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_BLKADDR, 64 FAULT_MAX, 65 }; 66 67 #ifdef CONFIG_F2FS_FAULT_INJECTION 68 #define F2FS_ALL_FAULT_TYPE ((1 << FAULT_MAX) - 1) 69 70 struct f2fs_fault_info { 71 atomic_t inject_ops; 72 unsigned int inject_rate; 73 unsigned int inject_type; 74 }; 75 76 extern const char *f2fs_fault_name[FAULT_MAX]; 77 #define IS_FAULT_SET(fi, type) ((fi)->inject_type & (1 << (type))) 78 #endif 79 80 /* 81 * For mount options 82 */ 83 #define F2FS_MOUNT_DISABLE_ROLL_FORWARD 0x00000002 84 #define F2FS_MOUNT_DISCARD 0x00000004 85 #define F2FS_MOUNT_NOHEAP 0x00000008 86 #define F2FS_MOUNT_XATTR_USER 0x00000010 87 #define F2FS_MOUNT_POSIX_ACL 0x00000020 88 #define F2FS_MOUNT_DISABLE_EXT_IDENTIFY 0x00000040 89 #define F2FS_MOUNT_INLINE_XATTR 0x00000080 90 #define F2FS_MOUNT_INLINE_DATA 0x00000100 91 #define F2FS_MOUNT_INLINE_DENTRY 0x00000200 92 #define F2FS_MOUNT_FLUSH_MERGE 0x00000400 93 #define F2FS_MOUNT_NOBARRIER 0x00000800 94 #define F2FS_MOUNT_FASTBOOT 0x00001000 95 #define F2FS_MOUNT_READ_EXTENT_CACHE 0x00002000 96 #define F2FS_MOUNT_DATA_FLUSH 0x00008000 97 #define F2FS_MOUNT_FAULT_INJECTION 0x00010000 98 #define F2FS_MOUNT_USRQUOTA 0x00080000 99 #define F2FS_MOUNT_GRPQUOTA 0x00100000 100 #define F2FS_MOUNT_PRJQUOTA 0x00200000 101 #define F2FS_MOUNT_QUOTA 0x00400000 102 #define F2FS_MOUNT_INLINE_XATTR_SIZE 0x00800000 103 #define F2FS_MOUNT_RESERVE_ROOT 0x01000000 104 #define F2FS_MOUNT_DISABLE_CHECKPOINT 0x02000000 105 #define F2FS_MOUNT_NORECOVERY 0x04000000 106 #define F2FS_MOUNT_ATGC 0x08000000 107 #define F2FS_MOUNT_MERGE_CHECKPOINT 0x10000000 108 #define F2FS_MOUNT_GC_MERGE 0x20000000 109 #define F2FS_MOUNT_COMPRESS_CACHE 0x40000000 110 #define F2FS_MOUNT_AGE_EXTENT_CACHE 0x80000000 111 112 #define F2FS_OPTION(sbi) ((sbi)->mount_opt) 113 #define clear_opt(sbi, option) (F2FS_OPTION(sbi).opt &= ~F2FS_MOUNT_##option) 114 #define set_opt(sbi, option) (F2FS_OPTION(sbi).opt |= F2FS_MOUNT_##option) 115 #define test_opt(sbi, option) (F2FS_OPTION(sbi).opt & F2FS_MOUNT_##option) 116 117 #define ver_after(a, b) (typecheck(unsigned long long, a) && \ 118 typecheck(unsigned long long, b) && \ 119 ((long long)((a) - (b)) > 0)) 120 121 typedef u32 block_t; /* 122 * should not change u32, since it is the on-disk block 123 * address format, __le32. 124 */ 125 typedef u32 nid_t; 126 127 #define COMPRESS_EXT_NUM 16 128 129 /* 130 * An implementation of an rwsem that is explicitly unfair to readers. This 131 * prevents priority inversion when a low-priority reader acquires the read lock 132 * while sleeping on the write lock but the write lock is needed by 133 * higher-priority clients. 134 */ 135 136 struct f2fs_rwsem { 137 struct rw_semaphore internal_rwsem; 138 #ifdef CONFIG_F2FS_UNFAIR_RWSEM 139 wait_queue_head_t read_waiters; 140 #endif 141 }; 142 143 struct f2fs_mount_info { 144 unsigned int opt; 145 int write_io_size_bits; /* Write IO size bits */ 146 block_t root_reserved_blocks; /* root reserved blocks */ 147 kuid_t s_resuid; /* reserved blocks for uid */ 148 kgid_t s_resgid; /* reserved blocks for gid */ 149 int active_logs; /* # of active logs */ 150 int inline_xattr_size; /* inline xattr size */ 151 #ifdef CONFIG_F2FS_FAULT_INJECTION 152 struct f2fs_fault_info fault_info; /* For fault injection */ 153 #endif 154 #ifdef CONFIG_QUOTA 155 /* Names of quota files with journalled quota */ 156 char *s_qf_names[MAXQUOTAS]; 157 int s_jquota_fmt; /* Format of quota to use */ 158 #endif 159 /* For which write hints are passed down to block layer */ 160 int alloc_mode; /* segment allocation policy */ 161 int fsync_mode; /* fsync policy */ 162 int fs_mode; /* fs mode: LFS or ADAPTIVE */ 163 int bggc_mode; /* bggc mode: off, on or sync */ 164 int memory_mode; /* memory mode */ 165 int discard_unit; /* 166 * discard command's offset/size should 167 * be aligned to this unit: block, 168 * segment or section 169 */ 170 struct fscrypt_dummy_policy dummy_enc_policy; /* test dummy encryption */ 171 block_t unusable_cap_perc; /* percentage for cap */ 172 block_t unusable_cap; /* Amount of space allowed to be 173 * unusable when disabling checkpoint 174 */ 175 176 /* For compression */ 177 unsigned char compress_algorithm; /* algorithm type */ 178 unsigned char compress_log_size; /* cluster log size */ 179 unsigned char compress_level; /* compress level */ 180 bool compress_chksum; /* compressed data chksum */ 181 unsigned char compress_ext_cnt; /* extension count */ 182 unsigned char nocompress_ext_cnt; /* nocompress extension count */ 183 int compress_mode; /* compression mode */ 184 unsigned char extensions[COMPRESS_EXT_NUM][F2FS_EXTENSION_LEN]; /* extensions */ 185 unsigned char noextensions[COMPRESS_EXT_NUM][F2FS_EXTENSION_LEN]; /* extensions */ 186 }; 187 188 #define F2FS_FEATURE_ENCRYPT 0x0001 189 #define F2FS_FEATURE_BLKZONED 0x0002 190 #define F2FS_FEATURE_ATOMIC_WRITE 0x0004 191 #define F2FS_FEATURE_EXTRA_ATTR 0x0008 192 #define F2FS_FEATURE_PRJQUOTA 0x0010 193 #define F2FS_FEATURE_INODE_CHKSUM 0x0020 194 #define F2FS_FEATURE_FLEXIBLE_INLINE_XATTR 0x0040 195 #define F2FS_FEATURE_QUOTA_INO 0x0080 196 #define F2FS_FEATURE_INODE_CRTIME 0x0100 197 #define F2FS_FEATURE_LOST_FOUND 0x0200 198 #define F2FS_FEATURE_VERITY 0x0400 199 #define F2FS_FEATURE_SB_CHKSUM 0x0800 200 #define F2FS_FEATURE_CASEFOLD 0x1000 201 #define F2FS_FEATURE_COMPRESSION 0x2000 202 #define F2FS_FEATURE_RO 0x4000 203 204 #define __F2FS_HAS_FEATURE(raw_super, mask) \ 205 ((raw_super->feature & cpu_to_le32(mask)) != 0) 206 #define F2FS_HAS_FEATURE(sbi, mask) __F2FS_HAS_FEATURE(sbi->raw_super, mask) 207 208 /* 209 * Default values for user and/or group using reserved blocks 210 */ 211 #define F2FS_DEF_RESUID 0 212 #define F2FS_DEF_RESGID 0 213 214 /* 215 * For checkpoint manager 216 */ 217 enum { 218 NAT_BITMAP, 219 SIT_BITMAP 220 }; 221 222 #define CP_UMOUNT 0x00000001 223 #define CP_FASTBOOT 0x00000002 224 #define CP_SYNC 0x00000004 225 #define CP_RECOVERY 0x00000008 226 #define CP_DISCARD 0x00000010 227 #define CP_TRIMMED 0x00000020 228 #define CP_PAUSE 0x00000040 229 #define CP_RESIZE 0x00000080 230 231 #define DEF_MAX_DISCARD_REQUEST 8 /* issue 8 discards per round */ 232 #define DEF_MIN_DISCARD_ISSUE_TIME 50 /* 50 ms, if exists */ 233 #define DEF_MID_DISCARD_ISSUE_TIME 500 /* 500 ms, if device busy */ 234 #define DEF_MAX_DISCARD_ISSUE_TIME 60000 /* 60 s, if no candidates */ 235 #define DEF_DISCARD_URGENT_UTIL 80 /* do more discard over 80% */ 236 #define DEF_CP_INTERVAL 60 /* 60 secs */ 237 #define DEF_IDLE_INTERVAL 5 /* 5 secs */ 238 #define DEF_DISABLE_INTERVAL 5 /* 5 secs */ 239 #define DEF_DISABLE_QUICK_INTERVAL 1 /* 1 secs */ 240 #define DEF_UMOUNT_DISCARD_TIMEOUT 5 /* 5 secs */ 241 242 struct cp_control { 243 int reason; 244 __u64 trim_start; 245 __u64 trim_end; 246 __u64 trim_minlen; 247 }; 248 249 /* 250 * indicate meta/data type 251 */ 252 enum { 253 META_CP, 254 META_NAT, 255 META_SIT, 256 META_SSA, 257 META_MAX, 258 META_POR, 259 DATA_GENERIC, /* check range only */ 260 DATA_GENERIC_ENHANCE, /* strong check on range and segment bitmap */ 261 DATA_GENERIC_ENHANCE_READ, /* 262 * strong check on range and segment 263 * bitmap but no warning due to race 264 * condition of read on truncated area 265 * by extent_cache 266 */ 267 DATA_GENERIC_ENHANCE_UPDATE, /* 268 * strong check on range and segment 269 * bitmap for update case 270 */ 271 META_GENERIC, 272 }; 273 274 /* for the list of ino */ 275 enum { 276 ORPHAN_INO, /* for orphan ino list */ 277 APPEND_INO, /* for append ino list */ 278 UPDATE_INO, /* for update ino list */ 279 TRANS_DIR_INO, /* for transactions dir ino list */ 280 FLUSH_INO, /* for multiple device flushing */ 281 MAX_INO_ENTRY, /* max. list */ 282 }; 283 284 struct ino_entry { 285 struct list_head list; /* list head */ 286 nid_t ino; /* inode number */ 287 unsigned int dirty_device; /* dirty device bitmap */ 288 }; 289 290 /* for the list of inodes to be GCed */ 291 struct inode_entry { 292 struct list_head list; /* list head */ 293 struct inode *inode; /* vfs inode pointer */ 294 }; 295 296 struct fsync_node_entry { 297 struct list_head list; /* list head */ 298 struct page *page; /* warm node page pointer */ 299 unsigned int seq_id; /* sequence id */ 300 }; 301 302 struct ckpt_req { 303 struct completion wait; /* completion for checkpoint done */ 304 struct llist_node llnode; /* llist_node to be linked in wait queue */ 305 int ret; /* return code of checkpoint */ 306 ktime_t queue_time; /* request queued time */ 307 }; 308 309 struct ckpt_req_control { 310 struct task_struct *f2fs_issue_ckpt; /* checkpoint task */ 311 int ckpt_thread_ioprio; /* checkpoint merge thread ioprio */ 312 wait_queue_head_t ckpt_wait_queue; /* waiting queue for wake-up */ 313 atomic_t issued_ckpt; /* # of actually issued ckpts */ 314 atomic_t total_ckpt; /* # of total ckpts */ 315 atomic_t queued_ckpt; /* # of queued ckpts */ 316 struct llist_head issue_list; /* list for command issue */ 317 spinlock_t stat_lock; /* lock for below checkpoint time stats */ 318 unsigned int cur_time; /* cur wait time in msec for currently issued checkpoint */ 319 unsigned int peak_time; /* peak wait time in msec until now */ 320 }; 321 322 /* for the bitmap indicate blocks to be discarded */ 323 struct discard_entry { 324 struct list_head list; /* list head */ 325 block_t start_blkaddr; /* start blockaddr of current segment */ 326 unsigned char discard_map[SIT_VBLOCK_MAP_SIZE]; /* segment discard bitmap */ 327 }; 328 329 /* minimum discard granularity, unit: block count */ 330 #define MIN_DISCARD_GRANULARITY 1 331 /* default discard granularity of inner discard thread, unit: block count */ 332 #define DEFAULT_DISCARD_GRANULARITY 16 333 /* default maximum discard granularity of ordered discard, unit: block count */ 334 #define DEFAULT_MAX_ORDERED_DISCARD_GRANULARITY 16 335 336 /* max discard pend list number */ 337 #define MAX_PLIST_NUM 512 338 #define plist_idx(blk_num) ((blk_num) >= MAX_PLIST_NUM ? \ 339 (MAX_PLIST_NUM - 1) : ((blk_num) - 1)) 340 341 enum { 342 D_PREP, /* initial */ 343 D_PARTIAL, /* partially submitted */ 344 D_SUBMIT, /* all submitted */ 345 D_DONE, /* finished */ 346 }; 347 348 struct discard_info { 349 block_t lstart; /* logical start address */ 350 block_t len; /* length */ 351 block_t start; /* actual start address in dev */ 352 }; 353 354 struct discard_cmd { 355 struct rb_node rb_node; /* rb node located in rb-tree */ 356 union { 357 struct { 358 block_t lstart; /* logical start address */ 359 block_t len; /* length */ 360 block_t start; /* actual start address in dev */ 361 }; 362 struct discard_info di; /* discard info */ 363 364 }; 365 struct list_head list; /* command list */ 366 struct completion wait; /* compleation */ 367 struct block_device *bdev; /* bdev */ 368 unsigned short ref; /* reference count */ 369 unsigned char state; /* state */ 370 unsigned char queued; /* queued discard */ 371 int error; /* bio error */ 372 spinlock_t lock; /* for state/bio_ref updating */ 373 unsigned short bio_ref; /* bio reference count */ 374 }; 375 376 enum { 377 DPOLICY_BG, 378 DPOLICY_FORCE, 379 DPOLICY_FSTRIM, 380 DPOLICY_UMOUNT, 381 MAX_DPOLICY, 382 }; 383 384 struct discard_policy { 385 int type; /* type of discard */ 386 unsigned int min_interval; /* used for candidates exist */ 387 unsigned int mid_interval; /* used for device busy */ 388 unsigned int max_interval; /* used for candidates not exist */ 389 unsigned int max_requests; /* # of discards issued per round */ 390 unsigned int io_aware_gran; /* minimum granularity discard not be aware of I/O */ 391 bool io_aware; /* issue discard in idle time */ 392 bool sync; /* submit discard with REQ_SYNC flag */ 393 bool ordered; /* issue discard by lba order */ 394 bool timeout; /* discard timeout for put_super */ 395 unsigned int granularity; /* discard granularity */ 396 }; 397 398 struct discard_cmd_control { 399 struct task_struct *f2fs_issue_discard; /* discard thread */ 400 struct list_head entry_list; /* 4KB discard entry list */ 401 struct list_head pend_list[MAX_PLIST_NUM];/* store pending entries */ 402 struct list_head wait_list; /* store on-flushing entries */ 403 struct list_head fstrim_list; /* in-flight discard from fstrim */ 404 wait_queue_head_t discard_wait_queue; /* waiting queue for wake-up */ 405 unsigned int discard_wake; /* to wake up discard thread */ 406 struct mutex cmd_lock; 407 unsigned int nr_discards; /* # of discards in the list */ 408 unsigned int max_discards; /* max. discards to be issued */ 409 unsigned int max_discard_request; /* max. discard request per round */ 410 unsigned int min_discard_issue_time; /* min. interval between discard issue */ 411 unsigned int mid_discard_issue_time; /* mid. interval between discard issue */ 412 unsigned int max_discard_issue_time; /* max. interval between discard issue */ 413 unsigned int discard_urgent_util; /* utilization which issue discard proactively */ 414 unsigned int discard_granularity; /* discard granularity */ 415 unsigned int max_ordered_discard; /* maximum discard granularity issued by lba order */ 416 unsigned int undiscard_blks; /* # of undiscard blocks */ 417 unsigned int next_pos; /* next discard position */ 418 atomic_t issued_discard; /* # of issued discard */ 419 atomic_t queued_discard; /* # of queued discard */ 420 atomic_t discard_cmd_cnt; /* # of cached cmd count */ 421 struct rb_root_cached root; /* root of discard rb-tree */ 422 bool rbtree_check; /* config for consistence check */ 423 }; 424 425 /* for the list of fsync inodes, used only during recovery */ 426 struct fsync_inode_entry { 427 struct list_head list; /* list head */ 428 struct inode *inode; /* vfs inode pointer */ 429 block_t blkaddr; /* block address locating the last fsync */ 430 block_t last_dentry; /* block address locating the last dentry */ 431 }; 432 433 #define nats_in_cursum(jnl) (le16_to_cpu((jnl)->n_nats)) 434 #define sits_in_cursum(jnl) (le16_to_cpu((jnl)->n_sits)) 435 436 #define nat_in_journal(jnl, i) ((jnl)->nat_j.entries[i].ne) 437 #define nid_in_journal(jnl, i) ((jnl)->nat_j.entries[i].nid) 438 #define sit_in_journal(jnl, i) ((jnl)->sit_j.entries[i].se) 439 #define segno_in_journal(jnl, i) ((jnl)->sit_j.entries[i].segno) 440 441 #define MAX_NAT_JENTRIES(jnl) (NAT_JOURNAL_ENTRIES - nats_in_cursum(jnl)) 442 #define MAX_SIT_JENTRIES(jnl) (SIT_JOURNAL_ENTRIES - sits_in_cursum(jnl)) 443 444 static inline int update_nats_in_cursum(struct f2fs_journal *journal, int i) 445 { 446 int before = nats_in_cursum(journal); 447 448 journal->n_nats = cpu_to_le16(before + i); 449 return before; 450 } 451 452 static inline int update_sits_in_cursum(struct f2fs_journal *journal, int i) 453 { 454 int before = sits_in_cursum(journal); 455 456 journal->n_sits = cpu_to_le16(before + i); 457 return before; 458 } 459 460 static inline bool __has_cursum_space(struct f2fs_journal *journal, 461 int size, int type) 462 { 463 if (type == NAT_JOURNAL) 464 return size <= MAX_NAT_JENTRIES(journal); 465 return size <= MAX_SIT_JENTRIES(journal); 466 } 467 468 /* for inline stuff */ 469 #define DEF_INLINE_RESERVED_SIZE 1 470 static inline int get_extra_isize(struct inode *inode); 471 static inline int get_inline_xattr_addrs(struct inode *inode); 472 #define MAX_INLINE_DATA(inode) (sizeof(__le32) * \ 473 (CUR_ADDRS_PER_INODE(inode) - \ 474 get_inline_xattr_addrs(inode) - \ 475 DEF_INLINE_RESERVED_SIZE)) 476 477 /* for inline dir */ 478 #define NR_INLINE_DENTRY(inode) (MAX_INLINE_DATA(inode) * BITS_PER_BYTE / \ 479 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \ 480 BITS_PER_BYTE + 1)) 481 #define INLINE_DENTRY_BITMAP_SIZE(inode) \ 482 DIV_ROUND_UP(NR_INLINE_DENTRY(inode), BITS_PER_BYTE) 483 #define INLINE_RESERVED_SIZE(inode) (MAX_INLINE_DATA(inode) - \ 484 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \ 485 NR_INLINE_DENTRY(inode) + \ 486 INLINE_DENTRY_BITMAP_SIZE(inode))) 487 488 /* 489 * For INODE and NODE manager 490 */ 491 /* for directory operations */ 492 493 struct f2fs_filename { 494 /* 495 * The filename the user specified. This is NULL for some 496 * filesystem-internal operations, e.g. converting an inline directory 497 * to a non-inline one, or roll-forward recovering an encrypted dentry. 498 */ 499 const struct qstr *usr_fname; 500 501 /* 502 * The on-disk filename. For encrypted directories, this is encrypted. 503 * This may be NULL for lookups in an encrypted dir without the key. 504 */ 505 struct fscrypt_str disk_name; 506 507 /* The dirhash of this filename */ 508 f2fs_hash_t hash; 509 510 #ifdef CONFIG_FS_ENCRYPTION 511 /* 512 * For lookups in encrypted directories: either the buffer backing 513 * disk_name, or a buffer that holds the decoded no-key name. 514 */ 515 struct fscrypt_str crypto_buf; 516 #endif 517 #if IS_ENABLED(CONFIG_UNICODE) 518 /* 519 * For casefolded directories: the casefolded name, but it's left NULL 520 * if the original name is not valid Unicode, if the original name is 521 * "." or "..", if the directory is both casefolded and encrypted and 522 * its encryption key is unavailable, or if the filesystem is doing an 523 * internal operation where usr_fname is also NULL. In all these cases 524 * we fall back to treating the name as an opaque byte sequence. 525 */ 526 struct fscrypt_str cf_name; 527 #endif 528 }; 529 530 struct f2fs_dentry_ptr { 531 struct inode *inode; 532 void *bitmap; 533 struct f2fs_dir_entry *dentry; 534 __u8 (*filename)[F2FS_SLOT_LEN]; 535 int max; 536 int nr_bitmap; 537 }; 538 539 static inline void make_dentry_ptr_block(struct inode *inode, 540 struct f2fs_dentry_ptr *d, struct f2fs_dentry_block *t) 541 { 542 d->inode = inode; 543 d->max = NR_DENTRY_IN_BLOCK; 544 d->nr_bitmap = SIZE_OF_DENTRY_BITMAP; 545 d->bitmap = t->dentry_bitmap; 546 d->dentry = t->dentry; 547 d->filename = t->filename; 548 } 549 550 static inline void make_dentry_ptr_inline(struct inode *inode, 551 struct f2fs_dentry_ptr *d, void *t) 552 { 553 int entry_cnt = NR_INLINE_DENTRY(inode); 554 int bitmap_size = INLINE_DENTRY_BITMAP_SIZE(inode); 555 int reserved_size = INLINE_RESERVED_SIZE(inode); 556 557 d->inode = inode; 558 d->max = entry_cnt; 559 d->nr_bitmap = bitmap_size; 560 d->bitmap = t; 561 d->dentry = t + bitmap_size + reserved_size; 562 d->filename = t + bitmap_size + reserved_size + 563 SIZE_OF_DIR_ENTRY * entry_cnt; 564 } 565 566 /* 567 * XATTR_NODE_OFFSET stores xattrs to one node block per file keeping -1 568 * as its node offset to distinguish from index node blocks. 569 * But some bits are used to mark the node block. 570 */ 571 #define XATTR_NODE_OFFSET ((((unsigned int)-1) << OFFSET_BIT_SHIFT) \ 572 >> OFFSET_BIT_SHIFT) 573 enum { 574 ALLOC_NODE, /* allocate a new node page if needed */ 575 LOOKUP_NODE, /* look up a node without readahead */ 576 LOOKUP_NODE_RA, /* 577 * look up a node with readahead called 578 * by get_data_block. 579 */ 580 }; 581 582 #define DEFAULT_RETRY_IO_COUNT 8 /* maximum retry read IO or flush count */ 583 584 /* congestion wait timeout value, default: 20ms */ 585 #define DEFAULT_IO_TIMEOUT (msecs_to_jiffies(20)) 586 587 /* maximum retry quota flush count */ 588 #define DEFAULT_RETRY_QUOTA_FLUSH_COUNT 8 589 590 /* maximum retry of EIO'ed page */ 591 #define MAX_RETRY_PAGE_EIO 100 592 593 #define F2FS_LINK_MAX 0xffffffff /* maximum link count per file */ 594 595 #define MAX_DIR_RA_PAGES 4 /* maximum ra pages of dir */ 596 597 /* dirty segments threshold for triggering CP */ 598 #define DEFAULT_DIRTY_THRESHOLD 4 599 600 #define RECOVERY_MAX_RA_BLOCKS BIO_MAX_VECS 601 #define RECOVERY_MIN_RA_BLOCKS 1 602 603 #define F2FS_ONSTACK_PAGES 16 /* nr of onstack pages */ 604 605 /* for in-memory extent cache entry */ 606 #define F2FS_MIN_EXTENT_LEN 64 /* minimum extent length */ 607 608 /* number of extent info in extent cache we try to shrink */ 609 #define READ_EXTENT_CACHE_SHRINK_NUMBER 128 610 611 /* number of age extent info in extent cache we try to shrink */ 612 #define AGE_EXTENT_CACHE_SHRINK_NUMBER 128 613 #define LAST_AGE_WEIGHT 30 614 #define SAME_AGE_REGION 1024 615 616 /* 617 * Define data block with age less than 1GB as hot data 618 * define data block with age less than 10GB but more than 1GB as warm data 619 */ 620 #define DEF_HOT_DATA_AGE_THRESHOLD 262144 621 #define DEF_WARM_DATA_AGE_THRESHOLD 2621440 622 623 /* extent cache type */ 624 enum extent_type { 625 EX_READ, 626 EX_BLOCK_AGE, 627 NR_EXTENT_CACHES, 628 }; 629 630 struct rb_entry { 631 struct rb_node rb_node; /* rb node located in rb-tree */ 632 union { 633 struct { 634 unsigned int ofs; /* start offset of the entry */ 635 unsigned int len; /* length of the entry */ 636 }; 637 unsigned long long key; /* 64-bits key */ 638 } __packed; 639 }; 640 641 struct extent_info { 642 unsigned int fofs; /* start offset in a file */ 643 unsigned int len; /* length of the extent */ 644 union { 645 /* read extent_cache */ 646 struct { 647 /* start block address of the extent */ 648 block_t blk; 649 #ifdef CONFIG_F2FS_FS_COMPRESSION 650 /* physical extent length of compressed blocks */ 651 unsigned int c_len; 652 #endif 653 }; 654 /* block age extent_cache */ 655 struct { 656 /* block age of the extent */ 657 unsigned long long age; 658 /* last total blocks allocated */ 659 unsigned long long last_blocks; 660 }; 661 }; 662 }; 663 664 struct extent_node { 665 struct rb_node rb_node; /* rb node located in rb-tree */ 666 struct extent_info ei; /* extent info */ 667 struct list_head list; /* node in global extent list of sbi */ 668 struct extent_tree *et; /* extent tree pointer */ 669 }; 670 671 struct extent_tree { 672 nid_t ino; /* inode number */ 673 enum extent_type type; /* keep the extent tree type */ 674 struct rb_root_cached root; /* root of extent info rb-tree */ 675 struct extent_node *cached_en; /* recently accessed extent node */ 676 struct list_head list; /* to be used by sbi->zombie_list */ 677 rwlock_t lock; /* protect extent info rb-tree */ 678 atomic_t node_cnt; /* # of extent node in rb-tree*/ 679 bool largest_updated; /* largest extent updated */ 680 struct extent_info largest; /* largest cached extent for EX_READ */ 681 }; 682 683 struct extent_tree_info { 684 struct radix_tree_root extent_tree_root;/* cache extent cache entries */ 685 struct mutex extent_tree_lock; /* locking extent radix tree */ 686 struct list_head extent_list; /* lru list for shrinker */ 687 spinlock_t extent_lock; /* locking extent lru list */ 688 atomic_t total_ext_tree; /* extent tree count */ 689 struct list_head zombie_list; /* extent zombie tree list */ 690 atomic_t total_zombie_tree; /* extent zombie tree count */ 691 atomic_t total_ext_node; /* extent info count */ 692 }; 693 694 /* 695 * This structure is taken from ext4_map_blocks. 696 * 697 * Note that, however, f2fs uses NEW and MAPPED flags for f2fs_map_blocks(). 698 */ 699 #define F2FS_MAP_NEW (1 << BH_New) 700 #define F2FS_MAP_MAPPED (1 << BH_Mapped) 701 #define F2FS_MAP_UNWRITTEN (1 << BH_Unwritten) 702 #define F2FS_MAP_FLAGS (F2FS_MAP_NEW | F2FS_MAP_MAPPED |\ 703 F2FS_MAP_UNWRITTEN) 704 705 struct f2fs_map_blocks { 706 struct block_device *m_bdev; /* for multi-device dio */ 707 block_t m_pblk; 708 block_t m_lblk; 709 unsigned int m_len; 710 unsigned int m_flags; 711 pgoff_t *m_next_pgofs; /* point next possible non-hole pgofs */ 712 pgoff_t *m_next_extent; /* point to next possible extent */ 713 int m_seg_type; 714 bool m_may_create; /* indicate it is from write path */ 715 bool m_multidev_dio; /* indicate it allows multi-device dio */ 716 }; 717 718 /* for flag in get_data_block */ 719 enum { 720 F2FS_GET_BLOCK_DEFAULT, 721 F2FS_GET_BLOCK_FIEMAP, 722 F2FS_GET_BLOCK_BMAP, 723 F2FS_GET_BLOCK_DIO, 724 F2FS_GET_BLOCK_PRE_DIO, 725 F2FS_GET_BLOCK_PRE_AIO, 726 F2FS_GET_BLOCK_PRECACHE, 727 }; 728 729 /* 730 * i_advise uses FADVISE_XXX_BIT. We can add additional hints later. 731 */ 732 #define FADVISE_COLD_BIT 0x01 733 #define FADVISE_LOST_PINO_BIT 0x02 734 #define FADVISE_ENCRYPT_BIT 0x04 735 #define FADVISE_ENC_NAME_BIT 0x08 736 #define FADVISE_KEEP_SIZE_BIT 0x10 737 #define FADVISE_HOT_BIT 0x20 738 #define FADVISE_VERITY_BIT 0x40 739 #define FADVISE_TRUNC_BIT 0x80 740 741 #define FADVISE_MODIFIABLE_BITS (FADVISE_COLD_BIT | FADVISE_HOT_BIT) 742 743 #define file_is_cold(inode) is_file(inode, FADVISE_COLD_BIT) 744 #define file_set_cold(inode) set_file(inode, FADVISE_COLD_BIT) 745 #define file_clear_cold(inode) clear_file(inode, FADVISE_COLD_BIT) 746 747 #define file_wrong_pino(inode) is_file(inode, FADVISE_LOST_PINO_BIT) 748 #define file_lost_pino(inode) set_file(inode, FADVISE_LOST_PINO_BIT) 749 #define file_got_pino(inode) clear_file(inode, FADVISE_LOST_PINO_BIT) 750 751 #define file_is_encrypt(inode) is_file(inode, FADVISE_ENCRYPT_BIT) 752 #define file_set_encrypt(inode) set_file(inode, FADVISE_ENCRYPT_BIT) 753 754 #define file_enc_name(inode) is_file(inode, FADVISE_ENC_NAME_BIT) 755 #define file_set_enc_name(inode) set_file(inode, FADVISE_ENC_NAME_BIT) 756 757 #define file_keep_isize(inode) is_file(inode, FADVISE_KEEP_SIZE_BIT) 758 #define file_set_keep_isize(inode) set_file(inode, FADVISE_KEEP_SIZE_BIT) 759 760 #define file_is_hot(inode) is_file(inode, FADVISE_HOT_BIT) 761 #define file_set_hot(inode) set_file(inode, FADVISE_HOT_BIT) 762 #define file_clear_hot(inode) clear_file(inode, FADVISE_HOT_BIT) 763 764 #define file_is_verity(inode) is_file(inode, FADVISE_VERITY_BIT) 765 #define file_set_verity(inode) set_file(inode, FADVISE_VERITY_BIT) 766 767 #define file_should_truncate(inode) is_file(inode, FADVISE_TRUNC_BIT) 768 #define file_need_truncate(inode) set_file(inode, FADVISE_TRUNC_BIT) 769 #define file_dont_truncate(inode) clear_file(inode, FADVISE_TRUNC_BIT) 770 771 #define DEF_DIR_LEVEL 0 772 773 enum { 774 GC_FAILURE_PIN, 775 MAX_GC_FAILURE 776 }; 777 778 /* used for f2fs_inode_info->flags */ 779 enum { 780 FI_NEW_INODE, /* indicate newly allocated inode */ 781 FI_DIRTY_INODE, /* indicate inode is dirty or not */ 782 FI_AUTO_RECOVER, /* indicate inode is recoverable */ 783 FI_DIRTY_DIR, /* indicate directory has dirty pages */ 784 FI_INC_LINK, /* need to increment i_nlink */ 785 FI_ACL_MODE, /* indicate acl mode */ 786 FI_NO_ALLOC, /* should not allocate any blocks */ 787 FI_FREE_NID, /* free allocated nide */ 788 FI_NO_EXTENT, /* not to use the extent cache */ 789 FI_INLINE_XATTR, /* used for inline xattr */ 790 FI_INLINE_DATA, /* used for inline data*/ 791 FI_INLINE_DENTRY, /* used for inline dentry */ 792 FI_APPEND_WRITE, /* inode has appended data */ 793 FI_UPDATE_WRITE, /* inode has in-place-update data */ 794 FI_NEED_IPU, /* used for ipu per file */ 795 FI_ATOMIC_FILE, /* indicate atomic file */ 796 FI_FIRST_BLOCK_WRITTEN, /* indicate #0 data block was written */ 797 FI_DROP_CACHE, /* drop dirty page cache */ 798 FI_DATA_EXIST, /* indicate data exists */ 799 FI_INLINE_DOTS, /* indicate inline dot dentries */ 800 FI_SKIP_WRITES, /* should skip data page writeback */ 801 FI_OPU_WRITE, /* used for opu per file */ 802 FI_DIRTY_FILE, /* indicate regular/symlink has dirty pages */ 803 FI_PREALLOCATED_ALL, /* all blocks for write were preallocated */ 804 FI_HOT_DATA, /* indicate file is hot */ 805 FI_EXTRA_ATTR, /* indicate file has extra attribute */ 806 FI_PROJ_INHERIT, /* indicate file inherits projectid */ 807 FI_PIN_FILE, /* indicate file should not be gced */ 808 FI_VERITY_IN_PROGRESS, /* building fs-verity Merkle tree */ 809 FI_COMPRESSED_FILE, /* indicate file's data can be compressed */ 810 FI_COMPRESS_CORRUPT, /* indicate compressed cluster is corrupted */ 811 FI_MMAP_FILE, /* indicate file was mmapped */ 812 FI_ENABLE_COMPRESS, /* enable compression in "user" compression mode */ 813 FI_COMPRESS_RELEASED, /* compressed blocks were released */ 814 FI_ALIGNED_WRITE, /* enable aligned write */ 815 FI_COW_FILE, /* indicate COW file */ 816 FI_ATOMIC_COMMITTED, /* indicate atomic commit completed except disk sync */ 817 FI_ATOMIC_REPLACE, /* indicate atomic replace */ 818 FI_MAX, /* max flag, never be used */ 819 }; 820 821 struct f2fs_inode_info { 822 struct inode vfs_inode; /* serve a vfs inode */ 823 unsigned long i_flags; /* keep an inode flags for ioctl */ 824 unsigned char i_advise; /* use to give file attribute hints */ 825 unsigned char i_dir_level; /* use for dentry level for large dir */ 826 unsigned int i_current_depth; /* only for directory depth */ 827 /* for gc failure statistic */ 828 unsigned int i_gc_failures[MAX_GC_FAILURE]; 829 unsigned int i_pino; /* parent inode number */ 830 umode_t i_acl_mode; /* keep file acl mode temporarily */ 831 832 /* Use below internally in f2fs*/ 833 unsigned long flags[BITS_TO_LONGS(FI_MAX)]; /* use to pass per-file flags */ 834 struct f2fs_rwsem i_sem; /* protect fi info */ 835 atomic_t dirty_pages; /* # of dirty pages */ 836 f2fs_hash_t chash; /* hash value of given file name */ 837 unsigned int clevel; /* maximum level of given file name */ 838 struct task_struct *task; /* lookup and create consistency */ 839 struct task_struct *cp_task; /* separate cp/wb IO stats*/ 840 struct task_struct *wb_task; /* indicate inode is in context of writeback */ 841 nid_t i_xattr_nid; /* node id that contains xattrs */ 842 loff_t last_disk_size; /* lastly written file size */ 843 spinlock_t i_size_lock; /* protect last_disk_size */ 844 845 #ifdef CONFIG_QUOTA 846 struct dquot *i_dquot[MAXQUOTAS]; 847 848 /* quota space reservation, managed internally by quota code */ 849 qsize_t i_reserved_quota; 850 #endif 851 struct list_head dirty_list; /* dirty list for dirs and files */ 852 struct list_head gdirty_list; /* linked in global dirty list */ 853 struct task_struct *atomic_write_task; /* store atomic write task */ 854 struct extent_tree *extent_tree[NR_EXTENT_CACHES]; 855 /* cached extent_tree entry */ 856 struct inode *cow_inode; /* copy-on-write inode for atomic write */ 857 858 /* avoid racing between foreground op and gc */ 859 struct f2fs_rwsem i_gc_rwsem[2]; 860 struct f2fs_rwsem i_xattr_sem; /* avoid racing between reading and changing EAs */ 861 862 int i_extra_isize; /* size of extra space located in i_addr */ 863 kprojid_t i_projid; /* id for project quota */ 864 int i_inline_xattr_size; /* inline xattr size */ 865 struct timespec64 i_crtime; /* inode creation time */ 866 struct timespec64 i_disk_time[4];/* inode disk times */ 867 868 /* for file compress */ 869 atomic_t i_compr_blocks; /* # of compressed blocks */ 870 unsigned char i_compress_algorithm; /* algorithm type */ 871 unsigned char i_log_cluster_size; /* log of cluster size */ 872 unsigned char i_compress_level; /* compress level (lz4hc,zstd) */ 873 unsigned short i_compress_flag; /* compress flag */ 874 unsigned int i_cluster_size; /* cluster size */ 875 876 unsigned int atomic_write_cnt; 877 loff_t original_i_size; /* original i_size before atomic write */ 878 }; 879 880 static inline void get_read_extent_info(struct extent_info *ext, 881 struct f2fs_extent *i_ext) 882 { 883 ext->fofs = le32_to_cpu(i_ext->fofs); 884 ext->blk = le32_to_cpu(i_ext->blk); 885 ext->len = le32_to_cpu(i_ext->len); 886 } 887 888 static inline void set_raw_read_extent(struct extent_info *ext, 889 struct f2fs_extent *i_ext) 890 { 891 i_ext->fofs = cpu_to_le32(ext->fofs); 892 i_ext->blk = cpu_to_le32(ext->blk); 893 i_ext->len = cpu_to_le32(ext->len); 894 } 895 896 static inline bool __is_discard_mergeable(struct discard_info *back, 897 struct discard_info *front, unsigned int max_len) 898 { 899 return (back->lstart + back->len == front->lstart) && 900 (back->len + front->len <= max_len); 901 } 902 903 static inline bool __is_discard_back_mergeable(struct discard_info *cur, 904 struct discard_info *back, unsigned int max_len) 905 { 906 return __is_discard_mergeable(back, cur, max_len); 907 } 908 909 static inline bool __is_discard_front_mergeable(struct discard_info *cur, 910 struct discard_info *front, unsigned int max_len) 911 { 912 return __is_discard_mergeable(cur, front, max_len); 913 } 914 915 /* 916 * For free nid management 917 */ 918 enum nid_state { 919 FREE_NID, /* newly added to free nid list */ 920 PREALLOC_NID, /* it is preallocated */ 921 MAX_NID_STATE, 922 }; 923 924 enum nat_state { 925 TOTAL_NAT, 926 DIRTY_NAT, 927 RECLAIMABLE_NAT, 928 MAX_NAT_STATE, 929 }; 930 931 struct f2fs_nm_info { 932 block_t nat_blkaddr; /* base disk address of NAT */ 933 nid_t max_nid; /* maximum possible node ids */ 934 nid_t available_nids; /* # of available node ids */ 935 nid_t next_scan_nid; /* the next nid to be scanned */ 936 nid_t max_rf_node_blocks; /* max # of nodes for recovery */ 937 unsigned int ram_thresh; /* control the memory footprint */ 938 unsigned int ra_nid_pages; /* # of nid pages to be readaheaded */ 939 unsigned int dirty_nats_ratio; /* control dirty nats ratio threshold */ 940 941 /* NAT cache management */ 942 struct radix_tree_root nat_root;/* root of the nat entry cache */ 943 struct radix_tree_root nat_set_root;/* root of the nat set cache */ 944 struct f2fs_rwsem nat_tree_lock; /* protect nat entry tree */ 945 struct list_head nat_entries; /* cached nat entry list (clean) */ 946 spinlock_t nat_list_lock; /* protect clean nat entry list */ 947 unsigned int nat_cnt[MAX_NAT_STATE]; /* the # of cached nat entries */ 948 unsigned int nat_blocks; /* # of nat blocks */ 949 950 /* free node ids management */ 951 struct radix_tree_root free_nid_root;/* root of the free_nid cache */ 952 struct list_head free_nid_list; /* list for free nids excluding preallocated nids */ 953 unsigned int nid_cnt[MAX_NID_STATE]; /* the number of free node id */ 954 spinlock_t nid_list_lock; /* protect nid lists ops */ 955 struct mutex build_lock; /* lock for build free nids */ 956 unsigned char **free_nid_bitmap; 957 unsigned char *nat_block_bitmap; 958 unsigned short *free_nid_count; /* free nid count of NAT block */ 959 960 /* for checkpoint */ 961 char *nat_bitmap; /* NAT bitmap pointer */ 962 963 unsigned int nat_bits_blocks; /* # of nat bits blocks */ 964 unsigned char *nat_bits; /* NAT bits blocks */ 965 unsigned char *full_nat_bits; /* full NAT pages */ 966 unsigned char *empty_nat_bits; /* empty NAT pages */ 967 #ifdef CONFIG_F2FS_CHECK_FS 968 char *nat_bitmap_mir; /* NAT bitmap mirror */ 969 #endif 970 int bitmap_size; /* bitmap size */ 971 }; 972 973 /* 974 * this structure is used as one of function parameters. 975 * all the information are dedicated to a given direct node block determined 976 * by the data offset in a file. 977 */ 978 struct dnode_of_data { 979 struct inode *inode; /* vfs inode pointer */ 980 struct page *inode_page; /* its inode page, NULL is possible */ 981 struct page *node_page; /* cached direct node page */ 982 nid_t nid; /* node id of the direct node block */ 983 unsigned int ofs_in_node; /* data offset in the node page */ 984 bool inode_page_locked; /* inode page is locked or not */ 985 bool node_changed; /* is node block changed */ 986 char cur_level; /* level of hole node page */ 987 char max_level; /* level of current page located */ 988 block_t data_blkaddr; /* block address of the node block */ 989 }; 990 991 static inline void set_new_dnode(struct dnode_of_data *dn, struct inode *inode, 992 struct page *ipage, struct page *npage, nid_t nid) 993 { 994 memset(dn, 0, sizeof(*dn)); 995 dn->inode = inode; 996 dn->inode_page = ipage; 997 dn->node_page = npage; 998 dn->nid = nid; 999 } 1000 1001 /* 1002 * For SIT manager 1003 * 1004 * By default, there are 6 active log areas across the whole main area. 1005 * When considering hot and cold data separation to reduce cleaning overhead, 1006 * we split 3 for data logs and 3 for node logs as hot, warm, and cold types, 1007 * respectively. 1008 * In the current design, you should not change the numbers intentionally. 1009 * Instead, as a mount option such as active_logs=x, you can use 2, 4, and 6 1010 * logs individually according to the underlying devices. (default: 6) 1011 * Just in case, on-disk layout covers maximum 16 logs that consist of 8 for 1012 * data and 8 for node logs. 1013 */ 1014 #define NR_CURSEG_DATA_TYPE (3) 1015 #define NR_CURSEG_NODE_TYPE (3) 1016 #define NR_CURSEG_INMEM_TYPE (2) 1017 #define NR_CURSEG_RO_TYPE (2) 1018 #define NR_CURSEG_PERSIST_TYPE (NR_CURSEG_DATA_TYPE + NR_CURSEG_NODE_TYPE) 1019 #define NR_CURSEG_TYPE (NR_CURSEG_INMEM_TYPE + NR_CURSEG_PERSIST_TYPE) 1020 1021 enum { 1022 CURSEG_HOT_DATA = 0, /* directory entry blocks */ 1023 CURSEG_WARM_DATA, /* data blocks */ 1024 CURSEG_COLD_DATA, /* multimedia or GCed data blocks */ 1025 CURSEG_HOT_NODE, /* direct node blocks of directory files */ 1026 CURSEG_WARM_NODE, /* direct node blocks of normal files */ 1027 CURSEG_COLD_NODE, /* indirect node blocks */ 1028 NR_PERSISTENT_LOG, /* number of persistent log */ 1029 CURSEG_COLD_DATA_PINNED = NR_PERSISTENT_LOG, 1030 /* pinned file that needs consecutive block address */ 1031 CURSEG_ALL_DATA_ATGC, /* SSR alloctor in hot/warm/cold data area */ 1032 NO_CHECK_TYPE, /* number of persistent & inmem log */ 1033 }; 1034 1035 struct flush_cmd { 1036 struct completion wait; 1037 struct llist_node llnode; 1038 nid_t ino; 1039 int ret; 1040 }; 1041 1042 struct flush_cmd_control { 1043 struct task_struct *f2fs_issue_flush; /* flush thread */ 1044 wait_queue_head_t flush_wait_queue; /* waiting queue for wake-up */ 1045 atomic_t issued_flush; /* # of issued flushes */ 1046 atomic_t queued_flush; /* # of queued flushes */ 1047 struct llist_head issue_list; /* list for command issue */ 1048 struct llist_node *dispatch_list; /* list for command dispatch */ 1049 }; 1050 1051 struct f2fs_sm_info { 1052 struct sit_info *sit_info; /* whole segment information */ 1053 struct free_segmap_info *free_info; /* free segment information */ 1054 struct dirty_seglist_info *dirty_info; /* dirty segment information */ 1055 struct curseg_info *curseg_array; /* active segment information */ 1056 1057 struct f2fs_rwsem curseg_lock; /* for preventing curseg change */ 1058 1059 block_t seg0_blkaddr; /* block address of 0'th segment */ 1060 block_t main_blkaddr; /* start block address of main area */ 1061 block_t ssa_blkaddr; /* start block address of SSA area */ 1062 1063 unsigned int segment_count; /* total # of segments */ 1064 unsigned int main_segments; /* # of segments in main area */ 1065 unsigned int reserved_segments; /* # of reserved segments */ 1066 unsigned int additional_reserved_segments;/* reserved segs for IO align feature */ 1067 unsigned int ovp_segments; /* # of overprovision segments */ 1068 1069 /* a threshold to reclaim prefree segments */ 1070 unsigned int rec_prefree_segments; 1071 1072 struct list_head sit_entry_set; /* sit entry set list */ 1073 1074 unsigned int ipu_policy; /* in-place-update policy */ 1075 unsigned int min_ipu_util; /* in-place-update threshold */ 1076 unsigned int min_fsync_blocks; /* threshold for fsync */ 1077 unsigned int min_seq_blocks; /* threshold for sequential blocks */ 1078 unsigned int min_hot_blocks; /* threshold for hot block allocation */ 1079 unsigned int min_ssr_sections; /* threshold to trigger SSR allocation */ 1080 1081 /* for flush command control */ 1082 struct flush_cmd_control *fcc_info; 1083 1084 /* for discard command control */ 1085 struct discard_cmd_control *dcc_info; 1086 }; 1087 1088 /* 1089 * For superblock 1090 */ 1091 /* 1092 * COUNT_TYPE for monitoring 1093 * 1094 * f2fs monitors the number of several block types such as on-writeback, 1095 * dirty dentry blocks, dirty node blocks, and dirty meta blocks. 1096 */ 1097 #define WB_DATA_TYPE(p) (__is_cp_guaranteed(p) ? F2FS_WB_CP_DATA : F2FS_WB_DATA) 1098 enum count_type { 1099 F2FS_DIRTY_DENTS, 1100 F2FS_DIRTY_DATA, 1101 F2FS_DIRTY_QDATA, 1102 F2FS_DIRTY_NODES, 1103 F2FS_DIRTY_META, 1104 F2FS_DIRTY_IMETA, 1105 F2FS_WB_CP_DATA, 1106 F2FS_WB_DATA, 1107 F2FS_RD_DATA, 1108 F2FS_RD_NODE, 1109 F2FS_RD_META, 1110 F2FS_DIO_WRITE, 1111 F2FS_DIO_READ, 1112 NR_COUNT_TYPE, 1113 }; 1114 1115 /* 1116 * The below are the page types of bios used in submit_bio(). 1117 * The available types are: 1118 * DATA User data pages. It operates as async mode. 1119 * NODE Node pages. It operates as async mode. 1120 * META FS metadata pages such as SIT, NAT, CP. 1121 * NR_PAGE_TYPE The number of page types. 1122 * META_FLUSH Make sure the previous pages are written 1123 * with waiting the bio's completion 1124 * ... Only can be used with META. 1125 */ 1126 #define PAGE_TYPE_OF_BIO(type) ((type) > META ? META : (type)) 1127 enum page_type { 1128 DATA = 0, 1129 NODE = 1, /* should not change this */ 1130 META, 1131 NR_PAGE_TYPE, 1132 META_FLUSH, 1133 IPU, /* the below types are used by tracepoints only. */ 1134 OPU, 1135 }; 1136 1137 enum temp_type { 1138 HOT = 0, /* must be zero for meta bio */ 1139 WARM, 1140 COLD, 1141 NR_TEMP_TYPE, 1142 }; 1143 1144 enum need_lock_type { 1145 LOCK_REQ = 0, 1146 LOCK_DONE, 1147 LOCK_RETRY, 1148 }; 1149 1150 enum cp_reason_type { 1151 CP_NO_NEEDED, 1152 CP_NON_REGULAR, 1153 CP_COMPRESSED, 1154 CP_HARDLINK, 1155 CP_SB_NEED_CP, 1156 CP_WRONG_PINO, 1157 CP_NO_SPC_ROLL, 1158 CP_NODE_NEED_CP, 1159 CP_FASTBOOT_MODE, 1160 CP_SPEC_LOG_NUM, 1161 CP_RECOVER_DIR, 1162 }; 1163 1164 enum iostat_type { 1165 /* WRITE IO */ 1166 APP_DIRECT_IO, /* app direct write IOs */ 1167 APP_BUFFERED_IO, /* app buffered write IOs */ 1168 APP_WRITE_IO, /* app write IOs */ 1169 APP_MAPPED_IO, /* app mapped IOs */ 1170 APP_BUFFERED_CDATA_IO, /* app buffered write IOs on compressed file */ 1171 APP_MAPPED_CDATA_IO, /* app mapped write IOs on compressed file */ 1172 FS_DATA_IO, /* data IOs from kworker/fsync/reclaimer */ 1173 FS_CDATA_IO, /* data IOs from kworker/fsync/reclaimer on compressed file */ 1174 FS_NODE_IO, /* node IOs from kworker/fsync/reclaimer */ 1175 FS_META_IO, /* meta IOs from kworker/reclaimer */ 1176 FS_GC_DATA_IO, /* data IOs from forground gc */ 1177 FS_GC_NODE_IO, /* node IOs from forground gc */ 1178 FS_CP_DATA_IO, /* data IOs from checkpoint */ 1179 FS_CP_NODE_IO, /* node IOs from checkpoint */ 1180 FS_CP_META_IO, /* meta IOs from checkpoint */ 1181 1182 /* READ IO */ 1183 APP_DIRECT_READ_IO, /* app direct read IOs */ 1184 APP_BUFFERED_READ_IO, /* app buffered read IOs */ 1185 APP_READ_IO, /* app read IOs */ 1186 APP_MAPPED_READ_IO, /* app mapped read IOs */ 1187 APP_BUFFERED_CDATA_READ_IO, /* app buffered read IOs on compressed file */ 1188 APP_MAPPED_CDATA_READ_IO, /* app mapped read IOs on compressed file */ 1189 FS_DATA_READ_IO, /* data read IOs */ 1190 FS_GDATA_READ_IO, /* data read IOs from background gc */ 1191 FS_CDATA_READ_IO, /* compressed data read IOs */ 1192 FS_NODE_READ_IO, /* node read IOs */ 1193 FS_META_READ_IO, /* meta read IOs */ 1194 1195 /* other */ 1196 FS_DISCARD, /* discard */ 1197 NR_IO_TYPE, 1198 }; 1199 1200 struct f2fs_io_info { 1201 struct f2fs_sb_info *sbi; /* f2fs_sb_info pointer */ 1202 nid_t ino; /* inode number */ 1203 enum page_type type; /* contains DATA/NODE/META/META_FLUSH */ 1204 enum temp_type temp; /* contains HOT/WARM/COLD */ 1205 enum req_op op; /* contains REQ_OP_ */ 1206 blk_opf_t op_flags; /* req_flag_bits */ 1207 block_t new_blkaddr; /* new block address to be written */ 1208 block_t old_blkaddr; /* old block address before Cow */ 1209 struct page *page; /* page to be written */ 1210 struct page *encrypted_page; /* encrypted page */ 1211 struct page *compressed_page; /* compressed page */ 1212 struct list_head list; /* serialize IOs */ 1213 bool submitted; /* indicate IO submission */ 1214 int need_lock; /* indicate we need to lock cp_rwsem */ 1215 bool in_list; /* indicate fio is in io_list */ 1216 bool is_por; /* indicate IO is from recovery or not */ 1217 bool retry; /* need to reallocate block address */ 1218 int compr_blocks; /* # of compressed block addresses */ 1219 bool encrypted; /* indicate file is encrypted */ 1220 bool post_read; /* require post read */ 1221 enum iostat_type io_type; /* io type */ 1222 struct writeback_control *io_wbc; /* writeback control */ 1223 struct bio **bio; /* bio for ipu */ 1224 sector_t *last_block; /* last block number in bio */ 1225 unsigned char version; /* version of the node */ 1226 }; 1227 1228 struct bio_entry { 1229 struct bio *bio; 1230 struct list_head list; 1231 }; 1232 1233 #define is_read_io(rw) ((rw) == READ) 1234 struct f2fs_bio_info { 1235 struct f2fs_sb_info *sbi; /* f2fs superblock */ 1236 struct bio *bio; /* bios to merge */ 1237 sector_t last_block_in_bio; /* last block number */ 1238 struct f2fs_io_info fio; /* store buffered io info. */ 1239 struct f2fs_rwsem io_rwsem; /* blocking op for bio */ 1240 spinlock_t io_lock; /* serialize DATA/NODE IOs */ 1241 struct list_head io_list; /* track fios */ 1242 struct list_head bio_list; /* bio entry list head */ 1243 struct f2fs_rwsem bio_list_lock; /* lock to protect bio entry list */ 1244 }; 1245 1246 #define FDEV(i) (sbi->devs[i]) 1247 #define RDEV(i) (raw_super->devs[i]) 1248 struct f2fs_dev_info { 1249 struct block_device *bdev; 1250 char path[MAX_PATH_LEN]; 1251 unsigned int total_segments; 1252 block_t start_blk; 1253 block_t end_blk; 1254 #ifdef CONFIG_BLK_DEV_ZONED 1255 unsigned int nr_blkz; /* Total number of zones */ 1256 unsigned long *blkz_seq; /* Bitmap indicating sequential zones */ 1257 #endif 1258 }; 1259 1260 enum inode_type { 1261 DIR_INODE, /* for dirty dir inode */ 1262 FILE_INODE, /* for dirty regular/symlink inode */ 1263 DIRTY_META, /* for all dirtied inode metadata */ 1264 NR_INODE_TYPE, 1265 }; 1266 1267 /* for inner inode cache management */ 1268 struct inode_management { 1269 struct radix_tree_root ino_root; /* ino entry array */ 1270 spinlock_t ino_lock; /* for ino entry lock */ 1271 struct list_head ino_list; /* inode list head */ 1272 unsigned long ino_num; /* number of entries */ 1273 }; 1274 1275 /* for GC_AT */ 1276 struct atgc_management { 1277 bool atgc_enabled; /* ATGC is enabled or not */ 1278 struct rb_root_cached root; /* root of victim rb-tree */ 1279 struct list_head victim_list; /* linked with all victim entries */ 1280 unsigned int victim_count; /* victim count in rb-tree */ 1281 unsigned int candidate_ratio; /* candidate ratio */ 1282 unsigned int max_candidate_count; /* max candidate count */ 1283 unsigned int age_weight; /* age weight, vblock_weight = 100 - age_weight */ 1284 unsigned long long age_threshold; /* age threshold */ 1285 }; 1286 1287 struct f2fs_gc_control { 1288 unsigned int victim_segno; /* target victim segment number */ 1289 int init_gc_type; /* FG_GC or BG_GC */ 1290 bool no_bg_gc; /* check the space and stop bg_gc */ 1291 bool should_migrate_blocks; /* should migrate blocks */ 1292 bool err_gc_skipped; /* return EAGAIN if GC skipped */ 1293 unsigned int nr_free_secs; /* # of free sections to do GC */ 1294 }; 1295 1296 /* For s_flag in struct f2fs_sb_info */ 1297 enum { 1298 SBI_IS_DIRTY, /* dirty flag for checkpoint */ 1299 SBI_IS_CLOSE, /* specify unmounting */ 1300 SBI_NEED_FSCK, /* need fsck.f2fs to fix */ 1301 SBI_POR_DOING, /* recovery is doing or not */ 1302 SBI_NEED_SB_WRITE, /* need to recover superblock */ 1303 SBI_NEED_CP, /* need to checkpoint */ 1304 SBI_IS_SHUTDOWN, /* shutdown by ioctl */ 1305 SBI_IS_RECOVERED, /* recovered orphan/data */ 1306 SBI_CP_DISABLED, /* CP was disabled last mount */ 1307 SBI_CP_DISABLED_QUICK, /* CP was disabled quickly */ 1308 SBI_QUOTA_NEED_FLUSH, /* need to flush quota info in CP */ 1309 SBI_QUOTA_SKIP_FLUSH, /* skip flushing quota in current CP */ 1310 SBI_QUOTA_NEED_REPAIR, /* quota file may be corrupted */ 1311 SBI_IS_RESIZEFS, /* resizefs is in process */ 1312 SBI_IS_FREEZING, /* freezefs is in process */ 1313 }; 1314 1315 enum { 1316 CP_TIME, 1317 REQ_TIME, 1318 DISCARD_TIME, 1319 GC_TIME, 1320 DISABLE_TIME, 1321 UMOUNT_DISCARD_TIMEOUT, 1322 MAX_TIME, 1323 }; 1324 1325 /* Note that you need to keep synchronization with this gc_mode_names array */ 1326 enum { 1327 GC_NORMAL, 1328 GC_IDLE_CB, 1329 GC_IDLE_GREEDY, 1330 GC_IDLE_AT, 1331 GC_URGENT_HIGH, 1332 GC_URGENT_LOW, 1333 GC_URGENT_MID, 1334 MAX_GC_MODE, 1335 }; 1336 1337 enum { 1338 BGGC_MODE_ON, /* background gc is on */ 1339 BGGC_MODE_OFF, /* background gc is off */ 1340 BGGC_MODE_SYNC, /* 1341 * background gc is on, migrating blocks 1342 * like foreground gc 1343 */ 1344 }; 1345 1346 enum { 1347 FS_MODE_ADAPTIVE, /* use both lfs/ssr allocation */ 1348 FS_MODE_LFS, /* use lfs allocation only */ 1349 FS_MODE_FRAGMENT_SEG, /* segment fragmentation mode */ 1350 FS_MODE_FRAGMENT_BLK, /* block fragmentation mode */ 1351 }; 1352 1353 enum { 1354 ALLOC_MODE_DEFAULT, /* stay default */ 1355 ALLOC_MODE_REUSE, /* reuse segments as much as possible */ 1356 }; 1357 1358 enum fsync_mode { 1359 FSYNC_MODE_POSIX, /* fsync follows posix semantics */ 1360 FSYNC_MODE_STRICT, /* fsync behaves in line with ext4 */ 1361 FSYNC_MODE_NOBARRIER, /* fsync behaves nobarrier based on posix */ 1362 }; 1363 1364 enum { 1365 COMPR_MODE_FS, /* 1366 * automatically compress compression 1367 * enabled files 1368 */ 1369 COMPR_MODE_USER, /* 1370 * automatical compression is disabled. 1371 * user can control the file compression 1372 * using ioctls 1373 */ 1374 }; 1375 1376 enum { 1377 DISCARD_UNIT_BLOCK, /* basic discard unit is block */ 1378 DISCARD_UNIT_SEGMENT, /* basic discard unit is segment */ 1379 DISCARD_UNIT_SECTION, /* basic discard unit is section */ 1380 }; 1381 1382 enum { 1383 MEMORY_MODE_NORMAL, /* memory mode for normal devices */ 1384 MEMORY_MODE_LOW, /* memory mode for low memry devices */ 1385 }; 1386 1387 1388 1389 static inline int f2fs_test_bit(unsigned int nr, char *addr); 1390 static inline void f2fs_set_bit(unsigned int nr, char *addr); 1391 static inline void f2fs_clear_bit(unsigned int nr, char *addr); 1392 1393 /* 1394 * Layout of f2fs page.private: 1395 * 1396 * Layout A: lowest bit should be 1 1397 * | bit0 = 1 | bit1 | bit2 | ... | bit MAX | private data .... | 1398 * bit 0 PAGE_PRIVATE_NOT_POINTER 1399 * bit 1 PAGE_PRIVATE_ATOMIC_WRITE 1400 * bit 2 PAGE_PRIVATE_DUMMY_WRITE 1401 * bit 3 PAGE_PRIVATE_ONGOING_MIGRATION 1402 * bit 4 PAGE_PRIVATE_INLINE_INODE 1403 * bit 5 PAGE_PRIVATE_REF_RESOURCE 1404 * bit 6- f2fs private data 1405 * 1406 * Layout B: lowest bit should be 0 1407 * page.private is a wrapped pointer. 1408 */ 1409 enum { 1410 PAGE_PRIVATE_NOT_POINTER, /* private contains non-pointer data */ 1411 PAGE_PRIVATE_ATOMIC_WRITE, /* data page from atomic write path */ 1412 PAGE_PRIVATE_DUMMY_WRITE, /* data page for padding aligned IO */ 1413 PAGE_PRIVATE_ONGOING_MIGRATION, /* data page which is on-going migrating */ 1414 PAGE_PRIVATE_INLINE_INODE, /* inode page contains inline data */ 1415 PAGE_PRIVATE_REF_RESOURCE, /* dirty page has referenced resources */ 1416 PAGE_PRIVATE_MAX 1417 }; 1418 1419 #define PAGE_PRIVATE_GET_FUNC(name, flagname) \ 1420 static inline bool page_private_##name(struct page *page) \ 1421 { \ 1422 return PagePrivate(page) && \ 1423 test_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)) && \ 1424 test_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \ 1425 } 1426 1427 #define PAGE_PRIVATE_SET_FUNC(name, flagname) \ 1428 static inline void set_page_private_##name(struct page *page) \ 1429 { \ 1430 if (!PagePrivate(page)) { \ 1431 get_page(page); \ 1432 SetPagePrivate(page); \ 1433 set_page_private(page, 0); \ 1434 } \ 1435 set_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)); \ 1436 set_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \ 1437 } 1438 1439 #define PAGE_PRIVATE_CLEAR_FUNC(name, flagname) \ 1440 static inline void clear_page_private_##name(struct page *page) \ 1441 { \ 1442 clear_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \ 1443 if (page_private(page) == 1 << PAGE_PRIVATE_NOT_POINTER) { \ 1444 set_page_private(page, 0); \ 1445 if (PagePrivate(page)) { \ 1446 ClearPagePrivate(page); \ 1447 put_page(page); \ 1448 }\ 1449 } \ 1450 } 1451 1452 PAGE_PRIVATE_GET_FUNC(nonpointer, NOT_POINTER); 1453 PAGE_PRIVATE_GET_FUNC(reference, REF_RESOURCE); 1454 PAGE_PRIVATE_GET_FUNC(inline, INLINE_INODE); 1455 PAGE_PRIVATE_GET_FUNC(gcing, ONGOING_MIGRATION); 1456 PAGE_PRIVATE_GET_FUNC(atomic, ATOMIC_WRITE); 1457 PAGE_PRIVATE_GET_FUNC(dummy, DUMMY_WRITE); 1458 1459 PAGE_PRIVATE_SET_FUNC(reference, REF_RESOURCE); 1460 PAGE_PRIVATE_SET_FUNC(inline, INLINE_INODE); 1461 PAGE_PRIVATE_SET_FUNC(gcing, ONGOING_MIGRATION); 1462 PAGE_PRIVATE_SET_FUNC(atomic, ATOMIC_WRITE); 1463 PAGE_PRIVATE_SET_FUNC(dummy, DUMMY_WRITE); 1464 1465 PAGE_PRIVATE_CLEAR_FUNC(reference, REF_RESOURCE); 1466 PAGE_PRIVATE_CLEAR_FUNC(inline, INLINE_INODE); 1467 PAGE_PRIVATE_CLEAR_FUNC(gcing, ONGOING_MIGRATION); 1468 PAGE_PRIVATE_CLEAR_FUNC(atomic, ATOMIC_WRITE); 1469 PAGE_PRIVATE_CLEAR_FUNC(dummy, DUMMY_WRITE); 1470 1471 static inline unsigned long get_page_private_data(struct page *page) 1472 { 1473 unsigned long data = page_private(page); 1474 1475 if (!test_bit(PAGE_PRIVATE_NOT_POINTER, &data)) 1476 return 0; 1477 return data >> PAGE_PRIVATE_MAX; 1478 } 1479 1480 static inline void set_page_private_data(struct page *page, unsigned long data) 1481 { 1482 if (!PagePrivate(page)) { 1483 get_page(page); 1484 SetPagePrivate(page); 1485 set_page_private(page, 0); 1486 } 1487 set_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)); 1488 page_private(page) |= data << PAGE_PRIVATE_MAX; 1489 } 1490 1491 static inline void clear_page_private_data(struct page *page) 1492 { 1493 page_private(page) &= (1 << PAGE_PRIVATE_MAX) - 1; 1494 if (page_private(page) == 1 << PAGE_PRIVATE_NOT_POINTER) { 1495 set_page_private(page, 0); 1496 if (PagePrivate(page)) { 1497 ClearPagePrivate(page); 1498 put_page(page); 1499 } 1500 } 1501 } 1502 1503 /* For compression */ 1504 enum compress_algorithm_type { 1505 COMPRESS_LZO, 1506 COMPRESS_LZ4, 1507 COMPRESS_ZSTD, 1508 COMPRESS_LZORLE, 1509 COMPRESS_MAX, 1510 }; 1511 1512 enum compress_flag { 1513 COMPRESS_CHKSUM, 1514 COMPRESS_MAX_FLAG, 1515 }; 1516 1517 #define COMPRESS_WATERMARK 20 1518 #define COMPRESS_PERCENT 20 1519 1520 #define COMPRESS_DATA_RESERVED_SIZE 4 1521 struct compress_data { 1522 __le32 clen; /* compressed data size */ 1523 __le32 chksum; /* compressed data chksum */ 1524 __le32 reserved[COMPRESS_DATA_RESERVED_SIZE]; /* reserved */ 1525 u8 cdata[]; /* compressed data */ 1526 }; 1527 1528 #define COMPRESS_HEADER_SIZE (sizeof(struct compress_data)) 1529 1530 #define F2FS_COMPRESSED_PAGE_MAGIC 0xF5F2C000 1531 1532 #define COMPRESS_LEVEL_OFFSET 8 1533 1534 /* compress context */ 1535 struct compress_ctx { 1536 struct inode *inode; /* inode the context belong to */ 1537 pgoff_t cluster_idx; /* cluster index number */ 1538 unsigned int cluster_size; /* page count in cluster */ 1539 unsigned int log_cluster_size; /* log of cluster size */ 1540 struct page **rpages; /* pages store raw data in cluster */ 1541 unsigned int nr_rpages; /* total page number in rpages */ 1542 struct page **cpages; /* pages store compressed data in cluster */ 1543 unsigned int nr_cpages; /* total page number in cpages */ 1544 unsigned int valid_nr_cpages; /* valid page number in cpages */ 1545 void *rbuf; /* virtual mapped address on rpages */ 1546 struct compress_data *cbuf; /* virtual mapped address on cpages */ 1547 size_t rlen; /* valid data length in rbuf */ 1548 size_t clen; /* valid data length in cbuf */ 1549 void *private; /* payload buffer for specified compression algorithm */ 1550 void *private2; /* extra payload buffer */ 1551 }; 1552 1553 /* compress context for write IO path */ 1554 struct compress_io_ctx { 1555 u32 magic; /* magic number to indicate page is compressed */ 1556 struct inode *inode; /* inode the context belong to */ 1557 struct page **rpages; /* pages store raw data in cluster */ 1558 unsigned int nr_rpages; /* total page number in rpages */ 1559 atomic_t pending_pages; /* in-flight compressed page count */ 1560 }; 1561 1562 /* Context for decompressing one cluster on the read IO path */ 1563 struct decompress_io_ctx { 1564 u32 magic; /* magic number to indicate page is compressed */ 1565 struct inode *inode; /* inode the context belong to */ 1566 pgoff_t cluster_idx; /* cluster index number */ 1567 unsigned int cluster_size; /* page count in cluster */ 1568 unsigned int log_cluster_size; /* log of cluster size */ 1569 struct page **rpages; /* pages store raw data in cluster */ 1570 unsigned int nr_rpages; /* total page number in rpages */ 1571 struct page **cpages; /* pages store compressed data in cluster */ 1572 unsigned int nr_cpages; /* total page number in cpages */ 1573 struct page **tpages; /* temp pages to pad holes in cluster */ 1574 void *rbuf; /* virtual mapped address on rpages */ 1575 struct compress_data *cbuf; /* virtual mapped address on cpages */ 1576 size_t rlen; /* valid data length in rbuf */ 1577 size_t clen; /* valid data length in cbuf */ 1578 1579 /* 1580 * The number of compressed pages remaining to be read in this cluster. 1581 * This is initially nr_cpages. It is decremented by 1 each time a page 1582 * has been read (or failed to be read). When it reaches 0, the cluster 1583 * is decompressed (or an error is reported). 1584 * 1585 * If an error occurs before all the pages have been submitted for I/O, 1586 * then this will never reach 0. In this case the I/O submitter is 1587 * responsible for calling f2fs_decompress_end_io() instead. 1588 */ 1589 atomic_t remaining_pages; 1590 1591 /* 1592 * Number of references to this decompress_io_ctx. 1593 * 1594 * One reference is held for I/O completion. This reference is dropped 1595 * after the pagecache pages are updated and unlocked -- either after 1596 * decompression (and verity if enabled), or after an error. 1597 * 1598 * In addition, each compressed page holds a reference while it is in a 1599 * bio. These references are necessary prevent compressed pages from 1600 * being freed while they are still in a bio. 1601 */ 1602 refcount_t refcnt; 1603 1604 bool failed; /* IO error occurred before decompression? */ 1605 bool need_verity; /* need fs-verity verification after decompression? */ 1606 void *private; /* payload buffer for specified decompression algorithm */ 1607 void *private2; /* extra payload buffer */ 1608 struct work_struct verity_work; /* work to verify the decompressed pages */ 1609 struct work_struct free_work; /* work for late free this structure itself */ 1610 }; 1611 1612 #define NULL_CLUSTER ((unsigned int)(~0)) 1613 #define MIN_COMPRESS_LOG_SIZE 2 1614 #define MAX_COMPRESS_LOG_SIZE 8 1615 #define MAX_COMPRESS_WINDOW_SIZE(log_size) ((PAGE_SIZE) << (log_size)) 1616 1617 struct f2fs_sb_info { 1618 struct super_block *sb; /* pointer to VFS super block */ 1619 struct proc_dir_entry *s_proc; /* proc entry */ 1620 struct f2fs_super_block *raw_super; /* raw super block pointer */ 1621 struct f2fs_rwsem sb_lock; /* lock for raw super block */ 1622 int valid_super_block; /* valid super block no */ 1623 unsigned long s_flag; /* flags for sbi */ 1624 struct mutex writepages; /* mutex for writepages() */ 1625 1626 #ifdef CONFIG_BLK_DEV_ZONED 1627 unsigned int blocks_per_blkz; /* F2FS blocks per zone */ 1628 unsigned int log_blocks_per_blkz; /* log2 F2FS blocks per zone */ 1629 #endif 1630 1631 /* for node-related operations */ 1632 struct f2fs_nm_info *nm_info; /* node manager */ 1633 struct inode *node_inode; /* cache node blocks */ 1634 1635 /* for segment-related operations */ 1636 struct f2fs_sm_info *sm_info; /* segment manager */ 1637 1638 /* for bio operations */ 1639 struct f2fs_bio_info *write_io[NR_PAGE_TYPE]; /* for write bios */ 1640 /* keep migration IO order for LFS mode */ 1641 struct f2fs_rwsem io_order_lock; 1642 mempool_t *write_io_dummy; /* Dummy pages */ 1643 pgoff_t page_eio_ofs[NR_PAGE_TYPE]; /* EIO page offset */ 1644 int page_eio_cnt[NR_PAGE_TYPE]; /* EIO count */ 1645 1646 /* for checkpoint */ 1647 struct f2fs_checkpoint *ckpt; /* raw checkpoint pointer */ 1648 int cur_cp_pack; /* remain current cp pack */ 1649 spinlock_t cp_lock; /* for flag in ckpt */ 1650 struct inode *meta_inode; /* cache meta blocks */ 1651 struct f2fs_rwsem cp_global_sem; /* checkpoint procedure lock */ 1652 struct f2fs_rwsem cp_rwsem; /* blocking FS operations */ 1653 struct f2fs_rwsem node_write; /* locking node writes */ 1654 struct f2fs_rwsem node_change; /* locking node change */ 1655 wait_queue_head_t cp_wait; 1656 unsigned long last_time[MAX_TIME]; /* to store time in jiffies */ 1657 long interval_time[MAX_TIME]; /* to store thresholds */ 1658 struct ckpt_req_control cprc_info; /* for checkpoint request control */ 1659 1660 struct inode_management im[MAX_INO_ENTRY]; /* manage inode cache */ 1661 1662 spinlock_t fsync_node_lock; /* for node entry lock */ 1663 struct list_head fsync_node_list; /* node list head */ 1664 unsigned int fsync_seg_id; /* sequence id */ 1665 unsigned int fsync_node_num; /* number of node entries */ 1666 1667 /* for orphan inode, use 0'th array */ 1668 unsigned int max_orphans; /* max orphan inodes */ 1669 1670 /* for inode management */ 1671 struct list_head inode_list[NR_INODE_TYPE]; /* dirty inode list */ 1672 spinlock_t inode_lock[NR_INODE_TYPE]; /* for dirty inode list lock */ 1673 struct mutex flush_lock; /* for flush exclusion */ 1674 1675 /* for extent tree cache */ 1676 struct extent_tree_info extent_tree[NR_EXTENT_CACHES]; 1677 atomic64_t allocated_data_blocks; /* for block age extent_cache */ 1678 1679 /* The threshold used for hot and warm data seperation*/ 1680 unsigned int hot_data_age_threshold; 1681 unsigned int warm_data_age_threshold; 1682 1683 /* basic filesystem units */ 1684 unsigned int log_sectors_per_block; /* log2 sectors per block */ 1685 unsigned int log_blocksize; /* log2 block size */ 1686 unsigned int blocksize; /* block size */ 1687 unsigned int root_ino_num; /* root inode number*/ 1688 unsigned int node_ino_num; /* node inode number*/ 1689 unsigned int meta_ino_num; /* meta inode number*/ 1690 unsigned int log_blocks_per_seg; /* log2 blocks per segment */ 1691 unsigned int blocks_per_seg; /* blocks per segment */ 1692 unsigned int unusable_blocks_per_sec; /* unusable blocks per section */ 1693 unsigned int segs_per_sec; /* segments per section */ 1694 unsigned int secs_per_zone; /* sections per zone */ 1695 unsigned int total_sections; /* total section count */ 1696 unsigned int total_node_count; /* total node block count */ 1697 unsigned int total_valid_node_count; /* valid node block count */ 1698 int dir_level; /* directory level */ 1699 bool readdir_ra; /* readahead inode in readdir */ 1700 u64 max_io_bytes; /* max io bytes to merge IOs */ 1701 1702 block_t user_block_count; /* # of user blocks */ 1703 block_t total_valid_block_count; /* # of valid blocks */ 1704 block_t discard_blks; /* discard command candidats */ 1705 block_t last_valid_block_count; /* for recovery */ 1706 block_t reserved_blocks; /* configurable reserved blocks */ 1707 block_t current_reserved_blocks; /* current reserved blocks */ 1708 1709 /* Additional tracking for no checkpoint mode */ 1710 block_t unusable_block_count; /* # of blocks saved by last cp */ 1711 1712 unsigned int nquota_files; /* # of quota sysfile */ 1713 struct f2fs_rwsem quota_sem; /* blocking cp for flags */ 1714 1715 /* # of pages, see count_type */ 1716 atomic_t nr_pages[NR_COUNT_TYPE]; 1717 /* # of allocated blocks */ 1718 struct percpu_counter alloc_valid_block_count; 1719 /* # of node block writes as roll forward recovery */ 1720 struct percpu_counter rf_node_block_count; 1721 1722 /* writeback control */ 1723 atomic_t wb_sync_req[META]; /* count # of WB_SYNC threads */ 1724 1725 /* valid inode count */ 1726 struct percpu_counter total_valid_inode_count; 1727 1728 struct f2fs_mount_info mount_opt; /* mount options */ 1729 1730 /* for cleaning operations */ 1731 struct f2fs_rwsem gc_lock; /* 1732 * semaphore for GC, avoid 1733 * race between GC and GC or CP 1734 */ 1735 struct f2fs_gc_kthread *gc_thread; /* GC thread */ 1736 struct atgc_management am; /* atgc management */ 1737 unsigned int cur_victim_sec; /* current victim section num */ 1738 unsigned int gc_mode; /* current GC state */ 1739 unsigned int next_victim_seg[2]; /* next segment in victim section */ 1740 spinlock_t gc_remaining_trials_lock; 1741 /* remaining trial count for GC_URGENT_* and GC_IDLE_* */ 1742 unsigned int gc_remaining_trials; 1743 1744 /* for skip statistic */ 1745 unsigned long long skipped_gc_rwsem; /* FG_GC only */ 1746 1747 /* threshold for gc trials on pinned files */ 1748 u64 gc_pin_file_threshold; 1749 struct f2fs_rwsem pin_sem; 1750 1751 /* maximum # of trials to find a victim segment for SSR and GC */ 1752 unsigned int max_victim_search; 1753 /* migration granularity of garbage collection, unit: segment */ 1754 unsigned int migration_granularity; 1755 1756 /* 1757 * for stat information. 1758 * one is for the LFS mode, and the other is for the SSR mode. 1759 */ 1760 #ifdef CONFIG_F2FS_STAT_FS 1761 struct f2fs_stat_info *stat_info; /* FS status information */ 1762 atomic_t meta_count[META_MAX]; /* # of meta blocks */ 1763 unsigned int segment_count[2]; /* # of allocated segments */ 1764 unsigned int block_count[2]; /* # of allocated blocks */ 1765 atomic_t inplace_count; /* # of inplace update */ 1766 /* # of lookup extent cache */ 1767 atomic64_t total_hit_ext[NR_EXTENT_CACHES]; 1768 /* # of hit rbtree extent node */ 1769 atomic64_t read_hit_rbtree[NR_EXTENT_CACHES]; 1770 /* # of hit cached extent node */ 1771 atomic64_t read_hit_cached[NR_EXTENT_CACHES]; 1772 /* # of hit largest extent node in read extent cache */ 1773 atomic64_t read_hit_largest; 1774 atomic_t inline_xattr; /* # of inline_xattr inodes */ 1775 atomic_t inline_inode; /* # of inline_data inodes */ 1776 atomic_t inline_dir; /* # of inline_dentry inodes */ 1777 atomic_t compr_inode; /* # of compressed inodes */ 1778 atomic64_t compr_blocks; /* # of compressed blocks */ 1779 atomic_t swapfile_inode; /* # of swapfile inodes */ 1780 atomic_t atomic_files; /* # of opened atomic file */ 1781 atomic_t max_aw_cnt; /* max # of atomic writes */ 1782 unsigned int io_skip_bggc; /* skip background gc for in-flight IO */ 1783 unsigned int other_skip_bggc; /* skip background gc for other reasons */ 1784 unsigned int ndirty_inode[NR_INODE_TYPE]; /* # of dirty inodes */ 1785 #endif 1786 spinlock_t stat_lock; /* lock for stat operations */ 1787 1788 /* to attach REQ_META|REQ_FUA flags */ 1789 unsigned int data_io_flag; 1790 unsigned int node_io_flag; 1791 1792 /* For sysfs support */ 1793 struct kobject s_kobj; /* /sys/fs/f2fs/<devname> */ 1794 struct completion s_kobj_unregister; 1795 1796 struct kobject s_stat_kobj; /* /sys/fs/f2fs/<devname>/stat */ 1797 struct completion s_stat_kobj_unregister; 1798 1799 struct kobject s_feature_list_kobj; /* /sys/fs/f2fs/<devname>/feature_list */ 1800 struct completion s_feature_list_kobj_unregister; 1801 1802 /* For shrinker support */ 1803 struct list_head s_list; 1804 struct mutex umount_mutex; 1805 unsigned int shrinker_run_no; 1806 1807 /* For multi devices */ 1808 int s_ndevs; /* number of devices */ 1809 struct f2fs_dev_info *devs; /* for device list */ 1810 unsigned int dirty_device; /* for checkpoint data flush */ 1811 spinlock_t dev_lock; /* protect dirty_device */ 1812 bool aligned_blksize; /* all devices has the same logical blksize */ 1813 1814 /* For write statistics */ 1815 u64 sectors_written_start; 1816 u64 kbytes_written; 1817 1818 /* Reference to checksum algorithm driver via cryptoapi */ 1819 struct crypto_shash *s_chksum_driver; 1820 1821 /* Precomputed FS UUID checksum for seeding other checksums */ 1822 __u32 s_chksum_seed; 1823 1824 struct workqueue_struct *post_read_wq; /* post read workqueue */ 1825 1826 unsigned char errors[MAX_F2FS_ERRORS]; /* error flags */ 1827 spinlock_t error_lock; /* protect errors array */ 1828 bool error_dirty; /* errors of sb is dirty */ 1829 1830 struct kmem_cache *inline_xattr_slab; /* inline xattr entry */ 1831 unsigned int inline_xattr_slab_size; /* default inline xattr slab size */ 1832 1833 /* For reclaimed segs statistics per each GC mode */ 1834 unsigned int gc_segment_mode; /* GC state for reclaimed segments */ 1835 unsigned int gc_reclaimed_segs[MAX_GC_MODE]; /* Reclaimed segs for each mode */ 1836 1837 unsigned long seq_file_ra_mul; /* multiplier for ra_pages of seq. files in fadvise */ 1838 1839 int max_fragment_chunk; /* max chunk size for block fragmentation mode */ 1840 int max_fragment_hole; /* max hole size for block fragmentation mode */ 1841 1842 /* For atomic write statistics */ 1843 atomic64_t current_atomic_write; 1844 s64 peak_atomic_write; 1845 u64 committed_atomic_block; 1846 u64 revoked_atomic_block; 1847 1848 #ifdef CONFIG_F2FS_FS_COMPRESSION 1849 struct kmem_cache *page_array_slab; /* page array entry */ 1850 unsigned int page_array_slab_size; /* default page array slab size */ 1851 1852 /* For runtime compression statistics */ 1853 u64 compr_written_block; 1854 u64 compr_saved_block; 1855 u32 compr_new_inode; 1856 1857 /* For compressed block cache */ 1858 struct inode *compress_inode; /* cache compressed blocks */ 1859 unsigned int compress_percent; /* cache page percentage */ 1860 unsigned int compress_watermark; /* cache page watermark */ 1861 atomic_t compress_page_hit; /* cache hit count */ 1862 #endif 1863 1864 #ifdef CONFIG_F2FS_IOSTAT 1865 /* For app/fs IO statistics */ 1866 spinlock_t iostat_lock; 1867 unsigned long long rw_iostat[NR_IO_TYPE]; 1868 unsigned long long prev_rw_iostat[NR_IO_TYPE]; 1869 bool iostat_enable; 1870 unsigned long iostat_next_period; 1871 unsigned int iostat_period_ms; 1872 1873 /* For io latency related statistics info in one iostat period */ 1874 spinlock_t iostat_lat_lock; 1875 struct iostat_lat_info *iostat_io_lat; 1876 #endif 1877 }; 1878 1879 #ifdef CONFIG_F2FS_FAULT_INJECTION 1880 #define f2fs_show_injection_info(sbi, type) \ 1881 printk_ratelimited("%sF2FS-fs (%s) : inject %s in %s of %pS\n", \ 1882 KERN_INFO, sbi->sb->s_id, \ 1883 f2fs_fault_name[type], \ 1884 __func__, __builtin_return_address(0)) 1885 static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type) 1886 { 1887 struct f2fs_fault_info *ffi = &F2FS_OPTION(sbi).fault_info; 1888 1889 if (!ffi->inject_rate) 1890 return false; 1891 1892 if (!IS_FAULT_SET(ffi, type)) 1893 return false; 1894 1895 atomic_inc(&ffi->inject_ops); 1896 if (atomic_read(&ffi->inject_ops) >= ffi->inject_rate) { 1897 atomic_set(&ffi->inject_ops, 0); 1898 return true; 1899 } 1900 return false; 1901 } 1902 #else 1903 #define f2fs_show_injection_info(sbi, type) do { } while (0) 1904 static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type) 1905 { 1906 return false; 1907 } 1908 #endif 1909 1910 /* 1911 * Test if the mounted volume is a multi-device volume. 1912 * - For a single regular disk volume, sbi->s_ndevs is 0. 1913 * - For a single zoned disk volume, sbi->s_ndevs is 1. 1914 * - For a multi-device volume, sbi->s_ndevs is always 2 or more. 1915 */ 1916 static inline bool f2fs_is_multi_device(struct f2fs_sb_info *sbi) 1917 { 1918 return sbi->s_ndevs > 1; 1919 } 1920 1921 static inline void f2fs_update_time(struct f2fs_sb_info *sbi, int type) 1922 { 1923 unsigned long now = jiffies; 1924 1925 sbi->last_time[type] = now; 1926 1927 /* DISCARD_TIME and GC_TIME are based on REQ_TIME */ 1928 if (type == REQ_TIME) { 1929 sbi->last_time[DISCARD_TIME] = now; 1930 sbi->last_time[GC_TIME] = now; 1931 } 1932 } 1933 1934 static inline bool f2fs_time_over(struct f2fs_sb_info *sbi, int type) 1935 { 1936 unsigned long interval = sbi->interval_time[type] * HZ; 1937 1938 return time_after(jiffies, sbi->last_time[type] + interval); 1939 } 1940 1941 static inline unsigned int f2fs_time_to_wait(struct f2fs_sb_info *sbi, 1942 int type) 1943 { 1944 unsigned long interval = sbi->interval_time[type] * HZ; 1945 unsigned int wait_ms = 0; 1946 long delta; 1947 1948 delta = (sbi->last_time[type] + interval) - jiffies; 1949 if (delta > 0) 1950 wait_ms = jiffies_to_msecs(delta); 1951 1952 return wait_ms; 1953 } 1954 1955 /* 1956 * Inline functions 1957 */ 1958 static inline u32 __f2fs_crc32(struct f2fs_sb_info *sbi, u32 crc, 1959 const void *address, unsigned int length) 1960 { 1961 struct { 1962 struct shash_desc shash; 1963 char ctx[4]; 1964 } desc; 1965 int err; 1966 1967 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver) != sizeof(desc.ctx)); 1968 1969 desc.shash.tfm = sbi->s_chksum_driver; 1970 *(u32 *)desc.ctx = crc; 1971 1972 err = crypto_shash_update(&desc.shash, address, length); 1973 BUG_ON(err); 1974 1975 return *(u32 *)desc.ctx; 1976 } 1977 1978 static inline u32 f2fs_crc32(struct f2fs_sb_info *sbi, const void *address, 1979 unsigned int length) 1980 { 1981 return __f2fs_crc32(sbi, F2FS_SUPER_MAGIC, address, length); 1982 } 1983 1984 static inline bool f2fs_crc_valid(struct f2fs_sb_info *sbi, __u32 blk_crc, 1985 void *buf, size_t buf_size) 1986 { 1987 return f2fs_crc32(sbi, buf, buf_size) == blk_crc; 1988 } 1989 1990 static inline u32 f2fs_chksum(struct f2fs_sb_info *sbi, u32 crc, 1991 const void *address, unsigned int length) 1992 { 1993 return __f2fs_crc32(sbi, crc, address, length); 1994 } 1995 1996 static inline struct f2fs_inode_info *F2FS_I(struct inode *inode) 1997 { 1998 return container_of(inode, struct f2fs_inode_info, vfs_inode); 1999 } 2000 2001 static inline struct f2fs_sb_info *F2FS_SB(struct super_block *sb) 2002 { 2003 return sb->s_fs_info; 2004 } 2005 2006 static inline struct f2fs_sb_info *F2FS_I_SB(struct inode *inode) 2007 { 2008 return F2FS_SB(inode->i_sb); 2009 } 2010 2011 static inline struct f2fs_sb_info *F2FS_M_SB(struct address_space *mapping) 2012 { 2013 return F2FS_I_SB(mapping->host); 2014 } 2015 2016 static inline struct f2fs_sb_info *F2FS_P_SB(struct page *page) 2017 { 2018 return F2FS_M_SB(page_file_mapping(page)); 2019 } 2020 2021 static inline struct f2fs_super_block *F2FS_RAW_SUPER(struct f2fs_sb_info *sbi) 2022 { 2023 return (struct f2fs_super_block *)(sbi->raw_super); 2024 } 2025 2026 static inline struct f2fs_checkpoint *F2FS_CKPT(struct f2fs_sb_info *sbi) 2027 { 2028 return (struct f2fs_checkpoint *)(sbi->ckpt); 2029 } 2030 2031 static inline struct f2fs_node *F2FS_NODE(struct page *page) 2032 { 2033 return (struct f2fs_node *)page_address(page); 2034 } 2035 2036 static inline struct f2fs_inode *F2FS_INODE(struct page *page) 2037 { 2038 return &((struct f2fs_node *)page_address(page))->i; 2039 } 2040 2041 static inline struct f2fs_nm_info *NM_I(struct f2fs_sb_info *sbi) 2042 { 2043 return (struct f2fs_nm_info *)(sbi->nm_info); 2044 } 2045 2046 static inline struct f2fs_sm_info *SM_I(struct f2fs_sb_info *sbi) 2047 { 2048 return (struct f2fs_sm_info *)(sbi->sm_info); 2049 } 2050 2051 static inline struct sit_info *SIT_I(struct f2fs_sb_info *sbi) 2052 { 2053 return (struct sit_info *)(SM_I(sbi)->sit_info); 2054 } 2055 2056 static inline struct free_segmap_info *FREE_I(struct f2fs_sb_info *sbi) 2057 { 2058 return (struct free_segmap_info *)(SM_I(sbi)->free_info); 2059 } 2060 2061 static inline struct dirty_seglist_info *DIRTY_I(struct f2fs_sb_info *sbi) 2062 { 2063 return (struct dirty_seglist_info *)(SM_I(sbi)->dirty_info); 2064 } 2065 2066 static inline struct address_space *META_MAPPING(struct f2fs_sb_info *sbi) 2067 { 2068 return sbi->meta_inode->i_mapping; 2069 } 2070 2071 static inline struct address_space *NODE_MAPPING(struct f2fs_sb_info *sbi) 2072 { 2073 return sbi->node_inode->i_mapping; 2074 } 2075 2076 static inline bool is_sbi_flag_set(struct f2fs_sb_info *sbi, unsigned int type) 2077 { 2078 return test_bit(type, &sbi->s_flag); 2079 } 2080 2081 static inline void set_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type) 2082 { 2083 set_bit(type, &sbi->s_flag); 2084 } 2085 2086 static inline void clear_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type) 2087 { 2088 clear_bit(type, &sbi->s_flag); 2089 } 2090 2091 static inline unsigned long long cur_cp_version(struct f2fs_checkpoint *cp) 2092 { 2093 return le64_to_cpu(cp->checkpoint_ver); 2094 } 2095 2096 static inline unsigned long f2fs_qf_ino(struct super_block *sb, int type) 2097 { 2098 if (type < F2FS_MAX_QUOTAS) 2099 return le32_to_cpu(F2FS_SB(sb)->raw_super->qf_ino[type]); 2100 return 0; 2101 } 2102 2103 static inline __u64 cur_cp_crc(struct f2fs_checkpoint *cp) 2104 { 2105 size_t crc_offset = le32_to_cpu(cp->checksum_offset); 2106 return le32_to_cpu(*((__le32 *)((unsigned char *)cp + crc_offset))); 2107 } 2108 2109 static inline bool __is_set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 2110 { 2111 unsigned int ckpt_flags = le32_to_cpu(cp->ckpt_flags); 2112 2113 return ckpt_flags & f; 2114 } 2115 2116 static inline bool is_set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 2117 { 2118 return __is_set_ckpt_flags(F2FS_CKPT(sbi), f); 2119 } 2120 2121 static inline void __set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 2122 { 2123 unsigned int ckpt_flags; 2124 2125 ckpt_flags = le32_to_cpu(cp->ckpt_flags); 2126 ckpt_flags |= f; 2127 cp->ckpt_flags = cpu_to_le32(ckpt_flags); 2128 } 2129 2130 static inline void set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 2131 { 2132 unsigned long flags; 2133 2134 spin_lock_irqsave(&sbi->cp_lock, flags); 2135 __set_ckpt_flags(F2FS_CKPT(sbi), f); 2136 spin_unlock_irqrestore(&sbi->cp_lock, flags); 2137 } 2138 2139 static inline void __clear_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 2140 { 2141 unsigned int ckpt_flags; 2142 2143 ckpt_flags = le32_to_cpu(cp->ckpt_flags); 2144 ckpt_flags &= (~f); 2145 cp->ckpt_flags = cpu_to_le32(ckpt_flags); 2146 } 2147 2148 static inline void clear_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 2149 { 2150 unsigned long flags; 2151 2152 spin_lock_irqsave(&sbi->cp_lock, flags); 2153 __clear_ckpt_flags(F2FS_CKPT(sbi), f); 2154 spin_unlock_irqrestore(&sbi->cp_lock, flags); 2155 } 2156 2157 #define init_f2fs_rwsem(sem) \ 2158 do { \ 2159 static struct lock_class_key __key; \ 2160 \ 2161 __init_f2fs_rwsem((sem), #sem, &__key); \ 2162 } while (0) 2163 2164 static inline void __init_f2fs_rwsem(struct f2fs_rwsem *sem, 2165 const char *sem_name, struct lock_class_key *key) 2166 { 2167 __init_rwsem(&sem->internal_rwsem, sem_name, key); 2168 #ifdef CONFIG_F2FS_UNFAIR_RWSEM 2169 init_waitqueue_head(&sem->read_waiters); 2170 #endif 2171 } 2172 2173 static inline int f2fs_rwsem_is_locked(struct f2fs_rwsem *sem) 2174 { 2175 return rwsem_is_locked(&sem->internal_rwsem); 2176 } 2177 2178 static inline int f2fs_rwsem_is_contended(struct f2fs_rwsem *sem) 2179 { 2180 return rwsem_is_contended(&sem->internal_rwsem); 2181 } 2182 2183 static inline void f2fs_down_read(struct f2fs_rwsem *sem) 2184 { 2185 #ifdef CONFIG_F2FS_UNFAIR_RWSEM 2186 wait_event(sem->read_waiters, down_read_trylock(&sem->internal_rwsem)); 2187 #else 2188 down_read(&sem->internal_rwsem); 2189 #endif 2190 } 2191 2192 static inline int f2fs_down_read_trylock(struct f2fs_rwsem *sem) 2193 { 2194 return down_read_trylock(&sem->internal_rwsem); 2195 } 2196 2197 #ifdef CONFIG_DEBUG_LOCK_ALLOC 2198 static inline void f2fs_down_read_nested(struct f2fs_rwsem *sem, int subclass) 2199 { 2200 down_read_nested(&sem->internal_rwsem, subclass); 2201 } 2202 #else 2203 #define f2fs_down_read_nested(sem, subclass) f2fs_down_read(sem) 2204 #endif 2205 2206 static inline void f2fs_up_read(struct f2fs_rwsem *sem) 2207 { 2208 up_read(&sem->internal_rwsem); 2209 } 2210 2211 static inline void f2fs_down_write(struct f2fs_rwsem *sem) 2212 { 2213 down_write(&sem->internal_rwsem); 2214 } 2215 2216 static inline int f2fs_down_write_trylock(struct f2fs_rwsem *sem) 2217 { 2218 return down_write_trylock(&sem->internal_rwsem); 2219 } 2220 2221 static inline void f2fs_up_write(struct f2fs_rwsem *sem) 2222 { 2223 up_write(&sem->internal_rwsem); 2224 #ifdef CONFIG_F2FS_UNFAIR_RWSEM 2225 wake_up_all(&sem->read_waiters); 2226 #endif 2227 } 2228 2229 static inline void f2fs_lock_op(struct f2fs_sb_info *sbi) 2230 { 2231 f2fs_down_read(&sbi->cp_rwsem); 2232 } 2233 2234 static inline int f2fs_trylock_op(struct f2fs_sb_info *sbi) 2235 { 2236 if (time_to_inject(sbi, FAULT_LOCK_OP)) { 2237 f2fs_show_injection_info(sbi, FAULT_LOCK_OP); 2238 return 0; 2239 } 2240 return f2fs_down_read_trylock(&sbi->cp_rwsem); 2241 } 2242 2243 static inline void f2fs_unlock_op(struct f2fs_sb_info *sbi) 2244 { 2245 f2fs_up_read(&sbi->cp_rwsem); 2246 } 2247 2248 static inline void f2fs_lock_all(struct f2fs_sb_info *sbi) 2249 { 2250 f2fs_down_write(&sbi->cp_rwsem); 2251 } 2252 2253 static inline void f2fs_unlock_all(struct f2fs_sb_info *sbi) 2254 { 2255 f2fs_up_write(&sbi->cp_rwsem); 2256 } 2257 2258 static inline int __get_cp_reason(struct f2fs_sb_info *sbi) 2259 { 2260 int reason = CP_SYNC; 2261 2262 if (test_opt(sbi, FASTBOOT)) 2263 reason = CP_FASTBOOT; 2264 if (is_sbi_flag_set(sbi, SBI_IS_CLOSE)) 2265 reason = CP_UMOUNT; 2266 return reason; 2267 } 2268 2269 static inline bool __remain_node_summaries(int reason) 2270 { 2271 return (reason & (CP_UMOUNT | CP_FASTBOOT)); 2272 } 2273 2274 static inline bool __exist_node_summaries(struct f2fs_sb_info *sbi) 2275 { 2276 return (is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG) || 2277 is_set_ckpt_flags(sbi, CP_FASTBOOT_FLAG)); 2278 } 2279 2280 /* 2281 * Check whether the inode has blocks or not 2282 */ 2283 static inline int F2FS_HAS_BLOCKS(struct inode *inode) 2284 { 2285 block_t xattr_block = F2FS_I(inode)->i_xattr_nid ? 1 : 0; 2286 2287 return (inode->i_blocks >> F2FS_LOG_SECTORS_PER_BLOCK) > xattr_block; 2288 } 2289 2290 static inline bool f2fs_has_xattr_block(unsigned int ofs) 2291 { 2292 return ofs == XATTR_NODE_OFFSET; 2293 } 2294 2295 static inline bool __allow_reserved_blocks(struct f2fs_sb_info *sbi, 2296 struct inode *inode, bool cap) 2297 { 2298 if (!inode) 2299 return true; 2300 if (!test_opt(sbi, RESERVE_ROOT)) 2301 return false; 2302 if (IS_NOQUOTA(inode)) 2303 return true; 2304 if (uid_eq(F2FS_OPTION(sbi).s_resuid, current_fsuid())) 2305 return true; 2306 if (!gid_eq(F2FS_OPTION(sbi).s_resgid, GLOBAL_ROOT_GID) && 2307 in_group_p(F2FS_OPTION(sbi).s_resgid)) 2308 return true; 2309 if (cap && capable(CAP_SYS_RESOURCE)) 2310 return true; 2311 return false; 2312 } 2313 2314 static inline void f2fs_i_blocks_write(struct inode *, block_t, bool, bool); 2315 static inline int inc_valid_block_count(struct f2fs_sb_info *sbi, 2316 struct inode *inode, blkcnt_t *count) 2317 { 2318 blkcnt_t diff = 0, release = 0; 2319 block_t avail_user_block_count; 2320 int ret; 2321 2322 ret = dquot_reserve_block(inode, *count); 2323 if (ret) 2324 return ret; 2325 2326 if (time_to_inject(sbi, FAULT_BLOCK)) { 2327 f2fs_show_injection_info(sbi, FAULT_BLOCK); 2328 release = *count; 2329 goto release_quota; 2330 } 2331 2332 /* 2333 * let's increase this in prior to actual block count change in order 2334 * for f2fs_sync_file to avoid data races when deciding checkpoint. 2335 */ 2336 percpu_counter_add(&sbi->alloc_valid_block_count, (*count)); 2337 2338 spin_lock(&sbi->stat_lock); 2339 sbi->total_valid_block_count += (block_t)(*count); 2340 avail_user_block_count = sbi->user_block_count - 2341 sbi->current_reserved_blocks; 2342 2343 if (!__allow_reserved_blocks(sbi, inode, true)) 2344 avail_user_block_count -= F2FS_OPTION(sbi).root_reserved_blocks; 2345 2346 if (F2FS_IO_ALIGNED(sbi)) 2347 avail_user_block_count -= sbi->blocks_per_seg * 2348 SM_I(sbi)->additional_reserved_segments; 2349 2350 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) { 2351 if (avail_user_block_count > sbi->unusable_block_count) 2352 avail_user_block_count -= sbi->unusable_block_count; 2353 else 2354 avail_user_block_count = 0; 2355 } 2356 if (unlikely(sbi->total_valid_block_count > avail_user_block_count)) { 2357 diff = sbi->total_valid_block_count - avail_user_block_count; 2358 if (diff > *count) 2359 diff = *count; 2360 *count -= diff; 2361 release = diff; 2362 sbi->total_valid_block_count -= diff; 2363 if (!*count) { 2364 spin_unlock(&sbi->stat_lock); 2365 goto enospc; 2366 } 2367 } 2368 spin_unlock(&sbi->stat_lock); 2369 2370 if (unlikely(release)) { 2371 percpu_counter_sub(&sbi->alloc_valid_block_count, release); 2372 dquot_release_reservation_block(inode, release); 2373 } 2374 f2fs_i_blocks_write(inode, *count, true, true); 2375 return 0; 2376 2377 enospc: 2378 percpu_counter_sub(&sbi->alloc_valid_block_count, release); 2379 release_quota: 2380 dquot_release_reservation_block(inode, release); 2381 return -ENOSPC; 2382 } 2383 2384 __printf(2, 3) 2385 void f2fs_printk(struct f2fs_sb_info *sbi, const char *fmt, ...); 2386 2387 #define f2fs_err(sbi, fmt, ...) \ 2388 f2fs_printk(sbi, KERN_ERR fmt, ##__VA_ARGS__) 2389 #define f2fs_warn(sbi, fmt, ...) \ 2390 f2fs_printk(sbi, KERN_WARNING fmt, ##__VA_ARGS__) 2391 #define f2fs_notice(sbi, fmt, ...) \ 2392 f2fs_printk(sbi, KERN_NOTICE fmt, ##__VA_ARGS__) 2393 #define f2fs_info(sbi, fmt, ...) \ 2394 f2fs_printk(sbi, KERN_INFO fmt, ##__VA_ARGS__) 2395 #define f2fs_debug(sbi, fmt, ...) \ 2396 f2fs_printk(sbi, KERN_DEBUG fmt, ##__VA_ARGS__) 2397 2398 static inline void dec_valid_block_count(struct f2fs_sb_info *sbi, 2399 struct inode *inode, 2400 block_t count) 2401 { 2402 blkcnt_t sectors = count << F2FS_LOG_SECTORS_PER_BLOCK; 2403 2404 spin_lock(&sbi->stat_lock); 2405 f2fs_bug_on(sbi, sbi->total_valid_block_count < (block_t) count); 2406 sbi->total_valid_block_count -= (block_t)count; 2407 if (sbi->reserved_blocks && 2408 sbi->current_reserved_blocks < sbi->reserved_blocks) 2409 sbi->current_reserved_blocks = min(sbi->reserved_blocks, 2410 sbi->current_reserved_blocks + count); 2411 spin_unlock(&sbi->stat_lock); 2412 if (unlikely(inode->i_blocks < sectors)) { 2413 f2fs_warn(sbi, "Inconsistent i_blocks, ino:%lu, iblocks:%llu, sectors:%llu", 2414 inode->i_ino, 2415 (unsigned long long)inode->i_blocks, 2416 (unsigned long long)sectors); 2417 set_sbi_flag(sbi, SBI_NEED_FSCK); 2418 return; 2419 } 2420 f2fs_i_blocks_write(inode, count, false, true); 2421 } 2422 2423 static inline void inc_page_count(struct f2fs_sb_info *sbi, int count_type) 2424 { 2425 atomic_inc(&sbi->nr_pages[count_type]); 2426 2427 if (count_type == F2FS_DIRTY_DENTS || 2428 count_type == F2FS_DIRTY_NODES || 2429 count_type == F2FS_DIRTY_META || 2430 count_type == F2FS_DIRTY_QDATA || 2431 count_type == F2FS_DIRTY_IMETA) 2432 set_sbi_flag(sbi, SBI_IS_DIRTY); 2433 } 2434 2435 static inline void inode_inc_dirty_pages(struct inode *inode) 2436 { 2437 atomic_inc(&F2FS_I(inode)->dirty_pages); 2438 inc_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ? 2439 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA); 2440 if (IS_NOQUOTA(inode)) 2441 inc_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA); 2442 } 2443 2444 static inline void dec_page_count(struct f2fs_sb_info *sbi, int count_type) 2445 { 2446 atomic_dec(&sbi->nr_pages[count_type]); 2447 } 2448 2449 static inline void inode_dec_dirty_pages(struct inode *inode) 2450 { 2451 if (!S_ISDIR(inode->i_mode) && !S_ISREG(inode->i_mode) && 2452 !S_ISLNK(inode->i_mode)) 2453 return; 2454 2455 atomic_dec(&F2FS_I(inode)->dirty_pages); 2456 dec_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ? 2457 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA); 2458 if (IS_NOQUOTA(inode)) 2459 dec_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA); 2460 } 2461 2462 static inline void inc_atomic_write_cnt(struct inode *inode) 2463 { 2464 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 2465 struct f2fs_inode_info *fi = F2FS_I(inode); 2466 u64 current_write; 2467 2468 fi->atomic_write_cnt++; 2469 atomic64_inc(&sbi->current_atomic_write); 2470 current_write = atomic64_read(&sbi->current_atomic_write); 2471 if (current_write > sbi->peak_atomic_write) 2472 sbi->peak_atomic_write = current_write; 2473 } 2474 2475 static inline void release_atomic_write_cnt(struct inode *inode) 2476 { 2477 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 2478 struct f2fs_inode_info *fi = F2FS_I(inode); 2479 2480 atomic64_sub(fi->atomic_write_cnt, &sbi->current_atomic_write); 2481 fi->atomic_write_cnt = 0; 2482 } 2483 2484 static inline s64 get_pages(struct f2fs_sb_info *sbi, int count_type) 2485 { 2486 return atomic_read(&sbi->nr_pages[count_type]); 2487 } 2488 2489 static inline int get_dirty_pages(struct inode *inode) 2490 { 2491 return atomic_read(&F2FS_I(inode)->dirty_pages); 2492 } 2493 2494 static inline int get_blocktype_secs(struct f2fs_sb_info *sbi, int block_type) 2495 { 2496 unsigned int pages_per_sec = sbi->segs_per_sec * sbi->blocks_per_seg; 2497 unsigned int segs = (get_pages(sbi, block_type) + pages_per_sec - 1) >> 2498 sbi->log_blocks_per_seg; 2499 2500 return segs / sbi->segs_per_sec; 2501 } 2502 2503 static inline block_t valid_user_blocks(struct f2fs_sb_info *sbi) 2504 { 2505 return sbi->total_valid_block_count; 2506 } 2507 2508 static inline block_t discard_blocks(struct f2fs_sb_info *sbi) 2509 { 2510 return sbi->discard_blks; 2511 } 2512 2513 static inline unsigned long __bitmap_size(struct f2fs_sb_info *sbi, int flag) 2514 { 2515 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi); 2516 2517 /* return NAT or SIT bitmap */ 2518 if (flag == NAT_BITMAP) 2519 return le32_to_cpu(ckpt->nat_ver_bitmap_bytesize); 2520 else if (flag == SIT_BITMAP) 2521 return le32_to_cpu(ckpt->sit_ver_bitmap_bytesize); 2522 2523 return 0; 2524 } 2525 2526 static inline block_t __cp_payload(struct f2fs_sb_info *sbi) 2527 { 2528 return le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_payload); 2529 } 2530 2531 static inline void *__bitmap_ptr(struct f2fs_sb_info *sbi, int flag) 2532 { 2533 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi); 2534 void *tmp_ptr = &ckpt->sit_nat_version_bitmap; 2535 int offset; 2536 2537 if (is_set_ckpt_flags(sbi, CP_LARGE_NAT_BITMAP_FLAG)) { 2538 offset = (flag == SIT_BITMAP) ? 2539 le32_to_cpu(ckpt->nat_ver_bitmap_bytesize) : 0; 2540 /* 2541 * if large_nat_bitmap feature is enabled, leave checksum 2542 * protection for all nat/sit bitmaps. 2543 */ 2544 return tmp_ptr + offset + sizeof(__le32); 2545 } 2546 2547 if (__cp_payload(sbi) > 0) { 2548 if (flag == NAT_BITMAP) 2549 return tmp_ptr; 2550 else 2551 return (unsigned char *)ckpt + F2FS_BLKSIZE; 2552 } else { 2553 offset = (flag == NAT_BITMAP) ? 2554 le32_to_cpu(ckpt->sit_ver_bitmap_bytesize) : 0; 2555 return tmp_ptr + offset; 2556 } 2557 } 2558 2559 static inline block_t __start_cp_addr(struct f2fs_sb_info *sbi) 2560 { 2561 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr); 2562 2563 if (sbi->cur_cp_pack == 2) 2564 start_addr += sbi->blocks_per_seg; 2565 return start_addr; 2566 } 2567 2568 static inline block_t __start_cp_next_addr(struct f2fs_sb_info *sbi) 2569 { 2570 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr); 2571 2572 if (sbi->cur_cp_pack == 1) 2573 start_addr += sbi->blocks_per_seg; 2574 return start_addr; 2575 } 2576 2577 static inline void __set_cp_next_pack(struct f2fs_sb_info *sbi) 2578 { 2579 sbi->cur_cp_pack = (sbi->cur_cp_pack == 1) ? 2 : 1; 2580 } 2581 2582 static inline block_t __start_sum_addr(struct f2fs_sb_info *sbi) 2583 { 2584 return le32_to_cpu(F2FS_CKPT(sbi)->cp_pack_start_sum); 2585 } 2586 2587 extern void f2fs_mark_inode_dirty_sync(struct inode *inode, bool sync); 2588 static inline int inc_valid_node_count(struct f2fs_sb_info *sbi, 2589 struct inode *inode, bool is_inode) 2590 { 2591 block_t valid_block_count; 2592 unsigned int valid_node_count, user_block_count; 2593 int err; 2594 2595 if (is_inode) { 2596 if (inode) { 2597 err = dquot_alloc_inode(inode); 2598 if (err) 2599 return err; 2600 } 2601 } else { 2602 err = dquot_reserve_block(inode, 1); 2603 if (err) 2604 return err; 2605 } 2606 2607 if (time_to_inject(sbi, FAULT_BLOCK)) { 2608 f2fs_show_injection_info(sbi, FAULT_BLOCK); 2609 goto enospc; 2610 } 2611 2612 spin_lock(&sbi->stat_lock); 2613 2614 valid_block_count = sbi->total_valid_block_count + 2615 sbi->current_reserved_blocks + 1; 2616 2617 if (!__allow_reserved_blocks(sbi, inode, false)) 2618 valid_block_count += F2FS_OPTION(sbi).root_reserved_blocks; 2619 2620 if (F2FS_IO_ALIGNED(sbi)) 2621 valid_block_count += sbi->blocks_per_seg * 2622 SM_I(sbi)->additional_reserved_segments; 2623 2624 user_block_count = sbi->user_block_count; 2625 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) 2626 user_block_count -= sbi->unusable_block_count; 2627 2628 if (unlikely(valid_block_count > user_block_count)) { 2629 spin_unlock(&sbi->stat_lock); 2630 goto enospc; 2631 } 2632 2633 valid_node_count = sbi->total_valid_node_count + 1; 2634 if (unlikely(valid_node_count > sbi->total_node_count)) { 2635 spin_unlock(&sbi->stat_lock); 2636 goto enospc; 2637 } 2638 2639 sbi->total_valid_node_count++; 2640 sbi->total_valid_block_count++; 2641 spin_unlock(&sbi->stat_lock); 2642 2643 if (inode) { 2644 if (is_inode) 2645 f2fs_mark_inode_dirty_sync(inode, true); 2646 else 2647 f2fs_i_blocks_write(inode, 1, true, true); 2648 } 2649 2650 percpu_counter_inc(&sbi->alloc_valid_block_count); 2651 return 0; 2652 2653 enospc: 2654 if (is_inode) { 2655 if (inode) 2656 dquot_free_inode(inode); 2657 } else { 2658 dquot_release_reservation_block(inode, 1); 2659 } 2660 return -ENOSPC; 2661 } 2662 2663 static inline void dec_valid_node_count(struct f2fs_sb_info *sbi, 2664 struct inode *inode, bool is_inode) 2665 { 2666 spin_lock(&sbi->stat_lock); 2667 2668 if (unlikely(!sbi->total_valid_block_count || 2669 !sbi->total_valid_node_count)) { 2670 f2fs_warn(sbi, "dec_valid_node_count: inconsistent block counts, total_valid_block:%u, total_valid_node:%u", 2671 sbi->total_valid_block_count, 2672 sbi->total_valid_node_count); 2673 set_sbi_flag(sbi, SBI_NEED_FSCK); 2674 } else { 2675 sbi->total_valid_block_count--; 2676 sbi->total_valid_node_count--; 2677 } 2678 2679 if (sbi->reserved_blocks && 2680 sbi->current_reserved_blocks < sbi->reserved_blocks) 2681 sbi->current_reserved_blocks++; 2682 2683 spin_unlock(&sbi->stat_lock); 2684 2685 if (is_inode) { 2686 dquot_free_inode(inode); 2687 } else { 2688 if (unlikely(inode->i_blocks == 0)) { 2689 f2fs_warn(sbi, "dec_valid_node_count: inconsistent i_blocks, ino:%lu, iblocks:%llu", 2690 inode->i_ino, 2691 (unsigned long long)inode->i_blocks); 2692 set_sbi_flag(sbi, SBI_NEED_FSCK); 2693 return; 2694 } 2695 f2fs_i_blocks_write(inode, 1, false, true); 2696 } 2697 } 2698 2699 static inline unsigned int valid_node_count(struct f2fs_sb_info *sbi) 2700 { 2701 return sbi->total_valid_node_count; 2702 } 2703 2704 static inline void inc_valid_inode_count(struct f2fs_sb_info *sbi) 2705 { 2706 percpu_counter_inc(&sbi->total_valid_inode_count); 2707 } 2708 2709 static inline void dec_valid_inode_count(struct f2fs_sb_info *sbi) 2710 { 2711 percpu_counter_dec(&sbi->total_valid_inode_count); 2712 } 2713 2714 static inline s64 valid_inode_count(struct f2fs_sb_info *sbi) 2715 { 2716 return percpu_counter_sum_positive(&sbi->total_valid_inode_count); 2717 } 2718 2719 static inline struct page *f2fs_grab_cache_page(struct address_space *mapping, 2720 pgoff_t index, bool for_write) 2721 { 2722 struct page *page; 2723 unsigned int flags; 2724 2725 if (IS_ENABLED(CONFIG_F2FS_FAULT_INJECTION)) { 2726 if (!for_write) 2727 page = find_get_page_flags(mapping, index, 2728 FGP_LOCK | FGP_ACCESSED); 2729 else 2730 page = find_lock_page(mapping, index); 2731 if (page) 2732 return page; 2733 2734 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_ALLOC)) { 2735 f2fs_show_injection_info(F2FS_M_SB(mapping), 2736 FAULT_PAGE_ALLOC); 2737 return NULL; 2738 } 2739 } 2740 2741 if (!for_write) 2742 return grab_cache_page(mapping, index); 2743 2744 flags = memalloc_nofs_save(); 2745 page = grab_cache_page_write_begin(mapping, index); 2746 memalloc_nofs_restore(flags); 2747 2748 return page; 2749 } 2750 2751 static inline struct page *f2fs_pagecache_get_page( 2752 struct address_space *mapping, pgoff_t index, 2753 int fgp_flags, gfp_t gfp_mask) 2754 { 2755 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_GET)) { 2756 f2fs_show_injection_info(F2FS_M_SB(mapping), FAULT_PAGE_GET); 2757 return NULL; 2758 } 2759 2760 return pagecache_get_page(mapping, index, fgp_flags, gfp_mask); 2761 } 2762 2763 static inline void f2fs_put_page(struct page *page, int unlock) 2764 { 2765 if (!page) 2766 return; 2767 2768 if (unlock) { 2769 f2fs_bug_on(F2FS_P_SB(page), !PageLocked(page)); 2770 unlock_page(page); 2771 } 2772 put_page(page); 2773 } 2774 2775 static inline void f2fs_put_dnode(struct dnode_of_data *dn) 2776 { 2777 if (dn->node_page) 2778 f2fs_put_page(dn->node_page, 1); 2779 if (dn->inode_page && dn->node_page != dn->inode_page) 2780 f2fs_put_page(dn->inode_page, 0); 2781 dn->node_page = NULL; 2782 dn->inode_page = NULL; 2783 } 2784 2785 static inline struct kmem_cache *f2fs_kmem_cache_create(const char *name, 2786 size_t size) 2787 { 2788 return kmem_cache_create(name, size, 0, SLAB_RECLAIM_ACCOUNT, NULL); 2789 } 2790 2791 static inline void *f2fs_kmem_cache_alloc_nofail(struct kmem_cache *cachep, 2792 gfp_t flags) 2793 { 2794 void *entry; 2795 2796 entry = kmem_cache_alloc(cachep, flags); 2797 if (!entry) 2798 entry = kmem_cache_alloc(cachep, flags | __GFP_NOFAIL); 2799 return entry; 2800 } 2801 2802 static inline void *f2fs_kmem_cache_alloc(struct kmem_cache *cachep, 2803 gfp_t flags, bool nofail, struct f2fs_sb_info *sbi) 2804 { 2805 if (nofail) 2806 return f2fs_kmem_cache_alloc_nofail(cachep, flags); 2807 2808 if (time_to_inject(sbi, FAULT_SLAB_ALLOC)) { 2809 f2fs_show_injection_info(sbi, FAULT_SLAB_ALLOC); 2810 return NULL; 2811 } 2812 2813 return kmem_cache_alloc(cachep, flags); 2814 } 2815 2816 static inline bool is_inflight_io(struct f2fs_sb_info *sbi, int type) 2817 { 2818 if (get_pages(sbi, F2FS_RD_DATA) || get_pages(sbi, F2FS_RD_NODE) || 2819 get_pages(sbi, F2FS_RD_META) || get_pages(sbi, F2FS_WB_DATA) || 2820 get_pages(sbi, F2FS_WB_CP_DATA) || 2821 get_pages(sbi, F2FS_DIO_READ) || 2822 get_pages(sbi, F2FS_DIO_WRITE)) 2823 return true; 2824 2825 if (type != DISCARD_TIME && SM_I(sbi) && SM_I(sbi)->dcc_info && 2826 atomic_read(&SM_I(sbi)->dcc_info->queued_discard)) 2827 return true; 2828 2829 if (SM_I(sbi) && SM_I(sbi)->fcc_info && 2830 atomic_read(&SM_I(sbi)->fcc_info->queued_flush)) 2831 return true; 2832 return false; 2833 } 2834 2835 static inline bool is_idle(struct f2fs_sb_info *sbi, int type) 2836 { 2837 if (sbi->gc_mode == GC_URGENT_HIGH) 2838 return true; 2839 2840 if (is_inflight_io(sbi, type)) 2841 return false; 2842 2843 if (sbi->gc_mode == GC_URGENT_MID) 2844 return true; 2845 2846 if (sbi->gc_mode == GC_URGENT_LOW && 2847 (type == DISCARD_TIME || type == GC_TIME)) 2848 return true; 2849 2850 return f2fs_time_over(sbi, type); 2851 } 2852 2853 static inline void f2fs_radix_tree_insert(struct radix_tree_root *root, 2854 unsigned long index, void *item) 2855 { 2856 while (radix_tree_insert(root, index, item)) 2857 cond_resched(); 2858 } 2859 2860 #define RAW_IS_INODE(p) ((p)->footer.nid == (p)->footer.ino) 2861 2862 static inline bool IS_INODE(struct page *page) 2863 { 2864 struct f2fs_node *p = F2FS_NODE(page); 2865 2866 return RAW_IS_INODE(p); 2867 } 2868 2869 static inline int offset_in_addr(struct f2fs_inode *i) 2870 { 2871 return (i->i_inline & F2FS_EXTRA_ATTR) ? 2872 (le16_to_cpu(i->i_extra_isize) / sizeof(__le32)) : 0; 2873 } 2874 2875 static inline __le32 *blkaddr_in_node(struct f2fs_node *node) 2876 { 2877 return RAW_IS_INODE(node) ? node->i.i_addr : node->dn.addr; 2878 } 2879 2880 static inline int f2fs_has_extra_attr(struct inode *inode); 2881 static inline block_t data_blkaddr(struct inode *inode, 2882 struct page *node_page, unsigned int offset) 2883 { 2884 struct f2fs_node *raw_node; 2885 __le32 *addr_array; 2886 int base = 0; 2887 bool is_inode = IS_INODE(node_page); 2888 2889 raw_node = F2FS_NODE(node_page); 2890 2891 if (is_inode) { 2892 if (!inode) 2893 /* from GC path only */ 2894 base = offset_in_addr(&raw_node->i); 2895 else if (f2fs_has_extra_attr(inode)) 2896 base = get_extra_isize(inode); 2897 } 2898 2899 addr_array = blkaddr_in_node(raw_node); 2900 return le32_to_cpu(addr_array[base + offset]); 2901 } 2902 2903 static inline block_t f2fs_data_blkaddr(struct dnode_of_data *dn) 2904 { 2905 return data_blkaddr(dn->inode, dn->node_page, dn->ofs_in_node); 2906 } 2907 2908 static inline int f2fs_test_bit(unsigned int nr, char *addr) 2909 { 2910 int mask; 2911 2912 addr += (nr >> 3); 2913 mask = 1 << (7 - (nr & 0x07)); 2914 return mask & *addr; 2915 } 2916 2917 static inline void f2fs_set_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 void f2fs_clear_bit(unsigned int nr, char *addr) 2927 { 2928 int mask; 2929 2930 addr += (nr >> 3); 2931 mask = 1 << (7 - (nr & 0x07)); 2932 *addr &= ~mask; 2933 } 2934 2935 static inline int f2fs_test_and_set_bit(unsigned int nr, char *addr) 2936 { 2937 int mask; 2938 int ret; 2939 2940 addr += (nr >> 3); 2941 mask = 1 << (7 - (nr & 0x07)); 2942 ret = mask & *addr; 2943 *addr |= mask; 2944 return ret; 2945 } 2946 2947 static inline int f2fs_test_and_clear_bit(unsigned int nr, char *addr) 2948 { 2949 int mask; 2950 int ret; 2951 2952 addr += (nr >> 3); 2953 mask = 1 << (7 - (nr & 0x07)); 2954 ret = mask & *addr; 2955 *addr &= ~mask; 2956 return ret; 2957 } 2958 2959 static inline void f2fs_change_bit(unsigned int nr, char *addr) 2960 { 2961 int mask; 2962 2963 addr += (nr >> 3); 2964 mask = 1 << (7 - (nr & 0x07)); 2965 *addr ^= mask; 2966 } 2967 2968 /* 2969 * On-disk inode flags (f2fs_inode::i_flags) 2970 */ 2971 #define F2FS_COMPR_FL 0x00000004 /* Compress file */ 2972 #define F2FS_SYNC_FL 0x00000008 /* Synchronous updates */ 2973 #define F2FS_IMMUTABLE_FL 0x00000010 /* Immutable file */ 2974 #define F2FS_APPEND_FL 0x00000020 /* writes to file may only append */ 2975 #define F2FS_NODUMP_FL 0x00000040 /* do not dump file */ 2976 #define F2FS_NOATIME_FL 0x00000080 /* do not update atime */ 2977 #define F2FS_NOCOMP_FL 0x00000400 /* Don't compress */ 2978 #define F2FS_INDEX_FL 0x00001000 /* hash-indexed directory */ 2979 #define F2FS_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */ 2980 #define F2FS_PROJINHERIT_FL 0x20000000 /* Create with parents projid */ 2981 #define F2FS_CASEFOLD_FL 0x40000000 /* Casefolded file */ 2982 2983 /* Flags that should be inherited by new inodes from their parent. */ 2984 #define F2FS_FL_INHERITED (F2FS_SYNC_FL | F2FS_NODUMP_FL | F2FS_NOATIME_FL | \ 2985 F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \ 2986 F2FS_CASEFOLD_FL) 2987 2988 /* Flags that are appropriate for regular files (all but dir-specific ones). */ 2989 #define F2FS_REG_FLMASK (~(F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \ 2990 F2FS_CASEFOLD_FL)) 2991 2992 /* Flags that are appropriate for non-directories/regular files. */ 2993 #define F2FS_OTHER_FLMASK (F2FS_NODUMP_FL | F2FS_NOATIME_FL) 2994 2995 static inline __u32 f2fs_mask_flags(umode_t mode, __u32 flags) 2996 { 2997 if (S_ISDIR(mode)) 2998 return flags; 2999 else if (S_ISREG(mode)) 3000 return flags & F2FS_REG_FLMASK; 3001 else 3002 return flags & F2FS_OTHER_FLMASK; 3003 } 3004 3005 static inline void __mark_inode_dirty_flag(struct inode *inode, 3006 int flag, bool set) 3007 { 3008 switch (flag) { 3009 case FI_INLINE_XATTR: 3010 case FI_INLINE_DATA: 3011 case FI_INLINE_DENTRY: 3012 case FI_NEW_INODE: 3013 if (set) 3014 return; 3015 fallthrough; 3016 case FI_DATA_EXIST: 3017 case FI_INLINE_DOTS: 3018 case FI_PIN_FILE: 3019 case FI_COMPRESS_RELEASED: 3020 f2fs_mark_inode_dirty_sync(inode, true); 3021 } 3022 } 3023 3024 static inline void set_inode_flag(struct inode *inode, int flag) 3025 { 3026 set_bit(flag, F2FS_I(inode)->flags); 3027 __mark_inode_dirty_flag(inode, flag, true); 3028 } 3029 3030 static inline int is_inode_flag_set(struct inode *inode, int flag) 3031 { 3032 return test_bit(flag, F2FS_I(inode)->flags); 3033 } 3034 3035 static inline void clear_inode_flag(struct inode *inode, int flag) 3036 { 3037 clear_bit(flag, F2FS_I(inode)->flags); 3038 __mark_inode_dirty_flag(inode, flag, false); 3039 } 3040 3041 static inline bool f2fs_verity_in_progress(struct inode *inode) 3042 { 3043 return IS_ENABLED(CONFIG_FS_VERITY) && 3044 is_inode_flag_set(inode, FI_VERITY_IN_PROGRESS); 3045 } 3046 3047 static inline void set_acl_inode(struct inode *inode, umode_t mode) 3048 { 3049 F2FS_I(inode)->i_acl_mode = mode; 3050 set_inode_flag(inode, FI_ACL_MODE); 3051 f2fs_mark_inode_dirty_sync(inode, false); 3052 } 3053 3054 static inline void f2fs_i_links_write(struct inode *inode, bool inc) 3055 { 3056 if (inc) 3057 inc_nlink(inode); 3058 else 3059 drop_nlink(inode); 3060 f2fs_mark_inode_dirty_sync(inode, true); 3061 } 3062 3063 static inline void f2fs_i_blocks_write(struct inode *inode, 3064 block_t diff, bool add, bool claim) 3065 { 3066 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE); 3067 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER); 3068 3069 /* add = 1, claim = 1 should be dquot_reserve_block in pair */ 3070 if (add) { 3071 if (claim) 3072 dquot_claim_block(inode, diff); 3073 else 3074 dquot_alloc_block_nofail(inode, diff); 3075 } else { 3076 dquot_free_block(inode, diff); 3077 } 3078 3079 f2fs_mark_inode_dirty_sync(inode, true); 3080 if (clean || recover) 3081 set_inode_flag(inode, FI_AUTO_RECOVER); 3082 } 3083 3084 static inline bool f2fs_is_atomic_file(struct inode *inode); 3085 3086 static inline void f2fs_i_size_write(struct inode *inode, loff_t i_size) 3087 { 3088 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE); 3089 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER); 3090 3091 if (i_size_read(inode) == i_size) 3092 return; 3093 3094 i_size_write(inode, i_size); 3095 3096 if (f2fs_is_atomic_file(inode)) 3097 return; 3098 3099 f2fs_mark_inode_dirty_sync(inode, true); 3100 if (clean || recover) 3101 set_inode_flag(inode, FI_AUTO_RECOVER); 3102 } 3103 3104 static inline void f2fs_i_depth_write(struct inode *inode, unsigned int depth) 3105 { 3106 F2FS_I(inode)->i_current_depth = depth; 3107 f2fs_mark_inode_dirty_sync(inode, true); 3108 } 3109 3110 static inline void f2fs_i_gc_failures_write(struct inode *inode, 3111 unsigned int count) 3112 { 3113 F2FS_I(inode)->i_gc_failures[GC_FAILURE_PIN] = count; 3114 f2fs_mark_inode_dirty_sync(inode, true); 3115 } 3116 3117 static inline void f2fs_i_xnid_write(struct inode *inode, nid_t xnid) 3118 { 3119 F2FS_I(inode)->i_xattr_nid = xnid; 3120 f2fs_mark_inode_dirty_sync(inode, true); 3121 } 3122 3123 static inline void f2fs_i_pino_write(struct inode *inode, nid_t pino) 3124 { 3125 F2FS_I(inode)->i_pino = pino; 3126 f2fs_mark_inode_dirty_sync(inode, true); 3127 } 3128 3129 static inline void get_inline_info(struct inode *inode, struct f2fs_inode *ri) 3130 { 3131 struct f2fs_inode_info *fi = F2FS_I(inode); 3132 3133 if (ri->i_inline & F2FS_INLINE_XATTR) 3134 set_bit(FI_INLINE_XATTR, fi->flags); 3135 if (ri->i_inline & F2FS_INLINE_DATA) 3136 set_bit(FI_INLINE_DATA, fi->flags); 3137 if (ri->i_inline & F2FS_INLINE_DENTRY) 3138 set_bit(FI_INLINE_DENTRY, fi->flags); 3139 if (ri->i_inline & F2FS_DATA_EXIST) 3140 set_bit(FI_DATA_EXIST, fi->flags); 3141 if (ri->i_inline & F2FS_INLINE_DOTS) 3142 set_bit(FI_INLINE_DOTS, fi->flags); 3143 if (ri->i_inline & F2FS_EXTRA_ATTR) 3144 set_bit(FI_EXTRA_ATTR, fi->flags); 3145 if (ri->i_inline & F2FS_PIN_FILE) 3146 set_bit(FI_PIN_FILE, fi->flags); 3147 if (ri->i_inline & F2FS_COMPRESS_RELEASED) 3148 set_bit(FI_COMPRESS_RELEASED, fi->flags); 3149 } 3150 3151 static inline void set_raw_inline(struct inode *inode, struct f2fs_inode *ri) 3152 { 3153 ri->i_inline = 0; 3154 3155 if (is_inode_flag_set(inode, FI_INLINE_XATTR)) 3156 ri->i_inline |= F2FS_INLINE_XATTR; 3157 if (is_inode_flag_set(inode, FI_INLINE_DATA)) 3158 ri->i_inline |= F2FS_INLINE_DATA; 3159 if (is_inode_flag_set(inode, FI_INLINE_DENTRY)) 3160 ri->i_inline |= F2FS_INLINE_DENTRY; 3161 if (is_inode_flag_set(inode, FI_DATA_EXIST)) 3162 ri->i_inline |= F2FS_DATA_EXIST; 3163 if (is_inode_flag_set(inode, FI_INLINE_DOTS)) 3164 ri->i_inline |= F2FS_INLINE_DOTS; 3165 if (is_inode_flag_set(inode, FI_EXTRA_ATTR)) 3166 ri->i_inline |= F2FS_EXTRA_ATTR; 3167 if (is_inode_flag_set(inode, FI_PIN_FILE)) 3168 ri->i_inline |= F2FS_PIN_FILE; 3169 if (is_inode_flag_set(inode, FI_COMPRESS_RELEASED)) 3170 ri->i_inline |= F2FS_COMPRESS_RELEASED; 3171 } 3172 3173 static inline int f2fs_has_extra_attr(struct inode *inode) 3174 { 3175 return is_inode_flag_set(inode, FI_EXTRA_ATTR); 3176 } 3177 3178 static inline int f2fs_has_inline_xattr(struct inode *inode) 3179 { 3180 return is_inode_flag_set(inode, FI_INLINE_XATTR); 3181 } 3182 3183 static inline int f2fs_compressed_file(struct inode *inode) 3184 { 3185 return S_ISREG(inode->i_mode) && 3186 is_inode_flag_set(inode, FI_COMPRESSED_FILE); 3187 } 3188 3189 static inline bool f2fs_need_compress_data(struct inode *inode) 3190 { 3191 int compress_mode = F2FS_OPTION(F2FS_I_SB(inode)).compress_mode; 3192 3193 if (!f2fs_compressed_file(inode)) 3194 return false; 3195 3196 if (compress_mode == COMPR_MODE_FS) 3197 return true; 3198 else if (compress_mode == COMPR_MODE_USER && 3199 is_inode_flag_set(inode, FI_ENABLE_COMPRESS)) 3200 return true; 3201 3202 return false; 3203 } 3204 3205 static inline unsigned int addrs_per_inode(struct inode *inode) 3206 { 3207 unsigned int addrs = CUR_ADDRS_PER_INODE(inode) - 3208 get_inline_xattr_addrs(inode); 3209 3210 if (!f2fs_compressed_file(inode)) 3211 return addrs; 3212 return ALIGN_DOWN(addrs, F2FS_I(inode)->i_cluster_size); 3213 } 3214 3215 static inline unsigned int addrs_per_block(struct inode *inode) 3216 { 3217 if (!f2fs_compressed_file(inode)) 3218 return DEF_ADDRS_PER_BLOCK; 3219 return ALIGN_DOWN(DEF_ADDRS_PER_BLOCK, F2FS_I(inode)->i_cluster_size); 3220 } 3221 3222 static inline void *inline_xattr_addr(struct inode *inode, struct page *page) 3223 { 3224 struct f2fs_inode *ri = F2FS_INODE(page); 3225 3226 return (void *)&(ri->i_addr[DEF_ADDRS_PER_INODE - 3227 get_inline_xattr_addrs(inode)]); 3228 } 3229 3230 static inline int inline_xattr_size(struct inode *inode) 3231 { 3232 if (f2fs_has_inline_xattr(inode)) 3233 return get_inline_xattr_addrs(inode) * sizeof(__le32); 3234 return 0; 3235 } 3236 3237 /* 3238 * Notice: check inline_data flag without inode page lock is unsafe. 3239 * It could change at any time by f2fs_convert_inline_page(). 3240 */ 3241 static inline int f2fs_has_inline_data(struct inode *inode) 3242 { 3243 return is_inode_flag_set(inode, FI_INLINE_DATA); 3244 } 3245 3246 static inline int f2fs_exist_data(struct inode *inode) 3247 { 3248 return is_inode_flag_set(inode, FI_DATA_EXIST); 3249 } 3250 3251 static inline int f2fs_has_inline_dots(struct inode *inode) 3252 { 3253 return is_inode_flag_set(inode, FI_INLINE_DOTS); 3254 } 3255 3256 static inline int f2fs_is_mmap_file(struct inode *inode) 3257 { 3258 return is_inode_flag_set(inode, FI_MMAP_FILE); 3259 } 3260 3261 static inline bool f2fs_is_pinned_file(struct inode *inode) 3262 { 3263 return is_inode_flag_set(inode, FI_PIN_FILE); 3264 } 3265 3266 static inline bool f2fs_is_atomic_file(struct inode *inode) 3267 { 3268 return is_inode_flag_set(inode, FI_ATOMIC_FILE); 3269 } 3270 3271 static inline bool f2fs_is_cow_file(struct inode *inode) 3272 { 3273 return is_inode_flag_set(inode, FI_COW_FILE); 3274 } 3275 3276 static inline bool f2fs_is_first_block_written(struct inode *inode) 3277 { 3278 return is_inode_flag_set(inode, FI_FIRST_BLOCK_WRITTEN); 3279 } 3280 3281 static inline bool f2fs_is_drop_cache(struct inode *inode) 3282 { 3283 return is_inode_flag_set(inode, FI_DROP_CACHE); 3284 } 3285 3286 static inline void *inline_data_addr(struct inode *inode, struct page *page) 3287 { 3288 struct f2fs_inode *ri = F2FS_INODE(page); 3289 int extra_size = get_extra_isize(inode); 3290 3291 return (void *)&(ri->i_addr[extra_size + DEF_INLINE_RESERVED_SIZE]); 3292 } 3293 3294 static inline int f2fs_has_inline_dentry(struct inode *inode) 3295 { 3296 return is_inode_flag_set(inode, FI_INLINE_DENTRY); 3297 } 3298 3299 static inline int is_file(struct inode *inode, int type) 3300 { 3301 return F2FS_I(inode)->i_advise & type; 3302 } 3303 3304 static inline void set_file(struct inode *inode, int type) 3305 { 3306 if (is_file(inode, type)) 3307 return; 3308 F2FS_I(inode)->i_advise |= type; 3309 f2fs_mark_inode_dirty_sync(inode, true); 3310 } 3311 3312 static inline void clear_file(struct inode *inode, int type) 3313 { 3314 if (!is_file(inode, type)) 3315 return; 3316 F2FS_I(inode)->i_advise &= ~type; 3317 f2fs_mark_inode_dirty_sync(inode, true); 3318 } 3319 3320 static inline bool f2fs_is_time_consistent(struct inode *inode) 3321 { 3322 if (!timespec64_equal(F2FS_I(inode)->i_disk_time, &inode->i_atime)) 3323 return false; 3324 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 1, &inode->i_ctime)) 3325 return false; 3326 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 2, &inode->i_mtime)) 3327 return false; 3328 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 3, 3329 &F2FS_I(inode)->i_crtime)) 3330 return false; 3331 return true; 3332 } 3333 3334 static inline bool f2fs_skip_inode_update(struct inode *inode, int dsync) 3335 { 3336 bool ret; 3337 3338 if (dsync) { 3339 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 3340 3341 spin_lock(&sbi->inode_lock[DIRTY_META]); 3342 ret = list_empty(&F2FS_I(inode)->gdirty_list); 3343 spin_unlock(&sbi->inode_lock[DIRTY_META]); 3344 return ret; 3345 } 3346 if (!is_inode_flag_set(inode, FI_AUTO_RECOVER) || 3347 file_keep_isize(inode) || 3348 i_size_read(inode) & ~PAGE_MASK) 3349 return false; 3350 3351 if (!f2fs_is_time_consistent(inode)) 3352 return false; 3353 3354 spin_lock(&F2FS_I(inode)->i_size_lock); 3355 ret = F2FS_I(inode)->last_disk_size == i_size_read(inode); 3356 spin_unlock(&F2FS_I(inode)->i_size_lock); 3357 3358 return ret; 3359 } 3360 3361 static inline bool f2fs_readonly(struct super_block *sb) 3362 { 3363 return sb_rdonly(sb); 3364 } 3365 3366 static inline bool f2fs_cp_error(struct f2fs_sb_info *sbi) 3367 { 3368 return is_set_ckpt_flags(sbi, CP_ERROR_FLAG); 3369 } 3370 3371 static inline bool is_dot_dotdot(const u8 *name, size_t len) 3372 { 3373 if (len == 1 && name[0] == '.') 3374 return true; 3375 3376 if (len == 2 && name[0] == '.' && name[1] == '.') 3377 return true; 3378 3379 return false; 3380 } 3381 3382 static inline void *f2fs_kmalloc(struct f2fs_sb_info *sbi, 3383 size_t size, gfp_t flags) 3384 { 3385 if (time_to_inject(sbi, FAULT_KMALLOC)) { 3386 f2fs_show_injection_info(sbi, FAULT_KMALLOC); 3387 return NULL; 3388 } 3389 3390 return kmalloc(size, flags); 3391 } 3392 3393 static inline void *f2fs_kzalloc(struct f2fs_sb_info *sbi, 3394 size_t size, gfp_t flags) 3395 { 3396 return f2fs_kmalloc(sbi, size, flags | __GFP_ZERO); 3397 } 3398 3399 static inline void *f2fs_kvmalloc(struct f2fs_sb_info *sbi, 3400 size_t size, gfp_t flags) 3401 { 3402 if (time_to_inject(sbi, FAULT_KVMALLOC)) { 3403 f2fs_show_injection_info(sbi, FAULT_KVMALLOC); 3404 return NULL; 3405 } 3406 3407 return kvmalloc(size, flags); 3408 } 3409 3410 static inline void *f2fs_kvzalloc(struct f2fs_sb_info *sbi, 3411 size_t size, gfp_t flags) 3412 { 3413 return f2fs_kvmalloc(sbi, size, flags | __GFP_ZERO); 3414 } 3415 3416 static inline int get_extra_isize(struct inode *inode) 3417 { 3418 return F2FS_I(inode)->i_extra_isize / sizeof(__le32); 3419 } 3420 3421 static inline int get_inline_xattr_addrs(struct inode *inode) 3422 { 3423 return F2FS_I(inode)->i_inline_xattr_size; 3424 } 3425 3426 #define f2fs_get_inode_mode(i) \ 3427 ((is_inode_flag_set(i, FI_ACL_MODE)) ? \ 3428 (F2FS_I(i)->i_acl_mode) : ((i)->i_mode)) 3429 3430 #define F2FS_TOTAL_EXTRA_ATTR_SIZE \ 3431 (offsetof(struct f2fs_inode, i_extra_end) - \ 3432 offsetof(struct f2fs_inode, i_extra_isize)) \ 3433 3434 #define F2FS_OLD_ATTRIBUTE_SIZE (offsetof(struct f2fs_inode, i_addr)) 3435 #define F2FS_FITS_IN_INODE(f2fs_inode, extra_isize, field) \ 3436 ((offsetof(typeof(*(f2fs_inode)), field) + \ 3437 sizeof((f2fs_inode)->field)) \ 3438 <= (F2FS_OLD_ATTRIBUTE_SIZE + (extra_isize))) \ 3439 3440 #define __is_large_section(sbi) ((sbi)->segs_per_sec > 1) 3441 3442 #define __is_meta_io(fio) (PAGE_TYPE_OF_BIO((fio)->type) == META) 3443 3444 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi, 3445 block_t blkaddr, int type); 3446 static inline void verify_blkaddr(struct f2fs_sb_info *sbi, 3447 block_t blkaddr, int type) 3448 { 3449 if (!f2fs_is_valid_blkaddr(sbi, blkaddr, type)) { 3450 f2fs_err(sbi, "invalid blkaddr: %u, type: %d, run fsck to fix.", 3451 blkaddr, type); 3452 f2fs_bug_on(sbi, 1); 3453 } 3454 } 3455 3456 static inline bool __is_valid_data_blkaddr(block_t blkaddr) 3457 { 3458 if (blkaddr == NEW_ADDR || blkaddr == NULL_ADDR || 3459 blkaddr == COMPRESS_ADDR) 3460 return false; 3461 return true; 3462 } 3463 3464 /* 3465 * file.c 3466 */ 3467 int f2fs_sync_file(struct file *file, loff_t start, loff_t end, int datasync); 3468 void f2fs_truncate_data_blocks(struct dnode_of_data *dn); 3469 int f2fs_do_truncate_blocks(struct inode *inode, u64 from, bool lock); 3470 int f2fs_truncate_blocks(struct inode *inode, u64 from, bool lock); 3471 int f2fs_truncate(struct inode *inode); 3472 int f2fs_getattr(struct user_namespace *mnt_userns, const struct path *path, 3473 struct kstat *stat, u32 request_mask, unsigned int flags); 3474 int f2fs_setattr(struct user_namespace *mnt_userns, struct dentry *dentry, 3475 struct iattr *attr); 3476 int f2fs_truncate_hole(struct inode *inode, pgoff_t pg_start, pgoff_t pg_end); 3477 void f2fs_truncate_data_blocks_range(struct dnode_of_data *dn, int count); 3478 int f2fs_precache_extents(struct inode *inode); 3479 int f2fs_fileattr_get(struct dentry *dentry, struct fileattr *fa); 3480 int f2fs_fileattr_set(struct user_namespace *mnt_userns, 3481 struct dentry *dentry, struct fileattr *fa); 3482 long f2fs_ioctl(struct file *filp, unsigned int cmd, unsigned long arg); 3483 long f2fs_compat_ioctl(struct file *file, unsigned int cmd, unsigned long arg); 3484 int f2fs_transfer_project_quota(struct inode *inode, kprojid_t kprojid); 3485 int f2fs_pin_file_control(struct inode *inode, bool inc); 3486 3487 /* 3488 * inode.c 3489 */ 3490 void f2fs_set_inode_flags(struct inode *inode); 3491 bool f2fs_inode_chksum_verify(struct f2fs_sb_info *sbi, struct page *page); 3492 void f2fs_inode_chksum_set(struct f2fs_sb_info *sbi, struct page *page); 3493 struct inode *f2fs_iget(struct super_block *sb, unsigned long ino); 3494 struct inode *f2fs_iget_retry(struct super_block *sb, unsigned long ino); 3495 int f2fs_try_to_free_nats(struct f2fs_sb_info *sbi, int nr_shrink); 3496 void f2fs_update_inode(struct inode *inode, struct page *node_page); 3497 void f2fs_update_inode_page(struct inode *inode); 3498 int f2fs_write_inode(struct inode *inode, struct writeback_control *wbc); 3499 void f2fs_evict_inode(struct inode *inode); 3500 void f2fs_handle_failed_inode(struct inode *inode); 3501 3502 /* 3503 * namei.c 3504 */ 3505 int f2fs_update_extension_list(struct f2fs_sb_info *sbi, const char *name, 3506 bool hot, bool set); 3507 struct dentry *f2fs_get_parent(struct dentry *child); 3508 int f2fs_get_tmpfile(struct user_namespace *mnt_userns, struct inode *dir, 3509 struct inode **new_inode); 3510 3511 /* 3512 * dir.c 3513 */ 3514 unsigned char f2fs_get_de_type(struct f2fs_dir_entry *de); 3515 int f2fs_init_casefolded_name(const struct inode *dir, 3516 struct f2fs_filename *fname); 3517 int f2fs_setup_filename(struct inode *dir, const struct qstr *iname, 3518 int lookup, struct f2fs_filename *fname); 3519 int f2fs_prepare_lookup(struct inode *dir, struct dentry *dentry, 3520 struct f2fs_filename *fname); 3521 void f2fs_free_filename(struct f2fs_filename *fname); 3522 struct f2fs_dir_entry *f2fs_find_target_dentry(const struct f2fs_dentry_ptr *d, 3523 const struct f2fs_filename *fname, int *max_slots); 3524 int f2fs_fill_dentries(struct dir_context *ctx, struct f2fs_dentry_ptr *d, 3525 unsigned int start_pos, struct fscrypt_str *fstr); 3526 void f2fs_do_make_empty_dir(struct inode *inode, struct inode *parent, 3527 struct f2fs_dentry_ptr *d); 3528 struct page *f2fs_init_inode_metadata(struct inode *inode, struct inode *dir, 3529 const struct f2fs_filename *fname, struct page *dpage); 3530 void f2fs_update_parent_metadata(struct inode *dir, struct inode *inode, 3531 unsigned int current_depth); 3532 int f2fs_room_for_filename(const void *bitmap, int slots, int max_slots); 3533 void f2fs_drop_nlink(struct inode *dir, struct inode *inode); 3534 struct f2fs_dir_entry *__f2fs_find_entry(struct inode *dir, 3535 const struct f2fs_filename *fname, 3536 struct page **res_page); 3537 struct f2fs_dir_entry *f2fs_find_entry(struct inode *dir, 3538 const struct qstr *child, struct page **res_page); 3539 struct f2fs_dir_entry *f2fs_parent_dir(struct inode *dir, struct page **p); 3540 ino_t f2fs_inode_by_name(struct inode *dir, const struct qstr *qstr, 3541 struct page **page); 3542 void f2fs_set_link(struct inode *dir, struct f2fs_dir_entry *de, 3543 struct page *page, struct inode *inode); 3544 bool f2fs_has_enough_room(struct inode *dir, struct page *ipage, 3545 const struct f2fs_filename *fname); 3546 void f2fs_update_dentry(nid_t ino, umode_t mode, struct f2fs_dentry_ptr *d, 3547 const struct fscrypt_str *name, f2fs_hash_t name_hash, 3548 unsigned int bit_pos); 3549 int f2fs_add_regular_entry(struct inode *dir, const struct f2fs_filename *fname, 3550 struct inode *inode, nid_t ino, umode_t mode); 3551 int f2fs_add_dentry(struct inode *dir, const struct f2fs_filename *fname, 3552 struct inode *inode, nid_t ino, umode_t mode); 3553 int f2fs_do_add_link(struct inode *dir, const struct qstr *name, 3554 struct inode *inode, nid_t ino, umode_t mode); 3555 void f2fs_delete_entry(struct f2fs_dir_entry *dentry, struct page *page, 3556 struct inode *dir, struct inode *inode); 3557 int f2fs_do_tmpfile(struct inode *inode, struct inode *dir); 3558 bool f2fs_empty_dir(struct inode *dir); 3559 3560 static inline int f2fs_add_link(struct dentry *dentry, struct inode *inode) 3561 { 3562 if (fscrypt_is_nokey_name(dentry)) 3563 return -ENOKEY; 3564 return f2fs_do_add_link(d_inode(dentry->d_parent), &dentry->d_name, 3565 inode, inode->i_ino, inode->i_mode); 3566 } 3567 3568 /* 3569 * super.c 3570 */ 3571 int f2fs_inode_dirtied(struct inode *inode, bool sync); 3572 void f2fs_inode_synced(struct inode *inode); 3573 int f2fs_dquot_initialize(struct inode *inode); 3574 int f2fs_enable_quota_files(struct f2fs_sb_info *sbi, bool rdonly); 3575 int f2fs_quota_sync(struct super_block *sb, int type); 3576 loff_t max_file_blocks(struct inode *inode); 3577 void f2fs_quota_off_umount(struct super_block *sb); 3578 void f2fs_handle_stop(struct f2fs_sb_info *sbi, unsigned char reason); 3579 void f2fs_handle_error(struct f2fs_sb_info *sbi, unsigned char error); 3580 int f2fs_commit_super(struct f2fs_sb_info *sbi, bool recover); 3581 int f2fs_sync_fs(struct super_block *sb, int sync); 3582 int f2fs_sanity_check_ckpt(struct f2fs_sb_info *sbi); 3583 3584 /* 3585 * hash.c 3586 */ 3587 void f2fs_hash_filename(const struct inode *dir, struct f2fs_filename *fname); 3588 3589 /* 3590 * node.c 3591 */ 3592 struct node_info; 3593 3594 int f2fs_check_nid_range(struct f2fs_sb_info *sbi, nid_t nid); 3595 bool f2fs_available_free_memory(struct f2fs_sb_info *sbi, int type); 3596 bool f2fs_in_warm_node_list(struct f2fs_sb_info *sbi, struct page *page); 3597 void f2fs_init_fsync_node_info(struct f2fs_sb_info *sbi); 3598 void f2fs_del_fsync_node_entry(struct f2fs_sb_info *sbi, struct page *page); 3599 void f2fs_reset_fsync_node_info(struct f2fs_sb_info *sbi); 3600 int f2fs_need_dentry_mark(struct f2fs_sb_info *sbi, nid_t nid); 3601 bool f2fs_is_checkpointed_node(struct f2fs_sb_info *sbi, nid_t nid); 3602 bool f2fs_need_inode_block_update(struct f2fs_sb_info *sbi, nid_t ino); 3603 int f2fs_get_node_info(struct f2fs_sb_info *sbi, nid_t nid, 3604 struct node_info *ni, bool checkpoint_context); 3605 pgoff_t f2fs_get_next_page_offset(struct dnode_of_data *dn, pgoff_t pgofs); 3606 int f2fs_get_dnode_of_data(struct dnode_of_data *dn, pgoff_t index, int mode); 3607 int f2fs_truncate_inode_blocks(struct inode *inode, pgoff_t from); 3608 int f2fs_truncate_xattr_node(struct inode *inode); 3609 int f2fs_wait_on_node_pages_writeback(struct f2fs_sb_info *sbi, 3610 unsigned int seq_id); 3611 bool f2fs_nat_bitmap_enabled(struct f2fs_sb_info *sbi); 3612 int f2fs_remove_inode_page(struct inode *inode); 3613 struct page *f2fs_new_inode_page(struct inode *inode); 3614 struct page *f2fs_new_node_page(struct dnode_of_data *dn, unsigned int ofs); 3615 void f2fs_ra_node_page(struct f2fs_sb_info *sbi, nid_t nid); 3616 struct page *f2fs_get_node_page(struct f2fs_sb_info *sbi, pgoff_t nid); 3617 struct page *f2fs_get_node_page_ra(struct page *parent, int start); 3618 int f2fs_move_node_page(struct page *node_page, int gc_type); 3619 void f2fs_flush_inline_data(struct f2fs_sb_info *sbi); 3620 int f2fs_fsync_node_pages(struct f2fs_sb_info *sbi, struct inode *inode, 3621 struct writeback_control *wbc, bool atomic, 3622 unsigned int *seq_id); 3623 int f2fs_sync_node_pages(struct f2fs_sb_info *sbi, 3624 struct writeback_control *wbc, 3625 bool do_balance, enum iostat_type io_type); 3626 int f2fs_build_free_nids(struct f2fs_sb_info *sbi, bool sync, bool mount); 3627 bool f2fs_alloc_nid(struct f2fs_sb_info *sbi, nid_t *nid); 3628 void f2fs_alloc_nid_done(struct f2fs_sb_info *sbi, nid_t nid); 3629 void f2fs_alloc_nid_failed(struct f2fs_sb_info *sbi, nid_t nid); 3630 int f2fs_try_to_free_nids(struct f2fs_sb_info *sbi, int nr_shrink); 3631 int f2fs_recover_inline_xattr(struct inode *inode, struct page *page); 3632 int f2fs_recover_xattr_data(struct inode *inode, struct page *page); 3633 int f2fs_recover_inode_page(struct f2fs_sb_info *sbi, struct page *page); 3634 int f2fs_restore_node_summary(struct f2fs_sb_info *sbi, 3635 unsigned int segno, struct f2fs_summary_block *sum); 3636 void f2fs_enable_nat_bits(struct f2fs_sb_info *sbi); 3637 int f2fs_flush_nat_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3638 int f2fs_build_node_manager(struct f2fs_sb_info *sbi); 3639 void f2fs_destroy_node_manager(struct f2fs_sb_info *sbi); 3640 int __init f2fs_create_node_manager_caches(void); 3641 void f2fs_destroy_node_manager_caches(void); 3642 3643 /* 3644 * segment.c 3645 */ 3646 bool f2fs_need_SSR(struct f2fs_sb_info *sbi); 3647 int f2fs_commit_atomic_write(struct inode *inode); 3648 void f2fs_abort_atomic_write(struct inode *inode, bool clean); 3649 void f2fs_balance_fs(struct f2fs_sb_info *sbi, bool need); 3650 void f2fs_balance_fs_bg(struct f2fs_sb_info *sbi, bool from_bg); 3651 int f2fs_issue_flush(struct f2fs_sb_info *sbi, nid_t ino); 3652 int f2fs_create_flush_cmd_control(struct f2fs_sb_info *sbi); 3653 int f2fs_flush_device_cache(struct f2fs_sb_info *sbi); 3654 void f2fs_destroy_flush_cmd_control(struct f2fs_sb_info *sbi, bool free); 3655 void f2fs_invalidate_blocks(struct f2fs_sb_info *sbi, block_t addr); 3656 bool f2fs_is_checkpointed_data(struct f2fs_sb_info *sbi, block_t blkaddr); 3657 int f2fs_start_discard_thread(struct f2fs_sb_info *sbi); 3658 void f2fs_drop_discard_cmd(struct f2fs_sb_info *sbi); 3659 void f2fs_stop_discard_thread(struct f2fs_sb_info *sbi); 3660 bool f2fs_issue_discard_timeout(struct f2fs_sb_info *sbi); 3661 void f2fs_clear_prefree_segments(struct f2fs_sb_info *sbi, 3662 struct cp_control *cpc); 3663 void f2fs_dirty_to_prefree(struct f2fs_sb_info *sbi); 3664 block_t f2fs_get_unusable_blocks(struct f2fs_sb_info *sbi); 3665 int f2fs_disable_cp_again(struct f2fs_sb_info *sbi, block_t unusable); 3666 void f2fs_release_discard_addrs(struct f2fs_sb_info *sbi); 3667 int f2fs_npages_for_summary_flush(struct f2fs_sb_info *sbi, bool for_ra); 3668 bool f2fs_segment_has_free_slot(struct f2fs_sb_info *sbi, int segno); 3669 void f2fs_init_inmem_curseg(struct f2fs_sb_info *sbi); 3670 void f2fs_save_inmem_curseg(struct f2fs_sb_info *sbi); 3671 void f2fs_restore_inmem_curseg(struct f2fs_sb_info *sbi); 3672 void f2fs_get_new_segment(struct f2fs_sb_info *sbi, 3673 unsigned int *newseg, bool new_sec, int dir); 3674 void f2fs_allocate_segment_for_resize(struct f2fs_sb_info *sbi, int type, 3675 unsigned int start, unsigned int end); 3676 void f2fs_allocate_new_section(struct f2fs_sb_info *sbi, int type, bool force); 3677 void f2fs_allocate_new_segments(struct f2fs_sb_info *sbi); 3678 int f2fs_trim_fs(struct f2fs_sb_info *sbi, struct fstrim_range *range); 3679 bool f2fs_exist_trim_candidates(struct f2fs_sb_info *sbi, 3680 struct cp_control *cpc); 3681 struct page *f2fs_get_sum_page(struct f2fs_sb_info *sbi, unsigned int segno); 3682 void f2fs_update_meta_page(struct f2fs_sb_info *sbi, void *src, 3683 block_t blk_addr); 3684 void f2fs_do_write_meta_page(struct f2fs_sb_info *sbi, struct page *page, 3685 enum iostat_type io_type); 3686 void f2fs_do_write_node_page(unsigned int nid, struct f2fs_io_info *fio); 3687 void f2fs_outplace_write_data(struct dnode_of_data *dn, 3688 struct f2fs_io_info *fio); 3689 int f2fs_inplace_write_data(struct f2fs_io_info *fio); 3690 void f2fs_do_replace_block(struct f2fs_sb_info *sbi, struct f2fs_summary *sum, 3691 block_t old_blkaddr, block_t new_blkaddr, 3692 bool recover_curseg, bool recover_newaddr, 3693 bool from_gc); 3694 void f2fs_replace_block(struct f2fs_sb_info *sbi, struct dnode_of_data *dn, 3695 block_t old_addr, block_t new_addr, 3696 unsigned char version, bool recover_curseg, 3697 bool recover_newaddr); 3698 void f2fs_allocate_data_block(struct f2fs_sb_info *sbi, struct page *page, 3699 block_t old_blkaddr, block_t *new_blkaddr, 3700 struct f2fs_summary *sum, int type, 3701 struct f2fs_io_info *fio); 3702 void f2fs_update_device_state(struct f2fs_sb_info *sbi, nid_t ino, 3703 block_t blkaddr, unsigned int blkcnt); 3704 void f2fs_wait_on_page_writeback(struct page *page, 3705 enum page_type type, bool ordered, bool locked); 3706 void f2fs_wait_on_block_writeback(struct inode *inode, block_t blkaddr); 3707 void f2fs_wait_on_block_writeback_range(struct inode *inode, block_t blkaddr, 3708 block_t len); 3709 void f2fs_write_data_summaries(struct f2fs_sb_info *sbi, block_t start_blk); 3710 void f2fs_write_node_summaries(struct f2fs_sb_info *sbi, block_t start_blk); 3711 int f2fs_lookup_journal_in_cursum(struct f2fs_journal *journal, int type, 3712 unsigned int val, int alloc); 3713 void f2fs_flush_sit_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3714 int f2fs_fix_curseg_write_pointer(struct f2fs_sb_info *sbi); 3715 int f2fs_check_write_pointer(struct f2fs_sb_info *sbi); 3716 int f2fs_build_segment_manager(struct f2fs_sb_info *sbi); 3717 void f2fs_destroy_segment_manager(struct f2fs_sb_info *sbi); 3718 int __init f2fs_create_segment_manager_caches(void); 3719 void f2fs_destroy_segment_manager_caches(void); 3720 int f2fs_rw_hint_to_seg_type(enum rw_hint hint); 3721 unsigned int f2fs_usable_segs_in_sec(struct f2fs_sb_info *sbi, 3722 unsigned int segno); 3723 unsigned int f2fs_usable_blks_in_seg(struct f2fs_sb_info *sbi, 3724 unsigned int segno); 3725 3726 #define DEF_FRAGMENT_SIZE 4 3727 #define MIN_FRAGMENT_SIZE 1 3728 #define MAX_FRAGMENT_SIZE 512 3729 3730 static inline bool f2fs_need_rand_seg(struct f2fs_sb_info *sbi) 3731 { 3732 return F2FS_OPTION(sbi).fs_mode == FS_MODE_FRAGMENT_SEG || 3733 F2FS_OPTION(sbi).fs_mode == FS_MODE_FRAGMENT_BLK; 3734 } 3735 3736 /* 3737 * checkpoint.c 3738 */ 3739 void f2fs_stop_checkpoint(struct f2fs_sb_info *sbi, bool end_io, 3740 unsigned char reason); 3741 void f2fs_flush_ckpt_thread(struct f2fs_sb_info *sbi); 3742 struct page *f2fs_grab_meta_page(struct f2fs_sb_info *sbi, pgoff_t index); 3743 struct page *f2fs_get_meta_page(struct f2fs_sb_info *sbi, pgoff_t index); 3744 struct page *f2fs_get_meta_page_retry(struct f2fs_sb_info *sbi, pgoff_t index); 3745 struct page *f2fs_get_tmp_page(struct f2fs_sb_info *sbi, pgoff_t index); 3746 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi, 3747 block_t blkaddr, int type); 3748 int f2fs_ra_meta_pages(struct f2fs_sb_info *sbi, block_t start, int nrpages, 3749 int type, bool sync); 3750 void f2fs_ra_meta_pages_cond(struct f2fs_sb_info *sbi, pgoff_t index, 3751 unsigned int ra_blocks); 3752 long f2fs_sync_meta_pages(struct f2fs_sb_info *sbi, enum page_type type, 3753 long nr_to_write, enum iostat_type io_type); 3754 void f2fs_add_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type); 3755 void f2fs_remove_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type); 3756 void f2fs_release_ino_entry(struct f2fs_sb_info *sbi, bool all); 3757 bool f2fs_exist_written_data(struct f2fs_sb_info *sbi, nid_t ino, int mode); 3758 void f2fs_set_dirty_device(struct f2fs_sb_info *sbi, nid_t ino, 3759 unsigned int devidx, int type); 3760 bool f2fs_is_dirty_device(struct f2fs_sb_info *sbi, nid_t ino, 3761 unsigned int devidx, int type); 3762 int f2fs_sync_inode_meta(struct f2fs_sb_info *sbi); 3763 int f2fs_acquire_orphan_inode(struct f2fs_sb_info *sbi); 3764 void f2fs_release_orphan_inode(struct f2fs_sb_info *sbi); 3765 void f2fs_add_orphan_inode(struct inode *inode); 3766 void f2fs_remove_orphan_inode(struct f2fs_sb_info *sbi, nid_t ino); 3767 int f2fs_recover_orphan_inodes(struct f2fs_sb_info *sbi); 3768 int f2fs_get_valid_checkpoint(struct f2fs_sb_info *sbi); 3769 void f2fs_update_dirty_folio(struct inode *inode, struct folio *folio); 3770 void f2fs_remove_dirty_inode(struct inode *inode); 3771 int f2fs_sync_dirty_inodes(struct f2fs_sb_info *sbi, enum inode_type type, 3772 bool from_cp); 3773 void f2fs_wait_on_all_pages(struct f2fs_sb_info *sbi, int type); 3774 u64 f2fs_get_sectors_written(struct f2fs_sb_info *sbi); 3775 int f2fs_write_checkpoint(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3776 void f2fs_init_ino_entry_info(struct f2fs_sb_info *sbi); 3777 int __init f2fs_create_checkpoint_caches(void); 3778 void f2fs_destroy_checkpoint_caches(void); 3779 int f2fs_issue_checkpoint(struct f2fs_sb_info *sbi); 3780 int f2fs_start_ckpt_thread(struct f2fs_sb_info *sbi); 3781 void f2fs_stop_ckpt_thread(struct f2fs_sb_info *sbi); 3782 void f2fs_init_ckpt_req_control(struct f2fs_sb_info *sbi); 3783 3784 /* 3785 * data.c 3786 */ 3787 int __init f2fs_init_bioset(void); 3788 void f2fs_destroy_bioset(void); 3789 int f2fs_init_bio_entry_cache(void); 3790 void f2fs_destroy_bio_entry_cache(void); 3791 void f2fs_submit_bio(struct f2fs_sb_info *sbi, 3792 struct bio *bio, enum page_type type); 3793 int f2fs_init_write_merge_io(struct f2fs_sb_info *sbi); 3794 void f2fs_submit_merged_write(struct f2fs_sb_info *sbi, enum page_type type); 3795 void f2fs_submit_merged_write_cond(struct f2fs_sb_info *sbi, 3796 struct inode *inode, struct page *page, 3797 nid_t ino, enum page_type type); 3798 void f2fs_submit_merged_ipu_write(struct f2fs_sb_info *sbi, 3799 struct bio **bio, struct page *page); 3800 void f2fs_flush_merged_writes(struct f2fs_sb_info *sbi); 3801 int f2fs_submit_page_bio(struct f2fs_io_info *fio); 3802 int f2fs_merge_page_bio(struct f2fs_io_info *fio); 3803 void f2fs_submit_page_write(struct f2fs_io_info *fio); 3804 struct block_device *f2fs_target_device(struct f2fs_sb_info *sbi, 3805 block_t blk_addr, sector_t *sector); 3806 int f2fs_target_device_index(struct f2fs_sb_info *sbi, block_t blkaddr); 3807 void f2fs_set_data_blkaddr(struct dnode_of_data *dn); 3808 void f2fs_update_data_blkaddr(struct dnode_of_data *dn, block_t blkaddr); 3809 int f2fs_reserve_new_blocks(struct dnode_of_data *dn, blkcnt_t count); 3810 int f2fs_reserve_new_block(struct dnode_of_data *dn); 3811 int f2fs_get_block(struct dnode_of_data *dn, pgoff_t index); 3812 int f2fs_reserve_block(struct dnode_of_data *dn, pgoff_t index); 3813 struct page *f2fs_get_read_data_page(struct inode *inode, pgoff_t index, 3814 blk_opf_t op_flags, bool for_write, pgoff_t *next_pgofs); 3815 struct page *f2fs_find_data_page(struct inode *inode, pgoff_t index, 3816 pgoff_t *next_pgofs); 3817 struct page *f2fs_get_lock_data_page(struct inode *inode, pgoff_t index, 3818 bool for_write); 3819 struct page *f2fs_get_new_data_page(struct inode *inode, 3820 struct page *ipage, pgoff_t index, bool new_i_size); 3821 int f2fs_do_write_data_page(struct f2fs_io_info *fio); 3822 void f2fs_do_map_lock(struct f2fs_sb_info *sbi, int flag, bool lock); 3823 int f2fs_map_blocks(struct inode *inode, struct f2fs_map_blocks *map, 3824 int create, int flag); 3825 int f2fs_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, 3826 u64 start, u64 len); 3827 int f2fs_encrypt_one_page(struct f2fs_io_info *fio); 3828 bool f2fs_should_update_inplace(struct inode *inode, struct f2fs_io_info *fio); 3829 bool f2fs_should_update_outplace(struct inode *inode, struct f2fs_io_info *fio); 3830 int f2fs_write_single_data_page(struct page *page, int *submitted, 3831 struct bio **bio, sector_t *last_block, 3832 struct writeback_control *wbc, 3833 enum iostat_type io_type, 3834 int compr_blocks, bool allow_balance); 3835 void f2fs_write_failed(struct inode *inode, loff_t to); 3836 void f2fs_invalidate_folio(struct folio *folio, size_t offset, size_t length); 3837 bool f2fs_release_folio(struct folio *folio, gfp_t wait); 3838 bool f2fs_overwrite_io(struct inode *inode, loff_t pos, size_t len); 3839 void f2fs_clear_page_cache_dirty_tag(struct page *page); 3840 int f2fs_init_post_read_processing(void); 3841 void f2fs_destroy_post_read_processing(void); 3842 int f2fs_init_post_read_wq(struct f2fs_sb_info *sbi); 3843 void f2fs_destroy_post_read_wq(struct f2fs_sb_info *sbi); 3844 extern const struct iomap_ops f2fs_iomap_ops; 3845 3846 /* 3847 * gc.c 3848 */ 3849 int f2fs_start_gc_thread(struct f2fs_sb_info *sbi); 3850 void f2fs_stop_gc_thread(struct f2fs_sb_info *sbi); 3851 block_t f2fs_start_bidx_of_node(unsigned int node_ofs, struct inode *inode); 3852 int f2fs_gc(struct f2fs_sb_info *sbi, struct f2fs_gc_control *gc_control); 3853 void f2fs_build_gc_manager(struct f2fs_sb_info *sbi); 3854 int f2fs_resize_fs(struct f2fs_sb_info *sbi, __u64 block_count); 3855 int __init f2fs_create_garbage_collection_cache(void); 3856 void f2fs_destroy_garbage_collection_cache(void); 3857 3858 /* 3859 * recovery.c 3860 */ 3861 int f2fs_recover_fsync_data(struct f2fs_sb_info *sbi, bool check_only); 3862 bool f2fs_space_for_roll_forward(struct f2fs_sb_info *sbi); 3863 int __init f2fs_create_recovery_cache(void); 3864 void f2fs_destroy_recovery_cache(void); 3865 3866 /* 3867 * debug.c 3868 */ 3869 #ifdef CONFIG_F2FS_STAT_FS 3870 struct f2fs_stat_info { 3871 struct list_head stat_list; 3872 struct f2fs_sb_info *sbi; 3873 int all_area_segs, sit_area_segs, nat_area_segs, ssa_area_segs; 3874 int main_area_segs, main_area_sections, main_area_zones; 3875 unsigned long long hit_cached[NR_EXTENT_CACHES]; 3876 unsigned long long hit_rbtree[NR_EXTENT_CACHES]; 3877 unsigned long long total_ext[NR_EXTENT_CACHES]; 3878 unsigned long long hit_total[NR_EXTENT_CACHES]; 3879 int ext_tree[NR_EXTENT_CACHES]; 3880 int zombie_tree[NR_EXTENT_CACHES]; 3881 int ext_node[NR_EXTENT_CACHES]; 3882 /* to count memory footprint */ 3883 unsigned long long ext_mem[NR_EXTENT_CACHES]; 3884 /* for read extent cache */ 3885 unsigned long long hit_largest; 3886 /* for block age extent cache */ 3887 unsigned long long allocated_data_blocks; 3888 int ndirty_node, ndirty_dent, ndirty_meta, ndirty_imeta; 3889 int ndirty_data, ndirty_qdata; 3890 unsigned int ndirty_dirs, ndirty_files, nquota_files, ndirty_all; 3891 int nats, dirty_nats, sits, dirty_sits; 3892 int free_nids, avail_nids, alloc_nids; 3893 int total_count, utilization; 3894 int bg_gc, nr_wb_cp_data, nr_wb_data; 3895 int nr_rd_data, nr_rd_node, nr_rd_meta; 3896 int nr_dio_read, nr_dio_write; 3897 unsigned int io_skip_bggc, other_skip_bggc; 3898 int nr_flushing, nr_flushed, flush_list_empty; 3899 int nr_discarding, nr_discarded; 3900 int nr_discard_cmd; 3901 unsigned int undiscard_blks; 3902 int nr_issued_ckpt, nr_total_ckpt, nr_queued_ckpt; 3903 unsigned int cur_ckpt_time, peak_ckpt_time; 3904 int inline_xattr, inline_inode, inline_dir, append, update, orphans; 3905 int compr_inode, swapfile_inode; 3906 unsigned long long compr_blocks; 3907 int aw_cnt, max_aw_cnt; 3908 unsigned int valid_count, valid_node_count, valid_inode_count, discard_blks; 3909 unsigned int bimodal, avg_vblocks; 3910 int util_free, util_valid, util_invalid; 3911 int rsvd_segs, overp_segs; 3912 int dirty_count, node_pages, meta_pages, compress_pages; 3913 int compress_page_hit; 3914 int prefree_count, call_count, cp_count, bg_cp_count; 3915 int tot_segs, node_segs, data_segs, free_segs, free_secs; 3916 int bg_node_segs, bg_data_segs; 3917 int tot_blks, data_blks, node_blks; 3918 int bg_data_blks, bg_node_blks; 3919 int curseg[NR_CURSEG_TYPE]; 3920 int cursec[NR_CURSEG_TYPE]; 3921 int curzone[NR_CURSEG_TYPE]; 3922 unsigned int dirty_seg[NR_CURSEG_TYPE]; 3923 unsigned int full_seg[NR_CURSEG_TYPE]; 3924 unsigned int valid_blks[NR_CURSEG_TYPE]; 3925 3926 unsigned int meta_count[META_MAX]; 3927 unsigned int segment_count[2]; 3928 unsigned int block_count[2]; 3929 unsigned int inplace_count; 3930 unsigned long long base_mem, cache_mem, page_mem; 3931 }; 3932 3933 static inline struct f2fs_stat_info *F2FS_STAT(struct f2fs_sb_info *sbi) 3934 { 3935 return (struct f2fs_stat_info *)sbi->stat_info; 3936 } 3937 3938 #define stat_inc_cp_count(si) ((si)->cp_count++) 3939 #define stat_inc_bg_cp_count(si) ((si)->bg_cp_count++) 3940 #define stat_inc_call_count(si) ((si)->call_count++) 3941 #define stat_inc_bggc_count(si) ((si)->bg_gc++) 3942 #define stat_io_skip_bggc_count(sbi) ((sbi)->io_skip_bggc++) 3943 #define stat_other_skip_bggc_count(sbi) ((sbi)->other_skip_bggc++) 3944 #define stat_inc_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]++) 3945 #define stat_dec_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]--) 3946 #define stat_inc_total_hit(sbi, type) (atomic64_inc(&(sbi)->total_hit_ext[type])) 3947 #define stat_inc_rbtree_node_hit(sbi, type) (atomic64_inc(&(sbi)->read_hit_rbtree[type])) 3948 #define stat_inc_largest_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_largest)) 3949 #define stat_inc_cached_node_hit(sbi, type) (atomic64_inc(&(sbi)->read_hit_cached[type])) 3950 #define stat_inc_inline_xattr(inode) \ 3951 do { \ 3952 if (f2fs_has_inline_xattr(inode)) \ 3953 (atomic_inc(&F2FS_I_SB(inode)->inline_xattr)); \ 3954 } while (0) 3955 #define stat_dec_inline_xattr(inode) \ 3956 do { \ 3957 if (f2fs_has_inline_xattr(inode)) \ 3958 (atomic_dec(&F2FS_I_SB(inode)->inline_xattr)); \ 3959 } while (0) 3960 #define stat_inc_inline_inode(inode) \ 3961 do { \ 3962 if (f2fs_has_inline_data(inode)) \ 3963 (atomic_inc(&F2FS_I_SB(inode)->inline_inode)); \ 3964 } while (0) 3965 #define stat_dec_inline_inode(inode) \ 3966 do { \ 3967 if (f2fs_has_inline_data(inode)) \ 3968 (atomic_dec(&F2FS_I_SB(inode)->inline_inode)); \ 3969 } while (0) 3970 #define stat_inc_inline_dir(inode) \ 3971 do { \ 3972 if (f2fs_has_inline_dentry(inode)) \ 3973 (atomic_inc(&F2FS_I_SB(inode)->inline_dir)); \ 3974 } while (0) 3975 #define stat_dec_inline_dir(inode) \ 3976 do { \ 3977 if (f2fs_has_inline_dentry(inode)) \ 3978 (atomic_dec(&F2FS_I_SB(inode)->inline_dir)); \ 3979 } while (0) 3980 #define stat_inc_compr_inode(inode) \ 3981 do { \ 3982 if (f2fs_compressed_file(inode)) \ 3983 (atomic_inc(&F2FS_I_SB(inode)->compr_inode)); \ 3984 } while (0) 3985 #define stat_dec_compr_inode(inode) \ 3986 do { \ 3987 if (f2fs_compressed_file(inode)) \ 3988 (atomic_dec(&F2FS_I_SB(inode)->compr_inode)); \ 3989 } while (0) 3990 #define stat_add_compr_blocks(inode, blocks) \ 3991 (atomic64_add(blocks, &F2FS_I_SB(inode)->compr_blocks)) 3992 #define stat_sub_compr_blocks(inode, blocks) \ 3993 (atomic64_sub(blocks, &F2FS_I_SB(inode)->compr_blocks)) 3994 #define stat_inc_swapfile_inode(inode) \ 3995 (atomic_inc(&F2FS_I_SB(inode)->swapfile_inode)) 3996 #define stat_dec_swapfile_inode(inode) \ 3997 (atomic_dec(&F2FS_I_SB(inode)->swapfile_inode)) 3998 #define stat_inc_atomic_inode(inode) \ 3999 (atomic_inc(&F2FS_I_SB(inode)->atomic_files)) 4000 #define stat_dec_atomic_inode(inode) \ 4001 (atomic_dec(&F2FS_I_SB(inode)->atomic_files)) 4002 #define stat_inc_meta_count(sbi, blkaddr) \ 4003 do { \ 4004 if (blkaddr < SIT_I(sbi)->sit_base_addr) \ 4005 atomic_inc(&(sbi)->meta_count[META_CP]); \ 4006 else if (blkaddr < NM_I(sbi)->nat_blkaddr) \ 4007 atomic_inc(&(sbi)->meta_count[META_SIT]); \ 4008 else if (blkaddr < SM_I(sbi)->ssa_blkaddr) \ 4009 atomic_inc(&(sbi)->meta_count[META_NAT]); \ 4010 else if (blkaddr < SM_I(sbi)->main_blkaddr) \ 4011 atomic_inc(&(sbi)->meta_count[META_SSA]); \ 4012 } while (0) 4013 #define stat_inc_seg_type(sbi, curseg) \ 4014 ((sbi)->segment_count[(curseg)->alloc_type]++) 4015 #define stat_inc_block_count(sbi, curseg) \ 4016 ((sbi)->block_count[(curseg)->alloc_type]++) 4017 #define stat_inc_inplace_blocks(sbi) \ 4018 (atomic_inc(&(sbi)->inplace_count)) 4019 #define stat_update_max_atomic_write(inode) \ 4020 do { \ 4021 int cur = atomic_read(&F2FS_I_SB(inode)->atomic_files); \ 4022 int max = atomic_read(&F2FS_I_SB(inode)->max_aw_cnt); \ 4023 if (cur > max) \ 4024 atomic_set(&F2FS_I_SB(inode)->max_aw_cnt, cur); \ 4025 } while (0) 4026 #define stat_inc_seg_count(sbi, type, gc_type) \ 4027 do { \ 4028 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 4029 si->tot_segs++; \ 4030 if ((type) == SUM_TYPE_DATA) { \ 4031 si->data_segs++; \ 4032 si->bg_data_segs += (gc_type == BG_GC) ? 1 : 0; \ 4033 } else { \ 4034 si->node_segs++; \ 4035 si->bg_node_segs += (gc_type == BG_GC) ? 1 : 0; \ 4036 } \ 4037 } while (0) 4038 4039 #define stat_inc_tot_blk_count(si, blks) \ 4040 ((si)->tot_blks += (blks)) 4041 4042 #define stat_inc_data_blk_count(sbi, blks, gc_type) \ 4043 do { \ 4044 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 4045 stat_inc_tot_blk_count(si, blks); \ 4046 si->data_blks += (blks); \ 4047 si->bg_data_blks += ((gc_type) == BG_GC) ? (blks) : 0; \ 4048 } while (0) 4049 4050 #define stat_inc_node_blk_count(sbi, blks, gc_type) \ 4051 do { \ 4052 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 4053 stat_inc_tot_blk_count(si, blks); \ 4054 si->node_blks += (blks); \ 4055 si->bg_node_blks += ((gc_type) == BG_GC) ? (blks) : 0; \ 4056 } while (0) 4057 4058 int f2fs_build_stats(struct f2fs_sb_info *sbi); 4059 void f2fs_destroy_stats(struct f2fs_sb_info *sbi); 4060 void __init f2fs_create_root_stats(void); 4061 void f2fs_destroy_root_stats(void); 4062 void f2fs_update_sit_info(struct f2fs_sb_info *sbi); 4063 #else 4064 #define stat_inc_cp_count(si) do { } while (0) 4065 #define stat_inc_bg_cp_count(si) do { } while (0) 4066 #define stat_inc_call_count(si) do { } while (0) 4067 #define stat_inc_bggc_count(si) do { } while (0) 4068 #define stat_io_skip_bggc_count(sbi) do { } while (0) 4069 #define stat_other_skip_bggc_count(sbi) do { } while (0) 4070 #define stat_inc_dirty_inode(sbi, type) do { } while (0) 4071 #define stat_dec_dirty_inode(sbi, type) do { } while (0) 4072 #define stat_inc_total_hit(sbi, type) do { } while (0) 4073 #define stat_inc_rbtree_node_hit(sbi, type) do { } while (0) 4074 #define stat_inc_largest_node_hit(sbi) do { } while (0) 4075 #define stat_inc_cached_node_hit(sbi, type) do { } while (0) 4076 #define stat_inc_inline_xattr(inode) do { } while (0) 4077 #define stat_dec_inline_xattr(inode) do { } while (0) 4078 #define stat_inc_inline_inode(inode) do { } while (0) 4079 #define stat_dec_inline_inode(inode) do { } while (0) 4080 #define stat_inc_inline_dir(inode) do { } while (0) 4081 #define stat_dec_inline_dir(inode) do { } while (0) 4082 #define stat_inc_compr_inode(inode) do { } while (0) 4083 #define stat_dec_compr_inode(inode) do { } while (0) 4084 #define stat_add_compr_blocks(inode, blocks) do { } while (0) 4085 #define stat_sub_compr_blocks(inode, blocks) do { } while (0) 4086 #define stat_inc_swapfile_inode(inode) do { } while (0) 4087 #define stat_dec_swapfile_inode(inode) do { } while (0) 4088 #define stat_inc_atomic_inode(inode) do { } while (0) 4089 #define stat_dec_atomic_inode(inode) do { } while (0) 4090 #define stat_update_max_atomic_write(inode) do { } while (0) 4091 #define stat_inc_meta_count(sbi, blkaddr) do { } while (0) 4092 #define stat_inc_seg_type(sbi, curseg) do { } while (0) 4093 #define stat_inc_block_count(sbi, curseg) do { } while (0) 4094 #define stat_inc_inplace_blocks(sbi) do { } while (0) 4095 #define stat_inc_seg_count(sbi, type, gc_type) do { } while (0) 4096 #define stat_inc_tot_blk_count(si, blks) do { } while (0) 4097 #define stat_inc_data_blk_count(sbi, blks, gc_type) do { } while (0) 4098 #define stat_inc_node_blk_count(sbi, blks, gc_type) do { } while (0) 4099 4100 static inline int f2fs_build_stats(struct f2fs_sb_info *sbi) { return 0; } 4101 static inline void f2fs_destroy_stats(struct f2fs_sb_info *sbi) { } 4102 static inline void __init f2fs_create_root_stats(void) { } 4103 static inline void f2fs_destroy_root_stats(void) { } 4104 static inline void f2fs_update_sit_info(struct f2fs_sb_info *sbi) {} 4105 #endif 4106 4107 extern const struct file_operations f2fs_dir_operations; 4108 extern const struct file_operations f2fs_file_operations; 4109 extern const struct inode_operations f2fs_file_inode_operations; 4110 extern const struct address_space_operations f2fs_dblock_aops; 4111 extern const struct address_space_operations f2fs_node_aops; 4112 extern const struct address_space_operations f2fs_meta_aops; 4113 extern const struct inode_operations f2fs_dir_inode_operations; 4114 extern const struct inode_operations f2fs_symlink_inode_operations; 4115 extern const struct inode_operations f2fs_encrypted_symlink_inode_operations; 4116 extern const struct inode_operations f2fs_special_inode_operations; 4117 extern struct kmem_cache *f2fs_inode_entry_slab; 4118 4119 /* 4120 * inline.c 4121 */ 4122 bool f2fs_may_inline_data(struct inode *inode); 4123 bool f2fs_sanity_check_inline_data(struct inode *inode); 4124 bool f2fs_may_inline_dentry(struct inode *inode); 4125 void f2fs_do_read_inline_data(struct page *page, struct page *ipage); 4126 void f2fs_truncate_inline_inode(struct inode *inode, 4127 struct page *ipage, u64 from); 4128 int f2fs_read_inline_data(struct inode *inode, struct page *page); 4129 int f2fs_convert_inline_page(struct dnode_of_data *dn, struct page *page); 4130 int f2fs_convert_inline_inode(struct inode *inode); 4131 int f2fs_try_convert_inline_dir(struct inode *dir, struct dentry *dentry); 4132 int f2fs_write_inline_data(struct inode *inode, struct page *page); 4133 int f2fs_recover_inline_data(struct inode *inode, struct page *npage); 4134 struct f2fs_dir_entry *f2fs_find_in_inline_dir(struct inode *dir, 4135 const struct f2fs_filename *fname, 4136 struct page **res_page); 4137 int f2fs_make_empty_inline_dir(struct inode *inode, struct inode *parent, 4138 struct page *ipage); 4139 int f2fs_add_inline_entry(struct inode *dir, const struct f2fs_filename *fname, 4140 struct inode *inode, nid_t ino, umode_t mode); 4141 void f2fs_delete_inline_entry(struct f2fs_dir_entry *dentry, 4142 struct page *page, struct inode *dir, 4143 struct inode *inode); 4144 bool f2fs_empty_inline_dir(struct inode *dir); 4145 int f2fs_read_inline_dir(struct file *file, struct dir_context *ctx, 4146 struct fscrypt_str *fstr); 4147 int f2fs_inline_data_fiemap(struct inode *inode, 4148 struct fiemap_extent_info *fieinfo, 4149 __u64 start, __u64 len); 4150 4151 /* 4152 * shrinker.c 4153 */ 4154 unsigned long f2fs_shrink_count(struct shrinker *shrink, 4155 struct shrink_control *sc); 4156 unsigned long f2fs_shrink_scan(struct shrinker *shrink, 4157 struct shrink_control *sc); 4158 void f2fs_join_shrinker(struct f2fs_sb_info *sbi); 4159 void f2fs_leave_shrinker(struct f2fs_sb_info *sbi); 4160 4161 /* 4162 * extent_cache.c 4163 */ 4164 struct rb_entry *f2fs_lookup_rb_tree(struct rb_root_cached *root, 4165 struct rb_entry *cached_re, unsigned int ofs); 4166 struct rb_node **f2fs_lookup_rb_tree_ext(struct f2fs_sb_info *sbi, 4167 struct rb_root_cached *root, 4168 struct rb_node **parent, 4169 unsigned long long key, bool *left_most); 4170 struct rb_node **f2fs_lookup_rb_tree_for_insert(struct f2fs_sb_info *sbi, 4171 struct rb_root_cached *root, 4172 struct rb_node **parent, 4173 unsigned int ofs, bool *leftmost); 4174 struct rb_entry *f2fs_lookup_rb_tree_ret(struct rb_root_cached *root, 4175 struct rb_entry *cached_re, unsigned int ofs, 4176 struct rb_entry **prev_entry, struct rb_entry **next_entry, 4177 struct rb_node ***insert_p, struct rb_node **insert_parent, 4178 bool force, bool *leftmost); 4179 bool f2fs_check_rb_tree_consistence(struct f2fs_sb_info *sbi, 4180 struct rb_root_cached *root, bool check_key); 4181 void f2fs_init_extent_tree(struct inode *inode); 4182 void f2fs_drop_extent_tree(struct inode *inode); 4183 void f2fs_destroy_extent_node(struct inode *inode); 4184 void f2fs_destroy_extent_tree(struct inode *inode); 4185 void f2fs_init_extent_cache_info(struct f2fs_sb_info *sbi); 4186 int __init f2fs_create_extent_cache(void); 4187 void f2fs_destroy_extent_cache(void); 4188 4189 /* read extent cache ops */ 4190 void f2fs_init_read_extent_tree(struct inode *inode, struct page *ipage); 4191 bool f2fs_lookup_read_extent_cache(struct inode *inode, pgoff_t pgofs, 4192 struct extent_info *ei); 4193 void f2fs_update_read_extent_cache(struct dnode_of_data *dn); 4194 void f2fs_update_read_extent_cache_range(struct dnode_of_data *dn, 4195 pgoff_t fofs, block_t blkaddr, unsigned int len); 4196 unsigned int f2fs_shrink_read_extent_tree(struct f2fs_sb_info *sbi, 4197 int nr_shrink); 4198 4199 /* block age extent cache ops */ 4200 void f2fs_init_age_extent_tree(struct inode *inode); 4201 bool f2fs_lookup_age_extent_cache(struct inode *inode, pgoff_t pgofs, 4202 struct extent_info *ei); 4203 void f2fs_update_age_extent_cache(struct dnode_of_data *dn); 4204 void f2fs_update_age_extent_cache_range(struct dnode_of_data *dn, 4205 pgoff_t fofs, unsigned int len); 4206 unsigned int f2fs_shrink_age_extent_tree(struct f2fs_sb_info *sbi, 4207 int nr_shrink); 4208 4209 /* 4210 * sysfs.c 4211 */ 4212 #define MIN_RA_MUL 2 4213 #define MAX_RA_MUL 256 4214 4215 int __init f2fs_init_sysfs(void); 4216 void f2fs_exit_sysfs(void); 4217 int f2fs_register_sysfs(struct f2fs_sb_info *sbi); 4218 void f2fs_unregister_sysfs(struct f2fs_sb_info *sbi); 4219 4220 /* verity.c */ 4221 extern const struct fsverity_operations f2fs_verityops; 4222 4223 /* 4224 * crypto support 4225 */ 4226 static inline bool f2fs_encrypted_file(struct inode *inode) 4227 { 4228 return IS_ENCRYPTED(inode) && S_ISREG(inode->i_mode); 4229 } 4230 4231 static inline void f2fs_set_encrypted_inode(struct inode *inode) 4232 { 4233 #ifdef CONFIG_FS_ENCRYPTION 4234 file_set_encrypt(inode); 4235 f2fs_set_inode_flags(inode); 4236 #endif 4237 } 4238 4239 /* 4240 * Returns true if the reads of the inode's data need to undergo some 4241 * postprocessing step, like decryption or authenticity verification. 4242 */ 4243 static inline bool f2fs_post_read_required(struct inode *inode) 4244 { 4245 return f2fs_encrypted_file(inode) || fsverity_active(inode) || 4246 f2fs_compressed_file(inode); 4247 } 4248 4249 /* 4250 * compress.c 4251 */ 4252 #ifdef CONFIG_F2FS_FS_COMPRESSION 4253 bool f2fs_is_compressed_page(struct page *page); 4254 struct page *f2fs_compress_control_page(struct page *page); 4255 int f2fs_prepare_compress_overwrite(struct inode *inode, 4256 struct page **pagep, pgoff_t index, void **fsdata); 4257 bool f2fs_compress_write_end(struct inode *inode, void *fsdata, 4258 pgoff_t index, unsigned copied); 4259 int f2fs_truncate_partial_cluster(struct inode *inode, u64 from, bool lock); 4260 void f2fs_compress_write_end_io(struct bio *bio, struct page *page); 4261 bool f2fs_is_compress_backend_ready(struct inode *inode); 4262 int f2fs_init_compress_mempool(void); 4263 void f2fs_destroy_compress_mempool(void); 4264 void f2fs_decompress_cluster(struct decompress_io_ctx *dic, bool in_task); 4265 void f2fs_end_read_compressed_page(struct page *page, bool failed, 4266 block_t blkaddr, bool in_task); 4267 bool f2fs_cluster_is_empty(struct compress_ctx *cc); 4268 bool f2fs_cluster_can_merge_page(struct compress_ctx *cc, pgoff_t index); 4269 bool f2fs_all_cluster_page_ready(struct compress_ctx *cc, struct page **pages, 4270 int index, int nr_pages, bool uptodate); 4271 bool f2fs_sanity_check_cluster(struct dnode_of_data *dn); 4272 void f2fs_compress_ctx_add_page(struct compress_ctx *cc, struct page *page); 4273 int f2fs_write_multi_pages(struct compress_ctx *cc, 4274 int *submitted, 4275 struct writeback_control *wbc, 4276 enum iostat_type io_type); 4277 int f2fs_is_compressed_cluster(struct inode *inode, pgoff_t index); 4278 void f2fs_update_read_extent_tree_range_compressed(struct inode *inode, 4279 pgoff_t fofs, block_t blkaddr, 4280 unsigned int llen, unsigned int c_len); 4281 int f2fs_read_multi_pages(struct compress_ctx *cc, struct bio **bio_ret, 4282 unsigned nr_pages, sector_t *last_block_in_bio, 4283 bool is_readahead, bool for_write); 4284 struct decompress_io_ctx *f2fs_alloc_dic(struct compress_ctx *cc); 4285 void f2fs_decompress_end_io(struct decompress_io_ctx *dic, bool failed, 4286 bool in_task); 4287 void f2fs_put_page_dic(struct page *page, bool in_task); 4288 unsigned int f2fs_cluster_blocks_are_contiguous(struct dnode_of_data *dn); 4289 int f2fs_init_compress_ctx(struct compress_ctx *cc); 4290 void f2fs_destroy_compress_ctx(struct compress_ctx *cc, bool reuse); 4291 void f2fs_init_compress_info(struct f2fs_sb_info *sbi); 4292 int f2fs_init_compress_inode(struct f2fs_sb_info *sbi); 4293 void f2fs_destroy_compress_inode(struct f2fs_sb_info *sbi); 4294 int f2fs_init_page_array_cache(struct f2fs_sb_info *sbi); 4295 void f2fs_destroy_page_array_cache(struct f2fs_sb_info *sbi); 4296 int __init f2fs_init_compress_cache(void); 4297 void f2fs_destroy_compress_cache(void); 4298 struct address_space *COMPRESS_MAPPING(struct f2fs_sb_info *sbi); 4299 void f2fs_invalidate_compress_page(struct f2fs_sb_info *sbi, block_t blkaddr); 4300 void f2fs_cache_compressed_page(struct f2fs_sb_info *sbi, struct page *page, 4301 nid_t ino, block_t blkaddr); 4302 bool f2fs_load_compressed_page(struct f2fs_sb_info *sbi, struct page *page, 4303 block_t blkaddr); 4304 void f2fs_invalidate_compress_pages(struct f2fs_sb_info *sbi, nid_t ino); 4305 #define inc_compr_inode_stat(inode) \ 4306 do { \ 4307 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); \ 4308 sbi->compr_new_inode++; \ 4309 } while (0) 4310 #define add_compr_block_stat(inode, blocks) \ 4311 do { \ 4312 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); \ 4313 int diff = F2FS_I(inode)->i_cluster_size - blocks; \ 4314 sbi->compr_written_block += blocks; \ 4315 sbi->compr_saved_block += diff; \ 4316 } while (0) 4317 #else 4318 static inline bool f2fs_is_compressed_page(struct page *page) { return false; } 4319 static inline bool f2fs_is_compress_backend_ready(struct inode *inode) 4320 { 4321 if (!f2fs_compressed_file(inode)) 4322 return true; 4323 /* not support compression */ 4324 return false; 4325 } 4326 static inline struct page *f2fs_compress_control_page(struct page *page) 4327 { 4328 WARN_ON_ONCE(1); 4329 return ERR_PTR(-EINVAL); 4330 } 4331 static inline int f2fs_init_compress_mempool(void) { return 0; } 4332 static inline void f2fs_destroy_compress_mempool(void) { } 4333 static inline void f2fs_decompress_cluster(struct decompress_io_ctx *dic, 4334 bool in_task) { } 4335 static inline void f2fs_end_read_compressed_page(struct page *page, 4336 bool failed, block_t blkaddr, bool in_task) 4337 { 4338 WARN_ON_ONCE(1); 4339 } 4340 static inline void f2fs_put_page_dic(struct page *page, bool in_task) 4341 { 4342 WARN_ON_ONCE(1); 4343 } 4344 static inline unsigned int f2fs_cluster_blocks_are_contiguous(struct dnode_of_data *dn) { return 0; } 4345 static inline bool f2fs_sanity_check_cluster(struct dnode_of_data *dn) { return false; } 4346 static inline int f2fs_init_compress_inode(struct f2fs_sb_info *sbi) { return 0; } 4347 static inline void f2fs_destroy_compress_inode(struct f2fs_sb_info *sbi) { } 4348 static inline int f2fs_init_page_array_cache(struct f2fs_sb_info *sbi) { return 0; } 4349 static inline void f2fs_destroy_page_array_cache(struct f2fs_sb_info *sbi) { } 4350 static inline int __init f2fs_init_compress_cache(void) { return 0; } 4351 static inline void f2fs_destroy_compress_cache(void) { } 4352 static inline void f2fs_invalidate_compress_page(struct f2fs_sb_info *sbi, 4353 block_t blkaddr) { } 4354 static inline void f2fs_cache_compressed_page(struct f2fs_sb_info *sbi, 4355 struct page *page, nid_t ino, block_t blkaddr) { } 4356 static inline bool f2fs_load_compressed_page(struct f2fs_sb_info *sbi, 4357 struct page *page, block_t blkaddr) { return false; } 4358 static inline void f2fs_invalidate_compress_pages(struct f2fs_sb_info *sbi, 4359 nid_t ino) { } 4360 #define inc_compr_inode_stat(inode) do { } while (0) 4361 static inline void f2fs_update_read_extent_tree_range_compressed( 4362 struct inode *inode, 4363 pgoff_t fofs, block_t blkaddr, 4364 unsigned int llen, unsigned int c_len) { } 4365 #endif 4366 4367 static inline int set_compress_context(struct inode *inode) 4368 { 4369 #ifdef CONFIG_F2FS_FS_COMPRESSION 4370 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 4371 4372 F2FS_I(inode)->i_compress_algorithm = 4373 F2FS_OPTION(sbi).compress_algorithm; 4374 F2FS_I(inode)->i_log_cluster_size = 4375 F2FS_OPTION(sbi).compress_log_size; 4376 F2FS_I(inode)->i_compress_flag = 4377 F2FS_OPTION(sbi).compress_chksum ? 4378 1 << COMPRESS_CHKSUM : 0; 4379 F2FS_I(inode)->i_cluster_size = 4380 1 << F2FS_I(inode)->i_log_cluster_size; 4381 if ((F2FS_I(inode)->i_compress_algorithm == COMPRESS_LZ4 || 4382 F2FS_I(inode)->i_compress_algorithm == COMPRESS_ZSTD) && 4383 F2FS_OPTION(sbi).compress_level) 4384 F2FS_I(inode)->i_compress_flag |= 4385 F2FS_OPTION(sbi).compress_level << 4386 COMPRESS_LEVEL_OFFSET; 4387 F2FS_I(inode)->i_flags |= F2FS_COMPR_FL; 4388 set_inode_flag(inode, FI_COMPRESSED_FILE); 4389 stat_inc_compr_inode(inode); 4390 inc_compr_inode_stat(inode); 4391 f2fs_mark_inode_dirty_sync(inode, true); 4392 return 0; 4393 #else 4394 return -EOPNOTSUPP; 4395 #endif 4396 } 4397 4398 static inline bool f2fs_disable_compressed_file(struct inode *inode) 4399 { 4400 struct f2fs_inode_info *fi = F2FS_I(inode); 4401 4402 if (!f2fs_compressed_file(inode)) 4403 return true; 4404 if (S_ISREG(inode->i_mode) && F2FS_HAS_BLOCKS(inode)) 4405 return false; 4406 4407 fi->i_flags &= ~F2FS_COMPR_FL; 4408 stat_dec_compr_inode(inode); 4409 clear_inode_flag(inode, FI_COMPRESSED_FILE); 4410 f2fs_mark_inode_dirty_sync(inode, true); 4411 return true; 4412 } 4413 4414 #define F2FS_FEATURE_FUNCS(name, flagname) \ 4415 static inline bool f2fs_sb_has_##name(struct f2fs_sb_info *sbi) \ 4416 { \ 4417 return F2FS_HAS_FEATURE(sbi, F2FS_FEATURE_##flagname); \ 4418 } 4419 4420 F2FS_FEATURE_FUNCS(encrypt, ENCRYPT); 4421 F2FS_FEATURE_FUNCS(blkzoned, BLKZONED); 4422 F2FS_FEATURE_FUNCS(extra_attr, EXTRA_ATTR); 4423 F2FS_FEATURE_FUNCS(project_quota, PRJQUOTA); 4424 F2FS_FEATURE_FUNCS(inode_chksum, INODE_CHKSUM); 4425 F2FS_FEATURE_FUNCS(flexible_inline_xattr, FLEXIBLE_INLINE_XATTR); 4426 F2FS_FEATURE_FUNCS(quota_ino, QUOTA_INO); 4427 F2FS_FEATURE_FUNCS(inode_crtime, INODE_CRTIME); 4428 F2FS_FEATURE_FUNCS(lost_found, LOST_FOUND); 4429 F2FS_FEATURE_FUNCS(verity, VERITY); 4430 F2FS_FEATURE_FUNCS(sb_chksum, SB_CHKSUM); 4431 F2FS_FEATURE_FUNCS(casefold, CASEFOLD); 4432 F2FS_FEATURE_FUNCS(compression, COMPRESSION); 4433 F2FS_FEATURE_FUNCS(readonly, RO); 4434 4435 #ifdef CONFIG_BLK_DEV_ZONED 4436 static inline bool f2fs_blkz_is_seq(struct f2fs_sb_info *sbi, int devi, 4437 block_t blkaddr) 4438 { 4439 unsigned int zno = blkaddr >> sbi->log_blocks_per_blkz; 4440 4441 return test_bit(zno, FDEV(devi).blkz_seq); 4442 } 4443 #endif 4444 4445 static inline bool f2fs_hw_should_discard(struct f2fs_sb_info *sbi) 4446 { 4447 return f2fs_sb_has_blkzoned(sbi); 4448 } 4449 4450 static inline bool f2fs_bdev_support_discard(struct block_device *bdev) 4451 { 4452 return bdev_max_discard_sectors(bdev) || bdev_is_zoned(bdev); 4453 } 4454 4455 static inline bool f2fs_hw_support_discard(struct f2fs_sb_info *sbi) 4456 { 4457 int i; 4458 4459 if (!f2fs_is_multi_device(sbi)) 4460 return f2fs_bdev_support_discard(sbi->sb->s_bdev); 4461 4462 for (i = 0; i < sbi->s_ndevs; i++) 4463 if (f2fs_bdev_support_discard(FDEV(i).bdev)) 4464 return true; 4465 return false; 4466 } 4467 4468 static inline bool f2fs_realtime_discard_enable(struct f2fs_sb_info *sbi) 4469 { 4470 return (test_opt(sbi, DISCARD) && f2fs_hw_support_discard(sbi)) || 4471 f2fs_hw_should_discard(sbi); 4472 } 4473 4474 static inline bool f2fs_hw_is_readonly(struct f2fs_sb_info *sbi) 4475 { 4476 int i; 4477 4478 if (!f2fs_is_multi_device(sbi)) 4479 return bdev_read_only(sbi->sb->s_bdev); 4480 4481 for (i = 0; i < sbi->s_ndevs; i++) 4482 if (bdev_read_only(FDEV(i).bdev)) 4483 return true; 4484 return false; 4485 } 4486 4487 static inline bool f2fs_lfs_mode(struct f2fs_sb_info *sbi) 4488 { 4489 return F2FS_OPTION(sbi).fs_mode == FS_MODE_LFS; 4490 } 4491 4492 static inline bool f2fs_low_mem_mode(struct f2fs_sb_info *sbi) 4493 { 4494 return F2FS_OPTION(sbi).memory_mode == MEMORY_MODE_LOW; 4495 } 4496 4497 static inline bool f2fs_may_compress(struct inode *inode) 4498 { 4499 if (IS_SWAPFILE(inode) || f2fs_is_pinned_file(inode) || 4500 f2fs_is_atomic_file(inode) || f2fs_has_inline_data(inode)) 4501 return false; 4502 return S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode); 4503 } 4504 4505 static inline void f2fs_i_compr_blocks_update(struct inode *inode, 4506 u64 blocks, bool add) 4507 { 4508 struct f2fs_inode_info *fi = F2FS_I(inode); 4509 int diff = fi->i_cluster_size - blocks; 4510 4511 /* don't update i_compr_blocks if saved blocks were released */ 4512 if (!add && !atomic_read(&fi->i_compr_blocks)) 4513 return; 4514 4515 if (add) { 4516 atomic_add(diff, &fi->i_compr_blocks); 4517 stat_add_compr_blocks(inode, diff); 4518 } else { 4519 atomic_sub(diff, &fi->i_compr_blocks); 4520 stat_sub_compr_blocks(inode, diff); 4521 } 4522 f2fs_mark_inode_dirty_sync(inode, true); 4523 } 4524 4525 static inline bool f2fs_allow_multi_device_dio(struct f2fs_sb_info *sbi, 4526 int flag) 4527 { 4528 if (!f2fs_is_multi_device(sbi)) 4529 return false; 4530 if (flag != F2FS_GET_BLOCK_DIO) 4531 return false; 4532 return sbi->aligned_blksize; 4533 } 4534 4535 static inline bool f2fs_need_verity(const struct inode *inode, pgoff_t idx) 4536 { 4537 return fsverity_active(inode) && 4538 idx < DIV_ROUND_UP(inode->i_size, PAGE_SIZE); 4539 } 4540 4541 #ifdef CONFIG_F2FS_FAULT_INJECTION 4542 extern void f2fs_build_fault_attr(struct f2fs_sb_info *sbi, unsigned int rate, 4543 unsigned int type); 4544 #else 4545 #define f2fs_build_fault_attr(sbi, rate, type) do { } while (0) 4546 #endif 4547 4548 static inline bool is_journalled_quota(struct f2fs_sb_info *sbi) 4549 { 4550 #ifdef CONFIG_QUOTA 4551 if (f2fs_sb_has_quota_ino(sbi)) 4552 return true; 4553 if (F2FS_OPTION(sbi).s_qf_names[USRQUOTA] || 4554 F2FS_OPTION(sbi).s_qf_names[GRPQUOTA] || 4555 F2FS_OPTION(sbi).s_qf_names[PRJQUOTA]) 4556 return true; 4557 #endif 4558 return false; 4559 } 4560 4561 static inline bool f2fs_block_unit_discard(struct f2fs_sb_info *sbi) 4562 { 4563 return F2FS_OPTION(sbi).discard_unit == DISCARD_UNIT_BLOCK; 4564 } 4565 4566 static inline void f2fs_io_schedule_timeout(long timeout) 4567 { 4568 set_current_state(TASK_UNINTERRUPTIBLE); 4569 io_schedule_timeout(timeout); 4570 } 4571 4572 static inline void f2fs_handle_page_eio(struct f2fs_sb_info *sbi, pgoff_t ofs, 4573 enum page_type type) 4574 { 4575 if (unlikely(f2fs_cp_error(sbi))) 4576 return; 4577 4578 if (ofs == sbi->page_eio_ofs[type]) { 4579 if (sbi->page_eio_cnt[type]++ == MAX_RETRY_PAGE_EIO) 4580 set_ckpt_flags(sbi, CP_ERROR_FLAG); 4581 } else { 4582 sbi->page_eio_ofs[type] = ofs; 4583 sbi->page_eio_cnt[type] = 0; 4584 } 4585 } 4586 4587 static inline bool f2fs_is_readonly(struct f2fs_sb_info *sbi) 4588 { 4589 return f2fs_sb_has_readonly(sbi) || f2fs_readonly(sbi->sb); 4590 } 4591 4592 #define EFSBADCRC EBADMSG /* Bad CRC detected */ 4593 #define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */ 4594 4595 #endif /* _LINUX_F2FS_H */ 4596