_BITMAP_SIZE(inode); 545 int reserved_size = INLINE_RESERVED_SIZE(inode); 546 547 d->inode = inode; 548 d->max = entry_cnt; 549 d->nr_bitmap = bitmap_size; 550 d->bitmap = t; 551 d->dentry = t + bitmap_size + reserved_size; 552 d->filename = t + bitmap_size + reserved_size + 553 SIZE_OF_DIR_ENTRY * entry_cnt; 554 } 555 556 /* 557 * XATTR_NODE_OFFSET stores xattrs to one node block per file keeping -1 558 * as its node offset to distinguish from index node blocks. 559 * But some bits are used to mark the node block. 560 */ 561 #define XATTR_NODE_OFFSET ((((unsigned int)-1) << OFFSET_BIT_SHIFT) \ 562 >> OFFSET_BIT_SHIFT) 563 enum { 564 ALLOC_NODE, /* allocate a new node page if needed */ 565 LOOKUP_NODE, /* look up a node without readahead */ 566 LOOKUP_NODE_RA, /* 567 * look up a node with readahead called 568 * by get_data_block. 569 */ 570 }; 571 572 #define DEFAULT_RETRY_IO_COUNT 8 /* maximum retry read IO or flush count */ 573 574 /* congestion wait timeout value, default: 20ms */ 575 #define DEFAULT_IO_TIMEOUT (msecs_to_jiffies(20)) 576 577 /* maximum retry quota flush count */ 578 #define DEFAULT_RETRY_QUOTA_FLUSH_COUNT 8 579 580 #define F2FS_LINK_MAX 0xffffffff /* maximum link count per file */ 581 582 #define MAX_DIR_RA_PAGES 4 /* maximum ra pages of dir */ 583 584 /* dirty segments threshold for triggering CP */ 585 #define DEFAULT_DIRTY_THRESHOLD 4 586 587 /* for in-memory extent cache entry */ 588 #define F2FS_MIN_EXTENT_LEN 64 /* minimum extent length */ 589 590 /* number of extent info in extent cache we try to shrink */ 591 #define EXTENT_CACHE_SHRINK_NUMBER 128 592 593 #define RECOVERY_MAX_RA_BLOCKS BIO_MAX_VECS 594 #define RECOVERY_MIN_RA_BLOCKS 1 595 596 struct rb_entry { 597 struct rb_node rb_node; /* rb node located in rb-tree */ 598 union { 599 struct { 600 unsigned int ofs; /* start offset of the entry */ 601 unsigned int len; /* length of the entry */ 602 }; 603 unsigned long long key; /* 64-bits key */ 604 } __packed; 605 }; 606 607 struct extent_info { 608 unsigned int fofs; /* start offset in a file */ 609 unsigned int len; /* length of the extent */ 610 u32 blk; /* start block address of the extent */ 611 #ifdef CONFIG_F2FS_FS_COMPRESSION 612 unsigned int c_len; /* physical extent length of compressed blocks */ 613 #endif 614 }; 615 616 struct extent_node { 617 struct rb_node rb_node; /* rb node located in rb-tree */ 618 struct extent_info ei; /* extent info */ 619 struct list_head list; /* node in global extent list of sbi */ 620 struct extent_tree *et; /* extent tree pointer */ 621 }; 622 623 struct extent_tree { 624 nid_t ino; /* inode number */ 625 struct rb_root_cached root; /* root of extent info rb-tree */ 626 struct extent_node *cached_en; /* recently accessed extent node */ 627 struct extent_info largest; /* largested extent info */ 628 struct list_head list; /* to be used by sbi->zombie_list */ 629 rwlock_t lock; /* protect extent info rb-tree */ 630 atomic_t node_cnt; /* # of extent node in rb-tree*/ 631 bool largest_updated; /* largest extent updated */ 632 }; 633 634 /* 635 * This structure is taken from ext4_map_blocks. 636 * 637 * Note that, however, f2fs uses NEW and MAPPED flags for f2fs_map_blocks(). 638 */ 639 #define F2FS_MAP_NEW (1 << BH_New) 640 #define F2FS_MAP_MAPPED (1 << BH_Mapped) 641 #define F2FS_MAP_UNWRITTEN (1 << BH_Unwritten) 642 #define F2FS_MAP_FLAGS (F2FS_MAP_NEW | F2FS_MAP_MAPPED |\ 643 F2FS_MAP_UNWRITTEN) 644 645 struct f2fs_map_blocks { 646 struct block_device *m_bdev; /* for multi-device dio */ 647 block_t m_pblk; 648 block_t m_lblk; 649 unsigned int m_len; 650 unsigned int m_flags; 651 pgoff_t *m_next_pgofs; /* point next possible non-hole pgofs */ 652 pgoff_t *m_next_extent; /* point to next possible extent */ 653 int m_seg_type; 654 bool m_may_create; /* indicate it is from write path */ 655 bool m_multidev_dio; /* indicate it allows multi-device dio */ 656 }; 657 658 /* for flag in get_data_block */ 659 enum { 660 F2FS_GET_BLOCK_DEFAULT, 661 F2FS_GET_BLOCK_FIEMAP, 662 F2FS_GET_BLOCK_BMAP, 663 F2FS_GET_BLOCK_DIO, 664 F2FS_GET_BLOCK_PRE_DIO, 665 F2FS_GET_BLOCK_PRE_AIO, 666 F2FS_GET_BLOCK_PRECACHE, 667 }; 668 669 /* 670 * i_advise uses FADVISE_XXX_BIT. We can add additional hints later. 671 */ 672 #define FADVISE_COLD_BIT 0x01 673 #define FADVISE_LOST_PINO_BIT 0x02 674 #define FADVISE_ENCRYPT_BIT 0x04 675 #define FADVISE_ENC_NAME_BIT 0x08 676 #define FADVISE_KEEP_SIZE_BIT 0x10 677 #define FADVISE_HOT_BIT 0x20 678 #define FADVISE_VERITY_BIT 0x40 679 #define FADVISE_TRUNC_BIT 0x80 680 681 #define FADVISE_MODIFIABLE_BITS (FADVISE_COLD_BIT | FADVISE_HOT_BIT) 682 683 #define file_is_cold(inode) is_file(inode, FADVISE_COLD_BIT) 684 #define file_set_cold(inode) set_file(inode, FADVISE_COLD_BIT) 685 #define file_clear_cold(inode) clear_file(inode, FADVISE_COLD_BIT) 686 687 #define file_wrong_pino(inode) is_file(inode, FADVISE_LOST_PINO_BIT) 688 #define file_lost_pino(inode) set_file(inode, FADVISE_LOST_PINO_BIT) 689 #define file_got_pino(inode) clear_file(inode, FADVISE_LOST_PINO_BIT) 690 691 #define file_is_encrypt(inode) is_file(inode, FADVISE_ENCRYPT_BIT) 692 #define file_set_encrypt(inode) set_file(inode, FADVISE_ENCRYPT_BIT) 693 694 #define file_enc_name(inode) is_file(inode, FADVISE_ENC_NAME_BIT) 695 #define file_set_enc_name(inode) set_file(inode, FADVISE_ENC_NAME_BIT) 696 697 #define file_keep_isize(inode) is_file(inode, FADVISE_KEEP_SIZE_BIT) 698 #define file_set_keep_isize(inode) set_file(inode, FADVISE_KEEP_SIZE_BIT) 699 700 #define file_is_hot(inode) is_file(inode, FADVISE_HOT_BIT) 701 #define file_set_hot(inode) set_file(inode, FADVISE_HOT_BIT) 702 #define file_clear_hot(inode) clear_file(inode, FADVISE_HOT_BIT) 703 704 #define file_is_verity(inode) is_file(inode, FADVISE_VERITY_BIT) 705 #define file_set_verity(inode) set_file(inode, FADVISE_VERITY_BIT) 706 707 #define file_should_truncate(inode) is_file(inode, FADVISE_TRUNC_BIT) 708 #define file_need_truncate(inode) set_file(inode, FADVISE_TRUNC_BIT) 709 #define file_dont_truncate(inode) clear_file(inode, FADVISE_TRUNC_BIT) 710 711 #define DEF_DIR_LEVEL 0 712 713 enum { 714 GC_FAILURE_PIN, 715 GC_FAILURE_ATOMIC, 716 MAX_GC_FAILURE 717 }; 718 719 /* used for f2fs_inode_info->flags */ 720 enum { 721 FI_NEW_INODE, /* indicate newly allocated inode */ 722 FI_DIRTY_INODE, /* indicate inode is dirty or not */ 723 FI_AUTO_RECOVER, /* indicate inode is recoverable */ 724 FI_DIRTY_DIR, /* indicate directory has dirty pages */ 725 FI_INC_LINK, /* need to increment i_nlink */ 726 FI_ACL_MODE, /* indicate acl mode */ 727 FI_NO_ALLOC, /* should not allocate any blocks */ 728 FI_FREE_NID, /* free allocated nide */ 729 FI_NO_EXTENT, /* not to use the extent cache */ 730 FI_INLINE_XATTR, /* used for inline xattr */ 731 FI_INLINE_DATA, /* used for inline data*/ 732 FI_INLINE_DENTRY, /* used for inline dentry */ 733 FI_APPEND_WRITE, /* inode has appended data */ 734 FI_UPDATE_WRITE, /* inode has in-place-update data */ 735 FI_NEED_IPU, /* used for ipu per file */ 736 FI_ATOMIC_FILE, /* indicate atomic file */ 737 FI_ATOMIC_COMMIT, /* indicate the state of atomical committing */ 738 FI_VOLATILE_FILE, /* indicate volatile file */ 739 FI_FIRST_BLOCK_WRITTEN, /* indicate #0 data block was written */ 740 FI_DROP_CACHE, /* drop dirty page cache */ 741 FI_DATA_EXIST, /* indicate data exists */ 742 FI_INLINE_DOTS, /* indicate inline dot dentries */ 743 FI_SKIP_WRITES, /* should skip data page writeback */ 744 FI_OPU_WRITE, /* used for opu per file */ 745 FI_DIRTY_FILE, /* indicate regular/symlink has dirty pages */ 746 FI_PREALLOCATED_ALL, /* all blocks for write were preallocated */ 747 FI_HOT_DATA, /* indicate file is hot */ 748 FI_EXTRA_ATTR, /* indicate file has extra attribute */ 749 FI_PROJ_INHERIT, /* indicate file inherits projectid */ 750 FI_PIN_FILE, /* indicate file should not be gced */ 751 FI_ATOMIC_REVOKE_REQUEST, /* request to drop atomic data */ 752 FI_VERITY_IN_PROGRESS, /* building fs-verity Merkle tree */ 753 FI_COMPRESSED_FILE, /* indicate file's data can be compressed */ 754 FI_COMPRESS_CORRUPT, /* indicate compressed cluster is corrupted */ 755 FI_MMAP_FILE, /* indicate file was mmapped */ 756 FI_ENABLE_COMPRESS, /* enable compression in "user" compression mode */ 757 FI_COMPRESS_RELEASED, /* compressed blocks were released */ 758 FI_ALIGNED_WRITE, /* enable aligned write */ 759 FI_MAX, /* max flag, never be used */ 760 }; 761 762 struct f2fs_inode_info { 763 struct inode vfs_inode; /* serve a vfs inode */ 764 unsigned long i_flags; /* keep an inode flags for ioctl */ 765 unsigned char i_advise; /* use to give file attribute hints */ 766 unsigned char i_dir_level; /* use for dentry level for large dir */ 767 unsigned int i_current_depth; /* only for directory depth */ 768 /* for gc failure statistic */ 769 unsigned int i_gc_failures[MAX_GC_FAILURE]; 770 unsigned int i_pino; /* parent inode number */ 771 umode_t i_acl_mode; /* keep file acl mode temporarily */ 772 773 /* Use below internally in f2fs*/ 774 unsigned long flags[BITS_TO_LONGS(FI_MAX)]; /* use to pass per-file flags */ 775 struct f2fs_rwsem i_sem; /* protect fi info */ 776 atomic_t dirty_pages; /* # of dirty pages */ 777 f2fs_hash_t chash; /* hash value of given file name */ 778 unsigned int clevel; /* maximum level of given file name */ 779 struct task_struct *task; /* lookup and create consistency */ 780 struct task_struct *cp_task; /* separate cp/wb IO stats*/ 781 nid_t i_xattr_nid; /* node id that contains xattrs */ 782 loff_t last_disk_size; /* lastly written file size */ 783 spinlock_t i_size_lock; /* protect last_disk_size */ 784 785 #ifdef CONFIG_QUOTA 786 struct dquot *i_dquot[MAXQUOTAS]; 787 788 /* quota space reservation, managed internally by quota code */ 789 qsize_t i_reserved_quota; 790 #endif 791 struct list_head dirty_list; /* dirty list for dirs and files */ 792 struct list_head gdirty_list; /* linked in global dirty list */ 793 struct list_head inmem_ilist; /* list for inmem inodes */ 794 struct list_head inmem_pages; /* inmemory pages managed by f2fs */ 795 struct task_struct *inmem_task; /* store inmemory task */ 796 struct mutex inmem_lock; /* lock for inmemory pages */ 797 struct extent_tree *extent_tree; /* cached extent_tree entry */ 798 799 /* avoid racing between foreground op and gc */ 800 struct f2fs_rwsem i_gc_rwsem[2]; 801 struct f2fs_rwsem i_xattr_sem; /* avoid racing between reading and changing EAs */ 802 803 int i_extra_isize; /* size of extra space located in i_addr */ 804 kprojid_t i_projid; /* id for project quota */ 805 int i_inline_xattr_size; /* inline xattr size */ 806 struct timespec64 i_crtime; /* inode creation time */ 807 struct timespec64 i_disk_time[4];/* inode disk times */ 808 809 /* for file compress */ 810 atomic_t i_compr_blocks; /* # of compressed blocks */ 811 unsigned char i_compress_algorithm; /* algorithm type */ 812 unsigned char i_log_cluster_size; /* log of cluster size */ 813 unsigned char i_compress_level; /* compress level (lz4hc,zstd) */ 814 unsigned short i_compress_flag; /* compress flag */ 815 unsigned int i_cluster_size; /* cluster size */ 816 }; 817 818 static inline void get_extent_info(struct extent_info *ext, 819 struct f2fs_extent *i_ext) 820 { 821 ext->fofs = le32_to_cpu(i_ext->fofs); 822 ext->blk = le32_to_cpu(i_ext->blk); 823 ext->len = le32_to_cpu(i_ext->len); 824 } 825 826 static inline void set_raw_extent(struct extent_info *ext, 827 struct f2fs_extent *i_ext) 828 { 829 i_ext->fofs = cpu_to_le32(ext->fofs); 830 i_ext->blk = cpu_to_le32(ext->blk); 831 i_ext->len = cpu_to_le32(ext->len); 832 } 833 834 static inline void set_extent_info(struct extent_info *ei, unsigned int fofs, 835 u32 blk, unsigned int len) 836 { 837 ei->fofs = fofs; 838 ei->blk = blk; 839 ei->len = len; 840 #ifdef CONFIG_F2FS_FS_COMPRESSION 841 ei->c_len = 0; 842 #endif 843 } 844 845 static inline bool __is_discard_mergeable(struct discard_info *back, 846 struct discard_info *front, unsigned int max_len) 847 { 848 return (back->lstart + back->len == front->lstart) && 849 (back->len + front->len <= max_len); 850 } 851 852 static inline bool __is_discard_back_mergeable(struct discard_info *cur, 853 struct discard_info *back, unsigned int max_len) 854 { 855 return __is_discard_mergeable(back, cur, max_len); 856 } 857 858 static inline bool __is_discard_front_mergeable(struct discard_info *cur, 859 struct discard_info *front, unsigned int max_len) 860 { 861 return __is_discard_mergeable(cur, front, max_len); 862 } 863 864 static inline bool __is_extent_mergeable(struct extent_info *back, 865 struct extent_info *front) 866 { 867 #ifdef CONFIG_F2FS_FS_COMPRESSION 868 if (back->c_len && back->len != back->c_len) 869 return false; 870 if (front->c_len && front->len != front->c_len) 871 return false; 872 #endif 873 return (back->fofs + back->len == front->fofs && 874 back->blk + back->len == front->blk); 875 } 876 877 static inline bool __is_back_mergeable(struct extent_info *cur, 878 struct extent_info *back) 879 { 880 return __is_extent_mergeable(back, cur); 881 } 882 883 static inline bool __is_front_mergeable(struct extent_info *cur, 884 struct extent_info *front) 885 { 886 return __is_extent_mergeable(cur, front); 887 } 888 889 extern void f2fs_mark_inode_dirty_sync(struct inode *inode, bool sync); 890 static inline void __try_update_largest_extent(struct extent_tree *et, 891 struct extent_node *en) 892 { 893 if (en->ei.len > et->largest.len) { 894 et->largest = en->ei; 895 et->largest_updated = true; 896 } 897 } 898 899 /* 900 * For free nid management 901 */ 902 enum nid_state { 903 FREE_NID, /* newly added to free nid list */ 904 PREALLOC_NID, /* it is preallocated */ 905 MAX_NID_STATE, 906 }; 907 908 enum nat_state { 909 TOTAL_NAT, 910 DIRTY_NAT, 911 RECLAIMABLE_NAT, 912 MAX_NAT_STATE, 913 }; 914 915 struct f2fs_nm_info { 916 block_t nat_blkaddr; /* base disk address of NAT */ 917 nid_t max_nid; /* maximum possible node ids */ 918 nid_t available_nids; /* # of available node ids */ 919 nid_t next_scan_nid; /* the next nid to be scanned */ 920 nid_t max_rf_node_blocks; /* max # of nodes for recovery */ 921 unsigned int ram_thresh; /* control the memory footprint */ 922 unsigned int ra_nid_pages; /* # of nid pages to be readaheaded */ 923 unsigned int dirty_nats_ratio; /* control dirty nats ratio threshold */ 924 925 /* NAT cache management */ 926 struct radix_tree_root nat_root;/* root of the nat entry cache */ 927 struct radix_tree_root nat_set_root;/* root of the nat set cache */ 928 struct f2fs_rwsem nat_tree_lock; /* protect nat entry tree */ 929 struct list_head nat_entries; /* cached nat entry list (clean) */ 930 spinlock_t nat_list_lock; /* protect clean nat entry list */ 931 unsigned int nat_cnt[MAX_NAT_STATE]; /* the # of cached nat entries */ 932 unsigned int nat_blocks; /* # of nat blocks */ 933 934 /* free node ids management */ 935 struct radix_tree_root free_nid_root;/* root of the free_nid cache */ 936 struct list_head free_nid_list; /* list for free nids excluding preallocated nids */ 937 unsigned int nid_cnt[MAX_NID_STATE]; /* the number of free node id */ 938 spinlock_t nid_list_lock; /* protect nid lists ops */ 939 struct mutex build_lock; /* lock for build free nids */ 940 unsigned char **free_nid_bitmap; 941 unsigned char *nat_block_bitmap; 942 unsigned short *free_nid_count; /* free nid count of NAT block */ 943 944 /* for checkpoint */ 945 char *nat_bitmap; /* NAT bitmap pointer */ 946 947 unsigned int nat_bits_blocks; /* # of nat bits blocks */ 948 unsigned char *nat_bits; /* NAT bits blocks */ 949 unsigned char *full_nat_bits; /* full NAT pages */ 950 unsigned char *empty_nat_bits; /* empty NAT pages */ 951 #ifdef CONFIG_F2FS_CHECK_FS 952 char *nat_bitmap_mir; /* NAT bitmap mirror */ 953 #endif 954 int bitmap_size; /* bitmap size */ 955 }; 956 957 /* 958 * this structure is used as one of function parameters. 959 * all the information are dedicated to a given direct node block determined 960 * by the data offset in a file. 961 */ 962 struct dnode_of_data { 963 struct inode *inode; /* vfs inode pointer */ 964 struct page *inode_page; /* its inode page, NULL is possible */ 965 struct page *node_page; /* cached direct node page */ 966 nid_t nid; /* node id of the direct node block */ 967 unsigned int ofs_in_node; /* data offset in the node page */ 968 bool inode_page_locked; /* inode page is locked or not */ 969 bool node_changed; /* is node block changed */ 970 char cur_level; /* level of hole node page */ 971 char max_level; /* level of current page located */ 972 block_t data_blkaddr; /* block address of the node block */ 973 }; 974 975 static inline void set_new_dnode(struct dnode_of_data *dn, struct inode *inode, 976 struct page *ipage, struct page *npage, nid_t nid) 977 { 978 memset(dn, 0, sizeof(*dn)); 979 dn->inode = inode; 980 dn->inode_page = ipage; 981 dn->node_page = npage; 982 dn->nid = nid; 983 } 984 985 /* 986 * For SIT manager 987 * 988 * By default, there are 6 active log areas across the whole main area. 989 * When considering hot and cold data separation to reduce cleaning overhead, 990 * we split 3 for data logs and 3 for node logs as hot, warm, and cold types, 991 * respectively. 992 * In the current design, you should not change the numbers intentionally. 993 * Instead, as a mount option such as active_logs=x, you can use 2, 4, and 6 994 * logs individually according to the underlying devices. (default: 6) 995 * Just in case, on-disk layout covers maximum 16 logs that consist of 8 for 996 * data and 8 for node logs. 997 */ 998 #define NR_CURSEG_DATA_TYPE (3) 999 #define NR_CURSEG_NODE_TYPE (3) 1000 #define NR_CURSEG_INMEM_TYPE (2) 1001 #define NR_CURSEG_RO_TYPE (2) 1002 #define NR_CURSEG_PERSIST_TYPE (NR_CURSEG_DATA_TYPE + NR_CURSEG_NODE_TYPE) 1003 #define NR_CURSEG_TYPE (NR_CURSEG_INMEM_TYPE + NR_CURSEG_PERSIST_TYPE) 1004 1005 enum { 1006 CURSEG_HOT_DATA = 0, /* directory entry blocks */ 1007 CURSEG_WARM_DATA, /* data blocks */ 1008 CURSEG_COLD_DATA, /* multimedia or GCed data blocks */ 1009 CURSEG_HOT_NODE, /* direct node blocks of directory files */ 1010 CURSEG_WARM_NODE, /* direct node blocks of normal files */ 1011 CURSEG_COLD_NODE, /* indirect node blocks */ 1012 NR_PERSISTENT_LOG, /* number of persistent log */ 1013 CURSEG_COLD_DATA_PINNED = NR_PERSISTENT_LOG, 1014 /* pinned file that needs consecutive block address */ 1015 CURSEG_ALL_DATA_ATGC, /* SSR alloctor in hot/warm/cold data area */ 1016 NO_CHECK_TYPE, /* number of persistent & inmem log */ 1017 }; 1018 1019 struct flush_cmd { 1020 struct completion wait; 1021 struct llist_node llnode; 1022 nid_t ino; 1023 int ret; 1024 }; 1025 1026 struct flush_cmd_control { 1027 struct task_struct *f2fs_issue_flush; /* flush thread */ 1028 wait_queue_head_t flush_wait_queue; /* waiting queue for wake-up */ 1029 atomic_t issued_flush; /* # of issued flushes */ 1030 atomic_t queued_flush; /* # of queued flushes */ 1031 struct llist_head issue_list; /* list for command issue */ 1032 struct llist_node *dispatch_list; /* list for command dispatch */ 1033 }; 1034 1035 struct f2fs_sm_info { 1036 struct sit_info *sit_info; /* whole segment information */ 1037 struct free_segmap_info *free_info; /* free segment information */ 1038 struct dirty_seglist_info *dirty_info; /* dirty segment information */ 1039 struct curseg_info *curseg_array; /* active segment information */ 1040 1041 struct f2fs_rwsem curseg_lock; /* for preventing curseg change */ 1042 1043 block_t seg0_blkaddr; /* block address of 0'th segment */ 1044 block_t main_blkaddr; /* start block address of main area */ 1045 block_t ssa_blkaddr; /* start block address of SSA area */ 1046 1047 unsigned int segment_count; /* total # of segments */ 1048 unsigned int main_segments; /* # of segments in main area */ 1049 unsigned int reserved_segments; /* # of reserved segments */ 1050 unsigned int additional_reserved_segments;/* reserved segs for IO align feature */ 1051 unsigned int ovp_segments; /* # of overprovision segments */ 1052 1053 /* a threshold to reclaim prefree segments */ 1054 unsigned int rec_prefree_segments; 1055 1056 /* for batched trimming */ 1057 unsigned int trim_sections; /* # of sections to trim */ 1058 1059 struct list_head sit_entry_set; /* sit entry set list */ 1060 1061 unsigned int ipu_policy; /* in-place-update policy */ 1062 unsigned int min_ipu_util; /* in-place-update threshold */ 1063 unsigned int min_fsync_blocks; /* threshold for fsync */ 1064 unsigned int min_seq_blocks; /* threshold for sequential blocks */ 1065 unsigned int min_hot_blocks; /* threshold for hot block allocation */ 1066 unsigned int min_ssr_sections; /* threshold to trigger SSR allocation */ 1067 1068 /* for flush command control */ 1069 struct flush_cmd_control *fcc_info; 1070 1071 /* for discard command control */ 1072 struct discard_cmd_control *dcc_info; 1073 }; 1074 1075 /* 1076 * For superblock 1077 */ 1078 /* 1079 * COUNT_TYPE for monitoring 1080 * 1081 * f2fs monitors the number of several block types such as on-writeback, 1082 * dirty dentry blocks, dirty node blocks, and dirty meta blocks. 1083 */ 1084 #define WB_DATA_TYPE(p) (__is_cp_guaranteed(p) ? F2FS_WB_CP_DATA : F2FS_WB_DATA) 1085 enum count_type { 1086 F2FS_DIRTY_DENTS, 1087 F2FS_DIRTY_DATA, 1088 F2FS_DIRTY_QDATA, 1089 F2FS_DIRTY_NODES, 1090 F2FS_DIRTY_META, 1091 F2FS_INMEM_PAGES, 1092 F2FS_DIRTY_IMETA, 1093 F2FS_WB_CP_DATA, 1094 F2FS_WB_DATA, 1095 F2FS_RD_DATA, 1096 F2FS_RD_NODE, 1097 F2FS_RD_META, 1098 F2FS_DIO_WRITE, 1099 F2FS_DIO_READ, 1100 NR_COUNT_TYPE, 1101 }; 1102 1103 /* 1104 * The below are the page types of bios used in submit_bio(). 1105 * The available types are: 1106 * DATA User data pages. It operates as async mode. 1107 * NODE Node pages. It operates as async mode. 1108 * META FS metadata pages such as SIT, NAT, CP. 1109 * NR_PAGE_TYPE The number of page types. 1110 * META_FLUSH Make sure the previous pages are written 1111 * with waiting the bio's completion 1112 * ... Only can be used with META. 1113 */ 1114 #define PAGE_TYPE_OF_BIO(type) ((type) > META ? META : (type)) 1115 enum page_type { 1116 DATA, 1117 NODE, 1118 META, 1119 NR_PAGE_TYPE, 1120 META_FLUSH, 1121 INMEM, /* the below types are used by tracepoints only. */ 1122 INMEM_DROP, 1123 INMEM_INVALIDATE, 1124 INMEM_REVOKE, 1125 IPU, 1126 OPU, 1127 }; 1128 1129 enum temp_type { 1130 HOT = 0, /* must be zero for meta bio */ 1131 WARM, 1132 COLD, 1133 NR_TEMP_TYPE, 1134 }; 1135 1136 enum need_lock_type { 1137 LOCK_REQ = 0, 1138 LOCK_DONE, 1139 LOCK_RETRY, 1140 }; 1141 1142 enum cp_reason_type { 1143 CP_NO_NEEDED, 1144 CP_NON_REGULAR, 1145 CP_COMPRESSED, 1146 CP_HARDLINK, 1147 CP_SB_NEED_CP, 1148 CP_WRONG_PINO, 1149 CP_NO_SPC_ROLL, 1150 CP_NODE_NEED_CP, 1151 CP_FASTBOOT_MODE, 1152 CP_SPEC_LOG_NUM, 1153 CP_RECOVER_DIR, 1154 }; 1155 1156 enum iostat_type { 1157 /* WRITE IO */ 1158 APP_DIRECT_IO, /* app direct write IOs */ 1159 APP_BUFFERED_IO, /* app buffered write IOs */ 1160 APP_WRITE_IO, /* app write IOs */ 1161 APP_MAPPED_IO, /* app mapped IOs */ 1162 FS_DATA_IO, /* data IOs from kworker/fsync/reclaimer */ 1163 FS_NODE_IO, /* node IOs from kworker/fsync/reclaimer */ 1164 FS_META_IO, /* meta IOs from kworker/reclaimer */ 1165 FS_GC_DATA_IO, /* data IOs from forground gc */ 1166 FS_GC_NODE_IO, /* node IOs from forground gc */ 1167 FS_CP_DATA_IO, /* data IOs from checkpoint */ 1168 FS_CP_NODE_IO, /* node IOs from checkpoint */ 1169 FS_CP_META_IO, /* meta IOs from checkpoint */ 1170 1171 /* READ IO */ 1172 APP_DIRECT_READ_IO, /* app direct read IOs */ 1173 APP_BUFFERED_READ_IO, /* app buffered read IOs */ 1174 APP_READ_IO, /* app read IOs */ 1175 APP_MAPPED_READ_IO, /* app mapped read IOs */ 1176 FS_DATA_READ_IO, /* data read IOs */ 1177 FS_GDATA_READ_IO, /* data read IOs from background gc */ 1178 FS_CDATA_READ_IO, /* compressed data read IOs */ 1179 FS_NODE_READ_IO, /* node read IOs */ 1180 FS_META_READ_IO, /* meta read IOs */ 1181 1182 /* other */ 1183 FS_DISCARD, /* discard */ 1184 NR_IO_TYPE, 1185 }; 1186 1187 struct f2fs_io_info { 1188 struct f2fs_sb_info *sbi; /* f2fs_sb_info pointer */ 1189 nid_t ino; /* inode number */ 1190 enum page_type type; /* contains DATA/NODE/META/META_FLUSH */ 1191 enum temp_type temp; /* contains HOT/WARM/COLD */ 1192 int op; /* contains REQ_OP_ */ 1193 int op_flags; /* req_flag_bits */ 1194 block_t new_blkaddr; /* new block address to be written */ 1195 block_t old_blkaddr; /* old block address before Cow */ 1196 struct page *page; /* page to be written */ 1197 struct page *encrypted_page; /* encrypted page */ 1198 struct page *compressed_page; /* compressed page */ 1199 struct list_head list; /* serialize IOs */ 1200 bool submitted; /* indicate IO submission */ 1201 int need_lock; /* indicate we need to lock cp_rwsem */ 1202 bool in_list; /* indicate fio is in io_list */ 1203 bool is_por; /* indicate IO is from recovery or not */ 1204 bool retry; /* need to reallocate block address */ 1205 int compr_blocks; /* # of compressed block addresses */ 1206 bool encrypted; /* indicate file is encrypted */ 1207 enum iostat_type io_type; /* io type */ 1208 struct writeback_control *io_wbc; /* writeback control */ 1209 struct bio **bio; /* bio for ipu */ 1210 sector_t *last_block; /* last block number in bio */ 1211 unsigned char version; /* version of the node */ 1212 }; 1213 1214 struct bio_entry { 1215 struct bio *bio; 1216 struct list_head list; 1217 }; 1218 1219 #define is_read_io(rw) ((rw) == READ) 1220 struct f2fs_bio_info { 1221 struct f2fs_sb_info *sbi; /* f2fs superblock */ 1222 struct bio *bio; /* bios to merge */ 1223 sector_t last_block_in_bio; /* last block number */ 1224 struct f2fs_io_info fio; /* store buffered io info. */ 1225 struct f2fs_rwsem io_rwsem; /* blocking op for bio */ 1226 spinlock_t io_lock; /* serialize DATA/NODE IOs */ 1227 struct list_head io_list; /* track fios */ 1228 struct list_head bio_list; /* bio entry list head */ 1229 struct f2fs_rwsem bio_list_lock; /* lock to protect bio entry list */ 1230 }; 1231 1232 #define FDEV(i) (sbi->devs[i]) 1233 #define RDEV(i) (raw_super->devs[i]) 1234 struct f2fs_dev_info { 1235 struct block_device *bdev; 1236 char path[MAX_PATH_LEN]; 1237 unsigned int total_segments; 1238 block_t start_blk; 1239 block_t end_blk; 1240 #ifdef CONFIG_BLK_DEV_ZONED 1241 unsigned int nr_blkz; /* Total number of zones */ 1242 unsigned long *blkz_seq; /* Bitmap indicating sequential zones */ 1243 block_t *zone_capacity_blocks; /* Array of zone capacity in blks */ 1244 #endif 1245 }; 1246 1247 enum inode_type { 1248 DIR_INODE, /* for dirty dir inode */ 1249 FILE_INODE, /* for dirty regular/symlink inode */ 1250 DIRTY_META, /* for all dirtied inode metadata */ 1251 ATOMIC_FILE, /* for all atomic files */ 1252 NR_INODE_TYPE, 1253 }; 1254 1255 /* for inner inode cache management */ 1256 struct inode_management { 1257 struct radix_tree_root ino_root; /* ino entry array */ 1258 spinlock_t ino_lock; /* for ino entry lock */ 1259 struct list_head ino_list; /* inode list head */ 1260 unsigned long ino_num; /* number of entries */ 1261 }; 1262 1263 /* for GC_AT */ 1264 struct atgc_management { 1265 bool atgc_enabled; /* ATGC is enabled or not */ 1266 struct rb_root_cached root; /* root of victim rb-tree */ 1267 struct list_head victim_list; /* linked with all victim entries */ 1268 unsigned int victim_count; /* victim count in rb-tree */ 1269 unsigned int candidate_ratio; /* candidate ratio */ 1270 unsigned int max_candidate_count; /* max candidate count */ 1271 unsigned int age_weight; /* age weight, vblock_weight = 100 - age_weight */ 1272 unsigned long long age_threshold; /* age threshold */ 1273 }; 1274 1275 /* For s_flag in struct f2fs_sb_info */ 1276 enum { 1277 SBI_IS_DIRTY, /* dirty flag for checkpoint */ 1278 SBI_IS_CLOSE, /* specify unmounting */ 1279 SBI_NEED_FSCK, /* need fsck.f2fs to fix */ 1280 SBI_POR_DOING, /* recovery is doing or not */ 1281 SBI_NEED_SB_WRITE, /* need to recover superblock */ 1282 SBI_NEED_CP, /* need to checkpoint */ 1283 SBI_IS_SHUTDOWN, /* shutdown by ioctl */ 1284 SBI_IS_RECOVERED, /* recovered orphan/data */ 1285 SBI_CP_DISABLED, /* CP was disabled last mount */ 1286 SBI_CP_DISABLED_QUICK, /* CP was disabled quickly */ 1287 SBI_QUOTA_NEED_FLUSH, /* need to flush quota info in CP */ 1288 SBI_QUOTA_SKIP_FLUSH, /* skip flushing quota in current CP */ 1289 SBI_QUOTA_NEED_REPAIR, /* quota file may be corrupted */ 1290 SBI_IS_RESIZEFS, /* resizefs is in process */ 1291 }; 1292 1293 enum { 1294 CP_TIME, 1295 REQ_TIME, 1296 DISCARD_TIME, 1297 GC_TIME, 1298 DISABLE_TIME, 1299 UMOUNT_DISCARD_TIMEOUT, 1300 MAX_TIME, 1301 }; 1302 1303 enum { 1304 GC_NORMAL, 1305 GC_IDLE_CB, 1306 GC_IDLE_GREEDY, 1307 GC_IDLE_AT, 1308 GC_URGENT_HIGH, 1309 GC_URGENT_LOW, 1310 MAX_GC_MODE, 1311 }; 1312 1313 enum { 1314 BGGC_MODE_ON, /* background gc is on */ 1315 BGGC_MODE_OFF, /* background gc is off */ 1316 BGGC_MODE_SYNC, /* 1317 * background gc is on, migrating blocks 1318 * like foreground gc 1319 */ 1320 }; 1321 1322 enum { 1323 FS_MODE_ADAPTIVE, /* use both lfs/ssr allocation */ 1324 FS_MODE_LFS, /* use lfs allocation only */ 1325 FS_MODE_FRAGMENT_SEG, /* segment fragmentation mode */ 1326 FS_MODE_FRAGMENT_BLK, /* block fragmentation mode */ 1327 }; 1328 1329 enum { 1330 WHINT_MODE_OFF, /* not pass down write hints */ 1331 WHINT_MODE_USER, /* try to pass down hints given by users */ 1332 WHINT_MODE_FS, /* pass down hints with F2FS policy */ 1333 }; 1334 1335 enum { 1336 ALLOC_MODE_DEFAULT, /* stay default */ 1337 ALLOC_MODE_REUSE, /* reuse segments as much as possible */ 1338 }; 1339 1340 enum fsync_mode { 1341 FSYNC_MODE_POSIX, /* fsync follows posix semantics */ 1342 FSYNC_MODE_STRICT, /* fsync behaves in line with ext4 */ 1343 FSYNC_MODE_NOBARRIER, /* fsync behaves nobarrier based on posix */ 1344 }; 1345 1346 enum { 1347 COMPR_MODE_FS, /* 1348 * automatically compress compression 1349 * enabled files 1350 */ 1351 COMPR_MODE_USER, /* 1352 * automatical compression is disabled. 1353 * user can control the file compression 1354 * using ioctls 1355 */ 1356 }; 1357 1358 enum { 1359 DISCARD_UNIT_BLOCK, /* basic discard unit is block */ 1360 DISCARD_UNIT_SEGMENT, /* basic discard unit is segment */ 1361 DISCARD_UNIT_SECTION, /* basic discard unit is section */ 1362 }; 1363 1364 static inline int f2fs_test_bit(unsigned int nr, char *addr); 1365 static inline void f2fs_set_bit(unsigned int nr, char *addr); 1366 static inline void f2fs_clear_bit(unsigned int nr, char *addr); 1367 1368 /* 1369 * Layout of f2fs page.private: 1370 * 1371 * Layout A: lowest bit should be 1 1372 * | bit0 = 1 | bit1 | bit2 | ... | bit MAX | private data .... | 1373 * bit 0 PAGE_PRIVATE_NOT_POINTER 1374 * bit 1 PAGE_PRIVATE_ATOMIC_WRITE 1375 * bit 2 PAGE_PRIVATE_DUMMY_WRITE 1376 * bit 3 PAGE_PRIVATE_ONGOING_MIGRATION 1377 * bit 4 PAGE_PRIVATE_INLINE_INODE 1378 * bit 5 PAGE_PRIVATE_REF_RESOURCE 1379 * bit 6- f2fs private data 1380 * 1381 * Layout B: lowest bit should be 0 1382 * page.private is a wrapped pointer. 1383 */ 1384 enum { 1385 PAGE_PRIVATE_NOT_POINTER, /* private contains non-pointer data */ 1386 PAGE_PRIVATE_ATOMIC_WRITE, /* data page from atomic write path */ 1387 PAGE_PRIVATE_DUMMY_WRITE, /* data page for padding aligned IO */ 1388 PAGE_PRIVATE_ONGOING_MIGRATION, /* data page which is on-going migrating */ 1389 PAGE_PRIVATE_INLINE_INODE, /* inode page contains inline data */ 1390 PAGE_PRIVATE_REF_RESOURCE, /* dirty page has referenced resources */ 1391 PAGE_PRIVATE_MAX 1392 }; 1393 1394 #define PAGE_PRIVATE_GET_FUNC(name, flagname) \ 1395 static inline bool page_private_##name(struct page *page) \ 1396 { \ 1397 return PagePrivate(page) && \ 1398 test_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)) && \ 1399 test_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \ 1400 } 1401 1402 #define PAGE_PRIVATE_SET_FUNC(name, flagname) \ 1403 static inline void set_page_private_##name(struct page *page) \ 1404 { \ 1405 if (!PagePrivate(page)) { \ 1406 get_page(page); \ 1407 SetPagePrivate(page); \ 1408 set_page_private(page, 0); \ 1409 } \ 1410 set_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)); \ 1411 set_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \ 1412 } 1413 1414 #define PAGE_PRIVATE_CLEAR_FUNC(name, flagname) \ 1415 static inline void clear_page_private_##name(struct page *page) \ 1416 { \ 1417 clear_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \ 1418 if (page_private(page) == 1 << PAGE_PRIVATE_NOT_POINTER) { \ 1419 set_page_private(page, 0); \ 1420 if (PagePrivate(page)) { \ 1421 ClearPagePrivate(page); \ 1422 put_page(page); \ 1423 }\ 1424 } \ 1425 } 1426 1427 PAGE_PRIVATE_GET_FUNC(nonpointer, NOT_POINTER); 1428 PAGE_PRIVATE_GET_FUNC(reference, REF_RESOURCE); 1429 PAGE_PRIVATE_GET_FUNC(inline, INLINE_INODE); 1430 PAGE_PRIVATE_GET_FUNC(gcing, ONGOING_MIGRATION); 1431 PAGE_PRIVATE_GET_FUNC(atomic, ATOMIC_WRITE); 1432 PAGE_PRIVATE_GET_FUNC(dummy, DUMMY_WRITE); 1433 1434 PAGE_PRIVATE_SET_FUNC(reference, REF_RESOURCE); 1435 PAGE_PRIVATE_SET_FUNC(inline, INLINE_INODE); 1436 PAGE_PRIVATE_SET_FUNC(gcing, ONGOING_MIGRATION); 1437 PAGE_PRIVATE_SET_FUNC(atomic, ATOMIC_WRITE); 1438 PAGE_PRIVATE_SET_FUNC(dummy, DUMMY_WRITE); 1439 1440 PAGE_PRIVATE_CLEAR_FUNC(reference, REF_RESOURCE); 1441 PAGE_PRIVATE_CLEAR_FUNC(inline, INLINE_INODE); 1442 PAGE_PRIVATE_CLEAR_FUNC(gcing, ONGOING_MIGRATION); 1443 PAGE_PRIVATE_CLEAR_FUNC(atomic, ATOMIC_WRITE); 1444 PAGE_PRIVATE_CLEAR_FUNC(dummy, DUMMY_WRITE); 1445 1446 static inline unsigned long get_page_private_data(struct page *page) 1447 { 1448 unsigned long data = page_private(page); 1449 1450 if (!test_bit(PAGE_PRIVATE_NOT_POINTER, &data)) 1451 return 0; 1452 return data >> PAGE_PRIVATE_MAX; 1453 } 1454 1455 static inline void set_page_private_data(struct page *page, unsigned long data) 1456 { 1457 if (!PagePrivate(page)) { 1458 get_page(page); 1459 SetPagePrivate(page); 1460 set_page_private(page, 0); 1461 } 1462 set_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)); 1463 page_private(page) |= data << PAGE_PRIVATE_MAX; 1464 } 1465 1466 static inline void clear_page_private_data(struct page *page) 1467 { 1468 page_private(page) &= (1 << PAGE_PRIVATE_MAX) - 1; 1469 if (page_private(page) == 1 << PAGE_PRIVATE_NOT_POINTER) { 1470 set_page_private(page, 0); 1471 if (PagePrivate(page)) { 1472 ClearPagePrivate(page); 1473 put_page(page); 1474 } 1475 } 1476 } 1477 1478 /* For compression */ 1479 enum compress_algorithm_type { 1480 COMPRESS_LZO, 1481 COMPRESS_LZ4, 1482 COMPRESS_ZSTD, 1483 COMPRESS_LZORLE, 1484 COMPRESS_MAX, 1485 }; 1486 1487 enum compress_flag { 1488 COMPRESS_CHKSUM, 1489 COMPRESS_MAX_FLAG, 1490 }; 1491 1492 #define COMPRESS_WATERMARK 20 1493 #define COMPRESS_PERCENT 20 1494 1495 #define COMPRESS_DATA_RESERVED_SIZE 4 1496 struct compress_data { 1497 __le32 clen; /* compressed data size */ 1498 __le32 chksum; /* compressed data chksum */ 1499 __le32 reserved[COMPRESS_DATA_RESERVED_SIZE]; /* reserved */ 1500 u8 cdata[]; /* compressed data */ 1501 }; 1502 1503 #define COMPRESS_HEADER_SIZE (sizeof(struct compress_data)) 1504 1505 #define F2FS_COMPRESSED_PAGE_MAGIC 0xF5F2C000 1506 1507 #define COMPRESS_LEVEL_OFFSET 8 1508 1509 /* compress context */ 1510 struct compress_ctx { 1511 struct inode *inode; /* inode the context belong to */ 1512 pgoff_t cluster_idx; /* cluster index number */ 1513 unsigned int cluster_size; /* page count in cluster */ 1514 unsigned int log_cluster_size; /* log of cluster size */ 1515 struct page **rpages; /* pages store raw data in cluster */ 1516 unsigned int nr_rpages; /* total page number in rpages */ 1517 struct page **cpages; /* pages store compressed data in cluster */ 1518 unsigned int nr_cpages; /* total page number in cpages */ 1519 unsigned int valid_nr_cpages; /* valid page number in cpages */ 1520 void *rbuf; /* virtual mapped address on rpages */ 1521 struct compress_data *cbuf; /* virtual mapped address on cpages */ 1522 size_t rlen; /* valid data length in rbuf */ 1523 size_t clen; /* valid data length in cbuf */ 1524 void *private; /* payload buffer for specified compression algorithm */ 1525 void *private2; /* extra payload buffer */ 1526 }; 1527 1528 /* compress context for write IO path */ 1529 struct compress_io_ctx { 1530 u32 magic; /* magic number to indicate page is compressed */ 1531 struct inode *inode; /* inode the context belong to */ 1532 struct page **rpages; /* pages store raw data in cluster */ 1533 unsigned int nr_rpages; /* total page number in rpages */ 1534 atomic_t pending_pages; /* in-flight compressed page count */ 1535 }; 1536 1537 /* Context for decompressing one cluster on the read IO path */ 1538 struct decompress_io_ctx { 1539 u32 magic; /* magic number to indicate page is compressed */ 1540 struct inode *inode; /* inode the context belong to */ 1541 pgoff_t cluster_idx; /* cluster index number */ 1542 unsigned int cluster_size; /* page count in cluster */ 1543 unsigned int log_cluster_size; /* log of cluster size */ 1544 struct page **rpages; /* pages store raw data in cluster */ 1545 unsigned int nr_rpages; /* total page number in rpages */ 1546 struct page **cpages; /* pages store compressed data in cluster */ 1547 unsigned int nr_cpages; /* total page number in cpages */ 1548 struct page **tpages; /* temp pages to pad holes in cluster */ 1549 void *rbuf; /* virtual mapped address on rpages */ 1550 struct compress_data *cbuf; /* virtual mapped address on cpages */ 1551 size_t rlen; /* valid data length in rbuf */ 1552 size_t clen; /* valid data length in cbuf */ 1553 1554 /* 1555 * The number of compressed pages remaining to be read in this cluster. 1556 * This is initially nr_cpages. It is decremented by 1 each time a page 1557 * has been read (or failed to be read). When it reaches 0, the cluster 1558 * is decompressed (or an error is reported). 1559 * 1560 * If an error occurs before all the pages have been submitted for I/O, 1561 * then this will never reach 0. In this case the I/O submitter is 1562 * responsible for calling f2fs_decompress_end_io() instead. 1563 */ 1564 atomic_t remaining_pages; 1565 1566 /* 1567 * Number of references to this decompress_io_ctx. 1568 * 1569 * One reference is held for I/O completion. This reference is dropped 1570 * after the pagecache pages are updated and unlocked -- either after 1571 * decompression (and verity if enabled), or after an error. 1572 * 1573 * In addition, each compressed page holds a reference while it is in a 1574 * bio. These references are necessary prevent compressed pages from 1575 * being freed while they are still in a bio. 1576 */ 1577 refcount_t refcnt; 1578 1579 bool failed; /* IO error occurred before decompression? */ 1580 bool need_verity; /* need fs-verity verification after decompression? */ 1581 void *private; /* payload buffer for specified decompression algorithm */ 1582 void *private2; /* extra payload buffer */ 1583 struct work_struct verity_work; /* work to verify the decompressed pages */ 1584 }; 1585 1586 #define NULL_CLUSTER ((unsigned int)(~0)) 1587 #define MIN_COMPRESS_LOG_SIZE 2 1588 #define MAX_COMPRESS_LOG_SIZE 8 1589 #define MAX_COMPRESS_WINDOW_SIZE(log_size) ((PAGE_SIZE) << (log_size)) 1590 1591 struct f2fs_sb_info { 1592 struct super_block *sb; /* pointer to VFS super block */ 1593 struct proc_dir_entry *s_proc; /* proc entry */ 1594 struct f2fs_super_block *raw_super; /* raw super block pointer */ 1595 struct f2fs_rwsem sb_lock; /* lock for raw super block */ 1596 int valid_super_block; /* valid super block no */ 1597 unsigned long s_flag; /* flags for sbi */ 1598 struct mutex writepages; /* mutex for writepages() */ 1599 1600 #ifdef CONFIG_BLK_DEV_ZONED 1601 unsigned int blocks_per_blkz; /* F2FS blocks per zone */ 1602 unsigned int log_blocks_per_blkz; /* log2 F2FS blocks per zone */ 1603 #endif 1604 1605 /* for node-related operations */ 1606 struct f2fs_nm_info *nm_info; /* node manager */ 1607 struct inode *node_inode; /* cache node blocks */ 1608 1609 /* for segment-related operations */ 1610 struct f2fs_sm_info *sm_info; /* segment manager */ 1611 1612 /* for bio operations */ 1613 struct f2fs_bio_info *write_io[NR_PAGE_TYPE]; /* for write bios */ 1614 /* keep migration IO order for LFS mode */ 1615 struct f2fs_rwsem io_order_lock; 1616 mempool_t *write_io_dummy; /* Dummy pages */ 1617 1618 /* for checkpoint */ 1619 struct f2fs_checkpoint *ckpt; /* raw checkpoint pointer */ 1620 int cur_cp_pack; /* remain current cp pack */ 1621 spinlock_t cp_lock; /* for flag in ckpt */ 1622 struct inode *meta_inode; /* cache meta blocks */ 1623 struct f2fs_rwsem cp_global_sem; /* checkpoint procedure lock */ 1624 struct f2fs_rwsem cp_rwsem; /* blocking FS operations */ 1625 struct f2fs_rwsem node_write; /* locking node writes */ 1626 struct f2fs_rwsem node_change; /* locking node change */ 1627 wait_queue_head_t cp_wait; 1628 unsigned long last_time[MAX_TIME]; /* to store time in jiffies */ 1629 long interval_time[MAX_TIME]; /* to store thresholds */ 1630 struct ckpt_req_control cprc_info; /* for checkpoint request control */ 1631 1632 struct inode_management im[MAX_INO_ENTRY]; /* manage inode cache */ 1633 1634 spinlock_t fsync_node_lock; /* for node entry lock */ 1635 struct list_head fsync_node_list; /* node list head */ 1636 unsigned int fsync_seg_id; /* sequence id */ 1637 unsigned int fsync_node_num; /* number of node entries */ 1638 1639 /* for orphan inode, use 0'th array */ 1640 unsigned int max_orphans; /* max orphan inodes */ 1641 1642 /* for inode management */ 1643 struct list_head inode_list[NR_INODE_TYPE]; /* dirty inode list */ 1644 spinlock_t inode_lock[NR_INODE_TYPE]; /* for dirty inode list lock */ 1645 struct mutex flush_lock; /* for flush exclusion */ 1646 1647 /* for extent tree cache */ 1648 struct radix_tree_root extent_tree_root;/* cache extent cache entries */ 1649 struct mutex extent_tree_lock; /* locking extent radix tree */ 1650 struct list_head extent_list; /* lru list for shrinker */ 1651 spinlock_t extent_lock; /* locking extent lru list */ 1652 atomic_t total_ext_tree; /* extent tree count */ 1653 struct list_head zombie_list; /* extent zombie tree list */ 1654 atomic_t total_zombie_tree; /* extent zombie tree count */ 1655 atomic_t total_ext_node; /* extent info count */ 1656 1657 /* basic filesystem units */ 1658 unsigned int log_sectors_per_block; /* log2 sectors per block */ 1659 unsigned int log_blocksize; /* log2 block size */ 1660 unsigned int blocksize; /* block size */ 1661 unsigned int root_ino_num; /* root inode number*/ 1662 unsigned int node_ino_num; /* node inode number*/ 1663 unsigned int meta_ino_num; /* meta inode number*/ 1664 unsigned int log_blocks_per_seg; /* log2 blocks per segment */ 1665 unsigned int blocks_per_seg; /* blocks per segment */ 1666 unsigned int segs_per_sec; /* segments per section */ 1667 unsigned int secs_per_zone; /* sections per zone */ 1668 unsigned int total_sections; /* total section count */ 1669 unsigned int total_node_count; /* total node block count */ 1670 unsigned int total_valid_node_count; /* valid node block count */ 1671 int dir_level; /* directory level */ 1672 int readdir_ra; /* readahead inode in readdir */ 1673 u64 max_io_bytes; /* max io bytes to merge IOs */ 1674 1675 block_t user_block_count; /* # of user blocks */ 1676 block_t total_valid_block_count; /* # of valid blocks */ 1677 block_t discard_blks; /* discard command candidats */ 1678 block_t last_valid_block_count; /* for recovery */ 1679 block_t reserved_blocks; /* configurable reserved blocks */ 1680 block_t current_reserved_blocks; /* current reserved blocks */ 1681 1682 /* Additional tracking for no checkpoint mode */ 1683 block_t unusable_block_count; /* # of blocks saved by last cp */ 1684 1685 unsigned int nquota_files; /* # of quota sysfile */ 1686 struct f2fs_rwsem quota_sem; /* blocking cp for flags */ 1687 1688 /* # of pages, see count_type */ 1689 atomic_t nr_pages[NR_COUNT_TYPE]; 1690 /* # of allocated blocks */ 1691 struct percpu_counter alloc_valid_block_count; 1692 /* # of node block writes as roll forward recovery */ 1693 struct percpu_counter rf_node_block_count; 1694 1695 /* writeback control */ 1696 atomic_t wb_sync_req[META]; /* count # of WB_SYNC threads */ 1697 1698 /* valid inode count */ 1699 struct percpu_counter total_valid_inode_count; 1700 1701 struct f2fs_mount_info mount_opt; /* mount options */ 1702 1703 /* for cleaning operations */ 1704 struct f2fs_rwsem gc_lock; /* 1705 * semaphore for GC, avoid 1706 * race between GC and GC or CP 1707 */ 1708 struct f2fs_gc_kthread *gc_thread; /* GC thread */ 1709 struct atgc_management am; /* atgc management */ 1710 unsigned int cur_victim_sec; /* current victim section num */ 1711 unsigned int gc_mode; /* current GC state */ 1712 unsigned int next_victim_seg[2]; /* next segment in victim section */ 1713 spinlock_t gc_urgent_high_lock; 1714 bool gc_urgent_high_limited; /* indicates having limited trial count */ 1715 unsigned int gc_urgent_high_remaining; /* remaining trial count for GC_URGENT_HIGH */ 1716 1717 /* for skip statistic */ 1718 unsigned int atomic_files; /* # of opened atomic file */ 1719 unsigned long long skipped_atomic_files[2]; /* FG_GC and BG_GC */ 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 vw_cnt; /* # of volatile writes */ 1751 atomic_t max_aw_cnt; /* max # of atomic writes */ 1752 atomic_t max_vw_cnt; /* max # of volatile writes */ 1753 unsigned int io_skip_bggc; /* skip background gc for in-flight IO */ 1754 unsigned int other_skip_bggc; /* skip background gc for other reasons */ 1755 unsigned int ndirty_inode[NR_INODE_TYPE]; /* # of dirty inodes */ 1756 #endif 1757 spinlock_t stat_lock; /* lock for stat operations */ 1758 1759 /* to attach REQ_META|REQ_FUA flags */ 1760 unsigned int data_io_flag; 1761 unsigned int node_io_flag; 1762 1763 /* For sysfs suppport */ 1764 struct kobject s_kobj; /* /sys/fs/f2fs/<devname> */ 1765 struct completion s_kobj_unregister; 1766 1767 struct kobject s_stat_kobj; /* /sys/fs/f2fs/<devname>/stat */ 1768 struct completion s_stat_kobj_unregister; 1769 1770 struct kobject s_feature_list_kobj; /* /sys/fs/f2fs/<devname>/feature_list */ 1771 struct completion s_feature_list_kobj_unregister; 1772 1773 /* For shrinker support */ 1774 struct list_head s_list; 1775 struct mutex umount_mutex; 1776 unsigned int shrinker_run_no; 1777 1778 /* For multi devices */ 1779 int s_ndevs; /* number of devices */ 1780 struct f2fs_dev_info *devs; /* for device list */ 1781 unsigned int dirty_device; /* for checkpoint data flush */ 1782 spinlock_t dev_lock; /* protect dirty_device */ 1783 bool aligned_blksize; /* all devices has the same logical blksize */ 1784 1785 /* For write statistics */ 1786 u64 sectors_written_start; 1787 u64 kbytes_written; 1788 1789 /* Reference to checksum algorithm driver via cryptoapi */ 1790 struct crypto_shash *s_chksum_driver; 1791 1792 /* Precomputed FS UUID checksum for seeding other checksums */ 1793 __u32 s_chksum_seed; 1794 1795 struct workqueue_struct *post_read_wq; /* post read workqueue */ 1796 1797 struct kmem_cache *inline_xattr_slab; /* inline xattr entry */ 1798 unsigned int inline_xattr_slab_size; /* default inline xattr slab size */ 1799 1800 /* For reclaimed segs statistics per each GC mode */ 1801 unsigned int gc_segment_mode; /* GC state for reclaimed segments */ 1802 unsigned int gc_reclaimed_segs[MAX_GC_MODE]; /* Reclaimed segs for each mode */ 1803 1804 unsigned long seq_file_ra_mul; /* multiplier for ra_pages of seq. files in fadvise */ 1805 1806 int max_fragment_chunk; /* max chunk size for block fragmentation mode */ 1807 int max_fragment_hole; /* max hole size for block fragmentation mode */ 1808 1809 #ifdef CONFIG_F2FS_FS_COMPRESSION 1810 struct kmem_cache *page_array_slab; /* page array entry */ 1811 unsigned int page_array_slab_size; /* default page array slab size */ 1812 1813 /* For runtime compression statistics */ 1814 u64 compr_written_block; 1815 u64 compr_saved_block; 1816 u32 compr_new_inode; 1817 1818 /* For compressed block cache */ 1819 struct inode *compress_inode; /* cache compressed blocks */ 1820 unsigned int compress_percent; /* cache page percentage */ 1821 unsigned int compress_watermark; /* cache page watermark */ 1822 atomic_t compress_page_hit; /* cache hit count */ 1823 #endif 1824 1825 #ifdef CONFIG_F2FS_IOSTAT 1826 /* For app/fs IO statistics */ 1827 spinlock_t iostat_lock; 1828 unsigned long long rw_iostat[NR_IO_TYPE]; 1829 unsigned long long prev_rw_iostat[NR_IO_TYPE]; 1830 bool iostat_enable; 1831 unsigned long iostat_next_period; 1832 unsigned int iostat_period_ms; 1833 1834 /* For io latency related statistics info in one iostat period */ 1835 spinlock_t iostat_lat_lock; 1836 struct iostat_lat_info *iostat_io_lat; 1837 #endif 1838 }; 1839 1840 #ifdef CONFIG_F2FS_FAULT_INJECTION 1841 #define f2fs_show_injection_info(sbi, type) \ 1842 printk_ratelimited("%sF2FS-fs (%s) : inject %s in %s of %pS\n", \ 1843 KERN_INFO, sbi->sb->s_id, \ 1844 f2fs_fault_name[type], \ 1845 __func__, __builtin_return_address(0)) 1846 static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type) 1847 { 1848 struct f2fs_fault_info *ffi = &F2FS_OPTION(sbi).fault_info; 1849 1850 if (!ffi->inject_rate) 1851 return false; 1852 1853 if (!IS_FAULT_SET(ffi, type)) 1854 return false; 1855 1856 atomic_inc(&ffi->inject_ops); 1857 if (atomic_read(&ffi->inject_ops) >= ffi->inject_rate) { 1858 atomic_set(&ffi->inject_ops, 0); 1859 return true; 1860 } 1861 return false; 1862 } 1863 #else 1864 #define f2fs_show_injection_info(sbi, type) do { } while (0) 1865 static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type) 1866 { 1867 return false; 1868 } 1869 #endif 1870 1871 /* 1872 * Test if the mounted volume is a multi-device volume. 1873 * - For a single regular disk volume, sbi->s_ndevs is 0. 1874 * - For a single zoned disk volume, sbi->s_ndevs is 1. 1875 * - For a multi-device volume, sbi->s_ndevs is always 2 or more. 1876 */ 1877 static inline bool f2fs_is_multi_device(struct f2fs_sb_info *sbi) 1878 { 1879 return sbi->s_ndevs > 1; 1880 } 1881 1882 static inline void f2fs_update_time(struct f2fs_sb_info *sbi, int type) 1883 { 1884 unsigned long now = jiffies; 1885 1886 sbi->last_time[type] = now; 1887 1888 /* DISCARD_TIME and GC_TIME are based on REQ_TIME */ 1889 if (type == REQ_TIME) { 1890 sbi->last_time[DISCARD_TIME] = now; 1891 sbi->last_time[GC_TIME] = now; 1892 } 1893 } 1894 1895 static inline bool f2fs_time_over(struct f2fs_sb_info *sbi, int type) 1896 { 1897 unsigned long interval = sbi->interval_time[type] * HZ; 1898 1899 return time_after(jiffies, sbi->last_time[type] + interval); 1900 } 1901 1902 static inline unsigned int f2fs_time_to_wait(struct f2fs_sb_info *sbi, 1903 int type) 1904 { 1905 unsigned long interval = sbi->interval_time[type] * HZ; 1906 unsigned int wait_ms = 0; 1907 long delta; 1908 1909 delta = (sbi->last_time[type] + interval) - jiffies; 1910 if (delta > 0) 1911 wait_ms = jiffies_to_msecs(delta); 1912 1913 return wait_ms; 1914 } 1915 1916 /* 1917 * Inline functions 1918 */ 1919 static inline u32 __f2fs_crc32(struct f2fs_sb_info *sbi, u32 crc, 1920 const void *address, unsigned int length) 1921 { 1922 struct { 1923 struct shash_desc shash; 1924 char ctx[4]; 1925 } desc; 1926 int err; 1927 1928 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver) != sizeof(desc.ctx)); 1929 1930 desc.shash.tfm = sbi->s_chksum_driver; 1931 *(u32 *)desc.ctx = crc; 1932 1933 err = crypto_shash_update(&desc.shash, address, length); 1934 BUG_ON(err); 1935 1936 return *(u32 *)desc.ctx; 1937 } 1938 1939 static inline u32 f2fs_crc32(struct f2fs_sb_info *sbi, const void *address, 1940 unsigned int length) 1941 { 1942 return __f2fs_crc32(sbi, F2FS_SUPER_MAGIC, address, length); 1943 } 1944 1945 static inline bool f2fs_crc_valid(struct f2fs_sb_info *sbi, __u32 blk_crc, 1946 void *buf, size_t buf_size) 1947 { 1948 return f2fs_crc32(sbi, buf, buf_size) == blk_crc; 1949 } 1950 1951 static inline u32 f2fs_chksum(struct f2fs_sb_info *sbi, u32 crc, 1952 const void *address, unsigned int length) 1953 { 1954 return __f2fs_crc32(sbi, crc, address, length); 1955 } 1956 1957 static inline struct f2fs_inode_info *F2FS_I(struct inode *inode) 1958 { 1959 return container_of(inode, struct f2fs_inode_info, vfs_inode); 1960 } 1961 1962 static inline struct f2fs_sb_info *F2FS_SB(struct super_block *sb) 1963 { 1964 return sb->s_fs_info; 1965 } 1966 1967 static inline struct f2fs_sb_info *F2FS_I_SB(struct inode *inode) 1968 { 1969 return F2FS_SB(inode->i_sb); 1970 } 1971 1972 static inline struct f2fs_sb_info *F2FS_M_SB(struct address_space *mapping) 1973 { 1974 return F2FS_I_SB(mapping->host); 1975 } 1976 1977 static inline struct f2fs_sb_info *F2FS_P_SB(struct page *page) 1978 { 1979 return F2FS_M_SB(page_file_mapping(page)); 1980 } 1981 1982 static inline struct f2fs_super_block *F2FS_RAW_SUPER(struct f2fs_sb_info *sbi) 1983 { 1984 return (struct f2fs_super_block *)(sbi->raw_super); 1985 } 1986 1987 static inline struct f2fs_checkpoint *F2FS_CKPT(struct f2fs_sb_info *sbi) 1988 { 1989 return (struct f2fs_checkpoint *)(sbi->ckpt); 1990 } 1991 1992 static inline struct f2fs_node *F2FS_NODE(struct page *page) 1993 { 1994 return (struct f2fs_node *)page_address(page); 1995 } 1996 1997 static inline struct f2fs_inode *F2FS_INODE(struct page *page) 1998 { 1999 return &((struct f2fs_node *)page_address(page))->i; 2000 } 2001 2002 static inline struct f2fs_nm_info *NM_I(struct f2fs_sb_info *sbi) 2003 { 2004 return (struct f2fs_nm_info *)(sbi->nm_info); 2005 } 2006 2007 static inline struct f2fs_sm_info *SM_I(struct f2fs_sb_info *sbi) 2008 { 2009 return (struct f2fs_sm_info *)(sbi->sm_info); 2010 } 2011 2012 static inline struct sit_info *SIT_I(struct f2fs_sb_info *sbi) 2013 { 2014 return (struct sit_info *)(SM_I(sbi)->sit_info); 2015 } 2016 2017 static inline struct free_segmap_info *FREE_I(struct f2fs_sb_info *sbi) 2018 { 2019 return (struct free_segmap_info *)(SM_I(sbi)->free_info); 2020 } 2021 2022 static inline struct dirty_seglist_info *DIRTY_I(struct f2fs_sb_info *sbi) 2023 { 2024 return (struct dirty_seglist_info *)(SM_I(sbi)->dirty_info); 2025 } 2026 2027 static inline struct address_space *META_MAPPING(struct f2fs_sb_info *sbi) 2028 { 2029 return sbi->meta_inode->i_mapping; 2030 } 2031 2032 static inline struct address_space *NODE_MAPPING(struct f2fs_sb_info *sbi) 2033 { 2034 return sbi->node_inode->i_mapping; 2035 } 2036 2037 static inline bool is_sbi_flag_set(struct f2fs_sb_info *sbi, unsigned int type) 2038 { 2039 return test_bit(type, &sbi->s_flag); 2040 } 2041 2042 static inline void set_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type) 2043 { 2044 set_bit(type, &sbi->s_flag); 2045 } 2046 2047 static inline void clear_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type) 2048 { 2049 clear_bit(type, &sbi->s_flag); 2050 } 2051 2052 static inline unsigned long long cur_cp_version(struct f2fs_checkpoint *cp) 2053 { 2054 return le64_to_cpu(cp->checkpoint_ver); 2055 } 2056 2057 static inline unsigned long f2fs_qf_ino(struct super_block *sb, int type) 2058 { 2059 if (type < F2FS_MAX_QUOTAS) 2060 return le32_to_cpu(F2FS_SB(sb)->raw_super->qf_ino[type]); 2061 return 0; 2062 } 2063 2064 static inline __u64 cur_cp_crc(struct f2fs_checkpoint *cp) 2065 { 2066 size_t crc_offset = le32_to_cpu(cp->checksum_offset); 2067 return le32_to_cpu(*((__le32 *)((unsigned char *)cp + crc_offset))); 2068 } 2069 2070 static inline bool __is_set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 2071 { 2072 unsigned int ckpt_flags = le32_to_cpu(cp->ckpt_flags); 2073 2074 return ckpt_flags & f; 2075 } 2076 2077 static inline bool is_set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 2078 { 2079 return __is_set_ckpt_flags(F2FS_CKPT(sbi), f); 2080 } 2081 2082 static inline void __set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 2083 { 2084 unsigned int ckpt_flags; 2085 2086 ckpt_flags = le32_to_cpu(cp->ckpt_flags); 2087 ckpt_flags |= f; 2088 cp->ckpt_flags = cpu_to_le32(ckpt_flags); 2089 } 2090 2091 static inline void set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 2092 { 2093 unsigned long flags; 2094 2095 spin_lock_irqsave(&sbi->cp_lock, flags); 2096 __set_ckpt_flags(F2FS_CKPT(sbi), f); 2097 spin_unlock_irqrestore(&sbi->cp_lock, flags); 2098 } 2099 2100 static inline void __clear_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 2101 { 2102 unsigned int ckpt_flags; 2103 2104 ckpt_flags = le32_to_cpu(cp->ckpt_flags); 2105 ckpt_flags &= (~f); 2106 cp->ckpt_flags = cpu_to_le32(ckpt_flags); 2107 } 2108 2109 static inline void clear_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 2110 { 2111 unsigned long flags; 2112 2113 spin_lock_irqsave(&sbi->cp_lock, flags); 2114 __clear_ckpt_flags(F2FS_CKPT(sbi), f); 2115 spin_unlock_irqrestore(&sbi->cp_lock, flags); 2116 } 2117 2118 #define init_f2fs_rwsem(sem) \ 2119 do { \ 2120 static struct lock_class_key __key; \ 2121 \ 2122 __init_f2fs_rwsem((sem), #sem, &__key); \ 2123 } while (0) 2124 2125 static inline void __init_f2fs_rwsem(struct f2fs_rwsem *sem, 2126 const char *sem_name, struct lock_class_key *key) 2127 { 2128 __init_rwsem(&sem->internal_rwsem, sem_name, key); 2129 init_waitqueue_head(&sem->read_waiters); 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 wait_event(sem->read_waiters, down_read_trylock(&sem->internal_rwsem)); 2145 } 2146 2147 static inline int f2fs_down_read_trylock(struct f2fs_rwsem *sem) 2148 { 2149 return down_read_trylock(&sem->internal_rwsem); 2150 } 2151 2152 #ifdef CONFIG_DEBUG_LOCK_ALLOC 2153 static inline void f2fs_down_read_nested(struct f2fs_rwsem *sem, int subclass) 2154 { 2155 down_read_nested(&sem->internal_rwsem, subclass); 2156 } 2157 #else 2158 #define f2fs_down_read_nested(sem, subclass) f2fs_down_read(sem) 2159 #endif 2160 2161 static inline void f2fs_up_read(struct f2fs_rwsem *sem) 2162 { 2163 up_read(&sem->internal_rwsem); 2164 } 2165 2166 static inline void f2fs_down_write(struct f2fs_rwsem *sem) 2167 { 2168 down_write(&sem->internal_rwsem); 2169 } 2170 2171 static inline int f2fs_down_write_trylock(struct f2fs_rwsem *sem) 2172 { 2173 return down_write_trylock(&sem->internal_rwsem); 2174 } 2175 2176 static inline void f2fs_up_write(struct f2fs_rwsem *sem) 2177 { 2178 up_write(&sem->internal_rwsem); 2179 wake_up_all(&sem->read_waiters); 2180 } 2181 2182 static inline void f2fs_lock_op(struct f2fs_sb_info *sbi) 2183 { 2184 f2fs_down_read(&sbi->cp_rwsem); 2185 } 2186 2187 static inline int f2fs_trylock_op(struct f2fs_sb_info *sbi) 2188 { 2189 if (time_to_inject(sbi, FAULT_LOCK_OP)) { 2190 f2fs_show_injection_info(sbi, FAULT_LOCK_OP); 2191 return 0; 2192 } 2193 return f2fs_down_read_trylock(&sbi->cp_rwsem); 2194 } 2195 2196 static inline void f2fs_unlock_op(struct f2fs_sb_info *sbi) 2197 { 2198 f2fs_up_read(&sbi->cp_rwsem); 2199 } 2200 2201 static inline void f2fs_lock_all(struct f2fs_sb_info *sbi) 2202 { 2203 f2fs_down_write(&sbi->cp_rwsem); 2204 } 2205 2206 static inline void f2fs_unlock_all(struct f2fs_sb_info *sbi) 2207 { 2208 f2fs_up_write(&sbi->cp_rwsem); 2209 } 2210 2211 static inline int __get_cp_reason(struct f2fs_sb_info *sbi) 2212 { 2213 int reason = CP_SYNC; 2214 2215 if (test_opt(sbi, FASTBOOT)) 2216 reason = CP_FASTBOOT; 2217 if (is_sbi_flag_set(sbi, SBI_IS_CLOSE)) 2218 reason = CP_UMOUNT; 2219 return reason; 2220 } 2221 2222 static inline bool __remain_node_summaries(int reason) 2223 { 2224 return (reason & (CP_UMOUNT | CP_FASTBOOT)); 2225 } 2226 2227 static inline bool __exist_node_summaries(struct f2fs_sb_info *sbi) 2228 { 2229 return (is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG) || 2230 is_set_ckpt_flags(sbi, CP_FASTBOOT_FLAG)); 2231 } 2232 2233 /* 2234 * Check whether the inode has blocks or not 2235 */ 2236 static inline int F2FS_HAS_BLOCKS(struct inode *inode) 2237 { 2238 block_t xattr_block = F2FS_I(inode)->i_xattr_nid ? 1 : 0; 2239 2240 return (inode->i_blocks >> F2FS_LOG_SECTORS_PER_BLOCK) > xattr_block; 2241 } 2242 2243 static inline bool f2fs_has_xattr_block(unsigned int ofs) 2244 { 2245 return ofs == XATTR_NODE_OFFSET; 2246 } 2247 2248 static inline bool __allow_reserved_blocks(struct f2fs_sb_info *sbi, 2249 struct inode *inode, bool cap) 2250 { 2251 if (!inode) 2252 return true; 2253 if (!test_opt(sbi, RESERVE_ROOT)) 2254 return false; 2255 if (IS_NOQUOTA(inode)) 2256 return true; 2257 if (uid_eq(F2FS_OPTION(sbi).s_resuid, current_fsuid())) 2258 return true; 2259 if (!gid_eq(F2FS_OPTION(sbi).s_resgid, GLOBAL_ROOT_GID) && 2260 in_group_p(F2FS_OPTION(sbi).s_resgid)) 2261 return true; 2262 if (cap && capable(CAP_SYS_RESOURCE)) 2263 return true; 2264 return false; 2265 } 2266 2267 static inline void f2fs_i_blocks_write(struct inode *, block_t, bool, bool); 2268 static inline int inc_valid_block_count(struct f2fs_sb_info *sbi, 2269 struct inode *inode, blkcnt_t *count) 2270 { 2271 blkcnt_t diff = 0, release = 0; 2272 block_t avail_user_block_count; 2273 int ret; 2274 2275 ret = dquot_reserve_block(inode, *count); 2276 if (ret) 2277 return ret; 2278 2279 if (time_to_inject(sbi, FAULT_BLOCK)) { 2280 f2fs_show_injection_info(sbi, FAULT_BLOCK); 2281 release = *count; 2282 goto release_quota; 2283 } 2284 2285 /* 2286 * let's increase this in prior to actual block count change in order 2287 * for f2fs_sync_file to avoid data races when deciding checkpoint. 2288 */ 2289 percpu_counter_add(&sbi->alloc_valid_block_count, (*count)); 2290 2291 spin_lock(&sbi->stat_lock); 2292 sbi->total_valid_block_count += (block_t)(*count); 2293 avail_user_block_count = sbi->user_block_count - 2294 sbi->current_reserved_blocks; 2295 2296 if (!__allow_reserved_blocks(sbi, inode, true)) 2297 avail_user_block_count -= F2FS_OPTION(sbi).root_reserved_blocks; 2298 2299 if (F2FS_IO_ALIGNED(sbi)) 2300 avail_user_block_count -= sbi->blocks_per_seg * 2301 SM_I(sbi)->additional_reserved_segments; 2302 2303 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) { 2304 if (avail_user_block_count > sbi->unusable_block_count) 2305 avail_user_block_count -= sbi->unusable_block_count; 2306 else 2307 avail_user_block_count = 0; 2308 } 2309 if (unlikely(sbi->total_valid_block_count > avail_user_block_count)) { 2310 diff = sbi->total_valid_block_count - avail_user_block_count; 2311 if (diff > *count) 2312 diff = *count; 2313 *count -= diff; 2314 release = diff; 2315 sbi->total_valid_block_count -= diff; 2316 if (!*count) { 2317 spin_unlock(&sbi->stat_lock); 2318 goto enospc; 2319 } 2320 } 2321 spin_unlock(&sbi->stat_lock); 2322 2323 if (unlikely(release)) { 2324 percpu_counter_sub(&sbi->alloc_valid_block_count, release); 2325 dquot_release_reservation_block(inode, release); 2326 } 2327 f2fs_i_blocks_write(inode, *count, true, true); 2328 return 0; 2329 2330 enospc: 2331 percpu_counter_sub(&sbi->alloc_valid_block_count, release); 2332 release_quota: 2333 dquot_release_reservation_block(inode, release); 2334 return -ENOSPC; 2335 } 2336 2337 __printf(2, 3) 2338 void f2fs_printk(struct f2fs_sb_info *sbi, const char *fmt, ...); 2339 2340 #define f2fs_err(sbi, fmt, ...) \ 2341 f2fs_printk(sbi, KERN_ERR fmt, ##__VA_ARGS__) 2342 #define f2fs_warn(sbi, fmt, ...) \ 2343 f2fs_printk(sbi, KERN_WARNING fmt, ##__VA_ARGS__) 2344 #define f2fs_notice(sbi, fmt, ...) \ 2345 f2fs_printk(sbi, KERN_NOTICE fmt, ##__VA_ARGS__) 2346 #define f2fs_info(sbi, fmt, ...) \ 2347 f2fs_printk(sbi, KERN_INFO fmt, ##__VA_ARGS__) 2348 #define f2fs_debug(sbi, fmt, ...) \ 2349 f2fs_printk(sbi, KERN_DEBUG fmt, ##__VA_ARGS__) 2350 2351 static inline void dec_valid_block_count(struct f2fs_sb_info *sbi, 2352 struct inode *inode, 2353 block_t count) 2354 { 2355 blkcnt_t sectors = count << F2FS_LOG_SECTORS_PER_BLOCK; 2356 2357 spin_lock(&sbi->stat_lock); 2358 f2fs_bug_on(sbi, sbi->total_valid_block_count < (block_t) count); 2359 sbi->total_valid_block_count -= (block_t)count; 2360 if (sbi->reserved_blocks && 2361 sbi->current_reserved_blocks < sbi->reserved_blocks) 2362 sbi->current_reserved_blocks = min(sbi->reserved_blocks, 2363 sbi->current_reserved_blocks + count); 2364 spin_unlock(&sbi->stat_lock); 2365 if (unlikely(inode->i_blocks < sectors)) { 2366 f2fs_warn(sbi, "Inconsistent i_blocks, ino:%lu, iblocks:%llu, sectors:%llu", 2367 inode->i_ino, 2368 (unsigned long long)inode->i_blocks, 2369 (unsigned long long)sectors); 2370 set_sbi_flag(sbi, SBI_NEED_FSCK); 2371 return; 2372 } 2373 f2fs_i_blocks_write(inode, count, false, true); 2374 } 2375 2376 static inline void inc_page_count(struct f2fs_sb_info *sbi, int count_type) 2377 { 2378 atomic_inc(&sbi->nr_pages[count_type]); 2379 2380 if (count_type == F2FS_DIRTY_DENTS || 2381 count_type == F2FS_DIRTY_NODES || 2382 count_type == F2FS_DIRTY_META || 2383 count_type == F2FS_DIRTY_QDATA || 2384 count_type == F2FS_DIRTY_IMETA) 2385 set_sbi_flag(sbi, SBI_IS_DIRTY); 2386 } 2387 2388 static inline void inode_inc_dirty_pages(struct inode *inode) 2389 { 2390 atomic_inc(&F2FS_I(inode)->dirty_pages); 2391 inc_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ? 2392 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA); 2393 if (IS_NOQUOTA(inode)) 2394 inc_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA); 2395 } 2396 2397 static inline void dec_page_count(struct f2fs_sb_info *sbi, int count_type) 2398 { 2399 atomic_dec(&sbi->nr_pages[count_type]); 2400 } 2401 2402 static inline void inode_dec_dirty_pages(struct inode *inode) 2403 { 2404 if (!S_ISDIR(inode->i_mode) && !S_ISREG(inode->i_mode) && 2405 !S_ISLNK(inode->i_mode)) 2406 return; 2407 2408 atomic_dec(&F2FS_I(inode)->dirty_pages); 2409 dec_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ? 2410 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA); 2411 if (IS_NOQUOTA(inode)) 2412 dec_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA); 2413 } 2414 2415 static inline s64 get_pages(struct f2fs_sb_info *sbi, int count_type) 2416 { 2417 return atomic_read(&sbi->nr_pages[count_type]); 2418 } 2419 2420 static inline int get_dirty_pages(struct inode *inode) 2421 { 2422 return atomic_read(&F2FS_I(inode)->dirty_pages); 2423 } 2424 2425 static inline int get_blocktype_secs(struct f2fs_sb_info *sbi, int block_type) 2426 { 2427 unsigned int pages_per_sec = sbi->segs_per_sec * sbi->blocks_per_seg; 2428 unsigned int segs = (get_pages(sbi, block_type) + pages_per_sec - 1) >> 2429 sbi->log_blocks_per_seg; 2430 2431 return segs / sbi->segs_per_sec; 2432 } 2433 2434 static inline block_t valid_user_blocks(struct f2fs_sb_info *sbi) 2435 { 2436 return sbi->total_valid_block_count; 2437 } 2438 2439 static inline block_t discard_blocks(struct f2fs_sb_info *sbi) 2440 { 2441 return sbi->discard_blks; 2442 } 2443 2444 static inline unsigned long __bitmap_size(struct f2fs_sb_info *sbi, int flag) 2445 { 2446 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi); 2447 2448 /* return NAT or SIT bitmap */ 2449 if (flag == NAT_BITMAP) 2450 return le32_to_cpu(ckpt->nat_ver_bitmap_bytesize); 2451 else if (flag == SIT_BITMAP) 2452 return le32_to_cpu(ckpt->sit_ver_bitmap_bytesize); 2453 2454 return 0; 2455 } 2456 2457 static inline block_t __cp_payload(struct f2fs_sb_info *sbi) 2458 { 2459 return le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_payload); 2460 } 2461 2462 static inline void *__bitmap_ptr(struct f2fs_sb_info *sbi, int flag) 2463 { 2464 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi); 2465 void *tmp_ptr = &ckpt->sit_nat_version_bitmap; 2466 int offset; 2467 2468 if (is_set_ckpt_flags(sbi, CP_LARGE_NAT_BITMAP_FLAG)) { 2469 offset = (flag == SIT_BITMAP) ? 2470 le32_to_cpu(ckpt->nat_ver_bitmap_bytesize) : 0; 2471 /* 2472 * if large_nat_bitmap feature is enabled, leave checksum 2473 * protection for all nat/sit bitmaps. 2474 */ 2475 return tmp_ptr + offset + sizeof(__le32); 2476 } 2477 2478 if (__cp_payload(sbi) > 0) { 2479 if (flag == NAT_BITMAP) 2480 return &ckpt->sit_nat_version_bitmap; 2481 else 2482 return (unsigned char *)ckpt + F2FS_BLKSIZE; 2483 } else { 2484 offset = (flag == NAT_BITMAP) ? 2485 le32_to_cpu(ckpt->sit_ver_bitmap_bytesize) : 0; 2486 return tmp_ptr + offset; 2487 } 2488 } 2489 2490 static inline block_t __start_cp_addr(struct f2fs_sb_info *sbi) 2491 { 2492 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr); 2493 2494 if (sbi->cur_cp_pack == 2) 2495 start_addr += sbi->blocks_per_seg; 2496 return start_addr; 2497 } 2498 2499 static inline block_t __start_cp_next_addr(struct f2fs_sb_info *sbi) 2500 { 2501 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr); 2502 2503 if (sbi->cur_cp_pack == 1) 2504 start_addr += sbi->blocks_per_seg; 2505 return start_addr; 2506 } 2507 2508 static inline void __set_cp_next_pack(struct f2fs_sb_info *sbi) 2509 { 2510 sbi->cur_cp_pack = (sbi->cur_cp_pack == 1) ? 2 : 1; 2511 } 2512 2513 static inline block_t __start_sum_addr(struct f2fs_sb_info *sbi) 2514 { 2515 return le32_to_cpu(F2FS_CKPT(sbi)->cp_pack_start_sum); 2516 } 2517 2518 static inline int inc_valid_node_count(struct f2fs_sb_info *sbi, 2519 struct inode *inode, bool is_inode) 2520 { 2521 block_t valid_block_count; 2522 unsigned int valid_node_count, user_block_count; 2523 int err; 2524 2525 if (is_inode) { 2526 if (inode) { 2527 err = dquot_alloc_inode(inode); 2528 if (err) 2529 return err; 2530 } 2531 } else { 2532 err = dquot_reserve_block(inode, 1); 2533 if (err) 2534 return err; 2535 } 2536 2537 if (time_to_inject(sbi, FAULT_BLOCK)) { 2538 f2fs_show_injection_info(sbi, FAULT_BLOCK); 2539 goto enospc; 2540 } 2541 2542 spin_lock(&sbi->stat_lock); 2543 2544 valid_block_count = sbi->total_valid_block_count + 2545 sbi->current_reserved_blocks + 1; 2546 2547 if (!__allow_reserved_blocks(sbi, inode, false)) 2548 valid_block_count += F2FS_OPTION(sbi).root_reserved_blocks; 2549 2550 if (F2FS_IO_ALIGNED(sbi)) 2551 valid_block_count += sbi->blocks_per_seg * 2552 SM_I(sbi)->additional_reserved_segments; 2553 2554 user_block_count = sbi->user_block_count; 2555 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) 2556 user_block_count -= sbi->unusable_block_count; 2557 2558 if (unlikely(valid_block_count > user_block_count)) { 2559 spin_unlock(&sbi->stat_lock); 2560 goto enospc; 2561 } 2562 2563 valid_node_count = sbi->total_valid_node_count + 1; 2564 if (unlikely(valid_node_count > sbi->total_node_count)) { 2565 spin_unlock(&sbi->stat_lock); 2566 goto enospc; 2567 } 2568 2569 sbi->total_valid_node_count++; 2570 sbi->total_valid_block_count++; 2571 spin_unlock(&sbi->stat_lock); 2572 2573 if (inode) { 2574 if (is_inode) 2575 f2fs_mark_inode_dirty_sync(inode, true); 2576 else 2577 f2fs_i_blocks_write(inode, 1, true, true); 2578 } 2579 2580 percpu_counter_inc(&sbi->alloc_valid_block_count); 2581 return 0; 2582 2583 enospc: 2584 if (is_inode) { 2585 if (inode) 2586 dquot_free_inode(inode); 2587 } else { 2588 dquot_release_reservation_block(inode, 1); 2589 } 2590 return -ENOSPC; 2591 } 2592 2593 static inline void dec_valid_node_count(struct f2fs_sb_info *sbi, 2594 struct inode *inode, bool is_inode) 2595 { 2596 spin_lock(&sbi->stat_lock); 2597 2598 f2fs_bug_on(sbi, !sbi->total_valid_block_count); 2599 f2fs_bug_on(sbi, !sbi->total_valid_node_count); 2600 2601 sbi->total_valid_node_count--; 2602 sbi->total_valid_block_count--; 2603 if (sbi->reserved_blocks && 2604 sbi->current_reserved_blocks < sbi->reserved_blocks) 2605 sbi->current_reserved_blocks++; 2606 2607 spin_unlock(&sbi->stat_lock); 2608 2609 if (is_inode) { 2610 dquot_free_inode(inode); 2611 } else { 2612 if (unlikely(inode->i_blocks == 0)) { 2613 f2fs_warn(sbi, "dec_valid_node_count: inconsistent i_blocks, ino:%lu, iblocks:%llu", 2614 inode->i_ino, 2615 (unsigned long long)inode->i_blocks); 2616 set_sbi_flag(sbi, SBI_NEED_FSCK); 2617 return; 2618 } 2619 f2fs_i_blocks_write(inode, 1, false, true); 2620 } 2621 } 2622 2623 static inline unsigned int valid_node_count(struct f2fs_sb_info *sbi) 2624 { 2625 return sbi->total_valid_node_count; 2626 } 2627 2628 static inline void inc_valid_inode_count(struct f2fs_sb_info *sbi) 2629 { 2630 percpu_counter_inc(&sbi->total_valid_inode_count); 2631 } 2632 2633 static inline void dec_valid_inode_count(struct f2fs_sb_info *sbi) 2634 { 2635 percpu_counter_dec(&sbi->total_valid_inode_count); 2636 } 2637 2638 static inline s64 valid_inode_count(struct f2fs_sb_info *sbi) 2639 { 2640 return percpu_counter_sum_positive(&sbi->total_valid_inode_count); 2641 } 2642 2643 static inline struct page *f2fs_grab_cache_page(struct address_space *mapping, 2644 pgoff_t index, bool for_write) 2645 { 2646 struct page *page; 2647 2648 if (IS_ENABLED(CONFIG_F2FS_FAULT_INJECTION)) { 2649 if (!for_write) 2650 page = find_get_page_flags(mapping, index, 2651 FGP_LOCK | FGP_ACCESSED); 2652 else 2653 page = find_lock_page(mapping, index); 2654 if (page) 2655 return page; 2656 2657 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_ALLOC)) { 2658 f2fs_show_injection_info(F2FS_M_SB(mapping), 2659 FAULT_PAGE_ALLOC); 2660 return NULL; 2661 } 2662 } 2663 2664 if (!for_write) 2665 return grab_cache_page(mapping, index); 2666 return grab_cache_page_write_begin(mapping, index, AOP_FLAG_NOFS); 2667 } 2668 2669 static inline struct page *f2fs_pagecache_get_page( 2670 struct address_space *mapping, pgoff_t index, 2671 int fgp_flags, gfp_t gfp_mask) 2672 { 2673 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_GET)) { 2674 f2fs_show_injection_info(F2FS_M_SB(mapping), FAULT_PAGE_GET); 2675 return NULL; 2676 } 2677 2678 return pagecache_get_page(mapping, index, fgp_flags, gfp_mask); 2679 } 2680 2681 static inline void f2fs_copy_page(struct page *src, struct page *dst) 2682 { 2683 char *src_kaddr = kmap(src); 2684 char *dst_kaddr = kmap(dst); 2685 2686 memcpy(dst_kaddr, src_kaddr, PAGE_SIZE); 2687 kunmap(dst); 2688 kunmap(src); 2689 } 2690 2691 static inline void f2fs_put_page(struct page *page, int unlock) 2692 { 2693 if (!page) 2694 return; 2695 2696 if (unlock) { 2697 f2fs_bug_on(F2FS_P_SB(page), !PageLocked(page)); 2698 unlock_page(page); 2699 } 2700 put_page(page); 2701 } 2702 2703 static inline void f2fs_put_dnode(struct dnode_of_data *dn) 2704 { 2705 if (dn->node_page) 2706 f2fs_put_page(dn->node_page, 1); 2707 if (dn->inode_page && dn->node_page != dn->inode_page) 2708 f2fs_put_page(dn->inode_page, 0); 2709 dn->node_page = NULL; 2710 dn->inode_page = NULL; 2711 } 2712 2713 static inline struct kmem_cache *f2fs_kmem_cache_create(const char *name, 2714 size_t size) 2715 { 2716 return kmem_cache_create(name, size, 0, SLAB_RECLAIM_ACCOUNT, NULL); 2717 } 2718 2719 static inline void *f2fs_kmem_cache_alloc_nofail(struct kmem_cache *cachep, 2720 gfp_t flags) 2721 { 2722 void *entry; 2723 2724 entry = kmem_cache_alloc(cachep, flags); 2725 if (!entry) 2726 entry = kmem_cache_alloc(cachep, flags | __GFP_NOFAIL); 2727 return entry; 2728 } 2729 2730 static inline void *f2fs_kmem_cache_alloc(struct kmem_cache *cachep, 2731 gfp_t flags, bool nofail, struct f2fs_sb_info *sbi) 2732 { 2733 if (nofail) 2734 return f2fs_kmem_cache_alloc_nofail(cachep, flags); 2735 2736 if (time_to_inject(sbi, FAULT_SLAB_ALLOC)) { 2737 f2fs_show_injection_info(sbi, FAULT_SLAB_ALLOC); 2738 return NULL; 2739 } 2740 2741 return kmem_cache_alloc(cachep, flags); 2742 } 2743 2744 static inline bool is_inflight_io(struct f2fs_sb_info *sbi, int type) 2745 { 2746 if (get_pages(sbi, F2FS_RD_DATA) || get_pages(sbi, F2FS_RD_NODE) || 2747 get_pages(sbi, F2FS_RD_META) || get_pages(sbi, F2FS_WB_DATA) || 2748 get_pages(sbi, F2FS_WB_CP_DATA) || 2749 get_pages(sbi, F2FS_DIO_READ) || 2750 get_pages(sbi, F2FS_DIO_WRITE)) 2751 return true; 2752 2753 if (type != DISCARD_TIME && SM_I(sbi) && SM_I(sbi)->dcc_info && 2754 atomic_read(&SM_I(sbi)->dcc_info->queued_discard)) 2755 return true; 2756 2757 if (SM_I(sbi) && SM_I(sbi)->fcc_info && 2758 atomic_read(&SM_I(sbi)->fcc_info->queued_flush)) 2759 return true; 2760 return false; 2761 } 2762 2763 static inline bool is_idle(struct f2fs_sb_info *sbi, int type) 2764 { 2765 if (sbi->gc_mode == GC_URGENT_HIGH) 2766 return true; 2767 2768 if (is_inflight_io(sbi, type)) 2769 return false; 2770 2771 if (sbi->gc_mode == GC_URGENT_LOW && 2772 (type == DISCARD_TIME || type == GC_TIME)) 2773 return true; 2774 2775 return f2fs_time_over(sbi, type); 2776 } 2777 2778 static inline void f2fs_radix_tree_insert(struct radix_tree_root *root, 2779 unsigned long index, void *item) 2780 { 2781 while (radix_tree_insert(root, index, item)) 2782 cond_resched(); 2783 } 2784 2785 #define RAW_IS_INODE(p) ((p)->footer.nid == (p)->footer.ino) 2786 2787 static inline bool IS_INODE(struct page *page) 2788 { 2789 struct f2fs_node *p = F2FS_NODE(page); 2790 2791 return RAW_IS_INODE(p); 2792 } 2793 2794 static inline int offset_in_addr(struct f2fs_inode *i) 2795 { 2796 return (i->i_inline & F2FS_EXTRA_ATTR) ? 2797 (le16_to_cpu(i->i_extra_isize) / sizeof(__le32)) : 0; 2798 } 2799 2800 static inline __le32 *blkaddr_in_node(struct f2fs_node *node) 2801 { 2802 return RAW_IS_INODE(node) ? node->i.i_addr : node->dn.addr; 2803 } 2804 2805 static inline int f2fs_has_extra_attr(struct inode *inode); 2806 static inline block_t data_blkaddr(struct inode *inode, 2807 struct page *node_page, unsigned int offset) 2808 { 2809 struct f2fs_node *raw_node; 2810 __le32 *addr_array; 2811 int base = 0; 2812 bool is_inode = IS_INODE(node_page); 2813 2814 raw_node = F2FS_NODE(node_page); 2815 2816 if (is_inode) { 2817 if (!inode) 2818 /* from GC path only */ 2819 base = offset_in_addr(&raw_node->i); 2820 else if (f2fs_has_extra_attr(inode)) 2821 base = get_extra_isize(inode); 2822 } 2823 2824 addr_array = blkaddr_in_node(raw_node); 2825 return le32_to_cpu(addr_array[base + offset]); 2826 } 2827 2828 static inline block_t f2fs_data_blkaddr(struct dnode_of_data *dn) 2829 { 2830 return data_blkaddr(dn->inode, dn->node_page, dn->ofs_in_node); 2831 } 2832 2833 static inline int f2fs_test_bit(unsigned int nr, char *addr) 2834 { 2835 int mask; 2836 2837 addr += (nr >> 3); 2838 mask = 1 << (7 - (nr & 0x07)); 2839 return mask & *addr; 2840 } 2841 2842 static inline void f2fs_set_bit(unsigned int nr, char *addr) 2843 { 2844 int mask; 2845 2846 addr += (nr >> 3); 2847 mask = 1 << (7 - (nr & 0x07)); 2848 *addr |= mask; 2849 } 2850 2851 static inline void f2fs_clear_bit(unsigned int nr, char *addr) 2852 { 2853 int mask; 2854 2855 addr += (nr >> 3); 2856 mask = 1 << (7 - (nr & 0x07)); 2857 *addr &= ~mask; 2858 } 2859 2860 static inline int f2fs_test_and_set_bit(unsigned int nr, char *addr) 2861 { 2862 int mask; 2863 int ret; 2864 2865 addr += (nr >> 3); 2866 mask = 1 << (7 - (nr & 0x07)); 2867 ret = mask & *addr; 2868 *addr |= mask; 2869 return ret; 2870 } 2871 2872 static inline int f2fs_test_and_clear_bit(unsigned int nr, char *addr) 2873 { 2874 int mask; 2875 int ret; 2876 2877 addr += (nr >> 3); 2878 mask = 1 << (7 - (nr & 0x07)); 2879 ret = mask & *addr; 2880 *addr &= ~mask; 2881 return ret; 2882 } 2883 2884 static inline void f2fs_change_bit(unsigned int nr, char *addr) 2885 { 2886 int mask; 2887 2888 addr += (nr >> 3); 2889 mask = 1 << (7 - (nr & 0x07)); 2890 *addr ^= mask; 2891 } 2892 2893 /* 2894 * On-disk inode flags (f2fs_inode::i_flags) 2895 */ 2896 #define F2FS_COMPR_FL 0x00000004 /* Compress file */ 2897 #define F2FS_SYNC_FL 0x00000008 /* Synchronous updates */ 2898 #define F2FS_IMMUTABLE_FL 0x00000010 /* Immutable file */ 2899 #define F2FS_APPEND_FL 0x00000020 /* writes to file may only append */ 2900 #define F2FS_NODUMP_FL 0x00000040 /* do not dump file */ 2901 #define F2FS_NOATIME_FL 0x00000080 /* do not update atime */ 2902 #define F2FS_NOCOMP_FL 0x00000400 /* Don't compress */ 2903 #define F2FS_INDEX_FL 0x00001000 /* hash-indexed directory */ 2904 #define F2FS_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */ 2905 #define F2FS_PROJINHERIT_FL 0x20000000 /* Create with parents projid */ 2906 #define F2FS_CASEFOLD_FL 0x40000000 /* Casefolded file */ 2907 2908 /* Flags that should be inherited by new inodes from their parent. */ 2909 #define F2FS_FL_INHERITED (F2FS_SYNC_FL | F2FS_NODUMP_FL | F2FS_NOATIME_FL | \ 2910 F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \ 2911 F2FS_CASEFOLD_FL | F2FS_COMPR_FL | F2FS_NOCOMP_FL) 2912 2913 /* Flags that are appropriate for regular files (all but dir-specific ones). */ 2914 #define F2FS_REG_FLMASK (~(F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \ 2915 F2FS_CASEFOLD_FL)) 2916 2917 /* Flags that are appropriate for non-directories/regular files. */ 2918 #define F2FS_OTHER_FLMASK (F2FS_NODUMP_FL | F2FS_NOATIME_FL) 2919 2920 static inline __u32 f2fs_mask_flags(umode_t mode, __u32 flags) 2921 { 2922 if (S_ISDIR(mode)) 2923 return flags; 2924 else if (S_ISREG(mode)) 2925 return flags & F2FS_REG_FLMASK; 2926 else 2927 return flags & F2FS_OTHER_FLMASK; 2928 } 2929 2930 static inline void __mark_inode_dirty_flag(struct inode *inode, 2931 int flag, bool set) 2932 { 2933 switch (flag) { 2934 case FI_INLINE_XATTR: 2935 case FI_INLINE_DATA: 2936 case FI_INLINE_DENTRY: 2937 case FI_NEW_INODE: 2938 if (set) 2939 return; 2940 fallthrough; 2941 case FI_DATA_EXIST: 2942 case FI_INLINE_DOTS: 2943 case FI_PIN_FILE: 2944 case FI_COMPRESS_RELEASED: 2945 f2fs_mark_inode_dirty_sync(inode, true); 2946 } 2947 } 2948 2949 static inline void set_inode_flag(struct inode *inode, int flag) 2950 { 2951 set_bit(flag, F2FS_I(inode)->flags); 2952 __mark_inode_dirty_flag(inode, flag, true); 2953 } 2954 2955 static inline int is_inode_flag_set(struct inode *inode, int flag) 2956 { 2957 return test_bit(flag, F2FS_I(inode)->flags); 2958 } 2959 2960 static inline void clear_inode_flag(struct inode *inode, int flag) 2961 { 2962 clear_bit(flag, F2FS_I(inode)->flags); 2963 __mark_inode_dirty_flag(inode, flag, false); 2964 } 2965 2966 static inline bool f2fs_verity_in_progress(struct inode *inode) 2967 { 2968 return IS_ENABLED(CONFIG_FS_VERITY) && 2969 is_inode_flag_set(inode, FI_VERITY_IN_PROGRESS); 2970 } 2971 2972 static inline void set_acl_inode(struct inode *inode, umode_t mode) 2973 { 2974 F2FS_I(inode)->i_acl_mode = mode; 2975 set_inode_flag(inode, FI_ACL_MODE); 2976 f2fs_mark_inode_dirty_sync(inode, false); 2977 } 2978 2979 static inline void f2fs_i_links_write(struct inode *inode, bool inc) 2980 { 2981 if (inc) 2982 inc_nlink(inode); 2983 else 2984 drop_nlink(inode); 2985 f2fs_mark_inode_dirty_sync(inode, true); 2986 } 2987 2988 static inline void f2fs_i_blocks_write(struct inode *inode, 2989 block_t diff, bool add, bool claim) 2990 { 2991 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE); 2992 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER); 2993 2994 /* add = 1, claim = 1 should be dquot_reserve_block in pair */ 2995 if (add) { 2996 if (claim) 2997 dquot_claim_block(inode, diff); 2998 else 2999 dquot_alloc_block_nofail(inode, diff); 3000 } else { 3001 dquot_free_block(inode, diff); 3002 } 3003 3004 f2fs_mark_inode_dirty_sync(inode, true); 3005 if (clean || recover) 3006 set_inode_flag(inode, FI_AUTO_RECOVER); 3007 } 3008 3009 static inline void f2fs_i_size_write(struct inode *inode, loff_t i_size) 3010 { 3011 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE); 3012 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER); 3013 3014 if (i_size_read(inode) == i_size) 3015 return; 3016 3017 i_size_write(inode, i_size); 3018 f2fs_mark_inode_dirty_sync(inode, true); 3019 if (clean || recover) 3020 set_inode_flag(inode, FI_AUTO_RECOVER); 3021 } 3022 3023 static inline void f2fs_i_depth_write(struct inode *inode, unsigned int depth) 3024 { 3025 F2FS_I(inode)->i_current_depth = depth; 3026 f2fs_mark_inode_dirty_sync(inode, true); 3027 } 3028 3029 static inline void f2fs_i_gc_failures_write(struct inode *inode, 3030 unsigned int count) 3031 { 3032 F2FS_I(inode)->i_gc_failures[GC_FAILURE_PIN] = count; 3033 f2fs_mark_inode_dirty_sync(inode, true); 3034 } 3035 3036 static inline void f2fs_i_xnid_write(struct inode *inode, nid_t xnid) 3037 { 3038 F2FS_I(inode)->i_xattr_nid = xnid; 3039 f2fs_mark_inode_dirty_sync(inode, true); 3040 } 3041 3042 static inline void f2fs_i_pino_write(struct inode *inode, nid_t pino) 3043 { 3044 F2FS_I(inode)->i_pino = pino; 3045 f2fs_mark_inode_dirty_sync(inode, true); 3046 } 3047 3048 static inline void get_inline_info(struct inode *inode, struct f2fs_inode *ri) 3049 { 3050 struct f2fs_inode_info *fi = F2FS_I(inode); 3051 3052 if (ri->i_inline & F2FS_INLINE_XATTR) 3053 set_bit(FI_INLINE_XATTR, fi->flags); 3054 if (ri->i_inline & F2FS_INLINE_DATA) 3055 set_bit(FI_INLINE_DATA, fi->flags); 3056 if (ri->i_inline & F2FS_INLINE_DENTRY) 3057 set_bit(FI_INLINE_DENTRY, fi->flags); 3058 if (ri->i_inline & F2FS_DATA_EXIST) 3059 set_bit(FI_DATA_EXIST, fi->flags); 3060 if (ri->i_inline & F2FS_INLINE_DOTS) 3061 set_bit(FI_INLINE_DOTS, fi->flags); 3062 if (ri->i_inline & F2FS_EXTRA_ATTR) 3063 set_bit(FI_EXTRA_ATTR, fi->flags); 3064 if (ri->i_inline & F2FS_PIN_FILE) 3065 set_bit(FI_PIN_FILE, fi->flags); 3066 if (ri->i_inline & F2FS_COMPRESS_RELEASED) 3067 set_bit(FI_COMPRESS_RELEASED, fi->flags); 3068 } 3069 3070 static inline void set_raw_inline(struct inode *inode, struct f2fs_inode *ri) 3071 { 3072 ri->i_inline = 0; 3073 3074 if (is_inode_flag_set(inode, FI_INLINE_XATTR)) 3075 ri->i_inline |= F2FS_INLINE_XATTR; 3076 if (is_inode_flag_set(inode, FI_INLINE_DATA)) 3077 ri->i_inline |= F2FS_INLINE_DATA; 3078 if (is_inode_flag_set(inode, FI_INLINE_DENTRY)) 3079 ri->i_inline |= F2FS_INLINE_DENTRY; 3080 if (is_inode_flag_set(inode, FI_DATA_EXIST)) 3081 ri->i_inline |= F2FS_DATA_EXIST; 3082 if (is_inode_flag_set(inode, FI_INLINE_DOTS)) 3083 ri->i_inline |= F2FS_INLINE_DOTS; 3084 if (is_inode_flag_set(inode, FI_EXTRA_ATTR)) 3085 ri->i_inline |= F2FS_EXTRA_ATTR; 3086 if (is_inode_flag_set(inode, FI_PIN_FILE)) 3087 ri->i_inline |= F2FS_PIN_FILE; 3088 if (is_inode_flag_set(inode, FI_COMPRESS_RELEASED)) 3089 ri->i_inline |= F2FS_COMPRESS_RELEASED; 3090 } 3091 3092 static inline int f2fs_has_extra_attr(struct inode *inode) 3093 { 3094 return is_inode_flag_set(inode, FI_EXTRA_ATTR); 3095 } 3096 3097 static inline int f2fs_has_inline_xattr(struct inode *inode) 3098 { 3099 return is_inode_flag_set(inode, FI_INLINE_XATTR); 3100 } 3101 3102 static inline int f2fs_compressed_file(struct inode *inode) 3103 { 3104 return S_ISREG(inode->i_mode) && 3105 is_inode_flag_set(inode, FI_COMPRESSED_FILE); 3106 } 3107 3108 static inline bool f2fs_need_compress_data(struct inode *inode) 3109 { 3110 int compress_mode = F2FS_OPTION(F2FS_I_SB(inode)).compress_mode; 3111 3112 if (!f2fs_compressed_file(inode)) 3113 return false; 3114 3115 if (compress_mode == COMPR_MODE_FS) 3116 return true; 3117 else if (compress_mode == COMPR_MODE_USER && 3118 is_inode_flag_set(inode, FI_ENABLE_COMPRESS)) 3119 return true; 3120 3121 return false; 3122 } 3123 3124 static inline unsigned int addrs_per_inode(struct inode *inode) 3125 { 3126 unsigned int addrs = CUR_ADDRS_PER_INODE(inode) - 3127 get_inline_xattr_addrs(inode); 3128 3129 if (!f2fs_compressed_file(inode)) 3130 return addrs; 3131 return ALIGN_DOWN(addrs, F2FS_I(inode)->i_cluster_size); 3132 } 3133 3134 static inline unsigned int addrs_per_block(struct inode *inode) 3135 { 3136 if (!f2fs_compressed_file(inode)) 3137 return DEF_ADDRS_PER_BLOCK; 3138 return ALIGN_DOWN(DEF_ADDRS_PER_BLOCK, F2FS_I(inode)->i_cluster_size); 3139 } 3140 3141 static inline void *inline_xattr_addr(struct inode *inode, struct page *page) 3142 { 3143 struct f2fs_inode *ri = F2FS_INODE(page); 3144 3145 return (void *)&(ri->i_addr[DEF_ADDRS_PER_INODE - 3146 get_inline_xattr_addrs(inode)]); 3147 } 3148 3149 static inline int inline_xattr_size(struct inode *inode) 3150 { 3151 if (f2fs_has_inline_xattr(inode)) 3152 return get_inline_xattr_addrs(inode) * sizeof(__le32); 3153 return 0; 3154 } 3155 3156 static inline int f2fs_has_inline_data(struct inode *inode) 3157 { 3158 return is_inode_flag_set(inode, FI_INLINE_DATA); 3159 } 3160 3161 static inline int f2fs_exist_data(struct inode *inode) 3162 { 3163 return is_inode_flag_set(inode, FI_DATA_EXIST); 3164 } 3165 3166 static inline int f2fs_has_inline_dots(struct inode *inode) 3167 { 3168 return is_inode_flag_set(inode, FI_INLINE_DOTS); 3169 } 3170 3171 static inline int f2fs_is_mmap_file(struct inode *inode) 3172 { 3173 return is_inode_flag_set(inode, FI_MMAP_FILE); 3174 } 3175 3176 static inline bool f2fs_is_pinned_file(struct inode *inode) 3177 { 3178 return is_inode_flag_set(inode, FI_PIN_FILE); 3179 } 3180 3181 static inline bool f2fs_is_atomic_file(struct inode *inode) 3182 { 3183 return is_inode_flag_set(inode, FI_ATOMIC_FILE); 3184 } 3185 3186 static inline bool f2fs_is_commit_atomic_write(struct inode *inode) 3187 { 3188 return is_inode_flag_set(inode, FI_ATOMIC_COMMIT); 3189 } 3190 3191 static inline bool f2fs_is_volatile_file(struct inode *inode) 3192 { 3193 return is_inode_flag_set(inode, FI_VOLATILE_FILE); 3194 } 3195 3196 static inline bool f2fs_is_first_block_written(struct inode *inode) 3197 { 3198 return is_inode_flag_set(inode, FI_FIRST_BLOCK_WRITTEN); 3199 } 3200 3201 static inline bool f2fs_is_drop_cache(struct inode *inode) 3202 { 3203 return is_inode_flag_set(inode, FI_DROP_CACHE); 3204 } 3205 3206 static inline void *inline_data_addr(struct inode *inode, struct page *page) 3207 { 3208 struct f2fs_inode *ri = F2FS_INODE(page); 3209 int extra_size = get_extra_isize(inode); 3210 3211 return (void *)&(ri->i_addr[extra_size + DEF_INLINE_RESERVED_SIZE]); 3212 } 3213 3214 static inline int f2fs_has_inline_dentry(struct inode *inode) 3215 { 3216 return is_inode_flag_set(inode, FI_INLINE_DENTRY); 3217 } 3218 3219 static inline int is_file(struct inode *inode, int type) 3220 { 3221 return F2FS_I(inode)->i_advise & type; 3222 } 3223 3224 static inline void set_file(struct inode *inode, int type) 3225 { 3226 if (is_file(inode, type)) 3227 return; 3228 F2FS_I(inode)->i_advise |= type; 3229 f2fs_mark_inode_dirty_sync(inode, true); 3230 } 3231 3232 static inline void clear_file(struct inode *inode, int type) 3233 { 3234 if (!is_file(inode, type)) 3235 return; 3236 F2FS_I(inode)->i_advise &= ~type; 3237 f2fs_mark_inode_dirty_sync(inode, true); 3238 } 3239 3240 static inline bool f2fs_is_time_consistent(struct inode *inode) 3241 { 3242 if (!timespec64_equal(F2FS_I(inode)->i_disk_time, &inode->i_atime)) 3243 return false; 3244 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 1, &inode->i_ctime)) 3245 return false; 3246 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 2, &inode->i_mtime)) 3247 return false; 3248 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 3, 3249 &F2FS_I(inode)->i_crtime)) 3250 return false; 3251 return true; 3252 } 3253 3254 static inline bool f2fs_skip_inode_update(struct inode *inode, int dsync) 3255 { 3256 bool ret; 3257 3258 if (dsync) { 3259 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 3260 3261 spin_lock(&sbi->inode_lock[DIRTY_META]); 3262 ret = list_empty(&F2FS_I(inode)->gdirty_list); 3263 spin_unlock(&sbi->inode_lock[DIRTY_META]); 3264 return ret; 3265 } 3266 if (!is_inode_flag_set(inode, FI_AUTO_RECOVER) || 3267 file_keep_isize(inode) || 3268 i_size_read(inode) & ~PAGE_MASK) 3269 return false; 3270 3271 if (!f2fs_is_time_consistent(inode)) 3272 return false; 3273 3274 spin_lock(&F2FS_I(inode)->i_size_lock); 3275 ret = F2FS_I(inode)->last_disk_size == i_size_read(inode); 3276 spin_unlock(&F2FS_I(inode)->i_size_lock); 3277 3278 return ret; 3279 } 3280 3281 static inline bool f2fs_readonly(struct super_block *sb) 3282 { 3283 return sb_rdonly(sb); 3284 } 3285 3286 static inline bool f2fs_cp_error(struct f2fs_sb_info *sbi) 3287 { 3288 return is_set_ckpt_flags(sbi, CP_ERROR_FLAG); 3289 } 3290 3291 static inline bool is_dot_dotdot(const u8 *name, size_t len) 3292 { 3293 if (len == 1 && name[0] == '.') 3294 return true; 3295 3296 if (len == 2 && name[0] == '.' && name[1] == '.') 3297 return true; 3298 3299 return false; 3300 } 3301 3302 static inline void *f2fs_kmalloc(struct f2fs_sb_info *sbi, 3303 size_t size, gfp_t flags) 3304 { 3305 if (time_to_inject(sbi, FAULT_KMALLOC)) { 3306 f2fs_show_injection_info(sbi, FAULT_KMALLOC); 3307 return NULL; 3308 } 3309 3310 return kmalloc(size, flags); 3311 } 3312 3313 static inline void *f2fs_kzalloc(struct f2fs_sb_info *sbi, 3314 size_t size, gfp_t flags) 3315 { 3316 return f2fs_kmalloc(sbi, size, flags | __GFP_ZERO); 3317 } 3318 3319 static inline void *f2fs_kvmalloc(struct f2fs_sb_info *sbi, 3320 size_t size, gfp_t flags) 3321 { 3322 if (time_to_inject(sbi, FAULT_KVMALLOC)) { 3323 f2fs_show_injection_info(sbi, FAULT_KVMALLOC); 3324 return NULL; 3325 } 3326 3327 return kvmalloc(size, flags); 3328 } 3329 3330 static inline void *f2fs_kvzalloc(struct f2fs_sb_info *sbi, 3331 size_t size, gfp_t flags) 3332 { 3333 return f2fs_kvmalloc(sbi, size, flags | __GFP_ZERO); 3334 } 3335 3336 static inline int get_extra_isize(struct inode *inode) 3337 { 3338 return F2FS_I(inode)->i_extra_isize / sizeof(__le32); 3339 } 3340 3341 static inline int get_inline_xattr_addrs(struct inode *inode) 3342 { 3343 return F2FS_I(inode)->i_inline_xattr_size; 3344 } 3345 3346 #define f2fs_get_inode_mode(i) \ 3347 ((is_inode_flag_set(i, FI_ACL_MODE)) ? \ 3348 (F2FS_I(i)->i_acl_mode) : ((i)->i_mode)) 3349 3350 #define F2FS_TOTAL_EXTRA_ATTR_SIZE \ 3351 (offsetof(struct f2fs_inode, i_extra_end) - \ 3352 offsetof(struct f2fs_inode, i_extra_isize)) \ 3353 3354 #define F2FS_OLD_ATTRIBUTE_SIZE (offsetof(struct f2fs_inode, i_addr)) 3355 #define F2FS_FITS_IN_INODE(f2fs_inode, extra_isize, field) \ 3356 ((offsetof(typeof(*(f2fs_inode)), field) + \ 3357 sizeof((f2fs_inode)->field)) \ 3358 <= (F2FS_OLD_ATTRIBUTE_SIZE + (extra_isize))) \ 3359 3360 #define __is_large_section(sbi) ((sbi)->segs_per_sec > 1) 3361 3362 #define __is_meta_io(fio) (PAGE_TYPE_OF_BIO((fio)->type) == META) 3363 3364 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi, 3365 block_t blkaddr, int type); 3366 static inline void verify_blkaddr(struct f2fs_sb_info *sbi, 3367 block_t blkaddr, int type) 3368 { 3369 if (!f2fs_is_valid_blkaddr(sbi, blkaddr, type)) { 3370 f2fs_err(sbi, "invalid blkaddr: %u, type: %d, run fsck to fix.", 3371 blkaddr, type); 3372 f2fs_bug_on(sbi, 1); 3373 } 3374 } 3375 3376 static inline bool __is_valid_data_blkaddr(block_t blkaddr) 3377 { 3378 if (blkaddr == NEW_ADDR || blkaddr == NULL_ADDR || 3379 blkaddr == COMPRESS_ADDR) 3380 return false; 3381 return true; 3382 } 3383 3384 /* 3385 * file.c 3386 */ 3387 int f2fs_sync_file(struct file *file, loff_t start, loff_t end, int datasync); 3388 void f2fs_truncate_data_blocks(struct dnode_of_data *dn); 3389 int f2fs_do_truncate_blocks(struct inode *inode, u64 from, bool lock); 3390 int f2fs_truncate_blocks(struct inode *inode, u64 from, bool lock); 3391 int f2fs_truncate(struct inode *inode); 3392 int f2fs_getattr(struct user_namespace *mnt_userns, const struct path *path, 3393 struct kstat *stat, u32 request_mask, unsigned int flags); 3394 int f2fs_setattr(struct user_namespace *mnt_userns, struct dentry *dentry, 3395 struct iattr *attr); 3396 int f2fs_truncate_hole(struct inode *inode, pgoff_t pg_start, pgoff_t pg_end); 3397 void f2fs_truncate_data_blocks_range(struct dnode_of_data *dn, int count); 3398 int f2fs_precache_extents(struct inode *inode); 3399 int f2fs_fileattr_get(struct dentry *dentry, struct fileattr *fa); 3400 int f2fs_fileattr_set(struct user_namespace *mnt_userns, 3401 struct dentry *dentry, struct fileattr *fa); 3402 long f2fs_ioctl(struct file *filp, unsigned int cmd, unsigned long arg); 3403 long f2fs_compat_ioctl(struct file *file, unsigned int cmd, unsigned long arg); 3404 int f2fs_transfer_project_quota(struct inode *inode, kprojid_t kprojid); 3405 int f2fs_pin_file_control(struct inode *inode, bool inc); 3406 3407 /* 3408 * inode.c 3409 */ 3410 void f2fs_set_inode_flags(struct inode *inode); 3411 bool f2fs_inode_chksum_verify(struct f2fs_sb_info *sbi, struct page *page); 3412 void f2fs_inode_chksum_set(struct f2fs_sb_info *sbi, struct page *page); 3413 struct inode *f2fs_iget(struct super_block *sb, unsigned long ino); 3414 struct inode *f2fs_iget_retry(struct super_block *sb, unsigned long ino); 3415 int f2fs_try_to_free_nats(struct f2fs_sb_info *sbi, int nr_shrink); 3416 void f2fs_update_inode(struct inode *inode, struct page *node_page); 3417 void f2fs_update_inode_page(struct inode *inode); 3418 int f2fs_write_inode(struct inode *inode, struct writeback_control *wbc); 3419 void f2fs_evict_inode(struct inode *inode); 3420 void f2fs_handle_failed_inode(struct inode *inode); 3421 3422 /* 3423 * namei.c 3424 */ 3425 int f2fs_update_extension_list(struct f2fs_sb_info *sbi, const char *name, 3426 bool hot, bool set); 3427 struct dentry *f2fs_get_parent(struct dentry *child); 3428 3429 /* 3430 * dir.c 3431 */ 3432 unsigned char f2fs_get_de_type(struct f2fs_dir_entry *de); 3433 int f2fs_init_casefolded_name(const struct inode *dir, 3434 struct f2fs_filename *fname); 3435 int f2fs_setup_filename(struct inode *dir, const struct qstr *iname, 3436 int lookup, struct f2fs_filename *fname); 3437 int f2fs_prepare_lookup(struct inode *dir, struct dentry *dentry, 3438 struct f2fs_filename *fname); 3439 void f2fs_free_filename(struct f2fs_filename *fname); 3440 struct f2fs_dir_entry *f2fs_find_target_dentry(const struct f2fs_dentry_ptr *d, 3441 const struct f2fs_filename *fname, int *max_slots); 3442 int f2fs_fill_dentries(struct dir_context *ctx, struct f2fs_dentry_ptr *d, 3443 unsigned int start_pos, struct fscrypt_str *fstr); 3444 void f2fs_do_make_empty_dir(struct inode *inode, struct inode *parent, 3445 struct f2fs_dentry_ptr *d); 3446 struct page *f2fs_init_inode_metadata(struct inode *inode, struct inode *dir, 3447 const struct f2fs_filename *fname, struct page *dpage); 3448 void f2fs_update_parent_metadata(struct inode *dir, struct inode *inode, 3449 unsigned int current_depth); 3450 int f2fs_room_for_filename(const void *bitmap, int slots, int max_slots); 3451 void f2fs_drop_nlink(struct inode *dir, struct inode *inode); 3452 struct f2fs_dir_entry *__f2fs_find_entry(struct inode *dir, 3453 const struct f2fs_filename *fname, 3454 struct page **res_page); 3455 struct f2fs_dir_entry *f2fs_find_entry(struct inode *dir, 3456 const struct qstr *child, struct page **res_page); 3457 struct f2fs_dir_entry *f2fs_parent_dir(struct inode *dir, struct page **p); 3458 ino_t f2fs_inode_by_name(struct inode *dir, const struct qstr *qstr, 3459 struct page **page); 3460 void f2fs_set_link(struct inode *dir, struct f2fs_dir_entry *de, 3461 struct page *page, struct inode *inode); 3462 bool f2fs_has_enough_room(struct inode *dir, struct page *ipage, 3463 const struct f2fs_filename *fname); 3464 void f2fs_update_dentry(nid_t ino, umode_t mode, struct f2fs_dentry_ptr *d, 3465 const struct fscrypt_str *name, f2fs_hash_t name_hash, 3466 unsigned int bit_pos); 3467 int f2fs_add_regular_entry(struct inode *dir, const struct f2fs_filename *fname, 3468 struct inode *inode, nid_t ino, umode_t mode); 3469 int f2fs_add_dentry(struct inode *dir, const struct f2fs_filename *fname, 3470 struct inode *inode, nid_t ino, umode_t mode); 3471 int f2fs_do_add_link(struct inode *dir, const struct qstr *name, 3472 struct inode *inode, nid_t ino, umode_t mode); 3473 void f2fs_delete_entry(struct f2fs_dir_entry *dentry, struct page *page, 3474 struct inode *dir, struct inode *inode); 3475 int f2fs_do_tmpfile(struct inode *inode, struct inode *dir); 3476 bool f2fs_empty_dir(struct inode *dir); 3477 3478 static inline int f2fs_add_link(struct dentry *dentry, struct inode *inode) 3479 { 3480 if (fscrypt_is_nokey_name(dentry)) 3481 return -ENOKEY; 3482 return f2fs_do_add_link(d_inode(dentry->d_parent), &dentry->d_name, 3483 inode, inode->i_ino, inode->i_mode); 3484 } 3485 3486 /* 3487 * super.c 3488 */ 3489 int f2fs_inode_dirtied(struct inode *inode, bool sync); 3490 void f2fs_inode_synced(struct inode *inode); 3491 int f2fs_dquot_initialize(struct inode *inode); 3492 int f2fs_enable_quota_files(struct f2fs_sb_info *sbi, bool rdonly); 3493 int f2fs_quota_sync(struct super_block *sb, int type); 3494 loff_t max_file_blocks(struct inode *inode); 3495 void f2fs_quota_off_umount(struct super_block *sb); 3496 int f2fs_commit_super(struct f2fs_sb_info *sbi, bool recover); 3497 int f2fs_sync_fs(struct super_block *sb, int sync); 3498 int f2fs_sanity_check_ckpt(struct f2fs_sb_info *sbi); 3499 3500 /* 3501 * hash.c 3502 */ 3503 void f2fs_hash_filename(const struct inode *dir, struct f2fs_filename *fname); 3504 3505 /* 3506 * node.c 3507 */ 3508 struct node_info; 3509 3510 int f2fs_check_nid_range(struct f2fs_sb_info *sbi, nid_t nid); 3511 bool f2fs_available_free_memory(struct f2fs_sb_info *sbi, int type); 3512 bool f2fs_in_warm_node_list(struct f2fs_sb_info *sbi, struct page *page); 3513 void f2fs_init_fsync_node_info(struct f2fs_sb_info *sbi); 3514 void f2fs_del_fsync_node_entry(struct f2fs_sb_info *sbi, struct page *page); 3515 void f2fs_reset_fsync_node_info(struct f2fs_sb_info *sbi); 3516 int f2fs_need_dentry_mark(struct f2fs_sb_info *sbi, nid_t nid); 3517 bool f2fs_is_checkpointed_node(struct f2fs_sb_info *sbi, nid_t nid); 3518 bool f2fs_need_inode_block_update(struct f2fs_sb_info *sbi, nid_t ino); 3519 int f2fs_get_node_info(struct f2fs_sb_info *sbi, nid_t nid, 3520 struct node_info *ni, bool checkpoint_context); 3521 pgoff_t f2fs_get_next_page_offset(struct dnode_of_data *dn, pgoff_t pgofs); 3522 int f2fs_get_dnode_of_data(struct dnode_of_data *dn, pgoff_t index, int mode); 3523 int f2fs_truncate_inode_blocks(struct inode *inode, pgoff_t from); 3524 int f2fs_truncate_xattr_node(struct inode *inode); 3525 int f2fs_wait_on_node_pages_writeback(struct f2fs_sb_info *sbi, 3526 unsigned int seq_id); 3527 bool f2fs_nat_bitmap_enabled(struct f2fs_sb_info *sbi); 3528 int f2fs_remove_inode_page(struct inode *inode); 3529 struct page *f2fs_new_inode_page(struct inode *inode); 3530 struct page *f2fs_new_node_page(struct dnode_of_data *dn, unsigned int ofs); 3531 void f2fs_ra_node_page(struct f2fs_sb_info *sbi, nid_t nid); 3532 struct page *f2fs_get_node_page(struct f2fs_sb_info *sbi, pgoff_t nid); 3533 struct page *f2fs_get_node_page_ra(struct page *parent, int start); 3534 int f2fs_move_node_page(struct page *node_page, int gc_type); 3535 void f2fs_flush_inline_data(struct f2fs_sb_info *sbi); 3536 int f2fs_fsync_node_pages(struct f2fs_sb_info *sbi, struct inode *inode, 3537 struct writeback_control *wbc, bool atomic, 3538 unsigned int *seq_id); 3539 int f2fs_sync_node_pages(struct f2fs_sb_info *sbi, 3540 struct writeback_control *wbc, 3541 bool do_balance, enum iostat_type io_type); 3542 int f2fs_build_free_nids(struct f2fs_sb_info *sbi, bool sync, bool mount); 3543 bool f2fs_alloc_nid(struct f2fs_sb_info *sbi, nid_t *nid); 3544 void f2fs_alloc_nid_done(struct f2fs_sb_info *sbi, nid_t nid); 3545 void f2fs_alloc_nid_failed(struct f2fs_sb_info *sbi, nid_t nid); 3546 int f2fs_try_to_free_nids(struct f2fs_sb_info *sbi, int nr_shrink); 3547 int f2fs_recover_inline_xattr(struct inode *inode, struct page *page); 3548 int f2fs_recover_xattr_data(struct inode *inode, struct page *page); 3549 int f2fs_recover_inode_page(struct f2fs_sb_info *sbi, struct page *page); 3550 int f2fs_restore_node_summary(struct f2fs_sb_info *sbi, 3551 unsigned int segno, struct f2fs_summary_block *sum); 3552 void f2fs_enable_nat_bits(struct f2fs_sb_info *sbi); 3553 int f2fs_flush_nat_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3554 int f2fs_build_node_manager(struct f2fs_sb_info *sbi); 3555 void f2fs_destroy_node_manager(struct f2fs_sb_info *sbi); 3556 int __init f2fs_create_node_manager_caches(void); 3557 void f2fs_destroy_node_manager_caches(void); 3558 3559 /* 3560 * segment.c 3561 */ 3562 bool f2fs_need_SSR(struct f2fs_sb_info *sbi); 3563 void f2fs_register_inmem_page(struct inode *inode, struct page *page); 3564 void f2fs_drop_inmem_pages_all(struct f2fs_sb_info *sbi, bool gc_failure); 3565 void f2fs_drop_inmem_pages(struct inode *inode); 3566 void f2fs_drop_inmem_page(struct inode *inode, struct page *page); 3567 int f2fs_commit_inmem_pages(struct inode *inode); 3568 void f2fs_balance_fs(struct f2fs_sb_info *sbi, bool need); 3569 void f2fs_balance_fs_bg(struct f2fs_sb_info *sbi, bool from_bg); 3570 int f2fs_issue_flush(struct f2fs_sb_info *sbi, nid_t ino); 3571 int f2fs_create_flush_cmd_control(struct f2fs_sb_info *sbi); 3572 int f2fs_flush_device_cache(struct f2fs_sb_info *sbi); 3573 void f2fs_destroy_flush_cmd_control(struct f2fs_sb_info *sbi, bool free); 3574 void f2fs_invalidate_blocks(struct f2fs_sb_info *sbi, block_t addr); 3575 bool f2fs_is_checkpointed_data(struct f2fs_sb_info *sbi, block_t blkaddr); 3576 int f2fs_start_discard_thread(struct f2fs_sb_info *sbi); 3577 void f2fs_drop_discard_cmd(struct f2fs_sb_info *sbi); 3578 void f2fs_stop_discard_thread(struct f2fs_sb_info *sbi); 3579 bool f2fs_issue_discard_timeout(struct f2fs_sb_info *sbi); 3580 void f2fs_clear_prefree_segments(struct f2fs_sb_info *sbi, 3581 struct cp_control *cpc); 3582 void f2fs_dirty_to_prefree(struct f2fs_sb_info *sbi); 3583 block_t f2fs_get_unusable_blocks(struct f2fs_sb_info *sbi); 3584 int f2fs_disable_cp_again(struct f2fs_sb_info *sbi, block_t unusable); 3585 void f2fs_release_discard_addrs(struct f2fs_sb_info *sbi); 3586 int f2fs_npages_for_summary_flush(struct f2fs_sb_info *sbi, bool for_ra); 3587 bool f2fs_segment_has_free_slot(struct f2fs_sb_info *sbi, int segno); 3588 void f2fs_init_inmem_curseg(struct f2fs_sb_info *sbi); 3589 void f2fs_save_inmem_curseg(struct f2fs_sb_info *sbi); 3590 void f2fs_restore_inmem_curseg(struct f2fs_sb_info *sbi); 3591 void f2fs_get_new_segment(struct f2fs_sb_info *sbi, 3592 unsigned int *newseg, bool new_sec, int dir); 3593 void f2fs_allocate_segment_for_resize(struct f2fs_sb_info *sbi, int type, 3594 unsigned int start, unsigned int end); 3595 void f2fs_allocate_new_section(struct f2fs_sb_info *sbi, int type, bool force); 3596 void f2fs_allocate_new_segments(struct f2fs_sb_info *sbi); 3597 int f2fs_trim_fs(struct f2fs_sb_info *sbi, struct fstrim_range *range); 3598 bool f2fs_exist_trim_candidates(struct f2fs_sb_info *sbi, 3599 struct cp_control *cpc); 3600 struct page *f2fs_get_sum_page(struct f2fs_sb_info *sbi, unsigned int segno); 3601 void f2fs_update_meta_page(struct f2fs_sb_info *sbi, void *src, 3602 block_t blk_addr); 3603 void f2fs_do_write_meta_page(struct f2fs_sb_info *sbi, struct page *page, 3604 enum iostat_type io_type); 3605 void f2fs_do_write_node_page(unsigned int nid, struct f2fs_io_info *fio); 3606 void f2fs_outplace_write_data(struct dnode_of_data *dn, 3607 struct f2fs_io_info *fio); 3608 int f2fs_inplace_write_data(struct f2fs_io_info *fio); 3609 void f2fs_do_replace_block(struct f2fs_sb_info *sbi, struct f2fs_summary *sum, 3610 block_t old_blkaddr, block_t new_blkaddr, 3611 bool recover_curseg, bool recover_newaddr, 3612 bool from_gc); 3613 void f2fs_replace_block(struct f2fs_sb_info *sbi, struct dnode_of_data *dn, 3614 block_t old_addr, block_t new_addr, 3615 unsigned char version, bool recover_curseg, 3616 bool recover_newaddr); 3617 void f2fs_allocate_data_block(struct f2fs_sb_info *sbi, struct page *page, 3618 block_t old_blkaddr, block_t *new_blkaddr, 3619 struct f2fs_summary *sum, int type, 3620 struct f2fs_io_info *fio); 3621 void f2fs_update_device_state(struct f2fs_sb_info *sbi, nid_t ino, 3622 block_t blkaddr, unsigned int blkcnt); 3623 void f2fs_wait_on_page_writeback(struct page *page, 3624 enum page_type type, bool ordered, bool locked); 3625 void f2fs_wait_on_block_writeback(struct inode *inode, block_t blkaddr); 3626 void f2fs_wait_on_block_writeback_range(struct inode *inode, block_t blkaddr, 3627 block_t len); 3628 void f2fs_write_data_summaries(struct f2fs_sb_info *sbi, block_t start_blk); 3629 void f2fs_write_node_summaries(struct f2fs_sb_info *sbi, block_t start_blk); 3630 int f2fs_lookup_journal_in_cursum(struct f2fs_journal *journal, int type, 3631 unsigned int val, int alloc); 3632 void f2fs_flush_sit_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3633 int f2fs_fix_curseg_write_pointer(struct f2fs_sb_info *sbi); 3634 int f2fs_check_write_pointer(struct f2fs_sb_info *sbi); 3635 int f2fs_build_segment_manager(struct f2fs_sb_info *sbi); 3636 void f2fs_destroy_segment_manager(struct f2fs_sb_info *sbi); 3637 int __init f2fs_create_segment_manager_caches(void); 3638 void f2fs_destroy_segment_manager_caches(void); 3639 int f2fs_rw_hint_to_seg_type(enum rw_hint hint); 3640 enum rw_hint f2fs_io_type_to_rw_hint(struct f2fs_sb_info *sbi, 3641 enum page_type type, enum temp_type temp); 3642 unsigned int f2fs_usable_segs_in_sec(struct f2fs_sb_info *sbi, 3643 unsigned int segno); 3644 unsigned int f2fs_usable_blks_in_seg(struct f2fs_sb_info *sbi, 3645 unsigned int segno); 3646 3647 #define DEF_FRAGMENT_SIZE 4 3648 #define MIN_FRAGMENT_SIZE 1 3649 #define MAX_FRAGMENT_SIZE 512 3650 3651 static inline bool f2fs_need_rand_seg(struct f2fs_sb_info *sbi) 3652 { 3653 return F2FS_OPTION(sbi).fs_mode == FS_MODE_FRAGMENT_SEG || 3654 F2FS_OPTION(sbi).fs_mode == FS_MODE_FRAGMENT_BLK; 3655 } 3656 3657 /* 3658 * checkpoint.c 3659 */ 3660 void f2fs_stop_checkpoint(struct f2fs_sb_info *sbi, bool end_io); 3661 struct page *f2fs_grab_meta_page(struct f2fs_sb_info *sbi, pgoff_t index); 3662 struct page *f2fs_get_meta_page(struct f2fs_sb_info *sbi, pgoff_t index); 3663 struct page *f2fs_get_meta_page_retry(struct f2fs_sb_info *sbi, pgoff_t index); 3664 struct page *f2fs_get_tmp_page(struct f2fs_sb_info *sbi, pgoff_t index); 3665 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi, 3666 block_t blkaddr, int type); 3667 int f2fs_ra_meta_pages(struct f2fs_sb_info *sbi, block_t start, int nrpages, 3668 int type, bool sync); 3669 void f2fs_ra_meta_pages_cond(struct f2fs_sb_info *sbi, pgoff_t index, 3670 unsigned int ra_blocks); 3671 long f2fs_sync_meta_pages(struct f2fs_sb_info *sbi, enum page_type type, 3672 long nr_to_write, enum iostat_type io_type); 3673 void f2fs_add_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type); 3674 void f2fs_remove_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type); 3675 void f2fs_release_ino_entry(struct f2fs_sb_info *sbi, bool all); 3676 bool f2fs_exist_written_data(struct f2fs_sb_info *sbi, nid_t ino, int mode); 3677 void f2fs_set_dirty_device(struct f2fs_sb_info *sbi, nid_t ino, 3678 unsigned int devidx, int type); 3679 bool f2fs_is_dirty_device(struct f2fs_sb_info *sbi, nid_t ino, 3680 unsigned int devidx, int type); 3681 int f2fs_sync_inode_meta(struct f2fs_sb_info *sbi); 3682 int f2fs_acquire_orphan_inode(struct f2fs_sb_info *sbi); 3683 void f2fs_release_orphan_inode(struct f2fs_sb_info *sbi); 3684 void f2fs_add_orphan_inode(struct inode *inode); 3685 void f2fs_remove_orphan_inode(struct f2fs_sb_info *sbi, nid_t ino); 3686 int f2fs_recover_orphan_inodes(struct f2fs_sb_info *sbi); 3687 int f2fs_get_valid_checkpoint(struct f2fs_sb_info *sbi); 3688 void f2fs_update_dirty_page(struct inode *inode, struct page *page); 3689 void f2fs_remove_dirty_inode(struct inode *inode); 3690 int f2fs_sync_dirty_inodes(struct f2fs_sb_info *sbi, enum inode_type type); 3691 void f2fs_wait_on_all_pages(struct f2fs_sb_info *sbi, int type); 3692 u64 f2fs_get_sectors_written(struct f2fs_sb_info *sbi); 3693 int f2fs_write_checkpoint(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3694 void f2fs_init_ino_entry_info(struct f2fs_sb_info *sbi); 3695 int __init f2fs_create_checkpoint_caches(void); 3696 void f2fs_destroy_checkpoint_caches(void); 3697 int f2fs_issue_checkpoint(struct f2fs_sb_info *sbi); 3698 int f2fs_start_ckpt_thread(struct f2fs_sb_info *sbi); 3699 void f2fs_stop_ckpt_thread(struct f2fs_sb_info *sbi); 3700 void f2fs_init_ckpt_req_control(struct f2fs_sb_info *sbi); 3701 3702 /* 3703 * data.c 3704 */ 3705 int __init f2fs_init_bioset(void); 3706 void f2fs_destroy_bioset(void); 3707 int f2fs_init_bio_entry_cache(void); 3708 void f2fs_destroy_bio_entry_cache(void); 3709 void f2fs_submit_bio(struct f2fs_sb_info *sbi, 3710 struct bio *bio, enum page_type type); 3711 void f2fs_submit_merged_write(struct f2fs_sb_info *sbi, enum page_type type); 3712 void f2fs_submit_merged_write_cond(struct f2fs_sb_info *sbi, 3713 struct inode *inode, struct page *page, 3714 nid_t ino, enum page_type type); 3715 void f2fs_submit_merged_ipu_write(struct f2fs_sb_info *sbi, 3716 struct bio **bio, struct page *page); 3717 void f2fs_flush_merged_writes(struct f2fs_sb_info *sbi); 3718 int f2fs_submit_page_bio(struct f2fs_io_info *fio); 3719 int f2fs_merge_page_bio(struct f2fs_io_info *fio); 3720 void f2fs_submit_page_write(struct f2fs_io_info *fio); 3721 struct block_device *f2fs_target_device(struct f2fs_sb_info *sbi, 3722 block_t blk_addr, struct bio *bio); 3723 int f2fs_target_device_index(struct f2fs_sb_info *sbi, block_t blkaddr); 3724 void f2fs_set_data_blkaddr(struct dnode_of_data *dn); 3725 void f2fs_update_data_blkaddr(struct dnode_of_data *dn, block_t blkaddr); 3726 int f2fs_reserve_new_blocks(struct dnode_of_data *dn, blkcnt_t count); 3727 int f2fs_reserve_new_block(struct dnode_of_data *dn); 3728 int f2fs_get_block(struct dnode_of_data *dn, pgoff_t index); 3729 int f2fs_reserve_block(struct dnode_of_data *dn, pgoff_t index); 3730 struct page *f2fs_get_read_data_page(struct inode *inode, pgoff_t index, 3731 int op_flags, bool for_write); 3732 struct page *f2fs_find_data_page(struct inode *inode, pgoff_t index); 3733 struct page *f2fs_get_lock_data_page(struct inode *inode, pgoff_t index, 3734 bool for_write); 3735 struct page *f2fs_get_new_data_page(struct inode *inode, 3736 struct page *ipage, pgoff_t index, bool new_i_size); 3737 int f2fs_do_write_data_page(struct f2fs_io_info *fio); 3738 void f2fs_do_map_lock(struct f2fs_sb_info *sbi, int flag, bool lock); 3739 int f2fs_map_blocks(struct inode *inode, struct f2fs_map_blocks *map, 3740 int create, int flag); 3741 int f2fs_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, 3742 u64 start, u64 len); 3743 int f2fs_encrypt_one_page(struct f2fs_io_info *fio); 3744 bool f2fs_should_update_inplace(struct inode *inode, struct f2fs_io_info *fio); 3745 bool f2fs_should_update_outplace(struct inode *inode, struct f2fs_io_info *fio); 3746 int f2fs_write_single_data_page(struct page *page, int *submitted, 3747 struct bio **bio, sector_t *last_block, 3748 struct writeback_control *wbc, 3749 enum iostat_type io_type, 3750 int compr_blocks, bool allow_balance); 3751 void f2fs_write_failed(struct inode *inode, loff_t to); 3752 void f2fs_invalidate_page(struct page *page, unsigned int offset, 3753 unsigned int length); 3754 int f2fs_release_page(struct page *page, gfp_t wait); 3755 #ifdef CONFIG_MIGRATION 3756 int f2fs_migrate_page(struct address_space *mapping, struct page *newpage, 3757 struct page *page, enum migrate_mode mode); 3758 #endif 3759 bool f2fs_overwrite_io(struct inode *inode, loff_t pos, size_t len); 3760 void f2fs_clear_page_cache_dirty_tag(struct page *page); 3761 int f2fs_init_post_read_processing(void); 3762 void f2fs_destroy_post_read_processing(void); 3763 int f2fs_init_post_read_wq(struct f2fs_sb_info *sbi); 3764 void f2fs_destroy_post_read_wq(struct f2fs_sb_info *sbi); 3765 extern const struct iomap_ops f2fs_iomap_ops; 3766 3767 /* 3768 * gc.c 3769 */ 3770 int f2fs_start_gc_thread(struct f2fs_sb_info *sbi); 3771 void f2fs_stop_gc_thread(struct f2fs_sb_info *sbi); 3772 block_t f2fs_start_bidx_of_node(unsigned int node_ofs, struct inode *inode); 3773 int f2fs_gc(struct f2fs_sb_info *sbi, bool sync, bool background, bool force, 3774 unsigned int segno); 3775 void f2fs_build_gc_manager(struct f2fs_sb_info *sbi); 3776 int f2fs_resize_fs(struct f2fs_sb_info *sbi, __u64 block_count); 3777 int __init f2fs_create_garbage_collection_cache(void); 3778 void f2fs_destroy_garbage_collection_cache(void); 3779 3780 /* 3781 * recovery.c 3782 */ 3783 int f2fs_recover_fsync_data(struct f2fs_sb_info *sbi, bool check_only); 3784 bool f2fs_space_for_roll_forward(struct f2fs_sb_info *sbi); 3785 int __init f2fs_create_recovery_cache(void); 3786 void f2fs_destroy_recovery_cache(void); 3787 3788 /* 3789 * debug.c 3790 */ 3791 #ifdef CONFIG_F2FS_STAT_FS 3792 struct f2fs_stat_info { 3793 struct list_head stat_list; 3794 struct f2fs_sb_info *sbi; 3795 int all_area_segs, sit_area_segs, nat_area_segs, ssa_area_segs; 3796 int main_area_segs, main_area_sections, main_area_zones; 3797 unsigned long long hit_largest, hit_cached, hit_rbtree; 3798 unsigned long long hit_total, total_ext; 3799 int ext_tree, zombie_tree, ext_node; 3800 int ndirty_node, ndirty_dent, ndirty_meta, ndirty_imeta; 3801 int ndirty_data, ndirty_qdata; 3802 int inmem_pages; 3803 unsigned int ndirty_dirs, ndirty_files, nquota_files, ndirty_all; 3804 int nats, dirty_nats, sits, dirty_sits; 3805 int free_nids, avail_nids, alloc_nids; 3806 int total_count, utilization; 3807 int bg_gc, nr_wb_cp_data, nr_wb_data; 3808 int nr_rd_data, nr_rd_node, nr_rd_meta; 3809 int nr_dio_read, nr_dio_write; 3810 unsigned int io_skip_bggc, other_skip_bggc; 3811 int nr_flushing, nr_flushed, flush_list_empty; 3812 int nr_discarding, nr_discarded; 3813 int nr_discard_cmd; 3814 unsigned int undiscard_blks; 3815 int nr_issued_ckpt, nr_total_ckpt, nr_queued_ckpt; 3816 unsigned int cur_ckpt_time, peak_ckpt_time; 3817 int inline_xattr, inline_inode, inline_dir, append, update, orphans; 3818 int compr_inode; 3819 unsigned long long compr_blocks; 3820 int aw_cnt, max_aw_cnt, vw_cnt, max_vw_cnt; 3821 unsigned int valid_count, valid_node_count, valid_inode_count, discard_blks; 3822 unsigned int bimodal, avg_vblocks; 3823 int util_free, util_valid, util_invalid; 3824 int rsvd_segs, overp_segs; 3825 int dirty_count, node_pages, meta_pages, compress_pages; 3826 int compress_page_hit; 3827 int prefree_count, call_count, cp_count, bg_cp_count; 3828 int tot_segs, node_segs, data_segs, free_segs, free_secs; 3829 int bg_node_segs, bg_data_segs; 3830 int tot_blks, data_blks, node_blks; 3831 int bg_data_blks, bg_node_blks; 3832 unsigned long long skipped_atomic_files[2]; 3833 int curseg[NR_CURSEG_TYPE]; 3834 int cursec[NR_CURSEG_TYPE]; 3835 int curzone[NR_CURSEG_TYPE]; 3836 unsigned int dirty_seg[NR_CURSEG_TYPE]; 3837 unsigned int full_seg[NR_CURSEG_TYPE]; 3838 unsigned int valid_blks[NR_CURSEG_TYPE]; 3839 3840 unsigned int meta_count[META_MAX]; 3841 unsigned int segment_count[2]; 3842 unsigned int block_count[2]; 3843 unsigned int inplace_count; 3844 unsigned long long base_mem, cache_mem, page_mem; 3845 }; 3846 3847 static inline struct f2fs_stat_info *F2FS_STAT(struct f2fs_sb_info *sbi) 3848 { 3849 return (struct f2fs_stat_info *)sbi->stat_info; 3850 } 3851 3852 #define stat_inc_cp_count(si) ((si)->cp_count++) 3853 #define stat_inc_bg_cp_count(si) ((si)->bg_cp_count++) 3854 #define stat_inc_call_count(si) ((si)->call_count++) 3855 #define stat_inc_bggc_count(si) ((si)->bg_gc++) 3856 #define stat_io_skip_bggc_count(sbi) ((sbi)->io_skip_bggc++) 3857 #define stat_other_skip_bggc_count(sbi) ((sbi)->other_skip_bggc++) 3858 #define stat_inc_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]++) 3859 #define stat_dec_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]--) 3860 #define stat_inc_total_hit(sbi) (atomic64_inc(&(sbi)->total_hit_ext)) 3861 #define stat_inc_rbtree_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_rbtree)) 3862 #define stat_inc_largest_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_largest)) 3863 #define stat_inc_cached_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_cached)) 3864 #define stat_inc_inline_xattr(inode) \ 3865 do { \ 3866 if (f2fs_has_inline_xattr(inode)) \ 3867 (atomic_inc(&F2FS_I_SB(inode)->inline_xattr)); \ 3868 } while (0) 3869 #define stat_dec_inline_xattr(inode) \ 3870 do { \ 3871 if (f2fs_has_inline_xattr(inode)) \ 3872 (atomic_dec(&F2FS_I_SB(inode)->inline_xattr)); \ 3873 } while (0) 3874 #define stat_inc_inline_inode(inode) \ 3875 do { \ 3876 if (f2fs_has_inline_data(inode)) \ 3877 (atomic_inc(&F2FS_I_SB(inode)->inline_inode)); \ 3878 } while (0) 3879 #define stat_dec_inline_inode(inode) \ 3880 do { \ 3881 if (f2fs_has_inline_data(inode)) \ 3882 (atomic_dec(&F2FS_I_SB(inode)->inline_inode)); \ 3883 } while (0) 3884 #define stat_inc_inline_dir(inode) \ 3885 do { \ 3886 if (f2fs_has_inline_dentry(inode)) \ 3887 (atomic_inc(&F2FS_I_SB(inode)->inline_dir)); \ 3888 } while (0) 3889 #define stat_dec_inline_dir(inode) \ 3890 do { \ 3891 if (f2fs_has_inline_dentry(inode)) \ 3892 (atomic_dec(&F2FS_I_SB(inode)->inline_dir)); \ 3893 } while (0) 3894 #define stat_inc_compr_inode(inode) \ 3895 do { \ 3896 if (f2fs_compressed_file(inode)) \ 3897 (atomic_inc(&F2FS_I_SB(inode)->compr_inode)); \ 3898 } while (0) 3899 #define stat_dec_compr_inode(inode) \ 3900 do { \ 3901 if (f2fs_compressed_file(inode)) \ 3902 (atomic_dec(&F2FS_I_SB(inode)->compr_inode)); \ 3903 } while (0) 3904 #define stat_add_compr_blocks(inode, blocks) \ 3905 (atomic64_add(blocks, &F2FS_I_SB(inode)->compr_blocks)) 3906 #define stat_sub_compr_blocks(inode, blocks) \ 3907 (atomic64_sub(blocks, &F2FS_I_SB(inode)->compr_blocks)) 3908 #define stat_inc_meta_count(sbi, blkaddr) \ 3909 do { \ 3910 if (blkaddr < SIT_I(sbi)->sit_base_addr) \ 3911 atomic_inc(&(sbi)->meta_count[META_CP]); \ 3912 else if (blkaddr < NM_I(sbi)->nat_blkaddr) \ 3913 atomic_inc(&(sbi)->meta_count[META_SIT]); \ 3914 else if (blkaddr < SM_I(sbi)->ssa_blkaddr) \ 3915 atomic_inc(&(sbi)->meta_count[META_NAT]); \ 3916 else if (blkaddr < SM_I(sbi)->main_blkaddr) \ 3917 atomic_inc(&(sbi)->meta_count[META_SSA]); \ 3918 } while (0) 3919 #define stat_inc_seg_type(sbi, curseg) \ 3920 ((sbi)->segment_count[(curseg)->alloc_type]++) 3921 #define stat_inc_block_count(sbi, curseg) \ 3922 ((sbi)->block_count[(curseg)->alloc_type]++) 3923 #define stat_inc_inplace_blocks(sbi) \ 3924 (atomic_inc(&(sbi)->inplace_count)) 3925 #define stat_update_max_atomic_write(inode) \ 3926 do { \ 3927 int cur = F2FS_I_SB(inode)->atomic_files; \ 3928 int max = atomic_read(&F2FS_I_SB(inode)->max_aw_cnt); \ 3929 if (cur > max) \ 3930 atomic_set(&F2FS_I_SB(inode)->max_aw_cnt, cur); \ 3931 } while (0) 3932 #define stat_inc_volatile_write(inode) \ 3933 (atomic_inc(&F2FS_I_SB(inode)->vw_cnt)) 3934 #define stat_dec_volatile_write(inode) \ 3935 (atomic_dec(&F2FS_I_SB(inode)->vw_cnt)) 3936 #define stat_update_max_volatile_write(inode) \ 3937 do { \ 3938 int cur = atomic_read(&F2FS_I_SB(inode)->vw_cnt); \ 3939 int max = atomic_read(&F2FS_I_SB(inode)->max_vw_cnt); \ 3940 if (cur > max) \ 3941 atomic_set(&F2FS_I_SB(inode)->max_vw_cnt, cur); \ 3942 } while (0) 3943 #define stat_inc_seg_count(sbi, type, gc_type) \ 3944 do { \ 3945 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 3946 si->tot_segs++; \ 3947 if ((type) == SUM_TYPE_DATA) { \ 3948 si->data_segs++; \ 3949 si->bg_data_segs += (gc_type == BG_GC) ? 1 : 0; \ 3950 } else { \ 3951 si->node_segs++; \ 3952 si->bg_node_segs += (gc_type == BG_GC) ? 1 : 0; \ 3953 } \ 3954 } while (0) 3955 3956 #define stat_inc_tot_blk_count(si, blks) \ 3957 ((si)->tot_blks += (blks)) 3958 3959 #define stat_inc_data_blk_count(sbi, blks, gc_type) \ 3960 do { \ 3961 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 3962 stat_inc_tot_blk_count(si, blks); \ 3963 si->data_blks += (blks); \ 3964 si->bg_data_blks += ((gc_type) == BG_GC) ? (blks) : 0; \ 3965 } while (0) 3966 3967 #define stat_inc_node_blk_count(sbi, blks, gc_type) \ 3968 do { \ 3969 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 3970 stat_inc_tot_blk_count(si, blks); \ 3971 si->node_blks += (blks); \ 3972 si->bg_node_blks += ((gc_type) == BG_GC) ? (blks) : 0; \ 3973 } while (0) 3974 3975 int f2fs_build_stats(struct f2fs_sb_info *sbi); 3976 void f2fs_destroy_stats(struct f2fs_sb_info *sbi); 3977 void __init f2fs_create_root_stats(void); 3978 void f2fs_destroy_root_stats(void); 3979 void f2fs_update_sit_info(struct f2fs_sb_info *sbi); 3980 #else 3981 #define stat_inc_cp_count(si) do { } while (0) 3982 #define stat_inc_bg_cp_count(si) do { } while (0) 3983 #define stat_inc_call_count(si) do { } while (0) 3984 #define stat_inc_bggc_count(si) do { } while (0) 3985 #define stat_io_skip_bggc_count(sbi) do { } while (0) 3986 #define stat_other_skip_bggc_count(sbi) do { } while (0) 3987 #define stat_inc_dirty_inode(sbi, type) do { } while (0) 3988 #define stat_dec_dirty_inode(sbi, type) do { } while (0) 3989 #define stat_inc_total_hit(sbi) do { } while (0) 3990 #define stat_inc_rbtree_node_hit(sbi) do { } while (0) 3991 #define stat_inc_largest_node_hit(sbi) do { } while (0) 3992 #define stat_inc_cached_node_hit(sbi) do { } while (0) 3993 #define stat_inc_inline_xattr(inode) do { } while (0) 3994 #define stat_dec_inline_xattr(inode) do { } while (0) 3995 #define stat_inc_inline_inode(inode) do { } while (0) 3996 #define stat_dec_inline_inode(inode) do { } while (0) 3997 #define stat_inc_inline_dir(inode) do { } while (0) 3998 #define stat_dec_inline_dir(inode) do { } while (0) 3999 #define stat_inc_compr_inode(inode) do { } while (0) 4000 #define stat_dec_compr_inode(inode) do { } while (0) 4001 #define stat_add_compr_blocks(inode, blocks) do { } while (0) 4002 #define stat_sub_compr_blocks(inode, blocks) do { } while (0) 4003 #define stat_update_max_atomic_write(inode) do { } while (0) 4004 #define stat_inc_volatile_write(inode) do { } while (0) 4005 #define stat_dec_volatile_write(inode) do { } while (0) 4006 #define stat_update_max_volatile_write(inode) do { } while (0) 4007 #define stat_inc_meta_count(sbi, blkaddr) do { } while (0) 4008 #define stat_inc_seg_type(sbi, curseg) do { } while (0) 4009 #define stat_inc_block_count(sbi, curseg) do { } while (0) 4010 #define stat_inc_inplace_blocks(sbi) do { } while (0) 4011 #define stat_inc_seg_count(sbi, type, gc_type) do { } while (0) 4012 #define stat_inc_tot_blk_count(si, blks) do { } while (0) 4013 #define stat_inc_data_blk_count(sbi, blks, gc_type) do { } while (0) 4014 #define stat_inc_node_blk_count(sbi, blks, gc_type) do { } while (0) 4015 4016 static inline int f2fs_build_stats(struct f2fs_sb_info *sbi) { return 0; } 4017 static inline void f2fs_destroy_stats(struct f2fs_sb_info *sbi) { } 4018 static inline void __init f2fs_create_root_stats(void) { } 4019 static inline void f2fs_destroy_root_stats(void) { } 4020 static inline void f2fs_update_sit_info(struct f2fs_sb_info *sbi) {} 4021 #endif 4022 4023 extern const struct file_operations f2fs_dir_operations; 4024 extern const struct file_operations f2fs_file_operations; 4025 extern const struct inode_operations f2fs_file_inode_operations; 4026 extern const struct address_space_operations f2fs_dblock_aops; 4027 extern const struct address_space_operations f2fs_node_aops; 4028 extern const struct address_space_operations f2fs_meta_aops; 4029 extern const struct inode_operations f2fs_dir_inode_operations; 4030 extern const struct inode_operations f2fs_symlink_inode_operations; 4031 extern const struct inode_operations f2fs_encrypted_symlink_inode_operations; 4032 extern const struct inode_operations f2fs_special_inode_operations; 4033 extern struct kmem_cache *f2fs_inode_entry_slab; 4034 4035 /* 4036 * inline.c 4037 */ 4038 bool f2fs_may_inline_data(struct inode *inode); 4039 bool f2fs_may_inline_dentry(struct inode *inode); 4040 void f2fs_do_read_inline_data(struct page *page, struct page *ipage); 4041 void f2fs_truncate_inline_inode(struct inode *inode, 4042 struct page *ipage, u64 from); 4043 int f2fs_read_inline_data(struct inode *inode, struct page *page); 4044 int f2fs_convert_inline_page(struct dnode_of_data *dn, struct page *page); 4045 int f2fs_convert_inline_inode(struct inode *inode); 4046 int f2fs_try_convert_inline_dir(struct inode *dir, struct dentry *dentry); 4047 int f2fs_write_inline_data(struct inode *inode, struct page *page); 4048 int f2fs_recover_inline_data(struct inode *inode, struct page *npage); 4049 struct f2fs_dir_entry *f2fs_find_in_inline_dir(struct inode *dir, 4050 const struct f2fs_filename *fname, 4051 struct page **res_page); 4052 int f2fs_make_empty_inline_dir(struct inode *inode, struct inode *parent, 4053 struct page *ipage); 4054 int f2fs_add_inline_entry(struct inode *dir, const struct f2fs_filename *fname, 4055 struct inode *inode, nid_t ino, umode_t mode); 4056 void f2fs_delete_inline_entry(struct f2fs_dir_entry *dentry, 4057 struct page *page, struct inode *dir, 4058 struct inode *inode); 4059 bool f2fs_empty_inline_dir(struct inode *dir); 4060 int f2fs_read_inline_dir(struct file *file, struct dir_context *ctx, 4061 struct fscrypt_str *fstr); 4062 int f2fs_inline_data_fiemap(struct inode *inode, 4063 struct fiemap_extent_info *fieinfo, 4064 __u64 start, __u64 len); 4065 4066 /* 4067 * shrinker.c 4068 */ 4069 unsigned long f2fs_shrink_count(struct shrinker *shrink, 4070 struct shrink_control *sc); 4071 unsigned long f2fs_shrink_scan(struct shrinker *shrink, 4072 struct shrink_control *sc); 4073 void f2fs_join_shrinker(struct f2fs_sb_info *sbi); 4074 void f2fs_leave_shrinker(struct f2fs_sb_info *sbi); 4075 4076 /* 4077 * extent_cache.c 4078 */ 4079 struct rb_entry *f2fs_lookup_rb_tree(struct rb_root_cached *root, 4080 struct rb_entry *cached_re, unsigned int ofs); 4081 struct rb_node **f2fs_lookup_rb_tree_ext(struct f2fs_sb_info *sbi, 4082 struct rb_root_cached *root, 4083 struct rb_node **parent, 4084 unsigned long long key, bool *left_most); 4085 struct rb_node **f2fs_lookup_rb_tree_for_insert(struct f2fs_sb_info *sbi, 4086 struct rb_root_cached *root, 4087 struct rb_node **parent, 4088 unsigned int ofs, bool *leftmost); 4089 struct rb_entry *f2fs_lookup_rb_tree_ret(struct rb_root_cached *root, 4090 struct rb_entry *cached_re, unsigned int ofs, 4091 struct rb_entry **prev_entry, struct rb_entry **next_entry, 4092 struct rb_node ***insert_p, struct rb_node **insert_parent, 4093 bool force, bool *leftmost); 4094 bool f2fs_check_rb_tree_consistence(struct f2fs_sb_info *sbi, 4095 struct rb_root_cached *root, bool check_key); 4096 unsigned int f2fs_shrink_extent_tree(struct f2fs_sb_info *sbi, int nr_shrink); 4097 void f2fs_init_extent_tree(struct inode *inode, struct page *ipage); 4098 void f2fs_drop_extent_tree(struct inode *inode); 4099 unsigned int f2fs_destroy_extent_node(struct inode *inode); 4100 void f2fs_destroy_extent_tree(struct inode *inode); 4101 bool f2fs_lookup_extent_cache(struct inode *inode, pgoff_t pgofs, 4102 struct extent_info *ei); 4103 void f2fs_update_extent_cache(struct dnode_of_data *dn); 4104 void f2fs_update_extent_cache_range(struct dnode_of_data *dn, 4105 pgoff_t fofs, block_t blkaddr, unsigned int len); 4106 void f2fs_init_extent_cache_info(struct f2fs_sb_info *sbi); 4107 int __init f2fs_create_extent_cache(void); 4108 void f2fs_destroy_extent_cache(void); 4109 4110 /* 4111 * sysfs.c 4112 */ 4113 #define MIN_RA_MUL 2 4114 #define MAX_RA_MUL 256 4115 4116 int __init f2fs_init_sysfs(void); 4117 void f2fs_exit_sysfs(void); 4118 int f2fs_register_sysfs(struct f2fs_sb_info *sbi); 4119 void f2fs_unregister_sysfs(struct f2fs_sb_info *sbi); 4120 4121 /* verity.c */ 4122 extern const struct fsverity_operations f2fs_verityops; 4123 4124 /* 4125 * crypto support 4126 */ 4127 static inline bool f2fs_encrypted_file(struct inode *inode) 4128 { 4129 return IS_ENCRYPTED(inode) && S_ISREG(inode->i_mode); 4130 } 4131 4132 static inline void f2fs_set_encrypted_inode(struct inode *inode) 4133 { 4134 #ifdef CONFIG_FS_ENCRYPTION 4135 file_set_encrypt(inode); 4136 f2fs_set_inode_flags(inode); 4137 #endif 4138 } 4139 4140 /* 4141 * Returns true if the reads of the inode's data need to undergo some 4142 * postprocessing step, like decryption or authenticity verification. 4143 */ 4144 static inline bool f2fs_post_read_required(struct inode *inode) 4145 { 4146 return f2fs_encrypted_file(inode) || fsverity_active(inode) || 4147 f2fs_compressed_file(inode); 4148 } 4149 4150 /* 4151 * compress.c 4152 */ 4153 #ifdef CONFIG_F2FS_FS_COMPRESSION 4154 bool f2fs_is_compressed_page(struct page *page); 4155 struct page *f2fs_compress_control_page(struct page *page); 4156 int f2fs_prepare_compress_overwrite(struct inode *inode, 4157 struct page **pagep, pgoff_t index, void **fsdata); 4158 bool f2fs_compress_write_end(struct inode *inode, void *fsdata, 4159 pgoff_t index, unsigned copied); 4160 int f2fs_truncate_partial_cluster(struct inode *inode, u64 from, bool lock); 4161 void f2fs_compress_write_end_io(struct bio *bio, struct page *page); 4162 bool f2fs_is_compress_backend_ready(struct inode *inode); 4163 int f2fs_init_compress_mempool(void); 4164 void f2fs_destroy_compress_mempool(void); 4165 void f2fs_decompress_cluster(struct decompress_io_ctx *dic); 4166 void f2fs_end_read_compressed_page(struct page *page, bool failed, 4167 block_t blkaddr); 4168 bool f2fs_cluster_is_empty(struct compress_ctx *cc); 4169 bool f2fs_cluster_can_merge_page(struct compress_ctx *cc, pgoff_t index); 4170 bool f2fs_all_cluster_page_loaded(struct compress_ctx *cc, struct pagevec *pvec, 4171 int index, int nr_pages); 4172 bool f2fs_sanity_check_cluster(struct dnode_of_data *dn); 4173 void f2fs_compress_ctx_add_page(struct compress_ctx *cc, struct page *page); 4174 int f2fs_write_multi_pages(struct compress_ctx *cc, 4175 int *submitted, 4176 struct writeback_control *wbc, 4177 enum iostat_type io_type); 4178 int f2fs_is_compressed_cluster(struct inode *inode, pgoff_t index); 4179 void f2fs_update_extent_tree_range_compressed(struct inode *inode, 4180 pgoff_t fofs, block_t blkaddr, unsigned int llen, 4181 unsigned int c_len); 4182 int f2fs_read_multi_pages(struct compress_ctx *cc, struct bio **bio_ret, 4183 unsigned nr_pages, sector_t *last_block_in_bio, 4184 bool is_readahead, bool for_write); 4185 struct decompress_io_ctx *f2fs_alloc_dic(struct compress_ctx *cc); 4186 void f2fs_decompress_end_io(struct decompress_io_ctx *dic, bool failed); 4187 void f2fs_put_page_dic(struct page *page); 4188 unsigned int f2fs_cluster_blocks_are_contiguous(struct dnode_of_data *dn); 4189 int f2fs_init_compress_ctx(struct compress_ctx *cc); 4190 void f2fs_destroy_compress_ctx(struct compress_ctx *cc, bool reuse); 4191 void f2fs_init_compress_info(struct f2fs_sb_info *sbi); 4192 int f2fs_init_compress_inode(struct f2fs_sb_info *sbi); 4193 void f2fs_destroy_compress_inode(struct f2fs_sb_info *sbi); 4194 int f2fs_init_page_array_cache(struct f2fs_sb_info *sbi); 4195 void f2fs_destroy_page_array_cache(struct f2fs_sb_info *sbi); 4196 int __init f2fs_init_compress_cache(void); 4197 void f2fs_destroy_compress_cache(void); 4198 struct address_space *COMPRESS_MAPPING(struct f2fs_sb_info *sbi); 4199 void f2fs_invalidate_compress_page(struct f2fs_sb_info *sbi, block_t blkaddr); 4200 void f2fs_cache_compressed_page(struct f2fs_sb_info *sbi, struct page *page, 4201 nid_t ino, block_t blkaddr); 4202 bool f2fs_load_compressed_page(struct f2fs_sb_info *sbi, struct page *page, 4203 block_t blkaddr); 4204 void f2fs_invalidate_compress_pages(struct f2fs_sb_info *sbi, nid_t ino); 4205 #define inc_compr_inode_stat(inode) \ 4206 do { \ 4207 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); \ 4208 sbi->compr_new_inode++; \ 4209 } while (0) 4210 #define add_compr_block_stat(inode, blocks) \ 4211 do { \ 4212 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); \ 4213 int diff = F2FS_I(inode)->i_cluster_size - blocks; \ 4214 sbi->compr_written_block += blocks; \ 4215 sbi->compr_saved_block += diff; \ 4216 } while (0) 4217 #else 4218 static inline bool f2fs_is_compressed_page(struct page *page) { return false; } 4219 static inline bool f2fs_is_compress_backend_ready(struct inode *inode) 4220 { 4221 if (!f2fs_compressed_file(inode)) 4222 return true; 4223 /* not support compression */ 4224 return false; 4225 } 4226 static inline struct page *f2fs_compress_control_page(struct page *page) 4227 { 4228 WARN_ON_ONCE(1); 4229 return ERR_PTR(-EINVAL); 4230 } 4231 static inline int f2fs_init_compress_mempool(void) { return 0; } 4232 static inline void f2fs_destroy_compress_mempool(void) { } 4233 static inline void f2fs_decompress_cluster(struct decompress_io_ctx *dic) { } 4234 static inline void f2fs_end_read_compressed_page(struct page *page, 4235 bool failed, block_t blkaddr) 4236 { 4237 WARN_ON_ONCE(1); 4238 } 4239 static inline void f2fs_put_page_dic(struct page *page) 4240 { 4241 WARN_ON_ONCE(1); 4242 } 4243 static inline unsigned int f2fs_cluster_blocks_are_contiguous(struct dnode_of_data *dn) { return 0; } 4244 static inline bool f2fs_sanity_check_cluster(struct dnode_of_data *dn) { return false; } 4245 static inline int f2fs_init_compress_inode(struct f2fs_sb_info *sbi) { return 0; } 4246 static inline void f2fs_destroy_compress_inode(struct f2fs_sb_info *sbi) { } 4247 static inline int f2fs_init_page_array_cache(struct f2fs_sb_info *sbi) { return 0; } 4248 static inline void f2fs_destroy_page_array_cache(struct f2fs_sb_info *sbi) { } 4249 static inline int __init f2fs_init_compress_cache(void) { return 0; } 4250 static inline void f2fs_destroy_compress_cache(void) { } 4251 static inline void f2fs_invalidate_compress_page(struct f2fs_sb_info *sbi, 4252 block_t blkaddr) { } 4253 static inline void f2fs_cache_compressed_page(struct f2fs_sb_info *sbi, 4254 struct page *page, nid_t ino, block_t blkaddr) { } 4255 static inline bool f2fs_load_compressed_page(struct f2fs_sb_info *sbi, 4256 struct page *page, block_t blkaddr) { return false; } 4257 static inline void f2fs_invalidate_compress_pages(struct f2fs_sb_info *sbi, 4258 nid_t ino) { } 4259 #define inc_compr_inode_stat(inode) do { } while (0) 4260 static inline void f2fs_update_extent_tree_range_compressed(struct inode *inode, 4261 pgoff_t fofs, block_t blkaddr, unsigned int llen, 4262 unsigned int c_len) { } 4263 #endif 4264 4265 static inline void set_compress_context(struct inode *inode) 4266 { 4267 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 4268 4269 F2FS_I(inode)->i_compress_algorithm = 4270 F2FS_OPTION(sbi).compress_algorithm; 4271 F2FS_I(inode)->i_log_cluster_size = 4272 F2FS_OPTION(sbi).compress_log_size; 4273 F2FS_I(inode)->i_compress_flag = 4274 F2FS_OPTION(sbi).compress_chksum ? 4275 1 << COMPRESS_CHKSUM : 0; 4276 F2FS_I(inode)->i_cluster_size = 4277 1 << F2FS_I(inode)->i_log_cluster_size; 4278 if ((F2FS_I(inode)->i_compress_algorithm == COMPRESS_LZ4 || 4279 F2FS_I(inode)->i_compress_algorithm == COMPRESS_ZSTD) && 4280 F2FS_OPTION(sbi).compress_level) 4281 F2FS_I(inode)->i_compress_flag |= 4282 F2FS_OPTION(sbi).compress_level << 4283 COMPRESS_LEVEL_OFFSET; 4284 F2FS_I(inode)->i_flags |= F2FS_COMPR_FL; 4285 set_inode_flag(inode, FI_COMPRESSED_FILE); 4286 stat_inc_compr_inode(inode); 4287 inc_compr_inode_stat(inode); 4288 f2fs_mark_inode_dirty_sync(inode, true); 4289 } 4290 4291 static inline bool f2fs_disable_compressed_file(struct inode *inode) 4292 { 4293 struct f2fs_inode_info *fi = F2FS_I(inode); 4294 4295 if (!f2fs_compressed_file(inode)) 4296 return true; 4297 if (S_ISREG(inode->i_mode) && F2FS_HAS_BLOCKS(inode)) 4298 return false; 4299 4300 fi->i_flags &= ~F2FS_COMPR_FL; 4301 stat_dec_compr_inode(inode); 4302 clear_inode_flag(inode, FI_COMPRESSED_FILE); 4303 f2fs_mark_inode_dirty_sync(inode, true); 4304 return true; 4305 } 4306 4307 #define F2FS_FEATURE_FUNCS(name, flagname) \ 4308 static inline int f2fs_sb_has_##name(struct f2fs_sb_info *sbi) \ 4309 { \ 4310 return F2FS_HAS_FEATURE(sbi, F2FS_FEATURE_##flagname); \ 4311 } 4312 4313 F2FS_FEATURE_FUNCS(encrypt, ENCRYPT); 4314 F2FS_FEATURE_FUNCS(blkzoned, BLKZONED); 4315 F2FS_FEATURE_FUNCS(extra_attr, EXTRA_ATTR); 4316 F2FS_FEATURE_FUNCS(project_quota, PRJQUOTA); 4317 F2FS_FEATURE_FUNCS(inode_chksum, INODE_CHKSUM); 4318 F2FS_FEATURE_FUNCS(flexible_inline_xattr, FLEXIBLE_INLINE_XATTR); 4319 F2FS_FEATURE_FUNCS(quota_ino, QUOTA_INO); 4320 F2FS_FEATURE_FUNCS(inode_crtime, INODE_CRTIME); 4321 F2FS_FEATURE_FUNCS(lost_found, LOST_FOUND); 4322 F2FS_FEATURE_FUNCS(verity, VERITY); 4323 F2FS_FEATURE_FUNCS(sb_chksum, SB_CHKSUM); 4324 F2FS_FEATURE_FUNCS(casefold, CASEFOLD); 4325 F2FS_FEATURE_FUNCS(compression, COMPRESSION); 4326 F2FS_FEATURE_FUNCS(readonly, RO); 4327 4328 static inline bool f2fs_may_extent_tree(struct inode *inode) 4329 { 4330 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 4331 4332 if (!test_opt(sbi, EXTENT_CACHE) || 4333 is_inode_flag_set(inode, FI_NO_EXTENT) || 4334 (is_inode_flag_set(inode, FI_COMPRESSED_FILE) && 4335 !f2fs_sb_has_readonly(sbi))) 4336 return false; 4337 4338 /* 4339 * for recovered files during mount do not create extents 4340 * if shrinker is not registered. 4341 */ 4342 if (list_empty(&sbi->s_list)) 4343 return false; 4344 4345 return S_ISREG(inode->i_mode); 4346 } 4347 4348 #ifdef CONFIG_BLK_DEV_ZONED 4349 static inline bool f2fs_blkz_is_seq(struct f2fs_sb_info *sbi, int devi, 4350 block_t blkaddr) 4351 { 4352 unsigned int zno = blkaddr >> sbi->log_blocks_per_blkz; 4353 4354 return test_bit(zno, FDEV(devi).blkz_seq); 4355 } 4356 #endif 4357 4358 static inline bool f2fs_hw_should_discard(struct f2fs_sb_info *sbi) 4359 { 4360 return f2fs_sb_has_blkzoned(sbi); 4361 } 4362 4363 static inline bool f2fs_bdev_support_discard(struct block_device *bdev) 4364 { 4365 return blk_queue_discard(bdev_get_queue(bdev)) || 4366 bdev_is_zoned(bdev); 4367 } 4368 4369 static inline bool f2fs_hw_support_discard(struct f2fs_sb_info *sbi) 4370 { 4371 int i; 4372 4373 if (!f2fs_is_multi_device(sbi)) 4374 return f2fs_bdev_support_discard(sbi->sb->s_bdev); 4375 4376 for (i = 0; i < sbi->s_ndevs; i++) 4377 if (f2fs_bdev_support_discard(FDEV(i).bdev)) 4378 return true; 4379 return false; 4380 } 4381 4382 static inline bool f2fs_realtime_discard_enable(struct f2fs_sb_info *sbi) 4383 { 4384 return (test_opt(sbi, DISCARD) && f2fs_hw_support_discard(sbi)) || 4385 f2fs_hw_should_discard(sbi); 4386 } 4387 4388 static inline bool f2fs_hw_is_readonly(struct f2fs_sb_info *sbi) 4389 { 4390 int i; 4391 4392 if (!f2fs_is_multi_device(sbi)) 4393 return bdev_read_only(sbi->sb->s_bdev); 4394 4395 for (i = 0; i < sbi->s_ndevs; i++) 4396 if (bdev_read_only(FDEV(i).bdev)) 4397 return true; 4398 return false; 4399 } 4400 4401 static inline bool f2fs_lfs_mode(struct f2fs_sb_info *sbi) 4402 { 4403 return F2FS_OPTION(sbi).fs_mode == FS_MODE_LFS; 4404 } 4405 4406 static inline bool f2fs_may_compress(struct inode *inode) 4407 { 4408 if (IS_SWAPFILE(inode) || f2fs_is_pinned_file(inode) || 4409 f2fs_is_atomic_file(inode) || 4410 f2fs_is_volatile_file(inode)) 4411 return false; 4412 return S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode); 4413 } 4414 4415 static inline void f2fs_i_compr_blocks_update(struct inode *inode, 4416 u64 blocks, bool add) 4417 { 4418 int diff = F2FS_I(inode)->i_cluster_size - blocks; 4419 struct f2fs_inode_info *fi = F2FS_I(inode); 4420 4421 /* don't update i_compr_blocks if saved blocks were released */ 4422 if (!add && !atomic_read(&fi->i_compr_blocks)) 4423 return; 4424 4425 if (add) { 4426 atomic_add(diff, &fi->i_compr_blocks); 4427 stat_add_compr_blocks(inode, diff); 4428 } else { 4429 atomic_sub(diff, &fi->i_compr_blocks); 4430 stat_sub_compr_blocks(inode, diff); 4431 } 4432 f2fs_mark_inode_dirty_sync(inode, true); 4433 } 4434 4435 static inline int block_unaligned_IO(struct inode *inode, 4436 struct kiocb *iocb, struct iov_iter *iter) 4437 { 4438 unsigned int i_blkbits = READ_ONCE(inode->i_blkbits); 4439 unsigned int blocksize_mask = (1 << i_blkbits) - 1; 4440 loff_t offset = iocb->ki_pos; 4441 unsigned long align = offset | iov_iter_alignment(iter); 4442 4443 return align & blocksize_mask; 4444 } 4445 4446 static inline bool f2fs_allow_multi_device_dio(struct f2fs_sb_info *sbi, 4447 int flag) 4448 { 4449 if (!f2fs_is_multi_device(sbi)) 4450 return false; 4451 if (flag != F2FS_GET_BLOCK_DIO) 4452 return false; 4453 return sbi->aligned_blksize; 4454 } 4455 4456 static inline bool f2fs_force_buffered_io(struct inode *inode, 4457 struct kiocb *iocb, struct iov_iter *iter) 4458 { 4459 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 4460 int rw = iov_iter_rw(iter); 4461 4462 if (f2fs_post_read_required(inode)) 4463 return true; 4464 4465 /* disallow direct IO if any of devices has unaligned blksize */ 4466 if (f2fs_is_multi_device(sbi) && !sbi->aligned_blksize) 4467 return true; 4468 /* 4469 * for blkzoned device, fallback direct IO to buffered IO, so 4470 * all IOs can be serialized by log-structured write. 4471 */ 4472 if (f2fs_sb_has_blkzoned(sbi)) 4473 return true; 4474 if (f2fs_lfs_mode(sbi) && (rw == WRITE)) { 4475 if (block_unaligned_IO(inode, iocb, iter)) 4476 return true; 4477 if (F2FS_IO_ALIGNED(sbi)) 4478 return true; 4479 } 4480 if (is_sbi_flag_set(F2FS_I_SB(inode), SBI_CP_DISABLED)) 4481 return true; 4482 4483 return false; 4484 } 4485 4486 static inline bool f2fs_need_verity(const struct inode *inode, pgoff_t idx) 4487 { 4488 return fsverity_active(inode) && 4489 idx < DIV_ROUND_UP(inode->i_size, PAGE_SIZE); 4490 } 4491 4492 #ifdef CONFIG_F2FS_FAULT_INJECTION 4493 extern void f2fs_build_fault_attr(struct f2fs_sb_info *sbi, unsigned int rate, 4494 unsigned int type); 4495 #else 4496 #define f2fs_build_fault_attr(sbi, rate, type) do { } while (0) 4497 #endif 4498 4499 static inline bool is_journalled_quota(struct f2fs_sb_info *sbi) 4500 { 4501 #ifdef CONFIG_QUOTA 4502 if (f2fs_sb_has_quota_ino(sbi)) 4503 return true; 4504 if (F2FS_OPTION(sbi).s_qf_names[USRQUOTA] || 4505 F2FS_OPTION(sbi).s_qf_names[GRPQUOTA] || 4506 F2FS_OPTION(sbi).s_qf_names[PRJQUOTA]) 4507 return true; 4508 #endif 4509 return false; 4510 } 4511 4512 static inline bool f2fs_block_unit_discard(struct f2fs_sb_info *sbi) 4513 { 4514 return F2FS_OPTION(sbi).discard_unit == DISCARD_UNIT_BLOCK; 4515 } 4516 4517 #define EFSBADCRC EBADMSG /* Bad CRC detected */ 4518 #define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */ 4519 4520 #endif /* _LINUX_F2FS_H */ 4521