1 /* SPDX-License-Identifier: GPL-2.0 */ 2 /* 3 * fs/f2fs/f2fs.h 4 * 5 * Copyright (c) 2012 Samsung Electronics Co., Ltd. 6 * http://www.samsung.com/ 7 */ 8 #ifndef _LINUX_F2FS_H 9 #define _LINUX_F2FS_H 10 11 #include <linux/uio.h> 12 #include <linux/types.h> 13 #include <linux/page-flags.h> 14 #include <linux/buffer_head.h> 15 #include <linux/slab.h> 16 #include <linux/crc32.h> 17 #include <linux/magic.h> 18 #include <linux/kobject.h> 19 #include <linux/sched.h> 20 #include <linux/cred.h> 21 #include <linux/sched/mm.h> 22 #include <linux/vmalloc.h> 23 #include <linux/bio.h> 24 #include <linux/blkdev.h> 25 #include <linux/quotaops.h> 26 #include <linux/part_stat.h> 27 #include <crypto/hash.h> 28 29 #include <linux/fscrypt.h> 30 #include <linux/fsverity.h> 31 32 struct pagevec; 33 34 #ifdef CONFIG_F2FS_CHECK_FS 35 #define f2fs_bug_on(sbi, condition) BUG_ON(condition) 36 #else 37 #define f2fs_bug_on(sbi, condition) \ 38 do { \ 39 if (WARN_ON(condition)) \ 40 set_sbi_flag(sbi, SBI_NEED_FSCK); \ 41 } while (0) 42 #endif 43 44 enum { 45 FAULT_KMALLOC, 46 FAULT_KVMALLOC, 47 FAULT_PAGE_ALLOC, 48 FAULT_PAGE_GET, 49 FAULT_ALLOC_BIO, /* it's obsolete due to bio_alloc() will never fail */ 50 FAULT_ALLOC_NID, 51 FAULT_ORPHAN, 52 FAULT_BLOCK, 53 FAULT_DIR_DEPTH, 54 FAULT_EVICT_INODE, 55 FAULT_TRUNCATE, 56 FAULT_READ_IO, 57 FAULT_CHECKPOINT, 58 FAULT_DISCARD, 59 FAULT_WRITE_IO, 60 FAULT_SLAB_ALLOC, 61 FAULT_DQUOT_INIT, 62 FAULT_LOCK_OP, 63 FAULT_BLKADDR, 64 FAULT_MAX, 65 }; 66 67 #ifdef CONFIG_F2FS_FAULT_INJECTION 68 #define F2FS_ALL_FAULT_TYPE (GENMASK(FAULT_MAX - 1, 0)) 69 70 struct f2fs_fault_info { 71 atomic_t inject_ops; 72 unsigned int inject_rate; 73 unsigned int inject_type; 74 }; 75 76 extern const char *f2fs_fault_name[FAULT_MAX]; 77 #define IS_FAULT_SET(fi, type) ((fi)->inject_type & BIT(type)) 78 #endif 79 80 /* 81 * For mount options 82 */ 83 #define F2FS_MOUNT_DISABLE_ROLL_FORWARD 0x00000001 84 #define F2FS_MOUNT_DISCARD 0x00000002 85 #define F2FS_MOUNT_NOHEAP 0x00000004 86 #define F2FS_MOUNT_XATTR_USER 0x00000008 87 #define F2FS_MOUNT_POSIX_ACL 0x00000010 88 #define F2FS_MOUNT_DISABLE_EXT_IDENTIFY 0x00000020 89 #define F2FS_MOUNT_INLINE_XATTR 0x00000040 90 #define F2FS_MOUNT_INLINE_DATA 0x00000080 91 #define F2FS_MOUNT_INLINE_DENTRY 0x00000100 92 #define F2FS_MOUNT_FLUSH_MERGE 0x00000200 93 #define F2FS_MOUNT_NOBARRIER 0x00000400 94 #define F2FS_MOUNT_FASTBOOT 0x00000800 95 #define F2FS_MOUNT_READ_EXTENT_CACHE 0x00001000 96 #define F2FS_MOUNT_DATA_FLUSH 0x00002000 97 #define F2FS_MOUNT_FAULT_INJECTION 0x00004000 98 #define F2FS_MOUNT_USRQUOTA 0x00008000 99 #define F2FS_MOUNT_GRPQUOTA 0x00010000 100 #define F2FS_MOUNT_PRJQUOTA 0x00020000 101 #define F2FS_MOUNT_QUOTA 0x00040000 102 #define F2FS_MOUNT_INLINE_XATTR_SIZE 0x00080000 103 #define F2FS_MOUNT_RESERVE_ROOT 0x00100000 104 #define F2FS_MOUNT_DISABLE_CHECKPOINT 0x00200000 105 #define F2FS_MOUNT_NORECOVERY 0x00400000 106 #define F2FS_MOUNT_ATGC 0x00800000 107 #define F2FS_MOUNT_MERGE_CHECKPOINT 0x01000000 108 #define F2FS_MOUNT_GC_MERGE 0x02000000 109 #define F2FS_MOUNT_COMPRESS_CACHE 0x04000000 110 #define F2FS_MOUNT_AGE_EXTENT_CACHE 0x08000000 111 112 #define F2FS_OPTION(sbi) ((sbi)->mount_opt) 113 #define clear_opt(sbi, option) (F2FS_OPTION(sbi).opt &= ~F2FS_MOUNT_##option) 114 #define set_opt(sbi, option) (F2FS_OPTION(sbi).opt |= F2FS_MOUNT_##option) 115 #define test_opt(sbi, option) (F2FS_OPTION(sbi).opt & F2FS_MOUNT_##option) 116 117 #define ver_after(a, b) (typecheck(unsigned long long, a) && \ 118 typecheck(unsigned long long, b) && \ 119 ((long long)((a) - (b)) > 0)) 120 121 typedef u32 block_t; /* 122 * should not change u32, since it is the on-disk block 123 * address format, __le32. 124 */ 125 typedef u32 nid_t; 126 127 #define COMPRESS_EXT_NUM 16 128 129 /* 130 * An implementation of an rwsem that is explicitly unfair to readers. This 131 * prevents priority inversion when a low-priority reader acquires the read lock 132 * while sleeping on the write lock but the write lock is needed by 133 * higher-priority clients. 134 */ 135 136 struct f2fs_rwsem { 137 struct rw_semaphore internal_rwsem; 138 #ifdef CONFIG_F2FS_UNFAIR_RWSEM 139 wait_queue_head_t read_waiters; 140 #endif 141 }; 142 143 struct f2fs_mount_info { 144 unsigned int opt; 145 int write_io_size_bits; /* Write IO size bits */ 146 block_t root_reserved_blocks; /* root reserved blocks */ 147 kuid_t s_resuid; /* reserved blocks for uid */ 148 kgid_t s_resgid; /* reserved blocks for gid */ 149 int active_logs; /* # of active logs */ 150 int inline_xattr_size; /* inline xattr size */ 151 #ifdef CONFIG_F2FS_FAULT_INJECTION 152 struct f2fs_fault_info fault_info; /* For fault injection */ 153 #endif 154 #ifdef CONFIG_QUOTA 155 /* Names of quota files with journalled quota */ 156 char *s_qf_names[MAXQUOTAS]; 157 int s_jquota_fmt; /* Format of quota to use */ 158 #endif 159 /* For which write hints are passed down to block layer */ 160 int alloc_mode; /* segment allocation policy */ 161 int fsync_mode; /* fsync policy */ 162 int fs_mode; /* fs mode: LFS or ADAPTIVE */ 163 int bggc_mode; /* bggc mode: off, on or sync */ 164 int memory_mode; /* memory mode */ 165 int errors; /* errors parameter */ 166 int discard_unit; /* 167 * discard command's offset/size should 168 * be aligned to this unit: block, 169 * segment or section 170 */ 171 struct fscrypt_dummy_policy dummy_enc_policy; /* test dummy encryption */ 172 block_t unusable_cap_perc; /* percentage for cap */ 173 block_t unusable_cap; /* Amount of space allowed to be 174 * unusable when disabling checkpoint 175 */ 176 177 /* For compression */ 178 unsigned char compress_algorithm; /* algorithm type */ 179 unsigned char compress_log_size; /* cluster log size */ 180 unsigned char compress_level; /* compress level */ 181 bool compress_chksum; /* compressed data chksum */ 182 unsigned char compress_ext_cnt; /* extension count */ 183 unsigned char nocompress_ext_cnt; /* nocompress extension count */ 184 int compress_mode; /* compression mode */ 185 unsigned char extensions[COMPRESS_EXT_NUM][F2FS_EXTENSION_LEN]; /* extensions */ 186 unsigned char noextensions[COMPRESS_EXT_NUM][F2FS_EXTENSION_LEN]; /* extensions */ 187 }; 188 189 #define F2FS_FEATURE_ENCRYPT 0x00000001 190 #define F2FS_FEATURE_BLKZONED 0x00000002 191 #define F2FS_FEATURE_ATOMIC_WRITE 0x00000004 192 #define F2FS_FEATURE_EXTRA_ATTR 0x00000008 193 #define F2FS_FEATURE_PRJQUOTA 0x00000010 194 #define F2FS_FEATURE_INODE_CHKSUM 0x00000020 195 #define F2FS_FEATURE_FLEXIBLE_INLINE_XATTR 0x00000040 196 #define F2FS_FEATURE_QUOTA_INO 0x00000080 197 #define F2FS_FEATURE_INODE_CRTIME 0x00000100 198 #define F2FS_FEATURE_LOST_FOUND 0x00000200 199 #define F2FS_FEATURE_VERITY 0x00000400 200 #define F2FS_FEATURE_SB_CHKSUM 0x00000800 201 #define F2FS_FEATURE_CASEFOLD 0x00001000 202 #define F2FS_FEATURE_COMPRESSION 0x00002000 203 #define F2FS_FEATURE_RO 0x00004000 204 205 #define __F2FS_HAS_FEATURE(raw_super, mask) \ 206 ((raw_super->feature & cpu_to_le32(mask)) != 0) 207 #define F2FS_HAS_FEATURE(sbi, mask) __F2FS_HAS_FEATURE(sbi->raw_super, mask) 208 209 /* 210 * Default values for user and/or group using reserved blocks 211 */ 212 #define F2FS_DEF_RESUID 0 213 #define F2FS_DEF_RESGID 0 214 215 /* 216 * For checkpoint manager 217 */ 218 enum { 219 NAT_BITMAP, 220 SIT_BITMAP 221 }; 222 223 #define CP_UMOUNT 0x00000001 224 #define CP_FASTBOOT 0x00000002 225 #define CP_SYNC 0x00000004 226 #define CP_RECOVERY 0x00000008 227 #define CP_DISCARD 0x00000010 228 #define CP_TRIMMED 0x00000020 229 #define CP_PAUSE 0x00000040 230 #define CP_RESIZE 0x00000080 231 232 #define DEF_MAX_DISCARD_REQUEST 8 /* issue 8 discards per round */ 233 #define DEF_MIN_DISCARD_ISSUE_TIME 50 /* 50 ms, if exists */ 234 #define DEF_MID_DISCARD_ISSUE_TIME 500 /* 500 ms, if device busy */ 235 #define DEF_MAX_DISCARD_ISSUE_TIME 60000 /* 60 s, if no candidates */ 236 #define DEF_DISCARD_URGENT_UTIL 80 /* do more discard over 80% */ 237 #define DEF_CP_INTERVAL 60 /* 60 secs */ 238 #define DEF_IDLE_INTERVAL 5 /* 5 secs */ 239 #define DEF_DISABLE_INTERVAL 5 /* 5 secs */ 240 #define DEF_DISABLE_QUICK_INTERVAL 1 /* 1 secs */ 241 #define DEF_UMOUNT_DISCARD_TIMEOUT 5 /* 5 secs */ 242 243 struct cp_control { 244 int reason; 245 __u64 trim_start; 246 __u64 trim_end; 247 __u64 trim_minlen; 248 }; 249 250 /* 251 * indicate meta/data type 252 */ 253 enum { 254 META_CP, 255 META_NAT, 256 META_SIT, 257 META_SSA, 258 META_MAX, 259 META_POR, 260 DATA_GENERIC, /* check range only */ 261 DATA_GENERIC_ENHANCE, /* strong check on range and segment bitmap */ 262 DATA_GENERIC_ENHANCE_READ, /* 263 * strong check on range and segment 264 * bitmap but no warning due to race 265 * condition of read on truncated area 266 * by extent_cache 267 */ 268 DATA_GENERIC_ENHANCE_UPDATE, /* 269 * strong check on range and segment 270 * bitmap for update case 271 */ 272 META_GENERIC, 273 }; 274 275 /* for the list of ino */ 276 enum { 277 ORPHAN_INO, /* for orphan ino list */ 278 APPEND_INO, /* for append ino list */ 279 UPDATE_INO, /* for update ino list */ 280 TRANS_DIR_INO, /* for transactions dir ino list */ 281 FLUSH_INO, /* for multiple device flushing */ 282 MAX_INO_ENTRY, /* max. list */ 283 }; 284 285 struct ino_entry { 286 struct list_head list; /* list head */ 287 nid_t ino; /* inode number */ 288 unsigned int dirty_device; /* dirty device bitmap */ 289 }; 290 291 /* for the list of inodes to be GCed */ 292 struct inode_entry { 293 struct list_head list; /* list head */ 294 struct inode *inode; /* vfs inode pointer */ 295 }; 296 297 struct fsync_node_entry { 298 struct list_head list; /* list head */ 299 struct page *page; /* warm node page pointer */ 300 unsigned int seq_id; /* sequence id */ 301 }; 302 303 struct ckpt_req { 304 struct completion wait; /* completion for checkpoint done */ 305 struct llist_node llnode; /* llist_node to be linked in wait queue */ 306 int ret; /* return code of checkpoint */ 307 ktime_t queue_time; /* request queued time */ 308 }; 309 310 struct ckpt_req_control { 311 struct task_struct *f2fs_issue_ckpt; /* checkpoint task */ 312 int ckpt_thread_ioprio; /* checkpoint merge thread ioprio */ 313 wait_queue_head_t ckpt_wait_queue; /* waiting queue for wake-up */ 314 atomic_t issued_ckpt; /* # of actually issued ckpts */ 315 atomic_t total_ckpt; /* # of total ckpts */ 316 atomic_t queued_ckpt; /* # of queued ckpts */ 317 struct llist_head issue_list; /* list for command issue */ 318 spinlock_t stat_lock; /* lock for below checkpoint time stats */ 319 unsigned int cur_time; /* cur wait time in msec for currently issued checkpoint */ 320 unsigned int peak_time; /* peak wait time in msec until now */ 321 }; 322 323 /* for the bitmap indicate blocks to be discarded */ 324 struct discard_entry { 325 struct list_head list; /* list head */ 326 block_t start_blkaddr; /* start blockaddr of current segment */ 327 unsigned char discard_map[SIT_VBLOCK_MAP_SIZE]; /* segment discard bitmap */ 328 }; 329 330 /* minimum discard granularity, unit: block count */ 331 #define MIN_DISCARD_GRANULARITY 1 332 /* default discard granularity of inner discard thread, unit: block count */ 333 #define DEFAULT_DISCARD_GRANULARITY 16 334 /* default maximum discard granularity of ordered discard, unit: block count */ 335 #define DEFAULT_MAX_ORDERED_DISCARD_GRANULARITY 16 336 337 /* max discard pend list number */ 338 #define MAX_PLIST_NUM 512 339 #define plist_idx(blk_num) ((blk_num) >= MAX_PLIST_NUM ? \ 340 (MAX_PLIST_NUM - 1) : ((blk_num) - 1)) 341 342 enum { 343 D_PREP, /* initial */ 344 D_PARTIAL, /* partially submitted */ 345 D_SUBMIT, /* all submitted */ 346 D_DONE, /* finished */ 347 }; 348 349 struct discard_info { 350 block_t lstart; /* logical start address */ 351 block_t len; /* length */ 352 block_t start; /* actual start address in dev */ 353 }; 354 355 struct discard_cmd { 356 struct rb_node rb_node; /* rb node located in rb-tree */ 357 struct discard_info di; /* discard info */ 358 struct list_head list; /* command list */ 359 struct completion wait; /* compleation */ 360 struct block_device *bdev; /* bdev */ 361 unsigned short ref; /* reference count */ 362 unsigned char state; /* state */ 363 unsigned char queued; /* queued discard */ 364 int error; /* bio error */ 365 spinlock_t lock; /* for state/bio_ref updating */ 366 unsigned short bio_ref; /* bio reference count */ 367 }; 368 369 enum { 370 DPOLICY_BG, 371 DPOLICY_FORCE, 372 DPOLICY_FSTRIM, 373 DPOLICY_UMOUNT, 374 MAX_DPOLICY, 375 }; 376 377 struct discard_policy { 378 int type; /* type of discard */ 379 unsigned int min_interval; /* used for candidates exist */ 380 unsigned int mid_interval; /* used for device busy */ 381 unsigned int max_interval; /* used for candidates not exist */ 382 unsigned int max_requests; /* # of discards issued per round */ 383 unsigned int io_aware_gran; /* minimum granularity discard not be aware of I/O */ 384 bool io_aware; /* issue discard in idle time */ 385 bool sync; /* submit discard with REQ_SYNC flag */ 386 bool ordered; /* issue discard by lba order */ 387 bool timeout; /* discard timeout for put_super */ 388 unsigned int granularity; /* discard granularity */ 389 }; 390 391 struct discard_cmd_control { 392 struct task_struct *f2fs_issue_discard; /* discard thread */ 393 struct list_head entry_list; /* 4KB discard entry list */ 394 struct list_head pend_list[MAX_PLIST_NUM];/* store pending entries */ 395 struct list_head wait_list; /* store on-flushing entries */ 396 struct list_head fstrim_list; /* in-flight discard from fstrim */ 397 wait_queue_head_t discard_wait_queue; /* waiting queue for wake-up */ 398 struct mutex cmd_lock; 399 unsigned int nr_discards; /* # of discards in the list */ 400 unsigned int max_discards; /* max. discards to be issued */ 401 unsigned int max_discard_request; /* max. discard request per round */ 402 unsigned int min_discard_issue_time; /* min. interval between discard issue */ 403 unsigned int mid_discard_issue_time; /* mid. interval between discard issue */ 404 unsigned int max_discard_issue_time; /* max. interval between discard issue */ 405 unsigned int discard_io_aware_gran; /* minimum discard granularity not be aware of I/O */ 406 unsigned int discard_urgent_util; /* utilization which issue discard proactively */ 407 unsigned int discard_granularity; /* discard granularity */ 408 unsigned int max_ordered_discard; /* maximum discard granularity issued by lba order */ 409 unsigned int undiscard_blks; /* # of undiscard blocks */ 410 unsigned int next_pos; /* next discard position */ 411 atomic_t issued_discard; /* # of issued discard */ 412 atomic_t queued_discard; /* # of queued discard */ 413 atomic_t discard_cmd_cnt; /* # of cached cmd count */ 414 struct rb_root_cached root; /* root of discard rb-tree */ 415 bool rbtree_check; /* config for consistence check */ 416 bool discard_wake; /* to wake up discard thread */ 417 }; 418 419 /* for the list of fsync inodes, used only during recovery */ 420 struct fsync_inode_entry { 421 struct list_head list; /* list head */ 422 struct inode *inode; /* vfs inode pointer */ 423 block_t blkaddr; /* block address locating the last fsync */ 424 block_t last_dentry; /* block address locating the last dentry */ 425 }; 426 427 #define nats_in_cursum(jnl) (le16_to_cpu((jnl)->n_nats)) 428 #define sits_in_cursum(jnl) (le16_to_cpu((jnl)->n_sits)) 429 430 #define nat_in_journal(jnl, i) ((jnl)->nat_j.entries[i].ne) 431 #define nid_in_journal(jnl, i) ((jnl)->nat_j.entries[i].nid) 432 #define sit_in_journal(jnl, i) ((jnl)->sit_j.entries[i].se) 433 #define segno_in_journal(jnl, i) ((jnl)->sit_j.entries[i].segno) 434 435 #define MAX_NAT_JENTRIES(jnl) (NAT_JOURNAL_ENTRIES - nats_in_cursum(jnl)) 436 #define MAX_SIT_JENTRIES(jnl) (SIT_JOURNAL_ENTRIES - sits_in_cursum(jnl)) 437 438 static inline int update_nats_in_cursum(struct f2fs_journal *journal, int i) 439 { 440 int before = nats_in_cursum(journal); 441 442 journal->n_nats = cpu_to_le16(before + i); 443 return before; 444 } 445 446 static inline int update_sits_in_cursum(struct f2fs_journal *journal, int i) 447 { 448 int before = sits_in_cursum(journal); 449 450 journal->n_sits = cpu_to_le16(before + i); 451 return before; 452 } 453 454 static inline bool __has_cursum_space(struct f2fs_journal *journal, 455 int size, int type) 456 { 457 if (type == NAT_JOURNAL) 458 return size <= MAX_NAT_JENTRIES(journal); 459 return size <= MAX_SIT_JENTRIES(journal); 460 } 461 462 /* for inline stuff */ 463 #define DEF_INLINE_RESERVED_SIZE 1 464 static inline int get_extra_isize(struct inode *inode); 465 static inline int get_inline_xattr_addrs(struct inode *inode); 466 #define MAX_INLINE_DATA(inode) (sizeof(__le32) * \ 467 (CUR_ADDRS_PER_INODE(inode) - \ 468 get_inline_xattr_addrs(inode) - \ 469 DEF_INLINE_RESERVED_SIZE)) 470 471 /* for inline dir */ 472 #define NR_INLINE_DENTRY(inode) (MAX_INLINE_DATA(inode) * BITS_PER_BYTE / \ 473 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \ 474 BITS_PER_BYTE + 1)) 475 #define INLINE_DENTRY_BITMAP_SIZE(inode) \ 476 DIV_ROUND_UP(NR_INLINE_DENTRY(inode), BITS_PER_BYTE) 477 #define INLINE_RESERVED_SIZE(inode) (MAX_INLINE_DATA(inode) - \ 478 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \ 479 NR_INLINE_DENTRY(inode) + \ 480 INLINE_DENTRY_BITMAP_SIZE(inode))) 481 482 /* 483 * For INODE and NODE manager 484 */ 485 /* for directory operations */ 486 487 struct f2fs_filename { 488 /* 489 * The filename the user specified. This is NULL for some 490 * filesystem-internal operations, e.g. converting an inline directory 491 * to a non-inline one, or roll-forward recovering an encrypted dentry. 492 */ 493 const struct qstr *usr_fname; 494 495 /* 496 * The on-disk filename. For encrypted directories, this is encrypted. 497 * This may be NULL for lookups in an encrypted dir without the key. 498 */ 499 struct fscrypt_str disk_name; 500 501 /* The dirhash of this filename */ 502 f2fs_hash_t hash; 503 504 #ifdef CONFIG_FS_ENCRYPTION 505 /* 506 * For lookups in encrypted directories: either the buffer backing 507 * disk_name, or a buffer that holds the decoded no-key name. 508 */ 509 struct fscrypt_str crypto_buf; 510 #endif 511 #if IS_ENABLED(CONFIG_UNICODE) 512 /* 513 * For casefolded directories: the casefolded name, but it's left NULL 514 * if the original name is not valid Unicode, if the original name is 515 * "." or "..", if the directory is both casefolded and encrypted and 516 * its encryption key is unavailable, or if the filesystem is doing an 517 * internal operation where usr_fname is also NULL. In all these cases 518 * we fall back to treating the name as an opaque byte sequence. 519 */ 520 struct fscrypt_str cf_name; 521 #endif 522 }; 523 524 struct f2fs_dentry_ptr { 525 struct inode *inode; 526 void *bitmap; 527 struct f2fs_dir_entry *dentry; 528 __u8 (*filename)[F2FS_SLOT_LEN]; 529 int max; 530 int nr_bitmap; 531 }; 532 533 static inline void make_dentry_ptr_block(struct inode *inode, 534 struct f2fs_dentry_ptr *d, struct f2fs_dentry_block *t) 535 { 536 d->inode = inode; 537 d->max = NR_DENTRY_IN_BLOCK; 538 d->nr_bitmap = SIZE_OF_DENTRY_BITMAP; 539 d->bitmap = t->dentry_bitmap; 540 d->dentry = t->dentry; 541 d->filename = t->filename; 542 } 543 544 static inline void make_dentry_ptr_inline(struct inode *inode, 545 struct f2fs_dentry_ptr *d, void *t) 546 { 547 int entry_cnt = NR_INLINE_DENTRY(inode); 548 int bitmap_size = INLINE_DENTRY_BITMAP_SIZE(inode); 549 int reserved_size = INLINE_RESERVED_SIZE(inode); 550 551 d->inode = inode; 552 d->max = entry_cnt; 553 d->nr_bitmap = bitmap_size; 554 d->bitmap = t; 555 d->dentry = t + bitmap_size + reserved_size; 556 d->filename = t + bitmap_size + reserved_size + 557 SIZE_OF_DIR_ENTRY * entry_cnt; 558 } 559 560 /* 561 * XATTR_NODE_OFFSET stores xattrs to one node block per file keeping -1 562 * as its node offset to distinguish from index node blocks. 563 * But some bits are used to mark the node block. 564 */ 565 #define XATTR_NODE_OFFSET ((((unsigned int)-1) << OFFSET_BIT_SHIFT) \ 566 >> OFFSET_BIT_SHIFT) 567 enum { 568 ALLOC_NODE, /* allocate a new node page if needed */ 569 LOOKUP_NODE, /* look up a node without readahead */ 570 LOOKUP_NODE_RA, /* 571 * look up a node with readahead called 572 * by get_data_block. 573 */ 574 }; 575 576 #define DEFAULT_RETRY_IO_COUNT 8 /* maximum retry read IO or flush count */ 577 578 /* congestion wait timeout value, default: 20ms */ 579 #define DEFAULT_IO_TIMEOUT (msecs_to_jiffies(20)) 580 581 /* maximum retry quota flush count */ 582 #define DEFAULT_RETRY_QUOTA_FLUSH_COUNT 8 583 584 /* maximum retry of EIO'ed page */ 585 #define MAX_RETRY_PAGE_EIO 100 586 587 #define F2FS_LINK_MAX 0xffffffff /* maximum link count per file */ 588 589 #define MAX_DIR_RA_PAGES 4 /* maximum ra pages of dir */ 590 591 /* dirty segments threshold for triggering CP */ 592 #define DEFAULT_DIRTY_THRESHOLD 4 593 594 #define RECOVERY_MAX_RA_BLOCKS BIO_MAX_VECS 595 #define RECOVERY_MIN_RA_BLOCKS 1 596 597 #define F2FS_ONSTACK_PAGES 16 /* nr of onstack pages */ 598 599 /* for in-memory extent cache entry */ 600 #define F2FS_MIN_EXTENT_LEN 64 /* minimum extent length */ 601 602 /* number of extent info in extent cache we try to shrink */ 603 #define READ_EXTENT_CACHE_SHRINK_NUMBER 128 604 605 /* number of age extent info in extent cache we try to shrink */ 606 #define AGE_EXTENT_CACHE_SHRINK_NUMBER 128 607 #define LAST_AGE_WEIGHT 30 608 #define SAME_AGE_REGION 1024 609 610 /* 611 * Define data block with age less than 1GB as hot data 612 * define data block with age less than 10GB but more than 1GB as warm data 613 */ 614 #define DEF_HOT_DATA_AGE_THRESHOLD 262144 615 #define DEF_WARM_DATA_AGE_THRESHOLD 2621440 616 617 /* extent cache type */ 618 enum extent_type { 619 EX_READ, 620 EX_BLOCK_AGE, 621 NR_EXTENT_CACHES, 622 }; 623 624 struct extent_info { 625 unsigned int fofs; /* start offset in a file */ 626 unsigned int len; /* length of the extent */ 627 union { 628 /* read extent_cache */ 629 struct { 630 /* start block address of the extent */ 631 block_t blk; 632 #ifdef CONFIG_F2FS_FS_COMPRESSION 633 /* physical extent length of compressed blocks */ 634 unsigned int c_len; 635 #endif 636 }; 637 /* block age extent_cache */ 638 struct { 639 /* block age of the extent */ 640 unsigned long long age; 641 /* last total blocks allocated */ 642 unsigned long long last_blocks; 643 }; 644 }; 645 }; 646 647 struct extent_node { 648 struct rb_node rb_node; /* rb node located in rb-tree */ 649 struct extent_info ei; /* extent info */ 650 struct list_head list; /* node in global extent list of sbi */ 651 struct extent_tree *et; /* extent tree pointer */ 652 }; 653 654 struct extent_tree { 655 nid_t ino; /* inode number */ 656 enum extent_type type; /* keep the extent tree type */ 657 struct rb_root_cached root; /* root of extent info rb-tree */ 658 struct extent_node *cached_en; /* recently accessed extent node */ 659 struct list_head list; /* to be used by sbi->zombie_list */ 660 rwlock_t lock; /* protect extent info rb-tree */ 661 atomic_t node_cnt; /* # of extent node in rb-tree*/ 662 bool largest_updated; /* largest extent updated */ 663 struct extent_info largest; /* largest cached extent for EX_READ */ 664 }; 665 666 struct extent_tree_info { 667 struct radix_tree_root extent_tree_root;/* cache extent cache entries */ 668 struct mutex extent_tree_lock; /* locking extent radix tree */ 669 struct list_head extent_list; /* lru list for shrinker */ 670 spinlock_t extent_lock; /* locking extent lru list */ 671 atomic_t total_ext_tree; /* extent tree count */ 672 struct list_head zombie_list; /* extent zombie tree list */ 673 atomic_t total_zombie_tree; /* extent zombie tree count */ 674 atomic_t total_ext_node; /* extent info count */ 675 }; 676 677 /* 678 * State of block returned by f2fs_map_blocks. 679 */ 680 #define F2FS_MAP_NEW (1U << 0) 681 #define F2FS_MAP_MAPPED (1U << 1) 682 #define F2FS_MAP_DELALLOC (1U << 2) 683 #define F2FS_MAP_FLAGS (F2FS_MAP_NEW | F2FS_MAP_MAPPED |\ 684 F2FS_MAP_DELALLOC) 685 686 struct f2fs_map_blocks { 687 struct block_device *m_bdev; /* for multi-device dio */ 688 block_t m_pblk; 689 block_t m_lblk; 690 unsigned int m_len; 691 unsigned int m_flags; 692 pgoff_t *m_next_pgofs; /* point next possible non-hole pgofs */ 693 pgoff_t *m_next_extent; /* point to next possible extent */ 694 int m_seg_type; 695 bool m_may_create; /* indicate it is from write path */ 696 bool m_multidev_dio; /* indicate it allows multi-device dio */ 697 }; 698 699 /* for flag in get_data_block */ 700 enum { 701 F2FS_GET_BLOCK_DEFAULT, 702 F2FS_GET_BLOCK_FIEMAP, 703 F2FS_GET_BLOCK_BMAP, 704 F2FS_GET_BLOCK_DIO, 705 F2FS_GET_BLOCK_PRE_DIO, 706 F2FS_GET_BLOCK_PRE_AIO, 707 F2FS_GET_BLOCK_PRECACHE, 708 }; 709 710 /* 711 * i_advise uses FADVISE_XXX_BIT. We can add additional hints later. 712 */ 713 #define FADVISE_COLD_BIT 0x01 714 #define FADVISE_LOST_PINO_BIT 0x02 715 #define FADVISE_ENCRYPT_BIT 0x04 716 #define FADVISE_ENC_NAME_BIT 0x08 717 #define FADVISE_KEEP_SIZE_BIT 0x10 718 #define FADVISE_HOT_BIT 0x20 719 #define FADVISE_VERITY_BIT 0x40 720 #define FADVISE_TRUNC_BIT 0x80 721 722 #define FADVISE_MODIFIABLE_BITS (FADVISE_COLD_BIT | FADVISE_HOT_BIT) 723 724 #define file_is_cold(inode) is_file(inode, FADVISE_COLD_BIT) 725 #define file_set_cold(inode) set_file(inode, FADVISE_COLD_BIT) 726 #define file_clear_cold(inode) clear_file(inode, FADVISE_COLD_BIT) 727 728 #define file_wrong_pino(inode) is_file(inode, FADVISE_LOST_PINO_BIT) 729 #define file_lost_pino(inode) set_file(inode, FADVISE_LOST_PINO_BIT) 730 #define file_got_pino(inode) clear_file(inode, FADVISE_LOST_PINO_BIT) 731 732 #define file_is_encrypt(inode) is_file(inode, FADVISE_ENCRYPT_BIT) 733 #define file_set_encrypt(inode) set_file(inode, FADVISE_ENCRYPT_BIT) 734 735 #define file_enc_name(inode) is_file(inode, FADVISE_ENC_NAME_BIT) 736 #define file_set_enc_name(inode) set_file(inode, FADVISE_ENC_NAME_BIT) 737 738 #define file_keep_isize(inode) is_file(inode, FADVISE_KEEP_SIZE_BIT) 739 #define file_set_keep_isize(inode) set_file(inode, FADVISE_KEEP_SIZE_BIT) 740 741 #define file_is_hot(inode) is_file(inode, FADVISE_HOT_BIT) 742 #define file_set_hot(inode) set_file(inode, FADVISE_HOT_BIT) 743 #define file_clear_hot(inode) clear_file(inode, FADVISE_HOT_BIT) 744 745 #define file_is_verity(inode) is_file(inode, FADVISE_VERITY_BIT) 746 #define file_set_verity(inode) set_file(inode, FADVISE_VERITY_BIT) 747 748 #define file_should_truncate(inode) is_file(inode, FADVISE_TRUNC_BIT) 749 #define file_need_truncate(inode) set_file(inode, FADVISE_TRUNC_BIT) 750 #define file_dont_truncate(inode) clear_file(inode, FADVISE_TRUNC_BIT) 751 752 #define DEF_DIR_LEVEL 0 753 754 enum { 755 GC_FAILURE_PIN, 756 MAX_GC_FAILURE 757 }; 758 759 /* used for f2fs_inode_info->flags */ 760 enum { 761 FI_NEW_INODE, /* indicate newly allocated inode */ 762 FI_DIRTY_INODE, /* indicate inode is dirty or not */ 763 FI_AUTO_RECOVER, /* indicate inode is recoverable */ 764 FI_DIRTY_DIR, /* indicate directory has dirty pages */ 765 FI_INC_LINK, /* need to increment i_nlink */ 766 FI_ACL_MODE, /* indicate acl mode */ 767 FI_NO_ALLOC, /* should not allocate any blocks */ 768 FI_FREE_NID, /* free allocated nide */ 769 FI_NO_EXTENT, /* not to use the extent cache */ 770 FI_INLINE_XATTR, /* used for inline xattr */ 771 FI_INLINE_DATA, /* used for inline data*/ 772 FI_INLINE_DENTRY, /* used for inline dentry */ 773 FI_APPEND_WRITE, /* inode has appended data */ 774 FI_UPDATE_WRITE, /* inode has in-place-update data */ 775 FI_NEED_IPU, /* used for ipu per file */ 776 FI_ATOMIC_FILE, /* indicate atomic file */ 777 FI_DATA_EXIST, /* indicate data exists */ 778 FI_INLINE_DOTS, /* indicate inline dot dentries */ 779 FI_SKIP_WRITES, /* should skip data page writeback */ 780 FI_OPU_WRITE, /* used for opu per file */ 781 FI_DIRTY_FILE, /* indicate regular/symlink has dirty pages */ 782 FI_PREALLOCATED_ALL, /* all blocks for write were preallocated */ 783 FI_HOT_DATA, /* indicate file is hot */ 784 FI_EXTRA_ATTR, /* indicate file has extra attribute */ 785 FI_PROJ_INHERIT, /* indicate file inherits projectid */ 786 FI_PIN_FILE, /* indicate file should not be gced */ 787 FI_VERITY_IN_PROGRESS, /* building fs-verity Merkle tree */ 788 FI_COMPRESSED_FILE, /* indicate file's data can be compressed */ 789 FI_COMPRESS_CORRUPT, /* indicate compressed cluster is corrupted */ 790 FI_MMAP_FILE, /* indicate file was mmapped */ 791 FI_ENABLE_COMPRESS, /* enable compression in "user" compression mode */ 792 FI_COMPRESS_RELEASED, /* compressed blocks were released */ 793 FI_ALIGNED_WRITE, /* enable aligned write */ 794 FI_COW_FILE, /* indicate COW file */ 795 FI_ATOMIC_COMMITTED, /* indicate atomic commit completed except disk sync */ 796 FI_ATOMIC_REPLACE, /* indicate atomic replace */ 797 FI_MAX, /* max flag, never be used */ 798 }; 799 800 struct f2fs_inode_info { 801 struct inode vfs_inode; /* serve a vfs inode */ 802 unsigned long i_flags; /* keep an inode flags for ioctl */ 803 unsigned char i_advise; /* use to give file attribute hints */ 804 unsigned char i_dir_level; /* use for dentry level for large dir */ 805 unsigned int i_current_depth; /* only for directory depth */ 806 /* for gc failure statistic */ 807 unsigned int i_gc_failures[MAX_GC_FAILURE]; 808 unsigned int i_pino; /* parent inode number */ 809 umode_t i_acl_mode; /* keep file acl mode temporarily */ 810 811 /* Use below internally in f2fs*/ 812 unsigned long flags[BITS_TO_LONGS(FI_MAX)]; /* use to pass per-file flags */ 813 struct f2fs_rwsem i_sem; /* protect fi info */ 814 atomic_t dirty_pages; /* # of dirty pages */ 815 f2fs_hash_t chash; /* hash value of given file name */ 816 unsigned int clevel; /* maximum level of given file name */ 817 struct task_struct *task; /* lookup and create consistency */ 818 struct task_struct *cp_task; /* separate cp/wb IO stats*/ 819 struct task_struct *wb_task; /* indicate inode is in context of writeback */ 820 nid_t i_xattr_nid; /* node id that contains xattrs */ 821 loff_t last_disk_size; /* lastly written file size */ 822 spinlock_t i_size_lock; /* protect last_disk_size */ 823 824 #ifdef CONFIG_QUOTA 825 struct dquot __rcu *i_dquot[MAXQUOTAS]; 826 827 /* quota space reservation, managed internally by quota code */ 828 qsize_t i_reserved_quota; 829 #endif 830 struct list_head dirty_list; /* dirty list for dirs and files */ 831 struct list_head gdirty_list; /* linked in global dirty list */ 832 struct task_struct *atomic_write_task; /* store atomic write task */ 833 struct extent_tree *extent_tree[NR_EXTENT_CACHES]; 834 /* cached extent_tree entry */ 835 struct inode *cow_inode; /* copy-on-write inode for atomic write */ 836 837 /* avoid racing between foreground op and gc */ 838 struct f2fs_rwsem i_gc_rwsem[2]; 839 struct f2fs_rwsem i_xattr_sem; /* avoid racing between reading and changing EAs */ 840 841 int i_extra_isize; /* size of extra space located in i_addr */ 842 kprojid_t i_projid; /* id for project quota */ 843 int i_inline_xattr_size; /* inline xattr size */ 844 struct timespec64 i_crtime; /* inode creation time */ 845 struct timespec64 i_disk_time[3];/* inode disk times */ 846 847 /* for file compress */ 848 atomic_t i_compr_blocks; /* # of compressed blocks */ 849 unsigned char i_compress_algorithm; /* algorithm type */ 850 unsigned char i_log_cluster_size; /* log of cluster size */ 851 unsigned char i_compress_level; /* compress level (lz4hc,zstd) */ 852 unsigned char i_compress_flag; /* compress flag */ 853 unsigned int i_cluster_size; /* cluster size */ 854 855 unsigned int atomic_write_cnt; 856 loff_t original_i_size; /* original i_size before atomic write */ 857 }; 858 859 static inline void get_read_extent_info(struct extent_info *ext, 860 struct f2fs_extent *i_ext) 861 { 862 ext->fofs = le32_to_cpu(i_ext->fofs); 863 ext->blk = le32_to_cpu(i_ext->blk); 864 ext->len = le32_to_cpu(i_ext->len); 865 } 866 867 static inline void set_raw_read_extent(struct extent_info *ext, 868 struct f2fs_extent *i_ext) 869 { 870 i_ext->fofs = cpu_to_le32(ext->fofs); 871 i_ext->blk = cpu_to_le32(ext->blk); 872 i_ext->len = cpu_to_le32(ext->len); 873 } 874 875 static inline bool __is_discard_mergeable(struct discard_info *back, 876 struct discard_info *front, unsigned int max_len) 877 { 878 return (back->lstart + back->len == front->lstart) && 879 (back->len + front->len <= max_len); 880 } 881 882 static inline bool __is_discard_back_mergeable(struct discard_info *cur, 883 struct discard_info *back, unsigned int max_len) 884 { 885 return __is_discard_mergeable(back, cur, max_len); 886 } 887 888 static inline bool __is_discard_front_mergeable(struct discard_info *cur, 889 struct discard_info *front, unsigned int max_len) 890 { 891 return __is_discard_mergeable(cur, front, max_len); 892 } 893 894 /* 895 * For free nid management 896 */ 897 enum nid_state { 898 FREE_NID, /* newly added to free nid list */ 899 PREALLOC_NID, /* it is preallocated */ 900 MAX_NID_STATE, 901 }; 902 903 enum nat_state { 904 TOTAL_NAT, 905 DIRTY_NAT, 906 RECLAIMABLE_NAT, 907 MAX_NAT_STATE, 908 }; 909 910 struct f2fs_nm_info { 911 block_t nat_blkaddr; /* base disk address of NAT */ 912 nid_t max_nid; /* maximum possible node ids */ 913 nid_t available_nids; /* # of available node ids */ 914 nid_t next_scan_nid; /* the next nid to be scanned */ 915 nid_t max_rf_node_blocks; /* max # of nodes for recovery */ 916 unsigned int ram_thresh; /* control the memory footprint */ 917 unsigned int ra_nid_pages; /* # of nid pages to be readaheaded */ 918 unsigned int dirty_nats_ratio; /* control dirty nats ratio threshold */ 919 920 /* NAT cache management */ 921 struct radix_tree_root nat_root;/* root of the nat entry cache */ 922 struct radix_tree_root nat_set_root;/* root of the nat set cache */ 923 struct f2fs_rwsem nat_tree_lock; /* protect nat entry tree */ 924 struct list_head nat_entries; /* cached nat entry list (clean) */ 925 spinlock_t nat_list_lock; /* protect clean nat entry list */ 926 unsigned int nat_cnt[MAX_NAT_STATE]; /* the # of cached nat entries */ 927 unsigned int nat_blocks; /* # of nat blocks */ 928 929 /* free node ids management */ 930 struct radix_tree_root free_nid_root;/* root of the free_nid cache */ 931 struct list_head free_nid_list; /* list for free nids excluding preallocated nids */ 932 unsigned int nid_cnt[MAX_NID_STATE]; /* the number of free node id */ 933 spinlock_t nid_list_lock; /* protect nid lists ops */ 934 struct mutex build_lock; /* lock for build free nids */ 935 unsigned char **free_nid_bitmap; 936 unsigned char *nat_block_bitmap; 937 unsigned short *free_nid_count; /* free nid count of NAT block */ 938 939 /* for checkpoint */ 940 char *nat_bitmap; /* NAT bitmap pointer */ 941 942 unsigned int nat_bits_blocks; /* # of nat bits blocks */ 943 unsigned char *nat_bits; /* NAT bits blocks */ 944 unsigned char *full_nat_bits; /* full NAT pages */ 945 unsigned char *empty_nat_bits; /* empty NAT pages */ 946 #ifdef CONFIG_F2FS_CHECK_FS 947 char *nat_bitmap_mir; /* NAT bitmap mirror */ 948 #endif 949 int bitmap_size; /* bitmap size */ 950 }; 951 952 /* 953 * this structure is used as one of function parameters. 954 * all the information are dedicated to a given direct node block determined 955 * by the data offset in a file. 956 */ 957 struct dnode_of_data { 958 struct inode *inode; /* vfs inode pointer */ 959 struct page *inode_page; /* its inode page, NULL is possible */ 960 struct page *node_page; /* cached direct node page */ 961 nid_t nid; /* node id of the direct node block */ 962 unsigned int ofs_in_node; /* data offset in the node page */ 963 bool inode_page_locked; /* inode page is locked or not */ 964 bool node_changed; /* is node block changed */ 965 char cur_level; /* level of hole node page */ 966 char max_level; /* level of current page located */ 967 block_t data_blkaddr; /* block address of the node block */ 968 }; 969 970 static inline void set_new_dnode(struct dnode_of_data *dn, struct inode *inode, 971 struct page *ipage, struct page *npage, nid_t nid) 972 { 973 memset(dn, 0, sizeof(*dn)); 974 dn->inode = inode; 975 dn->inode_page = ipage; 976 dn->node_page = npage; 977 dn->nid = nid; 978 } 979 980 /* 981 * For SIT manager 982 * 983 * By default, there are 6 active log areas across the whole main area. 984 * When considering hot and cold data separation to reduce cleaning overhead, 985 * we split 3 for data logs and 3 for node logs as hot, warm, and cold types, 986 * respectively. 987 * In the current design, you should not change the numbers intentionally. 988 * Instead, as a mount option such as active_logs=x, you can use 2, 4, and 6 989 * logs individually according to the underlying devices. (default: 6) 990 * Just in case, on-disk layout covers maximum 16 logs that consist of 8 for 991 * data and 8 for node logs. 992 */ 993 #define NR_CURSEG_DATA_TYPE (3) 994 #define NR_CURSEG_NODE_TYPE (3) 995 #define NR_CURSEG_INMEM_TYPE (2) 996 #define NR_CURSEG_RO_TYPE (2) 997 #define NR_CURSEG_PERSIST_TYPE (NR_CURSEG_DATA_TYPE + NR_CURSEG_NODE_TYPE) 998 #define NR_CURSEG_TYPE (NR_CURSEG_INMEM_TYPE + NR_CURSEG_PERSIST_TYPE) 999 1000 enum { 1001 CURSEG_HOT_DATA = 0, /* directory entry blocks */ 1002 CURSEG_WARM_DATA, /* data blocks */ 1003 CURSEG_COLD_DATA, /* multimedia or GCed data blocks */ 1004 CURSEG_HOT_NODE, /* direct node blocks of directory files */ 1005 CURSEG_WARM_NODE, /* direct node blocks of normal files */ 1006 CURSEG_COLD_NODE, /* indirect node blocks */ 1007 NR_PERSISTENT_LOG, /* number of persistent log */ 1008 CURSEG_COLD_DATA_PINNED = NR_PERSISTENT_LOG, 1009 /* pinned file that needs consecutive block address */ 1010 CURSEG_ALL_DATA_ATGC, /* SSR alloctor in hot/warm/cold data area */ 1011 NO_CHECK_TYPE, /* number of persistent & inmem log */ 1012 }; 1013 1014 struct flush_cmd { 1015 struct completion wait; 1016 struct llist_node llnode; 1017 nid_t ino; 1018 int ret; 1019 }; 1020 1021 struct flush_cmd_control { 1022 struct task_struct *f2fs_issue_flush; /* flush thread */ 1023 wait_queue_head_t flush_wait_queue; /* waiting queue for wake-up */ 1024 atomic_t issued_flush; /* # of issued flushes */ 1025 atomic_t queued_flush; /* # of queued flushes */ 1026 struct llist_head issue_list; /* list for command issue */ 1027 struct llist_node *dispatch_list; /* list for command dispatch */ 1028 }; 1029 1030 struct f2fs_sm_info { 1031 struct sit_info *sit_info; /* whole segment information */ 1032 struct free_segmap_info *free_info; /* free segment information */ 1033 struct dirty_seglist_info *dirty_info; /* dirty segment information */ 1034 struct curseg_info *curseg_array; /* active segment information */ 1035 1036 struct f2fs_rwsem curseg_lock; /* for preventing curseg change */ 1037 1038 block_t seg0_blkaddr; /* block address of 0'th segment */ 1039 block_t main_blkaddr; /* start block address of main area */ 1040 block_t ssa_blkaddr; /* start block address of SSA area */ 1041 1042 unsigned int segment_count; /* total # of segments */ 1043 unsigned int main_segments; /* # of segments in main area */ 1044 unsigned int reserved_segments; /* # of reserved segments */ 1045 unsigned int additional_reserved_segments;/* reserved segs for IO align feature */ 1046 unsigned int ovp_segments; /* # of overprovision segments */ 1047 1048 /* a threshold to reclaim prefree segments */ 1049 unsigned int rec_prefree_segments; 1050 1051 struct list_head sit_entry_set; /* sit entry set list */ 1052 1053 unsigned int ipu_policy; /* in-place-update policy */ 1054 unsigned int min_ipu_util; /* in-place-update threshold */ 1055 unsigned int min_fsync_blocks; /* threshold for fsync */ 1056 unsigned int min_seq_blocks; /* threshold for sequential blocks */ 1057 unsigned int min_hot_blocks; /* threshold for hot block allocation */ 1058 unsigned int min_ssr_sections; /* threshold to trigger SSR allocation */ 1059 1060 /* for flush command control */ 1061 struct flush_cmd_control *fcc_info; 1062 1063 /* for discard command control */ 1064 struct discard_cmd_control *dcc_info; 1065 }; 1066 1067 /* 1068 * For superblock 1069 */ 1070 /* 1071 * COUNT_TYPE for monitoring 1072 * 1073 * f2fs monitors the number of several block types such as on-writeback, 1074 * dirty dentry blocks, dirty node blocks, and dirty meta blocks. 1075 */ 1076 #define WB_DATA_TYPE(p, f) \ 1077 (f || f2fs_is_cp_guaranteed(p) ? F2FS_WB_CP_DATA : F2FS_WB_DATA) 1078 enum count_type { 1079 F2FS_DIRTY_DENTS, 1080 F2FS_DIRTY_DATA, 1081 F2FS_DIRTY_QDATA, 1082 F2FS_DIRTY_NODES, 1083 F2FS_DIRTY_META, 1084 F2FS_DIRTY_IMETA, 1085 F2FS_WB_CP_DATA, 1086 F2FS_WB_DATA, 1087 F2FS_RD_DATA, 1088 F2FS_RD_NODE, 1089 F2FS_RD_META, 1090 F2FS_DIO_WRITE, 1091 F2FS_DIO_READ, 1092 NR_COUNT_TYPE, 1093 }; 1094 1095 /* 1096 * The below are the page types of bios used in submit_bio(). 1097 * The available types are: 1098 * DATA User data pages. It operates as async mode. 1099 * NODE Node pages. It operates as async mode. 1100 * META FS metadata pages such as SIT, NAT, CP. 1101 * NR_PAGE_TYPE The number of page types. 1102 * META_FLUSH Make sure the previous pages are written 1103 * with waiting the bio's completion 1104 * ... Only can be used with META. 1105 */ 1106 #define PAGE_TYPE_OF_BIO(type) ((type) > META ? META : (type)) 1107 enum page_type { 1108 DATA = 0, 1109 NODE = 1, /* should not change this */ 1110 META, 1111 NR_PAGE_TYPE, 1112 META_FLUSH, 1113 IPU, /* the below types are used by tracepoints only. */ 1114 OPU, 1115 }; 1116 1117 enum temp_type { 1118 HOT = 0, /* must be zero for meta bio */ 1119 WARM, 1120 COLD, 1121 NR_TEMP_TYPE, 1122 }; 1123 1124 enum need_lock_type { 1125 LOCK_REQ = 0, 1126 LOCK_DONE, 1127 LOCK_RETRY, 1128 }; 1129 1130 enum cp_reason_type { 1131 CP_NO_NEEDED, 1132 CP_NON_REGULAR, 1133 CP_COMPRESSED, 1134 CP_HARDLINK, 1135 CP_SB_NEED_CP, 1136 CP_WRONG_PINO, 1137 CP_NO_SPC_ROLL, 1138 CP_NODE_NEED_CP, 1139 CP_FASTBOOT_MODE, 1140 CP_SPEC_LOG_NUM, 1141 CP_RECOVER_DIR, 1142 }; 1143 1144 enum iostat_type { 1145 /* WRITE IO */ 1146 APP_DIRECT_IO, /* app direct write IOs */ 1147 APP_BUFFERED_IO, /* app buffered write IOs */ 1148 APP_WRITE_IO, /* app write IOs */ 1149 APP_MAPPED_IO, /* app mapped IOs */ 1150 APP_BUFFERED_CDATA_IO, /* app buffered write IOs on compressed file */ 1151 APP_MAPPED_CDATA_IO, /* app mapped write IOs on compressed file */ 1152 FS_DATA_IO, /* data IOs from kworker/fsync/reclaimer */ 1153 FS_CDATA_IO, /* data IOs from kworker/fsync/reclaimer on compressed file */ 1154 FS_NODE_IO, /* node IOs from kworker/fsync/reclaimer */ 1155 FS_META_IO, /* meta IOs from kworker/reclaimer */ 1156 FS_GC_DATA_IO, /* data IOs from forground gc */ 1157 FS_GC_NODE_IO, /* node IOs from forground gc */ 1158 FS_CP_DATA_IO, /* data IOs from checkpoint */ 1159 FS_CP_NODE_IO, /* node IOs from checkpoint */ 1160 FS_CP_META_IO, /* meta IOs from checkpoint */ 1161 1162 /* READ IO */ 1163 APP_DIRECT_READ_IO, /* app direct read IOs */ 1164 APP_BUFFERED_READ_IO, /* app buffered read IOs */ 1165 APP_READ_IO, /* app read IOs */ 1166 APP_MAPPED_READ_IO, /* app mapped read IOs */ 1167 APP_BUFFERED_CDATA_READ_IO, /* app buffered read IOs on compressed file */ 1168 APP_MAPPED_CDATA_READ_IO, /* app mapped read IOs on compressed file */ 1169 FS_DATA_READ_IO, /* data read IOs */ 1170 FS_GDATA_READ_IO, /* data read IOs from background gc */ 1171 FS_CDATA_READ_IO, /* compressed data read IOs */ 1172 FS_NODE_READ_IO, /* node read IOs */ 1173 FS_META_READ_IO, /* meta read IOs */ 1174 1175 /* other */ 1176 FS_DISCARD_IO, /* discard */ 1177 FS_FLUSH_IO, /* flush */ 1178 FS_ZONE_RESET_IO, /* zone reset */ 1179 NR_IO_TYPE, 1180 }; 1181 1182 struct f2fs_io_info { 1183 struct f2fs_sb_info *sbi; /* f2fs_sb_info pointer */ 1184 nid_t ino; /* inode number */ 1185 enum page_type type; /* contains DATA/NODE/META/META_FLUSH */ 1186 enum temp_type temp; /* contains HOT/WARM/COLD */ 1187 enum req_op op; /* contains REQ_OP_ */ 1188 blk_opf_t op_flags; /* req_flag_bits */ 1189 block_t new_blkaddr; /* new block address to be written */ 1190 block_t old_blkaddr; /* old block address before Cow */ 1191 struct page *page; /* page to be written */ 1192 struct page *encrypted_page; /* encrypted page */ 1193 struct page *compressed_page; /* compressed page */ 1194 struct list_head list; /* serialize IOs */ 1195 unsigned int compr_blocks; /* # of compressed block addresses */ 1196 unsigned int need_lock:8; /* indicate we need to lock cp_rwsem */ 1197 unsigned int version:8; /* version of the node */ 1198 unsigned int submitted:1; /* indicate IO submission */ 1199 unsigned int in_list:1; /* indicate fio is in io_list */ 1200 unsigned int is_por:1; /* indicate IO is from recovery or not */ 1201 unsigned int retry:1; /* need to reallocate block address */ 1202 unsigned int encrypted:1; /* indicate file is encrypted */ 1203 unsigned int post_read:1; /* require post read */ 1204 enum iostat_type io_type; /* io type */ 1205 struct writeback_control *io_wbc; /* writeback control */ 1206 struct bio **bio; /* bio for ipu */ 1207 sector_t *last_block; /* last block number in bio */ 1208 }; 1209 1210 struct bio_entry { 1211 struct bio *bio; 1212 struct list_head list; 1213 }; 1214 1215 #define is_read_io(rw) ((rw) == READ) 1216 struct f2fs_bio_info { 1217 struct f2fs_sb_info *sbi; /* f2fs superblock */ 1218 struct bio *bio; /* bios to merge */ 1219 sector_t last_block_in_bio; /* last block number */ 1220 struct f2fs_io_info fio; /* store buffered io info. */ 1221 #ifdef CONFIG_BLK_DEV_ZONED 1222 struct completion zone_wait; /* condition value for the previous open zone to close */ 1223 struct bio *zone_pending_bio; /* pending bio for the previous zone */ 1224 void *bi_private; /* previous bi_private for pending bio */ 1225 #endif 1226 struct f2fs_rwsem io_rwsem; /* blocking op for bio */ 1227 spinlock_t io_lock; /* serialize DATA/NODE IOs */ 1228 struct list_head io_list; /* track fios */ 1229 struct list_head bio_list; /* bio entry list head */ 1230 struct f2fs_rwsem bio_list_lock; /* lock to protect bio entry list */ 1231 }; 1232 1233 #define FDEV(i) (sbi->devs[i]) 1234 #define RDEV(i) (raw_super->devs[i]) 1235 struct f2fs_dev_info { 1236 struct block_device *bdev; 1237 char path[MAX_PATH_LEN]; 1238 unsigned int total_segments; 1239 block_t start_blk; 1240 block_t end_blk; 1241 #ifdef CONFIG_BLK_DEV_ZONED 1242 unsigned int nr_blkz; /* Total number of zones */ 1243 unsigned long *blkz_seq; /* Bitmap indicating sequential zones */ 1244 #endif 1245 }; 1246 1247 enum inode_type { 1248 DIR_INODE, /* for dirty dir inode */ 1249 FILE_INODE, /* for dirty regular/symlink inode */ 1250 DIRTY_META, /* for all dirtied inode metadata */ 1251 NR_INODE_TYPE, 1252 }; 1253 1254 /* for inner inode cache management */ 1255 struct inode_management { 1256 struct radix_tree_root ino_root; /* ino entry array */ 1257 spinlock_t ino_lock; /* for ino entry lock */ 1258 struct list_head ino_list; /* inode list head */ 1259 unsigned long ino_num; /* number of entries */ 1260 }; 1261 1262 /* for GC_AT */ 1263 struct atgc_management { 1264 bool atgc_enabled; /* ATGC is enabled or not */ 1265 struct rb_root_cached root; /* root of victim rb-tree */ 1266 struct list_head victim_list; /* linked with all victim entries */ 1267 unsigned int victim_count; /* victim count in rb-tree */ 1268 unsigned int candidate_ratio; /* candidate ratio */ 1269 unsigned int max_candidate_count; /* max candidate count */ 1270 unsigned int age_weight; /* age weight, vblock_weight = 100 - age_weight */ 1271 unsigned long long age_threshold; /* age threshold */ 1272 }; 1273 1274 struct f2fs_gc_control { 1275 unsigned int victim_segno; /* target victim segment number */ 1276 int init_gc_type; /* FG_GC or BG_GC */ 1277 bool no_bg_gc; /* check the space and stop bg_gc */ 1278 bool should_migrate_blocks; /* should migrate blocks */ 1279 bool err_gc_skipped; /* return EAGAIN if GC skipped */ 1280 unsigned int nr_free_secs; /* # of free sections to do GC */ 1281 }; 1282 1283 /* 1284 * For s_flag in struct f2fs_sb_info 1285 * Modification on enum should be synchronized with s_flag array 1286 */ 1287 enum { 1288 SBI_IS_DIRTY, /* dirty flag for checkpoint */ 1289 SBI_IS_CLOSE, /* specify unmounting */ 1290 SBI_NEED_FSCK, /* need fsck.f2fs to fix */ 1291 SBI_POR_DOING, /* recovery is doing or not */ 1292 SBI_NEED_SB_WRITE, /* need to recover superblock */ 1293 SBI_NEED_CP, /* need to checkpoint */ 1294 SBI_IS_SHUTDOWN, /* shutdown by ioctl */ 1295 SBI_IS_RECOVERED, /* recovered orphan/data */ 1296 SBI_CP_DISABLED, /* CP was disabled last mount */ 1297 SBI_CP_DISABLED_QUICK, /* CP was disabled quickly */ 1298 SBI_QUOTA_NEED_FLUSH, /* need to flush quota info in CP */ 1299 SBI_QUOTA_SKIP_FLUSH, /* skip flushing quota in current CP */ 1300 SBI_QUOTA_NEED_REPAIR, /* quota file may be corrupted */ 1301 SBI_IS_RESIZEFS, /* resizefs is in process */ 1302 SBI_IS_FREEZING, /* freezefs is in process */ 1303 SBI_IS_WRITABLE, /* remove ro mountoption transiently */ 1304 MAX_SBI_FLAG, 1305 }; 1306 1307 enum { 1308 CP_TIME, 1309 REQ_TIME, 1310 DISCARD_TIME, 1311 GC_TIME, 1312 DISABLE_TIME, 1313 UMOUNT_DISCARD_TIMEOUT, 1314 MAX_TIME, 1315 }; 1316 1317 /* Note that you need to keep synchronization with this gc_mode_names array */ 1318 enum { 1319 GC_NORMAL, 1320 GC_IDLE_CB, 1321 GC_IDLE_GREEDY, 1322 GC_IDLE_AT, 1323 GC_URGENT_HIGH, 1324 GC_URGENT_LOW, 1325 GC_URGENT_MID, 1326 MAX_GC_MODE, 1327 }; 1328 1329 enum { 1330 BGGC_MODE_ON, /* background gc is on */ 1331 BGGC_MODE_OFF, /* background gc is off */ 1332 BGGC_MODE_SYNC, /* 1333 * background gc is on, migrating blocks 1334 * like foreground gc 1335 */ 1336 }; 1337 1338 enum { 1339 FS_MODE_ADAPTIVE, /* use both lfs/ssr allocation */ 1340 FS_MODE_LFS, /* use lfs allocation only */ 1341 FS_MODE_FRAGMENT_SEG, /* segment fragmentation mode */ 1342 FS_MODE_FRAGMENT_BLK, /* block fragmentation mode */ 1343 }; 1344 1345 enum { 1346 ALLOC_MODE_DEFAULT, /* stay default */ 1347 ALLOC_MODE_REUSE, /* reuse segments as much as possible */ 1348 }; 1349 1350 enum fsync_mode { 1351 FSYNC_MODE_POSIX, /* fsync follows posix semantics */ 1352 FSYNC_MODE_STRICT, /* fsync behaves in line with ext4 */ 1353 FSYNC_MODE_NOBARRIER, /* fsync behaves nobarrier based on posix */ 1354 }; 1355 1356 enum { 1357 COMPR_MODE_FS, /* 1358 * automatically compress compression 1359 * enabled files 1360 */ 1361 COMPR_MODE_USER, /* 1362 * automatical compression is disabled. 1363 * user can control the file compression 1364 * using ioctls 1365 */ 1366 }; 1367 1368 enum { 1369 DISCARD_UNIT_BLOCK, /* basic discard unit is block */ 1370 DISCARD_UNIT_SEGMENT, /* basic discard unit is segment */ 1371 DISCARD_UNIT_SECTION, /* basic discard unit is section */ 1372 }; 1373 1374 enum { 1375 MEMORY_MODE_NORMAL, /* memory mode for normal devices */ 1376 MEMORY_MODE_LOW, /* memory mode for low memry devices */ 1377 }; 1378 1379 enum errors_option { 1380 MOUNT_ERRORS_READONLY, /* remount fs ro on errors */ 1381 MOUNT_ERRORS_CONTINUE, /* continue on errors */ 1382 MOUNT_ERRORS_PANIC, /* panic on errors */ 1383 }; 1384 1385 enum { 1386 BACKGROUND, 1387 FOREGROUND, 1388 MAX_CALL_TYPE, 1389 TOTAL_CALL = FOREGROUND, 1390 }; 1391 1392 static inline int f2fs_test_bit(unsigned int nr, char *addr); 1393 static inline void f2fs_set_bit(unsigned int nr, char *addr); 1394 static inline void f2fs_clear_bit(unsigned int nr, char *addr); 1395 1396 /* 1397 * Layout of f2fs page.private: 1398 * 1399 * Layout A: lowest bit should be 1 1400 * | bit0 = 1 | bit1 | bit2 | ... | bit MAX | private data .... | 1401 * bit 0 PAGE_PRIVATE_NOT_POINTER 1402 * bit 1 PAGE_PRIVATE_DUMMY_WRITE 1403 * bit 2 PAGE_PRIVATE_ONGOING_MIGRATION 1404 * bit 3 PAGE_PRIVATE_INLINE_INODE 1405 * bit 4 PAGE_PRIVATE_REF_RESOURCE 1406 * bit 5- f2fs private data 1407 * 1408 * Layout B: lowest bit should be 0 1409 * page.private is a wrapped pointer. 1410 */ 1411 enum { 1412 PAGE_PRIVATE_NOT_POINTER, /* private contains non-pointer data */ 1413 PAGE_PRIVATE_DUMMY_WRITE, /* data page for padding aligned IO */ 1414 PAGE_PRIVATE_ONGOING_MIGRATION, /* data page which is on-going migrating */ 1415 PAGE_PRIVATE_INLINE_INODE, /* inode page contains inline data */ 1416 PAGE_PRIVATE_REF_RESOURCE, /* dirty page has referenced resources */ 1417 PAGE_PRIVATE_MAX 1418 }; 1419 1420 /* For compression */ 1421 enum compress_algorithm_type { 1422 COMPRESS_LZO, 1423 COMPRESS_LZ4, 1424 COMPRESS_ZSTD, 1425 COMPRESS_LZORLE, 1426 COMPRESS_MAX, 1427 }; 1428 1429 enum compress_flag { 1430 COMPRESS_CHKSUM, 1431 COMPRESS_MAX_FLAG, 1432 }; 1433 1434 #define COMPRESS_WATERMARK 20 1435 #define COMPRESS_PERCENT 20 1436 1437 #define COMPRESS_DATA_RESERVED_SIZE 4 1438 struct compress_data { 1439 __le32 clen; /* compressed data size */ 1440 __le32 chksum; /* compressed data chksum */ 1441 __le32 reserved[COMPRESS_DATA_RESERVED_SIZE]; /* reserved */ 1442 u8 cdata[]; /* compressed data */ 1443 }; 1444 1445 #define COMPRESS_HEADER_SIZE (sizeof(struct compress_data)) 1446 1447 #define F2FS_COMPRESSED_PAGE_MAGIC 0xF5F2C000 1448 1449 #define F2FS_ZSTD_DEFAULT_CLEVEL 1 1450 1451 #define COMPRESS_LEVEL_OFFSET 8 1452 1453 /* compress context */ 1454 struct compress_ctx { 1455 struct inode *inode; /* inode the context belong to */ 1456 pgoff_t cluster_idx; /* cluster index number */ 1457 unsigned int cluster_size; /* page count in cluster */ 1458 unsigned int log_cluster_size; /* log of cluster size */ 1459 struct page **rpages; /* pages store raw data in cluster */ 1460 unsigned int nr_rpages; /* total page number in rpages */ 1461 struct page **cpages; /* pages store compressed data in cluster */ 1462 unsigned int nr_cpages; /* total page number in cpages */ 1463 unsigned int valid_nr_cpages; /* valid page number in cpages */ 1464 void *rbuf; /* virtual mapped address on rpages */ 1465 struct compress_data *cbuf; /* virtual mapped address on cpages */ 1466 size_t rlen; /* valid data length in rbuf */ 1467 size_t clen; /* valid data length in cbuf */ 1468 void *private; /* payload buffer for specified compression algorithm */ 1469 void *private2; /* extra payload buffer */ 1470 }; 1471 1472 /* compress context for write IO path */ 1473 struct compress_io_ctx { 1474 u32 magic; /* magic number to indicate page is compressed */ 1475 struct inode *inode; /* inode the context belong to */ 1476 struct page **rpages; /* pages store raw data in cluster */ 1477 unsigned int nr_rpages; /* total page number in rpages */ 1478 atomic_t pending_pages; /* in-flight compressed page count */ 1479 }; 1480 1481 /* Context for decompressing one cluster on the read IO path */ 1482 struct decompress_io_ctx { 1483 u32 magic; /* magic number to indicate page is compressed */ 1484 struct inode *inode; /* inode the context belong to */ 1485 pgoff_t cluster_idx; /* cluster index number */ 1486 unsigned int cluster_size; /* page count in cluster */ 1487 unsigned int log_cluster_size; /* log of cluster size */ 1488 struct page **rpages; /* pages store raw data in cluster */ 1489 unsigned int nr_rpages; /* total page number in rpages */ 1490 struct page **cpages; /* pages store compressed data in cluster */ 1491 unsigned int nr_cpages; /* total page number in cpages */ 1492 struct page **tpages; /* temp pages to pad holes in cluster */ 1493 void *rbuf; /* virtual mapped address on rpages */ 1494 struct compress_data *cbuf; /* virtual mapped address on cpages */ 1495 size_t rlen; /* valid data length in rbuf */ 1496 size_t clen; /* valid data length in cbuf */ 1497 1498 /* 1499 * The number of compressed pages remaining to be read in this cluster. 1500 * This is initially nr_cpages. It is decremented by 1 each time a page 1501 * has been read (or failed to be read). When it reaches 0, the cluster 1502 * is decompressed (or an error is reported). 1503 * 1504 * If an error occurs before all the pages have been submitted for I/O, 1505 * then this will never reach 0. In this case the I/O submitter is 1506 * responsible for calling f2fs_decompress_end_io() instead. 1507 */ 1508 atomic_t remaining_pages; 1509 1510 /* 1511 * Number of references to this decompress_io_ctx. 1512 * 1513 * One reference is held for I/O completion. This reference is dropped 1514 * after the pagecache pages are updated and unlocked -- either after 1515 * decompression (and verity if enabled), or after an error. 1516 * 1517 * In addition, each compressed page holds a reference while it is in a 1518 * bio. These references are necessary prevent compressed pages from 1519 * being freed while they are still in a bio. 1520 */ 1521 refcount_t refcnt; 1522 1523 bool failed; /* IO error occurred before decompression? */ 1524 bool need_verity; /* need fs-verity verification after decompression? */ 1525 void *private; /* payload buffer for specified decompression algorithm */ 1526 void *private2; /* extra payload buffer */ 1527 struct work_struct verity_work; /* work to verify the decompressed pages */ 1528 struct work_struct free_work; /* work for late free this structure itself */ 1529 }; 1530 1531 #define NULL_CLUSTER ((unsigned int)(~0)) 1532 #define MIN_COMPRESS_LOG_SIZE 2 1533 #define MAX_COMPRESS_LOG_SIZE 8 1534 #define MAX_COMPRESS_WINDOW_SIZE(log_size) ((PAGE_SIZE) << (log_size)) 1535 1536 struct f2fs_sb_info { 1537 struct super_block *sb; /* pointer to VFS super block */ 1538 struct proc_dir_entry *s_proc; /* proc entry */ 1539 struct f2fs_super_block *raw_super; /* raw super block pointer */ 1540 struct f2fs_rwsem sb_lock; /* lock for raw super block */ 1541 int valid_super_block; /* valid super block no */ 1542 unsigned long s_flag; /* flags for sbi */ 1543 struct mutex writepages; /* mutex for writepages() */ 1544 1545 #ifdef CONFIG_BLK_DEV_ZONED 1546 unsigned int blocks_per_blkz; /* F2FS blocks per zone */ 1547 #endif 1548 1549 /* for node-related operations */ 1550 struct f2fs_nm_info *nm_info; /* node manager */ 1551 struct inode *node_inode; /* cache node blocks */ 1552 1553 /* for segment-related operations */ 1554 struct f2fs_sm_info *sm_info; /* segment manager */ 1555 1556 /* for bio operations */ 1557 struct f2fs_bio_info *write_io[NR_PAGE_TYPE]; /* for write bios */ 1558 /* keep migration IO order for LFS mode */ 1559 struct f2fs_rwsem io_order_lock; 1560 mempool_t *write_io_dummy; /* Dummy pages */ 1561 pgoff_t page_eio_ofs[NR_PAGE_TYPE]; /* EIO page offset */ 1562 int page_eio_cnt[NR_PAGE_TYPE]; /* EIO count */ 1563 1564 /* for checkpoint */ 1565 struct f2fs_checkpoint *ckpt; /* raw checkpoint pointer */ 1566 int cur_cp_pack; /* remain current cp pack */ 1567 spinlock_t cp_lock; /* for flag in ckpt */ 1568 struct inode *meta_inode; /* cache meta blocks */ 1569 struct f2fs_rwsem cp_global_sem; /* checkpoint procedure lock */ 1570 struct f2fs_rwsem cp_rwsem; /* blocking FS operations */ 1571 struct f2fs_rwsem node_write; /* locking node writes */ 1572 struct f2fs_rwsem node_change; /* locking node change */ 1573 wait_queue_head_t cp_wait; 1574 unsigned long last_time[MAX_TIME]; /* to store time in jiffies */ 1575 long interval_time[MAX_TIME]; /* to store thresholds */ 1576 struct ckpt_req_control cprc_info; /* for checkpoint request control */ 1577 1578 struct inode_management im[MAX_INO_ENTRY]; /* manage inode cache */ 1579 1580 spinlock_t fsync_node_lock; /* for node entry lock */ 1581 struct list_head fsync_node_list; /* node list head */ 1582 unsigned int fsync_seg_id; /* sequence id */ 1583 unsigned int fsync_node_num; /* number of node entries */ 1584 1585 /* for orphan inode, use 0'th array */ 1586 unsigned int max_orphans; /* max orphan inodes */ 1587 1588 /* for inode management */ 1589 struct list_head inode_list[NR_INODE_TYPE]; /* dirty inode list */ 1590 spinlock_t inode_lock[NR_INODE_TYPE]; /* for dirty inode list lock */ 1591 struct mutex flush_lock; /* for flush exclusion */ 1592 1593 /* for extent tree cache */ 1594 struct extent_tree_info extent_tree[NR_EXTENT_CACHES]; 1595 atomic64_t allocated_data_blocks; /* for block age extent_cache */ 1596 1597 /* The threshold used for hot and warm data seperation*/ 1598 unsigned int hot_data_age_threshold; 1599 unsigned int warm_data_age_threshold; 1600 unsigned int last_age_weight; 1601 1602 /* basic filesystem units */ 1603 unsigned int log_sectors_per_block; /* log2 sectors per block */ 1604 unsigned int log_blocksize; /* log2 block size */ 1605 unsigned int blocksize; /* block size */ 1606 unsigned int root_ino_num; /* root inode number*/ 1607 unsigned int node_ino_num; /* node inode number*/ 1608 unsigned int meta_ino_num; /* meta inode number*/ 1609 unsigned int log_blocks_per_seg; /* log2 blocks per segment */ 1610 unsigned int blocks_per_seg; /* blocks per segment */ 1611 unsigned int unusable_blocks_per_sec; /* unusable blocks per section */ 1612 unsigned int segs_per_sec; /* segments per section */ 1613 unsigned int secs_per_zone; /* sections per zone */ 1614 unsigned int total_sections; /* total section count */ 1615 unsigned int total_node_count; /* total node block count */ 1616 unsigned int total_valid_node_count; /* valid node block count */ 1617 int dir_level; /* directory level */ 1618 bool readdir_ra; /* readahead inode in readdir */ 1619 u64 max_io_bytes; /* max io bytes to merge IOs */ 1620 1621 block_t user_block_count; /* # of user blocks */ 1622 block_t total_valid_block_count; /* # of valid blocks */ 1623 block_t discard_blks; /* discard command candidats */ 1624 block_t last_valid_block_count; /* for recovery */ 1625 block_t reserved_blocks; /* configurable reserved blocks */ 1626 block_t current_reserved_blocks; /* current reserved blocks */ 1627 1628 /* Additional tracking for no checkpoint mode */ 1629 block_t unusable_block_count; /* # of blocks saved by last cp */ 1630 1631 unsigned int nquota_files; /* # of quota sysfile */ 1632 struct f2fs_rwsem quota_sem; /* blocking cp for flags */ 1633 1634 /* # of pages, see count_type */ 1635 atomic_t nr_pages[NR_COUNT_TYPE]; 1636 /* # of allocated blocks */ 1637 struct percpu_counter alloc_valid_block_count; 1638 /* # of node block writes as roll forward recovery */ 1639 struct percpu_counter rf_node_block_count; 1640 1641 /* writeback control */ 1642 atomic_t wb_sync_req[META]; /* count # of WB_SYNC threads */ 1643 1644 /* valid inode count */ 1645 struct percpu_counter total_valid_inode_count; 1646 1647 struct f2fs_mount_info mount_opt; /* mount options */ 1648 1649 /* for cleaning operations */ 1650 struct f2fs_rwsem gc_lock; /* 1651 * semaphore for GC, avoid 1652 * race between GC and GC or CP 1653 */ 1654 struct f2fs_gc_kthread *gc_thread; /* GC thread */ 1655 struct atgc_management am; /* atgc management */ 1656 unsigned int cur_victim_sec; /* current victim section num */ 1657 unsigned int gc_mode; /* current GC state */ 1658 unsigned int next_victim_seg[2]; /* next segment in victim section */ 1659 spinlock_t gc_remaining_trials_lock; 1660 /* remaining trial count for GC_URGENT_* and GC_IDLE_* */ 1661 unsigned int gc_remaining_trials; 1662 1663 /* for skip statistic */ 1664 unsigned long long skipped_gc_rwsem; /* FG_GC only */ 1665 1666 /* threshold for gc trials on pinned files */ 1667 u64 gc_pin_file_threshold; 1668 struct f2fs_rwsem pin_sem; 1669 1670 /* maximum # of trials to find a victim segment for SSR and GC */ 1671 unsigned int max_victim_search; 1672 /* migration granularity of garbage collection, unit: segment */ 1673 unsigned int migration_granularity; 1674 1675 /* 1676 * for stat information. 1677 * one is for the LFS mode, and the other is for the SSR mode. 1678 */ 1679 #ifdef CONFIG_F2FS_STAT_FS 1680 struct f2fs_stat_info *stat_info; /* FS status information */ 1681 atomic_t meta_count[META_MAX]; /* # of meta blocks */ 1682 unsigned int segment_count[2]; /* # of allocated segments */ 1683 unsigned int block_count[2]; /* # of allocated blocks */ 1684 atomic_t inplace_count; /* # of inplace update */ 1685 /* # of lookup extent cache */ 1686 atomic64_t total_hit_ext[NR_EXTENT_CACHES]; 1687 /* # of hit rbtree extent node */ 1688 atomic64_t read_hit_rbtree[NR_EXTENT_CACHES]; 1689 /* # of hit cached extent node */ 1690 atomic64_t read_hit_cached[NR_EXTENT_CACHES]; 1691 /* # of hit largest extent node in read extent cache */ 1692 atomic64_t read_hit_largest; 1693 atomic_t inline_xattr; /* # of inline_xattr inodes */ 1694 atomic_t inline_inode; /* # of inline_data inodes */ 1695 atomic_t inline_dir; /* # of inline_dentry inodes */ 1696 atomic_t compr_inode; /* # of compressed inodes */ 1697 atomic64_t compr_blocks; /* # of compressed blocks */ 1698 atomic_t swapfile_inode; /* # of swapfile inodes */ 1699 atomic_t atomic_files; /* # of opened atomic file */ 1700 atomic_t max_aw_cnt; /* max # of atomic writes */ 1701 unsigned int io_skip_bggc; /* skip background gc for in-flight IO */ 1702 unsigned int other_skip_bggc; /* skip background gc for other reasons */ 1703 unsigned int ndirty_inode[NR_INODE_TYPE]; /* # of dirty inodes */ 1704 atomic_t cp_call_count[MAX_CALL_TYPE]; /* # of cp call */ 1705 #endif 1706 spinlock_t stat_lock; /* lock for stat operations */ 1707 1708 /* to attach REQ_META|REQ_FUA flags */ 1709 unsigned int data_io_flag; 1710 unsigned int node_io_flag; 1711 1712 /* For sysfs support */ 1713 struct kobject s_kobj; /* /sys/fs/f2fs/<devname> */ 1714 struct completion s_kobj_unregister; 1715 1716 struct kobject s_stat_kobj; /* /sys/fs/f2fs/<devname>/stat */ 1717 struct completion s_stat_kobj_unregister; 1718 1719 struct kobject s_feature_list_kobj; /* /sys/fs/f2fs/<devname>/feature_list */ 1720 struct completion s_feature_list_kobj_unregister; 1721 1722 /* For shrinker support */ 1723 struct list_head s_list; 1724 struct mutex umount_mutex; 1725 unsigned int shrinker_run_no; 1726 1727 /* For multi devices */ 1728 int s_ndevs; /* number of devices */ 1729 struct f2fs_dev_info *devs; /* for device list */ 1730 unsigned int dirty_device; /* for checkpoint data flush */ 1731 spinlock_t dev_lock; /* protect dirty_device */ 1732 bool aligned_blksize; /* all devices has the same logical blksize */ 1733 1734 /* For write statistics */ 1735 u64 sectors_written_start; 1736 u64 kbytes_written; 1737 1738 /* Reference to checksum algorithm driver via cryptoapi */ 1739 struct crypto_shash *s_chksum_driver; 1740 1741 /* Precomputed FS UUID checksum for seeding other checksums */ 1742 __u32 s_chksum_seed; 1743 1744 struct workqueue_struct *post_read_wq; /* post read workqueue */ 1745 1746 /* 1747 * If we are in irq context, let's update error information into 1748 * on-disk superblock in the work. 1749 */ 1750 struct work_struct s_error_work; 1751 unsigned char errors[MAX_F2FS_ERRORS]; /* error flags */ 1752 unsigned char stop_reason[MAX_STOP_REASON]; /* stop reason */ 1753 spinlock_t error_lock; /* protect errors/stop_reason array */ 1754 bool error_dirty; /* errors of sb is dirty */ 1755 1756 struct kmem_cache *inline_xattr_slab; /* inline xattr entry */ 1757 unsigned int inline_xattr_slab_size; /* default inline xattr slab size */ 1758 1759 /* For reclaimed segs statistics per each GC mode */ 1760 unsigned int gc_segment_mode; /* GC state for reclaimed segments */ 1761 unsigned int gc_reclaimed_segs[MAX_GC_MODE]; /* Reclaimed segs for each mode */ 1762 1763 unsigned long seq_file_ra_mul; /* multiplier for ra_pages of seq. files in fadvise */ 1764 1765 int max_fragment_chunk; /* max chunk size for block fragmentation mode */ 1766 int max_fragment_hole; /* max hole size for block fragmentation mode */ 1767 1768 /* For atomic write statistics */ 1769 atomic64_t current_atomic_write; 1770 s64 peak_atomic_write; 1771 u64 committed_atomic_block; 1772 u64 revoked_atomic_block; 1773 1774 #ifdef CONFIG_F2FS_FS_COMPRESSION 1775 struct kmem_cache *page_array_slab; /* page array entry */ 1776 unsigned int page_array_slab_size; /* default page array slab size */ 1777 1778 /* For runtime compression statistics */ 1779 u64 compr_written_block; 1780 u64 compr_saved_block; 1781 u32 compr_new_inode; 1782 1783 /* For compressed block cache */ 1784 struct inode *compress_inode; /* cache compressed blocks */ 1785 unsigned int compress_percent; /* cache page percentage */ 1786 unsigned int compress_watermark; /* cache page watermark */ 1787 atomic_t compress_page_hit; /* cache hit count */ 1788 #endif 1789 1790 #ifdef CONFIG_F2FS_IOSTAT 1791 /* For app/fs IO statistics */ 1792 spinlock_t iostat_lock; 1793 unsigned long long iostat_count[NR_IO_TYPE]; 1794 unsigned long long iostat_bytes[NR_IO_TYPE]; 1795 unsigned long long prev_iostat_bytes[NR_IO_TYPE]; 1796 bool iostat_enable; 1797 unsigned long iostat_next_period; 1798 unsigned int iostat_period_ms; 1799 1800 /* For io latency related statistics info in one iostat period */ 1801 spinlock_t iostat_lat_lock; 1802 struct iostat_lat_info *iostat_io_lat; 1803 #endif 1804 }; 1805 1806 #ifdef CONFIG_F2FS_FAULT_INJECTION 1807 #define time_to_inject(sbi, type) __time_to_inject(sbi, type, __func__, \ 1808 __builtin_return_address(0)) 1809 static inline bool __time_to_inject(struct f2fs_sb_info *sbi, int type, 1810 const char *func, const char *parent_func) 1811 { 1812 struct f2fs_fault_info *ffi = &F2FS_OPTION(sbi).fault_info; 1813 1814 if (!ffi->inject_rate) 1815 return false; 1816 1817 if (!IS_FAULT_SET(ffi, type)) 1818 return false; 1819 1820 atomic_inc(&ffi->inject_ops); 1821 if (atomic_read(&ffi->inject_ops) >= ffi->inject_rate) { 1822 atomic_set(&ffi->inject_ops, 0); 1823 printk_ratelimited("%sF2FS-fs (%s) : inject %s in %s of %pS\n", 1824 KERN_INFO, sbi->sb->s_id, f2fs_fault_name[type], 1825 func, parent_func); 1826 return true; 1827 } 1828 return false; 1829 } 1830 #else 1831 static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type) 1832 { 1833 return false; 1834 } 1835 #endif 1836 1837 /* 1838 * Test if the mounted volume is a multi-device volume. 1839 * - For a single regular disk volume, sbi->s_ndevs is 0. 1840 * - For a single zoned disk volume, sbi->s_ndevs is 1. 1841 * - For a multi-device volume, sbi->s_ndevs is always 2 or more. 1842 */ 1843 static inline bool f2fs_is_multi_device(struct f2fs_sb_info *sbi) 1844 { 1845 return sbi->s_ndevs > 1; 1846 } 1847 1848 static inline void f2fs_update_time(struct f2fs_sb_info *sbi, int type) 1849 { 1850 unsigned long now = jiffies; 1851 1852 sbi->last_time[type] = now; 1853 1854 /* DISCARD_TIME and GC_TIME are based on REQ_TIME */ 1855 if (type == REQ_TIME) { 1856 sbi->last_time[DISCARD_TIME] = now; 1857 sbi->last_time[GC_TIME] = now; 1858 } 1859 } 1860 1861 static inline bool f2fs_time_over(struct f2fs_sb_info *sbi, int type) 1862 { 1863 unsigned long interval = sbi->interval_time[type] * HZ; 1864 1865 return time_after(jiffies, sbi->last_time[type] + interval); 1866 } 1867 1868 static inline unsigned int f2fs_time_to_wait(struct f2fs_sb_info *sbi, 1869 int type) 1870 { 1871 unsigned long interval = sbi->interval_time[type] * HZ; 1872 unsigned int wait_ms = 0; 1873 long delta; 1874 1875 delta = (sbi->last_time[type] + interval) - jiffies; 1876 if (delta > 0) 1877 wait_ms = jiffies_to_msecs(delta); 1878 1879 return wait_ms; 1880 } 1881 1882 /* 1883 * Inline functions 1884 */ 1885 static inline u32 __f2fs_crc32(struct f2fs_sb_info *sbi, u32 crc, 1886 const void *address, unsigned int length) 1887 { 1888 struct { 1889 struct shash_desc shash; 1890 char ctx[4]; 1891 } desc; 1892 int err; 1893 1894 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver) != sizeof(desc.ctx)); 1895 1896 desc.shash.tfm = sbi->s_chksum_driver; 1897 *(u32 *)desc.ctx = crc; 1898 1899 err = crypto_shash_update(&desc.shash, address, length); 1900 BUG_ON(err); 1901 1902 return *(u32 *)desc.ctx; 1903 } 1904 1905 static inline u32 f2fs_crc32(struct f2fs_sb_info *sbi, const void *address, 1906 unsigned int length) 1907 { 1908 return __f2fs_crc32(sbi, F2FS_SUPER_MAGIC, address, length); 1909 } 1910 1911 static inline bool f2fs_crc_valid(struct f2fs_sb_info *sbi, __u32 blk_crc, 1912 void *buf, size_t buf_size) 1913 { 1914 return f2fs_crc32(sbi, buf, buf_size) == blk_crc; 1915 } 1916 1917 static inline u32 f2fs_chksum(struct f2fs_sb_info *sbi, u32 crc, 1918 const void *address, unsigned int length) 1919 { 1920 return __f2fs_crc32(sbi, crc, address, length); 1921 } 1922 1923 static inline struct f2fs_inode_info *F2FS_I(struct inode *inode) 1924 { 1925 return container_of(inode, struct f2fs_inode_info, vfs_inode); 1926 } 1927 1928 static inline struct f2fs_sb_info *F2FS_SB(struct super_block *sb) 1929 { 1930 return sb->s_fs_info; 1931 } 1932 1933 static inline struct f2fs_sb_info *F2FS_I_SB(struct inode *inode) 1934 { 1935 return F2FS_SB(inode->i_sb); 1936 } 1937 1938 static inline struct f2fs_sb_info *F2FS_M_SB(struct address_space *mapping) 1939 { 1940 return F2FS_I_SB(mapping->host); 1941 } 1942 1943 static inline struct f2fs_sb_info *F2FS_P_SB(struct page *page) 1944 { 1945 return F2FS_M_SB(page_file_mapping(page)); 1946 } 1947 1948 static inline struct f2fs_super_block *F2FS_RAW_SUPER(struct f2fs_sb_info *sbi) 1949 { 1950 return (struct f2fs_super_block *)(sbi->raw_super); 1951 } 1952 1953 static inline struct f2fs_checkpoint *F2FS_CKPT(struct f2fs_sb_info *sbi) 1954 { 1955 return (struct f2fs_checkpoint *)(sbi->ckpt); 1956 } 1957 1958 static inline struct f2fs_node *F2FS_NODE(struct page *page) 1959 { 1960 return (struct f2fs_node *)page_address(page); 1961 } 1962 1963 static inline struct f2fs_inode *F2FS_INODE(struct page *page) 1964 { 1965 return &((struct f2fs_node *)page_address(page))->i; 1966 } 1967 1968 static inline struct f2fs_nm_info *NM_I(struct f2fs_sb_info *sbi) 1969 { 1970 return (struct f2fs_nm_info *)(sbi->nm_info); 1971 } 1972 1973 static inline struct f2fs_sm_info *SM_I(struct f2fs_sb_info *sbi) 1974 { 1975 return (struct f2fs_sm_info *)(sbi->sm_info); 1976 } 1977 1978 static inline struct sit_info *SIT_I(struct f2fs_sb_info *sbi) 1979 { 1980 return (struct sit_info *)(SM_I(sbi)->sit_info); 1981 } 1982 1983 static inline struct free_segmap_info *FREE_I(struct f2fs_sb_info *sbi) 1984 { 1985 return (struct free_segmap_info *)(SM_I(sbi)->free_info); 1986 } 1987 1988 static inline struct dirty_seglist_info *DIRTY_I(struct f2fs_sb_info *sbi) 1989 { 1990 return (struct dirty_seglist_info *)(SM_I(sbi)->dirty_info); 1991 } 1992 1993 static inline struct address_space *META_MAPPING(struct f2fs_sb_info *sbi) 1994 { 1995 return sbi->meta_inode->i_mapping; 1996 } 1997 1998 static inline struct address_space *NODE_MAPPING(struct f2fs_sb_info *sbi) 1999 { 2000 return sbi->node_inode->i_mapping; 2001 } 2002 2003 static inline bool is_sbi_flag_set(struct f2fs_sb_info *sbi, unsigned int type) 2004 { 2005 return test_bit(type, &sbi->s_flag); 2006 } 2007 2008 static inline void set_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type) 2009 { 2010 set_bit(type, &sbi->s_flag); 2011 } 2012 2013 static inline void clear_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type) 2014 { 2015 clear_bit(type, &sbi->s_flag); 2016 } 2017 2018 static inline unsigned long long cur_cp_version(struct f2fs_checkpoint *cp) 2019 { 2020 return le64_to_cpu(cp->checkpoint_ver); 2021 } 2022 2023 static inline unsigned long f2fs_qf_ino(struct super_block *sb, int type) 2024 { 2025 if (type < F2FS_MAX_QUOTAS) 2026 return le32_to_cpu(F2FS_SB(sb)->raw_super->qf_ino[type]); 2027 return 0; 2028 } 2029 2030 static inline __u64 cur_cp_crc(struct f2fs_checkpoint *cp) 2031 { 2032 size_t crc_offset = le32_to_cpu(cp->checksum_offset); 2033 return le32_to_cpu(*((__le32 *)((unsigned char *)cp + crc_offset))); 2034 } 2035 2036 static inline bool __is_set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 2037 { 2038 unsigned int ckpt_flags = le32_to_cpu(cp->ckpt_flags); 2039 2040 return ckpt_flags & f; 2041 } 2042 2043 static inline bool is_set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 2044 { 2045 return __is_set_ckpt_flags(F2FS_CKPT(sbi), f); 2046 } 2047 2048 static inline void __set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 2049 { 2050 unsigned int ckpt_flags; 2051 2052 ckpt_flags = le32_to_cpu(cp->ckpt_flags); 2053 ckpt_flags |= f; 2054 cp->ckpt_flags = cpu_to_le32(ckpt_flags); 2055 } 2056 2057 static inline void set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 2058 { 2059 unsigned long flags; 2060 2061 spin_lock_irqsave(&sbi->cp_lock, flags); 2062 __set_ckpt_flags(F2FS_CKPT(sbi), f); 2063 spin_unlock_irqrestore(&sbi->cp_lock, flags); 2064 } 2065 2066 static inline void __clear_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 2067 { 2068 unsigned int ckpt_flags; 2069 2070 ckpt_flags = le32_to_cpu(cp->ckpt_flags); 2071 ckpt_flags &= (~f); 2072 cp->ckpt_flags = cpu_to_le32(ckpt_flags); 2073 } 2074 2075 static inline void clear_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 2076 { 2077 unsigned long flags; 2078 2079 spin_lock_irqsave(&sbi->cp_lock, flags); 2080 __clear_ckpt_flags(F2FS_CKPT(sbi), f); 2081 spin_unlock_irqrestore(&sbi->cp_lock, flags); 2082 } 2083 2084 #define init_f2fs_rwsem(sem) \ 2085 do { \ 2086 static struct lock_class_key __key; \ 2087 \ 2088 __init_f2fs_rwsem((sem), #sem, &__key); \ 2089 } while (0) 2090 2091 static inline void __init_f2fs_rwsem(struct f2fs_rwsem *sem, 2092 const char *sem_name, struct lock_class_key *key) 2093 { 2094 __init_rwsem(&sem->internal_rwsem, sem_name, key); 2095 #ifdef CONFIG_F2FS_UNFAIR_RWSEM 2096 init_waitqueue_head(&sem->read_waiters); 2097 #endif 2098 } 2099 2100 static inline int f2fs_rwsem_is_locked(struct f2fs_rwsem *sem) 2101 { 2102 return rwsem_is_locked(&sem->internal_rwsem); 2103 } 2104 2105 static inline int f2fs_rwsem_is_contended(struct f2fs_rwsem *sem) 2106 { 2107 return rwsem_is_contended(&sem->internal_rwsem); 2108 } 2109 2110 static inline void f2fs_down_read(struct f2fs_rwsem *sem) 2111 { 2112 #ifdef CONFIG_F2FS_UNFAIR_RWSEM 2113 wait_event(sem->read_waiters, down_read_trylock(&sem->internal_rwsem)); 2114 #else 2115 down_read(&sem->internal_rwsem); 2116 #endif 2117 } 2118 2119 static inline int f2fs_down_read_trylock(struct f2fs_rwsem *sem) 2120 { 2121 return down_read_trylock(&sem->internal_rwsem); 2122 } 2123 2124 static inline void f2fs_up_read(struct f2fs_rwsem *sem) 2125 { 2126 up_read(&sem->internal_rwsem); 2127 } 2128 2129 static inline void f2fs_down_write(struct f2fs_rwsem *sem) 2130 { 2131 down_write(&sem->internal_rwsem); 2132 } 2133 2134 #ifdef CONFIG_DEBUG_LOCK_ALLOC 2135 static inline void f2fs_down_read_nested(struct f2fs_rwsem *sem, int subclass) 2136 { 2137 down_read_nested(&sem->internal_rwsem, subclass); 2138 } 2139 2140 static inline void f2fs_down_write_nested(struct f2fs_rwsem *sem, int subclass) 2141 { 2142 down_write_nested(&sem->internal_rwsem, subclass); 2143 } 2144 #else 2145 #define f2fs_down_read_nested(sem, subclass) f2fs_down_read(sem) 2146 #define f2fs_down_write_nested(sem, subclass) f2fs_down_write(sem) 2147 #endif 2148 2149 static inline int f2fs_down_write_trylock(struct f2fs_rwsem *sem) 2150 { 2151 return down_write_trylock(&sem->internal_rwsem); 2152 } 2153 2154 static inline void f2fs_up_write(struct f2fs_rwsem *sem) 2155 { 2156 up_write(&sem->internal_rwsem); 2157 #ifdef CONFIG_F2FS_UNFAIR_RWSEM 2158 wake_up_all(&sem->read_waiters); 2159 #endif 2160 } 2161 2162 static inline void f2fs_lock_op(struct f2fs_sb_info *sbi) 2163 { 2164 f2fs_down_read(&sbi->cp_rwsem); 2165 } 2166 2167 static inline int f2fs_trylock_op(struct f2fs_sb_info *sbi) 2168 { 2169 if (time_to_inject(sbi, FAULT_LOCK_OP)) 2170 return 0; 2171 return f2fs_down_read_trylock(&sbi->cp_rwsem); 2172 } 2173 2174 static inline void f2fs_unlock_op(struct f2fs_sb_info *sbi) 2175 { 2176 f2fs_up_read(&sbi->cp_rwsem); 2177 } 2178 2179 static inline void f2fs_lock_all(struct f2fs_sb_info *sbi) 2180 { 2181 f2fs_down_write(&sbi->cp_rwsem); 2182 } 2183 2184 static inline void f2fs_unlock_all(struct f2fs_sb_info *sbi) 2185 { 2186 f2fs_up_write(&sbi->cp_rwsem); 2187 } 2188 2189 static inline int __get_cp_reason(struct f2fs_sb_info *sbi) 2190 { 2191 int reason = CP_SYNC; 2192 2193 if (test_opt(sbi, FASTBOOT)) 2194 reason = CP_FASTBOOT; 2195 if (is_sbi_flag_set(sbi, SBI_IS_CLOSE)) 2196 reason = CP_UMOUNT; 2197 return reason; 2198 } 2199 2200 static inline bool __remain_node_summaries(int reason) 2201 { 2202 return (reason & (CP_UMOUNT | CP_FASTBOOT)); 2203 } 2204 2205 static inline bool __exist_node_summaries(struct f2fs_sb_info *sbi) 2206 { 2207 return (is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG) || 2208 is_set_ckpt_flags(sbi, CP_FASTBOOT_FLAG)); 2209 } 2210 2211 /* 2212 * Check whether the inode has blocks or not 2213 */ 2214 static inline int F2FS_HAS_BLOCKS(struct inode *inode) 2215 { 2216 block_t xattr_block = F2FS_I(inode)->i_xattr_nid ? 1 : 0; 2217 2218 return (inode->i_blocks >> F2FS_LOG_SECTORS_PER_BLOCK) > xattr_block; 2219 } 2220 2221 static inline bool f2fs_has_xattr_block(unsigned int ofs) 2222 { 2223 return ofs == XATTR_NODE_OFFSET; 2224 } 2225 2226 static inline bool __allow_reserved_blocks(struct f2fs_sb_info *sbi, 2227 struct inode *inode, bool cap) 2228 { 2229 if (!inode) 2230 return true; 2231 if (!test_opt(sbi, RESERVE_ROOT)) 2232 return false; 2233 if (IS_NOQUOTA(inode)) 2234 return true; 2235 if (uid_eq(F2FS_OPTION(sbi).s_resuid, current_fsuid())) 2236 return true; 2237 if (!gid_eq(F2FS_OPTION(sbi).s_resgid, GLOBAL_ROOT_GID) && 2238 in_group_p(F2FS_OPTION(sbi).s_resgid)) 2239 return true; 2240 if (cap && capable(CAP_SYS_RESOURCE)) 2241 return true; 2242 return false; 2243 } 2244 2245 static inline void f2fs_i_blocks_write(struct inode *, block_t, bool, bool); 2246 static inline int inc_valid_block_count(struct f2fs_sb_info *sbi, 2247 struct inode *inode, blkcnt_t *count, bool partial) 2248 { 2249 blkcnt_t diff = 0, release = 0; 2250 block_t avail_user_block_count; 2251 int ret; 2252 2253 ret = dquot_reserve_block(inode, *count); 2254 if (ret) 2255 return ret; 2256 2257 if (time_to_inject(sbi, FAULT_BLOCK)) { 2258 release = *count; 2259 goto release_quota; 2260 } 2261 2262 /* 2263 * let's increase this in prior to actual block count change in order 2264 * for f2fs_sync_file to avoid data races when deciding checkpoint. 2265 */ 2266 percpu_counter_add(&sbi->alloc_valid_block_count, (*count)); 2267 2268 spin_lock(&sbi->stat_lock); 2269 sbi->total_valid_block_count += (block_t)(*count); 2270 avail_user_block_count = sbi->user_block_count - 2271 sbi->current_reserved_blocks; 2272 2273 if (!__allow_reserved_blocks(sbi, inode, true)) 2274 avail_user_block_count -= F2FS_OPTION(sbi).root_reserved_blocks; 2275 2276 if (F2FS_IO_ALIGNED(sbi)) 2277 avail_user_block_count -= sbi->blocks_per_seg * 2278 SM_I(sbi)->additional_reserved_segments; 2279 2280 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) { 2281 if (avail_user_block_count > sbi->unusable_block_count) 2282 avail_user_block_count -= sbi->unusable_block_count; 2283 else 2284 avail_user_block_count = 0; 2285 } 2286 if (unlikely(sbi->total_valid_block_count > avail_user_block_count)) { 2287 if (!partial) { 2288 spin_unlock(&sbi->stat_lock); 2289 goto enospc; 2290 } 2291 2292 diff = sbi->total_valid_block_count - avail_user_block_count; 2293 if (diff > *count) 2294 diff = *count; 2295 *count -= diff; 2296 release = diff; 2297 sbi->total_valid_block_count -= diff; 2298 if (!*count) { 2299 spin_unlock(&sbi->stat_lock); 2300 goto enospc; 2301 } 2302 } 2303 spin_unlock(&sbi->stat_lock); 2304 2305 if (unlikely(release)) { 2306 percpu_counter_sub(&sbi->alloc_valid_block_count, release); 2307 dquot_release_reservation_block(inode, release); 2308 } 2309 f2fs_i_blocks_write(inode, *count, true, true); 2310 return 0; 2311 2312 enospc: 2313 percpu_counter_sub(&sbi->alloc_valid_block_count, release); 2314 release_quota: 2315 dquot_release_reservation_block(inode, release); 2316 return -ENOSPC; 2317 } 2318 2319 __printf(2, 3) 2320 void f2fs_printk(struct f2fs_sb_info *sbi, const char *fmt, ...); 2321 2322 #define f2fs_err(sbi, fmt, ...) \ 2323 f2fs_printk(sbi, KERN_ERR fmt, ##__VA_ARGS__) 2324 #define f2fs_warn(sbi, fmt, ...) \ 2325 f2fs_printk(sbi, KERN_WARNING fmt, ##__VA_ARGS__) 2326 #define f2fs_notice(sbi, fmt, ...) \ 2327 f2fs_printk(sbi, KERN_NOTICE fmt, ##__VA_ARGS__) 2328 #define f2fs_info(sbi, fmt, ...) \ 2329 f2fs_printk(sbi, KERN_INFO fmt, ##__VA_ARGS__) 2330 #define f2fs_debug(sbi, fmt, ...) \ 2331 f2fs_printk(sbi, KERN_DEBUG fmt, ##__VA_ARGS__) 2332 2333 #define PAGE_PRIVATE_GET_FUNC(name, flagname) \ 2334 static inline bool page_private_##name(struct page *page) \ 2335 { \ 2336 return PagePrivate(page) && \ 2337 test_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)) && \ 2338 test_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \ 2339 } 2340 2341 #define PAGE_PRIVATE_SET_FUNC(name, flagname) \ 2342 static inline void set_page_private_##name(struct page *page) \ 2343 { \ 2344 if (!PagePrivate(page)) \ 2345 attach_page_private(page, (void *)0); \ 2346 set_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)); \ 2347 set_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \ 2348 } 2349 2350 #define PAGE_PRIVATE_CLEAR_FUNC(name, flagname) \ 2351 static inline void clear_page_private_##name(struct page *page) \ 2352 { \ 2353 clear_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \ 2354 if (page_private(page) == BIT(PAGE_PRIVATE_NOT_POINTER)) \ 2355 detach_page_private(page); \ 2356 } 2357 2358 PAGE_PRIVATE_GET_FUNC(nonpointer, NOT_POINTER); 2359 PAGE_PRIVATE_GET_FUNC(inline, INLINE_INODE); 2360 PAGE_PRIVATE_GET_FUNC(gcing, ONGOING_MIGRATION); 2361 PAGE_PRIVATE_GET_FUNC(dummy, DUMMY_WRITE); 2362 2363 PAGE_PRIVATE_SET_FUNC(reference, REF_RESOURCE); 2364 PAGE_PRIVATE_SET_FUNC(inline, INLINE_INODE); 2365 PAGE_PRIVATE_SET_FUNC(gcing, ONGOING_MIGRATION); 2366 PAGE_PRIVATE_SET_FUNC(dummy, DUMMY_WRITE); 2367 2368 PAGE_PRIVATE_CLEAR_FUNC(reference, REF_RESOURCE); 2369 PAGE_PRIVATE_CLEAR_FUNC(inline, INLINE_INODE); 2370 PAGE_PRIVATE_CLEAR_FUNC(gcing, ONGOING_MIGRATION); 2371 PAGE_PRIVATE_CLEAR_FUNC(dummy, DUMMY_WRITE); 2372 2373 static inline unsigned long get_page_private_data(struct page *page) 2374 { 2375 unsigned long data = page_private(page); 2376 2377 if (!test_bit(PAGE_PRIVATE_NOT_POINTER, &data)) 2378 return 0; 2379 return data >> PAGE_PRIVATE_MAX; 2380 } 2381 2382 static inline void set_page_private_data(struct page *page, unsigned long data) 2383 { 2384 if (!PagePrivate(page)) 2385 attach_page_private(page, (void *)0); 2386 set_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)); 2387 page_private(page) |= data << PAGE_PRIVATE_MAX; 2388 } 2389 2390 static inline void clear_page_private_data(struct page *page) 2391 { 2392 page_private(page) &= GENMASK(PAGE_PRIVATE_MAX - 1, 0); 2393 if (page_private(page) == BIT(PAGE_PRIVATE_NOT_POINTER)) 2394 detach_page_private(page); 2395 } 2396 2397 static inline void clear_page_private_all(struct page *page) 2398 { 2399 clear_page_private_data(page); 2400 clear_page_private_reference(page); 2401 clear_page_private_gcing(page); 2402 clear_page_private_inline(page); 2403 2404 f2fs_bug_on(F2FS_P_SB(page), page_private(page)); 2405 } 2406 2407 static inline void dec_valid_block_count(struct f2fs_sb_info *sbi, 2408 struct inode *inode, 2409 block_t count) 2410 { 2411 blkcnt_t sectors = count << F2FS_LOG_SECTORS_PER_BLOCK; 2412 2413 spin_lock(&sbi->stat_lock); 2414 f2fs_bug_on(sbi, sbi->total_valid_block_count < (block_t) count); 2415 sbi->total_valid_block_count -= (block_t)count; 2416 if (sbi->reserved_blocks && 2417 sbi->current_reserved_blocks < sbi->reserved_blocks) 2418 sbi->current_reserved_blocks = min(sbi->reserved_blocks, 2419 sbi->current_reserved_blocks + count); 2420 spin_unlock(&sbi->stat_lock); 2421 if (unlikely(inode->i_blocks < sectors)) { 2422 f2fs_warn(sbi, "Inconsistent i_blocks, ino:%lu, iblocks:%llu, sectors:%llu", 2423 inode->i_ino, 2424 (unsigned long long)inode->i_blocks, 2425 (unsigned long long)sectors); 2426 set_sbi_flag(sbi, SBI_NEED_FSCK); 2427 return; 2428 } 2429 f2fs_i_blocks_write(inode, count, false, true); 2430 } 2431 2432 static inline void inc_page_count(struct f2fs_sb_info *sbi, int count_type) 2433 { 2434 atomic_inc(&sbi->nr_pages[count_type]); 2435 2436 if (count_type == F2FS_DIRTY_DENTS || 2437 count_type == F2FS_DIRTY_NODES || 2438 count_type == F2FS_DIRTY_META || 2439 count_type == F2FS_DIRTY_QDATA || 2440 count_type == F2FS_DIRTY_IMETA) 2441 set_sbi_flag(sbi, SBI_IS_DIRTY); 2442 } 2443 2444 static inline void inode_inc_dirty_pages(struct inode *inode) 2445 { 2446 atomic_inc(&F2FS_I(inode)->dirty_pages); 2447 inc_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ? 2448 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA); 2449 if (IS_NOQUOTA(inode)) 2450 inc_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA); 2451 } 2452 2453 static inline void dec_page_count(struct f2fs_sb_info *sbi, int count_type) 2454 { 2455 atomic_dec(&sbi->nr_pages[count_type]); 2456 } 2457 2458 static inline void inode_dec_dirty_pages(struct inode *inode) 2459 { 2460 if (!S_ISDIR(inode->i_mode) && !S_ISREG(inode->i_mode) && 2461 !S_ISLNK(inode->i_mode)) 2462 return; 2463 2464 atomic_dec(&F2FS_I(inode)->dirty_pages); 2465 dec_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ? 2466 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA); 2467 if (IS_NOQUOTA(inode)) 2468 dec_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA); 2469 } 2470 2471 static inline void inc_atomic_write_cnt(struct inode *inode) 2472 { 2473 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 2474 struct f2fs_inode_info *fi = F2FS_I(inode); 2475 u64 current_write; 2476 2477 fi->atomic_write_cnt++; 2478 atomic64_inc(&sbi->current_atomic_write); 2479 current_write = atomic64_read(&sbi->current_atomic_write); 2480 if (current_write > sbi->peak_atomic_write) 2481 sbi->peak_atomic_write = current_write; 2482 } 2483 2484 static inline void release_atomic_write_cnt(struct inode *inode) 2485 { 2486 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 2487 struct f2fs_inode_info *fi = F2FS_I(inode); 2488 2489 atomic64_sub(fi->atomic_write_cnt, &sbi->current_atomic_write); 2490 fi->atomic_write_cnt = 0; 2491 } 2492 2493 static inline s64 get_pages(struct f2fs_sb_info *sbi, int count_type) 2494 { 2495 return atomic_read(&sbi->nr_pages[count_type]); 2496 } 2497 2498 static inline int get_dirty_pages(struct inode *inode) 2499 { 2500 return atomic_read(&F2FS_I(inode)->dirty_pages); 2501 } 2502 2503 static inline int get_blocktype_secs(struct f2fs_sb_info *sbi, int block_type) 2504 { 2505 unsigned int pages_per_sec = sbi->segs_per_sec * sbi->blocks_per_seg; 2506 unsigned int segs = (get_pages(sbi, block_type) + pages_per_sec - 1) >> 2507 sbi->log_blocks_per_seg; 2508 2509 return segs / sbi->segs_per_sec; 2510 } 2511 2512 static inline block_t valid_user_blocks(struct f2fs_sb_info *sbi) 2513 { 2514 return sbi->total_valid_block_count; 2515 } 2516 2517 static inline block_t discard_blocks(struct f2fs_sb_info *sbi) 2518 { 2519 return sbi->discard_blks; 2520 } 2521 2522 static inline unsigned long __bitmap_size(struct f2fs_sb_info *sbi, int flag) 2523 { 2524 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi); 2525 2526 /* return NAT or SIT bitmap */ 2527 if (flag == NAT_BITMAP) 2528 return le32_to_cpu(ckpt->nat_ver_bitmap_bytesize); 2529 else if (flag == SIT_BITMAP) 2530 return le32_to_cpu(ckpt->sit_ver_bitmap_bytesize); 2531 2532 return 0; 2533 } 2534 2535 static inline block_t __cp_payload(struct f2fs_sb_info *sbi) 2536 { 2537 return le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_payload); 2538 } 2539 2540 static inline void *__bitmap_ptr(struct f2fs_sb_info *sbi, int flag) 2541 { 2542 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi); 2543 void *tmp_ptr = &ckpt->sit_nat_version_bitmap; 2544 int offset; 2545 2546 if (is_set_ckpt_flags(sbi, CP_LARGE_NAT_BITMAP_FLAG)) { 2547 offset = (flag == SIT_BITMAP) ? 2548 le32_to_cpu(ckpt->nat_ver_bitmap_bytesize) : 0; 2549 /* 2550 * if large_nat_bitmap feature is enabled, leave checksum 2551 * protection for all nat/sit bitmaps. 2552 */ 2553 return tmp_ptr + offset + sizeof(__le32); 2554 } 2555 2556 if (__cp_payload(sbi) > 0) { 2557 if (flag == NAT_BITMAP) 2558 return tmp_ptr; 2559 else 2560 return (unsigned char *)ckpt + F2FS_BLKSIZE; 2561 } else { 2562 offset = (flag == NAT_BITMAP) ? 2563 le32_to_cpu(ckpt->sit_ver_bitmap_bytesize) : 0; 2564 return tmp_ptr + offset; 2565 } 2566 } 2567 2568 static inline block_t __start_cp_addr(struct f2fs_sb_info *sbi) 2569 { 2570 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr); 2571 2572 if (sbi->cur_cp_pack == 2) 2573 start_addr += sbi->blocks_per_seg; 2574 return start_addr; 2575 } 2576 2577 static inline block_t __start_cp_next_addr(struct f2fs_sb_info *sbi) 2578 { 2579 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr); 2580 2581 if (sbi->cur_cp_pack == 1) 2582 start_addr += sbi->blocks_per_seg; 2583 return start_addr; 2584 } 2585 2586 static inline void __set_cp_next_pack(struct f2fs_sb_info *sbi) 2587 { 2588 sbi->cur_cp_pack = (sbi->cur_cp_pack == 1) ? 2 : 1; 2589 } 2590 2591 static inline block_t __start_sum_addr(struct f2fs_sb_info *sbi) 2592 { 2593 return le32_to_cpu(F2FS_CKPT(sbi)->cp_pack_start_sum); 2594 } 2595 2596 extern void f2fs_mark_inode_dirty_sync(struct inode *inode, bool sync); 2597 static inline int inc_valid_node_count(struct f2fs_sb_info *sbi, 2598 struct inode *inode, bool is_inode) 2599 { 2600 block_t valid_block_count; 2601 unsigned int valid_node_count, user_block_count; 2602 int err; 2603 2604 if (is_inode) { 2605 if (inode) { 2606 err = dquot_alloc_inode(inode); 2607 if (err) 2608 return err; 2609 } 2610 } else { 2611 err = dquot_reserve_block(inode, 1); 2612 if (err) 2613 return err; 2614 } 2615 2616 if (time_to_inject(sbi, FAULT_BLOCK)) 2617 goto enospc; 2618 2619 spin_lock(&sbi->stat_lock); 2620 2621 valid_block_count = sbi->total_valid_block_count + 2622 sbi->current_reserved_blocks + 1; 2623 2624 if (!__allow_reserved_blocks(sbi, inode, false)) 2625 valid_block_count += F2FS_OPTION(sbi).root_reserved_blocks; 2626 2627 if (F2FS_IO_ALIGNED(sbi)) 2628 valid_block_count += sbi->blocks_per_seg * 2629 SM_I(sbi)->additional_reserved_segments; 2630 2631 user_block_count = sbi->user_block_count; 2632 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) 2633 user_block_count -= sbi->unusable_block_count; 2634 2635 if (unlikely(valid_block_count > user_block_count)) { 2636 spin_unlock(&sbi->stat_lock); 2637 goto enospc; 2638 } 2639 2640 valid_node_count = sbi->total_valid_node_count + 1; 2641 if (unlikely(valid_node_count > sbi->total_node_count)) { 2642 spin_unlock(&sbi->stat_lock); 2643 goto enospc; 2644 } 2645 2646 sbi->total_valid_node_count++; 2647 sbi->total_valid_block_count++; 2648 spin_unlock(&sbi->stat_lock); 2649 2650 if (inode) { 2651 if (is_inode) 2652 f2fs_mark_inode_dirty_sync(inode, true); 2653 else 2654 f2fs_i_blocks_write(inode, 1, true, true); 2655 } 2656 2657 percpu_counter_inc(&sbi->alloc_valid_block_count); 2658 return 0; 2659 2660 enospc: 2661 if (is_inode) { 2662 if (inode) 2663 dquot_free_inode(inode); 2664 } else { 2665 dquot_release_reservation_block(inode, 1); 2666 } 2667 return -ENOSPC; 2668 } 2669 2670 static inline void dec_valid_node_count(struct f2fs_sb_info *sbi, 2671 struct inode *inode, bool is_inode) 2672 { 2673 spin_lock(&sbi->stat_lock); 2674 2675 if (unlikely(!sbi->total_valid_block_count || 2676 !sbi->total_valid_node_count)) { 2677 f2fs_warn(sbi, "dec_valid_node_count: inconsistent block counts, total_valid_block:%u, total_valid_node:%u", 2678 sbi->total_valid_block_count, 2679 sbi->total_valid_node_count); 2680 set_sbi_flag(sbi, SBI_NEED_FSCK); 2681 } else { 2682 sbi->total_valid_block_count--; 2683 sbi->total_valid_node_count--; 2684 } 2685 2686 if (sbi->reserved_blocks && 2687 sbi->current_reserved_blocks < sbi->reserved_blocks) 2688 sbi->current_reserved_blocks++; 2689 2690 spin_unlock(&sbi->stat_lock); 2691 2692 if (is_inode) { 2693 dquot_free_inode(inode); 2694 } else { 2695 if (unlikely(inode->i_blocks == 0)) { 2696 f2fs_warn(sbi, "dec_valid_node_count: inconsistent i_blocks, ino:%lu, iblocks:%llu", 2697 inode->i_ino, 2698 (unsigned long long)inode->i_blocks); 2699 set_sbi_flag(sbi, SBI_NEED_FSCK); 2700 return; 2701 } 2702 f2fs_i_blocks_write(inode, 1, false, true); 2703 } 2704 } 2705 2706 static inline unsigned int valid_node_count(struct f2fs_sb_info *sbi) 2707 { 2708 return sbi->total_valid_node_count; 2709 } 2710 2711 static inline void inc_valid_inode_count(struct f2fs_sb_info *sbi) 2712 { 2713 percpu_counter_inc(&sbi->total_valid_inode_count); 2714 } 2715 2716 static inline void dec_valid_inode_count(struct f2fs_sb_info *sbi) 2717 { 2718 percpu_counter_dec(&sbi->total_valid_inode_count); 2719 } 2720 2721 static inline s64 valid_inode_count(struct f2fs_sb_info *sbi) 2722 { 2723 return percpu_counter_sum_positive(&sbi->total_valid_inode_count); 2724 } 2725 2726 static inline struct page *f2fs_grab_cache_page(struct address_space *mapping, 2727 pgoff_t index, bool for_write) 2728 { 2729 struct page *page; 2730 unsigned int flags; 2731 2732 if (IS_ENABLED(CONFIG_F2FS_FAULT_INJECTION)) { 2733 if (!for_write) 2734 page = find_get_page_flags(mapping, index, 2735 FGP_LOCK | FGP_ACCESSED); 2736 else 2737 page = find_lock_page(mapping, index); 2738 if (page) 2739 return page; 2740 2741 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_ALLOC)) 2742 return NULL; 2743 } 2744 2745 if (!for_write) 2746 return grab_cache_page(mapping, index); 2747 2748 flags = memalloc_nofs_save(); 2749 page = grab_cache_page_write_begin(mapping, index); 2750 memalloc_nofs_restore(flags); 2751 2752 return page; 2753 } 2754 2755 static inline struct page *f2fs_pagecache_get_page( 2756 struct address_space *mapping, pgoff_t index, 2757 fgf_t fgp_flags, gfp_t gfp_mask) 2758 { 2759 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_GET)) 2760 return NULL; 2761 2762 return pagecache_get_page(mapping, index, fgp_flags, gfp_mask); 2763 } 2764 2765 static inline void f2fs_put_page(struct page *page, int unlock) 2766 { 2767 if (!page) 2768 return; 2769 2770 if (unlock) { 2771 f2fs_bug_on(F2FS_P_SB(page), !PageLocked(page)); 2772 unlock_page(page); 2773 } 2774 put_page(page); 2775 } 2776 2777 static inline void f2fs_put_dnode(struct dnode_of_data *dn) 2778 { 2779 if (dn->node_page) 2780 f2fs_put_page(dn->node_page, 1); 2781 if (dn->inode_page && dn->node_page != dn->inode_page) 2782 f2fs_put_page(dn->inode_page, 0); 2783 dn->node_page = NULL; 2784 dn->inode_page = NULL; 2785 } 2786 2787 static inline struct kmem_cache *f2fs_kmem_cache_create(const char *name, 2788 size_t size) 2789 { 2790 return kmem_cache_create(name, size, 0, SLAB_RECLAIM_ACCOUNT, NULL); 2791 } 2792 2793 static inline void *f2fs_kmem_cache_alloc_nofail(struct kmem_cache *cachep, 2794 gfp_t flags) 2795 { 2796 void *entry; 2797 2798 entry = kmem_cache_alloc(cachep, flags); 2799 if (!entry) 2800 entry = kmem_cache_alloc(cachep, flags | __GFP_NOFAIL); 2801 return entry; 2802 } 2803 2804 static inline void *f2fs_kmem_cache_alloc(struct kmem_cache *cachep, 2805 gfp_t flags, bool nofail, struct f2fs_sb_info *sbi) 2806 { 2807 if (nofail) 2808 return f2fs_kmem_cache_alloc_nofail(cachep, flags); 2809 2810 if (time_to_inject(sbi, FAULT_SLAB_ALLOC)) 2811 return NULL; 2812 2813 return kmem_cache_alloc(cachep, flags); 2814 } 2815 2816 static inline bool is_inflight_io(struct f2fs_sb_info *sbi, int type) 2817 { 2818 if (get_pages(sbi, F2FS_RD_DATA) || get_pages(sbi, F2FS_RD_NODE) || 2819 get_pages(sbi, F2FS_RD_META) || get_pages(sbi, F2FS_WB_DATA) || 2820 get_pages(sbi, F2FS_WB_CP_DATA) || 2821 get_pages(sbi, F2FS_DIO_READ) || 2822 get_pages(sbi, F2FS_DIO_WRITE)) 2823 return true; 2824 2825 if (type != DISCARD_TIME && SM_I(sbi) && SM_I(sbi)->dcc_info && 2826 atomic_read(&SM_I(sbi)->dcc_info->queued_discard)) 2827 return true; 2828 2829 if (SM_I(sbi) && SM_I(sbi)->fcc_info && 2830 atomic_read(&SM_I(sbi)->fcc_info->queued_flush)) 2831 return true; 2832 return false; 2833 } 2834 2835 static inline bool is_idle(struct f2fs_sb_info *sbi, int type) 2836 { 2837 if (sbi->gc_mode == GC_URGENT_HIGH) 2838 return true; 2839 2840 if (is_inflight_io(sbi, type)) 2841 return false; 2842 2843 if (sbi->gc_mode == GC_URGENT_MID) 2844 return true; 2845 2846 if (sbi->gc_mode == GC_URGENT_LOW && 2847 (type == DISCARD_TIME || type == GC_TIME)) 2848 return true; 2849 2850 return f2fs_time_over(sbi, type); 2851 } 2852 2853 static inline void f2fs_radix_tree_insert(struct radix_tree_root *root, 2854 unsigned long index, void *item) 2855 { 2856 while (radix_tree_insert(root, index, item)) 2857 cond_resched(); 2858 } 2859 2860 #define RAW_IS_INODE(p) ((p)->footer.nid == (p)->footer.ino) 2861 2862 static inline bool IS_INODE(struct page *page) 2863 { 2864 struct f2fs_node *p = F2FS_NODE(page); 2865 2866 return RAW_IS_INODE(p); 2867 } 2868 2869 static inline int offset_in_addr(struct f2fs_inode *i) 2870 { 2871 return (i->i_inline & F2FS_EXTRA_ATTR) ? 2872 (le16_to_cpu(i->i_extra_isize) / sizeof(__le32)) : 0; 2873 } 2874 2875 static inline __le32 *blkaddr_in_node(struct f2fs_node *node) 2876 { 2877 return RAW_IS_INODE(node) ? node->i.i_addr : node->dn.addr; 2878 } 2879 2880 static inline int f2fs_has_extra_attr(struct inode *inode); 2881 static inline block_t data_blkaddr(struct inode *inode, 2882 struct page *node_page, unsigned int offset) 2883 { 2884 struct f2fs_node *raw_node; 2885 __le32 *addr_array; 2886 int base = 0; 2887 bool is_inode = IS_INODE(node_page); 2888 2889 raw_node = F2FS_NODE(node_page); 2890 2891 if (is_inode) { 2892 if (!inode) 2893 /* from GC path only */ 2894 base = offset_in_addr(&raw_node->i); 2895 else if (f2fs_has_extra_attr(inode)) 2896 base = get_extra_isize(inode); 2897 } 2898 2899 addr_array = blkaddr_in_node(raw_node); 2900 return le32_to_cpu(addr_array[base + offset]); 2901 } 2902 2903 static inline block_t f2fs_data_blkaddr(struct dnode_of_data *dn) 2904 { 2905 return data_blkaddr(dn->inode, dn->node_page, dn->ofs_in_node); 2906 } 2907 2908 static inline int f2fs_test_bit(unsigned int nr, char *addr) 2909 { 2910 int mask; 2911 2912 addr += (nr >> 3); 2913 mask = BIT(7 - (nr & 0x07)); 2914 return mask & *addr; 2915 } 2916 2917 static inline void f2fs_set_bit(unsigned int nr, char *addr) 2918 { 2919 int mask; 2920 2921 addr += (nr >> 3); 2922 mask = BIT(7 - (nr & 0x07)); 2923 *addr |= mask; 2924 } 2925 2926 static inline void f2fs_clear_bit(unsigned int nr, char *addr) 2927 { 2928 int mask; 2929 2930 addr += (nr >> 3); 2931 mask = BIT(7 - (nr & 0x07)); 2932 *addr &= ~mask; 2933 } 2934 2935 static inline int f2fs_test_and_set_bit(unsigned int nr, char *addr) 2936 { 2937 int mask; 2938 int ret; 2939 2940 addr += (nr >> 3); 2941 mask = BIT(7 - (nr & 0x07)); 2942 ret = mask & *addr; 2943 *addr |= mask; 2944 return ret; 2945 } 2946 2947 static inline int f2fs_test_and_clear_bit(unsigned int nr, char *addr) 2948 { 2949 int mask; 2950 int ret; 2951 2952 addr += (nr >> 3); 2953 mask = BIT(7 - (nr & 0x07)); 2954 ret = mask & *addr; 2955 *addr &= ~mask; 2956 return ret; 2957 } 2958 2959 static inline void f2fs_change_bit(unsigned int nr, char *addr) 2960 { 2961 int mask; 2962 2963 addr += (nr >> 3); 2964 mask = BIT(7 - (nr & 0x07)); 2965 *addr ^= mask; 2966 } 2967 2968 /* 2969 * On-disk inode flags (f2fs_inode::i_flags) 2970 */ 2971 #define F2FS_COMPR_FL 0x00000004 /* Compress file */ 2972 #define F2FS_SYNC_FL 0x00000008 /* Synchronous updates */ 2973 #define F2FS_IMMUTABLE_FL 0x00000010 /* Immutable file */ 2974 #define F2FS_APPEND_FL 0x00000020 /* writes to file may only append */ 2975 #define F2FS_NODUMP_FL 0x00000040 /* do not dump file */ 2976 #define F2FS_NOATIME_FL 0x00000080 /* do not update atime */ 2977 #define F2FS_NOCOMP_FL 0x00000400 /* Don't compress */ 2978 #define F2FS_INDEX_FL 0x00001000 /* hash-indexed directory */ 2979 #define F2FS_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */ 2980 #define F2FS_PROJINHERIT_FL 0x20000000 /* Create with parents projid */ 2981 #define F2FS_CASEFOLD_FL 0x40000000 /* Casefolded file */ 2982 2983 #define F2FS_QUOTA_DEFAULT_FL (F2FS_NOATIME_FL | F2FS_IMMUTABLE_FL) 2984 2985 /* Flags that should be inherited by new inodes from their parent. */ 2986 #define F2FS_FL_INHERITED (F2FS_SYNC_FL | F2FS_NODUMP_FL | F2FS_NOATIME_FL | \ 2987 F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \ 2988 F2FS_CASEFOLD_FL) 2989 2990 /* Flags that are appropriate for regular files (all but dir-specific ones). */ 2991 #define F2FS_REG_FLMASK (~(F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \ 2992 F2FS_CASEFOLD_FL)) 2993 2994 /* Flags that are appropriate for non-directories/regular files. */ 2995 #define F2FS_OTHER_FLMASK (F2FS_NODUMP_FL | F2FS_NOATIME_FL) 2996 2997 static inline __u32 f2fs_mask_flags(umode_t mode, __u32 flags) 2998 { 2999 if (S_ISDIR(mode)) 3000 return flags; 3001 else if (S_ISREG(mode)) 3002 return flags & F2FS_REG_FLMASK; 3003 else 3004 return flags & F2FS_OTHER_FLMASK; 3005 } 3006 3007 static inline void __mark_inode_dirty_flag(struct inode *inode, 3008 int flag, bool set) 3009 { 3010 switch (flag) { 3011 case FI_INLINE_XATTR: 3012 case FI_INLINE_DATA: 3013 case FI_INLINE_DENTRY: 3014 case FI_NEW_INODE: 3015 if (set) 3016 return; 3017 fallthrough; 3018 case FI_DATA_EXIST: 3019 case FI_INLINE_DOTS: 3020 case FI_PIN_FILE: 3021 case FI_COMPRESS_RELEASED: 3022 f2fs_mark_inode_dirty_sync(inode, true); 3023 } 3024 } 3025 3026 static inline void set_inode_flag(struct inode *inode, int flag) 3027 { 3028 set_bit(flag, F2FS_I(inode)->flags); 3029 __mark_inode_dirty_flag(inode, flag, true); 3030 } 3031 3032 static inline int is_inode_flag_set(struct inode *inode, int flag) 3033 { 3034 return test_bit(flag, F2FS_I(inode)->flags); 3035 } 3036 3037 static inline void clear_inode_flag(struct inode *inode, int flag) 3038 { 3039 clear_bit(flag, F2FS_I(inode)->flags); 3040 __mark_inode_dirty_flag(inode, flag, false); 3041 } 3042 3043 static inline bool f2fs_verity_in_progress(struct inode *inode) 3044 { 3045 return IS_ENABLED(CONFIG_FS_VERITY) && 3046 is_inode_flag_set(inode, FI_VERITY_IN_PROGRESS); 3047 } 3048 3049 static inline void set_acl_inode(struct inode *inode, umode_t mode) 3050 { 3051 F2FS_I(inode)->i_acl_mode = mode; 3052 set_inode_flag(inode, FI_ACL_MODE); 3053 f2fs_mark_inode_dirty_sync(inode, false); 3054 } 3055 3056 static inline void f2fs_i_links_write(struct inode *inode, bool inc) 3057 { 3058 if (inc) 3059 inc_nlink(inode); 3060 else 3061 drop_nlink(inode); 3062 f2fs_mark_inode_dirty_sync(inode, true); 3063 } 3064 3065 static inline void f2fs_i_blocks_write(struct inode *inode, 3066 block_t diff, bool add, bool claim) 3067 { 3068 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE); 3069 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER); 3070 3071 /* add = 1, claim = 1 should be dquot_reserve_block in pair */ 3072 if (add) { 3073 if (claim) 3074 dquot_claim_block(inode, diff); 3075 else 3076 dquot_alloc_block_nofail(inode, diff); 3077 } else { 3078 dquot_free_block(inode, diff); 3079 } 3080 3081 f2fs_mark_inode_dirty_sync(inode, true); 3082 if (clean || recover) 3083 set_inode_flag(inode, FI_AUTO_RECOVER); 3084 } 3085 3086 static inline bool f2fs_is_atomic_file(struct inode *inode); 3087 3088 static inline void f2fs_i_size_write(struct inode *inode, loff_t i_size) 3089 { 3090 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE); 3091 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER); 3092 3093 if (i_size_read(inode) == i_size) 3094 return; 3095 3096 i_size_write(inode, i_size); 3097 3098 if (f2fs_is_atomic_file(inode)) 3099 return; 3100 3101 f2fs_mark_inode_dirty_sync(inode, true); 3102 if (clean || recover) 3103 set_inode_flag(inode, FI_AUTO_RECOVER); 3104 } 3105 3106 static inline void f2fs_i_depth_write(struct inode *inode, unsigned int depth) 3107 { 3108 F2FS_I(inode)->i_current_depth = depth; 3109 f2fs_mark_inode_dirty_sync(inode, true); 3110 } 3111 3112 static inline void f2fs_i_gc_failures_write(struct inode *inode, 3113 unsigned int count) 3114 { 3115 F2FS_I(inode)->i_gc_failures[GC_FAILURE_PIN] = count; 3116 f2fs_mark_inode_dirty_sync(inode, true); 3117 } 3118 3119 static inline void f2fs_i_xnid_write(struct inode *inode, nid_t xnid) 3120 { 3121 F2FS_I(inode)->i_xattr_nid = xnid; 3122 f2fs_mark_inode_dirty_sync(inode, true); 3123 } 3124 3125 static inline void f2fs_i_pino_write(struct inode *inode, nid_t pino) 3126 { 3127 F2FS_I(inode)->i_pino = pino; 3128 f2fs_mark_inode_dirty_sync(inode, true); 3129 } 3130 3131 static inline void get_inline_info(struct inode *inode, struct f2fs_inode *ri) 3132 { 3133 struct f2fs_inode_info *fi = F2FS_I(inode); 3134 3135 if (ri->i_inline & F2FS_INLINE_XATTR) 3136 set_bit(FI_INLINE_XATTR, fi->flags); 3137 if (ri->i_inline & F2FS_INLINE_DATA) 3138 set_bit(FI_INLINE_DATA, fi->flags); 3139 if (ri->i_inline & F2FS_INLINE_DENTRY) 3140 set_bit(FI_INLINE_DENTRY, fi->flags); 3141 if (ri->i_inline & F2FS_DATA_EXIST) 3142 set_bit(FI_DATA_EXIST, fi->flags); 3143 if (ri->i_inline & F2FS_INLINE_DOTS) 3144 set_bit(FI_INLINE_DOTS, fi->flags); 3145 if (ri->i_inline & F2FS_EXTRA_ATTR) 3146 set_bit(FI_EXTRA_ATTR, fi->flags); 3147 if (ri->i_inline & F2FS_PIN_FILE) 3148 set_bit(FI_PIN_FILE, fi->flags); 3149 if (ri->i_inline & F2FS_COMPRESS_RELEASED) 3150 set_bit(FI_COMPRESS_RELEASED, fi->flags); 3151 } 3152 3153 static inline void set_raw_inline(struct inode *inode, struct f2fs_inode *ri) 3154 { 3155 ri->i_inline = 0; 3156 3157 if (is_inode_flag_set(inode, FI_INLINE_XATTR)) 3158 ri->i_inline |= F2FS_INLINE_XATTR; 3159 if (is_inode_flag_set(inode, FI_INLINE_DATA)) 3160 ri->i_inline |= F2FS_INLINE_DATA; 3161 if (is_inode_flag_set(inode, FI_INLINE_DENTRY)) 3162 ri->i_inline |= F2FS_INLINE_DENTRY; 3163 if (is_inode_flag_set(inode, FI_DATA_EXIST)) 3164 ri->i_inline |= F2FS_DATA_EXIST; 3165 if (is_inode_flag_set(inode, FI_INLINE_DOTS)) 3166 ri->i_inline |= F2FS_INLINE_DOTS; 3167 if (is_inode_flag_set(inode, FI_EXTRA_ATTR)) 3168 ri->i_inline |= F2FS_EXTRA_ATTR; 3169 if (is_inode_flag_set(inode, FI_PIN_FILE)) 3170 ri->i_inline |= F2FS_PIN_FILE; 3171 if (is_inode_flag_set(inode, FI_COMPRESS_RELEASED)) 3172 ri->i_inline |= F2FS_COMPRESS_RELEASED; 3173 } 3174 3175 static inline int f2fs_has_extra_attr(struct inode *inode) 3176 { 3177 return is_inode_flag_set(inode, FI_EXTRA_ATTR); 3178 } 3179 3180 static inline int f2fs_has_inline_xattr(struct inode *inode) 3181 { 3182 return is_inode_flag_set(inode, FI_INLINE_XATTR); 3183 } 3184 3185 static inline int f2fs_compressed_file(struct inode *inode) 3186 { 3187 return S_ISREG(inode->i_mode) && 3188 is_inode_flag_set(inode, FI_COMPRESSED_FILE); 3189 } 3190 3191 static inline bool f2fs_need_compress_data(struct inode *inode) 3192 { 3193 int compress_mode = F2FS_OPTION(F2FS_I_SB(inode)).compress_mode; 3194 3195 if (!f2fs_compressed_file(inode)) 3196 return false; 3197 3198 if (compress_mode == COMPR_MODE_FS) 3199 return true; 3200 else if (compress_mode == COMPR_MODE_USER && 3201 is_inode_flag_set(inode, FI_ENABLE_COMPRESS)) 3202 return true; 3203 3204 return false; 3205 } 3206 3207 static inline unsigned int addrs_per_inode(struct inode *inode) 3208 { 3209 unsigned int addrs = CUR_ADDRS_PER_INODE(inode) - 3210 get_inline_xattr_addrs(inode); 3211 3212 if (!f2fs_compressed_file(inode)) 3213 return addrs; 3214 return ALIGN_DOWN(addrs, F2FS_I(inode)->i_cluster_size); 3215 } 3216 3217 static inline unsigned int addrs_per_block(struct inode *inode) 3218 { 3219 if (!f2fs_compressed_file(inode)) 3220 return DEF_ADDRS_PER_BLOCK; 3221 return ALIGN_DOWN(DEF_ADDRS_PER_BLOCK, F2FS_I(inode)->i_cluster_size); 3222 } 3223 3224 static inline void *inline_xattr_addr(struct inode *inode, struct page *page) 3225 { 3226 struct f2fs_inode *ri = F2FS_INODE(page); 3227 3228 return (void *)&(ri->i_addr[DEF_ADDRS_PER_INODE - 3229 get_inline_xattr_addrs(inode)]); 3230 } 3231 3232 static inline int inline_xattr_size(struct inode *inode) 3233 { 3234 if (f2fs_has_inline_xattr(inode)) 3235 return get_inline_xattr_addrs(inode) * sizeof(__le32); 3236 return 0; 3237 } 3238 3239 /* 3240 * Notice: check inline_data flag without inode page lock is unsafe. 3241 * It could change at any time by f2fs_convert_inline_page(). 3242 */ 3243 static inline int f2fs_has_inline_data(struct inode *inode) 3244 { 3245 return is_inode_flag_set(inode, FI_INLINE_DATA); 3246 } 3247 3248 static inline int f2fs_exist_data(struct inode *inode) 3249 { 3250 return is_inode_flag_set(inode, FI_DATA_EXIST); 3251 } 3252 3253 static inline int f2fs_has_inline_dots(struct inode *inode) 3254 { 3255 return is_inode_flag_set(inode, FI_INLINE_DOTS); 3256 } 3257 3258 static inline int f2fs_is_mmap_file(struct inode *inode) 3259 { 3260 return is_inode_flag_set(inode, FI_MMAP_FILE); 3261 } 3262 3263 static inline bool f2fs_is_pinned_file(struct inode *inode) 3264 { 3265 return is_inode_flag_set(inode, FI_PIN_FILE); 3266 } 3267 3268 static inline bool f2fs_is_atomic_file(struct inode *inode) 3269 { 3270 return is_inode_flag_set(inode, FI_ATOMIC_FILE); 3271 } 3272 3273 static inline bool f2fs_is_cow_file(struct inode *inode) 3274 { 3275 return is_inode_flag_set(inode, FI_COW_FILE); 3276 } 3277 3278 static inline __le32 *get_dnode_addr(struct inode *inode, 3279 struct page *node_page); 3280 static inline void *inline_data_addr(struct inode *inode, struct page *page) 3281 { 3282 __le32 *addr = get_dnode_addr(inode, page); 3283 3284 return (void *)(addr + DEF_INLINE_RESERVED_SIZE); 3285 } 3286 3287 static inline int f2fs_has_inline_dentry(struct inode *inode) 3288 { 3289 return is_inode_flag_set(inode, FI_INLINE_DENTRY); 3290 } 3291 3292 static inline int is_file(struct inode *inode, int type) 3293 { 3294 return F2FS_I(inode)->i_advise & type; 3295 } 3296 3297 static inline void set_file(struct inode *inode, int type) 3298 { 3299 if (is_file(inode, type)) 3300 return; 3301 F2FS_I(inode)->i_advise |= type; 3302 f2fs_mark_inode_dirty_sync(inode, true); 3303 } 3304 3305 static inline void clear_file(struct inode *inode, int type) 3306 { 3307 if (!is_file(inode, type)) 3308 return; 3309 F2FS_I(inode)->i_advise &= ~type; 3310 f2fs_mark_inode_dirty_sync(inode, true); 3311 } 3312 3313 static inline bool f2fs_is_time_consistent(struct inode *inode) 3314 { 3315 struct timespec64 ctime = inode_get_ctime(inode); 3316 3317 if (!timespec64_equal(F2FS_I(inode)->i_disk_time, &inode->i_atime)) 3318 return false; 3319 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 1, &ctime)) 3320 return false; 3321 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 2, &inode->i_mtime)) 3322 return false; 3323 return true; 3324 } 3325 3326 static inline bool f2fs_skip_inode_update(struct inode *inode, int dsync) 3327 { 3328 bool ret; 3329 3330 if (dsync) { 3331 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 3332 3333 spin_lock(&sbi->inode_lock[DIRTY_META]); 3334 ret = list_empty(&F2FS_I(inode)->gdirty_list); 3335 spin_unlock(&sbi->inode_lock[DIRTY_META]); 3336 return ret; 3337 } 3338 if (!is_inode_flag_set(inode, FI_AUTO_RECOVER) || 3339 file_keep_isize(inode) || 3340 i_size_read(inode) & ~PAGE_MASK) 3341 return false; 3342 3343 if (!f2fs_is_time_consistent(inode)) 3344 return false; 3345 3346 spin_lock(&F2FS_I(inode)->i_size_lock); 3347 ret = F2FS_I(inode)->last_disk_size == i_size_read(inode); 3348 spin_unlock(&F2FS_I(inode)->i_size_lock); 3349 3350 return ret; 3351 } 3352 3353 static inline bool f2fs_readonly(struct super_block *sb) 3354 { 3355 return sb_rdonly(sb); 3356 } 3357 3358 static inline bool f2fs_cp_error(struct f2fs_sb_info *sbi) 3359 { 3360 return is_set_ckpt_flags(sbi, CP_ERROR_FLAG); 3361 } 3362 3363 static inline bool is_dot_dotdot(const u8 *name, size_t len) 3364 { 3365 if (len == 1 && name[0] == '.') 3366 return true; 3367 3368 if (len == 2 && name[0] == '.' && name[1] == '.') 3369 return true; 3370 3371 return false; 3372 } 3373 3374 static inline void *f2fs_kmalloc(struct f2fs_sb_info *sbi, 3375 size_t size, gfp_t flags) 3376 { 3377 if (time_to_inject(sbi, FAULT_KMALLOC)) 3378 return NULL; 3379 3380 return kmalloc(size, flags); 3381 } 3382 3383 static inline void *f2fs_getname(struct f2fs_sb_info *sbi) 3384 { 3385 if (time_to_inject(sbi, FAULT_KMALLOC)) 3386 return NULL; 3387 3388 return __getname(); 3389 } 3390 3391 static inline void f2fs_putname(char *buf) 3392 { 3393 __putname(buf); 3394 } 3395 3396 static inline void *f2fs_kzalloc(struct f2fs_sb_info *sbi, 3397 size_t size, gfp_t flags) 3398 { 3399 return f2fs_kmalloc(sbi, size, flags | __GFP_ZERO); 3400 } 3401 3402 static inline void *f2fs_kvmalloc(struct f2fs_sb_info *sbi, 3403 size_t size, gfp_t flags) 3404 { 3405 if (time_to_inject(sbi, FAULT_KVMALLOC)) 3406 return NULL; 3407 3408 return kvmalloc(size, flags); 3409 } 3410 3411 static inline void *f2fs_kvzalloc(struct f2fs_sb_info *sbi, 3412 size_t size, gfp_t flags) 3413 { 3414 return f2fs_kvmalloc(sbi, size, flags | __GFP_ZERO); 3415 } 3416 3417 static inline int get_extra_isize(struct inode *inode) 3418 { 3419 return F2FS_I(inode)->i_extra_isize / sizeof(__le32); 3420 } 3421 3422 static inline int get_inline_xattr_addrs(struct inode *inode) 3423 { 3424 return F2FS_I(inode)->i_inline_xattr_size; 3425 } 3426 3427 static inline __le32 *get_dnode_addr(struct inode *inode, 3428 struct page *node_page) 3429 { 3430 int base = 0; 3431 3432 if (IS_INODE(node_page) && f2fs_has_extra_attr(inode)) 3433 base = get_extra_isize(inode); 3434 3435 return blkaddr_in_node(F2FS_NODE(node_page)) + base; 3436 } 3437 3438 #define f2fs_get_inode_mode(i) \ 3439 ((is_inode_flag_set(i, FI_ACL_MODE)) ? \ 3440 (F2FS_I(i)->i_acl_mode) : ((i)->i_mode)) 3441 3442 #define F2FS_MIN_EXTRA_ATTR_SIZE (sizeof(__le32)) 3443 3444 #define F2FS_TOTAL_EXTRA_ATTR_SIZE \ 3445 (offsetof(struct f2fs_inode, i_extra_end) - \ 3446 offsetof(struct f2fs_inode, i_extra_isize)) \ 3447 3448 #define F2FS_OLD_ATTRIBUTE_SIZE (offsetof(struct f2fs_inode, i_addr)) 3449 #define F2FS_FITS_IN_INODE(f2fs_inode, extra_isize, field) \ 3450 ((offsetof(typeof(*(f2fs_inode)), field) + \ 3451 sizeof((f2fs_inode)->field)) \ 3452 <= (F2FS_OLD_ATTRIBUTE_SIZE + (extra_isize))) \ 3453 3454 #define __is_large_section(sbi) ((sbi)->segs_per_sec > 1) 3455 3456 #define __is_meta_io(fio) (PAGE_TYPE_OF_BIO((fio)->type) == META) 3457 3458 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi, 3459 block_t blkaddr, int type); 3460 static inline void verify_blkaddr(struct f2fs_sb_info *sbi, 3461 block_t blkaddr, int type) 3462 { 3463 if (!f2fs_is_valid_blkaddr(sbi, blkaddr, type)) { 3464 f2fs_err(sbi, "invalid blkaddr: %u, type: %d, run fsck to fix.", 3465 blkaddr, type); 3466 f2fs_bug_on(sbi, 1); 3467 } 3468 } 3469 3470 static inline bool __is_valid_data_blkaddr(block_t blkaddr) 3471 { 3472 if (blkaddr == NEW_ADDR || blkaddr == NULL_ADDR || 3473 blkaddr == COMPRESS_ADDR) 3474 return false; 3475 return true; 3476 } 3477 3478 /* 3479 * file.c 3480 */ 3481 int f2fs_sync_file(struct file *file, loff_t start, loff_t end, int datasync); 3482 int f2fs_do_truncate_blocks(struct inode *inode, u64 from, bool lock); 3483 int f2fs_truncate_blocks(struct inode *inode, u64 from, bool lock); 3484 int f2fs_truncate(struct inode *inode); 3485 int f2fs_getattr(struct mnt_idmap *idmap, const struct path *path, 3486 struct kstat *stat, u32 request_mask, unsigned int flags); 3487 int f2fs_setattr(struct mnt_idmap *idmap, struct dentry *dentry, 3488 struct iattr *attr); 3489 int f2fs_truncate_hole(struct inode *inode, pgoff_t pg_start, pgoff_t pg_end); 3490 void f2fs_truncate_data_blocks_range(struct dnode_of_data *dn, int count); 3491 int f2fs_precache_extents(struct inode *inode); 3492 int f2fs_fileattr_get(struct dentry *dentry, struct fileattr *fa); 3493 int f2fs_fileattr_set(struct mnt_idmap *idmap, 3494 struct dentry *dentry, struct fileattr *fa); 3495 long f2fs_ioctl(struct file *filp, unsigned int cmd, unsigned long arg); 3496 long f2fs_compat_ioctl(struct file *file, unsigned int cmd, unsigned long arg); 3497 int f2fs_transfer_project_quota(struct inode *inode, kprojid_t kprojid); 3498 int f2fs_pin_file_control(struct inode *inode, bool inc); 3499 3500 /* 3501 * inode.c 3502 */ 3503 void f2fs_set_inode_flags(struct inode *inode); 3504 bool f2fs_inode_chksum_verify(struct f2fs_sb_info *sbi, struct page *page); 3505 void f2fs_inode_chksum_set(struct f2fs_sb_info *sbi, struct page *page); 3506 struct inode *f2fs_iget(struct super_block *sb, unsigned long ino); 3507 struct inode *f2fs_iget_retry(struct super_block *sb, unsigned long ino); 3508 int f2fs_try_to_free_nats(struct f2fs_sb_info *sbi, int nr_shrink); 3509 void f2fs_update_inode(struct inode *inode, struct page *node_page); 3510 void f2fs_update_inode_page(struct inode *inode); 3511 int f2fs_write_inode(struct inode *inode, struct writeback_control *wbc); 3512 void f2fs_evict_inode(struct inode *inode); 3513 void f2fs_handle_failed_inode(struct inode *inode); 3514 3515 /* 3516 * namei.c 3517 */ 3518 int f2fs_update_extension_list(struct f2fs_sb_info *sbi, const char *name, 3519 bool hot, bool set); 3520 struct dentry *f2fs_get_parent(struct dentry *child); 3521 int f2fs_get_tmpfile(struct mnt_idmap *idmap, struct inode *dir, 3522 struct inode **new_inode); 3523 3524 /* 3525 * dir.c 3526 */ 3527 int f2fs_init_casefolded_name(const struct inode *dir, 3528 struct f2fs_filename *fname); 3529 int f2fs_setup_filename(struct inode *dir, const struct qstr *iname, 3530 int lookup, struct f2fs_filename *fname); 3531 int f2fs_prepare_lookup(struct inode *dir, struct dentry *dentry, 3532 struct f2fs_filename *fname); 3533 void f2fs_free_filename(struct f2fs_filename *fname); 3534 struct f2fs_dir_entry *f2fs_find_target_dentry(const struct f2fs_dentry_ptr *d, 3535 const struct f2fs_filename *fname, int *max_slots); 3536 int f2fs_fill_dentries(struct dir_context *ctx, struct f2fs_dentry_ptr *d, 3537 unsigned int start_pos, struct fscrypt_str *fstr); 3538 void f2fs_do_make_empty_dir(struct inode *inode, struct inode *parent, 3539 struct f2fs_dentry_ptr *d); 3540 struct page *f2fs_init_inode_metadata(struct inode *inode, struct inode *dir, 3541 const struct f2fs_filename *fname, struct page *dpage); 3542 void f2fs_update_parent_metadata(struct inode *dir, struct inode *inode, 3543 unsigned int current_depth); 3544 int f2fs_room_for_filename(const void *bitmap, int slots, int max_slots); 3545 void f2fs_drop_nlink(struct inode *dir, struct inode *inode); 3546 struct f2fs_dir_entry *__f2fs_find_entry(struct inode *dir, 3547 const struct f2fs_filename *fname, 3548 struct page **res_page); 3549 struct f2fs_dir_entry *f2fs_find_entry(struct inode *dir, 3550 const struct qstr *child, struct page **res_page); 3551 struct f2fs_dir_entry *f2fs_parent_dir(struct inode *dir, struct page **p); 3552 ino_t f2fs_inode_by_name(struct inode *dir, const struct qstr *qstr, 3553 struct page **page); 3554 void f2fs_set_link(struct inode *dir, struct f2fs_dir_entry *de, 3555 struct page *page, struct inode *inode); 3556 bool f2fs_has_enough_room(struct inode *dir, struct page *ipage, 3557 const struct f2fs_filename *fname); 3558 void f2fs_update_dentry(nid_t ino, umode_t mode, struct f2fs_dentry_ptr *d, 3559 const struct fscrypt_str *name, f2fs_hash_t name_hash, 3560 unsigned int bit_pos); 3561 int f2fs_add_regular_entry(struct inode *dir, const struct f2fs_filename *fname, 3562 struct inode *inode, nid_t ino, umode_t mode); 3563 int f2fs_add_dentry(struct inode *dir, const struct f2fs_filename *fname, 3564 struct inode *inode, nid_t ino, umode_t mode); 3565 int f2fs_do_add_link(struct inode *dir, const struct qstr *name, 3566 struct inode *inode, nid_t ino, umode_t mode); 3567 void f2fs_delete_entry(struct f2fs_dir_entry *dentry, struct page *page, 3568 struct inode *dir, struct inode *inode); 3569 int f2fs_do_tmpfile(struct inode *inode, struct inode *dir); 3570 bool f2fs_empty_dir(struct inode *dir); 3571 3572 static inline int f2fs_add_link(struct dentry *dentry, struct inode *inode) 3573 { 3574 if (fscrypt_is_nokey_name(dentry)) 3575 return -ENOKEY; 3576 return f2fs_do_add_link(d_inode(dentry->d_parent), &dentry->d_name, 3577 inode, inode->i_ino, inode->i_mode); 3578 } 3579 3580 /* 3581 * super.c 3582 */ 3583 int f2fs_inode_dirtied(struct inode *inode, bool sync); 3584 void f2fs_inode_synced(struct inode *inode); 3585 int f2fs_dquot_initialize(struct inode *inode); 3586 int f2fs_enable_quota_files(struct f2fs_sb_info *sbi, bool rdonly); 3587 int f2fs_quota_sync(struct super_block *sb, int type); 3588 loff_t max_file_blocks(struct inode *inode); 3589 void f2fs_quota_off_umount(struct super_block *sb); 3590 void f2fs_save_errors(struct f2fs_sb_info *sbi, unsigned char flag); 3591 void f2fs_handle_critical_error(struct f2fs_sb_info *sbi, unsigned char reason, 3592 bool irq_context); 3593 void f2fs_handle_error(struct f2fs_sb_info *sbi, unsigned char error); 3594 void f2fs_handle_error_async(struct f2fs_sb_info *sbi, unsigned char error); 3595 int f2fs_commit_super(struct f2fs_sb_info *sbi, bool recover); 3596 int f2fs_sync_fs(struct super_block *sb, int sync); 3597 int f2fs_sanity_check_ckpt(struct f2fs_sb_info *sbi); 3598 3599 /* 3600 * hash.c 3601 */ 3602 void f2fs_hash_filename(const struct inode *dir, struct f2fs_filename *fname); 3603 3604 /* 3605 * node.c 3606 */ 3607 struct node_info; 3608 3609 int f2fs_check_nid_range(struct f2fs_sb_info *sbi, nid_t nid); 3610 bool f2fs_available_free_memory(struct f2fs_sb_info *sbi, int type); 3611 bool f2fs_in_warm_node_list(struct f2fs_sb_info *sbi, struct page *page); 3612 void f2fs_init_fsync_node_info(struct f2fs_sb_info *sbi); 3613 void f2fs_del_fsync_node_entry(struct f2fs_sb_info *sbi, struct page *page); 3614 void f2fs_reset_fsync_node_info(struct f2fs_sb_info *sbi); 3615 int f2fs_need_dentry_mark(struct f2fs_sb_info *sbi, nid_t nid); 3616 bool f2fs_is_checkpointed_node(struct f2fs_sb_info *sbi, nid_t nid); 3617 bool f2fs_need_inode_block_update(struct f2fs_sb_info *sbi, nid_t ino); 3618 int f2fs_get_node_info(struct f2fs_sb_info *sbi, nid_t nid, 3619 struct node_info *ni, bool checkpoint_context); 3620 pgoff_t f2fs_get_next_page_offset(struct dnode_of_data *dn, pgoff_t pgofs); 3621 int f2fs_get_dnode_of_data(struct dnode_of_data *dn, pgoff_t index, int mode); 3622 int f2fs_truncate_inode_blocks(struct inode *inode, pgoff_t from); 3623 int f2fs_truncate_xattr_node(struct inode *inode); 3624 int f2fs_wait_on_node_pages_writeback(struct f2fs_sb_info *sbi, 3625 unsigned int seq_id); 3626 bool f2fs_nat_bitmap_enabled(struct f2fs_sb_info *sbi); 3627 int f2fs_remove_inode_page(struct inode *inode); 3628 struct page *f2fs_new_inode_page(struct inode *inode); 3629 struct page *f2fs_new_node_page(struct dnode_of_data *dn, unsigned int ofs); 3630 void f2fs_ra_node_page(struct f2fs_sb_info *sbi, nid_t nid); 3631 struct page *f2fs_get_node_page(struct f2fs_sb_info *sbi, pgoff_t nid); 3632 struct page *f2fs_get_node_page_ra(struct page *parent, int start); 3633 int f2fs_move_node_page(struct page *node_page, int gc_type); 3634 void f2fs_flush_inline_data(struct f2fs_sb_info *sbi); 3635 int f2fs_fsync_node_pages(struct f2fs_sb_info *sbi, struct inode *inode, 3636 struct writeback_control *wbc, bool atomic, 3637 unsigned int *seq_id); 3638 int f2fs_sync_node_pages(struct f2fs_sb_info *sbi, 3639 struct writeback_control *wbc, 3640 bool do_balance, enum iostat_type io_type); 3641 int f2fs_build_free_nids(struct f2fs_sb_info *sbi, bool sync, bool mount); 3642 bool f2fs_alloc_nid(struct f2fs_sb_info *sbi, nid_t *nid); 3643 void f2fs_alloc_nid_done(struct f2fs_sb_info *sbi, nid_t nid); 3644 void f2fs_alloc_nid_failed(struct f2fs_sb_info *sbi, nid_t nid); 3645 int f2fs_try_to_free_nids(struct f2fs_sb_info *sbi, int nr_shrink); 3646 int f2fs_recover_inline_xattr(struct inode *inode, struct page *page); 3647 int f2fs_recover_xattr_data(struct inode *inode, struct page *page); 3648 int f2fs_recover_inode_page(struct f2fs_sb_info *sbi, struct page *page); 3649 int f2fs_restore_node_summary(struct f2fs_sb_info *sbi, 3650 unsigned int segno, struct f2fs_summary_block *sum); 3651 void f2fs_enable_nat_bits(struct f2fs_sb_info *sbi); 3652 int f2fs_flush_nat_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3653 int f2fs_build_node_manager(struct f2fs_sb_info *sbi); 3654 void f2fs_destroy_node_manager(struct f2fs_sb_info *sbi); 3655 int __init f2fs_create_node_manager_caches(void); 3656 void f2fs_destroy_node_manager_caches(void); 3657 3658 /* 3659 * segment.c 3660 */ 3661 bool f2fs_need_SSR(struct f2fs_sb_info *sbi); 3662 int f2fs_commit_atomic_write(struct inode *inode); 3663 void f2fs_abort_atomic_write(struct inode *inode, bool clean); 3664 void f2fs_balance_fs(struct f2fs_sb_info *sbi, bool need); 3665 void f2fs_balance_fs_bg(struct f2fs_sb_info *sbi, bool from_bg); 3666 int f2fs_issue_flush(struct f2fs_sb_info *sbi, nid_t ino); 3667 int f2fs_create_flush_cmd_control(struct f2fs_sb_info *sbi); 3668 int f2fs_flush_device_cache(struct f2fs_sb_info *sbi); 3669 void f2fs_destroy_flush_cmd_control(struct f2fs_sb_info *sbi, bool free); 3670 void f2fs_invalidate_blocks(struct f2fs_sb_info *sbi, block_t addr); 3671 bool f2fs_is_checkpointed_data(struct f2fs_sb_info *sbi, block_t blkaddr); 3672 int f2fs_start_discard_thread(struct f2fs_sb_info *sbi); 3673 void f2fs_drop_discard_cmd(struct f2fs_sb_info *sbi); 3674 void f2fs_stop_discard_thread(struct f2fs_sb_info *sbi); 3675 bool f2fs_issue_discard_timeout(struct f2fs_sb_info *sbi); 3676 void f2fs_clear_prefree_segments(struct f2fs_sb_info *sbi, 3677 struct cp_control *cpc); 3678 void f2fs_dirty_to_prefree(struct f2fs_sb_info *sbi); 3679 block_t f2fs_get_unusable_blocks(struct f2fs_sb_info *sbi); 3680 int f2fs_disable_cp_again(struct f2fs_sb_info *sbi, block_t unusable); 3681 void f2fs_release_discard_addrs(struct f2fs_sb_info *sbi); 3682 int f2fs_npages_for_summary_flush(struct f2fs_sb_info *sbi, bool for_ra); 3683 bool f2fs_segment_has_free_slot(struct f2fs_sb_info *sbi, int segno); 3684 void f2fs_init_inmem_curseg(struct f2fs_sb_info *sbi); 3685 void f2fs_save_inmem_curseg(struct f2fs_sb_info *sbi); 3686 void f2fs_restore_inmem_curseg(struct f2fs_sb_info *sbi); 3687 void f2fs_get_new_segment(struct f2fs_sb_info *sbi, 3688 unsigned int *newseg, bool new_sec, int dir); 3689 void f2fs_allocate_segment_for_resize(struct f2fs_sb_info *sbi, int type, 3690 unsigned int start, unsigned int end); 3691 void f2fs_allocate_new_section(struct f2fs_sb_info *sbi, int type, bool force); 3692 void f2fs_allocate_new_segments(struct f2fs_sb_info *sbi); 3693 int f2fs_trim_fs(struct f2fs_sb_info *sbi, struct fstrim_range *range); 3694 bool f2fs_exist_trim_candidates(struct f2fs_sb_info *sbi, 3695 struct cp_control *cpc); 3696 struct page *f2fs_get_sum_page(struct f2fs_sb_info *sbi, unsigned int segno); 3697 void f2fs_update_meta_page(struct f2fs_sb_info *sbi, void *src, 3698 block_t blk_addr); 3699 void f2fs_do_write_meta_page(struct f2fs_sb_info *sbi, struct page *page, 3700 enum iostat_type io_type); 3701 void f2fs_do_write_node_page(unsigned int nid, struct f2fs_io_info *fio); 3702 void f2fs_outplace_write_data(struct dnode_of_data *dn, 3703 struct f2fs_io_info *fio); 3704 int f2fs_inplace_write_data(struct f2fs_io_info *fio); 3705 void f2fs_do_replace_block(struct f2fs_sb_info *sbi, struct f2fs_summary *sum, 3706 block_t old_blkaddr, block_t new_blkaddr, 3707 bool recover_curseg, bool recover_newaddr, 3708 bool from_gc); 3709 void f2fs_replace_block(struct f2fs_sb_info *sbi, struct dnode_of_data *dn, 3710 block_t old_addr, block_t new_addr, 3711 unsigned char version, bool recover_curseg, 3712 bool recover_newaddr); 3713 void f2fs_allocate_data_block(struct f2fs_sb_info *sbi, struct page *page, 3714 block_t old_blkaddr, block_t *new_blkaddr, 3715 struct f2fs_summary *sum, int type, 3716 struct f2fs_io_info *fio); 3717 void f2fs_update_device_state(struct f2fs_sb_info *sbi, nid_t ino, 3718 block_t blkaddr, unsigned int blkcnt); 3719 void f2fs_wait_on_page_writeback(struct page *page, 3720 enum page_type type, bool ordered, bool locked); 3721 void f2fs_wait_on_block_writeback(struct inode *inode, block_t blkaddr); 3722 void f2fs_wait_on_block_writeback_range(struct inode *inode, block_t blkaddr, 3723 block_t len); 3724 void f2fs_write_data_summaries(struct f2fs_sb_info *sbi, block_t start_blk); 3725 void f2fs_write_node_summaries(struct f2fs_sb_info *sbi, block_t start_blk); 3726 int f2fs_lookup_journal_in_cursum(struct f2fs_journal *journal, int type, 3727 unsigned int val, int alloc); 3728 void f2fs_flush_sit_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3729 int f2fs_fix_curseg_write_pointer(struct f2fs_sb_info *sbi); 3730 int f2fs_check_write_pointer(struct f2fs_sb_info *sbi); 3731 int f2fs_build_segment_manager(struct f2fs_sb_info *sbi); 3732 void f2fs_destroy_segment_manager(struct f2fs_sb_info *sbi); 3733 int __init f2fs_create_segment_manager_caches(void); 3734 void f2fs_destroy_segment_manager_caches(void); 3735 int f2fs_rw_hint_to_seg_type(enum rw_hint hint); 3736 unsigned int f2fs_usable_segs_in_sec(struct f2fs_sb_info *sbi, 3737 unsigned int segno); 3738 unsigned int f2fs_usable_blks_in_seg(struct f2fs_sb_info *sbi, 3739 unsigned int segno); 3740 3741 #define DEF_FRAGMENT_SIZE 4 3742 #define MIN_FRAGMENT_SIZE 1 3743 #define MAX_FRAGMENT_SIZE 512 3744 3745 static inline bool f2fs_need_rand_seg(struct f2fs_sb_info *sbi) 3746 { 3747 return F2FS_OPTION(sbi).fs_mode == FS_MODE_FRAGMENT_SEG || 3748 F2FS_OPTION(sbi).fs_mode == FS_MODE_FRAGMENT_BLK; 3749 } 3750 3751 /* 3752 * checkpoint.c 3753 */ 3754 void f2fs_stop_checkpoint(struct f2fs_sb_info *sbi, bool end_io, 3755 unsigned char reason); 3756 void f2fs_flush_ckpt_thread(struct f2fs_sb_info *sbi); 3757 struct page *f2fs_grab_meta_page(struct f2fs_sb_info *sbi, pgoff_t index); 3758 struct page *f2fs_get_meta_page(struct f2fs_sb_info *sbi, pgoff_t index); 3759 struct page *f2fs_get_meta_page_retry(struct f2fs_sb_info *sbi, pgoff_t index); 3760 struct page *f2fs_get_tmp_page(struct f2fs_sb_info *sbi, pgoff_t index); 3761 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi, 3762 block_t blkaddr, int type); 3763 int f2fs_ra_meta_pages(struct f2fs_sb_info *sbi, block_t start, int nrpages, 3764 int type, bool sync); 3765 void f2fs_ra_meta_pages_cond(struct f2fs_sb_info *sbi, pgoff_t index, 3766 unsigned int ra_blocks); 3767 long f2fs_sync_meta_pages(struct f2fs_sb_info *sbi, enum page_type type, 3768 long nr_to_write, enum iostat_type io_type); 3769 void f2fs_add_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type); 3770 void f2fs_remove_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type); 3771 void f2fs_release_ino_entry(struct f2fs_sb_info *sbi, bool all); 3772 bool f2fs_exist_written_data(struct f2fs_sb_info *sbi, nid_t ino, int mode); 3773 void f2fs_set_dirty_device(struct f2fs_sb_info *sbi, nid_t ino, 3774 unsigned int devidx, int type); 3775 bool f2fs_is_dirty_device(struct f2fs_sb_info *sbi, nid_t ino, 3776 unsigned int devidx, int type); 3777 int f2fs_acquire_orphan_inode(struct f2fs_sb_info *sbi); 3778 void f2fs_release_orphan_inode(struct f2fs_sb_info *sbi); 3779 void f2fs_add_orphan_inode(struct inode *inode); 3780 void f2fs_remove_orphan_inode(struct f2fs_sb_info *sbi, nid_t ino); 3781 int f2fs_recover_orphan_inodes(struct f2fs_sb_info *sbi); 3782 int f2fs_get_valid_checkpoint(struct f2fs_sb_info *sbi); 3783 void f2fs_update_dirty_folio(struct inode *inode, struct folio *folio); 3784 void f2fs_remove_dirty_inode(struct inode *inode); 3785 int f2fs_sync_dirty_inodes(struct f2fs_sb_info *sbi, enum inode_type type, 3786 bool from_cp); 3787 void f2fs_wait_on_all_pages(struct f2fs_sb_info *sbi, int type); 3788 u64 f2fs_get_sectors_written(struct f2fs_sb_info *sbi); 3789 int f2fs_write_checkpoint(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3790 void f2fs_init_ino_entry_info(struct f2fs_sb_info *sbi); 3791 int __init f2fs_create_checkpoint_caches(void); 3792 void f2fs_destroy_checkpoint_caches(void); 3793 int f2fs_issue_checkpoint(struct f2fs_sb_info *sbi); 3794 int f2fs_start_ckpt_thread(struct f2fs_sb_info *sbi); 3795 void f2fs_stop_ckpt_thread(struct f2fs_sb_info *sbi); 3796 void f2fs_init_ckpt_req_control(struct f2fs_sb_info *sbi); 3797 3798 /* 3799 * data.c 3800 */ 3801 int __init f2fs_init_bioset(void); 3802 void f2fs_destroy_bioset(void); 3803 bool f2fs_is_cp_guaranteed(struct page *page); 3804 int f2fs_init_bio_entry_cache(void); 3805 void f2fs_destroy_bio_entry_cache(void); 3806 void f2fs_submit_read_bio(struct f2fs_sb_info *sbi, struct bio *bio, 3807 enum page_type type); 3808 int f2fs_init_write_merge_io(struct f2fs_sb_info *sbi); 3809 void f2fs_submit_merged_write(struct f2fs_sb_info *sbi, enum page_type type); 3810 void f2fs_submit_merged_write_cond(struct f2fs_sb_info *sbi, 3811 struct inode *inode, struct page *page, 3812 nid_t ino, enum page_type type); 3813 void f2fs_submit_merged_ipu_write(struct f2fs_sb_info *sbi, 3814 struct bio **bio, struct page *page); 3815 void f2fs_flush_merged_writes(struct f2fs_sb_info *sbi); 3816 int f2fs_submit_page_bio(struct f2fs_io_info *fio); 3817 int f2fs_merge_page_bio(struct f2fs_io_info *fio); 3818 void f2fs_submit_page_write(struct f2fs_io_info *fio); 3819 struct block_device *f2fs_target_device(struct f2fs_sb_info *sbi, 3820 block_t blk_addr, sector_t *sector); 3821 int f2fs_target_device_index(struct f2fs_sb_info *sbi, block_t blkaddr); 3822 void f2fs_set_data_blkaddr(struct dnode_of_data *dn, block_t blkaddr); 3823 void f2fs_update_data_blkaddr(struct dnode_of_data *dn, block_t blkaddr); 3824 int f2fs_reserve_new_blocks(struct dnode_of_data *dn, blkcnt_t count); 3825 int f2fs_reserve_new_block(struct dnode_of_data *dn); 3826 int f2fs_get_block_locked(struct dnode_of_data *dn, pgoff_t index); 3827 int f2fs_reserve_block(struct dnode_of_data *dn, pgoff_t index); 3828 struct page *f2fs_get_read_data_page(struct inode *inode, pgoff_t index, 3829 blk_opf_t op_flags, bool for_write, pgoff_t *next_pgofs); 3830 struct page *f2fs_find_data_page(struct inode *inode, pgoff_t index, 3831 pgoff_t *next_pgofs); 3832 struct page *f2fs_get_lock_data_page(struct inode *inode, pgoff_t index, 3833 bool for_write); 3834 struct page *f2fs_get_new_data_page(struct inode *inode, 3835 struct page *ipage, pgoff_t index, bool new_i_size); 3836 int f2fs_do_write_data_page(struct f2fs_io_info *fio); 3837 int f2fs_map_blocks(struct inode *inode, struct f2fs_map_blocks *map, int flag); 3838 int f2fs_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, 3839 u64 start, u64 len); 3840 int f2fs_encrypt_one_page(struct f2fs_io_info *fio); 3841 bool f2fs_should_update_inplace(struct inode *inode, struct f2fs_io_info *fio); 3842 bool f2fs_should_update_outplace(struct inode *inode, struct f2fs_io_info *fio); 3843 int f2fs_write_single_data_page(struct page *page, int *submitted, 3844 struct bio **bio, sector_t *last_block, 3845 struct writeback_control *wbc, 3846 enum iostat_type io_type, 3847 int compr_blocks, bool allow_balance); 3848 void f2fs_write_failed(struct inode *inode, loff_t to); 3849 void f2fs_invalidate_folio(struct folio *folio, size_t offset, size_t length); 3850 bool f2fs_release_folio(struct folio *folio, gfp_t wait); 3851 bool f2fs_overwrite_io(struct inode *inode, loff_t pos, size_t len); 3852 void f2fs_clear_page_cache_dirty_tag(struct page *page); 3853 int f2fs_init_post_read_processing(void); 3854 void f2fs_destroy_post_read_processing(void); 3855 int f2fs_init_post_read_wq(struct f2fs_sb_info *sbi); 3856 void f2fs_destroy_post_read_wq(struct f2fs_sb_info *sbi); 3857 extern const struct iomap_ops f2fs_iomap_ops; 3858 3859 /* 3860 * gc.c 3861 */ 3862 int f2fs_start_gc_thread(struct f2fs_sb_info *sbi); 3863 void f2fs_stop_gc_thread(struct f2fs_sb_info *sbi); 3864 block_t f2fs_start_bidx_of_node(unsigned int node_ofs, struct inode *inode); 3865 int f2fs_gc(struct f2fs_sb_info *sbi, struct f2fs_gc_control *gc_control); 3866 void f2fs_build_gc_manager(struct f2fs_sb_info *sbi); 3867 int f2fs_resize_fs(struct file *filp, __u64 block_count); 3868 int __init f2fs_create_garbage_collection_cache(void); 3869 void f2fs_destroy_garbage_collection_cache(void); 3870 /* victim selection function for cleaning and SSR */ 3871 int f2fs_get_victim(struct f2fs_sb_info *sbi, unsigned int *result, 3872 int gc_type, int type, char alloc_mode, 3873 unsigned long long age); 3874 3875 /* 3876 * recovery.c 3877 */ 3878 int f2fs_recover_fsync_data(struct f2fs_sb_info *sbi, bool check_only); 3879 bool f2fs_space_for_roll_forward(struct f2fs_sb_info *sbi); 3880 int __init f2fs_create_recovery_cache(void); 3881 void f2fs_destroy_recovery_cache(void); 3882 3883 /* 3884 * debug.c 3885 */ 3886 #ifdef CONFIG_F2FS_STAT_FS 3887 struct f2fs_stat_info { 3888 struct list_head stat_list; 3889 struct f2fs_sb_info *sbi; 3890 int all_area_segs, sit_area_segs, nat_area_segs, ssa_area_segs; 3891 int main_area_segs, main_area_sections, main_area_zones; 3892 unsigned long long hit_cached[NR_EXTENT_CACHES]; 3893 unsigned long long hit_rbtree[NR_EXTENT_CACHES]; 3894 unsigned long long total_ext[NR_EXTENT_CACHES]; 3895 unsigned long long hit_total[NR_EXTENT_CACHES]; 3896 int ext_tree[NR_EXTENT_CACHES]; 3897 int zombie_tree[NR_EXTENT_CACHES]; 3898 int ext_node[NR_EXTENT_CACHES]; 3899 /* to count memory footprint */ 3900 unsigned long long ext_mem[NR_EXTENT_CACHES]; 3901 /* for read extent cache */ 3902 unsigned long long hit_largest; 3903 /* for block age extent cache */ 3904 unsigned long long allocated_data_blocks; 3905 int ndirty_node, ndirty_dent, ndirty_meta, ndirty_imeta; 3906 int ndirty_data, ndirty_qdata; 3907 unsigned int ndirty_dirs, ndirty_files, nquota_files, ndirty_all; 3908 int nats, dirty_nats, sits, dirty_sits; 3909 int free_nids, avail_nids, alloc_nids; 3910 int total_count, utilization; 3911 int nr_wb_cp_data, nr_wb_data; 3912 int nr_rd_data, nr_rd_node, nr_rd_meta; 3913 int nr_dio_read, nr_dio_write; 3914 unsigned int io_skip_bggc, other_skip_bggc; 3915 int nr_flushing, nr_flushed, flush_list_empty; 3916 int nr_discarding, nr_discarded; 3917 int nr_discard_cmd; 3918 unsigned int undiscard_blks; 3919 int nr_issued_ckpt, nr_total_ckpt, nr_queued_ckpt; 3920 unsigned int cur_ckpt_time, peak_ckpt_time; 3921 int inline_xattr, inline_inode, inline_dir, append, update, orphans; 3922 int compr_inode, swapfile_inode; 3923 unsigned long long compr_blocks; 3924 int aw_cnt, max_aw_cnt; 3925 unsigned int valid_count, valid_node_count, valid_inode_count, discard_blks; 3926 unsigned int bimodal, avg_vblocks; 3927 int util_free, util_valid, util_invalid; 3928 int rsvd_segs, overp_segs; 3929 int dirty_count, node_pages, meta_pages, compress_pages; 3930 int compress_page_hit; 3931 int prefree_count, free_segs, free_secs; 3932 int cp_call_count[MAX_CALL_TYPE], cp_count; 3933 int gc_call_count[MAX_CALL_TYPE]; 3934 int gc_segs[2][2]; 3935 int gc_secs[2][2]; 3936 int tot_blks, data_blks, node_blks; 3937 int bg_data_blks, bg_node_blks; 3938 int curseg[NR_CURSEG_TYPE]; 3939 int cursec[NR_CURSEG_TYPE]; 3940 int curzone[NR_CURSEG_TYPE]; 3941 unsigned int dirty_seg[NR_CURSEG_TYPE]; 3942 unsigned int full_seg[NR_CURSEG_TYPE]; 3943 unsigned int valid_blks[NR_CURSEG_TYPE]; 3944 3945 unsigned int meta_count[META_MAX]; 3946 unsigned int segment_count[2]; 3947 unsigned int block_count[2]; 3948 unsigned int inplace_count; 3949 unsigned long long base_mem, cache_mem, page_mem; 3950 }; 3951 3952 static inline struct f2fs_stat_info *F2FS_STAT(struct f2fs_sb_info *sbi) 3953 { 3954 return (struct f2fs_stat_info *)sbi->stat_info; 3955 } 3956 3957 #define stat_inc_cp_call_count(sbi, foreground) \ 3958 atomic_inc(&sbi->cp_call_count[(foreground)]) 3959 #define stat_inc_cp_count(si) (F2FS_STAT(sbi)->cp_count++) 3960 #define stat_io_skip_bggc_count(sbi) ((sbi)->io_skip_bggc++) 3961 #define stat_other_skip_bggc_count(sbi) ((sbi)->other_skip_bggc++) 3962 #define stat_inc_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]++) 3963 #define stat_dec_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]--) 3964 #define stat_inc_total_hit(sbi, type) (atomic64_inc(&(sbi)->total_hit_ext[type])) 3965 #define stat_inc_rbtree_node_hit(sbi, type) (atomic64_inc(&(sbi)->read_hit_rbtree[type])) 3966 #define stat_inc_largest_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_largest)) 3967 #define stat_inc_cached_node_hit(sbi, type) (atomic64_inc(&(sbi)->read_hit_cached[type])) 3968 #define stat_inc_inline_xattr(inode) \ 3969 do { \ 3970 if (f2fs_has_inline_xattr(inode)) \ 3971 (atomic_inc(&F2FS_I_SB(inode)->inline_xattr)); \ 3972 } while (0) 3973 #define stat_dec_inline_xattr(inode) \ 3974 do { \ 3975 if (f2fs_has_inline_xattr(inode)) \ 3976 (atomic_dec(&F2FS_I_SB(inode)->inline_xattr)); \ 3977 } while (0) 3978 #define stat_inc_inline_inode(inode) \ 3979 do { \ 3980 if (f2fs_has_inline_data(inode)) \ 3981 (atomic_inc(&F2FS_I_SB(inode)->inline_inode)); \ 3982 } while (0) 3983 #define stat_dec_inline_inode(inode) \ 3984 do { \ 3985 if (f2fs_has_inline_data(inode)) \ 3986 (atomic_dec(&F2FS_I_SB(inode)->inline_inode)); \ 3987 } while (0) 3988 #define stat_inc_inline_dir(inode) \ 3989 do { \ 3990 if (f2fs_has_inline_dentry(inode)) \ 3991 (atomic_inc(&F2FS_I_SB(inode)->inline_dir)); \ 3992 } while (0) 3993 #define stat_dec_inline_dir(inode) \ 3994 do { \ 3995 if (f2fs_has_inline_dentry(inode)) \ 3996 (atomic_dec(&F2FS_I_SB(inode)->inline_dir)); \ 3997 } while (0) 3998 #define stat_inc_compr_inode(inode) \ 3999 do { \ 4000 if (f2fs_compressed_file(inode)) \ 4001 (atomic_inc(&F2FS_I_SB(inode)->compr_inode)); \ 4002 } while (0) 4003 #define stat_dec_compr_inode(inode) \ 4004 do { \ 4005 if (f2fs_compressed_file(inode)) \ 4006 (atomic_dec(&F2FS_I_SB(inode)->compr_inode)); \ 4007 } while (0) 4008 #define stat_add_compr_blocks(inode, blocks) \ 4009 (atomic64_add(blocks, &F2FS_I_SB(inode)->compr_blocks)) 4010 #define stat_sub_compr_blocks(inode, blocks) \ 4011 (atomic64_sub(blocks, &F2FS_I_SB(inode)->compr_blocks)) 4012 #define stat_inc_swapfile_inode(inode) \ 4013 (atomic_inc(&F2FS_I_SB(inode)->swapfile_inode)) 4014 #define stat_dec_swapfile_inode(inode) \ 4015 (atomic_dec(&F2FS_I_SB(inode)->swapfile_inode)) 4016 #define stat_inc_atomic_inode(inode) \ 4017 (atomic_inc(&F2FS_I_SB(inode)->atomic_files)) 4018 #define stat_dec_atomic_inode(inode) \ 4019 (atomic_dec(&F2FS_I_SB(inode)->atomic_files)) 4020 #define stat_inc_meta_count(sbi, blkaddr) \ 4021 do { \ 4022 if (blkaddr < SIT_I(sbi)->sit_base_addr) \ 4023 atomic_inc(&(sbi)->meta_count[META_CP]); \ 4024 else if (blkaddr < NM_I(sbi)->nat_blkaddr) \ 4025 atomic_inc(&(sbi)->meta_count[META_SIT]); \ 4026 else if (blkaddr < SM_I(sbi)->ssa_blkaddr) \ 4027 atomic_inc(&(sbi)->meta_count[META_NAT]); \ 4028 else if (blkaddr < SM_I(sbi)->main_blkaddr) \ 4029 atomic_inc(&(sbi)->meta_count[META_SSA]); \ 4030 } while (0) 4031 #define stat_inc_seg_type(sbi, curseg) \ 4032 ((sbi)->segment_count[(curseg)->alloc_type]++) 4033 #define stat_inc_block_count(sbi, curseg) \ 4034 ((sbi)->block_count[(curseg)->alloc_type]++) 4035 #define stat_inc_inplace_blocks(sbi) \ 4036 (atomic_inc(&(sbi)->inplace_count)) 4037 #define stat_update_max_atomic_write(inode) \ 4038 do { \ 4039 int cur = atomic_read(&F2FS_I_SB(inode)->atomic_files); \ 4040 int max = atomic_read(&F2FS_I_SB(inode)->max_aw_cnt); \ 4041 if (cur > max) \ 4042 atomic_set(&F2FS_I_SB(inode)->max_aw_cnt, cur); \ 4043 } while (0) 4044 #define stat_inc_gc_call_count(sbi, foreground) \ 4045 (F2FS_STAT(sbi)->gc_call_count[(foreground)]++) 4046 #define stat_inc_gc_sec_count(sbi, type, gc_type) \ 4047 (F2FS_STAT(sbi)->gc_secs[(type)][(gc_type)]++) 4048 #define stat_inc_gc_seg_count(sbi, type, gc_type) \ 4049 (F2FS_STAT(sbi)->gc_segs[(type)][(gc_type)]++) 4050 4051 #define stat_inc_tot_blk_count(si, blks) \ 4052 ((si)->tot_blks += (blks)) 4053 4054 #define stat_inc_data_blk_count(sbi, blks, gc_type) \ 4055 do { \ 4056 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 4057 stat_inc_tot_blk_count(si, blks); \ 4058 si->data_blks += (blks); \ 4059 si->bg_data_blks += ((gc_type) == BG_GC) ? (blks) : 0; \ 4060 } while (0) 4061 4062 #define stat_inc_node_blk_count(sbi, blks, gc_type) \ 4063 do { \ 4064 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 4065 stat_inc_tot_blk_count(si, blks); \ 4066 si->node_blks += (blks); \ 4067 si->bg_node_blks += ((gc_type) == BG_GC) ? (blks) : 0; \ 4068 } while (0) 4069 4070 int f2fs_build_stats(struct f2fs_sb_info *sbi); 4071 void f2fs_destroy_stats(struct f2fs_sb_info *sbi); 4072 void __init f2fs_create_root_stats(void); 4073 void f2fs_destroy_root_stats(void); 4074 void f2fs_update_sit_info(struct f2fs_sb_info *sbi); 4075 #else 4076 #define stat_inc_cp_call_count(sbi, foreground) do { } while (0) 4077 #define stat_inc_cp_count(sbi) do { } while (0) 4078 #define stat_io_skip_bggc_count(sbi) do { } while (0) 4079 #define stat_other_skip_bggc_count(sbi) do { } while (0) 4080 #define stat_inc_dirty_inode(sbi, type) do { } while (0) 4081 #define stat_dec_dirty_inode(sbi, type) do { } while (0) 4082 #define stat_inc_total_hit(sbi, type) do { } while (0) 4083 #define stat_inc_rbtree_node_hit(sbi, type) do { } while (0) 4084 #define stat_inc_largest_node_hit(sbi) do { } while (0) 4085 #define stat_inc_cached_node_hit(sbi, type) do { } while (0) 4086 #define stat_inc_inline_xattr(inode) do { } while (0) 4087 #define stat_dec_inline_xattr(inode) do { } while (0) 4088 #define stat_inc_inline_inode(inode) do { } while (0) 4089 #define stat_dec_inline_inode(inode) do { } while (0) 4090 #define stat_inc_inline_dir(inode) do { } while (0) 4091 #define stat_dec_inline_dir(inode) do { } while (0) 4092 #define stat_inc_compr_inode(inode) do { } while (0) 4093 #define stat_dec_compr_inode(inode) do { } while (0) 4094 #define stat_add_compr_blocks(inode, blocks) do { } while (0) 4095 #define stat_sub_compr_blocks(inode, blocks) do { } while (0) 4096 #define stat_inc_swapfile_inode(inode) do { } while (0) 4097 #define stat_dec_swapfile_inode(inode) do { } while (0) 4098 #define stat_inc_atomic_inode(inode) do { } while (0) 4099 #define stat_dec_atomic_inode(inode) do { } while (0) 4100 #define stat_update_max_atomic_write(inode) do { } while (0) 4101 #define stat_inc_meta_count(sbi, blkaddr) do { } while (0) 4102 #define stat_inc_seg_type(sbi, curseg) do { } while (0) 4103 #define stat_inc_block_count(sbi, curseg) do { } while (0) 4104 #define stat_inc_inplace_blocks(sbi) do { } while (0) 4105 #define stat_inc_gc_call_count(sbi, foreground) do { } while (0) 4106 #define stat_inc_gc_sec_count(sbi, type, gc_type) do { } while (0) 4107 #define stat_inc_gc_seg_count(sbi, type, gc_type) do { } while (0) 4108 #define stat_inc_tot_blk_count(si, blks) do { } while (0) 4109 #define stat_inc_data_blk_count(sbi, blks, gc_type) do { } while (0) 4110 #define stat_inc_node_blk_count(sbi, blks, gc_type) do { } while (0) 4111 4112 static inline int f2fs_build_stats(struct f2fs_sb_info *sbi) { return 0; } 4113 static inline void f2fs_destroy_stats(struct f2fs_sb_info *sbi) { } 4114 static inline void __init f2fs_create_root_stats(void) { } 4115 static inline void f2fs_destroy_root_stats(void) { } 4116 static inline void f2fs_update_sit_info(struct f2fs_sb_info *sbi) {} 4117 #endif 4118 4119 extern const struct file_operations f2fs_dir_operations; 4120 extern const struct file_operations f2fs_file_operations; 4121 extern const struct inode_operations f2fs_file_inode_operations; 4122 extern const struct address_space_operations f2fs_dblock_aops; 4123 extern const struct address_space_operations f2fs_node_aops; 4124 extern const struct address_space_operations f2fs_meta_aops; 4125 extern const struct inode_operations f2fs_dir_inode_operations; 4126 extern const struct inode_operations f2fs_symlink_inode_operations; 4127 extern const struct inode_operations f2fs_encrypted_symlink_inode_operations; 4128 extern const struct inode_operations f2fs_special_inode_operations; 4129 extern struct kmem_cache *f2fs_inode_entry_slab; 4130 4131 /* 4132 * inline.c 4133 */ 4134 bool f2fs_may_inline_data(struct inode *inode); 4135 bool f2fs_sanity_check_inline_data(struct inode *inode); 4136 bool f2fs_may_inline_dentry(struct inode *inode); 4137 void f2fs_do_read_inline_data(struct page *page, struct page *ipage); 4138 void f2fs_truncate_inline_inode(struct inode *inode, 4139 struct page *ipage, u64 from); 4140 int f2fs_read_inline_data(struct inode *inode, struct page *page); 4141 int f2fs_convert_inline_page(struct dnode_of_data *dn, struct page *page); 4142 int f2fs_convert_inline_inode(struct inode *inode); 4143 int f2fs_try_convert_inline_dir(struct inode *dir, struct dentry *dentry); 4144 int f2fs_write_inline_data(struct inode *inode, struct page *page); 4145 int f2fs_recover_inline_data(struct inode *inode, struct page *npage); 4146 struct f2fs_dir_entry *f2fs_find_in_inline_dir(struct inode *dir, 4147 const struct f2fs_filename *fname, 4148 struct page **res_page); 4149 int f2fs_make_empty_inline_dir(struct inode *inode, struct inode *parent, 4150 struct page *ipage); 4151 int f2fs_add_inline_entry(struct inode *dir, const struct f2fs_filename *fname, 4152 struct inode *inode, nid_t ino, umode_t mode); 4153 void f2fs_delete_inline_entry(struct f2fs_dir_entry *dentry, 4154 struct page *page, struct inode *dir, 4155 struct inode *inode); 4156 bool f2fs_empty_inline_dir(struct inode *dir); 4157 int f2fs_read_inline_dir(struct file *file, struct dir_context *ctx, 4158 struct fscrypt_str *fstr); 4159 int f2fs_inline_data_fiemap(struct inode *inode, 4160 struct fiemap_extent_info *fieinfo, 4161 __u64 start, __u64 len); 4162 4163 /* 4164 * shrinker.c 4165 */ 4166 unsigned long f2fs_shrink_count(struct shrinker *shrink, 4167 struct shrink_control *sc); 4168 unsigned long f2fs_shrink_scan(struct shrinker *shrink, 4169 struct shrink_control *sc); 4170 void f2fs_join_shrinker(struct f2fs_sb_info *sbi); 4171 void f2fs_leave_shrinker(struct f2fs_sb_info *sbi); 4172 4173 /* 4174 * extent_cache.c 4175 */ 4176 bool sanity_check_extent_cache(struct inode *inode); 4177 void f2fs_init_extent_tree(struct inode *inode); 4178 void f2fs_drop_extent_tree(struct inode *inode); 4179 void f2fs_destroy_extent_node(struct inode *inode); 4180 void f2fs_destroy_extent_tree(struct inode *inode); 4181 void f2fs_init_extent_cache_info(struct f2fs_sb_info *sbi); 4182 int __init f2fs_create_extent_cache(void); 4183 void f2fs_destroy_extent_cache(void); 4184 4185 /* read extent cache ops */ 4186 void f2fs_init_read_extent_tree(struct inode *inode, struct page *ipage); 4187 bool f2fs_lookup_read_extent_cache(struct inode *inode, pgoff_t pgofs, 4188 struct extent_info *ei); 4189 bool f2fs_lookup_read_extent_cache_block(struct inode *inode, pgoff_t index, 4190 block_t *blkaddr); 4191 void f2fs_update_read_extent_cache(struct dnode_of_data *dn); 4192 void f2fs_update_read_extent_cache_range(struct dnode_of_data *dn, 4193 pgoff_t fofs, block_t blkaddr, unsigned int len); 4194 unsigned int f2fs_shrink_read_extent_tree(struct f2fs_sb_info *sbi, 4195 int nr_shrink); 4196 4197 /* block age extent cache ops */ 4198 void f2fs_init_age_extent_tree(struct inode *inode); 4199 bool f2fs_lookup_age_extent_cache(struct inode *inode, pgoff_t pgofs, 4200 struct extent_info *ei); 4201 void f2fs_update_age_extent_cache(struct dnode_of_data *dn); 4202 void f2fs_update_age_extent_cache_range(struct dnode_of_data *dn, 4203 pgoff_t fofs, unsigned int len); 4204 unsigned int f2fs_shrink_age_extent_tree(struct f2fs_sb_info *sbi, 4205 int nr_shrink); 4206 4207 /* 4208 * sysfs.c 4209 */ 4210 #define MIN_RA_MUL 2 4211 #define MAX_RA_MUL 256 4212 4213 int __init f2fs_init_sysfs(void); 4214 void f2fs_exit_sysfs(void); 4215 int f2fs_register_sysfs(struct f2fs_sb_info *sbi); 4216 void f2fs_unregister_sysfs(struct f2fs_sb_info *sbi); 4217 4218 /* verity.c */ 4219 extern const struct fsverity_operations f2fs_verityops; 4220 4221 /* 4222 * crypto support 4223 */ 4224 static inline bool f2fs_encrypted_file(struct inode *inode) 4225 { 4226 return IS_ENCRYPTED(inode) && S_ISREG(inode->i_mode); 4227 } 4228 4229 static inline void f2fs_set_encrypted_inode(struct inode *inode) 4230 { 4231 #ifdef CONFIG_FS_ENCRYPTION 4232 file_set_encrypt(inode); 4233 f2fs_set_inode_flags(inode); 4234 #endif 4235 } 4236 4237 /* 4238 * Returns true if the reads of the inode's data need to undergo some 4239 * postprocessing step, like decryption or authenticity verification. 4240 */ 4241 static inline bool f2fs_post_read_required(struct inode *inode) 4242 { 4243 return f2fs_encrypted_file(inode) || fsverity_active(inode) || 4244 f2fs_compressed_file(inode); 4245 } 4246 4247 /* 4248 * compress.c 4249 */ 4250 #ifdef CONFIG_F2FS_FS_COMPRESSION 4251 bool f2fs_is_compressed_page(struct page *page); 4252 struct page *f2fs_compress_control_page(struct page *page); 4253 int f2fs_prepare_compress_overwrite(struct inode *inode, 4254 struct page **pagep, pgoff_t index, void **fsdata); 4255 bool f2fs_compress_write_end(struct inode *inode, void *fsdata, 4256 pgoff_t index, unsigned copied); 4257 int f2fs_truncate_partial_cluster(struct inode *inode, u64 from, bool lock); 4258 void f2fs_compress_write_end_io(struct bio *bio, struct page *page); 4259 bool f2fs_is_compress_backend_ready(struct inode *inode); 4260 bool f2fs_is_compress_level_valid(int alg, int lvl); 4261 int __init f2fs_init_compress_mempool(void); 4262 void f2fs_destroy_compress_mempool(void); 4263 void f2fs_decompress_cluster(struct decompress_io_ctx *dic, bool in_task); 4264 void f2fs_end_read_compressed_page(struct page *page, bool failed, 4265 block_t blkaddr, bool in_task); 4266 bool f2fs_cluster_is_empty(struct compress_ctx *cc); 4267 bool f2fs_cluster_can_merge_page(struct compress_ctx *cc, pgoff_t index); 4268 bool f2fs_all_cluster_page_ready(struct compress_ctx *cc, struct page **pages, 4269 int index, int nr_pages, bool uptodate); 4270 bool f2fs_sanity_check_cluster(struct dnode_of_data *dn); 4271 void f2fs_compress_ctx_add_page(struct compress_ctx *cc, struct page *page); 4272 int f2fs_write_multi_pages(struct compress_ctx *cc, 4273 int *submitted, 4274 struct writeback_control *wbc, 4275 enum iostat_type io_type); 4276 int f2fs_is_compressed_cluster(struct inode *inode, pgoff_t index); 4277 void f2fs_update_read_extent_tree_range_compressed(struct inode *inode, 4278 pgoff_t fofs, block_t blkaddr, 4279 unsigned int llen, unsigned int c_len); 4280 int f2fs_read_multi_pages(struct compress_ctx *cc, struct bio **bio_ret, 4281 unsigned nr_pages, sector_t *last_block_in_bio, 4282 bool is_readahead, bool for_write); 4283 struct decompress_io_ctx *f2fs_alloc_dic(struct compress_ctx *cc); 4284 void f2fs_decompress_end_io(struct decompress_io_ctx *dic, bool failed, 4285 bool in_task); 4286 void f2fs_put_page_dic(struct page *page, bool in_task); 4287 unsigned int f2fs_cluster_blocks_are_contiguous(struct dnode_of_data *dn); 4288 int f2fs_init_compress_ctx(struct compress_ctx *cc); 4289 void f2fs_destroy_compress_ctx(struct compress_ctx *cc, bool reuse); 4290 void f2fs_init_compress_info(struct f2fs_sb_info *sbi); 4291 int f2fs_init_compress_inode(struct f2fs_sb_info *sbi); 4292 void f2fs_destroy_compress_inode(struct f2fs_sb_info *sbi); 4293 int f2fs_init_page_array_cache(struct f2fs_sb_info *sbi); 4294 void f2fs_destroy_page_array_cache(struct f2fs_sb_info *sbi); 4295 int __init f2fs_init_compress_cache(void); 4296 void f2fs_destroy_compress_cache(void); 4297 struct address_space *COMPRESS_MAPPING(struct f2fs_sb_info *sbi); 4298 void f2fs_invalidate_compress_page(struct f2fs_sb_info *sbi, block_t blkaddr); 4299 void f2fs_cache_compressed_page(struct f2fs_sb_info *sbi, struct page *page, 4300 nid_t ino, block_t blkaddr); 4301 bool f2fs_load_compressed_page(struct f2fs_sb_info *sbi, struct page *page, 4302 block_t blkaddr); 4303 void f2fs_invalidate_compress_pages(struct f2fs_sb_info *sbi, nid_t ino); 4304 #define inc_compr_inode_stat(inode) \ 4305 do { \ 4306 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); \ 4307 sbi->compr_new_inode++; \ 4308 } while (0) 4309 #define add_compr_block_stat(inode, blocks) \ 4310 do { \ 4311 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); \ 4312 int diff = F2FS_I(inode)->i_cluster_size - blocks; \ 4313 sbi->compr_written_block += blocks; \ 4314 sbi->compr_saved_block += diff; \ 4315 } while (0) 4316 #else 4317 static inline bool f2fs_is_compressed_page(struct page *page) { return false; } 4318 static inline bool f2fs_is_compress_backend_ready(struct inode *inode) 4319 { 4320 if (!f2fs_compressed_file(inode)) 4321 return true; 4322 /* not support compression */ 4323 return false; 4324 } 4325 static inline bool f2fs_is_compress_level_valid(int alg, int lvl) { return false; } 4326 static inline struct page *f2fs_compress_control_page(struct page *page) 4327 { 4328 WARN_ON_ONCE(1); 4329 return ERR_PTR(-EINVAL); 4330 } 4331 static inline int __init f2fs_init_compress_mempool(void) { return 0; } 4332 static inline void f2fs_destroy_compress_mempool(void) { } 4333 static inline void f2fs_decompress_cluster(struct decompress_io_ctx *dic, 4334 bool in_task) { } 4335 static inline void f2fs_end_read_compressed_page(struct page *page, 4336 bool failed, block_t blkaddr, bool in_task) 4337 { 4338 WARN_ON_ONCE(1); 4339 } 4340 static inline void f2fs_put_page_dic(struct page *page, bool in_task) 4341 { 4342 WARN_ON_ONCE(1); 4343 } 4344 static inline unsigned int f2fs_cluster_blocks_are_contiguous(struct dnode_of_data *dn) { return 0; } 4345 static inline bool f2fs_sanity_check_cluster(struct dnode_of_data *dn) { return false; } 4346 static inline int f2fs_init_compress_inode(struct f2fs_sb_info *sbi) { return 0; } 4347 static inline void f2fs_destroy_compress_inode(struct f2fs_sb_info *sbi) { } 4348 static inline int f2fs_init_page_array_cache(struct f2fs_sb_info *sbi) { return 0; } 4349 static inline void f2fs_destroy_page_array_cache(struct f2fs_sb_info *sbi) { } 4350 static inline int __init f2fs_init_compress_cache(void) { return 0; } 4351 static inline void f2fs_destroy_compress_cache(void) { } 4352 static inline void f2fs_invalidate_compress_page(struct f2fs_sb_info *sbi, 4353 block_t blkaddr) { } 4354 static inline void f2fs_cache_compressed_page(struct f2fs_sb_info *sbi, 4355 struct page *page, nid_t ino, block_t blkaddr) { } 4356 static inline bool f2fs_load_compressed_page(struct f2fs_sb_info *sbi, 4357 struct page *page, block_t blkaddr) { return false; } 4358 static inline void f2fs_invalidate_compress_pages(struct f2fs_sb_info *sbi, 4359 nid_t ino) { } 4360 #define inc_compr_inode_stat(inode) do { } while (0) 4361 static inline void f2fs_update_read_extent_tree_range_compressed( 4362 struct inode *inode, 4363 pgoff_t fofs, block_t blkaddr, 4364 unsigned int llen, unsigned int c_len) { } 4365 #endif 4366 4367 static inline int set_compress_context(struct inode *inode) 4368 { 4369 #ifdef CONFIG_F2FS_FS_COMPRESSION 4370 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 4371 4372 F2FS_I(inode)->i_compress_algorithm = 4373 F2FS_OPTION(sbi).compress_algorithm; 4374 F2FS_I(inode)->i_log_cluster_size = 4375 F2FS_OPTION(sbi).compress_log_size; 4376 F2FS_I(inode)->i_compress_flag = 4377 F2FS_OPTION(sbi).compress_chksum ? 4378 BIT(COMPRESS_CHKSUM) : 0; 4379 F2FS_I(inode)->i_cluster_size = 4380 BIT(F2FS_I(inode)->i_log_cluster_size); 4381 if ((F2FS_I(inode)->i_compress_algorithm == COMPRESS_LZ4 || 4382 F2FS_I(inode)->i_compress_algorithm == COMPRESS_ZSTD) && 4383 F2FS_OPTION(sbi).compress_level) 4384 F2FS_I(inode)->i_compress_level = 4385 F2FS_OPTION(sbi).compress_level; 4386 F2FS_I(inode)->i_flags |= F2FS_COMPR_FL; 4387 set_inode_flag(inode, FI_COMPRESSED_FILE); 4388 stat_inc_compr_inode(inode); 4389 inc_compr_inode_stat(inode); 4390 f2fs_mark_inode_dirty_sync(inode, true); 4391 return 0; 4392 #else 4393 return -EOPNOTSUPP; 4394 #endif 4395 } 4396 4397 static inline bool f2fs_disable_compressed_file(struct inode *inode) 4398 { 4399 struct f2fs_inode_info *fi = F2FS_I(inode); 4400 4401 if (!f2fs_compressed_file(inode)) 4402 return true; 4403 if (S_ISREG(inode->i_mode) && F2FS_HAS_BLOCKS(inode)) 4404 return false; 4405 4406 fi->i_flags &= ~F2FS_COMPR_FL; 4407 stat_dec_compr_inode(inode); 4408 clear_inode_flag(inode, FI_COMPRESSED_FILE); 4409 f2fs_mark_inode_dirty_sync(inode, true); 4410 return true; 4411 } 4412 4413 #define F2FS_FEATURE_FUNCS(name, flagname) \ 4414 static inline bool f2fs_sb_has_##name(struct f2fs_sb_info *sbi) \ 4415 { \ 4416 return F2FS_HAS_FEATURE(sbi, F2FS_FEATURE_##flagname); \ 4417 } 4418 4419 F2FS_FEATURE_FUNCS(encrypt, ENCRYPT); 4420 F2FS_FEATURE_FUNCS(blkzoned, BLKZONED); 4421 F2FS_FEATURE_FUNCS(extra_attr, EXTRA_ATTR); 4422 F2FS_FEATURE_FUNCS(project_quota, PRJQUOTA); 4423 F2FS_FEATURE_FUNCS(inode_chksum, INODE_CHKSUM); 4424 F2FS_FEATURE_FUNCS(flexible_inline_xattr, FLEXIBLE_INLINE_XATTR); 4425 F2FS_FEATURE_FUNCS(quota_ino, QUOTA_INO); 4426 F2FS_FEATURE_FUNCS(inode_crtime, INODE_CRTIME); 4427 F2FS_FEATURE_FUNCS(lost_found, LOST_FOUND); 4428 F2FS_FEATURE_FUNCS(verity, VERITY); 4429 F2FS_FEATURE_FUNCS(sb_chksum, SB_CHKSUM); 4430 F2FS_FEATURE_FUNCS(casefold, CASEFOLD); 4431 F2FS_FEATURE_FUNCS(compression, COMPRESSION); 4432 F2FS_FEATURE_FUNCS(readonly, RO); 4433 4434 #ifdef CONFIG_BLK_DEV_ZONED 4435 static inline bool f2fs_blkz_is_seq(struct f2fs_sb_info *sbi, int devi, 4436 block_t blkaddr) 4437 { 4438 unsigned int zno = blkaddr / sbi->blocks_per_blkz; 4439 4440 return test_bit(zno, FDEV(devi).blkz_seq); 4441 } 4442 #endif 4443 4444 static inline int f2fs_bdev_index(struct f2fs_sb_info *sbi, 4445 struct block_device *bdev) 4446 { 4447 int i; 4448 4449 if (!f2fs_is_multi_device(sbi)) 4450 return 0; 4451 4452 for (i = 0; i < sbi->s_ndevs; i++) 4453 if (FDEV(i).bdev == bdev) 4454 return i; 4455 4456 WARN_ON(1); 4457 return -1; 4458 } 4459 4460 static inline bool f2fs_hw_should_discard(struct f2fs_sb_info *sbi) 4461 { 4462 return f2fs_sb_has_blkzoned(sbi); 4463 } 4464 4465 static inline bool f2fs_bdev_support_discard(struct block_device *bdev) 4466 { 4467 return bdev_max_discard_sectors(bdev) || bdev_is_zoned(bdev); 4468 } 4469 4470 static inline bool f2fs_hw_support_discard(struct f2fs_sb_info *sbi) 4471 { 4472 int i; 4473 4474 if (!f2fs_is_multi_device(sbi)) 4475 return f2fs_bdev_support_discard(sbi->sb->s_bdev); 4476 4477 for (i = 0; i < sbi->s_ndevs; i++) 4478 if (f2fs_bdev_support_discard(FDEV(i).bdev)) 4479 return true; 4480 return false; 4481 } 4482 4483 static inline bool f2fs_realtime_discard_enable(struct f2fs_sb_info *sbi) 4484 { 4485 return (test_opt(sbi, DISCARD) && f2fs_hw_support_discard(sbi)) || 4486 f2fs_hw_should_discard(sbi); 4487 } 4488 4489 static inline bool f2fs_hw_is_readonly(struct f2fs_sb_info *sbi) 4490 { 4491 int i; 4492 4493 if (!f2fs_is_multi_device(sbi)) 4494 return bdev_read_only(sbi->sb->s_bdev); 4495 4496 for (i = 0; i < sbi->s_ndevs; i++) 4497 if (bdev_read_only(FDEV(i).bdev)) 4498 return true; 4499 return false; 4500 } 4501 4502 static inline bool f2fs_dev_is_readonly(struct f2fs_sb_info *sbi) 4503 { 4504 return f2fs_sb_has_readonly(sbi) || f2fs_hw_is_readonly(sbi); 4505 } 4506 4507 static inline bool f2fs_lfs_mode(struct f2fs_sb_info *sbi) 4508 { 4509 return F2FS_OPTION(sbi).fs_mode == FS_MODE_LFS; 4510 } 4511 4512 static inline bool f2fs_low_mem_mode(struct f2fs_sb_info *sbi) 4513 { 4514 return F2FS_OPTION(sbi).memory_mode == MEMORY_MODE_LOW; 4515 } 4516 4517 static inline bool f2fs_may_compress(struct inode *inode) 4518 { 4519 if (IS_SWAPFILE(inode) || f2fs_is_pinned_file(inode) || 4520 f2fs_is_atomic_file(inode) || f2fs_has_inline_data(inode) || 4521 f2fs_is_mmap_file(inode)) 4522 return false; 4523 return S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode); 4524 } 4525 4526 static inline void f2fs_i_compr_blocks_update(struct inode *inode, 4527 u64 blocks, bool add) 4528 { 4529 struct f2fs_inode_info *fi = F2FS_I(inode); 4530 int diff = fi->i_cluster_size - blocks; 4531 4532 /* don't update i_compr_blocks if saved blocks were released */ 4533 if (!add && !atomic_read(&fi->i_compr_blocks)) 4534 return; 4535 4536 if (add) { 4537 atomic_add(diff, &fi->i_compr_blocks); 4538 stat_add_compr_blocks(inode, diff); 4539 } else { 4540 atomic_sub(diff, &fi->i_compr_blocks); 4541 stat_sub_compr_blocks(inode, diff); 4542 } 4543 f2fs_mark_inode_dirty_sync(inode, true); 4544 } 4545 4546 static inline bool f2fs_allow_multi_device_dio(struct f2fs_sb_info *sbi, 4547 int flag) 4548 { 4549 if (!f2fs_is_multi_device(sbi)) 4550 return false; 4551 if (flag != F2FS_GET_BLOCK_DIO) 4552 return false; 4553 return sbi->aligned_blksize; 4554 } 4555 4556 static inline bool f2fs_need_verity(const struct inode *inode, pgoff_t idx) 4557 { 4558 return fsverity_active(inode) && 4559 idx < DIV_ROUND_UP(inode->i_size, PAGE_SIZE); 4560 } 4561 4562 #ifdef CONFIG_F2FS_FAULT_INJECTION 4563 extern void f2fs_build_fault_attr(struct f2fs_sb_info *sbi, unsigned int rate, 4564 unsigned int type); 4565 #else 4566 #define f2fs_build_fault_attr(sbi, rate, type) do { } while (0) 4567 #endif 4568 4569 static inline bool is_journalled_quota(struct f2fs_sb_info *sbi) 4570 { 4571 #ifdef CONFIG_QUOTA 4572 if (f2fs_sb_has_quota_ino(sbi)) 4573 return true; 4574 if (F2FS_OPTION(sbi).s_qf_names[USRQUOTA] || 4575 F2FS_OPTION(sbi).s_qf_names[GRPQUOTA] || 4576 F2FS_OPTION(sbi).s_qf_names[PRJQUOTA]) 4577 return true; 4578 #endif 4579 return false; 4580 } 4581 4582 static inline bool f2fs_block_unit_discard(struct f2fs_sb_info *sbi) 4583 { 4584 return F2FS_OPTION(sbi).discard_unit == DISCARD_UNIT_BLOCK; 4585 } 4586 4587 static inline void f2fs_io_schedule_timeout(long timeout) 4588 { 4589 set_current_state(TASK_UNINTERRUPTIBLE); 4590 io_schedule_timeout(timeout); 4591 } 4592 4593 static inline void f2fs_handle_page_eio(struct f2fs_sb_info *sbi, pgoff_t ofs, 4594 enum page_type type) 4595 { 4596 if (unlikely(f2fs_cp_error(sbi))) 4597 return; 4598 4599 if (ofs == sbi->page_eio_ofs[type]) { 4600 if (sbi->page_eio_cnt[type]++ == MAX_RETRY_PAGE_EIO) 4601 set_ckpt_flags(sbi, CP_ERROR_FLAG); 4602 } else { 4603 sbi->page_eio_ofs[type] = ofs; 4604 sbi->page_eio_cnt[type] = 0; 4605 } 4606 } 4607 4608 static inline bool f2fs_is_readonly(struct f2fs_sb_info *sbi) 4609 { 4610 return f2fs_sb_has_readonly(sbi) || f2fs_readonly(sbi->sb); 4611 } 4612 4613 #define EFSBADCRC EBADMSG /* Bad CRC detected */ 4614 #define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */ 4615 4616 #endif /* _LINUX_F2FS_H */ 4617