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 101 102 #define F2FS_OPTION(sbi) ((sbi)->mount_opt) 103 #define clear_opt(sbi, option) (F2FS_OPTION(sbi).opt &= ~F2FS_MOUNT_##option) 104 #define set_opt(sbi, option) (F2FS_OPTION(sbi).opt |= F2FS_MOUNT_##option) 105 #define test_opt(sbi, option) (F2FS_OPTION(sbi).opt & F2FS_MOUNT_##option) 106 107 #define ver_after(a, b) (typecheck(unsigned long long, a) && \ 108 typecheck(unsigned long long, b) && \ 109 ((long long)((a) - (b)) > 0)) 110 111 typedef u32 block_t; /* 112 * should not change u32, since it is the on-disk block 113 * address format, __le32. 114 */ 115 typedef u32 nid_t; 116 117 #define COMPRESS_EXT_NUM 16 118 119 struct f2fs_mount_info { 120 unsigned int opt; 121 int write_io_size_bits; /* Write IO size bits */ 122 block_t root_reserved_blocks; /* root reserved blocks */ 123 kuid_t s_resuid; /* reserved blocks for uid */ 124 kgid_t s_resgid; /* reserved blocks for gid */ 125 int active_logs; /* # of active logs */ 126 int inline_xattr_size; /* inline xattr size */ 127 #ifdef CONFIG_F2FS_FAULT_INJECTION 128 struct f2fs_fault_info fault_info; /* For fault injection */ 129 #endif 130 #ifdef CONFIG_QUOTA 131 /* Names of quota files with journalled quota */ 132 char *s_qf_names[MAXQUOTAS]; 133 int s_jquota_fmt; /* Format of quota to use */ 134 #endif 135 /* For which write hints are passed down to block layer */ 136 int whint_mode; 137 int alloc_mode; /* segment allocation policy */ 138 int fsync_mode; /* fsync policy */ 139 int fs_mode; /* fs mode: LFS or ADAPTIVE */ 140 int bggc_mode; /* bggc mode: off, on or sync */ 141 struct fscrypt_dummy_policy dummy_enc_policy; /* test dummy encryption */ 142 block_t unusable_cap_perc; /* percentage for cap */ 143 block_t unusable_cap; /* Amount of space allowed to be 144 * unusable when disabling checkpoint 145 */ 146 147 /* For compression */ 148 unsigned char compress_algorithm; /* algorithm type */ 149 unsigned compress_log_size; /* cluster log size */ 150 unsigned char compress_ext_cnt; /* extension count */ 151 unsigned char extensions[COMPRESS_EXT_NUM][F2FS_EXTENSION_LEN]; /* extensions */ 152 }; 153 154 #define F2FS_FEATURE_ENCRYPT 0x0001 155 #define F2FS_FEATURE_BLKZONED 0x0002 156 #define F2FS_FEATURE_ATOMIC_WRITE 0x0004 157 #define F2FS_FEATURE_EXTRA_ATTR 0x0008 158 #define F2FS_FEATURE_PRJQUOTA 0x0010 159 #define F2FS_FEATURE_INODE_CHKSUM 0x0020 160 #define F2FS_FEATURE_FLEXIBLE_INLINE_XATTR 0x0040 161 #define F2FS_FEATURE_QUOTA_INO 0x0080 162 #define F2FS_FEATURE_INODE_CRTIME 0x0100 163 #define F2FS_FEATURE_LOST_FOUND 0x0200 164 #define F2FS_FEATURE_VERITY 0x0400 165 #define F2FS_FEATURE_SB_CHKSUM 0x0800 166 #define F2FS_FEATURE_CASEFOLD 0x1000 167 #define F2FS_FEATURE_COMPRESSION 0x2000 168 169 #define __F2FS_HAS_FEATURE(raw_super, mask) \ 170 ((raw_super->feature & cpu_to_le32(mask)) != 0) 171 #define F2FS_HAS_FEATURE(sbi, mask) __F2FS_HAS_FEATURE(sbi->raw_super, mask) 172 #define F2FS_SET_FEATURE(sbi, mask) \ 173 (sbi->raw_super->feature |= cpu_to_le32(mask)) 174 #define F2FS_CLEAR_FEATURE(sbi, mask) \ 175 (sbi->raw_super->feature &= ~cpu_to_le32(mask)) 176 177 /* 178 * Default values for user and/or group using reserved blocks 179 */ 180 #define F2FS_DEF_RESUID 0 181 #define F2FS_DEF_RESGID 0 182 183 /* 184 * For checkpoint manager 185 */ 186 enum { 187 NAT_BITMAP, 188 SIT_BITMAP 189 }; 190 191 #define CP_UMOUNT 0x00000001 192 #define CP_FASTBOOT 0x00000002 193 #define CP_SYNC 0x00000004 194 #define CP_RECOVERY 0x00000008 195 #define CP_DISCARD 0x00000010 196 #define CP_TRIMMED 0x00000020 197 #define CP_PAUSE 0x00000040 198 #define CP_RESIZE 0x00000080 199 200 #define MAX_DISCARD_BLOCKS(sbi) BLKS_PER_SEC(sbi) 201 #define DEF_MAX_DISCARD_REQUEST 8 /* issue 8 discards per round */ 202 #define DEF_MIN_DISCARD_ISSUE_TIME 50 /* 50 ms, if exists */ 203 #define DEF_MID_DISCARD_ISSUE_TIME 500 /* 500 ms, if device busy */ 204 #define DEF_MAX_DISCARD_ISSUE_TIME 60000 /* 60 s, if no candidates */ 205 #define DEF_DISCARD_URGENT_UTIL 80 /* do more discard over 80% */ 206 #define DEF_CP_INTERVAL 60 /* 60 secs */ 207 #define DEF_IDLE_INTERVAL 5 /* 5 secs */ 208 #define DEF_DISABLE_INTERVAL 5 /* 5 secs */ 209 #define DEF_DISABLE_QUICK_INTERVAL 1 /* 1 secs */ 210 #define DEF_UMOUNT_DISCARD_TIMEOUT 5 /* 5 secs */ 211 212 struct cp_control { 213 int reason; 214 __u64 trim_start; 215 __u64 trim_end; 216 __u64 trim_minlen; 217 }; 218 219 /* 220 * indicate meta/data type 221 */ 222 enum { 223 META_CP, 224 META_NAT, 225 META_SIT, 226 META_SSA, 227 META_MAX, 228 META_POR, 229 DATA_GENERIC, /* check range only */ 230 DATA_GENERIC_ENHANCE, /* strong check on range and segment bitmap */ 231 DATA_GENERIC_ENHANCE_READ, /* 232 * strong check on range and segment 233 * bitmap but no warning due to race 234 * condition of read on truncated area 235 * by extent_cache 236 */ 237 META_GENERIC, 238 }; 239 240 /* for the list of ino */ 241 enum { 242 ORPHAN_INO, /* for orphan ino list */ 243 APPEND_INO, /* for append ino list */ 244 UPDATE_INO, /* for update ino list */ 245 TRANS_DIR_INO, /* for trasactions dir ino list */ 246 FLUSH_INO, /* for multiple device flushing */ 247 MAX_INO_ENTRY, /* max. list */ 248 }; 249 250 struct ino_entry { 251 struct list_head list; /* list head */ 252 nid_t ino; /* inode number */ 253 unsigned int dirty_device; /* dirty device bitmap */ 254 }; 255 256 /* for the list of inodes to be GCed */ 257 struct inode_entry { 258 struct list_head list; /* list head */ 259 struct inode *inode; /* vfs inode pointer */ 260 }; 261 262 struct fsync_node_entry { 263 struct list_head list; /* list head */ 264 struct page *page; /* warm node page pointer */ 265 unsigned int seq_id; /* sequence id */ 266 }; 267 268 /* for the bitmap indicate blocks to be discarded */ 269 struct discard_entry { 270 struct list_head list; /* list head */ 271 block_t start_blkaddr; /* start blockaddr of current segment */ 272 unsigned char discard_map[SIT_VBLOCK_MAP_SIZE]; /* segment discard bitmap */ 273 }; 274 275 /* default discard granularity of inner discard thread, unit: block count */ 276 #define DEFAULT_DISCARD_GRANULARITY 16 277 278 /* max discard pend list number */ 279 #define MAX_PLIST_NUM 512 280 #define plist_idx(blk_num) ((blk_num) >= MAX_PLIST_NUM ? \ 281 (MAX_PLIST_NUM - 1) : ((blk_num) - 1)) 282 283 enum { 284 D_PREP, /* initial */ 285 D_PARTIAL, /* partially submitted */ 286 D_SUBMIT, /* all submitted */ 287 D_DONE, /* finished */ 288 }; 289 290 struct discard_info { 291 block_t lstart; /* logical start address */ 292 block_t len; /* length */ 293 block_t start; /* actual start address in dev */ 294 }; 295 296 struct discard_cmd { 297 struct rb_node rb_node; /* rb node located in rb-tree */ 298 union { 299 struct { 300 block_t lstart; /* logical start address */ 301 block_t len; /* length */ 302 block_t start; /* actual start address in dev */ 303 }; 304 struct discard_info di; /* discard info */ 305 306 }; 307 struct list_head list; /* command list */ 308 struct completion wait; /* compleation */ 309 struct block_device *bdev; /* bdev */ 310 unsigned short ref; /* reference count */ 311 unsigned char state; /* state */ 312 unsigned char queued; /* queued discard */ 313 int error; /* bio error */ 314 spinlock_t lock; /* for state/bio_ref updating */ 315 unsigned short bio_ref; /* bio reference count */ 316 }; 317 318 enum { 319 DPOLICY_BG, 320 DPOLICY_FORCE, 321 DPOLICY_FSTRIM, 322 DPOLICY_UMOUNT, 323 MAX_DPOLICY, 324 }; 325 326 struct discard_policy { 327 int type; /* type of discard */ 328 unsigned int min_interval; /* used for candidates exist */ 329 unsigned int mid_interval; /* used for device busy */ 330 unsigned int max_interval; /* used for candidates not exist */ 331 unsigned int max_requests; /* # of discards issued per round */ 332 unsigned int io_aware_gran; /* minimum granularity discard not be aware of I/O */ 333 bool io_aware; /* issue discard in idle time */ 334 bool sync; /* submit discard with REQ_SYNC flag */ 335 bool ordered; /* issue discard by lba order */ 336 bool timeout; /* discard timeout for put_super */ 337 unsigned int granularity; /* discard granularity */ 338 }; 339 340 struct discard_cmd_control { 341 struct task_struct *f2fs_issue_discard; /* discard thread */ 342 struct list_head entry_list; /* 4KB discard entry list */ 343 struct list_head pend_list[MAX_PLIST_NUM];/* store pending entries */ 344 struct list_head wait_list; /* store on-flushing entries */ 345 struct list_head fstrim_list; /* in-flight discard from fstrim */ 346 wait_queue_head_t discard_wait_queue; /* waiting queue for wake-up */ 347 unsigned int discard_wake; /* to wake up discard thread */ 348 struct mutex cmd_lock; 349 unsigned int nr_discards; /* # of discards in the list */ 350 unsigned int max_discards; /* max. discards to be issued */ 351 unsigned int discard_granularity; /* discard granularity */ 352 unsigned int undiscard_blks; /* # of undiscard blocks */ 353 unsigned int next_pos; /* next discard position */ 354 atomic_t issued_discard; /* # of issued discard */ 355 atomic_t queued_discard; /* # of queued discard */ 356 atomic_t discard_cmd_cnt; /* # of cached cmd count */ 357 struct rb_root_cached root; /* root of discard rb-tree */ 358 bool rbtree_check; /* config for consistence check */ 359 }; 360 361 /* for the list of fsync inodes, used only during recovery */ 362 struct fsync_inode_entry { 363 struct list_head list; /* list head */ 364 struct inode *inode; /* vfs inode pointer */ 365 block_t blkaddr; /* block address locating the last fsync */ 366 block_t last_dentry; /* block address locating the last dentry */ 367 }; 368 369 #define nats_in_cursum(jnl) (le16_to_cpu((jnl)->n_nats)) 370 #define sits_in_cursum(jnl) (le16_to_cpu((jnl)->n_sits)) 371 372 #define nat_in_journal(jnl, i) ((jnl)->nat_j.entries[i].ne) 373 #define nid_in_journal(jnl, i) ((jnl)->nat_j.entries[i].nid) 374 #define sit_in_journal(jnl, i) ((jnl)->sit_j.entries[i].se) 375 #define segno_in_journal(jnl, i) ((jnl)->sit_j.entries[i].segno) 376 377 #define MAX_NAT_JENTRIES(jnl) (NAT_JOURNAL_ENTRIES - nats_in_cursum(jnl)) 378 #define MAX_SIT_JENTRIES(jnl) (SIT_JOURNAL_ENTRIES - sits_in_cursum(jnl)) 379 380 static inline int update_nats_in_cursum(struct f2fs_journal *journal, int i) 381 { 382 int before = nats_in_cursum(journal); 383 384 journal->n_nats = cpu_to_le16(before + i); 385 return before; 386 } 387 388 static inline int update_sits_in_cursum(struct f2fs_journal *journal, int i) 389 { 390 int before = sits_in_cursum(journal); 391 392 journal->n_sits = cpu_to_le16(before + i); 393 return before; 394 } 395 396 static inline bool __has_cursum_space(struct f2fs_journal *journal, 397 int size, int type) 398 { 399 if (type == NAT_JOURNAL) 400 return size <= MAX_NAT_JENTRIES(journal); 401 return size <= MAX_SIT_JENTRIES(journal); 402 } 403 404 /* 405 * f2fs-specific ioctl commands 406 */ 407 #define F2FS_IOCTL_MAGIC 0xf5 408 #define F2FS_IOC_START_ATOMIC_WRITE _IO(F2FS_IOCTL_MAGIC, 1) 409 #define F2FS_IOC_COMMIT_ATOMIC_WRITE _IO(F2FS_IOCTL_MAGIC, 2) 410 #define F2FS_IOC_START_VOLATILE_WRITE _IO(F2FS_IOCTL_MAGIC, 3) 411 #define F2FS_IOC_RELEASE_VOLATILE_WRITE _IO(F2FS_IOCTL_MAGIC, 4) 412 #define F2FS_IOC_ABORT_VOLATILE_WRITE _IO(F2FS_IOCTL_MAGIC, 5) 413 #define F2FS_IOC_GARBAGE_COLLECT _IOW(F2FS_IOCTL_MAGIC, 6, __u32) 414 #define F2FS_IOC_WRITE_CHECKPOINT _IO(F2FS_IOCTL_MAGIC, 7) 415 #define F2FS_IOC_DEFRAGMENT _IOWR(F2FS_IOCTL_MAGIC, 8, \ 416 struct f2fs_defragment) 417 #define F2FS_IOC_MOVE_RANGE _IOWR(F2FS_IOCTL_MAGIC, 9, \ 418 struct f2fs_move_range) 419 #define F2FS_IOC_FLUSH_DEVICE _IOW(F2FS_IOCTL_MAGIC, 10, \ 420 struct f2fs_flush_device) 421 #define F2FS_IOC_GARBAGE_COLLECT_RANGE _IOW(F2FS_IOCTL_MAGIC, 11, \ 422 struct f2fs_gc_range) 423 #define F2FS_IOC_GET_FEATURES _IOR(F2FS_IOCTL_MAGIC, 12, __u32) 424 #define F2FS_IOC_SET_PIN_FILE _IOW(F2FS_IOCTL_MAGIC, 13, __u32) 425 #define F2FS_IOC_GET_PIN_FILE _IOR(F2FS_IOCTL_MAGIC, 14, __u32) 426 #define F2FS_IOC_PRECACHE_EXTENTS _IO(F2FS_IOCTL_MAGIC, 15) 427 #define F2FS_IOC_RESIZE_FS _IOW(F2FS_IOCTL_MAGIC, 16, __u64) 428 #define F2FS_IOC_GET_COMPRESS_BLOCKS _IOR(F2FS_IOCTL_MAGIC, 17, __u64) 429 #define F2FS_IOC_RELEASE_COMPRESS_BLOCKS \ 430 _IOR(F2FS_IOCTL_MAGIC, 18, __u64) 431 #define F2FS_IOC_RESERVE_COMPRESS_BLOCKS \ 432 _IOR(F2FS_IOCTL_MAGIC, 19, __u64) 433 #define F2FS_IOC_SEC_TRIM_FILE _IOW(F2FS_IOCTL_MAGIC, 20, \ 434 struct f2fs_sectrim_range) 435 436 /* 437 * should be same as XFS_IOC_GOINGDOWN. 438 * Flags for going down operation used by FS_IOC_GOINGDOWN 439 */ 440 #define F2FS_IOC_SHUTDOWN _IOR('X', 125, __u32) /* Shutdown */ 441 #define F2FS_GOING_DOWN_FULLSYNC 0x0 /* going down with full sync */ 442 #define F2FS_GOING_DOWN_METASYNC 0x1 /* going down with metadata */ 443 #define F2FS_GOING_DOWN_NOSYNC 0x2 /* going down */ 444 #define F2FS_GOING_DOWN_METAFLUSH 0x3 /* going down with meta flush */ 445 #define F2FS_GOING_DOWN_NEED_FSCK 0x4 /* going down to trigger fsck */ 446 447 /* 448 * Flags used by F2FS_IOC_SEC_TRIM_FILE 449 */ 450 #define F2FS_TRIM_FILE_DISCARD 0x1 /* send discard command */ 451 #define F2FS_TRIM_FILE_ZEROOUT 0x2 /* zero out */ 452 #define F2FS_TRIM_FILE_MASK 0x3 453 454 struct f2fs_gc_range { 455 u32 sync; 456 u64 start; 457 u64 len; 458 }; 459 460 struct f2fs_defragment { 461 u64 start; 462 u64 len; 463 }; 464 465 struct f2fs_move_range { 466 u32 dst_fd; /* destination fd */ 467 u64 pos_in; /* start position in src_fd */ 468 u64 pos_out; /* start position in dst_fd */ 469 u64 len; /* size to move */ 470 }; 471 472 struct f2fs_flush_device { 473 u32 dev_num; /* device number to flush */ 474 u32 segments; /* # of segments to flush */ 475 }; 476 477 struct f2fs_sectrim_range { 478 u64 start; 479 u64 len; 480 u64 flags; 481 }; 482 483 /* for inline stuff */ 484 #define DEF_INLINE_RESERVED_SIZE 1 485 static inline int get_extra_isize(struct inode *inode); 486 static inline int get_inline_xattr_addrs(struct inode *inode); 487 #define MAX_INLINE_DATA(inode) (sizeof(__le32) * \ 488 (CUR_ADDRS_PER_INODE(inode) - \ 489 get_inline_xattr_addrs(inode) - \ 490 DEF_INLINE_RESERVED_SIZE)) 491 492 /* for inline dir */ 493 #define NR_INLINE_DENTRY(inode) (MAX_INLINE_DATA(inode) * BITS_PER_BYTE / \ 494 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \ 495 BITS_PER_BYTE + 1)) 496 #define INLINE_DENTRY_BITMAP_SIZE(inode) \ 497 DIV_ROUND_UP(NR_INLINE_DENTRY(inode), BITS_PER_BYTE) 498 #define INLINE_RESERVED_SIZE(inode) (MAX_INLINE_DATA(inode) - \ 499 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \ 500 NR_INLINE_DENTRY(inode) + \ 501 INLINE_DENTRY_BITMAP_SIZE(inode))) 502 503 /* 504 * For INODE and NODE manager 505 */ 506 /* for directory operations */ 507 508 struct f2fs_filename { 509 /* 510 * The filename the user specified. This is NULL for some 511 * filesystem-internal operations, e.g. converting an inline directory 512 * to a non-inline one, or roll-forward recovering an encrypted dentry. 513 */ 514 const struct qstr *usr_fname; 515 516 /* 517 * The on-disk filename. For encrypted directories, this is encrypted. 518 * This may be NULL for lookups in an encrypted dir without the key. 519 */ 520 struct fscrypt_str disk_name; 521 522 /* The dirhash of this filename */ 523 f2fs_hash_t hash; 524 525 #ifdef CONFIG_FS_ENCRYPTION 526 /* 527 * For lookups in encrypted directories: either the buffer backing 528 * disk_name, or a buffer that holds the decoded no-key name. 529 */ 530 struct fscrypt_str crypto_buf; 531 #endif 532 #ifdef CONFIG_UNICODE 533 /* 534 * For casefolded directories: the casefolded name, but it's left NULL 535 * if the original name is not valid Unicode or if the filesystem is 536 * doing an internal operation where usr_fname is also NULL. In these 537 * cases we fall back to treating the name as an opaque byte sequence. 538 */ 539 struct fscrypt_str cf_name; 540 #endif 541 }; 542 543 struct f2fs_dentry_ptr { 544 struct inode *inode; 545 void *bitmap; 546 struct f2fs_dir_entry *dentry; 547 __u8 (*filename)[F2FS_SLOT_LEN]; 548 int max; 549 int nr_bitmap; 550 }; 551 552 static inline void make_dentry_ptr_block(struct inode *inode, 553 struct f2fs_dentry_ptr *d, struct f2fs_dentry_block *t) 554 { 555 d->inode = inode; 556 d->max = NR_DENTRY_IN_BLOCK; 557 d->nr_bitmap = SIZE_OF_DENTRY_BITMAP; 558 d->bitmap = t->dentry_bitmap; 559 d->dentry = t->dentry; 560 d->filename = t->filename; 561 } 562 563 static inline void make_dentry_ptr_inline(struct inode *inode, 564 struct f2fs_dentry_ptr *d, void *t) 565 { 566 int entry_cnt = NR_INLINE_DENTRY(inode); 567 int bitmap_size = INLINE_DENTRY_BITMAP_SIZE(inode); 568 int reserved_size = INLINE_RESERVED_SIZE(inode); 569 570 d->inode = inode; 571 d->max = entry_cnt; 572 d->nr_bitmap = bitmap_size; 573 d->bitmap = t; 574 d->dentry = t + bitmap_size + reserved_size; 575 d->filename = t + bitmap_size + reserved_size + 576 SIZE_OF_DIR_ENTRY * entry_cnt; 577 } 578 579 /* 580 * XATTR_NODE_OFFSET stores xattrs to one node block per file keeping -1 581 * as its node offset to distinguish from index node blocks. 582 * But some bits are used to mark the node block. 583 */ 584 #define XATTR_NODE_OFFSET ((((unsigned int)-1) << OFFSET_BIT_SHIFT) \ 585 >> OFFSET_BIT_SHIFT) 586 enum { 587 ALLOC_NODE, /* allocate a new node page if needed */ 588 LOOKUP_NODE, /* look up a node without readahead */ 589 LOOKUP_NODE_RA, /* 590 * look up a node with readahead called 591 * by get_data_block. 592 */ 593 }; 594 595 #define DEFAULT_RETRY_IO_COUNT 8 /* maximum retry read IO count */ 596 597 /* congestion wait timeout value, default: 20ms */ 598 #define DEFAULT_IO_TIMEOUT (msecs_to_jiffies(20)) 599 600 /* maximum retry quota flush count */ 601 #define DEFAULT_RETRY_QUOTA_FLUSH_COUNT 8 602 603 #define F2FS_LINK_MAX 0xffffffff /* maximum link count per file */ 604 605 #define MAX_DIR_RA_PAGES 4 /* maximum ra pages of dir */ 606 607 /* for in-memory extent cache entry */ 608 #define F2FS_MIN_EXTENT_LEN 64 /* minimum extent length */ 609 610 /* number of extent info in extent cache we try to shrink */ 611 #define EXTENT_CACHE_SHRINK_NUMBER 128 612 613 struct rb_entry { 614 struct rb_node rb_node; /* rb node located in rb-tree */ 615 unsigned int ofs; /* start offset of the entry */ 616 unsigned int len; /* length of the entry */ 617 }; 618 619 struct extent_info { 620 unsigned int fofs; /* start offset in a file */ 621 unsigned int len; /* length of the extent */ 622 u32 blk; /* start block address of the extent */ 623 }; 624 625 struct extent_node { 626 struct rb_node rb_node; /* rb node located in rb-tree */ 627 struct extent_info ei; /* extent info */ 628 struct list_head list; /* node in global extent list of sbi */ 629 struct extent_tree *et; /* extent tree pointer */ 630 }; 631 632 struct extent_tree { 633 nid_t ino; /* inode number */ 634 struct rb_root_cached root; /* root of extent info rb-tree */ 635 struct extent_node *cached_en; /* recently accessed extent node */ 636 struct extent_info largest; /* largested extent info */ 637 struct list_head list; /* to be used by sbi->zombie_list */ 638 rwlock_t lock; /* protect extent info rb-tree */ 639 atomic_t node_cnt; /* # of extent node in rb-tree*/ 640 bool largest_updated; /* largest extent updated */ 641 }; 642 643 /* 644 * This structure is taken from ext4_map_blocks. 645 * 646 * Note that, however, f2fs uses NEW and MAPPED flags for f2fs_map_blocks(). 647 */ 648 #define F2FS_MAP_NEW (1 << BH_New) 649 #define F2FS_MAP_MAPPED (1 << BH_Mapped) 650 #define F2FS_MAP_UNWRITTEN (1 << BH_Unwritten) 651 #define F2FS_MAP_FLAGS (F2FS_MAP_NEW | F2FS_MAP_MAPPED |\ 652 F2FS_MAP_UNWRITTEN) 653 654 struct f2fs_map_blocks { 655 block_t m_pblk; 656 block_t m_lblk; 657 unsigned int m_len; 658 unsigned int m_flags; 659 pgoff_t *m_next_pgofs; /* point next possible non-hole pgofs */ 660 pgoff_t *m_next_extent; /* point to next possible extent */ 661 int m_seg_type; 662 bool m_may_create; /* indicate it is from write path */ 663 }; 664 665 /* for flag in get_data_block */ 666 enum { 667 F2FS_GET_BLOCK_DEFAULT, 668 F2FS_GET_BLOCK_FIEMAP, 669 F2FS_GET_BLOCK_BMAP, 670 F2FS_GET_BLOCK_DIO, 671 F2FS_GET_BLOCK_PRE_DIO, 672 F2FS_GET_BLOCK_PRE_AIO, 673 F2FS_GET_BLOCK_PRECACHE, 674 }; 675 676 /* 677 * i_advise uses FADVISE_XXX_BIT. We can add additional hints later. 678 */ 679 #define FADVISE_COLD_BIT 0x01 680 #define FADVISE_LOST_PINO_BIT 0x02 681 #define FADVISE_ENCRYPT_BIT 0x04 682 #define FADVISE_ENC_NAME_BIT 0x08 683 #define FADVISE_KEEP_SIZE_BIT 0x10 684 #define FADVISE_HOT_BIT 0x20 685 #define FADVISE_VERITY_BIT 0x40 686 687 #define FADVISE_MODIFIABLE_BITS (FADVISE_COLD_BIT | FADVISE_HOT_BIT) 688 689 #define file_is_cold(inode) is_file(inode, FADVISE_COLD_BIT) 690 #define file_wrong_pino(inode) is_file(inode, FADVISE_LOST_PINO_BIT) 691 #define file_set_cold(inode) set_file(inode, FADVISE_COLD_BIT) 692 #define file_lost_pino(inode) set_file(inode, FADVISE_LOST_PINO_BIT) 693 #define file_clear_cold(inode) clear_file(inode, FADVISE_COLD_BIT) 694 #define file_got_pino(inode) clear_file(inode, FADVISE_LOST_PINO_BIT) 695 #define file_is_encrypt(inode) is_file(inode, FADVISE_ENCRYPT_BIT) 696 #define file_set_encrypt(inode) set_file(inode, FADVISE_ENCRYPT_BIT) 697 #define file_clear_encrypt(inode) clear_file(inode, FADVISE_ENCRYPT_BIT) 698 #define file_enc_name(inode) is_file(inode, FADVISE_ENC_NAME_BIT) 699 #define file_set_enc_name(inode) set_file(inode, FADVISE_ENC_NAME_BIT) 700 #define file_keep_isize(inode) is_file(inode, FADVISE_KEEP_SIZE_BIT) 701 #define file_set_keep_isize(inode) set_file(inode, FADVISE_KEEP_SIZE_BIT) 702 #define file_is_hot(inode) is_file(inode, FADVISE_HOT_BIT) 703 #define file_set_hot(inode) set_file(inode, FADVISE_HOT_BIT) 704 #define file_clear_hot(inode) clear_file(inode, FADVISE_HOT_BIT) 705 #define file_is_verity(inode) is_file(inode, FADVISE_VERITY_BIT) 706 #define file_set_verity(inode) set_file(inode, FADVISE_VERITY_BIT) 707 708 #define DEF_DIR_LEVEL 0 709 710 enum { 711 GC_FAILURE_PIN, 712 GC_FAILURE_ATOMIC, 713 MAX_GC_FAILURE 714 }; 715 716 /* used for f2fs_inode_info->flags */ 717 enum { 718 FI_NEW_INODE, /* indicate newly allocated inode */ 719 FI_DIRTY_INODE, /* indicate inode is dirty or not */ 720 FI_AUTO_RECOVER, /* indicate inode is recoverable */ 721 FI_DIRTY_DIR, /* indicate directory has dirty pages */ 722 FI_INC_LINK, /* need to increment i_nlink */ 723 FI_ACL_MODE, /* indicate acl mode */ 724 FI_NO_ALLOC, /* should not allocate any blocks */ 725 FI_FREE_NID, /* free allocated nide */ 726 FI_NO_EXTENT, /* not to use the extent cache */ 727 FI_INLINE_XATTR, /* used for inline xattr */ 728 FI_INLINE_DATA, /* used for inline data*/ 729 FI_INLINE_DENTRY, /* used for inline dentry */ 730 FI_APPEND_WRITE, /* inode has appended data */ 731 FI_UPDATE_WRITE, /* inode has in-place-update data */ 732 FI_NEED_IPU, /* used for ipu per file */ 733 FI_ATOMIC_FILE, /* indicate atomic file */ 734 FI_ATOMIC_COMMIT, /* indicate the state of atomical committing */ 735 FI_VOLATILE_FILE, /* indicate volatile file */ 736 FI_FIRST_BLOCK_WRITTEN, /* indicate #0 data block was written */ 737 FI_DROP_CACHE, /* drop dirty page cache */ 738 FI_DATA_EXIST, /* indicate data exists */ 739 FI_INLINE_DOTS, /* indicate inline dot dentries */ 740 FI_DO_DEFRAG, /* indicate defragment is running */ 741 FI_DIRTY_FILE, /* indicate regular/symlink has dirty pages */ 742 FI_NO_PREALLOC, /* indicate skipped preallocated blocks */ 743 FI_HOT_DATA, /* indicate file is hot */ 744 FI_EXTRA_ATTR, /* indicate file has extra attribute */ 745 FI_PROJ_INHERIT, /* indicate file inherits projectid */ 746 FI_PIN_FILE, /* indicate file should not be gced */ 747 FI_ATOMIC_REVOKE_REQUEST, /* request to drop atomic data */ 748 FI_VERITY_IN_PROGRESS, /* building fs-verity Merkle tree */ 749 FI_COMPRESSED_FILE, /* indicate file's data can be compressed */ 750 FI_MMAP_FILE, /* indicate file was mmapped */ 751 FI_MAX, /* max flag, never be used */ 752 }; 753 754 struct f2fs_inode_info { 755 struct inode vfs_inode; /* serve a vfs inode */ 756 unsigned long i_flags; /* keep an inode flags for ioctl */ 757 unsigned char i_advise; /* use to give file attribute hints */ 758 unsigned char i_dir_level; /* use for dentry level for large dir */ 759 unsigned int i_current_depth; /* only for directory depth */ 760 /* for gc failure statistic */ 761 unsigned int i_gc_failures[MAX_GC_FAILURE]; 762 unsigned int i_pino; /* parent inode number */ 763 umode_t i_acl_mode; /* keep file acl mode temporarily */ 764 765 /* Use below internally in f2fs*/ 766 unsigned long flags[BITS_TO_LONGS(FI_MAX)]; /* use to pass per-file flags */ 767 struct rw_semaphore i_sem; /* protect fi info */ 768 atomic_t dirty_pages; /* # of dirty pages */ 769 f2fs_hash_t chash; /* hash value of given file name */ 770 unsigned int clevel; /* maximum level of given file name */ 771 struct task_struct *task; /* lookup and create consistency */ 772 struct task_struct *cp_task; /* separate cp/wb IO stats*/ 773 nid_t i_xattr_nid; /* node id that contains xattrs */ 774 loff_t last_disk_size; /* lastly written file size */ 775 spinlock_t i_size_lock; /* protect last_disk_size */ 776 777 #ifdef CONFIG_QUOTA 778 struct dquot *i_dquot[MAXQUOTAS]; 779 780 /* quota space reservation, managed internally by quota code */ 781 qsize_t i_reserved_quota; 782 #endif 783 struct list_head dirty_list; /* dirty list for dirs and files */ 784 struct list_head gdirty_list; /* linked in global dirty list */ 785 struct list_head inmem_ilist; /* list for inmem inodes */ 786 struct list_head inmem_pages; /* inmemory pages managed by f2fs */ 787 struct task_struct *inmem_task; /* store inmemory task */ 788 struct mutex inmem_lock; /* lock for inmemory pages */ 789 pgoff_t ra_offset; /* ongoing readahead offset */ 790 struct extent_tree *extent_tree; /* cached extent_tree entry */ 791 792 /* avoid racing between foreground op and gc */ 793 struct rw_semaphore i_gc_rwsem[2]; 794 struct rw_semaphore i_mmap_sem; 795 struct rw_semaphore i_xattr_sem; /* avoid racing between reading and changing EAs */ 796 797 int i_extra_isize; /* size of extra space located in i_addr */ 798 kprojid_t i_projid; /* id for project quota */ 799 int i_inline_xattr_size; /* inline xattr size */ 800 struct timespec64 i_crtime; /* inode creation time */ 801 struct timespec64 i_disk_time[4];/* inode disk times */ 802 803 /* for file compress */ 804 u64 i_compr_blocks; /* # of compressed blocks */ 805 unsigned char i_compress_algorithm; /* algorithm type */ 806 unsigned char i_log_cluster_size; /* log of cluster size */ 807 unsigned int i_cluster_size; /* cluster size */ 808 }; 809 810 static inline void get_extent_info(struct extent_info *ext, 811 struct f2fs_extent *i_ext) 812 { 813 ext->fofs = le32_to_cpu(i_ext->fofs); 814 ext->blk = le32_to_cpu(i_ext->blk); 815 ext->len = le32_to_cpu(i_ext->len); 816 } 817 818 static inline void set_raw_extent(struct extent_info *ext, 819 struct f2fs_extent *i_ext) 820 { 821 i_ext->fofs = cpu_to_le32(ext->fofs); 822 i_ext->blk = cpu_to_le32(ext->blk); 823 i_ext->len = cpu_to_le32(ext->len); 824 } 825 826 static inline void set_extent_info(struct extent_info *ei, unsigned int fofs, 827 u32 blk, unsigned int len) 828 { 829 ei->fofs = fofs; 830 ei->blk = blk; 831 ei->len = len; 832 } 833 834 static inline bool __is_discard_mergeable(struct discard_info *back, 835 struct discard_info *front, unsigned int max_len) 836 { 837 return (back->lstart + back->len == front->lstart) && 838 (back->len + front->len <= max_len); 839 } 840 841 static inline bool __is_discard_back_mergeable(struct discard_info *cur, 842 struct discard_info *back, unsigned int max_len) 843 { 844 return __is_discard_mergeable(back, cur, max_len); 845 } 846 847 static inline bool __is_discard_front_mergeable(struct discard_info *cur, 848 struct discard_info *front, unsigned int max_len) 849 { 850 return __is_discard_mergeable(cur, front, max_len); 851 } 852 853 static inline bool __is_extent_mergeable(struct extent_info *back, 854 struct extent_info *front) 855 { 856 return (back->fofs + back->len == front->fofs && 857 back->blk + back->len == front->blk); 858 } 859 860 static inline bool __is_back_mergeable(struct extent_info *cur, 861 struct extent_info *back) 862 { 863 return __is_extent_mergeable(back, cur); 864 } 865 866 static inline bool __is_front_mergeable(struct extent_info *cur, 867 struct extent_info *front) 868 { 869 return __is_extent_mergeable(cur, front); 870 } 871 872 extern void f2fs_mark_inode_dirty_sync(struct inode *inode, bool sync); 873 static inline void __try_update_largest_extent(struct extent_tree *et, 874 struct extent_node *en) 875 { 876 if (en->ei.len > et->largest.len) { 877 et->largest = en->ei; 878 et->largest_updated = true; 879 } 880 } 881 882 /* 883 * For free nid management 884 */ 885 enum nid_state { 886 FREE_NID, /* newly added to free nid list */ 887 PREALLOC_NID, /* it is preallocated */ 888 MAX_NID_STATE, 889 }; 890 891 struct f2fs_nm_info { 892 block_t nat_blkaddr; /* base disk address of NAT */ 893 nid_t max_nid; /* maximum possible node ids */ 894 nid_t available_nids; /* # of available node ids */ 895 nid_t next_scan_nid; /* the next nid to be scanned */ 896 unsigned int ram_thresh; /* control the memory footprint */ 897 unsigned int ra_nid_pages; /* # of nid pages to be readaheaded */ 898 unsigned int dirty_nats_ratio; /* control dirty nats ratio threshold */ 899 900 /* NAT cache management */ 901 struct radix_tree_root nat_root;/* root of the nat entry cache */ 902 struct radix_tree_root nat_set_root;/* root of the nat set cache */ 903 struct rw_semaphore nat_tree_lock; /* protect nat_tree_lock */ 904 struct list_head nat_entries; /* cached nat entry list (clean) */ 905 spinlock_t nat_list_lock; /* protect clean nat entry list */ 906 unsigned int nat_cnt; /* the # of cached nat entries */ 907 unsigned int dirty_nat_cnt; /* total num of nat entries in set */ 908 unsigned int nat_blocks; /* # of nat blocks */ 909 910 /* free node ids management */ 911 struct radix_tree_root free_nid_root;/* root of the free_nid cache */ 912 struct list_head free_nid_list; /* list for free nids excluding preallocated nids */ 913 unsigned int nid_cnt[MAX_NID_STATE]; /* the number of free node id */ 914 spinlock_t nid_list_lock; /* protect nid lists ops */ 915 struct mutex build_lock; /* lock for build free nids */ 916 unsigned char **free_nid_bitmap; 917 unsigned char *nat_block_bitmap; 918 unsigned short *free_nid_count; /* free nid count of NAT block */ 919 920 /* for checkpoint */ 921 char *nat_bitmap; /* NAT bitmap pointer */ 922 923 unsigned int nat_bits_blocks; /* # of nat bits blocks */ 924 unsigned char *nat_bits; /* NAT bits blocks */ 925 unsigned char *full_nat_bits; /* full NAT pages */ 926 unsigned char *empty_nat_bits; /* empty NAT pages */ 927 #ifdef CONFIG_F2FS_CHECK_FS 928 char *nat_bitmap_mir; /* NAT bitmap mirror */ 929 #endif 930 int bitmap_size; /* bitmap size */ 931 }; 932 933 /* 934 * this structure is used as one of function parameters. 935 * all the information are dedicated to a given direct node block determined 936 * by the data offset in a file. 937 */ 938 struct dnode_of_data { 939 struct inode *inode; /* vfs inode pointer */ 940 struct page *inode_page; /* its inode page, NULL is possible */ 941 struct page *node_page; /* cached direct node page */ 942 nid_t nid; /* node id of the direct node block */ 943 unsigned int ofs_in_node; /* data offset in the node page */ 944 bool inode_page_locked; /* inode page is locked or not */ 945 bool node_changed; /* is node block changed */ 946 char cur_level; /* level of hole node page */ 947 char max_level; /* level of current page located */ 948 block_t data_blkaddr; /* block address of the node block */ 949 }; 950 951 static inline void set_new_dnode(struct dnode_of_data *dn, struct inode *inode, 952 struct page *ipage, struct page *npage, nid_t nid) 953 { 954 memset(dn, 0, sizeof(*dn)); 955 dn->inode = inode; 956 dn->inode_page = ipage; 957 dn->node_page = npage; 958 dn->nid = nid; 959 } 960 961 /* 962 * For SIT manager 963 * 964 * By default, there are 6 active log areas across the whole main area. 965 * When considering hot and cold data separation to reduce cleaning overhead, 966 * we split 3 for data logs and 3 for node logs as hot, warm, and cold types, 967 * respectively. 968 * In the current design, you should not change the numbers intentionally. 969 * Instead, as a mount option such as active_logs=x, you can use 2, 4, and 6 970 * logs individually according to the underlying devices. (default: 6) 971 * Just in case, on-disk layout covers maximum 16 logs that consist of 8 for 972 * data and 8 for node logs. 973 */ 974 #define NR_CURSEG_DATA_TYPE (3) 975 #define NR_CURSEG_NODE_TYPE (3) 976 #define NR_CURSEG_TYPE (NR_CURSEG_DATA_TYPE + NR_CURSEG_NODE_TYPE) 977 978 enum { 979 CURSEG_HOT_DATA = 0, /* directory entry blocks */ 980 CURSEG_WARM_DATA, /* data blocks */ 981 CURSEG_COLD_DATA, /* multimedia or GCed data blocks */ 982 CURSEG_HOT_NODE, /* direct node blocks of directory files */ 983 CURSEG_WARM_NODE, /* direct node blocks of normal files */ 984 CURSEG_COLD_NODE, /* indirect node blocks */ 985 NO_CHECK_TYPE, 986 CURSEG_COLD_DATA_PINNED,/* cold data for pinned file */ 987 }; 988 989 struct flush_cmd { 990 struct completion wait; 991 struct llist_node llnode; 992 nid_t ino; 993 int ret; 994 }; 995 996 struct flush_cmd_control { 997 struct task_struct *f2fs_issue_flush; /* flush thread */ 998 wait_queue_head_t flush_wait_queue; /* waiting queue for wake-up */ 999 atomic_t issued_flush; /* # of issued flushes */ 1000 atomic_t queued_flush; /* # of queued flushes */ 1001 struct llist_head issue_list; /* list for command issue */ 1002 struct llist_node *dispatch_list; /* list for command dispatch */ 1003 }; 1004 1005 struct f2fs_sm_info { 1006 struct sit_info *sit_info; /* whole segment information */ 1007 struct free_segmap_info *free_info; /* free segment information */ 1008 struct dirty_seglist_info *dirty_info; /* dirty segment information */ 1009 struct curseg_info *curseg_array; /* active segment information */ 1010 1011 struct rw_semaphore curseg_lock; /* for preventing curseg change */ 1012 1013 block_t seg0_blkaddr; /* block address of 0'th segment */ 1014 block_t main_blkaddr; /* start block address of main area */ 1015 block_t ssa_blkaddr; /* start block address of SSA area */ 1016 1017 unsigned int segment_count; /* total # of segments */ 1018 unsigned int main_segments; /* # of segments in main area */ 1019 unsigned int reserved_segments; /* # of reserved segments */ 1020 unsigned int ovp_segments; /* # of overprovision segments */ 1021 1022 /* a threshold to reclaim prefree segments */ 1023 unsigned int rec_prefree_segments; 1024 1025 /* for batched trimming */ 1026 unsigned int trim_sections; /* # of sections to trim */ 1027 1028 struct list_head sit_entry_set; /* sit entry set list */ 1029 1030 unsigned int ipu_policy; /* in-place-update policy */ 1031 unsigned int min_ipu_util; /* in-place-update threshold */ 1032 unsigned int min_fsync_blocks; /* threshold for fsync */ 1033 unsigned int min_seq_blocks; /* threshold for sequential blocks */ 1034 unsigned int min_hot_blocks; /* threshold for hot block allocation */ 1035 unsigned int min_ssr_sections; /* threshold to trigger SSR allocation */ 1036 1037 /* for flush command control */ 1038 struct flush_cmd_control *fcc_info; 1039 1040 /* for discard command control */ 1041 struct discard_cmd_control *dcc_info; 1042 }; 1043 1044 /* 1045 * For superblock 1046 */ 1047 /* 1048 * COUNT_TYPE for monitoring 1049 * 1050 * f2fs monitors the number of several block types such as on-writeback, 1051 * dirty dentry blocks, dirty node blocks, and dirty meta blocks. 1052 */ 1053 #define WB_DATA_TYPE(p) (__is_cp_guaranteed(p) ? F2FS_WB_CP_DATA : F2FS_WB_DATA) 1054 enum count_type { 1055 F2FS_DIRTY_DENTS, 1056 F2FS_DIRTY_DATA, 1057 F2FS_DIRTY_QDATA, 1058 F2FS_DIRTY_NODES, 1059 F2FS_DIRTY_META, 1060 F2FS_INMEM_PAGES, 1061 F2FS_DIRTY_IMETA, 1062 F2FS_WB_CP_DATA, 1063 F2FS_WB_DATA, 1064 F2FS_RD_DATA, 1065 F2FS_RD_NODE, 1066 F2FS_RD_META, 1067 F2FS_DIO_WRITE, 1068 F2FS_DIO_READ, 1069 NR_COUNT_TYPE, 1070 }; 1071 1072 /* 1073 * The below are the page types of bios used in submit_bio(). 1074 * The available types are: 1075 * DATA User data pages. It operates as async mode. 1076 * NODE Node pages. It operates as async mode. 1077 * META FS metadata pages such as SIT, NAT, CP. 1078 * NR_PAGE_TYPE The number of page types. 1079 * META_FLUSH Make sure the previous pages are written 1080 * with waiting the bio's completion 1081 * ... Only can be used with META. 1082 */ 1083 #define PAGE_TYPE_OF_BIO(type) ((type) > META ? META : (type)) 1084 enum page_type { 1085 DATA, 1086 NODE, 1087 META, 1088 NR_PAGE_TYPE, 1089 META_FLUSH, 1090 INMEM, /* the below types are used by tracepoints only. */ 1091 INMEM_DROP, 1092 INMEM_INVALIDATE, 1093 INMEM_REVOKE, 1094 IPU, 1095 OPU, 1096 }; 1097 1098 enum temp_type { 1099 HOT = 0, /* must be zero for meta bio */ 1100 WARM, 1101 COLD, 1102 NR_TEMP_TYPE, 1103 }; 1104 1105 enum need_lock_type { 1106 LOCK_REQ = 0, 1107 LOCK_DONE, 1108 LOCK_RETRY, 1109 }; 1110 1111 enum cp_reason_type { 1112 CP_NO_NEEDED, 1113 CP_NON_REGULAR, 1114 CP_COMPRESSED, 1115 CP_HARDLINK, 1116 CP_SB_NEED_CP, 1117 CP_WRONG_PINO, 1118 CP_NO_SPC_ROLL, 1119 CP_NODE_NEED_CP, 1120 CP_FASTBOOT_MODE, 1121 CP_SPEC_LOG_NUM, 1122 CP_RECOVER_DIR, 1123 }; 1124 1125 enum iostat_type { 1126 /* WRITE IO */ 1127 APP_DIRECT_IO, /* app direct write IOs */ 1128 APP_BUFFERED_IO, /* app buffered write IOs */ 1129 APP_WRITE_IO, /* app write IOs */ 1130 APP_MAPPED_IO, /* app mapped IOs */ 1131 FS_DATA_IO, /* data IOs from kworker/fsync/reclaimer */ 1132 FS_NODE_IO, /* node IOs from kworker/fsync/reclaimer */ 1133 FS_META_IO, /* meta IOs from kworker/reclaimer */ 1134 FS_GC_DATA_IO, /* data IOs from forground gc */ 1135 FS_GC_NODE_IO, /* node IOs from forground gc */ 1136 FS_CP_DATA_IO, /* data IOs from checkpoint */ 1137 FS_CP_NODE_IO, /* node IOs from checkpoint */ 1138 FS_CP_META_IO, /* meta IOs from checkpoint */ 1139 1140 /* READ IO */ 1141 APP_DIRECT_READ_IO, /* app direct read IOs */ 1142 APP_BUFFERED_READ_IO, /* app buffered read IOs */ 1143 APP_READ_IO, /* app read IOs */ 1144 APP_MAPPED_READ_IO, /* app mapped read IOs */ 1145 FS_DATA_READ_IO, /* data read IOs */ 1146 FS_GDATA_READ_IO, /* data read IOs from background gc */ 1147 FS_CDATA_READ_IO, /* compressed data read IOs */ 1148 FS_NODE_READ_IO, /* node read IOs */ 1149 FS_META_READ_IO, /* meta read IOs */ 1150 1151 /* other */ 1152 FS_DISCARD, /* discard */ 1153 NR_IO_TYPE, 1154 }; 1155 1156 struct f2fs_io_info { 1157 struct f2fs_sb_info *sbi; /* f2fs_sb_info pointer */ 1158 nid_t ino; /* inode number */ 1159 enum page_type type; /* contains DATA/NODE/META/META_FLUSH */ 1160 enum temp_type temp; /* contains HOT/WARM/COLD */ 1161 int op; /* contains REQ_OP_ */ 1162 int op_flags; /* req_flag_bits */ 1163 block_t new_blkaddr; /* new block address to be written */ 1164 block_t old_blkaddr; /* old block address before Cow */ 1165 struct page *page; /* page to be written */ 1166 struct page *encrypted_page; /* encrypted page */ 1167 struct page *compressed_page; /* compressed page */ 1168 struct list_head list; /* serialize IOs */ 1169 bool submitted; /* indicate IO submission */ 1170 int need_lock; /* indicate we need to lock cp_rwsem */ 1171 bool in_list; /* indicate fio is in io_list */ 1172 bool is_por; /* indicate IO is from recovery or not */ 1173 bool retry; /* need to reallocate block address */ 1174 int compr_blocks; /* # of compressed block addresses */ 1175 bool encrypted; /* indicate file is encrypted */ 1176 enum iostat_type io_type; /* io type */ 1177 struct writeback_control *io_wbc; /* writeback control */ 1178 struct bio **bio; /* bio for ipu */ 1179 sector_t *last_block; /* last block number in bio */ 1180 unsigned char version; /* version of the node */ 1181 }; 1182 1183 struct bio_entry { 1184 struct bio *bio; 1185 struct list_head list; 1186 }; 1187 1188 #define is_read_io(rw) ((rw) == READ) 1189 struct f2fs_bio_info { 1190 struct f2fs_sb_info *sbi; /* f2fs superblock */ 1191 struct bio *bio; /* bios to merge */ 1192 sector_t last_block_in_bio; /* last block number */ 1193 struct f2fs_io_info fio; /* store buffered io info. */ 1194 struct rw_semaphore io_rwsem; /* blocking op for bio */ 1195 spinlock_t io_lock; /* serialize DATA/NODE IOs */ 1196 struct list_head io_list; /* track fios */ 1197 struct list_head bio_list; /* bio entry list head */ 1198 struct rw_semaphore bio_list_lock; /* lock to protect bio entry list */ 1199 }; 1200 1201 #define FDEV(i) (sbi->devs[i]) 1202 #define RDEV(i) (raw_super->devs[i]) 1203 struct f2fs_dev_info { 1204 struct block_device *bdev; 1205 char path[MAX_PATH_LEN]; 1206 unsigned int total_segments; 1207 block_t start_blk; 1208 block_t end_blk; 1209 #ifdef CONFIG_BLK_DEV_ZONED 1210 unsigned int nr_blkz; /* Total number of zones */ 1211 unsigned long *blkz_seq; /* Bitmap indicating sequential zones */ 1212 #endif 1213 }; 1214 1215 enum inode_type { 1216 DIR_INODE, /* for dirty dir inode */ 1217 FILE_INODE, /* for dirty regular/symlink inode */ 1218 DIRTY_META, /* for all dirtied inode metadata */ 1219 ATOMIC_FILE, /* for all atomic files */ 1220 NR_INODE_TYPE, 1221 }; 1222 1223 /* for inner inode cache management */ 1224 struct inode_management { 1225 struct radix_tree_root ino_root; /* ino entry array */ 1226 spinlock_t ino_lock; /* for ino entry lock */ 1227 struct list_head ino_list; /* inode list head */ 1228 unsigned long ino_num; /* number of entries */ 1229 }; 1230 1231 /* For s_flag in struct f2fs_sb_info */ 1232 enum { 1233 SBI_IS_DIRTY, /* dirty flag for checkpoint */ 1234 SBI_IS_CLOSE, /* specify unmounting */ 1235 SBI_NEED_FSCK, /* need fsck.f2fs to fix */ 1236 SBI_POR_DOING, /* recovery is doing or not */ 1237 SBI_NEED_SB_WRITE, /* need to recover superblock */ 1238 SBI_NEED_CP, /* need to checkpoint */ 1239 SBI_IS_SHUTDOWN, /* shutdown by ioctl */ 1240 SBI_IS_RECOVERED, /* recovered orphan/data */ 1241 SBI_CP_DISABLED, /* CP was disabled last mount */ 1242 SBI_CP_DISABLED_QUICK, /* CP was disabled quickly */ 1243 SBI_QUOTA_NEED_FLUSH, /* need to flush quota info in CP */ 1244 SBI_QUOTA_SKIP_FLUSH, /* skip flushing quota in current CP */ 1245 SBI_QUOTA_NEED_REPAIR, /* quota file may be corrupted */ 1246 SBI_IS_RESIZEFS, /* resizefs is in process */ 1247 }; 1248 1249 enum { 1250 CP_TIME, 1251 REQ_TIME, 1252 DISCARD_TIME, 1253 GC_TIME, 1254 DISABLE_TIME, 1255 UMOUNT_DISCARD_TIMEOUT, 1256 MAX_TIME, 1257 }; 1258 1259 enum { 1260 GC_NORMAL, 1261 GC_IDLE_CB, 1262 GC_IDLE_GREEDY, 1263 GC_URGENT_HIGH, 1264 GC_URGENT_LOW, 1265 }; 1266 1267 enum { 1268 BGGC_MODE_ON, /* background gc is on */ 1269 BGGC_MODE_OFF, /* background gc is off */ 1270 BGGC_MODE_SYNC, /* 1271 * background gc is on, migrating blocks 1272 * like foreground gc 1273 */ 1274 }; 1275 1276 enum { 1277 FS_MODE_ADAPTIVE, /* use both lfs/ssr allocation */ 1278 FS_MODE_LFS, /* use lfs allocation only */ 1279 }; 1280 1281 enum { 1282 WHINT_MODE_OFF, /* not pass down write hints */ 1283 WHINT_MODE_USER, /* try to pass down hints given by users */ 1284 WHINT_MODE_FS, /* pass down hints with F2FS policy */ 1285 }; 1286 1287 enum { 1288 ALLOC_MODE_DEFAULT, /* stay default */ 1289 ALLOC_MODE_REUSE, /* reuse segments as much as possible */ 1290 }; 1291 1292 enum fsync_mode { 1293 FSYNC_MODE_POSIX, /* fsync follows posix semantics */ 1294 FSYNC_MODE_STRICT, /* fsync behaves in line with ext4 */ 1295 FSYNC_MODE_NOBARRIER, /* fsync behaves nobarrier based on posix */ 1296 }; 1297 1298 /* 1299 * this value is set in page as a private data which indicate that 1300 * the page is atomically written, and it is in inmem_pages list. 1301 */ 1302 #define ATOMIC_WRITTEN_PAGE ((unsigned long)-1) 1303 #define DUMMY_WRITTEN_PAGE ((unsigned long)-2) 1304 1305 #define IS_ATOMIC_WRITTEN_PAGE(page) \ 1306 (page_private(page) == (unsigned long)ATOMIC_WRITTEN_PAGE) 1307 #define IS_DUMMY_WRITTEN_PAGE(page) \ 1308 (page_private(page) == (unsigned long)DUMMY_WRITTEN_PAGE) 1309 1310 #ifdef CONFIG_F2FS_IO_TRACE 1311 #define IS_IO_TRACED_PAGE(page) \ 1312 (page_private(page) > 0 && \ 1313 page_private(page) < (unsigned long)PID_MAX_LIMIT) 1314 #else 1315 #define IS_IO_TRACED_PAGE(page) (0) 1316 #endif 1317 1318 /* For compression */ 1319 enum compress_algorithm_type { 1320 COMPRESS_LZO, 1321 COMPRESS_LZ4, 1322 COMPRESS_ZSTD, 1323 COMPRESS_LZORLE, 1324 COMPRESS_MAX, 1325 }; 1326 1327 #define COMPRESS_DATA_RESERVED_SIZE 5 1328 struct compress_data { 1329 __le32 clen; /* compressed data size */ 1330 __le32 reserved[COMPRESS_DATA_RESERVED_SIZE]; /* reserved */ 1331 u8 cdata[]; /* compressed data */ 1332 }; 1333 1334 #define COMPRESS_HEADER_SIZE (sizeof(struct compress_data)) 1335 1336 #define F2FS_COMPRESSED_PAGE_MAGIC 0xF5F2C000 1337 1338 /* compress context */ 1339 struct compress_ctx { 1340 struct inode *inode; /* inode the context belong to */ 1341 pgoff_t cluster_idx; /* cluster index number */ 1342 unsigned int cluster_size; /* page count in cluster */ 1343 unsigned int log_cluster_size; /* log of cluster size */ 1344 struct page **rpages; /* pages store raw data in cluster */ 1345 unsigned int nr_rpages; /* total page number in rpages */ 1346 struct page **cpages; /* pages store compressed data in cluster */ 1347 unsigned int nr_cpages; /* total page number in cpages */ 1348 void *rbuf; /* virtual mapped address on rpages */ 1349 struct compress_data *cbuf; /* virtual mapped address on cpages */ 1350 size_t rlen; /* valid data length in rbuf */ 1351 size_t clen; /* valid data length in cbuf */ 1352 void *private; /* payload buffer for specified compression algorithm */ 1353 void *private2; /* extra payload buffer */ 1354 }; 1355 1356 /* compress context for write IO path */ 1357 struct compress_io_ctx { 1358 u32 magic; /* magic number to indicate page is compressed */ 1359 struct inode *inode; /* inode the context belong to */ 1360 struct page **rpages; /* pages store raw data in cluster */ 1361 unsigned int nr_rpages; /* total page number in rpages */ 1362 refcount_t ref; /* referrence count of raw page */ 1363 }; 1364 1365 /* decompress io context for read IO path */ 1366 struct decompress_io_ctx { 1367 u32 magic; /* magic number to indicate page is compressed */ 1368 struct inode *inode; /* inode the context belong to */ 1369 pgoff_t cluster_idx; /* cluster index number */ 1370 unsigned int cluster_size; /* page count in cluster */ 1371 unsigned int log_cluster_size; /* log of cluster size */ 1372 struct page **rpages; /* pages store raw data in cluster */ 1373 unsigned int nr_rpages; /* total page number in rpages */ 1374 struct page **cpages; /* pages store compressed data in cluster */ 1375 unsigned int nr_cpages; /* total page number in cpages */ 1376 struct page **tpages; /* temp pages to pad holes in cluster */ 1377 void *rbuf; /* virtual mapped address on rpages */ 1378 struct compress_data *cbuf; /* virtual mapped address on cpages */ 1379 size_t rlen; /* valid data length in rbuf */ 1380 size_t clen; /* valid data length in cbuf */ 1381 refcount_t ref; /* referrence count of compressed page */ 1382 bool failed; /* indicate IO error during decompression */ 1383 void *private; /* payload buffer for specified decompression algorithm */ 1384 void *private2; /* extra payload buffer */ 1385 }; 1386 1387 #define NULL_CLUSTER ((unsigned int)(~0)) 1388 #define MIN_COMPRESS_LOG_SIZE 2 1389 #define MAX_COMPRESS_LOG_SIZE 8 1390 #define MAX_COMPRESS_WINDOW_SIZE ((PAGE_SIZE) << MAX_COMPRESS_LOG_SIZE) 1391 1392 struct f2fs_sb_info { 1393 struct super_block *sb; /* pointer to VFS super block */ 1394 struct proc_dir_entry *s_proc; /* proc entry */ 1395 struct f2fs_super_block *raw_super; /* raw super block pointer */ 1396 struct rw_semaphore sb_lock; /* lock for raw super block */ 1397 int valid_super_block; /* valid super block no */ 1398 unsigned long s_flag; /* flags for sbi */ 1399 struct mutex writepages; /* mutex for writepages() */ 1400 #ifdef CONFIG_UNICODE 1401 struct unicode_map *s_encoding; 1402 __u16 s_encoding_flags; 1403 #endif 1404 1405 #ifdef CONFIG_BLK_DEV_ZONED 1406 unsigned int blocks_per_blkz; /* F2FS blocks per zone */ 1407 unsigned int log_blocks_per_blkz; /* log2 F2FS blocks per zone */ 1408 #endif 1409 1410 /* for node-related operations */ 1411 struct f2fs_nm_info *nm_info; /* node manager */ 1412 struct inode *node_inode; /* cache node blocks */ 1413 1414 /* for segment-related operations */ 1415 struct f2fs_sm_info *sm_info; /* segment manager */ 1416 1417 /* for bio operations */ 1418 struct f2fs_bio_info *write_io[NR_PAGE_TYPE]; /* for write bios */ 1419 /* keep migration IO order for LFS mode */ 1420 struct rw_semaphore io_order_lock; 1421 mempool_t *write_io_dummy; /* Dummy pages */ 1422 1423 /* for checkpoint */ 1424 struct f2fs_checkpoint *ckpt; /* raw checkpoint pointer */ 1425 int cur_cp_pack; /* remain current cp pack */ 1426 spinlock_t cp_lock; /* for flag in ckpt */ 1427 struct inode *meta_inode; /* cache meta blocks */ 1428 struct mutex cp_mutex; /* checkpoint procedure lock */ 1429 struct rw_semaphore cp_rwsem; /* blocking FS operations */ 1430 struct rw_semaphore node_write; /* locking node writes */ 1431 struct rw_semaphore node_change; /* locking node change */ 1432 wait_queue_head_t cp_wait; 1433 unsigned long last_time[MAX_TIME]; /* to store time in jiffies */ 1434 long interval_time[MAX_TIME]; /* to store thresholds */ 1435 1436 struct inode_management im[MAX_INO_ENTRY]; /* manage inode cache */ 1437 1438 spinlock_t fsync_node_lock; /* for node entry lock */ 1439 struct list_head fsync_node_list; /* node list head */ 1440 unsigned int fsync_seg_id; /* sequence id */ 1441 unsigned int fsync_node_num; /* number of node entries */ 1442 1443 /* for orphan inode, use 0'th array */ 1444 unsigned int max_orphans; /* max orphan inodes */ 1445 1446 /* for inode management */ 1447 struct list_head inode_list[NR_INODE_TYPE]; /* dirty inode list */ 1448 spinlock_t inode_lock[NR_INODE_TYPE]; /* for dirty inode list lock */ 1449 struct mutex flush_lock; /* for flush exclusion */ 1450 1451 /* for extent tree cache */ 1452 struct radix_tree_root extent_tree_root;/* cache extent cache entries */ 1453 struct mutex extent_tree_lock; /* locking extent radix tree */ 1454 struct list_head extent_list; /* lru list for shrinker */ 1455 spinlock_t extent_lock; /* locking extent lru list */ 1456 atomic_t total_ext_tree; /* extent tree count */ 1457 struct list_head zombie_list; /* extent zombie tree list */ 1458 atomic_t total_zombie_tree; /* extent zombie tree count */ 1459 atomic_t total_ext_node; /* extent info count */ 1460 1461 /* basic filesystem units */ 1462 unsigned int log_sectors_per_block; /* log2 sectors per block */ 1463 unsigned int log_blocksize; /* log2 block size */ 1464 unsigned int blocksize; /* block size */ 1465 unsigned int root_ino_num; /* root inode number*/ 1466 unsigned int node_ino_num; /* node inode number*/ 1467 unsigned int meta_ino_num; /* meta inode number*/ 1468 unsigned int log_blocks_per_seg; /* log2 blocks per segment */ 1469 unsigned int blocks_per_seg; /* blocks per segment */ 1470 unsigned int segs_per_sec; /* segments per section */ 1471 unsigned int secs_per_zone; /* sections per zone */ 1472 unsigned int total_sections; /* total section count */ 1473 unsigned int total_node_count; /* total node block count */ 1474 unsigned int total_valid_node_count; /* valid node block count */ 1475 loff_t max_file_blocks; /* max block index of file */ 1476 int dir_level; /* directory level */ 1477 int readdir_ra; /* readahead inode in readdir */ 1478 1479 block_t user_block_count; /* # of user blocks */ 1480 block_t total_valid_block_count; /* # of valid blocks */ 1481 block_t discard_blks; /* discard command candidats */ 1482 block_t last_valid_block_count; /* for recovery */ 1483 block_t reserved_blocks; /* configurable reserved blocks */ 1484 block_t current_reserved_blocks; /* current reserved blocks */ 1485 1486 /* Additional tracking for no checkpoint mode */ 1487 block_t unusable_block_count; /* # of blocks saved by last cp */ 1488 1489 unsigned int nquota_files; /* # of quota sysfile */ 1490 struct rw_semaphore quota_sem; /* blocking cp for flags */ 1491 1492 /* # of pages, see count_type */ 1493 atomic_t nr_pages[NR_COUNT_TYPE]; 1494 /* # of allocated blocks */ 1495 struct percpu_counter alloc_valid_block_count; 1496 1497 /* writeback control */ 1498 atomic_t wb_sync_req[META]; /* count # of WB_SYNC threads */ 1499 1500 /* valid inode count */ 1501 struct percpu_counter total_valid_inode_count; 1502 1503 struct f2fs_mount_info mount_opt; /* mount options */ 1504 1505 /* for cleaning operations */ 1506 struct rw_semaphore gc_lock; /* 1507 * semaphore for GC, avoid 1508 * race between GC and GC or CP 1509 */ 1510 struct f2fs_gc_kthread *gc_thread; /* GC thread */ 1511 unsigned int cur_victim_sec; /* current victim section num */ 1512 unsigned int gc_mode; /* current GC state */ 1513 unsigned int next_victim_seg[2]; /* next segment in victim section */ 1514 1515 /* for skip statistic */ 1516 unsigned int atomic_files; /* # of opened atomic file */ 1517 unsigned long long skipped_atomic_files[2]; /* FG_GC and BG_GC */ 1518 unsigned long long skipped_gc_rwsem; /* FG_GC only */ 1519 1520 /* threshold for gc trials on pinned files */ 1521 u64 gc_pin_file_threshold; 1522 struct rw_semaphore pin_sem; 1523 1524 /* maximum # of trials to find a victim segment for SSR and GC */ 1525 unsigned int max_victim_search; 1526 /* migration granularity of garbage collection, unit: segment */ 1527 unsigned int migration_granularity; 1528 1529 /* 1530 * for stat information. 1531 * one is for the LFS mode, and the other is for the SSR mode. 1532 */ 1533 #ifdef CONFIG_F2FS_STAT_FS 1534 struct f2fs_stat_info *stat_info; /* FS status information */ 1535 atomic_t meta_count[META_MAX]; /* # of meta blocks */ 1536 unsigned int segment_count[2]; /* # of allocated segments */ 1537 unsigned int block_count[2]; /* # of allocated blocks */ 1538 atomic_t inplace_count; /* # of inplace update */ 1539 atomic64_t total_hit_ext; /* # of lookup extent cache */ 1540 atomic64_t read_hit_rbtree; /* # of hit rbtree extent node */ 1541 atomic64_t read_hit_largest; /* # of hit largest extent node */ 1542 atomic64_t read_hit_cached; /* # of hit cached extent node */ 1543 atomic_t inline_xattr; /* # of inline_xattr inodes */ 1544 atomic_t inline_inode; /* # of inline_data inodes */ 1545 atomic_t inline_dir; /* # of inline_dentry inodes */ 1546 atomic_t compr_inode; /* # of compressed inodes */ 1547 atomic_t compr_blocks; /* # of compressed blocks */ 1548 atomic_t vw_cnt; /* # of volatile writes */ 1549 atomic_t max_aw_cnt; /* max # of atomic writes */ 1550 atomic_t max_vw_cnt; /* max # of volatile writes */ 1551 unsigned int io_skip_bggc; /* skip background gc for in-flight IO */ 1552 unsigned int other_skip_bggc; /* skip background gc for other reasons */ 1553 unsigned int ndirty_inode[NR_INODE_TYPE]; /* # of dirty inodes */ 1554 #endif 1555 spinlock_t stat_lock; /* lock for stat operations */ 1556 1557 /* For app/fs IO statistics */ 1558 spinlock_t iostat_lock; 1559 unsigned long long rw_iostat[NR_IO_TYPE]; 1560 unsigned long long prev_rw_iostat[NR_IO_TYPE]; 1561 bool iostat_enable; 1562 unsigned long iostat_next_period; 1563 unsigned int iostat_period_ms; 1564 1565 /* to attach REQ_META|REQ_FUA flags */ 1566 unsigned int data_io_flag; 1567 unsigned int node_io_flag; 1568 1569 /* For sysfs suppport */ 1570 struct kobject s_kobj; 1571 struct completion s_kobj_unregister; 1572 1573 /* For shrinker support */ 1574 struct list_head s_list; 1575 int s_ndevs; /* number of devices */ 1576 struct f2fs_dev_info *devs; /* for device list */ 1577 unsigned int dirty_device; /* for checkpoint data flush */ 1578 spinlock_t dev_lock; /* protect dirty_device */ 1579 struct mutex umount_mutex; 1580 unsigned int shrinker_run_no; 1581 1582 /* For write statistics */ 1583 u64 sectors_written_start; 1584 u64 kbytes_written; 1585 1586 /* Reference to checksum algorithm driver via cryptoapi */ 1587 struct crypto_shash *s_chksum_driver; 1588 1589 /* Precomputed FS UUID checksum for seeding other checksums */ 1590 __u32 s_chksum_seed; 1591 1592 struct workqueue_struct *post_read_wq; /* post read workqueue */ 1593 1594 struct kmem_cache *inline_xattr_slab; /* inline xattr entry */ 1595 unsigned int inline_xattr_slab_size; /* default inline xattr slab size */ 1596 }; 1597 1598 struct f2fs_private_dio { 1599 struct inode *inode; 1600 void *orig_private; 1601 bio_end_io_t *orig_end_io; 1602 bool write; 1603 }; 1604 1605 #ifdef CONFIG_F2FS_FAULT_INJECTION 1606 #define f2fs_show_injection_info(sbi, type) \ 1607 printk_ratelimited("%sF2FS-fs (%s) : inject %s in %s of %pS\n", \ 1608 KERN_INFO, sbi->sb->s_id, \ 1609 f2fs_fault_name[type], \ 1610 __func__, __builtin_return_address(0)) 1611 static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type) 1612 { 1613 struct f2fs_fault_info *ffi = &F2FS_OPTION(sbi).fault_info; 1614 1615 if (!ffi->inject_rate) 1616 return false; 1617 1618 if (!IS_FAULT_SET(ffi, type)) 1619 return false; 1620 1621 atomic_inc(&ffi->inject_ops); 1622 if (atomic_read(&ffi->inject_ops) >= ffi->inject_rate) { 1623 atomic_set(&ffi->inject_ops, 0); 1624 return true; 1625 } 1626 return false; 1627 } 1628 #else 1629 #define f2fs_show_injection_info(sbi, type) do { } while (0) 1630 static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type) 1631 { 1632 return false; 1633 } 1634 #endif 1635 1636 /* 1637 * Test if the mounted volume is a multi-device volume. 1638 * - For a single regular disk volume, sbi->s_ndevs is 0. 1639 * - For a single zoned disk volume, sbi->s_ndevs is 1. 1640 * - For a multi-device volume, sbi->s_ndevs is always 2 or more. 1641 */ 1642 static inline bool f2fs_is_multi_device(struct f2fs_sb_info *sbi) 1643 { 1644 return sbi->s_ndevs > 1; 1645 } 1646 1647 /* For write statistics. Suppose sector size is 512 bytes, 1648 * and the return value is in kbytes. s is of struct f2fs_sb_info. 1649 */ 1650 #define BD_PART_WRITTEN(s) \ 1651 (((u64)part_stat_read((s)->sb->s_bdev->bd_part, sectors[STAT_WRITE]) - \ 1652 (s)->sectors_written_start) >> 1) 1653 1654 static inline void f2fs_update_time(struct f2fs_sb_info *sbi, int type) 1655 { 1656 unsigned long now = jiffies; 1657 1658 sbi->last_time[type] = now; 1659 1660 /* DISCARD_TIME and GC_TIME are based on REQ_TIME */ 1661 if (type == REQ_TIME) { 1662 sbi->last_time[DISCARD_TIME] = now; 1663 sbi->last_time[GC_TIME] = now; 1664 } 1665 } 1666 1667 static inline bool f2fs_time_over(struct f2fs_sb_info *sbi, int type) 1668 { 1669 unsigned long interval = sbi->interval_time[type] * HZ; 1670 1671 return time_after(jiffies, sbi->last_time[type] + interval); 1672 } 1673 1674 static inline unsigned int f2fs_time_to_wait(struct f2fs_sb_info *sbi, 1675 int type) 1676 { 1677 unsigned long interval = sbi->interval_time[type] * HZ; 1678 unsigned int wait_ms = 0; 1679 long delta; 1680 1681 delta = (sbi->last_time[type] + interval) - jiffies; 1682 if (delta > 0) 1683 wait_ms = jiffies_to_msecs(delta); 1684 1685 return wait_ms; 1686 } 1687 1688 /* 1689 * Inline functions 1690 */ 1691 static inline u32 __f2fs_crc32(struct f2fs_sb_info *sbi, u32 crc, 1692 const void *address, unsigned int length) 1693 { 1694 struct { 1695 struct shash_desc shash; 1696 char ctx[4]; 1697 } desc; 1698 int err; 1699 1700 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver) != sizeof(desc.ctx)); 1701 1702 desc.shash.tfm = sbi->s_chksum_driver; 1703 *(u32 *)desc.ctx = crc; 1704 1705 err = crypto_shash_update(&desc.shash, address, length); 1706 BUG_ON(err); 1707 1708 return *(u32 *)desc.ctx; 1709 } 1710 1711 static inline u32 f2fs_crc32(struct f2fs_sb_info *sbi, const void *address, 1712 unsigned int length) 1713 { 1714 return __f2fs_crc32(sbi, F2FS_SUPER_MAGIC, address, length); 1715 } 1716 1717 static inline bool f2fs_crc_valid(struct f2fs_sb_info *sbi, __u32 blk_crc, 1718 void *buf, size_t buf_size) 1719 { 1720 return f2fs_crc32(sbi, buf, buf_size) == blk_crc; 1721 } 1722 1723 static inline u32 f2fs_chksum(struct f2fs_sb_info *sbi, u32 crc, 1724 const void *address, unsigned int length) 1725 { 1726 return __f2fs_crc32(sbi, crc, address, length); 1727 } 1728 1729 static inline struct f2fs_inode_info *F2FS_I(struct inode *inode) 1730 { 1731 return container_of(inode, struct f2fs_inode_info, vfs_inode); 1732 } 1733 1734 static inline struct f2fs_sb_info *F2FS_SB(struct super_block *sb) 1735 { 1736 return sb->s_fs_info; 1737 } 1738 1739 static inline struct f2fs_sb_info *F2FS_I_SB(struct inode *inode) 1740 { 1741 return F2FS_SB(inode->i_sb); 1742 } 1743 1744 static inline struct f2fs_sb_info *F2FS_M_SB(struct address_space *mapping) 1745 { 1746 return F2FS_I_SB(mapping->host); 1747 } 1748 1749 static inline struct f2fs_sb_info *F2FS_P_SB(struct page *page) 1750 { 1751 return F2FS_M_SB(page_file_mapping(page)); 1752 } 1753 1754 static inline struct f2fs_super_block *F2FS_RAW_SUPER(struct f2fs_sb_info *sbi) 1755 { 1756 return (struct f2fs_super_block *)(sbi->raw_super); 1757 } 1758 1759 static inline struct f2fs_checkpoint *F2FS_CKPT(struct f2fs_sb_info *sbi) 1760 { 1761 return (struct f2fs_checkpoint *)(sbi->ckpt); 1762 } 1763 1764 static inline struct f2fs_node *F2FS_NODE(struct page *page) 1765 { 1766 return (struct f2fs_node *)page_address(page); 1767 } 1768 1769 static inline struct f2fs_inode *F2FS_INODE(struct page *page) 1770 { 1771 return &((struct f2fs_node *)page_address(page))->i; 1772 } 1773 1774 static inline struct f2fs_nm_info *NM_I(struct f2fs_sb_info *sbi) 1775 { 1776 return (struct f2fs_nm_info *)(sbi->nm_info); 1777 } 1778 1779 static inline struct f2fs_sm_info *SM_I(struct f2fs_sb_info *sbi) 1780 { 1781 return (struct f2fs_sm_info *)(sbi->sm_info); 1782 } 1783 1784 static inline struct sit_info *SIT_I(struct f2fs_sb_info *sbi) 1785 { 1786 return (struct sit_info *)(SM_I(sbi)->sit_info); 1787 } 1788 1789 static inline struct free_segmap_info *FREE_I(struct f2fs_sb_info *sbi) 1790 { 1791 return (struct free_segmap_info *)(SM_I(sbi)->free_info); 1792 } 1793 1794 static inline struct dirty_seglist_info *DIRTY_I(struct f2fs_sb_info *sbi) 1795 { 1796 return (struct dirty_seglist_info *)(SM_I(sbi)->dirty_info); 1797 } 1798 1799 static inline struct address_space *META_MAPPING(struct f2fs_sb_info *sbi) 1800 { 1801 return sbi->meta_inode->i_mapping; 1802 } 1803 1804 static inline struct address_space *NODE_MAPPING(struct f2fs_sb_info *sbi) 1805 { 1806 return sbi->node_inode->i_mapping; 1807 } 1808 1809 static inline bool is_sbi_flag_set(struct f2fs_sb_info *sbi, unsigned int type) 1810 { 1811 return test_bit(type, &sbi->s_flag); 1812 } 1813 1814 static inline void set_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type) 1815 { 1816 set_bit(type, &sbi->s_flag); 1817 } 1818 1819 static inline void clear_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type) 1820 { 1821 clear_bit(type, &sbi->s_flag); 1822 } 1823 1824 static inline unsigned long long cur_cp_version(struct f2fs_checkpoint *cp) 1825 { 1826 return le64_to_cpu(cp->checkpoint_ver); 1827 } 1828 1829 static inline unsigned long f2fs_qf_ino(struct super_block *sb, int type) 1830 { 1831 if (type < F2FS_MAX_QUOTAS) 1832 return le32_to_cpu(F2FS_SB(sb)->raw_super->qf_ino[type]); 1833 return 0; 1834 } 1835 1836 static inline __u64 cur_cp_crc(struct f2fs_checkpoint *cp) 1837 { 1838 size_t crc_offset = le32_to_cpu(cp->checksum_offset); 1839 return le32_to_cpu(*((__le32 *)((unsigned char *)cp + crc_offset))); 1840 } 1841 1842 static inline bool __is_set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 1843 { 1844 unsigned int ckpt_flags = le32_to_cpu(cp->ckpt_flags); 1845 1846 return ckpt_flags & f; 1847 } 1848 1849 static inline bool is_set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 1850 { 1851 return __is_set_ckpt_flags(F2FS_CKPT(sbi), f); 1852 } 1853 1854 static inline void __set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 1855 { 1856 unsigned int ckpt_flags; 1857 1858 ckpt_flags = le32_to_cpu(cp->ckpt_flags); 1859 ckpt_flags |= f; 1860 cp->ckpt_flags = cpu_to_le32(ckpt_flags); 1861 } 1862 1863 static inline void set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 1864 { 1865 unsigned long flags; 1866 1867 spin_lock_irqsave(&sbi->cp_lock, flags); 1868 __set_ckpt_flags(F2FS_CKPT(sbi), f); 1869 spin_unlock_irqrestore(&sbi->cp_lock, flags); 1870 } 1871 1872 static inline void __clear_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 1873 { 1874 unsigned int ckpt_flags; 1875 1876 ckpt_flags = le32_to_cpu(cp->ckpt_flags); 1877 ckpt_flags &= (~f); 1878 cp->ckpt_flags = cpu_to_le32(ckpt_flags); 1879 } 1880 1881 static inline void clear_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 1882 { 1883 unsigned long flags; 1884 1885 spin_lock_irqsave(&sbi->cp_lock, flags); 1886 __clear_ckpt_flags(F2FS_CKPT(sbi), f); 1887 spin_unlock_irqrestore(&sbi->cp_lock, flags); 1888 } 1889 1890 static inline void disable_nat_bits(struct f2fs_sb_info *sbi, bool lock) 1891 { 1892 unsigned long flags; 1893 unsigned char *nat_bits; 1894 1895 /* 1896 * In order to re-enable nat_bits we need to call fsck.f2fs by 1897 * set_sbi_flag(sbi, SBI_NEED_FSCK). But it may give huge cost, 1898 * so let's rely on regular fsck or unclean shutdown. 1899 */ 1900 1901 if (lock) 1902 spin_lock_irqsave(&sbi->cp_lock, flags); 1903 __clear_ckpt_flags(F2FS_CKPT(sbi), CP_NAT_BITS_FLAG); 1904 nat_bits = NM_I(sbi)->nat_bits; 1905 NM_I(sbi)->nat_bits = NULL; 1906 if (lock) 1907 spin_unlock_irqrestore(&sbi->cp_lock, flags); 1908 1909 kvfree(nat_bits); 1910 } 1911 1912 static inline bool enabled_nat_bits(struct f2fs_sb_info *sbi, 1913 struct cp_control *cpc) 1914 { 1915 bool set = is_set_ckpt_flags(sbi, CP_NAT_BITS_FLAG); 1916 1917 return (cpc) ? (cpc->reason & CP_UMOUNT) && set : set; 1918 } 1919 1920 static inline void f2fs_lock_op(struct f2fs_sb_info *sbi) 1921 { 1922 down_read(&sbi->cp_rwsem); 1923 } 1924 1925 static inline int f2fs_trylock_op(struct f2fs_sb_info *sbi) 1926 { 1927 return down_read_trylock(&sbi->cp_rwsem); 1928 } 1929 1930 static inline void f2fs_unlock_op(struct f2fs_sb_info *sbi) 1931 { 1932 up_read(&sbi->cp_rwsem); 1933 } 1934 1935 static inline void f2fs_lock_all(struct f2fs_sb_info *sbi) 1936 { 1937 down_write(&sbi->cp_rwsem); 1938 } 1939 1940 static inline void f2fs_unlock_all(struct f2fs_sb_info *sbi) 1941 { 1942 up_write(&sbi->cp_rwsem); 1943 } 1944 1945 static inline int __get_cp_reason(struct f2fs_sb_info *sbi) 1946 { 1947 int reason = CP_SYNC; 1948 1949 if (test_opt(sbi, FASTBOOT)) 1950 reason = CP_FASTBOOT; 1951 if (is_sbi_flag_set(sbi, SBI_IS_CLOSE)) 1952 reason = CP_UMOUNT; 1953 return reason; 1954 } 1955 1956 static inline bool __remain_node_summaries(int reason) 1957 { 1958 return (reason & (CP_UMOUNT | CP_FASTBOOT)); 1959 } 1960 1961 static inline bool __exist_node_summaries(struct f2fs_sb_info *sbi) 1962 { 1963 return (is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG) || 1964 is_set_ckpt_flags(sbi, CP_FASTBOOT_FLAG)); 1965 } 1966 1967 /* 1968 * Check whether the inode has blocks or not 1969 */ 1970 static inline int F2FS_HAS_BLOCKS(struct inode *inode) 1971 { 1972 block_t xattr_block = F2FS_I(inode)->i_xattr_nid ? 1 : 0; 1973 1974 return (inode->i_blocks >> F2FS_LOG_SECTORS_PER_BLOCK) > xattr_block; 1975 } 1976 1977 static inline bool f2fs_has_xattr_block(unsigned int ofs) 1978 { 1979 return ofs == XATTR_NODE_OFFSET; 1980 } 1981 1982 static inline bool __allow_reserved_blocks(struct f2fs_sb_info *sbi, 1983 struct inode *inode, bool cap) 1984 { 1985 if (!inode) 1986 return true; 1987 if (!test_opt(sbi, RESERVE_ROOT)) 1988 return false; 1989 if (IS_NOQUOTA(inode)) 1990 return true; 1991 if (uid_eq(F2FS_OPTION(sbi).s_resuid, current_fsuid())) 1992 return true; 1993 if (!gid_eq(F2FS_OPTION(sbi).s_resgid, GLOBAL_ROOT_GID) && 1994 in_group_p(F2FS_OPTION(sbi).s_resgid)) 1995 return true; 1996 if (cap && capable(CAP_SYS_RESOURCE)) 1997 return true; 1998 return false; 1999 } 2000 2001 static inline void f2fs_i_blocks_write(struct inode *, block_t, bool, bool); 2002 static inline int inc_valid_block_count(struct f2fs_sb_info *sbi, 2003 struct inode *inode, blkcnt_t *count) 2004 { 2005 blkcnt_t diff = 0, release = 0; 2006 block_t avail_user_block_count; 2007 int ret; 2008 2009 ret = dquot_reserve_block(inode, *count); 2010 if (ret) 2011 return ret; 2012 2013 if (time_to_inject(sbi, FAULT_BLOCK)) { 2014 f2fs_show_injection_info(sbi, FAULT_BLOCK); 2015 release = *count; 2016 goto release_quota; 2017 } 2018 2019 /* 2020 * let's increase this in prior to actual block count change in order 2021 * for f2fs_sync_file to avoid data races when deciding checkpoint. 2022 */ 2023 percpu_counter_add(&sbi->alloc_valid_block_count, (*count)); 2024 2025 spin_lock(&sbi->stat_lock); 2026 sbi->total_valid_block_count += (block_t)(*count); 2027 avail_user_block_count = sbi->user_block_count - 2028 sbi->current_reserved_blocks; 2029 2030 if (!__allow_reserved_blocks(sbi, inode, true)) 2031 avail_user_block_count -= F2FS_OPTION(sbi).root_reserved_blocks; 2032 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) { 2033 if (avail_user_block_count > sbi->unusable_block_count) 2034 avail_user_block_count -= sbi->unusable_block_count; 2035 else 2036 avail_user_block_count = 0; 2037 } 2038 if (unlikely(sbi->total_valid_block_count > avail_user_block_count)) { 2039 diff = sbi->total_valid_block_count - avail_user_block_count; 2040 if (diff > *count) 2041 diff = *count; 2042 *count -= diff; 2043 release = diff; 2044 sbi->total_valid_block_count -= diff; 2045 if (!*count) { 2046 spin_unlock(&sbi->stat_lock); 2047 goto enospc; 2048 } 2049 } 2050 spin_unlock(&sbi->stat_lock); 2051 2052 if (unlikely(release)) { 2053 percpu_counter_sub(&sbi->alloc_valid_block_count, release); 2054 dquot_release_reservation_block(inode, release); 2055 } 2056 f2fs_i_blocks_write(inode, *count, true, true); 2057 return 0; 2058 2059 enospc: 2060 percpu_counter_sub(&sbi->alloc_valid_block_count, release); 2061 release_quota: 2062 dquot_release_reservation_block(inode, release); 2063 return -ENOSPC; 2064 } 2065 2066 __printf(2, 3) 2067 void f2fs_printk(struct f2fs_sb_info *sbi, const char *fmt, ...); 2068 2069 #define f2fs_err(sbi, fmt, ...) \ 2070 f2fs_printk(sbi, KERN_ERR fmt, ##__VA_ARGS__) 2071 #define f2fs_warn(sbi, fmt, ...) \ 2072 f2fs_printk(sbi, KERN_WARNING fmt, ##__VA_ARGS__) 2073 #define f2fs_notice(sbi, fmt, ...) \ 2074 f2fs_printk(sbi, KERN_NOTICE fmt, ##__VA_ARGS__) 2075 #define f2fs_info(sbi, fmt, ...) \ 2076 f2fs_printk(sbi, KERN_INFO fmt, ##__VA_ARGS__) 2077 #define f2fs_debug(sbi, fmt, ...) \ 2078 f2fs_printk(sbi, KERN_DEBUG fmt, ##__VA_ARGS__) 2079 2080 static inline void dec_valid_block_count(struct f2fs_sb_info *sbi, 2081 struct inode *inode, 2082 block_t count) 2083 { 2084 blkcnt_t sectors = count << F2FS_LOG_SECTORS_PER_BLOCK; 2085 2086 spin_lock(&sbi->stat_lock); 2087 f2fs_bug_on(sbi, sbi->total_valid_block_count < (block_t) count); 2088 sbi->total_valid_block_count -= (block_t)count; 2089 if (sbi->reserved_blocks && 2090 sbi->current_reserved_blocks < sbi->reserved_blocks) 2091 sbi->current_reserved_blocks = min(sbi->reserved_blocks, 2092 sbi->current_reserved_blocks + count); 2093 spin_unlock(&sbi->stat_lock); 2094 if (unlikely(inode->i_blocks < sectors)) { 2095 f2fs_warn(sbi, "Inconsistent i_blocks, ino:%lu, iblocks:%llu, sectors:%llu", 2096 inode->i_ino, 2097 (unsigned long long)inode->i_blocks, 2098 (unsigned long long)sectors); 2099 set_sbi_flag(sbi, SBI_NEED_FSCK); 2100 return; 2101 } 2102 f2fs_i_blocks_write(inode, count, false, true); 2103 } 2104 2105 static inline void inc_page_count(struct f2fs_sb_info *sbi, int count_type) 2106 { 2107 atomic_inc(&sbi->nr_pages[count_type]); 2108 2109 if (count_type == F2FS_DIRTY_DENTS || 2110 count_type == F2FS_DIRTY_NODES || 2111 count_type == F2FS_DIRTY_META || 2112 count_type == F2FS_DIRTY_QDATA || 2113 count_type == F2FS_DIRTY_IMETA) 2114 set_sbi_flag(sbi, SBI_IS_DIRTY); 2115 } 2116 2117 static inline void inode_inc_dirty_pages(struct inode *inode) 2118 { 2119 atomic_inc(&F2FS_I(inode)->dirty_pages); 2120 inc_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ? 2121 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA); 2122 if (IS_NOQUOTA(inode)) 2123 inc_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA); 2124 } 2125 2126 static inline void dec_page_count(struct f2fs_sb_info *sbi, int count_type) 2127 { 2128 atomic_dec(&sbi->nr_pages[count_type]); 2129 } 2130 2131 static inline void inode_dec_dirty_pages(struct inode *inode) 2132 { 2133 if (!S_ISDIR(inode->i_mode) && !S_ISREG(inode->i_mode) && 2134 !S_ISLNK(inode->i_mode)) 2135 return; 2136 2137 atomic_dec(&F2FS_I(inode)->dirty_pages); 2138 dec_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ? 2139 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA); 2140 if (IS_NOQUOTA(inode)) 2141 dec_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA); 2142 } 2143 2144 static inline s64 get_pages(struct f2fs_sb_info *sbi, int count_type) 2145 { 2146 return atomic_read(&sbi->nr_pages[count_type]); 2147 } 2148 2149 static inline int get_dirty_pages(struct inode *inode) 2150 { 2151 return atomic_read(&F2FS_I(inode)->dirty_pages); 2152 } 2153 2154 static inline int get_blocktype_secs(struct f2fs_sb_info *sbi, int block_type) 2155 { 2156 unsigned int pages_per_sec = sbi->segs_per_sec * sbi->blocks_per_seg; 2157 unsigned int segs = (get_pages(sbi, block_type) + pages_per_sec - 1) >> 2158 sbi->log_blocks_per_seg; 2159 2160 return segs / sbi->segs_per_sec; 2161 } 2162 2163 static inline block_t valid_user_blocks(struct f2fs_sb_info *sbi) 2164 { 2165 return sbi->total_valid_block_count; 2166 } 2167 2168 static inline block_t discard_blocks(struct f2fs_sb_info *sbi) 2169 { 2170 return sbi->discard_blks; 2171 } 2172 2173 static inline unsigned long __bitmap_size(struct f2fs_sb_info *sbi, int flag) 2174 { 2175 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi); 2176 2177 /* return NAT or SIT bitmap */ 2178 if (flag == NAT_BITMAP) 2179 return le32_to_cpu(ckpt->nat_ver_bitmap_bytesize); 2180 else if (flag == SIT_BITMAP) 2181 return le32_to_cpu(ckpt->sit_ver_bitmap_bytesize); 2182 2183 return 0; 2184 } 2185 2186 static inline block_t __cp_payload(struct f2fs_sb_info *sbi) 2187 { 2188 return le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_payload); 2189 } 2190 2191 static inline void *__bitmap_ptr(struct f2fs_sb_info *sbi, int flag) 2192 { 2193 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi); 2194 int offset; 2195 2196 if (is_set_ckpt_flags(sbi, CP_LARGE_NAT_BITMAP_FLAG)) { 2197 offset = (flag == SIT_BITMAP) ? 2198 le32_to_cpu(ckpt->nat_ver_bitmap_bytesize) : 0; 2199 /* 2200 * if large_nat_bitmap feature is enabled, leave checksum 2201 * protection for all nat/sit bitmaps. 2202 */ 2203 return &ckpt->sit_nat_version_bitmap + offset + sizeof(__le32); 2204 } 2205 2206 if (__cp_payload(sbi) > 0) { 2207 if (flag == NAT_BITMAP) 2208 return &ckpt->sit_nat_version_bitmap; 2209 else 2210 return (unsigned char *)ckpt + F2FS_BLKSIZE; 2211 } else { 2212 offset = (flag == NAT_BITMAP) ? 2213 le32_to_cpu(ckpt->sit_ver_bitmap_bytesize) : 0; 2214 return &ckpt->sit_nat_version_bitmap + offset; 2215 } 2216 } 2217 2218 static inline block_t __start_cp_addr(struct f2fs_sb_info *sbi) 2219 { 2220 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr); 2221 2222 if (sbi->cur_cp_pack == 2) 2223 start_addr += sbi->blocks_per_seg; 2224 return start_addr; 2225 } 2226 2227 static inline block_t __start_cp_next_addr(struct f2fs_sb_info *sbi) 2228 { 2229 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr); 2230 2231 if (sbi->cur_cp_pack == 1) 2232 start_addr += sbi->blocks_per_seg; 2233 return start_addr; 2234 } 2235 2236 static inline void __set_cp_next_pack(struct f2fs_sb_info *sbi) 2237 { 2238 sbi->cur_cp_pack = (sbi->cur_cp_pack == 1) ? 2 : 1; 2239 } 2240 2241 static inline block_t __start_sum_addr(struct f2fs_sb_info *sbi) 2242 { 2243 return le32_to_cpu(F2FS_CKPT(sbi)->cp_pack_start_sum); 2244 } 2245 2246 static inline int inc_valid_node_count(struct f2fs_sb_info *sbi, 2247 struct inode *inode, bool is_inode) 2248 { 2249 block_t valid_block_count; 2250 unsigned int valid_node_count, user_block_count; 2251 int err; 2252 2253 if (is_inode) { 2254 if (inode) { 2255 err = dquot_alloc_inode(inode); 2256 if (err) 2257 return err; 2258 } 2259 } else { 2260 err = dquot_reserve_block(inode, 1); 2261 if (err) 2262 return err; 2263 } 2264 2265 if (time_to_inject(sbi, FAULT_BLOCK)) { 2266 f2fs_show_injection_info(sbi, FAULT_BLOCK); 2267 goto enospc; 2268 } 2269 2270 spin_lock(&sbi->stat_lock); 2271 2272 valid_block_count = sbi->total_valid_block_count + 2273 sbi->current_reserved_blocks + 1; 2274 2275 if (!__allow_reserved_blocks(sbi, inode, false)) 2276 valid_block_count += F2FS_OPTION(sbi).root_reserved_blocks; 2277 user_block_count = sbi->user_block_count; 2278 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) 2279 user_block_count -= sbi->unusable_block_count; 2280 2281 if (unlikely(valid_block_count > user_block_count)) { 2282 spin_unlock(&sbi->stat_lock); 2283 goto enospc; 2284 } 2285 2286 valid_node_count = sbi->total_valid_node_count + 1; 2287 if (unlikely(valid_node_count > sbi->total_node_count)) { 2288 spin_unlock(&sbi->stat_lock); 2289 goto enospc; 2290 } 2291 2292 sbi->total_valid_node_count++; 2293 sbi->total_valid_block_count++; 2294 spin_unlock(&sbi->stat_lock); 2295 2296 if (inode) { 2297 if (is_inode) 2298 f2fs_mark_inode_dirty_sync(inode, true); 2299 else 2300 f2fs_i_blocks_write(inode, 1, true, true); 2301 } 2302 2303 percpu_counter_inc(&sbi->alloc_valid_block_count); 2304 return 0; 2305 2306 enospc: 2307 if (is_inode) { 2308 if (inode) 2309 dquot_free_inode(inode); 2310 } else { 2311 dquot_release_reservation_block(inode, 1); 2312 } 2313 return -ENOSPC; 2314 } 2315 2316 static inline void dec_valid_node_count(struct f2fs_sb_info *sbi, 2317 struct inode *inode, bool is_inode) 2318 { 2319 spin_lock(&sbi->stat_lock); 2320 2321 f2fs_bug_on(sbi, !sbi->total_valid_block_count); 2322 f2fs_bug_on(sbi, !sbi->total_valid_node_count); 2323 2324 sbi->total_valid_node_count--; 2325 sbi->total_valid_block_count--; 2326 if (sbi->reserved_blocks && 2327 sbi->current_reserved_blocks < sbi->reserved_blocks) 2328 sbi->current_reserved_blocks++; 2329 2330 spin_unlock(&sbi->stat_lock); 2331 2332 if (is_inode) { 2333 dquot_free_inode(inode); 2334 } else { 2335 if (unlikely(inode->i_blocks == 0)) { 2336 f2fs_warn(sbi, "dec_valid_node_count: inconsistent i_blocks, ino:%lu, iblocks:%llu", 2337 inode->i_ino, 2338 (unsigned long long)inode->i_blocks); 2339 set_sbi_flag(sbi, SBI_NEED_FSCK); 2340 return; 2341 } 2342 f2fs_i_blocks_write(inode, 1, false, true); 2343 } 2344 } 2345 2346 static inline unsigned int valid_node_count(struct f2fs_sb_info *sbi) 2347 { 2348 return sbi->total_valid_node_count; 2349 } 2350 2351 static inline void inc_valid_inode_count(struct f2fs_sb_info *sbi) 2352 { 2353 percpu_counter_inc(&sbi->total_valid_inode_count); 2354 } 2355 2356 static inline void dec_valid_inode_count(struct f2fs_sb_info *sbi) 2357 { 2358 percpu_counter_dec(&sbi->total_valid_inode_count); 2359 } 2360 2361 static inline s64 valid_inode_count(struct f2fs_sb_info *sbi) 2362 { 2363 return percpu_counter_sum_positive(&sbi->total_valid_inode_count); 2364 } 2365 2366 static inline struct page *f2fs_grab_cache_page(struct address_space *mapping, 2367 pgoff_t index, bool for_write) 2368 { 2369 struct page *page; 2370 2371 if (IS_ENABLED(CONFIG_F2FS_FAULT_INJECTION)) { 2372 if (!for_write) 2373 page = find_get_page_flags(mapping, index, 2374 FGP_LOCK | FGP_ACCESSED); 2375 else 2376 page = find_lock_page(mapping, index); 2377 if (page) 2378 return page; 2379 2380 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_ALLOC)) { 2381 f2fs_show_injection_info(F2FS_M_SB(mapping), 2382 FAULT_PAGE_ALLOC); 2383 return NULL; 2384 } 2385 } 2386 2387 if (!for_write) 2388 return grab_cache_page(mapping, index); 2389 return grab_cache_page_write_begin(mapping, index, AOP_FLAG_NOFS); 2390 } 2391 2392 static inline struct page *f2fs_pagecache_get_page( 2393 struct address_space *mapping, pgoff_t index, 2394 int fgp_flags, gfp_t gfp_mask) 2395 { 2396 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_GET)) { 2397 f2fs_show_injection_info(F2FS_M_SB(mapping), FAULT_PAGE_GET); 2398 return NULL; 2399 } 2400 2401 return pagecache_get_page(mapping, index, fgp_flags, gfp_mask); 2402 } 2403 2404 static inline void f2fs_copy_page(struct page *src, struct page *dst) 2405 { 2406 char *src_kaddr = kmap(src); 2407 char *dst_kaddr = kmap(dst); 2408 2409 memcpy(dst_kaddr, src_kaddr, PAGE_SIZE); 2410 kunmap(dst); 2411 kunmap(src); 2412 } 2413 2414 static inline void f2fs_put_page(struct page *page, int unlock) 2415 { 2416 if (!page) 2417 return; 2418 2419 if (unlock) { 2420 f2fs_bug_on(F2FS_P_SB(page), !PageLocked(page)); 2421 unlock_page(page); 2422 } 2423 put_page(page); 2424 } 2425 2426 static inline void f2fs_put_dnode(struct dnode_of_data *dn) 2427 { 2428 if (dn->node_page) 2429 f2fs_put_page(dn->node_page, 1); 2430 if (dn->inode_page && dn->node_page != dn->inode_page) 2431 f2fs_put_page(dn->inode_page, 0); 2432 dn->node_page = NULL; 2433 dn->inode_page = NULL; 2434 } 2435 2436 static inline struct kmem_cache *f2fs_kmem_cache_create(const char *name, 2437 size_t size) 2438 { 2439 return kmem_cache_create(name, size, 0, SLAB_RECLAIM_ACCOUNT, NULL); 2440 } 2441 2442 static inline void *f2fs_kmem_cache_alloc(struct kmem_cache *cachep, 2443 gfp_t flags) 2444 { 2445 void *entry; 2446 2447 entry = kmem_cache_alloc(cachep, flags); 2448 if (!entry) 2449 entry = kmem_cache_alloc(cachep, flags | __GFP_NOFAIL); 2450 return entry; 2451 } 2452 2453 static inline bool is_idle(struct f2fs_sb_info *sbi, int type) 2454 { 2455 if (sbi->gc_mode == GC_URGENT_HIGH) 2456 return true; 2457 2458 if (get_pages(sbi, F2FS_RD_DATA) || get_pages(sbi, F2FS_RD_NODE) || 2459 get_pages(sbi, F2FS_RD_META) || get_pages(sbi, F2FS_WB_DATA) || 2460 get_pages(sbi, F2FS_WB_CP_DATA) || 2461 get_pages(sbi, F2FS_DIO_READ) || 2462 get_pages(sbi, F2FS_DIO_WRITE)) 2463 return false; 2464 2465 if (type != DISCARD_TIME && SM_I(sbi) && SM_I(sbi)->dcc_info && 2466 atomic_read(&SM_I(sbi)->dcc_info->queued_discard)) 2467 return false; 2468 2469 if (SM_I(sbi) && SM_I(sbi)->fcc_info && 2470 atomic_read(&SM_I(sbi)->fcc_info->queued_flush)) 2471 return false; 2472 2473 if (sbi->gc_mode == GC_URGENT_LOW && 2474 (type == DISCARD_TIME || type == GC_TIME)) 2475 return true; 2476 2477 return f2fs_time_over(sbi, type); 2478 } 2479 2480 static inline void f2fs_radix_tree_insert(struct radix_tree_root *root, 2481 unsigned long index, void *item) 2482 { 2483 while (radix_tree_insert(root, index, item)) 2484 cond_resched(); 2485 } 2486 2487 #define RAW_IS_INODE(p) ((p)->footer.nid == (p)->footer.ino) 2488 2489 static inline bool IS_INODE(struct page *page) 2490 { 2491 struct f2fs_node *p = F2FS_NODE(page); 2492 2493 return RAW_IS_INODE(p); 2494 } 2495 2496 static inline int offset_in_addr(struct f2fs_inode *i) 2497 { 2498 return (i->i_inline & F2FS_EXTRA_ATTR) ? 2499 (le16_to_cpu(i->i_extra_isize) / sizeof(__le32)) : 0; 2500 } 2501 2502 static inline __le32 *blkaddr_in_node(struct f2fs_node *node) 2503 { 2504 return RAW_IS_INODE(node) ? node->i.i_addr : node->dn.addr; 2505 } 2506 2507 static inline int f2fs_has_extra_attr(struct inode *inode); 2508 static inline block_t data_blkaddr(struct inode *inode, 2509 struct page *node_page, unsigned int offset) 2510 { 2511 struct f2fs_node *raw_node; 2512 __le32 *addr_array; 2513 int base = 0; 2514 bool is_inode = IS_INODE(node_page); 2515 2516 raw_node = F2FS_NODE(node_page); 2517 2518 if (is_inode) { 2519 if (!inode) 2520 /* from GC path only */ 2521 base = offset_in_addr(&raw_node->i); 2522 else if (f2fs_has_extra_attr(inode)) 2523 base = get_extra_isize(inode); 2524 } 2525 2526 addr_array = blkaddr_in_node(raw_node); 2527 return le32_to_cpu(addr_array[base + offset]); 2528 } 2529 2530 static inline block_t f2fs_data_blkaddr(struct dnode_of_data *dn) 2531 { 2532 return data_blkaddr(dn->inode, dn->node_page, dn->ofs_in_node); 2533 } 2534 2535 static inline int f2fs_test_bit(unsigned int nr, char *addr) 2536 { 2537 int mask; 2538 2539 addr += (nr >> 3); 2540 mask = 1 << (7 - (nr & 0x07)); 2541 return mask & *addr; 2542 } 2543 2544 static inline void f2fs_set_bit(unsigned int nr, char *addr) 2545 { 2546 int mask; 2547 2548 addr += (nr >> 3); 2549 mask = 1 << (7 - (nr & 0x07)); 2550 *addr |= mask; 2551 } 2552 2553 static inline void f2fs_clear_bit(unsigned int nr, char *addr) 2554 { 2555 int mask; 2556 2557 addr += (nr >> 3); 2558 mask = 1 << (7 - (nr & 0x07)); 2559 *addr &= ~mask; 2560 } 2561 2562 static inline int f2fs_test_and_set_bit(unsigned int nr, char *addr) 2563 { 2564 int mask; 2565 int ret; 2566 2567 addr += (nr >> 3); 2568 mask = 1 << (7 - (nr & 0x07)); 2569 ret = mask & *addr; 2570 *addr |= mask; 2571 return ret; 2572 } 2573 2574 static inline int f2fs_test_and_clear_bit(unsigned int nr, char *addr) 2575 { 2576 int mask; 2577 int ret; 2578 2579 addr += (nr >> 3); 2580 mask = 1 << (7 - (nr & 0x07)); 2581 ret = mask & *addr; 2582 *addr &= ~mask; 2583 return ret; 2584 } 2585 2586 static inline void f2fs_change_bit(unsigned int nr, char *addr) 2587 { 2588 int mask; 2589 2590 addr += (nr >> 3); 2591 mask = 1 << (7 - (nr & 0x07)); 2592 *addr ^= mask; 2593 } 2594 2595 /* 2596 * On-disk inode flags (f2fs_inode::i_flags) 2597 */ 2598 #define F2FS_COMPR_FL 0x00000004 /* Compress file */ 2599 #define F2FS_SYNC_FL 0x00000008 /* Synchronous updates */ 2600 #define F2FS_IMMUTABLE_FL 0x00000010 /* Immutable file */ 2601 #define F2FS_APPEND_FL 0x00000020 /* writes to file may only append */ 2602 #define F2FS_NODUMP_FL 0x00000040 /* do not dump file */ 2603 #define F2FS_NOATIME_FL 0x00000080 /* do not update atime */ 2604 #define F2FS_NOCOMP_FL 0x00000400 /* Don't compress */ 2605 #define F2FS_INDEX_FL 0x00001000 /* hash-indexed directory */ 2606 #define F2FS_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */ 2607 #define F2FS_PROJINHERIT_FL 0x20000000 /* Create with parents projid */ 2608 #define F2FS_CASEFOLD_FL 0x40000000 /* Casefolded file */ 2609 2610 /* Flags that should be inherited by new inodes from their parent. */ 2611 #define F2FS_FL_INHERITED (F2FS_SYNC_FL | F2FS_NODUMP_FL | F2FS_NOATIME_FL | \ 2612 F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \ 2613 F2FS_CASEFOLD_FL | F2FS_COMPR_FL | F2FS_NOCOMP_FL) 2614 2615 /* Flags that are appropriate for regular files (all but dir-specific ones). */ 2616 #define F2FS_REG_FLMASK (~(F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \ 2617 F2FS_CASEFOLD_FL)) 2618 2619 /* Flags that are appropriate for non-directories/regular files. */ 2620 #define F2FS_OTHER_FLMASK (F2FS_NODUMP_FL | F2FS_NOATIME_FL) 2621 2622 static inline __u32 f2fs_mask_flags(umode_t mode, __u32 flags) 2623 { 2624 if (S_ISDIR(mode)) 2625 return flags; 2626 else if (S_ISREG(mode)) 2627 return flags & F2FS_REG_FLMASK; 2628 else 2629 return flags & F2FS_OTHER_FLMASK; 2630 } 2631 2632 static inline void __mark_inode_dirty_flag(struct inode *inode, 2633 int flag, bool set) 2634 { 2635 switch (flag) { 2636 case FI_INLINE_XATTR: 2637 case FI_INLINE_DATA: 2638 case FI_INLINE_DENTRY: 2639 case FI_NEW_INODE: 2640 if (set) 2641 return; 2642 fallthrough; 2643 case FI_DATA_EXIST: 2644 case FI_INLINE_DOTS: 2645 case FI_PIN_FILE: 2646 f2fs_mark_inode_dirty_sync(inode, true); 2647 } 2648 } 2649 2650 static inline void set_inode_flag(struct inode *inode, int flag) 2651 { 2652 set_bit(flag, F2FS_I(inode)->flags); 2653 __mark_inode_dirty_flag(inode, flag, true); 2654 } 2655 2656 static inline int is_inode_flag_set(struct inode *inode, int flag) 2657 { 2658 return test_bit(flag, F2FS_I(inode)->flags); 2659 } 2660 2661 static inline void clear_inode_flag(struct inode *inode, int flag) 2662 { 2663 clear_bit(flag, F2FS_I(inode)->flags); 2664 __mark_inode_dirty_flag(inode, flag, false); 2665 } 2666 2667 static inline bool f2fs_verity_in_progress(struct inode *inode) 2668 { 2669 return IS_ENABLED(CONFIG_FS_VERITY) && 2670 is_inode_flag_set(inode, FI_VERITY_IN_PROGRESS); 2671 } 2672 2673 static inline void set_acl_inode(struct inode *inode, umode_t mode) 2674 { 2675 F2FS_I(inode)->i_acl_mode = mode; 2676 set_inode_flag(inode, FI_ACL_MODE); 2677 f2fs_mark_inode_dirty_sync(inode, false); 2678 } 2679 2680 static inline void f2fs_i_links_write(struct inode *inode, bool inc) 2681 { 2682 if (inc) 2683 inc_nlink(inode); 2684 else 2685 drop_nlink(inode); 2686 f2fs_mark_inode_dirty_sync(inode, true); 2687 } 2688 2689 static inline void f2fs_i_blocks_write(struct inode *inode, 2690 block_t diff, bool add, bool claim) 2691 { 2692 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE); 2693 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER); 2694 2695 /* add = 1, claim = 1 should be dquot_reserve_block in pair */ 2696 if (add) { 2697 if (claim) 2698 dquot_claim_block(inode, diff); 2699 else 2700 dquot_alloc_block_nofail(inode, diff); 2701 } else { 2702 dquot_free_block(inode, diff); 2703 } 2704 2705 f2fs_mark_inode_dirty_sync(inode, true); 2706 if (clean || recover) 2707 set_inode_flag(inode, FI_AUTO_RECOVER); 2708 } 2709 2710 static inline void f2fs_i_size_write(struct inode *inode, loff_t i_size) 2711 { 2712 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE); 2713 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER); 2714 2715 if (i_size_read(inode) == i_size) 2716 return; 2717 2718 i_size_write(inode, i_size); 2719 f2fs_mark_inode_dirty_sync(inode, true); 2720 if (clean || recover) 2721 set_inode_flag(inode, FI_AUTO_RECOVER); 2722 } 2723 2724 static inline void f2fs_i_depth_write(struct inode *inode, unsigned int depth) 2725 { 2726 F2FS_I(inode)->i_current_depth = depth; 2727 f2fs_mark_inode_dirty_sync(inode, true); 2728 } 2729 2730 static inline void f2fs_i_gc_failures_write(struct inode *inode, 2731 unsigned int count) 2732 { 2733 F2FS_I(inode)->i_gc_failures[GC_FAILURE_PIN] = count; 2734 f2fs_mark_inode_dirty_sync(inode, true); 2735 } 2736 2737 static inline void f2fs_i_xnid_write(struct inode *inode, nid_t xnid) 2738 { 2739 F2FS_I(inode)->i_xattr_nid = xnid; 2740 f2fs_mark_inode_dirty_sync(inode, true); 2741 } 2742 2743 static inline void f2fs_i_pino_write(struct inode *inode, nid_t pino) 2744 { 2745 F2FS_I(inode)->i_pino = pino; 2746 f2fs_mark_inode_dirty_sync(inode, true); 2747 } 2748 2749 static inline void get_inline_info(struct inode *inode, struct f2fs_inode *ri) 2750 { 2751 struct f2fs_inode_info *fi = F2FS_I(inode); 2752 2753 if (ri->i_inline & F2FS_INLINE_XATTR) 2754 set_bit(FI_INLINE_XATTR, fi->flags); 2755 if (ri->i_inline & F2FS_INLINE_DATA) 2756 set_bit(FI_INLINE_DATA, fi->flags); 2757 if (ri->i_inline & F2FS_INLINE_DENTRY) 2758 set_bit(FI_INLINE_DENTRY, fi->flags); 2759 if (ri->i_inline & F2FS_DATA_EXIST) 2760 set_bit(FI_DATA_EXIST, fi->flags); 2761 if (ri->i_inline & F2FS_INLINE_DOTS) 2762 set_bit(FI_INLINE_DOTS, fi->flags); 2763 if (ri->i_inline & F2FS_EXTRA_ATTR) 2764 set_bit(FI_EXTRA_ATTR, fi->flags); 2765 if (ri->i_inline & F2FS_PIN_FILE) 2766 set_bit(FI_PIN_FILE, fi->flags); 2767 } 2768 2769 static inline void set_raw_inline(struct inode *inode, struct f2fs_inode *ri) 2770 { 2771 ri->i_inline = 0; 2772 2773 if (is_inode_flag_set(inode, FI_INLINE_XATTR)) 2774 ri->i_inline |= F2FS_INLINE_XATTR; 2775 if (is_inode_flag_set(inode, FI_INLINE_DATA)) 2776 ri->i_inline |= F2FS_INLINE_DATA; 2777 if (is_inode_flag_set(inode, FI_INLINE_DENTRY)) 2778 ri->i_inline |= F2FS_INLINE_DENTRY; 2779 if (is_inode_flag_set(inode, FI_DATA_EXIST)) 2780 ri->i_inline |= F2FS_DATA_EXIST; 2781 if (is_inode_flag_set(inode, FI_INLINE_DOTS)) 2782 ri->i_inline |= F2FS_INLINE_DOTS; 2783 if (is_inode_flag_set(inode, FI_EXTRA_ATTR)) 2784 ri->i_inline |= F2FS_EXTRA_ATTR; 2785 if (is_inode_flag_set(inode, FI_PIN_FILE)) 2786 ri->i_inline |= F2FS_PIN_FILE; 2787 } 2788 2789 static inline int f2fs_has_extra_attr(struct inode *inode) 2790 { 2791 return is_inode_flag_set(inode, FI_EXTRA_ATTR); 2792 } 2793 2794 static inline int f2fs_has_inline_xattr(struct inode *inode) 2795 { 2796 return is_inode_flag_set(inode, FI_INLINE_XATTR); 2797 } 2798 2799 static inline int f2fs_compressed_file(struct inode *inode) 2800 { 2801 return S_ISREG(inode->i_mode) && 2802 is_inode_flag_set(inode, FI_COMPRESSED_FILE); 2803 } 2804 2805 static inline unsigned int addrs_per_inode(struct inode *inode) 2806 { 2807 unsigned int addrs = CUR_ADDRS_PER_INODE(inode) - 2808 get_inline_xattr_addrs(inode); 2809 2810 if (!f2fs_compressed_file(inode)) 2811 return addrs; 2812 return ALIGN_DOWN(addrs, F2FS_I(inode)->i_cluster_size); 2813 } 2814 2815 static inline unsigned int addrs_per_block(struct inode *inode) 2816 { 2817 if (!f2fs_compressed_file(inode)) 2818 return DEF_ADDRS_PER_BLOCK; 2819 return ALIGN_DOWN(DEF_ADDRS_PER_BLOCK, F2FS_I(inode)->i_cluster_size); 2820 } 2821 2822 static inline void *inline_xattr_addr(struct inode *inode, struct page *page) 2823 { 2824 struct f2fs_inode *ri = F2FS_INODE(page); 2825 2826 return (void *)&(ri->i_addr[DEF_ADDRS_PER_INODE - 2827 get_inline_xattr_addrs(inode)]); 2828 } 2829 2830 static inline int inline_xattr_size(struct inode *inode) 2831 { 2832 if (f2fs_has_inline_xattr(inode)) 2833 return get_inline_xattr_addrs(inode) * sizeof(__le32); 2834 return 0; 2835 } 2836 2837 static inline int f2fs_has_inline_data(struct inode *inode) 2838 { 2839 return is_inode_flag_set(inode, FI_INLINE_DATA); 2840 } 2841 2842 static inline int f2fs_exist_data(struct inode *inode) 2843 { 2844 return is_inode_flag_set(inode, FI_DATA_EXIST); 2845 } 2846 2847 static inline int f2fs_has_inline_dots(struct inode *inode) 2848 { 2849 return is_inode_flag_set(inode, FI_INLINE_DOTS); 2850 } 2851 2852 static inline int f2fs_is_mmap_file(struct inode *inode) 2853 { 2854 return is_inode_flag_set(inode, FI_MMAP_FILE); 2855 } 2856 2857 static inline bool f2fs_is_pinned_file(struct inode *inode) 2858 { 2859 return is_inode_flag_set(inode, FI_PIN_FILE); 2860 } 2861 2862 static inline bool f2fs_is_atomic_file(struct inode *inode) 2863 { 2864 return is_inode_flag_set(inode, FI_ATOMIC_FILE); 2865 } 2866 2867 static inline bool f2fs_is_commit_atomic_write(struct inode *inode) 2868 { 2869 return is_inode_flag_set(inode, FI_ATOMIC_COMMIT); 2870 } 2871 2872 static inline bool f2fs_is_volatile_file(struct inode *inode) 2873 { 2874 return is_inode_flag_set(inode, FI_VOLATILE_FILE); 2875 } 2876 2877 static inline bool f2fs_is_first_block_written(struct inode *inode) 2878 { 2879 return is_inode_flag_set(inode, FI_FIRST_BLOCK_WRITTEN); 2880 } 2881 2882 static inline bool f2fs_is_drop_cache(struct inode *inode) 2883 { 2884 return is_inode_flag_set(inode, FI_DROP_CACHE); 2885 } 2886 2887 static inline void *inline_data_addr(struct inode *inode, struct page *page) 2888 { 2889 struct f2fs_inode *ri = F2FS_INODE(page); 2890 int extra_size = get_extra_isize(inode); 2891 2892 return (void *)&(ri->i_addr[extra_size + DEF_INLINE_RESERVED_SIZE]); 2893 } 2894 2895 static inline int f2fs_has_inline_dentry(struct inode *inode) 2896 { 2897 return is_inode_flag_set(inode, FI_INLINE_DENTRY); 2898 } 2899 2900 static inline int is_file(struct inode *inode, int type) 2901 { 2902 return F2FS_I(inode)->i_advise & type; 2903 } 2904 2905 static inline void set_file(struct inode *inode, int type) 2906 { 2907 F2FS_I(inode)->i_advise |= type; 2908 f2fs_mark_inode_dirty_sync(inode, true); 2909 } 2910 2911 static inline void clear_file(struct inode *inode, int type) 2912 { 2913 F2FS_I(inode)->i_advise &= ~type; 2914 f2fs_mark_inode_dirty_sync(inode, true); 2915 } 2916 2917 static inline bool f2fs_is_time_consistent(struct inode *inode) 2918 { 2919 if (!timespec64_equal(F2FS_I(inode)->i_disk_time, &inode->i_atime)) 2920 return false; 2921 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 1, &inode->i_ctime)) 2922 return false; 2923 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 2, &inode->i_mtime)) 2924 return false; 2925 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 3, 2926 &F2FS_I(inode)->i_crtime)) 2927 return false; 2928 return true; 2929 } 2930 2931 static inline bool f2fs_skip_inode_update(struct inode *inode, int dsync) 2932 { 2933 bool ret; 2934 2935 if (dsync) { 2936 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 2937 2938 spin_lock(&sbi->inode_lock[DIRTY_META]); 2939 ret = list_empty(&F2FS_I(inode)->gdirty_list); 2940 spin_unlock(&sbi->inode_lock[DIRTY_META]); 2941 return ret; 2942 } 2943 if (!is_inode_flag_set(inode, FI_AUTO_RECOVER) || 2944 file_keep_isize(inode) || 2945 i_size_read(inode) & ~PAGE_MASK) 2946 return false; 2947 2948 if (!f2fs_is_time_consistent(inode)) 2949 return false; 2950 2951 spin_lock(&F2FS_I(inode)->i_size_lock); 2952 ret = F2FS_I(inode)->last_disk_size == i_size_read(inode); 2953 spin_unlock(&F2FS_I(inode)->i_size_lock); 2954 2955 return ret; 2956 } 2957 2958 static inline bool f2fs_readonly(struct super_block *sb) 2959 { 2960 return sb_rdonly(sb); 2961 } 2962 2963 static inline bool f2fs_cp_error(struct f2fs_sb_info *sbi) 2964 { 2965 return is_set_ckpt_flags(sbi, CP_ERROR_FLAG); 2966 } 2967 2968 static inline bool is_dot_dotdot(const u8 *name, size_t len) 2969 { 2970 if (len == 1 && name[0] == '.') 2971 return true; 2972 2973 if (len == 2 && name[0] == '.' && name[1] == '.') 2974 return true; 2975 2976 return false; 2977 } 2978 2979 static inline bool f2fs_may_extent_tree(struct inode *inode) 2980 { 2981 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 2982 2983 if (!test_opt(sbi, EXTENT_CACHE) || 2984 is_inode_flag_set(inode, FI_NO_EXTENT) || 2985 is_inode_flag_set(inode, FI_COMPRESSED_FILE)) 2986 return false; 2987 2988 /* 2989 * for recovered files during mount do not create extents 2990 * if shrinker is not registered. 2991 */ 2992 if (list_empty(&sbi->s_list)) 2993 return false; 2994 2995 return S_ISREG(inode->i_mode); 2996 } 2997 2998 static inline void *f2fs_kmalloc(struct f2fs_sb_info *sbi, 2999 size_t size, gfp_t flags) 3000 { 3001 if (time_to_inject(sbi, FAULT_KMALLOC)) { 3002 f2fs_show_injection_info(sbi, FAULT_KMALLOC); 3003 return NULL; 3004 } 3005 3006 return kmalloc(size, flags); 3007 } 3008 3009 static inline void *f2fs_kzalloc(struct f2fs_sb_info *sbi, 3010 size_t size, gfp_t flags) 3011 { 3012 return f2fs_kmalloc(sbi, size, flags | __GFP_ZERO); 3013 } 3014 3015 static inline void *f2fs_kvmalloc(struct f2fs_sb_info *sbi, 3016 size_t size, gfp_t flags) 3017 { 3018 if (time_to_inject(sbi, FAULT_KVMALLOC)) { 3019 f2fs_show_injection_info(sbi, FAULT_KVMALLOC); 3020 return NULL; 3021 } 3022 3023 return kvmalloc(size, flags); 3024 } 3025 3026 static inline void *f2fs_kvzalloc(struct f2fs_sb_info *sbi, 3027 size_t size, gfp_t flags) 3028 { 3029 return f2fs_kvmalloc(sbi, size, flags | __GFP_ZERO); 3030 } 3031 3032 static inline int get_extra_isize(struct inode *inode) 3033 { 3034 return F2FS_I(inode)->i_extra_isize / sizeof(__le32); 3035 } 3036 3037 static inline int get_inline_xattr_addrs(struct inode *inode) 3038 { 3039 return F2FS_I(inode)->i_inline_xattr_size; 3040 } 3041 3042 #define f2fs_get_inode_mode(i) \ 3043 ((is_inode_flag_set(i, FI_ACL_MODE)) ? \ 3044 (F2FS_I(i)->i_acl_mode) : ((i)->i_mode)) 3045 3046 #define F2FS_TOTAL_EXTRA_ATTR_SIZE \ 3047 (offsetof(struct f2fs_inode, i_extra_end) - \ 3048 offsetof(struct f2fs_inode, i_extra_isize)) \ 3049 3050 #define F2FS_OLD_ATTRIBUTE_SIZE (offsetof(struct f2fs_inode, i_addr)) 3051 #define F2FS_FITS_IN_INODE(f2fs_inode, extra_isize, field) \ 3052 ((offsetof(typeof(*(f2fs_inode)), field) + \ 3053 sizeof((f2fs_inode)->field)) \ 3054 <= (F2FS_OLD_ATTRIBUTE_SIZE + (extra_isize))) \ 3055 3056 #define DEFAULT_IOSTAT_PERIOD_MS 3000 3057 #define MIN_IOSTAT_PERIOD_MS 100 3058 /* maximum period of iostat tracing is 1 day */ 3059 #define MAX_IOSTAT_PERIOD_MS 8640000 3060 3061 static inline void f2fs_reset_iostat(struct f2fs_sb_info *sbi) 3062 { 3063 int i; 3064 3065 spin_lock(&sbi->iostat_lock); 3066 for (i = 0; i < NR_IO_TYPE; i++) { 3067 sbi->rw_iostat[i] = 0; 3068 sbi->prev_rw_iostat[i] = 0; 3069 } 3070 spin_unlock(&sbi->iostat_lock); 3071 } 3072 3073 extern void f2fs_record_iostat(struct f2fs_sb_info *sbi); 3074 3075 static inline void f2fs_update_iostat(struct f2fs_sb_info *sbi, 3076 enum iostat_type type, unsigned long long io_bytes) 3077 { 3078 if (!sbi->iostat_enable) 3079 return; 3080 spin_lock(&sbi->iostat_lock); 3081 sbi->rw_iostat[type] += io_bytes; 3082 3083 if (type == APP_WRITE_IO || type == APP_DIRECT_IO) 3084 sbi->rw_iostat[APP_BUFFERED_IO] = 3085 sbi->rw_iostat[APP_WRITE_IO] - 3086 sbi->rw_iostat[APP_DIRECT_IO]; 3087 3088 if (type == APP_READ_IO || type == APP_DIRECT_READ_IO) 3089 sbi->rw_iostat[APP_BUFFERED_READ_IO] = 3090 sbi->rw_iostat[APP_READ_IO] - 3091 sbi->rw_iostat[APP_DIRECT_READ_IO]; 3092 spin_unlock(&sbi->iostat_lock); 3093 3094 f2fs_record_iostat(sbi); 3095 } 3096 3097 #define __is_large_section(sbi) ((sbi)->segs_per_sec > 1) 3098 3099 #define __is_meta_io(fio) (PAGE_TYPE_OF_BIO((fio)->type) == META) 3100 3101 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi, 3102 block_t blkaddr, int type); 3103 static inline void verify_blkaddr(struct f2fs_sb_info *sbi, 3104 block_t blkaddr, int type) 3105 { 3106 if (!f2fs_is_valid_blkaddr(sbi, blkaddr, type)) { 3107 f2fs_err(sbi, "invalid blkaddr: %u, type: %d, run fsck to fix.", 3108 blkaddr, type); 3109 f2fs_bug_on(sbi, 1); 3110 } 3111 } 3112 3113 static inline bool __is_valid_data_blkaddr(block_t blkaddr) 3114 { 3115 if (blkaddr == NEW_ADDR || blkaddr == NULL_ADDR || 3116 blkaddr == COMPRESS_ADDR) 3117 return false; 3118 return true; 3119 } 3120 3121 static inline void f2fs_set_page_private(struct page *page, 3122 unsigned long data) 3123 { 3124 if (PagePrivate(page)) 3125 return; 3126 3127 attach_page_private(page, (void *)data); 3128 } 3129 3130 static inline void f2fs_clear_page_private(struct page *page) 3131 { 3132 detach_page_private(page); 3133 } 3134 3135 /* 3136 * file.c 3137 */ 3138 int f2fs_sync_file(struct file *file, loff_t start, loff_t end, int datasync); 3139 void f2fs_truncate_data_blocks(struct dnode_of_data *dn); 3140 int f2fs_do_truncate_blocks(struct inode *inode, u64 from, bool lock); 3141 int f2fs_truncate_blocks(struct inode *inode, u64 from, bool lock); 3142 int f2fs_truncate(struct inode *inode); 3143 int f2fs_getattr(const struct path *path, struct kstat *stat, 3144 u32 request_mask, unsigned int flags); 3145 int f2fs_setattr(struct dentry *dentry, struct iattr *attr); 3146 int f2fs_truncate_hole(struct inode *inode, pgoff_t pg_start, pgoff_t pg_end); 3147 void f2fs_truncate_data_blocks_range(struct dnode_of_data *dn, int count); 3148 int f2fs_precache_extents(struct inode *inode); 3149 long f2fs_ioctl(struct file *filp, unsigned int cmd, unsigned long arg); 3150 long f2fs_compat_ioctl(struct file *file, unsigned int cmd, unsigned long arg); 3151 int f2fs_transfer_project_quota(struct inode *inode, kprojid_t kprojid); 3152 int f2fs_pin_file_control(struct inode *inode, bool inc); 3153 3154 /* 3155 * inode.c 3156 */ 3157 void f2fs_set_inode_flags(struct inode *inode); 3158 bool f2fs_inode_chksum_verify(struct f2fs_sb_info *sbi, struct page *page); 3159 void f2fs_inode_chksum_set(struct f2fs_sb_info *sbi, struct page *page); 3160 struct inode *f2fs_iget(struct super_block *sb, unsigned long ino); 3161 struct inode *f2fs_iget_retry(struct super_block *sb, unsigned long ino); 3162 int f2fs_try_to_free_nats(struct f2fs_sb_info *sbi, int nr_shrink); 3163 void f2fs_update_inode(struct inode *inode, struct page *node_page); 3164 void f2fs_update_inode_page(struct inode *inode); 3165 int f2fs_write_inode(struct inode *inode, struct writeback_control *wbc); 3166 void f2fs_evict_inode(struct inode *inode); 3167 void f2fs_handle_failed_inode(struct inode *inode); 3168 3169 /* 3170 * namei.c 3171 */ 3172 int f2fs_update_extension_list(struct f2fs_sb_info *sbi, const char *name, 3173 bool hot, bool set); 3174 struct dentry *f2fs_get_parent(struct dentry *child); 3175 3176 /* 3177 * dir.c 3178 */ 3179 unsigned char f2fs_get_de_type(struct f2fs_dir_entry *de); 3180 int f2fs_init_casefolded_name(const struct inode *dir, 3181 struct f2fs_filename *fname); 3182 int f2fs_setup_filename(struct inode *dir, const struct qstr *iname, 3183 int lookup, struct f2fs_filename *fname); 3184 int f2fs_prepare_lookup(struct inode *dir, struct dentry *dentry, 3185 struct f2fs_filename *fname); 3186 void f2fs_free_filename(struct f2fs_filename *fname); 3187 struct f2fs_dir_entry *f2fs_find_target_dentry(const struct f2fs_dentry_ptr *d, 3188 const struct f2fs_filename *fname, int *max_slots); 3189 int f2fs_fill_dentries(struct dir_context *ctx, struct f2fs_dentry_ptr *d, 3190 unsigned int start_pos, struct fscrypt_str *fstr); 3191 void f2fs_do_make_empty_dir(struct inode *inode, struct inode *parent, 3192 struct f2fs_dentry_ptr *d); 3193 struct page *f2fs_init_inode_metadata(struct inode *inode, struct inode *dir, 3194 const struct f2fs_filename *fname, struct page *dpage); 3195 void f2fs_update_parent_metadata(struct inode *dir, struct inode *inode, 3196 unsigned int current_depth); 3197 int f2fs_room_for_filename(const void *bitmap, int slots, int max_slots); 3198 void f2fs_drop_nlink(struct inode *dir, struct inode *inode); 3199 struct f2fs_dir_entry *__f2fs_find_entry(struct inode *dir, 3200 const struct f2fs_filename *fname, 3201 struct page **res_page); 3202 struct f2fs_dir_entry *f2fs_find_entry(struct inode *dir, 3203 const struct qstr *child, struct page **res_page); 3204 struct f2fs_dir_entry *f2fs_parent_dir(struct inode *dir, struct page **p); 3205 ino_t f2fs_inode_by_name(struct inode *dir, const struct qstr *qstr, 3206 struct page **page); 3207 void f2fs_set_link(struct inode *dir, struct f2fs_dir_entry *de, 3208 struct page *page, struct inode *inode); 3209 bool f2fs_has_enough_room(struct inode *dir, struct page *ipage, 3210 const struct f2fs_filename *fname); 3211 void f2fs_update_dentry(nid_t ino, umode_t mode, struct f2fs_dentry_ptr *d, 3212 const struct fscrypt_str *name, f2fs_hash_t name_hash, 3213 unsigned int bit_pos); 3214 int f2fs_add_regular_entry(struct inode *dir, const struct f2fs_filename *fname, 3215 struct inode *inode, nid_t ino, umode_t mode); 3216 int f2fs_add_dentry(struct inode *dir, const struct f2fs_filename *fname, 3217 struct inode *inode, nid_t ino, umode_t mode); 3218 int f2fs_do_add_link(struct inode *dir, const struct qstr *name, 3219 struct inode *inode, nid_t ino, umode_t mode); 3220 void f2fs_delete_entry(struct f2fs_dir_entry *dentry, struct page *page, 3221 struct inode *dir, struct inode *inode); 3222 int f2fs_do_tmpfile(struct inode *inode, struct inode *dir); 3223 bool f2fs_empty_dir(struct inode *dir); 3224 3225 static inline int f2fs_add_link(struct dentry *dentry, struct inode *inode) 3226 { 3227 return f2fs_do_add_link(d_inode(dentry->d_parent), &dentry->d_name, 3228 inode, inode->i_ino, inode->i_mode); 3229 } 3230 3231 /* 3232 * super.c 3233 */ 3234 int f2fs_inode_dirtied(struct inode *inode, bool sync); 3235 void f2fs_inode_synced(struct inode *inode); 3236 int f2fs_enable_quota_files(struct f2fs_sb_info *sbi, bool rdonly); 3237 int f2fs_quota_sync(struct super_block *sb, int type); 3238 void f2fs_quota_off_umount(struct super_block *sb); 3239 int f2fs_commit_super(struct f2fs_sb_info *sbi, bool recover); 3240 int f2fs_sync_fs(struct super_block *sb, int sync); 3241 int f2fs_sanity_check_ckpt(struct f2fs_sb_info *sbi); 3242 3243 /* 3244 * hash.c 3245 */ 3246 void f2fs_hash_filename(const struct inode *dir, struct f2fs_filename *fname); 3247 3248 /* 3249 * node.c 3250 */ 3251 struct dnode_of_data; 3252 struct node_info; 3253 3254 int f2fs_check_nid_range(struct f2fs_sb_info *sbi, nid_t nid); 3255 bool f2fs_available_free_memory(struct f2fs_sb_info *sbi, int type); 3256 bool f2fs_in_warm_node_list(struct f2fs_sb_info *sbi, struct page *page); 3257 void f2fs_init_fsync_node_info(struct f2fs_sb_info *sbi); 3258 void f2fs_del_fsync_node_entry(struct f2fs_sb_info *sbi, struct page *page); 3259 void f2fs_reset_fsync_node_info(struct f2fs_sb_info *sbi); 3260 int f2fs_need_dentry_mark(struct f2fs_sb_info *sbi, nid_t nid); 3261 bool f2fs_is_checkpointed_node(struct f2fs_sb_info *sbi, nid_t nid); 3262 bool f2fs_need_inode_block_update(struct f2fs_sb_info *sbi, nid_t ino); 3263 int f2fs_get_node_info(struct f2fs_sb_info *sbi, nid_t nid, 3264 struct node_info *ni); 3265 pgoff_t f2fs_get_next_page_offset(struct dnode_of_data *dn, pgoff_t pgofs); 3266 int f2fs_get_dnode_of_data(struct dnode_of_data *dn, pgoff_t index, int mode); 3267 int f2fs_truncate_inode_blocks(struct inode *inode, pgoff_t from); 3268 int f2fs_truncate_xattr_node(struct inode *inode); 3269 int f2fs_wait_on_node_pages_writeback(struct f2fs_sb_info *sbi, 3270 unsigned int seq_id); 3271 int f2fs_remove_inode_page(struct inode *inode); 3272 struct page *f2fs_new_inode_page(struct inode *inode); 3273 struct page *f2fs_new_node_page(struct dnode_of_data *dn, unsigned int ofs); 3274 void f2fs_ra_node_page(struct f2fs_sb_info *sbi, nid_t nid); 3275 struct page *f2fs_get_node_page(struct f2fs_sb_info *sbi, pgoff_t nid); 3276 struct page *f2fs_get_node_page_ra(struct page *parent, int start); 3277 int f2fs_move_node_page(struct page *node_page, int gc_type); 3278 void f2fs_flush_inline_data(struct f2fs_sb_info *sbi); 3279 int f2fs_fsync_node_pages(struct f2fs_sb_info *sbi, struct inode *inode, 3280 struct writeback_control *wbc, bool atomic, 3281 unsigned int *seq_id); 3282 int f2fs_sync_node_pages(struct f2fs_sb_info *sbi, 3283 struct writeback_control *wbc, 3284 bool do_balance, enum iostat_type io_type); 3285 int f2fs_build_free_nids(struct f2fs_sb_info *sbi, bool sync, bool mount); 3286 bool f2fs_alloc_nid(struct f2fs_sb_info *sbi, nid_t *nid); 3287 void f2fs_alloc_nid_done(struct f2fs_sb_info *sbi, nid_t nid); 3288 void f2fs_alloc_nid_failed(struct f2fs_sb_info *sbi, nid_t nid); 3289 int f2fs_try_to_free_nids(struct f2fs_sb_info *sbi, int nr_shrink); 3290 int f2fs_recover_inline_xattr(struct inode *inode, struct page *page); 3291 int f2fs_recover_xattr_data(struct inode *inode, struct page *page); 3292 int f2fs_recover_inode_page(struct f2fs_sb_info *sbi, struct page *page); 3293 int f2fs_restore_node_summary(struct f2fs_sb_info *sbi, 3294 unsigned int segno, struct f2fs_summary_block *sum); 3295 int f2fs_flush_nat_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3296 int f2fs_build_node_manager(struct f2fs_sb_info *sbi); 3297 void f2fs_destroy_node_manager(struct f2fs_sb_info *sbi); 3298 int __init f2fs_create_node_manager_caches(void); 3299 void f2fs_destroy_node_manager_caches(void); 3300 3301 /* 3302 * segment.c 3303 */ 3304 bool f2fs_need_SSR(struct f2fs_sb_info *sbi); 3305 void f2fs_register_inmem_page(struct inode *inode, struct page *page); 3306 void f2fs_drop_inmem_pages_all(struct f2fs_sb_info *sbi, bool gc_failure); 3307 void f2fs_drop_inmem_pages(struct inode *inode); 3308 void f2fs_drop_inmem_page(struct inode *inode, struct page *page); 3309 int f2fs_commit_inmem_pages(struct inode *inode); 3310 void f2fs_balance_fs(struct f2fs_sb_info *sbi, bool need); 3311 void f2fs_balance_fs_bg(struct f2fs_sb_info *sbi, bool from_bg); 3312 int f2fs_issue_flush(struct f2fs_sb_info *sbi, nid_t ino); 3313 int f2fs_create_flush_cmd_control(struct f2fs_sb_info *sbi); 3314 int f2fs_flush_device_cache(struct f2fs_sb_info *sbi); 3315 void f2fs_destroy_flush_cmd_control(struct f2fs_sb_info *sbi, bool free); 3316 void f2fs_invalidate_blocks(struct f2fs_sb_info *sbi, block_t addr); 3317 bool f2fs_is_checkpointed_data(struct f2fs_sb_info *sbi, block_t blkaddr); 3318 void f2fs_drop_discard_cmd(struct f2fs_sb_info *sbi); 3319 void f2fs_stop_discard_thread(struct f2fs_sb_info *sbi); 3320 bool f2fs_issue_discard_timeout(struct f2fs_sb_info *sbi); 3321 void f2fs_clear_prefree_segments(struct f2fs_sb_info *sbi, 3322 struct cp_control *cpc); 3323 void f2fs_dirty_to_prefree(struct f2fs_sb_info *sbi); 3324 block_t f2fs_get_unusable_blocks(struct f2fs_sb_info *sbi); 3325 int f2fs_disable_cp_again(struct f2fs_sb_info *sbi, block_t unusable); 3326 void f2fs_release_discard_addrs(struct f2fs_sb_info *sbi); 3327 int f2fs_npages_for_summary_flush(struct f2fs_sb_info *sbi, bool for_ra); 3328 void f2fs_allocate_segment_for_resize(struct f2fs_sb_info *sbi, int type, 3329 unsigned int start, unsigned int end); 3330 void f2fs_allocate_new_segment(struct f2fs_sb_info *sbi, int type); 3331 void f2fs_allocate_new_segments(struct f2fs_sb_info *sbi); 3332 int f2fs_trim_fs(struct f2fs_sb_info *sbi, struct fstrim_range *range); 3333 bool f2fs_exist_trim_candidates(struct f2fs_sb_info *sbi, 3334 struct cp_control *cpc); 3335 struct page *f2fs_get_sum_page(struct f2fs_sb_info *sbi, unsigned int segno); 3336 void f2fs_update_meta_page(struct f2fs_sb_info *sbi, void *src, 3337 block_t blk_addr); 3338 void f2fs_do_write_meta_page(struct f2fs_sb_info *sbi, struct page *page, 3339 enum iostat_type io_type); 3340 void f2fs_do_write_node_page(unsigned int nid, struct f2fs_io_info *fio); 3341 void f2fs_outplace_write_data(struct dnode_of_data *dn, 3342 struct f2fs_io_info *fio); 3343 int f2fs_inplace_write_data(struct f2fs_io_info *fio); 3344 void f2fs_do_replace_block(struct f2fs_sb_info *sbi, struct f2fs_summary *sum, 3345 block_t old_blkaddr, block_t new_blkaddr, 3346 bool recover_curseg, bool recover_newaddr); 3347 void f2fs_replace_block(struct f2fs_sb_info *sbi, struct dnode_of_data *dn, 3348 block_t old_addr, block_t new_addr, 3349 unsigned char version, bool recover_curseg, 3350 bool recover_newaddr); 3351 void f2fs_allocate_data_block(struct f2fs_sb_info *sbi, struct page *page, 3352 block_t old_blkaddr, block_t *new_blkaddr, 3353 struct f2fs_summary *sum, int type, 3354 struct f2fs_io_info *fio); 3355 void f2fs_wait_on_page_writeback(struct page *page, 3356 enum page_type type, bool ordered, bool locked); 3357 void f2fs_wait_on_block_writeback(struct inode *inode, block_t blkaddr); 3358 void f2fs_wait_on_block_writeback_range(struct inode *inode, block_t blkaddr, 3359 block_t len); 3360 void f2fs_write_data_summaries(struct f2fs_sb_info *sbi, block_t start_blk); 3361 void f2fs_write_node_summaries(struct f2fs_sb_info *sbi, block_t start_blk); 3362 int f2fs_lookup_journal_in_cursum(struct f2fs_journal *journal, int type, 3363 unsigned int val, int alloc); 3364 void f2fs_flush_sit_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3365 int f2fs_fix_curseg_write_pointer(struct f2fs_sb_info *sbi); 3366 int f2fs_check_write_pointer(struct f2fs_sb_info *sbi); 3367 int f2fs_build_segment_manager(struct f2fs_sb_info *sbi); 3368 void f2fs_destroy_segment_manager(struct f2fs_sb_info *sbi); 3369 int __init f2fs_create_segment_manager_caches(void); 3370 void f2fs_destroy_segment_manager_caches(void); 3371 int f2fs_rw_hint_to_seg_type(enum rw_hint hint); 3372 enum rw_hint f2fs_io_type_to_rw_hint(struct f2fs_sb_info *sbi, 3373 enum page_type type, enum temp_type temp); 3374 3375 /* 3376 * checkpoint.c 3377 */ 3378 void f2fs_stop_checkpoint(struct f2fs_sb_info *sbi, bool end_io); 3379 struct page *f2fs_grab_meta_page(struct f2fs_sb_info *sbi, pgoff_t index); 3380 struct page *f2fs_get_meta_page(struct f2fs_sb_info *sbi, pgoff_t index); 3381 struct page *f2fs_get_meta_page_nofail(struct f2fs_sb_info *sbi, pgoff_t index); 3382 struct page *f2fs_get_tmp_page(struct f2fs_sb_info *sbi, pgoff_t index); 3383 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi, 3384 block_t blkaddr, int type); 3385 int f2fs_ra_meta_pages(struct f2fs_sb_info *sbi, block_t start, int nrpages, 3386 int type, bool sync); 3387 void f2fs_ra_meta_pages_cond(struct f2fs_sb_info *sbi, pgoff_t index); 3388 long f2fs_sync_meta_pages(struct f2fs_sb_info *sbi, enum page_type type, 3389 long nr_to_write, enum iostat_type io_type); 3390 void f2fs_add_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type); 3391 void f2fs_remove_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type); 3392 void f2fs_release_ino_entry(struct f2fs_sb_info *sbi, bool all); 3393 bool f2fs_exist_written_data(struct f2fs_sb_info *sbi, nid_t ino, int mode); 3394 void f2fs_set_dirty_device(struct f2fs_sb_info *sbi, nid_t ino, 3395 unsigned int devidx, int type); 3396 bool f2fs_is_dirty_device(struct f2fs_sb_info *sbi, nid_t ino, 3397 unsigned int devidx, int type); 3398 int f2fs_sync_inode_meta(struct f2fs_sb_info *sbi); 3399 int f2fs_acquire_orphan_inode(struct f2fs_sb_info *sbi); 3400 void f2fs_release_orphan_inode(struct f2fs_sb_info *sbi); 3401 void f2fs_add_orphan_inode(struct inode *inode); 3402 void f2fs_remove_orphan_inode(struct f2fs_sb_info *sbi, nid_t ino); 3403 int f2fs_recover_orphan_inodes(struct f2fs_sb_info *sbi); 3404 int f2fs_get_valid_checkpoint(struct f2fs_sb_info *sbi); 3405 void f2fs_update_dirty_page(struct inode *inode, struct page *page); 3406 void f2fs_remove_dirty_inode(struct inode *inode); 3407 int f2fs_sync_dirty_inodes(struct f2fs_sb_info *sbi, enum inode_type type); 3408 void f2fs_wait_on_all_pages(struct f2fs_sb_info *sbi, int type); 3409 int f2fs_write_checkpoint(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3410 void f2fs_init_ino_entry_info(struct f2fs_sb_info *sbi); 3411 int __init f2fs_create_checkpoint_caches(void); 3412 void f2fs_destroy_checkpoint_caches(void); 3413 3414 /* 3415 * data.c 3416 */ 3417 int __init f2fs_init_bioset(void); 3418 void f2fs_destroy_bioset(void); 3419 struct bio *f2fs_bio_alloc(struct f2fs_sb_info *sbi, int npages, bool noio); 3420 int f2fs_init_bio_entry_cache(void); 3421 void f2fs_destroy_bio_entry_cache(void); 3422 void f2fs_submit_bio(struct f2fs_sb_info *sbi, 3423 struct bio *bio, enum page_type type); 3424 void f2fs_submit_merged_write(struct f2fs_sb_info *sbi, enum page_type type); 3425 void f2fs_submit_merged_write_cond(struct f2fs_sb_info *sbi, 3426 struct inode *inode, struct page *page, 3427 nid_t ino, enum page_type type); 3428 void f2fs_submit_merged_ipu_write(struct f2fs_sb_info *sbi, 3429 struct bio **bio, struct page *page); 3430 void f2fs_flush_merged_writes(struct f2fs_sb_info *sbi); 3431 int f2fs_submit_page_bio(struct f2fs_io_info *fio); 3432 int f2fs_merge_page_bio(struct f2fs_io_info *fio); 3433 void f2fs_submit_page_write(struct f2fs_io_info *fio); 3434 struct block_device *f2fs_target_device(struct f2fs_sb_info *sbi, 3435 block_t blk_addr, struct bio *bio); 3436 int f2fs_target_device_index(struct f2fs_sb_info *sbi, block_t blkaddr); 3437 void f2fs_set_data_blkaddr(struct dnode_of_data *dn); 3438 void f2fs_update_data_blkaddr(struct dnode_of_data *dn, block_t blkaddr); 3439 int f2fs_reserve_new_blocks(struct dnode_of_data *dn, blkcnt_t count); 3440 int f2fs_reserve_new_block(struct dnode_of_data *dn); 3441 int f2fs_get_block(struct dnode_of_data *dn, pgoff_t index); 3442 int f2fs_preallocate_blocks(struct kiocb *iocb, struct iov_iter *from); 3443 int f2fs_reserve_block(struct dnode_of_data *dn, pgoff_t index); 3444 struct page *f2fs_get_read_data_page(struct inode *inode, pgoff_t index, 3445 int op_flags, bool for_write); 3446 struct page *f2fs_find_data_page(struct inode *inode, pgoff_t index); 3447 struct page *f2fs_get_lock_data_page(struct inode *inode, pgoff_t index, 3448 bool for_write); 3449 struct page *f2fs_get_new_data_page(struct inode *inode, 3450 struct page *ipage, pgoff_t index, bool new_i_size); 3451 int f2fs_do_write_data_page(struct f2fs_io_info *fio); 3452 void f2fs_do_map_lock(struct f2fs_sb_info *sbi, int flag, bool lock); 3453 int f2fs_map_blocks(struct inode *inode, struct f2fs_map_blocks *map, 3454 int create, int flag); 3455 int f2fs_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, 3456 u64 start, u64 len); 3457 int f2fs_encrypt_one_page(struct f2fs_io_info *fio); 3458 bool f2fs_should_update_inplace(struct inode *inode, struct f2fs_io_info *fio); 3459 bool f2fs_should_update_outplace(struct inode *inode, struct f2fs_io_info *fio); 3460 int f2fs_write_single_data_page(struct page *page, int *submitted, 3461 struct bio **bio, sector_t *last_block, 3462 struct writeback_control *wbc, 3463 enum iostat_type io_type, 3464 int compr_blocks); 3465 void f2fs_invalidate_page(struct page *page, unsigned int offset, 3466 unsigned int length); 3467 int f2fs_release_page(struct page *page, gfp_t wait); 3468 #ifdef CONFIG_MIGRATION 3469 int f2fs_migrate_page(struct address_space *mapping, struct page *newpage, 3470 struct page *page, enum migrate_mode mode); 3471 #endif 3472 bool f2fs_overwrite_io(struct inode *inode, loff_t pos, size_t len); 3473 void f2fs_clear_page_cache_dirty_tag(struct page *page); 3474 int f2fs_init_post_read_processing(void); 3475 void f2fs_destroy_post_read_processing(void); 3476 int f2fs_init_post_read_wq(struct f2fs_sb_info *sbi); 3477 void f2fs_destroy_post_read_wq(struct f2fs_sb_info *sbi); 3478 3479 /* 3480 * gc.c 3481 */ 3482 int f2fs_start_gc_thread(struct f2fs_sb_info *sbi); 3483 void f2fs_stop_gc_thread(struct f2fs_sb_info *sbi); 3484 block_t f2fs_start_bidx_of_node(unsigned int node_ofs, struct inode *inode); 3485 int f2fs_gc(struct f2fs_sb_info *sbi, bool sync, bool background, 3486 unsigned int segno); 3487 void f2fs_build_gc_manager(struct f2fs_sb_info *sbi); 3488 int f2fs_resize_fs(struct f2fs_sb_info *sbi, __u64 block_count); 3489 3490 /* 3491 * recovery.c 3492 */ 3493 int f2fs_recover_fsync_data(struct f2fs_sb_info *sbi, bool check_only); 3494 bool f2fs_space_for_roll_forward(struct f2fs_sb_info *sbi); 3495 3496 /* 3497 * debug.c 3498 */ 3499 #ifdef CONFIG_F2FS_STAT_FS 3500 struct f2fs_stat_info { 3501 struct list_head stat_list; 3502 struct f2fs_sb_info *sbi; 3503 int all_area_segs, sit_area_segs, nat_area_segs, ssa_area_segs; 3504 int main_area_segs, main_area_sections, main_area_zones; 3505 unsigned long long hit_largest, hit_cached, hit_rbtree; 3506 unsigned long long hit_total, total_ext; 3507 int ext_tree, zombie_tree, ext_node; 3508 int ndirty_node, ndirty_dent, ndirty_meta, ndirty_imeta; 3509 int ndirty_data, ndirty_qdata; 3510 int inmem_pages; 3511 unsigned int ndirty_dirs, ndirty_files, nquota_files, ndirty_all; 3512 int nats, dirty_nats, sits, dirty_sits; 3513 int free_nids, avail_nids, alloc_nids; 3514 int total_count, utilization; 3515 int bg_gc, nr_wb_cp_data, nr_wb_data; 3516 int nr_rd_data, nr_rd_node, nr_rd_meta; 3517 int nr_dio_read, nr_dio_write; 3518 unsigned int io_skip_bggc, other_skip_bggc; 3519 int nr_flushing, nr_flushed, flush_list_empty; 3520 int nr_discarding, nr_discarded; 3521 int nr_discard_cmd; 3522 unsigned int undiscard_blks; 3523 int inline_xattr, inline_inode, inline_dir, append, update, orphans; 3524 int compr_inode, compr_blocks; 3525 int aw_cnt, max_aw_cnt, vw_cnt, max_vw_cnt; 3526 unsigned int valid_count, valid_node_count, valid_inode_count, discard_blks; 3527 unsigned int bimodal, avg_vblocks; 3528 int util_free, util_valid, util_invalid; 3529 int rsvd_segs, overp_segs; 3530 int dirty_count, node_pages, meta_pages; 3531 int prefree_count, call_count, cp_count, bg_cp_count; 3532 int tot_segs, node_segs, data_segs, free_segs, free_secs; 3533 int bg_node_segs, bg_data_segs; 3534 int tot_blks, data_blks, node_blks; 3535 int bg_data_blks, bg_node_blks; 3536 unsigned long long skipped_atomic_files[2]; 3537 int curseg[NR_CURSEG_TYPE]; 3538 int cursec[NR_CURSEG_TYPE]; 3539 int curzone[NR_CURSEG_TYPE]; 3540 unsigned int dirty_seg[NR_CURSEG_TYPE]; 3541 unsigned int full_seg[NR_CURSEG_TYPE]; 3542 unsigned int valid_blks[NR_CURSEG_TYPE]; 3543 3544 unsigned int meta_count[META_MAX]; 3545 unsigned int segment_count[2]; 3546 unsigned int block_count[2]; 3547 unsigned int inplace_count; 3548 unsigned long long base_mem, cache_mem, page_mem; 3549 }; 3550 3551 static inline struct f2fs_stat_info *F2FS_STAT(struct f2fs_sb_info *sbi) 3552 { 3553 return (struct f2fs_stat_info *)sbi->stat_info; 3554 } 3555 3556 #define stat_inc_cp_count(si) ((si)->cp_count++) 3557 #define stat_inc_bg_cp_count(si) ((si)->bg_cp_count++) 3558 #define stat_inc_call_count(si) ((si)->call_count++) 3559 #define stat_inc_bggc_count(si) ((si)->bg_gc++) 3560 #define stat_io_skip_bggc_count(sbi) ((sbi)->io_skip_bggc++) 3561 #define stat_other_skip_bggc_count(sbi) ((sbi)->other_skip_bggc++) 3562 #define stat_inc_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]++) 3563 #define stat_dec_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]--) 3564 #define stat_inc_total_hit(sbi) (atomic64_inc(&(sbi)->total_hit_ext)) 3565 #define stat_inc_rbtree_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_rbtree)) 3566 #define stat_inc_largest_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_largest)) 3567 #define stat_inc_cached_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_cached)) 3568 #define stat_inc_inline_xattr(inode) \ 3569 do { \ 3570 if (f2fs_has_inline_xattr(inode)) \ 3571 (atomic_inc(&F2FS_I_SB(inode)->inline_xattr)); \ 3572 } while (0) 3573 #define stat_dec_inline_xattr(inode) \ 3574 do { \ 3575 if (f2fs_has_inline_xattr(inode)) \ 3576 (atomic_dec(&F2FS_I_SB(inode)->inline_xattr)); \ 3577 } while (0) 3578 #define stat_inc_inline_inode(inode) \ 3579 do { \ 3580 if (f2fs_has_inline_data(inode)) \ 3581 (atomic_inc(&F2FS_I_SB(inode)->inline_inode)); \ 3582 } while (0) 3583 #define stat_dec_inline_inode(inode) \ 3584 do { \ 3585 if (f2fs_has_inline_data(inode)) \ 3586 (atomic_dec(&F2FS_I_SB(inode)->inline_inode)); \ 3587 } while (0) 3588 #define stat_inc_inline_dir(inode) \ 3589 do { \ 3590 if (f2fs_has_inline_dentry(inode)) \ 3591 (atomic_inc(&F2FS_I_SB(inode)->inline_dir)); \ 3592 } while (0) 3593 #define stat_dec_inline_dir(inode) \ 3594 do { \ 3595 if (f2fs_has_inline_dentry(inode)) \ 3596 (atomic_dec(&F2FS_I_SB(inode)->inline_dir)); \ 3597 } while (0) 3598 #define stat_inc_compr_inode(inode) \ 3599 do { \ 3600 if (f2fs_compressed_file(inode)) \ 3601 (atomic_inc(&F2FS_I_SB(inode)->compr_inode)); \ 3602 } while (0) 3603 #define stat_dec_compr_inode(inode) \ 3604 do { \ 3605 if (f2fs_compressed_file(inode)) \ 3606 (atomic_dec(&F2FS_I_SB(inode)->compr_inode)); \ 3607 } while (0) 3608 #define stat_add_compr_blocks(inode, blocks) \ 3609 (atomic_add(blocks, &F2FS_I_SB(inode)->compr_blocks)) 3610 #define stat_sub_compr_blocks(inode, blocks) \ 3611 (atomic_sub(blocks, &F2FS_I_SB(inode)->compr_blocks)) 3612 #define stat_inc_meta_count(sbi, blkaddr) \ 3613 do { \ 3614 if (blkaddr < SIT_I(sbi)->sit_base_addr) \ 3615 atomic_inc(&(sbi)->meta_count[META_CP]); \ 3616 else if (blkaddr < NM_I(sbi)->nat_blkaddr) \ 3617 atomic_inc(&(sbi)->meta_count[META_SIT]); \ 3618 else if (blkaddr < SM_I(sbi)->ssa_blkaddr) \ 3619 atomic_inc(&(sbi)->meta_count[META_NAT]); \ 3620 else if (blkaddr < SM_I(sbi)->main_blkaddr) \ 3621 atomic_inc(&(sbi)->meta_count[META_SSA]); \ 3622 } while (0) 3623 #define stat_inc_seg_type(sbi, curseg) \ 3624 ((sbi)->segment_count[(curseg)->alloc_type]++) 3625 #define stat_inc_block_count(sbi, curseg) \ 3626 ((sbi)->block_count[(curseg)->alloc_type]++) 3627 #define stat_inc_inplace_blocks(sbi) \ 3628 (atomic_inc(&(sbi)->inplace_count)) 3629 #define stat_update_max_atomic_write(inode) \ 3630 do { \ 3631 int cur = F2FS_I_SB(inode)->atomic_files; \ 3632 int max = atomic_read(&F2FS_I_SB(inode)->max_aw_cnt); \ 3633 if (cur > max) \ 3634 atomic_set(&F2FS_I_SB(inode)->max_aw_cnt, cur); \ 3635 } while (0) 3636 #define stat_inc_volatile_write(inode) \ 3637 (atomic_inc(&F2FS_I_SB(inode)->vw_cnt)) 3638 #define stat_dec_volatile_write(inode) \ 3639 (atomic_dec(&F2FS_I_SB(inode)->vw_cnt)) 3640 #define stat_update_max_volatile_write(inode) \ 3641 do { \ 3642 int cur = atomic_read(&F2FS_I_SB(inode)->vw_cnt); \ 3643 int max = atomic_read(&F2FS_I_SB(inode)->max_vw_cnt); \ 3644 if (cur > max) \ 3645 atomic_set(&F2FS_I_SB(inode)->max_vw_cnt, cur); \ 3646 } while (0) 3647 #define stat_inc_seg_count(sbi, type, gc_type) \ 3648 do { \ 3649 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 3650 si->tot_segs++; \ 3651 if ((type) == SUM_TYPE_DATA) { \ 3652 si->data_segs++; \ 3653 si->bg_data_segs += (gc_type == BG_GC) ? 1 : 0; \ 3654 } else { \ 3655 si->node_segs++; \ 3656 si->bg_node_segs += (gc_type == BG_GC) ? 1 : 0; \ 3657 } \ 3658 } while (0) 3659 3660 #define stat_inc_tot_blk_count(si, blks) \ 3661 ((si)->tot_blks += (blks)) 3662 3663 #define stat_inc_data_blk_count(sbi, blks, gc_type) \ 3664 do { \ 3665 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 3666 stat_inc_tot_blk_count(si, blks); \ 3667 si->data_blks += (blks); \ 3668 si->bg_data_blks += ((gc_type) == BG_GC) ? (blks) : 0; \ 3669 } while (0) 3670 3671 #define stat_inc_node_blk_count(sbi, blks, gc_type) \ 3672 do { \ 3673 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 3674 stat_inc_tot_blk_count(si, blks); \ 3675 si->node_blks += (blks); \ 3676 si->bg_node_blks += ((gc_type) == BG_GC) ? (blks) : 0; \ 3677 } while (0) 3678 3679 int f2fs_build_stats(struct f2fs_sb_info *sbi); 3680 void f2fs_destroy_stats(struct f2fs_sb_info *sbi); 3681 void __init f2fs_create_root_stats(void); 3682 void f2fs_destroy_root_stats(void); 3683 void f2fs_update_sit_info(struct f2fs_sb_info *sbi); 3684 #else 3685 #define stat_inc_cp_count(si) do { } while (0) 3686 #define stat_inc_bg_cp_count(si) do { } while (0) 3687 #define stat_inc_call_count(si) do { } while (0) 3688 #define stat_inc_bggc_count(si) do { } while (0) 3689 #define stat_io_skip_bggc_count(sbi) do { } while (0) 3690 #define stat_other_skip_bggc_count(sbi) do { } while (0) 3691 #define stat_inc_dirty_inode(sbi, type) do { } while (0) 3692 #define stat_dec_dirty_inode(sbi, type) do { } while (0) 3693 #define stat_inc_total_hit(sbi) do { } while (0) 3694 #define stat_inc_rbtree_node_hit(sbi) do { } while (0) 3695 #define stat_inc_largest_node_hit(sbi) do { } while (0) 3696 #define stat_inc_cached_node_hit(sbi) do { } while (0) 3697 #define stat_inc_inline_xattr(inode) do { } while (0) 3698 #define stat_dec_inline_xattr(inode) do { } while (0) 3699 #define stat_inc_inline_inode(inode) do { } while (0) 3700 #define stat_dec_inline_inode(inode) do { } while (0) 3701 #define stat_inc_inline_dir(inode) do { } while (0) 3702 #define stat_dec_inline_dir(inode) do { } while (0) 3703 #define stat_inc_compr_inode(inode) do { } while (0) 3704 #define stat_dec_compr_inode(inode) do { } while (0) 3705 #define stat_add_compr_blocks(inode, blocks) do { } while (0) 3706 #define stat_sub_compr_blocks(inode, blocks) do { } while (0) 3707 #define stat_inc_atomic_write(inode) do { } while (0) 3708 #define stat_dec_atomic_write(inode) do { } while (0) 3709 #define stat_update_max_atomic_write(inode) do { } while (0) 3710 #define stat_inc_volatile_write(inode) do { } while (0) 3711 #define stat_dec_volatile_write(inode) do { } while (0) 3712 #define stat_update_max_volatile_write(inode) do { } while (0) 3713 #define stat_inc_meta_count(sbi, blkaddr) do { } while (0) 3714 #define stat_inc_seg_type(sbi, curseg) do { } while (0) 3715 #define stat_inc_block_count(sbi, curseg) do { } while (0) 3716 #define stat_inc_inplace_blocks(sbi) do { } while (0) 3717 #define stat_inc_seg_count(sbi, type, gc_type) do { } while (0) 3718 #define stat_inc_tot_blk_count(si, blks) do { } while (0) 3719 #define stat_inc_data_blk_count(sbi, blks, gc_type) do { } while (0) 3720 #define stat_inc_node_blk_count(sbi, blks, gc_type) do { } while (0) 3721 3722 static inline int f2fs_build_stats(struct f2fs_sb_info *sbi) { return 0; } 3723 static inline void f2fs_destroy_stats(struct f2fs_sb_info *sbi) { } 3724 static inline void __init f2fs_create_root_stats(void) { } 3725 static inline void f2fs_destroy_root_stats(void) { } 3726 static inline void f2fs_update_sit_info(struct f2fs_sb_info *sbi) {} 3727 #endif 3728 3729 extern const struct file_operations f2fs_dir_operations; 3730 #ifdef CONFIG_UNICODE 3731 extern const struct dentry_operations f2fs_dentry_ops; 3732 #endif 3733 extern const struct file_operations f2fs_file_operations; 3734 extern const struct inode_operations f2fs_file_inode_operations; 3735 extern const struct address_space_operations f2fs_dblock_aops; 3736 extern const struct address_space_operations f2fs_node_aops; 3737 extern const struct address_space_operations f2fs_meta_aops; 3738 extern const struct inode_operations f2fs_dir_inode_operations; 3739 extern const struct inode_operations f2fs_symlink_inode_operations; 3740 extern const struct inode_operations f2fs_encrypted_symlink_inode_operations; 3741 extern const struct inode_operations f2fs_special_inode_operations; 3742 extern struct kmem_cache *f2fs_inode_entry_slab; 3743 3744 /* 3745 * inline.c 3746 */ 3747 bool f2fs_may_inline_data(struct inode *inode); 3748 bool f2fs_may_inline_dentry(struct inode *inode); 3749 void f2fs_do_read_inline_data(struct page *page, struct page *ipage); 3750 void f2fs_truncate_inline_inode(struct inode *inode, 3751 struct page *ipage, u64 from); 3752 int f2fs_read_inline_data(struct inode *inode, struct page *page); 3753 int f2fs_convert_inline_page(struct dnode_of_data *dn, struct page *page); 3754 int f2fs_convert_inline_inode(struct inode *inode); 3755 int f2fs_try_convert_inline_dir(struct inode *dir, struct dentry *dentry); 3756 int f2fs_write_inline_data(struct inode *inode, struct page *page); 3757 int f2fs_recover_inline_data(struct inode *inode, struct page *npage); 3758 struct f2fs_dir_entry *f2fs_find_in_inline_dir(struct inode *dir, 3759 const struct f2fs_filename *fname, 3760 struct page **res_page); 3761 int f2fs_make_empty_inline_dir(struct inode *inode, struct inode *parent, 3762 struct page *ipage); 3763 int f2fs_add_inline_entry(struct inode *dir, const struct f2fs_filename *fname, 3764 struct inode *inode, nid_t ino, umode_t mode); 3765 void f2fs_delete_inline_entry(struct f2fs_dir_entry *dentry, 3766 struct page *page, struct inode *dir, 3767 struct inode *inode); 3768 bool f2fs_empty_inline_dir(struct inode *dir); 3769 int f2fs_read_inline_dir(struct file *file, struct dir_context *ctx, 3770 struct fscrypt_str *fstr); 3771 int f2fs_inline_data_fiemap(struct inode *inode, 3772 struct fiemap_extent_info *fieinfo, 3773 __u64 start, __u64 len); 3774 3775 /* 3776 * shrinker.c 3777 */ 3778 unsigned long f2fs_shrink_count(struct shrinker *shrink, 3779 struct shrink_control *sc); 3780 unsigned long f2fs_shrink_scan(struct shrinker *shrink, 3781 struct shrink_control *sc); 3782 void f2fs_join_shrinker(struct f2fs_sb_info *sbi); 3783 void f2fs_leave_shrinker(struct f2fs_sb_info *sbi); 3784 3785 /* 3786 * extent_cache.c 3787 */ 3788 struct rb_entry *f2fs_lookup_rb_tree(struct rb_root_cached *root, 3789 struct rb_entry *cached_re, unsigned int ofs); 3790 struct rb_node **f2fs_lookup_rb_tree_for_insert(struct f2fs_sb_info *sbi, 3791 struct rb_root_cached *root, 3792 struct rb_node **parent, 3793 unsigned int ofs, bool *leftmost); 3794 struct rb_entry *f2fs_lookup_rb_tree_ret(struct rb_root_cached *root, 3795 struct rb_entry *cached_re, unsigned int ofs, 3796 struct rb_entry **prev_entry, struct rb_entry **next_entry, 3797 struct rb_node ***insert_p, struct rb_node **insert_parent, 3798 bool force, bool *leftmost); 3799 bool f2fs_check_rb_tree_consistence(struct f2fs_sb_info *sbi, 3800 struct rb_root_cached *root); 3801 unsigned int f2fs_shrink_extent_tree(struct f2fs_sb_info *sbi, int nr_shrink); 3802 void f2fs_init_extent_tree(struct inode *inode, struct page *ipage); 3803 void f2fs_drop_extent_tree(struct inode *inode); 3804 unsigned int f2fs_destroy_extent_node(struct inode *inode); 3805 void f2fs_destroy_extent_tree(struct inode *inode); 3806 bool f2fs_lookup_extent_cache(struct inode *inode, pgoff_t pgofs, 3807 struct extent_info *ei); 3808 void f2fs_update_extent_cache(struct dnode_of_data *dn); 3809 void f2fs_update_extent_cache_range(struct dnode_of_data *dn, 3810 pgoff_t fofs, block_t blkaddr, unsigned int len); 3811 void f2fs_init_extent_cache_info(struct f2fs_sb_info *sbi); 3812 int __init f2fs_create_extent_cache(void); 3813 void f2fs_destroy_extent_cache(void); 3814 3815 /* 3816 * sysfs.c 3817 */ 3818 int __init f2fs_init_sysfs(void); 3819 void f2fs_exit_sysfs(void); 3820 int f2fs_register_sysfs(struct f2fs_sb_info *sbi); 3821 void f2fs_unregister_sysfs(struct f2fs_sb_info *sbi); 3822 3823 /* verity.c */ 3824 extern const struct fsverity_operations f2fs_verityops; 3825 3826 /* 3827 * crypto support 3828 */ 3829 static inline bool f2fs_encrypted_file(struct inode *inode) 3830 { 3831 return IS_ENCRYPTED(inode) && S_ISREG(inode->i_mode); 3832 } 3833 3834 static inline void f2fs_set_encrypted_inode(struct inode *inode) 3835 { 3836 #ifdef CONFIG_FS_ENCRYPTION 3837 file_set_encrypt(inode); 3838 f2fs_set_inode_flags(inode); 3839 #endif 3840 } 3841 3842 /* 3843 * Returns true if the reads of the inode's data need to undergo some 3844 * postprocessing step, like decryption or authenticity verification. 3845 */ 3846 static inline bool f2fs_post_read_required(struct inode *inode) 3847 { 3848 return f2fs_encrypted_file(inode) || fsverity_active(inode) || 3849 f2fs_compressed_file(inode); 3850 } 3851 3852 /* 3853 * compress.c 3854 */ 3855 #ifdef CONFIG_F2FS_FS_COMPRESSION 3856 bool f2fs_is_compressed_page(struct page *page); 3857 struct page *f2fs_compress_control_page(struct page *page); 3858 int f2fs_prepare_compress_overwrite(struct inode *inode, 3859 struct page **pagep, pgoff_t index, void **fsdata); 3860 bool f2fs_compress_write_end(struct inode *inode, void *fsdata, 3861 pgoff_t index, unsigned copied); 3862 int f2fs_truncate_partial_cluster(struct inode *inode, u64 from, bool lock); 3863 void f2fs_compress_write_end_io(struct bio *bio, struct page *page); 3864 bool f2fs_is_compress_backend_ready(struct inode *inode); 3865 int f2fs_init_compress_mempool(void); 3866 void f2fs_destroy_compress_mempool(void); 3867 void f2fs_decompress_pages(struct bio *bio, struct page *page, bool verity); 3868 bool f2fs_cluster_is_empty(struct compress_ctx *cc); 3869 bool f2fs_cluster_can_merge_page(struct compress_ctx *cc, pgoff_t index); 3870 void f2fs_compress_ctx_add_page(struct compress_ctx *cc, struct page *page); 3871 int f2fs_write_multi_pages(struct compress_ctx *cc, 3872 int *submitted, 3873 struct writeback_control *wbc, 3874 enum iostat_type io_type); 3875 int f2fs_is_compressed_cluster(struct inode *inode, pgoff_t index); 3876 int f2fs_read_multi_pages(struct compress_ctx *cc, struct bio **bio_ret, 3877 unsigned nr_pages, sector_t *last_block_in_bio, 3878 bool is_readahead, bool for_write); 3879 struct decompress_io_ctx *f2fs_alloc_dic(struct compress_ctx *cc); 3880 void f2fs_free_dic(struct decompress_io_ctx *dic); 3881 void f2fs_decompress_end_io(struct page **rpages, 3882 unsigned int cluster_size, bool err, bool verity); 3883 int f2fs_init_compress_ctx(struct compress_ctx *cc); 3884 void f2fs_destroy_compress_ctx(struct compress_ctx *cc); 3885 void f2fs_init_compress_info(struct f2fs_sb_info *sbi); 3886 #else 3887 static inline bool f2fs_is_compressed_page(struct page *page) { return false; } 3888 static inline bool f2fs_is_compress_backend_ready(struct inode *inode) 3889 { 3890 if (!f2fs_compressed_file(inode)) 3891 return true; 3892 /* not support compression */ 3893 return false; 3894 } 3895 static inline struct page *f2fs_compress_control_page(struct page *page) 3896 { 3897 WARN_ON_ONCE(1); 3898 return ERR_PTR(-EINVAL); 3899 } 3900 static inline int f2fs_init_compress_mempool(void) { return 0; } 3901 static inline void f2fs_destroy_compress_mempool(void) { } 3902 #endif 3903 3904 static inline void set_compress_context(struct inode *inode) 3905 { 3906 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 3907 3908 F2FS_I(inode)->i_compress_algorithm = 3909 F2FS_OPTION(sbi).compress_algorithm; 3910 F2FS_I(inode)->i_log_cluster_size = 3911 F2FS_OPTION(sbi).compress_log_size; 3912 F2FS_I(inode)->i_cluster_size = 3913 1 << F2FS_I(inode)->i_log_cluster_size; 3914 F2FS_I(inode)->i_flags |= F2FS_COMPR_FL; 3915 set_inode_flag(inode, FI_COMPRESSED_FILE); 3916 stat_inc_compr_inode(inode); 3917 f2fs_mark_inode_dirty_sync(inode, true); 3918 } 3919 3920 static inline u64 f2fs_disable_compressed_file(struct inode *inode) 3921 { 3922 struct f2fs_inode_info *fi = F2FS_I(inode); 3923 3924 if (!f2fs_compressed_file(inode)) 3925 return 0; 3926 if (S_ISREG(inode->i_mode)) { 3927 if (get_dirty_pages(inode)) 3928 return 1; 3929 if (fi->i_compr_blocks) 3930 return fi->i_compr_blocks; 3931 } 3932 3933 fi->i_flags &= ~F2FS_COMPR_FL; 3934 stat_dec_compr_inode(inode); 3935 clear_inode_flag(inode, FI_COMPRESSED_FILE); 3936 f2fs_mark_inode_dirty_sync(inode, true); 3937 return 0; 3938 } 3939 3940 #define F2FS_FEATURE_FUNCS(name, flagname) \ 3941 static inline int f2fs_sb_has_##name(struct f2fs_sb_info *sbi) \ 3942 { \ 3943 return F2FS_HAS_FEATURE(sbi, F2FS_FEATURE_##flagname); \ 3944 } 3945 3946 F2FS_FEATURE_FUNCS(encrypt, ENCRYPT); 3947 F2FS_FEATURE_FUNCS(blkzoned, BLKZONED); 3948 F2FS_FEATURE_FUNCS(extra_attr, EXTRA_ATTR); 3949 F2FS_FEATURE_FUNCS(project_quota, PRJQUOTA); 3950 F2FS_FEATURE_FUNCS(inode_chksum, INODE_CHKSUM); 3951 F2FS_FEATURE_FUNCS(flexible_inline_xattr, FLEXIBLE_INLINE_XATTR); 3952 F2FS_FEATURE_FUNCS(quota_ino, QUOTA_INO); 3953 F2FS_FEATURE_FUNCS(inode_crtime, INODE_CRTIME); 3954 F2FS_FEATURE_FUNCS(lost_found, LOST_FOUND); 3955 F2FS_FEATURE_FUNCS(verity, VERITY); 3956 F2FS_FEATURE_FUNCS(sb_chksum, SB_CHKSUM); 3957 F2FS_FEATURE_FUNCS(casefold, CASEFOLD); 3958 F2FS_FEATURE_FUNCS(compression, COMPRESSION); 3959 3960 #ifdef CONFIG_BLK_DEV_ZONED 3961 static inline bool f2fs_blkz_is_seq(struct f2fs_sb_info *sbi, int devi, 3962 block_t blkaddr) 3963 { 3964 unsigned int zno = blkaddr >> sbi->log_blocks_per_blkz; 3965 3966 return test_bit(zno, FDEV(devi).blkz_seq); 3967 } 3968 #endif 3969 3970 static inline bool f2fs_hw_should_discard(struct f2fs_sb_info *sbi) 3971 { 3972 return f2fs_sb_has_blkzoned(sbi); 3973 } 3974 3975 static inline bool f2fs_bdev_support_discard(struct block_device *bdev) 3976 { 3977 return blk_queue_discard(bdev_get_queue(bdev)) || 3978 bdev_is_zoned(bdev); 3979 } 3980 3981 static inline bool f2fs_hw_support_discard(struct f2fs_sb_info *sbi) 3982 { 3983 int i; 3984 3985 if (!f2fs_is_multi_device(sbi)) 3986 return f2fs_bdev_support_discard(sbi->sb->s_bdev); 3987 3988 for (i = 0; i < sbi->s_ndevs; i++) 3989 if (f2fs_bdev_support_discard(FDEV(i).bdev)) 3990 return true; 3991 return false; 3992 } 3993 3994 static inline bool f2fs_realtime_discard_enable(struct f2fs_sb_info *sbi) 3995 { 3996 return (test_opt(sbi, DISCARD) && f2fs_hw_support_discard(sbi)) || 3997 f2fs_hw_should_discard(sbi); 3998 } 3999 4000 static inline bool f2fs_hw_is_readonly(struct f2fs_sb_info *sbi) 4001 { 4002 int i; 4003 4004 if (!f2fs_is_multi_device(sbi)) 4005 return bdev_read_only(sbi->sb->s_bdev); 4006 4007 for (i = 0; i < sbi->s_ndevs; i++) 4008 if (bdev_read_only(FDEV(i).bdev)) 4009 return true; 4010 return false; 4011 } 4012 4013 static inline bool f2fs_lfs_mode(struct f2fs_sb_info *sbi) 4014 { 4015 return F2FS_OPTION(sbi).fs_mode == FS_MODE_LFS; 4016 } 4017 4018 static inline bool f2fs_may_compress(struct inode *inode) 4019 { 4020 if (IS_SWAPFILE(inode) || f2fs_is_pinned_file(inode) || 4021 f2fs_is_atomic_file(inode) || 4022 f2fs_is_volatile_file(inode)) 4023 return false; 4024 return S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode); 4025 } 4026 4027 static inline void f2fs_i_compr_blocks_update(struct inode *inode, 4028 u64 blocks, bool add) 4029 { 4030 int diff = F2FS_I(inode)->i_cluster_size - blocks; 4031 4032 /* don't update i_compr_blocks if saved blocks were released */ 4033 if (!add && !F2FS_I(inode)->i_compr_blocks) 4034 return; 4035 4036 if (add) { 4037 F2FS_I(inode)->i_compr_blocks += diff; 4038 stat_add_compr_blocks(inode, diff); 4039 } else { 4040 F2FS_I(inode)->i_compr_blocks -= diff; 4041 stat_sub_compr_blocks(inode, diff); 4042 } 4043 f2fs_mark_inode_dirty_sync(inode, true); 4044 } 4045 4046 static inline int block_unaligned_IO(struct inode *inode, 4047 struct kiocb *iocb, struct iov_iter *iter) 4048 { 4049 unsigned int i_blkbits = READ_ONCE(inode->i_blkbits); 4050 unsigned int blocksize_mask = (1 << i_blkbits) - 1; 4051 loff_t offset = iocb->ki_pos; 4052 unsigned long align = offset | iov_iter_alignment(iter); 4053 4054 return align & blocksize_mask; 4055 } 4056 4057 static inline int allow_outplace_dio(struct inode *inode, 4058 struct kiocb *iocb, struct iov_iter *iter) 4059 { 4060 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 4061 int rw = iov_iter_rw(iter); 4062 4063 return (f2fs_lfs_mode(sbi) && (rw == WRITE) && 4064 !block_unaligned_IO(inode, iocb, iter)); 4065 } 4066 4067 static inline bool f2fs_force_buffered_io(struct inode *inode, 4068 struct kiocb *iocb, struct iov_iter *iter) 4069 { 4070 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 4071 int rw = iov_iter_rw(iter); 4072 4073 if (f2fs_post_read_required(inode)) 4074 return true; 4075 if (f2fs_is_multi_device(sbi)) 4076 return true; 4077 /* 4078 * for blkzoned device, fallback direct IO to buffered IO, so 4079 * all IOs can be serialized by log-structured write. 4080 */ 4081 if (f2fs_sb_has_blkzoned(sbi)) 4082 return true; 4083 if (f2fs_lfs_mode(sbi) && (rw == WRITE)) { 4084 if (block_unaligned_IO(inode, iocb, iter)) 4085 return true; 4086 if (F2FS_IO_ALIGNED(sbi)) 4087 return true; 4088 } 4089 if (is_sbi_flag_set(F2FS_I_SB(inode), SBI_CP_DISABLED) && 4090 !IS_SWAPFILE(inode)) 4091 return true; 4092 4093 return false; 4094 } 4095 4096 #ifdef CONFIG_F2FS_FAULT_INJECTION 4097 extern void f2fs_build_fault_attr(struct f2fs_sb_info *sbi, unsigned int rate, 4098 unsigned int type); 4099 #else 4100 #define f2fs_build_fault_attr(sbi, rate, type) do { } while (0) 4101 #endif 4102 4103 static inline bool is_journalled_quota(struct f2fs_sb_info *sbi) 4104 { 4105 #ifdef CONFIG_QUOTA 4106 if (f2fs_sb_has_quota_ino(sbi)) 4107 return true; 4108 if (F2FS_OPTION(sbi).s_qf_names[USRQUOTA] || 4109 F2FS_OPTION(sbi).s_qf_names[GRPQUOTA] || 4110 F2FS_OPTION(sbi).s_qf_names[PRJQUOTA]) 4111 return true; 4112 #endif 4113 return false; 4114 } 4115 4116 #define EFSBADCRC EBADMSG /* Bad CRC detected */ 4117 #define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */ 4118 4119 #endif /* _LINUX_F2FS_H */ 4120