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 #ifdef CONFIG_F2FS_CHECK_FS 32 #define f2fs_bug_on(sbi, condition) BUG_ON(condition) 33 #else 34 #define f2fs_bug_on(sbi, condition) \ 35 do { \ 36 if (WARN_ON(condition)) \ 37 set_sbi_flag(sbi, SBI_NEED_FSCK); \ 38 } while (0) 39 #endif 40 41 enum { 42 FAULT_KMALLOC, 43 FAULT_KVMALLOC, 44 FAULT_PAGE_ALLOC, 45 FAULT_PAGE_GET, 46 FAULT_ALLOC_NID, 47 FAULT_ORPHAN, 48 FAULT_BLOCK, 49 FAULT_DIR_DEPTH, 50 FAULT_EVICT_INODE, 51 FAULT_TRUNCATE, 52 FAULT_READ_IO, 53 FAULT_CHECKPOINT, 54 FAULT_DISCARD, 55 FAULT_WRITE_IO, 56 FAULT_MAX, 57 }; 58 59 #ifdef CONFIG_F2FS_FAULT_INJECTION 60 #define F2FS_ALL_FAULT_TYPE ((1 << FAULT_MAX) - 1) 61 62 struct f2fs_fault_info { 63 atomic_t inject_ops; 64 unsigned int inject_rate; 65 unsigned int inject_type; 66 }; 67 68 extern const char *f2fs_fault_name[FAULT_MAX]; 69 #define IS_FAULT_SET(fi, type) ((fi)->inject_type & (1 << (type))) 70 #endif 71 72 /* 73 * For mount options 74 */ 75 #define F2FS_MOUNT_DISABLE_ROLL_FORWARD 0x00000002 76 #define F2FS_MOUNT_DISCARD 0x00000004 77 #define F2FS_MOUNT_NOHEAP 0x00000008 78 #define F2FS_MOUNT_XATTR_USER 0x00000010 79 #define F2FS_MOUNT_POSIX_ACL 0x00000020 80 #define F2FS_MOUNT_DISABLE_EXT_IDENTIFY 0x00000040 81 #define F2FS_MOUNT_INLINE_XATTR 0x00000080 82 #define F2FS_MOUNT_INLINE_DATA 0x00000100 83 #define F2FS_MOUNT_INLINE_DENTRY 0x00000200 84 #define F2FS_MOUNT_FLUSH_MERGE 0x00000400 85 #define F2FS_MOUNT_NOBARRIER 0x00000800 86 #define F2FS_MOUNT_FASTBOOT 0x00001000 87 #define F2FS_MOUNT_EXTENT_CACHE 0x00002000 88 #define F2FS_MOUNT_DATA_FLUSH 0x00008000 89 #define F2FS_MOUNT_FAULT_INJECTION 0x00010000 90 #define F2FS_MOUNT_USRQUOTA 0x00080000 91 #define F2FS_MOUNT_GRPQUOTA 0x00100000 92 #define F2FS_MOUNT_PRJQUOTA 0x00200000 93 #define F2FS_MOUNT_QUOTA 0x00400000 94 #define F2FS_MOUNT_INLINE_XATTR_SIZE 0x00800000 95 #define F2FS_MOUNT_RESERVE_ROOT 0x01000000 96 #define F2FS_MOUNT_DISABLE_CHECKPOINT 0x02000000 97 #define F2FS_MOUNT_NORECOVERY 0x04000000 98 #define F2FS_MOUNT_ATGC 0x08000000 99 #define F2FS_MOUNT_MERGE_CHECKPOINT 0x10000000 100 #define F2FS_MOUNT_GC_MERGE 0x20000000 101 #define F2FS_MOUNT_COMPRESS_CACHE 0x40000000 102 103 #define F2FS_OPTION(sbi) ((sbi)->mount_opt) 104 #define clear_opt(sbi, option) (F2FS_OPTION(sbi).opt &= ~F2FS_MOUNT_##option) 105 #define set_opt(sbi, option) (F2FS_OPTION(sbi).opt |= F2FS_MOUNT_##option) 106 #define test_opt(sbi, option) (F2FS_OPTION(sbi).opt & F2FS_MOUNT_##option) 107 108 #define ver_after(a, b) (typecheck(unsigned long long, a) && \ 109 typecheck(unsigned long long, b) && \ 110 ((long long)((a) - (b)) > 0)) 111 112 typedef u32 block_t; /* 113 * should not change u32, since it is the on-disk block 114 * address format, __le32. 115 */ 116 typedef u32 nid_t; 117 118 #define COMPRESS_EXT_NUM 16 119 120 struct f2fs_mount_info { 121 unsigned int opt; 122 int write_io_size_bits; /* Write IO size bits */ 123 block_t root_reserved_blocks; /* root reserved blocks */ 124 kuid_t s_resuid; /* reserved blocks for uid */ 125 kgid_t s_resgid; /* reserved blocks for gid */ 126 int active_logs; /* # of active logs */ 127 int inline_xattr_size; /* inline xattr size */ 128 #ifdef CONFIG_F2FS_FAULT_INJECTION 129 struct f2fs_fault_info fault_info; /* For fault injection */ 130 #endif 131 #ifdef CONFIG_QUOTA 132 /* Names of quota files with journalled quota */ 133 char *s_qf_names[MAXQUOTAS]; 134 int s_jquota_fmt; /* Format of quota to use */ 135 #endif 136 /* For which write hints are passed down to block layer */ 137 int whint_mode; 138 int alloc_mode; /* segment allocation policy */ 139 int fsync_mode; /* fsync policy */ 140 int fs_mode; /* fs mode: LFS or ADAPTIVE */ 141 int bggc_mode; /* bggc mode: off, on or sync */ 142 struct fscrypt_dummy_policy dummy_enc_policy; /* test dummy encryption */ 143 block_t unusable_cap_perc; /* percentage for cap */ 144 block_t unusable_cap; /* Amount of space allowed to be 145 * unusable when disabling checkpoint 146 */ 147 148 /* For compression */ 149 unsigned char compress_algorithm; /* algorithm type */ 150 unsigned char compress_log_size; /* cluster log size */ 151 unsigned char compress_level; /* compress level */ 152 bool compress_chksum; /* compressed data chksum */ 153 unsigned char compress_ext_cnt; /* extension count */ 154 unsigned char nocompress_ext_cnt; /* nocompress extension count */ 155 int compress_mode; /* compression mode */ 156 unsigned char extensions[COMPRESS_EXT_NUM][F2FS_EXTENSION_LEN]; /* extensions */ 157 unsigned char noextensions[COMPRESS_EXT_NUM][F2FS_EXTENSION_LEN]; /* extensions */ 158 }; 159 160 #define F2FS_FEATURE_ENCRYPT 0x0001 161 #define F2FS_FEATURE_BLKZONED 0x0002 162 #define F2FS_FEATURE_ATOMIC_WRITE 0x0004 163 #define F2FS_FEATURE_EXTRA_ATTR 0x0008 164 #define F2FS_FEATURE_PRJQUOTA 0x0010 165 #define F2FS_FEATURE_INODE_CHKSUM 0x0020 166 #define F2FS_FEATURE_FLEXIBLE_INLINE_XATTR 0x0040 167 #define F2FS_FEATURE_QUOTA_INO 0x0080 168 #define F2FS_FEATURE_INODE_CRTIME 0x0100 169 #define F2FS_FEATURE_LOST_FOUND 0x0200 170 #define F2FS_FEATURE_VERITY 0x0400 171 #define F2FS_FEATURE_SB_CHKSUM 0x0800 172 #define F2FS_FEATURE_CASEFOLD 0x1000 173 #define F2FS_FEATURE_COMPRESSION 0x2000 174 #define F2FS_FEATURE_RO 0x4000 175 176 #define __F2FS_HAS_FEATURE(raw_super, mask) \ 177 ((raw_super->feature & cpu_to_le32(mask)) != 0) 178 #define F2FS_HAS_FEATURE(sbi, mask) __F2FS_HAS_FEATURE(sbi->raw_super, mask) 179 #define F2FS_SET_FEATURE(sbi, mask) \ 180 (sbi->raw_super->feature |= cpu_to_le32(mask)) 181 #define F2FS_CLEAR_FEATURE(sbi, mask) \ 182 (sbi->raw_super->feature &= ~cpu_to_le32(mask)) 183 184 /* 185 * Default values for user and/or group using reserved blocks 186 */ 187 #define F2FS_DEF_RESUID 0 188 #define F2FS_DEF_RESGID 0 189 190 /* 191 * For checkpoint manager 192 */ 193 enum { 194 NAT_BITMAP, 195 SIT_BITMAP 196 }; 197 198 #define CP_UMOUNT 0x00000001 199 #define CP_FASTBOOT 0x00000002 200 #define CP_SYNC 0x00000004 201 #define CP_RECOVERY 0x00000008 202 #define CP_DISCARD 0x00000010 203 #define CP_TRIMMED 0x00000020 204 #define CP_PAUSE 0x00000040 205 #define CP_RESIZE 0x00000080 206 207 #define MAX_DISCARD_BLOCKS(sbi) BLKS_PER_SEC(sbi) 208 #define DEF_MAX_DISCARD_REQUEST 8 /* issue 8 discards per round */ 209 #define DEF_MIN_DISCARD_ISSUE_TIME 50 /* 50 ms, if exists */ 210 #define DEF_MID_DISCARD_ISSUE_TIME 500 /* 500 ms, if device busy */ 211 #define DEF_MAX_DISCARD_ISSUE_TIME 60000 /* 60 s, if no candidates */ 212 #define DEF_DISCARD_URGENT_UTIL 80 /* do more discard over 80% */ 213 #define DEF_CP_INTERVAL 60 /* 60 secs */ 214 #define DEF_IDLE_INTERVAL 5 /* 5 secs */ 215 #define DEF_DISABLE_INTERVAL 5 /* 5 secs */ 216 #define DEF_DISABLE_QUICK_INTERVAL 1 /* 1 secs */ 217 #define DEF_UMOUNT_DISCARD_TIMEOUT 5 /* 5 secs */ 218 219 struct cp_control { 220 int reason; 221 __u64 trim_start; 222 __u64 trim_end; 223 __u64 trim_minlen; 224 }; 225 226 /* 227 * indicate meta/data type 228 */ 229 enum { 230 META_CP, 231 META_NAT, 232 META_SIT, 233 META_SSA, 234 META_MAX, 235 META_POR, 236 DATA_GENERIC, /* check range only */ 237 DATA_GENERIC_ENHANCE, /* strong check on range and segment bitmap */ 238 DATA_GENERIC_ENHANCE_READ, /* 239 * strong check on range and segment 240 * bitmap but no warning due to race 241 * condition of read on truncated area 242 * by extent_cache 243 */ 244 META_GENERIC, 245 }; 246 247 /* for the list of ino */ 248 enum { 249 ORPHAN_INO, /* for orphan ino list */ 250 APPEND_INO, /* for append ino list */ 251 UPDATE_INO, /* for update ino list */ 252 TRANS_DIR_INO, /* for trasactions dir ino list */ 253 FLUSH_INO, /* for multiple device flushing */ 254 MAX_INO_ENTRY, /* max. list */ 255 }; 256 257 struct ino_entry { 258 struct list_head list; /* list head */ 259 nid_t ino; /* inode number */ 260 unsigned int dirty_device; /* dirty device bitmap */ 261 }; 262 263 /* for the list of inodes to be GCed */ 264 struct inode_entry { 265 struct list_head list; /* list head */ 266 struct inode *inode; /* vfs inode pointer */ 267 }; 268 269 struct fsync_node_entry { 270 struct list_head list; /* list head */ 271 struct page *page; /* warm node page pointer */ 272 unsigned int seq_id; /* sequence id */ 273 }; 274 275 struct ckpt_req { 276 struct completion wait; /* completion for checkpoint done */ 277 struct llist_node llnode; /* llist_node to be linked in wait queue */ 278 int ret; /* return code of checkpoint */ 279 ktime_t queue_time; /* request queued time */ 280 }; 281 282 struct ckpt_req_control { 283 struct task_struct *f2fs_issue_ckpt; /* checkpoint task */ 284 int ckpt_thread_ioprio; /* checkpoint merge thread ioprio */ 285 wait_queue_head_t ckpt_wait_queue; /* waiting queue for wake-up */ 286 atomic_t issued_ckpt; /* # of actually issued ckpts */ 287 atomic_t total_ckpt; /* # of total ckpts */ 288 atomic_t queued_ckpt; /* # of queued ckpts */ 289 struct llist_head issue_list; /* list for command issue */ 290 spinlock_t stat_lock; /* lock for below checkpoint time stats */ 291 unsigned int cur_time; /* cur wait time in msec for currently issued checkpoint */ 292 unsigned int peak_time; /* peak wait time in msec until now */ 293 }; 294 295 /* for the bitmap indicate blocks to be discarded */ 296 struct discard_entry { 297 struct list_head list; /* list head */ 298 block_t start_blkaddr; /* start blockaddr of current segment */ 299 unsigned char discard_map[SIT_VBLOCK_MAP_SIZE]; /* segment discard bitmap */ 300 }; 301 302 /* default discard granularity of inner discard thread, unit: block count */ 303 #define DEFAULT_DISCARD_GRANULARITY 16 304 305 /* max discard pend list number */ 306 #define MAX_PLIST_NUM 512 307 #define plist_idx(blk_num) ((blk_num) >= MAX_PLIST_NUM ? \ 308 (MAX_PLIST_NUM - 1) : ((blk_num) - 1)) 309 310 enum { 311 D_PREP, /* initial */ 312 D_PARTIAL, /* partially submitted */ 313 D_SUBMIT, /* all submitted */ 314 D_DONE, /* finished */ 315 }; 316 317 struct discard_info { 318 block_t lstart; /* logical start address */ 319 block_t len; /* length */ 320 block_t start; /* actual start address in dev */ 321 }; 322 323 struct discard_cmd { 324 struct rb_node rb_node; /* rb node located in rb-tree */ 325 union { 326 struct { 327 block_t lstart; /* logical start address */ 328 block_t len; /* length */ 329 block_t start; /* actual start address in dev */ 330 }; 331 struct discard_info di; /* discard info */ 332 333 }; 334 struct list_head list; /* command list */ 335 struct completion wait; /* compleation */ 336 struct block_device *bdev; /* bdev */ 337 unsigned short ref; /* reference count */ 338 unsigned char state; /* state */ 339 unsigned char queued; /* queued discard */ 340 int error; /* bio error */ 341 spinlock_t lock; /* for state/bio_ref updating */ 342 unsigned short bio_ref; /* bio reference count */ 343 }; 344 345 enum { 346 DPOLICY_BG, 347 DPOLICY_FORCE, 348 DPOLICY_FSTRIM, 349 DPOLICY_UMOUNT, 350 MAX_DPOLICY, 351 }; 352 353 struct discard_policy { 354 int type; /* type of discard */ 355 unsigned int min_interval; /* used for candidates exist */ 356 unsigned int mid_interval; /* used for device busy */ 357 unsigned int max_interval; /* used for candidates not exist */ 358 unsigned int max_requests; /* # of discards issued per round */ 359 unsigned int io_aware_gran; /* minimum granularity discard not be aware of I/O */ 360 bool io_aware; /* issue discard in idle time */ 361 bool sync; /* submit discard with REQ_SYNC flag */ 362 bool ordered; /* issue discard by lba order */ 363 bool timeout; /* discard timeout for put_super */ 364 unsigned int granularity; /* discard granularity */ 365 }; 366 367 struct discard_cmd_control { 368 struct task_struct *f2fs_issue_discard; /* discard thread */ 369 struct list_head entry_list; /* 4KB discard entry list */ 370 struct list_head pend_list[MAX_PLIST_NUM];/* store pending entries */ 371 struct list_head wait_list; /* store on-flushing entries */ 372 struct list_head fstrim_list; /* in-flight discard from fstrim */ 373 wait_queue_head_t discard_wait_queue; /* waiting queue for wake-up */ 374 unsigned int discard_wake; /* to wake up discard thread */ 375 struct mutex cmd_lock; 376 unsigned int nr_discards; /* # of discards in the list */ 377 unsigned int max_discards; /* max. discards to be issued */ 378 unsigned int discard_granularity; /* discard granularity */ 379 unsigned int undiscard_blks; /* # of undiscard blocks */ 380 unsigned int next_pos; /* next discard position */ 381 atomic_t issued_discard; /* # of issued discard */ 382 atomic_t queued_discard; /* # of queued discard */ 383 atomic_t discard_cmd_cnt; /* # of cached cmd count */ 384 struct rb_root_cached root; /* root of discard rb-tree */ 385 bool rbtree_check; /* config for consistence check */ 386 }; 387 388 /* for the list of fsync inodes, used only during recovery */ 389 struct fsync_inode_entry { 390 struct list_head list; /* list head */ 391 struct inode *inode; /* vfs inode pointer */ 392 block_t blkaddr; /* block address locating the last fsync */ 393 block_t last_dentry; /* block address locating the last dentry */ 394 }; 395 396 #define nats_in_cursum(jnl) (le16_to_cpu((jnl)->n_nats)) 397 #define sits_in_cursum(jnl) (le16_to_cpu((jnl)->n_sits)) 398 399 #define nat_in_journal(jnl, i) ((jnl)->nat_j.entries[i].ne) 400 #define nid_in_journal(jnl, i) ((jnl)->nat_j.entries[i].nid) 401 #define sit_in_journal(jnl, i) ((jnl)->sit_j.entries[i].se) 402 #define segno_in_journal(jnl, i) ((jnl)->sit_j.entries[i].segno) 403 404 #define MAX_NAT_JENTRIES(jnl) (NAT_JOURNAL_ENTRIES - nats_in_cursum(jnl)) 405 #define MAX_SIT_JENTRIES(jnl) (SIT_JOURNAL_ENTRIES - sits_in_cursum(jnl)) 406 407 static inline int update_nats_in_cursum(struct f2fs_journal *journal, int i) 408 { 409 int before = nats_in_cursum(journal); 410 411 journal->n_nats = cpu_to_le16(before + i); 412 return before; 413 } 414 415 static inline int update_sits_in_cursum(struct f2fs_journal *journal, int i) 416 { 417 int before = sits_in_cursum(journal); 418 419 journal->n_sits = cpu_to_le16(before + i); 420 return before; 421 } 422 423 static inline bool __has_cursum_space(struct f2fs_journal *journal, 424 int size, int type) 425 { 426 if (type == NAT_JOURNAL) 427 return size <= MAX_NAT_JENTRIES(journal); 428 return size <= MAX_SIT_JENTRIES(journal); 429 } 430 431 /* for inline stuff */ 432 #define DEF_INLINE_RESERVED_SIZE 1 433 static inline int get_extra_isize(struct inode *inode); 434 static inline int get_inline_xattr_addrs(struct inode *inode); 435 #define MAX_INLINE_DATA(inode) (sizeof(__le32) * \ 436 (CUR_ADDRS_PER_INODE(inode) - \ 437 get_inline_xattr_addrs(inode) - \ 438 DEF_INLINE_RESERVED_SIZE)) 439 440 /* for inline dir */ 441 #define NR_INLINE_DENTRY(inode) (MAX_INLINE_DATA(inode) * BITS_PER_BYTE / \ 442 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \ 443 BITS_PER_BYTE + 1)) 444 #define INLINE_DENTRY_BITMAP_SIZE(inode) \ 445 DIV_ROUND_UP(NR_INLINE_DENTRY(inode), BITS_PER_BYTE) 446 #define INLINE_RESERVED_SIZE(inode) (MAX_INLINE_DATA(inode) - \ 447 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \ 448 NR_INLINE_DENTRY(inode) + \ 449 INLINE_DENTRY_BITMAP_SIZE(inode))) 450 451 /* 452 * For INODE and NODE manager 453 */ 454 /* for directory operations */ 455 456 struct f2fs_filename { 457 /* 458 * The filename the user specified. This is NULL for some 459 * filesystem-internal operations, e.g. converting an inline directory 460 * to a non-inline one, or roll-forward recovering an encrypted dentry. 461 */ 462 const struct qstr *usr_fname; 463 464 /* 465 * The on-disk filename. For encrypted directories, this is encrypted. 466 * This may be NULL for lookups in an encrypted dir without the key. 467 */ 468 struct fscrypt_str disk_name; 469 470 /* The dirhash of this filename */ 471 f2fs_hash_t hash; 472 473 #ifdef CONFIG_FS_ENCRYPTION 474 /* 475 * For lookups in encrypted directories: either the buffer backing 476 * disk_name, or a buffer that holds the decoded no-key name. 477 */ 478 struct fscrypt_str crypto_buf; 479 #endif 480 #ifdef CONFIG_UNICODE 481 /* 482 * For casefolded directories: the casefolded name, but it's left NULL 483 * if the original name is not valid Unicode, if the directory is both 484 * casefolded and encrypted and its encryption key is unavailable, or if 485 * the filesystem is doing an internal operation where usr_fname is also 486 * NULL. In all these cases we fall back to treating the name as an 487 * opaque byte sequence. 488 */ 489 struct fscrypt_str cf_name; 490 #endif 491 }; 492 493 struct f2fs_dentry_ptr { 494 struct inode *inode; 495 void *bitmap; 496 struct f2fs_dir_entry *dentry; 497 __u8 (*filename)[F2FS_SLOT_LEN]; 498 int max; 499 int nr_bitmap; 500 }; 501 502 static inline void make_dentry_ptr_block(struct inode *inode, 503 struct f2fs_dentry_ptr *d, struct f2fs_dentry_block *t) 504 { 505 d->inode = inode; 506 d->max = NR_DENTRY_IN_BLOCK; 507 d->nr_bitmap = SIZE_OF_DENTRY_BITMAP; 508 d->bitmap = t->dentry_bitmap; 509 d->dentry = t->dentry; 510 d->filename = t->filename; 511 } 512 513 static inline void make_dentry_ptr_inline(struct inode *inode, 514 struct f2fs_dentry_ptr *d, void *t) 515 { 516 int entry_cnt = NR_INLINE_DENTRY(inode); 517 int bitmap_size = INLINE_DENTRY_BITMAP_SIZE(inode); 518 int reserved_size = INLINE_RESERVED_SIZE(inode); 519 520 d->inode = inode; 521 d->max = entry_cnt; 522 d->nr_bitmap = bitmap_size; 523 d->bitmap = t; 524 d->dentry = t + bitmap_size + reserved_size; 525 d->filename = t + bitmap_size + reserved_size + 526 SIZE_OF_DIR_ENTRY * entry_cnt; 527 } 528 529 /* 530 * XATTR_NODE_OFFSET stores xattrs to one node block per file keeping -1 531 * as its node offset to distinguish from index node blocks. 532 * But some bits are used to mark the node block. 533 */ 534 #define XATTR_NODE_OFFSET ((((unsigned int)-1) << OFFSET_BIT_SHIFT) \ 535 >> OFFSET_BIT_SHIFT) 536 enum { 537 ALLOC_NODE, /* allocate a new node page if needed */ 538 LOOKUP_NODE, /* look up a node without readahead */ 539 LOOKUP_NODE_RA, /* 540 * look up a node with readahead called 541 * by get_data_block. 542 */ 543 }; 544 545 #define DEFAULT_RETRY_IO_COUNT 8 /* maximum retry read IO count */ 546 547 /* congestion wait timeout value, default: 20ms */ 548 #define DEFAULT_IO_TIMEOUT (msecs_to_jiffies(20)) 549 550 /* maximum retry quota flush count */ 551 #define DEFAULT_RETRY_QUOTA_FLUSH_COUNT 8 552 553 #define F2FS_LINK_MAX 0xffffffff /* maximum link count per file */ 554 555 #define MAX_DIR_RA_PAGES 4 /* maximum ra pages of dir */ 556 557 /* for in-memory extent cache entry */ 558 #define F2FS_MIN_EXTENT_LEN 64 /* minimum extent length */ 559 560 /* number of extent info in extent cache we try to shrink */ 561 #define EXTENT_CACHE_SHRINK_NUMBER 128 562 563 struct rb_entry { 564 struct rb_node rb_node; /* rb node located in rb-tree */ 565 union { 566 struct { 567 unsigned int ofs; /* start offset of the entry */ 568 unsigned int len; /* length of the entry */ 569 }; 570 unsigned long long key; /* 64-bits key */ 571 } __packed; 572 }; 573 574 struct extent_info { 575 unsigned int fofs; /* start offset in a file */ 576 unsigned int len; /* length of the extent */ 577 u32 blk; /* start block address of the extent */ 578 }; 579 580 struct extent_node { 581 struct rb_node rb_node; /* rb node located in rb-tree */ 582 struct extent_info ei; /* extent info */ 583 struct list_head list; /* node in global extent list of sbi */ 584 struct extent_tree *et; /* extent tree pointer */ 585 }; 586 587 struct extent_tree { 588 nid_t ino; /* inode number */ 589 struct rb_root_cached root; /* root of extent info rb-tree */ 590 struct extent_node *cached_en; /* recently accessed extent node */ 591 struct extent_info largest; /* largested extent info */ 592 struct list_head list; /* to be used by sbi->zombie_list */ 593 rwlock_t lock; /* protect extent info rb-tree */ 594 atomic_t node_cnt; /* # of extent node in rb-tree*/ 595 bool largest_updated; /* largest extent updated */ 596 }; 597 598 /* 599 * This structure is taken from ext4_map_blocks. 600 * 601 * Note that, however, f2fs uses NEW and MAPPED flags for f2fs_map_blocks(). 602 */ 603 #define F2FS_MAP_NEW (1 << BH_New) 604 #define F2FS_MAP_MAPPED (1 << BH_Mapped) 605 #define F2FS_MAP_UNWRITTEN (1 << BH_Unwritten) 606 #define F2FS_MAP_FLAGS (F2FS_MAP_NEW | F2FS_MAP_MAPPED |\ 607 F2FS_MAP_UNWRITTEN) 608 609 struct f2fs_map_blocks { 610 block_t m_pblk; 611 block_t m_lblk; 612 unsigned int m_len; 613 unsigned int m_flags; 614 pgoff_t *m_next_pgofs; /* point next possible non-hole pgofs */ 615 pgoff_t *m_next_extent; /* point to next possible extent */ 616 int m_seg_type; 617 bool m_may_create; /* indicate it is from write path */ 618 }; 619 620 /* for flag in get_data_block */ 621 enum { 622 F2FS_GET_BLOCK_DEFAULT, 623 F2FS_GET_BLOCK_FIEMAP, 624 F2FS_GET_BLOCK_BMAP, 625 F2FS_GET_BLOCK_DIO, 626 F2FS_GET_BLOCK_PRE_DIO, 627 F2FS_GET_BLOCK_PRE_AIO, 628 F2FS_GET_BLOCK_PRECACHE, 629 }; 630 631 /* 632 * i_advise uses FADVISE_XXX_BIT. We can add additional hints later. 633 */ 634 #define FADVISE_COLD_BIT 0x01 635 #define FADVISE_LOST_PINO_BIT 0x02 636 #define FADVISE_ENCRYPT_BIT 0x04 637 #define FADVISE_ENC_NAME_BIT 0x08 638 #define FADVISE_KEEP_SIZE_BIT 0x10 639 #define FADVISE_HOT_BIT 0x20 640 #define FADVISE_VERITY_BIT 0x40 641 642 #define FADVISE_MODIFIABLE_BITS (FADVISE_COLD_BIT | FADVISE_HOT_BIT) 643 644 #define file_is_cold(inode) is_file(inode, FADVISE_COLD_BIT) 645 #define file_set_cold(inode) set_file(inode, FADVISE_COLD_BIT) 646 #define file_clear_cold(inode) clear_file(inode, FADVISE_COLD_BIT) 647 648 #define file_wrong_pino(inode) is_file(inode, FADVISE_LOST_PINO_BIT) 649 #define file_lost_pino(inode) set_file(inode, FADVISE_LOST_PINO_BIT) 650 #define file_got_pino(inode) clear_file(inode, FADVISE_LOST_PINO_BIT) 651 652 #define file_is_encrypt(inode) is_file(inode, FADVISE_ENCRYPT_BIT) 653 #define file_set_encrypt(inode) set_file(inode, FADVISE_ENCRYPT_BIT) 654 655 #define file_enc_name(inode) is_file(inode, FADVISE_ENC_NAME_BIT) 656 #define file_set_enc_name(inode) set_file(inode, FADVISE_ENC_NAME_BIT) 657 658 #define file_keep_isize(inode) is_file(inode, FADVISE_KEEP_SIZE_BIT) 659 #define file_set_keep_isize(inode) set_file(inode, FADVISE_KEEP_SIZE_BIT) 660 661 #define file_is_hot(inode) is_file(inode, FADVISE_HOT_BIT) 662 #define file_set_hot(inode) set_file(inode, FADVISE_HOT_BIT) 663 #define file_clear_hot(inode) clear_file(inode, FADVISE_HOT_BIT) 664 665 #define file_is_verity(inode) is_file(inode, FADVISE_VERITY_BIT) 666 #define file_set_verity(inode) set_file(inode, FADVISE_VERITY_BIT) 667 668 #define DEF_DIR_LEVEL 0 669 670 enum { 671 GC_FAILURE_PIN, 672 GC_FAILURE_ATOMIC, 673 MAX_GC_FAILURE 674 }; 675 676 /* used for f2fs_inode_info->flags */ 677 enum { 678 FI_NEW_INODE, /* indicate newly allocated inode */ 679 FI_DIRTY_INODE, /* indicate inode is dirty or not */ 680 FI_AUTO_RECOVER, /* indicate inode is recoverable */ 681 FI_DIRTY_DIR, /* indicate directory has dirty pages */ 682 FI_INC_LINK, /* need to increment i_nlink */ 683 FI_ACL_MODE, /* indicate acl mode */ 684 FI_NO_ALLOC, /* should not allocate any blocks */ 685 FI_FREE_NID, /* free allocated nide */ 686 FI_NO_EXTENT, /* not to use the extent cache */ 687 FI_INLINE_XATTR, /* used for inline xattr */ 688 FI_INLINE_DATA, /* used for inline data*/ 689 FI_INLINE_DENTRY, /* used for inline dentry */ 690 FI_APPEND_WRITE, /* inode has appended data */ 691 FI_UPDATE_WRITE, /* inode has in-place-update data */ 692 FI_NEED_IPU, /* used for ipu per file */ 693 FI_ATOMIC_FILE, /* indicate atomic file */ 694 FI_ATOMIC_COMMIT, /* indicate the state of atomical committing */ 695 FI_VOLATILE_FILE, /* indicate volatile file */ 696 FI_FIRST_BLOCK_WRITTEN, /* indicate #0 data block was written */ 697 FI_DROP_CACHE, /* drop dirty page cache */ 698 FI_DATA_EXIST, /* indicate data exists */ 699 FI_INLINE_DOTS, /* indicate inline dot dentries */ 700 FI_DO_DEFRAG, /* indicate defragment is running */ 701 FI_DIRTY_FILE, /* indicate regular/symlink has dirty pages */ 702 FI_NO_PREALLOC, /* indicate skipped preallocated blocks */ 703 FI_HOT_DATA, /* indicate file is hot */ 704 FI_EXTRA_ATTR, /* indicate file has extra attribute */ 705 FI_PROJ_INHERIT, /* indicate file inherits projectid */ 706 FI_PIN_FILE, /* indicate file should not be gced */ 707 FI_ATOMIC_REVOKE_REQUEST, /* request to drop atomic data */ 708 FI_VERITY_IN_PROGRESS, /* building fs-verity Merkle tree */ 709 FI_COMPRESSED_FILE, /* indicate file's data can be compressed */ 710 FI_COMPRESS_CORRUPT, /* indicate compressed cluster is corrupted */ 711 FI_MMAP_FILE, /* indicate file was mmapped */ 712 FI_ENABLE_COMPRESS, /* enable compression in "user" compression mode */ 713 FI_COMPRESS_RELEASED, /* compressed blocks were released */ 714 FI_ALIGNED_WRITE, /* enable aligned write */ 715 FI_MAX, /* max flag, never be used */ 716 }; 717 718 struct f2fs_inode_info { 719 struct inode vfs_inode; /* serve a vfs inode */ 720 unsigned long i_flags; /* keep an inode flags for ioctl */ 721 unsigned char i_advise; /* use to give file attribute hints */ 722 unsigned char i_dir_level; /* use for dentry level for large dir */ 723 unsigned int i_current_depth; /* only for directory depth */ 724 /* for gc failure statistic */ 725 unsigned int i_gc_failures[MAX_GC_FAILURE]; 726 unsigned int i_pino; /* parent inode number */ 727 umode_t i_acl_mode; /* keep file acl mode temporarily */ 728 729 /* Use below internally in f2fs*/ 730 unsigned long flags[BITS_TO_LONGS(FI_MAX)]; /* use to pass per-file flags */ 731 struct rw_semaphore i_sem; /* protect fi info */ 732 atomic_t dirty_pages; /* # of dirty pages */ 733 f2fs_hash_t chash; /* hash value of given file name */ 734 unsigned int clevel; /* maximum level of given file name */ 735 struct task_struct *task; /* lookup and create consistency */ 736 struct task_struct *cp_task; /* separate cp/wb IO stats*/ 737 nid_t i_xattr_nid; /* node id that contains xattrs */ 738 loff_t last_disk_size; /* lastly written file size */ 739 spinlock_t i_size_lock; /* protect last_disk_size */ 740 741 #ifdef CONFIG_QUOTA 742 struct dquot *i_dquot[MAXQUOTAS]; 743 744 /* quota space reservation, managed internally by quota code */ 745 qsize_t i_reserved_quota; 746 #endif 747 struct list_head dirty_list; /* dirty list for dirs and files */ 748 struct list_head gdirty_list; /* linked in global dirty list */ 749 struct list_head inmem_ilist; /* list for inmem inodes */ 750 struct list_head inmem_pages; /* inmemory pages managed by f2fs */ 751 struct task_struct *inmem_task; /* store inmemory task */ 752 struct mutex inmem_lock; /* lock for inmemory pages */ 753 struct extent_tree *extent_tree; /* cached extent_tree entry */ 754 755 /* avoid racing between foreground op and gc */ 756 struct rw_semaphore i_gc_rwsem[2]; 757 struct rw_semaphore i_mmap_sem; 758 struct rw_semaphore i_xattr_sem; /* avoid racing between reading and changing EAs */ 759 760 int i_extra_isize; /* size of extra space located in i_addr */ 761 kprojid_t i_projid; /* id for project quota */ 762 int i_inline_xattr_size; /* inline xattr size */ 763 struct timespec64 i_crtime; /* inode creation time */ 764 struct timespec64 i_disk_time[4];/* inode disk times */ 765 766 /* for file compress */ 767 atomic_t i_compr_blocks; /* # of compressed blocks */ 768 unsigned char i_compress_algorithm; /* algorithm type */ 769 unsigned char i_log_cluster_size; /* log of cluster size */ 770 unsigned char i_compress_level; /* compress level (lz4hc,zstd) */ 771 unsigned short i_compress_flag; /* compress flag */ 772 unsigned int i_cluster_size; /* cluster size */ 773 }; 774 775 static inline void get_extent_info(struct extent_info *ext, 776 struct f2fs_extent *i_ext) 777 { 778 ext->fofs = le32_to_cpu(i_ext->fofs); 779 ext->blk = le32_to_cpu(i_ext->blk); 780 ext->len = le32_to_cpu(i_ext->len); 781 } 782 783 static inline void set_raw_extent(struct extent_info *ext, 784 struct f2fs_extent *i_ext) 785 { 786 i_ext->fofs = cpu_to_le32(ext->fofs); 787 i_ext->blk = cpu_to_le32(ext->blk); 788 i_ext->len = cpu_to_le32(ext->len); 789 } 790 791 static inline void set_extent_info(struct extent_info *ei, unsigned int fofs, 792 u32 blk, unsigned int len) 793 { 794 ei->fofs = fofs; 795 ei->blk = blk; 796 ei->len = len; 797 } 798 799 static inline bool __is_discard_mergeable(struct discard_info *back, 800 struct discard_info *front, unsigned int max_len) 801 { 802 return (back->lstart + back->len == front->lstart) && 803 (back->len + front->len <= max_len); 804 } 805 806 static inline bool __is_discard_back_mergeable(struct discard_info *cur, 807 struct discard_info *back, unsigned int max_len) 808 { 809 return __is_discard_mergeable(back, cur, max_len); 810 } 811 812 static inline bool __is_discard_front_mergeable(struct discard_info *cur, 813 struct discard_info *front, unsigned int max_len) 814 { 815 return __is_discard_mergeable(cur, front, max_len); 816 } 817 818 static inline bool __is_extent_mergeable(struct extent_info *back, 819 struct extent_info *front) 820 { 821 return (back->fofs + back->len == front->fofs && 822 back->blk + back->len == front->blk); 823 } 824 825 static inline bool __is_back_mergeable(struct extent_info *cur, 826 struct extent_info *back) 827 { 828 return __is_extent_mergeable(back, cur); 829 } 830 831 static inline bool __is_front_mergeable(struct extent_info *cur, 832 struct extent_info *front) 833 { 834 return __is_extent_mergeable(cur, front); 835 } 836 837 extern void f2fs_mark_inode_dirty_sync(struct inode *inode, bool sync); 838 static inline void __try_update_largest_extent(struct extent_tree *et, 839 struct extent_node *en) 840 { 841 if (en->ei.len > et->largest.len) { 842 et->largest = en->ei; 843 et->largest_updated = true; 844 } 845 } 846 847 /* 848 * For free nid management 849 */ 850 enum nid_state { 851 FREE_NID, /* newly added to free nid list */ 852 PREALLOC_NID, /* it is preallocated */ 853 MAX_NID_STATE, 854 }; 855 856 enum nat_state { 857 TOTAL_NAT, 858 DIRTY_NAT, 859 RECLAIMABLE_NAT, 860 MAX_NAT_STATE, 861 }; 862 863 struct f2fs_nm_info { 864 block_t nat_blkaddr; /* base disk address of NAT */ 865 nid_t max_nid; /* maximum possible node ids */ 866 nid_t available_nids; /* # of available node ids */ 867 nid_t next_scan_nid; /* the next nid to be scanned */ 868 unsigned int ram_thresh; /* control the memory footprint */ 869 unsigned int ra_nid_pages; /* # of nid pages to be readaheaded */ 870 unsigned int dirty_nats_ratio; /* control dirty nats ratio threshold */ 871 872 /* NAT cache management */ 873 struct radix_tree_root nat_root;/* root of the nat entry cache */ 874 struct radix_tree_root nat_set_root;/* root of the nat set cache */ 875 struct rw_semaphore nat_tree_lock; /* protect nat entry tree */ 876 struct list_head nat_entries; /* cached nat entry list (clean) */ 877 spinlock_t nat_list_lock; /* protect clean nat entry list */ 878 unsigned int nat_cnt[MAX_NAT_STATE]; /* the # of cached nat entries */ 879 unsigned int nat_blocks; /* # of nat blocks */ 880 881 /* free node ids management */ 882 struct radix_tree_root free_nid_root;/* root of the free_nid cache */ 883 struct list_head free_nid_list; /* list for free nids excluding preallocated nids */ 884 unsigned int nid_cnt[MAX_NID_STATE]; /* the number of free node id */ 885 spinlock_t nid_list_lock; /* protect nid lists ops */ 886 struct mutex build_lock; /* lock for build free nids */ 887 unsigned char **free_nid_bitmap; 888 unsigned char *nat_block_bitmap; 889 unsigned short *free_nid_count; /* free nid count of NAT block */ 890 891 /* for checkpoint */ 892 char *nat_bitmap; /* NAT bitmap pointer */ 893 894 unsigned int nat_bits_blocks; /* # of nat bits blocks */ 895 unsigned char *nat_bits; /* NAT bits blocks */ 896 unsigned char *full_nat_bits; /* full NAT pages */ 897 unsigned char *empty_nat_bits; /* empty NAT pages */ 898 #ifdef CONFIG_F2FS_CHECK_FS 899 char *nat_bitmap_mir; /* NAT bitmap mirror */ 900 #endif 901 int bitmap_size; /* bitmap size */ 902 }; 903 904 /* 905 * this structure is used as one of function parameters. 906 * all the information are dedicated to a given direct node block determined 907 * by the data offset in a file. 908 */ 909 struct dnode_of_data { 910 struct inode *inode; /* vfs inode pointer */ 911 struct page *inode_page; /* its inode page, NULL is possible */ 912 struct page *node_page; /* cached direct node page */ 913 nid_t nid; /* node id of the direct node block */ 914 unsigned int ofs_in_node; /* data offset in the node page */ 915 bool inode_page_locked; /* inode page is locked or not */ 916 bool node_changed; /* is node block changed */ 917 char cur_level; /* level of hole node page */ 918 char max_level; /* level of current page located */ 919 block_t data_blkaddr; /* block address of the node block */ 920 }; 921 922 static inline void set_new_dnode(struct dnode_of_data *dn, struct inode *inode, 923 struct page *ipage, struct page *npage, nid_t nid) 924 { 925 memset(dn, 0, sizeof(*dn)); 926 dn->inode = inode; 927 dn->inode_page = ipage; 928 dn->node_page = npage; 929 dn->nid = nid; 930 } 931 932 /* 933 * For SIT manager 934 * 935 * By default, there are 6 active log areas across the whole main area. 936 * When considering hot and cold data separation to reduce cleaning overhead, 937 * we split 3 for data logs and 3 for node logs as hot, warm, and cold types, 938 * respectively. 939 * In the current design, you should not change the numbers intentionally. 940 * Instead, as a mount option such as active_logs=x, you can use 2, 4, and 6 941 * logs individually according to the underlying devices. (default: 6) 942 * Just in case, on-disk layout covers maximum 16 logs that consist of 8 for 943 * data and 8 for node logs. 944 */ 945 #define NR_CURSEG_DATA_TYPE (3) 946 #define NR_CURSEG_NODE_TYPE (3) 947 #define NR_CURSEG_INMEM_TYPE (2) 948 #define NR_CURSEG_RO_TYPE (2) 949 #define NR_CURSEG_PERSIST_TYPE (NR_CURSEG_DATA_TYPE + NR_CURSEG_NODE_TYPE) 950 #define NR_CURSEG_TYPE (NR_CURSEG_INMEM_TYPE + NR_CURSEG_PERSIST_TYPE) 951 952 enum { 953 CURSEG_HOT_DATA = 0, /* directory entry blocks */ 954 CURSEG_WARM_DATA, /* data blocks */ 955 CURSEG_COLD_DATA, /* multimedia or GCed data blocks */ 956 CURSEG_HOT_NODE, /* direct node blocks of directory files */ 957 CURSEG_WARM_NODE, /* direct node blocks of normal files */ 958 CURSEG_COLD_NODE, /* indirect node blocks */ 959 NR_PERSISTENT_LOG, /* number of persistent log */ 960 CURSEG_COLD_DATA_PINNED = NR_PERSISTENT_LOG, 961 /* pinned file that needs consecutive block address */ 962 CURSEG_ALL_DATA_ATGC, /* SSR alloctor in hot/warm/cold data area */ 963 NO_CHECK_TYPE, /* number of persistent & inmem log */ 964 }; 965 966 struct flush_cmd { 967 struct completion wait; 968 struct llist_node llnode; 969 nid_t ino; 970 int ret; 971 }; 972 973 struct flush_cmd_control { 974 struct task_struct *f2fs_issue_flush; /* flush thread */ 975 wait_queue_head_t flush_wait_queue; /* waiting queue for wake-up */ 976 atomic_t issued_flush; /* # of issued flushes */ 977 atomic_t queued_flush; /* # of queued flushes */ 978 struct llist_head issue_list; /* list for command issue */ 979 struct llist_node *dispatch_list; /* list for command dispatch */ 980 }; 981 982 struct f2fs_sm_info { 983 struct sit_info *sit_info; /* whole segment information */ 984 struct free_segmap_info *free_info; /* free segment information */ 985 struct dirty_seglist_info *dirty_info; /* dirty segment information */ 986 struct curseg_info *curseg_array; /* active segment information */ 987 988 struct rw_semaphore curseg_lock; /* for preventing curseg change */ 989 990 block_t seg0_blkaddr; /* block address of 0'th segment */ 991 block_t main_blkaddr; /* start block address of main area */ 992 block_t ssa_blkaddr; /* start block address of SSA area */ 993 994 unsigned int segment_count; /* total # of segments */ 995 unsigned int main_segments; /* # of segments in main area */ 996 unsigned int reserved_segments; /* # of reserved segments */ 997 unsigned int ovp_segments; /* # of overprovision segments */ 998 999 /* a threshold to reclaim prefree segments */ 1000 unsigned int rec_prefree_segments; 1001 1002 /* for batched trimming */ 1003 unsigned int trim_sections; /* # of sections to trim */ 1004 1005 struct list_head sit_entry_set; /* sit entry set list */ 1006 1007 unsigned int ipu_policy; /* in-place-update policy */ 1008 unsigned int min_ipu_util; /* in-place-update threshold */ 1009 unsigned int min_fsync_blocks; /* threshold for fsync */ 1010 unsigned int min_seq_blocks; /* threshold for sequential blocks */ 1011 unsigned int min_hot_blocks; /* threshold for hot block allocation */ 1012 unsigned int min_ssr_sections; /* threshold to trigger SSR allocation */ 1013 1014 /* for flush command control */ 1015 struct flush_cmd_control *fcc_info; 1016 1017 /* for discard command control */ 1018 struct discard_cmd_control *dcc_info; 1019 }; 1020 1021 /* 1022 * For superblock 1023 */ 1024 /* 1025 * COUNT_TYPE for monitoring 1026 * 1027 * f2fs monitors the number of several block types such as on-writeback, 1028 * dirty dentry blocks, dirty node blocks, and dirty meta blocks. 1029 */ 1030 #define WB_DATA_TYPE(p) (__is_cp_guaranteed(p) ? F2FS_WB_CP_DATA : F2FS_WB_DATA) 1031 enum count_type { 1032 F2FS_DIRTY_DENTS, 1033 F2FS_DIRTY_DATA, 1034 F2FS_DIRTY_QDATA, 1035 F2FS_DIRTY_NODES, 1036 F2FS_DIRTY_META, 1037 F2FS_INMEM_PAGES, 1038 F2FS_DIRTY_IMETA, 1039 F2FS_WB_CP_DATA, 1040 F2FS_WB_DATA, 1041 F2FS_RD_DATA, 1042 F2FS_RD_NODE, 1043 F2FS_RD_META, 1044 F2FS_DIO_WRITE, 1045 F2FS_DIO_READ, 1046 NR_COUNT_TYPE, 1047 }; 1048 1049 /* 1050 * The below are the page types of bios used in submit_bio(). 1051 * The available types are: 1052 * DATA User data pages. It operates as async mode. 1053 * NODE Node pages. It operates as async mode. 1054 * META FS metadata pages such as SIT, NAT, CP. 1055 * NR_PAGE_TYPE The number of page types. 1056 * META_FLUSH Make sure the previous pages are written 1057 * with waiting the bio's completion 1058 * ... Only can be used with META. 1059 */ 1060 #define PAGE_TYPE_OF_BIO(type) ((type) > META ? META : (type)) 1061 enum page_type { 1062 DATA, 1063 NODE, 1064 META, 1065 NR_PAGE_TYPE, 1066 META_FLUSH, 1067 INMEM, /* the below types are used by tracepoints only. */ 1068 INMEM_DROP, 1069 INMEM_INVALIDATE, 1070 INMEM_REVOKE, 1071 IPU, 1072 OPU, 1073 }; 1074 1075 enum temp_type { 1076 HOT = 0, /* must be zero for meta bio */ 1077 WARM, 1078 COLD, 1079 NR_TEMP_TYPE, 1080 }; 1081 1082 enum need_lock_type { 1083 LOCK_REQ = 0, 1084 LOCK_DONE, 1085 LOCK_RETRY, 1086 }; 1087 1088 enum cp_reason_type { 1089 CP_NO_NEEDED, 1090 CP_NON_REGULAR, 1091 CP_COMPRESSED, 1092 CP_HARDLINK, 1093 CP_SB_NEED_CP, 1094 CP_WRONG_PINO, 1095 CP_NO_SPC_ROLL, 1096 CP_NODE_NEED_CP, 1097 CP_FASTBOOT_MODE, 1098 CP_SPEC_LOG_NUM, 1099 CP_RECOVER_DIR, 1100 }; 1101 1102 enum iostat_type { 1103 /* WRITE IO */ 1104 APP_DIRECT_IO, /* app direct write IOs */ 1105 APP_BUFFERED_IO, /* app buffered write IOs */ 1106 APP_WRITE_IO, /* app write IOs */ 1107 APP_MAPPED_IO, /* app mapped IOs */ 1108 FS_DATA_IO, /* data IOs from kworker/fsync/reclaimer */ 1109 FS_NODE_IO, /* node IOs from kworker/fsync/reclaimer */ 1110 FS_META_IO, /* meta IOs from kworker/reclaimer */ 1111 FS_GC_DATA_IO, /* data IOs from forground gc */ 1112 FS_GC_NODE_IO, /* node IOs from forground gc */ 1113 FS_CP_DATA_IO, /* data IOs from checkpoint */ 1114 FS_CP_NODE_IO, /* node IOs from checkpoint */ 1115 FS_CP_META_IO, /* meta IOs from checkpoint */ 1116 1117 /* READ IO */ 1118 APP_DIRECT_READ_IO, /* app direct read IOs */ 1119 APP_BUFFERED_READ_IO, /* app buffered read IOs */ 1120 APP_READ_IO, /* app read IOs */ 1121 APP_MAPPED_READ_IO, /* app mapped read IOs */ 1122 FS_DATA_READ_IO, /* data read IOs */ 1123 FS_GDATA_READ_IO, /* data read IOs from background gc */ 1124 FS_CDATA_READ_IO, /* compressed data read IOs */ 1125 FS_NODE_READ_IO, /* node read IOs */ 1126 FS_META_READ_IO, /* meta read IOs */ 1127 1128 /* other */ 1129 FS_DISCARD, /* discard */ 1130 NR_IO_TYPE, 1131 }; 1132 1133 struct f2fs_io_info { 1134 struct f2fs_sb_info *sbi; /* f2fs_sb_info pointer */ 1135 nid_t ino; /* inode number */ 1136 enum page_type type; /* contains DATA/NODE/META/META_FLUSH */ 1137 enum temp_type temp; /* contains HOT/WARM/COLD */ 1138 int op; /* contains REQ_OP_ */ 1139 int op_flags; /* req_flag_bits */ 1140 block_t new_blkaddr; /* new block address to be written */ 1141 block_t old_blkaddr; /* old block address before Cow */ 1142 struct page *page; /* page to be written */ 1143 struct page *encrypted_page; /* encrypted page */ 1144 struct page *compressed_page; /* compressed page */ 1145 struct list_head list; /* serialize IOs */ 1146 bool submitted; /* indicate IO submission */ 1147 int need_lock; /* indicate we need to lock cp_rwsem */ 1148 bool in_list; /* indicate fio is in io_list */ 1149 bool is_por; /* indicate IO is from recovery or not */ 1150 bool retry; /* need to reallocate block address */ 1151 int compr_blocks; /* # of compressed block addresses */ 1152 bool encrypted; /* indicate file is encrypted */ 1153 enum iostat_type io_type; /* io type */ 1154 struct writeback_control *io_wbc; /* writeback control */ 1155 struct bio **bio; /* bio for ipu */ 1156 sector_t *last_block; /* last block number in bio */ 1157 unsigned char version; /* version of the node */ 1158 }; 1159 1160 struct bio_entry { 1161 struct bio *bio; 1162 struct list_head list; 1163 }; 1164 1165 #define is_read_io(rw) ((rw) == READ) 1166 struct f2fs_bio_info { 1167 struct f2fs_sb_info *sbi; /* f2fs superblock */ 1168 struct bio *bio; /* bios to merge */ 1169 sector_t last_block_in_bio; /* last block number */ 1170 struct f2fs_io_info fio; /* store buffered io info. */ 1171 struct rw_semaphore io_rwsem; /* blocking op for bio */ 1172 spinlock_t io_lock; /* serialize DATA/NODE IOs */ 1173 struct list_head io_list; /* track fios */ 1174 struct list_head bio_list; /* bio entry list head */ 1175 struct rw_semaphore bio_list_lock; /* lock to protect bio entry list */ 1176 }; 1177 1178 #define FDEV(i) (sbi->devs[i]) 1179 #define RDEV(i) (raw_super->devs[i]) 1180 struct f2fs_dev_info { 1181 struct block_device *bdev; 1182 char path[MAX_PATH_LEN]; 1183 unsigned int total_segments; 1184 block_t start_blk; 1185 block_t end_blk; 1186 #ifdef CONFIG_BLK_DEV_ZONED 1187 unsigned int nr_blkz; /* Total number of zones */ 1188 unsigned long *blkz_seq; /* Bitmap indicating sequential zones */ 1189 block_t *zone_capacity_blocks; /* Array of zone capacity in blks */ 1190 #endif 1191 }; 1192 1193 enum inode_type { 1194 DIR_INODE, /* for dirty dir inode */ 1195 FILE_INODE, /* for dirty regular/symlink inode */ 1196 DIRTY_META, /* for all dirtied inode metadata */ 1197 ATOMIC_FILE, /* for all atomic files */ 1198 NR_INODE_TYPE, 1199 }; 1200 1201 /* for inner inode cache management */ 1202 struct inode_management { 1203 struct radix_tree_root ino_root; /* ino entry array */ 1204 spinlock_t ino_lock; /* for ino entry lock */ 1205 struct list_head ino_list; /* inode list head */ 1206 unsigned long ino_num; /* number of entries */ 1207 }; 1208 1209 /* for GC_AT */ 1210 struct atgc_management { 1211 bool atgc_enabled; /* ATGC is enabled or not */ 1212 struct rb_root_cached root; /* root of victim rb-tree */ 1213 struct list_head victim_list; /* linked with all victim entries */ 1214 unsigned int victim_count; /* victim count in rb-tree */ 1215 unsigned int candidate_ratio; /* candidate ratio */ 1216 unsigned int max_candidate_count; /* max candidate count */ 1217 unsigned int age_weight; /* age weight, vblock_weight = 100 - age_weight */ 1218 unsigned long long age_threshold; /* age threshold */ 1219 }; 1220 1221 /* For s_flag in struct f2fs_sb_info */ 1222 enum { 1223 SBI_IS_DIRTY, /* dirty flag for checkpoint */ 1224 SBI_IS_CLOSE, /* specify unmounting */ 1225 SBI_NEED_FSCK, /* need fsck.f2fs to fix */ 1226 SBI_POR_DOING, /* recovery is doing or not */ 1227 SBI_NEED_SB_WRITE, /* need to recover superblock */ 1228 SBI_NEED_CP, /* need to checkpoint */ 1229 SBI_IS_SHUTDOWN, /* shutdown by ioctl */ 1230 SBI_IS_RECOVERED, /* recovered orphan/data */ 1231 SBI_CP_DISABLED, /* CP was disabled last mount */ 1232 SBI_CP_DISABLED_QUICK, /* CP was disabled quickly */ 1233 SBI_QUOTA_NEED_FLUSH, /* need to flush quota info in CP */ 1234 SBI_QUOTA_SKIP_FLUSH, /* skip flushing quota in current CP */ 1235 SBI_QUOTA_NEED_REPAIR, /* quota file may be corrupted */ 1236 SBI_IS_RESIZEFS, /* resizefs is in process */ 1237 }; 1238 1239 enum { 1240 CP_TIME, 1241 REQ_TIME, 1242 DISCARD_TIME, 1243 GC_TIME, 1244 DISABLE_TIME, 1245 UMOUNT_DISCARD_TIMEOUT, 1246 MAX_TIME, 1247 }; 1248 1249 enum { 1250 GC_NORMAL, 1251 GC_IDLE_CB, 1252 GC_IDLE_GREEDY, 1253 GC_IDLE_AT, 1254 GC_URGENT_HIGH, 1255 GC_URGENT_LOW, 1256 }; 1257 1258 enum { 1259 BGGC_MODE_ON, /* background gc is on */ 1260 BGGC_MODE_OFF, /* background gc is off */ 1261 BGGC_MODE_SYNC, /* 1262 * background gc is on, migrating blocks 1263 * like foreground gc 1264 */ 1265 }; 1266 1267 enum { 1268 FS_MODE_ADAPTIVE, /* use both lfs/ssr allocation */ 1269 FS_MODE_LFS, /* use lfs allocation only */ 1270 }; 1271 1272 enum { 1273 WHINT_MODE_OFF, /* not pass down write hints */ 1274 WHINT_MODE_USER, /* try to pass down hints given by users */ 1275 WHINT_MODE_FS, /* pass down hints with F2FS policy */ 1276 }; 1277 1278 enum { 1279 ALLOC_MODE_DEFAULT, /* stay default */ 1280 ALLOC_MODE_REUSE, /* reuse segments as much as possible */ 1281 }; 1282 1283 enum fsync_mode { 1284 FSYNC_MODE_POSIX, /* fsync follows posix semantics */ 1285 FSYNC_MODE_STRICT, /* fsync behaves in line with ext4 */ 1286 FSYNC_MODE_NOBARRIER, /* fsync behaves nobarrier based on posix */ 1287 }; 1288 1289 enum { 1290 COMPR_MODE_FS, /* 1291 * automatically compress compression 1292 * enabled files 1293 */ 1294 COMPR_MODE_USER, /* 1295 * automatical compression is disabled. 1296 * user can control the file compression 1297 * using ioctls 1298 */ 1299 }; 1300 1301 static inline int f2fs_test_bit(unsigned int nr, char *addr); 1302 static inline void f2fs_set_bit(unsigned int nr, char *addr); 1303 static inline void f2fs_clear_bit(unsigned int nr, char *addr); 1304 1305 /* 1306 * Layout of f2fs page.private: 1307 * 1308 * Layout A: lowest bit should be 1 1309 * | bit0 = 1 | bit1 | bit2 | ... | bit MAX | private data .... | 1310 * bit 0 PAGE_PRIVATE_NOT_POINTER 1311 * bit 1 PAGE_PRIVATE_ATOMIC_WRITE 1312 * bit 2 PAGE_PRIVATE_DUMMY_WRITE 1313 * bit 3 PAGE_PRIVATE_ONGOING_MIGRATION 1314 * bit 4 PAGE_PRIVATE_INLINE_INODE 1315 * bit 5 PAGE_PRIVATE_REF_RESOURCE 1316 * bit 6- f2fs private data 1317 * 1318 * Layout B: lowest bit should be 0 1319 * page.private is a wrapped pointer. 1320 */ 1321 enum { 1322 PAGE_PRIVATE_NOT_POINTER, /* private contains non-pointer data */ 1323 PAGE_PRIVATE_ATOMIC_WRITE, /* data page from atomic write path */ 1324 PAGE_PRIVATE_DUMMY_WRITE, /* data page for padding aligned IO */ 1325 PAGE_PRIVATE_ONGOING_MIGRATION, /* data page which is on-going migrating */ 1326 PAGE_PRIVATE_INLINE_INODE, /* inode page contains inline data */ 1327 PAGE_PRIVATE_REF_RESOURCE, /* dirty page has referenced resources */ 1328 PAGE_PRIVATE_MAX 1329 }; 1330 1331 #define PAGE_PRIVATE_GET_FUNC(name, flagname) \ 1332 static inline bool page_private_##name(struct page *page) \ 1333 { \ 1334 return PagePrivate(page) && \ 1335 test_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)) && \ 1336 test_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \ 1337 } 1338 1339 #define PAGE_PRIVATE_SET_FUNC(name, flagname) \ 1340 static inline void set_page_private_##name(struct page *page) \ 1341 { \ 1342 if (!PagePrivate(page)) { \ 1343 get_page(page); \ 1344 SetPagePrivate(page); \ 1345 set_page_private(page, 0); \ 1346 } \ 1347 set_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)); \ 1348 set_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \ 1349 } 1350 1351 #define PAGE_PRIVATE_CLEAR_FUNC(name, flagname) \ 1352 static inline void clear_page_private_##name(struct page *page) \ 1353 { \ 1354 clear_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \ 1355 if (page_private(page) == 1 << PAGE_PRIVATE_NOT_POINTER) { \ 1356 set_page_private(page, 0); \ 1357 if (PagePrivate(page)) { \ 1358 ClearPagePrivate(page); \ 1359 put_page(page); \ 1360 }\ 1361 } \ 1362 } 1363 1364 PAGE_PRIVATE_GET_FUNC(nonpointer, NOT_POINTER); 1365 PAGE_PRIVATE_GET_FUNC(reference, REF_RESOURCE); 1366 PAGE_PRIVATE_GET_FUNC(inline, INLINE_INODE); 1367 PAGE_PRIVATE_GET_FUNC(gcing, ONGOING_MIGRATION); 1368 PAGE_PRIVATE_GET_FUNC(atomic, ATOMIC_WRITE); 1369 PAGE_PRIVATE_GET_FUNC(dummy, DUMMY_WRITE); 1370 1371 PAGE_PRIVATE_SET_FUNC(reference, REF_RESOURCE); 1372 PAGE_PRIVATE_SET_FUNC(inline, INLINE_INODE); 1373 PAGE_PRIVATE_SET_FUNC(gcing, ONGOING_MIGRATION); 1374 PAGE_PRIVATE_SET_FUNC(atomic, ATOMIC_WRITE); 1375 PAGE_PRIVATE_SET_FUNC(dummy, DUMMY_WRITE); 1376 1377 PAGE_PRIVATE_CLEAR_FUNC(reference, REF_RESOURCE); 1378 PAGE_PRIVATE_CLEAR_FUNC(inline, INLINE_INODE); 1379 PAGE_PRIVATE_CLEAR_FUNC(gcing, ONGOING_MIGRATION); 1380 PAGE_PRIVATE_CLEAR_FUNC(atomic, ATOMIC_WRITE); 1381 PAGE_PRIVATE_CLEAR_FUNC(dummy, DUMMY_WRITE); 1382 1383 static inline unsigned long get_page_private_data(struct page *page) 1384 { 1385 unsigned long data = page_private(page); 1386 1387 if (!test_bit(PAGE_PRIVATE_NOT_POINTER, &data)) 1388 return 0; 1389 return data >> PAGE_PRIVATE_MAX; 1390 } 1391 1392 static inline void set_page_private_data(struct page *page, unsigned long data) 1393 { 1394 if (!PagePrivate(page)) { 1395 get_page(page); 1396 SetPagePrivate(page); 1397 set_page_private(page, 0); 1398 } 1399 set_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)); 1400 page_private(page) |= data << PAGE_PRIVATE_MAX; 1401 } 1402 1403 static inline void clear_page_private_data(struct page *page) 1404 { 1405 page_private(page) &= (1 << PAGE_PRIVATE_MAX) - 1; 1406 if (page_private(page) == 1 << PAGE_PRIVATE_NOT_POINTER) { 1407 set_page_private(page, 0); 1408 if (PagePrivate(page)) { 1409 ClearPagePrivate(page); 1410 put_page(page); 1411 } 1412 } 1413 } 1414 1415 /* For compression */ 1416 enum compress_algorithm_type { 1417 COMPRESS_LZO, 1418 COMPRESS_LZ4, 1419 COMPRESS_ZSTD, 1420 COMPRESS_LZORLE, 1421 COMPRESS_MAX, 1422 }; 1423 1424 enum compress_flag { 1425 COMPRESS_CHKSUM, 1426 COMPRESS_MAX_FLAG, 1427 }; 1428 1429 #define COMPRESS_WATERMARK 20 1430 #define COMPRESS_PERCENT 20 1431 1432 #define COMPRESS_DATA_RESERVED_SIZE 4 1433 struct compress_data { 1434 __le32 clen; /* compressed data size */ 1435 __le32 chksum; /* compressed data chksum */ 1436 __le32 reserved[COMPRESS_DATA_RESERVED_SIZE]; /* reserved */ 1437 u8 cdata[]; /* compressed data */ 1438 }; 1439 1440 #define COMPRESS_HEADER_SIZE (sizeof(struct compress_data)) 1441 1442 #define F2FS_COMPRESSED_PAGE_MAGIC 0xF5F2C000 1443 1444 #define COMPRESS_LEVEL_OFFSET 8 1445 1446 /* compress context */ 1447 struct compress_ctx { 1448 struct inode *inode; /* inode the context belong to */ 1449 pgoff_t cluster_idx; /* cluster index number */ 1450 unsigned int cluster_size; /* page count in cluster */ 1451 unsigned int log_cluster_size; /* log of cluster size */ 1452 struct page **rpages; /* pages store raw data in cluster */ 1453 unsigned int nr_rpages; /* total page number in rpages */ 1454 struct page **cpages; /* pages store compressed data in cluster */ 1455 unsigned int nr_cpages; /* total page number in cpages */ 1456 void *rbuf; /* virtual mapped address on rpages */ 1457 struct compress_data *cbuf; /* virtual mapped address on cpages */ 1458 size_t rlen; /* valid data length in rbuf */ 1459 size_t clen; /* valid data length in cbuf */ 1460 void *private; /* payload buffer for specified compression algorithm */ 1461 void *private2; /* extra payload buffer */ 1462 }; 1463 1464 /* compress context for write IO path */ 1465 struct compress_io_ctx { 1466 u32 magic; /* magic number to indicate page is compressed */ 1467 struct inode *inode; /* inode the context belong to */ 1468 struct page **rpages; /* pages store raw data in cluster */ 1469 unsigned int nr_rpages; /* total page number in rpages */ 1470 atomic_t pending_pages; /* in-flight compressed page count */ 1471 }; 1472 1473 /* Context for decompressing one cluster on the read IO path */ 1474 struct decompress_io_ctx { 1475 u32 magic; /* magic number to indicate page is compressed */ 1476 struct inode *inode; /* inode the context belong to */ 1477 pgoff_t cluster_idx; /* cluster index number */ 1478 unsigned int cluster_size; /* page count in cluster */ 1479 unsigned int log_cluster_size; /* log of cluster size */ 1480 struct page **rpages; /* pages store raw data in cluster */ 1481 unsigned int nr_rpages; /* total page number in rpages */ 1482 struct page **cpages; /* pages store compressed data in cluster */ 1483 unsigned int nr_cpages; /* total page number in cpages */ 1484 struct page **tpages; /* temp pages to pad holes in cluster */ 1485 void *rbuf; /* virtual mapped address on rpages */ 1486 struct compress_data *cbuf; /* virtual mapped address on cpages */ 1487 size_t rlen; /* valid data length in rbuf */ 1488 size_t clen; /* valid data length in cbuf */ 1489 1490 /* 1491 * The number of compressed pages remaining to be read in this cluster. 1492 * This is initially nr_cpages. It is decremented by 1 each time a page 1493 * has been read (or failed to be read). When it reaches 0, the cluster 1494 * is decompressed (or an error is reported). 1495 * 1496 * If an error occurs before all the pages have been submitted for I/O, 1497 * then this will never reach 0. In this case the I/O submitter is 1498 * responsible for calling f2fs_decompress_end_io() instead. 1499 */ 1500 atomic_t remaining_pages; 1501 1502 /* 1503 * Number of references to this decompress_io_ctx. 1504 * 1505 * One reference is held for I/O completion. This reference is dropped 1506 * after the pagecache pages are updated and unlocked -- either after 1507 * decompression (and verity if enabled), or after an error. 1508 * 1509 * In addition, each compressed page holds a reference while it is in a 1510 * bio. These references are necessary prevent compressed pages from 1511 * being freed while they are still in a bio. 1512 */ 1513 refcount_t refcnt; 1514 1515 bool failed; /* IO error occurred before decompression? */ 1516 bool need_verity; /* need fs-verity verification after decompression? */ 1517 void *private; /* payload buffer for specified decompression algorithm */ 1518 void *private2; /* extra payload buffer */ 1519 struct work_struct verity_work; /* work to verify the decompressed pages */ 1520 }; 1521 1522 #define NULL_CLUSTER ((unsigned int)(~0)) 1523 #define MIN_COMPRESS_LOG_SIZE 2 1524 #define MAX_COMPRESS_LOG_SIZE 8 1525 #define MAX_COMPRESS_WINDOW_SIZE(log_size) ((PAGE_SIZE) << (log_size)) 1526 1527 struct f2fs_sb_info { 1528 struct super_block *sb; /* pointer to VFS super block */ 1529 struct proc_dir_entry *s_proc; /* proc entry */ 1530 struct f2fs_super_block *raw_super; /* raw super block pointer */ 1531 struct rw_semaphore sb_lock; /* lock for raw super block */ 1532 int valid_super_block; /* valid super block no */ 1533 unsigned long s_flag; /* flags for sbi */ 1534 struct mutex writepages; /* mutex for writepages() */ 1535 1536 #ifdef CONFIG_BLK_DEV_ZONED 1537 unsigned int blocks_per_blkz; /* F2FS blocks per zone */ 1538 unsigned int log_blocks_per_blkz; /* log2 F2FS blocks per zone */ 1539 #endif 1540 1541 /* for node-related operations */ 1542 struct f2fs_nm_info *nm_info; /* node manager */ 1543 struct inode *node_inode; /* cache node blocks */ 1544 1545 /* for segment-related operations */ 1546 struct f2fs_sm_info *sm_info; /* segment manager */ 1547 1548 /* for bio operations */ 1549 struct f2fs_bio_info *write_io[NR_PAGE_TYPE]; /* for write bios */ 1550 /* keep migration IO order for LFS mode */ 1551 struct rw_semaphore io_order_lock; 1552 mempool_t *write_io_dummy; /* Dummy pages */ 1553 1554 /* for checkpoint */ 1555 struct f2fs_checkpoint *ckpt; /* raw checkpoint pointer */ 1556 int cur_cp_pack; /* remain current cp pack */ 1557 spinlock_t cp_lock; /* for flag in ckpt */ 1558 struct inode *meta_inode; /* cache meta blocks */ 1559 struct rw_semaphore cp_global_sem; /* checkpoint procedure lock */ 1560 struct rw_semaphore cp_rwsem; /* blocking FS operations */ 1561 struct rw_semaphore node_write; /* locking node writes */ 1562 struct rw_semaphore node_change; /* locking node change */ 1563 wait_queue_head_t cp_wait; 1564 unsigned long last_time[MAX_TIME]; /* to store time in jiffies */ 1565 long interval_time[MAX_TIME]; /* to store thresholds */ 1566 struct ckpt_req_control cprc_info; /* for checkpoint request control */ 1567 1568 struct inode_management im[MAX_INO_ENTRY]; /* manage inode cache */ 1569 1570 spinlock_t fsync_node_lock; /* for node entry lock */ 1571 struct list_head fsync_node_list; /* node list head */ 1572 unsigned int fsync_seg_id; /* sequence id */ 1573 unsigned int fsync_node_num; /* number of node entries */ 1574 1575 /* for orphan inode, use 0'th array */ 1576 unsigned int max_orphans; /* max orphan inodes */ 1577 1578 /* for inode management */ 1579 struct list_head inode_list[NR_INODE_TYPE]; /* dirty inode list */ 1580 spinlock_t inode_lock[NR_INODE_TYPE]; /* for dirty inode list lock */ 1581 struct mutex flush_lock; /* for flush exclusion */ 1582 1583 /* for extent tree cache */ 1584 struct radix_tree_root extent_tree_root;/* cache extent cache entries */ 1585 struct mutex extent_tree_lock; /* locking extent radix tree */ 1586 struct list_head extent_list; /* lru list for shrinker */ 1587 spinlock_t extent_lock; /* locking extent lru list */ 1588 atomic_t total_ext_tree; /* extent tree count */ 1589 struct list_head zombie_list; /* extent zombie tree list */ 1590 atomic_t total_zombie_tree; /* extent zombie tree count */ 1591 atomic_t total_ext_node; /* extent info count */ 1592 1593 /* basic filesystem units */ 1594 unsigned int log_sectors_per_block; /* log2 sectors per block */ 1595 unsigned int log_blocksize; /* log2 block size */ 1596 unsigned int blocksize; /* block size */ 1597 unsigned int root_ino_num; /* root inode number*/ 1598 unsigned int node_ino_num; /* node inode number*/ 1599 unsigned int meta_ino_num; /* meta inode number*/ 1600 unsigned int log_blocks_per_seg; /* log2 blocks per segment */ 1601 unsigned int blocks_per_seg; /* blocks per segment */ 1602 unsigned int segs_per_sec; /* segments per section */ 1603 unsigned int secs_per_zone; /* sections per zone */ 1604 unsigned int total_sections; /* total section count */ 1605 unsigned int total_node_count; /* total node block count */ 1606 unsigned int total_valid_node_count; /* valid node block count */ 1607 int dir_level; /* directory level */ 1608 int readdir_ra; /* readahead inode in readdir */ 1609 u64 max_io_bytes; /* max io bytes to merge IOs */ 1610 1611 block_t user_block_count; /* # of user blocks */ 1612 block_t total_valid_block_count; /* # of valid blocks */ 1613 block_t discard_blks; /* discard command candidats */ 1614 block_t last_valid_block_count; /* for recovery */ 1615 block_t reserved_blocks; /* configurable reserved blocks */ 1616 block_t current_reserved_blocks; /* current reserved blocks */ 1617 1618 /* Additional tracking for no checkpoint mode */ 1619 block_t unusable_block_count; /* # of blocks saved by last cp */ 1620 1621 unsigned int nquota_files; /* # of quota sysfile */ 1622 struct rw_semaphore quota_sem; /* blocking cp for flags */ 1623 1624 /* # of pages, see count_type */ 1625 atomic_t nr_pages[NR_COUNT_TYPE]; 1626 /* # of allocated blocks */ 1627 struct percpu_counter alloc_valid_block_count; 1628 1629 /* writeback control */ 1630 atomic_t wb_sync_req[META]; /* count # of WB_SYNC threads */ 1631 1632 /* valid inode count */ 1633 struct percpu_counter total_valid_inode_count; 1634 1635 struct f2fs_mount_info mount_opt; /* mount options */ 1636 1637 /* for cleaning operations */ 1638 struct rw_semaphore gc_lock; /* 1639 * semaphore for GC, avoid 1640 * race between GC and GC or CP 1641 */ 1642 struct f2fs_gc_kthread *gc_thread; /* GC thread */ 1643 struct atgc_management am; /* atgc management */ 1644 unsigned int cur_victim_sec; /* current victim section num */ 1645 unsigned int gc_mode; /* current GC state */ 1646 unsigned int next_victim_seg[2]; /* next segment in victim section */ 1647 1648 /* for skip statistic */ 1649 unsigned int atomic_files; /* # of opened atomic file */ 1650 unsigned long long skipped_atomic_files[2]; /* FG_GC and BG_GC */ 1651 unsigned long long skipped_gc_rwsem; /* FG_GC only */ 1652 1653 /* threshold for gc trials on pinned files */ 1654 u64 gc_pin_file_threshold; 1655 struct rw_semaphore pin_sem; 1656 1657 /* maximum # of trials to find a victim segment for SSR and GC */ 1658 unsigned int max_victim_search; 1659 /* migration granularity of garbage collection, unit: segment */ 1660 unsigned int migration_granularity; 1661 1662 /* 1663 * for stat information. 1664 * one is for the LFS mode, and the other is for the SSR mode. 1665 */ 1666 #ifdef CONFIG_F2FS_STAT_FS 1667 struct f2fs_stat_info *stat_info; /* FS status information */ 1668 atomic_t meta_count[META_MAX]; /* # of meta blocks */ 1669 unsigned int segment_count[2]; /* # of allocated segments */ 1670 unsigned int block_count[2]; /* # of allocated blocks */ 1671 atomic_t inplace_count; /* # of inplace update */ 1672 atomic64_t total_hit_ext; /* # of lookup extent cache */ 1673 atomic64_t read_hit_rbtree; /* # of hit rbtree extent node */ 1674 atomic64_t read_hit_largest; /* # of hit largest extent node */ 1675 atomic64_t read_hit_cached; /* # of hit cached extent node */ 1676 atomic_t inline_xattr; /* # of inline_xattr inodes */ 1677 atomic_t inline_inode; /* # of inline_data inodes */ 1678 atomic_t inline_dir; /* # of inline_dentry inodes */ 1679 atomic_t compr_inode; /* # of compressed inodes */ 1680 atomic64_t compr_blocks; /* # of compressed blocks */ 1681 atomic_t vw_cnt; /* # of volatile writes */ 1682 atomic_t max_aw_cnt; /* max # of atomic writes */ 1683 atomic_t max_vw_cnt; /* max # of volatile writes */ 1684 unsigned int io_skip_bggc; /* skip background gc for in-flight IO */ 1685 unsigned int other_skip_bggc; /* skip background gc for other reasons */ 1686 unsigned int ndirty_inode[NR_INODE_TYPE]; /* # of dirty inodes */ 1687 #endif 1688 spinlock_t stat_lock; /* lock for stat operations */ 1689 1690 /* For app/fs IO statistics */ 1691 spinlock_t iostat_lock; 1692 unsigned long long rw_iostat[NR_IO_TYPE]; 1693 unsigned long long prev_rw_iostat[NR_IO_TYPE]; 1694 bool iostat_enable; 1695 unsigned long iostat_next_period; 1696 unsigned int iostat_period_ms; 1697 1698 /* to attach REQ_META|REQ_FUA flags */ 1699 unsigned int data_io_flag; 1700 unsigned int node_io_flag; 1701 1702 /* For sysfs suppport */ 1703 struct kobject s_kobj; /* /sys/fs/f2fs/<devname> */ 1704 struct completion s_kobj_unregister; 1705 1706 struct kobject s_stat_kobj; /* /sys/fs/f2fs/<devname>/stat */ 1707 struct completion s_stat_kobj_unregister; 1708 1709 struct kobject s_feature_list_kobj; /* /sys/fs/f2fs/<devname>/feature_list */ 1710 struct completion s_feature_list_kobj_unregister; 1711 1712 /* For shrinker support */ 1713 struct list_head s_list; 1714 int s_ndevs; /* number of devices */ 1715 struct f2fs_dev_info *devs; /* for device list */ 1716 unsigned int dirty_device; /* for checkpoint data flush */ 1717 spinlock_t dev_lock; /* protect dirty_device */ 1718 struct mutex umount_mutex; 1719 unsigned int shrinker_run_no; 1720 1721 /* For write statistics */ 1722 u64 sectors_written_start; 1723 u64 kbytes_written; 1724 1725 /* Reference to checksum algorithm driver via cryptoapi */ 1726 struct crypto_shash *s_chksum_driver; 1727 1728 /* Precomputed FS UUID checksum for seeding other checksums */ 1729 __u32 s_chksum_seed; 1730 1731 struct workqueue_struct *post_read_wq; /* post read workqueue */ 1732 1733 struct kmem_cache *inline_xattr_slab; /* inline xattr entry */ 1734 unsigned int inline_xattr_slab_size; /* default inline xattr slab size */ 1735 1736 #ifdef CONFIG_F2FS_FS_COMPRESSION 1737 struct kmem_cache *page_array_slab; /* page array entry */ 1738 unsigned int page_array_slab_size; /* default page array slab size */ 1739 1740 /* For runtime compression statistics */ 1741 u64 compr_written_block; 1742 u64 compr_saved_block; 1743 u32 compr_new_inode; 1744 1745 /* For compressed block cache */ 1746 struct inode *compress_inode; /* cache compressed blocks */ 1747 unsigned int compress_percent; /* cache page percentage */ 1748 unsigned int compress_watermark; /* cache page watermark */ 1749 atomic_t compress_page_hit; /* cache hit count */ 1750 #endif 1751 }; 1752 1753 struct f2fs_private_dio { 1754 struct inode *inode; 1755 void *orig_private; 1756 bio_end_io_t *orig_end_io; 1757 bool write; 1758 }; 1759 1760 #ifdef CONFIG_F2FS_FAULT_INJECTION 1761 #define f2fs_show_injection_info(sbi, type) \ 1762 printk_ratelimited("%sF2FS-fs (%s) : inject %s in %s of %pS\n", \ 1763 KERN_INFO, sbi->sb->s_id, \ 1764 f2fs_fault_name[type], \ 1765 __func__, __builtin_return_address(0)) 1766 static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type) 1767 { 1768 struct f2fs_fault_info *ffi = &F2FS_OPTION(sbi).fault_info; 1769 1770 if (!ffi->inject_rate) 1771 return false; 1772 1773 if (!IS_FAULT_SET(ffi, type)) 1774 return false; 1775 1776 atomic_inc(&ffi->inject_ops); 1777 if (atomic_read(&ffi->inject_ops) >= ffi->inject_rate) { 1778 atomic_set(&ffi->inject_ops, 0); 1779 return true; 1780 } 1781 return false; 1782 } 1783 #else 1784 #define f2fs_show_injection_info(sbi, type) do { } while (0) 1785 static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type) 1786 { 1787 return false; 1788 } 1789 #endif 1790 1791 /* 1792 * Test if the mounted volume is a multi-device volume. 1793 * - For a single regular disk volume, sbi->s_ndevs is 0. 1794 * - For a single zoned disk volume, sbi->s_ndevs is 1. 1795 * - For a multi-device volume, sbi->s_ndevs is always 2 or more. 1796 */ 1797 static inline bool f2fs_is_multi_device(struct f2fs_sb_info *sbi) 1798 { 1799 return sbi->s_ndevs > 1; 1800 } 1801 1802 static inline void f2fs_update_time(struct f2fs_sb_info *sbi, int type) 1803 { 1804 unsigned long now = jiffies; 1805 1806 sbi->last_time[type] = now; 1807 1808 /* DISCARD_TIME and GC_TIME are based on REQ_TIME */ 1809 if (type == REQ_TIME) { 1810 sbi->last_time[DISCARD_TIME] = now; 1811 sbi->last_time[GC_TIME] = now; 1812 } 1813 } 1814 1815 static inline bool f2fs_time_over(struct f2fs_sb_info *sbi, int type) 1816 { 1817 unsigned long interval = sbi->interval_time[type] * HZ; 1818 1819 return time_after(jiffies, sbi->last_time[type] + interval); 1820 } 1821 1822 static inline unsigned int f2fs_time_to_wait(struct f2fs_sb_info *sbi, 1823 int type) 1824 { 1825 unsigned long interval = sbi->interval_time[type] * HZ; 1826 unsigned int wait_ms = 0; 1827 long delta; 1828 1829 delta = (sbi->last_time[type] + interval) - jiffies; 1830 if (delta > 0) 1831 wait_ms = jiffies_to_msecs(delta); 1832 1833 return wait_ms; 1834 } 1835 1836 /* 1837 * Inline functions 1838 */ 1839 static inline u32 __f2fs_crc32(struct f2fs_sb_info *sbi, u32 crc, 1840 const void *address, unsigned int length) 1841 { 1842 struct { 1843 struct shash_desc shash; 1844 char ctx[4]; 1845 } desc; 1846 int err; 1847 1848 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver) != sizeof(desc.ctx)); 1849 1850 desc.shash.tfm = sbi->s_chksum_driver; 1851 *(u32 *)desc.ctx = crc; 1852 1853 err = crypto_shash_update(&desc.shash, address, length); 1854 BUG_ON(err); 1855 1856 return *(u32 *)desc.ctx; 1857 } 1858 1859 static inline u32 f2fs_crc32(struct f2fs_sb_info *sbi, const void *address, 1860 unsigned int length) 1861 { 1862 return __f2fs_crc32(sbi, F2FS_SUPER_MAGIC, address, length); 1863 } 1864 1865 static inline bool f2fs_crc_valid(struct f2fs_sb_info *sbi, __u32 blk_crc, 1866 void *buf, size_t buf_size) 1867 { 1868 return f2fs_crc32(sbi, buf, buf_size) == blk_crc; 1869 } 1870 1871 static inline u32 f2fs_chksum(struct f2fs_sb_info *sbi, u32 crc, 1872 const void *address, unsigned int length) 1873 { 1874 return __f2fs_crc32(sbi, crc, address, length); 1875 } 1876 1877 static inline struct f2fs_inode_info *F2FS_I(struct inode *inode) 1878 { 1879 return container_of(inode, struct f2fs_inode_info, vfs_inode); 1880 } 1881 1882 static inline struct f2fs_sb_info *F2FS_SB(struct super_block *sb) 1883 { 1884 return sb->s_fs_info; 1885 } 1886 1887 static inline struct f2fs_sb_info *F2FS_I_SB(struct inode *inode) 1888 { 1889 return F2FS_SB(inode->i_sb); 1890 } 1891 1892 static inline struct f2fs_sb_info *F2FS_M_SB(struct address_space *mapping) 1893 { 1894 return F2FS_I_SB(mapping->host); 1895 } 1896 1897 static inline struct f2fs_sb_info *F2FS_P_SB(struct page *page) 1898 { 1899 return F2FS_M_SB(page_file_mapping(page)); 1900 } 1901 1902 static inline struct f2fs_super_block *F2FS_RAW_SUPER(struct f2fs_sb_info *sbi) 1903 { 1904 return (struct f2fs_super_block *)(sbi->raw_super); 1905 } 1906 1907 static inline struct f2fs_checkpoint *F2FS_CKPT(struct f2fs_sb_info *sbi) 1908 { 1909 return (struct f2fs_checkpoint *)(sbi->ckpt); 1910 } 1911 1912 static inline struct f2fs_node *F2FS_NODE(struct page *page) 1913 { 1914 return (struct f2fs_node *)page_address(page); 1915 } 1916 1917 static inline struct f2fs_inode *F2FS_INODE(struct page *page) 1918 { 1919 return &((struct f2fs_node *)page_address(page))->i; 1920 } 1921 1922 static inline struct f2fs_nm_info *NM_I(struct f2fs_sb_info *sbi) 1923 { 1924 return (struct f2fs_nm_info *)(sbi->nm_info); 1925 } 1926 1927 static inline struct f2fs_sm_info *SM_I(struct f2fs_sb_info *sbi) 1928 { 1929 return (struct f2fs_sm_info *)(sbi->sm_info); 1930 } 1931 1932 static inline struct sit_info *SIT_I(struct f2fs_sb_info *sbi) 1933 { 1934 return (struct sit_info *)(SM_I(sbi)->sit_info); 1935 } 1936 1937 static inline struct free_segmap_info *FREE_I(struct f2fs_sb_info *sbi) 1938 { 1939 return (struct free_segmap_info *)(SM_I(sbi)->free_info); 1940 } 1941 1942 static inline struct dirty_seglist_info *DIRTY_I(struct f2fs_sb_info *sbi) 1943 { 1944 return (struct dirty_seglist_info *)(SM_I(sbi)->dirty_info); 1945 } 1946 1947 static inline struct address_space *META_MAPPING(struct f2fs_sb_info *sbi) 1948 { 1949 return sbi->meta_inode->i_mapping; 1950 } 1951 1952 static inline struct address_space *NODE_MAPPING(struct f2fs_sb_info *sbi) 1953 { 1954 return sbi->node_inode->i_mapping; 1955 } 1956 1957 static inline bool is_sbi_flag_set(struct f2fs_sb_info *sbi, unsigned int type) 1958 { 1959 return test_bit(type, &sbi->s_flag); 1960 } 1961 1962 static inline void set_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type) 1963 { 1964 set_bit(type, &sbi->s_flag); 1965 } 1966 1967 static inline void clear_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type) 1968 { 1969 clear_bit(type, &sbi->s_flag); 1970 } 1971 1972 static inline unsigned long long cur_cp_version(struct f2fs_checkpoint *cp) 1973 { 1974 return le64_to_cpu(cp->checkpoint_ver); 1975 } 1976 1977 static inline unsigned long f2fs_qf_ino(struct super_block *sb, int type) 1978 { 1979 if (type < F2FS_MAX_QUOTAS) 1980 return le32_to_cpu(F2FS_SB(sb)->raw_super->qf_ino[type]); 1981 return 0; 1982 } 1983 1984 static inline __u64 cur_cp_crc(struct f2fs_checkpoint *cp) 1985 { 1986 size_t crc_offset = le32_to_cpu(cp->checksum_offset); 1987 return le32_to_cpu(*((__le32 *)((unsigned char *)cp + crc_offset))); 1988 } 1989 1990 static inline bool __is_set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 1991 { 1992 unsigned int ckpt_flags = le32_to_cpu(cp->ckpt_flags); 1993 1994 return ckpt_flags & f; 1995 } 1996 1997 static inline bool is_set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 1998 { 1999 return __is_set_ckpt_flags(F2FS_CKPT(sbi), f); 2000 } 2001 2002 static inline void __set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 2003 { 2004 unsigned int ckpt_flags; 2005 2006 ckpt_flags = le32_to_cpu(cp->ckpt_flags); 2007 ckpt_flags |= f; 2008 cp->ckpt_flags = cpu_to_le32(ckpt_flags); 2009 } 2010 2011 static inline void set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 2012 { 2013 unsigned long flags; 2014 2015 spin_lock_irqsave(&sbi->cp_lock, flags); 2016 __set_ckpt_flags(F2FS_CKPT(sbi), f); 2017 spin_unlock_irqrestore(&sbi->cp_lock, flags); 2018 } 2019 2020 static inline void __clear_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 2021 { 2022 unsigned int ckpt_flags; 2023 2024 ckpt_flags = le32_to_cpu(cp->ckpt_flags); 2025 ckpt_flags &= (~f); 2026 cp->ckpt_flags = cpu_to_le32(ckpt_flags); 2027 } 2028 2029 static inline void clear_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 2030 { 2031 unsigned long flags; 2032 2033 spin_lock_irqsave(&sbi->cp_lock, flags); 2034 __clear_ckpt_flags(F2FS_CKPT(sbi), f); 2035 spin_unlock_irqrestore(&sbi->cp_lock, flags); 2036 } 2037 2038 static inline void disable_nat_bits(struct f2fs_sb_info *sbi, bool lock) 2039 { 2040 unsigned long flags; 2041 unsigned char *nat_bits; 2042 2043 /* 2044 * In order to re-enable nat_bits we need to call fsck.f2fs by 2045 * set_sbi_flag(sbi, SBI_NEED_FSCK). But it may give huge cost, 2046 * so let's rely on regular fsck or unclean shutdown. 2047 */ 2048 2049 if (lock) 2050 spin_lock_irqsave(&sbi->cp_lock, flags); 2051 __clear_ckpt_flags(F2FS_CKPT(sbi), CP_NAT_BITS_FLAG); 2052 nat_bits = NM_I(sbi)->nat_bits; 2053 NM_I(sbi)->nat_bits = NULL; 2054 if (lock) 2055 spin_unlock_irqrestore(&sbi->cp_lock, flags); 2056 2057 kvfree(nat_bits); 2058 } 2059 2060 static inline bool enabled_nat_bits(struct f2fs_sb_info *sbi, 2061 struct cp_control *cpc) 2062 { 2063 bool set = is_set_ckpt_flags(sbi, CP_NAT_BITS_FLAG); 2064 2065 return (cpc) ? (cpc->reason & CP_UMOUNT) && set : set; 2066 } 2067 2068 static inline void f2fs_lock_op(struct f2fs_sb_info *sbi) 2069 { 2070 down_read(&sbi->cp_rwsem); 2071 } 2072 2073 static inline int f2fs_trylock_op(struct f2fs_sb_info *sbi) 2074 { 2075 return down_read_trylock(&sbi->cp_rwsem); 2076 } 2077 2078 static inline void f2fs_unlock_op(struct f2fs_sb_info *sbi) 2079 { 2080 up_read(&sbi->cp_rwsem); 2081 } 2082 2083 static inline void f2fs_lock_all(struct f2fs_sb_info *sbi) 2084 { 2085 down_write(&sbi->cp_rwsem); 2086 } 2087 2088 static inline void f2fs_unlock_all(struct f2fs_sb_info *sbi) 2089 { 2090 up_write(&sbi->cp_rwsem); 2091 } 2092 2093 static inline int __get_cp_reason(struct f2fs_sb_info *sbi) 2094 { 2095 int reason = CP_SYNC; 2096 2097 if (test_opt(sbi, FASTBOOT)) 2098 reason = CP_FASTBOOT; 2099 if (is_sbi_flag_set(sbi, SBI_IS_CLOSE)) 2100 reason = CP_UMOUNT; 2101 return reason; 2102 } 2103 2104 static inline bool __remain_node_summaries(int reason) 2105 { 2106 return (reason & (CP_UMOUNT | CP_FASTBOOT)); 2107 } 2108 2109 static inline bool __exist_node_summaries(struct f2fs_sb_info *sbi) 2110 { 2111 return (is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG) || 2112 is_set_ckpt_flags(sbi, CP_FASTBOOT_FLAG)); 2113 } 2114 2115 /* 2116 * Check whether the inode has blocks or not 2117 */ 2118 static inline int F2FS_HAS_BLOCKS(struct inode *inode) 2119 { 2120 block_t xattr_block = F2FS_I(inode)->i_xattr_nid ? 1 : 0; 2121 2122 return (inode->i_blocks >> F2FS_LOG_SECTORS_PER_BLOCK) > xattr_block; 2123 } 2124 2125 static inline bool f2fs_has_xattr_block(unsigned int ofs) 2126 { 2127 return ofs == XATTR_NODE_OFFSET; 2128 } 2129 2130 static inline bool __allow_reserved_blocks(struct f2fs_sb_info *sbi, 2131 struct inode *inode, bool cap) 2132 { 2133 if (!inode) 2134 return true; 2135 if (!test_opt(sbi, RESERVE_ROOT)) 2136 return false; 2137 if (IS_NOQUOTA(inode)) 2138 return true; 2139 if (uid_eq(F2FS_OPTION(sbi).s_resuid, current_fsuid())) 2140 return true; 2141 if (!gid_eq(F2FS_OPTION(sbi).s_resgid, GLOBAL_ROOT_GID) && 2142 in_group_p(F2FS_OPTION(sbi).s_resgid)) 2143 return true; 2144 if (cap && capable(CAP_SYS_RESOURCE)) 2145 return true; 2146 return false; 2147 } 2148 2149 static inline void f2fs_i_blocks_write(struct inode *, block_t, bool, bool); 2150 static inline int inc_valid_block_count(struct f2fs_sb_info *sbi, 2151 struct inode *inode, blkcnt_t *count) 2152 { 2153 blkcnt_t diff = 0, release = 0; 2154 block_t avail_user_block_count; 2155 int ret; 2156 2157 ret = dquot_reserve_block(inode, *count); 2158 if (ret) 2159 return ret; 2160 2161 if (time_to_inject(sbi, FAULT_BLOCK)) { 2162 f2fs_show_injection_info(sbi, FAULT_BLOCK); 2163 release = *count; 2164 goto release_quota; 2165 } 2166 2167 /* 2168 * let's increase this in prior to actual block count change in order 2169 * for f2fs_sync_file to avoid data races when deciding checkpoint. 2170 */ 2171 percpu_counter_add(&sbi->alloc_valid_block_count, (*count)); 2172 2173 spin_lock(&sbi->stat_lock); 2174 sbi->total_valid_block_count += (block_t)(*count); 2175 avail_user_block_count = sbi->user_block_count - 2176 sbi->current_reserved_blocks; 2177 2178 if (!__allow_reserved_blocks(sbi, inode, true)) 2179 avail_user_block_count -= F2FS_OPTION(sbi).root_reserved_blocks; 2180 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) { 2181 if (avail_user_block_count > sbi->unusable_block_count) 2182 avail_user_block_count -= sbi->unusable_block_count; 2183 else 2184 avail_user_block_count = 0; 2185 } 2186 if (unlikely(sbi->total_valid_block_count > avail_user_block_count)) { 2187 diff = sbi->total_valid_block_count - avail_user_block_count; 2188 if (diff > *count) 2189 diff = *count; 2190 *count -= diff; 2191 release = diff; 2192 sbi->total_valid_block_count -= diff; 2193 if (!*count) { 2194 spin_unlock(&sbi->stat_lock); 2195 goto enospc; 2196 } 2197 } 2198 spin_unlock(&sbi->stat_lock); 2199 2200 if (unlikely(release)) { 2201 percpu_counter_sub(&sbi->alloc_valid_block_count, release); 2202 dquot_release_reservation_block(inode, release); 2203 } 2204 f2fs_i_blocks_write(inode, *count, true, true); 2205 return 0; 2206 2207 enospc: 2208 percpu_counter_sub(&sbi->alloc_valid_block_count, release); 2209 release_quota: 2210 dquot_release_reservation_block(inode, release); 2211 return -ENOSPC; 2212 } 2213 2214 __printf(2, 3) 2215 void f2fs_printk(struct f2fs_sb_info *sbi, const char *fmt, ...); 2216 2217 #define f2fs_err(sbi, fmt, ...) \ 2218 f2fs_printk(sbi, KERN_ERR fmt, ##__VA_ARGS__) 2219 #define f2fs_warn(sbi, fmt, ...) \ 2220 f2fs_printk(sbi, KERN_WARNING fmt, ##__VA_ARGS__) 2221 #define f2fs_notice(sbi, fmt, ...) \ 2222 f2fs_printk(sbi, KERN_NOTICE fmt, ##__VA_ARGS__) 2223 #define f2fs_info(sbi, fmt, ...) \ 2224 f2fs_printk(sbi, KERN_INFO fmt, ##__VA_ARGS__) 2225 #define f2fs_debug(sbi, fmt, ...) \ 2226 f2fs_printk(sbi, KERN_DEBUG fmt, ##__VA_ARGS__) 2227 2228 static inline void dec_valid_block_count(struct f2fs_sb_info *sbi, 2229 struct inode *inode, 2230 block_t count) 2231 { 2232 blkcnt_t sectors = count << F2FS_LOG_SECTORS_PER_BLOCK; 2233 2234 spin_lock(&sbi->stat_lock); 2235 f2fs_bug_on(sbi, sbi->total_valid_block_count < (block_t) count); 2236 sbi->total_valid_block_count -= (block_t)count; 2237 if (sbi->reserved_blocks && 2238 sbi->current_reserved_blocks < sbi->reserved_blocks) 2239 sbi->current_reserved_blocks = min(sbi->reserved_blocks, 2240 sbi->current_reserved_blocks + count); 2241 spin_unlock(&sbi->stat_lock); 2242 if (unlikely(inode->i_blocks < sectors)) { 2243 f2fs_warn(sbi, "Inconsistent i_blocks, ino:%lu, iblocks:%llu, sectors:%llu", 2244 inode->i_ino, 2245 (unsigned long long)inode->i_blocks, 2246 (unsigned long long)sectors); 2247 set_sbi_flag(sbi, SBI_NEED_FSCK); 2248 return; 2249 } 2250 f2fs_i_blocks_write(inode, count, false, true); 2251 } 2252 2253 static inline void inc_page_count(struct f2fs_sb_info *sbi, int count_type) 2254 { 2255 atomic_inc(&sbi->nr_pages[count_type]); 2256 2257 if (count_type == F2FS_DIRTY_DENTS || 2258 count_type == F2FS_DIRTY_NODES || 2259 count_type == F2FS_DIRTY_META || 2260 count_type == F2FS_DIRTY_QDATA || 2261 count_type == F2FS_DIRTY_IMETA) 2262 set_sbi_flag(sbi, SBI_IS_DIRTY); 2263 } 2264 2265 static inline void inode_inc_dirty_pages(struct inode *inode) 2266 { 2267 atomic_inc(&F2FS_I(inode)->dirty_pages); 2268 inc_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ? 2269 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA); 2270 if (IS_NOQUOTA(inode)) 2271 inc_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA); 2272 } 2273 2274 static inline void dec_page_count(struct f2fs_sb_info *sbi, int count_type) 2275 { 2276 atomic_dec(&sbi->nr_pages[count_type]); 2277 } 2278 2279 static inline void inode_dec_dirty_pages(struct inode *inode) 2280 { 2281 if (!S_ISDIR(inode->i_mode) && !S_ISREG(inode->i_mode) && 2282 !S_ISLNK(inode->i_mode)) 2283 return; 2284 2285 atomic_dec(&F2FS_I(inode)->dirty_pages); 2286 dec_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ? 2287 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA); 2288 if (IS_NOQUOTA(inode)) 2289 dec_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA); 2290 } 2291 2292 static inline s64 get_pages(struct f2fs_sb_info *sbi, int count_type) 2293 { 2294 return atomic_read(&sbi->nr_pages[count_type]); 2295 } 2296 2297 static inline int get_dirty_pages(struct inode *inode) 2298 { 2299 return atomic_read(&F2FS_I(inode)->dirty_pages); 2300 } 2301 2302 static inline int get_blocktype_secs(struct f2fs_sb_info *sbi, int block_type) 2303 { 2304 unsigned int pages_per_sec = sbi->segs_per_sec * sbi->blocks_per_seg; 2305 unsigned int segs = (get_pages(sbi, block_type) + pages_per_sec - 1) >> 2306 sbi->log_blocks_per_seg; 2307 2308 return segs / sbi->segs_per_sec; 2309 } 2310 2311 static inline block_t valid_user_blocks(struct f2fs_sb_info *sbi) 2312 { 2313 return sbi->total_valid_block_count; 2314 } 2315 2316 static inline block_t discard_blocks(struct f2fs_sb_info *sbi) 2317 { 2318 return sbi->discard_blks; 2319 } 2320 2321 static inline unsigned long __bitmap_size(struct f2fs_sb_info *sbi, int flag) 2322 { 2323 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi); 2324 2325 /* return NAT or SIT bitmap */ 2326 if (flag == NAT_BITMAP) 2327 return le32_to_cpu(ckpt->nat_ver_bitmap_bytesize); 2328 else if (flag == SIT_BITMAP) 2329 return le32_to_cpu(ckpt->sit_ver_bitmap_bytesize); 2330 2331 return 0; 2332 } 2333 2334 static inline block_t __cp_payload(struct f2fs_sb_info *sbi) 2335 { 2336 return le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_payload); 2337 } 2338 2339 static inline void *__bitmap_ptr(struct f2fs_sb_info *sbi, int flag) 2340 { 2341 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi); 2342 void *tmp_ptr = &ckpt->sit_nat_version_bitmap; 2343 int offset; 2344 2345 if (is_set_ckpt_flags(sbi, CP_LARGE_NAT_BITMAP_FLAG)) { 2346 offset = (flag == SIT_BITMAP) ? 2347 le32_to_cpu(ckpt->nat_ver_bitmap_bytesize) : 0; 2348 /* 2349 * if large_nat_bitmap feature is enabled, leave checksum 2350 * protection for all nat/sit bitmaps. 2351 */ 2352 return tmp_ptr + offset + sizeof(__le32); 2353 } 2354 2355 if (__cp_payload(sbi) > 0) { 2356 if (flag == NAT_BITMAP) 2357 return &ckpt->sit_nat_version_bitmap; 2358 else 2359 return (unsigned char *)ckpt + F2FS_BLKSIZE; 2360 } else { 2361 offset = (flag == NAT_BITMAP) ? 2362 le32_to_cpu(ckpt->sit_ver_bitmap_bytesize) : 0; 2363 return tmp_ptr + offset; 2364 } 2365 } 2366 2367 static inline block_t __start_cp_addr(struct f2fs_sb_info *sbi) 2368 { 2369 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr); 2370 2371 if (sbi->cur_cp_pack == 2) 2372 start_addr += sbi->blocks_per_seg; 2373 return start_addr; 2374 } 2375 2376 static inline block_t __start_cp_next_addr(struct f2fs_sb_info *sbi) 2377 { 2378 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr); 2379 2380 if (sbi->cur_cp_pack == 1) 2381 start_addr += sbi->blocks_per_seg; 2382 return start_addr; 2383 } 2384 2385 static inline void __set_cp_next_pack(struct f2fs_sb_info *sbi) 2386 { 2387 sbi->cur_cp_pack = (sbi->cur_cp_pack == 1) ? 2 : 1; 2388 } 2389 2390 static inline block_t __start_sum_addr(struct f2fs_sb_info *sbi) 2391 { 2392 return le32_to_cpu(F2FS_CKPT(sbi)->cp_pack_start_sum); 2393 } 2394 2395 static inline int inc_valid_node_count(struct f2fs_sb_info *sbi, 2396 struct inode *inode, bool is_inode) 2397 { 2398 block_t valid_block_count; 2399 unsigned int valid_node_count, user_block_count; 2400 int err; 2401 2402 if (is_inode) { 2403 if (inode) { 2404 err = dquot_alloc_inode(inode); 2405 if (err) 2406 return err; 2407 } 2408 } else { 2409 err = dquot_reserve_block(inode, 1); 2410 if (err) 2411 return err; 2412 } 2413 2414 if (time_to_inject(sbi, FAULT_BLOCK)) { 2415 f2fs_show_injection_info(sbi, FAULT_BLOCK); 2416 goto enospc; 2417 } 2418 2419 spin_lock(&sbi->stat_lock); 2420 2421 valid_block_count = sbi->total_valid_block_count + 2422 sbi->current_reserved_blocks + 1; 2423 2424 if (!__allow_reserved_blocks(sbi, inode, false)) 2425 valid_block_count += F2FS_OPTION(sbi).root_reserved_blocks; 2426 user_block_count = sbi->user_block_count; 2427 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) 2428 user_block_count -= sbi->unusable_block_count; 2429 2430 if (unlikely(valid_block_count > user_block_count)) { 2431 spin_unlock(&sbi->stat_lock); 2432 goto enospc; 2433 } 2434 2435 valid_node_count = sbi->total_valid_node_count + 1; 2436 if (unlikely(valid_node_count > sbi->total_node_count)) { 2437 spin_unlock(&sbi->stat_lock); 2438 goto enospc; 2439 } 2440 2441 sbi->total_valid_node_count++; 2442 sbi->total_valid_block_count++; 2443 spin_unlock(&sbi->stat_lock); 2444 2445 if (inode) { 2446 if (is_inode) 2447 f2fs_mark_inode_dirty_sync(inode, true); 2448 else 2449 f2fs_i_blocks_write(inode, 1, true, true); 2450 } 2451 2452 percpu_counter_inc(&sbi->alloc_valid_block_count); 2453 return 0; 2454 2455 enospc: 2456 if (is_inode) { 2457 if (inode) 2458 dquot_free_inode(inode); 2459 } else { 2460 dquot_release_reservation_block(inode, 1); 2461 } 2462 return -ENOSPC; 2463 } 2464 2465 static inline void dec_valid_node_count(struct f2fs_sb_info *sbi, 2466 struct inode *inode, bool is_inode) 2467 { 2468 spin_lock(&sbi->stat_lock); 2469 2470 f2fs_bug_on(sbi, !sbi->total_valid_block_count); 2471 f2fs_bug_on(sbi, !sbi->total_valid_node_count); 2472 2473 sbi->total_valid_node_count--; 2474 sbi->total_valid_block_count--; 2475 if (sbi->reserved_blocks && 2476 sbi->current_reserved_blocks < sbi->reserved_blocks) 2477 sbi->current_reserved_blocks++; 2478 2479 spin_unlock(&sbi->stat_lock); 2480 2481 if (is_inode) { 2482 dquot_free_inode(inode); 2483 } else { 2484 if (unlikely(inode->i_blocks == 0)) { 2485 f2fs_warn(sbi, "dec_valid_node_count: inconsistent i_blocks, ino:%lu, iblocks:%llu", 2486 inode->i_ino, 2487 (unsigned long long)inode->i_blocks); 2488 set_sbi_flag(sbi, SBI_NEED_FSCK); 2489 return; 2490 } 2491 f2fs_i_blocks_write(inode, 1, false, true); 2492 } 2493 } 2494 2495 static inline unsigned int valid_node_count(struct f2fs_sb_info *sbi) 2496 { 2497 return sbi->total_valid_node_count; 2498 } 2499 2500 static inline void inc_valid_inode_count(struct f2fs_sb_info *sbi) 2501 { 2502 percpu_counter_inc(&sbi->total_valid_inode_count); 2503 } 2504 2505 static inline void dec_valid_inode_count(struct f2fs_sb_info *sbi) 2506 { 2507 percpu_counter_dec(&sbi->total_valid_inode_count); 2508 } 2509 2510 static inline s64 valid_inode_count(struct f2fs_sb_info *sbi) 2511 { 2512 return percpu_counter_sum_positive(&sbi->total_valid_inode_count); 2513 } 2514 2515 static inline struct page *f2fs_grab_cache_page(struct address_space *mapping, 2516 pgoff_t index, bool for_write) 2517 { 2518 struct page *page; 2519 2520 if (IS_ENABLED(CONFIG_F2FS_FAULT_INJECTION)) { 2521 if (!for_write) 2522 page = find_get_page_flags(mapping, index, 2523 FGP_LOCK | FGP_ACCESSED); 2524 else 2525 page = find_lock_page(mapping, index); 2526 if (page) 2527 return page; 2528 2529 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_ALLOC)) { 2530 f2fs_show_injection_info(F2FS_M_SB(mapping), 2531 FAULT_PAGE_ALLOC); 2532 return NULL; 2533 } 2534 } 2535 2536 if (!for_write) 2537 return grab_cache_page(mapping, index); 2538 return grab_cache_page_write_begin(mapping, index, AOP_FLAG_NOFS); 2539 } 2540 2541 static inline struct page *f2fs_pagecache_get_page( 2542 struct address_space *mapping, pgoff_t index, 2543 int fgp_flags, gfp_t gfp_mask) 2544 { 2545 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_GET)) { 2546 f2fs_show_injection_info(F2FS_M_SB(mapping), FAULT_PAGE_GET); 2547 return NULL; 2548 } 2549 2550 return pagecache_get_page(mapping, index, fgp_flags, gfp_mask); 2551 } 2552 2553 static inline void f2fs_copy_page(struct page *src, struct page *dst) 2554 { 2555 char *src_kaddr = kmap(src); 2556 char *dst_kaddr = kmap(dst); 2557 2558 memcpy(dst_kaddr, src_kaddr, PAGE_SIZE); 2559 kunmap(dst); 2560 kunmap(src); 2561 } 2562 2563 static inline void f2fs_put_page(struct page *page, int unlock) 2564 { 2565 if (!page) 2566 return; 2567 2568 if (unlock) { 2569 f2fs_bug_on(F2FS_P_SB(page), !PageLocked(page)); 2570 unlock_page(page); 2571 } 2572 put_page(page); 2573 } 2574 2575 static inline void f2fs_put_dnode(struct dnode_of_data *dn) 2576 { 2577 if (dn->node_page) 2578 f2fs_put_page(dn->node_page, 1); 2579 if (dn->inode_page && dn->node_page != dn->inode_page) 2580 f2fs_put_page(dn->inode_page, 0); 2581 dn->node_page = NULL; 2582 dn->inode_page = NULL; 2583 } 2584 2585 static inline struct kmem_cache *f2fs_kmem_cache_create(const char *name, 2586 size_t size) 2587 { 2588 return kmem_cache_create(name, size, 0, SLAB_RECLAIM_ACCOUNT, NULL); 2589 } 2590 2591 static inline void *f2fs_kmem_cache_alloc(struct kmem_cache *cachep, 2592 gfp_t flags) 2593 { 2594 void *entry; 2595 2596 entry = kmem_cache_alloc(cachep, flags); 2597 if (!entry) 2598 entry = kmem_cache_alloc(cachep, flags | __GFP_NOFAIL); 2599 return entry; 2600 } 2601 2602 static inline bool is_inflight_io(struct f2fs_sb_info *sbi, int type) 2603 { 2604 if (get_pages(sbi, F2FS_RD_DATA) || get_pages(sbi, F2FS_RD_NODE) || 2605 get_pages(sbi, F2FS_RD_META) || get_pages(sbi, F2FS_WB_DATA) || 2606 get_pages(sbi, F2FS_WB_CP_DATA) || 2607 get_pages(sbi, F2FS_DIO_READ) || 2608 get_pages(sbi, F2FS_DIO_WRITE)) 2609 return true; 2610 2611 if (type != DISCARD_TIME && SM_I(sbi) && SM_I(sbi)->dcc_info && 2612 atomic_read(&SM_I(sbi)->dcc_info->queued_discard)) 2613 return true; 2614 2615 if (SM_I(sbi) && SM_I(sbi)->fcc_info && 2616 atomic_read(&SM_I(sbi)->fcc_info->queued_flush)) 2617 return true; 2618 return false; 2619 } 2620 2621 static inline bool is_idle(struct f2fs_sb_info *sbi, int type) 2622 { 2623 if (sbi->gc_mode == GC_URGENT_HIGH) 2624 return true; 2625 2626 if (is_inflight_io(sbi, type)) 2627 return false; 2628 2629 if (sbi->gc_mode == GC_URGENT_LOW && 2630 (type == DISCARD_TIME || type == GC_TIME)) 2631 return true; 2632 2633 return f2fs_time_over(sbi, type); 2634 } 2635 2636 static inline void f2fs_radix_tree_insert(struct radix_tree_root *root, 2637 unsigned long index, void *item) 2638 { 2639 while (radix_tree_insert(root, index, item)) 2640 cond_resched(); 2641 } 2642 2643 #define RAW_IS_INODE(p) ((p)->footer.nid == (p)->footer.ino) 2644 2645 static inline bool IS_INODE(struct page *page) 2646 { 2647 struct f2fs_node *p = F2FS_NODE(page); 2648 2649 return RAW_IS_INODE(p); 2650 } 2651 2652 static inline int offset_in_addr(struct f2fs_inode *i) 2653 { 2654 return (i->i_inline & F2FS_EXTRA_ATTR) ? 2655 (le16_to_cpu(i->i_extra_isize) / sizeof(__le32)) : 0; 2656 } 2657 2658 static inline __le32 *blkaddr_in_node(struct f2fs_node *node) 2659 { 2660 return RAW_IS_INODE(node) ? node->i.i_addr : node->dn.addr; 2661 } 2662 2663 static inline int f2fs_has_extra_attr(struct inode *inode); 2664 static inline block_t data_blkaddr(struct inode *inode, 2665 struct page *node_page, unsigned int offset) 2666 { 2667 struct f2fs_node *raw_node; 2668 __le32 *addr_array; 2669 int base = 0; 2670 bool is_inode = IS_INODE(node_page); 2671 2672 raw_node = F2FS_NODE(node_page); 2673 2674 if (is_inode) { 2675 if (!inode) 2676 /* from GC path only */ 2677 base = offset_in_addr(&raw_node->i); 2678 else if (f2fs_has_extra_attr(inode)) 2679 base = get_extra_isize(inode); 2680 } 2681 2682 addr_array = blkaddr_in_node(raw_node); 2683 return le32_to_cpu(addr_array[base + offset]); 2684 } 2685 2686 static inline block_t f2fs_data_blkaddr(struct dnode_of_data *dn) 2687 { 2688 return data_blkaddr(dn->inode, dn->node_page, dn->ofs_in_node); 2689 } 2690 2691 static inline int f2fs_test_bit(unsigned int nr, char *addr) 2692 { 2693 int mask; 2694 2695 addr += (nr >> 3); 2696 mask = 1 << (7 - (nr & 0x07)); 2697 return mask & *addr; 2698 } 2699 2700 static inline void f2fs_set_bit(unsigned int nr, char *addr) 2701 { 2702 int mask; 2703 2704 addr += (nr >> 3); 2705 mask = 1 << (7 - (nr & 0x07)); 2706 *addr |= mask; 2707 } 2708 2709 static inline void f2fs_clear_bit(unsigned int nr, char *addr) 2710 { 2711 int mask; 2712 2713 addr += (nr >> 3); 2714 mask = 1 << (7 - (nr & 0x07)); 2715 *addr &= ~mask; 2716 } 2717 2718 static inline int f2fs_test_and_set_bit(unsigned int nr, char *addr) 2719 { 2720 int mask; 2721 int ret; 2722 2723 addr += (nr >> 3); 2724 mask = 1 << (7 - (nr & 0x07)); 2725 ret = mask & *addr; 2726 *addr |= mask; 2727 return ret; 2728 } 2729 2730 static inline int f2fs_test_and_clear_bit(unsigned int nr, char *addr) 2731 { 2732 int mask; 2733 int ret; 2734 2735 addr += (nr >> 3); 2736 mask = 1 << (7 - (nr & 0x07)); 2737 ret = mask & *addr; 2738 *addr &= ~mask; 2739 return ret; 2740 } 2741 2742 static inline void f2fs_change_bit(unsigned int nr, char *addr) 2743 { 2744 int mask; 2745 2746 addr += (nr >> 3); 2747 mask = 1 << (7 - (nr & 0x07)); 2748 *addr ^= mask; 2749 } 2750 2751 /* 2752 * On-disk inode flags (f2fs_inode::i_flags) 2753 */ 2754 #define F2FS_COMPR_FL 0x00000004 /* Compress file */ 2755 #define F2FS_SYNC_FL 0x00000008 /* Synchronous updates */ 2756 #define F2FS_IMMUTABLE_FL 0x00000010 /* Immutable file */ 2757 #define F2FS_APPEND_FL 0x00000020 /* writes to file may only append */ 2758 #define F2FS_NODUMP_FL 0x00000040 /* do not dump file */ 2759 #define F2FS_NOATIME_FL 0x00000080 /* do not update atime */ 2760 #define F2FS_NOCOMP_FL 0x00000400 /* Don't compress */ 2761 #define F2FS_INDEX_FL 0x00001000 /* hash-indexed directory */ 2762 #define F2FS_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */ 2763 #define F2FS_PROJINHERIT_FL 0x20000000 /* Create with parents projid */ 2764 #define F2FS_CASEFOLD_FL 0x40000000 /* Casefolded file */ 2765 2766 /* Flags that should be inherited by new inodes from their parent. */ 2767 #define F2FS_FL_INHERITED (F2FS_SYNC_FL | F2FS_NODUMP_FL | F2FS_NOATIME_FL | \ 2768 F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \ 2769 F2FS_CASEFOLD_FL | F2FS_COMPR_FL | F2FS_NOCOMP_FL) 2770 2771 /* Flags that are appropriate for regular files (all but dir-specific ones). */ 2772 #define F2FS_REG_FLMASK (~(F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \ 2773 F2FS_CASEFOLD_FL)) 2774 2775 /* Flags that are appropriate for non-directories/regular files. */ 2776 #define F2FS_OTHER_FLMASK (F2FS_NODUMP_FL | F2FS_NOATIME_FL) 2777 2778 static inline __u32 f2fs_mask_flags(umode_t mode, __u32 flags) 2779 { 2780 if (S_ISDIR(mode)) 2781 return flags; 2782 else if (S_ISREG(mode)) 2783 return flags & F2FS_REG_FLMASK; 2784 else 2785 return flags & F2FS_OTHER_FLMASK; 2786 } 2787 2788 static inline void __mark_inode_dirty_flag(struct inode *inode, 2789 int flag, bool set) 2790 { 2791 switch (flag) { 2792 case FI_INLINE_XATTR: 2793 case FI_INLINE_DATA: 2794 case FI_INLINE_DENTRY: 2795 case FI_NEW_INODE: 2796 if (set) 2797 return; 2798 fallthrough; 2799 case FI_DATA_EXIST: 2800 case FI_INLINE_DOTS: 2801 case FI_PIN_FILE: 2802 case FI_COMPRESS_RELEASED: 2803 f2fs_mark_inode_dirty_sync(inode, true); 2804 } 2805 } 2806 2807 static inline void set_inode_flag(struct inode *inode, int flag) 2808 { 2809 set_bit(flag, F2FS_I(inode)->flags); 2810 __mark_inode_dirty_flag(inode, flag, true); 2811 } 2812 2813 static inline int is_inode_flag_set(struct inode *inode, int flag) 2814 { 2815 return test_bit(flag, F2FS_I(inode)->flags); 2816 } 2817 2818 static inline void clear_inode_flag(struct inode *inode, int flag) 2819 { 2820 clear_bit(flag, F2FS_I(inode)->flags); 2821 __mark_inode_dirty_flag(inode, flag, false); 2822 } 2823 2824 static inline bool f2fs_verity_in_progress(struct inode *inode) 2825 { 2826 return IS_ENABLED(CONFIG_FS_VERITY) && 2827 is_inode_flag_set(inode, FI_VERITY_IN_PROGRESS); 2828 } 2829 2830 static inline void set_acl_inode(struct inode *inode, umode_t mode) 2831 { 2832 F2FS_I(inode)->i_acl_mode = mode; 2833 set_inode_flag(inode, FI_ACL_MODE); 2834 f2fs_mark_inode_dirty_sync(inode, false); 2835 } 2836 2837 static inline void f2fs_i_links_write(struct inode *inode, bool inc) 2838 { 2839 if (inc) 2840 inc_nlink(inode); 2841 else 2842 drop_nlink(inode); 2843 f2fs_mark_inode_dirty_sync(inode, true); 2844 } 2845 2846 static inline void f2fs_i_blocks_write(struct inode *inode, 2847 block_t diff, bool add, bool claim) 2848 { 2849 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE); 2850 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER); 2851 2852 /* add = 1, claim = 1 should be dquot_reserve_block in pair */ 2853 if (add) { 2854 if (claim) 2855 dquot_claim_block(inode, diff); 2856 else 2857 dquot_alloc_block_nofail(inode, diff); 2858 } else { 2859 dquot_free_block(inode, diff); 2860 } 2861 2862 f2fs_mark_inode_dirty_sync(inode, true); 2863 if (clean || recover) 2864 set_inode_flag(inode, FI_AUTO_RECOVER); 2865 } 2866 2867 static inline void f2fs_i_size_write(struct inode *inode, loff_t i_size) 2868 { 2869 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE); 2870 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER); 2871 2872 if (i_size_read(inode) == i_size) 2873 return; 2874 2875 i_size_write(inode, i_size); 2876 f2fs_mark_inode_dirty_sync(inode, true); 2877 if (clean || recover) 2878 set_inode_flag(inode, FI_AUTO_RECOVER); 2879 } 2880 2881 static inline void f2fs_i_depth_write(struct inode *inode, unsigned int depth) 2882 { 2883 F2FS_I(inode)->i_current_depth = depth; 2884 f2fs_mark_inode_dirty_sync(inode, true); 2885 } 2886 2887 static inline void f2fs_i_gc_failures_write(struct inode *inode, 2888 unsigned int count) 2889 { 2890 F2FS_I(inode)->i_gc_failures[GC_FAILURE_PIN] = count; 2891 f2fs_mark_inode_dirty_sync(inode, true); 2892 } 2893 2894 static inline void f2fs_i_xnid_write(struct inode *inode, nid_t xnid) 2895 { 2896 F2FS_I(inode)->i_xattr_nid = xnid; 2897 f2fs_mark_inode_dirty_sync(inode, true); 2898 } 2899 2900 static inline void f2fs_i_pino_write(struct inode *inode, nid_t pino) 2901 { 2902 F2FS_I(inode)->i_pino = pino; 2903 f2fs_mark_inode_dirty_sync(inode, true); 2904 } 2905 2906 static inline void get_inline_info(struct inode *inode, struct f2fs_inode *ri) 2907 { 2908 struct f2fs_inode_info *fi = F2FS_I(inode); 2909 2910 if (ri->i_inline & F2FS_INLINE_XATTR) 2911 set_bit(FI_INLINE_XATTR, fi->flags); 2912 if (ri->i_inline & F2FS_INLINE_DATA) 2913 set_bit(FI_INLINE_DATA, fi->flags); 2914 if (ri->i_inline & F2FS_INLINE_DENTRY) 2915 set_bit(FI_INLINE_DENTRY, fi->flags); 2916 if (ri->i_inline & F2FS_DATA_EXIST) 2917 set_bit(FI_DATA_EXIST, fi->flags); 2918 if (ri->i_inline & F2FS_INLINE_DOTS) 2919 set_bit(FI_INLINE_DOTS, fi->flags); 2920 if (ri->i_inline & F2FS_EXTRA_ATTR) 2921 set_bit(FI_EXTRA_ATTR, fi->flags); 2922 if (ri->i_inline & F2FS_PIN_FILE) 2923 set_bit(FI_PIN_FILE, fi->flags); 2924 if (ri->i_inline & F2FS_COMPRESS_RELEASED) 2925 set_bit(FI_COMPRESS_RELEASED, fi->flags); 2926 } 2927 2928 static inline void set_raw_inline(struct inode *inode, struct f2fs_inode *ri) 2929 { 2930 ri->i_inline = 0; 2931 2932 if (is_inode_flag_set(inode, FI_INLINE_XATTR)) 2933 ri->i_inline |= F2FS_INLINE_XATTR; 2934 if (is_inode_flag_set(inode, FI_INLINE_DATA)) 2935 ri->i_inline |= F2FS_INLINE_DATA; 2936 if (is_inode_flag_set(inode, FI_INLINE_DENTRY)) 2937 ri->i_inline |= F2FS_INLINE_DENTRY; 2938 if (is_inode_flag_set(inode, FI_DATA_EXIST)) 2939 ri->i_inline |= F2FS_DATA_EXIST; 2940 if (is_inode_flag_set(inode, FI_INLINE_DOTS)) 2941 ri->i_inline |= F2FS_INLINE_DOTS; 2942 if (is_inode_flag_set(inode, FI_EXTRA_ATTR)) 2943 ri->i_inline |= F2FS_EXTRA_ATTR; 2944 if (is_inode_flag_set(inode, FI_PIN_FILE)) 2945 ri->i_inline |= F2FS_PIN_FILE; 2946 if (is_inode_flag_set(inode, FI_COMPRESS_RELEASED)) 2947 ri->i_inline |= F2FS_COMPRESS_RELEASED; 2948 } 2949 2950 static inline int f2fs_has_extra_attr(struct inode *inode) 2951 { 2952 return is_inode_flag_set(inode, FI_EXTRA_ATTR); 2953 } 2954 2955 static inline int f2fs_has_inline_xattr(struct inode *inode) 2956 { 2957 return is_inode_flag_set(inode, FI_INLINE_XATTR); 2958 } 2959 2960 static inline int f2fs_compressed_file(struct inode *inode) 2961 { 2962 return S_ISREG(inode->i_mode) && 2963 is_inode_flag_set(inode, FI_COMPRESSED_FILE); 2964 } 2965 2966 static inline bool f2fs_need_compress_data(struct inode *inode) 2967 { 2968 int compress_mode = F2FS_OPTION(F2FS_I_SB(inode)).compress_mode; 2969 2970 if (!f2fs_compressed_file(inode)) 2971 return false; 2972 2973 if (compress_mode == COMPR_MODE_FS) 2974 return true; 2975 else if (compress_mode == COMPR_MODE_USER && 2976 is_inode_flag_set(inode, FI_ENABLE_COMPRESS)) 2977 return true; 2978 2979 return false; 2980 } 2981 2982 static inline unsigned int addrs_per_inode(struct inode *inode) 2983 { 2984 unsigned int addrs = CUR_ADDRS_PER_INODE(inode) - 2985 get_inline_xattr_addrs(inode); 2986 2987 if (!f2fs_compressed_file(inode)) 2988 return addrs; 2989 return ALIGN_DOWN(addrs, F2FS_I(inode)->i_cluster_size); 2990 } 2991 2992 static inline unsigned int addrs_per_block(struct inode *inode) 2993 { 2994 if (!f2fs_compressed_file(inode)) 2995 return DEF_ADDRS_PER_BLOCK; 2996 return ALIGN_DOWN(DEF_ADDRS_PER_BLOCK, F2FS_I(inode)->i_cluster_size); 2997 } 2998 2999 static inline void *inline_xattr_addr(struct inode *inode, struct page *page) 3000 { 3001 struct f2fs_inode *ri = F2FS_INODE(page); 3002 3003 return (void *)&(ri->i_addr[DEF_ADDRS_PER_INODE - 3004 get_inline_xattr_addrs(inode)]); 3005 } 3006 3007 static inline int inline_xattr_size(struct inode *inode) 3008 { 3009 if (f2fs_has_inline_xattr(inode)) 3010 return get_inline_xattr_addrs(inode) * sizeof(__le32); 3011 return 0; 3012 } 3013 3014 static inline int f2fs_has_inline_data(struct inode *inode) 3015 { 3016 return is_inode_flag_set(inode, FI_INLINE_DATA); 3017 } 3018 3019 static inline int f2fs_exist_data(struct inode *inode) 3020 { 3021 return is_inode_flag_set(inode, FI_DATA_EXIST); 3022 } 3023 3024 static inline int f2fs_has_inline_dots(struct inode *inode) 3025 { 3026 return is_inode_flag_set(inode, FI_INLINE_DOTS); 3027 } 3028 3029 static inline int f2fs_is_mmap_file(struct inode *inode) 3030 { 3031 return is_inode_flag_set(inode, FI_MMAP_FILE); 3032 } 3033 3034 static inline bool f2fs_is_pinned_file(struct inode *inode) 3035 { 3036 return is_inode_flag_set(inode, FI_PIN_FILE); 3037 } 3038 3039 static inline bool f2fs_is_atomic_file(struct inode *inode) 3040 { 3041 return is_inode_flag_set(inode, FI_ATOMIC_FILE); 3042 } 3043 3044 static inline bool f2fs_is_commit_atomic_write(struct inode *inode) 3045 { 3046 return is_inode_flag_set(inode, FI_ATOMIC_COMMIT); 3047 } 3048 3049 static inline bool f2fs_is_volatile_file(struct inode *inode) 3050 { 3051 return is_inode_flag_set(inode, FI_VOLATILE_FILE); 3052 } 3053 3054 static inline bool f2fs_is_first_block_written(struct inode *inode) 3055 { 3056 return is_inode_flag_set(inode, FI_FIRST_BLOCK_WRITTEN); 3057 } 3058 3059 static inline bool f2fs_is_drop_cache(struct inode *inode) 3060 { 3061 return is_inode_flag_set(inode, FI_DROP_CACHE); 3062 } 3063 3064 static inline void *inline_data_addr(struct inode *inode, struct page *page) 3065 { 3066 struct f2fs_inode *ri = F2FS_INODE(page); 3067 int extra_size = get_extra_isize(inode); 3068 3069 return (void *)&(ri->i_addr[extra_size + DEF_INLINE_RESERVED_SIZE]); 3070 } 3071 3072 static inline int f2fs_has_inline_dentry(struct inode *inode) 3073 { 3074 return is_inode_flag_set(inode, FI_INLINE_DENTRY); 3075 } 3076 3077 static inline int is_file(struct inode *inode, int type) 3078 { 3079 return F2FS_I(inode)->i_advise & type; 3080 } 3081 3082 static inline void set_file(struct inode *inode, int type) 3083 { 3084 F2FS_I(inode)->i_advise |= type; 3085 f2fs_mark_inode_dirty_sync(inode, true); 3086 } 3087 3088 static inline void clear_file(struct inode *inode, int type) 3089 { 3090 F2FS_I(inode)->i_advise &= ~type; 3091 f2fs_mark_inode_dirty_sync(inode, true); 3092 } 3093 3094 static inline bool f2fs_is_time_consistent(struct inode *inode) 3095 { 3096 if (!timespec64_equal(F2FS_I(inode)->i_disk_time, &inode->i_atime)) 3097 return false; 3098 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 1, &inode->i_ctime)) 3099 return false; 3100 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 2, &inode->i_mtime)) 3101 return false; 3102 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 3, 3103 &F2FS_I(inode)->i_crtime)) 3104 return false; 3105 return true; 3106 } 3107 3108 static inline bool f2fs_skip_inode_update(struct inode *inode, int dsync) 3109 { 3110 bool ret; 3111 3112 if (dsync) { 3113 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 3114 3115 spin_lock(&sbi->inode_lock[DIRTY_META]); 3116 ret = list_empty(&F2FS_I(inode)->gdirty_list); 3117 spin_unlock(&sbi->inode_lock[DIRTY_META]); 3118 return ret; 3119 } 3120 if (!is_inode_flag_set(inode, FI_AUTO_RECOVER) || 3121 file_keep_isize(inode) || 3122 i_size_read(inode) & ~PAGE_MASK) 3123 return false; 3124 3125 if (!f2fs_is_time_consistent(inode)) 3126 return false; 3127 3128 spin_lock(&F2FS_I(inode)->i_size_lock); 3129 ret = F2FS_I(inode)->last_disk_size == i_size_read(inode); 3130 spin_unlock(&F2FS_I(inode)->i_size_lock); 3131 3132 return ret; 3133 } 3134 3135 static inline bool f2fs_readonly(struct super_block *sb) 3136 { 3137 return sb_rdonly(sb); 3138 } 3139 3140 static inline bool f2fs_cp_error(struct f2fs_sb_info *sbi) 3141 { 3142 return is_set_ckpt_flags(sbi, CP_ERROR_FLAG); 3143 } 3144 3145 static inline bool is_dot_dotdot(const u8 *name, size_t len) 3146 { 3147 if (len == 1 && name[0] == '.') 3148 return true; 3149 3150 if (len == 2 && name[0] == '.' && name[1] == '.') 3151 return true; 3152 3153 return false; 3154 } 3155 3156 static inline void *f2fs_kmalloc(struct f2fs_sb_info *sbi, 3157 size_t size, gfp_t flags) 3158 { 3159 if (time_to_inject(sbi, FAULT_KMALLOC)) { 3160 f2fs_show_injection_info(sbi, FAULT_KMALLOC); 3161 return NULL; 3162 } 3163 3164 return kmalloc(size, flags); 3165 } 3166 3167 static inline void *f2fs_kzalloc(struct f2fs_sb_info *sbi, 3168 size_t size, gfp_t flags) 3169 { 3170 return f2fs_kmalloc(sbi, size, flags | __GFP_ZERO); 3171 } 3172 3173 static inline void *f2fs_kvmalloc(struct f2fs_sb_info *sbi, 3174 size_t size, gfp_t flags) 3175 { 3176 if (time_to_inject(sbi, FAULT_KVMALLOC)) { 3177 f2fs_show_injection_info(sbi, FAULT_KVMALLOC); 3178 return NULL; 3179 } 3180 3181 return kvmalloc(size, flags); 3182 } 3183 3184 static inline void *f2fs_kvzalloc(struct f2fs_sb_info *sbi, 3185 size_t size, gfp_t flags) 3186 { 3187 return f2fs_kvmalloc(sbi, size, flags | __GFP_ZERO); 3188 } 3189 3190 static inline int get_extra_isize(struct inode *inode) 3191 { 3192 return F2FS_I(inode)->i_extra_isize / sizeof(__le32); 3193 } 3194 3195 static inline int get_inline_xattr_addrs(struct inode *inode) 3196 { 3197 return F2FS_I(inode)->i_inline_xattr_size; 3198 } 3199 3200 #define f2fs_get_inode_mode(i) \ 3201 ((is_inode_flag_set(i, FI_ACL_MODE)) ? \ 3202 (F2FS_I(i)->i_acl_mode) : ((i)->i_mode)) 3203 3204 #define F2FS_TOTAL_EXTRA_ATTR_SIZE \ 3205 (offsetof(struct f2fs_inode, i_extra_end) - \ 3206 offsetof(struct f2fs_inode, i_extra_isize)) \ 3207 3208 #define F2FS_OLD_ATTRIBUTE_SIZE (offsetof(struct f2fs_inode, i_addr)) 3209 #define F2FS_FITS_IN_INODE(f2fs_inode, extra_isize, field) \ 3210 ((offsetof(typeof(*(f2fs_inode)), field) + \ 3211 sizeof((f2fs_inode)->field)) \ 3212 <= (F2FS_OLD_ATTRIBUTE_SIZE + (extra_isize))) \ 3213 3214 #define DEFAULT_IOSTAT_PERIOD_MS 3000 3215 #define MIN_IOSTAT_PERIOD_MS 100 3216 /* maximum period of iostat tracing is 1 day */ 3217 #define MAX_IOSTAT_PERIOD_MS 8640000 3218 3219 static inline void f2fs_reset_iostat(struct f2fs_sb_info *sbi) 3220 { 3221 int i; 3222 3223 spin_lock(&sbi->iostat_lock); 3224 for (i = 0; i < NR_IO_TYPE; i++) { 3225 sbi->rw_iostat[i] = 0; 3226 sbi->prev_rw_iostat[i] = 0; 3227 } 3228 spin_unlock(&sbi->iostat_lock); 3229 } 3230 3231 extern void f2fs_record_iostat(struct f2fs_sb_info *sbi); 3232 3233 static inline void f2fs_update_iostat(struct f2fs_sb_info *sbi, 3234 enum iostat_type type, unsigned long long io_bytes) 3235 { 3236 if (!sbi->iostat_enable) 3237 return; 3238 spin_lock(&sbi->iostat_lock); 3239 sbi->rw_iostat[type] += io_bytes; 3240 3241 if (type == APP_WRITE_IO || type == APP_DIRECT_IO) 3242 sbi->rw_iostat[APP_BUFFERED_IO] = 3243 sbi->rw_iostat[APP_WRITE_IO] - 3244 sbi->rw_iostat[APP_DIRECT_IO]; 3245 3246 if (type == APP_READ_IO || type == APP_DIRECT_READ_IO) 3247 sbi->rw_iostat[APP_BUFFERED_READ_IO] = 3248 sbi->rw_iostat[APP_READ_IO] - 3249 sbi->rw_iostat[APP_DIRECT_READ_IO]; 3250 spin_unlock(&sbi->iostat_lock); 3251 3252 f2fs_record_iostat(sbi); 3253 } 3254 3255 #define __is_large_section(sbi) ((sbi)->segs_per_sec > 1) 3256 3257 #define __is_meta_io(fio) (PAGE_TYPE_OF_BIO((fio)->type) == META) 3258 3259 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi, 3260 block_t blkaddr, int type); 3261 static inline void verify_blkaddr(struct f2fs_sb_info *sbi, 3262 block_t blkaddr, int type) 3263 { 3264 if (!f2fs_is_valid_blkaddr(sbi, blkaddr, type)) { 3265 f2fs_err(sbi, "invalid blkaddr: %u, type: %d, run fsck to fix.", 3266 blkaddr, type); 3267 f2fs_bug_on(sbi, 1); 3268 } 3269 } 3270 3271 static inline bool __is_valid_data_blkaddr(block_t blkaddr) 3272 { 3273 if (blkaddr == NEW_ADDR || blkaddr == NULL_ADDR || 3274 blkaddr == COMPRESS_ADDR) 3275 return false; 3276 return true; 3277 } 3278 3279 /* 3280 * file.c 3281 */ 3282 int f2fs_sync_file(struct file *file, loff_t start, loff_t end, int datasync); 3283 void f2fs_truncate_data_blocks(struct dnode_of_data *dn); 3284 int f2fs_do_truncate_blocks(struct inode *inode, u64 from, bool lock); 3285 int f2fs_truncate_blocks(struct inode *inode, u64 from, bool lock); 3286 int f2fs_truncate(struct inode *inode); 3287 int f2fs_getattr(struct user_namespace *mnt_userns, const struct path *path, 3288 struct kstat *stat, u32 request_mask, unsigned int flags); 3289 int f2fs_setattr(struct user_namespace *mnt_userns, struct dentry *dentry, 3290 struct iattr *attr); 3291 int f2fs_truncate_hole(struct inode *inode, pgoff_t pg_start, pgoff_t pg_end); 3292 void f2fs_truncate_data_blocks_range(struct dnode_of_data *dn, int count); 3293 int f2fs_precache_extents(struct inode *inode); 3294 int f2fs_fileattr_get(struct dentry *dentry, struct fileattr *fa); 3295 int f2fs_fileattr_set(struct user_namespace *mnt_userns, 3296 struct dentry *dentry, struct fileattr *fa); 3297 long f2fs_ioctl(struct file *filp, unsigned int cmd, unsigned long arg); 3298 long f2fs_compat_ioctl(struct file *file, unsigned int cmd, unsigned long arg); 3299 int f2fs_transfer_project_quota(struct inode *inode, kprojid_t kprojid); 3300 int f2fs_pin_file_control(struct inode *inode, bool inc); 3301 3302 /* 3303 * inode.c 3304 */ 3305 void f2fs_set_inode_flags(struct inode *inode); 3306 bool f2fs_inode_chksum_verify(struct f2fs_sb_info *sbi, struct page *page); 3307 void f2fs_inode_chksum_set(struct f2fs_sb_info *sbi, struct page *page); 3308 struct inode *f2fs_iget(struct super_block *sb, unsigned long ino); 3309 struct inode *f2fs_iget_retry(struct super_block *sb, unsigned long ino); 3310 int f2fs_try_to_free_nats(struct f2fs_sb_info *sbi, int nr_shrink); 3311 void f2fs_update_inode(struct inode *inode, struct page *node_page); 3312 void f2fs_update_inode_page(struct inode *inode); 3313 int f2fs_write_inode(struct inode *inode, struct writeback_control *wbc); 3314 void f2fs_evict_inode(struct inode *inode); 3315 void f2fs_handle_failed_inode(struct inode *inode); 3316 3317 /* 3318 * namei.c 3319 */ 3320 int f2fs_update_extension_list(struct f2fs_sb_info *sbi, const char *name, 3321 bool hot, bool set); 3322 struct dentry *f2fs_get_parent(struct dentry *child); 3323 3324 /* 3325 * dir.c 3326 */ 3327 unsigned char f2fs_get_de_type(struct f2fs_dir_entry *de); 3328 int f2fs_init_casefolded_name(const struct inode *dir, 3329 struct f2fs_filename *fname); 3330 int f2fs_setup_filename(struct inode *dir, const struct qstr *iname, 3331 int lookup, struct f2fs_filename *fname); 3332 int f2fs_prepare_lookup(struct inode *dir, struct dentry *dentry, 3333 struct f2fs_filename *fname); 3334 void f2fs_free_filename(struct f2fs_filename *fname); 3335 struct f2fs_dir_entry *f2fs_find_target_dentry(const struct f2fs_dentry_ptr *d, 3336 const struct f2fs_filename *fname, int *max_slots); 3337 int f2fs_fill_dentries(struct dir_context *ctx, struct f2fs_dentry_ptr *d, 3338 unsigned int start_pos, struct fscrypt_str *fstr); 3339 void f2fs_do_make_empty_dir(struct inode *inode, struct inode *parent, 3340 struct f2fs_dentry_ptr *d); 3341 struct page *f2fs_init_inode_metadata(struct inode *inode, struct inode *dir, 3342 const struct f2fs_filename *fname, struct page *dpage); 3343 void f2fs_update_parent_metadata(struct inode *dir, struct inode *inode, 3344 unsigned int current_depth); 3345 int f2fs_room_for_filename(const void *bitmap, int slots, int max_slots); 3346 void f2fs_drop_nlink(struct inode *dir, struct inode *inode); 3347 struct f2fs_dir_entry *__f2fs_find_entry(struct inode *dir, 3348 const struct f2fs_filename *fname, 3349 struct page **res_page); 3350 struct f2fs_dir_entry *f2fs_find_entry(struct inode *dir, 3351 const struct qstr *child, struct page **res_page); 3352 struct f2fs_dir_entry *f2fs_parent_dir(struct inode *dir, struct page **p); 3353 ino_t f2fs_inode_by_name(struct inode *dir, const struct qstr *qstr, 3354 struct page **page); 3355 void f2fs_set_link(struct inode *dir, struct f2fs_dir_entry *de, 3356 struct page *page, struct inode *inode); 3357 bool f2fs_has_enough_room(struct inode *dir, struct page *ipage, 3358 const struct f2fs_filename *fname); 3359 void f2fs_update_dentry(nid_t ino, umode_t mode, struct f2fs_dentry_ptr *d, 3360 const struct fscrypt_str *name, f2fs_hash_t name_hash, 3361 unsigned int bit_pos); 3362 int f2fs_add_regular_entry(struct inode *dir, const struct f2fs_filename *fname, 3363 struct inode *inode, nid_t ino, umode_t mode); 3364 int f2fs_add_dentry(struct inode *dir, const struct f2fs_filename *fname, 3365 struct inode *inode, nid_t ino, umode_t mode); 3366 int f2fs_do_add_link(struct inode *dir, const struct qstr *name, 3367 struct inode *inode, nid_t ino, umode_t mode); 3368 void f2fs_delete_entry(struct f2fs_dir_entry *dentry, struct page *page, 3369 struct inode *dir, struct inode *inode); 3370 int f2fs_do_tmpfile(struct inode *inode, struct inode *dir); 3371 bool f2fs_empty_dir(struct inode *dir); 3372 3373 static inline int f2fs_add_link(struct dentry *dentry, struct inode *inode) 3374 { 3375 if (fscrypt_is_nokey_name(dentry)) 3376 return -ENOKEY; 3377 return f2fs_do_add_link(d_inode(dentry->d_parent), &dentry->d_name, 3378 inode, inode->i_ino, inode->i_mode); 3379 } 3380 3381 /* 3382 * super.c 3383 */ 3384 int f2fs_inode_dirtied(struct inode *inode, bool sync); 3385 void f2fs_inode_synced(struct inode *inode); 3386 int f2fs_enable_quota_files(struct f2fs_sb_info *sbi, bool rdonly); 3387 int f2fs_quota_sync(struct super_block *sb, int type); 3388 loff_t max_file_blocks(struct inode *inode); 3389 void f2fs_quota_off_umount(struct super_block *sb); 3390 int f2fs_commit_super(struct f2fs_sb_info *sbi, bool recover); 3391 int f2fs_sync_fs(struct super_block *sb, int sync); 3392 int f2fs_sanity_check_ckpt(struct f2fs_sb_info *sbi); 3393 3394 /* 3395 * hash.c 3396 */ 3397 void f2fs_hash_filename(const struct inode *dir, struct f2fs_filename *fname); 3398 3399 /* 3400 * node.c 3401 */ 3402 struct node_info; 3403 3404 int f2fs_check_nid_range(struct f2fs_sb_info *sbi, nid_t nid); 3405 bool f2fs_available_free_memory(struct f2fs_sb_info *sbi, int type); 3406 bool f2fs_in_warm_node_list(struct f2fs_sb_info *sbi, struct page *page); 3407 void f2fs_init_fsync_node_info(struct f2fs_sb_info *sbi); 3408 void f2fs_del_fsync_node_entry(struct f2fs_sb_info *sbi, struct page *page); 3409 void f2fs_reset_fsync_node_info(struct f2fs_sb_info *sbi); 3410 int f2fs_need_dentry_mark(struct f2fs_sb_info *sbi, nid_t nid); 3411 bool f2fs_is_checkpointed_node(struct f2fs_sb_info *sbi, nid_t nid); 3412 bool f2fs_need_inode_block_update(struct f2fs_sb_info *sbi, nid_t ino); 3413 int f2fs_get_node_info(struct f2fs_sb_info *sbi, nid_t nid, 3414 struct node_info *ni); 3415 pgoff_t f2fs_get_next_page_offset(struct dnode_of_data *dn, pgoff_t pgofs); 3416 int f2fs_get_dnode_of_data(struct dnode_of_data *dn, pgoff_t index, int mode); 3417 int f2fs_truncate_inode_blocks(struct inode *inode, pgoff_t from); 3418 int f2fs_truncate_xattr_node(struct inode *inode); 3419 int f2fs_wait_on_node_pages_writeback(struct f2fs_sb_info *sbi, 3420 unsigned int seq_id); 3421 int f2fs_remove_inode_page(struct inode *inode); 3422 struct page *f2fs_new_inode_page(struct inode *inode); 3423 struct page *f2fs_new_node_page(struct dnode_of_data *dn, unsigned int ofs); 3424 void f2fs_ra_node_page(struct f2fs_sb_info *sbi, nid_t nid); 3425 struct page *f2fs_get_node_page(struct f2fs_sb_info *sbi, pgoff_t nid); 3426 struct page *f2fs_get_node_page_ra(struct page *parent, int start); 3427 int f2fs_move_node_page(struct page *node_page, int gc_type); 3428 void f2fs_flush_inline_data(struct f2fs_sb_info *sbi); 3429 int f2fs_fsync_node_pages(struct f2fs_sb_info *sbi, struct inode *inode, 3430 struct writeback_control *wbc, bool atomic, 3431 unsigned int *seq_id); 3432 int f2fs_sync_node_pages(struct f2fs_sb_info *sbi, 3433 struct writeback_control *wbc, 3434 bool do_balance, enum iostat_type io_type); 3435 int f2fs_build_free_nids(struct f2fs_sb_info *sbi, bool sync, bool mount); 3436 bool f2fs_alloc_nid(struct f2fs_sb_info *sbi, nid_t *nid); 3437 void f2fs_alloc_nid_done(struct f2fs_sb_info *sbi, nid_t nid); 3438 void f2fs_alloc_nid_failed(struct f2fs_sb_info *sbi, nid_t nid); 3439 int f2fs_try_to_free_nids(struct f2fs_sb_info *sbi, int nr_shrink); 3440 int f2fs_recover_inline_xattr(struct inode *inode, struct page *page); 3441 int f2fs_recover_xattr_data(struct inode *inode, struct page *page); 3442 int f2fs_recover_inode_page(struct f2fs_sb_info *sbi, struct page *page); 3443 int f2fs_restore_node_summary(struct f2fs_sb_info *sbi, 3444 unsigned int segno, struct f2fs_summary_block *sum); 3445 int f2fs_flush_nat_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3446 int f2fs_build_node_manager(struct f2fs_sb_info *sbi); 3447 void f2fs_destroy_node_manager(struct f2fs_sb_info *sbi); 3448 int __init f2fs_create_node_manager_caches(void); 3449 void f2fs_destroy_node_manager_caches(void); 3450 3451 /* 3452 * segment.c 3453 */ 3454 bool f2fs_need_SSR(struct f2fs_sb_info *sbi); 3455 void f2fs_register_inmem_page(struct inode *inode, struct page *page); 3456 void f2fs_drop_inmem_pages_all(struct f2fs_sb_info *sbi, bool gc_failure); 3457 void f2fs_drop_inmem_pages(struct inode *inode); 3458 void f2fs_drop_inmem_page(struct inode *inode, struct page *page); 3459 int f2fs_commit_inmem_pages(struct inode *inode); 3460 void f2fs_balance_fs(struct f2fs_sb_info *sbi, bool need); 3461 void f2fs_balance_fs_bg(struct f2fs_sb_info *sbi, bool from_bg); 3462 int f2fs_issue_flush(struct f2fs_sb_info *sbi, nid_t ino); 3463 int f2fs_create_flush_cmd_control(struct f2fs_sb_info *sbi); 3464 int f2fs_flush_device_cache(struct f2fs_sb_info *sbi); 3465 void f2fs_destroy_flush_cmd_control(struct f2fs_sb_info *sbi, bool free); 3466 void f2fs_invalidate_blocks(struct f2fs_sb_info *sbi, block_t addr); 3467 bool f2fs_is_checkpointed_data(struct f2fs_sb_info *sbi, block_t blkaddr); 3468 void f2fs_drop_discard_cmd(struct f2fs_sb_info *sbi); 3469 void f2fs_stop_discard_thread(struct f2fs_sb_info *sbi); 3470 bool f2fs_issue_discard_timeout(struct f2fs_sb_info *sbi); 3471 void f2fs_clear_prefree_segments(struct f2fs_sb_info *sbi, 3472 struct cp_control *cpc); 3473 void f2fs_dirty_to_prefree(struct f2fs_sb_info *sbi); 3474 block_t f2fs_get_unusable_blocks(struct f2fs_sb_info *sbi); 3475 int f2fs_disable_cp_again(struct f2fs_sb_info *sbi, block_t unusable); 3476 void f2fs_release_discard_addrs(struct f2fs_sb_info *sbi); 3477 int f2fs_npages_for_summary_flush(struct f2fs_sb_info *sbi, bool for_ra); 3478 bool f2fs_segment_has_free_slot(struct f2fs_sb_info *sbi, int segno); 3479 void f2fs_init_inmem_curseg(struct f2fs_sb_info *sbi); 3480 void f2fs_save_inmem_curseg(struct f2fs_sb_info *sbi); 3481 void f2fs_restore_inmem_curseg(struct f2fs_sb_info *sbi); 3482 void f2fs_get_new_segment(struct f2fs_sb_info *sbi, 3483 unsigned int *newseg, bool new_sec, int dir); 3484 void f2fs_allocate_segment_for_resize(struct f2fs_sb_info *sbi, int type, 3485 unsigned int start, unsigned int end); 3486 void f2fs_allocate_new_section(struct f2fs_sb_info *sbi, int type, bool force); 3487 void f2fs_allocate_new_segments(struct f2fs_sb_info *sbi); 3488 int f2fs_trim_fs(struct f2fs_sb_info *sbi, struct fstrim_range *range); 3489 bool f2fs_exist_trim_candidates(struct f2fs_sb_info *sbi, 3490 struct cp_control *cpc); 3491 struct page *f2fs_get_sum_page(struct f2fs_sb_info *sbi, unsigned int segno); 3492 void f2fs_update_meta_page(struct f2fs_sb_info *sbi, void *src, 3493 block_t blk_addr); 3494 void f2fs_do_write_meta_page(struct f2fs_sb_info *sbi, struct page *page, 3495 enum iostat_type io_type); 3496 void f2fs_do_write_node_page(unsigned int nid, struct f2fs_io_info *fio); 3497 void f2fs_outplace_write_data(struct dnode_of_data *dn, 3498 struct f2fs_io_info *fio); 3499 int f2fs_inplace_write_data(struct f2fs_io_info *fio); 3500 void f2fs_do_replace_block(struct f2fs_sb_info *sbi, struct f2fs_summary *sum, 3501 block_t old_blkaddr, block_t new_blkaddr, 3502 bool recover_curseg, bool recover_newaddr, 3503 bool from_gc); 3504 void f2fs_replace_block(struct f2fs_sb_info *sbi, struct dnode_of_data *dn, 3505 block_t old_addr, block_t new_addr, 3506 unsigned char version, bool recover_curseg, 3507 bool recover_newaddr); 3508 void f2fs_allocate_data_block(struct f2fs_sb_info *sbi, struct page *page, 3509 block_t old_blkaddr, block_t *new_blkaddr, 3510 struct f2fs_summary *sum, int type, 3511 struct f2fs_io_info *fio); 3512 void f2fs_wait_on_page_writeback(struct page *page, 3513 enum page_type type, bool ordered, bool locked); 3514 void f2fs_wait_on_block_writeback(struct inode *inode, block_t blkaddr); 3515 void f2fs_wait_on_block_writeback_range(struct inode *inode, block_t blkaddr, 3516 block_t len); 3517 void f2fs_write_data_summaries(struct f2fs_sb_info *sbi, block_t start_blk); 3518 void f2fs_write_node_summaries(struct f2fs_sb_info *sbi, block_t start_blk); 3519 int f2fs_lookup_journal_in_cursum(struct f2fs_journal *journal, int type, 3520 unsigned int val, int alloc); 3521 void f2fs_flush_sit_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3522 int f2fs_fix_curseg_write_pointer(struct f2fs_sb_info *sbi); 3523 int f2fs_check_write_pointer(struct f2fs_sb_info *sbi); 3524 int f2fs_build_segment_manager(struct f2fs_sb_info *sbi); 3525 void f2fs_destroy_segment_manager(struct f2fs_sb_info *sbi); 3526 int __init f2fs_create_segment_manager_caches(void); 3527 void f2fs_destroy_segment_manager_caches(void); 3528 int f2fs_rw_hint_to_seg_type(enum rw_hint hint); 3529 enum rw_hint f2fs_io_type_to_rw_hint(struct f2fs_sb_info *sbi, 3530 enum page_type type, enum temp_type temp); 3531 unsigned int f2fs_usable_segs_in_sec(struct f2fs_sb_info *sbi, 3532 unsigned int segno); 3533 unsigned int f2fs_usable_blks_in_seg(struct f2fs_sb_info *sbi, 3534 unsigned int segno); 3535 3536 /* 3537 * checkpoint.c 3538 */ 3539 void f2fs_stop_checkpoint(struct f2fs_sb_info *sbi, bool end_io); 3540 struct page *f2fs_grab_meta_page(struct f2fs_sb_info *sbi, pgoff_t index); 3541 struct page *f2fs_get_meta_page(struct f2fs_sb_info *sbi, pgoff_t index); 3542 struct page *f2fs_get_meta_page_retry(struct f2fs_sb_info *sbi, pgoff_t index); 3543 struct page *f2fs_get_tmp_page(struct f2fs_sb_info *sbi, pgoff_t index); 3544 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi, 3545 block_t blkaddr, int type); 3546 int f2fs_ra_meta_pages(struct f2fs_sb_info *sbi, block_t start, int nrpages, 3547 int type, bool sync); 3548 void f2fs_ra_meta_pages_cond(struct f2fs_sb_info *sbi, pgoff_t index); 3549 long f2fs_sync_meta_pages(struct f2fs_sb_info *sbi, enum page_type type, 3550 long nr_to_write, enum iostat_type io_type); 3551 void f2fs_add_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type); 3552 void f2fs_remove_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type); 3553 void f2fs_release_ino_entry(struct f2fs_sb_info *sbi, bool all); 3554 bool f2fs_exist_written_data(struct f2fs_sb_info *sbi, nid_t ino, int mode); 3555 void f2fs_set_dirty_device(struct f2fs_sb_info *sbi, nid_t ino, 3556 unsigned int devidx, int type); 3557 bool f2fs_is_dirty_device(struct f2fs_sb_info *sbi, nid_t ino, 3558 unsigned int devidx, int type); 3559 int f2fs_sync_inode_meta(struct f2fs_sb_info *sbi); 3560 int f2fs_acquire_orphan_inode(struct f2fs_sb_info *sbi); 3561 void f2fs_release_orphan_inode(struct f2fs_sb_info *sbi); 3562 void f2fs_add_orphan_inode(struct inode *inode); 3563 void f2fs_remove_orphan_inode(struct f2fs_sb_info *sbi, nid_t ino); 3564 int f2fs_recover_orphan_inodes(struct f2fs_sb_info *sbi); 3565 int f2fs_get_valid_checkpoint(struct f2fs_sb_info *sbi); 3566 void f2fs_update_dirty_page(struct inode *inode, struct page *page); 3567 void f2fs_remove_dirty_inode(struct inode *inode); 3568 int f2fs_sync_dirty_inodes(struct f2fs_sb_info *sbi, enum inode_type type); 3569 void f2fs_wait_on_all_pages(struct f2fs_sb_info *sbi, int type); 3570 u64 f2fs_get_sectors_written(struct f2fs_sb_info *sbi); 3571 int f2fs_write_checkpoint(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3572 void f2fs_init_ino_entry_info(struct f2fs_sb_info *sbi); 3573 int __init f2fs_create_checkpoint_caches(void); 3574 void f2fs_destroy_checkpoint_caches(void); 3575 int f2fs_issue_checkpoint(struct f2fs_sb_info *sbi); 3576 int f2fs_start_ckpt_thread(struct f2fs_sb_info *sbi); 3577 void f2fs_stop_ckpt_thread(struct f2fs_sb_info *sbi); 3578 void f2fs_init_ckpt_req_control(struct f2fs_sb_info *sbi); 3579 3580 /* 3581 * data.c 3582 */ 3583 int __init f2fs_init_bioset(void); 3584 void f2fs_destroy_bioset(void); 3585 int f2fs_init_bio_entry_cache(void); 3586 void f2fs_destroy_bio_entry_cache(void); 3587 void f2fs_submit_bio(struct f2fs_sb_info *sbi, 3588 struct bio *bio, enum page_type type); 3589 void f2fs_submit_merged_write(struct f2fs_sb_info *sbi, enum page_type type); 3590 void f2fs_submit_merged_write_cond(struct f2fs_sb_info *sbi, 3591 struct inode *inode, struct page *page, 3592 nid_t ino, enum page_type type); 3593 void f2fs_submit_merged_ipu_write(struct f2fs_sb_info *sbi, 3594 struct bio **bio, struct page *page); 3595 void f2fs_flush_merged_writes(struct f2fs_sb_info *sbi); 3596 int f2fs_submit_page_bio(struct f2fs_io_info *fio); 3597 int f2fs_merge_page_bio(struct f2fs_io_info *fio); 3598 void f2fs_submit_page_write(struct f2fs_io_info *fio); 3599 struct block_device *f2fs_target_device(struct f2fs_sb_info *sbi, 3600 block_t blk_addr, struct bio *bio); 3601 int f2fs_target_device_index(struct f2fs_sb_info *sbi, block_t blkaddr); 3602 void f2fs_set_data_blkaddr(struct dnode_of_data *dn); 3603 void f2fs_update_data_blkaddr(struct dnode_of_data *dn, block_t blkaddr); 3604 int f2fs_reserve_new_blocks(struct dnode_of_data *dn, blkcnt_t count); 3605 int f2fs_reserve_new_block(struct dnode_of_data *dn); 3606 int f2fs_get_block(struct dnode_of_data *dn, pgoff_t index); 3607 int f2fs_preallocate_blocks(struct kiocb *iocb, struct iov_iter *from); 3608 int f2fs_reserve_block(struct dnode_of_data *dn, pgoff_t index); 3609 struct page *f2fs_get_read_data_page(struct inode *inode, pgoff_t index, 3610 int op_flags, bool for_write); 3611 struct page *f2fs_find_data_page(struct inode *inode, pgoff_t index); 3612 struct page *f2fs_get_lock_data_page(struct inode *inode, pgoff_t index, 3613 bool for_write); 3614 struct page *f2fs_get_new_data_page(struct inode *inode, 3615 struct page *ipage, pgoff_t index, bool new_i_size); 3616 int f2fs_do_write_data_page(struct f2fs_io_info *fio); 3617 void f2fs_do_map_lock(struct f2fs_sb_info *sbi, int flag, bool lock); 3618 int f2fs_map_blocks(struct inode *inode, struct f2fs_map_blocks *map, 3619 int create, int flag); 3620 int f2fs_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, 3621 u64 start, u64 len); 3622 int f2fs_encrypt_one_page(struct f2fs_io_info *fio); 3623 bool f2fs_should_update_inplace(struct inode *inode, struct f2fs_io_info *fio); 3624 bool f2fs_should_update_outplace(struct inode *inode, struct f2fs_io_info *fio); 3625 int f2fs_write_single_data_page(struct page *page, int *submitted, 3626 struct bio **bio, sector_t *last_block, 3627 struct writeback_control *wbc, 3628 enum iostat_type io_type, 3629 int compr_blocks, bool allow_balance); 3630 void f2fs_invalidate_page(struct page *page, unsigned int offset, 3631 unsigned int length); 3632 int f2fs_release_page(struct page *page, gfp_t wait); 3633 #ifdef CONFIG_MIGRATION 3634 int f2fs_migrate_page(struct address_space *mapping, struct page *newpage, 3635 struct page *page, enum migrate_mode mode); 3636 #endif 3637 bool f2fs_overwrite_io(struct inode *inode, loff_t pos, size_t len); 3638 void f2fs_clear_page_cache_dirty_tag(struct page *page); 3639 int f2fs_init_post_read_processing(void); 3640 void f2fs_destroy_post_read_processing(void); 3641 int f2fs_init_post_read_wq(struct f2fs_sb_info *sbi); 3642 void f2fs_destroy_post_read_wq(struct f2fs_sb_info *sbi); 3643 3644 /* 3645 * gc.c 3646 */ 3647 int f2fs_start_gc_thread(struct f2fs_sb_info *sbi); 3648 void f2fs_stop_gc_thread(struct f2fs_sb_info *sbi); 3649 block_t f2fs_start_bidx_of_node(unsigned int node_ofs, struct inode *inode); 3650 int f2fs_gc(struct f2fs_sb_info *sbi, bool sync, bool background, bool force, 3651 unsigned int segno); 3652 void f2fs_build_gc_manager(struct f2fs_sb_info *sbi); 3653 int f2fs_resize_fs(struct f2fs_sb_info *sbi, __u64 block_count); 3654 int __init f2fs_create_garbage_collection_cache(void); 3655 void f2fs_destroy_garbage_collection_cache(void); 3656 3657 /* 3658 * recovery.c 3659 */ 3660 int f2fs_recover_fsync_data(struct f2fs_sb_info *sbi, bool check_only); 3661 bool f2fs_space_for_roll_forward(struct f2fs_sb_info *sbi); 3662 int __init f2fs_create_recovery_cache(void); 3663 void f2fs_destroy_recovery_cache(void); 3664 3665 /* 3666 * debug.c 3667 */ 3668 #ifdef CONFIG_F2FS_STAT_FS 3669 struct f2fs_stat_info { 3670 struct list_head stat_list; 3671 struct f2fs_sb_info *sbi; 3672 int all_area_segs, sit_area_segs, nat_area_segs, ssa_area_segs; 3673 int main_area_segs, main_area_sections, main_area_zones; 3674 unsigned long long hit_largest, hit_cached, hit_rbtree; 3675 unsigned long long hit_total, total_ext; 3676 int ext_tree, zombie_tree, ext_node; 3677 int ndirty_node, ndirty_dent, ndirty_meta, ndirty_imeta; 3678 int ndirty_data, ndirty_qdata; 3679 int inmem_pages; 3680 unsigned int ndirty_dirs, ndirty_files, nquota_files, ndirty_all; 3681 int nats, dirty_nats, sits, dirty_sits; 3682 int free_nids, avail_nids, alloc_nids; 3683 int total_count, utilization; 3684 int bg_gc, nr_wb_cp_data, nr_wb_data; 3685 int nr_rd_data, nr_rd_node, nr_rd_meta; 3686 int nr_dio_read, nr_dio_write; 3687 unsigned int io_skip_bggc, other_skip_bggc; 3688 int nr_flushing, nr_flushed, flush_list_empty; 3689 int nr_discarding, nr_discarded; 3690 int nr_discard_cmd; 3691 unsigned int undiscard_blks; 3692 int nr_issued_ckpt, nr_total_ckpt, nr_queued_ckpt; 3693 unsigned int cur_ckpt_time, peak_ckpt_time; 3694 int inline_xattr, inline_inode, inline_dir, append, update, orphans; 3695 int compr_inode; 3696 unsigned long long compr_blocks; 3697 int aw_cnt, max_aw_cnt, vw_cnt, max_vw_cnt; 3698 unsigned int valid_count, valid_node_count, valid_inode_count, discard_blks; 3699 unsigned int bimodal, avg_vblocks; 3700 int util_free, util_valid, util_invalid; 3701 int rsvd_segs, overp_segs; 3702 int dirty_count, node_pages, meta_pages, compress_pages; 3703 int compress_page_hit; 3704 int prefree_count, call_count, cp_count, bg_cp_count; 3705 int tot_segs, node_segs, data_segs, free_segs, free_secs; 3706 int bg_node_segs, bg_data_segs; 3707 int tot_blks, data_blks, node_blks; 3708 int bg_data_blks, bg_node_blks; 3709 unsigned long long skipped_atomic_files[2]; 3710 int curseg[NR_CURSEG_TYPE]; 3711 int cursec[NR_CURSEG_TYPE]; 3712 int curzone[NR_CURSEG_TYPE]; 3713 unsigned int dirty_seg[NR_CURSEG_TYPE]; 3714 unsigned int full_seg[NR_CURSEG_TYPE]; 3715 unsigned int valid_blks[NR_CURSEG_TYPE]; 3716 3717 unsigned int meta_count[META_MAX]; 3718 unsigned int segment_count[2]; 3719 unsigned int block_count[2]; 3720 unsigned int inplace_count; 3721 unsigned long long base_mem, cache_mem, page_mem; 3722 }; 3723 3724 static inline struct f2fs_stat_info *F2FS_STAT(struct f2fs_sb_info *sbi) 3725 { 3726 return (struct f2fs_stat_info *)sbi->stat_info; 3727 } 3728 3729 #define stat_inc_cp_count(si) ((si)->cp_count++) 3730 #define stat_inc_bg_cp_count(si) ((si)->bg_cp_count++) 3731 #define stat_inc_call_count(si) ((si)->call_count++) 3732 #define stat_inc_bggc_count(si) ((si)->bg_gc++) 3733 #define stat_io_skip_bggc_count(sbi) ((sbi)->io_skip_bggc++) 3734 #define stat_other_skip_bggc_count(sbi) ((sbi)->other_skip_bggc++) 3735 #define stat_inc_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]++) 3736 #define stat_dec_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]--) 3737 #define stat_inc_total_hit(sbi) (atomic64_inc(&(sbi)->total_hit_ext)) 3738 #define stat_inc_rbtree_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_rbtree)) 3739 #define stat_inc_largest_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_largest)) 3740 #define stat_inc_cached_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_cached)) 3741 #define stat_inc_inline_xattr(inode) \ 3742 do { \ 3743 if (f2fs_has_inline_xattr(inode)) \ 3744 (atomic_inc(&F2FS_I_SB(inode)->inline_xattr)); \ 3745 } while (0) 3746 #define stat_dec_inline_xattr(inode) \ 3747 do { \ 3748 if (f2fs_has_inline_xattr(inode)) \ 3749 (atomic_dec(&F2FS_I_SB(inode)->inline_xattr)); \ 3750 } while (0) 3751 #define stat_inc_inline_inode(inode) \ 3752 do { \ 3753 if (f2fs_has_inline_data(inode)) \ 3754 (atomic_inc(&F2FS_I_SB(inode)->inline_inode)); \ 3755 } while (0) 3756 #define stat_dec_inline_inode(inode) \ 3757 do { \ 3758 if (f2fs_has_inline_data(inode)) \ 3759 (atomic_dec(&F2FS_I_SB(inode)->inline_inode)); \ 3760 } while (0) 3761 #define stat_inc_inline_dir(inode) \ 3762 do { \ 3763 if (f2fs_has_inline_dentry(inode)) \ 3764 (atomic_inc(&F2FS_I_SB(inode)->inline_dir)); \ 3765 } while (0) 3766 #define stat_dec_inline_dir(inode) \ 3767 do { \ 3768 if (f2fs_has_inline_dentry(inode)) \ 3769 (atomic_dec(&F2FS_I_SB(inode)->inline_dir)); \ 3770 } while (0) 3771 #define stat_inc_compr_inode(inode) \ 3772 do { \ 3773 if (f2fs_compressed_file(inode)) \ 3774 (atomic_inc(&F2FS_I_SB(inode)->compr_inode)); \ 3775 } while (0) 3776 #define stat_dec_compr_inode(inode) \ 3777 do { \ 3778 if (f2fs_compressed_file(inode)) \ 3779 (atomic_dec(&F2FS_I_SB(inode)->compr_inode)); \ 3780 } while (0) 3781 #define stat_add_compr_blocks(inode, blocks) \ 3782 (atomic64_add(blocks, &F2FS_I_SB(inode)->compr_blocks)) 3783 #define stat_sub_compr_blocks(inode, blocks) \ 3784 (atomic64_sub(blocks, &F2FS_I_SB(inode)->compr_blocks)) 3785 #define stat_inc_meta_count(sbi, blkaddr) \ 3786 do { \ 3787 if (blkaddr < SIT_I(sbi)->sit_base_addr) \ 3788 atomic_inc(&(sbi)->meta_count[META_CP]); \ 3789 else if (blkaddr < NM_I(sbi)->nat_blkaddr) \ 3790 atomic_inc(&(sbi)->meta_count[META_SIT]); \ 3791 else if (blkaddr < SM_I(sbi)->ssa_blkaddr) \ 3792 atomic_inc(&(sbi)->meta_count[META_NAT]); \ 3793 else if (blkaddr < SM_I(sbi)->main_blkaddr) \ 3794 atomic_inc(&(sbi)->meta_count[META_SSA]); \ 3795 } while (0) 3796 #define stat_inc_seg_type(sbi, curseg) \ 3797 ((sbi)->segment_count[(curseg)->alloc_type]++) 3798 #define stat_inc_block_count(sbi, curseg) \ 3799 ((sbi)->block_count[(curseg)->alloc_type]++) 3800 #define stat_inc_inplace_blocks(sbi) \ 3801 (atomic_inc(&(sbi)->inplace_count)) 3802 #define stat_update_max_atomic_write(inode) \ 3803 do { \ 3804 int cur = F2FS_I_SB(inode)->atomic_files; \ 3805 int max = atomic_read(&F2FS_I_SB(inode)->max_aw_cnt); \ 3806 if (cur > max) \ 3807 atomic_set(&F2FS_I_SB(inode)->max_aw_cnt, cur); \ 3808 } while (0) 3809 #define stat_inc_volatile_write(inode) \ 3810 (atomic_inc(&F2FS_I_SB(inode)->vw_cnt)) 3811 #define stat_dec_volatile_write(inode) \ 3812 (atomic_dec(&F2FS_I_SB(inode)->vw_cnt)) 3813 #define stat_update_max_volatile_write(inode) \ 3814 do { \ 3815 int cur = atomic_read(&F2FS_I_SB(inode)->vw_cnt); \ 3816 int max = atomic_read(&F2FS_I_SB(inode)->max_vw_cnt); \ 3817 if (cur > max) \ 3818 atomic_set(&F2FS_I_SB(inode)->max_vw_cnt, cur); \ 3819 } while (0) 3820 #define stat_inc_seg_count(sbi, type, gc_type) \ 3821 do { \ 3822 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 3823 si->tot_segs++; \ 3824 if ((type) == SUM_TYPE_DATA) { \ 3825 si->data_segs++; \ 3826 si->bg_data_segs += (gc_type == BG_GC) ? 1 : 0; \ 3827 } else { \ 3828 si->node_segs++; \ 3829 si->bg_node_segs += (gc_type == BG_GC) ? 1 : 0; \ 3830 } \ 3831 } while (0) 3832 3833 #define stat_inc_tot_blk_count(si, blks) \ 3834 ((si)->tot_blks += (blks)) 3835 3836 #define stat_inc_data_blk_count(sbi, blks, gc_type) \ 3837 do { \ 3838 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 3839 stat_inc_tot_blk_count(si, blks); \ 3840 si->data_blks += (blks); \ 3841 si->bg_data_blks += ((gc_type) == BG_GC) ? (blks) : 0; \ 3842 } while (0) 3843 3844 #define stat_inc_node_blk_count(sbi, blks, gc_type) \ 3845 do { \ 3846 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 3847 stat_inc_tot_blk_count(si, blks); \ 3848 si->node_blks += (blks); \ 3849 si->bg_node_blks += ((gc_type) == BG_GC) ? (blks) : 0; \ 3850 } while (0) 3851 3852 int f2fs_build_stats(struct f2fs_sb_info *sbi); 3853 void f2fs_destroy_stats(struct f2fs_sb_info *sbi); 3854 void __init f2fs_create_root_stats(void); 3855 void f2fs_destroy_root_stats(void); 3856 void f2fs_update_sit_info(struct f2fs_sb_info *sbi); 3857 #else 3858 #define stat_inc_cp_count(si) do { } while (0) 3859 #define stat_inc_bg_cp_count(si) do { } while (0) 3860 #define stat_inc_call_count(si) do { } while (0) 3861 #define stat_inc_bggc_count(si) do { } while (0) 3862 #define stat_io_skip_bggc_count(sbi) do { } while (0) 3863 #define stat_other_skip_bggc_count(sbi) do { } while (0) 3864 #define stat_inc_dirty_inode(sbi, type) do { } while (0) 3865 #define stat_dec_dirty_inode(sbi, type) do { } while (0) 3866 #define stat_inc_total_hit(sbi) do { } while (0) 3867 #define stat_inc_rbtree_node_hit(sbi) do { } while (0) 3868 #define stat_inc_largest_node_hit(sbi) do { } while (0) 3869 #define stat_inc_cached_node_hit(sbi) do { } while (0) 3870 #define stat_inc_inline_xattr(inode) do { } while (0) 3871 #define stat_dec_inline_xattr(inode) do { } while (0) 3872 #define stat_inc_inline_inode(inode) do { } while (0) 3873 #define stat_dec_inline_inode(inode) do { } while (0) 3874 #define stat_inc_inline_dir(inode) do { } while (0) 3875 #define stat_dec_inline_dir(inode) do { } while (0) 3876 #define stat_inc_compr_inode(inode) do { } while (0) 3877 #define stat_dec_compr_inode(inode) do { } while (0) 3878 #define stat_add_compr_blocks(inode, blocks) do { } while (0) 3879 #define stat_sub_compr_blocks(inode, blocks) do { } while (0) 3880 #define stat_update_max_atomic_write(inode) do { } while (0) 3881 #define stat_inc_volatile_write(inode) do { } while (0) 3882 #define stat_dec_volatile_write(inode) do { } while (0) 3883 #define stat_update_max_volatile_write(inode) do { } while (0) 3884 #define stat_inc_meta_count(sbi, blkaddr) do { } while (0) 3885 #define stat_inc_seg_type(sbi, curseg) do { } while (0) 3886 #define stat_inc_block_count(sbi, curseg) do { } while (0) 3887 #define stat_inc_inplace_blocks(sbi) do { } while (0) 3888 #define stat_inc_seg_count(sbi, type, gc_type) do { } while (0) 3889 #define stat_inc_tot_blk_count(si, blks) do { } while (0) 3890 #define stat_inc_data_blk_count(sbi, blks, gc_type) do { } while (0) 3891 #define stat_inc_node_blk_count(sbi, blks, gc_type) do { } while (0) 3892 3893 static inline int f2fs_build_stats(struct f2fs_sb_info *sbi) { return 0; } 3894 static inline void f2fs_destroy_stats(struct f2fs_sb_info *sbi) { } 3895 static inline void __init f2fs_create_root_stats(void) { } 3896 static inline void f2fs_destroy_root_stats(void) { } 3897 static inline void f2fs_update_sit_info(struct f2fs_sb_info *sbi) {} 3898 #endif 3899 3900 extern const struct file_operations f2fs_dir_operations; 3901 extern const struct file_operations f2fs_file_operations; 3902 extern const struct inode_operations f2fs_file_inode_operations; 3903 extern const struct address_space_operations f2fs_dblock_aops; 3904 extern const struct address_space_operations f2fs_node_aops; 3905 extern const struct address_space_operations f2fs_meta_aops; 3906 extern const struct inode_operations f2fs_dir_inode_operations; 3907 extern const struct inode_operations f2fs_symlink_inode_operations; 3908 extern const struct inode_operations f2fs_encrypted_symlink_inode_operations; 3909 extern const struct inode_operations f2fs_special_inode_operations; 3910 extern struct kmem_cache *f2fs_inode_entry_slab; 3911 3912 /* 3913 * inline.c 3914 */ 3915 bool f2fs_may_inline_data(struct inode *inode); 3916 bool f2fs_may_inline_dentry(struct inode *inode); 3917 void f2fs_do_read_inline_data(struct page *page, struct page *ipage); 3918 void f2fs_truncate_inline_inode(struct inode *inode, 3919 struct page *ipage, u64 from); 3920 int f2fs_read_inline_data(struct inode *inode, struct page *page); 3921 int f2fs_convert_inline_page(struct dnode_of_data *dn, struct page *page); 3922 int f2fs_convert_inline_inode(struct inode *inode); 3923 int f2fs_try_convert_inline_dir(struct inode *dir, struct dentry *dentry); 3924 int f2fs_write_inline_data(struct inode *inode, struct page *page); 3925 int f2fs_recover_inline_data(struct inode *inode, struct page *npage); 3926 struct f2fs_dir_entry *f2fs_find_in_inline_dir(struct inode *dir, 3927 const struct f2fs_filename *fname, 3928 struct page **res_page); 3929 int f2fs_make_empty_inline_dir(struct inode *inode, struct inode *parent, 3930 struct page *ipage); 3931 int f2fs_add_inline_entry(struct inode *dir, const struct f2fs_filename *fname, 3932 struct inode *inode, nid_t ino, umode_t mode); 3933 void f2fs_delete_inline_entry(struct f2fs_dir_entry *dentry, 3934 struct page *page, struct inode *dir, 3935 struct inode *inode); 3936 bool f2fs_empty_inline_dir(struct inode *dir); 3937 int f2fs_read_inline_dir(struct file *file, struct dir_context *ctx, 3938 struct fscrypt_str *fstr); 3939 int f2fs_inline_data_fiemap(struct inode *inode, 3940 struct fiemap_extent_info *fieinfo, 3941 __u64 start, __u64 len); 3942 3943 /* 3944 * shrinker.c 3945 */ 3946 unsigned long f2fs_shrink_count(struct shrinker *shrink, 3947 struct shrink_control *sc); 3948 unsigned long f2fs_shrink_scan(struct shrinker *shrink, 3949 struct shrink_control *sc); 3950 void f2fs_join_shrinker(struct f2fs_sb_info *sbi); 3951 void f2fs_leave_shrinker(struct f2fs_sb_info *sbi); 3952 3953 /* 3954 * extent_cache.c 3955 */ 3956 struct rb_entry *f2fs_lookup_rb_tree(struct rb_root_cached *root, 3957 struct rb_entry *cached_re, unsigned int ofs); 3958 struct rb_node **f2fs_lookup_rb_tree_ext(struct f2fs_sb_info *sbi, 3959 struct rb_root_cached *root, 3960 struct rb_node **parent, 3961 unsigned long long key, bool *left_most); 3962 struct rb_node **f2fs_lookup_rb_tree_for_insert(struct f2fs_sb_info *sbi, 3963 struct rb_root_cached *root, 3964 struct rb_node **parent, 3965 unsigned int ofs, bool *leftmost); 3966 struct rb_entry *f2fs_lookup_rb_tree_ret(struct rb_root_cached *root, 3967 struct rb_entry *cached_re, unsigned int ofs, 3968 struct rb_entry **prev_entry, struct rb_entry **next_entry, 3969 struct rb_node ***insert_p, struct rb_node **insert_parent, 3970 bool force, bool *leftmost); 3971 bool f2fs_check_rb_tree_consistence(struct f2fs_sb_info *sbi, 3972 struct rb_root_cached *root, bool check_key); 3973 unsigned int f2fs_shrink_extent_tree(struct f2fs_sb_info *sbi, int nr_shrink); 3974 void f2fs_init_extent_tree(struct inode *inode, struct page *ipage); 3975 void f2fs_drop_extent_tree(struct inode *inode); 3976 unsigned int f2fs_destroy_extent_node(struct inode *inode); 3977 void f2fs_destroy_extent_tree(struct inode *inode); 3978 bool f2fs_lookup_extent_cache(struct inode *inode, pgoff_t pgofs, 3979 struct extent_info *ei); 3980 void f2fs_update_extent_cache(struct dnode_of_data *dn); 3981 void f2fs_update_extent_cache_range(struct dnode_of_data *dn, 3982 pgoff_t fofs, block_t blkaddr, unsigned int len); 3983 void f2fs_init_extent_cache_info(struct f2fs_sb_info *sbi); 3984 int __init f2fs_create_extent_cache(void); 3985 void f2fs_destroy_extent_cache(void); 3986 3987 /* 3988 * sysfs.c 3989 */ 3990 int __init f2fs_init_sysfs(void); 3991 void f2fs_exit_sysfs(void); 3992 int f2fs_register_sysfs(struct f2fs_sb_info *sbi); 3993 void f2fs_unregister_sysfs(struct f2fs_sb_info *sbi); 3994 3995 /* verity.c */ 3996 extern const struct fsverity_operations f2fs_verityops; 3997 3998 /* 3999 * crypto support 4000 */ 4001 static inline bool f2fs_encrypted_file(struct inode *inode) 4002 { 4003 return IS_ENCRYPTED(inode) && S_ISREG(inode->i_mode); 4004 } 4005 4006 static inline void f2fs_set_encrypted_inode(struct inode *inode) 4007 { 4008 #ifdef CONFIG_FS_ENCRYPTION 4009 file_set_encrypt(inode); 4010 f2fs_set_inode_flags(inode); 4011 #endif 4012 } 4013 4014 /* 4015 * Returns true if the reads of the inode's data need to undergo some 4016 * postprocessing step, like decryption or authenticity verification. 4017 */ 4018 static inline bool f2fs_post_read_required(struct inode *inode) 4019 { 4020 return f2fs_encrypted_file(inode) || fsverity_active(inode) || 4021 f2fs_compressed_file(inode); 4022 } 4023 4024 /* 4025 * compress.c 4026 */ 4027 #ifdef CONFIG_F2FS_FS_COMPRESSION 4028 bool f2fs_is_compressed_page(struct page *page); 4029 struct page *f2fs_compress_control_page(struct page *page); 4030 int f2fs_prepare_compress_overwrite(struct inode *inode, 4031 struct page **pagep, pgoff_t index, void **fsdata); 4032 bool f2fs_compress_write_end(struct inode *inode, void *fsdata, 4033 pgoff_t index, unsigned copied); 4034 int f2fs_truncate_partial_cluster(struct inode *inode, u64 from, bool lock); 4035 void f2fs_compress_write_end_io(struct bio *bio, struct page *page); 4036 bool f2fs_is_compress_backend_ready(struct inode *inode); 4037 int f2fs_init_compress_mempool(void); 4038 void f2fs_destroy_compress_mempool(void); 4039 void f2fs_decompress_cluster(struct decompress_io_ctx *dic); 4040 void f2fs_end_read_compressed_page(struct page *page, bool failed, 4041 block_t blkaddr); 4042 bool f2fs_cluster_is_empty(struct compress_ctx *cc); 4043 bool f2fs_cluster_can_merge_page(struct compress_ctx *cc, pgoff_t index); 4044 void f2fs_compress_ctx_add_page(struct compress_ctx *cc, struct page *page); 4045 int f2fs_write_multi_pages(struct compress_ctx *cc, 4046 int *submitted, 4047 struct writeback_control *wbc, 4048 enum iostat_type io_type); 4049 int f2fs_is_compressed_cluster(struct inode *inode, pgoff_t index); 4050 int f2fs_read_multi_pages(struct compress_ctx *cc, struct bio **bio_ret, 4051 unsigned nr_pages, sector_t *last_block_in_bio, 4052 bool is_readahead, bool for_write); 4053 struct decompress_io_ctx *f2fs_alloc_dic(struct compress_ctx *cc); 4054 void f2fs_decompress_end_io(struct decompress_io_ctx *dic, bool failed); 4055 void f2fs_put_page_dic(struct page *page); 4056 int f2fs_init_compress_ctx(struct compress_ctx *cc); 4057 void f2fs_destroy_compress_ctx(struct compress_ctx *cc, bool reuse); 4058 void f2fs_init_compress_info(struct f2fs_sb_info *sbi); 4059 int f2fs_init_compress_inode(struct f2fs_sb_info *sbi); 4060 void f2fs_destroy_compress_inode(struct f2fs_sb_info *sbi); 4061 int f2fs_init_page_array_cache(struct f2fs_sb_info *sbi); 4062 void f2fs_destroy_page_array_cache(struct f2fs_sb_info *sbi); 4063 int __init f2fs_init_compress_cache(void); 4064 void f2fs_destroy_compress_cache(void); 4065 struct address_space *COMPRESS_MAPPING(struct f2fs_sb_info *sbi); 4066 void f2fs_invalidate_compress_page(struct f2fs_sb_info *sbi, block_t blkaddr); 4067 void f2fs_cache_compressed_page(struct f2fs_sb_info *sbi, struct page *page, 4068 nid_t ino, block_t blkaddr); 4069 bool f2fs_load_compressed_page(struct f2fs_sb_info *sbi, struct page *page, 4070 block_t blkaddr); 4071 void f2fs_invalidate_compress_pages(struct f2fs_sb_info *sbi, nid_t ino); 4072 #define inc_compr_inode_stat(inode) \ 4073 do { \ 4074 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); \ 4075 sbi->compr_new_inode++; \ 4076 } while (0) 4077 #define add_compr_block_stat(inode, blocks) \ 4078 do { \ 4079 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); \ 4080 int diff = F2FS_I(inode)->i_cluster_size - blocks; \ 4081 sbi->compr_written_block += blocks; \ 4082 sbi->compr_saved_block += diff; \ 4083 } while (0) 4084 #else 4085 static inline bool f2fs_is_compressed_page(struct page *page) { return false; } 4086 static inline bool f2fs_is_compress_backend_ready(struct inode *inode) 4087 { 4088 if (!f2fs_compressed_file(inode)) 4089 return true; 4090 /* not support compression */ 4091 return false; 4092 } 4093 static inline struct page *f2fs_compress_control_page(struct page *page) 4094 { 4095 WARN_ON_ONCE(1); 4096 return ERR_PTR(-EINVAL); 4097 } 4098 static inline int f2fs_init_compress_mempool(void) { return 0; } 4099 static inline void f2fs_destroy_compress_mempool(void) { } 4100 static inline void f2fs_decompress_cluster(struct decompress_io_ctx *dic) { } 4101 static inline void f2fs_end_read_compressed_page(struct page *page, 4102 bool failed, block_t blkaddr) 4103 { 4104 WARN_ON_ONCE(1); 4105 } 4106 static inline void f2fs_put_page_dic(struct page *page) 4107 { 4108 WARN_ON_ONCE(1); 4109 } 4110 static inline int f2fs_init_compress_inode(struct f2fs_sb_info *sbi) { return 0; } 4111 static inline void f2fs_destroy_compress_inode(struct f2fs_sb_info *sbi) { } 4112 static inline int f2fs_init_page_array_cache(struct f2fs_sb_info *sbi) { return 0; } 4113 static inline void f2fs_destroy_page_array_cache(struct f2fs_sb_info *sbi) { } 4114 static inline int __init f2fs_init_compress_cache(void) { return 0; } 4115 static inline void f2fs_destroy_compress_cache(void) { } 4116 static inline void f2fs_invalidate_compress_page(struct f2fs_sb_info *sbi, 4117 block_t blkaddr) { } 4118 static inline void f2fs_cache_compressed_page(struct f2fs_sb_info *sbi, 4119 struct page *page, nid_t ino, block_t blkaddr) { } 4120 static inline bool f2fs_load_compressed_page(struct f2fs_sb_info *sbi, 4121 struct page *page, block_t blkaddr) { return false; } 4122 static inline void f2fs_invalidate_compress_pages(struct f2fs_sb_info *sbi, 4123 nid_t ino) { } 4124 #define inc_compr_inode_stat(inode) do { } while (0) 4125 #endif 4126 4127 static inline void set_compress_context(struct inode *inode) 4128 { 4129 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 4130 4131 F2FS_I(inode)->i_compress_algorithm = 4132 F2FS_OPTION(sbi).compress_algorithm; 4133 F2FS_I(inode)->i_log_cluster_size = 4134 F2FS_OPTION(sbi).compress_log_size; 4135 F2FS_I(inode)->i_compress_flag = 4136 F2FS_OPTION(sbi).compress_chksum ? 4137 1 << COMPRESS_CHKSUM : 0; 4138 F2FS_I(inode)->i_cluster_size = 4139 1 << F2FS_I(inode)->i_log_cluster_size; 4140 if (F2FS_I(inode)->i_compress_algorithm == COMPRESS_LZ4 && 4141 F2FS_OPTION(sbi).compress_level) 4142 F2FS_I(inode)->i_compress_flag |= 4143 F2FS_OPTION(sbi).compress_level << 4144 COMPRESS_LEVEL_OFFSET; 4145 F2FS_I(inode)->i_flags |= F2FS_COMPR_FL; 4146 set_inode_flag(inode, FI_COMPRESSED_FILE); 4147 stat_inc_compr_inode(inode); 4148 inc_compr_inode_stat(inode); 4149 f2fs_mark_inode_dirty_sync(inode, true); 4150 } 4151 4152 static inline bool f2fs_disable_compressed_file(struct inode *inode) 4153 { 4154 struct f2fs_inode_info *fi = F2FS_I(inode); 4155 4156 if (!f2fs_compressed_file(inode)) 4157 return true; 4158 if (S_ISREG(inode->i_mode) && 4159 (get_dirty_pages(inode) || atomic_read(&fi->i_compr_blocks))) 4160 return false; 4161 4162 fi->i_flags &= ~F2FS_COMPR_FL; 4163 stat_dec_compr_inode(inode); 4164 clear_inode_flag(inode, FI_COMPRESSED_FILE); 4165 f2fs_mark_inode_dirty_sync(inode, true); 4166 return true; 4167 } 4168 4169 #define F2FS_FEATURE_FUNCS(name, flagname) \ 4170 static inline int f2fs_sb_has_##name(struct f2fs_sb_info *sbi) \ 4171 { \ 4172 return F2FS_HAS_FEATURE(sbi, F2FS_FEATURE_##flagname); \ 4173 } 4174 4175 F2FS_FEATURE_FUNCS(encrypt, ENCRYPT); 4176 F2FS_FEATURE_FUNCS(blkzoned, BLKZONED); 4177 F2FS_FEATURE_FUNCS(extra_attr, EXTRA_ATTR); 4178 F2FS_FEATURE_FUNCS(project_quota, PRJQUOTA); 4179 F2FS_FEATURE_FUNCS(inode_chksum, INODE_CHKSUM); 4180 F2FS_FEATURE_FUNCS(flexible_inline_xattr, FLEXIBLE_INLINE_XATTR); 4181 F2FS_FEATURE_FUNCS(quota_ino, QUOTA_INO); 4182 F2FS_FEATURE_FUNCS(inode_crtime, INODE_CRTIME); 4183 F2FS_FEATURE_FUNCS(lost_found, LOST_FOUND); 4184 F2FS_FEATURE_FUNCS(verity, VERITY); 4185 F2FS_FEATURE_FUNCS(sb_chksum, SB_CHKSUM); 4186 F2FS_FEATURE_FUNCS(casefold, CASEFOLD); 4187 F2FS_FEATURE_FUNCS(compression, COMPRESSION); 4188 F2FS_FEATURE_FUNCS(readonly, RO); 4189 4190 static inline bool f2fs_may_extent_tree(struct inode *inode) 4191 { 4192 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 4193 4194 if (!test_opt(sbi, EXTENT_CACHE) || 4195 is_inode_flag_set(inode, FI_NO_EXTENT) || 4196 (is_inode_flag_set(inode, FI_COMPRESSED_FILE) && 4197 !f2fs_sb_has_readonly(sbi))) 4198 return false; 4199 4200 /* 4201 * for recovered files during mount do not create extents 4202 * if shrinker is not registered. 4203 */ 4204 if (list_empty(&sbi->s_list)) 4205 return false; 4206 4207 return S_ISREG(inode->i_mode); 4208 } 4209 4210 #ifdef CONFIG_BLK_DEV_ZONED 4211 static inline bool f2fs_blkz_is_seq(struct f2fs_sb_info *sbi, int devi, 4212 block_t blkaddr) 4213 { 4214 unsigned int zno = blkaddr >> sbi->log_blocks_per_blkz; 4215 4216 return test_bit(zno, FDEV(devi).blkz_seq); 4217 } 4218 #endif 4219 4220 static inline bool f2fs_hw_should_discard(struct f2fs_sb_info *sbi) 4221 { 4222 return f2fs_sb_has_blkzoned(sbi); 4223 } 4224 4225 static inline bool f2fs_bdev_support_discard(struct block_device *bdev) 4226 { 4227 return blk_queue_discard(bdev_get_queue(bdev)) || 4228 bdev_is_zoned(bdev); 4229 } 4230 4231 static inline bool f2fs_hw_support_discard(struct f2fs_sb_info *sbi) 4232 { 4233 int i; 4234 4235 if (!f2fs_is_multi_device(sbi)) 4236 return f2fs_bdev_support_discard(sbi->sb->s_bdev); 4237 4238 for (i = 0; i < sbi->s_ndevs; i++) 4239 if (f2fs_bdev_support_discard(FDEV(i).bdev)) 4240 return true; 4241 return false; 4242 } 4243 4244 static inline bool f2fs_realtime_discard_enable(struct f2fs_sb_info *sbi) 4245 { 4246 return (test_opt(sbi, DISCARD) && f2fs_hw_support_discard(sbi)) || 4247 f2fs_hw_should_discard(sbi); 4248 } 4249 4250 static inline bool f2fs_hw_is_readonly(struct f2fs_sb_info *sbi) 4251 { 4252 int i; 4253 4254 if (!f2fs_is_multi_device(sbi)) 4255 return bdev_read_only(sbi->sb->s_bdev); 4256 4257 for (i = 0; i < sbi->s_ndevs; i++) 4258 if (bdev_read_only(FDEV(i).bdev)) 4259 return true; 4260 return false; 4261 } 4262 4263 static inline bool f2fs_lfs_mode(struct f2fs_sb_info *sbi) 4264 { 4265 return F2FS_OPTION(sbi).fs_mode == FS_MODE_LFS; 4266 } 4267 4268 static inline bool f2fs_may_compress(struct inode *inode) 4269 { 4270 if (IS_SWAPFILE(inode) || f2fs_is_pinned_file(inode) || 4271 f2fs_is_atomic_file(inode) || 4272 f2fs_is_volatile_file(inode)) 4273 return false; 4274 return S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode); 4275 } 4276 4277 static inline void f2fs_i_compr_blocks_update(struct inode *inode, 4278 u64 blocks, bool add) 4279 { 4280 int diff = F2FS_I(inode)->i_cluster_size - blocks; 4281 struct f2fs_inode_info *fi = F2FS_I(inode); 4282 4283 /* don't update i_compr_blocks if saved blocks were released */ 4284 if (!add && !atomic_read(&fi->i_compr_blocks)) 4285 return; 4286 4287 if (add) { 4288 atomic_add(diff, &fi->i_compr_blocks); 4289 stat_add_compr_blocks(inode, diff); 4290 } else { 4291 atomic_sub(diff, &fi->i_compr_blocks); 4292 stat_sub_compr_blocks(inode, diff); 4293 } 4294 f2fs_mark_inode_dirty_sync(inode, true); 4295 } 4296 4297 static inline int block_unaligned_IO(struct inode *inode, 4298 struct kiocb *iocb, struct iov_iter *iter) 4299 { 4300 unsigned int i_blkbits = READ_ONCE(inode->i_blkbits); 4301 unsigned int blocksize_mask = (1 << i_blkbits) - 1; 4302 loff_t offset = iocb->ki_pos; 4303 unsigned long align = offset | iov_iter_alignment(iter); 4304 4305 return align & blocksize_mask; 4306 } 4307 4308 static inline int allow_outplace_dio(struct inode *inode, 4309 struct kiocb *iocb, struct iov_iter *iter) 4310 { 4311 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 4312 int rw = iov_iter_rw(iter); 4313 4314 return (f2fs_lfs_mode(sbi) && (rw == WRITE) && 4315 !block_unaligned_IO(inode, iocb, iter)); 4316 } 4317 4318 static inline bool f2fs_force_buffered_io(struct inode *inode, 4319 struct kiocb *iocb, struct iov_iter *iter) 4320 { 4321 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 4322 int rw = iov_iter_rw(iter); 4323 4324 if (f2fs_post_read_required(inode)) 4325 return true; 4326 if (f2fs_is_multi_device(sbi)) 4327 return true; 4328 /* 4329 * for blkzoned device, fallback direct IO to buffered IO, so 4330 * all IOs can be serialized by log-structured write. 4331 */ 4332 if (f2fs_sb_has_blkzoned(sbi)) 4333 return true; 4334 if (f2fs_lfs_mode(sbi) && (rw == WRITE)) { 4335 if (block_unaligned_IO(inode, iocb, iter)) 4336 return true; 4337 if (F2FS_IO_ALIGNED(sbi)) 4338 return true; 4339 } 4340 if (is_sbi_flag_set(F2FS_I_SB(inode), SBI_CP_DISABLED)) 4341 return true; 4342 4343 return false; 4344 } 4345 4346 static inline bool f2fs_need_verity(const struct inode *inode, pgoff_t idx) 4347 { 4348 return fsverity_active(inode) && 4349 idx < DIV_ROUND_UP(inode->i_size, PAGE_SIZE); 4350 } 4351 4352 #ifdef CONFIG_F2FS_FAULT_INJECTION 4353 extern void f2fs_build_fault_attr(struct f2fs_sb_info *sbi, unsigned int rate, 4354 unsigned int type); 4355 #else 4356 #define f2fs_build_fault_attr(sbi, rate, type) do { } while (0) 4357 #endif 4358 4359 static inline bool is_journalled_quota(struct f2fs_sb_info *sbi) 4360 { 4361 #ifdef CONFIG_QUOTA 4362 if (f2fs_sb_has_quota_ino(sbi)) 4363 return true; 4364 if (F2FS_OPTION(sbi).s_qf_names[USRQUOTA] || 4365 F2FS_OPTION(sbi).s_qf_names[GRPQUOTA] || 4366 F2FS_OPTION(sbi).s_qf_names[PRJQUOTA]) 4367 return true; 4368 #endif 4369 return false; 4370 } 4371 4372 #define EFSBADCRC EBADMSG /* Bad CRC detected */ 4373 #define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */ 4374 4375 #endif /* _LINUX_F2FS_H */ 4376