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 (GENMASK(FAULT_MAX - 1, 0)) 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 & BIT(type)) 78 79 /* maximum retry count for injected failure */ 80 #define DEFAULT_FAILURE_RETRY_COUNT 8 81 #else 82 #define DEFAULT_FAILURE_RETRY_COUNT 1 83 #endif 84 85 /* 86 * For mount options 87 */ 88 #define F2FS_MOUNT_DISABLE_ROLL_FORWARD 0x00000001 89 #define F2FS_MOUNT_DISCARD 0x00000002 90 #define F2FS_MOUNT_NOHEAP 0x00000004 91 #define F2FS_MOUNT_XATTR_USER 0x00000008 92 #define F2FS_MOUNT_POSIX_ACL 0x00000010 93 #define F2FS_MOUNT_DISABLE_EXT_IDENTIFY 0x00000020 94 #define F2FS_MOUNT_INLINE_XATTR 0x00000040 95 #define F2FS_MOUNT_INLINE_DATA 0x00000080 96 #define F2FS_MOUNT_INLINE_DENTRY 0x00000100 97 #define F2FS_MOUNT_FLUSH_MERGE 0x00000200 98 #define F2FS_MOUNT_NOBARRIER 0x00000400 99 #define F2FS_MOUNT_FASTBOOT 0x00000800 100 #define F2FS_MOUNT_READ_EXTENT_CACHE 0x00001000 101 #define F2FS_MOUNT_DATA_FLUSH 0x00002000 102 #define F2FS_MOUNT_FAULT_INJECTION 0x00004000 103 #define F2FS_MOUNT_USRQUOTA 0x00008000 104 #define F2FS_MOUNT_GRPQUOTA 0x00010000 105 #define F2FS_MOUNT_PRJQUOTA 0x00020000 106 #define F2FS_MOUNT_QUOTA 0x00040000 107 #define F2FS_MOUNT_INLINE_XATTR_SIZE 0x00080000 108 #define F2FS_MOUNT_RESERVE_ROOT 0x00100000 109 #define F2FS_MOUNT_DISABLE_CHECKPOINT 0x00200000 110 #define F2FS_MOUNT_NORECOVERY 0x00400000 111 #define F2FS_MOUNT_ATGC 0x00800000 112 #define F2FS_MOUNT_MERGE_CHECKPOINT 0x01000000 113 #define F2FS_MOUNT_GC_MERGE 0x02000000 114 #define F2FS_MOUNT_COMPRESS_CACHE 0x04000000 115 #define F2FS_MOUNT_AGE_EXTENT_CACHE 0x08000000 116 117 #define F2FS_OPTION(sbi) ((sbi)->mount_opt) 118 #define clear_opt(sbi, option) (F2FS_OPTION(sbi).opt &= ~F2FS_MOUNT_##option) 119 #define set_opt(sbi, option) (F2FS_OPTION(sbi).opt |= F2FS_MOUNT_##option) 120 #define test_opt(sbi, option) (F2FS_OPTION(sbi).opt & F2FS_MOUNT_##option) 121 122 #define ver_after(a, b) (typecheck(unsigned long long, a) && \ 123 typecheck(unsigned long long, b) && \ 124 ((long long)((a) - (b)) > 0)) 125 126 typedef u32 block_t; /* 127 * should not change u32, since it is the on-disk block 128 * address format, __le32. 129 */ 130 typedef u32 nid_t; 131 132 #define COMPRESS_EXT_NUM 16 133 134 /* 135 * An implementation of an rwsem that is explicitly unfair to readers. This 136 * prevents priority inversion when a low-priority reader acquires the read lock 137 * while sleeping on the write lock but the write lock is needed by 138 * higher-priority clients. 139 */ 140 141 struct f2fs_rwsem { 142 struct rw_semaphore internal_rwsem; 143 #ifdef CONFIG_F2FS_UNFAIR_RWSEM 144 wait_queue_head_t read_waiters; 145 #endif 146 }; 147 148 struct f2fs_mount_info { 149 unsigned int opt; 150 block_t root_reserved_blocks; /* root reserved blocks */ 151 kuid_t s_resuid; /* reserved blocks for uid */ 152 kgid_t s_resgid; /* reserved blocks for gid */ 153 int active_logs; /* # of active logs */ 154 int inline_xattr_size; /* inline xattr size */ 155 #ifdef CONFIG_F2FS_FAULT_INJECTION 156 struct f2fs_fault_info fault_info; /* For fault injection */ 157 #endif 158 #ifdef CONFIG_QUOTA 159 /* Names of quota files with journalled quota */ 160 char *s_qf_names[MAXQUOTAS]; 161 int s_jquota_fmt; /* Format of quota to use */ 162 #endif 163 /* For which write hints are passed down to block layer */ 164 int alloc_mode; /* segment allocation policy */ 165 int fsync_mode; /* fsync policy */ 166 int fs_mode; /* fs mode: LFS or ADAPTIVE */ 167 int bggc_mode; /* bggc mode: off, on or sync */ 168 int memory_mode; /* memory mode */ 169 int errors; /* errors parameter */ 170 int discard_unit; /* 171 * discard command's offset/size should 172 * be aligned to this unit: block, 173 * segment or section 174 */ 175 struct fscrypt_dummy_policy dummy_enc_policy; /* test dummy encryption */ 176 block_t unusable_cap_perc; /* percentage for cap */ 177 block_t unusable_cap; /* Amount of space allowed to be 178 * unusable when disabling checkpoint 179 */ 180 181 /* For compression */ 182 unsigned char compress_algorithm; /* algorithm type */ 183 unsigned char compress_log_size; /* cluster log size */ 184 unsigned char compress_level; /* compress level */ 185 bool compress_chksum; /* compressed data chksum */ 186 unsigned char compress_ext_cnt; /* extension count */ 187 unsigned char nocompress_ext_cnt; /* nocompress extension count */ 188 int compress_mode; /* compression mode */ 189 unsigned char extensions[COMPRESS_EXT_NUM][F2FS_EXTENSION_LEN]; /* extensions */ 190 unsigned char noextensions[COMPRESS_EXT_NUM][F2FS_EXTENSION_LEN]; /* extensions */ 191 }; 192 193 #define F2FS_FEATURE_ENCRYPT 0x00000001 194 #define F2FS_FEATURE_BLKZONED 0x00000002 195 #define F2FS_FEATURE_ATOMIC_WRITE 0x00000004 196 #define F2FS_FEATURE_EXTRA_ATTR 0x00000008 197 #define F2FS_FEATURE_PRJQUOTA 0x00000010 198 #define F2FS_FEATURE_INODE_CHKSUM 0x00000020 199 #define F2FS_FEATURE_FLEXIBLE_INLINE_XATTR 0x00000040 200 #define F2FS_FEATURE_QUOTA_INO 0x00000080 201 #define F2FS_FEATURE_INODE_CRTIME 0x00000100 202 #define F2FS_FEATURE_LOST_FOUND 0x00000200 203 #define F2FS_FEATURE_VERITY 0x00000400 204 #define F2FS_FEATURE_SB_CHKSUM 0x00000800 205 #define F2FS_FEATURE_CASEFOLD 0x00001000 206 #define F2FS_FEATURE_COMPRESSION 0x00002000 207 #define F2FS_FEATURE_RO 0x00004000 208 209 #define __F2FS_HAS_FEATURE(raw_super, mask) \ 210 ((raw_super->feature & cpu_to_le32(mask)) != 0) 211 #define F2FS_HAS_FEATURE(sbi, mask) __F2FS_HAS_FEATURE(sbi->raw_super, mask) 212 213 /* 214 * Default values for user and/or group using reserved blocks 215 */ 216 #define F2FS_DEF_RESUID 0 217 #define F2FS_DEF_RESGID 0 218 219 /* 220 * For checkpoint manager 221 */ 222 enum { 223 NAT_BITMAP, 224 SIT_BITMAP 225 }; 226 227 #define CP_UMOUNT 0x00000001 228 #define CP_FASTBOOT 0x00000002 229 #define CP_SYNC 0x00000004 230 #define CP_RECOVERY 0x00000008 231 #define CP_DISCARD 0x00000010 232 #define CP_TRIMMED 0x00000020 233 #define CP_PAUSE 0x00000040 234 #define CP_RESIZE 0x00000080 235 236 #define DEF_MAX_DISCARD_REQUEST 8 /* issue 8 discards per round */ 237 #define DEF_MIN_DISCARD_ISSUE_TIME 50 /* 50 ms, if exists */ 238 #define DEF_MID_DISCARD_ISSUE_TIME 500 /* 500 ms, if device busy */ 239 #define DEF_MAX_DISCARD_ISSUE_TIME 60000 /* 60 s, if no candidates */ 240 #define DEF_DISCARD_URGENT_UTIL 80 /* do more discard over 80% */ 241 #define DEF_CP_INTERVAL 60 /* 60 secs */ 242 #define DEF_IDLE_INTERVAL 5 /* 5 secs */ 243 #define DEF_DISABLE_INTERVAL 5 /* 5 secs */ 244 #define DEF_DISABLE_QUICK_INTERVAL 1 /* 1 secs */ 245 #define DEF_UMOUNT_DISCARD_TIMEOUT 5 /* 5 secs */ 246 247 struct cp_control { 248 int reason; 249 __u64 trim_start; 250 __u64 trim_end; 251 __u64 trim_minlen; 252 }; 253 254 /* 255 * indicate meta/data type 256 */ 257 enum { 258 META_CP, 259 META_NAT, 260 META_SIT, 261 META_SSA, 262 META_MAX, 263 META_POR, 264 DATA_GENERIC, /* check range only */ 265 DATA_GENERIC_ENHANCE, /* strong check on range and segment bitmap */ 266 DATA_GENERIC_ENHANCE_READ, /* 267 * strong check on range and segment 268 * bitmap but no warning due to race 269 * condition of read on truncated area 270 * by extent_cache 271 */ 272 DATA_GENERIC_ENHANCE_UPDATE, /* 273 * strong check on range and segment 274 * bitmap for update case 275 */ 276 META_GENERIC, 277 }; 278 279 /* for the list of ino */ 280 enum { 281 ORPHAN_INO, /* for orphan ino list */ 282 APPEND_INO, /* for append ino list */ 283 UPDATE_INO, /* for update ino list */ 284 TRANS_DIR_INO, /* for transactions dir ino list */ 285 FLUSH_INO, /* for multiple device flushing */ 286 MAX_INO_ENTRY, /* max. list */ 287 }; 288 289 struct ino_entry { 290 struct list_head list; /* list head */ 291 nid_t ino; /* inode number */ 292 unsigned int dirty_device; /* dirty device bitmap */ 293 }; 294 295 /* for the list of inodes to be GCed */ 296 struct inode_entry { 297 struct list_head list; /* list head */ 298 struct inode *inode; /* vfs inode pointer */ 299 }; 300 301 struct fsync_node_entry { 302 struct list_head list; /* list head */ 303 struct page *page; /* warm node page pointer */ 304 unsigned int seq_id; /* sequence id */ 305 }; 306 307 struct ckpt_req { 308 struct completion wait; /* completion for checkpoint done */ 309 struct llist_node llnode; /* llist_node to be linked in wait queue */ 310 int ret; /* return code of checkpoint */ 311 ktime_t queue_time; /* request queued time */ 312 }; 313 314 struct ckpt_req_control { 315 struct task_struct *f2fs_issue_ckpt; /* checkpoint task */ 316 int ckpt_thread_ioprio; /* checkpoint merge thread ioprio */ 317 wait_queue_head_t ckpt_wait_queue; /* waiting queue for wake-up */ 318 atomic_t issued_ckpt; /* # of actually issued ckpts */ 319 atomic_t total_ckpt; /* # of total ckpts */ 320 atomic_t queued_ckpt; /* # of queued ckpts */ 321 struct llist_head issue_list; /* list for command issue */ 322 spinlock_t stat_lock; /* lock for below checkpoint time stats */ 323 unsigned int cur_time; /* cur wait time in msec for currently issued checkpoint */ 324 unsigned int peak_time; /* peak wait time in msec until now */ 325 }; 326 327 /* for the bitmap indicate blocks to be discarded */ 328 struct discard_entry { 329 struct list_head list; /* list head */ 330 block_t start_blkaddr; /* start blockaddr of current segment */ 331 unsigned char discard_map[SIT_VBLOCK_MAP_SIZE]; /* segment discard bitmap */ 332 }; 333 334 /* minimum discard granularity, unit: block count */ 335 #define MIN_DISCARD_GRANULARITY 1 336 /* default discard granularity of inner discard thread, unit: block count */ 337 #define DEFAULT_DISCARD_GRANULARITY 16 338 /* default maximum discard granularity of ordered discard, unit: block count */ 339 #define DEFAULT_MAX_ORDERED_DISCARD_GRANULARITY 16 340 341 /* max discard pend list number */ 342 #define MAX_PLIST_NUM 512 343 #define plist_idx(blk_num) ((blk_num) >= MAX_PLIST_NUM ? \ 344 (MAX_PLIST_NUM - 1) : ((blk_num) - 1)) 345 346 enum { 347 D_PREP, /* initial */ 348 D_PARTIAL, /* partially submitted */ 349 D_SUBMIT, /* all submitted */ 350 D_DONE, /* finished */ 351 }; 352 353 struct discard_info { 354 block_t lstart; /* logical start address */ 355 block_t len; /* length */ 356 block_t start; /* actual start address in dev */ 357 }; 358 359 struct discard_cmd { 360 struct rb_node rb_node; /* rb node located in rb-tree */ 361 struct discard_info di; /* discard info */ 362 struct list_head list; /* command list */ 363 struct completion wait; /* compleation */ 364 struct block_device *bdev; /* bdev */ 365 unsigned short ref; /* reference count */ 366 unsigned char state; /* state */ 367 unsigned char queued; /* queued discard */ 368 int error; /* bio error */ 369 spinlock_t lock; /* for state/bio_ref updating */ 370 unsigned short bio_ref; /* bio reference count */ 371 }; 372 373 enum { 374 DPOLICY_BG, 375 DPOLICY_FORCE, 376 DPOLICY_FSTRIM, 377 DPOLICY_UMOUNT, 378 MAX_DPOLICY, 379 }; 380 381 struct discard_policy { 382 int type; /* type of discard */ 383 unsigned int min_interval; /* used for candidates exist */ 384 unsigned int mid_interval; /* used for device busy */ 385 unsigned int max_interval; /* used for candidates not exist */ 386 unsigned int max_requests; /* # of discards issued per round */ 387 unsigned int io_aware_gran; /* minimum granularity discard not be aware of I/O */ 388 bool io_aware; /* issue discard in idle time */ 389 bool sync; /* submit discard with REQ_SYNC flag */ 390 bool ordered; /* issue discard by lba order */ 391 bool timeout; /* discard timeout for put_super */ 392 unsigned int granularity; /* discard granularity */ 393 }; 394 395 struct discard_cmd_control { 396 struct task_struct *f2fs_issue_discard; /* discard thread */ 397 struct list_head entry_list; /* 4KB discard entry list */ 398 struct list_head pend_list[MAX_PLIST_NUM];/* store pending entries */ 399 struct list_head wait_list; /* store on-flushing entries */ 400 struct list_head fstrim_list; /* in-flight discard from fstrim */ 401 wait_queue_head_t discard_wait_queue; /* waiting queue for wake-up */ 402 struct mutex cmd_lock; 403 unsigned int nr_discards; /* # of discards in the list */ 404 unsigned int max_discards; /* max. discards to be issued */ 405 unsigned int max_discard_request; /* max. discard request per round */ 406 unsigned int min_discard_issue_time; /* min. interval between discard issue */ 407 unsigned int mid_discard_issue_time; /* mid. interval between discard issue */ 408 unsigned int max_discard_issue_time; /* max. interval between discard issue */ 409 unsigned int discard_io_aware_gran; /* minimum discard granularity not be aware of I/O */ 410 unsigned int discard_urgent_util; /* utilization which issue discard proactively */ 411 unsigned int discard_granularity; /* discard granularity */ 412 unsigned int max_ordered_discard; /* maximum discard granularity issued by lba order */ 413 unsigned int undiscard_blks; /* # of undiscard blocks */ 414 unsigned int next_pos; /* next discard position */ 415 atomic_t issued_discard; /* # of issued discard */ 416 atomic_t queued_discard; /* # of queued discard */ 417 atomic_t discard_cmd_cnt; /* # of cached cmd count */ 418 struct rb_root_cached root; /* root of discard rb-tree */ 419 bool rbtree_check; /* config for consistence check */ 420 bool discard_wake; /* to wake up discard thread */ 421 }; 422 423 /* for the list of fsync inodes, used only during recovery */ 424 struct fsync_inode_entry { 425 struct list_head list; /* list head */ 426 struct inode *inode; /* vfs inode pointer */ 427 block_t blkaddr; /* block address locating the last fsync */ 428 block_t last_dentry; /* block address locating the last dentry */ 429 }; 430 431 #define nats_in_cursum(jnl) (le16_to_cpu((jnl)->n_nats)) 432 #define sits_in_cursum(jnl) (le16_to_cpu((jnl)->n_sits)) 433 434 #define nat_in_journal(jnl, i) ((jnl)->nat_j.entries[i].ne) 435 #define nid_in_journal(jnl, i) ((jnl)->nat_j.entries[i].nid) 436 #define sit_in_journal(jnl, i) ((jnl)->sit_j.entries[i].se) 437 #define segno_in_journal(jnl, i) ((jnl)->sit_j.entries[i].segno) 438 439 #define MAX_NAT_JENTRIES(jnl) (NAT_JOURNAL_ENTRIES - nats_in_cursum(jnl)) 440 #define MAX_SIT_JENTRIES(jnl) (SIT_JOURNAL_ENTRIES - sits_in_cursum(jnl)) 441 442 static inline int update_nats_in_cursum(struct f2fs_journal *journal, int i) 443 { 444 int before = nats_in_cursum(journal); 445 446 journal->n_nats = cpu_to_le16(before + i); 447 return before; 448 } 449 450 static inline int update_sits_in_cursum(struct f2fs_journal *journal, int i) 451 { 452 int before = sits_in_cursum(journal); 453 454 journal->n_sits = cpu_to_le16(before + i); 455 return before; 456 } 457 458 static inline bool __has_cursum_space(struct f2fs_journal *journal, 459 int size, int type) 460 { 461 if (type == NAT_JOURNAL) 462 return size <= MAX_NAT_JENTRIES(journal); 463 return size <= MAX_SIT_JENTRIES(journal); 464 } 465 466 /* for inline stuff */ 467 #define DEF_INLINE_RESERVED_SIZE 1 468 static inline int get_extra_isize(struct inode *inode); 469 static inline int get_inline_xattr_addrs(struct inode *inode); 470 #define MAX_INLINE_DATA(inode) (sizeof(__le32) * \ 471 (CUR_ADDRS_PER_INODE(inode) - \ 472 get_inline_xattr_addrs(inode) - \ 473 DEF_INLINE_RESERVED_SIZE)) 474 475 /* for inline dir */ 476 #define NR_INLINE_DENTRY(inode) (MAX_INLINE_DATA(inode) * BITS_PER_BYTE / \ 477 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \ 478 BITS_PER_BYTE + 1)) 479 #define INLINE_DENTRY_BITMAP_SIZE(inode) \ 480 DIV_ROUND_UP(NR_INLINE_DENTRY(inode), BITS_PER_BYTE) 481 #define INLINE_RESERVED_SIZE(inode) (MAX_INLINE_DATA(inode) - \ 482 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \ 483 NR_INLINE_DENTRY(inode) + \ 484 INLINE_DENTRY_BITMAP_SIZE(inode))) 485 486 /* 487 * For INODE and NODE manager 488 */ 489 /* for directory operations */ 490 491 struct f2fs_filename { 492 /* 493 * The filename the user specified. This is NULL for some 494 * filesystem-internal operations, e.g. converting an inline directory 495 * to a non-inline one, or roll-forward recovering an encrypted dentry. 496 */ 497 const struct qstr *usr_fname; 498 499 /* 500 * The on-disk filename. For encrypted directories, this is encrypted. 501 * This may be NULL for lookups in an encrypted dir without the key. 502 */ 503 struct fscrypt_str disk_name; 504 505 /* The dirhash of this filename */ 506 f2fs_hash_t hash; 507 508 #ifdef CONFIG_FS_ENCRYPTION 509 /* 510 * For lookups in encrypted directories: either the buffer backing 511 * disk_name, or a buffer that holds the decoded no-key name. 512 */ 513 struct fscrypt_str crypto_buf; 514 #endif 515 #if IS_ENABLED(CONFIG_UNICODE) 516 /* 517 * For casefolded directories: the casefolded name, but it's left NULL 518 * if the original name is not valid Unicode, if the original name is 519 * "." or "..", if the directory is both casefolded and encrypted and 520 * its encryption key is unavailable, or if the filesystem is doing an 521 * internal operation where usr_fname is also NULL. In all these cases 522 * we fall back to treating the name as an opaque byte sequence. 523 */ 524 struct fscrypt_str cf_name; 525 #endif 526 }; 527 528 struct f2fs_dentry_ptr { 529 struct inode *inode; 530 void *bitmap; 531 struct f2fs_dir_entry *dentry; 532 __u8 (*filename)[F2FS_SLOT_LEN]; 533 int max; 534 int nr_bitmap; 535 }; 536 537 static inline void make_dentry_ptr_block(struct inode *inode, 538 struct f2fs_dentry_ptr *d, struct f2fs_dentry_block *t) 539 { 540 d->inode = inode; 541 d->max = NR_DENTRY_IN_BLOCK; 542 d->nr_bitmap = SIZE_OF_DENTRY_BITMAP; 543 d->bitmap = t->dentry_bitmap; 544 d->dentry = t->dentry; 545 d->filename = t->filename; 546 } 547 548 static inline void make_dentry_ptr_inline(struct inode *inode, 549 struct f2fs_dentry_ptr *d, void *t) 550 { 551 int entry_cnt = NR_INLINE_DENTRY(inode); 552 int bitmap_size = INLINE_DENTRY_BITMAP_SIZE(inode); 553 int reserved_size = INLINE_RESERVED_SIZE(inode); 554 555 d->inode = inode; 556 d->max = entry_cnt; 557 d->nr_bitmap = bitmap_size; 558 d->bitmap = t; 559 d->dentry = t + bitmap_size + reserved_size; 560 d->filename = t + bitmap_size + reserved_size + 561 SIZE_OF_DIR_ENTRY * entry_cnt; 562 } 563 564 /* 565 * XATTR_NODE_OFFSET stores xattrs to one node block per file keeping -1 566 * as its node offset to distinguish from index node blocks. 567 * But some bits are used to mark the node block. 568 */ 569 #define XATTR_NODE_OFFSET ((((unsigned int)-1) << OFFSET_BIT_SHIFT) \ 570 >> OFFSET_BIT_SHIFT) 571 enum { 572 ALLOC_NODE, /* allocate a new node page if needed */ 573 LOOKUP_NODE, /* look up a node without readahead */ 574 LOOKUP_NODE_RA, /* 575 * look up a node with readahead called 576 * by get_data_block. 577 */ 578 }; 579 580 #define DEFAULT_RETRY_IO_COUNT 8 /* maximum retry read IO or flush count */ 581 582 /* congestion wait timeout value, default: 20ms */ 583 #define DEFAULT_IO_TIMEOUT (msecs_to_jiffies(20)) 584 585 /* maximum retry quota flush count */ 586 #define DEFAULT_RETRY_QUOTA_FLUSH_COUNT 8 587 588 /* maximum retry of EIO'ed page */ 589 #define MAX_RETRY_PAGE_EIO 100 590 591 #define F2FS_LINK_MAX 0xffffffff /* maximum link count per file */ 592 593 #define MAX_DIR_RA_PAGES 4 /* maximum ra pages of dir */ 594 595 /* dirty segments threshold for triggering CP */ 596 #define DEFAULT_DIRTY_THRESHOLD 4 597 598 #define RECOVERY_MAX_RA_BLOCKS BIO_MAX_VECS 599 #define RECOVERY_MIN_RA_BLOCKS 1 600 601 #define F2FS_ONSTACK_PAGES 16 /* nr of onstack pages */ 602 603 /* for in-memory extent cache entry */ 604 #define F2FS_MIN_EXTENT_LEN 64 /* minimum extent length */ 605 606 /* number of extent info in extent cache we try to shrink */ 607 #define READ_EXTENT_CACHE_SHRINK_NUMBER 128 608 609 /* number of age extent info in extent cache we try to shrink */ 610 #define AGE_EXTENT_CACHE_SHRINK_NUMBER 128 611 #define LAST_AGE_WEIGHT 30 612 #define SAME_AGE_REGION 1024 613 614 /* 615 * Define data block with age less than 1GB as hot data 616 * define data block with age less than 10GB but more than 1GB as warm data 617 */ 618 #define DEF_HOT_DATA_AGE_THRESHOLD 262144 619 #define DEF_WARM_DATA_AGE_THRESHOLD 2621440 620 621 /* extent cache type */ 622 enum extent_type { 623 EX_READ, 624 EX_BLOCK_AGE, 625 NR_EXTENT_CACHES, 626 }; 627 628 struct extent_info { 629 unsigned int fofs; /* start offset in a file */ 630 unsigned int len; /* length of the extent */ 631 union { 632 /* read extent_cache */ 633 struct { 634 /* start block address of the extent */ 635 block_t blk; 636 #ifdef CONFIG_F2FS_FS_COMPRESSION 637 /* physical extent length of compressed blocks */ 638 unsigned int c_len; 639 #endif 640 }; 641 /* block age extent_cache */ 642 struct { 643 /* block age of the extent */ 644 unsigned long long age; 645 /* last total blocks allocated */ 646 unsigned long long last_blocks; 647 }; 648 }; 649 }; 650 651 struct extent_node { 652 struct rb_node rb_node; /* rb node located in rb-tree */ 653 struct extent_info ei; /* extent info */ 654 struct list_head list; /* node in global extent list of sbi */ 655 struct extent_tree *et; /* extent tree pointer */ 656 }; 657 658 struct extent_tree { 659 nid_t ino; /* inode number */ 660 enum extent_type type; /* keep the extent tree type */ 661 struct rb_root_cached root; /* root of extent info rb-tree */ 662 struct extent_node *cached_en; /* recently accessed extent node */ 663 struct list_head list; /* to be used by sbi->zombie_list */ 664 rwlock_t lock; /* protect extent info rb-tree */ 665 atomic_t node_cnt; /* # of extent node in rb-tree*/ 666 bool largest_updated; /* largest extent updated */ 667 struct extent_info largest; /* largest cached extent for EX_READ */ 668 }; 669 670 struct extent_tree_info { 671 struct radix_tree_root extent_tree_root;/* cache extent cache entries */ 672 struct mutex extent_tree_lock; /* locking extent radix tree */ 673 struct list_head extent_list; /* lru list for shrinker */ 674 spinlock_t extent_lock; /* locking extent lru list */ 675 atomic_t total_ext_tree; /* extent tree count */ 676 struct list_head zombie_list; /* extent zombie tree list */ 677 atomic_t total_zombie_tree; /* extent zombie tree count */ 678 atomic_t total_ext_node; /* extent info count */ 679 }; 680 681 /* 682 * State of block returned by f2fs_map_blocks. 683 */ 684 #define F2FS_MAP_NEW (1U << 0) 685 #define F2FS_MAP_MAPPED (1U << 1) 686 #define F2FS_MAP_DELALLOC (1U << 2) 687 #define F2FS_MAP_FLAGS (F2FS_MAP_NEW | F2FS_MAP_MAPPED |\ 688 F2FS_MAP_DELALLOC) 689 690 struct f2fs_map_blocks { 691 struct block_device *m_bdev; /* for multi-device dio */ 692 block_t m_pblk; 693 block_t m_lblk; 694 unsigned int m_len; 695 unsigned int m_flags; 696 pgoff_t *m_next_pgofs; /* point next possible non-hole pgofs */ 697 pgoff_t *m_next_extent; /* point to next possible extent */ 698 int m_seg_type; 699 bool m_may_create; /* indicate it is from write path */ 700 bool m_multidev_dio; /* indicate it allows multi-device dio */ 701 }; 702 703 /* for flag in get_data_block */ 704 enum { 705 F2FS_GET_BLOCK_DEFAULT, 706 F2FS_GET_BLOCK_FIEMAP, 707 F2FS_GET_BLOCK_BMAP, 708 F2FS_GET_BLOCK_DIO, 709 F2FS_GET_BLOCK_PRE_DIO, 710 F2FS_GET_BLOCK_PRE_AIO, 711 F2FS_GET_BLOCK_PRECACHE, 712 }; 713 714 /* 715 * i_advise uses FADVISE_XXX_BIT. We can add additional hints later. 716 */ 717 #define FADVISE_COLD_BIT 0x01 718 #define FADVISE_LOST_PINO_BIT 0x02 719 #define FADVISE_ENCRYPT_BIT 0x04 720 #define FADVISE_ENC_NAME_BIT 0x08 721 #define FADVISE_KEEP_SIZE_BIT 0x10 722 #define FADVISE_HOT_BIT 0x20 723 #define FADVISE_VERITY_BIT 0x40 724 #define FADVISE_TRUNC_BIT 0x80 725 726 #define FADVISE_MODIFIABLE_BITS (FADVISE_COLD_BIT | FADVISE_HOT_BIT) 727 728 #define file_is_cold(inode) is_file(inode, FADVISE_COLD_BIT) 729 #define file_set_cold(inode) set_file(inode, FADVISE_COLD_BIT) 730 #define file_clear_cold(inode) clear_file(inode, FADVISE_COLD_BIT) 731 732 #define file_wrong_pino(inode) is_file(inode, FADVISE_LOST_PINO_BIT) 733 #define file_lost_pino(inode) set_file(inode, FADVISE_LOST_PINO_BIT) 734 #define file_got_pino(inode) clear_file(inode, FADVISE_LOST_PINO_BIT) 735 736 #define file_is_encrypt(inode) is_file(inode, FADVISE_ENCRYPT_BIT) 737 #define file_set_encrypt(inode) set_file(inode, FADVISE_ENCRYPT_BIT) 738 739 #define file_enc_name(inode) is_file(inode, FADVISE_ENC_NAME_BIT) 740 #define file_set_enc_name(inode) set_file(inode, FADVISE_ENC_NAME_BIT) 741 742 #define file_keep_isize(inode) is_file(inode, FADVISE_KEEP_SIZE_BIT) 743 #define file_set_keep_isize(inode) set_file(inode, FADVISE_KEEP_SIZE_BIT) 744 745 #define file_is_hot(inode) is_file(inode, FADVISE_HOT_BIT) 746 #define file_set_hot(inode) set_file(inode, FADVISE_HOT_BIT) 747 #define file_clear_hot(inode) clear_file(inode, FADVISE_HOT_BIT) 748 749 #define file_is_verity(inode) is_file(inode, FADVISE_VERITY_BIT) 750 #define file_set_verity(inode) set_file(inode, FADVISE_VERITY_BIT) 751 752 #define file_should_truncate(inode) is_file(inode, FADVISE_TRUNC_BIT) 753 #define file_need_truncate(inode) set_file(inode, FADVISE_TRUNC_BIT) 754 #define file_dont_truncate(inode) clear_file(inode, FADVISE_TRUNC_BIT) 755 756 #define DEF_DIR_LEVEL 0 757 758 enum { 759 GC_FAILURE_PIN, 760 MAX_GC_FAILURE 761 }; 762 763 /* used for f2fs_inode_info->flags */ 764 enum { 765 FI_NEW_INODE, /* indicate newly allocated inode */ 766 FI_DIRTY_INODE, /* indicate inode is dirty or not */ 767 FI_AUTO_RECOVER, /* indicate inode is recoverable */ 768 FI_DIRTY_DIR, /* indicate directory has dirty pages */ 769 FI_INC_LINK, /* need to increment i_nlink */ 770 FI_ACL_MODE, /* indicate acl mode */ 771 FI_NO_ALLOC, /* should not allocate any blocks */ 772 FI_FREE_NID, /* free allocated nide */ 773 FI_NO_EXTENT, /* not to use the extent cache */ 774 FI_INLINE_XATTR, /* used for inline xattr */ 775 FI_INLINE_DATA, /* used for inline data*/ 776 FI_INLINE_DENTRY, /* used for inline dentry */ 777 FI_APPEND_WRITE, /* inode has appended data */ 778 FI_UPDATE_WRITE, /* inode has in-place-update data */ 779 FI_NEED_IPU, /* used for ipu per file */ 780 FI_ATOMIC_FILE, /* indicate atomic file */ 781 FI_DATA_EXIST, /* indicate data exists */ 782 FI_INLINE_DOTS, /* indicate inline dot dentries */ 783 FI_SKIP_WRITES, /* should skip data page writeback */ 784 FI_OPU_WRITE, /* used for opu per file */ 785 FI_DIRTY_FILE, /* indicate regular/symlink has dirty pages */ 786 FI_PREALLOCATED_ALL, /* all blocks for write were preallocated */ 787 FI_HOT_DATA, /* indicate file is hot */ 788 FI_EXTRA_ATTR, /* indicate file has extra attribute */ 789 FI_PROJ_INHERIT, /* indicate file inherits projectid */ 790 FI_PIN_FILE, /* indicate file should not be gced */ 791 FI_VERITY_IN_PROGRESS, /* building fs-verity Merkle tree */ 792 FI_COMPRESSED_FILE, /* indicate file's data can be compressed */ 793 FI_COMPRESS_CORRUPT, /* indicate compressed cluster is corrupted */ 794 FI_MMAP_FILE, /* indicate file was mmapped */ 795 FI_ENABLE_COMPRESS, /* enable compression in "user" compression mode */ 796 FI_COMPRESS_RELEASED, /* compressed blocks were released */ 797 FI_ALIGNED_WRITE, /* enable aligned write */ 798 FI_COW_FILE, /* indicate COW file */ 799 FI_ATOMIC_COMMITTED, /* indicate atomic commit completed except disk sync */ 800 FI_ATOMIC_REPLACE, /* indicate atomic replace */ 801 FI_MAX, /* max flag, never be used */ 802 }; 803 804 struct f2fs_inode_info { 805 struct inode vfs_inode; /* serve a vfs inode */ 806 unsigned long i_flags; /* keep an inode flags for ioctl */ 807 unsigned char i_advise; /* use to give file attribute hints */ 808 unsigned char i_dir_level; /* use for dentry level for large dir */ 809 unsigned int i_current_depth; /* only for directory depth */ 810 /* for gc failure statistic */ 811 unsigned int i_gc_failures[MAX_GC_FAILURE]; 812 unsigned int i_pino; /* parent inode number */ 813 umode_t i_acl_mode; /* keep file acl mode temporarily */ 814 815 /* Use below internally in f2fs*/ 816 unsigned long flags[BITS_TO_LONGS(FI_MAX)]; /* use to pass per-file flags */ 817 struct f2fs_rwsem i_sem; /* protect fi info */ 818 atomic_t dirty_pages; /* # of dirty pages */ 819 f2fs_hash_t chash; /* hash value of given file name */ 820 unsigned int clevel; /* maximum level of given file name */ 821 struct task_struct *task; /* lookup and create consistency */ 822 struct task_struct *cp_task; /* separate cp/wb IO stats*/ 823 struct task_struct *wb_task; /* indicate inode is in context of writeback */ 824 nid_t i_xattr_nid; /* node id that contains xattrs */ 825 loff_t last_disk_size; /* lastly written file size */ 826 spinlock_t i_size_lock; /* protect last_disk_size */ 827 828 #ifdef CONFIG_QUOTA 829 struct dquot __rcu *i_dquot[MAXQUOTAS]; 830 831 /* quota space reservation, managed internally by quota code */ 832 qsize_t i_reserved_quota; 833 #endif 834 struct list_head dirty_list; /* dirty list for dirs and files */ 835 struct list_head gdirty_list; /* linked in global dirty list */ 836 struct task_struct *atomic_write_task; /* store atomic write task */ 837 struct extent_tree *extent_tree[NR_EXTENT_CACHES]; 838 /* cached extent_tree entry */ 839 struct inode *cow_inode; /* copy-on-write inode for atomic write */ 840 841 /* avoid racing between foreground op and gc */ 842 struct f2fs_rwsem i_gc_rwsem[2]; 843 struct f2fs_rwsem i_xattr_sem; /* avoid racing between reading and changing EAs */ 844 845 int i_extra_isize; /* size of extra space located in i_addr */ 846 kprojid_t i_projid; /* id for project quota */ 847 int i_inline_xattr_size; /* inline xattr size */ 848 struct timespec64 i_crtime; /* inode creation time */ 849 struct timespec64 i_disk_time[3];/* inode disk times */ 850 851 /* for file compress */ 852 atomic_t i_compr_blocks; /* # of compressed blocks */ 853 unsigned char i_compress_algorithm; /* algorithm type */ 854 unsigned char i_log_cluster_size; /* log of cluster size */ 855 unsigned char i_compress_level; /* compress level (lz4hc,zstd) */ 856 unsigned char i_compress_flag; /* compress flag */ 857 unsigned int i_cluster_size; /* cluster size */ 858 859 unsigned int atomic_write_cnt; 860 loff_t original_i_size; /* original i_size before atomic write */ 861 }; 862 863 static inline void get_read_extent_info(struct extent_info *ext, 864 struct f2fs_extent *i_ext) 865 { 866 ext->fofs = le32_to_cpu(i_ext->fofs); 867 ext->blk = le32_to_cpu(i_ext->blk); 868 ext->len = le32_to_cpu(i_ext->len); 869 } 870 871 static inline void set_raw_read_extent(struct extent_info *ext, 872 struct f2fs_extent *i_ext) 873 { 874 i_ext->fofs = cpu_to_le32(ext->fofs); 875 i_ext->blk = cpu_to_le32(ext->blk); 876 i_ext->len = cpu_to_le32(ext->len); 877 } 878 879 static inline bool __is_discard_mergeable(struct discard_info *back, 880 struct discard_info *front, unsigned int max_len) 881 { 882 return (back->lstart + back->len == front->lstart) && 883 (back->len + front->len <= max_len); 884 } 885 886 static inline bool __is_discard_back_mergeable(struct discard_info *cur, 887 struct discard_info *back, unsigned int max_len) 888 { 889 return __is_discard_mergeable(back, cur, max_len); 890 } 891 892 static inline bool __is_discard_front_mergeable(struct discard_info *cur, 893 struct discard_info *front, unsigned int max_len) 894 { 895 return __is_discard_mergeable(cur, front, max_len); 896 } 897 898 /* 899 * For free nid management 900 */ 901 enum nid_state { 902 FREE_NID, /* newly added to free nid list */ 903 PREALLOC_NID, /* it is preallocated */ 904 MAX_NID_STATE, 905 }; 906 907 enum nat_state { 908 TOTAL_NAT, 909 DIRTY_NAT, 910 RECLAIMABLE_NAT, 911 MAX_NAT_STATE, 912 }; 913 914 struct f2fs_nm_info { 915 block_t nat_blkaddr; /* base disk address of NAT */ 916 nid_t max_nid; /* maximum possible node ids */ 917 nid_t available_nids; /* # of available node ids */ 918 nid_t next_scan_nid; /* the next nid to be scanned */ 919 nid_t max_rf_node_blocks; /* max # of nodes for recovery */ 920 unsigned int ram_thresh; /* control the memory footprint */ 921 unsigned int ra_nid_pages; /* # of nid pages to be readaheaded */ 922 unsigned int dirty_nats_ratio; /* control dirty nats ratio threshold */ 923 924 /* NAT cache management */ 925 struct radix_tree_root nat_root;/* root of the nat entry cache */ 926 struct radix_tree_root nat_set_root;/* root of the nat set cache */ 927 struct f2fs_rwsem nat_tree_lock; /* protect nat entry tree */ 928 struct list_head nat_entries; /* cached nat entry list (clean) */ 929 spinlock_t nat_list_lock; /* protect clean nat entry list */ 930 unsigned int nat_cnt[MAX_NAT_STATE]; /* the # of cached nat entries */ 931 unsigned int nat_blocks; /* # of nat blocks */ 932 933 /* free node ids management */ 934 struct radix_tree_root free_nid_root;/* root of the free_nid cache */ 935 struct list_head free_nid_list; /* list for free nids excluding preallocated nids */ 936 unsigned int nid_cnt[MAX_NID_STATE]; /* the number of free node id */ 937 spinlock_t nid_list_lock; /* protect nid lists ops */ 938 struct mutex build_lock; /* lock for build free nids */ 939 unsigned char **free_nid_bitmap; 940 unsigned char *nat_block_bitmap; 941 unsigned short *free_nid_count; /* free nid count of NAT block */ 942 943 /* for checkpoint */ 944 char *nat_bitmap; /* NAT bitmap pointer */ 945 946 unsigned int nat_bits_blocks; /* # of nat bits blocks */ 947 unsigned char *nat_bits; /* NAT bits blocks */ 948 unsigned char *full_nat_bits; /* full NAT pages */ 949 unsigned char *empty_nat_bits; /* empty NAT pages */ 950 #ifdef CONFIG_F2FS_CHECK_FS 951 char *nat_bitmap_mir; /* NAT bitmap mirror */ 952 #endif 953 int bitmap_size; /* bitmap size */ 954 }; 955 956 /* 957 * this structure is used as one of function parameters. 958 * all the information are dedicated to a given direct node block determined 959 * by the data offset in a file. 960 */ 961 struct dnode_of_data { 962 struct inode *inode; /* vfs inode pointer */ 963 struct page *inode_page; /* its inode page, NULL is possible */ 964 struct page *node_page; /* cached direct node page */ 965 nid_t nid; /* node id of the direct node block */ 966 unsigned int ofs_in_node; /* data offset in the node page */ 967 bool inode_page_locked; /* inode page is locked or not */ 968 bool node_changed; /* is node block changed */ 969 char cur_level; /* level of hole node page */ 970 char max_level; /* level of current page located */ 971 block_t data_blkaddr; /* block address of the node block */ 972 }; 973 974 static inline void set_new_dnode(struct dnode_of_data *dn, struct inode *inode, 975 struct page *ipage, struct page *npage, nid_t nid) 976 { 977 memset(dn, 0, sizeof(*dn)); 978 dn->inode = inode; 979 dn->inode_page = ipage; 980 dn->node_page = npage; 981 dn->nid = nid; 982 } 983 984 /* 985 * For SIT manager 986 * 987 * By default, there are 6 active log areas across the whole main area. 988 * When considering hot and cold data separation to reduce cleaning overhead, 989 * we split 3 for data logs and 3 for node logs as hot, warm, and cold types, 990 * respectively. 991 * In the current design, you should not change the numbers intentionally. 992 * Instead, as a mount option such as active_logs=x, you can use 2, 4, and 6 993 * logs individually according to the underlying devices. (default: 6) 994 * Just in case, on-disk layout covers maximum 16 logs that consist of 8 for 995 * data and 8 for node logs. 996 */ 997 #define NR_CURSEG_DATA_TYPE (3) 998 #define NR_CURSEG_NODE_TYPE (3) 999 #define NR_CURSEG_INMEM_TYPE (2) 1000 #define NR_CURSEG_RO_TYPE (2) 1001 #define NR_CURSEG_PERSIST_TYPE (NR_CURSEG_DATA_TYPE + NR_CURSEG_NODE_TYPE) 1002 #define NR_CURSEG_TYPE (NR_CURSEG_INMEM_TYPE + NR_CURSEG_PERSIST_TYPE) 1003 1004 enum { 1005 CURSEG_HOT_DATA = 0, /* directory entry blocks */ 1006 CURSEG_WARM_DATA, /* data blocks */ 1007 CURSEG_COLD_DATA, /* multimedia or GCed data blocks */ 1008 CURSEG_HOT_NODE, /* direct node blocks of directory files */ 1009 CURSEG_WARM_NODE, /* direct node blocks of normal files */ 1010 CURSEG_COLD_NODE, /* indirect node blocks */ 1011 NR_PERSISTENT_LOG, /* number of persistent log */ 1012 CURSEG_COLD_DATA_PINNED = NR_PERSISTENT_LOG, 1013 /* pinned file that needs consecutive block address */ 1014 CURSEG_ALL_DATA_ATGC, /* SSR alloctor in hot/warm/cold data area */ 1015 NO_CHECK_TYPE, /* number of persistent & inmem log */ 1016 }; 1017 1018 struct flush_cmd { 1019 struct completion wait; 1020 struct llist_node llnode; 1021 nid_t ino; 1022 int ret; 1023 }; 1024 1025 struct flush_cmd_control { 1026 struct task_struct *f2fs_issue_flush; /* flush thread */ 1027 wait_queue_head_t flush_wait_queue; /* waiting queue for wake-up */ 1028 atomic_t issued_flush; /* # of issued flushes */ 1029 atomic_t queued_flush; /* # of queued flushes */ 1030 struct llist_head issue_list; /* list for command issue */ 1031 struct llist_node *dispatch_list; /* list for command dispatch */ 1032 }; 1033 1034 struct f2fs_sm_info { 1035 struct sit_info *sit_info; /* whole segment information */ 1036 struct free_segmap_info *free_info; /* free segment information */ 1037 struct dirty_seglist_info *dirty_info; /* dirty segment information */ 1038 struct curseg_info *curseg_array; /* active segment information */ 1039 1040 struct f2fs_rwsem curseg_lock; /* for preventing curseg change */ 1041 1042 block_t seg0_blkaddr; /* block address of 0'th segment */ 1043 block_t main_blkaddr; /* start block address of main area */ 1044 block_t ssa_blkaddr; /* start block address of SSA area */ 1045 1046 unsigned int segment_count; /* total # of segments */ 1047 unsigned int main_segments; /* # of segments in main area */ 1048 unsigned int reserved_segments; /* # of reserved segments */ 1049 unsigned int additional_reserved_segments;/* reserved segs for IO align feature */ 1050 unsigned int ovp_segments; /* # of overprovision segments */ 1051 1052 /* a threshold to reclaim prefree segments */ 1053 unsigned int rec_prefree_segments; 1054 1055 struct list_head sit_entry_set; /* sit entry set list */ 1056 1057 unsigned int ipu_policy; /* in-place-update policy */ 1058 unsigned int min_ipu_util; /* in-place-update threshold */ 1059 unsigned int min_fsync_blocks; /* threshold for fsync */ 1060 unsigned int min_seq_blocks; /* threshold for sequential blocks */ 1061 unsigned int min_hot_blocks; /* threshold for hot block allocation */ 1062 unsigned int min_ssr_sections; /* threshold to trigger SSR allocation */ 1063 1064 /* for flush command control */ 1065 struct flush_cmd_control *fcc_info; 1066 1067 /* for discard command control */ 1068 struct discard_cmd_control *dcc_info; 1069 }; 1070 1071 /* 1072 * For superblock 1073 */ 1074 /* 1075 * COUNT_TYPE for monitoring 1076 * 1077 * f2fs monitors the number of several block types such as on-writeback, 1078 * dirty dentry blocks, dirty node blocks, and dirty meta blocks. 1079 */ 1080 #define WB_DATA_TYPE(p, f) \ 1081 (f || f2fs_is_cp_guaranteed(p) ? F2FS_WB_CP_DATA : F2FS_WB_DATA) 1082 enum count_type { 1083 F2FS_DIRTY_DENTS, 1084 F2FS_DIRTY_DATA, 1085 F2FS_DIRTY_QDATA, 1086 F2FS_DIRTY_NODES, 1087 F2FS_DIRTY_META, 1088 F2FS_DIRTY_IMETA, 1089 F2FS_WB_CP_DATA, 1090 F2FS_WB_DATA, 1091 F2FS_RD_DATA, 1092 F2FS_RD_NODE, 1093 F2FS_RD_META, 1094 F2FS_DIO_WRITE, 1095 F2FS_DIO_READ, 1096 NR_COUNT_TYPE, 1097 }; 1098 1099 /* 1100 * The below are the page types of bios used in submit_bio(). 1101 * The available types are: 1102 * DATA User data pages. It operates as async mode. 1103 * NODE Node pages. It operates as async mode. 1104 * META FS metadata pages such as SIT, NAT, CP. 1105 * NR_PAGE_TYPE The number of page types. 1106 * META_FLUSH Make sure the previous pages are written 1107 * with waiting the bio's completion 1108 * ... Only can be used with META. 1109 */ 1110 #define PAGE_TYPE_OF_BIO(type) ((type) > META ? META : (type)) 1111 #define PAGE_TYPE_ON_MAIN(type) ((type) == DATA || (type) == NODE) 1112 enum page_type { 1113 DATA = 0, 1114 NODE = 1, /* should not change this */ 1115 META, 1116 NR_PAGE_TYPE, 1117 META_FLUSH, 1118 IPU, /* the below types are used by tracepoints only. */ 1119 OPU, 1120 }; 1121 1122 enum temp_type { 1123 HOT = 0, /* must be zero for meta bio */ 1124 WARM, 1125 COLD, 1126 NR_TEMP_TYPE, 1127 }; 1128 1129 enum need_lock_type { 1130 LOCK_REQ = 0, 1131 LOCK_DONE, 1132 LOCK_RETRY, 1133 }; 1134 1135 enum cp_reason_type { 1136 CP_NO_NEEDED, 1137 CP_NON_REGULAR, 1138 CP_COMPRESSED, 1139 CP_HARDLINK, 1140 CP_SB_NEED_CP, 1141 CP_WRONG_PINO, 1142 CP_NO_SPC_ROLL, 1143 CP_NODE_NEED_CP, 1144 CP_FASTBOOT_MODE, 1145 CP_SPEC_LOG_NUM, 1146 CP_RECOVER_DIR, 1147 }; 1148 1149 enum iostat_type { 1150 /* WRITE IO */ 1151 APP_DIRECT_IO, /* app direct write IOs */ 1152 APP_BUFFERED_IO, /* app buffered write IOs */ 1153 APP_WRITE_IO, /* app write IOs */ 1154 APP_MAPPED_IO, /* app mapped IOs */ 1155 APP_BUFFERED_CDATA_IO, /* app buffered write IOs on compressed file */ 1156 APP_MAPPED_CDATA_IO, /* app mapped write IOs on compressed file */ 1157 FS_DATA_IO, /* data IOs from kworker/fsync/reclaimer */ 1158 FS_CDATA_IO, /* data IOs from kworker/fsync/reclaimer on compressed file */ 1159 FS_NODE_IO, /* node IOs from kworker/fsync/reclaimer */ 1160 FS_META_IO, /* meta IOs from kworker/reclaimer */ 1161 FS_GC_DATA_IO, /* data IOs from forground gc */ 1162 FS_GC_NODE_IO, /* node IOs from forground gc */ 1163 FS_CP_DATA_IO, /* data IOs from checkpoint */ 1164 FS_CP_NODE_IO, /* node IOs from checkpoint */ 1165 FS_CP_META_IO, /* meta IOs from checkpoint */ 1166 1167 /* READ IO */ 1168 APP_DIRECT_READ_IO, /* app direct read IOs */ 1169 APP_BUFFERED_READ_IO, /* app buffered read IOs */ 1170 APP_READ_IO, /* app read IOs */ 1171 APP_MAPPED_READ_IO, /* app mapped read IOs */ 1172 APP_BUFFERED_CDATA_READ_IO, /* app buffered read IOs on compressed file */ 1173 APP_MAPPED_CDATA_READ_IO, /* app mapped read IOs on compressed file */ 1174 FS_DATA_READ_IO, /* data read IOs */ 1175 FS_GDATA_READ_IO, /* data read IOs from background gc */ 1176 FS_CDATA_READ_IO, /* compressed data read IOs */ 1177 FS_NODE_READ_IO, /* node read IOs */ 1178 FS_META_READ_IO, /* meta read IOs */ 1179 1180 /* other */ 1181 FS_DISCARD_IO, /* discard */ 1182 FS_FLUSH_IO, /* flush */ 1183 FS_ZONE_RESET_IO, /* zone reset */ 1184 NR_IO_TYPE, 1185 }; 1186 1187 struct f2fs_io_info { 1188 struct f2fs_sb_info *sbi; /* f2fs_sb_info pointer */ 1189 nid_t ino; /* inode number */ 1190 enum page_type type; /* contains DATA/NODE/META/META_FLUSH */ 1191 enum temp_type temp; /* contains HOT/WARM/COLD */ 1192 enum req_op op; /* contains REQ_OP_ */ 1193 blk_opf_t op_flags; /* req_flag_bits */ 1194 block_t new_blkaddr; /* new block address to be written */ 1195 block_t old_blkaddr; /* old block address before Cow */ 1196 struct page *page; /* page to be written */ 1197 struct page *encrypted_page; /* encrypted page */ 1198 struct page *compressed_page; /* compressed page */ 1199 struct list_head list; /* serialize IOs */ 1200 unsigned int compr_blocks; /* # of compressed block addresses */ 1201 unsigned int need_lock:8; /* indicate we need to lock cp_rwsem */ 1202 unsigned int version:8; /* version of the node */ 1203 unsigned int submitted:1; /* indicate IO submission */ 1204 unsigned int in_list:1; /* indicate fio is in io_list */ 1205 unsigned int is_por:1; /* indicate IO is from recovery or not */ 1206 unsigned int encrypted:1; /* indicate file is encrypted */ 1207 unsigned int post_read:1; /* require post read */ 1208 enum iostat_type io_type; /* io type */ 1209 struct writeback_control *io_wbc; /* writeback control */ 1210 struct bio **bio; /* bio for ipu */ 1211 sector_t *last_block; /* last block number in bio */ 1212 }; 1213 1214 struct bio_entry { 1215 struct bio *bio; 1216 struct list_head list; 1217 }; 1218 1219 #define is_read_io(rw) ((rw) == READ) 1220 struct f2fs_bio_info { 1221 struct f2fs_sb_info *sbi; /* f2fs superblock */ 1222 struct bio *bio; /* bios to merge */ 1223 sector_t last_block_in_bio; /* last block number */ 1224 struct f2fs_io_info fio; /* store buffered io info. */ 1225 #ifdef CONFIG_BLK_DEV_ZONED 1226 struct completion zone_wait; /* condition value for the previous open zone to close */ 1227 struct bio *zone_pending_bio; /* pending bio for the previous zone */ 1228 void *bi_private; /* previous bi_private for pending bio */ 1229 #endif 1230 struct f2fs_rwsem io_rwsem; /* blocking op for bio */ 1231 spinlock_t io_lock; /* serialize DATA/NODE IOs */ 1232 struct list_head io_list; /* track fios */ 1233 struct list_head bio_list; /* bio entry list head */ 1234 struct f2fs_rwsem bio_list_lock; /* lock to protect bio entry list */ 1235 }; 1236 1237 #define FDEV(i) (sbi->devs[i]) 1238 #define RDEV(i) (raw_super->devs[i]) 1239 struct f2fs_dev_info { 1240 struct block_device *bdev; 1241 char path[MAX_PATH_LEN]; 1242 unsigned int total_segments; 1243 block_t start_blk; 1244 block_t end_blk; 1245 #ifdef CONFIG_BLK_DEV_ZONED 1246 unsigned int nr_blkz; /* Total number of zones */ 1247 unsigned long *blkz_seq; /* Bitmap indicating sequential zones */ 1248 #endif 1249 }; 1250 1251 enum inode_type { 1252 DIR_INODE, /* for dirty dir inode */ 1253 FILE_INODE, /* for dirty regular/symlink inode */ 1254 DIRTY_META, /* for all dirtied inode metadata */ 1255 NR_INODE_TYPE, 1256 }; 1257 1258 /* for inner inode cache management */ 1259 struct inode_management { 1260 struct radix_tree_root ino_root; /* ino entry array */ 1261 spinlock_t ino_lock; /* for ino entry lock */ 1262 struct list_head ino_list; /* inode list head */ 1263 unsigned long ino_num; /* number of entries */ 1264 }; 1265 1266 /* for GC_AT */ 1267 struct atgc_management { 1268 bool atgc_enabled; /* ATGC is enabled or not */ 1269 struct rb_root_cached root; /* root of victim rb-tree */ 1270 struct list_head victim_list; /* linked with all victim entries */ 1271 unsigned int victim_count; /* victim count in rb-tree */ 1272 unsigned int candidate_ratio; /* candidate ratio */ 1273 unsigned int max_candidate_count; /* max candidate count */ 1274 unsigned int age_weight; /* age weight, vblock_weight = 100 - age_weight */ 1275 unsigned long long age_threshold; /* age threshold */ 1276 }; 1277 1278 struct f2fs_gc_control { 1279 unsigned int victim_segno; /* target victim segment number */ 1280 int init_gc_type; /* FG_GC or BG_GC */ 1281 bool no_bg_gc; /* check the space and stop bg_gc */ 1282 bool should_migrate_blocks; /* should migrate blocks */ 1283 bool err_gc_skipped; /* return EAGAIN if GC skipped */ 1284 unsigned int nr_free_secs; /* # of free sections to do GC */ 1285 }; 1286 1287 /* 1288 * For s_flag in struct f2fs_sb_info 1289 * Modification on enum should be synchronized with s_flag array 1290 */ 1291 enum { 1292 SBI_IS_DIRTY, /* dirty flag for checkpoint */ 1293 SBI_IS_CLOSE, /* specify unmounting */ 1294 SBI_NEED_FSCK, /* need fsck.f2fs to fix */ 1295 SBI_POR_DOING, /* recovery is doing or not */ 1296 SBI_NEED_SB_WRITE, /* need to recover superblock */ 1297 SBI_NEED_CP, /* need to checkpoint */ 1298 SBI_IS_SHUTDOWN, /* shutdown by ioctl */ 1299 SBI_IS_RECOVERED, /* recovered orphan/data */ 1300 SBI_CP_DISABLED, /* CP was disabled last mount */ 1301 SBI_CP_DISABLED_QUICK, /* CP was disabled quickly */ 1302 SBI_QUOTA_NEED_FLUSH, /* need to flush quota info in CP */ 1303 SBI_QUOTA_SKIP_FLUSH, /* skip flushing quota in current CP */ 1304 SBI_QUOTA_NEED_REPAIR, /* quota file may be corrupted */ 1305 SBI_IS_RESIZEFS, /* resizefs is in process */ 1306 SBI_IS_FREEZING, /* freezefs is in process */ 1307 SBI_IS_WRITABLE, /* remove ro mountoption transiently */ 1308 MAX_SBI_FLAG, 1309 }; 1310 1311 enum { 1312 CP_TIME, 1313 REQ_TIME, 1314 DISCARD_TIME, 1315 GC_TIME, 1316 DISABLE_TIME, 1317 UMOUNT_DISCARD_TIMEOUT, 1318 MAX_TIME, 1319 }; 1320 1321 /* Note that you need to keep synchronization with this gc_mode_names array */ 1322 enum { 1323 GC_NORMAL, 1324 GC_IDLE_CB, 1325 GC_IDLE_GREEDY, 1326 GC_IDLE_AT, 1327 GC_URGENT_HIGH, 1328 GC_URGENT_LOW, 1329 GC_URGENT_MID, 1330 MAX_GC_MODE, 1331 }; 1332 1333 enum { 1334 BGGC_MODE_ON, /* background gc is on */ 1335 BGGC_MODE_OFF, /* background gc is off */ 1336 BGGC_MODE_SYNC, /* 1337 * background gc is on, migrating blocks 1338 * like foreground gc 1339 */ 1340 }; 1341 1342 enum { 1343 FS_MODE_ADAPTIVE, /* use both lfs/ssr allocation */ 1344 FS_MODE_LFS, /* use lfs allocation only */ 1345 FS_MODE_FRAGMENT_SEG, /* segment fragmentation mode */ 1346 FS_MODE_FRAGMENT_BLK, /* block fragmentation mode */ 1347 }; 1348 1349 enum { 1350 ALLOC_MODE_DEFAULT, /* stay default */ 1351 ALLOC_MODE_REUSE, /* reuse segments as much as possible */ 1352 }; 1353 1354 enum fsync_mode { 1355 FSYNC_MODE_POSIX, /* fsync follows posix semantics */ 1356 FSYNC_MODE_STRICT, /* fsync behaves in line with ext4 */ 1357 FSYNC_MODE_NOBARRIER, /* fsync behaves nobarrier based on posix */ 1358 }; 1359 1360 enum { 1361 COMPR_MODE_FS, /* 1362 * automatically compress compression 1363 * enabled files 1364 */ 1365 COMPR_MODE_USER, /* 1366 * automatical compression is disabled. 1367 * user can control the file compression 1368 * using ioctls 1369 */ 1370 }; 1371 1372 enum { 1373 DISCARD_UNIT_BLOCK, /* basic discard unit is block */ 1374 DISCARD_UNIT_SEGMENT, /* basic discard unit is segment */ 1375 DISCARD_UNIT_SECTION, /* basic discard unit is section */ 1376 }; 1377 1378 enum { 1379 MEMORY_MODE_NORMAL, /* memory mode for normal devices */ 1380 MEMORY_MODE_LOW, /* memory mode for low memry devices */ 1381 }; 1382 1383 enum errors_option { 1384 MOUNT_ERRORS_READONLY, /* remount fs ro on errors */ 1385 MOUNT_ERRORS_CONTINUE, /* continue on errors */ 1386 MOUNT_ERRORS_PANIC, /* panic on errors */ 1387 }; 1388 1389 enum { 1390 BACKGROUND, 1391 FOREGROUND, 1392 MAX_CALL_TYPE, 1393 TOTAL_CALL = FOREGROUND, 1394 }; 1395 1396 static inline int f2fs_test_bit(unsigned int nr, char *addr); 1397 static inline void f2fs_set_bit(unsigned int nr, char *addr); 1398 static inline void f2fs_clear_bit(unsigned int nr, char *addr); 1399 1400 /* 1401 * Layout of f2fs page.private: 1402 * 1403 * Layout A: lowest bit should be 1 1404 * | bit0 = 1 | bit1 | bit2 | ... | bit MAX | private data .... | 1405 * bit 0 PAGE_PRIVATE_NOT_POINTER 1406 * bit 1 PAGE_PRIVATE_ONGOING_MIGRATION 1407 * bit 2 PAGE_PRIVATE_INLINE_INODE 1408 * bit 3 PAGE_PRIVATE_REF_RESOURCE 1409 * bit 4- f2fs private data 1410 * 1411 * Layout B: lowest bit should be 0 1412 * page.private is a wrapped pointer. 1413 */ 1414 enum { 1415 PAGE_PRIVATE_NOT_POINTER, /* private contains non-pointer data */ 1416 PAGE_PRIVATE_ONGOING_MIGRATION, /* data page which is on-going migrating */ 1417 PAGE_PRIVATE_INLINE_INODE, /* inode page contains inline data */ 1418 PAGE_PRIVATE_REF_RESOURCE, /* dirty page has referenced resources */ 1419 PAGE_PRIVATE_MAX 1420 }; 1421 1422 /* For compression */ 1423 enum compress_algorithm_type { 1424 COMPRESS_LZO, 1425 COMPRESS_LZ4, 1426 COMPRESS_ZSTD, 1427 COMPRESS_LZORLE, 1428 COMPRESS_MAX, 1429 }; 1430 1431 enum compress_flag { 1432 COMPRESS_CHKSUM, 1433 COMPRESS_MAX_FLAG, 1434 }; 1435 1436 #define COMPRESS_WATERMARK 20 1437 #define COMPRESS_PERCENT 20 1438 1439 #define COMPRESS_DATA_RESERVED_SIZE 4 1440 struct compress_data { 1441 __le32 clen; /* compressed data size */ 1442 __le32 chksum; /* compressed data chksum */ 1443 __le32 reserved[COMPRESS_DATA_RESERVED_SIZE]; /* reserved */ 1444 u8 cdata[]; /* compressed data */ 1445 }; 1446 1447 #define COMPRESS_HEADER_SIZE (sizeof(struct compress_data)) 1448 1449 #define F2FS_COMPRESSED_PAGE_MAGIC 0xF5F2C000 1450 1451 #define F2FS_ZSTD_DEFAULT_CLEVEL 1 1452 1453 #define COMPRESS_LEVEL_OFFSET 8 1454 1455 /* compress context */ 1456 struct compress_ctx { 1457 struct inode *inode; /* inode the context belong to */ 1458 pgoff_t cluster_idx; /* cluster index number */ 1459 unsigned int cluster_size; /* page count in cluster */ 1460 unsigned int log_cluster_size; /* log of cluster size */ 1461 struct page **rpages; /* pages store raw data in cluster */ 1462 unsigned int nr_rpages; /* total page number in rpages */ 1463 struct page **cpages; /* pages store compressed data in cluster */ 1464 unsigned int nr_cpages; /* total page number in cpages */ 1465 unsigned int valid_nr_cpages; /* valid page number in cpages */ 1466 void *rbuf; /* virtual mapped address on rpages */ 1467 struct compress_data *cbuf; /* virtual mapped address on cpages */ 1468 size_t rlen; /* valid data length in rbuf */ 1469 size_t clen; /* valid data length in cbuf */ 1470 void *private; /* payload buffer for specified compression algorithm */ 1471 void *private2; /* extra payload buffer */ 1472 }; 1473 1474 /* compress context for write IO path */ 1475 struct compress_io_ctx { 1476 u32 magic; /* magic number to indicate page is compressed */ 1477 struct inode *inode; /* inode the context belong to */ 1478 struct page **rpages; /* pages store raw data in cluster */ 1479 unsigned int nr_rpages; /* total page number in rpages */ 1480 atomic_t pending_pages; /* in-flight compressed page count */ 1481 }; 1482 1483 /* Context for decompressing one cluster on the read IO path */ 1484 struct decompress_io_ctx { 1485 u32 magic; /* magic number to indicate page is compressed */ 1486 struct inode *inode; /* inode the context belong to */ 1487 pgoff_t cluster_idx; /* cluster index number */ 1488 unsigned int cluster_size; /* page count in cluster */ 1489 unsigned int log_cluster_size; /* log of cluster size */ 1490 struct page **rpages; /* pages store raw data in cluster */ 1491 unsigned int nr_rpages; /* total page number in rpages */ 1492 struct page **cpages; /* pages store compressed data in cluster */ 1493 unsigned int nr_cpages; /* total page number in cpages */ 1494 struct page **tpages; /* temp pages to pad holes in cluster */ 1495 void *rbuf; /* virtual mapped address on rpages */ 1496 struct compress_data *cbuf; /* virtual mapped address on cpages */ 1497 size_t rlen; /* valid data length in rbuf */ 1498 size_t clen; /* valid data length in cbuf */ 1499 1500 /* 1501 * The number of compressed pages remaining to be read in this cluster. 1502 * This is initially nr_cpages. It is decremented by 1 each time a page 1503 * has been read (or failed to be read). When it reaches 0, the cluster 1504 * is decompressed (or an error is reported). 1505 * 1506 * If an error occurs before all the pages have been submitted for I/O, 1507 * then this will never reach 0. In this case the I/O submitter is 1508 * responsible for calling f2fs_decompress_end_io() instead. 1509 */ 1510 atomic_t remaining_pages; 1511 1512 /* 1513 * Number of references to this decompress_io_ctx. 1514 * 1515 * One reference is held for I/O completion. This reference is dropped 1516 * after the pagecache pages are updated and unlocked -- either after 1517 * decompression (and verity if enabled), or after an error. 1518 * 1519 * In addition, each compressed page holds a reference while it is in a 1520 * bio. These references are necessary prevent compressed pages from 1521 * being freed while they are still in a bio. 1522 */ 1523 refcount_t refcnt; 1524 1525 bool failed; /* IO error occurred before decompression? */ 1526 bool need_verity; /* need fs-verity verification after decompression? */ 1527 void *private; /* payload buffer for specified decompression algorithm */ 1528 void *private2; /* extra payload buffer */ 1529 struct work_struct verity_work; /* work to verify the decompressed pages */ 1530 struct work_struct free_work; /* work for late free this structure itself */ 1531 }; 1532 1533 #define NULL_CLUSTER ((unsigned int)(~0)) 1534 #define MIN_COMPRESS_LOG_SIZE 2 1535 #define MAX_COMPRESS_LOG_SIZE 8 1536 #define MAX_COMPRESS_WINDOW_SIZE(log_size) ((PAGE_SIZE) << (log_size)) 1537 1538 struct f2fs_sb_info { 1539 struct super_block *sb; /* pointer to VFS super block */ 1540 struct proc_dir_entry *s_proc; /* proc entry */ 1541 struct f2fs_super_block *raw_super; /* raw super block pointer */ 1542 struct f2fs_rwsem sb_lock; /* lock for raw super block */ 1543 int valid_super_block; /* valid super block no */ 1544 unsigned long s_flag; /* flags for sbi */ 1545 struct mutex writepages; /* mutex for writepages() */ 1546 1547 #ifdef CONFIG_BLK_DEV_ZONED 1548 unsigned int blocks_per_blkz; /* F2FS blocks per zone */ 1549 #endif 1550 1551 /* for node-related operations */ 1552 struct f2fs_nm_info *nm_info; /* node manager */ 1553 struct inode *node_inode; /* cache node blocks */ 1554 1555 /* for segment-related operations */ 1556 struct f2fs_sm_info *sm_info; /* segment manager */ 1557 1558 /* for bio operations */ 1559 struct f2fs_bio_info *write_io[NR_PAGE_TYPE]; /* for write bios */ 1560 /* keep migration IO order for LFS mode */ 1561 struct f2fs_rwsem io_order_lock; 1562 pgoff_t page_eio_ofs[NR_PAGE_TYPE]; /* EIO page offset */ 1563 int page_eio_cnt[NR_PAGE_TYPE]; /* EIO count */ 1564 1565 /* for checkpoint */ 1566 struct f2fs_checkpoint *ckpt; /* raw checkpoint pointer */ 1567 int cur_cp_pack; /* remain current cp pack */ 1568 spinlock_t cp_lock; /* for flag in ckpt */ 1569 struct inode *meta_inode; /* cache meta blocks */ 1570 struct f2fs_rwsem cp_global_sem; /* checkpoint procedure lock */ 1571 struct f2fs_rwsem cp_rwsem; /* blocking FS operations */ 1572 struct f2fs_rwsem node_write; /* locking node writes */ 1573 struct f2fs_rwsem node_change; /* locking node change */ 1574 wait_queue_head_t cp_wait; 1575 unsigned long last_time[MAX_TIME]; /* to store time in jiffies */ 1576 long interval_time[MAX_TIME]; /* to store thresholds */ 1577 struct ckpt_req_control cprc_info; /* for checkpoint request control */ 1578 1579 struct inode_management im[MAX_INO_ENTRY]; /* manage inode cache */ 1580 1581 spinlock_t fsync_node_lock; /* for node entry lock */ 1582 struct list_head fsync_node_list; /* node list head */ 1583 unsigned int fsync_seg_id; /* sequence id */ 1584 unsigned int fsync_node_num; /* number of node entries */ 1585 1586 /* for orphan inode, use 0'th array */ 1587 unsigned int max_orphans; /* max orphan inodes */ 1588 1589 /* for inode management */ 1590 struct list_head inode_list[NR_INODE_TYPE]; /* dirty inode list */ 1591 spinlock_t inode_lock[NR_INODE_TYPE]; /* for dirty inode list lock */ 1592 struct mutex flush_lock; /* for flush exclusion */ 1593 1594 /* for extent tree cache */ 1595 struct extent_tree_info extent_tree[NR_EXTENT_CACHES]; 1596 atomic64_t allocated_data_blocks; /* for block age extent_cache */ 1597 1598 /* The threshold used for hot and warm data seperation*/ 1599 unsigned int hot_data_age_threshold; 1600 unsigned int warm_data_age_threshold; 1601 unsigned int last_age_weight; 1602 1603 /* basic filesystem units */ 1604 unsigned int log_sectors_per_block; /* log2 sectors per block */ 1605 unsigned int log_blocksize; /* log2 block size */ 1606 unsigned int blocksize; /* block size */ 1607 unsigned int root_ino_num; /* root inode number*/ 1608 unsigned int node_ino_num; /* node inode number*/ 1609 unsigned int meta_ino_num; /* meta inode number*/ 1610 unsigned int log_blocks_per_seg; /* log2 blocks per segment */ 1611 unsigned int blocks_per_seg; /* blocks per segment */ 1612 unsigned int unusable_blocks_per_sec; /* unusable blocks per section */ 1613 unsigned int segs_per_sec; /* segments per section */ 1614 unsigned int secs_per_zone; /* sections per zone */ 1615 unsigned int total_sections; /* total section count */ 1616 unsigned int total_node_count; /* total node block count */ 1617 unsigned int total_valid_node_count; /* valid node block count */ 1618 int dir_level; /* directory level */ 1619 bool readdir_ra; /* readahead inode in readdir */ 1620 u64 max_io_bytes; /* max io bytes to merge IOs */ 1621 1622 block_t user_block_count; /* # of user blocks */ 1623 block_t total_valid_block_count; /* # of valid blocks */ 1624 block_t discard_blks; /* discard command candidats */ 1625 block_t last_valid_block_count; /* for recovery */ 1626 block_t reserved_blocks; /* configurable reserved blocks */ 1627 block_t current_reserved_blocks; /* current reserved blocks */ 1628 1629 /* Additional tracking for no checkpoint mode */ 1630 block_t unusable_block_count; /* # of blocks saved by last cp */ 1631 1632 unsigned int nquota_files; /* # of quota sysfile */ 1633 struct f2fs_rwsem quota_sem; /* blocking cp for flags */ 1634 1635 /* # of pages, see count_type */ 1636 atomic_t nr_pages[NR_COUNT_TYPE]; 1637 /* # of allocated blocks */ 1638 struct percpu_counter alloc_valid_block_count; 1639 /* # of node block writes as roll forward recovery */ 1640 struct percpu_counter rf_node_block_count; 1641 1642 /* writeback control */ 1643 atomic_t wb_sync_req[META]; /* count # of WB_SYNC threads */ 1644 1645 /* valid inode count */ 1646 struct percpu_counter total_valid_inode_count; 1647 1648 struct f2fs_mount_info mount_opt; /* mount options */ 1649 1650 /* for cleaning operations */ 1651 struct f2fs_rwsem gc_lock; /* 1652 * semaphore for GC, avoid 1653 * race between GC and GC or CP 1654 */ 1655 struct f2fs_gc_kthread *gc_thread; /* GC thread */ 1656 struct atgc_management am; /* atgc management */ 1657 unsigned int cur_victim_sec; /* current victim section num */ 1658 unsigned int gc_mode; /* current GC state */ 1659 unsigned int next_victim_seg[2]; /* next segment in victim section */ 1660 spinlock_t gc_remaining_trials_lock; 1661 /* remaining trial count for GC_URGENT_* and GC_IDLE_* */ 1662 unsigned int gc_remaining_trials; 1663 1664 /* for skip statistic */ 1665 unsigned long long skipped_gc_rwsem; /* FG_GC only */ 1666 1667 /* threshold for gc trials on pinned files */ 1668 u64 gc_pin_file_threshold; 1669 struct f2fs_rwsem pin_sem; 1670 1671 /* maximum # of trials to find a victim segment for SSR and GC */ 1672 unsigned int max_victim_search; 1673 /* migration granularity of garbage collection, unit: segment */ 1674 unsigned int migration_granularity; 1675 1676 /* 1677 * for stat information. 1678 * one is for the LFS mode, and the other is for the SSR mode. 1679 */ 1680 #ifdef CONFIG_F2FS_STAT_FS 1681 struct f2fs_stat_info *stat_info; /* FS status information */ 1682 atomic_t meta_count[META_MAX]; /* # of meta blocks */ 1683 unsigned int segment_count[2]; /* # of allocated segments */ 1684 unsigned int block_count[2]; /* # of allocated blocks */ 1685 atomic_t inplace_count; /* # of inplace update */ 1686 /* # of lookup extent cache */ 1687 atomic64_t total_hit_ext[NR_EXTENT_CACHES]; 1688 /* # of hit rbtree extent node */ 1689 atomic64_t read_hit_rbtree[NR_EXTENT_CACHES]; 1690 /* # of hit cached extent node */ 1691 atomic64_t read_hit_cached[NR_EXTENT_CACHES]; 1692 /* # of hit largest extent node in read extent cache */ 1693 atomic64_t read_hit_largest; 1694 atomic_t inline_xattr; /* # of inline_xattr inodes */ 1695 atomic_t inline_inode; /* # of inline_data inodes */ 1696 atomic_t inline_dir; /* # of inline_dentry inodes */ 1697 atomic_t compr_inode; /* # of compressed inodes */ 1698 atomic64_t compr_blocks; /* # of compressed blocks */ 1699 atomic_t swapfile_inode; /* # of swapfile inodes */ 1700 atomic_t atomic_files; /* # of opened atomic file */ 1701 atomic_t max_aw_cnt; /* max # of atomic writes */ 1702 unsigned int io_skip_bggc; /* skip background gc for in-flight IO */ 1703 unsigned int other_skip_bggc; /* skip background gc for other reasons */ 1704 unsigned int ndirty_inode[NR_INODE_TYPE]; /* # of dirty inodes */ 1705 atomic_t cp_call_count[MAX_CALL_TYPE]; /* # of cp call */ 1706 #endif 1707 spinlock_t stat_lock; /* lock for stat operations */ 1708 1709 /* to attach REQ_META|REQ_FUA flags */ 1710 unsigned int data_io_flag; 1711 unsigned int node_io_flag; 1712 1713 /* For sysfs support */ 1714 struct kobject s_kobj; /* /sys/fs/f2fs/<devname> */ 1715 struct completion s_kobj_unregister; 1716 1717 struct kobject s_stat_kobj; /* /sys/fs/f2fs/<devname>/stat */ 1718 struct completion s_stat_kobj_unregister; 1719 1720 struct kobject s_feature_list_kobj; /* /sys/fs/f2fs/<devname>/feature_list */ 1721 struct completion s_feature_list_kobj_unregister; 1722 1723 /* For shrinker support */ 1724 struct list_head s_list; 1725 struct mutex umount_mutex; 1726 unsigned int shrinker_run_no; 1727 1728 /* For multi devices */ 1729 int s_ndevs; /* number of devices */ 1730 struct f2fs_dev_info *devs; /* for device list */ 1731 unsigned int dirty_device; /* for checkpoint data flush */ 1732 spinlock_t dev_lock; /* protect dirty_device */ 1733 bool aligned_blksize; /* all devices has the same logical blksize */ 1734 1735 /* For write statistics */ 1736 u64 sectors_written_start; 1737 u64 kbytes_written; 1738 1739 /* Reference to checksum algorithm driver via cryptoapi */ 1740 struct crypto_shash *s_chksum_driver; 1741 1742 /* Precomputed FS UUID checksum for seeding other checksums */ 1743 __u32 s_chksum_seed; 1744 1745 struct workqueue_struct *post_read_wq; /* post read workqueue */ 1746 1747 /* 1748 * If we are in irq context, let's update error information into 1749 * on-disk superblock in the work. 1750 */ 1751 struct work_struct s_error_work; 1752 unsigned char errors[MAX_F2FS_ERRORS]; /* error flags */ 1753 unsigned char stop_reason[MAX_STOP_REASON]; /* stop reason */ 1754 spinlock_t error_lock; /* protect errors/stop_reason array */ 1755 bool error_dirty; /* errors of sb is dirty */ 1756 1757 struct kmem_cache *inline_xattr_slab; /* inline xattr entry */ 1758 unsigned int inline_xattr_slab_size; /* default inline xattr slab size */ 1759 1760 /* For reclaimed segs statistics per each GC mode */ 1761 unsigned int gc_segment_mode; /* GC state for reclaimed segments */ 1762 unsigned int gc_reclaimed_segs[MAX_GC_MODE]; /* Reclaimed segs for each mode */ 1763 1764 unsigned long seq_file_ra_mul; /* multiplier for ra_pages of seq. files in fadvise */ 1765 1766 int max_fragment_chunk; /* max chunk size for block fragmentation mode */ 1767 int max_fragment_hole; /* max hole size for block fragmentation mode */ 1768 1769 /* For atomic write statistics */ 1770 atomic64_t current_atomic_write; 1771 s64 peak_atomic_write; 1772 u64 committed_atomic_block; 1773 u64 revoked_atomic_block; 1774 1775 #ifdef CONFIG_F2FS_FS_COMPRESSION 1776 struct kmem_cache *page_array_slab; /* page array entry */ 1777 unsigned int page_array_slab_size; /* default page array slab size */ 1778 1779 /* For runtime compression statistics */ 1780 u64 compr_written_block; 1781 u64 compr_saved_block; 1782 u32 compr_new_inode; 1783 1784 /* For compressed block cache */ 1785 struct inode *compress_inode; /* cache compressed blocks */ 1786 unsigned int compress_percent; /* cache page percentage */ 1787 unsigned int compress_watermark; /* cache page watermark */ 1788 atomic_t compress_page_hit; /* cache hit count */ 1789 #endif 1790 1791 #ifdef CONFIG_F2FS_IOSTAT 1792 /* For app/fs IO statistics */ 1793 spinlock_t iostat_lock; 1794 unsigned long long iostat_count[NR_IO_TYPE]; 1795 unsigned long long iostat_bytes[NR_IO_TYPE]; 1796 unsigned long long prev_iostat_bytes[NR_IO_TYPE]; 1797 bool iostat_enable; 1798 unsigned long iostat_next_period; 1799 unsigned int iostat_period_ms; 1800 1801 /* For io latency related statistics info in one iostat period */ 1802 spinlock_t iostat_lat_lock; 1803 struct iostat_lat_info *iostat_io_lat; 1804 #endif 1805 }; 1806 1807 /* Definitions to access f2fs_sb_info */ 1808 #define BLKS_PER_SEG(sbi) \ 1809 ((sbi)->blocks_per_seg) 1810 #define BLKS_PER_SEC(sbi) \ 1811 ((sbi)->segs_per_sec << (sbi)->log_blocks_per_seg) 1812 #define SEGS_PER_SEC(sbi) \ 1813 ((sbi)->segs_per_sec) 1814 1815 __printf(3, 4) 1816 void f2fs_printk(struct f2fs_sb_info *sbi, bool limit_rate, const char *fmt, ...); 1817 1818 #define f2fs_err(sbi, fmt, ...) \ 1819 f2fs_printk(sbi, false, KERN_ERR fmt, ##__VA_ARGS__) 1820 #define f2fs_warn(sbi, fmt, ...) \ 1821 f2fs_printk(sbi, false, KERN_WARNING fmt, ##__VA_ARGS__) 1822 #define f2fs_notice(sbi, fmt, ...) \ 1823 f2fs_printk(sbi, false, KERN_NOTICE fmt, ##__VA_ARGS__) 1824 #define f2fs_info(sbi, fmt, ...) \ 1825 f2fs_printk(sbi, false, KERN_INFO fmt, ##__VA_ARGS__) 1826 #define f2fs_debug(sbi, fmt, ...) \ 1827 f2fs_printk(sbi, false, KERN_DEBUG fmt, ##__VA_ARGS__) 1828 1829 #define f2fs_err_ratelimited(sbi, fmt, ...) \ 1830 f2fs_printk(sbi, true, KERN_ERR fmt, ##__VA_ARGS__) 1831 #define f2fs_warn_ratelimited(sbi, fmt, ...) \ 1832 f2fs_printk(sbi, true, KERN_WARNING fmt, ##__VA_ARGS__) 1833 #define f2fs_info_ratelimited(sbi, fmt, ...) \ 1834 f2fs_printk(sbi, true, KERN_INFO fmt, ##__VA_ARGS__) 1835 1836 #ifdef CONFIG_F2FS_FAULT_INJECTION 1837 #define time_to_inject(sbi, type) __time_to_inject(sbi, type, __func__, \ 1838 __builtin_return_address(0)) 1839 static inline bool __time_to_inject(struct f2fs_sb_info *sbi, int type, 1840 const char *func, const char *parent_func) 1841 { 1842 struct f2fs_fault_info *ffi = &F2FS_OPTION(sbi).fault_info; 1843 1844 if (!ffi->inject_rate) 1845 return false; 1846 1847 if (!IS_FAULT_SET(ffi, type)) 1848 return false; 1849 1850 atomic_inc(&ffi->inject_ops); 1851 if (atomic_read(&ffi->inject_ops) >= ffi->inject_rate) { 1852 atomic_set(&ffi->inject_ops, 0); 1853 f2fs_info_ratelimited(sbi, "inject %s in %s of %pS", 1854 f2fs_fault_name[type], func, parent_func); 1855 return true; 1856 } 1857 return false; 1858 } 1859 #else 1860 static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type) 1861 { 1862 return false; 1863 } 1864 #endif 1865 1866 /* 1867 * Test if the mounted volume is a multi-device volume. 1868 * - For a single regular disk volume, sbi->s_ndevs is 0. 1869 * - For a single zoned disk volume, sbi->s_ndevs is 1. 1870 * - For a multi-device volume, sbi->s_ndevs is always 2 or more. 1871 */ 1872 static inline bool f2fs_is_multi_device(struct f2fs_sb_info *sbi) 1873 { 1874 return sbi->s_ndevs > 1; 1875 } 1876 1877 static inline void f2fs_update_time(struct f2fs_sb_info *sbi, int type) 1878 { 1879 unsigned long now = jiffies; 1880 1881 sbi->last_time[type] = now; 1882 1883 /* DISCARD_TIME and GC_TIME are based on REQ_TIME */ 1884 if (type == REQ_TIME) { 1885 sbi->last_time[DISCARD_TIME] = now; 1886 sbi->last_time[GC_TIME] = now; 1887 } 1888 } 1889 1890 static inline bool f2fs_time_over(struct f2fs_sb_info *sbi, int type) 1891 { 1892 unsigned long interval = sbi->interval_time[type] * HZ; 1893 1894 return time_after(jiffies, sbi->last_time[type] + interval); 1895 } 1896 1897 static inline unsigned int f2fs_time_to_wait(struct f2fs_sb_info *sbi, 1898 int type) 1899 { 1900 unsigned long interval = sbi->interval_time[type] * HZ; 1901 unsigned int wait_ms = 0; 1902 long delta; 1903 1904 delta = (sbi->last_time[type] + interval) - jiffies; 1905 if (delta > 0) 1906 wait_ms = jiffies_to_msecs(delta); 1907 1908 return wait_ms; 1909 } 1910 1911 /* 1912 * Inline functions 1913 */ 1914 static inline u32 __f2fs_crc32(struct f2fs_sb_info *sbi, u32 crc, 1915 const void *address, unsigned int length) 1916 { 1917 struct { 1918 struct shash_desc shash; 1919 char ctx[4]; 1920 } desc; 1921 int err; 1922 1923 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver) != sizeof(desc.ctx)); 1924 1925 desc.shash.tfm = sbi->s_chksum_driver; 1926 *(u32 *)desc.ctx = crc; 1927 1928 err = crypto_shash_update(&desc.shash, address, length); 1929 BUG_ON(err); 1930 1931 return *(u32 *)desc.ctx; 1932 } 1933 1934 static inline u32 f2fs_crc32(struct f2fs_sb_info *sbi, const void *address, 1935 unsigned int length) 1936 { 1937 return __f2fs_crc32(sbi, F2FS_SUPER_MAGIC, address, length); 1938 } 1939 1940 static inline bool f2fs_crc_valid(struct f2fs_sb_info *sbi, __u32 blk_crc, 1941 void *buf, size_t buf_size) 1942 { 1943 return f2fs_crc32(sbi, buf, buf_size) == blk_crc; 1944 } 1945 1946 static inline u32 f2fs_chksum(struct f2fs_sb_info *sbi, u32 crc, 1947 const void *address, unsigned int length) 1948 { 1949 return __f2fs_crc32(sbi, crc, address, length); 1950 } 1951 1952 static inline struct f2fs_inode_info *F2FS_I(struct inode *inode) 1953 { 1954 return container_of(inode, struct f2fs_inode_info, vfs_inode); 1955 } 1956 1957 static inline struct f2fs_sb_info *F2FS_SB(struct super_block *sb) 1958 { 1959 return sb->s_fs_info; 1960 } 1961 1962 static inline struct f2fs_sb_info *F2FS_I_SB(struct inode *inode) 1963 { 1964 return F2FS_SB(inode->i_sb); 1965 } 1966 1967 static inline struct f2fs_sb_info *F2FS_M_SB(struct address_space *mapping) 1968 { 1969 return F2FS_I_SB(mapping->host); 1970 } 1971 1972 static inline struct f2fs_sb_info *F2FS_P_SB(struct page *page) 1973 { 1974 return F2FS_M_SB(page_file_mapping(page)); 1975 } 1976 1977 static inline struct f2fs_super_block *F2FS_RAW_SUPER(struct f2fs_sb_info *sbi) 1978 { 1979 return (struct f2fs_super_block *)(sbi->raw_super); 1980 } 1981 1982 static inline struct f2fs_checkpoint *F2FS_CKPT(struct f2fs_sb_info *sbi) 1983 { 1984 return (struct f2fs_checkpoint *)(sbi->ckpt); 1985 } 1986 1987 static inline struct f2fs_node *F2FS_NODE(struct page *page) 1988 { 1989 return (struct f2fs_node *)page_address(page); 1990 } 1991 1992 static inline struct f2fs_inode *F2FS_INODE(struct page *page) 1993 { 1994 return &((struct f2fs_node *)page_address(page))->i; 1995 } 1996 1997 static inline struct f2fs_nm_info *NM_I(struct f2fs_sb_info *sbi) 1998 { 1999 return (struct f2fs_nm_info *)(sbi->nm_info); 2000 } 2001 2002 static inline struct f2fs_sm_info *SM_I(struct f2fs_sb_info *sbi) 2003 { 2004 return (struct f2fs_sm_info *)(sbi->sm_info); 2005 } 2006 2007 static inline struct sit_info *SIT_I(struct f2fs_sb_info *sbi) 2008 { 2009 return (struct sit_info *)(SM_I(sbi)->sit_info); 2010 } 2011 2012 static inline struct free_segmap_info *FREE_I(struct f2fs_sb_info *sbi) 2013 { 2014 return (struct free_segmap_info *)(SM_I(sbi)->free_info); 2015 } 2016 2017 static inline struct dirty_seglist_info *DIRTY_I(struct f2fs_sb_info *sbi) 2018 { 2019 return (struct dirty_seglist_info *)(SM_I(sbi)->dirty_info); 2020 } 2021 2022 static inline struct address_space *META_MAPPING(struct f2fs_sb_info *sbi) 2023 { 2024 return sbi->meta_inode->i_mapping; 2025 } 2026 2027 static inline struct address_space *NODE_MAPPING(struct f2fs_sb_info *sbi) 2028 { 2029 return sbi->node_inode->i_mapping; 2030 } 2031 2032 static inline bool is_sbi_flag_set(struct f2fs_sb_info *sbi, unsigned int type) 2033 { 2034 return test_bit(type, &sbi->s_flag); 2035 } 2036 2037 static inline void set_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type) 2038 { 2039 set_bit(type, &sbi->s_flag); 2040 } 2041 2042 static inline void clear_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type) 2043 { 2044 clear_bit(type, &sbi->s_flag); 2045 } 2046 2047 static inline unsigned long long cur_cp_version(struct f2fs_checkpoint *cp) 2048 { 2049 return le64_to_cpu(cp->checkpoint_ver); 2050 } 2051 2052 static inline unsigned long f2fs_qf_ino(struct super_block *sb, int type) 2053 { 2054 if (type < F2FS_MAX_QUOTAS) 2055 return le32_to_cpu(F2FS_SB(sb)->raw_super->qf_ino[type]); 2056 return 0; 2057 } 2058 2059 static inline __u64 cur_cp_crc(struct f2fs_checkpoint *cp) 2060 { 2061 size_t crc_offset = le32_to_cpu(cp->checksum_offset); 2062 return le32_to_cpu(*((__le32 *)((unsigned char *)cp + crc_offset))); 2063 } 2064 2065 static inline bool __is_set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 2066 { 2067 unsigned int ckpt_flags = le32_to_cpu(cp->ckpt_flags); 2068 2069 return ckpt_flags & f; 2070 } 2071 2072 static inline bool is_set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 2073 { 2074 return __is_set_ckpt_flags(F2FS_CKPT(sbi), f); 2075 } 2076 2077 static inline void __set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 2078 { 2079 unsigned int ckpt_flags; 2080 2081 ckpt_flags = le32_to_cpu(cp->ckpt_flags); 2082 ckpt_flags |= f; 2083 cp->ckpt_flags = cpu_to_le32(ckpt_flags); 2084 } 2085 2086 static inline void set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 2087 { 2088 unsigned long flags; 2089 2090 spin_lock_irqsave(&sbi->cp_lock, flags); 2091 __set_ckpt_flags(F2FS_CKPT(sbi), f); 2092 spin_unlock_irqrestore(&sbi->cp_lock, flags); 2093 } 2094 2095 static inline void __clear_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 2096 { 2097 unsigned int ckpt_flags; 2098 2099 ckpt_flags = le32_to_cpu(cp->ckpt_flags); 2100 ckpt_flags &= (~f); 2101 cp->ckpt_flags = cpu_to_le32(ckpt_flags); 2102 } 2103 2104 static inline void clear_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 2105 { 2106 unsigned long flags; 2107 2108 spin_lock_irqsave(&sbi->cp_lock, flags); 2109 __clear_ckpt_flags(F2FS_CKPT(sbi), f); 2110 spin_unlock_irqrestore(&sbi->cp_lock, flags); 2111 } 2112 2113 #define init_f2fs_rwsem(sem) \ 2114 do { \ 2115 static struct lock_class_key __key; \ 2116 \ 2117 __init_f2fs_rwsem((sem), #sem, &__key); \ 2118 } while (0) 2119 2120 static inline void __init_f2fs_rwsem(struct f2fs_rwsem *sem, 2121 const char *sem_name, struct lock_class_key *key) 2122 { 2123 __init_rwsem(&sem->internal_rwsem, sem_name, key); 2124 #ifdef CONFIG_F2FS_UNFAIR_RWSEM 2125 init_waitqueue_head(&sem->read_waiters); 2126 #endif 2127 } 2128 2129 static inline int f2fs_rwsem_is_locked(struct f2fs_rwsem *sem) 2130 { 2131 return rwsem_is_locked(&sem->internal_rwsem); 2132 } 2133 2134 static inline int f2fs_rwsem_is_contended(struct f2fs_rwsem *sem) 2135 { 2136 return rwsem_is_contended(&sem->internal_rwsem); 2137 } 2138 2139 static inline void f2fs_down_read(struct f2fs_rwsem *sem) 2140 { 2141 #ifdef CONFIG_F2FS_UNFAIR_RWSEM 2142 wait_event(sem->read_waiters, down_read_trylock(&sem->internal_rwsem)); 2143 #else 2144 down_read(&sem->internal_rwsem); 2145 #endif 2146 } 2147 2148 static inline int f2fs_down_read_trylock(struct f2fs_rwsem *sem) 2149 { 2150 return down_read_trylock(&sem->internal_rwsem); 2151 } 2152 2153 static inline void f2fs_up_read(struct f2fs_rwsem *sem) 2154 { 2155 up_read(&sem->internal_rwsem); 2156 } 2157 2158 static inline void f2fs_down_write(struct f2fs_rwsem *sem) 2159 { 2160 down_write(&sem->internal_rwsem); 2161 } 2162 2163 #ifdef CONFIG_DEBUG_LOCK_ALLOC 2164 static inline void f2fs_down_read_nested(struct f2fs_rwsem *sem, int subclass) 2165 { 2166 down_read_nested(&sem->internal_rwsem, subclass); 2167 } 2168 2169 static inline void f2fs_down_write_nested(struct f2fs_rwsem *sem, int subclass) 2170 { 2171 down_write_nested(&sem->internal_rwsem, subclass); 2172 } 2173 #else 2174 #define f2fs_down_read_nested(sem, subclass) f2fs_down_read(sem) 2175 #define f2fs_down_write_nested(sem, subclass) f2fs_down_write(sem) 2176 #endif 2177 2178 static inline int f2fs_down_write_trylock(struct f2fs_rwsem *sem) 2179 { 2180 return down_write_trylock(&sem->internal_rwsem); 2181 } 2182 2183 static inline void f2fs_up_write(struct f2fs_rwsem *sem) 2184 { 2185 up_write(&sem->internal_rwsem); 2186 #ifdef CONFIG_F2FS_UNFAIR_RWSEM 2187 wake_up_all(&sem->read_waiters); 2188 #endif 2189 } 2190 2191 static inline void f2fs_lock_op(struct f2fs_sb_info *sbi) 2192 { 2193 f2fs_down_read(&sbi->cp_rwsem); 2194 } 2195 2196 static inline int f2fs_trylock_op(struct f2fs_sb_info *sbi) 2197 { 2198 if (time_to_inject(sbi, FAULT_LOCK_OP)) 2199 return 0; 2200 return f2fs_down_read_trylock(&sbi->cp_rwsem); 2201 } 2202 2203 static inline void f2fs_unlock_op(struct f2fs_sb_info *sbi) 2204 { 2205 f2fs_up_read(&sbi->cp_rwsem); 2206 } 2207 2208 static inline void f2fs_lock_all(struct f2fs_sb_info *sbi) 2209 { 2210 f2fs_down_write(&sbi->cp_rwsem); 2211 } 2212 2213 static inline void f2fs_unlock_all(struct f2fs_sb_info *sbi) 2214 { 2215 f2fs_up_write(&sbi->cp_rwsem); 2216 } 2217 2218 static inline int __get_cp_reason(struct f2fs_sb_info *sbi) 2219 { 2220 int reason = CP_SYNC; 2221 2222 if (test_opt(sbi, FASTBOOT)) 2223 reason = CP_FASTBOOT; 2224 if (is_sbi_flag_set(sbi, SBI_IS_CLOSE)) 2225 reason = CP_UMOUNT; 2226 return reason; 2227 } 2228 2229 static inline bool __remain_node_summaries(int reason) 2230 { 2231 return (reason & (CP_UMOUNT | CP_FASTBOOT)); 2232 } 2233 2234 static inline bool __exist_node_summaries(struct f2fs_sb_info *sbi) 2235 { 2236 return (is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG) || 2237 is_set_ckpt_flags(sbi, CP_FASTBOOT_FLAG)); 2238 } 2239 2240 /* 2241 * Check whether the inode has blocks or not 2242 */ 2243 static inline int F2FS_HAS_BLOCKS(struct inode *inode) 2244 { 2245 block_t xattr_block = F2FS_I(inode)->i_xattr_nid ? 1 : 0; 2246 2247 return (inode->i_blocks >> F2FS_LOG_SECTORS_PER_BLOCK) > xattr_block; 2248 } 2249 2250 static inline bool f2fs_has_xattr_block(unsigned int ofs) 2251 { 2252 return ofs == XATTR_NODE_OFFSET; 2253 } 2254 2255 static inline bool __allow_reserved_blocks(struct f2fs_sb_info *sbi, 2256 struct inode *inode, bool cap) 2257 { 2258 if (!inode) 2259 return true; 2260 if (!test_opt(sbi, RESERVE_ROOT)) 2261 return false; 2262 if (IS_NOQUOTA(inode)) 2263 return true; 2264 if (uid_eq(F2FS_OPTION(sbi).s_resuid, current_fsuid())) 2265 return true; 2266 if (!gid_eq(F2FS_OPTION(sbi).s_resgid, GLOBAL_ROOT_GID) && 2267 in_group_p(F2FS_OPTION(sbi).s_resgid)) 2268 return true; 2269 if (cap && capable(CAP_SYS_RESOURCE)) 2270 return true; 2271 return false; 2272 } 2273 2274 static inline void f2fs_i_blocks_write(struct inode *, block_t, bool, bool); 2275 static inline int inc_valid_block_count(struct f2fs_sb_info *sbi, 2276 struct inode *inode, blkcnt_t *count, bool partial) 2277 { 2278 blkcnt_t diff = 0, release = 0; 2279 block_t avail_user_block_count; 2280 int ret; 2281 2282 ret = dquot_reserve_block(inode, *count); 2283 if (ret) 2284 return ret; 2285 2286 if (time_to_inject(sbi, FAULT_BLOCK)) { 2287 release = *count; 2288 goto release_quota; 2289 } 2290 2291 /* 2292 * let's increase this in prior to actual block count change in order 2293 * for f2fs_sync_file to avoid data races when deciding checkpoint. 2294 */ 2295 percpu_counter_add(&sbi->alloc_valid_block_count, (*count)); 2296 2297 spin_lock(&sbi->stat_lock); 2298 sbi->total_valid_block_count += (block_t)(*count); 2299 avail_user_block_count = sbi->user_block_count - 2300 sbi->current_reserved_blocks; 2301 2302 if (!__allow_reserved_blocks(sbi, inode, true)) 2303 avail_user_block_count -= F2FS_OPTION(sbi).root_reserved_blocks; 2304 2305 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) { 2306 if (avail_user_block_count > sbi->unusable_block_count) 2307 avail_user_block_count -= sbi->unusable_block_count; 2308 else 2309 avail_user_block_count = 0; 2310 } 2311 if (unlikely(sbi->total_valid_block_count > avail_user_block_count)) { 2312 if (!partial) { 2313 spin_unlock(&sbi->stat_lock); 2314 goto enospc; 2315 } 2316 2317 diff = sbi->total_valid_block_count - avail_user_block_count; 2318 if (diff > *count) 2319 diff = *count; 2320 *count -= diff; 2321 release = diff; 2322 sbi->total_valid_block_count -= diff; 2323 if (!*count) { 2324 spin_unlock(&sbi->stat_lock); 2325 goto enospc; 2326 } 2327 } 2328 spin_unlock(&sbi->stat_lock); 2329 2330 if (unlikely(release)) { 2331 percpu_counter_sub(&sbi->alloc_valid_block_count, release); 2332 dquot_release_reservation_block(inode, release); 2333 } 2334 f2fs_i_blocks_write(inode, *count, true, true); 2335 return 0; 2336 2337 enospc: 2338 percpu_counter_sub(&sbi->alloc_valid_block_count, release); 2339 release_quota: 2340 dquot_release_reservation_block(inode, release); 2341 return -ENOSPC; 2342 } 2343 2344 #define PAGE_PRIVATE_GET_FUNC(name, flagname) \ 2345 static inline bool page_private_##name(struct page *page) \ 2346 { \ 2347 return PagePrivate(page) && \ 2348 test_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)) && \ 2349 test_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \ 2350 } 2351 2352 #define PAGE_PRIVATE_SET_FUNC(name, flagname) \ 2353 static inline void set_page_private_##name(struct page *page) \ 2354 { \ 2355 if (!PagePrivate(page)) \ 2356 attach_page_private(page, (void *)0); \ 2357 set_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)); \ 2358 set_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \ 2359 } 2360 2361 #define PAGE_PRIVATE_CLEAR_FUNC(name, flagname) \ 2362 static inline void clear_page_private_##name(struct page *page) \ 2363 { \ 2364 clear_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \ 2365 if (page_private(page) == BIT(PAGE_PRIVATE_NOT_POINTER)) \ 2366 detach_page_private(page); \ 2367 } 2368 2369 PAGE_PRIVATE_GET_FUNC(nonpointer, NOT_POINTER); 2370 PAGE_PRIVATE_GET_FUNC(inline, INLINE_INODE); 2371 PAGE_PRIVATE_GET_FUNC(gcing, ONGOING_MIGRATION); 2372 2373 PAGE_PRIVATE_SET_FUNC(reference, REF_RESOURCE); 2374 PAGE_PRIVATE_SET_FUNC(inline, INLINE_INODE); 2375 PAGE_PRIVATE_SET_FUNC(gcing, ONGOING_MIGRATION); 2376 2377 PAGE_PRIVATE_CLEAR_FUNC(reference, REF_RESOURCE); 2378 PAGE_PRIVATE_CLEAR_FUNC(inline, INLINE_INODE); 2379 PAGE_PRIVATE_CLEAR_FUNC(gcing, ONGOING_MIGRATION); 2380 2381 static inline unsigned long get_page_private_data(struct page *page) 2382 { 2383 unsigned long data = page_private(page); 2384 2385 if (!test_bit(PAGE_PRIVATE_NOT_POINTER, &data)) 2386 return 0; 2387 return data >> PAGE_PRIVATE_MAX; 2388 } 2389 2390 static inline void set_page_private_data(struct page *page, unsigned long data) 2391 { 2392 if (!PagePrivate(page)) 2393 attach_page_private(page, (void *)0); 2394 set_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)); 2395 page_private(page) |= data << PAGE_PRIVATE_MAX; 2396 } 2397 2398 static inline void clear_page_private_data(struct page *page) 2399 { 2400 page_private(page) &= GENMASK(PAGE_PRIVATE_MAX - 1, 0); 2401 if (page_private(page) == BIT(PAGE_PRIVATE_NOT_POINTER)) 2402 detach_page_private(page); 2403 } 2404 2405 static inline void clear_page_private_all(struct page *page) 2406 { 2407 clear_page_private_data(page); 2408 clear_page_private_reference(page); 2409 clear_page_private_gcing(page); 2410 clear_page_private_inline(page); 2411 2412 f2fs_bug_on(F2FS_P_SB(page), page_private(page)); 2413 } 2414 2415 static inline void dec_valid_block_count(struct f2fs_sb_info *sbi, 2416 struct inode *inode, 2417 block_t count) 2418 { 2419 blkcnt_t sectors = count << F2FS_LOG_SECTORS_PER_BLOCK; 2420 2421 spin_lock(&sbi->stat_lock); 2422 f2fs_bug_on(sbi, sbi->total_valid_block_count < (block_t) count); 2423 sbi->total_valid_block_count -= (block_t)count; 2424 if (sbi->reserved_blocks && 2425 sbi->current_reserved_blocks < sbi->reserved_blocks) 2426 sbi->current_reserved_blocks = min(sbi->reserved_blocks, 2427 sbi->current_reserved_blocks + count); 2428 spin_unlock(&sbi->stat_lock); 2429 if (unlikely(inode->i_blocks < sectors)) { 2430 f2fs_warn(sbi, "Inconsistent i_blocks, ino:%lu, iblocks:%llu, sectors:%llu", 2431 inode->i_ino, 2432 (unsigned long long)inode->i_blocks, 2433 (unsigned long long)sectors); 2434 set_sbi_flag(sbi, SBI_NEED_FSCK); 2435 return; 2436 } 2437 f2fs_i_blocks_write(inode, count, false, true); 2438 } 2439 2440 static inline void inc_page_count(struct f2fs_sb_info *sbi, int count_type) 2441 { 2442 atomic_inc(&sbi->nr_pages[count_type]); 2443 2444 if (count_type == F2FS_DIRTY_DENTS || 2445 count_type == F2FS_DIRTY_NODES || 2446 count_type == F2FS_DIRTY_META || 2447 count_type == F2FS_DIRTY_QDATA || 2448 count_type == F2FS_DIRTY_IMETA) 2449 set_sbi_flag(sbi, SBI_IS_DIRTY); 2450 } 2451 2452 static inline void inode_inc_dirty_pages(struct inode *inode) 2453 { 2454 atomic_inc(&F2FS_I(inode)->dirty_pages); 2455 inc_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ? 2456 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA); 2457 if (IS_NOQUOTA(inode)) 2458 inc_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA); 2459 } 2460 2461 static inline void dec_page_count(struct f2fs_sb_info *sbi, int count_type) 2462 { 2463 atomic_dec(&sbi->nr_pages[count_type]); 2464 } 2465 2466 static inline void inode_dec_dirty_pages(struct inode *inode) 2467 { 2468 if (!S_ISDIR(inode->i_mode) && !S_ISREG(inode->i_mode) && 2469 !S_ISLNK(inode->i_mode)) 2470 return; 2471 2472 atomic_dec(&F2FS_I(inode)->dirty_pages); 2473 dec_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ? 2474 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA); 2475 if (IS_NOQUOTA(inode)) 2476 dec_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA); 2477 } 2478 2479 static inline void inc_atomic_write_cnt(struct inode *inode) 2480 { 2481 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 2482 struct f2fs_inode_info *fi = F2FS_I(inode); 2483 u64 current_write; 2484 2485 fi->atomic_write_cnt++; 2486 atomic64_inc(&sbi->current_atomic_write); 2487 current_write = atomic64_read(&sbi->current_atomic_write); 2488 if (current_write > sbi->peak_atomic_write) 2489 sbi->peak_atomic_write = current_write; 2490 } 2491 2492 static inline void release_atomic_write_cnt(struct inode *inode) 2493 { 2494 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 2495 struct f2fs_inode_info *fi = F2FS_I(inode); 2496 2497 atomic64_sub(fi->atomic_write_cnt, &sbi->current_atomic_write); 2498 fi->atomic_write_cnt = 0; 2499 } 2500 2501 static inline s64 get_pages(struct f2fs_sb_info *sbi, int count_type) 2502 { 2503 return atomic_read(&sbi->nr_pages[count_type]); 2504 } 2505 2506 static inline int get_dirty_pages(struct inode *inode) 2507 { 2508 return atomic_read(&F2FS_I(inode)->dirty_pages); 2509 } 2510 2511 static inline int get_blocktype_secs(struct f2fs_sb_info *sbi, int block_type) 2512 { 2513 return div_u64(get_pages(sbi, block_type) + BLKS_PER_SEC(sbi) - 1, 2514 BLKS_PER_SEC(sbi)); 2515 } 2516 2517 static inline block_t valid_user_blocks(struct f2fs_sb_info *sbi) 2518 { 2519 return sbi->total_valid_block_count; 2520 } 2521 2522 static inline block_t discard_blocks(struct f2fs_sb_info *sbi) 2523 { 2524 return sbi->discard_blks; 2525 } 2526 2527 static inline unsigned long __bitmap_size(struct f2fs_sb_info *sbi, int flag) 2528 { 2529 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi); 2530 2531 /* return NAT or SIT bitmap */ 2532 if (flag == NAT_BITMAP) 2533 return le32_to_cpu(ckpt->nat_ver_bitmap_bytesize); 2534 else if (flag == SIT_BITMAP) 2535 return le32_to_cpu(ckpt->sit_ver_bitmap_bytesize); 2536 2537 return 0; 2538 } 2539 2540 static inline block_t __cp_payload(struct f2fs_sb_info *sbi) 2541 { 2542 return le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_payload); 2543 } 2544 2545 static inline void *__bitmap_ptr(struct f2fs_sb_info *sbi, int flag) 2546 { 2547 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi); 2548 void *tmp_ptr = &ckpt->sit_nat_version_bitmap; 2549 int offset; 2550 2551 if (is_set_ckpt_flags(sbi, CP_LARGE_NAT_BITMAP_FLAG)) { 2552 offset = (flag == SIT_BITMAP) ? 2553 le32_to_cpu(ckpt->nat_ver_bitmap_bytesize) : 0; 2554 /* 2555 * if large_nat_bitmap feature is enabled, leave checksum 2556 * protection for all nat/sit bitmaps. 2557 */ 2558 return tmp_ptr + offset + sizeof(__le32); 2559 } 2560 2561 if (__cp_payload(sbi) > 0) { 2562 if (flag == NAT_BITMAP) 2563 return tmp_ptr; 2564 else 2565 return (unsigned char *)ckpt + F2FS_BLKSIZE; 2566 } else { 2567 offset = (flag == NAT_BITMAP) ? 2568 le32_to_cpu(ckpt->sit_ver_bitmap_bytesize) : 0; 2569 return tmp_ptr + offset; 2570 } 2571 } 2572 2573 static inline block_t __start_cp_addr(struct f2fs_sb_info *sbi) 2574 { 2575 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr); 2576 2577 if (sbi->cur_cp_pack == 2) 2578 start_addr += BLKS_PER_SEG(sbi); 2579 return start_addr; 2580 } 2581 2582 static inline block_t __start_cp_next_addr(struct f2fs_sb_info *sbi) 2583 { 2584 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr); 2585 2586 if (sbi->cur_cp_pack == 1) 2587 start_addr += BLKS_PER_SEG(sbi); 2588 return start_addr; 2589 } 2590 2591 static inline void __set_cp_next_pack(struct f2fs_sb_info *sbi) 2592 { 2593 sbi->cur_cp_pack = (sbi->cur_cp_pack == 1) ? 2 : 1; 2594 } 2595 2596 static inline block_t __start_sum_addr(struct f2fs_sb_info *sbi) 2597 { 2598 return le32_to_cpu(F2FS_CKPT(sbi)->cp_pack_start_sum); 2599 } 2600 2601 extern void f2fs_mark_inode_dirty_sync(struct inode *inode, bool sync); 2602 static inline int inc_valid_node_count(struct f2fs_sb_info *sbi, 2603 struct inode *inode, bool is_inode) 2604 { 2605 block_t valid_block_count; 2606 unsigned int valid_node_count, user_block_count; 2607 int err; 2608 2609 if (is_inode) { 2610 if (inode) { 2611 err = dquot_alloc_inode(inode); 2612 if (err) 2613 return err; 2614 } 2615 } else { 2616 err = dquot_reserve_block(inode, 1); 2617 if (err) 2618 return err; 2619 } 2620 2621 if (time_to_inject(sbi, FAULT_BLOCK)) 2622 goto enospc; 2623 2624 spin_lock(&sbi->stat_lock); 2625 2626 valid_block_count = sbi->total_valid_block_count + 2627 sbi->current_reserved_blocks + 1; 2628 2629 if (!__allow_reserved_blocks(sbi, inode, false)) 2630 valid_block_count += F2FS_OPTION(sbi).root_reserved_blocks; 2631 2632 user_block_count = sbi->user_block_count; 2633 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) 2634 user_block_count -= sbi->unusable_block_count; 2635 2636 if (unlikely(valid_block_count > user_block_count)) { 2637 spin_unlock(&sbi->stat_lock); 2638 goto enospc; 2639 } 2640 2641 valid_node_count = sbi->total_valid_node_count + 1; 2642 if (unlikely(valid_node_count > sbi->total_node_count)) { 2643 spin_unlock(&sbi->stat_lock); 2644 goto enospc; 2645 } 2646 2647 sbi->total_valid_node_count++; 2648 sbi->total_valid_block_count++; 2649 spin_unlock(&sbi->stat_lock); 2650 2651 if (inode) { 2652 if (is_inode) 2653 f2fs_mark_inode_dirty_sync(inode, true); 2654 else 2655 f2fs_i_blocks_write(inode, 1, true, true); 2656 } 2657 2658 percpu_counter_inc(&sbi->alloc_valid_block_count); 2659 return 0; 2660 2661 enospc: 2662 if (is_inode) { 2663 if (inode) 2664 dquot_free_inode(inode); 2665 } else { 2666 dquot_release_reservation_block(inode, 1); 2667 } 2668 return -ENOSPC; 2669 } 2670 2671 static inline void dec_valid_node_count(struct f2fs_sb_info *sbi, 2672 struct inode *inode, bool is_inode) 2673 { 2674 spin_lock(&sbi->stat_lock); 2675 2676 if (unlikely(!sbi->total_valid_block_count || 2677 !sbi->total_valid_node_count)) { 2678 f2fs_warn(sbi, "dec_valid_node_count: inconsistent block counts, total_valid_block:%u, total_valid_node:%u", 2679 sbi->total_valid_block_count, 2680 sbi->total_valid_node_count); 2681 set_sbi_flag(sbi, SBI_NEED_FSCK); 2682 } else { 2683 sbi->total_valid_block_count--; 2684 sbi->total_valid_node_count--; 2685 } 2686 2687 if (sbi->reserved_blocks && 2688 sbi->current_reserved_blocks < sbi->reserved_blocks) 2689 sbi->current_reserved_blocks++; 2690 2691 spin_unlock(&sbi->stat_lock); 2692 2693 if (is_inode) { 2694 dquot_free_inode(inode); 2695 } else { 2696 if (unlikely(inode->i_blocks == 0)) { 2697 f2fs_warn(sbi, "dec_valid_node_count: inconsistent i_blocks, ino:%lu, iblocks:%llu", 2698 inode->i_ino, 2699 (unsigned long long)inode->i_blocks); 2700 set_sbi_flag(sbi, SBI_NEED_FSCK); 2701 return; 2702 } 2703 f2fs_i_blocks_write(inode, 1, false, true); 2704 } 2705 } 2706 2707 static inline unsigned int valid_node_count(struct f2fs_sb_info *sbi) 2708 { 2709 return sbi->total_valid_node_count; 2710 } 2711 2712 static inline void inc_valid_inode_count(struct f2fs_sb_info *sbi) 2713 { 2714 percpu_counter_inc(&sbi->total_valid_inode_count); 2715 } 2716 2717 static inline void dec_valid_inode_count(struct f2fs_sb_info *sbi) 2718 { 2719 percpu_counter_dec(&sbi->total_valid_inode_count); 2720 } 2721 2722 static inline s64 valid_inode_count(struct f2fs_sb_info *sbi) 2723 { 2724 return percpu_counter_sum_positive(&sbi->total_valid_inode_count); 2725 } 2726 2727 static inline struct page *f2fs_grab_cache_page(struct address_space *mapping, 2728 pgoff_t index, bool for_write) 2729 { 2730 struct page *page; 2731 unsigned int flags; 2732 2733 if (IS_ENABLED(CONFIG_F2FS_FAULT_INJECTION)) { 2734 if (!for_write) 2735 page = find_get_page_flags(mapping, index, 2736 FGP_LOCK | FGP_ACCESSED); 2737 else 2738 page = find_lock_page(mapping, index); 2739 if (page) 2740 return page; 2741 2742 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_ALLOC)) 2743 return NULL; 2744 } 2745 2746 if (!for_write) 2747 return grab_cache_page(mapping, index); 2748 2749 flags = memalloc_nofs_save(); 2750 page = grab_cache_page_write_begin(mapping, index); 2751 memalloc_nofs_restore(flags); 2752 2753 return page; 2754 } 2755 2756 static inline struct page *f2fs_pagecache_get_page( 2757 struct address_space *mapping, pgoff_t index, 2758 fgf_t fgp_flags, gfp_t gfp_mask) 2759 { 2760 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_GET)) 2761 return NULL; 2762 2763 return pagecache_get_page(mapping, index, fgp_flags, gfp_mask); 2764 } 2765 2766 static inline void f2fs_put_page(struct page *page, int unlock) 2767 { 2768 if (!page) 2769 return; 2770 2771 if (unlock) { 2772 f2fs_bug_on(F2FS_P_SB(page), !PageLocked(page)); 2773 unlock_page(page); 2774 } 2775 put_page(page); 2776 } 2777 2778 static inline void f2fs_put_dnode(struct dnode_of_data *dn) 2779 { 2780 if (dn->node_page) 2781 f2fs_put_page(dn->node_page, 1); 2782 if (dn->inode_page && dn->node_page != dn->inode_page) 2783 f2fs_put_page(dn->inode_page, 0); 2784 dn->node_page = NULL; 2785 dn->inode_page = NULL; 2786 } 2787 2788 static inline struct kmem_cache *f2fs_kmem_cache_create(const char *name, 2789 size_t size) 2790 { 2791 return kmem_cache_create(name, size, 0, SLAB_RECLAIM_ACCOUNT, NULL); 2792 } 2793 2794 static inline void *f2fs_kmem_cache_alloc_nofail(struct kmem_cache *cachep, 2795 gfp_t flags) 2796 { 2797 void *entry; 2798 2799 entry = kmem_cache_alloc(cachep, flags); 2800 if (!entry) 2801 entry = kmem_cache_alloc(cachep, flags | __GFP_NOFAIL); 2802 return entry; 2803 } 2804 2805 static inline void *f2fs_kmem_cache_alloc(struct kmem_cache *cachep, 2806 gfp_t flags, bool nofail, struct f2fs_sb_info *sbi) 2807 { 2808 if (nofail) 2809 return f2fs_kmem_cache_alloc_nofail(cachep, flags); 2810 2811 if (time_to_inject(sbi, FAULT_SLAB_ALLOC)) 2812 return NULL; 2813 2814 return kmem_cache_alloc(cachep, flags); 2815 } 2816 2817 static inline bool is_inflight_io(struct f2fs_sb_info *sbi, int type) 2818 { 2819 if (get_pages(sbi, F2FS_RD_DATA) || get_pages(sbi, F2FS_RD_NODE) || 2820 get_pages(sbi, F2FS_RD_META) || get_pages(sbi, F2FS_WB_DATA) || 2821 get_pages(sbi, F2FS_WB_CP_DATA) || 2822 get_pages(sbi, F2FS_DIO_READ) || 2823 get_pages(sbi, F2FS_DIO_WRITE)) 2824 return true; 2825 2826 if (type != DISCARD_TIME && SM_I(sbi) && SM_I(sbi)->dcc_info && 2827 atomic_read(&SM_I(sbi)->dcc_info->queued_discard)) 2828 return true; 2829 2830 if (SM_I(sbi) && SM_I(sbi)->fcc_info && 2831 atomic_read(&SM_I(sbi)->fcc_info->queued_flush)) 2832 return true; 2833 return false; 2834 } 2835 2836 static inline bool is_idle(struct f2fs_sb_info *sbi, int type) 2837 { 2838 if (sbi->gc_mode == GC_URGENT_HIGH) 2839 return true; 2840 2841 if (is_inflight_io(sbi, type)) 2842 return false; 2843 2844 if (sbi->gc_mode == GC_URGENT_MID) 2845 return true; 2846 2847 if (sbi->gc_mode == GC_URGENT_LOW && 2848 (type == DISCARD_TIME || type == GC_TIME)) 2849 return true; 2850 2851 return f2fs_time_over(sbi, type); 2852 } 2853 2854 static inline void f2fs_radix_tree_insert(struct radix_tree_root *root, 2855 unsigned long index, void *item) 2856 { 2857 while (radix_tree_insert(root, index, item)) 2858 cond_resched(); 2859 } 2860 2861 #define RAW_IS_INODE(p) ((p)->footer.nid == (p)->footer.ino) 2862 2863 static inline bool IS_INODE(struct page *page) 2864 { 2865 struct f2fs_node *p = F2FS_NODE(page); 2866 2867 return RAW_IS_INODE(p); 2868 } 2869 2870 static inline int offset_in_addr(struct f2fs_inode *i) 2871 { 2872 return (i->i_inline & F2FS_EXTRA_ATTR) ? 2873 (le16_to_cpu(i->i_extra_isize) / sizeof(__le32)) : 0; 2874 } 2875 2876 static inline __le32 *blkaddr_in_node(struct f2fs_node *node) 2877 { 2878 return RAW_IS_INODE(node) ? node->i.i_addr : node->dn.addr; 2879 } 2880 2881 static inline int f2fs_has_extra_attr(struct inode *inode); 2882 static inline block_t data_blkaddr(struct inode *inode, 2883 struct page *node_page, unsigned int offset) 2884 { 2885 struct f2fs_node *raw_node; 2886 __le32 *addr_array; 2887 int base = 0; 2888 bool is_inode = IS_INODE(node_page); 2889 2890 raw_node = F2FS_NODE(node_page); 2891 2892 if (is_inode) { 2893 if (!inode) 2894 /* from GC path only */ 2895 base = offset_in_addr(&raw_node->i); 2896 else if (f2fs_has_extra_attr(inode)) 2897 base = get_extra_isize(inode); 2898 } 2899 2900 addr_array = blkaddr_in_node(raw_node); 2901 return le32_to_cpu(addr_array[base + offset]); 2902 } 2903 2904 static inline block_t f2fs_data_blkaddr(struct dnode_of_data *dn) 2905 { 2906 return data_blkaddr(dn->inode, dn->node_page, dn->ofs_in_node); 2907 } 2908 2909 static inline int f2fs_test_bit(unsigned int nr, char *addr) 2910 { 2911 int mask; 2912 2913 addr += (nr >> 3); 2914 mask = BIT(7 - (nr & 0x07)); 2915 return mask & *addr; 2916 } 2917 2918 static inline void f2fs_set_bit(unsigned int nr, char *addr) 2919 { 2920 int mask; 2921 2922 addr += (nr >> 3); 2923 mask = BIT(7 - (nr & 0x07)); 2924 *addr |= mask; 2925 } 2926 2927 static inline void f2fs_clear_bit(unsigned int nr, char *addr) 2928 { 2929 int mask; 2930 2931 addr += (nr >> 3); 2932 mask = BIT(7 - (nr & 0x07)); 2933 *addr &= ~mask; 2934 } 2935 2936 static inline int f2fs_test_and_set_bit(unsigned int nr, char *addr) 2937 { 2938 int mask; 2939 int ret; 2940 2941 addr += (nr >> 3); 2942 mask = BIT(7 - (nr & 0x07)); 2943 ret = mask & *addr; 2944 *addr |= mask; 2945 return ret; 2946 } 2947 2948 static inline int f2fs_test_and_clear_bit(unsigned int nr, char *addr) 2949 { 2950 int mask; 2951 int ret; 2952 2953 addr += (nr >> 3); 2954 mask = BIT(7 - (nr & 0x07)); 2955 ret = mask & *addr; 2956 *addr &= ~mask; 2957 return ret; 2958 } 2959 2960 static inline void f2fs_change_bit(unsigned int nr, char *addr) 2961 { 2962 int mask; 2963 2964 addr += (nr >> 3); 2965 mask = BIT(7 - (nr & 0x07)); 2966 *addr ^= mask; 2967 } 2968 2969 /* 2970 * On-disk inode flags (f2fs_inode::i_flags) 2971 */ 2972 #define F2FS_COMPR_FL 0x00000004 /* Compress file */ 2973 #define F2FS_SYNC_FL 0x00000008 /* Synchronous updates */ 2974 #define F2FS_IMMUTABLE_FL 0x00000010 /* Immutable file */ 2975 #define F2FS_APPEND_FL 0x00000020 /* writes to file may only append */ 2976 #define F2FS_NODUMP_FL 0x00000040 /* do not dump file */ 2977 #define F2FS_NOATIME_FL 0x00000080 /* do not update atime */ 2978 #define F2FS_NOCOMP_FL 0x00000400 /* Don't compress */ 2979 #define F2FS_INDEX_FL 0x00001000 /* hash-indexed directory */ 2980 #define F2FS_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */ 2981 #define F2FS_PROJINHERIT_FL 0x20000000 /* Create with parents projid */ 2982 #define F2FS_CASEFOLD_FL 0x40000000 /* Casefolded file */ 2983 2984 #define F2FS_QUOTA_DEFAULT_FL (F2FS_NOATIME_FL | F2FS_IMMUTABLE_FL) 2985 2986 /* Flags that should be inherited by new inodes from their parent. */ 2987 #define F2FS_FL_INHERITED (F2FS_SYNC_FL | F2FS_NODUMP_FL | F2FS_NOATIME_FL | \ 2988 F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \ 2989 F2FS_CASEFOLD_FL) 2990 2991 /* Flags that are appropriate for regular files (all but dir-specific ones). */ 2992 #define F2FS_REG_FLMASK (~(F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \ 2993 F2FS_CASEFOLD_FL)) 2994 2995 /* Flags that are appropriate for non-directories/regular files. */ 2996 #define F2FS_OTHER_FLMASK (F2FS_NODUMP_FL | F2FS_NOATIME_FL) 2997 2998 static inline __u32 f2fs_mask_flags(umode_t mode, __u32 flags) 2999 { 3000 if (S_ISDIR(mode)) 3001 return flags; 3002 else if (S_ISREG(mode)) 3003 return flags & F2FS_REG_FLMASK; 3004 else 3005 return flags & F2FS_OTHER_FLMASK; 3006 } 3007 3008 static inline void __mark_inode_dirty_flag(struct inode *inode, 3009 int flag, bool set) 3010 { 3011 switch (flag) { 3012 case FI_INLINE_XATTR: 3013 case FI_INLINE_DATA: 3014 case FI_INLINE_DENTRY: 3015 case FI_NEW_INODE: 3016 if (set) 3017 return; 3018 fallthrough; 3019 case FI_DATA_EXIST: 3020 case FI_INLINE_DOTS: 3021 case FI_PIN_FILE: 3022 case FI_COMPRESS_RELEASED: 3023 case FI_ATOMIC_COMMITTED: 3024 f2fs_mark_inode_dirty_sync(inode, true); 3025 } 3026 } 3027 3028 static inline void set_inode_flag(struct inode *inode, int flag) 3029 { 3030 set_bit(flag, F2FS_I(inode)->flags); 3031 __mark_inode_dirty_flag(inode, flag, true); 3032 } 3033 3034 static inline int is_inode_flag_set(struct inode *inode, int flag) 3035 { 3036 return test_bit(flag, F2FS_I(inode)->flags); 3037 } 3038 3039 static inline void clear_inode_flag(struct inode *inode, int flag) 3040 { 3041 clear_bit(flag, F2FS_I(inode)->flags); 3042 __mark_inode_dirty_flag(inode, flag, false); 3043 } 3044 3045 static inline bool f2fs_verity_in_progress(struct inode *inode) 3046 { 3047 return IS_ENABLED(CONFIG_FS_VERITY) && 3048 is_inode_flag_set(inode, FI_VERITY_IN_PROGRESS); 3049 } 3050 3051 static inline void set_acl_inode(struct inode *inode, umode_t mode) 3052 { 3053 F2FS_I(inode)->i_acl_mode = mode; 3054 set_inode_flag(inode, FI_ACL_MODE); 3055 f2fs_mark_inode_dirty_sync(inode, false); 3056 } 3057 3058 static inline void f2fs_i_links_write(struct inode *inode, bool inc) 3059 { 3060 if (inc) 3061 inc_nlink(inode); 3062 else 3063 drop_nlink(inode); 3064 f2fs_mark_inode_dirty_sync(inode, true); 3065 } 3066 3067 static inline void f2fs_i_blocks_write(struct inode *inode, 3068 block_t diff, bool add, bool claim) 3069 { 3070 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE); 3071 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER); 3072 3073 /* add = 1, claim = 1 should be dquot_reserve_block in pair */ 3074 if (add) { 3075 if (claim) 3076 dquot_claim_block(inode, diff); 3077 else 3078 dquot_alloc_block_nofail(inode, diff); 3079 } else { 3080 dquot_free_block(inode, diff); 3081 } 3082 3083 f2fs_mark_inode_dirty_sync(inode, true); 3084 if (clean || recover) 3085 set_inode_flag(inode, FI_AUTO_RECOVER); 3086 } 3087 3088 static inline bool f2fs_is_atomic_file(struct inode *inode); 3089 3090 static inline void f2fs_i_size_write(struct inode *inode, loff_t i_size) 3091 { 3092 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE); 3093 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER); 3094 3095 if (i_size_read(inode) == i_size) 3096 return; 3097 3098 i_size_write(inode, i_size); 3099 3100 if (f2fs_is_atomic_file(inode)) 3101 return; 3102 3103 f2fs_mark_inode_dirty_sync(inode, true); 3104 if (clean || recover) 3105 set_inode_flag(inode, FI_AUTO_RECOVER); 3106 } 3107 3108 static inline void f2fs_i_depth_write(struct inode *inode, unsigned int depth) 3109 { 3110 F2FS_I(inode)->i_current_depth = depth; 3111 f2fs_mark_inode_dirty_sync(inode, true); 3112 } 3113 3114 static inline void f2fs_i_gc_failures_write(struct inode *inode, 3115 unsigned int count) 3116 { 3117 F2FS_I(inode)->i_gc_failures[GC_FAILURE_PIN] = count; 3118 f2fs_mark_inode_dirty_sync(inode, true); 3119 } 3120 3121 static inline void f2fs_i_xnid_write(struct inode *inode, nid_t xnid) 3122 { 3123 F2FS_I(inode)->i_xattr_nid = xnid; 3124 f2fs_mark_inode_dirty_sync(inode, true); 3125 } 3126 3127 static inline void f2fs_i_pino_write(struct inode *inode, nid_t pino) 3128 { 3129 F2FS_I(inode)->i_pino = pino; 3130 f2fs_mark_inode_dirty_sync(inode, true); 3131 } 3132 3133 static inline void get_inline_info(struct inode *inode, struct f2fs_inode *ri) 3134 { 3135 struct f2fs_inode_info *fi = F2FS_I(inode); 3136 3137 if (ri->i_inline & F2FS_INLINE_XATTR) 3138 set_bit(FI_INLINE_XATTR, fi->flags); 3139 if (ri->i_inline & F2FS_INLINE_DATA) 3140 set_bit(FI_INLINE_DATA, fi->flags); 3141 if (ri->i_inline & F2FS_INLINE_DENTRY) 3142 set_bit(FI_INLINE_DENTRY, fi->flags); 3143 if (ri->i_inline & F2FS_DATA_EXIST) 3144 set_bit(FI_DATA_EXIST, fi->flags); 3145 if (ri->i_inline & F2FS_INLINE_DOTS) 3146 set_bit(FI_INLINE_DOTS, fi->flags); 3147 if (ri->i_inline & F2FS_EXTRA_ATTR) 3148 set_bit(FI_EXTRA_ATTR, fi->flags); 3149 if (ri->i_inline & F2FS_PIN_FILE) 3150 set_bit(FI_PIN_FILE, fi->flags); 3151 if (ri->i_inline & F2FS_COMPRESS_RELEASED) 3152 set_bit(FI_COMPRESS_RELEASED, fi->flags); 3153 } 3154 3155 static inline void set_raw_inline(struct inode *inode, struct f2fs_inode *ri) 3156 { 3157 ri->i_inline = 0; 3158 3159 if (is_inode_flag_set(inode, FI_INLINE_XATTR)) 3160 ri->i_inline |= F2FS_INLINE_XATTR; 3161 if (is_inode_flag_set(inode, FI_INLINE_DATA)) 3162 ri->i_inline |= F2FS_INLINE_DATA; 3163 if (is_inode_flag_set(inode, FI_INLINE_DENTRY)) 3164 ri->i_inline |= F2FS_INLINE_DENTRY; 3165 if (is_inode_flag_set(inode, FI_DATA_EXIST)) 3166 ri->i_inline |= F2FS_DATA_EXIST; 3167 if (is_inode_flag_set(inode, FI_INLINE_DOTS)) 3168 ri->i_inline |= F2FS_INLINE_DOTS; 3169 if (is_inode_flag_set(inode, FI_EXTRA_ATTR)) 3170 ri->i_inline |= F2FS_EXTRA_ATTR; 3171 if (is_inode_flag_set(inode, FI_PIN_FILE)) 3172 ri->i_inline |= F2FS_PIN_FILE; 3173 if (is_inode_flag_set(inode, FI_COMPRESS_RELEASED)) 3174 ri->i_inline |= F2FS_COMPRESS_RELEASED; 3175 } 3176 3177 static inline int f2fs_has_extra_attr(struct inode *inode) 3178 { 3179 return is_inode_flag_set(inode, FI_EXTRA_ATTR); 3180 } 3181 3182 static inline int f2fs_has_inline_xattr(struct inode *inode) 3183 { 3184 return is_inode_flag_set(inode, FI_INLINE_XATTR); 3185 } 3186 3187 static inline int f2fs_compressed_file(struct inode *inode) 3188 { 3189 return S_ISREG(inode->i_mode) && 3190 is_inode_flag_set(inode, FI_COMPRESSED_FILE); 3191 } 3192 3193 static inline bool f2fs_need_compress_data(struct inode *inode) 3194 { 3195 int compress_mode = F2FS_OPTION(F2FS_I_SB(inode)).compress_mode; 3196 3197 if (!f2fs_compressed_file(inode)) 3198 return false; 3199 3200 if (compress_mode == COMPR_MODE_FS) 3201 return true; 3202 else if (compress_mode == COMPR_MODE_USER && 3203 is_inode_flag_set(inode, FI_ENABLE_COMPRESS)) 3204 return true; 3205 3206 return false; 3207 } 3208 3209 static inline unsigned int addrs_per_inode(struct inode *inode) 3210 { 3211 unsigned int addrs = CUR_ADDRS_PER_INODE(inode) - 3212 get_inline_xattr_addrs(inode); 3213 3214 if (!f2fs_compressed_file(inode)) 3215 return addrs; 3216 return ALIGN_DOWN(addrs, F2FS_I(inode)->i_cluster_size); 3217 } 3218 3219 static inline unsigned int addrs_per_block(struct inode *inode) 3220 { 3221 if (!f2fs_compressed_file(inode)) 3222 return DEF_ADDRS_PER_BLOCK; 3223 return ALIGN_DOWN(DEF_ADDRS_PER_BLOCK, F2FS_I(inode)->i_cluster_size); 3224 } 3225 3226 static inline void *inline_xattr_addr(struct inode *inode, struct page *page) 3227 { 3228 struct f2fs_inode *ri = F2FS_INODE(page); 3229 3230 return (void *)&(ri->i_addr[DEF_ADDRS_PER_INODE - 3231 get_inline_xattr_addrs(inode)]); 3232 } 3233 3234 static inline int inline_xattr_size(struct inode *inode) 3235 { 3236 if (f2fs_has_inline_xattr(inode)) 3237 return get_inline_xattr_addrs(inode) * sizeof(__le32); 3238 return 0; 3239 } 3240 3241 /* 3242 * Notice: check inline_data flag without inode page lock is unsafe. 3243 * It could change at any time by f2fs_convert_inline_page(). 3244 */ 3245 static inline int f2fs_has_inline_data(struct inode *inode) 3246 { 3247 return is_inode_flag_set(inode, FI_INLINE_DATA); 3248 } 3249 3250 static inline int f2fs_exist_data(struct inode *inode) 3251 { 3252 return is_inode_flag_set(inode, FI_DATA_EXIST); 3253 } 3254 3255 static inline int f2fs_has_inline_dots(struct inode *inode) 3256 { 3257 return is_inode_flag_set(inode, FI_INLINE_DOTS); 3258 } 3259 3260 static inline int f2fs_is_mmap_file(struct inode *inode) 3261 { 3262 return is_inode_flag_set(inode, FI_MMAP_FILE); 3263 } 3264 3265 static inline bool f2fs_is_pinned_file(struct inode *inode) 3266 { 3267 return is_inode_flag_set(inode, FI_PIN_FILE); 3268 } 3269 3270 static inline bool f2fs_is_atomic_file(struct inode *inode) 3271 { 3272 return is_inode_flag_set(inode, FI_ATOMIC_FILE); 3273 } 3274 3275 static inline bool f2fs_is_cow_file(struct inode *inode) 3276 { 3277 return is_inode_flag_set(inode, FI_COW_FILE); 3278 } 3279 3280 static inline __le32 *get_dnode_addr(struct inode *inode, 3281 struct page *node_page); 3282 static inline void *inline_data_addr(struct inode *inode, struct page *page) 3283 { 3284 __le32 *addr = get_dnode_addr(inode, page); 3285 3286 return (void *)(addr + DEF_INLINE_RESERVED_SIZE); 3287 } 3288 3289 static inline int f2fs_has_inline_dentry(struct inode *inode) 3290 { 3291 return is_inode_flag_set(inode, FI_INLINE_DENTRY); 3292 } 3293 3294 static inline int is_file(struct inode *inode, int type) 3295 { 3296 return F2FS_I(inode)->i_advise & type; 3297 } 3298 3299 static inline void set_file(struct inode *inode, int type) 3300 { 3301 if (is_file(inode, type)) 3302 return; 3303 F2FS_I(inode)->i_advise |= type; 3304 f2fs_mark_inode_dirty_sync(inode, true); 3305 } 3306 3307 static inline void clear_file(struct inode *inode, int type) 3308 { 3309 if (!is_file(inode, type)) 3310 return; 3311 F2FS_I(inode)->i_advise &= ~type; 3312 f2fs_mark_inode_dirty_sync(inode, true); 3313 } 3314 3315 static inline bool f2fs_is_time_consistent(struct inode *inode) 3316 { 3317 struct timespec64 ctime = inode_get_ctime(inode); 3318 3319 if (!timespec64_equal(F2FS_I(inode)->i_disk_time, &inode->i_atime)) 3320 return false; 3321 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 1, &ctime)) 3322 return false; 3323 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 2, &inode->i_mtime)) 3324 return false; 3325 return true; 3326 } 3327 3328 static inline bool f2fs_skip_inode_update(struct inode *inode, int dsync) 3329 { 3330 bool ret; 3331 3332 if (dsync) { 3333 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 3334 3335 spin_lock(&sbi->inode_lock[DIRTY_META]); 3336 ret = list_empty(&F2FS_I(inode)->gdirty_list); 3337 spin_unlock(&sbi->inode_lock[DIRTY_META]); 3338 return ret; 3339 } 3340 if (!is_inode_flag_set(inode, FI_AUTO_RECOVER) || 3341 file_keep_isize(inode) || 3342 i_size_read(inode) & ~PAGE_MASK) 3343 return false; 3344 3345 if (!f2fs_is_time_consistent(inode)) 3346 return false; 3347 3348 spin_lock(&F2FS_I(inode)->i_size_lock); 3349 ret = F2FS_I(inode)->last_disk_size == i_size_read(inode); 3350 spin_unlock(&F2FS_I(inode)->i_size_lock); 3351 3352 return ret; 3353 } 3354 3355 static inline bool f2fs_readonly(struct super_block *sb) 3356 { 3357 return sb_rdonly(sb); 3358 } 3359 3360 static inline bool f2fs_cp_error(struct f2fs_sb_info *sbi) 3361 { 3362 return is_set_ckpt_flags(sbi, CP_ERROR_FLAG); 3363 } 3364 3365 static inline bool is_dot_dotdot(const u8 *name, size_t len) 3366 { 3367 if (len == 1 && name[0] == '.') 3368 return true; 3369 3370 if (len == 2 && name[0] == '.' && name[1] == '.') 3371 return true; 3372 3373 return false; 3374 } 3375 3376 static inline void *f2fs_kmalloc(struct f2fs_sb_info *sbi, 3377 size_t size, gfp_t flags) 3378 { 3379 if (time_to_inject(sbi, FAULT_KMALLOC)) 3380 return NULL; 3381 3382 return kmalloc(size, flags); 3383 } 3384 3385 static inline void *f2fs_getname(struct f2fs_sb_info *sbi) 3386 { 3387 if (time_to_inject(sbi, FAULT_KMALLOC)) 3388 return NULL; 3389 3390 return __getname(); 3391 } 3392 3393 static inline void f2fs_putname(char *buf) 3394 { 3395 __putname(buf); 3396 } 3397 3398 static inline void *f2fs_kzalloc(struct f2fs_sb_info *sbi, 3399 size_t size, gfp_t flags) 3400 { 3401 return f2fs_kmalloc(sbi, size, flags | __GFP_ZERO); 3402 } 3403 3404 static inline void *f2fs_kvmalloc(struct f2fs_sb_info *sbi, 3405 size_t size, gfp_t flags) 3406 { 3407 if (time_to_inject(sbi, FAULT_KVMALLOC)) 3408 return NULL; 3409 3410 return kvmalloc(size, flags); 3411 } 3412 3413 static inline void *f2fs_kvzalloc(struct f2fs_sb_info *sbi, 3414 size_t size, gfp_t flags) 3415 { 3416 return f2fs_kvmalloc(sbi, size, flags | __GFP_ZERO); 3417 } 3418 3419 static inline int get_extra_isize(struct inode *inode) 3420 { 3421 return F2FS_I(inode)->i_extra_isize / sizeof(__le32); 3422 } 3423 3424 static inline int get_inline_xattr_addrs(struct inode *inode) 3425 { 3426 return F2FS_I(inode)->i_inline_xattr_size; 3427 } 3428 3429 static inline __le32 *get_dnode_addr(struct inode *inode, 3430 struct page *node_page) 3431 { 3432 int base = 0; 3433 3434 if (IS_INODE(node_page) && f2fs_has_extra_attr(inode)) 3435 base = get_extra_isize(inode); 3436 3437 return blkaddr_in_node(F2FS_NODE(node_page)) + base; 3438 } 3439 3440 #define f2fs_get_inode_mode(i) \ 3441 ((is_inode_flag_set(i, FI_ACL_MODE)) ? \ 3442 (F2FS_I(i)->i_acl_mode) : ((i)->i_mode)) 3443 3444 #define F2FS_MIN_EXTRA_ATTR_SIZE (sizeof(__le32)) 3445 3446 #define F2FS_TOTAL_EXTRA_ATTR_SIZE \ 3447 (offsetof(struct f2fs_inode, i_extra_end) - \ 3448 offsetof(struct f2fs_inode, i_extra_isize)) \ 3449 3450 #define F2FS_OLD_ATTRIBUTE_SIZE (offsetof(struct f2fs_inode, i_addr)) 3451 #define F2FS_FITS_IN_INODE(f2fs_inode, extra_isize, field) \ 3452 ((offsetof(typeof(*(f2fs_inode)), field) + \ 3453 sizeof((f2fs_inode)->field)) \ 3454 <= (F2FS_OLD_ATTRIBUTE_SIZE + (extra_isize))) \ 3455 3456 #define __is_large_section(sbi) (SEGS_PER_SEC(sbi) > 1) 3457 3458 #define __is_meta_io(fio) (PAGE_TYPE_OF_BIO((fio)->type) == META) 3459 3460 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi, 3461 block_t blkaddr, int type); 3462 static inline void verify_blkaddr(struct f2fs_sb_info *sbi, 3463 block_t blkaddr, int type) 3464 { 3465 if (!f2fs_is_valid_blkaddr(sbi, blkaddr, type)) 3466 f2fs_err(sbi, "invalid blkaddr: %u, type: %d, run fsck to fix.", 3467 blkaddr, type); 3468 } 3469 3470 static inline bool __is_valid_data_blkaddr(block_t blkaddr) 3471 { 3472 if (blkaddr == NEW_ADDR || blkaddr == NULL_ADDR || 3473 blkaddr == COMPRESS_ADDR) 3474 return false; 3475 return true; 3476 } 3477 3478 /* 3479 * file.c 3480 */ 3481 int f2fs_sync_file(struct file *file, loff_t start, loff_t end, int datasync); 3482 int f2fs_do_truncate_blocks(struct inode *inode, u64 from, bool lock); 3483 int f2fs_truncate_blocks(struct inode *inode, u64 from, bool lock); 3484 int f2fs_truncate(struct inode *inode); 3485 int f2fs_getattr(struct mnt_idmap *idmap, const struct path *path, 3486 struct kstat *stat, u32 request_mask, unsigned int flags); 3487 int f2fs_setattr(struct mnt_idmap *idmap, struct dentry *dentry, 3488 struct iattr *attr); 3489 int f2fs_truncate_hole(struct inode *inode, pgoff_t pg_start, pgoff_t pg_end); 3490 void f2fs_truncate_data_blocks_range(struct dnode_of_data *dn, int count); 3491 int f2fs_precache_extents(struct inode *inode); 3492 int f2fs_fileattr_get(struct dentry *dentry, struct fileattr *fa); 3493 int f2fs_fileattr_set(struct mnt_idmap *idmap, 3494 struct dentry *dentry, struct fileattr *fa); 3495 long f2fs_ioctl(struct file *filp, unsigned int cmd, unsigned long arg); 3496 long f2fs_compat_ioctl(struct file *file, unsigned int cmd, unsigned long arg); 3497 int f2fs_transfer_project_quota(struct inode *inode, kprojid_t kprojid); 3498 int f2fs_pin_file_control(struct inode *inode, bool inc); 3499 3500 /* 3501 * inode.c 3502 */ 3503 void f2fs_set_inode_flags(struct inode *inode); 3504 bool f2fs_inode_chksum_verify(struct f2fs_sb_info *sbi, struct page *page); 3505 void f2fs_inode_chksum_set(struct f2fs_sb_info *sbi, struct page *page); 3506 struct inode *f2fs_iget(struct super_block *sb, unsigned long ino); 3507 struct inode *f2fs_iget_retry(struct super_block *sb, unsigned long ino); 3508 int f2fs_try_to_free_nats(struct f2fs_sb_info *sbi, int nr_shrink); 3509 void f2fs_update_inode(struct inode *inode, struct page *node_page); 3510 void f2fs_update_inode_page(struct inode *inode); 3511 int f2fs_write_inode(struct inode *inode, struct writeback_control *wbc); 3512 void f2fs_evict_inode(struct inode *inode); 3513 void f2fs_handle_failed_inode(struct inode *inode); 3514 3515 /* 3516 * namei.c 3517 */ 3518 int f2fs_update_extension_list(struct f2fs_sb_info *sbi, const char *name, 3519 bool hot, bool set); 3520 struct dentry *f2fs_get_parent(struct dentry *child); 3521 int f2fs_get_tmpfile(struct mnt_idmap *idmap, struct inode *dir, 3522 struct inode **new_inode); 3523 3524 /* 3525 * dir.c 3526 */ 3527 int f2fs_init_casefolded_name(const struct inode *dir, 3528 struct f2fs_filename *fname); 3529 int f2fs_setup_filename(struct inode *dir, const struct qstr *iname, 3530 int lookup, struct f2fs_filename *fname); 3531 int f2fs_prepare_lookup(struct inode *dir, struct dentry *dentry, 3532 struct f2fs_filename *fname); 3533 void f2fs_free_filename(struct f2fs_filename *fname); 3534 struct f2fs_dir_entry *f2fs_find_target_dentry(const struct f2fs_dentry_ptr *d, 3535 const struct f2fs_filename *fname, int *max_slots); 3536 int f2fs_fill_dentries(struct dir_context *ctx, struct f2fs_dentry_ptr *d, 3537 unsigned int start_pos, struct fscrypt_str *fstr); 3538 void f2fs_do_make_empty_dir(struct inode *inode, struct inode *parent, 3539 struct f2fs_dentry_ptr *d); 3540 struct page *f2fs_init_inode_metadata(struct inode *inode, struct inode *dir, 3541 const struct f2fs_filename *fname, struct page *dpage); 3542 void f2fs_update_parent_metadata(struct inode *dir, struct inode *inode, 3543 unsigned int current_depth); 3544 int f2fs_room_for_filename(const void *bitmap, int slots, int max_slots); 3545 void f2fs_drop_nlink(struct inode *dir, struct inode *inode); 3546 struct f2fs_dir_entry *__f2fs_find_entry(struct inode *dir, 3547 const struct f2fs_filename *fname, 3548 struct page **res_page); 3549 struct f2fs_dir_entry *f2fs_find_entry(struct inode *dir, 3550 const struct qstr *child, struct page **res_page); 3551 struct f2fs_dir_entry *f2fs_parent_dir(struct inode *dir, struct page **p); 3552 ino_t f2fs_inode_by_name(struct inode *dir, const struct qstr *qstr, 3553 struct page **page); 3554 void f2fs_set_link(struct inode *dir, struct f2fs_dir_entry *de, 3555 struct page *page, struct inode *inode); 3556 bool f2fs_has_enough_room(struct inode *dir, struct page *ipage, 3557 const struct f2fs_filename *fname); 3558 void f2fs_update_dentry(nid_t ino, umode_t mode, struct f2fs_dentry_ptr *d, 3559 const struct fscrypt_str *name, f2fs_hash_t name_hash, 3560 unsigned int bit_pos); 3561 int f2fs_add_regular_entry(struct inode *dir, const struct f2fs_filename *fname, 3562 struct inode *inode, nid_t ino, umode_t mode); 3563 int f2fs_add_dentry(struct inode *dir, const struct f2fs_filename *fname, 3564 struct inode *inode, nid_t ino, umode_t mode); 3565 int f2fs_do_add_link(struct inode *dir, const struct qstr *name, 3566 struct inode *inode, nid_t ino, umode_t mode); 3567 void f2fs_delete_entry(struct f2fs_dir_entry *dentry, struct page *page, 3568 struct inode *dir, struct inode *inode); 3569 int f2fs_do_tmpfile(struct inode *inode, struct inode *dir, 3570 struct f2fs_filename *fname); 3571 bool f2fs_empty_dir(struct inode *dir); 3572 3573 static inline int f2fs_add_link(struct dentry *dentry, struct inode *inode) 3574 { 3575 if (fscrypt_is_nokey_name(dentry)) 3576 return -ENOKEY; 3577 return f2fs_do_add_link(d_inode(dentry->d_parent), &dentry->d_name, 3578 inode, inode->i_ino, inode->i_mode); 3579 } 3580 3581 /* 3582 * super.c 3583 */ 3584 int f2fs_inode_dirtied(struct inode *inode, bool sync); 3585 void f2fs_inode_synced(struct inode *inode); 3586 int f2fs_dquot_initialize(struct inode *inode); 3587 int f2fs_enable_quota_files(struct f2fs_sb_info *sbi, bool rdonly); 3588 int f2fs_quota_sync(struct super_block *sb, int type); 3589 loff_t max_file_blocks(struct inode *inode); 3590 void f2fs_quota_off_umount(struct super_block *sb); 3591 void f2fs_save_errors(struct f2fs_sb_info *sbi, unsigned char flag); 3592 void f2fs_handle_critical_error(struct f2fs_sb_info *sbi, unsigned char reason, 3593 bool irq_context); 3594 void f2fs_handle_error(struct f2fs_sb_info *sbi, unsigned char error); 3595 void f2fs_handle_error_async(struct f2fs_sb_info *sbi, unsigned char error); 3596 int f2fs_commit_super(struct f2fs_sb_info *sbi, bool recover); 3597 int f2fs_sync_fs(struct super_block *sb, int sync); 3598 int f2fs_sanity_check_ckpt(struct f2fs_sb_info *sbi); 3599 3600 /* 3601 * hash.c 3602 */ 3603 void f2fs_hash_filename(const struct inode *dir, struct f2fs_filename *fname); 3604 3605 /* 3606 * node.c 3607 */ 3608 struct node_info; 3609 3610 int f2fs_check_nid_range(struct f2fs_sb_info *sbi, nid_t nid); 3611 bool f2fs_available_free_memory(struct f2fs_sb_info *sbi, int type); 3612 bool f2fs_in_warm_node_list(struct f2fs_sb_info *sbi, struct page *page); 3613 void f2fs_init_fsync_node_info(struct f2fs_sb_info *sbi); 3614 void f2fs_del_fsync_node_entry(struct f2fs_sb_info *sbi, struct page *page); 3615 void f2fs_reset_fsync_node_info(struct f2fs_sb_info *sbi); 3616 int f2fs_need_dentry_mark(struct f2fs_sb_info *sbi, nid_t nid); 3617 bool f2fs_is_checkpointed_node(struct f2fs_sb_info *sbi, nid_t nid); 3618 bool f2fs_need_inode_block_update(struct f2fs_sb_info *sbi, nid_t ino); 3619 int f2fs_get_node_info(struct f2fs_sb_info *sbi, nid_t nid, 3620 struct node_info *ni, bool checkpoint_context); 3621 pgoff_t f2fs_get_next_page_offset(struct dnode_of_data *dn, pgoff_t pgofs); 3622 int f2fs_get_dnode_of_data(struct dnode_of_data *dn, pgoff_t index, int mode); 3623 int f2fs_truncate_inode_blocks(struct inode *inode, pgoff_t from); 3624 int f2fs_truncate_xattr_node(struct inode *inode); 3625 int f2fs_wait_on_node_pages_writeback(struct f2fs_sb_info *sbi, 3626 unsigned int seq_id); 3627 bool f2fs_nat_bitmap_enabled(struct f2fs_sb_info *sbi); 3628 int f2fs_remove_inode_page(struct inode *inode); 3629 struct page *f2fs_new_inode_page(struct inode *inode); 3630 struct page *f2fs_new_node_page(struct dnode_of_data *dn, unsigned int ofs); 3631 void f2fs_ra_node_page(struct f2fs_sb_info *sbi, nid_t nid); 3632 struct page *f2fs_get_node_page(struct f2fs_sb_info *sbi, pgoff_t nid); 3633 struct page *f2fs_get_node_page_ra(struct page *parent, int start); 3634 int f2fs_move_node_page(struct page *node_page, int gc_type); 3635 void f2fs_flush_inline_data(struct f2fs_sb_info *sbi); 3636 int f2fs_fsync_node_pages(struct f2fs_sb_info *sbi, struct inode *inode, 3637 struct writeback_control *wbc, bool atomic, 3638 unsigned int *seq_id); 3639 int f2fs_sync_node_pages(struct f2fs_sb_info *sbi, 3640 struct writeback_control *wbc, 3641 bool do_balance, enum iostat_type io_type); 3642 int f2fs_build_free_nids(struct f2fs_sb_info *sbi, bool sync, bool mount); 3643 bool f2fs_alloc_nid(struct f2fs_sb_info *sbi, nid_t *nid); 3644 void f2fs_alloc_nid_done(struct f2fs_sb_info *sbi, nid_t nid); 3645 void f2fs_alloc_nid_failed(struct f2fs_sb_info *sbi, nid_t nid); 3646 int f2fs_try_to_free_nids(struct f2fs_sb_info *sbi, int nr_shrink); 3647 int f2fs_recover_inline_xattr(struct inode *inode, struct page *page); 3648 int f2fs_recover_xattr_data(struct inode *inode, struct page *page); 3649 int f2fs_recover_inode_page(struct f2fs_sb_info *sbi, struct page *page); 3650 int f2fs_restore_node_summary(struct f2fs_sb_info *sbi, 3651 unsigned int segno, struct f2fs_summary_block *sum); 3652 void f2fs_enable_nat_bits(struct f2fs_sb_info *sbi); 3653 int f2fs_flush_nat_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3654 int f2fs_build_node_manager(struct f2fs_sb_info *sbi); 3655 void f2fs_destroy_node_manager(struct f2fs_sb_info *sbi); 3656 int __init f2fs_create_node_manager_caches(void); 3657 void f2fs_destroy_node_manager_caches(void); 3658 3659 /* 3660 * segment.c 3661 */ 3662 bool f2fs_need_SSR(struct f2fs_sb_info *sbi); 3663 int f2fs_commit_atomic_write(struct inode *inode); 3664 void f2fs_abort_atomic_write(struct inode *inode, bool clean); 3665 void f2fs_balance_fs(struct f2fs_sb_info *sbi, bool need); 3666 void f2fs_balance_fs_bg(struct f2fs_sb_info *sbi, bool from_bg); 3667 int f2fs_issue_flush(struct f2fs_sb_info *sbi, nid_t ino); 3668 int f2fs_create_flush_cmd_control(struct f2fs_sb_info *sbi); 3669 int f2fs_flush_device_cache(struct f2fs_sb_info *sbi); 3670 void f2fs_destroy_flush_cmd_control(struct f2fs_sb_info *sbi, bool free); 3671 void f2fs_invalidate_blocks(struct f2fs_sb_info *sbi, block_t addr); 3672 bool f2fs_is_checkpointed_data(struct f2fs_sb_info *sbi, block_t blkaddr); 3673 int f2fs_start_discard_thread(struct f2fs_sb_info *sbi); 3674 void f2fs_drop_discard_cmd(struct f2fs_sb_info *sbi); 3675 void f2fs_stop_discard_thread(struct f2fs_sb_info *sbi); 3676 bool f2fs_issue_discard_timeout(struct f2fs_sb_info *sbi); 3677 void f2fs_clear_prefree_segments(struct f2fs_sb_info *sbi, 3678 struct cp_control *cpc); 3679 void f2fs_dirty_to_prefree(struct f2fs_sb_info *sbi); 3680 block_t f2fs_get_unusable_blocks(struct f2fs_sb_info *sbi); 3681 int f2fs_disable_cp_again(struct f2fs_sb_info *sbi, block_t unusable); 3682 void f2fs_release_discard_addrs(struct f2fs_sb_info *sbi); 3683 int f2fs_npages_for_summary_flush(struct f2fs_sb_info *sbi, bool for_ra); 3684 bool f2fs_segment_has_free_slot(struct f2fs_sb_info *sbi, int segno); 3685 void f2fs_init_inmem_curseg(struct f2fs_sb_info *sbi); 3686 void f2fs_save_inmem_curseg(struct f2fs_sb_info *sbi); 3687 void f2fs_restore_inmem_curseg(struct f2fs_sb_info *sbi); 3688 void f2fs_get_new_segment(struct f2fs_sb_info *sbi, 3689 unsigned int *newseg, bool new_sec, int dir); 3690 void f2fs_allocate_segment_for_resize(struct f2fs_sb_info *sbi, int type, 3691 unsigned int start, unsigned int end); 3692 int f2fs_allocate_new_section(struct f2fs_sb_info *sbi, int type, bool force); 3693 int f2fs_allocate_pinning_section(struct f2fs_sb_info *sbi); 3694 void f2fs_allocate_new_segments(struct f2fs_sb_info *sbi); 3695 int f2fs_trim_fs(struct f2fs_sb_info *sbi, struct fstrim_range *range); 3696 bool f2fs_exist_trim_candidates(struct f2fs_sb_info *sbi, 3697 struct cp_control *cpc); 3698 struct page *f2fs_get_sum_page(struct f2fs_sb_info *sbi, unsigned int segno); 3699 void f2fs_update_meta_page(struct f2fs_sb_info *sbi, void *src, 3700 block_t blk_addr); 3701 void f2fs_do_write_meta_page(struct f2fs_sb_info *sbi, struct page *page, 3702 enum iostat_type io_type); 3703 void f2fs_do_write_node_page(unsigned int nid, struct f2fs_io_info *fio); 3704 void f2fs_outplace_write_data(struct dnode_of_data *dn, 3705 struct f2fs_io_info *fio); 3706 int f2fs_inplace_write_data(struct f2fs_io_info *fio); 3707 void f2fs_do_replace_block(struct f2fs_sb_info *sbi, struct f2fs_summary *sum, 3708 block_t old_blkaddr, block_t new_blkaddr, 3709 bool recover_curseg, bool recover_newaddr, 3710 bool from_gc); 3711 void f2fs_replace_block(struct f2fs_sb_info *sbi, struct dnode_of_data *dn, 3712 block_t old_addr, block_t new_addr, 3713 unsigned char version, bool recover_curseg, 3714 bool recover_newaddr); 3715 void f2fs_allocate_data_block(struct f2fs_sb_info *sbi, struct page *page, 3716 block_t old_blkaddr, block_t *new_blkaddr, 3717 struct f2fs_summary *sum, int type, 3718 struct f2fs_io_info *fio); 3719 void f2fs_update_device_state(struct f2fs_sb_info *sbi, nid_t ino, 3720 block_t blkaddr, unsigned int blkcnt); 3721 void f2fs_wait_on_page_writeback(struct page *page, 3722 enum page_type type, bool ordered, bool locked); 3723 void f2fs_wait_on_block_writeback(struct inode *inode, block_t blkaddr); 3724 void f2fs_wait_on_block_writeback_range(struct inode *inode, block_t blkaddr, 3725 block_t len); 3726 void f2fs_write_data_summaries(struct f2fs_sb_info *sbi, block_t start_blk); 3727 void f2fs_write_node_summaries(struct f2fs_sb_info *sbi, block_t start_blk); 3728 int f2fs_lookup_journal_in_cursum(struct f2fs_journal *journal, int type, 3729 unsigned int val, int alloc); 3730 void f2fs_flush_sit_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3731 int f2fs_fix_curseg_write_pointer(struct f2fs_sb_info *sbi); 3732 int f2fs_check_write_pointer(struct f2fs_sb_info *sbi); 3733 int f2fs_build_segment_manager(struct f2fs_sb_info *sbi); 3734 void f2fs_destroy_segment_manager(struct f2fs_sb_info *sbi); 3735 int __init f2fs_create_segment_manager_caches(void); 3736 void f2fs_destroy_segment_manager_caches(void); 3737 int f2fs_rw_hint_to_seg_type(enum rw_hint hint); 3738 unsigned int f2fs_usable_segs_in_sec(struct f2fs_sb_info *sbi, 3739 unsigned int segno); 3740 unsigned int f2fs_usable_blks_in_seg(struct f2fs_sb_info *sbi, 3741 unsigned int segno); 3742 3743 #define DEF_FRAGMENT_SIZE 4 3744 #define MIN_FRAGMENT_SIZE 1 3745 #define MAX_FRAGMENT_SIZE 512 3746 3747 static inline bool f2fs_need_rand_seg(struct f2fs_sb_info *sbi) 3748 { 3749 return F2FS_OPTION(sbi).fs_mode == FS_MODE_FRAGMENT_SEG || 3750 F2FS_OPTION(sbi).fs_mode == FS_MODE_FRAGMENT_BLK; 3751 } 3752 3753 /* 3754 * checkpoint.c 3755 */ 3756 void f2fs_stop_checkpoint(struct f2fs_sb_info *sbi, bool end_io, 3757 unsigned char reason); 3758 void f2fs_flush_ckpt_thread(struct f2fs_sb_info *sbi); 3759 struct page *f2fs_grab_meta_page(struct f2fs_sb_info *sbi, pgoff_t index); 3760 struct page *f2fs_get_meta_page(struct f2fs_sb_info *sbi, pgoff_t index); 3761 struct page *f2fs_get_meta_page_retry(struct f2fs_sb_info *sbi, pgoff_t index); 3762 struct page *f2fs_get_tmp_page(struct f2fs_sb_info *sbi, pgoff_t index); 3763 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi, 3764 block_t blkaddr, int type); 3765 int f2fs_ra_meta_pages(struct f2fs_sb_info *sbi, block_t start, int nrpages, 3766 int type, bool sync); 3767 void f2fs_ra_meta_pages_cond(struct f2fs_sb_info *sbi, pgoff_t index, 3768 unsigned int ra_blocks); 3769 long f2fs_sync_meta_pages(struct f2fs_sb_info *sbi, enum page_type type, 3770 long nr_to_write, enum iostat_type io_type); 3771 void f2fs_add_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type); 3772 void f2fs_remove_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type); 3773 void f2fs_release_ino_entry(struct f2fs_sb_info *sbi, bool all); 3774 bool f2fs_exist_written_data(struct f2fs_sb_info *sbi, nid_t ino, int mode); 3775 void f2fs_set_dirty_device(struct f2fs_sb_info *sbi, nid_t ino, 3776 unsigned int devidx, int type); 3777 bool f2fs_is_dirty_device(struct f2fs_sb_info *sbi, nid_t ino, 3778 unsigned int devidx, int type); 3779 int f2fs_acquire_orphan_inode(struct f2fs_sb_info *sbi); 3780 void f2fs_release_orphan_inode(struct f2fs_sb_info *sbi); 3781 void f2fs_add_orphan_inode(struct inode *inode); 3782 void f2fs_remove_orphan_inode(struct f2fs_sb_info *sbi, nid_t ino); 3783 int f2fs_recover_orphan_inodes(struct f2fs_sb_info *sbi); 3784 int f2fs_get_valid_checkpoint(struct f2fs_sb_info *sbi); 3785 void f2fs_update_dirty_folio(struct inode *inode, struct folio *folio); 3786 void f2fs_remove_dirty_inode(struct inode *inode); 3787 int f2fs_sync_dirty_inodes(struct f2fs_sb_info *sbi, enum inode_type type, 3788 bool from_cp); 3789 void f2fs_wait_on_all_pages(struct f2fs_sb_info *sbi, int type); 3790 u64 f2fs_get_sectors_written(struct f2fs_sb_info *sbi); 3791 int f2fs_write_checkpoint(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3792 void f2fs_init_ino_entry_info(struct f2fs_sb_info *sbi); 3793 int __init f2fs_create_checkpoint_caches(void); 3794 void f2fs_destroy_checkpoint_caches(void); 3795 int f2fs_issue_checkpoint(struct f2fs_sb_info *sbi); 3796 int f2fs_start_ckpt_thread(struct f2fs_sb_info *sbi); 3797 void f2fs_stop_ckpt_thread(struct f2fs_sb_info *sbi); 3798 void f2fs_init_ckpt_req_control(struct f2fs_sb_info *sbi); 3799 3800 /* 3801 * data.c 3802 */ 3803 int __init f2fs_init_bioset(void); 3804 void f2fs_destroy_bioset(void); 3805 bool f2fs_is_cp_guaranteed(struct page *page); 3806 int f2fs_init_bio_entry_cache(void); 3807 void f2fs_destroy_bio_entry_cache(void); 3808 void f2fs_submit_read_bio(struct f2fs_sb_info *sbi, struct bio *bio, 3809 enum page_type type); 3810 int f2fs_init_write_merge_io(struct f2fs_sb_info *sbi); 3811 void f2fs_submit_merged_write(struct f2fs_sb_info *sbi, enum page_type type); 3812 void f2fs_submit_merged_write_cond(struct f2fs_sb_info *sbi, 3813 struct inode *inode, struct page *page, 3814 nid_t ino, enum page_type type); 3815 void f2fs_submit_merged_ipu_write(struct f2fs_sb_info *sbi, 3816 struct bio **bio, struct page *page); 3817 void f2fs_flush_merged_writes(struct f2fs_sb_info *sbi); 3818 int f2fs_submit_page_bio(struct f2fs_io_info *fio); 3819 int f2fs_merge_page_bio(struct f2fs_io_info *fio); 3820 void f2fs_submit_page_write(struct f2fs_io_info *fio); 3821 struct block_device *f2fs_target_device(struct f2fs_sb_info *sbi, 3822 block_t blk_addr, sector_t *sector); 3823 int f2fs_target_device_index(struct f2fs_sb_info *sbi, block_t blkaddr); 3824 void f2fs_set_data_blkaddr(struct dnode_of_data *dn, block_t blkaddr); 3825 void f2fs_update_data_blkaddr(struct dnode_of_data *dn, block_t blkaddr); 3826 int f2fs_reserve_new_blocks(struct dnode_of_data *dn, blkcnt_t count); 3827 int f2fs_reserve_new_block(struct dnode_of_data *dn); 3828 int f2fs_get_block_locked(struct dnode_of_data *dn, pgoff_t index); 3829 int f2fs_reserve_block(struct dnode_of_data *dn, pgoff_t index); 3830 struct page *f2fs_get_read_data_page(struct inode *inode, pgoff_t index, 3831 blk_opf_t op_flags, bool for_write, pgoff_t *next_pgofs); 3832 struct page *f2fs_find_data_page(struct inode *inode, pgoff_t index, 3833 pgoff_t *next_pgofs); 3834 struct page *f2fs_get_lock_data_page(struct inode *inode, pgoff_t index, 3835 bool for_write); 3836 struct page *f2fs_get_new_data_page(struct inode *inode, 3837 struct page *ipage, pgoff_t index, bool new_i_size); 3838 int f2fs_do_write_data_page(struct f2fs_io_info *fio); 3839 int f2fs_map_blocks(struct inode *inode, struct f2fs_map_blocks *map, int flag); 3840 int f2fs_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, 3841 u64 start, u64 len); 3842 int f2fs_encrypt_one_page(struct f2fs_io_info *fio); 3843 bool f2fs_should_update_inplace(struct inode *inode, struct f2fs_io_info *fio); 3844 bool f2fs_should_update_outplace(struct inode *inode, struct f2fs_io_info *fio); 3845 int f2fs_write_single_data_page(struct page *page, int *submitted, 3846 struct bio **bio, sector_t *last_block, 3847 struct writeback_control *wbc, 3848 enum iostat_type io_type, 3849 int compr_blocks, bool allow_balance); 3850 void f2fs_write_failed(struct inode *inode, loff_t to); 3851 void f2fs_invalidate_folio(struct folio *folio, size_t offset, size_t length); 3852 bool f2fs_release_folio(struct folio *folio, gfp_t wait); 3853 bool f2fs_overwrite_io(struct inode *inode, loff_t pos, size_t len); 3854 void f2fs_clear_page_cache_dirty_tag(struct page *page); 3855 int f2fs_init_post_read_processing(void); 3856 void f2fs_destroy_post_read_processing(void); 3857 int f2fs_init_post_read_wq(struct f2fs_sb_info *sbi); 3858 void f2fs_destroy_post_read_wq(struct f2fs_sb_info *sbi); 3859 extern const struct iomap_ops f2fs_iomap_ops; 3860 3861 /* 3862 * gc.c 3863 */ 3864 int f2fs_start_gc_thread(struct f2fs_sb_info *sbi); 3865 void f2fs_stop_gc_thread(struct f2fs_sb_info *sbi); 3866 block_t f2fs_start_bidx_of_node(unsigned int node_ofs, struct inode *inode); 3867 int f2fs_gc(struct f2fs_sb_info *sbi, struct f2fs_gc_control *gc_control); 3868 void f2fs_build_gc_manager(struct f2fs_sb_info *sbi); 3869 int f2fs_gc_range(struct f2fs_sb_info *sbi, 3870 unsigned int start_seg, unsigned int end_seg, 3871 bool dry_run, unsigned int dry_run_sections); 3872 int f2fs_resize_fs(struct file *filp, __u64 block_count); 3873 int __init f2fs_create_garbage_collection_cache(void); 3874 void f2fs_destroy_garbage_collection_cache(void); 3875 /* victim selection function for cleaning and SSR */ 3876 int f2fs_get_victim(struct f2fs_sb_info *sbi, unsigned int *result, 3877 int gc_type, int type, char alloc_mode, 3878 unsigned long long age); 3879 3880 /* 3881 * recovery.c 3882 */ 3883 int f2fs_recover_fsync_data(struct f2fs_sb_info *sbi, bool check_only); 3884 bool f2fs_space_for_roll_forward(struct f2fs_sb_info *sbi); 3885 int __init f2fs_create_recovery_cache(void); 3886 void f2fs_destroy_recovery_cache(void); 3887 3888 /* 3889 * debug.c 3890 */ 3891 #ifdef CONFIG_F2FS_STAT_FS 3892 struct f2fs_stat_info { 3893 struct list_head stat_list; 3894 struct f2fs_sb_info *sbi; 3895 int all_area_segs, sit_area_segs, nat_area_segs, ssa_area_segs; 3896 int main_area_segs, main_area_sections, main_area_zones; 3897 unsigned long long hit_cached[NR_EXTENT_CACHES]; 3898 unsigned long long hit_rbtree[NR_EXTENT_CACHES]; 3899 unsigned long long total_ext[NR_EXTENT_CACHES]; 3900 unsigned long long hit_total[NR_EXTENT_CACHES]; 3901 int ext_tree[NR_EXTENT_CACHES]; 3902 int zombie_tree[NR_EXTENT_CACHES]; 3903 int ext_node[NR_EXTENT_CACHES]; 3904 /* to count memory footprint */ 3905 unsigned long long ext_mem[NR_EXTENT_CACHES]; 3906 /* for read extent cache */ 3907 unsigned long long hit_largest; 3908 /* for block age extent cache */ 3909 unsigned long long allocated_data_blocks; 3910 int ndirty_node, ndirty_dent, ndirty_meta, ndirty_imeta; 3911 int ndirty_data, ndirty_qdata; 3912 unsigned int ndirty_dirs, ndirty_files, nquota_files, ndirty_all; 3913 int nats, dirty_nats, sits, dirty_sits; 3914 int free_nids, avail_nids, alloc_nids; 3915 int total_count, utilization; 3916 int nr_wb_cp_data, nr_wb_data; 3917 int nr_rd_data, nr_rd_node, nr_rd_meta; 3918 int nr_dio_read, nr_dio_write; 3919 unsigned int io_skip_bggc, other_skip_bggc; 3920 int nr_flushing, nr_flushed, flush_list_empty; 3921 int nr_discarding, nr_discarded; 3922 int nr_discard_cmd; 3923 unsigned int undiscard_blks; 3924 int nr_issued_ckpt, nr_total_ckpt, nr_queued_ckpt; 3925 unsigned int cur_ckpt_time, peak_ckpt_time; 3926 int inline_xattr, inline_inode, inline_dir, append, update, orphans; 3927 int compr_inode, swapfile_inode; 3928 unsigned long long compr_blocks; 3929 int aw_cnt, max_aw_cnt; 3930 unsigned int valid_count, valid_node_count, valid_inode_count, discard_blks; 3931 unsigned int bimodal, avg_vblocks; 3932 int util_free, util_valid, util_invalid; 3933 int rsvd_segs, overp_segs; 3934 int dirty_count, node_pages, meta_pages, compress_pages; 3935 int compress_page_hit; 3936 int prefree_count, free_segs, free_secs; 3937 int cp_call_count[MAX_CALL_TYPE], cp_count; 3938 int gc_call_count[MAX_CALL_TYPE]; 3939 int gc_segs[2][2]; 3940 int gc_secs[2][2]; 3941 int tot_blks, data_blks, node_blks; 3942 int bg_data_blks, bg_node_blks; 3943 int curseg[NR_CURSEG_TYPE]; 3944 int cursec[NR_CURSEG_TYPE]; 3945 int curzone[NR_CURSEG_TYPE]; 3946 unsigned int dirty_seg[NR_CURSEG_TYPE]; 3947 unsigned int full_seg[NR_CURSEG_TYPE]; 3948 unsigned int valid_blks[NR_CURSEG_TYPE]; 3949 3950 unsigned int meta_count[META_MAX]; 3951 unsigned int segment_count[2]; 3952 unsigned int block_count[2]; 3953 unsigned int inplace_count; 3954 unsigned long long base_mem, cache_mem, page_mem; 3955 }; 3956 3957 static inline struct f2fs_stat_info *F2FS_STAT(struct f2fs_sb_info *sbi) 3958 { 3959 return (struct f2fs_stat_info *)sbi->stat_info; 3960 } 3961 3962 #define stat_inc_cp_call_count(sbi, foreground) \ 3963 atomic_inc(&sbi->cp_call_count[(foreground)]) 3964 #define stat_inc_cp_count(si) (F2FS_STAT(sbi)->cp_count++) 3965 #define stat_io_skip_bggc_count(sbi) ((sbi)->io_skip_bggc++) 3966 #define stat_other_skip_bggc_count(sbi) ((sbi)->other_skip_bggc++) 3967 #define stat_inc_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]++) 3968 #define stat_dec_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]--) 3969 #define stat_inc_total_hit(sbi, type) (atomic64_inc(&(sbi)->total_hit_ext[type])) 3970 #define stat_inc_rbtree_node_hit(sbi, type) (atomic64_inc(&(sbi)->read_hit_rbtree[type])) 3971 #define stat_inc_largest_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_largest)) 3972 #define stat_inc_cached_node_hit(sbi, type) (atomic64_inc(&(sbi)->read_hit_cached[type])) 3973 #define stat_inc_inline_xattr(inode) \ 3974 do { \ 3975 if (f2fs_has_inline_xattr(inode)) \ 3976 (atomic_inc(&F2FS_I_SB(inode)->inline_xattr)); \ 3977 } while (0) 3978 #define stat_dec_inline_xattr(inode) \ 3979 do { \ 3980 if (f2fs_has_inline_xattr(inode)) \ 3981 (atomic_dec(&F2FS_I_SB(inode)->inline_xattr)); \ 3982 } while (0) 3983 #define stat_inc_inline_inode(inode) \ 3984 do { \ 3985 if (f2fs_has_inline_data(inode)) \ 3986 (atomic_inc(&F2FS_I_SB(inode)->inline_inode)); \ 3987 } while (0) 3988 #define stat_dec_inline_inode(inode) \ 3989 do { \ 3990 if (f2fs_has_inline_data(inode)) \ 3991 (atomic_dec(&F2FS_I_SB(inode)->inline_inode)); \ 3992 } while (0) 3993 #define stat_inc_inline_dir(inode) \ 3994 do { \ 3995 if (f2fs_has_inline_dentry(inode)) \ 3996 (atomic_inc(&F2FS_I_SB(inode)->inline_dir)); \ 3997 } while (0) 3998 #define stat_dec_inline_dir(inode) \ 3999 do { \ 4000 if (f2fs_has_inline_dentry(inode)) \ 4001 (atomic_dec(&F2FS_I_SB(inode)->inline_dir)); \ 4002 } while (0) 4003 #define stat_inc_compr_inode(inode) \ 4004 do { \ 4005 if (f2fs_compressed_file(inode)) \ 4006 (atomic_inc(&F2FS_I_SB(inode)->compr_inode)); \ 4007 } while (0) 4008 #define stat_dec_compr_inode(inode) \ 4009 do { \ 4010 if (f2fs_compressed_file(inode)) \ 4011 (atomic_dec(&F2FS_I_SB(inode)->compr_inode)); \ 4012 } while (0) 4013 #define stat_add_compr_blocks(inode, blocks) \ 4014 (atomic64_add(blocks, &F2FS_I_SB(inode)->compr_blocks)) 4015 #define stat_sub_compr_blocks(inode, blocks) \ 4016 (atomic64_sub(blocks, &F2FS_I_SB(inode)->compr_blocks)) 4017 #define stat_inc_swapfile_inode(inode) \ 4018 (atomic_inc(&F2FS_I_SB(inode)->swapfile_inode)) 4019 #define stat_dec_swapfile_inode(inode) \ 4020 (atomic_dec(&F2FS_I_SB(inode)->swapfile_inode)) 4021 #define stat_inc_atomic_inode(inode) \ 4022 (atomic_inc(&F2FS_I_SB(inode)->atomic_files)) 4023 #define stat_dec_atomic_inode(inode) \ 4024 (atomic_dec(&F2FS_I_SB(inode)->atomic_files)) 4025 #define stat_inc_meta_count(sbi, blkaddr) \ 4026 do { \ 4027 if (blkaddr < SIT_I(sbi)->sit_base_addr) \ 4028 atomic_inc(&(sbi)->meta_count[META_CP]); \ 4029 else if (blkaddr < NM_I(sbi)->nat_blkaddr) \ 4030 atomic_inc(&(sbi)->meta_count[META_SIT]); \ 4031 else if (blkaddr < SM_I(sbi)->ssa_blkaddr) \ 4032 atomic_inc(&(sbi)->meta_count[META_NAT]); \ 4033 else if (blkaddr < SM_I(sbi)->main_blkaddr) \ 4034 atomic_inc(&(sbi)->meta_count[META_SSA]); \ 4035 } while (0) 4036 #define stat_inc_seg_type(sbi, curseg) \ 4037 ((sbi)->segment_count[(curseg)->alloc_type]++) 4038 #define stat_inc_block_count(sbi, curseg) \ 4039 ((sbi)->block_count[(curseg)->alloc_type]++) 4040 #define stat_inc_inplace_blocks(sbi) \ 4041 (atomic_inc(&(sbi)->inplace_count)) 4042 #define stat_update_max_atomic_write(inode) \ 4043 do { \ 4044 int cur = atomic_read(&F2FS_I_SB(inode)->atomic_files); \ 4045 int max = atomic_read(&F2FS_I_SB(inode)->max_aw_cnt); \ 4046 if (cur > max) \ 4047 atomic_set(&F2FS_I_SB(inode)->max_aw_cnt, cur); \ 4048 } while (0) 4049 #define stat_inc_gc_call_count(sbi, foreground) \ 4050 (F2FS_STAT(sbi)->gc_call_count[(foreground)]++) 4051 #define stat_inc_gc_sec_count(sbi, type, gc_type) \ 4052 (F2FS_STAT(sbi)->gc_secs[(type)][(gc_type)]++) 4053 #define stat_inc_gc_seg_count(sbi, type, gc_type) \ 4054 (F2FS_STAT(sbi)->gc_segs[(type)][(gc_type)]++) 4055 4056 #define stat_inc_tot_blk_count(si, blks) \ 4057 ((si)->tot_blks += (blks)) 4058 4059 #define stat_inc_data_blk_count(sbi, blks, gc_type) \ 4060 do { \ 4061 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 4062 stat_inc_tot_blk_count(si, blks); \ 4063 si->data_blks += (blks); \ 4064 si->bg_data_blks += ((gc_type) == BG_GC) ? (blks) : 0; \ 4065 } while (0) 4066 4067 #define stat_inc_node_blk_count(sbi, blks, gc_type) \ 4068 do { \ 4069 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 4070 stat_inc_tot_blk_count(si, blks); \ 4071 si->node_blks += (blks); \ 4072 si->bg_node_blks += ((gc_type) == BG_GC) ? (blks) : 0; \ 4073 } while (0) 4074 4075 int f2fs_build_stats(struct f2fs_sb_info *sbi); 4076 void f2fs_destroy_stats(struct f2fs_sb_info *sbi); 4077 void __init f2fs_create_root_stats(void); 4078 void f2fs_destroy_root_stats(void); 4079 void f2fs_update_sit_info(struct f2fs_sb_info *sbi); 4080 #else 4081 #define stat_inc_cp_call_count(sbi, foreground) do { } while (0) 4082 #define stat_inc_cp_count(sbi) do { } while (0) 4083 #define stat_io_skip_bggc_count(sbi) do { } while (0) 4084 #define stat_other_skip_bggc_count(sbi) do { } while (0) 4085 #define stat_inc_dirty_inode(sbi, type) do { } while (0) 4086 #define stat_dec_dirty_inode(sbi, type) do { } while (0) 4087 #define stat_inc_total_hit(sbi, type) do { } while (0) 4088 #define stat_inc_rbtree_node_hit(sbi, type) do { } while (0) 4089 #define stat_inc_largest_node_hit(sbi) do { } while (0) 4090 #define stat_inc_cached_node_hit(sbi, type) do { } while (0) 4091 #define stat_inc_inline_xattr(inode) do { } while (0) 4092 #define stat_dec_inline_xattr(inode) do { } while (0) 4093 #define stat_inc_inline_inode(inode) do { } while (0) 4094 #define stat_dec_inline_inode(inode) do { } while (0) 4095 #define stat_inc_inline_dir(inode) do { } while (0) 4096 #define stat_dec_inline_dir(inode) do { } while (0) 4097 #define stat_inc_compr_inode(inode) do { } while (0) 4098 #define stat_dec_compr_inode(inode) do { } while (0) 4099 #define stat_add_compr_blocks(inode, blocks) do { } while (0) 4100 #define stat_sub_compr_blocks(inode, blocks) do { } while (0) 4101 #define stat_inc_swapfile_inode(inode) do { } while (0) 4102 #define stat_dec_swapfile_inode(inode) do { } while (0) 4103 #define stat_inc_atomic_inode(inode) do { } while (0) 4104 #define stat_dec_atomic_inode(inode) do { } while (0) 4105 #define stat_update_max_atomic_write(inode) do { } while (0) 4106 #define stat_inc_meta_count(sbi, blkaddr) do { } while (0) 4107 #define stat_inc_seg_type(sbi, curseg) do { } while (0) 4108 #define stat_inc_block_count(sbi, curseg) do { } while (0) 4109 #define stat_inc_inplace_blocks(sbi) do { } while (0) 4110 #define stat_inc_gc_call_count(sbi, foreground) do { } while (0) 4111 #define stat_inc_gc_sec_count(sbi, type, gc_type) do { } while (0) 4112 #define stat_inc_gc_seg_count(sbi, type, gc_type) do { } while (0) 4113 #define stat_inc_tot_blk_count(si, blks) do { } while (0) 4114 #define stat_inc_data_blk_count(sbi, blks, gc_type) do { } while (0) 4115 #define stat_inc_node_blk_count(sbi, blks, gc_type) do { } while (0) 4116 4117 static inline int f2fs_build_stats(struct f2fs_sb_info *sbi) { return 0; } 4118 static inline void f2fs_destroy_stats(struct f2fs_sb_info *sbi) { } 4119 static inline void __init f2fs_create_root_stats(void) { } 4120 static inline void f2fs_destroy_root_stats(void) { } 4121 static inline void f2fs_update_sit_info(struct f2fs_sb_info *sbi) {} 4122 #endif 4123 4124 extern const struct file_operations f2fs_dir_operations; 4125 extern const struct file_operations f2fs_file_operations; 4126 extern const struct inode_operations f2fs_file_inode_operations; 4127 extern const struct address_space_operations f2fs_dblock_aops; 4128 extern const struct address_space_operations f2fs_node_aops; 4129 extern const struct address_space_operations f2fs_meta_aops; 4130 extern const struct inode_operations f2fs_dir_inode_operations; 4131 extern const struct inode_operations f2fs_symlink_inode_operations; 4132 extern const struct inode_operations f2fs_encrypted_symlink_inode_operations; 4133 extern const struct inode_operations f2fs_special_inode_operations; 4134 extern struct kmem_cache *f2fs_inode_entry_slab; 4135 4136 /* 4137 * inline.c 4138 */ 4139 bool f2fs_may_inline_data(struct inode *inode); 4140 bool f2fs_sanity_check_inline_data(struct inode *inode); 4141 bool f2fs_may_inline_dentry(struct inode *inode); 4142 void f2fs_do_read_inline_data(struct page *page, struct page *ipage); 4143 void f2fs_truncate_inline_inode(struct inode *inode, 4144 struct page *ipage, u64 from); 4145 int f2fs_read_inline_data(struct inode *inode, struct page *page); 4146 int f2fs_convert_inline_page(struct dnode_of_data *dn, struct page *page); 4147 int f2fs_convert_inline_inode(struct inode *inode); 4148 int f2fs_try_convert_inline_dir(struct inode *dir, struct dentry *dentry); 4149 int f2fs_write_inline_data(struct inode *inode, struct page *page); 4150 int f2fs_recover_inline_data(struct inode *inode, struct page *npage); 4151 struct f2fs_dir_entry *f2fs_find_in_inline_dir(struct inode *dir, 4152 const struct f2fs_filename *fname, 4153 struct page **res_page); 4154 int f2fs_make_empty_inline_dir(struct inode *inode, struct inode *parent, 4155 struct page *ipage); 4156 int f2fs_add_inline_entry(struct inode *dir, const struct f2fs_filename *fname, 4157 struct inode *inode, nid_t ino, umode_t mode); 4158 void f2fs_delete_inline_entry(struct f2fs_dir_entry *dentry, 4159 struct page *page, struct inode *dir, 4160 struct inode *inode); 4161 bool f2fs_empty_inline_dir(struct inode *dir); 4162 int f2fs_read_inline_dir(struct file *file, struct dir_context *ctx, 4163 struct fscrypt_str *fstr); 4164 int f2fs_inline_data_fiemap(struct inode *inode, 4165 struct fiemap_extent_info *fieinfo, 4166 __u64 start, __u64 len); 4167 4168 /* 4169 * shrinker.c 4170 */ 4171 unsigned long f2fs_shrink_count(struct shrinker *shrink, 4172 struct shrink_control *sc); 4173 unsigned long f2fs_shrink_scan(struct shrinker *shrink, 4174 struct shrink_control *sc); 4175 void f2fs_join_shrinker(struct f2fs_sb_info *sbi); 4176 void f2fs_leave_shrinker(struct f2fs_sb_info *sbi); 4177 4178 /* 4179 * extent_cache.c 4180 */ 4181 bool sanity_check_extent_cache(struct inode *inode); 4182 void f2fs_init_extent_tree(struct inode *inode); 4183 void f2fs_drop_extent_tree(struct inode *inode); 4184 void f2fs_destroy_extent_node(struct inode *inode); 4185 void f2fs_destroy_extent_tree(struct inode *inode); 4186 void f2fs_init_extent_cache_info(struct f2fs_sb_info *sbi); 4187 int __init f2fs_create_extent_cache(void); 4188 void f2fs_destroy_extent_cache(void); 4189 4190 /* read extent cache ops */ 4191 void f2fs_init_read_extent_tree(struct inode *inode, struct page *ipage); 4192 bool f2fs_lookup_read_extent_cache(struct inode *inode, pgoff_t pgofs, 4193 struct extent_info *ei); 4194 bool f2fs_lookup_read_extent_cache_block(struct inode *inode, pgoff_t index, 4195 block_t *blkaddr); 4196 void f2fs_update_read_extent_cache(struct dnode_of_data *dn); 4197 void f2fs_update_read_extent_cache_range(struct dnode_of_data *dn, 4198 pgoff_t fofs, block_t blkaddr, unsigned int len); 4199 unsigned int f2fs_shrink_read_extent_tree(struct f2fs_sb_info *sbi, 4200 int nr_shrink); 4201 4202 /* block age extent cache ops */ 4203 void f2fs_init_age_extent_tree(struct inode *inode); 4204 bool f2fs_lookup_age_extent_cache(struct inode *inode, pgoff_t pgofs, 4205 struct extent_info *ei); 4206 void f2fs_update_age_extent_cache(struct dnode_of_data *dn); 4207 void f2fs_update_age_extent_cache_range(struct dnode_of_data *dn, 4208 pgoff_t fofs, unsigned int len); 4209 unsigned int f2fs_shrink_age_extent_tree(struct f2fs_sb_info *sbi, 4210 int nr_shrink); 4211 4212 /* 4213 * sysfs.c 4214 */ 4215 #define MIN_RA_MUL 2 4216 #define MAX_RA_MUL 256 4217 4218 int __init f2fs_init_sysfs(void); 4219 void f2fs_exit_sysfs(void); 4220 int f2fs_register_sysfs(struct f2fs_sb_info *sbi); 4221 void f2fs_unregister_sysfs(struct f2fs_sb_info *sbi); 4222 4223 /* verity.c */ 4224 extern const struct fsverity_operations f2fs_verityops; 4225 4226 /* 4227 * crypto support 4228 */ 4229 static inline bool f2fs_encrypted_file(struct inode *inode) 4230 { 4231 return IS_ENCRYPTED(inode) && S_ISREG(inode->i_mode); 4232 } 4233 4234 static inline void f2fs_set_encrypted_inode(struct inode *inode) 4235 { 4236 #ifdef CONFIG_FS_ENCRYPTION 4237 file_set_encrypt(inode); 4238 f2fs_set_inode_flags(inode); 4239 #endif 4240 } 4241 4242 /* 4243 * Returns true if the reads of the inode's data need to undergo some 4244 * postprocessing step, like decryption or authenticity verification. 4245 */ 4246 static inline bool f2fs_post_read_required(struct inode *inode) 4247 { 4248 return f2fs_encrypted_file(inode) || fsverity_active(inode) || 4249 f2fs_compressed_file(inode); 4250 } 4251 4252 /* 4253 * compress.c 4254 */ 4255 #ifdef CONFIG_F2FS_FS_COMPRESSION 4256 bool f2fs_is_compressed_page(struct page *page); 4257 struct page *f2fs_compress_control_page(struct page *page); 4258 int f2fs_prepare_compress_overwrite(struct inode *inode, 4259 struct page **pagep, pgoff_t index, void **fsdata); 4260 bool f2fs_compress_write_end(struct inode *inode, void *fsdata, 4261 pgoff_t index, unsigned copied); 4262 int f2fs_truncate_partial_cluster(struct inode *inode, u64 from, bool lock); 4263 void f2fs_compress_write_end_io(struct bio *bio, struct page *page); 4264 bool f2fs_is_compress_backend_ready(struct inode *inode); 4265 bool f2fs_is_compress_level_valid(int alg, int lvl); 4266 int __init f2fs_init_compress_mempool(void); 4267 void f2fs_destroy_compress_mempool(void); 4268 void f2fs_decompress_cluster(struct decompress_io_ctx *dic, bool in_task); 4269 void f2fs_end_read_compressed_page(struct page *page, bool failed, 4270 block_t blkaddr, bool in_task); 4271 bool f2fs_cluster_is_empty(struct compress_ctx *cc); 4272 bool f2fs_cluster_can_merge_page(struct compress_ctx *cc, pgoff_t index); 4273 bool f2fs_all_cluster_page_ready(struct compress_ctx *cc, struct page **pages, 4274 int index, int nr_pages, bool uptodate); 4275 bool f2fs_sanity_check_cluster(struct dnode_of_data *dn); 4276 void f2fs_compress_ctx_add_page(struct compress_ctx *cc, struct page *page); 4277 int f2fs_write_multi_pages(struct compress_ctx *cc, 4278 int *submitted, 4279 struct writeback_control *wbc, 4280 enum iostat_type io_type); 4281 int f2fs_is_compressed_cluster(struct inode *inode, pgoff_t index); 4282 void f2fs_update_read_extent_tree_range_compressed(struct inode *inode, 4283 pgoff_t fofs, block_t blkaddr, 4284 unsigned int llen, unsigned int c_len); 4285 int f2fs_read_multi_pages(struct compress_ctx *cc, struct bio **bio_ret, 4286 unsigned nr_pages, sector_t *last_block_in_bio, 4287 bool is_readahead, bool for_write); 4288 struct decompress_io_ctx *f2fs_alloc_dic(struct compress_ctx *cc); 4289 void f2fs_decompress_end_io(struct decompress_io_ctx *dic, bool failed, 4290 bool in_task); 4291 void f2fs_put_page_dic(struct page *page, bool in_task); 4292 unsigned int f2fs_cluster_blocks_are_contiguous(struct dnode_of_data *dn, 4293 unsigned int ofs_in_node); 4294 int f2fs_init_compress_ctx(struct compress_ctx *cc); 4295 void f2fs_destroy_compress_ctx(struct compress_ctx *cc, bool reuse); 4296 void f2fs_init_compress_info(struct f2fs_sb_info *sbi); 4297 int f2fs_init_compress_inode(struct f2fs_sb_info *sbi); 4298 void f2fs_destroy_compress_inode(struct f2fs_sb_info *sbi); 4299 int f2fs_init_page_array_cache(struct f2fs_sb_info *sbi); 4300 void f2fs_destroy_page_array_cache(struct f2fs_sb_info *sbi); 4301 int __init f2fs_init_compress_cache(void); 4302 void f2fs_destroy_compress_cache(void); 4303 struct address_space *COMPRESS_MAPPING(struct f2fs_sb_info *sbi); 4304 void f2fs_invalidate_compress_page(struct f2fs_sb_info *sbi, block_t blkaddr); 4305 void f2fs_cache_compressed_page(struct f2fs_sb_info *sbi, struct page *page, 4306 nid_t ino, block_t blkaddr); 4307 bool f2fs_load_compressed_page(struct f2fs_sb_info *sbi, struct page *page, 4308 block_t blkaddr); 4309 void f2fs_invalidate_compress_pages(struct f2fs_sb_info *sbi, nid_t ino); 4310 #define inc_compr_inode_stat(inode) \ 4311 do { \ 4312 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); \ 4313 sbi->compr_new_inode++; \ 4314 } while (0) 4315 #define add_compr_block_stat(inode, blocks) \ 4316 do { \ 4317 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); \ 4318 int diff = F2FS_I(inode)->i_cluster_size - blocks; \ 4319 sbi->compr_written_block += blocks; \ 4320 sbi->compr_saved_block += diff; \ 4321 } while (0) 4322 #else 4323 static inline bool f2fs_is_compressed_page(struct page *page) { return false; } 4324 static inline bool f2fs_is_compress_backend_ready(struct inode *inode) 4325 { 4326 if (!f2fs_compressed_file(inode)) 4327 return true; 4328 /* not support compression */ 4329 return false; 4330 } 4331 static inline bool f2fs_is_compress_level_valid(int alg, int lvl) { return false; } 4332 static inline struct page *f2fs_compress_control_page(struct page *page) 4333 { 4334 WARN_ON_ONCE(1); 4335 return ERR_PTR(-EINVAL); 4336 } 4337 static inline int __init f2fs_init_compress_mempool(void) { return 0; } 4338 static inline void f2fs_destroy_compress_mempool(void) { } 4339 static inline void f2fs_decompress_cluster(struct decompress_io_ctx *dic, 4340 bool in_task) { } 4341 static inline void f2fs_end_read_compressed_page(struct page *page, 4342 bool failed, block_t blkaddr, bool in_task) 4343 { 4344 WARN_ON_ONCE(1); 4345 } 4346 static inline void f2fs_put_page_dic(struct page *page, bool in_task) 4347 { 4348 WARN_ON_ONCE(1); 4349 } 4350 static inline unsigned int f2fs_cluster_blocks_are_contiguous( 4351 struct dnode_of_data *dn, unsigned int ofs_in_node) { return 0; } 4352 static inline bool f2fs_sanity_check_cluster(struct dnode_of_data *dn) { return false; } 4353 static inline int f2fs_init_compress_inode(struct f2fs_sb_info *sbi) { return 0; } 4354 static inline void f2fs_destroy_compress_inode(struct f2fs_sb_info *sbi) { } 4355 static inline int f2fs_init_page_array_cache(struct f2fs_sb_info *sbi) { return 0; } 4356 static inline void f2fs_destroy_page_array_cache(struct f2fs_sb_info *sbi) { } 4357 static inline int __init f2fs_init_compress_cache(void) { return 0; } 4358 static inline void f2fs_destroy_compress_cache(void) { } 4359 static inline void f2fs_invalidate_compress_page(struct f2fs_sb_info *sbi, 4360 block_t blkaddr) { } 4361 static inline void f2fs_cache_compressed_page(struct f2fs_sb_info *sbi, 4362 struct page *page, nid_t ino, block_t blkaddr) { } 4363 static inline bool f2fs_load_compressed_page(struct f2fs_sb_info *sbi, 4364 struct page *page, block_t blkaddr) { return false; } 4365 static inline void f2fs_invalidate_compress_pages(struct f2fs_sb_info *sbi, 4366 nid_t ino) { } 4367 #define inc_compr_inode_stat(inode) do { } while (0) 4368 static inline void f2fs_update_read_extent_tree_range_compressed( 4369 struct inode *inode, 4370 pgoff_t fofs, block_t blkaddr, 4371 unsigned int llen, unsigned int c_len) { } 4372 #endif 4373 4374 static inline int set_compress_context(struct inode *inode) 4375 { 4376 #ifdef CONFIG_F2FS_FS_COMPRESSION 4377 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 4378 4379 F2FS_I(inode)->i_compress_algorithm = 4380 F2FS_OPTION(sbi).compress_algorithm; 4381 F2FS_I(inode)->i_log_cluster_size = 4382 F2FS_OPTION(sbi).compress_log_size; 4383 F2FS_I(inode)->i_compress_flag = 4384 F2FS_OPTION(sbi).compress_chksum ? 4385 BIT(COMPRESS_CHKSUM) : 0; 4386 F2FS_I(inode)->i_cluster_size = 4387 BIT(F2FS_I(inode)->i_log_cluster_size); 4388 if ((F2FS_I(inode)->i_compress_algorithm == COMPRESS_LZ4 || 4389 F2FS_I(inode)->i_compress_algorithm == COMPRESS_ZSTD) && 4390 F2FS_OPTION(sbi).compress_level) 4391 F2FS_I(inode)->i_compress_level = 4392 F2FS_OPTION(sbi).compress_level; 4393 F2FS_I(inode)->i_flags |= F2FS_COMPR_FL; 4394 set_inode_flag(inode, FI_COMPRESSED_FILE); 4395 stat_inc_compr_inode(inode); 4396 inc_compr_inode_stat(inode); 4397 f2fs_mark_inode_dirty_sync(inode, true); 4398 return 0; 4399 #else 4400 return -EOPNOTSUPP; 4401 #endif 4402 } 4403 4404 static inline bool f2fs_disable_compressed_file(struct inode *inode) 4405 { 4406 struct f2fs_inode_info *fi = F2FS_I(inode); 4407 4408 f2fs_down_write(&F2FS_I(inode)->i_sem); 4409 4410 if (!f2fs_compressed_file(inode)) { 4411 f2fs_up_write(&F2FS_I(inode)->i_sem); 4412 return true; 4413 } 4414 if (f2fs_is_mmap_file(inode) || 4415 (S_ISREG(inode->i_mode) && F2FS_HAS_BLOCKS(inode))) { 4416 f2fs_up_write(&F2FS_I(inode)->i_sem); 4417 return false; 4418 } 4419 4420 fi->i_flags &= ~F2FS_COMPR_FL; 4421 stat_dec_compr_inode(inode); 4422 clear_inode_flag(inode, FI_COMPRESSED_FILE); 4423 f2fs_mark_inode_dirty_sync(inode, true); 4424 4425 f2fs_up_write(&F2FS_I(inode)->i_sem); 4426 return true; 4427 } 4428 4429 #define F2FS_FEATURE_FUNCS(name, flagname) \ 4430 static inline bool f2fs_sb_has_##name(struct f2fs_sb_info *sbi) \ 4431 { \ 4432 return F2FS_HAS_FEATURE(sbi, F2FS_FEATURE_##flagname); \ 4433 } 4434 4435 F2FS_FEATURE_FUNCS(encrypt, ENCRYPT); 4436 F2FS_FEATURE_FUNCS(blkzoned, BLKZONED); 4437 F2FS_FEATURE_FUNCS(extra_attr, EXTRA_ATTR); 4438 F2FS_FEATURE_FUNCS(project_quota, PRJQUOTA); 4439 F2FS_FEATURE_FUNCS(inode_chksum, INODE_CHKSUM); 4440 F2FS_FEATURE_FUNCS(flexible_inline_xattr, FLEXIBLE_INLINE_XATTR); 4441 F2FS_FEATURE_FUNCS(quota_ino, QUOTA_INO); 4442 F2FS_FEATURE_FUNCS(inode_crtime, INODE_CRTIME); 4443 F2FS_FEATURE_FUNCS(lost_found, LOST_FOUND); 4444 F2FS_FEATURE_FUNCS(verity, VERITY); 4445 F2FS_FEATURE_FUNCS(sb_chksum, SB_CHKSUM); 4446 F2FS_FEATURE_FUNCS(casefold, CASEFOLD); 4447 F2FS_FEATURE_FUNCS(compression, COMPRESSION); 4448 F2FS_FEATURE_FUNCS(readonly, RO); 4449 4450 #ifdef CONFIG_BLK_DEV_ZONED 4451 static inline bool f2fs_blkz_is_seq(struct f2fs_sb_info *sbi, int devi, 4452 block_t blkaddr) 4453 { 4454 unsigned int zno = blkaddr / sbi->blocks_per_blkz; 4455 4456 return test_bit(zno, FDEV(devi).blkz_seq); 4457 } 4458 #endif 4459 4460 static inline int f2fs_bdev_index(struct f2fs_sb_info *sbi, 4461 struct block_device *bdev) 4462 { 4463 int i; 4464 4465 if (!f2fs_is_multi_device(sbi)) 4466 return 0; 4467 4468 for (i = 0; i < sbi->s_ndevs; i++) 4469 if (FDEV(i).bdev == bdev) 4470 return i; 4471 4472 WARN_ON(1); 4473 return -1; 4474 } 4475 4476 static inline bool f2fs_hw_should_discard(struct f2fs_sb_info *sbi) 4477 { 4478 return f2fs_sb_has_blkzoned(sbi); 4479 } 4480 4481 static inline bool f2fs_bdev_support_discard(struct block_device *bdev) 4482 { 4483 return bdev_max_discard_sectors(bdev) || bdev_is_zoned(bdev); 4484 } 4485 4486 static inline bool f2fs_hw_support_discard(struct f2fs_sb_info *sbi) 4487 { 4488 int i; 4489 4490 if (!f2fs_is_multi_device(sbi)) 4491 return f2fs_bdev_support_discard(sbi->sb->s_bdev); 4492 4493 for (i = 0; i < sbi->s_ndevs; i++) 4494 if (f2fs_bdev_support_discard(FDEV(i).bdev)) 4495 return true; 4496 return false; 4497 } 4498 4499 static inline bool f2fs_realtime_discard_enable(struct f2fs_sb_info *sbi) 4500 { 4501 return (test_opt(sbi, DISCARD) && f2fs_hw_support_discard(sbi)) || 4502 f2fs_hw_should_discard(sbi); 4503 } 4504 4505 static inline bool f2fs_hw_is_readonly(struct f2fs_sb_info *sbi) 4506 { 4507 int i; 4508 4509 if (!f2fs_is_multi_device(sbi)) 4510 return bdev_read_only(sbi->sb->s_bdev); 4511 4512 for (i = 0; i < sbi->s_ndevs; i++) 4513 if (bdev_read_only(FDEV(i).bdev)) 4514 return true; 4515 return false; 4516 } 4517 4518 static inline bool f2fs_dev_is_readonly(struct f2fs_sb_info *sbi) 4519 { 4520 return f2fs_sb_has_readonly(sbi) || f2fs_hw_is_readonly(sbi); 4521 } 4522 4523 static inline bool f2fs_lfs_mode(struct f2fs_sb_info *sbi) 4524 { 4525 return F2FS_OPTION(sbi).fs_mode == FS_MODE_LFS; 4526 } 4527 4528 static inline bool f2fs_valid_pinned_area(struct f2fs_sb_info *sbi, 4529 block_t blkaddr) 4530 { 4531 if (f2fs_sb_has_blkzoned(sbi)) { 4532 int devi = f2fs_target_device_index(sbi, blkaddr); 4533 4534 return !bdev_is_zoned(FDEV(devi).bdev); 4535 } 4536 return true; 4537 } 4538 4539 static inline bool f2fs_low_mem_mode(struct f2fs_sb_info *sbi) 4540 { 4541 return F2FS_OPTION(sbi).memory_mode == MEMORY_MODE_LOW; 4542 } 4543 4544 static inline bool f2fs_may_compress(struct inode *inode) 4545 { 4546 if (IS_SWAPFILE(inode) || f2fs_is_pinned_file(inode) || 4547 f2fs_is_atomic_file(inode) || f2fs_has_inline_data(inode) || 4548 f2fs_is_mmap_file(inode)) 4549 return false; 4550 return S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode); 4551 } 4552 4553 static inline void f2fs_i_compr_blocks_update(struct inode *inode, 4554 u64 blocks, bool add) 4555 { 4556 struct f2fs_inode_info *fi = F2FS_I(inode); 4557 int diff = fi->i_cluster_size - blocks; 4558 4559 /* don't update i_compr_blocks if saved blocks were released */ 4560 if (!add && !atomic_read(&fi->i_compr_blocks)) 4561 return; 4562 4563 if (add) { 4564 atomic_add(diff, &fi->i_compr_blocks); 4565 stat_add_compr_blocks(inode, diff); 4566 } else { 4567 atomic_sub(diff, &fi->i_compr_blocks); 4568 stat_sub_compr_blocks(inode, diff); 4569 } 4570 f2fs_mark_inode_dirty_sync(inode, true); 4571 } 4572 4573 static inline bool f2fs_allow_multi_device_dio(struct f2fs_sb_info *sbi, 4574 int flag) 4575 { 4576 if (!f2fs_is_multi_device(sbi)) 4577 return false; 4578 if (flag != F2FS_GET_BLOCK_DIO) 4579 return false; 4580 return sbi->aligned_blksize; 4581 } 4582 4583 static inline bool f2fs_need_verity(const struct inode *inode, pgoff_t idx) 4584 { 4585 return fsverity_active(inode) && 4586 idx < DIV_ROUND_UP(inode->i_size, PAGE_SIZE); 4587 } 4588 4589 #ifdef CONFIG_F2FS_FAULT_INJECTION 4590 extern void f2fs_build_fault_attr(struct f2fs_sb_info *sbi, unsigned int rate, 4591 unsigned int type); 4592 #else 4593 #define f2fs_build_fault_attr(sbi, rate, type) do { } while (0) 4594 #endif 4595 4596 static inline bool is_journalled_quota(struct f2fs_sb_info *sbi) 4597 { 4598 #ifdef CONFIG_QUOTA 4599 if (f2fs_sb_has_quota_ino(sbi)) 4600 return true; 4601 if (F2FS_OPTION(sbi).s_qf_names[USRQUOTA] || 4602 F2FS_OPTION(sbi).s_qf_names[GRPQUOTA] || 4603 F2FS_OPTION(sbi).s_qf_names[PRJQUOTA]) 4604 return true; 4605 #endif 4606 return false; 4607 } 4608 4609 static inline bool f2fs_block_unit_discard(struct f2fs_sb_info *sbi) 4610 { 4611 return F2FS_OPTION(sbi).discard_unit == DISCARD_UNIT_BLOCK; 4612 } 4613 4614 static inline void f2fs_io_schedule_timeout(long timeout) 4615 { 4616 set_current_state(TASK_UNINTERRUPTIBLE); 4617 io_schedule_timeout(timeout); 4618 } 4619 4620 static inline void f2fs_handle_page_eio(struct f2fs_sb_info *sbi, pgoff_t ofs, 4621 enum page_type type) 4622 { 4623 if (unlikely(f2fs_cp_error(sbi))) 4624 return; 4625 4626 if (ofs == sbi->page_eio_ofs[type]) { 4627 if (sbi->page_eio_cnt[type]++ == MAX_RETRY_PAGE_EIO) 4628 set_ckpt_flags(sbi, CP_ERROR_FLAG); 4629 } else { 4630 sbi->page_eio_ofs[type] = ofs; 4631 sbi->page_eio_cnt[type] = 0; 4632 } 4633 } 4634 4635 static inline bool f2fs_is_readonly(struct f2fs_sb_info *sbi) 4636 { 4637 return f2fs_sb_has_readonly(sbi) || f2fs_readonly(sbi->sb); 4638 } 4639 4640 static inline void f2fs_truncate_meta_inode_pages(struct f2fs_sb_info *sbi, 4641 block_t blkaddr, unsigned int cnt) 4642 { 4643 bool need_submit = false; 4644 int i = 0; 4645 4646 do { 4647 struct page *page; 4648 4649 page = find_get_page(META_MAPPING(sbi), blkaddr + i); 4650 if (page) { 4651 if (PageWriteback(page)) 4652 need_submit = true; 4653 f2fs_put_page(page, 0); 4654 } 4655 } while (++i < cnt && !need_submit); 4656 4657 if (need_submit) 4658 f2fs_submit_merged_write_cond(sbi, sbi->meta_inode, 4659 NULL, 0, DATA); 4660 4661 truncate_inode_pages_range(META_MAPPING(sbi), 4662 F2FS_BLK_TO_BYTES((loff_t)blkaddr), 4663 F2FS_BLK_END_BYTES((loff_t)(blkaddr + cnt - 1))); 4664 } 4665 4666 static inline void f2fs_invalidate_internal_cache(struct f2fs_sb_info *sbi, 4667 block_t blkaddr) 4668 { 4669 f2fs_truncate_meta_inode_pages(sbi, blkaddr, 1); 4670 f2fs_invalidate_compress_page(sbi, blkaddr); 4671 } 4672 4673 #define EFSBADCRC EBADMSG /* Bad CRC detected */ 4674 #define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */ 4675 4676 #endif /* _LINUX_F2FS_H */ 4677