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 657 658 #define FADVISE_MODIFIABLE_BITS (FADVISE_COLD_BIT | FADVISE_HOT_BIT) 659 660 #define file_is_cold(inode) is_file(inode, FADVISE_COLD_BIT) 661 #define file_set_cold(inode) set_file(inode, FADVISE_COLD_BIT) 662 #define file_clear_cold(inode) clear_file(inode, FADVISE_COLD_BIT) 663 664 #define file_wrong_pino(inode) is_file(inode, FADVISE_LOST_PINO_BIT) 665 #define file_lost_pino(inode) set_file(inode, FADVISE_LOST_PINO_BIT) 666 #define file_got_pino(inode) clear_file(inode, FADVISE_LOST_PINO_BIT) 667 668 #define file_is_encrypt(inode) is_file(inode, FADVISE_ENCRYPT_BIT) 669 #define file_set_encrypt(inode) set_file(inode, FADVISE_ENCRYPT_BIT) 670 671 #define file_enc_name(inode) is_file(inode, FADVISE_ENC_NAME_BIT) 672 #define file_set_enc_name(inode) set_file(inode, FADVISE_ENC_NAME_BIT) 673 674 #define file_keep_isize(inode) is_file(inode, FADVISE_KEEP_SIZE_BIT) 675 #define file_set_keep_isize(inode) set_file(inode, FADVISE_KEEP_SIZE_BIT) 676 677 #define file_is_hot(inode) is_file(inode, FADVISE_HOT_BIT) 678 #define file_set_hot(inode) set_file(inode, FADVISE_HOT_BIT) 679 #define file_clear_hot(inode) clear_file(inode, FADVISE_HOT_BIT) 680 681 #define file_is_verity(inode) is_file(inode, FADVISE_VERITY_BIT) 682 #define file_set_verity(inode) set_file(inode, FADVISE_VERITY_BIT) 683 684 #define DEF_DIR_LEVEL 0 685 686 enum { 687 GC_FAILURE_PIN, 688 GC_FAILURE_ATOMIC, 689 MAX_GC_FAILURE 690 }; 691 692 /* used for f2fs_inode_info->flags */ 693 enum { 694 FI_NEW_INODE, /* indicate newly allocated inode */ 695 FI_DIRTY_INODE, /* indicate inode is dirty or not */ 696 FI_AUTO_RECOVER, /* indicate inode is recoverable */ 697 FI_DIRTY_DIR, /* indicate directory has dirty pages */ 698 FI_INC_LINK, /* need to increment i_nlink */ 699 FI_ACL_MODE, /* indicate acl mode */ 700 FI_NO_ALLOC, /* should not allocate any blocks */ 701 FI_FREE_NID, /* free allocated nide */ 702 FI_NO_EXTENT, /* not to use the extent cache */ 703 FI_INLINE_XATTR, /* used for inline xattr */ 704 FI_INLINE_DATA, /* used for inline data*/ 705 FI_INLINE_DENTRY, /* used for inline dentry */ 706 FI_APPEND_WRITE, /* inode has appended data */ 707 FI_UPDATE_WRITE, /* inode has in-place-update data */ 708 FI_NEED_IPU, /* used for ipu per file */ 709 FI_ATOMIC_FILE, /* indicate atomic file */ 710 FI_ATOMIC_COMMIT, /* indicate the state of atomical committing */ 711 FI_VOLATILE_FILE, /* indicate volatile file */ 712 FI_FIRST_BLOCK_WRITTEN, /* indicate #0 data block was written */ 713 FI_DROP_CACHE, /* drop dirty page cache */ 714 FI_DATA_EXIST, /* indicate data exists */ 715 FI_INLINE_DOTS, /* indicate inline dot dentries */ 716 FI_DO_DEFRAG, /* indicate defragment is running */ 717 FI_DIRTY_FILE, /* indicate regular/symlink has dirty pages */ 718 FI_NO_PREALLOC, /* indicate skipped preallocated blocks */ 719 FI_HOT_DATA, /* indicate file is hot */ 720 FI_EXTRA_ATTR, /* indicate file has extra attribute */ 721 FI_PROJ_INHERIT, /* indicate file inherits projectid */ 722 FI_PIN_FILE, /* indicate file should not be gced */ 723 FI_ATOMIC_REVOKE_REQUEST, /* request to drop atomic data */ 724 FI_VERITY_IN_PROGRESS, /* building fs-verity Merkle tree */ 725 FI_COMPRESSED_FILE, /* indicate file's data can be compressed */ 726 FI_COMPRESS_CORRUPT, /* indicate compressed cluster is corrupted */ 727 FI_MMAP_FILE, /* indicate file was mmapped */ 728 FI_ENABLE_COMPRESS, /* enable compression in "user" compression mode */ 729 FI_COMPRESS_RELEASED, /* compressed blocks were released */ 730 FI_ALIGNED_WRITE, /* enable aligned write */ 731 FI_MAX, /* max flag, never be used */ 732 }; 733 734 struct f2fs_inode_info { 735 struct inode vfs_inode; /* serve a vfs inode */ 736 unsigned long i_flags; /* keep an inode flags for ioctl */ 737 unsigned char i_advise; /* use to give file attribute hints */ 738 unsigned char i_dir_level; /* use for dentry level for large dir */ 739 unsigned int i_current_depth; /* only for directory depth */ 740 /* for gc failure statistic */ 741 unsigned int i_gc_failures[MAX_GC_FAILURE]; 742 unsigned int i_pino; /* parent inode number */ 743 umode_t i_acl_mode; /* keep file acl mode temporarily */ 744 745 /* Use below internally in f2fs*/ 746 unsigned long flags[BITS_TO_LONGS(FI_MAX)]; /* use to pass per-file flags */ 747 struct rw_semaphore i_sem; /* protect fi info */ 748 atomic_t dirty_pages; /* # of dirty pages */ 749 f2fs_hash_t chash; /* hash value of given file name */ 750 unsigned int clevel; /* maximum level of given file name */ 751 struct task_struct *task; /* lookup and create consistency */ 752 struct task_struct *cp_task; /* separate cp/wb IO stats*/ 753 nid_t i_xattr_nid; /* node id that contains xattrs */ 754 loff_t last_disk_size; /* lastly written file size */ 755 spinlock_t i_size_lock; /* protect last_disk_size */ 756 757 #ifdef CONFIG_QUOTA 758 struct dquot *i_dquot[MAXQUOTAS]; 759 760 /* quota space reservation, managed internally by quota code */ 761 qsize_t i_reserved_quota; 762 #endif 763 struct list_head dirty_list; /* dirty list for dirs and files */ 764 struct list_head gdirty_list; /* linked in global dirty list */ 765 struct list_head inmem_ilist; /* list for inmem inodes */ 766 struct list_head inmem_pages; /* inmemory pages managed by f2fs */ 767 struct task_struct *inmem_task; /* store inmemory task */ 768 struct mutex inmem_lock; /* lock for inmemory pages */ 769 struct extent_tree *extent_tree; /* cached extent_tree entry */ 770 771 /* avoid racing between foreground op and gc */ 772 struct rw_semaphore i_gc_rwsem[2]; 773 struct rw_semaphore i_xattr_sem; /* avoid racing between reading and changing EAs */ 774 775 int i_extra_isize; /* size of extra space located in i_addr */ 776 kprojid_t i_projid; /* id for project quota */ 777 int i_inline_xattr_size; /* inline xattr size */ 778 struct timespec64 i_crtime; /* inode creation time */ 779 struct timespec64 i_disk_time[4];/* inode disk times */ 780 781 /* for file compress */ 782 atomic_t i_compr_blocks; /* # of compressed blocks */ 783 unsigned char i_compress_algorithm; /* algorithm type */ 784 unsigned char i_log_cluster_size; /* log of cluster size */ 785 unsigned char i_compress_level; /* compress level (lz4hc,zstd) */ 786 unsigned short i_compress_flag; /* compress flag */ 787 unsigned int i_cluster_size; /* cluster size */ 788 }; 789 790 static inline void get_extent_info(struct extent_info *ext, 791 struct f2fs_extent *i_ext) 792 { 793 ext->fofs = le32_to_cpu(i_ext->fofs); 794 ext->blk = le32_to_cpu(i_ext->blk); 795 ext->len = le32_to_cpu(i_ext->len); 796 } 797 798 static inline void set_raw_extent(struct extent_info *ext, 799 struct f2fs_extent *i_ext) 800 { 801 i_ext->fofs = cpu_to_le32(ext->fofs); 802 i_ext->blk = cpu_to_le32(ext->blk); 803 i_ext->len = cpu_to_le32(ext->len); 804 } 805 806 static inline void set_extent_info(struct extent_info *ei, unsigned int fofs, 807 u32 blk, unsigned int len) 808 { 809 ei->fofs = fofs; 810 ei->blk = blk; 811 ei->len = len; 812 #ifdef CONFIG_F2FS_FS_COMPRESSION 813 ei->c_len = 0; 814 #endif 815 } 816 817 static inline bool __is_discard_mergeable(struct discard_info *back, 818 struct discard_info *front, unsigned int max_len) 819 { 820 return (back->lstart + back->len == front->lstart) && 821 (back->len + front->len <= max_len); 822 } 823 824 static inline bool __is_discard_back_mergeable(struct discard_info *cur, 825 struct discard_info *back, unsigned int max_len) 826 { 827 return __is_discard_mergeable(back, cur, max_len); 828 } 829 830 static inline bool __is_discard_front_mergeable(struct discard_info *cur, 831 struct discard_info *front, unsigned int max_len) 832 { 833 return __is_discard_mergeable(cur, front, max_len); 834 } 835 836 static inline bool __is_extent_mergeable(struct extent_info *back, 837 struct extent_info *front) 838 { 839 #ifdef CONFIG_F2FS_FS_COMPRESSION 840 if (back->c_len && back->len != back->c_len) 841 return false; 842 if (front->c_len && front->len != front->c_len) 843 return false; 844 #endif 845 return (back->fofs + back->len == front->fofs && 846 back->blk + back->len == front->blk); 847 } 848 849 static inline bool __is_back_mergeable(struct extent_info *cur, 850 struct extent_info *back) 851 { 852 return __is_extent_mergeable(back, cur); 853 } 854 855 static inline bool __is_front_mergeable(struct extent_info *cur, 856 struct extent_info *front) 857 { 858 return __is_extent_mergeable(cur, front); 859 } 860 861 extern void f2fs_mark_inode_dirty_sync(struct inode *inode, bool sync); 862 static inline void __try_update_largest_extent(struct extent_tree *et, 863 struct extent_node *en) 864 { 865 if (en->ei.len > et->largest.len) { 866 et->largest = en->ei; 867 et->largest_updated = true; 868 } 869 } 870 871 /* 872 * For free nid management 873 */ 874 enum nid_state { 875 FREE_NID, /* newly added to free nid list */ 876 PREALLOC_NID, /* it is preallocated */ 877 MAX_NID_STATE, 878 }; 879 880 enum nat_state { 881 TOTAL_NAT, 882 DIRTY_NAT, 883 RECLAIMABLE_NAT, 884 MAX_NAT_STATE, 885 }; 886 887 struct f2fs_nm_info { 888 block_t nat_blkaddr; /* base disk address of NAT */ 889 nid_t max_nid; /* maximum possible node ids */ 890 nid_t available_nids; /* # of available node ids */ 891 nid_t next_scan_nid; /* the next nid to be scanned */ 892 unsigned int ram_thresh; /* control the memory footprint */ 893 unsigned int ra_nid_pages; /* # of nid pages to be readaheaded */ 894 unsigned int dirty_nats_ratio; /* control dirty nats ratio threshold */ 895 896 /* NAT cache management */ 897 struct radix_tree_root nat_root;/* root of the nat entry cache */ 898 struct radix_tree_root nat_set_root;/* root of the nat set cache */ 899 struct rw_semaphore nat_tree_lock; /* protect nat entry tree */ 900 struct list_head nat_entries; /* cached nat entry list (clean) */ 901 spinlock_t nat_list_lock; /* protect clean nat entry list */ 902 unsigned int nat_cnt[MAX_NAT_STATE]; /* the # of cached nat entries */ 903 unsigned int nat_blocks; /* # of nat blocks */ 904 905 /* free node ids management */ 906 struct radix_tree_root free_nid_root;/* root of the free_nid cache */ 907 struct list_head free_nid_list; /* list for free nids excluding preallocated nids */ 908 unsigned int nid_cnt[MAX_NID_STATE]; /* the number of free node id */ 909 spinlock_t nid_list_lock; /* protect nid lists ops */ 910 struct mutex build_lock; /* lock for build free nids */ 911 unsigned char **free_nid_bitmap; 912 unsigned char *nat_block_bitmap; 913 unsigned short *free_nid_count; /* free nid count of NAT block */ 914 915 /* for checkpoint */ 916 char *nat_bitmap; /* NAT bitmap pointer */ 917 918 unsigned int nat_bits_blocks; /* # of nat bits blocks */ 919 unsigned char *nat_bits; /* NAT bits blocks */ 920 unsigned char *full_nat_bits; /* full NAT pages */ 921 unsigned char *empty_nat_bits; /* empty NAT pages */ 922 #ifdef CONFIG_F2FS_CHECK_FS 923 char *nat_bitmap_mir; /* NAT bitmap mirror */ 924 #endif 925 int bitmap_size; /* bitmap size */ 926 }; 927 928 /* 929 * this structure is used as one of function parameters. 930 * all the information are dedicated to a given direct node block determined 931 * by the data offset in a file. 932 */ 933 struct dnode_of_data { 934 struct inode *inode; /* vfs inode pointer */ 935 struct page *inode_page; /* its inode page, NULL is possible */ 936 struct page *node_page; /* cached direct node page */ 937 nid_t nid; /* node id of the direct node block */ 938 unsigned int ofs_in_node; /* data offset in the node page */ 939 bool inode_page_locked; /* inode page is locked or not */ 940 bool node_changed; /* is node block changed */ 941 char cur_level; /* level of hole node page */ 942 char max_level; /* level of current page located */ 943 block_t data_blkaddr; /* block address of the node block */ 944 }; 945 946 static inline void set_new_dnode(struct dnode_of_data *dn, struct inode *inode, 947 struct page *ipage, struct page *npage, nid_t nid) 948 { 949 memset(dn, 0, sizeof(*dn)); 950 dn->inode = inode; 951 dn->inode_page = ipage; 952 dn->node_page = npage; 953 dn->nid = nid; 954 } 955 956 /* 957 * For SIT manager 958 * 959 * By default, there are 6 active log areas across the whole main area. 960 * When considering hot and cold data separation to reduce cleaning overhead, 961 * we split 3 for data logs and 3 for node logs as hot, warm, and cold types, 962 * respectively. 963 * In the current design, you should not change the numbers intentionally. 964 * Instead, as a mount option such as active_logs=x, you can use 2, 4, and 6 965 * logs individually according to the underlying devices. (default: 6) 966 * Just in case, on-disk layout covers maximum 16 logs that consist of 8 for 967 * data and 8 for node logs. 968 */ 969 #define NR_CURSEG_DATA_TYPE (3) 970 #define NR_CURSEG_NODE_TYPE (3) 971 #define NR_CURSEG_INMEM_TYPE (2) 972 #define NR_CURSEG_RO_TYPE (2) 973 #define NR_CURSEG_PERSIST_TYPE (NR_CURSEG_DATA_TYPE + NR_CURSEG_NODE_TYPE) 974 #define NR_CURSEG_TYPE (NR_CURSEG_INMEM_TYPE + NR_CURSEG_PERSIST_TYPE) 975 976 enum { 977 CURSEG_HOT_DATA = 0, /* directory entry blocks */ 978 CURSEG_WARM_DATA, /* data blocks */ 979 CURSEG_COLD_DATA, /* multimedia or GCed data blocks */ 980 CURSEG_HOT_NODE, /* direct node blocks of directory files */ 981 CURSEG_WARM_NODE, /* direct node blocks of normal files */ 982 CURSEG_COLD_NODE, /* indirect node blocks */ 983 NR_PERSISTENT_LOG, /* number of persistent log */ 984 CURSEG_COLD_DATA_PINNED = NR_PERSISTENT_LOG, 985 /* pinned file that needs consecutive block address */ 986 CURSEG_ALL_DATA_ATGC, /* SSR alloctor in hot/warm/cold data area */ 987 NO_CHECK_TYPE, /* number of persistent & inmem log */ 988 }; 989 990 struct flush_cmd { 991 struct completion wait; 992 struct llist_node llnode; 993 nid_t ino; 994 int ret; 995 }; 996 997 struct flush_cmd_control { 998 struct task_struct *f2fs_issue_flush; /* flush thread */ 999 wait_queue_head_t flush_wait_queue; /* waiting queue for wake-up */ 1000 atomic_t issued_flush; /* # of issued flushes */ 1001 atomic_t queued_flush; /* # of queued flushes */ 1002 struct llist_head issue_list; /* list for command issue */ 1003 struct llist_node *dispatch_list; /* list for command dispatch */ 1004 }; 1005 1006 struct f2fs_sm_info { 1007 struct sit_info *sit_info; /* whole segment information */ 1008 struct free_segmap_info *free_info; /* free segment information */ 1009 struct dirty_seglist_info *dirty_info; /* dirty segment information */ 1010 struct curseg_info *curseg_array; /* active segment information */ 1011 1012 struct rw_semaphore curseg_lock; /* for preventing curseg change */ 1013 1014 block_t seg0_blkaddr; /* block address of 0'th segment */ 1015 block_t main_blkaddr; /* start block address of main area */ 1016 block_t ssa_blkaddr; /* start block address of SSA area */ 1017 1018 unsigned int segment_count; /* total # of segments */ 1019 unsigned int main_segments; /* # of segments in main area */ 1020 unsigned int reserved_segments; /* # of reserved segments */ 1021 unsigned int ovp_segments; /* # of overprovision segments */ 1022 1023 /* a threshold to reclaim prefree segments */ 1024 unsigned int rec_prefree_segments; 1025 1026 /* for batched trimming */ 1027 unsigned int trim_sections; /* # of sections to trim */ 1028 1029 struct list_head sit_entry_set; /* sit entry set list */ 1030 1031 unsigned int ipu_policy; /* in-place-update policy */ 1032 unsigned int min_ipu_util; /* in-place-update threshold */ 1033 unsigned int min_fsync_blocks; /* threshold for fsync */ 1034 unsigned int min_seq_blocks; /* threshold for sequential blocks */ 1035 unsigned int min_hot_blocks; /* threshold for hot block allocation */ 1036 unsigned int min_ssr_sections; /* threshold to trigger SSR allocation */ 1037 1038 /* for flush command control */ 1039 struct flush_cmd_control *fcc_info; 1040 1041 /* for discard command control */ 1042 struct discard_cmd_control *dcc_info; 1043 }; 1044 1045 /* 1046 * For superblock 1047 */ 1048 /* 1049 * COUNT_TYPE for monitoring 1050 * 1051 * f2fs monitors the number of several block types such as on-writeback, 1052 * dirty dentry blocks, dirty node blocks, and dirty meta blocks. 1053 */ 1054 #define WB_DATA_TYPE(p) (__is_cp_guaranteed(p) ? F2FS_WB_CP_DATA : F2FS_WB_DATA) 1055 enum count_type { 1056 F2FS_DIRTY_DENTS, 1057 F2FS_DIRTY_DATA, 1058 F2FS_DIRTY_QDATA, 1059 F2FS_DIRTY_NODES, 1060 F2FS_DIRTY_META, 1061 F2FS_INMEM_PAGES, 1062 F2FS_DIRTY_IMETA, 1063 F2FS_WB_CP_DATA, 1064 F2FS_WB_DATA, 1065 F2FS_RD_DATA, 1066 F2FS_RD_NODE, 1067 F2FS_RD_META, 1068 F2FS_DIO_WRITE, 1069 F2FS_DIO_READ, 1070 NR_COUNT_TYPE, 1071 }; 1072 1073 /* 1074 * The below are the page types of bios used in submit_bio(). 1075 * The available types are: 1076 * DATA User data pages. It operates as async mode. 1077 * NODE Node pages. It operates as async mode. 1078 * META FS metadata pages such as SIT, NAT, CP. 1079 * NR_PAGE_TYPE The number of page types. 1080 * META_FLUSH Make sure the previous pages are written 1081 * with waiting the bio's completion 1082 * ... Only can be used with META. 1083 */ 1084 #define PAGE_TYPE_OF_BIO(type) ((type) > META ? META : (type)) 1085 enum page_type { 1086 DATA, 1087 NODE, 1088 META, 1089 NR_PAGE_TYPE, 1090 META_FLUSH, 1091 INMEM, /* the below types are used by tracepoints only. */ 1092 INMEM_DROP, 1093 INMEM_INVALIDATE, 1094 INMEM_REVOKE, 1095 IPU, 1096 OPU, 1097 }; 1098 1099 enum temp_type { 1100 HOT = 0, /* must be zero for meta bio */ 1101 WARM, 1102 COLD, 1103 NR_TEMP_TYPE, 1104 }; 1105 1106 enum need_lock_type { 1107 LOCK_REQ = 0, 1108 LOCK_DONE, 1109 LOCK_RETRY, 1110 }; 1111 1112 enum cp_reason_type { 1113 CP_NO_NEEDED, 1114 CP_NON_REGULAR, 1115 CP_COMPRESSED, 1116 CP_HARDLINK, 1117 CP_SB_NEED_CP, 1118 CP_WRONG_PINO, 1119 CP_NO_SPC_ROLL, 1120 CP_NODE_NEED_CP, 1121 CP_FASTBOOT_MODE, 1122 CP_SPEC_LOG_NUM, 1123 CP_RECOVER_DIR, 1124 }; 1125 1126 enum iostat_type { 1127 /* WRITE IO */ 1128 APP_DIRECT_IO, /* app direct write IOs */ 1129 APP_BUFFERED_IO, /* app buffered write IOs */ 1130 APP_WRITE_IO, /* app write IOs */ 1131 APP_MAPPED_IO, /* app mapped IOs */ 1132 FS_DATA_IO, /* data IOs from kworker/fsync/reclaimer */ 1133 FS_NODE_IO, /* node IOs from kworker/fsync/reclaimer */ 1134 FS_META_IO, /* meta IOs from kworker/reclaimer */ 1135 FS_GC_DATA_IO, /* data IOs from forground gc */ 1136 FS_GC_NODE_IO, /* node IOs from forground gc */ 1137 FS_CP_DATA_IO, /* data IOs from checkpoint */ 1138 FS_CP_NODE_IO, /* node IOs from checkpoint */ 1139 FS_CP_META_IO, /* meta IOs from checkpoint */ 1140 1141 /* READ IO */ 1142 APP_DIRECT_READ_IO, /* app direct read IOs */ 1143 APP_BUFFERED_READ_IO, /* app buffered read IOs */ 1144 APP_READ_IO, /* app read IOs */ 1145 APP_MAPPED_READ_IO, /* app mapped read IOs */ 1146 FS_DATA_READ_IO, /* data read IOs */ 1147 FS_GDATA_READ_IO, /* data read IOs from background gc */ 1148 FS_CDATA_READ_IO, /* compressed data read IOs */ 1149 FS_NODE_READ_IO, /* node read IOs */ 1150 FS_META_READ_IO, /* meta read IOs */ 1151 1152 /* other */ 1153 FS_DISCARD, /* discard */ 1154 NR_IO_TYPE, 1155 }; 1156 1157 struct f2fs_io_info { 1158 struct f2fs_sb_info *sbi; /* f2fs_sb_info pointer */ 1159 nid_t ino; /* inode number */ 1160 enum page_type type; /* contains DATA/NODE/META/META_FLUSH */ 1161 enum temp_type temp; /* contains HOT/WARM/COLD */ 1162 int op; /* contains REQ_OP_ */ 1163 int op_flags; /* req_flag_bits */ 1164 block_t new_blkaddr; /* new block address to be written */ 1165 block_t old_blkaddr; /* old block address before Cow */ 1166 struct page *page; /* page to be written */ 1167 struct page *encrypted_page; /* encrypted page */ 1168 struct page *compressed_page; /* compressed page */ 1169 struct list_head list; /* serialize IOs */ 1170 bool submitted; /* indicate IO submission */ 1171 int need_lock; /* indicate we need to lock cp_rwsem */ 1172 bool in_list; /* indicate fio is in io_list */ 1173 bool is_por; /* indicate IO is from recovery or not */ 1174 bool retry; /* need to reallocate block address */ 1175 int compr_blocks; /* # of compressed block addresses */ 1176 bool encrypted; /* indicate file is encrypted */ 1177 enum iostat_type io_type; /* io type */ 1178 struct writeback_control *io_wbc; /* writeback control */ 1179 struct bio **bio; /* bio for ipu */ 1180 sector_t *last_block; /* last block number in bio */ 1181 unsigned char version; /* version of the node */ 1182 }; 1183 1184 struct bio_entry { 1185 struct bio *bio; 1186 struct list_head list; 1187 }; 1188 1189 #define is_read_io(rw) ((rw) == READ) 1190 struct f2fs_bio_info { 1191 struct f2fs_sb_info *sbi; /* f2fs superblock */ 1192 struct bio *bio; /* bios to merge */ 1193 sector_t last_block_in_bio; /* last block number */ 1194 struct f2fs_io_info fio; /* store buffered io info. */ 1195 struct rw_semaphore io_rwsem; /* blocking op for bio */ 1196 spinlock_t io_lock; /* serialize DATA/NODE IOs */ 1197 struct list_head io_list; /* track fios */ 1198 struct list_head bio_list; /* bio entry list head */ 1199 struct rw_semaphore bio_list_lock; /* lock to protect bio entry list */ 1200 }; 1201 1202 #define FDEV(i) (sbi->devs[i]) 1203 #define RDEV(i) (raw_super->devs[i]) 1204 struct f2fs_dev_info { 1205 struct block_device *bdev; 1206 char path[MAX_PATH_LEN]; 1207 unsigned int total_segments; 1208 block_t start_blk; 1209 block_t end_blk; 1210 #ifdef CONFIG_BLK_DEV_ZONED 1211 unsigned int nr_blkz; /* Total number of zones */ 1212 unsigned long *blkz_seq; /* Bitmap indicating sequential zones */ 1213 block_t *zone_capacity_blocks; /* Array of zone capacity in blks */ 1214 #endif 1215 }; 1216 1217 enum inode_type { 1218 DIR_INODE, /* for dirty dir inode */ 1219 FILE_INODE, /* for dirty regular/symlink inode */ 1220 DIRTY_META, /* for all dirtied inode metadata */ 1221 ATOMIC_FILE, /* for all atomic files */ 1222 NR_INODE_TYPE, 1223 }; 1224 1225 /* for inner inode cache management */ 1226 struct inode_management { 1227 struct radix_tree_root ino_root; /* ino entry array */ 1228 spinlock_t ino_lock; /* for ino entry lock */ 1229 struct list_head ino_list; /* inode list head */ 1230 unsigned long ino_num; /* number of entries */ 1231 }; 1232 1233 /* for GC_AT */ 1234 struct atgc_management { 1235 bool atgc_enabled; /* ATGC is enabled or not */ 1236 struct rb_root_cached root; /* root of victim rb-tree */ 1237 struct list_head victim_list; /* linked with all victim entries */ 1238 unsigned int victim_count; /* victim count in rb-tree */ 1239 unsigned int candidate_ratio; /* candidate ratio */ 1240 unsigned int max_candidate_count; /* max candidate count */ 1241 unsigned int age_weight; /* age weight, vblock_weight = 100 - age_weight */ 1242 unsigned long long age_threshold; /* age threshold */ 1243 }; 1244 1245 /* For s_flag in struct f2fs_sb_info */ 1246 enum { 1247 SBI_IS_DIRTY, /* dirty flag for checkpoint */ 1248 SBI_IS_CLOSE, /* specify unmounting */ 1249 SBI_NEED_FSCK, /* need fsck.f2fs to fix */ 1250 SBI_POR_DOING, /* recovery is doing or not */ 1251 SBI_NEED_SB_WRITE, /* need to recover superblock */ 1252 SBI_NEED_CP, /* need to checkpoint */ 1253 SBI_IS_SHUTDOWN, /* shutdown by ioctl */ 1254 SBI_IS_RECOVERED, /* recovered orphan/data */ 1255 SBI_CP_DISABLED, /* CP was disabled last mount */ 1256 SBI_CP_DISABLED_QUICK, /* CP was disabled quickly */ 1257 SBI_QUOTA_NEED_FLUSH, /* need to flush quota info in CP */ 1258 SBI_QUOTA_SKIP_FLUSH, /* skip flushing quota in current CP */ 1259 SBI_QUOTA_NEED_REPAIR, /* quota file may be corrupted */ 1260 SBI_IS_RESIZEFS, /* resizefs is in process */ 1261 }; 1262 1263 enum { 1264 CP_TIME, 1265 REQ_TIME, 1266 DISCARD_TIME, 1267 GC_TIME, 1268 DISABLE_TIME, 1269 UMOUNT_DISCARD_TIMEOUT, 1270 MAX_TIME, 1271 }; 1272 1273 enum { 1274 GC_NORMAL, 1275 GC_IDLE_CB, 1276 GC_IDLE_GREEDY, 1277 GC_IDLE_AT, 1278 GC_URGENT_HIGH, 1279 GC_URGENT_LOW, 1280 MAX_GC_MODE, 1281 }; 1282 1283 enum { 1284 BGGC_MODE_ON, /* background gc is on */ 1285 BGGC_MODE_OFF, /* background gc is off */ 1286 BGGC_MODE_SYNC, /* 1287 * background gc is on, migrating blocks 1288 * like foreground gc 1289 */ 1290 }; 1291 1292 enum { 1293 FS_MODE_ADAPTIVE, /* use both lfs/ssr allocation */ 1294 FS_MODE_LFS, /* use lfs allocation only */ 1295 FS_MODE_FRAGMENT_SEG, /* segment fragmentation mode */ 1296 FS_MODE_FRAGMENT_BLK, /* block fragmentation mode */ 1297 }; 1298 1299 enum { 1300 WHINT_MODE_OFF, /* not pass down write hints */ 1301 WHINT_MODE_USER, /* try to pass down hints given by users */ 1302 WHINT_MODE_FS, /* pass down hints with F2FS policy */ 1303 }; 1304 1305 enum { 1306 ALLOC_MODE_DEFAULT, /* stay default */ 1307 ALLOC_MODE_REUSE, /* reuse segments as much as possible */ 1308 }; 1309 1310 enum fsync_mode { 1311 FSYNC_MODE_POSIX, /* fsync follows posix semantics */ 1312 FSYNC_MODE_STRICT, /* fsync behaves in line with ext4 */ 1313 FSYNC_MODE_NOBARRIER, /* fsync behaves nobarrier based on posix */ 1314 }; 1315 1316 enum { 1317 COMPR_MODE_FS, /* 1318 * automatically compress compression 1319 * enabled files 1320 */ 1321 COMPR_MODE_USER, /* 1322 * automatical compression is disabled. 1323 * user can control the file compression 1324 * using ioctls 1325 */ 1326 }; 1327 1328 enum { 1329 DISCARD_UNIT_BLOCK, /* basic discard unit is block */ 1330 DISCARD_UNIT_SEGMENT, /* basic discard unit is segment */ 1331 DISCARD_UNIT_SECTION, /* basic discard unit is section */ 1332 }; 1333 1334 static inline int f2fs_test_bit(unsigned int nr, char *addr); 1335 static inline void f2fs_set_bit(unsigned int nr, char *addr); 1336 static inline void f2fs_clear_bit(unsigned int nr, char *addr); 1337 1338 /* 1339 * Layout of f2fs page.private: 1340 * 1341 * Layout A: lowest bit should be 1 1342 * | bit0 = 1 | bit1 | bit2 | ... | bit MAX | private data .... | 1343 * bit 0 PAGE_PRIVATE_NOT_POINTER 1344 * bit 1 PAGE_PRIVATE_ATOMIC_WRITE 1345 * bit 2 PAGE_PRIVATE_DUMMY_WRITE 1346 * bit 3 PAGE_PRIVATE_ONGOING_MIGRATION 1347 * bit 4 PAGE_PRIVATE_INLINE_INODE 1348 * bit 5 PAGE_PRIVATE_REF_RESOURCE 1349 * bit 6- f2fs private data 1350 * 1351 * Layout B: lowest bit should be 0 1352 * page.private is a wrapped pointer. 1353 */ 1354 enum { 1355 PAGE_PRIVATE_NOT_POINTER, /* private contains non-pointer data */ 1356 PAGE_PRIVATE_ATOMIC_WRITE, /* data page from atomic write path */ 1357 PAGE_PRIVATE_DUMMY_WRITE, /* data page for padding aligned IO */ 1358 PAGE_PRIVATE_ONGOING_MIGRATION, /* data page which is on-going migrating */ 1359 PAGE_PRIVATE_INLINE_INODE, /* inode page contains inline data */ 1360 PAGE_PRIVATE_REF_RESOURCE, /* dirty page has referenced resources */ 1361 PAGE_PRIVATE_MAX 1362 }; 1363 1364 #define PAGE_PRIVATE_GET_FUNC(name, flagname) \ 1365 static inline bool page_private_##name(struct page *page) \ 1366 { \ 1367 return PagePrivate(page) && \ 1368 test_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)) && \ 1369 test_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \ 1370 } 1371 1372 #define PAGE_PRIVATE_SET_FUNC(name, flagname) \ 1373 static inline void set_page_private_##name(struct page *page) \ 1374 { \ 1375 if (!PagePrivate(page)) { \ 1376 get_page(page); \ 1377 SetPagePrivate(page); \ 1378 set_page_private(page, 0); \ 1379 } \ 1380 set_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)); \ 1381 set_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \ 1382 } 1383 1384 #define PAGE_PRIVATE_CLEAR_FUNC(name, flagname) \ 1385 static inline void clear_page_private_##name(struct page *page) \ 1386 { \ 1387 clear_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \ 1388 if (page_private(page) == 1 << PAGE_PRIVATE_NOT_POINTER) { \ 1389 set_page_private(page, 0); \ 1390 if (PagePrivate(page)) { \ 1391 ClearPagePrivate(page); \ 1392 put_page(page); \ 1393 }\ 1394 } \ 1395 } 1396 1397 PAGE_PRIVATE_GET_FUNC(nonpointer, NOT_POINTER); 1398 PAGE_PRIVATE_GET_FUNC(reference, REF_RESOURCE); 1399 PAGE_PRIVATE_GET_FUNC(inline, INLINE_INODE); 1400 PAGE_PRIVATE_GET_FUNC(gcing, ONGOING_MIGRATION); 1401 PAGE_PRIVATE_GET_FUNC(atomic, ATOMIC_WRITE); 1402 PAGE_PRIVATE_GET_FUNC(dummy, DUMMY_WRITE); 1403 1404 PAGE_PRIVATE_SET_FUNC(reference, REF_RESOURCE); 1405 PAGE_PRIVATE_SET_FUNC(inline, INLINE_INODE); 1406 PAGE_PRIVATE_SET_FUNC(gcing, ONGOING_MIGRATION); 1407 PAGE_PRIVATE_SET_FUNC(atomic, ATOMIC_WRITE); 1408 PAGE_PRIVATE_SET_FUNC(dummy, DUMMY_WRITE); 1409 1410 PAGE_PRIVATE_CLEAR_FUNC(reference, REF_RESOURCE); 1411 PAGE_PRIVATE_CLEAR_FUNC(inline, INLINE_INODE); 1412 PAGE_PRIVATE_CLEAR_FUNC(gcing, ONGOING_MIGRATION); 1413 PAGE_PRIVATE_CLEAR_FUNC(atomic, ATOMIC_WRITE); 1414 PAGE_PRIVATE_CLEAR_FUNC(dummy, DUMMY_WRITE); 1415 1416 static inline unsigned long get_page_private_data(struct page *page) 1417 { 1418 unsigned long data = page_private(page); 1419 1420 if (!test_bit(PAGE_PRIVATE_NOT_POINTER, &data)) 1421 return 0; 1422 return data >> PAGE_PRIVATE_MAX; 1423 } 1424 1425 static inline void set_page_private_data(struct page *page, unsigned long data) 1426 { 1427 if (!PagePrivate(page)) { 1428 get_page(page); 1429 SetPagePrivate(page); 1430 set_page_private(page, 0); 1431 } 1432 set_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)); 1433 page_private(page) |= data << PAGE_PRIVATE_MAX; 1434 } 1435 1436 static inline void clear_page_private_data(struct page *page) 1437 { 1438 page_private(page) &= (1 << PAGE_PRIVATE_MAX) - 1; 1439 if (page_private(page) == 1 << PAGE_PRIVATE_NOT_POINTER) { 1440 set_page_private(page, 0); 1441 if (PagePrivate(page)) { 1442 ClearPagePrivate(page); 1443 put_page(page); 1444 } 1445 } 1446 } 1447 1448 /* For compression */ 1449 enum compress_algorithm_type { 1450 COMPRESS_LZO, 1451 COMPRESS_LZ4, 1452 COMPRESS_ZSTD, 1453 COMPRESS_LZORLE, 1454 COMPRESS_MAX, 1455 }; 1456 1457 enum compress_flag { 1458 COMPRESS_CHKSUM, 1459 COMPRESS_MAX_FLAG, 1460 }; 1461 1462 #define COMPRESS_WATERMARK 20 1463 #define COMPRESS_PERCENT 20 1464 1465 #define COMPRESS_DATA_RESERVED_SIZE 4 1466 struct compress_data { 1467 __le32 clen; /* compressed data size */ 1468 __le32 chksum; /* compressed data chksum */ 1469 __le32 reserved[COMPRESS_DATA_RESERVED_SIZE]; /* reserved */ 1470 u8 cdata[]; /* compressed data */ 1471 }; 1472 1473 #define COMPRESS_HEADER_SIZE (sizeof(struct compress_data)) 1474 1475 #define F2FS_COMPRESSED_PAGE_MAGIC 0xF5F2C000 1476 1477 #define COMPRESS_LEVEL_OFFSET 8 1478 1479 /* compress context */ 1480 struct compress_ctx { 1481 struct inode *inode; /* inode the context belong to */ 1482 pgoff_t cluster_idx; /* cluster index number */ 1483 unsigned int cluster_size; /* page count in cluster */ 1484 unsigned int log_cluster_size; /* log of cluster size */ 1485 struct page **rpages; /* pages store raw data in cluster */ 1486 unsigned int nr_rpages; /* total page number in rpages */ 1487 struct page **cpages; /* pages store compressed data in cluster */ 1488 unsigned int nr_cpages; /* total page number in cpages */ 1489 void *rbuf; /* virtual mapped address on rpages */ 1490 struct compress_data *cbuf; /* virtual mapped address on cpages */ 1491 size_t rlen; /* valid data length in rbuf */ 1492 size_t clen; /* valid data length in cbuf */ 1493 void *private; /* payload buffer for specified compression algorithm */ 1494 void *private2; /* extra payload buffer */ 1495 }; 1496 1497 /* compress context for write IO path */ 1498 struct compress_io_ctx { 1499 u32 magic; /* magic number to indicate page is compressed */ 1500 struct inode *inode; /* inode the context belong to */ 1501 struct page **rpages; /* pages store raw data in cluster */ 1502 unsigned int nr_rpages; /* total page number in rpages */ 1503 atomic_t pending_pages; /* in-flight compressed page count */ 1504 }; 1505 1506 /* Context for decompressing one cluster on the read IO path */ 1507 struct decompress_io_ctx { 1508 u32 magic; /* magic number to indicate page is compressed */ 1509 struct inode *inode; /* inode the context belong to */ 1510 pgoff_t cluster_idx; /* cluster index number */ 1511 unsigned int cluster_size; /* page count in cluster */ 1512 unsigned int log_cluster_size; /* log of cluster size */ 1513 struct page **rpages; /* pages store raw data in cluster */ 1514 unsigned int nr_rpages; /* total page number in rpages */ 1515 struct page **cpages; /* pages store compressed data in cluster */ 1516 unsigned int nr_cpages; /* total page number in cpages */ 1517 struct page **tpages; /* temp pages to pad holes in cluster */ 1518 void *rbuf; /* virtual mapped address on rpages */ 1519 struct compress_data *cbuf; /* virtual mapped address on cpages */ 1520 size_t rlen; /* valid data length in rbuf */ 1521 size_t clen; /* valid data length in cbuf */ 1522 1523 /* 1524 * The number of compressed pages remaining to be read in this cluster. 1525 * This is initially nr_cpages. It is decremented by 1 each time a page 1526 * has been read (or failed to be read). When it reaches 0, the cluster 1527 * is decompressed (or an error is reported). 1528 * 1529 * If an error occurs before all the pages have been submitted for I/O, 1530 * then this will never reach 0. In this case the I/O submitter is 1531 * responsible for calling f2fs_decompress_end_io() instead. 1532 */ 1533 atomic_t remaining_pages; 1534 1535 /* 1536 * Number of references to this decompress_io_ctx. 1537 * 1538 * One reference is held for I/O completion. This reference is dropped 1539 * after the pagecache pages are updated and unlocked -- either after 1540 * decompression (and verity if enabled), or after an error. 1541 * 1542 * In addition, each compressed page holds a reference while it is in a 1543 * bio. These references are necessary prevent compressed pages from 1544 * being freed while they are still in a bio. 1545 */ 1546 refcount_t refcnt; 1547 1548 bool failed; /* IO error occurred before decompression? */ 1549 bool need_verity; /* need fs-verity verification after decompression? */ 1550 void *private; /* payload buffer for specified decompression algorithm */ 1551 void *private2; /* extra payload buffer */ 1552 struct work_struct verity_work; /* work to verify the decompressed pages */ 1553 }; 1554 1555 #define NULL_CLUSTER ((unsigned int)(~0)) 1556 #define MIN_COMPRESS_LOG_SIZE 2 1557 #define MAX_COMPRESS_LOG_SIZE 8 1558 #define MAX_COMPRESS_WINDOW_SIZE(log_size) ((PAGE_SIZE) << (log_size)) 1559 1560 struct f2fs_sb_info { 1561 struct super_block *sb; /* pointer to VFS super block */ 1562 struct proc_dir_entry *s_proc; /* proc entry */ 1563 struct f2fs_super_block *raw_super; /* raw super block pointer */ 1564 struct rw_semaphore sb_lock; /* lock for raw super block */ 1565 int valid_super_block; /* valid super block no */ 1566 unsigned long s_flag; /* flags for sbi */ 1567 struct mutex writepages; /* mutex for writepages() */ 1568 1569 #ifdef CONFIG_BLK_DEV_ZONED 1570 unsigned int blocks_per_blkz; /* F2FS blocks per zone */ 1571 unsigned int log_blocks_per_blkz; /* log2 F2FS blocks per zone */ 1572 #endif 1573 1574 /* for node-related operations */ 1575 struct f2fs_nm_info *nm_info; /* node manager */ 1576 struct inode *node_inode; /* cache node blocks */ 1577 1578 /* for segment-related operations */ 1579 struct f2fs_sm_info *sm_info; /* segment manager */ 1580 1581 /* for bio operations */ 1582 struct f2fs_bio_info *write_io[NR_PAGE_TYPE]; /* for write bios */ 1583 /* keep migration IO order for LFS mode */ 1584 struct rw_semaphore io_order_lock; 1585 mempool_t *write_io_dummy; /* Dummy pages */ 1586 1587 /* for checkpoint */ 1588 struct f2fs_checkpoint *ckpt; /* raw checkpoint pointer */ 1589 int cur_cp_pack; /* remain current cp pack */ 1590 spinlock_t cp_lock; /* for flag in ckpt */ 1591 struct inode *meta_inode; /* cache meta blocks */ 1592 struct rw_semaphore cp_global_sem; /* checkpoint procedure lock */ 1593 struct rw_semaphore cp_rwsem; /* blocking FS operations */ 1594 struct rw_semaphore node_write; /* locking node writes */ 1595 struct rw_semaphore node_change; /* locking node change */ 1596 wait_queue_head_t cp_wait; 1597 unsigned long last_time[MAX_TIME]; /* to store time in jiffies */ 1598 long interval_time[MAX_TIME]; /* to store thresholds */ 1599 struct ckpt_req_control cprc_info; /* for checkpoint request control */ 1600 1601 struct inode_management im[MAX_INO_ENTRY]; /* manage inode cache */ 1602 1603 spinlock_t fsync_node_lock; /* for node entry lock */ 1604 struct list_head fsync_node_list; /* node list head */ 1605 unsigned int fsync_seg_id; /* sequence id */ 1606 unsigned int fsync_node_num; /* number of node entries */ 1607 1608 /* for orphan inode, use 0'th array */ 1609 unsigned int max_orphans; /* max orphan inodes */ 1610 1611 /* for inode management */ 1612 struct list_head inode_list[NR_INODE_TYPE]; /* dirty inode list */ 1613 spinlock_t inode_lock[NR_INODE_TYPE]; /* for dirty inode list lock */ 1614 struct mutex flush_lock; /* for flush exclusion */ 1615 1616 /* for extent tree cache */ 1617 struct radix_tree_root extent_tree_root;/* cache extent cache entries */ 1618 struct mutex extent_tree_lock; /* locking extent radix tree */ 1619 struct list_head extent_list; /* lru list for shrinker */ 1620 spinlock_t extent_lock; /* locking extent lru list */ 1621 atomic_t total_ext_tree; /* extent tree count */ 1622 struct list_head zombie_list; /* extent zombie tree list */ 1623 atomic_t total_zombie_tree; /* extent zombie tree count */ 1624 atomic_t total_ext_node; /* extent info count */ 1625 1626 /* basic filesystem units */ 1627 unsigned int log_sectors_per_block; /* log2 sectors per block */ 1628 unsigned int log_blocksize; /* log2 block size */ 1629 unsigned int blocksize; /* block size */ 1630 unsigned int root_ino_num; /* root inode number*/ 1631 unsigned int node_ino_num; /* node inode number*/ 1632 unsigned int meta_ino_num; /* meta inode number*/ 1633 unsigned int log_blocks_per_seg; /* log2 blocks per segment */ 1634 unsigned int blocks_per_seg; /* blocks per segment */ 1635 unsigned int segs_per_sec; /* segments per section */ 1636 unsigned int secs_per_zone; /* sections per zone */ 1637 unsigned int total_sections; /* total section count */ 1638 unsigned int total_node_count; /* total node block count */ 1639 unsigned int total_valid_node_count; /* valid node block count */ 1640 int dir_level; /* directory level */ 1641 int readdir_ra; /* readahead inode in readdir */ 1642 u64 max_io_bytes; /* max io bytes to merge IOs */ 1643 1644 block_t user_block_count; /* # of user blocks */ 1645 block_t total_valid_block_count; /* # of valid blocks */ 1646 block_t discard_blks; /* discard command candidats */ 1647 block_t last_valid_block_count; /* for recovery */ 1648 block_t reserved_blocks; /* configurable reserved blocks */ 1649 block_t current_reserved_blocks; /* current reserved blocks */ 1650 1651 /* Additional tracking for no checkpoint mode */ 1652 block_t unusable_block_count; /* # of blocks saved by last cp */ 1653 1654 unsigned int nquota_files; /* # of quota sysfile */ 1655 struct rw_semaphore quota_sem; /* blocking cp for flags */ 1656 1657 /* # of pages, see count_type */ 1658 atomic_t nr_pages[NR_COUNT_TYPE]; 1659 /* # of allocated blocks */ 1660 struct percpu_counter alloc_valid_block_count; 1661 1662 /* writeback control */ 1663 atomic_t wb_sync_req[META]; /* count # of WB_SYNC threads */ 1664 1665 /* valid inode count */ 1666 struct percpu_counter total_valid_inode_count; 1667 1668 struct f2fs_mount_info mount_opt; /* mount options */ 1669 1670 /* for cleaning operations */ 1671 struct rw_semaphore gc_lock; /* 1672 * semaphore for GC, avoid 1673 * race between GC and GC or CP 1674 */ 1675 struct f2fs_gc_kthread *gc_thread; /* GC thread */ 1676 struct atgc_management am; /* atgc management */ 1677 unsigned int cur_victim_sec; /* current victim section num */ 1678 unsigned int gc_mode; /* current GC state */ 1679 unsigned int next_victim_seg[2]; /* next segment in victim section */ 1680 1681 /* for skip statistic */ 1682 unsigned int atomic_files; /* # of opened atomic file */ 1683 unsigned long long skipped_atomic_files[2]; /* FG_GC and BG_GC */ 1684 unsigned long long skipped_gc_rwsem; /* FG_GC only */ 1685 1686 /* threshold for gc trials on pinned files */ 1687 u64 gc_pin_file_threshold; 1688 struct rw_semaphore pin_sem; 1689 1690 /* maximum # of trials to find a victim segment for SSR and GC */ 1691 unsigned int max_victim_search; 1692 /* migration granularity of garbage collection, unit: segment */ 1693 unsigned int migration_granularity; 1694 1695 /* 1696 * for stat information. 1697 * one is for the LFS mode, and the other is for the SSR mode. 1698 */ 1699 #ifdef CONFIG_F2FS_STAT_FS 1700 struct f2fs_stat_info *stat_info; /* FS status information */ 1701 atomic_t meta_count[META_MAX]; /* # of meta blocks */ 1702 unsigned int segment_count[2]; /* # of allocated segments */ 1703 unsigned int block_count[2]; /* # of allocated blocks */ 1704 atomic_t inplace_count; /* # of inplace update */ 1705 atomic64_t total_hit_ext; /* # of lookup extent cache */ 1706 atomic64_t read_hit_rbtree; /* # of hit rbtree extent node */ 1707 atomic64_t read_hit_largest; /* # of hit largest extent node */ 1708 atomic64_t read_hit_cached; /* # of hit cached extent node */ 1709 atomic_t inline_xattr; /* # of inline_xattr inodes */ 1710 atomic_t inline_inode; /* # of inline_data inodes */ 1711 atomic_t inline_dir; /* # of inline_dentry inodes */ 1712 atomic_t compr_inode; /* # of compressed inodes */ 1713 atomic64_t compr_blocks; /* # of compressed blocks */ 1714 atomic_t vw_cnt; /* # of volatile writes */ 1715 atomic_t max_aw_cnt; /* max # of atomic writes */ 1716 atomic_t max_vw_cnt; /* max # of volatile writes */ 1717 unsigned int io_skip_bggc; /* skip background gc for in-flight IO */ 1718 unsigned int other_skip_bggc; /* skip background gc for other reasons */ 1719 unsigned int ndirty_inode[NR_INODE_TYPE]; /* # of dirty inodes */ 1720 #endif 1721 spinlock_t stat_lock; /* lock for stat operations */ 1722 1723 /* to attach REQ_META|REQ_FUA flags */ 1724 unsigned int data_io_flag; 1725 unsigned int node_io_flag; 1726 1727 /* For sysfs suppport */ 1728 struct kobject s_kobj; /* /sys/fs/f2fs/<devname> */ 1729 struct completion s_kobj_unregister; 1730 1731 struct kobject s_stat_kobj; /* /sys/fs/f2fs/<devname>/stat */ 1732 struct completion s_stat_kobj_unregister; 1733 1734 struct kobject s_feature_list_kobj; /* /sys/fs/f2fs/<devname>/feature_list */ 1735 struct completion s_feature_list_kobj_unregister; 1736 1737 /* For shrinker support */ 1738 struct list_head s_list; 1739 struct mutex umount_mutex; 1740 unsigned int shrinker_run_no; 1741 1742 /* For multi devices */ 1743 int s_ndevs; /* number of devices */ 1744 struct f2fs_dev_info *devs; /* for device list */ 1745 unsigned int dirty_device; /* for checkpoint data flush */ 1746 spinlock_t dev_lock; /* protect dirty_device */ 1747 bool aligned_blksize; /* all devices has the same logical blksize */ 1748 1749 /* For write statistics */ 1750 u64 sectors_written_start; 1751 u64 kbytes_written; 1752 1753 /* Reference to checksum algorithm driver via cryptoapi */ 1754 struct crypto_shash *s_chksum_driver; 1755 1756 /* Precomputed FS UUID checksum for seeding other checksums */ 1757 __u32 s_chksum_seed; 1758 1759 struct workqueue_struct *post_read_wq; /* post read workqueue */ 1760 1761 struct kmem_cache *inline_xattr_slab; /* inline xattr entry */ 1762 unsigned int inline_xattr_slab_size; /* default inline xattr slab size */ 1763 1764 /* For reclaimed segs statistics per each GC mode */ 1765 unsigned int gc_segment_mode; /* GC state for reclaimed segments */ 1766 unsigned int gc_reclaimed_segs[MAX_GC_MODE]; /* Reclaimed segs for each mode */ 1767 1768 unsigned long seq_file_ra_mul; /* multiplier for ra_pages of seq. files in fadvise */ 1769 1770 int max_fragment_chunk; /* max chunk size for block fragmentation mode */ 1771 int max_fragment_hole; /* max hole size for block fragmentation mode */ 1772 1773 #ifdef CONFIG_F2FS_FS_COMPRESSION 1774 struct kmem_cache *page_array_slab; /* page array entry */ 1775 unsigned int page_array_slab_size; /* default page array slab size */ 1776 1777 /* For runtime compression statistics */ 1778 u64 compr_written_block; 1779 u64 compr_saved_block; 1780 u32 compr_new_inode; 1781 1782 /* For compressed block cache */ 1783 struct inode *compress_inode; /* cache compressed blocks */ 1784 unsigned int compress_percent; /* cache page percentage */ 1785 unsigned int compress_watermark; /* cache page watermark */ 1786 atomic_t compress_page_hit; /* cache hit count */ 1787 #endif 1788 1789 #ifdef CONFIG_F2FS_IOSTAT 1790 /* For app/fs IO statistics */ 1791 spinlock_t iostat_lock; 1792 unsigned long long rw_iostat[NR_IO_TYPE]; 1793 unsigned long long prev_rw_iostat[NR_IO_TYPE]; 1794 bool iostat_enable; 1795 unsigned long iostat_next_period; 1796 unsigned int iostat_period_ms; 1797 1798 /* For io latency related statistics info in one iostat period */ 1799 spinlock_t iostat_lat_lock; 1800 struct iostat_lat_info *iostat_io_lat; 1801 #endif 1802 }; 1803 1804 struct f2fs_private_dio { 1805 struct inode *inode; 1806 void *orig_private; 1807 bio_end_io_t *orig_end_io; 1808 bool write; 1809 }; 1810 1811 #ifdef CONFIG_F2FS_FAULT_INJECTION 1812 #define f2fs_show_injection_info(sbi, type) \ 1813 printk_ratelimited("%sF2FS-fs (%s) : inject %s in %s of %pS\n", \ 1814 KERN_INFO, sbi->sb->s_id, \ 1815 f2fs_fault_name[type], \ 1816 __func__, __builtin_return_address(0)) 1817 static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type) 1818 { 1819 struct f2fs_fault_info *ffi = &F2FS_OPTION(sbi).fault_info; 1820 1821 if (!ffi->inject_rate) 1822 return false; 1823 1824 if (!IS_FAULT_SET(ffi, type)) 1825 return false; 1826 1827 atomic_inc(&ffi->inject_ops); 1828 if (atomic_read(&ffi->inject_ops) >= ffi->inject_rate) { 1829 atomic_set(&ffi->inject_ops, 0); 1830 return true; 1831 } 1832 return false; 1833 } 1834 #else 1835 #define f2fs_show_injection_info(sbi, type) do { } while (0) 1836 static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type) 1837 { 1838 return false; 1839 } 1840 #endif 1841 1842 /* 1843 * Test if the mounted volume is a multi-device volume. 1844 * - For a single regular disk volume, sbi->s_ndevs is 0. 1845 * - For a single zoned disk volume, sbi->s_ndevs is 1. 1846 * - For a multi-device volume, sbi->s_ndevs is always 2 or more. 1847 */ 1848 static inline bool f2fs_is_multi_device(struct f2fs_sb_info *sbi) 1849 { 1850 return sbi->s_ndevs > 1; 1851 } 1852 1853 static inline void f2fs_update_time(struct f2fs_sb_info *sbi, int type) 1854 { 1855 unsigned long now = jiffies; 1856 1857 sbi->last_time[type] = now; 1858 1859 /* DISCARD_TIME and GC_TIME are based on REQ_TIME */ 1860 if (type == REQ_TIME) { 1861 sbi->last_time[DISCARD_TIME] = now; 1862 sbi->last_time[GC_TIME] = now; 1863 } 1864 } 1865 1866 static inline bool f2fs_time_over(struct f2fs_sb_info *sbi, int type) 1867 { 1868 unsigned long interval = sbi->interval_time[type] * HZ; 1869 1870 return time_after(jiffies, sbi->last_time[type] + interval); 1871 } 1872 1873 static inline unsigned int f2fs_time_to_wait(struct f2fs_sb_info *sbi, 1874 int type) 1875 { 1876 unsigned long interval = sbi->interval_time[type] * HZ; 1877 unsigned int wait_ms = 0; 1878 long delta; 1879 1880 delta = (sbi->last_time[type] + interval) - jiffies; 1881 if (delta > 0) 1882 wait_ms = jiffies_to_msecs(delta); 1883 1884 return wait_ms; 1885 } 1886 1887 /* 1888 * Inline functions 1889 */ 1890 static inline u32 __f2fs_crc32(struct f2fs_sb_info *sbi, u32 crc, 1891 const void *address, unsigned int length) 1892 { 1893 struct { 1894 struct shash_desc shash; 1895 char ctx[4]; 1896 } desc; 1897 int err; 1898 1899 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver) != sizeof(desc.ctx)); 1900 1901 desc.shash.tfm = sbi->s_chksum_driver; 1902 *(u32 *)desc.ctx = crc; 1903 1904 err = crypto_shash_update(&desc.shash, address, length); 1905 BUG_ON(err); 1906 1907 return *(u32 *)desc.ctx; 1908 } 1909 1910 static inline u32 f2fs_crc32(struct f2fs_sb_info *sbi, const void *address, 1911 unsigned int length) 1912 { 1913 return __f2fs_crc32(sbi, F2FS_SUPER_MAGIC, address, length); 1914 } 1915 1916 static inline bool f2fs_crc_valid(struct f2fs_sb_info *sbi, __u32 blk_crc, 1917 void *buf, size_t buf_size) 1918 { 1919 return f2fs_crc32(sbi, buf, buf_size) == blk_crc; 1920 } 1921 1922 static inline u32 f2fs_chksum(struct f2fs_sb_info *sbi, u32 crc, 1923 const void *address, unsigned int length) 1924 { 1925 return __f2fs_crc32(sbi, crc, address, length); 1926 } 1927 1928 static inline struct f2fs_inode_info *F2FS_I(struct inode *inode) 1929 { 1930 return container_of(inode, struct f2fs_inode_info, vfs_inode); 1931 } 1932 1933 static inline struct f2fs_sb_info *F2FS_SB(struct super_block *sb) 1934 { 1935 return sb->s_fs_info; 1936 } 1937 1938 static inline struct f2fs_sb_info *F2FS_I_SB(struct inode *inode) 1939 { 1940 return F2FS_SB(inode->i_sb); 1941 } 1942 1943 static inline struct f2fs_sb_info *F2FS_M_SB(struct address_space *mapping) 1944 { 1945 return F2FS_I_SB(mapping->host); 1946 } 1947 1948 static inline struct f2fs_sb_info *F2FS_P_SB(struct page *page) 1949 { 1950 return F2FS_M_SB(page_file_mapping(page)); 1951 } 1952 1953 static inline struct f2fs_super_block *F2FS_RAW_SUPER(struct f2fs_sb_info *sbi) 1954 { 1955 return (struct f2fs_super_block *)(sbi->raw_super); 1956 } 1957 1958 static inline struct f2fs_checkpoint *F2FS_CKPT(struct f2fs_sb_info *sbi) 1959 { 1960 return (struct f2fs_checkpoint *)(sbi->ckpt); 1961 } 1962 1963 static inline struct f2fs_node *F2FS_NODE(struct page *page) 1964 { 1965 return (struct f2fs_node *)page_address(page); 1966 } 1967 1968 static inline struct f2fs_inode *F2FS_INODE(struct page *page) 1969 { 1970 return &((struct f2fs_node *)page_address(page))->i; 1971 } 1972 1973 static inline struct f2fs_nm_info *NM_I(struct f2fs_sb_info *sbi) 1974 { 1975 return (struct f2fs_nm_info *)(sbi->nm_info); 1976 } 1977 1978 static inline struct f2fs_sm_info *SM_I(struct f2fs_sb_info *sbi) 1979 { 1980 return (struct f2fs_sm_info *)(sbi->sm_info); 1981 } 1982 1983 static inline struct sit_info *SIT_I(struct f2fs_sb_info *sbi) 1984 { 1985 return (struct sit_info *)(SM_I(sbi)->sit_info); 1986 } 1987 1988 static inline struct free_segmap_info *FREE_I(struct f2fs_sb_info *sbi) 1989 { 1990 return (struct free_segmap_info *)(SM_I(sbi)->free_info); 1991 } 1992 1993 static inline struct dirty_seglist_info *DIRTY_I(struct f2fs_sb_info *sbi) 1994 { 1995 return (struct dirty_seglist_info *)(SM_I(sbi)->dirty_info); 1996 } 1997 1998 static inline struct address_space *META_MAPPING(struct f2fs_sb_info *sbi) 1999 { 2000 return sbi->meta_inode->i_mapping; 2001 } 2002 2003 static inline struct address_space *NODE_MAPPING(struct f2fs_sb_info *sbi) 2004 { 2005 return sbi->node_inode->i_mapping; 2006 } 2007 2008 static inline bool is_sbi_flag_set(struct f2fs_sb_info *sbi, unsigned int type) 2009 { 2010 return test_bit(type, &sbi->s_flag); 2011 } 2012 2013 static inline void set_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type) 2014 { 2015 set_bit(type, &sbi->s_flag); 2016 } 2017 2018 static inline void clear_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type) 2019 { 2020 clear_bit(type, &sbi->s_flag); 2021 } 2022 2023 static inline unsigned long long cur_cp_version(struct f2fs_checkpoint *cp) 2024 { 2025 return le64_to_cpu(cp->checkpoint_ver); 2026 } 2027 2028 static inline unsigned long f2fs_qf_ino(struct super_block *sb, int type) 2029 { 2030 if (type < F2FS_MAX_QUOTAS) 2031 return le32_to_cpu(F2FS_SB(sb)->raw_super->qf_ino[type]); 2032 return 0; 2033 } 2034 2035 static inline __u64 cur_cp_crc(struct f2fs_checkpoint *cp) 2036 { 2037 size_t crc_offset = le32_to_cpu(cp->checksum_offset); 2038 return le32_to_cpu(*((__le32 *)((unsigned char *)cp + crc_offset))); 2039 } 2040 2041 static inline bool __is_set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 2042 { 2043 unsigned int ckpt_flags = le32_to_cpu(cp->ckpt_flags); 2044 2045 return ckpt_flags & f; 2046 } 2047 2048 static inline bool is_set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 2049 { 2050 return __is_set_ckpt_flags(F2FS_CKPT(sbi), f); 2051 } 2052 2053 static inline void __set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 2054 { 2055 unsigned int ckpt_flags; 2056 2057 ckpt_flags = le32_to_cpu(cp->ckpt_flags); 2058 ckpt_flags |= f; 2059 cp->ckpt_flags = cpu_to_le32(ckpt_flags); 2060 } 2061 2062 static inline void set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 2063 { 2064 unsigned long flags; 2065 2066 spin_lock_irqsave(&sbi->cp_lock, flags); 2067 __set_ckpt_flags(F2FS_CKPT(sbi), f); 2068 spin_unlock_irqrestore(&sbi->cp_lock, flags); 2069 } 2070 2071 static inline void __clear_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 2072 { 2073 unsigned int ckpt_flags; 2074 2075 ckpt_flags = le32_to_cpu(cp->ckpt_flags); 2076 ckpt_flags &= (~f); 2077 cp->ckpt_flags = cpu_to_le32(ckpt_flags); 2078 } 2079 2080 static inline void clear_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 2081 { 2082 unsigned long flags; 2083 2084 spin_lock_irqsave(&sbi->cp_lock, flags); 2085 __clear_ckpt_flags(F2FS_CKPT(sbi), f); 2086 spin_unlock_irqrestore(&sbi->cp_lock, flags); 2087 } 2088 2089 static inline void f2fs_lock_op(struct f2fs_sb_info *sbi) 2090 { 2091 down_read(&sbi->cp_rwsem); 2092 } 2093 2094 static inline int f2fs_trylock_op(struct f2fs_sb_info *sbi) 2095 { 2096 return down_read_trylock(&sbi->cp_rwsem); 2097 } 2098 2099 static inline void f2fs_unlock_op(struct f2fs_sb_info *sbi) 2100 { 2101 up_read(&sbi->cp_rwsem); 2102 } 2103 2104 static inline void f2fs_lock_all(struct f2fs_sb_info *sbi) 2105 { 2106 down_write(&sbi->cp_rwsem); 2107 } 2108 2109 static inline void f2fs_unlock_all(struct f2fs_sb_info *sbi) 2110 { 2111 up_write(&sbi->cp_rwsem); 2112 } 2113 2114 static inline int __get_cp_reason(struct f2fs_sb_info *sbi) 2115 { 2116 int reason = CP_SYNC; 2117 2118 if (test_opt(sbi, FASTBOOT)) 2119 reason = CP_FASTBOOT; 2120 if (is_sbi_flag_set(sbi, SBI_IS_CLOSE)) 2121 reason = CP_UMOUNT; 2122 return reason; 2123 } 2124 2125 static inline bool __remain_node_summaries(int reason) 2126 { 2127 return (reason & (CP_UMOUNT | CP_FASTBOOT)); 2128 } 2129 2130 static inline bool __exist_node_summaries(struct f2fs_sb_info *sbi) 2131 { 2132 return (is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG) || 2133 is_set_ckpt_flags(sbi, CP_FASTBOOT_FLAG)); 2134 } 2135 2136 /* 2137 * Check whether the inode has blocks or not 2138 */ 2139 static inline int F2FS_HAS_BLOCKS(struct inode *inode) 2140 { 2141 block_t xattr_block = F2FS_I(inode)->i_xattr_nid ? 1 : 0; 2142 2143 return (inode->i_blocks >> F2FS_LOG_SECTORS_PER_BLOCK) > xattr_block; 2144 } 2145 2146 static inline bool f2fs_has_xattr_block(unsigned int ofs) 2147 { 2148 return ofs == XATTR_NODE_OFFSET; 2149 } 2150 2151 static inline bool __allow_reserved_blocks(struct f2fs_sb_info *sbi, 2152 struct inode *inode, bool cap) 2153 { 2154 if (!inode) 2155 return true; 2156 if (!test_opt(sbi, RESERVE_ROOT)) 2157 return false; 2158 if (IS_NOQUOTA(inode)) 2159 return true; 2160 if (uid_eq(F2FS_OPTION(sbi).s_resuid, current_fsuid())) 2161 return true; 2162 if (!gid_eq(F2FS_OPTION(sbi).s_resgid, GLOBAL_ROOT_GID) && 2163 in_group_p(F2FS_OPTION(sbi).s_resgid)) 2164 return true; 2165 if (cap && capable(CAP_SYS_RESOURCE)) 2166 return true; 2167 return false; 2168 } 2169 2170 static inline void f2fs_i_blocks_write(struct inode *, block_t, bool, bool); 2171 static inline int inc_valid_block_count(struct f2fs_sb_info *sbi, 2172 struct inode *inode, blkcnt_t *count) 2173 { 2174 blkcnt_t diff = 0, release = 0; 2175 block_t avail_user_block_count; 2176 int ret; 2177 2178 ret = dquot_reserve_block(inode, *count); 2179 if (ret) 2180 return ret; 2181 2182 if (time_to_inject(sbi, FAULT_BLOCK)) { 2183 f2fs_show_injection_info(sbi, FAULT_BLOCK); 2184 release = *count; 2185 goto release_quota; 2186 } 2187 2188 /* 2189 * let's increase this in prior to actual block count change in order 2190 * for f2fs_sync_file to avoid data races when deciding checkpoint. 2191 */ 2192 percpu_counter_add(&sbi->alloc_valid_block_count, (*count)); 2193 2194 spin_lock(&sbi->stat_lock); 2195 sbi->total_valid_block_count += (block_t)(*count); 2196 avail_user_block_count = sbi->user_block_count - 2197 sbi->current_reserved_blocks; 2198 2199 if (!__allow_reserved_blocks(sbi, inode, true)) 2200 avail_user_block_count -= F2FS_OPTION(sbi).root_reserved_blocks; 2201 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) { 2202 if (avail_user_block_count > sbi->unusable_block_count) 2203 avail_user_block_count -= sbi->unusable_block_count; 2204 else 2205 avail_user_block_count = 0; 2206 } 2207 if (unlikely(sbi->total_valid_block_count > avail_user_block_count)) { 2208 diff = sbi->total_valid_block_count - avail_user_block_count; 2209 if (diff > *count) 2210 diff = *count; 2211 *count -= diff; 2212 release = diff; 2213 sbi->total_valid_block_count -= diff; 2214 if (!*count) { 2215 spin_unlock(&sbi->stat_lock); 2216 goto enospc; 2217 } 2218 } 2219 spin_unlock(&sbi->stat_lock); 2220 2221 if (unlikely(release)) { 2222 percpu_counter_sub(&sbi->alloc_valid_block_count, release); 2223 dquot_release_reservation_block(inode, release); 2224 } 2225 f2fs_i_blocks_write(inode, *count, true, true); 2226 return 0; 2227 2228 enospc: 2229 percpu_counter_sub(&sbi->alloc_valid_block_count, release); 2230 release_quota: 2231 dquot_release_reservation_block(inode, release); 2232 return -ENOSPC; 2233 } 2234 2235 __printf(2, 3) 2236 void f2fs_printk(struct f2fs_sb_info *sbi, const char *fmt, ...); 2237 2238 #define f2fs_err(sbi, fmt, ...) \ 2239 f2fs_printk(sbi, KERN_ERR fmt, ##__VA_ARGS__) 2240 #define f2fs_warn(sbi, fmt, ...) \ 2241 f2fs_printk(sbi, KERN_WARNING fmt, ##__VA_ARGS__) 2242 #define f2fs_notice(sbi, fmt, ...) \ 2243 f2fs_printk(sbi, KERN_NOTICE fmt, ##__VA_ARGS__) 2244 #define f2fs_info(sbi, fmt, ...) \ 2245 f2fs_printk(sbi, KERN_INFO fmt, ##__VA_ARGS__) 2246 #define f2fs_debug(sbi, fmt, ...) \ 2247 f2fs_printk(sbi, KERN_DEBUG fmt, ##__VA_ARGS__) 2248 2249 static inline void dec_valid_block_count(struct f2fs_sb_info *sbi, 2250 struct inode *inode, 2251 block_t count) 2252 { 2253 blkcnt_t sectors = count << F2FS_LOG_SECTORS_PER_BLOCK; 2254 2255 spin_lock(&sbi->stat_lock); 2256 f2fs_bug_on(sbi, sbi->total_valid_block_count < (block_t) count); 2257 sbi->total_valid_block_count -= (block_t)count; 2258 if (sbi->reserved_blocks && 2259 sbi->current_reserved_blocks < sbi->reserved_blocks) 2260 sbi->current_reserved_blocks = min(sbi->reserved_blocks, 2261 sbi->current_reserved_blocks + count); 2262 spin_unlock(&sbi->stat_lock); 2263 if (unlikely(inode->i_blocks < sectors)) { 2264 f2fs_warn(sbi, "Inconsistent i_blocks, ino:%lu, iblocks:%llu, sectors:%llu", 2265 inode->i_ino, 2266 (unsigned long long)inode->i_blocks, 2267 (unsigned long long)sectors); 2268 set_sbi_flag(sbi, SBI_NEED_FSCK); 2269 return; 2270 } 2271 f2fs_i_blocks_write(inode, count, false, true); 2272 } 2273 2274 static inline void inc_page_count(struct f2fs_sb_info *sbi, int count_type) 2275 { 2276 atomic_inc(&sbi->nr_pages[count_type]); 2277 2278 if (count_type == F2FS_DIRTY_DENTS || 2279 count_type == F2FS_DIRTY_NODES || 2280 count_type == F2FS_DIRTY_META || 2281 count_type == F2FS_DIRTY_QDATA || 2282 count_type == F2FS_DIRTY_IMETA) 2283 set_sbi_flag(sbi, SBI_IS_DIRTY); 2284 } 2285 2286 static inline void inode_inc_dirty_pages(struct inode *inode) 2287 { 2288 atomic_inc(&F2FS_I(inode)->dirty_pages); 2289 inc_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ? 2290 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA); 2291 if (IS_NOQUOTA(inode)) 2292 inc_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA); 2293 } 2294 2295 static inline void dec_page_count(struct f2fs_sb_info *sbi, int count_type) 2296 { 2297 atomic_dec(&sbi->nr_pages[count_type]); 2298 } 2299 2300 static inline void inode_dec_dirty_pages(struct inode *inode) 2301 { 2302 if (!S_ISDIR(inode->i_mode) && !S_ISREG(inode->i_mode) && 2303 !S_ISLNK(inode->i_mode)) 2304 return; 2305 2306 atomic_dec(&F2FS_I(inode)->dirty_pages); 2307 dec_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ? 2308 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA); 2309 if (IS_NOQUOTA(inode)) 2310 dec_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA); 2311 } 2312 2313 static inline s64 get_pages(struct f2fs_sb_info *sbi, int count_type) 2314 { 2315 return atomic_read(&sbi->nr_pages[count_type]); 2316 } 2317 2318 static inline int get_dirty_pages(struct inode *inode) 2319 { 2320 return atomic_read(&F2FS_I(inode)->dirty_pages); 2321 } 2322 2323 static inline int get_blocktype_secs(struct f2fs_sb_info *sbi, int block_type) 2324 { 2325 unsigned int pages_per_sec = sbi->segs_per_sec * sbi->blocks_per_seg; 2326 unsigned int segs = (get_pages(sbi, block_type) + pages_per_sec - 1) >> 2327 sbi->log_blocks_per_seg; 2328 2329 return segs / sbi->segs_per_sec; 2330 } 2331 2332 static inline block_t valid_user_blocks(struct f2fs_sb_info *sbi) 2333 { 2334 return sbi->total_valid_block_count; 2335 } 2336 2337 static inline block_t discard_blocks(struct f2fs_sb_info *sbi) 2338 { 2339 return sbi->discard_blks; 2340 } 2341 2342 static inline unsigned long __bitmap_size(struct f2fs_sb_info *sbi, int flag) 2343 { 2344 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi); 2345 2346 /* return NAT or SIT bitmap */ 2347 if (flag == NAT_BITMAP) 2348 return le32_to_cpu(ckpt->nat_ver_bitmap_bytesize); 2349 else if (flag == SIT_BITMAP) 2350 return le32_to_cpu(ckpt->sit_ver_bitmap_bytesize); 2351 2352 return 0; 2353 } 2354 2355 static inline block_t __cp_payload(struct f2fs_sb_info *sbi) 2356 { 2357 return le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_payload); 2358 } 2359 2360 static inline void *__bitmap_ptr(struct f2fs_sb_info *sbi, int flag) 2361 { 2362 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi); 2363 void *tmp_ptr = &ckpt->sit_nat_version_bitmap; 2364 int offset; 2365 2366 if (is_set_ckpt_flags(sbi, CP_LARGE_NAT_BITMAP_FLAG)) { 2367 offset = (flag == SIT_BITMAP) ? 2368 le32_to_cpu(ckpt->nat_ver_bitmap_bytesize) : 0; 2369 /* 2370 * if large_nat_bitmap feature is enabled, leave checksum 2371 * protection for all nat/sit bitmaps. 2372 */ 2373 return tmp_ptr + offset + sizeof(__le32); 2374 } 2375 2376 if (__cp_payload(sbi) > 0) { 2377 if (flag == NAT_BITMAP) 2378 return &ckpt->sit_nat_version_bitmap; 2379 else 2380 return (unsigned char *)ckpt + F2FS_BLKSIZE; 2381 } else { 2382 offset = (flag == NAT_BITMAP) ? 2383 le32_to_cpu(ckpt->sit_ver_bitmap_bytesize) : 0; 2384 return tmp_ptr + offset; 2385 } 2386 } 2387 2388 static inline block_t __start_cp_addr(struct f2fs_sb_info *sbi) 2389 { 2390 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr); 2391 2392 if (sbi->cur_cp_pack == 2) 2393 start_addr += sbi->blocks_per_seg; 2394 return start_addr; 2395 } 2396 2397 static inline block_t __start_cp_next_addr(struct f2fs_sb_info *sbi) 2398 { 2399 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr); 2400 2401 if (sbi->cur_cp_pack == 1) 2402 start_addr += sbi->blocks_per_seg; 2403 return start_addr; 2404 } 2405 2406 static inline void __set_cp_next_pack(struct f2fs_sb_info *sbi) 2407 { 2408 sbi->cur_cp_pack = (sbi->cur_cp_pack == 1) ? 2 : 1; 2409 } 2410 2411 static inline block_t __start_sum_addr(struct f2fs_sb_info *sbi) 2412 { 2413 return le32_to_cpu(F2FS_CKPT(sbi)->cp_pack_start_sum); 2414 } 2415 2416 static inline int inc_valid_node_count(struct f2fs_sb_info *sbi, 2417 struct inode *inode, bool is_inode) 2418 { 2419 block_t valid_block_count; 2420 unsigned int valid_node_count, user_block_count; 2421 int err; 2422 2423 if (is_inode) { 2424 if (inode) { 2425 err = dquot_alloc_inode(inode); 2426 if (err) 2427 return err; 2428 } 2429 } else { 2430 err = dquot_reserve_block(inode, 1); 2431 if (err) 2432 return err; 2433 } 2434 2435 if (time_to_inject(sbi, FAULT_BLOCK)) { 2436 f2fs_show_injection_info(sbi, FAULT_BLOCK); 2437 goto enospc; 2438 } 2439 2440 spin_lock(&sbi->stat_lock); 2441 2442 valid_block_count = sbi->total_valid_block_count + 2443 sbi->current_reserved_blocks + 1; 2444 2445 if (!__allow_reserved_blocks(sbi, inode, false)) 2446 valid_block_count += F2FS_OPTION(sbi).root_reserved_blocks; 2447 user_block_count = sbi->user_block_count; 2448 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) 2449 user_block_count -= sbi->unusable_block_count; 2450 2451 if (unlikely(valid_block_count > user_block_count)) { 2452 spin_unlock(&sbi->stat_lock); 2453 goto enospc; 2454 } 2455 2456 valid_node_count = sbi->total_valid_node_count + 1; 2457 if (unlikely(valid_node_count > sbi->total_node_count)) { 2458 spin_unlock(&sbi->stat_lock); 2459 goto enospc; 2460 } 2461 2462 sbi->total_valid_node_count++; 2463 sbi->total_valid_block_count++; 2464 spin_unlock(&sbi->stat_lock); 2465 2466 if (inode) { 2467 if (is_inode) 2468 f2fs_mark_inode_dirty_sync(inode, true); 2469 else 2470 f2fs_i_blocks_write(inode, 1, true, true); 2471 } 2472 2473 percpu_counter_inc(&sbi->alloc_valid_block_count); 2474 return 0; 2475 2476 enospc: 2477 if (is_inode) { 2478 if (inode) 2479 dquot_free_inode(inode); 2480 } else { 2481 dquot_release_reservation_block(inode, 1); 2482 } 2483 return -ENOSPC; 2484 } 2485 2486 static inline void dec_valid_node_count(struct f2fs_sb_info *sbi, 2487 struct inode *inode, bool is_inode) 2488 { 2489 spin_lock(&sbi->stat_lock); 2490 2491 f2fs_bug_on(sbi, !sbi->total_valid_block_count); 2492 f2fs_bug_on(sbi, !sbi->total_valid_node_count); 2493 2494 sbi->total_valid_node_count--; 2495 sbi->total_valid_block_count--; 2496 if (sbi->reserved_blocks && 2497 sbi->current_reserved_blocks < sbi->reserved_blocks) 2498 sbi->current_reserved_blocks++; 2499 2500 spin_unlock(&sbi->stat_lock); 2501 2502 if (is_inode) { 2503 dquot_free_inode(inode); 2504 } else { 2505 if (unlikely(inode->i_blocks == 0)) { 2506 f2fs_warn(sbi, "dec_valid_node_count: inconsistent i_blocks, ino:%lu, iblocks:%llu", 2507 inode->i_ino, 2508 (unsigned long long)inode->i_blocks); 2509 set_sbi_flag(sbi, SBI_NEED_FSCK); 2510 return; 2511 } 2512 f2fs_i_blocks_write(inode, 1, false, true); 2513 } 2514 } 2515 2516 static inline unsigned int valid_node_count(struct f2fs_sb_info *sbi) 2517 { 2518 return sbi->total_valid_node_count; 2519 } 2520 2521 static inline void inc_valid_inode_count(struct f2fs_sb_info *sbi) 2522 { 2523 percpu_counter_inc(&sbi->total_valid_inode_count); 2524 } 2525 2526 static inline void dec_valid_inode_count(struct f2fs_sb_info *sbi) 2527 { 2528 percpu_counter_dec(&sbi->total_valid_inode_count); 2529 } 2530 2531 static inline s64 valid_inode_count(struct f2fs_sb_info *sbi) 2532 { 2533 return percpu_counter_sum_positive(&sbi->total_valid_inode_count); 2534 } 2535 2536 static inline struct page *f2fs_grab_cache_page(struct address_space *mapping, 2537 pgoff_t index, bool for_write) 2538 { 2539 struct page *page; 2540 2541 if (IS_ENABLED(CONFIG_F2FS_FAULT_INJECTION)) { 2542 if (!for_write) 2543 page = find_get_page_flags(mapping, index, 2544 FGP_LOCK | FGP_ACCESSED); 2545 else 2546 page = find_lock_page(mapping, index); 2547 if (page) 2548 return page; 2549 2550 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_ALLOC)) { 2551 f2fs_show_injection_info(F2FS_M_SB(mapping), 2552 FAULT_PAGE_ALLOC); 2553 return NULL; 2554 } 2555 } 2556 2557 if (!for_write) 2558 return grab_cache_page(mapping, index); 2559 return grab_cache_page_write_begin(mapping, index, AOP_FLAG_NOFS); 2560 } 2561 2562 static inline struct page *f2fs_pagecache_get_page( 2563 struct address_space *mapping, pgoff_t index, 2564 int fgp_flags, gfp_t gfp_mask) 2565 { 2566 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_GET)) { 2567 f2fs_show_injection_info(F2FS_M_SB(mapping), FAULT_PAGE_GET); 2568 return NULL; 2569 } 2570 2571 return pagecache_get_page(mapping, index, fgp_flags, gfp_mask); 2572 } 2573 2574 static inline void f2fs_copy_page(struct page *src, struct page *dst) 2575 { 2576 char *src_kaddr = kmap(src); 2577 char *dst_kaddr = kmap(dst); 2578 2579 memcpy(dst_kaddr, src_kaddr, PAGE_SIZE); 2580 kunmap(dst); 2581 kunmap(src); 2582 } 2583 2584 static inline void f2fs_put_page(struct page *page, int unlock) 2585 { 2586 if (!page) 2587 return; 2588 2589 if (unlock) { 2590 f2fs_bug_on(F2FS_P_SB(page), !PageLocked(page)); 2591 unlock_page(page); 2592 } 2593 put_page(page); 2594 } 2595 2596 static inline void f2fs_put_dnode(struct dnode_of_data *dn) 2597 { 2598 if (dn->node_page) 2599 f2fs_put_page(dn->node_page, 1); 2600 if (dn->inode_page && dn->node_page != dn->inode_page) 2601 f2fs_put_page(dn->inode_page, 0); 2602 dn->node_page = NULL; 2603 dn->inode_page = NULL; 2604 } 2605 2606 static inline struct kmem_cache *f2fs_kmem_cache_create(const char *name, 2607 size_t size) 2608 { 2609 return kmem_cache_create(name, size, 0, SLAB_RECLAIM_ACCOUNT, NULL); 2610 } 2611 2612 static inline void *f2fs_kmem_cache_alloc_nofail(struct kmem_cache *cachep, 2613 gfp_t flags) 2614 { 2615 void *entry; 2616 2617 entry = kmem_cache_alloc(cachep, flags); 2618 if (!entry) 2619 entry = kmem_cache_alloc(cachep, flags | __GFP_NOFAIL); 2620 return entry; 2621 } 2622 2623 static inline void *f2fs_kmem_cache_alloc(struct kmem_cache *cachep, 2624 gfp_t flags, bool nofail, struct f2fs_sb_info *sbi) 2625 { 2626 if (nofail) 2627 return f2fs_kmem_cache_alloc_nofail(cachep, flags); 2628 2629 if (time_to_inject(sbi, FAULT_SLAB_ALLOC)) { 2630 f2fs_show_injection_info(sbi, FAULT_SLAB_ALLOC); 2631 return NULL; 2632 } 2633 2634 return kmem_cache_alloc(cachep, flags); 2635 } 2636 2637 static inline bool is_inflight_io(struct f2fs_sb_info *sbi, int type) 2638 { 2639 if (get_pages(sbi, F2FS_RD_DATA) || get_pages(sbi, F2FS_RD_NODE) || 2640 get_pages(sbi, F2FS_RD_META) || get_pages(sbi, F2FS_WB_DATA) || 2641 get_pages(sbi, F2FS_WB_CP_DATA) || 2642 get_pages(sbi, F2FS_DIO_READ) || 2643 get_pages(sbi, F2FS_DIO_WRITE)) 2644 return true; 2645 2646 if (type != DISCARD_TIME && SM_I(sbi) && SM_I(sbi)->dcc_info && 2647 atomic_read(&SM_I(sbi)->dcc_info->queued_discard)) 2648 return true; 2649 2650 if (SM_I(sbi) && SM_I(sbi)->fcc_info && 2651 atomic_read(&SM_I(sbi)->fcc_info->queued_flush)) 2652 return true; 2653 return false; 2654 } 2655 2656 static inline bool is_idle(struct f2fs_sb_info *sbi, int type) 2657 { 2658 if (sbi->gc_mode == GC_URGENT_HIGH) 2659 return true; 2660 2661 if (is_inflight_io(sbi, type)) 2662 return false; 2663 2664 if (sbi->gc_mode == GC_URGENT_LOW && 2665 (type == DISCARD_TIME || type == GC_TIME)) 2666 return true; 2667 2668 return f2fs_time_over(sbi, type); 2669 } 2670 2671 static inline void f2fs_radix_tree_insert(struct radix_tree_root *root, 2672 unsigned long index, void *item) 2673 { 2674 while (radix_tree_insert(root, index, item)) 2675 cond_resched(); 2676 } 2677 2678 #define RAW_IS_INODE(p) ((p)->footer.nid == (p)->footer.ino) 2679 2680 static inline bool IS_INODE(struct page *page) 2681 { 2682 struct f2fs_node *p = F2FS_NODE(page); 2683 2684 return RAW_IS_INODE(p); 2685 } 2686 2687 static inline int offset_in_addr(struct f2fs_inode *i) 2688 { 2689 return (i->i_inline & F2FS_EXTRA_ATTR) ? 2690 (le16_to_cpu(i->i_extra_isize) / sizeof(__le32)) : 0; 2691 } 2692 2693 static inline __le32 *blkaddr_in_node(struct f2fs_node *node) 2694 { 2695 return RAW_IS_INODE(node) ? node->i.i_addr : node->dn.addr; 2696 } 2697 2698 static inline int f2fs_has_extra_attr(struct inode *inode); 2699 static inline block_t data_blkaddr(struct inode *inode, 2700 struct page *node_page, unsigned int offset) 2701 { 2702 struct f2fs_node *raw_node; 2703 __le32 *addr_array; 2704 int base = 0; 2705 bool is_inode = IS_INODE(node_page); 2706 2707 raw_node = F2FS_NODE(node_page); 2708 2709 if (is_inode) { 2710 if (!inode) 2711 /* from GC path only */ 2712 base = offset_in_addr(&raw_node->i); 2713 else if (f2fs_has_extra_attr(inode)) 2714 base = get_extra_isize(inode); 2715 } 2716 2717 addr_array = blkaddr_in_node(raw_node); 2718 return le32_to_cpu(addr_array[base + offset]); 2719 } 2720 2721 static inline block_t f2fs_data_blkaddr(struct dnode_of_data *dn) 2722 { 2723 return data_blkaddr(dn->inode, dn->node_page, dn->ofs_in_node); 2724 } 2725 2726 static inline int f2fs_test_bit(unsigned int nr, char *addr) 2727 { 2728 int mask; 2729 2730 addr += (nr >> 3); 2731 mask = 1 << (7 - (nr & 0x07)); 2732 return mask & *addr; 2733 } 2734 2735 static inline void f2fs_set_bit(unsigned int nr, char *addr) 2736 { 2737 int mask; 2738 2739 addr += (nr >> 3); 2740 mask = 1 << (7 - (nr & 0x07)); 2741 *addr |= mask; 2742 } 2743 2744 static inline void f2fs_clear_bit(unsigned int nr, char *addr) 2745 { 2746 int mask; 2747 2748 addr += (nr >> 3); 2749 mask = 1 << (7 - (nr & 0x07)); 2750 *addr &= ~mask; 2751 } 2752 2753 static inline int f2fs_test_and_set_bit(unsigned int nr, char *addr) 2754 { 2755 int mask; 2756 int ret; 2757 2758 addr += (nr >> 3); 2759 mask = 1 << (7 - (nr & 0x07)); 2760 ret = mask & *addr; 2761 *addr |= mask; 2762 return ret; 2763 } 2764 2765 static inline int f2fs_test_and_clear_bit(unsigned int nr, char *addr) 2766 { 2767 int mask; 2768 int ret; 2769 2770 addr += (nr >> 3); 2771 mask = 1 << (7 - (nr & 0x07)); 2772 ret = mask & *addr; 2773 *addr &= ~mask; 2774 return ret; 2775 } 2776 2777 static inline void f2fs_change_bit(unsigned int nr, char *addr) 2778 { 2779 int mask; 2780 2781 addr += (nr >> 3); 2782 mask = 1 << (7 - (nr & 0x07)); 2783 *addr ^= mask; 2784 } 2785 2786 /* 2787 * On-disk inode flags (f2fs_inode::i_flags) 2788 */ 2789 #define F2FS_COMPR_FL 0x00000004 /* Compress file */ 2790 #define F2FS_SYNC_FL 0x00000008 /* Synchronous updates */ 2791 #define F2FS_IMMUTABLE_FL 0x00000010 /* Immutable file */ 2792 #define F2FS_APPEND_FL 0x00000020 /* writes to file may only append */ 2793 #define F2FS_NODUMP_FL 0x00000040 /* do not dump file */ 2794 #define F2FS_NOATIME_FL 0x00000080 /* do not update atime */ 2795 #define F2FS_NOCOMP_FL 0x00000400 /* Don't compress */ 2796 #define F2FS_INDEX_FL 0x00001000 /* hash-indexed directory */ 2797 #define F2FS_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */ 2798 #define F2FS_PROJINHERIT_FL 0x20000000 /* Create with parents projid */ 2799 #define F2FS_CASEFOLD_FL 0x40000000 /* Casefolded file */ 2800 2801 /* Flags that should be inherited by new inodes from their parent. */ 2802 #define F2FS_FL_INHERITED (F2FS_SYNC_FL | F2FS_NODUMP_FL | F2FS_NOATIME_FL | \ 2803 F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \ 2804 F2FS_CASEFOLD_FL | F2FS_COMPR_FL | F2FS_NOCOMP_FL) 2805 2806 /* Flags that are appropriate for regular files (all but dir-specific ones). */ 2807 #define F2FS_REG_FLMASK (~(F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \ 2808 F2FS_CASEFOLD_FL)) 2809 2810 /* Flags that are appropriate for non-directories/regular files. */ 2811 #define F2FS_OTHER_FLMASK (F2FS_NODUMP_FL | F2FS_NOATIME_FL) 2812 2813 static inline __u32 f2fs_mask_flags(umode_t mode, __u32 flags) 2814 { 2815 if (S_ISDIR(mode)) 2816 return flags; 2817 else if (S_ISREG(mode)) 2818 return flags & F2FS_REG_FLMASK; 2819 else 2820 return flags & F2FS_OTHER_FLMASK; 2821 } 2822 2823 static inline void __mark_inode_dirty_flag(struct inode *inode, 2824 int flag, bool set) 2825 { 2826 switch (flag) { 2827 case FI_INLINE_XATTR: 2828 case FI_INLINE_DATA: 2829 case FI_INLINE_DENTRY: 2830 case FI_NEW_INODE: 2831 if (set) 2832 return; 2833 fallthrough; 2834 case FI_DATA_EXIST: 2835 case FI_INLINE_DOTS: 2836 case FI_PIN_FILE: 2837 case FI_COMPRESS_RELEASED: 2838 f2fs_mark_inode_dirty_sync(inode, true); 2839 } 2840 } 2841 2842 static inline void set_inode_flag(struct inode *inode, int flag) 2843 { 2844 set_bit(flag, F2FS_I(inode)->flags); 2845 __mark_inode_dirty_flag(inode, flag, true); 2846 } 2847 2848 static inline int is_inode_flag_set(struct inode *inode, int flag) 2849 { 2850 return test_bit(flag, F2FS_I(inode)->flags); 2851 } 2852 2853 static inline void clear_inode_flag(struct inode *inode, int flag) 2854 { 2855 clear_bit(flag, F2FS_I(inode)->flags); 2856 __mark_inode_dirty_flag(inode, flag, false); 2857 } 2858 2859 static inline bool f2fs_verity_in_progress(struct inode *inode) 2860 { 2861 return IS_ENABLED(CONFIG_FS_VERITY) && 2862 is_inode_flag_set(inode, FI_VERITY_IN_PROGRESS); 2863 } 2864 2865 static inline void set_acl_inode(struct inode *inode, umode_t mode) 2866 { 2867 F2FS_I(inode)->i_acl_mode = mode; 2868 set_inode_flag(inode, FI_ACL_MODE); 2869 f2fs_mark_inode_dirty_sync(inode, false); 2870 } 2871 2872 static inline void f2fs_i_links_write(struct inode *inode, bool inc) 2873 { 2874 if (inc) 2875 inc_nlink(inode); 2876 else 2877 drop_nlink(inode); 2878 f2fs_mark_inode_dirty_sync(inode, true); 2879 } 2880 2881 static inline void f2fs_i_blocks_write(struct inode *inode, 2882 block_t diff, bool add, bool claim) 2883 { 2884 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE); 2885 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER); 2886 2887 /* add = 1, claim = 1 should be dquot_reserve_block in pair */ 2888 if (add) { 2889 if (claim) 2890 dquot_claim_block(inode, diff); 2891 else 2892 dquot_alloc_block_nofail(inode, diff); 2893 } else { 2894 dquot_free_block(inode, diff); 2895 } 2896 2897 f2fs_mark_inode_dirty_sync(inode, true); 2898 if (clean || recover) 2899 set_inode_flag(inode, FI_AUTO_RECOVER); 2900 } 2901 2902 static inline void f2fs_i_size_write(struct inode *inode, loff_t i_size) 2903 { 2904 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE); 2905 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER); 2906 2907 if (i_size_read(inode) == i_size) 2908 return; 2909 2910 i_size_write(inode, i_size); 2911 f2fs_mark_inode_dirty_sync(inode, true); 2912 if (clean || recover) 2913 set_inode_flag(inode, FI_AUTO_RECOVER); 2914 } 2915 2916 static inline void f2fs_i_depth_write(struct inode *inode, unsigned int depth) 2917 { 2918 F2FS_I(inode)->i_current_depth = depth; 2919 f2fs_mark_inode_dirty_sync(inode, true); 2920 } 2921 2922 static inline void f2fs_i_gc_failures_write(struct inode *inode, 2923 unsigned int count) 2924 { 2925 F2FS_I(inode)->i_gc_failures[GC_FAILURE_PIN] = count; 2926 f2fs_mark_inode_dirty_sync(inode, true); 2927 } 2928 2929 static inline void f2fs_i_xnid_write(struct inode *inode, nid_t xnid) 2930 { 2931 F2FS_I(inode)->i_xattr_nid = xnid; 2932 f2fs_mark_inode_dirty_sync(inode, true); 2933 } 2934 2935 static inline void f2fs_i_pino_write(struct inode *inode, nid_t pino) 2936 { 2937 F2FS_I(inode)->i_pino = pino; 2938 f2fs_mark_inode_dirty_sync(inode, true); 2939 } 2940 2941 static inline void get_inline_info(struct inode *inode, struct f2fs_inode *ri) 2942 { 2943 struct f2fs_inode_info *fi = F2FS_I(inode); 2944 2945 if (ri->i_inline & F2FS_INLINE_XATTR) 2946 set_bit(FI_INLINE_XATTR, fi->flags); 2947 if (ri->i_inline & F2FS_INLINE_DATA) 2948 set_bit(FI_INLINE_DATA, fi->flags); 2949 if (ri->i_inline & F2FS_INLINE_DENTRY) 2950 set_bit(FI_INLINE_DENTRY, fi->flags); 2951 if (ri->i_inline & F2FS_DATA_EXIST) 2952 set_bit(FI_DATA_EXIST, fi->flags); 2953 if (ri->i_inline & F2FS_INLINE_DOTS) 2954 set_bit(FI_INLINE_DOTS, fi->flags); 2955 if (ri->i_inline & F2FS_EXTRA_ATTR) 2956 set_bit(FI_EXTRA_ATTR, fi->flags); 2957 if (ri->i_inline & F2FS_PIN_FILE) 2958 set_bit(FI_PIN_FILE, fi->flags); 2959 if (ri->i_inline & F2FS_COMPRESS_RELEASED) 2960 set_bit(FI_COMPRESS_RELEASED, fi->flags); 2961 } 2962 2963 static inline void set_raw_inline(struct inode *inode, struct f2fs_inode *ri) 2964 { 2965 ri->i_inline = 0; 2966 2967 if (is_inode_flag_set(inode, FI_INLINE_XATTR)) 2968 ri->i_inline |= F2FS_INLINE_XATTR; 2969 if (is_inode_flag_set(inode, FI_INLINE_DATA)) 2970 ri->i_inline |= F2FS_INLINE_DATA; 2971 if (is_inode_flag_set(inode, FI_INLINE_DENTRY)) 2972 ri->i_inline |= F2FS_INLINE_DENTRY; 2973 if (is_inode_flag_set(inode, FI_DATA_EXIST)) 2974 ri->i_inline |= F2FS_DATA_EXIST; 2975 if (is_inode_flag_set(inode, FI_INLINE_DOTS)) 2976 ri->i_inline |= F2FS_INLINE_DOTS; 2977 if (is_inode_flag_set(inode, FI_EXTRA_ATTR)) 2978 ri->i_inline |= F2FS_EXTRA_ATTR; 2979 if (is_inode_flag_set(inode, FI_PIN_FILE)) 2980 ri->i_inline |= F2FS_PIN_FILE; 2981 if (is_inode_flag_set(inode, FI_COMPRESS_RELEASED)) 2982 ri->i_inline |= F2FS_COMPRESS_RELEASED; 2983 } 2984 2985 static inline int f2fs_has_extra_attr(struct inode *inode) 2986 { 2987 return is_inode_flag_set(inode, FI_EXTRA_ATTR); 2988 } 2989 2990 static inline int f2fs_has_inline_xattr(struct inode *inode) 2991 { 2992 return is_inode_flag_set(inode, FI_INLINE_XATTR); 2993 } 2994 2995 static inline int f2fs_compressed_file(struct inode *inode) 2996 { 2997 return S_ISREG(inode->i_mode) && 2998 is_inode_flag_set(inode, FI_COMPRESSED_FILE); 2999 } 3000 3001 static inline bool f2fs_need_compress_data(struct inode *inode) 3002 { 3003 int compress_mode = F2FS_OPTION(F2FS_I_SB(inode)).compress_mode; 3004 3005 if (!f2fs_compressed_file(inode)) 3006 return false; 3007 3008 if (compress_mode == COMPR_MODE_FS) 3009 return true; 3010 else if (compress_mode == COMPR_MODE_USER && 3011 is_inode_flag_set(inode, FI_ENABLE_COMPRESS)) 3012 return true; 3013 3014 return false; 3015 } 3016 3017 static inline unsigned int addrs_per_inode(struct inode *inode) 3018 { 3019 unsigned int addrs = CUR_ADDRS_PER_INODE(inode) - 3020 get_inline_xattr_addrs(inode); 3021 3022 if (!f2fs_compressed_file(inode)) 3023 return addrs; 3024 return ALIGN_DOWN(addrs, F2FS_I(inode)->i_cluster_size); 3025 } 3026 3027 static inline unsigned int addrs_per_block(struct inode *inode) 3028 { 3029 if (!f2fs_compressed_file(inode)) 3030 return DEF_ADDRS_PER_BLOCK; 3031 return ALIGN_DOWN(DEF_ADDRS_PER_BLOCK, F2FS_I(inode)->i_cluster_size); 3032 } 3033 3034 static inline void *inline_xattr_addr(struct inode *inode, struct page *page) 3035 { 3036 struct f2fs_inode *ri = F2FS_INODE(page); 3037 3038 return (void *)&(ri->i_addr[DEF_ADDRS_PER_INODE - 3039 get_inline_xattr_addrs(inode)]); 3040 } 3041 3042 static inline int inline_xattr_size(struct inode *inode) 3043 { 3044 if (f2fs_has_inline_xattr(inode)) 3045 return get_inline_xattr_addrs(inode) * sizeof(__le32); 3046 return 0; 3047 } 3048 3049 static inline int f2fs_has_inline_data(struct inode *inode) 3050 { 3051 return is_inode_flag_set(inode, FI_INLINE_DATA); 3052 } 3053 3054 static inline int f2fs_exist_data(struct inode *inode) 3055 { 3056 return is_inode_flag_set(inode, FI_DATA_EXIST); 3057 } 3058 3059 static inline int f2fs_has_inline_dots(struct inode *inode) 3060 { 3061 return is_inode_flag_set(inode, FI_INLINE_DOTS); 3062 } 3063 3064 static inline int f2fs_is_mmap_file(struct inode *inode) 3065 { 3066 return is_inode_flag_set(inode, FI_MMAP_FILE); 3067 } 3068 3069 static inline bool f2fs_is_pinned_file(struct inode *inode) 3070 { 3071 return is_inode_flag_set(inode, FI_PIN_FILE); 3072 } 3073 3074 static inline bool f2fs_is_atomic_file(struct inode *inode) 3075 { 3076 return is_inode_flag_set(inode, FI_ATOMIC_FILE); 3077 } 3078 3079 static inline bool f2fs_is_commit_atomic_write(struct inode *inode) 3080 { 3081 return is_inode_flag_set(inode, FI_ATOMIC_COMMIT); 3082 } 3083 3084 static inline bool f2fs_is_volatile_file(struct inode *inode) 3085 { 3086 return is_inode_flag_set(inode, FI_VOLATILE_FILE); 3087 } 3088 3089 static inline bool f2fs_is_first_block_written(struct inode *inode) 3090 { 3091 return is_inode_flag_set(inode, FI_FIRST_BLOCK_WRITTEN); 3092 } 3093 3094 static inline bool f2fs_is_drop_cache(struct inode *inode) 3095 { 3096 return is_inode_flag_set(inode, FI_DROP_CACHE); 3097 } 3098 3099 static inline void *inline_data_addr(struct inode *inode, struct page *page) 3100 { 3101 struct f2fs_inode *ri = F2FS_INODE(page); 3102 int extra_size = get_extra_isize(inode); 3103 3104 return (void *)&(ri->i_addr[extra_size + DEF_INLINE_RESERVED_SIZE]); 3105 } 3106 3107 static inline int f2fs_has_inline_dentry(struct inode *inode) 3108 { 3109 return is_inode_flag_set(inode, FI_INLINE_DENTRY); 3110 } 3111 3112 static inline int is_file(struct inode *inode, int type) 3113 { 3114 return F2FS_I(inode)->i_advise & type; 3115 } 3116 3117 static inline void set_file(struct inode *inode, int type) 3118 { 3119 F2FS_I(inode)->i_advise |= type; 3120 f2fs_mark_inode_dirty_sync(inode, true); 3121 } 3122 3123 static inline void clear_file(struct inode *inode, int type) 3124 { 3125 F2FS_I(inode)->i_advise &= ~type; 3126 f2fs_mark_inode_dirty_sync(inode, true); 3127 } 3128 3129 static inline bool f2fs_is_time_consistent(struct inode *inode) 3130 { 3131 if (!timespec64_equal(F2FS_I(inode)->i_disk_time, &inode->i_atime)) 3132 return false; 3133 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 1, &inode->i_ctime)) 3134 return false; 3135 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 2, &inode->i_mtime)) 3136 return false; 3137 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 3, 3138 &F2FS_I(inode)->i_crtime)) 3139 return false; 3140 return true; 3141 } 3142 3143 static inline bool f2fs_skip_inode_update(struct inode *inode, int dsync) 3144 { 3145 bool ret; 3146 3147 if (dsync) { 3148 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 3149 3150 spin_lock(&sbi->inode_lock[DIRTY_META]); 3151 ret = list_empty(&F2FS_I(inode)->gdirty_list); 3152 spin_unlock(&sbi->inode_lock[DIRTY_META]); 3153 return ret; 3154 } 3155 if (!is_inode_flag_set(inode, FI_AUTO_RECOVER) || 3156 file_keep_isize(inode) || 3157 i_size_read(inode) & ~PAGE_MASK) 3158 return false; 3159 3160 if (!f2fs_is_time_consistent(inode)) 3161 return false; 3162 3163 spin_lock(&F2FS_I(inode)->i_size_lock); 3164 ret = F2FS_I(inode)->last_disk_size == i_size_read(inode); 3165 spin_unlock(&F2FS_I(inode)->i_size_lock); 3166 3167 return ret; 3168 } 3169 3170 static inline bool f2fs_readonly(struct super_block *sb) 3171 { 3172 return sb_rdonly(sb); 3173 } 3174 3175 static inline bool f2fs_cp_error(struct f2fs_sb_info *sbi) 3176 { 3177 return is_set_ckpt_flags(sbi, CP_ERROR_FLAG); 3178 } 3179 3180 static inline bool is_dot_dotdot(const u8 *name, size_t len) 3181 { 3182 if (len == 1 && name[0] == '.') 3183 return true; 3184 3185 if (len == 2 && name[0] == '.' && name[1] == '.') 3186 return true; 3187 3188 return false; 3189 } 3190 3191 static inline void *f2fs_kmalloc(struct f2fs_sb_info *sbi, 3192 size_t size, gfp_t flags) 3193 { 3194 if (time_to_inject(sbi, FAULT_KMALLOC)) { 3195 f2fs_show_injection_info(sbi, FAULT_KMALLOC); 3196 return NULL; 3197 } 3198 3199 return kmalloc(size, flags); 3200 } 3201 3202 static inline void *f2fs_kzalloc(struct f2fs_sb_info *sbi, 3203 size_t size, gfp_t flags) 3204 { 3205 return f2fs_kmalloc(sbi, size, flags | __GFP_ZERO); 3206 } 3207 3208 static inline void *f2fs_kvmalloc(struct f2fs_sb_info *sbi, 3209 size_t size, gfp_t flags) 3210 { 3211 if (time_to_inject(sbi, FAULT_KVMALLOC)) { 3212 f2fs_show_injection_info(sbi, FAULT_KVMALLOC); 3213 return NULL; 3214 } 3215 3216 return kvmalloc(size, flags); 3217 } 3218 3219 static inline void *f2fs_kvzalloc(struct f2fs_sb_info *sbi, 3220 size_t size, gfp_t flags) 3221 { 3222 return f2fs_kvmalloc(sbi, size, flags | __GFP_ZERO); 3223 } 3224 3225 static inline int get_extra_isize(struct inode *inode) 3226 { 3227 return F2FS_I(inode)->i_extra_isize / sizeof(__le32); 3228 } 3229 3230 static inline int get_inline_xattr_addrs(struct inode *inode) 3231 { 3232 return F2FS_I(inode)->i_inline_xattr_size; 3233 } 3234 3235 #define f2fs_get_inode_mode(i) \ 3236 ((is_inode_flag_set(i, FI_ACL_MODE)) ? \ 3237 (F2FS_I(i)->i_acl_mode) : ((i)->i_mode)) 3238 3239 #define F2FS_TOTAL_EXTRA_ATTR_SIZE \ 3240 (offsetof(struct f2fs_inode, i_extra_end) - \ 3241 offsetof(struct f2fs_inode, i_extra_isize)) \ 3242 3243 #define F2FS_OLD_ATTRIBUTE_SIZE (offsetof(struct f2fs_inode, i_addr)) 3244 #define F2FS_FITS_IN_INODE(f2fs_inode, extra_isize, field) \ 3245 ((offsetof(typeof(*(f2fs_inode)), field) + \ 3246 sizeof((f2fs_inode)->field)) \ 3247 <= (F2FS_OLD_ATTRIBUTE_SIZE + (extra_isize))) \ 3248 3249 #define __is_large_section(sbi) ((sbi)->segs_per_sec > 1) 3250 3251 #define __is_meta_io(fio) (PAGE_TYPE_OF_BIO((fio)->type) == META) 3252 3253 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi, 3254 block_t blkaddr, int type); 3255 static inline void verify_blkaddr(struct f2fs_sb_info *sbi, 3256 block_t blkaddr, int type) 3257 { 3258 if (!f2fs_is_valid_blkaddr(sbi, blkaddr, type)) { 3259 f2fs_err(sbi, "invalid blkaddr: %u, type: %d, run fsck to fix.", 3260 blkaddr, type); 3261 f2fs_bug_on(sbi, 1); 3262 } 3263 } 3264 3265 static inline bool __is_valid_data_blkaddr(block_t blkaddr) 3266 { 3267 if (blkaddr == NEW_ADDR || blkaddr == NULL_ADDR || 3268 blkaddr == COMPRESS_ADDR) 3269 return false; 3270 return true; 3271 } 3272 3273 /* 3274 * file.c 3275 */ 3276 int f2fs_sync_file(struct file *file, loff_t start, loff_t end, int datasync); 3277 void f2fs_truncate_data_blocks(struct dnode_of_data *dn); 3278 int f2fs_do_truncate_blocks(struct inode *inode, u64 from, bool lock); 3279 int f2fs_truncate_blocks(struct inode *inode, u64 from, bool lock); 3280 int f2fs_truncate(struct inode *inode); 3281 int f2fs_getattr(struct user_namespace *mnt_userns, const struct path *path, 3282 struct kstat *stat, u32 request_mask, unsigned int flags); 3283 int f2fs_setattr(struct user_namespace *mnt_userns, struct dentry *dentry, 3284 struct iattr *attr); 3285 int f2fs_truncate_hole(struct inode *inode, pgoff_t pg_start, pgoff_t pg_end); 3286 void f2fs_truncate_data_blocks_range(struct dnode_of_data *dn, int count); 3287 int f2fs_precache_extents(struct inode *inode); 3288 int f2fs_fileattr_get(struct dentry *dentry, struct fileattr *fa); 3289 int f2fs_fileattr_set(struct user_namespace *mnt_userns, 3290 struct dentry *dentry, struct fileattr *fa); 3291 long f2fs_ioctl(struct file *filp, unsigned int cmd, unsigned long arg); 3292 long f2fs_compat_ioctl(struct file *file, unsigned int cmd, unsigned long arg); 3293 int f2fs_transfer_project_quota(struct inode *inode, kprojid_t kprojid); 3294 int f2fs_pin_file_control(struct inode *inode, bool inc); 3295 3296 /* 3297 * inode.c 3298 */ 3299 void f2fs_set_inode_flags(struct inode *inode); 3300 bool f2fs_inode_chksum_verify(struct f2fs_sb_info *sbi, struct page *page); 3301 void f2fs_inode_chksum_set(struct f2fs_sb_info *sbi, struct page *page); 3302 struct inode *f2fs_iget(struct super_block *sb, unsigned long ino); 3303 struct inode *f2fs_iget_retry(struct super_block *sb, unsigned long ino); 3304 int f2fs_try_to_free_nats(struct f2fs_sb_info *sbi, int nr_shrink); 3305 void f2fs_update_inode(struct inode *inode, struct page *node_page); 3306 void f2fs_update_inode_page(struct inode *inode); 3307 int f2fs_write_inode(struct inode *inode, struct writeback_control *wbc); 3308 void f2fs_evict_inode(struct inode *inode); 3309 void f2fs_handle_failed_inode(struct inode *inode); 3310 3311 /* 3312 * namei.c 3313 */ 3314 int f2fs_update_extension_list(struct f2fs_sb_info *sbi, const char *name, 3315 bool hot, bool set); 3316 struct dentry *f2fs_get_parent(struct dentry *child); 3317 3318 /* 3319 * dir.c 3320 */ 3321 unsigned char f2fs_get_de_type(struct f2fs_dir_entry *de); 3322 int f2fs_init_casefolded_name(const struct inode *dir, 3323 struct f2fs_filename *fname); 3324 int f2fs_setup_filename(struct inode *dir, const struct qstr *iname, 3325 int lookup, struct f2fs_filename *fname); 3326 int f2fs_prepare_lookup(struct inode *dir, struct dentry *dentry, 3327 struct f2fs_filename *fname); 3328 void f2fs_free_filename(struct f2fs_filename *fname); 3329 struct f2fs_dir_entry *f2fs_find_target_dentry(const struct f2fs_dentry_ptr *d, 3330 const struct f2fs_filename *fname, int *max_slots); 3331 int f2fs_fill_dentries(struct dir_context *ctx, struct f2fs_dentry_ptr *d, 3332 unsigned int start_pos, struct fscrypt_str *fstr); 3333 void f2fs_do_make_empty_dir(struct inode *inode, struct inode *parent, 3334 struct f2fs_dentry_ptr *d); 3335 struct page *f2fs_init_inode_metadata(struct inode *inode, struct inode *dir, 3336 const struct f2fs_filename *fname, struct page *dpage); 3337 void f2fs_update_parent_metadata(struct inode *dir, struct inode *inode, 3338 unsigned int current_depth); 3339 int f2fs_room_for_filename(const void *bitmap, int slots, int max_slots); 3340 void f2fs_drop_nlink(struct inode *dir, struct inode *inode); 3341 struct f2fs_dir_entry *__f2fs_find_entry(struct inode *dir, 3342 const struct f2fs_filename *fname, 3343 struct page **res_page); 3344 struct f2fs_dir_entry *f2fs_find_entry(struct inode *dir, 3345 const struct qstr *child, struct page **res_page); 3346 struct f2fs_dir_entry *f2fs_parent_dir(struct inode *dir, struct page **p); 3347 ino_t f2fs_inode_by_name(struct inode *dir, const struct qstr *qstr, 3348 struct page **page); 3349 void f2fs_set_link(struct inode *dir, struct f2fs_dir_entry *de, 3350 struct page *page, struct inode *inode); 3351 bool f2fs_has_enough_room(struct inode *dir, struct page *ipage, 3352 const struct f2fs_filename *fname); 3353 void f2fs_update_dentry(nid_t ino, umode_t mode, struct f2fs_dentry_ptr *d, 3354 const struct fscrypt_str *name, f2fs_hash_t name_hash, 3355 unsigned int bit_pos); 3356 int f2fs_add_regular_entry(struct inode *dir, const struct f2fs_filename *fname, 3357 struct inode *inode, nid_t ino, umode_t mode); 3358 int f2fs_add_dentry(struct inode *dir, const struct f2fs_filename *fname, 3359 struct inode *inode, nid_t ino, umode_t mode); 3360 int f2fs_do_add_link(struct inode *dir, const struct qstr *name, 3361 struct inode *inode, nid_t ino, umode_t mode); 3362 void f2fs_delete_entry(struct f2fs_dir_entry *dentry, struct page *page, 3363 struct inode *dir, struct inode *inode); 3364 int f2fs_do_tmpfile(struct inode *inode, struct inode *dir); 3365 bool f2fs_empty_dir(struct inode *dir); 3366 3367 static inline int f2fs_add_link(struct dentry *dentry, struct inode *inode) 3368 { 3369 if (fscrypt_is_nokey_name(dentry)) 3370 return -ENOKEY; 3371 return f2fs_do_add_link(d_inode(dentry->d_parent), &dentry->d_name, 3372 inode, inode->i_ino, inode->i_mode); 3373 } 3374 3375 /* 3376 * super.c 3377 */ 3378 int f2fs_inode_dirtied(struct inode *inode, bool sync); 3379 void f2fs_inode_synced(struct inode *inode); 3380 int f2fs_dquot_initialize(struct inode *inode); 3381 int f2fs_enable_quota_files(struct f2fs_sb_info *sbi, bool rdonly); 3382 int f2fs_quota_sync(struct super_block *sb, int type); 3383 loff_t max_file_blocks(struct inode *inode); 3384 void f2fs_quota_off_umount(struct super_block *sb); 3385 int f2fs_commit_super(struct f2fs_sb_info *sbi, bool recover); 3386 int f2fs_sync_fs(struct super_block *sb, int sync); 3387 int f2fs_sanity_check_ckpt(struct f2fs_sb_info *sbi); 3388 3389 /* 3390 * hash.c 3391 */ 3392 void f2fs_hash_filename(const struct inode *dir, struct f2fs_filename *fname); 3393 3394 /* 3395 * node.c 3396 */ 3397 struct node_info; 3398 3399 int f2fs_check_nid_range(struct f2fs_sb_info *sbi, nid_t nid); 3400 bool f2fs_available_free_memory(struct f2fs_sb_info *sbi, int type); 3401 bool f2fs_in_warm_node_list(struct f2fs_sb_info *sbi, struct page *page); 3402 void f2fs_init_fsync_node_info(struct f2fs_sb_info *sbi); 3403 void f2fs_del_fsync_node_entry(struct f2fs_sb_info *sbi, struct page *page); 3404 void f2fs_reset_fsync_node_info(struct f2fs_sb_info *sbi); 3405 int f2fs_need_dentry_mark(struct f2fs_sb_info *sbi, nid_t nid); 3406 bool f2fs_is_checkpointed_node(struct f2fs_sb_info *sbi, nid_t nid); 3407 bool f2fs_need_inode_block_update(struct f2fs_sb_info *sbi, nid_t ino); 3408 int f2fs_get_node_info(struct f2fs_sb_info *sbi, nid_t nid, 3409 struct node_info *ni); 3410 pgoff_t f2fs_get_next_page_offset(struct dnode_of_data *dn, pgoff_t pgofs); 3411 int f2fs_get_dnode_of_data(struct dnode_of_data *dn, pgoff_t index, int mode); 3412 int f2fs_truncate_inode_blocks(struct inode *inode, pgoff_t from); 3413 int f2fs_truncate_xattr_node(struct inode *inode); 3414 int f2fs_wait_on_node_pages_writeback(struct f2fs_sb_info *sbi, 3415 unsigned int seq_id); 3416 bool f2fs_nat_bitmap_enabled(struct f2fs_sb_info *sbi); 3417 int f2fs_remove_inode_page(struct inode *inode); 3418 struct page *f2fs_new_inode_page(struct inode *inode); 3419 struct page *f2fs_new_node_page(struct dnode_of_data *dn, unsigned int ofs); 3420 void f2fs_ra_node_page(struct f2fs_sb_info *sbi, nid_t nid); 3421 struct page *f2fs_get_node_page(struct f2fs_sb_info *sbi, pgoff_t nid); 3422 struct page *f2fs_get_node_page_ra(struct page *parent, int start); 3423 int f2fs_move_node_page(struct page *node_page, int gc_type); 3424 void f2fs_flush_inline_data(struct f2fs_sb_info *sbi); 3425 int f2fs_fsync_node_pages(struct f2fs_sb_info *sbi, struct inode *inode, 3426 struct writeback_control *wbc, bool atomic, 3427 unsigned int *seq_id); 3428 int f2fs_sync_node_pages(struct f2fs_sb_info *sbi, 3429 struct writeback_control *wbc, 3430 bool do_balance, enum iostat_type io_type); 3431 int f2fs_build_free_nids(struct f2fs_sb_info *sbi, bool sync, bool mount); 3432 bool f2fs_alloc_nid(struct f2fs_sb_info *sbi, nid_t *nid); 3433 void f2fs_alloc_nid_done(struct f2fs_sb_info *sbi, nid_t nid); 3434 void f2fs_alloc_nid_failed(struct f2fs_sb_info *sbi, nid_t nid); 3435 int f2fs_try_to_free_nids(struct f2fs_sb_info *sbi, int nr_shrink); 3436 int f2fs_recover_inline_xattr(struct inode *inode, struct page *page); 3437 int f2fs_recover_xattr_data(struct inode *inode, struct page *page); 3438 int f2fs_recover_inode_page(struct f2fs_sb_info *sbi, struct page *page); 3439 int f2fs_restore_node_summary(struct f2fs_sb_info *sbi, 3440 unsigned int segno, struct f2fs_summary_block *sum); 3441 void f2fs_enable_nat_bits(struct f2fs_sb_info *sbi); 3442 int f2fs_flush_nat_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3443 int f2fs_build_node_manager(struct f2fs_sb_info *sbi); 3444 void f2fs_destroy_node_manager(struct f2fs_sb_info *sbi); 3445 int __init f2fs_create_node_manager_caches(void); 3446 void f2fs_destroy_node_manager_caches(void); 3447 3448 /* 3449 * segment.c 3450 */ 3451 bool f2fs_need_SSR(struct f2fs_sb_info *sbi); 3452 void f2fs_register_inmem_page(struct inode *inode, struct page *page); 3453 void f2fs_drop_inmem_pages_all(struct f2fs_sb_info *sbi, bool gc_failure); 3454 void f2fs_drop_inmem_pages(struct inode *inode); 3455 void f2fs_drop_inmem_page(struct inode *inode, struct page *page); 3456 int f2fs_commit_inmem_pages(struct inode *inode); 3457 void f2fs_balance_fs(struct f2fs_sb_info *sbi, bool need); 3458 void f2fs_balance_fs_bg(struct f2fs_sb_info *sbi, bool from_bg); 3459 int f2fs_issue_flush(struct f2fs_sb_info *sbi, nid_t ino); 3460 int f2fs_create_flush_cmd_control(struct f2fs_sb_info *sbi); 3461 int f2fs_flush_device_cache(struct f2fs_sb_info *sbi); 3462 void f2fs_destroy_flush_cmd_control(struct f2fs_sb_info *sbi, bool free); 3463 void f2fs_invalidate_blocks(struct f2fs_sb_info *sbi, block_t addr); 3464 bool f2fs_is_checkpointed_data(struct f2fs_sb_info *sbi, block_t blkaddr); 3465 int f2fs_start_discard_thread(struct f2fs_sb_info *sbi); 3466 void f2fs_drop_discard_cmd(struct f2fs_sb_info *sbi); 3467 void f2fs_stop_discard_thread(struct f2fs_sb_info *sbi); 3468 bool f2fs_issue_discard_timeout(struct f2fs_sb_info *sbi); 3469 void f2fs_clear_prefree_segments(struct f2fs_sb_info *sbi, 3470 struct cp_control *cpc); 3471 void f2fs_dirty_to_prefree(struct f2fs_sb_info *sbi); 3472 block_t f2fs_get_unusable_blocks(struct f2fs_sb_info *sbi); 3473 int f2fs_disable_cp_again(struct f2fs_sb_info *sbi, block_t unusable); 3474 void f2fs_release_discard_addrs(struct f2fs_sb_info *sbi); 3475 int f2fs_npages_for_summary_flush(struct f2fs_sb_info *sbi, bool for_ra); 3476 bool f2fs_segment_has_free_slot(struct f2fs_sb_info *sbi, int segno); 3477 void f2fs_init_inmem_curseg(struct f2fs_sb_info *sbi); 3478 void f2fs_save_inmem_curseg(struct f2fs_sb_info *sbi); 3479 void f2fs_restore_inmem_curseg(struct f2fs_sb_info *sbi); 3480 void f2fs_get_new_segment(struct f2fs_sb_info *sbi, 3481 unsigned int *newseg, bool new_sec, int dir); 3482 void f2fs_allocate_segment_for_resize(struct f2fs_sb_info *sbi, int type, 3483 unsigned int start, unsigned int end); 3484 void f2fs_allocate_new_section(struct f2fs_sb_info *sbi, int type, bool force); 3485 void f2fs_allocate_new_segments(struct f2fs_sb_info *sbi); 3486 int f2fs_trim_fs(struct f2fs_sb_info *sbi, struct fstrim_range *range); 3487 bool f2fs_exist_trim_candidates(struct f2fs_sb_info *sbi, 3488 struct cp_control *cpc); 3489 struct page *f2fs_get_sum_page(struct f2fs_sb_info *sbi, unsigned int segno); 3490 void f2fs_update_meta_page(struct f2fs_sb_info *sbi, void *src, 3491 block_t blk_addr); 3492 void f2fs_do_write_meta_page(struct f2fs_sb_info *sbi, struct page *page, 3493 enum iostat_type io_type); 3494 void f2fs_do_write_node_page(unsigned int nid, struct f2fs_io_info *fio); 3495 void f2fs_outplace_write_data(struct dnode_of_data *dn, 3496 struct f2fs_io_info *fio); 3497 int f2fs_inplace_write_data(struct f2fs_io_info *fio); 3498 void f2fs_do_replace_block(struct f2fs_sb_info *sbi, struct f2fs_summary *sum, 3499 block_t old_blkaddr, block_t new_blkaddr, 3500 bool recover_curseg, bool recover_newaddr, 3501 bool from_gc); 3502 void f2fs_replace_block(struct f2fs_sb_info *sbi, struct dnode_of_data *dn, 3503 block_t old_addr, block_t new_addr, 3504 unsigned char version, bool recover_curseg, 3505 bool recover_newaddr); 3506 void f2fs_allocate_data_block(struct f2fs_sb_info *sbi, struct page *page, 3507 block_t old_blkaddr, block_t *new_blkaddr, 3508 struct f2fs_summary *sum, int type, 3509 struct f2fs_io_info *fio); 3510 void f2fs_update_device_state(struct f2fs_sb_info *sbi, nid_t ino, 3511 block_t blkaddr, unsigned int blkcnt); 3512 void f2fs_wait_on_page_writeback(struct page *page, 3513 enum page_type type, bool ordered, bool locked); 3514 void f2fs_wait_on_block_writeback(struct inode *inode, block_t blkaddr); 3515 void f2fs_wait_on_block_writeback_range(struct inode *inode, block_t blkaddr, 3516 block_t len); 3517 void f2fs_write_data_summaries(struct f2fs_sb_info *sbi, block_t start_blk); 3518 void f2fs_write_node_summaries(struct f2fs_sb_info *sbi, block_t start_blk); 3519 int f2fs_lookup_journal_in_cursum(struct f2fs_journal *journal, int type, 3520 unsigned int val, int alloc); 3521 void f2fs_flush_sit_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3522 int f2fs_fix_curseg_write_pointer(struct f2fs_sb_info *sbi); 3523 int f2fs_check_write_pointer(struct f2fs_sb_info *sbi); 3524 int f2fs_build_segment_manager(struct f2fs_sb_info *sbi); 3525 void f2fs_destroy_segment_manager(struct f2fs_sb_info *sbi); 3526 int __init f2fs_create_segment_manager_caches(void); 3527 void f2fs_destroy_segment_manager_caches(void); 3528 int f2fs_rw_hint_to_seg_type(enum rw_hint hint); 3529 enum rw_hint f2fs_io_type_to_rw_hint(struct f2fs_sb_info *sbi, 3530 enum page_type type, enum temp_type temp); 3531 unsigned int f2fs_usable_segs_in_sec(struct f2fs_sb_info *sbi, 3532 unsigned int segno); 3533 unsigned int f2fs_usable_blks_in_seg(struct f2fs_sb_info *sbi, 3534 unsigned int segno); 3535 3536 #define DEF_FRAGMENT_SIZE 4 3537 #define MIN_FRAGMENT_SIZE 1 3538 #define MAX_FRAGMENT_SIZE 512 3539 3540 static inline bool f2fs_need_rand_seg(struct f2fs_sb_info *sbi) 3541 { 3542 return F2FS_OPTION(sbi).fs_mode == FS_MODE_FRAGMENT_SEG || 3543 F2FS_OPTION(sbi).fs_mode == FS_MODE_FRAGMENT_BLK; 3544 } 3545 3546 /* 3547 * checkpoint.c 3548 */ 3549 void f2fs_stop_checkpoint(struct f2fs_sb_info *sbi, bool end_io); 3550 struct page *f2fs_grab_meta_page(struct f2fs_sb_info *sbi, pgoff_t index); 3551 struct page *f2fs_get_meta_page(struct f2fs_sb_info *sbi, pgoff_t index); 3552 struct page *f2fs_get_meta_page_retry(struct f2fs_sb_info *sbi, pgoff_t index); 3553 struct page *f2fs_get_tmp_page(struct f2fs_sb_info *sbi, pgoff_t index); 3554 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi, 3555 block_t blkaddr, int type); 3556 int f2fs_ra_meta_pages(struct f2fs_sb_info *sbi, block_t start, int nrpages, 3557 int type, bool sync); 3558 void f2fs_ra_meta_pages_cond(struct f2fs_sb_info *sbi, pgoff_t index); 3559 long f2fs_sync_meta_pages(struct f2fs_sb_info *sbi, enum page_type type, 3560 long nr_to_write, enum iostat_type io_type); 3561 void f2fs_add_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type); 3562 void f2fs_remove_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type); 3563 void f2fs_release_ino_entry(struct f2fs_sb_info *sbi, bool all); 3564 bool f2fs_exist_written_data(struct f2fs_sb_info *sbi, nid_t ino, int mode); 3565 void f2fs_set_dirty_device(struct f2fs_sb_info *sbi, nid_t ino, 3566 unsigned int devidx, int type); 3567 bool f2fs_is_dirty_device(struct f2fs_sb_info *sbi, nid_t ino, 3568 unsigned int devidx, int type); 3569 int f2fs_sync_inode_meta(struct f2fs_sb_info *sbi); 3570 int f2fs_acquire_orphan_inode(struct f2fs_sb_info *sbi); 3571 void f2fs_release_orphan_inode(struct f2fs_sb_info *sbi); 3572 void f2fs_add_orphan_inode(struct inode *inode); 3573 void f2fs_remove_orphan_inode(struct f2fs_sb_info *sbi, nid_t ino); 3574 int f2fs_recover_orphan_inodes(struct f2fs_sb_info *sbi); 3575 int f2fs_get_valid_checkpoint(struct f2fs_sb_info *sbi); 3576 void f2fs_update_dirty_page(struct inode *inode, struct page *page); 3577 void f2fs_remove_dirty_inode(struct inode *inode); 3578 int f2fs_sync_dirty_inodes(struct f2fs_sb_info *sbi, enum inode_type type); 3579 void f2fs_wait_on_all_pages(struct f2fs_sb_info *sbi, int type); 3580 u64 f2fs_get_sectors_written(struct f2fs_sb_info *sbi); 3581 int f2fs_write_checkpoint(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3582 void f2fs_init_ino_entry_info(struct f2fs_sb_info *sbi); 3583 int __init f2fs_create_checkpoint_caches(void); 3584 void f2fs_destroy_checkpoint_caches(void); 3585 int f2fs_issue_checkpoint(struct f2fs_sb_info *sbi); 3586 int f2fs_start_ckpt_thread(struct f2fs_sb_info *sbi); 3587 void f2fs_stop_ckpt_thread(struct f2fs_sb_info *sbi); 3588 void f2fs_init_ckpt_req_control(struct f2fs_sb_info *sbi); 3589 3590 /* 3591 * data.c 3592 */ 3593 int __init f2fs_init_bioset(void); 3594 void f2fs_destroy_bioset(void); 3595 int f2fs_init_bio_entry_cache(void); 3596 void f2fs_destroy_bio_entry_cache(void); 3597 void f2fs_submit_bio(struct f2fs_sb_info *sbi, 3598 struct bio *bio, enum page_type type); 3599 void f2fs_submit_merged_write(struct f2fs_sb_info *sbi, enum page_type type); 3600 void f2fs_submit_merged_write_cond(struct f2fs_sb_info *sbi, 3601 struct inode *inode, struct page *page, 3602 nid_t ino, enum page_type type); 3603 void f2fs_submit_merged_ipu_write(struct f2fs_sb_info *sbi, 3604 struct bio **bio, struct page *page); 3605 void f2fs_flush_merged_writes(struct f2fs_sb_info *sbi); 3606 int f2fs_submit_page_bio(struct f2fs_io_info *fio); 3607 int f2fs_merge_page_bio(struct f2fs_io_info *fio); 3608 void f2fs_submit_page_write(struct f2fs_io_info *fio); 3609 struct block_device *f2fs_target_device(struct f2fs_sb_info *sbi, 3610 block_t blk_addr, struct bio *bio); 3611 int f2fs_target_device_index(struct f2fs_sb_info *sbi, block_t blkaddr); 3612 void f2fs_set_data_blkaddr(struct dnode_of_data *dn); 3613 void f2fs_update_data_blkaddr(struct dnode_of_data *dn, block_t blkaddr); 3614 int f2fs_reserve_new_blocks(struct dnode_of_data *dn, blkcnt_t count); 3615 int f2fs_reserve_new_block(struct dnode_of_data *dn); 3616 int f2fs_get_block(struct dnode_of_data *dn, pgoff_t index); 3617 int f2fs_preallocate_blocks(struct kiocb *iocb, struct iov_iter *from); 3618 int f2fs_reserve_block(struct dnode_of_data *dn, pgoff_t index); 3619 struct page *f2fs_get_read_data_page(struct inode *inode, pgoff_t index, 3620 int op_flags, bool for_write); 3621 struct page *f2fs_find_data_page(struct inode *inode, pgoff_t index); 3622 struct page *f2fs_get_lock_data_page(struct inode *inode, pgoff_t index, 3623 bool for_write); 3624 struct page *f2fs_get_new_data_page(struct inode *inode, 3625 struct page *ipage, pgoff_t index, bool new_i_size); 3626 int f2fs_do_write_data_page(struct f2fs_io_info *fio); 3627 void f2fs_do_map_lock(struct f2fs_sb_info *sbi, int flag, bool lock); 3628 int f2fs_map_blocks(struct inode *inode, struct f2fs_map_blocks *map, 3629 int create, int flag); 3630 int f2fs_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, 3631 u64 start, u64 len); 3632 int f2fs_encrypt_one_page(struct f2fs_io_info *fio); 3633 bool f2fs_should_update_inplace(struct inode *inode, struct f2fs_io_info *fio); 3634 bool f2fs_should_update_outplace(struct inode *inode, struct f2fs_io_info *fio); 3635 int f2fs_write_single_data_page(struct page *page, int *submitted, 3636 struct bio **bio, sector_t *last_block, 3637 struct writeback_control *wbc, 3638 enum iostat_type io_type, 3639 int compr_blocks, bool allow_balance); 3640 void f2fs_invalidate_page(struct page *page, unsigned int offset, 3641 unsigned int length); 3642 int f2fs_release_page(struct page *page, gfp_t wait); 3643 #ifdef CONFIG_MIGRATION 3644 int f2fs_migrate_page(struct address_space *mapping, struct page *newpage, 3645 struct page *page, enum migrate_mode mode); 3646 #endif 3647 bool f2fs_overwrite_io(struct inode *inode, loff_t pos, size_t len); 3648 void f2fs_clear_page_cache_dirty_tag(struct page *page); 3649 int f2fs_init_post_read_processing(void); 3650 void f2fs_destroy_post_read_processing(void); 3651 int f2fs_init_post_read_wq(struct f2fs_sb_info *sbi); 3652 void f2fs_destroy_post_read_wq(struct f2fs_sb_info *sbi); 3653 3654 /* 3655 * gc.c 3656 */ 3657 int f2fs_start_gc_thread(struct f2fs_sb_info *sbi); 3658 void f2fs_stop_gc_thread(struct f2fs_sb_info *sbi); 3659 block_t f2fs_start_bidx_of_node(unsigned int node_ofs, struct inode *inode); 3660 int f2fs_gc(struct f2fs_sb_info *sbi, bool sync, bool background, bool force, 3661 unsigned int segno); 3662 void f2fs_build_gc_manager(struct f2fs_sb_info *sbi); 3663 int f2fs_resize_fs(struct f2fs_sb_info *sbi, __u64 block_count); 3664 int __init f2fs_create_garbage_collection_cache(void); 3665 void f2fs_destroy_garbage_collection_cache(void); 3666 3667 /* 3668 * recovery.c 3669 */ 3670 int f2fs_recover_fsync_data(struct f2fs_sb_info *sbi, bool check_only); 3671 bool f2fs_space_for_roll_forward(struct f2fs_sb_info *sbi); 3672 int __init f2fs_create_recovery_cache(void); 3673 void f2fs_destroy_recovery_cache(void); 3674 3675 /* 3676 * debug.c 3677 */ 3678 #ifdef CONFIG_F2FS_STAT_FS 3679 struct f2fs_stat_info { 3680 struct list_head stat_list; 3681 struct f2fs_sb_info *sbi; 3682 int all_area_segs, sit_area_segs, nat_area_segs, ssa_area_segs; 3683 int main_area_segs, main_area_sections, main_area_zones; 3684 unsigned long long hit_largest, hit_cached, hit_rbtree; 3685 unsigned long long hit_total, total_ext; 3686 int ext_tree, zombie_tree, ext_node; 3687 int ndirty_node, ndirty_dent, ndirty_meta, ndirty_imeta; 3688 int ndirty_data, ndirty_qdata; 3689 int inmem_pages; 3690 unsigned int ndirty_dirs, ndirty_files, nquota_files, ndirty_all; 3691 int nats, dirty_nats, sits, dirty_sits; 3692 int free_nids, avail_nids, alloc_nids; 3693 int total_count, utilization; 3694 int bg_gc, nr_wb_cp_data, nr_wb_data; 3695 int nr_rd_data, nr_rd_node, nr_rd_meta; 3696 int nr_dio_read, nr_dio_write; 3697 unsigned int io_skip_bggc, other_skip_bggc; 3698 int nr_flushing, nr_flushed, flush_list_empty; 3699 int nr_discarding, nr_discarded; 3700 int nr_discard_cmd; 3701 unsigned int undiscard_blks; 3702 int nr_issued_ckpt, nr_total_ckpt, nr_queued_ckpt; 3703 unsigned int cur_ckpt_time, peak_ckpt_time; 3704 int inline_xattr, inline_inode, inline_dir, append, update, orphans; 3705 int compr_inode; 3706 unsigned long long compr_blocks; 3707 int aw_cnt, max_aw_cnt, vw_cnt, max_vw_cnt; 3708 unsigned int valid_count, valid_node_count, valid_inode_count, discard_blks; 3709 unsigned int bimodal, avg_vblocks; 3710 int util_free, util_valid, util_invalid; 3711 int rsvd_segs, overp_segs; 3712 int dirty_count, node_pages, meta_pages, compress_pages; 3713 int compress_page_hit; 3714 int prefree_count, call_count, cp_count, bg_cp_count; 3715 int tot_segs, node_segs, data_segs, free_segs, free_secs; 3716 int bg_node_segs, bg_data_segs; 3717 int tot_blks, data_blks, node_blks; 3718 int bg_data_blks, bg_node_blks; 3719 unsigned long long skipped_atomic_files[2]; 3720 int curseg[NR_CURSEG_TYPE]; 3721 int cursec[NR_CURSEG_TYPE]; 3722 int curzone[NR_CURSEG_TYPE]; 3723 unsigned int dirty_seg[NR_CURSEG_TYPE]; 3724 unsigned int full_seg[NR_CURSEG_TYPE]; 3725 unsigned int valid_blks[NR_CURSEG_TYPE]; 3726 3727 unsigned int meta_count[META_MAX]; 3728 unsigned int segment_count[2]; 3729 unsigned int block_count[2]; 3730 unsigned int inplace_count; 3731 unsigned long long base_mem, cache_mem, page_mem; 3732 }; 3733 3734 static inline struct f2fs_stat_info *F2FS_STAT(struct f2fs_sb_info *sbi) 3735 { 3736 return (struct f2fs_stat_info *)sbi->stat_info; 3737 } 3738 3739 #define stat_inc_cp_count(si) ((si)->cp_count++) 3740 #define stat_inc_bg_cp_count(si) ((si)->bg_cp_count++) 3741 #define stat_inc_call_count(si) ((si)->call_count++) 3742 #define stat_inc_bggc_count(si) ((si)->bg_gc++) 3743 #define stat_io_skip_bggc_count(sbi) ((sbi)->io_skip_bggc++) 3744 #define stat_other_skip_bggc_count(sbi) ((sbi)->other_skip_bggc++) 3745 #define stat_inc_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]++) 3746 #define stat_dec_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]--) 3747 #define stat_inc_total_hit(sbi) (atomic64_inc(&(sbi)->total_hit_ext)) 3748 #define stat_inc_rbtree_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_rbtree)) 3749 #define stat_inc_largest_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_largest)) 3750 #define stat_inc_cached_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_cached)) 3751 #define stat_inc_inline_xattr(inode) \ 3752 do { \ 3753 if (f2fs_has_inline_xattr(inode)) \ 3754 (atomic_inc(&F2FS_I_SB(inode)->inline_xattr)); \ 3755 } while (0) 3756 #define stat_dec_inline_xattr(inode) \ 3757 do { \ 3758 if (f2fs_has_inline_xattr(inode)) \ 3759 (atomic_dec(&F2FS_I_SB(inode)->inline_xattr)); \ 3760 } while (0) 3761 #define stat_inc_inline_inode(inode) \ 3762 do { \ 3763 if (f2fs_has_inline_data(inode)) \ 3764 (atomic_inc(&F2FS_I_SB(inode)->inline_inode)); \ 3765 } while (0) 3766 #define stat_dec_inline_inode(inode) \ 3767 do { \ 3768 if (f2fs_has_inline_data(inode)) \ 3769 (atomic_dec(&F2FS_I_SB(inode)->inline_inode)); \ 3770 } while (0) 3771 #define stat_inc_inline_dir(inode) \ 3772 do { \ 3773 if (f2fs_has_inline_dentry(inode)) \ 3774 (atomic_inc(&F2FS_I_SB(inode)->inline_dir)); \ 3775 } while (0) 3776 #define stat_dec_inline_dir(inode) \ 3777 do { \ 3778 if (f2fs_has_inline_dentry(inode)) \ 3779 (atomic_dec(&F2FS_I_SB(inode)->inline_dir)); \ 3780 } while (0) 3781 #define stat_inc_compr_inode(inode) \ 3782 do { \ 3783 if (f2fs_compressed_file(inode)) \ 3784 (atomic_inc(&F2FS_I_SB(inode)->compr_inode)); \ 3785 } while (0) 3786 #define stat_dec_compr_inode(inode) \ 3787 do { \ 3788 if (f2fs_compressed_file(inode)) \ 3789 (atomic_dec(&F2FS_I_SB(inode)->compr_inode)); \ 3790 } while (0) 3791 #define stat_add_compr_blocks(inode, blocks) \ 3792 (atomic64_add(blocks, &F2FS_I_SB(inode)->compr_blocks)) 3793 #define stat_sub_compr_blocks(inode, blocks) \ 3794 (atomic64_sub(blocks, &F2FS_I_SB(inode)->compr_blocks)) 3795 #define stat_inc_meta_count(sbi, blkaddr) \ 3796 do { \ 3797 if (blkaddr < SIT_I(sbi)->sit_base_addr) \ 3798 atomic_inc(&(sbi)->meta_count[META_CP]); \ 3799 else if (blkaddr < NM_I(sbi)->nat_blkaddr) \ 3800 atomic_inc(&(sbi)->meta_count[META_SIT]); \ 3801 else if (blkaddr < SM_I(sbi)->ssa_blkaddr) \ 3802 atomic_inc(&(sbi)->meta_count[META_NAT]); \ 3803 else if (blkaddr < SM_I(sbi)->main_blkaddr) \ 3804 atomic_inc(&(sbi)->meta_count[META_SSA]); \ 3805 } while (0) 3806 #define stat_inc_seg_type(sbi, curseg) \ 3807 ((sbi)->segment_count[(curseg)->alloc_type]++) 3808 #define stat_inc_block_count(sbi, curseg) \ 3809 ((sbi)->block_count[(curseg)->alloc_type]++) 3810 #define stat_inc_inplace_blocks(sbi) \ 3811 (atomic_inc(&(sbi)->inplace_count)) 3812 #define stat_update_max_atomic_write(inode) \ 3813 do { \ 3814 int cur = F2FS_I_SB(inode)->atomic_files; \ 3815 int max = atomic_read(&F2FS_I_SB(inode)->max_aw_cnt); \ 3816 if (cur > max) \ 3817 atomic_set(&F2FS_I_SB(inode)->max_aw_cnt, cur); \ 3818 } while (0) 3819 #define stat_inc_volatile_write(inode) \ 3820 (atomic_inc(&F2FS_I_SB(inode)->vw_cnt)) 3821 #define stat_dec_volatile_write(inode) \ 3822 (atomic_dec(&F2FS_I_SB(inode)->vw_cnt)) 3823 #define stat_update_max_volatile_write(inode) \ 3824 do { \ 3825 int cur = atomic_read(&F2FS_I_SB(inode)->vw_cnt); \ 3826 int max = atomic_read(&F2FS_I_SB(inode)->max_vw_cnt); \ 3827 if (cur > max) \ 3828 atomic_set(&F2FS_I_SB(inode)->max_vw_cnt, cur); \ 3829 } while (0) 3830 #define stat_inc_seg_count(sbi, type, gc_type) \ 3831 do { \ 3832 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 3833 si->tot_segs++; \ 3834 if ((type) == SUM_TYPE_DATA) { \ 3835 si->data_segs++; \ 3836 si->bg_data_segs += (gc_type == BG_GC) ? 1 : 0; \ 3837 } else { \ 3838 si->node_segs++; \ 3839 si->bg_node_segs += (gc_type == BG_GC) ? 1 : 0; \ 3840 } \ 3841 } while (0) 3842 3843 #define stat_inc_tot_blk_count(si, blks) \ 3844 ((si)->tot_blks += (blks)) 3845 3846 #define stat_inc_data_blk_count(sbi, blks, gc_type) \ 3847 do { \ 3848 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 3849 stat_inc_tot_blk_count(si, blks); \ 3850 si->data_blks += (blks); \ 3851 si->bg_data_blks += ((gc_type) == BG_GC) ? (blks) : 0; \ 3852 } while (0) 3853 3854 #define stat_inc_node_blk_count(sbi, blks, gc_type) \ 3855 do { \ 3856 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 3857 stat_inc_tot_blk_count(si, blks); \ 3858 si->node_blks += (blks); \ 3859 si->bg_node_blks += ((gc_type) == BG_GC) ? (blks) : 0; \ 3860 } while (0) 3861 3862 int f2fs_build_stats(struct f2fs_sb_info *sbi); 3863 void f2fs_destroy_stats(struct f2fs_sb_info *sbi); 3864 void __init f2fs_create_root_stats(void); 3865 void f2fs_destroy_root_stats(void); 3866 void f2fs_update_sit_info(struct f2fs_sb_info *sbi); 3867 #else 3868 #define stat_inc_cp_count(si) do { } while (0) 3869 #define stat_inc_bg_cp_count(si) do { } while (0) 3870 #define stat_inc_call_count(si) do { } while (0) 3871 #define stat_inc_bggc_count(si) do { } while (0) 3872 #define stat_io_skip_bggc_count(sbi) do { } while (0) 3873 #define stat_other_skip_bggc_count(sbi) do { } while (0) 3874 #define stat_inc_dirty_inode(sbi, type) do { } while (0) 3875 #define stat_dec_dirty_inode(sbi, type) do { } while (0) 3876 #define stat_inc_total_hit(sbi) do { } while (0) 3877 #define stat_inc_rbtree_node_hit(sbi) do { } while (0) 3878 #define stat_inc_largest_node_hit(sbi) do { } while (0) 3879 #define stat_inc_cached_node_hit(sbi) do { } while (0) 3880 #define stat_inc_inline_xattr(inode) do { } while (0) 3881 #define stat_dec_inline_xattr(inode) do { } while (0) 3882 #define stat_inc_inline_inode(inode) do { } while (0) 3883 #define stat_dec_inline_inode(inode) do { } while (0) 3884 #define stat_inc_inline_dir(inode) do { } while (0) 3885 #define stat_dec_inline_dir(inode) do { } while (0) 3886 #define stat_inc_compr_inode(inode) do { } while (0) 3887 #define stat_dec_compr_inode(inode) do { } while (0) 3888 #define stat_add_compr_blocks(inode, blocks) do { } while (0) 3889 #define stat_sub_compr_blocks(inode, blocks) do { } while (0) 3890 #define stat_update_max_atomic_write(inode) do { } while (0) 3891 #define stat_inc_volatile_write(inode) do { } while (0) 3892 #define stat_dec_volatile_write(inode) do { } while (0) 3893 #define stat_update_max_volatile_write(inode) do { } while (0) 3894 #define stat_inc_meta_count(sbi, blkaddr) do { } while (0) 3895 #define stat_inc_seg_type(sbi, curseg) do { } while (0) 3896 #define stat_inc_block_count(sbi, curseg) do { } while (0) 3897 #define stat_inc_inplace_blocks(sbi) do { } while (0) 3898 #define stat_inc_seg_count(sbi, type, gc_type) do { } while (0) 3899 #define stat_inc_tot_blk_count(si, blks) do { } while (0) 3900 #define stat_inc_data_blk_count(sbi, blks, gc_type) do { } while (0) 3901 #define stat_inc_node_blk_count(sbi, blks, gc_type) do { } while (0) 3902 3903 static inline int f2fs_build_stats(struct f2fs_sb_info *sbi) { return 0; } 3904 static inline void f2fs_destroy_stats(struct f2fs_sb_info *sbi) { } 3905 static inline void __init f2fs_create_root_stats(void) { } 3906 static inline void f2fs_destroy_root_stats(void) { } 3907 static inline void f2fs_update_sit_info(struct f2fs_sb_info *sbi) {} 3908 #endif 3909 3910 extern const struct file_operations f2fs_dir_operations; 3911 extern const struct file_operations f2fs_file_operations; 3912 extern const struct inode_operations f2fs_file_inode_operations; 3913 extern const struct address_space_operations f2fs_dblock_aops; 3914 extern const struct address_space_operations f2fs_node_aops; 3915 extern const struct address_space_operations f2fs_meta_aops; 3916 extern const struct inode_operations f2fs_dir_inode_operations; 3917 extern const struct inode_operations f2fs_symlink_inode_operations; 3918 extern const struct inode_operations f2fs_encrypted_symlink_inode_operations; 3919 extern const struct inode_operations f2fs_special_inode_operations; 3920 extern struct kmem_cache *f2fs_inode_entry_slab; 3921 3922 /* 3923 * inline.c 3924 */ 3925 bool f2fs_may_inline_data(struct inode *inode); 3926 bool f2fs_may_inline_dentry(struct inode *inode); 3927 void f2fs_do_read_inline_data(struct page *page, struct page *ipage); 3928 void f2fs_truncate_inline_inode(struct inode *inode, 3929 struct page *ipage, u64 from); 3930 int f2fs_read_inline_data(struct inode *inode, struct page *page); 3931 int f2fs_convert_inline_page(struct dnode_of_data *dn, struct page *page); 3932 int f2fs_convert_inline_inode(struct inode *inode); 3933 int f2fs_try_convert_inline_dir(struct inode *dir, struct dentry *dentry); 3934 int f2fs_write_inline_data(struct inode *inode, struct page *page); 3935 int f2fs_recover_inline_data(struct inode *inode, struct page *npage); 3936 struct f2fs_dir_entry *f2fs_find_in_inline_dir(struct inode *dir, 3937 const struct f2fs_filename *fname, 3938 struct page **res_page); 3939 int f2fs_make_empty_inline_dir(struct inode *inode, struct inode *parent, 3940 struct page *ipage); 3941 int f2fs_add_inline_entry(struct inode *dir, const struct f2fs_filename *fname, 3942 struct inode *inode, nid_t ino, umode_t mode); 3943 void f2fs_delete_inline_entry(struct f2fs_dir_entry *dentry, 3944 struct page *page, struct inode *dir, 3945 struct inode *inode); 3946 bool f2fs_empty_inline_dir(struct inode *dir); 3947 int f2fs_read_inline_dir(struct file *file, struct dir_context *ctx, 3948 struct fscrypt_str *fstr); 3949 int f2fs_inline_data_fiemap(struct inode *inode, 3950 struct fiemap_extent_info *fieinfo, 3951 __u64 start, __u64 len); 3952 3953 /* 3954 * shrinker.c 3955 */ 3956 unsigned long f2fs_shrink_count(struct shrinker *shrink, 3957 struct shrink_control *sc); 3958 unsigned long f2fs_shrink_scan(struct shrinker *shrink, 3959 struct shrink_control *sc); 3960 void f2fs_join_shrinker(struct f2fs_sb_info *sbi); 3961 void f2fs_leave_shrinker(struct f2fs_sb_info *sbi); 3962 3963 /* 3964 * extent_cache.c 3965 */ 3966 struct rb_entry *f2fs_lookup_rb_tree(struct rb_root_cached *root, 3967 struct rb_entry *cached_re, unsigned int ofs); 3968 struct rb_node **f2fs_lookup_rb_tree_ext(struct f2fs_sb_info *sbi, 3969 struct rb_root_cached *root, 3970 struct rb_node **parent, 3971 unsigned long long key, bool *left_most); 3972 struct rb_node **f2fs_lookup_rb_tree_for_insert(struct f2fs_sb_info *sbi, 3973 struct rb_root_cached *root, 3974 struct rb_node **parent, 3975 unsigned int ofs, bool *leftmost); 3976 struct rb_entry *f2fs_lookup_rb_tree_ret(struct rb_root_cached *root, 3977 struct rb_entry *cached_re, unsigned int ofs, 3978 struct rb_entry **prev_entry, struct rb_entry **next_entry, 3979 struct rb_node ***insert_p, struct rb_node **insert_parent, 3980 bool force, bool *leftmost); 3981 bool f2fs_check_rb_tree_consistence(struct f2fs_sb_info *sbi, 3982 struct rb_root_cached *root, bool check_key); 3983 unsigned int f2fs_shrink_extent_tree(struct f2fs_sb_info *sbi, int nr_shrink); 3984 void f2fs_init_extent_tree(struct inode *inode, struct page *ipage); 3985 void f2fs_drop_extent_tree(struct inode *inode); 3986 unsigned int f2fs_destroy_extent_node(struct inode *inode); 3987 void f2fs_destroy_extent_tree(struct inode *inode); 3988 bool f2fs_lookup_extent_cache(struct inode *inode, pgoff_t pgofs, 3989 struct extent_info *ei); 3990 void f2fs_update_extent_cache(struct dnode_of_data *dn); 3991 void f2fs_update_extent_cache_range(struct dnode_of_data *dn, 3992 pgoff_t fofs, block_t blkaddr, unsigned int len); 3993 void f2fs_init_extent_cache_info(struct f2fs_sb_info *sbi); 3994 int __init f2fs_create_extent_cache(void); 3995 void f2fs_destroy_extent_cache(void); 3996 3997 /* 3998 * sysfs.c 3999 */ 4000 #define MIN_RA_MUL 2 4001 #define MAX_RA_MUL 256 4002 4003 int __init f2fs_init_sysfs(void); 4004 void f2fs_exit_sysfs(void); 4005 int f2fs_register_sysfs(struct f2fs_sb_info *sbi); 4006 void f2fs_unregister_sysfs(struct f2fs_sb_info *sbi); 4007 4008 /* verity.c */ 4009 extern const struct fsverity_operations f2fs_verityops; 4010 4011 /* 4012 * crypto support 4013 */ 4014 static inline bool f2fs_encrypted_file(struct inode *inode) 4015 { 4016 return IS_ENCRYPTED(inode) && S_ISREG(inode->i_mode); 4017 } 4018 4019 static inline void f2fs_set_encrypted_inode(struct inode *inode) 4020 { 4021 #ifdef CONFIG_FS_ENCRYPTION 4022 file_set_encrypt(inode); 4023 f2fs_set_inode_flags(inode); 4024 #endif 4025 } 4026 4027 /* 4028 * Returns true if the reads of the inode's data need to undergo some 4029 * postprocessing step, like decryption or authenticity verification. 4030 */ 4031 static inline bool f2fs_post_read_required(struct inode *inode) 4032 { 4033 return f2fs_encrypted_file(inode) || fsverity_active(inode) || 4034 f2fs_compressed_file(inode); 4035 } 4036 4037 /* 4038 * compress.c 4039 */ 4040 #ifdef CONFIG_F2FS_FS_COMPRESSION 4041 bool f2fs_is_compressed_page(struct page *page); 4042 struct page *f2fs_compress_control_page(struct page *page); 4043 int f2fs_prepare_compress_overwrite(struct inode *inode, 4044 struct page **pagep, pgoff_t index, void **fsdata); 4045 bool f2fs_compress_write_end(struct inode *inode, void *fsdata, 4046 pgoff_t index, unsigned copied); 4047 int f2fs_truncate_partial_cluster(struct inode *inode, u64 from, bool lock); 4048 void f2fs_compress_write_end_io(struct bio *bio, struct page *page); 4049 bool f2fs_is_compress_backend_ready(struct inode *inode); 4050 int f2fs_init_compress_mempool(void); 4051 void f2fs_destroy_compress_mempool(void); 4052 void f2fs_decompress_cluster(struct decompress_io_ctx *dic); 4053 void f2fs_end_read_compressed_page(struct page *page, bool failed, 4054 block_t blkaddr); 4055 bool f2fs_cluster_is_empty(struct compress_ctx *cc); 4056 bool f2fs_cluster_can_merge_page(struct compress_ctx *cc, pgoff_t index); 4057 bool f2fs_all_cluster_page_loaded(struct compress_ctx *cc, struct pagevec *pvec, 4058 int index, int nr_pages); 4059 bool f2fs_sanity_check_cluster(struct dnode_of_data *dn); 4060 void f2fs_compress_ctx_add_page(struct compress_ctx *cc, struct page *page); 4061 int f2fs_write_multi_pages(struct compress_ctx *cc, 4062 int *submitted, 4063 struct writeback_control *wbc, 4064 enum iostat_type io_type); 4065 int f2fs_is_compressed_cluster(struct inode *inode, pgoff_t index); 4066 void f2fs_update_extent_tree_range_compressed(struct inode *inode, 4067 pgoff_t fofs, block_t blkaddr, unsigned int llen, 4068 unsigned int c_len); 4069 int f2fs_read_multi_pages(struct compress_ctx *cc, struct bio **bio_ret, 4070 unsigned nr_pages, sector_t *last_block_in_bio, 4071 bool is_readahead, bool for_write); 4072 struct decompress_io_ctx *f2fs_alloc_dic(struct compress_ctx *cc); 4073 void f2fs_decompress_end_io(struct decompress_io_ctx *dic, bool failed); 4074 void f2fs_put_page_dic(struct page *page); 4075 unsigned int f2fs_cluster_blocks_are_contiguous(struct dnode_of_data *dn); 4076 int f2fs_init_compress_ctx(struct compress_ctx *cc); 4077 void f2fs_destroy_compress_ctx(struct compress_ctx *cc, bool reuse); 4078 void f2fs_init_compress_info(struct f2fs_sb_info *sbi); 4079 int f2fs_init_compress_inode(struct f2fs_sb_info *sbi); 4080 void f2fs_destroy_compress_inode(struct f2fs_sb_info *sbi); 4081 int f2fs_init_page_array_cache(struct f2fs_sb_info *sbi); 4082 void f2fs_destroy_page_array_cache(struct f2fs_sb_info *sbi); 4083 int __init f2fs_init_compress_cache(void); 4084 void f2fs_destroy_compress_cache(void); 4085 struct address_space *COMPRESS_MAPPING(struct f2fs_sb_info *sbi); 4086 void f2fs_invalidate_compress_page(struct f2fs_sb_info *sbi, block_t blkaddr); 4087 void f2fs_cache_compressed_page(struct f2fs_sb_info *sbi, struct page *page, 4088 nid_t ino, block_t blkaddr); 4089 bool f2fs_load_compressed_page(struct f2fs_sb_info *sbi, struct page *page, 4090 block_t blkaddr); 4091 void f2fs_invalidate_compress_pages(struct f2fs_sb_info *sbi, nid_t ino); 4092 #define inc_compr_inode_stat(inode) \ 4093 do { \ 4094 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); \ 4095 sbi->compr_new_inode++; \ 4096 } while (0) 4097 #define add_compr_block_stat(inode, blocks) \ 4098 do { \ 4099 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); \ 4100 int diff = F2FS_I(inode)->i_cluster_size - blocks; \ 4101 sbi->compr_written_block += blocks; \ 4102 sbi->compr_saved_block += diff; \ 4103 } while (0) 4104 #else 4105 static inline bool f2fs_is_compressed_page(struct page *page) { return false; } 4106 static inline bool f2fs_is_compress_backend_ready(struct inode *inode) 4107 { 4108 if (!f2fs_compressed_file(inode)) 4109 return true; 4110 /* not support compression */ 4111 return false; 4112 } 4113 static inline struct page *f2fs_compress_control_page(struct page *page) 4114 { 4115 WARN_ON_ONCE(1); 4116 return ERR_PTR(-EINVAL); 4117 } 4118 static inline int f2fs_init_compress_mempool(void) { return 0; } 4119 static inline void f2fs_destroy_compress_mempool(void) { } 4120 static inline void f2fs_decompress_cluster(struct decompress_io_ctx *dic) { } 4121 static inline void f2fs_end_read_compressed_page(struct page *page, 4122 bool failed, block_t blkaddr) 4123 { 4124 WARN_ON_ONCE(1); 4125 } 4126 static inline void f2fs_put_page_dic(struct page *page) 4127 { 4128 WARN_ON_ONCE(1); 4129 } 4130 static inline unsigned int f2fs_cluster_blocks_are_contiguous(struct dnode_of_data *dn) { return 0; } 4131 static inline bool f2fs_sanity_check_cluster(struct dnode_of_data *dn) { return false; } 4132 static inline int f2fs_init_compress_inode(struct f2fs_sb_info *sbi) { return 0; } 4133 static inline void f2fs_destroy_compress_inode(struct f2fs_sb_info *sbi) { } 4134 static inline int f2fs_init_page_array_cache(struct f2fs_sb_info *sbi) { return 0; } 4135 static inline void f2fs_destroy_page_array_cache(struct f2fs_sb_info *sbi) { } 4136 static inline int __init f2fs_init_compress_cache(void) { return 0; } 4137 static inline void f2fs_destroy_compress_cache(void) { } 4138 static inline void f2fs_invalidate_compress_page(struct f2fs_sb_info *sbi, 4139 block_t blkaddr) { } 4140 static inline void f2fs_cache_compressed_page(struct f2fs_sb_info *sbi, 4141 struct page *page, nid_t ino, block_t blkaddr) { } 4142 static inline bool f2fs_load_compressed_page(struct f2fs_sb_info *sbi, 4143 struct page *page, block_t blkaddr) { return false; } 4144 static inline void f2fs_invalidate_compress_pages(struct f2fs_sb_info *sbi, 4145 nid_t ino) { } 4146 #define inc_compr_inode_stat(inode) do { } while (0) 4147 static inline void f2fs_update_extent_tree_range_compressed(struct inode *inode, 4148 pgoff_t fofs, block_t blkaddr, unsigned int llen, 4149 unsigned int c_len) { } 4150 #endif 4151 4152 static inline void set_compress_context(struct inode *inode) 4153 { 4154 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 4155 4156 F2FS_I(inode)->i_compress_algorithm = 4157 F2FS_OPTION(sbi).compress_algorithm; 4158 F2FS_I(inode)->i_log_cluster_size = 4159 F2FS_OPTION(sbi).compress_log_size; 4160 F2FS_I(inode)->i_compress_flag = 4161 F2FS_OPTION(sbi).compress_chksum ? 4162 1 << COMPRESS_CHKSUM : 0; 4163 F2FS_I(inode)->i_cluster_size = 4164 1 << F2FS_I(inode)->i_log_cluster_size; 4165 if ((F2FS_I(inode)->i_compress_algorithm == COMPRESS_LZ4 || 4166 F2FS_I(inode)->i_compress_algorithm == COMPRESS_ZSTD) && 4167 F2FS_OPTION(sbi).compress_level) 4168 F2FS_I(inode)->i_compress_flag |= 4169 F2FS_OPTION(sbi).compress_level << 4170 COMPRESS_LEVEL_OFFSET; 4171 F2FS_I(inode)->i_flags |= F2FS_COMPR_FL; 4172 set_inode_flag(inode, FI_COMPRESSED_FILE); 4173 stat_inc_compr_inode(inode); 4174 inc_compr_inode_stat(inode); 4175 f2fs_mark_inode_dirty_sync(inode, true); 4176 } 4177 4178 static inline bool f2fs_disable_compressed_file(struct inode *inode) 4179 { 4180 struct f2fs_inode_info *fi = F2FS_I(inode); 4181 4182 if (!f2fs_compressed_file(inode)) 4183 return true; 4184 if (S_ISREG(inode->i_mode) && F2FS_HAS_BLOCKS(inode)) 4185 return false; 4186 4187 fi->i_flags &= ~F2FS_COMPR_FL; 4188 stat_dec_compr_inode(inode); 4189 clear_inode_flag(inode, FI_COMPRESSED_FILE); 4190 f2fs_mark_inode_dirty_sync(inode, true); 4191 return true; 4192 } 4193 4194 #define F2FS_FEATURE_FUNCS(name, flagname) \ 4195 static inline int f2fs_sb_has_##name(struct f2fs_sb_info *sbi) \ 4196 { \ 4197 return F2FS_HAS_FEATURE(sbi, F2FS_FEATURE_##flagname); \ 4198 } 4199 4200 F2FS_FEATURE_FUNCS(encrypt, ENCRYPT); 4201 F2FS_FEATURE_FUNCS(blkzoned, BLKZONED); 4202 F2FS_FEATURE_FUNCS(extra_attr, EXTRA_ATTR); 4203 F2FS_FEATURE_FUNCS(project_quota, PRJQUOTA); 4204 F2FS_FEATURE_FUNCS(inode_chksum, INODE_CHKSUM); 4205 F2FS_FEATURE_FUNCS(flexible_inline_xattr, FLEXIBLE_INLINE_XATTR); 4206 F2FS_FEATURE_FUNCS(quota_ino, QUOTA_INO); 4207 F2FS_FEATURE_FUNCS(inode_crtime, INODE_CRTIME); 4208 F2FS_FEATURE_FUNCS(lost_found, LOST_FOUND); 4209 F2FS_FEATURE_FUNCS(verity, VERITY); 4210 F2FS_FEATURE_FUNCS(sb_chksum, SB_CHKSUM); 4211 F2FS_FEATURE_FUNCS(casefold, CASEFOLD); 4212 F2FS_FEATURE_FUNCS(compression, COMPRESSION); 4213 F2FS_FEATURE_FUNCS(readonly, RO); 4214 4215 static inline bool f2fs_may_extent_tree(struct inode *inode) 4216 { 4217 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 4218 4219 if (!test_opt(sbi, EXTENT_CACHE) || 4220 is_inode_flag_set(inode, FI_NO_EXTENT) || 4221 (is_inode_flag_set(inode, FI_COMPRESSED_FILE) && 4222 !f2fs_sb_has_readonly(sbi))) 4223 return false; 4224 4225 /* 4226 * for recovered files during mount do not create extents 4227 * if shrinker is not registered. 4228 */ 4229 if (list_empty(&sbi->s_list)) 4230 return false; 4231 4232 return S_ISREG(inode->i_mode); 4233 } 4234 4235 #ifdef CONFIG_BLK_DEV_ZONED 4236 static inline bool f2fs_blkz_is_seq(struct f2fs_sb_info *sbi, int devi, 4237 block_t blkaddr) 4238 { 4239 unsigned int zno = blkaddr >> sbi->log_blocks_per_blkz; 4240 4241 return test_bit(zno, FDEV(devi).blkz_seq); 4242 } 4243 #endif 4244 4245 static inline bool f2fs_hw_should_discard(struct f2fs_sb_info *sbi) 4246 { 4247 return f2fs_sb_has_blkzoned(sbi); 4248 } 4249 4250 static inline bool f2fs_bdev_support_discard(struct block_device *bdev) 4251 { 4252 return blk_queue_discard(bdev_get_queue(bdev)) || 4253 bdev_is_zoned(bdev); 4254 } 4255 4256 static inline bool f2fs_hw_support_discard(struct f2fs_sb_info *sbi) 4257 { 4258 int i; 4259 4260 if (!f2fs_is_multi_device(sbi)) 4261 return f2fs_bdev_support_discard(sbi->sb->s_bdev); 4262 4263 for (i = 0; i < sbi->s_ndevs; i++) 4264 if (f2fs_bdev_support_discard(FDEV(i).bdev)) 4265 return true; 4266 return false; 4267 } 4268 4269 static inline bool f2fs_realtime_discard_enable(struct f2fs_sb_info *sbi) 4270 { 4271 return (test_opt(sbi, DISCARD) && f2fs_hw_support_discard(sbi)) || 4272 f2fs_hw_should_discard(sbi); 4273 } 4274 4275 static inline bool f2fs_hw_is_readonly(struct f2fs_sb_info *sbi) 4276 { 4277 int i; 4278 4279 if (!f2fs_is_multi_device(sbi)) 4280 return bdev_read_only(sbi->sb->s_bdev); 4281 4282 for (i = 0; i < sbi->s_ndevs; i++) 4283 if (bdev_read_only(FDEV(i).bdev)) 4284 return true; 4285 return false; 4286 } 4287 4288 static inline bool f2fs_lfs_mode(struct f2fs_sb_info *sbi) 4289 { 4290 return F2FS_OPTION(sbi).fs_mode == FS_MODE_LFS; 4291 } 4292 4293 static inline bool f2fs_may_compress(struct inode *inode) 4294 { 4295 if (IS_SWAPFILE(inode) || f2fs_is_pinned_file(inode) || 4296 f2fs_is_atomic_file(inode) || 4297 f2fs_is_volatile_file(inode)) 4298 return false; 4299 return S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode); 4300 } 4301 4302 static inline void f2fs_i_compr_blocks_update(struct inode *inode, 4303 u64 blocks, bool add) 4304 { 4305 int diff = F2FS_I(inode)->i_cluster_size - blocks; 4306 struct f2fs_inode_info *fi = F2FS_I(inode); 4307 4308 /* don't update i_compr_blocks if saved blocks were released */ 4309 if (!add && !atomic_read(&fi->i_compr_blocks)) 4310 return; 4311 4312 if (add) { 4313 atomic_add(diff, &fi->i_compr_blocks); 4314 stat_add_compr_blocks(inode, diff); 4315 } else { 4316 atomic_sub(diff, &fi->i_compr_blocks); 4317 stat_sub_compr_blocks(inode, diff); 4318 } 4319 f2fs_mark_inode_dirty_sync(inode, true); 4320 } 4321 4322 static inline int block_unaligned_IO(struct inode *inode, 4323 struct kiocb *iocb, struct iov_iter *iter) 4324 { 4325 unsigned int i_blkbits = READ_ONCE(inode->i_blkbits); 4326 unsigned int blocksize_mask = (1 << i_blkbits) - 1; 4327 loff_t offset = iocb->ki_pos; 4328 unsigned long align = offset | iov_iter_alignment(iter); 4329 4330 return align & blocksize_mask; 4331 } 4332 4333 static inline bool f2fs_allow_multi_device_dio(struct f2fs_sb_info *sbi, 4334 int flag) 4335 { 4336 if (!f2fs_is_multi_device(sbi)) 4337 return false; 4338 if (flag != F2FS_GET_BLOCK_DIO) 4339 return false; 4340 return sbi->aligned_blksize; 4341 } 4342 4343 static inline bool f2fs_force_buffered_io(struct inode *inode, 4344 struct kiocb *iocb, struct iov_iter *iter) 4345 { 4346 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 4347 int rw = iov_iter_rw(iter); 4348 4349 if (f2fs_post_read_required(inode)) 4350 return true; 4351 4352 /* disallow direct IO if any of devices has unaligned blksize */ 4353 if (f2fs_is_multi_device(sbi) && !sbi->aligned_blksize) 4354 return true; 4355 /* 4356 * for blkzoned device, fallback direct IO to buffered IO, so 4357 * all IOs can be serialized by log-structured write. 4358 */ 4359 if (f2fs_sb_has_blkzoned(sbi)) 4360 return true; 4361 if (f2fs_lfs_mode(sbi) && (rw == WRITE)) { 4362 if (block_unaligned_IO(inode, iocb, iter)) 4363 return true; 4364 if (F2FS_IO_ALIGNED(sbi)) 4365 return true; 4366 } 4367 if (is_sbi_flag_set(F2FS_I_SB(inode), SBI_CP_DISABLED)) 4368 return true; 4369 4370 return false; 4371 } 4372 4373 static inline bool f2fs_need_verity(const struct inode *inode, pgoff_t idx) 4374 { 4375 return fsverity_active(inode) && 4376 idx < DIV_ROUND_UP(inode->i_size, PAGE_SIZE); 4377 } 4378 4379 #ifdef CONFIG_F2FS_FAULT_INJECTION 4380 extern void f2fs_build_fault_attr(struct f2fs_sb_info *sbi, unsigned int rate, 4381 unsigned int type); 4382 #else 4383 #define f2fs_build_fault_attr(sbi, rate, type) do { } while (0) 4384 #endif 4385 4386 static inline bool is_journalled_quota(struct f2fs_sb_info *sbi) 4387 { 4388 #ifdef CONFIG_QUOTA 4389 if (f2fs_sb_has_quota_ino(sbi)) 4390 return true; 4391 if (F2FS_OPTION(sbi).s_qf_names[USRQUOTA] || 4392 F2FS_OPTION(sbi).s_qf_names[GRPQUOTA] || 4393 F2FS_OPTION(sbi).s_qf_names[PRJQUOTA]) 4394 return true; 4395 #endif 4396 return false; 4397 } 4398 4399 static inline bool f2fs_block_unit_discard(struct f2fs_sb_info *sbi) 4400 { 4401 return F2FS_OPTION(sbi).discard_unit == DISCARD_UNIT_BLOCK; 4402 } 4403 4404 #define EFSBADCRC EBADMSG /* Bad CRC detected */ 4405 #define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */ 4406 4407 #endif /* _LINUX_F2FS_H */ 4408