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_OPENED_FILE, /* indicate file has been opened */ 802 FI_MAX, /* max flag, never be used */ 803 }; 804 805 struct f2fs_inode_info { 806 struct inode vfs_inode; /* serve a vfs inode */ 807 unsigned long i_flags; /* keep an inode flags for ioctl */ 808 unsigned char i_advise; /* use to give file attribute hints */ 809 unsigned char i_dir_level; /* use for dentry level for large dir */ 810 unsigned int i_current_depth; /* only for directory depth */ 811 /* for gc failure statistic */ 812 unsigned int i_gc_failures[MAX_GC_FAILURE]; 813 unsigned int i_pino; /* parent inode number */ 814 umode_t i_acl_mode; /* keep file acl mode temporarily */ 815 816 /* Use below internally in f2fs*/ 817 unsigned long flags[BITS_TO_LONGS(FI_MAX)]; /* use to pass per-file flags */ 818 struct f2fs_rwsem i_sem; /* protect fi info */ 819 atomic_t dirty_pages; /* # of dirty pages */ 820 f2fs_hash_t chash; /* hash value of given file name */ 821 unsigned int clevel; /* maximum level of given file name */ 822 struct task_struct *task; /* lookup and create consistency */ 823 struct task_struct *cp_task; /* separate cp/wb IO stats*/ 824 struct task_struct *wb_task; /* indicate inode is in context of writeback */ 825 nid_t i_xattr_nid; /* node id that contains xattrs */ 826 loff_t last_disk_size; /* lastly written file size */ 827 spinlock_t i_size_lock; /* protect last_disk_size */ 828 829 #ifdef CONFIG_QUOTA 830 struct dquot __rcu *i_dquot[MAXQUOTAS]; 831 832 /* quota space reservation, managed internally by quota code */ 833 qsize_t i_reserved_quota; 834 #endif 835 struct list_head dirty_list; /* dirty list for dirs and files */ 836 struct list_head gdirty_list; /* linked in global dirty list */ 837 struct task_struct *atomic_write_task; /* store atomic write task */ 838 struct extent_tree *extent_tree[NR_EXTENT_CACHES]; 839 /* cached extent_tree entry */ 840 union { 841 struct inode *cow_inode; /* copy-on-write inode for atomic write */ 842 struct inode *atomic_inode; 843 /* point to atomic_inode, available only for cow_inode */ 844 }; 845 846 /* avoid racing between foreground op and gc */ 847 struct f2fs_rwsem i_gc_rwsem[2]; 848 struct f2fs_rwsem i_xattr_sem; /* avoid racing between reading and changing EAs */ 849 850 int i_extra_isize; /* size of extra space located in i_addr */ 851 kprojid_t i_projid; /* id for project quota */ 852 int i_inline_xattr_size; /* inline xattr size */ 853 struct timespec64 i_crtime; /* inode creation time */ 854 struct timespec64 i_disk_time[3];/* inode disk times */ 855 856 /* for file compress */ 857 atomic_t i_compr_blocks; /* # of compressed blocks */ 858 unsigned char i_compress_algorithm; /* algorithm type */ 859 unsigned char i_log_cluster_size; /* log of cluster size */ 860 unsigned char i_compress_level; /* compress level (lz4hc,zstd) */ 861 unsigned char i_compress_flag; /* compress flag */ 862 unsigned int i_cluster_size; /* cluster size */ 863 864 unsigned int atomic_write_cnt; 865 loff_t original_i_size; /* original i_size before atomic write */ 866 }; 867 868 static inline void get_read_extent_info(struct extent_info *ext, 869 struct f2fs_extent *i_ext) 870 { 871 ext->fofs = le32_to_cpu(i_ext->fofs); 872 ext->blk = le32_to_cpu(i_ext->blk); 873 ext->len = le32_to_cpu(i_ext->len); 874 } 875 876 static inline void set_raw_read_extent(struct extent_info *ext, 877 struct f2fs_extent *i_ext) 878 { 879 i_ext->fofs = cpu_to_le32(ext->fofs); 880 i_ext->blk = cpu_to_le32(ext->blk); 881 i_ext->len = cpu_to_le32(ext->len); 882 } 883 884 static inline bool __is_discard_mergeable(struct discard_info *back, 885 struct discard_info *front, unsigned int max_len) 886 { 887 return (back->lstart + back->len == front->lstart) && 888 (back->len + front->len <= max_len); 889 } 890 891 static inline bool __is_discard_back_mergeable(struct discard_info *cur, 892 struct discard_info *back, unsigned int max_len) 893 { 894 return __is_discard_mergeable(back, cur, max_len); 895 } 896 897 static inline bool __is_discard_front_mergeable(struct discard_info *cur, 898 struct discard_info *front, unsigned int max_len) 899 { 900 return __is_discard_mergeable(cur, front, max_len); 901 } 902 903 /* 904 * For free nid management 905 */ 906 enum nid_state { 907 FREE_NID, /* newly added to free nid list */ 908 PREALLOC_NID, /* it is preallocated */ 909 MAX_NID_STATE, 910 }; 911 912 enum nat_state { 913 TOTAL_NAT, 914 DIRTY_NAT, 915 RECLAIMABLE_NAT, 916 MAX_NAT_STATE, 917 }; 918 919 struct f2fs_nm_info { 920 block_t nat_blkaddr; /* base disk address of NAT */ 921 nid_t max_nid; /* maximum possible node ids */ 922 nid_t available_nids; /* # of available node ids */ 923 nid_t next_scan_nid; /* the next nid to be scanned */ 924 nid_t max_rf_node_blocks; /* max # of nodes for recovery */ 925 unsigned int ram_thresh; /* control the memory footprint */ 926 unsigned int ra_nid_pages; /* # of nid pages to be readaheaded */ 927 unsigned int dirty_nats_ratio; /* control dirty nats ratio threshold */ 928 929 /* NAT cache management */ 930 struct radix_tree_root nat_root;/* root of the nat entry cache */ 931 struct radix_tree_root nat_set_root;/* root of the nat set cache */ 932 struct f2fs_rwsem nat_tree_lock; /* protect nat entry tree */ 933 struct list_head nat_entries; /* cached nat entry list (clean) */ 934 spinlock_t nat_list_lock; /* protect clean nat entry list */ 935 unsigned int nat_cnt[MAX_NAT_STATE]; /* the # of cached nat entries */ 936 unsigned int nat_blocks; /* # of nat blocks */ 937 938 /* free node ids management */ 939 struct radix_tree_root free_nid_root;/* root of the free_nid cache */ 940 struct list_head free_nid_list; /* list for free nids excluding preallocated nids */ 941 unsigned int nid_cnt[MAX_NID_STATE]; /* the number of free node id */ 942 spinlock_t nid_list_lock; /* protect nid lists ops */ 943 struct mutex build_lock; /* lock for build free nids */ 944 unsigned char **free_nid_bitmap; 945 unsigned char *nat_block_bitmap; 946 unsigned short *free_nid_count; /* free nid count of NAT block */ 947 948 /* for checkpoint */ 949 char *nat_bitmap; /* NAT bitmap pointer */ 950 951 unsigned int nat_bits_blocks; /* # of nat bits blocks */ 952 unsigned char *nat_bits; /* NAT bits blocks */ 953 unsigned char *full_nat_bits; /* full NAT pages */ 954 unsigned char *empty_nat_bits; /* empty NAT pages */ 955 #ifdef CONFIG_F2FS_CHECK_FS 956 char *nat_bitmap_mir; /* NAT bitmap mirror */ 957 #endif 958 int bitmap_size; /* bitmap size */ 959 }; 960 961 /* 962 * this structure is used as one of function parameters. 963 * all the information are dedicated to a given direct node block determined 964 * by the data offset in a file. 965 */ 966 struct dnode_of_data { 967 struct inode *inode; /* vfs inode pointer */ 968 struct page *inode_page; /* its inode page, NULL is possible */ 969 struct page *node_page; /* cached direct node page */ 970 nid_t nid; /* node id of the direct node block */ 971 unsigned int ofs_in_node; /* data offset in the node page */ 972 bool inode_page_locked; /* inode page is locked or not */ 973 bool node_changed; /* is node block changed */ 974 char cur_level; /* level of hole node page */ 975 char max_level; /* level of current page located */ 976 block_t data_blkaddr; /* block address of the node block */ 977 }; 978 979 static inline void set_new_dnode(struct dnode_of_data *dn, struct inode *inode, 980 struct page *ipage, struct page *npage, nid_t nid) 981 { 982 memset(dn, 0, sizeof(*dn)); 983 dn->inode = inode; 984 dn->inode_page = ipage; 985 dn->node_page = npage; 986 dn->nid = nid; 987 } 988 989 /* 990 * For SIT manager 991 * 992 * By default, there are 6 active log areas across the whole main area. 993 * When considering hot and cold data separation to reduce cleaning overhead, 994 * we split 3 for data logs and 3 for node logs as hot, warm, and cold types, 995 * respectively. 996 * In the current design, you should not change the numbers intentionally. 997 * Instead, as a mount option such as active_logs=x, you can use 2, 4, and 6 998 * logs individually according to the underlying devices. (default: 6) 999 * Just in case, on-disk layout covers maximum 16 logs that consist of 8 for 1000 * data and 8 for node logs. 1001 */ 1002 #define NR_CURSEG_DATA_TYPE (3) 1003 #define NR_CURSEG_NODE_TYPE (3) 1004 #define NR_CURSEG_INMEM_TYPE (2) 1005 #define NR_CURSEG_RO_TYPE (2) 1006 #define NR_CURSEG_PERSIST_TYPE (NR_CURSEG_DATA_TYPE + NR_CURSEG_NODE_TYPE) 1007 #define NR_CURSEG_TYPE (NR_CURSEG_INMEM_TYPE + NR_CURSEG_PERSIST_TYPE) 1008 1009 enum { 1010 CURSEG_HOT_DATA = 0, /* directory entry blocks */ 1011 CURSEG_WARM_DATA, /* data blocks */ 1012 CURSEG_COLD_DATA, /* multimedia or GCed data blocks */ 1013 CURSEG_HOT_NODE, /* direct node blocks of directory files */ 1014 CURSEG_WARM_NODE, /* direct node blocks of normal files */ 1015 CURSEG_COLD_NODE, /* indirect node blocks */ 1016 NR_PERSISTENT_LOG, /* number of persistent log */ 1017 CURSEG_COLD_DATA_PINNED = NR_PERSISTENT_LOG, 1018 /* pinned file that needs consecutive block address */ 1019 CURSEG_ALL_DATA_ATGC, /* SSR alloctor in hot/warm/cold data area */ 1020 NO_CHECK_TYPE, /* number of persistent & inmem log */ 1021 }; 1022 1023 struct flush_cmd { 1024 struct completion wait; 1025 struct llist_node llnode; 1026 nid_t ino; 1027 int ret; 1028 }; 1029 1030 struct flush_cmd_control { 1031 struct task_struct *f2fs_issue_flush; /* flush thread */ 1032 wait_queue_head_t flush_wait_queue; /* waiting queue for wake-up */ 1033 atomic_t issued_flush; /* # of issued flushes */ 1034 atomic_t queued_flush; /* # of queued flushes */ 1035 struct llist_head issue_list; /* list for command issue */ 1036 struct llist_node *dispatch_list; /* list for command dispatch */ 1037 }; 1038 1039 struct f2fs_sm_info { 1040 struct sit_info *sit_info; /* whole segment information */ 1041 struct free_segmap_info *free_info; /* free segment information */ 1042 struct dirty_seglist_info *dirty_info; /* dirty segment information */ 1043 struct curseg_info *curseg_array; /* active segment information */ 1044 1045 struct f2fs_rwsem curseg_lock; /* for preventing curseg change */ 1046 1047 block_t seg0_blkaddr; /* block address of 0'th segment */ 1048 block_t main_blkaddr; /* start block address of main area */ 1049 block_t ssa_blkaddr; /* start block address of SSA area */ 1050 1051 unsigned int segment_count; /* total # of segments */ 1052 unsigned int main_segments; /* # of segments in main area */ 1053 unsigned int reserved_segments; /* # of reserved segments */ 1054 unsigned int additional_reserved_segments;/* reserved segs for IO align feature */ 1055 unsigned int ovp_segments; /* # of overprovision segments */ 1056 1057 /* a threshold to reclaim prefree segments */ 1058 unsigned int rec_prefree_segments; 1059 1060 struct list_head sit_entry_set; /* sit entry set list */ 1061 1062 unsigned int ipu_policy; /* in-place-update policy */ 1063 unsigned int min_ipu_util; /* in-place-update threshold */ 1064 unsigned int min_fsync_blocks; /* threshold for fsync */ 1065 unsigned int min_seq_blocks; /* threshold for sequential blocks */ 1066 unsigned int min_hot_blocks; /* threshold for hot block allocation */ 1067 unsigned int min_ssr_sections; /* threshold to trigger SSR allocation */ 1068 1069 /* for flush command control */ 1070 struct flush_cmd_control *fcc_info; 1071 1072 /* for discard command control */ 1073 struct discard_cmd_control *dcc_info; 1074 }; 1075 1076 /* 1077 * For superblock 1078 */ 1079 /* 1080 * COUNT_TYPE for monitoring 1081 * 1082 * f2fs monitors the number of several block types such as on-writeback, 1083 * dirty dentry blocks, dirty node blocks, and dirty meta blocks. 1084 */ 1085 #define WB_DATA_TYPE(p, f) \ 1086 (f || f2fs_is_cp_guaranteed(p) ? F2FS_WB_CP_DATA : F2FS_WB_DATA) 1087 enum count_type { 1088 F2FS_DIRTY_DENTS, 1089 F2FS_DIRTY_DATA, 1090 F2FS_DIRTY_QDATA, 1091 F2FS_DIRTY_NODES, 1092 F2FS_DIRTY_META, 1093 F2FS_DIRTY_IMETA, 1094 F2FS_WB_CP_DATA, 1095 F2FS_WB_DATA, 1096 F2FS_RD_DATA, 1097 F2FS_RD_NODE, 1098 F2FS_RD_META, 1099 F2FS_DIO_WRITE, 1100 F2FS_DIO_READ, 1101 NR_COUNT_TYPE, 1102 }; 1103 1104 /* 1105 * The below are the page types of bios used in submit_bio(). 1106 * The available types are: 1107 * DATA User data pages. It operates as async mode. 1108 * NODE Node pages. It operates as async mode. 1109 * META FS metadata pages such as SIT, NAT, CP. 1110 * NR_PAGE_TYPE The number of page types. 1111 * META_FLUSH Make sure the previous pages are written 1112 * with waiting the bio's completion 1113 * ... Only can be used with META. 1114 */ 1115 #define PAGE_TYPE_OF_BIO(type) ((type) > META ? META : (type)) 1116 #define PAGE_TYPE_ON_MAIN(type) ((type) == DATA || (type) == NODE) 1117 enum page_type { 1118 DATA = 0, 1119 NODE = 1, /* should not change this */ 1120 META, 1121 NR_PAGE_TYPE, 1122 META_FLUSH, 1123 IPU, /* the below types are used by tracepoints only. */ 1124 OPU, 1125 }; 1126 1127 enum temp_type { 1128 HOT = 0, /* must be zero for meta bio */ 1129 WARM, 1130 COLD, 1131 NR_TEMP_TYPE, 1132 }; 1133 1134 enum need_lock_type { 1135 LOCK_REQ = 0, 1136 LOCK_DONE, 1137 LOCK_RETRY, 1138 }; 1139 1140 enum cp_reason_type { 1141 CP_NO_NEEDED, 1142 CP_NON_REGULAR, 1143 CP_COMPRESSED, 1144 CP_HARDLINK, 1145 CP_SB_NEED_CP, 1146 CP_WRONG_PINO, 1147 CP_NO_SPC_ROLL, 1148 CP_NODE_NEED_CP, 1149 CP_FASTBOOT_MODE, 1150 CP_SPEC_LOG_NUM, 1151 CP_RECOVER_DIR, 1152 }; 1153 1154 enum iostat_type { 1155 /* WRITE IO */ 1156 APP_DIRECT_IO, /* app direct write IOs */ 1157 APP_BUFFERED_IO, /* app buffered write IOs */ 1158 APP_WRITE_IO, /* app write IOs */ 1159 APP_MAPPED_IO, /* app mapped IOs */ 1160 APP_BUFFERED_CDATA_IO, /* app buffered write IOs on compressed file */ 1161 APP_MAPPED_CDATA_IO, /* app mapped write IOs on compressed file */ 1162 FS_DATA_IO, /* data IOs from kworker/fsync/reclaimer */ 1163 FS_CDATA_IO, /* data IOs from kworker/fsync/reclaimer on compressed file */ 1164 FS_NODE_IO, /* node IOs from kworker/fsync/reclaimer */ 1165 FS_META_IO, /* meta IOs from kworker/reclaimer */ 1166 FS_GC_DATA_IO, /* data IOs from forground gc */ 1167 FS_GC_NODE_IO, /* node IOs from forground gc */ 1168 FS_CP_DATA_IO, /* data IOs from checkpoint */ 1169 FS_CP_NODE_IO, /* node IOs from checkpoint */ 1170 FS_CP_META_IO, /* meta IOs from checkpoint */ 1171 1172 /* READ IO */ 1173 APP_DIRECT_READ_IO, /* app direct read IOs */ 1174 APP_BUFFERED_READ_IO, /* app buffered read IOs */ 1175 APP_READ_IO, /* app read IOs */ 1176 APP_MAPPED_READ_IO, /* app mapped read IOs */ 1177 APP_BUFFERED_CDATA_READ_IO, /* app buffered read IOs on compressed file */ 1178 APP_MAPPED_CDATA_READ_IO, /* app mapped read IOs on compressed file */ 1179 FS_DATA_READ_IO, /* data read IOs */ 1180 FS_GDATA_READ_IO, /* data read IOs from background gc */ 1181 FS_CDATA_READ_IO, /* compressed data read IOs */ 1182 FS_NODE_READ_IO, /* node read IOs */ 1183 FS_META_READ_IO, /* meta read IOs */ 1184 1185 /* other */ 1186 FS_DISCARD_IO, /* discard */ 1187 FS_FLUSH_IO, /* flush */ 1188 FS_ZONE_RESET_IO, /* zone reset */ 1189 NR_IO_TYPE, 1190 }; 1191 1192 struct f2fs_io_info { 1193 struct f2fs_sb_info *sbi; /* f2fs_sb_info pointer */ 1194 nid_t ino; /* inode number */ 1195 enum page_type type; /* contains DATA/NODE/META/META_FLUSH */ 1196 enum temp_type temp; /* contains HOT/WARM/COLD */ 1197 enum req_op op; /* contains REQ_OP_ */ 1198 blk_opf_t op_flags; /* req_flag_bits */ 1199 block_t new_blkaddr; /* new block address to be written */ 1200 block_t old_blkaddr; /* old block address before Cow */ 1201 struct page *page; /* page to be written */ 1202 struct page *encrypted_page; /* encrypted page */ 1203 struct page *compressed_page; /* compressed page */ 1204 struct list_head list; /* serialize IOs */ 1205 unsigned int compr_blocks; /* # of compressed block addresses */ 1206 unsigned int need_lock:8; /* indicate we need to lock cp_rwsem */ 1207 unsigned int version:8; /* version of the node */ 1208 unsigned int submitted:1; /* indicate IO submission */ 1209 unsigned int in_list:1; /* indicate fio is in io_list */ 1210 unsigned int is_por:1; /* indicate IO is from recovery or not */ 1211 unsigned int encrypted:1; /* indicate file is encrypted */ 1212 unsigned int meta_gc:1; /* require meta inode GC */ 1213 enum iostat_type io_type; /* io type */ 1214 struct writeback_control *io_wbc; /* writeback control */ 1215 struct bio **bio; /* bio for ipu */ 1216 sector_t *last_block; /* last block number in bio */ 1217 }; 1218 1219 struct bio_entry { 1220 struct bio *bio; 1221 struct list_head list; 1222 }; 1223 1224 #define is_read_io(rw) ((rw) == READ) 1225 struct f2fs_bio_info { 1226 struct f2fs_sb_info *sbi; /* f2fs superblock */ 1227 struct bio *bio; /* bios to merge */ 1228 sector_t last_block_in_bio; /* last block number */ 1229 struct f2fs_io_info fio; /* store buffered io info. */ 1230 #ifdef CONFIG_BLK_DEV_ZONED 1231 struct completion zone_wait; /* condition value for the previous open zone to close */ 1232 struct bio *zone_pending_bio; /* pending bio for the previous zone */ 1233 void *bi_private; /* previous bi_private for pending bio */ 1234 #endif 1235 struct f2fs_rwsem io_rwsem; /* blocking op for bio */ 1236 spinlock_t io_lock; /* serialize DATA/NODE IOs */ 1237 struct list_head io_list; /* track fios */ 1238 struct list_head bio_list; /* bio entry list head */ 1239 struct f2fs_rwsem bio_list_lock; /* lock to protect bio entry list */ 1240 }; 1241 1242 #define FDEV(i) (sbi->devs[i]) 1243 #define RDEV(i) (raw_super->devs[i]) 1244 struct f2fs_dev_info { 1245 struct block_device *bdev; 1246 char path[MAX_PATH_LEN]; 1247 unsigned int total_segments; 1248 block_t start_blk; 1249 block_t end_blk; 1250 #ifdef CONFIG_BLK_DEV_ZONED 1251 unsigned int nr_blkz; /* Total number of zones */ 1252 unsigned long *blkz_seq; /* Bitmap indicating sequential zones */ 1253 #endif 1254 }; 1255 1256 enum inode_type { 1257 DIR_INODE, /* for dirty dir inode */ 1258 FILE_INODE, /* for dirty regular/symlink inode */ 1259 DIRTY_META, /* for all dirtied inode metadata */ 1260 NR_INODE_TYPE, 1261 }; 1262 1263 /* for inner inode cache management */ 1264 struct inode_management { 1265 struct radix_tree_root ino_root; /* ino entry array */ 1266 spinlock_t ino_lock; /* for ino entry lock */ 1267 struct list_head ino_list; /* inode list head */ 1268 unsigned long ino_num; /* number of entries */ 1269 }; 1270 1271 /* for GC_AT */ 1272 struct atgc_management { 1273 bool atgc_enabled; /* ATGC is enabled or not */ 1274 struct rb_root_cached root; /* root of victim rb-tree */ 1275 struct list_head victim_list; /* linked with all victim entries */ 1276 unsigned int victim_count; /* victim count in rb-tree */ 1277 unsigned int candidate_ratio; /* candidate ratio */ 1278 unsigned int max_candidate_count; /* max candidate count */ 1279 unsigned int age_weight; /* age weight, vblock_weight = 100 - age_weight */ 1280 unsigned long long age_threshold; /* age threshold */ 1281 }; 1282 1283 struct f2fs_gc_control { 1284 unsigned int victim_segno; /* target victim segment number */ 1285 int init_gc_type; /* FG_GC or BG_GC */ 1286 bool no_bg_gc; /* check the space and stop bg_gc */ 1287 bool should_migrate_blocks; /* should migrate blocks */ 1288 bool err_gc_skipped; /* return EAGAIN if GC skipped */ 1289 unsigned int nr_free_secs; /* # of free sections to do GC */ 1290 }; 1291 1292 /* 1293 * For s_flag in struct f2fs_sb_info 1294 * Modification on enum should be synchronized with s_flag array 1295 */ 1296 enum { 1297 SBI_IS_DIRTY, /* dirty flag for checkpoint */ 1298 SBI_IS_CLOSE, /* specify unmounting */ 1299 SBI_NEED_FSCK, /* need fsck.f2fs to fix */ 1300 SBI_POR_DOING, /* recovery is doing or not */ 1301 SBI_NEED_SB_WRITE, /* need to recover superblock */ 1302 SBI_NEED_CP, /* need to checkpoint */ 1303 SBI_IS_SHUTDOWN, /* shutdown by ioctl */ 1304 SBI_IS_RECOVERED, /* recovered orphan/data */ 1305 SBI_CP_DISABLED, /* CP was disabled last mount */ 1306 SBI_CP_DISABLED_QUICK, /* CP was disabled quickly */ 1307 SBI_QUOTA_NEED_FLUSH, /* need to flush quota info in CP */ 1308 SBI_QUOTA_SKIP_FLUSH, /* skip flushing quota in current CP */ 1309 SBI_QUOTA_NEED_REPAIR, /* quota file may be corrupted */ 1310 SBI_IS_RESIZEFS, /* resizefs is in process */ 1311 SBI_IS_FREEZING, /* freezefs is in process */ 1312 SBI_IS_WRITABLE, /* remove ro mountoption transiently */ 1313 MAX_SBI_FLAG, 1314 }; 1315 1316 enum { 1317 CP_TIME, 1318 REQ_TIME, 1319 DISCARD_TIME, 1320 GC_TIME, 1321 DISABLE_TIME, 1322 UMOUNT_DISCARD_TIMEOUT, 1323 MAX_TIME, 1324 }; 1325 1326 /* Note that you need to keep synchronization with this gc_mode_names array */ 1327 enum { 1328 GC_NORMAL, 1329 GC_IDLE_CB, 1330 GC_IDLE_GREEDY, 1331 GC_IDLE_AT, 1332 GC_URGENT_HIGH, 1333 GC_URGENT_LOW, 1334 GC_URGENT_MID, 1335 MAX_GC_MODE, 1336 }; 1337 1338 enum { 1339 BGGC_MODE_ON, /* background gc is on */ 1340 BGGC_MODE_OFF, /* background gc is off */ 1341 BGGC_MODE_SYNC, /* 1342 * background gc is on, migrating blocks 1343 * like foreground gc 1344 */ 1345 }; 1346 1347 enum { 1348 FS_MODE_ADAPTIVE, /* use both lfs/ssr allocation */ 1349 FS_MODE_LFS, /* use lfs allocation only */ 1350 FS_MODE_FRAGMENT_SEG, /* segment fragmentation mode */ 1351 FS_MODE_FRAGMENT_BLK, /* block fragmentation mode */ 1352 }; 1353 1354 enum { 1355 ALLOC_MODE_DEFAULT, /* stay default */ 1356 ALLOC_MODE_REUSE, /* reuse segments as much as possible */ 1357 }; 1358 1359 enum fsync_mode { 1360 FSYNC_MODE_POSIX, /* fsync follows posix semantics */ 1361 FSYNC_MODE_STRICT, /* fsync behaves in line with ext4 */ 1362 FSYNC_MODE_NOBARRIER, /* fsync behaves nobarrier based on posix */ 1363 }; 1364 1365 enum { 1366 COMPR_MODE_FS, /* 1367 * automatically compress compression 1368 * enabled files 1369 */ 1370 COMPR_MODE_USER, /* 1371 * automatical compression is disabled. 1372 * user can control the file compression 1373 * using ioctls 1374 */ 1375 }; 1376 1377 enum { 1378 DISCARD_UNIT_BLOCK, /* basic discard unit is block */ 1379 DISCARD_UNIT_SEGMENT, /* basic discard unit is segment */ 1380 DISCARD_UNIT_SECTION, /* basic discard unit is section */ 1381 }; 1382 1383 enum { 1384 MEMORY_MODE_NORMAL, /* memory mode for normal devices */ 1385 MEMORY_MODE_LOW, /* memory mode for low memry devices */ 1386 }; 1387 1388 enum errors_option { 1389 MOUNT_ERRORS_READONLY, /* remount fs ro on errors */ 1390 MOUNT_ERRORS_CONTINUE, /* continue on errors */ 1391 MOUNT_ERRORS_PANIC, /* panic on errors */ 1392 }; 1393 1394 enum { 1395 BACKGROUND, 1396 FOREGROUND, 1397 MAX_CALL_TYPE, 1398 TOTAL_CALL = FOREGROUND, 1399 }; 1400 1401 static inline int f2fs_test_bit(unsigned int nr, char *addr); 1402 static inline void f2fs_set_bit(unsigned int nr, char *addr); 1403 static inline void f2fs_clear_bit(unsigned int nr, char *addr); 1404 1405 /* 1406 * Layout of f2fs page.private: 1407 * 1408 * Layout A: lowest bit should be 1 1409 * | bit0 = 1 | bit1 | bit2 | ... | bit MAX | private data .... | 1410 * bit 0 PAGE_PRIVATE_NOT_POINTER 1411 * bit 1 PAGE_PRIVATE_ONGOING_MIGRATION 1412 * bit 2 PAGE_PRIVATE_INLINE_INODE 1413 * bit 3 PAGE_PRIVATE_REF_RESOURCE 1414 * bit 4- f2fs private data 1415 * 1416 * Layout B: lowest bit should be 0 1417 * page.private is a wrapped pointer. 1418 */ 1419 enum { 1420 PAGE_PRIVATE_NOT_POINTER, /* private contains non-pointer data */ 1421 PAGE_PRIVATE_ONGOING_MIGRATION, /* data page which is on-going migrating */ 1422 PAGE_PRIVATE_INLINE_INODE, /* inode page contains inline data */ 1423 PAGE_PRIVATE_REF_RESOURCE, /* dirty page has referenced resources */ 1424 PAGE_PRIVATE_MAX 1425 }; 1426 1427 /* For compression */ 1428 enum compress_algorithm_type { 1429 COMPRESS_LZO, 1430 COMPRESS_LZ4, 1431 COMPRESS_ZSTD, 1432 COMPRESS_LZORLE, 1433 COMPRESS_MAX, 1434 }; 1435 1436 enum compress_flag { 1437 COMPRESS_CHKSUM, 1438 COMPRESS_MAX_FLAG, 1439 }; 1440 1441 #define COMPRESS_WATERMARK 20 1442 #define COMPRESS_PERCENT 20 1443 1444 #define COMPRESS_DATA_RESERVED_SIZE 4 1445 struct compress_data { 1446 __le32 clen; /* compressed data size */ 1447 __le32 chksum; /* compressed data chksum */ 1448 __le32 reserved[COMPRESS_DATA_RESERVED_SIZE]; /* reserved */ 1449 u8 cdata[]; /* compressed data */ 1450 }; 1451 1452 #define COMPRESS_HEADER_SIZE (sizeof(struct compress_data)) 1453 1454 #define F2FS_COMPRESSED_PAGE_MAGIC 0xF5F2C000 1455 1456 #define F2FS_ZSTD_DEFAULT_CLEVEL 1 1457 1458 #define COMPRESS_LEVEL_OFFSET 8 1459 1460 /* compress context */ 1461 struct compress_ctx { 1462 struct inode *inode; /* inode the context belong to */ 1463 pgoff_t cluster_idx; /* cluster index number */ 1464 unsigned int cluster_size; /* page count in cluster */ 1465 unsigned int log_cluster_size; /* log of cluster size */ 1466 struct page **rpages; /* pages store raw data in cluster */ 1467 unsigned int nr_rpages; /* total page number in rpages */ 1468 struct page **cpages; /* pages store compressed data in cluster */ 1469 unsigned int nr_cpages; /* total page number in cpages */ 1470 unsigned int valid_nr_cpages; /* valid page number in cpages */ 1471 void *rbuf; /* virtual mapped address on rpages */ 1472 struct compress_data *cbuf; /* virtual mapped address on cpages */ 1473 size_t rlen; /* valid data length in rbuf */ 1474 size_t clen; /* valid data length in cbuf */ 1475 void *private; /* payload buffer for specified compression algorithm */ 1476 void *private2; /* extra payload buffer */ 1477 }; 1478 1479 /* compress context for write IO path */ 1480 struct compress_io_ctx { 1481 u32 magic; /* magic number to indicate page is compressed */ 1482 struct inode *inode; /* inode the context belong to */ 1483 struct page **rpages; /* pages store raw data in cluster */ 1484 unsigned int nr_rpages; /* total page number in rpages */ 1485 atomic_t pending_pages; /* in-flight compressed page count */ 1486 }; 1487 1488 /* Context for decompressing one cluster on the read IO path */ 1489 struct decompress_io_ctx { 1490 u32 magic; /* magic number to indicate page is compressed */ 1491 struct inode *inode; /* inode the context belong to */ 1492 pgoff_t cluster_idx; /* cluster index number */ 1493 unsigned int cluster_size; /* page count in cluster */ 1494 unsigned int log_cluster_size; /* log of cluster size */ 1495 struct page **rpages; /* pages store raw data in cluster */ 1496 unsigned int nr_rpages; /* total page number in rpages */ 1497 struct page **cpages; /* pages store compressed data in cluster */ 1498 unsigned int nr_cpages; /* total page number in cpages */ 1499 struct page **tpages; /* temp pages to pad holes in cluster */ 1500 void *rbuf; /* virtual mapped address on rpages */ 1501 struct compress_data *cbuf; /* virtual mapped address on cpages */ 1502 size_t rlen; /* valid data length in rbuf */ 1503 size_t clen; /* valid data length in cbuf */ 1504 1505 /* 1506 * The number of compressed pages remaining to be read in this cluster. 1507 * This is initially nr_cpages. It is decremented by 1 each time a page 1508 * has been read (or failed to be read). When it reaches 0, the cluster 1509 * is decompressed (or an error is reported). 1510 * 1511 * If an error occurs before all the pages have been submitted for I/O, 1512 * then this will never reach 0. In this case the I/O submitter is 1513 * responsible for calling f2fs_decompress_end_io() instead. 1514 */ 1515 atomic_t remaining_pages; 1516 1517 /* 1518 * Number of references to this decompress_io_ctx. 1519 * 1520 * One reference is held for I/O completion. This reference is dropped 1521 * after the pagecache pages are updated and unlocked -- either after 1522 * decompression (and verity if enabled), or after an error. 1523 * 1524 * In addition, each compressed page holds a reference while it is in a 1525 * bio. These references are necessary prevent compressed pages from 1526 * being freed while they are still in a bio. 1527 */ 1528 refcount_t refcnt; 1529 1530 bool failed; /* IO error occurred before decompression? */ 1531 bool need_verity; /* need fs-verity verification after decompression? */ 1532 void *private; /* payload buffer for specified decompression algorithm */ 1533 void *private2; /* extra payload buffer */ 1534 struct work_struct verity_work; /* work to verify the decompressed pages */ 1535 struct work_struct free_work; /* work for late free this structure itself */ 1536 }; 1537 1538 #define NULL_CLUSTER ((unsigned int)(~0)) 1539 #define MIN_COMPRESS_LOG_SIZE 2 1540 #define MAX_COMPRESS_LOG_SIZE 8 1541 #define MAX_COMPRESS_WINDOW_SIZE(log_size) ((PAGE_SIZE) << (log_size)) 1542 1543 struct f2fs_sb_info { 1544 struct super_block *sb; /* pointer to VFS super block */ 1545 struct proc_dir_entry *s_proc; /* proc entry */ 1546 struct f2fs_super_block *raw_super; /* raw super block pointer */ 1547 struct f2fs_rwsem sb_lock; /* lock for raw super block */ 1548 int valid_super_block; /* valid super block no */ 1549 unsigned long s_flag; /* flags for sbi */ 1550 struct mutex writepages; /* mutex for writepages() */ 1551 1552 #ifdef CONFIG_BLK_DEV_ZONED 1553 unsigned int blocks_per_blkz; /* F2FS blocks per zone */ 1554 #endif 1555 1556 /* for node-related operations */ 1557 struct f2fs_nm_info *nm_info; /* node manager */ 1558 struct inode *node_inode; /* cache node blocks */ 1559 1560 /* for segment-related operations */ 1561 struct f2fs_sm_info *sm_info; /* segment manager */ 1562 1563 /* for bio operations */ 1564 struct f2fs_bio_info *write_io[NR_PAGE_TYPE]; /* for write bios */ 1565 /* keep migration IO order for LFS mode */ 1566 struct f2fs_rwsem io_order_lock; 1567 pgoff_t page_eio_ofs[NR_PAGE_TYPE]; /* EIO page offset */ 1568 int page_eio_cnt[NR_PAGE_TYPE]; /* EIO count */ 1569 1570 /* for checkpoint */ 1571 struct f2fs_checkpoint *ckpt; /* raw checkpoint pointer */ 1572 int cur_cp_pack; /* remain current cp pack */ 1573 spinlock_t cp_lock; /* for flag in ckpt */ 1574 struct inode *meta_inode; /* cache meta blocks */ 1575 struct f2fs_rwsem cp_global_sem; /* checkpoint procedure lock */ 1576 struct f2fs_rwsem cp_rwsem; /* blocking FS operations */ 1577 struct f2fs_rwsem node_write; /* locking node writes */ 1578 struct f2fs_rwsem node_change; /* locking node change */ 1579 wait_queue_head_t cp_wait; 1580 unsigned long last_time[MAX_TIME]; /* to store time in jiffies */ 1581 long interval_time[MAX_TIME]; /* to store thresholds */ 1582 struct ckpt_req_control cprc_info; /* for checkpoint request control */ 1583 1584 struct inode_management im[MAX_INO_ENTRY]; /* manage inode cache */ 1585 1586 spinlock_t fsync_node_lock; /* for node entry lock */ 1587 struct list_head fsync_node_list; /* node list head */ 1588 unsigned int fsync_seg_id; /* sequence id */ 1589 unsigned int fsync_node_num; /* number of node entries */ 1590 1591 /* for orphan inode, use 0'th array */ 1592 unsigned int max_orphans; /* max orphan inodes */ 1593 1594 /* for inode management */ 1595 struct list_head inode_list[NR_INODE_TYPE]; /* dirty inode list */ 1596 spinlock_t inode_lock[NR_INODE_TYPE]; /* for dirty inode list lock */ 1597 struct mutex flush_lock; /* for flush exclusion */ 1598 1599 /* for extent tree cache */ 1600 struct extent_tree_info extent_tree[NR_EXTENT_CACHES]; 1601 atomic64_t allocated_data_blocks; /* for block age extent_cache */ 1602 1603 /* The threshold used for hot and warm data seperation*/ 1604 unsigned int hot_data_age_threshold; 1605 unsigned int warm_data_age_threshold; 1606 unsigned int last_age_weight; 1607 1608 /* basic filesystem units */ 1609 unsigned int log_sectors_per_block; /* log2 sectors per block */ 1610 unsigned int log_blocksize; /* log2 block size */ 1611 unsigned int blocksize; /* block size */ 1612 unsigned int root_ino_num; /* root inode number*/ 1613 unsigned int node_ino_num; /* node inode number*/ 1614 unsigned int meta_ino_num; /* meta inode number*/ 1615 unsigned int log_blocks_per_seg; /* log2 blocks per segment */ 1616 unsigned int blocks_per_seg; /* blocks per segment */ 1617 unsigned int unusable_blocks_per_sec; /* unusable blocks per section */ 1618 unsigned int segs_per_sec; /* segments per section */ 1619 unsigned int secs_per_zone; /* sections per zone */ 1620 unsigned int total_sections; /* total section count */ 1621 unsigned int total_node_count; /* total node block count */ 1622 unsigned int total_valid_node_count; /* valid node block count */ 1623 int dir_level; /* directory level */ 1624 bool readdir_ra; /* readahead inode in readdir */ 1625 u64 max_io_bytes; /* max io bytes to merge IOs */ 1626 1627 block_t user_block_count; /* # of user blocks */ 1628 block_t total_valid_block_count; /* # of valid blocks */ 1629 block_t discard_blks; /* discard command candidats */ 1630 block_t last_valid_block_count; /* for recovery */ 1631 block_t reserved_blocks; /* configurable reserved blocks */ 1632 block_t current_reserved_blocks; /* current reserved blocks */ 1633 1634 /* Additional tracking for no checkpoint mode */ 1635 block_t unusable_block_count; /* # of blocks saved by last cp */ 1636 1637 unsigned int nquota_files; /* # of quota sysfile */ 1638 struct f2fs_rwsem quota_sem; /* blocking cp for flags */ 1639 1640 /* # of pages, see count_type */ 1641 atomic_t nr_pages[NR_COUNT_TYPE]; 1642 /* # of allocated blocks */ 1643 struct percpu_counter alloc_valid_block_count; 1644 /* # of node block writes as roll forward recovery */ 1645 struct percpu_counter rf_node_block_count; 1646 1647 /* writeback control */ 1648 atomic_t wb_sync_req[META]; /* count # of WB_SYNC threads */ 1649 1650 /* valid inode count */ 1651 struct percpu_counter total_valid_inode_count; 1652 1653 struct f2fs_mount_info mount_opt; /* mount options */ 1654 1655 /* for cleaning operations */ 1656 struct f2fs_rwsem gc_lock; /* 1657 * semaphore for GC, avoid 1658 * race between GC and GC or CP 1659 */ 1660 struct f2fs_gc_kthread *gc_thread; /* GC thread */ 1661 struct atgc_management am; /* atgc management */ 1662 unsigned int cur_victim_sec; /* current victim section num */ 1663 unsigned int gc_mode; /* current GC state */ 1664 unsigned int next_victim_seg[2]; /* next segment in victim section */ 1665 spinlock_t gc_remaining_trials_lock; 1666 /* remaining trial count for GC_URGENT_* and GC_IDLE_* */ 1667 unsigned int gc_remaining_trials; 1668 1669 /* for skip statistic */ 1670 unsigned long long skipped_gc_rwsem; /* FG_GC only */ 1671 1672 /* threshold for gc trials on pinned files */ 1673 u64 gc_pin_file_threshold; 1674 struct f2fs_rwsem pin_sem; 1675 1676 /* maximum # of trials to find a victim segment for SSR and GC */ 1677 unsigned int max_victim_search; 1678 /* migration granularity of garbage collection, unit: segment */ 1679 unsigned int migration_granularity; 1680 1681 /* 1682 * for stat information. 1683 * one is for the LFS mode, and the other is for the SSR mode. 1684 */ 1685 #ifdef CONFIG_F2FS_STAT_FS 1686 struct f2fs_stat_info *stat_info; /* FS status information */ 1687 atomic_t meta_count[META_MAX]; /* # of meta blocks */ 1688 unsigned int segment_count[2]; /* # of allocated segments */ 1689 unsigned int block_count[2]; /* # of allocated blocks */ 1690 atomic_t inplace_count; /* # of inplace update */ 1691 /* # of lookup extent cache */ 1692 atomic64_t total_hit_ext[NR_EXTENT_CACHES]; 1693 /* # of hit rbtree extent node */ 1694 atomic64_t read_hit_rbtree[NR_EXTENT_CACHES]; 1695 /* # of hit cached extent node */ 1696 atomic64_t read_hit_cached[NR_EXTENT_CACHES]; 1697 /* # of hit largest extent node in read extent cache */ 1698 atomic64_t read_hit_largest; 1699 atomic_t inline_xattr; /* # of inline_xattr inodes */ 1700 atomic_t inline_inode; /* # of inline_data inodes */ 1701 atomic_t inline_dir; /* # of inline_dentry inodes */ 1702 atomic_t compr_inode; /* # of compressed inodes */ 1703 atomic64_t compr_blocks; /* # of compressed blocks */ 1704 atomic_t swapfile_inode; /* # of swapfile inodes */ 1705 atomic_t atomic_files; /* # of opened atomic file */ 1706 atomic_t max_aw_cnt; /* max # of atomic writes */ 1707 unsigned int io_skip_bggc; /* skip background gc for in-flight IO */ 1708 unsigned int other_skip_bggc; /* skip background gc for other reasons */ 1709 unsigned int ndirty_inode[NR_INODE_TYPE]; /* # of dirty inodes */ 1710 atomic_t cp_call_count[MAX_CALL_TYPE]; /* # of cp call */ 1711 #endif 1712 spinlock_t stat_lock; /* lock for stat operations */ 1713 1714 /* to attach REQ_META|REQ_FUA flags */ 1715 unsigned int data_io_flag; 1716 unsigned int node_io_flag; 1717 1718 /* For sysfs support */ 1719 struct kobject s_kobj; /* /sys/fs/f2fs/<devname> */ 1720 struct completion s_kobj_unregister; 1721 1722 struct kobject s_stat_kobj; /* /sys/fs/f2fs/<devname>/stat */ 1723 struct completion s_stat_kobj_unregister; 1724 1725 struct kobject s_feature_list_kobj; /* /sys/fs/f2fs/<devname>/feature_list */ 1726 struct completion s_feature_list_kobj_unregister; 1727 1728 /* For shrinker support */ 1729 struct list_head s_list; 1730 struct mutex umount_mutex; 1731 unsigned int shrinker_run_no; 1732 1733 /* For multi devices */ 1734 int s_ndevs; /* number of devices */ 1735 struct f2fs_dev_info *devs; /* for device list */ 1736 unsigned int dirty_device; /* for checkpoint data flush */ 1737 spinlock_t dev_lock; /* protect dirty_device */ 1738 bool aligned_blksize; /* all devices has the same logical blksize */ 1739 1740 /* For write statistics */ 1741 u64 sectors_written_start; 1742 u64 kbytes_written; 1743 1744 /* Reference to checksum algorithm driver via cryptoapi */ 1745 struct crypto_shash *s_chksum_driver; 1746 1747 /* Precomputed FS UUID checksum for seeding other checksums */ 1748 __u32 s_chksum_seed; 1749 1750 struct workqueue_struct *post_read_wq; /* post read workqueue */ 1751 1752 /* 1753 * If we are in irq context, let's update error information into 1754 * on-disk superblock in the work. 1755 */ 1756 struct work_struct s_error_work; 1757 unsigned char errors[MAX_F2FS_ERRORS]; /* error flags */ 1758 unsigned char stop_reason[MAX_STOP_REASON]; /* stop reason */ 1759 spinlock_t error_lock; /* protect errors/stop_reason array */ 1760 bool error_dirty; /* errors of sb is dirty */ 1761 1762 struct kmem_cache *inline_xattr_slab; /* inline xattr entry */ 1763 unsigned int inline_xattr_slab_size; /* default inline xattr slab size */ 1764 1765 /* For reclaimed segs statistics per each GC mode */ 1766 unsigned int gc_segment_mode; /* GC state for reclaimed segments */ 1767 unsigned int gc_reclaimed_segs[MAX_GC_MODE]; /* Reclaimed segs for each mode */ 1768 1769 unsigned long seq_file_ra_mul; /* multiplier for ra_pages of seq. files in fadvise */ 1770 1771 int max_fragment_chunk; /* max chunk size for block fragmentation mode */ 1772 int max_fragment_hole; /* max hole size for block fragmentation mode */ 1773 1774 /* For atomic write statistics */ 1775 atomic64_t current_atomic_write; 1776 s64 peak_atomic_write; 1777 u64 committed_atomic_block; 1778 u64 revoked_atomic_block; 1779 1780 #ifdef CONFIG_F2FS_FS_COMPRESSION 1781 struct kmem_cache *page_array_slab; /* page array entry */ 1782 unsigned int page_array_slab_size; /* default page array slab size */ 1783 1784 /* For runtime compression statistics */ 1785 u64 compr_written_block; 1786 u64 compr_saved_block; 1787 u32 compr_new_inode; 1788 1789 /* For compressed block cache */ 1790 struct inode *compress_inode; /* cache compressed blocks */ 1791 unsigned int compress_percent; /* cache page percentage */ 1792 unsigned int compress_watermark; /* cache page watermark */ 1793 atomic_t compress_page_hit; /* cache hit count */ 1794 #endif 1795 1796 #ifdef CONFIG_F2FS_IOSTAT 1797 /* For app/fs IO statistics */ 1798 spinlock_t iostat_lock; 1799 unsigned long long iostat_count[NR_IO_TYPE]; 1800 unsigned long long iostat_bytes[NR_IO_TYPE]; 1801 unsigned long long prev_iostat_bytes[NR_IO_TYPE]; 1802 bool iostat_enable; 1803 unsigned long iostat_next_period; 1804 unsigned int iostat_period_ms; 1805 1806 /* For io latency related statistics info in one iostat period */ 1807 spinlock_t iostat_lat_lock; 1808 struct iostat_lat_info *iostat_io_lat; 1809 #endif 1810 }; 1811 1812 /* Definitions to access f2fs_sb_info */ 1813 #define BLKS_PER_SEG(sbi) \ 1814 ((sbi)->blocks_per_seg) 1815 #define BLKS_PER_SEC(sbi) \ 1816 ((sbi)->segs_per_sec << (sbi)->log_blocks_per_seg) 1817 #define SEGS_PER_SEC(sbi) \ 1818 ((sbi)->segs_per_sec) 1819 1820 __printf(3, 4) 1821 void f2fs_printk(struct f2fs_sb_info *sbi, bool limit_rate, const char *fmt, ...); 1822 1823 #define f2fs_err(sbi, fmt, ...) \ 1824 f2fs_printk(sbi, false, KERN_ERR fmt, ##__VA_ARGS__) 1825 #define f2fs_warn(sbi, fmt, ...) \ 1826 f2fs_printk(sbi, false, KERN_WARNING fmt, ##__VA_ARGS__) 1827 #define f2fs_notice(sbi, fmt, ...) \ 1828 f2fs_printk(sbi, false, KERN_NOTICE fmt, ##__VA_ARGS__) 1829 #define f2fs_info(sbi, fmt, ...) \ 1830 f2fs_printk(sbi, false, KERN_INFO fmt, ##__VA_ARGS__) 1831 #define f2fs_debug(sbi, fmt, ...) \ 1832 f2fs_printk(sbi, false, KERN_DEBUG fmt, ##__VA_ARGS__) 1833 1834 #define f2fs_err_ratelimited(sbi, fmt, ...) \ 1835 f2fs_printk(sbi, true, KERN_ERR fmt, ##__VA_ARGS__) 1836 #define f2fs_warn_ratelimited(sbi, fmt, ...) \ 1837 f2fs_printk(sbi, true, KERN_WARNING fmt, ##__VA_ARGS__) 1838 #define f2fs_info_ratelimited(sbi, fmt, ...) \ 1839 f2fs_printk(sbi, true, KERN_INFO fmt, ##__VA_ARGS__) 1840 1841 #ifdef CONFIG_F2FS_FAULT_INJECTION 1842 #define time_to_inject(sbi, type) __time_to_inject(sbi, type, __func__, \ 1843 __builtin_return_address(0)) 1844 static inline bool __time_to_inject(struct f2fs_sb_info *sbi, int type, 1845 const char *func, const char *parent_func) 1846 { 1847 struct f2fs_fault_info *ffi = &F2FS_OPTION(sbi).fault_info; 1848 1849 if (!ffi->inject_rate) 1850 return false; 1851 1852 if (!IS_FAULT_SET(ffi, type)) 1853 return false; 1854 1855 atomic_inc(&ffi->inject_ops); 1856 if (atomic_read(&ffi->inject_ops) >= ffi->inject_rate) { 1857 atomic_set(&ffi->inject_ops, 0); 1858 f2fs_info_ratelimited(sbi, "inject %s in %s of %pS", 1859 f2fs_fault_name[type], func, parent_func); 1860 return true; 1861 } 1862 return false; 1863 } 1864 #else 1865 static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type) 1866 { 1867 return false; 1868 } 1869 #endif 1870 1871 /* 1872 * Test if the mounted volume is a multi-device volume. 1873 * - For a single regular disk volume, sbi->s_ndevs is 0. 1874 * - For a single zoned disk volume, sbi->s_ndevs is 1. 1875 * - For a multi-device volume, sbi->s_ndevs is always 2 or more. 1876 */ 1877 static inline bool f2fs_is_multi_device(struct f2fs_sb_info *sbi) 1878 { 1879 return sbi->s_ndevs > 1; 1880 } 1881 1882 static inline void f2fs_update_time(struct f2fs_sb_info *sbi, int type) 1883 { 1884 unsigned long now = jiffies; 1885 1886 sbi->last_time[type] = now; 1887 1888 /* DISCARD_TIME and GC_TIME are based on REQ_TIME */ 1889 if (type == REQ_TIME) { 1890 sbi->last_time[DISCARD_TIME] = now; 1891 sbi->last_time[GC_TIME] = now; 1892 } 1893 } 1894 1895 static inline bool f2fs_time_over(struct f2fs_sb_info *sbi, int type) 1896 { 1897 unsigned long interval = sbi->interval_time[type] * HZ; 1898 1899 return time_after(jiffies, sbi->last_time[type] + interval); 1900 } 1901 1902 static inline unsigned int f2fs_time_to_wait(struct f2fs_sb_info *sbi, 1903 int type) 1904 { 1905 unsigned long interval = sbi->interval_time[type] * HZ; 1906 unsigned int wait_ms = 0; 1907 long delta; 1908 1909 delta = (sbi->last_time[type] + interval) - jiffies; 1910 if (delta > 0) 1911 wait_ms = jiffies_to_msecs(delta); 1912 1913 return wait_ms; 1914 } 1915 1916 /* 1917 * Inline functions 1918 */ 1919 static inline u32 __f2fs_crc32(struct f2fs_sb_info *sbi, u32 crc, 1920 const void *address, unsigned int length) 1921 { 1922 struct { 1923 struct shash_desc shash; 1924 char ctx[4]; 1925 } desc; 1926 int err; 1927 1928 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver) != sizeof(desc.ctx)); 1929 1930 desc.shash.tfm = sbi->s_chksum_driver; 1931 *(u32 *)desc.ctx = crc; 1932 1933 err = crypto_shash_update(&desc.shash, address, length); 1934 BUG_ON(err); 1935 1936 return *(u32 *)desc.ctx; 1937 } 1938 1939 static inline u32 f2fs_crc32(struct f2fs_sb_info *sbi, const void *address, 1940 unsigned int length) 1941 { 1942 return __f2fs_crc32(sbi, F2FS_SUPER_MAGIC, address, length); 1943 } 1944 1945 static inline bool f2fs_crc_valid(struct f2fs_sb_info *sbi, __u32 blk_crc, 1946 void *buf, size_t buf_size) 1947 { 1948 return f2fs_crc32(sbi, buf, buf_size) == blk_crc; 1949 } 1950 1951 static inline u32 f2fs_chksum(struct f2fs_sb_info *sbi, u32 crc, 1952 const void *address, unsigned int length) 1953 { 1954 return __f2fs_crc32(sbi, crc, address, length); 1955 } 1956 1957 static inline struct f2fs_inode_info *F2FS_I(struct inode *inode) 1958 { 1959 return container_of(inode, struct f2fs_inode_info, vfs_inode); 1960 } 1961 1962 static inline struct f2fs_sb_info *F2FS_SB(struct super_block *sb) 1963 { 1964 return sb->s_fs_info; 1965 } 1966 1967 static inline struct f2fs_sb_info *F2FS_I_SB(struct inode *inode) 1968 { 1969 return F2FS_SB(inode->i_sb); 1970 } 1971 1972 static inline struct f2fs_sb_info *F2FS_M_SB(struct address_space *mapping) 1973 { 1974 return F2FS_I_SB(mapping->host); 1975 } 1976 1977 static inline struct f2fs_sb_info *F2FS_P_SB(struct page *page) 1978 { 1979 return F2FS_M_SB(page_file_mapping(page)); 1980 } 1981 1982 static inline struct f2fs_super_block *F2FS_RAW_SUPER(struct f2fs_sb_info *sbi) 1983 { 1984 return (struct f2fs_super_block *)(sbi->raw_super); 1985 } 1986 1987 static inline struct f2fs_checkpoint *F2FS_CKPT(struct f2fs_sb_info *sbi) 1988 { 1989 return (struct f2fs_checkpoint *)(sbi->ckpt); 1990 } 1991 1992 static inline struct f2fs_node *F2FS_NODE(struct page *page) 1993 { 1994 return (struct f2fs_node *)page_address(page); 1995 } 1996 1997 static inline struct f2fs_inode *F2FS_INODE(struct page *page) 1998 { 1999 return &((struct f2fs_node *)page_address(page))->i; 2000 } 2001 2002 static inline struct f2fs_nm_info *NM_I(struct f2fs_sb_info *sbi) 2003 { 2004 return (struct f2fs_nm_info *)(sbi->nm_info); 2005 } 2006 2007 static inline struct f2fs_sm_info *SM_I(struct f2fs_sb_info *sbi) 2008 { 2009 return (struct f2fs_sm_info *)(sbi->sm_info); 2010 } 2011 2012 static inline struct sit_info *SIT_I(struct f2fs_sb_info *sbi) 2013 { 2014 return (struct sit_info *)(SM_I(sbi)->sit_info); 2015 } 2016 2017 static inline struct free_segmap_info *FREE_I(struct f2fs_sb_info *sbi) 2018 { 2019 return (struct free_segmap_info *)(SM_I(sbi)->free_info); 2020 } 2021 2022 static inline struct dirty_seglist_info *DIRTY_I(struct f2fs_sb_info *sbi) 2023 { 2024 return (struct dirty_seglist_info *)(SM_I(sbi)->dirty_info); 2025 } 2026 2027 static inline struct address_space *META_MAPPING(struct f2fs_sb_info *sbi) 2028 { 2029 return sbi->meta_inode->i_mapping; 2030 } 2031 2032 static inline struct address_space *NODE_MAPPING(struct f2fs_sb_info *sbi) 2033 { 2034 return sbi->node_inode->i_mapping; 2035 } 2036 2037 static inline bool is_sbi_flag_set(struct f2fs_sb_info *sbi, unsigned int type) 2038 { 2039 return test_bit(type, &sbi->s_flag); 2040 } 2041 2042 static inline void set_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type) 2043 { 2044 set_bit(type, &sbi->s_flag); 2045 } 2046 2047 static inline void clear_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type) 2048 { 2049 clear_bit(type, &sbi->s_flag); 2050 } 2051 2052 static inline unsigned long long cur_cp_version(struct f2fs_checkpoint *cp) 2053 { 2054 return le64_to_cpu(cp->checkpoint_ver); 2055 } 2056 2057 static inline unsigned long f2fs_qf_ino(struct super_block *sb, int type) 2058 { 2059 if (type < F2FS_MAX_QUOTAS) 2060 return le32_to_cpu(F2FS_SB(sb)->raw_super->qf_ino[type]); 2061 return 0; 2062 } 2063 2064 static inline __u64 cur_cp_crc(struct f2fs_checkpoint *cp) 2065 { 2066 size_t crc_offset = le32_to_cpu(cp->checksum_offset); 2067 return le32_to_cpu(*((__le32 *)((unsigned char *)cp + crc_offset))); 2068 } 2069 2070 static inline bool __is_set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 2071 { 2072 unsigned int ckpt_flags = le32_to_cpu(cp->ckpt_flags); 2073 2074 return ckpt_flags & f; 2075 } 2076 2077 static inline bool is_set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 2078 { 2079 return __is_set_ckpt_flags(F2FS_CKPT(sbi), f); 2080 } 2081 2082 static inline void __set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 2083 { 2084 unsigned int ckpt_flags; 2085 2086 ckpt_flags = le32_to_cpu(cp->ckpt_flags); 2087 ckpt_flags |= f; 2088 cp->ckpt_flags = cpu_to_le32(ckpt_flags); 2089 } 2090 2091 static inline void set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 2092 { 2093 unsigned long flags; 2094 2095 spin_lock_irqsave(&sbi->cp_lock, flags); 2096 __set_ckpt_flags(F2FS_CKPT(sbi), f); 2097 spin_unlock_irqrestore(&sbi->cp_lock, flags); 2098 } 2099 2100 static inline void __clear_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 2101 { 2102 unsigned int ckpt_flags; 2103 2104 ckpt_flags = le32_to_cpu(cp->ckpt_flags); 2105 ckpt_flags &= (~f); 2106 cp->ckpt_flags = cpu_to_le32(ckpt_flags); 2107 } 2108 2109 static inline void clear_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 2110 { 2111 unsigned long flags; 2112 2113 spin_lock_irqsave(&sbi->cp_lock, flags); 2114 __clear_ckpt_flags(F2FS_CKPT(sbi), f); 2115 spin_unlock_irqrestore(&sbi->cp_lock, flags); 2116 } 2117 2118 #define init_f2fs_rwsem(sem) \ 2119 do { \ 2120 static struct lock_class_key __key; \ 2121 \ 2122 __init_f2fs_rwsem((sem), #sem, &__key); \ 2123 } while (0) 2124 2125 static inline void __init_f2fs_rwsem(struct f2fs_rwsem *sem, 2126 const char *sem_name, struct lock_class_key *key) 2127 { 2128 __init_rwsem(&sem->internal_rwsem, sem_name, key); 2129 #ifdef CONFIG_F2FS_UNFAIR_RWSEM 2130 init_waitqueue_head(&sem->read_waiters); 2131 #endif 2132 } 2133 2134 static inline int f2fs_rwsem_is_locked(struct f2fs_rwsem *sem) 2135 { 2136 return rwsem_is_locked(&sem->internal_rwsem); 2137 } 2138 2139 static inline int f2fs_rwsem_is_contended(struct f2fs_rwsem *sem) 2140 { 2141 return rwsem_is_contended(&sem->internal_rwsem); 2142 } 2143 2144 static inline void f2fs_down_read(struct f2fs_rwsem *sem) 2145 { 2146 #ifdef CONFIG_F2FS_UNFAIR_RWSEM 2147 wait_event(sem->read_waiters, down_read_trylock(&sem->internal_rwsem)); 2148 #else 2149 down_read(&sem->internal_rwsem); 2150 #endif 2151 } 2152 2153 static inline int f2fs_down_read_trylock(struct f2fs_rwsem *sem) 2154 { 2155 return down_read_trylock(&sem->internal_rwsem); 2156 } 2157 2158 static inline void f2fs_up_read(struct f2fs_rwsem *sem) 2159 { 2160 up_read(&sem->internal_rwsem); 2161 } 2162 2163 static inline void f2fs_down_write(struct f2fs_rwsem *sem) 2164 { 2165 down_write(&sem->internal_rwsem); 2166 } 2167 2168 #ifdef CONFIG_DEBUG_LOCK_ALLOC 2169 static inline void f2fs_down_read_nested(struct f2fs_rwsem *sem, int subclass) 2170 { 2171 down_read_nested(&sem->internal_rwsem, subclass); 2172 } 2173 2174 static inline void f2fs_down_write_nested(struct f2fs_rwsem *sem, int subclass) 2175 { 2176 down_write_nested(&sem->internal_rwsem, subclass); 2177 } 2178 #else 2179 #define f2fs_down_read_nested(sem, subclass) f2fs_down_read(sem) 2180 #define f2fs_down_write_nested(sem, subclass) f2fs_down_write(sem) 2181 #endif 2182 2183 static inline int f2fs_down_write_trylock(struct f2fs_rwsem *sem) 2184 { 2185 return down_write_trylock(&sem->internal_rwsem); 2186 } 2187 2188 static inline void f2fs_up_write(struct f2fs_rwsem *sem) 2189 { 2190 up_write(&sem->internal_rwsem); 2191 #ifdef CONFIG_F2FS_UNFAIR_RWSEM 2192 wake_up_all(&sem->read_waiters); 2193 #endif 2194 } 2195 2196 static inline void f2fs_lock_op(struct f2fs_sb_info *sbi) 2197 { 2198 f2fs_down_read(&sbi->cp_rwsem); 2199 } 2200 2201 static inline int f2fs_trylock_op(struct f2fs_sb_info *sbi) 2202 { 2203 if (time_to_inject(sbi, FAULT_LOCK_OP)) 2204 return 0; 2205 return f2fs_down_read_trylock(&sbi->cp_rwsem); 2206 } 2207 2208 static inline void f2fs_unlock_op(struct f2fs_sb_info *sbi) 2209 { 2210 f2fs_up_read(&sbi->cp_rwsem); 2211 } 2212 2213 static inline void f2fs_lock_all(struct f2fs_sb_info *sbi) 2214 { 2215 f2fs_down_write(&sbi->cp_rwsem); 2216 } 2217 2218 static inline void f2fs_unlock_all(struct f2fs_sb_info *sbi) 2219 { 2220 f2fs_up_write(&sbi->cp_rwsem); 2221 } 2222 2223 static inline int __get_cp_reason(struct f2fs_sb_info *sbi) 2224 { 2225 int reason = CP_SYNC; 2226 2227 if (test_opt(sbi, FASTBOOT)) 2228 reason = CP_FASTBOOT; 2229 if (is_sbi_flag_set(sbi, SBI_IS_CLOSE)) 2230 reason = CP_UMOUNT; 2231 return reason; 2232 } 2233 2234 static inline bool __remain_node_summaries(int reason) 2235 { 2236 return (reason & (CP_UMOUNT | CP_FASTBOOT)); 2237 } 2238 2239 static inline bool __exist_node_summaries(struct f2fs_sb_info *sbi) 2240 { 2241 return (is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG) || 2242 is_set_ckpt_flags(sbi, CP_FASTBOOT_FLAG)); 2243 } 2244 2245 /* 2246 * Check whether the inode has blocks or not 2247 */ 2248 static inline int F2FS_HAS_BLOCKS(struct inode *inode) 2249 { 2250 block_t xattr_block = F2FS_I(inode)->i_xattr_nid ? 1 : 0; 2251 2252 return (inode->i_blocks >> F2FS_LOG_SECTORS_PER_BLOCK) > xattr_block; 2253 } 2254 2255 static inline bool f2fs_has_xattr_block(unsigned int ofs) 2256 { 2257 return ofs == XATTR_NODE_OFFSET; 2258 } 2259 2260 static inline bool __allow_reserved_blocks(struct f2fs_sb_info *sbi, 2261 struct inode *inode, bool cap) 2262 { 2263 if (!inode) 2264 return true; 2265 if (!test_opt(sbi, RESERVE_ROOT)) 2266 return false; 2267 if (IS_NOQUOTA(inode)) 2268 return true; 2269 if (uid_eq(F2FS_OPTION(sbi).s_resuid, current_fsuid())) 2270 return true; 2271 if (!gid_eq(F2FS_OPTION(sbi).s_resgid, GLOBAL_ROOT_GID) && 2272 in_group_p(F2FS_OPTION(sbi).s_resgid)) 2273 return true; 2274 if (cap && capable(CAP_SYS_RESOURCE)) 2275 return true; 2276 return false; 2277 } 2278 2279 static inline unsigned int get_available_block_count(struct f2fs_sb_info *sbi, 2280 struct inode *inode, bool cap) 2281 { 2282 block_t avail_user_block_count; 2283 2284 avail_user_block_count = sbi->user_block_count - 2285 sbi->current_reserved_blocks; 2286 2287 if (!__allow_reserved_blocks(sbi, inode, cap)) 2288 avail_user_block_count -= F2FS_OPTION(sbi).root_reserved_blocks; 2289 2290 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) { 2291 if (avail_user_block_count > sbi->unusable_block_count) 2292 avail_user_block_count -= sbi->unusable_block_count; 2293 else 2294 avail_user_block_count = 0; 2295 } 2296 2297 return avail_user_block_count; 2298 } 2299 2300 static inline void f2fs_i_blocks_write(struct inode *, block_t, bool, bool); 2301 static inline int inc_valid_block_count(struct f2fs_sb_info *sbi, 2302 struct inode *inode, blkcnt_t *count, bool partial) 2303 { 2304 long long diff = 0, release = 0; 2305 block_t avail_user_block_count; 2306 int ret; 2307 2308 ret = dquot_reserve_block(inode, *count); 2309 if (ret) 2310 return ret; 2311 2312 if (time_to_inject(sbi, FAULT_BLOCK)) { 2313 release = *count; 2314 goto release_quota; 2315 } 2316 2317 /* 2318 * let's increase this in prior to actual block count change in order 2319 * for f2fs_sync_file to avoid data races when deciding checkpoint. 2320 */ 2321 percpu_counter_add(&sbi->alloc_valid_block_count, (*count)); 2322 2323 spin_lock(&sbi->stat_lock); 2324 2325 avail_user_block_count = get_available_block_count(sbi, inode, true); 2326 diff = (long long)sbi->total_valid_block_count + *count - 2327 avail_user_block_count; 2328 if (unlikely(diff > 0)) { 2329 if (!partial) { 2330 spin_unlock(&sbi->stat_lock); 2331 release = *count; 2332 goto enospc; 2333 } 2334 if (diff > *count) 2335 diff = *count; 2336 *count -= diff; 2337 release = diff; 2338 if (!*count) { 2339 spin_unlock(&sbi->stat_lock); 2340 goto enospc; 2341 } 2342 } 2343 sbi->total_valid_block_count += (block_t)(*count); 2344 2345 spin_unlock(&sbi->stat_lock); 2346 2347 if (unlikely(release)) { 2348 percpu_counter_sub(&sbi->alloc_valid_block_count, release); 2349 dquot_release_reservation_block(inode, release); 2350 } 2351 f2fs_i_blocks_write(inode, *count, true, true); 2352 return 0; 2353 2354 enospc: 2355 percpu_counter_sub(&sbi->alloc_valid_block_count, release); 2356 release_quota: 2357 dquot_release_reservation_block(inode, release); 2358 return -ENOSPC; 2359 } 2360 2361 #define PAGE_PRIVATE_GET_FUNC(name, flagname) \ 2362 static inline bool page_private_##name(struct page *page) \ 2363 { \ 2364 return PagePrivate(page) && \ 2365 test_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)) && \ 2366 test_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \ 2367 } 2368 2369 #define PAGE_PRIVATE_SET_FUNC(name, flagname) \ 2370 static inline void set_page_private_##name(struct page *page) \ 2371 { \ 2372 if (!PagePrivate(page)) \ 2373 attach_page_private(page, (void *)0); \ 2374 set_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)); \ 2375 set_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \ 2376 } 2377 2378 #define PAGE_PRIVATE_CLEAR_FUNC(name, flagname) \ 2379 static inline void clear_page_private_##name(struct page *page) \ 2380 { \ 2381 clear_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \ 2382 if (page_private(page) == BIT(PAGE_PRIVATE_NOT_POINTER)) \ 2383 detach_page_private(page); \ 2384 } 2385 2386 PAGE_PRIVATE_GET_FUNC(nonpointer, NOT_POINTER); 2387 PAGE_PRIVATE_GET_FUNC(inline, INLINE_INODE); 2388 PAGE_PRIVATE_GET_FUNC(gcing, ONGOING_MIGRATION); 2389 2390 PAGE_PRIVATE_SET_FUNC(reference, REF_RESOURCE); 2391 PAGE_PRIVATE_SET_FUNC(inline, INLINE_INODE); 2392 PAGE_PRIVATE_SET_FUNC(gcing, ONGOING_MIGRATION); 2393 2394 PAGE_PRIVATE_CLEAR_FUNC(reference, REF_RESOURCE); 2395 PAGE_PRIVATE_CLEAR_FUNC(inline, INLINE_INODE); 2396 PAGE_PRIVATE_CLEAR_FUNC(gcing, ONGOING_MIGRATION); 2397 2398 static inline unsigned long get_page_private_data(struct page *page) 2399 { 2400 unsigned long data = page_private(page); 2401 2402 if (!test_bit(PAGE_PRIVATE_NOT_POINTER, &data)) 2403 return 0; 2404 return data >> PAGE_PRIVATE_MAX; 2405 } 2406 2407 static inline void set_page_private_data(struct page *page, unsigned long data) 2408 { 2409 if (!PagePrivate(page)) 2410 attach_page_private(page, (void *)0); 2411 set_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)); 2412 page_private(page) |= data << PAGE_PRIVATE_MAX; 2413 } 2414 2415 static inline void clear_page_private_data(struct page *page) 2416 { 2417 page_private(page) &= GENMASK(PAGE_PRIVATE_MAX - 1, 0); 2418 if (page_private(page) == BIT(PAGE_PRIVATE_NOT_POINTER)) 2419 detach_page_private(page); 2420 } 2421 2422 static inline void clear_page_private_all(struct page *page) 2423 { 2424 clear_page_private_data(page); 2425 clear_page_private_reference(page); 2426 clear_page_private_gcing(page); 2427 clear_page_private_inline(page); 2428 2429 f2fs_bug_on(F2FS_P_SB(page), page_private(page)); 2430 } 2431 2432 static inline void dec_valid_block_count(struct f2fs_sb_info *sbi, 2433 struct inode *inode, 2434 block_t count) 2435 { 2436 blkcnt_t sectors = count << F2FS_LOG_SECTORS_PER_BLOCK; 2437 2438 spin_lock(&sbi->stat_lock); 2439 f2fs_bug_on(sbi, sbi->total_valid_block_count < (block_t) count); 2440 sbi->total_valid_block_count -= (block_t)count; 2441 if (sbi->reserved_blocks && 2442 sbi->current_reserved_blocks < sbi->reserved_blocks) 2443 sbi->current_reserved_blocks = min(sbi->reserved_blocks, 2444 sbi->current_reserved_blocks + count); 2445 spin_unlock(&sbi->stat_lock); 2446 if (unlikely(inode->i_blocks < sectors)) { 2447 f2fs_warn(sbi, "Inconsistent i_blocks, ino:%lu, iblocks:%llu, sectors:%llu", 2448 inode->i_ino, 2449 (unsigned long long)inode->i_blocks, 2450 (unsigned long long)sectors); 2451 set_sbi_flag(sbi, SBI_NEED_FSCK); 2452 return; 2453 } 2454 f2fs_i_blocks_write(inode, count, false, true); 2455 } 2456 2457 static inline void inc_page_count(struct f2fs_sb_info *sbi, int count_type) 2458 { 2459 atomic_inc(&sbi->nr_pages[count_type]); 2460 2461 if (count_type == F2FS_DIRTY_DENTS || 2462 count_type == F2FS_DIRTY_NODES || 2463 count_type == F2FS_DIRTY_META || 2464 count_type == F2FS_DIRTY_QDATA || 2465 count_type == F2FS_DIRTY_IMETA) 2466 set_sbi_flag(sbi, SBI_IS_DIRTY); 2467 } 2468 2469 static inline void inode_inc_dirty_pages(struct inode *inode) 2470 { 2471 atomic_inc(&F2FS_I(inode)->dirty_pages); 2472 inc_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ? 2473 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA); 2474 if (IS_NOQUOTA(inode)) 2475 inc_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA); 2476 } 2477 2478 static inline void dec_page_count(struct f2fs_sb_info *sbi, int count_type) 2479 { 2480 atomic_dec(&sbi->nr_pages[count_type]); 2481 } 2482 2483 static inline void inode_dec_dirty_pages(struct inode *inode) 2484 { 2485 if (!S_ISDIR(inode->i_mode) && !S_ISREG(inode->i_mode) && 2486 !S_ISLNK(inode->i_mode)) 2487 return; 2488 2489 atomic_dec(&F2FS_I(inode)->dirty_pages); 2490 dec_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ? 2491 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA); 2492 if (IS_NOQUOTA(inode)) 2493 dec_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA); 2494 } 2495 2496 static inline void inc_atomic_write_cnt(struct inode *inode) 2497 { 2498 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 2499 struct f2fs_inode_info *fi = F2FS_I(inode); 2500 u64 current_write; 2501 2502 fi->atomic_write_cnt++; 2503 atomic64_inc(&sbi->current_atomic_write); 2504 current_write = atomic64_read(&sbi->current_atomic_write); 2505 if (current_write > sbi->peak_atomic_write) 2506 sbi->peak_atomic_write = current_write; 2507 } 2508 2509 static inline void release_atomic_write_cnt(struct inode *inode) 2510 { 2511 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 2512 struct f2fs_inode_info *fi = F2FS_I(inode); 2513 2514 atomic64_sub(fi->atomic_write_cnt, &sbi->current_atomic_write); 2515 fi->atomic_write_cnt = 0; 2516 } 2517 2518 static inline s64 get_pages(struct f2fs_sb_info *sbi, int count_type) 2519 { 2520 return atomic_read(&sbi->nr_pages[count_type]); 2521 } 2522 2523 static inline int get_dirty_pages(struct inode *inode) 2524 { 2525 return atomic_read(&F2FS_I(inode)->dirty_pages); 2526 } 2527 2528 static inline int get_blocktype_secs(struct f2fs_sb_info *sbi, int block_type) 2529 { 2530 return div_u64(get_pages(sbi, block_type) + BLKS_PER_SEC(sbi) - 1, 2531 BLKS_PER_SEC(sbi)); 2532 } 2533 2534 static inline block_t valid_user_blocks(struct f2fs_sb_info *sbi) 2535 { 2536 return sbi->total_valid_block_count; 2537 } 2538 2539 static inline block_t discard_blocks(struct f2fs_sb_info *sbi) 2540 { 2541 return sbi->discard_blks; 2542 } 2543 2544 static inline unsigned long __bitmap_size(struct f2fs_sb_info *sbi, int flag) 2545 { 2546 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi); 2547 2548 /* return NAT or SIT bitmap */ 2549 if (flag == NAT_BITMAP) 2550 return le32_to_cpu(ckpt->nat_ver_bitmap_bytesize); 2551 else if (flag == SIT_BITMAP) 2552 return le32_to_cpu(ckpt->sit_ver_bitmap_bytesize); 2553 2554 return 0; 2555 } 2556 2557 static inline block_t __cp_payload(struct f2fs_sb_info *sbi) 2558 { 2559 return le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_payload); 2560 } 2561 2562 static inline void *__bitmap_ptr(struct f2fs_sb_info *sbi, int flag) 2563 { 2564 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi); 2565 void *tmp_ptr = &ckpt->sit_nat_version_bitmap; 2566 int offset; 2567 2568 if (is_set_ckpt_flags(sbi, CP_LARGE_NAT_BITMAP_FLAG)) { 2569 offset = (flag == SIT_BITMAP) ? 2570 le32_to_cpu(ckpt->nat_ver_bitmap_bytesize) : 0; 2571 /* 2572 * if large_nat_bitmap feature is enabled, leave checksum 2573 * protection for all nat/sit bitmaps. 2574 */ 2575 return tmp_ptr + offset + sizeof(__le32); 2576 } 2577 2578 if (__cp_payload(sbi) > 0) { 2579 if (flag == NAT_BITMAP) 2580 return tmp_ptr; 2581 else 2582 return (unsigned char *)ckpt + F2FS_BLKSIZE; 2583 } else { 2584 offset = (flag == NAT_BITMAP) ? 2585 le32_to_cpu(ckpt->sit_ver_bitmap_bytesize) : 0; 2586 return tmp_ptr + offset; 2587 } 2588 } 2589 2590 static inline block_t __start_cp_addr(struct f2fs_sb_info *sbi) 2591 { 2592 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr); 2593 2594 if (sbi->cur_cp_pack == 2) 2595 start_addr += BLKS_PER_SEG(sbi); 2596 return start_addr; 2597 } 2598 2599 static inline block_t __start_cp_next_addr(struct f2fs_sb_info *sbi) 2600 { 2601 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr); 2602 2603 if (sbi->cur_cp_pack == 1) 2604 start_addr += BLKS_PER_SEG(sbi); 2605 return start_addr; 2606 } 2607 2608 static inline void __set_cp_next_pack(struct f2fs_sb_info *sbi) 2609 { 2610 sbi->cur_cp_pack = (sbi->cur_cp_pack == 1) ? 2 : 1; 2611 } 2612 2613 static inline block_t __start_sum_addr(struct f2fs_sb_info *sbi) 2614 { 2615 return le32_to_cpu(F2FS_CKPT(sbi)->cp_pack_start_sum); 2616 } 2617 2618 extern void f2fs_mark_inode_dirty_sync(struct inode *inode, bool sync); 2619 static inline int inc_valid_node_count(struct f2fs_sb_info *sbi, 2620 struct inode *inode, bool is_inode) 2621 { 2622 block_t valid_block_count; 2623 unsigned int valid_node_count; 2624 unsigned int avail_user_block_count; 2625 int err; 2626 2627 if (is_inode) { 2628 if (inode) { 2629 err = dquot_alloc_inode(inode); 2630 if (err) 2631 return err; 2632 } 2633 } else { 2634 err = dquot_reserve_block(inode, 1); 2635 if (err) 2636 return err; 2637 } 2638 2639 if (time_to_inject(sbi, FAULT_BLOCK)) 2640 goto enospc; 2641 2642 spin_lock(&sbi->stat_lock); 2643 2644 valid_block_count = sbi->total_valid_block_count + 1; 2645 avail_user_block_count = get_available_block_count(sbi, inode, false); 2646 2647 if (unlikely(valid_block_count > avail_user_block_count)) { 2648 spin_unlock(&sbi->stat_lock); 2649 goto enospc; 2650 } 2651 2652 valid_node_count = sbi->total_valid_node_count + 1; 2653 if (unlikely(valid_node_count > sbi->total_node_count)) { 2654 spin_unlock(&sbi->stat_lock); 2655 goto enospc; 2656 } 2657 2658 sbi->total_valid_node_count++; 2659 sbi->total_valid_block_count++; 2660 spin_unlock(&sbi->stat_lock); 2661 2662 if (inode) { 2663 if (is_inode) 2664 f2fs_mark_inode_dirty_sync(inode, true); 2665 else 2666 f2fs_i_blocks_write(inode, 1, true, true); 2667 } 2668 2669 percpu_counter_inc(&sbi->alloc_valid_block_count); 2670 return 0; 2671 2672 enospc: 2673 if (is_inode) { 2674 if (inode) 2675 dquot_free_inode(inode); 2676 } else { 2677 dquot_release_reservation_block(inode, 1); 2678 } 2679 return -ENOSPC; 2680 } 2681 2682 static inline void dec_valid_node_count(struct f2fs_sb_info *sbi, 2683 struct inode *inode, bool is_inode) 2684 { 2685 spin_lock(&sbi->stat_lock); 2686 2687 if (unlikely(!sbi->total_valid_block_count || 2688 !sbi->total_valid_node_count)) { 2689 f2fs_warn(sbi, "dec_valid_node_count: inconsistent block counts, total_valid_block:%u, total_valid_node:%u", 2690 sbi->total_valid_block_count, 2691 sbi->total_valid_node_count); 2692 set_sbi_flag(sbi, SBI_NEED_FSCK); 2693 } else { 2694 sbi->total_valid_block_count--; 2695 sbi->total_valid_node_count--; 2696 } 2697 2698 if (sbi->reserved_blocks && 2699 sbi->current_reserved_blocks < sbi->reserved_blocks) 2700 sbi->current_reserved_blocks++; 2701 2702 spin_unlock(&sbi->stat_lock); 2703 2704 if (is_inode) { 2705 dquot_free_inode(inode); 2706 } else { 2707 if (unlikely(inode->i_blocks == 0)) { 2708 f2fs_warn(sbi, "dec_valid_node_count: inconsistent i_blocks, ino:%lu, iblocks:%llu", 2709 inode->i_ino, 2710 (unsigned long long)inode->i_blocks); 2711 set_sbi_flag(sbi, SBI_NEED_FSCK); 2712 return; 2713 } 2714 f2fs_i_blocks_write(inode, 1, false, true); 2715 } 2716 } 2717 2718 static inline unsigned int valid_node_count(struct f2fs_sb_info *sbi) 2719 { 2720 return sbi->total_valid_node_count; 2721 } 2722 2723 static inline void inc_valid_inode_count(struct f2fs_sb_info *sbi) 2724 { 2725 percpu_counter_inc(&sbi->total_valid_inode_count); 2726 } 2727 2728 static inline void dec_valid_inode_count(struct f2fs_sb_info *sbi) 2729 { 2730 percpu_counter_dec(&sbi->total_valid_inode_count); 2731 } 2732 2733 static inline s64 valid_inode_count(struct f2fs_sb_info *sbi) 2734 { 2735 return percpu_counter_sum_positive(&sbi->total_valid_inode_count); 2736 } 2737 2738 static inline struct page *f2fs_grab_cache_page(struct address_space *mapping, 2739 pgoff_t index, bool for_write) 2740 { 2741 struct page *page; 2742 unsigned int flags; 2743 2744 if (IS_ENABLED(CONFIG_F2FS_FAULT_INJECTION)) { 2745 if (!for_write) 2746 page = find_get_page_flags(mapping, index, 2747 FGP_LOCK | FGP_ACCESSED); 2748 else 2749 page = find_lock_page(mapping, index); 2750 if (page) 2751 return page; 2752 2753 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_ALLOC)) 2754 return NULL; 2755 } 2756 2757 if (!for_write) 2758 return grab_cache_page(mapping, index); 2759 2760 flags = memalloc_nofs_save(); 2761 page = grab_cache_page_write_begin(mapping, index); 2762 memalloc_nofs_restore(flags); 2763 2764 return page; 2765 } 2766 2767 static inline struct page *f2fs_pagecache_get_page( 2768 struct address_space *mapping, pgoff_t index, 2769 fgf_t fgp_flags, gfp_t gfp_mask) 2770 { 2771 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_GET)) 2772 return NULL; 2773 2774 return pagecache_get_page(mapping, index, fgp_flags, gfp_mask); 2775 } 2776 2777 static inline void f2fs_put_page(struct page *page, int unlock) 2778 { 2779 if (!page) 2780 return; 2781 2782 if (unlock) { 2783 f2fs_bug_on(F2FS_P_SB(page), !PageLocked(page)); 2784 unlock_page(page); 2785 } 2786 put_page(page); 2787 } 2788 2789 static inline void f2fs_put_dnode(struct dnode_of_data *dn) 2790 { 2791 if (dn->node_page) 2792 f2fs_put_page(dn->node_page, 1); 2793 if (dn->inode_page && dn->node_page != dn->inode_page) 2794 f2fs_put_page(dn->inode_page, 0); 2795 dn->node_page = NULL; 2796 dn->inode_page = NULL; 2797 } 2798 2799 static inline struct kmem_cache *f2fs_kmem_cache_create(const char *name, 2800 size_t size) 2801 { 2802 return kmem_cache_create(name, size, 0, SLAB_RECLAIM_ACCOUNT, NULL); 2803 } 2804 2805 static inline void *f2fs_kmem_cache_alloc_nofail(struct kmem_cache *cachep, 2806 gfp_t flags) 2807 { 2808 void *entry; 2809 2810 entry = kmem_cache_alloc(cachep, flags); 2811 if (!entry) 2812 entry = kmem_cache_alloc(cachep, flags | __GFP_NOFAIL); 2813 return entry; 2814 } 2815 2816 static inline void *f2fs_kmem_cache_alloc(struct kmem_cache *cachep, 2817 gfp_t flags, bool nofail, struct f2fs_sb_info *sbi) 2818 { 2819 if (nofail) 2820 return f2fs_kmem_cache_alloc_nofail(cachep, flags); 2821 2822 if (time_to_inject(sbi, FAULT_SLAB_ALLOC)) 2823 return NULL; 2824 2825 return kmem_cache_alloc(cachep, flags); 2826 } 2827 2828 static inline bool is_inflight_io(struct f2fs_sb_info *sbi, int type) 2829 { 2830 if (get_pages(sbi, F2FS_RD_DATA) || get_pages(sbi, F2FS_RD_NODE) || 2831 get_pages(sbi, F2FS_RD_META) || get_pages(sbi, F2FS_WB_DATA) || 2832 get_pages(sbi, F2FS_WB_CP_DATA) || 2833 get_pages(sbi, F2FS_DIO_READ) || 2834 get_pages(sbi, F2FS_DIO_WRITE)) 2835 return true; 2836 2837 if (type != DISCARD_TIME && SM_I(sbi) && SM_I(sbi)->dcc_info && 2838 atomic_read(&SM_I(sbi)->dcc_info->queued_discard)) 2839 return true; 2840 2841 if (SM_I(sbi) && SM_I(sbi)->fcc_info && 2842 atomic_read(&SM_I(sbi)->fcc_info->queued_flush)) 2843 return true; 2844 return false; 2845 } 2846 2847 static inline bool is_idle(struct f2fs_sb_info *sbi, int type) 2848 { 2849 if (sbi->gc_mode == GC_URGENT_HIGH) 2850 return true; 2851 2852 if (is_inflight_io(sbi, type)) 2853 return false; 2854 2855 if (sbi->gc_mode == GC_URGENT_MID) 2856 return true; 2857 2858 if (sbi->gc_mode == GC_URGENT_LOW && 2859 (type == DISCARD_TIME || type == GC_TIME)) 2860 return true; 2861 2862 return f2fs_time_over(sbi, type); 2863 } 2864 2865 static inline void f2fs_radix_tree_insert(struct radix_tree_root *root, 2866 unsigned long index, void *item) 2867 { 2868 while (radix_tree_insert(root, index, item)) 2869 cond_resched(); 2870 } 2871 2872 #define RAW_IS_INODE(p) ((p)->footer.nid == (p)->footer.ino) 2873 2874 static inline bool IS_INODE(struct page *page) 2875 { 2876 struct f2fs_node *p = F2FS_NODE(page); 2877 2878 return RAW_IS_INODE(p); 2879 } 2880 2881 static inline int offset_in_addr(struct f2fs_inode *i) 2882 { 2883 return (i->i_inline & F2FS_EXTRA_ATTR) ? 2884 (le16_to_cpu(i->i_extra_isize) / sizeof(__le32)) : 0; 2885 } 2886 2887 static inline __le32 *blkaddr_in_node(struct f2fs_node *node) 2888 { 2889 return RAW_IS_INODE(node) ? node->i.i_addr : node->dn.addr; 2890 } 2891 2892 static inline int f2fs_has_extra_attr(struct inode *inode); 2893 static inline block_t data_blkaddr(struct inode *inode, 2894 struct page *node_page, unsigned int offset) 2895 { 2896 struct f2fs_node *raw_node; 2897 __le32 *addr_array; 2898 int base = 0; 2899 bool is_inode = IS_INODE(node_page); 2900 2901 raw_node = F2FS_NODE(node_page); 2902 2903 if (is_inode) { 2904 if (!inode) 2905 /* from GC path only */ 2906 base = offset_in_addr(&raw_node->i); 2907 else if (f2fs_has_extra_attr(inode)) 2908 base = get_extra_isize(inode); 2909 } 2910 2911 addr_array = blkaddr_in_node(raw_node); 2912 return le32_to_cpu(addr_array[base + offset]); 2913 } 2914 2915 static inline block_t f2fs_data_blkaddr(struct dnode_of_data *dn) 2916 { 2917 return data_blkaddr(dn->inode, dn->node_page, dn->ofs_in_node); 2918 } 2919 2920 static inline int f2fs_test_bit(unsigned int nr, char *addr) 2921 { 2922 int mask; 2923 2924 addr += (nr >> 3); 2925 mask = BIT(7 - (nr & 0x07)); 2926 return mask & *addr; 2927 } 2928 2929 static inline void f2fs_set_bit(unsigned int nr, char *addr) 2930 { 2931 int mask; 2932 2933 addr += (nr >> 3); 2934 mask = BIT(7 - (nr & 0x07)); 2935 *addr |= mask; 2936 } 2937 2938 static inline void f2fs_clear_bit(unsigned int nr, char *addr) 2939 { 2940 int mask; 2941 2942 addr += (nr >> 3); 2943 mask = BIT(7 - (nr & 0x07)); 2944 *addr &= ~mask; 2945 } 2946 2947 static inline int f2fs_test_and_set_bit(unsigned int nr, char *addr) 2948 { 2949 int mask; 2950 int ret; 2951 2952 addr += (nr >> 3); 2953 mask = BIT(7 - (nr & 0x07)); 2954 ret = mask & *addr; 2955 *addr |= mask; 2956 return ret; 2957 } 2958 2959 static inline int f2fs_test_and_clear_bit(unsigned int nr, char *addr) 2960 { 2961 int mask; 2962 int ret; 2963 2964 addr += (nr >> 3); 2965 mask = BIT(7 - (nr & 0x07)); 2966 ret = mask & *addr; 2967 *addr &= ~mask; 2968 return ret; 2969 } 2970 2971 static inline void f2fs_change_bit(unsigned int nr, char *addr) 2972 { 2973 int mask; 2974 2975 addr += (nr >> 3); 2976 mask = BIT(7 - (nr & 0x07)); 2977 *addr ^= mask; 2978 } 2979 2980 /* 2981 * On-disk inode flags (f2fs_inode::i_flags) 2982 */ 2983 #define F2FS_COMPR_FL 0x00000004 /* Compress file */ 2984 #define F2FS_SYNC_FL 0x00000008 /* Synchronous updates */ 2985 #define F2FS_IMMUTABLE_FL 0x00000010 /* Immutable file */ 2986 #define F2FS_APPEND_FL 0x00000020 /* writes to file may only append */ 2987 #define F2FS_NODUMP_FL 0x00000040 /* do not dump file */ 2988 #define F2FS_NOATIME_FL 0x00000080 /* do not update atime */ 2989 #define F2FS_NOCOMP_FL 0x00000400 /* Don't compress */ 2990 #define F2FS_INDEX_FL 0x00001000 /* hash-indexed directory */ 2991 #define F2FS_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */ 2992 #define F2FS_PROJINHERIT_FL 0x20000000 /* Create with parents projid */ 2993 #define F2FS_CASEFOLD_FL 0x40000000 /* Casefolded file */ 2994 2995 #define F2FS_QUOTA_DEFAULT_FL (F2FS_NOATIME_FL | F2FS_IMMUTABLE_FL) 2996 2997 /* Flags that should be inherited by new inodes from their parent. */ 2998 #define F2FS_FL_INHERITED (F2FS_SYNC_FL | F2FS_NODUMP_FL | F2FS_NOATIME_FL | \ 2999 F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \ 3000 F2FS_CASEFOLD_FL) 3001 3002 /* Flags that are appropriate for regular files (all but dir-specific ones). */ 3003 #define F2FS_REG_FLMASK (~(F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \ 3004 F2FS_CASEFOLD_FL)) 3005 3006 /* Flags that are appropriate for non-directories/regular files. */ 3007 #define F2FS_OTHER_FLMASK (F2FS_NODUMP_FL | F2FS_NOATIME_FL) 3008 3009 static inline __u32 f2fs_mask_flags(umode_t mode, __u32 flags) 3010 { 3011 if (S_ISDIR(mode)) 3012 return flags; 3013 else if (S_ISREG(mode)) 3014 return flags & F2FS_REG_FLMASK; 3015 else 3016 return flags & F2FS_OTHER_FLMASK; 3017 } 3018 3019 static inline void __mark_inode_dirty_flag(struct inode *inode, 3020 int flag, bool set) 3021 { 3022 switch (flag) { 3023 case FI_INLINE_XATTR: 3024 case FI_INLINE_DATA: 3025 case FI_INLINE_DENTRY: 3026 case FI_NEW_INODE: 3027 if (set) 3028 return; 3029 fallthrough; 3030 case FI_DATA_EXIST: 3031 case FI_INLINE_DOTS: 3032 case FI_PIN_FILE: 3033 case FI_COMPRESS_RELEASED: 3034 case FI_ATOMIC_COMMITTED: 3035 f2fs_mark_inode_dirty_sync(inode, true); 3036 } 3037 } 3038 3039 static inline void set_inode_flag(struct inode *inode, int flag) 3040 { 3041 set_bit(flag, F2FS_I(inode)->flags); 3042 __mark_inode_dirty_flag(inode, flag, true); 3043 } 3044 3045 static inline int is_inode_flag_set(struct inode *inode, int flag) 3046 { 3047 return test_bit(flag, F2FS_I(inode)->flags); 3048 } 3049 3050 static inline void clear_inode_flag(struct inode *inode, int flag) 3051 { 3052 clear_bit(flag, F2FS_I(inode)->flags); 3053 __mark_inode_dirty_flag(inode, flag, false); 3054 } 3055 3056 static inline bool f2fs_verity_in_progress(struct inode *inode) 3057 { 3058 return IS_ENABLED(CONFIG_FS_VERITY) && 3059 is_inode_flag_set(inode, FI_VERITY_IN_PROGRESS); 3060 } 3061 3062 static inline void set_acl_inode(struct inode *inode, umode_t mode) 3063 { 3064 F2FS_I(inode)->i_acl_mode = mode; 3065 set_inode_flag(inode, FI_ACL_MODE); 3066 f2fs_mark_inode_dirty_sync(inode, false); 3067 } 3068 3069 static inline void f2fs_i_links_write(struct inode *inode, bool inc) 3070 { 3071 if (inc) 3072 inc_nlink(inode); 3073 else 3074 drop_nlink(inode); 3075 f2fs_mark_inode_dirty_sync(inode, true); 3076 } 3077 3078 static inline void f2fs_i_blocks_write(struct inode *inode, 3079 block_t diff, bool add, bool claim) 3080 { 3081 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE); 3082 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER); 3083 3084 /* add = 1, claim = 1 should be dquot_reserve_block in pair */ 3085 if (add) { 3086 if (claim) 3087 dquot_claim_block(inode, diff); 3088 else 3089 dquot_alloc_block_nofail(inode, diff); 3090 } else { 3091 dquot_free_block(inode, diff); 3092 } 3093 3094 f2fs_mark_inode_dirty_sync(inode, true); 3095 if (clean || recover) 3096 set_inode_flag(inode, FI_AUTO_RECOVER); 3097 } 3098 3099 static inline bool f2fs_is_atomic_file(struct inode *inode); 3100 3101 static inline void f2fs_i_size_write(struct inode *inode, loff_t i_size) 3102 { 3103 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE); 3104 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER); 3105 3106 if (i_size_read(inode) == i_size) 3107 return; 3108 3109 i_size_write(inode, i_size); 3110 3111 if (f2fs_is_atomic_file(inode)) 3112 return; 3113 3114 f2fs_mark_inode_dirty_sync(inode, true); 3115 if (clean || recover) 3116 set_inode_flag(inode, FI_AUTO_RECOVER); 3117 } 3118 3119 static inline void f2fs_i_depth_write(struct inode *inode, unsigned int depth) 3120 { 3121 F2FS_I(inode)->i_current_depth = depth; 3122 f2fs_mark_inode_dirty_sync(inode, true); 3123 } 3124 3125 static inline void f2fs_i_gc_failures_write(struct inode *inode, 3126 unsigned int count) 3127 { 3128 F2FS_I(inode)->i_gc_failures[GC_FAILURE_PIN] = count; 3129 f2fs_mark_inode_dirty_sync(inode, true); 3130 } 3131 3132 static inline void f2fs_i_xnid_write(struct inode *inode, nid_t xnid) 3133 { 3134 F2FS_I(inode)->i_xattr_nid = xnid; 3135 f2fs_mark_inode_dirty_sync(inode, true); 3136 } 3137 3138 static inline void f2fs_i_pino_write(struct inode *inode, nid_t pino) 3139 { 3140 F2FS_I(inode)->i_pino = pino; 3141 f2fs_mark_inode_dirty_sync(inode, true); 3142 } 3143 3144 static inline void get_inline_info(struct inode *inode, struct f2fs_inode *ri) 3145 { 3146 struct f2fs_inode_info *fi = F2FS_I(inode); 3147 3148 if (ri->i_inline & F2FS_INLINE_XATTR) 3149 set_bit(FI_INLINE_XATTR, fi->flags); 3150 if (ri->i_inline & F2FS_INLINE_DATA) 3151 set_bit(FI_INLINE_DATA, fi->flags); 3152 if (ri->i_inline & F2FS_INLINE_DENTRY) 3153 set_bit(FI_INLINE_DENTRY, fi->flags); 3154 if (ri->i_inline & F2FS_DATA_EXIST) 3155 set_bit(FI_DATA_EXIST, fi->flags); 3156 if (ri->i_inline & F2FS_INLINE_DOTS) 3157 set_bit(FI_INLINE_DOTS, fi->flags); 3158 if (ri->i_inline & F2FS_EXTRA_ATTR) 3159 set_bit(FI_EXTRA_ATTR, fi->flags); 3160 if (ri->i_inline & F2FS_PIN_FILE) 3161 set_bit(FI_PIN_FILE, fi->flags); 3162 if (ri->i_inline & F2FS_COMPRESS_RELEASED) 3163 set_bit(FI_COMPRESS_RELEASED, fi->flags); 3164 } 3165 3166 static inline void set_raw_inline(struct inode *inode, struct f2fs_inode *ri) 3167 { 3168 ri->i_inline = 0; 3169 3170 if (is_inode_flag_set(inode, FI_INLINE_XATTR)) 3171 ri->i_inline |= F2FS_INLINE_XATTR; 3172 if (is_inode_flag_set(inode, FI_INLINE_DATA)) 3173 ri->i_inline |= F2FS_INLINE_DATA; 3174 if (is_inode_flag_set(inode, FI_INLINE_DENTRY)) 3175 ri->i_inline |= F2FS_INLINE_DENTRY; 3176 if (is_inode_flag_set(inode, FI_DATA_EXIST)) 3177 ri->i_inline |= F2FS_DATA_EXIST; 3178 if (is_inode_flag_set(inode, FI_INLINE_DOTS)) 3179 ri->i_inline |= F2FS_INLINE_DOTS; 3180 if (is_inode_flag_set(inode, FI_EXTRA_ATTR)) 3181 ri->i_inline |= F2FS_EXTRA_ATTR; 3182 if (is_inode_flag_set(inode, FI_PIN_FILE)) 3183 ri->i_inline |= F2FS_PIN_FILE; 3184 if (is_inode_flag_set(inode, FI_COMPRESS_RELEASED)) 3185 ri->i_inline |= F2FS_COMPRESS_RELEASED; 3186 } 3187 3188 static inline int f2fs_has_extra_attr(struct inode *inode) 3189 { 3190 return is_inode_flag_set(inode, FI_EXTRA_ATTR); 3191 } 3192 3193 static inline int f2fs_has_inline_xattr(struct inode *inode) 3194 { 3195 return is_inode_flag_set(inode, FI_INLINE_XATTR); 3196 } 3197 3198 static inline int f2fs_compressed_file(struct inode *inode) 3199 { 3200 return S_ISREG(inode->i_mode) && 3201 is_inode_flag_set(inode, FI_COMPRESSED_FILE); 3202 } 3203 3204 static inline bool f2fs_need_compress_data(struct inode *inode) 3205 { 3206 int compress_mode = F2FS_OPTION(F2FS_I_SB(inode)).compress_mode; 3207 3208 if (!f2fs_compressed_file(inode)) 3209 return false; 3210 3211 if (compress_mode == COMPR_MODE_FS) 3212 return true; 3213 else if (compress_mode == COMPR_MODE_USER && 3214 is_inode_flag_set(inode, FI_ENABLE_COMPRESS)) 3215 return true; 3216 3217 return false; 3218 } 3219 3220 static inline unsigned int addrs_per_inode(struct inode *inode) 3221 { 3222 unsigned int addrs = CUR_ADDRS_PER_INODE(inode) - 3223 get_inline_xattr_addrs(inode); 3224 3225 if (!f2fs_compressed_file(inode)) 3226 return addrs; 3227 return ALIGN_DOWN(addrs, F2FS_I(inode)->i_cluster_size); 3228 } 3229 3230 static inline unsigned int addrs_per_block(struct inode *inode) 3231 { 3232 if (!f2fs_compressed_file(inode)) 3233 return DEF_ADDRS_PER_BLOCK; 3234 return ALIGN_DOWN(DEF_ADDRS_PER_BLOCK, F2FS_I(inode)->i_cluster_size); 3235 } 3236 3237 static inline void *inline_xattr_addr(struct inode *inode, struct page *page) 3238 { 3239 struct f2fs_inode *ri = F2FS_INODE(page); 3240 3241 return (void *)&(ri->i_addr[DEF_ADDRS_PER_INODE - 3242 get_inline_xattr_addrs(inode)]); 3243 } 3244 3245 static inline int inline_xattr_size(struct inode *inode) 3246 { 3247 if (f2fs_has_inline_xattr(inode)) 3248 return get_inline_xattr_addrs(inode) * sizeof(__le32); 3249 return 0; 3250 } 3251 3252 /* 3253 * Notice: check inline_data flag without inode page lock is unsafe. 3254 * It could change at any time by f2fs_convert_inline_page(). 3255 */ 3256 static inline int f2fs_has_inline_data(struct inode *inode) 3257 { 3258 return is_inode_flag_set(inode, FI_INLINE_DATA); 3259 } 3260 3261 static inline int f2fs_exist_data(struct inode *inode) 3262 { 3263 return is_inode_flag_set(inode, FI_DATA_EXIST); 3264 } 3265 3266 static inline int f2fs_has_inline_dots(struct inode *inode) 3267 { 3268 return is_inode_flag_set(inode, FI_INLINE_DOTS); 3269 } 3270 3271 static inline int f2fs_is_mmap_file(struct inode *inode) 3272 { 3273 return is_inode_flag_set(inode, FI_MMAP_FILE); 3274 } 3275 3276 static inline bool f2fs_is_pinned_file(struct inode *inode) 3277 { 3278 return is_inode_flag_set(inode, FI_PIN_FILE); 3279 } 3280 3281 static inline bool f2fs_is_atomic_file(struct inode *inode) 3282 { 3283 return is_inode_flag_set(inode, FI_ATOMIC_FILE); 3284 } 3285 3286 static inline bool f2fs_is_cow_file(struct inode *inode) 3287 { 3288 return is_inode_flag_set(inode, FI_COW_FILE); 3289 } 3290 3291 static inline __le32 *get_dnode_addr(struct inode *inode, 3292 struct page *node_page); 3293 static inline void *inline_data_addr(struct inode *inode, struct page *page) 3294 { 3295 __le32 *addr = get_dnode_addr(inode, page); 3296 3297 return (void *)(addr + DEF_INLINE_RESERVED_SIZE); 3298 } 3299 3300 static inline int f2fs_has_inline_dentry(struct inode *inode) 3301 { 3302 return is_inode_flag_set(inode, FI_INLINE_DENTRY); 3303 } 3304 3305 static inline int is_file(struct inode *inode, int type) 3306 { 3307 return F2FS_I(inode)->i_advise & type; 3308 } 3309 3310 static inline void set_file(struct inode *inode, int type) 3311 { 3312 if (is_file(inode, type)) 3313 return; 3314 F2FS_I(inode)->i_advise |= type; 3315 f2fs_mark_inode_dirty_sync(inode, true); 3316 } 3317 3318 static inline void clear_file(struct inode *inode, int type) 3319 { 3320 if (!is_file(inode, type)) 3321 return; 3322 F2FS_I(inode)->i_advise &= ~type; 3323 f2fs_mark_inode_dirty_sync(inode, true); 3324 } 3325 3326 static inline bool f2fs_is_time_consistent(struct inode *inode) 3327 { 3328 struct timespec64 ctime = inode_get_ctime(inode); 3329 3330 if (!timespec64_equal(F2FS_I(inode)->i_disk_time, &inode->i_atime)) 3331 return false; 3332 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 1, &ctime)) 3333 return false; 3334 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 2, &inode->i_mtime)) 3335 return false; 3336 return true; 3337 } 3338 3339 static inline bool f2fs_skip_inode_update(struct inode *inode, int dsync) 3340 { 3341 bool ret; 3342 3343 if (dsync) { 3344 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 3345 3346 spin_lock(&sbi->inode_lock[DIRTY_META]); 3347 ret = list_empty(&F2FS_I(inode)->gdirty_list); 3348 spin_unlock(&sbi->inode_lock[DIRTY_META]); 3349 return ret; 3350 } 3351 if (!is_inode_flag_set(inode, FI_AUTO_RECOVER) || 3352 file_keep_isize(inode) || 3353 i_size_read(inode) & ~PAGE_MASK) 3354 return false; 3355 3356 if (!f2fs_is_time_consistent(inode)) 3357 return false; 3358 3359 spin_lock(&F2FS_I(inode)->i_size_lock); 3360 ret = F2FS_I(inode)->last_disk_size == i_size_read(inode); 3361 spin_unlock(&F2FS_I(inode)->i_size_lock); 3362 3363 return ret; 3364 } 3365 3366 static inline bool f2fs_readonly(struct super_block *sb) 3367 { 3368 return sb_rdonly(sb); 3369 } 3370 3371 static inline bool f2fs_cp_error(struct f2fs_sb_info *sbi) 3372 { 3373 return is_set_ckpt_flags(sbi, CP_ERROR_FLAG); 3374 } 3375 3376 static inline bool is_dot_dotdot(const u8 *name, size_t len) 3377 { 3378 if (len == 1 && name[0] == '.') 3379 return true; 3380 3381 if (len == 2 && name[0] == '.' && name[1] == '.') 3382 return true; 3383 3384 return false; 3385 } 3386 3387 static inline void *f2fs_kmalloc(struct f2fs_sb_info *sbi, 3388 size_t size, gfp_t flags) 3389 { 3390 if (time_to_inject(sbi, FAULT_KMALLOC)) 3391 return NULL; 3392 3393 return kmalloc(size, flags); 3394 } 3395 3396 static inline void *f2fs_getname(struct f2fs_sb_info *sbi) 3397 { 3398 if (time_to_inject(sbi, FAULT_KMALLOC)) 3399 return NULL; 3400 3401 return __getname(); 3402 } 3403 3404 static inline void f2fs_putname(char *buf) 3405 { 3406 __putname(buf); 3407 } 3408 3409 static inline void *f2fs_kzalloc(struct f2fs_sb_info *sbi, 3410 size_t size, gfp_t flags) 3411 { 3412 return f2fs_kmalloc(sbi, size, flags | __GFP_ZERO); 3413 } 3414 3415 static inline void *f2fs_kvmalloc(struct f2fs_sb_info *sbi, 3416 size_t size, gfp_t flags) 3417 { 3418 if (time_to_inject(sbi, FAULT_KVMALLOC)) 3419 return NULL; 3420 3421 return kvmalloc(size, flags); 3422 } 3423 3424 static inline void *f2fs_kvzalloc(struct f2fs_sb_info *sbi, 3425 size_t size, gfp_t flags) 3426 { 3427 return f2fs_kvmalloc(sbi, size, flags | __GFP_ZERO); 3428 } 3429 3430 static inline int get_extra_isize(struct inode *inode) 3431 { 3432 return F2FS_I(inode)->i_extra_isize / sizeof(__le32); 3433 } 3434 3435 static inline int get_inline_xattr_addrs(struct inode *inode) 3436 { 3437 return F2FS_I(inode)->i_inline_xattr_size; 3438 } 3439 3440 static inline __le32 *get_dnode_addr(struct inode *inode, 3441 struct page *node_page) 3442 { 3443 int base = 0; 3444 3445 if (IS_INODE(node_page) && f2fs_has_extra_attr(inode)) 3446 base = get_extra_isize(inode); 3447 3448 return blkaddr_in_node(F2FS_NODE(node_page)) + base; 3449 } 3450 3451 #define f2fs_get_inode_mode(i) \ 3452 ((is_inode_flag_set(i, FI_ACL_MODE)) ? \ 3453 (F2FS_I(i)->i_acl_mode) : ((i)->i_mode)) 3454 3455 #define F2FS_MIN_EXTRA_ATTR_SIZE (sizeof(__le32)) 3456 3457 #define F2FS_TOTAL_EXTRA_ATTR_SIZE \ 3458 (offsetof(struct f2fs_inode, i_extra_end) - \ 3459 offsetof(struct f2fs_inode, i_extra_isize)) \ 3460 3461 #define F2FS_OLD_ATTRIBUTE_SIZE (offsetof(struct f2fs_inode, i_addr)) 3462 #define F2FS_FITS_IN_INODE(f2fs_inode, extra_isize, field) \ 3463 ((offsetof(typeof(*(f2fs_inode)), field) + \ 3464 sizeof((f2fs_inode)->field)) \ 3465 <= (F2FS_OLD_ATTRIBUTE_SIZE + (extra_isize))) \ 3466 3467 #define __is_large_section(sbi) (SEGS_PER_SEC(sbi) > 1) 3468 3469 #define __is_meta_io(fio) (PAGE_TYPE_OF_BIO((fio)->type) == META) 3470 3471 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi, 3472 block_t blkaddr, int type); 3473 static inline void verify_blkaddr(struct f2fs_sb_info *sbi, 3474 block_t blkaddr, int type) 3475 { 3476 if (!f2fs_is_valid_blkaddr(sbi, blkaddr, type)) 3477 f2fs_err(sbi, "invalid blkaddr: %u, type: %d, run fsck to fix.", 3478 blkaddr, type); 3479 } 3480 3481 static inline bool __is_valid_data_blkaddr(block_t blkaddr) 3482 { 3483 if (blkaddr == NEW_ADDR || blkaddr == NULL_ADDR || 3484 blkaddr == COMPRESS_ADDR) 3485 return false; 3486 return true; 3487 } 3488 3489 /* 3490 * file.c 3491 */ 3492 int f2fs_sync_file(struct file *file, loff_t start, loff_t end, int datasync); 3493 int f2fs_do_truncate_blocks(struct inode *inode, u64 from, bool lock); 3494 int f2fs_truncate_blocks(struct inode *inode, u64 from, bool lock); 3495 int f2fs_truncate(struct inode *inode); 3496 int f2fs_getattr(struct mnt_idmap *idmap, const struct path *path, 3497 struct kstat *stat, u32 request_mask, unsigned int flags); 3498 int f2fs_setattr(struct mnt_idmap *idmap, struct dentry *dentry, 3499 struct iattr *attr); 3500 int f2fs_truncate_hole(struct inode *inode, pgoff_t pg_start, pgoff_t pg_end); 3501 void f2fs_truncate_data_blocks_range(struct dnode_of_data *dn, int count); 3502 int f2fs_precache_extents(struct inode *inode); 3503 int f2fs_fileattr_get(struct dentry *dentry, struct fileattr *fa); 3504 int f2fs_fileattr_set(struct mnt_idmap *idmap, 3505 struct dentry *dentry, struct fileattr *fa); 3506 long f2fs_ioctl(struct file *filp, unsigned int cmd, unsigned long arg); 3507 long f2fs_compat_ioctl(struct file *file, unsigned int cmd, unsigned long arg); 3508 int f2fs_transfer_project_quota(struct inode *inode, kprojid_t kprojid); 3509 int f2fs_pin_file_control(struct inode *inode, bool inc); 3510 3511 /* 3512 * inode.c 3513 */ 3514 void f2fs_set_inode_flags(struct inode *inode); 3515 bool f2fs_inode_chksum_verify(struct f2fs_sb_info *sbi, struct page *page); 3516 void f2fs_inode_chksum_set(struct f2fs_sb_info *sbi, struct page *page); 3517 struct inode *f2fs_iget(struct super_block *sb, unsigned long ino); 3518 struct inode *f2fs_iget_retry(struct super_block *sb, unsigned long ino); 3519 int f2fs_try_to_free_nats(struct f2fs_sb_info *sbi, int nr_shrink); 3520 void f2fs_update_inode(struct inode *inode, struct page *node_page); 3521 void f2fs_update_inode_page(struct inode *inode); 3522 int f2fs_write_inode(struct inode *inode, struct writeback_control *wbc); 3523 void f2fs_evict_inode(struct inode *inode); 3524 void f2fs_handle_failed_inode(struct inode *inode); 3525 3526 /* 3527 * namei.c 3528 */ 3529 int f2fs_update_extension_list(struct f2fs_sb_info *sbi, const char *name, 3530 bool hot, bool set); 3531 struct dentry *f2fs_get_parent(struct dentry *child); 3532 int f2fs_get_tmpfile(struct mnt_idmap *idmap, struct inode *dir, 3533 struct inode **new_inode); 3534 3535 /* 3536 * dir.c 3537 */ 3538 int f2fs_init_casefolded_name(const struct inode *dir, 3539 struct f2fs_filename *fname); 3540 int f2fs_setup_filename(struct inode *dir, const struct qstr *iname, 3541 int lookup, struct f2fs_filename *fname); 3542 int f2fs_prepare_lookup(struct inode *dir, struct dentry *dentry, 3543 struct f2fs_filename *fname); 3544 void f2fs_free_filename(struct f2fs_filename *fname); 3545 struct f2fs_dir_entry *f2fs_find_target_dentry(const struct f2fs_dentry_ptr *d, 3546 const struct f2fs_filename *fname, int *max_slots); 3547 int f2fs_fill_dentries(struct dir_context *ctx, struct f2fs_dentry_ptr *d, 3548 unsigned int start_pos, struct fscrypt_str *fstr); 3549 void f2fs_do_make_empty_dir(struct inode *inode, struct inode *parent, 3550 struct f2fs_dentry_ptr *d); 3551 struct page *f2fs_init_inode_metadata(struct inode *inode, struct inode *dir, 3552 const struct f2fs_filename *fname, struct page *dpage); 3553 void f2fs_update_parent_metadata(struct inode *dir, struct inode *inode, 3554 unsigned int current_depth); 3555 int f2fs_room_for_filename(const void *bitmap, int slots, int max_slots); 3556 void f2fs_drop_nlink(struct inode *dir, struct inode *inode); 3557 struct f2fs_dir_entry *__f2fs_find_entry(struct inode *dir, 3558 const struct f2fs_filename *fname, 3559 struct page **res_page); 3560 struct f2fs_dir_entry *f2fs_find_entry(struct inode *dir, 3561 const struct qstr *child, struct page **res_page); 3562 struct f2fs_dir_entry *f2fs_parent_dir(struct inode *dir, struct page **p); 3563 ino_t f2fs_inode_by_name(struct inode *dir, const struct qstr *qstr, 3564 struct page **page); 3565 void f2fs_set_link(struct inode *dir, struct f2fs_dir_entry *de, 3566 struct page *page, struct inode *inode); 3567 bool f2fs_has_enough_room(struct inode *dir, struct page *ipage, 3568 const struct f2fs_filename *fname); 3569 void f2fs_update_dentry(nid_t ino, umode_t mode, struct f2fs_dentry_ptr *d, 3570 const struct fscrypt_str *name, f2fs_hash_t name_hash, 3571 unsigned int bit_pos); 3572 int f2fs_add_regular_entry(struct inode *dir, const struct f2fs_filename *fname, 3573 struct inode *inode, nid_t ino, umode_t mode); 3574 int f2fs_add_dentry(struct inode *dir, const struct f2fs_filename *fname, 3575 struct inode *inode, nid_t ino, umode_t mode); 3576 int f2fs_do_add_link(struct inode *dir, const struct qstr *name, 3577 struct inode *inode, nid_t ino, umode_t mode); 3578 void f2fs_delete_entry(struct f2fs_dir_entry *dentry, struct page *page, 3579 struct inode *dir, struct inode *inode); 3580 int f2fs_do_tmpfile(struct inode *inode, struct inode *dir, 3581 struct f2fs_filename *fname); 3582 bool f2fs_empty_dir(struct inode *dir); 3583 3584 static inline int f2fs_add_link(struct dentry *dentry, struct inode *inode) 3585 { 3586 if (fscrypt_is_nokey_name(dentry)) 3587 return -ENOKEY; 3588 return f2fs_do_add_link(d_inode(dentry->d_parent), &dentry->d_name, 3589 inode, inode->i_ino, inode->i_mode); 3590 } 3591 3592 /* 3593 * super.c 3594 */ 3595 int f2fs_inode_dirtied(struct inode *inode, bool sync); 3596 void f2fs_inode_synced(struct inode *inode); 3597 int f2fs_dquot_initialize(struct inode *inode); 3598 int f2fs_enable_quota_files(struct f2fs_sb_info *sbi, bool rdonly); 3599 int f2fs_quota_sync(struct super_block *sb, int type); 3600 loff_t max_file_blocks(struct inode *inode); 3601 void f2fs_quota_off_umount(struct super_block *sb); 3602 void f2fs_save_errors(struct f2fs_sb_info *sbi, unsigned char flag); 3603 void f2fs_handle_critical_error(struct f2fs_sb_info *sbi, unsigned char reason, 3604 bool irq_context); 3605 void f2fs_handle_error(struct f2fs_sb_info *sbi, unsigned char error); 3606 void f2fs_handle_error_async(struct f2fs_sb_info *sbi, unsigned char error); 3607 int f2fs_commit_super(struct f2fs_sb_info *sbi, bool recover); 3608 int f2fs_sync_fs(struct super_block *sb, int sync); 3609 int f2fs_sanity_check_ckpt(struct f2fs_sb_info *sbi); 3610 3611 /* 3612 * hash.c 3613 */ 3614 void f2fs_hash_filename(const struct inode *dir, struct f2fs_filename *fname); 3615 3616 /* 3617 * node.c 3618 */ 3619 struct node_info; 3620 3621 int f2fs_check_nid_range(struct f2fs_sb_info *sbi, nid_t nid); 3622 bool f2fs_available_free_memory(struct f2fs_sb_info *sbi, int type); 3623 bool f2fs_in_warm_node_list(struct f2fs_sb_info *sbi, struct page *page); 3624 void f2fs_init_fsync_node_info(struct f2fs_sb_info *sbi); 3625 void f2fs_del_fsync_node_entry(struct f2fs_sb_info *sbi, struct page *page); 3626 void f2fs_reset_fsync_node_info(struct f2fs_sb_info *sbi); 3627 int f2fs_need_dentry_mark(struct f2fs_sb_info *sbi, nid_t nid); 3628 bool f2fs_is_checkpointed_node(struct f2fs_sb_info *sbi, nid_t nid); 3629 bool f2fs_need_inode_block_update(struct f2fs_sb_info *sbi, nid_t ino); 3630 int f2fs_get_node_info(struct f2fs_sb_info *sbi, nid_t nid, 3631 struct node_info *ni, bool checkpoint_context); 3632 pgoff_t f2fs_get_next_page_offset(struct dnode_of_data *dn, pgoff_t pgofs); 3633 int f2fs_get_dnode_of_data(struct dnode_of_data *dn, pgoff_t index, int mode); 3634 int f2fs_truncate_inode_blocks(struct inode *inode, pgoff_t from); 3635 int f2fs_truncate_xattr_node(struct inode *inode); 3636 int f2fs_wait_on_node_pages_writeback(struct f2fs_sb_info *sbi, 3637 unsigned int seq_id); 3638 bool f2fs_nat_bitmap_enabled(struct f2fs_sb_info *sbi); 3639 int f2fs_remove_inode_page(struct inode *inode); 3640 struct page *f2fs_new_inode_page(struct inode *inode); 3641 struct page *f2fs_new_node_page(struct dnode_of_data *dn, unsigned int ofs); 3642 void f2fs_ra_node_page(struct f2fs_sb_info *sbi, nid_t nid); 3643 struct page *f2fs_get_node_page(struct f2fs_sb_info *sbi, pgoff_t nid); 3644 struct page *f2fs_get_node_page_ra(struct page *parent, int start); 3645 int f2fs_move_node_page(struct page *node_page, int gc_type); 3646 void f2fs_flush_inline_data(struct f2fs_sb_info *sbi); 3647 int f2fs_fsync_node_pages(struct f2fs_sb_info *sbi, struct inode *inode, 3648 struct writeback_control *wbc, bool atomic, 3649 unsigned int *seq_id); 3650 int f2fs_sync_node_pages(struct f2fs_sb_info *sbi, 3651 struct writeback_control *wbc, 3652 bool do_balance, enum iostat_type io_type); 3653 int f2fs_build_free_nids(struct f2fs_sb_info *sbi, bool sync, bool mount); 3654 bool f2fs_alloc_nid(struct f2fs_sb_info *sbi, nid_t *nid); 3655 void f2fs_alloc_nid_done(struct f2fs_sb_info *sbi, nid_t nid); 3656 void f2fs_alloc_nid_failed(struct f2fs_sb_info *sbi, nid_t nid); 3657 int f2fs_try_to_free_nids(struct f2fs_sb_info *sbi, int nr_shrink); 3658 int f2fs_recover_inline_xattr(struct inode *inode, struct page *page); 3659 int f2fs_recover_xattr_data(struct inode *inode, struct page *page); 3660 int f2fs_recover_inode_page(struct f2fs_sb_info *sbi, struct page *page); 3661 int f2fs_restore_node_summary(struct f2fs_sb_info *sbi, 3662 unsigned int segno, struct f2fs_summary_block *sum); 3663 void f2fs_enable_nat_bits(struct f2fs_sb_info *sbi); 3664 int f2fs_flush_nat_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3665 int f2fs_build_node_manager(struct f2fs_sb_info *sbi); 3666 void f2fs_destroy_node_manager(struct f2fs_sb_info *sbi); 3667 int __init f2fs_create_node_manager_caches(void); 3668 void f2fs_destroy_node_manager_caches(void); 3669 3670 /* 3671 * segment.c 3672 */ 3673 bool f2fs_need_SSR(struct f2fs_sb_info *sbi); 3674 int f2fs_commit_atomic_write(struct inode *inode); 3675 void f2fs_abort_atomic_write(struct inode *inode, bool clean); 3676 void f2fs_balance_fs(struct f2fs_sb_info *sbi, bool need); 3677 void f2fs_balance_fs_bg(struct f2fs_sb_info *sbi, bool from_bg); 3678 int f2fs_issue_flush(struct f2fs_sb_info *sbi, nid_t ino); 3679 int f2fs_create_flush_cmd_control(struct f2fs_sb_info *sbi); 3680 int f2fs_flush_device_cache(struct f2fs_sb_info *sbi); 3681 void f2fs_destroy_flush_cmd_control(struct f2fs_sb_info *sbi, bool free); 3682 void f2fs_invalidate_blocks(struct f2fs_sb_info *sbi, block_t addr); 3683 bool f2fs_is_checkpointed_data(struct f2fs_sb_info *sbi, block_t blkaddr); 3684 int f2fs_start_discard_thread(struct f2fs_sb_info *sbi); 3685 void f2fs_drop_discard_cmd(struct f2fs_sb_info *sbi); 3686 void f2fs_stop_discard_thread(struct f2fs_sb_info *sbi); 3687 bool f2fs_issue_discard_timeout(struct f2fs_sb_info *sbi); 3688 void f2fs_clear_prefree_segments(struct f2fs_sb_info *sbi, 3689 struct cp_control *cpc); 3690 void f2fs_dirty_to_prefree(struct f2fs_sb_info *sbi); 3691 block_t f2fs_get_unusable_blocks(struct f2fs_sb_info *sbi); 3692 int f2fs_disable_cp_again(struct f2fs_sb_info *sbi, block_t unusable); 3693 void f2fs_release_discard_addrs(struct f2fs_sb_info *sbi); 3694 int f2fs_npages_for_summary_flush(struct f2fs_sb_info *sbi, bool for_ra); 3695 bool f2fs_segment_has_free_slot(struct f2fs_sb_info *sbi, int segno); 3696 void f2fs_init_inmem_curseg(struct f2fs_sb_info *sbi); 3697 void f2fs_save_inmem_curseg(struct f2fs_sb_info *sbi); 3698 void f2fs_restore_inmem_curseg(struct f2fs_sb_info *sbi); 3699 void f2fs_get_new_segment(struct f2fs_sb_info *sbi, 3700 unsigned int *newseg, bool new_sec, int dir); 3701 void f2fs_allocate_segment_for_resize(struct f2fs_sb_info *sbi, int type, 3702 unsigned int start, unsigned int end); 3703 int f2fs_allocate_new_section(struct f2fs_sb_info *sbi, int type, bool force); 3704 int f2fs_allocate_pinning_section(struct f2fs_sb_info *sbi); 3705 void f2fs_allocate_new_segments(struct f2fs_sb_info *sbi); 3706 int f2fs_trim_fs(struct f2fs_sb_info *sbi, struct fstrim_range *range); 3707 bool f2fs_exist_trim_candidates(struct f2fs_sb_info *sbi, 3708 struct cp_control *cpc); 3709 struct page *f2fs_get_sum_page(struct f2fs_sb_info *sbi, unsigned int segno); 3710 void f2fs_update_meta_page(struct f2fs_sb_info *sbi, void *src, 3711 block_t blk_addr); 3712 void f2fs_do_write_meta_page(struct f2fs_sb_info *sbi, struct page *page, 3713 enum iostat_type io_type); 3714 void f2fs_do_write_node_page(unsigned int nid, struct f2fs_io_info *fio); 3715 void f2fs_outplace_write_data(struct dnode_of_data *dn, 3716 struct f2fs_io_info *fio); 3717 int f2fs_inplace_write_data(struct f2fs_io_info *fio); 3718 void f2fs_do_replace_block(struct f2fs_sb_info *sbi, struct f2fs_summary *sum, 3719 block_t old_blkaddr, block_t new_blkaddr, 3720 bool recover_curseg, bool recover_newaddr, 3721 bool from_gc); 3722 void f2fs_replace_block(struct f2fs_sb_info *sbi, struct dnode_of_data *dn, 3723 block_t old_addr, block_t new_addr, 3724 unsigned char version, bool recover_curseg, 3725 bool recover_newaddr); 3726 void f2fs_allocate_data_block(struct f2fs_sb_info *sbi, struct page *page, 3727 block_t old_blkaddr, block_t *new_blkaddr, 3728 struct f2fs_summary *sum, int type, 3729 struct f2fs_io_info *fio); 3730 void f2fs_update_device_state(struct f2fs_sb_info *sbi, nid_t ino, 3731 block_t blkaddr, unsigned int blkcnt); 3732 void f2fs_wait_on_page_writeback(struct page *page, 3733 enum page_type type, bool ordered, bool locked); 3734 void f2fs_wait_on_block_writeback(struct inode *inode, block_t blkaddr); 3735 void f2fs_wait_on_block_writeback_range(struct inode *inode, block_t blkaddr, 3736 block_t len); 3737 void f2fs_write_data_summaries(struct f2fs_sb_info *sbi, block_t start_blk); 3738 void f2fs_write_node_summaries(struct f2fs_sb_info *sbi, block_t start_blk); 3739 int f2fs_lookup_journal_in_cursum(struct f2fs_journal *journal, int type, 3740 unsigned int val, int alloc); 3741 void f2fs_flush_sit_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3742 int f2fs_fix_curseg_write_pointer(struct f2fs_sb_info *sbi); 3743 int f2fs_check_write_pointer(struct f2fs_sb_info *sbi); 3744 int f2fs_build_segment_manager(struct f2fs_sb_info *sbi); 3745 void f2fs_destroy_segment_manager(struct f2fs_sb_info *sbi); 3746 int __init f2fs_create_segment_manager_caches(void); 3747 void f2fs_destroy_segment_manager_caches(void); 3748 int f2fs_rw_hint_to_seg_type(enum rw_hint hint); 3749 unsigned int f2fs_usable_segs_in_sec(struct f2fs_sb_info *sbi, 3750 unsigned int segno); 3751 unsigned int f2fs_usable_blks_in_seg(struct f2fs_sb_info *sbi, 3752 unsigned int segno); 3753 3754 #define DEF_FRAGMENT_SIZE 4 3755 #define MIN_FRAGMENT_SIZE 1 3756 #define MAX_FRAGMENT_SIZE 512 3757 3758 static inline bool f2fs_need_rand_seg(struct f2fs_sb_info *sbi) 3759 { 3760 return F2FS_OPTION(sbi).fs_mode == FS_MODE_FRAGMENT_SEG || 3761 F2FS_OPTION(sbi).fs_mode == FS_MODE_FRAGMENT_BLK; 3762 } 3763 3764 /* 3765 * checkpoint.c 3766 */ 3767 void f2fs_stop_checkpoint(struct f2fs_sb_info *sbi, bool end_io, 3768 unsigned char reason); 3769 void f2fs_flush_ckpt_thread(struct f2fs_sb_info *sbi); 3770 struct page *f2fs_grab_meta_page(struct f2fs_sb_info *sbi, pgoff_t index); 3771 struct page *f2fs_get_meta_page(struct f2fs_sb_info *sbi, pgoff_t index); 3772 struct page *f2fs_get_meta_page_retry(struct f2fs_sb_info *sbi, pgoff_t index); 3773 struct page *f2fs_get_tmp_page(struct f2fs_sb_info *sbi, pgoff_t index); 3774 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi, 3775 block_t blkaddr, int type); 3776 int f2fs_ra_meta_pages(struct f2fs_sb_info *sbi, block_t start, int nrpages, 3777 int type, bool sync); 3778 void f2fs_ra_meta_pages_cond(struct f2fs_sb_info *sbi, pgoff_t index, 3779 unsigned int ra_blocks); 3780 long f2fs_sync_meta_pages(struct f2fs_sb_info *sbi, enum page_type type, 3781 long nr_to_write, enum iostat_type io_type); 3782 void f2fs_add_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type); 3783 void f2fs_remove_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type); 3784 void f2fs_release_ino_entry(struct f2fs_sb_info *sbi, bool all); 3785 bool f2fs_exist_written_data(struct f2fs_sb_info *sbi, nid_t ino, int mode); 3786 void f2fs_set_dirty_device(struct f2fs_sb_info *sbi, nid_t ino, 3787 unsigned int devidx, int type); 3788 bool f2fs_is_dirty_device(struct f2fs_sb_info *sbi, nid_t ino, 3789 unsigned int devidx, int type); 3790 int f2fs_acquire_orphan_inode(struct f2fs_sb_info *sbi); 3791 void f2fs_release_orphan_inode(struct f2fs_sb_info *sbi); 3792 void f2fs_add_orphan_inode(struct inode *inode); 3793 void f2fs_remove_orphan_inode(struct f2fs_sb_info *sbi, nid_t ino); 3794 int f2fs_recover_orphan_inodes(struct f2fs_sb_info *sbi); 3795 int f2fs_get_valid_checkpoint(struct f2fs_sb_info *sbi); 3796 void f2fs_update_dirty_folio(struct inode *inode, struct folio *folio); 3797 void f2fs_remove_dirty_inode(struct inode *inode); 3798 int f2fs_sync_dirty_inodes(struct f2fs_sb_info *sbi, enum inode_type type, 3799 bool from_cp); 3800 void f2fs_wait_on_all_pages(struct f2fs_sb_info *sbi, int type); 3801 u64 f2fs_get_sectors_written(struct f2fs_sb_info *sbi); 3802 int f2fs_write_checkpoint(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3803 void f2fs_init_ino_entry_info(struct f2fs_sb_info *sbi); 3804 int __init f2fs_create_checkpoint_caches(void); 3805 void f2fs_destroy_checkpoint_caches(void); 3806 int f2fs_issue_checkpoint(struct f2fs_sb_info *sbi); 3807 int f2fs_start_ckpt_thread(struct f2fs_sb_info *sbi); 3808 void f2fs_stop_ckpt_thread(struct f2fs_sb_info *sbi); 3809 void f2fs_init_ckpt_req_control(struct f2fs_sb_info *sbi); 3810 3811 /* 3812 * data.c 3813 */ 3814 int __init f2fs_init_bioset(void); 3815 void f2fs_destroy_bioset(void); 3816 bool f2fs_is_cp_guaranteed(struct page *page); 3817 int f2fs_init_bio_entry_cache(void); 3818 void f2fs_destroy_bio_entry_cache(void); 3819 void f2fs_submit_read_bio(struct f2fs_sb_info *sbi, struct bio *bio, 3820 enum page_type type); 3821 int f2fs_init_write_merge_io(struct f2fs_sb_info *sbi); 3822 void f2fs_submit_merged_write(struct f2fs_sb_info *sbi, enum page_type type); 3823 void f2fs_submit_merged_write_cond(struct f2fs_sb_info *sbi, 3824 struct inode *inode, struct page *page, 3825 nid_t ino, enum page_type type); 3826 void f2fs_submit_merged_ipu_write(struct f2fs_sb_info *sbi, 3827 struct bio **bio, struct page *page); 3828 void f2fs_flush_merged_writes(struct f2fs_sb_info *sbi); 3829 int f2fs_submit_page_bio(struct f2fs_io_info *fio); 3830 int f2fs_merge_page_bio(struct f2fs_io_info *fio); 3831 void f2fs_submit_page_write(struct f2fs_io_info *fio); 3832 struct block_device *f2fs_target_device(struct f2fs_sb_info *sbi, 3833 block_t blk_addr, sector_t *sector); 3834 int f2fs_target_device_index(struct f2fs_sb_info *sbi, block_t blkaddr); 3835 void f2fs_set_data_blkaddr(struct dnode_of_data *dn, block_t blkaddr); 3836 void f2fs_update_data_blkaddr(struct dnode_of_data *dn, block_t blkaddr); 3837 int f2fs_reserve_new_blocks(struct dnode_of_data *dn, blkcnt_t count); 3838 int f2fs_reserve_new_block(struct dnode_of_data *dn); 3839 int f2fs_get_block_locked(struct dnode_of_data *dn, pgoff_t index); 3840 int f2fs_reserve_block(struct dnode_of_data *dn, pgoff_t index); 3841 struct page *f2fs_get_read_data_page(struct inode *inode, pgoff_t index, 3842 blk_opf_t op_flags, bool for_write, pgoff_t *next_pgofs); 3843 struct page *f2fs_find_data_page(struct inode *inode, pgoff_t index, 3844 pgoff_t *next_pgofs); 3845 struct page *f2fs_get_lock_data_page(struct inode *inode, pgoff_t index, 3846 bool for_write); 3847 struct page *f2fs_get_new_data_page(struct inode *inode, 3848 struct page *ipage, pgoff_t index, bool new_i_size); 3849 int f2fs_do_write_data_page(struct f2fs_io_info *fio); 3850 int f2fs_map_blocks(struct inode *inode, struct f2fs_map_blocks *map, int flag); 3851 int f2fs_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, 3852 u64 start, u64 len); 3853 int f2fs_encrypt_one_page(struct f2fs_io_info *fio); 3854 bool f2fs_should_update_inplace(struct inode *inode, struct f2fs_io_info *fio); 3855 bool f2fs_should_update_outplace(struct inode *inode, struct f2fs_io_info *fio); 3856 int f2fs_write_single_data_page(struct page *page, int *submitted, 3857 struct bio **bio, sector_t *last_block, 3858 struct writeback_control *wbc, 3859 enum iostat_type io_type, 3860 int compr_blocks, bool allow_balance); 3861 void f2fs_write_failed(struct inode *inode, loff_t to); 3862 void f2fs_invalidate_folio(struct folio *folio, size_t offset, size_t length); 3863 bool f2fs_release_folio(struct folio *folio, gfp_t wait); 3864 bool f2fs_overwrite_io(struct inode *inode, loff_t pos, size_t len); 3865 void f2fs_clear_page_cache_dirty_tag(struct page *page); 3866 int f2fs_init_post_read_processing(void); 3867 void f2fs_destroy_post_read_processing(void); 3868 int f2fs_init_post_read_wq(struct f2fs_sb_info *sbi); 3869 void f2fs_destroy_post_read_wq(struct f2fs_sb_info *sbi); 3870 extern const struct iomap_ops f2fs_iomap_ops; 3871 3872 /* 3873 * gc.c 3874 */ 3875 int f2fs_start_gc_thread(struct f2fs_sb_info *sbi); 3876 void f2fs_stop_gc_thread(struct f2fs_sb_info *sbi); 3877 block_t f2fs_start_bidx_of_node(unsigned int node_ofs, struct inode *inode); 3878 int f2fs_gc(struct f2fs_sb_info *sbi, struct f2fs_gc_control *gc_control); 3879 void f2fs_build_gc_manager(struct f2fs_sb_info *sbi); 3880 int f2fs_gc_range(struct f2fs_sb_info *sbi, 3881 unsigned int start_seg, unsigned int end_seg, 3882 bool dry_run, unsigned int dry_run_sections); 3883 int f2fs_resize_fs(struct file *filp, __u64 block_count); 3884 int __init f2fs_create_garbage_collection_cache(void); 3885 void f2fs_destroy_garbage_collection_cache(void); 3886 /* victim selection function for cleaning and SSR */ 3887 int f2fs_get_victim(struct f2fs_sb_info *sbi, unsigned int *result, 3888 int gc_type, int type, char alloc_mode, 3889 unsigned long long age); 3890 3891 /* 3892 * recovery.c 3893 */ 3894 int f2fs_recover_fsync_data(struct f2fs_sb_info *sbi, bool check_only); 3895 bool f2fs_space_for_roll_forward(struct f2fs_sb_info *sbi); 3896 int __init f2fs_create_recovery_cache(void); 3897 void f2fs_destroy_recovery_cache(void); 3898 3899 /* 3900 * debug.c 3901 */ 3902 #ifdef CONFIG_F2FS_STAT_FS 3903 struct f2fs_stat_info { 3904 struct list_head stat_list; 3905 struct f2fs_sb_info *sbi; 3906 int all_area_segs, sit_area_segs, nat_area_segs, ssa_area_segs; 3907 int main_area_segs, main_area_sections, main_area_zones; 3908 unsigned long long hit_cached[NR_EXTENT_CACHES]; 3909 unsigned long long hit_rbtree[NR_EXTENT_CACHES]; 3910 unsigned long long total_ext[NR_EXTENT_CACHES]; 3911 unsigned long long hit_total[NR_EXTENT_CACHES]; 3912 int ext_tree[NR_EXTENT_CACHES]; 3913 int zombie_tree[NR_EXTENT_CACHES]; 3914 int ext_node[NR_EXTENT_CACHES]; 3915 /* to count memory footprint */ 3916 unsigned long long ext_mem[NR_EXTENT_CACHES]; 3917 /* for read extent cache */ 3918 unsigned long long hit_largest; 3919 /* for block age extent cache */ 3920 unsigned long long allocated_data_blocks; 3921 int ndirty_node, ndirty_dent, ndirty_meta, ndirty_imeta; 3922 int ndirty_data, ndirty_qdata; 3923 unsigned int ndirty_dirs, ndirty_files, nquota_files, ndirty_all; 3924 int nats, dirty_nats, sits, dirty_sits; 3925 int free_nids, avail_nids, alloc_nids; 3926 int total_count, utilization; 3927 int nr_wb_cp_data, nr_wb_data; 3928 int nr_rd_data, nr_rd_node, nr_rd_meta; 3929 int nr_dio_read, nr_dio_write; 3930 unsigned int io_skip_bggc, other_skip_bggc; 3931 int nr_flushing, nr_flushed, flush_list_empty; 3932 int nr_discarding, nr_discarded; 3933 int nr_discard_cmd; 3934 unsigned int undiscard_blks; 3935 int nr_issued_ckpt, nr_total_ckpt, nr_queued_ckpt; 3936 unsigned int cur_ckpt_time, peak_ckpt_time; 3937 int inline_xattr, inline_inode, inline_dir, append, update, orphans; 3938 int compr_inode, swapfile_inode; 3939 unsigned long long compr_blocks; 3940 int aw_cnt, max_aw_cnt; 3941 unsigned int valid_count, valid_node_count, valid_inode_count, discard_blks; 3942 unsigned int bimodal, avg_vblocks; 3943 int util_free, util_valid, util_invalid; 3944 int rsvd_segs, overp_segs; 3945 int dirty_count, node_pages, meta_pages, compress_pages; 3946 int compress_page_hit; 3947 int prefree_count, free_segs, free_secs; 3948 int cp_call_count[MAX_CALL_TYPE], cp_count; 3949 int gc_call_count[MAX_CALL_TYPE]; 3950 int gc_segs[2][2]; 3951 int gc_secs[2][2]; 3952 int tot_blks, data_blks, node_blks; 3953 int bg_data_blks, bg_node_blks; 3954 int curseg[NR_CURSEG_TYPE]; 3955 int cursec[NR_CURSEG_TYPE]; 3956 int curzone[NR_CURSEG_TYPE]; 3957 unsigned int dirty_seg[NR_CURSEG_TYPE]; 3958 unsigned int full_seg[NR_CURSEG_TYPE]; 3959 unsigned int valid_blks[NR_CURSEG_TYPE]; 3960 3961 unsigned int meta_count[META_MAX]; 3962 unsigned int segment_count[2]; 3963 unsigned int block_count[2]; 3964 unsigned int inplace_count; 3965 unsigned long long base_mem, cache_mem, page_mem; 3966 }; 3967 3968 static inline struct f2fs_stat_info *F2FS_STAT(struct f2fs_sb_info *sbi) 3969 { 3970 return (struct f2fs_stat_info *)sbi->stat_info; 3971 } 3972 3973 #define stat_inc_cp_call_count(sbi, foreground) \ 3974 atomic_inc(&sbi->cp_call_count[(foreground)]) 3975 #define stat_inc_cp_count(si) (F2FS_STAT(sbi)->cp_count++) 3976 #define stat_io_skip_bggc_count(sbi) ((sbi)->io_skip_bggc++) 3977 #define stat_other_skip_bggc_count(sbi) ((sbi)->other_skip_bggc++) 3978 #define stat_inc_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]++) 3979 #define stat_dec_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]--) 3980 #define stat_inc_total_hit(sbi, type) (atomic64_inc(&(sbi)->total_hit_ext[type])) 3981 #define stat_inc_rbtree_node_hit(sbi, type) (atomic64_inc(&(sbi)->read_hit_rbtree[type])) 3982 #define stat_inc_largest_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_largest)) 3983 #define stat_inc_cached_node_hit(sbi, type) (atomic64_inc(&(sbi)->read_hit_cached[type])) 3984 #define stat_inc_inline_xattr(inode) \ 3985 do { \ 3986 if (f2fs_has_inline_xattr(inode)) \ 3987 (atomic_inc(&F2FS_I_SB(inode)->inline_xattr)); \ 3988 } while (0) 3989 #define stat_dec_inline_xattr(inode) \ 3990 do { \ 3991 if (f2fs_has_inline_xattr(inode)) \ 3992 (atomic_dec(&F2FS_I_SB(inode)->inline_xattr)); \ 3993 } while (0) 3994 #define stat_inc_inline_inode(inode) \ 3995 do { \ 3996 if (f2fs_has_inline_data(inode)) \ 3997 (atomic_inc(&F2FS_I_SB(inode)->inline_inode)); \ 3998 } while (0) 3999 #define stat_dec_inline_inode(inode) \ 4000 do { \ 4001 if (f2fs_has_inline_data(inode)) \ 4002 (atomic_dec(&F2FS_I_SB(inode)->inline_inode)); \ 4003 } while (0) 4004 #define stat_inc_inline_dir(inode) \ 4005 do { \ 4006 if (f2fs_has_inline_dentry(inode)) \ 4007 (atomic_inc(&F2FS_I_SB(inode)->inline_dir)); \ 4008 } while (0) 4009 #define stat_dec_inline_dir(inode) \ 4010 do { \ 4011 if (f2fs_has_inline_dentry(inode)) \ 4012 (atomic_dec(&F2FS_I_SB(inode)->inline_dir)); \ 4013 } while (0) 4014 #define stat_inc_compr_inode(inode) \ 4015 do { \ 4016 if (f2fs_compressed_file(inode)) \ 4017 (atomic_inc(&F2FS_I_SB(inode)->compr_inode)); \ 4018 } while (0) 4019 #define stat_dec_compr_inode(inode) \ 4020 do { \ 4021 if (f2fs_compressed_file(inode)) \ 4022 (atomic_dec(&F2FS_I_SB(inode)->compr_inode)); \ 4023 } while (0) 4024 #define stat_add_compr_blocks(inode, blocks) \ 4025 (atomic64_add(blocks, &F2FS_I_SB(inode)->compr_blocks)) 4026 #define stat_sub_compr_blocks(inode, blocks) \ 4027 (atomic64_sub(blocks, &F2FS_I_SB(inode)->compr_blocks)) 4028 #define stat_inc_swapfile_inode(inode) \ 4029 (atomic_inc(&F2FS_I_SB(inode)->swapfile_inode)) 4030 #define stat_dec_swapfile_inode(inode) \ 4031 (atomic_dec(&F2FS_I_SB(inode)->swapfile_inode)) 4032 #define stat_inc_atomic_inode(inode) \ 4033 (atomic_inc(&F2FS_I_SB(inode)->atomic_files)) 4034 #define stat_dec_atomic_inode(inode) \ 4035 (atomic_dec(&F2FS_I_SB(inode)->atomic_files)) 4036 #define stat_inc_meta_count(sbi, blkaddr) \ 4037 do { \ 4038 if (blkaddr < SIT_I(sbi)->sit_base_addr) \ 4039 atomic_inc(&(sbi)->meta_count[META_CP]); \ 4040 else if (blkaddr < NM_I(sbi)->nat_blkaddr) \ 4041 atomic_inc(&(sbi)->meta_count[META_SIT]); \ 4042 else if (blkaddr < SM_I(sbi)->ssa_blkaddr) \ 4043 atomic_inc(&(sbi)->meta_count[META_NAT]); \ 4044 else if (blkaddr < SM_I(sbi)->main_blkaddr) \ 4045 atomic_inc(&(sbi)->meta_count[META_SSA]); \ 4046 } while (0) 4047 #define stat_inc_seg_type(sbi, curseg) \ 4048 ((sbi)->segment_count[(curseg)->alloc_type]++) 4049 #define stat_inc_block_count(sbi, curseg) \ 4050 ((sbi)->block_count[(curseg)->alloc_type]++) 4051 #define stat_inc_inplace_blocks(sbi) \ 4052 (atomic_inc(&(sbi)->inplace_count)) 4053 #define stat_update_max_atomic_write(inode) \ 4054 do { \ 4055 int cur = atomic_read(&F2FS_I_SB(inode)->atomic_files); \ 4056 int max = atomic_read(&F2FS_I_SB(inode)->max_aw_cnt); \ 4057 if (cur > max) \ 4058 atomic_set(&F2FS_I_SB(inode)->max_aw_cnt, cur); \ 4059 } while (0) 4060 #define stat_inc_gc_call_count(sbi, foreground) \ 4061 (F2FS_STAT(sbi)->gc_call_count[(foreground)]++) 4062 #define stat_inc_gc_sec_count(sbi, type, gc_type) \ 4063 (F2FS_STAT(sbi)->gc_secs[(type)][(gc_type)]++) 4064 #define stat_inc_gc_seg_count(sbi, type, gc_type) \ 4065 (F2FS_STAT(sbi)->gc_segs[(type)][(gc_type)]++) 4066 4067 #define stat_inc_tot_blk_count(si, blks) \ 4068 ((si)->tot_blks += (blks)) 4069 4070 #define stat_inc_data_blk_count(sbi, blks, gc_type) \ 4071 do { \ 4072 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 4073 stat_inc_tot_blk_count(si, blks); \ 4074 si->data_blks += (blks); \ 4075 si->bg_data_blks += ((gc_type) == BG_GC) ? (blks) : 0; \ 4076 } while (0) 4077 4078 #define stat_inc_node_blk_count(sbi, blks, gc_type) \ 4079 do { \ 4080 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 4081 stat_inc_tot_blk_count(si, blks); \ 4082 si->node_blks += (blks); \ 4083 si->bg_node_blks += ((gc_type) == BG_GC) ? (blks) : 0; \ 4084 } while (0) 4085 4086 int f2fs_build_stats(struct f2fs_sb_info *sbi); 4087 void f2fs_destroy_stats(struct f2fs_sb_info *sbi); 4088 void __init f2fs_create_root_stats(void); 4089 void f2fs_destroy_root_stats(void); 4090 void f2fs_update_sit_info(struct f2fs_sb_info *sbi); 4091 #else 4092 #define stat_inc_cp_call_count(sbi, foreground) do { } while (0) 4093 #define stat_inc_cp_count(sbi) do { } while (0) 4094 #define stat_io_skip_bggc_count(sbi) do { } while (0) 4095 #define stat_other_skip_bggc_count(sbi) do { } while (0) 4096 #define stat_inc_dirty_inode(sbi, type) do { } while (0) 4097 #define stat_dec_dirty_inode(sbi, type) do { } while (0) 4098 #define stat_inc_total_hit(sbi, type) do { } while (0) 4099 #define stat_inc_rbtree_node_hit(sbi, type) do { } while (0) 4100 #define stat_inc_largest_node_hit(sbi) do { } while (0) 4101 #define stat_inc_cached_node_hit(sbi, type) do { } while (0) 4102 #define stat_inc_inline_xattr(inode) do { } while (0) 4103 #define stat_dec_inline_xattr(inode) do { } while (0) 4104 #define stat_inc_inline_inode(inode) do { } while (0) 4105 #define stat_dec_inline_inode(inode) do { } while (0) 4106 #define stat_inc_inline_dir(inode) do { } while (0) 4107 #define stat_dec_inline_dir(inode) do { } while (0) 4108 #define stat_inc_compr_inode(inode) do { } while (0) 4109 #define stat_dec_compr_inode(inode) do { } while (0) 4110 #define stat_add_compr_blocks(inode, blocks) do { } while (0) 4111 #define stat_sub_compr_blocks(inode, blocks) do { } while (0) 4112 #define stat_inc_swapfile_inode(inode) do { } while (0) 4113 #define stat_dec_swapfile_inode(inode) do { } while (0) 4114 #define stat_inc_atomic_inode(inode) do { } while (0) 4115 #define stat_dec_atomic_inode(inode) do { } while (0) 4116 #define stat_update_max_atomic_write(inode) do { } while (0) 4117 #define stat_inc_meta_count(sbi, blkaddr) do { } while (0) 4118 #define stat_inc_seg_type(sbi, curseg) do { } while (0) 4119 #define stat_inc_block_count(sbi, curseg) do { } while (0) 4120 #define stat_inc_inplace_blocks(sbi) do { } while (0) 4121 #define stat_inc_gc_call_count(sbi, foreground) do { } while (0) 4122 #define stat_inc_gc_sec_count(sbi, type, gc_type) do { } while (0) 4123 #define stat_inc_gc_seg_count(sbi, type, gc_type) do { } while (0) 4124 #define stat_inc_tot_blk_count(si, blks) do { } while (0) 4125 #define stat_inc_data_blk_count(sbi, blks, gc_type) do { } while (0) 4126 #define stat_inc_node_blk_count(sbi, blks, gc_type) do { } while (0) 4127 4128 static inline int f2fs_build_stats(struct f2fs_sb_info *sbi) { return 0; } 4129 static inline void f2fs_destroy_stats(struct f2fs_sb_info *sbi) { } 4130 static inline void __init f2fs_create_root_stats(void) { } 4131 static inline void f2fs_destroy_root_stats(void) { } 4132 static inline void f2fs_update_sit_info(struct f2fs_sb_info *sbi) {} 4133 #endif 4134 4135 extern const struct file_operations f2fs_dir_operations; 4136 extern const struct file_operations f2fs_file_operations; 4137 extern const struct inode_operations f2fs_file_inode_operations; 4138 extern const struct address_space_operations f2fs_dblock_aops; 4139 extern const struct address_space_operations f2fs_node_aops; 4140 extern const struct address_space_operations f2fs_meta_aops; 4141 extern const struct inode_operations f2fs_dir_inode_operations; 4142 extern const struct inode_operations f2fs_symlink_inode_operations; 4143 extern const struct inode_operations f2fs_encrypted_symlink_inode_operations; 4144 extern const struct inode_operations f2fs_special_inode_operations; 4145 extern struct kmem_cache *f2fs_inode_entry_slab; 4146 4147 /* 4148 * inline.c 4149 */ 4150 bool f2fs_may_inline_data(struct inode *inode); 4151 bool f2fs_sanity_check_inline_data(struct inode *inode, struct page *ipage); 4152 bool f2fs_may_inline_dentry(struct inode *inode); 4153 void f2fs_do_read_inline_data(struct page *page, struct page *ipage); 4154 void f2fs_truncate_inline_inode(struct inode *inode, 4155 struct page *ipage, u64 from); 4156 int f2fs_read_inline_data(struct inode *inode, struct page *page); 4157 int f2fs_convert_inline_page(struct dnode_of_data *dn, struct page *page); 4158 int f2fs_convert_inline_inode(struct inode *inode); 4159 int f2fs_try_convert_inline_dir(struct inode *dir, struct dentry *dentry); 4160 int f2fs_write_inline_data(struct inode *inode, struct page *page); 4161 int f2fs_recover_inline_data(struct inode *inode, struct page *npage); 4162 struct f2fs_dir_entry *f2fs_find_in_inline_dir(struct inode *dir, 4163 const struct f2fs_filename *fname, 4164 struct page **res_page); 4165 int f2fs_make_empty_inline_dir(struct inode *inode, struct inode *parent, 4166 struct page *ipage); 4167 int f2fs_add_inline_entry(struct inode *dir, const struct f2fs_filename *fname, 4168 struct inode *inode, nid_t ino, umode_t mode); 4169 void f2fs_delete_inline_entry(struct f2fs_dir_entry *dentry, 4170 struct page *page, struct inode *dir, 4171 struct inode *inode); 4172 bool f2fs_empty_inline_dir(struct inode *dir); 4173 int f2fs_read_inline_dir(struct file *file, struct dir_context *ctx, 4174 struct fscrypt_str *fstr); 4175 int f2fs_inline_data_fiemap(struct inode *inode, 4176 struct fiemap_extent_info *fieinfo, 4177 __u64 start, __u64 len); 4178 4179 /* 4180 * shrinker.c 4181 */ 4182 unsigned long f2fs_shrink_count(struct shrinker *shrink, 4183 struct shrink_control *sc); 4184 unsigned long f2fs_shrink_scan(struct shrinker *shrink, 4185 struct shrink_control *sc); 4186 void f2fs_join_shrinker(struct f2fs_sb_info *sbi); 4187 void f2fs_leave_shrinker(struct f2fs_sb_info *sbi); 4188 4189 /* 4190 * extent_cache.c 4191 */ 4192 bool sanity_check_extent_cache(struct inode *inode, struct page *ipage); 4193 void f2fs_init_extent_tree(struct inode *inode); 4194 void f2fs_drop_extent_tree(struct inode *inode); 4195 void f2fs_destroy_extent_node(struct inode *inode); 4196 void f2fs_destroy_extent_tree(struct inode *inode); 4197 void f2fs_init_extent_cache_info(struct f2fs_sb_info *sbi); 4198 int __init f2fs_create_extent_cache(void); 4199 void f2fs_destroy_extent_cache(void); 4200 4201 /* read extent cache ops */ 4202 void f2fs_init_read_extent_tree(struct inode *inode, struct page *ipage); 4203 bool f2fs_lookup_read_extent_cache(struct inode *inode, pgoff_t pgofs, 4204 struct extent_info *ei); 4205 bool f2fs_lookup_read_extent_cache_block(struct inode *inode, pgoff_t index, 4206 block_t *blkaddr); 4207 void f2fs_update_read_extent_cache(struct dnode_of_data *dn); 4208 void f2fs_update_read_extent_cache_range(struct dnode_of_data *dn, 4209 pgoff_t fofs, block_t blkaddr, unsigned int len); 4210 unsigned int f2fs_shrink_read_extent_tree(struct f2fs_sb_info *sbi, 4211 int nr_shrink); 4212 4213 /* block age extent cache ops */ 4214 void f2fs_init_age_extent_tree(struct inode *inode); 4215 bool f2fs_lookup_age_extent_cache(struct inode *inode, pgoff_t pgofs, 4216 struct extent_info *ei); 4217 void f2fs_update_age_extent_cache(struct dnode_of_data *dn); 4218 void f2fs_update_age_extent_cache_range(struct dnode_of_data *dn, 4219 pgoff_t fofs, unsigned int len); 4220 unsigned int f2fs_shrink_age_extent_tree(struct f2fs_sb_info *sbi, 4221 int nr_shrink); 4222 4223 /* 4224 * sysfs.c 4225 */ 4226 #define MIN_RA_MUL 2 4227 #define MAX_RA_MUL 256 4228 4229 int __init f2fs_init_sysfs(void); 4230 void f2fs_exit_sysfs(void); 4231 int f2fs_register_sysfs(struct f2fs_sb_info *sbi); 4232 void f2fs_unregister_sysfs(struct f2fs_sb_info *sbi); 4233 4234 /* verity.c */ 4235 extern const struct fsverity_operations f2fs_verityops; 4236 4237 /* 4238 * crypto support 4239 */ 4240 static inline bool f2fs_encrypted_file(struct inode *inode) 4241 { 4242 return IS_ENCRYPTED(inode) && S_ISREG(inode->i_mode); 4243 } 4244 4245 static inline void f2fs_set_encrypted_inode(struct inode *inode) 4246 { 4247 #ifdef CONFIG_FS_ENCRYPTION 4248 file_set_encrypt(inode); 4249 f2fs_set_inode_flags(inode); 4250 #endif 4251 } 4252 4253 /* 4254 * Returns true if the reads of the inode's data need to undergo some 4255 * postprocessing step, like decryption or authenticity verification. 4256 */ 4257 static inline bool f2fs_post_read_required(struct inode *inode) 4258 { 4259 return f2fs_encrypted_file(inode) || fsverity_active(inode) || 4260 f2fs_compressed_file(inode); 4261 } 4262 4263 static inline bool f2fs_used_in_atomic_write(struct inode *inode) 4264 { 4265 return f2fs_is_atomic_file(inode) || f2fs_is_cow_file(inode); 4266 } 4267 4268 static inline bool f2fs_meta_inode_gc_required(struct inode *inode) 4269 { 4270 return f2fs_post_read_required(inode) || f2fs_used_in_atomic_write(inode); 4271 } 4272 4273 /* 4274 * compress.c 4275 */ 4276 #ifdef CONFIG_F2FS_FS_COMPRESSION 4277 bool f2fs_is_compressed_page(struct page *page); 4278 struct page *f2fs_compress_control_page(struct page *page); 4279 int f2fs_prepare_compress_overwrite(struct inode *inode, 4280 struct page **pagep, pgoff_t index, void **fsdata); 4281 bool f2fs_compress_write_end(struct inode *inode, void *fsdata, 4282 pgoff_t index, unsigned copied); 4283 int f2fs_truncate_partial_cluster(struct inode *inode, u64 from, bool lock); 4284 void f2fs_compress_write_end_io(struct bio *bio, struct page *page); 4285 bool f2fs_is_compress_backend_ready(struct inode *inode); 4286 bool f2fs_is_compress_level_valid(int alg, int lvl); 4287 int __init f2fs_init_compress_mempool(void); 4288 void f2fs_destroy_compress_mempool(void); 4289 void f2fs_decompress_cluster(struct decompress_io_ctx *dic, bool in_task); 4290 void f2fs_end_read_compressed_page(struct page *page, bool failed, 4291 block_t blkaddr, bool in_task); 4292 bool f2fs_cluster_is_empty(struct compress_ctx *cc); 4293 bool f2fs_cluster_can_merge_page(struct compress_ctx *cc, pgoff_t index); 4294 bool f2fs_all_cluster_page_ready(struct compress_ctx *cc, struct page **pages, 4295 int index, int nr_pages, bool uptodate); 4296 bool f2fs_sanity_check_cluster(struct dnode_of_data *dn); 4297 void f2fs_compress_ctx_add_page(struct compress_ctx *cc, struct page *page); 4298 int f2fs_write_multi_pages(struct compress_ctx *cc, 4299 int *submitted, 4300 struct writeback_control *wbc, 4301 enum iostat_type io_type); 4302 int f2fs_is_compressed_cluster(struct inode *inode, pgoff_t index); 4303 void f2fs_update_read_extent_tree_range_compressed(struct inode *inode, 4304 pgoff_t fofs, block_t blkaddr, 4305 unsigned int llen, unsigned int c_len); 4306 int f2fs_read_multi_pages(struct compress_ctx *cc, struct bio **bio_ret, 4307 unsigned nr_pages, sector_t *last_block_in_bio, 4308 bool is_readahead, bool for_write); 4309 struct decompress_io_ctx *f2fs_alloc_dic(struct compress_ctx *cc); 4310 void f2fs_decompress_end_io(struct decompress_io_ctx *dic, bool failed, 4311 bool in_task); 4312 void f2fs_put_page_dic(struct page *page, bool in_task); 4313 unsigned int f2fs_cluster_blocks_are_contiguous(struct dnode_of_data *dn, 4314 unsigned int ofs_in_node); 4315 int f2fs_init_compress_ctx(struct compress_ctx *cc); 4316 void f2fs_destroy_compress_ctx(struct compress_ctx *cc, bool reuse); 4317 void f2fs_init_compress_info(struct f2fs_sb_info *sbi); 4318 int f2fs_init_compress_inode(struct f2fs_sb_info *sbi); 4319 void f2fs_destroy_compress_inode(struct f2fs_sb_info *sbi); 4320 int f2fs_init_page_array_cache(struct f2fs_sb_info *sbi); 4321 void f2fs_destroy_page_array_cache(struct f2fs_sb_info *sbi); 4322 int __init f2fs_init_compress_cache(void); 4323 void f2fs_destroy_compress_cache(void); 4324 struct address_space *COMPRESS_MAPPING(struct f2fs_sb_info *sbi); 4325 void f2fs_invalidate_compress_page(struct f2fs_sb_info *sbi, block_t blkaddr); 4326 void f2fs_cache_compressed_page(struct f2fs_sb_info *sbi, struct page *page, 4327 nid_t ino, block_t blkaddr); 4328 bool f2fs_load_compressed_page(struct f2fs_sb_info *sbi, struct page *page, 4329 block_t blkaddr); 4330 void f2fs_invalidate_compress_pages(struct f2fs_sb_info *sbi, nid_t ino); 4331 #define inc_compr_inode_stat(inode) \ 4332 do { \ 4333 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); \ 4334 sbi->compr_new_inode++; \ 4335 } while (0) 4336 #define add_compr_block_stat(inode, blocks) \ 4337 do { \ 4338 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); \ 4339 int diff = F2FS_I(inode)->i_cluster_size - blocks; \ 4340 sbi->compr_written_block += blocks; \ 4341 sbi->compr_saved_block += diff; \ 4342 } while (0) 4343 #else 4344 static inline bool f2fs_is_compressed_page(struct page *page) { return false; } 4345 static inline bool f2fs_is_compress_backend_ready(struct inode *inode) 4346 { 4347 if (!f2fs_compressed_file(inode)) 4348 return true; 4349 /* not support compression */ 4350 return false; 4351 } 4352 static inline bool f2fs_is_compress_level_valid(int alg, int lvl) { return false; } 4353 static inline struct page *f2fs_compress_control_page(struct page *page) 4354 { 4355 WARN_ON_ONCE(1); 4356 return ERR_PTR(-EINVAL); 4357 } 4358 static inline int __init f2fs_init_compress_mempool(void) { return 0; } 4359 static inline void f2fs_destroy_compress_mempool(void) { } 4360 static inline void f2fs_decompress_cluster(struct decompress_io_ctx *dic, 4361 bool in_task) { } 4362 static inline void f2fs_end_read_compressed_page(struct page *page, 4363 bool failed, block_t blkaddr, bool in_task) 4364 { 4365 WARN_ON_ONCE(1); 4366 } 4367 static inline void f2fs_put_page_dic(struct page *page, bool in_task) 4368 { 4369 WARN_ON_ONCE(1); 4370 } 4371 static inline unsigned int f2fs_cluster_blocks_are_contiguous( 4372 struct dnode_of_data *dn, unsigned int ofs_in_node) { return 0; } 4373 static inline bool f2fs_sanity_check_cluster(struct dnode_of_data *dn) { return false; } 4374 static inline int f2fs_init_compress_inode(struct f2fs_sb_info *sbi) { return 0; } 4375 static inline void f2fs_destroy_compress_inode(struct f2fs_sb_info *sbi) { } 4376 static inline int f2fs_init_page_array_cache(struct f2fs_sb_info *sbi) { return 0; } 4377 static inline void f2fs_destroy_page_array_cache(struct f2fs_sb_info *sbi) { } 4378 static inline int __init f2fs_init_compress_cache(void) { return 0; } 4379 static inline void f2fs_destroy_compress_cache(void) { } 4380 static inline void f2fs_invalidate_compress_page(struct f2fs_sb_info *sbi, 4381 block_t blkaddr) { } 4382 static inline void f2fs_cache_compressed_page(struct f2fs_sb_info *sbi, 4383 struct page *page, nid_t ino, block_t blkaddr) { } 4384 static inline bool f2fs_load_compressed_page(struct f2fs_sb_info *sbi, 4385 struct page *page, block_t blkaddr) { return false; } 4386 static inline void f2fs_invalidate_compress_pages(struct f2fs_sb_info *sbi, 4387 nid_t ino) { } 4388 #define inc_compr_inode_stat(inode) do { } while (0) 4389 static inline void f2fs_update_read_extent_tree_range_compressed( 4390 struct inode *inode, 4391 pgoff_t fofs, block_t blkaddr, 4392 unsigned int llen, unsigned int c_len) { } 4393 #endif 4394 4395 static inline int set_compress_context(struct inode *inode) 4396 { 4397 #ifdef CONFIG_F2FS_FS_COMPRESSION 4398 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 4399 4400 F2FS_I(inode)->i_compress_algorithm = 4401 F2FS_OPTION(sbi).compress_algorithm; 4402 F2FS_I(inode)->i_log_cluster_size = 4403 F2FS_OPTION(sbi).compress_log_size; 4404 F2FS_I(inode)->i_compress_flag = 4405 F2FS_OPTION(sbi).compress_chksum ? 4406 BIT(COMPRESS_CHKSUM) : 0; 4407 F2FS_I(inode)->i_cluster_size = 4408 BIT(F2FS_I(inode)->i_log_cluster_size); 4409 if ((F2FS_I(inode)->i_compress_algorithm == COMPRESS_LZ4 || 4410 F2FS_I(inode)->i_compress_algorithm == COMPRESS_ZSTD) && 4411 F2FS_OPTION(sbi).compress_level) 4412 F2FS_I(inode)->i_compress_level = 4413 F2FS_OPTION(sbi).compress_level; 4414 F2FS_I(inode)->i_flags |= F2FS_COMPR_FL; 4415 set_inode_flag(inode, FI_COMPRESSED_FILE); 4416 stat_inc_compr_inode(inode); 4417 inc_compr_inode_stat(inode); 4418 f2fs_mark_inode_dirty_sync(inode, true); 4419 return 0; 4420 #else 4421 return -EOPNOTSUPP; 4422 #endif 4423 } 4424 4425 static inline bool f2fs_disable_compressed_file(struct inode *inode) 4426 { 4427 struct f2fs_inode_info *fi = F2FS_I(inode); 4428 4429 f2fs_down_write(&F2FS_I(inode)->i_sem); 4430 4431 if (!f2fs_compressed_file(inode)) { 4432 f2fs_up_write(&F2FS_I(inode)->i_sem); 4433 return true; 4434 } 4435 if (f2fs_is_mmap_file(inode) || 4436 (S_ISREG(inode->i_mode) && F2FS_HAS_BLOCKS(inode))) { 4437 f2fs_up_write(&F2FS_I(inode)->i_sem); 4438 return false; 4439 } 4440 4441 fi->i_flags &= ~F2FS_COMPR_FL; 4442 stat_dec_compr_inode(inode); 4443 clear_inode_flag(inode, FI_COMPRESSED_FILE); 4444 f2fs_mark_inode_dirty_sync(inode, true); 4445 4446 f2fs_up_write(&F2FS_I(inode)->i_sem); 4447 return true; 4448 } 4449 4450 #define F2FS_FEATURE_FUNCS(name, flagname) \ 4451 static inline bool f2fs_sb_has_##name(struct f2fs_sb_info *sbi) \ 4452 { \ 4453 return F2FS_HAS_FEATURE(sbi, F2FS_FEATURE_##flagname); \ 4454 } 4455 4456 F2FS_FEATURE_FUNCS(encrypt, ENCRYPT); 4457 F2FS_FEATURE_FUNCS(blkzoned, BLKZONED); 4458 F2FS_FEATURE_FUNCS(extra_attr, EXTRA_ATTR); 4459 F2FS_FEATURE_FUNCS(project_quota, PRJQUOTA); 4460 F2FS_FEATURE_FUNCS(inode_chksum, INODE_CHKSUM); 4461 F2FS_FEATURE_FUNCS(flexible_inline_xattr, FLEXIBLE_INLINE_XATTR); 4462 F2FS_FEATURE_FUNCS(quota_ino, QUOTA_INO); 4463 F2FS_FEATURE_FUNCS(inode_crtime, INODE_CRTIME); 4464 F2FS_FEATURE_FUNCS(lost_found, LOST_FOUND); 4465 F2FS_FEATURE_FUNCS(verity, VERITY); 4466 F2FS_FEATURE_FUNCS(sb_chksum, SB_CHKSUM); 4467 F2FS_FEATURE_FUNCS(casefold, CASEFOLD); 4468 F2FS_FEATURE_FUNCS(compression, COMPRESSION); 4469 F2FS_FEATURE_FUNCS(readonly, RO); 4470 4471 #ifdef CONFIG_BLK_DEV_ZONED 4472 static inline bool f2fs_blkz_is_seq(struct f2fs_sb_info *sbi, int devi, 4473 block_t blkaddr) 4474 { 4475 unsigned int zno = blkaddr / sbi->blocks_per_blkz; 4476 4477 return test_bit(zno, FDEV(devi).blkz_seq); 4478 } 4479 #endif 4480 4481 static inline int f2fs_bdev_index(struct f2fs_sb_info *sbi, 4482 struct block_device *bdev) 4483 { 4484 int i; 4485 4486 if (!f2fs_is_multi_device(sbi)) 4487 return 0; 4488 4489 for (i = 0; i < sbi->s_ndevs; i++) 4490 if (FDEV(i).bdev == bdev) 4491 return i; 4492 4493 WARN_ON(1); 4494 return -1; 4495 } 4496 4497 static inline bool f2fs_hw_should_discard(struct f2fs_sb_info *sbi) 4498 { 4499 return f2fs_sb_has_blkzoned(sbi); 4500 } 4501 4502 static inline bool f2fs_bdev_support_discard(struct block_device *bdev) 4503 { 4504 return bdev_max_discard_sectors(bdev) || bdev_is_zoned(bdev); 4505 } 4506 4507 static inline bool f2fs_hw_support_discard(struct f2fs_sb_info *sbi) 4508 { 4509 int i; 4510 4511 if (!f2fs_is_multi_device(sbi)) 4512 return f2fs_bdev_support_discard(sbi->sb->s_bdev); 4513 4514 for (i = 0; i < sbi->s_ndevs; i++) 4515 if (f2fs_bdev_support_discard(FDEV(i).bdev)) 4516 return true; 4517 return false; 4518 } 4519 4520 static inline bool f2fs_realtime_discard_enable(struct f2fs_sb_info *sbi) 4521 { 4522 return (test_opt(sbi, DISCARD) && f2fs_hw_support_discard(sbi)) || 4523 f2fs_hw_should_discard(sbi); 4524 } 4525 4526 static inline bool f2fs_hw_is_readonly(struct f2fs_sb_info *sbi) 4527 { 4528 int i; 4529 4530 if (!f2fs_is_multi_device(sbi)) 4531 return bdev_read_only(sbi->sb->s_bdev); 4532 4533 for (i = 0; i < sbi->s_ndevs; i++) 4534 if (bdev_read_only(FDEV(i).bdev)) 4535 return true; 4536 return false; 4537 } 4538 4539 static inline bool f2fs_dev_is_readonly(struct f2fs_sb_info *sbi) 4540 { 4541 return f2fs_sb_has_readonly(sbi) || f2fs_hw_is_readonly(sbi); 4542 } 4543 4544 static inline bool f2fs_lfs_mode(struct f2fs_sb_info *sbi) 4545 { 4546 return F2FS_OPTION(sbi).fs_mode == FS_MODE_LFS; 4547 } 4548 4549 static inline bool f2fs_valid_pinned_area(struct f2fs_sb_info *sbi, 4550 block_t blkaddr) 4551 { 4552 if (f2fs_sb_has_blkzoned(sbi)) { 4553 int devi = f2fs_target_device_index(sbi, blkaddr); 4554 4555 return !bdev_is_zoned(FDEV(devi).bdev); 4556 } 4557 return true; 4558 } 4559 4560 static inline bool f2fs_low_mem_mode(struct f2fs_sb_info *sbi) 4561 { 4562 return F2FS_OPTION(sbi).memory_mode == MEMORY_MODE_LOW; 4563 } 4564 4565 static inline bool f2fs_may_compress(struct inode *inode) 4566 { 4567 if (IS_SWAPFILE(inode) || f2fs_is_pinned_file(inode) || 4568 f2fs_is_atomic_file(inode) || f2fs_has_inline_data(inode) || 4569 f2fs_is_mmap_file(inode)) 4570 return false; 4571 return S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode); 4572 } 4573 4574 static inline void f2fs_i_compr_blocks_update(struct inode *inode, 4575 u64 blocks, bool add) 4576 { 4577 struct f2fs_inode_info *fi = F2FS_I(inode); 4578 int diff = fi->i_cluster_size - blocks; 4579 4580 /* don't update i_compr_blocks if saved blocks were released */ 4581 if (!add && !atomic_read(&fi->i_compr_blocks)) 4582 return; 4583 4584 if (add) { 4585 atomic_add(diff, &fi->i_compr_blocks); 4586 stat_add_compr_blocks(inode, diff); 4587 } else { 4588 atomic_sub(diff, &fi->i_compr_blocks); 4589 stat_sub_compr_blocks(inode, diff); 4590 } 4591 f2fs_mark_inode_dirty_sync(inode, true); 4592 } 4593 4594 static inline bool f2fs_allow_multi_device_dio(struct f2fs_sb_info *sbi, 4595 int flag) 4596 { 4597 if (!f2fs_is_multi_device(sbi)) 4598 return false; 4599 if (flag != F2FS_GET_BLOCK_DIO) 4600 return false; 4601 return sbi->aligned_blksize; 4602 } 4603 4604 static inline bool f2fs_need_verity(const struct inode *inode, pgoff_t idx) 4605 { 4606 return fsverity_active(inode) && 4607 idx < DIV_ROUND_UP(inode->i_size, PAGE_SIZE); 4608 } 4609 4610 #ifdef CONFIG_F2FS_FAULT_INJECTION 4611 extern int f2fs_build_fault_attr(struct f2fs_sb_info *sbi, unsigned long rate, 4612 unsigned long type); 4613 #else 4614 static inline int f2fs_build_fault_attr(struct f2fs_sb_info *sbi, 4615 unsigned long rate, unsigned long type) 4616 { 4617 return 0; 4618 } 4619 #endif 4620 4621 static inline bool is_journalled_quota(struct f2fs_sb_info *sbi) 4622 { 4623 #ifdef CONFIG_QUOTA 4624 if (f2fs_sb_has_quota_ino(sbi)) 4625 return true; 4626 if (F2FS_OPTION(sbi).s_qf_names[USRQUOTA] || 4627 F2FS_OPTION(sbi).s_qf_names[GRPQUOTA] || 4628 F2FS_OPTION(sbi).s_qf_names[PRJQUOTA]) 4629 return true; 4630 #endif 4631 return false; 4632 } 4633 4634 static inline bool f2fs_block_unit_discard(struct f2fs_sb_info *sbi) 4635 { 4636 return F2FS_OPTION(sbi).discard_unit == DISCARD_UNIT_BLOCK; 4637 } 4638 4639 static inline void f2fs_io_schedule_timeout(long timeout) 4640 { 4641 set_current_state(TASK_UNINTERRUPTIBLE); 4642 io_schedule_timeout(timeout); 4643 } 4644 4645 static inline void f2fs_handle_page_eio(struct f2fs_sb_info *sbi, pgoff_t ofs, 4646 enum page_type type) 4647 { 4648 if (unlikely(f2fs_cp_error(sbi))) 4649 return; 4650 4651 if (ofs == sbi->page_eio_ofs[type]) { 4652 if (sbi->page_eio_cnt[type]++ == MAX_RETRY_PAGE_EIO) 4653 set_ckpt_flags(sbi, CP_ERROR_FLAG); 4654 } else { 4655 sbi->page_eio_ofs[type] = ofs; 4656 sbi->page_eio_cnt[type] = 0; 4657 } 4658 } 4659 4660 static inline bool f2fs_is_readonly(struct f2fs_sb_info *sbi) 4661 { 4662 return f2fs_sb_has_readonly(sbi) || f2fs_readonly(sbi->sb); 4663 } 4664 4665 static inline void f2fs_truncate_meta_inode_pages(struct f2fs_sb_info *sbi, 4666 block_t blkaddr, unsigned int cnt) 4667 { 4668 bool need_submit = false; 4669 int i = 0; 4670 4671 do { 4672 struct page *page; 4673 4674 page = find_get_page(META_MAPPING(sbi), blkaddr + i); 4675 if (page) { 4676 if (PageWriteback(page)) 4677 need_submit = true; 4678 f2fs_put_page(page, 0); 4679 } 4680 } while (++i < cnt && !need_submit); 4681 4682 if (need_submit) 4683 f2fs_submit_merged_write_cond(sbi, sbi->meta_inode, 4684 NULL, 0, DATA); 4685 4686 truncate_inode_pages_range(META_MAPPING(sbi), 4687 F2FS_BLK_TO_BYTES((loff_t)blkaddr), 4688 F2FS_BLK_END_BYTES((loff_t)(blkaddr + cnt - 1))); 4689 } 4690 4691 static inline void f2fs_invalidate_internal_cache(struct f2fs_sb_info *sbi, 4692 block_t blkaddr) 4693 { 4694 f2fs_truncate_meta_inode_pages(sbi, blkaddr, 1); 4695 f2fs_invalidate_compress_page(sbi, blkaddr); 4696 } 4697 4698 #define EFSBADCRC EBADMSG /* Bad CRC detected */ 4699 #define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */ 4700 4701 #endif /* _LINUX_F2FS_H */ 4702