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_DO_DEFRAG, /* indicate defragment is running */ 744 FI_DIRTY_FILE, /* indicate regular/symlink has dirty pages */ 745 FI_PREALLOCATED_ALL, /* all blocks for write were preallocated */ 746 FI_HOT_DATA, /* indicate file is hot */ 747 FI_EXTRA_ATTR, /* indicate file has extra attribute */ 748 FI_PROJ_INHERIT, /* indicate file inherits projectid */ 749 FI_PIN_FILE, /* indicate file should not be gced */ 750 FI_ATOMIC_REVOKE_REQUEST, /* request to drop atomic data */ 751 FI_VERITY_IN_PROGRESS, /* building fs-verity Merkle tree */ 752 FI_COMPRESSED_FILE, /* indicate file's data can be compressed */ 753 FI_COMPRESS_CORRUPT, /* indicate compressed cluster is corrupted */ 754 FI_MMAP_FILE, /* indicate file was mmapped */ 755 FI_ENABLE_COMPRESS, /* enable compression in "user" compression mode */ 756 FI_COMPRESS_RELEASED, /* compressed blocks were released */ 757 FI_ALIGNED_WRITE, /* enable aligned write */ 758 FI_MAX, /* max flag, never be used */ 759 }; 760 761 struct f2fs_inode_info { 762 struct inode vfs_inode; /* serve a vfs inode */ 763 unsigned long i_flags; /* keep an inode flags for ioctl */ 764 unsigned char i_advise; /* use to give file attribute hints */ 765 unsigned char i_dir_level; /* use for dentry level for large dir */ 766 unsigned int i_current_depth; /* only for directory depth */ 767 /* for gc failure statistic */ 768 unsigned int i_gc_failures[MAX_GC_FAILURE]; 769 unsigned int i_pino; /* parent inode number */ 770 umode_t i_acl_mode; /* keep file acl mode temporarily */ 771 772 /* Use below internally in f2fs*/ 773 unsigned long flags[BITS_TO_LONGS(FI_MAX)]; /* use to pass per-file flags */ 774 struct f2fs_rwsem i_sem; /* protect fi info */ 775 atomic_t dirty_pages; /* # of dirty pages */ 776 f2fs_hash_t chash; /* hash value of given file name */ 777 unsigned int clevel; /* maximum level of given file name */ 778 struct task_struct *task; /* lookup and create consistency */ 779 struct task_struct *cp_task; /* separate cp/wb IO stats*/ 780 nid_t i_xattr_nid; /* node id that contains xattrs */ 781 loff_t last_disk_size; /* lastly written file size */ 782 spinlock_t i_size_lock; /* protect last_disk_size */ 783 784 #ifdef CONFIG_QUOTA 785 struct dquot *i_dquot[MAXQUOTAS]; 786 787 /* quota space reservation, managed internally by quota code */ 788 qsize_t i_reserved_quota; 789 #endif 790 struct list_head dirty_list; /* dirty list for dirs and files */ 791 struct list_head gdirty_list; /* linked in global dirty list */ 792 struct list_head inmem_ilist; /* list for inmem inodes */ 793 struct list_head inmem_pages; /* inmemory pages managed by f2fs */ 794 struct task_struct *inmem_task; /* store inmemory task */ 795 struct mutex inmem_lock; /* lock for inmemory pages */ 796 struct extent_tree *extent_tree; /* cached extent_tree entry */ 797 798 /* avoid racing between foreground op and gc */ 799 struct f2fs_rwsem i_gc_rwsem[2]; 800 struct f2fs_rwsem i_xattr_sem; /* avoid racing between reading and changing EAs */ 801 802 int i_extra_isize; /* size of extra space located in i_addr */ 803 kprojid_t i_projid; /* id for project quota */ 804 int i_inline_xattr_size; /* inline xattr size */ 805 struct timespec64 i_crtime; /* inode creation time */ 806 struct timespec64 i_disk_time[4];/* inode disk times */ 807 808 /* for file compress */ 809 atomic_t i_compr_blocks; /* # of compressed blocks */ 810 unsigned char i_compress_algorithm; /* algorithm type */ 811 unsigned char i_log_cluster_size; /* log of cluster size */ 812 unsigned char i_compress_level; /* compress level (lz4hc,zstd) */ 813 unsigned short i_compress_flag; /* compress flag */ 814 unsigned int i_cluster_size; /* cluster size */ 815 }; 816 817 static inline void get_extent_info(struct extent_info *ext, 818 struct f2fs_extent *i_ext) 819 { 820 ext->fofs = le32_to_cpu(i_ext->fofs); 821 ext->blk = le32_to_cpu(i_ext->blk); 822 ext->len = le32_to_cpu(i_ext->len); 823 } 824 825 static inline void set_raw_extent(struct extent_info *ext, 826 struct f2fs_extent *i_ext) 827 { 828 i_ext->fofs = cpu_to_le32(ext->fofs); 829 i_ext->blk = cpu_to_le32(ext->blk); 830 i_ext->len = cpu_to_le32(ext->len); 831 } 832 833 static inline void set_extent_info(struct extent_info *ei, unsigned int fofs, 834 u32 blk, unsigned int len) 835 { 836 ei->fofs = fofs; 837 ei->blk = blk; 838 ei->len = len; 839 #ifdef CONFIG_F2FS_FS_COMPRESSION 840 ei->c_len = 0; 841 #endif 842 } 843 844 static inline bool __is_discard_mergeable(struct discard_info *back, 845 struct discard_info *front, unsigned int max_len) 846 { 847 return (back->lstart + back->len == front->lstart) && 848 (back->len + front->len <= max_len); 849 } 850 851 static inline bool __is_discard_back_mergeable(struct discard_info *cur, 852 struct discard_info *back, unsigned int max_len) 853 { 854 return __is_discard_mergeable(back, cur, max_len); 855 } 856 857 static inline bool __is_discard_front_mergeable(struct discard_info *cur, 858 struct discard_info *front, unsigned int max_len) 859 { 860 return __is_discard_mergeable(cur, front, max_len); 861 } 862 863 static inline bool __is_extent_mergeable(struct extent_info *back, 864 struct extent_info *front) 865 { 866 #ifdef CONFIG_F2FS_FS_COMPRESSION 867 if (back->c_len && back->len != back->c_len) 868 return false; 869 if (front->c_len && front->len != front->c_len) 870 return false; 871 #endif 872 return (back->fofs + back->len == front->fofs && 873 back->blk + back->len == front->blk); 874 } 875 876 static inline bool __is_back_mergeable(struct extent_info *cur, 877 struct extent_info *back) 878 { 879 return __is_extent_mergeable(back, cur); 880 } 881 882 static inline bool __is_front_mergeable(struct extent_info *cur, 883 struct extent_info *front) 884 { 885 return __is_extent_mergeable(cur, front); 886 } 887 888 extern void f2fs_mark_inode_dirty_sync(struct inode *inode, bool sync); 889 static inline void __try_update_largest_extent(struct extent_tree *et, 890 struct extent_node *en) 891 { 892 if (en->ei.len > et->largest.len) { 893 et->largest = en->ei; 894 et->largest_updated = true; 895 } 896 } 897 898 /* 899 * For free nid management 900 */ 901 enum nid_state { 902 FREE_NID, /* newly added to free nid list */ 903 PREALLOC_NID, /* it is preallocated */ 904 MAX_NID_STATE, 905 }; 906 907 enum nat_state { 908 TOTAL_NAT, 909 DIRTY_NAT, 910 RECLAIMABLE_NAT, 911 MAX_NAT_STATE, 912 }; 913 914 struct f2fs_nm_info { 915 block_t nat_blkaddr; /* base disk address of NAT */ 916 nid_t max_nid; /* maximum possible node ids */ 917 nid_t available_nids; /* # of available node ids */ 918 nid_t next_scan_nid; /* the next nid to be scanned */ 919 unsigned int ram_thresh; /* control the memory footprint */ 920 unsigned int ra_nid_pages; /* # of nid pages to be readaheaded */ 921 unsigned int dirty_nats_ratio; /* control dirty nats ratio threshold */ 922 923 /* NAT cache management */ 924 struct radix_tree_root nat_root;/* root of the nat entry cache */ 925 struct radix_tree_root nat_set_root;/* root of the nat set cache */ 926 struct f2fs_rwsem nat_tree_lock; /* protect nat entry tree */ 927 struct list_head nat_entries; /* cached nat entry list (clean) */ 928 spinlock_t nat_list_lock; /* protect clean nat entry list */ 929 unsigned int nat_cnt[MAX_NAT_STATE]; /* the # of cached nat entries */ 930 unsigned int nat_blocks; /* # of nat blocks */ 931 932 /* free node ids management */ 933 struct radix_tree_root free_nid_root;/* root of the free_nid cache */ 934 struct list_head free_nid_list; /* list for free nids excluding preallocated nids */ 935 unsigned int nid_cnt[MAX_NID_STATE]; /* the number of free node id */ 936 spinlock_t nid_list_lock; /* protect nid lists ops */ 937 struct mutex build_lock; /* lock for build free nids */ 938 unsigned char **free_nid_bitmap; 939 unsigned char *nat_block_bitmap; 940 unsigned short *free_nid_count; /* free nid count of NAT block */ 941 942 /* for checkpoint */ 943 char *nat_bitmap; /* NAT bitmap pointer */ 944 945 unsigned int nat_bits_blocks; /* # of nat bits blocks */ 946 unsigned char *nat_bits; /* NAT bits blocks */ 947 unsigned char *full_nat_bits; /* full NAT pages */ 948 unsigned char *empty_nat_bits; /* empty NAT pages */ 949 #ifdef CONFIG_F2FS_CHECK_FS 950 char *nat_bitmap_mir; /* NAT bitmap mirror */ 951 #endif 952 int bitmap_size; /* bitmap size */ 953 }; 954 955 /* 956 * this structure is used as one of function parameters. 957 * all the information are dedicated to a given direct node block determined 958 * by the data offset in a file. 959 */ 960 struct dnode_of_data { 961 struct inode *inode; /* vfs inode pointer */ 962 struct page *inode_page; /* its inode page, NULL is possible */ 963 struct page *node_page; /* cached direct node page */ 964 nid_t nid; /* node id of the direct node block */ 965 unsigned int ofs_in_node; /* data offset in the node page */ 966 bool inode_page_locked; /* inode page is locked or not */ 967 bool node_changed; /* is node block changed */ 968 char cur_level; /* level of hole node page */ 969 char max_level; /* level of current page located */ 970 block_t data_blkaddr; /* block address of the node block */ 971 }; 972 973 static inline void set_new_dnode(struct dnode_of_data *dn, struct inode *inode, 974 struct page *ipage, struct page *npage, nid_t nid) 975 { 976 memset(dn, 0, sizeof(*dn)); 977 dn->inode = inode; 978 dn->inode_page = ipage; 979 dn->node_page = npage; 980 dn->nid = nid; 981 } 982 983 /* 984 * For SIT manager 985 * 986 * By default, there are 6 active log areas across the whole main area. 987 * When considering hot and cold data separation to reduce cleaning overhead, 988 * we split 3 for data logs and 3 for node logs as hot, warm, and cold types, 989 * respectively. 990 * In the current design, you should not change the numbers intentionally. 991 * Instead, as a mount option such as active_logs=x, you can use 2, 4, and 6 992 * logs individually according to the underlying devices. (default: 6) 993 * Just in case, on-disk layout covers maximum 16 logs that consist of 8 for 994 * data and 8 for node logs. 995 */ 996 #define NR_CURSEG_DATA_TYPE (3) 997 #define NR_CURSEG_NODE_TYPE (3) 998 #define NR_CURSEG_INMEM_TYPE (2) 999 #define NR_CURSEG_RO_TYPE (2) 1000 #define NR_CURSEG_PERSIST_TYPE (NR_CURSEG_DATA_TYPE + NR_CURSEG_NODE_TYPE) 1001 #define NR_CURSEG_TYPE (NR_CURSEG_INMEM_TYPE + NR_CURSEG_PERSIST_TYPE) 1002 1003 enum { 1004 CURSEG_HOT_DATA = 0, /* directory entry blocks */ 1005 CURSEG_WARM_DATA, /* data blocks */ 1006 CURSEG_COLD_DATA, /* multimedia or GCed data blocks */ 1007 CURSEG_HOT_NODE, /* direct node blocks of directory files */ 1008 CURSEG_WARM_NODE, /* direct node blocks of normal files */ 1009 CURSEG_COLD_NODE, /* indirect node blocks */ 1010 NR_PERSISTENT_LOG, /* number of persistent log */ 1011 CURSEG_COLD_DATA_PINNED = NR_PERSISTENT_LOG, 1012 /* pinned file that needs consecutive block address */ 1013 CURSEG_ALL_DATA_ATGC, /* SSR alloctor in hot/warm/cold data area */ 1014 NO_CHECK_TYPE, /* number of persistent & inmem log */ 1015 }; 1016 1017 struct flush_cmd { 1018 struct completion wait; 1019 struct llist_node llnode; 1020 nid_t ino; 1021 int ret; 1022 }; 1023 1024 struct flush_cmd_control { 1025 struct task_struct *f2fs_issue_flush; /* flush thread */ 1026 wait_queue_head_t flush_wait_queue; /* waiting queue for wake-up */ 1027 atomic_t issued_flush; /* # of issued flushes */ 1028 atomic_t queued_flush; /* # of queued flushes */ 1029 struct llist_head issue_list; /* list for command issue */ 1030 struct llist_node *dispatch_list; /* list for command dispatch */ 1031 }; 1032 1033 struct f2fs_sm_info { 1034 struct sit_info *sit_info; /* whole segment information */ 1035 struct free_segmap_info *free_info; /* free segment information */ 1036 struct dirty_seglist_info *dirty_info; /* dirty segment information */ 1037 struct curseg_info *curseg_array; /* active segment information */ 1038 1039 struct f2fs_rwsem curseg_lock; /* for preventing curseg change */ 1040 1041 block_t seg0_blkaddr; /* block address of 0'th segment */ 1042 block_t main_blkaddr; /* start block address of main area */ 1043 block_t ssa_blkaddr; /* start block address of SSA area */ 1044 1045 unsigned int segment_count; /* total # of segments */ 1046 unsigned int main_segments; /* # of segments in main area */ 1047 unsigned int reserved_segments; /* # of reserved segments */ 1048 unsigned int additional_reserved_segments;/* reserved segs for IO align feature */ 1049 unsigned int ovp_segments; /* # of overprovision segments */ 1050 1051 /* a threshold to reclaim prefree segments */ 1052 unsigned int rec_prefree_segments; 1053 1054 /* for batched trimming */ 1055 unsigned int trim_sections; /* # of sections to trim */ 1056 1057 struct list_head sit_entry_set; /* sit entry set list */ 1058 1059 unsigned int ipu_policy; /* in-place-update policy */ 1060 unsigned int min_ipu_util; /* in-place-update threshold */ 1061 unsigned int min_fsync_blocks; /* threshold for fsync */ 1062 unsigned int min_seq_blocks; /* threshold for sequential blocks */ 1063 unsigned int min_hot_blocks; /* threshold for hot block allocation */ 1064 unsigned int min_ssr_sections; /* threshold to trigger SSR allocation */ 1065 1066 /* for flush command control */ 1067 struct flush_cmd_control *fcc_info; 1068 1069 /* for discard command control */ 1070 struct discard_cmd_control *dcc_info; 1071 }; 1072 1073 /* 1074 * For superblock 1075 */ 1076 /* 1077 * COUNT_TYPE for monitoring 1078 * 1079 * f2fs monitors the number of several block types such as on-writeback, 1080 * dirty dentry blocks, dirty node blocks, and dirty meta blocks. 1081 */ 1082 #define WB_DATA_TYPE(p) (__is_cp_guaranteed(p) ? F2FS_WB_CP_DATA : F2FS_WB_DATA) 1083 enum count_type { 1084 F2FS_DIRTY_DENTS, 1085 F2FS_DIRTY_DATA, 1086 F2FS_DIRTY_QDATA, 1087 F2FS_DIRTY_NODES, 1088 F2FS_DIRTY_META, 1089 F2FS_INMEM_PAGES, 1090 F2FS_DIRTY_IMETA, 1091 F2FS_WB_CP_DATA, 1092 F2FS_WB_DATA, 1093 F2FS_RD_DATA, 1094 F2FS_RD_NODE, 1095 F2FS_RD_META, 1096 F2FS_DIO_WRITE, 1097 F2FS_DIO_READ, 1098 NR_COUNT_TYPE, 1099 }; 1100 1101 /* 1102 * The below are the page types of bios used in submit_bio(). 1103 * The available types are: 1104 * DATA User data pages. It operates as async mode. 1105 * NODE Node pages. It operates as async mode. 1106 * META FS metadata pages such as SIT, NAT, CP. 1107 * NR_PAGE_TYPE The number of page types. 1108 * META_FLUSH Make sure the previous pages are written 1109 * with waiting the bio's completion 1110 * ... Only can be used with META. 1111 */ 1112 #define PAGE_TYPE_OF_BIO(type) ((type) > META ? META : (type)) 1113 enum page_type { 1114 DATA, 1115 NODE, 1116 META, 1117 NR_PAGE_TYPE, 1118 META_FLUSH, 1119 INMEM, /* the below types are used by tracepoints only. */ 1120 INMEM_DROP, 1121 INMEM_INVALIDATE, 1122 INMEM_REVOKE, 1123 IPU, 1124 OPU, 1125 }; 1126 1127 enum temp_type { 1128 HOT = 0, /* must be zero for meta bio */ 1129 WARM, 1130 COLD, 1131 NR_TEMP_TYPE, 1132 }; 1133 1134 enum need_lock_type { 1135 LOCK_REQ = 0, 1136 LOCK_DONE, 1137 LOCK_RETRY, 1138 }; 1139 1140 enum cp_reason_type { 1141 CP_NO_NEEDED, 1142 CP_NON_REGULAR, 1143 CP_COMPRESSED, 1144 CP_HARDLINK, 1145 CP_SB_NEED_CP, 1146 CP_WRONG_PINO, 1147 CP_NO_SPC_ROLL, 1148 CP_NODE_NEED_CP, 1149 CP_FASTBOOT_MODE, 1150 CP_SPEC_LOG_NUM, 1151 CP_RECOVER_DIR, 1152 }; 1153 1154 enum iostat_type { 1155 /* WRITE IO */ 1156 APP_DIRECT_IO, /* app direct write IOs */ 1157 APP_BUFFERED_IO, /* app buffered write IOs */ 1158 APP_WRITE_IO, /* app write IOs */ 1159 APP_MAPPED_IO, /* app mapped IOs */ 1160 FS_DATA_IO, /* data IOs from kworker/fsync/reclaimer */ 1161 FS_NODE_IO, /* node IOs from kworker/fsync/reclaimer */ 1162 FS_META_IO, /* meta IOs from kworker/reclaimer */ 1163 FS_GC_DATA_IO, /* data IOs from forground gc */ 1164 FS_GC_NODE_IO, /* node IOs from forground gc */ 1165 FS_CP_DATA_IO, /* data IOs from checkpoint */ 1166 FS_CP_NODE_IO, /* node IOs from checkpoint */ 1167 FS_CP_META_IO, /* meta IOs from checkpoint */ 1168 1169 /* READ IO */ 1170 APP_DIRECT_READ_IO, /* app direct read IOs */ 1171 APP_BUFFERED_READ_IO, /* app buffered read IOs */ 1172 APP_READ_IO, /* app read IOs */ 1173 APP_MAPPED_READ_IO, /* app mapped read IOs */ 1174 FS_DATA_READ_IO, /* data read IOs */ 1175 FS_GDATA_READ_IO, /* data read IOs from background gc */ 1176 FS_CDATA_READ_IO, /* compressed data read IOs */ 1177 FS_NODE_READ_IO, /* node read IOs */ 1178 FS_META_READ_IO, /* meta read IOs */ 1179 1180 /* other */ 1181 FS_DISCARD, /* discard */ 1182 NR_IO_TYPE, 1183 }; 1184 1185 struct f2fs_io_info { 1186 struct f2fs_sb_info *sbi; /* f2fs_sb_info pointer */ 1187 nid_t ino; /* inode number */ 1188 enum page_type type; /* contains DATA/NODE/META/META_FLUSH */ 1189 enum temp_type temp; /* contains HOT/WARM/COLD */ 1190 int op; /* contains REQ_OP_ */ 1191 int op_flags; /* req_flag_bits */ 1192 block_t new_blkaddr; /* new block address to be written */ 1193 block_t old_blkaddr; /* old block address before Cow */ 1194 struct page *page; /* page to be written */ 1195 struct page *encrypted_page; /* encrypted page */ 1196 struct page *compressed_page; /* compressed page */ 1197 struct list_head list; /* serialize IOs */ 1198 bool submitted; /* indicate IO submission */ 1199 int need_lock; /* indicate we need to lock cp_rwsem */ 1200 bool in_list; /* indicate fio is in io_list */ 1201 bool is_por; /* indicate IO is from recovery or not */ 1202 bool retry; /* need to reallocate block address */ 1203 int compr_blocks; /* # of compressed block addresses */ 1204 bool encrypted; /* indicate file is encrypted */ 1205 enum iostat_type io_type; /* io type */ 1206 struct writeback_control *io_wbc; /* writeback control */ 1207 struct bio **bio; /* bio for ipu */ 1208 sector_t *last_block; /* last block number in bio */ 1209 unsigned char version; /* version of the node */ 1210 }; 1211 1212 struct bio_entry { 1213 struct bio *bio; 1214 struct list_head list; 1215 }; 1216 1217 #define is_read_io(rw) ((rw) == READ) 1218 struct f2fs_bio_info { 1219 struct f2fs_sb_info *sbi; /* f2fs superblock */ 1220 struct bio *bio; /* bios to merge */ 1221 sector_t last_block_in_bio; /* last block number */ 1222 struct f2fs_io_info fio; /* store buffered io info. */ 1223 struct f2fs_rwsem io_rwsem; /* blocking op for bio */ 1224 spinlock_t io_lock; /* serialize DATA/NODE IOs */ 1225 struct list_head io_list; /* track fios */ 1226 struct list_head bio_list; /* bio entry list head */ 1227 struct f2fs_rwsem bio_list_lock; /* lock to protect bio entry list */ 1228 }; 1229 1230 #define FDEV(i) (sbi->devs[i]) 1231 #define RDEV(i) (raw_super->devs[i]) 1232 struct f2fs_dev_info { 1233 struct block_device *bdev; 1234 char path[MAX_PATH_LEN]; 1235 unsigned int total_segments; 1236 block_t start_blk; 1237 block_t end_blk; 1238 #ifdef CONFIG_BLK_DEV_ZONED 1239 unsigned int nr_blkz; /* Total number of zones */ 1240 unsigned long *blkz_seq; /* Bitmap indicating sequential zones */ 1241 block_t *zone_capacity_blocks; /* Array of zone capacity in blks */ 1242 #endif 1243 }; 1244 1245 enum inode_type { 1246 DIR_INODE, /* for dirty dir inode */ 1247 FILE_INODE, /* for dirty regular/symlink inode */ 1248 DIRTY_META, /* for all dirtied inode metadata */ 1249 ATOMIC_FILE, /* for all atomic files */ 1250 NR_INODE_TYPE, 1251 }; 1252 1253 /* for inner inode cache management */ 1254 struct inode_management { 1255 struct radix_tree_root ino_root; /* ino entry array */ 1256 spinlock_t ino_lock; /* for ino entry lock */ 1257 struct list_head ino_list; /* inode list head */ 1258 unsigned long ino_num; /* number of entries */ 1259 }; 1260 1261 /* for GC_AT */ 1262 struct atgc_management { 1263 bool atgc_enabled; /* ATGC is enabled or not */ 1264 struct rb_root_cached root; /* root of victim rb-tree */ 1265 struct list_head victim_list; /* linked with all victim entries */ 1266 unsigned int victim_count; /* victim count in rb-tree */ 1267 unsigned int candidate_ratio; /* candidate ratio */ 1268 unsigned int max_candidate_count; /* max candidate count */ 1269 unsigned int age_weight; /* age weight, vblock_weight = 100 - age_weight */ 1270 unsigned long long age_threshold; /* age threshold */ 1271 }; 1272 1273 /* For s_flag in struct f2fs_sb_info */ 1274 enum { 1275 SBI_IS_DIRTY, /* dirty flag for checkpoint */ 1276 SBI_IS_CLOSE, /* specify unmounting */ 1277 SBI_NEED_FSCK, /* need fsck.f2fs to fix */ 1278 SBI_POR_DOING, /* recovery is doing or not */ 1279 SBI_NEED_SB_WRITE, /* need to recover superblock */ 1280 SBI_NEED_CP, /* need to checkpoint */ 1281 SBI_IS_SHUTDOWN, /* shutdown by ioctl */ 1282 SBI_IS_RECOVERED, /* recovered orphan/data */ 1283 SBI_CP_DISABLED, /* CP was disabled last mount */ 1284 SBI_CP_DISABLED_QUICK, /* CP was disabled quickly */ 1285 SBI_QUOTA_NEED_FLUSH, /* need to flush quota info in CP */ 1286 SBI_QUOTA_SKIP_FLUSH, /* skip flushing quota in current CP */ 1287 SBI_QUOTA_NEED_REPAIR, /* quota file may be corrupted */ 1288 SBI_IS_RESIZEFS, /* resizefs is in process */ 1289 }; 1290 1291 enum { 1292 CP_TIME, 1293 REQ_TIME, 1294 DISCARD_TIME, 1295 GC_TIME, 1296 DISABLE_TIME, 1297 UMOUNT_DISCARD_TIMEOUT, 1298 MAX_TIME, 1299 }; 1300 1301 enum { 1302 GC_NORMAL, 1303 GC_IDLE_CB, 1304 GC_IDLE_GREEDY, 1305 GC_IDLE_AT, 1306 GC_URGENT_HIGH, 1307 GC_URGENT_LOW, 1308 MAX_GC_MODE, 1309 }; 1310 1311 enum { 1312 BGGC_MODE_ON, /* background gc is on */ 1313 BGGC_MODE_OFF, /* background gc is off */ 1314 BGGC_MODE_SYNC, /* 1315 * background gc is on, migrating blocks 1316 * like foreground gc 1317 */ 1318 }; 1319 1320 enum { 1321 FS_MODE_ADAPTIVE, /* use both lfs/ssr allocation */ 1322 FS_MODE_LFS, /* use lfs allocation only */ 1323 FS_MODE_FRAGMENT_SEG, /* segment fragmentation mode */ 1324 FS_MODE_FRAGMENT_BLK, /* block fragmentation mode */ 1325 }; 1326 1327 enum { 1328 WHINT_MODE_OFF, /* not pass down write hints */ 1329 WHINT_MODE_USER, /* try to pass down hints given by users */ 1330 WHINT_MODE_FS, /* pass down hints with F2FS policy */ 1331 }; 1332 1333 enum { 1334 ALLOC_MODE_DEFAULT, /* stay default */ 1335 ALLOC_MODE_REUSE, /* reuse segments as much as possible */ 1336 }; 1337 1338 enum fsync_mode { 1339 FSYNC_MODE_POSIX, /* fsync follows posix semantics */ 1340 FSYNC_MODE_STRICT, /* fsync behaves in line with ext4 */ 1341 FSYNC_MODE_NOBARRIER, /* fsync behaves nobarrier based on posix */ 1342 }; 1343 1344 enum { 1345 COMPR_MODE_FS, /* 1346 * automatically compress compression 1347 * enabled files 1348 */ 1349 COMPR_MODE_USER, /* 1350 * automatical compression is disabled. 1351 * user can control the file compression 1352 * using ioctls 1353 */ 1354 }; 1355 1356 enum { 1357 DISCARD_UNIT_BLOCK, /* basic discard unit is block */ 1358 DISCARD_UNIT_SEGMENT, /* basic discard unit is segment */ 1359 DISCARD_UNIT_SECTION, /* basic discard unit is section */ 1360 }; 1361 1362 static inline int f2fs_test_bit(unsigned int nr, char *addr); 1363 static inline void f2fs_set_bit(unsigned int nr, char *addr); 1364 static inline void f2fs_clear_bit(unsigned int nr, char *addr); 1365 1366 /* 1367 * Layout of f2fs page.private: 1368 * 1369 * Layout A: lowest bit should be 1 1370 * | bit0 = 1 | bit1 | bit2 | ... | bit MAX | private data .... | 1371 * bit 0 PAGE_PRIVATE_NOT_POINTER 1372 * bit 1 PAGE_PRIVATE_ATOMIC_WRITE 1373 * bit 2 PAGE_PRIVATE_DUMMY_WRITE 1374 * bit 3 PAGE_PRIVATE_ONGOING_MIGRATION 1375 * bit 4 PAGE_PRIVATE_INLINE_INODE 1376 * bit 5 PAGE_PRIVATE_REF_RESOURCE 1377 * bit 6- f2fs private data 1378 * 1379 * Layout B: lowest bit should be 0 1380 * page.private is a wrapped pointer. 1381 */ 1382 enum { 1383 PAGE_PRIVATE_NOT_POINTER, /* private contains non-pointer data */ 1384 PAGE_PRIVATE_ATOMIC_WRITE, /* data page from atomic write path */ 1385 PAGE_PRIVATE_DUMMY_WRITE, /* data page for padding aligned IO */ 1386 PAGE_PRIVATE_ONGOING_MIGRATION, /* data page which is on-going migrating */ 1387 PAGE_PRIVATE_INLINE_INODE, /* inode page contains inline data */ 1388 PAGE_PRIVATE_REF_RESOURCE, /* dirty page has referenced resources */ 1389 PAGE_PRIVATE_MAX 1390 }; 1391 1392 #define PAGE_PRIVATE_GET_FUNC(name, flagname) \ 1393 static inline bool page_private_##name(struct page *page) \ 1394 { \ 1395 return PagePrivate(page) && \ 1396 test_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)) && \ 1397 test_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \ 1398 } 1399 1400 #define PAGE_PRIVATE_SET_FUNC(name, flagname) \ 1401 static inline void set_page_private_##name(struct page *page) \ 1402 { \ 1403 if (!PagePrivate(page)) { \ 1404 get_page(page); \ 1405 SetPagePrivate(page); \ 1406 set_page_private(page, 0); \ 1407 } \ 1408 set_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)); \ 1409 set_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \ 1410 } 1411 1412 #define PAGE_PRIVATE_CLEAR_FUNC(name, flagname) \ 1413 static inline void clear_page_private_##name(struct page *page) \ 1414 { \ 1415 clear_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \ 1416 if (page_private(page) == 1 << PAGE_PRIVATE_NOT_POINTER) { \ 1417 set_page_private(page, 0); \ 1418 if (PagePrivate(page)) { \ 1419 ClearPagePrivate(page); \ 1420 put_page(page); \ 1421 }\ 1422 } \ 1423 } 1424 1425 PAGE_PRIVATE_GET_FUNC(nonpointer, NOT_POINTER); 1426 PAGE_PRIVATE_GET_FUNC(reference, REF_RESOURCE); 1427 PAGE_PRIVATE_GET_FUNC(inline, INLINE_INODE); 1428 PAGE_PRIVATE_GET_FUNC(gcing, ONGOING_MIGRATION); 1429 PAGE_PRIVATE_GET_FUNC(atomic, ATOMIC_WRITE); 1430 PAGE_PRIVATE_GET_FUNC(dummy, DUMMY_WRITE); 1431 1432 PAGE_PRIVATE_SET_FUNC(reference, REF_RESOURCE); 1433 PAGE_PRIVATE_SET_FUNC(inline, INLINE_INODE); 1434 PAGE_PRIVATE_SET_FUNC(gcing, ONGOING_MIGRATION); 1435 PAGE_PRIVATE_SET_FUNC(atomic, ATOMIC_WRITE); 1436 PAGE_PRIVATE_SET_FUNC(dummy, DUMMY_WRITE); 1437 1438 PAGE_PRIVATE_CLEAR_FUNC(reference, REF_RESOURCE); 1439 PAGE_PRIVATE_CLEAR_FUNC(inline, INLINE_INODE); 1440 PAGE_PRIVATE_CLEAR_FUNC(gcing, ONGOING_MIGRATION); 1441 PAGE_PRIVATE_CLEAR_FUNC(atomic, ATOMIC_WRITE); 1442 PAGE_PRIVATE_CLEAR_FUNC(dummy, DUMMY_WRITE); 1443 1444 static inline unsigned long get_page_private_data(struct page *page) 1445 { 1446 unsigned long data = page_private(page); 1447 1448 if (!test_bit(PAGE_PRIVATE_NOT_POINTER, &data)) 1449 return 0; 1450 return data >> PAGE_PRIVATE_MAX; 1451 } 1452 1453 static inline void set_page_private_data(struct page *page, unsigned long data) 1454 { 1455 if (!PagePrivate(page)) { 1456 get_page(page); 1457 SetPagePrivate(page); 1458 set_page_private(page, 0); 1459 } 1460 set_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)); 1461 page_private(page) |= data << PAGE_PRIVATE_MAX; 1462 } 1463 1464 static inline void clear_page_private_data(struct page *page) 1465 { 1466 page_private(page) &= (1 << PAGE_PRIVATE_MAX) - 1; 1467 if (page_private(page) == 1 << PAGE_PRIVATE_NOT_POINTER) { 1468 set_page_private(page, 0); 1469 if (PagePrivate(page)) { 1470 ClearPagePrivate(page); 1471 put_page(page); 1472 } 1473 } 1474 } 1475 1476 /* For compression */ 1477 enum compress_algorithm_type { 1478 COMPRESS_LZO, 1479 COMPRESS_LZ4, 1480 COMPRESS_ZSTD, 1481 COMPRESS_LZORLE, 1482 COMPRESS_MAX, 1483 }; 1484 1485 enum compress_flag { 1486 COMPRESS_CHKSUM, 1487 COMPRESS_MAX_FLAG, 1488 }; 1489 1490 #define COMPRESS_WATERMARK 20 1491 #define COMPRESS_PERCENT 20 1492 1493 #define COMPRESS_DATA_RESERVED_SIZE 4 1494 struct compress_data { 1495 __le32 clen; /* compressed data size */ 1496 __le32 chksum; /* compressed data chksum */ 1497 __le32 reserved[COMPRESS_DATA_RESERVED_SIZE]; /* reserved */ 1498 u8 cdata[]; /* compressed data */ 1499 }; 1500 1501 #define COMPRESS_HEADER_SIZE (sizeof(struct compress_data)) 1502 1503 #define F2FS_COMPRESSED_PAGE_MAGIC 0xF5F2C000 1504 1505 #define COMPRESS_LEVEL_OFFSET 8 1506 1507 /* compress context */ 1508 struct compress_ctx { 1509 struct inode *inode; /* inode the context belong to */ 1510 pgoff_t cluster_idx; /* cluster index number */ 1511 unsigned int cluster_size; /* page count in cluster */ 1512 unsigned int log_cluster_size; /* log of cluster size */ 1513 struct page **rpages; /* pages store raw data in cluster */ 1514 unsigned int nr_rpages; /* total page number in rpages */ 1515 struct page **cpages; /* pages store compressed data in cluster */ 1516 unsigned int nr_cpages; /* total page number in cpages */ 1517 unsigned int valid_nr_cpages; /* valid page number in cpages */ 1518 void *rbuf; /* virtual mapped address on rpages */ 1519 struct compress_data *cbuf; /* virtual mapped address on cpages */ 1520 size_t rlen; /* valid data length in rbuf */ 1521 size_t clen; /* valid data length in cbuf */ 1522 void *private; /* payload buffer for specified compression algorithm */ 1523 void *private2; /* extra payload buffer */ 1524 }; 1525 1526 /* compress context for write IO path */ 1527 struct compress_io_ctx { 1528 u32 magic; /* magic number to indicate page is compressed */ 1529 struct inode *inode; /* inode the context belong to */ 1530 struct page **rpages; /* pages store raw data in cluster */ 1531 unsigned int nr_rpages; /* total page number in rpages */ 1532 atomic_t pending_pages; /* in-flight compressed page count */ 1533 }; 1534 1535 /* Context for decompressing one cluster on the read IO path */ 1536 struct decompress_io_ctx { 1537 u32 magic; /* magic number to indicate page is compressed */ 1538 struct inode *inode; /* inode the context belong to */ 1539 pgoff_t cluster_idx; /* cluster index number */ 1540 unsigned int cluster_size; /* page count in cluster */ 1541 unsigned int log_cluster_size; /* log of cluster size */ 1542 struct page **rpages; /* pages store raw data in cluster */ 1543 unsigned int nr_rpages; /* total page number in rpages */ 1544 struct page **cpages; /* pages store compressed data in cluster */ 1545 unsigned int nr_cpages; /* total page number in cpages */ 1546 struct page **tpages; /* temp pages to pad holes in cluster */ 1547 void *rbuf; /* virtual mapped address on rpages */ 1548 struct compress_data *cbuf; /* virtual mapped address on cpages */ 1549 size_t rlen; /* valid data length in rbuf */ 1550 size_t clen; /* valid data length in cbuf */ 1551 1552 /* 1553 * The number of compressed pages remaining to be read in this cluster. 1554 * This is initially nr_cpages. It is decremented by 1 each time a page 1555 * has been read (or failed to be read). When it reaches 0, the cluster 1556 * is decompressed (or an error is reported). 1557 * 1558 * If an error occurs before all the pages have been submitted for I/O, 1559 * then this will never reach 0. In this case the I/O submitter is 1560 * responsible for calling f2fs_decompress_end_io() instead. 1561 */ 1562 atomic_t remaining_pages; 1563 1564 /* 1565 * Number of references to this decompress_io_ctx. 1566 * 1567 * One reference is held for I/O completion. This reference is dropped 1568 * after the pagecache pages are updated and unlocked -- either after 1569 * decompression (and verity if enabled), or after an error. 1570 * 1571 * In addition, each compressed page holds a reference while it is in a 1572 * bio. These references are necessary prevent compressed pages from 1573 * being freed while they are still in a bio. 1574 */ 1575 refcount_t refcnt; 1576 1577 bool failed; /* IO error occurred before decompression? */ 1578 bool need_verity; /* need fs-verity verification after decompression? */ 1579 void *private; /* payload buffer for specified decompression algorithm */ 1580 void *private2; /* extra payload buffer */ 1581 struct work_struct verity_work; /* work to verify the decompressed pages */ 1582 }; 1583 1584 #define NULL_CLUSTER ((unsigned int)(~0)) 1585 #define MIN_COMPRESS_LOG_SIZE 2 1586 #define MAX_COMPRESS_LOG_SIZE 8 1587 #define MAX_COMPRESS_WINDOW_SIZE(log_size) ((PAGE_SIZE) << (log_size)) 1588 1589 struct f2fs_sb_info { 1590 struct super_block *sb; /* pointer to VFS super block */ 1591 struct proc_dir_entry *s_proc; /* proc entry */ 1592 struct f2fs_super_block *raw_super; /* raw super block pointer */ 1593 struct f2fs_rwsem sb_lock; /* lock for raw super block */ 1594 int valid_super_block; /* valid super block no */ 1595 unsigned long s_flag; /* flags for sbi */ 1596 struct mutex writepages; /* mutex for writepages() */ 1597 1598 #ifdef CONFIG_BLK_DEV_ZONED 1599 unsigned int blocks_per_blkz; /* F2FS blocks per zone */ 1600 unsigned int log_blocks_per_blkz; /* log2 F2FS blocks per zone */ 1601 #endif 1602 1603 /* for node-related operations */ 1604 struct f2fs_nm_info *nm_info; /* node manager */ 1605 struct inode *node_inode; /* cache node blocks */ 1606 1607 /* for segment-related operations */ 1608 struct f2fs_sm_info *sm_info; /* segment manager */ 1609 1610 /* for bio operations */ 1611 struct f2fs_bio_info *write_io[NR_PAGE_TYPE]; /* for write bios */ 1612 /* keep migration IO order for LFS mode */ 1613 struct f2fs_rwsem io_order_lock; 1614 mempool_t *write_io_dummy; /* Dummy pages */ 1615 1616 /* for checkpoint */ 1617 struct f2fs_checkpoint *ckpt; /* raw checkpoint pointer */ 1618 int cur_cp_pack; /* remain current cp pack */ 1619 spinlock_t cp_lock; /* for flag in ckpt */ 1620 struct inode *meta_inode; /* cache meta blocks */ 1621 struct f2fs_rwsem cp_global_sem; /* checkpoint procedure lock */ 1622 struct f2fs_rwsem cp_rwsem; /* blocking FS operations */ 1623 struct f2fs_rwsem node_write; /* locking node writes */ 1624 struct f2fs_rwsem node_change; /* locking node change */ 1625 wait_queue_head_t cp_wait; 1626 unsigned long last_time[MAX_TIME]; /* to store time in jiffies */ 1627 long interval_time[MAX_TIME]; /* to store thresholds */ 1628 struct ckpt_req_control cprc_info; /* for checkpoint request control */ 1629 1630 struct inode_management im[MAX_INO_ENTRY]; /* manage inode cache */ 1631 1632 spinlock_t fsync_node_lock; /* for node entry lock */ 1633 struct list_head fsync_node_list; /* node list head */ 1634 unsigned int fsync_seg_id; /* sequence id */ 1635 unsigned int fsync_node_num; /* number of node entries */ 1636 1637 /* for orphan inode, use 0'th array */ 1638 unsigned int max_orphans; /* max orphan inodes */ 1639 1640 /* for inode management */ 1641 struct list_head inode_list[NR_INODE_TYPE]; /* dirty inode list */ 1642 spinlock_t inode_lock[NR_INODE_TYPE]; /* for dirty inode list lock */ 1643 struct mutex flush_lock; /* for flush exclusion */ 1644 1645 /* for extent tree cache */ 1646 struct radix_tree_root extent_tree_root;/* cache extent cache entries */ 1647 struct mutex extent_tree_lock; /* locking extent radix tree */ 1648 struct list_head extent_list; /* lru list for shrinker */ 1649 spinlock_t extent_lock; /* locking extent lru list */ 1650 atomic_t total_ext_tree; /* extent tree count */ 1651 struct list_head zombie_list; /* extent zombie tree list */ 1652 atomic_t total_zombie_tree; /* extent zombie tree count */ 1653 atomic_t total_ext_node; /* extent info count */ 1654 1655 /* basic filesystem units */ 1656 unsigned int log_sectors_per_block; /* log2 sectors per block */ 1657 unsigned int log_blocksize; /* log2 block size */ 1658 unsigned int blocksize; /* block size */ 1659 unsigned int root_ino_num; /* root inode number*/ 1660 unsigned int node_ino_num; /* node inode number*/ 1661 unsigned int meta_ino_num; /* meta inode number*/ 1662 unsigned int log_blocks_per_seg; /* log2 blocks per segment */ 1663 unsigned int blocks_per_seg; /* blocks per segment */ 1664 unsigned int segs_per_sec; /* segments per section */ 1665 unsigned int secs_per_zone; /* sections per zone */ 1666 unsigned int total_sections; /* total section count */ 1667 unsigned int total_node_count; /* total node block count */ 1668 unsigned int total_valid_node_count; /* valid node block count */ 1669 int dir_level; /* directory level */ 1670 int readdir_ra; /* readahead inode in readdir */ 1671 u64 max_io_bytes; /* max io bytes to merge IOs */ 1672 1673 block_t user_block_count; /* # of user blocks */ 1674 block_t total_valid_block_count; /* # of valid blocks */ 1675 block_t discard_blks; /* discard command candidats */ 1676 block_t last_valid_block_count; /* for recovery */ 1677 block_t reserved_blocks; /* configurable reserved blocks */ 1678 block_t current_reserved_blocks; /* current reserved blocks */ 1679 1680 /* Additional tracking for no checkpoint mode */ 1681 block_t unusable_block_count; /* # of blocks saved by last cp */ 1682 1683 unsigned int nquota_files; /* # of quota sysfile */ 1684 struct f2fs_rwsem quota_sem; /* blocking cp for flags */ 1685 1686 /* # of pages, see count_type */ 1687 atomic_t nr_pages[NR_COUNT_TYPE]; 1688 /* # of allocated blocks */ 1689 struct percpu_counter alloc_valid_block_count; 1690 1691 /* writeback control */ 1692 atomic_t wb_sync_req[META]; /* count # of WB_SYNC threads */ 1693 1694 /* valid inode count */ 1695 struct percpu_counter total_valid_inode_count; 1696 1697 struct f2fs_mount_info mount_opt; /* mount options */ 1698 1699 /* for cleaning operations */ 1700 struct f2fs_rwsem gc_lock; /* 1701 * semaphore for GC, avoid 1702 * race between GC and GC or CP 1703 */ 1704 struct f2fs_gc_kthread *gc_thread; /* GC thread */ 1705 struct atgc_management am; /* atgc management */ 1706 unsigned int cur_victim_sec; /* current victim section num */ 1707 unsigned int gc_mode; /* current GC state */ 1708 unsigned int next_victim_seg[2]; /* next segment in victim section */ 1709 spinlock_t gc_urgent_high_lock; 1710 bool gc_urgent_high_limited; /* indicates having limited trial count */ 1711 unsigned int gc_urgent_high_remaining; /* remaining trial count for GC_URGENT_HIGH */ 1712 1713 /* for skip statistic */ 1714 unsigned int atomic_files; /* # of opened atomic file */ 1715 unsigned long long skipped_atomic_files[2]; /* FG_GC and BG_GC */ 1716 unsigned long long skipped_gc_rwsem; /* FG_GC only */ 1717 1718 /* threshold for gc trials on pinned files */ 1719 u64 gc_pin_file_threshold; 1720 struct f2fs_rwsem pin_sem; 1721 1722 /* maximum # of trials to find a victim segment for SSR and GC */ 1723 unsigned int max_victim_search; 1724 /* migration granularity of garbage collection, unit: segment */ 1725 unsigned int migration_granularity; 1726 1727 /* 1728 * for stat information. 1729 * one is for the LFS mode, and the other is for the SSR mode. 1730 */ 1731 #ifdef CONFIG_F2FS_STAT_FS 1732 struct f2fs_stat_info *stat_info; /* FS status information */ 1733 atomic_t meta_count[META_MAX]; /* # of meta blocks */ 1734 unsigned int segment_count[2]; /* # of allocated segments */ 1735 unsigned int block_count[2]; /* # of allocated blocks */ 1736 atomic_t inplace_count; /* # of inplace update */ 1737 atomic64_t total_hit_ext; /* # of lookup extent cache */ 1738 atomic64_t read_hit_rbtree; /* # of hit rbtree extent node */ 1739 atomic64_t read_hit_largest; /* # of hit largest extent node */ 1740 atomic64_t read_hit_cached; /* # of hit cached extent node */ 1741 atomic_t inline_xattr; /* # of inline_xattr inodes */ 1742 atomic_t inline_inode; /* # of inline_data inodes */ 1743 atomic_t inline_dir; /* # of inline_dentry inodes */ 1744 atomic_t compr_inode; /* # of compressed inodes */ 1745 atomic64_t compr_blocks; /* # of compressed blocks */ 1746 atomic_t vw_cnt; /* # of volatile writes */ 1747 atomic_t max_aw_cnt; /* max # of atomic writes */ 1748 atomic_t max_vw_cnt; /* max # of volatile writes */ 1749 unsigned int io_skip_bggc; /* skip background gc for in-flight IO */ 1750 unsigned int other_skip_bggc; /* skip background gc for other reasons */ 1751 unsigned int ndirty_inode[NR_INODE_TYPE]; /* # of dirty inodes */ 1752 #endif 1753 spinlock_t stat_lock; /* lock for stat operations */ 1754 1755 /* to attach REQ_META|REQ_FUA flags */ 1756 unsigned int data_io_flag; 1757 unsigned int node_io_flag; 1758 1759 /* For sysfs suppport */ 1760 struct kobject s_kobj; /* /sys/fs/f2fs/<devname> */ 1761 struct completion s_kobj_unregister; 1762 1763 struct kobject s_stat_kobj; /* /sys/fs/f2fs/<devname>/stat */ 1764 struct completion s_stat_kobj_unregister; 1765 1766 struct kobject s_feature_list_kobj; /* /sys/fs/f2fs/<devname>/feature_list */ 1767 struct completion s_feature_list_kobj_unregister; 1768 1769 /* For shrinker support */ 1770 struct list_head s_list; 1771 struct mutex umount_mutex; 1772 unsigned int shrinker_run_no; 1773 1774 /* For multi devices */ 1775 int s_ndevs; /* number of devices */ 1776 struct f2fs_dev_info *devs; /* for device list */ 1777 unsigned int dirty_device; /* for checkpoint data flush */ 1778 spinlock_t dev_lock; /* protect dirty_device */ 1779 bool aligned_blksize; /* all devices has the same logical blksize */ 1780 1781 /* For write statistics */ 1782 u64 sectors_written_start; 1783 u64 kbytes_written; 1784 1785 /* Reference to checksum algorithm driver via cryptoapi */ 1786 struct crypto_shash *s_chksum_driver; 1787 1788 /* Precomputed FS UUID checksum for seeding other checksums */ 1789 __u32 s_chksum_seed; 1790 1791 struct workqueue_struct *post_read_wq; /* post read workqueue */ 1792 1793 struct kmem_cache *inline_xattr_slab; /* inline xattr entry */ 1794 unsigned int inline_xattr_slab_size; /* default inline xattr slab size */ 1795 1796 /* For reclaimed segs statistics per each GC mode */ 1797 unsigned int gc_segment_mode; /* GC state for reclaimed segments */ 1798 unsigned int gc_reclaimed_segs[MAX_GC_MODE]; /* Reclaimed segs for each mode */ 1799 1800 unsigned long seq_file_ra_mul; /* multiplier for ra_pages of seq. files in fadvise */ 1801 1802 int max_fragment_chunk; /* max chunk size for block fragmentation mode */ 1803 int max_fragment_hole; /* max hole size for block fragmentation mode */ 1804 1805 #ifdef CONFIG_F2FS_FS_COMPRESSION 1806 struct kmem_cache *page_array_slab; /* page array entry */ 1807 unsigned int page_array_slab_size; /* default page array slab size */ 1808 1809 /* For runtime compression statistics */ 1810 u64 compr_written_block; 1811 u64 compr_saved_block; 1812 u32 compr_new_inode; 1813 1814 /* For compressed block cache */ 1815 struct inode *compress_inode; /* cache compressed blocks */ 1816 unsigned int compress_percent; /* cache page percentage */ 1817 unsigned int compress_watermark; /* cache page watermark */ 1818 atomic_t compress_page_hit; /* cache hit count */ 1819 #endif 1820 1821 #ifdef CONFIG_F2FS_IOSTAT 1822 /* For app/fs IO statistics */ 1823 spinlock_t iostat_lock; 1824 unsigned long long rw_iostat[NR_IO_TYPE]; 1825 unsigned long long prev_rw_iostat[NR_IO_TYPE]; 1826 bool iostat_enable; 1827 unsigned long iostat_next_period; 1828 unsigned int iostat_period_ms; 1829 1830 /* For io latency related statistics info in one iostat period */ 1831 spinlock_t iostat_lat_lock; 1832 struct iostat_lat_info *iostat_io_lat; 1833 #endif 1834 }; 1835 1836 #ifdef CONFIG_F2FS_FAULT_INJECTION 1837 #define f2fs_show_injection_info(sbi, type) \ 1838 printk_ratelimited("%sF2FS-fs (%s) : inject %s in %s of %pS\n", \ 1839 KERN_INFO, sbi->sb->s_id, \ 1840 f2fs_fault_name[type], \ 1841 __func__, __builtin_return_address(0)) 1842 static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type) 1843 { 1844 struct f2fs_fault_info *ffi = &F2FS_OPTION(sbi).fault_info; 1845 1846 if (!ffi->inject_rate) 1847 return false; 1848 1849 if (!IS_FAULT_SET(ffi, type)) 1850 return false; 1851 1852 atomic_inc(&ffi->inject_ops); 1853 if (atomic_read(&ffi->inject_ops) >= ffi->inject_rate) { 1854 atomic_set(&ffi->inject_ops, 0); 1855 return true; 1856 } 1857 return false; 1858 } 1859 #else 1860 #define f2fs_show_injection_info(sbi, type) do { } while (0) 1861 static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type) 1862 { 1863 return false; 1864 } 1865 #endif 1866 1867 /* 1868 * Test if the mounted volume is a multi-device volume. 1869 * - For a single regular disk volume, sbi->s_ndevs is 0. 1870 * - For a single zoned disk volume, sbi->s_ndevs is 1. 1871 * - For a multi-device volume, sbi->s_ndevs is always 2 or more. 1872 */ 1873 static inline bool f2fs_is_multi_device(struct f2fs_sb_info *sbi) 1874 { 1875 return sbi->s_ndevs > 1; 1876 } 1877 1878 static inline void f2fs_update_time(struct f2fs_sb_info *sbi, int type) 1879 { 1880 unsigned long now = jiffies; 1881 1882 sbi->last_time[type] = now; 1883 1884 /* DISCARD_TIME and GC_TIME are based on REQ_TIME */ 1885 if (type == REQ_TIME) { 1886 sbi->last_time[DISCARD_TIME] = now; 1887 sbi->last_time[GC_TIME] = now; 1888 } 1889 } 1890 1891 static inline bool f2fs_time_over(struct f2fs_sb_info *sbi, int type) 1892 { 1893 unsigned long interval = sbi->interval_time[type] * HZ; 1894 1895 return time_after(jiffies, sbi->last_time[type] + interval); 1896 } 1897 1898 static inline unsigned int f2fs_time_to_wait(struct f2fs_sb_info *sbi, 1899 int type) 1900 { 1901 unsigned long interval = sbi->interval_time[type] * HZ; 1902 unsigned int wait_ms = 0; 1903 long delta; 1904 1905 delta = (sbi->last_time[type] + interval) - jiffies; 1906 if (delta > 0) 1907 wait_ms = jiffies_to_msecs(delta); 1908 1909 return wait_ms; 1910 } 1911 1912 /* 1913 * Inline functions 1914 */ 1915 static inline u32 __f2fs_crc32(struct f2fs_sb_info *sbi, u32 crc, 1916 const void *address, unsigned int length) 1917 { 1918 struct { 1919 struct shash_desc shash; 1920 char ctx[4]; 1921 } desc; 1922 int err; 1923 1924 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver) != sizeof(desc.ctx)); 1925 1926 desc.shash.tfm = sbi->s_chksum_driver; 1927 *(u32 *)desc.ctx = crc; 1928 1929 err = crypto_shash_update(&desc.shash, address, length); 1930 BUG_ON(err); 1931 1932 return *(u32 *)desc.ctx; 1933 } 1934 1935 static inline u32 f2fs_crc32(struct f2fs_sb_info *sbi, const void *address, 1936 unsigned int length) 1937 { 1938 return __f2fs_crc32(sbi, F2FS_SUPER_MAGIC, address, length); 1939 } 1940 1941 static inline bool f2fs_crc_valid(struct f2fs_sb_info *sbi, __u32 blk_crc, 1942 void *buf, size_t buf_size) 1943 { 1944 return f2fs_crc32(sbi, buf, buf_size) == blk_crc; 1945 } 1946 1947 static inline u32 f2fs_chksum(struct f2fs_sb_info *sbi, u32 crc, 1948 const void *address, unsigned int length) 1949 { 1950 return __f2fs_crc32(sbi, crc, address, length); 1951 } 1952 1953 static inline struct f2fs_inode_info *F2FS_I(struct inode *inode) 1954 { 1955 return container_of(inode, struct f2fs_inode_info, vfs_inode); 1956 } 1957 1958 static inline struct f2fs_sb_info *F2FS_SB(struct super_block *sb) 1959 { 1960 return sb->s_fs_info; 1961 } 1962 1963 static inline struct f2fs_sb_info *F2FS_I_SB(struct inode *inode) 1964 { 1965 return F2FS_SB(inode->i_sb); 1966 } 1967 1968 static inline struct f2fs_sb_info *F2FS_M_SB(struct address_space *mapping) 1969 { 1970 return F2FS_I_SB(mapping->host); 1971 } 1972 1973 static inline struct f2fs_sb_info *F2FS_P_SB(struct page *page) 1974 { 1975 return F2FS_M_SB(page_file_mapping(page)); 1976 } 1977 1978 static inline struct f2fs_super_block *F2FS_RAW_SUPER(struct f2fs_sb_info *sbi) 1979 { 1980 return (struct f2fs_super_block *)(sbi->raw_super); 1981 } 1982 1983 static inline struct f2fs_checkpoint *F2FS_CKPT(struct f2fs_sb_info *sbi) 1984 { 1985 return (struct f2fs_checkpoint *)(sbi->ckpt); 1986 } 1987 1988 static inline struct f2fs_node *F2FS_NODE(struct page *page) 1989 { 1990 return (struct f2fs_node *)page_address(page); 1991 } 1992 1993 static inline struct f2fs_inode *F2FS_INODE(struct page *page) 1994 { 1995 return &((struct f2fs_node *)page_address(page))->i; 1996 } 1997 1998 static inline struct f2fs_nm_info *NM_I(struct f2fs_sb_info *sbi) 1999 { 2000 return (struct f2fs_nm_info *)(sbi->nm_info); 2001 } 2002 2003 static inline struct f2fs_sm_info *SM_I(struct f2fs_sb_info *sbi) 2004 { 2005 return (struct f2fs_sm_info *)(sbi->sm_info); 2006 } 2007 2008 static inline struct sit_info *SIT_I(struct f2fs_sb_info *sbi) 2009 { 2010 return (struct sit_info *)(SM_I(sbi)->sit_info); 2011 } 2012 2013 static inline struct free_segmap_info *FREE_I(struct f2fs_sb_info *sbi) 2014 { 2015 return (struct free_segmap_info *)(SM_I(sbi)->free_info); 2016 } 2017 2018 static inline struct dirty_seglist_info *DIRTY_I(struct f2fs_sb_info *sbi) 2019 { 2020 return (struct dirty_seglist_info *)(SM_I(sbi)->dirty_info); 2021 } 2022 2023 static inline struct address_space *META_MAPPING(struct f2fs_sb_info *sbi) 2024 { 2025 return sbi->meta_inode->i_mapping; 2026 } 2027 2028 static inline struct address_space *NODE_MAPPING(struct f2fs_sb_info *sbi) 2029 { 2030 return sbi->node_inode->i_mapping; 2031 } 2032 2033 static inline bool is_sbi_flag_set(struct f2fs_sb_info *sbi, unsigned int type) 2034 { 2035 return test_bit(type, &sbi->s_flag); 2036 } 2037 2038 static inline void set_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type) 2039 { 2040 set_bit(type, &sbi->s_flag); 2041 } 2042 2043 static inline void clear_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type) 2044 { 2045 clear_bit(type, &sbi->s_flag); 2046 } 2047 2048 static inline unsigned long long cur_cp_version(struct f2fs_checkpoint *cp) 2049 { 2050 return le64_to_cpu(cp->checkpoint_ver); 2051 } 2052 2053 static inline unsigned long f2fs_qf_ino(struct super_block *sb, int type) 2054 { 2055 if (type < F2FS_MAX_QUOTAS) 2056 return le32_to_cpu(F2FS_SB(sb)->raw_super->qf_ino[type]); 2057 return 0; 2058 } 2059 2060 static inline __u64 cur_cp_crc(struct f2fs_checkpoint *cp) 2061 { 2062 size_t crc_offset = le32_to_cpu(cp->checksum_offset); 2063 return le32_to_cpu(*((__le32 *)((unsigned char *)cp + crc_offset))); 2064 } 2065 2066 static inline bool __is_set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 2067 { 2068 unsigned int ckpt_flags = le32_to_cpu(cp->ckpt_flags); 2069 2070 return ckpt_flags & f; 2071 } 2072 2073 static inline bool is_set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 2074 { 2075 return __is_set_ckpt_flags(F2FS_CKPT(sbi), f); 2076 } 2077 2078 static inline void __set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 2079 { 2080 unsigned int ckpt_flags; 2081 2082 ckpt_flags = le32_to_cpu(cp->ckpt_flags); 2083 ckpt_flags |= f; 2084 cp->ckpt_flags = cpu_to_le32(ckpt_flags); 2085 } 2086 2087 static inline void set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 2088 { 2089 unsigned long flags; 2090 2091 spin_lock_irqsave(&sbi->cp_lock, flags); 2092 __set_ckpt_flags(F2FS_CKPT(sbi), f); 2093 spin_unlock_irqrestore(&sbi->cp_lock, flags); 2094 } 2095 2096 static inline void __clear_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 2097 { 2098 unsigned int ckpt_flags; 2099 2100 ckpt_flags = le32_to_cpu(cp->ckpt_flags); 2101 ckpt_flags &= (~f); 2102 cp->ckpt_flags = cpu_to_le32(ckpt_flags); 2103 } 2104 2105 static inline void clear_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 2106 { 2107 unsigned long flags; 2108 2109 spin_lock_irqsave(&sbi->cp_lock, flags); 2110 __clear_ckpt_flags(F2FS_CKPT(sbi), f); 2111 spin_unlock_irqrestore(&sbi->cp_lock, flags); 2112 } 2113 2114 static inline void init_f2fs_rwsem(struct f2fs_rwsem *sem) 2115 { 2116 init_rwsem(&sem->internal_rwsem); 2117 init_waitqueue_head(&sem->read_waiters); 2118 } 2119 2120 static inline int f2fs_rwsem_is_locked(struct f2fs_rwsem *sem) 2121 { 2122 return rwsem_is_locked(&sem->internal_rwsem); 2123 } 2124 2125 static inline int f2fs_rwsem_is_contended(struct f2fs_rwsem *sem) 2126 { 2127 return rwsem_is_contended(&sem->internal_rwsem); 2128 } 2129 2130 static inline void f2fs_down_read(struct f2fs_rwsem *sem) 2131 { 2132 wait_event(sem->read_waiters, down_read_trylock(&sem->internal_rwsem)); 2133 } 2134 2135 static inline int f2fs_down_read_trylock(struct f2fs_rwsem *sem) 2136 { 2137 return down_read_trylock(&sem->internal_rwsem); 2138 } 2139 2140 #ifdef CONFIG_DEBUG_LOCK_ALLOC 2141 static inline void f2fs_down_read_nested(struct f2fs_rwsem *sem, int subclass) 2142 { 2143 down_read_nested(&sem->internal_rwsem, subclass); 2144 } 2145 #else 2146 #define f2fs_down_read_nested(sem, subclass) f2fs_down_read(sem) 2147 #endif 2148 2149 static inline void f2fs_up_read(struct f2fs_rwsem *sem) 2150 { 2151 up_read(&sem->internal_rwsem); 2152 } 2153 2154 static inline void f2fs_down_write(struct f2fs_rwsem *sem) 2155 { 2156 down_write(&sem->internal_rwsem); 2157 } 2158 2159 static inline int f2fs_down_write_trylock(struct f2fs_rwsem *sem) 2160 { 2161 return down_write_trylock(&sem->internal_rwsem); 2162 } 2163 2164 static inline void f2fs_up_write(struct f2fs_rwsem *sem) 2165 { 2166 up_write(&sem->internal_rwsem); 2167 wake_up_all(&sem->read_waiters); 2168 } 2169 2170 static inline void f2fs_lock_op(struct f2fs_sb_info *sbi) 2171 { 2172 f2fs_down_read(&sbi->cp_rwsem); 2173 } 2174 2175 static inline int f2fs_trylock_op(struct f2fs_sb_info *sbi) 2176 { 2177 if (time_to_inject(sbi, FAULT_LOCK_OP)) { 2178 f2fs_show_injection_info(sbi, FAULT_LOCK_OP); 2179 return 0; 2180 } 2181 return f2fs_down_read_trylock(&sbi->cp_rwsem); 2182 } 2183 2184 static inline void f2fs_unlock_op(struct f2fs_sb_info *sbi) 2185 { 2186 f2fs_up_read(&sbi->cp_rwsem); 2187 } 2188 2189 static inline void f2fs_lock_all(struct f2fs_sb_info *sbi) 2190 { 2191 f2fs_down_write(&sbi->cp_rwsem); 2192 } 2193 2194 static inline void f2fs_unlock_all(struct f2fs_sb_info *sbi) 2195 { 2196 f2fs_up_write(&sbi->cp_rwsem); 2197 } 2198 2199 static inline int __get_cp_reason(struct f2fs_sb_info *sbi) 2200 { 2201 int reason = CP_SYNC; 2202 2203 if (test_opt(sbi, FASTBOOT)) 2204 reason = CP_FASTBOOT; 2205 if (is_sbi_flag_set(sbi, SBI_IS_CLOSE)) 2206 reason = CP_UMOUNT; 2207 return reason; 2208 } 2209 2210 static inline bool __remain_node_summaries(int reason) 2211 { 2212 return (reason & (CP_UMOUNT | CP_FASTBOOT)); 2213 } 2214 2215 static inline bool __exist_node_summaries(struct f2fs_sb_info *sbi) 2216 { 2217 return (is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG) || 2218 is_set_ckpt_flags(sbi, CP_FASTBOOT_FLAG)); 2219 } 2220 2221 /* 2222 * Check whether the inode has blocks or not 2223 */ 2224 static inline int F2FS_HAS_BLOCKS(struct inode *inode) 2225 { 2226 block_t xattr_block = F2FS_I(inode)->i_xattr_nid ? 1 : 0; 2227 2228 return (inode->i_blocks >> F2FS_LOG_SECTORS_PER_BLOCK) > xattr_block; 2229 } 2230 2231 static inline bool f2fs_has_xattr_block(unsigned int ofs) 2232 { 2233 return ofs == XATTR_NODE_OFFSET; 2234 } 2235 2236 static inline bool __allow_reserved_blocks(struct f2fs_sb_info *sbi, 2237 struct inode *inode, bool cap) 2238 { 2239 if (!inode) 2240 return true; 2241 if (!test_opt(sbi, RESERVE_ROOT)) 2242 return false; 2243 if (IS_NOQUOTA(inode)) 2244 return true; 2245 if (uid_eq(F2FS_OPTION(sbi).s_resuid, current_fsuid())) 2246 return true; 2247 if (!gid_eq(F2FS_OPTION(sbi).s_resgid, GLOBAL_ROOT_GID) && 2248 in_group_p(F2FS_OPTION(sbi).s_resgid)) 2249 return true; 2250 if (cap && capable(CAP_SYS_RESOURCE)) 2251 return true; 2252 return false; 2253 } 2254 2255 static inline void f2fs_i_blocks_write(struct inode *, block_t, bool, bool); 2256 static inline int inc_valid_block_count(struct f2fs_sb_info *sbi, 2257 struct inode *inode, blkcnt_t *count) 2258 { 2259 blkcnt_t diff = 0, release = 0; 2260 block_t avail_user_block_count; 2261 int ret; 2262 2263 ret = dquot_reserve_block(inode, *count); 2264 if (ret) 2265 return ret; 2266 2267 if (time_to_inject(sbi, FAULT_BLOCK)) { 2268 f2fs_show_injection_info(sbi, FAULT_BLOCK); 2269 release = *count; 2270 goto release_quota; 2271 } 2272 2273 /* 2274 * let's increase this in prior to actual block count change in order 2275 * for f2fs_sync_file to avoid data races when deciding checkpoint. 2276 */ 2277 percpu_counter_add(&sbi->alloc_valid_block_count, (*count)); 2278 2279 spin_lock(&sbi->stat_lock); 2280 sbi->total_valid_block_count += (block_t)(*count); 2281 avail_user_block_count = sbi->user_block_count - 2282 sbi->current_reserved_blocks; 2283 2284 if (!__allow_reserved_blocks(sbi, inode, true)) 2285 avail_user_block_count -= F2FS_OPTION(sbi).root_reserved_blocks; 2286 2287 if (F2FS_IO_ALIGNED(sbi)) 2288 avail_user_block_count -= sbi->blocks_per_seg * 2289 SM_I(sbi)->additional_reserved_segments; 2290 2291 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) { 2292 if (avail_user_block_count > sbi->unusable_block_count) 2293 avail_user_block_count -= sbi->unusable_block_count; 2294 else 2295 avail_user_block_count = 0; 2296 } 2297 if (unlikely(sbi->total_valid_block_count > avail_user_block_count)) { 2298 diff = sbi->total_valid_block_count - avail_user_block_count; 2299 if (diff > *count) 2300 diff = *count; 2301 *count -= diff; 2302 release = diff; 2303 sbi->total_valid_block_count -= diff; 2304 if (!*count) { 2305 spin_unlock(&sbi->stat_lock); 2306 goto enospc; 2307 } 2308 } 2309 spin_unlock(&sbi->stat_lock); 2310 2311 if (unlikely(release)) { 2312 percpu_counter_sub(&sbi->alloc_valid_block_count, release); 2313 dquot_release_reservation_block(inode, release); 2314 } 2315 f2fs_i_blocks_write(inode, *count, true, true); 2316 return 0; 2317 2318 enospc: 2319 percpu_counter_sub(&sbi->alloc_valid_block_count, release); 2320 release_quota: 2321 dquot_release_reservation_block(inode, release); 2322 return -ENOSPC; 2323 } 2324 2325 __printf(2, 3) 2326 void f2fs_printk(struct f2fs_sb_info *sbi, const char *fmt, ...); 2327 2328 #define f2fs_err(sbi, fmt, ...) \ 2329 f2fs_printk(sbi, KERN_ERR fmt, ##__VA_ARGS__) 2330 #define f2fs_warn(sbi, fmt, ...) \ 2331 f2fs_printk(sbi, KERN_WARNING fmt, ##__VA_ARGS__) 2332 #define f2fs_notice(sbi, fmt, ...) \ 2333 f2fs_printk(sbi, KERN_NOTICE fmt, ##__VA_ARGS__) 2334 #define f2fs_info(sbi, fmt, ...) \ 2335 f2fs_printk(sbi, KERN_INFO fmt, ##__VA_ARGS__) 2336 #define f2fs_debug(sbi, fmt, ...) \ 2337 f2fs_printk(sbi, KERN_DEBUG fmt, ##__VA_ARGS__) 2338 2339 static inline void dec_valid_block_count(struct f2fs_sb_info *sbi, 2340 struct inode *inode, 2341 block_t count) 2342 { 2343 blkcnt_t sectors = count << F2FS_LOG_SECTORS_PER_BLOCK; 2344 2345 spin_lock(&sbi->stat_lock); 2346 f2fs_bug_on(sbi, sbi->total_valid_block_count < (block_t) count); 2347 sbi->total_valid_block_count -= (block_t)count; 2348 if (sbi->reserved_blocks && 2349 sbi->current_reserved_blocks < sbi->reserved_blocks) 2350 sbi->current_reserved_blocks = min(sbi->reserved_blocks, 2351 sbi->current_reserved_blocks + count); 2352 spin_unlock(&sbi->stat_lock); 2353 if (unlikely(inode->i_blocks < sectors)) { 2354 f2fs_warn(sbi, "Inconsistent i_blocks, ino:%lu, iblocks:%llu, sectors:%llu", 2355 inode->i_ino, 2356 (unsigned long long)inode->i_blocks, 2357 (unsigned long long)sectors); 2358 set_sbi_flag(sbi, SBI_NEED_FSCK); 2359 return; 2360 } 2361 f2fs_i_blocks_write(inode, count, false, true); 2362 } 2363 2364 static inline void inc_page_count(struct f2fs_sb_info *sbi, int count_type) 2365 { 2366 atomic_inc(&sbi->nr_pages[count_type]); 2367 2368 if (count_type == F2FS_DIRTY_DENTS || 2369 count_type == F2FS_DIRTY_NODES || 2370 count_type == F2FS_DIRTY_META || 2371 count_type == F2FS_DIRTY_QDATA || 2372 count_type == F2FS_DIRTY_IMETA) 2373 set_sbi_flag(sbi, SBI_IS_DIRTY); 2374 } 2375 2376 static inline void inode_inc_dirty_pages(struct inode *inode) 2377 { 2378 atomic_inc(&F2FS_I(inode)->dirty_pages); 2379 inc_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ? 2380 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA); 2381 if (IS_NOQUOTA(inode)) 2382 inc_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA); 2383 } 2384 2385 static inline void dec_page_count(struct f2fs_sb_info *sbi, int count_type) 2386 { 2387 atomic_dec(&sbi->nr_pages[count_type]); 2388 } 2389 2390 static inline void inode_dec_dirty_pages(struct inode *inode) 2391 { 2392 if (!S_ISDIR(inode->i_mode) && !S_ISREG(inode->i_mode) && 2393 !S_ISLNK(inode->i_mode)) 2394 return; 2395 2396 atomic_dec(&F2FS_I(inode)->dirty_pages); 2397 dec_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ? 2398 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA); 2399 if (IS_NOQUOTA(inode)) 2400 dec_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA); 2401 } 2402 2403 static inline s64 get_pages(struct f2fs_sb_info *sbi, int count_type) 2404 { 2405 return atomic_read(&sbi->nr_pages[count_type]); 2406 } 2407 2408 static inline int get_dirty_pages(struct inode *inode) 2409 { 2410 return atomic_read(&F2FS_I(inode)->dirty_pages); 2411 } 2412 2413 static inline int get_blocktype_secs(struct f2fs_sb_info *sbi, int block_type) 2414 { 2415 unsigned int pages_per_sec = sbi->segs_per_sec * sbi->blocks_per_seg; 2416 unsigned int segs = (get_pages(sbi, block_type) + pages_per_sec - 1) >> 2417 sbi->log_blocks_per_seg; 2418 2419 return segs / sbi->segs_per_sec; 2420 } 2421 2422 static inline block_t valid_user_blocks(struct f2fs_sb_info *sbi) 2423 { 2424 return sbi->total_valid_block_count; 2425 } 2426 2427 static inline block_t discard_blocks(struct f2fs_sb_info *sbi) 2428 { 2429 return sbi->discard_blks; 2430 } 2431 2432 static inline unsigned long __bitmap_size(struct f2fs_sb_info *sbi, int flag) 2433 { 2434 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi); 2435 2436 /* return NAT or SIT bitmap */ 2437 if (flag == NAT_BITMAP) 2438 return le32_to_cpu(ckpt->nat_ver_bitmap_bytesize); 2439 else if (flag == SIT_BITMAP) 2440 return le32_to_cpu(ckpt->sit_ver_bitmap_bytesize); 2441 2442 return 0; 2443 } 2444 2445 static inline block_t __cp_payload(struct f2fs_sb_info *sbi) 2446 { 2447 return le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_payload); 2448 } 2449 2450 static inline void *__bitmap_ptr(struct f2fs_sb_info *sbi, int flag) 2451 { 2452 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi); 2453 void *tmp_ptr = &ckpt->sit_nat_version_bitmap; 2454 int offset; 2455 2456 if (is_set_ckpt_flags(sbi, CP_LARGE_NAT_BITMAP_FLAG)) { 2457 offset = (flag == SIT_BITMAP) ? 2458 le32_to_cpu(ckpt->nat_ver_bitmap_bytesize) : 0; 2459 /* 2460 * if large_nat_bitmap feature is enabled, leave checksum 2461 * protection for all nat/sit bitmaps. 2462 */ 2463 return tmp_ptr + offset + sizeof(__le32); 2464 } 2465 2466 if (__cp_payload(sbi) > 0) { 2467 if (flag == NAT_BITMAP) 2468 return &ckpt->sit_nat_version_bitmap; 2469 else 2470 return (unsigned char *)ckpt + F2FS_BLKSIZE; 2471 } else { 2472 offset = (flag == NAT_BITMAP) ? 2473 le32_to_cpu(ckpt->sit_ver_bitmap_bytesize) : 0; 2474 return tmp_ptr + offset; 2475 } 2476 } 2477 2478 static inline block_t __start_cp_addr(struct f2fs_sb_info *sbi) 2479 { 2480 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr); 2481 2482 if (sbi->cur_cp_pack == 2) 2483 start_addr += sbi->blocks_per_seg; 2484 return start_addr; 2485 } 2486 2487 static inline block_t __start_cp_next_addr(struct f2fs_sb_info *sbi) 2488 { 2489 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr); 2490 2491 if (sbi->cur_cp_pack == 1) 2492 start_addr += sbi->blocks_per_seg; 2493 return start_addr; 2494 } 2495 2496 static inline void __set_cp_next_pack(struct f2fs_sb_info *sbi) 2497 { 2498 sbi->cur_cp_pack = (sbi->cur_cp_pack == 1) ? 2 : 1; 2499 } 2500 2501 static inline block_t __start_sum_addr(struct f2fs_sb_info *sbi) 2502 { 2503 return le32_to_cpu(F2FS_CKPT(sbi)->cp_pack_start_sum); 2504 } 2505 2506 static inline int inc_valid_node_count(struct f2fs_sb_info *sbi, 2507 struct inode *inode, bool is_inode) 2508 { 2509 block_t valid_block_count; 2510 unsigned int valid_node_count, user_block_count; 2511 int err; 2512 2513 if (is_inode) { 2514 if (inode) { 2515 err = dquot_alloc_inode(inode); 2516 if (err) 2517 return err; 2518 } 2519 } else { 2520 err = dquot_reserve_block(inode, 1); 2521 if (err) 2522 return err; 2523 } 2524 2525 if (time_to_inject(sbi, FAULT_BLOCK)) { 2526 f2fs_show_injection_info(sbi, FAULT_BLOCK); 2527 goto enospc; 2528 } 2529 2530 spin_lock(&sbi->stat_lock); 2531 2532 valid_block_count = sbi->total_valid_block_count + 2533 sbi->current_reserved_blocks + 1; 2534 2535 if (!__allow_reserved_blocks(sbi, inode, false)) 2536 valid_block_count += F2FS_OPTION(sbi).root_reserved_blocks; 2537 2538 if (F2FS_IO_ALIGNED(sbi)) 2539 valid_block_count += sbi->blocks_per_seg * 2540 SM_I(sbi)->additional_reserved_segments; 2541 2542 user_block_count = sbi->user_block_count; 2543 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) 2544 user_block_count -= sbi->unusable_block_count; 2545 2546 if (unlikely(valid_block_count > user_block_count)) { 2547 spin_unlock(&sbi->stat_lock); 2548 goto enospc; 2549 } 2550 2551 valid_node_count = sbi->total_valid_node_count + 1; 2552 if (unlikely(valid_node_count > sbi->total_node_count)) { 2553 spin_unlock(&sbi->stat_lock); 2554 goto enospc; 2555 } 2556 2557 sbi->total_valid_node_count++; 2558 sbi->total_valid_block_count++; 2559 spin_unlock(&sbi->stat_lock); 2560 2561 if (inode) { 2562 if (is_inode) 2563 f2fs_mark_inode_dirty_sync(inode, true); 2564 else 2565 f2fs_i_blocks_write(inode, 1, true, true); 2566 } 2567 2568 percpu_counter_inc(&sbi->alloc_valid_block_count); 2569 return 0; 2570 2571 enospc: 2572 if (is_inode) { 2573 if (inode) 2574 dquot_free_inode(inode); 2575 } else { 2576 dquot_release_reservation_block(inode, 1); 2577 } 2578 return -ENOSPC; 2579 } 2580 2581 static inline void dec_valid_node_count(struct f2fs_sb_info *sbi, 2582 struct inode *inode, bool is_inode) 2583 { 2584 spin_lock(&sbi->stat_lock); 2585 2586 f2fs_bug_on(sbi, !sbi->total_valid_block_count); 2587 f2fs_bug_on(sbi, !sbi->total_valid_node_count); 2588 2589 sbi->total_valid_node_count--; 2590 sbi->total_valid_block_count--; 2591 if (sbi->reserved_blocks && 2592 sbi->current_reserved_blocks < sbi->reserved_blocks) 2593 sbi->current_reserved_blocks++; 2594 2595 spin_unlock(&sbi->stat_lock); 2596 2597 if (is_inode) { 2598 dquot_free_inode(inode); 2599 } else { 2600 if (unlikely(inode->i_blocks == 0)) { 2601 f2fs_warn(sbi, "dec_valid_node_count: inconsistent i_blocks, ino:%lu, iblocks:%llu", 2602 inode->i_ino, 2603 (unsigned long long)inode->i_blocks); 2604 set_sbi_flag(sbi, SBI_NEED_FSCK); 2605 return; 2606 } 2607 f2fs_i_blocks_write(inode, 1, false, true); 2608 } 2609 } 2610 2611 static inline unsigned int valid_node_count(struct f2fs_sb_info *sbi) 2612 { 2613 return sbi->total_valid_node_count; 2614 } 2615 2616 static inline void inc_valid_inode_count(struct f2fs_sb_info *sbi) 2617 { 2618 percpu_counter_inc(&sbi->total_valid_inode_count); 2619 } 2620 2621 static inline void dec_valid_inode_count(struct f2fs_sb_info *sbi) 2622 { 2623 percpu_counter_dec(&sbi->total_valid_inode_count); 2624 } 2625 2626 static inline s64 valid_inode_count(struct f2fs_sb_info *sbi) 2627 { 2628 return percpu_counter_sum_positive(&sbi->total_valid_inode_count); 2629 } 2630 2631 static inline struct page *f2fs_grab_cache_page(struct address_space *mapping, 2632 pgoff_t index, bool for_write) 2633 { 2634 struct page *page; 2635 2636 if (IS_ENABLED(CONFIG_F2FS_FAULT_INJECTION)) { 2637 if (!for_write) 2638 page = find_get_page_flags(mapping, index, 2639 FGP_LOCK | FGP_ACCESSED); 2640 else 2641 page = find_lock_page(mapping, index); 2642 if (page) 2643 return page; 2644 2645 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_ALLOC)) { 2646 f2fs_show_injection_info(F2FS_M_SB(mapping), 2647 FAULT_PAGE_ALLOC); 2648 return NULL; 2649 } 2650 } 2651 2652 if (!for_write) 2653 return grab_cache_page(mapping, index); 2654 return grab_cache_page_write_begin(mapping, index, AOP_FLAG_NOFS); 2655 } 2656 2657 static inline struct page *f2fs_pagecache_get_page( 2658 struct address_space *mapping, pgoff_t index, 2659 int fgp_flags, gfp_t gfp_mask) 2660 { 2661 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_GET)) { 2662 f2fs_show_injection_info(F2FS_M_SB(mapping), FAULT_PAGE_GET); 2663 return NULL; 2664 } 2665 2666 return pagecache_get_page(mapping, index, fgp_flags, gfp_mask); 2667 } 2668 2669 static inline void f2fs_copy_page(struct page *src, struct page *dst) 2670 { 2671 char *src_kaddr = kmap(src); 2672 char *dst_kaddr = kmap(dst); 2673 2674 memcpy(dst_kaddr, src_kaddr, PAGE_SIZE); 2675 kunmap(dst); 2676 kunmap(src); 2677 } 2678 2679 static inline void f2fs_put_page(struct page *page, int unlock) 2680 { 2681 if (!page) 2682 return; 2683 2684 if (unlock) { 2685 f2fs_bug_on(F2FS_P_SB(page), !PageLocked(page)); 2686 unlock_page(page); 2687 } 2688 put_page(page); 2689 } 2690 2691 static inline void f2fs_put_dnode(struct dnode_of_data *dn) 2692 { 2693 if (dn->node_page) 2694 f2fs_put_page(dn->node_page, 1); 2695 if (dn->inode_page && dn->node_page != dn->inode_page) 2696 f2fs_put_page(dn->inode_page, 0); 2697 dn->node_page = NULL; 2698 dn->inode_page = NULL; 2699 } 2700 2701 static inline struct kmem_cache *f2fs_kmem_cache_create(const char *name, 2702 size_t size) 2703 { 2704 return kmem_cache_create(name, size, 0, SLAB_RECLAIM_ACCOUNT, NULL); 2705 } 2706 2707 static inline void *f2fs_kmem_cache_alloc_nofail(struct kmem_cache *cachep, 2708 gfp_t flags) 2709 { 2710 void *entry; 2711 2712 entry = kmem_cache_alloc(cachep, flags); 2713 if (!entry) 2714 entry = kmem_cache_alloc(cachep, flags | __GFP_NOFAIL); 2715 return entry; 2716 } 2717 2718 static inline void *f2fs_kmem_cache_alloc(struct kmem_cache *cachep, 2719 gfp_t flags, bool nofail, struct f2fs_sb_info *sbi) 2720 { 2721 if (nofail) 2722 return f2fs_kmem_cache_alloc_nofail(cachep, flags); 2723 2724 if (time_to_inject(sbi, FAULT_SLAB_ALLOC)) { 2725 f2fs_show_injection_info(sbi, FAULT_SLAB_ALLOC); 2726 return NULL; 2727 } 2728 2729 return kmem_cache_alloc(cachep, flags); 2730 } 2731 2732 static inline bool is_inflight_io(struct f2fs_sb_info *sbi, int type) 2733 { 2734 if (get_pages(sbi, F2FS_RD_DATA) || get_pages(sbi, F2FS_RD_NODE) || 2735 get_pages(sbi, F2FS_RD_META) || get_pages(sbi, F2FS_WB_DATA) || 2736 get_pages(sbi, F2FS_WB_CP_DATA) || 2737 get_pages(sbi, F2FS_DIO_READ) || 2738 get_pages(sbi, F2FS_DIO_WRITE)) 2739 return true; 2740 2741 if (type != DISCARD_TIME && SM_I(sbi) && SM_I(sbi)->dcc_info && 2742 atomic_read(&SM_I(sbi)->dcc_info->queued_discard)) 2743 return true; 2744 2745 if (SM_I(sbi) && SM_I(sbi)->fcc_info && 2746 atomic_read(&SM_I(sbi)->fcc_info->queued_flush)) 2747 return true; 2748 return false; 2749 } 2750 2751 static inline bool is_idle(struct f2fs_sb_info *sbi, int type) 2752 { 2753 if (sbi->gc_mode == GC_URGENT_HIGH) 2754 return true; 2755 2756 if (is_inflight_io(sbi, type)) 2757 return false; 2758 2759 if (sbi->gc_mode == GC_URGENT_LOW && 2760 (type == DISCARD_TIME || type == GC_TIME)) 2761 return true; 2762 2763 return f2fs_time_over(sbi, type); 2764 } 2765 2766 static inline void f2fs_radix_tree_insert(struct radix_tree_root *root, 2767 unsigned long index, void *item) 2768 { 2769 while (radix_tree_insert(root, index, item)) 2770 cond_resched(); 2771 } 2772 2773 #define RAW_IS_INODE(p) ((p)->footer.nid == (p)->footer.ino) 2774 2775 static inline bool IS_INODE(struct page *page) 2776 { 2777 struct f2fs_node *p = F2FS_NODE(page); 2778 2779 return RAW_IS_INODE(p); 2780 } 2781 2782 static inline int offset_in_addr(struct f2fs_inode *i) 2783 { 2784 return (i->i_inline & F2FS_EXTRA_ATTR) ? 2785 (le16_to_cpu(i->i_extra_isize) / sizeof(__le32)) : 0; 2786 } 2787 2788 static inline __le32 *blkaddr_in_node(struct f2fs_node *node) 2789 { 2790 return RAW_IS_INODE(node) ? node->i.i_addr : node->dn.addr; 2791 } 2792 2793 static inline int f2fs_has_extra_attr(struct inode *inode); 2794 static inline block_t data_blkaddr(struct inode *inode, 2795 struct page *node_page, unsigned int offset) 2796 { 2797 struct f2fs_node *raw_node; 2798 __le32 *addr_array; 2799 int base = 0; 2800 bool is_inode = IS_INODE(node_page); 2801 2802 raw_node = F2FS_NODE(node_page); 2803 2804 if (is_inode) { 2805 if (!inode) 2806 /* from GC path only */ 2807 base = offset_in_addr(&raw_node->i); 2808 else if (f2fs_has_extra_attr(inode)) 2809 base = get_extra_isize(inode); 2810 } 2811 2812 addr_array = blkaddr_in_node(raw_node); 2813 return le32_to_cpu(addr_array[base + offset]); 2814 } 2815 2816 static inline block_t f2fs_data_blkaddr(struct dnode_of_data *dn) 2817 { 2818 return data_blkaddr(dn->inode, dn->node_page, dn->ofs_in_node); 2819 } 2820 2821 static inline int f2fs_test_bit(unsigned int nr, char *addr) 2822 { 2823 int mask; 2824 2825 addr += (nr >> 3); 2826 mask = 1 << (7 - (nr & 0x07)); 2827 return mask & *addr; 2828 } 2829 2830 static inline void f2fs_set_bit(unsigned int nr, char *addr) 2831 { 2832 int mask; 2833 2834 addr += (nr >> 3); 2835 mask = 1 << (7 - (nr & 0x07)); 2836 *addr |= mask; 2837 } 2838 2839 static inline void f2fs_clear_bit(unsigned int nr, char *addr) 2840 { 2841 int mask; 2842 2843 addr += (nr >> 3); 2844 mask = 1 << (7 - (nr & 0x07)); 2845 *addr &= ~mask; 2846 } 2847 2848 static inline int f2fs_test_and_set_bit(unsigned int nr, char *addr) 2849 { 2850 int mask; 2851 int ret; 2852 2853 addr += (nr >> 3); 2854 mask = 1 << (7 - (nr & 0x07)); 2855 ret = mask & *addr; 2856 *addr |= mask; 2857 return ret; 2858 } 2859 2860 static inline int f2fs_test_and_clear_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 void f2fs_change_bit(unsigned int nr, char *addr) 2873 { 2874 int mask; 2875 2876 addr += (nr >> 3); 2877 mask = 1 << (7 - (nr & 0x07)); 2878 *addr ^= mask; 2879 } 2880 2881 /* 2882 * On-disk inode flags (f2fs_inode::i_flags) 2883 */ 2884 #define F2FS_COMPR_FL 0x00000004 /* Compress file */ 2885 #define F2FS_SYNC_FL 0x00000008 /* Synchronous updates */ 2886 #define F2FS_IMMUTABLE_FL 0x00000010 /* Immutable file */ 2887 #define F2FS_APPEND_FL 0x00000020 /* writes to file may only append */ 2888 #define F2FS_NODUMP_FL 0x00000040 /* do not dump file */ 2889 #define F2FS_NOATIME_FL 0x00000080 /* do not update atime */ 2890 #define F2FS_NOCOMP_FL 0x00000400 /* Don't compress */ 2891 #define F2FS_INDEX_FL 0x00001000 /* hash-indexed directory */ 2892 #define F2FS_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */ 2893 #define F2FS_PROJINHERIT_FL 0x20000000 /* Create with parents projid */ 2894 #define F2FS_CASEFOLD_FL 0x40000000 /* Casefolded file */ 2895 2896 /* Flags that should be inherited by new inodes from their parent. */ 2897 #define F2FS_FL_INHERITED (F2FS_SYNC_FL | F2FS_NODUMP_FL | F2FS_NOATIME_FL | \ 2898 F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \ 2899 F2FS_CASEFOLD_FL | F2FS_COMPR_FL | F2FS_NOCOMP_FL) 2900 2901 /* Flags that are appropriate for regular files (all but dir-specific ones). */ 2902 #define F2FS_REG_FLMASK (~(F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \ 2903 F2FS_CASEFOLD_FL)) 2904 2905 /* Flags that are appropriate for non-directories/regular files. */ 2906 #define F2FS_OTHER_FLMASK (F2FS_NODUMP_FL | F2FS_NOATIME_FL) 2907 2908 static inline __u32 f2fs_mask_flags(umode_t mode, __u32 flags) 2909 { 2910 if (S_ISDIR(mode)) 2911 return flags; 2912 else if (S_ISREG(mode)) 2913 return flags & F2FS_REG_FLMASK; 2914 else 2915 return flags & F2FS_OTHER_FLMASK; 2916 } 2917 2918 static inline void __mark_inode_dirty_flag(struct inode *inode, 2919 int flag, bool set) 2920 { 2921 switch (flag) { 2922 case FI_INLINE_XATTR: 2923 case FI_INLINE_DATA: 2924 case FI_INLINE_DENTRY: 2925 case FI_NEW_INODE: 2926 if (set) 2927 return; 2928 fallthrough; 2929 case FI_DATA_EXIST: 2930 case FI_INLINE_DOTS: 2931 case FI_PIN_FILE: 2932 case FI_COMPRESS_RELEASED: 2933 f2fs_mark_inode_dirty_sync(inode, true); 2934 } 2935 } 2936 2937 static inline void set_inode_flag(struct inode *inode, int flag) 2938 { 2939 set_bit(flag, F2FS_I(inode)->flags); 2940 __mark_inode_dirty_flag(inode, flag, true); 2941 } 2942 2943 static inline int is_inode_flag_set(struct inode *inode, int flag) 2944 { 2945 return test_bit(flag, F2FS_I(inode)->flags); 2946 } 2947 2948 static inline void clear_inode_flag(struct inode *inode, int flag) 2949 { 2950 clear_bit(flag, F2FS_I(inode)->flags); 2951 __mark_inode_dirty_flag(inode, flag, false); 2952 } 2953 2954 static inline bool f2fs_verity_in_progress(struct inode *inode) 2955 { 2956 return IS_ENABLED(CONFIG_FS_VERITY) && 2957 is_inode_flag_set(inode, FI_VERITY_IN_PROGRESS); 2958 } 2959 2960 static inline void set_acl_inode(struct inode *inode, umode_t mode) 2961 { 2962 F2FS_I(inode)->i_acl_mode = mode; 2963 set_inode_flag(inode, FI_ACL_MODE); 2964 f2fs_mark_inode_dirty_sync(inode, false); 2965 } 2966 2967 static inline void f2fs_i_links_write(struct inode *inode, bool inc) 2968 { 2969 if (inc) 2970 inc_nlink(inode); 2971 else 2972 drop_nlink(inode); 2973 f2fs_mark_inode_dirty_sync(inode, true); 2974 } 2975 2976 static inline void f2fs_i_blocks_write(struct inode *inode, 2977 block_t diff, bool add, bool claim) 2978 { 2979 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE); 2980 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER); 2981 2982 /* add = 1, claim = 1 should be dquot_reserve_block in pair */ 2983 if (add) { 2984 if (claim) 2985 dquot_claim_block(inode, diff); 2986 else 2987 dquot_alloc_block_nofail(inode, diff); 2988 } else { 2989 dquot_free_block(inode, diff); 2990 } 2991 2992 f2fs_mark_inode_dirty_sync(inode, true); 2993 if (clean || recover) 2994 set_inode_flag(inode, FI_AUTO_RECOVER); 2995 } 2996 2997 static inline void f2fs_i_size_write(struct inode *inode, loff_t i_size) 2998 { 2999 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE); 3000 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER); 3001 3002 if (i_size_read(inode) == i_size) 3003 return; 3004 3005 i_size_write(inode, i_size); 3006 f2fs_mark_inode_dirty_sync(inode, true); 3007 if (clean || recover) 3008 set_inode_flag(inode, FI_AUTO_RECOVER); 3009 } 3010 3011 static inline void f2fs_i_depth_write(struct inode *inode, unsigned int depth) 3012 { 3013 F2FS_I(inode)->i_current_depth = depth; 3014 f2fs_mark_inode_dirty_sync(inode, true); 3015 } 3016 3017 static inline void f2fs_i_gc_failures_write(struct inode *inode, 3018 unsigned int count) 3019 { 3020 F2FS_I(inode)->i_gc_failures[GC_FAILURE_PIN] = count; 3021 f2fs_mark_inode_dirty_sync(inode, true); 3022 } 3023 3024 static inline void f2fs_i_xnid_write(struct inode *inode, nid_t xnid) 3025 { 3026 F2FS_I(inode)->i_xattr_nid = xnid; 3027 f2fs_mark_inode_dirty_sync(inode, true); 3028 } 3029 3030 static inline void f2fs_i_pino_write(struct inode *inode, nid_t pino) 3031 { 3032 F2FS_I(inode)->i_pino = pino; 3033 f2fs_mark_inode_dirty_sync(inode, true); 3034 } 3035 3036 static inline void get_inline_info(struct inode *inode, struct f2fs_inode *ri) 3037 { 3038 struct f2fs_inode_info *fi = F2FS_I(inode); 3039 3040 if (ri->i_inline & F2FS_INLINE_XATTR) 3041 set_bit(FI_INLINE_XATTR, fi->flags); 3042 if (ri->i_inline & F2FS_INLINE_DATA) 3043 set_bit(FI_INLINE_DATA, fi->flags); 3044 if (ri->i_inline & F2FS_INLINE_DENTRY) 3045 set_bit(FI_INLINE_DENTRY, fi->flags); 3046 if (ri->i_inline & F2FS_DATA_EXIST) 3047 set_bit(FI_DATA_EXIST, fi->flags); 3048 if (ri->i_inline & F2FS_INLINE_DOTS) 3049 set_bit(FI_INLINE_DOTS, fi->flags); 3050 if (ri->i_inline & F2FS_EXTRA_ATTR) 3051 set_bit(FI_EXTRA_ATTR, fi->flags); 3052 if (ri->i_inline & F2FS_PIN_FILE) 3053 set_bit(FI_PIN_FILE, fi->flags); 3054 if (ri->i_inline & F2FS_COMPRESS_RELEASED) 3055 set_bit(FI_COMPRESS_RELEASED, fi->flags); 3056 } 3057 3058 static inline void set_raw_inline(struct inode *inode, struct f2fs_inode *ri) 3059 { 3060 ri->i_inline = 0; 3061 3062 if (is_inode_flag_set(inode, FI_INLINE_XATTR)) 3063 ri->i_inline |= F2FS_INLINE_XATTR; 3064 if (is_inode_flag_set(inode, FI_INLINE_DATA)) 3065 ri->i_inline |= F2FS_INLINE_DATA; 3066 if (is_inode_flag_set(inode, FI_INLINE_DENTRY)) 3067 ri->i_inline |= F2FS_INLINE_DENTRY; 3068 if (is_inode_flag_set(inode, FI_DATA_EXIST)) 3069 ri->i_inline |= F2FS_DATA_EXIST; 3070 if (is_inode_flag_set(inode, FI_INLINE_DOTS)) 3071 ri->i_inline |= F2FS_INLINE_DOTS; 3072 if (is_inode_flag_set(inode, FI_EXTRA_ATTR)) 3073 ri->i_inline |= F2FS_EXTRA_ATTR; 3074 if (is_inode_flag_set(inode, FI_PIN_FILE)) 3075 ri->i_inline |= F2FS_PIN_FILE; 3076 if (is_inode_flag_set(inode, FI_COMPRESS_RELEASED)) 3077 ri->i_inline |= F2FS_COMPRESS_RELEASED; 3078 } 3079 3080 static inline int f2fs_has_extra_attr(struct inode *inode) 3081 { 3082 return is_inode_flag_set(inode, FI_EXTRA_ATTR); 3083 } 3084 3085 static inline int f2fs_has_inline_xattr(struct inode *inode) 3086 { 3087 return is_inode_flag_set(inode, FI_INLINE_XATTR); 3088 } 3089 3090 static inline int f2fs_compressed_file(struct inode *inode) 3091 { 3092 return S_ISREG(inode->i_mode) && 3093 is_inode_flag_set(inode, FI_COMPRESSED_FILE); 3094 } 3095 3096 static inline bool f2fs_need_compress_data(struct inode *inode) 3097 { 3098 int compress_mode = F2FS_OPTION(F2FS_I_SB(inode)).compress_mode; 3099 3100 if (!f2fs_compressed_file(inode)) 3101 return false; 3102 3103 if (compress_mode == COMPR_MODE_FS) 3104 return true; 3105 else if (compress_mode == COMPR_MODE_USER && 3106 is_inode_flag_set(inode, FI_ENABLE_COMPRESS)) 3107 return true; 3108 3109 return false; 3110 } 3111 3112 static inline unsigned int addrs_per_inode(struct inode *inode) 3113 { 3114 unsigned int addrs = CUR_ADDRS_PER_INODE(inode) - 3115 get_inline_xattr_addrs(inode); 3116 3117 if (!f2fs_compressed_file(inode)) 3118 return addrs; 3119 return ALIGN_DOWN(addrs, F2FS_I(inode)->i_cluster_size); 3120 } 3121 3122 static inline unsigned int addrs_per_block(struct inode *inode) 3123 { 3124 if (!f2fs_compressed_file(inode)) 3125 return DEF_ADDRS_PER_BLOCK; 3126 return ALIGN_DOWN(DEF_ADDRS_PER_BLOCK, F2FS_I(inode)->i_cluster_size); 3127 } 3128 3129 static inline void *inline_xattr_addr(struct inode *inode, struct page *page) 3130 { 3131 struct f2fs_inode *ri = F2FS_INODE(page); 3132 3133 return (void *)&(ri->i_addr[DEF_ADDRS_PER_INODE - 3134 get_inline_xattr_addrs(inode)]); 3135 } 3136 3137 static inline int inline_xattr_size(struct inode *inode) 3138 { 3139 if (f2fs_has_inline_xattr(inode)) 3140 return get_inline_xattr_addrs(inode) * sizeof(__le32); 3141 return 0; 3142 } 3143 3144 static inline int f2fs_has_inline_data(struct inode *inode) 3145 { 3146 return is_inode_flag_set(inode, FI_INLINE_DATA); 3147 } 3148 3149 static inline int f2fs_exist_data(struct inode *inode) 3150 { 3151 return is_inode_flag_set(inode, FI_DATA_EXIST); 3152 } 3153 3154 static inline int f2fs_has_inline_dots(struct inode *inode) 3155 { 3156 return is_inode_flag_set(inode, FI_INLINE_DOTS); 3157 } 3158 3159 static inline int f2fs_is_mmap_file(struct inode *inode) 3160 { 3161 return is_inode_flag_set(inode, FI_MMAP_FILE); 3162 } 3163 3164 static inline bool f2fs_is_pinned_file(struct inode *inode) 3165 { 3166 return is_inode_flag_set(inode, FI_PIN_FILE); 3167 } 3168 3169 static inline bool f2fs_is_atomic_file(struct inode *inode) 3170 { 3171 return is_inode_flag_set(inode, FI_ATOMIC_FILE); 3172 } 3173 3174 static inline bool f2fs_is_commit_atomic_write(struct inode *inode) 3175 { 3176 return is_inode_flag_set(inode, FI_ATOMIC_COMMIT); 3177 } 3178 3179 static inline bool f2fs_is_volatile_file(struct inode *inode) 3180 { 3181 return is_inode_flag_set(inode, FI_VOLATILE_FILE); 3182 } 3183 3184 static inline bool f2fs_is_first_block_written(struct inode *inode) 3185 { 3186 return is_inode_flag_set(inode, FI_FIRST_BLOCK_WRITTEN); 3187 } 3188 3189 static inline bool f2fs_is_drop_cache(struct inode *inode) 3190 { 3191 return is_inode_flag_set(inode, FI_DROP_CACHE); 3192 } 3193 3194 static inline void *inline_data_addr(struct inode *inode, struct page *page) 3195 { 3196 struct f2fs_inode *ri = F2FS_INODE(page); 3197 int extra_size = get_extra_isize(inode); 3198 3199 return (void *)&(ri->i_addr[extra_size + DEF_INLINE_RESERVED_SIZE]); 3200 } 3201 3202 static inline int f2fs_has_inline_dentry(struct inode *inode) 3203 { 3204 return is_inode_flag_set(inode, FI_INLINE_DENTRY); 3205 } 3206 3207 static inline int is_file(struct inode *inode, int type) 3208 { 3209 return F2FS_I(inode)->i_advise & type; 3210 } 3211 3212 static inline void set_file(struct inode *inode, int type) 3213 { 3214 if (is_file(inode, type)) 3215 return; 3216 F2FS_I(inode)->i_advise |= type; 3217 f2fs_mark_inode_dirty_sync(inode, true); 3218 } 3219 3220 static inline void clear_file(struct inode *inode, int type) 3221 { 3222 if (!is_file(inode, type)) 3223 return; 3224 F2FS_I(inode)->i_advise &= ~type; 3225 f2fs_mark_inode_dirty_sync(inode, true); 3226 } 3227 3228 static inline bool f2fs_is_time_consistent(struct inode *inode) 3229 { 3230 if (!timespec64_equal(F2FS_I(inode)->i_disk_time, &inode->i_atime)) 3231 return false; 3232 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 1, &inode->i_ctime)) 3233 return false; 3234 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 2, &inode->i_mtime)) 3235 return false; 3236 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 3, 3237 &F2FS_I(inode)->i_crtime)) 3238 return false; 3239 return true; 3240 } 3241 3242 static inline bool f2fs_skip_inode_update(struct inode *inode, int dsync) 3243 { 3244 bool ret; 3245 3246 if (dsync) { 3247 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 3248 3249 spin_lock(&sbi->inode_lock[DIRTY_META]); 3250 ret = list_empty(&F2FS_I(inode)->gdirty_list); 3251 spin_unlock(&sbi->inode_lock[DIRTY_META]); 3252 return ret; 3253 } 3254 if (!is_inode_flag_set(inode, FI_AUTO_RECOVER) || 3255 file_keep_isize(inode) || 3256 i_size_read(inode) & ~PAGE_MASK) 3257 return false; 3258 3259 if (!f2fs_is_time_consistent(inode)) 3260 return false; 3261 3262 spin_lock(&F2FS_I(inode)->i_size_lock); 3263 ret = F2FS_I(inode)->last_disk_size == i_size_read(inode); 3264 spin_unlock(&F2FS_I(inode)->i_size_lock); 3265 3266 return ret; 3267 } 3268 3269 static inline bool f2fs_readonly(struct super_block *sb) 3270 { 3271 return sb_rdonly(sb); 3272 } 3273 3274 static inline bool f2fs_cp_error(struct f2fs_sb_info *sbi) 3275 { 3276 return is_set_ckpt_flags(sbi, CP_ERROR_FLAG); 3277 } 3278 3279 static inline bool is_dot_dotdot(const u8 *name, size_t len) 3280 { 3281 if (len == 1 && name[0] == '.') 3282 return true; 3283 3284 if (len == 2 && name[0] == '.' && name[1] == '.') 3285 return true; 3286 3287 return false; 3288 } 3289 3290 static inline void *f2fs_kmalloc(struct f2fs_sb_info *sbi, 3291 size_t size, gfp_t flags) 3292 { 3293 if (time_to_inject(sbi, FAULT_KMALLOC)) { 3294 f2fs_show_injection_info(sbi, FAULT_KMALLOC); 3295 return NULL; 3296 } 3297 3298 return kmalloc(size, flags); 3299 } 3300 3301 static inline void *f2fs_kzalloc(struct f2fs_sb_info *sbi, 3302 size_t size, gfp_t flags) 3303 { 3304 return f2fs_kmalloc(sbi, size, flags | __GFP_ZERO); 3305 } 3306 3307 static inline void *f2fs_kvmalloc(struct f2fs_sb_info *sbi, 3308 size_t size, gfp_t flags) 3309 { 3310 if (time_to_inject(sbi, FAULT_KVMALLOC)) { 3311 f2fs_show_injection_info(sbi, FAULT_KVMALLOC); 3312 return NULL; 3313 } 3314 3315 return kvmalloc(size, flags); 3316 } 3317 3318 static inline void *f2fs_kvzalloc(struct f2fs_sb_info *sbi, 3319 size_t size, gfp_t flags) 3320 { 3321 return f2fs_kvmalloc(sbi, size, flags | __GFP_ZERO); 3322 } 3323 3324 static inline int get_extra_isize(struct inode *inode) 3325 { 3326 return F2FS_I(inode)->i_extra_isize / sizeof(__le32); 3327 } 3328 3329 static inline int get_inline_xattr_addrs(struct inode *inode) 3330 { 3331 return F2FS_I(inode)->i_inline_xattr_size; 3332 } 3333 3334 #define f2fs_get_inode_mode(i) \ 3335 ((is_inode_flag_set(i, FI_ACL_MODE)) ? \ 3336 (F2FS_I(i)->i_acl_mode) : ((i)->i_mode)) 3337 3338 #define F2FS_TOTAL_EXTRA_ATTR_SIZE \ 3339 (offsetof(struct f2fs_inode, i_extra_end) - \ 3340 offsetof(struct f2fs_inode, i_extra_isize)) \ 3341 3342 #define F2FS_OLD_ATTRIBUTE_SIZE (offsetof(struct f2fs_inode, i_addr)) 3343 #define F2FS_FITS_IN_INODE(f2fs_inode, extra_isize, field) \ 3344 ((offsetof(typeof(*(f2fs_inode)), field) + \ 3345 sizeof((f2fs_inode)->field)) \ 3346 <= (F2FS_OLD_ATTRIBUTE_SIZE + (extra_isize))) \ 3347 3348 #define __is_large_section(sbi) ((sbi)->segs_per_sec > 1) 3349 3350 #define __is_meta_io(fio) (PAGE_TYPE_OF_BIO((fio)->type) == META) 3351 3352 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi, 3353 block_t blkaddr, int type); 3354 static inline void verify_blkaddr(struct f2fs_sb_info *sbi, 3355 block_t blkaddr, int type) 3356 { 3357 if (!f2fs_is_valid_blkaddr(sbi, blkaddr, type)) { 3358 f2fs_err(sbi, "invalid blkaddr: %u, type: %d, run fsck to fix.", 3359 blkaddr, type); 3360 f2fs_bug_on(sbi, 1); 3361 } 3362 } 3363 3364 static inline bool __is_valid_data_blkaddr(block_t blkaddr) 3365 { 3366 if (blkaddr == NEW_ADDR || blkaddr == NULL_ADDR || 3367 blkaddr == COMPRESS_ADDR) 3368 return false; 3369 return true; 3370 } 3371 3372 /* 3373 * file.c 3374 */ 3375 int f2fs_sync_file(struct file *file, loff_t start, loff_t end, int datasync); 3376 void f2fs_truncate_data_blocks(struct dnode_of_data *dn); 3377 int f2fs_do_truncate_blocks(struct inode *inode, u64 from, bool lock); 3378 int f2fs_truncate_blocks(struct inode *inode, u64 from, bool lock); 3379 int f2fs_truncate(struct inode *inode); 3380 int f2fs_getattr(struct user_namespace *mnt_userns, const struct path *path, 3381 struct kstat *stat, u32 request_mask, unsigned int flags); 3382 int f2fs_setattr(struct user_namespace *mnt_userns, struct dentry *dentry, 3383 struct iattr *attr); 3384 int f2fs_truncate_hole(struct inode *inode, pgoff_t pg_start, pgoff_t pg_end); 3385 void f2fs_truncate_data_blocks_range(struct dnode_of_data *dn, int count); 3386 int f2fs_precache_extents(struct inode *inode); 3387 int f2fs_fileattr_get(struct dentry *dentry, struct fileattr *fa); 3388 int f2fs_fileattr_set(struct user_namespace *mnt_userns, 3389 struct dentry *dentry, struct fileattr *fa); 3390 long f2fs_ioctl(struct file *filp, unsigned int cmd, unsigned long arg); 3391 long f2fs_compat_ioctl(struct file *file, unsigned int cmd, unsigned long arg); 3392 int f2fs_transfer_project_quota(struct inode *inode, kprojid_t kprojid); 3393 int f2fs_pin_file_control(struct inode *inode, bool inc); 3394 3395 /* 3396 * inode.c 3397 */ 3398 void f2fs_set_inode_flags(struct inode *inode); 3399 bool f2fs_inode_chksum_verify(struct f2fs_sb_info *sbi, struct page *page); 3400 void f2fs_inode_chksum_set(struct f2fs_sb_info *sbi, struct page *page); 3401 struct inode *f2fs_iget(struct super_block *sb, unsigned long ino); 3402 struct inode *f2fs_iget_retry(struct super_block *sb, unsigned long ino); 3403 int f2fs_try_to_free_nats(struct f2fs_sb_info *sbi, int nr_shrink); 3404 void f2fs_update_inode(struct inode *inode, struct page *node_page); 3405 void f2fs_update_inode_page(struct inode *inode); 3406 int f2fs_write_inode(struct inode *inode, struct writeback_control *wbc); 3407 void f2fs_evict_inode(struct inode *inode); 3408 void f2fs_handle_failed_inode(struct inode *inode); 3409 3410 /* 3411 * namei.c 3412 */ 3413 int f2fs_update_extension_list(struct f2fs_sb_info *sbi, const char *name, 3414 bool hot, bool set); 3415 struct dentry *f2fs_get_parent(struct dentry *child); 3416 3417 /* 3418 * dir.c 3419 */ 3420 unsigned char f2fs_get_de_type(struct f2fs_dir_entry *de); 3421 int f2fs_init_casefolded_name(const struct inode *dir, 3422 struct f2fs_filename *fname); 3423 int f2fs_setup_filename(struct inode *dir, const struct qstr *iname, 3424 int lookup, struct f2fs_filename *fname); 3425 int f2fs_prepare_lookup(struct inode *dir, struct dentry *dentry, 3426 struct f2fs_filename *fname); 3427 void f2fs_free_filename(struct f2fs_filename *fname); 3428 struct f2fs_dir_entry *f2fs_find_target_dentry(const struct f2fs_dentry_ptr *d, 3429 const struct f2fs_filename *fname, int *max_slots); 3430 int f2fs_fill_dentries(struct dir_context *ctx, struct f2fs_dentry_ptr *d, 3431 unsigned int start_pos, struct fscrypt_str *fstr); 3432 void f2fs_do_make_empty_dir(struct inode *inode, struct inode *parent, 3433 struct f2fs_dentry_ptr *d); 3434 struct page *f2fs_init_inode_metadata(struct inode *inode, struct inode *dir, 3435 const struct f2fs_filename *fname, struct page *dpage); 3436 void f2fs_update_parent_metadata(struct inode *dir, struct inode *inode, 3437 unsigned int current_depth); 3438 int f2fs_room_for_filename(const void *bitmap, int slots, int max_slots); 3439 void f2fs_drop_nlink(struct inode *dir, struct inode *inode); 3440 struct f2fs_dir_entry *__f2fs_find_entry(struct inode *dir, 3441 const struct f2fs_filename *fname, 3442 struct page **res_page); 3443 struct f2fs_dir_entry *f2fs_find_entry(struct inode *dir, 3444 const struct qstr *child, struct page **res_page); 3445 struct f2fs_dir_entry *f2fs_parent_dir(struct inode *dir, struct page **p); 3446 ino_t f2fs_inode_by_name(struct inode *dir, const struct qstr *qstr, 3447 struct page **page); 3448 void f2fs_set_link(struct inode *dir, struct f2fs_dir_entry *de, 3449 struct page *page, struct inode *inode); 3450 bool f2fs_has_enough_room(struct inode *dir, struct page *ipage, 3451 const struct f2fs_filename *fname); 3452 void f2fs_update_dentry(nid_t ino, umode_t mode, struct f2fs_dentry_ptr *d, 3453 const struct fscrypt_str *name, f2fs_hash_t name_hash, 3454 unsigned int bit_pos); 3455 int f2fs_add_regular_entry(struct inode *dir, const struct f2fs_filename *fname, 3456 struct inode *inode, nid_t ino, umode_t mode); 3457 int f2fs_add_dentry(struct inode *dir, const struct f2fs_filename *fname, 3458 struct inode *inode, nid_t ino, umode_t mode); 3459 int f2fs_do_add_link(struct inode *dir, const struct qstr *name, 3460 struct inode *inode, nid_t ino, umode_t mode); 3461 void f2fs_delete_entry(struct f2fs_dir_entry *dentry, struct page *page, 3462 struct inode *dir, struct inode *inode); 3463 int f2fs_do_tmpfile(struct inode *inode, struct inode *dir); 3464 bool f2fs_empty_dir(struct inode *dir); 3465 3466 static inline int f2fs_add_link(struct dentry *dentry, struct inode *inode) 3467 { 3468 if (fscrypt_is_nokey_name(dentry)) 3469 return -ENOKEY; 3470 return f2fs_do_add_link(d_inode(dentry->d_parent), &dentry->d_name, 3471 inode, inode->i_ino, inode->i_mode); 3472 } 3473 3474 /* 3475 * super.c 3476 */ 3477 int f2fs_inode_dirtied(struct inode *inode, bool sync); 3478 void f2fs_inode_synced(struct inode *inode); 3479 int f2fs_dquot_initialize(struct inode *inode); 3480 int f2fs_enable_quota_files(struct f2fs_sb_info *sbi, bool rdonly); 3481 int f2fs_quota_sync(struct super_block *sb, int type); 3482 loff_t max_file_blocks(struct inode *inode); 3483 void f2fs_quota_off_umount(struct super_block *sb); 3484 int f2fs_commit_super(struct f2fs_sb_info *sbi, bool recover); 3485 int f2fs_sync_fs(struct super_block *sb, int sync); 3486 int f2fs_sanity_check_ckpt(struct f2fs_sb_info *sbi); 3487 3488 /* 3489 * hash.c 3490 */ 3491 void f2fs_hash_filename(const struct inode *dir, struct f2fs_filename *fname); 3492 3493 /* 3494 * node.c 3495 */ 3496 struct node_info; 3497 3498 int f2fs_check_nid_range(struct f2fs_sb_info *sbi, nid_t nid); 3499 bool f2fs_available_free_memory(struct f2fs_sb_info *sbi, int type); 3500 bool f2fs_in_warm_node_list(struct f2fs_sb_info *sbi, struct page *page); 3501 void f2fs_init_fsync_node_info(struct f2fs_sb_info *sbi); 3502 void f2fs_del_fsync_node_entry(struct f2fs_sb_info *sbi, struct page *page); 3503 void f2fs_reset_fsync_node_info(struct f2fs_sb_info *sbi); 3504 int f2fs_need_dentry_mark(struct f2fs_sb_info *sbi, nid_t nid); 3505 bool f2fs_is_checkpointed_node(struct f2fs_sb_info *sbi, nid_t nid); 3506 bool f2fs_need_inode_block_update(struct f2fs_sb_info *sbi, nid_t ino); 3507 int f2fs_get_node_info(struct f2fs_sb_info *sbi, nid_t nid, 3508 struct node_info *ni, bool checkpoint_context); 3509 pgoff_t f2fs_get_next_page_offset(struct dnode_of_data *dn, pgoff_t pgofs); 3510 int f2fs_get_dnode_of_data(struct dnode_of_data *dn, pgoff_t index, int mode); 3511 int f2fs_truncate_inode_blocks(struct inode *inode, pgoff_t from); 3512 int f2fs_truncate_xattr_node(struct inode *inode); 3513 int f2fs_wait_on_node_pages_writeback(struct f2fs_sb_info *sbi, 3514 unsigned int seq_id); 3515 bool f2fs_nat_bitmap_enabled(struct f2fs_sb_info *sbi); 3516 int f2fs_remove_inode_page(struct inode *inode); 3517 struct page *f2fs_new_inode_page(struct inode *inode); 3518 struct page *f2fs_new_node_page(struct dnode_of_data *dn, unsigned int ofs); 3519 void f2fs_ra_node_page(struct f2fs_sb_info *sbi, nid_t nid); 3520 struct page *f2fs_get_node_page(struct f2fs_sb_info *sbi, pgoff_t nid); 3521 struct page *f2fs_get_node_page_ra(struct page *parent, int start); 3522 int f2fs_move_node_page(struct page *node_page, int gc_type); 3523 void f2fs_flush_inline_data(struct f2fs_sb_info *sbi); 3524 int f2fs_fsync_node_pages(struct f2fs_sb_info *sbi, struct inode *inode, 3525 struct writeback_control *wbc, bool atomic, 3526 unsigned int *seq_id); 3527 int f2fs_sync_node_pages(struct f2fs_sb_info *sbi, 3528 struct writeback_control *wbc, 3529 bool do_balance, enum iostat_type io_type); 3530 int f2fs_build_free_nids(struct f2fs_sb_info *sbi, bool sync, bool mount); 3531 bool f2fs_alloc_nid(struct f2fs_sb_info *sbi, nid_t *nid); 3532 void f2fs_alloc_nid_done(struct f2fs_sb_info *sbi, nid_t nid); 3533 void f2fs_alloc_nid_failed(struct f2fs_sb_info *sbi, nid_t nid); 3534 int f2fs_try_to_free_nids(struct f2fs_sb_info *sbi, int nr_shrink); 3535 int f2fs_recover_inline_xattr(struct inode *inode, struct page *page); 3536 int f2fs_recover_xattr_data(struct inode *inode, struct page *page); 3537 int f2fs_recover_inode_page(struct f2fs_sb_info *sbi, struct page *page); 3538 int f2fs_restore_node_summary(struct f2fs_sb_info *sbi, 3539 unsigned int segno, struct f2fs_summary_block *sum); 3540 void f2fs_enable_nat_bits(struct f2fs_sb_info *sbi); 3541 int f2fs_flush_nat_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3542 int f2fs_build_node_manager(struct f2fs_sb_info *sbi); 3543 void f2fs_destroy_node_manager(struct f2fs_sb_info *sbi); 3544 int __init f2fs_create_node_manager_caches(void); 3545 void f2fs_destroy_node_manager_caches(void); 3546 3547 /* 3548 * segment.c 3549 */ 3550 bool f2fs_need_SSR(struct f2fs_sb_info *sbi); 3551 void f2fs_register_inmem_page(struct inode *inode, struct page *page); 3552 void f2fs_drop_inmem_pages_all(struct f2fs_sb_info *sbi, bool gc_failure); 3553 void f2fs_drop_inmem_pages(struct inode *inode); 3554 void f2fs_drop_inmem_page(struct inode *inode, struct page *page); 3555 int f2fs_commit_inmem_pages(struct inode *inode); 3556 void f2fs_balance_fs(struct f2fs_sb_info *sbi, bool need); 3557 void f2fs_balance_fs_bg(struct f2fs_sb_info *sbi, bool from_bg); 3558 int f2fs_issue_flush(struct f2fs_sb_info *sbi, nid_t ino); 3559 int f2fs_create_flush_cmd_control(struct f2fs_sb_info *sbi); 3560 int f2fs_flush_device_cache(struct f2fs_sb_info *sbi); 3561 void f2fs_destroy_flush_cmd_control(struct f2fs_sb_info *sbi, bool free); 3562 void f2fs_invalidate_blocks(struct f2fs_sb_info *sbi, block_t addr); 3563 bool f2fs_is_checkpointed_data(struct f2fs_sb_info *sbi, block_t blkaddr); 3564 int f2fs_start_discard_thread(struct f2fs_sb_info *sbi); 3565 void f2fs_drop_discard_cmd(struct f2fs_sb_info *sbi); 3566 void f2fs_stop_discard_thread(struct f2fs_sb_info *sbi); 3567 bool f2fs_issue_discard_timeout(struct f2fs_sb_info *sbi); 3568 void f2fs_clear_prefree_segments(struct f2fs_sb_info *sbi, 3569 struct cp_control *cpc); 3570 void f2fs_dirty_to_prefree(struct f2fs_sb_info *sbi); 3571 block_t f2fs_get_unusable_blocks(struct f2fs_sb_info *sbi); 3572 int f2fs_disable_cp_again(struct f2fs_sb_info *sbi, block_t unusable); 3573 void f2fs_release_discard_addrs(struct f2fs_sb_info *sbi); 3574 int f2fs_npages_for_summary_flush(struct f2fs_sb_info *sbi, bool for_ra); 3575 bool f2fs_segment_has_free_slot(struct f2fs_sb_info *sbi, int segno); 3576 void f2fs_init_inmem_curseg(struct f2fs_sb_info *sbi); 3577 void f2fs_save_inmem_curseg(struct f2fs_sb_info *sbi); 3578 void f2fs_restore_inmem_curseg(struct f2fs_sb_info *sbi); 3579 void f2fs_get_new_segment(struct f2fs_sb_info *sbi, 3580 unsigned int *newseg, bool new_sec, int dir); 3581 void f2fs_allocate_segment_for_resize(struct f2fs_sb_info *sbi, int type, 3582 unsigned int start, unsigned int end); 3583 void f2fs_allocate_new_section(struct f2fs_sb_info *sbi, int type, bool force); 3584 void f2fs_allocate_new_segments(struct f2fs_sb_info *sbi); 3585 int f2fs_trim_fs(struct f2fs_sb_info *sbi, struct fstrim_range *range); 3586 bool f2fs_exist_trim_candidates(struct f2fs_sb_info *sbi, 3587 struct cp_control *cpc); 3588 struct page *f2fs_get_sum_page(struct f2fs_sb_info *sbi, unsigned int segno); 3589 void f2fs_update_meta_page(struct f2fs_sb_info *sbi, void *src, 3590 block_t blk_addr); 3591 void f2fs_do_write_meta_page(struct f2fs_sb_info *sbi, struct page *page, 3592 enum iostat_type io_type); 3593 void f2fs_do_write_node_page(unsigned int nid, struct f2fs_io_info *fio); 3594 void f2fs_outplace_write_data(struct dnode_of_data *dn, 3595 struct f2fs_io_info *fio); 3596 int f2fs_inplace_write_data(struct f2fs_io_info *fio); 3597 void f2fs_do_replace_block(struct f2fs_sb_info *sbi, struct f2fs_summary *sum, 3598 block_t old_blkaddr, block_t new_blkaddr, 3599 bool recover_curseg, bool recover_newaddr, 3600 bool from_gc); 3601 void f2fs_replace_block(struct f2fs_sb_info *sbi, struct dnode_of_data *dn, 3602 block_t old_addr, block_t new_addr, 3603 unsigned char version, bool recover_curseg, 3604 bool recover_newaddr); 3605 void f2fs_allocate_data_block(struct f2fs_sb_info *sbi, struct page *page, 3606 block_t old_blkaddr, block_t *new_blkaddr, 3607 struct f2fs_summary *sum, int type, 3608 struct f2fs_io_info *fio); 3609 void f2fs_update_device_state(struct f2fs_sb_info *sbi, nid_t ino, 3610 block_t blkaddr, unsigned int blkcnt); 3611 void f2fs_wait_on_page_writeback(struct page *page, 3612 enum page_type type, bool ordered, bool locked); 3613 void f2fs_wait_on_block_writeback(struct inode *inode, block_t blkaddr); 3614 void f2fs_wait_on_block_writeback_range(struct inode *inode, block_t blkaddr, 3615 block_t len); 3616 void f2fs_write_data_summaries(struct f2fs_sb_info *sbi, block_t start_blk); 3617 void f2fs_write_node_summaries(struct f2fs_sb_info *sbi, block_t start_blk); 3618 int f2fs_lookup_journal_in_cursum(struct f2fs_journal *journal, int type, 3619 unsigned int val, int alloc); 3620 void f2fs_flush_sit_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3621 int f2fs_fix_curseg_write_pointer(struct f2fs_sb_info *sbi); 3622 int f2fs_check_write_pointer(struct f2fs_sb_info *sbi); 3623 int f2fs_build_segment_manager(struct f2fs_sb_info *sbi); 3624 void f2fs_destroy_segment_manager(struct f2fs_sb_info *sbi); 3625 int __init f2fs_create_segment_manager_caches(void); 3626 void f2fs_destroy_segment_manager_caches(void); 3627 int f2fs_rw_hint_to_seg_type(enum rw_hint hint); 3628 enum rw_hint f2fs_io_type_to_rw_hint(struct f2fs_sb_info *sbi, 3629 enum page_type type, enum temp_type temp); 3630 unsigned int f2fs_usable_segs_in_sec(struct f2fs_sb_info *sbi, 3631 unsigned int segno); 3632 unsigned int f2fs_usable_blks_in_seg(struct f2fs_sb_info *sbi, 3633 unsigned int segno); 3634 3635 #define DEF_FRAGMENT_SIZE 4 3636 #define MIN_FRAGMENT_SIZE 1 3637 #define MAX_FRAGMENT_SIZE 512 3638 3639 static inline bool f2fs_need_rand_seg(struct f2fs_sb_info *sbi) 3640 { 3641 return F2FS_OPTION(sbi).fs_mode == FS_MODE_FRAGMENT_SEG || 3642 F2FS_OPTION(sbi).fs_mode == FS_MODE_FRAGMENT_BLK; 3643 } 3644 3645 /* 3646 * checkpoint.c 3647 */ 3648 void f2fs_stop_checkpoint(struct f2fs_sb_info *sbi, bool end_io); 3649 struct page *f2fs_grab_meta_page(struct f2fs_sb_info *sbi, pgoff_t index); 3650 struct page *f2fs_get_meta_page(struct f2fs_sb_info *sbi, pgoff_t index); 3651 struct page *f2fs_get_meta_page_retry(struct f2fs_sb_info *sbi, pgoff_t index); 3652 struct page *f2fs_get_tmp_page(struct f2fs_sb_info *sbi, pgoff_t index); 3653 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi, 3654 block_t blkaddr, int type); 3655 int f2fs_ra_meta_pages(struct f2fs_sb_info *sbi, block_t start, int nrpages, 3656 int type, bool sync); 3657 void f2fs_ra_meta_pages_cond(struct f2fs_sb_info *sbi, pgoff_t index, 3658 unsigned int ra_blocks); 3659 long f2fs_sync_meta_pages(struct f2fs_sb_info *sbi, enum page_type type, 3660 long nr_to_write, enum iostat_type io_type); 3661 void f2fs_add_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type); 3662 void f2fs_remove_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type); 3663 void f2fs_release_ino_entry(struct f2fs_sb_info *sbi, bool all); 3664 bool f2fs_exist_written_data(struct f2fs_sb_info *sbi, nid_t ino, int mode); 3665 void f2fs_set_dirty_device(struct f2fs_sb_info *sbi, nid_t ino, 3666 unsigned int devidx, int type); 3667 bool f2fs_is_dirty_device(struct f2fs_sb_info *sbi, nid_t ino, 3668 unsigned int devidx, int type); 3669 int f2fs_sync_inode_meta(struct f2fs_sb_info *sbi); 3670 int f2fs_acquire_orphan_inode(struct f2fs_sb_info *sbi); 3671 void f2fs_release_orphan_inode(struct f2fs_sb_info *sbi); 3672 void f2fs_add_orphan_inode(struct inode *inode); 3673 void f2fs_remove_orphan_inode(struct f2fs_sb_info *sbi, nid_t ino); 3674 int f2fs_recover_orphan_inodes(struct f2fs_sb_info *sbi); 3675 int f2fs_get_valid_checkpoint(struct f2fs_sb_info *sbi); 3676 void f2fs_update_dirty_page(struct inode *inode, struct page *page); 3677 void f2fs_remove_dirty_inode(struct inode *inode); 3678 int f2fs_sync_dirty_inodes(struct f2fs_sb_info *sbi, enum inode_type type); 3679 void f2fs_wait_on_all_pages(struct f2fs_sb_info *sbi, int type); 3680 u64 f2fs_get_sectors_written(struct f2fs_sb_info *sbi); 3681 int f2fs_write_checkpoint(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3682 void f2fs_init_ino_entry_info(struct f2fs_sb_info *sbi); 3683 int __init f2fs_create_checkpoint_caches(void); 3684 void f2fs_destroy_checkpoint_caches(void); 3685 int f2fs_issue_checkpoint(struct f2fs_sb_info *sbi); 3686 int f2fs_start_ckpt_thread(struct f2fs_sb_info *sbi); 3687 void f2fs_stop_ckpt_thread(struct f2fs_sb_info *sbi); 3688 void f2fs_init_ckpt_req_control(struct f2fs_sb_info *sbi); 3689 3690 /* 3691 * data.c 3692 */ 3693 int __init f2fs_init_bioset(void); 3694 void f2fs_destroy_bioset(void); 3695 int f2fs_init_bio_entry_cache(void); 3696 void f2fs_destroy_bio_entry_cache(void); 3697 void f2fs_submit_bio(struct f2fs_sb_info *sbi, 3698 struct bio *bio, enum page_type type); 3699 void f2fs_submit_merged_write(struct f2fs_sb_info *sbi, enum page_type type); 3700 void f2fs_submit_merged_write_cond(struct f2fs_sb_info *sbi, 3701 struct inode *inode, struct page *page, 3702 nid_t ino, enum page_type type); 3703 void f2fs_submit_merged_ipu_write(struct f2fs_sb_info *sbi, 3704 struct bio **bio, struct page *page); 3705 void f2fs_flush_merged_writes(struct f2fs_sb_info *sbi); 3706 int f2fs_submit_page_bio(struct f2fs_io_info *fio); 3707 int f2fs_merge_page_bio(struct f2fs_io_info *fio); 3708 void f2fs_submit_page_write(struct f2fs_io_info *fio); 3709 struct block_device *f2fs_target_device(struct f2fs_sb_info *sbi, 3710 block_t blk_addr, struct bio *bio); 3711 int f2fs_target_device_index(struct f2fs_sb_info *sbi, block_t blkaddr); 3712 void f2fs_set_data_blkaddr(struct dnode_of_data *dn); 3713 void f2fs_update_data_blkaddr(struct dnode_of_data *dn, block_t blkaddr); 3714 int f2fs_reserve_new_blocks(struct dnode_of_data *dn, blkcnt_t count); 3715 int f2fs_reserve_new_block(struct dnode_of_data *dn); 3716 int f2fs_get_block(struct dnode_of_data *dn, pgoff_t index); 3717 int f2fs_reserve_block(struct dnode_of_data *dn, pgoff_t index); 3718 struct page *f2fs_get_read_data_page(struct inode *inode, pgoff_t index, 3719 int op_flags, bool for_write); 3720 struct page *f2fs_find_data_page(struct inode *inode, pgoff_t index); 3721 struct page *f2fs_get_lock_data_page(struct inode *inode, pgoff_t index, 3722 bool for_write); 3723 struct page *f2fs_get_new_data_page(struct inode *inode, 3724 struct page *ipage, pgoff_t index, bool new_i_size); 3725 int f2fs_do_write_data_page(struct f2fs_io_info *fio); 3726 void f2fs_do_map_lock(struct f2fs_sb_info *sbi, int flag, bool lock); 3727 int f2fs_map_blocks(struct inode *inode, struct f2fs_map_blocks *map, 3728 int create, int flag); 3729 int f2fs_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, 3730 u64 start, u64 len); 3731 int f2fs_encrypt_one_page(struct f2fs_io_info *fio); 3732 bool f2fs_should_update_inplace(struct inode *inode, struct f2fs_io_info *fio); 3733 bool f2fs_should_update_outplace(struct inode *inode, struct f2fs_io_info *fio); 3734 int f2fs_write_single_data_page(struct page *page, int *submitted, 3735 struct bio **bio, sector_t *last_block, 3736 struct writeback_control *wbc, 3737 enum iostat_type io_type, 3738 int compr_blocks, bool allow_balance); 3739 void f2fs_write_failed(struct inode *inode, loff_t to); 3740 void f2fs_invalidate_page(struct page *page, unsigned int offset, 3741 unsigned int length); 3742 int f2fs_release_page(struct page *page, gfp_t wait); 3743 #ifdef CONFIG_MIGRATION 3744 int f2fs_migrate_page(struct address_space *mapping, struct page *newpage, 3745 struct page *page, enum migrate_mode mode); 3746 #endif 3747 bool f2fs_overwrite_io(struct inode *inode, loff_t pos, size_t len); 3748 void f2fs_clear_page_cache_dirty_tag(struct page *page); 3749 int f2fs_init_post_read_processing(void); 3750 void f2fs_destroy_post_read_processing(void); 3751 int f2fs_init_post_read_wq(struct f2fs_sb_info *sbi); 3752 void f2fs_destroy_post_read_wq(struct f2fs_sb_info *sbi); 3753 extern const struct iomap_ops f2fs_iomap_ops; 3754 3755 /* 3756 * gc.c 3757 */ 3758 int f2fs_start_gc_thread(struct f2fs_sb_info *sbi); 3759 void f2fs_stop_gc_thread(struct f2fs_sb_info *sbi); 3760 block_t f2fs_start_bidx_of_node(unsigned int node_ofs, struct inode *inode); 3761 int f2fs_gc(struct f2fs_sb_info *sbi, bool sync, bool background, bool force, 3762 unsigned int segno); 3763 void f2fs_build_gc_manager(struct f2fs_sb_info *sbi); 3764 int f2fs_resize_fs(struct f2fs_sb_info *sbi, __u64 block_count); 3765 int __init f2fs_create_garbage_collection_cache(void); 3766 void f2fs_destroy_garbage_collection_cache(void); 3767 3768 /* 3769 * recovery.c 3770 */ 3771 int f2fs_recover_fsync_data(struct f2fs_sb_info *sbi, bool check_only); 3772 bool f2fs_space_for_roll_forward(struct f2fs_sb_info *sbi); 3773 int __init f2fs_create_recovery_cache(void); 3774 void f2fs_destroy_recovery_cache(void); 3775 3776 /* 3777 * debug.c 3778 */ 3779 #ifdef CONFIG_F2FS_STAT_FS 3780 struct f2fs_stat_info { 3781 struct list_head stat_list; 3782 struct f2fs_sb_info *sbi; 3783 int all_area_segs, sit_area_segs, nat_area_segs, ssa_area_segs; 3784 int main_area_segs, main_area_sections, main_area_zones; 3785 unsigned long long hit_largest, hit_cached, hit_rbtree; 3786 unsigned long long hit_total, total_ext; 3787 int ext_tree, zombie_tree, ext_node; 3788 int ndirty_node, ndirty_dent, ndirty_meta, ndirty_imeta; 3789 int ndirty_data, ndirty_qdata; 3790 int inmem_pages; 3791 unsigned int ndirty_dirs, ndirty_files, nquota_files, ndirty_all; 3792 int nats, dirty_nats, sits, dirty_sits; 3793 int free_nids, avail_nids, alloc_nids; 3794 int total_count, utilization; 3795 int bg_gc, nr_wb_cp_data, nr_wb_data; 3796 int nr_rd_data, nr_rd_node, nr_rd_meta; 3797 int nr_dio_read, nr_dio_write; 3798 unsigned int io_skip_bggc, other_skip_bggc; 3799 int nr_flushing, nr_flushed, flush_list_empty; 3800 int nr_discarding, nr_discarded; 3801 int nr_discard_cmd; 3802 unsigned int undiscard_blks; 3803 int nr_issued_ckpt, nr_total_ckpt, nr_queued_ckpt; 3804 unsigned int cur_ckpt_time, peak_ckpt_time; 3805 int inline_xattr, inline_inode, inline_dir, append, update, orphans; 3806 int compr_inode; 3807 unsigned long long compr_blocks; 3808 int aw_cnt, max_aw_cnt, vw_cnt, max_vw_cnt; 3809 unsigned int valid_count, valid_node_count, valid_inode_count, discard_blks; 3810 unsigned int bimodal, avg_vblocks; 3811 int util_free, util_valid, util_invalid; 3812 int rsvd_segs, overp_segs; 3813 int dirty_count, node_pages, meta_pages, compress_pages; 3814 int compress_page_hit; 3815 int prefree_count, call_count, cp_count, bg_cp_count; 3816 int tot_segs, node_segs, data_segs, free_segs, free_secs; 3817 int bg_node_segs, bg_data_segs; 3818 int tot_blks, data_blks, node_blks; 3819 int bg_data_blks, bg_node_blks; 3820 unsigned long long skipped_atomic_files[2]; 3821 int curseg[NR_CURSEG_TYPE]; 3822 int cursec[NR_CURSEG_TYPE]; 3823 int curzone[NR_CURSEG_TYPE]; 3824 unsigned int dirty_seg[NR_CURSEG_TYPE]; 3825 unsigned int full_seg[NR_CURSEG_TYPE]; 3826 unsigned int valid_blks[NR_CURSEG_TYPE]; 3827 3828 unsigned int meta_count[META_MAX]; 3829 unsigned int segment_count[2]; 3830 unsigned int block_count[2]; 3831 unsigned int inplace_count; 3832 unsigned long long base_mem, cache_mem, page_mem; 3833 }; 3834 3835 static inline struct f2fs_stat_info *F2FS_STAT(struct f2fs_sb_info *sbi) 3836 { 3837 return (struct f2fs_stat_info *)sbi->stat_info; 3838 } 3839 3840 #define stat_inc_cp_count(si) ((si)->cp_count++) 3841 #define stat_inc_bg_cp_count(si) ((si)->bg_cp_count++) 3842 #define stat_inc_call_count(si) ((si)->call_count++) 3843 #define stat_inc_bggc_count(si) ((si)->bg_gc++) 3844 #define stat_io_skip_bggc_count(sbi) ((sbi)->io_skip_bggc++) 3845 #define stat_other_skip_bggc_count(sbi) ((sbi)->other_skip_bggc++) 3846 #define stat_inc_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]++) 3847 #define stat_dec_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]--) 3848 #define stat_inc_total_hit(sbi) (atomic64_inc(&(sbi)->total_hit_ext)) 3849 #define stat_inc_rbtree_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_rbtree)) 3850 #define stat_inc_largest_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_largest)) 3851 #define stat_inc_cached_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_cached)) 3852 #define stat_inc_inline_xattr(inode) \ 3853 do { \ 3854 if (f2fs_has_inline_xattr(inode)) \ 3855 (atomic_inc(&F2FS_I_SB(inode)->inline_xattr)); \ 3856 } while (0) 3857 #define stat_dec_inline_xattr(inode) \ 3858 do { \ 3859 if (f2fs_has_inline_xattr(inode)) \ 3860 (atomic_dec(&F2FS_I_SB(inode)->inline_xattr)); \ 3861 } while (0) 3862 #define stat_inc_inline_inode(inode) \ 3863 do { \ 3864 if (f2fs_has_inline_data(inode)) \ 3865 (atomic_inc(&F2FS_I_SB(inode)->inline_inode)); \ 3866 } while (0) 3867 #define stat_dec_inline_inode(inode) \ 3868 do { \ 3869 if (f2fs_has_inline_data(inode)) \ 3870 (atomic_dec(&F2FS_I_SB(inode)->inline_inode)); \ 3871 } while (0) 3872 #define stat_inc_inline_dir(inode) \ 3873 do { \ 3874 if (f2fs_has_inline_dentry(inode)) \ 3875 (atomic_inc(&F2FS_I_SB(inode)->inline_dir)); \ 3876 } while (0) 3877 #define stat_dec_inline_dir(inode) \ 3878 do { \ 3879 if (f2fs_has_inline_dentry(inode)) \ 3880 (atomic_dec(&F2FS_I_SB(inode)->inline_dir)); \ 3881 } while (0) 3882 #define stat_inc_compr_inode(inode) \ 3883 do { \ 3884 if (f2fs_compressed_file(inode)) \ 3885 (atomic_inc(&F2FS_I_SB(inode)->compr_inode)); \ 3886 } while (0) 3887 #define stat_dec_compr_inode(inode) \ 3888 do { \ 3889 if (f2fs_compressed_file(inode)) \ 3890 (atomic_dec(&F2FS_I_SB(inode)->compr_inode)); \ 3891 } while (0) 3892 #define stat_add_compr_blocks(inode, blocks) \ 3893 (atomic64_add(blocks, &F2FS_I_SB(inode)->compr_blocks)) 3894 #define stat_sub_compr_blocks(inode, blocks) \ 3895 (atomic64_sub(blocks, &F2FS_I_SB(inode)->compr_blocks)) 3896 #define stat_inc_meta_count(sbi, blkaddr) \ 3897 do { \ 3898 if (blkaddr < SIT_I(sbi)->sit_base_addr) \ 3899 atomic_inc(&(sbi)->meta_count[META_CP]); \ 3900 else if (blkaddr < NM_I(sbi)->nat_blkaddr) \ 3901 atomic_inc(&(sbi)->meta_count[META_SIT]); \ 3902 else if (blkaddr < SM_I(sbi)->ssa_blkaddr) \ 3903 atomic_inc(&(sbi)->meta_count[META_NAT]); \ 3904 else if (blkaddr < SM_I(sbi)->main_blkaddr) \ 3905 atomic_inc(&(sbi)->meta_count[META_SSA]); \ 3906 } while (0) 3907 #define stat_inc_seg_type(sbi, curseg) \ 3908 ((sbi)->segment_count[(curseg)->alloc_type]++) 3909 #define stat_inc_block_count(sbi, curseg) \ 3910 ((sbi)->block_count[(curseg)->alloc_type]++) 3911 #define stat_inc_inplace_blocks(sbi) \ 3912 (atomic_inc(&(sbi)->inplace_count)) 3913 #define stat_update_max_atomic_write(inode) \ 3914 do { \ 3915 int cur = F2FS_I_SB(inode)->atomic_files; \ 3916 int max = atomic_read(&F2FS_I_SB(inode)->max_aw_cnt); \ 3917 if (cur > max) \ 3918 atomic_set(&F2FS_I_SB(inode)->max_aw_cnt, cur); \ 3919 } while (0) 3920 #define stat_inc_volatile_write(inode) \ 3921 (atomic_inc(&F2FS_I_SB(inode)->vw_cnt)) 3922 #define stat_dec_volatile_write(inode) \ 3923 (atomic_dec(&F2FS_I_SB(inode)->vw_cnt)) 3924 #define stat_update_max_volatile_write(inode) \ 3925 do { \ 3926 int cur = atomic_read(&F2FS_I_SB(inode)->vw_cnt); \ 3927 int max = atomic_read(&F2FS_I_SB(inode)->max_vw_cnt); \ 3928 if (cur > max) \ 3929 atomic_set(&F2FS_I_SB(inode)->max_vw_cnt, cur); \ 3930 } while (0) 3931 #define stat_inc_seg_count(sbi, type, gc_type) \ 3932 do { \ 3933 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 3934 si->tot_segs++; \ 3935 if ((type) == SUM_TYPE_DATA) { \ 3936 si->data_segs++; \ 3937 si->bg_data_segs += (gc_type == BG_GC) ? 1 : 0; \ 3938 } else { \ 3939 si->node_segs++; \ 3940 si->bg_node_segs += (gc_type == BG_GC) ? 1 : 0; \ 3941 } \ 3942 } while (0) 3943 3944 #define stat_inc_tot_blk_count(si, blks) \ 3945 ((si)->tot_blks += (blks)) 3946 3947 #define stat_inc_data_blk_count(sbi, blks, gc_type) \ 3948 do { \ 3949 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 3950 stat_inc_tot_blk_count(si, blks); \ 3951 si->data_blks += (blks); \ 3952 si->bg_data_blks += ((gc_type) == BG_GC) ? (blks) : 0; \ 3953 } while (0) 3954 3955 #define stat_inc_node_blk_count(sbi, blks, gc_type) \ 3956 do { \ 3957 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 3958 stat_inc_tot_blk_count(si, blks); \ 3959 si->node_blks += (blks); \ 3960 si->bg_node_blks += ((gc_type) == BG_GC) ? (blks) : 0; \ 3961 } while (0) 3962 3963 int f2fs_build_stats(struct f2fs_sb_info *sbi); 3964 void f2fs_destroy_stats(struct f2fs_sb_info *sbi); 3965 void __init f2fs_create_root_stats(void); 3966 void f2fs_destroy_root_stats(void); 3967 void f2fs_update_sit_info(struct f2fs_sb_info *sbi); 3968 #else 3969 #define stat_inc_cp_count(si) do { } while (0) 3970 #define stat_inc_bg_cp_count(si) do { } while (0) 3971 #define stat_inc_call_count(si) do { } while (0) 3972 #define stat_inc_bggc_count(si) do { } while (0) 3973 #define stat_io_skip_bggc_count(sbi) do { } while (0) 3974 #define stat_other_skip_bggc_count(sbi) do { } while (0) 3975 #define stat_inc_dirty_inode(sbi, type) do { } while (0) 3976 #define stat_dec_dirty_inode(sbi, type) do { } while (0) 3977 #define stat_inc_total_hit(sbi) do { } while (0) 3978 #define stat_inc_rbtree_node_hit(sbi) do { } while (0) 3979 #define stat_inc_largest_node_hit(sbi) do { } while (0) 3980 #define stat_inc_cached_node_hit(sbi) do { } while (0) 3981 #define stat_inc_inline_xattr(inode) do { } while (0) 3982 #define stat_dec_inline_xattr(inode) do { } while (0) 3983 #define stat_inc_inline_inode(inode) do { } while (0) 3984 #define stat_dec_inline_inode(inode) do { } while (0) 3985 #define stat_inc_inline_dir(inode) do { } while (0) 3986 #define stat_dec_inline_dir(inode) do { } while (0) 3987 #define stat_inc_compr_inode(inode) do { } while (0) 3988 #define stat_dec_compr_inode(inode) do { } while (0) 3989 #define stat_add_compr_blocks(inode, blocks) do { } while (0) 3990 #define stat_sub_compr_blocks(inode, blocks) do { } while (0) 3991 #define stat_update_max_atomic_write(inode) do { } while (0) 3992 #define stat_inc_volatile_write(inode) do { } while (0) 3993 #define stat_dec_volatile_write(inode) do { } while (0) 3994 #define stat_update_max_volatile_write(inode) do { } while (0) 3995 #define stat_inc_meta_count(sbi, blkaddr) do { } while (0) 3996 #define stat_inc_seg_type(sbi, curseg) do { } while (0) 3997 #define stat_inc_block_count(sbi, curseg) do { } while (0) 3998 #define stat_inc_inplace_blocks(sbi) do { } while (0) 3999 #define stat_inc_seg_count(sbi, type, gc_type) do { } while (0) 4000 #define stat_inc_tot_blk_count(si, blks) do { } while (0) 4001 #define stat_inc_data_blk_count(sbi, blks, gc_type) do { } while (0) 4002 #define stat_inc_node_blk_count(sbi, blks, gc_type) do { } while (0) 4003 4004 static inline int f2fs_build_stats(struct f2fs_sb_info *sbi) { return 0; } 4005 static inline void f2fs_destroy_stats(struct f2fs_sb_info *sbi) { } 4006 static inline void __init f2fs_create_root_stats(void) { } 4007 static inline void f2fs_destroy_root_stats(void) { } 4008 static inline void f2fs_update_sit_info(struct f2fs_sb_info *sbi) {} 4009 #endif 4010 4011 extern const struct file_operations f2fs_dir_operations; 4012 extern const struct file_operations f2fs_file_operations; 4013 extern const struct inode_operations f2fs_file_inode_operations; 4014 extern const struct address_space_operations f2fs_dblock_aops; 4015 extern const struct address_space_operations f2fs_node_aops; 4016 extern const struct address_space_operations f2fs_meta_aops; 4017 extern const struct inode_operations f2fs_dir_inode_operations; 4018 extern const struct inode_operations f2fs_symlink_inode_operations; 4019 extern const struct inode_operations f2fs_encrypted_symlink_inode_operations; 4020 extern const struct inode_operations f2fs_special_inode_operations; 4021 extern struct kmem_cache *f2fs_inode_entry_slab; 4022 4023 /* 4024 * inline.c 4025 */ 4026 bool f2fs_may_inline_data(struct inode *inode); 4027 bool f2fs_may_inline_dentry(struct inode *inode); 4028 void f2fs_do_read_inline_data(struct page *page, struct page *ipage); 4029 void f2fs_truncate_inline_inode(struct inode *inode, 4030 struct page *ipage, u64 from); 4031 int f2fs_read_inline_data(struct inode *inode, struct page *page); 4032 int f2fs_convert_inline_page(struct dnode_of_data *dn, struct page *page); 4033 int f2fs_convert_inline_inode(struct inode *inode); 4034 int f2fs_try_convert_inline_dir(struct inode *dir, struct dentry *dentry); 4035 int f2fs_write_inline_data(struct inode *inode, struct page *page); 4036 int f2fs_recover_inline_data(struct inode *inode, struct page *npage); 4037 struct f2fs_dir_entry *f2fs_find_in_inline_dir(struct inode *dir, 4038 const struct f2fs_filename *fname, 4039 struct page **res_page); 4040 int f2fs_make_empty_inline_dir(struct inode *inode, struct inode *parent, 4041 struct page *ipage); 4042 int f2fs_add_inline_entry(struct inode *dir, const struct f2fs_filename *fname, 4043 struct inode *inode, nid_t ino, umode_t mode); 4044 void f2fs_delete_inline_entry(struct f2fs_dir_entry *dentry, 4045 struct page *page, struct inode *dir, 4046 struct inode *inode); 4047 bool f2fs_empty_inline_dir(struct inode *dir); 4048 int f2fs_read_inline_dir(struct file *file, struct dir_context *ctx, 4049 struct fscrypt_str *fstr); 4050 int f2fs_inline_data_fiemap(struct inode *inode, 4051 struct fiemap_extent_info *fieinfo, 4052 __u64 start, __u64 len); 4053 4054 /* 4055 * shrinker.c 4056 */ 4057 unsigned long f2fs_shrink_count(struct shrinker *shrink, 4058 struct shrink_control *sc); 4059 unsigned long f2fs_shrink_scan(struct shrinker *shrink, 4060 struct shrink_control *sc); 4061 void f2fs_join_shrinker(struct f2fs_sb_info *sbi); 4062 void f2fs_leave_shrinker(struct f2fs_sb_info *sbi); 4063 4064 /* 4065 * extent_cache.c 4066 */ 4067 struct rb_entry *f2fs_lookup_rb_tree(struct rb_root_cached *root, 4068 struct rb_entry *cached_re, unsigned int ofs); 4069 struct rb_node **f2fs_lookup_rb_tree_ext(struct f2fs_sb_info *sbi, 4070 struct rb_root_cached *root, 4071 struct rb_node **parent, 4072 unsigned long long key, bool *left_most); 4073 struct rb_node **f2fs_lookup_rb_tree_for_insert(struct f2fs_sb_info *sbi, 4074 struct rb_root_cached *root, 4075 struct rb_node **parent, 4076 unsigned int ofs, bool *leftmost); 4077 struct rb_entry *f2fs_lookup_rb_tree_ret(struct rb_root_cached *root, 4078 struct rb_entry *cached_re, unsigned int ofs, 4079 struct rb_entry **prev_entry, struct rb_entry **next_entry, 4080 struct rb_node ***insert_p, struct rb_node **insert_parent, 4081 bool force, bool *leftmost); 4082 bool f2fs_check_rb_tree_consistence(struct f2fs_sb_info *sbi, 4083 struct rb_root_cached *root, bool check_key); 4084 unsigned int f2fs_shrink_extent_tree(struct f2fs_sb_info *sbi, int nr_shrink); 4085 void f2fs_init_extent_tree(struct inode *inode, struct page *ipage); 4086 void f2fs_drop_extent_tree(struct inode *inode); 4087 unsigned int f2fs_destroy_extent_node(struct inode *inode); 4088 void f2fs_destroy_extent_tree(struct inode *inode); 4089 bool f2fs_lookup_extent_cache(struct inode *inode, pgoff_t pgofs, 4090 struct extent_info *ei); 4091 void f2fs_update_extent_cache(struct dnode_of_data *dn); 4092 void f2fs_update_extent_cache_range(struct dnode_of_data *dn, 4093 pgoff_t fofs, block_t blkaddr, unsigned int len); 4094 void f2fs_init_extent_cache_info(struct f2fs_sb_info *sbi); 4095 int __init f2fs_create_extent_cache(void); 4096 void f2fs_destroy_extent_cache(void); 4097 4098 /* 4099 * sysfs.c 4100 */ 4101 #define MIN_RA_MUL 2 4102 #define MAX_RA_MUL 256 4103 4104 int __init f2fs_init_sysfs(void); 4105 void f2fs_exit_sysfs(void); 4106 int f2fs_register_sysfs(struct f2fs_sb_info *sbi); 4107 void f2fs_unregister_sysfs(struct f2fs_sb_info *sbi); 4108 4109 /* verity.c */ 4110 extern const struct fsverity_operations f2fs_verityops; 4111 4112 /* 4113 * crypto support 4114 */ 4115 static inline bool f2fs_encrypted_file(struct inode *inode) 4116 { 4117 return IS_ENCRYPTED(inode) && S_ISREG(inode->i_mode); 4118 } 4119 4120 static inline void f2fs_set_encrypted_inode(struct inode *inode) 4121 { 4122 #ifdef CONFIG_FS_ENCRYPTION 4123 file_set_encrypt(inode); 4124 f2fs_set_inode_flags(inode); 4125 #endif 4126 } 4127 4128 /* 4129 * Returns true if the reads of the inode's data need to undergo some 4130 * postprocessing step, like decryption or authenticity verification. 4131 */ 4132 static inline bool f2fs_post_read_required(struct inode *inode) 4133 { 4134 return f2fs_encrypted_file(inode) || fsverity_active(inode) || 4135 f2fs_compressed_file(inode); 4136 } 4137 4138 /* 4139 * compress.c 4140 */ 4141 #ifdef CONFIG_F2FS_FS_COMPRESSION 4142 bool f2fs_is_compressed_page(struct page *page); 4143 struct page *f2fs_compress_control_page(struct page *page); 4144 int f2fs_prepare_compress_overwrite(struct inode *inode, 4145 struct page **pagep, pgoff_t index, void **fsdata); 4146 bool f2fs_compress_write_end(struct inode *inode, void *fsdata, 4147 pgoff_t index, unsigned copied); 4148 int f2fs_truncate_partial_cluster(struct inode *inode, u64 from, bool lock); 4149 void f2fs_compress_write_end_io(struct bio *bio, struct page *page); 4150 bool f2fs_is_compress_backend_ready(struct inode *inode); 4151 int f2fs_init_compress_mempool(void); 4152 void f2fs_destroy_compress_mempool(void); 4153 void f2fs_decompress_cluster(struct decompress_io_ctx *dic); 4154 void f2fs_end_read_compressed_page(struct page *page, bool failed, 4155 block_t blkaddr); 4156 bool f2fs_cluster_is_empty(struct compress_ctx *cc); 4157 bool f2fs_cluster_can_merge_page(struct compress_ctx *cc, pgoff_t index); 4158 bool f2fs_all_cluster_page_loaded(struct compress_ctx *cc, struct pagevec *pvec, 4159 int index, int nr_pages); 4160 bool f2fs_sanity_check_cluster(struct dnode_of_data *dn); 4161 void f2fs_compress_ctx_add_page(struct compress_ctx *cc, struct page *page); 4162 int f2fs_write_multi_pages(struct compress_ctx *cc, 4163 int *submitted, 4164 struct writeback_control *wbc, 4165 enum iostat_type io_type); 4166 int f2fs_is_compressed_cluster(struct inode *inode, pgoff_t index); 4167 void f2fs_update_extent_tree_range_compressed(struct inode *inode, 4168 pgoff_t fofs, block_t blkaddr, unsigned int llen, 4169 unsigned int c_len); 4170 int f2fs_read_multi_pages(struct compress_ctx *cc, struct bio **bio_ret, 4171 unsigned nr_pages, sector_t *last_block_in_bio, 4172 bool is_readahead, bool for_write); 4173 struct decompress_io_ctx *f2fs_alloc_dic(struct compress_ctx *cc); 4174 void f2fs_decompress_end_io(struct decompress_io_ctx *dic, bool failed); 4175 void f2fs_put_page_dic(struct page *page); 4176 unsigned int f2fs_cluster_blocks_are_contiguous(struct dnode_of_data *dn); 4177 int f2fs_init_compress_ctx(struct compress_ctx *cc); 4178 void f2fs_destroy_compress_ctx(struct compress_ctx *cc, bool reuse); 4179 void f2fs_init_compress_info(struct f2fs_sb_info *sbi); 4180 int f2fs_init_compress_inode(struct f2fs_sb_info *sbi); 4181 void f2fs_destroy_compress_inode(struct f2fs_sb_info *sbi); 4182 int f2fs_init_page_array_cache(struct f2fs_sb_info *sbi); 4183 void f2fs_destroy_page_array_cache(struct f2fs_sb_info *sbi); 4184 int __init f2fs_init_compress_cache(void); 4185 void f2fs_destroy_compress_cache(void); 4186 struct address_space *COMPRESS_MAPPING(struct f2fs_sb_info *sbi); 4187 void f2fs_invalidate_compress_page(struct f2fs_sb_info *sbi, block_t blkaddr); 4188 void f2fs_cache_compressed_page(struct f2fs_sb_info *sbi, struct page *page, 4189 nid_t ino, block_t blkaddr); 4190 bool f2fs_load_compressed_page(struct f2fs_sb_info *sbi, struct page *page, 4191 block_t blkaddr); 4192 void f2fs_invalidate_compress_pages(struct f2fs_sb_info *sbi, nid_t ino); 4193 #define inc_compr_inode_stat(inode) \ 4194 do { \ 4195 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); \ 4196 sbi->compr_new_inode++; \ 4197 } while (0) 4198 #define add_compr_block_stat(inode, blocks) \ 4199 do { \ 4200 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); \ 4201 int diff = F2FS_I(inode)->i_cluster_size - blocks; \ 4202 sbi->compr_written_block += blocks; \ 4203 sbi->compr_saved_block += diff; \ 4204 } while (0) 4205 #else 4206 static inline bool f2fs_is_compressed_page(struct page *page) { return false; } 4207 static inline bool f2fs_is_compress_backend_ready(struct inode *inode) 4208 { 4209 if (!f2fs_compressed_file(inode)) 4210 return true; 4211 /* not support compression */ 4212 return false; 4213 } 4214 static inline struct page *f2fs_compress_control_page(struct page *page) 4215 { 4216 WARN_ON_ONCE(1); 4217 return ERR_PTR(-EINVAL); 4218 } 4219 static inline int f2fs_init_compress_mempool(void) { return 0; } 4220 static inline void f2fs_destroy_compress_mempool(void) { } 4221 static inline void f2fs_decompress_cluster(struct decompress_io_ctx *dic) { } 4222 static inline void f2fs_end_read_compressed_page(struct page *page, 4223 bool failed, block_t blkaddr) 4224 { 4225 WARN_ON_ONCE(1); 4226 } 4227 static inline void f2fs_put_page_dic(struct page *page) 4228 { 4229 WARN_ON_ONCE(1); 4230 } 4231 static inline unsigned int f2fs_cluster_blocks_are_contiguous(struct dnode_of_data *dn) { return 0; } 4232 static inline bool f2fs_sanity_check_cluster(struct dnode_of_data *dn) { return false; } 4233 static inline int f2fs_init_compress_inode(struct f2fs_sb_info *sbi) { return 0; } 4234 static inline void f2fs_destroy_compress_inode(struct f2fs_sb_info *sbi) { } 4235 static inline int f2fs_init_page_array_cache(struct f2fs_sb_info *sbi) { return 0; } 4236 static inline void f2fs_destroy_page_array_cache(struct f2fs_sb_info *sbi) { } 4237 static inline int __init f2fs_init_compress_cache(void) { return 0; } 4238 static inline void f2fs_destroy_compress_cache(void) { } 4239 static inline void f2fs_invalidate_compress_page(struct f2fs_sb_info *sbi, 4240 block_t blkaddr) { } 4241 static inline void f2fs_cache_compressed_page(struct f2fs_sb_info *sbi, 4242 struct page *page, nid_t ino, block_t blkaddr) { } 4243 static inline bool f2fs_load_compressed_page(struct f2fs_sb_info *sbi, 4244 struct page *page, block_t blkaddr) { return false; } 4245 static inline void f2fs_invalidate_compress_pages(struct f2fs_sb_info *sbi, 4246 nid_t ino) { } 4247 #define inc_compr_inode_stat(inode) do { } while (0) 4248 static inline void f2fs_update_extent_tree_range_compressed(struct inode *inode, 4249 pgoff_t fofs, block_t blkaddr, unsigned int llen, 4250 unsigned int c_len) { } 4251 #endif 4252 4253 static inline void set_compress_context(struct inode *inode) 4254 { 4255 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 4256 4257 F2FS_I(inode)->i_compress_algorithm = 4258 F2FS_OPTION(sbi).compress_algorithm; 4259 F2FS_I(inode)->i_log_cluster_size = 4260 F2FS_OPTION(sbi).compress_log_size; 4261 F2FS_I(inode)->i_compress_flag = 4262 F2FS_OPTION(sbi).compress_chksum ? 4263 1 << COMPRESS_CHKSUM : 0; 4264 F2FS_I(inode)->i_cluster_size = 4265 1 << F2FS_I(inode)->i_log_cluster_size; 4266 if ((F2FS_I(inode)->i_compress_algorithm == COMPRESS_LZ4 || 4267 F2FS_I(inode)->i_compress_algorithm == COMPRESS_ZSTD) && 4268 F2FS_OPTION(sbi).compress_level) 4269 F2FS_I(inode)->i_compress_flag |= 4270 F2FS_OPTION(sbi).compress_level << 4271 COMPRESS_LEVEL_OFFSET; 4272 F2FS_I(inode)->i_flags |= F2FS_COMPR_FL; 4273 set_inode_flag(inode, FI_COMPRESSED_FILE); 4274 stat_inc_compr_inode(inode); 4275 inc_compr_inode_stat(inode); 4276 f2fs_mark_inode_dirty_sync(inode, true); 4277 } 4278 4279 static inline bool f2fs_disable_compressed_file(struct inode *inode) 4280 { 4281 struct f2fs_inode_info *fi = F2FS_I(inode); 4282 4283 if (!f2fs_compressed_file(inode)) 4284 return true; 4285 if (S_ISREG(inode->i_mode) && F2FS_HAS_BLOCKS(inode)) 4286 return false; 4287 4288 fi->i_flags &= ~F2FS_COMPR_FL; 4289 stat_dec_compr_inode(inode); 4290 clear_inode_flag(inode, FI_COMPRESSED_FILE); 4291 f2fs_mark_inode_dirty_sync(inode, true); 4292 return true; 4293 } 4294 4295 #define F2FS_FEATURE_FUNCS(name, flagname) \ 4296 static inline int f2fs_sb_has_##name(struct f2fs_sb_info *sbi) \ 4297 { \ 4298 return F2FS_HAS_FEATURE(sbi, F2FS_FEATURE_##flagname); \ 4299 } 4300 4301 F2FS_FEATURE_FUNCS(encrypt, ENCRYPT); 4302 F2FS_FEATURE_FUNCS(blkzoned, BLKZONED); 4303 F2FS_FEATURE_FUNCS(extra_attr, EXTRA_ATTR); 4304 F2FS_FEATURE_FUNCS(project_quota, PRJQUOTA); 4305 F2FS_FEATURE_FUNCS(inode_chksum, INODE_CHKSUM); 4306 F2FS_FEATURE_FUNCS(flexible_inline_xattr, FLEXIBLE_INLINE_XATTR); 4307 F2FS_FEATURE_FUNCS(quota_ino, QUOTA_INO); 4308 F2FS_FEATURE_FUNCS(inode_crtime, INODE_CRTIME); 4309 F2FS_FEATURE_FUNCS(lost_found, LOST_FOUND); 4310 F2FS_FEATURE_FUNCS(verity, VERITY); 4311 F2FS_FEATURE_FUNCS(sb_chksum, SB_CHKSUM); 4312 F2FS_FEATURE_FUNCS(casefold, CASEFOLD); 4313 F2FS_FEATURE_FUNCS(compression, COMPRESSION); 4314 F2FS_FEATURE_FUNCS(readonly, RO); 4315 4316 static inline bool f2fs_may_extent_tree(struct inode *inode) 4317 { 4318 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 4319 4320 if (!test_opt(sbi, EXTENT_CACHE) || 4321 is_inode_flag_set(inode, FI_NO_EXTENT) || 4322 (is_inode_flag_set(inode, FI_COMPRESSED_FILE) && 4323 !f2fs_sb_has_readonly(sbi))) 4324 return false; 4325 4326 /* 4327 * for recovered files during mount do not create extents 4328 * if shrinker is not registered. 4329 */ 4330 if (list_empty(&sbi->s_list)) 4331 return false; 4332 4333 return S_ISREG(inode->i_mode); 4334 } 4335 4336 #ifdef CONFIG_BLK_DEV_ZONED 4337 static inline bool f2fs_blkz_is_seq(struct f2fs_sb_info *sbi, int devi, 4338 block_t blkaddr) 4339 { 4340 unsigned int zno = blkaddr >> sbi->log_blocks_per_blkz; 4341 4342 return test_bit(zno, FDEV(devi).blkz_seq); 4343 } 4344 #endif 4345 4346 static inline bool f2fs_hw_should_discard(struct f2fs_sb_info *sbi) 4347 { 4348 return f2fs_sb_has_blkzoned(sbi); 4349 } 4350 4351 static inline bool f2fs_bdev_support_discard(struct block_device *bdev) 4352 { 4353 return blk_queue_discard(bdev_get_queue(bdev)) || 4354 bdev_is_zoned(bdev); 4355 } 4356 4357 static inline bool f2fs_hw_support_discard(struct f2fs_sb_info *sbi) 4358 { 4359 int i; 4360 4361 if (!f2fs_is_multi_device(sbi)) 4362 return f2fs_bdev_support_discard(sbi->sb->s_bdev); 4363 4364 for (i = 0; i < sbi->s_ndevs; i++) 4365 if (f2fs_bdev_support_discard(FDEV(i).bdev)) 4366 return true; 4367 return false; 4368 } 4369 4370 static inline bool f2fs_realtime_discard_enable(struct f2fs_sb_info *sbi) 4371 { 4372 return (test_opt(sbi, DISCARD) && f2fs_hw_support_discard(sbi)) || 4373 f2fs_hw_should_discard(sbi); 4374 } 4375 4376 static inline bool f2fs_hw_is_readonly(struct f2fs_sb_info *sbi) 4377 { 4378 int i; 4379 4380 if (!f2fs_is_multi_device(sbi)) 4381 return bdev_read_only(sbi->sb->s_bdev); 4382 4383 for (i = 0; i < sbi->s_ndevs; i++) 4384 if (bdev_read_only(FDEV(i).bdev)) 4385 return true; 4386 return false; 4387 } 4388 4389 static inline bool f2fs_lfs_mode(struct f2fs_sb_info *sbi) 4390 { 4391 return F2FS_OPTION(sbi).fs_mode == FS_MODE_LFS; 4392 } 4393 4394 static inline bool f2fs_may_compress(struct inode *inode) 4395 { 4396 if (IS_SWAPFILE(inode) || f2fs_is_pinned_file(inode) || 4397 f2fs_is_atomic_file(inode) || 4398 f2fs_is_volatile_file(inode)) 4399 return false; 4400 return S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode); 4401 } 4402 4403 static inline void f2fs_i_compr_blocks_update(struct inode *inode, 4404 u64 blocks, bool add) 4405 { 4406 int diff = F2FS_I(inode)->i_cluster_size - blocks; 4407 struct f2fs_inode_info *fi = F2FS_I(inode); 4408 4409 /* don't update i_compr_blocks if saved blocks were released */ 4410 if (!add && !atomic_read(&fi->i_compr_blocks)) 4411 return; 4412 4413 if (add) { 4414 atomic_add(diff, &fi->i_compr_blocks); 4415 stat_add_compr_blocks(inode, diff); 4416 } else { 4417 atomic_sub(diff, &fi->i_compr_blocks); 4418 stat_sub_compr_blocks(inode, diff); 4419 } 4420 f2fs_mark_inode_dirty_sync(inode, true); 4421 } 4422 4423 static inline int block_unaligned_IO(struct inode *inode, 4424 struct kiocb *iocb, struct iov_iter *iter) 4425 { 4426 unsigned int i_blkbits = READ_ONCE(inode->i_blkbits); 4427 unsigned int blocksize_mask = (1 << i_blkbits) - 1; 4428 loff_t offset = iocb->ki_pos; 4429 unsigned long align = offset | iov_iter_alignment(iter); 4430 4431 return align & blocksize_mask; 4432 } 4433 4434 static inline bool f2fs_allow_multi_device_dio(struct f2fs_sb_info *sbi, 4435 int flag) 4436 { 4437 if (!f2fs_is_multi_device(sbi)) 4438 return false; 4439 if (flag != F2FS_GET_BLOCK_DIO) 4440 return false; 4441 return sbi->aligned_blksize; 4442 } 4443 4444 static inline bool f2fs_force_buffered_io(struct inode *inode, 4445 struct kiocb *iocb, struct iov_iter *iter) 4446 { 4447 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 4448 int rw = iov_iter_rw(iter); 4449 4450 if (f2fs_post_read_required(inode)) 4451 return true; 4452 4453 /* disallow direct IO if any of devices has unaligned blksize */ 4454 if (f2fs_is_multi_device(sbi) && !sbi->aligned_blksize) 4455 return true; 4456 /* 4457 * for blkzoned device, fallback direct IO to buffered IO, so 4458 * all IOs can be serialized by log-structured write. 4459 */ 4460 if (f2fs_sb_has_blkzoned(sbi)) 4461 return true; 4462 if (f2fs_lfs_mode(sbi) && (rw == WRITE)) { 4463 if (block_unaligned_IO(inode, iocb, iter)) 4464 return true; 4465 if (F2FS_IO_ALIGNED(sbi)) 4466 return true; 4467 } 4468 if (is_sbi_flag_set(F2FS_I_SB(inode), SBI_CP_DISABLED)) 4469 return true; 4470 4471 return false; 4472 } 4473 4474 static inline bool f2fs_need_verity(const struct inode *inode, pgoff_t idx) 4475 { 4476 return fsverity_active(inode) && 4477 idx < DIV_ROUND_UP(inode->i_size, PAGE_SIZE); 4478 } 4479 4480 #ifdef CONFIG_F2FS_FAULT_INJECTION 4481 extern void f2fs_build_fault_attr(struct f2fs_sb_info *sbi, unsigned int rate, 4482 unsigned int type); 4483 #else 4484 #define f2fs_build_fault_attr(sbi, rate, type) do { } while (0) 4485 #endif 4486 4487 static inline bool is_journalled_quota(struct f2fs_sb_info *sbi) 4488 { 4489 #ifdef CONFIG_QUOTA 4490 if (f2fs_sb_has_quota_ino(sbi)) 4491 return true; 4492 if (F2FS_OPTION(sbi).s_qf_names[USRQUOTA] || 4493 F2FS_OPTION(sbi).s_qf_names[GRPQUOTA] || 4494 F2FS_OPTION(sbi).s_qf_names[PRJQUOTA]) 4495 return true; 4496 #endif 4497 return false; 4498 } 4499 4500 static inline bool f2fs_block_unit_discard(struct f2fs_sb_info *sbi) 4501 { 4502 return F2FS_OPTION(sbi).discard_unit == DISCARD_UNIT_BLOCK; 4503 } 4504 4505 #define EFSBADCRC EBADMSG /* Bad CRC detected */ 4506 #define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */ 4507 4508 #endif /* _LINUX_F2FS_H */ 4509