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