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 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 union { 840 struct inode *cow_inode; /* copy-on-write inode for atomic write */ 841 struct inode *atomic_inode; 842 /* point to atomic_inode, available only for cow_inode */ 843 }; 844 845 /* avoid racing between foreground op and gc */ 846 struct f2fs_rwsem i_gc_rwsem[2]; 847 struct f2fs_rwsem i_xattr_sem; /* avoid racing between reading and changing EAs */ 848 849 int i_extra_isize; /* size of extra space located in i_addr */ 850 kprojid_t i_projid; /* id for project quota */ 851 int i_inline_xattr_size; /* inline xattr size */ 852 struct timespec64 i_crtime; /* inode creation time */ 853 struct timespec64 i_disk_time[3];/* inode disk times */ 854 855 /* for file compress */ 856 atomic_t i_compr_blocks; /* # of compressed blocks */ 857 unsigned char i_compress_algorithm; /* algorithm type */ 858 unsigned char i_log_cluster_size; /* log of cluster size */ 859 unsigned char i_compress_level; /* compress level (lz4hc,zstd) */ 860 unsigned char i_compress_flag; /* compress flag */ 861 unsigned int i_cluster_size; /* cluster size */ 862 863 unsigned int atomic_write_cnt; 864 loff_t original_i_size; /* original i_size before atomic write */ 865 }; 866 867 static inline void get_read_extent_info(struct extent_info *ext, 868 struct f2fs_extent *i_ext) 869 { 870 ext->fofs = le32_to_cpu(i_ext->fofs); 871 ext->blk = le32_to_cpu(i_ext->blk); 872 ext->len = le32_to_cpu(i_ext->len); 873 } 874 875 static inline void set_raw_read_extent(struct extent_info *ext, 876 struct f2fs_extent *i_ext) 877 { 878 i_ext->fofs = cpu_to_le32(ext->fofs); 879 i_ext->blk = cpu_to_le32(ext->blk); 880 i_ext->len = cpu_to_le32(ext->len); 881 } 882 883 static inline bool __is_discard_mergeable(struct discard_info *back, 884 struct discard_info *front, unsigned int max_len) 885 { 886 return (back->lstart + back->len == front->lstart) && 887 (back->len + front->len <= max_len); 888 } 889 890 static inline bool __is_discard_back_mergeable(struct discard_info *cur, 891 struct discard_info *back, unsigned int max_len) 892 { 893 return __is_discard_mergeable(back, cur, max_len); 894 } 895 896 static inline bool __is_discard_front_mergeable(struct discard_info *cur, 897 struct discard_info *front, unsigned int max_len) 898 { 899 return __is_discard_mergeable(cur, front, max_len); 900 } 901 902 /* 903 * For free nid management 904 */ 905 enum nid_state { 906 FREE_NID, /* newly added to free nid list */ 907 PREALLOC_NID, /* it is preallocated */ 908 MAX_NID_STATE, 909 }; 910 911 enum nat_state { 912 TOTAL_NAT, 913 DIRTY_NAT, 914 RECLAIMABLE_NAT, 915 MAX_NAT_STATE, 916 }; 917 918 struct f2fs_nm_info { 919 block_t nat_blkaddr; /* base disk address of NAT */ 920 nid_t max_nid; /* maximum possible node ids */ 921 nid_t available_nids; /* # of available node ids */ 922 nid_t next_scan_nid; /* the next nid to be scanned */ 923 nid_t max_rf_node_blocks; /* max # of nodes for recovery */ 924 unsigned int ram_thresh; /* control the memory footprint */ 925 unsigned int ra_nid_pages; /* # of nid pages to be readaheaded */ 926 unsigned int dirty_nats_ratio; /* control dirty nats ratio threshold */ 927 928 /* NAT cache management */ 929 struct radix_tree_root nat_root;/* root of the nat entry cache */ 930 struct radix_tree_root nat_set_root;/* root of the nat set cache */ 931 struct f2fs_rwsem nat_tree_lock; /* protect nat entry tree */ 932 struct list_head nat_entries; /* cached nat entry list (clean) */ 933 spinlock_t nat_list_lock; /* protect clean nat entry list */ 934 unsigned int nat_cnt[MAX_NAT_STATE]; /* the # of cached nat entries */ 935 unsigned int nat_blocks; /* # of nat blocks */ 936 937 /* free node ids management */ 938 struct radix_tree_root free_nid_root;/* root of the free_nid cache */ 939 struct list_head free_nid_list; /* list for free nids excluding preallocated nids */ 940 unsigned int nid_cnt[MAX_NID_STATE]; /* the number of free node id */ 941 spinlock_t nid_list_lock; /* protect nid lists ops */ 942 struct mutex build_lock; /* lock for build free nids */ 943 unsigned char **free_nid_bitmap; 944 unsigned char *nat_block_bitmap; 945 unsigned short *free_nid_count; /* free nid count of NAT block */ 946 947 /* for checkpoint */ 948 char *nat_bitmap; /* NAT bitmap pointer */ 949 950 unsigned int nat_bits_blocks; /* # of nat bits blocks */ 951 unsigned char *nat_bits; /* NAT bits blocks */ 952 unsigned char *full_nat_bits; /* full NAT pages */ 953 unsigned char *empty_nat_bits; /* empty NAT pages */ 954 #ifdef CONFIG_F2FS_CHECK_FS 955 char *nat_bitmap_mir; /* NAT bitmap mirror */ 956 #endif 957 int bitmap_size; /* bitmap size */ 958 }; 959 960 /* 961 * this structure is used as one of function parameters. 962 * all the information are dedicated to a given direct node block determined 963 * by the data offset in a file. 964 */ 965 struct dnode_of_data { 966 struct inode *inode; /* vfs inode pointer */ 967 struct page *inode_page; /* its inode page, NULL is possible */ 968 struct page *node_page; /* cached direct node page */ 969 nid_t nid; /* node id of the direct node block */ 970 unsigned int ofs_in_node; /* data offset in the node page */ 971 bool inode_page_locked; /* inode page is locked or not */ 972 bool node_changed; /* is node block changed */ 973 char cur_level; /* level of hole node page */ 974 char max_level; /* level of current page located */ 975 block_t data_blkaddr; /* block address of the node block */ 976 }; 977 978 static inline void set_new_dnode(struct dnode_of_data *dn, struct inode *inode, 979 struct page *ipage, struct page *npage, nid_t nid) 980 { 981 memset(dn, 0, sizeof(*dn)); 982 dn->inode = inode; 983 dn->inode_page = ipage; 984 dn->node_page = npage; 985 dn->nid = nid; 986 } 987 988 /* 989 * For SIT manager 990 * 991 * By default, there are 6 active log areas across the whole main area. 992 * When considering hot and cold data separation to reduce cleaning overhead, 993 * we split 3 for data logs and 3 for node logs as hot, warm, and cold types, 994 * respectively. 995 * In the current design, you should not change the numbers intentionally. 996 * Instead, as a mount option such as active_logs=x, you can use 2, 4, and 6 997 * logs individually according to the underlying devices. (default: 6) 998 * Just in case, on-disk layout covers maximum 16 logs that consist of 8 for 999 * data and 8 for node logs. 1000 */ 1001 #define NR_CURSEG_DATA_TYPE (3) 1002 #define NR_CURSEG_NODE_TYPE (3) 1003 #define NR_CURSEG_INMEM_TYPE (2) 1004 #define NR_CURSEG_RO_TYPE (2) 1005 #define NR_CURSEG_PERSIST_TYPE (NR_CURSEG_DATA_TYPE + NR_CURSEG_NODE_TYPE) 1006 #define NR_CURSEG_TYPE (NR_CURSEG_INMEM_TYPE + NR_CURSEG_PERSIST_TYPE) 1007 1008 enum { 1009 CURSEG_HOT_DATA = 0, /* directory entry blocks */ 1010 CURSEG_WARM_DATA, /* data blocks */ 1011 CURSEG_COLD_DATA, /* multimedia or GCed data blocks */ 1012 CURSEG_HOT_NODE, /* direct node blocks of directory files */ 1013 CURSEG_WARM_NODE, /* direct node blocks of normal files */ 1014 CURSEG_COLD_NODE, /* indirect node blocks */ 1015 NR_PERSISTENT_LOG, /* number of persistent log */ 1016 CURSEG_COLD_DATA_PINNED = NR_PERSISTENT_LOG, 1017 /* pinned file that needs consecutive block address */ 1018 CURSEG_ALL_DATA_ATGC, /* SSR alloctor in hot/warm/cold data area */ 1019 NO_CHECK_TYPE, /* number of persistent & inmem log */ 1020 }; 1021 1022 struct flush_cmd { 1023 struct completion wait; 1024 struct llist_node llnode; 1025 nid_t ino; 1026 int ret; 1027 }; 1028 1029 struct flush_cmd_control { 1030 struct task_struct *f2fs_issue_flush; /* flush thread */ 1031 wait_queue_head_t flush_wait_queue; /* waiting queue for wake-up */ 1032 atomic_t issued_flush; /* # of issued flushes */ 1033 atomic_t queued_flush; /* # of queued flushes */ 1034 struct llist_head issue_list; /* list for command issue */ 1035 struct llist_node *dispatch_list; /* list for command dispatch */ 1036 }; 1037 1038 struct f2fs_sm_info { 1039 struct sit_info *sit_info; /* whole segment information */ 1040 struct free_segmap_info *free_info; /* free segment information */ 1041 struct dirty_seglist_info *dirty_info; /* dirty segment information */ 1042 struct curseg_info *curseg_array; /* active segment information */ 1043 1044 struct f2fs_rwsem curseg_lock; /* for preventing curseg change */ 1045 1046 block_t seg0_blkaddr; /* block address of 0'th segment */ 1047 block_t main_blkaddr; /* start block address of main area */ 1048 block_t ssa_blkaddr; /* start block address of SSA area */ 1049 1050 unsigned int segment_count; /* total # of segments */ 1051 unsigned int main_segments; /* # of segments in main area */ 1052 unsigned int reserved_segments; /* # of reserved segments */ 1053 unsigned int additional_reserved_segments;/* reserved segs for IO align feature */ 1054 unsigned int ovp_segments; /* # of overprovision segments */ 1055 1056 /* a threshold to reclaim prefree segments */ 1057 unsigned int rec_prefree_segments; 1058 1059 struct list_head sit_entry_set; /* sit entry set list */ 1060 1061 unsigned int ipu_policy; /* in-place-update policy */ 1062 unsigned int min_ipu_util; /* in-place-update threshold */ 1063 unsigned int min_fsync_blocks; /* threshold for fsync */ 1064 unsigned int min_seq_blocks; /* threshold for sequential blocks */ 1065 unsigned int min_hot_blocks; /* threshold for hot block allocation */ 1066 unsigned int min_ssr_sections; /* threshold to trigger SSR allocation */ 1067 1068 /* for flush command control */ 1069 struct flush_cmd_control *fcc_info; 1070 1071 /* for discard command control */ 1072 struct discard_cmd_control *dcc_info; 1073 }; 1074 1075 /* 1076 * For superblock 1077 */ 1078 /* 1079 * COUNT_TYPE for monitoring 1080 * 1081 * f2fs monitors the number of several block types such as on-writeback, 1082 * dirty dentry blocks, dirty node blocks, and dirty meta blocks. 1083 */ 1084 #define WB_DATA_TYPE(p, f) \ 1085 (f || f2fs_is_cp_guaranteed(p) ? F2FS_WB_CP_DATA : F2FS_WB_DATA) 1086 enum count_type { 1087 F2FS_DIRTY_DENTS, 1088 F2FS_DIRTY_DATA, 1089 F2FS_DIRTY_QDATA, 1090 F2FS_DIRTY_NODES, 1091 F2FS_DIRTY_META, 1092 F2FS_DIRTY_IMETA, 1093 F2FS_WB_CP_DATA, 1094 F2FS_WB_DATA, 1095 F2FS_RD_DATA, 1096 F2FS_RD_NODE, 1097 F2FS_RD_META, 1098 F2FS_DIO_WRITE, 1099 F2FS_DIO_READ, 1100 NR_COUNT_TYPE, 1101 }; 1102 1103 /* 1104 * The below are the page types of bios used in submit_bio(). 1105 * The available types are: 1106 * DATA User data pages. It operates as async mode. 1107 * NODE Node pages. It operates as async mode. 1108 * META FS metadata pages such as SIT, NAT, CP. 1109 * NR_PAGE_TYPE The number of page types. 1110 * META_FLUSH Make sure the previous pages are written 1111 * with waiting the bio's completion 1112 * ... Only can be used with META. 1113 */ 1114 #define PAGE_TYPE_OF_BIO(type) ((type) > META ? META : (type)) 1115 #define PAGE_TYPE_ON_MAIN(type) ((type) == DATA || (type) == NODE) 1116 enum page_type { 1117 DATA = 0, 1118 NODE = 1, /* should not change this */ 1119 META, 1120 NR_PAGE_TYPE, 1121 META_FLUSH, 1122 IPU, /* the below types are used by tracepoints only. */ 1123 OPU, 1124 }; 1125 1126 enum temp_type { 1127 HOT = 0, /* must be zero for meta bio */ 1128 WARM, 1129 COLD, 1130 NR_TEMP_TYPE, 1131 }; 1132 1133 enum need_lock_type { 1134 LOCK_REQ = 0, 1135 LOCK_DONE, 1136 LOCK_RETRY, 1137 }; 1138 1139 enum cp_reason_type { 1140 CP_NO_NEEDED, 1141 CP_NON_REGULAR, 1142 CP_COMPRESSED, 1143 CP_HARDLINK, 1144 CP_SB_NEED_CP, 1145 CP_WRONG_PINO, 1146 CP_NO_SPC_ROLL, 1147 CP_NODE_NEED_CP, 1148 CP_FASTBOOT_MODE, 1149 CP_SPEC_LOG_NUM, 1150 CP_RECOVER_DIR, 1151 }; 1152 1153 enum iostat_type { 1154 /* WRITE IO */ 1155 APP_DIRECT_IO, /* app direct write IOs */ 1156 APP_BUFFERED_IO, /* app buffered write IOs */ 1157 APP_WRITE_IO, /* app write IOs */ 1158 APP_MAPPED_IO, /* app mapped IOs */ 1159 APP_BUFFERED_CDATA_IO, /* app buffered write IOs on compressed file */ 1160 APP_MAPPED_CDATA_IO, /* app mapped write IOs on compressed file */ 1161 FS_DATA_IO, /* data IOs from kworker/fsync/reclaimer */ 1162 FS_CDATA_IO, /* data IOs from kworker/fsync/reclaimer on compressed file */ 1163 FS_NODE_IO, /* node IOs from kworker/fsync/reclaimer */ 1164 FS_META_IO, /* meta IOs from kworker/reclaimer */ 1165 FS_GC_DATA_IO, /* data IOs from forground gc */ 1166 FS_GC_NODE_IO, /* node IOs from forground gc */ 1167 FS_CP_DATA_IO, /* data IOs from checkpoint */ 1168 FS_CP_NODE_IO, /* node IOs from checkpoint */ 1169 FS_CP_META_IO, /* meta IOs from checkpoint */ 1170 1171 /* READ IO */ 1172 APP_DIRECT_READ_IO, /* app direct read IOs */ 1173 APP_BUFFERED_READ_IO, /* app buffered read IOs */ 1174 APP_READ_IO, /* app read IOs */ 1175 APP_MAPPED_READ_IO, /* app mapped read IOs */ 1176 APP_BUFFERED_CDATA_READ_IO, /* app buffered read IOs on compressed file */ 1177 APP_MAPPED_CDATA_READ_IO, /* app mapped read IOs on compressed file */ 1178 FS_DATA_READ_IO, /* data read IOs */ 1179 FS_GDATA_READ_IO, /* data read IOs from background gc */ 1180 FS_CDATA_READ_IO, /* compressed data read IOs */ 1181 FS_NODE_READ_IO, /* node read IOs */ 1182 FS_META_READ_IO, /* meta read IOs */ 1183 1184 /* other */ 1185 FS_DISCARD_IO, /* discard */ 1186 FS_FLUSH_IO, /* flush */ 1187 FS_ZONE_RESET_IO, /* zone reset */ 1188 NR_IO_TYPE, 1189 }; 1190 1191 struct f2fs_io_info { 1192 struct f2fs_sb_info *sbi; /* f2fs_sb_info pointer */ 1193 nid_t ino; /* inode number */ 1194 enum page_type type; /* contains DATA/NODE/META/META_FLUSH */ 1195 enum temp_type temp; /* contains HOT/WARM/COLD */ 1196 enum req_op op; /* contains REQ_OP_ */ 1197 blk_opf_t op_flags; /* req_flag_bits */ 1198 block_t new_blkaddr; /* new block address to be written */ 1199 block_t old_blkaddr; /* old block address before Cow */ 1200 struct page *page; /* page to be written */ 1201 struct page *encrypted_page; /* encrypted page */ 1202 struct page *compressed_page; /* compressed page */ 1203 struct list_head list; /* serialize IOs */ 1204 unsigned int compr_blocks; /* # of compressed block addresses */ 1205 unsigned int need_lock:8; /* indicate we need to lock cp_rwsem */ 1206 unsigned int version:8; /* version of the node */ 1207 unsigned int submitted:1; /* indicate IO submission */ 1208 unsigned int in_list:1; /* indicate fio is in io_list */ 1209 unsigned int is_por:1; /* indicate IO is from recovery or not */ 1210 unsigned int encrypted:1; /* indicate file is encrypted */ 1211 unsigned int meta_gc:1; /* require meta inode GC */ 1212 enum iostat_type io_type; /* io type */ 1213 struct writeback_control *io_wbc; /* writeback control */ 1214 struct bio **bio; /* bio for ipu */ 1215 sector_t *last_block; /* last block number in bio */ 1216 }; 1217 1218 struct bio_entry { 1219 struct bio *bio; 1220 struct list_head list; 1221 }; 1222 1223 #define is_read_io(rw) ((rw) == READ) 1224 struct f2fs_bio_info { 1225 struct f2fs_sb_info *sbi; /* f2fs superblock */ 1226 struct bio *bio; /* bios to merge */ 1227 sector_t last_block_in_bio; /* last block number */ 1228 struct f2fs_io_info fio; /* store buffered io info. */ 1229 #ifdef CONFIG_BLK_DEV_ZONED 1230 struct completion zone_wait; /* condition value for the previous open zone to close */ 1231 struct bio *zone_pending_bio; /* pending bio for the previous zone */ 1232 void *bi_private; /* previous bi_private for pending bio */ 1233 #endif 1234 struct f2fs_rwsem io_rwsem; /* blocking op for bio */ 1235 spinlock_t io_lock; /* serialize DATA/NODE IOs */ 1236 struct list_head io_list; /* track fios */ 1237 struct list_head bio_list; /* bio entry list head */ 1238 struct f2fs_rwsem bio_list_lock; /* lock to protect bio entry list */ 1239 }; 1240 1241 #define FDEV(i) (sbi->devs[i]) 1242 #define RDEV(i) (raw_super->devs[i]) 1243 struct f2fs_dev_info { 1244 struct block_device *bdev; 1245 char path[MAX_PATH_LEN]; 1246 unsigned int total_segments; 1247 block_t start_blk; 1248 block_t end_blk; 1249 #ifdef CONFIG_BLK_DEV_ZONED 1250 unsigned int nr_blkz; /* Total number of zones */ 1251 unsigned long *blkz_seq; /* Bitmap indicating sequential zones */ 1252 #endif 1253 }; 1254 1255 enum inode_type { 1256 DIR_INODE, /* for dirty dir inode */ 1257 FILE_INODE, /* for dirty regular/symlink inode */ 1258 DIRTY_META, /* for all dirtied inode metadata */ 1259 NR_INODE_TYPE, 1260 }; 1261 1262 /* for inner inode cache management */ 1263 struct inode_management { 1264 struct radix_tree_root ino_root; /* ino entry array */ 1265 spinlock_t ino_lock; /* for ino entry lock */ 1266 struct list_head ino_list; /* inode list head */ 1267 unsigned long ino_num; /* number of entries */ 1268 }; 1269 1270 /* for GC_AT */ 1271 struct atgc_management { 1272 bool atgc_enabled; /* ATGC is enabled or not */ 1273 struct rb_root_cached root; /* root of victim rb-tree */ 1274 struct list_head victim_list; /* linked with all victim entries */ 1275 unsigned int victim_count; /* victim count in rb-tree */ 1276 unsigned int candidate_ratio; /* candidate ratio */ 1277 unsigned int max_candidate_count; /* max candidate count */ 1278 unsigned int age_weight; /* age weight, vblock_weight = 100 - age_weight */ 1279 unsigned long long age_threshold; /* age threshold */ 1280 }; 1281 1282 struct f2fs_gc_control { 1283 unsigned int victim_segno; /* target victim segment number */ 1284 int init_gc_type; /* FG_GC or BG_GC */ 1285 bool no_bg_gc; /* check the space and stop bg_gc */ 1286 bool should_migrate_blocks; /* should migrate blocks */ 1287 bool err_gc_skipped; /* return EAGAIN if GC skipped */ 1288 unsigned int nr_free_secs; /* # of free sections to do GC */ 1289 }; 1290 1291 /* 1292 * For s_flag in struct f2fs_sb_info 1293 * Modification on enum should be synchronized with s_flag array 1294 */ 1295 enum { 1296 SBI_IS_DIRTY, /* dirty flag for checkpoint */ 1297 SBI_IS_CLOSE, /* specify unmounting */ 1298 SBI_NEED_FSCK, /* need fsck.f2fs to fix */ 1299 SBI_POR_DOING, /* recovery is doing or not */ 1300 SBI_NEED_SB_WRITE, /* need to recover superblock */ 1301 SBI_NEED_CP, /* need to checkpoint */ 1302 SBI_IS_SHUTDOWN, /* shutdown by ioctl */ 1303 SBI_IS_RECOVERED, /* recovered orphan/data */ 1304 SBI_CP_DISABLED, /* CP was disabled last mount */ 1305 SBI_CP_DISABLED_QUICK, /* CP was disabled quickly */ 1306 SBI_QUOTA_NEED_FLUSH, /* need to flush quota info in CP */ 1307 SBI_QUOTA_SKIP_FLUSH, /* skip flushing quota in current CP */ 1308 SBI_QUOTA_NEED_REPAIR, /* quota file may be corrupted */ 1309 SBI_IS_RESIZEFS, /* resizefs is in process */ 1310 SBI_IS_FREEZING, /* freezefs is in process */ 1311 SBI_IS_WRITABLE, /* remove ro mountoption transiently */ 1312 MAX_SBI_FLAG, 1313 }; 1314 1315 enum { 1316 CP_TIME, 1317 REQ_TIME, 1318 DISCARD_TIME, 1319 GC_TIME, 1320 DISABLE_TIME, 1321 UMOUNT_DISCARD_TIMEOUT, 1322 MAX_TIME, 1323 }; 1324 1325 /* Note that you need to keep synchronization with this gc_mode_names array */ 1326 enum { 1327 GC_NORMAL, 1328 GC_IDLE_CB, 1329 GC_IDLE_GREEDY, 1330 GC_IDLE_AT, 1331 GC_URGENT_HIGH, 1332 GC_URGENT_LOW, 1333 GC_URGENT_MID, 1334 MAX_GC_MODE, 1335 }; 1336 1337 enum { 1338 BGGC_MODE_ON, /* background gc is on */ 1339 BGGC_MODE_OFF, /* background gc is off */ 1340 BGGC_MODE_SYNC, /* 1341 * background gc is on, migrating blocks 1342 * like foreground gc 1343 */ 1344 }; 1345 1346 enum { 1347 FS_MODE_ADAPTIVE, /* use both lfs/ssr allocation */ 1348 FS_MODE_LFS, /* use lfs allocation only */ 1349 FS_MODE_FRAGMENT_SEG, /* segment fragmentation mode */ 1350 FS_MODE_FRAGMENT_BLK, /* block fragmentation mode */ 1351 }; 1352 1353 enum { 1354 ALLOC_MODE_DEFAULT, /* stay default */ 1355 ALLOC_MODE_REUSE, /* reuse segments as much as possible */ 1356 }; 1357 1358 enum fsync_mode { 1359 FSYNC_MODE_POSIX, /* fsync follows posix semantics */ 1360 FSYNC_MODE_STRICT, /* fsync behaves in line with ext4 */ 1361 FSYNC_MODE_NOBARRIER, /* fsync behaves nobarrier based on posix */ 1362 }; 1363 1364 enum { 1365 COMPR_MODE_FS, /* 1366 * automatically compress compression 1367 * enabled files 1368 */ 1369 COMPR_MODE_USER, /* 1370 * automatical compression is disabled. 1371 * user can control the file compression 1372 * using ioctls 1373 */ 1374 }; 1375 1376 enum { 1377 DISCARD_UNIT_BLOCK, /* basic discard unit is block */ 1378 DISCARD_UNIT_SEGMENT, /* basic discard unit is segment */ 1379 DISCARD_UNIT_SECTION, /* basic discard unit is section */ 1380 }; 1381 1382 enum { 1383 MEMORY_MODE_NORMAL, /* memory mode for normal devices */ 1384 MEMORY_MODE_LOW, /* memory mode for low memry devices */ 1385 }; 1386 1387 enum errors_option { 1388 MOUNT_ERRORS_READONLY, /* remount fs ro on errors */ 1389 MOUNT_ERRORS_CONTINUE, /* continue on errors */ 1390 MOUNT_ERRORS_PANIC, /* panic on errors */ 1391 }; 1392 1393 enum { 1394 BACKGROUND, 1395 FOREGROUND, 1396 MAX_CALL_TYPE, 1397 TOTAL_CALL = FOREGROUND, 1398 }; 1399 1400 static inline int f2fs_test_bit(unsigned int nr, char *addr); 1401 static inline void f2fs_set_bit(unsigned int nr, char *addr); 1402 static inline void f2fs_clear_bit(unsigned int nr, char *addr); 1403 1404 /* 1405 * Layout of f2fs page.private: 1406 * 1407 * Layout A: lowest bit should be 1 1408 * | bit0 = 1 | bit1 | bit2 | ... | bit MAX | private data .... | 1409 * bit 0 PAGE_PRIVATE_NOT_POINTER 1410 * bit 1 PAGE_PRIVATE_ONGOING_MIGRATION 1411 * bit 2 PAGE_PRIVATE_INLINE_INODE 1412 * bit 3 PAGE_PRIVATE_REF_RESOURCE 1413 * bit 4- f2fs private data 1414 * 1415 * Layout B: lowest bit should be 0 1416 * page.private is a wrapped pointer. 1417 */ 1418 enum { 1419 PAGE_PRIVATE_NOT_POINTER, /* private contains non-pointer data */ 1420 PAGE_PRIVATE_ONGOING_MIGRATION, /* data page which is on-going migrating */ 1421 PAGE_PRIVATE_INLINE_INODE, /* inode page contains inline data */ 1422 PAGE_PRIVATE_REF_RESOURCE, /* dirty page has referenced resources */ 1423 PAGE_PRIVATE_MAX 1424 }; 1425 1426 /* For compression */ 1427 enum compress_algorithm_type { 1428 COMPRESS_LZO, 1429 COMPRESS_LZ4, 1430 COMPRESS_ZSTD, 1431 COMPRESS_LZORLE, 1432 COMPRESS_MAX, 1433 }; 1434 1435 enum compress_flag { 1436 COMPRESS_CHKSUM, 1437 COMPRESS_MAX_FLAG, 1438 }; 1439 1440 #define COMPRESS_WATERMARK 20 1441 #define COMPRESS_PERCENT 20 1442 1443 #define COMPRESS_DATA_RESERVED_SIZE 4 1444 struct compress_data { 1445 __le32 clen; /* compressed data size */ 1446 __le32 chksum; /* compressed data chksum */ 1447 __le32 reserved[COMPRESS_DATA_RESERVED_SIZE]; /* reserved */ 1448 u8 cdata[]; /* compressed data */ 1449 }; 1450 1451 #define COMPRESS_HEADER_SIZE (sizeof(struct compress_data)) 1452 1453 #define F2FS_COMPRESSED_PAGE_MAGIC 0xF5F2C000 1454 1455 #define F2FS_ZSTD_DEFAULT_CLEVEL 1 1456 1457 #define COMPRESS_LEVEL_OFFSET 8 1458 1459 /* compress context */ 1460 struct compress_ctx { 1461 struct inode *inode; /* inode the context belong to */ 1462 pgoff_t cluster_idx; /* cluster index number */ 1463 unsigned int cluster_size; /* page count in cluster */ 1464 unsigned int log_cluster_size; /* log of cluster size */ 1465 struct page **rpages; /* pages store raw data in cluster */ 1466 unsigned int nr_rpages; /* total page number in rpages */ 1467 struct page **cpages; /* pages store compressed data in cluster */ 1468 unsigned int nr_cpages; /* total page number in cpages */ 1469 unsigned int valid_nr_cpages; /* valid page number in cpages */ 1470 void *rbuf; /* virtual mapped address on rpages */ 1471 struct compress_data *cbuf; /* virtual mapped address on cpages */ 1472 size_t rlen; /* valid data length in rbuf */ 1473 size_t clen; /* valid data length in cbuf */ 1474 void *private; /* payload buffer for specified compression algorithm */ 1475 void *private2; /* extra payload buffer */ 1476 }; 1477 1478 /* compress context for write IO path */ 1479 struct compress_io_ctx { 1480 u32 magic; /* magic number to indicate page is compressed */ 1481 struct inode *inode; /* inode the context belong to */ 1482 struct page **rpages; /* pages store raw data in cluster */ 1483 unsigned int nr_rpages; /* total page number in rpages */ 1484 atomic_t pending_pages; /* in-flight compressed page count */ 1485 }; 1486 1487 /* Context for decompressing one cluster on the read IO path */ 1488 struct decompress_io_ctx { 1489 u32 magic; /* magic number to indicate page is compressed */ 1490 struct inode *inode; /* inode the context belong to */ 1491 pgoff_t cluster_idx; /* cluster index number */ 1492 unsigned int cluster_size; /* page count in cluster */ 1493 unsigned int log_cluster_size; /* log of cluster size */ 1494 struct page **rpages; /* pages store raw data in cluster */ 1495 unsigned int nr_rpages; /* total page number in rpages */ 1496 struct page **cpages; /* pages store compressed data in cluster */ 1497 unsigned int nr_cpages; /* total page number in cpages */ 1498 struct page **tpages; /* temp pages to pad holes in cluster */ 1499 void *rbuf; /* virtual mapped address on rpages */ 1500 struct compress_data *cbuf; /* virtual mapped address on cpages */ 1501 size_t rlen; /* valid data length in rbuf */ 1502 size_t clen; /* valid data length in cbuf */ 1503 1504 /* 1505 * The number of compressed pages remaining to be read in this cluster. 1506 * This is initially nr_cpages. It is decremented by 1 each time a page 1507 * has been read (or failed to be read). When it reaches 0, the cluster 1508 * is decompressed (or an error is reported). 1509 * 1510 * If an error occurs before all the pages have been submitted for I/O, 1511 * then this will never reach 0. In this case the I/O submitter is 1512 * responsible for calling f2fs_decompress_end_io() instead. 1513 */ 1514 atomic_t remaining_pages; 1515 1516 /* 1517 * Number of references to this decompress_io_ctx. 1518 * 1519 * One reference is held for I/O completion. This reference is dropped 1520 * after the pagecache pages are updated and unlocked -- either after 1521 * decompression (and verity if enabled), or after an error. 1522 * 1523 * In addition, each compressed page holds a reference while it is in a 1524 * bio. These references are necessary prevent compressed pages from 1525 * being freed while they are still in a bio. 1526 */ 1527 refcount_t refcnt; 1528 1529 bool failed; /* IO error occurred before decompression? */ 1530 bool need_verity; /* need fs-verity verification after decompression? */ 1531 void *private; /* payload buffer for specified decompression algorithm */ 1532 void *private2; /* extra payload buffer */ 1533 struct work_struct verity_work; /* work to verify the decompressed pages */ 1534 struct work_struct free_work; /* work for late free this structure itself */ 1535 }; 1536 1537 #define NULL_CLUSTER ((unsigned int)(~0)) 1538 #define MIN_COMPRESS_LOG_SIZE 2 1539 #define MAX_COMPRESS_LOG_SIZE 8 1540 #define MAX_COMPRESS_WINDOW_SIZE(log_size) ((PAGE_SIZE) << (log_size)) 1541 1542 struct f2fs_sb_info { 1543 struct super_block *sb; /* pointer to VFS super block */ 1544 struct proc_dir_entry *s_proc; /* proc entry */ 1545 struct f2fs_super_block *raw_super; /* raw super block pointer */ 1546 struct f2fs_rwsem sb_lock; /* lock for raw super block */ 1547 int valid_super_block; /* valid super block no */ 1548 unsigned long s_flag; /* flags for sbi */ 1549 struct mutex writepages; /* mutex for writepages() */ 1550 1551 #ifdef CONFIG_BLK_DEV_ZONED 1552 unsigned int blocks_per_blkz; /* F2FS blocks per zone */ 1553 #endif 1554 1555 /* for node-related operations */ 1556 struct f2fs_nm_info *nm_info; /* node manager */ 1557 struct inode *node_inode; /* cache node blocks */ 1558 1559 /* for segment-related operations */ 1560 struct f2fs_sm_info *sm_info; /* segment manager */ 1561 1562 /* for bio operations */ 1563 struct f2fs_bio_info *write_io[NR_PAGE_TYPE]; /* for write bios */ 1564 /* keep migration IO order for LFS mode */ 1565 struct f2fs_rwsem io_order_lock; 1566 pgoff_t page_eio_ofs[NR_PAGE_TYPE]; /* EIO page offset */ 1567 int page_eio_cnt[NR_PAGE_TYPE]; /* EIO count */ 1568 1569 /* for checkpoint */ 1570 struct f2fs_checkpoint *ckpt; /* raw checkpoint pointer */ 1571 int cur_cp_pack; /* remain current cp pack */ 1572 spinlock_t cp_lock; /* for flag in ckpt */ 1573 struct inode *meta_inode; /* cache meta blocks */ 1574 struct f2fs_rwsem cp_global_sem; /* checkpoint procedure lock */ 1575 struct f2fs_rwsem cp_rwsem; /* blocking FS operations */ 1576 struct f2fs_rwsem node_write; /* locking node writes */ 1577 struct f2fs_rwsem node_change; /* locking node change */ 1578 wait_queue_head_t cp_wait; 1579 unsigned long last_time[MAX_TIME]; /* to store time in jiffies */ 1580 long interval_time[MAX_TIME]; /* to store thresholds */ 1581 struct ckpt_req_control cprc_info; /* for checkpoint request control */ 1582 1583 struct inode_management im[MAX_INO_ENTRY]; /* manage inode cache */ 1584 1585 spinlock_t fsync_node_lock; /* for node entry lock */ 1586 struct list_head fsync_node_list; /* node list head */ 1587 unsigned int fsync_seg_id; /* sequence id */ 1588 unsigned int fsync_node_num; /* number of node entries */ 1589 1590 /* for orphan inode, use 0'th array */ 1591 unsigned int max_orphans; /* max orphan inodes */ 1592 1593 /* for inode management */ 1594 struct list_head inode_list[NR_INODE_TYPE]; /* dirty inode list */ 1595 spinlock_t inode_lock[NR_INODE_TYPE]; /* for dirty inode list lock */ 1596 struct mutex flush_lock; /* for flush exclusion */ 1597 1598 /* for extent tree cache */ 1599 struct extent_tree_info extent_tree[NR_EXTENT_CACHES]; 1600 atomic64_t allocated_data_blocks; /* for block age extent_cache */ 1601 1602 /* The threshold used for hot and warm data seperation*/ 1603 unsigned int hot_data_age_threshold; 1604 unsigned int warm_data_age_threshold; 1605 unsigned int last_age_weight; 1606 1607 /* basic filesystem units */ 1608 unsigned int log_sectors_per_block; /* log2 sectors per block */ 1609 unsigned int log_blocksize; /* log2 block size */ 1610 unsigned int blocksize; /* block size */ 1611 unsigned int root_ino_num; /* root inode number*/ 1612 unsigned int node_ino_num; /* node inode number*/ 1613 unsigned int meta_ino_num; /* meta inode number*/ 1614 unsigned int log_blocks_per_seg; /* log2 blocks per segment */ 1615 unsigned int blocks_per_seg; /* blocks per segment */ 1616 unsigned int unusable_blocks_per_sec; /* unusable blocks per section */ 1617 unsigned int segs_per_sec; /* segments per section */ 1618 unsigned int secs_per_zone; /* sections per zone */ 1619 unsigned int total_sections; /* total section count */ 1620 unsigned int total_node_count; /* total node block count */ 1621 unsigned int total_valid_node_count; /* valid node block count */ 1622 int dir_level; /* directory level */ 1623 bool readdir_ra; /* readahead inode in readdir */ 1624 u64 max_io_bytes; /* max io bytes to merge IOs */ 1625 1626 block_t user_block_count; /* # of user blocks */ 1627 block_t total_valid_block_count; /* # of valid blocks */ 1628 block_t discard_blks; /* discard command candidats */ 1629 block_t last_valid_block_count; /* for recovery */ 1630 block_t reserved_blocks; /* configurable reserved blocks */ 1631 block_t current_reserved_blocks; /* current reserved blocks */ 1632 1633 /* Additional tracking for no checkpoint mode */ 1634 block_t unusable_block_count; /* # of blocks saved by last cp */ 1635 1636 unsigned int nquota_files; /* # of quota sysfile */ 1637 struct f2fs_rwsem quota_sem; /* blocking cp for flags */ 1638 1639 /* # of pages, see count_type */ 1640 atomic_t nr_pages[NR_COUNT_TYPE]; 1641 /* # of allocated blocks */ 1642 struct percpu_counter alloc_valid_block_count; 1643 /* # of node block writes as roll forward recovery */ 1644 struct percpu_counter rf_node_block_count; 1645 1646 /* writeback control */ 1647 atomic_t wb_sync_req[META]; /* count # of WB_SYNC threads */ 1648 1649 /* valid inode count */ 1650 struct percpu_counter total_valid_inode_count; 1651 1652 struct f2fs_mount_info mount_opt; /* mount options */ 1653 1654 /* for cleaning operations */ 1655 struct f2fs_rwsem gc_lock; /* 1656 * semaphore for GC, avoid 1657 * race between GC and GC or CP 1658 */ 1659 struct f2fs_gc_kthread *gc_thread; /* GC thread */ 1660 struct atgc_management am; /* atgc management */ 1661 unsigned int cur_victim_sec; /* current victim section num */ 1662 unsigned int gc_mode; /* current GC state */ 1663 unsigned int next_victim_seg[2]; /* next segment in victim section */ 1664 spinlock_t gc_remaining_trials_lock; 1665 /* remaining trial count for GC_URGENT_* and GC_IDLE_* */ 1666 unsigned int gc_remaining_trials; 1667 1668 /* for skip statistic */ 1669 unsigned long long skipped_gc_rwsem; /* FG_GC only */ 1670 1671 /* threshold for gc trials on pinned files */ 1672 u64 gc_pin_file_threshold; 1673 struct f2fs_rwsem pin_sem; 1674 1675 /* maximum # of trials to find a victim segment for SSR and GC */ 1676 unsigned int max_victim_search; 1677 /* migration granularity of garbage collection, unit: segment */ 1678 unsigned int migration_granularity; 1679 1680 /* 1681 * for stat information. 1682 * one is for the LFS mode, and the other is for the SSR mode. 1683 */ 1684 #ifdef CONFIG_F2FS_STAT_FS 1685 struct f2fs_stat_info *stat_info; /* FS status information */ 1686 atomic_t meta_count[META_MAX]; /* # of meta blocks */ 1687 unsigned int segment_count[2]; /* # of allocated segments */ 1688 unsigned int block_count[2]; /* # of allocated blocks */ 1689 atomic_t inplace_count; /* # of inplace update */ 1690 /* # of lookup extent cache */ 1691 atomic64_t total_hit_ext[NR_EXTENT_CACHES]; 1692 /* # of hit rbtree extent node */ 1693 atomic64_t read_hit_rbtree[NR_EXTENT_CACHES]; 1694 /* # of hit cached extent node */ 1695 atomic64_t read_hit_cached[NR_EXTENT_CACHES]; 1696 /* # of hit largest extent node in read extent cache */ 1697 atomic64_t read_hit_largest; 1698 atomic_t inline_xattr; /* # of inline_xattr inodes */ 1699 atomic_t inline_inode; /* # of inline_data inodes */ 1700 atomic_t inline_dir; /* # of inline_dentry inodes */ 1701 atomic_t compr_inode; /* # of compressed inodes */ 1702 atomic64_t compr_blocks; /* # of compressed blocks */ 1703 atomic_t swapfile_inode; /* # of swapfile inodes */ 1704 atomic_t atomic_files; /* # of opened atomic file */ 1705 atomic_t max_aw_cnt; /* max # of atomic writes */ 1706 unsigned int io_skip_bggc; /* skip background gc for in-flight IO */ 1707 unsigned int other_skip_bggc; /* skip background gc for other reasons */ 1708 unsigned int ndirty_inode[NR_INODE_TYPE]; /* # of dirty inodes */ 1709 atomic_t cp_call_count[MAX_CALL_TYPE]; /* # of cp call */ 1710 #endif 1711 spinlock_t stat_lock; /* lock for stat operations */ 1712 1713 /* to attach REQ_META|REQ_FUA flags */ 1714 unsigned int data_io_flag; 1715 unsigned int node_io_flag; 1716 1717 /* For sysfs support */ 1718 struct kobject s_kobj; /* /sys/fs/f2fs/<devname> */ 1719 struct completion s_kobj_unregister; 1720 1721 struct kobject s_stat_kobj; /* /sys/fs/f2fs/<devname>/stat */ 1722 struct completion s_stat_kobj_unregister; 1723 1724 struct kobject s_feature_list_kobj; /* /sys/fs/f2fs/<devname>/feature_list */ 1725 struct completion s_feature_list_kobj_unregister; 1726 1727 /* For shrinker support */ 1728 struct list_head s_list; 1729 struct mutex umount_mutex; 1730 unsigned int shrinker_run_no; 1731 1732 /* For multi devices */ 1733 int s_ndevs; /* number of devices */ 1734 struct f2fs_dev_info *devs; /* for device list */ 1735 unsigned int dirty_device; /* for checkpoint data flush */ 1736 spinlock_t dev_lock; /* protect dirty_device */ 1737 bool aligned_blksize; /* all devices has the same logical blksize */ 1738 1739 /* For write statistics */ 1740 u64 sectors_written_start; 1741 u64 kbytes_written; 1742 1743 /* Reference to checksum algorithm driver via cryptoapi */ 1744 struct crypto_shash *s_chksum_driver; 1745 1746 /* Precomputed FS UUID checksum for seeding other checksums */ 1747 __u32 s_chksum_seed; 1748 1749 struct workqueue_struct *post_read_wq; /* post read workqueue */ 1750 1751 /* 1752 * If we are in irq context, let's update error information into 1753 * on-disk superblock in the work. 1754 */ 1755 struct work_struct s_error_work; 1756 unsigned char errors[MAX_F2FS_ERRORS]; /* error flags */ 1757 unsigned char stop_reason[MAX_STOP_REASON]; /* stop reason */ 1758 spinlock_t error_lock; /* protect errors/stop_reason array */ 1759 bool error_dirty; /* errors of sb is dirty */ 1760 1761 struct kmem_cache *inline_xattr_slab; /* inline xattr entry */ 1762 unsigned int inline_xattr_slab_size; /* default inline xattr slab size */ 1763 1764 /* For reclaimed segs statistics per each GC mode */ 1765 unsigned int gc_segment_mode; /* GC state for reclaimed segments */ 1766 unsigned int gc_reclaimed_segs[MAX_GC_MODE]; /* Reclaimed segs for each mode */ 1767 1768 unsigned long seq_file_ra_mul; /* multiplier for ra_pages of seq. files in fadvise */ 1769 1770 int max_fragment_chunk; /* max chunk size for block fragmentation mode */ 1771 int max_fragment_hole; /* max hole size for block fragmentation mode */ 1772 1773 /* For atomic write statistics */ 1774 atomic64_t current_atomic_write; 1775 s64 peak_atomic_write; 1776 u64 committed_atomic_block; 1777 u64 revoked_atomic_block; 1778 1779 #ifdef CONFIG_F2FS_FS_COMPRESSION 1780 struct kmem_cache *page_array_slab; /* page array entry */ 1781 unsigned int page_array_slab_size; /* default page array slab size */ 1782 1783 /* For runtime compression statistics */ 1784 u64 compr_written_block; 1785 u64 compr_saved_block; 1786 u32 compr_new_inode; 1787 1788 /* For compressed block cache */ 1789 struct inode *compress_inode; /* cache compressed blocks */ 1790 unsigned int compress_percent; /* cache page percentage */ 1791 unsigned int compress_watermark; /* cache page watermark */ 1792 atomic_t compress_page_hit; /* cache hit count */ 1793 #endif 1794 1795 #ifdef CONFIG_F2FS_IOSTAT 1796 /* For app/fs IO statistics */ 1797 spinlock_t iostat_lock; 1798 unsigned long long iostat_count[NR_IO_TYPE]; 1799 unsigned long long iostat_bytes[NR_IO_TYPE]; 1800 unsigned long long prev_iostat_bytes[NR_IO_TYPE]; 1801 bool iostat_enable; 1802 unsigned long iostat_next_period; 1803 unsigned int iostat_period_ms; 1804 1805 /* For io latency related statistics info in one iostat period */ 1806 spinlock_t iostat_lat_lock; 1807 struct iostat_lat_info *iostat_io_lat; 1808 #endif 1809 }; 1810 1811 /* Definitions to access f2fs_sb_info */ 1812 #define BLKS_PER_SEG(sbi) \ 1813 ((sbi)->blocks_per_seg) 1814 #define BLKS_PER_SEC(sbi) \ 1815 ((sbi)->segs_per_sec << (sbi)->log_blocks_per_seg) 1816 #define SEGS_PER_SEC(sbi) \ 1817 ((sbi)->segs_per_sec) 1818 1819 __printf(3, 4) 1820 void f2fs_printk(struct f2fs_sb_info *sbi, bool limit_rate, const char *fmt, ...); 1821 1822 #define f2fs_err(sbi, fmt, ...) \ 1823 f2fs_printk(sbi, false, KERN_ERR fmt, ##__VA_ARGS__) 1824 #define f2fs_warn(sbi, fmt, ...) \ 1825 f2fs_printk(sbi, false, KERN_WARNING fmt, ##__VA_ARGS__) 1826 #define f2fs_notice(sbi, fmt, ...) \ 1827 f2fs_printk(sbi, false, KERN_NOTICE fmt, ##__VA_ARGS__) 1828 #define f2fs_info(sbi, fmt, ...) \ 1829 f2fs_printk(sbi, false, KERN_INFO fmt, ##__VA_ARGS__) 1830 #define f2fs_debug(sbi, fmt, ...) \ 1831 f2fs_printk(sbi, false, KERN_DEBUG fmt, ##__VA_ARGS__) 1832 1833 #define f2fs_err_ratelimited(sbi, fmt, ...) \ 1834 f2fs_printk(sbi, true, KERN_ERR fmt, ##__VA_ARGS__) 1835 #define f2fs_warn_ratelimited(sbi, fmt, ...) \ 1836 f2fs_printk(sbi, true, KERN_WARNING fmt, ##__VA_ARGS__) 1837 #define f2fs_info_ratelimited(sbi, fmt, ...) \ 1838 f2fs_printk(sbi, true, KERN_INFO fmt, ##__VA_ARGS__) 1839 1840 #ifdef CONFIG_F2FS_FAULT_INJECTION 1841 #define time_to_inject(sbi, type) __time_to_inject(sbi, type, __func__, \ 1842 __builtin_return_address(0)) 1843 static inline bool __time_to_inject(struct f2fs_sb_info *sbi, int type, 1844 const char *func, const char *parent_func) 1845 { 1846 struct f2fs_fault_info *ffi = &F2FS_OPTION(sbi).fault_info; 1847 1848 if (!ffi->inject_rate) 1849 return false; 1850 1851 if (!IS_FAULT_SET(ffi, type)) 1852 return false; 1853 1854 atomic_inc(&ffi->inject_ops); 1855 if (atomic_read(&ffi->inject_ops) >= ffi->inject_rate) { 1856 atomic_set(&ffi->inject_ops, 0); 1857 f2fs_info_ratelimited(sbi, "inject %s in %s of %pS", 1858 f2fs_fault_name[type], func, parent_func); 1859 return true; 1860 } 1861 return false; 1862 } 1863 #else 1864 static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type) 1865 { 1866 return false; 1867 } 1868 #endif 1869 1870 /* 1871 * Test if the mounted volume is a multi-device volume. 1872 * - For a single regular disk volume, sbi->s_ndevs is 0. 1873 * - For a single zoned disk volume, sbi->s_ndevs is 1. 1874 * - For a multi-device volume, sbi->s_ndevs is always 2 or more. 1875 */ 1876 static inline bool f2fs_is_multi_device(struct f2fs_sb_info *sbi) 1877 { 1878 return sbi->s_ndevs > 1; 1879 } 1880 1881 static inline void f2fs_update_time(struct f2fs_sb_info *sbi, int type) 1882 { 1883 unsigned long now = jiffies; 1884 1885 sbi->last_time[type] = now; 1886 1887 /* DISCARD_TIME and GC_TIME are based on REQ_TIME */ 1888 if (type == REQ_TIME) { 1889 sbi->last_time[DISCARD_TIME] = now; 1890 sbi->last_time[GC_TIME] = now; 1891 } 1892 } 1893 1894 static inline bool f2fs_time_over(struct f2fs_sb_info *sbi, int type) 1895 { 1896 unsigned long interval = sbi->interval_time[type] * HZ; 1897 1898 return time_after(jiffies, sbi->last_time[type] + interval); 1899 } 1900 1901 static inline unsigned int f2fs_time_to_wait(struct f2fs_sb_info *sbi, 1902 int type) 1903 { 1904 unsigned long interval = sbi->interval_time[type] * HZ; 1905 unsigned int wait_ms = 0; 1906 long delta; 1907 1908 delta = (sbi->last_time[type] + interval) - jiffies; 1909 if (delta > 0) 1910 wait_ms = jiffies_to_msecs(delta); 1911 1912 return wait_ms; 1913 } 1914 1915 /* 1916 * Inline functions 1917 */ 1918 static inline u32 __f2fs_crc32(struct f2fs_sb_info *sbi, u32 crc, 1919 const void *address, unsigned int length) 1920 { 1921 struct { 1922 struct shash_desc shash; 1923 char ctx[4]; 1924 } desc; 1925 int err; 1926 1927 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver) != sizeof(desc.ctx)); 1928 1929 desc.shash.tfm = sbi->s_chksum_driver; 1930 *(u32 *)desc.ctx = crc; 1931 1932 err = crypto_shash_update(&desc.shash, address, length); 1933 BUG_ON(err); 1934 1935 return *(u32 *)desc.ctx; 1936 } 1937 1938 static inline u32 f2fs_crc32(struct f2fs_sb_info *sbi, const void *address, 1939 unsigned int length) 1940 { 1941 return __f2fs_crc32(sbi, F2FS_SUPER_MAGIC, address, length); 1942 } 1943 1944 static inline bool f2fs_crc_valid(struct f2fs_sb_info *sbi, __u32 blk_crc, 1945 void *buf, size_t buf_size) 1946 { 1947 return f2fs_crc32(sbi, buf, buf_size) == blk_crc; 1948 } 1949 1950 static inline u32 f2fs_chksum(struct f2fs_sb_info *sbi, u32 crc, 1951 const void *address, unsigned int length) 1952 { 1953 return __f2fs_crc32(sbi, crc, address, length); 1954 } 1955 1956 static inline struct f2fs_inode_info *F2FS_I(struct inode *inode) 1957 { 1958 return container_of(inode, struct f2fs_inode_info, vfs_inode); 1959 } 1960 1961 static inline struct f2fs_sb_info *F2FS_SB(struct super_block *sb) 1962 { 1963 return sb->s_fs_info; 1964 } 1965 1966 static inline struct f2fs_sb_info *F2FS_I_SB(struct inode *inode) 1967 { 1968 return F2FS_SB(inode->i_sb); 1969 } 1970 1971 static inline struct f2fs_sb_info *F2FS_M_SB(struct address_space *mapping) 1972 { 1973 return F2FS_I_SB(mapping->host); 1974 } 1975 1976 static inline struct f2fs_sb_info *F2FS_P_SB(struct page *page) 1977 { 1978 return F2FS_M_SB(page_file_mapping(page)); 1979 } 1980 1981 static inline struct f2fs_super_block *F2FS_RAW_SUPER(struct f2fs_sb_info *sbi) 1982 { 1983 return (struct f2fs_super_block *)(sbi->raw_super); 1984 } 1985 1986 static inline struct f2fs_checkpoint *F2FS_CKPT(struct f2fs_sb_info *sbi) 1987 { 1988 return (struct f2fs_checkpoint *)(sbi->ckpt); 1989 } 1990 1991 static inline struct f2fs_node *F2FS_NODE(struct page *page) 1992 { 1993 return (struct f2fs_node *)page_address(page); 1994 } 1995 1996 static inline struct f2fs_inode *F2FS_INODE(struct page *page) 1997 { 1998 return &((struct f2fs_node *)page_address(page))->i; 1999 } 2000 2001 static inline struct f2fs_nm_info *NM_I(struct f2fs_sb_info *sbi) 2002 { 2003 return (struct f2fs_nm_info *)(sbi->nm_info); 2004 } 2005 2006 static inline struct f2fs_sm_info *SM_I(struct f2fs_sb_info *sbi) 2007 { 2008 return (struct f2fs_sm_info *)(sbi->sm_info); 2009 } 2010 2011 static inline struct sit_info *SIT_I(struct f2fs_sb_info *sbi) 2012 { 2013 return (struct sit_info *)(SM_I(sbi)->sit_info); 2014 } 2015 2016 static inline struct free_segmap_info *FREE_I(struct f2fs_sb_info *sbi) 2017 { 2018 return (struct free_segmap_info *)(SM_I(sbi)->free_info); 2019 } 2020 2021 static inline struct dirty_seglist_info *DIRTY_I(struct f2fs_sb_info *sbi) 2022 { 2023 return (struct dirty_seglist_info *)(SM_I(sbi)->dirty_info); 2024 } 2025 2026 static inline struct address_space *META_MAPPING(struct f2fs_sb_info *sbi) 2027 { 2028 return sbi->meta_inode->i_mapping; 2029 } 2030 2031 static inline struct address_space *NODE_MAPPING(struct f2fs_sb_info *sbi) 2032 { 2033 return sbi->node_inode->i_mapping; 2034 } 2035 2036 static inline bool is_sbi_flag_set(struct f2fs_sb_info *sbi, unsigned int type) 2037 { 2038 return test_bit(type, &sbi->s_flag); 2039 } 2040 2041 static inline void set_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type) 2042 { 2043 set_bit(type, &sbi->s_flag); 2044 } 2045 2046 static inline void clear_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type) 2047 { 2048 clear_bit(type, &sbi->s_flag); 2049 } 2050 2051 static inline unsigned long long cur_cp_version(struct f2fs_checkpoint *cp) 2052 { 2053 return le64_to_cpu(cp->checkpoint_ver); 2054 } 2055 2056 static inline unsigned long f2fs_qf_ino(struct super_block *sb, int type) 2057 { 2058 if (type < F2FS_MAX_QUOTAS) 2059 return le32_to_cpu(F2FS_SB(sb)->raw_super->qf_ino[type]); 2060 return 0; 2061 } 2062 2063 static inline __u64 cur_cp_crc(struct f2fs_checkpoint *cp) 2064 { 2065 size_t crc_offset = le32_to_cpu(cp->checksum_offset); 2066 return le32_to_cpu(*((__le32 *)((unsigned char *)cp + crc_offset))); 2067 } 2068 2069 static inline bool __is_set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 2070 { 2071 unsigned int ckpt_flags = le32_to_cpu(cp->ckpt_flags); 2072 2073 return ckpt_flags & f; 2074 } 2075 2076 static inline bool is_set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 2077 { 2078 return __is_set_ckpt_flags(F2FS_CKPT(sbi), f); 2079 } 2080 2081 static inline void __set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 2082 { 2083 unsigned int ckpt_flags; 2084 2085 ckpt_flags = le32_to_cpu(cp->ckpt_flags); 2086 ckpt_flags |= f; 2087 cp->ckpt_flags = cpu_to_le32(ckpt_flags); 2088 } 2089 2090 static inline void set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 2091 { 2092 unsigned long flags; 2093 2094 spin_lock_irqsave(&sbi->cp_lock, flags); 2095 __set_ckpt_flags(F2FS_CKPT(sbi), f); 2096 spin_unlock_irqrestore(&sbi->cp_lock, flags); 2097 } 2098 2099 static inline void __clear_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 2100 { 2101 unsigned int ckpt_flags; 2102 2103 ckpt_flags = le32_to_cpu(cp->ckpt_flags); 2104 ckpt_flags &= (~f); 2105 cp->ckpt_flags = cpu_to_le32(ckpt_flags); 2106 } 2107 2108 static inline void clear_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 2109 { 2110 unsigned long flags; 2111 2112 spin_lock_irqsave(&sbi->cp_lock, flags); 2113 __clear_ckpt_flags(F2FS_CKPT(sbi), f); 2114 spin_unlock_irqrestore(&sbi->cp_lock, flags); 2115 } 2116 2117 #define init_f2fs_rwsem(sem) \ 2118 do { \ 2119 static struct lock_class_key __key; \ 2120 \ 2121 __init_f2fs_rwsem((sem), #sem, &__key); \ 2122 } while (0) 2123 2124 static inline void __init_f2fs_rwsem(struct f2fs_rwsem *sem, 2125 const char *sem_name, struct lock_class_key *key) 2126 { 2127 __init_rwsem(&sem->internal_rwsem, sem_name, key); 2128 #ifdef CONFIG_F2FS_UNFAIR_RWSEM 2129 init_waitqueue_head(&sem->read_waiters); 2130 #endif 2131 } 2132 2133 static inline int f2fs_rwsem_is_locked(struct f2fs_rwsem *sem) 2134 { 2135 return rwsem_is_locked(&sem->internal_rwsem); 2136 } 2137 2138 static inline int f2fs_rwsem_is_contended(struct f2fs_rwsem *sem) 2139 { 2140 return rwsem_is_contended(&sem->internal_rwsem); 2141 } 2142 2143 static inline void f2fs_down_read(struct f2fs_rwsem *sem) 2144 { 2145 #ifdef CONFIG_F2FS_UNFAIR_RWSEM 2146 wait_event(sem->read_waiters, down_read_trylock(&sem->internal_rwsem)); 2147 #else 2148 down_read(&sem->internal_rwsem); 2149 #endif 2150 } 2151 2152 static inline int f2fs_down_read_trylock(struct f2fs_rwsem *sem) 2153 { 2154 return down_read_trylock(&sem->internal_rwsem); 2155 } 2156 2157 static inline void f2fs_up_read(struct f2fs_rwsem *sem) 2158 { 2159 up_read(&sem->internal_rwsem); 2160 } 2161 2162 static inline void f2fs_down_write(struct f2fs_rwsem *sem) 2163 { 2164 down_write(&sem->internal_rwsem); 2165 } 2166 2167 #ifdef CONFIG_DEBUG_LOCK_ALLOC 2168 static inline void f2fs_down_read_nested(struct f2fs_rwsem *sem, int subclass) 2169 { 2170 down_read_nested(&sem->internal_rwsem, subclass); 2171 } 2172 2173 static inline void f2fs_down_write_nested(struct f2fs_rwsem *sem, int subclass) 2174 { 2175 down_write_nested(&sem->internal_rwsem, subclass); 2176 } 2177 #else 2178 #define f2fs_down_read_nested(sem, subclass) f2fs_down_read(sem) 2179 #define f2fs_down_write_nested(sem, subclass) f2fs_down_write(sem) 2180 #endif 2181 2182 static inline int f2fs_down_write_trylock(struct f2fs_rwsem *sem) 2183 { 2184 return down_write_trylock(&sem->internal_rwsem); 2185 } 2186 2187 static inline void f2fs_up_write(struct f2fs_rwsem *sem) 2188 { 2189 up_write(&sem->internal_rwsem); 2190 #ifdef CONFIG_F2FS_UNFAIR_RWSEM 2191 wake_up_all(&sem->read_waiters); 2192 #endif 2193 } 2194 2195 static inline void f2fs_lock_op(struct f2fs_sb_info *sbi) 2196 { 2197 f2fs_down_read(&sbi->cp_rwsem); 2198 } 2199 2200 static inline int f2fs_trylock_op(struct f2fs_sb_info *sbi) 2201 { 2202 if (time_to_inject(sbi, FAULT_LOCK_OP)) 2203 return 0; 2204 return f2fs_down_read_trylock(&sbi->cp_rwsem); 2205 } 2206 2207 static inline void f2fs_unlock_op(struct f2fs_sb_info *sbi) 2208 { 2209 f2fs_up_read(&sbi->cp_rwsem); 2210 } 2211 2212 static inline void f2fs_lock_all(struct f2fs_sb_info *sbi) 2213 { 2214 f2fs_down_write(&sbi->cp_rwsem); 2215 } 2216 2217 static inline void f2fs_unlock_all(struct f2fs_sb_info *sbi) 2218 { 2219 f2fs_up_write(&sbi->cp_rwsem); 2220 } 2221 2222 static inline int __get_cp_reason(struct f2fs_sb_info *sbi) 2223 { 2224 int reason = CP_SYNC; 2225 2226 if (test_opt(sbi, FASTBOOT)) 2227 reason = CP_FASTBOOT; 2228 if (is_sbi_flag_set(sbi, SBI_IS_CLOSE)) 2229 reason = CP_UMOUNT; 2230 return reason; 2231 } 2232 2233 static inline bool __remain_node_summaries(int reason) 2234 { 2235 return (reason & (CP_UMOUNT | CP_FASTBOOT)); 2236 } 2237 2238 static inline bool __exist_node_summaries(struct f2fs_sb_info *sbi) 2239 { 2240 return (is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG) || 2241 is_set_ckpt_flags(sbi, CP_FASTBOOT_FLAG)); 2242 } 2243 2244 /* 2245 * Check whether the inode has blocks or not 2246 */ 2247 static inline int F2FS_HAS_BLOCKS(struct inode *inode) 2248 { 2249 block_t xattr_block = F2FS_I(inode)->i_xattr_nid ? 1 : 0; 2250 2251 return (inode->i_blocks >> F2FS_LOG_SECTORS_PER_BLOCK) > xattr_block; 2252 } 2253 2254 static inline bool f2fs_has_xattr_block(unsigned int ofs) 2255 { 2256 return ofs == XATTR_NODE_OFFSET; 2257 } 2258 2259 static inline bool __allow_reserved_blocks(struct f2fs_sb_info *sbi, 2260 struct inode *inode, bool cap) 2261 { 2262 if (!inode) 2263 return true; 2264 if (!test_opt(sbi, RESERVE_ROOT)) 2265 return false; 2266 if (IS_NOQUOTA(inode)) 2267 return true; 2268 if (uid_eq(F2FS_OPTION(sbi).s_resuid, current_fsuid())) 2269 return true; 2270 if (!gid_eq(F2FS_OPTION(sbi).s_resgid, GLOBAL_ROOT_GID) && 2271 in_group_p(F2FS_OPTION(sbi).s_resgid)) 2272 return true; 2273 if (cap && capable(CAP_SYS_RESOURCE)) 2274 return true; 2275 return false; 2276 } 2277 2278 static inline unsigned int get_available_block_count(struct f2fs_sb_info *sbi, 2279 struct inode *inode, bool cap) 2280 { 2281 block_t avail_user_block_count; 2282 2283 avail_user_block_count = sbi->user_block_count - 2284 sbi->current_reserved_blocks; 2285 2286 if (!__allow_reserved_blocks(sbi, inode, cap)) 2287 avail_user_block_count -= F2FS_OPTION(sbi).root_reserved_blocks; 2288 2289 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) { 2290 if (avail_user_block_count > sbi->unusable_block_count) 2291 avail_user_block_count -= sbi->unusable_block_count; 2292 else 2293 avail_user_block_count = 0; 2294 } 2295 2296 return avail_user_block_count; 2297 } 2298 2299 static inline void f2fs_i_blocks_write(struct inode *, block_t, bool, bool); 2300 static inline int inc_valid_block_count(struct f2fs_sb_info *sbi, 2301 struct inode *inode, blkcnt_t *count, bool partial) 2302 { 2303 long long diff = 0, release = 0; 2304 block_t avail_user_block_count; 2305 int ret; 2306 2307 ret = dquot_reserve_block(inode, *count); 2308 if (ret) 2309 return ret; 2310 2311 if (time_to_inject(sbi, FAULT_BLOCK)) { 2312 release = *count; 2313 goto release_quota; 2314 } 2315 2316 /* 2317 * let's increase this in prior to actual block count change in order 2318 * for f2fs_sync_file to avoid data races when deciding checkpoint. 2319 */ 2320 percpu_counter_add(&sbi->alloc_valid_block_count, (*count)); 2321 2322 spin_lock(&sbi->stat_lock); 2323 2324 avail_user_block_count = get_available_block_count(sbi, inode, true); 2325 diff = (long long)sbi->total_valid_block_count + *count - 2326 avail_user_block_count; 2327 if (unlikely(diff > 0)) { 2328 if (!partial) { 2329 spin_unlock(&sbi->stat_lock); 2330 release = *count; 2331 goto enospc; 2332 } 2333 if (diff > *count) 2334 diff = *count; 2335 *count -= diff; 2336 release = diff; 2337 if (!*count) { 2338 spin_unlock(&sbi->stat_lock); 2339 goto enospc; 2340 } 2341 } 2342 sbi->total_valid_block_count += (block_t)(*count); 2343 2344 spin_unlock(&sbi->stat_lock); 2345 2346 if (unlikely(release)) { 2347 percpu_counter_sub(&sbi->alloc_valid_block_count, release); 2348 dquot_release_reservation_block(inode, release); 2349 } 2350 f2fs_i_blocks_write(inode, *count, true, true); 2351 return 0; 2352 2353 enospc: 2354 percpu_counter_sub(&sbi->alloc_valid_block_count, release); 2355 release_quota: 2356 dquot_release_reservation_block(inode, release); 2357 return -ENOSPC; 2358 } 2359 2360 #define PAGE_PRIVATE_GET_FUNC(name, flagname) \ 2361 static inline bool page_private_##name(struct page *page) \ 2362 { \ 2363 return PagePrivate(page) && \ 2364 test_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)) && \ 2365 test_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \ 2366 } 2367 2368 #define PAGE_PRIVATE_SET_FUNC(name, flagname) \ 2369 static inline void set_page_private_##name(struct page *page) \ 2370 { \ 2371 if (!PagePrivate(page)) \ 2372 attach_page_private(page, (void *)0); \ 2373 set_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)); \ 2374 set_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \ 2375 } 2376 2377 #define PAGE_PRIVATE_CLEAR_FUNC(name, flagname) \ 2378 static inline void clear_page_private_##name(struct page *page) \ 2379 { \ 2380 clear_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \ 2381 if (page_private(page) == BIT(PAGE_PRIVATE_NOT_POINTER)) \ 2382 detach_page_private(page); \ 2383 } 2384 2385 PAGE_PRIVATE_GET_FUNC(nonpointer, NOT_POINTER); 2386 PAGE_PRIVATE_GET_FUNC(inline, INLINE_INODE); 2387 PAGE_PRIVATE_GET_FUNC(gcing, ONGOING_MIGRATION); 2388 2389 PAGE_PRIVATE_SET_FUNC(reference, REF_RESOURCE); 2390 PAGE_PRIVATE_SET_FUNC(inline, INLINE_INODE); 2391 PAGE_PRIVATE_SET_FUNC(gcing, ONGOING_MIGRATION); 2392 2393 PAGE_PRIVATE_CLEAR_FUNC(reference, REF_RESOURCE); 2394 PAGE_PRIVATE_CLEAR_FUNC(inline, INLINE_INODE); 2395 PAGE_PRIVATE_CLEAR_FUNC(gcing, ONGOING_MIGRATION); 2396 2397 static inline unsigned long get_page_private_data(struct page *page) 2398 { 2399 unsigned long data = page_private(page); 2400 2401 if (!test_bit(PAGE_PRIVATE_NOT_POINTER, &data)) 2402 return 0; 2403 return data >> PAGE_PRIVATE_MAX; 2404 } 2405 2406 static inline void set_page_private_data(struct page *page, unsigned long data) 2407 { 2408 if (!PagePrivate(page)) 2409 attach_page_private(page, (void *)0); 2410 set_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)); 2411 page_private(page) |= data << PAGE_PRIVATE_MAX; 2412 } 2413 2414 static inline void clear_page_private_data(struct page *page) 2415 { 2416 page_private(page) &= GENMASK(PAGE_PRIVATE_MAX - 1, 0); 2417 if (page_private(page) == BIT(PAGE_PRIVATE_NOT_POINTER)) 2418 detach_page_private(page); 2419 } 2420 2421 static inline void clear_page_private_all(struct page *page) 2422 { 2423 clear_page_private_data(page); 2424 clear_page_private_reference(page); 2425 clear_page_private_gcing(page); 2426 clear_page_private_inline(page); 2427 2428 f2fs_bug_on(F2FS_P_SB(page), page_private(page)); 2429 } 2430 2431 static inline void dec_valid_block_count(struct f2fs_sb_info *sbi, 2432 struct inode *inode, 2433 block_t count) 2434 { 2435 blkcnt_t sectors = count << F2FS_LOG_SECTORS_PER_BLOCK; 2436 2437 spin_lock(&sbi->stat_lock); 2438 f2fs_bug_on(sbi, sbi->total_valid_block_count < (block_t) count); 2439 sbi->total_valid_block_count -= (block_t)count; 2440 if (sbi->reserved_blocks && 2441 sbi->current_reserved_blocks < sbi->reserved_blocks) 2442 sbi->current_reserved_blocks = min(sbi->reserved_blocks, 2443 sbi->current_reserved_blocks + count); 2444 spin_unlock(&sbi->stat_lock); 2445 if (unlikely(inode->i_blocks < sectors)) { 2446 f2fs_warn(sbi, "Inconsistent i_blocks, ino:%lu, iblocks:%llu, sectors:%llu", 2447 inode->i_ino, 2448 (unsigned long long)inode->i_blocks, 2449 (unsigned long long)sectors); 2450 set_sbi_flag(sbi, SBI_NEED_FSCK); 2451 return; 2452 } 2453 f2fs_i_blocks_write(inode, count, false, true); 2454 } 2455 2456 static inline void inc_page_count(struct f2fs_sb_info *sbi, int count_type) 2457 { 2458 atomic_inc(&sbi->nr_pages[count_type]); 2459 2460 if (count_type == F2FS_DIRTY_DENTS || 2461 count_type == F2FS_DIRTY_NODES || 2462 count_type == F2FS_DIRTY_META || 2463 count_type == F2FS_DIRTY_QDATA || 2464 count_type == F2FS_DIRTY_IMETA) 2465 set_sbi_flag(sbi, SBI_IS_DIRTY); 2466 } 2467 2468 static inline void inode_inc_dirty_pages(struct inode *inode) 2469 { 2470 atomic_inc(&F2FS_I(inode)->dirty_pages); 2471 inc_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ? 2472 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA); 2473 if (IS_NOQUOTA(inode)) 2474 inc_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA); 2475 } 2476 2477 static inline void dec_page_count(struct f2fs_sb_info *sbi, int count_type) 2478 { 2479 atomic_dec(&sbi->nr_pages[count_type]); 2480 } 2481 2482 static inline void inode_dec_dirty_pages(struct inode *inode) 2483 { 2484 if (!S_ISDIR(inode->i_mode) && !S_ISREG(inode->i_mode) && 2485 !S_ISLNK(inode->i_mode)) 2486 return; 2487 2488 atomic_dec(&F2FS_I(inode)->dirty_pages); 2489 dec_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ? 2490 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA); 2491 if (IS_NOQUOTA(inode)) 2492 dec_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA); 2493 } 2494 2495 static inline void inc_atomic_write_cnt(struct inode *inode) 2496 { 2497 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 2498 struct f2fs_inode_info *fi = F2FS_I(inode); 2499 u64 current_write; 2500 2501 fi->atomic_write_cnt++; 2502 atomic64_inc(&sbi->current_atomic_write); 2503 current_write = atomic64_read(&sbi->current_atomic_write); 2504 if (current_write > sbi->peak_atomic_write) 2505 sbi->peak_atomic_write = current_write; 2506 } 2507 2508 static inline void release_atomic_write_cnt(struct inode *inode) 2509 { 2510 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 2511 struct f2fs_inode_info *fi = F2FS_I(inode); 2512 2513 atomic64_sub(fi->atomic_write_cnt, &sbi->current_atomic_write); 2514 fi->atomic_write_cnt = 0; 2515 } 2516 2517 static inline s64 get_pages(struct f2fs_sb_info *sbi, int count_type) 2518 { 2519 return atomic_read(&sbi->nr_pages[count_type]); 2520 } 2521 2522 static inline int get_dirty_pages(struct inode *inode) 2523 { 2524 return atomic_read(&F2FS_I(inode)->dirty_pages); 2525 } 2526 2527 static inline int get_blocktype_secs(struct f2fs_sb_info *sbi, int block_type) 2528 { 2529 return div_u64(get_pages(sbi, block_type) + BLKS_PER_SEC(sbi) - 1, 2530 BLKS_PER_SEC(sbi)); 2531 } 2532 2533 static inline block_t valid_user_blocks(struct f2fs_sb_info *sbi) 2534 { 2535 return sbi->total_valid_block_count; 2536 } 2537 2538 static inline block_t discard_blocks(struct f2fs_sb_info *sbi) 2539 { 2540 return sbi->discard_blks; 2541 } 2542 2543 static inline unsigned long __bitmap_size(struct f2fs_sb_info *sbi, int flag) 2544 { 2545 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi); 2546 2547 /* return NAT or SIT bitmap */ 2548 if (flag == NAT_BITMAP) 2549 return le32_to_cpu(ckpt->nat_ver_bitmap_bytesize); 2550 else if (flag == SIT_BITMAP) 2551 return le32_to_cpu(ckpt->sit_ver_bitmap_bytesize); 2552 2553 return 0; 2554 } 2555 2556 static inline block_t __cp_payload(struct f2fs_sb_info *sbi) 2557 { 2558 return le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_payload); 2559 } 2560 2561 static inline void *__bitmap_ptr(struct f2fs_sb_info *sbi, int flag) 2562 { 2563 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi); 2564 void *tmp_ptr = &ckpt->sit_nat_version_bitmap; 2565 int offset; 2566 2567 if (is_set_ckpt_flags(sbi, CP_LARGE_NAT_BITMAP_FLAG)) { 2568 offset = (flag == SIT_BITMAP) ? 2569 le32_to_cpu(ckpt->nat_ver_bitmap_bytesize) : 0; 2570 /* 2571 * if large_nat_bitmap feature is enabled, leave checksum 2572 * protection for all nat/sit bitmaps. 2573 */ 2574 return tmp_ptr + offset + sizeof(__le32); 2575 } 2576 2577 if (__cp_payload(sbi) > 0) { 2578 if (flag == NAT_BITMAP) 2579 return tmp_ptr; 2580 else 2581 return (unsigned char *)ckpt + F2FS_BLKSIZE; 2582 } else { 2583 offset = (flag == NAT_BITMAP) ? 2584 le32_to_cpu(ckpt->sit_ver_bitmap_bytesize) : 0; 2585 return tmp_ptr + offset; 2586 } 2587 } 2588 2589 static inline block_t __start_cp_addr(struct f2fs_sb_info *sbi) 2590 { 2591 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr); 2592 2593 if (sbi->cur_cp_pack == 2) 2594 start_addr += BLKS_PER_SEG(sbi); 2595 return start_addr; 2596 } 2597 2598 static inline block_t __start_cp_next_addr(struct f2fs_sb_info *sbi) 2599 { 2600 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr); 2601 2602 if (sbi->cur_cp_pack == 1) 2603 start_addr += BLKS_PER_SEG(sbi); 2604 return start_addr; 2605 } 2606 2607 static inline void __set_cp_next_pack(struct f2fs_sb_info *sbi) 2608 { 2609 sbi->cur_cp_pack = (sbi->cur_cp_pack == 1) ? 2 : 1; 2610 } 2611 2612 static inline block_t __start_sum_addr(struct f2fs_sb_info *sbi) 2613 { 2614 return le32_to_cpu(F2FS_CKPT(sbi)->cp_pack_start_sum); 2615 } 2616 2617 extern void f2fs_mark_inode_dirty_sync(struct inode *inode, bool sync); 2618 static inline int inc_valid_node_count(struct f2fs_sb_info *sbi, 2619 struct inode *inode, bool is_inode) 2620 { 2621 block_t valid_block_count; 2622 unsigned int valid_node_count; 2623 unsigned int avail_user_block_count; 2624 int err; 2625 2626 if (is_inode) { 2627 if (inode) { 2628 err = dquot_alloc_inode(inode); 2629 if (err) 2630 return err; 2631 } 2632 } else { 2633 err = dquot_reserve_block(inode, 1); 2634 if (err) 2635 return err; 2636 } 2637 2638 if (time_to_inject(sbi, FAULT_BLOCK)) 2639 goto enospc; 2640 2641 spin_lock(&sbi->stat_lock); 2642 2643 valid_block_count = sbi->total_valid_block_count + 1; 2644 avail_user_block_count = get_available_block_count(sbi, inode, false); 2645 2646 if (unlikely(valid_block_count > avail_user_block_count)) { 2647 spin_unlock(&sbi->stat_lock); 2648 goto enospc; 2649 } 2650 2651 valid_node_count = sbi->total_valid_node_count + 1; 2652 if (unlikely(valid_node_count > sbi->total_node_count)) { 2653 spin_unlock(&sbi->stat_lock); 2654 goto enospc; 2655 } 2656 2657 sbi->total_valid_node_count++; 2658 sbi->total_valid_block_count++; 2659 spin_unlock(&sbi->stat_lock); 2660 2661 if (inode) { 2662 if (is_inode) 2663 f2fs_mark_inode_dirty_sync(inode, true); 2664 else 2665 f2fs_i_blocks_write(inode, 1, true, true); 2666 } 2667 2668 percpu_counter_inc(&sbi->alloc_valid_block_count); 2669 return 0; 2670 2671 enospc: 2672 if (is_inode) { 2673 if (inode) 2674 dquot_free_inode(inode); 2675 } else { 2676 dquot_release_reservation_block(inode, 1); 2677 } 2678 return -ENOSPC; 2679 } 2680 2681 static inline void dec_valid_node_count(struct f2fs_sb_info *sbi, 2682 struct inode *inode, bool is_inode) 2683 { 2684 spin_lock(&sbi->stat_lock); 2685 2686 if (unlikely(!sbi->total_valid_block_count || 2687 !sbi->total_valid_node_count)) { 2688 f2fs_warn(sbi, "dec_valid_node_count: inconsistent block counts, total_valid_block:%u, total_valid_node:%u", 2689 sbi->total_valid_block_count, 2690 sbi->total_valid_node_count); 2691 set_sbi_flag(sbi, SBI_NEED_FSCK); 2692 } else { 2693 sbi->total_valid_block_count--; 2694 sbi->total_valid_node_count--; 2695 } 2696 2697 if (sbi->reserved_blocks && 2698 sbi->current_reserved_blocks < sbi->reserved_blocks) 2699 sbi->current_reserved_blocks++; 2700 2701 spin_unlock(&sbi->stat_lock); 2702 2703 if (is_inode) { 2704 dquot_free_inode(inode); 2705 } else { 2706 if (unlikely(inode->i_blocks == 0)) { 2707 f2fs_warn(sbi, "dec_valid_node_count: inconsistent i_blocks, ino:%lu, iblocks:%llu", 2708 inode->i_ino, 2709 (unsigned long long)inode->i_blocks); 2710 set_sbi_flag(sbi, SBI_NEED_FSCK); 2711 return; 2712 } 2713 f2fs_i_blocks_write(inode, 1, false, true); 2714 } 2715 } 2716 2717 static inline unsigned int valid_node_count(struct f2fs_sb_info *sbi) 2718 { 2719 return sbi->total_valid_node_count; 2720 } 2721 2722 static inline void inc_valid_inode_count(struct f2fs_sb_info *sbi) 2723 { 2724 percpu_counter_inc(&sbi->total_valid_inode_count); 2725 } 2726 2727 static inline void dec_valid_inode_count(struct f2fs_sb_info *sbi) 2728 { 2729 percpu_counter_dec(&sbi->total_valid_inode_count); 2730 } 2731 2732 static inline s64 valid_inode_count(struct f2fs_sb_info *sbi) 2733 { 2734 return percpu_counter_sum_positive(&sbi->total_valid_inode_count); 2735 } 2736 2737 static inline struct page *f2fs_grab_cache_page(struct address_space *mapping, 2738 pgoff_t index, bool for_write) 2739 { 2740 struct page *page; 2741 unsigned int flags; 2742 2743 if (IS_ENABLED(CONFIG_F2FS_FAULT_INJECTION)) { 2744 if (!for_write) 2745 page = find_get_page_flags(mapping, index, 2746 FGP_LOCK | FGP_ACCESSED); 2747 else 2748 page = find_lock_page(mapping, index); 2749 if (page) 2750 return page; 2751 2752 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_ALLOC)) 2753 return NULL; 2754 } 2755 2756 if (!for_write) 2757 return grab_cache_page(mapping, index); 2758 2759 flags = memalloc_nofs_save(); 2760 page = grab_cache_page_write_begin(mapping, index); 2761 memalloc_nofs_restore(flags); 2762 2763 return page; 2764 } 2765 2766 static inline struct page *f2fs_pagecache_get_page( 2767 struct address_space *mapping, pgoff_t index, 2768 fgf_t fgp_flags, gfp_t gfp_mask) 2769 { 2770 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_GET)) 2771 return NULL; 2772 2773 return pagecache_get_page(mapping, index, fgp_flags, gfp_mask); 2774 } 2775 2776 static inline void f2fs_put_page(struct page *page, int unlock) 2777 { 2778 if (!page) 2779 return; 2780 2781 if (unlock) { 2782 f2fs_bug_on(F2FS_P_SB(page), !PageLocked(page)); 2783 unlock_page(page); 2784 } 2785 put_page(page); 2786 } 2787 2788 static inline void f2fs_put_dnode(struct dnode_of_data *dn) 2789 { 2790 if (dn->node_page) 2791 f2fs_put_page(dn->node_page, 1); 2792 if (dn->inode_page && dn->node_page != dn->inode_page) 2793 f2fs_put_page(dn->inode_page, 0); 2794 dn->node_page = NULL; 2795 dn->inode_page = NULL; 2796 } 2797 2798 static inline struct kmem_cache *f2fs_kmem_cache_create(const char *name, 2799 size_t size) 2800 { 2801 return kmem_cache_create(name, size, 0, SLAB_RECLAIM_ACCOUNT, NULL); 2802 } 2803 2804 static inline void *f2fs_kmem_cache_alloc_nofail(struct kmem_cache *cachep, 2805 gfp_t flags) 2806 { 2807 void *entry; 2808 2809 entry = kmem_cache_alloc(cachep, flags); 2810 if (!entry) 2811 entry = kmem_cache_alloc(cachep, flags | __GFP_NOFAIL); 2812 return entry; 2813 } 2814 2815 static inline void *f2fs_kmem_cache_alloc(struct kmem_cache *cachep, 2816 gfp_t flags, bool nofail, struct f2fs_sb_info *sbi) 2817 { 2818 if (nofail) 2819 return f2fs_kmem_cache_alloc_nofail(cachep, flags); 2820 2821 if (time_to_inject(sbi, FAULT_SLAB_ALLOC)) 2822 return NULL; 2823 2824 return kmem_cache_alloc(cachep, flags); 2825 } 2826 2827 static inline bool is_inflight_io(struct f2fs_sb_info *sbi, int type) 2828 { 2829 if (get_pages(sbi, F2FS_RD_DATA) || get_pages(sbi, F2FS_RD_NODE) || 2830 get_pages(sbi, F2FS_RD_META) || get_pages(sbi, F2FS_WB_DATA) || 2831 get_pages(sbi, F2FS_WB_CP_DATA) || 2832 get_pages(sbi, F2FS_DIO_READ) || 2833 get_pages(sbi, F2FS_DIO_WRITE)) 2834 return true; 2835 2836 if (type != DISCARD_TIME && SM_I(sbi) && SM_I(sbi)->dcc_info && 2837 atomic_read(&SM_I(sbi)->dcc_info->queued_discard)) 2838 return true; 2839 2840 if (SM_I(sbi) && SM_I(sbi)->fcc_info && 2841 atomic_read(&SM_I(sbi)->fcc_info->queued_flush)) 2842 return true; 2843 return false; 2844 } 2845 2846 static inline bool is_idle(struct f2fs_sb_info *sbi, int type) 2847 { 2848 if (sbi->gc_mode == GC_URGENT_HIGH) 2849 return true; 2850 2851 if (is_inflight_io(sbi, type)) 2852 return false; 2853 2854 if (sbi->gc_mode == GC_URGENT_MID) 2855 return true; 2856 2857 if (sbi->gc_mode == GC_URGENT_LOW && 2858 (type == DISCARD_TIME || type == GC_TIME)) 2859 return true; 2860 2861 return f2fs_time_over(sbi, type); 2862 } 2863 2864 static inline void f2fs_radix_tree_insert(struct radix_tree_root *root, 2865 unsigned long index, void *item) 2866 { 2867 while (radix_tree_insert(root, index, item)) 2868 cond_resched(); 2869 } 2870 2871 #define RAW_IS_INODE(p) ((p)->footer.nid == (p)->footer.ino) 2872 2873 static inline bool IS_INODE(struct page *page) 2874 { 2875 struct f2fs_node *p = F2FS_NODE(page); 2876 2877 return RAW_IS_INODE(p); 2878 } 2879 2880 static inline int offset_in_addr(struct f2fs_inode *i) 2881 { 2882 return (i->i_inline & F2FS_EXTRA_ATTR) ? 2883 (le16_to_cpu(i->i_extra_isize) / sizeof(__le32)) : 0; 2884 } 2885 2886 static inline __le32 *blkaddr_in_node(struct f2fs_node *node) 2887 { 2888 return RAW_IS_INODE(node) ? node->i.i_addr : node->dn.addr; 2889 } 2890 2891 static inline int f2fs_has_extra_attr(struct inode *inode); 2892 static inline block_t data_blkaddr(struct inode *inode, 2893 struct page *node_page, unsigned int offset) 2894 { 2895 struct f2fs_node *raw_node; 2896 __le32 *addr_array; 2897 int base = 0; 2898 bool is_inode = IS_INODE(node_page); 2899 2900 raw_node = F2FS_NODE(node_page); 2901 2902 if (is_inode) { 2903 if (!inode) 2904 /* from GC path only */ 2905 base = offset_in_addr(&raw_node->i); 2906 else if (f2fs_has_extra_attr(inode)) 2907 base = get_extra_isize(inode); 2908 } 2909 2910 addr_array = blkaddr_in_node(raw_node); 2911 return le32_to_cpu(addr_array[base + offset]); 2912 } 2913 2914 static inline block_t f2fs_data_blkaddr(struct dnode_of_data *dn) 2915 { 2916 return data_blkaddr(dn->inode, dn->node_page, dn->ofs_in_node); 2917 } 2918 2919 static inline int f2fs_test_bit(unsigned int nr, char *addr) 2920 { 2921 int mask; 2922 2923 addr += (nr >> 3); 2924 mask = BIT(7 - (nr & 0x07)); 2925 return mask & *addr; 2926 } 2927 2928 static inline void f2fs_set_bit(unsigned int nr, char *addr) 2929 { 2930 int mask; 2931 2932 addr += (nr >> 3); 2933 mask = BIT(7 - (nr & 0x07)); 2934 *addr |= mask; 2935 } 2936 2937 static inline void f2fs_clear_bit(unsigned int nr, char *addr) 2938 { 2939 int mask; 2940 2941 addr += (nr >> 3); 2942 mask = BIT(7 - (nr & 0x07)); 2943 *addr &= ~mask; 2944 } 2945 2946 static inline int f2fs_test_and_set_bit(unsigned int nr, char *addr) 2947 { 2948 int mask; 2949 int ret; 2950 2951 addr += (nr >> 3); 2952 mask = BIT(7 - (nr & 0x07)); 2953 ret = mask & *addr; 2954 *addr |= mask; 2955 return ret; 2956 } 2957 2958 static inline int f2fs_test_and_clear_bit(unsigned int nr, char *addr) 2959 { 2960 int mask; 2961 int ret; 2962 2963 addr += (nr >> 3); 2964 mask = BIT(7 - (nr & 0x07)); 2965 ret = mask & *addr; 2966 *addr &= ~mask; 2967 return ret; 2968 } 2969 2970 static inline void f2fs_change_bit(unsigned int nr, char *addr) 2971 { 2972 int mask; 2973 2974 addr += (nr >> 3); 2975 mask = BIT(7 - (nr & 0x07)); 2976 *addr ^= mask; 2977 } 2978 2979 /* 2980 * On-disk inode flags (f2fs_inode::i_flags) 2981 */ 2982 #define F2FS_COMPR_FL 0x00000004 /* Compress file */ 2983 #define F2FS_SYNC_FL 0x00000008 /* Synchronous updates */ 2984 #define F2FS_IMMUTABLE_FL 0x00000010 /* Immutable file */ 2985 #define F2FS_APPEND_FL 0x00000020 /* writes to file may only append */ 2986 #define F2FS_NODUMP_FL 0x00000040 /* do not dump file */ 2987 #define F2FS_NOATIME_FL 0x00000080 /* do not update atime */ 2988 #define F2FS_NOCOMP_FL 0x00000400 /* Don't compress */ 2989 #define F2FS_INDEX_FL 0x00001000 /* hash-indexed directory */ 2990 #define F2FS_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */ 2991 #define F2FS_PROJINHERIT_FL 0x20000000 /* Create with parents projid */ 2992 #define F2FS_CASEFOLD_FL 0x40000000 /* Casefolded file */ 2993 2994 #define F2FS_QUOTA_DEFAULT_FL (F2FS_NOATIME_FL | F2FS_IMMUTABLE_FL) 2995 2996 /* Flags that should be inherited by new inodes from their parent. */ 2997 #define F2FS_FL_INHERITED (F2FS_SYNC_FL | F2FS_NODUMP_FL | F2FS_NOATIME_FL | \ 2998 F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \ 2999 F2FS_CASEFOLD_FL) 3000 3001 /* Flags that are appropriate for regular files (all but dir-specific ones). */ 3002 #define F2FS_REG_FLMASK (~(F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \ 3003 F2FS_CASEFOLD_FL)) 3004 3005 /* Flags that are appropriate for non-directories/regular files. */ 3006 #define F2FS_OTHER_FLMASK (F2FS_NODUMP_FL | F2FS_NOATIME_FL) 3007 3008 static inline __u32 f2fs_mask_flags(umode_t mode, __u32 flags) 3009 { 3010 if (S_ISDIR(mode)) 3011 return flags; 3012 else if (S_ISREG(mode)) 3013 return flags & F2FS_REG_FLMASK; 3014 else 3015 return flags & F2FS_OTHER_FLMASK; 3016 } 3017 3018 static inline void __mark_inode_dirty_flag(struct inode *inode, 3019 int flag, bool set) 3020 { 3021 switch (flag) { 3022 case FI_INLINE_XATTR: 3023 case FI_INLINE_DATA: 3024 case FI_INLINE_DENTRY: 3025 case FI_NEW_INODE: 3026 if (set) 3027 return; 3028 fallthrough; 3029 case FI_DATA_EXIST: 3030 case FI_INLINE_DOTS: 3031 case FI_PIN_FILE: 3032 case FI_COMPRESS_RELEASED: 3033 case FI_ATOMIC_COMMITTED: 3034 f2fs_mark_inode_dirty_sync(inode, true); 3035 } 3036 } 3037 3038 static inline void set_inode_flag(struct inode *inode, int flag) 3039 { 3040 set_bit(flag, F2FS_I(inode)->flags); 3041 __mark_inode_dirty_flag(inode, flag, true); 3042 } 3043 3044 static inline int is_inode_flag_set(struct inode *inode, int flag) 3045 { 3046 return test_bit(flag, F2FS_I(inode)->flags); 3047 } 3048 3049 static inline void clear_inode_flag(struct inode *inode, int flag) 3050 { 3051 clear_bit(flag, F2FS_I(inode)->flags); 3052 __mark_inode_dirty_flag(inode, flag, false); 3053 } 3054 3055 static inline bool f2fs_verity_in_progress(struct inode *inode) 3056 { 3057 return IS_ENABLED(CONFIG_FS_VERITY) && 3058 is_inode_flag_set(inode, FI_VERITY_IN_PROGRESS); 3059 } 3060 3061 static inline void set_acl_inode(struct inode *inode, umode_t mode) 3062 { 3063 F2FS_I(inode)->i_acl_mode = mode; 3064 set_inode_flag(inode, FI_ACL_MODE); 3065 f2fs_mark_inode_dirty_sync(inode, false); 3066 } 3067 3068 static inline void f2fs_i_links_write(struct inode *inode, bool inc) 3069 { 3070 if (inc) 3071 inc_nlink(inode); 3072 else 3073 drop_nlink(inode); 3074 f2fs_mark_inode_dirty_sync(inode, true); 3075 } 3076 3077 static inline void f2fs_i_blocks_write(struct inode *inode, 3078 block_t diff, bool add, bool claim) 3079 { 3080 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE); 3081 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER); 3082 3083 /* add = 1, claim = 1 should be dquot_reserve_block in pair */ 3084 if (add) { 3085 if (claim) 3086 dquot_claim_block(inode, diff); 3087 else 3088 dquot_alloc_block_nofail(inode, diff); 3089 } else { 3090 dquot_free_block(inode, diff); 3091 } 3092 3093 f2fs_mark_inode_dirty_sync(inode, true); 3094 if (clean || recover) 3095 set_inode_flag(inode, FI_AUTO_RECOVER); 3096 } 3097 3098 static inline bool f2fs_is_atomic_file(struct inode *inode); 3099 3100 static inline void f2fs_i_size_write(struct inode *inode, loff_t i_size) 3101 { 3102 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE); 3103 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER); 3104 3105 if (i_size_read(inode) == i_size) 3106 return; 3107 3108 i_size_write(inode, i_size); 3109 3110 if (f2fs_is_atomic_file(inode)) 3111 return; 3112 3113 f2fs_mark_inode_dirty_sync(inode, true); 3114 if (clean || recover) 3115 set_inode_flag(inode, FI_AUTO_RECOVER); 3116 } 3117 3118 static inline void f2fs_i_depth_write(struct inode *inode, unsigned int depth) 3119 { 3120 F2FS_I(inode)->i_current_depth = depth; 3121 f2fs_mark_inode_dirty_sync(inode, true); 3122 } 3123 3124 static inline void f2fs_i_gc_failures_write(struct inode *inode, 3125 unsigned int count) 3126 { 3127 F2FS_I(inode)->i_gc_failures[GC_FAILURE_PIN] = count; 3128 f2fs_mark_inode_dirty_sync(inode, true); 3129 } 3130 3131 static inline void f2fs_i_xnid_write(struct inode *inode, nid_t xnid) 3132 { 3133 F2FS_I(inode)->i_xattr_nid = xnid; 3134 f2fs_mark_inode_dirty_sync(inode, true); 3135 } 3136 3137 static inline void f2fs_i_pino_write(struct inode *inode, nid_t pino) 3138 { 3139 F2FS_I(inode)->i_pino = pino; 3140 f2fs_mark_inode_dirty_sync(inode, true); 3141 } 3142 3143 static inline void get_inline_info(struct inode *inode, struct f2fs_inode *ri) 3144 { 3145 struct f2fs_inode_info *fi = F2FS_I(inode); 3146 3147 if (ri->i_inline & F2FS_INLINE_XATTR) 3148 set_bit(FI_INLINE_XATTR, fi->flags); 3149 if (ri->i_inline & F2FS_INLINE_DATA) 3150 set_bit(FI_INLINE_DATA, fi->flags); 3151 if (ri->i_inline & F2FS_INLINE_DENTRY) 3152 set_bit(FI_INLINE_DENTRY, fi->flags); 3153 if (ri->i_inline & F2FS_DATA_EXIST) 3154 set_bit(FI_DATA_EXIST, fi->flags); 3155 if (ri->i_inline & F2FS_INLINE_DOTS) 3156 set_bit(FI_INLINE_DOTS, fi->flags); 3157 if (ri->i_inline & F2FS_EXTRA_ATTR) 3158 set_bit(FI_EXTRA_ATTR, fi->flags); 3159 if (ri->i_inline & F2FS_PIN_FILE) 3160 set_bit(FI_PIN_FILE, fi->flags); 3161 if (ri->i_inline & F2FS_COMPRESS_RELEASED) 3162 set_bit(FI_COMPRESS_RELEASED, fi->flags); 3163 } 3164 3165 static inline void set_raw_inline(struct inode *inode, struct f2fs_inode *ri) 3166 { 3167 ri->i_inline = 0; 3168 3169 if (is_inode_flag_set(inode, FI_INLINE_XATTR)) 3170 ri->i_inline |= F2FS_INLINE_XATTR; 3171 if (is_inode_flag_set(inode, FI_INLINE_DATA)) 3172 ri->i_inline |= F2FS_INLINE_DATA; 3173 if (is_inode_flag_set(inode, FI_INLINE_DENTRY)) 3174 ri->i_inline |= F2FS_INLINE_DENTRY; 3175 if (is_inode_flag_set(inode, FI_DATA_EXIST)) 3176 ri->i_inline |= F2FS_DATA_EXIST; 3177 if (is_inode_flag_set(inode, FI_INLINE_DOTS)) 3178 ri->i_inline |= F2FS_INLINE_DOTS; 3179 if (is_inode_flag_set(inode, FI_EXTRA_ATTR)) 3180 ri->i_inline |= F2FS_EXTRA_ATTR; 3181 if (is_inode_flag_set(inode, FI_PIN_FILE)) 3182 ri->i_inline |= F2FS_PIN_FILE; 3183 if (is_inode_flag_set(inode, FI_COMPRESS_RELEASED)) 3184 ri->i_inline |= F2FS_COMPRESS_RELEASED; 3185 } 3186 3187 static inline int f2fs_has_extra_attr(struct inode *inode) 3188 { 3189 return is_inode_flag_set(inode, FI_EXTRA_ATTR); 3190 } 3191 3192 static inline int f2fs_has_inline_xattr(struct inode *inode) 3193 { 3194 return is_inode_flag_set(inode, FI_INLINE_XATTR); 3195 } 3196 3197 static inline int f2fs_compressed_file(struct inode *inode) 3198 { 3199 return S_ISREG(inode->i_mode) && 3200 is_inode_flag_set(inode, FI_COMPRESSED_FILE); 3201 } 3202 3203 static inline bool f2fs_need_compress_data(struct inode *inode) 3204 { 3205 int compress_mode = F2FS_OPTION(F2FS_I_SB(inode)).compress_mode; 3206 3207 if (!f2fs_compressed_file(inode)) 3208 return false; 3209 3210 if (compress_mode == COMPR_MODE_FS) 3211 return true; 3212 else if (compress_mode == COMPR_MODE_USER && 3213 is_inode_flag_set(inode, FI_ENABLE_COMPRESS)) 3214 return true; 3215 3216 return false; 3217 } 3218 3219 static inline unsigned int addrs_per_inode(struct inode *inode) 3220 { 3221 unsigned int addrs = CUR_ADDRS_PER_INODE(inode) - 3222 get_inline_xattr_addrs(inode); 3223 3224 if (!f2fs_compressed_file(inode)) 3225 return addrs; 3226 return ALIGN_DOWN(addrs, F2FS_I(inode)->i_cluster_size); 3227 } 3228 3229 static inline unsigned int addrs_per_block(struct inode *inode) 3230 { 3231 if (!f2fs_compressed_file(inode)) 3232 return DEF_ADDRS_PER_BLOCK; 3233 return ALIGN_DOWN(DEF_ADDRS_PER_BLOCK, F2FS_I(inode)->i_cluster_size); 3234 } 3235 3236 static inline void *inline_xattr_addr(struct inode *inode, struct page *page) 3237 { 3238 struct f2fs_inode *ri = F2FS_INODE(page); 3239 3240 return (void *)&(ri->i_addr[DEF_ADDRS_PER_INODE - 3241 get_inline_xattr_addrs(inode)]); 3242 } 3243 3244 static inline int inline_xattr_size(struct inode *inode) 3245 { 3246 if (f2fs_has_inline_xattr(inode)) 3247 return get_inline_xattr_addrs(inode) * sizeof(__le32); 3248 return 0; 3249 } 3250 3251 /* 3252 * Notice: check inline_data flag without inode page lock is unsafe. 3253 * It could change at any time by f2fs_convert_inline_page(). 3254 */ 3255 static inline int f2fs_has_inline_data(struct inode *inode) 3256 { 3257 return is_inode_flag_set(inode, FI_INLINE_DATA); 3258 } 3259 3260 static inline int f2fs_exist_data(struct inode *inode) 3261 { 3262 return is_inode_flag_set(inode, FI_DATA_EXIST); 3263 } 3264 3265 static inline int f2fs_has_inline_dots(struct inode *inode) 3266 { 3267 return is_inode_flag_set(inode, FI_INLINE_DOTS); 3268 } 3269 3270 static inline int f2fs_is_mmap_file(struct inode *inode) 3271 { 3272 return is_inode_flag_set(inode, FI_MMAP_FILE); 3273 } 3274 3275 static inline bool f2fs_is_pinned_file(struct inode *inode) 3276 { 3277 return is_inode_flag_set(inode, FI_PIN_FILE); 3278 } 3279 3280 static inline bool f2fs_is_atomic_file(struct inode *inode) 3281 { 3282 return is_inode_flag_set(inode, FI_ATOMIC_FILE); 3283 } 3284 3285 static inline bool f2fs_is_cow_file(struct inode *inode) 3286 { 3287 return is_inode_flag_set(inode, FI_COW_FILE); 3288 } 3289 3290 static inline __le32 *get_dnode_addr(struct inode *inode, 3291 struct page *node_page); 3292 static inline void *inline_data_addr(struct inode *inode, struct page *page) 3293 { 3294 __le32 *addr = get_dnode_addr(inode, page); 3295 3296 return (void *)(addr + DEF_INLINE_RESERVED_SIZE); 3297 } 3298 3299 static inline int f2fs_has_inline_dentry(struct inode *inode) 3300 { 3301 return is_inode_flag_set(inode, FI_INLINE_DENTRY); 3302 } 3303 3304 static inline int is_file(struct inode *inode, int type) 3305 { 3306 return F2FS_I(inode)->i_advise & type; 3307 } 3308 3309 static inline void set_file(struct inode *inode, int type) 3310 { 3311 if (is_file(inode, type)) 3312 return; 3313 F2FS_I(inode)->i_advise |= type; 3314 f2fs_mark_inode_dirty_sync(inode, true); 3315 } 3316 3317 static inline void clear_file(struct inode *inode, int type) 3318 { 3319 if (!is_file(inode, type)) 3320 return; 3321 F2FS_I(inode)->i_advise &= ~type; 3322 f2fs_mark_inode_dirty_sync(inode, true); 3323 } 3324 3325 static inline bool f2fs_is_time_consistent(struct inode *inode) 3326 { 3327 struct timespec64 ctime = inode_get_ctime(inode); 3328 3329 if (!timespec64_equal(F2FS_I(inode)->i_disk_time, &inode->i_atime)) 3330 return false; 3331 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 1, &ctime)) 3332 return false; 3333 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 2, &inode->i_mtime)) 3334 return false; 3335 return true; 3336 } 3337 3338 static inline bool f2fs_skip_inode_update(struct inode *inode, int dsync) 3339 { 3340 bool ret; 3341 3342 if (dsync) { 3343 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 3344 3345 spin_lock(&sbi->inode_lock[DIRTY_META]); 3346 ret = list_empty(&F2FS_I(inode)->gdirty_list); 3347 spin_unlock(&sbi->inode_lock[DIRTY_META]); 3348 return ret; 3349 } 3350 if (!is_inode_flag_set(inode, FI_AUTO_RECOVER) || 3351 file_keep_isize(inode) || 3352 i_size_read(inode) & ~PAGE_MASK) 3353 return false; 3354 3355 if (!f2fs_is_time_consistent(inode)) 3356 return false; 3357 3358 spin_lock(&F2FS_I(inode)->i_size_lock); 3359 ret = F2FS_I(inode)->last_disk_size == i_size_read(inode); 3360 spin_unlock(&F2FS_I(inode)->i_size_lock); 3361 3362 return ret; 3363 } 3364 3365 static inline bool f2fs_readonly(struct super_block *sb) 3366 { 3367 return sb_rdonly(sb); 3368 } 3369 3370 static inline bool f2fs_cp_error(struct f2fs_sb_info *sbi) 3371 { 3372 return is_set_ckpt_flags(sbi, CP_ERROR_FLAG); 3373 } 3374 3375 static inline bool is_dot_dotdot(const u8 *name, size_t len) 3376 { 3377 if (len == 1 && name[0] == '.') 3378 return true; 3379 3380 if (len == 2 && name[0] == '.' && name[1] == '.') 3381 return true; 3382 3383 return false; 3384 } 3385 3386 static inline void *f2fs_kmalloc(struct f2fs_sb_info *sbi, 3387 size_t size, gfp_t flags) 3388 { 3389 if (time_to_inject(sbi, FAULT_KMALLOC)) 3390 return NULL; 3391 3392 return kmalloc(size, flags); 3393 } 3394 3395 static inline void *f2fs_getname(struct f2fs_sb_info *sbi) 3396 { 3397 if (time_to_inject(sbi, FAULT_KMALLOC)) 3398 return NULL; 3399 3400 return __getname(); 3401 } 3402 3403 static inline void f2fs_putname(char *buf) 3404 { 3405 __putname(buf); 3406 } 3407 3408 static inline void *f2fs_kzalloc(struct f2fs_sb_info *sbi, 3409 size_t size, gfp_t flags) 3410 { 3411 return f2fs_kmalloc(sbi, size, flags | __GFP_ZERO); 3412 } 3413 3414 static inline void *f2fs_kvmalloc(struct f2fs_sb_info *sbi, 3415 size_t size, gfp_t flags) 3416 { 3417 if (time_to_inject(sbi, FAULT_KVMALLOC)) 3418 return NULL; 3419 3420 return kvmalloc(size, flags); 3421 } 3422 3423 static inline void *f2fs_kvzalloc(struct f2fs_sb_info *sbi, 3424 size_t size, gfp_t flags) 3425 { 3426 return f2fs_kvmalloc(sbi, size, flags | __GFP_ZERO); 3427 } 3428 3429 static inline int get_extra_isize(struct inode *inode) 3430 { 3431 return F2FS_I(inode)->i_extra_isize / sizeof(__le32); 3432 } 3433 3434 static inline int get_inline_xattr_addrs(struct inode *inode) 3435 { 3436 return F2FS_I(inode)->i_inline_xattr_size; 3437 } 3438 3439 static inline __le32 *get_dnode_addr(struct inode *inode, 3440 struct page *node_page) 3441 { 3442 int base = 0; 3443 3444 if (IS_INODE(node_page) && f2fs_has_extra_attr(inode)) 3445 base = get_extra_isize(inode); 3446 3447 return blkaddr_in_node(F2FS_NODE(node_page)) + base; 3448 } 3449 3450 #define f2fs_get_inode_mode(i) \ 3451 ((is_inode_flag_set(i, FI_ACL_MODE)) ? \ 3452 (F2FS_I(i)->i_acl_mode) : ((i)->i_mode)) 3453 3454 #define F2FS_MIN_EXTRA_ATTR_SIZE (sizeof(__le32)) 3455 3456 #define F2FS_TOTAL_EXTRA_ATTR_SIZE \ 3457 (offsetof(struct f2fs_inode, i_extra_end) - \ 3458 offsetof(struct f2fs_inode, i_extra_isize)) \ 3459 3460 #define F2FS_OLD_ATTRIBUTE_SIZE (offsetof(struct f2fs_inode, i_addr)) 3461 #define F2FS_FITS_IN_INODE(f2fs_inode, extra_isize, field) \ 3462 ((offsetof(typeof(*(f2fs_inode)), field) + \ 3463 sizeof((f2fs_inode)->field)) \ 3464 <= (F2FS_OLD_ATTRIBUTE_SIZE + (extra_isize))) \ 3465 3466 #define __is_large_section(sbi) (SEGS_PER_SEC(sbi) > 1) 3467 3468 #define __is_meta_io(fio) (PAGE_TYPE_OF_BIO((fio)->type) == META) 3469 3470 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi, 3471 block_t blkaddr, int type); 3472 static inline void verify_blkaddr(struct f2fs_sb_info *sbi, 3473 block_t blkaddr, int type) 3474 { 3475 if (!f2fs_is_valid_blkaddr(sbi, blkaddr, type)) 3476 f2fs_err(sbi, "invalid blkaddr: %u, type: %d, run fsck to fix.", 3477 blkaddr, type); 3478 } 3479 3480 static inline bool __is_valid_data_blkaddr(block_t blkaddr) 3481 { 3482 if (blkaddr == NEW_ADDR || blkaddr == NULL_ADDR || 3483 blkaddr == COMPRESS_ADDR) 3484 return false; 3485 return true; 3486 } 3487 3488 /* 3489 * file.c 3490 */ 3491 int f2fs_sync_file(struct file *file, loff_t start, loff_t end, int datasync); 3492 int f2fs_do_truncate_blocks(struct inode *inode, u64 from, bool lock); 3493 int f2fs_truncate_blocks(struct inode *inode, u64 from, bool lock); 3494 int f2fs_truncate(struct inode *inode); 3495 int f2fs_getattr(struct mnt_idmap *idmap, const struct path *path, 3496 struct kstat *stat, u32 request_mask, unsigned int flags); 3497 int f2fs_setattr(struct mnt_idmap *idmap, struct dentry *dentry, 3498 struct iattr *attr); 3499 int f2fs_truncate_hole(struct inode *inode, pgoff_t pg_start, pgoff_t pg_end); 3500 void f2fs_truncate_data_blocks_range(struct dnode_of_data *dn, int count); 3501 int f2fs_precache_extents(struct inode *inode); 3502 int f2fs_fileattr_get(struct dentry *dentry, struct fileattr *fa); 3503 int f2fs_fileattr_set(struct mnt_idmap *idmap, 3504 struct dentry *dentry, struct fileattr *fa); 3505 long f2fs_ioctl(struct file *filp, unsigned int cmd, unsigned long arg); 3506 long f2fs_compat_ioctl(struct file *file, unsigned int cmd, unsigned long arg); 3507 int f2fs_transfer_project_quota(struct inode *inode, kprojid_t kprojid); 3508 int f2fs_pin_file_control(struct inode *inode, bool inc); 3509 3510 /* 3511 * inode.c 3512 */ 3513 void f2fs_set_inode_flags(struct inode *inode); 3514 bool f2fs_inode_chksum_verify(struct f2fs_sb_info *sbi, struct page *page); 3515 void f2fs_inode_chksum_set(struct f2fs_sb_info *sbi, struct page *page); 3516 struct inode *f2fs_iget(struct super_block *sb, unsigned long ino); 3517 struct inode *f2fs_iget_retry(struct super_block *sb, unsigned long ino); 3518 int f2fs_try_to_free_nats(struct f2fs_sb_info *sbi, int nr_shrink); 3519 void f2fs_update_inode(struct inode *inode, struct page *node_page); 3520 void f2fs_update_inode_page(struct inode *inode); 3521 int f2fs_write_inode(struct inode *inode, struct writeback_control *wbc); 3522 void f2fs_evict_inode(struct inode *inode); 3523 void f2fs_handle_failed_inode(struct inode *inode); 3524 3525 /* 3526 * namei.c 3527 */ 3528 int f2fs_update_extension_list(struct f2fs_sb_info *sbi, const char *name, 3529 bool hot, bool set); 3530 struct dentry *f2fs_get_parent(struct dentry *child); 3531 int f2fs_get_tmpfile(struct mnt_idmap *idmap, struct inode *dir, 3532 struct inode **new_inode); 3533 3534 /* 3535 * dir.c 3536 */ 3537 int f2fs_init_casefolded_name(const struct inode *dir, 3538 struct f2fs_filename *fname); 3539 int f2fs_setup_filename(struct inode *dir, const struct qstr *iname, 3540 int lookup, struct f2fs_filename *fname); 3541 int f2fs_prepare_lookup(struct inode *dir, struct dentry *dentry, 3542 struct f2fs_filename *fname); 3543 void f2fs_free_filename(struct f2fs_filename *fname); 3544 struct f2fs_dir_entry *f2fs_find_target_dentry(const struct f2fs_dentry_ptr *d, 3545 const struct f2fs_filename *fname, int *max_slots); 3546 int f2fs_fill_dentries(struct dir_context *ctx, struct f2fs_dentry_ptr *d, 3547 unsigned int start_pos, struct fscrypt_str *fstr); 3548 void f2fs_do_make_empty_dir(struct inode *inode, struct inode *parent, 3549 struct f2fs_dentry_ptr *d); 3550 struct page *f2fs_init_inode_metadata(struct inode *inode, struct inode *dir, 3551 const struct f2fs_filename *fname, struct page *dpage); 3552 void f2fs_update_parent_metadata(struct inode *dir, struct inode *inode, 3553 unsigned int current_depth); 3554 int f2fs_room_for_filename(const void *bitmap, int slots, int max_slots); 3555 void f2fs_drop_nlink(struct inode *dir, struct inode *inode); 3556 struct f2fs_dir_entry *__f2fs_find_entry(struct inode *dir, 3557 const struct f2fs_filename *fname, 3558 struct page **res_page); 3559 struct f2fs_dir_entry *f2fs_find_entry(struct inode *dir, 3560 const struct qstr *child, struct page **res_page); 3561 struct f2fs_dir_entry *f2fs_parent_dir(struct inode *dir, struct page **p); 3562 ino_t f2fs_inode_by_name(struct inode *dir, const struct qstr *qstr, 3563 struct page **page); 3564 void f2fs_set_link(struct inode *dir, struct f2fs_dir_entry *de, 3565 struct page *page, struct inode *inode); 3566 bool f2fs_has_enough_room(struct inode *dir, struct page *ipage, 3567 const struct f2fs_filename *fname); 3568 void f2fs_update_dentry(nid_t ino, umode_t mode, struct f2fs_dentry_ptr *d, 3569 const struct fscrypt_str *name, f2fs_hash_t name_hash, 3570 unsigned int bit_pos); 3571 int f2fs_add_regular_entry(struct inode *dir, const struct f2fs_filename *fname, 3572 struct inode *inode, nid_t ino, umode_t mode); 3573 int f2fs_add_dentry(struct inode *dir, const struct f2fs_filename *fname, 3574 struct inode *inode, nid_t ino, umode_t mode); 3575 int f2fs_do_add_link(struct inode *dir, const struct qstr *name, 3576 struct inode *inode, nid_t ino, umode_t mode); 3577 void f2fs_delete_entry(struct f2fs_dir_entry *dentry, struct page *page, 3578 struct inode *dir, struct inode *inode); 3579 int f2fs_do_tmpfile(struct inode *inode, struct inode *dir, 3580 struct f2fs_filename *fname); 3581 bool f2fs_empty_dir(struct inode *dir); 3582 3583 static inline int f2fs_add_link(struct dentry *dentry, struct inode *inode) 3584 { 3585 if (fscrypt_is_nokey_name(dentry)) 3586 return -ENOKEY; 3587 return f2fs_do_add_link(d_inode(dentry->d_parent), &dentry->d_name, 3588 inode, inode->i_ino, inode->i_mode); 3589 } 3590 3591 /* 3592 * super.c 3593 */ 3594 int f2fs_inode_dirtied(struct inode *inode, bool sync); 3595 void f2fs_inode_synced(struct inode *inode); 3596 int f2fs_dquot_initialize(struct inode *inode); 3597 int f2fs_enable_quota_files(struct f2fs_sb_info *sbi, bool rdonly); 3598 int f2fs_quota_sync(struct super_block *sb, int type); 3599 loff_t max_file_blocks(struct inode *inode); 3600 void f2fs_quota_off_umount(struct super_block *sb); 3601 void f2fs_save_errors(struct f2fs_sb_info *sbi, unsigned char flag); 3602 void f2fs_handle_critical_error(struct f2fs_sb_info *sbi, unsigned char reason, 3603 bool irq_context); 3604 void f2fs_handle_error(struct f2fs_sb_info *sbi, unsigned char error); 3605 void f2fs_handle_error_async(struct f2fs_sb_info *sbi, unsigned char error); 3606 int f2fs_commit_super(struct f2fs_sb_info *sbi, bool recover); 3607 int f2fs_sync_fs(struct super_block *sb, int sync); 3608 int f2fs_sanity_check_ckpt(struct f2fs_sb_info *sbi); 3609 3610 /* 3611 * hash.c 3612 */ 3613 void f2fs_hash_filename(const struct inode *dir, struct f2fs_filename *fname); 3614 3615 /* 3616 * node.c 3617 */ 3618 struct node_info; 3619 3620 int f2fs_check_nid_range(struct f2fs_sb_info *sbi, nid_t nid); 3621 bool f2fs_available_free_memory(struct f2fs_sb_info *sbi, int type); 3622 bool f2fs_in_warm_node_list(struct f2fs_sb_info *sbi, struct page *page); 3623 void f2fs_init_fsync_node_info(struct f2fs_sb_info *sbi); 3624 void f2fs_del_fsync_node_entry(struct f2fs_sb_info *sbi, struct page *page); 3625 void f2fs_reset_fsync_node_info(struct f2fs_sb_info *sbi); 3626 int f2fs_need_dentry_mark(struct f2fs_sb_info *sbi, nid_t nid); 3627 bool f2fs_is_checkpointed_node(struct f2fs_sb_info *sbi, nid_t nid); 3628 bool f2fs_need_inode_block_update(struct f2fs_sb_info *sbi, nid_t ino); 3629 int f2fs_get_node_info(struct f2fs_sb_info *sbi, nid_t nid, 3630 struct node_info *ni, bool checkpoint_context); 3631 pgoff_t f2fs_get_next_page_offset(struct dnode_of_data *dn, pgoff_t pgofs); 3632 int f2fs_get_dnode_of_data(struct dnode_of_data *dn, pgoff_t index, int mode); 3633 int f2fs_truncate_inode_blocks(struct inode *inode, pgoff_t from); 3634 int f2fs_truncate_xattr_node(struct inode *inode); 3635 int f2fs_wait_on_node_pages_writeback(struct f2fs_sb_info *sbi, 3636 unsigned int seq_id); 3637 bool f2fs_nat_bitmap_enabled(struct f2fs_sb_info *sbi); 3638 int f2fs_remove_inode_page(struct inode *inode); 3639 struct page *f2fs_new_inode_page(struct inode *inode); 3640 struct page *f2fs_new_node_page(struct dnode_of_data *dn, unsigned int ofs); 3641 void f2fs_ra_node_page(struct f2fs_sb_info *sbi, nid_t nid); 3642 struct page *f2fs_get_node_page(struct f2fs_sb_info *sbi, pgoff_t nid); 3643 struct page *f2fs_get_node_page_ra(struct page *parent, int start); 3644 int f2fs_move_node_page(struct page *node_page, int gc_type); 3645 void f2fs_flush_inline_data(struct f2fs_sb_info *sbi); 3646 int f2fs_fsync_node_pages(struct f2fs_sb_info *sbi, struct inode *inode, 3647 struct writeback_control *wbc, bool atomic, 3648 unsigned int *seq_id); 3649 int f2fs_sync_node_pages(struct f2fs_sb_info *sbi, 3650 struct writeback_control *wbc, 3651 bool do_balance, enum iostat_type io_type); 3652 int f2fs_build_free_nids(struct f2fs_sb_info *sbi, bool sync, bool mount); 3653 bool f2fs_alloc_nid(struct f2fs_sb_info *sbi, nid_t *nid); 3654 void f2fs_alloc_nid_done(struct f2fs_sb_info *sbi, nid_t nid); 3655 void f2fs_alloc_nid_failed(struct f2fs_sb_info *sbi, nid_t nid); 3656 int f2fs_try_to_free_nids(struct f2fs_sb_info *sbi, int nr_shrink); 3657 int f2fs_recover_inline_xattr(struct inode *inode, struct page *page); 3658 int f2fs_recover_xattr_data(struct inode *inode, struct page *page); 3659 int f2fs_recover_inode_page(struct f2fs_sb_info *sbi, struct page *page); 3660 int f2fs_restore_node_summary(struct f2fs_sb_info *sbi, 3661 unsigned int segno, struct f2fs_summary_block *sum); 3662 void f2fs_enable_nat_bits(struct f2fs_sb_info *sbi); 3663 int f2fs_flush_nat_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3664 int f2fs_build_node_manager(struct f2fs_sb_info *sbi); 3665 void f2fs_destroy_node_manager(struct f2fs_sb_info *sbi); 3666 int __init f2fs_create_node_manager_caches(void); 3667 void f2fs_destroy_node_manager_caches(void); 3668 3669 /* 3670 * segment.c 3671 */ 3672 bool f2fs_need_SSR(struct f2fs_sb_info *sbi); 3673 int f2fs_commit_atomic_write(struct inode *inode); 3674 void f2fs_abort_atomic_write(struct inode *inode, bool clean); 3675 void f2fs_balance_fs(struct f2fs_sb_info *sbi, bool need); 3676 void f2fs_balance_fs_bg(struct f2fs_sb_info *sbi, bool from_bg); 3677 int f2fs_issue_flush(struct f2fs_sb_info *sbi, nid_t ino); 3678 int f2fs_create_flush_cmd_control(struct f2fs_sb_info *sbi); 3679 int f2fs_flush_device_cache(struct f2fs_sb_info *sbi); 3680 void f2fs_destroy_flush_cmd_control(struct f2fs_sb_info *sbi, bool free); 3681 void f2fs_invalidate_blocks(struct f2fs_sb_info *sbi, block_t addr); 3682 bool f2fs_is_checkpointed_data(struct f2fs_sb_info *sbi, block_t blkaddr); 3683 int f2fs_start_discard_thread(struct f2fs_sb_info *sbi); 3684 void f2fs_drop_discard_cmd(struct f2fs_sb_info *sbi); 3685 void f2fs_stop_discard_thread(struct f2fs_sb_info *sbi); 3686 bool f2fs_issue_discard_timeout(struct f2fs_sb_info *sbi); 3687 void f2fs_clear_prefree_segments(struct f2fs_sb_info *sbi, 3688 struct cp_control *cpc); 3689 void f2fs_dirty_to_prefree(struct f2fs_sb_info *sbi); 3690 block_t f2fs_get_unusable_blocks(struct f2fs_sb_info *sbi); 3691 int f2fs_disable_cp_again(struct f2fs_sb_info *sbi, block_t unusable); 3692 void f2fs_release_discard_addrs(struct f2fs_sb_info *sbi); 3693 int f2fs_npages_for_summary_flush(struct f2fs_sb_info *sbi, bool for_ra); 3694 bool f2fs_segment_has_free_slot(struct f2fs_sb_info *sbi, int segno); 3695 void f2fs_init_inmem_curseg(struct f2fs_sb_info *sbi); 3696 void f2fs_save_inmem_curseg(struct f2fs_sb_info *sbi); 3697 void f2fs_restore_inmem_curseg(struct f2fs_sb_info *sbi); 3698 void f2fs_get_new_segment(struct f2fs_sb_info *sbi, 3699 unsigned int *newseg, bool new_sec, int dir); 3700 void f2fs_allocate_segment_for_resize(struct f2fs_sb_info *sbi, int type, 3701 unsigned int start, unsigned int end); 3702 int f2fs_allocate_new_section(struct f2fs_sb_info *sbi, int type, bool force); 3703 int f2fs_allocate_pinning_section(struct f2fs_sb_info *sbi); 3704 void f2fs_allocate_new_segments(struct f2fs_sb_info *sbi); 3705 int f2fs_trim_fs(struct f2fs_sb_info *sbi, struct fstrim_range *range); 3706 bool f2fs_exist_trim_candidates(struct f2fs_sb_info *sbi, 3707 struct cp_control *cpc); 3708 struct page *f2fs_get_sum_page(struct f2fs_sb_info *sbi, unsigned int segno); 3709 void f2fs_update_meta_page(struct f2fs_sb_info *sbi, void *src, 3710 block_t blk_addr); 3711 void f2fs_do_write_meta_page(struct f2fs_sb_info *sbi, struct page *page, 3712 enum iostat_type io_type); 3713 void f2fs_do_write_node_page(unsigned int nid, struct f2fs_io_info *fio); 3714 void f2fs_outplace_write_data(struct dnode_of_data *dn, 3715 struct f2fs_io_info *fio); 3716 int f2fs_inplace_write_data(struct f2fs_io_info *fio); 3717 void f2fs_do_replace_block(struct f2fs_sb_info *sbi, struct f2fs_summary *sum, 3718 block_t old_blkaddr, block_t new_blkaddr, 3719 bool recover_curseg, bool recover_newaddr, 3720 bool from_gc); 3721 void f2fs_replace_block(struct f2fs_sb_info *sbi, struct dnode_of_data *dn, 3722 block_t old_addr, block_t new_addr, 3723 unsigned char version, bool recover_curseg, 3724 bool recover_newaddr); 3725 void f2fs_allocate_data_block(struct f2fs_sb_info *sbi, struct page *page, 3726 block_t old_blkaddr, block_t *new_blkaddr, 3727 struct f2fs_summary *sum, int type, 3728 struct f2fs_io_info *fio); 3729 void f2fs_update_device_state(struct f2fs_sb_info *sbi, nid_t ino, 3730 block_t blkaddr, unsigned int blkcnt); 3731 void f2fs_wait_on_page_writeback(struct page *page, 3732 enum page_type type, bool ordered, bool locked); 3733 void f2fs_wait_on_block_writeback(struct inode *inode, block_t blkaddr); 3734 void f2fs_wait_on_block_writeback_range(struct inode *inode, block_t blkaddr, 3735 block_t len); 3736 void f2fs_write_data_summaries(struct f2fs_sb_info *sbi, block_t start_blk); 3737 void f2fs_write_node_summaries(struct f2fs_sb_info *sbi, block_t start_blk); 3738 int f2fs_lookup_journal_in_cursum(struct f2fs_journal *journal, int type, 3739 unsigned int val, int alloc); 3740 void f2fs_flush_sit_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3741 int f2fs_fix_curseg_write_pointer(struct f2fs_sb_info *sbi); 3742 int f2fs_check_write_pointer(struct f2fs_sb_info *sbi); 3743 int f2fs_build_segment_manager(struct f2fs_sb_info *sbi); 3744 void f2fs_destroy_segment_manager(struct f2fs_sb_info *sbi); 3745 int __init f2fs_create_segment_manager_caches(void); 3746 void f2fs_destroy_segment_manager_caches(void); 3747 int f2fs_rw_hint_to_seg_type(enum rw_hint hint); 3748 unsigned int f2fs_usable_segs_in_sec(struct f2fs_sb_info *sbi, 3749 unsigned int segno); 3750 unsigned int f2fs_usable_blks_in_seg(struct f2fs_sb_info *sbi, 3751 unsigned int segno); 3752 3753 #define DEF_FRAGMENT_SIZE 4 3754 #define MIN_FRAGMENT_SIZE 1 3755 #define MAX_FRAGMENT_SIZE 512 3756 3757 static inline bool f2fs_need_rand_seg(struct f2fs_sb_info *sbi) 3758 { 3759 return F2FS_OPTION(sbi).fs_mode == FS_MODE_FRAGMENT_SEG || 3760 F2FS_OPTION(sbi).fs_mode == FS_MODE_FRAGMENT_BLK; 3761 } 3762 3763 /* 3764 * checkpoint.c 3765 */ 3766 void f2fs_stop_checkpoint(struct f2fs_sb_info *sbi, bool end_io, 3767 unsigned char reason); 3768 void f2fs_flush_ckpt_thread(struct f2fs_sb_info *sbi); 3769 struct page *f2fs_grab_meta_page(struct f2fs_sb_info *sbi, pgoff_t index); 3770 struct page *f2fs_get_meta_page(struct f2fs_sb_info *sbi, pgoff_t index); 3771 struct page *f2fs_get_meta_page_retry(struct f2fs_sb_info *sbi, pgoff_t index); 3772 struct page *f2fs_get_tmp_page(struct f2fs_sb_info *sbi, pgoff_t index); 3773 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi, 3774 block_t blkaddr, int type); 3775 int f2fs_ra_meta_pages(struct f2fs_sb_info *sbi, block_t start, int nrpages, 3776 int type, bool sync); 3777 void f2fs_ra_meta_pages_cond(struct f2fs_sb_info *sbi, pgoff_t index, 3778 unsigned int ra_blocks); 3779 long f2fs_sync_meta_pages(struct f2fs_sb_info *sbi, enum page_type type, 3780 long nr_to_write, enum iostat_type io_type); 3781 void f2fs_add_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type); 3782 void f2fs_remove_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type); 3783 void f2fs_release_ino_entry(struct f2fs_sb_info *sbi, bool all); 3784 bool f2fs_exist_written_data(struct f2fs_sb_info *sbi, nid_t ino, int mode); 3785 void f2fs_set_dirty_device(struct f2fs_sb_info *sbi, nid_t ino, 3786 unsigned int devidx, int type); 3787 bool f2fs_is_dirty_device(struct f2fs_sb_info *sbi, nid_t ino, 3788 unsigned int devidx, int type); 3789 int f2fs_acquire_orphan_inode(struct f2fs_sb_info *sbi); 3790 void f2fs_release_orphan_inode(struct f2fs_sb_info *sbi); 3791 void f2fs_add_orphan_inode(struct inode *inode); 3792 void f2fs_remove_orphan_inode(struct f2fs_sb_info *sbi, nid_t ino); 3793 int f2fs_recover_orphan_inodes(struct f2fs_sb_info *sbi); 3794 int f2fs_get_valid_checkpoint(struct f2fs_sb_info *sbi); 3795 void f2fs_update_dirty_folio(struct inode *inode, struct folio *folio); 3796 void f2fs_remove_dirty_inode(struct inode *inode); 3797 int f2fs_sync_dirty_inodes(struct f2fs_sb_info *sbi, enum inode_type type, 3798 bool from_cp); 3799 void f2fs_wait_on_all_pages(struct f2fs_sb_info *sbi, int type); 3800 u64 f2fs_get_sectors_written(struct f2fs_sb_info *sbi); 3801 int f2fs_write_checkpoint(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3802 void f2fs_init_ino_entry_info(struct f2fs_sb_info *sbi); 3803 int __init f2fs_create_checkpoint_caches(void); 3804 void f2fs_destroy_checkpoint_caches(void); 3805 int f2fs_issue_checkpoint(struct f2fs_sb_info *sbi); 3806 int f2fs_start_ckpt_thread(struct f2fs_sb_info *sbi); 3807 void f2fs_stop_ckpt_thread(struct f2fs_sb_info *sbi); 3808 void f2fs_init_ckpt_req_control(struct f2fs_sb_info *sbi); 3809 3810 /* 3811 * data.c 3812 */ 3813 int __init f2fs_init_bioset(void); 3814 void f2fs_destroy_bioset(void); 3815 bool f2fs_is_cp_guaranteed(struct page *page); 3816 int f2fs_init_bio_entry_cache(void); 3817 void f2fs_destroy_bio_entry_cache(void); 3818 void f2fs_submit_read_bio(struct f2fs_sb_info *sbi, struct bio *bio, 3819 enum page_type type); 3820 int f2fs_init_write_merge_io(struct f2fs_sb_info *sbi); 3821 void f2fs_submit_merged_write(struct f2fs_sb_info *sbi, enum page_type type); 3822 void f2fs_submit_merged_write_cond(struct f2fs_sb_info *sbi, 3823 struct inode *inode, struct page *page, 3824 nid_t ino, enum page_type type); 3825 void f2fs_submit_merged_ipu_write(struct f2fs_sb_info *sbi, 3826 struct bio **bio, struct page *page); 3827 void f2fs_flush_merged_writes(struct f2fs_sb_info *sbi); 3828 int f2fs_submit_page_bio(struct f2fs_io_info *fio); 3829 int f2fs_merge_page_bio(struct f2fs_io_info *fio); 3830 void f2fs_submit_page_write(struct f2fs_io_info *fio); 3831 struct block_device *f2fs_target_device(struct f2fs_sb_info *sbi, 3832 block_t blk_addr, sector_t *sector); 3833 int f2fs_target_device_index(struct f2fs_sb_info *sbi, block_t blkaddr); 3834 void f2fs_set_data_blkaddr(struct dnode_of_data *dn, block_t blkaddr); 3835 void f2fs_update_data_blkaddr(struct dnode_of_data *dn, block_t blkaddr); 3836 int f2fs_reserve_new_blocks(struct dnode_of_data *dn, blkcnt_t count); 3837 int f2fs_reserve_new_block(struct dnode_of_data *dn); 3838 int f2fs_get_block_locked(struct dnode_of_data *dn, pgoff_t index); 3839 int f2fs_reserve_block(struct dnode_of_data *dn, pgoff_t index); 3840 struct page *f2fs_get_read_data_page(struct inode *inode, pgoff_t index, 3841 blk_opf_t op_flags, bool for_write, pgoff_t *next_pgofs); 3842 struct page *f2fs_find_data_page(struct inode *inode, pgoff_t index, 3843 pgoff_t *next_pgofs); 3844 struct page *f2fs_get_lock_data_page(struct inode *inode, pgoff_t index, 3845 bool for_write); 3846 struct page *f2fs_get_new_data_page(struct inode *inode, 3847 struct page *ipage, pgoff_t index, bool new_i_size); 3848 int f2fs_do_write_data_page(struct f2fs_io_info *fio); 3849 int f2fs_map_blocks(struct inode *inode, struct f2fs_map_blocks *map, int flag); 3850 int f2fs_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, 3851 u64 start, u64 len); 3852 int f2fs_encrypt_one_page(struct f2fs_io_info *fio); 3853 bool f2fs_should_update_inplace(struct inode *inode, struct f2fs_io_info *fio); 3854 bool f2fs_should_update_outplace(struct inode *inode, struct f2fs_io_info *fio); 3855 int f2fs_write_single_data_page(struct page *page, int *submitted, 3856 struct bio **bio, sector_t *last_block, 3857 struct writeback_control *wbc, 3858 enum iostat_type io_type, 3859 int compr_blocks, bool allow_balance); 3860 void f2fs_write_failed(struct inode *inode, loff_t to); 3861 void f2fs_invalidate_folio(struct folio *folio, size_t offset, size_t length); 3862 bool f2fs_release_folio(struct folio *folio, gfp_t wait); 3863 bool f2fs_overwrite_io(struct inode *inode, loff_t pos, size_t len); 3864 void f2fs_clear_page_cache_dirty_tag(struct page *page); 3865 int f2fs_init_post_read_processing(void); 3866 void f2fs_destroy_post_read_processing(void); 3867 int f2fs_init_post_read_wq(struct f2fs_sb_info *sbi); 3868 void f2fs_destroy_post_read_wq(struct f2fs_sb_info *sbi); 3869 extern const struct iomap_ops f2fs_iomap_ops; 3870 3871 /* 3872 * gc.c 3873 */ 3874 int f2fs_start_gc_thread(struct f2fs_sb_info *sbi); 3875 void f2fs_stop_gc_thread(struct f2fs_sb_info *sbi); 3876 block_t f2fs_start_bidx_of_node(unsigned int node_ofs, struct inode *inode); 3877 int f2fs_gc(struct f2fs_sb_info *sbi, struct f2fs_gc_control *gc_control); 3878 void f2fs_build_gc_manager(struct f2fs_sb_info *sbi); 3879 int f2fs_gc_range(struct f2fs_sb_info *sbi, 3880 unsigned int start_seg, unsigned int end_seg, 3881 bool dry_run, unsigned int dry_run_sections); 3882 int f2fs_resize_fs(struct file *filp, __u64 block_count); 3883 int __init f2fs_create_garbage_collection_cache(void); 3884 void f2fs_destroy_garbage_collection_cache(void); 3885 /* victim selection function for cleaning and SSR */ 3886 int f2fs_get_victim(struct f2fs_sb_info *sbi, unsigned int *result, 3887 int gc_type, int type, char alloc_mode, 3888 unsigned long long age); 3889 3890 /* 3891 * recovery.c 3892 */ 3893 int f2fs_recover_fsync_data(struct f2fs_sb_info *sbi, bool check_only); 3894 bool f2fs_space_for_roll_forward(struct f2fs_sb_info *sbi); 3895 int __init f2fs_create_recovery_cache(void); 3896 void f2fs_destroy_recovery_cache(void); 3897 3898 /* 3899 * debug.c 3900 */ 3901 #ifdef CONFIG_F2FS_STAT_FS 3902 struct f2fs_stat_info { 3903 struct list_head stat_list; 3904 struct f2fs_sb_info *sbi; 3905 int all_area_segs, sit_area_segs, nat_area_segs, ssa_area_segs; 3906 int main_area_segs, main_area_sections, main_area_zones; 3907 unsigned long long hit_cached[NR_EXTENT_CACHES]; 3908 unsigned long long hit_rbtree[NR_EXTENT_CACHES]; 3909 unsigned long long total_ext[NR_EXTENT_CACHES]; 3910 unsigned long long hit_total[NR_EXTENT_CACHES]; 3911 int ext_tree[NR_EXTENT_CACHES]; 3912 int zombie_tree[NR_EXTENT_CACHES]; 3913 int ext_node[NR_EXTENT_CACHES]; 3914 /* to count memory footprint */ 3915 unsigned long long ext_mem[NR_EXTENT_CACHES]; 3916 /* for read extent cache */ 3917 unsigned long long hit_largest; 3918 /* for block age extent cache */ 3919 unsigned long long allocated_data_blocks; 3920 int ndirty_node, ndirty_dent, ndirty_meta, ndirty_imeta; 3921 int ndirty_data, ndirty_qdata; 3922 unsigned int ndirty_dirs, ndirty_files, nquota_files, ndirty_all; 3923 int nats, dirty_nats, sits, dirty_sits; 3924 int free_nids, avail_nids, alloc_nids; 3925 int total_count, utilization; 3926 int nr_wb_cp_data, nr_wb_data; 3927 int nr_rd_data, nr_rd_node, nr_rd_meta; 3928 int nr_dio_read, nr_dio_write; 3929 unsigned int io_skip_bggc, other_skip_bggc; 3930 int nr_flushing, nr_flushed, flush_list_empty; 3931 int nr_discarding, nr_discarded; 3932 int nr_discard_cmd; 3933 unsigned int undiscard_blks; 3934 int nr_issued_ckpt, nr_total_ckpt, nr_queued_ckpt; 3935 unsigned int cur_ckpt_time, peak_ckpt_time; 3936 int inline_xattr, inline_inode, inline_dir, append, update, orphans; 3937 int compr_inode, swapfile_inode; 3938 unsigned long long compr_blocks; 3939 int aw_cnt, max_aw_cnt; 3940 unsigned int valid_count, valid_node_count, valid_inode_count, discard_blks; 3941 unsigned int bimodal, avg_vblocks; 3942 int util_free, util_valid, util_invalid; 3943 int rsvd_segs, overp_segs; 3944 int dirty_count, node_pages, meta_pages, compress_pages; 3945 int compress_page_hit; 3946 int prefree_count, free_segs, free_secs; 3947 int cp_call_count[MAX_CALL_TYPE], cp_count; 3948 int gc_call_count[MAX_CALL_TYPE]; 3949 int gc_segs[2][2]; 3950 int gc_secs[2][2]; 3951 int tot_blks, data_blks, node_blks; 3952 int bg_data_blks, bg_node_blks; 3953 int curseg[NR_CURSEG_TYPE]; 3954 int cursec[NR_CURSEG_TYPE]; 3955 int curzone[NR_CURSEG_TYPE]; 3956 unsigned int dirty_seg[NR_CURSEG_TYPE]; 3957 unsigned int full_seg[NR_CURSEG_TYPE]; 3958 unsigned int valid_blks[NR_CURSEG_TYPE]; 3959 3960 unsigned int meta_count[META_MAX]; 3961 unsigned int segment_count[2]; 3962 unsigned int block_count[2]; 3963 unsigned int inplace_count; 3964 unsigned long long base_mem, cache_mem, page_mem; 3965 }; 3966 3967 static inline struct f2fs_stat_info *F2FS_STAT(struct f2fs_sb_info *sbi) 3968 { 3969 return (struct f2fs_stat_info *)sbi->stat_info; 3970 } 3971 3972 #define stat_inc_cp_call_count(sbi, foreground) \ 3973 atomic_inc(&sbi->cp_call_count[(foreground)]) 3974 #define stat_inc_cp_count(si) (F2FS_STAT(sbi)->cp_count++) 3975 #define stat_io_skip_bggc_count(sbi) ((sbi)->io_skip_bggc++) 3976 #define stat_other_skip_bggc_count(sbi) ((sbi)->other_skip_bggc++) 3977 #define stat_inc_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]++) 3978 #define stat_dec_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]--) 3979 #define stat_inc_total_hit(sbi, type) (atomic64_inc(&(sbi)->total_hit_ext[type])) 3980 #define stat_inc_rbtree_node_hit(sbi, type) (atomic64_inc(&(sbi)->read_hit_rbtree[type])) 3981 #define stat_inc_largest_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_largest)) 3982 #define stat_inc_cached_node_hit(sbi, type) (atomic64_inc(&(sbi)->read_hit_cached[type])) 3983 #define stat_inc_inline_xattr(inode) \ 3984 do { \ 3985 if (f2fs_has_inline_xattr(inode)) \ 3986 (atomic_inc(&F2FS_I_SB(inode)->inline_xattr)); \ 3987 } while (0) 3988 #define stat_dec_inline_xattr(inode) \ 3989 do { \ 3990 if (f2fs_has_inline_xattr(inode)) \ 3991 (atomic_dec(&F2FS_I_SB(inode)->inline_xattr)); \ 3992 } while (0) 3993 #define stat_inc_inline_inode(inode) \ 3994 do { \ 3995 if (f2fs_has_inline_data(inode)) \ 3996 (atomic_inc(&F2FS_I_SB(inode)->inline_inode)); \ 3997 } while (0) 3998 #define stat_dec_inline_inode(inode) \ 3999 do { \ 4000 if (f2fs_has_inline_data(inode)) \ 4001 (atomic_dec(&F2FS_I_SB(inode)->inline_inode)); \ 4002 } while (0) 4003 #define stat_inc_inline_dir(inode) \ 4004 do { \ 4005 if (f2fs_has_inline_dentry(inode)) \ 4006 (atomic_inc(&F2FS_I_SB(inode)->inline_dir)); \ 4007 } while (0) 4008 #define stat_dec_inline_dir(inode) \ 4009 do { \ 4010 if (f2fs_has_inline_dentry(inode)) \ 4011 (atomic_dec(&F2FS_I_SB(inode)->inline_dir)); \ 4012 } while (0) 4013 #define stat_inc_compr_inode(inode) \ 4014 do { \ 4015 if (f2fs_compressed_file(inode)) \ 4016 (atomic_inc(&F2FS_I_SB(inode)->compr_inode)); \ 4017 } while (0) 4018 #define stat_dec_compr_inode(inode) \ 4019 do { \ 4020 if (f2fs_compressed_file(inode)) \ 4021 (atomic_dec(&F2FS_I_SB(inode)->compr_inode)); \ 4022 } while (0) 4023 #define stat_add_compr_blocks(inode, blocks) \ 4024 (atomic64_add(blocks, &F2FS_I_SB(inode)->compr_blocks)) 4025 #define stat_sub_compr_blocks(inode, blocks) \ 4026 (atomic64_sub(blocks, &F2FS_I_SB(inode)->compr_blocks)) 4027 #define stat_inc_swapfile_inode(inode) \ 4028 (atomic_inc(&F2FS_I_SB(inode)->swapfile_inode)) 4029 #define stat_dec_swapfile_inode(inode) \ 4030 (atomic_dec(&F2FS_I_SB(inode)->swapfile_inode)) 4031 #define stat_inc_atomic_inode(inode) \ 4032 (atomic_inc(&F2FS_I_SB(inode)->atomic_files)) 4033 #define stat_dec_atomic_inode(inode) \ 4034 (atomic_dec(&F2FS_I_SB(inode)->atomic_files)) 4035 #define stat_inc_meta_count(sbi, blkaddr) \ 4036 do { \ 4037 if (blkaddr < SIT_I(sbi)->sit_base_addr) \ 4038 atomic_inc(&(sbi)->meta_count[META_CP]); \ 4039 else if (blkaddr < NM_I(sbi)->nat_blkaddr) \ 4040 atomic_inc(&(sbi)->meta_count[META_SIT]); \ 4041 else if (blkaddr < SM_I(sbi)->ssa_blkaddr) \ 4042 atomic_inc(&(sbi)->meta_count[META_NAT]); \ 4043 else if (blkaddr < SM_I(sbi)->main_blkaddr) \ 4044 atomic_inc(&(sbi)->meta_count[META_SSA]); \ 4045 } while (0) 4046 #define stat_inc_seg_type(sbi, curseg) \ 4047 ((sbi)->segment_count[(curseg)->alloc_type]++) 4048 #define stat_inc_block_count(sbi, curseg) \ 4049 ((sbi)->block_count[(curseg)->alloc_type]++) 4050 #define stat_inc_inplace_blocks(sbi) \ 4051 (atomic_inc(&(sbi)->inplace_count)) 4052 #define stat_update_max_atomic_write(inode) \ 4053 do { \ 4054 int cur = atomic_read(&F2FS_I_SB(inode)->atomic_files); \ 4055 int max = atomic_read(&F2FS_I_SB(inode)->max_aw_cnt); \ 4056 if (cur > max) \ 4057 atomic_set(&F2FS_I_SB(inode)->max_aw_cnt, cur); \ 4058 } while (0) 4059 #define stat_inc_gc_call_count(sbi, foreground) \ 4060 (F2FS_STAT(sbi)->gc_call_count[(foreground)]++) 4061 #define stat_inc_gc_sec_count(sbi, type, gc_type) \ 4062 (F2FS_STAT(sbi)->gc_secs[(type)][(gc_type)]++) 4063 #define stat_inc_gc_seg_count(sbi, type, gc_type) \ 4064 (F2FS_STAT(sbi)->gc_segs[(type)][(gc_type)]++) 4065 4066 #define stat_inc_tot_blk_count(si, blks) \ 4067 ((si)->tot_blks += (blks)) 4068 4069 #define stat_inc_data_blk_count(sbi, blks, gc_type) \ 4070 do { \ 4071 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 4072 stat_inc_tot_blk_count(si, blks); \ 4073 si->data_blks += (blks); \ 4074 si->bg_data_blks += ((gc_type) == BG_GC) ? (blks) : 0; \ 4075 } while (0) 4076 4077 #define stat_inc_node_blk_count(sbi, blks, gc_type) \ 4078 do { \ 4079 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 4080 stat_inc_tot_blk_count(si, blks); \ 4081 si->node_blks += (blks); \ 4082 si->bg_node_blks += ((gc_type) == BG_GC) ? (blks) : 0; \ 4083 } while (0) 4084 4085 int f2fs_build_stats(struct f2fs_sb_info *sbi); 4086 void f2fs_destroy_stats(struct f2fs_sb_info *sbi); 4087 void __init f2fs_create_root_stats(void); 4088 void f2fs_destroy_root_stats(void); 4089 void f2fs_update_sit_info(struct f2fs_sb_info *sbi); 4090 #else 4091 #define stat_inc_cp_call_count(sbi, foreground) do { } while (0) 4092 #define stat_inc_cp_count(sbi) do { } while (0) 4093 #define stat_io_skip_bggc_count(sbi) do { } while (0) 4094 #define stat_other_skip_bggc_count(sbi) do { } while (0) 4095 #define stat_inc_dirty_inode(sbi, type) do { } while (0) 4096 #define stat_dec_dirty_inode(sbi, type) do { } while (0) 4097 #define stat_inc_total_hit(sbi, type) do { } while (0) 4098 #define stat_inc_rbtree_node_hit(sbi, type) do { } while (0) 4099 #define stat_inc_largest_node_hit(sbi) do { } while (0) 4100 #define stat_inc_cached_node_hit(sbi, type) do { } while (0) 4101 #define stat_inc_inline_xattr(inode) do { } while (0) 4102 #define stat_dec_inline_xattr(inode) do { } while (0) 4103 #define stat_inc_inline_inode(inode) do { } while (0) 4104 #define stat_dec_inline_inode(inode) do { } while (0) 4105 #define stat_inc_inline_dir(inode) do { } while (0) 4106 #define stat_dec_inline_dir(inode) do { } while (0) 4107 #define stat_inc_compr_inode(inode) do { } while (0) 4108 #define stat_dec_compr_inode(inode) do { } while (0) 4109 #define stat_add_compr_blocks(inode, blocks) do { } while (0) 4110 #define stat_sub_compr_blocks(inode, blocks) do { } while (0) 4111 #define stat_inc_swapfile_inode(inode) do { } while (0) 4112 #define stat_dec_swapfile_inode(inode) do { } while (0) 4113 #define stat_inc_atomic_inode(inode) do { } while (0) 4114 #define stat_dec_atomic_inode(inode) do { } while (0) 4115 #define stat_update_max_atomic_write(inode) do { } while (0) 4116 #define stat_inc_meta_count(sbi, blkaddr) do { } while (0) 4117 #define stat_inc_seg_type(sbi, curseg) do { } while (0) 4118 #define stat_inc_block_count(sbi, curseg) do { } while (0) 4119 #define stat_inc_inplace_blocks(sbi) do { } while (0) 4120 #define stat_inc_gc_call_count(sbi, foreground) do { } while (0) 4121 #define stat_inc_gc_sec_count(sbi, type, gc_type) do { } while (0) 4122 #define stat_inc_gc_seg_count(sbi, type, gc_type) do { } while (0) 4123 #define stat_inc_tot_blk_count(si, blks) do { } while (0) 4124 #define stat_inc_data_blk_count(sbi, blks, gc_type) do { } while (0) 4125 #define stat_inc_node_blk_count(sbi, blks, gc_type) do { } while (0) 4126 4127 static inline int f2fs_build_stats(struct f2fs_sb_info *sbi) { return 0; } 4128 static inline void f2fs_destroy_stats(struct f2fs_sb_info *sbi) { } 4129 static inline void __init f2fs_create_root_stats(void) { } 4130 static inline void f2fs_destroy_root_stats(void) { } 4131 static inline void f2fs_update_sit_info(struct f2fs_sb_info *sbi) {} 4132 #endif 4133 4134 extern const struct file_operations f2fs_dir_operations; 4135 extern const struct file_operations f2fs_file_operations; 4136 extern const struct inode_operations f2fs_file_inode_operations; 4137 extern const struct address_space_operations f2fs_dblock_aops; 4138 extern const struct address_space_operations f2fs_node_aops; 4139 extern const struct address_space_operations f2fs_meta_aops; 4140 extern const struct inode_operations f2fs_dir_inode_operations; 4141 extern const struct inode_operations f2fs_symlink_inode_operations; 4142 extern const struct inode_operations f2fs_encrypted_symlink_inode_operations; 4143 extern const struct inode_operations f2fs_special_inode_operations; 4144 extern struct kmem_cache *f2fs_inode_entry_slab; 4145 4146 /* 4147 * inline.c 4148 */ 4149 bool f2fs_may_inline_data(struct inode *inode); 4150 bool f2fs_sanity_check_inline_data(struct inode *inode); 4151 bool f2fs_may_inline_dentry(struct inode *inode); 4152 void f2fs_do_read_inline_data(struct page *page, struct page *ipage); 4153 void f2fs_truncate_inline_inode(struct inode *inode, 4154 struct page *ipage, u64 from); 4155 int f2fs_read_inline_data(struct inode *inode, struct page *page); 4156 int f2fs_convert_inline_page(struct dnode_of_data *dn, struct page *page); 4157 int f2fs_convert_inline_inode(struct inode *inode); 4158 int f2fs_try_convert_inline_dir(struct inode *dir, struct dentry *dentry); 4159 int f2fs_write_inline_data(struct inode *inode, struct page *page); 4160 int f2fs_recover_inline_data(struct inode *inode, struct page *npage); 4161 struct f2fs_dir_entry *f2fs_find_in_inline_dir(struct inode *dir, 4162 const struct f2fs_filename *fname, 4163 struct page **res_page); 4164 int f2fs_make_empty_inline_dir(struct inode *inode, struct inode *parent, 4165 struct page *ipage); 4166 int f2fs_add_inline_entry(struct inode *dir, const struct f2fs_filename *fname, 4167 struct inode *inode, nid_t ino, umode_t mode); 4168 void f2fs_delete_inline_entry(struct f2fs_dir_entry *dentry, 4169 struct page *page, struct inode *dir, 4170 struct inode *inode); 4171 bool f2fs_empty_inline_dir(struct inode *dir); 4172 int f2fs_read_inline_dir(struct file *file, struct dir_context *ctx, 4173 struct fscrypt_str *fstr); 4174 int f2fs_inline_data_fiemap(struct inode *inode, 4175 struct fiemap_extent_info *fieinfo, 4176 __u64 start, __u64 len); 4177 4178 /* 4179 * shrinker.c 4180 */ 4181 unsigned long f2fs_shrink_count(struct shrinker *shrink, 4182 struct shrink_control *sc); 4183 unsigned long f2fs_shrink_scan(struct shrinker *shrink, 4184 struct shrink_control *sc); 4185 void f2fs_join_shrinker(struct f2fs_sb_info *sbi); 4186 void f2fs_leave_shrinker(struct f2fs_sb_info *sbi); 4187 4188 /* 4189 * extent_cache.c 4190 */ 4191 bool sanity_check_extent_cache(struct inode *inode); 4192 void f2fs_init_extent_tree(struct inode *inode); 4193 void f2fs_drop_extent_tree(struct inode *inode); 4194 void f2fs_destroy_extent_node(struct inode *inode); 4195 void f2fs_destroy_extent_tree(struct inode *inode); 4196 void f2fs_init_extent_cache_info(struct f2fs_sb_info *sbi); 4197 int __init f2fs_create_extent_cache(void); 4198 void f2fs_destroy_extent_cache(void); 4199 4200 /* read extent cache ops */ 4201 void f2fs_init_read_extent_tree(struct inode *inode, struct page *ipage); 4202 bool f2fs_lookup_read_extent_cache(struct inode *inode, pgoff_t pgofs, 4203 struct extent_info *ei); 4204 bool f2fs_lookup_read_extent_cache_block(struct inode *inode, pgoff_t index, 4205 block_t *blkaddr); 4206 void f2fs_update_read_extent_cache(struct dnode_of_data *dn); 4207 void f2fs_update_read_extent_cache_range(struct dnode_of_data *dn, 4208 pgoff_t fofs, block_t blkaddr, unsigned int len); 4209 unsigned int f2fs_shrink_read_extent_tree(struct f2fs_sb_info *sbi, 4210 int nr_shrink); 4211 4212 /* block age extent cache ops */ 4213 void f2fs_init_age_extent_tree(struct inode *inode); 4214 bool f2fs_lookup_age_extent_cache(struct inode *inode, pgoff_t pgofs, 4215 struct extent_info *ei); 4216 void f2fs_update_age_extent_cache(struct dnode_of_data *dn); 4217 void f2fs_update_age_extent_cache_range(struct dnode_of_data *dn, 4218 pgoff_t fofs, unsigned int len); 4219 unsigned int f2fs_shrink_age_extent_tree(struct f2fs_sb_info *sbi, 4220 int nr_shrink); 4221 4222 /* 4223 * sysfs.c 4224 */ 4225 #define MIN_RA_MUL 2 4226 #define MAX_RA_MUL 256 4227 4228 int __init f2fs_init_sysfs(void); 4229 void f2fs_exit_sysfs(void); 4230 int f2fs_register_sysfs(struct f2fs_sb_info *sbi); 4231 void f2fs_unregister_sysfs(struct f2fs_sb_info *sbi); 4232 4233 /* verity.c */ 4234 extern const struct fsverity_operations f2fs_verityops; 4235 4236 /* 4237 * crypto support 4238 */ 4239 static inline bool f2fs_encrypted_file(struct inode *inode) 4240 { 4241 return IS_ENCRYPTED(inode) && S_ISREG(inode->i_mode); 4242 } 4243 4244 static inline void f2fs_set_encrypted_inode(struct inode *inode) 4245 { 4246 #ifdef CONFIG_FS_ENCRYPTION 4247 file_set_encrypt(inode); 4248 f2fs_set_inode_flags(inode); 4249 #endif 4250 } 4251 4252 /* 4253 * Returns true if the reads of the inode's data need to undergo some 4254 * postprocessing step, like decryption or authenticity verification. 4255 */ 4256 static inline bool f2fs_post_read_required(struct inode *inode) 4257 { 4258 return f2fs_encrypted_file(inode) || fsverity_active(inode) || 4259 f2fs_compressed_file(inode); 4260 } 4261 4262 static inline bool f2fs_used_in_atomic_write(struct inode *inode) 4263 { 4264 return f2fs_is_atomic_file(inode) || f2fs_is_cow_file(inode); 4265 } 4266 4267 static inline bool f2fs_meta_inode_gc_required(struct inode *inode) 4268 { 4269 return f2fs_post_read_required(inode) || f2fs_used_in_atomic_write(inode); 4270 } 4271 4272 /* 4273 * compress.c 4274 */ 4275 #ifdef CONFIG_F2FS_FS_COMPRESSION 4276 bool f2fs_is_compressed_page(struct page *page); 4277 struct page *f2fs_compress_control_page(struct page *page); 4278 int f2fs_prepare_compress_overwrite(struct inode *inode, 4279 struct page **pagep, pgoff_t index, void **fsdata); 4280 bool f2fs_compress_write_end(struct inode *inode, void *fsdata, 4281 pgoff_t index, unsigned copied); 4282 int f2fs_truncate_partial_cluster(struct inode *inode, u64 from, bool lock); 4283 void f2fs_compress_write_end_io(struct bio *bio, struct page *page); 4284 bool f2fs_is_compress_backend_ready(struct inode *inode); 4285 bool f2fs_is_compress_level_valid(int alg, int lvl); 4286 int __init f2fs_init_compress_mempool(void); 4287 void f2fs_destroy_compress_mempool(void); 4288 void f2fs_decompress_cluster(struct decompress_io_ctx *dic, bool in_task); 4289 void f2fs_end_read_compressed_page(struct page *page, bool failed, 4290 block_t blkaddr, bool in_task); 4291 bool f2fs_cluster_is_empty(struct compress_ctx *cc); 4292 bool f2fs_cluster_can_merge_page(struct compress_ctx *cc, pgoff_t index); 4293 bool f2fs_all_cluster_page_ready(struct compress_ctx *cc, struct page **pages, 4294 int index, int nr_pages, bool uptodate); 4295 bool f2fs_sanity_check_cluster(struct dnode_of_data *dn); 4296 void f2fs_compress_ctx_add_page(struct compress_ctx *cc, struct page *page); 4297 int f2fs_write_multi_pages(struct compress_ctx *cc, 4298 int *submitted, 4299 struct writeback_control *wbc, 4300 enum iostat_type io_type); 4301 int f2fs_is_compressed_cluster(struct inode *inode, pgoff_t index); 4302 void f2fs_update_read_extent_tree_range_compressed(struct inode *inode, 4303 pgoff_t fofs, block_t blkaddr, 4304 unsigned int llen, unsigned int c_len); 4305 int f2fs_read_multi_pages(struct compress_ctx *cc, struct bio **bio_ret, 4306 unsigned nr_pages, sector_t *last_block_in_bio, 4307 bool is_readahead, bool for_write); 4308 struct decompress_io_ctx *f2fs_alloc_dic(struct compress_ctx *cc); 4309 void f2fs_decompress_end_io(struct decompress_io_ctx *dic, bool failed, 4310 bool in_task); 4311 void f2fs_put_page_dic(struct page *page, bool in_task); 4312 unsigned int f2fs_cluster_blocks_are_contiguous(struct dnode_of_data *dn, 4313 unsigned int ofs_in_node); 4314 int f2fs_init_compress_ctx(struct compress_ctx *cc); 4315 void f2fs_destroy_compress_ctx(struct compress_ctx *cc, bool reuse); 4316 void f2fs_init_compress_info(struct f2fs_sb_info *sbi); 4317 int f2fs_init_compress_inode(struct f2fs_sb_info *sbi); 4318 void f2fs_destroy_compress_inode(struct f2fs_sb_info *sbi); 4319 int f2fs_init_page_array_cache(struct f2fs_sb_info *sbi); 4320 void f2fs_destroy_page_array_cache(struct f2fs_sb_info *sbi); 4321 int __init f2fs_init_compress_cache(void); 4322 void f2fs_destroy_compress_cache(void); 4323 struct address_space *COMPRESS_MAPPING(struct f2fs_sb_info *sbi); 4324 void f2fs_invalidate_compress_page(struct f2fs_sb_info *sbi, block_t blkaddr); 4325 void f2fs_cache_compressed_page(struct f2fs_sb_info *sbi, struct page *page, 4326 nid_t ino, block_t blkaddr); 4327 bool f2fs_load_compressed_page(struct f2fs_sb_info *sbi, struct page *page, 4328 block_t blkaddr); 4329 void f2fs_invalidate_compress_pages(struct f2fs_sb_info *sbi, nid_t ino); 4330 #define inc_compr_inode_stat(inode) \ 4331 do { \ 4332 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); \ 4333 sbi->compr_new_inode++; \ 4334 } while (0) 4335 #define add_compr_block_stat(inode, blocks) \ 4336 do { \ 4337 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); \ 4338 int diff = F2FS_I(inode)->i_cluster_size - blocks; \ 4339 sbi->compr_written_block += blocks; \ 4340 sbi->compr_saved_block += diff; \ 4341 } while (0) 4342 #else 4343 static inline bool f2fs_is_compressed_page(struct page *page) { return false; } 4344 static inline bool f2fs_is_compress_backend_ready(struct inode *inode) 4345 { 4346 if (!f2fs_compressed_file(inode)) 4347 return true; 4348 /* not support compression */ 4349 return false; 4350 } 4351 static inline bool f2fs_is_compress_level_valid(int alg, int lvl) { return false; } 4352 static inline struct page *f2fs_compress_control_page(struct page *page) 4353 { 4354 WARN_ON_ONCE(1); 4355 return ERR_PTR(-EINVAL); 4356 } 4357 static inline int __init f2fs_init_compress_mempool(void) { return 0; } 4358 static inline void f2fs_destroy_compress_mempool(void) { } 4359 static inline void f2fs_decompress_cluster(struct decompress_io_ctx *dic, 4360 bool in_task) { } 4361 static inline void f2fs_end_read_compressed_page(struct page *page, 4362 bool failed, block_t blkaddr, bool in_task) 4363 { 4364 WARN_ON_ONCE(1); 4365 } 4366 static inline void f2fs_put_page_dic(struct page *page, bool in_task) 4367 { 4368 WARN_ON_ONCE(1); 4369 } 4370 static inline unsigned int f2fs_cluster_blocks_are_contiguous( 4371 struct dnode_of_data *dn, unsigned int ofs_in_node) { return 0; } 4372 static inline bool f2fs_sanity_check_cluster(struct dnode_of_data *dn) { return false; } 4373 static inline int f2fs_init_compress_inode(struct f2fs_sb_info *sbi) { return 0; } 4374 static inline void f2fs_destroy_compress_inode(struct f2fs_sb_info *sbi) { } 4375 static inline int f2fs_init_page_array_cache(struct f2fs_sb_info *sbi) { return 0; } 4376 static inline void f2fs_destroy_page_array_cache(struct f2fs_sb_info *sbi) { } 4377 static inline int __init f2fs_init_compress_cache(void) { return 0; } 4378 static inline void f2fs_destroy_compress_cache(void) { } 4379 static inline void f2fs_invalidate_compress_page(struct f2fs_sb_info *sbi, 4380 block_t blkaddr) { } 4381 static inline void f2fs_cache_compressed_page(struct f2fs_sb_info *sbi, 4382 struct page *page, nid_t ino, block_t blkaddr) { } 4383 static inline bool f2fs_load_compressed_page(struct f2fs_sb_info *sbi, 4384 struct page *page, block_t blkaddr) { return false; } 4385 static inline void f2fs_invalidate_compress_pages(struct f2fs_sb_info *sbi, 4386 nid_t ino) { } 4387 #define inc_compr_inode_stat(inode) do { } while (0) 4388 static inline void f2fs_update_read_extent_tree_range_compressed( 4389 struct inode *inode, 4390 pgoff_t fofs, block_t blkaddr, 4391 unsigned int llen, unsigned int c_len) { } 4392 #endif 4393 4394 static inline int set_compress_context(struct inode *inode) 4395 { 4396 #ifdef CONFIG_F2FS_FS_COMPRESSION 4397 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 4398 4399 F2FS_I(inode)->i_compress_algorithm = 4400 F2FS_OPTION(sbi).compress_algorithm; 4401 F2FS_I(inode)->i_log_cluster_size = 4402 F2FS_OPTION(sbi).compress_log_size; 4403 F2FS_I(inode)->i_compress_flag = 4404 F2FS_OPTION(sbi).compress_chksum ? 4405 BIT(COMPRESS_CHKSUM) : 0; 4406 F2FS_I(inode)->i_cluster_size = 4407 BIT(F2FS_I(inode)->i_log_cluster_size); 4408 if ((F2FS_I(inode)->i_compress_algorithm == COMPRESS_LZ4 || 4409 F2FS_I(inode)->i_compress_algorithm == COMPRESS_ZSTD) && 4410 F2FS_OPTION(sbi).compress_level) 4411 F2FS_I(inode)->i_compress_level = 4412 F2FS_OPTION(sbi).compress_level; 4413 F2FS_I(inode)->i_flags |= F2FS_COMPR_FL; 4414 set_inode_flag(inode, FI_COMPRESSED_FILE); 4415 stat_inc_compr_inode(inode); 4416 inc_compr_inode_stat(inode); 4417 f2fs_mark_inode_dirty_sync(inode, true); 4418 return 0; 4419 #else 4420 return -EOPNOTSUPP; 4421 #endif 4422 } 4423 4424 static inline bool f2fs_disable_compressed_file(struct inode *inode) 4425 { 4426 struct f2fs_inode_info *fi = F2FS_I(inode); 4427 4428 f2fs_down_write(&F2FS_I(inode)->i_sem); 4429 4430 if (!f2fs_compressed_file(inode)) { 4431 f2fs_up_write(&F2FS_I(inode)->i_sem); 4432 return true; 4433 } 4434 if (f2fs_is_mmap_file(inode) || 4435 (S_ISREG(inode->i_mode) && F2FS_HAS_BLOCKS(inode))) { 4436 f2fs_up_write(&F2FS_I(inode)->i_sem); 4437 return false; 4438 } 4439 4440 fi->i_flags &= ~F2FS_COMPR_FL; 4441 stat_dec_compr_inode(inode); 4442 clear_inode_flag(inode, FI_COMPRESSED_FILE); 4443 f2fs_mark_inode_dirty_sync(inode, true); 4444 4445 f2fs_up_write(&F2FS_I(inode)->i_sem); 4446 return true; 4447 } 4448 4449 #define F2FS_FEATURE_FUNCS(name, flagname) \ 4450 static inline bool f2fs_sb_has_##name(struct f2fs_sb_info *sbi) \ 4451 { \ 4452 return F2FS_HAS_FEATURE(sbi, F2FS_FEATURE_##flagname); \ 4453 } 4454 4455 F2FS_FEATURE_FUNCS(encrypt, ENCRYPT); 4456 F2FS_FEATURE_FUNCS(blkzoned, BLKZONED); 4457 F2FS_FEATURE_FUNCS(extra_attr, EXTRA_ATTR); 4458 F2FS_FEATURE_FUNCS(project_quota, PRJQUOTA); 4459 F2FS_FEATURE_FUNCS(inode_chksum, INODE_CHKSUM); 4460 F2FS_FEATURE_FUNCS(flexible_inline_xattr, FLEXIBLE_INLINE_XATTR); 4461 F2FS_FEATURE_FUNCS(quota_ino, QUOTA_INO); 4462 F2FS_FEATURE_FUNCS(inode_crtime, INODE_CRTIME); 4463 F2FS_FEATURE_FUNCS(lost_found, LOST_FOUND); 4464 F2FS_FEATURE_FUNCS(verity, VERITY); 4465 F2FS_FEATURE_FUNCS(sb_chksum, SB_CHKSUM); 4466 F2FS_FEATURE_FUNCS(casefold, CASEFOLD); 4467 F2FS_FEATURE_FUNCS(compression, COMPRESSION); 4468 F2FS_FEATURE_FUNCS(readonly, RO); 4469 4470 #ifdef CONFIG_BLK_DEV_ZONED 4471 static inline bool f2fs_blkz_is_seq(struct f2fs_sb_info *sbi, int devi, 4472 block_t blkaddr) 4473 { 4474 unsigned int zno = blkaddr / sbi->blocks_per_blkz; 4475 4476 return test_bit(zno, FDEV(devi).blkz_seq); 4477 } 4478 #endif 4479 4480 static inline int f2fs_bdev_index(struct f2fs_sb_info *sbi, 4481 struct block_device *bdev) 4482 { 4483 int i; 4484 4485 if (!f2fs_is_multi_device(sbi)) 4486 return 0; 4487 4488 for (i = 0; i < sbi->s_ndevs; i++) 4489 if (FDEV(i).bdev == bdev) 4490 return i; 4491 4492 WARN_ON(1); 4493 return -1; 4494 } 4495 4496 static inline bool f2fs_hw_should_discard(struct f2fs_sb_info *sbi) 4497 { 4498 return f2fs_sb_has_blkzoned(sbi); 4499 } 4500 4501 static inline bool f2fs_bdev_support_discard(struct block_device *bdev) 4502 { 4503 return bdev_max_discard_sectors(bdev) || bdev_is_zoned(bdev); 4504 } 4505 4506 static inline bool f2fs_hw_support_discard(struct f2fs_sb_info *sbi) 4507 { 4508 int i; 4509 4510 if (!f2fs_is_multi_device(sbi)) 4511 return f2fs_bdev_support_discard(sbi->sb->s_bdev); 4512 4513 for (i = 0; i < sbi->s_ndevs; i++) 4514 if (f2fs_bdev_support_discard(FDEV(i).bdev)) 4515 return true; 4516 return false; 4517 } 4518 4519 static inline bool f2fs_realtime_discard_enable(struct f2fs_sb_info *sbi) 4520 { 4521 return (test_opt(sbi, DISCARD) && f2fs_hw_support_discard(sbi)) || 4522 f2fs_hw_should_discard(sbi); 4523 } 4524 4525 static inline bool f2fs_hw_is_readonly(struct f2fs_sb_info *sbi) 4526 { 4527 int i; 4528 4529 if (!f2fs_is_multi_device(sbi)) 4530 return bdev_read_only(sbi->sb->s_bdev); 4531 4532 for (i = 0; i < sbi->s_ndevs; i++) 4533 if (bdev_read_only(FDEV(i).bdev)) 4534 return true; 4535 return false; 4536 } 4537 4538 static inline bool f2fs_dev_is_readonly(struct f2fs_sb_info *sbi) 4539 { 4540 return f2fs_sb_has_readonly(sbi) || f2fs_hw_is_readonly(sbi); 4541 } 4542 4543 static inline bool f2fs_lfs_mode(struct f2fs_sb_info *sbi) 4544 { 4545 return F2FS_OPTION(sbi).fs_mode == FS_MODE_LFS; 4546 } 4547 4548 static inline bool f2fs_valid_pinned_area(struct f2fs_sb_info *sbi, 4549 block_t blkaddr) 4550 { 4551 if (f2fs_sb_has_blkzoned(sbi)) { 4552 int devi = f2fs_target_device_index(sbi, blkaddr); 4553 4554 return !bdev_is_zoned(FDEV(devi).bdev); 4555 } 4556 return true; 4557 } 4558 4559 static inline bool f2fs_low_mem_mode(struct f2fs_sb_info *sbi) 4560 { 4561 return F2FS_OPTION(sbi).memory_mode == MEMORY_MODE_LOW; 4562 } 4563 4564 static inline bool f2fs_may_compress(struct inode *inode) 4565 { 4566 if (IS_SWAPFILE(inode) || f2fs_is_pinned_file(inode) || 4567 f2fs_is_atomic_file(inode) || f2fs_has_inline_data(inode) || 4568 f2fs_is_mmap_file(inode)) 4569 return false; 4570 return S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode); 4571 } 4572 4573 static inline void f2fs_i_compr_blocks_update(struct inode *inode, 4574 u64 blocks, bool add) 4575 { 4576 struct f2fs_inode_info *fi = F2FS_I(inode); 4577 int diff = fi->i_cluster_size - blocks; 4578 4579 /* don't update i_compr_blocks if saved blocks were released */ 4580 if (!add && !atomic_read(&fi->i_compr_blocks)) 4581 return; 4582 4583 if (add) { 4584 atomic_add(diff, &fi->i_compr_blocks); 4585 stat_add_compr_blocks(inode, diff); 4586 } else { 4587 atomic_sub(diff, &fi->i_compr_blocks); 4588 stat_sub_compr_blocks(inode, diff); 4589 } 4590 f2fs_mark_inode_dirty_sync(inode, true); 4591 } 4592 4593 static inline bool f2fs_allow_multi_device_dio(struct f2fs_sb_info *sbi, 4594 int flag) 4595 { 4596 if (!f2fs_is_multi_device(sbi)) 4597 return false; 4598 if (flag != F2FS_GET_BLOCK_DIO) 4599 return false; 4600 return sbi->aligned_blksize; 4601 } 4602 4603 static inline bool f2fs_need_verity(const struct inode *inode, pgoff_t idx) 4604 { 4605 return fsverity_active(inode) && 4606 idx < DIV_ROUND_UP(inode->i_size, PAGE_SIZE); 4607 } 4608 4609 #ifdef CONFIG_F2FS_FAULT_INJECTION 4610 extern int f2fs_build_fault_attr(struct f2fs_sb_info *sbi, unsigned long rate, 4611 unsigned long type); 4612 #else 4613 static inline int f2fs_build_fault_attr(struct f2fs_sb_info *sbi, 4614 unsigned long rate, unsigned long type) 4615 { 4616 return 0; 4617 } 4618 #endif 4619 4620 static inline bool is_journalled_quota(struct f2fs_sb_info *sbi) 4621 { 4622 #ifdef CONFIG_QUOTA 4623 if (f2fs_sb_has_quota_ino(sbi)) 4624 return true; 4625 if (F2FS_OPTION(sbi).s_qf_names[USRQUOTA] || 4626 F2FS_OPTION(sbi).s_qf_names[GRPQUOTA] || 4627 F2FS_OPTION(sbi).s_qf_names[PRJQUOTA]) 4628 return true; 4629 #endif 4630 return false; 4631 } 4632 4633 static inline bool f2fs_block_unit_discard(struct f2fs_sb_info *sbi) 4634 { 4635 return F2FS_OPTION(sbi).discard_unit == DISCARD_UNIT_BLOCK; 4636 } 4637 4638 static inline void f2fs_io_schedule_timeout(long timeout) 4639 { 4640 set_current_state(TASK_UNINTERRUPTIBLE); 4641 io_schedule_timeout(timeout); 4642 } 4643 4644 static inline void f2fs_handle_page_eio(struct f2fs_sb_info *sbi, pgoff_t ofs, 4645 enum page_type type) 4646 { 4647 if (unlikely(f2fs_cp_error(sbi))) 4648 return; 4649 4650 if (ofs == sbi->page_eio_ofs[type]) { 4651 if (sbi->page_eio_cnt[type]++ == MAX_RETRY_PAGE_EIO) 4652 set_ckpt_flags(sbi, CP_ERROR_FLAG); 4653 } else { 4654 sbi->page_eio_ofs[type] = ofs; 4655 sbi->page_eio_cnt[type] = 0; 4656 } 4657 } 4658 4659 static inline bool f2fs_is_readonly(struct f2fs_sb_info *sbi) 4660 { 4661 return f2fs_sb_has_readonly(sbi) || f2fs_readonly(sbi->sb); 4662 } 4663 4664 static inline void f2fs_truncate_meta_inode_pages(struct f2fs_sb_info *sbi, 4665 block_t blkaddr, unsigned int cnt) 4666 { 4667 bool need_submit = false; 4668 int i = 0; 4669 4670 do { 4671 struct page *page; 4672 4673 page = find_get_page(META_MAPPING(sbi), blkaddr + i); 4674 if (page) { 4675 if (PageWriteback(page)) 4676 need_submit = true; 4677 f2fs_put_page(page, 0); 4678 } 4679 } while (++i < cnt && !need_submit); 4680 4681 if (need_submit) 4682 f2fs_submit_merged_write_cond(sbi, sbi->meta_inode, 4683 NULL, 0, DATA); 4684 4685 truncate_inode_pages_range(META_MAPPING(sbi), 4686 F2FS_BLK_TO_BYTES((loff_t)blkaddr), 4687 F2FS_BLK_END_BYTES((loff_t)(blkaddr + cnt - 1))); 4688 } 4689 4690 static inline void f2fs_invalidate_internal_cache(struct f2fs_sb_info *sbi, 4691 block_t blkaddr) 4692 { 4693 f2fs_truncate_meta_inode_pages(sbi, blkaddr, 1); 4694 f2fs_invalidate_compress_page(sbi, blkaddr); 4695 } 4696 4697 #define EFSBADCRC EBADMSG /* Bad CRC detected */ 4698 #define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */ 4699 4700 #endif /* _LINUX_F2FS_H */ 4701