1 /* SPDX-License-Identifier: GPL-2.0 */ 2 /* 3 * fs/f2fs/f2fs.h 4 * 5 * Copyright (c) 2012 Samsung Electronics Co., Ltd. 6 * http://www.samsung.com/ 7 */ 8 #ifndef _LINUX_F2FS_H 9 #define _LINUX_F2FS_H 10 11 #include <linux/uio.h> 12 #include <linux/types.h> 13 #include <linux/page-flags.h> 14 #include <linux/buffer_head.h> 15 #include <linux/slab.h> 16 #include <linux/crc32.h> 17 #include <linux/magic.h> 18 #include <linux/kobject.h> 19 #include <linux/sched.h> 20 #include <linux/cred.h> 21 #include <linux/sched/mm.h> 22 #include <linux/vmalloc.h> 23 #include <linux/bio.h> 24 #include <linux/blkdev.h> 25 #include <linux/quotaops.h> 26 #include <linux/part_stat.h> 27 #include <crypto/hash.h> 28 29 #include <linux/fscrypt.h> 30 #include <linux/fsverity.h> 31 32 struct pagevec; 33 34 #ifdef CONFIG_F2FS_CHECK_FS 35 #define f2fs_bug_on(sbi, condition) BUG_ON(condition) 36 #else 37 #define f2fs_bug_on(sbi, condition) \ 38 do { \ 39 if (WARN_ON(condition)) \ 40 set_sbi_flag(sbi, SBI_NEED_FSCK); \ 41 } while (0) 42 #endif 43 44 enum { 45 FAULT_KMALLOC, 46 FAULT_KVMALLOC, 47 FAULT_PAGE_ALLOC, 48 FAULT_PAGE_GET, 49 FAULT_ALLOC_BIO, /* it's obsolete due to bio_alloc() will never fail */ 50 FAULT_ALLOC_NID, 51 FAULT_ORPHAN, 52 FAULT_BLOCK, 53 FAULT_DIR_DEPTH, 54 FAULT_EVICT_INODE, 55 FAULT_TRUNCATE, 56 FAULT_READ_IO, 57 FAULT_CHECKPOINT, 58 FAULT_DISCARD, 59 FAULT_WRITE_IO, 60 FAULT_SLAB_ALLOC, 61 FAULT_DQUOT_INIT, 62 FAULT_LOCK_OP, 63 FAULT_MAX, 64 }; 65 66 #ifdef CONFIG_F2FS_FAULT_INJECTION 67 #define F2FS_ALL_FAULT_TYPE ((1 << FAULT_MAX) - 1) 68 69 struct f2fs_fault_info { 70 atomic_t inject_ops; 71 unsigned int inject_rate; 72 unsigned int inject_type; 73 }; 74 75 extern const char *f2fs_fault_name[FAULT_MAX]; 76 #define IS_FAULT_SET(fi, type) ((fi)->inject_type & (1 << (type))) 77 #endif 78 79 /* 80 * For mount options 81 */ 82 #define F2FS_MOUNT_DISABLE_ROLL_FORWARD 0x00000002 83 #define F2FS_MOUNT_DISCARD 0x00000004 84 #define F2FS_MOUNT_NOHEAP 0x00000008 85 #define F2FS_MOUNT_XATTR_USER 0x00000010 86 #define F2FS_MOUNT_POSIX_ACL 0x00000020 87 #define F2FS_MOUNT_DISABLE_EXT_IDENTIFY 0x00000040 88 #define F2FS_MOUNT_INLINE_XATTR 0x00000080 89 #define F2FS_MOUNT_INLINE_DATA 0x00000100 90 #define F2FS_MOUNT_INLINE_DENTRY 0x00000200 91 #define F2FS_MOUNT_FLUSH_MERGE 0x00000400 92 #define F2FS_MOUNT_NOBARRIER 0x00000800 93 #define F2FS_MOUNT_FASTBOOT 0x00001000 94 #define F2FS_MOUNT_EXTENT_CACHE 0x00002000 95 #define F2FS_MOUNT_DATA_FLUSH 0x00008000 96 #define F2FS_MOUNT_FAULT_INJECTION 0x00010000 97 #define F2FS_MOUNT_USRQUOTA 0x00080000 98 #define F2FS_MOUNT_GRPQUOTA 0x00100000 99 #define F2FS_MOUNT_PRJQUOTA 0x00200000 100 #define F2FS_MOUNT_QUOTA 0x00400000 101 #define F2FS_MOUNT_INLINE_XATTR_SIZE 0x00800000 102 #define F2FS_MOUNT_RESERVE_ROOT 0x01000000 103 #define F2FS_MOUNT_DISABLE_CHECKPOINT 0x02000000 104 #define F2FS_MOUNT_NORECOVERY 0x04000000 105 #define F2FS_MOUNT_ATGC 0x08000000 106 #define F2FS_MOUNT_MERGE_CHECKPOINT 0x10000000 107 #define F2FS_MOUNT_GC_MERGE 0x20000000 108 #define F2FS_MOUNT_COMPRESS_CACHE 0x40000000 109 110 #define F2FS_OPTION(sbi) ((sbi)->mount_opt) 111 #define clear_opt(sbi, option) (F2FS_OPTION(sbi).opt &= ~F2FS_MOUNT_##option) 112 #define set_opt(sbi, option) (F2FS_OPTION(sbi).opt |= F2FS_MOUNT_##option) 113 #define test_opt(sbi, option) (F2FS_OPTION(sbi).opt & F2FS_MOUNT_##option) 114 115 #define ver_after(a, b) (typecheck(unsigned long long, a) && \ 116 typecheck(unsigned long long, b) && \ 117 ((long long)((a) - (b)) > 0)) 118 119 typedef u32 block_t; /* 120 * should not change u32, since it is the on-disk block 121 * address format, __le32. 122 */ 123 typedef u32 nid_t; 124 125 #define COMPRESS_EXT_NUM 16 126 127 /* 128 * An implementation of an rwsem that is explicitly unfair to readers. This 129 * prevents priority inversion when a low-priority reader acquires the read lock 130 * while sleeping on the write lock but the write lock is needed by 131 * higher-priority clients. 132 */ 133 134 struct f2fs_rwsem { 135 struct rw_semaphore internal_rwsem; 136 #ifdef CONFIG_F2FS_UNFAIR_RWSEM 137 wait_queue_head_t read_waiters; 138 #endif 139 }; 140 141 struct f2fs_mount_info { 142 unsigned int opt; 143 int write_io_size_bits; /* Write IO size bits */ 144 block_t root_reserved_blocks; /* root reserved blocks */ 145 kuid_t s_resuid; /* reserved blocks for uid */ 146 kgid_t s_resgid; /* reserved blocks for gid */ 147 int active_logs; /* # of active logs */ 148 int inline_xattr_size; /* inline xattr size */ 149 #ifdef CONFIG_F2FS_FAULT_INJECTION 150 struct f2fs_fault_info fault_info; /* For fault injection */ 151 #endif 152 #ifdef CONFIG_QUOTA 153 /* Names of quota files with journalled quota */ 154 char *s_qf_names[MAXQUOTAS]; 155 int s_jquota_fmt; /* Format of quota to use */ 156 #endif 157 /* For which write hints are passed down to block layer */ 158 int alloc_mode; /* segment allocation policy */ 159 int fsync_mode; /* fsync policy */ 160 int fs_mode; /* fs mode: LFS or ADAPTIVE */ 161 int bggc_mode; /* bggc mode: off, on or sync */ 162 int memory_mode; /* memory mode */ 163 int discard_unit; /* 164 * discard command's offset/size should 165 * be aligned to this unit: block, 166 * segment or section 167 */ 168 struct fscrypt_dummy_policy dummy_enc_policy; /* test dummy encryption */ 169 block_t unusable_cap_perc; /* percentage for cap */ 170 block_t unusable_cap; /* Amount of space allowed to be 171 * unusable when disabling checkpoint 172 */ 173 174 /* For compression */ 175 unsigned char compress_algorithm; /* algorithm type */ 176 unsigned char compress_log_size; /* cluster log size */ 177 unsigned char compress_level; /* compress level */ 178 bool compress_chksum; /* compressed data chksum */ 179 unsigned char compress_ext_cnt; /* extension count */ 180 unsigned char nocompress_ext_cnt; /* nocompress extension count */ 181 int compress_mode; /* compression mode */ 182 unsigned char extensions[COMPRESS_EXT_NUM][F2FS_EXTENSION_LEN]; /* extensions */ 183 unsigned char noextensions[COMPRESS_EXT_NUM][F2FS_EXTENSION_LEN]; /* extensions */ 184 }; 185 186 #define F2FS_FEATURE_ENCRYPT 0x0001 187 #define F2FS_FEATURE_BLKZONED 0x0002 188 #define F2FS_FEATURE_ATOMIC_WRITE 0x0004 189 #define F2FS_FEATURE_EXTRA_ATTR 0x0008 190 #define F2FS_FEATURE_PRJQUOTA 0x0010 191 #define F2FS_FEATURE_INODE_CHKSUM 0x0020 192 #define F2FS_FEATURE_FLEXIBLE_INLINE_XATTR 0x0040 193 #define F2FS_FEATURE_QUOTA_INO 0x0080 194 #define F2FS_FEATURE_INODE_CRTIME 0x0100 195 #define F2FS_FEATURE_LOST_FOUND 0x0200 196 #define F2FS_FEATURE_VERITY 0x0400 197 #define F2FS_FEATURE_SB_CHKSUM 0x0800 198 #define F2FS_FEATURE_CASEFOLD 0x1000 199 #define F2FS_FEATURE_COMPRESSION 0x2000 200 #define F2FS_FEATURE_RO 0x4000 201 202 #define __F2FS_HAS_FEATURE(raw_super, mask) \ 203 ((raw_super->feature & cpu_to_le32(mask)) != 0) 204 #define F2FS_HAS_FEATURE(sbi, mask) __F2FS_HAS_FEATURE(sbi->raw_super, mask) 205 #define F2FS_SET_FEATURE(sbi, mask) \ 206 (sbi->raw_super->feature |= cpu_to_le32(mask)) 207 #define F2FS_CLEAR_FEATURE(sbi, mask) \ 208 (sbi->raw_super->feature &= ~cpu_to_le32(mask)) 209 210 /* 211 * Default values for user and/or group using reserved blocks 212 */ 213 #define F2FS_DEF_RESUID 0 214 #define F2FS_DEF_RESGID 0 215 216 /* 217 * For checkpoint manager 218 */ 219 enum { 220 NAT_BITMAP, 221 SIT_BITMAP 222 }; 223 224 #define CP_UMOUNT 0x00000001 225 #define CP_FASTBOOT 0x00000002 226 #define CP_SYNC 0x00000004 227 #define CP_RECOVERY 0x00000008 228 #define CP_DISCARD 0x00000010 229 #define CP_TRIMMED 0x00000020 230 #define CP_PAUSE 0x00000040 231 #define CP_RESIZE 0x00000080 232 233 #define DEF_MAX_DISCARD_REQUEST 8 /* issue 8 discards per round */ 234 #define DEF_MIN_DISCARD_ISSUE_TIME 50 /* 50 ms, if exists */ 235 #define DEF_MID_DISCARD_ISSUE_TIME 500 /* 500 ms, if device busy */ 236 #define DEF_MAX_DISCARD_ISSUE_TIME 60000 /* 60 s, if no candidates */ 237 #define DEF_DISCARD_URGENT_UTIL 80 /* do more discard over 80% */ 238 #define DEF_CP_INTERVAL 60 /* 60 secs */ 239 #define DEF_IDLE_INTERVAL 5 /* 5 secs */ 240 #define DEF_DISABLE_INTERVAL 5 /* 5 secs */ 241 #define DEF_DISABLE_QUICK_INTERVAL 1 /* 1 secs */ 242 #define DEF_UMOUNT_DISCARD_TIMEOUT 5 /* 5 secs */ 243 244 struct cp_control { 245 int reason; 246 __u64 trim_start; 247 __u64 trim_end; 248 __u64 trim_minlen; 249 }; 250 251 /* 252 * indicate meta/data type 253 */ 254 enum { 255 META_CP, 256 META_NAT, 257 META_SIT, 258 META_SSA, 259 META_MAX, 260 META_POR, 261 DATA_GENERIC, /* check range only */ 262 DATA_GENERIC_ENHANCE, /* strong check on range and segment bitmap */ 263 DATA_GENERIC_ENHANCE_READ, /* 264 * strong check on range and segment 265 * bitmap but no warning due to race 266 * condition of read on truncated area 267 * by extent_cache 268 */ 269 DATA_GENERIC_ENHANCE_UPDATE, /* 270 * strong check on range and segment 271 * bitmap for update case 272 */ 273 META_GENERIC, 274 }; 275 276 /* for the list of ino */ 277 enum { 278 ORPHAN_INO, /* for orphan ino list */ 279 APPEND_INO, /* for append ino list */ 280 UPDATE_INO, /* for update ino list */ 281 TRANS_DIR_INO, /* for transactions dir ino list */ 282 FLUSH_INO, /* for multiple device flushing */ 283 MAX_INO_ENTRY, /* max. list */ 284 }; 285 286 struct ino_entry { 287 struct list_head list; /* list head */ 288 nid_t ino; /* inode number */ 289 unsigned int dirty_device; /* dirty device bitmap */ 290 }; 291 292 /* for the list of inodes to be GCed */ 293 struct inode_entry { 294 struct list_head list; /* list head */ 295 struct inode *inode; /* vfs inode pointer */ 296 }; 297 298 struct fsync_node_entry { 299 struct list_head list; /* list head */ 300 struct page *page; /* warm node page pointer */ 301 unsigned int seq_id; /* sequence id */ 302 }; 303 304 struct ckpt_req { 305 struct completion wait; /* completion for checkpoint done */ 306 struct llist_node llnode; /* llist_node to be linked in wait queue */ 307 int ret; /* return code of checkpoint */ 308 ktime_t queue_time; /* request queued time */ 309 }; 310 311 struct ckpt_req_control { 312 struct task_struct *f2fs_issue_ckpt; /* checkpoint task */ 313 int ckpt_thread_ioprio; /* checkpoint merge thread ioprio */ 314 wait_queue_head_t ckpt_wait_queue; /* waiting queue for wake-up */ 315 atomic_t issued_ckpt; /* # of actually issued ckpts */ 316 atomic_t total_ckpt; /* # of total ckpts */ 317 atomic_t queued_ckpt; /* # of queued ckpts */ 318 struct llist_head issue_list; /* list for command issue */ 319 spinlock_t stat_lock; /* lock for below checkpoint time stats */ 320 unsigned int cur_time; /* cur wait time in msec for currently issued checkpoint */ 321 unsigned int peak_time; /* peak wait time in msec until now */ 322 }; 323 324 /* for the bitmap indicate blocks to be discarded */ 325 struct discard_entry { 326 struct list_head list; /* list head */ 327 block_t start_blkaddr; /* start blockaddr of current segment */ 328 unsigned char discard_map[SIT_VBLOCK_MAP_SIZE]; /* segment discard bitmap */ 329 }; 330 331 /* default discard granularity of inner discard thread, unit: block count */ 332 #define DEFAULT_DISCARD_GRANULARITY 16 333 334 /* max discard pend list number */ 335 #define MAX_PLIST_NUM 512 336 #define plist_idx(blk_num) ((blk_num) >= MAX_PLIST_NUM ? \ 337 (MAX_PLIST_NUM - 1) : ((blk_num) - 1)) 338 339 enum { 340 D_PREP, /* initial */ 341 D_PARTIAL, /* partially submitted */ 342 D_SUBMIT, /* all submitted */ 343 D_DONE, /* finished */ 344 }; 345 346 struct discard_info { 347 block_t lstart; /* logical start address */ 348 block_t len; /* length */ 349 block_t start; /* actual start address in dev */ 350 }; 351 352 struct discard_cmd { 353 struct rb_node rb_node; /* rb node located in rb-tree */ 354 union { 355 struct { 356 block_t lstart; /* logical start address */ 357 block_t len; /* length */ 358 block_t start; /* actual start address in dev */ 359 }; 360 struct discard_info di; /* discard info */ 361 362 }; 363 struct list_head list; /* command list */ 364 struct completion wait; /* compleation */ 365 struct block_device *bdev; /* bdev */ 366 unsigned short ref; /* reference count */ 367 unsigned char state; /* state */ 368 unsigned char queued; /* queued discard */ 369 int error; /* bio error */ 370 spinlock_t lock; /* for state/bio_ref updating */ 371 unsigned short bio_ref; /* bio reference count */ 372 }; 373 374 enum { 375 DPOLICY_BG, 376 DPOLICY_FORCE, 377 DPOLICY_FSTRIM, 378 DPOLICY_UMOUNT, 379 MAX_DPOLICY, 380 }; 381 382 struct discard_policy { 383 int type; /* type of discard */ 384 unsigned int min_interval; /* used for candidates exist */ 385 unsigned int mid_interval; /* used for device busy */ 386 unsigned int max_interval; /* used for candidates not exist */ 387 unsigned int max_requests; /* # of discards issued per round */ 388 unsigned int io_aware_gran; /* minimum granularity discard not be aware of I/O */ 389 bool io_aware; /* issue discard in idle time */ 390 bool sync; /* submit discard with REQ_SYNC flag */ 391 bool ordered; /* issue discard by lba order */ 392 bool timeout; /* discard timeout for put_super */ 393 unsigned int granularity; /* discard granularity */ 394 }; 395 396 struct discard_cmd_control { 397 struct task_struct *f2fs_issue_discard; /* discard thread */ 398 struct list_head entry_list; /* 4KB discard entry list */ 399 struct list_head pend_list[MAX_PLIST_NUM];/* store pending entries */ 400 struct list_head wait_list; /* store on-flushing entries */ 401 struct list_head fstrim_list; /* in-flight discard from fstrim */ 402 wait_queue_head_t discard_wait_queue; /* waiting queue for wake-up */ 403 unsigned int discard_wake; /* to wake up discard thread */ 404 struct mutex cmd_lock; 405 unsigned int nr_discards; /* # of discards in the list */ 406 unsigned int max_discards; /* max. discards to be issued */ 407 unsigned int max_discard_request; /* max. discard request per round */ 408 unsigned int min_discard_issue_time; /* min. interval between discard issue */ 409 unsigned int mid_discard_issue_time; /* mid. interval between discard issue */ 410 unsigned int max_discard_issue_time; /* max. interval between discard issue */ 411 unsigned int discard_granularity; /* discard granularity */ 412 unsigned int undiscard_blks; /* # of undiscard blocks */ 413 unsigned int next_pos; /* next discard position */ 414 atomic_t issued_discard; /* # of issued discard */ 415 atomic_t queued_discard; /* # of queued discard */ 416 atomic_t discard_cmd_cnt; /* # of cached cmd count */ 417 struct rb_root_cached root; /* root of discard rb-tree */ 418 bool rbtree_check; /* config for consistence check */ 419 }; 420 421 /* for the list of fsync inodes, used only during recovery */ 422 struct fsync_inode_entry { 423 struct list_head list; /* list head */ 424 struct inode *inode; /* vfs inode pointer */ 425 block_t blkaddr; /* block address locating the last fsync */ 426 block_t last_dentry; /* block address locating the last dentry */ 427 }; 428 429 #define nats_in_cursum(jnl) (le16_to_cpu((jnl)->n_nats)) 430 #define sits_in_cursum(jnl) (le16_to_cpu((jnl)->n_sits)) 431 432 #define nat_in_journal(jnl, i) ((jnl)->nat_j.entries[i].ne) 433 #define nid_in_journal(jnl, i) ((jnl)->nat_j.entries[i].nid) 434 #define sit_in_journal(jnl, i) ((jnl)->sit_j.entries[i].se) 435 #define segno_in_journal(jnl, i) ((jnl)->sit_j.entries[i].segno) 436 437 #define MAX_NAT_JENTRIES(jnl) (NAT_JOURNAL_ENTRIES - nats_in_cursum(jnl)) 438 #define MAX_SIT_JENTRIES(jnl) (SIT_JOURNAL_ENTRIES - sits_in_cursum(jnl)) 439 440 static inline int update_nats_in_cursum(struct f2fs_journal *journal, int i) 441 { 442 int before = nats_in_cursum(journal); 443 444 journal->n_nats = cpu_to_le16(before + i); 445 return before; 446 } 447 448 static inline int update_sits_in_cursum(struct f2fs_journal *journal, int i) 449 { 450 int before = sits_in_cursum(journal); 451 452 journal->n_sits = cpu_to_le16(before + i); 453 return before; 454 } 455 456 static inline bool __has_cursum_space(struct f2fs_journal *journal, 457 int size, int type) 458 { 459 if (type == NAT_JOURNAL) 460 return size <= MAX_NAT_JENTRIES(journal); 461 return size <= MAX_SIT_JENTRIES(journal); 462 } 463 464 /* for inline stuff */ 465 #define DEF_INLINE_RESERVED_SIZE 1 466 static inline int get_extra_isize(struct inode *inode); 467 static inline int get_inline_xattr_addrs(struct inode *inode); 468 #define MAX_INLINE_DATA(inode) (sizeof(__le32) * \ 469 (CUR_ADDRS_PER_INODE(inode) - \ 470 get_inline_xattr_addrs(inode) - \ 471 DEF_INLINE_RESERVED_SIZE)) 472 473 /* for inline dir */ 474 #define NR_INLINE_DENTRY(inode) (MAX_INLINE_DATA(inode) * BITS_PER_BYTE / \ 475 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \ 476 BITS_PER_BYTE + 1)) 477 #define INLINE_DENTRY_BITMAP_SIZE(inode) \ 478 DIV_ROUND_UP(NR_INLINE_DENTRY(inode), BITS_PER_BYTE) 479 #define INLINE_RESERVED_SIZE(inode) (MAX_INLINE_DATA(inode) - \ 480 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \ 481 NR_INLINE_DENTRY(inode) + \ 482 INLINE_DENTRY_BITMAP_SIZE(inode))) 483 484 /* 485 * For INODE and NODE manager 486 */ 487 /* for directory operations */ 488 489 struct f2fs_filename { 490 /* 491 * The filename the user specified. This is NULL for some 492 * filesystem-internal operations, e.g. converting an inline directory 493 * to a non-inline one, or roll-forward recovering an encrypted dentry. 494 */ 495 const struct qstr *usr_fname; 496 497 /* 498 * The on-disk filename. For encrypted directories, this is encrypted. 499 * This may be NULL for lookups in an encrypted dir without the key. 500 */ 501 struct fscrypt_str disk_name; 502 503 /* The dirhash of this filename */ 504 f2fs_hash_t hash; 505 506 #ifdef CONFIG_FS_ENCRYPTION 507 /* 508 * For lookups in encrypted directories: either the buffer backing 509 * disk_name, or a buffer that holds the decoded no-key name. 510 */ 511 struct fscrypt_str crypto_buf; 512 #endif 513 #if IS_ENABLED(CONFIG_UNICODE) 514 /* 515 * For casefolded directories: the casefolded name, but it's left NULL 516 * if the original name is not valid Unicode, if the original name is 517 * "." or "..", if the directory is both casefolded and encrypted and 518 * its encryption key is unavailable, or if the filesystem is doing an 519 * internal operation where usr_fname is also NULL. In all these cases 520 * we fall back to treating the name as an opaque byte sequence. 521 */ 522 struct fscrypt_str cf_name; 523 #endif 524 }; 525 526 struct f2fs_dentry_ptr { 527 struct inode *inode; 528 void *bitmap; 529 struct f2fs_dir_entry *dentry; 530 __u8 (*filename)[F2FS_SLOT_LEN]; 531 int max; 532 int nr_bitmap; 533 }; 534 535 static inline void make_dentry_ptr_block(struct inode *inode, 536 struct f2fs_dentry_ptr *d, struct f2fs_dentry_block *t) 537 { 538 d->inode = inode; 539 d->max = NR_DENTRY_IN_BLOCK; 540 d->nr_bitmap = SIZE_OF_DENTRY_BITMAP; 541 d->bitmap = t->dentry_bitmap; 542 d->dentry = t->dentry; 543 d->filename = t->filename; 544 } 545 546 static inline void make_dentry_ptr_inline(struct inode *inode, 547 struct f2fs_dentry_ptr *d, void *t) 548 { 549 int entry_cnt = NR_INLINE_DENTRY(inode); 550 int bitmap_size = INLINE_DENTRY_BITMAP_SIZE(inode); 551 int reserved_size = INLINE_RESERVED_SIZE(inode); 552 553 d->inode = inode; 554 d->max = entry_cnt; 555 d->nr_bitmap = bitmap_size; 556 d->bitmap = t; 557 d->dentry = t + bitmap_size + reserved_size; 558 d->filename = t + bitmap_size + reserved_size + 559 SIZE_OF_DIR_ENTRY * entry_cnt; 560 } 561 562 /* 563 * XATTR_NODE_OFFSET stores xattrs to one node block per file keeping -1 564 * as its node offset to distinguish from index node blocks. 565 * But some bits are used to mark the node block. 566 */ 567 #define XATTR_NODE_OFFSET ((((unsigned int)-1) << OFFSET_BIT_SHIFT) \ 568 >> OFFSET_BIT_SHIFT) 569 enum { 570 ALLOC_NODE, /* allocate a new node page if needed */ 571 LOOKUP_NODE, /* look up a node without readahead */ 572 LOOKUP_NODE_RA, /* 573 * look up a node with readahead called 574 * by get_data_block. 575 */ 576 }; 577 578 #define DEFAULT_RETRY_IO_COUNT 8 /* maximum retry read IO or flush count */ 579 580 /* congestion wait timeout value, default: 20ms */ 581 #define DEFAULT_IO_TIMEOUT (msecs_to_jiffies(20)) 582 583 /* maximum retry quota flush count */ 584 #define DEFAULT_RETRY_QUOTA_FLUSH_COUNT 8 585 586 /* maximum retry of EIO'ed page */ 587 #define MAX_RETRY_PAGE_EIO 100 588 589 #define F2FS_LINK_MAX 0xffffffff /* maximum link count per file */ 590 591 #define MAX_DIR_RA_PAGES 4 /* maximum ra pages of dir */ 592 593 /* dirty segments threshold for triggering CP */ 594 #define DEFAULT_DIRTY_THRESHOLD 4 595 596 /* for in-memory extent cache entry */ 597 #define F2FS_MIN_EXTENT_LEN 64 /* minimum extent length */ 598 599 /* number of extent info in extent cache we try to shrink */ 600 #define EXTENT_CACHE_SHRINK_NUMBER 128 601 602 #define RECOVERY_MAX_RA_BLOCKS BIO_MAX_VECS 603 #define RECOVERY_MIN_RA_BLOCKS 1 604 605 #define F2FS_ONSTACK_PAGES 16 /* nr of onstack pages */ 606 607 struct rb_entry { 608 struct rb_node rb_node; /* rb node located in rb-tree */ 609 union { 610 struct { 611 unsigned int ofs; /* start offset of the entry */ 612 unsigned int len; /* length of the entry */ 613 }; 614 unsigned long long key; /* 64-bits key */ 615 } __packed; 616 }; 617 618 struct extent_info { 619 unsigned int fofs; /* start offset in a file */ 620 unsigned int len; /* length of the extent */ 621 u32 blk; /* start block address of the extent */ 622 #ifdef CONFIG_F2FS_FS_COMPRESSION 623 unsigned int c_len; /* physical extent length of compressed blocks */ 624 #endif 625 }; 626 627 struct extent_node { 628 struct rb_node rb_node; /* rb node located in rb-tree */ 629 struct extent_info ei; /* extent info */ 630 struct list_head list; /* node in global extent list of sbi */ 631 struct extent_tree *et; /* extent tree pointer */ 632 }; 633 634 struct extent_tree { 635 nid_t ino; /* inode number */ 636 struct rb_root_cached root; /* root of extent info rb-tree */ 637 struct extent_node *cached_en; /* recently accessed extent node */ 638 struct extent_info largest; /* largested extent info */ 639 struct list_head list; /* to be used by sbi->zombie_list */ 640 rwlock_t lock; /* protect extent info rb-tree */ 641 atomic_t node_cnt; /* # of extent node in rb-tree*/ 642 bool largest_updated; /* largest extent updated */ 643 }; 644 645 /* 646 * This structure is taken from ext4_map_blocks. 647 * 648 * Note that, however, f2fs uses NEW and MAPPED flags for f2fs_map_blocks(). 649 */ 650 #define F2FS_MAP_NEW (1 << BH_New) 651 #define F2FS_MAP_MAPPED (1 << BH_Mapped) 652 #define F2FS_MAP_UNWRITTEN (1 << BH_Unwritten) 653 #define F2FS_MAP_FLAGS (F2FS_MAP_NEW | F2FS_MAP_MAPPED |\ 654 F2FS_MAP_UNWRITTEN) 655 656 struct f2fs_map_blocks { 657 struct block_device *m_bdev; /* for multi-device dio */ 658 block_t m_pblk; 659 block_t m_lblk; 660 unsigned int m_len; 661 unsigned int m_flags; 662 pgoff_t *m_next_pgofs; /* point next possible non-hole pgofs */ 663 pgoff_t *m_next_extent; /* point to next possible extent */ 664 int m_seg_type; 665 bool m_may_create; /* indicate it is from write path */ 666 bool m_multidev_dio; /* indicate it allows multi-device dio */ 667 }; 668 669 /* for flag in get_data_block */ 670 enum { 671 F2FS_GET_BLOCK_DEFAULT, 672 F2FS_GET_BLOCK_FIEMAP, 673 F2FS_GET_BLOCK_BMAP, 674 F2FS_GET_BLOCK_DIO, 675 F2FS_GET_BLOCK_PRE_DIO, 676 F2FS_GET_BLOCK_PRE_AIO, 677 F2FS_GET_BLOCK_PRECACHE, 678 }; 679 680 /* 681 * i_advise uses FADVISE_XXX_BIT. We can add additional hints later. 682 */ 683 #define FADVISE_COLD_BIT 0x01 684 #define FADVISE_LOST_PINO_BIT 0x02 685 #define FADVISE_ENCRYPT_BIT 0x04 686 #define FADVISE_ENC_NAME_BIT 0x08 687 #define FADVISE_KEEP_SIZE_BIT 0x10 688 #define FADVISE_HOT_BIT 0x20 689 #define FADVISE_VERITY_BIT 0x40 690 #define FADVISE_TRUNC_BIT 0x80 691 692 #define FADVISE_MODIFIABLE_BITS (FADVISE_COLD_BIT | FADVISE_HOT_BIT) 693 694 #define file_is_cold(inode) is_file(inode, FADVISE_COLD_BIT) 695 #define file_set_cold(inode) set_file(inode, FADVISE_COLD_BIT) 696 #define file_clear_cold(inode) clear_file(inode, FADVISE_COLD_BIT) 697 698 #define file_wrong_pino(inode) is_file(inode, FADVISE_LOST_PINO_BIT) 699 #define file_lost_pino(inode) set_file(inode, FADVISE_LOST_PINO_BIT) 700 #define file_got_pino(inode) clear_file(inode, FADVISE_LOST_PINO_BIT) 701 702 #define file_is_encrypt(inode) is_file(inode, FADVISE_ENCRYPT_BIT) 703 #define file_set_encrypt(inode) set_file(inode, FADVISE_ENCRYPT_BIT) 704 705 #define file_enc_name(inode) is_file(inode, FADVISE_ENC_NAME_BIT) 706 #define file_set_enc_name(inode) set_file(inode, FADVISE_ENC_NAME_BIT) 707 708 #define file_keep_isize(inode) is_file(inode, FADVISE_KEEP_SIZE_BIT) 709 #define file_set_keep_isize(inode) set_file(inode, FADVISE_KEEP_SIZE_BIT) 710 711 #define file_is_hot(inode) is_file(inode, FADVISE_HOT_BIT) 712 #define file_set_hot(inode) set_file(inode, FADVISE_HOT_BIT) 713 #define file_clear_hot(inode) clear_file(inode, FADVISE_HOT_BIT) 714 715 #define file_is_verity(inode) is_file(inode, FADVISE_VERITY_BIT) 716 #define file_set_verity(inode) set_file(inode, FADVISE_VERITY_BIT) 717 718 #define file_should_truncate(inode) is_file(inode, FADVISE_TRUNC_BIT) 719 #define file_need_truncate(inode) set_file(inode, FADVISE_TRUNC_BIT) 720 #define file_dont_truncate(inode) clear_file(inode, FADVISE_TRUNC_BIT) 721 722 #define DEF_DIR_LEVEL 0 723 724 enum { 725 GC_FAILURE_PIN, 726 MAX_GC_FAILURE 727 }; 728 729 /* used for f2fs_inode_info->flags */ 730 enum { 731 FI_NEW_INODE, /* indicate newly allocated inode */ 732 FI_DIRTY_INODE, /* indicate inode is dirty or not */ 733 FI_AUTO_RECOVER, /* indicate inode is recoverable */ 734 FI_DIRTY_DIR, /* indicate directory has dirty pages */ 735 FI_INC_LINK, /* need to increment i_nlink */ 736 FI_ACL_MODE, /* indicate acl mode */ 737 FI_NO_ALLOC, /* should not allocate any blocks */ 738 FI_FREE_NID, /* free allocated nide */ 739 FI_NO_EXTENT, /* not to use the extent cache */ 740 FI_INLINE_XATTR, /* used for inline xattr */ 741 FI_INLINE_DATA, /* used for inline data*/ 742 FI_INLINE_DENTRY, /* used for inline dentry */ 743 FI_APPEND_WRITE, /* inode has appended data */ 744 FI_UPDATE_WRITE, /* inode has in-place-update data */ 745 FI_NEED_IPU, /* used for ipu per file */ 746 FI_ATOMIC_FILE, /* indicate atomic file */ 747 FI_FIRST_BLOCK_WRITTEN, /* indicate #0 data block was written */ 748 FI_DROP_CACHE, /* drop dirty page cache */ 749 FI_DATA_EXIST, /* indicate data exists */ 750 FI_INLINE_DOTS, /* indicate inline dot dentries */ 751 FI_SKIP_WRITES, /* should skip data page writeback */ 752 FI_OPU_WRITE, /* used for opu per file */ 753 FI_DIRTY_FILE, /* indicate regular/symlink has dirty pages */ 754 FI_PREALLOCATED_ALL, /* all blocks for write were preallocated */ 755 FI_HOT_DATA, /* indicate file is hot */ 756 FI_EXTRA_ATTR, /* indicate file has extra attribute */ 757 FI_PROJ_INHERIT, /* indicate file inherits projectid */ 758 FI_PIN_FILE, /* indicate file should not be gced */ 759 FI_VERITY_IN_PROGRESS, /* building fs-verity Merkle tree */ 760 FI_COMPRESSED_FILE, /* indicate file's data can be compressed */ 761 FI_COMPRESS_CORRUPT, /* indicate compressed cluster is corrupted */ 762 FI_MMAP_FILE, /* indicate file was mmapped */ 763 FI_ENABLE_COMPRESS, /* enable compression in "user" compression mode */ 764 FI_COMPRESS_RELEASED, /* compressed blocks were released */ 765 FI_ALIGNED_WRITE, /* enable aligned write */ 766 FI_COW_FILE, /* indicate COW file */ 767 FI_MAX, /* max flag, never be used */ 768 }; 769 770 struct f2fs_inode_info { 771 struct inode vfs_inode; /* serve a vfs inode */ 772 unsigned long i_flags; /* keep an inode flags for ioctl */ 773 unsigned char i_advise; /* use to give file attribute hints */ 774 unsigned char i_dir_level; /* use for dentry level for large dir */ 775 unsigned int i_current_depth; /* only for directory depth */ 776 /* for gc failure statistic */ 777 unsigned int i_gc_failures[MAX_GC_FAILURE]; 778 unsigned int i_pino; /* parent inode number */ 779 umode_t i_acl_mode; /* keep file acl mode temporarily */ 780 781 /* Use below internally in f2fs*/ 782 unsigned long flags[BITS_TO_LONGS(FI_MAX)]; /* use to pass per-file flags */ 783 struct f2fs_rwsem i_sem; /* protect fi info */ 784 atomic_t dirty_pages; /* # of dirty pages */ 785 f2fs_hash_t chash; /* hash value of given file name */ 786 unsigned int clevel; /* maximum level of given file name */ 787 struct task_struct *task; /* lookup and create consistency */ 788 struct task_struct *cp_task; /* separate cp/wb IO stats*/ 789 struct task_struct *wb_task; /* indicate inode is in context of writeback */ 790 nid_t i_xattr_nid; /* node id that contains xattrs */ 791 loff_t last_disk_size; /* lastly written file size */ 792 spinlock_t i_size_lock; /* protect last_disk_size */ 793 794 #ifdef CONFIG_QUOTA 795 struct dquot *i_dquot[MAXQUOTAS]; 796 797 /* quota space reservation, managed internally by quota code */ 798 qsize_t i_reserved_quota; 799 #endif 800 struct list_head dirty_list; /* dirty list for dirs and files */ 801 struct list_head gdirty_list; /* linked in global dirty list */ 802 struct task_struct *atomic_write_task; /* store atomic write task */ 803 struct extent_tree *extent_tree; /* cached extent_tree entry */ 804 struct inode *cow_inode; /* copy-on-write inode for atomic write */ 805 806 /* avoid racing between foreground op and gc */ 807 struct f2fs_rwsem i_gc_rwsem[2]; 808 struct f2fs_rwsem i_xattr_sem; /* avoid racing between reading and changing EAs */ 809 810 int i_extra_isize; /* size of extra space located in i_addr */ 811 kprojid_t i_projid; /* id for project quota */ 812 int i_inline_xattr_size; /* inline xattr size */ 813 struct timespec64 i_crtime; /* inode creation time */ 814 struct timespec64 i_disk_time[4];/* inode disk times */ 815 816 /* for file compress */ 817 atomic_t i_compr_blocks; /* # of compressed blocks */ 818 unsigned char i_compress_algorithm; /* algorithm type */ 819 unsigned char i_log_cluster_size; /* log of cluster size */ 820 unsigned char i_compress_level; /* compress level (lz4hc,zstd) */ 821 unsigned short i_compress_flag; /* compress flag */ 822 unsigned int i_cluster_size; /* cluster size */ 823 824 unsigned int atomic_write_cnt; 825 }; 826 827 static inline void get_extent_info(struct extent_info *ext, 828 struct f2fs_extent *i_ext) 829 { 830 ext->fofs = le32_to_cpu(i_ext->fofs); 831 ext->blk = le32_to_cpu(i_ext->blk); 832 ext->len = le32_to_cpu(i_ext->len); 833 } 834 835 static inline void set_raw_extent(struct extent_info *ext, 836 struct f2fs_extent *i_ext) 837 { 838 i_ext->fofs = cpu_to_le32(ext->fofs); 839 i_ext->blk = cpu_to_le32(ext->blk); 840 i_ext->len = cpu_to_le32(ext->len); 841 } 842 843 static inline void set_extent_info(struct extent_info *ei, unsigned int fofs, 844 u32 blk, unsigned int len) 845 { 846 ei->fofs = fofs; 847 ei->blk = blk; 848 ei->len = len; 849 #ifdef CONFIG_F2FS_FS_COMPRESSION 850 ei->c_len = 0; 851 #endif 852 } 853 854 static inline bool __is_discard_mergeable(struct discard_info *back, 855 struct discard_info *front, unsigned int max_len) 856 { 857 return (back->lstart + back->len == front->lstart) && 858 (back->len + front->len <= max_len); 859 } 860 861 static inline bool __is_discard_back_mergeable(struct discard_info *cur, 862 struct discard_info *back, unsigned int max_len) 863 { 864 return __is_discard_mergeable(back, cur, max_len); 865 } 866 867 static inline bool __is_discard_front_mergeable(struct discard_info *cur, 868 struct discard_info *front, unsigned int max_len) 869 { 870 return __is_discard_mergeable(cur, front, max_len); 871 } 872 873 static inline bool __is_extent_mergeable(struct extent_info *back, 874 struct extent_info *front) 875 { 876 #ifdef CONFIG_F2FS_FS_COMPRESSION 877 if (back->c_len && back->len != back->c_len) 878 return false; 879 if (front->c_len && front->len != front->c_len) 880 return false; 881 #endif 882 return (back->fofs + back->len == front->fofs && 883 back->blk + back->len == front->blk); 884 } 885 886 static inline bool __is_back_mergeable(struct extent_info *cur, 887 struct extent_info *back) 888 { 889 return __is_extent_mergeable(back, cur); 890 } 891 892 static inline bool __is_front_mergeable(struct extent_info *cur, 893 struct extent_info *front) 894 { 895 return __is_extent_mergeable(cur, front); 896 } 897 898 extern void f2fs_mark_inode_dirty_sync(struct inode *inode, bool sync); 899 static inline void __try_update_largest_extent(struct extent_tree *et, 900 struct extent_node *en) 901 { 902 if (en->ei.len > et->largest.len) { 903 et->largest = en->ei; 904 et->largest_updated = true; 905 } 906 } 907 908 /* 909 * For free nid management 910 */ 911 enum nid_state { 912 FREE_NID, /* newly added to free nid list */ 913 PREALLOC_NID, /* it is preallocated */ 914 MAX_NID_STATE, 915 }; 916 917 enum nat_state { 918 TOTAL_NAT, 919 DIRTY_NAT, 920 RECLAIMABLE_NAT, 921 MAX_NAT_STATE, 922 }; 923 924 struct f2fs_nm_info { 925 block_t nat_blkaddr; /* base disk address of NAT */ 926 nid_t max_nid; /* maximum possible node ids */ 927 nid_t available_nids; /* # of available node ids */ 928 nid_t next_scan_nid; /* the next nid to be scanned */ 929 nid_t max_rf_node_blocks; /* max # of nodes for recovery */ 930 unsigned int ram_thresh; /* control the memory footprint */ 931 unsigned int ra_nid_pages; /* # of nid pages to be readaheaded */ 932 unsigned int dirty_nats_ratio; /* control dirty nats ratio threshold */ 933 934 /* NAT cache management */ 935 struct radix_tree_root nat_root;/* root of the nat entry cache */ 936 struct radix_tree_root nat_set_root;/* root of the nat set cache */ 937 struct f2fs_rwsem nat_tree_lock; /* protect nat entry tree */ 938 struct list_head nat_entries; /* cached nat entry list (clean) */ 939 spinlock_t nat_list_lock; /* protect clean nat entry list */ 940 unsigned int nat_cnt[MAX_NAT_STATE]; /* the # of cached nat entries */ 941 unsigned int nat_blocks; /* # of nat blocks */ 942 943 /* free node ids management */ 944 struct radix_tree_root free_nid_root;/* root of the free_nid cache */ 945 struct list_head free_nid_list; /* list for free nids excluding preallocated nids */ 946 unsigned int nid_cnt[MAX_NID_STATE]; /* the number of free node id */ 947 spinlock_t nid_list_lock; /* protect nid lists ops */ 948 struct mutex build_lock; /* lock for build free nids */ 949 unsigned char **free_nid_bitmap; 950 unsigned char *nat_block_bitmap; 951 unsigned short *free_nid_count; /* free nid count of NAT block */ 952 953 /* for checkpoint */ 954 char *nat_bitmap; /* NAT bitmap pointer */ 955 956 unsigned int nat_bits_blocks; /* # of nat bits blocks */ 957 unsigned char *nat_bits; /* NAT bits blocks */ 958 unsigned char *full_nat_bits; /* full NAT pages */ 959 unsigned char *empty_nat_bits; /* empty NAT pages */ 960 #ifdef CONFIG_F2FS_CHECK_FS 961 char *nat_bitmap_mir; /* NAT bitmap mirror */ 962 #endif 963 int bitmap_size; /* bitmap size */ 964 }; 965 966 /* 967 * this structure is used as one of function parameters. 968 * all the information are dedicated to a given direct node block determined 969 * by the data offset in a file. 970 */ 971 struct dnode_of_data { 972 struct inode *inode; /* vfs inode pointer */ 973 struct page *inode_page; /* its inode page, NULL is possible */ 974 struct page *node_page; /* cached direct node page */ 975 nid_t nid; /* node id of the direct node block */ 976 unsigned int ofs_in_node; /* data offset in the node page */ 977 bool inode_page_locked; /* inode page is locked or not */ 978 bool node_changed; /* is node block changed */ 979 char cur_level; /* level of hole node page */ 980 char max_level; /* level of current page located */ 981 block_t data_blkaddr; /* block address of the node block */ 982 }; 983 984 static inline void set_new_dnode(struct dnode_of_data *dn, struct inode *inode, 985 struct page *ipage, struct page *npage, nid_t nid) 986 { 987 memset(dn, 0, sizeof(*dn)); 988 dn->inode = inode; 989 dn->inode_page = ipage; 990 dn->node_page = npage; 991 dn->nid = nid; 992 } 993 994 /* 995 * For SIT manager 996 * 997 * By default, there are 6 active log areas across the whole main area. 998 * When considering hot and cold data separation to reduce cleaning overhead, 999 * we split 3 for data logs and 3 for node logs as hot, warm, and cold types, 1000 * respectively. 1001 * In the current design, you should not change the numbers intentionally. 1002 * Instead, as a mount option such as active_logs=x, you can use 2, 4, and 6 1003 * logs individually according to the underlying devices. (default: 6) 1004 * Just in case, on-disk layout covers maximum 16 logs that consist of 8 for 1005 * data and 8 for node logs. 1006 */ 1007 #define NR_CURSEG_DATA_TYPE (3) 1008 #define NR_CURSEG_NODE_TYPE (3) 1009 #define NR_CURSEG_INMEM_TYPE (2) 1010 #define NR_CURSEG_RO_TYPE (2) 1011 #define NR_CURSEG_PERSIST_TYPE (NR_CURSEG_DATA_TYPE + NR_CURSEG_NODE_TYPE) 1012 #define NR_CURSEG_TYPE (NR_CURSEG_INMEM_TYPE + NR_CURSEG_PERSIST_TYPE) 1013 1014 enum { 1015 CURSEG_HOT_DATA = 0, /* directory entry blocks */ 1016 CURSEG_WARM_DATA, /* data blocks */ 1017 CURSEG_COLD_DATA, /* multimedia or GCed data blocks */ 1018 CURSEG_HOT_NODE, /* direct node blocks of directory files */ 1019 CURSEG_WARM_NODE, /* direct node blocks of normal files */ 1020 CURSEG_COLD_NODE, /* indirect node blocks */ 1021 NR_PERSISTENT_LOG, /* number of persistent log */ 1022 CURSEG_COLD_DATA_PINNED = NR_PERSISTENT_LOG, 1023 /* pinned file that needs consecutive block address */ 1024 CURSEG_ALL_DATA_ATGC, /* SSR alloctor in hot/warm/cold data area */ 1025 NO_CHECK_TYPE, /* number of persistent & inmem log */ 1026 }; 1027 1028 struct flush_cmd { 1029 struct completion wait; 1030 struct llist_node llnode; 1031 nid_t ino; 1032 int ret; 1033 }; 1034 1035 struct flush_cmd_control { 1036 struct task_struct *f2fs_issue_flush; /* flush thread */ 1037 wait_queue_head_t flush_wait_queue; /* waiting queue for wake-up */ 1038 atomic_t issued_flush; /* # of issued flushes */ 1039 atomic_t queued_flush; /* # of queued flushes */ 1040 struct llist_head issue_list; /* list for command issue */ 1041 struct llist_node *dispatch_list; /* list for command dispatch */ 1042 }; 1043 1044 struct f2fs_sm_info { 1045 struct sit_info *sit_info; /* whole segment information */ 1046 struct free_segmap_info *free_info; /* free segment information */ 1047 struct dirty_seglist_info *dirty_info; /* dirty segment information */ 1048 struct curseg_info *curseg_array; /* active segment information */ 1049 1050 struct f2fs_rwsem curseg_lock; /* for preventing curseg change */ 1051 1052 block_t seg0_blkaddr; /* block address of 0'th segment */ 1053 block_t main_blkaddr; /* start block address of main area */ 1054 block_t ssa_blkaddr; /* start block address of SSA area */ 1055 1056 unsigned int segment_count; /* total # of segments */ 1057 unsigned int main_segments; /* # of segments in main area */ 1058 unsigned int reserved_segments; /* # of reserved segments */ 1059 unsigned int additional_reserved_segments;/* reserved segs for IO align feature */ 1060 unsigned int ovp_segments; /* # of overprovision segments */ 1061 1062 /* a threshold to reclaim prefree segments */ 1063 unsigned int rec_prefree_segments; 1064 1065 /* for batched trimming */ 1066 unsigned int trim_sections; /* # of sections to trim */ 1067 1068 struct list_head sit_entry_set; /* sit entry set list */ 1069 1070 unsigned int ipu_policy; /* in-place-update policy */ 1071 unsigned int min_ipu_util; /* in-place-update threshold */ 1072 unsigned int min_fsync_blocks; /* threshold for fsync */ 1073 unsigned int min_seq_blocks; /* threshold for sequential blocks */ 1074 unsigned int min_hot_blocks; /* threshold for hot block allocation */ 1075 unsigned int min_ssr_sections; /* threshold to trigger SSR allocation */ 1076 1077 /* for flush command control */ 1078 struct flush_cmd_control *fcc_info; 1079 1080 /* for discard command control */ 1081 struct discard_cmd_control *dcc_info; 1082 }; 1083 1084 /* 1085 * For superblock 1086 */ 1087 /* 1088 * COUNT_TYPE for monitoring 1089 * 1090 * f2fs monitors the number of several block types such as on-writeback, 1091 * dirty dentry blocks, dirty node blocks, and dirty meta blocks. 1092 */ 1093 #define WB_DATA_TYPE(p) (__is_cp_guaranteed(p) ? F2FS_WB_CP_DATA : F2FS_WB_DATA) 1094 enum count_type { 1095 F2FS_DIRTY_DENTS, 1096 F2FS_DIRTY_DATA, 1097 F2FS_DIRTY_QDATA, 1098 F2FS_DIRTY_NODES, 1099 F2FS_DIRTY_META, 1100 F2FS_DIRTY_IMETA, 1101 F2FS_WB_CP_DATA, 1102 F2FS_WB_DATA, 1103 F2FS_RD_DATA, 1104 F2FS_RD_NODE, 1105 F2FS_RD_META, 1106 F2FS_DIO_WRITE, 1107 F2FS_DIO_READ, 1108 NR_COUNT_TYPE, 1109 }; 1110 1111 /* 1112 * The below are the page types of bios used in submit_bio(). 1113 * The available types are: 1114 * DATA User data pages. It operates as async mode. 1115 * NODE Node pages. It operates as async mode. 1116 * META FS metadata pages such as SIT, NAT, CP. 1117 * NR_PAGE_TYPE The number of page types. 1118 * META_FLUSH Make sure the previous pages are written 1119 * with waiting the bio's completion 1120 * ... Only can be used with META. 1121 */ 1122 #define PAGE_TYPE_OF_BIO(type) ((type) > META ? META : (type)) 1123 enum page_type { 1124 DATA = 0, 1125 NODE = 1, /* should not change this */ 1126 META, 1127 NR_PAGE_TYPE, 1128 META_FLUSH, 1129 IPU, /* the below types are used by tracepoints only. */ 1130 OPU, 1131 }; 1132 1133 enum temp_type { 1134 HOT = 0, /* must be zero for meta bio */ 1135 WARM, 1136 COLD, 1137 NR_TEMP_TYPE, 1138 }; 1139 1140 enum need_lock_type { 1141 LOCK_REQ = 0, 1142 LOCK_DONE, 1143 LOCK_RETRY, 1144 }; 1145 1146 enum cp_reason_type { 1147 CP_NO_NEEDED, 1148 CP_NON_REGULAR, 1149 CP_COMPRESSED, 1150 CP_HARDLINK, 1151 CP_SB_NEED_CP, 1152 CP_WRONG_PINO, 1153 CP_NO_SPC_ROLL, 1154 CP_NODE_NEED_CP, 1155 CP_FASTBOOT_MODE, 1156 CP_SPEC_LOG_NUM, 1157 CP_RECOVER_DIR, 1158 }; 1159 1160 enum iostat_type { 1161 /* WRITE IO */ 1162 APP_DIRECT_IO, /* app direct write IOs */ 1163 APP_BUFFERED_IO, /* app buffered write IOs */ 1164 APP_WRITE_IO, /* app write IOs */ 1165 APP_MAPPED_IO, /* app mapped IOs */ 1166 APP_BUFFERED_CDATA_IO, /* app buffered write IOs on compressed file */ 1167 APP_MAPPED_CDATA_IO, /* app mapped write IOs on compressed file */ 1168 FS_DATA_IO, /* data IOs from kworker/fsync/reclaimer */ 1169 FS_CDATA_IO, /* data IOs from kworker/fsync/reclaimer on compressed file */ 1170 FS_NODE_IO, /* node IOs from kworker/fsync/reclaimer */ 1171 FS_META_IO, /* meta IOs from kworker/reclaimer */ 1172 FS_GC_DATA_IO, /* data IOs from forground gc */ 1173 FS_GC_NODE_IO, /* node IOs from forground gc */ 1174 FS_CP_DATA_IO, /* data IOs from checkpoint */ 1175 FS_CP_NODE_IO, /* node IOs from checkpoint */ 1176 FS_CP_META_IO, /* meta IOs from checkpoint */ 1177 1178 /* READ IO */ 1179 APP_DIRECT_READ_IO, /* app direct read IOs */ 1180 APP_BUFFERED_READ_IO, /* app buffered read IOs */ 1181 APP_READ_IO, /* app read IOs */ 1182 APP_MAPPED_READ_IO, /* app mapped read IOs */ 1183 APP_BUFFERED_CDATA_READ_IO, /* app buffered read IOs on compressed file */ 1184 APP_MAPPED_CDATA_READ_IO, /* app mapped read IOs on compressed file */ 1185 FS_DATA_READ_IO, /* data read IOs */ 1186 FS_GDATA_READ_IO, /* data read IOs from background gc */ 1187 FS_CDATA_READ_IO, /* compressed data read IOs */ 1188 FS_NODE_READ_IO, /* node read IOs */ 1189 FS_META_READ_IO, /* meta read IOs */ 1190 1191 /* other */ 1192 FS_DISCARD, /* discard */ 1193 NR_IO_TYPE, 1194 }; 1195 1196 struct f2fs_io_info { 1197 struct f2fs_sb_info *sbi; /* f2fs_sb_info pointer */ 1198 nid_t ino; /* inode number */ 1199 enum page_type type; /* contains DATA/NODE/META/META_FLUSH */ 1200 enum temp_type temp; /* contains HOT/WARM/COLD */ 1201 enum req_op op; /* contains REQ_OP_ */ 1202 blk_opf_t op_flags; /* req_flag_bits */ 1203 block_t new_blkaddr; /* new block address to be written */ 1204 block_t old_blkaddr; /* old block address before Cow */ 1205 struct page *page; /* page to be written */ 1206 struct page *encrypted_page; /* encrypted page */ 1207 struct page *compressed_page; /* compressed page */ 1208 struct list_head list; /* serialize IOs */ 1209 bool submitted; /* indicate IO submission */ 1210 int need_lock; /* indicate we need to lock cp_rwsem */ 1211 bool in_list; /* indicate fio is in io_list */ 1212 bool is_por; /* indicate IO is from recovery or not */ 1213 bool retry; /* need to reallocate block address */ 1214 int compr_blocks; /* # of compressed block addresses */ 1215 bool encrypted; /* indicate file is encrypted */ 1216 bool post_read; /* require post read */ 1217 enum iostat_type io_type; /* io type */ 1218 struct writeback_control *io_wbc; /* writeback control */ 1219 struct bio **bio; /* bio for ipu */ 1220 sector_t *last_block; /* last block number in bio */ 1221 unsigned char version; /* version of the node */ 1222 }; 1223 1224 struct bio_entry { 1225 struct bio *bio; 1226 struct list_head list; 1227 }; 1228 1229 #define is_read_io(rw) ((rw) == READ) 1230 struct f2fs_bio_info { 1231 struct f2fs_sb_info *sbi; /* f2fs superblock */ 1232 struct bio *bio; /* bios to merge */ 1233 sector_t last_block_in_bio; /* last block number */ 1234 struct f2fs_io_info fio; /* store buffered io info. */ 1235 struct f2fs_rwsem io_rwsem; /* blocking op for bio */ 1236 spinlock_t io_lock; /* serialize DATA/NODE IOs */ 1237 struct list_head io_list; /* track fios */ 1238 struct list_head bio_list; /* bio entry list head */ 1239 struct f2fs_rwsem bio_list_lock; /* lock to protect bio entry list */ 1240 }; 1241 1242 #define FDEV(i) (sbi->devs[i]) 1243 #define RDEV(i) (raw_super->devs[i]) 1244 struct f2fs_dev_info { 1245 struct block_device *bdev; 1246 char path[MAX_PATH_LEN]; 1247 unsigned int total_segments; 1248 block_t start_blk; 1249 block_t end_blk; 1250 #ifdef CONFIG_BLK_DEV_ZONED 1251 unsigned int nr_blkz; /* Total number of zones */ 1252 unsigned long *blkz_seq; /* Bitmap indicating sequential zones */ 1253 #endif 1254 }; 1255 1256 enum inode_type { 1257 DIR_INODE, /* for dirty dir inode */ 1258 FILE_INODE, /* for dirty regular/symlink inode */ 1259 DIRTY_META, /* for all dirtied inode metadata */ 1260 NR_INODE_TYPE, 1261 }; 1262 1263 /* for inner inode cache management */ 1264 struct inode_management { 1265 struct radix_tree_root ino_root; /* ino entry array */ 1266 spinlock_t ino_lock; /* for ino entry lock */ 1267 struct list_head ino_list; /* inode list head */ 1268 unsigned long ino_num; /* number of entries */ 1269 }; 1270 1271 /* for GC_AT */ 1272 struct atgc_management { 1273 bool atgc_enabled; /* ATGC is enabled or not */ 1274 struct rb_root_cached root; /* root of victim rb-tree */ 1275 struct list_head victim_list; /* linked with all victim entries */ 1276 unsigned int victim_count; /* victim count in rb-tree */ 1277 unsigned int candidate_ratio; /* candidate ratio */ 1278 unsigned int max_candidate_count; /* max candidate count */ 1279 unsigned int age_weight; /* age weight, vblock_weight = 100 - age_weight */ 1280 unsigned long long age_threshold; /* age threshold */ 1281 }; 1282 1283 struct f2fs_gc_control { 1284 unsigned int victim_segno; /* target victim segment number */ 1285 int init_gc_type; /* FG_GC or BG_GC */ 1286 bool no_bg_gc; /* check the space and stop bg_gc */ 1287 bool should_migrate_blocks; /* should migrate blocks */ 1288 bool err_gc_skipped; /* return EAGAIN if GC skipped */ 1289 unsigned int nr_free_secs; /* # of free sections to do GC */ 1290 }; 1291 1292 /* For s_flag in struct f2fs_sb_info */ 1293 enum { 1294 SBI_IS_DIRTY, /* dirty flag for checkpoint */ 1295 SBI_IS_CLOSE, /* specify unmounting */ 1296 SBI_NEED_FSCK, /* need fsck.f2fs to fix */ 1297 SBI_POR_DOING, /* recovery is doing or not */ 1298 SBI_NEED_SB_WRITE, /* need to recover superblock */ 1299 SBI_NEED_CP, /* need to checkpoint */ 1300 SBI_IS_SHUTDOWN, /* shutdown by ioctl */ 1301 SBI_IS_RECOVERED, /* recovered orphan/data */ 1302 SBI_CP_DISABLED, /* CP was disabled last mount */ 1303 SBI_CP_DISABLED_QUICK, /* CP was disabled quickly */ 1304 SBI_QUOTA_NEED_FLUSH, /* need to flush quota info in CP */ 1305 SBI_QUOTA_SKIP_FLUSH, /* skip flushing quota in current CP */ 1306 SBI_QUOTA_NEED_REPAIR, /* quota file may be corrupted */ 1307 SBI_IS_RESIZEFS, /* resizefs is in process */ 1308 SBI_IS_FREEZING, /* freezefs is in process */ 1309 }; 1310 1311 enum { 1312 CP_TIME, 1313 REQ_TIME, 1314 DISCARD_TIME, 1315 GC_TIME, 1316 DISABLE_TIME, 1317 UMOUNT_DISCARD_TIMEOUT, 1318 MAX_TIME, 1319 }; 1320 1321 enum { 1322 GC_NORMAL, 1323 GC_IDLE_CB, 1324 GC_IDLE_GREEDY, 1325 GC_IDLE_AT, 1326 GC_URGENT_HIGH, 1327 GC_URGENT_LOW, 1328 GC_URGENT_MID, 1329 MAX_GC_MODE, 1330 }; 1331 1332 enum { 1333 BGGC_MODE_ON, /* background gc is on */ 1334 BGGC_MODE_OFF, /* background gc is off */ 1335 BGGC_MODE_SYNC, /* 1336 * background gc is on, migrating blocks 1337 * like foreground gc 1338 */ 1339 }; 1340 1341 enum { 1342 FS_MODE_ADAPTIVE, /* use both lfs/ssr allocation */ 1343 FS_MODE_LFS, /* use lfs allocation only */ 1344 FS_MODE_FRAGMENT_SEG, /* segment fragmentation mode */ 1345 FS_MODE_FRAGMENT_BLK, /* block fragmentation mode */ 1346 }; 1347 1348 enum { 1349 ALLOC_MODE_DEFAULT, /* stay default */ 1350 ALLOC_MODE_REUSE, /* reuse segments as much as possible */ 1351 }; 1352 1353 enum fsync_mode { 1354 FSYNC_MODE_POSIX, /* fsync follows posix semantics */ 1355 FSYNC_MODE_STRICT, /* fsync behaves in line with ext4 */ 1356 FSYNC_MODE_NOBARRIER, /* fsync behaves nobarrier based on posix */ 1357 }; 1358 1359 enum { 1360 COMPR_MODE_FS, /* 1361 * automatically compress compression 1362 * enabled files 1363 */ 1364 COMPR_MODE_USER, /* 1365 * automatical compression is disabled. 1366 * user can control the file compression 1367 * using ioctls 1368 */ 1369 }; 1370 1371 enum { 1372 DISCARD_UNIT_BLOCK, /* basic discard unit is block */ 1373 DISCARD_UNIT_SEGMENT, /* basic discard unit is segment */ 1374 DISCARD_UNIT_SECTION, /* basic discard unit is section */ 1375 }; 1376 1377 enum { 1378 MEMORY_MODE_NORMAL, /* memory mode for normal devices */ 1379 MEMORY_MODE_LOW, /* memory mode for low memry devices */ 1380 }; 1381 1382 1383 1384 static inline int f2fs_test_bit(unsigned int nr, char *addr); 1385 static inline void f2fs_set_bit(unsigned int nr, char *addr); 1386 static inline void f2fs_clear_bit(unsigned int nr, char *addr); 1387 1388 /* 1389 * Layout of f2fs page.private: 1390 * 1391 * Layout A: lowest bit should be 1 1392 * | bit0 = 1 | bit1 | bit2 | ... | bit MAX | private data .... | 1393 * bit 0 PAGE_PRIVATE_NOT_POINTER 1394 * bit 1 PAGE_PRIVATE_ATOMIC_WRITE 1395 * bit 2 PAGE_PRIVATE_DUMMY_WRITE 1396 * bit 3 PAGE_PRIVATE_ONGOING_MIGRATION 1397 * bit 4 PAGE_PRIVATE_INLINE_INODE 1398 * bit 5 PAGE_PRIVATE_REF_RESOURCE 1399 * bit 6- f2fs private data 1400 * 1401 * Layout B: lowest bit should be 0 1402 * page.private is a wrapped pointer. 1403 */ 1404 enum { 1405 PAGE_PRIVATE_NOT_POINTER, /* private contains non-pointer data */ 1406 PAGE_PRIVATE_ATOMIC_WRITE, /* data page from atomic write path */ 1407 PAGE_PRIVATE_DUMMY_WRITE, /* data page for padding aligned IO */ 1408 PAGE_PRIVATE_ONGOING_MIGRATION, /* data page which is on-going migrating */ 1409 PAGE_PRIVATE_INLINE_INODE, /* inode page contains inline data */ 1410 PAGE_PRIVATE_REF_RESOURCE, /* dirty page has referenced resources */ 1411 PAGE_PRIVATE_MAX 1412 }; 1413 1414 #define PAGE_PRIVATE_GET_FUNC(name, flagname) \ 1415 static inline bool page_private_##name(struct page *page) \ 1416 { \ 1417 return PagePrivate(page) && \ 1418 test_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)) && \ 1419 test_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \ 1420 } 1421 1422 #define PAGE_PRIVATE_SET_FUNC(name, flagname) \ 1423 static inline void set_page_private_##name(struct page *page) \ 1424 { \ 1425 if (!PagePrivate(page)) { \ 1426 get_page(page); \ 1427 SetPagePrivate(page); \ 1428 set_page_private(page, 0); \ 1429 } \ 1430 set_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)); \ 1431 set_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \ 1432 } 1433 1434 #define PAGE_PRIVATE_CLEAR_FUNC(name, flagname) \ 1435 static inline void clear_page_private_##name(struct page *page) \ 1436 { \ 1437 clear_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \ 1438 if (page_private(page) == 1 << PAGE_PRIVATE_NOT_POINTER) { \ 1439 set_page_private(page, 0); \ 1440 if (PagePrivate(page)) { \ 1441 ClearPagePrivate(page); \ 1442 put_page(page); \ 1443 }\ 1444 } \ 1445 } 1446 1447 PAGE_PRIVATE_GET_FUNC(nonpointer, NOT_POINTER); 1448 PAGE_PRIVATE_GET_FUNC(reference, REF_RESOURCE); 1449 PAGE_PRIVATE_GET_FUNC(inline, INLINE_INODE); 1450 PAGE_PRIVATE_GET_FUNC(gcing, ONGOING_MIGRATION); 1451 PAGE_PRIVATE_GET_FUNC(atomic, ATOMIC_WRITE); 1452 PAGE_PRIVATE_GET_FUNC(dummy, DUMMY_WRITE); 1453 1454 PAGE_PRIVATE_SET_FUNC(reference, REF_RESOURCE); 1455 PAGE_PRIVATE_SET_FUNC(inline, INLINE_INODE); 1456 PAGE_PRIVATE_SET_FUNC(gcing, ONGOING_MIGRATION); 1457 PAGE_PRIVATE_SET_FUNC(atomic, ATOMIC_WRITE); 1458 PAGE_PRIVATE_SET_FUNC(dummy, DUMMY_WRITE); 1459 1460 PAGE_PRIVATE_CLEAR_FUNC(reference, REF_RESOURCE); 1461 PAGE_PRIVATE_CLEAR_FUNC(inline, INLINE_INODE); 1462 PAGE_PRIVATE_CLEAR_FUNC(gcing, ONGOING_MIGRATION); 1463 PAGE_PRIVATE_CLEAR_FUNC(atomic, ATOMIC_WRITE); 1464 PAGE_PRIVATE_CLEAR_FUNC(dummy, DUMMY_WRITE); 1465 1466 static inline unsigned long get_page_private_data(struct page *page) 1467 { 1468 unsigned long data = page_private(page); 1469 1470 if (!test_bit(PAGE_PRIVATE_NOT_POINTER, &data)) 1471 return 0; 1472 return data >> PAGE_PRIVATE_MAX; 1473 } 1474 1475 static inline void set_page_private_data(struct page *page, unsigned long data) 1476 { 1477 if (!PagePrivate(page)) { 1478 get_page(page); 1479 SetPagePrivate(page); 1480 set_page_private(page, 0); 1481 } 1482 set_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)); 1483 page_private(page) |= data << PAGE_PRIVATE_MAX; 1484 } 1485 1486 static inline void clear_page_private_data(struct page *page) 1487 { 1488 page_private(page) &= (1 << PAGE_PRIVATE_MAX) - 1; 1489 if (page_private(page) == 1 << PAGE_PRIVATE_NOT_POINTER) { 1490 set_page_private(page, 0); 1491 if (PagePrivate(page)) { 1492 ClearPagePrivate(page); 1493 put_page(page); 1494 } 1495 } 1496 } 1497 1498 /* For compression */ 1499 enum compress_algorithm_type { 1500 COMPRESS_LZO, 1501 COMPRESS_LZ4, 1502 COMPRESS_ZSTD, 1503 COMPRESS_LZORLE, 1504 COMPRESS_MAX, 1505 }; 1506 1507 enum compress_flag { 1508 COMPRESS_CHKSUM, 1509 COMPRESS_MAX_FLAG, 1510 }; 1511 1512 #define COMPRESS_WATERMARK 20 1513 #define COMPRESS_PERCENT 20 1514 1515 #define COMPRESS_DATA_RESERVED_SIZE 4 1516 struct compress_data { 1517 __le32 clen; /* compressed data size */ 1518 __le32 chksum; /* compressed data chksum */ 1519 __le32 reserved[COMPRESS_DATA_RESERVED_SIZE]; /* reserved */ 1520 u8 cdata[]; /* compressed data */ 1521 }; 1522 1523 #define COMPRESS_HEADER_SIZE (sizeof(struct compress_data)) 1524 1525 #define F2FS_COMPRESSED_PAGE_MAGIC 0xF5F2C000 1526 1527 #define COMPRESS_LEVEL_OFFSET 8 1528 1529 /* compress context */ 1530 struct compress_ctx { 1531 struct inode *inode; /* inode the context belong to */ 1532 pgoff_t cluster_idx; /* cluster index number */ 1533 unsigned int cluster_size; /* page count in cluster */ 1534 unsigned int log_cluster_size; /* log of cluster size */ 1535 struct page **rpages; /* pages store raw data in cluster */ 1536 unsigned int nr_rpages; /* total page number in rpages */ 1537 struct page **cpages; /* pages store compressed data in cluster */ 1538 unsigned int nr_cpages; /* total page number in cpages */ 1539 unsigned int valid_nr_cpages; /* valid page number in cpages */ 1540 void *rbuf; /* virtual mapped address on rpages */ 1541 struct compress_data *cbuf; /* virtual mapped address on cpages */ 1542 size_t rlen; /* valid data length in rbuf */ 1543 size_t clen; /* valid data length in cbuf */ 1544 void *private; /* payload buffer for specified compression algorithm */ 1545 void *private2; /* extra payload buffer */ 1546 }; 1547 1548 /* compress context for write IO path */ 1549 struct compress_io_ctx { 1550 u32 magic; /* magic number to indicate page is compressed */ 1551 struct inode *inode; /* inode the context belong to */ 1552 struct page **rpages; /* pages store raw data in cluster */ 1553 unsigned int nr_rpages; /* total page number in rpages */ 1554 atomic_t pending_pages; /* in-flight compressed page count */ 1555 }; 1556 1557 /* Context for decompressing one cluster on the read IO path */ 1558 struct decompress_io_ctx { 1559 u32 magic; /* magic number to indicate page is compressed */ 1560 struct inode *inode; /* inode the context belong to */ 1561 pgoff_t cluster_idx; /* cluster index number */ 1562 unsigned int cluster_size; /* page count in cluster */ 1563 unsigned int log_cluster_size; /* log of cluster size */ 1564 struct page **rpages; /* pages store raw data in cluster */ 1565 unsigned int nr_rpages; /* total page number in rpages */ 1566 struct page **cpages; /* pages store compressed data in cluster */ 1567 unsigned int nr_cpages; /* total page number in cpages */ 1568 struct page **tpages; /* temp pages to pad holes in cluster */ 1569 void *rbuf; /* virtual mapped address on rpages */ 1570 struct compress_data *cbuf; /* virtual mapped address on cpages */ 1571 size_t rlen; /* valid data length in rbuf */ 1572 size_t clen; /* valid data length in cbuf */ 1573 1574 /* 1575 * The number of compressed pages remaining to be read in this cluster. 1576 * This is initially nr_cpages. It is decremented by 1 each time a page 1577 * has been read (or failed to be read). When it reaches 0, the cluster 1578 * is decompressed (or an error is reported). 1579 * 1580 * If an error occurs before all the pages have been submitted for I/O, 1581 * then this will never reach 0. In this case the I/O submitter is 1582 * responsible for calling f2fs_decompress_end_io() instead. 1583 */ 1584 atomic_t remaining_pages; 1585 1586 /* 1587 * Number of references to this decompress_io_ctx. 1588 * 1589 * One reference is held for I/O completion. This reference is dropped 1590 * after the pagecache pages are updated and unlocked -- either after 1591 * decompression (and verity if enabled), or after an error. 1592 * 1593 * In addition, each compressed page holds a reference while it is in a 1594 * bio. These references are necessary prevent compressed pages from 1595 * being freed while they are still in a bio. 1596 */ 1597 refcount_t refcnt; 1598 1599 bool failed; /* IO error occurred before decompression? */ 1600 bool need_verity; /* need fs-verity verification after decompression? */ 1601 void *private; /* payload buffer for specified decompression algorithm */ 1602 void *private2; /* extra payload buffer */ 1603 struct work_struct verity_work; /* work to verify the decompressed pages */ 1604 struct work_struct free_work; /* work for late free this structure itself */ 1605 }; 1606 1607 #define NULL_CLUSTER ((unsigned int)(~0)) 1608 #define MIN_COMPRESS_LOG_SIZE 2 1609 #define MAX_COMPRESS_LOG_SIZE 8 1610 #define MAX_COMPRESS_WINDOW_SIZE(log_size) ((PAGE_SIZE) << (log_size)) 1611 1612 struct f2fs_sb_info { 1613 struct super_block *sb; /* pointer to VFS super block */ 1614 struct proc_dir_entry *s_proc; /* proc entry */ 1615 struct f2fs_super_block *raw_super; /* raw super block pointer */ 1616 struct f2fs_rwsem sb_lock; /* lock for raw super block */ 1617 int valid_super_block; /* valid super block no */ 1618 unsigned long s_flag; /* flags for sbi */ 1619 struct mutex writepages; /* mutex for writepages() */ 1620 1621 #ifdef CONFIG_BLK_DEV_ZONED 1622 unsigned int blocks_per_blkz; /* F2FS blocks per zone */ 1623 unsigned int log_blocks_per_blkz; /* log2 F2FS blocks per zone */ 1624 #endif 1625 1626 /* for node-related operations */ 1627 struct f2fs_nm_info *nm_info; /* node manager */ 1628 struct inode *node_inode; /* cache node blocks */ 1629 1630 /* for segment-related operations */ 1631 struct f2fs_sm_info *sm_info; /* segment manager */ 1632 1633 /* for bio operations */ 1634 struct f2fs_bio_info *write_io[NR_PAGE_TYPE]; /* for write bios */ 1635 /* keep migration IO order for LFS mode */ 1636 struct f2fs_rwsem io_order_lock; 1637 mempool_t *write_io_dummy; /* Dummy pages */ 1638 pgoff_t page_eio_ofs[NR_PAGE_TYPE]; /* EIO page offset */ 1639 int page_eio_cnt[NR_PAGE_TYPE]; /* EIO count */ 1640 1641 /* for checkpoint */ 1642 struct f2fs_checkpoint *ckpt; /* raw checkpoint pointer */ 1643 int cur_cp_pack; /* remain current cp pack */ 1644 spinlock_t cp_lock; /* for flag in ckpt */ 1645 struct inode *meta_inode; /* cache meta blocks */ 1646 struct f2fs_rwsem cp_global_sem; /* checkpoint procedure lock */ 1647 struct f2fs_rwsem cp_rwsem; /* blocking FS operations */ 1648 struct f2fs_rwsem node_write; /* locking node writes */ 1649 struct f2fs_rwsem node_change; /* locking node change */ 1650 wait_queue_head_t cp_wait; 1651 unsigned long last_time[MAX_TIME]; /* to store time in jiffies */ 1652 long interval_time[MAX_TIME]; /* to store thresholds */ 1653 struct ckpt_req_control cprc_info; /* for checkpoint request control */ 1654 1655 struct inode_management im[MAX_INO_ENTRY]; /* manage inode cache */ 1656 1657 spinlock_t fsync_node_lock; /* for node entry lock */ 1658 struct list_head fsync_node_list; /* node list head */ 1659 unsigned int fsync_seg_id; /* sequence id */ 1660 unsigned int fsync_node_num; /* number of node entries */ 1661 1662 /* for orphan inode, use 0'th array */ 1663 unsigned int max_orphans; /* max orphan inodes */ 1664 1665 /* for inode management */ 1666 struct list_head inode_list[NR_INODE_TYPE]; /* dirty inode list */ 1667 spinlock_t inode_lock[NR_INODE_TYPE]; /* for dirty inode list lock */ 1668 struct mutex flush_lock; /* for flush exclusion */ 1669 1670 /* for extent tree cache */ 1671 struct radix_tree_root extent_tree_root;/* cache extent cache entries */ 1672 struct mutex extent_tree_lock; /* locking extent radix tree */ 1673 struct list_head extent_list; /* lru list for shrinker */ 1674 spinlock_t extent_lock; /* locking extent lru list */ 1675 atomic_t total_ext_tree; /* extent tree count */ 1676 struct list_head zombie_list; /* extent zombie tree list */ 1677 atomic_t total_zombie_tree; /* extent zombie tree count */ 1678 atomic_t total_ext_node; /* extent info count */ 1679 1680 /* basic filesystem units */ 1681 unsigned int log_sectors_per_block; /* log2 sectors per block */ 1682 unsigned int log_blocksize; /* log2 block size */ 1683 unsigned int blocksize; /* block size */ 1684 unsigned int root_ino_num; /* root inode number*/ 1685 unsigned int node_ino_num; /* node inode number*/ 1686 unsigned int meta_ino_num; /* meta inode number*/ 1687 unsigned int log_blocks_per_seg; /* log2 blocks per segment */ 1688 unsigned int blocks_per_seg; /* blocks per segment */ 1689 unsigned int unusable_blocks_per_sec; /* unusable blocks per section */ 1690 unsigned int segs_per_sec; /* segments per section */ 1691 unsigned int secs_per_zone; /* sections per zone */ 1692 unsigned int total_sections; /* total section count */ 1693 unsigned int total_node_count; /* total node block count */ 1694 unsigned int total_valid_node_count; /* valid node block count */ 1695 int dir_level; /* directory level */ 1696 int readdir_ra; /* readahead inode in readdir */ 1697 u64 max_io_bytes; /* max io bytes to merge IOs */ 1698 1699 block_t user_block_count; /* # of user blocks */ 1700 block_t total_valid_block_count; /* # of valid blocks */ 1701 block_t discard_blks; /* discard command candidats */ 1702 block_t last_valid_block_count; /* for recovery */ 1703 block_t reserved_blocks; /* configurable reserved blocks */ 1704 block_t current_reserved_blocks; /* current reserved blocks */ 1705 1706 /* Additional tracking for no checkpoint mode */ 1707 block_t unusable_block_count; /* # of blocks saved by last cp */ 1708 1709 unsigned int nquota_files; /* # of quota sysfile */ 1710 struct f2fs_rwsem quota_sem; /* blocking cp for flags */ 1711 1712 /* # of pages, see count_type */ 1713 atomic_t nr_pages[NR_COUNT_TYPE]; 1714 /* # of allocated blocks */ 1715 struct percpu_counter alloc_valid_block_count; 1716 /* # of node block writes as roll forward recovery */ 1717 struct percpu_counter rf_node_block_count; 1718 1719 /* writeback control */ 1720 atomic_t wb_sync_req[META]; /* count # of WB_SYNC threads */ 1721 1722 /* valid inode count */ 1723 struct percpu_counter total_valid_inode_count; 1724 1725 struct f2fs_mount_info mount_opt; /* mount options */ 1726 1727 /* for cleaning operations */ 1728 struct f2fs_rwsem gc_lock; /* 1729 * semaphore for GC, avoid 1730 * race between GC and GC or CP 1731 */ 1732 struct f2fs_gc_kthread *gc_thread; /* GC thread */ 1733 struct atgc_management am; /* atgc management */ 1734 unsigned int cur_victim_sec; /* current victim section num */ 1735 unsigned int gc_mode; /* current GC state */ 1736 unsigned int next_victim_seg[2]; /* next segment in victim section */ 1737 spinlock_t gc_urgent_high_lock; 1738 unsigned int gc_urgent_high_remaining; /* remaining trial count for GC_URGENT_HIGH */ 1739 1740 /* for skip statistic */ 1741 unsigned long long skipped_gc_rwsem; /* FG_GC only */ 1742 1743 /* threshold for gc trials on pinned files */ 1744 u64 gc_pin_file_threshold; 1745 struct f2fs_rwsem pin_sem; 1746 1747 /* maximum # of trials to find a victim segment for SSR and GC */ 1748 unsigned int max_victim_search; 1749 /* migration granularity of garbage collection, unit: segment */ 1750 unsigned int migration_granularity; 1751 1752 /* 1753 * for stat information. 1754 * one is for the LFS mode, and the other is for the SSR mode. 1755 */ 1756 #ifdef CONFIG_F2FS_STAT_FS 1757 struct f2fs_stat_info *stat_info; /* FS status information */ 1758 atomic_t meta_count[META_MAX]; /* # of meta blocks */ 1759 unsigned int segment_count[2]; /* # of allocated segments */ 1760 unsigned int block_count[2]; /* # of allocated blocks */ 1761 atomic_t inplace_count; /* # of inplace update */ 1762 atomic64_t total_hit_ext; /* # of lookup extent cache */ 1763 atomic64_t read_hit_rbtree; /* # of hit rbtree extent node */ 1764 atomic64_t read_hit_largest; /* # of hit largest extent node */ 1765 atomic64_t read_hit_cached; /* # of hit cached extent node */ 1766 atomic_t inline_xattr; /* # of inline_xattr inodes */ 1767 atomic_t inline_inode; /* # of inline_data inodes */ 1768 atomic_t inline_dir; /* # of inline_dentry inodes */ 1769 atomic_t compr_inode; /* # of compressed inodes */ 1770 atomic64_t compr_blocks; /* # of compressed blocks */ 1771 atomic_t swapfile_inode; /* # of swapfile inodes */ 1772 atomic_t atomic_files; /* # of opened atomic file */ 1773 atomic_t max_aw_cnt; /* max # of atomic writes */ 1774 unsigned int io_skip_bggc; /* skip background gc for in-flight IO */ 1775 unsigned int other_skip_bggc; /* skip background gc for other reasons */ 1776 unsigned int ndirty_inode[NR_INODE_TYPE]; /* # of dirty inodes */ 1777 #endif 1778 spinlock_t stat_lock; /* lock for stat operations */ 1779 1780 /* to attach REQ_META|REQ_FUA flags */ 1781 unsigned int data_io_flag; 1782 unsigned int node_io_flag; 1783 1784 /* For sysfs support */ 1785 struct kobject s_kobj; /* /sys/fs/f2fs/<devname> */ 1786 struct completion s_kobj_unregister; 1787 1788 struct kobject s_stat_kobj; /* /sys/fs/f2fs/<devname>/stat */ 1789 struct completion s_stat_kobj_unregister; 1790 1791 struct kobject s_feature_list_kobj; /* /sys/fs/f2fs/<devname>/feature_list */ 1792 struct completion s_feature_list_kobj_unregister; 1793 1794 /* For shrinker support */ 1795 struct list_head s_list; 1796 struct mutex umount_mutex; 1797 unsigned int shrinker_run_no; 1798 1799 /* For multi devices */ 1800 int s_ndevs; /* number of devices */ 1801 struct f2fs_dev_info *devs; /* for device list */ 1802 unsigned int dirty_device; /* for checkpoint data flush */ 1803 spinlock_t dev_lock; /* protect dirty_device */ 1804 bool aligned_blksize; /* all devices has the same logical blksize */ 1805 1806 /* For write statistics */ 1807 u64 sectors_written_start; 1808 u64 kbytes_written; 1809 1810 /* Reference to checksum algorithm driver via cryptoapi */ 1811 struct crypto_shash *s_chksum_driver; 1812 1813 /* Precomputed FS UUID checksum for seeding other checksums */ 1814 __u32 s_chksum_seed; 1815 1816 struct workqueue_struct *post_read_wq; /* post read workqueue */ 1817 1818 unsigned char errors[MAX_F2FS_ERRORS]; /* error flags */ 1819 spinlock_t error_lock; /* protect errors array */ 1820 bool error_dirty; /* errors of sb is dirty */ 1821 1822 struct kmem_cache *inline_xattr_slab; /* inline xattr entry */ 1823 unsigned int inline_xattr_slab_size; /* default inline xattr slab size */ 1824 1825 /* For reclaimed segs statistics per each GC mode */ 1826 unsigned int gc_segment_mode; /* GC state for reclaimed segments */ 1827 unsigned int gc_reclaimed_segs[MAX_GC_MODE]; /* Reclaimed segs for each mode */ 1828 1829 unsigned long seq_file_ra_mul; /* multiplier for ra_pages of seq. files in fadvise */ 1830 1831 int max_fragment_chunk; /* max chunk size for block fragmentation mode */ 1832 int max_fragment_hole; /* max hole size for block fragmentation mode */ 1833 1834 /* For atomic write statistics */ 1835 atomic64_t current_atomic_write; 1836 s64 peak_atomic_write; 1837 u64 committed_atomic_block; 1838 u64 revoked_atomic_block; 1839 1840 #ifdef CONFIG_F2FS_FS_COMPRESSION 1841 struct kmem_cache *page_array_slab; /* page array entry */ 1842 unsigned int page_array_slab_size; /* default page array slab size */ 1843 1844 /* For runtime compression statistics */ 1845 u64 compr_written_block; 1846 u64 compr_saved_block; 1847 u32 compr_new_inode; 1848 1849 /* For compressed block cache */ 1850 struct inode *compress_inode; /* cache compressed blocks */ 1851 unsigned int compress_percent; /* cache page percentage */ 1852 unsigned int compress_watermark; /* cache page watermark */ 1853 atomic_t compress_page_hit; /* cache hit count */ 1854 #endif 1855 1856 #ifdef CONFIG_F2FS_IOSTAT 1857 /* For app/fs IO statistics */ 1858 spinlock_t iostat_lock; 1859 unsigned long long rw_iostat[NR_IO_TYPE]; 1860 unsigned long long prev_rw_iostat[NR_IO_TYPE]; 1861 bool iostat_enable; 1862 unsigned long iostat_next_period; 1863 unsigned int iostat_period_ms; 1864 1865 /* For io latency related statistics info in one iostat period */ 1866 spinlock_t iostat_lat_lock; 1867 struct iostat_lat_info *iostat_io_lat; 1868 #endif 1869 }; 1870 1871 #ifdef CONFIG_F2FS_FAULT_INJECTION 1872 #define f2fs_show_injection_info(sbi, type) \ 1873 printk_ratelimited("%sF2FS-fs (%s) : inject %s in %s of %pS\n", \ 1874 KERN_INFO, sbi->sb->s_id, \ 1875 f2fs_fault_name[type], \ 1876 __func__, __builtin_return_address(0)) 1877 static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type) 1878 { 1879 struct f2fs_fault_info *ffi = &F2FS_OPTION(sbi).fault_info; 1880 1881 if (!ffi->inject_rate) 1882 return false; 1883 1884 if (!IS_FAULT_SET(ffi, type)) 1885 return false; 1886 1887 atomic_inc(&ffi->inject_ops); 1888 if (atomic_read(&ffi->inject_ops) >= ffi->inject_rate) { 1889 atomic_set(&ffi->inject_ops, 0); 1890 return true; 1891 } 1892 return false; 1893 } 1894 #else 1895 #define f2fs_show_injection_info(sbi, type) do { } while (0) 1896 static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type) 1897 { 1898 return false; 1899 } 1900 #endif 1901 1902 /* 1903 * Test if the mounted volume is a multi-device volume. 1904 * - For a single regular disk volume, sbi->s_ndevs is 0. 1905 * - For a single zoned disk volume, sbi->s_ndevs is 1. 1906 * - For a multi-device volume, sbi->s_ndevs is always 2 or more. 1907 */ 1908 static inline bool f2fs_is_multi_device(struct f2fs_sb_info *sbi) 1909 { 1910 return sbi->s_ndevs > 1; 1911 } 1912 1913 static inline void f2fs_update_time(struct f2fs_sb_info *sbi, int type) 1914 { 1915 unsigned long now = jiffies; 1916 1917 sbi->last_time[type] = now; 1918 1919 /* DISCARD_TIME and GC_TIME are based on REQ_TIME */ 1920 if (type == REQ_TIME) { 1921 sbi->last_time[DISCARD_TIME] = now; 1922 sbi->last_time[GC_TIME] = now; 1923 } 1924 } 1925 1926 static inline bool f2fs_time_over(struct f2fs_sb_info *sbi, int type) 1927 { 1928 unsigned long interval = sbi->interval_time[type] * HZ; 1929 1930 return time_after(jiffies, sbi->last_time[type] + interval); 1931 } 1932 1933 static inline unsigned int f2fs_time_to_wait(struct f2fs_sb_info *sbi, 1934 int type) 1935 { 1936 unsigned long interval = sbi->interval_time[type] * HZ; 1937 unsigned int wait_ms = 0; 1938 long delta; 1939 1940 delta = (sbi->last_time[type] + interval) - jiffies; 1941 if (delta > 0) 1942 wait_ms = jiffies_to_msecs(delta); 1943 1944 return wait_ms; 1945 } 1946 1947 /* 1948 * Inline functions 1949 */ 1950 static inline u32 __f2fs_crc32(struct f2fs_sb_info *sbi, u32 crc, 1951 const void *address, unsigned int length) 1952 { 1953 struct { 1954 struct shash_desc shash; 1955 char ctx[4]; 1956 } desc; 1957 int err; 1958 1959 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver) != sizeof(desc.ctx)); 1960 1961 desc.shash.tfm = sbi->s_chksum_driver; 1962 *(u32 *)desc.ctx = crc; 1963 1964 err = crypto_shash_update(&desc.shash, address, length); 1965 BUG_ON(err); 1966 1967 return *(u32 *)desc.ctx; 1968 } 1969 1970 static inline u32 f2fs_crc32(struct f2fs_sb_info *sbi, const void *address, 1971 unsigned int length) 1972 { 1973 return __f2fs_crc32(sbi, F2FS_SUPER_MAGIC, address, length); 1974 } 1975 1976 static inline bool f2fs_crc_valid(struct f2fs_sb_info *sbi, __u32 blk_crc, 1977 void *buf, size_t buf_size) 1978 { 1979 return f2fs_crc32(sbi, buf, buf_size) == blk_crc; 1980 } 1981 1982 static inline u32 f2fs_chksum(struct f2fs_sb_info *sbi, u32 crc, 1983 const void *address, unsigned int length) 1984 { 1985 return __f2fs_crc32(sbi, crc, address, length); 1986 } 1987 1988 static inline struct f2fs_inode_info *F2FS_I(struct inode *inode) 1989 { 1990 return container_of(inode, struct f2fs_inode_info, vfs_inode); 1991 } 1992 1993 static inline struct f2fs_sb_info *F2FS_SB(struct super_block *sb) 1994 { 1995 return sb->s_fs_info; 1996 } 1997 1998 static inline struct f2fs_sb_info *F2FS_I_SB(struct inode *inode) 1999 { 2000 return F2FS_SB(inode->i_sb); 2001 } 2002 2003 static inline struct f2fs_sb_info *F2FS_M_SB(struct address_space *mapping) 2004 { 2005 return F2FS_I_SB(mapping->host); 2006 } 2007 2008 static inline struct f2fs_sb_info *F2FS_P_SB(struct page *page) 2009 { 2010 return F2FS_M_SB(page_file_mapping(page)); 2011 } 2012 2013 static inline struct f2fs_super_block *F2FS_RAW_SUPER(struct f2fs_sb_info *sbi) 2014 { 2015 return (struct f2fs_super_block *)(sbi->raw_super); 2016 } 2017 2018 static inline struct f2fs_checkpoint *F2FS_CKPT(struct f2fs_sb_info *sbi) 2019 { 2020 return (struct f2fs_checkpoint *)(sbi->ckpt); 2021 } 2022 2023 static inline struct f2fs_node *F2FS_NODE(struct page *page) 2024 { 2025 return (struct f2fs_node *)page_address(page); 2026 } 2027 2028 static inline struct f2fs_inode *F2FS_INODE(struct page *page) 2029 { 2030 return &((struct f2fs_node *)page_address(page))->i; 2031 } 2032 2033 static inline struct f2fs_nm_info *NM_I(struct f2fs_sb_info *sbi) 2034 { 2035 return (struct f2fs_nm_info *)(sbi->nm_info); 2036 } 2037 2038 static inline struct f2fs_sm_info *SM_I(struct f2fs_sb_info *sbi) 2039 { 2040 return (struct f2fs_sm_info *)(sbi->sm_info); 2041 } 2042 2043 static inline struct sit_info *SIT_I(struct f2fs_sb_info *sbi) 2044 { 2045 return (struct sit_info *)(SM_I(sbi)->sit_info); 2046 } 2047 2048 static inline struct free_segmap_info *FREE_I(struct f2fs_sb_info *sbi) 2049 { 2050 return (struct free_segmap_info *)(SM_I(sbi)->free_info); 2051 } 2052 2053 static inline struct dirty_seglist_info *DIRTY_I(struct f2fs_sb_info *sbi) 2054 { 2055 return (struct dirty_seglist_info *)(SM_I(sbi)->dirty_info); 2056 } 2057 2058 static inline struct address_space *META_MAPPING(struct f2fs_sb_info *sbi) 2059 { 2060 return sbi->meta_inode->i_mapping; 2061 } 2062 2063 static inline struct address_space *NODE_MAPPING(struct f2fs_sb_info *sbi) 2064 { 2065 return sbi->node_inode->i_mapping; 2066 } 2067 2068 static inline bool is_sbi_flag_set(struct f2fs_sb_info *sbi, unsigned int type) 2069 { 2070 return test_bit(type, &sbi->s_flag); 2071 } 2072 2073 static inline void set_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type) 2074 { 2075 set_bit(type, &sbi->s_flag); 2076 } 2077 2078 static inline void clear_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type) 2079 { 2080 clear_bit(type, &sbi->s_flag); 2081 } 2082 2083 static inline unsigned long long cur_cp_version(struct f2fs_checkpoint *cp) 2084 { 2085 return le64_to_cpu(cp->checkpoint_ver); 2086 } 2087 2088 static inline unsigned long f2fs_qf_ino(struct super_block *sb, int type) 2089 { 2090 if (type < F2FS_MAX_QUOTAS) 2091 return le32_to_cpu(F2FS_SB(sb)->raw_super->qf_ino[type]); 2092 return 0; 2093 } 2094 2095 static inline __u64 cur_cp_crc(struct f2fs_checkpoint *cp) 2096 { 2097 size_t crc_offset = le32_to_cpu(cp->checksum_offset); 2098 return le32_to_cpu(*((__le32 *)((unsigned char *)cp + crc_offset))); 2099 } 2100 2101 static inline bool __is_set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 2102 { 2103 unsigned int ckpt_flags = le32_to_cpu(cp->ckpt_flags); 2104 2105 return ckpt_flags & f; 2106 } 2107 2108 static inline bool is_set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 2109 { 2110 return __is_set_ckpt_flags(F2FS_CKPT(sbi), f); 2111 } 2112 2113 static inline void __set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 2114 { 2115 unsigned int ckpt_flags; 2116 2117 ckpt_flags = le32_to_cpu(cp->ckpt_flags); 2118 ckpt_flags |= f; 2119 cp->ckpt_flags = cpu_to_le32(ckpt_flags); 2120 } 2121 2122 static inline void set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 2123 { 2124 unsigned long flags; 2125 2126 spin_lock_irqsave(&sbi->cp_lock, flags); 2127 __set_ckpt_flags(F2FS_CKPT(sbi), f); 2128 spin_unlock_irqrestore(&sbi->cp_lock, flags); 2129 } 2130 2131 static inline void __clear_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 2132 { 2133 unsigned int ckpt_flags; 2134 2135 ckpt_flags = le32_to_cpu(cp->ckpt_flags); 2136 ckpt_flags &= (~f); 2137 cp->ckpt_flags = cpu_to_le32(ckpt_flags); 2138 } 2139 2140 static inline void clear_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 2141 { 2142 unsigned long flags; 2143 2144 spin_lock_irqsave(&sbi->cp_lock, flags); 2145 __clear_ckpt_flags(F2FS_CKPT(sbi), f); 2146 spin_unlock_irqrestore(&sbi->cp_lock, flags); 2147 } 2148 2149 #define init_f2fs_rwsem(sem) \ 2150 do { \ 2151 static struct lock_class_key __key; \ 2152 \ 2153 __init_f2fs_rwsem((sem), #sem, &__key); \ 2154 } while (0) 2155 2156 static inline void __init_f2fs_rwsem(struct f2fs_rwsem *sem, 2157 const char *sem_name, struct lock_class_key *key) 2158 { 2159 __init_rwsem(&sem->internal_rwsem, sem_name, key); 2160 #ifdef CONFIG_F2FS_UNFAIR_RWSEM 2161 init_waitqueue_head(&sem->read_waiters); 2162 #endif 2163 } 2164 2165 static inline int f2fs_rwsem_is_locked(struct f2fs_rwsem *sem) 2166 { 2167 return rwsem_is_locked(&sem->internal_rwsem); 2168 } 2169 2170 static inline int f2fs_rwsem_is_contended(struct f2fs_rwsem *sem) 2171 { 2172 return rwsem_is_contended(&sem->internal_rwsem); 2173 } 2174 2175 static inline void f2fs_down_read(struct f2fs_rwsem *sem) 2176 { 2177 #ifdef CONFIG_F2FS_UNFAIR_RWSEM 2178 wait_event(sem->read_waiters, down_read_trylock(&sem->internal_rwsem)); 2179 #else 2180 down_read(&sem->internal_rwsem); 2181 #endif 2182 } 2183 2184 static inline int f2fs_down_read_trylock(struct f2fs_rwsem *sem) 2185 { 2186 return down_read_trylock(&sem->internal_rwsem); 2187 } 2188 2189 #ifdef CONFIG_DEBUG_LOCK_ALLOC 2190 static inline void f2fs_down_read_nested(struct f2fs_rwsem *sem, int subclass) 2191 { 2192 down_read_nested(&sem->internal_rwsem, subclass); 2193 } 2194 #else 2195 #define f2fs_down_read_nested(sem, subclass) f2fs_down_read(sem) 2196 #endif 2197 2198 static inline void f2fs_up_read(struct f2fs_rwsem *sem) 2199 { 2200 up_read(&sem->internal_rwsem); 2201 } 2202 2203 static inline void f2fs_down_write(struct f2fs_rwsem *sem) 2204 { 2205 down_write(&sem->internal_rwsem); 2206 } 2207 2208 static inline int f2fs_down_write_trylock(struct f2fs_rwsem *sem) 2209 { 2210 return down_write_trylock(&sem->internal_rwsem); 2211 } 2212 2213 static inline void f2fs_up_write(struct f2fs_rwsem *sem) 2214 { 2215 up_write(&sem->internal_rwsem); 2216 #ifdef CONFIG_F2FS_UNFAIR_RWSEM 2217 wake_up_all(&sem->read_waiters); 2218 #endif 2219 } 2220 2221 static inline void f2fs_lock_op(struct f2fs_sb_info *sbi) 2222 { 2223 f2fs_down_read(&sbi->cp_rwsem); 2224 } 2225 2226 static inline int f2fs_trylock_op(struct f2fs_sb_info *sbi) 2227 { 2228 if (time_to_inject(sbi, FAULT_LOCK_OP)) { 2229 f2fs_show_injection_info(sbi, FAULT_LOCK_OP); 2230 return 0; 2231 } 2232 return f2fs_down_read_trylock(&sbi->cp_rwsem); 2233 } 2234 2235 static inline void f2fs_unlock_op(struct f2fs_sb_info *sbi) 2236 { 2237 f2fs_up_read(&sbi->cp_rwsem); 2238 } 2239 2240 static inline void f2fs_lock_all(struct f2fs_sb_info *sbi) 2241 { 2242 f2fs_down_write(&sbi->cp_rwsem); 2243 } 2244 2245 static inline void f2fs_unlock_all(struct f2fs_sb_info *sbi) 2246 { 2247 f2fs_up_write(&sbi->cp_rwsem); 2248 } 2249 2250 static inline int __get_cp_reason(struct f2fs_sb_info *sbi) 2251 { 2252 int reason = CP_SYNC; 2253 2254 if (test_opt(sbi, FASTBOOT)) 2255 reason = CP_FASTBOOT; 2256 if (is_sbi_flag_set(sbi, SBI_IS_CLOSE)) 2257 reason = CP_UMOUNT; 2258 return reason; 2259 } 2260 2261 static inline bool __remain_node_summaries(int reason) 2262 { 2263 return (reason & (CP_UMOUNT | CP_FASTBOOT)); 2264 } 2265 2266 static inline bool __exist_node_summaries(struct f2fs_sb_info *sbi) 2267 { 2268 return (is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG) || 2269 is_set_ckpt_flags(sbi, CP_FASTBOOT_FLAG)); 2270 } 2271 2272 /* 2273 * Check whether the inode has blocks or not 2274 */ 2275 static inline int F2FS_HAS_BLOCKS(struct inode *inode) 2276 { 2277 block_t xattr_block = F2FS_I(inode)->i_xattr_nid ? 1 : 0; 2278 2279 return (inode->i_blocks >> F2FS_LOG_SECTORS_PER_BLOCK) > xattr_block; 2280 } 2281 2282 static inline bool f2fs_has_xattr_block(unsigned int ofs) 2283 { 2284 return ofs == XATTR_NODE_OFFSET; 2285 } 2286 2287 static inline bool __allow_reserved_blocks(struct f2fs_sb_info *sbi, 2288 struct inode *inode, bool cap) 2289 { 2290 if (!inode) 2291 return true; 2292 if (!test_opt(sbi, RESERVE_ROOT)) 2293 return false; 2294 if (IS_NOQUOTA(inode)) 2295 return true; 2296 if (uid_eq(F2FS_OPTION(sbi).s_resuid, current_fsuid())) 2297 return true; 2298 if (!gid_eq(F2FS_OPTION(sbi).s_resgid, GLOBAL_ROOT_GID) && 2299 in_group_p(F2FS_OPTION(sbi).s_resgid)) 2300 return true; 2301 if (cap && capable(CAP_SYS_RESOURCE)) 2302 return true; 2303 return false; 2304 } 2305 2306 static inline void f2fs_i_blocks_write(struct inode *, block_t, bool, bool); 2307 static inline int inc_valid_block_count(struct f2fs_sb_info *sbi, 2308 struct inode *inode, blkcnt_t *count) 2309 { 2310 blkcnt_t diff = 0, release = 0; 2311 block_t avail_user_block_count; 2312 int ret; 2313 2314 ret = dquot_reserve_block(inode, *count); 2315 if (ret) 2316 return ret; 2317 2318 if (time_to_inject(sbi, FAULT_BLOCK)) { 2319 f2fs_show_injection_info(sbi, FAULT_BLOCK); 2320 release = *count; 2321 goto release_quota; 2322 } 2323 2324 /* 2325 * let's increase this in prior to actual block count change in order 2326 * for f2fs_sync_file to avoid data races when deciding checkpoint. 2327 */ 2328 percpu_counter_add(&sbi->alloc_valid_block_count, (*count)); 2329 2330 spin_lock(&sbi->stat_lock); 2331 sbi->total_valid_block_count += (block_t)(*count); 2332 avail_user_block_count = sbi->user_block_count - 2333 sbi->current_reserved_blocks; 2334 2335 if (!__allow_reserved_blocks(sbi, inode, true)) 2336 avail_user_block_count -= F2FS_OPTION(sbi).root_reserved_blocks; 2337 2338 if (F2FS_IO_ALIGNED(sbi)) 2339 avail_user_block_count -= sbi->blocks_per_seg * 2340 SM_I(sbi)->additional_reserved_segments; 2341 2342 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) { 2343 if (avail_user_block_count > sbi->unusable_block_count) 2344 avail_user_block_count -= sbi->unusable_block_count; 2345 else 2346 avail_user_block_count = 0; 2347 } 2348 if (unlikely(sbi->total_valid_block_count > avail_user_block_count)) { 2349 diff = sbi->total_valid_block_count - avail_user_block_count; 2350 if (diff > *count) 2351 diff = *count; 2352 *count -= diff; 2353 release = diff; 2354 sbi->total_valid_block_count -= diff; 2355 if (!*count) { 2356 spin_unlock(&sbi->stat_lock); 2357 goto enospc; 2358 } 2359 } 2360 spin_unlock(&sbi->stat_lock); 2361 2362 if (unlikely(release)) { 2363 percpu_counter_sub(&sbi->alloc_valid_block_count, release); 2364 dquot_release_reservation_block(inode, release); 2365 } 2366 f2fs_i_blocks_write(inode, *count, true, true); 2367 return 0; 2368 2369 enospc: 2370 percpu_counter_sub(&sbi->alloc_valid_block_count, release); 2371 release_quota: 2372 dquot_release_reservation_block(inode, release); 2373 return -ENOSPC; 2374 } 2375 2376 __printf(2, 3) 2377 void f2fs_printk(struct f2fs_sb_info *sbi, const char *fmt, ...); 2378 2379 #define f2fs_err(sbi, fmt, ...) \ 2380 f2fs_printk(sbi, KERN_ERR fmt, ##__VA_ARGS__) 2381 #define f2fs_warn(sbi, fmt, ...) \ 2382 f2fs_printk(sbi, KERN_WARNING fmt, ##__VA_ARGS__) 2383 #define f2fs_notice(sbi, fmt, ...) \ 2384 f2fs_printk(sbi, KERN_NOTICE fmt, ##__VA_ARGS__) 2385 #define f2fs_info(sbi, fmt, ...) \ 2386 f2fs_printk(sbi, KERN_INFO fmt, ##__VA_ARGS__) 2387 #define f2fs_debug(sbi, fmt, ...) \ 2388 f2fs_printk(sbi, KERN_DEBUG fmt, ##__VA_ARGS__) 2389 2390 static inline void dec_valid_block_count(struct f2fs_sb_info *sbi, 2391 struct inode *inode, 2392 block_t count) 2393 { 2394 blkcnt_t sectors = count << F2FS_LOG_SECTORS_PER_BLOCK; 2395 2396 spin_lock(&sbi->stat_lock); 2397 f2fs_bug_on(sbi, sbi->total_valid_block_count < (block_t) count); 2398 sbi->total_valid_block_count -= (block_t)count; 2399 if (sbi->reserved_blocks && 2400 sbi->current_reserved_blocks < sbi->reserved_blocks) 2401 sbi->current_reserved_blocks = min(sbi->reserved_blocks, 2402 sbi->current_reserved_blocks + count); 2403 spin_unlock(&sbi->stat_lock); 2404 if (unlikely(inode->i_blocks < sectors)) { 2405 f2fs_warn(sbi, "Inconsistent i_blocks, ino:%lu, iblocks:%llu, sectors:%llu", 2406 inode->i_ino, 2407 (unsigned long long)inode->i_blocks, 2408 (unsigned long long)sectors); 2409 set_sbi_flag(sbi, SBI_NEED_FSCK); 2410 return; 2411 } 2412 f2fs_i_blocks_write(inode, count, false, true); 2413 } 2414 2415 static inline void inc_page_count(struct f2fs_sb_info *sbi, int count_type) 2416 { 2417 atomic_inc(&sbi->nr_pages[count_type]); 2418 2419 if (count_type == F2FS_DIRTY_DENTS || 2420 count_type == F2FS_DIRTY_NODES || 2421 count_type == F2FS_DIRTY_META || 2422 count_type == F2FS_DIRTY_QDATA || 2423 count_type == F2FS_DIRTY_IMETA) 2424 set_sbi_flag(sbi, SBI_IS_DIRTY); 2425 } 2426 2427 static inline void inode_inc_dirty_pages(struct inode *inode) 2428 { 2429 atomic_inc(&F2FS_I(inode)->dirty_pages); 2430 inc_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ? 2431 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA); 2432 if (IS_NOQUOTA(inode)) 2433 inc_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA); 2434 } 2435 2436 static inline void dec_page_count(struct f2fs_sb_info *sbi, int count_type) 2437 { 2438 atomic_dec(&sbi->nr_pages[count_type]); 2439 } 2440 2441 static inline void inode_dec_dirty_pages(struct inode *inode) 2442 { 2443 if (!S_ISDIR(inode->i_mode) && !S_ISREG(inode->i_mode) && 2444 !S_ISLNK(inode->i_mode)) 2445 return; 2446 2447 atomic_dec(&F2FS_I(inode)->dirty_pages); 2448 dec_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ? 2449 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA); 2450 if (IS_NOQUOTA(inode)) 2451 dec_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA); 2452 } 2453 2454 static inline void inc_atomic_write_cnt(struct inode *inode) 2455 { 2456 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 2457 struct f2fs_inode_info *fi = F2FS_I(inode); 2458 u64 current_write; 2459 2460 fi->atomic_write_cnt++; 2461 atomic64_inc(&sbi->current_atomic_write); 2462 current_write = atomic64_read(&sbi->current_atomic_write); 2463 if (current_write > sbi->peak_atomic_write) 2464 sbi->peak_atomic_write = current_write; 2465 } 2466 2467 static inline void release_atomic_write_cnt(struct inode *inode) 2468 { 2469 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 2470 struct f2fs_inode_info *fi = F2FS_I(inode); 2471 2472 atomic64_sub(fi->atomic_write_cnt, &sbi->current_atomic_write); 2473 fi->atomic_write_cnt = 0; 2474 } 2475 2476 static inline s64 get_pages(struct f2fs_sb_info *sbi, int count_type) 2477 { 2478 return atomic_read(&sbi->nr_pages[count_type]); 2479 } 2480 2481 static inline int get_dirty_pages(struct inode *inode) 2482 { 2483 return atomic_read(&F2FS_I(inode)->dirty_pages); 2484 } 2485 2486 static inline int get_blocktype_secs(struct f2fs_sb_info *sbi, int block_type) 2487 { 2488 unsigned int pages_per_sec = sbi->segs_per_sec * sbi->blocks_per_seg; 2489 unsigned int segs = (get_pages(sbi, block_type) + pages_per_sec - 1) >> 2490 sbi->log_blocks_per_seg; 2491 2492 return segs / sbi->segs_per_sec; 2493 } 2494 2495 static inline block_t valid_user_blocks(struct f2fs_sb_info *sbi) 2496 { 2497 return sbi->total_valid_block_count; 2498 } 2499 2500 static inline block_t discard_blocks(struct f2fs_sb_info *sbi) 2501 { 2502 return sbi->discard_blks; 2503 } 2504 2505 static inline unsigned long __bitmap_size(struct f2fs_sb_info *sbi, int flag) 2506 { 2507 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi); 2508 2509 /* return NAT or SIT bitmap */ 2510 if (flag == NAT_BITMAP) 2511 return le32_to_cpu(ckpt->nat_ver_bitmap_bytesize); 2512 else if (flag == SIT_BITMAP) 2513 return le32_to_cpu(ckpt->sit_ver_bitmap_bytesize); 2514 2515 return 0; 2516 } 2517 2518 static inline block_t __cp_payload(struct f2fs_sb_info *sbi) 2519 { 2520 return le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_payload); 2521 } 2522 2523 static inline void *__bitmap_ptr(struct f2fs_sb_info *sbi, int flag) 2524 { 2525 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi); 2526 void *tmp_ptr = &ckpt->sit_nat_version_bitmap; 2527 int offset; 2528 2529 if (is_set_ckpt_flags(sbi, CP_LARGE_NAT_BITMAP_FLAG)) { 2530 offset = (flag == SIT_BITMAP) ? 2531 le32_to_cpu(ckpt->nat_ver_bitmap_bytesize) : 0; 2532 /* 2533 * if large_nat_bitmap feature is enabled, leave checksum 2534 * protection for all nat/sit bitmaps. 2535 */ 2536 return tmp_ptr + offset + sizeof(__le32); 2537 } 2538 2539 if (__cp_payload(sbi) > 0) { 2540 if (flag == NAT_BITMAP) 2541 return tmp_ptr; 2542 else 2543 return (unsigned char *)ckpt + F2FS_BLKSIZE; 2544 } else { 2545 offset = (flag == NAT_BITMAP) ? 2546 le32_to_cpu(ckpt->sit_ver_bitmap_bytesize) : 0; 2547 return tmp_ptr + offset; 2548 } 2549 } 2550 2551 static inline block_t __start_cp_addr(struct f2fs_sb_info *sbi) 2552 { 2553 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr); 2554 2555 if (sbi->cur_cp_pack == 2) 2556 start_addr += sbi->blocks_per_seg; 2557 return start_addr; 2558 } 2559 2560 static inline block_t __start_cp_next_addr(struct f2fs_sb_info *sbi) 2561 { 2562 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr); 2563 2564 if (sbi->cur_cp_pack == 1) 2565 start_addr += sbi->blocks_per_seg; 2566 return start_addr; 2567 } 2568 2569 static inline void __set_cp_next_pack(struct f2fs_sb_info *sbi) 2570 { 2571 sbi->cur_cp_pack = (sbi->cur_cp_pack == 1) ? 2 : 1; 2572 } 2573 2574 static inline block_t __start_sum_addr(struct f2fs_sb_info *sbi) 2575 { 2576 return le32_to_cpu(F2FS_CKPT(sbi)->cp_pack_start_sum); 2577 } 2578 2579 static inline int inc_valid_node_count(struct f2fs_sb_info *sbi, 2580 struct inode *inode, bool is_inode) 2581 { 2582 block_t valid_block_count; 2583 unsigned int valid_node_count, user_block_count; 2584 int err; 2585 2586 if (is_inode) { 2587 if (inode) { 2588 err = dquot_alloc_inode(inode); 2589 if (err) 2590 return err; 2591 } 2592 } else { 2593 err = dquot_reserve_block(inode, 1); 2594 if (err) 2595 return err; 2596 } 2597 2598 if (time_to_inject(sbi, FAULT_BLOCK)) { 2599 f2fs_show_injection_info(sbi, FAULT_BLOCK); 2600 goto enospc; 2601 } 2602 2603 spin_lock(&sbi->stat_lock); 2604 2605 valid_block_count = sbi->total_valid_block_count + 2606 sbi->current_reserved_blocks + 1; 2607 2608 if (!__allow_reserved_blocks(sbi, inode, false)) 2609 valid_block_count += F2FS_OPTION(sbi).root_reserved_blocks; 2610 2611 if (F2FS_IO_ALIGNED(sbi)) 2612 valid_block_count += sbi->blocks_per_seg * 2613 SM_I(sbi)->additional_reserved_segments; 2614 2615 user_block_count = sbi->user_block_count; 2616 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) 2617 user_block_count -= sbi->unusable_block_count; 2618 2619 if (unlikely(valid_block_count > user_block_count)) { 2620 spin_unlock(&sbi->stat_lock); 2621 goto enospc; 2622 } 2623 2624 valid_node_count = sbi->total_valid_node_count + 1; 2625 if (unlikely(valid_node_count > sbi->total_node_count)) { 2626 spin_unlock(&sbi->stat_lock); 2627 goto enospc; 2628 } 2629 2630 sbi->total_valid_node_count++; 2631 sbi->total_valid_block_count++; 2632 spin_unlock(&sbi->stat_lock); 2633 2634 if (inode) { 2635 if (is_inode) 2636 f2fs_mark_inode_dirty_sync(inode, true); 2637 else 2638 f2fs_i_blocks_write(inode, 1, true, true); 2639 } 2640 2641 percpu_counter_inc(&sbi->alloc_valid_block_count); 2642 return 0; 2643 2644 enospc: 2645 if (is_inode) { 2646 if (inode) 2647 dquot_free_inode(inode); 2648 } else { 2649 dquot_release_reservation_block(inode, 1); 2650 } 2651 return -ENOSPC; 2652 } 2653 2654 static inline void dec_valid_node_count(struct f2fs_sb_info *sbi, 2655 struct inode *inode, bool is_inode) 2656 { 2657 spin_lock(&sbi->stat_lock); 2658 2659 if (unlikely(!sbi->total_valid_block_count || 2660 !sbi->total_valid_node_count)) { 2661 f2fs_warn(sbi, "dec_valid_node_count: inconsistent block counts, total_valid_block:%u, total_valid_node:%u", 2662 sbi->total_valid_block_count, 2663 sbi->total_valid_node_count); 2664 set_sbi_flag(sbi, SBI_NEED_FSCK); 2665 } else { 2666 sbi->total_valid_block_count--; 2667 sbi->total_valid_node_count--; 2668 } 2669 2670 if (sbi->reserved_blocks && 2671 sbi->current_reserved_blocks < sbi->reserved_blocks) 2672 sbi->current_reserved_blocks++; 2673 2674 spin_unlock(&sbi->stat_lock); 2675 2676 if (is_inode) { 2677 dquot_free_inode(inode); 2678 } else { 2679 if (unlikely(inode->i_blocks == 0)) { 2680 f2fs_warn(sbi, "dec_valid_node_count: inconsistent i_blocks, ino:%lu, iblocks:%llu", 2681 inode->i_ino, 2682 (unsigned long long)inode->i_blocks); 2683 set_sbi_flag(sbi, SBI_NEED_FSCK); 2684 return; 2685 } 2686 f2fs_i_blocks_write(inode, 1, false, true); 2687 } 2688 } 2689 2690 static inline unsigned int valid_node_count(struct f2fs_sb_info *sbi) 2691 { 2692 return sbi->total_valid_node_count; 2693 } 2694 2695 static inline void inc_valid_inode_count(struct f2fs_sb_info *sbi) 2696 { 2697 percpu_counter_inc(&sbi->total_valid_inode_count); 2698 } 2699 2700 static inline void dec_valid_inode_count(struct f2fs_sb_info *sbi) 2701 { 2702 percpu_counter_dec(&sbi->total_valid_inode_count); 2703 } 2704 2705 static inline s64 valid_inode_count(struct f2fs_sb_info *sbi) 2706 { 2707 return percpu_counter_sum_positive(&sbi->total_valid_inode_count); 2708 } 2709 2710 static inline struct page *f2fs_grab_cache_page(struct address_space *mapping, 2711 pgoff_t index, bool for_write) 2712 { 2713 struct page *page; 2714 unsigned int flags; 2715 2716 if (IS_ENABLED(CONFIG_F2FS_FAULT_INJECTION)) { 2717 if (!for_write) 2718 page = find_get_page_flags(mapping, index, 2719 FGP_LOCK | FGP_ACCESSED); 2720 else 2721 page = find_lock_page(mapping, index); 2722 if (page) 2723 return page; 2724 2725 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_ALLOC)) { 2726 f2fs_show_injection_info(F2FS_M_SB(mapping), 2727 FAULT_PAGE_ALLOC); 2728 return NULL; 2729 } 2730 } 2731 2732 if (!for_write) 2733 return grab_cache_page(mapping, index); 2734 2735 flags = memalloc_nofs_save(); 2736 page = grab_cache_page_write_begin(mapping, index); 2737 memalloc_nofs_restore(flags); 2738 2739 return page; 2740 } 2741 2742 static inline struct page *f2fs_pagecache_get_page( 2743 struct address_space *mapping, pgoff_t index, 2744 int fgp_flags, gfp_t gfp_mask) 2745 { 2746 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_GET)) { 2747 f2fs_show_injection_info(F2FS_M_SB(mapping), FAULT_PAGE_GET); 2748 return NULL; 2749 } 2750 2751 return pagecache_get_page(mapping, index, fgp_flags, gfp_mask); 2752 } 2753 2754 static inline void f2fs_put_page(struct page *page, int unlock) 2755 { 2756 if (!page) 2757 return; 2758 2759 if (unlock) { 2760 f2fs_bug_on(F2FS_P_SB(page), !PageLocked(page)); 2761 unlock_page(page); 2762 } 2763 put_page(page); 2764 } 2765 2766 static inline void f2fs_put_dnode(struct dnode_of_data *dn) 2767 { 2768 if (dn->node_page) 2769 f2fs_put_page(dn->node_page, 1); 2770 if (dn->inode_page && dn->node_page != dn->inode_page) 2771 f2fs_put_page(dn->inode_page, 0); 2772 dn->node_page = NULL; 2773 dn->inode_page = NULL; 2774 } 2775 2776 static inline struct kmem_cache *f2fs_kmem_cache_create(const char *name, 2777 size_t size) 2778 { 2779 return kmem_cache_create(name, size, 0, SLAB_RECLAIM_ACCOUNT, NULL); 2780 } 2781 2782 static inline void *f2fs_kmem_cache_alloc_nofail(struct kmem_cache *cachep, 2783 gfp_t flags) 2784 { 2785 void *entry; 2786 2787 entry = kmem_cache_alloc(cachep, flags); 2788 if (!entry) 2789 entry = kmem_cache_alloc(cachep, flags | __GFP_NOFAIL); 2790 return entry; 2791 } 2792 2793 static inline void *f2fs_kmem_cache_alloc(struct kmem_cache *cachep, 2794 gfp_t flags, bool nofail, struct f2fs_sb_info *sbi) 2795 { 2796 if (nofail) 2797 return f2fs_kmem_cache_alloc_nofail(cachep, flags); 2798 2799 if (time_to_inject(sbi, FAULT_SLAB_ALLOC)) { 2800 f2fs_show_injection_info(sbi, FAULT_SLAB_ALLOC); 2801 return NULL; 2802 } 2803 2804 return kmem_cache_alloc(cachep, flags); 2805 } 2806 2807 static inline bool is_inflight_io(struct f2fs_sb_info *sbi, int type) 2808 { 2809 if (get_pages(sbi, F2FS_RD_DATA) || get_pages(sbi, F2FS_RD_NODE) || 2810 get_pages(sbi, F2FS_RD_META) || get_pages(sbi, F2FS_WB_DATA) || 2811 get_pages(sbi, F2FS_WB_CP_DATA) || 2812 get_pages(sbi, F2FS_DIO_READ) || 2813 get_pages(sbi, F2FS_DIO_WRITE)) 2814 return true; 2815 2816 if (type != DISCARD_TIME && SM_I(sbi) && SM_I(sbi)->dcc_info && 2817 atomic_read(&SM_I(sbi)->dcc_info->queued_discard)) 2818 return true; 2819 2820 if (SM_I(sbi) && SM_I(sbi)->fcc_info && 2821 atomic_read(&SM_I(sbi)->fcc_info->queued_flush)) 2822 return true; 2823 return false; 2824 } 2825 2826 static inline bool is_idle(struct f2fs_sb_info *sbi, int type) 2827 { 2828 if (sbi->gc_mode == GC_URGENT_HIGH) 2829 return true; 2830 2831 if (is_inflight_io(sbi, type)) 2832 return false; 2833 2834 if (sbi->gc_mode == GC_URGENT_MID) 2835 return true; 2836 2837 if (sbi->gc_mode == GC_URGENT_LOW && 2838 (type == DISCARD_TIME || type == GC_TIME)) 2839 return true; 2840 2841 return f2fs_time_over(sbi, type); 2842 } 2843 2844 static inline void f2fs_radix_tree_insert(struct radix_tree_root *root, 2845 unsigned long index, void *item) 2846 { 2847 while (radix_tree_insert(root, index, item)) 2848 cond_resched(); 2849 } 2850 2851 #define RAW_IS_INODE(p) ((p)->footer.nid == (p)->footer.ino) 2852 2853 static inline bool IS_INODE(struct page *page) 2854 { 2855 struct f2fs_node *p = F2FS_NODE(page); 2856 2857 return RAW_IS_INODE(p); 2858 } 2859 2860 static inline int offset_in_addr(struct f2fs_inode *i) 2861 { 2862 return (i->i_inline & F2FS_EXTRA_ATTR) ? 2863 (le16_to_cpu(i->i_extra_isize) / sizeof(__le32)) : 0; 2864 } 2865 2866 static inline __le32 *blkaddr_in_node(struct f2fs_node *node) 2867 { 2868 return RAW_IS_INODE(node) ? node->i.i_addr : node->dn.addr; 2869 } 2870 2871 static inline int f2fs_has_extra_attr(struct inode *inode); 2872 static inline block_t data_blkaddr(struct inode *inode, 2873 struct page *node_page, unsigned int offset) 2874 { 2875 struct f2fs_node *raw_node; 2876 __le32 *addr_array; 2877 int base = 0; 2878 bool is_inode = IS_INODE(node_page); 2879 2880 raw_node = F2FS_NODE(node_page); 2881 2882 if (is_inode) { 2883 if (!inode) 2884 /* from GC path only */ 2885 base = offset_in_addr(&raw_node->i); 2886 else if (f2fs_has_extra_attr(inode)) 2887 base = get_extra_isize(inode); 2888 } 2889 2890 addr_array = blkaddr_in_node(raw_node); 2891 return le32_to_cpu(addr_array[base + offset]); 2892 } 2893 2894 static inline block_t f2fs_data_blkaddr(struct dnode_of_data *dn) 2895 { 2896 return data_blkaddr(dn->inode, dn->node_page, dn->ofs_in_node); 2897 } 2898 2899 static inline int f2fs_test_bit(unsigned int nr, char *addr) 2900 { 2901 int mask; 2902 2903 addr += (nr >> 3); 2904 mask = 1 << (7 - (nr & 0x07)); 2905 return mask & *addr; 2906 } 2907 2908 static inline void f2fs_set_bit(unsigned int nr, char *addr) 2909 { 2910 int mask; 2911 2912 addr += (nr >> 3); 2913 mask = 1 << (7 - (nr & 0x07)); 2914 *addr |= mask; 2915 } 2916 2917 static inline void f2fs_clear_bit(unsigned int nr, char *addr) 2918 { 2919 int mask; 2920 2921 addr += (nr >> 3); 2922 mask = 1 << (7 - (nr & 0x07)); 2923 *addr &= ~mask; 2924 } 2925 2926 static inline int f2fs_test_and_set_bit(unsigned int nr, char *addr) 2927 { 2928 int mask; 2929 int ret; 2930 2931 addr += (nr >> 3); 2932 mask = 1 << (7 - (nr & 0x07)); 2933 ret = mask & *addr; 2934 *addr |= mask; 2935 return ret; 2936 } 2937 2938 static inline int f2fs_test_and_clear_bit(unsigned int nr, char *addr) 2939 { 2940 int mask; 2941 int ret; 2942 2943 addr += (nr >> 3); 2944 mask = 1 << (7 - (nr & 0x07)); 2945 ret = mask & *addr; 2946 *addr &= ~mask; 2947 return ret; 2948 } 2949 2950 static inline void f2fs_change_bit(unsigned int nr, char *addr) 2951 { 2952 int mask; 2953 2954 addr += (nr >> 3); 2955 mask = 1 << (7 - (nr & 0x07)); 2956 *addr ^= mask; 2957 } 2958 2959 /* 2960 * On-disk inode flags (f2fs_inode::i_flags) 2961 */ 2962 #define F2FS_COMPR_FL 0x00000004 /* Compress file */ 2963 #define F2FS_SYNC_FL 0x00000008 /* Synchronous updates */ 2964 #define F2FS_IMMUTABLE_FL 0x00000010 /* Immutable file */ 2965 #define F2FS_APPEND_FL 0x00000020 /* writes to file may only append */ 2966 #define F2FS_NODUMP_FL 0x00000040 /* do not dump file */ 2967 #define F2FS_NOATIME_FL 0x00000080 /* do not update atime */ 2968 #define F2FS_NOCOMP_FL 0x00000400 /* Don't compress */ 2969 #define F2FS_INDEX_FL 0x00001000 /* hash-indexed directory */ 2970 #define F2FS_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */ 2971 #define F2FS_PROJINHERIT_FL 0x20000000 /* Create with parents projid */ 2972 #define F2FS_CASEFOLD_FL 0x40000000 /* Casefolded file */ 2973 2974 /* Flags that should be inherited by new inodes from their parent. */ 2975 #define F2FS_FL_INHERITED (F2FS_SYNC_FL | F2FS_NODUMP_FL | F2FS_NOATIME_FL | \ 2976 F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \ 2977 F2FS_CASEFOLD_FL | F2FS_COMPR_FL | F2FS_NOCOMP_FL) 2978 2979 /* Flags that are appropriate for regular files (all but dir-specific ones). */ 2980 #define F2FS_REG_FLMASK (~(F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \ 2981 F2FS_CASEFOLD_FL)) 2982 2983 /* Flags that are appropriate for non-directories/regular files. */ 2984 #define F2FS_OTHER_FLMASK (F2FS_NODUMP_FL | F2FS_NOATIME_FL) 2985 2986 static inline __u32 f2fs_mask_flags(umode_t mode, __u32 flags) 2987 { 2988 if (S_ISDIR(mode)) 2989 return flags; 2990 else if (S_ISREG(mode)) 2991 return flags & F2FS_REG_FLMASK; 2992 else 2993 return flags & F2FS_OTHER_FLMASK; 2994 } 2995 2996 static inline void __mark_inode_dirty_flag(struct inode *inode, 2997 int flag, bool set) 2998 { 2999 switch (flag) { 3000 case FI_INLINE_XATTR: 3001 case FI_INLINE_DATA: 3002 case FI_INLINE_DENTRY: 3003 case FI_NEW_INODE: 3004 if (set) 3005 return; 3006 fallthrough; 3007 case FI_DATA_EXIST: 3008 case FI_INLINE_DOTS: 3009 case FI_PIN_FILE: 3010 case FI_COMPRESS_RELEASED: 3011 f2fs_mark_inode_dirty_sync(inode, true); 3012 } 3013 } 3014 3015 static inline void set_inode_flag(struct inode *inode, int flag) 3016 { 3017 set_bit(flag, F2FS_I(inode)->flags); 3018 __mark_inode_dirty_flag(inode, flag, true); 3019 } 3020 3021 static inline int is_inode_flag_set(struct inode *inode, int flag) 3022 { 3023 return test_bit(flag, F2FS_I(inode)->flags); 3024 } 3025 3026 static inline void clear_inode_flag(struct inode *inode, int flag) 3027 { 3028 clear_bit(flag, F2FS_I(inode)->flags); 3029 __mark_inode_dirty_flag(inode, flag, false); 3030 } 3031 3032 static inline bool f2fs_verity_in_progress(struct inode *inode) 3033 { 3034 return IS_ENABLED(CONFIG_FS_VERITY) && 3035 is_inode_flag_set(inode, FI_VERITY_IN_PROGRESS); 3036 } 3037 3038 static inline void set_acl_inode(struct inode *inode, umode_t mode) 3039 { 3040 F2FS_I(inode)->i_acl_mode = mode; 3041 set_inode_flag(inode, FI_ACL_MODE); 3042 f2fs_mark_inode_dirty_sync(inode, false); 3043 } 3044 3045 static inline void f2fs_i_links_write(struct inode *inode, bool inc) 3046 { 3047 if (inc) 3048 inc_nlink(inode); 3049 else 3050 drop_nlink(inode); 3051 f2fs_mark_inode_dirty_sync(inode, true); 3052 } 3053 3054 static inline void f2fs_i_blocks_write(struct inode *inode, 3055 block_t diff, bool add, bool claim) 3056 { 3057 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE); 3058 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER); 3059 3060 /* add = 1, claim = 1 should be dquot_reserve_block in pair */ 3061 if (add) { 3062 if (claim) 3063 dquot_claim_block(inode, diff); 3064 else 3065 dquot_alloc_block_nofail(inode, diff); 3066 } else { 3067 dquot_free_block(inode, diff); 3068 } 3069 3070 f2fs_mark_inode_dirty_sync(inode, true); 3071 if (clean || recover) 3072 set_inode_flag(inode, FI_AUTO_RECOVER); 3073 } 3074 3075 static inline void f2fs_i_size_write(struct inode *inode, loff_t i_size) 3076 { 3077 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE); 3078 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER); 3079 3080 if (i_size_read(inode) == i_size) 3081 return; 3082 3083 i_size_write(inode, i_size); 3084 f2fs_mark_inode_dirty_sync(inode, true); 3085 if (clean || recover) 3086 set_inode_flag(inode, FI_AUTO_RECOVER); 3087 } 3088 3089 static inline void f2fs_i_depth_write(struct inode *inode, unsigned int depth) 3090 { 3091 F2FS_I(inode)->i_current_depth = depth; 3092 f2fs_mark_inode_dirty_sync(inode, true); 3093 } 3094 3095 static inline void f2fs_i_gc_failures_write(struct inode *inode, 3096 unsigned int count) 3097 { 3098 F2FS_I(inode)->i_gc_failures[GC_FAILURE_PIN] = count; 3099 f2fs_mark_inode_dirty_sync(inode, true); 3100 } 3101 3102 static inline void f2fs_i_xnid_write(struct inode *inode, nid_t xnid) 3103 { 3104 F2FS_I(inode)->i_xattr_nid = xnid; 3105 f2fs_mark_inode_dirty_sync(inode, true); 3106 } 3107 3108 static inline void f2fs_i_pino_write(struct inode *inode, nid_t pino) 3109 { 3110 F2FS_I(inode)->i_pino = pino; 3111 f2fs_mark_inode_dirty_sync(inode, true); 3112 } 3113 3114 static inline void get_inline_info(struct inode *inode, struct f2fs_inode *ri) 3115 { 3116 struct f2fs_inode_info *fi = F2FS_I(inode); 3117 3118 if (ri->i_inline & F2FS_INLINE_XATTR) 3119 set_bit(FI_INLINE_XATTR, fi->flags); 3120 if (ri->i_inline & F2FS_INLINE_DATA) 3121 set_bit(FI_INLINE_DATA, fi->flags); 3122 if (ri->i_inline & F2FS_INLINE_DENTRY) 3123 set_bit(FI_INLINE_DENTRY, fi->flags); 3124 if (ri->i_inline & F2FS_DATA_EXIST) 3125 set_bit(FI_DATA_EXIST, fi->flags); 3126 if (ri->i_inline & F2FS_INLINE_DOTS) 3127 set_bit(FI_INLINE_DOTS, fi->flags); 3128 if (ri->i_inline & F2FS_EXTRA_ATTR) 3129 set_bit(FI_EXTRA_ATTR, fi->flags); 3130 if (ri->i_inline & F2FS_PIN_FILE) 3131 set_bit(FI_PIN_FILE, fi->flags); 3132 if (ri->i_inline & F2FS_COMPRESS_RELEASED) 3133 set_bit(FI_COMPRESS_RELEASED, fi->flags); 3134 } 3135 3136 static inline void set_raw_inline(struct inode *inode, struct f2fs_inode *ri) 3137 { 3138 ri->i_inline = 0; 3139 3140 if (is_inode_flag_set(inode, FI_INLINE_XATTR)) 3141 ri->i_inline |= F2FS_INLINE_XATTR; 3142 if (is_inode_flag_set(inode, FI_INLINE_DATA)) 3143 ri->i_inline |= F2FS_INLINE_DATA; 3144 if (is_inode_flag_set(inode, FI_INLINE_DENTRY)) 3145 ri->i_inline |= F2FS_INLINE_DENTRY; 3146 if (is_inode_flag_set(inode, FI_DATA_EXIST)) 3147 ri->i_inline |= F2FS_DATA_EXIST; 3148 if (is_inode_flag_set(inode, FI_INLINE_DOTS)) 3149 ri->i_inline |= F2FS_INLINE_DOTS; 3150 if (is_inode_flag_set(inode, FI_EXTRA_ATTR)) 3151 ri->i_inline |= F2FS_EXTRA_ATTR; 3152 if (is_inode_flag_set(inode, FI_PIN_FILE)) 3153 ri->i_inline |= F2FS_PIN_FILE; 3154 if (is_inode_flag_set(inode, FI_COMPRESS_RELEASED)) 3155 ri->i_inline |= F2FS_COMPRESS_RELEASED; 3156 } 3157 3158 static inline int f2fs_has_extra_attr(struct inode *inode) 3159 { 3160 return is_inode_flag_set(inode, FI_EXTRA_ATTR); 3161 } 3162 3163 static inline int f2fs_has_inline_xattr(struct inode *inode) 3164 { 3165 return is_inode_flag_set(inode, FI_INLINE_XATTR); 3166 } 3167 3168 static inline int f2fs_compressed_file(struct inode *inode) 3169 { 3170 return S_ISREG(inode->i_mode) && 3171 is_inode_flag_set(inode, FI_COMPRESSED_FILE); 3172 } 3173 3174 static inline bool f2fs_need_compress_data(struct inode *inode) 3175 { 3176 int compress_mode = F2FS_OPTION(F2FS_I_SB(inode)).compress_mode; 3177 3178 if (!f2fs_compressed_file(inode)) 3179 return false; 3180 3181 if (compress_mode == COMPR_MODE_FS) 3182 return true; 3183 else if (compress_mode == COMPR_MODE_USER && 3184 is_inode_flag_set(inode, FI_ENABLE_COMPRESS)) 3185 return true; 3186 3187 return false; 3188 } 3189 3190 static inline unsigned int addrs_per_inode(struct inode *inode) 3191 { 3192 unsigned int addrs = CUR_ADDRS_PER_INODE(inode) - 3193 get_inline_xattr_addrs(inode); 3194 3195 if (!f2fs_compressed_file(inode)) 3196 return addrs; 3197 return ALIGN_DOWN(addrs, F2FS_I(inode)->i_cluster_size); 3198 } 3199 3200 static inline unsigned int addrs_per_block(struct inode *inode) 3201 { 3202 if (!f2fs_compressed_file(inode)) 3203 return DEF_ADDRS_PER_BLOCK; 3204 return ALIGN_DOWN(DEF_ADDRS_PER_BLOCK, F2FS_I(inode)->i_cluster_size); 3205 } 3206 3207 static inline void *inline_xattr_addr(struct inode *inode, struct page *page) 3208 { 3209 struct f2fs_inode *ri = F2FS_INODE(page); 3210 3211 return (void *)&(ri->i_addr[DEF_ADDRS_PER_INODE - 3212 get_inline_xattr_addrs(inode)]); 3213 } 3214 3215 static inline int inline_xattr_size(struct inode *inode) 3216 { 3217 if (f2fs_has_inline_xattr(inode)) 3218 return get_inline_xattr_addrs(inode) * sizeof(__le32); 3219 return 0; 3220 } 3221 3222 /* 3223 * Notice: check inline_data flag without inode page lock is unsafe. 3224 * It could change at any time by f2fs_convert_inline_page(). 3225 */ 3226 static inline int f2fs_has_inline_data(struct inode *inode) 3227 { 3228 return is_inode_flag_set(inode, FI_INLINE_DATA); 3229 } 3230 3231 static inline int f2fs_exist_data(struct inode *inode) 3232 { 3233 return is_inode_flag_set(inode, FI_DATA_EXIST); 3234 } 3235 3236 static inline int f2fs_has_inline_dots(struct inode *inode) 3237 { 3238 return is_inode_flag_set(inode, FI_INLINE_DOTS); 3239 } 3240 3241 static inline int f2fs_is_mmap_file(struct inode *inode) 3242 { 3243 return is_inode_flag_set(inode, FI_MMAP_FILE); 3244 } 3245 3246 static inline bool f2fs_is_pinned_file(struct inode *inode) 3247 { 3248 return is_inode_flag_set(inode, FI_PIN_FILE); 3249 } 3250 3251 static inline bool f2fs_is_atomic_file(struct inode *inode) 3252 { 3253 return is_inode_flag_set(inode, FI_ATOMIC_FILE); 3254 } 3255 3256 static inline bool f2fs_is_cow_file(struct inode *inode) 3257 { 3258 return is_inode_flag_set(inode, FI_COW_FILE); 3259 } 3260 3261 static inline bool f2fs_is_first_block_written(struct inode *inode) 3262 { 3263 return is_inode_flag_set(inode, FI_FIRST_BLOCK_WRITTEN); 3264 } 3265 3266 static inline bool f2fs_is_drop_cache(struct inode *inode) 3267 { 3268 return is_inode_flag_set(inode, FI_DROP_CACHE); 3269 } 3270 3271 static inline void *inline_data_addr(struct inode *inode, struct page *page) 3272 { 3273 struct f2fs_inode *ri = F2FS_INODE(page); 3274 int extra_size = get_extra_isize(inode); 3275 3276 return (void *)&(ri->i_addr[extra_size + DEF_INLINE_RESERVED_SIZE]); 3277 } 3278 3279 static inline int f2fs_has_inline_dentry(struct inode *inode) 3280 { 3281 return is_inode_flag_set(inode, FI_INLINE_DENTRY); 3282 } 3283 3284 static inline int is_file(struct inode *inode, int type) 3285 { 3286 return F2FS_I(inode)->i_advise & type; 3287 } 3288 3289 static inline void set_file(struct inode *inode, int type) 3290 { 3291 if (is_file(inode, type)) 3292 return; 3293 F2FS_I(inode)->i_advise |= type; 3294 f2fs_mark_inode_dirty_sync(inode, true); 3295 } 3296 3297 static inline void clear_file(struct inode *inode, int type) 3298 { 3299 if (!is_file(inode, type)) 3300 return; 3301 F2FS_I(inode)->i_advise &= ~type; 3302 f2fs_mark_inode_dirty_sync(inode, true); 3303 } 3304 3305 static inline bool f2fs_is_time_consistent(struct inode *inode) 3306 { 3307 if (!timespec64_equal(F2FS_I(inode)->i_disk_time, &inode->i_atime)) 3308 return false; 3309 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 1, &inode->i_ctime)) 3310 return false; 3311 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 2, &inode->i_mtime)) 3312 return false; 3313 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 3, 3314 &F2FS_I(inode)->i_crtime)) 3315 return false; 3316 return true; 3317 } 3318 3319 static inline bool f2fs_skip_inode_update(struct inode *inode, int dsync) 3320 { 3321 bool ret; 3322 3323 if (dsync) { 3324 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 3325 3326 spin_lock(&sbi->inode_lock[DIRTY_META]); 3327 ret = list_empty(&F2FS_I(inode)->gdirty_list); 3328 spin_unlock(&sbi->inode_lock[DIRTY_META]); 3329 return ret; 3330 } 3331 if (!is_inode_flag_set(inode, FI_AUTO_RECOVER) || 3332 file_keep_isize(inode) || 3333 i_size_read(inode) & ~PAGE_MASK) 3334 return false; 3335 3336 if (!f2fs_is_time_consistent(inode)) 3337 return false; 3338 3339 spin_lock(&F2FS_I(inode)->i_size_lock); 3340 ret = F2FS_I(inode)->last_disk_size == i_size_read(inode); 3341 spin_unlock(&F2FS_I(inode)->i_size_lock); 3342 3343 return ret; 3344 } 3345 3346 static inline bool f2fs_readonly(struct super_block *sb) 3347 { 3348 return sb_rdonly(sb); 3349 } 3350 3351 static inline bool f2fs_cp_error(struct f2fs_sb_info *sbi) 3352 { 3353 return is_set_ckpt_flags(sbi, CP_ERROR_FLAG); 3354 } 3355 3356 static inline bool is_dot_dotdot(const u8 *name, size_t len) 3357 { 3358 if (len == 1 && name[0] == '.') 3359 return true; 3360 3361 if (len == 2 && name[0] == '.' && name[1] == '.') 3362 return true; 3363 3364 return false; 3365 } 3366 3367 static inline void *f2fs_kmalloc(struct f2fs_sb_info *sbi, 3368 size_t size, gfp_t flags) 3369 { 3370 if (time_to_inject(sbi, FAULT_KMALLOC)) { 3371 f2fs_show_injection_info(sbi, FAULT_KMALLOC); 3372 return NULL; 3373 } 3374 3375 return kmalloc(size, flags); 3376 } 3377 3378 static inline void *f2fs_kzalloc(struct f2fs_sb_info *sbi, 3379 size_t size, gfp_t flags) 3380 { 3381 return f2fs_kmalloc(sbi, size, flags | __GFP_ZERO); 3382 } 3383 3384 static inline void *f2fs_kvmalloc(struct f2fs_sb_info *sbi, 3385 size_t size, gfp_t flags) 3386 { 3387 if (time_to_inject(sbi, FAULT_KVMALLOC)) { 3388 f2fs_show_injection_info(sbi, FAULT_KVMALLOC); 3389 return NULL; 3390 } 3391 3392 return kvmalloc(size, flags); 3393 } 3394 3395 static inline void *f2fs_kvzalloc(struct f2fs_sb_info *sbi, 3396 size_t size, gfp_t flags) 3397 { 3398 return f2fs_kvmalloc(sbi, size, flags | __GFP_ZERO); 3399 } 3400 3401 static inline int get_extra_isize(struct inode *inode) 3402 { 3403 return F2FS_I(inode)->i_extra_isize / sizeof(__le32); 3404 } 3405 3406 static inline int get_inline_xattr_addrs(struct inode *inode) 3407 { 3408 return F2FS_I(inode)->i_inline_xattr_size; 3409 } 3410 3411 #define f2fs_get_inode_mode(i) \ 3412 ((is_inode_flag_set(i, FI_ACL_MODE)) ? \ 3413 (F2FS_I(i)->i_acl_mode) : ((i)->i_mode)) 3414 3415 #define F2FS_TOTAL_EXTRA_ATTR_SIZE \ 3416 (offsetof(struct f2fs_inode, i_extra_end) - \ 3417 offsetof(struct f2fs_inode, i_extra_isize)) \ 3418 3419 #define F2FS_OLD_ATTRIBUTE_SIZE (offsetof(struct f2fs_inode, i_addr)) 3420 #define F2FS_FITS_IN_INODE(f2fs_inode, extra_isize, field) \ 3421 ((offsetof(typeof(*(f2fs_inode)), field) + \ 3422 sizeof((f2fs_inode)->field)) \ 3423 <= (F2FS_OLD_ATTRIBUTE_SIZE + (extra_isize))) \ 3424 3425 #define __is_large_section(sbi) ((sbi)->segs_per_sec > 1) 3426 3427 #define __is_meta_io(fio) (PAGE_TYPE_OF_BIO((fio)->type) == META) 3428 3429 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi, 3430 block_t blkaddr, int type); 3431 static inline void verify_blkaddr(struct f2fs_sb_info *sbi, 3432 block_t blkaddr, int type) 3433 { 3434 if (!f2fs_is_valid_blkaddr(sbi, blkaddr, type)) { 3435 f2fs_err(sbi, "invalid blkaddr: %u, type: %d, run fsck to fix.", 3436 blkaddr, type); 3437 f2fs_bug_on(sbi, 1); 3438 } 3439 } 3440 3441 static inline bool __is_valid_data_blkaddr(block_t blkaddr) 3442 { 3443 if (blkaddr == NEW_ADDR || blkaddr == NULL_ADDR || 3444 blkaddr == COMPRESS_ADDR) 3445 return false; 3446 return true; 3447 } 3448 3449 /* 3450 * file.c 3451 */ 3452 int f2fs_sync_file(struct file *file, loff_t start, loff_t end, int datasync); 3453 void f2fs_truncate_data_blocks(struct dnode_of_data *dn); 3454 int f2fs_do_truncate_blocks(struct inode *inode, u64 from, bool lock); 3455 int f2fs_truncate_blocks(struct inode *inode, u64 from, bool lock); 3456 int f2fs_truncate(struct inode *inode); 3457 int f2fs_getattr(struct user_namespace *mnt_userns, const struct path *path, 3458 struct kstat *stat, u32 request_mask, unsigned int flags); 3459 int f2fs_setattr(struct user_namespace *mnt_userns, struct dentry *dentry, 3460 struct iattr *attr); 3461 int f2fs_truncate_hole(struct inode *inode, pgoff_t pg_start, pgoff_t pg_end); 3462 void f2fs_truncate_data_blocks_range(struct dnode_of_data *dn, int count); 3463 int f2fs_precache_extents(struct inode *inode); 3464 int f2fs_fileattr_get(struct dentry *dentry, struct fileattr *fa); 3465 int f2fs_fileattr_set(struct user_namespace *mnt_userns, 3466 struct dentry *dentry, struct fileattr *fa); 3467 long f2fs_ioctl(struct file *filp, unsigned int cmd, unsigned long arg); 3468 long f2fs_compat_ioctl(struct file *file, unsigned int cmd, unsigned long arg); 3469 int f2fs_transfer_project_quota(struct inode *inode, kprojid_t kprojid); 3470 int f2fs_pin_file_control(struct inode *inode, bool inc); 3471 3472 /* 3473 * inode.c 3474 */ 3475 void f2fs_set_inode_flags(struct inode *inode); 3476 bool f2fs_inode_chksum_verify(struct f2fs_sb_info *sbi, struct page *page); 3477 void f2fs_inode_chksum_set(struct f2fs_sb_info *sbi, struct page *page); 3478 struct inode *f2fs_iget(struct super_block *sb, unsigned long ino); 3479 struct inode *f2fs_iget_retry(struct super_block *sb, unsigned long ino); 3480 int f2fs_try_to_free_nats(struct f2fs_sb_info *sbi, int nr_shrink); 3481 void f2fs_update_inode(struct inode *inode, struct page *node_page); 3482 void f2fs_update_inode_page(struct inode *inode); 3483 int f2fs_write_inode(struct inode *inode, struct writeback_control *wbc); 3484 void f2fs_evict_inode(struct inode *inode); 3485 void f2fs_handle_failed_inode(struct inode *inode); 3486 3487 /* 3488 * namei.c 3489 */ 3490 int f2fs_update_extension_list(struct f2fs_sb_info *sbi, const char *name, 3491 bool hot, bool set); 3492 struct dentry *f2fs_get_parent(struct dentry *child); 3493 int f2fs_get_tmpfile(struct user_namespace *mnt_userns, struct inode *dir, 3494 struct inode **new_inode); 3495 3496 /* 3497 * dir.c 3498 */ 3499 unsigned char f2fs_get_de_type(struct f2fs_dir_entry *de); 3500 int f2fs_init_casefolded_name(const struct inode *dir, 3501 struct f2fs_filename *fname); 3502 int f2fs_setup_filename(struct inode *dir, const struct qstr *iname, 3503 int lookup, struct f2fs_filename *fname); 3504 int f2fs_prepare_lookup(struct inode *dir, struct dentry *dentry, 3505 struct f2fs_filename *fname); 3506 void f2fs_free_filename(struct f2fs_filename *fname); 3507 struct f2fs_dir_entry *f2fs_find_target_dentry(const struct f2fs_dentry_ptr *d, 3508 const struct f2fs_filename *fname, int *max_slots); 3509 int f2fs_fill_dentries(struct dir_context *ctx, struct f2fs_dentry_ptr *d, 3510 unsigned int start_pos, struct fscrypt_str *fstr); 3511 void f2fs_do_make_empty_dir(struct inode *inode, struct inode *parent, 3512 struct f2fs_dentry_ptr *d); 3513 struct page *f2fs_init_inode_metadata(struct inode *inode, struct inode *dir, 3514 const struct f2fs_filename *fname, struct page *dpage); 3515 void f2fs_update_parent_metadata(struct inode *dir, struct inode *inode, 3516 unsigned int current_depth); 3517 int f2fs_room_for_filename(const void *bitmap, int slots, int max_slots); 3518 void f2fs_drop_nlink(struct inode *dir, struct inode *inode); 3519 struct f2fs_dir_entry *__f2fs_find_entry(struct inode *dir, 3520 const struct f2fs_filename *fname, 3521 struct page **res_page); 3522 struct f2fs_dir_entry *f2fs_find_entry(struct inode *dir, 3523 const struct qstr *child, struct page **res_page); 3524 struct f2fs_dir_entry *f2fs_parent_dir(struct inode *dir, struct page **p); 3525 ino_t f2fs_inode_by_name(struct inode *dir, const struct qstr *qstr, 3526 struct page **page); 3527 void f2fs_set_link(struct inode *dir, struct f2fs_dir_entry *de, 3528 struct page *page, struct inode *inode); 3529 bool f2fs_has_enough_room(struct inode *dir, struct page *ipage, 3530 const struct f2fs_filename *fname); 3531 void f2fs_update_dentry(nid_t ino, umode_t mode, struct f2fs_dentry_ptr *d, 3532 const struct fscrypt_str *name, f2fs_hash_t name_hash, 3533 unsigned int bit_pos); 3534 int f2fs_add_regular_entry(struct inode *dir, const struct f2fs_filename *fname, 3535 struct inode *inode, nid_t ino, umode_t mode); 3536 int f2fs_add_dentry(struct inode *dir, const struct f2fs_filename *fname, 3537 struct inode *inode, nid_t ino, umode_t mode); 3538 int f2fs_do_add_link(struct inode *dir, const struct qstr *name, 3539 struct inode *inode, nid_t ino, umode_t mode); 3540 void f2fs_delete_entry(struct f2fs_dir_entry *dentry, struct page *page, 3541 struct inode *dir, struct inode *inode); 3542 int f2fs_do_tmpfile(struct inode *inode, struct inode *dir); 3543 bool f2fs_empty_dir(struct inode *dir); 3544 3545 static inline int f2fs_add_link(struct dentry *dentry, struct inode *inode) 3546 { 3547 if (fscrypt_is_nokey_name(dentry)) 3548 return -ENOKEY; 3549 return f2fs_do_add_link(d_inode(dentry->d_parent), &dentry->d_name, 3550 inode, inode->i_ino, inode->i_mode); 3551 } 3552 3553 /* 3554 * super.c 3555 */ 3556 int f2fs_inode_dirtied(struct inode *inode, bool sync); 3557 void f2fs_inode_synced(struct inode *inode); 3558 int f2fs_dquot_initialize(struct inode *inode); 3559 int f2fs_enable_quota_files(struct f2fs_sb_info *sbi, bool rdonly); 3560 int f2fs_quota_sync(struct super_block *sb, int type); 3561 loff_t max_file_blocks(struct inode *inode); 3562 void f2fs_quota_off_umount(struct super_block *sb); 3563 void f2fs_handle_stop(struct f2fs_sb_info *sbi, unsigned char reason); 3564 void f2fs_handle_error(struct f2fs_sb_info *sbi, unsigned char error); 3565 int f2fs_commit_super(struct f2fs_sb_info *sbi, bool recover); 3566 int f2fs_sync_fs(struct super_block *sb, int sync); 3567 int f2fs_sanity_check_ckpt(struct f2fs_sb_info *sbi); 3568 3569 /* 3570 * hash.c 3571 */ 3572 void f2fs_hash_filename(const struct inode *dir, struct f2fs_filename *fname); 3573 3574 /* 3575 * node.c 3576 */ 3577 struct node_info; 3578 3579 int f2fs_check_nid_range(struct f2fs_sb_info *sbi, nid_t nid); 3580 bool f2fs_available_free_memory(struct f2fs_sb_info *sbi, int type); 3581 bool f2fs_in_warm_node_list(struct f2fs_sb_info *sbi, struct page *page); 3582 void f2fs_init_fsync_node_info(struct f2fs_sb_info *sbi); 3583 void f2fs_del_fsync_node_entry(struct f2fs_sb_info *sbi, struct page *page); 3584 void f2fs_reset_fsync_node_info(struct f2fs_sb_info *sbi); 3585 int f2fs_need_dentry_mark(struct f2fs_sb_info *sbi, nid_t nid); 3586 bool f2fs_is_checkpointed_node(struct f2fs_sb_info *sbi, nid_t nid); 3587 bool f2fs_need_inode_block_update(struct f2fs_sb_info *sbi, nid_t ino); 3588 int f2fs_get_node_info(struct f2fs_sb_info *sbi, nid_t nid, 3589 struct node_info *ni, bool checkpoint_context); 3590 pgoff_t f2fs_get_next_page_offset(struct dnode_of_data *dn, pgoff_t pgofs); 3591 int f2fs_get_dnode_of_data(struct dnode_of_data *dn, pgoff_t index, int mode); 3592 int f2fs_truncate_inode_blocks(struct inode *inode, pgoff_t from); 3593 int f2fs_truncate_xattr_node(struct inode *inode); 3594 int f2fs_wait_on_node_pages_writeback(struct f2fs_sb_info *sbi, 3595 unsigned int seq_id); 3596 bool f2fs_nat_bitmap_enabled(struct f2fs_sb_info *sbi); 3597 int f2fs_remove_inode_page(struct inode *inode); 3598 struct page *f2fs_new_inode_page(struct inode *inode); 3599 struct page *f2fs_new_node_page(struct dnode_of_data *dn, unsigned int ofs); 3600 void f2fs_ra_node_page(struct f2fs_sb_info *sbi, nid_t nid); 3601 struct page *f2fs_get_node_page(struct f2fs_sb_info *sbi, pgoff_t nid); 3602 struct page *f2fs_get_node_page_ra(struct page *parent, int start); 3603 int f2fs_move_node_page(struct page *node_page, int gc_type); 3604 void f2fs_flush_inline_data(struct f2fs_sb_info *sbi); 3605 int f2fs_fsync_node_pages(struct f2fs_sb_info *sbi, struct inode *inode, 3606 struct writeback_control *wbc, bool atomic, 3607 unsigned int *seq_id); 3608 int f2fs_sync_node_pages(struct f2fs_sb_info *sbi, 3609 struct writeback_control *wbc, 3610 bool do_balance, enum iostat_type io_type); 3611 int f2fs_build_free_nids(struct f2fs_sb_info *sbi, bool sync, bool mount); 3612 bool f2fs_alloc_nid(struct f2fs_sb_info *sbi, nid_t *nid); 3613 void f2fs_alloc_nid_done(struct f2fs_sb_info *sbi, nid_t nid); 3614 void f2fs_alloc_nid_failed(struct f2fs_sb_info *sbi, nid_t nid); 3615 int f2fs_try_to_free_nids(struct f2fs_sb_info *sbi, int nr_shrink); 3616 int f2fs_recover_inline_xattr(struct inode *inode, struct page *page); 3617 int f2fs_recover_xattr_data(struct inode *inode, struct page *page); 3618 int f2fs_recover_inode_page(struct f2fs_sb_info *sbi, struct page *page); 3619 int f2fs_restore_node_summary(struct f2fs_sb_info *sbi, 3620 unsigned int segno, struct f2fs_summary_block *sum); 3621 void f2fs_enable_nat_bits(struct f2fs_sb_info *sbi); 3622 int f2fs_flush_nat_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3623 int f2fs_build_node_manager(struct f2fs_sb_info *sbi); 3624 void f2fs_destroy_node_manager(struct f2fs_sb_info *sbi); 3625 int __init f2fs_create_node_manager_caches(void); 3626 void f2fs_destroy_node_manager_caches(void); 3627 3628 /* 3629 * segment.c 3630 */ 3631 bool f2fs_need_SSR(struct f2fs_sb_info *sbi); 3632 int f2fs_commit_atomic_write(struct inode *inode); 3633 void f2fs_abort_atomic_write(struct inode *inode, bool clean); 3634 void f2fs_balance_fs(struct f2fs_sb_info *sbi, bool need); 3635 void f2fs_balance_fs_bg(struct f2fs_sb_info *sbi, bool from_bg); 3636 int f2fs_issue_flush(struct f2fs_sb_info *sbi, nid_t ino); 3637 int f2fs_create_flush_cmd_control(struct f2fs_sb_info *sbi); 3638 int f2fs_flush_device_cache(struct f2fs_sb_info *sbi); 3639 void f2fs_destroy_flush_cmd_control(struct f2fs_sb_info *sbi, bool free); 3640 void f2fs_invalidate_blocks(struct f2fs_sb_info *sbi, block_t addr); 3641 bool f2fs_is_checkpointed_data(struct f2fs_sb_info *sbi, block_t blkaddr); 3642 int f2fs_start_discard_thread(struct f2fs_sb_info *sbi); 3643 void f2fs_drop_discard_cmd(struct f2fs_sb_info *sbi); 3644 void f2fs_stop_discard_thread(struct f2fs_sb_info *sbi); 3645 bool f2fs_issue_discard_timeout(struct f2fs_sb_info *sbi); 3646 void f2fs_clear_prefree_segments(struct f2fs_sb_info *sbi, 3647 struct cp_control *cpc); 3648 void f2fs_dirty_to_prefree(struct f2fs_sb_info *sbi); 3649 block_t f2fs_get_unusable_blocks(struct f2fs_sb_info *sbi); 3650 int f2fs_disable_cp_again(struct f2fs_sb_info *sbi, block_t unusable); 3651 void f2fs_release_discard_addrs(struct f2fs_sb_info *sbi); 3652 int f2fs_npages_for_summary_flush(struct f2fs_sb_info *sbi, bool for_ra); 3653 bool f2fs_segment_has_free_slot(struct f2fs_sb_info *sbi, int segno); 3654 void f2fs_init_inmem_curseg(struct f2fs_sb_info *sbi); 3655 void f2fs_save_inmem_curseg(struct f2fs_sb_info *sbi); 3656 void f2fs_restore_inmem_curseg(struct f2fs_sb_info *sbi); 3657 void f2fs_get_new_segment(struct f2fs_sb_info *sbi, 3658 unsigned int *newseg, bool new_sec, int dir); 3659 void f2fs_allocate_segment_for_resize(struct f2fs_sb_info *sbi, int type, 3660 unsigned int start, unsigned int end); 3661 void f2fs_allocate_new_section(struct f2fs_sb_info *sbi, int type, bool force); 3662 void f2fs_allocate_new_segments(struct f2fs_sb_info *sbi); 3663 int f2fs_trim_fs(struct f2fs_sb_info *sbi, struct fstrim_range *range); 3664 bool f2fs_exist_trim_candidates(struct f2fs_sb_info *sbi, 3665 struct cp_control *cpc); 3666 struct page *f2fs_get_sum_page(struct f2fs_sb_info *sbi, unsigned int segno); 3667 void f2fs_update_meta_page(struct f2fs_sb_info *sbi, void *src, 3668 block_t blk_addr); 3669 void f2fs_do_write_meta_page(struct f2fs_sb_info *sbi, struct page *page, 3670 enum iostat_type io_type); 3671 void f2fs_do_write_node_page(unsigned int nid, struct f2fs_io_info *fio); 3672 void f2fs_outplace_write_data(struct dnode_of_data *dn, 3673 struct f2fs_io_info *fio); 3674 int f2fs_inplace_write_data(struct f2fs_io_info *fio); 3675 void f2fs_do_replace_block(struct f2fs_sb_info *sbi, struct f2fs_summary *sum, 3676 block_t old_blkaddr, block_t new_blkaddr, 3677 bool recover_curseg, bool recover_newaddr, 3678 bool from_gc); 3679 void f2fs_replace_block(struct f2fs_sb_info *sbi, struct dnode_of_data *dn, 3680 block_t old_addr, block_t new_addr, 3681 unsigned char version, bool recover_curseg, 3682 bool recover_newaddr); 3683 void f2fs_allocate_data_block(struct f2fs_sb_info *sbi, struct page *page, 3684 block_t old_blkaddr, block_t *new_blkaddr, 3685 struct f2fs_summary *sum, int type, 3686 struct f2fs_io_info *fio); 3687 void f2fs_update_device_state(struct f2fs_sb_info *sbi, nid_t ino, 3688 block_t blkaddr, unsigned int blkcnt); 3689 void f2fs_wait_on_page_writeback(struct page *page, 3690 enum page_type type, bool ordered, bool locked); 3691 void f2fs_wait_on_block_writeback(struct inode *inode, block_t blkaddr); 3692 void f2fs_wait_on_block_writeback_range(struct inode *inode, block_t blkaddr, 3693 block_t len); 3694 void f2fs_write_data_summaries(struct f2fs_sb_info *sbi, block_t start_blk); 3695 void f2fs_write_node_summaries(struct f2fs_sb_info *sbi, block_t start_blk); 3696 int f2fs_lookup_journal_in_cursum(struct f2fs_journal *journal, int type, 3697 unsigned int val, int alloc); 3698 void f2fs_flush_sit_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3699 int f2fs_fix_curseg_write_pointer(struct f2fs_sb_info *sbi); 3700 int f2fs_check_write_pointer(struct f2fs_sb_info *sbi); 3701 int f2fs_build_segment_manager(struct f2fs_sb_info *sbi); 3702 void f2fs_destroy_segment_manager(struct f2fs_sb_info *sbi); 3703 int __init f2fs_create_segment_manager_caches(void); 3704 void f2fs_destroy_segment_manager_caches(void); 3705 int f2fs_rw_hint_to_seg_type(enum rw_hint hint); 3706 unsigned int f2fs_usable_segs_in_sec(struct f2fs_sb_info *sbi, 3707 unsigned int segno); 3708 unsigned int f2fs_usable_blks_in_seg(struct f2fs_sb_info *sbi, 3709 unsigned int segno); 3710 3711 #define DEF_FRAGMENT_SIZE 4 3712 #define MIN_FRAGMENT_SIZE 1 3713 #define MAX_FRAGMENT_SIZE 512 3714 3715 static inline bool f2fs_need_rand_seg(struct f2fs_sb_info *sbi) 3716 { 3717 return F2FS_OPTION(sbi).fs_mode == FS_MODE_FRAGMENT_SEG || 3718 F2FS_OPTION(sbi).fs_mode == FS_MODE_FRAGMENT_BLK; 3719 } 3720 3721 /* 3722 * checkpoint.c 3723 */ 3724 void f2fs_stop_checkpoint(struct f2fs_sb_info *sbi, bool end_io, 3725 unsigned char reason); 3726 void f2fs_flush_ckpt_thread(struct f2fs_sb_info *sbi); 3727 struct page *f2fs_grab_meta_page(struct f2fs_sb_info *sbi, pgoff_t index); 3728 struct page *f2fs_get_meta_page(struct f2fs_sb_info *sbi, pgoff_t index); 3729 struct page *f2fs_get_meta_page_retry(struct f2fs_sb_info *sbi, pgoff_t index); 3730 struct page *f2fs_get_tmp_page(struct f2fs_sb_info *sbi, pgoff_t index); 3731 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi, 3732 block_t blkaddr, int type); 3733 int f2fs_ra_meta_pages(struct f2fs_sb_info *sbi, block_t start, int nrpages, 3734 int type, bool sync); 3735 void f2fs_ra_meta_pages_cond(struct f2fs_sb_info *sbi, pgoff_t index, 3736 unsigned int ra_blocks); 3737 long f2fs_sync_meta_pages(struct f2fs_sb_info *sbi, enum page_type type, 3738 long nr_to_write, enum iostat_type io_type); 3739 void f2fs_add_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type); 3740 void f2fs_remove_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type); 3741 void f2fs_release_ino_entry(struct f2fs_sb_info *sbi, bool all); 3742 bool f2fs_exist_written_data(struct f2fs_sb_info *sbi, nid_t ino, int mode); 3743 void f2fs_set_dirty_device(struct f2fs_sb_info *sbi, nid_t ino, 3744 unsigned int devidx, int type); 3745 bool f2fs_is_dirty_device(struct f2fs_sb_info *sbi, nid_t ino, 3746 unsigned int devidx, int type); 3747 int f2fs_sync_inode_meta(struct f2fs_sb_info *sbi); 3748 int f2fs_acquire_orphan_inode(struct f2fs_sb_info *sbi); 3749 void f2fs_release_orphan_inode(struct f2fs_sb_info *sbi); 3750 void f2fs_add_orphan_inode(struct inode *inode); 3751 void f2fs_remove_orphan_inode(struct f2fs_sb_info *sbi, nid_t ino); 3752 int f2fs_recover_orphan_inodes(struct f2fs_sb_info *sbi); 3753 int f2fs_get_valid_checkpoint(struct f2fs_sb_info *sbi); 3754 void f2fs_update_dirty_folio(struct inode *inode, struct folio *folio); 3755 void f2fs_remove_dirty_inode(struct inode *inode); 3756 int f2fs_sync_dirty_inodes(struct f2fs_sb_info *sbi, enum inode_type type, 3757 bool from_cp); 3758 void f2fs_wait_on_all_pages(struct f2fs_sb_info *sbi, int type); 3759 u64 f2fs_get_sectors_written(struct f2fs_sb_info *sbi); 3760 int f2fs_write_checkpoint(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3761 void f2fs_init_ino_entry_info(struct f2fs_sb_info *sbi); 3762 int __init f2fs_create_checkpoint_caches(void); 3763 void f2fs_destroy_checkpoint_caches(void); 3764 int f2fs_issue_checkpoint(struct f2fs_sb_info *sbi); 3765 int f2fs_start_ckpt_thread(struct f2fs_sb_info *sbi); 3766 void f2fs_stop_ckpt_thread(struct f2fs_sb_info *sbi); 3767 void f2fs_init_ckpt_req_control(struct f2fs_sb_info *sbi); 3768 3769 /* 3770 * data.c 3771 */ 3772 int __init f2fs_init_bioset(void); 3773 void f2fs_destroy_bioset(void); 3774 int f2fs_init_bio_entry_cache(void); 3775 void f2fs_destroy_bio_entry_cache(void); 3776 void f2fs_submit_bio(struct f2fs_sb_info *sbi, 3777 struct bio *bio, enum page_type type); 3778 int f2fs_init_write_merge_io(struct f2fs_sb_info *sbi); 3779 void f2fs_submit_merged_write(struct f2fs_sb_info *sbi, enum page_type type); 3780 void f2fs_submit_merged_write_cond(struct f2fs_sb_info *sbi, 3781 struct inode *inode, struct page *page, 3782 nid_t ino, enum page_type type); 3783 void f2fs_submit_merged_ipu_write(struct f2fs_sb_info *sbi, 3784 struct bio **bio, struct page *page); 3785 void f2fs_flush_merged_writes(struct f2fs_sb_info *sbi); 3786 int f2fs_submit_page_bio(struct f2fs_io_info *fio); 3787 int f2fs_merge_page_bio(struct f2fs_io_info *fio); 3788 void f2fs_submit_page_write(struct f2fs_io_info *fio); 3789 struct block_device *f2fs_target_device(struct f2fs_sb_info *sbi, 3790 block_t blk_addr, sector_t *sector); 3791 int f2fs_target_device_index(struct f2fs_sb_info *sbi, block_t blkaddr); 3792 void f2fs_set_data_blkaddr(struct dnode_of_data *dn); 3793 void f2fs_update_data_blkaddr(struct dnode_of_data *dn, block_t blkaddr); 3794 int f2fs_reserve_new_blocks(struct dnode_of_data *dn, blkcnt_t count); 3795 int f2fs_reserve_new_block(struct dnode_of_data *dn); 3796 int f2fs_get_block(struct dnode_of_data *dn, pgoff_t index); 3797 int f2fs_reserve_block(struct dnode_of_data *dn, pgoff_t index); 3798 struct page *f2fs_get_read_data_page(struct inode *inode, pgoff_t index, 3799 blk_opf_t op_flags, bool for_write); 3800 struct page *f2fs_find_data_page(struct inode *inode, pgoff_t index); 3801 struct page *f2fs_get_lock_data_page(struct inode *inode, pgoff_t index, 3802 bool for_write); 3803 struct page *f2fs_get_new_data_page(struct inode *inode, 3804 struct page *ipage, pgoff_t index, bool new_i_size); 3805 int f2fs_do_write_data_page(struct f2fs_io_info *fio); 3806 void f2fs_do_map_lock(struct f2fs_sb_info *sbi, int flag, bool lock); 3807 int f2fs_map_blocks(struct inode *inode, struct f2fs_map_blocks *map, 3808 int create, int flag); 3809 int f2fs_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, 3810 u64 start, u64 len); 3811 int f2fs_encrypt_one_page(struct f2fs_io_info *fio); 3812 bool f2fs_should_update_inplace(struct inode *inode, struct f2fs_io_info *fio); 3813 bool f2fs_should_update_outplace(struct inode *inode, struct f2fs_io_info *fio); 3814 int f2fs_write_single_data_page(struct page *page, int *submitted, 3815 struct bio **bio, sector_t *last_block, 3816 struct writeback_control *wbc, 3817 enum iostat_type io_type, 3818 int compr_blocks, bool allow_balance); 3819 void f2fs_write_failed(struct inode *inode, loff_t to); 3820 void f2fs_invalidate_folio(struct folio *folio, size_t offset, size_t length); 3821 bool f2fs_release_folio(struct folio *folio, gfp_t wait); 3822 bool f2fs_overwrite_io(struct inode *inode, loff_t pos, size_t len); 3823 void f2fs_clear_page_cache_dirty_tag(struct page *page); 3824 int f2fs_init_post_read_processing(void); 3825 void f2fs_destroy_post_read_processing(void); 3826 int f2fs_init_post_read_wq(struct f2fs_sb_info *sbi); 3827 void f2fs_destroy_post_read_wq(struct f2fs_sb_info *sbi); 3828 extern const struct iomap_ops f2fs_iomap_ops; 3829 3830 /* 3831 * gc.c 3832 */ 3833 int f2fs_start_gc_thread(struct f2fs_sb_info *sbi); 3834 void f2fs_stop_gc_thread(struct f2fs_sb_info *sbi); 3835 block_t f2fs_start_bidx_of_node(unsigned int node_ofs, struct inode *inode); 3836 int f2fs_gc(struct f2fs_sb_info *sbi, struct f2fs_gc_control *gc_control); 3837 void f2fs_build_gc_manager(struct f2fs_sb_info *sbi); 3838 int f2fs_resize_fs(struct f2fs_sb_info *sbi, __u64 block_count); 3839 int __init f2fs_create_garbage_collection_cache(void); 3840 void f2fs_destroy_garbage_collection_cache(void); 3841 3842 /* 3843 * recovery.c 3844 */ 3845 int f2fs_recover_fsync_data(struct f2fs_sb_info *sbi, bool check_only); 3846 bool f2fs_space_for_roll_forward(struct f2fs_sb_info *sbi); 3847 int __init f2fs_create_recovery_cache(void); 3848 void f2fs_destroy_recovery_cache(void); 3849 3850 /* 3851 * debug.c 3852 */ 3853 #ifdef CONFIG_F2FS_STAT_FS 3854 struct f2fs_stat_info { 3855 struct list_head stat_list; 3856 struct f2fs_sb_info *sbi; 3857 int all_area_segs, sit_area_segs, nat_area_segs, ssa_area_segs; 3858 int main_area_segs, main_area_sections, main_area_zones; 3859 unsigned long long hit_largest, hit_cached, hit_rbtree; 3860 unsigned long long hit_total, total_ext; 3861 int ext_tree, zombie_tree, ext_node; 3862 int ndirty_node, ndirty_dent, ndirty_meta, ndirty_imeta; 3863 int ndirty_data, ndirty_qdata; 3864 unsigned int ndirty_dirs, ndirty_files, nquota_files, ndirty_all; 3865 int nats, dirty_nats, sits, dirty_sits; 3866 int free_nids, avail_nids, alloc_nids; 3867 int total_count, utilization; 3868 int bg_gc, nr_wb_cp_data, nr_wb_data; 3869 int nr_rd_data, nr_rd_node, nr_rd_meta; 3870 int nr_dio_read, nr_dio_write; 3871 unsigned int io_skip_bggc, other_skip_bggc; 3872 int nr_flushing, nr_flushed, flush_list_empty; 3873 int nr_discarding, nr_discarded; 3874 int nr_discard_cmd; 3875 unsigned int undiscard_blks; 3876 int nr_issued_ckpt, nr_total_ckpt, nr_queued_ckpt; 3877 unsigned int cur_ckpt_time, peak_ckpt_time; 3878 int inline_xattr, inline_inode, inline_dir, append, update, orphans; 3879 int compr_inode, swapfile_inode; 3880 unsigned long long compr_blocks; 3881 int aw_cnt, max_aw_cnt; 3882 unsigned int valid_count, valid_node_count, valid_inode_count, discard_blks; 3883 unsigned int bimodal, avg_vblocks; 3884 int util_free, util_valid, util_invalid; 3885 int rsvd_segs, overp_segs; 3886 int dirty_count, node_pages, meta_pages, compress_pages; 3887 int compress_page_hit; 3888 int prefree_count, call_count, cp_count, bg_cp_count; 3889 int tot_segs, node_segs, data_segs, free_segs, free_secs; 3890 int bg_node_segs, bg_data_segs; 3891 int tot_blks, data_blks, node_blks; 3892 int bg_data_blks, bg_node_blks; 3893 int curseg[NR_CURSEG_TYPE]; 3894 int cursec[NR_CURSEG_TYPE]; 3895 int curzone[NR_CURSEG_TYPE]; 3896 unsigned int dirty_seg[NR_CURSEG_TYPE]; 3897 unsigned int full_seg[NR_CURSEG_TYPE]; 3898 unsigned int valid_blks[NR_CURSEG_TYPE]; 3899 3900 unsigned int meta_count[META_MAX]; 3901 unsigned int segment_count[2]; 3902 unsigned int block_count[2]; 3903 unsigned int inplace_count; 3904 unsigned long long base_mem, cache_mem, page_mem; 3905 }; 3906 3907 static inline struct f2fs_stat_info *F2FS_STAT(struct f2fs_sb_info *sbi) 3908 { 3909 return (struct f2fs_stat_info *)sbi->stat_info; 3910 } 3911 3912 #define stat_inc_cp_count(si) ((si)->cp_count++) 3913 #define stat_inc_bg_cp_count(si) ((si)->bg_cp_count++) 3914 #define stat_inc_call_count(si) ((si)->call_count++) 3915 #define stat_inc_bggc_count(si) ((si)->bg_gc++) 3916 #define stat_io_skip_bggc_count(sbi) ((sbi)->io_skip_bggc++) 3917 #define stat_other_skip_bggc_count(sbi) ((sbi)->other_skip_bggc++) 3918 #define stat_inc_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]++) 3919 #define stat_dec_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]--) 3920 #define stat_inc_total_hit(sbi) (atomic64_inc(&(sbi)->total_hit_ext)) 3921 #define stat_inc_rbtree_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_rbtree)) 3922 #define stat_inc_largest_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_largest)) 3923 #define stat_inc_cached_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_cached)) 3924 #define stat_inc_inline_xattr(inode) \ 3925 do { \ 3926 if (f2fs_has_inline_xattr(inode)) \ 3927 (atomic_inc(&F2FS_I_SB(inode)->inline_xattr)); \ 3928 } while (0) 3929 #define stat_dec_inline_xattr(inode) \ 3930 do { \ 3931 if (f2fs_has_inline_xattr(inode)) \ 3932 (atomic_dec(&F2FS_I_SB(inode)->inline_xattr)); \ 3933 } while (0) 3934 #define stat_inc_inline_inode(inode) \ 3935 do { \ 3936 if (f2fs_has_inline_data(inode)) \ 3937 (atomic_inc(&F2FS_I_SB(inode)->inline_inode)); \ 3938 } while (0) 3939 #define stat_dec_inline_inode(inode) \ 3940 do { \ 3941 if (f2fs_has_inline_data(inode)) \ 3942 (atomic_dec(&F2FS_I_SB(inode)->inline_inode)); \ 3943 } while (0) 3944 #define stat_inc_inline_dir(inode) \ 3945 do { \ 3946 if (f2fs_has_inline_dentry(inode)) \ 3947 (atomic_inc(&F2FS_I_SB(inode)->inline_dir)); \ 3948 } while (0) 3949 #define stat_dec_inline_dir(inode) \ 3950 do { \ 3951 if (f2fs_has_inline_dentry(inode)) \ 3952 (atomic_dec(&F2FS_I_SB(inode)->inline_dir)); \ 3953 } while (0) 3954 #define stat_inc_compr_inode(inode) \ 3955 do { \ 3956 if (f2fs_compressed_file(inode)) \ 3957 (atomic_inc(&F2FS_I_SB(inode)->compr_inode)); \ 3958 } while (0) 3959 #define stat_dec_compr_inode(inode) \ 3960 do { \ 3961 if (f2fs_compressed_file(inode)) \ 3962 (atomic_dec(&F2FS_I_SB(inode)->compr_inode)); \ 3963 } while (0) 3964 #define stat_add_compr_blocks(inode, blocks) \ 3965 (atomic64_add(blocks, &F2FS_I_SB(inode)->compr_blocks)) 3966 #define stat_sub_compr_blocks(inode, blocks) \ 3967 (atomic64_sub(blocks, &F2FS_I_SB(inode)->compr_blocks)) 3968 #define stat_inc_swapfile_inode(inode) \ 3969 (atomic_inc(&F2FS_I_SB(inode)->swapfile_inode)) 3970 #define stat_dec_swapfile_inode(inode) \ 3971 (atomic_dec(&F2FS_I_SB(inode)->swapfile_inode)) 3972 #define stat_inc_atomic_inode(inode) \ 3973 (atomic_inc(&F2FS_I_SB(inode)->atomic_files)) 3974 #define stat_dec_atomic_inode(inode) \ 3975 (atomic_dec(&F2FS_I_SB(inode)->atomic_files)) 3976 #define stat_inc_meta_count(sbi, blkaddr) \ 3977 do { \ 3978 if (blkaddr < SIT_I(sbi)->sit_base_addr) \ 3979 atomic_inc(&(sbi)->meta_count[META_CP]); \ 3980 else if (blkaddr < NM_I(sbi)->nat_blkaddr) \ 3981 atomic_inc(&(sbi)->meta_count[META_SIT]); \ 3982 else if (blkaddr < SM_I(sbi)->ssa_blkaddr) \ 3983 atomic_inc(&(sbi)->meta_count[META_NAT]); \ 3984 else if (blkaddr < SM_I(sbi)->main_blkaddr) \ 3985 atomic_inc(&(sbi)->meta_count[META_SSA]); \ 3986 } while (0) 3987 #define stat_inc_seg_type(sbi, curseg) \ 3988 ((sbi)->segment_count[(curseg)->alloc_type]++) 3989 #define stat_inc_block_count(sbi, curseg) \ 3990 ((sbi)->block_count[(curseg)->alloc_type]++) 3991 #define stat_inc_inplace_blocks(sbi) \ 3992 (atomic_inc(&(sbi)->inplace_count)) 3993 #define stat_update_max_atomic_write(inode) \ 3994 do { \ 3995 int cur = atomic_read(&F2FS_I_SB(inode)->atomic_files); \ 3996 int max = atomic_read(&F2FS_I_SB(inode)->max_aw_cnt); \ 3997 if (cur > max) \ 3998 atomic_set(&F2FS_I_SB(inode)->max_aw_cnt, cur); \ 3999 } while (0) 4000 #define stat_inc_seg_count(sbi, type, gc_type) \ 4001 do { \ 4002 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 4003 si->tot_segs++; \ 4004 if ((type) == SUM_TYPE_DATA) { \ 4005 si->data_segs++; \ 4006 si->bg_data_segs += (gc_type == BG_GC) ? 1 : 0; \ 4007 } else { \ 4008 si->node_segs++; \ 4009 si->bg_node_segs += (gc_type == BG_GC) ? 1 : 0; \ 4010 } \ 4011 } while (0) 4012 4013 #define stat_inc_tot_blk_count(si, blks) \ 4014 ((si)->tot_blks += (blks)) 4015 4016 #define stat_inc_data_blk_count(sbi, blks, gc_type) \ 4017 do { \ 4018 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 4019 stat_inc_tot_blk_count(si, blks); \ 4020 si->data_blks += (blks); \ 4021 si->bg_data_blks += ((gc_type) == BG_GC) ? (blks) : 0; \ 4022 } while (0) 4023 4024 #define stat_inc_node_blk_count(sbi, blks, gc_type) \ 4025 do { \ 4026 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 4027 stat_inc_tot_blk_count(si, blks); \ 4028 si->node_blks += (blks); \ 4029 si->bg_node_blks += ((gc_type) == BG_GC) ? (blks) : 0; \ 4030 } while (0) 4031 4032 int f2fs_build_stats(struct f2fs_sb_info *sbi); 4033 void f2fs_destroy_stats(struct f2fs_sb_info *sbi); 4034 void __init f2fs_create_root_stats(void); 4035 void f2fs_destroy_root_stats(void); 4036 void f2fs_update_sit_info(struct f2fs_sb_info *sbi); 4037 #else 4038 #define stat_inc_cp_count(si) do { } while (0) 4039 #define stat_inc_bg_cp_count(si) do { } while (0) 4040 #define stat_inc_call_count(si) do { } while (0) 4041 #define stat_inc_bggc_count(si) do { } while (0) 4042 #define stat_io_skip_bggc_count(sbi) do { } while (0) 4043 #define stat_other_skip_bggc_count(sbi) do { } while (0) 4044 #define stat_inc_dirty_inode(sbi, type) do { } while (0) 4045 #define stat_dec_dirty_inode(sbi, type) do { } while (0) 4046 #define stat_inc_total_hit(sbi) do { } while (0) 4047 #define stat_inc_rbtree_node_hit(sbi) do { } while (0) 4048 #define stat_inc_largest_node_hit(sbi) do { } while (0) 4049 #define stat_inc_cached_node_hit(sbi) do { } while (0) 4050 #define stat_inc_inline_xattr(inode) do { } while (0) 4051 #define stat_dec_inline_xattr(inode) do { } while (0) 4052 #define stat_inc_inline_inode(inode) do { } while (0) 4053 #define stat_dec_inline_inode(inode) do { } while (0) 4054 #define stat_inc_inline_dir(inode) do { } while (0) 4055 #define stat_dec_inline_dir(inode) do { } while (0) 4056 #define stat_inc_compr_inode(inode) do { } while (0) 4057 #define stat_dec_compr_inode(inode) do { } while (0) 4058 #define stat_add_compr_blocks(inode, blocks) do { } while (0) 4059 #define stat_sub_compr_blocks(inode, blocks) do { } while (0) 4060 #define stat_inc_swapfile_inode(inode) do { } while (0) 4061 #define stat_dec_swapfile_inode(inode) do { } while (0) 4062 #define stat_inc_atomic_inode(inode) do { } while (0) 4063 #define stat_dec_atomic_inode(inode) do { } while (0) 4064 #define stat_update_max_atomic_write(inode) do { } while (0) 4065 #define stat_inc_meta_count(sbi, blkaddr) do { } while (0) 4066 #define stat_inc_seg_type(sbi, curseg) do { } while (0) 4067 #define stat_inc_block_count(sbi, curseg) do { } while (0) 4068 #define stat_inc_inplace_blocks(sbi) do { } while (0) 4069 #define stat_inc_seg_count(sbi, type, gc_type) do { } while (0) 4070 #define stat_inc_tot_blk_count(si, blks) do { } while (0) 4071 #define stat_inc_data_blk_count(sbi, blks, gc_type) do { } while (0) 4072 #define stat_inc_node_blk_count(sbi, blks, gc_type) do { } while (0) 4073 4074 static inline int f2fs_build_stats(struct f2fs_sb_info *sbi) { return 0; } 4075 static inline void f2fs_destroy_stats(struct f2fs_sb_info *sbi) { } 4076 static inline void __init f2fs_create_root_stats(void) { } 4077 static inline void f2fs_destroy_root_stats(void) { } 4078 static inline void f2fs_update_sit_info(struct f2fs_sb_info *sbi) {} 4079 #endif 4080 4081 extern const struct file_operations f2fs_dir_operations; 4082 extern const struct file_operations f2fs_file_operations; 4083 extern const struct inode_operations f2fs_file_inode_operations; 4084 extern const struct address_space_operations f2fs_dblock_aops; 4085 extern const struct address_space_operations f2fs_node_aops; 4086 extern const struct address_space_operations f2fs_meta_aops; 4087 extern const struct inode_operations f2fs_dir_inode_operations; 4088 extern const struct inode_operations f2fs_symlink_inode_operations; 4089 extern const struct inode_operations f2fs_encrypted_symlink_inode_operations; 4090 extern const struct inode_operations f2fs_special_inode_operations; 4091 extern struct kmem_cache *f2fs_inode_entry_slab; 4092 4093 /* 4094 * inline.c 4095 */ 4096 bool f2fs_may_inline_data(struct inode *inode); 4097 bool f2fs_sanity_check_inline_data(struct inode *inode); 4098 bool f2fs_may_inline_dentry(struct inode *inode); 4099 void f2fs_do_read_inline_data(struct page *page, struct page *ipage); 4100 void f2fs_truncate_inline_inode(struct inode *inode, 4101 struct page *ipage, u64 from); 4102 int f2fs_read_inline_data(struct inode *inode, struct page *page); 4103 int f2fs_convert_inline_page(struct dnode_of_data *dn, struct page *page); 4104 int f2fs_convert_inline_inode(struct inode *inode); 4105 int f2fs_try_convert_inline_dir(struct inode *dir, struct dentry *dentry); 4106 int f2fs_write_inline_data(struct inode *inode, struct page *page); 4107 int f2fs_recover_inline_data(struct inode *inode, struct page *npage); 4108 struct f2fs_dir_entry *f2fs_find_in_inline_dir(struct inode *dir, 4109 const struct f2fs_filename *fname, 4110 struct page **res_page); 4111 int f2fs_make_empty_inline_dir(struct inode *inode, struct inode *parent, 4112 struct page *ipage); 4113 int f2fs_add_inline_entry(struct inode *dir, const struct f2fs_filename *fname, 4114 struct inode *inode, nid_t ino, umode_t mode); 4115 void f2fs_delete_inline_entry(struct f2fs_dir_entry *dentry, 4116 struct page *page, struct inode *dir, 4117 struct inode *inode); 4118 bool f2fs_empty_inline_dir(struct inode *dir); 4119 int f2fs_read_inline_dir(struct file *file, struct dir_context *ctx, 4120 struct fscrypt_str *fstr); 4121 int f2fs_inline_data_fiemap(struct inode *inode, 4122 struct fiemap_extent_info *fieinfo, 4123 __u64 start, __u64 len); 4124 4125 /* 4126 * shrinker.c 4127 */ 4128 unsigned long f2fs_shrink_count(struct shrinker *shrink, 4129 struct shrink_control *sc); 4130 unsigned long f2fs_shrink_scan(struct shrinker *shrink, 4131 struct shrink_control *sc); 4132 void f2fs_join_shrinker(struct f2fs_sb_info *sbi); 4133 void f2fs_leave_shrinker(struct f2fs_sb_info *sbi); 4134 4135 /* 4136 * extent_cache.c 4137 */ 4138 struct rb_entry *f2fs_lookup_rb_tree(struct rb_root_cached *root, 4139 struct rb_entry *cached_re, unsigned int ofs); 4140 struct rb_node **f2fs_lookup_rb_tree_ext(struct f2fs_sb_info *sbi, 4141 struct rb_root_cached *root, 4142 struct rb_node **parent, 4143 unsigned long long key, bool *left_most); 4144 struct rb_node **f2fs_lookup_rb_tree_for_insert(struct f2fs_sb_info *sbi, 4145 struct rb_root_cached *root, 4146 struct rb_node **parent, 4147 unsigned int ofs, bool *leftmost); 4148 struct rb_entry *f2fs_lookup_rb_tree_ret(struct rb_root_cached *root, 4149 struct rb_entry *cached_re, unsigned int ofs, 4150 struct rb_entry **prev_entry, struct rb_entry **next_entry, 4151 struct rb_node ***insert_p, struct rb_node **insert_parent, 4152 bool force, bool *leftmost); 4153 bool f2fs_check_rb_tree_consistence(struct f2fs_sb_info *sbi, 4154 struct rb_root_cached *root, bool check_key); 4155 unsigned int f2fs_shrink_extent_tree(struct f2fs_sb_info *sbi, int nr_shrink); 4156 void f2fs_init_extent_tree(struct inode *inode, struct page *ipage); 4157 void f2fs_drop_extent_tree(struct inode *inode); 4158 unsigned int f2fs_destroy_extent_node(struct inode *inode); 4159 void f2fs_destroy_extent_tree(struct inode *inode); 4160 bool f2fs_lookup_extent_cache(struct inode *inode, pgoff_t pgofs, 4161 struct extent_info *ei); 4162 void f2fs_update_extent_cache(struct dnode_of_data *dn); 4163 void f2fs_update_extent_cache_range(struct dnode_of_data *dn, 4164 pgoff_t fofs, block_t blkaddr, unsigned int len); 4165 void f2fs_init_extent_cache_info(struct f2fs_sb_info *sbi); 4166 int __init f2fs_create_extent_cache(void); 4167 void f2fs_destroy_extent_cache(void); 4168 4169 /* 4170 * sysfs.c 4171 */ 4172 #define MIN_RA_MUL 2 4173 #define MAX_RA_MUL 256 4174 4175 int __init f2fs_init_sysfs(void); 4176 void f2fs_exit_sysfs(void); 4177 int f2fs_register_sysfs(struct f2fs_sb_info *sbi); 4178 void f2fs_unregister_sysfs(struct f2fs_sb_info *sbi); 4179 4180 /* verity.c */ 4181 extern const struct fsverity_operations f2fs_verityops; 4182 4183 /* 4184 * crypto support 4185 */ 4186 static inline bool f2fs_encrypted_file(struct inode *inode) 4187 { 4188 return IS_ENCRYPTED(inode) && S_ISREG(inode->i_mode); 4189 } 4190 4191 static inline void f2fs_set_encrypted_inode(struct inode *inode) 4192 { 4193 #ifdef CONFIG_FS_ENCRYPTION 4194 file_set_encrypt(inode); 4195 f2fs_set_inode_flags(inode); 4196 #endif 4197 } 4198 4199 /* 4200 * Returns true if the reads of the inode's data need to undergo some 4201 * postprocessing step, like decryption or authenticity verification. 4202 */ 4203 static inline bool f2fs_post_read_required(struct inode *inode) 4204 { 4205 return f2fs_encrypted_file(inode) || fsverity_active(inode) || 4206 f2fs_compressed_file(inode); 4207 } 4208 4209 /* 4210 * compress.c 4211 */ 4212 #ifdef CONFIG_F2FS_FS_COMPRESSION 4213 bool f2fs_is_compressed_page(struct page *page); 4214 struct page *f2fs_compress_control_page(struct page *page); 4215 int f2fs_prepare_compress_overwrite(struct inode *inode, 4216 struct page **pagep, pgoff_t index, void **fsdata); 4217 bool f2fs_compress_write_end(struct inode *inode, void *fsdata, 4218 pgoff_t index, unsigned copied); 4219 int f2fs_truncate_partial_cluster(struct inode *inode, u64 from, bool lock); 4220 void f2fs_compress_write_end_io(struct bio *bio, struct page *page); 4221 bool f2fs_is_compress_backend_ready(struct inode *inode); 4222 int f2fs_init_compress_mempool(void); 4223 void f2fs_destroy_compress_mempool(void); 4224 void f2fs_decompress_cluster(struct decompress_io_ctx *dic, bool in_task); 4225 void f2fs_end_read_compressed_page(struct page *page, bool failed, 4226 block_t blkaddr, bool in_task); 4227 bool f2fs_cluster_is_empty(struct compress_ctx *cc); 4228 bool f2fs_cluster_can_merge_page(struct compress_ctx *cc, pgoff_t index); 4229 bool f2fs_all_cluster_page_ready(struct compress_ctx *cc, struct page **pages, 4230 int index, int nr_pages, bool uptodate); 4231 bool f2fs_sanity_check_cluster(struct dnode_of_data *dn); 4232 void f2fs_compress_ctx_add_page(struct compress_ctx *cc, struct page *page); 4233 int f2fs_write_multi_pages(struct compress_ctx *cc, 4234 int *submitted, 4235 struct writeback_control *wbc, 4236 enum iostat_type io_type); 4237 int f2fs_is_compressed_cluster(struct inode *inode, pgoff_t index); 4238 void f2fs_update_extent_tree_range_compressed(struct inode *inode, 4239 pgoff_t fofs, block_t blkaddr, unsigned int llen, 4240 unsigned int c_len); 4241 int f2fs_read_multi_pages(struct compress_ctx *cc, struct bio **bio_ret, 4242 unsigned nr_pages, sector_t *last_block_in_bio, 4243 bool is_readahead, bool for_write); 4244 struct decompress_io_ctx *f2fs_alloc_dic(struct compress_ctx *cc); 4245 void f2fs_decompress_end_io(struct decompress_io_ctx *dic, bool failed, 4246 bool in_task); 4247 void f2fs_put_page_dic(struct page *page, bool in_task); 4248 unsigned int f2fs_cluster_blocks_are_contiguous(struct dnode_of_data *dn); 4249 int f2fs_init_compress_ctx(struct compress_ctx *cc); 4250 void f2fs_destroy_compress_ctx(struct compress_ctx *cc, bool reuse); 4251 void f2fs_init_compress_info(struct f2fs_sb_info *sbi); 4252 int f2fs_init_compress_inode(struct f2fs_sb_info *sbi); 4253 void f2fs_destroy_compress_inode(struct f2fs_sb_info *sbi); 4254 int f2fs_init_page_array_cache(struct f2fs_sb_info *sbi); 4255 void f2fs_destroy_page_array_cache(struct f2fs_sb_info *sbi); 4256 int __init f2fs_init_compress_cache(void); 4257 void f2fs_destroy_compress_cache(void); 4258 struct address_space *COMPRESS_MAPPING(struct f2fs_sb_info *sbi); 4259 void f2fs_invalidate_compress_page(struct f2fs_sb_info *sbi, block_t blkaddr); 4260 void f2fs_cache_compressed_page(struct f2fs_sb_info *sbi, struct page *page, 4261 nid_t ino, block_t blkaddr); 4262 bool f2fs_load_compressed_page(struct f2fs_sb_info *sbi, struct page *page, 4263 block_t blkaddr); 4264 void f2fs_invalidate_compress_pages(struct f2fs_sb_info *sbi, nid_t ino); 4265 #define inc_compr_inode_stat(inode) \ 4266 do { \ 4267 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); \ 4268 sbi->compr_new_inode++; \ 4269 } while (0) 4270 #define add_compr_block_stat(inode, blocks) \ 4271 do { \ 4272 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); \ 4273 int diff = F2FS_I(inode)->i_cluster_size - blocks; \ 4274 sbi->compr_written_block += blocks; \ 4275 sbi->compr_saved_block += diff; \ 4276 } while (0) 4277 #else 4278 static inline bool f2fs_is_compressed_page(struct page *page) { return false; } 4279 static inline bool f2fs_is_compress_backend_ready(struct inode *inode) 4280 { 4281 if (!f2fs_compressed_file(inode)) 4282 return true; 4283 /* not support compression */ 4284 return false; 4285 } 4286 static inline struct page *f2fs_compress_control_page(struct page *page) 4287 { 4288 WARN_ON_ONCE(1); 4289 return ERR_PTR(-EINVAL); 4290 } 4291 static inline int f2fs_init_compress_mempool(void) { return 0; } 4292 static inline void f2fs_destroy_compress_mempool(void) { } 4293 static inline void f2fs_decompress_cluster(struct decompress_io_ctx *dic, 4294 bool in_task) { } 4295 static inline void f2fs_end_read_compressed_page(struct page *page, 4296 bool failed, block_t blkaddr, bool in_task) 4297 { 4298 WARN_ON_ONCE(1); 4299 } 4300 static inline void f2fs_put_page_dic(struct page *page, bool in_task) 4301 { 4302 WARN_ON_ONCE(1); 4303 } 4304 static inline unsigned int f2fs_cluster_blocks_are_contiguous(struct dnode_of_data *dn) { return 0; } 4305 static inline bool f2fs_sanity_check_cluster(struct dnode_of_data *dn) { return false; } 4306 static inline int f2fs_init_compress_inode(struct f2fs_sb_info *sbi) { return 0; } 4307 static inline void f2fs_destroy_compress_inode(struct f2fs_sb_info *sbi) { } 4308 static inline int f2fs_init_page_array_cache(struct f2fs_sb_info *sbi) { return 0; } 4309 static inline void f2fs_destroy_page_array_cache(struct f2fs_sb_info *sbi) { } 4310 static inline int __init f2fs_init_compress_cache(void) { return 0; } 4311 static inline void f2fs_destroy_compress_cache(void) { } 4312 static inline void f2fs_invalidate_compress_page(struct f2fs_sb_info *sbi, 4313 block_t blkaddr) { } 4314 static inline void f2fs_cache_compressed_page(struct f2fs_sb_info *sbi, 4315 struct page *page, nid_t ino, block_t blkaddr) { } 4316 static inline bool f2fs_load_compressed_page(struct f2fs_sb_info *sbi, 4317 struct page *page, block_t blkaddr) { return false; } 4318 static inline void f2fs_invalidate_compress_pages(struct f2fs_sb_info *sbi, 4319 nid_t ino) { } 4320 #define inc_compr_inode_stat(inode) do { } while (0) 4321 static inline void f2fs_update_extent_tree_range_compressed(struct inode *inode, 4322 pgoff_t fofs, block_t blkaddr, unsigned int llen, 4323 unsigned int c_len) { } 4324 #endif 4325 4326 static inline int set_compress_context(struct inode *inode) 4327 { 4328 #ifdef CONFIG_F2FS_FS_COMPRESSION 4329 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 4330 4331 F2FS_I(inode)->i_compress_algorithm = 4332 F2FS_OPTION(sbi).compress_algorithm; 4333 F2FS_I(inode)->i_log_cluster_size = 4334 F2FS_OPTION(sbi).compress_log_size; 4335 F2FS_I(inode)->i_compress_flag = 4336 F2FS_OPTION(sbi).compress_chksum ? 4337 1 << COMPRESS_CHKSUM : 0; 4338 F2FS_I(inode)->i_cluster_size = 4339 1 << F2FS_I(inode)->i_log_cluster_size; 4340 if ((F2FS_I(inode)->i_compress_algorithm == COMPRESS_LZ4 || 4341 F2FS_I(inode)->i_compress_algorithm == COMPRESS_ZSTD) && 4342 F2FS_OPTION(sbi).compress_level) 4343 F2FS_I(inode)->i_compress_flag |= 4344 F2FS_OPTION(sbi).compress_level << 4345 COMPRESS_LEVEL_OFFSET; 4346 F2FS_I(inode)->i_flags |= F2FS_COMPR_FL; 4347 set_inode_flag(inode, FI_COMPRESSED_FILE); 4348 stat_inc_compr_inode(inode); 4349 inc_compr_inode_stat(inode); 4350 f2fs_mark_inode_dirty_sync(inode, true); 4351 return 0; 4352 #else 4353 return -EOPNOTSUPP; 4354 #endif 4355 } 4356 4357 static inline bool f2fs_disable_compressed_file(struct inode *inode) 4358 { 4359 struct f2fs_inode_info *fi = F2FS_I(inode); 4360 4361 if (!f2fs_compressed_file(inode)) 4362 return true; 4363 if (S_ISREG(inode->i_mode) && F2FS_HAS_BLOCKS(inode)) 4364 return false; 4365 4366 fi->i_flags &= ~F2FS_COMPR_FL; 4367 stat_dec_compr_inode(inode); 4368 clear_inode_flag(inode, FI_COMPRESSED_FILE); 4369 f2fs_mark_inode_dirty_sync(inode, true); 4370 return true; 4371 } 4372 4373 #define F2FS_FEATURE_FUNCS(name, flagname) \ 4374 static inline int f2fs_sb_has_##name(struct f2fs_sb_info *sbi) \ 4375 { \ 4376 return F2FS_HAS_FEATURE(sbi, F2FS_FEATURE_##flagname); \ 4377 } 4378 4379 F2FS_FEATURE_FUNCS(encrypt, ENCRYPT); 4380 F2FS_FEATURE_FUNCS(blkzoned, BLKZONED); 4381 F2FS_FEATURE_FUNCS(extra_attr, EXTRA_ATTR); 4382 F2FS_FEATURE_FUNCS(project_quota, PRJQUOTA); 4383 F2FS_FEATURE_FUNCS(inode_chksum, INODE_CHKSUM); 4384 F2FS_FEATURE_FUNCS(flexible_inline_xattr, FLEXIBLE_INLINE_XATTR); 4385 F2FS_FEATURE_FUNCS(quota_ino, QUOTA_INO); 4386 F2FS_FEATURE_FUNCS(inode_crtime, INODE_CRTIME); 4387 F2FS_FEATURE_FUNCS(lost_found, LOST_FOUND); 4388 F2FS_FEATURE_FUNCS(verity, VERITY); 4389 F2FS_FEATURE_FUNCS(sb_chksum, SB_CHKSUM); 4390 F2FS_FEATURE_FUNCS(casefold, CASEFOLD); 4391 F2FS_FEATURE_FUNCS(compression, COMPRESSION); 4392 F2FS_FEATURE_FUNCS(readonly, RO); 4393 4394 static inline bool f2fs_may_extent_tree(struct inode *inode) 4395 { 4396 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 4397 4398 if (!test_opt(sbi, EXTENT_CACHE) || 4399 is_inode_flag_set(inode, FI_NO_EXTENT) || 4400 (is_inode_flag_set(inode, FI_COMPRESSED_FILE) && 4401 !f2fs_sb_has_readonly(sbi))) 4402 return false; 4403 4404 /* 4405 * for recovered files during mount do not create extents 4406 * if shrinker is not registered. 4407 */ 4408 if (list_empty(&sbi->s_list)) 4409 return false; 4410 4411 return S_ISREG(inode->i_mode); 4412 } 4413 4414 #ifdef CONFIG_BLK_DEV_ZONED 4415 static inline bool f2fs_blkz_is_seq(struct f2fs_sb_info *sbi, int devi, 4416 block_t blkaddr) 4417 { 4418 unsigned int zno = blkaddr >> sbi->log_blocks_per_blkz; 4419 4420 return test_bit(zno, FDEV(devi).blkz_seq); 4421 } 4422 #endif 4423 4424 static inline bool f2fs_hw_should_discard(struct f2fs_sb_info *sbi) 4425 { 4426 return f2fs_sb_has_blkzoned(sbi); 4427 } 4428 4429 static inline bool f2fs_bdev_support_discard(struct block_device *bdev) 4430 { 4431 return bdev_max_discard_sectors(bdev) || bdev_is_zoned(bdev); 4432 } 4433 4434 static inline bool f2fs_hw_support_discard(struct f2fs_sb_info *sbi) 4435 { 4436 int i; 4437 4438 if (!f2fs_is_multi_device(sbi)) 4439 return f2fs_bdev_support_discard(sbi->sb->s_bdev); 4440 4441 for (i = 0; i < sbi->s_ndevs; i++) 4442 if (f2fs_bdev_support_discard(FDEV(i).bdev)) 4443 return true; 4444 return false; 4445 } 4446 4447 static inline bool f2fs_realtime_discard_enable(struct f2fs_sb_info *sbi) 4448 { 4449 return (test_opt(sbi, DISCARD) && f2fs_hw_support_discard(sbi)) || 4450 f2fs_hw_should_discard(sbi); 4451 } 4452 4453 static inline bool f2fs_hw_is_readonly(struct f2fs_sb_info *sbi) 4454 { 4455 int i; 4456 4457 if (!f2fs_is_multi_device(sbi)) 4458 return bdev_read_only(sbi->sb->s_bdev); 4459 4460 for (i = 0; i < sbi->s_ndevs; i++) 4461 if (bdev_read_only(FDEV(i).bdev)) 4462 return true; 4463 return false; 4464 } 4465 4466 static inline bool f2fs_lfs_mode(struct f2fs_sb_info *sbi) 4467 { 4468 return F2FS_OPTION(sbi).fs_mode == FS_MODE_LFS; 4469 } 4470 4471 static inline bool f2fs_low_mem_mode(struct f2fs_sb_info *sbi) 4472 { 4473 return F2FS_OPTION(sbi).memory_mode == MEMORY_MODE_LOW; 4474 } 4475 4476 static inline bool f2fs_may_compress(struct inode *inode) 4477 { 4478 if (IS_SWAPFILE(inode) || f2fs_is_pinned_file(inode) || 4479 f2fs_is_atomic_file(inode) || f2fs_has_inline_data(inode)) 4480 return false; 4481 return S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode); 4482 } 4483 4484 static inline void f2fs_i_compr_blocks_update(struct inode *inode, 4485 u64 blocks, bool add) 4486 { 4487 struct f2fs_inode_info *fi = F2FS_I(inode); 4488 int diff = fi->i_cluster_size - blocks; 4489 4490 /* don't update i_compr_blocks if saved blocks were released */ 4491 if (!add && !atomic_read(&fi->i_compr_blocks)) 4492 return; 4493 4494 if (add) { 4495 atomic_add(diff, &fi->i_compr_blocks); 4496 stat_add_compr_blocks(inode, diff); 4497 } else { 4498 atomic_sub(diff, &fi->i_compr_blocks); 4499 stat_sub_compr_blocks(inode, diff); 4500 } 4501 f2fs_mark_inode_dirty_sync(inode, true); 4502 } 4503 4504 static inline bool f2fs_allow_multi_device_dio(struct f2fs_sb_info *sbi, 4505 int flag) 4506 { 4507 if (!f2fs_is_multi_device(sbi)) 4508 return false; 4509 if (flag != F2FS_GET_BLOCK_DIO) 4510 return false; 4511 return sbi->aligned_blksize; 4512 } 4513 4514 static inline bool f2fs_need_verity(const struct inode *inode, pgoff_t idx) 4515 { 4516 return fsverity_active(inode) && 4517 idx < DIV_ROUND_UP(inode->i_size, PAGE_SIZE); 4518 } 4519 4520 #ifdef CONFIG_F2FS_FAULT_INJECTION 4521 extern void f2fs_build_fault_attr(struct f2fs_sb_info *sbi, unsigned int rate, 4522 unsigned int type); 4523 #else 4524 #define f2fs_build_fault_attr(sbi, rate, type) do { } while (0) 4525 #endif 4526 4527 static inline bool is_journalled_quota(struct f2fs_sb_info *sbi) 4528 { 4529 #ifdef CONFIG_QUOTA 4530 if (f2fs_sb_has_quota_ino(sbi)) 4531 return true; 4532 if (F2FS_OPTION(sbi).s_qf_names[USRQUOTA] || 4533 F2FS_OPTION(sbi).s_qf_names[GRPQUOTA] || 4534 F2FS_OPTION(sbi).s_qf_names[PRJQUOTA]) 4535 return true; 4536 #endif 4537 return false; 4538 } 4539 4540 static inline bool f2fs_block_unit_discard(struct f2fs_sb_info *sbi) 4541 { 4542 return F2FS_OPTION(sbi).discard_unit == DISCARD_UNIT_BLOCK; 4543 } 4544 4545 static inline void f2fs_io_schedule_timeout(long timeout) 4546 { 4547 set_current_state(TASK_UNINTERRUPTIBLE); 4548 io_schedule_timeout(timeout); 4549 } 4550 4551 static inline void f2fs_handle_page_eio(struct f2fs_sb_info *sbi, pgoff_t ofs, 4552 enum page_type type) 4553 { 4554 if (unlikely(f2fs_cp_error(sbi))) 4555 return; 4556 4557 if (ofs == sbi->page_eio_ofs[type]) { 4558 if (sbi->page_eio_cnt[type]++ == MAX_RETRY_PAGE_EIO) 4559 set_ckpt_flags(sbi, CP_ERROR_FLAG); 4560 } else { 4561 sbi->page_eio_ofs[type] = ofs; 4562 sbi->page_eio_cnt[type] = 0; 4563 } 4564 } 4565 4566 #define EFSBADCRC EBADMSG /* Bad CRC detected */ 4567 #define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */ 4568 4569 #endif /* _LINUX_F2FS_H */ 4570