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 (unlikely(condition)) { \ 37 WARN_ON(1); \ 38 set_sbi_flag(sbi, SBI_NEED_FSCK); \ 39 } \ 40 } while (0) 41 #endif 42 43 enum { 44 FAULT_KMALLOC, 45 FAULT_KVMALLOC, 46 FAULT_PAGE_ALLOC, 47 FAULT_PAGE_GET, 48 FAULT_ALLOC_BIO, 49 FAULT_ALLOC_NID, 50 FAULT_ORPHAN, 51 FAULT_BLOCK, 52 FAULT_DIR_DEPTH, 53 FAULT_EVICT_INODE, 54 FAULT_TRUNCATE, 55 FAULT_READ_IO, 56 FAULT_CHECKPOINT, 57 FAULT_DISCARD, 58 FAULT_WRITE_IO, 59 FAULT_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 102 103 #define F2FS_OPTION(sbi) ((sbi)->mount_opt) 104 #define clear_opt(sbi, option) (F2FS_OPTION(sbi).opt &= ~F2FS_MOUNT_##option) 105 #define set_opt(sbi, option) (F2FS_OPTION(sbi).opt |= F2FS_MOUNT_##option) 106 #define test_opt(sbi, option) (F2FS_OPTION(sbi).opt & F2FS_MOUNT_##option) 107 108 #define ver_after(a, b) (typecheck(unsigned long long, a) && \ 109 typecheck(unsigned long long, b) && \ 110 ((long long)((a) - (b)) > 0)) 111 112 typedef u32 block_t; /* 113 * should not change u32, since it is the on-disk block 114 * address format, __le32. 115 */ 116 typedef u32 nid_t; 117 118 #define COMPRESS_EXT_NUM 16 119 120 struct f2fs_mount_info { 121 unsigned int opt; 122 int write_io_size_bits; /* Write IO size bits */ 123 block_t root_reserved_blocks; /* root reserved blocks */ 124 kuid_t s_resuid; /* reserved blocks for uid */ 125 kgid_t s_resgid; /* reserved blocks for gid */ 126 int active_logs; /* # of active logs */ 127 int inline_xattr_size; /* inline xattr size */ 128 #ifdef CONFIG_F2FS_FAULT_INJECTION 129 struct f2fs_fault_info fault_info; /* For fault injection */ 130 #endif 131 #ifdef CONFIG_QUOTA 132 /* Names of quota files with journalled quota */ 133 char *s_qf_names[MAXQUOTAS]; 134 int s_jquota_fmt; /* Format of quota to use */ 135 #endif 136 /* For which write hints are passed down to block layer */ 137 int whint_mode; 138 int alloc_mode; /* segment allocation policy */ 139 int fsync_mode; /* fsync policy */ 140 int fs_mode; /* fs mode: LFS or ADAPTIVE */ 141 int bggc_mode; /* bggc mode: off, on or sync */ 142 struct fscrypt_dummy_policy dummy_enc_policy; /* test dummy encryption */ 143 block_t unusable_cap_perc; /* percentage for cap */ 144 block_t unusable_cap; /* Amount of space allowed to be 145 * unusable when disabling checkpoint 146 */ 147 148 /* For compression */ 149 unsigned char compress_algorithm; /* algorithm type */ 150 unsigned compress_log_size; /* cluster log size */ 151 unsigned char compress_ext_cnt; /* extension count */ 152 unsigned char extensions[COMPRESS_EXT_NUM][F2FS_EXTENSION_LEN]; /* extensions */ 153 }; 154 155 #define F2FS_FEATURE_ENCRYPT 0x0001 156 #define F2FS_FEATURE_BLKZONED 0x0002 157 #define F2FS_FEATURE_ATOMIC_WRITE 0x0004 158 #define F2FS_FEATURE_EXTRA_ATTR 0x0008 159 #define F2FS_FEATURE_PRJQUOTA 0x0010 160 #define F2FS_FEATURE_INODE_CHKSUM 0x0020 161 #define F2FS_FEATURE_FLEXIBLE_INLINE_XATTR 0x0040 162 #define F2FS_FEATURE_QUOTA_INO 0x0080 163 #define F2FS_FEATURE_INODE_CRTIME 0x0100 164 #define F2FS_FEATURE_LOST_FOUND 0x0200 165 #define F2FS_FEATURE_VERITY 0x0400 166 #define F2FS_FEATURE_SB_CHKSUM 0x0800 167 #define F2FS_FEATURE_CASEFOLD 0x1000 168 #define F2FS_FEATURE_COMPRESSION 0x2000 169 170 #define __F2FS_HAS_FEATURE(raw_super, mask) \ 171 ((raw_super->feature & cpu_to_le32(mask)) != 0) 172 #define F2FS_HAS_FEATURE(sbi, mask) __F2FS_HAS_FEATURE(sbi->raw_super, mask) 173 #define F2FS_SET_FEATURE(sbi, mask) \ 174 (sbi->raw_super->feature |= cpu_to_le32(mask)) 175 #define F2FS_CLEAR_FEATURE(sbi, mask) \ 176 (sbi->raw_super->feature &= ~cpu_to_le32(mask)) 177 178 /* 179 * Default values for user and/or group using reserved blocks 180 */ 181 #define F2FS_DEF_RESUID 0 182 #define F2FS_DEF_RESGID 0 183 184 /* 185 * For checkpoint manager 186 */ 187 enum { 188 NAT_BITMAP, 189 SIT_BITMAP 190 }; 191 192 #define CP_UMOUNT 0x00000001 193 #define CP_FASTBOOT 0x00000002 194 #define CP_SYNC 0x00000004 195 #define CP_RECOVERY 0x00000008 196 #define CP_DISCARD 0x00000010 197 #define CP_TRIMMED 0x00000020 198 #define CP_PAUSE 0x00000040 199 #define CP_RESIZE 0x00000080 200 201 #define MAX_DISCARD_BLOCKS(sbi) BLKS_PER_SEC(sbi) 202 #define DEF_MAX_DISCARD_REQUEST 8 /* issue 8 discards per round */ 203 #define DEF_MIN_DISCARD_ISSUE_TIME 50 /* 50 ms, if exists */ 204 #define DEF_MID_DISCARD_ISSUE_TIME 500 /* 500 ms, if device busy */ 205 #define DEF_MAX_DISCARD_ISSUE_TIME 60000 /* 60 s, if no candidates */ 206 #define DEF_DISCARD_URGENT_UTIL 80 /* do more discard over 80% */ 207 #define DEF_CP_INTERVAL 60 /* 60 secs */ 208 #define DEF_IDLE_INTERVAL 5 /* 5 secs */ 209 #define DEF_DISABLE_INTERVAL 5 /* 5 secs */ 210 #define DEF_DISABLE_QUICK_INTERVAL 1 /* 1 secs */ 211 #define DEF_UMOUNT_DISCARD_TIMEOUT 5 /* 5 secs */ 212 213 struct cp_control { 214 int reason; 215 __u64 trim_start; 216 __u64 trim_end; 217 __u64 trim_minlen; 218 }; 219 220 /* 221 * indicate meta/data type 222 */ 223 enum { 224 META_CP, 225 META_NAT, 226 META_SIT, 227 META_SSA, 228 META_MAX, 229 META_POR, 230 DATA_GENERIC, /* check range only */ 231 DATA_GENERIC_ENHANCE, /* strong check on range and segment bitmap */ 232 DATA_GENERIC_ENHANCE_READ, /* 233 * strong check on range and segment 234 * bitmap but no warning due to race 235 * condition of read on truncated area 236 * by extent_cache 237 */ 238 META_GENERIC, 239 }; 240 241 /* for the list of ino */ 242 enum { 243 ORPHAN_INO, /* for orphan ino list */ 244 APPEND_INO, /* for append ino list */ 245 UPDATE_INO, /* for update ino list */ 246 TRANS_DIR_INO, /* for trasactions dir ino list */ 247 FLUSH_INO, /* for multiple device flushing */ 248 MAX_INO_ENTRY, /* max. list */ 249 }; 250 251 struct ino_entry { 252 struct list_head list; /* list head */ 253 nid_t ino; /* inode number */ 254 unsigned int dirty_device; /* dirty device bitmap */ 255 }; 256 257 /* for the list of inodes to be GCed */ 258 struct inode_entry { 259 struct list_head list; /* list head */ 260 struct inode *inode; /* vfs inode pointer */ 261 }; 262 263 struct fsync_node_entry { 264 struct list_head list; /* list head */ 265 struct page *page; /* warm node page pointer */ 266 unsigned int seq_id; /* sequence id */ 267 }; 268 269 /* for the bitmap indicate blocks to be discarded */ 270 struct discard_entry { 271 struct list_head list; /* list head */ 272 block_t start_blkaddr; /* start blockaddr of current segment */ 273 unsigned char discard_map[SIT_VBLOCK_MAP_SIZE]; /* segment discard bitmap */ 274 }; 275 276 /* default discard granularity of inner discard thread, unit: block count */ 277 #define DEFAULT_DISCARD_GRANULARITY 16 278 279 /* max discard pend list number */ 280 #define MAX_PLIST_NUM 512 281 #define plist_idx(blk_num) ((blk_num) >= MAX_PLIST_NUM ? \ 282 (MAX_PLIST_NUM - 1) : ((blk_num) - 1)) 283 284 enum { 285 D_PREP, /* initial */ 286 D_PARTIAL, /* partially submitted */ 287 D_SUBMIT, /* all submitted */ 288 D_DONE, /* finished */ 289 }; 290 291 struct discard_info { 292 block_t lstart; /* logical start address */ 293 block_t len; /* length */ 294 block_t start; /* actual start address in dev */ 295 }; 296 297 struct discard_cmd { 298 struct rb_node rb_node; /* rb node located in rb-tree */ 299 union { 300 struct { 301 block_t lstart; /* logical start address */ 302 block_t len; /* length */ 303 block_t start; /* actual start address in dev */ 304 }; 305 struct discard_info di; /* discard info */ 306 307 }; 308 struct list_head list; /* command list */ 309 struct completion wait; /* compleation */ 310 struct block_device *bdev; /* bdev */ 311 unsigned short ref; /* reference count */ 312 unsigned char state; /* state */ 313 unsigned char queued; /* queued discard */ 314 int error; /* bio error */ 315 spinlock_t lock; /* for state/bio_ref updating */ 316 unsigned short bio_ref; /* bio reference count */ 317 }; 318 319 enum { 320 DPOLICY_BG, 321 DPOLICY_FORCE, 322 DPOLICY_FSTRIM, 323 DPOLICY_UMOUNT, 324 MAX_DPOLICY, 325 }; 326 327 struct discard_policy { 328 int type; /* type of discard */ 329 unsigned int min_interval; /* used for candidates exist */ 330 unsigned int mid_interval; /* used for device busy */ 331 unsigned int max_interval; /* used for candidates not exist */ 332 unsigned int max_requests; /* # of discards issued per round */ 333 unsigned int io_aware_gran; /* minimum granularity discard not be aware of I/O */ 334 bool io_aware; /* issue discard in idle time */ 335 bool sync; /* submit discard with REQ_SYNC flag */ 336 bool ordered; /* issue discard by lba order */ 337 bool timeout; /* discard timeout for put_super */ 338 unsigned int granularity; /* discard granularity */ 339 }; 340 341 struct discard_cmd_control { 342 struct task_struct *f2fs_issue_discard; /* discard thread */ 343 struct list_head entry_list; /* 4KB discard entry list */ 344 struct list_head pend_list[MAX_PLIST_NUM];/* store pending entries */ 345 struct list_head wait_list; /* store on-flushing entries */ 346 struct list_head fstrim_list; /* in-flight discard from fstrim */ 347 wait_queue_head_t discard_wait_queue; /* waiting queue for wake-up */ 348 unsigned int discard_wake; /* to wake up discard thread */ 349 struct mutex cmd_lock; 350 unsigned int nr_discards; /* # of discards in the list */ 351 unsigned int max_discards; /* max. discards to be issued */ 352 unsigned int discard_granularity; /* discard granularity */ 353 unsigned int undiscard_blks; /* # of undiscard blocks */ 354 unsigned int next_pos; /* next discard position */ 355 atomic_t issued_discard; /* # of issued discard */ 356 atomic_t queued_discard; /* # of queued discard */ 357 atomic_t discard_cmd_cnt; /* # of cached cmd count */ 358 struct rb_root_cached root; /* root of discard rb-tree */ 359 bool rbtree_check; /* config for consistence check */ 360 }; 361 362 /* for the list of fsync inodes, used only during recovery */ 363 struct fsync_inode_entry { 364 struct list_head list; /* list head */ 365 struct inode *inode; /* vfs inode pointer */ 366 block_t blkaddr; /* block address locating the last fsync */ 367 block_t last_dentry; /* block address locating the last dentry */ 368 }; 369 370 #define nats_in_cursum(jnl) (le16_to_cpu((jnl)->n_nats)) 371 #define sits_in_cursum(jnl) (le16_to_cpu((jnl)->n_sits)) 372 373 #define nat_in_journal(jnl, i) ((jnl)->nat_j.entries[i].ne) 374 #define nid_in_journal(jnl, i) ((jnl)->nat_j.entries[i].nid) 375 #define sit_in_journal(jnl, i) ((jnl)->sit_j.entries[i].se) 376 #define segno_in_journal(jnl, i) ((jnl)->sit_j.entries[i].segno) 377 378 #define MAX_NAT_JENTRIES(jnl) (NAT_JOURNAL_ENTRIES - nats_in_cursum(jnl)) 379 #define MAX_SIT_JENTRIES(jnl) (SIT_JOURNAL_ENTRIES - sits_in_cursum(jnl)) 380 381 static inline int update_nats_in_cursum(struct f2fs_journal *journal, int i) 382 { 383 int before = nats_in_cursum(journal); 384 385 journal->n_nats = cpu_to_le16(before + i); 386 return before; 387 } 388 389 static inline int update_sits_in_cursum(struct f2fs_journal *journal, int i) 390 { 391 int before = sits_in_cursum(journal); 392 393 journal->n_sits = cpu_to_le16(before + i); 394 return before; 395 } 396 397 static inline bool __has_cursum_space(struct f2fs_journal *journal, 398 int size, int type) 399 { 400 if (type == NAT_JOURNAL) 401 return size <= MAX_NAT_JENTRIES(journal); 402 return size <= MAX_SIT_JENTRIES(journal); 403 } 404 405 /* 406 * f2fs-specific ioctl commands 407 */ 408 #define F2FS_IOCTL_MAGIC 0xf5 409 #define F2FS_IOC_START_ATOMIC_WRITE _IO(F2FS_IOCTL_MAGIC, 1) 410 #define F2FS_IOC_COMMIT_ATOMIC_WRITE _IO(F2FS_IOCTL_MAGIC, 2) 411 #define F2FS_IOC_START_VOLATILE_WRITE _IO(F2FS_IOCTL_MAGIC, 3) 412 #define F2FS_IOC_RELEASE_VOLATILE_WRITE _IO(F2FS_IOCTL_MAGIC, 4) 413 #define F2FS_IOC_ABORT_VOLATILE_WRITE _IO(F2FS_IOCTL_MAGIC, 5) 414 #define F2FS_IOC_GARBAGE_COLLECT _IOW(F2FS_IOCTL_MAGIC, 6, __u32) 415 #define F2FS_IOC_WRITE_CHECKPOINT _IO(F2FS_IOCTL_MAGIC, 7) 416 #define F2FS_IOC_DEFRAGMENT _IOWR(F2FS_IOCTL_MAGIC, 8, \ 417 struct f2fs_defragment) 418 #define F2FS_IOC_MOVE_RANGE _IOWR(F2FS_IOCTL_MAGIC, 9, \ 419 struct f2fs_move_range) 420 #define F2FS_IOC_FLUSH_DEVICE _IOW(F2FS_IOCTL_MAGIC, 10, \ 421 struct f2fs_flush_device) 422 #define F2FS_IOC_GARBAGE_COLLECT_RANGE _IOW(F2FS_IOCTL_MAGIC, 11, \ 423 struct f2fs_gc_range) 424 #define F2FS_IOC_GET_FEATURES _IOR(F2FS_IOCTL_MAGIC, 12, __u32) 425 #define F2FS_IOC_SET_PIN_FILE _IOW(F2FS_IOCTL_MAGIC, 13, __u32) 426 #define F2FS_IOC_GET_PIN_FILE _IOR(F2FS_IOCTL_MAGIC, 14, __u32) 427 #define F2FS_IOC_PRECACHE_EXTENTS _IO(F2FS_IOCTL_MAGIC, 15) 428 #define F2FS_IOC_RESIZE_FS _IOW(F2FS_IOCTL_MAGIC, 16, __u64) 429 #define F2FS_IOC_GET_COMPRESS_BLOCKS _IOR(F2FS_IOCTL_MAGIC, 17, __u64) 430 #define F2FS_IOC_RELEASE_COMPRESS_BLOCKS \ 431 _IOR(F2FS_IOCTL_MAGIC, 18, __u64) 432 #define F2FS_IOC_RESERVE_COMPRESS_BLOCKS \ 433 _IOR(F2FS_IOCTL_MAGIC, 19, __u64) 434 #define F2FS_IOC_SEC_TRIM_FILE _IOW(F2FS_IOCTL_MAGIC, 20, \ 435 struct f2fs_sectrim_range) 436 437 /* 438 * should be same as XFS_IOC_GOINGDOWN. 439 * Flags for going down operation used by FS_IOC_GOINGDOWN 440 */ 441 #define F2FS_IOC_SHUTDOWN _IOR('X', 125, __u32) /* Shutdown */ 442 #define F2FS_GOING_DOWN_FULLSYNC 0x0 /* going down with full sync */ 443 #define F2FS_GOING_DOWN_METASYNC 0x1 /* going down with metadata */ 444 #define F2FS_GOING_DOWN_NOSYNC 0x2 /* going down */ 445 #define F2FS_GOING_DOWN_METAFLUSH 0x3 /* going down with meta flush */ 446 #define F2FS_GOING_DOWN_NEED_FSCK 0x4 /* going down to trigger fsck */ 447 448 /* 449 * Flags used by F2FS_IOC_SEC_TRIM_FILE 450 */ 451 #define F2FS_TRIM_FILE_DISCARD 0x1 /* send discard command */ 452 #define F2FS_TRIM_FILE_ZEROOUT 0x2 /* zero out */ 453 #define F2FS_TRIM_FILE_MASK 0x3 454 455 struct f2fs_gc_range { 456 u32 sync; 457 u64 start; 458 u64 len; 459 }; 460 461 struct f2fs_defragment { 462 u64 start; 463 u64 len; 464 }; 465 466 struct f2fs_move_range { 467 u32 dst_fd; /* destination fd */ 468 u64 pos_in; /* start position in src_fd */ 469 u64 pos_out; /* start position in dst_fd */ 470 u64 len; /* size to move */ 471 }; 472 473 struct f2fs_flush_device { 474 u32 dev_num; /* device number to flush */ 475 u32 segments; /* # of segments to flush */ 476 }; 477 478 struct f2fs_sectrim_range { 479 u64 start; 480 u64 len; 481 u64 flags; 482 }; 483 484 /* for inline stuff */ 485 #define DEF_INLINE_RESERVED_SIZE 1 486 static inline int get_extra_isize(struct inode *inode); 487 static inline int get_inline_xattr_addrs(struct inode *inode); 488 #define MAX_INLINE_DATA(inode) (sizeof(__le32) * \ 489 (CUR_ADDRS_PER_INODE(inode) - \ 490 get_inline_xattr_addrs(inode) - \ 491 DEF_INLINE_RESERVED_SIZE)) 492 493 /* for inline dir */ 494 #define NR_INLINE_DENTRY(inode) (MAX_INLINE_DATA(inode) * BITS_PER_BYTE / \ 495 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \ 496 BITS_PER_BYTE + 1)) 497 #define INLINE_DENTRY_BITMAP_SIZE(inode) \ 498 DIV_ROUND_UP(NR_INLINE_DENTRY(inode), BITS_PER_BYTE) 499 #define INLINE_RESERVED_SIZE(inode) (MAX_INLINE_DATA(inode) - \ 500 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \ 501 NR_INLINE_DENTRY(inode) + \ 502 INLINE_DENTRY_BITMAP_SIZE(inode))) 503 504 /* 505 * For INODE and NODE manager 506 */ 507 /* for directory operations */ 508 509 struct f2fs_filename { 510 /* 511 * The filename the user specified. This is NULL for some 512 * filesystem-internal operations, e.g. converting an inline directory 513 * to a non-inline one, or roll-forward recovering an encrypted dentry. 514 */ 515 const struct qstr *usr_fname; 516 517 /* 518 * The on-disk filename. For encrypted directories, this is encrypted. 519 * This may be NULL for lookups in an encrypted dir without the key. 520 */ 521 struct fscrypt_str disk_name; 522 523 /* The dirhash of this filename */ 524 f2fs_hash_t hash; 525 526 #ifdef CONFIG_FS_ENCRYPTION 527 /* 528 * For lookups in encrypted directories: either the buffer backing 529 * disk_name, or a buffer that holds the decoded no-key name. 530 */ 531 struct fscrypt_str crypto_buf; 532 #endif 533 #ifdef CONFIG_UNICODE 534 /* 535 * For casefolded directories: the casefolded name, but it's left NULL 536 * if the original name is not valid Unicode or if the filesystem is 537 * doing an internal operation where usr_fname is also NULL. In these 538 * cases we fall back to treating the name as an opaque byte sequence. 539 */ 540 struct fscrypt_str cf_name; 541 #endif 542 }; 543 544 struct f2fs_dentry_ptr { 545 struct inode *inode; 546 void *bitmap; 547 struct f2fs_dir_entry *dentry; 548 __u8 (*filename)[F2FS_SLOT_LEN]; 549 int max; 550 int nr_bitmap; 551 }; 552 553 static inline void make_dentry_ptr_block(struct inode *inode, 554 struct f2fs_dentry_ptr *d, struct f2fs_dentry_block *t) 555 { 556 d->inode = inode; 557 d->max = NR_DENTRY_IN_BLOCK; 558 d->nr_bitmap = SIZE_OF_DENTRY_BITMAP; 559 d->bitmap = t->dentry_bitmap; 560 d->dentry = t->dentry; 561 d->filename = t->filename; 562 } 563 564 static inline void make_dentry_ptr_inline(struct inode *inode, 565 struct f2fs_dentry_ptr *d, void *t) 566 { 567 int entry_cnt = NR_INLINE_DENTRY(inode); 568 int bitmap_size = INLINE_DENTRY_BITMAP_SIZE(inode); 569 int reserved_size = INLINE_RESERVED_SIZE(inode); 570 571 d->inode = inode; 572 d->max = entry_cnt; 573 d->nr_bitmap = bitmap_size; 574 d->bitmap = t; 575 d->dentry = t + bitmap_size + reserved_size; 576 d->filename = t + bitmap_size + reserved_size + 577 SIZE_OF_DIR_ENTRY * entry_cnt; 578 } 579 580 /* 581 * XATTR_NODE_OFFSET stores xattrs to one node block per file keeping -1 582 * as its node offset to distinguish from index node blocks. 583 * But some bits are used to mark the node block. 584 */ 585 #define XATTR_NODE_OFFSET ((((unsigned int)-1) << OFFSET_BIT_SHIFT) \ 586 >> OFFSET_BIT_SHIFT) 587 enum { 588 ALLOC_NODE, /* allocate a new node page if needed */ 589 LOOKUP_NODE, /* look up a node without readahead */ 590 LOOKUP_NODE_RA, /* 591 * look up a node with readahead called 592 * by get_data_block. 593 */ 594 }; 595 596 #define DEFAULT_RETRY_IO_COUNT 8 /* maximum retry read IO count */ 597 598 /* congestion wait timeout value, default: 20ms */ 599 #define DEFAULT_IO_TIMEOUT (msecs_to_jiffies(20)) 600 601 /* maximum retry quota flush count */ 602 #define DEFAULT_RETRY_QUOTA_FLUSH_COUNT 8 603 604 #define F2FS_LINK_MAX 0xffffffff /* maximum link count per file */ 605 606 #define MAX_DIR_RA_PAGES 4 /* maximum ra pages of dir */ 607 608 /* for in-memory extent cache entry */ 609 #define F2FS_MIN_EXTENT_LEN 64 /* minimum extent length */ 610 611 /* number of extent info in extent cache we try to shrink */ 612 #define EXTENT_CACHE_SHRINK_NUMBER 128 613 614 struct rb_entry { 615 struct rb_node rb_node; /* rb node located in rb-tree */ 616 union { 617 struct { 618 unsigned int ofs; /* start offset of the entry */ 619 unsigned int len; /* length of the entry */ 620 }; 621 unsigned long long key; /* 64-bits key */ 622 } __packed; 623 }; 624 625 struct extent_info { 626 unsigned int fofs; /* start offset in a file */ 627 unsigned int len; /* length of the extent */ 628 u32 blk; /* start block address of the extent */ 629 }; 630 631 struct extent_node { 632 struct rb_node rb_node; /* rb node located in rb-tree */ 633 struct extent_info ei; /* extent info */ 634 struct list_head list; /* node in global extent list of sbi */ 635 struct extent_tree *et; /* extent tree pointer */ 636 }; 637 638 struct extent_tree { 639 nid_t ino; /* inode number */ 640 struct rb_root_cached root; /* root of extent info rb-tree */ 641 struct extent_node *cached_en; /* recently accessed extent node */ 642 struct extent_info largest; /* largested extent info */ 643 struct list_head list; /* to be used by sbi->zombie_list */ 644 rwlock_t lock; /* protect extent info rb-tree */ 645 atomic_t node_cnt; /* # of extent node in rb-tree*/ 646 bool largest_updated; /* largest extent updated */ 647 }; 648 649 /* 650 * This structure is taken from ext4_map_blocks. 651 * 652 * Note that, however, f2fs uses NEW and MAPPED flags for f2fs_map_blocks(). 653 */ 654 #define F2FS_MAP_NEW (1 << BH_New) 655 #define F2FS_MAP_MAPPED (1 << BH_Mapped) 656 #define F2FS_MAP_UNWRITTEN (1 << BH_Unwritten) 657 #define F2FS_MAP_FLAGS (F2FS_MAP_NEW | F2FS_MAP_MAPPED |\ 658 F2FS_MAP_UNWRITTEN) 659 660 struct f2fs_map_blocks { 661 block_t m_pblk; 662 block_t m_lblk; 663 unsigned int m_len; 664 unsigned int m_flags; 665 pgoff_t *m_next_pgofs; /* point next possible non-hole pgofs */ 666 pgoff_t *m_next_extent; /* point to next possible extent */ 667 int m_seg_type; 668 bool m_may_create; /* indicate it is from write path */ 669 }; 670 671 /* for flag in get_data_block */ 672 enum { 673 F2FS_GET_BLOCK_DEFAULT, 674 F2FS_GET_BLOCK_FIEMAP, 675 F2FS_GET_BLOCK_BMAP, 676 F2FS_GET_BLOCK_DIO, 677 F2FS_GET_BLOCK_PRE_DIO, 678 F2FS_GET_BLOCK_PRE_AIO, 679 F2FS_GET_BLOCK_PRECACHE, 680 }; 681 682 /* 683 * i_advise uses FADVISE_XXX_BIT. We can add additional hints later. 684 */ 685 #define FADVISE_COLD_BIT 0x01 686 #define FADVISE_LOST_PINO_BIT 0x02 687 #define FADVISE_ENCRYPT_BIT 0x04 688 #define FADVISE_ENC_NAME_BIT 0x08 689 #define FADVISE_KEEP_SIZE_BIT 0x10 690 #define FADVISE_HOT_BIT 0x20 691 #define FADVISE_VERITY_BIT 0x40 692 693 #define FADVISE_MODIFIABLE_BITS (FADVISE_COLD_BIT | FADVISE_HOT_BIT) 694 695 #define file_is_cold(inode) is_file(inode, FADVISE_COLD_BIT) 696 #define file_wrong_pino(inode) is_file(inode, FADVISE_LOST_PINO_BIT) 697 #define file_set_cold(inode) set_file(inode, FADVISE_COLD_BIT) 698 #define file_lost_pino(inode) set_file(inode, FADVISE_LOST_PINO_BIT) 699 #define file_clear_cold(inode) clear_file(inode, FADVISE_COLD_BIT) 700 #define file_got_pino(inode) clear_file(inode, FADVISE_LOST_PINO_BIT) 701 #define file_is_encrypt(inode) is_file(inode, FADVISE_ENCRYPT_BIT) 702 #define file_set_encrypt(inode) set_file(inode, FADVISE_ENCRYPT_BIT) 703 #define file_clear_encrypt(inode) clear_file(inode, FADVISE_ENCRYPT_BIT) 704 #define file_enc_name(inode) is_file(inode, FADVISE_ENC_NAME_BIT) 705 #define file_set_enc_name(inode) set_file(inode, FADVISE_ENC_NAME_BIT) 706 #define file_keep_isize(inode) is_file(inode, FADVISE_KEEP_SIZE_BIT) 707 #define file_set_keep_isize(inode) set_file(inode, FADVISE_KEEP_SIZE_BIT) 708 #define file_is_hot(inode) is_file(inode, FADVISE_HOT_BIT) 709 #define file_set_hot(inode) set_file(inode, FADVISE_HOT_BIT) 710 #define file_clear_hot(inode) clear_file(inode, FADVISE_HOT_BIT) 711 #define file_is_verity(inode) is_file(inode, FADVISE_VERITY_BIT) 712 #define file_set_verity(inode) set_file(inode, FADVISE_VERITY_BIT) 713 714 #define DEF_DIR_LEVEL 0 715 716 enum { 717 GC_FAILURE_PIN, 718 GC_FAILURE_ATOMIC, 719 MAX_GC_FAILURE 720 }; 721 722 /* used for f2fs_inode_info->flags */ 723 enum { 724 FI_NEW_INODE, /* indicate newly allocated inode */ 725 FI_DIRTY_INODE, /* indicate inode is dirty or not */ 726 FI_AUTO_RECOVER, /* indicate inode is recoverable */ 727 FI_DIRTY_DIR, /* indicate directory has dirty pages */ 728 FI_INC_LINK, /* need to increment i_nlink */ 729 FI_ACL_MODE, /* indicate acl mode */ 730 FI_NO_ALLOC, /* should not allocate any blocks */ 731 FI_FREE_NID, /* free allocated nide */ 732 FI_NO_EXTENT, /* not to use the extent cache */ 733 FI_INLINE_XATTR, /* used for inline xattr */ 734 FI_INLINE_DATA, /* used for inline data*/ 735 FI_INLINE_DENTRY, /* used for inline dentry */ 736 FI_APPEND_WRITE, /* inode has appended data */ 737 FI_UPDATE_WRITE, /* inode has in-place-update data */ 738 FI_NEED_IPU, /* used for ipu per file */ 739 FI_ATOMIC_FILE, /* indicate atomic file */ 740 FI_ATOMIC_COMMIT, /* indicate the state of atomical committing */ 741 FI_VOLATILE_FILE, /* indicate volatile file */ 742 FI_FIRST_BLOCK_WRITTEN, /* indicate #0 data block was written */ 743 FI_DROP_CACHE, /* drop dirty page cache */ 744 FI_DATA_EXIST, /* indicate data exists */ 745 FI_INLINE_DOTS, /* indicate inline dot dentries */ 746 FI_DO_DEFRAG, /* indicate defragment is running */ 747 FI_DIRTY_FILE, /* indicate regular/symlink has dirty pages */ 748 FI_NO_PREALLOC, /* indicate skipped preallocated blocks */ 749 FI_HOT_DATA, /* indicate file is hot */ 750 FI_EXTRA_ATTR, /* indicate file has extra attribute */ 751 FI_PROJ_INHERIT, /* indicate file inherits projectid */ 752 FI_PIN_FILE, /* indicate file should not be gced */ 753 FI_ATOMIC_REVOKE_REQUEST, /* request to drop atomic data */ 754 FI_VERITY_IN_PROGRESS, /* building fs-verity Merkle tree */ 755 FI_COMPRESSED_FILE, /* indicate file's data can be compressed */ 756 FI_MMAP_FILE, /* indicate file was mmapped */ 757 FI_MAX, /* max flag, never be used */ 758 }; 759 760 struct f2fs_inode_info { 761 struct inode vfs_inode; /* serve a vfs inode */ 762 unsigned long i_flags; /* keep an inode flags for ioctl */ 763 unsigned char i_advise; /* use to give file attribute hints */ 764 unsigned char i_dir_level; /* use for dentry level for large dir */ 765 unsigned int i_current_depth; /* only for directory depth */ 766 /* for gc failure statistic */ 767 unsigned int i_gc_failures[MAX_GC_FAILURE]; 768 unsigned int i_pino; /* parent inode number */ 769 umode_t i_acl_mode; /* keep file acl mode temporarily */ 770 771 /* Use below internally in f2fs*/ 772 unsigned long flags[BITS_TO_LONGS(FI_MAX)]; /* use to pass per-file flags */ 773 struct rw_semaphore i_sem; /* protect fi info */ 774 atomic_t dirty_pages; /* # of dirty pages */ 775 f2fs_hash_t chash; /* hash value of given file name */ 776 unsigned int clevel; /* maximum level of given file name */ 777 struct task_struct *task; /* lookup and create consistency */ 778 struct task_struct *cp_task; /* separate cp/wb IO stats*/ 779 nid_t i_xattr_nid; /* node id that contains xattrs */ 780 loff_t last_disk_size; /* lastly written file size */ 781 spinlock_t i_size_lock; /* protect last_disk_size */ 782 783 #ifdef CONFIG_QUOTA 784 struct dquot *i_dquot[MAXQUOTAS]; 785 786 /* quota space reservation, managed internally by quota code */ 787 qsize_t i_reserved_quota; 788 #endif 789 struct list_head dirty_list; /* dirty list for dirs and files */ 790 struct list_head gdirty_list; /* linked in global dirty list */ 791 struct list_head inmem_ilist; /* list for inmem inodes */ 792 struct list_head inmem_pages; /* inmemory pages managed by f2fs */ 793 struct task_struct *inmem_task; /* store inmemory task */ 794 struct mutex inmem_lock; /* lock for inmemory pages */ 795 pgoff_t ra_offset; /* ongoing readahead offset */ 796 struct extent_tree *extent_tree; /* cached extent_tree entry */ 797 798 /* avoid racing between foreground op and gc */ 799 struct rw_semaphore i_gc_rwsem[2]; 800 struct rw_semaphore i_mmap_sem; 801 struct rw_semaphore i_xattr_sem; /* avoid racing between reading and changing EAs */ 802 803 int i_extra_isize; /* size of extra space located in i_addr */ 804 kprojid_t i_projid; /* id for project quota */ 805 int i_inline_xattr_size; /* inline xattr size */ 806 struct timespec64 i_crtime; /* inode creation time */ 807 struct timespec64 i_disk_time[4];/* inode disk times */ 808 809 /* for file compress */ 810 atomic_t i_compr_blocks; /* # of compressed blocks */ 811 unsigned char i_compress_algorithm; /* algorithm type */ 812 unsigned char i_log_cluster_size; /* log of cluster size */ 813 unsigned int i_cluster_size; /* cluster size */ 814 }; 815 816 static inline void get_extent_info(struct extent_info *ext, 817 struct f2fs_extent *i_ext) 818 { 819 ext->fofs = le32_to_cpu(i_ext->fofs); 820 ext->blk = le32_to_cpu(i_ext->blk); 821 ext->len = le32_to_cpu(i_ext->len); 822 } 823 824 static inline void set_raw_extent(struct extent_info *ext, 825 struct f2fs_extent *i_ext) 826 { 827 i_ext->fofs = cpu_to_le32(ext->fofs); 828 i_ext->blk = cpu_to_le32(ext->blk); 829 i_ext->len = cpu_to_le32(ext->len); 830 } 831 832 static inline void set_extent_info(struct extent_info *ei, unsigned int fofs, 833 u32 blk, unsigned int len) 834 { 835 ei->fofs = fofs; 836 ei->blk = blk; 837 ei->len = len; 838 } 839 840 static inline bool __is_discard_mergeable(struct discard_info *back, 841 struct discard_info *front, unsigned int max_len) 842 { 843 return (back->lstart + back->len == front->lstart) && 844 (back->len + front->len <= max_len); 845 } 846 847 static inline bool __is_discard_back_mergeable(struct discard_info *cur, 848 struct discard_info *back, unsigned int max_len) 849 { 850 return __is_discard_mergeable(back, cur, max_len); 851 } 852 853 static inline bool __is_discard_front_mergeable(struct discard_info *cur, 854 struct discard_info *front, unsigned int max_len) 855 { 856 return __is_discard_mergeable(cur, front, max_len); 857 } 858 859 static inline bool __is_extent_mergeable(struct extent_info *back, 860 struct extent_info *front) 861 { 862 return (back->fofs + back->len == front->fofs && 863 back->blk + back->len == front->blk); 864 } 865 866 static inline bool __is_back_mergeable(struct extent_info *cur, 867 struct extent_info *back) 868 { 869 return __is_extent_mergeable(back, cur); 870 } 871 872 static inline bool __is_front_mergeable(struct extent_info *cur, 873 struct extent_info *front) 874 { 875 return __is_extent_mergeable(cur, front); 876 } 877 878 extern void f2fs_mark_inode_dirty_sync(struct inode *inode, bool sync); 879 static inline void __try_update_largest_extent(struct extent_tree *et, 880 struct extent_node *en) 881 { 882 if (en->ei.len > et->largest.len) { 883 et->largest = en->ei; 884 et->largest_updated = true; 885 } 886 } 887 888 /* 889 * For free nid management 890 */ 891 enum nid_state { 892 FREE_NID, /* newly added to free nid list */ 893 PREALLOC_NID, /* it is preallocated */ 894 MAX_NID_STATE, 895 }; 896 897 struct f2fs_nm_info { 898 block_t nat_blkaddr; /* base disk address of NAT */ 899 nid_t max_nid; /* maximum possible node ids */ 900 nid_t available_nids; /* # of available node ids */ 901 nid_t next_scan_nid; /* the next nid to be scanned */ 902 unsigned int ram_thresh; /* control the memory footprint */ 903 unsigned int ra_nid_pages; /* # of nid pages to be readaheaded */ 904 unsigned int dirty_nats_ratio; /* control dirty nats ratio threshold */ 905 906 /* NAT cache management */ 907 struct radix_tree_root nat_root;/* root of the nat entry cache */ 908 struct radix_tree_root nat_set_root;/* root of the nat set cache */ 909 struct rw_semaphore nat_tree_lock; /* protect nat_tree_lock */ 910 struct list_head nat_entries; /* cached nat entry list (clean) */ 911 spinlock_t nat_list_lock; /* protect clean nat entry list */ 912 unsigned int nat_cnt; /* the # of cached nat entries */ 913 unsigned int dirty_nat_cnt; /* total num of nat entries in set */ 914 unsigned int nat_blocks; /* # of nat blocks */ 915 916 /* free node ids management */ 917 struct radix_tree_root free_nid_root;/* root of the free_nid cache */ 918 struct list_head free_nid_list; /* list for free nids excluding preallocated nids */ 919 unsigned int nid_cnt[MAX_NID_STATE]; /* the number of free node id */ 920 spinlock_t nid_list_lock; /* protect nid lists ops */ 921 struct mutex build_lock; /* lock for build free nids */ 922 unsigned char **free_nid_bitmap; 923 unsigned char *nat_block_bitmap; 924 unsigned short *free_nid_count; /* free nid count of NAT block */ 925 926 /* for checkpoint */ 927 char *nat_bitmap; /* NAT bitmap pointer */ 928 929 unsigned int nat_bits_blocks; /* # of nat bits blocks */ 930 unsigned char *nat_bits; /* NAT bits blocks */ 931 unsigned char *full_nat_bits; /* full NAT pages */ 932 unsigned char *empty_nat_bits; /* empty NAT pages */ 933 #ifdef CONFIG_F2FS_CHECK_FS 934 char *nat_bitmap_mir; /* NAT bitmap mirror */ 935 #endif 936 int bitmap_size; /* bitmap size */ 937 }; 938 939 /* 940 * this structure is used as one of function parameters. 941 * all the information are dedicated to a given direct node block determined 942 * by the data offset in a file. 943 */ 944 struct dnode_of_data { 945 struct inode *inode; /* vfs inode pointer */ 946 struct page *inode_page; /* its inode page, NULL is possible */ 947 struct page *node_page; /* cached direct node page */ 948 nid_t nid; /* node id of the direct node block */ 949 unsigned int ofs_in_node; /* data offset in the node page */ 950 bool inode_page_locked; /* inode page is locked or not */ 951 bool node_changed; /* is node block changed */ 952 char cur_level; /* level of hole node page */ 953 char max_level; /* level of current page located */ 954 block_t data_blkaddr; /* block address of the node block */ 955 }; 956 957 static inline void set_new_dnode(struct dnode_of_data *dn, struct inode *inode, 958 struct page *ipage, struct page *npage, nid_t nid) 959 { 960 memset(dn, 0, sizeof(*dn)); 961 dn->inode = inode; 962 dn->inode_page = ipage; 963 dn->node_page = npage; 964 dn->nid = nid; 965 } 966 967 /* 968 * For SIT manager 969 * 970 * By default, there are 6 active log areas across the whole main area. 971 * When considering hot and cold data separation to reduce cleaning overhead, 972 * we split 3 for data logs and 3 for node logs as hot, warm, and cold types, 973 * respectively. 974 * In the current design, you should not change the numbers intentionally. 975 * Instead, as a mount option such as active_logs=x, you can use 2, 4, and 6 976 * logs individually according to the underlying devices. (default: 6) 977 * Just in case, on-disk layout covers maximum 16 logs that consist of 8 for 978 * data and 8 for node logs. 979 */ 980 #define NR_CURSEG_DATA_TYPE (3) 981 #define NR_CURSEG_NODE_TYPE (3) 982 #define NR_CURSEG_INMEM_TYPE (2) 983 #define NR_CURSEG_PERSIST_TYPE (NR_CURSEG_DATA_TYPE + NR_CURSEG_NODE_TYPE) 984 #define NR_CURSEG_TYPE (NR_CURSEG_INMEM_TYPE + NR_CURSEG_PERSIST_TYPE) 985 986 enum { 987 CURSEG_HOT_DATA = 0, /* directory entry blocks */ 988 CURSEG_WARM_DATA, /* data blocks */ 989 CURSEG_COLD_DATA, /* multimedia or GCed data blocks */ 990 CURSEG_HOT_NODE, /* direct node blocks of directory files */ 991 CURSEG_WARM_NODE, /* direct node blocks of normal files */ 992 CURSEG_COLD_NODE, /* indirect node blocks */ 993 NR_PERSISTENT_LOG, /* number of persistent log */ 994 CURSEG_COLD_DATA_PINNED = NR_PERSISTENT_LOG, 995 /* pinned file that needs consecutive block address */ 996 CURSEG_ALL_DATA_ATGC, /* SSR alloctor in hot/warm/cold data area */ 997 NO_CHECK_TYPE, /* number of persistent & inmem log */ 998 }; 999 1000 struct flush_cmd { 1001 struct completion wait; 1002 struct llist_node llnode; 1003 nid_t ino; 1004 int ret; 1005 }; 1006 1007 struct flush_cmd_control { 1008 struct task_struct *f2fs_issue_flush; /* flush thread */ 1009 wait_queue_head_t flush_wait_queue; /* waiting queue for wake-up */ 1010 atomic_t issued_flush; /* # of issued flushes */ 1011 atomic_t queued_flush; /* # of queued flushes */ 1012 struct llist_head issue_list; /* list for command issue */ 1013 struct llist_node *dispatch_list; /* list for command dispatch */ 1014 }; 1015 1016 struct f2fs_sm_info { 1017 struct sit_info *sit_info; /* whole segment information */ 1018 struct free_segmap_info *free_info; /* free segment information */ 1019 struct dirty_seglist_info *dirty_info; /* dirty segment information */ 1020 struct curseg_info *curseg_array; /* active segment information */ 1021 1022 struct rw_semaphore curseg_lock; /* for preventing curseg change */ 1023 1024 block_t seg0_blkaddr; /* block address of 0'th segment */ 1025 block_t main_blkaddr; /* start block address of main area */ 1026 block_t ssa_blkaddr; /* start block address of SSA area */ 1027 1028 unsigned int segment_count; /* total # of segments */ 1029 unsigned int main_segments; /* # of segments in main area */ 1030 unsigned int reserved_segments; /* # of reserved segments */ 1031 unsigned int ovp_segments; /* # of overprovision segments */ 1032 1033 /* a threshold to reclaim prefree segments */ 1034 unsigned int rec_prefree_segments; 1035 1036 /* for batched trimming */ 1037 unsigned int trim_sections; /* # of sections to trim */ 1038 1039 struct list_head sit_entry_set; /* sit entry set list */ 1040 1041 unsigned int ipu_policy; /* in-place-update policy */ 1042 unsigned int min_ipu_util; /* in-place-update threshold */ 1043 unsigned int min_fsync_blocks; /* threshold for fsync */ 1044 unsigned int min_seq_blocks; /* threshold for sequential blocks */ 1045 unsigned int min_hot_blocks; /* threshold for hot block allocation */ 1046 unsigned int min_ssr_sections; /* threshold to trigger SSR allocation */ 1047 1048 /* for flush command control */ 1049 struct flush_cmd_control *fcc_info; 1050 1051 /* for discard command control */ 1052 struct discard_cmd_control *dcc_info; 1053 }; 1054 1055 /* 1056 * For superblock 1057 */ 1058 /* 1059 * COUNT_TYPE for monitoring 1060 * 1061 * f2fs monitors the number of several block types such as on-writeback, 1062 * dirty dentry blocks, dirty node blocks, and dirty meta blocks. 1063 */ 1064 #define WB_DATA_TYPE(p) (__is_cp_guaranteed(p) ? F2FS_WB_CP_DATA : F2FS_WB_DATA) 1065 enum count_type { 1066 F2FS_DIRTY_DENTS, 1067 F2FS_DIRTY_DATA, 1068 F2FS_DIRTY_QDATA, 1069 F2FS_DIRTY_NODES, 1070 F2FS_DIRTY_META, 1071 F2FS_INMEM_PAGES, 1072 F2FS_DIRTY_IMETA, 1073 F2FS_WB_CP_DATA, 1074 F2FS_WB_DATA, 1075 F2FS_RD_DATA, 1076 F2FS_RD_NODE, 1077 F2FS_RD_META, 1078 F2FS_DIO_WRITE, 1079 F2FS_DIO_READ, 1080 NR_COUNT_TYPE, 1081 }; 1082 1083 /* 1084 * The below are the page types of bios used in submit_bio(). 1085 * The available types are: 1086 * DATA User data pages. It operates as async mode. 1087 * NODE Node pages. It operates as async mode. 1088 * META FS metadata pages such as SIT, NAT, CP. 1089 * NR_PAGE_TYPE The number of page types. 1090 * META_FLUSH Make sure the previous pages are written 1091 * with waiting the bio's completion 1092 * ... Only can be used with META. 1093 */ 1094 #define PAGE_TYPE_OF_BIO(type) ((type) > META ? META : (type)) 1095 enum page_type { 1096 DATA, 1097 NODE, 1098 META, 1099 NR_PAGE_TYPE, 1100 META_FLUSH, 1101 INMEM, /* the below types are used by tracepoints only. */ 1102 INMEM_DROP, 1103 INMEM_INVALIDATE, 1104 INMEM_REVOKE, 1105 IPU, 1106 OPU, 1107 }; 1108 1109 enum temp_type { 1110 HOT = 0, /* must be zero for meta bio */ 1111 WARM, 1112 COLD, 1113 NR_TEMP_TYPE, 1114 }; 1115 1116 enum need_lock_type { 1117 LOCK_REQ = 0, 1118 LOCK_DONE, 1119 LOCK_RETRY, 1120 }; 1121 1122 enum cp_reason_type { 1123 CP_NO_NEEDED, 1124 CP_NON_REGULAR, 1125 CP_COMPRESSED, 1126 CP_HARDLINK, 1127 CP_SB_NEED_CP, 1128 CP_WRONG_PINO, 1129 CP_NO_SPC_ROLL, 1130 CP_NODE_NEED_CP, 1131 CP_FASTBOOT_MODE, 1132 CP_SPEC_LOG_NUM, 1133 CP_RECOVER_DIR, 1134 }; 1135 1136 enum iostat_type { 1137 /* WRITE IO */ 1138 APP_DIRECT_IO, /* app direct write IOs */ 1139 APP_BUFFERED_IO, /* app buffered write IOs */ 1140 APP_WRITE_IO, /* app write IOs */ 1141 APP_MAPPED_IO, /* app mapped IOs */ 1142 FS_DATA_IO, /* data IOs from kworker/fsync/reclaimer */ 1143 FS_NODE_IO, /* node IOs from kworker/fsync/reclaimer */ 1144 FS_META_IO, /* meta IOs from kworker/reclaimer */ 1145 FS_GC_DATA_IO, /* data IOs from forground gc */ 1146 FS_GC_NODE_IO, /* node IOs from forground gc */ 1147 FS_CP_DATA_IO, /* data IOs from checkpoint */ 1148 FS_CP_NODE_IO, /* node IOs from checkpoint */ 1149 FS_CP_META_IO, /* meta IOs from checkpoint */ 1150 1151 /* READ IO */ 1152 APP_DIRECT_READ_IO, /* app direct read IOs */ 1153 APP_BUFFERED_READ_IO, /* app buffered read IOs */ 1154 APP_READ_IO, /* app read IOs */ 1155 APP_MAPPED_READ_IO, /* app mapped read IOs */ 1156 FS_DATA_READ_IO, /* data read IOs */ 1157 FS_GDATA_READ_IO, /* data read IOs from background gc */ 1158 FS_CDATA_READ_IO, /* compressed data read IOs */ 1159 FS_NODE_READ_IO, /* node read IOs */ 1160 FS_META_READ_IO, /* meta read IOs */ 1161 1162 /* other */ 1163 FS_DISCARD, /* discard */ 1164 NR_IO_TYPE, 1165 }; 1166 1167 struct f2fs_io_info { 1168 struct f2fs_sb_info *sbi; /* f2fs_sb_info pointer */ 1169 nid_t ino; /* inode number */ 1170 enum page_type type; /* contains DATA/NODE/META/META_FLUSH */ 1171 enum temp_type temp; /* contains HOT/WARM/COLD */ 1172 int op; /* contains REQ_OP_ */ 1173 int op_flags; /* req_flag_bits */ 1174 block_t new_blkaddr; /* new block address to be written */ 1175 block_t old_blkaddr; /* old block address before Cow */ 1176 struct page *page; /* page to be written */ 1177 struct page *encrypted_page; /* encrypted page */ 1178 struct page *compressed_page; /* compressed page */ 1179 struct list_head list; /* serialize IOs */ 1180 bool submitted; /* indicate IO submission */ 1181 int need_lock; /* indicate we need to lock cp_rwsem */ 1182 bool in_list; /* indicate fio is in io_list */ 1183 bool is_por; /* indicate IO is from recovery or not */ 1184 bool retry; /* need to reallocate block address */ 1185 int compr_blocks; /* # of compressed block addresses */ 1186 bool encrypted; /* indicate file is encrypted */ 1187 enum iostat_type io_type; /* io type */ 1188 struct writeback_control *io_wbc; /* writeback control */ 1189 struct bio **bio; /* bio for ipu */ 1190 sector_t *last_block; /* last block number in bio */ 1191 unsigned char version; /* version of the node */ 1192 }; 1193 1194 struct bio_entry { 1195 struct bio *bio; 1196 struct list_head list; 1197 }; 1198 1199 #define is_read_io(rw) ((rw) == READ) 1200 struct f2fs_bio_info { 1201 struct f2fs_sb_info *sbi; /* f2fs superblock */ 1202 struct bio *bio; /* bios to merge */ 1203 sector_t last_block_in_bio; /* last block number */ 1204 struct f2fs_io_info fio; /* store buffered io info. */ 1205 struct rw_semaphore io_rwsem; /* blocking op for bio */ 1206 spinlock_t io_lock; /* serialize DATA/NODE IOs */ 1207 struct list_head io_list; /* track fios */ 1208 struct list_head bio_list; /* bio entry list head */ 1209 struct rw_semaphore bio_list_lock; /* lock to protect bio entry list */ 1210 }; 1211 1212 #define FDEV(i) (sbi->devs[i]) 1213 #define RDEV(i) (raw_super->devs[i]) 1214 struct f2fs_dev_info { 1215 struct block_device *bdev; 1216 char path[MAX_PATH_LEN]; 1217 unsigned int total_segments; 1218 block_t start_blk; 1219 block_t end_blk; 1220 #ifdef CONFIG_BLK_DEV_ZONED 1221 unsigned int nr_blkz; /* Total number of zones */ 1222 unsigned long *blkz_seq; /* Bitmap indicating sequential zones */ 1223 block_t *zone_capacity_blocks; /* Array of zone capacity in blks */ 1224 #endif 1225 }; 1226 1227 enum inode_type { 1228 DIR_INODE, /* for dirty dir inode */ 1229 FILE_INODE, /* for dirty regular/symlink inode */ 1230 DIRTY_META, /* for all dirtied inode metadata */ 1231 ATOMIC_FILE, /* for all atomic files */ 1232 NR_INODE_TYPE, 1233 }; 1234 1235 /* for inner inode cache management */ 1236 struct inode_management { 1237 struct radix_tree_root ino_root; /* ino entry array */ 1238 spinlock_t ino_lock; /* for ino entry lock */ 1239 struct list_head ino_list; /* inode list head */ 1240 unsigned long ino_num; /* number of entries */ 1241 }; 1242 1243 /* for GC_AT */ 1244 struct atgc_management { 1245 bool atgc_enabled; /* ATGC is enabled or not */ 1246 struct rb_root_cached root; /* root of victim rb-tree */ 1247 struct list_head victim_list; /* linked with all victim entries */ 1248 unsigned int victim_count; /* victim count in rb-tree */ 1249 unsigned int candidate_ratio; /* candidate ratio */ 1250 unsigned int max_candidate_count; /* max candidate count */ 1251 unsigned int age_weight; /* age weight, vblock_weight = 100 - age_weight */ 1252 unsigned long long age_threshold; /* age threshold */ 1253 }; 1254 1255 /* For s_flag in struct f2fs_sb_info */ 1256 enum { 1257 SBI_IS_DIRTY, /* dirty flag for checkpoint */ 1258 SBI_IS_CLOSE, /* specify unmounting */ 1259 SBI_NEED_FSCK, /* need fsck.f2fs to fix */ 1260 SBI_POR_DOING, /* recovery is doing or not */ 1261 SBI_NEED_SB_WRITE, /* need to recover superblock */ 1262 SBI_NEED_CP, /* need to checkpoint */ 1263 SBI_IS_SHUTDOWN, /* shutdown by ioctl */ 1264 SBI_IS_RECOVERED, /* recovered orphan/data */ 1265 SBI_CP_DISABLED, /* CP was disabled last mount */ 1266 SBI_CP_DISABLED_QUICK, /* CP was disabled quickly */ 1267 SBI_QUOTA_NEED_FLUSH, /* need to flush quota info in CP */ 1268 SBI_QUOTA_SKIP_FLUSH, /* skip flushing quota in current CP */ 1269 SBI_QUOTA_NEED_REPAIR, /* quota file may be corrupted */ 1270 SBI_IS_RESIZEFS, /* resizefs is in process */ 1271 }; 1272 1273 enum { 1274 CP_TIME, 1275 REQ_TIME, 1276 DISCARD_TIME, 1277 GC_TIME, 1278 DISABLE_TIME, 1279 UMOUNT_DISCARD_TIMEOUT, 1280 MAX_TIME, 1281 }; 1282 1283 enum { 1284 GC_NORMAL, 1285 GC_IDLE_CB, 1286 GC_IDLE_GREEDY, 1287 GC_IDLE_AT, 1288 GC_URGENT_HIGH, 1289 GC_URGENT_LOW, 1290 }; 1291 1292 enum { 1293 BGGC_MODE_ON, /* background gc is on */ 1294 BGGC_MODE_OFF, /* background gc is off */ 1295 BGGC_MODE_SYNC, /* 1296 * background gc is on, migrating blocks 1297 * like foreground gc 1298 */ 1299 }; 1300 1301 enum { 1302 FS_MODE_ADAPTIVE, /* use both lfs/ssr allocation */ 1303 FS_MODE_LFS, /* use lfs allocation only */ 1304 }; 1305 1306 enum { 1307 WHINT_MODE_OFF, /* not pass down write hints */ 1308 WHINT_MODE_USER, /* try to pass down hints given by users */ 1309 WHINT_MODE_FS, /* pass down hints with F2FS policy */ 1310 }; 1311 1312 enum { 1313 ALLOC_MODE_DEFAULT, /* stay default */ 1314 ALLOC_MODE_REUSE, /* reuse segments as much as possible */ 1315 }; 1316 1317 enum fsync_mode { 1318 FSYNC_MODE_POSIX, /* fsync follows posix semantics */ 1319 FSYNC_MODE_STRICT, /* fsync behaves in line with ext4 */ 1320 FSYNC_MODE_NOBARRIER, /* fsync behaves nobarrier based on posix */ 1321 }; 1322 1323 /* 1324 * this value is set in page as a private data which indicate that 1325 * the page is atomically written, and it is in inmem_pages list. 1326 */ 1327 #define ATOMIC_WRITTEN_PAGE ((unsigned long)-1) 1328 #define DUMMY_WRITTEN_PAGE ((unsigned long)-2) 1329 1330 #define IS_ATOMIC_WRITTEN_PAGE(page) \ 1331 (page_private(page) == ATOMIC_WRITTEN_PAGE) 1332 #define IS_DUMMY_WRITTEN_PAGE(page) \ 1333 (page_private(page) == DUMMY_WRITTEN_PAGE) 1334 1335 #ifdef CONFIG_F2FS_IO_TRACE 1336 #define IS_IO_TRACED_PAGE(page) \ 1337 (page_private(page) > 0 && \ 1338 page_private(page) < (unsigned long)PID_MAX_LIMIT) 1339 #else 1340 #define IS_IO_TRACED_PAGE(page) (0) 1341 #endif 1342 1343 /* For compression */ 1344 enum compress_algorithm_type { 1345 COMPRESS_LZO, 1346 COMPRESS_LZ4, 1347 COMPRESS_ZSTD, 1348 COMPRESS_LZORLE, 1349 COMPRESS_MAX, 1350 }; 1351 1352 #define COMPRESS_DATA_RESERVED_SIZE 5 1353 struct compress_data { 1354 __le32 clen; /* compressed data size */ 1355 __le32 reserved[COMPRESS_DATA_RESERVED_SIZE]; /* reserved */ 1356 u8 cdata[]; /* compressed data */ 1357 }; 1358 1359 #define COMPRESS_HEADER_SIZE (sizeof(struct compress_data)) 1360 1361 #define F2FS_COMPRESSED_PAGE_MAGIC 0xF5F2C000 1362 1363 /* compress context */ 1364 struct compress_ctx { 1365 struct inode *inode; /* inode the context belong to */ 1366 pgoff_t cluster_idx; /* cluster index number */ 1367 unsigned int cluster_size; /* page count in cluster */ 1368 unsigned int log_cluster_size; /* log of cluster size */ 1369 struct page **rpages; /* pages store raw data in cluster */ 1370 unsigned int nr_rpages; /* total page number in rpages */ 1371 struct page **cpages; /* pages store compressed data in cluster */ 1372 unsigned int nr_cpages; /* total page number in cpages */ 1373 void *rbuf; /* virtual mapped address on rpages */ 1374 struct compress_data *cbuf; /* virtual mapped address on cpages */ 1375 size_t rlen; /* valid data length in rbuf */ 1376 size_t clen; /* valid data length in cbuf */ 1377 void *private; /* payload buffer for specified compression algorithm */ 1378 void *private2; /* extra payload buffer */ 1379 }; 1380 1381 /* compress context for write IO path */ 1382 struct compress_io_ctx { 1383 u32 magic; /* magic number to indicate page is compressed */ 1384 struct inode *inode; /* inode the context belong to */ 1385 struct page **rpages; /* pages store raw data in cluster */ 1386 unsigned int nr_rpages; /* total page number in rpages */ 1387 atomic_t pending_pages; /* in-flight compressed page count */ 1388 }; 1389 1390 /* decompress io context for read IO path */ 1391 struct decompress_io_ctx { 1392 u32 magic; /* magic number to indicate page is compressed */ 1393 struct inode *inode; /* inode the context belong to */ 1394 pgoff_t cluster_idx; /* cluster index number */ 1395 unsigned int cluster_size; /* page count in cluster */ 1396 unsigned int log_cluster_size; /* log of cluster size */ 1397 struct page **rpages; /* pages store raw data in cluster */ 1398 unsigned int nr_rpages; /* total page number in rpages */ 1399 struct page **cpages; /* pages store compressed data in cluster */ 1400 unsigned int nr_cpages; /* total page number in cpages */ 1401 struct page **tpages; /* temp pages to pad holes in cluster */ 1402 void *rbuf; /* virtual mapped address on rpages */ 1403 struct compress_data *cbuf; /* virtual mapped address on cpages */ 1404 size_t rlen; /* valid data length in rbuf */ 1405 size_t clen; /* valid data length in cbuf */ 1406 atomic_t pending_pages; /* in-flight compressed page count */ 1407 bool failed; /* indicate IO error during decompression */ 1408 void *private; /* payload buffer for specified decompression algorithm */ 1409 void *private2; /* extra payload buffer */ 1410 }; 1411 1412 #define NULL_CLUSTER ((unsigned int)(~0)) 1413 #define MIN_COMPRESS_LOG_SIZE 2 1414 #define MAX_COMPRESS_LOG_SIZE 8 1415 #define MAX_COMPRESS_WINDOW_SIZE(log_size) ((PAGE_SIZE) << (log_size)) 1416 1417 struct f2fs_sb_info { 1418 struct super_block *sb; /* pointer to VFS super block */ 1419 struct proc_dir_entry *s_proc; /* proc entry */ 1420 struct f2fs_super_block *raw_super; /* raw super block pointer */ 1421 struct rw_semaphore sb_lock; /* lock for raw super block */ 1422 int valid_super_block; /* valid super block no */ 1423 unsigned long s_flag; /* flags for sbi */ 1424 struct mutex writepages; /* mutex for writepages() */ 1425 1426 #ifdef CONFIG_BLK_DEV_ZONED 1427 unsigned int blocks_per_blkz; /* F2FS blocks per zone */ 1428 unsigned int log_blocks_per_blkz; /* log2 F2FS blocks per zone */ 1429 #endif 1430 1431 /* for node-related operations */ 1432 struct f2fs_nm_info *nm_info; /* node manager */ 1433 struct inode *node_inode; /* cache node blocks */ 1434 1435 /* for segment-related operations */ 1436 struct f2fs_sm_info *sm_info; /* segment manager */ 1437 1438 /* for bio operations */ 1439 struct f2fs_bio_info *write_io[NR_PAGE_TYPE]; /* for write bios */ 1440 /* keep migration IO order for LFS mode */ 1441 struct rw_semaphore io_order_lock; 1442 mempool_t *write_io_dummy; /* Dummy pages */ 1443 1444 /* for checkpoint */ 1445 struct f2fs_checkpoint *ckpt; /* raw checkpoint pointer */ 1446 int cur_cp_pack; /* remain current cp pack */ 1447 spinlock_t cp_lock; /* for flag in ckpt */ 1448 struct inode *meta_inode; /* cache meta blocks */ 1449 struct mutex cp_mutex; /* checkpoint procedure lock */ 1450 struct rw_semaphore cp_rwsem; /* blocking FS operations */ 1451 struct rw_semaphore node_write; /* locking node writes */ 1452 struct rw_semaphore node_change; /* locking node change */ 1453 wait_queue_head_t cp_wait; 1454 unsigned long last_time[MAX_TIME]; /* to store time in jiffies */ 1455 long interval_time[MAX_TIME]; /* to store thresholds */ 1456 1457 struct inode_management im[MAX_INO_ENTRY]; /* manage inode cache */ 1458 1459 spinlock_t fsync_node_lock; /* for node entry lock */ 1460 struct list_head fsync_node_list; /* node list head */ 1461 unsigned int fsync_seg_id; /* sequence id */ 1462 unsigned int fsync_node_num; /* number of node entries */ 1463 1464 /* for orphan inode, use 0'th array */ 1465 unsigned int max_orphans; /* max orphan inodes */ 1466 1467 /* for inode management */ 1468 struct list_head inode_list[NR_INODE_TYPE]; /* dirty inode list */ 1469 spinlock_t inode_lock[NR_INODE_TYPE]; /* for dirty inode list lock */ 1470 struct mutex flush_lock; /* for flush exclusion */ 1471 1472 /* for extent tree cache */ 1473 struct radix_tree_root extent_tree_root;/* cache extent cache entries */ 1474 struct mutex extent_tree_lock; /* locking extent radix tree */ 1475 struct list_head extent_list; /* lru list for shrinker */ 1476 spinlock_t extent_lock; /* locking extent lru list */ 1477 atomic_t total_ext_tree; /* extent tree count */ 1478 struct list_head zombie_list; /* extent zombie tree list */ 1479 atomic_t total_zombie_tree; /* extent zombie tree count */ 1480 atomic_t total_ext_node; /* extent info count */ 1481 1482 /* basic filesystem units */ 1483 unsigned int log_sectors_per_block; /* log2 sectors per block */ 1484 unsigned int log_blocksize; /* log2 block size */ 1485 unsigned int blocksize; /* block size */ 1486 unsigned int root_ino_num; /* root inode number*/ 1487 unsigned int node_ino_num; /* node inode number*/ 1488 unsigned int meta_ino_num; /* meta inode number*/ 1489 unsigned int log_blocks_per_seg; /* log2 blocks per segment */ 1490 unsigned int blocks_per_seg; /* blocks per segment */ 1491 unsigned int segs_per_sec; /* segments per section */ 1492 unsigned int secs_per_zone; /* sections per zone */ 1493 unsigned int total_sections; /* total section count */ 1494 unsigned int total_node_count; /* total node block count */ 1495 unsigned int total_valid_node_count; /* valid node block count */ 1496 loff_t max_file_blocks; /* max block index of file */ 1497 int dir_level; /* directory level */ 1498 int readdir_ra; /* readahead inode in readdir */ 1499 1500 block_t user_block_count; /* # of user blocks */ 1501 block_t total_valid_block_count; /* # of valid blocks */ 1502 block_t discard_blks; /* discard command candidats */ 1503 block_t last_valid_block_count; /* for recovery */ 1504 block_t reserved_blocks; /* configurable reserved blocks */ 1505 block_t current_reserved_blocks; /* current reserved blocks */ 1506 1507 /* Additional tracking for no checkpoint mode */ 1508 block_t unusable_block_count; /* # of blocks saved by last cp */ 1509 1510 unsigned int nquota_files; /* # of quota sysfile */ 1511 struct rw_semaphore quota_sem; /* blocking cp for flags */ 1512 1513 /* # of pages, see count_type */ 1514 atomic_t nr_pages[NR_COUNT_TYPE]; 1515 /* # of allocated blocks */ 1516 struct percpu_counter alloc_valid_block_count; 1517 1518 /* writeback control */ 1519 atomic_t wb_sync_req[META]; /* count # of WB_SYNC threads */ 1520 1521 /* valid inode count */ 1522 struct percpu_counter total_valid_inode_count; 1523 1524 struct f2fs_mount_info mount_opt; /* mount options */ 1525 1526 /* for cleaning operations */ 1527 struct rw_semaphore gc_lock; /* 1528 * semaphore for GC, avoid 1529 * race between GC and GC or CP 1530 */ 1531 struct f2fs_gc_kthread *gc_thread; /* GC thread */ 1532 struct atgc_management am; /* atgc management */ 1533 unsigned int cur_victim_sec; /* current victim section num */ 1534 unsigned int gc_mode; /* current GC state */ 1535 unsigned int next_victim_seg[2]; /* next segment in victim section */ 1536 1537 /* for skip statistic */ 1538 unsigned int atomic_files; /* # of opened atomic file */ 1539 unsigned long long skipped_atomic_files[2]; /* FG_GC and BG_GC */ 1540 unsigned long long skipped_gc_rwsem; /* FG_GC only */ 1541 1542 /* threshold for gc trials on pinned files */ 1543 u64 gc_pin_file_threshold; 1544 struct rw_semaphore pin_sem; 1545 1546 /* maximum # of trials to find a victim segment for SSR and GC */ 1547 unsigned int max_victim_search; 1548 /* migration granularity of garbage collection, unit: segment */ 1549 unsigned int migration_granularity; 1550 1551 /* 1552 * for stat information. 1553 * one is for the LFS mode, and the other is for the SSR mode. 1554 */ 1555 #ifdef CONFIG_F2FS_STAT_FS 1556 struct f2fs_stat_info *stat_info; /* FS status information */ 1557 atomic_t meta_count[META_MAX]; /* # of meta blocks */ 1558 unsigned int segment_count[2]; /* # of allocated segments */ 1559 unsigned int block_count[2]; /* # of allocated blocks */ 1560 atomic_t inplace_count; /* # of inplace update */ 1561 atomic64_t total_hit_ext; /* # of lookup extent cache */ 1562 atomic64_t read_hit_rbtree; /* # of hit rbtree extent node */ 1563 atomic64_t read_hit_largest; /* # of hit largest extent node */ 1564 atomic64_t read_hit_cached; /* # of hit cached extent node */ 1565 atomic_t inline_xattr; /* # of inline_xattr inodes */ 1566 atomic_t inline_inode; /* # of inline_data inodes */ 1567 atomic_t inline_dir; /* # of inline_dentry inodes */ 1568 atomic_t compr_inode; /* # of compressed inodes */ 1569 atomic64_t compr_blocks; /* # of compressed blocks */ 1570 atomic_t vw_cnt; /* # of volatile writes */ 1571 atomic_t max_aw_cnt; /* max # of atomic writes */ 1572 atomic_t max_vw_cnt; /* max # of volatile writes */ 1573 unsigned int io_skip_bggc; /* skip background gc for in-flight IO */ 1574 unsigned int other_skip_bggc; /* skip background gc for other reasons */ 1575 unsigned int ndirty_inode[NR_INODE_TYPE]; /* # of dirty inodes */ 1576 #endif 1577 spinlock_t stat_lock; /* lock for stat operations */ 1578 1579 /* For app/fs IO statistics */ 1580 spinlock_t iostat_lock; 1581 unsigned long long rw_iostat[NR_IO_TYPE]; 1582 unsigned long long prev_rw_iostat[NR_IO_TYPE]; 1583 bool iostat_enable; 1584 unsigned long iostat_next_period; 1585 unsigned int iostat_period_ms; 1586 1587 /* to attach REQ_META|REQ_FUA flags */ 1588 unsigned int data_io_flag; 1589 unsigned int node_io_flag; 1590 1591 /* For sysfs suppport */ 1592 struct kobject s_kobj; 1593 struct completion s_kobj_unregister; 1594 1595 /* For shrinker support */ 1596 struct list_head s_list; 1597 int s_ndevs; /* number of devices */ 1598 struct f2fs_dev_info *devs; /* for device list */ 1599 unsigned int dirty_device; /* for checkpoint data flush */ 1600 spinlock_t dev_lock; /* protect dirty_device */ 1601 struct mutex umount_mutex; 1602 unsigned int shrinker_run_no; 1603 1604 /* For write statistics */ 1605 u64 sectors_written_start; 1606 u64 kbytes_written; 1607 1608 /* Reference to checksum algorithm driver via cryptoapi */ 1609 struct crypto_shash *s_chksum_driver; 1610 1611 /* Precomputed FS UUID checksum for seeding other checksums */ 1612 __u32 s_chksum_seed; 1613 1614 struct workqueue_struct *post_read_wq; /* post read workqueue */ 1615 1616 struct kmem_cache *inline_xattr_slab; /* inline xattr entry */ 1617 unsigned int inline_xattr_slab_size; /* default inline xattr slab size */ 1618 1619 #ifdef CONFIG_F2FS_FS_COMPRESSION 1620 struct kmem_cache *page_array_slab; /* page array entry */ 1621 unsigned int page_array_slab_size; /* default page array slab size */ 1622 #endif 1623 }; 1624 1625 struct f2fs_private_dio { 1626 struct inode *inode; 1627 void *orig_private; 1628 bio_end_io_t *orig_end_io; 1629 bool write; 1630 }; 1631 1632 #ifdef CONFIG_F2FS_FAULT_INJECTION 1633 #define f2fs_show_injection_info(sbi, type) \ 1634 printk_ratelimited("%sF2FS-fs (%s) : inject %s in %s of %pS\n", \ 1635 KERN_INFO, sbi->sb->s_id, \ 1636 f2fs_fault_name[type], \ 1637 __func__, __builtin_return_address(0)) 1638 static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type) 1639 { 1640 struct f2fs_fault_info *ffi = &F2FS_OPTION(sbi).fault_info; 1641 1642 if (!ffi->inject_rate) 1643 return false; 1644 1645 if (!IS_FAULT_SET(ffi, type)) 1646 return false; 1647 1648 atomic_inc(&ffi->inject_ops); 1649 if (atomic_read(&ffi->inject_ops) >= ffi->inject_rate) { 1650 atomic_set(&ffi->inject_ops, 0); 1651 return true; 1652 } 1653 return false; 1654 } 1655 #else 1656 #define f2fs_show_injection_info(sbi, type) do { } while (0) 1657 static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type) 1658 { 1659 return false; 1660 } 1661 #endif 1662 1663 /* 1664 * Test if the mounted volume is a multi-device volume. 1665 * - For a single regular disk volume, sbi->s_ndevs is 0. 1666 * - For a single zoned disk volume, sbi->s_ndevs is 1. 1667 * - For a multi-device volume, sbi->s_ndevs is always 2 or more. 1668 */ 1669 static inline bool f2fs_is_multi_device(struct f2fs_sb_info *sbi) 1670 { 1671 return sbi->s_ndevs > 1; 1672 } 1673 1674 /* For write statistics. Suppose sector size is 512 bytes, 1675 * and the return value is in kbytes. s is of struct f2fs_sb_info. 1676 */ 1677 #define BD_PART_WRITTEN(s) \ 1678 (((u64)part_stat_read((s)->sb->s_bdev->bd_part, sectors[STAT_WRITE]) - \ 1679 (s)->sectors_written_start) >> 1) 1680 1681 static inline void f2fs_update_time(struct f2fs_sb_info *sbi, int type) 1682 { 1683 unsigned long now = jiffies; 1684 1685 sbi->last_time[type] = now; 1686 1687 /* DISCARD_TIME and GC_TIME are based on REQ_TIME */ 1688 if (type == REQ_TIME) { 1689 sbi->last_time[DISCARD_TIME] = now; 1690 sbi->last_time[GC_TIME] = now; 1691 } 1692 } 1693 1694 static inline bool f2fs_time_over(struct f2fs_sb_info *sbi, int type) 1695 { 1696 unsigned long interval = sbi->interval_time[type] * HZ; 1697 1698 return time_after(jiffies, sbi->last_time[type] + interval); 1699 } 1700 1701 static inline unsigned int f2fs_time_to_wait(struct f2fs_sb_info *sbi, 1702 int type) 1703 { 1704 unsigned long interval = sbi->interval_time[type] * HZ; 1705 unsigned int wait_ms = 0; 1706 long delta; 1707 1708 delta = (sbi->last_time[type] + interval) - jiffies; 1709 if (delta > 0) 1710 wait_ms = jiffies_to_msecs(delta); 1711 1712 return wait_ms; 1713 } 1714 1715 /* 1716 * Inline functions 1717 */ 1718 static inline u32 __f2fs_crc32(struct f2fs_sb_info *sbi, u32 crc, 1719 const void *address, unsigned int length) 1720 { 1721 struct { 1722 struct shash_desc shash; 1723 char ctx[4]; 1724 } desc; 1725 int err; 1726 1727 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver) != sizeof(desc.ctx)); 1728 1729 desc.shash.tfm = sbi->s_chksum_driver; 1730 *(u32 *)desc.ctx = crc; 1731 1732 err = crypto_shash_update(&desc.shash, address, length); 1733 BUG_ON(err); 1734 1735 return *(u32 *)desc.ctx; 1736 } 1737 1738 static inline u32 f2fs_crc32(struct f2fs_sb_info *sbi, const void *address, 1739 unsigned int length) 1740 { 1741 return __f2fs_crc32(sbi, F2FS_SUPER_MAGIC, address, length); 1742 } 1743 1744 static inline bool f2fs_crc_valid(struct f2fs_sb_info *sbi, __u32 blk_crc, 1745 void *buf, size_t buf_size) 1746 { 1747 return f2fs_crc32(sbi, buf, buf_size) == blk_crc; 1748 } 1749 1750 static inline u32 f2fs_chksum(struct f2fs_sb_info *sbi, u32 crc, 1751 const void *address, unsigned int length) 1752 { 1753 return __f2fs_crc32(sbi, crc, address, length); 1754 } 1755 1756 static inline struct f2fs_inode_info *F2FS_I(struct inode *inode) 1757 { 1758 return container_of(inode, struct f2fs_inode_info, vfs_inode); 1759 } 1760 1761 static inline struct f2fs_sb_info *F2FS_SB(struct super_block *sb) 1762 { 1763 return sb->s_fs_info; 1764 } 1765 1766 static inline struct f2fs_sb_info *F2FS_I_SB(struct inode *inode) 1767 { 1768 return F2FS_SB(inode->i_sb); 1769 } 1770 1771 static inline struct f2fs_sb_info *F2FS_M_SB(struct address_space *mapping) 1772 { 1773 return F2FS_I_SB(mapping->host); 1774 } 1775 1776 static inline struct f2fs_sb_info *F2FS_P_SB(struct page *page) 1777 { 1778 return F2FS_M_SB(page_file_mapping(page)); 1779 } 1780 1781 static inline struct f2fs_super_block *F2FS_RAW_SUPER(struct f2fs_sb_info *sbi) 1782 { 1783 return (struct f2fs_super_block *)(sbi->raw_super); 1784 } 1785 1786 static inline struct f2fs_checkpoint *F2FS_CKPT(struct f2fs_sb_info *sbi) 1787 { 1788 return (struct f2fs_checkpoint *)(sbi->ckpt); 1789 } 1790 1791 static inline struct f2fs_node *F2FS_NODE(struct page *page) 1792 { 1793 return (struct f2fs_node *)page_address(page); 1794 } 1795 1796 static inline struct f2fs_inode *F2FS_INODE(struct page *page) 1797 { 1798 return &((struct f2fs_node *)page_address(page))->i; 1799 } 1800 1801 static inline struct f2fs_nm_info *NM_I(struct f2fs_sb_info *sbi) 1802 { 1803 return (struct f2fs_nm_info *)(sbi->nm_info); 1804 } 1805 1806 static inline struct f2fs_sm_info *SM_I(struct f2fs_sb_info *sbi) 1807 { 1808 return (struct f2fs_sm_info *)(sbi->sm_info); 1809 } 1810 1811 static inline struct sit_info *SIT_I(struct f2fs_sb_info *sbi) 1812 { 1813 return (struct sit_info *)(SM_I(sbi)->sit_info); 1814 } 1815 1816 static inline struct free_segmap_info *FREE_I(struct f2fs_sb_info *sbi) 1817 { 1818 return (struct free_segmap_info *)(SM_I(sbi)->free_info); 1819 } 1820 1821 static inline struct dirty_seglist_info *DIRTY_I(struct f2fs_sb_info *sbi) 1822 { 1823 return (struct dirty_seglist_info *)(SM_I(sbi)->dirty_info); 1824 } 1825 1826 static inline struct address_space *META_MAPPING(struct f2fs_sb_info *sbi) 1827 { 1828 return sbi->meta_inode->i_mapping; 1829 } 1830 1831 static inline struct address_space *NODE_MAPPING(struct f2fs_sb_info *sbi) 1832 { 1833 return sbi->node_inode->i_mapping; 1834 } 1835 1836 static inline bool is_sbi_flag_set(struct f2fs_sb_info *sbi, unsigned int type) 1837 { 1838 return test_bit(type, &sbi->s_flag); 1839 } 1840 1841 static inline void set_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type) 1842 { 1843 set_bit(type, &sbi->s_flag); 1844 } 1845 1846 static inline void clear_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type) 1847 { 1848 clear_bit(type, &sbi->s_flag); 1849 } 1850 1851 static inline unsigned long long cur_cp_version(struct f2fs_checkpoint *cp) 1852 { 1853 return le64_to_cpu(cp->checkpoint_ver); 1854 } 1855 1856 static inline unsigned long f2fs_qf_ino(struct super_block *sb, int type) 1857 { 1858 if (type < F2FS_MAX_QUOTAS) 1859 return le32_to_cpu(F2FS_SB(sb)->raw_super->qf_ino[type]); 1860 return 0; 1861 } 1862 1863 static inline __u64 cur_cp_crc(struct f2fs_checkpoint *cp) 1864 { 1865 size_t crc_offset = le32_to_cpu(cp->checksum_offset); 1866 return le32_to_cpu(*((__le32 *)((unsigned char *)cp + crc_offset))); 1867 } 1868 1869 static inline bool __is_set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 1870 { 1871 unsigned int ckpt_flags = le32_to_cpu(cp->ckpt_flags); 1872 1873 return ckpt_flags & f; 1874 } 1875 1876 static inline bool is_set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 1877 { 1878 return __is_set_ckpt_flags(F2FS_CKPT(sbi), f); 1879 } 1880 1881 static inline void __set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 1882 { 1883 unsigned int ckpt_flags; 1884 1885 ckpt_flags = le32_to_cpu(cp->ckpt_flags); 1886 ckpt_flags |= f; 1887 cp->ckpt_flags = cpu_to_le32(ckpt_flags); 1888 } 1889 1890 static inline void set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 1891 { 1892 unsigned long flags; 1893 1894 spin_lock_irqsave(&sbi->cp_lock, flags); 1895 __set_ckpt_flags(F2FS_CKPT(sbi), f); 1896 spin_unlock_irqrestore(&sbi->cp_lock, flags); 1897 } 1898 1899 static inline void __clear_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 1900 { 1901 unsigned int ckpt_flags; 1902 1903 ckpt_flags = le32_to_cpu(cp->ckpt_flags); 1904 ckpt_flags &= (~f); 1905 cp->ckpt_flags = cpu_to_le32(ckpt_flags); 1906 } 1907 1908 static inline void clear_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 1909 { 1910 unsigned long flags; 1911 1912 spin_lock_irqsave(&sbi->cp_lock, flags); 1913 __clear_ckpt_flags(F2FS_CKPT(sbi), f); 1914 spin_unlock_irqrestore(&sbi->cp_lock, flags); 1915 } 1916 1917 static inline void disable_nat_bits(struct f2fs_sb_info *sbi, bool lock) 1918 { 1919 unsigned long flags; 1920 unsigned char *nat_bits; 1921 1922 /* 1923 * In order to re-enable nat_bits we need to call fsck.f2fs by 1924 * set_sbi_flag(sbi, SBI_NEED_FSCK). But it may give huge cost, 1925 * so let's rely on regular fsck or unclean shutdown. 1926 */ 1927 1928 if (lock) 1929 spin_lock_irqsave(&sbi->cp_lock, flags); 1930 __clear_ckpt_flags(F2FS_CKPT(sbi), CP_NAT_BITS_FLAG); 1931 nat_bits = NM_I(sbi)->nat_bits; 1932 NM_I(sbi)->nat_bits = NULL; 1933 if (lock) 1934 spin_unlock_irqrestore(&sbi->cp_lock, flags); 1935 1936 kvfree(nat_bits); 1937 } 1938 1939 static inline bool enabled_nat_bits(struct f2fs_sb_info *sbi, 1940 struct cp_control *cpc) 1941 { 1942 bool set = is_set_ckpt_flags(sbi, CP_NAT_BITS_FLAG); 1943 1944 return (cpc) ? (cpc->reason & CP_UMOUNT) && set : set; 1945 } 1946 1947 static inline void f2fs_lock_op(struct f2fs_sb_info *sbi) 1948 { 1949 down_read(&sbi->cp_rwsem); 1950 } 1951 1952 static inline int f2fs_trylock_op(struct f2fs_sb_info *sbi) 1953 { 1954 return down_read_trylock(&sbi->cp_rwsem); 1955 } 1956 1957 static inline void f2fs_unlock_op(struct f2fs_sb_info *sbi) 1958 { 1959 up_read(&sbi->cp_rwsem); 1960 } 1961 1962 static inline void f2fs_lock_all(struct f2fs_sb_info *sbi) 1963 { 1964 down_write(&sbi->cp_rwsem); 1965 } 1966 1967 static inline void f2fs_unlock_all(struct f2fs_sb_info *sbi) 1968 { 1969 up_write(&sbi->cp_rwsem); 1970 } 1971 1972 static inline int __get_cp_reason(struct f2fs_sb_info *sbi) 1973 { 1974 int reason = CP_SYNC; 1975 1976 if (test_opt(sbi, FASTBOOT)) 1977 reason = CP_FASTBOOT; 1978 if (is_sbi_flag_set(sbi, SBI_IS_CLOSE)) 1979 reason = CP_UMOUNT; 1980 return reason; 1981 } 1982 1983 static inline bool __remain_node_summaries(int reason) 1984 { 1985 return (reason & (CP_UMOUNT | CP_FASTBOOT)); 1986 } 1987 1988 static inline bool __exist_node_summaries(struct f2fs_sb_info *sbi) 1989 { 1990 return (is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG) || 1991 is_set_ckpt_flags(sbi, CP_FASTBOOT_FLAG)); 1992 } 1993 1994 /* 1995 * Check whether the inode has blocks or not 1996 */ 1997 static inline int F2FS_HAS_BLOCKS(struct inode *inode) 1998 { 1999 block_t xattr_block = F2FS_I(inode)->i_xattr_nid ? 1 : 0; 2000 2001 return (inode->i_blocks >> F2FS_LOG_SECTORS_PER_BLOCK) > xattr_block; 2002 } 2003 2004 static inline bool f2fs_has_xattr_block(unsigned int ofs) 2005 { 2006 return ofs == XATTR_NODE_OFFSET; 2007 } 2008 2009 static inline bool __allow_reserved_blocks(struct f2fs_sb_info *sbi, 2010 struct inode *inode, bool cap) 2011 { 2012 if (!inode) 2013 return true; 2014 if (!test_opt(sbi, RESERVE_ROOT)) 2015 return false; 2016 if (IS_NOQUOTA(inode)) 2017 return true; 2018 if (uid_eq(F2FS_OPTION(sbi).s_resuid, current_fsuid())) 2019 return true; 2020 if (!gid_eq(F2FS_OPTION(sbi).s_resgid, GLOBAL_ROOT_GID) && 2021 in_group_p(F2FS_OPTION(sbi).s_resgid)) 2022 return true; 2023 if (cap && capable(CAP_SYS_RESOURCE)) 2024 return true; 2025 return false; 2026 } 2027 2028 static inline void f2fs_i_blocks_write(struct inode *, block_t, bool, bool); 2029 static inline int inc_valid_block_count(struct f2fs_sb_info *sbi, 2030 struct inode *inode, blkcnt_t *count) 2031 { 2032 blkcnt_t diff = 0, release = 0; 2033 block_t avail_user_block_count; 2034 int ret; 2035 2036 ret = dquot_reserve_block(inode, *count); 2037 if (ret) 2038 return ret; 2039 2040 if (time_to_inject(sbi, FAULT_BLOCK)) { 2041 f2fs_show_injection_info(sbi, FAULT_BLOCK); 2042 release = *count; 2043 goto release_quota; 2044 } 2045 2046 /* 2047 * let's increase this in prior to actual block count change in order 2048 * for f2fs_sync_file to avoid data races when deciding checkpoint. 2049 */ 2050 percpu_counter_add(&sbi->alloc_valid_block_count, (*count)); 2051 2052 spin_lock(&sbi->stat_lock); 2053 sbi->total_valid_block_count += (block_t)(*count); 2054 avail_user_block_count = sbi->user_block_count - 2055 sbi->current_reserved_blocks; 2056 2057 if (!__allow_reserved_blocks(sbi, inode, true)) 2058 avail_user_block_count -= F2FS_OPTION(sbi).root_reserved_blocks; 2059 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) { 2060 if (avail_user_block_count > sbi->unusable_block_count) 2061 avail_user_block_count -= sbi->unusable_block_count; 2062 else 2063 avail_user_block_count = 0; 2064 } 2065 if (unlikely(sbi->total_valid_block_count > avail_user_block_count)) { 2066 diff = sbi->total_valid_block_count - avail_user_block_count; 2067 if (diff > *count) 2068 diff = *count; 2069 *count -= diff; 2070 release = diff; 2071 sbi->total_valid_block_count -= diff; 2072 if (!*count) { 2073 spin_unlock(&sbi->stat_lock); 2074 goto enospc; 2075 } 2076 } 2077 spin_unlock(&sbi->stat_lock); 2078 2079 if (unlikely(release)) { 2080 percpu_counter_sub(&sbi->alloc_valid_block_count, release); 2081 dquot_release_reservation_block(inode, release); 2082 } 2083 f2fs_i_blocks_write(inode, *count, true, true); 2084 return 0; 2085 2086 enospc: 2087 percpu_counter_sub(&sbi->alloc_valid_block_count, release); 2088 release_quota: 2089 dquot_release_reservation_block(inode, release); 2090 return -ENOSPC; 2091 } 2092 2093 __printf(2, 3) 2094 void f2fs_printk(struct f2fs_sb_info *sbi, const char *fmt, ...); 2095 2096 #define f2fs_err(sbi, fmt, ...) \ 2097 f2fs_printk(sbi, KERN_ERR fmt, ##__VA_ARGS__) 2098 #define f2fs_warn(sbi, fmt, ...) \ 2099 f2fs_printk(sbi, KERN_WARNING fmt, ##__VA_ARGS__) 2100 #define f2fs_notice(sbi, fmt, ...) \ 2101 f2fs_printk(sbi, KERN_NOTICE fmt, ##__VA_ARGS__) 2102 #define f2fs_info(sbi, fmt, ...) \ 2103 f2fs_printk(sbi, KERN_INFO fmt, ##__VA_ARGS__) 2104 #define f2fs_debug(sbi, fmt, ...) \ 2105 f2fs_printk(sbi, KERN_DEBUG fmt, ##__VA_ARGS__) 2106 2107 static inline void dec_valid_block_count(struct f2fs_sb_info *sbi, 2108 struct inode *inode, 2109 block_t count) 2110 { 2111 blkcnt_t sectors = count << F2FS_LOG_SECTORS_PER_BLOCK; 2112 2113 spin_lock(&sbi->stat_lock); 2114 f2fs_bug_on(sbi, sbi->total_valid_block_count < (block_t) count); 2115 sbi->total_valid_block_count -= (block_t)count; 2116 if (sbi->reserved_blocks && 2117 sbi->current_reserved_blocks < sbi->reserved_blocks) 2118 sbi->current_reserved_blocks = min(sbi->reserved_blocks, 2119 sbi->current_reserved_blocks + count); 2120 spin_unlock(&sbi->stat_lock); 2121 if (unlikely(inode->i_blocks < sectors)) { 2122 f2fs_warn(sbi, "Inconsistent i_blocks, ino:%lu, iblocks:%llu, sectors:%llu", 2123 inode->i_ino, 2124 (unsigned long long)inode->i_blocks, 2125 (unsigned long long)sectors); 2126 set_sbi_flag(sbi, SBI_NEED_FSCK); 2127 return; 2128 } 2129 f2fs_i_blocks_write(inode, count, false, true); 2130 } 2131 2132 static inline void inc_page_count(struct f2fs_sb_info *sbi, int count_type) 2133 { 2134 atomic_inc(&sbi->nr_pages[count_type]); 2135 2136 if (count_type == F2FS_DIRTY_DENTS || 2137 count_type == F2FS_DIRTY_NODES || 2138 count_type == F2FS_DIRTY_META || 2139 count_type == F2FS_DIRTY_QDATA || 2140 count_type == F2FS_DIRTY_IMETA) 2141 set_sbi_flag(sbi, SBI_IS_DIRTY); 2142 } 2143 2144 static inline void inode_inc_dirty_pages(struct inode *inode) 2145 { 2146 atomic_inc(&F2FS_I(inode)->dirty_pages); 2147 inc_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ? 2148 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA); 2149 if (IS_NOQUOTA(inode)) 2150 inc_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA); 2151 } 2152 2153 static inline void dec_page_count(struct f2fs_sb_info *sbi, int count_type) 2154 { 2155 atomic_dec(&sbi->nr_pages[count_type]); 2156 } 2157 2158 static inline void inode_dec_dirty_pages(struct inode *inode) 2159 { 2160 if (!S_ISDIR(inode->i_mode) && !S_ISREG(inode->i_mode) && 2161 !S_ISLNK(inode->i_mode)) 2162 return; 2163 2164 atomic_dec(&F2FS_I(inode)->dirty_pages); 2165 dec_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ? 2166 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA); 2167 if (IS_NOQUOTA(inode)) 2168 dec_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA); 2169 } 2170 2171 static inline s64 get_pages(struct f2fs_sb_info *sbi, int count_type) 2172 { 2173 return atomic_read(&sbi->nr_pages[count_type]); 2174 } 2175 2176 static inline int get_dirty_pages(struct inode *inode) 2177 { 2178 return atomic_read(&F2FS_I(inode)->dirty_pages); 2179 } 2180 2181 static inline int get_blocktype_secs(struct f2fs_sb_info *sbi, int block_type) 2182 { 2183 unsigned int pages_per_sec = sbi->segs_per_sec * sbi->blocks_per_seg; 2184 unsigned int segs = (get_pages(sbi, block_type) + pages_per_sec - 1) >> 2185 sbi->log_blocks_per_seg; 2186 2187 return segs / sbi->segs_per_sec; 2188 } 2189 2190 static inline block_t valid_user_blocks(struct f2fs_sb_info *sbi) 2191 { 2192 return sbi->total_valid_block_count; 2193 } 2194 2195 static inline block_t discard_blocks(struct f2fs_sb_info *sbi) 2196 { 2197 return sbi->discard_blks; 2198 } 2199 2200 static inline unsigned long __bitmap_size(struct f2fs_sb_info *sbi, int flag) 2201 { 2202 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi); 2203 2204 /* return NAT or SIT bitmap */ 2205 if (flag == NAT_BITMAP) 2206 return le32_to_cpu(ckpt->nat_ver_bitmap_bytesize); 2207 else if (flag == SIT_BITMAP) 2208 return le32_to_cpu(ckpt->sit_ver_bitmap_bytesize); 2209 2210 return 0; 2211 } 2212 2213 static inline block_t __cp_payload(struct f2fs_sb_info *sbi) 2214 { 2215 return le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_payload); 2216 } 2217 2218 static inline void *__bitmap_ptr(struct f2fs_sb_info *sbi, int flag) 2219 { 2220 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi); 2221 int offset; 2222 2223 if (is_set_ckpt_flags(sbi, CP_LARGE_NAT_BITMAP_FLAG)) { 2224 offset = (flag == SIT_BITMAP) ? 2225 le32_to_cpu(ckpt->nat_ver_bitmap_bytesize) : 0; 2226 /* 2227 * if large_nat_bitmap feature is enabled, leave checksum 2228 * protection for all nat/sit bitmaps. 2229 */ 2230 return &ckpt->sit_nat_version_bitmap + offset + sizeof(__le32); 2231 } 2232 2233 if (__cp_payload(sbi) > 0) { 2234 if (flag == NAT_BITMAP) 2235 return &ckpt->sit_nat_version_bitmap; 2236 else 2237 return (unsigned char *)ckpt + F2FS_BLKSIZE; 2238 } else { 2239 offset = (flag == NAT_BITMAP) ? 2240 le32_to_cpu(ckpt->sit_ver_bitmap_bytesize) : 0; 2241 return &ckpt->sit_nat_version_bitmap + offset; 2242 } 2243 } 2244 2245 static inline block_t __start_cp_addr(struct f2fs_sb_info *sbi) 2246 { 2247 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr); 2248 2249 if (sbi->cur_cp_pack == 2) 2250 start_addr += sbi->blocks_per_seg; 2251 return start_addr; 2252 } 2253 2254 static inline block_t __start_cp_next_addr(struct f2fs_sb_info *sbi) 2255 { 2256 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr); 2257 2258 if (sbi->cur_cp_pack == 1) 2259 start_addr += sbi->blocks_per_seg; 2260 return start_addr; 2261 } 2262 2263 static inline void __set_cp_next_pack(struct f2fs_sb_info *sbi) 2264 { 2265 sbi->cur_cp_pack = (sbi->cur_cp_pack == 1) ? 2 : 1; 2266 } 2267 2268 static inline block_t __start_sum_addr(struct f2fs_sb_info *sbi) 2269 { 2270 return le32_to_cpu(F2FS_CKPT(sbi)->cp_pack_start_sum); 2271 } 2272 2273 static inline int inc_valid_node_count(struct f2fs_sb_info *sbi, 2274 struct inode *inode, bool is_inode) 2275 { 2276 block_t valid_block_count; 2277 unsigned int valid_node_count, user_block_count; 2278 int err; 2279 2280 if (is_inode) { 2281 if (inode) { 2282 err = dquot_alloc_inode(inode); 2283 if (err) 2284 return err; 2285 } 2286 } else { 2287 err = dquot_reserve_block(inode, 1); 2288 if (err) 2289 return err; 2290 } 2291 2292 if (time_to_inject(sbi, FAULT_BLOCK)) { 2293 f2fs_show_injection_info(sbi, FAULT_BLOCK); 2294 goto enospc; 2295 } 2296 2297 spin_lock(&sbi->stat_lock); 2298 2299 valid_block_count = sbi->total_valid_block_count + 2300 sbi->current_reserved_blocks + 1; 2301 2302 if (!__allow_reserved_blocks(sbi, inode, false)) 2303 valid_block_count += F2FS_OPTION(sbi).root_reserved_blocks; 2304 user_block_count = sbi->user_block_count; 2305 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) 2306 user_block_count -= sbi->unusable_block_count; 2307 2308 if (unlikely(valid_block_count > user_block_count)) { 2309 spin_unlock(&sbi->stat_lock); 2310 goto enospc; 2311 } 2312 2313 valid_node_count = sbi->total_valid_node_count + 1; 2314 if (unlikely(valid_node_count > sbi->total_node_count)) { 2315 spin_unlock(&sbi->stat_lock); 2316 goto enospc; 2317 } 2318 2319 sbi->total_valid_node_count++; 2320 sbi->total_valid_block_count++; 2321 spin_unlock(&sbi->stat_lock); 2322 2323 if (inode) { 2324 if (is_inode) 2325 f2fs_mark_inode_dirty_sync(inode, true); 2326 else 2327 f2fs_i_blocks_write(inode, 1, true, true); 2328 } 2329 2330 percpu_counter_inc(&sbi->alloc_valid_block_count); 2331 return 0; 2332 2333 enospc: 2334 if (is_inode) { 2335 if (inode) 2336 dquot_free_inode(inode); 2337 } else { 2338 dquot_release_reservation_block(inode, 1); 2339 } 2340 return -ENOSPC; 2341 } 2342 2343 static inline void dec_valid_node_count(struct f2fs_sb_info *sbi, 2344 struct inode *inode, bool is_inode) 2345 { 2346 spin_lock(&sbi->stat_lock); 2347 2348 f2fs_bug_on(sbi, !sbi->total_valid_block_count); 2349 f2fs_bug_on(sbi, !sbi->total_valid_node_count); 2350 2351 sbi->total_valid_node_count--; 2352 sbi->total_valid_block_count--; 2353 if (sbi->reserved_blocks && 2354 sbi->current_reserved_blocks < sbi->reserved_blocks) 2355 sbi->current_reserved_blocks++; 2356 2357 spin_unlock(&sbi->stat_lock); 2358 2359 if (is_inode) { 2360 dquot_free_inode(inode); 2361 } else { 2362 if (unlikely(inode->i_blocks == 0)) { 2363 f2fs_warn(sbi, "dec_valid_node_count: inconsistent i_blocks, ino:%lu, iblocks:%llu", 2364 inode->i_ino, 2365 (unsigned long long)inode->i_blocks); 2366 set_sbi_flag(sbi, SBI_NEED_FSCK); 2367 return; 2368 } 2369 f2fs_i_blocks_write(inode, 1, false, true); 2370 } 2371 } 2372 2373 static inline unsigned int valid_node_count(struct f2fs_sb_info *sbi) 2374 { 2375 return sbi->total_valid_node_count; 2376 } 2377 2378 static inline void inc_valid_inode_count(struct f2fs_sb_info *sbi) 2379 { 2380 percpu_counter_inc(&sbi->total_valid_inode_count); 2381 } 2382 2383 static inline void dec_valid_inode_count(struct f2fs_sb_info *sbi) 2384 { 2385 percpu_counter_dec(&sbi->total_valid_inode_count); 2386 } 2387 2388 static inline s64 valid_inode_count(struct f2fs_sb_info *sbi) 2389 { 2390 return percpu_counter_sum_positive(&sbi->total_valid_inode_count); 2391 } 2392 2393 static inline struct page *f2fs_grab_cache_page(struct address_space *mapping, 2394 pgoff_t index, bool for_write) 2395 { 2396 struct page *page; 2397 2398 if (IS_ENABLED(CONFIG_F2FS_FAULT_INJECTION)) { 2399 if (!for_write) 2400 page = find_get_page_flags(mapping, index, 2401 FGP_LOCK | FGP_ACCESSED); 2402 else 2403 page = find_lock_page(mapping, index); 2404 if (page) 2405 return page; 2406 2407 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_ALLOC)) { 2408 f2fs_show_injection_info(F2FS_M_SB(mapping), 2409 FAULT_PAGE_ALLOC); 2410 return NULL; 2411 } 2412 } 2413 2414 if (!for_write) 2415 return grab_cache_page(mapping, index); 2416 return grab_cache_page_write_begin(mapping, index, AOP_FLAG_NOFS); 2417 } 2418 2419 static inline struct page *f2fs_pagecache_get_page( 2420 struct address_space *mapping, pgoff_t index, 2421 int fgp_flags, gfp_t gfp_mask) 2422 { 2423 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_GET)) { 2424 f2fs_show_injection_info(F2FS_M_SB(mapping), FAULT_PAGE_GET); 2425 return NULL; 2426 } 2427 2428 return pagecache_get_page(mapping, index, fgp_flags, gfp_mask); 2429 } 2430 2431 static inline void f2fs_copy_page(struct page *src, struct page *dst) 2432 { 2433 char *src_kaddr = kmap(src); 2434 char *dst_kaddr = kmap(dst); 2435 2436 memcpy(dst_kaddr, src_kaddr, PAGE_SIZE); 2437 kunmap(dst); 2438 kunmap(src); 2439 } 2440 2441 static inline void f2fs_put_page(struct page *page, int unlock) 2442 { 2443 if (!page) 2444 return; 2445 2446 if (unlock) { 2447 f2fs_bug_on(F2FS_P_SB(page), !PageLocked(page)); 2448 unlock_page(page); 2449 } 2450 put_page(page); 2451 } 2452 2453 static inline void f2fs_put_dnode(struct dnode_of_data *dn) 2454 { 2455 if (dn->node_page) 2456 f2fs_put_page(dn->node_page, 1); 2457 if (dn->inode_page && dn->node_page != dn->inode_page) 2458 f2fs_put_page(dn->inode_page, 0); 2459 dn->node_page = NULL; 2460 dn->inode_page = NULL; 2461 } 2462 2463 static inline struct kmem_cache *f2fs_kmem_cache_create(const char *name, 2464 size_t size) 2465 { 2466 return kmem_cache_create(name, size, 0, SLAB_RECLAIM_ACCOUNT, NULL); 2467 } 2468 2469 static inline void *f2fs_kmem_cache_alloc(struct kmem_cache *cachep, 2470 gfp_t flags) 2471 { 2472 void *entry; 2473 2474 entry = kmem_cache_alloc(cachep, flags); 2475 if (!entry) 2476 entry = kmem_cache_alloc(cachep, flags | __GFP_NOFAIL); 2477 return entry; 2478 } 2479 2480 static inline bool is_idle(struct f2fs_sb_info *sbi, int type) 2481 { 2482 if (sbi->gc_mode == GC_URGENT_HIGH) 2483 return true; 2484 2485 if (get_pages(sbi, F2FS_RD_DATA) || get_pages(sbi, F2FS_RD_NODE) || 2486 get_pages(sbi, F2FS_RD_META) || get_pages(sbi, F2FS_WB_DATA) || 2487 get_pages(sbi, F2FS_WB_CP_DATA) || 2488 get_pages(sbi, F2FS_DIO_READ) || 2489 get_pages(sbi, F2FS_DIO_WRITE)) 2490 return false; 2491 2492 if (type != DISCARD_TIME && SM_I(sbi) && SM_I(sbi)->dcc_info && 2493 atomic_read(&SM_I(sbi)->dcc_info->queued_discard)) 2494 return false; 2495 2496 if (SM_I(sbi) && SM_I(sbi)->fcc_info && 2497 atomic_read(&SM_I(sbi)->fcc_info->queued_flush)) 2498 return false; 2499 2500 if (sbi->gc_mode == GC_URGENT_LOW && 2501 (type == DISCARD_TIME || type == GC_TIME)) 2502 return true; 2503 2504 return f2fs_time_over(sbi, type); 2505 } 2506 2507 static inline void f2fs_radix_tree_insert(struct radix_tree_root *root, 2508 unsigned long index, void *item) 2509 { 2510 while (radix_tree_insert(root, index, item)) 2511 cond_resched(); 2512 } 2513 2514 #define RAW_IS_INODE(p) ((p)->footer.nid == (p)->footer.ino) 2515 2516 static inline bool IS_INODE(struct page *page) 2517 { 2518 struct f2fs_node *p = F2FS_NODE(page); 2519 2520 return RAW_IS_INODE(p); 2521 } 2522 2523 static inline int offset_in_addr(struct f2fs_inode *i) 2524 { 2525 return (i->i_inline & F2FS_EXTRA_ATTR) ? 2526 (le16_to_cpu(i->i_extra_isize) / sizeof(__le32)) : 0; 2527 } 2528 2529 static inline __le32 *blkaddr_in_node(struct f2fs_node *node) 2530 { 2531 return RAW_IS_INODE(node) ? node->i.i_addr : node->dn.addr; 2532 } 2533 2534 static inline int f2fs_has_extra_attr(struct inode *inode); 2535 static inline block_t data_blkaddr(struct inode *inode, 2536 struct page *node_page, unsigned int offset) 2537 { 2538 struct f2fs_node *raw_node; 2539 __le32 *addr_array; 2540 int base = 0; 2541 bool is_inode = IS_INODE(node_page); 2542 2543 raw_node = F2FS_NODE(node_page); 2544 2545 if (is_inode) { 2546 if (!inode) 2547 /* from GC path only */ 2548 base = offset_in_addr(&raw_node->i); 2549 else if (f2fs_has_extra_attr(inode)) 2550 base = get_extra_isize(inode); 2551 } 2552 2553 addr_array = blkaddr_in_node(raw_node); 2554 return le32_to_cpu(addr_array[base + offset]); 2555 } 2556 2557 static inline block_t f2fs_data_blkaddr(struct dnode_of_data *dn) 2558 { 2559 return data_blkaddr(dn->inode, dn->node_page, dn->ofs_in_node); 2560 } 2561 2562 static inline int f2fs_test_bit(unsigned int nr, char *addr) 2563 { 2564 int mask; 2565 2566 addr += (nr >> 3); 2567 mask = 1 << (7 - (nr & 0x07)); 2568 return mask & *addr; 2569 } 2570 2571 static inline void f2fs_set_bit(unsigned int nr, char *addr) 2572 { 2573 int mask; 2574 2575 addr += (nr >> 3); 2576 mask = 1 << (7 - (nr & 0x07)); 2577 *addr |= mask; 2578 } 2579 2580 static inline void f2fs_clear_bit(unsigned int nr, char *addr) 2581 { 2582 int mask; 2583 2584 addr += (nr >> 3); 2585 mask = 1 << (7 - (nr & 0x07)); 2586 *addr &= ~mask; 2587 } 2588 2589 static inline int f2fs_test_and_set_bit(unsigned int nr, char *addr) 2590 { 2591 int mask; 2592 int ret; 2593 2594 addr += (nr >> 3); 2595 mask = 1 << (7 - (nr & 0x07)); 2596 ret = mask & *addr; 2597 *addr |= mask; 2598 return ret; 2599 } 2600 2601 static inline int f2fs_test_and_clear_bit(unsigned int nr, char *addr) 2602 { 2603 int mask; 2604 int ret; 2605 2606 addr += (nr >> 3); 2607 mask = 1 << (7 - (nr & 0x07)); 2608 ret = mask & *addr; 2609 *addr &= ~mask; 2610 return ret; 2611 } 2612 2613 static inline void f2fs_change_bit(unsigned int nr, char *addr) 2614 { 2615 int mask; 2616 2617 addr += (nr >> 3); 2618 mask = 1 << (7 - (nr & 0x07)); 2619 *addr ^= mask; 2620 } 2621 2622 /* 2623 * On-disk inode flags (f2fs_inode::i_flags) 2624 */ 2625 #define F2FS_COMPR_FL 0x00000004 /* Compress file */ 2626 #define F2FS_SYNC_FL 0x00000008 /* Synchronous updates */ 2627 #define F2FS_IMMUTABLE_FL 0x00000010 /* Immutable file */ 2628 #define F2FS_APPEND_FL 0x00000020 /* writes to file may only append */ 2629 #define F2FS_NODUMP_FL 0x00000040 /* do not dump file */ 2630 #define F2FS_NOATIME_FL 0x00000080 /* do not update atime */ 2631 #define F2FS_NOCOMP_FL 0x00000400 /* Don't compress */ 2632 #define F2FS_INDEX_FL 0x00001000 /* hash-indexed directory */ 2633 #define F2FS_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */ 2634 #define F2FS_PROJINHERIT_FL 0x20000000 /* Create with parents projid */ 2635 #define F2FS_CASEFOLD_FL 0x40000000 /* Casefolded file */ 2636 2637 /* Flags that should be inherited by new inodes from their parent. */ 2638 #define F2FS_FL_INHERITED (F2FS_SYNC_FL | F2FS_NODUMP_FL | F2FS_NOATIME_FL | \ 2639 F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \ 2640 F2FS_CASEFOLD_FL | F2FS_COMPR_FL | F2FS_NOCOMP_FL) 2641 2642 /* Flags that are appropriate for regular files (all but dir-specific ones). */ 2643 #define F2FS_REG_FLMASK (~(F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \ 2644 F2FS_CASEFOLD_FL)) 2645 2646 /* Flags that are appropriate for non-directories/regular files. */ 2647 #define F2FS_OTHER_FLMASK (F2FS_NODUMP_FL | F2FS_NOATIME_FL) 2648 2649 static inline __u32 f2fs_mask_flags(umode_t mode, __u32 flags) 2650 { 2651 if (S_ISDIR(mode)) 2652 return flags; 2653 else if (S_ISREG(mode)) 2654 return flags & F2FS_REG_FLMASK; 2655 else 2656 return flags & F2FS_OTHER_FLMASK; 2657 } 2658 2659 static inline void __mark_inode_dirty_flag(struct inode *inode, 2660 int flag, bool set) 2661 { 2662 switch (flag) { 2663 case FI_INLINE_XATTR: 2664 case FI_INLINE_DATA: 2665 case FI_INLINE_DENTRY: 2666 case FI_NEW_INODE: 2667 if (set) 2668 return; 2669 fallthrough; 2670 case FI_DATA_EXIST: 2671 case FI_INLINE_DOTS: 2672 case FI_PIN_FILE: 2673 f2fs_mark_inode_dirty_sync(inode, true); 2674 } 2675 } 2676 2677 static inline void set_inode_flag(struct inode *inode, int flag) 2678 { 2679 set_bit(flag, F2FS_I(inode)->flags); 2680 __mark_inode_dirty_flag(inode, flag, true); 2681 } 2682 2683 static inline int is_inode_flag_set(struct inode *inode, int flag) 2684 { 2685 return test_bit(flag, F2FS_I(inode)->flags); 2686 } 2687 2688 static inline void clear_inode_flag(struct inode *inode, int flag) 2689 { 2690 clear_bit(flag, F2FS_I(inode)->flags); 2691 __mark_inode_dirty_flag(inode, flag, false); 2692 } 2693 2694 static inline bool f2fs_verity_in_progress(struct inode *inode) 2695 { 2696 return IS_ENABLED(CONFIG_FS_VERITY) && 2697 is_inode_flag_set(inode, FI_VERITY_IN_PROGRESS); 2698 } 2699 2700 static inline void set_acl_inode(struct inode *inode, umode_t mode) 2701 { 2702 F2FS_I(inode)->i_acl_mode = mode; 2703 set_inode_flag(inode, FI_ACL_MODE); 2704 f2fs_mark_inode_dirty_sync(inode, false); 2705 } 2706 2707 static inline void f2fs_i_links_write(struct inode *inode, bool inc) 2708 { 2709 if (inc) 2710 inc_nlink(inode); 2711 else 2712 drop_nlink(inode); 2713 f2fs_mark_inode_dirty_sync(inode, true); 2714 } 2715 2716 static inline void f2fs_i_blocks_write(struct inode *inode, 2717 block_t diff, bool add, bool claim) 2718 { 2719 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE); 2720 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER); 2721 2722 /* add = 1, claim = 1 should be dquot_reserve_block in pair */ 2723 if (add) { 2724 if (claim) 2725 dquot_claim_block(inode, diff); 2726 else 2727 dquot_alloc_block_nofail(inode, diff); 2728 } else { 2729 dquot_free_block(inode, diff); 2730 } 2731 2732 f2fs_mark_inode_dirty_sync(inode, true); 2733 if (clean || recover) 2734 set_inode_flag(inode, FI_AUTO_RECOVER); 2735 } 2736 2737 static inline void f2fs_i_size_write(struct inode *inode, loff_t i_size) 2738 { 2739 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE); 2740 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER); 2741 2742 if (i_size_read(inode) == i_size) 2743 return; 2744 2745 i_size_write(inode, i_size); 2746 f2fs_mark_inode_dirty_sync(inode, true); 2747 if (clean || recover) 2748 set_inode_flag(inode, FI_AUTO_RECOVER); 2749 } 2750 2751 static inline void f2fs_i_depth_write(struct inode *inode, unsigned int depth) 2752 { 2753 F2FS_I(inode)->i_current_depth = depth; 2754 f2fs_mark_inode_dirty_sync(inode, true); 2755 } 2756 2757 static inline void f2fs_i_gc_failures_write(struct inode *inode, 2758 unsigned int count) 2759 { 2760 F2FS_I(inode)->i_gc_failures[GC_FAILURE_PIN] = count; 2761 f2fs_mark_inode_dirty_sync(inode, true); 2762 } 2763 2764 static inline void f2fs_i_xnid_write(struct inode *inode, nid_t xnid) 2765 { 2766 F2FS_I(inode)->i_xattr_nid = xnid; 2767 f2fs_mark_inode_dirty_sync(inode, true); 2768 } 2769 2770 static inline void f2fs_i_pino_write(struct inode *inode, nid_t pino) 2771 { 2772 F2FS_I(inode)->i_pino = pino; 2773 f2fs_mark_inode_dirty_sync(inode, true); 2774 } 2775 2776 static inline void get_inline_info(struct inode *inode, struct f2fs_inode *ri) 2777 { 2778 struct f2fs_inode_info *fi = F2FS_I(inode); 2779 2780 if (ri->i_inline & F2FS_INLINE_XATTR) 2781 set_bit(FI_INLINE_XATTR, fi->flags); 2782 if (ri->i_inline & F2FS_INLINE_DATA) 2783 set_bit(FI_INLINE_DATA, fi->flags); 2784 if (ri->i_inline & F2FS_INLINE_DENTRY) 2785 set_bit(FI_INLINE_DENTRY, fi->flags); 2786 if (ri->i_inline & F2FS_DATA_EXIST) 2787 set_bit(FI_DATA_EXIST, fi->flags); 2788 if (ri->i_inline & F2FS_INLINE_DOTS) 2789 set_bit(FI_INLINE_DOTS, fi->flags); 2790 if (ri->i_inline & F2FS_EXTRA_ATTR) 2791 set_bit(FI_EXTRA_ATTR, fi->flags); 2792 if (ri->i_inline & F2FS_PIN_FILE) 2793 set_bit(FI_PIN_FILE, fi->flags); 2794 } 2795 2796 static inline void set_raw_inline(struct inode *inode, struct f2fs_inode *ri) 2797 { 2798 ri->i_inline = 0; 2799 2800 if (is_inode_flag_set(inode, FI_INLINE_XATTR)) 2801 ri->i_inline |= F2FS_INLINE_XATTR; 2802 if (is_inode_flag_set(inode, FI_INLINE_DATA)) 2803 ri->i_inline |= F2FS_INLINE_DATA; 2804 if (is_inode_flag_set(inode, FI_INLINE_DENTRY)) 2805 ri->i_inline |= F2FS_INLINE_DENTRY; 2806 if (is_inode_flag_set(inode, FI_DATA_EXIST)) 2807 ri->i_inline |= F2FS_DATA_EXIST; 2808 if (is_inode_flag_set(inode, FI_INLINE_DOTS)) 2809 ri->i_inline |= F2FS_INLINE_DOTS; 2810 if (is_inode_flag_set(inode, FI_EXTRA_ATTR)) 2811 ri->i_inline |= F2FS_EXTRA_ATTR; 2812 if (is_inode_flag_set(inode, FI_PIN_FILE)) 2813 ri->i_inline |= F2FS_PIN_FILE; 2814 } 2815 2816 static inline int f2fs_has_extra_attr(struct inode *inode) 2817 { 2818 return is_inode_flag_set(inode, FI_EXTRA_ATTR); 2819 } 2820 2821 static inline int f2fs_has_inline_xattr(struct inode *inode) 2822 { 2823 return is_inode_flag_set(inode, FI_INLINE_XATTR); 2824 } 2825 2826 static inline int f2fs_compressed_file(struct inode *inode) 2827 { 2828 return S_ISREG(inode->i_mode) && 2829 is_inode_flag_set(inode, FI_COMPRESSED_FILE); 2830 } 2831 2832 static inline unsigned int addrs_per_inode(struct inode *inode) 2833 { 2834 unsigned int addrs = CUR_ADDRS_PER_INODE(inode) - 2835 get_inline_xattr_addrs(inode); 2836 2837 if (!f2fs_compressed_file(inode)) 2838 return addrs; 2839 return ALIGN_DOWN(addrs, F2FS_I(inode)->i_cluster_size); 2840 } 2841 2842 static inline unsigned int addrs_per_block(struct inode *inode) 2843 { 2844 if (!f2fs_compressed_file(inode)) 2845 return DEF_ADDRS_PER_BLOCK; 2846 return ALIGN_DOWN(DEF_ADDRS_PER_BLOCK, F2FS_I(inode)->i_cluster_size); 2847 } 2848 2849 static inline void *inline_xattr_addr(struct inode *inode, struct page *page) 2850 { 2851 struct f2fs_inode *ri = F2FS_INODE(page); 2852 2853 return (void *)&(ri->i_addr[DEF_ADDRS_PER_INODE - 2854 get_inline_xattr_addrs(inode)]); 2855 } 2856 2857 static inline int inline_xattr_size(struct inode *inode) 2858 { 2859 if (f2fs_has_inline_xattr(inode)) 2860 return get_inline_xattr_addrs(inode) * sizeof(__le32); 2861 return 0; 2862 } 2863 2864 static inline int f2fs_has_inline_data(struct inode *inode) 2865 { 2866 return is_inode_flag_set(inode, FI_INLINE_DATA); 2867 } 2868 2869 static inline int f2fs_exist_data(struct inode *inode) 2870 { 2871 return is_inode_flag_set(inode, FI_DATA_EXIST); 2872 } 2873 2874 static inline int f2fs_has_inline_dots(struct inode *inode) 2875 { 2876 return is_inode_flag_set(inode, FI_INLINE_DOTS); 2877 } 2878 2879 static inline int f2fs_is_mmap_file(struct inode *inode) 2880 { 2881 return is_inode_flag_set(inode, FI_MMAP_FILE); 2882 } 2883 2884 static inline bool f2fs_is_pinned_file(struct inode *inode) 2885 { 2886 return is_inode_flag_set(inode, FI_PIN_FILE); 2887 } 2888 2889 static inline bool f2fs_is_atomic_file(struct inode *inode) 2890 { 2891 return is_inode_flag_set(inode, FI_ATOMIC_FILE); 2892 } 2893 2894 static inline bool f2fs_is_commit_atomic_write(struct inode *inode) 2895 { 2896 return is_inode_flag_set(inode, FI_ATOMIC_COMMIT); 2897 } 2898 2899 static inline bool f2fs_is_volatile_file(struct inode *inode) 2900 { 2901 return is_inode_flag_set(inode, FI_VOLATILE_FILE); 2902 } 2903 2904 static inline bool f2fs_is_first_block_written(struct inode *inode) 2905 { 2906 return is_inode_flag_set(inode, FI_FIRST_BLOCK_WRITTEN); 2907 } 2908 2909 static inline bool f2fs_is_drop_cache(struct inode *inode) 2910 { 2911 return is_inode_flag_set(inode, FI_DROP_CACHE); 2912 } 2913 2914 static inline void *inline_data_addr(struct inode *inode, struct page *page) 2915 { 2916 struct f2fs_inode *ri = F2FS_INODE(page); 2917 int extra_size = get_extra_isize(inode); 2918 2919 return (void *)&(ri->i_addr[extra_size + DEF_INLINE_RESERVED_SIZE]); 2920 } 2921 2922 static inline int f2fs_has_inline_dentry(struct inode *inode) 2923 { 2924 return is_inode_flag_set(inode, FI_INLINE_DENTRY); 2925 } 2926 2927 static inline int is_file(struct inode *inode, int type) 2928 { 2929 return F2FS_I(inode)->i_advise & type; 2930 } 2931 2932 static inline void set_file(struct inode *inode, int type) 2933 { 2934 F2FS_I(inode)->i_advise |= type; 2935 f2fs_mark_inode_dirty_sync(inode, true); 2936 } 2937 2938 static inline void clear_file(struct inode *inode, int type) 2939 { 2940 F2FS_I(inode)->i_advise &= ~type; 2941 f2fs_mark_inode_dirty_sync(inode, true); 2942 } 2943 2944 static inline bool f2fs_is_time_consistent(struct inode *inode) 2945 { 2946 if (!timespec64_equal(F2FS_I(inode)->i_disk_time, &inode->i_atime)) 2947 return false; 2948 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 1, &inode->i_ctime)) 2949 return false; 2950 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 2, &inode->i_mtime)) 2951 return false; 2952 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 3, 2953 &F2FS_I(inode)->i_crtime)) 2954 return false; 2955 return true; 2956 } 2957 2958 static inline bool f2fs_skip_inode_update(struct inode *inode, int dsync) 2959 { 2960 bool ret; 2961 2962 if (dsync) { 2963 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 2964 2965 spin_lock(&sbi->inode_lock[DIRTY_META]); 2966 ret = list_empty(&F2FS_I(inode)->gdirty_list); 2967 spin_unlock(&sbi->inode_lock[DIRTY_META]); 2968 return ret; 2969 } 2970 if (!is_inode_flag_set(inode, FI_AUTO_RECOVER) || 2971 file_keep_isize(inode) || 2972 i_size_read(inode) & ~PAGE_MASK) 2973 return false; 2974 2975 if (!f2fs_is_time_consistent(inode)) 2976 return false; 2977 2978 spin_lock(&F2FS_I(inode)->i_size_lock); 2979 ret = F2FS_I(inode)->last_disk_size == i_size_read(inode); 2980 spin_unlock(&F2FS_I(inode)->i_size_lock); 2981 2982 return ret; 2983 } 2984 2985 static inline bool f2fs_readonly(struct super_block *sb) 2986 { 2987 return sb_rdonly(sb); 2988 } 2989 2990 static inline bool f2fs_cp_error(struct f2fs_sb_info *sbi) 2991 { 2992 return is_set_ckpt_flags(sbi, CP_ERROR_FLAG); 2993 } 2994 2995 static inline bool is_dot_dotdot(const u8 *name, size_t len) 2996 { 2997 if (len == 1 && name[0] == '.') 2998 return true; 2999 3000 if (len == 2 && name[0] == '.' && name[1] == '.') 3001 return true; 3002 3003 return false; 3004 } 3005 3006 static inline bool f2fs_may_extent_tree(struct inode *inode) 3007 { 3008 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 3009 3010 if (!test_opt(sbi, EXTENT_CACHE) || 3011 is_inode_flag_set(inode, FI_NO_EXTENT) || 3012 is_inode_flag_set(inode, FI_COMPRESSED_FILE)) 3013 return false; 3014 3015 /* 3016 * for recovered files during mount do not create extents 3017 * if shrinker is not registered. 3018 */ 3019 if (list_empty(&sbi->s_list)) 3020 return false; 3021 3022 return S_ISREG(inode->i_mode); 3023 } 3024 3025 static inline void *f2fs_kmalloc(struct f2fs_sb_info *sbi, 3026 size_t size, gfp_t flags) 3027 { 3028 if (time_to_inject(sbi, FAULT_KMALLOC)) { 3029 f2fs_show_injection_info(sbi, FAULT_KMALLOC); 3030 return NULL; 3031 } 3032 3033 return kmalloc(size, flags); 3034 } 3035 3036 static inline void *f2fs_kzalloc(struct f2fs_sb_info *sbi, 3037 size_t size, gfp_t flags) 3038 { 3039 return f2fs_kmalloc(sbi, size, flags | __GFP_ZERO); 3040 } 3041 3042 static inline void *f2fs_kvmalloc(struct f2fs_sb_info *sbi, 3043 size_t size, gfp_t flags) 3044 { 3045 if (time_to_inject(sbi, FAULT_KVMALLOC)) { 3046 f2fs_show_injection_info(sbi, FAULT_KVMALLOC); 3047 return NULL; 3048 } 3049 3050 return kvmalloc(size, flags); 3051 } 3052 3053 static inline void *f2fs_kvzalloc(struct f2fs_sb_info *sbi, 3054 size_t size, gfp_t flags) 3055 { 3056 return f2fs_kvmalloc(sbi, size, flags | __GFP_ZERO); 3057 } 3058 3059 static inline int get_extra_isize(struct inode *inode) 3060 { 3061 return F2FS_I(inode)->i_extra_isize / sizeof(__le32); 3062 } 3063 3064 static inline int get_inline_xattr_addrs(struct inode *inode) 3065 { 3066 return F2FS_I(inode)->i_inline_xattr_size; 3067 } 3068 3069 #define f2fs_get_inode_mode(i) \ 3070 ((is_inode_flag_set(i, FI_ACL_MODE)) ? \ 3071 (F2FS_I(i)->i_acl_mode) : ((i)->i_mode)) 3072 3073 #define F2FS_TOTAL_EXTRA_ATTR_SIZE \ 3074 (offsetof(struct f2fs_inode, i_extra_end) - \ 3075 offsetof(struct f2fs_inode, i_extra_isize)) \ 3076 3077 #define F2FS_OLD_ATTRIBUTE_SIZE (offsetof(struct f2fs_inode, i_addr)) 3078 #define F2FS_FITS_IN_INODE(f2fs_inode, extra_isize, field) \ 3079 ((offsetof(typeof(*(f2fs_inode)), field) + \ 3080 sizeof((f2fs_inode)->field)) \ 3081 <= (F2FS_OLD_ATTRIBUTE_SIZE + (extra_isize))) \ 3082 3083 #define DEFAULT_IOSTAT_PERIOD_MS 3000 3084 #define MIN_IOSTAT_PERIOD_MS 100 3085 /* maximum period of iostat tracing is 1 day */ 3086 #define MAX_IOSTAT_PERIOD_MS 8640000 3087 3088 static inline void f2fs_reset_iostat(struct f2fs_sb_info *sbi) 3089 { 3090 int i; 3091 3092 spin_lock(&sbi->iostat_lock); 3093 for (i = 0; i < NR_IO_TYPE; i++) { 3094 sbi->rw_iostat[i] = 0; 3095 sbi->prev_rw_iostat[i] = 0; 3096 } 3097 spin_unlock(&sbi->iostat_lock); 3098 } 3099 3100 extern void f2fs_record_iostat(struct f2fs_sb_info *sbi); 3101 3102 static inline void f2fs_update_iostat(struct f2fs_sb_info *sbi, 3103 enum iostat_type type, unsigned long long io_bytes) 3104 { 3105 if (!sbi->iostat_enable) 3106 return; 3107 spin_lock(&sbi->iostat_lock); 3108 sbi->rw_iostat[type] += io_bytes; 3109 3110 if (type == APP_WRITE_IO || type == APP_DIRECT_IO) 3111 sbi->rw_iostat[APP_BUFFERED_IO] = 3112 sbi->rw_iostat[APP_WRITE_IO] - 3113 sbi->rw_iostat[APP_DIRECT_IO]; 3114 3115 if (type == APP_READ_IO || type == APP_DIRECT_READ_IO) 3116 sbi->rw_iostat[APP_BUFFERED_READ_IO] = 3117 sbi->rw_iostat[APP_READ_IO] - 3118 sbi->rw_iostat[APP_DIRECT_READ_IO]; 3119 spin_unlock(&sbi->iostat_lock); 3120 3121 f2fs_record_iostat(sbi); 3122 } 3123 3124 #define __is_large_section(sbi) ((sbi)->segs_per_sec > 1) 3125 3126 #define __is_meta_io(fio) (PAGE_TYPE_OF_BIO((fio)->type) == META) 3127 3128 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi, 3129 block_t blkaddr, int type); 3130 static inline void verify_blkaddr(struct f2fs_sb_info *sbi, 3131 block_t blkaddr, int type) 3132 { 3133 if (!f2fs_is_valid_blkaddr(sbi, blkaddr, type)) { 3134 f2fs_err(sbi, "invalid blkaddr: %u, type: %d, run fsck to fix.", 3135 blkaddr, type); 3136 f2fs_bug_on(sbi, 1); 3137 } 3138 } 3139 3140 static inline bool __is_valid_data_blkaddr(block_t blkaddr) 3141 { 3142 if (blkaddr == NEW_ADDR || blkaddr == NULL_ADDR || 3143 blkaddr == COMPRESS_ADDR) 3144 return false; 3145 return true; 3146 } 3147 3148 static inline void f2fs_set_page_private(struct page *page, 3149 unsigned long data) 3150 { 3151 if (PagePrivate(page)) 3152 return; 3153 3154 attach_page_private(page, (void *)data); 3155 } 3156 3157 static inline void f2fs_clear_page_private(struct page *page) 3158 { 3159 detach_page_private(page); 3160 } 3161 3162 /* 3163 * file.c 3164 */ 3165 int f2fs_sync_file(struct file *file, loff_t start, loff_t end, int datasync); 3166 void f2fs_truncate_data_blocks(struct dnode_of_data *dn); 3167 int f2fs_do_truncate_blocks(struct inode *inode, u64 from, bool lock); 3168 int f2fs_truncate_blocks(struct inode *inode, u64 from, bool lock); 3169 int f2fs_truncate(struct inode *inode); 3170 int f2fs_getattr(const struct path *path, struct kstat *stat, 3171 u32 request_mask, unsigned int flags); 3172 int f2fs_setattr(struct dentry *dentry, struct iattr *attr); 3173 int f2fs_truncate_hole(struct inode *inode, pgoff_t pg_start, pgoff_t pg_end); 3174 void f2fs_truncate_data_blocks_range(struct dnode_of_data *dn, int count); 3175 int f2fs_precache_extents(struct inode *inode); 3176 long f2fs_ioctl(struct file *filp, unsigned int cmd, unsigned long arg); 3177 long f2fs_compat_ioctl(struct file *file, unsigned int cmd, unsigned long arg); 3178 int f2fs_transfer_project_quota(struct inode *inode, kprojid_t kprojid); 3179 int f2fs_pin_file_control(struct inode *inode, bool inc); 3180 3181 /* 3182 * inode.c 3183 */ 3184 void f2fs_set_inode_flags(struct inode *inode); 3185 bool f2fs_inode_chksum_verify(struct f2fs_sb_info *sbi, struct page *page); 3186 void f2fs_inode_chksum_set(struct f2fs_sb_info *sbi, struct page *page); 3187 struct inode *f2fs_iget(struct super_block *sb, unsigned long ino); 3188 struct inode *f2fs_iget_retry(struct super_block *sb, unsigned long ino); 3189 int f2fs_try_to_free_nats(struct f2fs_sb_info *sbi, int nr_shrink); 3190 void f2fs_update_inode(struct inode *inode, struct page *node_page); 3191 void f2fs_update_inode_page(struct inode *inode); 3192 int f2fs_write_inode(struct inode *inode, struct writeback_control *wbc); 3193 void f2fs_evict_inode(struct inode *inode); 3194 void f2fs_handle_failed_inode(struct inode *inode); 3195 3196 /* 3197 * namei.c 3198 */ 3199 int f2fs_update_extension_list(struct f2fs_sb_info *sbi, const char *name, 3200 bool hot, bool set); 3201 struct dentry *f2fs_get_parent(struct dentry *child); 3202 3203 /* 3204 * dir.c 3205 */ 3206 unsigned char f2fs_get_de_type(struct f2fs_dir_entry *de); 3207 int f2fs_init_casefolded_name(const struct inode *dir, 3208 struct f2fs_filename *fname); 3209 int f2fs_setup_filename(struct inode *dir, const struct qstr *iname, 3210 int lookup, struct f2fs_filename *fname); 3211 int f2fs_prepare_lookup(struct inode *dir, struct dentry *dentry, 3212 struct f2fs_filename *fname); 3213 void f2fs_free_filename(struct f2fs_filename *fname); 3214 struct f2fs_dir_entry *f2fs_find_target_dentry(const struct f2fs_dentry_ptr *d, 3215 const struct f2fs_filename *fname, int *max_slots); 3216 int f2fs_fill_dentries(struct dir_context *ctx, struct f2fs_dentry_ptr *d, 3217 unsigned int start_pos, struct fscrypt_str *fstr); 3218 void f2fs_do_make_empty_dir(struct inode *inode, struct inode *parent, 3219 struct f2fs_dentry_ptr *d); 3220 struct page *f2fs_init_inode_metadata(struct inode *inode, struct inode *dir, 3221 const struct f2fs_filename *fname, struct page *dpage); 3222 void f2fs_update_parent_metadata(struct inode *dir, struct inode *inode, 3223 unsigned int current_depth); 3224 int f2fs_room_for_filename(const void *bitmap, int slots, int max_slots); 3225 void f2fs_drop_nlink(struct inode *dir, struct inode *inode); 3226 struct f2fs_dir_entry *__f2fs_find_entry(struct inode *dir, 3227 const struct f2fs_filename *fname, 3228 struct page **res_page); 3229 struct f2fs_dir_entry *f2fs_find_entry(struct inode *dir, 3230 const struct qstr *child, struct page **res_page); 3231 struct f2fs_dir_entry *f2fs_parent_dir(struct inode *dir, struct page **p); 3232 ino_t f2fs_inode_by_name(struct inode *dir, const struct qstr *qstr, 3233 struct page **page); 3234 void f2fs_set_link(struct inode *dir, struct f2fs_dir_entry *de, 3235 struct page *page, struct inode *inode); 3236 bool f2fs_has_enough_room(struct inode *dir, struct page *ipage, 3237 const struct f2fs_filename *fname); 3238 void f2fs_update_dentry(nid_t ino, umode_t mode, struct f2fs_dentry_ptr *d, 3239 const struct fscrypt_str *name, f2fs_hash_t name_hash, 3240 unsigned int bit_pos); 3241 int f2fs_add_regular_entry(struct inode *dir, const struct f2fs_filename *fname, 3242 struct inode *inode, nid_t ino, umode_t mode); 3243 int f2fs_add_dentry(struct inode *dir, const struct f2fs_filename *fname, 3244 struct inode *inode, nid_t ino, umode_t mode); 3245 int f2fs_do_add_link(struct inode *dir, const struct qstr *name, 3246 struct inode *inode, nid_t ino, umode_t mode); 3247 void f2fs_delete_entry(struct f2fs_dir_entry *dentry, struct page *page, 3248 struct inode *dir, struct inode *inode); 3249 int f2fs_do_tmpfile(struct inode *inode, struct inode *dir); 3250 bool f2fs_empty_dir(struct inode *dir); 3251 3252 static inline int f2fs_add_link(struct dentry *dentry, struct inode *inode) 3253 { 3254 return f2fs_do_add_link(d_inode(dentry->d_parent), &dentry->d_name, 3255 inode, inode->i_ino, inode->i_mode); 3256 } 3257 3258 /* 3259 * super.c 3260 */ 3261 int f2fs_inode_dirtied(struct inode *inode, bool sync); 3262 void f2fs_inode_synced(struct inode *inode); 3263 int f2fs_enable_quota_files(struct f2fs_sb_info *sbi, bool rdonly); 3264 int f2fs_quota_sync(struct super_block *sb, int type); 3265 void f2fs_quota_off_umount(struct super_block *sb); 3266 int f2fs_commit_super(struct f2fs_sb_info *sbi, bool recover); 3267 int f2fs_sync_fs(struct super_block *sb, int sync); 3268 int f2fs_sanity_check_ckpt(struct f2fs_sb_info *sbi); 3269 3270 /* 3271 * hash.c 3272 */ 3273 void f2fs_hash_filename(const struct inode *dir, struct f2fs_filename *fname); 3274 3275 /* 3276 * node.c 3277 */ 3278 struct dnode_of_data; 3279 struct node_info; 3280 3281 int f2fs_check_nid_range(struct f2fs_sb_info *sbi, nid_t nid); 3282 bool f2fs_available_free_memory(struct f2fs_sb_info *sbi, int type); 3283 bool f2fs_in_warm_node_list(struct f2fs_sb_info *sbi, struct page *page); 3284 void f2fs_init_fsync_node_info(struct f2fs_sb_info *sbi); 3285 void f2fs_del_fsync_node_entry(struct f2fs_sb_info *sbi, struct page *page); 3286 void f2fs_reset_fsync_node_info(struct f2fs_sb_info *sbi); 3287 int f2fs_need_dentry_mark(struct f2fs_sb_info *sbi, nid_t nid); 3288 bool f2fs_is_checkpointed_node(struct f2fs_sb_info *sbi, nid_t nid); 3289 bool f2fs_need_inode_block_update(struct f2fs_sb_info *sbi, nid_t ino); 3290 int f2fs_get_node_info(struct f2fs_sb_info *sbi, nid_t nid, 3291 struct node_info *ni); 3292 pgoff_t f2fs_get_next_page_offset(struct dnode_of_data *dn, pgoff_t pgofs); 3293 int f2fs_get_dnode_of_data(struct dnode_of_data *dn, pgoff_t index, int mode); 3294 int f2fs_truncate_inode_blocks(struct inode *inode, pgoff_t from); 3295 int f2fs_truncate_xattr_node(struct inode *inode); 3296 int f2fs_wait_on_node_pages_writeback(struct f2fs_sb_info *sbi, 3297 unsigned int seq_id); 3298 int f2fs_remove_inode_page(struct inode *inode); 3299 struct page *f2fs_new_inode_page(struct inode *inode); 3300 struct page *f2fs_new_node_page(struct dnode_of_data *dn, unsigned int ofs); 3301 void f2fs_ra_node_page(struct f2fs_sb_info *sbi, nid_t nid); 3302 struct page *f2fs_get_node_page(struct f2fs_sb_info *sbi, pgoff_t nid); 3303 struct page *f2fs_get_node_page_ra(struct page *parent, int start); 3304 int f2fs_move_node_page(struct page *node_page, int gc_type); 3305 void f2fs_flush_inline_data(struct f2fs_sb_info *sbi); 3306 int f2fs_fsync_node_pages(struct f2fs_sb_info *sbi, struct inode *inode, 3307 struct writeback_control *wbc, bool atomic, 3308 unsigned int *seq_id); 3309 int f2fs_sync_node_pages(struct f2fs_sb_info *sbi, 3310 struct writeback_control *wbc, 3311 bool do_balance, enum iostat_type io_type); 3312 int f2fs_build_free_nids(struct f2fs_sb_info *sbi, bool sync, bool mount); 3313 bool f2fs_alloc_nid(struct f2fs_sb_info *sbi, nid_t *nid); 3314 void f2fs_alloc_nid_done(struct f2fs_sb_info *sbi, nid_t nid); 3315 void f2fs_alloc_nid_failed(struct f2fs_sb_info *sbi, nid_t nid); 3316 int f2fs_try_to_free_nids(struct f2fs_sb_info *sbi, int nr_shrink); 3317 int f2fs_recover_inline_xattr(struct inode *inode, struct page *page); 3318 int f2fs_recover_xattr_data(struct inode *inode, struct page *page); 3319 int f2fs_recover_inode_page(struct f2fs_sb_info *sbi, struct page *page); 3320 int f2fs_restore_node_summary(struct f2fs_sb_info *sbi, 3321 unsigned int segno, struct f2fs_summary_block *sum); 3322 int f2fs_flush_nat_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3323 int f2fs_build_node_manager(struct f2fs_sb_info *sbi); 3324 void f2fs_destroy_node_manager(struct f2fs_sb_info *sbi); 3325 int __init f2fs_create_node_manager_caches(void); 3326 void f2fs_destroy_node_manager_caches(void); 3327 3328 /* 3329 * segment.c 3330 */ 3331 bool f2fs_need_SSR(struct f2fs_sb_info *sbi); 3332 void f2fs_register_inmem_page(struct inode *inode, struct page *page); 3333 void f2fs_drop_inmem_pages_all(struct f2fs_sb_info *sbi, bool gc_failure); 3334 void f2fs_drop_inmem_pages(struct inode *inode); 3335 void f2fs_drop_inmem_page(struct inode *inode, struct page *page); 3336 int f2fs_commit_inmem_pages(struct inode *inode); 3337 void f2fs_balance_fs(struct f2fs_sb_info *sbi, bool need); 3338 void f2fs_balance_fs_bg(struct f2fs_sb_info *sbi, bool from_bg); 3339 int f2fs_issue_flush(struct f2fs_sb_info *sbi, nid_t ino); 3340 int f2fs_create_flush_cmd_control(struct f2fs_sb_info *sbi); 3341 int f2fs_flush_device_cache(struct f2fs_sb_info *sbi); 3342 void f2fs_destroy_flush_cmd_control(struct f2fs_sb_info *sbi, bool free); 3343 void f2fs_invalidate_blocks(struct f2fs_sb_info *sbi, block_t addr); 3344 bool f2fs_is_checkpointed_data(struct f2fs_sb_info *sbi, block_t blkaddr); 3345 void f2fs_drop_discard_cmd(struct f2fs_sb_info *sbi); 3346 void f2fs_stop_discard_thread(struct f2fs_sb_info *sbi); 3347 bool f2fs_issue_discard_timeout(struct f2fs_sb_info *sbi); 3348 void f2fs_clear_prefree_segments(struct f2fs_sb_info *sbi, 3349 struct cp_control *cpc); 3350 void f2fs_dirty_to_prefree(struct f2fs_sb_info *sbi); 3351 block_t f2fs_get_unusable_blocks(struct f2fs_sb_info *sbi); 3352 int f2fs_disable_cp_again(struct f2fs_sb_info *sbi, block_t unusable); 3353 void f2fs_release_discard_addrs(struct f2fs_sb_info *sbi); 3354 int f2fs_npages_for_summary_flush(struct f2fs_sb_info *sbi, bool for_ra); 3355 void f2fs_init_inmem_curseg(struct f2fs_sb_info *sbi); 3356 void f2fs_save_inmem_curseg(struct f2fs_sb_info *sbi); 3357 void f2fs_restore_inmem_curseg(struct f2fs_sb_info *sbi); 3358 void f2fs_get_new_segment(struct f2fs_sb_info *sbi, 3359 unsigned int *newseg, bool new_sec, int dir); 3360 void f2fs_allocate_segment_for_resize(struct f2fs_sb_info *sbi, int type, 3361 unsigned int start, unsigned int end); 3362 void f2fs_allocate_new_segment(struct f2fs_sb_info *sbi, int type); 3363 void f2fs_allocate_new_segments(struct f2fs_sb_info *sbi); 3364 int f2fs_trim_fs(struct f2fs_sb_info *sbi, struct fstrim_range *range); 3365 bool f2fs_exist_trim_candidates(struct f2fs_sb_info *sbi, 3366 struct cp_control *cpc); 3367 struct page *f2fs_get_sum_page(struct f2fs_sb_info *sbi, unsigned int segno); 3368 void f2fs_update_meta_page(struct f2fs_sb_info *sbi, void *src, 3369 block_t blk_addr); 3370 void f2fs_do_write_meta_page(struct f2fs_sb_info *sbi, struct page *page, 3371 enum iostat_type io_type); 3372 void f2fs_do_write_node_page(unsigned int nid, struct f2fs_io_info *fio); 3373 void f2fs_outplace_write_data(struct dnode_of_data *dn, 3374 struct f2fs_io_info *fio); 3375 int f2fs_inplace_write_data(struct f2fs_io_info *fio); 3376 void f2fs_do_replace_block(struct f2fs_sb_info *sbi, struct f2fs_summary *sum, 3377 block_t old_blkaddr, block_t new_blkaddr, 3378 bool recover_curseg, bool recover_newaddr, 3379 bool from_gc); 3380 void f2fs_replace_block(struct f2fs_sb_info *sbi, struct dnode_of_data *dn, 3381 block_t old_addr, block_t new_addr, 3382 unsigned char version, bool recover_curseg, 3383 bool recover_newaddr); 3384 void f2fs_allocate_data_block(struct f2fs_sb_info *sbi, struct page *page, 3385 block_t old_blkaddr, block_t *new_blkaddr, 3386 struct f2fs_summary *sum, int type, 3387 struct f2fs_io_info *fio); 3388 void f2fs_wait_on_page_writeback(struct page *page, 3389 enum page_type type, bool ordered, bool locked); 3390 void f2fs_wait_on_block_writeback(struct inode *inode, block_t blkaddr); 3391 void f2fs_wait_on_block_writeback_range(struct inode *inode, block_t blkaddr, 3392 block_t len); 3393 void f2fs_write_data_summaries(struct f2fs_sb_info *sbi, block_t start_blk); 3394 void f2fs_write_node_summaries(struct f2fs_sb_info *sbi, block_t start_blk); 3395 int f2fs_lookup_journal_in_cursum(struct f2fs_journal *journal, int type, 3396 unsigned int val, int alloc); 3397 void f2fs_flush_sit_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3398 int f2fs_fix_curseg_write_pointer(struct f2fs_sb_info *sbi); 3399 int f2fs_check_write_pointer(struct f2fs_sb_info *sbi); 3400 int f2fs_build_segment_manager(struct f2fs_sb_info *sbi); 3401 void f2fs_destroy_segment_manager(struct f2fs_sb_info *sbi); 3402 int __init f2fs_create_segment_manager_caches(void); 3403 void f2fs_destroy_segment_manager_caches(void); 3404 int f2fs_rw_hint_to_seg_type(enum rw_hint hint); 3405 enum rw_hint f2fs_io_type_to_rw_hint(struct f2fs_sb_info *sbi, 3406 enum page_type type, enum temp_type temp); 3407 unsigned int f2fs_usable_segs_in_sec(struct f2fs_sb_info *sbi, 3408 unsigned int segno); 3409 unsigned int f2fs_usable_blks_in_seg(struct f2fs_sb_info *sbi, 3410 unsigned int segno); 3411 3412 /* 3413 * checkpoint.c 3414 */ 3415 void f2fs_stop_checkpoint(struct f2fs_sb_info *sbi, bool end_io); 3416 struct page *f2fs_grab_meta_page(struct f2fs_sb_info *sbi, pgoff_t index); 3417 struct page *f2fs_get_meta_page(struct f2fs_sb_info *sbi, pgoff_t index); 3418 struct page *f2fs_get_meta_page_retry(struct f2fs_sb_info *sbi, pgoff_t index); 3419 struct page *f2fs_get_tmp_page(struct f2fs_sb_info *sbi, pgoff_t index); 3420 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi, 3421 block_t blkaddr, int type); 3422 int f2fs_ra_meta_pages(struct f2fs_sb_info *sbi, block_t start, int nrpages, 3423 int type, bool sync); 3424 void f2fs_ra_meta_pages_cond(struct f2fs_sb_info *sbi, pgoff_t index); 3425 long f2fs_sync_meta_pages(struct f2fs_sb_info *sbi, enum page_type type, 3426 long nr_to_write, enum iostat_type io_type); 3427 void f2fs_add_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type); 3428 void f2fs_remove_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type); 3429 void f2fs_release_ino_entry(struct f2fs_sb_info *sbi, bool all); 3430 bool f2fs_exist_written_data(struct f2fs_sb_info *sbi, nid_t ino, int mode); 3431 void f2fs_set_dirty_device(struct f2fs_sb_info *sbi, nid_t ino, 3432 unsigned int devidx, int type); 3433 bool f2fs_is_dirty_device(struct f2fs_sb_info *sbi, nid_t ino, 3434 unsigned int devidx, int type); 3435 int f2fs_sync_inode_meta(struct f2fs_sb_info *sbi); 3436 int f2fs_acquire_orphan_inode(struct f2fs_sb_info *sbi); 3437 void f2fs_release_orphan_inode(struct f2fs_sb_info *sbi); 3438 void f2fs_add_orphan_inode(struct inode *inode); 3439 void f2fs_remove_orphan_inode(struct f2fs_sb_info *sbi, nid_t ino); 3440 int f2fs_recover_orphan_inodes(struct f2fs_sb_info *sbi); 3441 int f2fs_get_valid_checkpoint(struct f2fs_sb_info *sbi); 3442 void f2fs_update_dirty_page(struct inode *inode, struct page *page); 3443 void f2fs_remove_dirty_inode(struct inode *inode); 3444 int f2fs_sync_dirty_inodes(struct f2fs_sb_info *sbi, enum inode_type type); 3445 void f2fs_wait_on_all_pages(struct f2fs_sb_info *sbi, int type); 3446 int f2fs_write_checkpoint(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3447 void f2fs_init_ino_entry_info(struct f2fs_sb_info *sbi); 3448 int __init f2fs_create_checkpoint_caches(void); 3449 void f2fs_destroy_checkpoint_caches(void); 3450 3451 /* 3452 * data.c 3453 */ 3454 int __init f2fs_init_bioset(void); 3455 void f2fs_destroy_bioset(void); 3456 struct bio *f2fs_bio_alloc(struct f2fs_sb_info *sbi, int npages, bool noio); 3457 int f2fs_init_bio_entry_cache(void); 3458 void f2fs_destroy_bio_entry_cache(void); 3459 void f2fs_submit_bio(struct f2fs_sb_info *sbi, 3460 struct bio *bio, enum page_type type); 3461 void f2fs_submit_merged_write(struct f2fs_sb_info *sbi, enum page_type type); 3462 void f2fs_submit_merged_write_cond(struct f2fs_sb_info *sbi, 3463 struct inode *inode, struct page *page, 3464 nid_t ino, enum page_type type); 3465 void f2fs_submit_merged_ipu_write(struct f2fs_sb_info *sbi, 3466 struct bio **bio, struct page *page); 3467 void f2fs_flush_merged_writes(struct f2fs_sb_info *sbi); 3468 int f2fs_submit_page_bio(struct f2fs_io_info *fio); 3469 int f2fs_merge_page_bio(struct f2fs_io_info *fio); 3470 void f2fs_submit_page_write(struct f2fs_io_info *fio); 3471 struct block_device *f2fs_target_device(struct f2fs_sb_info *sbi, 3472 block_t blk_addr, struct bio *bio); 3473 int f2fs_target_device_index(struct f2fs_sb_info *sbi, block_t blkaddr); 3474 void f2fs_set_data_blkaddr(struct dnode_of_data *dn); 3475 void f2fs_update_data_blkaddr(struct dnode_of_data *dn, block_t blkaddr); 3476 int f2fs_reserve_new_blocks(struct dnode_of_data *dn, blkcnt_t count); 3477 int f2fs_reserve_new_block(struct dnode_of_data *dn); 3478 int f2fs_get_block(struct dnode_of_data *dn, pgoff_t index); 3479 int f2fs_preallocate_blocks(struct kiocb *iocb, struct iov_iter *from); 3480 int f2fs_reserve_block(struct dnode_of_data *dn, pgoff_t index); 3481 struct page *f2fs_get_read_data_page(struct inode *inode, pgoff_t index, 3482 int op_flags, bool for_write); 3483 struct page *f2fs_find_data_page(struct inode *inode, pgoff_t index); 3484 struct page *f2fs_get_lock_data_page(struct inode *inode, pgoff_t index, 3485 bool for_write); 3486 struct page *f2fs_get_new_data_page(struct inode *inode, 3487 struct page *ipage, pgoff_t index, bool new_i_size); 3488 int f2fs_do_write_data_page(struct f2fs_io_info *fio); 3489 void f2fs_do_map_lock(struct f2fs_sb_info *sbi, int flag, bool lock); 3490 int f2fs_map_blocks(struct inode *inode, struct f2fs_map_blocks *map, 3491 int create, int flag); 3492 int f2fs_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, 3493 u64 start, u64 len); 3494 int f2fs_encrypt_one_page(struct f2fs_io_info *fio); 3495 bool f2fs_should_update_inplace(struct inode *inode, struct f2fs_io_info *fio); 3496 bool f2fs_should_update_outplace(struct inode *inode, struct f2fs_io_info *fio); 3497 int f2fs_write_single_data_page(struct page *page, int *submitted, 3498 struct bio **bio, sector_t *last_block, 3499 struct writeback_control *wbc, 3500 enum iostat_type io_type, 3501 int compr_blocks); 3502 void f2fs_invalidate_page(struct page *page, unsigned int offset, 3503 unsigned int length); 3504 int f2fs_release_page(struct page *page, gfp_t wait); 3505 #ifdef CONFIG_MIGRATION 3506 int f2fs_migrate_page(struct address_space *mapping, struct page *newpage, 3507 struct page *page, enum migrate_mode mode); 3508 #endif 3509 bool f2fs_overwrite_io(struct inode *inode, loff_t pos, size_t len); 3510 void f2fs_clear_page_cache_dirty_tag(struct page *page); 3511 int f2fs_init_post_read_processing(void); 3512 void f2fs_destroy_post_read_processing(void); 3513 int f2fs_init_post_read_wq(struct f2fs_sb_info *sbi); 3514 void f2fs_destroy_post_read_wq(struct f2fs_sb_info *sbi); 3515 3516 /* 3517 * gc.c 3518 */ 3519 int f2fs_start_gc_thread(struct f2fs_sb_info *sbi); 3520 void f2fs_stop_gc_thread(struct f2fs_sb_info *sbi); 3521 block_t f2fs_start_bidx_of_node(unsigned int node_ofs, struct inode *inode); 3522 int f2fs_gc(struct f2fs_sb_info *sbi, bool sync, bool background, 3523 unsigned int segno); 3524 void f2fs_build_gc_manager(struct f2fs_sb_info *sbi); 3525 int f2fs_resize_fs(struct f2fs_sb_info *sbi, __u64 block_count); 3526 int __init f2fs_create_garbage_collection_cache(void); 3527 void f2fs_destroy_garbage_collection_cache(void); 3528 3529 /* 3530 * recovery.c 3531 */ 3532 int f2fs_recover_fsync_data(struct f2fs_sb_info *sbi, bool check_only); 3533 bool f2fs_space_for_roll_forward(struct f2fs_sb_info *sbi); 3534 3535 /* 3536 * debug.c 3537 */ 3538 #ifdef CONFIG_F2FS_STAT_FS 3539 struct f2fs_stat_info { 3540 struct list_head stat_list; 3541 struct f2fs_sb_info *sbi; 3542 int all_area_segs, sit_area_segs, nat_area_segs, ssa_area_segs; 3543 int main_area_segs, main_area_sections, main_area_zones; 3544 unsigned long long hit_largest, hit_cached, hit_rbtree; 3545 unsigned long long hit_total, total_ext; 3546 int ext_tree, zombie_tree, ext_node; 3547 int ndirty_node, ndirty_dent, ndirty_meta, ndirty_imeta; 3548 int ndirty_data, ndirty_qdata; 3549 int inmem_pages; 3550 unsigned int ndirty_dirs, ndirty_files, nquota_files, ndirty_all; 3551 int nats, dirty_nats, sits, dirty_sits; 3552 int free_nids, avail_nids, alloc_nids; 3553 int total_count, utilization; 3554 int bg_gc, nr_wb_cp_data, nr_wb_data; 3555 int nr_rd_data, nr_rd_node, nr_rd_meta; 3556 int nr_dio_read, nr_dio_write; 3557 unsigned int io_skip_bggc, other_skip_bggc; 3558 int nr_flushing, nr_flushed, flush_list_empty; 3559 int nr_discarding, nr_discarded; 3560 int nr_discard_cmd; 3561 unsigned int undiscard_blks; 3562 int inline_xattr, inline_inode, inline_dir, append, update, orphans; 3563 int compr_inode; 3564 unsigned long long compr_blocks; 3565 int aw_cnt, max_aw_cnt, vw_cnt, max_vw_cnt; 3566 unsigned int valid_count, valid_node_count, valid_inode_count, discard_blks; 3567 unsigned int bimodal, avg_vblocks; 3568 int util_free, util_valid, util_invalid; 3569 int rsvd_segs, overp_segs; 3570 int dirty_count, node_pages, meta_pages; 3571 int prefree_count, call_count, cp_count, bg_cp_count; 3572 int tot_segs, node_segs, data_segs, free_segs, free_secs; 3573 int bg_node_segs, bg_data_segs; 3574 int tot_blks, data_blks, node_blks; 3575 int bg_data_blks, bg_node_blks; 3576 unsigned long long skipped_atomic_files[2]; 3577 int curseg[NR_CURSEG_TYPE]; 3578 int cursec[NR_CURSEG_TYPE]; 3579 int curzone[NR_CURSEG_TYPE]; 3580 unsigned int dirty_seg[NR_CURSEG_TYPE]; 3581 unsigned int full_seg[NR_CURSEG_TYPE]; 3582 unsigned int valid_blks[NR_CURSEG_TYPE]; 3583 3584 unsigned int meta_count[META_MAX]; 3585 unsigned int segment_count[2]; 3586 unsigned int block_count[2]; 3587 unsigned int inplace_count; 3588 unsigned long long base_mem, cache_mem, page_mem; 3589 }; 3590 3591 static inline struct f2fs_stat_info *F2FS_STAT(struct f2fs_sb_info *sbi) 3592 { 3593 return (struct f2fs_stat_info *)sbi->stat_info; 3594 } 3595 3596 #define stat_inc_cp_count(si) ((si)->cp_count++) 3597 #define stat_inc_bg_cp_count(si) ((si)->bg_cp_count++) 3598 #define stat_inc_call_count(si) ((si)->call_count++) 3599 #define stat_inc_bggc_count(si) ((si)->bg_gc++) 3600 #define stat_io_skip_bggc_count(sbi) ((sbi)->io_skip_bggc++) 3601 #define stat_other_skip_bggc_count(sbi) ((sbi)->other_skip_bggc++) 3602 #define stat_inc_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]++) 3603 #define stat_dec_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]--) 3604 #define stat_inc_total_hit(sbi) (atomic64_inc(&(sbi)->total_hit_ext)) 3605 #define stat_inc_rbtree_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_rbtree)) 3606 #define stat_inc_largest_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_largest)) 3607 #define stat_inc_cached_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_cached)) 3608 #define stat_inc_inline_xattr(inode) \ 3609 do { \ 3610 if (f2fs_has_inline_xattr(inode)) \ 3611 (atomic_inc(&F2FS_I_SB(inode)->inline_xattr)); \ 3612 } while (0) 3613 #define stat_dec_inline_xattr(inode) \ 3614 do { \ 3615 if (f2fs_has_inline_xattr(inode)) \ 3616 (atomic_dec(&F2FS_I_SB(inode)->inline_xattr)); \ 3617 } while (0) 3618 #define stat_inc_inline_inode(inode) \ 3619 do { \ 3620 if (f2fs_has_inline_data(inode)) \ 3621 (atomic_inc(&F2FS_I_SB(inode)->inline_inode)); \ 3622 } while (0) 3623 #define stat_dec_inline_inode(inode) \ 3624 do { \ 3625 if (f2fs_has_inline_data(inode)) \ 3626 (atomic_dec(&F2FS_I_SB(inode)->inline_inode)); \ 3627 } while (0) 3628 #define stat_inc_inline_dir(inode) \ 3629 do { \ 3630 if (f2fs_has_inline_dentry(inode)) \ 3631 (atomic_inc(&F2FS_I_SB(inode)->inline_dir)); \ 3632 } while (0) 3633 #define stat_dec_inline_dir(inode) \ 3634 do { \ 3635 if (f2fs_has_inline_dentry(inode)) \ 3636 (atomic_dec(&F2FS_I_SB(inode)->inline_dir)); \ 3637 } while (0) 3638 #define stat_inc_compr_inode(inode) \ 3639 do { \ 3640 if (f2fs_compressed_file(inode)) \ 3641 (atomic_inc(&F2FS_I_SB(inode)->compr_inode)); \ 3642 } while (0) 3643 #define stat_dec_compr_inode(inode) \ 3644 do { \ 3645 if (f2fs_compressed_file(inode)) \ 3646 (atomic_dec(&F2FS_I_SB(inode)->compr_inode)); \ 3647 } while (0) 3648 #define stat_add_compr_blocks(inode, blocks) \ 3649 (atomic64_add(blocks, &F2FS_I_SB(inode)->compr_blocks)) 3650 #define stat_sub_compr_blocks(inode, blocks) \ 3651 (atomic64_sub(blocks, &F2FS_I_SB(inode)->compr_blocks)) 3652 #define stat_inc_meta_count(sbi, blkaddr) \ 3653 do { \ 3654 if (blkaddr < SIT_I(sbi)->sit_base_addr) \ 3655 atomic_inc(&(sbi)->meta_count[META_CP]); \ 3656 else if (blkaddr < NM_I(sbi)->nat_blkaddr) \ 3657 atomic_inc(&(sbi)->meta_count[META_SIT]); \ 3658 else if (blkaddr < SM_I(sbi)->ssa_blkaddr) \ 3659 atomic_inc(&(sbi)->meta_count[META_NAT]); \ 3660 else if (blkaddr < SM_I(sbi)->main_blkaddr) \ 3661 atomic_inc(&(sbi)->meta_count[META_SSA]); \ 3662 } while (0) 3663 #define stat_inc_seg_type(sbi, curseg) \ 3664 ((sbi)->segment_count[(curseg)->alloc_type]++) 3665 #define stat_inc_block_count(sbi, curseg) \ 3666 ((sbi)->block_count[(curseg)->alloc_type]++) 3667 #define stat_inc_inplace_blocks(sbi) \ 3668 (atomic_inc(&(sbi)->inplace_count)) 3669 #define stat_update_max_atomic_write(inode) \ 3670 do { \ 3671 int cur = F2FS_I_SB(inode)->atomic_files; \ 3672 int max = atomic_read(&F2FS_I_SB(inode)->max_aw_cnt); \ 3673 if (cur > max) \ 3674 atomic_set(&F2FS_I_SB(inode)->max_aw_cnt, cur); \ 3675 } while (0) 3676 #define stat_inc_volatile_write(inode) \ 3677 (atomic_inc(&F2FS_I_SB(inode)->vw_cnt)) 3678 #define stat_dec_volatile_write(inode) \ 3679 (atomic_dec(&F2FS_I_SB(inode)->vw_cnt)) 3680 #define stat_update_max_volatile_write(inode) \ 3681 do { \ 3682 int cur = atomic_read(&F2FS_I_SB(inode)->vw_cnt); \ 3683 int max = atomic_read(&F2FS_I_SB(inode)->max_vw_cnt); \ 3684 if (cur > max) \ 3685 atomic_set(&F2FS_I_SB(inode)->max_vw_cnt, cur); \ 3686 } while (0) 3687 #define stat_inc_seg_count(sbi, type, gc_type) \ 3688 do { \ 3689 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 3690 si->tot_segs++; \ 3691 if ((type) == SUM_TYPE_DATA) { \ 3692 si->data_segs++; \ 3693 si->bg_data_segs += (gc_type == BG_GC) ? 1 : 0; \ 3694 } else { \ 3695 si->node_segs++; \ 3696 si->bg_node_segs += (gc_type == BG_GC) ? 1 : 0; \ 3697 } \ 3698 } while (0) 3699 3700 #define stat_inc_tot_blk_count(si, blks) \ 3701 ((si)->tot_blks += (blks)) 3702 3703 #define stat_inc_data_blk_count(sbi, blks, gc_type) \ 3704 do { \ 3705 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 3706 stat_inc_tot_blk_count(si, blks); \ 3707 si->data_blks += (blks); \ 3708 si->bg_data_blks += ((gc_type) == BG_GC) ? (blks) : 0; \ 3709 } while (0) 3710 3711 #define stat_inc_node_blk_count(sbi, blks, gc_type) \ 3712 do { \ 3713 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 3714 stat_inc_tot_blk_count(si, blks); \ 3715 si->node_blks += (blks); \ 3716 si->bg_node_blks += ((gc_type) == BG_GC) ? (blks) : 0; \ 3717 } while (0) 3718 3719 int f2fs_build_stats(struct f2fs_sb_info *sbi); 3720 void f2fs_destroy_stats(struct f2fs_sb_info *sbi); 3721 void __init f2fs_create_root_stats(void); 3722 void f2fs_destroy_root_stats(void); 3723 void f2fs_update_sit_info(struct f2fs_sb_info *sbi); 3724 #else 3725 #define stat_inc_cp_count(si) do { } while (0) 3726 #define stat_inc_bg_cp_count(si) do { } while (0) 3727 #define stat_inc_call_count(si) do { } while (0) 3728 #define stat_inc_bggc_count(si) do { } while (0) 3729 #define stat_io_skip_bggc_count(sbi) do { } while (0) 3730 #define stat_other_skip_bggc_count(sbi) do { } while (0) 3731 #define stat_inc_dirty_inode(sbi, type) do { } while (0) 3732 #define stat_dec_dirty_inode(sbi, type) do { } while (0) 3733 #define stat_inc_total_hit(sbi) do { } while (0) 3734 #define stat_inc_rbtree_node_hit(sbi) do { } while (0) 3735 #define stat_inc_largest_node_hit(sbi) do { } while (0) 3736 #define stat_inc_cached_node_hit(sbi) do { } while (0) 3737 #define stat_inc_inline_xattr(inode) do { } while (0) 3738 #define stat_dec_inline_xattr(inode) do { } while (0) 3739 #define stat_inc_inline_inode(inode) do { } while (0) 3740 #define stat_dec_inline_inode(inode) do { } while (0) 3741 #define stat_inc_inline_dir(inode) do { } while (0) 3742 #define stat_dec_inline_dir(inode) do { } while (0) 3743 #define stat_inc_compr_inode(inode) do { } while (0) 3744 #define stat_dec_compr_inode(inode) do { } while (0) 3745 #define stat_add_compr_blocks(inode, blocks) do { } while (0) 3746 #define stat_sub_compr_blocks(inode, blocks) do { } while (0) 3747 #define stat_inc_atomic_write(inode) do { } while (0) 3748 #define stat_dec_atomic_write(inode) do { } while (0) 3749 #define stat_update_max_atomic_write(inode) do { } while (0) 3750 #define stat_inc_volatile_write(inode) do { } while (0) 3751 #define stat_dec_volatile_write(inode) do { } while (0) 3752 #define stat_update_max_volatile_write(inode) do { } while (0) 3753 #define stat_inc_meta_count(sbi, blkaddr) do { } while (0) 3754 #define stat_inc_seg_type(sbi, curseg) do { } while (0) 3755 #define stat_inc_block_count(sbi, curseg) do { } while (0) 3756 #define stat_inc_inplace_blocks(sbi) do { } while (0) 3757 #define stat_inc_seg_count(sbi, type, gc_type) do { } while (0) 3758 #define stat_inc_tot_blk_count(si, blks) do { } while (0) 3759 #define stat_inc_data_blk_count(sbi, blks, gc_type) do { } while (0) 3760 #define stat_inc_node_blk_count(sbi, blks, gc_type) do { } while (0) 3761 3762 static inline int f2fs_build_stats(struct f2fs_sb_info *sbi) { return 0; } 3763 static inline void f2fs_destroy_stats(struct f2fs_sb_info *sbi) { } 3764 static inline void __init f2fs_create_root_stats(void) { } 3765 static inline void f2fs_destroy_root_stats(void) { } 3766 static inline void f2fs_update_sit_info(struct f2fs_sb_info *sbi) {} 3767 #endif 3768 3769 extern const struct file_operations f2fs_dir_operations; 3770 #ifdef CONFIG_UNICODE 3771 extern const struct dentry_operations f2fs_dentry_ops; 3772 #endif 3773 extern const struct file_operations f2fs_file_operations; 3774 extern const struct inode_operations f2fs_file_inode_operations; 3775 extern const struct address_space_operations f2fs_dblock_aops; 3776 extern const struct address_space_operations f2fs_node_aops; 3777 extern const struct address_space_operations f2fs_meta_aops; 3778 extern const struct inode_operations f2fs_dir_inode_operations; 3779 extern const struct inode_operations f2fs_symlink_inode_operations; 3780 extern const struct inode_operations f2fs_encrypted_symlink_inode_operations; 3781 extern const struct inode_operations f2fs_special_inode_operations; 3782 extern struct kmem_cache *f2fs_inode_entry_slab; 3783 3784 /* 3785 * inline.c 3786 */ 3787 bool f2fs_may_inline_data(struct inode *inode); 3788 bool f2fs_may_inline_dentry(struct inode *inode); 3789 void f2fs_do_read_inline_data(struct page *page, struct page *ipage); 3790 void f2fs_truncate_inline_inode(struct inode *inode, 3791 struct page *ipage, u64 from); 3792 int f2fs_read_inline_data(struct inode *inode, struct page *page); 3793 int f2fs_convert_inline_page(struct dnode_of_data *dn, struct page *page); 3794 int f2fs_convert_inline_inode(struct inode *inode); 3795 int f2fs_try_convert_inline_dir(struct inode *dir, struct dentry *dentry); 3796 int f2fs_write_inline_data(struct inode *inode, struct page *page); 3797 int f2fs_recover_inline_data(struct inode *inode, struct page *npage); 3798 struct f2fs_dir_entry *f2fs_find_in_inline_dir(struct inode *dir, 3799 const struct f2fs_filename *fname, 3800 struct page **res_page); 3801 int f2fs_make_empty_inline_dir(struct inode *inode, struct inode *parent, 3802 struct page *ipage); 3803 int f2fs_add_inline_entry(struct inode *dir, const struct f2fs_filename *fname, 3804 struct inode *inode, nid_t ino, umode_t mode); 3805 void f2fs_delete_inline_entry(struct f2fs_dir_entry *dentry, 3806 struct page *page, struct inode *dir, 3807 struct inode *inode); 3808 bool f2fs_empty_inline_dir(struct inode *dir); 3809 int f2fs_read_inline_dir(struct file *file, struct dir_context *ctx, 3810 struct fscrypt_str *fstr); 3811 int f2fs_inline_data_fiemap(struct inode *inode, 3812 struct fiemap_extent_info *fieinfo, 3813 __u64 start, __u64 len); 3814 3815 /* 3816 * shrinker.c 3817 */ 3818 unsigned long f2fs_shrink_count(struct shrinker *shrink, 3819 struct shrink_control *sc); 3820 unsigned long f2fs_shrink_scan(struct shrinker *shrink, 3821 struct shrink_control *sc); 3822 void f2fs_join_shrinker(struct f2fs_sb_info *sbi); 3823 void f2fs_leave_shrinker(struct f2fs_sb_info *sbi); 3824 3825 /* 3826 * extent_cache.c 3827 */ 3828 struct rb_entry *f2fs_lookup_rb_tree(struct rb_root_cached *root, 3829 struct rb_entry *cached_re, unsigned int ofs); 3830 struct rb_node **f2fs_lookup_rb_tree_ext(struct f2fs_sb_info *sbi, 3831 struct rb_root_cached *root, 3832 struct rb_node **parent, 3833 unsigned long long key, bool *left_most); 3834 struct rb_node **f2fs_lookup_rb_tree_for_insert(struct f2fs_sb_info *sbi, 3835 struct rb_root_cached *root, 3836 struct rb_node **parent, 3837 unsigned int ofs, bool *leftmost); 3838 struct rb_entry *f2fs_lookup_rb_tree_ret(struct rb_root_cached *root, 3839 struct rb_entry *cached_re, unsigned int ofs, 3840 struct rb_entry **prev_entry, struct rb_entry **next_entry, 3841 struct rb_node ***insert_p, struct rb_node **insert_parent, 3842 bool force, bool *leftmost); 3843 bool f2fs_check_rb_tree_consistence(struct f2fs_sb_info *sbi, 3844 struct rb_root_cached *root, bool check_key); 3845 unsigned int f2fs_shrink_extent_tree(struct f2fs_sb_info *sbi, int nr_shrink); 3846 void f2fs_init_extent_tree(struct inode *inode, struct page *ipage); 3847 void f2fs_drop_extent_tree(struct inode *inode); 3848 unsigned int f2fs_destroy_extent_node(struct inode *inode); 3849 void f2fs_destroy_extent_tree(struct inode *inode); 3850 bool f2fs_lookup_extent_cache(struct inode *inode, pgoff_t pgofs, 3851 struct extent_info *ei); 3852 void f2fs_update_extent_cache(struct dnode_of_data *dn); 3853 void f2fs_update_extent_cache_range(struct dnode_of_data *dn, 3854 pgoff_t fofs, block_t blkaddr, unsigned int len); 3855 void f2fs_init_extent_cache_info(struct f2fs_sb_info *sbi); 3856 int __init f2fs_create_extent_cache(void); 3857 void f2fs_destroy_extent_cache(void); 3858 3859 /* 3860 * sysfs.c 3861 */ 3862 int __init f2fs_init_sysfs(void); 3863 void f2fs_exit_sysfs(void); 3864 int f2fs_register_sysfs(struct f2fs_sb_info *sbi); 3865 void f2fs_unregister_sysfs(struct f2fs_sb_info *sbi); 3866 3867 /* verity.c */ 3868 extern const struct fsverity_operations f2fs_verityops; 3869 3870 /* 3871 * crypto support 3872 */ 3873 static inline bool f2fs_encrypted_file(struct inode *inode) 3874 { 3875 return IS_ENCRYPTED(inode) && S_ISREG(inode->i_mode); 3876 } 3877 3878 static inline void f2fs_set_encrypted_inode(struct inode *inode) 3879 { 3880 #ifdef CONFIG_FS_ENCRYPTION 3881 file_set_encrypt(inode); 3882 f2fs_set_inode_flags(inode); 3883 #endif 3884 } 3885 3886 /* 3887 * Returns true if the reads of the inode's data need to undergo some 3888 * postprocessing step, like decryption or authenticity verification. 3889 */ 3890 static inline bool f2fs_post_read_required(struct inode *inode) 3891 { 3892 return f2fs_encrypted_file(inode) || fsverity_active(inode) || 3893 f2fs_compressed_file(inode); 3894 } 3895 3896 /* 3897 * compress.c 3898 */ 3899 #ifdef CONFIG_F2FS_FS_COMPRESSION 3900 bool f2fs_is_compressed_page(struct page *page); 3901 struct page *f2fs_compress_control_page(struct page *page); 3902 int f2fs_prepare_compress_overwrite(struct inode *inode, 3903 struct page **pagep, pgoff_t index, void **fsdata); 3904 bool f2fs_compress_write_end(struct inode *inode, void *fsdata, 3905 pgoff_t index, unsigned copied); 3906 int f2fs_truncate_partial_cluster(struct inode *inode, u64 from, bool lock); 3907 void f2fs_compress_write_end_io(struct bio *bio, struct page *page); 3908 bool f2fs_is_compress_backend_ready(struct inode *inode); 3909 int f2fs_init_compress_mempool(void); 3910 void f2fs_destroy_compress_mempool(void); 3911 void f2fs_decompress_pages(struct bio *bio, struct page *page, bool verity); 3912 bool f2fs_cluster_is_empty(struct compress_ctx *cc); 3913 bool f2fs_cluster_can_merge_page(struct compress_ctx *cc, pgoff_t index); 3914 void f2fs_compress_ctx_add_page(struct compress_ctx *cc, struct page *page); 3915 int f2fs_write_multi_pages(struct compress_ctx *cc, 3916 int *submitted, 3917 struct writeback_control *wbc, 3918 enum iostat_type io_type); 3919 int f2fs_is_compressed_cluster(struct inode *inode, pgoff_t index); 3920 int f2fs_read_multi_pages(struct compress_ctx *cc, struct bio **bio_ret, 3921 unsigned nr_pages, sector_t *last_block_in_bio, 3922 bool is_readahead, bool for_write); 3923 struct decompress_io_ctx *f2fs_alloc_dic(struct compress_ctx *cc); 3924 void f2fs_free_dic(struct decompress_io_ctx *dic); 3925 void f2fs_decompress_end_io(struct page **rpages, 3926 unsigned int cluster_size, bool err, bool verity); 3927 int f2fs_init_compress_ctx(struct compress_ctx *cc); 3928 void f2fs_destroy_compress_ctx(struct compress_ctx *cc); 3929 void f2fs_init_compress_info(struct f2fs_sb_info *sbi); 3930 int f2fs_init_page_array_cache(struct f2fs_sb_info *sbi); 3931 void f2fs_destroy_page_array_cache(struct f2fs_sb_info *sbi); 3932 int __init f2fs_init_compress_cache(void); 3933 void f2fs_destroy_compress_cache(void); 3934 #else 3935 static inline bool f2fs_is_compressed_page(struct page *page) { return false; } 3936 static inline bool f2fs_is_compress_backend_ready(struct inode *inode) 3937 { 3938 if (!f2fs_compressed_file(inode)) 3939 return true; 3940 /* not support compression */ 3941 return false; 3942 } 3943 static inline struct page *f2fs_compress_control_page(struct page *page) 3944 { 3945 WARN_ON_ONCE(1); 3946 return ERR_PTR(-EINVAL); 3947 } 3948 static inline int f2fs_init_compress_mempool(void) { return 0; } 3949 static inline void f2fs_destroy_compress_mempool(void) { } 3950 static inline int f2fs_init_page_array_cache(struct f2fs_sb_info *sbi) { return 0; } 3951 static inline void f2fs_destroy_page_array_cache(struct f2fs_sb_info *sbi) { } 3952 static inline int __init f2fs_init_compress_cache(void) { return 0; } 3953 static inline void f2fs_destroy_compress_cache(void) { } 3954 #endif 3955 3956 static inline void set_compress_context(struct inode *inode) 3957 { 3958 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 3959 3960 F2FS_I(inode)->i_compress_algorithm = 3961 F2FS_OPTION(sbi).compress_algorithm; 3962 F2FS_I(inode)->i_log_cluster_size = 3963 F2FS_OPTION(sbi).compress_log_size; 3964 F2FS_I(inode)->i_cluster_size = 3965 1 << F2FS_I(inode)->i_log_cluster_size; 3966 F2FS_I(inode)->i_flags |= F2FS_COMPR_FL; 3967 set_inode_flag(inode, FI_COMPRESSED_FILE); 3968 stat_inc_compr_inode(inode); 3969 f2fs_mark_inode_dirty_sync(inode, true); 3970 } 3971 3972 static inline bool f2fs_disable_compressed_file(struct inode *inode) 3973 { 3974 struct f2fs_inode_info *fi = F2FS_I(inode); 3975 3976 if (!f2fs_compressed_file(inode)) 3977 return true; 3978 if (S_ISREG(inode->i_mode) && 3979 (get_dirty_pages(inode) || atomic_read(&fi->i_compr_blocks))) 3980 return false; 3981 3982 fi->i_flags &= ~F2FS_COMPR_FL; 3983 stat_dec_compr_inode(inode); 3984 clear_inode_flag(inode, FI_COMPRESSED_FILE); 3985 f2fs_mark_inode_dirty_sync(inode, true); 3986 return true; 3987 } 3988 3989 #define F2FS_FEATURE_FUNCS(name, flagname) \ 3990 static inline int f2fs_sb_has_##name(struct f2fs_sb_info *sbi) \ 3991 { \ 3992 return F2FS_HAS_FEATURE(sbi, F2FS_FEATURE_##flagname); \ 3993 } 3994 3995 F2FS_FEATURE_FUNCS(encrypt, ENCRYPT); 3996 F2FS_FEATURE_FUNCS(blkzoned, BLKZONED); 3997 F2FS_FEATURE_FUNCS(extra_attr, EXTRA_ATTR); 3998 F2FS_FEATURE_FUNCS(project_quota, PRJQUOTA); 3999 F2FS_FEATURE_FUNCS(inode_chksum, INODE_CHKSUM); 4000 F2FS_FEATURE_FUNCS(flexible_inline_xattr, FLEXIBLE_INLINE_XATTR); 4001 F2FS_FEATURE_FUNCS(quota_ino, QUOTA_INO); 4002 F2FS_FEATURE_FUNCS(inode_crtime, INODE_CRTIME); 4003 F2FS_FEATURE_FUNCS(lost_found, LOST_FOUND); 4004 F2FS_FEATURE_FUNCS(verity, VERITY); 4005 F2FS_FEATURE_FUNCS(sb_chksum, SB_CHKSUM); 4006 F2FS_FEATURE_FUNCS(casefold, CASEFOLD); 4007 F2FS_FEATURE_FUNCS(compression, COMPRESSION); 4008 4009 #ifdef CONFIG_BLK_DEV_ZONED 4010 static inline bool f2fs_blkz_is_seq(struct f2fs_sb_info *sbi, int devi, 4011 block_t blkaddr) 4012 { 4013 unsigned int zno = blkaddr >> sbi->log_blocks_per_blkz; 4014 4015 return test_bit(zno, FDEV(devi).blkz_seq); 4016 } 4017 #endif 4018 4019 static inline bool f2fs_hw_should_discard(struct f2fs_sb_info *sbi) 4020 { 4021 return f2fs_sb_has_blkzoned(sbi); 4022 } 4023 4024 static inline bool f2fs_bdev_support_discard(struct block_device *bdev) 4025 { 4026 return blk_queue_discard(bdev_get_queue(bdev)) || 4027 bdev_is_zoned(bdev); 4028 } 4029 4030 static inline bool f2fs_hw_support_discard(struct f2fs_sb_info *sbi) 4031 { 4032 int i; 4033 4034 if (!f2fs_is_multi_device(sbi)) 4035 return f2fs_bdev_support_discard(sbi->sb->s_bdev); 4036 4037 for (i = 0; i < sbi->s_ndevs; i++) 4038 if (f2fs_bdev_support_discard(FDEV(i).bdev)) 4039 return true; 4040 return false; 4041 } 4042 4043 static inline bool f2fs_realtime_discard_enable(struct f2fs_sb_info *sbi) 4044 { 4045 return (test_opt(sbi, DISCARD) && f2fs_hw_support_discard(sbi)) || 4046 f2fs_hw_should_discard(sbi); 4047 } 4048 4049 static inline bool f2fs_hw_is_readonly(struct f2fs_sb_info *sbi) 4050 { 4051 int i; 4052 4053 if (!f2fs_is_multi_device(sbi)) 4054 return bdev_read_only(sbi->sb->s_bdev); 4055 4056 for (i = 0; i < sbi->s_ndevs; i++) 4057 if (bdev_read_only(FDEV(i).bdev)) 4058 return true; 4059 return false; 4060 } 4061 4062 static inline bool f2fs_lfs_mode(struct f2fs_sb_info *sbi) 4063 { 4064 return F2FS_OPTION(sbi).fs_mode == FS_MODE_LFS; 4065 } 4066 4067 static inline bool f2fs_may_compress(struct inode *inode) 4068 { 4069 if (IS_SWAPFILE(inode) || f2fs_is_pinned_file(inode) || 4070 f2fs_is_atomic_file(inode) || 4071 f2fs_is_volatile_file(inode)) 4072 return false; 4073 return S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode); 4074 } 4075 4076 static inline void f2fs_i_compr_blocks_update(struct inode *inode, 4077 u64 blocks, bool add) 4078 { 4079 int diff = F2FS_I(inode)->i_cluster_size - blocks; 4080 struct f2fs_inode_info *fi = F2FS_I(inode); 4081 4082 /* don't update i_compr_blocks if saved blocks were released */ 4083 if (!add && !atomic_read(&fi->i_compr_blocks)) 4084 return; 4085 4086 if (add) { 4087 atomic_add(diff, &fi->i_compr_blocks); 4088 stat_add_compr_blocks(inode, diff); 4089 } else { 4090 atomic_sub(diff, &fi->i_compr_blocks); 4091 stat_sub_compr_blocks(inode, diff); 4092 } 4093 f2fs_mark_inode_dirty_sync(inode, true); 4094 } 4095 4096 static inline int block_unaligned_IO(struct inode *inode, 4097 struct kiocb *iocb, struct iov_iter *iter) 4098 { 4099 unsigned int i_blkbits = READ_ONCE(inode->i_blkbits); 4100 unsigned int blocksize_mask = (1 << i_blkbits) - 1; 4101 loff_t offset = iocb->ki_pos; 4102 unsigned long align = offset | iov_iter_alignment(iter); 4103 4104 return align & blocksize_mask; 4105 } 4106 4107 static inline int allow_outplace_dio(struct inode *inode, 4108 struct kiocb *iocb, struct iov_iter *iter) 4109 { 4110 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 4111 int rw = iov_iter_rw(iter); 4112 4113 return (f2fs_lfs_mode(sbi) && (rw == WRITE) && 4114 !block_unaligned_IO(inode, iocb, iter)); 4115 } 4116 4117 static inline bool f2fs_force_buffered_io(struct inode *inode, 4118 struct kiocb *iocb, struct iov_iter *iter) 4119 { 4120 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 4121 int rw = iov_iter_rw(iter); 4122 4123 if (f2fs_post_read_required(inode)) 4124 return true; 4125 if (f2fs_is_multi_device(sbi)) 4126 return true; 4127 /* 4128 * for blkzoned device, fallback direct IO to buffered IO, so 4129 * all IOs can be serialized by log-structured write. 4130 */ 4131 if (f2fs_sb_has_blkzoned(sbi)) 4132 return true; 4133 if (f2fs_lfs_mode(sbi) && (rw == WRITE)) { 4134 if (block_unaligned_IO(inode, iocb, iter)) 4135 return true; 4136 if (F2FS_IO_ALIGNED(sbi)) 4137 return true; 4138 } 4139 if (is_sbi_flag_set(F2FS_I_SB(inode), SBI_CP_DISABLED) && 4140 !IS_SWAPFILE(inode)) 4141 return true; 4142 4143 return false; 4144 } 4145 4146 #ifdef CONFIG_F2FS_FAULT_INJECTION 4147 extern void f2fs_build_fault_attr(struct f2fs_sb_info *sbi, unsigned int rate, 4148 unsigned int type); 4149 #else 4150 #define f2fs_build_fault_attr(sbi, rate, type) do { } while (0) 4151 #endif 4152 4153 static inline bool is_journalled_quota(struct f2fs_sb_info *sbi) 4154 { 4155 #ifdef CONFIG_QUOTA 4156 if (f2fs_sb_has_quota_ino(sbi)) 4157 return true; 4158 if (F2FS_OPTION(sbi).s_qf_names[USRQUOTA] || 4159 F2FS_OPTION(sbi).s_qf_names[GRPQUOTA] || 4160 F2FS_OPTION(sbi).s_qf_names[PRJQUOTA]) 4161 return true; 4162 #endif 4163 return false; 4164 } 4165 4166 #define EFSBADCRC EBADMSG /* Bad CRC detected */ 4167 #define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */ 4168 4169 #endif /* _LINUX_F2FS_H */ 4170