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 #ifdef CONFIG_F2FS_UNFAIR_RWSEM 136 wait_queue_head_t read_waiters; 137 #endif 138 }; 139 140 struct f2fs_mount_info { 141 unsigned int opt; 142 int write_io_size_bits; /* Write IO size bits */ 143 block_t root_reserved_blocks; /* root reserved blocks */ 144 kuid_t s_resuid; /* reserved blocks for uid */ 145 kgid_t s_resgid; /* reserved blocks for gid */ 146 int active_logs; /* # of active logs */ 147 int inline_xattr_size; /* inline xattr size */ 148 #ifdef CONFIG_F2FS_FAULT_INJECTION 149 struct f2fs_fault_info fault_info; /* For fault injection */ 150 #endif 151 #ifdef CONFIG_QUOTA 152 /* Names of quota files with journalled quota */ 153 char *s_qf_names[MAXQUOTAS]; 154 int s_jquota_fmt; /* Format of quota to use */ 155 #endif 156 /* For which write hints are passed down to block layer */ 157 int whint_mode; 158 int alloc_mode; /* segment allocation policy */ 159 int fsync_mode; /* fsync policy */ 160 int fs_mode; /* fs mode: LFS or ADAPTIVE */ 161 int bggc_mode; /* bggc mode: off, on or sync */ 162 int discard_unit; /* 163 * discard command's offset/size should 164 * be aligned to this unit: block, 165 * segment or section 166 */ 167 struct fscrypt_dummy_policy dummy_enc_policy; /* test dummy encryption */ 168 block_t unusable_cap_perc; /* percentage for cap */ 169 block_t unusable_cap; /* Amount of space allowed to be 170 * unusable when disabling checkpoint 171 */ 172 173 /* For compression */ 174 unsigned char compress_algorithm; /* algorithm type */ 175 unsigned char compress_log_size; /* cluster log size */ 176 unsigned char compress_level; /* compress level */ 177 bool compress_chksum; /* compressed data chksum */ 178 unsigned char compress_ext_cnt; /* extension count */ 179 unsigned char nocompress_ext_cnt; /* nocompress extension count */ 180 int compress_mode; /* compression mode */ 181 unsigned char extensions[COMPRESS_EXT_NUM][F2FS_EXTENSION_LEN]; /* extensions */ 182 unsigned char noextensions[COMPRESS_EXT_NUM][F2FS_EXTENSION_LEN]; /* extensions */ 183 }; 184 185 #define F2FS_FEATURE_ENCRYPT 0x0001 186 #define F2FS_FEATURE_BLKZONED 0x0002 187 #define F2FS_FEATURE_ATOMIC_WRITE 0x0004 188 #define F2FS_FEATURE_EXTRA_ATTR 0x0008 189 #define F2FS_FEATURE_PRJQUOTA 0x0010 190 #define F2FS_FEATURE_INODE_CHKSUM 0x0020 191 #define F2FS_FEATURE_FLEXIBLE_INLINE_XATTR 0x0040 192 #define F2FS_FEATURE_QUOTA_INO 0x0080 193 #define F2FS_FEATURE_INODE_CRTIME 0x0100 194 #define F2FS_FEATURE_LOST_FOUND 0x0200 195 #define F2FS_FEATURE_VERITY 0x0400 196 #define F2FS_FEATURE_SB_CHKSUM 0x0800 197 #define F2FS_FEATURE_CASEFOLD 0x1000 198 #define F2FS_FEATURE_COMPRESSION 0x2000 199 #define F2FS_FEATURE_RO 0x4000 200 201 #define __F2FS_HAS_FEATURE(raw_super, mask) \ 202 ((raw_super->feature & cpu_to_le32(mask)) != 0) 203 #define F2FS_HAS_FEATURE(sbi, mask) __F2FS_HAS_FEATURE(sbi->raw_super, mask) 204 #define F2FS_SET_FEATURE(sbi, mask) \ 205 (sbi->raw_super->feature |= cpu_to_le32(mask)) 206 #define F2FS_CLEAR_FEATURE(sbi, mask) \ 207 (sbi->raw_super->feature &= ~cpu_to_le32(mask)) 208 209 /* 210 * Default values for user and/or group using reserved blocks 211 */ 212 #define F2FS_DEF_RESUID 0 213 #define F2FS_DEF_RESGID 0 214 215 /* 216 * For checkpoint manager 217 */ 218 enum { 219 NAT_BITMAP, 220 SIT_BITMAP 221 }; 222 223 #define CP_UMOUNT 0x00000001 224 #define CP_FASTBOOT 0x00000002 225 #define CP_SYNC 0x00000004 226 #define CP_RECOVERY 0x00000008 227 #define CP_DISCARD 0x00000010 228 #define CP_TRIMMED 0x00000020 229 #define CP_PAUSE 0x00000040 230 #define CP_RESIZE 0x00000080 231 232 #define MAX_DISCARD_BLOCKS(sbi) BLKS_PER_SEC(sbi) 233 #define DEF_MAX_DISCARD_REQUEST 8 /* issue 8 discards per round */ 234 #define DEF_MIN_DISCARD_ISSUE_TIME 50 /* 50 ms, if exists */ 235 #define DEF_MID_DISCARD_ISSUE_TIME 500 /* 500 ms, if device busy */ 236 #define DEF_MAX_DISCARD_ISSUE_TIME 60000 /* 60 s, if no candidates */ 237 #define DEF_DISCARD_URGENT_UTIL 80 /* do more discard over 80% */ 238 #define DEF_CP_INTERVAL 60 /* 60 secs */ 239 #define DEF_IDLE_INTERVAL 5 /* 5 secs */ 240 #define DEF_DISABLE_INTERVAL 5 /* 5 secs */ 241 #define DEF_DISABLE_QUICK_INTERVAL 1 /* 1 secs */ 242 #define DEF_UMOUNT_DISCARD_TIMEOUT 5 /* 5 secs */ 243 244 struct cp_control { 245 int reason; 246 __u64 trim_start; 247 __u64 trim_end; 248 __u64 trim_minlen; 249 }; 250 251 /* 252 * indicate meta/data type 253 */ 254 enum { 255 META_CP, 256 META_NAT, 257 META_SIT, 258 META_SSA, 259 META_MAX, 260 META_POR, 261 DATA_GENERIC, /* check range only */ 262 DATA_GENERIC_ENHANCE, /* strong check on range and segment bitmap */ 263 DATA_GENERIC_ENHANCE_READ, /* 264 * strong check on range and segment 265 * bitmap but no warning due to race 266 * condition of read on truncated area 267 * by extent_cache 268 */ 269 META_GENERIC, 270 }; 271 272 /* for the list of ino */ 273 enum { 274 ORPHAN_INO, /* for orphan ino list */ 275 APPEND_INO, /* for append ino list */ 276 UPDATE_INO, /* for update ino list */ 277 TRANS_DIR_INO, /* for trasactions dir ino list */ 278 FLUSH_INO, /* for multiple device flushing */ 279 MAX_INO_ENTRY, /* max. list */ 280 }; 281 282 struct ino_entry { 283 struct list_head list; /* list head */ 284 nid_t ino; /* inode number */ 285 unsigned int dirty_device; /* dirty device bitmap */ 286 }; 287 288 /* for the list of inodes to be GCed */ 289 struct inode_entry { 290 struct list_head list; /* list head */ 291 struct inode *inode; /* vfs inode pointer */ 292 }; 293 294 struct fsync_node_entry { 295 struct list_head list; /* list head */ 296 struct page *page; /* warm node page pointer */ 297 unsigned int seq_id; /* sequence id */ 298 }; 299 300 struct ckpt_req { 301 struct completion wait; /* completion for checkpoint done */ 302 struct llist_node llnode; /* llist_node to be linked in wait queue */ 303 int ret; /* return code of checkpoint */ 304 ktime_t queue_time; /* request queued time */ 305 }; 306 307 struct ckpt_req_control { 308 struct task_struct *f2fs_issue_ckpt; /* checkpoint task */ 309 int ckpt_thread_ioprio; /* checkpoint merge thread ioprio */ 310 wait_queue_head_t ckpt_wait_queue; /* waiting queue for wake-up */ 311 atomic_t issued_ckpt; /* # of actually issued ckpts */ 312 atomic_t total_ckpt; /* # of total ckpts */ 313 atomic_t queued_ckpt; /* # of queued ckpts */ 314 struct llist_head issue_list; /* list for command issue */ 315 spinlock_t stat_lock; /* lock for below checkpoint time stats */ 316 unsigned int cur_time; /* cur wait time in msec for currently issued checkpoint */ 317 unsigned int peak_time; /* peak wait time in msec until now */ 318 }; 319 320 /* for the bitmap indicate blocks to be discarded */ 321 struct discard_entry { 322 struct list_head list; /* list head */ 323 block_t start_blkaddr; /* start blockaddr of current segment */ 324 unsigned char discard_map[SIT_VBLOCK_MAP_SIZE]; /* segment discard bitmap */ 325 }; 326 327 /* default discard granularity of inner discard thread, unit: block count */ 328 #define DEFAULT_DISCARD_GRANULARITY 16 329 330 /* max discard pend list number */ 331 #define MAX_PLIST_NUM 512 332 #define plist_idx(blk_num) ((blk_num) >= MAX_PLIST_NUM ? \ 333 (MAX_PLIST_NUM - 1) : ((blk_num) - 1)) 334 335 enum { 336 D_PREP, /* initial */ 337 D_PARTIAL, /* partially submitted */ 338 D_SUBMIT, /* all submitted */ 339 D_DONE, /* finished */ 340 }; 341 342 struct discard_info { 343 block_t lstart; /* logical start address */ 344 block_t len; /* length */ 345 block_t start; /* actual start address in dev */ 346 }; 347 348 struct discard_cmd { 349 struct rb_node rb_node; /* rb node located in rb-tree */ 350 union { 351 struct { 352 block_t lstart; /* logical start address */ 353 block_t len; /* length */ 354 block_t start; /* actual start address in dev */ 355 }; 356 struct discard_info di; /* discard info */ 357 358 }; 359 struct list_head list; /* command list */ 360 struct completion wait; /* compleation */ 361 struct block_device *bdev; /* bdev */ 362 unsigned short ref; /* reference count */ 363 unsigned char state; /* state */ 364 unsigned char queued; /* queued discard */ 365 int error; /* bio error */ 366 spinlock_t lock; /* for state/bio_ref updating */ 367 unsigned short bio_ref; /* bio reference count */ 368 }; 369 370 enum { 371 DPOLICY_BG, 372 DPOLICY_FORCE, 373 DPOLICY_FSTRIM, 374 DPOLICY_UMOUNT, 375 MAX_DPOLICY, 376 }; 377 378 struct discard_policy { 379 int type; /* type of discard */ 380 unsigned int min_interval; /* used for candidates exist */ 381 unsigned int mid_interval; /* used for device busy */ 382 unsigned int max_interval; /* used for candidates not exist */ 383 unsigned int max_requests; /* # of discards issued per round */ 384 unsigned int io_aware_gran; /* minimum granularity discard not be aware of I/O */ 385 bool io_aware; /* issue discard in idle time */ 386 bool sync; /* submit discard with REQ_SYNC flag */ 387 bool ordered; /* issue discard by lba order */ 388 bool timeout; /* discard timeout for put_super */ 389 unsigned int granularity; /* discard granularity */ 390 }; 391 392 struct discard_cmd_control { 393 struct task_struct *f2fs_issue_discard; /* discard thread */ 394 struct list_head entry_list; /* 4KB discard entry list */ 395 struct list_head pend_list[MAX_PLIST_NUM];/* store pending entries */ 396 struct list_head wait_list; /* store on-flushing entries */ 397 struct list_head fstrim_list; /* in-flight discard from fstrim */ 398 wait_queue_head_t discard_wait_queue; /* waiting queue for wake-up */ 399 unsigned int discard_wake; /* to wake up discard thread */ 400 struct mutex cmd_lock; 401 unsigned int nr_discards; /* # of discards in the list */ 402 unsigned int max_discards; /* max. discards to be issued */ 403 unsigned int max_discard_request; /* max. discard request per round */ 404 unsigned int min_discard_issue_time; /* min. interval between discard issue */ 405 unsigned int mid_discard_issue_time; /* mid. interval between discard issue */ 406 unsigned int max_discard_issue_time; /* max. interval between discard issue */ 407 unsigned int discard_granularity; /* discard granularity */ 408 unsigned int undiscard_blks; /* # of undiscard blocks */ 409 unsigned int next_pos; /* next discard position */ 410 atomic_t issued_discard; /* # of issued discard */ 411 atomic_t queued_discard; /* # of queued discard */ 412 atomic_t discard_cmd_cnt; /* # of cached cmd count */ 413 struct rb_root_cached root; /* root of discard rb-tree */ 414 bool rbtree_check; /* config for consistence check */ 415 }; 416 417 /* for the list of fsync inodes, used only during recovery */ 418 struct fsync_inode_entry { 419 struct list_head list; /* list head */ 420 struct inode *inode; /* vfs inode pointer */ 421 block_t blkaddr; /* block address locating the last fsync */ 422 block_t last_dentry; /* block address locating the last dentry */ 423 }; 424 425 #define nats_in_cursum(jnl) (le16_to_cpu((jnl)->n_nats)) 426 #define sits_in_cursum(jnl) (le16_to_cpu((jnl)->n_sits)) 427 428 #define nat_in_journal(jnl, i) ((jnl)->nat_j.entries[i].ne) 429 #define nid_in_journal(jnl, i) ((jnl)->nat_j.entries[i].nid) 430 #define sit_in_journal(jnl, i) ((jnl)->sit_j.entries[i].se) 431 #define segno_in_journal(jnl, i) ((jnl)->sit_j.entries[i].segno) 432 433 #define MAX_NAT_JENTRIES(jnl) (NAT_JOURNAL_ENTRIES - nats_in_cursum(jnl)) 434 #define MAX_SIT_JENTRIES(jnl) (SIT_JOURNAL_ENTRIES - sits_in_cursum(jnl)) 435 436 static inline int update_nats_in_cursum(struct f2fs_journal *journal, int i) 437 { 438 int before = nats_in_cursum(journal); 439 440 journal->n_nats = cpu_to_le16(before + i); 441 return before; 442 } 443 444 static inline int update_sits_in_cursum(struct f2fs_journal *journal, int i) 445 { 446 int before = sits_in_cursum(journal); 447 448 journal->n_sits = cpu_to_le16(before + i); 449 return before; 450 } 451 452 static inline bool __has_cursum_space(struct f2fs_journal *journal, 453 int size, int type) 454 { 455 if (type == NAT_JOURNAL) 456 return size <= MAX_NAT_JENTRIES(journal); 457 return size <= MAX_SIT_JENTRIES(journal); 458 } 459 460 /* for inline stuff */ 461 #define DEF_INLINE_RESERVED_SIZE 1 462 static inline int get_extra_isize(struct inode *inode); 463 static inline int get_inline_xattr_addrs(struct inode *inode); 464 #define MAX_INLINE_DATA(inode) (sizeof(__le32) * \ 465 (CUR_ADDRS_PER_INODE(inode) - \ 466 get_inline_xattr_addrs(inode) - \ 467 DEF_INLINE_RESERVED_SIZE)) 468 469 /* for inline dir */ 470 #define NR_INLINE_DENTRY(inode) (MAX_INLINE_DATA(inode) * BITS_PER_BYTE / \ 471 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \ 472 BITS_PER_BYTE + 1)) 473 #define INLINE_DENTRY_BITMAP_SIZE(inode) \ 474 DIV_ROUND_UP(NR_INLINE_DENTRY(inode), BITS_PER_BYTE) 475 #define INLINE_RESERVED_SIZE(inode) (MAX_INLINE_DATA(inode) - \ 476 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \ 477 NR_INLINE_DENTRY(inode) + \ 478 INLINE_DENTRY_BITMAP_SIZE(inode))) 479 480 /* 481 * For INODE and NODE manager 482 */ 483 /* for directory operations */ 484 485 struct f2fs_filename { 486 /* 487 * The filename the user specified. This is NULL for some 488 * filesystem-internal operations, e.g. converting an inline directory 489 * to a non-inline one, or roll-forward recovering an encrypted dentry. 490 */ 491 const struct qstr *usr_fname; 492 493 /* 494 * The on-disk filename. For encrypted directories, this is encrypted. 495 * This may be NULL for lookups in an encrypted dir without the key. 496 */ 497 struct fscrypt_str disk_name; 498 499 /* The dirhash of this filename */ 500 f2fs_hash_t hash; 501 502 #ifdef CONFIG_FS_ENCRYPTION 503 /* 504 * For lookups in encrypted directories: either the buffer backing 505 * disk_name, or a buffer that holds the decoded no-key name. 506 */ 507 struct fscrypt_str crypto_buf; 508 #endif 509 #ifdef CONFIG_UNICODE 510 /* 511 * For casefolded directories: the casefolded name, but it's left NULL 512 * if the original name is not valid Unicode, if the directory is both 513 * casefolded and encrypted and its encryption key is unavailable, or if 514 * the filesystem is doing an internal operation where usr_fname is also 515 * NULL. In all these cases we fall back to treating the name as an 516 * opaque byte sequence. 517 */ 518 struct fscrypt_str cf_name; 519 #endif 520 }; 521 522 struct f2fs_dentry_ptr { 523 struct inode *inode; 524 void *bitmap; 525 struct f2fs_dir_entry *dentry; 526 __u8 (*filename)[F2FS_SLOT_LEN]; 527 int max; 528 int nr_bitmap; 529 }; 530 531 static inline void make_dentry_ptr_block(struct inode *inode, 532 struct f2fs_dentry_ptr *d, struct f2fs_dentry_block *t) 533 { 534 d->inode = inode; 535 d->max = NR_DENTRY_IN_BLOCK; 536 d->nr_bitmap = SIZE_OF_DENTRY_BITMAP; 537 d->bitmap = t->dentry_bitmap; 538 d->dentry = t->dentry; 539 d->filename = t->filename; 540 } 541 542 static inline void make_dentry_ptr_inline(struct inode *inode, 543 struct f2fs_dentry_ptr *d, void *t) 544 { 545 int entry_cnt = NR_INLINE_DENTRY(inode); 546 int bitmap_size = INLINE_DENTRY_BITMAP_SIZE(inode); 547 int reserved_size = INLINE_RESERVED_SIZE(inode); 548 549 d->inode = inode; 550 d->max = entry_cnt; 551 d->nr_bitmap = bitmap_size; 552 d->bitmap = t; 553 d->dentry = t + bitmap_size + reserved_size; 554 d->filename = t + bitmap_size + reserved_size + 555 SIZE_OF_DIR_ENTRY * entry_cnt; 556 } 557 558 /* 559 * XATTR_NODE_OFFSET stores xattrs to one node block per file keeping -1 560 * as its node offset to distinguish from index node blocks. 561 * But some bits are used to mark the node block. 562 */ 563 #define XATTR_NODE_OFFSET ((((unsigned int)-1) << OFFSET_BIT_SHIFT) \ 564 >> OFFSET_BIT_SHIFT) 565 enum { 566 ALLOC_NODE, /* allocate a new node page if needed */ 567 LOOKUP_NODE, /* look up a node without readahead */ 568 LOOKUP_NODE_RA, /* 569 * look up a node with readahead called 570 * by get_data_block. 571 */ 572 }; 573 574 #define DEFAULT_RETRY_IO_COUNT 8 /* maximum retry read IO or flush count */ 575 576 /* congestion wait timeout value, default: 20ms */ 577 #define DEFAULT_IO_TIMEOUT (msecs_to_jiffies(20)) 578 579 /* maximum retry quota flush count */ 580 #define DEFAULT_RETRY_QUOTA_FLUSH_COUNT 8 581 582 /* maximum retry of EIO'ed meta page */ 583 #define MAX_RETRY_META_PAGE_EIO 100 584 585 #define F2FS_LINK_MAX 0xffffffff /* maximum link count per file */ 586 587 #define MAX_DIR_RA_PAGES 4 /* maximum ra pages of dir */ 588 589 /* dirty segments threshold for triggering CP */ 590 #define DEFAULT_DIRTY_THRESHOLD 4 591 592 /* for in-memory extent cache entry */ 593 #define F2FS_MIN_EXTENT_LEN 64 /* minimum extent length */ 594 595 /* number of extent info in extent cache we try to shrink */ 596 #define EXTENT_CACHE_SHRINK_NUMBER 128 597 598 #define RECOVERY_MAX_RA_BLOCKS BIO_MAX_VECS 599 #define RECOVERY_MIN_RA_BLOCKS 1 600 601 struct rb_entry { 602 struct rb_node rb_node; /* rb node located in rb-tree */ 603 union { 604 struct { 605 unsigned int ofs; /* start offset of the entry */ 606 unsigned int len; /* length of the entry */ 607 }; 608 unsigned long long key; /* 64-bits key */ 609 } __packed; 610 }; 611 612 struct extent_info { 613 unsigned int fofs; /* start offset in a file */ 614 unsigned int len; /* length of the extent */ 615 u32 blk; /* start block address of the extent */ 616 #ifdef CONFIG_F2FS_FS_COMPRESSION 617 unsigned int c_len; /* physical extent length of compressed blocks */ 618 #endif 619 }; 620 621 struct extent_node { 622 struct rb_node rb_node; /* rb node located in rb-tree */ 623 struct extent_info ei; /* extent info */ 624 struct list_head list; /* node in global extent list of sbi */ 625 struct extent_tree *et; /* extent tree pointer */ 626 }; 627 628 struct extent_tree { 629 nid_t ino; /* inode number */ 630 struct rb_root_cached root; /* root of extent info rb-tree */ 631 struct extent_node *cached_en; /* recently accessed extent node */ 632 struct extent_info largest; /* largested extent info */ 633 struct list_head list; /* to be used by sbi->zombie_list */ 634 rwlock_t lock; /* protect extent info rb-tree */ 635 atomic_t node_cnt; /* # of extent node in rb-tree*/ 636 bool largest_updated; /* largest extent updated */ 637 }; 638 639 /* 640 * This structure is taken from ext4_map_blocks. 641 * 642 * Note that, however, f2fs uses NEW and MAPPED flags for f2fs_map_blocks(). 643 */ 644 #define F2FS_MAP_NEW (1 << BH_New) 645 #define F2FS_MAP_MAPPED (1 << BH_Mapped) 646 #define F2FS_MAP_UNWRITTEN (1 << BH_Unwritten) 647 #define F2FS_MAP_FLAGS (F2FS_MAP_NEW | F2FS_MAP_MAPPED |\ 648 F2FS_MAP_UNWRITTEN) 649 650 struct f2fs_map_blocks { 651 struct block_device *m_bdev; /* for multi-device dio */ 652 block_t m_pblk; 653 block_t m_lblk; 654 unsigned int m_len; 655 unsigned int m_flags; 656 pgoff_t *m_next_pgofs; /* point next possible non-hole pgofs */ 657 pgoff_t *m_next_extent; /* point to next possible extent */ 658 int m_seg_type; 659 bool m_may_create; /* indicate it is from write path */ 660 bool m_multidev_dio; /* indicate it allows multi-device dio */ 661 }; 662 663 /* for flag in get_data_block */ 664 enum { 665 F2FS_GET_BLOCK_DEFAULT, 666 F2FS_GET_BLOCK_FIEMAP, 667 F2FS_GET_BLOCK_BMAP, 668 F2FS_GET_BLOCK_DIO, 669 F2FS_GET_BLOCK_PRE_DIO, 670 F2FS_GET_BLOCK_PRE_AIO, 671 F2FS_GET_BLOCK_PRECACHE, 672 }; 673 674 /* 675 * i_advise uses FADVISE_XXX_BIT. We can add additional hints later. 676 */ 677 #define FADVISE_COLD_BIT 0x01 678 #define FADVISE_LOST_PINO_BIT 0x02 679 #define FADVISE_ENCRYPT_BIT 0x04 680 #define FADVISE_ENC_NAME_BIT 0x08 681 #define FADVISE_KEEP_SIZE_BIT 0x10 682 #define FADVISE_HOT_BIT 0x20 683 #define FADVISE_VERITY_BIT 0x40 684 #define FADVISE_TRUNC_BIT 0x80 685 686 #define FADVISE_MODIFIABLE_BITS (FADVISE_COLD_BIT | FADVISE_HOT_BIT) 687 688 #define file_is_cold(inode) is_file(inode, FADVISE_COLD_BIT) 689 #define file_set_cold(inode) set_file(inode, FADVISE_COLD_BIT) 690 #define file_clear_cold(inode) clear_file(inode, FADVISE_COLD_BIT) 691 692 #define file_wrong_pino(inode) is_file(inode, FADVISE_LOST_PINO_BIT) 693 #define file_lost_pino(inode) set_file(inode, FADVISE_LOST_PINO_BIT) 694 #define file_got_pino(inode) clear_file(inode, FADVISE_LOST_PINO_BIT) 695 696 #define file_is_encrypt(inode) is_file(inode, FADVISE_ENCRYPT_BIT) 697 #define file_set_encrypt(inode) set_file(inode, FADVISE_ENCRYPT_BIT) 698 699 #define file_enc_name(inode) is_file(inode, FADVISE_ENC_NAME_BIT) 700 #define file_set_enc_name(inode) set_file(inode, FADVISE_ENC_NAME_BIT) 701 702 #define file_keep_isize(inode) is_file(inode, FADVISE_KEEP_SIZE_BIT) 703 #define file_set_keep_isize(inode) set_file(inode, FADVISE_KEEP_SIZE_BIT) 704 705 #define file_is_hot(inode) is_file(inode, FADVISE_HOT_BIT) 706 #define file_set_hot(inode) set_file(inode, FADVISE_HOT_BIT) 707 #define file_clear_hot(inode) clear_file(inode, FADVISE_HOT_BIT) 708 709 #define file_is_verity(inode) is_file(inode, FADVISE_VERITY_BIT) 710 #define file_set_verity(inode) set_file(inode, FADVISE_VERITY_BIT) 711 712 #define file_should_truncate(inode) is_file(inode, FADVISE_TRUNC_BIT) 713 #define file_need_truncate(inode) set_file(inode, FADVISE_TRUNC_BIT) 714 #define file_dont_truncate(inode) clear_file(inode, FADVISE_TRUNC_BIT) 715 716 #define DEF_DIR_LEVEL 0 717 718 enum { 719 GC_FAILURE_PIN, 720 GC_FAILURE_ATOMIC, 721 MAX_GC_FAILURE 722 }; 723 724 /* used for f2fs_inode_info->flags */ 725 enum { 726 FI_NEW_INODE, /* indicate newly allocated inode */ 727 FI_DIRTY_INODE, /* indicate inode is dirty or not */ 728 FI_AUTO_RECOVER, /* indicate inode is recoverable */ 729 FI_DIRTY_DIR, /* indicate directory has dirty pages */ 730 FI_INC_LINK, /* need to increment i_nlink */ 731 FI_ACL_MODE, /* indicate acl mode */ 732 FI_NO_ALLOC, /* should not allocate any blocks */ 733 FI_FREE_NID, /* free allocated nide */ 734 FI_NO_EXTENT, /* not to use the extent cache */ 735 FI_INLINE_XATTR, /* used for inline xattr */ 736 FI_INLINE_DATA, /* used for inline data*/ 737 FI_INLINE_DENTRY, /* used for inline dentry */ 738 FI_APPEND_WRITE, /* inode has appended data */ 739 FI_UPDATE_WRITE, /* inode has in-place-update data */ 740 FI_NEED_IPU, /* used for ipu per file */ 741 FI_ATOMIC_FILE, /* indicate atomic file */ 742 FI_ATOMIC_COMMIT, /* indicate the state of atomical committing */ 743 FI_VOLATILE_FILE, /* indicate volatile file */ 744 FI_FIRST_BLOCK_WRITTEN, /* indicate #0 data block was written */ 745 FI_DROP_CACHE, /* drop dirty page cache */ 746 FI_DATA_EXIST, /* indicate data exists */ 747 FI_INLINE_DOTS, /* indicate inline dot dentries */ 748 FI_SKIP_WRITES, /* should skip data page writeback */ 749 FI_OPU_WRITE, /* used for opu per file */ 750 FI_DIRTY_FILE, /* indicate regular/symlink has dirty pages */ 751 FI_PREALLOCATED_ALL, /* all blocks for write were preallocated */ 752 FI_HOT_DATA, /* indicate file is hot */ 753 FI_EXTRA_ATTR, /* indicate file has extra attribute */ 754 FI_PROJ_INHERIT, /* indicate file inherits projectid */ 755 FI_PIN_FILE, /* indicate file should not be gced */ 756 FI_ATOMIC_REVOKE_REQUEST, /* request to drop atomic data */ 757 FI_VERITY_IN_PROGRESS, /* building fs-verity Merkle tree */ 758 FI_COMPRESSED_FILE, /* indicate file's data can be compressed */ 759 FI_COMPRESS_CORRUPT, /* indicate compressed cluster is corrupted */ 760 FI_MMAP_FILE, /* indicate file was mmapped */ 761 FI_ENABLE_COMPRESS, /* enable compression in "user" compression mode */ 762 FI_COMPRESS_RELEASED, /* compressed blocks were released */ 763 FI_ALIGNED_WRITE, /* enable aligned write */ 764 FI_MAX, /* max flag, never be used */ 765 }; 766 767 struct f2fs_inode_info { 768 struct inode vfs_inode; /* serve a vfs inode */ 769 unsigned long i_flags; /* keep an inode flags for ioctl */ 770 unsigned char i_advise; /* use to give file attribute hints */ 771 unsigned char i_dir_level; /* use for dentry level for large dir */ 772 unsigned int i_current_depth; /* only for directory depth */ 773 /* for gc failure statistic */ 774 unsigned int i_gc_failures[MAX_GC_FAILURE]; 775 unsigned int i_pino; /* parent inode number */ 776 umode_t i_acl_mode; /* keep file acl mode temporarily */ 777 778 /* Use below internally in f2fs*/ 779 unsigned long flags[BITS_TO_LONGS(FI_MAX)]; /* use to pass per-file flags */ 780 struct f2fs_rwsem i_sem; /* protect fi info */ 781 atomic_t dirty_pages; /* # of dirty pages */ 782 f2fs_hash_t chash; /* hash value of given file name */ 783 unsigned int clevel; /* maximum level of given file name */ 784 struct task_struct *task; /* lookup and create consistency */ 785 struct task_struct *cp_task; /* separate cp/wb IO stats*/ 786 nid_t i_xattr_nid; /* node id that contains xattrs */ 787 loff_t last_disk_size; /* lastly written file size */ 788 spinlock_t i_size_lock; /* protect last_disk_size */ 789 790 #ifdef CONFIG_QUOTA 791 struct dquot *i_dquot[MAXQUOTAS]; 792 793 /* quota space reservation, managed internally by quota code */ 794 qsize_t i_reserved_quota; 795 #endif 796 struct list_head dirty_list; /* dirty list for dirs and files */ 797 struct list_head gdirty_list; /* linked in global dirty list */ 798 struct list_head inmem_ilist; /* list for inmem inodes */ 799 struct list_head inmem_pages; /* inmemory pages managed by f2fs */ 800 struct task_struct *inmem_task; /* store inmemory task */ 801 struct mutex inmem_lock; /* lock for inmemory pages */ 802 struct extent_tree *extent_tree; /* cached extent_tree entry */ 803 804 /* avoid racing between foreground op and gc */ 805 struct f2fs_rwsem i_gc_rwsem[2]; 806 struct f2fs_rwsem i_xattr_sem; /* avoid racing between reading and changing EAs */ 807 808 int i_extra_isize; /* size of extra space located in i_addr */ 809 kprojid_t i_projid; /* id for project quota */ 810 int i_inline_xattr_size; /* inline xattr size */ 811 struct timespec64 i_crtime; /* inode creation time */ 812 struct timespec64 i_disk_time[4];/* inode disk times */ 813 814 /* for file compress */ 815 atomic_t i_compr_blocks; /* # of compressed blocks */ 816 unsigned char i_compress_algorithm; /* algorithm type */ 817 unsigned char i_log_cluster_size; /* log of cluster size */ 818 unsigned char i_compress_level; /* compress level (lz4hc,zstd) */ 819 unsigned short i_compress_flag; /* compress flag */ 820 unsigned int i_cluster_size; /* cluster size */ 821 }; 822 823 static inline void get_extent_info(struct extent_info *ext, 824 struct f2fs_extent *i_ext) 825 { 826 ext->fofs = le32_to_cpu(i_ext->fofs); 827 ext->blk = le32_to_cpu(i_ext->blk); 828 ext->len = le32_to_cpu(i_ext->len); 829 } 830 831 static inline void set_raw_extent(struct extent_info *ext, 832 struct f2fs_extent *i_ext) 833 { 834 i_ext->fofs = cpu_to_le32(ext->fofs); 835 i_ext->blk = cpu_to_le32(ext->blk); 836 i_ext->len = cpu_to_le32(ext->len); 837 } 838 839 static inline void set_extent_info(struct extent_info *ei, unsigned int fofs, 840 u32 blk, unsigned int len) 841 { 842 ei->fofs = fofs; 843 ei->blk = blk; 844 ei->len = len; 845 #ifdef CONFIG_F2FS_FS_COMPRESSION 846 ei->c_len = 0; 847 #endif 848 } 849 850 static inline bool __is_discard_mergeable(struct discard_info *back, 851 struct discard_info *front, unsigned int max_len) 852 { 853 return (back->lstart + back->len == front->lstart) && 854 (back->len + front->len <= max_len); 855 } 856 857 static inline bool __is_discard_back_mergeable(struct discard_info *cur, 858 struct discard_info *back, unsigned int max_len) 859 { 860 return __is_discard_mergeable(back, cur, max_len); 861 } 862 863 static inline bool __is_discard_front_mergeable(struct discard_info *cur, 864 struct discard_info *front, unsigned int max_len) 865 { 866 return __is_discard_mergeable(cur, front, max_len); 867 } 868 869 static inline bool __is_extent_mergeable(struct extent_info *back, 870 struct extent_info *front) 871 { 872 #ifdef CONFIG_F2FS_FS_COMPRESSION 873 if (back->c_len && back->len != back->c_len) 874 return false; 875 if (front->c_len && front->len != front->c_len) 876 return false; 877 #endif 878 return (back->fofs + back->len == front->fofs && 879 back->blk + back->len == front->blk); 880 } 881 882 static inline bool __is_back_mergeable(struct extent_info *cur, 883 struct extent_info *back) 884 { 885 return __is_extent_mergeable(back, cur); 886 } 887 888 static inline bool __is_front_mergeable(struct extent_info *cur, 889 struct extent_info *front) 890 { 891 return __is_extent_mergeable(cur, front); 892 } 893 894 extern void f2fs_mark_inode_dirty_sync(struct inode *inode, bool sync); 895 static inline void __try_update_largest_extent(struct extent_tree *et, 896 struct extent_node *en) 897 { 898 if (en->ei.len > et->largest.len) { 899 et->largest = en->ei; 900 et->largest_updated = true; 901 } 902 } 903 904 /* 905 * For free nid management 906 */ 907 enum nid_state { 908 FREE_NID, /* newly added to free nid list */ 909 PREALLOC_NID, /* it is preallocated */ 910 MAX_NID_STATE, 911 }; 912 913 enum nat_state { 914 TOTAL_NAT, 915 DIRTY_NAT, 916 RECLAIMABLE_NAT, 917 MAX_NAT_STATE, 918 }; 919 920 struct f2fs_nm_info { 921 block_t nat_blkaddr; /* base disk address of NAT */ 922 nid_t max_nid; /* maximum possible node ids */ 923 nid_t available_nids; /* # of available node ids */ 924 nid_t next_scan_nid; /* the next nid to be scanned */ 925 nid_t max_rf_node_blocks; /* max # of nodes for recovery */ 926 unsigned int ram_thresh; /* control the memory footprint */ 927 unsigned int ra_nid_pages; /* # of nid pages to be readaheaded */ 928 unsigned int dirty_nats_ratio; /* control dirty nats ratio threshold */ 929 930 /* NAT cache management */ 931 struct radix_tree_root nat_root;/* root of the nat entry cache */ 932 struct radix_tree_root nat_set_root;/* root of the nat set cache */ 933 struct f2fs_rwsem nat_tree_lock; /* protect nat entry tree */ 934 struct list_head nat_entries; /* cached nat entry list (clean) */ 935 spinlock_t nat_list_lock; /* protect clean nat entry list */ 936 unsigned int nat_cnt[MAX_NAT_STATE]; /* the # of cached nat entries */ 937 unsigned int nat_blocks; /* # of nat blocks */ 938 939 /* free node ids management */ 940 struct radix_tree_root free_nid_root;/* root of the free_nid cache */ 941 struct list_head free_nid_list; /* list for free nids excluding preallocated nids */ 942 unsigned int nid_cnt[MAX_NID_STATE]; /* the number of free node id */ 943 spinlock_t nid_list_lock; /* protect nid lists ops */ 944 struct mutex build_lock; /* lock for build free nids */ 945 unsigned char **free_nid_bitmap; 946 unsigned char *nat_block_bitmap; 947 unsigned short *free_nid_count; /* free nid count of NAT block */ 948 949 /* for checkpoint */ 950 char *nat_bitmap; /* NAT bitmap pointer */ 951 952 unsigned int nat_bits_blocks; /* # of nat bits blocks */ 953 unsigned char *nat_bits; /* NAT bits blocks */ 954 unsigned char *full_nat_bits; /* full NAT pages */ 955 unsigned char *empty_nat_bits; /* empty NAT pages */ 956 #ifdef CONFIG_F2FS_CHECK_FS 957 char *nat_bitmap_mir; /* NAT bitmap mirror */ 958 #endif 959 int bitmap_size; /* bitmap size */ 960 }; 961 962 /* 963 * this structure is used as one of function parameters. 964 * all the information are dedicated to a given direct node block determined 965 * by the data offset in a file. 966 */ 967 struct dnode_of_data { 968 struct inode *inode; /* vfs inode pointer */ 969 struct page *inode_page; /* its inode page, NULL is possible */ 970 struct page *node_page; /* cached direct node page */ 971 nid_t nid; /* node id of the direct node block */ 972 unsigned int ofs_in_node; /* data offset in the node page */ 973 bool inode_page_locked; /* inode page is locked or not */ 974 bool node_changed; /* is node block changed */ 975 char cur_level; /* level of hole node page */ 976 char max_level; /* level of current page located */ 977 block_t data_blkaddr; /* block address of the node block */ 978 }; 979 980 static inline void set_new_dnode(struct dnode_of_data *dn, struct inode *inode, 981 struct page *ipage, struct page *npage, nid_t nid) 982 { 983 memset(dn, 0, sizeof(*dn)); 984 dn->inode = inode; 985 dn->inode_page = ipage; 986 dn->node_page = npage; 987 dn->nid = nid; 988 } 989 990 /* 991 * For SIT manager 992 * 993 * By default, there are 6 active log areas across the whole main area. 994 * When considering hot and cold data separation to reduce cleaning overhead, 995 * we split 3 for data logs and 3 for node logs as hot, warm, and cold types, 996 * respectively. 997 * In the current design, you should not change the numbers intentionally. 998 * Instead, as a mount option such as active_logs=x, you can use 2, 4, and 6 999 * logs individually according to the underlying devices. (default: 6) 1000 * Just in case, on-disk layout covers maximum 16 logs that consist of 8 for 1001 * data and 8 for node logs. 1002 */ 1003 #define NR_CURSEG_DATA_TYPE (3) 1004 #define NR_CURSEG_NODE_TYPE (3) 1005 #define NR_CURSEG_INMEM_TYPE (2) 1006 #define NR_CURSEG_RO_TYPE (2) 1007 #define NR_CURSEG_PERSIST_TYPE (NR_CURSEG_DATA_TYPE + NR_CURSEG_NODE_TYPE) 1008 #define NR_CURSEG_TYPE (NR_CURSEG_INMEM_TYPE + NR_CURSEG_PERSIST_TYPE) 1009 1010 enum { 1011 CURSEG_HOT_DATA = 0, /* directory entry blocks */ 1012 CURSEG_WARM_DATA, /* data blocks */ 1013 CURSEG_COLD_DATA, /* multimedia or GCed data blocks */ 1014 CURSEG_HOT_NODE, /* direct node blocks of directory files */ 1015 CURSEG_WARM_NODE, /* direct node blocks of normal files */ 1016 CURSEG_COLD_NODE, /* indirect node blocks */ 1017 NR_PERSISTENT_LOG, /* number of persistent log */ 1018 CURSEG_COLD_DATA_PINNED = NR_PERSISTENT_LOG, 1019 /* pinned file that needs consecutive block address */ 1020 CURSEG_ALL_DATA_ATGC, /* SSR alloctor in hot/warm/cold data area */ 1021 NO_CHECK_TYPE, /* number of persistent & inmem log */ 1022 }; 1023 1024 struct flush_cmd { 1025 struct completion wait; 1026 struct llist_node llnode; 1027 nid_t ino; 1028 int ret; 1029 }; 1030 1031 struct flush_cmd_control { 1032 struct task_struct *f2fs_issue_flush; /* flush thread */ 1033 wait_queue_head_t flush_wait_queue; /* waiting queue for wake-up */ 1034 atomic_t issued_flush; /* # of issued flushes */ 1035 atomic_t queued_flush; /* # of queued flushes */ 1036 struct llist_head issue_list; /* list for command issue */ 1037 struct llist_node *dispatch_list; /* list for command dispatch */ 1038 }; 1039 1040 struct f2fs_sm_info { 1041 struct sit_info *sit_info; /* whole segment information */ 1042 struct free_segmap_info *free_info; /* free segment information */ 1043 struct dirty_seglist_info *dirty_info; /* dirty segment information */ 1044 struct curseg_info *curseg_array; /* active segment information */ 1045 1046 struct f2fs_rwsem curseg_lock; /* for preventing curseg change */ 1047 1048 block_t seg0_blkaddr; /* block address of 0'th segment */ 1049 block_t main_blkaddr; /* start block address of main area */ 1050 block_t ssa_blkaddr; /* start block address of SSA area */ 1051 1052 unsigned int segment_count; /* total # of segments */ 1053 unsigned int main_segments; /* # of segments in main area */ 1054 unsigned int reserved_segments; /* # of reserved segments */ 1055 unsigned int additional_reserved_segments;/* reserved segs for IO align feature */ 1056 unsigned int ovp_segments; /* # of overprovision segments */ 1057 1058 /* a threshold to reclaim prefree segments */ 1059 unsigned int rec_prefree_segments; 1060 1061 /* for batched trimming */ 1062 unsigned int trim_sections; /* # of sections to trim */ 1063 1064 struct list_head sit_entry_set; /* sit entry set list */ 1065 1066 unsigned int ipu_policy; /* in-place-update policy */ 1067 unsigned int min_ipu_util; /* in-place-update threshold */ 1068 unsigned int min_fsync_blocks; /* threshold for fsync */ 1069 unsigned int min_seq_blocks; /* threshold for sequential blocks */ 1070 unsigned int min_hot_blocks; /* threshold for hot block allocation */ 1071 unsigned int min_ssr_sections; /* threshold to trigger SSR allocation */ 1072 1073 /* for flush command control */ 1074 struct flush_cmd_control *fcc_info; 1075 1076 /* for discard command control */ 1077 struct discard_cmd_control *dcc_info; 1078 }; 1079 1080 /* 1081 * For superblock 1082 */ 1083 /* 1084 * COUNT_TYPE for monitoring 1085 * 1086 * f2fs monitors the number of several block types such as on-writeback, 1087 * dirty dentry blocks, dirty node blocks, and dirty meta blocks. 1088 */ 1089 #define WB_DATA_TYPE(p) (__is_cp_guaranteed(p) ? F2FS_WB_CP_DATA : F2FS_WB_DATA) 1090 enum count_type { 1091 F2FS_DIRTY_DENTS, 1092 F2FS_DIRTY_DATA, 1093 F2FS_DIRTY_QDATA, 1094 F2FS_DIRTY_NODES, 1095 F2FS_DIRTY_META, 1096 F2FS_INMEM_PAGES, 1097 F2FS_DIRTY_IMETA, 1098 F2FS_WB_CP_DATA, 1099 F2FS_WB_DATA, 1100 F2FS_RD_DATA, 1101 F2FS_RD_NODE, 1102 F2FS_RD_META, 1103 F2FS_DIO_WRITE, 1104 F2FS_DIO_READ, 1105 NR_COUNT_TYPE, 1106 }; 1107 1108 /* 1109 * The below are the page types of bios used in submit_bio(). 1110 * The available types are: 1111 * DATA User data pages. It operates as async mode. 1112 * NODE Node pages. It operates as async mode. 1113 * META FS metadata pages such as SIT, NAT, CP. 1114 * NR_PAGE_TYPE The number of page types. 1115 * META_FLUSH Make sure the previous pages are written 1116 * with waiting the bio's completion 1117 * ... Only can be used with META. 1118 */ 1119 #define PAGE_TYPE_OF_BIO(type) ((type) > META ? META : (type)) 1120 enum page_type { 1121 DATA, 1122 NODE, 1123 META, 1124 NR_PAGE_TYPE, 1125 META_FLUSH, 1126 INMEM, /* the below types are used by tracepoints only. */ 1127 INMEM_DROP, 1128 INMEM_INVALIDATE, 1129 INMEM_REVOKE, 1130 IPU, 1131 OPU, 1132 }; 1133 1134 enum temp_type { 1135 HOT = 0, /* must be zero for meta bio */ 1136 WARM, 1137 COLD, 1138 NR_TEMP_TYPE, 1139 }; 1140 1141 enum need_lock_type { 1142 LOCK_REQ = 0, 1143 LOCK_DONE, 1144 LOCK_RETRY, 1145 }; 1146 1147 enum cp_reason_type { 1148 CP_NO_NEEDED, 1149 CP_NON_REGULAR, 1150 CP_COMPRESSED, 1151 CP_HARDLINK, 1152 CP_SB_NEED_CP, 1153 CP_WRONG_PINO, 1154 CP_NO_SPC_ROLL, 1155 CP_NODE_NEED_CP, 1156 CP_FASTBOOT_MODE, 1157 CP_SPEC_LOG_NUM, 1158 CP_RECOVER_DIR, 1159 }; 1160 1161 enum iostat_type { 1162 /* WRITE IO */ 1163 APP_DIRECT_IO, /* app direct write IOs */ 1164 APP_BUFFERED_IO, /* app buffered write IOs */ 1165 APP_WRITE_IO, /* app write IOs */ 1166 APP_MAPPED_IO, /* app mapped IOs */ 1167 FS_DATA_IO, /* data IOs from kworker/fsync/reclaimer */ 1168 FS_NODE_IO, /* node IOs from kworker/fsync/reclaimer */ 1169 FS_META_IO, /* meta IOs from kworker/reclaimer */ 1170 FS_GC_DATA_IO, /* data IOs from forground gc */ 1171 FS_GC_NODE_IO, /* node IOs from forground gc */ 1172 FS_CP_DATA_IO, /* data IOs from checkpoint */ 1173 FS_CP_NODE_IO, /* node IOs from checkpoint */ 1174 FS_CP_META_IO, /* meta IOs from checkpoint */ 1175 1176 /* READ IO */ 1177 APP_DIRECT_READ_IO, /* app direct read IOs */ 1178 APP_BUFFERED_READ_IO, /* app buffered read IOs */ 1179 APP_READ_IO, /* app read IOs */ 1180 APP_MAPPED_READ_IO, /* app mapped read IOs */ 1181 FS_DATA_READ_IO, /* data read IOs */ 1182 FS_GDATA_READ_IO, /* data read IOs from background gc */ 1183 FS_CDATA_READ_IO, /* compressed data read IOs */ 1184 FS_NODE_READ_IO, /* node read IOs */ 1185 FS_META_READ_IO, /* meta read IOs */ 1186 1187 /* other */ 1188 FS_DISCARD, /* discard */ 1189 NR_IO_TYPE, 1190 }; 1191 1192 struct f2fs_io_info { 1193 struct f2fs_sb_info *sbi; /* f2fs_sb_info pointer */ 1194 nid_t ino; /* inode number */ 1195 enum page_type type; /* contains DATA/NODE/META/META_FLUSH */ 1196 enum temp_type temp; /* contains HOT/WARM/COLD */ 1197 int op; /* contains REQ_OP_ */ 1198 int op_flags; /* req_flag_bits */ 1199 block_t new_blkaddr; /* new block address to be written */ 1200 block_t old_blkaddr; /* old block address before Cow */ 1201 struct page *page; /* page to be written */ 1202 struct page *encrypted_page; /* encrypted page */ 1203 struct page *compressed_page; /* compressed page */ 1204 struct list_head list; /* serialize IOs */ 1205 bool submitted; /* indicate IO submission */ 1206 int need_lock; /* indicate we need to lock cp_rwsem */ 1207 bool in_list; /* indicate fio is in io_list */ 1208 bool is_por; /* indicate IO is from recovery or not */ 1209 bool retry; /* need to reallocate block address */ 1210 int compr_blocks; /* # of compressed block addresses */ 1211 bool encrypted; /* indicate file is encrypted */ 1212 enum iostat_type io_type; /* io type */ 1213 struct writeback_control *io_wbc; /* writeback control */ 1214 struct bio **bio; /* bio for ipu */ 1215 sector_t *last_block; /* last block number in bio */ 1216 unsigned char version; /* version of the node */ 1217 }; 1218 1219 struct bio_entry { 1220 struct bio *bio; 1221 struct list_head list; 1222 }; 1223 1224 #define is_read_io(rw) ((rw) == READ) 1225 struct f2fs_bio_info { 1226 struct f2fs_sb_info *sbi; /* f2fs superblock */ 1227 struct bio *bio; /* bios to merge */ 1228 sector_t last_block_in_bio; /* last block number */ 1229 struct f2fs_io_info fio; /* store buffered io info. */ 1230 struct f2fs_rwsem io_rwsem; /* blocking op for bio */ 1231 spinlock_t io_lock; /* serialize DATA/NODE IOs */ 1232 struct list_head io_list; /* track fios */ 1233 struct list_head bio_list; /* bio entry list head */ 1234 struct f2fs_rwsem bio_list_lock; /* lock to protect bio entry list */ 1235 }; 1236 1237 #define FDEV(i) (sbi->devs[i]) 1238 #define RDEV(i) (raw_super->devs[i]) 1239 struct f2fs_dev_info { 1240 struct block_device *bdev; 1241 char path[MAX_PATH_LEN]; 1242 unsigned int total_segments; 1243 block_t start_blk; 1244 block_t end_blk; 1245 #ifdef CONFIG_BLK_DEV_ZONED 1246 unsigned int nr_blkz; /* Total number of zones */ 1247 unsigned long *blkz_seq; /* Bitmap indicating sequential zones */ 1248 block_t *zone_capacity_blocks; /* Array of zone capacity in blks */ 1249 #endif 1250 }; 1251 1252 enum inode_type { 1253 DIR_INODE, /* for dirty dir inode */ 1254 FILE_INODE, /* for dirty regular/symlink inode */ 1255 DIRTY_META, /* for all dirtied inode metadata */ 1256 ATOMIC_FILE, /* for all atomic files */ 1257 NR_INODE_TYPE, 1258 }; 1259 1260 /* for inner inode cache management */ 1261 struct inode_management { 1262 struct radix_tree_root ino_root; /* ino entry array */ 1263 spinlock_t ino_lock; /* for ino entry lock */ 1264 struct list_head ino_list; /* inode list head */ 1265 unsigned long ino_num; /* number of entries */ 1266 }; 1267 1268 /* for GC_AT */ 1269 struct atgc_management { 1270 bool atgc_enabled; /* ATGC is enabled or not */ 1271 struct rb_root_cached root; /* root of victim rb-tree */ 1272 struct list_head victim_list; /* linked with all victim entries */ 1273 unsigned int victim_count; /* victim count in rb-tree */ 1274 unsigned int candidate_ratio; /* candidate ratio */ 1275 unsigned int max_candidate_count; /* max candidate count */ 1276 unsigned int age_weight; /* age weight, vblock_weight = 100 - age_weight */ 1277 unsigned long long age_threshold; /* age threshold */ 1278 }; 1279 1280 /* For s_flag in struct f2fs_sb_info */ 1281 enum { 1282 SBI_IS_DIRTY, /* dirty flag for checkpoint */ 1283 SBI_IS_CLOSE, /* specify unmounting */ 1284 SBI_NEED_FSCK, /* need fsck.f2fs to fix */ 1285 SBI_POR_DOING, /* recovery is doing or not */ 1286 SBI_NEED_SB_WRITE, /* need to recover superblock */ 1287 SBI_NEED_CP, /* need to checkpoint */ 1288 SBI_IS_SHUTDOWN, /* shutdown by ioctl */ 1289 SBI_IS_RECOVERED, /* recovered orphan/data */ 1290 SBI_CP_DISABLED, /* CP was disabled last mount */ 1291 SBI_CP_DISABLED_QUICK, /* CP was disabled quickly */ 1292 SBI_QUOTA_NEED_FLUSH, /* need to flush quota info in CP */ 1293 SBI_QUOTA_SKIP_FLUSH, /* skip flushing quota in current CP */ 1294 SBI_QUOTA_NEED_REPAIR, /* quota file may be corrupted */ 1295 SBI_IS_RESIZEFS, /* resizefs is in process */ 1296 }; 1297 1298 enum { 1299 CP_TIME, 1300 REQ_TIME, 1301 DISCARD_TIME, 1302 GC_TIME, 1303 DISABLE_TIME, 1304 UMOUNT_DISCARD_TIMEOUT, 1305 MAX_TIME, 1306 }; 1307 1308 enum { 1309 GC_NORMAL, 1310 GC_IDLE_CB, 1311 GC_IDLE_GREEDY, 1312 GC_IDLE_AT, 1313 GC_URGENT_HIGH, 1314 GC_URGENT_LOW, 1315 MAX_GC_MODE, 1316 }; 1317 1318 enum { 1319 BGGC_MODE_ON, /* background gc is on */ 1320 BGGC_MODE_OFF, /* background gc is off */ 1321 BGGC_MODE_SYNC, /* 1322 * background gc is on, migrating blocks 1323 * like foreground gc 1324 */ 1325 }; 1326 1327 enum { 1328 FS_MODE_ADAPTIVE, /* use both lfs/ssr allocation */ 1329 FS_MODE_LFS, /* use lfs allocation only */ 1330 FS_MODE_FRAGMENT_SEG, /* segment fragmentation mode */ 1331 FS_MODE_FRAGMENT_BLK, /* block fragmentation mode */ 1332 }; 1333 1334 enum { 1335 WHINT_MODE_OFF, /* not pass down write hints */ 1336 WHINT_MODE_USER, /* try to pass down hints given by users */ 1337 WHINT_MODE_FS, /* pass down hints with F2FS policy */ 1338 }; 1339 1340 enum { 1341 ALLOC_MODE_DEFAULT, /* stay default */ 1342 ALLOC_MODE_REUSE, /* reuse segments as much as possible */ 1343 }; 1344 1345 enum fsync_mode { 1346 FSYNC_MODE_POSIX, /* fsync follows posix semantics */ 1347 FSYNC_MODE_STRICT, /* fsync behaves in line with ext4 */ 1348 FSYNC_MODE_NOBARRIER, /* fsync behaves nobarrier based on posix */ 1349 }; 1350 1351 enum { 1352 COMPR_MODE_FS, /* 1353 * automatically compress compression 1354 * enabled files 1355 */ 1356 COMPR_MODE_USER, /* 1357 * automatical compression is disabled. 1358 * user can control the file compression 1359 * using ioctls 1360 */ 1361 }; 1362 1363 enum { 1364 DISCARD_UNIT_BLOCK, /* basic discard unit is block */ 1365 DISCARD_UNIT_SEGMENT, /* basic discard unit is segment */ 1366 DISCARD_UNIT_SECTION, /* basic discard unit is section */ 1367 }; 1368 1369 static inline int f2fs_test_bit(unsigned int nr, char *addr); 1370 static inline void f2fs_set_bit(unsigned int nr, char *addr); 1371 static inline void f2fs_clear_bit(unsigned int nr, char *addr); 1372 1373 /* 1374 * Layout of f2fs page.private: 1375 * 1376 * Layout A: lowest bit should be 1 1377 * | bit0 = 1 | bit1 | bit2 | ... | bit MAX | private data .... | 1378 * bit 0 PAGE_PRIVATE_NOT_POINTER 1379 * bit 1 PAGE_PRIVATE_ATOMIC_WRITE 1380 * bit 2 PAGE_PRIVATE_DUMMY_WRITE 1381 * bit 3 PAGE_PRIVATE_ONGOING_MIGRATION 1382 * bit 4 PAGE_PRIVATE_INLINE_INODE 1383 * bit 5 PAGE_PRIVATE_REF_RESOURCE 1384 * bit 6- f2fs private data 1385 * 1386 * Layout B: lowest bit should be 0 1387 * page.private is a wrapped pointer. 1388 */ 1389 enum { 1390 PAGE_PRIVATE_NOT_POINTER, /* private contains non-pointer data */ 1391 PAGE_PRIVATE_ATOMIC_WRITE, /* data page from atomic write path */ 1392 PAGE_PRIVATE_DUMMY_WRITE, /* data page for padding aligned IO */ 1393 PAGE_PRIVATE_ONGOING_MIGRATION, /* data page which is on-going migrating */ 1394 PAGE_PRIVATE_INLINE_INODE, /* inode page contains inline data */ 1395 PAGE_PRIVATE_REF_RESOURCE, /* dirty page has referenced resources */ 1396 PAGE_PRIVATE_MAX 1397 }; 1398 1399 #define PAGE_PRIVATE_GET_FUNC(name, flagname) \ 1400 static inline bool page_private_##name(struct page *page) \ 1401 { \ 1402 return PagePrivate(page) && \ 1403 test_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)) && \ 1404 test_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \ 1405 } 1406 1407 #define PAGE_PRIVATE_SET_FUNC(name, flagname) \ 1408 static inline void set_page_private_##name(struct page *page) \ 1409 { \ 1410 if (!PagePrivate(page)) { \ 1411 get_page(page); \ 1412 SetPagePrivate(page); \ 1413 set_page_private(page, 0); \ 1414 } \ 1415 set_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)); \ 1416 set_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \ 1417 } 1418 1419 #define PAGE_PRIVATE_CLEAR_FUNC(name, flagname) \ 1420 static inline void clear_page_private_##name(struct page *page) \ 1421 { \ 1422 clear_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \ 1423 if (page_private(page) == 1 << PAGE_PRIVATE_NOT_POINTER) { \ 1424 set_page_private(page, 0); \ 1425 if (PagePrivate(page)) { \ 1426 ClearPagePrivate(page); \ 1427 put_page(page); \ 1428 }\ 1429 } \ 1430 } 1431 1432 PAGE_PRIVATE_GET_FUNC(nonpointer, NOT_POINTER); 1433 PAGE_PRIVATE_GET_FUNC(reference, REF_RESOURCE); 1434 PAGE_PRIVATE_GET_FUNC(inline, INLINE_INODE); 1435 PAGE_PRIVATE_GET_FUNC(gcing, ONGOING_MIGRATION); 1436 PAGE_PRIVATE_GET_FUNC(atomic, ATOMIC_WRITE); 1437 PAGE_PRIVATE_GET_FUNC(dummy, DUMMY_WRITE); 1438 1439 PAGE_PRIVATE_SET_FUNC(reference, REF_RESOURCE); 1440 PAGE_PRIVATE_SET_FUNC(inline, INLINE_INODE); 1441 PAGE_PRIVATE_SET_FUNC(gcing, ONGOING_MIGRATION); 1442 PAGE_PRIVATE_SET_FUNC(atomic, ATOMIC_WRITE); 1443 PAGE_PRIVATE_SET_FUNC(dummy, DUMMY_WRITE); 1444 1445 PAGE_PRIVATE_CLEAR_FUNC(reference, REF_RESOURCE); 1446 PAGE_PRIVATE_CLEAR_FUNC(inline, INLINE_INODE); 1447 PAGE_PRIVATE_CLEAR_FUNC(gcing, ONGOING_MIGRATION); 1448 PAGE_PRIVATE_CLEAR_FUNC(atomic, ATOMIC_WRITE); 1449 PAGE_PRIVATE_CLEAR_FUNC(dummy, DUMMY_WRITE); 1450 1451 static inline unsigned long get_page_private_data(struct page *page) 1452 { 1453 unsigned long data = page_private(page); 1454 1455 if (!test_bit(PAGE_PRIVATE_NOT_POINTER, &data)) 1456 return 0; 1457 return data >> PAGE_PRIVATE_MAX; 1458 } 1459 1460 static inline void set_page_private_data(struct page *page, unsigned long data) 1461 { 1462 if (!PagePrivate(page)) { 1463 get_page(page); 1464 SetPagePrivate(page); 1465 set_page_private(page, 0); 1466 } 1467 set_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)); 1468 page_private(page) |= data << PAGE_PRIVATE_MAX; 1469 } 1470 1471 static inline void clear_page_private_data(struct page *page) 1472 { 1473 page_private(page) &= (1 << PAGE_PRIVATE_MAX) - 1; 1474 if (page_private(page) == 1 << PAGE_PRIVATE_NOT_POINTER) { 1475 set_page_private(page, 0); 1476 if (PagePrivate(page)) { 1477 ClearPagePrivate(page); 1478 put_page(page); 1479 } 1480 } 1481 } 1482 1483 /* For compression */ 1484 enum compress_algorithm_type { 1485 COMPRESS_LZO, 1486 COMPRESS_LZ4, 1487 COMPRESS_ZSTD, 1488 COMPRESS_LZORLE, 1489 COMPRESS_MAX, 1490 }; 1491 1492 enum compress_flag { 1493 COMPRESS_CHKSUM, 1494 COMPRESS_MAX_FLAG, 1495 }; 1496 1497 #define COMPRESS_WATERMARK 20 1498 #define COMPRESS_PERCENT 20 1499 1500 #define COMPRESS_DATA_RESERVED_SIZE 4 1501 struct compress_data { 1502 __le32 clen; /* compressed data size */ 1503 __le32 chksum; /* compressed data chksum */ 1504 __le32 reserved[COMPRESS_DATA_RESERVED_SIZE]; /* reserved */ 1505 u8 cdata[]; /* compressed data */ 1506 }; 1507 1508 #define COMPRESS_HEADER_SIZE (sizeof(struct compress_data)) 1509 1510 #define F2FS_COMPRESSED_PAGE_MAGIC 0xF5F2C000 1511 1512 #define COMPRESS_LEVEL_OFFSET 8 1513 1514 /* compress context */ 1515 struct compress_ctx { 1516 struct inode *inode; /* inode the context belong to */ 1517 pgoff_t cluster_idx; /* cluster index number */ 1518 unsigned int cluster_size; /* page count in cluster */ 1519 unsigned int log_cluster_size; /* log of cluster size */ 1520 struct page **rpages; /* pages store raw data in cluster */ 1521 unsigned int nr_rpages; /* total page number in rpages */ 1522 struct page **cpages; /* pages store compressed data in cluster */ 1523 unsigned int nr_cpages; /* total page number in cpages */ 1524 unsigned int valid_nr_cpages; /* valid page number in cpages */ 1525 void *rbuf; /* virtual mapped address on rpages */ 1526 struct compress_data *cbuf; /* virtual mapped address on cpages */ 1527 size_t rlen; /* valid data length in rbuf */ 1528 size_t clen; /* valid data length in cbuf */ 1529 void *private; /* payload buffer for specified compression algorithm */ 1530 void *private2; /* extra payload buffer */ 1531 }; 1532 1533 /* compress context for write IO path */ 1534 struct compress_io_ctx { 1535 u32 magic; /* magic number to indicate page is compressed */ 1536 struct inode *inode; /* inode the context belong to */ 1537 struct page **rpages; /* pages store raw data in cluster */ 1538 unsigned int nr_rpages; /* total page number in rpages */ 1539 atomic_t pending_pages; /* in-flight compressed page count */ 1540 }; 1541 1542 /* Context for decompressing one cluster on the read IO path */ 1543 struct decompress_io_ctx { 1544 u32 magic; /* magic number to indicate page is compressed */ 1545 struct inode *inode; /* inode the context belong to */ 1546 pgoff_t cluster_idx; /* cluster index number */ 1547 unsigned int cluster_size; /* page count in cluster */ 1548 unsigned int log_cluster_size; /* log of cluster size */ 1549 struct page **rpages; /* pages store raw data in cluster */ 1550 unsigned int nr_rpages; /* total page number in rpages */ 1551 struct page **cpages; /* pages store compressed data in cluster */ 1552 unsigned int nr_cpages; /* total page number in cpages */ 1553 struct page **tpages; /* temp pages to pad holes in cluster */ 1554 void *rbuf; /* virtual mapped address on rpages */ 1555 struct compress_data *cbuf; /* virtual mapped address on cpages */ 1556 size_t rlen; /* valid data length in rbuf */ 1557 size_t clen; /* valid data length in cbuf */ 1558 1559 /* 1560 * The number of compressed pages remaining to be read in this cluster. 1561 * This is initially nr_cpages. It is decremented by 1 each time a page 1562 * has been read (or failed to be read). When it reaches 0, the cluster 1563 * is decompressed (or an error is reported). 1564 * 1565 * If an error occurs before all the pages have been submitted for I/O, 1566 * then this will never reach 0. In this case the I/O submitter is 1567 * responsible for calling f2fs_decompress_end_io() instead. 1568 */ 1569 atomic_t remaining_pages; 1570 1571 /* 1572 * Number of references to this decompress_io_ctx. 1573 * 1574 * One reference is held for I/O completion. This reference is dropped 1575 * after the pagecache pages are updated and unlocked -- either after 1576 * decompression (and verity if enabled), or after an error. 1577 * 1578 * In addition, each compressed page holds a reference while it is in a 1579 * bio. These references are necessary prevent compressed pages from 1580 * being freed while they are still in a bio. 1581 */ 1582 refcount_t refcnt; 1583 1584 bool failed; /* IO error occurred before decompression? */ 1585 bool need_verity; /* need fs-verity verification after decompression? */ 1586 void *private; /* payload buffer for specified decompression algorithm */ 1587 void *private2; /* extra payload buffer */ 1588 struct work_struct verity_work; /* work to verify the decompressed pages */ 1589 }; 1590 1591 #define NULL_CLUSTER ((unsigned int)(~0)) 1592 #define MIN_COMPRESS_LOG_SIZE 2 1593 #define MAX_COMPRESS_LOG_SIZE 8 1594 #define MAX_COMPRESS_WINDOW_SIZE(log_size) ((PAGE_SIZE) << (log_size)) 1595 1596 struct f2fs_sb_info { 1597 struct super_block *sb; /* pointer to VFS super block */ 1598 struct proc_dir_entry *s_proc; /* proc entry */ 1599 struct f2fs_super_block *raw_super; /* raw super block pointer */ 1600 struct f2fs_rwsem sb_lock; /* lock for raw super block */ 1601 int valid_super_block; /* valid super block no */ 1602 unsigned long s_flag; /* flags for sbi */ 1603 struct mutex writepages; /* mutex for writepages() */ 1604 1605 #ifdef CONFIG_BLK_DEV_ZONED 1606 unsigned int blocks_per_blkz; /* F2FS blocks per zone */ 1607 unsigned int log_blocks_per_blkz; /* log2 F2FS blocks per zone */ 1608 #endif 1609 1610 /* for node-related operations */ 1611 struct f2fs_nm_info *nm_info; /* node manager */ 1612 struct inode *node_inode; /* cache node blocks */ 1613 1614 /* for segment-related operations */ 1615 struct f2fs_sm_info *sm_info; /* segment manager */ 1616 1617 /* for bio operations */ 1618 struct f2fs_bio_info *write_io[NR_PAGE_TYPE]; /* for write bios */ 1619 /* keep migration IO order for LFS mode */ 1620 struct f2fs_rwsem io_order_lock; 1621 mempool_t *write_io_dummy; /* Dummy pages */ 1622 pgoff_t metapage_eio_ofs; /* EIO page offset */ 1623 int metapage_eio_cnt; /* EIO count */ 1624 1625 /* for checkpoint */ 1626 struct f2fs_checkpoint *ckpt; /* raw checkpoint pointer */ 1627 int cur_cp_pack; /* remain current cp pack */ 1628 spinlock_t cp_lock; /* for flag in ckpt */ 1629 struct inode *meta_inode; /* cache meta blocks */ 1630 struct f2fs_rwsem cp_global_sem; /* checkpoint procedure lock */ 1631 struct f2fs_rwsem cp_rwsem; /* blocking FS operations */ 1632 struct f2fs_rwsem node_write; /* locking node writes */ 1633 struct f2fs_rwsem node_change; /* locking node change */ 1634 wait_queue_head_t cp_wait; 1635 unsigned long last_time[MAX_TIME]; /* to store time in jiffies */ 1636 long interval_time[MAX_TIME]; /* to store thresholds */ 1637 struct ckpt_req_control cprc_info; /* for checkpoint request control */ 1638 1639 struct inode_management im[MAX_INO_ENTRY]; /* manage inode cache */ 1640 1641 spinlock_t fsync_node_lock; /* for node entry lock */ 1642 struct list_head fsync_node_list; /* node list head */ 1643 unsigned int fsync_seg_id; /* sequence id */ 1644 unsigned int fsync_node_num; /* number of node entries */ 1645 1646 /* for orphan inode, use 0'th array */ 1647 unsigned int max_orphans; /* max orphan inodes */ 1648 1649 /* for inode management */ 1650 struct list_head inode_list[NR_INODE_TYPE]; /* dirty inode list */ 1651 spinlock_t inode_lock[NR_INODE_TYPE]; /* for dirty inode list lock */ 1652 struct mutex flush_lock; /* for flush exclusion */ 1653 1654 /* for extent tree cache */ 1655 struct radix_tree_root extent_tree_root;/* cache extent cache entries */ 1656 struct mutex extent_tree_lock; /* locking extent radix tree */ 1657 struct list_head extent_list; /* lru list for shrinker */ 1658 spinlock_t extent_lock; /* locking extent lru list */ 1659 atomic_t total_ext_tree; /* extent tree count */ 1660 struct list_head zombie_list; /* extent zombie tree list */ 1661 atomic_t total_zombie_tree; /* extent zombie tree count */ 1662 atomic_t total_ext_node; /* extent info count */ 1663 1664 /* basic filesystem units */ 1665 unsigned int log_sectors_per_block; /* log2 sectors per block */ 1666 unsigned int log_blocksize; /* log2 block size */ 1667 unsigned int blocksize; /* block size */ 1668 unsigned int root_ino_num; /* root inode number*/ 1669 unsigned int node_ino_num; /* node inode number*/ 1670 unsigned int meta_ino_num; /* meta inode number*/ 1671 unsigned int log_blocks_per_seg; /* log2 blocks per segment */ 1672 unsigned int blocks_per_seg; /* blocks per segment */ 1673 unsigned int segs_per_sec; /* segments per section */ 1674 unsigned int secs_per_zone; /* sections per zone */ 1675 unsigned int total_sections; /* total section count */ 1676 unsigned int total_node_count; /* total node block count */ 1677 unsigned int total_valid_node_count; /* valid node block count */ 1678 int dir_level; /* directory level */ 1679 int readdir_ra; /* readahead inode in readdir */ 1680 u64 max_io_bytes; /* max io bytes to merge IOs */ 1681 1682 block_t user_block_count; /* # of user blocks */ 1683 block_t total_valid_block_count; /* # of valid blocks */ 1684 block_t discard_blks; /* discard command candidats */ 1685 block_t last_valid_block_count; /* for recovery */ 1686 block_t reserved_blocks; /* configurable reserved blocks */ 1687 block_t current_reserved_blocks; /* current reserved blocks */ 1688 1689 /* Additional tracking for no checkpoint mode */ 1690 block_t unusable_block_count; /* # of blocks saved by last cp */ 1691 1692 unsigned int nquota_files; /* # of quota sysfile */ 1693 struct f2fs_rwsem quota_sem; /* blocking cp for flags */ 1694 1695 /* # of pages, see count_type */ 1696 atomic_t nr_pages[NR_COUNT_TYPE]; 1697 /* # of allocated blocks */ 1698 struct percpu_counter alloc_valid_block_count; 1699 /* # of node block writes as roll forward recovery */ 1700 struct percpu_counter rf_node_block_count; 1701 1702 /* writeback control */ 1703 atomic_t wb_sync_req[META]; /* count # of WB_SYNC threads */ 1704 1705 /* valid inode count */ 1706 struct percpu_counter total_valid_inode_count; 1707 1708 struct f2fs_mount_info mount_opt; /* mount options */ 1709 1710 /* for cleaning operations */ 1711 struct f2fs_rwsem gc_lock; /* 1712 * semaphore for GC, avoid 1713 * race between GC and GC or CP 1714 */ 1715 struct f2fs_gc_kthread *gc_thread; /* GC thread */ 1716 struct atgc_management am; /* atgc management */ 1717 unsigned int cur_victim_sec; /* current victim section num */ 1718 unsigned int gc_mode; /* current GC state */ 1719 unsigned int next_victim_seg[2]; /* next segment in victim section */ 1720 spinlock_t gc_urgent_high_lock; 1721 bool gc_urgent_high_limited; /* indicates having limited trial count */ 1722 unsigned int gc_urgent_high_remaining; /* remaining trial count for GC_URGENT_HIGH */ 1723 1724 /* for skip statistic */ 1725 unsigned int atomic_files; /* # of opened atomic file */ 1726 unsigned long long skipped_atomic_files[2]; /* FG_GC and BG_GC */ 1727 unsigned long long skipped_gc_rwsem; /* FG_GC only */ 1728 1729 /* threshold for gc trials on pinned files */ 1730 u64 gc_pin_file_threshold; 1731 struct f2fs_rwsem pin_sem; 1732 1733 /* maximum # of trials to find a victim segment for SSR and GC */ 1734 unsigned int max_victim_search; 1735 /* migration granularity of garbage collection, unit: segment */ 1736 unsigned int migration_granularity; 1737 1738 /* 1739 * for stat information. 1740 * one is for the LFS mode, and the other is for the SSR mode. 1741 */ 1742 #ifdef CONFIG_F2FS_STAT_FS 1743 struct f2fs_stat_info *stat_info; /* FS status information */ 1744 atomic_t meta_count[META_MAX]; /* # of meta blocks */ 1745 unsigned int segment_count[2]; /* # of allocated segments */ 1746 unsigned int block_count[2]; /* # of allocated blocks */ 1747 atomic_t inplace_count; /* # of inplace update */ 1748 atomic64_t total_hit_ext; /* # of lookup extent cache */ 1749 atomic64_t read_hit_rbtree; /* # of hit rbtree extent node */ 1750 atomic64_t read_hit_largest; /* # of hit largest extent node */ 1751 atomic64_t read_hit_cached; /* # of hit cached extent node */ 1752 atomic_t inline_xattr; /* # of inline_xattr inodes */ 1753 atomic_t inline_inode; /* # of inline_data inodes */ 1754 atomic_t inline_dir; /* # of inline_dentry inodes */ 1755 atomic_t compr_inode; /* # of compressed inodes */ 1756 atomic64_t compr_blocks; /* # of compressed blocks */ 1757 atomic_t vw_cnt; /* # of volatile writes */ 1758 atomic_t max_aw_cnt; /* max # of atomic writes */ 1759 atomic_t max_vw_cnt; /* max # of volatile writes */ 1760 unsigned int io_skip_bggc; /* skip background gc for in-flight IO */ 1761 unsigned int other_skip_bggc; /* skip background gc for other reasons */ 1762 unsigned int ndirty_inode[NR_INODE_TYPE]; /* # of dirty inodes */ 1763 #endif 1764 spinlock_t stat_lock; /* lock for stat operations */ 1765 1766 /* to attach REQ_META|REQ_FUA flags */ 1767 unsigned int data_io_flag; 1768 unsigned int node_io_flag; 1769 1770 /* For sysfs suppport */ 1771 struct kobject s_kobj; /* /sys/fs/f2fs/<devname> */ 1772 struct completion s_kobj_unregister; 1773 1774 struct kobject s_stat_kobj; /* /sys/fs/f2fs/<devname>/stat */ 1775 struct completion s_stat_kobj_unregister; 1776 1777 struct kobject s_feature_list_kobj; /* /sys/fs/f2fs/<devname>/feature_list */ 1778 struct completion s_feature_list_kobj_unregister; 1779 1780 /* For shrinker support */ 1781 struct list_head s_list; 1782 struct mutex umount_mutex; 1783 unsigned int shrinker_run_no; 1784 1785 /* For multi devices */ 1786 int s_ndevs; /* number of devices */ 1787 struct f2fs_dev_info *devs; /* for device list */ 1788 unsigned int dirty_device; /* for checkpoint data flush */ 1789 spinlock_t dev_lock; /* protect dirty_device */ 1790 bool aligned_blksize; /* all devices has the same logical blksize */ 1791 1792 /* For write statistics */ 1793 u64 sectors_written_start; 1794 u64 kbytes_written; 1795 1796 /* Reference to checksum algorithm driver via cryptoapi */ 1797 struct crypto_shash *s_chksum_driver; 1798 1799 /* Precomputed FS UUID checksum for seeding other checksums */ 1800 __u32 s_chksum_seed; 1801 1802 struct workqueue_struct *post_read_wq; /* post read workqueue */ 1803 1804 struct kmem_cache *inline_xattr_slab; /* inline xattr entry */ 1805 unsigned int inline_xattr_slab_size; /* default inline xattr slab size */ 1806 1807 /* For reclaimed segs statistics per each GC mode */ 1808 unsigned int gc_segment_mode; /* GC state for reclaimed segments */ 1809 unsigned int gc_reclaimed_segs[MAX_GC_MODE]; /* Reclaimed segs for each mode */ 1810 1811 unsigned long seq_file_ra_mul; /* multiplier for ra_pages of seq. files in fadvise */ 1812 1813 int max_fragment_chunk; /* max chunk size for block fragmentation mode */ 1814 int max_fragment_hole; /* max hole size for block fragmentation mode */ 1815 1816 #ifdef CONFIG_F2FS_FS_COMPRESSION 1817 struct kmem_cache *page_array_slab; /* page array entry */ 1818 unsigned int page_array_slab_size; /* default page array slab size */ 1819 1820 /* For runtime compression statistics */ 1821 u64 compr_written_block; 1822 u64 compr_saved_block; 1823 u32 compr_new_inode; 1824 1825 /* For compressed block cache */ 1826 struct inode *compress_inode; /* cache compressed blocks */ 1827 unsigned int compress_percent; /* cache page percentage */ 1828 unsigned int compress_watermark; /* cache page watermark */ 1829 atomic_t compress_page_hit; /* cache hit count */ 1830 #endif 1831 1832 #ifdef CONFIG_F2FS_IOSTAT 1833 /* For app/fs IO statistics */ 1834 spinlock_t iostat_lock; 1835 unsigned long long rw_iostat[NR_IO_TYPE]; 1836 unsigned long long prev_rw_iostat[NR_IO_TYPE]; 1837 bool iostat_enable; 1838 unsigned long iostat_next_period; 1839 unsigned int iostat_period_ms; 1840 1841 /* For io latency related statistics info in one iostat period */ 1842 spinlock_t iostat_lat_lock; 1843 struct iostat_lat_info *iostat_io_lat; 1844 #endif 1845 }; 1846 1847 #ifdef CONFIG_F2FS_FAULT_INJECTION 1848 #define f2fs_show_injection_info(sbi, type) \ 1849 printk_ratelimited("%sF2FS-fs (%s) : inject %s in %s of %pS\n", \ 1850 KERN_INFO, sbi->sb->s_id, \ 1851 f2fs_fault_name[type], \ 1852 __func__, __builtin_return_address(0)) 1853 static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type) 1854 { 1855 struct f2fs_fault_info *ffi = &F2FS_OPTION(sbi).fault_info; 1856 1857 if (!ffi->inject_rate) 1858 return false; 1859 1860 if (!IS_FAULT_SET(ffi, type)) 1861 return false; 1862 1863 atomic_inc(&ffi->inject_ops); 1864 if (atomic_read(&ffi->inject_ops) >= ffi->inject_rate) { 1865 atomic_set(&ffi->inject_ops, 0); 1866 return true; 1867 } 1868 return false; 1869 } 1870 #else 1871 #define f2fs_show_injection_info(sbi, type) do { } while (0) 1872 static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type) 1873 { 1874 return false; 1875 } 1876 #endif 1877 1878 /* 1879 * Test if the mounted volume is a multi-device volume. 1880 * - For a single regular disk volume, sbi->s_ndevs is 0. 1881 * - For a single zoned disk volume, sbi->s_ndevs is 1. 1882 * - For a multi-device volume, sbi->s_ndevs is always 2 or more. 1883 */ 1884 static inline bool f2fs_is_multi_device(struct f2fs_sb_info *sbi) 1885 { 1886 return sbi->s_ndevs > 1; 1887 } 1888 1889 static inline void f2fs_update_time(struct f2fs_sb_info *sbi, int type) 1890 { 1891 unsigned long now = jiffies; 1892 1893 sbi->last_time[type] = now; 1894 1895 /* DISCARD_TIME and GC_TIME are based on REQ_TIME */ 1896 if (type == REQ_TIME) { 1897 sbi->last_time[DISCARD_TIME] = now; 1898 sbi->last_time[GC_TIME] = now; 1899 } 1900 } 1901 1902 static inline bool f2fs_time_over(struct f2fs_sb_info *sbi, int type) 1903 { 1904 unsigned long interval = sbi->interval_time[type] * HZ; 1905 1906 return time_after(jiffies, sbi->last_time[type] + interval); 1907 } 1908 1909 static inline unsigned int f2fs_time_to_wait(struct f2fs_sb_info *sbi, 1910 int type) 1911 { 1912 unsigned long interval = sbi->interval_time[type] * HZ; 1913 unsigned int wait_ms = 0; 1914 long delta; 1915 1916 delta = (sbi->last_time[type] + interval) - jiffies; 1917 if (delta > 0) 1918 wait_ms = jiffies_to_msecs(delta); 1919 1920 return wait_ms; 1921 } 1922 1923 /* 1924 * Inline functions 1925 */ 1926 static inline u32 __f2fs_crc32(struct f2fs_sb_info *sbi, u32 crc, 1927 const void *address, unsigned int length) 1928 { 1929 struct { 1930 struct shash_desc shash; 1931 char ctx[4]; 1932 } desc; 1933 int err; 1934 1935 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver) != sizeof(desc.ctx)); 1936 1937 desc.shash.tfm = sbi->s_chksum_driver; 1938 *(u32 *)desc.ctx = crc; 1939 1940 err = crypto_shash_update(&desc.shash, address, length); 1941 BUG_ON(err); 1942 1943 return *(u32 *)desc.ctx; 1944 } 1945 1946 static inline u32 f2fs_crc32(struct f2fs_sb_info *sbi, const void *address, 1947 unsigned int length) 1948 { 1949 return __f2fs_crc32(sbi, F2FS_SUPER_MAGIC, address, length); 1950 } 1951 1952 static inline bool f2fs_crc_valid(struct f2fs_sb_info *sbi, __u32 blk_crc, 1953 void *buf, size_t buf_size) 1954 { 1955 return f2fs_crc32(sbi, buf, buf_size) == blk_crc; 1956 } 1957 1958 static inline u32 f2fs_chksum(struct f2fs_sb_info *sbi, u32 crc, 1959 const void *address, unsigned int length) 1960 { 1961 return __f2fs_crc32(sbi, crc, address, length); 1962 } 1963 1964 static inline struct f2fs_inode_info *F2FS_I(struct inode *inode) 1965 { 1966 return container_of(inode, struct f2fs_inode_info, vfs_inode); 1967 } 1968 1969 static inline struct f2fs_sb_info *F2FS_SB(struct super_block *sb) 1970 { 1971 return sb->s_fs_info; 1972 } 1973 1974 static inline struct f2fs_sb_info *F2FS_I_SB(struct inode *inode) 1975 { 1976 return F2FS_SB(inode->i_sb); 1977 } 1978 1979 static inline struct f2fs_sb_info *F2FS_M_SB(struct address_space *mapping) 1980 { 1981 return F2FS_I_SB(mapping->host); 1982 } 1983 1984 static inline struct f2fs_sb_info *F2FS_P_SB(struct page *page) 1985 { 1986 return F2FS_M_SB(page_file_mapping(page)); 1987 } 1988 1989 static inline struct f2fs_super_block *F2FS_RAW_SUPER(struct f2fs_sb_info *sbi) 1990 { 1991 return (struct f2fs_super_block *)(sbi->raw_super); 1992 } 1993 1994 static inline struct f2fs_checkpoint *F2FS_CKPT(struct f2fs_sb_info *sbi) 1995 { 1996 return (struct f2fs_checkpoint *)(sbi->ckpt); 1997 } 1998 1999 static inline struct f2fs_node *F2FS_NODE(struct page *page) 2000 { 2001 return (struct f2fs_node *)page_address(page); 2002 } 2003 2004 static inline struct f2fs_inode *F2FS_INODE(struct page *page) 2005 { 2006 return &((struct f2fs_node *)page_address(page))->i; 2007 } 2008 2009 static inline struct f2fs_nm_info *NM_I(struct f2fs_sb_info *sbi) 2010 { 2011 return (struct f2fs_nm_info *)(sbi->nm_info); 2012 } 2013 2014 static inline struct f2fs_sm_info *SM_I(struct f2fs_sb_info *sbi) 2015 { 2016 return (struct f2fs_sm_info *)(sbi->sm_info); 2017 } 2018 2019 static inline struct sit_info *SIT_I(struct f2fs_sb_info *sbi) 2020 { 2021 return (struct sit_info *)(SM_I(sbi)->sit_info); 2022 } 2023 2024 static inline struct free_segmap_info *FREE_I(struct f2fs_sb_info *sbi) 2025 { 2026 return (struct free_segmap_info *)(SM_I(sbi)->free_info); 2027 } 2028 2029 static inline struct dirty_seglist_info *DIRTY_I(struct f2fs_sb_info *sbi) 2030 { 2031 return (struct dirty_seglist_info *)(SM_I(sbi)->dirty_info); 2032 } 2033 2034 static inline struct address_space *META_MAPPING(struct f2fs_sb_info *sbi) 2035 { 2036 return sbi->meta_inode->i_mapping; 2037 } 2038 2039 static inline struct address_space *NODE_MAPPING(struct f2fs_sb_info *sbi) 2040 { 2041 return sbi->node_inode->i_mapping; 2042 } 2043 2044 static inline bool is_sbi_flag_set(struct f2fs_sb_info *sbi, unsigned int type) 2045 { 2046 return test_bit(type, &sbi->s_flag); 2047 } 2048 2049 static inline void set_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type) 2050 { 2051 set_bit(type, &sbi->s_flag); 2052 } 2053 2054 static inline void clear_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type) 2055 { 2056 clear_bit(type, &sbi->s_flag); 2057 } 2058 2059 static inline unsigned long long cur_cp_version(struct f2fs_checkpoint *cp) 2060 { 2061 return le64_to_cpu(cp->checkpoint_ver); 2062 } 2063 2064 static inline unsigned long f2fs_qf_ino(struct super_block *sb, int type) 2065 { 2066 if (type < F2FS_MAX_QUOTAS) 2067 return le32_to_cpu(F2FS_SB(sb)->raw_super->qf_ino[type]); 2068 return 0; 2069 } 2070 2071 static inline __u64 cur_cp_crc(struct f2fs_checkpoint *cp) 2072 { 2073 size_t crc_offset = le32_to_cpu(cp->checksum_offset); 2074 return le32_to_cpu(*((__le32 *)((unsigned char *)cp + crc_offset))); 2075 } 2076 2077 static inline bool __is_set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 2078 { 2079 unsigned int ckpt_flags = le32_to_cpu(cp->ckpt_flags); 2080 2081 return ckpt_flags & f; 2082 } 2083 2084 static inline bool is_set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 2085 { 2086 return __is_set_ckpt_flags(F2FS_CKPT(sbi), f); 2087 } 2088 2089 static inline void __set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 2090 { 2091 unsigned int ckpt_flags; 2092 2093 ckpt_flags = le32_to_cpu(cp->ckpt_flags); 2094 ckpt_flags |= f; 2095 cp->ckpt_flags = cpu_to_le32(ckpt_flags); 2096 } 2097 2098 static inline void set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 2099 { 2100 unsigned long flags; 2101 2102 spin_lock_irqsave(&sbi->cp_lock, flags); 2103 __set_ckpt_flags(F2FS_CKPT(sbi), f); 2104 spin_unlock_irqrestore(&sbi->cp_lock, flags); 2105 } 2106 2107 static inline void __clear_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 2108 { 2109 unsigned int ckpt_flags; 2110 2111 ckpt_flags = le32_to_cpu(cp->ckpt_flags); 2112 ckpt_flags &= (~f); 2113 cp->ckpt_flags = cpu_to_le32(ckpt_flags); 2114 } 2115 2116 static inline void clear_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 2117 { 2118 unsigned long flags; 2119 2120 spin_lock_irqsave(&sbi->cp_lock, flags); 2121 __clear_ckpt_flags(F2FS_CKPT(sbi), f); 2122 spin_unlock_irqrestore(&sbi->cp_lock, flags); 2123 } 2124 2125 #define init_f2fs_rwsem(sem) \ 2126 do { \ 2127 static struct lock_class_key __key; \ 2128 \ 2129 __init_f2fs_rwsem((sem), #sem, &__key); \ 2130 } while (0) 2131 2132 static inline void __init_f2fs_rwsem(struct f2fs_rwsem *sem, 2133 const char *sem_name, struct lock_class_key *key) 2134 { 2135 __init_rwsem(&sem->internal_rwsem, sem_name, key); 2136 #ifdef CONFIG_F2FS_UNFAIR_RWSEM 2137 init_waitqueue_head(&sem->read_waiters); 2138 #endif 2139 } 2140 2141 static inline int f2fs_rwsem_is_locked(struct f2fs_rwsem *sem) 2142 { 2143 return rwsem_is_locked(&sem->internal_rwsem); 2144 } 2145 2146 static inline int f2fs_rwsem_is_contended(struct f2fs_rwsem *sem) 2147 { 2148 return rwsem_is_contended(&sem->internal_rwsem); 2149 } 2150 2151 static inline void f2fs_down_read(struct f2fs_rwsem *sem) 2152 { 2153 #ifdef CONFIG_F2FS_UNFAIR_RWSEM 2154 wait_event(sem->read_waiters, down_read_trylock(&sem->internal_rwsem)); 2155 #else 2156 down_read(&sem->internal_rwsem); 2157 #endif 2158 } 2159 2160 static inline int f2fs_down_read_trylock(struct f2fs_rwsem *sem) 2161 { 2162 return down_read_trylock(&sem->internal_rwsem); 2163 } 2164 2165 #ifdef CONFIG_DEBUG_LOCK_ALLOC 2166 static inline void f2fs_down_read_nested(struct f2fs_rwsem *sem, int subclass) 2167 { 2168 down_read_nested(&sem->internal_rwsem, subclass); 2169 } 2170 #else 2171 #define f2fs_down_read_nested(sem, subclass) f2fs_down_read(sem) 2172 #endif 2173 2174 static inline void f2fs_up_read(struct f2fs_rwsem *sem) 2175 { 2176 up_read(&sem->internal_rwsem); 2177 } 2178 2179 static inline void f2fs_down_write(struct f2fs_rwsem *sem) 2180 { 2181 down_write(&sem->internal_rwsem); 2182 } 2183 2184 static inline int f2fs_down_write_trylock(struct f2fs_rwsem *sem) 2185 { 2186 return down_write_trylock(&sem->internal_rwsem); 2187 } 2188 2189 static inline void f2fs_up_write(struct f2fs_rwsem *sem) 2190 { 2191 up_write(&sem->internal_rwsem); 2192 #ifdef CONFIG_F2FS_UNFAIR_RWSEM 2193 wake_up_all(&sem->read_waiters); 2194 #endif 2195 } 2196 2197 static inline void f2fs_lock_op(struct f2fs_sb_info *sbi) 2198 { 2199 f2fs_down_read(&sbi->cp_rwsem); 2200 } 2201 2202 static inline int f2fs_trylock_op(struct f2fs_sb_info *sbi) 2203 { 2204 if (time_to_inject(sbi, FAULT_LOCK_OP)) { 2205 f2fs_show_injection_info(sbi, FAULT_LOCK_OP); 2206 return 0; 2207 } 2208 return f2fs_down_read_trylock(&sbi->cp_rwsem); 2209 } 2210 2211 static inline void f2fs_unlock_op(struct f2fs_sb_info *sbi) 2212 { 2213 f2fs_up_read(&sbi->cp_rwsem); 2214 } 2215 2216 static inline void f2fs_lock_all(struct f2fs_sb_info *sbi) 2217 { 2218 f2fs_down_write(&sbi->cp_rwsem); 2219 } 2220 2221 static inline void f2fs_unlock_all(struct f2fs_sb_info *sbi) 2222 { 2223 f2fs_up_write(&sbi->cp_rwsem); 2224 } 2225 2226 static inline int __get_cp_reason(struct f2fs_sb_info *sbi) 2227 { 2228 int reason = CP_SYNC; 2229 2230 if (test_opt(sbi, FASTBOOT)) 2231 reason = CP_FASTBOOT; 2232 if (is_sbi_flag_set(sbi, SBI_IS_CLOSE)) 2233 reason = CP_UMOUNT; 2234 return reason; 2235 } 2236 2237 static inline bool __remain_node_summaries(int reason) 2238 { 2239 return (reason & (CP_UMOUNT | CP_FASTBOOT)); 2240 } 2241 2242 static inline bool __exist_node_summaries(struct f2fs_sb_info *sbi) 2243 { 2244 return (is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG) || 2245 is_set_ckpt_flags(sbi, CP_FASTBOOT_FLAG)); 2246 } 2247 2248 /* 2249 * Check whether the inode has blocks or not 2250 */ 2251 static inline int F2FS_HAS_BLOCKS(struct inode *inode) 2252 { 2253 block_t xattr_block = F2FS_I(inode)->i_xattr_nid ? 1 : 0; 2254 2255 return (inode->i_blocks >> F2FS_LOG_SECTORS_PER_BLOCK) > xattr_block; 2256 } 2257 2258 static inline bool f2fs_has_xattr_block(unsigned int ofs) 2259 { 2260 return ofs == XATTR_NODE_OFFSET; 2261 } 2262 2263 static inline bool __allow_reserved_blocks(struct f2fs_sb_info *sbi, 2264 struct inode *inode, bool cap) 2265 { 2266 if (!inode) 2267 return true; 2268 if (!test_opt(sbi, RESERVE_ROOT)) 2269 return false; 2270 if (IS_NOQUOTA(inode)) 2271 return true; 2272 if (uid_eq(F2FS_OPTION(sbi).s_resuid, current_fsuid())) 2273 return true; 2274 if (!gid_eq(F2FS_OPTION(sbi).s_resgid, GLOBAL_ROOT_GID) && 2275 in_group_p(F2FS_OPTION(sbi).s_resgid)) 2276 return true; 2277 if (cap && capable(CAP_SYS_RESOURCE)) 2278 return true; 2279 return false; 2280 } 2281 2282 static inline void f2fs_i_blocks_write(struct inode *, block_t, bool, bool); 2283 static inline int inc_valid_block_count(struct f2fs_sb_info *sbi, 2284 struct inode *inode, blkcnt_t *count) 2285 { 2286 blkcnt_t diff = 0, release = 0; 2287 block_t avail_user_block_count; 2288 int ret; 2289 2290 ret = dquot_reserve_block(inode, *count); 2291 if (ret) 2292 return ret; 2293 2294 if (time_to_inject(sbi, FAULT_BLOCK)) { 2295 f2fs_show_injection_info(sbi, FAULT_BLOCK); 2296 release = *count; 2297 goto release_quota; 2298 } 2299 2300 /* 2301 * let's increase this in prior to actual block count change in order 2302 * for f2fs_sync_file to avoid data races when deciding checkpoint. 2303 */ 2304 percpu_counter_add(&sbi->alloc_valid_block_count, (*count)); 2305 2306 spin_lock(&sbi->stat_lock); 2307 sbi->total_valid_block_count += (block_t)(*count); 2308 avail_user_block_count = sbi->user_block_count - 2309 sbi->current_reserved_blocks; 2310 2311 if (!__allow_reserved_blocks(sbi, inode, true)) 2312 avail_user_block_count -= F2FS_OPTION(sbi).root_reserved_blocks; 2313 2314 if (F2FS_IO_ALIGNED(sbi)) 2315 avail_user_block_count -= sbi->blocks_per_seg * 2316 SM_I(sbi)->additional_reserved_segments; 2317 2318 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) { 2319 if (avail_user_block_count > sbi->unusable_block_count) 2320 avail_user_block_count -= sbi->unusable_block_count; 2321 else 2322 avail_user_block_count = 0; 2323 } 2324 if (unlikely(sbi->total_valid_block_count > avail_user_block_count)) { 2325 diff = sbi->total_valid_block_count - avail_user_block_count; 2326 if (diff > *count) 2327 diff = *count; 2328 *count -= diff; 2329 release = diff; 2330 sbi->total_valid_block_count -= diff; 2331 if (!*count) { 2332 spin_unlock(&sbi->stat_lock); 2333 goto enospc; 2334 } 2335 } 2336 spin_unlock(&sbi->stat_lock); 2337 2338 if (unlikely(release)) { 2339 percpu_counter_sub(&sbi->alloc_valid_block_count, release); 2340 dquot_release_reservation_block(inode, release); 2341 } 2342 f2fs_i_blocks_write(inode, *count, true, true); 2343 return 0; 2344 2345 enospc: 2346 percpu_counter_sub(&sbi->alloc_valid_block_count, release); 2347 release_quota: 2348 dquot_release_reservation_block(inode, release); 2349 return -ENOSPC; 2350 } 2351 2352 __printf(2, 3) 2353 void f2fs_printk(struct f2fs_sb_info *sbi, const char *fmt, ...); 2354 2355 #define f2fs_err(sbi, fmt, ...) \ 2356 f2fs_printk(sbi, KERN_ERR fmt, ##__VA_ARGS__) 2357 #define f2fs_warn(sbi, fmt, ...) \ 2358 f2fs_printk(sbi, KERN_WARNING fmt, ##__VA_ARGS__) 2359 #define f2fs_notice(sbi, fmt, ...) \ 2360 f2fs_printk(sbi, KERN_NOTICE fmt, ##__VA_ARGS__) 2361 #define f2fs_info(sbi, fmt, ...) \ 2362 f2fs_printk(sbi, KERN_INFO fmt, ##__VA_ARGS__) 2363 #define f2fs_debug(sbi, fmt, ...) \ 2364 f2fs_printk(sbi, KERN_DEBUG fmt, ##__VA_ARGS__) 2365 2366 static inline void dec_valid_block_count(struct f2fs_sb_info *sbi, 2367 struct inode *inode, 2368 block_t count) 2369 { 2370 blkcnt_t sectors = count << F2FS_LOG_SECTORS_PER_BLOCK; 2371 2372 spin_lock(&sbi->stat_lock); 2373 f2fs_bug_on(sbi, sbi->total_valid_block_count < (block_t) count); 2374 sbi->total_valid_block_count -= (block_t)count; 2375 if (sbi->reserved_blocks && 2376 sbi->current_reserved_blocks < sbi->reserved_blocks) 2377 sbi->current_reserved_blocks = min(sbi->reserved_blocks, 2378 sbi->current_reserved_blocks + count); 2379 spin_unlock(&sbi->stat_lock); 2380 if (unlikely(inode->i_blocks < sectors)) { 2381 f2fs_warn(sbi, "Inconsistent i_blocks, ino:%lu, iblocks:%llu, sectors:%llu", 2382 inode->i_ino, 2383 (unsigned long long)inode->i_blocks, 2384 (unsigned long long)sectors); 2385 set_sbi_flag(sbi, SBI_NEED_FSCK); 2386 return; 2387 } 2388 f2fs_i_blocks_write(inode, count, false, true); 2389 } 2390 2391 static inline void inc_page_count(struct f2fs_sb_info *sbi, int count_type) 2392 { 2393 atomic_inc(&sbi->nr_pages[count_type]); 2394 2395 if (count_type == F2FS_DIRTY_DENTS || 2396 count_type == F2FS_DIRTY_NODES || 2397 count_type == F2FS_DIRTY_META || 2398 count_type == F2FS_DIRTY_QDATA || 2399 count_type == F2FS_DIRTY_IMETA) 2400 set_sbi_flag(sbi, SBI_IS_DIRTY); 2401 } 2402 2403 static inline void inode_inc_dirty_pages(struct inode *inode) 2404 { 2405 atomic_inc(&F2FS_I(inode)->dirty_pages); 2406 inc_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ? 2407 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA); 2408 if (IS_NOQUOTA(inode)) 2409 inc_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA); 2410 } 2411 2412 static inline void dec_page_count(struct f2fs_sb_info *sbi, int count_type) 2413 { 2414 atomic_dec(&sbi->nr_pages[count_type]); 2415 } 2416 2417 static inline void inode_dec_dirty_pages(struct inode *inode) 2418 { 2419 if (!S_ISDIR(inode->i_mode) && !S_ISREG(inode->i_mode) && 2420 !S_ISLNK(inode->i_mode)) 2421 return; 2422 2423 atomic_dec(&F2FS_I(inode)->dirty_pages); 2424 dec_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ? 2425 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA); 2426 if (IS_NOQUOTA(inode)) 2427 dec_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA); 2428 } 2429 2430 static inline s64 get_pages(struct f2fs_sb_info *sbi, int count_type) 2431 { 2432 return atomic_read(&sbi->nr_pages[count_type]); 2433 } 2434 2435 static inline int get_dirty_pages(struct inode *inode) 2436 { 2437 return atomic_read(&F2FS_I(inode)->dirty_pages); 2438 } 2439 2440 static inline int get_blocktype_secs(struct f2fs_sb_info *sbi, int block_type) 2441 { 2442 unsigned int pages_per_sec = sbi->segs_per_sec * sbi->blocks_per_seg; 2443 unsigned int segs = (get_pages(sbi, block_type) + pages_per_sec - 1) >> 2444 sbi->log_blocks_per_seg; 2445 2446 return segs / sbi->segs_per_sec; 2447 } 2448 2449 static inline block_t valid_user_blocks(struct f2fs_sb_info *sbi) 2450 { 2451 return sbi->total_valid_block_count; 2452 } 2453 2454 static inline block_t discard_blocks(struct f2fs_sb_info *sbi) 2455 { 2456 return sbi->discard_blks; 2457 } 2458 2459 static inline unsigned long __bitmap_size(struct f2fs_sb_info *sbi, int flag) 2460 { 2461 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi); 2462 2463 /* return NAT or SIT bitmap */ 2464 if (flag == NAT_BITMAP) 2465 return le32_to_cpu(ckpt->nat_ver_bitmap_bytesize); 2466 else if (flag == SIT_BITMAP) 2467 return le32_to_cpu(ckpt->sit_ver_bitmap_bytesize); 2468 2469 return 0; 2470 } 2471 2472 static inline block_t __cp_payload(struct f2fs_sb_info *sbi) 2473 { 2474 return le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_payload); 2475 } 2476 2477 static inline void *__bitmap_ptr(struct f2fs_sb_info *sbi, int flag) 2478 { 2479 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi); 2480 void *tmp_ptr = &ckpt->sit_nat_version_bitmap; 2481 int offset; 2482 2483 if (is_set_ckpt_flags(sbi, CP_LARGE_NAT_BITMAP_FLAG)) { 2484 offset = (flag == SIT_BITMAP) ? 2485 le32_to_cpu(ckpt->nat_ver_bitmap_bytesize) : 0; 2486 /* 2487 * if large_nat_bitmap feature is enabled, leave checksum 2488 * protection for all nat/sit bitmaps. 2489 */ 2490 return tmp_ptr + offset + sizeof(__le32); 2491 } 2492 2493 if (__cp_payload(sbi) > 0) { 2494 if (flag == NAT_BITMAP) 2495 return &ckpt->sit_nat_version_bitmap; 2496 else 2497 return (unsigned char *)ckpt + F2FS_BLKSIZE; 2498 } else { 2499 offset = (flag == NAT_BITMAP) ? 2500 le32_to_cpu(ckpt->sit_ver_bitmap_bytesize) : 0; 2501 return tmp_ptr + offset; 2502 } 2503 } 2504 2505 static inline block_t __start_cp_addr(struct f2fs_sb_info *sbi) 2506 { 2507 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr); 2508 2509 if (sbi->cur_cp_pack == 2) 2510 start_addr += sbi->blocks_per_seg; 2511 return start_addr; 2512 } 2513 2514 static inline block_t __start_cp_next_addr(struct f2fs_sb_info *sbi) 2515 { 2516 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr); 2517 2518 if (sbi->cur_cp_pack == 1) 2519 start_addr += sbi->blocks_per_seg; 2520 return start_addr; 2521 } 2522 2523 static inline void __set_cp_next_pack(struct f2fs_sb_info *sbi) 2524 { 2525 sbi->cur_cp_pack = (sbi->cur_cp_pack == 1) ? 2 : 1; 2526 } 2527 2528 static inline block_t __start_sum_addr(struct f2fs_sb_info *sbi) 2529 { 2530 return le32_to_cpu(F2FS_CKPT(sbi)->cp_pack_start_sum); 2531 } 2532 2533 static inline int inc_valid_node_count(struct f2fs_sb_info *sbi, 2534 struct inode *inode, bool is_inode) 2535 { 2536 block_t valid_block_count; 2537 unsigned int valid_node_count, user_block_count; 2538 int err; 2539 2540 if (is_inode) { 2541 if (inode) { 2542 err = dquot_alloc_inode(inode); 2543 if (err) 2544 return err; 2545 } 2546 } else { 2547 err = dquot_reserve_block(inode, 1); 2548 if (err) 2549 return err; 2550 } 2551 2552 if (time_to_inject(sbi, FAULT_BLOCK)) { 2553 f2fs_show_injection_info(sbi, FAULT_BLOCK); 2554 goto enospc; 2555 } 2556 2557 spin_lock(&sbi->stat_lock); 2558 2559 valid_block_count = sbi->total_valid_block_count + 2560 sbi->current_reserved_blocks + 1; 2561 2562 if (!__allow_reserved_blocks(sbi, inode, false)) 2563 valid_block_count += F2FS_OPTION(sbi).root_reserved_blocks; 2564 2565 if (F2FS_IO_ALIGNED(sbi)) 2566 valid_block_count += sbi->blocks_per_seg * 2567 SM_I(sbi)->additional_reserved_segments; 2568 2569 user_block_count = sbi->user_block_count; 2570 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) 2571 user_block_count -= sbi->unusable_block_count; 2572 2573 if (unlikely(valid_block_count > user_block_count)) { 2574 spin_unlock(&sbi->stat_lock); 2575 goto enospc; 2576 } 2577 2578 valid_node_count = sbi->total_valid_node_count + 1; 2579 if (unlikely(valid_node_count > sbi->total_node_count)) { 2580 spin_unlock(&sbi->stat_lock); 2581 goto enospc; 2582 } 2583 2584 sbi->total_valid_node_count++; 2585 sbi->total_valid_block_count++; 2586 spin_unlock(&sbi->stat_lock); 2587 2588 if (inode) { 2589 if (is_inode) 2590 f2fs_mark_inode_dirty_sync(inode, true); 2591 else 2592 f2fs_i_blocks_write(inode, 1, true, true); 2593 } 2594 2595 percpu_counter_inc(&sbi->alloc_valid_block_count); 2596 return 0; 2597 2598 enospc: 2599 if (is_inode) { 2600 if (inode) 2601 dquot_free_inode(inode); 2602 } else { 2603 dquot_release_reservation_block(inode, 1); 2604 } 2605 return -ENOSPC; 2606 } 2607 2608 static inline void dec_valid_node_count(struct f2fs_sb_info *sbi, 2609 struct inode *inode, bool is_inode) 2610 { 2611 spin_lock(&sbi->stat_lock); 2612 2613 f2fs_bug_on(sbi, !sbi->total_valid_block_count); 2614 f2fs_bug_on(sbi, !sbi->total_valid_node_count); 2615 2616 sbi->total_valid_node_count--; 2617 sbi->total_valid_block_count--; 2618 if (sbi->reserved_blocks && 2619 sbi->current_reserved_blocks < sbi->reserved_blocks) 2620 sbi->current_reserved_blocks++; 2621 2622 spin_unlock(&sbi->stat_lock); 2623 2624 if (is_inode) { 2625 dquot_free_inode(inode); 2626 } else { 2627 if (unlikely(inode->i_blocks == 0)) { 2628 f2fs_warn(sbi, "dec_valid_node_count: inconsistent i_blocks, ino:%lu, iblocks:%llu", 2629 inode->i_ino, 2630 (unsigned long long)inode->i_blocks); 2631 set_sbi_flag(sbi, SBI_NEED_FSCK); 2632 return; 2633 } 2634 f2fs_i_blocks_write(inode, 1, false, true); 2635 } 2636 } 2637 2638 static inline unsigned int valid_node_count(struct f2fs_sb_info *sbi) 2639 { 2640 return sbi->total_valid_node_count; 2641 } 2642 2643 static inline void inc_valid_inode_count(struct f2fs_sb_info *sbi) 2644 { 2645 percpu_counter_inc(&sbi->total_valid_inode_count); 2646 } 2647 2648 static inline void dec_valid_inode_count(struct f2fs_sb_info *sbi) 2649 { 2650 percpu_counter_dec(&sbi->total_valid_inode_count); 2651 } 2652 2653 static inline s64 valid_inode_count(struct f2fs_sb_info *sbi) 2654 { 2655 return percpu_counter_sum_positive(&sbi->total_valid_inode_count); 2656 } 2657 2658 static inline struct page *f2fs_grab_cache_page(struct address_space *mapping, 2659 pgoff_t index, bool for_write) 2660 { 2661 struct page *page; 2662 2663 if (IS_ENABLED(CONFIG_F2FS_FAULT_INJECTION)) { 2664 if (!for_write) 2665 page = find_get_page_flags(mapping, index, 2666 FGP_LOCK | FGP_ACCESSED); 2667 else 2668 page = find_lock_page(mapping, index); 2669 if (page) 2670 return page; 2671 2672 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_ALLOC)) { 2673 f2fs_show_injection_info(F2FS_M_SB(mapping), 2674 FAULT_PAGE_ALLOC); 2675 return NULL; 2676 } 2677 } 2678 2679 if (!for_write) 2680 return grab_cache_page(mapping, index); 2681 return grab_cache_page_write_begin(mapping, index, AOP_FLAG_NOFS); 2682 } 2683 2684 static inline struct page *f2fs_pagecache_get_page( 2685 struct address_space *mapping, pgoff_t index, 2686 int fgp_flags, gfp_t gfp_mask) 2687 { 2688 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_GET)) { 2689 f2fs_show_injection_info(F2FS_M_SB(mapping), FAULT_PAGE_GET); 2690 return NULL; 2691 } 2692 2693 return pagecache_get_page(mapping, index, fgp_flags, gfp_mask); 2694 } 2695 2696 static inline void f2fs_copy_page(struct page *src, struct page *dst) 2697 { 2698 char *src_kaddr = kmap(src); 2699 char *dst_kaddr = kmap(dst); 2700 2701 memcpy(dst_kaddr, src_kaddr, PAGE_SIZE); 2702 kunmap(dst); 2703 kunmap(src); 2704 } 2705 2706 static inline void f2fs_put_page(struct page *page, int unlock) 2707 { 2708 if (!page) 2709 return; 2710 2711 if (unlock) { 2712 f2fs_bug_on(F2FS_P_SB(page), !PageLocked(page)); 2713 unlock_page(page); 2714 } 2715 put_page(page); 2716 } 2717 2718 static inline void f2fs_put_dnode(struct dnode_of_data *dn) 2719 { 2720 if (dn->node_page) 2721 f2fs_put_page(dn->node_page, 1); 2722 if (dn->inode_page && dn->node_page != dn->inode_page) 2723 f2fs_put_page(dn->inode_page, 0); 2724 dn->node_page = NULL; 2725 dn->inode_page = NULL; 2726 } 2727 2728 static inline struct kmem_cache *f2fs_kmem_cache_create(const char *name, 2729 size_t size) 2730 { 2731 return kmem_cache_create(name, size, 0, SLAB_RECLAIM_ACCOUNT, NULL); 2732 } 2733 2734 static inline void *f2fs_kmem_cache_alloc_nofail(struct kmem_cache *cachep, 2735 gfp_t flags) 2736 { 2737 void *entry; 2738 2739 entry = kmem_cache_alloc(cachep, flags); 2740 if (!entry) 2741 entry = kmem_cache_alloc(cachep, flags | __GFP_NOFAIL); 2742 return entry; 2743 } 2744 2745 static inline void *f2fs_kmem_cache_alloc(struct kmem_cache *cachep, 2746 gfp_t flags, bool nofail, struct f2fs_sb_info *sbi) 2747 { 2748 if (nofail) 2749 return f2fs_kmem_cache_alloc_nofail(cachep, flags); 2750 2751 if (time_to_inject(sbi, FAULT_SLAB_ALLOC)) { 2752 f2fs_show_injection_info(sbi, FAULT_SLAB_ALLOC); 2753 return NULL; 2754 } 2755 2756 return kmem_cache_alloc(cachep, flags); 2757 } 2758 2759 static inline bool is_inflight_io(struct f2fs_sb_info *sbi, int type) 2760 { 2761 if (get_pages(sbi, F2FS_RD_DATA) || get_pages(sbi, F2FS_RD_NODE) || 2762 get_pages(sbi, F2FS_RD_META) || get_pages(sbi, F2FS_WB_DATA) || 2763 get_pages(sbi, F2FS_WB_CP_DATA) || 2764 get_pages(sbi, F2FS_DIO_READ) || 2765 get_pages(sbi, F2FS_DIO_WRITE)) 2766 return true; 2767 2768 if (type != DISCARD_TIME && SM_I(sbi) && SM_I(sbi)->dcc_info && 2769 atomic_read(&SM_I(sbi)->dcc_info->queued_discard)) 2770 return true; 2771 2772 if (SM_I(sbi) && SM_I(sbi)->fcc_info && 2773 atomic_read(&SM_I(sbi)->fcc_info->queued_flush)) 2774 return true; 2775 return false; 2776 } 2777 2778 static inline bool is_idle(struct f2fs_sb_info *sbi, int type) 2779 { 2780 if (sbi->gc_mode == GC_URGENT_HIGH) 2781 return true; 2782 2783 if (is_inflight_io(sbi, type)) 2784 return false; 2785 2786 if (sbi->gc_mode == GC_URGENT_LOW && 2787 (type == DISCARD_TIME || type == GC_TIME)) 2788 return true; 2789 2790 return f2fs_time_over(sbi, type); 2791 } 2792 2793 static inline void f2fs_radix_tree_insert(struct radix_tree_root *root, 2794 unsigned long index, void *item) 2795 { 2796 while (radix_tree_insert(root, index, item)) 2797 cond_resched(); 2798 } 2799 2800 #define RAW_IS_INODE(p) ((p)->footer.nid == (p)->footer.ino) 2801 2802 static inline bool IS_INODE(struct page *page) 2803 { 2804 struct f2fs_node *p = F2FS_NODE(page); 2805 2806 return RAW_IS_INODE(p); 2807 } 2808 2809 static inline int offset_in_addr(struct f2fs_inode *i) 2810 { 2811 return (i->i_inline & F2FS_EXTRA_ATTR) ? 2812 (le16_to_cpu(i->i_extra_isize) / sizeof(__le32)) : 0; 2813 } 2814 2815 static inline __le32 *blkaddr_in_node(struct f2fs_node *node) 2816 { 2817 return RAW_IS_INODE(node) ? node->i.i_addr : node->dn.addr; 2818 } 2819 2820 static inline int f2fs_has_extra_attr(struct inode *inode); 2821 static inline block_t data_blkaddr(struct inode *inode, 2822 struct page *node_page, unsigned int offset) 2823 { 2824 struct f2fs_node *raw_node; 2825 __le32 *addr_array; 2826 int base = 0; 2827 bool is_inode = IS_INODE(node_page); 2828 2829 raw_node = F2FS_NODE(node_page); 2830 2831 if (is_inode) { 2832 if (!inode) 2833 /* from GC path only */ 2834 base = offset_in_addr(&raw_node->i); 2835 else if (f2fs_has_extra_attr(inode)) 2836 base = get_extra_isize(inode); 2837 } 2838 2839 addr_array = blkaddr_in_node(raw_node); 2840 return le32_to_cpu(addr_array[base + offset]); 2841 } 2842 2843 static inline block_t f2fs_data_blkaddr(struct dnode_of_data *dn) 2844 { 2845 return data_blkaddr(dn->inode, dn->node_page, dn->ofs_in_node); 2846 } 2847 2848 static inline int f2fs_test_bit(unsigned int nr, char *addr) 2849 { 2850 int mask; 2851 2852 addr += (nr >> 3); 2853 mask = 1 << (7 - (nr & 0x07)); 2854 return mask & *addr; 2855 } 2856 2857 static inline void f2fs_set_bit(unsigned int nr, char *addr) 2858 { 2859 int mask; 2860 2861 addr += (nr >> 3); 2862 mask = 1 << (7 - (nr & 0x07)); 2863 *addr |= mask; 2864 } 2865 2866 static inline void f2fs_clear_bit(unsigned int nr, char *addr) 2867 { 2868 int mask; 2869 2870 addr += (nr >> 3); 2871 mask = 1 << (7 - (nr & 0x07)); 2872 *addr &= ~mask; 2873 } 2874 2875 static inline int f2fs_test_and_set_bit(unsigned int nr, char *addr) 2876 { 2877 int mask; 2878 int ret; 2879 2880 addr += (nr >> 3); 2881 mask = 1 << (7 - (nr & 0x07)); 2882 ret = mask & *addr; 2883 *addr |= mask; 2884 return ret; 2885 } 2886 2887 static inline int f2fs_test_and_clear_bit(unsigned int nr, char *addr) 2888 { 2889 int mask; 2890 int ret; 2891 2892 addr += (nr >> 3); 2893 mask = 1 << (7 - (nr & 0x07)); 2894 ret = mask & *addr; 2895 *addr &= ~mask; 2896 return ret; 2897 } 2898 2899 static inline void f2fs_change_bit(unsigned int nr, char *addr) 2900 { 2901 int mask; 2902 2903 addr += (nr >> 3); 2904 mask = 1 << (7 - (nr & 0x07)); 2905 *addr ^= mask; 2906 } 2907 2908 /* 2909 * On-disk inode flags (f2fs_inode::i_flags) 2910 */ 2911 #define F2FS_COMPR_FL 0x00000004 /* Compress file */ 2912 #define F2FS_SYNC_FL 0x00000008 /* Synchronous updates */ 2913 #define F2FS_IMMUTABLE_FL 0x00000010 /* Immutable file */ 2914 #define F2FS_APPEND_FL 0x00000020 /* writes to file may only append */ 2915 #define F2FS_NODUMP_FL 0x00000040 /* do not dump file */ 2916 #define F2FS_NOATIME_FL 0x00000080 /* do not update atime */ 2917 #define F2FS_NOCOMP_FL 0x00000400 /* Don't compress */ 2918 #define F2FS_INDEX_FL 0x00001000 /* hash-indexed directory */ 2919 #define F2FS_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */ 2920 #define F2FS_PROJINHERIT_FL 0x20000000 /* Create with parents projid */ 2921 #define F2FS_CASEFOLD_FL 0x40000000 /* Casefolded file */ 2922 2923 /* Flags that should be inherited by new inodes from their parent. */ 2924 #define F2FS_FL_INHERITED (F2FS_SYNC_FL | F2FS_NODUMP_FL | F2FS_NOATIME_FL | \ 2925 F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \ 2926 F2FS_CASEFOLD_FL | F2FS_COMPR_FL | F2FS_NOCOMP_FL) 2927 2928 /* Flags that are appropriate for regular files (all but dir-specific ones). */ 2929 #define F2FS_REG_FLMASK (~(F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \ 2930 F2FS_CASEFOLD_FL)) 2931 2932 /* Flags that are appropriate for non-directories/regular files. */ 2933 #define F2FS_OTHER_FLMASK (F2FS_NODUMP_FL | F2FS_NOATIME_FL) 2934 2935 static inline __u32 f2fs_mask_flags(umode_t mode, __u32 flags) 2936 { 2937 if (S_ISDIR(mode)) 2938 return flags; 2939 else if (S_ISREG(mode)) 2940 return flags & F2FS_REG_FLMASK; 2941 else 2942 return flags & F2FS_OTHER_FLMASK; 2943 } 2944 2945 static inline void __mark_inode_dirty_flag(struct inode *inode, 2946 int flag, bool set) 2947 { 2948 switch (flag) { 2949 case FI_INLINE_XATTR: 2950 case FI_INLINE_DATA: 2951 case FI_INLINE_DENTRY: 2952 case FI_NEW_INODE: 2953 if (set) 2954 return; 2955 fallthrough; 2956 case FI_DATA_EXIST: 2957 case FI_INLINE_DOTS: 2958 case FI_PIN_FILE: 2959 case FI_COMPRESS_RELEASED: 2960 f2fs_mark_inode_dirty_sync(inode, true); 2961 } 2962 } 2963 2964 static inline void set_inode_flag(struct inode *inode, int flag) 2965 { 2966 set_bit(flag, F2FS_I(inode)->flags); 2967 __mark_inode_dirty_flag(inode, flag, true); 2968 } 2969 2970 static inline int is_inode_flag_set(struct inode *inode, int flag) 2971 { 2972 return test_bit(flag, F2FS_I(inode)->flags); 2973 } 2974 2975 static inline void clear_inode_flag(struct inode *inode, int flag) 2976 { 2977 clear_bit(flag, F2FS_I(inode)->flags); 2978 __mark_inode_dirty_flag(inode, flag, false); 2979 } 2980 2981 static inline bool f2fs_verity_in_progress(struct inode *inode) 2982 { 2983 return IS_ENABLED(CONFIG_FS_VERITY) && 2984 is_inode_flag_set(inode, FI_VERITY_IN_PROGRESS); 2985 } 2986 2987 static inline void set_acl_inode(struct inode *inode, umode_t mode) 2988 { 2989 F2FS_I(inode)->i_acl_mode = mode; 2990 set_inode_flag(inode, FI_ACL_MODE); 2991 f2fs_mark_inode_dirty_sync(inode, false); 2992 } 2993 2994 static inline void f2fs_i_links_write(struct inode *inode, bool inc) 2995 { 2996 if (inc) 2997 inc_nlink(inode); 2998 else 2999 drop_nlink(inode); 3000 f2fs_mark_inode_dirty_sync(inode, true); 3001 } 3002 3003 static inline void f2fs_i_blocks_write(struct inode *inode, 3004 block_t diff, bool add, bool claim) 3005 { 3006 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE); 3007 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER); 3008 3009 /* add = 1, claim = 1 should be dquot_reserve_block in pair */ 3010 if (add) { 3011 if (claim) 3012 dquot_claim_block(inode, diff); 3013 else 3014 dquot_alloc_block_nofail(inode, diff); 3015 } else { 3016 dquot_free_block(inode, diff); 3017 } 3018 3019 f2fs_mark_inode_dirty_sync(inode, true); 3020 if (clean || recover) 3021 set_inode_flag(inode, FI_AUTO_RECOVER); 3022 } 3023 3024 static inline void f2fs_i_size_write(struct inode *inode, loff_t i_size) 3025 { 3026 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE); 3027 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER); 3028 3029 if (i_size_read(inode) == i_size) 3030 return; 3031 3032 i_size_write(inode, i_size); 3033 f2fs_mark_inode_dirty_sync(inode, true); 3034 if (clean || recover) 3035 set_inode_flag(inode, FI_AUTO_RECOVER); 3036 } 3037 3038 static inline void f2fs_i_depth_write(struct inode *inode, unsigned int depth) 3039 { 3040 F2FS_I(inode)->i_current_depth = depth; 3041 f2fs_mark_inode_dirty_sync(inode, true); 3042 } 3043 3044 static inline void f2fs_i_gc_failures_write(struct inode *inode, 3045 unsigned int count) 3046 { 3047 F2FS_I(inode)->i_gc_failures[GC_FAILURE_PIN] = count; 3048 f2fs_mark_inode_dirty_sync(inode, true); 3049 } 3050 3051 static inline void f2fs_i_xnid_write(struct inode *inode, nid_t xnid) 3052 { 3053 F2FS_I(inode)->i_xattr_nid = xnid; 3054 f2fs_mark_inode_dirty_sync(inode, true); 3055 } 3056 3057 static inline void f2fs_i_pino_write(struct inode *inode, nid_t pino) 3058 { 3059 F2FS_I(inode)->i_pino = pino; 3060 f2fs_mark_inode_dirty_sync(inode, true); 3061 } 3062 3063 static inline void get_inline_info(struct inode *inode, struct f2fs_inode *ri) 3064 { 3065 struct f2fs_inode_info *fi = F2FS_I(inode); 3066 3067 if (ri->i_inline & F2FS_INLINE_XATTR) 3068 set_bit(FI_INLINE_XATTR, fi->flags); 3069 if (ri->i_inline & F2FS_INLINE_DATA) 3070 set_bit(FI_INLINE_DATA, fi->flags); 3071 if (ri->i_inline & F2FS_INLINE_DENTRY) 3072 set_bit(FI_INLINE_DENTRY, fi->flags); 3073 if (ri->i_inline & F2FS_DATA_EXIST) 3074 set_bit(FI_DATA_EXIST, fi->flags); 3075 if (ri->i_inline & F2FS_INLINE_DOTS) 3076 set_bit(FI_INLINE_DOTS, fi->flags); 3077 if (ri->i_inline & F2FS_EXTRA_ATTR) 3078 set_bit(FI_EXTRA_ATTR, fi->flags); 3079 if (ri->i_inline & F2FS_PIN_FILE) 3080 set_bit(FI_PIN_FILE, fi->flags); 3081 if (ri->i_inline & F2FS_COMPRESS_RELEASED) 3082 set_bit(FI_COMPRESS_RELEASED, fi->flags); 3083 } 3084 3085 static inline void set_raw_inline(struct inode *inode, struct f2fs_inode *ri) 3086 { 3087 ri->i_inline = 0; 3088 3089 if (is_inode_flag_set(inode, FI_INLINE_XATTR)) 3090 ri->i_inline |= F2FS_INLINE_XATTR; 3091 if (is_inode_flag_set(inode, FI_INLINE_DATA)) 3092 ri->i_inline |= F2FS_INLINE_DATA; 3093 if (is_inode_flag_set(inode, FI_INLINE_DENTRY)) 3094 ri->i_inline |= F2FS_INLINE_DENTRY; 3095 if (is_inode_flag_set(inode, FI_DATA_EXIST)) 3096 ri->i_inline |= F2FS_DATA_EXIST; 3097 if (is_inode_flag_set(inode, FI_INLINE_DOTS)) 3098 ri->i_inline |= F2FS_INLINE_DOTS; 3099 if (is_inode_flag_set(inode, FI_EXTRA_ATTR)) 3100 ri->i_inline |= F2FS_EXTRA_ATTR; 3101 if (is_inode_flag_set(inode, FI_PIN_FILE)) 3102 ri->i_inline |= F2FS_PIN_FILE; 3103 if (is_inode_flag_set(inode, FI_COMPRESS_RELEASED)) 3104 ri->i_inline |= F2FS_COMPRESS_RELEASED; 3105 } 3106 3107 static inline int f2fs_has_extra_attr(struct inode *inode) 3108 { 3109 return is_inode_flag_set(inode, FI_EXTRA_ATTR); 3110 } 3111 3112 static inline int f2fs_has_inline_xattr(struct inode *inode) 3113 { 3114 return is_inode_flag_set(inode, FI_INLINE_XATTR); 3115 } 3116 3117 static inline int f2fs_compressed_file(struct inode *inode) 3118 { 3119 return S_ISREG(inode->i_mode) && 3120 is_inode_flag_set(inode, FI_COMPRESSED_FILE); 3121 } 3122 3123 static inline bool f2fs_need_compress_data(struct inode *inode) 3124 { 3125 int compress_mode = F2FS_OPTION(F2FS_I_SB(inode)).compress_mode; 3126 3127 if (!f2fs_compressed_file(inode)) 3128 return false; 3129 3130 if (compress_mode == COMPR_MODE_FS) 3131 return true; 3132 else if (compress_mode == COMPR_MODE_USER && 3133 is_inode_flag_set(inode, FI_ENABLE_COMPRESS)) 3134 return true; 3135 3136 return false; 3137 } 3138 3139 static inline unsigned int addrs_per_inode(struct inode *inode) 3140 { 3141 unsigned int addrs = CUR_ADDRS_PER_INODE(inode) - 3142 get_inline_xattr_addrs(inode); 3143 3144 if (!f2fs_compressed_file(inode)) 3145 return addrs; 3146 return ALIGN_DOWN(addrs, F2FS_I(inode)->i_cluster_size); 3147 } 3148 3149 static inline unsigned int addrs_per_block(struct inode *inode) 3150 { 3151 if (!f2fs_compressed_file(inode)) 3152 return DEF_ADDRS_PER_BLOCK; 3153 return ALIGN_DOWN(DEF_ADDRS_PER_BLOCK, F2FS_I(inode)->i_cluster_size); 3154 } 3155 3156 static inline void *inline_xattr_addr(struct inode *inode, struct page *page) 3157 { 3158 struct f2fs_inode *ri = F2FS_INODE(page); 3159 3160 return (void *)&(ri->i_addr[DEF_ADDRS_PER_INODE - 3161 get_inline_xattr_addrs(inode)]); 3162 } 3163 3164 static inline int inline_xattr_size(struct inode *inode) 3165 { 3166 if (f2fs_has_inline_xattr(inode)) 3167 return get_inline_xattr_addrs(inode) * sizeof(__le32); 3168 return 0; 3169 } 3170 3171 static inline int f2fs_has_inline_data(struct inode *inode) 3172 { 3173 return is_inode_flag_set(inode, FI_INLINE_DATA); 3174 } 3175 3176 static inline int f2fs_exist_data(struct inode *inode) 3177 { 3178 return is_inode_flag_set(inode, FI_DATA_EXIST); 3179 } 3180 3181 static inline int f2fs_has_inline_dots(struct inode *inode) 3182 { 3183 return is_inode_flag_set(inode, FI_INLINE_DOTS); 3184 } 3185 3186 static inline int f2fs_is_mmap_file(struct inode *inode) 3187 { 3188 return is_inode_flag_set(inode, FI_MMAP_FILE); 3189 } 3190 3191 static inline bool f2fs_is_pinned_file(struct inode *inode) 3192 { 3193 return is_inode_flag_set(inode, FI_PIN_FILE); 3194 } 3195 3196 static inline bool f2fs_is_atomic_file(struct inode *inode) 3197 { 3198 return is_inode_flag_set(inode, FI_ATOMIC_FILE); 3199 } 3200 3201 static inline bool f2fs_is_commit_atomic_write(struct inode *inode) 3202 { 3203 return is_inode_flag_set(inode, FI_ATOMIC_COMMIT); 3204 } 3205 3206 static inline bool f2fs_is_volatile_file(struct inode *inode) 3207 { 3208 return is_inode_flag_set(inode, FI_VOLATILE_FILE); 3209 } 3210 3211 static inline bool f2fs_is_first_block_written(struct inode *inode) 3212 { 3213 return is_inode_flag_set(inode, FI_FIRST_BLOCK_WRITTEN); 3214 } 3215 3216 static inline bool f2fs_is_drop_cache(struct inode *inode) 3217 { 3218 return is_inode_flag_set(inode, FI_DROP_CACHE); 3219 } 3220 3221 static inline void *inline_data_addr(struct inode *inode, struct page *page) 3222 { 3223 struct f2fs_inode *ri = F2FS_INODE(page); 3224 int extra_size = get_extra_isize(inode); 3225 3226 return (void *)&(ri->i_addr[extra_size + DEF_INLINE_RESERVED_SIZE]); 3227 } 3228 3229 static inline int f2fs_has_inline_dentry(struct inode *inode) 3230 { 3231 return is_inode_flag_set(inode, FI_INLINE_DENTRY); 3232 } 3233 3234 static inline int is_file(struct inode *inode, int type) 3235 { 3236 return F2FS_I(inode)->i_advise & type; 3237 } 3238 3239 static inline void set_file(struct inode *inode, int type) 3240 { 3241 if (is_file(inode, type)) 3242 return; 3243 F2FS_I(inode)->i_advise |= type; 3244 f2fs_mark_inode_dirty_sync(inode, true); 3245 } 3246 3247 static inline void clear_file(struct inode *inode, int type) 3248 { 3249 if (!is_file(inode, type)) 3250 return; 3251 F2FS_I(inode)->i_advise &= ~type; 3252 f2fs_mark_inode_dirty_sync(inode, true); 3253 } 3254 3255 static inline bool f2fs_is_time_consistent(struct inode *inode) 3256 { 3257 if (!timespec64_equal(F2FS_I(inode)->i_disk_time, &inode->i_atime)) 3258 return false; 3259 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 1, &inode->i_ctime)) 3260 return false; 3261 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 2, &inode->i_mtime)) 3262 return false; 3263 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 3, 3264 &F2FS_I(inode)->i_crtime)) 3265 return false; 3266 return true; 3267 } 3268 3269 static inline bool f2fs_skip_inode_update(struct inode *inode, int dsync) 3270 { 3271 bool ret; 3272 3273 if (dsync) { 3274 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 3275 3276 spin_lock(&sbi->inode_lock[DIRTY_META]); 3277 ret = list_empty(&F2FS_I(inode)->gdirty_list); 3278 spin_unlock(&sbi->inode_lock[DIRTY_META]); 3279 return ret; 3280 } 3281 if (!is_inode_flag_set(inode, FI_AUTO_RECOVER) || 3282 file_keep_isize(inode) || 3283 i_size_read(inode) & ~PAGE_MASK) 3284 return false; 3285 3286 if (!f2fs_is_time_consistent(inode)) 3287 return false; 3288 3289 spin_lock(&F2FS_I(inode)->i_size_lock); 3290 ret = F2FS_I(inode)->last_disk_size == i_size_read(inode); 3291 spin_unlock(&F2FS_I(inode)->i_size_lock); 3292 3293 return ret; 3294 } 3295 3296 static inline bool f2fs_readonly(struct super_block *sb) 3297 { 3298 return sb_rdonly(sb); 3299 } 3300 3301 static inline bool f2fs_cp_error(struct f2fs_sb_info *sbi) 3302 { 3303 return is_set_ckpt_flags(sbi, CP_ERROR_FLAG); 3304 } 3305 3306 static inline bool is_dot_dotdot(const u8 *name, size_t len) 3307 { 3308 if (len == 1 && name[0] == '.') 3309 return true; 3310 3311 if (len == 2 && name[0] == '.' && name[1] == '.') 3312 return true; 3313 3314 return false; 3315 } 3316 3317 static inline void *f2fs_kmalloc(struct f2fs_sb_info *sbi, 3318 size_t size, gfp_t flags) 3319 { 3320 if (time_to_inject(sbi, FAULT_KMALLOC)) { 3321 f2fs_show_injection_info(sbi, FAULT_KMALLOC); 3322 return NULL; 3323 } 3324 3325 return kmalloc(size, flags); 3326 } 3327 3328 static inline void *f2fs_kzalloc(struct f2fs_sb_info *sbi, 3329 size_t size, gfp_t flags) 3330 { 3331 return f2fs_kmalloc(sbi, size, flags | __GFP_ZERO); 3332 } 3333 3334 static inline void *f2fs_kvmalloc(struct f2fs_sb_info *sbi, 3335 size_t size, gfp_t flags) 3336 { 3337 if (time_to_inject(sbi, FAULT_KVMALLOC)) { 3338 f2fs_show_injection_info(sbi, FAULT_KVMALLOC); 3339 return NULL; 3340 } 3341 3342 return kvmalloc(size, flags); 3343 } 3344 3345 static inline void *f2fs_kvzalloc(struct f2fs_sb_info *sbi, 3346 size_t size, gfp_t flags) 3347 { 3348 return f2fs_kvmalloc(sbi, size, flags | __GFP_ZERO); 3349 } 3350 3351 static inline int get_extra_isize(struct inode *inode) 3352 { 3353 return F2FS_I(inode)->i_extra_isize / sizeof(__le32); 3354 } 3355 3356 static inline int get_inline_xattr_addrs(struct inode *inode) 3357 { 3358 return F2FS_I(inode)->i_inline_xattr_size; 3359 } 3360 3361 #define f2fs_get_inode_mode(i) \ 3362 ((is_inode_flag_set(i, FI_ACL_MODE)) ? \ 3363 (F2FS_I(i)->i_acl_mode) : ((i)->i_mode)) 3364 3365 #define F2FS_TOTAL_EXTRA_ATTR_SIZE \ 3366 (offsetof(struct f2fs_inode, i_extra_end) - \ 3367 offsetof(struct f2fs_inode, i_extra_isize)) \ 3368 3369 #define F2FS_OLD_ATTRIBUTE_SIZE (offsetof(struct f2fs_inode, i_addr)) 3370 #define F2FS_FITS_IN_INODE(f2fs_inode, extra_isize, field) \ 3371 ((offsetof(typeof(*(f2fs_inode)), field) + \ 3372 sizeof((f2fs_inode)->field)) \ 3373 <= (F2FS_OLD_ATTRIBUTE_SIZE + (extra_isize))) \ 3374 3375 #define __is_large_section(sbi) ((sbi)->segs_per_sec > 1) 3376 3377 #define __is_meta_io(fio) (PAGE_TYPE_OF_BIO((fio)->type) == META) 3378 3379 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi, 3380 block_t blkaddr, int type); 3381 static inline void verify_blkaddr(struct f2fs_sb_info *sbi, 3382 block_t blkaddr, int type) 3383 { 3384 if (!f2fs_is_valid_blkaddr(sbi, blkaddr, type)) { 3385 f2fs_err(sbi, "invalid blkaddr: %u, type: %d, run fsck to fix.", 3386 blkaddr, type); 3387 f2fs_bug_on(sbi, 1); 3388 } 3389 } 3390 3391 static inline bool __is_valid_data_blkaddr(block_t blkaddr) 3392 { 3393 if (blkaddr == NEW_ADDR || blkaddr == NULL_ADDR || 3394 blkaddr == COMPRESS_ADDR) 3395 return false; 3396 return true; 3397 } 3398 3399 /* 3400 * file.c 3401 */ 3402 int f2fs_sync_file(struct file *file, loff_t start, loff_t end, int datasync); 3403 void f2fs_truncate_data_blocks(struct dnode_of_data *dn); 3404 int f2fs_do_truncate_blocks(struct inode *inode, u64 from, bool lock); 3405 int f2fs_truncate_blocks(struct inode *inode, u64 from, bool lock); 3406 int f2fs_truncate(struct inode *inode); 3407 int f2fs_getattr(struct user_namespace *mnt_userns, const struct path *path, 3408 struct kstat *stat, u32 request_mask, unsigned int flags); 3409 int f2fs_setattr(struct user_namespace *mnt_userns, struct dentry *dentry, 3410 struct iattr *attr); 3411 int f2fs_truncate_hole(struct inode *inode, pgoff_t pg_start, pgoff_t pg_end); 3412 void f2fs_truncate_data_blocks_range(struct dnode_of_data *dn, int count); 3413 int f2fs_precache_extents(struct inode *inode); 3414 int f2fs_fileattr_get(struct dentry *dentry, struct fileattr *fa); 3415 int f2fs_fileattr_set(struct user_namespace *mnt_userns, 3416 struct dentry *dentry, struct fileattr *fa); 3417 long f2fs_ioctl(struct file *filp, unsigned int cmd, unsigned long arg); 3418 long f2fs_compat_ioctl(struct file *file, unsigned int cmd, unsigned long arg); 3419 int f2fs_transfer_project_quota(struct inode *inode, kprojid_t kprojid); 3420 int f2fs_pin_file_control(struct inode *inode, bool inc); 3421 3422 /* 3423 * inode.c 3424 */ 3425 void f2fs_set_inode_flags(struct inode *inode); 3426 bool f2fs_inode_chksum_verify(struct f2fs_sb_info *sbi, struct page *page); 3427 void f2fs_inode_chksum_set(struct f2fs_sb_info *sbi, struct page *page); 3428 struct inode *f2fs_iget(struct super_block *sb, unsigned long ino); 3429 struct inode *f2fs_iget_retry(struct super_block *sb, unsigned long ino); 3430 int f2fs_try_to_free_nats(struct f2fs_sb_info *sbi, int nr_shrink); 3431 void f2fs_update_inode(struct inode *inode, struct page *node_page); 3432 void f2fs_update_inode_page(struct inode *inode); 3433 int f2fs_write_inode(struct inode *inode, struct writeback_control *wbc); 3434 void f2fs_evict_inode(struct inode *inode); 3435 void f2fs_handle_failed_inode(struct inode *inode); 3436 3437 /* 3438 * namei.c 3439 */ 3440 int f2fs_update_extension_list(struct f2fs_sb_info *sbi, const char *name, 3441 bool hot, bool set); 3442 struct dentry *f2fs_get_parent(struct dentry *child); 3443 3444 /* 3445 * dir.c 3446 */ 3447 unsigned char f2fs_get_de_type(struct f2fs_dir_entry *de); 3448 int f2fs_init_casefolded_name(const struct inode *dir, 3449 struct f2fs_filename *fname); 3450 int f2fs_setup_filename(struct inode *dir, const struct qstr *iname, 3451 int lookup, struct f2fs_filename *fname); 3452 int f2fs_prepare_lookup(struct inode *dir, struct dentry *dentry, 3453 struct f2fs_filename *fname); 3454 void f2fs_free_filename(struct f2fs_filename *fname); 3455 struct f2fs_dir_entry *f2fs_find_target_dentry(const struct f2fs_dentry_ptr *d, 3456 const struct f2fs_filename *fname, int *max_slots); 3457 int f2fs_fill_dentries(struct dir_context *ctx, struct f2fs_dentry_ptr *d, 3458 unsigned int start_pos, struct fscrypt_str *fstr); 3459 void f2fs_do_make_empty_dir(struct inode *inode, struct inode *parent, 3460 struct f2fs_dentry_ptr *d); 3461 struct page *f2fs_init_inode_metadata(struct inode *inode, struct inode *dir, 3462 const struct f2fs_filename *fname, struct page *dpage); 3463 void f2fs_update_parent_metadata(struct inode *dir, struct inode *inode, 3464 unsigned int current_depth); 3465 int f2fs_room_for_filename(const void *bitmap, int slots, int max_slots); 3466 void f2fs_drop_nlink(struct inode *dir, struct inode *inode); 3467 struct f2fs_dir_entry *__f2fs_find_entry(struct inode *dir, 3468 const struct f2fs_filename *fname, 3469 struct page **res_page); 3470 struct f2fs_dir_entry *f2fs_find_entry(struct inode *dir, 3471 const struct qstr *child, struct page **res_page); 3472 struct f2fs_dir_entry *f2fs_parent_dir(struct inode *dir, struct page **p); 3473 ino_t f2fs_inode_by_name(struct inode *dir, const struct qstr *qstr, 3474 struct page **page); 3475 void f2fs_set_link(struct inode *dir, struct f2fs_dir_entry *de, 3476 struct page *page, struct inode *inode); 3477 bool f2fs_has_enough_room(struct inode *dir, struct page *ipage, 3478 const struct f2fs_filename *fname); 3479 void f2fs_update_dentry(nid_t ino, umode_t mode, struct f2fs_dentry_ptr *d, 3480 const struct fscrypt_str *name, f2fs_hash_t name_hash, 3481 unsigned int bit_pos); 3482 int f2fs_add_regular_entry(struct inode *dir, const struct f2fs_filename *fname, 3483 struct inode *inode, nid_t ino, umode_t mode); 3484 int f2fs_add_dentry(struct inode *dir, const struct f2fs_filename *fname, 3485 struct inode *inode, nid_t ino, umode_t mode); 3486 int f2fs_do_add_link(struct inode *dir, const struct qstr *name, 3487 struct inode *inode, nid_t ino, umode_t mode); 3488 void f2fs_delete_entry(struct f2fs_dir_entry *dentry, struct page *page, 3489 struct inode *dir, struct inode *inode); 3490 int f2fs_do_tmpfile(struct inode *inode, struct inode *dir); 3491 bool f2fs_empty_dir(struct inode *dir); 3492 3493 static inline int f2fs_add_link(struct dentry *dentry, struct inode *inode) 3494 { 3495 if (fscrypt_is_nokey_name(dentry)) 3496 return -ENOKEY; 3497 return f2fs_do_add_link(d_inode(dentry->d_parent), &dentry->d_name, 3498 inode, inode->i_ino, inode->i_mode); 3499 } 3500 3501 /* 3502 * super.c 3503 */ 3504 int f2fs_inode_dirtied(struct inode *inode, bool sync); 3505 void f2fs_inode_synced(struct inode *inode); 3506 int f2fs_dquot_initialize(struct inode *inode); 3507 int f2fs_enable_quota_files(struct f2fs_sb_info *sbi, bool rdonly); 3508 int f2fs_quota_sync(struct super_block *sb, int type); 3509 loff_t max_file_blocks(struct inode *inode); 3510 void f2fs_quota_off_umount(struct super_block *sb); 3511 int f2fs_commit_super(struct f2fs_sb_info *sbi, bool recover); 3512 int f2fs_sync_fs(struct super_block *sb, int sync); 3513 int f2fs_sanity_check_ckpt(struct f2fs_sb_info *sbi); 3514 3515 /* 3516 * hash.c 3517 */ 3518 void f2fs_hash_filename(const struct inode *dir, struct f2fs_filename *fname); 3519 3520 /* 3521 * node.c 3522 */ 3523 struct node_info; 3524 3525 int f2fs_check_nid_range(struct f2fs_sb_info *sbi, nid_t nid); 3526 bool f2fs_available_free_memory(struct f2fs_sb_info *sbi, int type); 3527 bool f2fs_in_warm_node_list(struct f2fs_sb_info *sbi, struct page *page); 3528 void f2fs_init_fsync_node_info(struct f2fs_sb_info *sbi); 3529 void f2fs_del_fsync_node_entry(struct f2fs_sb_info *sbi, struct page *page); 3530 void f2fs_reset_fsync_node_info(struct f2fs_sb_info *sbi); 3531 int f2fs_need_dentry_mark(struct f2fs_sb_info *sbi, nid_t nid); 3532 bool f2fs_is_checkpointed_node(struct f2fs_sb_info *sbi, nid_t nid); 3533 bool f2fs_need_inode_block_update(struct f2fs_sb_info *sbi, nid_t ino); 3534 int f2fs_get_node_info(struct f2fs_sb_info *sbi, nid_t nid, 3535 struct node_info *ni, bool checkpoint_context); 3536 pgoff_t f2fs_get_next_page_offset(struct dnode_of_data *dn, pgoff_t pgofs); 3537 int f2fs_get_dnode_of_data(struct dnode_of_data *dn, pgoff_t index, int mode); 3538 int f2fs_truncate_inode_blocks(struct inode *inode, pgoff_t from); 3539 int f2fs_truncate_xattr_node(struct inode *inode); 3540 int f2fs_wait_on_node_pages_writeback(struct f2fs_sb_info *sbi, 3541 unsigned int seq_id); 3542 bool f2fs_nat_bitmap_enabled(struct f2fs_sb_info *sbi); 3543 int f2fs_remove_inode_page(struct inode *inode); 3544 struct page *f2fs_new_inode_page(struct inode *inode); 3545 struct page *f2fs_new_node_page(struct dnode_of_data *dn, unsigned int ofs); 3546 void f2fs_ra_node_page(struct f2fs_sb_info *sbi, nid_t nid); 3547 struct page *f2fs_get_node_page(struct f2fs_sb_info *sbi, pgoff_t nid); 3548 struct page *f2fs_get_node_page_ra(struct page *parent, int start); 3549 int f2fs_move_node_page(struct page *node_page, int gc_type); 3550 void f2fs_flush_inline_data(struct f2fs_sb_info *sbi); 3551 int f2fs_fsync_node_pages(struct f2fs_sb_info *sbi, struct inode *inode, 3552 struct writeback_control *wbc, bool atomic, 3553 unsigned int *seq_id); 3554 int f2fs_sync_node_pages(struct f2fs_sb_info *sbi, 3555 struct writeback_control *wbc, 3556 bool do_balance, enum iostat_type io_type); 3557 int f2fs_build_free_nids(struct f2fs_sb_info *sbi, bool sync, bool mount); 3558 bool f2fs_alloc_nid(struct f2fs_sb_info *sbi, nid_t *nid); 3559 void f2fs_alloc_nid_done(struct f2fs_sb_info *sbi, nid_t nid); 3560 void f2fs_alloc_nid_failed(struct f2fs_sb_info *sbi, nid_t nid); 3561 int f2fs_try_to_free_nids(struct f2fs_sb_info *sbi, int nr_shrink); 3562 int f2fs_recover_inline_xattr(struct inode *inode, struct page *page); 3563 int f2fs_recover_xattr_data(struct inode *inode, struct page *page); 3564 int f2fs_recover_inode_page(struct f2fs_sb_info *sbi, struct page *page); 3565 int f2fs_restore_node_summary(struct f2fs_sb_info *sbi, 3566 unsigned int segno, struct f2fs_summary_block *sum); 3567 void f2fs_enable_nat_bits(struct f2fs_sb_info *sbi); 3568 int f2fs_flush_nat_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3569 int f2fs_build_node_manager(struct f2fs_sb_info *sbi); 3570 void f2fs_destroy_node_manager(struct f2fs_sb_info *sbi); 3571 int __init f2fs_create_node_manager_caches(void); 3572 void f2fs_destroy_node_manager_caches(void); 3573 3574 /* 3575 * segment.c 3576 */ 3577 bool f2fs_need_SSR(struct f2fs_sb_info *sbi); 3578 void f2fs_register_inmem_page(struct inode *inode, struct page *page); 3579 void f2fs_drop_inmem_pages_all(struct f2fs_sb_info *sbi, bool gc_failure); 3580 void f2fs_drop_inmem_pages(struct inode *inode); 3581 void f2fs_drop_inmem_page(struct inode *inode, struct page *page); 3582 int f2fs_commit_inmem_pages(struct inode *inode); 3583 void f2fs_balance_fs(struct f2fs_sb_info *sbi, bool need); 3584 void f2fs_balance_fs_bg(struct f2fs_sb_info *sbi, bool from_bg); 3585 int f2fs_issue_flush(struct f2fs_sb_info *sbi, nid_t ino); 3586 int f2fs_create_flush_cmd_control(struct f2fs_sb_info *sbi); 3587 int f2fs_flush_device_cache(struct f2fs_sb_info *sbi); 3588 void f2fs_destroy_flush_cmd_control(struct f2fs_sb_info *sbi, bool free); 3589 void f2fs_invalidate_blocks(struct f2fs_sb_info *sbi, block_t addr); 3590 bool f2fs_is_checkpointed_data(struct f2fs_sb_info *sbi, block_t blkaddr); 3591 int f2fs_start_discard_thread(struct f2fs_sb_info *sbi); 3592 void f2fs_drop_discard_cmd(struct f2fs_sb_info *sbi); 3593 void f2fs_stop_discard_thread(struct f2fs_sb_info *sbi); 3594 bool f2fs_issue_discard_timeout(struct f2fs_sb_info *sbi); 3595 void f2fs_clear_prefree_segments(struct f2fs_sb_info *sbi, 3596 struct cp_control *cpc); 3597 void f2fs_dirty_to_prefree(struct f2fs_sb_info *sbi); 3598 block_t f2fs_get_unusable_blocks(struct f2fs_sb_info *sbi); 3599 int f2fs_disable_cp_again(struct f2fs_sb_info *sbi, block_t unusable); 3600 void f2fs_release_discard_addrs(struct f2fs_sb_info *sbi); 3601 int f2fs_npages_for_summary_flush(struct f2fs_sb_info *sbi, bool for_ra); 3602 bool f2fs_segment_has_free_slot(struct f2fs_sb_info *sbi, int segno); 3603 void f2fs_init_inmem_curseg(struct f2fs_sb_info *sbi); 3604 void f2fs_save_inmem_curseg(struct f2fs_sb_info *sbi); 3605 void f2fs_restore_inmem_curseg(struct f2fs_sb_info *sbi); 3606 void f2fs_get_new_segment(struct f2fs_sb_info *sbi, 3607 unsigned int *newseg, bool new_sec, int dir); 3608 void f2fs_allocate_segment_for_resize(struct f2fs_sb_info *sbi, int type, 3609 unsigned int start, unsigned int end); 3610 void f2fs_allocate_new_section(struct f2fs_sb_info *sbi, int type, bool force); 3611 void f2fs_allocate_new_segments(struct f2fs_sb_info *sbi); 3612 int f2fs_trim_fs(struct f2fs_sb_info *sbi, struct fstrim_range *range); 3613 bool f2fs_exist_trim_candidates(struct f2fs_sb_info *sbi, 3614 struct cp_control *cpc); 3615 struct page *f2fs_get_sum_page(struct f2fs_sb_info *sbi, unsigned int segno); 3616 void f2fs_update_meta_page(struct f2fs_sb_info *sbi, void *src, 3617 block_t blk_addr); 3618 void f2fs_do_write_meta_page(struct f2fs_sb_info *sbi, struct page *page, 3619 enum iostat_type io_type); 3620 void f2fs_do_write_node_page(unsigned int nid, struct f2fs_io_info *fio); 3621 void f2fs_outplace_write_data(struct dnode_of_data *dn, 3622 struct f2fs_io_info *fio); 3623 int f2fs_inplace_write_data(struct f2fs_io_info *fio); 3624 void f2fs_do_replace_block(struct f2fs_sb_info *sbi, struct f2fs_summary *sum, 3625 block_t old_blkaddr, block_t new_blkaddr, 3626 bool recover_curseg, bool recover_newaddr, 3627 bool from_gc); 3628 void f2fs_replace_block(struct f2fs_sb_info *sbi, struct dnode_of_data *dn, 3629 block_t old_addr, block_t new_addr, 3630 unsigned char version, bool recover_curseg, 3631 bool recover_newaddr); 3632 void f2fs_allocate_data_block(struct f2fs_sb_info *sbi, struct page *page, 3633 block_t old_blkaddr, block_t *new_blkaddr, 3634 struct f2fs_summary *sum, int type, 3635 struct f2fs_io_info *fio); 3636 void f2fs_update_device_state(struct f2fs_sb_info *sbi, nid_t ino, 3637 block_t blkaddr, unsigned int blkcnt); 3638 void f2fs_wait_on_page_writeback(struct page *page, 3639 enum page_type type, bool ordered, bool locked); 3640 void f2fs_wait_on_block_writeback(struct inode *inode, block_t blkaddr); 3641 void f2fs_wait_on_block_writeback_range(struct inode *inode, block_t blkaddr, 3642 block_t len); 3643 void f2fs_write_data_summaries(struct f2fs_sb_info *sbi, block_t start_blk); 3644 void f2fs_write_node_summaries(struct f2fs_sb_info *sbi, block_t start_blk); 3645 int f2fs_lookup_journal_in_cursum(struct f2fs_journal *journal, int type, 3646 unsigned int val, int alloc); 3647 void f2fs_flush_sit_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3648 int f2fs_fix_curseg_write_pointer(struct f2fs_sb_info *sbi); 3649 int f2fs_check_write_pointer(struct f2fs_sb_info *sbi); 3650 int f2fs_build_segment_manager(struct f2fs_sb_info *sbi); 3651 void f2fs_destroy_segment_manager(struct f2fs_sb_info *sbi); 3652 int __init f2fs_create_segment_manager_caches(void); 3653 void f2fs_destroy_segment_manager_caches(void); 3654 int f2fs_rw_hint_to_seg_type(enum rw_hint hint); 3655 enum rw_hint f2fs_io_type_to_rw_hint(struct f2fs_sb_info *sbi, 3656 enum page_type type, enum temp_type temp); 3657 unsigned int f2fs_usable_segs_in_sec(struct f2fs_sb_info *sbi, 3658 unsigned int segno); 3659 unsigned int f2fs_usable_blks_in_seg(struct f2fs_sb_info *sbi, 3660 unsigned int segno); 3661 3662 #define DEF_FRAGMENT_SIZE 4 3663 #define MIN_FRAGMENT_SIZE 1 3664 #define MAX_FRAGMENT_SIZE 512 3665 3666 static inline bool f2fs_need_rand_seg(struct f2fs_sb_info *sbi) 3667 { 3668 return F2FS_OPTION(sbi).fs_mode == FS_MODE_FRAGMENT_SEG || 3669 F2FS_OPTION(sbi).fs_mode == FS_MODE_FRAGMENT_BLK; 3670 } 3671 3672 /* 3673 * checkpoint.c 3674 */ 3675 void f2fs_stop_checkpoint(struct f2fs_sb_info *sbi, bool end_io); 3676 struct page *f2fs_grab_meta_page(struct f2fs_sb_info *sbi, pgoff_t index); 3677 struct page *f2fs_get_meta_page(struct f2fs_sb_info *sbi, pgoff_t index); 3678 struct page *f2fs_get_meta_page_retry(struct f2fs_sb_info *sbi, pgoff_t index); 3679 struct page *f2fs_get_tmp_page(struct f2fs_sb_info *sbi, pgoff_t index); 3680 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi, 3681 block_t blkaddr, int type); 3682 int f2fs_ra_meta_pages(struct f2fs_sb_info *sbi, block_t start, int nrpages, 3683 int type, bool sync); 3684 void f2fs_ra_meta_pages_cond(struct f2fs_sb_info *sbi, pgoff_t index, 3685 unsigned int ra_blocks); 3686 long f2fs_sync_meta_pages(struct f2fs_sb_info *sbi, enum page_type type, 3687 long nr_to_write, enum iostat_type io_type); 3688 void f2fs_add_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type); 3689 void f2fs_remove_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type); 3690 void f2fs_release_ino_entry(struct f2fs_sb_info *sbi, bool all); 3691 bool f2fs_exist_written_data(struct f2fs_sb_info *sbi, nid_t ino, int mode); 3692 void f2fs_set_dirty_device(struct f2fs_sb_info *sbi, nid_t ino, 3693 unsigned int devidx, int type); 3694 bool f2fs_is_dirty_device(struct f2fs_sb_info *sbi, nid_t ino, 3695 unsigned int devidx, int type); 3696 int f2fs_sync_inode_meta(struct f2fs_sb_info *sbi); 3697 int f2fs_acquire_orphan_inode(struct f2fs_sb_info *sbi); 3698 void f2fs_release_orphan_inode(struct f2fs_sb_info *sbi); 3699 void f2fs_add_orphan_inode(struct inode *inode); 3700 void f2fs_remove_orphan_inode(struct f2fs_sb_info *sbi, nid_t ino); 3701 int f2fs_recover_orphan_inodes(struct f2fs_sb_info *sbi); 3702 int f2fs_get_valid_checkpoint(struct f2fs_sb_info *sbi); 3703 void f2fs_update_dirty_page(struct inode *inode, struct page *page); 3704 void f2fs_remove_dirty_inode(struct inode *inode); 3705 int f2fs_sync_dirty_inodes(struct f2fs_sb_info *sbi, enum inode_type type); 3706 void f2fs_wait_on_all_pages(struct f2fs_sb_info *sbi, int type); 3707 u64 f2fs_get_sectors_written(struct f2fs_sb_info *sbi); 3708 int f2fs_write_checkpoint(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3709 void f2fs_init_ino_entry_info(struct f2fs_sb_info *sbi); 3710 int __init f2fs_create_checkpoint_caches(void); 3711 void f2fs_destroy_checkpoint_caches(void); 3712 int f2fs_issue_checkpoint(struct f2fs_sb_info *sbi); 3713 int f2fs_start_ckpt_thread(struct f2fs_sb_info *sbi); 3714 void f2fs_stop_ckpt_thread(struct f2fs_sb_info *sbi); 3715 void f2fs_init_ckpt_req_control(struct f2fs_sb_info *sbi); 3716 3717 /* 3718 * data.c 3719 */ 3720 int __init f2fs_init_bioset(void); 3721 void f2fs_destroy_bioset(void); 3722 int f2fs_init_bio_entry_cache(void); 3723 void f2fs_destroy_bio_entry_cache(void); 3724 void f2fs_submit_bio(struct f2fs_sb_info *sbi, 3725 struct bio *bio, enum page_type type); 3726 void f2fs_submit_merged_write(struct f2fs_sb_info *sbi, enum page_type type); 3727 void f2fs_submit_merged_write_cond(struct f2fs_sb_info *sbi, 3728 struct inode *inode, struct page *page, 3729 nid_t ino, enum page_type type); 3730 void f2fs_submit_merged_ipu_write(struct f2fs_sb_info *sbi, 3731 struct bio **bio, struct page *page); 3732 void f2fs_flush_merged_writes(struct f2fs_sb_info *sbi); 3733 int f2fs_submit_page_bio(struct f2fs_io_info *fio); 3734 int f2fs_merge_page_bio(struct f2fs_io_info *fio); 3735 void f2fs_submit_page_write(struct f2fs_io_info *fio); 3736 struct block_device *f2fs_target_device(struct f2fs_sb_info *sbi, 3737 block_t blk_addr, struct bio *bio); 3738 int f2fs_target_device_index(struct f2fs_sb_info *sbi, block_t blkaddr); 3739 void f2fs_set_data_blkaddr(struct dnode_of_data *dn); 3740 void f2fs_update_data_blkaddr(struct dnode_of_data *dn, block_t blkaddr); 3741 int f2fs_reserve_new_blocks(struct dnode_of_data *dn, blkcnt_t count); 3742 int f2fs_reserve_new_block(struct dnode_of_data *dn); 3743 int f2fs_get_block(struct dnode_of_data *dn, pgoff_t index); 3744 int f2fs_reserve_block(struct dnode_of_data *dn, pgoff_t index); 3745 struct page *f2fs_get_read_data_page(struct inode *inode, pgoff_t index, 3746 int op_flags, bool for_write); 3747 struct page *f2fs_find_data_page(struct inode *inode, pgoff_t index); 3748 struct page *f2fs_get_lock_data_page(struct inode *inode, pgoff_t index, 3749 bool for_write); 3750 struct page *f2fs_get_new_data_page(struct inode *inode, 3751 struct page *ipage, pgoff_t index, bool new_i_size); 3752 int f2fs_do_write_data_page(struct f2fs_io_info *fio); 3753 void f2fs_do_map_lock(struct f2fs_sb_info *sbi, int flag, bool lock); 3754 int f2fs_map_blocks(struct inode *inode, struct f2fs_map_blocks *map, 3755 int create, int flag); 3756 int f2fs_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, 3757 u64 start, u64 len); 3758 int f2fs_encrypt_one_page(struct f2fs_io_info *fio); 3759 bool f2fs_should_update_inplace(struct inode *inode, struct f2fs_io_info *fio); 3760 bool f2fs_should_update_outplace(struct inode *inode, struct f2fs_io_info *fio); 3761 int f2fs_write_single_data_page(struct page *page, int *submitted, 3762 struct bio **bio, sector_t *last_block, 3763 struct writeback_control *wbc, 3764 enum iostat_type io_type, 3765 int compr_blocks, bool allow_balance); 3766 void f2fs_write_failed(struct inode *inode, loff_t to); 3767 void f2fs_invalidate_page(struct page *page, unsigned int offset, 3768 unsigned int length); 3769 int f2fs_release_page(struct page *page, gfp_t wait); 3770 #ifdef CONFIG_MIGRATION 3771 int f2fs_migrate_page(struct address_space *mapping, struct page *newpage, 3772 struct page *page, enum migrate_mode mode); 3773 #endif 3774 bool f2fs_overwrite_io(struct inode *inode, loff_t pos, size_t len); 3775 void f2fs_clear_page_cache_dirty_tag(struct page *page); 3776 int f2fs_init_post_read_processing(void); 3777 void f2fs_destroy_post_read_processing(void); 3778 int f2fs_init_post_read_wq(struct f2fs_sb_info *sbi); 3779 void f2fs_destroy_post_read_wq(struct f2fs_sb_info *sbi); 3780 extern const struct iomap_ops f2fs_iomap_ops; 3781 3782 /* 3783 * gc.c 3784 */ 3785 int f2fs_start_gc_thread(struct f2fs_sb_info *sbi); 3786 void f2fs_stop_gc_thread(struct f2fs_sb_info *sbi); 3787 block_t f2fs_start_bidx_of_node(unsigned int node_ofs, struct inode *inode); 3788 int f2fs_gc(struct f2fs_sb_info *sbi, bool sync, bool background, bool force, 3789 unsigned int segno); 3790 void f2fs_build_gc_manager(struct f2fs_sb_info *sbi); 3791 int f2fs_resize_fs(struct f2fs_sb_info *sbi, __u64 block_count); 3792 int __init f2fs_create_garbage_collection_cache(void); 3793 void f2fs_destroy_garbage_collection_cache(void); 3794 3795 /* 3796 * recovery.c 3797 */ 3798 int f2fs_recover_fsync_data(struct f2fs_sb_info *sbi, bool check_only); 3799 bool f2fs_space_for_roll_forward(struct f2fs_sb_info *sbi); 3800 int __init f2fs_create_recovery_cache(void); 3801 void f2fs_destroy_recovery_cache(void); 3802 3803 /* 3804 * debug.c 3805 */ 3806 #ifdef CONFIG_F2FS_STAT_FS 3807 struct f2fs_stat_info { 3808 struct list_head stat_list; 3809 struct f2fs_sb_info *sbi; 3810 int all_area_segs, sit_area_segs, nat_area_segs, ssa_area_segs; 3811 int main_area_segs, main_area_sections, main_area_zones; 3812 unsigned long long hit_largest, hit_cached, hit_rbtree; 3813 unsigned long long hit_total, total_ext; 3814 int ext_tree, zombie_tree, ext_node; 3815 int ndirty_node, ndirty_dent, ndirty_meta, ndirty_imeta; 3816 int ndirty_data, ndirty_qdata; 3817 int inmem_pages; 3818 unsigned int ndirty_dirs, ndirty_files, nquota_files, ndirty_all; 3819 int nats, dirty_nats, sits, dirty_sits; 3820 int free_nids, avail_nids, alloc_nids; 3821 int total_count, utilization; 3822 int bg_gc, nr_wb_cp_data, nr_wb_data; 3823 int nr_rd_data, nr_rd_node, nr_rd_meta; 3824 int nr_dio_read, nr_dio_write; 3825 unsigned int io_skip_bggc, other_skip_bggc; 3826 int nr_flushing, nr_flushed, flush_list_empty; 3827 int nr_discarding, nr_discarded; 3828 int nr_discard_cmd; 3829 unsigned int undiscard_blks; 3830 int nr_issued_ckpt, nr_total_ckpt, nr_queued_ckpt; 3831 unsigned int cur_ckpt_time, peak_ckpt_time; 3832 int inline_xattr, inline_inode, inline_dir, append, update, orphans; 3833 int compr_inode; 3834 unsigned long long compr_blocks; 3835 int aw_cnt, max_aw_cnt, vw_cnt, max_vw_cnt; 3836 unsigned int valid_count, valid_node_count, valid_inode_count, discard_blks; 3837 unsigned int bimodal, avg_vblocks; 3838 int util_free, util_valid, util_invalid; 3839 int rsvd_segs, overp_segs; 3840 int dirty_count, node_pages, meta_pages, compress_pages; 3841 int compress_page_hit; 3842 int prefree_count, call_count, cp_count, bg_cp_count; 3843 int tot_segs, node_segs, data_segs, free_segs, free_secs; 3844 int bg_node_segs, bg_data_segs; 3845 int tot_blks, data_blks, node_blks; 3846 int bg_data_blks, bg_node_blks; 3847 unsigned long long skipped_atomic_files[2]; 3848 int curseg[NR_CURSEG_TYPE]; 3849 int cursec[NR_CURSEG_TYPE]; 3850 int curzone[NR_CURSEG_TYPE]; 3851 unsigned int dirty_seg[NR_CURSEG_TYPE]; 3852 unsigned int full_seg[NR_CURSEG_TYPE]; 3853 unsigned int valid_blks[NR_CURSEG_TYPE]; 3854 3855 unsigned int meta_count[META_MAX]; 3856 unsigned int segment_count[2]; 3857 unsigned int block_count[2]; 3858 unsigned int inplace_count; 3859 unsigned long long base_mem, cache_mem, page_mem; 3860 }; 3861 3862 static inline struct f2fs_stat_info *F2FS_STAT(struct f2fs_sb_info *sbi) 3863 { 3864 return (struct f2fs_stat_info *)sbi->stat_info; 3865 } 3866 3867 #define stat_inc_cp_count(si) ((si)->cp_count++) 3868 #define stat_inc_bg_cp_count(si) ((si)->bg_cp_count++) 3869 #define stat_inc_call_count(si) ((si)->call_count++) 3870 #define stat_inc_bggc_count(si) ((si)->bg_gc++) 3871 #define stat_io_skip_bggc_count(sbi) ((sbi)->io_skip_bggc++) 3872 #define stat_other_skip_bggc_count(sbi) ((sbi)->other_skip_bggc++) 3873 #define stat_inc_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]++) 3874 #define stat_dec_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]--) 3875 #define stat_inc_total_hit(sbi) (atomic64_inc(&(sbi)->total_hit_ext)) 3876 #define stat_inc_rbtree_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_rbtree)) 3877 #define stat_inc_largest_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_largest)) 3878 #define stat_inc_cached_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_cached)) 3879 #define stat_inc_inline_xattr(inode) \ 3880 do { \ 3881 if (f2fs_has_inline_xattr(inode)) \ 3882 (atomic_inc(&F2FS_I_SB(inode)->inline_xattr)); \ 3883 } while (0) 3884 #define stat_dec_inline_xattr(inode) \ 3885 do { \ 3886 if (f2fs_has_inline_xattr(inode)) \ 3887 (atomic_dec(&F2FS_I_SB(inode)->inline_xattr)); \ 3888 } while (0) 3889 #define stat_inc_inline_inode(inode) \ 3890 do { \ 3891 if (f2fs_has_inline_data(inode)) \ 3892 (atomic_inc(&F2FS_I_SB(inode)->inline_inode)); \ 3893 } while (0) 3894 #define stat_dec_inline_inode(inode) \ 3895 do { \ 3896 if (f2fs_has_inline_data(inode)) \ 3897 (atomic_dec(&F2FS_I_SB(inode)->inline_inode)); \ 3898 } while (0) 3899 #define stat_inc_inline_dir(inode) \ 3900 do { \ 3901 if (f2fs_has_inline_dentry(inode)) \ 3902 (atomic_inc(&F2FS_I_SB(inode)->inline_dir)); \ 3903 } while (0) 3904 #define stat_dec_inline_dir(inode) \ 3905 do { \ 3906 if (f2fs_has_inline_dentry(inode)) \ 3907 (atomic_dec(&F2FS_I_SB(inode)->inline_dir)); \ 3908 } while (0) 3909 #define stat_inc_compr_inode(inode) \ 3910 do { \ 3911 if (f2fs_compressed_file(inode)) \ 3912 (atomic_inc(&F2FS_I_SB(inode)->compr_inode)); \ 3913 } while (0) 3914 #define stat_dec_compr_inode(inode) \ 3915 do { \ 3916 if (f2fs_compressed_file(inode)) \ 3917 (atomic_dec(&F2FS_I_SB(inode)->compr_inode)); \ 3918 } while (0) 3919 #define stat_add_compr_blocks(inode, blocks) \ 3920 (atomic64_add(blocks, &F2FS_I_SB(inode)->compr_blocks)) 3921 #define stat_sub_compr_blocks(inode, blocks) \ 3922 (atomic64_sub(blocks, &F2FS_I_SB(inode)->compr_blocks)) 3923 #define stat_inc_meta_count(sbi, blkaddr) \ 3924 do { \ 3925 if (blkaddr < SIT_I(sbi)->sit_base_addr) \ 3926 atomic_inc(&(sbi)->meta_count[META_CP]); \ 3927 else if (blkaddr < NM_I(sbi)->nat_blkaddr) \ 3928 atomic_inc(&(sbi)->meta_count[META_SIT]); \ 3929 else if (blkaddr < SM_I(sbi)->ssa_blkaddr) \ 3930 atomic_inc(&(sbi)->meta_count[META_NAT]); \ 3931 else if (blkaddr < SM_I(sbi)->main_blkaddr) \ 3932 atomic_inc(&(sbi)->meta_count[META_SSA]); \ 3933 } while (0) 3934 #define stat_inc_seg_type(sbi, curseg) \ 3935 ((sbi)->segment_count[(curseg)->alloc_type]++) 3936 #define stat_inc_block_count(sbi, curseg) \ 3937 ((sbi)->block_count[(curseg)->alloc_type]++) 3938 #define stat_inc_inplace_blocks(sbi) \ 3939 (atomic_inc(&(sbi)->inplace_count)) 3940 #define stat_update_max_atomic_write(inode) \ 3941 do { \ 3942 int cur = F2FS_I_SB(inode)->atomic_files; \ 3943 int max = atomic_read(&F2FS_I_SB(inode)->max_aw_cnt); \ 3944 if (cur > max) \ 3945 atomic_set(&F2FS_I_SB(inode)->max_aw_cnt, cur); \ 3946 } while (0) 3947 #define stat_inc_volatile_write(inode) \ 3948 (atomic_inc(&F2FS_I_SB(inode)->vw_cnt)) 3949 #define stat_dec_volatile_write(inode) \ 3950 (atomic_dec(&F2FS_I_SB(inode)->vw_cnt)) 3951 #define stat_update_max_volatile_write(inode) \ 3952 do { \ 3953 int cur = atomic_read(&F2FS_I_SB(inode)->vw_cnt); \ 3954 int max = atomic_read(&F2FS_I_SB(inode)->max_vw_cnt); \ 3955 if (cur > max) \ 3956 atomic_set(&F2FS_I_SB(inode)->max_vw_cnt, cur); \ 3957 } while (0) 3958 #define stat_inc_seg_count(sbi, type, gc_type) \ 3959 do { \ 3960 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 3961 si->tot_segs++; \ 3962 if ((type) == SUM_TYPE_DATA) { \ 3963 si->data_segs++; \ 3964 si->bg_data_segs += (gc_type == BG_GC) ? 1 : 0; \ 3965 } else { \ 3966 si->node_segs++; \ 3967 si->bg_node_segs += (gc_type == BG_GC) ? 1 : 0; \ 3968 } \ 3969 } while (0) 3970 3971 #define stat_inc_tot_blk_count(si, blks) \ 3972 ((si)->tot_blks += (blks)) 3973 3974 #define stat_inc_data_blk_count(sbi, blks, gc_type) \ 3975 do { \ 3976 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 3977 stat_inc_tot_blk_count(si, blks); \ 3978 si->data_blks += (blks); \ 3979 si->bg_data_blks += ((gc_type) == BG_GC) ? (blks) : 0; \ 3980 } while (0) 3981 3982 #define stat_inc_node_blk_count(sbi, blks, gc_type) \ 3983 do { \ 3984 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 3985 stat_inc_tot_blk_count(si, blks); \ 3986 si->node_blks += (blks); \ 3987 si->bg_node_blks += ((gc_type) == BG_GC) ? (blks) : 0; \ 3988 } while (0) 3989 3990 int f2fs_build_stats(struct f2fs_sb_info *sbi); 3991 void f2fs_destroy_stats(struct f2fs_sb_info *sbi); 3992 void __init f2fs_create_root_stats(void); 3993 void f2fs_destroy_root_stats(void); 3994 void f2fs_update_sit_info(struct f2fs_sb_info *sbi); 3995 #else 3996 #define stat_inc_cp_count(si) do { } while (0) 3997 #define stat_inc_bg_cp_count(si) do { } while (0) 3998 #define stat_inc_call_count(si) do { } while (0) 3999 #define stat_inc_bggc_count(si) do { } while (0) 4000 #define stat_io_skip_bggc_count(sbi) do { } while (0) 4001 #define stat_other_skip_bggc_count(sbi) do { } while (0) 4002 #define stat_inc_dirty_inode(sbi, type) do { } while (0) 4003 #define stat_dec_dirty_inode(sbi, type) do { } while (0) 4004 #define stat_inc_total_hit(sbi) do { } while (0) 4005 #define stat_inc_rbtree_node_hit(sbi) do { } while (0) 4006 #define stat_inc_largest_node_hit(sbi) do { } while (0) 4007 #define stat_inc_cached_node_hit(sbi) do { } while (0) 4008 #define stat_inc_inline_xattr(inode) do { } while (0) 4009 #define stat_dec_inline_xattr(inode) do { } while (0) 4010 #define stat_inc_inline_inode(inode) do { } while (0) 4011 #define stat_dec_inline_inode(inode) do { } while (0) 4012 #define stat_inc_inline_dir(inode) do { } while (0) 4013 #define stat_dec_inline_dir(inode) do { } while (0) 4014 #define stat_inc_compr_inode(inode) do { } while (0) 4015 #define stat_dec_compr_inode(inode) do { } while (0) 4016 #define stat_add_compr_blocks(inode, blocks) do { } while (0) 4017 #define stat_sub_compr_blocks(inode, blocks) do { } while (0) 4018 #define stat_update_max_atomic_write(inode) do { } while (0) 4019 #define stat_inc_volatile_write(inode) do { } while (0) 4020 #define stat_dec_volatile_write(inode) do { } while (0) 4021 #define stat_update_max_volatile_write(inode) do { } while (0) 4022 #define stat_inc_meta_count(sbi, blkaddr) do { } while (0) 4023 #define stat_inc_seg_type(sbi, curseg) do { } while (0) 4024 #define stat_inc_block_count(sbi, curseg) do { } while (0) 4025 #define stat_inc_inplace_blocks(sbi) do { } while (0) 4026 #define stat_inc_seg_count(sbi, type, gc_type) do { } while (0) 4027 #define stat_inc_tot_blk_count(si, blks) do { } while (0) 4028 #define stat_inc_data_blk_count(sbi, blks, gc_type) do { } while (0) 4029 #define stat_inc_node_blk_count(sbi, blks, gc_type) do { } while (0) 4030 4031 static inline int f2fs_build_stats(struct f2fs_sb_info *sbi) { return 0; } 4032 static inline void f2fs_destroy_stats(struct f2fs_sb_info *sbi) { } 4033 static inline void __init f2fs_create_root_stats(void) { } 4034 static inline void f2fs_destroy_root_stats(void) { } 4035 static inline void f2fs_update_sit_info(struct f2fs_sb_info *sbi) {} 4036 #endif 4037 4038 extern const struct file_operations f2fs_dir_operations; 4039 extern const struct file_operations f2fs_file_operations; 4040 extern const struct inode_operations f2fs_file_inode_operations; 4041 extern const struct address_space_operations f2fs_dblock_aops; 4042 extern const struct address_space_operations f2fs_node_aops; 4043 extern const struct address_space_operations f2fs_meta_aops; 4044 extern const struct inode_operations f2fs_dir_inode_operations; 4045 extern const struct inode_operations f2fs_symlink_inode_operations; 4046 extern const struct inode_operations f2fs_encrypted_symlink_inode_operations; 4047 extern const struct inode_operations f2fs_special_inode_operations; 4048 extern struct kmem_cache *f2fs_inode_entry_slab; 4049 4050 /* 4051 * inline.c 4052 */ 4053 bool f2fs_may_inline_data(struct inode *inode); 4054 bool f2fs_may_inline_dentry(struct inode *inode); 4055 void f2fs_do_read_inline_data(struct page *page, struct page *ipage); 4056 void f2fs_truncate_inline_inode(struct inode *inode, 4057 struct page *ipage, u64 from); 4058 int f2fs_read_inline_data(struct inode *inode, struct page *page); 4059 int f2fs_convert_inline_page(struct dnode_of_data *dn, struct page *page); 4060 int f2fs_convert_inline_inode(struct inode *inode); 4061 int f2fs_try_convert_inline_dir(struct inode *dir, struct dentry *dentry); 4062 int f2fs_write_inline_data(struct inode *inode, struct page *page); 4063 int f2fs_recover_inline_data(struct inode *inode, struct page *npage); 4064 struct f2fs_dir_entry *f2fs_find_in_inline_dir(struct inode *dir, 4065 const struct f2fs_filename *fname, 4066 struct page **res_page); 4067 int f2fs_make_empty_inline_dir(struct inode *inode, struct inode *parent, 4068 struct page *ipage); 4069 int f2fs_add_inline_entry(struct inode *dir, const struct f2fs_filename *fname, 4070 struct inode *inode, nid_t ino, umode_t mode); 4071 void f2fs_delete_inline_entry(struct f2fs_dir_entry *dentry, 4072 struct page *page, struct inode *dir, 4073 struct inode *inode); 4074 bool f2fs_empty_inline_dir(struct inode *dir); 4075 int f2fs_read_inline_dir(struct file *file, struct dir_context *ctx, 4076 struct fscrypt_str *fstr); 4077 int f2fs_inline_data_fiemap(struct inode *inode, 4078 struct fiemap_extent_info *fieinfo, 4079 __u64 start, __u64 len); 4080 4081 /* 4082 * shrinker.c 4083 */ 4084 unsigned long f2fs_shrink_count(struct shrinker *shrink, 4085 struct shrink_control *sc); 4086 unsigned long f2fs_shrink_scan(struct shrinker *shrink, 4087 struct shrink_control *sc); 4088 void f2fs_join_shrinker(struct f2fs_sb_info *sbi); 4089 void f2fs_leave_shrinker(struct f2fs_sb_info *sbi); 4090 4091 /* 4092 * extent_cache.c 4093 */ 4094 struct rb_entry *f2fs_lookup_rb_tree(struct rb_root_cached *root, 4095 struct rb_entry *cached_re, unsigned int ofs); 4096 struct rb_node **f2fs_lookup_rb_tree_ext(struct f2fs_sb_info *sbi, 4097 struct rb_root_cached *root, 4098 struct rb_node **parent, 4099 unsigned long long key, bool *left_most); 4100 struct rb_node **f2fs_lookup_rb_tree_for_insert(struct f2fs_sb_info *sbi, 4101 struct rb_root_cached *root, 4102 struct rb_node **parent, 4103 unsigned int ofs, bool *leftmost); 4104 struct rb_entry *f2fs_lookup_rb_tree_ret(struct rb_root_cached *root, 4105 struct rb_entry *cached_re, unsigned int ofs, 4106 struct rb_entry **prev_entry, struct rb_entry **next_entry, 4107 struct rb_node ***insert_p, struct rb_node **insert_parent, 4108 bool force, bool *leftmost); 4109 bool f2fs_check_rb_tree_consistence(struct f2fs_sb_info *sbi, 4110 struct rb_root_cached *root, bool check_key); 4111 unsigned int f2fs_shrink_extent_tree(struct f2fs_sb_info *sbi, int nr_shrink); 4112 void f2fs_init_extent_tree(struct inode *inode, struct page *ipage); 4113 void f2fs_drop_extent_tree(struct inode *inode); 4114 unsigned int f2fs_destroy_extent_node(struct inode *inode); 4115 void f2fs_destroy_extent_tree(struct inode *inode); 4116 bool f2fs_lookup_extent_cache(struct inode *inode, pgoff_t pgofs, 4117 struct extent_info *ei); 4118 void f2fs_update_extent_cache(struct dnode_of_data *dn); 4119 void f2fs_update_extent_cache_range(struct dnode_of_data *dn, 4120 pgoff_t fofs, block_t blkaddr, unsigned int len); 4121 void f2fs_init_extent_cache_info(struct f2fs_sb_info *sbi); 4122 int __init f2fs_create_extent_cache(void); 4123 void f2fs_destroy_extent_cache(void); 4124 4125 /* 4126 * sysfs.c 4127 */ 4128 #define MIN_RA_MUL 2 4129 #define MAX_RA_MUL 256 4130 4131 int __init f2fs_init_sysfs(void); 4132 void f2fs_exit_sysfs(void); 4133 int f2fs_register_sysfs(struct f2fs_sb_info *sbi); 4134 void f2fs_unregister_sysfs(struct f2fs_sb_info *sbi); 4135 4136 /* verity.c */ 4137 extern const struct fsverity_operations f2fs_verityops; 4138 4139 /* 4140 * crypto support 4141 */ 4142 static inline bool f2fs_encrypted_file(struct inode *inode) 4143 { 4144 return IS_ENCRYPTED(inode) && S_ISREG(inode->i_mode); 4145 } 4146 4147 static inline void f2fs_set_encrypted_inode(struct inode *inode) 4148 { 4149 #ifdef CONFIG_FS_ENCRYPTION 4150 file_set_encrypt(inode); 4151 f2fs_set_inode_flags(inode); 4152 #endif 4153 } 4154 4155 /* 4156 * Returns true if the reads of the inode's data need to undergo some 4157 * postprocessing step, like decryption or authenticity verification. 4158 */ 4159 static inline bool f2fs_post_read_required(struct inode *inode) 4160 { 4161 return f2fs_encrypted_file(inode) || fsverity_active(inode) || 4162 f2fs_compressed_file(inode); 4163 } 4164 4165 /* 4166 * compress.c 4167 */ 4168 #ifdef CONFIG_F2FS_FS_COMPRESSION 4169 bool f2fs_is_compressed_page(struct page *page); 4170 struct page *f2fs_compress_control_page(struct page *page); 4171 int f2fs_prepare_compress_overwrite(struct inode *inode, 4172 struct page **pagep, pgoff_t index, void **fsdata); 4173 bool f2fs_compress_write_end(struct inode *inode, void *fsdata, 4174 pgoff_t index, unsigned copied); 4175 int f2fs_truncate_partial_cluster(struct inode *inode, u64 from, bool lock); 4176 void f2fs_compress_write_end_io(struct bio *bio, struct page *page); 4177 bool f2fs_is_compress_backend_ready(struct inode *inode); 4178 int f2fs_init_compress_mempool(void); 4179 void f2fs_destroy_compress_mempool(void); 4180 void f2fs_decompress_cluster(struct decompress_io_ctx *dic); 4181 void f2fs_end_read_compressed_page(struct page *page, bool failed, 4182 block_t blkaddr); 4183 bool f2fs_cluster_is_empty(struct compress_ctx *cc); 4184 bool f2fs_cluster_can_merge_page(struct compress_ctx *cc, pgoff_t index); 4185 bool f2fs_all_cluster_page_loaded(struct compress_ctx *cc, struct pagevec *pvec, 4186 int index, int nr_pages); 4187 bool f2fs_sanity_check_cluster(struct dnode_of_data *dn); 4188 void f2fs_compress_ctx_add_page(struct compress_ctx *cc, struct page *page); 4189 int f2fs_write_multi_pages(struct compress_ctx *cc, 4190 int *submitted, 4191 struct writeback_control *wbc, 4192 enum iostat_type io_type); 4193 int f2fs_is_compressed_cluster(struct inode *inode, pgoff_t index); 4194 void f2fs_update_extent_tree_range_compressed(struct inode *inode, 4195 pgoff_t fofs, block_t blkaddr, unsigned int llen, 4196 unsigned int c_len); 4197 int f2fs_read_multi_pages(struct compress_ctx *cc, struct bio **bio_ret, 4198 unsigned nr_pages, sector_t *last_block_in_bio, 4199 bool is_readahead, bool for_write); 4200 struct decompress_io_ctx *f2fs_alloc_dic(struct compress_ctx *cc); 4201 void f2fs_decompress_end_io(struct decompress_io_ctx *dic, bool failed); 4202 void f2fs_put_page_dic(struct page *page); 4203 unsigned int f2fs_cluster_blocks_are_contiguous(struct dnode_of_data *dn); 4204 int f2fs_init_compress_ctx(struct compress_ctx *cc); 4205 void f2fs_destroy_compress_ctx(struct compress_ctx *cc, bool reuse); 4206 void f2fs_init_compress_info(struct f2fs_sb_info *sbi); 4207 int f2fs_init_compress_inode(struct f2fs_sb_info *sbi); 4208 void f2fs_destroy_compress_inode(struct f2fs_sb_info *sbi); 4209 int f2fs_init_page_array_cache(struct f2fs_sb_info *sbi); 4210 void f2fs_destroy_page_array_cache(struct f2fs_sb_info *sbi); 4211 int __init f2fs_init_compress_cache(void); 4212 void f2fs_destroy_compress_cache(void); 4213 struct address_space *COMPRESS_MAPPING(struct f2fs_sb_info *sbi); 4214 void f2fs_invalidate_compress_page(struct f2fs_sb_info *sbi, block_t blkaddr); 4215 void f2fs_cache_compressed_page(struct f2fs_sb_info *sbi, struct page *page, 4216 nid_t ino, block_t blkaddr); 4217 bool f2fs_load_compressed_page(struct f2fs_sb_info *sbi, struct page *page, 4218 block_t blkaddr); 4219 void f2fs_invalidate_compress_pages(struct f2fs_sb_info *sbi, nid_t ino); 4220 #define inc_compr_inode_stat(inode) \ 4221 do { \ 4222 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); \ 4223 sbi->compr_new_inode++; \ 4224 } while (0) 4225 #define add_compr_block_stat(inode, blocks) \ 4226 do { \ 4227 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); \ 4228 int diff = F2FS_I(inode)->i_cluster_size - blocks; \ 4229 sbi->compr_written_block += blocks; \ 4230 sbi->compr_saved_block += diff; \ 4231 } while (0) 4232 #else 4233 static inline bool f2fs_is_compressed_page(struct page *page) { return false; } 4234 static inline bool f2fs_is_compress_backend_ready(struct inode *inode) 4235 { 4236 if (!f2fs_compressed_file(inode)) 4237 return true; 4238 /* not support compression */ 4239 return false; 4240 } 4241 static inline struct page *f2fs_compress_control_page(struct page *page) 4242 { 4243 WARN_ON_ONCE(1); 4244 return ERR_PTR(-EINVAL); 4245 } 4246 static inline int f2fs_init_compress_mempool(void) { return 0; } 4247 static inline void f2fs_destroy_compress_mempool(void) { } 4248 static inline void f2fs_decompress_cluster(struct decompress_io_ctx *dic) { } 4249 static inline void f2fs_end_read_compressed_page(struct page *page, 4250 bool failed, block_t blkaddr) 4251 { 4252 WARN_ON_ONCE(1); 4253 } 4254 static inline void f2fs_put_page_dic(struct page *page) 4255 { 4256 WARN_ON_ONCE(1); 4257 } 4258 static inline unsigned int f2fs_cluster_blocks_are_contiguous(struct dnode_of_data *dn) { return 0; } 4259 static inline bool f2fs_sanity_check_cluster(struct dnode_of_data *dn) { return false; } 4260 static inline int f2fs_init_compress_inode(struct f2fs_sb_info *sbi) { return 0; } 4261 static inline void f2fs_destroy_compress_inode(struct f2fs_sb_info *sbi) { } 4262 static inline int f2fs_init_page_array_cache(struct f2fs_sb_info *sbi) { return 0; } 4263 static inline void f2fs_destroy_page_array_cache(struct f2fs_sb_info *sbi) { } 4264 static inline int __init f2fs_init_compress_cache(void) { return 0; } 4265 static inline void f2fs_destroy_compress_cache(void) { } 4266 static inline void f2fs_invalidate_compress_page(struct f2fs_sb_info *sbi, 4267 block_t blkaddr) { } 4268 static inline void f2fs_cache_compressed_page(struct f2fs_sb_info *sbi, 4269 struct page *page, nid_t ino, block_t blkaddr) { } 4270 static inline bool f2fs_load_compressed_page(struct f2fs_sb_info *sbi, 4271 struct page *page, block_t blkaddr) { return false; } 4272 static inline void f2fs_invalidate_compress_pages(struct f2fs_sb_info *sbi, 4273 nid_t ino) { } 4274 #define inc_compr_inode_stat(inode) do { } while (0) 4275 static inline void f2fs_update_extent_tree_range_compressed(struct inode *inode, 4276 pgoff_t fofs, block_t blkaddr, unsigned int llen, 4277 unsigned int c_len) { } 4278 #endif 4279 4280 static inline void set_compress_context(struct inode *inode) 4281 { 4282 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 4283 4284 F2FS_I(inode)->i_compress_algorithm = 4285 F2FS_OPTION(sbi).compress_algorithm; 4286 F2FS_I(inode)->i_log_cluster_size = 4287 F2FS_OPTION(sbi).compress_log_size; 4288 F2FS_I(inode)->i_compress_flag = 4289 F2FS_OPTION(sbi).compress_chksum ? 4290 1 << COMPRESS_CHKSUM : 0; 4291 F2FS_I(inode)->i_cluster_size = 4292 1 << F2FS_I(inode)->i_log_cluster_size; 4293 if ((F2FS_I(inode)->i_compress_algorithm == COMPRESS_LZ4 || 4294 F2FS_I(inode)->i_compress_algorithm == COMPRESS_ZSTD) && 4295 F2FS_OPTION(sbi).compress_level) 4296 F2FS_I(inode)->i_compress_flag |= 4297 F2FS_OPTION(sbi).compress_level << 4298 COMPRESS_LEVEL_OFFSET; 4299 F2FS_I(inode)->i_flags |= F2FS_COMPR_FL; 4300 set_inode_flag(inode, FI_COMPRESSED_FILE); 4301 stat_inc_compr_inode(inode); 4302 inc_compr_inode_stat(inode); 4303 f2fs_mark_inode_dirty_sync(inode, true); 4304 } 4305 4306 static inline bool f2fs_disable_compressed_file(struct inode *inode) 4307 { 4308 struct f2fs_inode_info *fi = F2FS_I(inode); 4309 4310 if (!f2fs_compressed_file(inode)) 4311 return true; 4312 if (S_ISREG(inode->i_mode) && F2FS_HAS_BLOCKS(inode)) 4313 return false; 4314 4315 fi->i_flags &= ~F2FS_COMPR_FL; 4316 stat_dec_compr_inode(inode); 4317 clear_inode_flag(inode, FI_COMPRESSED_FILE); 4318 f2fs_mark_inode_dirty_sync(inode, true); 4319 return true; 4320 } 4321 4322 #define F2FS_FEATURE_FUNCS(name, flagname) \ 4323 static inline int f2fs_sb_has_##name(struct f2fs_sb_info *sbi) \ 4324 { \ 4325 return F2FS_HAS_FEATURE(sbi, F2FS_FEATURE_##flagname); \ 4326 } 4327 4328 F2FS_FEATURE_FUNCS(encrypt, ENCRYPT); 4329 F2FS_FEATURE_FUNCS(blkzoned, BLKZONED); 4330 F2FS_FEATURE_FUNCS(extra_attr, EXTRA_ATTR); 4331 F2FS_FEATURE_FUNCS(project_quota, PRJQUOTA); 4332 F2FS_FEATURE_FUNCS(inode_chksum, INODE_CHKSUM); 4333 F2FS_FEATURE_FUNCS(flexible_inline_xattr, FLEXIBLE_INLINE_XATTR); 4334 F2FS_FEATURE_FUNCS(quota_ino, QUOTA_INO); 4335 F2FS_FEATURE_FUNCS(inode_crtime, INODE_CRTIME); 4336 F2FS_FEATURE_FUNCS(lost_found, LOST_FOUND); 4337 F2FS_FEATURE_FUNCS(verity, VERITY); 4338 F2FS_FEATURE_FUNCS(sb_chksum, SB_CHKSUM); 4339 F2FS_FEATURE_FUNCS(casefold, CASEFOLD); 4340 F2FS_FEATURE_FUNCS(compression, COMPRESSION); 4341 F2FS_FEATURE_FUNCS(readonly, RO); 4342 4343 static inline bool f2fs_may_extent_tree(struct inode *inode) 4344 { 4345 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 4346 4347 if (!test_opt(sbi, EXTENT_CACHE) || 4348 is_inode_flag_set(inode, FI_NO_EXTENT) || 4349 (is_inode_flag_set(inode, FI_COMPRESSED_FILE) && 4350 !f2fs_sb_has_readonly(sbi))) 4351 return false; 4352 4353 /* 4354 * for recovered files during mount do not create extents 4355 * if shrinker is not registered. 4356 */ 4357 if (list_empty(&sbi->s_list)) 4358 return false; 4359 4360 return S_ISREG(inode->i_mode); 4361 } 4362 4363 #ifdef CONFIG_BLK_DEV_ZONED 4364 static inline bool f2fs_blkz_is_seq(struct f2fs_sb_info *sbi, int devi, 4365 block_t blkaddr) 4366 { 4367 unsigned int zno = blkaddr >> sbi->log_blocks_per_blkz; 4368 4369 return test_bit(zno, FDEV(devi).blkz_seq); 4370 } 4371 #endif 4372 4373 static inline bool f2fs_hw_should_discard(struct f2fs_sb_info *sbi) 4374 { 4375 return f2fs_sb_has_blkzoned(sbi); 4376 } 4377 4378 static inline bool f2fs_bdev_support_discard(struct block_device *bdev) 4379 { 4380 return blk_queue_discard(bdev_get_queue(bdev)) || 4381 bdev_is_zoned(bdev); 4382 } 4383 4384 static inline bool f2fs_hw_support_discard(struct f2fs_sb_info *sbi) 4385 { 4386 int i; 4387 4388 if (!f2fs_is_multi_device(sbi)) 4389 return f2fs_bdev_support_discard(sbi->sb->s_bdev); 4390 4391 for (i = 0; i < sbi->s_ndevs; i++) 4392 if (f2fs_bdev_support_discard(FDEV(i).bdev)) 4393 return true; 4394 return false; 4395 } 4396 4397 static inline bool f2fs_realtime_discard_enable(struct f2fs_sb_info *sbi) 4398 { 4399 return (test_opt(sbi, DISCARD) && f2fs_hw_support_discard(sbi)) || 4400 f2fs_hw_should_discard(sbi); 4401 } 4402 4403 static inline bool f2fs_hw_is_readonly(struct f2fs_sb_info *sbi) 4404 { 4405 int i; 4406 4407 if (!f2fs_is_multi_device(sbi)) 4408 return bdev_read_only(sbi->sb->s_bdev); 4409 4410 for (i = 0; i < sbi->s_ndevs; i++) 4411 if (bdev_read_only(FDEV(i).bdev)) 4412 return true; 4413 return false; 4414 } 4415 4416 static inline bool f2fs_lfs_mode(struct f2fs_sb_info *sbi) 4417 { 4418 return F2FS_OPTION(sbi).fs_mode == FS_MODE_LFS; 4419 } 4420 4421 static inline bool f2fs_may_compress(struct inode *inode) 4422 { 4423 if (IS_SWAPFILE(inode) || f2fs_is_pinned_file(inode) || 4424 f2fs_is_atomic_file(inode) || 4425 f2fs_is_volatile_file(inode)) 4426 return false; 4427 return S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode); 4428 } 4429 4430 static inline void f2fs_i_compr_blocks_update(struct inode *inode, 4431 u64 blocks, bool add) 4432 { 4433 int diff = F2FS_I(inode)->i_cluster_size - blocks; 4434 struct f2fs_inode_info *fi = F2FS_I(inode); 4435 4436 /* don't update i_compr_blocks if saved blocks were released */ 4437 if (!add && !atomic_read(&fi->i_compr_blocks)) 4438 return; 4439 4440 if (add) { 4441 atomic_add(diff, &fi->i_compr_blocks); 4442 stat_add_compr_blocks(inode, diff); 4443 } else { 4444 atomic_sub(diff, &fi->i_compr_blocks); 4445 stat_sub_compr_blocks(inode, diff); 4446 } 4447 f2fs_mark_inode_dirty_sync(inode, true); 4448 } 4449 4450 static inline int block_unaligned_IO(struct inode *inode, 4451 struct kiocb *iocb, struct iov_iter *iter) 4452 { 4453 unsigned int i_blkbits = READ_ONCE(inode->i_blkbits); 4454 unsigned int blocksize_mask = (1 << i_blkbits) - 1; 4455 loff_t offset = iocb->ki_pos; 4456 unsigned long align = offset | iov_iter_alignment(iter); 4457 4458 return align & blocksize_mask; 4459 } 4460 4461 static inline bool f2fs_allow_multi_device_dio(struct f2fs_sb_info *sbi, 4462 int flag) 4463 { 4464 if (!f2fs_is_multi_device(sbi)) 4465 return false; 4466 if (flag != F2FS_GET_BLOCK_DIO) 4467 return false; 4468 return sbi->aligned_blksize; 4469 } 4470 4471 static inline bool f2fs_force_buffered_io(struct inode *inode, 4472 struct kiocb *iocb, struct iov_iter *iter) 4473 { 4474 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 4475 int rw = iov_iter_rw(iter); 4476 4477 if (f2fs_post_read_required(inode)) 4478 return true; 4479 4480 /* disallow direct IO if any of devices has unaligned blksize */ 4481 if (f2fs_is_multi_device(sbi) && !sbi->aligned_blksize) 4482 return true; 4483 /* 4484 * for blkzoned device, fallback direct IO to buffered IO, so 4485 * all IOs can be serialized by log-structured write. 4486 */ 4487 if (f2fs_sb_has_blkzoned(sbi)) 4488 return true; 4489 if (f2fs_lfs_mode(sbi) && (rw == WRITE)) { 4490 if (block_unaligned_IO(inode, iocb, iter)) 4491 return true; 4492 if (F2FS_IO_ALIGNED(sbi)) 4493 return true; 4494 } 4495 if (is_sbi_flag_set(F2FS_I_SB(inode), SBI_CP_DISABLED)) 4496 return true; 4497 4498 return false; 4499 } 4500 4501 static inline bool f2fs_need_verity(const struct inode *inode, pgoff_t idx) 4502 { 4503 return fsverity_active(inode) && 4504 idx < DIV_ROUND_UP(inode->i_size, PAGE_SIZE); 4505 } 4506 4507 #ifdef CONFIG_F2FS_FAULT_INJECTION 4508 extern void f2fs_build_fault_attr(struct f2fs_sb_info *sbi, unsigned int rate, 4509 unsigned int type); 4510 #else 4511 #define f2fs_build_fault_attr(sbi, rate, type) do { } while (0) 4512 #endif 4513 4514 static inline bool is_journalled_quota(struct f2fs_sb_info *sbi) 4515 { 4516 #ifdef CONFIG_QUOTA 4517 if (f2fs_sb_has_quota_ino(sbi)) 4518 return true; 4519 if (F2FS_OPTION(sbi).s_qf_names[USRQUOTA] || 4520 F2FS_OPTION(sbi).s_qf_names[GRPQUOTA] || 4521 F2FS_OPTION(sbi).s_qf_names[PRJQUOTA]) 4522 return true; 4523 #endif 4524 return false; 4525 } 4526 4527 static inline bool f2fs_block_unit_discard(struct f2fs_sb_info *sbi) 4528 { 4529 return F2FS_OPTION(sbi).discard_unit == DISCARD_UNIT_BLOCK; 4530 } 4531 4532 #define EFSBADCRC EBADMSG /* Bad CRC detected */ 4533 #define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */ 4534 4535 #endif /* _LINUX_F2FS_H */ 4536