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/vmalloc.h> 22 #include <linux/bio.h> 23 #include <linux/blkdev.h> 24 #include <linux/quotaops.h> 25 #include <linux/part_stat.h> 26 #include <crypto/hash.h> 27 28 #include <linux/fscrypt.h> 29 #include <linux/fsverity.h> 30 31 struct pagevec; 32 33 #ifdef CONFIG_F2FS_CHECK_FS 34 #define f2fs_bug_on(sbi, condition) BUG_ON(condition) 35 #else 36 #define f2fs_bug_on(sbi, condition) \ 37 do { \ 38 if (WARN_ON(condition)) \ 39 set_sbi_flag(sbi, SBI_NEED_FSCK); \ 40 } while (0) 41 #endif 42 43 enum { 44 FAULT_KMALLOC, 45 FAULT_KVMALLOC, 46 FAULT_PAGE_ALLOC, 47 FAULT_PAGE_GET, 48 FAULT_ALLOC_BIO, /* it's obsolete due to bio_alloc() will never fail */ 49 FAULT_ALLOC_NID, 50 FAULT_ORPHAN, 51 FAULT_BLOCK, 52 FAULT_DIR_DEPTH, 53 FAULT_EVICT_INODE, 54 FAULT_TRUNCATE, 55 FAULT_READ_IO, 56 FAULT_CHECKPOINT, 57 FAULT_DISCARD, 58 FAULT_WRITE_IO, 59 FAULT_SLAB_ALLOC, 60 FAULT_DQUOT_INIT, 61 FAULT_LOCK_OP, 62 FAULT_MAX, 63 }; 64 65 #ifdef CONFIG_F2FS_FAULT_INJECTION 66 #define F2FS_ALL_FAULT_TYPE ((1 << FAULT_MAX) - 1) 67 68 struct f2fs_fault_info { 69 atomic_t inject_ops; 70 unsigned int inject_rate; 71 unsigned int inject_type; 72 }; 73 74 extern const char *f2fs_fault_name[FAULT_MAX]; 75 #define IS_FAULT_SET(fi, type) ((fi)->inject_type & (1 << (type))) 76 #endif 77 78 /* 79 * For mount options 80 */ 81 #define F2FS_MOUNT_DISABLE_ROLL_FORWARD 0x00000002 82 #define F2FS_MOUNT_DISCARD 0x00000004 83 #define F2FS_MOUNT_NOHEAP 0x00000008 84 #define F2FS_MOUNT_XATTR_USER 0x00000010 85 #define F2FS_MOUNT_POSIX_ACL 0x00000020 86 #define F2FS_MOUNT_DISABLE_EXT_IDENTIFY 0x00000040 87 #define F2FS_MOUNT_INLINE_XATTR 0x00000080 88 #define F2FS_MOUNT_INLINE_DATA 0x00000100 89 #define F2FS_MOUNT_INLINE_DENTRY 0x00000200 90 #define F2FS_MOUNT_FLUSH_MERGE 0x00000400 91 #define F2FS_MOUNT_NOBARRIER 0x00000800 92 #define F2FS_MOUNT_FASTBOOT 0x00001000 93 #define F2FS_MOUNT_EXTENT_CACHE 0x00002000 94 #define F2FS_MOUNT_DATA_FLUSH 0x00008000 95 #define F2FS_MOUNT_FAULT_INJECTION 0x00010000 96 #define F2FS_MOUNT_USRQUOTA 0x00080000 97 #define F2FS_MOUNT_GRPQUOTA 0x00100000 98 #define F2FS_MOUNT_PRJQUOTA 0x00200000 99 #define F2FS_MOUNT_QUOTA 0x00400000 100 #define F2FS_MOUNT_INLINE_XATTR_SIZE 0x00800000 101 #define F2FS_MOUNT_RESERVE_ROOT 0x01000000 102 #define F2FS_MOUNT_DISABLE_CHECKPOINT 0x02000000 103 #define F2FS_MOUNT_NORECOVERY 0x04000000 104 #define F2FS_MOUNT_ATGC 0x08000000 105 #define F2FS_MOUNT_MERGE_CHECKPOINT 0x10000000 106 #define F2FS_MOUNT_GC_MERGE 0x20000000 107 #define F2FS_MOUNT_COMPRESS_CACHE 0x40000000 108 109 #define F2FS_OPTION(sbi) ((sbi)->mount_opt) 110 #define clear_opt(sbi, option) (F2FS_OPTION(sbi).opt &= ~F2FS_MOUNT_##option) 111 #define set_opt(sbi, option) (F2FS_OPTION(sbi).opt |= F2FS_MOUNT_##option) 112 #define test_opt(sbi, option) (F2FS_OPTION(sbi).opt & F2FS_MOUNT_##option) 113 114 #define ver_after(a, b) (typecheck(unsigned long long, a) && \ 115 typecheck(unsigned long long, b) && \ 116 ((long long)((a) - (b)) > 0)) 117 118 typedef u32 block_t; /* 119 * should not change u32, since it is the on-disk block 120 * address format, __le32. 121 */ 122 typedef u32 nid_t; 123 124 #define COMPRESS_EXT_NUM 16 125 126 struct f2fs_mount_info { 127 unsigned int opt; 128 int write_io_size_bits; /* Write IO size bits */ 129 block_t root_reserved_blocks; /* root reserved blocks */ 130 kuid_t s_resuid; /* reserved blocks for uid */ 131 kgid_t s_resgid; /* reserved blocks for gid */ 132 int active_logs; /* # of active logs */ 133 int inline_xattr_size; /* inline xattr size */ 134 #ifdef CONFIG_F2FS_FAULT_INJECTION 135 struct f2fs_fault_info fault_info; /* For fault injection */ 136 #endif 137 #ifdef CONFIG_QUOTA 138 /* Names of quota files with journalled quota */ 139 char *s_qf_names[MAXQUOTAS]; 140 int s_jquota_fmt; /* Format of quota to use */ 141 #endif 142 /* For which write hints are passed down to block layer */ 143 int whint_mode; 144 int alloc_mode; /* segment allocation policy */ 145 int fsync_mode; /* fsync policy */ 146 int fs_mode; /* fs mode: LFS or ADAPTIVE */ 147 int bggc_mode; /* bggc mode: off, on or sync */ 148 int discard_unit; /* 149 * discard command's offset/size should 150 * be aligned to this unit: block, 151 * segment or section 152 */ 153 struct fscrypt_dummy_policy dummy_enc_policy; /* test dummy encryption */ 154 block_t unusable_cap_perc; /* percentage for cap */ 155 block_t unusable_cap; /* Amount of space allowed to be 156 * unusable when disabling checkpoint 157 */ 158 159 /* For compression */ 160 unsigned char compress_algorithm; /* algorithm type */ 161 unsigned char compress_log_size; /* cluster log size */ 162 unsigned char compress_level; /* compress level */ 163 bool compress_chksum; /* compressed data chksum */ 164 unsigned char compress_ext_cnt; /* extension count */ 165 unsigned char nocompress_ext_cnt; /* nocompress extension count */ 166 int compress_mode; /* compression mode */ 167 unsigned char extensions[COMPRESS_EXT_NUM][F2FS_EXTENSION_LEN]; /* extensions */ 168 unsigned char noextensions[COMPRESS_EXT_NUM][F2FS_EXTENSION_LEN]; /* extensions */ 169 }; 170 171 #define F2FS_FEATURE_ENCRYPT 0x0001 172 #define F2FS_FEATURE_BLKZONED 0x0002 173 #define F2FS_FEATURE_ATOMIC_WRITE 0x0004 174 #define F2FS_FEATURE_EXTRA_ATTR 0x0008 175 #define F2FS_FEATURE_PRJQUOTA 0x0010 176 #define F2FS_FEATURE_INODE_CHKSUM 0x0020 177 #define F2FS_FEATURE_FLEXIBLE_INLINE_XATTR 0x0040 178 #define F2FS_FEATURE_QUOTA_INO 0x0080 179 #define F2FS_FEATURE_INODE_CRTIME 0x0100 180 #define F2FS_FEATURE_LOST_FOUND 0x0200 181 #define F2FS_FEATURE_VERITY 0x0400 182 #define F2FS_FEATURE_SB_CHKSUM 0x0800 183 #define F2FS_FEATURE_CASEFOLD 0x1000 184 #define F2FS_FEATURE_COMPRESSION 0x2000 185 #define F2FS_FEATURE_RO 0x4000 186 187 #define __F2FS_HAS_FEATURE(raw_super, mask) \ 188 ((raw_super->feature & cpu_to_le32(mask)) != 0) 189 #define F2FS_HAS_FEATURE(sbi, mask) __F2FS_HAS_FEATURE(sbi->raw_super, mask) 190 #define F2FS_SET_FEATURE(sbi, mask) \ 191 (sbi->raw_super->feature |= cpu_to_le32(mask)) 192 #define F2FS_CLEAR_FEATURE(sbi, mask) \ 193 (sbi->raw_super->feature &= ~cpu_to_le32(mask)) 194 195 /* 196 * Default values for user and/or group using reserved blocks 197 */ 198 #define F2FS_DEF_RESUID 0 199 #define F2FS_DEF_RESGID 0 200 201 /* 202 * For checkpoint manager 203 */ 204 enum { 205 NAT_BITMAP, 206 SIT_BITMAP 207 }; 208 209 #define CP_UMOUNT 0x00000001 210 #define CP_FASTBOOT 0x00000002 211 #define CP_SYNC 0x00000004 212 #define CP_RECOVERY 0x00000008 213 #define CP_DISCARD 0x00000010 214 #define CP_TRIMMED 0x00000020 215 #define CP_PAUSE 0x00000040 216 #define CP_RESIZE 0x00000080 217 218 #define MAX_DISCARD_BLOCKS(sbi) BLKS_PER_SEC(sbi) 219 #define DEF_MAX_DISCARD_REQUEST 8 /* issue 8 discards per round */ 220 #define DEF_MIN_DISCARD_ISSUE_TIME 50 /* 50 ms, if exists */ 221 #define DEF_MID_DISCARD_ISSUE_TIME 500 /* 500 ms, if device busy */ 222 #define DEF_MAX_DISCARD_ISSUE_TIME 60000 /* 60 s, if no candidates */ 223 #define DEF_DISCARD_URGENT_UTIL 80 /* do more discard over 80% */ 224 #define DEF_CP_INTERVAL 60 /* 60 secs */ 225 #define DEF_IDLE_INTERVAL 5 /* 5 secs */ 226 #define DEF_DISABLE_INTERVAL 5 /* 5 secs */ 227 #define DEF_DISABLE_QUICK_INTERVAL 1 /* 1 secs */ 228 #define DEF_UMOUNT_DISCARD_TIMEOUT 5 /* 5 secs */ 229 230 struct cp_control { 231 int reason; 232 __u64 trim_start; 233 __u64 trim_end; 234 __u64 trim_minlen; 235 }; 236 237 /* 238 * indicate meta/data type 239 */ 240 enum { 241 META_CP, 242 META_NAT, 243 META_SIT, 244 META_SSA, 245 META_MAX, 246 META_POR, 247 DATA_GENERIC, /* check range only */ 248 DATA_GENERIC_ENHANCE, /* strong check on range and segment bitmap */ 249 DATA_GENERIC_ENHANCE_READ, /* 250 * strong check on range and segment 251 * bitmap but no warning due to race 252 * condition of read on truncated area 253 * by extent_cache 254 */ 255 META_GENERIC, 256 }; 257 258 /* for the list of ino */ 259 enum { 260 ORPHAN_INO, /* for orphan ino list */ 261 APPEND_INO, /* for append ino list */ 262 UPDATE_INO, /* for update ino list */ 263 TRANS_DIR_INO, /* for trasactions dir ino list */ 264 FLUSH_INO, /* for multiple device flushing */ 265 MAX_INO_ENTRY, /* max. list */ 266 }; 267 268 struct ino_entry { 269 struct list_head list; /* list head */ 270 nid_t ino; /* inode number */ 271 unsigned int dirty_device; /* dirty device bitmap */ 272 }; 273 274 /* for the list of inodes to be GCed */ 275 struct inode_entry { 276 struct list_head list; /* list head */ 277 struct inode *inode; /* vfs inode pointer */ 278 }; 279 280 struct fsync_node_entry { 281 struct list_head list; /* list head */ 282 struct page *page; /* warm node page pointer */ 283 unsigned int seq_id; /* sequence id */ 284 }; 285 286 struct ckpt_req { 287 struct completion wait; /* completion for checkpoint done */ 288 struct llist_node llnode; /* llist_node to be linked in wait queue */ 289 int ret; /* return code of checkpoint */ 290 ktime_t queue_time; /* request queued time */ 291 }; 292 293 struct ckpt_req_control { 294 struct task_struct *f2fs_issue_ckpt; /* checkpoint task */ 295 int ckpt_thread_ioprio; /* checkpoint merge thread ioprio */ 296 wait_queue_head_t ckpt_wait_queue; /* waiting queue for wake-up */ 297 atomic_t issued_ckpt; /* # of actually issued ckpts */ 298 atomic_t total_ckpt; /* # of total ckpts */ 299 atomic_t queued_ckpt; /* # of queued ckpts */ 300 struct llist_head issue_list; /* list for command issue */ 301 spinlock_t stat_lock; /* lock for below checkpoint time stats */ 302 unsigned int cur_time; /* cur wait time in msec for currently issued checkpoint */ 303 unsigned int peak_time; /* peak wait time in msec until now */ 304 }; 305 306 /* for the bitmap indicate blocks to be discarded */ 307 struct discard_entry { 308 struct list_head list; /* list head */ 309 block_t start_blkaddr; /* start blockaddr of current segment */ 310 unsigned char discard_map[SIT_VBLOCK_MAP_SIZE]; /* segment discard bitmap */ 311 }; 312 313 /* default discard granularity of inner discard thread, unit: block count */ 314 #define DEFAULT_DISCARD_GRANULARITY 16 315 316 /* max discard pend list number */ 317 #define MAX_PLIST_NUM 512 318 #define plist_idx(blk_num) ((blk_num) >= MAX_PLIST_NUM ? \ 319 (MAX_PLIST_NUM - 1) : ((blk_num) - 1)) 320 321 enum { 322 D_PREP, /* initial */ 323 D_PARTIAL, /* partially submitted */ 324 D_SUBMIT, /* all submitted */ 325 D_DONE, /* finished */ 326 }; 327 328 struct discard_info { 329 block_t lstart; /* logical start address */ 330 block_t len; /* length */ 331 block_t start; /* actual start address in dev */ 332 }; 333 334 struct discard_cmd { 335 struct rb_node rb_node; /* rb node located in rb-tree */ 336 union { 337 struct { 338 block_t lstart; /* logical start address */ 339 block_t len; /* length */ 340 block_t start; /* actual start address in dev */ 341 }; 342 struct discard_info di; /* discard info */ 343 344 }; 345 struct list_head list; /* command list */ 346 struct completion wait; /* compleation */ 347 struct block_device *bdev; /* bdev */ 348 unsigned short ref; /* reference count */ 349 unsigned char state; /* state */ 350 unsigned char queued; /* queued discard */ 351 int error; /* bio error */ 352 spinlock_t lock; /* for state/bio_ref updating */ 353 unsigned short bio_ref; /* bio reference count */ 354 }; 355 356 enum { 357 DPOLICY_BG, 358 DPOLICY_FORCE, 359 DPOLICY_FSTRIM, 360 DPOLICY_UMOUNT, 361 MAX_DPOLICY, 362 }; 363 364 struct discard_policy { 365 int type; /* type of discard */ 366 unsigned int min_interval; /* used for candidates exist */ 367 unsigned int mid_interval; /* used for device busy */ 368 unsigned int max_interval; /* used for candidates not exist */ 369 unsigned int max_requests; /* # of discards issued per round */ 370 unsigned int io_aware_gran; /* minimum granularity discard not be aware of I/O */ 371 bool io_aware; /* issue discard in idle time */ 372 bool sync; /* submit discard with REQ_SYNC flag */ 373 bool ordered; /* issue discard by lba order */ 374 bool timeout; /* discard timeout for put_super */ 375 unsigned int granularity; /* discard granularity */ 376 }; 377 378 struct discard_cmd_control { 379 struct task_struct *f2fs_issue_discard; /* discard thread */ 380 struct list_head entry_list; /* 4KB discard entry list */ 381 struct list_head pend_list[MAX_PLIST_NUM];/* store pending entries */ 382 struct list_head wait_list; /* store on-flushing entries */ 383 struct list_head fstrim_list; /* in-flight discard from fstrim */ 384 wait_queue_head_t discard_wait_queue; /* waiting queue for wake-up */ 385 unsigned int discard_wake; /* to wake up discard thread */ 386 struct mutex cmd_lock; 387 unsigned int nr_discards; /* # of discards in the list */ 388 unsigned int max_discards; /* max. discards to be issued */ 389 unsigned int discard_granularity; /* discard granularity */ 390 unsigned int undiscard_blks; /* # of undiscard blocks */ 391 unsigned int next_pos; /* next discard position */ 392 atomic_t issued_discard; /* # of issued discard */ 393 atomic_t queued_discard; /* # of queued discard */ 394 atomic_t discard_cmd_cnt; /* # of cached cmd count */ 395 struct rb_root_cached root; /* root of discard rb-tree */ 396 bool rbtree_check; /* config for consistence check */ 397 }; 398 399 /* for the list of fsync inodes, used only during recovery */ 400 struct fsync_inode_entry { 401 struct list_head list; /* list head */ 402 struct inode *inode; /* vfs inode pointer */ 403 block_t blkaddr; /* block address locating the last fsync */ 404 block_t last_dentry; /* block address locating the last dentry */ 405 }; 406 407 #define nats_in_cursum(jnl) (le16_to_cpu((jnl)->n_nats)) 408 #define sits_in_cursum(jnl) (le16_to_cpu((jnl)->n_sits)) 409 410 #define nat_in_journal(jnl, i) ((jnl)->nat_j.entries[i].ne) 411 #define nid_in_journal(jnl, i) ((jnl)->nat_j.entries[i].nid) 412 #define sit_in_journal(jnl, i) ((jnl)->sit_j.entries[i].se) 413 #define segno_in_journal(jnl, i) ((jnl)->sit_j.entries[i].segno) 414 415 #define MAX_NAT_JENTRIES(jnl) (NAT_JOURNAL_ENTRIES - nats_in_cursum(jnl)) 416 #define MAX_SIT_JENTRIES(jnl) (SIT_JOURNAL_ENTRIES - sits_in_cursum(jnl)) 417 418 static inline int update_nats_in_cursum(struct f2fs_journal *journal, int i) 419 { 420 int before = nats_in_cursum(journal); 421 422 journal->n_nats = cpu_to_le16(before + i); 423 return before; 424 } 425 426 static inline int update_sits_in_cursum(struct f2fs_journal *journal, int i) 427 { 428 int before = sits_in_cursum(journal); 429 430 journal->n_sits = cpu_to_le16(before + i); 431 return before; 432 } 433 434 static inline bool __has_cursum_space(struct f2fs_journal *journal, 435 int size, int type) 436 { 437 if (type == NAT_JOURNAL) 438 return size <= MAX_NAT_JENTRIES(journal); 439 return size <= MAX_SIT_JENTRIES(journal); 440 } 441 442 /* for inline stuff */ 443 #define DEF_INLINE_RESERVED_SIZE 1 444 static inline int get_extra_isize(struct inode *inode); 445 static inline int get_inline_xattr_addrs(struct inode *inode); 446 #define MAX_INLINE_DATA(inode) (sizeof(__le32) * \ 447 (CUR_ADDRS_PER_INODE(inode) - \ 448 get_inline_xattr_addrs(inode) - \ 449 DEF_INLINE_RESERVED_SIZE)) 450 451 /* for inline dir */ 452 #define NR_INLINE_DENTRY(inode) (MAX_INLINE_DATA(inode) * BITS_PER_BYTE / \ 453 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \ 454 BITS_PER_BYTE + 1)) 455 #define INLINE_DENTRY_BITMAP_SIZE(inode) \ 456 DIV_ROUND_UP(NR_INLINE_DENTRY(inode), BITS_PER_BYTE) 457 #define INLINE_RESERVED_SIZE(inode) (MAX_INLINE_DATA(inode) - \ 458 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \ 459 NR_INLINE_DENTRY(inode) + \ 460 INLINE_DENTRY_BITMAP_SIZE(inode))) 461 462 /* 463 * For INODE and NODE manager 464 */ 465 /* for directory operations */ 466 467 struct f2fs_filename { 468 /* 469 * The filename the user specified. This is NULL for some 470 * filesystem-internal operations, e.g. converting an inline directory 471 * to a non-inline one, or roll-forward recovering an encrypted dentry. 472 */ 473 const struct qstr *usr_fname; 474 475 /* 476 * The on-disk filename. For encrypted directories, this is encrypted. 477 * This may be NULL for lookups in an encrypted dir without the key. 478 */ 479 struct fscrypt_str disk_name; 480 481 /* The dirhash of this filename */ 482 f2fs_hash_t hash; 483 484 #ifdef CONFIG_FS_ENCRYPTION 485 /* 486 * For lookups in encrypted directories: either the buffer backing 487 * disk_name, or a buffer that holds the decoded no-key name. 488 */ 489 struct fscrypt_str crypto_buf; 490 #endif 491 #if IS_ENABLED(CONFIG_UNICODE) 492 /* 493 * For casefolded directories: the casefolded name, but it's left NULL 494 * if the original name is not valid Unicode, if the directory is both 495 * casefolded and encrypted and its encryption key is unavailable, or if 496 * the filesystem is doing an internal operation where usr_fname is also 497 * NULL. In all these cases we fall back to treating the name as an 498 * opaque byte sequence. 499 */ 500 struct fscrypt_str cf_name; 501 #endif 502 }; 503 504 struct f2fs_dentry_ptr { 505 struct inode *inode; 506 void *bitmap; 507 struct f2fs_dir_entry *dentry; 508 __u8 (*filename)[F2FS_SLOT_LEN]; 509 int max; 510 int nr_bitmap; 511 }; 512 513 static inline void make_dentry_ptr_block(struct inode *inode, 514 struct f2fs_dentry_ptr *d, struct f2fs_dentry_block *t) 515 { 516 d->inode = inode; 517 d->max = NR_DENTRY_IN_BLOCK; 518 d->nr_bitmap = SIZE_OF_DENTRY_BITMAP; 519 d->bitmap = t->dentry_bitmap; 520 d->dentry = t->dentry; 521 d->filename = t->filename; 522 } 523 524 static inline void make_dentry_ptr_inline(struct inode *inode, 525 struct f2fs_dentry_ptr *d, void *t) 526 { 527 int entry_cnt = NR_INLINE_DENTRY(inode); 528 int bitmap_size = INLINE_DENTRY_BITMAP_SIZE(inode); 529 int reserved_size = INLINE_RESERVED_SIZE(inode); 530 531 d->inode = inode; 532 d->max = entry_cnt; 533 d->nr_bitmap = bitmap_size; 534 d->bitmap = t; 535 d->dentry = t + bitmap_size + reserved_size; 536 d->filename = t + bitmap_size + reserved_size + 537 SIZE_OF_DIR_ENTRY * entry_cnt; 538 } 539 540 /* 541 * XATTR_NODE_OFFSET stores xattrs to one node block per file keeping -1 542 * as its node offset to distinguish from index node blocks. 543 * But some bits are used to mark the node block. 544 */ 545 #define XATTR_NODE_OFFSET ((((unsigned int)-1) << OFFSET_BIT_SHIFT) \ 546 >> OFFSET_BIT_SHIFT) 547 enum { 548 ALLOC_NODE, /* allocate a new node page if needed */ 549 LOOKUP_NODE, /* look up a node without readahead */ 550 LOOKUP_NODE_RA, /* 551 * look up a node with readahead called 552 * by get_data_block. 553 */ 554 }; 555 556 #define DEFAULT_RETRY_IO_COUNT 8 /* maximum retry read IO or flush count */ 557 558 /* congestion wait timeout value, default: 20ms */ 559 #define DEFAULT_IO_TIMEOUT (msecs_to_jiffies(20)) 560 561 /* maximum retry quota flush count */ 562 #define DEFAULT_RETRY_QUOTA_FLUSH_COUNT 8 563 564 #define F2FS_LINK_MAX 0xffffffff /* maximum link count per file */ 565 566 #define MAX_DIR_RA_PAGES 4 /* maximum ra pages of dir */ 567 568 /* dirty segments threshold for triggering CP */ 569 #define DEFAULT_DIRTY_THRESHOLD 4 570 571 /* for in-memory extent cache entry */ 572 #define F2FS_MIN_EXTENT_LEN 64 /* minimum extent length */ 573 574 /* number of extent info in extent cache we try to shrink */ 575 #define EXTENT_CACHE_SHRINK_NUMBER 128 576 577 struct rb_entry { 578 struct rb_node rb_node; /* rb node located in rb-tree */ 579 union { 580 struct { 581 unsigned int ofs; /* start offset of the entry */ 582 unsigned int len; /* length of the entry */ 583 }; 584 unsigned long long key; /* 64-bits key */ 585 } __packed; 586 }; 587 588 struct extent_info { 589 unsigned int fofs; /* start offset in a file */ 590 unsigned int len; /* length of the extent */ 591 u32 blk; /* start block address of the extent */ 592 #ifdef CONFIG_F2FS_FS_COMPRESSION 593 unsigned int c_len; /* physical extent length of compressed blocks */ 594 #endif 595 }; 596 597 struct extent_node { 598 struct rb_node rb_node; /* rb node located in rb-tree */ 599 struct extent_info ei; /* extent info */ 600 struct list_head list; /* node in global extent list of sbi */ 601 struct extent_tree *et; /* extent tree pointer */ 602 }; 603 604 struct extent_tree { 605 nid_t ino; /* inode number */ 606 struct rb_root_cached root; /* root of extent info rb-tree */ 607 struct extent_node *cached_en; /* recently accessed extent node */ 608 struct extent_info largest; /* largested extent info */ 609 struct list_head list; /* to be used by sbi->zombie_list */ 610 rwlock_t lock; /* protect extent info rb-tree */ 611 atomic_t node_cnt; /* # of extent node in rb-tree*/ 612 bool largest_updated; /* largest extent updated */ 613 }; 614 615 /* 616 * This structure is taken from ext4_map_blocks. 617 * 618 * Note that, however, f2fs uses NEW and MAPPED flags for f2fs_map_blocks(). 619 */ 620 #define F2FS_MAP_NEW (1 << BH_New) 621 #define F2FS_MAP_MAPPED (1 << BH_Mapped) 622 #define F2FS_MAP_UNWRITTEN (1 << BH_Unwritten) 623 #define F2FS_MAP_FLAGS (F2FS_MAP_NEW | F2FS_MAP_MAPPED |\ 624 F2FS_MAP_UNWRITTEN) 625 626 struct f2fs_map_blocks { 627 struct block_device *m_bdev; /* for multi-device dio */ 628 block_t m_pblk; 629 block_t m_lblk; 630 unsigned int m_len; 631 unsigned int m_flags; 632 pgoff_t *m_next_pgofs; /* point next possible non-hole pgofs */ 633 pgoff_t *m_next_extent; /* point to next possible extent */ 634 int m_seg_type; 635 bool m_may_create; /* indicate it is from write path */ 636 bool m_multidev_dio; /* indicate it allows multi-device dio */ 637 }; 638 639 /* for flag in get_data_block */ 640 enum { 641 F2FS_GET_BLOCK_DEFAULT, 642 F2FS_GET_BLOCK_FIEMAP, 643 F2FS_GET_BLOCK_BMAP, 644 F2FS_GET_BLOCK_DIO, 645 F2FS_GET_BLOCK_PRE_DIO, 646 F2FS_GET_BLOCK_PRE_AIO, 647 F2FS_GET_BLOCK_PRECACHE, 648 }; 649 650 /* 651 * i_advise uses FADVISE_XXX_BIT. We can add additional hints later. 652 */ 653 #define FADVISE_COLD_BIT 0x01 654 #define FADVISE_LOST_PINO_BIT 0x02 655 #define FADVISE_ENCRYPT_BIT 0x04 656 #define FADVISE_ENC_NAME_BIT 0x08 657 #define FADVISE_KEEP_SIZE_BIT 0x10 658 #define FADVISE_HOT_BIT 0x20 659 #define FADVISE_VERITY_BIT 0x40 660 #define FADVISE_TRUNC_BIT 0x80 661 662 #define FADVISE_MODIFIABLE_BITS (FADVISE_COLD_BIT | FADVISE_HOT_BIT) 663 664 #define file_is_cold(inode) is_file(inode, FADVISE_COLD_BIT) 665 #define file_set_cold(inode) set_file(inode, FADVISE_COLD_BIT) 666 #define file_clear_cold(inode) clear_file(inode, FADVISE_COLD_BIT) 667 668 #define file_wrong_pino(inode) is_file(inode, FADVISE_LOST_PINO_BIT) 669 #define file_lost_pino(inode) set_file(inode, FADVISE_LOST_PINO_BIT) 670 #define file_got_pino(inode) clear_file(inode, FADVISE_LOST_PINO_BIT) 671 672 #define file_is_encrypt(inode) is_file(inode, FADVISE_ENCRYPT_BIT) 673 #define file_set_encrypt(inode) set_file(inode, FADVISE_ENCRYPT_BIT) 674 675 #define file_enc_name(inode) is_file(inode, FADVISE_ENC_NAME_BIT) 676 #define file_set_enc_name(inode) set_file(inode, FADVISE_ENC_NAME_BIT) 677 678 #define file_keep_isize(inode) is_file(inode, FADVISE_KEEP_SIZE_BIT) 679 #define file_set_keep_isize(inode) set_file(inode, FADVISE_KEEP_SIZE_BIT) 680 681 #define file_is_hot(inode) is_file(inode, FADVISE_HOT_BIT) 682 #define file_set_hot(inode) set_file(inode, FADVISE_HOT_BIT) 683 #define file_clear_hot(inode) clear_file(inode, FADVISE_HOT_BIT) 684 685 #define file_is_verity(inode) is_file(inode, FADVISE_VERITY_BIT) 686 #define file_set_verity(inode) set_file(inode, FADVISE_VERITY_BIT) 687 688 #define file_should_truncate(inode) is_file(inode, FADVISE_TRUNC_BIT) 689 #define file_need_truncate(inode) set_file(inode, FADVISE_TRUNC_BIT) 690 #define file_dont_truncate(inode) clear_file(inode, FADVISE_TRUNC_BIT) 691 692 #define DEF_DIR_LEVEL 0 693 694 enum { 695 GC_FAILURE_PIN, 696 GC_FAILURE_ATOMIC, 697 MAX_GC_FAILURE 698 }; 699 700 /* used for f2fs_inode_info->flags */ 701 enum { 702 FI_NEW_INODE, /* indicate newly allocated inode */ 703 FI_DIRTY_INODE, /* indicate inode is dirty or not */ 704 FI_AUTO_RECOVER, /* indicate inode is recoverable */ 705 FI_DIRTY_DIR, /* indicate directory has dirty pages */ 706 FI_INC_LINK, /* need to increment i_nlink */ 707 FI_ACL_MODE, /* indicate acl mode */ 708 FI_NO_ALLOC, /* should not allocate any blocks */ 709 FI_FREE_NID, /* free allocated nide */ 710 FI_NO_EXTENT, /* not to use the extent cache */ 711 FI_INLINE_XATTR, /* used for inline xattr */ 712 FI_INLINE_DATA, /* used for inline data*/ 713 FI_INLINE_DENTRY, /* used for inline dentry */ 714 FI_APPEND_WRITE, /* inode has appended data */ 715 FI_UPDATE_WRITE, /* inode has in-place-update data */ 716 FI_NEED_IPU, /* used for ipu per file */ 717 FI_ATOMIC_FILE, /* indicate atomic file */ 718 FI_ATOMIC_COMMIT, /* indicate the state of atomical committing */ 719 FI_VOLATILE_FILE, /* indicate volatile file */ 720 FI_FIRST_BLOCK_WRITTEN, /* indicate #0 data block was written */ 721 FI_DROP_CACHE, /* drop dirty page cache */ 722 FI_DATA_EXIST, /* indicate data exists */ 723 FI_INLINE_DOTS, /* indicate inline dot dentries */ 724 FI_DO_DEFRAG, /* indicate defragment is running */ 725 FI_DIRTY_FILE, /* indicate regular/symlink has dirty pages */ 726 FI_PREALLOCATED_ALL, /* all blocks for write were preallocated */ 727 FI_HOT_DATA, /* indicate file is hot */ 728 FI_EXTRA_ATTR, /* indicate file has extra attribute */ 729 FI_PROJ_INHERIT, /* indicate file inherits projectid */ 730 FI_PIN_FILE, /* indicate file should not be gced */ 731 FI_ATOMIC_REVOKE_REQUEST, /* request to drop atomic data */ 732 FI_VERITY_IN_PROGRESS, /* building fs-verity Merkle tree */ 733 FI_COMPRESSED_FILE, /* indicate file's data can be compressed */ 734 FI_COMPRESS_CORRUPT, /* indicate compressed cluster is corrupted */ 735 FI_MMAP_FILE, /* indicate file was mmapped */ 736 FI_ENABLE_COMPRESS, /* enable compression in "user" compression mode */ 737 FI_COMPRESS_RELEASED, /* compressed blocks were released */ 738 FI_ALIGNED_WRITE, /* enable aligned write */ 739 FI_MAX, /* max flag, never be used */ 740 }; 741 742 struct f2fs_inode_info { 743 struct inode vfs_inode; /* serve a vfs inode */ 744 unsigned long i_flags; /* keep an inode flags for ioctl */ 745 unsigned char i_advise; /* use to give file attribute hints */ 746 unsigned char i_dir_level; /* use for dentry level for large dir */ 747 unsigned int i_current_depth; /* only for directory depth */ 748 /* for gc failure statistic */ 749 unsigned int i_gc_failures[MAX_GC_FAILURE]; 750 unsigned int i_pino; /* parent inode number */ 751 umode_t i_acl_mode; /* keep file acl mode temporarily */ 752 753 /* Use below internally in f2fs*/ 754 unsigned long flags[BITS_TO_LONGS(FI_MAX)]; /* use to pass per-file flags */ 755 struct rw_semaphore i_sem; /* protect fi info */ 756 atomic_t dirty_pages; /* # of dirty pages */ 757 f2fs_hash_t chash; /* hash value of given file name */ 758 unsigned int clevel; /* maximum level of given file name */ 759 struct task_struct *task; /* lookup and create consistency */ 760 struct task_struct *cp_task; /* separate cp/wb IO stats*/ 761 nid_t i_xattr_nid; /* node id that contains xattrs */ 762 loff_t last_disk_size; /* lastly written file size */ 763 spinlock_t i_size_lock; /* protect last_disk_size */ 764 765 #ifdef CONFIG_QUOTA 766 struct dquot *i_dquot[MAXQUOTAS]; 767 768 /* quota space reservation, managed internally by quota code */ 769 qsize_t i_reserved_quota; 770 #endif 771 struct list_head dirty_list; /* dirty list for dirs and files */ 772 struct list_head gdirty_list; /* linked in global dirty list */ 773 struct list_head inmem_ilist; /* list for inmem inodes */ 774 struct list_head inmem_pages; /* inmemory pages managed by f2fs */ 775 struct task_struct *inmem_task; /* store inmemory task */ 776 struct mutex inmem_lock; /* lock for inmemory pages */ 777 struct extent_tree *extent_tree; /* cached extent_tree entry */ 778 779 /* avoid racing between foreground op and gc */ 780 struct rw_semaphore i_gc_rwsem[2]; 781 struct rw_semaphore i_xattr_sem; /* avoid racing between reading and changing EAs */ 782 783 int i_extra_isize; /* size of extra space located in i_addr */ 784 kprojid_t i_projid; /* id for project quota */ 785 int i_inline_xattr_size; /* inline xattr size */ 786 struct timespec64 i_crtime; /* inode creation time */ 787 struct timespec64 i_disk_time[4];/* inode disk times */ 788 789 /* for file compress */ 790 atomic_t i_compr_blocks; /* # of compressed blocks */ 791 unsigned char i_compress_algorithm; /* algorithm type */ 792 unsigned char i_log_cluster_size; /* log of cluster size */ 793 unsigned char i_compress_level; /* compress level (lz4hc,zstd) */ 794 unsigned short i_compress_flag; /* compress flag */ 795 unsigned int i_cluster_size; /* cluster size */ 796 }; 797 798 static inline void get_extent_info(struct extent_info *ext, 799 struct f2fs_extent *i_ext) 800 { 801 ext->fofs = le32_to_cpu(i_ext->fofs); 802 ext->blk = le32_to_cpu(i_ext->blk); 803 ext->len = le32_to_cpu(i_ext->len); 804 } 805 806 static inline void set_raw_extent(struct extent_info *ext, 807 struct f2fs_extent *i_ext) 808 { 809 i_ext->fofs = cpu_to_le32(ext->fofs); 810 i_ext->blk = cpu_to_le32(ext->blk); 811 i_ext->len = cpu_to_le32(ext->len); 812 } 813 814 static inline void set_extent_info(struct extent_info *ei, unsigned int fofs, 815 u32 blk, unsigned int len) 816 { 817 ei->fofs = fofs; 818 ei->blk = blk; 819 ei->len = len; 820 #ifdef CONFIG_F2FS_FS_COMPRESSION 821 ei->c_len = 0; 822 #endif 823 } 824 825 static inline bool __is_discard_mergeable(struct discard_info *back, 826 struct discard_info *front, unsigned int max_len) 827 { 828 return (back->lstart + back->len == front->lstart) && 829 (back->len + front->len <= max_len); 830 } 831 832 static inline bool __is_discard_back_mergeable(struct discard_info *cur, 833 struct discard_info *back, unsigned int max_len) 834 { 835 return __is_discard_mergeable(back, cur, max_len); 836 } 837 838 static inline bool __is_discard_front_mergeable(struct discard_info *cur, 839 struct discard_info *front, unsigned int max_len) 840 { 841 return __is_discard_mergeable(cur, front, max_len); 842 } 843 844 static inline bool __is_extent_mergeable(struct extent_info *back, 845 struct extent_info *front) 846 { 847 #ifdef CONFIG_F2FS_FS_COMPRESSION 848 if (back->c_len && back->len != back->c_len) 849 return false; 850 if (front->c_len && front->len != front->c_len) 851 return false; 852 #endif 853 return (back->fofs + back->len == front->fofs && 854 back->blk + back->len == front->blk); 855 } 856 857 static inline bool __is_back_mergeable(struct extent_info *cur, 858 struct extent_info *back) 859 { 860 return __is_extent_mergeable(back, cur); 861 } 862 863 static inline bool __is_front_mergeable(struct extent_info *cur, 864 struct extent_info *front) 865 { 866 return __is_extent_mergeable(cur, front); 867 } 868 869 extern void f2fs_mark_inode_dirty_sync(struct inode *inode, bool sync); 870 static inline void __try_update_largest_extent(struct extent_tree *et, 871 struct extent_node *en) 872 { 873 if (en->ei.len > et->largest.len) { 874 et->largest = en->ei; 875 et->largest_updated = true; 876 } 877 } 878 879 /* 880 * For free nid management 881 */ 882 enum nid_state { 883 FREE_NID, /* newly added to free nid list */ 884 PREALLOC_NID, /* it is preallocated */ 885 MAX_NID_STATE, 886 }; 887 888 enum nat_state { 889 TOTAL_NAT, 890 DIRTY_NAT, 891 RECLAIMABLE_NAT, 892 MAX_NAT_STATE, 893 }; 894 895 struct f2fs_nm_info { 896 block_t nat_blkaddr; /* base disk address of NAT */ 897 nid_t max_nid; /* maximum possible node ids */ 898 nid_t available_nids; /* # of available node ids */ 899 nid_t next_scan_nid; /* the next nid to be scanned */ 900 unsigned int ram_thresh; /* control the memory footprint */ 901 unsigned int ra_nid_pages; /* # of nid pages to be readaheaded */ 902 unsigned int dirty_nats_ratio; /* control dirty nats ratio threshold */ 903 904 /* NAT cache management */ 905 struct radix_tree_root nat_root;/* root of the nat entry cache */ 906 struct radix_tree_root nat_set_root;/* root of the nat set cache */ 907 struct rw_semaphore nat_tree_lock; /* protect nat entry tree */ 908 struct list_head nat_entries; /* cached nat entry list (clean) */ 909 spinlock_t nat_list_lock; /* protect clean nat entry list */ 910 unsigned int nat_cnt[MAX_NAT_STATE]; /* the # of cached nat entries */ 911 unsigned int nat_blocks; /* # of nat blocks */ 912 913 /* free node ids management */ 914 struct radix_tree_root free_nid_root;/* root of the free_nid cache */ 915 struct list_head free_nid_list; /* list for free nids excluding preallocated nids */ 916 unsigned int nid_cnt[MAX_NID_STATE]; /* the number of free node id */ 917 spinlock_t nid_list_lock; /* protect nid lists ops */ 918 struct mutex build_lock; /* lock for build free nids */ 919 unsigned char **free_nid_bitmap; 920 unsigned char *nat_block_bitmap; 921 unsigned short *free_nid_count; /* free nid count of NAT block */ 922 923 /* for checkpoint */ 924 char *nat_bitmap; /* NAT bitmap pointer */ 925 926 unsigned int nat_bits_blocks; /* # of nat bits blocks */ 927 unsigned char *nat_bits; /* NAT bits blocks */ 928 unsigned char *full_nat_bits; /* full NAT pages */ 929 unsigned char *empty_nat_bits; /* empty NAT pages */ 930 #ifdef CONFIG_F2FS_CHECK_FS 931 char *nat_bitmap_mir; /* NAT bitmap mirror */ 932 #endif 933 int bitmap_size; /* bitmap size */ 934 }; 935 936 /* 937 * this structure is used as one of function parameters. 938 * all the information are dedicated to a given direct node block determined 939 * by the data offset in a file. 940 */ 941 struct dnode_of_data { 942 struct inode *inode; /* vfs inode pointer */ 943 struct page *inode_page; /* its inode page, NULL is possible */ 944 struct page *node_page; /* cached direct node page */ 945 nid_t nid; /* node id of the direct node block */ 946 unsigned int ofs_in_node; /* data offset in the node page */ 947 bool inode_page_locked; /* inode page is locked or not */ 948 bool node_changed; /* is node block changed */ 949 char cur_level; /* level of hole node page */ 950 char max_level; /* level of current page located */ 951 block_t data_blkaddr; /* block address of the node block */ 952 }; 953 954 static inline void set_new_dnode(struct dnode_of_data *dn, struct inode *inode, 955 struct page *ipage, struct page *npage, nid_t nid) 956 { 957 memset(dn, 0, sizeof(*dn)); 958 dn->inode = inode; 959 dn->inode_page = ipage; 960 dn->node_page = npage; 961 dn->nid = nid; 962 } 963 964 /* 965 * For SIT manager 966 * 967 * By default, there are 6 active log areas across the whole main area. 968 * When considering hot and cold data separation to reduce cleaning overhead, 969 * we split 3 for data logs and 3 for node logs as hot, warm, and cold types, 970 * respectively. 971 * In the current design, you should not change the numbers intentionally. 972 * Instead, as a mount option such as active_logs=x, you can use 2, 4, and 6 973 * logs individually according to the underlying devices. (default: 6) 974 * Just in case, on-disk layout covers maximum 16 logs that consist of 8 for 975 * data and 8 for node logs. 976 */ 977 #define NR_CURSEG_DATA_TYPE (3) 978 #define NR_CURSEG_NODE_TYPE (3) 979 #define NR_CURSEG_INMEM_TYPE (2) 980 #define NR_CURSEG_RO_TYPE (2) 981 #define NR_CURSEG_PERSIST_TYPE (NR_CURSEG_DATA_TYPE + NR_CURSEG_NODE_TYPE) 982 #define NR_CURSEG_TYPE (NR_CURSEG_INMEM_TYPE + NR_CURSEG_PERSIST_TYPE) 983 984 enum { 985 CURSEG_HOT_DATA = 0, /* directory entry blocks */ 986 CURSEG_WARM_DATA, /* data blocks */ 987 CURSEG_COLD_DATA, /* multimedia or GCed data blocks */ 988 CURSEG_HOT_NODE, /* direct node blocks of directory files */ 989 CURSEG_WARM_NODE, /* direct node blocks of normal files */ 990 CURSEG_COLD_NODE, /* indirect node blocks */ 991 NR_PERSISTENT_LOG, /* number of persistent log */ 992 CURSEG_COLD_DATA_PINNED = NR_PERSISTENT_LOG, 993 /* pinned file that needs consecutive block address */ 994 CURSEG_ALL_DATA_ATGC, /* SSR alloctor in hot/warm/cold data area */ 995 NO_CHECK_TYPE, /* number of persistent & inmem log */ 996 }; 997 998 struct flush_cmd { 999 struct completion wait; 1000 struct llist_node llnode; 1001 nid_t ino; 1002 int ret; 1003 }; 1004 1005 struct flush_cmd_control { 1006 struct task_struct *f2fs_issue_flush; /* flush thread */ 1007 wait_queue_head_t flush_wait_queue; /* waiting queue for wake-up */ 1008 atomic_t issued_flush; /* # of issued flushes */ 1009 atomic_t queued_flush; /* # of queued flushes */ 1010 struct llist_head issue_list; /* list for command issue */ 1011 struct llist_node *dispatch_list; /* list for command dispatch */ 1012 }; 1013 1014 struct f2fs_sm_info { 1015 struct sit_info *sit_info; /* whole segment information */ 1016 struct free_segmap_info *free_info; /* free segment information */ 1017 struct dirty_seglist_info *dirty_info; /* dirty segment information */ 1018 struct curseg_info *curseg_array; /* active segment information */ 1019 1020 struct rw_semaphore curseg_lock; /* for preventing curseg change */ 1021 1022 block_t seg0_blkaddr; /* block address of 0'th segment */ 1023 block_t main_blkaddr; /* start block address of main area */ 1024 block_t ssa_blkaddr; /* start block address of SSA area */ 1025 1026 unsigned int segment_count; /* total # of segments */ 1027 unsigned int main_segments; /* # of segments in main area */ 1028 unsigned int reserved_segments; /* # of reserved segments */ 1029 unsigned int additional_reserved_segments;/* reserved segs for IO align feature */ 1030 unsigned int ovp_segments; /* # of overprovision segments */ 1031 1032 /* a threshold to reclaim prefree segments */ 1033 unsigned int rec_prefree_segments; 1034 1035 /* for batched trimming */ 1036 unsigned int trim_sections; /* # of sections to trim */ 1037 1038 struct list_head sit_entry_set; /* sit entry set list */ 1039 1040 unsigned int ipu_policy; /* in-place-update policy */ 1041 unsigned int min_ipu_util; /* in-place-update threshold */ 1042 unsigned int min_fsync_blocks; /* threshold for fsync */ 1043 unsigned int min_seq_blocks; /* threshold for sequential blocks */ 1044 unsigned int min_hot_blocks; /* threshold for hot block allocation */ 1045 unsigned int min_ssr_sections; /* threshold to trigger SSR allocation */ 1046 1047 /* for flush command control */ 1048 struct flush_cmd_control *fcc_info; 1049 1050 /* for discard command control */ 1051 struct discard_cmd_control *dcc_info; 1052 }; 1053 1054 /* 1055 * For superblock 1056 */ 1057 /* 1058 * COUNT_TYPE for monitoring 1059 * 1060 * f2fs monitors the number of several block types such as on-writeback, 1061 * dirty dentry blocks, dirty node blocks, and dirty meta blocks. 1062 */ 1063 #define WB_DATA_TYPE(p) (__is_cp_guaranteed(p) ? F2FS_WB_CP_DATA : F2FS_WB_DATA) 1064 enum count_type { 1065 F2FS_DIRTY_DENTS, 1066 F2FS_DIRTY_DATA, 1067 F2FS_DIRTY_QDATA, 1068 F2FS_DIRTY_NODES, 1069 F2FS_DIRTY_META, 1070 F2FS_INMEM_PAGES, 1071 F2FS_DIRTY_IMETA, 1072 F2FS_WB_CP_DATA, 1073 F2FS_WB_DATA, 1074 F2FS_RD_DATA, 1075 F2FS_RD_NODE, 1076 F2FS_RD_META, 1077 F2FS_DIO_WRITE, 1078 F2FS_DIO_READ, 1079 NR_COUNT_TYPE, 1080 }; 1081 1082 /* 1083 * The below are the page types of bios used in submit_bio(). 1084 * The available types are: 1085 * DATA User data pages. It operates as async mode. 1086 * NODE Node pages. It operates as async mode. 1087 * META FS metadata pages such as SIT, NAT, CP. 1088 * NR_PAGE_TYPE The number of page types. 1089 * META_FLUSH Make sure the previous pages are written 1090 * with waiting the bio's completion 1091 * ... Only can be used with META. 1092 */ 1093 #define PAGE_TYPE_OF_BIO(type) ((type) > META ? META : (type)) 1094 enum page_type { 1095 DATA, 1096 NODE, 1097 META, 1098 NR_PAGE_TYPE, 1099 META_FLUSH, 1100 INMEM, /* the below types are used by tracepoints only. */ 1101 INMEM_DROP, 1102 INMEM_INVALIDATE, 1103 INMEM_REVOKE, 1104 IPU, 1105 OPU, 1106 }; 1107 1108 enum temp_type { 1109 HOT = 0, /* must be zero for meta bio */ 1110 WARM, 1111 COLD, 1112 NR_TEMP_TYPE, 1113 }; 1114 1115 enum need_lock_type { 1116 LOCK_REQ = 0, 1117 LOCK_DONE, 1118 LOCK_RETRY, 1119 }; 1120 1121 enum cp_reason_type { 1122 CP_NO_NEEDED, 1123 CP_NON_REGULAR, 1124 CP_COMPRESSED, 1125 CP_HARDLINK, 1126 CP_SB_NEED_CP, 1127 CP_WRONG_PINO, 1128 CP_NO_SPC_ROLL, 1129 CP_NODE_NEED_CP, 1130 CP_FASTBOOT_MODE, 1131 CP_SPEC_LOG_NUM, 1132 CP_RECOVER_DIR, 1133 }; 1134 1135 enum iostat_type { 1136 /* WRITE IO */ 1137 APP_DIRECT_IO, /* app direct write IOs */ 1138 APP_BUFFERED_IO, /* app buffered write IOs */ 1139 APP_WRITE_IO, /* app write IOs */ 1140 APP_MAPPED_IO, /* app mapped IOs */ 1141 FS_DATA_IO, /* data IOs from kworker/fsync/reclaimer */ 1142 FS_NODE_IO, /* node IOs from kworker/fsync/reclaimer */ 1143 FS_META_IO, /* meta IOs from kworker/reclaimer */ 1144 FS_GC_DATA_IO, /* data IOs from forground gc */ 1145 FS_GC_NODE_IO, /* node IOs from forground gc */ 1146 FS_CP_DATA_IO, /* data IOs from checkpoint */ 1147 FS_CP_NODE_IO, /* node IOs from checkpoint */ 1148 FS_CP_META_IO, /* meta IOs from checkpoint */ 1149 1150 /* READ IO */ 1151 APP_DIRECT_READ_IO, /* app direct read IOs */ 1152 APP_BUFFERED_READ_IO, /* app buffered read IOs */ 1153 APP_READ_IO, /* app read IOs */ 1154 APP_MAPPED_READ_IO, /* app mapped read IOs */ 1155 FS_DATA_READ_IO, /* data read IOs */ 1156 FS_GDATA_READ_IO, /* data read IOs from background gc */ 1157 FS_CDATA_READ_IO, /* compressed data read IOs */ 1158 FS_NODE_READ_IO, /* node read IOs */ 1159 FS_META_READ_IO, /* meta read IOs */ 1160 1161 /* other */ 1162 FS_DISCARD, /* discard */ 1163 NR_IO_TYPE, 1164 }; 1165 1166 struct f2fs_io_info { 1167 struct f2fs_sb_info *sbi; /* f2fs_sb_info pointer */ 1168 nid_t ino; /* inode number */ 1169 enum page_type type; /* contains DATA/NODE/META/META_FLUSH */ 1170 enum temp_type temp; /* contains HOT/WARM/COLD */ 1171 int op; /* contains REQ_OP_ */ 1172 int op_flags; /* req_flag_bits */ 1173 block_t new_blkaddr; /* new block address to be written */ 1174 block_t old_blkaddr; /* old block address before Cow */ 1175 struct page *page; /* page to be written */ 1176 struct page *encrypted_page; /* encrypted page */ 1177 struct page *compressed_page; /* compressed page */ 1178 struct list_head list; /* serialize IOs */ 1179 bool submitted; /* indicate IO submission */ 1180 int need_lock; /* indicate we need to lock cp_rwsem */ 1181 bool in_list; /* indicate fio is in io_list */ 1182 bool is_por; /* indicate IO is from recovery or not */ 1183 bool retry; /* need to reallocate block address */ 1184 int compr_blocks; /* # of compressed block addresses */ 1185 bool encrypted; /* indicate file is encrypted */ 1186 enum iostat_type io_type; /* io type */ 1187 struct writeback_control *io_wbc; /* writeback control */ 1188 struct bio **bio; /* bio for ipu */ 1189 sector_t *last_block; /* last block number in bio */ 1190 unsigned char version; /* version of the node */ 1191 }; 1192 1193 struct bio_entry { 1194 struct bio *bio; 1195 struct list_head list; 1196 }; 1197 1198 #define is_read_io(rw) ((rw) == READ) 1199 struct f2fs_bio_info { 1200 struct f2fs_sb_info *sbi; /* f2fs superblock */ 1201 struct bio *bio; /* bios to merge */ 1202 sector_t last_block_in_bio; /* last block number */ 1203 struct f2fs_io_info fio; /* store buffered io info. */ 1204 struct rw_semaphore io_rwsem; /* blocking op for bio */ 1205 spinlock_t io_lock; /* serialize DATA/NODE IOs */ 1206 struct list_head io_list; /* track fios */ 1207 struct list_head bio_list; /* bio entry list head */ 1208 struct rw_semaphore bio_list_lock; /* lock to protect bio entry list */ 1209 }; 1210 1211 #define FDEV(i) (sbi->devs[i]) 1212 #define RDEV(i) (raw_super->devs[i]) 1213 struct f2fs_dev_info { 1214 struct block_device *bdev; 1215 char path[MAX_PATH_LEN]; 1216 unsigned int total_segments; 1217 block_t start_blk; 1218 block_t end_blk; 1219 #ifdef CONFIG_BLK_DEV_ZONED 1220 unsigned int nr_blkz; /* Total number of zones */ 1221 unsigned long *blkz_seq; /* Bitmap indicating sequential zones */ 1222 block_t *zone_capacity_blocks; /* Array of zone capacity in blks */ 1223 #endif 1224 }; 1225 1226 enum inode_type { 1227 DIR_INODE, /* for dirty dir inode */ 1228 FILE_INODE, /* for dirty regular/symlink inode */ 1229 DIRTY_META, /* for all dirtied inode metadata */ 1230 ATOMIC_FILE, /* for all atomic files */ 1231 NR_INODE_TYPE, 1232 }; 1233 1234 /* for inner inode cache management */ 1235 struct inode_management { 1236 struct radix_tree_root ino_root; /* ino entry array */ 1237 spinlock_t ino_lock; /* for ino entry lock */ 1238 struct list_head ino_list; /* inode list head */ 1239 unsigned long ino_num; /* number of entries */ 1240 }; 1241 1242 /* for GC_AT */ 1243 struct atgc_management { 1244 bool atgc_enabled; /* ATGC is enabled or not */ 1245 struct rb_root_cached root; /* root of victim rb-tree */ 1246 struct list_head victim_list; /* linked with all victim entries */ 1247 unsigned int victim_count; /* victim count in rb-tree */ 1248 unsigned int candidate_ratio; /* candidate ratio */ 1249 unsigned int max_candidate_count; /* max candidate count */ 1250 unsigned int age_weight; /* age weight, vblock_weight = 100 - age_weight */ 1251 unsigned long long age_threshold; /* age threshold */ 1252 }; 1253 1254 /* For s_flag in struct f2fs_sb_info */ 1255 enum { 1256 SBI_IS_DIRTY, /* dirty flag for checkpoint */ 1257 SBI_IS_CLOSE, /* specify unmounting */ 1258 SBI_NEED_FSCK, /* need fsck.f2fs to fix */ 1259 SBI_POR_DOING, /* recovery is doing or not */ 1260 SBI_NEED_SB_WRITE, /* need to recover superblock */ 1261 SBI_NEED_CP, /* need to checkpoint */ 1262 SBI_IS_SHUTDOWN, /* shutdown by ioctl */ 1263 SBI_IS_RECOVERED, /* recovered orphan/data */ 1264 SBI_CP_DISABLED, /* CP was disabled last mount */ 1265 SBI_CP_DISABLED_QUICK, /* CP was disabled quickly */ 1266 SBI_QUOTA_NEED_FLUSH, /* need to flush quota info in CP */ 1267 SBI_QUOTA_SKIP_FLUSH, /* skip flushing quota in current CP */ 1268 SBI_QUOTA_NEED_REPAIR, /* quota file may be corrupted */ 1269 SBI_IS_RESIZEFS, /* resizefs is in process */ 1270 }; 1271 1272 enum { 1273 CP_TIME, 1274 REQ_TIME, 1275 DISCARD_TIME, 1276 GC_TIME, 1277 DISABLE_TIME, 1278 UMOUNT_DISCARD_TIMEOUT, 1279 MAX_TIME, 1280 }; 1281 1282 enum { 1283 GC_NORMAL, 1284 GC_IDLE_CB, 1285 GC_IDLE_GREEDY, 1286 GC_IDLE_AT, 1287 GC_URGENT_HIGH, 1288 GC_URGENT_LOW, 1289 MAX_GC_MODE, 1290 }; 1291 1292 enum { 1293 BGGC_MODE_ON, /* background gc is on */ 1294 BGGC_MODE_OFF, /* background gc is off */ 1295 BGGC_MODE_SYNC, /* 1296 * background gc is on, migrating blocks 1297 * like foreground gc 1298 */ 1299 }; 1300 1301 enum { 1302 FS_MODE_ADAPTIVE, /* use both lfs/ssr allocation */ 1303 FS_MODE_LFS, /* use lfs allocation only */ 1304 FS_MODE_FRAGMENT_SEG, /* segment fragmentation mode */ 1305 FS_MODE_FRAGMENT_BLK, /* block fragmentation mode */ 1306 }; 1307 1308 enum { 1309 WHINT_MODE_OFF, /* not pass down write hints */ 1310 WHINT_MODE_USER, /* try to pass down hints given by users */ 1311 WHINT_MODE_FS, /* pass down hints with F2FS policy */ 1312 }; 1313 1314 enum { 1315 ALLOC_MODE_DEFAULT, /* stay default */ 1316 ALLOC_MODE_REUSE, /* reuse segments as much as possible */ 1317 }; 1318 1319 enum fsync_mode { 1320 FSYNC_MODE_POSIX, /* fsync follows posix semantics */ 1321 FSYNC_MODE_STRICT, /* fsync behaves in line with ext4 */ 1322 FSYNC_MODE_NOBARRIER, /* fsync behaves nobarrier based on posix */ 1323 }; 1324 1325 enum { 1326 COMPR_MODE_FS, /* 1327 * automatically compress compression 1328 * enabled files 1329 */ 1330 COMPR_MODE_USER, /* 1331 * automatical compression is disabled. 1332 * user can control the file compression 1333 * using ioctls 1334 */ 1335 }; 1336 1337 enum { 1338 DISCARD_UNIT_BLOCK, /* basic discard unit is block */ 1339 DISCARD_UNIT_SEGMENT, /* basic discard unit is segment */ 1340 DISCARD_UNIT_SECTION, /* basic discard unit is section */ 1341 }; 1342 1343 static inline int f2fs_test_bit(unsigned int nr, char *addr); 1344 static inline void f2fs_set_bit(unsigned int nr, char *addr); 1345 static inline void f2fs_clear_bit(unsigned int nr, char *addr); 1346 1347 /* 1348 * Layout of f2fs page.private: 1349 * 1350 * Layout A: lowest bit should be 1 1351 * | bit0 = 1 | bit1 | bit2 | ... | bit MAX | private data .... | 1352 * bit 0 PAGE_PRIVATE_NOT_POINTER 1353 * bit 1 PAGE_PRIVATE_ATOMIC_WRITE 1354 * bit 2 PAGE_PRIVATE_DUMMY_WRITE 1355 * bit 3 PAGE_PRIVATE_ONGOING_MIGRATION 1356 * bit 4 PAGE_PRIVATE_INLINE_INODE 1357 * bit 5 PAGE_PRIVATE_REF_RESOURCE 1358 * bit 6- f2fs private data 1359 * 1360 * Layout B: lowest bit should be 0 1361 * page.private is a wrapped pointer. 1362 */ 1363 enum { 1364 PAGE_PRIVATE_NOT_POINTER, /* private contains non-pointer data */ 1365 PAGE_PRIVATE_ATOMIC_WRITE, /* data page from atomic write path */ 1366 PAGE_PRIVATE_DUMMY_WRITE, /* data page for padding aligned IO */ 1367 PAGE_PRIVATE_ONGOING_MIGRATION, /* data page which is on-going migrating */ 1368 PAGE_PRIVATE_INLINE_INODE, /* inode page contains inline data */ 1369 PAGE_PRIVATE_REF_RESOURCE, /* dirty page has referenced resources */ 1370 PAGE_PRIVATE_MAX 1371 }; 1372 1373 #define PAGE_PRIVATE_GET_FUNC(name, flagname) \ 1374 static inline bool page_private_##name(struct page *page) \ 1375 { \ 1376 return PagePrivate(page) && \ 1377 test_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)) && \ 1378 test_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \ 1379 } 1380 1381 #define PAGE_PRIVATE_SET_FUNC(name, flagname) \ 1382 static inline void set_page_private_##name(struct page *page) \ 1383 { \ 1384 if (!PagePrivate(page)) { \ 1385 get_page(page); \ 1386 SetPagePrivate(page); \ 1387 set_page_private(page, 0); \ 1388 } \ 1389 set_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)); \ 1390 set_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \ 1391 } 1392 1393 #define PAGE_PRIVATE_CLEAR_FUNC(name, flagname) \ 1394 static inline void clear_page_private_##name(struct page *page) \ 1395 { \ 1396 clear_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \ 1397 if (page_private(page) == 1 << PAGE_PRIVATE_NOT_POINTER) { \ 1398 set_page_private(page, 0); \ 1399 if (PagePrivate(page)) { \ 1400 ClearPagePrivate(page); \ 1401 put_page(page); \ 1402 }\ 1403 } \ 1404 } 1405 1406 PAGE_PRIVATE_GET_FUNC(nonpointer, NOT_POINTER); 1407 PAGE_PRIVATE_GET_FUNC(reference, REF_RESOURCE); 1408 PAGE_PRIVATE_GET_FUNC(inline, INLINE_INODE); 1409 PAGE_PRIVATE_GET_FUNC(gcing, ONGOING_MIGRATION); 1410 PAGE_PRIVATE_GET_FUNC(atomic, ATOMIC_WRITE); 1411 PAGE_PRIVATE_GET_FUNC(dummy, DUMMY_WRITE); 1412 1413 PAGE_PRIVATE_SET_FUNC(reference, REF_RESOURCE); 1414 PAGE_PRIVATE_SET_FUNC(inline, INLINE_INODE); 1415 PAGE_PRIVATE_SET_FUNC(gcing, ONGOING_MIGRATION); 1416 PAGE_PRIVATE_SET_FUNC(atomic, ATOMIC_WRITE); 1417 PAGE_PRIVATE_SET_FUNC(dummy, DUMMY_WRITE); 1418 1419 PAGE_PRIVATE_CLEAR_FUNC(reference, REF_RESOURCE); 1420 PAGE_PRIVATE_CLEAR_FUNC(inline, INLINE_INODE); 1421 PAGE_PRIVATE_CLEAR_FUNC(gcing, ONGOING_MIGRATION); 1422 PAGE_PRIVATE_CLEAR_FUNC(atomic, ATOMIC_WRITE); 1423 PAGE_PRIVATE_CLEAR_FUNC(dummy, DUMMY_WRITE); 1424 1425 static inline unsigned long get_page_private_data(struct page *page) 1426 { 1427 unsigned long data = page_private(page); 1428 1429 if (!test_bit(PAGE_PRIVATE_NOT_POINTER, &data)) 1430 return 0; 1431 return data >> PAGE_PRIVATE_MAX; 1432 } 1433 1434 static inline void set_page_private_data(struct page *page, unsigned long data) 1435 { 1436 if (!PagePrivate(page)) { 1437 get_page(page); 1438 SetPagePrivate(page); 1439 set_page_private(page, 0); 1440 } 1441 set_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)); 1442 page_private(page) |= data << PAGE_PRIVATE_MAX; 1443 } 1444 1445 static inline void clear_page_private_data(struct page *page) 1446 { 1447 page_private(page) &= (1 << PAGE_PRIVATE_MAX) - 1; 1448 if (page_private(page) == 1 << PAGE_PRIVATE_NOT_POINTER) { 1449 set_page_private(page, 0); 1450 if (PagePrivate(page)) { 1451 ClearPagePrivate(page); 1452 put_page(page); 1453 } 1454 } 1455 } 1456 1457 /* For compression */ 1458 enum compress_algorithm_type { 1459 COMPRESS_LZO, 1460 COMPRESS_LZ4, 1461 COMPRESS_ZSTD, 1462 COMPRESS_LZORLE, 1463 COMPRESS_MAX, 1464 }; 1465 1466 enum compress_flag { 1467 COMPRESS_CHKSUM, 1468 COMPRESS_MAX_FLAG, 1469 }; 1470 1471 #define COMPRESS_WATERMARK 20 1472 #define COMPRESS_PERCENT 20 1473 1474 #define COMPRESS_DATA_RESERVED_SIZE 4 1475 struct compress_data { 1476 __le32 clen; /* compressed data size */ 1477 __le32 chksum; /* compressed data chksum */ 1478 __le32 reserved[COMPRESS_DATA_RESERVED_SIZE]; /* reserved */ 1479 u8 cdata[]; /* compressed data */ 1480 }; 1481 1482 #define COMPRESS_HEADER_SIZE (sizeof(struct compress_data)) 1483 1484 #define F2FS_COMPRESSED_PAGE_MAGIC 0xF5F2C000 1485 1486 #define COMPRESS_LEVEL_OFFSET 8 1487 1488 /* compress context */ 1489 struct compress_ctx { 1490 struct inode *inode; /* inode the context belong to */ 1491 pgoff_t cluster_idx; /* cluster index number */ 1492 unsigned int cluster_size; /* page count in cluster */ 1493 unsigned int log_cluster_size; /* log of cluster size */ 1494 struct page **rpages; /* pages store raw data in cluster */ 1495 unsigned int nr_rpages; /* total page number in rpages */ 1496 struct page **cpages; /* pages store compressed data in cluster */ 1497 unsigned int nr_cpages; /* total page number in cpages */ 1498 unsigned int valid_nr_cpages; /* valid page number in cpages */ 1499 void *rbuf; /* virtual mapped address on rpages */ 1500 struct compress_data *cbuf; /* virtual mapped address on cpages */ 1501 size_t rlen; /* valid data length in rbuf */ 1502 size_t clen; /* valid data length in cbuf */ 1503 void *private; /* payload buffer for specified compression algorithm */ 1504 void *private2; /* extra payload buffer */ 1505 }; 1506 1507 /* compress context for write IO path */ 1508 struct compress_io_ctx { 1509 u32 magic; /* magic number to indicate page is compressed */ 1510 struct inode *inode; /* inode the context belong to */ 1511 struct page **rpages; /* pages store raw data in cluster */ 1512 unsigned int nr_rpages; /* total page number in rpages */ 1513 atomic_t pending_pages; /* in-flight compressed page count */ 1514 }; 1515 1516 /* Context for decompressing one cluster on the read IO path */ 1517 struct decompress_io_ctx { 1518 u32 magic; /* magic number to indicate page is compressed */ 1519 struct inode *inode; /* inode the context belong to */ 1520 pgoff_t cluster_idx; /* cluster index number */ 1521 unsigned int cluster_size; /* page count in cluster */ 1522 unsigned int log_cluster_size; /* log of cluster size */ 1523 struct page **rpages; /* pages store raw data in cluster */ 1524 unsigned int nr_rpages; /* total page number in rpages */ 1525 struct page **cpages; /* pages store compressed data in cluster */ 1526 unsigned int nr_cpages; /* total page number in cpages */ 1527 struct page **tpages; /* temp pages to pad holes in cluster */ 1528 void *rbuf; /* virtual mapped address on rpages */ 1529 struct compress_data *cbuf; /* virtual mapped address on cpages */ 1530 size_t rlen; /* valid data length in rbuf */ 1531 size_t clen; /* valid data length in cbuf */ 1532 1533 /* 1534 * The number of compressed pages remaining to be read in this cluster. 1535 * This is initially nr_cpages. It is decremented by 1 each time a page 1536 * has been read (or failed to be read). When it reaches 0, the cluster 1537 * is decompressed (or an error is reported). 1538 * 1539 * If an error occurs before all the pages have been submitted for I/O, 1540 * then this will never reach 0. In this case the I/O submitter is 1541 * responsible for calling f2fs_decompress_end_io() instead. 1542 */ 1543 atomic_t remaining_pages; 1544 1545 /* 1546 * Number of references to this decompress_io_ctx. 1547 * 1548 * One reference is held for I/O completion. This reference is dropped 1549 * after the pagecache pages are updated and unlocked -- either after 1550 * decompression (and verity if enabled), or after an error. 1551 * 1552 * In addition, each compressed page holds a reference while it is in a 1553 * bio. These references are necessary prevent compressed pages from 1554 * being freed while they are still in a bio. 1555 */ 1556 refcount_t refcnt; 1557 1558 bool failed; /* IO error occurred before decompression? */ 1559 bool need_verity; /* need fs-verity verification after decompression? */ 1560 void *private; /* payload buffer for specified decompression algorithm */ 1561 void *private2; /* extra payload buffer */ 1562 struct work_struct verity_work; /* work to verify the decompressed pages */ 1563 }; 1564 1565 #define NULL_CLUSTER ((unsigned int)(~0)) 1566 #define MIN_COMPRESS_LOG_SIZE 2 1567 #define MAX_COMPRESS_LOG_SIZE 8 1568 #define MAX_COMPRESS_WINDOW_SIZE(log_size) ((PAGE_SIZE) << (log_size)) 1569 1570 struct f2fs_sb_info { 1571 struct super_block *sb; /* pointer to VFS super block */ 1572 struct proc_dir_entry *s_proc; /* proc entry */ 1573 struct f2fs_super_block *raw_super; /* raw super block pointer */ 1574 struct rw_semaphore sb_lock; /* lock for raw super block */ 1575 int valid_super_block; /* valid super block no */ 1576 unsigned long s_flag; /* flags for sbi */ 1577 struct mutex writepages; /* mutex for writepages() */ 1578 1579 #ifdef CONFIG_BLK_DEV_ZONED 1580 unsigned int blocks_per_blkz; /* F2FS blocks per zone */ 1581 unsigned int log_blocks_per_blkz; /* log2 F2FS blocks per zone */ 1582 #endif 1583 1584 /* for node-related operations */ 1585 struct f2fs_nm_info *nm_info; /* node manager */ 1586 struct inode *node_inode; /* cache node blocks */ 1587 1588 /* for segment-related operations */ 1589 struct f2fs_sm_info *sm_info; /* segment manager */ 1590 1591 /* for bio operations */ 1592 struct f2fs_bio_info *write_io[NR_PAGE_TYPE]; /* for write bios */ 1593 /* keep migration IO order for LFS mode */ 1594 struct rw_semaphore io_order_lock; 1595 mempool_t *write_io_dummy; /* Dummy pages */ 1596 1597 /* for checkpoint */ 1598 struct f2fs_checkpoint *ckpt; /* raw checkpoint pointer */ 1599 int cur_cp_pack; /* remain current cp pack */ 1600 spinlock_t cp_lock; /* for flag in ckpt */ 1601 struct inode *meta_inode; /* cache meta blocks */ 1602 struct rw_semaphore cp_global_sem; /* checkpoint procedure lock */ 1603 struct rw_semaphore cp_rwsem; /* blocking FS operations */ 1604 struct rw_semaphore node_write; /* locking node writes */ 1605 struct rw_semaphore node_change; /* locking node change */ 1606 wait_queue_head_t cp_wait; 1607 unsigned long last_time[MAX_TIME]; /* to store time in jiffies */ 1608 long interval_time[MAX_TIME]; /* to store thresholds */ 1609 struct ckpt_req_control cprc_info; /* for checkpoint request control */ 1610 1611 struct inode_management im[MAX_INO_ENTRY]; /* manage inode cache */ 1612 1613 spinlock_t fsync_node_lock; /* for node entry lock */ 1614 struct list_head fsync_node_list; /* node list head */ 1615 unsigned int fsync_seg_id; /* sequence id */ 1616 unsigned int fsync_node_num; /* number of node entries */ 1617 1618 /* for orphan inode, use 0'th array */ 1619 unsigned int max_orphans; /* max orphan inodes */ 1620 1621 /* for inode management */ 1622 struct list_head inode_list[NR_INODE_TYPE]; /* dirty inode list */ 1623 spinlock_t inode_lock[NR_INODE_TYPE]; /* for dirty inode list lock */ 1624 struct mutex flush_lock; /* for flush exclusion */ 1625 1626 /* for extent tree cache */ 1627 struct radix_tree_root extent_tree_root;/* cache extent cache entries */ 1628 struct mutex extent_tree_lock; /* locking extent radix tree */ 1629 struct list_head extent_list; /* lru list for shrinker */ 1630 spinlock_t extent_lock; /* locking extent lru list */ 1631 atomic_t total_ext_tree; /* extent tree count */ 1632 struct list_head zombie_list; /* extent zombie tree list */ 1633 atomic_t total_zombie_tree; /* extent zombie tree count */ 1634 atomic_t total_ext_node; /* extent info count */ 1635 1636 /* basic filesystem units */ 1637 unsigned int log_sectors_per_block; /* log2 sectors per block */ 1638 unsigned int log_blocksize; /* log2 block size */ 1639 unsigned int blocksize; /* block size */ 1640 unsigned int root_ino_num; /* root inode number*/ 1641 unsigned int node_ino_num; /* node inode number*/ 1642 unsigned int meta_ino_num; /* meta inode number*/ 1643 unsigned int log_blocks_per_seg; /* log2 blocks per segment */ 1644 unsigned int blocks_per_seg; /* blocks per segment */ 1645 unsigned int segs_per_sec; /* segments per section */ 1646 unsigned int secs_per_zone; /* sections per zone */ 1647 unsigned int total_sections; /* total section count */ 1648 unsigned int total_node_count; /* total node block count */ 1649 unsigned int total_valid_node_count; /* valid node block count */ 1650 int dir_level; /* directory level */ 1651 int readdir_ra; /* readahead inode in readdir */ 1652 u64 max_io_bytes; /* max io bytes to merge IOs */ 1653 1654 block_t user_block_count; /* # of user blocks */ 1655 block_t total_valid_block_count; /* # of valid blocks */ 1656 block_t discard_blks; /* discard command candidats */ 1657 block_t last_valid_block_count; /* for recovery */ 1658 block_t reserved_blocks; /* configurable reserved blocks */ 1659 block_t current_reserved_blocks; /* current reserved blocks */ 1660 1661 /* Additional tracking for no checkpoint mode */ 1662 block_t unusable_block_count; /* # of blocks saved by last cp */ 1663 1664 unsigned int nquota_files; /* # of quota sysfile */ 1665 struct rw_semaphore quota_sem; /* blocking cp for flags */ 1666 1667 /* # of pages, see count_type */ 1668 atomic_t nr_pages[NR_COUNT_TYPE]; 1669 /* # of allocated blocks */ 1670 struct percpu_counter alloc_valid_block_count; 1671 1672 /* writeback control */ 1673 atomic_t wb_sync_req[META]; /* count # of WB_SYNC threads */ 1674 1675 /* valid inode count */ 1676 struct percpu_counter total_valid_inode_count; 1677 1678 struct f2fs_mount_info mount_opt; /* mount options */ 1679 1680 /* for cleaning operations */ 1681 struct rw_semaphore gc_lock; /* 1682 * semaphore for GC, avoid 1683 * race between GC and GC or CP 1684 */ 1685 struct f2fs_gc_kthread *gc_thread; /* GC thread */ 1686 struct atgc_management am; /* atgc management */ 1687 unsigned int cur_victim_sec; /* current victim section num */ 1688 unsigned int gc_mode; /* current GC state */ 1689 unsigned int next_victim_seg[2]; /* next segment in victim section */ 1690 spinlock_t gc_urgent_high_lock; 1691 bool gc_urgent_high_limited; /* indicates having limited trial count */ 1692 unsigned int gc_urgent_high_remaining; /* remaining trial count for GC_URGENT_HIGH */ 1693 1694 /* for skip statistic */ 1695 unsigned int atomic_files; /* # of opened atomic file */ 1696 unsigned long long skipped_atomic_files[2]; /* FG_GC and BG_GC */ 1697 unsigned long long skipped_gc_rwsem; /* FG_GC only */ 1698 1699 /* threshold for gc trials on pinned files */ 1700 u64 gc_pin_file_threshold; 1701 struct rw_semaphore pin_sem; 1702 1703 /* maximum # of trials to find a victim segment for SSR and GC */ 1704 unsigned int max_victim_search; 1705 /* migration granularity of garbage collection, unit: segment */ 1706 unsigned int migration_granularity; 1707 1708 /* 1709 * for stat information. 1710 * one is for the LFS mode, and the other is for the SSR mode. 1711 */ 1712 #ifdef CONFIG_F2FS_STAT_FS 1713 struct f2fs_stat_info *stat_info; /* FS status information */ 1714 atomic_t meta_count[META_MAX]; /* # of meta blocks */ 1715 unsigned int segment_count[2]; /* # of allocated segments */ 1716 unsigned int block_count[2]; /* # of allocated blocks */ 1717 atomic_t inplace_count; /* # of inplace update */ 1718 atomic64_t total_hit_ext; /* # of lookup extent cache */ 1719 atomic64_t read_hit_rbtree; /* # of hit rbtree extent node */ 1720 atomic64_t read_hit_largest; /* # of hit largest extent node */ 1721 atomic64_t read_hit_cached; /* # of hit cached extent node */ 1722 atomic_t inline_xattr; /* # of inline_xattr inodes */ 1723 atomic_t inline_inode; /* # of inline_data inodes */ 1724 atomic_t inline_dir; /* # of inline_dentry inodes */ 1725 atomic_t compr_inode; /* # of compressed inodes */ 1726 atomic64_t compr_blocks; /* # of compressed blocks */ 1727 atomic_t vw_cnt; /* # of volatile writes */ 1728 atomic_t max_aw_cnt; /* max # of atomic writes */ 1729 atomic_t max_vw_cnt; /* max # of volatile writes */ 1730 unsigned int io_skip_bggc; /* skip background gc for in-flight IO */ 1731 unsigned int other_skip_bggc; /* skip background gc for other reasons */ 1732 unsigned int ndirty_inode[NR_INODE_TYPE]; /* # of dirty inodes */ 1733 #endif 1734 spinlock_t stat_lock; /* lock for stat operations */ 1735 1736 /* to attach REQ_META|REQ_FUA flags */ 1737 unsigned int data_io_flag; 1738 unsigned int node_io_flag; 1739 1740 /* For sysfs suppport */ 1741 struct kobject s_kobj; /* /sys/fs/f2fs/<devname> */ 1742 struct completion s_kobj_unregister; 1743 1744 struct kobject s_stat_kobj; /* /sys/fs/f2fs/<devname>/stat */ 1745 struct completion s_stat_kobj_unregister; 1746 1747 struct kobject s_feature_list_kobj; /* /sys/fs/f2fs/<devname>/feature_list */ 1748 struct completion s_feature_list_kobj_unregister; 1749 1750 /* For shrinker support */ 1751 struct list_head s_list; 1752 struct mutex umount_mutex; 1753 unsigned int shrinker_run_no; 1754 1755 /* For multi devices */ 1756 int s_ndevs; /* number of devices */ 1757 struct f2fs_dev_info *devs; /* for device list */ 1758 unsigned int dirty_device; /* for checkpoint data flush */ 1759 spinlock_t dev_lock; /* protect dirty_device */ 1760 bool aligned_blksize; /* all devices has the same logical blksize */ 1761 1762 /* For write statistics */ 1763 u64 sectors_written_start; 1764 u64 kbytes_written; 1765 1766 /* Reference to checksum algorithm driver via cryptoapi */ 1767 struct crypto_shash *s_chksum_driver; 1768 1769 /* Precomputed FS UUID checksum for seeding other checksums */ 1770 __u32 s_chksum_seed; 1771 1772 struct workqueue_struct *post_read_wq; /* post read workqueue */ 1773 1774 struct kmem_cache *inline_xattr_slab; /* inline xattr entry */ 1775 unsigned int inline_xattr_slab_size; /* default inline xattr slab size */ 1776 1777 /* For reclaimed segs statistics per each GC mode */ 1778 unsigned int gc_segment_mode; /* GC state for reclaimed segments */ 1779 unsigned int gc_reclaimed_segs[MAX_GC_MODE]; /* Reclaimed segs for each mode */ 1780 1781 unsigned long seq_file_ra_mul; /* multiplier for ra_pages of seq. files in fadvise */ 1782 1783 int max_fragment_chunk; /* max chunk size for block fragmentation mode */ 1784 int max_fragment_hole; /* max hole size for block fragmentation mode */ 1785 1786 #ifdef CONFIG_F2FS_FS_COMPRESSION 1787 struct kmem_cache *page_array_slab; /* page array entry */ 1788 unsigned int page_array_slab_size; /* default page array slab size */ 1789 1790 /* For runtime compression statistics */ 1791 u64 compr_written_block; 1792 u64 compr_saved_block; 1793 u32 compr_new_inode; 1794 1795 /* For compressed block cache */ 1796 struct inode *compress_inode; /* cache compressed blocks */ 1797 unsigned int compress_percent; /* cache page percentage */ 1798 unsigned int compress_watermark; /* cache page watermark */ 1799 atomic_t compress_page_hit; /* cache hit count */ 1800 #endif 1801 1802 #ifdef CONFIG_F2FS_IOSTAT 1803 /* For app/fs IO statistics */ 1804 spinlock_t iostat_lock; 1805 unsigned long long rw_iostat[NR_IO_TYPE]; 1806 unsigned long long prev_rw_iostat[NR_IO_TYPE]; 1807 bool iostat_enable; 1808 unsigned long iostat_next_period; 1809 unsigned int iostat_period_ms; 1810 1811 /* For io latency related statistics info in one iostat period */ 1812 spinlock_t iostat_lat_lock; 1813 struct iostat_lat_info *iostat_io_lat; 1814 #endif 1815 }; 1816 1817 #ifdef CONFIG_F2FS_FAULT_INJECTION 1818 #define f2fs_show_injection_info(sbi, type) \ 1819 printk_ratelimited("%sF2FS-fs (%s) : inject %s in %s of %pS\n", \ 1820 KERN_INFO, sbi->sb->s_id, \ 1821 f2fs_fault_name[type], \ 1822 __func__, __builtin_return_address(0)) 1823 static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type) 1824 { 1825 struct f2fs_fault_info *ffi = &F2FS_OPTION(sbi).fault_info; 1826 1827 if (!ffi->inject_rate) 1828 return false; 1829 1830 if (!IS_FAULT_SET(ffi, type)) 1831 return false; 1832 1833 atomic_inc(&ffi->inject_ops); 1834 if (atomic_read(&ffi->inject_ops) >= ffi->inject_rate) { 1835 atomic_set(&ffi->inject_ops, 0); 1836 return true; 1837 } 1838 return false; 1839 } 1840 #else 1841 #define f2fs_show_injection_info(sbi, type) do { } while (0) 1842 static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type) 1843 { 1844 return false; 1845 } 1846 #endif 1847 1848 /* 1849 * Test if the mounted volume is a multi-device volume. 1850 * - For a single regular disk volume, sbi->s_ndevs is 0. 1851 * - For a single zoned disk volume, sbi->s_ndevs is 1. 1852 * - For a multi-device volume, sbi->s_ndevs is always 2 or more. 1853 */ 1854 static inline bool f2fs_is_multi_device(struct f2fs_sb_info *sbi) 1855 { 1856 return sbi->s_ndevs > 1; 1857 } 1858 1859 static inline void f2fs_update_time(struct f2fs_sb_info *sbi, int type) 1860 { 1861 unsigned long now = jiffies; 1862 1863 sbi->last_time[type] = now; 1864 1865 /* DISCARD_TIME and GC_TIME are based on REQ_TIME */ 1866 if (type == REQ_TIME) { 1867 sbi->last_time[DISCARD_TIME] = now; 1868 sbi->last_time[GC_TIME] = now; 1869 } 1870 } 1871 1872 static inline bool f2fs_time_over(struct f2fs_sb_info *sbi, int type) 1873 { 1874 unsigned long interval = sbi->interval_time[type] * HZ; 1875 1876 return time_after(jiffies, sbi->last_time[type] + interval); 1877 } 1878 1879 static inline unsigned int f2fs_time_to_wait(struct f2fs_sb_info *sbi, 1880 int type) 1881 { 1882 unsigned long interval = sbi->interval_time[type] * HZ; 1883 unsigned int wait_ms = 0; 1884 long delta; 1885 1886 delta = (sbi->last_time[type] + interval) - jiffies; 1887 if (delta > 0) 1888 wait_ms = jiffies_to_msecs(delta); 1889 1890 return wait_ms; 1891 } 1892 1893 /* 1894 * Inline functions 1895 */ 1896 static inline u32 __f2fs_crc32(struct f2fs_sb_info *sbi, u32 crc, 1897 const void *address, unsigned int length) 1898 { 1899 struct { 1900 struct shash_desc shash; 1901 char ctx[4]; 1902 } desc; 1903 int err; 1904 1905 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver) != sizeof(desc.ctx)); 1906 1907 desc.shash.tfm = sbi->s_chksum_driver; 1908 *(u32 *)desc.ctx = crc; 1909 1910 err = crypto_shash_update(&desc.shash, address, length); 1911 BUG_ON(err); 1912 1913 return *(u32 *)desc.ctx; 1914 } 1915 1916 static inline u32 f2fs_crc32(struct f2fs_sb_info *sbi, const void *address, 1917 unsigned int length) 1918 { 1919 return __f2fs_crc32(sbi, F2FS_SUPER_MAGIC, address, length); 1920 } 1921 1922 static inline bool f2fs_crc_valid(struct f2fs_sb_info *sbi, __u32 blk_crc, 1923 void *buf, size_t buf_size) 1924 { 1925 return f2fs_crc32(sbi, buf, buf_size) == blk_crc; 1926 } 1927 1928 static inline u32 f2fs_chksum(struct f2fs_sb_info *sbi, u32 crc, 1929 const void *address, unsigned int length) 1930 { 1931 return __f2fs_crc32(sbi, crc, address, length); 1932 } 1933 1934 static inline struct f2fs_inode_info *F2FS_I(struct inode *inode) 1935 { 1936 return container_of(inode, struct f2fs_inode_info, vfs_inode); 1937 } 1938 1939 static inline struct f2fs_sb_info *F2FS_SB(struct super_block *sb) 1940 { 1941 return sb->s_fs_info; 1942 } 1943 1944 static inline struct f2fs_sb_info *F2FS_I_SB(struct inode *inode) 1945 { 1946 return F2FS_SB(inode->i_sb); 1947 } 1948 1949 static inline struct f2fs_sb_info *F2FS_M_SB(struct address_space *mapping) 1950 { 1951 return F2FS_I_SB(mapping->host); 1952 } 1953 1954 static inline struct f2fs_sb_info *F2FS_P_SB(struct page *page) 1955 { 1956 return F2FS_M_SB(page_file_mapping(page)); 1957 } 1958 1959 static inline struct f2fs_super_block *F2FS_RAW_SUPER(struct f2fs_sb_info *sbi) 1960 { 1961 return (struct f2fs_super_block *)(sbi->raw_super); 1962 } 1963 1964 static inline struct f2fs_checkpoint *F2FS_CKPT(struct f2fs_sb_info *sbi) 1965 { 1966 return (struct f2fs_checkpoint *)(sbi->ckpt); 1967 } 1968 1969 static inline struct f2fs_node *F2FS_NODE(struct page *page) 1970 { 1971 return (struct f2fs_node *)page_address(page); 1972 } 1973 1974 static inline struct f2fs_inode *F2FS_INODE(struct page *page) 1975 { 1976 return &((struct f2fs_node *)page_address(page))->i; 1977 } 1978 1979 static inline struct f2fs_nm_info *NM_I(struct f2fs_sb_info *sbi) 1980 { 1981 return (struct f2fs_nm_info *)(sbi->nm_info); 1982 } 1983 1984 static inline struct f2fs_sm_info *SM_I(struct f2fs_sb_info *sbi) 1985 { 1986 return (struct f2fs_sm_info *)(sbi->sm_info); 1987 } 1988 1989 static inline struct sit_info *SIT_I(struct f2fs_sb_info *sbi) 1990 { 1991 return (struct sit_info *)(SM_I(sbi)->sit_info); 1992 } 1993 1994 static inline struct free_segmap_info *FREE_I(struct f2fs_sb_info *sbi) 1995 { 1996 return (struct free_segmap_info *)(SM_I(sbi)->free_info); 1997 } 1998 1999 static inline struct dirty_seglist_info *DIRTY_I(struct f2fs_sb_info *sbi) 2000 { 2001 return (struct dirty_seglist_info *)(SM_I(sbi)->dirty_info); 2002 } 2003 2004 static inline struct address_space *META_MAPPING(struct f2fs_sb_info *sbi) 2005 { 2006 return sbi->meta_inode->i_mapping; 2007 } 2008 2009 static inline struct address_space *NODE_MAPPING(struct f2fs_sb_info *sbi) 2010 { 2011 return sbi->node_inode->i_mapping; 2012 } 2013 2014 static inline bool is_sbi_flag_set(struct f2fs_sb_info *sbi, unsigned int type) 2015 { 2016 return test_bit(type, &sbi->s_flag); 2017 } 2018 2019 static inline void set_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type) 2020 { 2021 set_bit(type, &sbi->s_flag); 2022 } 2023 2024 static inline void clear_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type) 2025 { 2026 clear_bit(type, &sbi->s_flag); 2027 } 2028 2029 static inline unsigned long long cur_cp_version(struct f2fs_checkpoint *cp) 2030 { 2031 return le64_to_cpu(cp->checkpoint_ver); 2032 } 2033 2034 static inline unsigned long f2fs_qf_ino(struct super_block *sb, int type) 2035 { 2036 if (type < F2FS_MAX_QUOTAS) 2037 return le32_to_cpu(F2FS_SB(sb)->raw_super->qf_ino[type]); 2038 return 0; 2039 } 2040 2041 static inline __u64 cur_cp_crc(struct f2fs_checkpoint *cp) 2042 { 2043 size_t crc_offset = le32_to_cpu(cp->checksum_offset); 2044 return le32_to_cpu(*((__le32 *)((unsigned char *)cp + crc_offset))); 2045 } 2046 2047 static inline bool __is_set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 2048 { 2049 unsigned int ckpt_flags = le32_to_cpu(cp->ckpt_flags); 2050 2051 return ckpt_flags & f; 2052 } 2053 2054 static inline bool is_set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 2055 { 2056 return __is_set_ckpt_flags(F2FS_CKPT(sbi), f); 2057 } 2058 2059 static inline void __set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 2060 { 2061 unsigned int ckpt_flags; 2062 2063 ckpt_flags = le32_to_cpu(cp->ckpt_flags); 2064 ckpt_flags |= f; 2065 cp->ckpt_flags = cpu_to_le32(ckpt_flags); 2066 } 2067 2068 static inline void set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 2069 { 2070 unsigned long flags; 2071 2072 spin_lock_irqsave(&sbi->cp_lock, flags); 2073 __set_ckpt_flags(F2FS_CKPT(sbi), f); 2074 spin_unlock_irqrestore(&sbi->cp_lock, flags); 2075 } 2076 2077 static inline void __clear_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 2078 { 2079 unsigned int ckpt_flags; 2080 2081 ckpt_flags = le32_to_cpu(cp->ckpt_flags); 2082 ckpt_flags &= (~f); 2083 cp->ckpt_flags = cpu_to_le32(ckpt_flags); 2084 } 2085 2086 static inline void clear_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 2087 { 2088 unsigned long flags; 2089 2090 spin_lock_irqsave(&sbi->cp_lock, flags); 2091 __clear_ckpt_flags(F2FS_CKPT(sbi), f); 2092 spin_unlock_irqrestore(&sbi->cp_lock, flags); 2093 } 2094 2095 static inline void f2fs_lock_op(struct f2fs_sb_info *sbi) 2096 { 2097 down_read(&sbi->cp_rwsem); 2098 } 2099 2100 static inline int f2fs_trylock_op(struct f2fs_sb_info *sbi) 2101 { 2102 if (time_to_inject(sbi, FAULT_LOCK_OP)) { 2103 f2fs_show_injection_info(sbi, FAULT_LOCK_OP); 2104 return 0; 2105 } 2106 return down_read_trylock(&sbi->cp_rwsem); 2107 } 2108 2109 static inline void f2fs_unlock_op(struct f2fs_sb_info *sbi) 2110 { 2111 up_read(&sbi->cp_rwsem); 2112 } 2113 2114 static inline void f2fs_lock_all(struct f2fs_sb_info *sbi) 2115 { 2116 down_write(&sbi->cp_rwsem); 2117 } 2118 2119 static inline void f2fs_unlock_all(struct f2fs_sb_info *sbi) 2120 { 2121 up_write(&sbi->cp_rwsem); 2122 } 2123 2124 static inline int __get_cp_reason(struct f2fs_sb_info *sbi) 2125 { 2126 int reason = CP_SYNC; 2127 2128 if (test_opt(sbi, FASTBOOT)) 2129 reason = CP_FASTBOOT; 2130 if (is_sbi_flag_set(sbi, SBI_IS_CLOSE)) 2131 reason = CP_UMOUNT; 2132 return reason; 2133 } 2134 2135 static inline bool __remain_node_summaries(int reason) 2136 { 2137 return (reason & (CP_UMOUNT | CP_FASTBOOT)); 2138 } 2139 2140 static inline bool __exist_node_summaries(struct f2fs_sb_info *sbi) 2141 { 2142 return (is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG) || 2143 is_set_ckpt_flags(sbi, CP_FASTBOOT_FLAG)); 2144 } 2145 2146 /* 2147 * Check whether the inode has blocks or not 2148 */ 2149 static inline int F2FS_HAS_BLOCKS(struct inode *inode) 2150 { 2151 block_t xattr_block = F2FS_I(inode)->i_xattr_nid ? 1 : 0; 2152 2153 return (inode->i_blocks >> F2FS_LOG_SECTORS_PER_BLOCK) > xattr_block; 2154 } 2155 2156 static inline bool f2fs_has_xattr_block(unsigned int ofs) 2157 { 2158 return ofs == XATTR_NODE_OFFSET; 2159 } 2160 2161 static inline bool __allow_reserved_blocks(struct f2fs_sb_info *sbi, 2162 struct inode *inode, bool cap) 2163 { 2164 if (!inode) 2165 return true; 2166 if (!test_opt(sbi, RESERVE_ROOT)) 2167 return false; 2168 if (IS_NOQUOTA(inode)) 2169 return true; 2170 if (uid_eq(F2FS_OPTION(sbi).s_resuid, current_fsuid())) 2171 return true; 2172 if (!gid_eq(F2FS_OPTION(sbi).s_resgid, GLOBAL_ROOT_GID) && 2173 in_group_p(F2FS_OPTION(sbi).s_resgid)) 2174 return true; 2175 if (cap && capable(CAP_SYS_RESOURCE)) 2176 return true; 2177 return false; 2178 } 2179 2180 static inline void f2fs_i_blocks_write(struct inode *, block_t, bool, bool); 2181 static inline int inc_valid_block_count(struct f2fs_sb_info *sbi, 2182 struct inode *inode, blkcnt_t *count) 2183 { 2184 blkcnt_t diff = 0, release = 0; 2185 block_t avail_user_block_count; 2186 int ret; 2187 2188 ret = dquot_reserve_block(inode, *count); 2189 if (ret) 2190 return ret; 2191 2192 if (time_to_inject(sbi, FAULT_BLOCK)) { 2193 f2fs_show_injection_info(sbi, FAULT_BLOCK); 2194 release = *count; 2195 goto release_quota; 2196 } 2197 2198 /* 2199 * let's increase this in prior to actual block count change in order 2200 * for f2fs_sync_file to avoid data races when deciding checkpoint. 2201 */ 2202 percpu_counter_add(&sbi->alloc_valid_block_count, (*count)); 2203 2204 spin_lock(&sbi->stat_lock); 2205 sbi->total_valid_block_count += (block_t)(*count); 2206 avail_user_block_count = sbi->user_block_count - 2207 sbi->current_reserved_blocks; 2208 2209 if (!__allow_reserved_blocks(sbi, inode, true)) 2210 avail_user_block_count -= F2FS_OPTION(sbi).root_reserved_blocks; 2211 2212 if (F2FS_IO_ALIGNED(sbi)) 2213 avail_user_block_count -= sbi->blocks_per_seg * 2214 SM_I(sbi)->additional_reserved_segments; 2215 2216 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) { 2217 if (avail_user_block_count > sbi->unusable_block_count) 2218 avail_user_block_count -= sbi->unusable_block_count; 2219 else 2220 avail_user_block_count = 0; 2221 } 2222 if (unlikely(sbi->total_valid_block_count > avail_user_block_count)) { 2223 diff = sbi->total_valid_block_count - avail_user_block_count; 2224 if (diff > *count) 2225 diff = *count; 2226 *count -= diff; 2227 release = diff; 2228 sbi->total_valid_block_count -= diff; 2229 if (!*count) { 2230 spin_unlock(&sbi->stat_lock); 2231 goto enospc; 2232 } 2233 } 2234 spin_unlock(&sbi->stat_lock); 2235 2236 if (unlikely(release)) { 2237 percpu_counter_sub(&sbi->alloc_valid_block_count, release); 2238 dquot_release_reservation_block(inode, release); 2239 } 2240 f2fs_i_blocks_write(inode, *count, true, true); 2241 return 0; 2242 2243 enospc: 2244 percpu_counter_sub(&sbi->alloc_valid_block_count, release); 2245 release_quota: 2246 dquot_release_reservation_block(inode, release); 2247 return -ENOSPC; 2248 } 2249 2250 __printf(2, 3) 2251 void f2fs_printk(struct f2fs_sb_info *sbi, const char *fmt, ...); 2252 2253 #define f2fs_err(sbi, fmt, ...) \ 2254 f2fs_printk(sbi, KERN_ERR fmt, ##__VA_ARGS__) 2255 #define f2fs_warn(sbi, fmt, ...) \ 2256 f2fs_printk(sbi, KERN_WARNING fmt, ##__VA_ARGS__) 2257 #define f2fs_notice(sbi, fmt, ...) \ 2258 f2fs_printk(sbi, KERN_NOTICE fmt, ##__VA_ARGS__) 2259 #define f2fs_info(sbi, fmt, ...) \ 2260 f2fs_printk(sbi, KERN_INFO fmt, ##__VA_ARGS__) 2261 #define f2fs_debug(sbi, fmt, ...) \ 2262 f2fs_printk(sbi, KERN_DEBUG fmt, ##__VA_ARGS__) 2263 2264 static inline void dec_valid_block_count(struct f2fs_sb_info *sbi, 2265 struct inode *inode, 2266 block_t count) 2267 { 2268 blkcnt_t sectors = count << F2FS_LOG_SECTORS_PER_BLOCK; 2269 2270 spin_lock(&sbi->stat_lock); 2271 f2fs_bug_on(sbi, sbi->total_valid_block_count < (block_t) count); 2272 sbi->total_valid_block_count -= (block_t)count; 2273 if (sbi->reserved_blocks && 2274 sbi->current_reserved_blocks < sbi->reserved_blocks) 2275 sbi->current_reserved_blocks = min(sbi->reserved_blocks, 2276 sbi->current_reserved_blocks + count); 2277 spin_unlock(&sbi->stat_lock); 2278 if (unlikely(inode->i_blocks < sectors)) { 2279 f2fs_warn(sbi, "Inconsistent i_blocks, ino:%lu, iblocks:%llu, sectors:%llu", 2280 inode->i_ino, 2281 (unsigned long long)inode->i_blocks, 2282 (unsigned long long)sectors); 2283 set_sbi_flag(sbi, SBI_NEED_FSCK); 2284 return; 2285 } 2286 f2fs_i_blocks_write(inode, count, false, true); 2287 } 2288 2289 static inline void inc_page_count(struct f2fs_sb_info *sbi, int count_type) 2290 { 2291 atomic_inc(&sbi->nr_pages[count_type]); 2292 2293 if (count_type == F2FS_DIRTY_DENTS || 2294 count_type == F2FS_DIRTY_NODES || 2295 count_type == F2FS_DIRTY_META || 2296 count_type == F2FS_DIRTY_QDATA || 2297 count_type == F2FS_DIRTY_IMETA) 2298 set_sbi_flag(sbi, SBI_IS_DIRTY); 2299 } 2300 2301 static inline void inode_inc_dirty_pages(struct inode *inode) 2302 { 2303 atomic_inc(&F2FS_I(inode)->dirty_pages); 2304 inc_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ? 2305 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA); 2306 if (IS_NOQUOTA(inode)) 2307 inc_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA); 2308 } 2309 2310 static inline void dec_page_count(struct f2fs_sb_info *sbi, int count_type) 2311 { 2312 atomic_dec(&sbi->nr_pages[count_type]); 2313 } 2314 2315 static inline void inode_dec_dirty_pages(struct inode *inode) 2316 { 2317 if (!S_ISDIR(inode->i_mode) && !S_ISREG(inode->i_mode) && 2318 !S_ISLNK(inode->i_mode)) 2319 return; 2320 2321 atomic_dec(&F2FS_I(inode)->dirty_pages); 2322 dec_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ? 2323 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA); 2324 if (IS_NOQUOTA(inode)) 2325 dec_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA); 2326 } 2327 2328 static inline s64 get_pages(struct f2fs_sb_info *sbi, int count_type) 2329 { 2330 return atomic_read(&sbi->nr_pages[count_type]); 2331 } 2332 2333 static inline int get_dirty_pages(struct inode *inode) 2334 { 2335 return atomic_read(&F2FS_I(inode)->dirty_pages); 2336 } 2337 2338 static inline int get_blocktype_secs(struct f2fs_sb_info *sbi, int block_type) 2339 { 2340 unsigned int pages_per_sec = sbi->segs_per_sec * sbi->blocks_per_seg; 2341 unsigned int segs = (get_pages(sbi, block_type) + pages_per_sec - 1) >> 2342 sbi->log_blocks_per_seg; 2343 2344 return segs / sbi->segs_per_sec; 2345 } 2346 2347 static inline block_t valid_user_blocks(struct f2fs_sb_info *sbi) 2348 { 2349 return sbi->total_valid_block_count; 2350 } 2351 2352 static inline block_t discard_blocks(struct f2fs_sb_info *sbi) 2353 { 2354 return sbi->discard_blks; 2355 } 2356 2357 static inline unsigned long __bitmap_size(struct f2fs_sb_info *sbi, int flag) 2358 { 2359 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi); 2360 2361 /* return NAT or SIT bitmap */ 2362 if (flag == NAT_BITMAP) 2363 return le32_to_cpu(ckpt->nat_ver_bitmap_bytesize); 2364 else if (flag == SIT_BITMAP) 2365 return le32_to_cpu(ckpt->sit_ver_bitmap_bytesize); 2366 2367 return 0; 2368 } 2369 2370 static inline block_t __cp_payload(struct f2fs_sb_info *sbi) 2371 { 2372 return le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_payload); 2373 } 2374 2375 static inline void *__bitmap_ptr(struct f2fs_sb_info *sbi, int flag) 2376 { 2377 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi); 2378 void *tmp_ptr = &ckpt->sit_nat_version_bitmap; 2379 int offset; 2380 2381 if (is_set_ckpt_flags(sbi, CP_LARGE_NAT_BITMAP_FLAG)) { 2382 offset = (flag == SIT_BITMAP) ? 2383 le32_to_cpu(ckpt->nat_ver_bitmap_bytesize) : 0; 2384 /* 2385 * if large_nat_bitmap feature is enabled, leave checksum 2386 * protection for all nat/sit bitmaps. 2387 */ 2388 return tmp_ptr + offset + sizeof(__le32); 2389 } 2390 2391 if (__cp_payload(sbi) > 0) { 2392 if (flag == NAT_BITMAP) 2393 return &ckpt->sit_nat_version_bitmap; 2394 else 2395 return (unsigned char *)ckpt + F2FS_BLKSIZE; 2396 } else { 2397 offset = (flag == NAT_BITMAP) ? 2398 le32_to_cpu(ckpt->sit_ver_bitmap_bytesize) : 0; 2399 return tmp_ptr + offset; 2400 } 2401 } 2402 2403 static inline block_t __start_cp_addr(struct f2fs_sb_info *sbi) 2404 { 2405 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr); 2406 2407 if (sbi->cur_cp_pack == 2) 2408 start_addr += sbi->blocks_per_seg; 2409 return start_addr; 2410 } 2411 2412 static inline block_t __start_cp_next_addr(struct f2fs_sb_info *sbi) 2413 { 2414 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr); 2415 2416 if (sbi->cur_cp_pack == 1) 2417 start_addr += sbi->blocks_per_seg; 2418 return start_addr; 2419 } 2420 2421 static inline void __set_cp_next_pack(struct f2fs_sb_info *sbi) 2422 { 2423 sbi->cur_cp_pack = (sbi->cur_cp_pack == 1) ? 2 : 1; 2424 } 2425 2426 static inline block_t __start_sum_addr(struct f2fs_sb_info *sbi) 2427 { 2428 return le32_to_cpu(F2FS_CKPT(sbi)->cp_pack_start_sum); 2429 } 2430 2431 static inline int inc_valid_node_count(struct f2fs_sb_info *sbi, 2432 struct inode *inode, bool is_inode) 2433 { 2434 block_t valid_block_count; 2435 unsigned int valid_node_count, user_block_count; 2436 int err; 2437 2438 if (is_inode) { 2439 if (inode) { 2440 err = dquot_alloc_inode(inode); 2441 if (err) 2442 return err; 2443 } 2444 } else { 2445 err = dquot_reserve_block(inode, 1); 2446 if (err) 2447 return err; 2448 } 2449 2450 if (time_to_inject(sbi, FAULT_BLOCK)) { 2451 f2fs_show_injection_info(sbi, FAULT_BLOCK); 2452 goto enospc; 2453 } 2454 2455 spin_lock(&sbi->stat_lock); 2456 2457 valid_block_count = sbi->total_valid_block_count + 2458 sbi->current_reserved_blocks + 1; 2459 2460 if (!__allow_reserved_blocks(sbi, inode, false)) 2461 valid_block_count += F2FS_OPTION(sbi).root_reserved_blocks; 2462 2463 if (F2FS_IO_ALIGNED(sbi)) 2464 valid_block_count += sbi->blocks_per_seg * 2465 SM_I(sbi)->additional_reserved_segments; 2466 2467 user_block_count = sbi->user_block_count; 2468 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) 2469 user_block_count -= sbi->unusable_block_count; 2470 2471 if (unlikely(valid_block_count > user_block_count)) { 2472 spin_unlock(&sbi->stat_lock); 2473 goto enospc; 2474 } 2475 2476 valid_node_count = sbi->total_valid_node_count + 1; 2477 if (unlikely(valid_node_count > sbi->total_node_count)) { 2478 spin_unlock(&sbi->stat_lock); 2479 goto enospc; 2480 } 2481 2482 sbi->total_valid_node_count++; 2483 sbi->total_valid_block_count++; 2484 spin_unlock(&sbi->stat_lock); 2485 2486 if (inode) { 2487 if (is_inode) 2488 f2fs_mark_inode_dirty_sync(inode, true); 2489 else 2490 f2fs_i_blocks_write(inode, 1, true, true); 2491 } 2492 2493 percpu_counter_inc(&sbi->alloc_valid_block_count); 2494 return 0; 2495 2496 enospc: 2497 if (is_inode) { 2498 if (inode) 2499 dquot_free_inode(inode); 2500 } else { 2501 dquot_release_reservation_block(inode, 1); 2502 } 2503 return -ENOSPC; 2504 } 2505 2506 static inline void dec_valid_node_count(struct f2fs_sb_info *sbi, 2507 struct inode *inode, bool is_inode) 2508 { 2509 spin_lock(&sbi->stat_lock); 2510 2511 f2fs_bug_on(sbi, !sbi->total_valid_block_count); 2512 f2fs_bug_on(sbi, !sbi->total_valid_node_count); 2513 2514 sbi->total_valid_node_count--; 2515 sbi->total_valid_block_count--; 2516 if (sbi->reserved_blocks && 2517 sbi->current_reserved_blocks < sbi->reserved_blocks) 2518 sbi->current_reserved_blocks++; 2519 2520 spin_unlock(&sbi->stat_lock); 2521 2522 if (is_inode) { 2523 dquot_free_inode(inode); 2524 } else { 2525 if (unlikely(inode->i_blocks == 0)) { 2526 f2fs_warn(sbi, "dec_valid_node_count: inconsistent i_blocks, ino:%lu, iblocks:%llu", 2527 inode->i_ino, 2528 (unsigned long long)inode->i_blocks); 2529 set_sbi_flag(sbi, SBI_NEED_FSCK); 2530 return; 2531 } 2532 f2fs_i_blocks_write(inode, 1, false, true); 2533 } 2534 } 2535 2536 static inline unsigned int valid_node_count(struct f2fs_sb_info *sbi) 2537 { 2538 return sbi->total_valid_node_count; 2539 } 2540 2541 static inline void inc_valid_inode_count(struct f2fs_sb_info *sbi) 2542 { 2543 percpu_counter_inc(&sbi->total_valid_inode_count); 2544 } 2545 2546 static inline void dec_valid_inode_count(struct f2fs_sb_info *sbi) 2547 { 2548 percpu_counter_dec(&sbi->total_valid_inode_count); 2549 } 2550 2551 static inline s64 valid_inode_count(struct f2fs_sb_info *sbi) 2552 { 2553 return percpu_counter_sum_positive(&sbi->total_valid_inode_count); 2554 } 2555 2556 static inline struct page *f2fs_grab_cache_page(struct address_space *mapping, 2557 pgoff_t index, bool for_write) 2558 { 2559 struct page *page; 2560 2561 if (IS_ENABLED(CONFIG_F2FS_FAULT_INJECTION)) { 2562 if (!for_write) 2563 page = find_get_page_flags(mapping, index, 2564 FGP_LOCK | FGP_ACCESSED); 2565 else 2566 page = find_lock_page(mapping, index); 2567 if (page) 2568 return page; 2569 2570 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_ALLOC)) { 2571 f2fs_show_injection_info(F2FS_M_SB(mapping), 2572 FAULT_PAGE_ALLOC); 2573 return NULL; 2574 } 2575 } 2576 2577 if (!for_write) 2578 return grab_cache_page(mapping, index); 2579 return grab_cache_page_write_begin(mapping, index, AOP_FLAG_NOFS); 2580 } 2581 2582 static inline struct page *f2fs_pagecache_get_page( 2583 struct address_space *mapping, pgoff_t index, 2584 int fgp_flags, gfp_t gfp_mask) 2585 { 2586 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_GET)) { 2587 f2fs_show_injection_info(F2FS_M_SB(mapping), FAULT_PAGE_GET); 2588 return NULL; 2589 } 2590 2591 return pagecache_get_page(mapping, index, fgp_flags, gfp_mask); 2592 } 2593 2594 static inline void f2fs_copy_page(struct page *src, struct page *dst) 2595 { 2596 char *src_kaddr = kmap(src); 2597 char *dst_kaddr = kmap(dst); 2598 2599 memcpy(dst_kaddr, src_kaddr, PAGE_SIZE); 2600 kunmap(dst); 2601 kunmap(src); 2602 } 2603 2604 static inline void f2fs_put_page(struct page *page, int unlock) 2605 { 2606 if (!page) 2607 return; 2608 2609 if (unlock) { 2610 f2fs_bug_on(F2FS_P_SB(page), !PageLocked(page)); 2611 unlock_page(page); 2612 } 2613 put_page(page); 2614 } 2615 2616 static inline void f2fs_put_dnode(struct dnode_of_data *dn) 2617 { 2618 if (dn->node_page) 2619 f2fs_put_page(dn->node_page, 1); 2620 if (dn->inode_page && dn->node_page != dn->inode_page) 2621 f2fs_put_page(dn->inode_page, 0); 2622 dn->node_page = NULL; 2623 dn->inode_page = NULL; 2624 } 2625 2626 static inline struct kmem_cache *f2fs_kmem_cache_create(const char *name, 2627 size_t size) 2628 { 2629 return kmem_cache_create(name, size, 0, SLAB_RECLAIM_ACCOUNT, NULL); 2630 } 2631 2632 static inline void *f2fs_kmem_cache_alloc_nofail(struct kmem_cache *cachep, 2633 gfp_t flags) 2634 { 2635 void *entry; 2636 2637 entry = kmem_cache_alloc(cachep, flags); 2638 if (!entry) 2639 entry = kmem_cache_alloc(cachep, flags | __GFP_NOFAIL); 2640 return entry; 2641 } 2642 2643 static inline void *f2fs_kmem_cache_alloc(struct kmem_cache *cachep, 2644 gfp_t flags, bool nofail, struct f2fs_sb_info *sbi) 2645 { 2646 if (nofail) 2647 return f2fs_kmem_cache_alloc_nofail(cachep, flags); 2648 2649 if (time_to_inject(sbi, FAULT_SLAB_ALLOC)) { 2650 f2fs_show_injection_info(sbi, FAULT_SLAB_ALLOC); 2651 return NULL; 2652 } 2653 2654 return kmem_cache_alloc(cachep, flags); 2655 } 2656 2657 static inline bool is_inflight_io(struct f2fs_sb_info *sbi, int type) 2658 { 2659 if (get_pages(sbi, F2FS_RD_DATA) || get_pages(sbi, F2FS_RD_NODE) || 2660 get_pages(sbi, F2FS_RD_META) || get_pages(sbi, F2FS_WB_DATA) || 2661 get_pages(sbi, F2FS_WB_CP_DATA) || 2662 get_pages(sbi, F2FS_DIO_READ) || 2663 get_pages(sbi, F2FS_DIO_WRITE)) 2664 return true; 2665 2666 if (type != DISCARD_TIME && SM_I(sbi) && SM_I(sbi)->dcc_info && 2667 atomic_read(&SM_I(sbi)->dcc_info->queued_discard)) 2668 return true; 2669 2670 if (SM_I(sbi) && SM_I(sbi)->fcc_info && 2671 atomic_read(&SM_I(sbi)->fcc_info->queued_flush)) 2672 return true; 2673 return false; 2674 } 2675 2676 static inline bool is_idle(struct f2fs_sb_info *sbi, int type) 2677 { 2678 if (sbi->gc_mode == GC_URGENT_HIGH) 2679 return true; 2680 2681 if (is_inflight_io(sbi, type)) 2682 return false; 2683 2684 if (sbi->gc_mode == GC_URGENT_LOW && 2685 (type == DISCARD_TIME || type == GC_TIME)) 2686 return true; 2687 2688 return f2fs_time_over(sbi, type); 2689 } 2690 2691 static inline void f2fs_radix_tree_insert(struct radix_tree_root *root, 2692 unsigned long index, void *item) 2693 { 2694 while (radix_tree_insert(root, index, item)) 2695 cond_resched(); 2696 } 2697 2698 #define RAW_IS_INODE(p) ((p)->footer.nid == (p)->footer.ino) 2699 2700 static inline bool IS_INODE(struct page *page) 2701 { 2702 struct f2fs_node *p = F2FS_NODE(page); 2703 2704 return RAW_IS_INODE(p); 2705 } 2706 2707 static inline int offset_in_addr(struct f2fs_inode *i) 2708 { 2709 return (i->i_inline & F2FS_EXTRA_ATTR) ? 2710 (le16_to_cpu(i->i_extra_isize) / sizeof(__le32)) : 0; 2711 } 2712 2713 static inline __le32 *blkaddr_in_node(struct f2fs_node *node) 2714 { 2715 return RAW_IS_INODE(node) ? node->i.i_addr : node->dn.addr; 2716 } 2717 2718 static inline int f2fs_has_extra_attr(struct inode *inode); 2719 static inline block_t data_blkaddr(struct inode *inode, 2720 struct page *node_page, unsigned int offset) 2721 { 2722 struct f2fs_node *raw_node; 2723 __le32 *addr_array; 2724 int base = 0; 2725 bool is_inode = IS_INODE(node_page); 2726 2727 raw_node = F2FS_NODE(node_page); 2728 2729 if (is_inode) { 2730 if (!inode) 2731 /* from GC path only */ 2732 base = offset_in_addr(&raw_node->i); 2733 else if (f2fs_has_extra_attr(inode)) 2734 base = get_extra_isize(inode); 2735 } 2736 2737 addr_array = blkaddr_in_node(raw_node); 2738 return le32_to_cpu(addr_array[base + offset]); 2739 } 2740 2741 static inline block_t f2fs_data_blkaddr(struct dnode_of_data *dn) 2742 { 2743 return data_blkaddr(dn->inode, dn->node_page, dn->ofs_in_node); 2744 } 2745 2746 static inline int f2fs_test_bit(unsigned int nr, char *addr) 2747 { 2748 int mask; 2749 2750 addr += (nr >> 3); 2751 mask = 1 << (7 - (nr & 0x07)); 2752 return mask & *addr; 2753 } 2754 2755 static inline void f2fs_set_bit(unsigned int nr, char *addr) 2756 { 2757 int mask; 2758 2759 addr += (nr >> 3); 2760 mask = 1 << (7 - (nr & 0x07)); 2761 *addr |= mask; 2762 } 2763 2764 static inline void f2fs_clear_bit(unsigned int nr, char *addr) 2765 { 2766 int mask; 2767 2768 addr += (nr >> 3); 2769 mask = 1 << (7 - (nr & 0x07)); 2770 *addr &= ~mask; 2771 } 2772 2773 static inline int f2fs_test_and_set_bit(unsigned int nr, char *addr) 2774 { 2775 int mask; 2776 int ret; 2777 2778 addr += (nr >> 3); 2779 mask = 1 << (7 - (nr & 0x07)); 2780 ret = mask & *addr; 2781 *addr |= mask; 2782 return ret; 2783 } 2784 2785 static inline int f2fs_test_and_clear_bit(unsigned int nr, char *addr) 2786 { 2787 int mask; 2788 int ret; 2789 2790 addr += (nr >> 3); 2791 mask = 1 << (7 - (nr & 0x07)); 2792 ret = mask & *addr; 2793 *addr &= ~mask; 2794 return ret; 2795 } 2796 2797 static inline void f2fs_change_bit(unsigned int nr, char *addr) 2798 { 2799 int mask; 2800 2801 addr += (nr >> 3); 2802 mask = 1 << (7 - (nr & 0x07)); 2803 *addr ^= mask; 2804 } 2805 2806 /* 2807 * On-disk inode flags (f2fs_inode::i_flags) 2808 */ 2809 #define F2FS_COMPR_FL 0x00000004 /* Compress file */ 2810 #define F2FS_SYNC_FL 0x00000008 /* Synchronous updates */ 2811 #define F2FS_IMMUTABLE_FL 0x00000010 /* Immutable file */ 2812 #define F2FS_APPEND_FL 0x00000020 /* writes to file may only append */ 2813 #define F2FS_NODUMP_FL 0x00000040 /* do not dump file */ 2814 #define F2FS_NOATIME_FL 0x00000080 /* do not update atime */ 2815 #define F2FS_NOCOMP_FL 0x00000400 /* Don't compress */ 2816 #define F2FS_INDEX_FL 0x00001000 /* hash-indexed directory */ 2817 #define F2FS_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */ 2818 #define F2FS_PROJINHERIT_FL 0x20000000 /* Create with parents projid */ 2819 #define F2FS_CASEFOLD_FL 0x40000000 /* Casefolded file */ 2820 2821 /* Flags that should be inherited by new inodes from their parent. */ 2822 #define F2FS_FL_INHERITED (F2FS_SYNC_FL | F2FS_NODUMP_FL | F2FS_NOATIME_FL | \ 2823 F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \ 2824 F2FS_CASEFOLD_FL | F2FS_COMPR_FL | F2FS_NOCOMP_FL) 2825 2826 /* Flags that are appropriate for regular files (all but dir-specific ones). */ 2827 #define F2FS_REG_FLMASK (~(F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \ 2828 F2FS_CASEFOLD_FL)) 2829 2830 /* Flags that are appropriate for non-directories/regular files. */ 2831 #define F2FS_OTHER_FLMASK (F2FS_NODUMP_FL | F2FS_NOATIME_FL) 2832 2833 static inline __u32 f2fs_mask_flags(umode_t mode, __u32 flags) 2834 { 2835 if (S_ISDIR(mode)) 2836 return flags; 2837 else if (S_ISREG(mode)) 2838 return flags & F2FS_REG_FLMASK; 2839 else 2840 return flags & F2FS_OTHER_FLMASK; 2841 } 2842 2843 static inline void __mark_inode_dirty_flag(struct inode *inode, 2844 int flag, bool set) 2845 { 2846 switch (flag) { 2847 case FI_INLINE_XATTR: 2848 case FI_INLINE_DATA: 2849 case FI_INLINE_DENTRY: 2850 case FI_NEW_INODE: 2851 if (set) 2852 return; 2853 fallthrough; 2854 case FI_DATA_EXIST: 2855 case FI_INLINE_DOTS: 2856 case FI_PIN_FILE: 2857 case FI_COMPRESS_RELEASED: 2858 f2fs_mark_inode_dirty_sync(inode, true); 2859 } 2860 } 2861 2862 static inline void set_inode_flag(struct inode *inode, int flag) 2863 { 2864 set_bit(flag, F2FS_I(inode)->flags); 2865 __mark_inode_dirty_flag(inode, flag, true); 2866 } 2867 2868 static inline int is_inode_flag_set(struct inode *inode, int flag) 2869 { 2870 return test_bit(flag, F2FS_I(inode)->flags); 2871 } 2872 2873 static inline void clear_inode_flag(struct inode *inode, int flag) 2874 { 2875 clear_bit(flag, F2FS_I(inode)->flags); 2876 __mark_inode_dirty_flag(inode, flag, false); 2877 } 2878 2879 static inline bool f2fs_verity_in_progress(struct inode *inode) 2880 { 2881 return IS_ENABLED(CONFIG_FS_VERITY) && 2882 is_inode_flag_set(inode, FI_VERITY_IN_PROGRESS); 2883 } 2884 2885 static inline void set_acl_inode(struct inode *inode, umode_t mode) 2886 { 2887 F2FS_I(inode)->i_acl_mode = mode; 2888 set_inode_flag(inode, FI_ACL_MODE); 2889 f2fs_mark_inode_dirty_sync(inode, false); 2890 } 2891 2892 static inline void f2fs_i_links_write(struct inode *inode, bool inc) 2893 { 2894 if (inc) 2895 inc_nlink(inode); 2896 else 2897 drop_nlink(inode); 2898 f2fs_mark_inode_dirty_sync(inode, true); 2899 } 2900 2901 static inline void f2fs_i_blocks_write(struct inode *inode, 2902 block_t diff, bool add, bool claim) 2903 { 2904 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE); 2905 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER); 2906 2907 /* add = 1, claim = 1 should be dquot_reserve_block in pair */ 2908 if (add) { 2909 if (claim) 2910 dquot_claim_block(inode, diff); 2911 else 2912 dquot_alloc_block_nofail(inode, diff); 2913 } else { 2914 dquot_free_block(inode, diff); 2915 } 2916 2917 f2fs_mark_inode_dirty_sync(inode, true); 2918 if (clean || recover) 2919 set_inode_flag(inode, FI_AUTO_RECOVER); 2920 } 2921 2922 static inline void f2fs_i_size_write(struct inode *inode, loff_t i_size) 2923 { 2924 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE); 2925 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER); 2926 2927 if (i_size_read(inode) == i_size) 2928 return; 2929 2930 i_size_write(inode, i_size); 2931 f2fs_mark_inode_dirty_sync(inode, true); 2932 if (clean || recover) 2933 set_inode_flag(inode, FI_AUTO_RECOVER); 2934 } 2935 2936 static inline void f2fs_i_depth_write(struct inode *inode, unsigned int depth) 2937 { 2938 F2FS_I(inode)->i_current_depth = depth; 2939 f2fs_mark_inode_dirty_sync(inode, true); 2940 } 2941 2942 static inline void f2fs_i_gc_failures_write(struct inode *inode, 2943 unsigned int count) 2944 { 2945 F2FS_I(inode)->i_gc_failures[GC_FAILURE_PIN] = count; 2946 f2fs_mark_inode_dirty_sync(inode, true); 2947 } 2948 2949 static inline void f2fs_i_xnid_write(struct inode *inode, nid_t xnid) 2950 { 2951 F2FS_I(inode)->i_xattr_nid = xnid; 2952 f2fs_mark_inode_dirty_sync(inode, true); 2953 } 2954 2955 static inline void f2fs_i_pino_write(struct inode *inode, nid_t pino) 2956 { 2957 F2FS_I(inode)->i_pino = pino; 2958 f2fs_mark_inode_dirty_sync(inode, true); 2959 } 2960 2961 static inline void get_inline_info(struct inode *inode, struct f2fs_inode *ri) 2962 { 2963 struct f2fs_inode_info *fi = F2FS_I(inode); 2964 2965 if (ri->i_inline & F2FS_INLINE_XATTR) 2966 set_bit(FI_INLINE_XATTR, fi->flags); 2967 if (ri->i_inline & F2FS_INLINE_DATA) 2968 set_bit(FI_INLINE_DATA, fi->flags); 2969 if (ri->i_inline & F2FS_INLINE_DENTRY) 2970 set_bit(FI_INLINE_DENTRY, fi->flags); 2971 if (ri->i_inline & F2FS_DATA_EXIST) 2972 set_bit(FI_DATA_EXIST, fi->flags); 2973 if (ri->i_inline & F2FS_INLINE_DOTS) 2974 set_bit(FI_INLINE_DOTS, fi->flags); 2975 if (ri->i_inline & F2FS_EXTRA_ATTR) 2976 set_bit(FI_EXTRA_ATTR, fi->flags); 2977 if (ri->i_inline & F2FS_PIN_FILE) 2978 set_bit(FI_PIN_FILE, fi->flags); 2979 if (ri->i_inline & F2FS_COMPRESS_RELEASED) 2980 set_bit(FI_COMPRESS_RELEASED, fi->flags); 2981 } 2982 2983 static inline void set_raw_inline(struct inode *inode, struct f2fs_inode *ri) 2984 { 2985 ri->i_inline = 0; 2986 2987 if (is_inode_flag_set(inode, FI_INLINE_XATTR)) 2988 ri->i_inline |= F2FS_INLINE_XATTR; 2989 if (is_inode_flag_set(inode, FI_INLINE_DATA)) 2990 ri->i_inline |= F2FS_INLINE_DATA; 2991 if (is_inode_flag_set(inode, FI_INLINE_DENTRY)) 2992 ri->i_inline |= F2FS_INLINE_DENTRY; 2993 if (is_inode_flag_set(inode, FI_DATA_EXIST)) 2994 ri->i_inline |= F2FS_DATA_EXIST; 2995 if (is_inode_flag_set(inode, FI_INLINE_DOTS)) 2996 ri->i_inline |= F2FS_INLINE_DOTS; 2997 if (is_inode_flag_set(inode, FI_EXTRA_ATTR)) 2998 ri->i_inline |= F2FS_EXTRA_ATTR; 2999 if (is_inode_flag_set(inode, FI_PIN_FILE)) 3000 ri->i_inline |= F2FS_PIN_FILE; 3001 if (is_inode_flag_set(inode, FI_COMPRESS_RELEASED)) 3002 ri->i_inline |= F2FS_COMPRESS_RELEASED; 3003 } 3004 3005 static inline int f2fs_has_extra_attr(struct inode *inode) 3006 { 3007 return is_inode_flag_set(inode, FI_EXTRA_ATTR); 3008 } 3009 3010 static inline int f2fs_has_inline_xattr(struct inode *inode) 3011 { 3012 return is_inode_flag_set(inode, FI_INLINE_XATTR); 3013 } 3014 3015 static inline int f2fs_compressed_file(struct inode *inode) 3016 { 3017 return S_ISREG(inode->i_mode) && 3018 is_inode_flag_set(inode, FI_COMPRESSED_FILE); 3019 } 3020 3021 static inline bool f2fs_need_compress_data(struct inode *inode) 3022 { 3023 int compress_mode = F2FS_OPTION(F2FS_I_SB(inode)).compress_mode; 3024 3025 if (!f2fs_compressed_file(inode)) 3026 return false; 3027 3028 if (compress_mode == COMPR_MODE_FS) 3029 return true; 3030 else if (compress_mode == COMPR_MODE_USER && 3031 is_inode_flag_set(inode, FI_ENABLE_COMPRESS)) 3032 return true; 3033 3034 return false; 3035 } 3036 3037 static inline unsigned int addrs_per_inode(struct inode *inode) 3038 { 3039 unsigned int addrs = CUR_ADDRS_PER_INODE(inode) - 3040 get_inline_xattr_addrs(inode); 3041 3042 if (!f2fs_compressed_file(inode)) 3043 return addrs; 3044 return ALIGN_DOWN(addrs, F2FS_I(inode)->i_cluster_size); 3045 } 3046 3047 static inline unsigned int addrs_per_block(struct inode *inode) 3048 { 3049 if (!f2fs_compressed_file(inode)) 3050 return DEF_ADDRS_PER_BLOCK; 3051 return ALIGN_DOWN(DEF_ADDRS_PER_BLOCK, F2FS_I(inode)->i_cluster_size); 3052 } 3053 3054 static inline void *inline_xattr_addr(struct inode *inode, struct page *page) 3055 { 3056 struct f2fs_inode *ri = F2FS_INODE(page); 3057 3058 return (void *)&(ri->i_addr[DEF_ADDRS_PER_INODE - 3059 get_inline_xattr_addrs(inode)]); 3060 } 3061 3062 static inline int inline_xattr_size(struct inode *inode) 3063 { 3064 if (f2fs_has_inline_xattr(inode)) 3065 return get_inline_xattr_addrs(inode) * sizeof(__le32); 3066 return 0; 3067 } 3068 3069 static inline int f2fs_has_inline_data(struct inode *inode) 3070 { 3071 return is_inode_flag_set(inode, FI_INLINE_DATA); 3072 } 3073 3074 static inline int f2fs_exist_data(struct inode *inode) 3075 { 3076 return is_inode_flag_set(inode, FI_DATA_EXIST); 3077 } 3078 3079 static inline int f2fs_has_inline_dots(struct inode *inode) 3080 { 3081 return is_inode_flag_set(inode, FI_INLINE_DOTS); 3082 } 3083 3084 static inline int f2fs_is_mmap_file(struct inode *inode) 3085 { 3086 return is_inode_flag_set(inode, FI_MMAP_FILE); 3087 } 3088 3089 static inline bool f2fs_is_pinned_file(struct inode *inode) 3090 { 3091 return is_inode_flag_set(inode, FI_PIN_FILE); 3092 } 3093 3094 static inline bool f2fs_is_atomic_file(struct inode *inode) 3095 { 3096 return is_inode_flag_set(inode, FI_ATOMIC_FILE); 3097 } 3098 3099 static inline bool f2fs_is_commit_atomic_write(struct inode *inode) 3100 { 3101 return is_inode_flag_set(inode, FI_ATOMIC_COMMIT); 3102 } 3103 3104 static inline bool f2fs_is_volatile_file(struct inode *inode) 3105 { 3106 return is_inode_flag_set(inode, FI_VOLATILE_FILE); 3107 } 3108 3109 static inline bool f2fs_is_first_block_written(struct inode *inode) 3110 { 3111 return is_inode_flag_set(inode, FI_FIRST_BLOCK_WRITTEN); 3112 } 3113 3114 static inline bool f2fs_is_drop_cache(struct inode *inode) 3115 { 3116 return is_inode_flag_set(inode, FI_DROP_CACHE); 3117 } 3118 3119 static inline void *inline_data_addr(struct inode *inode, struct page *page) 3120 { 3121 struct f2fs_inode *ri = F2FS_INODE(page); 3122 int extra_size = get_extra_isize(inode); 3123 3124 return (void *)&(ri->i_addr[extra_size + DEF_INLINE_RESERVED_SIZE]); 3125 } 3126 3127 static inline int f2fs_has_inline_dentry(struct inode *inode) 3128 { 3129 return is_inode_flag_set(inode, FI_INLINE_DENTRY); 3130 } 3131 3132 static inline int is_file(struct inode *inode, int type) 3133 { 3134 return F2FS_I(inode)->i_advise & type; 3135 } 3136 3137 static inline void set_file(struct inode *inode, int type) 3138 { 3139 if (is_file(inode, type)) 3140 return; 3141 F2FS_I(inode)->i_advise |= type; 3142 f2fs_mark_inode_dirty_sync(inode, true); 3143 } 3144 3145 static inline void clear_file(struct inode *inode, int type) 3146 { 3147 if (!is_file(inode, type)) 3148 return; 3149 F2FS_I(inode)->i_advise &= ~type; 3150 f2fs_mark_inode_dirty_sync(inode, true); 3151 } 3152 3153 static inline bool f2fs_is_time_consistent(struct inode *inode) 3154 { 3155 if (!timespec64_equal(F2FS_I(inode)->i_disk_time, &inode->i_atime)) 3156 return false; 3157 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 1, &inode->i_ctime)) 3158 return false; 3159 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 2, &inode->i_mtime)) 3160 return false; 3161 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 3, 3162 &F2FS_I(inode)->i_crtime)) 3163 return false; 3164 return true; 3165 } 3166 3167 static inline bool f2fs_skip_inode_update(struct inode *inode, int dsync) 3168 { 3169 bool ret; 3170 3171 if (dsync) { 3172 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 3173 3174 spin_lock(&sbi->inode_lock[DIRTY_META]); 3175 ret = list_empty(&F2FS_I(inode)->gdirty_list); 3176 spin_unlock(&sbi->inode_lock[DIRTY_META]); 3177 return ret; 3178 } 3179 if (!is_inode_flag_set(inode, FI_AUTO_RECOVER) || 3180 file_keep_isize(inode) || 3181 i_size_read(inode) & ~PAGE_MASK) 3182 return false; 3183 3184 if (!f2fs_is_time_consistent(inode)) 3185 return false; 3186 3187 spin_lock(&F2FS_I(inode)->i_size_lock); 3188 ret = F2FS_I(inode)->last_disk_size == i_size_read(inode); 3189 spin_unlock(&F2FS_I(inode)->i_size_lock); 3190 3191 return ret; 3192 } 3193 3194 static inline bool f2fs_readonly(struct super_block *sb) 3195 { 3196 return sb_rdonly(sb); 3197 } 3198 3199 static inline bool f2fs_cp_error(struct f2fs_sb_info *sbi) 3200 { 3201 return is_set_ckpt_flags(sbi, CP_ERROR_FLAG); 3202 } 3203 3204 static inline bool is_dot_dotdot(const u8 *name, size_t len) 3205 { 3206 if (len == 1 && name[0] == '.') 3207 return true; 3208 3209 if (len == 2 && name[0] == '.' && name[1] == '.') 3210 return true; 3211 3212 return false; 3213 } 3214 3215 static inline void *f2fs_kmalloc(struct f2fs_sb_info *sbi, 3216 size_t size, gfp_t flags) 3217 { 3218 if (time_to_inject(sbi, FAULT_KMALLOC)) { 3219 f2fs_show_injection_info(sbi, FAULT_KMALLOC); 3220 return NULL; 3221 } 3222 3223 return kmalloc(size, flags); 3224 } 3225 3226 static inline void *f2fs_kzalloc(struct f2fs_sb_info *sbi, 3227 size_t size, gfp_t flags) 3228 { 3229 return f2fs_kmalloc(sbi, size, flags | __GFP_ZERO); 3230 } 3231 3232 static inline void *f2fs_kvmalloc(struct f2fs_sb_info *sbi, 3233 size_t size, gfp_t flags) 3234 { 3235 if (time_to_inject(sbi, FAULT_KVMALLOC)) { 3236 f2fs_show_injection_info(sbi, FAULT_KVMALLOC); 3237 return NULL; 3238 } 3239 3240 return kvmalloc(size, flags); 3241 } 3242 3243 static inline void *f2fs_kvzalloc(struct f2fs_sb_info *sbi, 3244 size_t size, gfp_t flags) 3245 { 3246 return f2fs_kvmalloc(sbi, size, flags | __GFP_ZERO); 3247 } 3248 3249 static inline int get_extra_isize(struct inode *inode) 3250 { 3251 return F2FS_I(inode)->i_extra_isize / sizeof(__le32); 3252 } 3253 3254 static inline int get_inline_xattr_addrs(struct inode *inode) 3255 { 3256 return F2FS_I(inode)->i_inline_xattr_size; 3257 } 3258 3259 #define f2fs_get_inode_mode(i) \ 3260 ((is_inode_flag_set(i, FI_ACL_MODE)) ? \ 3261 (F2FS_I(i)->i_acl_mode) : ((i)->i_mode)) 3262 3263 #define F2FS_TOTAL_EXTRA_ATTR_SIZE \ 3264 (offsetof(struct f2fs_inode, i_extra_end) - \ 3265 offsetof(struct f2fs_inode, i_extra_isize)) \ 3266 3267 #define F2FS_OLD_ATTRIBUTE_SIZE (offsetof(struct f2fs_inode, i_addr)) 3268 #define F2FS_FITS_IN_INODE(f2fs_inode, extra_isize, field) \ 3269 ((offsetof(typeof(*(f2fs_inode)), field) + \ 3270 sizeof((f2fs_inode)->field)) \ 3271 <= (F2FS_OLD_ATTRIBUTE_SIZE + (extra_isize))) \ 3272 3273 #define __is_large_section(sbi) ((sbi)->segs_per_sec > 1) 3274 3275 #define __is_meta_io(fio) (PAGE_TYPE_OF_BIO((fio)->type) == META) 3276 3277 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi, 3278 block_t blkaddr, int type); 3279 static inline void verify_blkaddr(struct f2fs_sb_info *sbi, 3280 block_t blkaddr, int type) 3281 { 3282 if (!f2fs_is_valid_blkaddr(sbi, blkaddr, type)) { 3283 f2fs_err(sbi, "invalid blkaddr: %u, type: %d, run fsck to fix.", 3284 blkaddr, type); 3285 f2fs_bug_on(sbi, 1); 3286 } 3287 } 3288 3289 static inline bool __is_valid_data_blkaddr(block_t blkaddr) 3290 { 3291 if (blkaddr == NEW_ADDR || blkaddr == NULL_ADDR || 3292 blkaddr == COMPRESS_ADDR) 3293 return false; 3294 return true; 3295 } 3296 3297 /* 3298 * file.c 3299 */ 3300 int f2fs_sync_file(struct file *file, loff_t start, loff_t end, int datasync); 3301 void f2fs_truncate_data_blocks(struct dnode_of_data *dn); 3302 int f2fs_do_truncate_blocks(struct inode *inode, u64 from, bool lock); 3303 int f2fs_truncate_blocks(struct inode *inode, u64 from, bool lock); 3304 int f2fs_truncate(struct inode *inode); 3305 int f2fs_getattr(struct user_namespace *mnt_userns, const struct path *path, 3306 struct kstat *stat, u32 request_mask, unsigned int flags); 3307 int f2fs_setattr(struct user_namespace *mnt_userns, struct dentry *dentry, 3308 struct iattr *attr); 3309 int f2fs_truncate_hole(struct inode *inode, pgoff_t pg_start, pgoff_t pg_end); 3310 void f2fs_truncate_data_blocks_range(struct dnode_of_data *dn, int count); 3311 int f2fs_precache_extents(struct inode *inode); 3312 int f2fs_fileattr_get(struct dentry *dentry, struct fileattr *fa); 3313 int f2fs_fileattr_set(struct user_namespace *mnt_userns, 3314 struct dentry *dentry, struct fileattr *fa); 3315 long f2fs_ioctl(struct file *filp, unsigned int cmd, unsigned long arg); 3316 long f2fs_compat_ioctl(struct file *file, unsigned int cmd, unsigned long arg); 3317 int f2fs_transfer_project_quota(struct inode *inode, kprojid_t kprojid); 3318 int f2fs_pin_file_control(struct inode *inode, bool inc); 3319 3320 /* 3321 * inode.c 3322 */ 3323 void f2fs_set_inode_flags(struct inode *inode); 3324 bool f2fs_inode_chksum_verify(struct f2fs_sb_info *sbi, struct page *page); 3325 void f2fs_inode_chksum_set(struct f2fs_sb_info *sbi, struct page *page); 3326 struct inode *f2fs_iget(struct super_block *sb, unsigned long ino); 3327 struct inode *f2fs_iget_retry(struct super_block *sb, unsigned long ino); 3328 int f2fs_try_to_free_nats(struct f2fs_sb_info *sbi, int nr_shrink); 3329 void f2fs_update_inode(struct inode *inode, struct page *node_page); 3330 void f2fs_update_inode_page(struct inode *inode); 3331 int f2fs_write_inode(struct inode *inode, struct writeback_control *wbc); 3332 void f2fs_evict_inode(struct inode *inode); 3333 void f2fs_handle_failed_inode(struct inode *inode); 3334 3335 /* 3336 * namei.c 3337 */ 3338 int f2fs_update_extension_list(struct f2fs_sb_info *sbi, const char *name, 3339 bool hot, bool set); 3340 struct dentry *f2fs_get_parent(struct dentry *child); 3341 3342 /* 3343 * dir.c 3344 */ 3345 unsigned char f2fs_get_de_type(struct f2fs_dir_entry *de); 3346 int f2fs_init_casefolded_name(const struct inode *dir, 3347 struct f2fs_filename *fname); 3348 int f2fs_setup_filename(struct inode *dir, const struct qstr *iname, 3349 int lookup, struct f2fs_filename *fname); 3350 int f2fs_prepare_lookup(struct inode *dir, struct dentry *dentry, 3351 struct f2fs_filename *fname); 3352 void f2fs_free_filename(struct f2fs_filename *fname); 3353 struct f2fs_dir_entry *f2fs_find_target_dentry(const struct f2fs_dentry_ptr *d, 3354 const struct f2fs_filename *fname, int *max_slots); 3355 int f2fs_fill_dentries(struct dir_context *ctx, struct f2fs_dentry_ptr *d, 3356 unsigned int start_pos, struct fscrypt_str *fstr); 3357 void f2fs_do_make_empty_dir(struct inode *inode, struct inode *parent, 3358 struct f2fs_dentry_ptr *d); 3359 struct page *f2fs_init_inode_metadata(struct inode *inode, struct inode *dir, 3360 const struct f2fs_filename *fname, struct page *dpage); 3361 void f2fs_update_parent_metadata(struct inode *dir, struct inode *inode, 3362 unsigned int current_depth); 3363 int f2fs_room_for_filename(const void *bitmap, int slots, int max_slots); 3364 void f2fs_drop_nlink(struct inode *dir, struct inode *inode); 3365 struct f2fs_dir_entry *__f2fs_find_entry(struct inode *dir, 3366 const struct f2fs_filename *fname, 3367 struct page **res_page); 3368 struct f2fs_dir_entry *f2fs_find_entry(struct inode *dir, 3369 const struct qstr *child, struct page **res_page); 3370 struct f2fs_dir_entry *f2fs_parent_dir(struct inode *dir, struct page **p); 3371 ino_t f2fs_inode_by_name(struct inode *dir, const struct qstr *qstr, 3372 struct page **page); 3373 void f2fs_set_link(struct inode *dir, struct f2fs_dir_entry *de, 3374 struct page *page, struct inode *inode); 3375 bool f2fs_has_enough_room(struct inode *dir, struct page *ipage, 3376 const struct f2fs_filename *fname); 3377 void f2fs_update_dentry(nid_t ino, umode_t mode, struct f2fs_dentry_ptr *d, 3378 const struct fscrypt_str *name, f2fs_hash_t name_hash, 3379 unsigned int bit_pos); 3380 int f2fs_add_regular_entry(struct inode *dir, const struct f2fs_filename *fname, 3381 struct inode *inode, nid_t ino, umode_t mode); 3382 int f2fs_add_dentry(struct inode *dir, const struct f2fs_filename *fname, 3383 struct inode *inode, nid_t ino, umode_t mode); 3384 int f2fs_do_add_link(struct inode *dir, const struct qstr *name, 3385 struct inode *inode, nid_t ino, umode_t mode); 3386 void f2fs_delete_entry(struct f2fs_dir_entry *dentry, struct page *page, 3387 struct inode *dir, struct inode *inode); 3388 int f2fs_do_tmpfile(struct inode *inode, struct inode *dir); 3389 bool f2fs_empty_dir(struct inode *dir); 3390 3391 static inline int f2fs_add_link(struct dentry *dentry, struct inode *inode) 3392 { 3393 if (fscrypt_is_nokey_name(dentry)) 3394 return -ENOKEY; 3395 return f2fs_do_add_link(d_inode(dentry->d_parent), &dentry->d_name, 3396 inode, inode->i_ino, inode->i_mode); 3397 } 3398 3399 /* 3400 * super.c 3401 */ 3402 int f2fs_inode_dirtied(struct inode *inode, bool sync); 3403 void f2fs_inode_synced(struct inode *inode); 3404 int f2fs_dquot_initialize(struct inode *inode); 3405 int f2fs_enable_quota_files(struct f2fs_sb_info *sbi, bool rdonly); 3406 int f2fs_quota_sync(struct super_block *sb, int type); 3407 loff_t max_file_blocks(struct inode *inode); 3408 void f2fs_quota_off_umount(struct super_block *sb); 3409 int f2fs_commit_super(struct f2fs_sb_info *sbi, bool recover); 3410 int f2fs_sync_fs(struct super_block *sb, int sync); 3411 int f2fs_sanity_check_ckpt(struct f2fs_sb_info *sbi); 3412 3413 /* 3414 * hash.c 3415 */ 3416 void f2fs_hash_filename(const struct inode *dir, struct f2fs_filename *fname); 3417 3418 /* 3419 * node.c 3420 */ 3421 struct node_info; 3422 3423 int f2fs_check_nid_range(struct f2fs_sb_info *sbi, nid_t nid); 3424 bool f2fs_available_free_memory(struct f2fs_sb_info *sbi, int type); 3425 bool f2fs_in_warm_node_list(struct f2fs_sb_info *sbi, struct page *page); 3426 void f2fs_init_fsync_node_info(struct f2fs_sb_info *sbi); 3427 void f2fs_del_fsync_node_entry(struct f2fs_sb_info *sbi, struct page *page); 3428 void f2fs_reset_fsync_node_info(struct f2fs_sb_info *sbi); 3429 int f2fs_need_dentry_mark(struct f2fs_sb_info *sbi, nid_t nid); 3430 bool f2fs_is_checkpointed_node(struct f2fs_sb_info *sbi, nid_t nid); 3431 bool f2fs_need_inode_block_update(struct f2fs_sb_info *sbi, nid_t ino); 3432 int f2fs_get_node_info(struct f2fs_sb_info *sbi, nid_t nid, 3433 struct node_info *ni, bool checkpoint_context); 3434 pgoff_t f2fs_get_next_page_offset(struct dnode_of_data *dn, pgoff_t pgofs); 3435 int f2fs_get_dnode_of_data(struct dnode_of_data *dn, pgoff_t index, int mode); 3436 int f2fs_truncate_inode_blocks(struct inode *inode, pgoff_t from); 3437 int f2fs_truncate_xattr_node(struct inode *inode); 3438 int f2fs_wait_on_node_pages_writeback(struct f2fs_sb_info *sbi, 3439 unsigned int seq_id); 3440 bool f2fs_nat_bitmap_enabled(struct f2fs_sb_info *sbi); 3441 int f2fs_remove_inode_page(struct inode *inode); 3442 struct page *f2fs_new_inode_page(struct inode *inode); 3443 struct page *f2fs_new_node_page(struct dnode_of_data *dn, unsigned int ofs); 3444 void f2fs_ra_node_page(struct f2fs_sb_info *sbi, nid_t nid); 3445 struct page *f2fs_get_node_page(struct f2fs_sb_info *sbi, pgoff_t nid); 3446 struct page *f2fs_get_node_page_ra(struct page *parent, int start); 3447 int f2fs_move_node_page(struct page *node_page, int gc_type); 3448 void f2fs_flush_inline_data(struct f2fs_sb_info *sbi); 3449 int f2fs_fsync_node_pages(struct f2fs_sb_info *sbi, struct inode *inode, 3450 struct writeback_control *wbc, bool atomic, 3451 unsigned int *seq_id); 3452 int f2fs_sync_node_pages(struct f2fs_sb_info *sbi, 3453 struct writeback_control *wbc, 3454 bool do_balance, enum iostat_type io_type); 3455 int f2fs_build_free_nids(struct f2fs_sb_info *sbi, bool sync, bool mount); 3456 bool f2fs_alloc_nid(struct f2fs_sb_info *sbi, nid_t *nid); 3457 void f2fs_alloc_nid_done(struct f2fs_sb_info *sbi, nid_t nid); 3458 void f2fs_alloc_nid_failed(struct f2fs_sb_info *sbi, nid_t nid); 3459 int f2fs_try_to_free_nids(struct f2fs_sb_info *sbi, int nr_shrink); 3460 int f2fs_recover_inline_xattr(struct inode *inode, struct page *page); 3461 int f2fs_recover_xattr_data(struct inode *inode, struct page *page); 3462 int f2fs_recover_inode_page(struct f2fs_sb_info *sbi, struct page *page); 3463 int f2fs_restore_node_summary(struct f2fs_sb_info *sbi, 3464 unsigned int segno, struct f2fs_summary_block *sum); 3465 void f2fs_enable_nat_bits(struct f2fs_sb_info *sbi); 3466 int f2fs_flush_nat_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3467 int f2fs_build_node_manager(struct f2fs_sb_info *sbi); 3468 void f2fs_destroy_node_manager(struct f2fs_sb_info *sbi); 3469 int __init f2fs_create_node_manager_caches(void); 3470 void f2fs_destroy_node_manager_caches(void); 3471 3472 /* 3473 * segment.c 3474 */ 3475 bool f2fs_need_SSR(struct f2fs_sb_info *sbi); 3476 void f2fs_register_inmem_page(struct inode *inode, struct page *page); 3477 void f2fs_drop_inmem_pages_all(struct f2fs_sb_info *sbi, bool gc_failure); 3478 void f2fs_drop_inmem_pages(struct inode *inode); 3479 void f2fs_drop_inmem_page(struct inode *inode, struct page *page); 3480 int f2fs_commit_inmem_pages(struct inode *inode); 3481 void f2fs_balance_fs(struct f2fs_sb_info *sbi, bool need); 3482 void f2fs_balance_fs_bg(struct f2fs_sb_info *sbi, bool from_bg); 3483 int f2fs_issue_flush(struct f2fs_sb_info *sbi, nid_t ino); 3484 int f2fs_create_flush_cmd_control(struct f2fs_sb_info *sbi); 3485 int f2fs_flush_device_cache(struct f2fs_sb_info *sbi); 3486 void f2fs_destroy_flush_cmd_control(struct f2fs_sb_info *sbi, bool free); 3487 void f2fs_invalidate_blocks(struct f2fs_sb_info *sbi, block_t addr); 3488 bool f2fs_is_checkpointed_data(struct f2fs_sb_info *sbi, block_t blkaddr); 3489 int f2fs_start_discard_thread(struct f2fs_sb_info *sbi); 3490 void f2fs_drop_discard_cmd(struct f2fs_sb_info *sbi); 3491 void f2fs_stop_discard_thread(struct f2fs_sb_info *sbi); 3492 bool f2fs_issue_discard_timeout(struct f2fs_sb_info *sbi); 3493 void f2fs_clear_prefree_segments(struct f2fs_sb_info *sbi, 3494 struct cp_control *cpc); 3495 void f2fs_dirty_to_prefree(struct f2fs_sb_info *sbi); 3496 block_t f2fs_get_unusable_blocks(struct f2fs_sb_info *sbi); 3497 int f2fs_disable_cp_again(struct f2fs_sb_info *sbi, block_t unusable); 3498 void f2fs_release_discard_addrs(struct f2fs_sb_info *sbi); 3499 int f2fs_npages_for_summary_flush(struct f2fs_sb_info *sbi, bool for_ra); 3500 bool f2fs_segment_has_free_slot(struct f2fs_sb_info *sbi, int segno); 3501 void f2fs_init_inmem_curseg(struct f2fs_sb_info *sbi); 3502 void f2fs_save_inmem_curseg(struct f2fs_sb_info *sbi); 3503 void f2fs_restore_inmem_curseg(struct f2fs_sb_info *sbi); 3504 void f2fs_get_new_segment(struct f2fs_sb_info *sbi, 3505 unsigned int *newseg, bool new_sec, int dir); 3506 void f2fs_allocate_segment_for_resize(struct f2fs_sb_info *sbi, int type, 3507 unsigned int start, unsigned int end); 3508 void f2fs_allocate_new_section(struct f2fs_sb_info *sbi, int type, bool force); 3509 void f2fs_allocate_new_segments(struct f2fs_sb_info *sbi); 3510 int f2fs_trim_fs(struct f2fs_sb_info *sbi, struct fstrim_range *range); 3511 bool f2fs_exist_trim_candidates(struct f2fs_sb_info *sbi, 3512 struct cp_control *cpc); 3513 struct page *f2fs_get_sum_page(struct f2fs_sb_info *sbi, unsigned int segno); 3514 void f2fs_update_meta_page(struct f2fs_sb_info *sbi, void *src, 3515 block_t blk_addr); 3516 void f2fs_do_write_meta_page(struct f2fs_sb_info *sbi, struct page *page, 3517 enum iostat_type io_type); 3518 void f2fs_do_write_node_page(unsigned int nid, struct f2fs_io_info *fio); 3519 void f2fs_outplace_write_data(struct dnode_of_data *dn, 3520 struct f2fs_io_info *fio); 3521 int f2fs_inplace_write_data(struct f2fs_io_info *fio); 3522 void f2fs_do_replace_block(struct f2fs_sb_info *sbi, struct f2fs_summary *sum, 3523 block_t old_blkaddr, block_t new_blkaddr, 3524 bool recover_curseg, bool recover_newaddr, 3525 bool from_gc); 3526 void f2fs_replace_block(struct f2fs_sb_info *sbi, struct dnode_of_data *dn, 3527 block_t old_addr, block_t new_addr, 3528 unsigned char version, bool recover_curseg, 3529 bool recover_newaddr); 3530 void f2fs_allocate_data_block(struct f2fs_sb_info *sbi, struct page *page, 3531 block_t old_blkaddr, block_t *new_blkaddr, 3532 struct f2fs_summary *sum, int type, 3533 struct f2fs_io_info *fio); 3534 void f2fs_update_device_state(struct f2fs_sb_info *sbi, nid_t ino, 3535 block_t blkaddr, unsigned int blkcnt); 3536 void f2fs_wait_on_page_writeback(struct page *page, 3537 enum page_type type, bool ordered, bool locked); 3538 void f2fs_wait_on_block_writeback(struct inode *inode, block_t blkaddr); 3539 void f2fs_wait_on_block_writeback_range(struct inode *inode, block_t blkaddr, 3540 block_t len); 3541 void f2fs_write_data_summaries(struct f2fs_sb_info *sbi, block_t start_blk); 3542 void f2fs_write_node_summaries(struct f2fs_sb_info *sbi, block_t start_blk); 3543 int f2fs_lookup_journal_in_cursum(struct f2fs_journal *journal, int type, 3544 unsigned int val, int alloc); 3545 void f2fs_flush_sit_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3546 int f2fs_fix_curseg_write_pointer(struct f2fs_sb_info *sbi); 3547 int f2fs_check_write_pointer(struct f2fs_sb_info *sbi); 3548 int f2fs_build_segment_manager(struct f2fs_sb_info *sbi); 3549 void f2fs_destroy_segment_manager(struct f2fs_sb_info *sbi); 3550 int __init f2fs_create_segment_manager_caches(void); 3551 void f2fs_destroy_segment_manager_caches(void); 3552 int f2fs_rw_hint_to_seg_type(enum rw_hint hint); 3553 enum rw_hint f2fs_io_type_to_rw_hint(struct f2fs_sb_info *sbi, 3554 enum page_type type, enum temp_type temp); 3555 unsigned int f2fs_usable_segs_in_sec(struct f2fs_sb_info *sbi, 3556 unsigned int segno); 3557 unsigned int f2fs_usable_blks_in_seg(struct f2fs_sb_info *sbi, 3558 unsigned int segno); 3559 3560 #define DEF_FRAGMENT_SIZE 4 3561 #define MIN_FRAGMENT_SIZE 1 3562 #define MAX_FRAGMENT_SIZE 512 3563 3564 static inline bool f2fs_need_rand_seg(struct f2fs_sb_info *sbi) 3565 { 3566 return F2FS_OPTION(sbi).fs_mode == FS_MODE_FRAGMENT_SEG || 3567 F2FS_OPTION(sbi).fs_mode == FS_MODE_FRAGMENT_BLK; 3568 } 3569 3570 /* 3571 * checkpoint.c 3572 */ 3573 void f2fs_stop_checkpoint(struct f2fs_sb_info *sbi, bool end_io); 3574 struct page *f2fs_grab_meta_page(struct f2fs_sb_info *sbi, pgoff_t index); 3575 struct page *f2fs_get_meta_page(struct f2fs_sb_info *sbi, pgoff_t index); 3576 struct page *f2fs_get_meta_page_retry(struct f2fs_sb_info *sbi, pgoff_t index); 3577 struct page *f2fs_get_tmp_page(struct f2fs_sb_info *sbi, pgoff_t index); 3578 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi, 3579 block_t blkaddr, int type); 3580 int f2fs_ra_meta_pages(struct f2fs_sb_info *sbi, block_t start, int nrpages, 3581 int type, bool sync); 3582 void f2fs_ra_meta_pages_cond(struct f2fs_sb_info *sbi, pgoff_t index); 3583 long f2fs_sync_meta_pages(struct f2fs_sb_info *sbi, enum page_type type, 3584 long nr_to_write, enum iostat_type io_type); 3585 void f2fs_add_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type); 3586 void f2fs_remove_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type); 3587 void f2fs_release_ino_entry(struct f2fs_sb_info *sbi, bool all); 3588 bool f2fs_exist_written_data(struct f2fs_sb_info *sbi, nid_t ino, int mode); 3589 void f2fs_set_dirty_device(struct f2fs_sb_info *sbi, nid_t ino, 3590 unsigned int devidx, int type); 3591 bool f2fs_is_dirty_device(struct f2fs_sb_info *sbi, nid_t ino, 3592 unsigned int devidx, int type); 3593 int f2fs_sync_inode_meta(struct f2fs_sb_info *sbi); 3594 int f2fs_acquire_orphan_inode(struct f2fs_sb_info *sbi); 3595 void f2fs_release_orphan_inode(struct f2fs_sb_info *sbi); 3596 void f2fs_add_orphan_inode(struct inode *inode); 3597 void f2fs_remove_orphan_inode(struct f2fs_sb_info *sbi, nid_t ino); 3598 int f2fs_recover_orphan_inodes(struct f2fs_sb_info *sbi); 3599 int f2fs_get_valid_checkpoint(struct f2fs_sb_info *sbi); 3600 void f2fs_update_dirty_page(struct inode *inode, struct page *page); 3601 void f2fs_remove_dirty_inode(struct inode *inode); 3602 int f2fs_sync_dirty_inodes(struct f2fs_sb_info *sbi, enum inode_type type); 3603 void f2fs_wait_on_all_pages(struct f2fs_sb_info *sbi, int type); 3604 u64 f2fs_get_sectors_written(struct f2fs_sb_info *sbi); 3605 int f2fs_write_checkpoint(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3606 void f2fs_init_ino_entry_info(struct f2fs_sb_info *sbi); 3607 int __init f2fs_create_checkpoint_caches(void); 3608 void f2fs_destroy_checkpoint_caches(void); 3609 int f2fs_issue_checkpoint(struct f2fs_sb_info *sbi); 3610 int f2fs_start_ckpt_thread(struct f2fs_sb_info *sbi); 3611 void f2fs_stop_ckpt_thread(struct f2fs_sb_info *sbi); 3612 void f2fs_init_ckpt_req_control(struct f2fs_sb_info *sbi); 3613 3614 /* 3615 * data.c 3616 */ 3617 int __init f2fs_init_bioset(void); 3618 void f2fs_destroy_bioset(void); 3619 int f2fs_init_bio_entry_cache(void); 3620 void f2fs_destroy_bio_entry_cache(void); 3621 void f2fs_submit_bio(struct f2fs_sb_info *sbi, 3622 struct bio *bio, enum page_type type); 3623 void f2fs_submit_merged_write(struct f2fs_sb_info *sbi, enum page_type type); 3624 void f2fs_submit_merged_write_cond(struct f2fs_sb_info *sbi, 3625 struct inode *inode, struct page *page, 3626 nid_t ino, enum page_type type); 3627 void f2fs_submit_merged_ipu_write(struct f2fs_sb_info *sbi, 3628 struct bio **bio, struct page *page); 3629 void f2fs_flush_merged_writes(struct f2fs_sb_info *sbi); 3630 int f2fs_submit_page_bio(struct f2fs_io_info *fio); 3631 int f2fs_merge_page_bio(struct f2fs_io_info *fio); 3632 void f2fs_submit_page_write(struct f2fs_io_info *fio); 3633 struct block_device *f2fs_target_device(struct f2fs_sb_info *sbi, 3634 block_t blk_addr, struct bio *bio); 3635 int f2fs_target_device_index(struct f2fs_sb_info *sbi, block_t blkaddr); 3636 void f2fs_set_data_blkaddr(struct dnode_of_data *dn); 3637 void f2fs_update_data_blkaddr(struct dnode_of_data *dn, block_t blkaddr); 3638 int f2fs_reserve_new_blocks(struct dnode_of_data *dn, blkcnt_t count); 3639 int f2fs_reserve_new_block(struct dnode_of_data *dn); 3640 int f2fs_get_block(struct dnode_of_data *dn, pgoff_t index); 3641 int f2fs_reserve_block(struct dnode_of_data *dn, pgoff_t index); 3642 struct page *f2fs_get_read_data_page(struct inode *inode, pgoff_t index, 3643 int op_flags, bool for_write); 3644 struct page *f2fs_find_data_page(struct inode *inode, pgoff_t index); 3645 struct page *f2fs_get_lock_data_page(struct inode *inode, pgoff_t index, 3646 bool for_write); 3647 struct page *f2fs_get_new_data_page(struct inode *inode, 3648 struct page *ipage, pgoff_t index, bool new_i_size); 3649 int f2fs_do_write_data_page(struct f2fs_io_info *fio); 3650 void f2fs_do_map_lock(struct f2fs_sb_info *sbi, int flag, bool lock); 3651 int f2fs_map_blocks(struct inode *inode, struct f2fs_map_blocks *map, 3652 int create, int flag); 3653 int f2fs_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, 3654 u64 start, u64 len); 3655 int f2fs_encrypt_one_page(struct f2fs_io_info *fio); 3656 bool f2fs_should_update_inplace(struct inode *inode, struct f2fs_io_info *fio); 3657 bool f2fs_should_update_outplace(struct inode *inode, struct f2fs_io_info *fio); 3658 int f2fs_write_single_data_page(struct page *page, int *submitted, 3659 struct bio **bio, sector_t *last_block, 3660 struct writeback_control *wbc, 3661 enum iostat_type io_type, 3662 int compr_blocks, bool allow_balance); 3663 void f2fs_write_failed(struct inode *inode, loff_t to); 3664 void f2fs_invalidate_page(struct page *page, unsigned int offset, 3665 unsigned int length); 3666 int f2fs_release_page(struct page *page, gfp_t wait); 3667 #ifdef CONFIG_MIGRATION 3668 int f2fs_migrate_page(struct address_space *mapping, struct page *newpage, 3669 struct page *page, enum migrate_mode mode); 3670 #endif 3671 bool f2fs_overwrite_io(struct inode *inode, loff_t pos, size_t len); 3672 void f2fs_clear_page_cache_dirty_tag(struct page *page); 3673 int f2fs_init_post_read_processing(void); 3674 void f2fs_destroy_post_read_processing(void); 3675 int f2fs_init_post_read_wq(struct f2fs_sb_info *sbi); 3676 void f2fs_destroy_post_read_wq(struct f2fs_sb_info *sbi); 3677 extern const struct iomap_ops f2fs_iomap_ops; 3678 3679 /* 3680 * gc.c 3681 */ 3682 int f2fs_start_gc_thread(struct f2fs_sb_info *sbi); 3683 void f2fs_stop_gc_thread(struct f2fs_sb_info *sbi); 3684 block_t f2fs_start_bidx_of_node(unsigned int node_ofs, struct inode *inode); 3685 int f2fs_gc(struct f2fs_sb_info *sbi, bool sync, bool background, bool force, 3686 unsigned int segno); 3687 void f2fs_build_gc_manager(struct f2fs_sb_info *sbi); 3688 int f2fs_resize_fs(struct f2fs_sb_info *sbi, __u64 block_count); 3689 int __init f2fs_create_garbage_collection_cache(void); 3690 void f2fs_destroy_garbage_collection_cache(void); 3691 3692 /* 3693 * recovery.c 3694 */ 3695 int f2fs_recover_fsync_data(struct f2fs_sb_info *sbi, bool check_only); 3696 bool f2fs_space_for_roll_forward(struct f2fs_sb_info *sbi); 3697 int __init f2fs_create_recovery_cache(void); 3698 void f2fs_destroy_recovery_cache(void); 3699 3700 /* 3701 * debug.c 3702 */ 3703 #ifdef CONFIG_F2FS_STAT_FS 3704 struct f2fs_stat_info { 3705 struct list_head stat_list; 3706 struct f2fs_sb_info *sbi; 3707 int all_area_segs, sit_area_segs, nat_area_segs, ssa_area_segs; 3708 int main_area_segs, main_area_sections, main_area_zones; 3709 unsigned long long hit_largest, hit_cached, hit_rbtree; 3710 unsigned long long hit_total, total_ext; 3711 int ext_tree, zombie_tree, ext_node; 3712 int ndirty_node, ndirty_dent, ndirty_meta, ndirty_imeta; 3713 int ndirty_data, ndirty_qdata; 3714 int inmem_pages; 3715 unsigned int ndirty_dirs, ndirty_files, nquota_files, ndirty_all; 3716 int nats, dirty_nats, sits, dirty_sits; 3717 int free_nids, avail_nids, alloc_nids; 3718 int total_count, utilization; 3719 int bg_gc, nr_wb_cp_data, nr_wb_data; 3720 int nr_rd_data, nr_rd_node, nr_rd_meta; 3721 int nr_dio_read, nr_dio_write; 3722 unsigned int io_skip_bggc, other_skip_bggc; 3723 int nr_flushing, nr_flushed, flush_list_empty; 3724 int nr_discarding, nr_discarded; 3725 int nr_discard_cmd; 3726 unsigned int undiscard_blks; 3727 int nr_issued_ckpt, nr_total_ckpt, nr_queued_ckpt; 3728 unsigned int cur_ckpt_time, peak_ckpt_time; 3729 int inline_xattr, inline_inode, inline_dir, append, update, orphans; 3730 int compr_inode; 3731 unsigned long long compr_blocks; 3732 int aw_cnt, max_aw_cnt, vw_cnt, max_vw_cnt; 3733 unsigned int valid_count, valid_node_count, valid_inode_count, discard_blks; 3734 unsigned int bimodal, avg_vblocks; 3735 int util_free, util_valid, util_invalid; 3736 int rsvd_segs, overp_segs; 3737 int dirty_count, node_pages, meta_pages, compress_pages; 3738 int compress_page_hit; 3739 int prefree_count, call_count, cp_count, bg_cp_count; 3740 int tot_segs, node_segs, data_segs, free_segs, free_secs; 3741 int bg_node_segs, bg_data_segs; 3742 int tot_blks, data_blks, node_blks; 3743 int bg_data_blks, bg_node_blks; 3744 unsigned long long skipped_atomic_files[2]; 3745 int curseg[NR_CURSEG_TYPE]; 3746 int cursec[NR_CURSEG_TYPE]; 3747 int curzone[NR_CURSEG_TYPE]; 3748 unsigned int dirty_seg[NR_CURSEG_TYPE]; 3749 unsigned int full_seg[NR_CURSEG_TYPE]; 3750 unsigned int valid_blks[NR_CURSEG_TYPE]; 3751 3752 unsigned int meta_count[META_MAX]; 3753 unsigned int segment_count[2]; 3754 unsigned int block_count[2]; 3755 unsigned int inplace_count; 3756 unsigned long long base_mem, cache_mem, page_mem; 3757 }; 3758 3759 static inline struct f2fs_stat_info *F2FS_STAT(struct f2fs_sb_info *sbi) 3760 { 3761 return (struct f2fs_stat_info *)sbi->stat_info; 3762 } 3763 3764 #define stat_inc_cp_count(si) ((si)->cp_count++) 3765 #define stat_inc_bg_cp_count(si) ((si)->bg_cp_count++) 3766 #define stat_inc_call_count(si) ((si)->call_count++) 3767 #define stat_inc_bggc_count(si) ((si)->bg_gc++) 3768 #define stat_io_skip_bggc_count(sbi) ((sbi)->io_skip_bggc++) 3769 #define stat_other_skip_bggc_count(sbi) ((sbi)->other_skip_bggc++) 3770 #define stat_inc_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]++) 3771 #define stat_dec_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]--) 3772 #define stat_inc_total_hit(sbi) (atomic64_inc(&(sbi)->total_hit_ext)) 3773 #define stat_inc_rbtree_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_rbtree)) 3774 #define stat_inc_largest_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_largest)) 3775 #define stat_inc_cached_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_cached)) 3776 #define stat_inc_inline_xattr(inode) \ 3777 do { \ 3778 if (f2fs_has_inline_xattr(inode)) \ 3779 (atomic_inc(&F2FS_I_SB(inode)->inline_xattr)); \ 3780 } while (0) 3781 #define stat_dec_inline_xattr(inode) \ 3782 do { \ 3783 if (f2fs_has_inline_xattr(inode)) \ 3784 (atomic_dec(&F2FS_I_SB(inode)->inline_xattr)); \ 3785 } while (0) 3786 #define stat_inc_inline_inode(inode) \ 3787 do { \ 3788 if (f2fs_has_inline_data(inode)) \ 3789 (atomic_inc(&F2FS_I_SB(inode)->inline_inode)); \ 3790 } while (0) 3791 #define stat_dec_inline_inode(inode) \ 3792 do { \ 3793 if (f2fs_has_inline_data(inode)) \ 3794 (atomic_dec(&F2FS_I_SB(inode)->inline_inode)); \ 3795 } while (0) 3796 #define stat_inc_inline_dir(inode) \ 3797 do { \ 3798 if (f2fs_has_inline_dentry(inode)) \ 3799 (atomic_inc(&F2FS_I_SB(inode)->inline_dir)); \ 3800 } while (0) 3801 #define stat_dec_inline_dir(inode) \ 3802 do { \ 3803 if (f2fs_has_inline_dentry(inode)) \ 3804 (atomic_dec(&F2FS_I_SB(inode)->inline_dir)); \ 3805 } while (0) 3806 #define stat_inc_compr_inode(inode) \ 3807 do { \ 3808 if (f2fs_compressed_file(inode)) \ 3809 (atomic_inc(&F2FS_I_SB(inode)->compr_inode)); \ 3810 } while (0) 3811 #define stat_dec_compr_inode(inode) \ 3812 do { \ 3813 if (f2fs_compressed_file(inode)) \ 3814 (atomic_dec(&F2FS_I_SB(inode)->compr_inode)); \ 3815 } while (0) 3816 #define stat_add_compr_blocks(inode, blocks) \ 3817 (atomic64_add(blocks, &F2FS_I_SB(inode)->compr_blocks)) 3818 #define stat_sub_compr_blocks(inode, blocks) \ 3819 (atomic64_sub(blocks, &F2FS_I_SB(inode)->compr_blocks)) 3820 #define stat_inc_meta_count(sbi, blkaddr) \ 3821 do { \ 3822 if (blkaddr < SIT_I(sbi)->sit_base_addr) \ 3823 atomic_inc(&(sbi)->meta_count[META_CP]); \ 3824 else if (blkaddr < NM_I(sbi)->nat_blkaddr) \ 3825 atomic_inc(&(sbi)->meta_count[META_SIT]); \ 3826 else if (blkaddr < SM_I(sbi)->ssa_blkaddr) \ 3827 atomic_inc(&(sbi)->meta_count[META_NAT]); \ 3828 else if (blkaddr < SM_I(sbi)->main_blkaddr) \ 3829 atomic_inc(&(sbi)->meta_count[META_SSA]); \ 3830 } while (0) 3831 #define stat_inc_seg_type(sbi, curseg) \ 3832 ((sbi)->segment_count[(curseg)->alloc_type]++) 3833 #define stat_inc_block_count(sbi, curseg) \ 3834 ((sbi)->block_count[(curseg)->alloc_type]++) 3835 #define stat_inc_inplace_blocks(sbi) \ 3836 (atomic_inc(&(sbi)->inplace_count)) 3837 #define stat_update_max_atomic_write(inode) \ 3838 do { \ 3839 int cur = F2FS_I_SB(inode)->atomic_files; \ 3840 int max = atomic_read(&F2FS_I_SB(inode)->max_aw_cnt); \ 3841 if (cur > max) \ 3842 atomic_set(&F2FS_I_SB(inode)->max_aw_cnt, cur); \ 3843 } while (0) 3844 #define stat_inc_volatile_write(inode) \ 3845 (atomic_inc(&F2FS_I_SB(inode)->vw_cnt)) 3846 #define stat_dec_volatile_write(inode) \ 3847 (atomic_dec(&F2FS_I_SB(inode)->vw_cnt)) 3848 #define stat_update_max_volatile_write(inode) \ 3849 do { \ 3850 int cur = atomic_read(&F2FS_I_SB(inode)->vw_cnt); \ 3851 int max = atomic_read(&F2FS_I_SB(inode)->max_vw_cnt); \ 3852 if (cur > max) \ 3853 atomic_set(&F2FS_I_SB(inode)->max_vw_cnt, cur); \ 3854 } while (0) 3855 #define stat_inc_seg_count(sbi, type, gc_type) \ 3856 do { \ 3857 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 3858 si->tot_segs++; \ 3859 if ((type) == SUM_TYPE_DATA) { \ 3860 si->data_segs++; \ 3861 si->bg_data_segs += (gc_type == BG_GC) ? 1 : 0; \ 3862 } else { \ 3863 si->node_segs++; \ 3864 si->bg_node_segs += (gc_type == BG_GC) ? 1 : 0; \ 3865 } \ 3866 } while (0) 3867 3868 #define stat_inc_tot_blk_count(si, blks) \ 3869 ((si)->tot_blks += (blks)) 3870 3871 #define stat_inc_data_blk_count(sbi, blks, gc_type) \ 3872 do { \ 3873 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 3874 stat_inc_tot_blk_count(si, blks); \ 3875 si->data_blks += (blks); \ 3876 si->bg_data_blks += ((gc_type) == BG_GC) ? (blks) : 0; \ 3877 } while (0) 3878 3879 #define stat_inc_node_blk_count(sbi, blks, gc_type) \ 3880 do { \ 3881 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 3882 stat_inc_tot_blk_count(si, blks); \ 3883 si->node_blks += (blks); \ 3884 si->bg_node_blks += ((gc_type) == BG_GC) ? (blks) : 0; \ 3885 } while (0) 3886 3887 int f2fs_build_stats(struct f2fs_sb_info *sbi); 3888 void f2fs_destroy_stats(struct f2fs_sb_info *sbi); 3889 void __init f2fs_create_root_stats(void); 3890 void f2fs_destroy_root_stats(void); 3891 void f2fs_update_sit_info(struct f2fs_sb_info *sbi); 3892 #else 3893 #define stat_inc_cp_count(si) do { } while (0) 3894 #define stat_inc_bg_cp_count(si) do { } while (0) 3895 #define stat_inc_call_count(si) do { } while (0) 3896 #define stat_inc_bggc_count(si) do { } while (0) 3897 #define stat_io_skip_bggc_count(sbi) do { } while (0) 3898 #define stat_other_skip_bggc_count(sbi) do { } while (0) 3899 #define stat_inc_dirty_inode(sbi, type) do { } while (0) 3900 #define stat_dec_dirty_inode(sbi, type) do { } while (0) 3901 #define stat_inc_total_hit(sbi) do { } while (0) 3902 #define stat_inc_rbtree_node_hit(sbi) do { } while (0) 3903 #define stat_inc_largest_node_hit(sbi) do { } while (0) 3904 #define stat_inc_cached_node_hit(sbi) do { } while (0) 3905 #define stat_inc_inline_xattr(inode) do { } while (0) 3906 #define stat_dec_inline_xattr(inode) do { } while (0) 3907 #define stat_inc_inline_inode(inode) do { } while (0) 3908 #define stat_dec_inline_inode(inode) do { } while (0) 3909 #define stat_inc_inline_dir(inode) do { } while (0) 3910 #define stat_dec_inline_dir(inode) do { } while (0) 3911 #define stat_inc_compr_inode(inode) do { } while (0) 3912 #define stat_dec_compr_inode(inode) do { } while (0) 3913 #define stat_add_compr_blocks(inode, blocks) do { } while (0) 3914 #define stat_sub_compr_blocks(inode, blocks) do { } while (0) 3915 #define stat_update_max_atomic_write(inode) do { } while (0) 3916 #define stat_inc_volatile_write(inode) do { } while (0) 3917 #define stat_dec_volatile_write(inode) do { } while (0) 3918 #define stat_update_max_volatile_write(inode) do { } while (0) 3919 #define stat_inc_meta_count(sbi, blkaddr) do { } while (0) 3920 #define stat_inc_seg_type(sbi, curseg) do { } while (0) 3921 #define stat_inc_block_count(sbi, curseg) do { } while (0) 3922 #define stat_inc_inplace_blocks(sbi) do { } while (0) 3923 #define stat_inc_seg_count(sbi, type, gc_type) do { } while (0) 3924 #define stat_inc_tot_blk_count(si, blks) do { } while (0) 3925 #define stat_inc_data_blk_count(sbi, blks, gc_type) do { } while (0) 3926 #define stat_inc_node_blk_count(sbi, blks, gc_type) do { } while (0) 3927 3928 static inline int f2fs_build_stats(struct f2fs_sb_info *sbi) { return 0; } 3929 static inline void f2fs_destroy_stats(struct f2fs_sb_info *sbi) { } 3930 static inline void __init f2fs_create_root_stats(void) { } 3931 static inline void f2fs_destroy_root_stats(void) { } 3932 static inline void f2fs_update_sit_info(struct f2fs_sb_info *sbi) {} 3933 #endif 3934 3935 extern const struct file_operations f2fs_dir_operations; 3936 extern const struct file_operations f2fs_file_operations; 3937 extern const struct inode_operations f2fs_file_inode_operations; 3938 extern const struct address_space_operations f2fs_dblock_aops; 3939 extern const struct address_space_operations f2fs_node_aops; 3940 extern const struct address_space_operations f2fs_meta_aops; 3941 extern const struct inode_operations f2fs_dir_inode_operations; 3942 extern const struct inode_operations f2fs_symlink_inode_operations; 3943 extern const struct inode_operations f2fs_encrypted_symlink_inode_operations; 3944 extern const struct inode_operations f2fs_special_inode_operations; 3945 extern struct kmem_cache *f2fs_inode_entry_slab; 3946 3947 /* 3948 * inline.c 3949 */ 3950 bool f2fs_may_inline_data(struct inode *inode); 3951 bool f2fs_may_inline_dentry(struct inode *inode); 3952 void f2fs_do_read_inline_data(struct page *page, struct page *ipage); 3953 void f2fs_truncate_inline_inode(struct inode *inode, 3954 struct page *ipage, u64 from); 3955 int f2fs_read_inline_data(struct inode *inode, struct page *page); 3956 int f2fs_convert_inline_page(struct dnode_of_data *dn, struct page *page); 3957 int f2fs_convert_inline_inode(struct inode *inode); 3958 int f2fs_try_convert_inline_dir(struct inode *dir, struct dentry *dentry); 3959 int f2fs_write_inline_data(struct inode *inode, struct page *page); 3960 int f2fs_recover_inline_data(struct inode *inode, struct page *npage); 3961 struct f2fs_dir_entry *f2fs_find_in_inline_dir(struct inode *dir, 3962 const struct f2fs_filename *fname, 3963 struct page **res_page); 3964 int f2fs_make_empty_inline_dir(struct inode *inode, struct inode *parent, 3965 struct page *ipage); 3966 int f2fs_add_inline_entry(struct inode *dir, const struct f2fs_filename *fname, 3967 struct inode *inode, nid_t ino, umode_t mode); 3968 void f2fs_delete_inline_entry(struct f2fs_dir_entry *dentry, 3969 struct page *page, struct inode *dir, 3970 struct inode *inode); 3971 bool f2fs_empty_inline_dir(struct inode *dir); 3972 int f2fs_read_inline_dir(struct file *file, struct dir_context *ctx, 3973 struct fscrypt_str *fstr); 3974 int f2fs_inline_data_fiemap(struct inode *inode, 3975 struct fiemap_extent_info *fieinfo, 3976 __u64 start, __u64 len); 3977 3978 /* 3979 * shrinker.c 3980 */ 3981 unsigned long f2fs_shrink_count(struct shrinker *shrink, 3982 struct shrink_control *sc); 3983 unsigned long f2fs_shrink_scan(struct shrinker *shrink, 3984 struct shrink_control *sc); 3985 void f2fs_join_shrinker(struct f2fs_sb_info *sbi); 3986 void f2fs_leave_shrinker(struct f2fs_sb_info *sbi); 3987 3988 /* 3989 * extent_cache.c 3990 */ 3991 struct rb_entry *f2fs_lookup_rb_tree(struct rb_root_cached *root, 3992 struct rb_entry *cached_re, unsigned int ofs); 3993 struct rb_node **f2fs_lookup_rb_tree_ext(struct f2fs_sb_info *sbi, 3994 struct rb_root_cached *root, 3995 struct rb_node **parent, 3996 unsigned long long key, bool *left_most); 3997 struct rb_node **f2fs_lookup_rb_tree_for_insert(struct f2fs_sb_info *sbi, 3998 struct rb_root_cached *root, 3999 struct rb_node **parent, 4000 unsigned int ofs, bool *leftmost); 4001 struct rb_entry *f2fs_lookup_rb_tree_ret(struct rb_root_cached *root, 4002 struct rb_entry *cached_re, unsigned int ofs, 4003 struct rb_entry **prev_entry, struct rb_entry **next_entry, 4004 struct rb_node ***insert_p, struct rb_node **insert_parent, 4005 bool force, bool *leftmost); 4006 bool f2fs_check_rb_tree_consistence(struct f2fs_sb_info *sbi, 4007 struct rb_root_cached *root, bool check_key); 4008 unsigned int f2fs_shrink_extent_tree(struct f2fs_sb_info *sbi, int nr_shrink); 4009 void f2fs_init_extent_tree(struct inode *inode, struct page *ipage); 4010 void f2fs_drop_extent_tree(struct inode *inode); 4011 unsigned int f2fs_destroy_extent_node(struct inode *inode); 4012 void f2fs_destroy_extent_tree(struct inode *inode); 4013 bool f2fs_lookup_extent_cache(struct inode *inode, pgoff_t pgofs, 4014 struct extent_info *ei); 4015 void f2fs_update_extent_cache(struct dnode_of_data *dn); 4016 void f2fs_update_extent_cache_range(struct dnode_of_data *dn, 4017 pgoff_t fofs, block_t blkaddr, unsigned int len); 4018 void f2fs_init_extent_cache_info(struct f2fs_sb_info *sbi); 4019 int __init f2fs_create_extent_cache(void); 4020 void f2fs_destroy_extent_cache(void); 4021 4022 /* 4023 * sysfs.c 4024 */ 4025 #define MIN_RA_MUL 2 4026 #define MAX_RA_MUL 256 4027 4028 int __init f2fs_init_sysfs(void); 4029 void f2fs_exit_sysfs(void); 4030 int f2fs_register_sysfs(struct f2fs_sb_info *sbi); 4031 void f2fs_unregister_sysfs(struct f2fs_sb_info *sbi); 4032 4033 /* verity.c */ 4034 extern const struct fsverity_operations f2fs_verityops; 4035 4036 /* 4037 * crypto support 4038 */ 4039 static inline bool f2fs_encrypted_file(struct inode *inode) 4040 { 4041 return IS_ENCRYPTED(inode) && S_ISREG(inode->i_mode); 4042 } 4043 4044 static inline void f2fs_set_encrypted_inode(struct inode *inode) 4045 { 4046 #ifdef CONFIG_FS_ENCRYPTION 4047 file_set_encrypt(inode); 4048 f2fs_set_inode_flags(inode); 4049 #endif 4050 } 4051 4052 /* 4053 * Returns true if the reads of the inode's data need to undergo some 4054 * postprocessing step, like decryption or authenticity verification. 4055 */ 4056 static inline bool f2fs_post_read_required(struct inode *inode) 4057 { 4058 return f2fs_encrypted_file(inode) || fsverity_active(inode) || 4059 f2fs_compressed_file(inode); 4060 } 4061 4062 /* 4063 * compress.c 4064 */ 4065 #ifdef CONFIG_F2FS_FS_COMPRESSION 4066 bool f2fs_is_compressed_page(struct page *page); 4067 struct page *f2fs_compress_control_page(struct page *page); 4068 int f2fs_prepare_compress_overwrite(struct inode *inode, 4069 struct page **pagep, pgoff_t index, void **fsdata); 4070 bool f2fs_compress_write_end(struct inode *inode, void *fsdata, 4071 pgoff_t index, unsigned copied); 4072 int f2fs_truncate_partial_cluster(struct inode *inode, u64 from, bool lock); 4073 void f2fs_compress_write_end_io(struct bio *bio, struct page *page); 4074 bool f2fs_is_compress_backend_ready(struct inode *inode); 4075 int f2fs_init_compress_mempool(void); 4076 void f2fs_destroy_compress_mempool(void); 4077 void f2fs_decompress_cluster(struct decompress_io_ctx *dic); 4078 void f2fs_end_read_compressed_page(struct page *page, bool failed, 4079 block_t blkaddr); 4080 bool f2fs_cluster_is_empty(struct compress_ctx *cc); 4081 bool f2fs_cluster_can_merge_page(struct compress_ctx *cc, pgoff_t index); 4082 bool f2fs_all_cluster_page_loaded(struct compress_ctx *cc, struct pagevec *pvec, 4083 int index, int nr_pages); 4084 bool f2fs_sanity_check_cluster(struct dnode_of_data *dn); 4085 void f2fs_compress_ctx_add_page(struct compress_ctx *cc, struct page *page); 4086 int f2fs_write_multi_pages(struct compress_ctx *cc, 4087 int *submitted, 4088 struct writeback_control *wbc, 4089 enum iostat_type io_type); 4090 int f2fs_is_compressed_cluster(struct inode *inode, pgoff_t index); 4091 void f2fs_update_extent_tree_range_compressed(struct inode *inode, 4092 pgoff_t fofs, block_t blkaddr, unsigned int llen, 4093 unsigned int c_len); 4094 int f2fs_read_multi_pages(struct compress_ctx *cc, struct bio **bio_ret, 4095 unsigned nr_pages, sector_t *last_block_in_bio, 4096 bool is_readahead, bool for_write); 4097 struct decompress_io_ctx *f2fs_alloc_dic(struct compress_ctx *cc); 4098 void f2fs_decompress_end_io(struct decompress_io_ctx *dic, bool failed); 4099 void f2fs_put_page_dic(struct page *page); 4100 unsigned int f2fs_cluster_blocks_are_contiguous(struct dnode_of_data *dn); 4101 int f2fs_init_compress_ctx(struct compress_ctx *cc); 4102 void f2fs_destroy_compress_ctx(struct compress_ctx *cc, bool reuse); 4103 void f2fs_init_compress_info(struct f2fs_sb_info *sbi); 4104 int f2fs_init_compress_inode(struct f2fs_sb_info *sbi); 4105 void f2fs_destroy_compress_inode(struct f2fs_sb_info *sbi); 4106 int f2fs_init_page_array_cache(struct f2fs_sb_info *sbi); 4107 void f2fs_destroy_page_array_cache(struct f2fs_sb_info *sbi); 4108 int __init f2fs_init_compress_cache(void); 4109 void f2fs_destroy_compress_cache(void); 4110 struct address_space *COMPRESS_MAPPING(struct f2fs_sb_info *sbi); 4111 void f2fs_invalidate_compress_page(struct f2fs_sb_info *sbi, block_t blkaddr); 4112 void f2fs_cache_compressed_page(struct f2fs_sb_info *sbi, struct page *page, 4113 nid_t ino, block_t blkaddr); 4114 bool f2fs_load_compressed_page(struct f2fs_sb_info *sbi, struct page *page, 4115 block_t blkaddr); 4116 void f2fs_invalidate_compress_pages(struct f2fs_sb_info *sbi, nid_t ino); 4117 #define inc_compr_inode_stat(inode) \ 4118 do { \ 4119 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); \ 4120 sbi->compr_new_inode++; \ 4121 } while (0) 4122 #define add_compr_block_stat(inode, blocks) \ 4123 do { \ 4124 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); \ 4125 int diff = F2FS_I(inode)->i_cluster_size - blocks; \ 4126 sbi->compr_written_block += blocks; \ 4127 sbi->compr_saved_block += diff; \ 4128 } while (0) 4129 #else 4130 static inline bool f2fs_is_compressed_page(struct page *page) { return false; } 4131 static inline bool f2fs_is_compress_backend_ready(struct inode *inode) 4132 { 4133 if (!f2fs_compressed_file(inode)) 4134 return true; 4135 /* not support compression */ 4136 return false; 4137 } 4138 static inline struct page *f2fs_compress_control_page(struct page *page) 4139 { 4140 WARN_ON_ONCE(1); 4141 return ERR_PTR(-EINVAL); 4142 } 4143 static inline int f2fs_init_compress_mempool(void) { return 0; } 4144 static inline void f2fs_destroy_compress_mempool(void) { } 4145 static inline void f2fs_decompress_cluster(struct decompress_io_ctx *dic) { } 4146 static inline void f2fs_end_read_compressed_page(struct page *page, 4147 bool failed, block_t blkaddr) 4148 { 4149 WARN_ON_ONCE(1); 4150 } 4151 static inline void f2fs_put_page_dic(struct page *page) 4152 { 4153 WARN_ON_ONCE(1); 4154 } 4155 static inline unsigned int f2fs_cluster_blocks_are_contiguous(struct dnode_of_data *dn) { return 0; } 4156 static inline bool f2fs_sanity_check_cluster(struct dnode_of_data *dn) { return false; } 4157 static inline int f2fs_init_compress_inode(struct f2fs_sb_info *sbi) { return 0; } 4158 static inline void f2fs_destroy_compress_inode(struct f2fs_sb_info *sbi) { } 4159 static inline int f2fs_init_page_array_cache(struct f2fs_sb_info *sbi) { return 0; } 4160 static inline void f2fs_destroy_page_array_cache(struct f2fs_sb_info *sbi) { } 4161 static inline int __init f2fs_init_compress_cache(void) { return 0; } 4162 static inline void f2fs_destroy_compress_cache(void) { } 4163 static inline void f2fs_invalidate_compress_page(struct f2fs_sb_info *sbi, 4164 block_t blkaddr) { } 4165 static inline void f2fs_cache_compressed_page(struct f2fs_sb_info *sbi, 4166 struct page *page, nid_t ino, block_t blkaddr) { } 4167 static inline bool f2fs_load_compressed_page(struct f2fs_sb_info *sbi, 4168 struct page *page, block_t blkaddr) { return false; } 4169 static inline void f2fs_invalidate_compress_pages(struct f2fs_sb_info *sbi, 4170 nid_t ino) { } 4171 #define inc_compr_inode_stat(inode) do { } while (0) 4172 static inline void f2fs_update_extent_tree_range_compressed(struct inode *inode, 4173 pgoff_t fofs, block_t blkaddr, unsigned int llen, 4174 unsigned int c_len) { } 4175 #endif 4176 4177 static inline void set_compress_context(struct inode *inode) 4178 { 4179 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 4180 4181 F2FS_I(inode)->i_compress_algorithm = 4182 F2FS_OPTION(sbi).compress_algorithm; 4183 F2FS_I(inode)->i_log_cluster_size = 4184 F2FS_OPTION(sbi).compress_log_size; 4185 F2FS_I(inode)->i_compress_flag = 4186 F2FS_OPTION(sbi).compress_chksum ? 4187 1 << COMPRESS_CHKSUM : 0; 4188 F2FS_I(inode)->i_cluster_size = 4189 1 << F2FS_I(inode)->i_log_cluster_size; 4190 if ((F2FS_I(inode)->i_compress_algorithm == COMPRESS_LZ4 || 4191 F2FS_I(inode)->i_compress_algorithm == COMPRESS_ZSTD) && 4192 F2FS_OPTION(sbi).compress_level) 4193 F2FS_I(inode)->i_compress_flag |= 4194 F2FS_OPTION(sbi).compress_level << 4195 COMPRESS_LEVEL_OFFSET; 4196 F2FS_I(inode)->i_flags |= F2FS_COMPR_FL; 4197 set_inode_flag(inode, FI_COMPRESSED_FILE); 4198 stat_inc_compr_inode(inode); 4199 inc_compr_inode_stat(inode); 4200 f2fs_mark_inode_dirty_sync(inode, true); 4201 } 4202 4203 static inline bool f2fs_disable_compressed_file(struct inode *inode) 4204 { 4205 struct f2fs_inode_info *fi = F2FS_I(inode); 4206 4207 if (!f2fs_compressed_file(inode)) 4208 return true; 4209 if (S_ISREG(inode->i_mode) && F2FS_HAS_BLOCKS(inode)) 4210 return false; 4211 4212 fi->i_flags &= ~F2FS_COMPR_FL; 4213 stat_dec_compr_inode(inode); 4214 clear_inode_flag(inode, FI_COMPRESSED_FILE); 4215 f2fs_mark_inode_dirty_sync(inode, true); 4216 return true; 4217 } 4218 4219 #define F2FS_FEATURE_FUNCS(name, flagname) \ 4220 static inline int f2fs_sb_has_##name(struct f2fs_sb_info *sbi) \ 4221 { \ 4222 return F2FS_HAS_FEATURE(sbi, F2FS_FEATURE_##flagname); \ 4223 } 4224 4225 F2FS_FEATURE_FUNCS(encrypt, ENCRYPT); 4226 F2FS_FEATURE_FUNCS(blkzoned, BLKZONED); 4227 F2FS_FEATURE_FUNCS(extra_attr, EXTRA_ATTR); 4228 F2FS_FEATURE_FUNCS(project_quota, PRJQUOTA); 4229 F2FS_FEATURE_FUNCS(inode_chksum, INODE_CHKSUM); 4230 F2FS_FEATURE_FUNCS(flexible_inline_xattr, FLEXIBLE_INLINE_XATTR); 4231 F2FS_FEATURE_FUNCS(quota_ino, QUOTA_INO); 4232 F2FS_FEATURE_FUNCS(inode_crtime, INODE_CRTIME); 4233 F2FS_FEATURE_FUNCS(lost_found, LOST_FOUND); 4234 F2FS_FEATURE_FUNCS(verity, VERITY); 4235 F2FS_FEATURE_FUNCS(sb_chksum, SB_CHKSUM); 4236 F2FS_FEATURE_FUNCS(casefold, CASEFOLD); 4237 F2FS_FEATURE_FUNCS(compression, COMPRESSION); 4238 F2FS_FEATURE_FUNCS(readonly, RO); 4239 4240 static inline bool f2fs_may_extent_tree(struct inode *inode) 4241 { 4242 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 4243 4244 if (!test_opt(sbi, EXTENT_CACHE) || 4245 is_inode_flag_set(inode, FI_NO_EXTENT) || 4246 (is_inode_flag_set(inode, FI_COMPRESSED_FILE) && 4247 !f2fs_sb_has_readonly(sbi))) 4248 return false; 4249 4250 /* 4251 * for recovered files during mount do not create extents 4252 * if shrinker is not registered. 4253 */ 4254 if (list_empty(&sbi->s_list)) 4255 return false; 4256 4257 return S_ISREG(inode->i_mode); 4258 } 4259 4260 #ifdef CONFIG_BLK_DEV_ZONED 4261 static inline bool f2fs_blkz_is_seq(struct f2fs_sb_info *sbi, int devi, 4262 block_t blkaddr) 4263 { 4264 unsigned int zno = blkaddr >> sbi->log_blocks_per_blkz; 4265 4266 return test_bit(zno, FDEV(devi).blkz_seq); 4267 } 4268 #endif 4269 4270 static inline bool f2fs_hw_should_discard(struct f2fs_sb_info *sbi) 4271 { 4272 return f2fs_sb_has_blkzoned(sbi); 4273 } 4274 4275 static inline bool f2fs_bdev_support_discard(struct block_device *bdev) 4276 { 4277 return blk_queue_discard(bdev_get_queue(bdev)) || 4278 bdev_is_zoned(bdev); 4279 } 4280 4281 static inline bool f2fs_hw_support_discard(struct f2fs_sb_info *sbi) 4282 { 4283 int i; 4284 4285 if (!f2fs_is_multi_device(sbi)) 4286 return f2fs_bdev_support_discard(sbi->sb->s_bdev); 4287 4288 for (i = 0; i < sbi->s_ndevs; i++) 4289 if (f2fs_bdev_support_discard(FDEV(i).bdev)) 4290 return true; 4291 return false; 4292 } 4293 4294 static inline bool f2fs_realtime_discard_enable(struct f2fs_sb_info *sbi) 4295 { 4296 return (test_opt(sbi, DISCARD) && f2fs_hw_support_discard(sbi)) || 4297 f2fs_hw_should_discard(sbi); 4298 } 4299 4300 static inline bool f2fs_hw_is_readonly(struct f2fs_sb_info *sbi) 4301 { 4302 int i; 4303 4304 if (!f2fs_is_multi_device(sbi)) 4305 return bdev_read_only(sbi->sb->s_bdev); 4306 4307 for (i = 0; i < sbi->s_ndevs; i++) 4308 if (bdev_read_only(FDEV(i).bdev)) 4309 return true; 4310 return false; 4311 } 4312 4313 static inline bool f2fs_lfs_mode(struct f2fs_sb_info *sbi) 4314 { 4315 return F2FS_OPTION(sbi).fs_mode == FS_MODE_LFS; 4316 } 4317 4318 static inline bool f2fs_may_compress(struct inode *inode) 4319 { 4320 if (IS_SWAPFILE(inode) || f2fs_is_pinned_file(inode) || 4321 f2fs_is_atomic_file(inode) || 4322 f2fs_is_volatile_file(inode)) 4323 return false; 4324 return S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode); 4325 } 4326 4327 static inline void f2fs_i_compr_blocks_update(struct inode *inode, 4328 u64 blocks, bool add) 4329 { 4330 int diff = F2FS_I(inode)->i_cluster_size - blocks; 4331 struct f2fs_inode_info *fi = F2FS_I(inode); 4332 4333 /* don't update i_compr_blocks if saved blocks were released */ 4334 if (!add && !atomic_read(&fi->i_compr_blocks)) 4335 return; 4336 4337 if (add) { 4338 atomic_add(diff, &fi->i_compr_blocks); 4339 stat_add_compr_blocks(inode, diff); 4340 } else { 4341 atomic_sub(diff, &fi->i_compr_blocks); 4342 stat_sub_compr_blocks(inode, diff); 4343 } 4344 f2fs_mark_inode_dirty_sync(inode, true); 4345 } 4346 4347 static inline int block_unaligned_IO(struct inode *inode, 4348 struct kiocb *iocb, struct iov_iter *iter) 4349 { 4350 unsigned int i_blkbits = READ_ONCE(inode->i_blkbits); 4351 unsigned int blocksize_mask = (1 << i_blkbits) - 1; 4352 loff_t offset = iocb->ki_pos; 4353 unsigned long align = offset | iov_iter_alignment(iter); 4354 4355 return align & blocksize_mask; 4356 } 4357 4358 static inline bool f2fs_allow_multi_device_dio(struct f2fs_sb_info *sbi, 4359 int flag) 4360 { 4361 if (!f2fs_is_multi_device(sbi)) 4362 return false; 4363 if (flag != F2FS_GET_BLOCK_DIO) 4364 return false; 4365 return sbi->aligned_blksize; 4366 } 4367 4368 static inline bool f2fs_force_buffered_io(struct inode *inode, 4369 struct kiocb *iocb, struct iov_iter *iter) 4370 { 4371 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 4372 int rw = iov_iter_rw(iter); 4373 4374 if (f2fs_post_read_required(inode)) 4375 return true; 4376 4377 /* disallow direct IO if any of devices has unaligned blksize */ 4378 if (f2fs_is_multi_device(sbi) && !sbi->aligned_blksize) 4379 return true; 4380 /* 4381 * for blkzoned device, fallback direct IO to buffered IO, so 4382 * all IOs can be serialized by log-structured write. 4383 */ 4384 if (f2fs_sb_has_blkzoned(sbi)) 4385 return true; 4386 if (f2fs_lfs_mode(sbi) && (rw == WRITE)) { 4387 if (block_unaligned_IO(inode, iocb, iter)) 4388 return true; 4389 if (F2FS_IO_ALIGNED(sbi)) 4390 return true; 4391 } 4392 if (is_sbi_flag_set(F2FS_I_SB(inode), SBI_CP_DISABLED)) 4393 return true; 4394 4395 return false; 4396 } 4397 4398 static inline bool f2fs_need_verity(const struct inode *inode, pgoff_t idx) 4399 { 4400 return fsverity_active(inode) && 4401 idx < DIV_ROUND_UP(inode->i_size, PAGE_SIZE); 4402 } 4403 4404 #ifdef CONFIG_F2FS_FAULT_INJECTION 4405 extern void f2fs_build_fault_attr(struct f2fs_sb_info *sbi, unsigned int rate, 4406 unsigned int type); 4407 #else 4408 #define f2fs_build_fault_attr(sbi, rate, type) do { } while (0) 4409 #endif 4410 4411 static inline bool is_journalled_quota(struct f2fs_sb_info *sbi) 4412 { 4413 #ifdef CONFIG_QUOTA 4414 if (f2fs_sb_has_quota_ino(sbi)) 4415 return true; 4416 if (F2FS_OPTION(sbi).s_qf_names[USRQUOTA] || 4417 F2FS_OPTION(sbi).s_qf_names[GRPQUOTA] || 4418 F2FS_OPTION(sbi).s_qf_names[PRJQUOTA]) 4419 return true; 4420 #endif 4421 return false; 4422 } 4423 4424 static inline bool f2fs_block_unit_discard(struct f2fs_sb_info *sbi) 4425 { 4426 return F2FS_OPTION(sbi).discard_unit == DISCARD_UNIT_BLOCK; 4427 } 4428 4429 #define EFSBADCRC EBADMSG /* Bad CRC detected */ 4430 #define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */ 4431 4432 #endif /* _LINUX_F2FS_H */ 4433