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