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 bool test_dummy_encryption; /* test dummy encryption */ 142 block_t unusable_cap; /* Amount of space allowed to be 143 * unusable when disabling checkpoint 144 */ 145 146 /* For compression */ 147 unsigned char compress_algorithm; /* algorithm type */ 148 unsigned compress_log_size; /* cluster log size */ 149 unsigned char compress_ext_cnt; /* extension count */ 150 unsigned char extensions[COMPRESS_EXT_NUM][F2FS_EXTENSION_LEN]; /* extensions */ 151 }; 152 153 #define F2FS_FEATURE_ENCRYPT 0x0001 154 #define F2FS_FEATURE_BLKZONED 0x0002 155 #define F2FS_FEATURE_ATOMIC_WRITE 0x0004 156 #define F2FS_FEATURE_EXTRA_ATTR 0x0008 157 #define F2FS_FEATURE_PRJQUOTA 0x0010 158 #define F2FS_FEATURE_INODE_CHKSUM 0x0020 159 #define F2FS_FEATURE_FLEXIBLE_INLINE_XATTR 0x0040 160 #define F2FS_FEATURE_QUOTA_INO 0x0080 161 #define F2FS_FEATURE_INODE_CRTIME 0x0100 162 #define F2FS_FEATURE_LOST_FOUND 0x0200 163 #define F2FS_FEATURE_VERITY 0x0400 164 #define F2FS_FEATURE_SB_CHKSUM 0x0800 165 #define F2FS_FEATURE_CASEFOLD 0x1000 166 #define F2FS_FEATURE_COMPRESSION 0x2000 167 168 #define __F2FS_HAS_FEATURE(raw_super, mask) \ 169 ((raw_super->feature & cpu_to_le32(mask)) != 0) 170 #define F2FS_HAS_FEATURE(sbi, mask) __F2FS_HAS_FEATURE(sbi->raw_super, mask) 171 #define F2FS_SET_FEATURE(sbi, mask) \ 172 (sbi->raw_super->feature |= cpu_to_le32(mask)) 173 #define F2FS_CLEAR_FEATURE(sbi, mask) \ 174 (sbi->raw_super->feature &= ~cpu_to_le32(mask)) 175 176 /* 177 * Default values for user and/or group using reserved blocks 178 */ 179 #define F2FS_DEF_RESUID 0 180 #define F2FS_DEF_RESGID 0 181 182 /* 183 * For checkpoint manager 184 */ 185 enum { 186 NAT_BITMAP, 187 SIT_BITMAP 188 }; 189 190 #define CP_UMOUNT 0x00000001 191 #define CP_FASTBOOT 0x00000002 192 #define CP_SYNC 0x00000004 193 #define CP_RECOVERY 0x00000008 194 #define CP_DISCARD 0x00000010 195 #define CP_TRIMMED 0x00000020 196 #define CP_PAUSE 0x00000040 197 198 #define MAX_DISCARD_BLOCKS(sbi) BLKS_PER_SEC(sbi) 199 #define DEF_MAX_DISCARD_REQUEST 8 /* issue 8 discards per round */ 200 #define DEF_MIN_DISCARD_ISSUE_TIME 50 /* 50 ms, if exists */ 201 #define DEF_MID_DISCARD_ISSUE_TIME 500 /* 500 ms, if device busy */ 202 #define DEF_MAX_DISCARD_ISSUE_TIME 60000 /* 60 s, if no candidates */ 203 #define DEF_DISCARD_URGENT_UTIL 80 /* do more discard over 80% */ 204 #define DEF_CP_INTERVAL 60 /* 60 secs */ 205 #define DEF_IDLE_INTERVAL 5 /* 5 secs */ 206 #define DEF_DISABLE_INTERVAL 5 /* 5 secs */ 207 #define DEF_DISABLE_QUICK_INTERVAL 1 /* 1 secs */ 208 #define DEF_UMOUNT_DISCARD_TIMEOUT 5 /* 5 secs */ 209 210 struct cp_control { 211 int reason; 212 __u64 trim_start; 213 __u64 trim_end; 214 __u64 trim_minlen; 215 }; 216 217 /* 218 * indicate meta/data type 219 */ 220 enum { 221 META_CP, 222 META_NAT, 223 META_SIT, 224 META_SSA, 225 META_MAX, 226 META_POR, 227 DATA_GENERIC, /* check range only */ 228 DATA_GENERIC_ENHANCE, /* strong check on range and segment bitmap */ 229 DATA_GENERIC_ENHANCE_READ, /* 230 * strong check on range and segment 231 * bitmap but no warning due to race 232 * condition of read on truncated area 233 * by extent_cache 234 */ 235 META_GENERIC, 236 }; 237 238 /* for the list of ino */ 239 enum { 240 ORPHAN_INO, /* for orphan ino list */ 241 APPEND_INO, /* for append ino list */ 242 UPDATE_INO, /* for update ino list */ 243 TRANS_DIR_INO, /* for trasactions dir ino list */ 244 FLUSH_INO, /* for multiple device flushing */ 245 MAX_INO_ENTRY, /* max. list */ 246 }; 247 248 struct ino_entry { 249 struct list_head list; /* list head */ 250 nid_t ino; /* inode number */ 251 unsigned int dirty_device; /* dirty device bitmap */ 252 }; 253 254 /* for the list of inodes to be GCed */ 255 struct inode_entry { 256 struct list_head list; /* list head */ 257 struct inode *inode; /* vfs inode pointer */ 258 }; 259 260 struct fsync_node_entry { 261 struct list_head list; /* list head */ 262 struct page *page; /* warm node page pointer */ 263 unsigned int seq_id; /* sequence id */ 264 }; 265 266 /* for the bitmap indicate blocks to be discarded */ 267 struct discard_entry { 268 struct list_head list; /* list head */ 269 block_t start_blkaddr; /* start blockaddr of current segment */ 270 unsigned char discard_map[SIT_VBLOCK_MAP_SIZE]; /* segment discard bitmap */ 271 }; 272 273 /* default discard granularity of inner discard thread, unit: block count */ 274 #define DEFAULT_DISCARD_GRANULARITY 16 275 276 /* max discard pend list number */ 277 #define MAX_PLIST_NUM 512 278 #define plist_idx(blk_num) ((blk_num) >= MAX_PLIST_NUM ? \ 279 (MAX_PLIST_NUM - 1) : ((blk_num) - 1)) 280 281 enum { 282 D_PREP, /* initial */ 283 D_PARTIAL, /* partially submitted */ 284 D_SUBMIT, /* all submitted */ 285 D_DONE, /* finished */ 286 }; 287 288 struct discard_info { 289 block_t lstart; /* logical start address */ 290 block_t len; /* length */ 291 block_t start; /* actual start address in dev */ 292 }; 293 294 struct discard_cmd { 295 struct rb_node rb_node; /* rb node located in rb-tree */ 296 union { 297 struct { 298 block_t lstart; /* logical start address */ 299 block_t len; /* length */ 300 block_t start; /* actual start address in dev */ 301 }; 302 struct discard_info di; /* discard info */ 303 304 }; 305 struct list_head list; /* command list */ 306 struct completion wait; /* compleation */ 307 struct block_device *bdev; /* bdev */ 308 unsigned short ref; /* reference count */ 309 unsigned char state; /* state */ 310 unsigned char queued; /* queued discard */ 311 int error; /* bio error */ 312 spinlock_t lock; /* for state/bio_ref updating */ 313 unsigned short bio_ref; /* bio reference count */ 314 }; 315 316 enum { 317 DPOLICY_BG, 318 DPOLICY_FORCE, 319 DPOLICY_FSTRIM, 320 DPOLICY_UMOUNT, 321 MAX_DPOLICY, 322 }; 323 324 struct discard_policy { 325 int type; /* type of discard */ 326 unsigned int min_interval; /* used for candidates exist */ 327 unsigned int mid_interval; /* used for device busy */ 328 unsigned int max_interval; /* used for candidates not exist */ 329 unsigned int max_requests; /* # of discards issued per round */ 330 unsigned int io_aware_gran; /* minimum granularity discard not be aware of I/O */ 331 bool io_aware; /* issue discard in idle time */ 332 bool sync; /* submit discard with REQ_SYNC flag */ 333 bool ordered; /* issue discard by lba order */ 334 bool timeout; /* discard timeout for put_super */ 335 unsigned int granularity; /* discard granularity */ 336 }; 337 338 struct discard_cmd_control { 339 struct task_struct *f2fs_issue_discard; /* discard thread */ 340 struct list_head entry_list; /* 4KB discard entry list */ 341 struct list_head pend_list[MAX_PLIST_NUM];/* store pending entries */ 342 struct list_head wait_list; /* store on-flushing entries */ 343 struct list_head fstrim_list; /* in-flight discard from fstrim */ 344 wait_queue_head_t discard_wait_queue; /* waiting queue for wake-up */ 345 unsigned int discard_wake; /* to wake up discard thread */ 346 struct mutex cmd_lock; 347 unsigned int nr_discards; /* # of discards in the list */ 348 unsigned int max_discards; /* max. discards to be issued */ 349 unsigned int discard_granularity; /* discard granularity */ 350 unsigned int undiscard_blks; /* # of undiscard blocks */ 351 unsigned int next_pos; /* next discard position */ 352 atomic_t issued_discard; /* # of issued discard */ 353 atomic_t queued_discard; /* # of queued discard */ 354 atomic_t discard_cmd_cnt; /* # of cached cmd count */ 355 struct rb_root_cached root; /* root of discard rb-tree */ 356 bool rbtree_check; /* config for consistence check */ 357 }; 358 359 /* for the list of fsync inodes, used only during recovery */ 360 struct fsync_inode_entry { 361 struct list_head list; /* list head */ 362 struct inode *inode; /* vfs inode pointer */ 363 block_t blkaddr; /* block address locating the last fsync */ 364 block_t last_dentry; /* block address locating the last dentry */ 365 }; 366 367 #define nats_in_cursum(jnl) (le16_to_cpu((jnl)->n_nats)) 368 #define sits_in_cursum(jnl) (le16_to_cpu((jnl)->n_sits)) 369 370 #define nat_in_journal(jnl, i) ((jnl)->nat_j.entries[i].ne) 371 #define nid_in_journal(jnl, i) ((jnl)->nat_j.entries[i].nid) 372 #define sit_in_journal(jnl, i) ((jnl)->sit_j.entries[i].se) 373 #define segno_in_journal(jnl, i) ((jnl)->sit_j.entries[i].segno) 374 375 #define MAX_NAT_JENTRIES(jnl) (NAT_JOURNAL_ENTRIES - nats_in_cursum(jnl)) 376 #define MAX_SIT_JENTRIES(jnl) (SIT_JOURNAL_ENTRIES - sits_in_cursum(jnl)) 377 378 static inline int update_nats_in_cursum(struct f2fs_journal *journal, int i) 379 { 380 int before = nats_in_cursum(journal); 381 382 journal->n_nats = cpu_to_le16(before + i); 383 return before; 384 } 385 386 static inline int update_sits_in_cursum(struct f2fs_journal *journal, int i) 387 { 388 int before = sits_in_cursum(journal); 389 390 journal->n_sits = cpu_to_le16(before + i); 391 return before; 392 } 393 394 static inline bool __has_cursum_space(struct f2fs_journal *journal, 395 int size, int type) 396 { 397 if (type == NAT_JOURNAL) 398 return size <= MAX_NAT_JENTRIES(journal); 399 return size <= MAX_SIT_JENTRIES(journal); 400 } 401 402 /* 403 * ioctl commands 404 */ 405 #define F2FS_IOC_GETFLAGS FS_IOC_GETFLAGS 406 #define F2FS_IOC_SETFLAGS FS_IOC_SETFLAGS 407 #define F2FS_IOC_GETVERSION FS_IOC_GETVERSION 408 409 #define F2FS_IOCTL_MAGIC 0xf5 410 #define F2FS_IOC_START_ATOMIC_WRITE _IO(F2FS_IOCTL_MAGIC, 1) 411 #define F2FS_IOC_COMMIT_ATOMIC_WRITE _IO(F2FS_IOCTL_MAGIC, 2) 412 #define F2FS_IOC_START_VOLATILE_WRITE _IO(F2FS_IOCTL_MAGIC, 3) 413 #define F2FS_IOC_RELEASE_VOLATILE_WRITE _IO(F2FS_IOCTL_MAGIC, 4) 414 #define F2FS_IOC_ABORT_VOLATILE_WRITE _IO(F2FS_IOCTL_MAGIC, 5) 415 #define F2FS_IOC_GARBAGE_COLLECT _IOW(F2FS_IOCTL_MAGIC, 6, __u32) 416 #define F2FS_IOC_WRITE_CHECKPOINT _IO(F2FS_IOCTL_MAGIC, 7) 417 #define F2FS_IOC_DEFRAGMENT _IOWR(F2FS_IOCTL_MAGIC, 8, \ 418 struct f2fs_defragment) 419 #define F2FS_IOC_MOVE_RANGE _IOWR(F2FS_IOCTL_MAGIC, 9, \ 420 struct f2fs_move_range) 421 #define F2FS_IOC_FLUSH_DEVICE _IOW(F2FS_IOCTL_MAGIC, 10, \ 422 struct f2fs_flush_device) 423 #define F2FS_IOC_GARBAGE_COLLECT_RANGE _IOW(F2FS_IOCTL_MAGIC, 11, \ 424 struct f2fs_gc_range) 425 #define F2FS_IOC_GET_FEATURES _IOR(F2FS_IOCTL_MAGIC, 12, __u32) 426 #define F2FS_IOC_SET_PIN_FILE _IOW(F2FS_IOCTL_MAGIC, 13, __u32) 427 #define F2FS_IOC_GET_PIN_FILE _IOR(F2FS_IOCTL_MAGIC, 14, __u32) 428 #define F2FS_IOC_PRECACHE_EXTENTS _IO(F2FS_IOCTL_MAGIC, 15) 429 #define F2FS_IOC_RESIZE_FS _IOW(F2FS_IOCTL_MAGIC, 16, __u64) 430 #define F2FS_IOC_GET_COMPRESS_BLOCKS _IOR(F2FS_IOCTL_MAGIC, 17, __u64) 431 #define F2FS_IOC_RELEASE_COMPRESS_BLOCKS \ 432 _IOR(F2FS_IOCTL_MAGIC, 18, __u64) 433 434 #define F2FS_IOC_GET_VOLUME_NAME FS_IOC_GETFSLABEL 435 #define F2FS_IOC_SET_VOLUME_NAME FS_IOC_SETFSLABEL 436 437 #define F2FS_IOC_SET_ENCRYPTION_POLICY FS_IOC_SET_ENCRYPTION_POLICY 438 #define F2FS_IOC_GET_ENCRYPTION_POLICY FS_IOC_GET_ENCRYPTION_POLICY 439 #define F2FS_IOC_GET_ENCRYPTION_PWSALT FS_IOC_GET_ENCRYPTION_PWSALT 440 441 /* 442 * should be same as XFS_IOC_GOINGDOWN. 443 * Flags for going down operation used by FS_IOC_GOINGDOWN 444 */ 445 #define F2FS_IOC_SHUTDOWN _IOR('X', 125, __u32) /* Shutdown */ 446 #define F2FS_GOING_DOWN_FULLSYNC 0x0 /* going down with full sync */ 447 #define F2FS_GOING_DOWN_METASYNC 0x1 /* going down with metadata */ 448 #define F2FS_GOING_DOWN_NOSYNC 0x2 /* going down */ 449 #define F2FS_GOING_DOWN_METAFLUSH 0x3 /* going down with meta flush */ 450 #define F2FS_GOING_DOWN_NEED_FSCK 0x4 /* going down to trigger fsck */ 451 452 #if defined(__KERNEL__) && defined(CONFIG_COMPAT) 453 /* 454 * ioctl commands in 32 bit emulation 455 */ 456 #define F2FS_IOC32_GETFLAGS FS_IOC32_GETFLAGS 457 #define F2FS_IOC32_SETFLAGS FS_IOC32_SETFLAGS 458 #define F2FS_IOC32_GETVERSION FS_IOC32_GETVERSION 459 #endif 460 461 #define F2FS_IOC_FSGETXATTR FS_IOC_FSGETXATTR 462 #define F2FS_IOC_FSSETXATTR FS_IOC_FSSETXATTR 463 464 struct f2fs_gc_range { 465 u32 sync; 466 u64 start; 467 u64 len; 468 }; 469 470 struct f2fs_defragment { 471 u64 start; 472 u64 len; 473 }; 474 475 struct f2fs_move_range { 476 u32 dst_fd; /* destination fd */ 477 u64 pos_in; /* start position in src_fd */ 478 u64 pos_out; /* start position in dst_fd */ 479 u64 len; /* size to move */ 480 }; 481 482 struct f2fs_flush_device { 483 u32 dev_num; /* device number to flush */ 484 u32 segments; /* # of segments to flush */ 485 }; 486 487 /* for inline stuff */ 488 #define DEF_INLINE_RESERVED_SIZE 1 489 static inline int get_extra_isize(struct inode *inode); 490 static inline int get_inline_xattr_addrs(struct inode *inode); 491 #define MAX_INLINE_DATA(inode) (sizeof(__le32) * \ 492 (CUR_ADDRS_PER_INODE(inode) - \ 493 get_inline_xattr_addrs(inode) - \ 494 DEF_INLINE_RESERVED_SIZE)) 495 496 /* for inline dir */ 497 #define NR_INLINE_DENTRY(inode) (MAX_INLINE_DATA(inode) * BITS_PER_BYTE / \ 498 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \ 499 BITS_PER_BYTE + 1)) 500 #define INLINE_DENTRY_BITMAP_SIZE(inode) \ 501 DIV_ROUND_UP(NR_INLINE_DENTRY(inode), BITS_PER_BYTE) 502 #define INLINE_RESERVED_SIZE(inode) (MAX_INLINE_DATA(inode) - \ 503 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \ 504 NR_INLINE_DENTRY(inode) + \ 505 INLINE_DENTRY_BITMAP_SIZE(inode))) 506 507 /* 508 * For INODE and NODE manager 509 */ 510 /* for directory operations */ 511 512 struct f2fs_filename { 513 /* 514 * The filename the user specified. This is NULL for some 515 * filesystem-internal operations, e.g. converting an inline directory 516 * to a non-inline one, or roll-forward recovering an encrypted dentry. 517 */ 518 const struct qstr *usr_fname; 519 520 /* 521 * The on-disk filename. For encrypted directories, this is encrypted. 522 * This may be NULL for lookups in an encrypted dir without the key. 523 */ 524 struct fscrypt_str disk_name; 525 526 /* The dirhash of this filename */ 527 f2fs_hash_t hash; 528 529 #ifdef CONFIG_FS_ENCRYPTION 530 /* 531 * For lookups in encrypted directories: either the buffer backing 532 * disk_name, or a buffer that holds the decoded no-key name. 533 */ 534 struct fscrypt_str crypto_buf; 535 #endif 536 #ifdef CONFIG_UNICODE 537 /* 538 * For casefolded directories: the casefolded name, but it's left NULL 539 * if the original name is not valid Unicode or if the filesystem is 540 * doing an internal operation where usr_fname is also NULL. In these 541 * cases we fall back to treating the name as an opaque byte sequence. 542 */ 543 struct fscrypt_str cf_name; 544 #endif 545 }; 546 547 struct f2fs_dentry_ptr { 548 struct inode *inode; 549 void *bitmap; 550 struct f2fs_dir_entry *dentry; 551 __u8 (*filename)[F2FS_SLOT_LEN]; 552 int max; 553 int nr_bitmap; 554 }; 555 556 static inline void make_dentry_ptr_block(struct inode *inode, 557 struct f2fs_dentry_ptr *d, struct f2fs_dentry_block *t) 558 { 559 d->inode = inode; 560 d->max = NR_DENTRY_IN_BLOCK; 561 d->nr_bitmap = SIZE_OF_DENTRY_BITMAP; 562 d->bitmap = t->dentry_bitmap; 563 d->dentry = t->dentry; 564 d->filename = t->filename; 565 } 566 567 static inline void make_dentry_ptr_inline(struct inode *inode, 568 struct f2fs_dentry_ptr *d, void *t) 569 { 570 int entry_cnt = NR_INLINE_DENTRY(inode); 571 int bitmap_size = INLINE_DENTRY_BITMAP_SIZE(inode); 572 int reserved_size = INLINE_RESERVED_SIZE(inode); 573 574 d->inode = inode; 575 d->max = entry_cnt; 576 d->nr_bitmap = bitmap_size; 577 d->bitmap = t; 578 d->dentry = t + bitmap_size + reserved_size; 579 d->filename = t + bitmap_size + reserved_size + 580 SIZE_OF_DIR_ENTRY * entry_cnt; 581 } 582 583 /* 584 * XATTR_NODE_OFFSET stores xattrs to one node block per file keeping -1 585 * as its node offset to distinguish from index node blocks. 586 * But some bits are used to mark the node block. 587 */ 588 #define XATTR_NODE_OFFSET ((((unsigned int)-1) << OFFSET_BIT_SHIFT) \ 589 >> OFFSET_BIT_SHIFT) 590 enum { 591 ALLOC_NODE, /* allocate a new node page if needed */ 592 LOOKUP_NODE, /* look up a node without readahead */ 593 LOOKUP_NODE_RA, /* 594 * look up a node with readahead called 595 * by get_data_block. 596 */ 597 }; 598 599 #define DEFAULT_RETRY_IO_COUNT 8 /* maximum retry read IO count */ 600 601 /* congestion wait timeout value, default: 20ms */ 602 #define DEFAULT_IO_TIMEOUT (msecs_to_jiffies(20)) 603 604 /* maximum retry quota flush count */ 605 #define DEFAULT_RETRY_QUOTA_FLUSH_COUNT 8 606 607 #define F2FS_LINK_MAX 0xffffffff /* maximum link count per file */ 608 609 #define MAX_DIR_RA_PAGES 4 /* maximum ra pages of dir */ 610 611 /* for in-memory extent cache entry */ 612 #define F2FS_MIN_EXTENT_LEN 64 /* minimum extent length */ 613 614 /* number of extent info in extent cache we try to shrink */ 615 #define EXTENT_CACHE_SHRINK_NUMBER 128 616 617 struct rb_entry { 618 struct rb_node rb_node; /* rb node located in rb-tree */ 619 unsigned int ofs; /* start offset of the entry */ 620 unsigned int len; /* length of the entry */ 621 }; 622 623 struct extent_info { 624 unsigned int fofs; /* start offset in a file */ 625 unsigned int len; /* length of the extent */ 626 u32 blk; /* start block address of the extent */ 627 }; 628 629 struct extent_node { 630 struct rb_node rb_node; /* rb node located in rb-tree */ 631 struct extent_info ei; /* extent info */ 632 struct list_head list; /* node in global extent list of sbi */ 633 struct extent_tree *et; /* extent tree pointer */ 634 }; 635 636 struct extent_tree { 637 nid_t ino; /* inode number */ 638 struct rb_root_cached root; /* root of extent info rb-tree */ 639 struct extent_node *cached_en; /* recently accessed extent node */ 640 struct extent_info largest; /* largested extent info */ 641 struct list_head list; /* to be used by sbi->zombie_list */ 642 rwlock_t lock; /* protect extent info rb-tree */ 643 atomic_t node_cnt; /* # of extent node in rb-tree*/ 644 bool largest_updated; /* largest extent updated */ 645 }; 646 647 /* 648 * This structure is taken from ext4_map_blocks. 649 * 650 * Note that, however, f2fs uses NEW and MAPPED flags for f2fs_map_blocks(). 651 */ 652 #define F2FS_MAP_NEW (1 << BH_New) 653 #define F2FS_MAP_MAPPED (1 << BH_Mapped) 654 #define F2FS_MAP_UNWRITTEN (1 << BH_Unwritten) 655 #define F2FS_MAP_FLAGS (F2FS_MAP_NEW | F2FS_MAP_MAPPED |\ 656 F2FS_MAP_UNWRITTEN) 657 658 struct f2fs_map_blocks { 659 block_t m_pblk; 660 block_t m_lblk; 661 unsigned int m_len; 662 unsigned int m_flags; 663 pgoff_t *m_next_pgofs; /* point next possible non-hole pgofs */ 664 pgoff_t *m_next_extent; /* point to next possible extent */ 665 int m_seg_type; 666 bool m_may_create; /* indicate it is from write path */ 667 }; 668 669 /* for flag in get_data_block */ 670 enum { 671 F2FS_GET_BLOCK_DEFAULT, 672 F2FS_GET_BLOCK_FIEMAP, 673 F2FS_GET_BLOCK_BMAP, 674 F2FS_GET_BLOCK_DIO, 675 F2FS_GET_BLOCK_PRE_DIO, 676 F2FS_GET_BLOCK_PRE_AIO, 677 F2FS_GET_BLOCK_PRECACHE, 678 }; 679 680 /* 681 * i_advise uses FADVISE_XXX_BIT. We can add additional hints later. 682 */ 683 #define FADVISE_COLD_BIT 0x01 684 #define FADVISE_LOST_PINO_BIT 0x02 685 #define FADVISE_ENCRYPT_BIT 0x04 686 #define FADVISE_ENC_NAME_BIT 0x08 687 #define FADVISE_KEEP_SIZE_BIT 0x10 688 #define FADVISE_HOT_BIT 0x20 689 #define FADVISE_VERITY_BIT 0x40 690 691 #define FADVISE_MODIFIABLE_BITS (FADVISE_COLD_BIT | FADVISE_HOT_BIT) 692 693 #define file_is_cold(inode) is_file(inode, FADVISE_COLD_BIT) 694 #define file_wrong_pino(inode) is_file(inode, FADVISE_LOST_PINO_BIT) 695 #define file_set_cold(inode) set_file(inode, FADVISE_COLD_BIT) 696 #define file_lost_pino(inode) set_file(inode, FADVISE_LOST_PINO_BIT) 697 #define file_clear_cold(inode) clear_file(inode, FADVISE_COLD_BIT) 698 #define file_got_pino(inode) clear_file(inode, FADVISE_LOST_PINO_BIT) 699 #define file_is_encrypt(inode) is_file(inode, FADVISE_ENCRYPT_BIT) 700 #define file_set_encrypt(inode) set_file(inode, FADVISE_ENCRYPT_BIT) 701 #define file_clear_encrypt(inode) clear_file(inode, FADVISE_ENCRYPT_BIT) 702 #define file_enc_name(inode) is_file(inode, FADVISE_ENC_NAME_BIT) 703 #define file_set_enc_name(inode) set_file(inode, FADVISE_ENC_NAME_BIT) 704 #define file_keep_isize(inode) is_file(inode, FADVISE_KEEP_SIZE_BIT) 705 #define file_set_keep_isize(inode) set_file(inode, FADVISE_KEEP_SIZE_BIT) 706 #define file_is_hot(inode) is_file(inode, FADVISE_HOT_BIT) 707 #define file_set_hot(inode) set_file(inode, FADVISE_HOT_BIT) 708 #define file_clear_hot(inode) clear_file(inode, FADVISE_HOT_BIT) 709 #define file_is_verity(inode) is_file(inode, FADVISE_VERITY_BIT) 710 #define file_set_verity(inode) set_file(inode, FADVISE_VERITY_BIT) 711 712 #define DEF_DIR_LEVEL 0 713 714 enum { 715 GC_FAILURE_PIN, 716 GC_FAILURE_ATOMIC, 717 MAX_GC_FAILURE 718 }; 719 720 /* used for f2fs_inode_info->flags */ 721 enum { 722 FI_NEW_INODE, /* indicate newly allocated inode */ 723 FI_DIRTY_INODE, /* indicate inode is dirty or not */ 724 FI_AUTO_RECOVER, /* indicate inode is recoverable */ 725 FI_DIRTY_DIR, /* indicate directory has dirty pages */ 726 FI_INC_LINK, /* need to increment i_nlink */ 727 FI_ACL_MODE, /* indicate acl mode */ 728 FI_NO_ALLOC, /* should not allocate any blocks */ 729 FI_FREE_NID, /* free allocated nide */ 730 FI_NO_EXTENT, /* not to use the extent cache */ 731 FI_INLINE_XATTR, /* used for inline xattr */ 732 FI_INLINE_DATA, /* used for inline data*/ 733 FI_INLINE_DENTRY, /* used for inline dentry */ 734 FI_APPEND_WRITE, /* inode has appended data */ 735 FI_UPDATE_WRITE, /* inode has in-place-update data */ 736 FI_NEED_IPU, /* used for ipu per file */ 737 FI_ATOMIC_FILE, /* indicate atomic file */ 738 FI_ATOMIC_COMMIT, /* indicate the state of atomical committing */ 739 FI_VOLATILE_FILE, /* indicate volatile file */ 740 FI_FIRST_BLOCK_WRITTEN, /* indicate #0 data block was written */ 741 FI_DROP_CACHE, /* drop dirty page cache */ 742 FI_DATA_EXIST, /* indicate data exists */ 743 FI_INLINE_DOTS, /* indicate inline dot dentries */ 744 FI_DO_DEFRAG, /* indicate defragment is running */ 745 FI_DIRTY_FILE, /* indicate regular/symlink has dirty pages */ 746 FI_NO_PREALLOC, /* indicate skipped preallocated blocks */ 747 FI_HOT_DATA, /* indicate file is hot */ 748 FI_EXTRA_ATTR, /* indicate file has extra attribute */ 749 FI_PROJ_INHERIT, /* indicate file inherits projectid */ 750 FI_PIN_FILE, /* indicate file should not be gced */ 751 FI_ATOMIC_REVOKE_REQUEST, /* request to drop atomic data */ 752 FI_VERITY_IN_PROGRESS, /* building fs-verity Merkle tree */ 753 FI_COMPRESSED_FILE, /* indicate file's data can be compressed */ 754 FI_MMAP_FILE, /* indicate file was mmapped */ 755 FI_MAX, /* max flag, never be used */ 756 }; 757 758 struct f2fs_inode_info { 759 struct inode vfs_inode; /* serve a vfs inode */ 760 unsigned long i_flags; /* keep an inode flags for ioctl */ 761 unsigned char i_advise; /* use to give file attribute hints */ 762 unsigned char i_dir_level; /* use for dentry level for large dir */ 763 unsigned int i_current_depth; /* only for directory depth */ 764 /* for gc failure statistic */ 765 unsigned int i_gc_failures[MAX_GC_FAILURE]; 766 unsigned int i_pino; /* parent inode number */ 767 umode_t i_acl_mode; /* keep file acl mode temporarily */ 768 769 /* Use below internally in f2fs*/ 770 unsigned long flags[BITS_TO_LONGS(FI_MAX)]; /* use to pass per-file flags */ 771 struct rw_semaphore i_sem; /* protect fi info */ 772 atomic_t dirty_pages; /* # of dirty pages */ 773 f2fs_hash_t chash; /* hash value of given file name */ 774 unsigned int clevel; /* maximum level of given file name */ 775 struct task_struct *task; /* lookup and create consistency */ 776 struct task_struct *cp_task; /* separate cp/wb IO stats*/ 777 nid_t i_xattr_nid; /* node id that contains xattrs */ 778 loff_t last_disk_size; /* lastly written file size */ 779 spinlock_t i_size_lock; /* protect last_disk_size */ 780 781 #ifdef CONFIG_QUOTA 782 struct dquot *i_dquot[MAXQUOTAS]; 783 784 /* quota space reservation, managed internally by quota code */ 785 qsize_t i_reserved_quota; 786 #endif 787 struct list_head dirty_list; /* dirty list for dirs and files */ 788 struct list_head gdirty_list; /* linked in global dirty list */ 789 struct list_head inmem_ilist; /* list for inmem inodes */ 790 struct list_head inmem_pages; /* inmemory pages managed by f2fs */ 791 struct task_struct *inmem_task; /* store inmemory task */ 792 struct mutex inmem_lock; /* lock for inmemory pages */ 793 struct extent_tree *extent_tree; /* cached extent_tree entry */ 794 795 /* avoid racing between foreground op and gc */ 796 struct rw_semaphore i_gc_rwsem[2]; 797 struct rw_semaphore i_mmap_sem; 798 struct rw_semaphore i_xattr_sem; /* avoid racing between reading and changing EAs */ 799 800 int i_extra_isize; /* size of extra space located in i_addr */ 801 kprojid_t i_projid; /* id for project quota */ 802 int i_inline_xattr_size; /* inline xattr size */ 803 struct timespec64 i_crtime; /* inode creation time */ 804 struct timespec64 i_disk_time[4];/* inode disk times */ 805 806 /* for file compress */ 807 u64 i_compr_blocks; /* # of compressed blocks */ 808 unsigned char i_compress_algorithm; /* algorithm type */ 809 unsigned char i_log_cluster_size; /* log of cluster size */ 810 unsigned int i_cluster_size; /* cluster size */ 811 }; 812 813 static inline void get_extent_info(struct extent_info *ext, 814 struct f2fs_extent *i_ext) 815 { 816 ext->fofs = le32_to_cpu(i_ext->fofs); 817 ext->blk = le32_to_cpu(i_ext->blk); 818 ext->len = le32_to_cpu(i_ext->len); 819 } 820 821 static inline void set_raw_extent(struct extent_info *ext, 822 struct f2fs_extent *i_ext) 823 { 824 i_ext->fofs = cpu_to_le32(ext->fofs); 825 i_ext->blk = cpu_to_le32(ext->blk); 826 i_ext->len = cpu_to_le32(ext->len); 827 } 828 829 static inline void set_extent_info(struct extent_info *ei, unsigned int fofs, 830 u32 blk, unsigned int len) 831 { 832 ei->fofs = fofs; 833 ei->blk = blk; 834 ei->len = len; 835 } 836 837 static inline bool __is_discard_mergeable(struct discard_info *back, 838 struct discard_info *front, unsigned int max_len) 839 { 840 return (back->lstart + back->len == front->lstart) && 841 (back->len + front->len <= max_len); 842 } 843 844 static inline bool __is_discard_back_mergeable(struct discard_info *cur, 845 struct discard_info *back, unsigned int max_len) 846 { 847 return __is_discard_mergeable(back, cur, max_len); 848 } 849 850 static inline bool __is_discard_front_mergeable(struct discard_info *cur, 851 struct discard_info *front, unsigned int max_len) 852 { 853 return __is_discard_mergeable(cur, front, max_len); 854 } 855 856 static inline bool __is_extent_mergeable(struct extent_info *back, 857 struct extent_info *front) 858 { 859 return (back->fofs + back->len == front->fofs && 860 back->blk + back->len == front->blk); 861 } 862 863 static inline bool __is_back_mergeable(struct extent_info *cur, 864 struct extent_info *back) 865 { 866 return __is_extent_mergeable(back, cur); 867 } 868 869 static inline bool __is_front_mergeable(struct extent_info *cur, 870 struct extent_info *front) 871 { 872 return __is_extent_mergeable(cur, front); 873 } 874 875 extern void f2fs_mark_inode_dirty_sync(struct inode *inode, bool sync); 876 static inline void __try_update_largest_extent(struct extent_tree *et, 877 struct extent_node *en) 878 { 879 if (en->ei.len > et->largest.len) { 880 et->largest = en->ei; 881 et->largest_updated = true; 882 } 883 } 884 885 /* 886 * For free nid management 887 */ 888 enum nid_state { 889 FREE_NID, /* newly added to free nid list */ 890 PREALLOC_NID, /* it is preallocated */ 891 MAX_NID_STATE, 892 }; 893 894 struct f2fs_nm_info { 895 block_t nat_blkaddr; /* base disk address of NAT */ 896 nid_t max_nid; /* maximum possible node ids */ 897 nid_t available_nids; /* # of available node ids */ 898 nid_t next_scan_nid; /* the next nid to be scanned */ 899 unsigned int ram_thresh; /* control the memory footprint */ 900 unsigned int ra_nid_pages; /* # of nid pages to be readaheaded */ 901 unsigned int dirty_nats_ratio; /* control dirty nats ratio threshold */ 902 903 /* NAT cache management */ 904 struct radix_tree_root nat_root;/* root of the nat entry cache */ 905 struct radix_tree_root nat_set_root;/* root of the nat set cache */ 906 struct rw_semaphore nat_tree_lock; /* protect nat_tree_lock */ 907 struct list_head nat_entries; /* cached nat entry list (clean) */ 908 spinlock_t nat_list_lock; /* protect clean nat entry list */ 909 unsigned int nat_cnt; /* the # of cached nat entries */ 910 unsigned int dirty_nat_cnt; /* total num of nat entries in set */ 911 unsigned int nat_blocks; /* # of nat blocks */ 912 913 /* free node ids management */ 914 struct radix_tree_root free_nid_root;/* root of the free_nid cache */ 915 struct list_head free_nid_list; /* list for free nids excluding preallocated nids */ 916 unsigned int nid_cnt[MAX_NID_STATE]; /* the number of free node id */ 917 spinlock_t nid_list_lock; /* protect nid lists ops */ 918 struct mutex build_lock; /* lock for build free nids */ 919 unsigned char **free_nid_bitmap; 920 unsigned char *nat_block_bitmap; 921 unsigned short *free_nid_count; /* free nid count of NAT block */ 922 923 /* for checkpoint */ 924 char *nat_bitmap; /* NAT bitmap pointer */ 925 926 unsigned int nat_bits_blocks; /* # of nat bits blocks */ 927 unsigned char *nat_bits; /* NAT bits blocks */ 928 unsigned char *full_nat_bits; /* full NAT pages */ 929 unsigned char *empty_nat_bits; /* empty NAT pages */ 930 #ifdef CONFIG_F2FS_CHECK_FS 931 char *nat_bitmap_mir; /* NAT bitmap mirror */ 932 #endif 933 int bitmap_size; /* bitmap size */ 934 }; 935 936 /* 937 * this structure is used as one of function parameters. 938 * all the information are dedicated to a given direct node block determined 939 * by the data offset in a file. 940 */ 941 struct dnode_of_data { 942 struct inode *inode; /* vfs inode pointer */ 943 struct page *inode_page; /* its inode page, NULL is possible */ 944 struct page *node_page; /* cached direct node page */ 945 nid_t nid; /* node id of the direct node block */ 946 unsigned int ofs_in_node; /* data offset in the node page */ 947 bool inode_page_locked; /* inode page is locked or not */ 948 bool node_changed; /* is node block changed */ 949 char cur_level; /* level of hole node page */ 950 char max_level; /* level of current page located */ 951 block_t data_blkaddr; /* block address of the node block */ 952 }; 953 954 static inline void set_new_dnode(struct dnode_of_data *dn, struct inode *inode, 955 struct page *ipage, struct page *npage, nid_t nid) 956 { 957 memset(dn, 0, sizeof(*dn)); 958 dn->inode = inode; 959 dn->inode_page = ipage; 960 dn->node_page = npage; 961 dn->nid = nid; 962 } 963 964 /* 965 * For SIT manager 966 * 967 * By default, there are 6 active log areas across the whole main area. 968 * When considering hot and cold data separation to reduce cleaning overhead, 969 * we split 3 for data logs and 3 for node logs as hot, warm, and cold types, 970 * respectively. 971 * In the current design, you should not change the numbers intentionally. 972 * Instead, as a mount option such as active_logs=x, you can use 2, 4, and 6 973 * logs individually according to the underlying devices. (default: 6) 974 * Just in case, on-disk layout covers maximum 16 logs that consist of 8 for 975 * data and 8 for node logs. 976 */ 977 #define NR_CURSEG_DATA_TYPE (3) 978 #define NR_CURSEG_NODE_TYPE (3) 979 #define NR_CURSEG_TYPE (NR_CURSEG_DATA_TYPE + NR_CURSEG_NODE_TYPE) 980 981 enum { 982 CURSEG_HOT_DATA = 0, /* directory entry blocks */ 983 CURSEG_WARM_DATA, /* data blocks */ 984 CURSEG_COLD_DATA, /* multimedia or GCed data blocks */ 985 CURSEG_HOT_NODE, /* direct node blocks of directory files */ 986 CURSEG_WARM_NODE, /* direct node blocks of normal files */ 987 CURSEG_COLD_NODE, /* indirect node blocks */ 988 NO_CHECK_TYPE, 989 CURSEG_COLD_DATA_PINNED,/* cold data for pinned file */ 990 }; 991 992 struct flush_cmd { 993 struct completion wait; 994 struct llist_node llnode; 995 nid_t ino; 996 int ret; 997 }; 998 999 struct flush_cmd_control { 1000 struct task_struct *f2fs_issue_flush; /* flush thread */ 1001 wait_queue_head_t flush_wait_queue; /* waiting queue for wake-up */ 1002 atomic_t issued_flush; /* # of issued flushes */ 1003 atomic_t queued_flush; /* # of queued flushes */ 1004 struct llist_head issue_list; /* list for command issue */ 1005 struct llist_node *dispatch_list; /* list for command dispatch */ 1006 }; 1007 1008 struct f2fs_sm_info { 1009 struct sit_info *sit_info; /* whole segment information */ 1010 struct free_segmap_info *free_info; /* free segment information */ 1011 struct dirty_seglist_info *dirty_info; /* dirty segment information */ 1012 struct curseg_info *curseg_array; /* active segment information */ 1013 1014 struct rw_semaphore curseg_lock; /* for preventing curseg change */ 1015 1016 block_t seg0_blkaddr; /* block address of 0'th segment */ 1017 block_t main_blkaddr; /* start block address of main area */ 1018 block_t ssa_blkaddr; /* start block address of SSA area */ 1019 1020 unsigned int segment_count; /* total # of segments */ 1021 unsigned int main_segments; /* # of segments in main area */ 1022 unsigned int reserved_segments; /* # of reserved segments */ 1023 unsigned int ovp_segments; /* # of overprovision segments */ 1024 1025 /* a threshold to reclaim prefree segments */ 1026 unsigned int rec_prefree_segments; 1027 1028 /* for batched trimming */ 1029 unsigned int trim_sections; /* # of sections to trim */ 1030 1031 struct list_head sit_entry_set; /* sit entry set list */ 1032 1033 unsigned int ipu_policy; /* in-place-update policy */ 1034 unsigned int min_ipu_util; /* in-place-update threshold */ 1035 unsigned int min_fsync_blocks; /* threshold for fsync */ 1036 unsigned int min_seq_blocks; /* threshold for sequential blocks */ 1037 unsigned int min_hot_blocks; /* threshold for hot block allocation */ 1038 unsigned int min_ssr_sections; /* threshold to trigger SSR allocation */ 1039 1040 /* for flush command control */ 1041 struct flush_cmd_control *fcc_info; 1042 1043 /* for discard command control */ 1044 struct discard_cmd_control *dcc_info; 1045 }; 1046 1047 /* 1048 * For superblock 1049 */ 1050 /* 1051 * COUNT_TYPE for monitoring 1052 * 1053 * f2fs monitors the number of several block types such as on-writeback, 1054 * dirty dentry blocks, dirty node blocks, and dirty meta blocks. 1055 */ 1056 #define WB_DATA_TYPE(p) (__is_cp_guaranteed(p) ? F2FS_WB_CP_DATA : F2FS_WB_DATA) 1057 enum count_type { 1058 F2FS_DIRTY_DENTS, 1059 F2FS_DIRTY_DATA, 1060 F2FS_DIRTY_QDATA, 1061 F2FS_DIRTY_NODES, 1062 F2FS_DIRTY_META, 1063 F2FS_INMEM_PAGES, 1064 F2FS_DIRTY_IMETA, 1065 F2FS_WB_CP_DATA, 1066 F2FS_WB_DATA, 1067 F2FS_RD_DATA, 1068 F2FS_RD_NODE, 1069 F2FS_RD_META, 1070 F2FS_DIO_WRITE, 1071 F2FS_DIO_READ, 1072 NR_COUNT_TYPE, 1073 }; 1074 1075 /* 1076 * The below are the page types of bios used in submit_bio(). 1077 * The available types are: 1078 * DATA User data pages. It operates as async mode. 1079 * NODE Node pages. It operates as async mode. 1080 * META FS metadata pages such as SIT, NAT, CP. 1081 * NR_PAGE_TYPE The number of page types. 1082 * META_FLUSH Make sure the previous pages are written 1083 * with waiting the bio's completion 1084 * ... Only can be used with META. 1085 */ 1086 #define PAGE_TYPE_OF_BIO(type) ((type) > META ? META : (type)) 1087 enum page_type { 1088 DATA, 1089 NODE, 1090 META, 1091 NR_PAGE_TYPE, 1092 META_FLUSH, 1093 INMEM, /* the below types are used by tracepoints only. */ 1094 INMEM_DROP, 1095 INMEM_INVALIDATE, 1096 INMEM_REVOKE, 1097 IPU, 1098 OPU, 1099 }; 1100 1101 enum temp_type { 1102 HOT = 0, /* must be zero for meta bio */ 1103 WARM, 1104 COLD, 1105 NR_TEMP_TYPE, 1106 }; 1107 1108 enum need_lock_type { 1109 LOCK_REQ = 0, 1110 LOCK_DONE, 1111 LOCK_RETRY, 1112 }; 1113 1114 enum cp_reason_type { 1115 CP_NO_NEEDED, 1116 CP_NON_REGULAR, 1117 CP_COMPRESSED, 1118 CP_HARDLINK, 1119 CP_SB_NEED_CP, 1120 CP_WRONG_PINO, 1121 CP_NO_SPC_ROLL, 1122 CP_NODE_NEED_CP, 1123 CP_FASTBOOT_MODE, 1124 CP_SPEC_LOG_NUM, 1125 CP_RECOVER_DIR, 1126 }; 1127 1128 enum iostat_type { 1129 /* WRITE IO */ 1130 APP_DIRECT_IO, /* app direct write IOs */ 1131 APP_BUFFERED_IO, /* app buffered write IOs */ 1132 APP_WRITE_IO, /* app write IOs */ 1133 APP_MAPPED_IO, /* app mapped IOs */ 1134 FS_DATA_IO, /* data IOs from kworker/fsync/reclaimer */ 1135 FS_NODE_IO, /* node IOs from kworker/fsync/reclaimer */ 1136 FS_META_IO, /* meta IOs from kworker/reclaimer */ 1137 FS_GC_DATA_IO, /* data IOs from forground gc */ 1138 FS_GC_NODE_IO, /* node IOs from forground gc */ 1139 FS_CP_DATA_IO, /* data IOs from checkpoint */ 1140 FS_CP_NODE_IO, /* node IOs from checkpoint */ 1141 FS_CP_META_IO, /* meta IOs from checkpoint */ 1142 1143 /* READ IO */ 1144 APP_DIRECT_READ_IO, /* app direct read IOs */ 1145 APP_BUFFERED_READ_IO, /* app buffered read IOs */ 1146 APP_READ_IO, /* app read IOs */ 1147 APP_MAPPED_READ_IO, /* app mapped read IOs */ 1148 FS_DATA_READ_IO, /* data read IOs */ 1149 FS_NODE_READ_IO, /* node read IOs */ 1150 FS_META_READ_IO, /* meta read IOs */ 1151 1152 /* other */ 1153 FS_DISCARD, /* discard */ 1154 NR_IO_TYPE, 1155 }; 1156 1157 struct f2fs_io_info { 1158 struct f2fs_sb_info *sbi; /* f2fs_sb_info pointer */ 1159 nid_t ino; /* inode number */ 1160 enum page_type type; /* contains DATA/NODE/META/META_FLUSH */ 1161 enum temp_type temp; /* contains HOT/WARM/COLD */ 1162 int op; /* contains REQ_OP_ */ 1163 int op_flags; /* req_flag_bits */ 1164 block_t new_blkaddr; /* new block address to be written */ 1165 block_t old_blkaddr; /* old block address before Cow */ 1166 struct page *page; /* page to be written */ 1167 struct page *encrypted_page; /* encrypted page */ 1168 struct page *compressed_page; /* compressed page */ 1169 struct list_head list; /* serialize IOs */ 1170 bool submitted; /* indicate IO submission */ 1171 int need_lock; /* indicate we need to lock cp_rwsem */ 1172 bool in_list; /* indicate fio is in io_list */ 1173 bool is_por; /* indicate IO is from recovery or not */ 1174 bool retry; /* need to reallocate block address */ 1175 int compr_blocks; /* # of compressed block addresses */ 1176 bool encrypted; /* indicate file is encrypted */ 1177 enum iostat_type io_type; /* io type */ 1178 struct writeback_control *io_wbc; /* writeback control */ 1179 struct bio **bio; /* bio for ipu */ 1180 sector_t *last_block; /* last block number in bio */ 1181 unsigned char version; /* version of the node */ 1182 }; 1183 1184 struct bio_entry { 1185 struct bio *bio; 1186 struct list_head list; 1187 }; 1188 1189 #define is_read_io(rw) ((rw) == READ) 1190 struct f2fs_bio_info { 1191 struct f2fs_sb_info *sbi; /* f2fs superblock */ 1192 struct bio *bio; /* bios to merge */ 1193 sector_t last_block_in_bio; /* last block number */ 1194 struct f2fs_io_info fio; /* store buffered io info. */ 1195 struct rw_semaphore io_rwsem; /* blocking op for bio */ 1196 spinlock_t io_lock; /* serialize DATA/NODE IOs */ 1197 struct list_head io_list; /* track fios */ 1198 struct list_head bio_list; /* bio entry list head */ 1199 struct rw_semaphore bio_list_lock; /* lock to protect bio entry list */ 1200 }; 1201 1202 #define FDEV(i) (sbi->devs[i]) 1203 #define RDEV(i) (raw_super->devs[i]) 1204 struct f2fs_dev_info { 1205 struct block_device *bdev; 1206 char path[MAX_PATH_LEN]; 1207 unsigned int total_segments; 1208 block_t start_blk; 1209 block_t end_blk; 1210 #ifdef CONFIG_BLK_DEV_ZONED 1211 unsigned int nr_blkz; /* Total number of zones */ 1212 unsigned long *blkz_seq; /* Bitmap indicating sequential zones */ 1213 #endif 1214 }; 1215 1216 enum inode_type { 1217 DIR_INODE, /* for dirty dir inode */ 1218 FILE_INODE, /* for dirty regular/symlink inode */ 1219 DIRTY_META, /* for all dirtied inode metadata */ 1220 ATOMIC_FILE, /* for all atomic files */ 1221 NR_INODE_TYPE, 1222 }; 1223 1224 /* for inner inode cache management */ 1225 struct inode_management { 1226 struct radix_tree_root ino_root; /* ino entry array */ 1227 spinlock_t ino_lock; /* for ino entry lock */ 1228 struct list_head ino_list; /* inode list head */ 1229 unsigned long ino_num; /* number of entries */ 1230 }; 1231 1232 /* For s_flag in struct f2fs_sb_info */ 1233 enum { 1234 SBI_IS_DIRTY, /* dirty flag for checkpoint */ 1235 SBI_IS_CLOSE, /* specify unmounting */ 1236 SBI_NEED_FSCK, /* need fsck.f2fs to fix */ 1237 SBI_POR_DOING, /* recovery is doing or not */ 1238 SBI_NEED_SB_WRITE, /* need to recover superblock */ 1239 SBI_NEED_CP, /* need to checkpoint */ 1240 SBI_IS_SHUTDOWN, /* shutdown by ioctl */ 1241 SBI_IS_RECOVERED, /* recovered orphan/data */ 1242 SBI_CP_DISABLED, /* CP was disabled last mount */ 1243 SBI_CP_DISABLED_QUICK, /* CP was disabled quickly */ 1244 SBI_QUOTA_NEED_FLUSH, /* need to flush quota info in CP */ 1245 SBI_QUOTA_SKIP_FLUSH, /* skip flushing quota in current CP */ 1246 SBI_QUOTA_NEED_REPAIR, /* quota file may be corrupted */ 1247 SBI_IS_RESIZEFS, /* resizefs is in process */ 1248 }; 1249 1250 enum { 1251 CP_TIME, 1252 REQ_TIME, 1253 DISCARD_TIME, 1254 GC_TIME, 1255 DISABLE_TIME, 1256 UMOUNT_DISCARD_TIMEOUT, 1257 MAX_TIME, 1258 }; 1259 1260 enum { 1261 GC_NORMAL, 1262 GC_IDLE_CB, 1263 GC_IDLE_GREEDY, 1264 GC_URGENT, 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_FS_ENCRYPTION 1311 #define DUMMY_ENCRYPTION_ENABLED(sbi) \ 1312 (unlikely(F2FS_OPTION(sbi).test_dummy_encryption)) 1313 #else 1314 #define DUMMY_ENCRYPTION_ENABLED(sbi) (0) 1315 #endif 1316 1317 /* For compression */ 1318 enum compress_algorithm_type { 1319 COMPRESS_LZO, 1320 COMPRESS_LZ4, 1321 COMPRESS_ZSTD, 1322 COMPRESS_LZORLE, 1323 COMPRESS_MAX, 1324 }; 1325 1326 #define COMPRESS_DATA_RESERVED_SIZE 5 1327 struct compress_data { 1328 __le32 clen; /* compressed data size */ 1329 __le32 reserved[COMPRESS_DATA_RESERVED_SIZE]; /* reserved */ 1330 u8 cdata[]; /* compressed data */ 1331 }; 1332 1333 #define COMPRESS_HEADER_SIZE (sizeof(struct compress_data)) 1334 1335 #define F2FS_COMPRESSED_PAGE_MAGIC 0xF5F2C000 1336 1337 /* compress context */ 1338 struct compress_ctx { 1339 struct inode *inode; /* inode the context belong to */ 1340 pgoff_t cluster_idx; /* cluster index number */ 1341 unsigned int cluster_size; /* page count in cluster */ 1342 unsigned int log_cluster_size; /* log of cluster size */ 1343 struct page **rpages; /* pages store raw data in cluster */ 1344 unsigned int nr_rpages; /* total page number in rpages */ 1345 struct page **cpages; /* pages store compressed data in cluster */ 1346 unsigned int nr_cpages; /* total page number in cpages */ 1347 void *rbuf; /* virtual mapped address on rpages */ 1348 struct compress_data *cbuf; /* virtual mapped address on cpages */ 1349 size_t rlen; /* valid data length in rbuf */ 1350 size_t clen; /* valid data length in cbuf */ 1351 void *private; /* payload buffer for specified compression algorithm */ 1352 void *private2; /* extra payload buffer */ 1353 }; 1354 1355 /* compress context for write IO path */ 1356 struct compress_io_ctx { 1357 u32 magic; /* magic number to indicate page is compressed */ 1358 struct inode *inode; /* inode the context belong to */ 1359 struct page **rpages; /* pages store raw data in cluster */ 1360 unsigned int nr_rpages; /* total page number in rpages */ 1361 refcount_t ref; /* referrence count of raw page */ 1362 }; 1363 1364 /* decompress io context for read IO path */ 1365 struct decompress_io_ctx { 1366 u32 magic; /* magic number to indicate page is compressed */ 1367 struct inode *inode; /* inode the context belong to */ 1368 pgoff_t cluster_idx; /* cluster index number */ 1369 unsigned int cluster_size; /* page count in cluster */ 1370 unsigned int log_cluster_size; /* log of cluster size */ 1371 struct page **rpages; /* pages store raw data in cluster */ 1372 unsigned int nr_rpages; /* total page number in rpages */ 1373 struct page **cpages; /* pages store compressed data in cluster */ 1374 unsigned int nr_cpages; /* total page number in cpages */ 1375 struct page **tpages; /* temp pages to pad holes in cluster */ 1376 void *rbuf; /* virtual mapped address on rpages */ 1377 struct compress_data *cbuf; /* virtual mapped address on cpages */ 1378 size_t rlen; /* valid data length in rbuf */ 1379 size_t clen; /* valid data length in cbuf */ 1380 refcount_t ref; /* referrence count of compressed page */ 1381 bool failed; /* indicate IO error during decompression */ 1382 void *private; /* payload buffer for specified decompression algorithm */ 1383 void *private2; /* extra payload buffer */ 1384 }; 1385 1386 #define NULL_CLUSTER ((unsigned int)(~0)) 1387 #define MIN_COMPRESS_LOG_SIZE 2 1388 #define MAX_COMPRESS_LOG_SIZE 8 1389 #define MAX_COMPRESS_WINDOW_SIZE ((PAGE_SIZE) << MAX_COMPRESS_LOG_SIZE) 1390 1391 struct f2fs_sb_info { 1392 struct super_block *sb; /* pointer to VFS super block */ 1393 struct proc_dir_entry *s_proc; /* proc entry */ 1394 struct f2fs_super_block *raw_super; /* raw super block pointer */ 1395 struct rw_semaphore sb_lock; /* lock for raw super block */ 1396 int valid_super_block; /* valid super block no */ 1397 unsigned long s_flag; /* flags for sbi */ 1398 struct mutex writepages; /* mutex for writepages() */ 1399 #ifdef CONFIG_UNICODE 1400 struct unicode_map *s_encoding; 1401 __u16 s_encoding_flags; 1402 #endif 1403 1404 #ifdef CONFIG_BLK_DEV_ZONED 1405 unsigned int blocks_per_blkz; /* F2FS blocks per zone */ 1406 unsigned int log_blocks_per_blkz; /* log2 F2FS blocks per zone */ 1407 #endif 1408 1409 /* for node-related operations */ 1410 struct f2fs_nm_info *nm_info; /* node manager */ 1411 struct inode *node_inode; /* cache node blocks */ 1412 1413 /* for segment-related operations */ 1414 struct f2fs_sm_info *sm_info; /* segment manager */ 1415 1416 /* for bio operations */ 1417 struct f2fs_bio_info *write_io[NR_PAGE_TYPE]; /* for write bios */ 1418 /* keep migration IO order for LFS mode */ 1419 struct rw_semaphore io_order_lock; 1420 mempool_t *write_io_dummy; /* Dummy pages */ 1421 1422 /* for checkpoint */ 1423 struct f2fs_checkpoint *ckpt; /* raw checkpoint pointer */ 1424 int cur_cp_pack; /* remain current cp pack */ 1425 spinlock_t cp_lock; /* for flag in ckpt */ 1426 struct inode *meta_inode; /* cache meta blocks */ 1427 struct mutex cp_mutex; /* checkpoint procedure lock */ 1428 struct rw_semaphore cp_rwsem; /* blocking FS operations */ 1429 struct rw_semaphore node_write; /* locking node writes */ 1430 struct rw_semaphore node_change; /* locking node change */ 1431 wait_queue_head_t cp_wait; 1432 unsigned long last_time[MAX_TIME]; /* to store time in jiffies */ 1433 long interval_time[MAX_TIME]; /* to store thresholds */ 1434 1435 struct inode_management im[MAX_INO_ENTRY]; /* manage inode cache */ 1436 1437 spinlock_t fsync_node_lock; /* for node entry lock */ 1438 struct list_head fsync_node_list; /* node list head */ 1439 unsigned int fsync_seg_id; /* sequence id */ 1440 unsigned int fsync_node_num; /* number of node entries */ 1441 1442 /* for orphan inode, use 0'th array */ 1443 unsigned int max_orphans; /* max orphan inodes */ 1444 1445 /* for inode management */ 1446 struct list_head inode_list[NR_INODE_TYPE]; /* dirty inode list */ 1447 spinlock_t inode_lock[NR_INODE_TYPE]; /* for dirty inode list lock */ 1448 struct mutex flush_lock; /* for flush exclusion */ 1449 1450 /* for extent tree cache */ 1451 struct radix_tree_root extent_tree_root;/* cache extent cache entries */ 1452 struct mutex extent_tree_lock; /* locking extent radix tree */ 1453 struct list_head extent_list; /* lru list for shrinker */ 1454 spinlock_t extent_lock; /* locking extent lru list */ 1455 atomic_t total_ext_tree; /* extent tree count */ 1456 struct list_head zombie_list; /* extent zombie tree list */ 1457 atomic_t total_zombie_tree; /* extent zombie tree count */ 1458 atomic_t total_ext_node; /* extent info count */ 1459 1460 /* basic filesystem units */ 1461 unsigned int log_sectors_per_block; /* log2 sectors per block */ 1462 unsigned int log_blocksize; /* log2 block size */ 1463 unsigned int blocksize; /* block size */ 1464 unsigned int root_ino_num; /* root inode number*/ 1465 unsigned int node_ino_num; /* node inode number*/ 1466 unsigned int meta_ino_num; /* meta inode number*/ 1467 unsigned int log_blocks_per_seg; /* log2 blocks per segment */ 1468 unsigned int blocks_per_seg; /* blocks per segment */ 1469 unsigned int segs_per_sec; /* segments per section */ 1470 unsigned int secs_per_zone; /* sections per zone */ 1471 unsigned int total_sections; /* total section count */ 1472 struct mutex resize_mutex; /* for resize exclusion */ 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 /* for skip statistic */ 1515 unsigned int atomic_files; /* # of opened atomic file */ 1516 unsigned long long skipped_atomic_files[2]; /* FG_GC and BG_GC */ 1517 unsigned long long skipped_gc_rwsem; /* FG_GC only */ 1518 1519 /* threshold for gc trials on pinned files */ 1520 u64 gc_pin_file_threshold; 1521 struct rw_semaphore pin_sem; 1522 1523 /* maximum # of trials to find a victim segment for SSR and GC */ 1524 unsigned int max_victim_search; 1525 /* migration granularity of garbage collection, unit: segment */ 1526 unsigned int migration_granularity; 1527 1528 /* 1529 * for stat information. 1530 * one is for the LFS mode, and the other is for the SSR mode. 1531 */ 1532 #ifdef CONFIG_F2FS_STAT_FS 1533 struct f2fs_stat_info *stat_info; /* FS status information */ 1534 atomic_t meta_count[META_MAX]; /* # of meta blocks */ 1535 unsigned int segment_count[2]; /* # of allocated segments */ 1536 unsigned int block_count[2]; /* # of allocated blocks */ 1537 atomic_t inplace_count; /* # of inplace update */ 1538 atomic64_t total_hit_ext; /* # of lookup extent cache */ 1539 atomic64_t read_hit_rbtree; /* # of hit rbtree extent node */ 1540 atomic64_t read_hit_largest; /* # of hit largest extent node */ 1541 atomic64_t read_hit_cached; /* # of hit cached extent node */ 1542 atomic_t inline_xattr; /* # of inline_xattr inodes */ 1543 atomic_t inline_inode; /* # of inline_data inodes */ 1544 atomic_t inline_dir; /* # of inline_dentry inodes */ 1545 atomic_t compr_inode; /* # of compressed inodes */ 1546 atomic_t compr_blocks; /* # of compressed blocks */ 1547 atomic_t vw_cnt; /* # of volatile writes */ 1548 atomic_t max_aw_cnt; /* max # of atomic writes */ 1549 atomic_t max_vw_cnt; /* max # of volatile writes */ 1550 unsigned int io_skip_bggc; /* skip background gc for in-flight IO */ 1551 unsigned int other_skip_bggc; /* skip background gc for other reasons */ 1552 unsigned int ndirty_inode[NR_INODE_TYPE]; /* # of dirty inodes */ 1553 #endif 1554 spinlock_t stat_lock; /* lock for stat operations */ 1555 1556 /* For app/fs IO statistics */ 1557 spinlock_t iostat_lock; 1558 unsigned long long rw_iostat[NR_IO_TYPE]; 1559 unsigned long long prev_rw_iostat[NR_IO_TYPE]; 1560 bool iostat_enable; 1561 unsigned long iostat_next_period; 1562 unsigned int iostat_period_ms; 1563 1564 /* to attach REQ_META|REQ_FUA flags */ 1565 unsigned int data_io_flag; 1566 1567 /* For sysfs suppport */ 1568 struct kobject s_kobj; 1569 struct completion s_kobj_unregister; 1570 1571 /* For shrinker support */ 1572 struct list_head s_list; 1573 int s_ndevs; /* number of devices */ 1574 struct f2fs_dev_info *devs; /* for device list */ 1575 unsigned int dirty_device; /* for checkpoint data flush */ 1576 spinlock_t dev_lock; /* protect dirty_device */ 1577 struct mutex umount_mutex; 1578 unsigned int shrinker_run_no; 1579 1580 /* For write statistics */ 1581 u64 sectors_written_start; 1582 u64 kbytes_written; 1583 1584 /* Reference to checksum algorithm driver via cryptoapi */ 1585 struct crypto_shash *s_chksum_driver; 1586 1587 /* Precomputed FS UUID checksum for seeding other checksums */ 1588 __u32 s_chksum_seed; 1589 1590 struct workqueue_struct *post_read_wq; /* post read workqueue */ 1591 1592 struct kmem_cache *inline_xattr_slab; /* inline xattr entry */ 1593 unsigned int inline_xattr_slab_size; /* default inline xattr slab size */ 1594 }; 1595 1596 struct f2fs_private_dio { 1597 struct inode *inode; 1598 void *orig_private; 1599 bio_end_io_t *orig_end_io; 1600 bool write; 1601 }; 1602 1603 #ifdef CONFIG_F2FS_FAULT_INJECTION 1604 #define f2fs_show_injection_info(sbi, type) \ 1605 printk_ratelimited("%sF2FS-fs (%s) : inject %s in %s of %pS\n", \ 1606 KERN_INFO, sbi->sb->s_id, \ 1607 f2fs_fault_name[type], \ 1608 __func__, __builtin_return_address(0)) 1609 static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type) 1610 { 1611 struct f2fs_fault_info *ffi = &F2FS_OPTION(sbi).fault_info; 1612 1613 if (!ffi->inject_rate) 1614 return false; 1615 1616 if (!IS_FAULT_SET(ffi, type)) 1617 return false; 1618 1619 atomic_inc(&ffi->inject_ops); 1620 if (atomic_read(&ffi->inject_ops) >= ffi->inject_rate) { 1621 atomic_set(&ffi->inject_ops, 0); 1622 return true; 1623 } 1624 return false; 1625 } 1626 #else 1627 #define f2fs_show_injection_info(sbi, type) do { } while (0) 1628 static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type) 1629 { 1630 return false; 1631 } 1632 #endif 1633 1634 /* 1635 * Test if the mounted volume is a multi-device volume. 1636 * - For a single regular disk volume, sbi->s_ndevs is 0. 1637 * - For a single zoned disk volume, sbi->s_ndevs is 1. 1638 * - For a multi-device volume, sbi->s_ndevs is always 2 or more. 1639 */ 1640 static inline bool f2fs_is_multi_device(struct f2fs_sb_info *sbi) 1641 { 1642 return sbi->s_ndevs > 1; 1643 } 1644 1645 /* For write statistics. Suppose sector size is 512 bytes, 1646 * and the return value is in kbytes. s is of struct f2fs_sb_info. 1647 */ 1648 #define BD_PART_WRITTEN(s) \ 1649 (((u64)part_stat_read((s)->sb->s_bdev->bd_part, sectors[STAT_WRITE]) - \ 1650 (s)->sectors_written_start) >> 1) 1651 1652 static inline void f2fs_update_time(struct f2fs_sb_info *sbi, int type) 1653 { 1654 unsigned long now = jiffies; 1655 1656 sbi->last_time[type] = now; 1657 1658 /* DISCARD_TIME and GC_TIME are based on REQ_TIME */ 1659 if (type == REQ_TIME) { 1660 sbi->last_time[DISCARD_TIME] = now; 1661 sbi->last_time[GC_TIME] = now; 1662 } 1663 } 1664 1665 static inline bool f2fs_time_over(struct f2fs_sb_info *sbi, int type) 1666 { 1667 unsigned long interval = sbi->interval_time[type] * HZ; 1668 1669 return time_after(jiffies, sbi->last_time[type] + interval); 1670 } 1671 1672 static inline unsigned int f2fs_time_to_wait(struct f2fs_sb_info *sbi, 1673 int type) 1674 { 1675 unsigned long interval = sbi->interval_time[type] * HZ; 1676 unsigned int wait_ms = 0; 1677 long delta; 1678 1679 delta = (sbi->last_time[type] + interval) - jiffies; 1680 if (delta > 0) 1681 wait_ms = jiffies_to_msecs(delta); 1682 1683 return wait_ms; 1684 } 1685 1686 /* 1687 * Inline functions 1688 */ 1689 static inline u32 __f2fs_crc32(struct f2fs_sb_info *sbi, u32 crc, 1690 const void *address, unsigned int length) 1691 { 1692 struct { 1693 struct shash_desc shash; 1694 char ctx[4]; 1695 } desc; 1696 int err; 1697 1698 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver) != sizeof(desc.ctx)); 1699 1700 desc.shash.tfm = sbi->s_chksum_driver; 1701 *(u32 *)desc.ctx = crc; 1702 1703 err = crypto_shash_update(&desc.shash, address, length); 1704 BUG_ON(err); 1705 1706 return *(u32 *)desc.ctx; 1707 } 1708 1709 static inline u32 f2fs_crc32(struct f2fs_sb_info *sbi, const void *address, 1710 unsigned int length) 1711 { 1712 return __f2fs_crc32(sbi, F2FS_SUPER_MAGIC, address, length); 1713 } 1714 1715 static inline bool f2fs_crc_valid(struct f2fs_sb_info *sbi, __u32 blk_crc, 1716 void *buf, size_t buf_size) 1717 { 1718 return f2fs_crc32(sbi, buf, buf_size) == blk_crc; 1719 } 1720 1721 static inline u32 f2fs_chksum(struct f2fs_sb_info *sbi, u32 crc, 1722 const void *address, unsigned int length) 1723 { 1724 return __f2fs_crc32(sbi, crc, address, length); 1725 } 1726 1727 static inline struct f2fs_inode_info *F2FS_I(struct inode *inode) 1728 { 1729 return container_of(inode, struct f2fs_inode_info, vfs_inode); 1730 } 1731 1732 static inline struct f2fs_sb_info *F2FS_SB(struct super_block *sb) 1733 { 1734 return sb->s_fs_info; 1735 } 1736 1737 static inline struct f2fs_sb_info *F2FS_I_SB(struct inode *inode) 1738 { 1739 return F2FS_SB(inode->i_sb); 1740 } 1741 1742 static inline struct f2fs_sb_info *F2FS_M_SB(struct address_space *mapping) 1743 { 1744 return F2FS_I_SB(mapping->host); 1745 } 1746 1747 static inline struct f2fs_sb_info *F2FS_P_SB(struct page *page) 1748 { 1749 return F2FS_M_SB(page_file_mapping(page)); 1750 } 1751 1752 static inline struct f2fs_super_block *F2FS_RAW_SUPER(struct f2fs_sb_info *sbi) 1753 { 1754 return (struct f2fs_super_block *)(sbi->raw_super); 1755 } 1756 1757 static inline struct f2fs_checkpoint *F2FS_CKPT(struct f2fs_sb_info *sbi) 1758 { 1759 return (struct f2fs_checkpoint *)(sbi->ckpt); 1760 } 1761 1762 static inline struct f2fs_node *F2FS_NODE(struct page *page) 1763 { 1764 return (struct f2fs_node *)page_address(page); 1765 } 1766 1767 static inline struct f2fs_inode *F2FS_INODE(struct page *page) 1768 { 1769 return &((struct f2fs_node *)page_address(page))->i; 1770 } 1771 1772 static inline struct f2fs_nm_info *NM_I(struct f2fs_sb_info *sbi) 1773 { 1774 return (struct f2fs_nm_info *)(sbi->nm_info); 1775 } 1776 1777 static inline struct f2fs_sm_info *SM_I(struct f2fs_sb_info *sbi) 1778 { 1779 return (struct f2fs_sm_info *)(sbi->sm_info); 1780 } 1781 1782 static inline struct sit_info *SIT_I(struct f2fs_sb_info *sbi) 1783 { 1784 return (struct sit_info *)(SM_I(sbi)->sit_info); 1785 } 1786 1787 static inline struct free_segmap_info *FREE_I(struct f2fs_sb_info *sbi) 1788 { 1789 return (struct free_segmap_info *)(SM_I(sbi)->free_info); 1790 } 1791 1792 static inline struct dirty_seglist_info *DIRTY_I(struct f2fs_sb_info *sbi) 1793 { 1794 return (struct dirty_seglist_info *)(SM_I(sbi)->dirty_info); 1795 } 1796 1797 static inline struct address_space *META_MAPPING(struct f2fs_sb_info *sbi) 1798 { 1799 return sbi->meta_inode->i_mapping; 1800 } 1801 1802 static inline struct address_space *NODE_MAPPING(struct f2fs_sb_info *sbi) 1803 { 1804 return sbi->node_inode->i_mapping; 1805 } 1806 1807 static inline bool is_sbi_flag_set(struct f2fs_sb_info *sbi, unsigned int type) 1808 { 1809 return test_bit(type, &sbi->s_flag); 1810 } 1811 1812 static inline void set_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type) 1813 { 1814 set_bit(type, &sbi->s_flag); 1815 } 1816 1817 static inline void clear_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type) 1818 { 1819 clear_bit(type, &sbi->s_flag); 1820 } 1821 1822 static inline unsigned long long cur_cp_version(struct f2fs_checkpoint *cp) 1823 { 1824 return le64_to_cpu(cp->checkpoint_ver); 1825 } 1826 1827 static inline unsigned long f2fs_qf_ino(struct super_block *sb, int type) 1828 { 1829 if (type < F2FS_MAX_QUOTAS) 1830 return le32_to_cpu(F2FS_SB(sb)->raw_super->qf_ino[type]); 1831 return 0; 1832 } 1833 1834 static inline __u64 cur_cp_crc(struct f2fs_checkpoint *cp) 1835 { 1836 size_t crc_offset = le32_to_cpu(cp->checksum_offset); 1837 return le32_to_cpu(*((__le32 *)((unsigned char *)cp + crc_offset))); 1838 } 1839 1840 static inline bool __is_set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 1841 { 1842 unsigned int ckpt_flags = le32_to_cpu(cp->ckpt_flags); 1843 1844 return ckpt_flags & f; 1845 } 1846 1847 static inline bool is_set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 1848 { 1849 return __is_set_ckpt_flags(F2FS_CKPT(sbi), f); 1850 } 1851 1852 static inline void __set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 1853 { 1854 unsigned int ckpt_flags; 1855 1856 ckpt_flags = le32_to_cpu(cp->ckpt_flags); 1857 ckpt_flags |= f; 1858 cp->ckpt_flags = cpu_to_le32(ckpt_flags); 1859 } 1860 1861 static inline void set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 1862 { 1863 unsigned long flags; 1864 1865 spin_lock_irqsave(&sbi->cp_lock, flags); 1866 __set_ckpt_flags(F2FS_CKPT(sbi), f); 1867 spin_unlock_irqrestore(&sbi->cp_lock, flags); 1868 } 1869 1870 static inline void __clear_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 1871 { 1872 unsigned int ckpt_flags; 1873 1874 ckpt_flags = le32_to_cpu(cp->ckpt_flags); 1875 ckpt_flags &= (~f); 1876 cp->ckpt_flags = cpu_to_le32(ckpt_flags); 1877 } 1878 1879 static inline void clear_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 1880 { 1881 unsigned long flags; 1882 1883 spin_lock_irqsave(&sbi->cp_lock, flags); 1884 __clear_ckpt_flags(F2FS_CKPT(sbi), f); 1885 spin_unlock_irqrestore(&sbi->cp_lock, flags); 1886 } 1887 1888 static inline void disable_nat_bits(struct f2fs_sb_info *sbi, bool lock) 1889 { 1890 unsigned long flags; 1891 unsigned char *nat_bits; 1892 1893 /* 1894 * In order to re-enable nat_bits we need to call fsck.f2fs by 1895 * set_sbi_flag(sbi, SBI_NEED_FSCK). But it may give huge cost, 1896 * so let's rely on regular fsck or unclean shutdown. 1897 */ 1898 1899 if (lock) 1900 spin_lock_irqsave(&sbi->cp_lock, flags); 1901 __clear_ckpt_flags(F2FS_CKPT(sbi), CP_NAT_BITS_FLAG); 1902 nat_bits = NM_I(sbi)->nat_bits; 1903 NM_I(sbi)->nat_bits = NULL; 1904 if (lock) 1905 spin_unlock_irqrestore(&sbi->cp_lock, flags); 1906 1907 kvfree(nat_bits); 1908 } 1909 1910 static inline bool enabled_nat_bits(struct f2fs_sb_info *sbi, 1911 struct cp_control *cpc) 1912 { 1913 bool set = is_set_ckpt_flags(sbi, CP_NAT_BITS_FLAG); 1914 1915 return (cpc) ? (cpc->reason & CP_UMOUNT) && set : set; 1916 } 1917 1918 static inline void f2fs_lock_op(struct f2fs_sb_info *sbi) 1919 { 1920 down_read(&sbi->cp_rwsem); 1921 } 1922 1923 static inline int f2fs_trylock_op(struct f2fs_sb_info *sbi) 1924 { 1925 return down_read_trylock(&sbi->cp_rwsem); 1926 } 1927 1928 static inline void f2fs_unlock_op(struct f2fs_sb_info *sbi) 1929 { 1930 up_read(&sbi->cp_rwsem); 1931 } 1932 1933 static inline void f2fs_lock_all(struct f2fs_sb_info *sbi) 1934 { 1935 down_write(&sbi->cp_rwsem); 1936 } 1937 1938 static inline void f2fs_unlock_all(struct f2fs_sb_info *sbi) 1939 { 1940 up_write(&sbi->cp_rwsem); 1941 } 1942 1943 static inline int __get_cp_reason(struct f2fs_sb_info *sbi) 1944 { 1945 int reason = CP_SYNC; 1946 1947 if (test_opt(sbi, FASTBOOT)) 1948 reason = CP_FASTBOOT; 1949 if (is_sbi_flag_set(sbi, SBI_IS_CLOSE)) 1950 reason = CP_UMOUNT; 1951 return reason; 1952 } 1953 1954 static inline bool __remain_node_summaries(int reason) 1955 { 1956 return (reason & (CP_UMOUNT | CP_FASTBOOT)); 1957 } 1958 1959 static inline bool __exist_node_summaries(struct f2fs_sb_info *sbi) 1960 { 1961 return (is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG) || 1962 is_set_ckpt_flags(sbi, CP_FASTBOOT_FLAG)); 1963 } 1964 1965 /* 1966 * Check whether the inode has blocks or not 1967 */ 1968 static inline int F2FS_HAS_BLOCKS(struct inode *inode) 1969 { 1970 block_t xattr_block = F2FS_I(inode)->i_xattr_nid ? 1 : 0; 1971 1972 return (inode->i_blocks >> F2FS_LOG_SECTORS_PER_BLOCK) > xattr_block; 1973 } 1974 1975 static inline bool f2fs_has_xattr_block(unsigned int ofs) 1976 { 1977 return ofs == XATTR_NODE_OFFSET; 1978 } 1979 1980 static inline bool __allow_reserved_blocks(struct f2fs_sb_info *sbi, 1981 struct inode *inode, bool cap) 1982 { 1983 if (!inode) 1984 return true; 1985 if (!test_opt(sbi, RESERVE_ROOT)) 1986 return false; 1987 if (IS_NOQUOTA(inode)) 1988 return true; 1989 if (uid_eq(F2FS_OPTION(sbi).s_resuid, current_fsuid())) 1990 return true; 1991 if (!gid_eq(F2FS_OPTION(sbi).s_resgid, GLOBAL_ROOT_GID) && 1992 in_group_p(F2FS_OPTION(sbi).s_resgid)) 1993 return true; 1994 if (cap && capable(CAP_SYS_RESOURCE)) 1995 return true; 1996 return false; 1997 } 1998 1999 static inline void f2fs_i_blocks_write(struct inode *, block_t, bool, bool); 2000 static inline int inc_valid_block_count(struct f2fs_sb_info *sbi, 2001 struct inode *inode, blkcnt_t *count) 2002 { 2003 blkcnt_t diff = 0, release = 0; 2004 block_t avail_user_block_count; 2005 int ret; 2006 2007 ret = dquot_reserve_block(inode, *count); 2008 if (ret) 2009 return ret; 2010 2011 if (time_to_inject(sbi, FAULT_BLOCK)) { 2012 f2fs_show_injection_info(sbi, FAULT_BLOCK); 2013 release = *count; 2014 goto release_quota; 2015 } 2016 2017 /* 2018 * let's increase this in prior to actual block count change in order 2019 * for f2fs_sync_file to avoid data races when deciding checkpoint. 2020 */ 2021 percpu_counter_add(&sbi->alloc_valid_block_count, (*count)); 2022 2023 spin_lock(&sbi->stat_lock); 2024 sbi->total_valid_block_count += (block_t)(*count); 2025 avail_user_block_count = sbi->user_block_count - 2026 sbi->current_reserved_blocks; 2027 2028 if (!__allow_reserved_blocks(sbi, inode, true)) 2029 avail_user_block_count -= F2FS_OPTION(sbi).root_reserved_blocks; 2030 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) { 2031 if (avail_user_block_count > sbi->unusable_block_count) 2032 avail_user_block_count -= sbi->unusable_block_count; 2033 else 2034 avail_user_block_count = 0; 2035 } 2036 if (unlikely(sbi->total_valid_block_count > avail_user_block_count)) { 2037 diff = sbi->total_valid_block_count - avail_user_block_count; 2038 if (diff > *count) 2039 diff = *count; 2040 *count -= diff; 2041 release = diff; 2042 sbi->total_valid_block_count -= diff; 2043 if (!*count) { 2044 spin_unlock(&sbi->stat_lock); 2045 goto enospc; 2046 } 2047 } 2048 spin_unlock(&sbi->stat_lock); 2049 2050 if (unlikely(release)) { 2051 percpu_counter_sub(&sbi->alloc_valid_block_count, release); 2052 dquot_release_reservation_block(inode, release); 2053 } 2054 f2fs_i_blocks_write(inode, *count, true, true); 2055 return 0; 2056 2057 enospc: 2058 percpu_counter_sub(&sbi->alloc_valid_block_count, release); 2059 release_quota: 2060 dquot_release_reservation_block(inode, release); 2061 return -ENOSPC; 2062 } 2063 2064 __printf(2, 3) 2065 void f2fs_printk(struct f2fs_sb_info *sbi, const char *fmt, ...); 2066 2067 #define f2fs_err(sbi, fmt, ...) \ 2068 f2fs_printk(sbi, KERN_ERR fmt, ##__VA_ARGS__) 2069 #define f2fs_warn(sbi, fmt, ...) \ 2070 f2fs_printk(sbi, KERN_WARNING fmt, ##__VA_ARGS__) 2071 #define f2fs_notice(sbi, fmt, ...) \ 2072 f2fs_printk(sbi, KERN_NOTICE fmt, ##__VA_ARGS__) 2073 #define f2fs_info(sbi, fmt, ...) \ 2074 f2fs_printk(sbi, KERN_INFO fmt, ##__VA_ARGS__) 2075 #define f2fs_debug(sbi, fmt, ...) \ 2076 f2fs_printk(sbi, KERN_DEBUG fmt, ##__VA_ARGS__) 2077 2078 static inline void dec_valid_block_count(struct f2fs_sb_info *sbi, 2079 struct inode *inode, 2080 block_t count) 2081 { 2082 blkcnt_t sectors = count << F2FS_LOG_SECTORS_PER_BLOCK; 2083 2084 spin_lock(&sbi->stat_lock); 2085 f2fs_bug_on(sbi, sbi->total_valid_block_count < (block_t) count); 2086 sbi->total_valid_block_count -= (block_t)count; 2087 if (sbi->reserved_blocks && 2088 sbi->current_reserved_blocks < sbi->reserved_blocks) 2089 sbi->current_reserved_blocks = min(sbi->reserved_blocks, 2090 sbi->current_reserved_blocks + count); 2091 spin_unlock(&sbi->stat_lock); 2092 if (unlikely(inode->i_blocks < sectors)) { 2093 f2fs_warn(sbi, "Inconsistent i_blocks, ino:%lu, iblocks:%llu, sectors:%llu", 2094 inode->i_ino, 2095 (unsigned long long)inode->i_blocks, 2096 (unsigned long long)sectors); 2097 set_sbi_flag(sbi, SBI_NEED_FSCK); 2098 return; 2099 } 2100 f2fs_i_blocks_write(inode, count, false, true); 2101 } 2102 2103 static inline void inc_page_count(struct f2fs_sb_info *sbi, int count_type) 2104 { 2105 atomic_inc(&sbi->nr_pages[count_type]); 2106 2107 if (count_type == F2FS_DIRTY_DENTS || 2108 count_type == F2FS_DIRTY_NODES || 2109 count_type == F2FS_DIRTY_META || 2110 count_type == F2FS_DIRTY_QDATA || 2111 count_type == F2FS_DIRTY_IMETA) 2112 set_sbi_flag(sbi, SBI_IS_DIRTY); 2113 } 2114 2115 static inline void inode_inc_dirty_pages(struct inode *inode) 2116 { 2117 atomic_inc(&F2FS_I(inode)->dirty_pages); 2118 inc_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ? 2119 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA); 2120 if (IS_NOQUOTA(inode)) 2121 inc_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA); 2122 } 2123 2124 static inline void dec_page_count(struct f2fs_sb_info *sbi, int count_type) 2125 { 2126 atomic_dec(&sbi->nr_pages[count_type]); 2127 } 2128 2129 static inline void inode_dec_dirty_pages(struct inode *inode) 2130 { 2131 if (!S_ISDIR(inode->i_mode) && !S_ISREG(inode->i_mode) && 2132 !S_ISLNK(inode->i_mode)) 2133 return; 2134 2135 atomic_dec(&F2FS_I(inode)->dirty_pages); 2136 dec_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ? 2137 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA); 2138 if (IS_NOQUOTA(inode)) 2139 dec_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA); 2140 } 2141 2142 static inline s64 get_pages(struct f2fs_sb_info *sbi, int count_type) 2143 { 2144 return atomic_read(&sbi->nr_pages[count_type]); 2145 } 2146 2147 static inline int get_dirty_pages(struct inode *inode) 2148 { 2149 return atomic_read(&F2FS_I(inode)->dirty_pages); 2150 } 2151 2152 static inline int get_blocktype_secs(struct f2fs_sb_info *sbi, int block_type) 2153 { 2154 unsigned int pages_per_sec = sbi->segs_per_sec * sbi->blocks_per_seg; 2155 unsigned int segs = (get_pages(sbi, block_type) + pages_per_sec - 1) >> 2156 sbi->log_blocks_per_seg; 2157 2158 return segs / sbi->segs_per_sec; 2159 } 2160 2161 static inline block_t valid_user_blocks(struct f2fs_sb_info *sbi) 2162 { 2163 return sbi->total_valid_block_count; 2164 } 2165 2166 static inline block_t discard_blocks(struct f2fs_sb_info *sbi) 2167 { 2168 return sbi->discard_blks; 2169 } 2170 2171 static inline unsigned long __bitmap_size(struct f2fs_sb_info *sbi, int flag) 2172 { 2173 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi); 2174 2175 /* return NAT or SIT bitmap */ 2176 if (flag == NAT_BITMAP) 2177 return le32_to_cpu(ckpt->nat_ver_bitmap_bytesize); 2178 else if (flag == SIT_BITMAP) 2179 return le32_to_cpu(ckpt->sit_ver_bitmap_bytesize); 2180 2181 return 0; 2182 } 2183 2184 static inline block_t __cp_payload(struct f2fs_sb_info *sbi) 2185 { 2186 return le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_payload); 2187 } 2188 2189 static inline void *__bitmap_ptr(struct f2fs_sb_info *sbi, int flag) 2190 { 2191 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi); 2192 int offset; 2193 2194 if (is_set_ckpt_flags(sbi, CP_LARGE_NAT_BITMAP_FLAG)) { 2195 offset = (flag == SIT_BITMAP) ? 2196 le32_to_cpu(ckpt->nat_ver_bitmap_bytesize) : 0; 2197 /* 2198 * if large_nat_bitmap feature is enabled, leave checksum 2199 * protection for all nat/sit bitmaps. 2200 */ 2201 return &ckpt->sit_nat_version_bitmap + offset + sizeof(__le32); 2202 } 2203 2204 if (__cp_payload(sbi) > 0) { 2205 if (flag == NAT_BITMAP) 2206 return &ckpt->sit_nat_version_bitmap; 2207 else 2208 return (unsigned char *)ckpt + F2FS_BLKSIZE; 2209 } else { 2210 offset = (flag == NAT_BITMAP) ? 2211 le32_to_cpu(ckpt->sit_ver_bitmap_bytesize) : 0; 2212 return &ckpt->sit_nat_version_bitmap + offset; 2213 } 2214 } 2215 2216 static inline block_t __start_cp_addr(struct f2fs_sb_info *sbi) 2217 { 2218 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr); 2219 2220 if (sbi->cur_cp_pack == 2) 2221 start_addr += sbi->blocks_per_seg; 2222 return start_addr; 2223 } 2224 2225 static inline block_t __start_cp_next_addr(struct f2fs_sb_info *sbi) 2226 { 2227 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr); 2228 2229 if (sbi->cur_cp_pack == 1) 2230 start_addr += sbi->blocks_per_seg; 2231 return start_addr; 2232 } 2233 2234 static inline void __set_cp_next_pack(struct f2fs_sb_info *sbi) 2235 { 2236 sbi->cur_cp_pack = (sbi->cur_cp_pack == 1) ? 2 : 1; 2237 } 2238 2239 static inline block_t __start_sum_addr(struct f2fs_sb_info *sbi) 2240 { 2241 return le32_to_cpu(F2FS_CKPT(sbi)->cp_pack_start_sum); 2242 } 2243 2244 static inline int inc_valid_node_count(struct f2fs_sb_info *sbi, 2245 struct inode *inode, bool is_inode) 2246 { 2247 block_t valid_block_count; 2248 unsigned int valid_node_count, user_block_count; 2249 int err; 2250 2251 if (is_inode) { 2252 if (inode) { 2253 err = dquot_alloc_inode(inode); 2254 if (err) 2255 return err; 2256 } 2257 } else { 2258 err = dquot_reserve_block(inode, 1); 2259 if (err) 2260 return err; 2261 } 2262 2263 if (time_to_inject(sbi, FAULT_BLOCK)) { 2264 f2fs_show_injection_info(sbi, FAULT_BLOCK); 2265 goto enospc; 2266 } 2267 2268 spin_lock(&sbi->stat_lock); 2269 2270 valid_block_count = sbi->total_valid_block_count + 2271 sbi->current_reserved_blocks + 1; 2272 2273 if (!__allow_reserved_blocks(sbi, inode, false)) 2274 valid_block_count += F2FS_OPTION(sbi).root_reserved_blocks; 2275 user_block_count = sbi->user_block_count; 2276 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) 2277 user_block_count -= sbi->unusable_block_count; 2278 2279 if (unlikely(valid_block_count > user_block_count)) { 2280 spin_unlock(&sbi->stat_lock); 2281 goto enospc; 2282 } 2283 2284 valid_node_count = sbi->total_valid_node_count + 1; 2285 if (unlikely(valid_node_count > sbi->total_node_count)) { 2286 spin_unlock(&sbi->stat_lock); 2287 goto enospc; 2288 } 2289 2290 sbi->total_valid_node_count++; 2291 sbi->total_valid_block_count++; 2292 spin_unlock(&sbi->stat_lock); 2293 2294 if (inode) { 2295 if (is_inode) 2296 f2fs_mark_inode_dirty_sync(inode, true); 2297 else 2298 f2fs_i_blocks_write(inode, 1, true, true); 2299 } 2300 2301 percpu_counter_inc(&sbi->alloc_valid_block_count); 2302 return 0; 2303 2304 enospc: 2305 if (is_inode) { 2306 if (inode) 2307 dquot_free_inode(inode); 2308 } else { 2309 dquot_release_reservation_block(inode, 1); 2310 } 2311 return -ENOSPC; 2312 } 2313 2314 static inline void dec_valid_node_count(struct f2fs_sb_info *sbi, 2315 struct inode *inode, bool is_inode) 2316 { 2317 spin_lock(&sbi->stat_lock); 2318 2319 f2fs_bug_on(sbi, !sbi->total_valid_block_count); 2320 f2fs_bug_on(sbi, !sbi->total_valid_node_count); 2321 2322 sbi->total_valid_node_count--; 2323 sbi->total_valid_block_count--; 2324 if (sbi->reserved_blocks && 2325 sbi->current_reserved_blocks < sbi->reserved_blocks) 2326 sbi->current_reserved_blocks++; 2327 2328 spin_unlock(&sbi->stat_lock); 2329 2330 if (is_inode) { 2331 dquot_free_inode(inode); 2332 } else { 2333 if (unlikely(inode->i_blocks == 0)) { 2334 f2fs_warn(sbi, "dec_valid_node_count: inconsistent i_blocks, ino:%lu, iblocks:%llu", 2335 inode->i_ino, 2336 (unsigned long long)inode->i_blocks); 2337 set_sbi_flag(sbi, SBI_NEED_FSCK); 2338 return; 2339 } 2340 f2fs_i_blocks_write(inode, 1, false, true); 2341 } 2342 } 2343 2344 static inline unsigned int valid_node_count(struct f2fs_sb_info *sbi) 2345 { 2346 return sbi->total_valid_node_count; 2347 } 2348 2349 static inline void inc_valid_inode_count(struct f2fs_sb_info *sbi) 2350 { 2351 percpu_counter_inc(&sbi->total_valid_inode_count); 2352 } 2353 2354 static inline void dec_valid_inode_count(struct f2fs_sb_info *sbi) 2355 { 2356 percpu_counter_dec(&sbi->total_valid_inode_count); 2357 } 2358 2359 static inline s64 valid_inode_count(struct f2fs_sb_info *sbi) 2360 { 2361 return percpu_counter_sum_positive(&sbi->total_valid_inode_count); 2362 } 2363 2364 static inline struct page *f2fs_grab_cache_page(struct address_space *mapping, 2365 pgoff_t index, bool for_write) 2366 { 2367 struct page *page; 2368 2369 if (IS_ENABLED(CONFIG_F2FS_FAULT_INJECTION)) { 2370 if (!for_write) 2371 page = find_get_page_flags(mapping, index, 2372 FGP_LOCK | FGP_ACCESSED); 2373 else 2374 page = find_lock_page(mapping, index); 2375 if (page) 2376 return page; 2377 2378 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_ALLOC)) { 2379 f2fs_show_injection_info(F2FS_M_SB(mapping), 2380 FAULT_PAGE_ALLOC); 2381 return NULL; 2382 } 2383 } 2384 2385 if (!for_write) 2386 return grab_cache_page(mapping, index); 2387 return grab_cache_page_write_begin(mapping, index, AOP_FLAG_NOFS); 2388 } 2389 2390 static inline struct page *f2fs_pagecache_get_page( 2391 struct address_space *mapping, pgoff_t index, 2392 int fgp_flags, gfp_t gfp_mask) 2393 { 2394 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_GET)) { 2395 f2fs_show_injection_info(F2FS_M_SB(mapping), FAULT_PAGE_GET); 2396 return NULL; 2397 } 2398 2399 return pagecache_get_page(mapping, index, fgp_flags, gfp_mask); 2400 } 2401 2402 static inline void f2fs_copy_page(struct page *src, struct page *dst) 2403 { 2404 char *src_kaddr = kmap(src); 2405 char *dst_kaddr = kmap(dst); 2406 2407 memcpy(dst_kaddr, src_kaddr, PAGE_SIZE); 2408 kunmap(dst); 2409 kunmap(src); 2410 } 2411 2412 static inline void f2fs_put_page(struct page *page, int unlock) 2413 { 2414 if (!page) 2415 return; 2416 2417 if (unlock) { 2418 f2fs_bug_on(F2FS_P_SB(page), !PageLocked(page)); 2419 unlock_page(page); 2420 } 2421 put_page(page); 2422 } 2423 2424 static inline void f2fs_put_dnode(struct dnode_of_data *dn) 2425 { 2426 if (dn->node_page) 2427 f2fs_put_page(dn->node_page, 1); 2428 if (dn->inode_page && dn->node_page != dn->inode_page) 2429 f2fs_put_page(dn->inode_page, 0); 2430 dn->node_page = NULL; 2431 dn->inode_page = NULL; 2432 } 2433 2434 static inline struct kmem_cache *f2fs_kmem_cache_create(const char *name, 2435 size_t size) 2436 { 2437 return kmem_cache_create(name, size, 0, SLAB_RECLAIM_ACCOUNT, NULL); 2438 } 2439 2440 static inline void *f2fs_kmem_cache_alloc(struct kmem_cache *cachep, 2441 gfp_t flags) 2442 { 2443 void *entry; 2444 2445 entry = kmem_cache_alloc(cachep, flags); 2446 if (!entry) 2447 entry = kmem_cache_alloc(cachep, flags | __GFP_NOFAIL); 2448 return entry; 2449 } 2450 2451 static inline bool is_idle(struct f2fs_sb_info *sbi, int type) 2452 { 2453 if (sbi->gc_mode == GC_URGENT) 2454 return true; 2455 2456 if (get_pages(sbi, F2FS_RD_DATA) || get_pages(sbi, F2FS_RD_NODE) || 2457 get_pages(sbi, F2FS_RD_META) || get_pages(sbi, F2FS_WB_DATA) || 2458 get_pages(sbi, F2FS_WB_CP_DATA) || 2459 get_pages(sbi, F2FS_DIO_READ) || 2460 get_pages(sbi, F2FS_DIO_WRITE)) 2461 return false; 2462 2463 if (type != DISCARD_TIME && SM_I(sbi) && SM_I(sbi)->dcc_info && 2464 atomic_read(&SM_I(sbi)->dcc_info->queued_discard)) 2465 return false; 2466 2467 if (SM_I(sbi) && SM_I(sbi)->fcc_info && 2468 atomic_read(&SM_I(sbi)->fcc_info->queued_flush)) 2469 return false; 2470 2471 return f2fs_time_over(sbi, type); 2472 } 2473 2474 static inline void f2fs_radix_tree_insert(struct radix_tree_root *root, 2475 unsigned long index, void *item) 2476 { 2477 while (radix_tree_insert(root, index, item)) 2478 cond_resched(); 2479 } 2480 2481 #define RAW_IS_INODE(p) ((p)->footer.nid == (p)->footer.ino) 2482 2483 static inline bool IS_INODE(struct page *page) 2484 { 2485 struct f2fs_node *p = F2FS_NODE(page); 2486 2487 return RAW_IS_INODE(p); 2488 } 2489 2490 static inline int offset_in_addr(struct f2fs_inode *i) 2491 { 2492 return (i->i_inline & F2FS_EXTRA_ATTR) ? 2493 (le16_to_cpu(i->i_extra_isize) / sizeof(__le32)) : 0; 2494 } 2495 2496 static inline __le32 *blkaddr_in_node(struct f2fs_node *node) 2497 { 2498 return RAW_IS_INODE(node) ? node->i.i_addr : node->dn.addr; 2499 } 2500 2501 static inline int f2fs_has_extra_attr(struct inode *inode); 2502 static inline block_t data_blkaddr(struct inode *inode, 2503 struct page *node_page, unsigned int offset) 2504 { 2505 struct f2fs_node *raw_node; 2506 __le32 *addr_array; 2507 int base = 0; 2508 bool is_inode = IS_INODE(node_page); 2509 2510 raw_node = F2FS_NODE(node_page); 2511 2512 if (is_inode) { 2513 if (!inode) 2514 /* from GC path only */ 2515 base = offset_in_addr(&raw_node->i); 2516 else if (f2fs_has_extra_attr(inode)) 2517 base = get_extra_isize(inode); 2518 } 2519 2520 addr_array = blkaddr_in_node(raw_node); 2521 return le32_to_cpu(addr_array[base + offset]); 2522 } 2523 2524 static inline block_t f2fs_data_blkaddr(struct dnode_of_data *dn) 2525 { 2526 return data_blkaddr(dn->inode, dn->node_page, dn->ofs_in_node); 2527 } 2528 2529 static inline int f2fs_test_bit(unsigned int nr, char *addr) 2530 { 2531 int mask; 2532 2533 addr += (nr >> 3); 2534 mask = 1 << (7 - (nr & 0x07)); 2535 return mask & *addr; 2536 } 2537 2538 static inline void f2fs_set_bit(unsigned int nr, char *addr) 2539 { 2540 int mask; 2541 2542 addr += (nr >> 3); 2543 mask = 1 << (7 - (nr & 0x07)); 2544 *addr |= mask; 2545 } 2546 2547 static inline void f2fs_clear_bit(unsigned int nr, char *addr) 2548 { 2549 int mask; 2550 2551 addr += (nr >> 3); 2552 mask = 1 << (7 - (nr & 0x07)); 2553 *addr &= ~mask; 2554 } 2555 2556 static inline int f2fs_test_and_set_bit(unsigned int nr, char *addr) 2557 { 2558 int mask; 2559 int ret; 2560 2561 addr += (nr >> 3); 2562 mask = 1 << (7 - (nr & 0x07)); 2563 ret = mask & *addr; 2564 *addr |= mask; 2565 return ret; 2566 } 2567 2568 static inline int f2fs_test_and_clear_bit(unsigned int nr, char *addr) 2569 { 2570 int mask; 2571 int ret; 2572 2573 addr += (nr >> 3); 2574 mask = 1 << (7 - (nr & 0x07)); 2575 ret = mask & *addr; 2576 *addr &= ~mask; 2577 return ret; 2578 } 2579 2580 static inline void f2fs_change_bit(unsigned int nr, char *addr) 2581 { 2582 int mask; 2583 2584 addr += (nr >> 3); 2585 mask = 1 << (7 - (nr & 0x07)); 2586 *addr ^= mask; 2587 } 2588 2589 /* 2590 * On-disk inode flags (f2fs_inode::i_flags) 2591 */ 2592 #define F2FS_COMPR_FL 0x00000004 /* Compress file */ 2593 #define F2FS_SYNC_FL 0x00000008 /* Synchronous updates */ 2594 #define F2FS_IMMUTABLE_FL 0x00000010 /* Immutable file */ 2595 #define F2FS_APPEND_FL 0x00000020 /* writes to file may only append */ 2596 #define F2FS_NODUMP_FL 0x00000040 /* do not dump file */ 2597 #define F2FS_NOATIME_FL 0x00000080 /* do not update atime */ 2598 #define F2FS_NOCOMP_FL 0x00000400 /* Don't compress */ 2599 #define F2FS_INDEX_FL 0x00001000 /* hash-indexed directory */ 2600 #define F2FS_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */ 2601 #define F2FS_PROJINHERIT_FL 0x20000000 /* Create with parents projid */ 2602 #define F2FS_CASEFOLD_FL 0x40000000 /* Casefolded file */ 2603 2604 /* Flags that should be inherited by new inodes from their parent. */ 2605 #define F2FS_FL_INHERITED (F2FS_SYNC_FL | F2FS_NODUMP_FL | F2FS_NOATIME_FL | \ 2606 F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \ 2607 F2FS_CASEFOLD_FL | F2FS_COMPR_FL | F2FS_NOCOMP_FL) 2608 2609 /* Flags that are appropriate for regular files (all but dir-specific ones). */ 2610 #define F2FS_REG_FLMASK (~(F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \ 2611 F2FS_CASEFOLD_FL)) 2612 2613 /* Flags that are appropriate for non-directories/regular files. */ 2614 #define F2FS_OTHER_FLMASK (F2FS_NODUMP_FL | F2FS_NOATIME_FL) 2615 2616 static inline __u32 f2fs_mask_flags(umode_t mode, __u32 flags) 2617 { 2618 if (S_ISDIR(mode)) 2619 return flags; 2620 else if (S_ISREG(mode)) 2621 return flags & F2FS_REG_FLMASK; 2622 else 2623 return flags & F2FS_OTHER_FLMASK; 2624 } 2625 2626 static inline void __mark_inode_dirty_flag(struct inode *inode, 2627 int flag, bool set) 2628 { 2629 switch (flag) { 2630 case FI_INLINE_XATTR: 2631 case FI_INLINE_DATA: 2632 case FI_INLINE_DENTRY: 2633 case FI_NEW_INODE: 2634 if (set) 2635 return; 2636 /* fall through */ 2637 case FI_DATA_EXIST: 2638 case FI_INLINE_DOTS: 2639 case FI_PIN_FILE: 2640 f2fs_mark_inode_dirty_sync(inode, true); 2641 } 2642 } 2643 2644 static inline void set_inode_flag(struct inode *inode, int flag) 2645 { 2646 test_and_set_bit(flag, F2FS_I(inode)->flags); 2647 __mark_inode_dirty_flag(inode, flag, true); 2648 } 2649 2650 static inline int is_inode_flag_set(struct inode *inode, int flag) 2651 { 2652 return test_bit(flag, F2FS_I(inode)->flags); 2653 } 2654 2655 static inline void clear_inode_flag(struct inode *inode, int flag) 2656 { 2657 test_and_clear_bit(flag, F2FS_I(inode)->flags); 2658 __mark_inode_dirty_flag(inode, flag, false); 2659 } 2660 2661 static inline bool f2fs_verity_in_progress(struct inode *inode) 2662 { 2663 return IS_ENABLED(CONFIG_FS_VERITY) && 2664 is_inode_flag_set(inode, FI_VERITY_IN_PROGRESS); 2665 } 2666 2667 static inline void set_acl_inode(struct inode *inode, umode_t mode) 2668 { 2669 F2FS_I(inode)->i_acl_mode = mode; 2670 set_inode_flag(inode, FI_ACL_MODE); 2671 f2fs_mark_inode_dirty_sync(inode, false); 2672 } 2673 2674 static inline void f2fs_i_links_write(struct inode *inode, bool inc) 2675 { 2676 if (inc) 2677 inc_nlink(inode); 2678 else 2679 drop_nlink(inode); 2680 f2fs_mark_inode_dirty_sync(inode, true); 2681 } 2682 2683 static inline void f2fs_i_blocks_write(struct inode *inode, 2684 block_t diff, bool add, bool claim) 2685 { 2686 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE); 2687 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER); 2688 2689 /* add = 1, claim = 1 should be dquot_reserve_block in pair */ 2690 if (add) { 2691 if (claim) 2692 dquot_claim_block(inode, diff); 2693 else 2694 dquot_alloc_block_nofail(inode, diff); 2695 } else { 2696 dquot_free_block(inode, diff); 2697 } 2698 2699 f2fs_mark_inode_dirty_sync(inode, true); 2700 if (clean || recover) 2701 set_inode_flag(inode, FI_AUTO_RECOVER); 2702 } 2703 2704 static inline void f2fs_i_size_write(struct inode *inode, loff_t i_size) 2705 { 2706 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE); 2707 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER); 2708 2709 if (i_size_read(inode) == i_size) 2710 return; 2711 2712 i_size_write(inode, i_size); 2713 f2fs_mark_inode_dirty_sync(inode, true); 2714 if (clean || recover) 2715 set_inode_flag(inode, FI_AUTO_RECOVER); 2716 } 2717 2718 static inline void f2fs_i_depth_write(struct inode *inode, unsigned int depth) 2719 { 2720 F2FS_I(inode)->i_current_depth = depth; 2721 f2fs_mark_inode_dirty_sync(inode, true); 2722 } 2723 2724 static inline void f2fs_i_gc_failures_write(struct inode *inode, 2725 unsigned int count) 2726 { 2727 F2FS_I(inode)->i_gc_failures[GC_FAILURE_PIN] = count; 2728 f2fs_mark_inode_dirty_sync(inode, true); 2729 } 2730 2731 static inline void f2fs_i_xnid_write(struct inode *inode, nid_t xnid) 2732 { 2733 F2FS_I(inode)->i_xattr_nid = xnid; 2734 f2fs_mark_inode_dirty_sync(inode, true); 2735 } 2736 2737 static inline void f2fs_i_pino_write(struct inode *inode, nid_t pino) 2738 { 2739 F2FS_I(inode)->i_pino = pino; 2740 f2fs_mark_inode_dirty_sync(inode, true); 2741 } 2742 2743 static inline void get_inline_info(struct inode *inode, struct f2fs_inode *ri) 2744 { 2745 struct f2fs_inode_info *fi = F2FS_I(inode); 2746 2747 if (ri->i_inline & F2FS_INLINE_XATTR) 2748 set_bit(FI_INLINE_XATTR, fi->flags); 2749 if (ri->i_inline & F2FS_INLINE_DATA) 2750 set_bit(FI_INLINE_DATA, fi->flags); 2751 if (ri->i_inline & F2FS_INLINE_DENTRY) 2752 set_bit(FI_INLINE_DENTRY, fi->flags); 2753 if (ri->i_inline & F2FS_DATA_EXIST) 2754 set_bit(FI_DATA_EXIST, fi->flags); 2755 if (ri->i_inline & F2FS_INLINE_DOTS) 2756 set_bit(FI_INLINE_DOTS, fi->flags); 2757 if (ri->i_inline & F2FS_EXTRA_ATTR) 2758 set_bit(FI_EXTRA_ATTR, fi->flags); 2759 if (ri->i_inline & F2FS_PIN_FILE) 2760 set_bit(FI_PIN_FILE, fi->flags); 2761 } 2762 2763 static inline void set_raw_inline(struct inode *inode, struct f2fs_inode *ri) 2764 { 2765 ri->i_inline = 0; 2766 2767 if (is_inode_flag_set(inode, FI_INLINE_XATTR)) 2768 ri->i_inline |= F2FS_INLINE_XATTR; 2769 if (is_inode_flag_set(inode, FI_INLINE_DATA)) 2770 ri->i_inline |= F2FS_INLINE_DATA; 2771 if (is_inode_flag_set(inode, FI_INLINE_DENTRY)) 2772 ri->i_inline |= F2FS_INLINE_DENTRY; 2773 if (is_inode_flag_set(inode, FI_DATA_EXIST)) 2774 ri->i_inline |= F2FS_DATA_EXIST; 2775 if (is_inode_flag_set(inode, FI_INLINE_DOTS)) 2776 ri->i_inline |= F2FS_INLINE_DOTS; 2777 if (is_inode_flag_set(inode, FI_EXTRA_ATTR)) 2778 ri->i_inline |= F2FS_EXTRA_ATTR; 2779 if (is_inode_flag_set(inode, FI_PIN_FILE)) 2780 ri->i_inline |= F2FS_PIN_FILE; 2781 } 2782 2783 static inline int f2fs_has_extra_attr(struct inode *inode) 2784 { 2785 return is_inode_flag_set(inode, FI_EXTRA_ATTR); 2786 } 2787 2788 static inline int f2fs_has_inline_xattr(struct inode *inode) 2789 { 2790 return is_inode_flag_set(inode, FI_INLINE_XATTR); 2791 } 2792 2793 static inline int f2fs_compressed_file(struct inode *inode) 2794 { 2795 return S_ISREG(inode->i_mode) && 2796 is_inode_flag_set(inode, FI_COMPRESSED_FILE); 2797 } 2798 2799 static inline unsigned int addrs_per_inode(struct inode *inode) 2800 { 2801 unsigned int addrs = CUR_ADDRS_PER_INODE(inode) - 2802 get_inline_xattr_addrs(inode); 2803 2804 if (!f2fs_compressed_file(inode)) 2805 return addrs; 2806 return ALIGN_DOWN(addrs, F2FS_I(inode)->i_cluster_size); 2807 } 2808 2809 static inline unsigned int addrs_per_block(struct inode *inode) 2810 { 2811 if (!f2fs_compressed_file(inode)) 2812 return DEF_ADDRS_PER_BLOCK; 2813 return ALIGN_DOWN(DEF_ADDRS_PER_BLOCK, F2FS_I(inode)->i_cluster_size); 2814 } 2815 2816 static inline void *inline_xattr_addr(struct inode *inode, struct page *page) 2817 { 2818 struct f2fs_inode *ri = F2FS_INODE(page); 2819 2820 return (void *)&(ri->i_addr[DEF_ADDRS_PER_INODE - 2821 get_inline_xattr_addrs(inode)]); 2822 } 2823 2824 static inline int inline_xattr_size(struct inode *inode) 2825 { 2826 if (f2fs_has_inline_xattr(inode)) 2827 return get_inline_xattr_addrs(inode) * sizeof(__le32); 2828 return 0; 2829 } 2830 2831 static inline int f2fs_has_inline_data(struct inode *inode) 2832 { 2833 return is_inode_flag_set(inode, FI_INLINE_DATA); 2834 } 2835 2836 static inline int f2fs_exist_data(struct inode *inode) 2837 { 2838 return is_inode_flag_set(inode, FI_DATA_EXIST); 2839 } 2840 2841 static inline int f2fs_has_inline_dots(struct inode *inode) 2842 { 2843 return is_inode_flag_set(inode, FI_INLINE_DOTS); 2844 } 2845 2846 static inline int f2fs_is_mmap_file(struct inode *inode) 2847 { 2848 return is_inode_flag_set(inode, FI_MMAP_FILE); 2849 } 2850 2851 static inline bool f2fs_is_pinned_file(struct inode *inode) 2852 { 2853 return is_inode_flag_set(inode, FI_PIN_FILE); 2854 } 2855 2856 static inline bool f2fs_is_atomic_file(struct inode *inode) 2857 { 2858 return is_inode_flag_set(inode, FI_ATOMIC_FILE); 2859 } 2860 2861 static inline bool f2fs_is_commit_atomic_write(struct inode *inode) 2862 { 2863 return is_inode_flag_set(inode, FI_ATOMIC_COMMIT); 2864 } 2865 2866 static inline bool f2fs_is_volatile_file(struct inode *inode) 2867 { 2868 return is_inode_flag_set(inode, FI_VOLATILE_FILE); 2869 } 2870 2871 static inline bool f2fs_is_first_block_written(struct inode *inode) 2872 { 2873 return is_inode_flag_set(inode, FI_FIRST_BLOCK_WRITTEN); 2874 } 2875 2876 static inline bool f2fs_is_drop_cache(struct inode *inode) 2877 { 2878 return is_inode_flag_set(inode, FI_DROP_CACHE); 2879 } 2880 2881 static inline void *inline_data_addr(struct inode *inode, struct page *page) 2882 { 2883 struct f2fs_inode *ri = F2FS_INODE(page); 2884 int extra_size = get_extra_isize(inode); 2885 2886 return (void *)&(ri->i_addr[extra_size + DEF_INLINE_RESERVED_SIZE]); 2887 } 2888 2889 static inline int f2fs_has_inline_dentry(struct inode *inode) 2890 { 2891 return is_inode_flag_set(inode, FI_INLINE_DENTRY); 2892 } 2893 2894 static inline int is_file(struct inode *inode, int type) 2895 { 2896 return F2FS_I(inode)->i_advise & type; 2897 } 2898 2899 static inline void set_file(struct inode *inode, int type) 2900 { 2901 F2FS_I(inode)->i_advise |= type; 2902 f2fs_mark_inode_dirty_sync(inode, true); 2903 } 2904 2905 static inline void clear_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 bool f2fs_is_time_consistent(struct inode *inode) 2912 { 2913 if (!timespec64_equal(F2FS_I(inode)->i_disk_time, &inode->i_atime)) 2914 return false; 2915 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 1, &inode->i_ctime)) 2916 return false; 2917 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 2, &inode->i_mtime)) 2918 return false; 2919 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 3, 2920 &F2FS_I(inode)->i_crtime)) 2921 return false; 2922 return true; 2923 } 2924 2925 static inline bool f2fs_skip_inode_update(struct inode *inode, int dsync) 2926 { 2927 bool ret; 2928 2929 if (dsync) { 2930 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 2931 2932 spin_lock(&sbi->inode_lock[DIRTY_META]); 2933 ret = list_empty(&F2FS_I(inode)->gdirty_list); 2934 spin_unlock(&sbi->inode_lock[DIRTY_META]); 2935 return ret; 2936 } 2937 if (!is_inode_flag_set(inode, FI_AUTO_RECOVER) || 2938 file_keep_isize(inode) || 2939 i_size_read(inode) & ~PAGE_MASK) 2940 return false; 2941 2942 if (!f2fs_is_time_consistent(inode)) 2943 return false; 2944 2945 spin_lock(&F2FS_I(inode)->i_size_lock); 2946 ret = F2FS_I(inode)->last_disk_size == i_size_read(inode); 2947 spin_unlock(&F2FS_I(inode)->i_size_lock); 2948 2949 return ret; 2950 } 2951 2952 static inline bool f2fs_readonly(struct super_block *sb) 2953 { 2954 return sb_rdonly(sb); 2955 } 2956 2957 static inline bool f2fs_cp_error(struct f2fs_sb_info *sbi) 2958 { 2959 return is_set_ckpt_flags(sbi, CP_ERROR_FLAG); 2960 } 2961 2962 static inline bool is_dot_dotdot(const u8 *name, size_t len) 2963 { 2964 if (len == 1 && name[0] == '.') 2965 return true; 2966 2967 if (len == 2 && name[0] == '.' && name[1] == '.') 2968 return true; 2969 2970 return false; 2971 } 2972 2973 static inline bool f2fs_may_extent_tree(struct inode *inode) 2974 { 2975 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 2976 2977 if (!test_opt(sbi, EXTENT_CACHE) || 2978 is_inode_flag_set(inode, FI_NO_EXTENT) || 2979 is_inode_flag_set(inode, FI_COMPRESSED_FILE)) 2980 return false; 2981 2982 /* 2983 * for recovered files during mount do not create extents 2984 * if shrinker is not registered. 2985 */ 2986 if (list_empty(&sbi->s_list)) 2987 return false; 2988 2989 return S_ISREG(inode->i_mode); 2990 } 2991 2992 static inline void *f2fs_kmalloc(struct f2fs_sb_info *sbi, 2993 size_t size, gfp_t flags) 2994 { 2995 void *ret; 2996 2997 if (time_to_inject(sbi, FAULT_KMALLOC)) { 2998 f2fs_show_injection_info(sbi, FAULT_KMALLOC); 2999 return NULL; 3000 } 3001 3002 ret = kmalloc(size, flags); 3003 if (ret) 3004 return ret; 3005 3006 return kvmalloc(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 get_page(page); 3128 SetPagePrivate(page); 3129 set_page_private(page, data); 3130 } 3131 3132 static inline void f2fs_clear_page_private(struct page *page) 3133 { 3134 if (!PagePrivate(page)) 3135 return; 3136 3137 set_page_private(page, 0); 3138 ClearPagePrivate(page); 3139 f2fs_put_page(page, 0); 3140 } 3141 3142 /* 3143 * file.c 3144 */ 3145 int f2fs_sync_file(struct file *file, loff_t start, loff_t end, int datasync); 3146 void f2fs_truncate_data_blocks(struct dnode_of_data *dn); 3147 int f2fs_do_truncate_blocks(struct inode *inode, u64 from, bool lock); 3148 int f2fs_truncate_blocks(struct inode *inode, u64 from, bool lock); 3149 int f2fs_truncate(struct inode *inode); 3150 int f2fs_getattr(const struct path *path, struct kstat *stat, 3151 u32 request_mask, unsigned int flags); 3152 int f2fs_setattr(struct dentry *dentry, struct iattr *attr); 3153 int f2fs_truncate_hole(struct inode *inode, pgoff_t pg_start, pgoff_t pg_end); 3154 void f2fs_truncate_data_blocks_range(struct dnode_of_data *dn, int count); 3155 int f2fs_precache_extents(struct inode *inode); 3156 long f2fs_ioctl(struct file *filp, unsigned int cmd, unsigned long arg); 3157 long f2fs_compat_ioctl(struct file *file, unsigned int cmd, unsigned long arg); 3158 int f2fs_transfer_project_quota(struct inode *inode, kprojid_t kprojid); 3159 int f2fs_pin_file_control(struct inode *inode, bool inc); 3160 3161 /* 3162 * inode.c 3163 */ 3164 void f2fs_set_inode_flags(struct inode *inode); 3165 bool f2fs_inode_chksum_verify(struct f2fs_sb_info *sbi, struct page *page); 3166 void f2fs_inode_chksum_set(struct f2fs_sb_info *sbi, struct page *page); 3167 struct inode *f2fs_iget(struct super_block *sb, unsigned long ino); 3168 struct inode *f2fs_iget_retry(struct super_block *sb, unsigned long ino); 3169 int f2fs_try_to_free_nats(struct f2fs_sb_info *sbi, int nr_shrink); 3170 void f2fs_update_inode(struct inode *inode, struct page *node_page); 3171 void f2fs_update_inode_page(struct inode *inode); 3172 int f2fs_write_inode(struct inode *inode, struct writeback_control *wbc); 3173 void f2fs_evict_inode(struct inode *inode); 3174 void f2fs_handle_failed_inode(struct inode *inode); 3175 3176 /* 3177 * namei.c 3178 */ 3179 int f2fs_update_extension_list(struct f2fs_sb_info *sbi, const char *name, 3180 bool hot, bool set); 3181 struct dentry *f2fs_get_parent(struct dentry *child); 3182 3183 /* 3184 * dir.c 3185 */ 3186 unsigned char f2fs_get_de_type(struct f2fs_dir_entry *de); 3187 int f2fs_init_casefolded_name(const struct inode *dir, 3188 struct f2fs_filename *fname); 3189 int f2fs_setup_filename(struct inode *dir, const struct qstr *iname, 3190 int lookup, struct f2fs_filename *fname); 3191 int f2fs_prepare_lookup(struct inode *dir, struct dentry *dentry, 3192 struct f2fs_filename *fname); 3193 void f2fs_free_filename(struct f2fs_filename *fname); 3194 struct f2fs_dir_entry *f2fs_find_target_dentry(const struct f2fs_dentry_ptr *d, 3195 const struct f2fs_filename *fname, int *max_slots); 3196 int f2fs_fill_dentries(struct dir_context *ctx, struct f2fs_dentry_ptr *d, 3197 unsigned int start_pos, struct fscrypt_str *fstr); 3198 void f2fs_do_make_empty_dir(struct inode *inode, struct inode *parent, 3199 struct f2fs_dentry_ptr *d); 3200 struct page *f2fs_init_inode_metadata(struct inode *inode, struct inode *dir, 3201 const struct f2fs_filename *fname, struct page *dpage); 3202 void f2fs_update_parent_metadata(struct inode *dir, struct inode *inode, 3203 unsigned int current_depth); 3204 int f2fs_room_for_filename(const void *bitmap, int slots, int max_slots); 3205 void f2fs_drop_nlink(struct inode *dir, struct inode *inode); 3206 struct f2fs_dir_entry *__f2fs_find_entry(struct inode *dir, 3207 const struct f2fs_filename *fname, 3208 struct page **res_page); 3209 struct f2fs_dir_entry *f2fs_find_entry(struct inode *dir, 3210 const struct qstr *child, struct page **res_page); 3211 struct f2fs_dir_entry *f2fs_parent_dir(struct inode *dir, struct page **p); 3212 ino_t f2fs_inode_by_name(struct inode *dir, const struct qstr *qstr, 3213 struct page **page); 3214 void f2fs_set_link(struct inode *dir, struct f2fs_dir_entry *de, 3215 struct page *page, struct inode *inode); 3216 bool f2fs_has_enough_room(struct inode *dir, struct page *ipage, 3217 const struct f2fs_filename *fname); 3218 void f2fs_update_dentry(nid_t ino, umode_t mode, struct f2fs_dentry_ptr *d, 3219 const struct fscrypt_str *name, f2fs_hash_t name_hash, 3220 unsigned int bit_pos); 3221 int f2fs_add_regular_entry(struct inode *dir, const struct f2fs_filename *fname, 3222 struct inode *inode, nid_t ino, umode_t mode); 3223 int f2fs_add_dentry(struct inode *dir, const struct f2fs_filename *fname, 3224 struct inode *inode, nid_t ino, umode_t mode); 3225 int f2fs_do_add_link(struct inode *dir, const struct qstr *name, 3226 struct inode *inode, nid_t ino, umode_t mode); 3227 void f2fs_delete_entry(struct f2fs_dir_entry *dentry, struct page *page, 3228 struct inode *dir, struct inode *inode); 3229 int f2fs_do_tmpfile(struct inode *inode, struct inode *dir); 3230 bool f2fs_empty_dir(struct inode *dir); 3231 3232 static inline int f2fs_add_link(struct dentry *dentry, struct inode *inode) 3233 { 3234 return f2fs_do_add_link(d_inode(dentry->d_parent), &dentry->d_name, 3235 inode, inode->i_ino, inode->i_mode); 3236 } 3237 3238 /* 3239 * super.c 3240 */ 3241 int f2fs_inode_dirtied(struct inode *inode, bool sync); 3242 void f2fs_inode_synced(struct inode *inode); 3243 int f2fs_enable_quota_files(struct f2fs_sb_info *sbi, bool rdonly); 3244 int f2fs_quota_sync(struct super_block *sb, int type); 3245 void f2fs_quota_off_umount(struct super_block *sb); 3246 int f2fs_commit_super(struct f2fs_sb_info *sbi, bool recover); 3247 int f2fs_sync_fs(struct super_block *sb, int sync); 3248 int f2fs_sanity_check_ckpt(struct f2fs_sb_info *sbi); 3249 3250 /* 3251 * hash.c 3252 */ 3253 void f2fs_hash_filename(const struct inode *dir, struct f2fs_filename *fname); 3254 3255 /* 3256 * node.c 3257 */ 3258 struct dnode_of_data; 3259 struct node_info; 3260 3261 int f2fs_check_nid_range(struct f2fs_sb_info *sbi, nid_t nid); 3262 bool f2fs_available_free_memory(struct f2fs_sb_info *sbi, int type); 3263 bool f2fs_in_warm_node_list(struct f2fs_sb_info *sbi, struct page *page); 3264 void f2fs_init_fsync_node_info(struct f2fs_sb_info *sbi); 3265 void f2fs_del_fsync_node_entry(struct f2fs_sb_info *sbi, struct page *page); 3266 void f2fs_reset_fsync_node_info(struct f2fs_sb_info *sbi); 3267 int f2fs_need_dentry_mark(struct f2fs_sb_info *sbi, nid_t nid); 3268 bool f2fs_is_checkpointed_node(struct f2fs_sb_info *sbi, nid_t nid); 3269 bool f2fs_need_inode_block_update(struct f2fs_sb_info *sbi, nid_t ino); 3270 int f2fs_get_node_info(struct f2fs_sb_info *sbi, nid_t nid, 3271 struct node_info *ni); 3272 pgoff_t f2fs_get_next_page_offset(struct dnode_of_data *dn, pgoff_t pgofs); 3273 int f2fs_get_dnode_of_data(struct dnode_of_data *dn, pgoff_t index, int mode); 3274 int f2fs_truncate_inode_blocks(struct inode *inode, pgoff_t from); 3275 int f2fs_truncate_xattr_node(struct inode *inode); 3276 int f2fs_wait_on_node_pages_writeback(struct f2fs_sb_info *sbi, 3277 unsigned int seq_id); 3278 int f2fs_remove_inode_page(struct inode *inode); 3279 struct page *f2fs_new_inode_page(struct inode *inode); 3280 struct page *f2fs_new_node_page(struct dnode_of_data *dn, unsigned int ofs); 3281 void f2fs_ra_node_page(struct f2fs_sb_info *sbi, nid_t nid); 3282 struct page *f2fs_get_node_page(struct f2fs_sb_info *sbi, pgoff_t nid); 3283 struct page *f2fs_get_node_page_ra(struct page *parent, int start); 3284 int f2fs_move_node_page(struct page *node_page, int gc_type); 3285 int f2fs_fsync_node_pages(struct f2fs_sb_info *sbi, struct inode *inode, 3286 struct writeback_control *wbc, bool atomic, 3287 unsigned int *seq_id); 3288 int f2fs_sync_node_pages(struct f2fs_sb_info *sbi, 3289 struct writeback_control *wbc, 3290 bool do_balance, enum iostat_type io_type); 3291 int f2fs_build_free_nids(struct f2fs_sb_info *sbi, bool sync, bool mount); 3292 bool f2fs_alloc_nid(struct f2fs_sb_info *sbi, nid_t *nid); 3293 void f2fs_alloc_nid_done(struct f2fs_sb_info *sbi, nid_t nid); 3294 void f2fs_alloc_nid_failed(struct f2fs_sb_info *sbi, nid_t nid); 3295 int f2fs_try_to_free_nids(struct f2fs_sb_info *sbi, int nr_shrink); 3296 void f2fs_recover_inline_xattr(struct inode *inode, struct page *page); 3297 int f2fs_recover_xattr_data(struct inode *inode, struct page *page); 3298 int f2fs_recover_inode_page(struct f2fs_sb_info *sbi, struct page *page); 3299 int f2fs_restore_node_summary(struct f2fs_sb_info *sbi, 3300 unsigned int segno, struct f2fs_summary_block *sum); 3301 int f2fs_flush_nat_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3302 int f2fs_build_node_manager(struct f2fs_sb_info *sbi); 3303 void f2fs_destroy_node_manager(struct f2fs_sb_info *sbi); 3304 int __init f2fs_create_node_manager_caches(void); 3305 void f2fs_destroy_node_manager_caches(void); 3306 3307 /* 3308 * segment.c 3309 */ 3310 bool f2fs_need_SSR(struct f2fs_sb_info *sbi); 3311 void f2fs_register_inmem_page(struct inode *inode, struct page *page); 3312 void f2fs_drop_inmem_pages_all(struct f2fs_sb_info *sbi, bool gc_failure); 3313 void f2fs_drop_inmem_pages(struct inode *inode); 3314 void f2fs_drop_inmem_page(struct inode *inode, struct page *page); 3315 int f2fs_commit_inmem_pages(struct inode *inode); 3316 void f2fs_balance_fs(struct f2fs_sb_info *sbi, bool need); 3317 void f2fs_balance_fs_bg(struct f2fs_sb_info *sbi, bool from_bg); 3318 int f2fs_issue_flush(struct f2fs_sb_info *sbi, nid_t ino); 3319 int f2fs_create_flush_cmd_control(struct f2fs_sb_info *sbi); 3320 int f2fs_flush_device_cache(struct f2fs_sb_info *sbi); 3321 void f2fs_destroy_flush_cmd_control(struct f2fs_sb_info *sbi, bool free); 3322 void f2fs_invalidate_blocks(struct f2fs_sb_info *sbi, block_t addr); 3323 bool f2fs_is_checkpointed_data(struct f2fs_sb_info *sbi, block_t blkaddr); 3324 void f2fs_drop_discard_cmd(struct f2fs_sb_info *sbi); 3325 void f2fs_stop_discard_thread(struct f2fs_sb_info *sbi); 3326 bool f2fs_issue_discard_timeout(struct f2fs_sb_info *sbi); 3327 void f2fs_clear_prefree_segments(struct f2fs_sb_info *sbi, 3328 struct cp_control *cpc); 3329 void f2fs_dirty_to_prefree(struct f2fs_sb_info *sbi); 3330 block_t f2fs_get_unusable_blocks(struct f2fs_sb_info *sbi); 3331 int f2fs_disable_cp_again(struct f2fs_sb_info *sbi, block_t unusable); 3332 void f2fs_release_discard_addrs(struct f2fs_sb_info *sbi); 3333 int f2fs_npages_for_summary_flush(struct f2fs_sb_info *sbi, bool for_ra); 3334 void allocate_segment_for_resize(struct f2fs_sb_info *sbi, int type, 3335 unsigned int start, unsigned int end); 3336 void f2fs_allocate_new_segments(struct f2fs_sb_info *sbi, int type); 3337 int f2fs_trim_fs(struct f2fs_sb_info *sbi, struct fstrim_range *range); 3338 bool f2fs_exist_trim_candidates(struct f2fs_sb_info *sbi, 3339 struct cp_control *cpc); 3340 struct page *f2fs_get_sum_page(struct f2fs_sb_info *sbi, unsigned int segno); 3341 void f2fs_update_meta_page(struct f2fs_sb_info *sbi, void *src, 3342 block_t blk_addr); 3343 void f2fs_do_write_meta_page(struct f2fs_sb_info *sbi, struct page *page, 3344 enum iostat_type io_type); 3345 void f2fs_do_write_node_page(unsigned int nid, struct f2fs_io_info *fio); 3346 void f2fs_outplace_write_data(struct dnode_of_data *dn, 3347 struct f2fs_io_info *fio); 3348 int f2fs_inplace_write_data(struct f2fs_io_info *fio); 3349 void f2fs_do_replace_block(struct f2fs_sb_info *sbi, struct f2fs_summary *sum, 3350 block_t old_blkaddr, block_t new_blkaddr, 3351 bool recover_curseg, bool recover_newaddr); 3352 void f2fs_replace_block(struct f2fs_sb_info *sbi, struct dnode_of_data *dn, 3353 block_t old_addr, block_t new_addr, 3354 unsigned char version, bool recover_curseg, 3355 bool recover_newaddr); 3356 void f2fs_allocate_data_block(struct f2fs_sb_info *sbi, struct page *page, 3357 block_t old_blkaddr, block_t *new_blkaddr, 3358 struct f2fs_summary *sum, int type, 3359 struct f2fs_io_info *fio, bool add_list); 3360 void f2fs_wait_on_page_writeback(struct page *page, 3361 enum page_type type, bool ordered, bool locked); 3362 void f2fs_wait_on_block_writeback(struct inode *inode, block_t blkaddr); 3363 void f2fs_wait_on_block_writeback_range(struct inode *inode, block_t blkaddr, 3364 block_t len); 3365 void f2fs_write_data_summaries(struct f2fs_sb_info *sbi, block_t start_blk); 3366 void f2fs_write_node_summaries(struct f2fs_sb_info *sbi, block_t start_blk); 3367 int f2fs_lookup_journal_in_cursum(struct f2fs_journal *journal, int type, 3368 unsigned int val, int alloc); 3369 void f2fs_flush_sit_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3370 int f2fs_fix_curseg_write_pointer(struct f2fs_sb_info *sbi); 3371 int f2fs_check_write_pointer(struct f2fs_sb_info *sbi); 3372 int f2fs_build_segment_manager(struct f2fs_sb_info *sbi); 3373 void f2fs_destroy_segment_manager(struct f2fs_sb_info *sbi); 3374 int __init f2fs_create_segment_manager_caches(void); 3375 void f2fs_destroy_segment_manager_caches(void); 3376 int f2fs_rw_hint_to_seg_type(enum rw_hint hint); 3377 enum rw_hint f2fs_io_type_to_rw_hint(struct f2fs_sb_info *sbi, 3378 enum page_type type, enum temp_type temp); 3379 3380 /* 3381 * checkpoint.c 3382 */ 3383 void f2fs_stop_checkpoint(struct f2fs_sb_info *sbi, bool end_io); 3384 struct page *f2fs_grab_meta_page(struct f2fs_sb_info *sbi, pgoff_t index); 3385 struct page *f2fs_get_meta_page(struct f2fs_sb_info *sbi, pgoff_t index); 3386 struct page *f2fs_get_meta_page_nofail(struct f2fs_sb_info *sbi, pgoff_t index); 3387 struct page *f2fs_get_tmp_page(struct f2fs_sb_info *sbi, pgoff_t index); 3388 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi, 3389 block_t blkaddr, int type); 3390 int f2fs_ra_meta_pages(struct f2fs_sb_info *sbi, block_t start, int nrpages, 3391 int type, bool sync); 3392 void f2fs_ra_meta_pages_cond(struct f2fs_sb_info *sbi, pgoff_t index); 3393 long f2fs_sync_meta_pages(struct f2fs_sb_info *sbi, enum page_type type, 3394 long nr_to_write, enum iostat_type io_type); 3395 void f2fs_add_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type); 3396 void f2fs_remove_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type); 3397 void f2fs_release_ino_entry(struct f2fs_sb_info *sbi, bool all); 3398 bool f2fs_exist_written_data(struct f2fs_sb_info *sbi, nid_t ino, int mode); 3399 void f2fs_set_dirty_device(struct f2fs_sb_info *sbi, nid_t ino, 3400 unsigned int devidx, int type); 3401 bool f2fs_is_dirty_device(struct f2fs_sb_info *sbi, nid_t ino, 3402 unsigned int devidx, int type); 3403 int f2fs_sync_inode_meta(struct f2fs_sb_info *sbi); 3404 int f2fs_acquire_orphan_inode(struct f2fs_sb_info *sbi); 3405 void f2fs_release_orphan_inode(struct f2fs_sb_info *sbi); 3406 void f2fs_add_orphan_inode(struct inode *inode); 3407 void f2fs_remove_orphan_inode(struct f2fs_sb_info *sbi, nid_t ino); 3408 int f2fs_recover_orphan_inodes(struct f2fs_sb_info *sbi); 3409 int f2fs_get_valid_checkpoint(struct f2fs_sb_info *sbi); 3410 void f2fs_update_dirty_page(struct inode *inode, struct page *page); 3411 void f2fs_remove_dirty_inode(struct inode *inode); 3412 int f2fs_sync_dirty_inodes(struct f2fs_sb_info *sbi, enum inode_type type); 3413 void f2fs_wait_on_all_pages(struct f2fs_sb_info *sbi, int type); 3414 int f2fs_write_checkpoint(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3415 void f2fs_init_ino_entry_info(struct f2fs_sb_info *sbi); 3416 int __init f2fs_create_checkpoint_caches(void); 3417 void f2fs_destroy_checkpoint_caches(void); 3418 3419 /* 3420 * data.c 3421 */ 3422 int __init f2fs_init_bioset(void); 3423 void f2fs_destroy_bioset(void); 3424 struct bio *f2fs_bio_alloc(struct f2fs_sb_info *sbi, int npages, bool noio); 3425 int f2fs_init_bio_entry_cache(void); 3426 void f2fs_destroy_bio_entry_cache(void); 3427 void f2fs_submit_bio(struct f2fs_sb_info *sbi, 3428 struct bio *bio, enum page_type type); 3429 void f2fs_submit_merged_write(struct f2fs_sb_info *sbi, enum page_type type); 3430 void f2fs_submit_merged_write_cond(struct f2fs_sb_info *sbi, 3431 struct inode *inode, struct page *page, 3432 nid_t ino, enum page_type type); 3433 void f2fs_submit_merged_ipu_write(struct f2fs_sb_info *sbi, 3434 struct bio **bio, struct page *page); 3435 void f2fs_flush_merged_writes(struct f2fs_sb_info *sbi); 3436 int f2fs_submit_page_bio(struct f2fs_io_info *fio); 3437 int f2fs_merge_page_bio(struct f2fs_io_info *fio); 3438 void f2fs_submit_page_write(struct f2fs_io_info *fio); 3439 struct block_device *f2fs_target_device(struct f2fs_sb_info *sbi, 3440 block_t blk_addr, struct bio *bio); 3441 int f2fs_target_device_index(struct f2fs_sb_info *sbi, block_t blkaddr); 3442 void f2fs_set_data_blkaddr(struct dnode_of_data *dn); 3443 void f2fs_update_data_blkaddr(struct dnode_of_data *dn, block_t blkaddr); 3444 int f2fs_reserve_new_blocks(struct dnode_of_data *dn, blkcnt_t count); 3445 int f2fs_reserve_new_block(struct dnode_of_data *dn); 3446 int f2fs_get_block(struct dnode_of_data *dn, pgoff_t index); 3447 int f2fs_preallocate_blocks(struct kiocb *iocb, struct iov_iter *from); 3448 int f2fs_reserve_block(struct dnode_of_data *dn, pgoff_t index); 3449 int f2fs_mpage_readpages(struct address_space *mapping, 3450 struct list_head *pages, struct page *page, 3451 unsigned nr_pages, bool is_readahead); 3452 struct page *f2fs_get_read_data_page(struct inode *inode, pgoff_t index, 3453 int op_flags, bool for_write); 3454 struct page *f2fs_find_data_page(struct inode *inode, pgoff_t index); 3455 struct page *f2fs_get_lock_data_page(struct inode *inode, pgoff_t index, 3456 bool for_write); 3457 struct page *f2fs_get_new_data_page(struct inode *inode, 3458 struct page *ipage, pgoff_t index, bool new_i_size); 3459 int f2fs_do_write_data_page(struct f2fs_io_info *fio); 3460 void __do_map_lock(struct f2fs_sb_info *sbi, int flag, bool lock); 3461 int f2fs_map_blocks(struct inode *inode, struct f2fs_map_blocks *map, 3462 int create, int flag); 3463 int f2fs_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, 3464 u64 start, u64 len); 3465 int f2fs_encrypt_one_page(struct f2fs_io_info *fio); 3466 bool f2fs_should_update_inplace(struct inode *inode, struct f2fs_io_info *fio); 3467 bool f2fs_should_update_outplace(struct inode *inode, struct f2fs_io_info *fio); 3468 int f2fs_write_single_data_page(struct page *page, int *submitted, 3469 struct bio **bio, sector_t *last_block, 3470 struct writeback_control *wbc, 3471 enum iostat_type io_type, 3472 int compr_blocks); 3473 void f2fs_invalidate_page(struct page *page, unsigned int offset, 3474 unsigned int length); 3475 int f2fs_release_page(struct page *page, gfp_t wait); 3476 #ifdef CONFIG_MIGRATION 3477 int f2fs_migrate_page(struct address_space *mapping, struct page *newpage, 3478 struct page *page, enum migrate_mode mode); 3479 #endif 3480 bool f2fs_overwrite_io(struct inode *inode, loff_t pos, size_t len); 3481 void f2fs_clear_page_cache_dirty_tag(struct page *page); 3482 int f2fs_init_post_read_processing(void); 3483 void f2fs_destroy_post_read_processing(void); 3484 int f2fs_init_post_read_wq(struct f2fs_sb_info *sbi); 3485 void f2fs_destroy_post_read_wq(struct f2fs_sb_info *sbi); 3486 3487 /* 3488 * gc.c 3489 */ 3490 int f2fs_start_gc_thread(struct f2fs_sb_info *sbi); 3491 void f2fs_stop_gc_thread(struct f2fs_sb_info *sbi); 3492 block_t f2fs_start_bidx_of_node(unsigned int node_ofs, struct inode *inode); 3493 int f2fs_gc(struct f2fs_sb_info *sbi, bool sync, bool background, 3494 unsigned int segno); 3495 void f2fs_build_gc_manager(struct f2fs_sb_info *sbi); 3496 int f2fs_resize_fs(struct f2fs_sb_info *sbi, __u64 block_count); 3497 3498 /* 3499 * recovery.c 3500 */ 3501 int f2fs_recover_fsync_data(struct f2fs_sb_info *sbi, bool check_only); 3502 bool f2fs_space_for_roll_forward(struct f2fs_sb_info *sbi); 3503 3504 /* 3505 * debug.c 3506 */ 3507 #ifdef CONFIG_F2FS_STAT_FS 3508 struct f2fs_stat_info { 3509 struct list_head stat_list; 3510 struct f2fs_sb_info *sbi; 3511 int all_area_segs, sit_area_segs, nat_area_segs, ssa_area_segs; 3512 int main_area_segs, main_area_sections, main_area_zones; 3513 unsigned long long hit_largest, hit_cached, hit_rbtree; 3514 unsigned long long hit_total, total_ext; 3515 int ext_tree, zombie_tree, ext_node; 3516 int ndirty_node, ndirty_dent, ndirty_meta, ndirty_imeta; 3517 int ndirty_data, ndirty_qdata; 3518 int inmem_pages; 3519 unsigned int ndirty_dirs, ndirty_files, nquota_files, ndirty_all; 3520 int nats, dirty_nats, sits, dirty_sits; 3521 int free_nids, avail_nids, alloc_nids; 3522 int total_count, utilization; 3523 int bg_gc, nr_wb_cp_data, nr_wb_data; 3524 int nr_rd_data, nr_rd_node, nr_rd_meta; 3525 int nr_dio_read, nr_dio_write; 3526 unsigned int io_skip_bggc, other_skip_bggc; 3527 int nr_flushing, nr_flushed, flush_list_empty; 3528 int nr_discarding, nr_discarded; 3529 int nr_discard_cmd; 3530 unsigned int undiscard_blks; 3531 int inline_xattr, inline_inode, inline_dir, append, update, orphans; 3532 int compr_inode, compr_blocks; 3533 int aw_cnt, max_aw_cnt, vw_cnt, max_vw_cnt; 3534 unsigned int valid_count, valid_node_count, valid_inode_count, discard_blks; 3535 unsigned int bimodal, avg_vblocks; 3536 int util_free, util_valid, util_invalid; 3537 int rsvd_segs, overp_segs; 3538 int dirty_count, node_pages, meta_pages; 3539 int prefree_count, call_count, cp_count, bg_cp_count; 3540 int tot_segs, node_segs, data_segs, free_segs, free_secs; 3541 int bg_node_segs, bg_data_segs; 3542 int tot_blks, data_blks, node_blks; 3543 int bg_data_blks, bg_node_blks; 3544 unsigned long long skipped_atomic_files[2]; 3545 int curseg[NR_CURSEG_TYPE]; 3546 int cursec[NR_CURSEG_TYPE]; 3547 int curzone[NR_CURSEG_TYPE]; 3548 3549 unsigned int meta_count[META_MAX]; 3550 unsigned int segment_count[2]; 3551 unsigned int block_count[2]; 3552 unsigned int inplace_count; 3553 unsigned long long base_mem, cache_mem, page_mem; 3554 }; 3555 3556 static inline struct f2fs_stat_info *F2FS_STAT(struct f2fs_sb_info *sbi) 3557 { 3558 return (struct f2fs_stat_info *)sbi->stat_info; 3559 } 3560 3561 #define stat_inc_cp_count(si) ((si)->cp_count++) 3562 #define stat_inc_bg_cp_count(si) ((si)->bg_cp_count++) 3563 #define stat_inc_call_count(si) ((si)->call_count++) 3564 #define stat_inc_bggc_count(si) ((si)->bg_gc++) 3565 #define stat_io_skip_bggc_count(sbi) ((sbi)->io_skip_bggc++) 3566 #define stat_other_skip_bggc_count(sbi) ((sbi)->other_skip_bggc++) 3567 #define stat_inc_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]++) 3568 #define stat_dec_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]--) 3569 #define stat_inc_total_hit(sbi) (atomic64_inc(&(sbi)->total_hit_ext)) 3570 #define stat_inc_rbtree_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_rbtree)) 3571 #define stat_inc_largest_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_largest)) 3572 #define stat_inc_cached_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_cached)) 3573 #define stat_inc_inline_xattr(inode) \ 3574 do { \ 3575 if (f2fs_has_inline_xattr(inode)) \ 3576 (atomic_inc(&F2FS_I_SB(inode)->inline_xattr)); \ 3577 } while (0) 3578 #define stat_dec_inline_xattr(inode) \ 3579 do { \ 3580 if (f2fs_has_inline_xattr(inode)) \ 3581 (atomic_dec(&F2FS_I_SB(inode)->inline_xattr)); \ 3582 } while (0) 3583 #define stat_inc_inline_inode(inode) \ 3584 do { \ 3585 if (f2fs_has_inline_data(inode)) \ 3586 (atomic_inc(&F2FS_I_SB(inode)->inline_inode)); \ 3587 } while (0) 3588 #define stat_dec_inline_inode(inode) \ 3589 do { \ 3590 if (f2fs_has_inline_data(inode)) \ 3591 (atomic_dec(&F2FS_I_SB(inode)->inline_inode)); \ 3592 } while (0) 3593 #define stat_inc_inline_dir(inode) \ 3594 do { \ 3595 if (f2fs_has_inline_dentry(inode)) \ 3596 (atomic_inc(&F2FS_I_SB(inode)->inline_dir)); \ 3597 } while (0) 3598 #define stat_dec_inline_dir(inode) \ 3599 do { \ 3600 if (f2fs_has_inline_dentry(inode)) \ 3601 (atomic_dec(&F2FS_I_SB(inode)->inline_dir)); \ 3602 } while (0) 3603 #define stat_inc_compr_inode(inode) \ 3604 do { \ 3605 if (f2fs_compressed_file(inode)) \ 3606 (atomic_inc(&F2FS_I_SB(inode)->compr_inode)); \ 3607 } while (0) 3608 #define stat_dec_compr_inode(inode) \ 3609 do { \ 3610 if (f2fs_compressed_file(inode)) \ 3611 (atomic_dec(&F2FS_I_SB(inode)->compr_inode)); \ 3612 } while (0) 3613 #define stat_add_compr_blocks(inode, blocks) \ 3614 (atomic_add(blocks, &F2FS_I_SB(inode)->compr_blocks)) 3615 #define stat_sub_compr_blocks(inode, blocks) \ 3616 (atomic_sub(blocks, &F2FS_I_SB(inode)->compr_blocks)) 3617 #define stat_inc_meta_count(sbi, blkaddr) \ 3618 do { \ 3619 if (blkaddr < SIT_I(sbi)->sit_base_addr) \ 3620 atomic_inc(&(sbi)->meta_count[META_CP]); \ 3621 else if (blkaddr < NM_I(sbi)->nat_blkaddr) \ 3622 atomic_inc(&(sbi)->meta_count[META_SIT]); \ 3623 else if (blkaddr < SM_I(sbi)->ssa_blkaddr) \ 3624 atomic_inc(&(sbi)->meta_count[META_NAT]); \ 3625 else if (blkaddr < SM_I(sbi)->main_blkaddr) \ 3626 atomic_inc(&(sbi)->meta_count[META_SSA]); \ 3627 } while (0) 3628 #define stat_inc_seg_type(sbi, curseg) \ 3629 ((sbi)->segment_count[(curseg)->alloc_type]++) 3630 #define stat_inc_block_count(sbi, curseg) \ 3631 ((sbi)->block_count[(curseg)->alloc_type]++) 3632 #define stat_inc_inplace_blocks(sbi) \ 3633 (atomic_inc(&(sbi)->inplace_count)) 3634 #define stat_update_max_atomic_write(inode) \ 3635 do { \ 3636 int cur = F2FS_I_SB(inode)->atomic_files; \ 3637 int max = atomic_read(&F2FS_I_SB(inode)->max_aw_cnt); \ 3638 if (cur > max) \ 3639 atomic_set(&F2FS_I_SB(inode)->max_aw_cnt, cur); \ 3640 } while (0) 3641 #define stat_inc_volatile_write(inode) \ 3642 (atomic_inc(&F2FS_I_SB(inode)->vw_cnt)) 3643 #define stat_dec_volatile_write(inode) \ 3644 (atomic_dec(&F2FS_I_SB(inode)->vw_cnt)) 3645 #define stat_update_max_volatile_write(inode) \ 3646 do { \ 3647 int cur = atomic_read(&F2FS_I_SB(inode)->vw_cnt); \ 3648 int max = atomic_read(&F2FS_I_SB(inode)->max_vw_cnt); \ 3649 if (cur > max) \ 3650 atomic_set(&F2FS_I_SB(inode)->max_vw_cnt, cur); \ 3651 } while (0) 3652 #define stat_inc_seg_count(sbi, type, gc_type) \ 3653 do { \ 3654 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 3655 si->tot_segs++; \ 3656 if ((type) == SUM_TYPE_DATA) { \ 3657 si->data_segs++; \ 3658 si->bg_data_segs += (gc_type == BG_GC) ? 1 : 0; \ 3659 } else { \ 3660 si->node_segs++; \ 3661 si->bg_node_segs += (gc_type == BG_GC) ? 1 : 0; \ 3662 } \ 3663 } while (0) 3664 3665 #define stat_inc_tot_blk_count(si, blks) \ 3666 ((si)->tot_blks += (blks)) 3667 3668 #define stat_inc_data_blk_count(sbi, blks, gc_type) \ 3669 do { \ 3670 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 3671 stat_inc_tot_blk_count(si, blks); \ 3672 si->data_blks += (blks); \ 3673 si->bg_data_blks += ((gc_type) == BG_GC) ? (blks) : 0; \ 3674 } while (0) 3675 3676 #define stat_inc_node_blk_count(sbi, blks, gc_type) \ 3677 do { \ 3678 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 3679 stat_inc_tot_blk_count(si, blks); \ 3680 si->node_blks += (blks); \ 3681 si->bg_node_blks += ((gc_type) == BG_GC) ? (blks) : 0; \ 3682 } while (0) 3683 3684 int f2fs_build_stats(struct f2fs_sb_info *sbi); 3685 void f2fs_destroy_stats(struct f2fs_sb_info *sbi); 3686 void __init f2fs_create_root_stats(void); 3687 void f2fs_destroy_root_stats(void); 3688 void f2fs_update_sit_info(struct f2fs_sb_info *sbi); 3689 #else 3690 #define stat_inc_cp_count(si) do { } while (0) 3691 #define stat_inc_bg_cp_count(si) do { } while (0) 3692 #define stat_inc_call_count(si) do { } while (0) 3693 #define stat_inc_bggc_count(si) do { } while (0) 3694 #define stat_io_skip_bggc_count(sbi) do { } while (0) 3695 #define stat_other_skip_bggc_count(sbi) do { } while (0) 3696 #define stat_inc_dirty_inode(sbi, type) do { } while (0) 3697 #define stat_dec_dirty_inode(sbi, type) do { } while (0) 3698 #define stat_inc_total_hit(sbi) do { } while (0) 3699 #define stat_inc_rbtree_node_hit(sbi) do { } while (0) 3700 #define stat_inc_largest_node_hit(sbi) do { } while (0) 3701 #define stat_inc_cached_node_hit(sbi) do { } while (0) 3702 #define stat_inc_inline_xattr(inode) do { } while (0) 3703 #define stat_dec_inline_xattr(inode) do { } while (0) 3704 #define stat_inc_inline_inode(inode) do { } while (0) 3705 #define stat_dec_inline_inode(inode) do { } while (0) 3706 #define stat_inc_inline_dir(inode) do { } while (0) 3707 #define stat_dec_inline_dir(inode) do { } while (0) 3708 #define stat_inc_compr_inode(inode) do { } while (0) 3709 #define stat_dec_compr_inode(inode) do { } while (0) 3710 #define stat_add_compr_blocks(inode, blocks) do { } while (0) 3711 #define stat_sub_compr_blocks(inode, blocks) do { } while (0) 3712 #define stat_inc_atomic_write(inode) do { } while (0) 3713 #define stat_dec_atomic_write(inode) do { } while (0) 3714 #define stat_update_max_atomic_write(inode) do { } while (0) 3715 #define stat_inc_volatile_write(inode) do { } while (0) 3716 #define stat_dec_volatile_write(inode) do { } while (0) 3717 #define stat_update_max_volatile_write(inode) do { } while (0) 3718 #define stat_inc_meta_count(sbi, blkaddr) do { } while (0) 3719 #define stat_inc_seg_type(sbi, curseg) do { } while (0) 3720 #define stat_inc_block_count(sbi, curseg) do { } while (0) 3721 #define stat_inc_inplace_blocks(sbi) do { } while (0) 3722 #define stat_inc_seg_count(sbi, type, gc_type) do { } while (0) 3723 #define stat_inc_tot_blk_count(si, blks) do { } while (0) 3724 #define stat_inc_data_blk_count(sbi, blks, gc_type) do { } while (0) 3725 #define stat_inc_node_blk_count(sbi, blks, gc_type) do { } while (0) 3726 3727 static inline int f2fs_build_stats(struct f2fs_sb_info *sbi) { return 0; } 3728 static inline void f2fs_destroy_stats(struct f2fs_sb_info *sbi) { } 3729 static inline void __init f2fs_create_root_stats(void) { } 3730 static inline void f2fs_destroy_root_stats(void) { } 3731 static inline void update_sit_info(struct f2fs_sb_info *sbi) {} 3732 #endif 3733 3734 extern const struct file_operations f2fs_dir_operations; 3735 #ifdef CONFIG_UNICODE 3736 extern const struct dentry_operations f2fs_dentry_ops; 3737 #endif 3738 extern const struct file_operations f2fs_file_operations; 3739 extern const struct inode_operations f2fs_file_inode_operations; 3740 extern const struct address_space_operations f2fs_dblock_aops; 3741 extern const struct address_space_operations f2fs_node_aops; 3742 extern const struct address_space_operations f2fs_meta_aops; 3743 extern const struct inode_operations f2fs_dir_inode_operations; 3744 extern const struct inode_operations f2fs_symlink_inode_operations; 3745 extern const struct inode_operations f2fs_encrypted_symlink_inode_operations; 3746 extern const struct inode_operations f2fs_special_inode_operations; 3747 extern struct kmem_cache *f2fs_inode_entry_slab; 3748 3749 /* 3750 * inline.c 3751 */ 3752 bool f2fs_may_inline_data(struct inode *inode); 3753 bool f2fs_may_inline_dentry(struct inode *inode); 3754 void f2fs_do_read_inline_data(struct page *page, struct page *ipage); 3755 void f2fs_truncate_inline_inode(struct inode *inode, 3756 struct page *ipage, u64 from); 3757 int f2fs_read_inline_data(struct inode *inode, struct page *page); 3758 int f2fs_convert_inline_page(struct dnode_of_data *dn, struct page *page); 3759 int f2fs_convert_inline_inode(struct inode *inode); 3760 int f2fs_try_convert_inline_dir(struct inode *dir, struct dentry *dentry); 3761 int f2fs_write_inline_data(struct inode *inode, struct page *page); 3762 bool f2fs_recover_inline_data(struct inode *inode, struct page *npage); 3763 struct f2fs_dir_entry *f2fs_find_in_inline_dir(struct inode *dir, 3764 const struct f2fs_filename *fname, 3765 struct page **res_page); 3766 int f2fs_make_empty_inline_dir(struct inode *inode, struct inode *parent, 3767 struct page *ipage); 3768 int f2fs_add_inline_entry(struct inode *dir, const struct f2fs_filename *fname, 3769 struct inode *inode, nid_t ino, umode_t mode); 3770 void f2fs_delete_inline_entry(struct f2fs_dir_entry *dentry, 3771 struct page *page, struct inode *dir, 3772 struct inode *inode); 3773 bool f2fs_empty_inline_dir(struct inode *dir); 3774 int f2fs_read_inline_dir(struct file *file, struct dir_context *ctx, 3775 struct fscrypt_str *fstr); 3776 int f2fs_inline_data_fiemap(struct inode *inode, 3777 struct fiemap_extent_info *fieinfo, 3778 __u64 start, __u64 len); 3779 3780 /* 3781 * shrinker.c 3782 */ 3783 unsigned long f2fs_shrink_count(struct shrinker *shrink, 3784 struct shrink_control *sc); 3785 unsigned long f2fs_shrink_scan(struct shrinker *shrink, 3786 struct shrink_control *sc); 3787 void f2fs_join_shrinker(struct f2fs_sb_info *sbi); 3788 void f2fs_leave_shrinker(struct f2fs_sb_info *sbi); 3789 3790 /* 3791 * extent_cache.c 3792 */ 3793 struct rb_entry *f2fs_lookup_rb_tree(struct rb_root_cached *root, 3794 struct rb_entry *cached_re, unsigned int ofs); 3795 struct rb_node **f2fs_lookup_rb_tree_for_insert(struct f2fs_sb_info *sbi, 3796 struct rb_root_cached *root, 3797 struct rb_node **parent, 3798 unsigned int ofs, bool *leftmost); 3799 struct rb_entry *f2fs_lookup_rb_tree_ret(struct rb_root_cached *root, 3800 struct rb_entry *cached_re, unsigned int ofs, 3801 struct rb_entry **prev_entry, struct rb_entry **next_entry, 3802 struct rb_node ***insert_p, struct rb_node **insert_parent, 3803 bool force, bool *leftmost); 3804 bool f2fs_check_rb_tree_consistence(struct f2fs_sb_info *sbi, 3805 struct rb_root_cached *root); 3806 unsigned int f2fs_shrink_extent_tree(struct f2fs_sb_info *sbi, int nr_shrink); 3807 bool f2fs_init_extent_tree(struct inode *inode, struct f2fs_extent *i_ext); 3808 void f2fs_drop_extent_tree(struct inode *inode); 3809 unsigned int f2fs_destroy_extent_node(struct inode *inode); 3810 void f2fs_destroy_extent_tree(struct inode *inode); 3811 bool f2fs_lookup_extent_cache(struct inode *inode, pgoff_t pgofs, 3812 struct extent_info *ei); 3813 void f2fs_update_extent_cache(struct dnode_of_data *dn); 3814 void f2fs_update_extent_cache_range(struct dnode_of_data *dn, 3815 pgoff_t fofs, block_t blkaddr, unsigned int len); 3816 void f2fs_init_extent_cache_info(struct f2fs_sb_info *sbi); 3817 int __init f2fs_create_extent_cache(void); 3818 void f2fs_destroy_extent_cache(void); 3819 3820 /* 3821 * sysfs.c 3822 */ 3823 int __init f2fs_init_sysfs(void); 3824 void f2fs_exit_sysfs(void); 3825 int f2fs_register_sysfs(struct f2fs_sb_info *sbi); 3826 void f2fs_unregister_sysfs(struct f2fs_sb_info *sbi); 3827 3828 /* verity.c */ 3829 extern const struct fsverity_operations f2fs_verityops; 3830 3831 /* 3832 * crypto support 3833 */ 3834 static inline bool f2fs_encrypted_file(struct inode *inode) 3835 { 3836 return IS_ENCRYPTED(inode) && S_ISREG(inode->i_mode); 3837 } 3838 3839 static inline void f2fs_set_encrypted_inode(struct inode *inode) 3840 { 3841 #ifdef CONFIG_FS_ENCRYPTION 3842 file_set_encrypt(inode); 3843 f2fs_set_inode_flags(inode); 3844 #endif 3845 } 3846 3847 /* 3848 * Returns true if the reads of the inode's data need to undergo some 3849 * postprocessing step, like decryption or authenticity verification. 3850 */ 3851 static inline bool f2fs_post_read_required(struct inode *inode) 3852 { 3853 return f2fs_encrypted_file(inode) || fsverity_active(inode) || 3854 f2fs_compressed_file(inode); 3855 } 3856 3857 /* 3858 * compress.c 3859 */ 3860 #ifdef CONFIG_F2FS_FS_COMPRESSION 3861 bool f2fs_is_compressed_page(struct page *page); 3862 struct page *f2fs_compress_control_page(struct page *page); 3863 int f2fs_prepare_compress_overwrite(struct inode *inode, 3864 struct page **pagep, pgoff_t index, void **fsdata); 3865 bool f2fs_compress_write_end(struct inode *inode, void *fsdata, 3866 pgoff_t index, unsigned copied); 3867 int f2fs_truncate_partial_cluster(struct inode *inode, u64 from, bool lock); 3868 void f2fs_compress_write_end_io(struct bio *bio, struct page *page); 3869 bool f2fs_is_compress_backend_ready(struct inode *inode); 3870 int f2fs_init_compress_mempool(void); 3871 void f2fs_destroy_compress_mempool(void); 3872 void f2fs_decompress_pages(struct bio *bio, struct page *page, bool verity); 3873 bool f2fs_cluster_is_empty(struct compress_ctx *cc); 3874 bool f2fs_cluster_can_merge_page(struct compress_ctx *cc, pgoff_t index); 3875 void f2fs_compress_ctx_add_page(struct compress_ctx *cc, struct page *page); 3876 int f2fs_write_multi_pages(struct compress_ctx *cc, 3877 int *submitted, 3878 struct writeback_control *wbc, 3879 enum iostat_type io_type); 3880 int f2fs_is_compressed_cluster(struct inode *inode, pgoff_t index); 3881 int f2fs_read_multi_pages(struct compress_ctx *cc, struct bio **bio_ret, 3882 unsigned nr_pages, sector_t *last_block_in_bio, 3883 bool is_readahead, bool for_write); 3884 struct decompress_io_ctx *f2fs_alloc_dic(struct compress_ctx *cc); 3885 void f2fs_free_dic(struct decompress_io_ctx *dic); 3886 void f2fs_decompress_end_io(struct page **rpages, 3887 unsigned int cluster_size, bool err, bool verity); 3888 int f2fs_init_compress_ctx(struct compress_ctx *cc); 3889 void f2fs_destroy_compress_ctx(struct compress_ctx *cc); 3890 void f2fs_init_compress_info(struct f2fs_sb_info *sbi); 3891 #else 3892 static inline bool f2fs_is_compressed_page(struct page *page) { return false; } 3893 static inline bool f2fs_is_compress_backend_ready(struct inode *inode) 3894 { 3895 if (!f2fs_compressed_file(inode)) 3896 return true; 3897 /* not support compression */ 3898 return false; 3899 } 3900 static inline struct page *f2fs_compress_control_page(struct page *page) 3901 { 3902 WARN_ON_ONCE(1); 3903 return ERR_PTR(-EINVAL); 3904 } 3905 static inline int f2fs_init_compress_mempool(void) { return 0; } 3906 static inline void f2fs_destroy_compress_mempool(void) { } 3907 #endif 3908 3909 static inline void set_compress_context(struct inode *inode) 3910 { 3911 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 3912 3913 F2FS_I(inode)->i_compress_algorithm = 3914 F2FS_OPTION(sbi).compress_algorithm; 3915 F2FS_I(inode)->i_log_cluster_size = 3916 F2FS_OPTION(sbi).compress_log_size; 3917 F2FS_I(inode)->i_cluster_size = 3918 1 << F2FS_I(inode)->i_log_cluster_size; 3919 F2FS_I(inode)->i_flags |= F2FS_COMPR_FL; 3920 set_inode_flag(inode, FI_COMPRESSED_FILE); 3921 stat_inc_compr_inode(inode); 3922 f2fs_mark_inode_dirty_sync(inode, true); 3923 } 3924 3925 static inline u64 f2fs_disable_compressed_file(struct inode *inode) 3926 { 3927 struct f2fs_inode_info *fi = F2FS_I(inode); 3928 3929 if (!f2fs_compressed_file(inode)) 3930 return 0; 3931 if (S_ISREG(inode->i_mode)) { 3932 if (get_dirty_pages(inode)) 3933 return 1; 3934 if (fi->i_compr_blocks) 3935 return fi->i_compr_blocks; 3936 } 3937 3938 fi->i_flags &= ~F2FS_COMPR_FL; 3939 stat_dec_compr_inode(inode); 3940 clear_inode_flag(inode, FI_COMPRESSED_FILE); 3941 f2fs_mark_inode_dirty_sync(inode, true); 3942 return 0; 3943 } 3944 3945 #define F2FS_FEATURE_FUNCS(name, flagname) \ 3946 static inline int f2fs_sb_has_##name(struct f2fs_sb_info *sbi) \ 3947 { \ 3948 return F2FS_HAS_FEATURE(sbi, F2FS_FEATURE_##flagname); \ 3949 } 3950 3951 F2FS_FEATURE_FUNCS(encrypt, ENCRYPT); 3952 F2FS_FEATURE_FUNCS(blkzoned, BLKZONED); 3953 F2FS_FEATURE_FUNCS(extra_attr, EXTRA_ATTR); 3954 F2FS_FEATURE_FUNCS(project_quota, PRJQUOTA); 3955 F2FS_FEATURE_FUNCS(inode_chksum, INODE_CHKSUM); 3956 F2FS_FEATURE_FUNCS(flexible_inline_xattr, FLEXIBLE_INLINE_XATTR); 3957 F2FS_FEATURE_FUNCS(quota_ino, QUOTA_INO); 3958 F2FS_FEATURE_FUNCS(inode_crtime, INODE_CRTIME); 3959 F2FS_FEATURE_FUNCS(lost_found, LOST_FOUND); 3960 F2FS_FEATURE_FUNCS(verity, VERITY); 3961 F2FS_FEATURE_FUNCS(sb_chksum, SB_CHKSUM); 3962 F2FS_FEATURE_FUNCS(casefold, CASEFOLD); 3963 F2FS_FEATURE_FUNCS(compression, COMPRESSION); 3964 3965 #ifdef CONFIG_BLK_DEV_ZONED 3966 static inline bool f2fs_blkz_is_seq(struct f2fs_sb_info *sbi, int devi, 3967 block_t blkaddr) 3968 { 3969 unsigned int zno = blkaddr >> sbi->log_blocks_per_blkz; 3970 3971 return test_bit(zno, FDEV(devi).blkz_seq); 3972 } 3973 #endif 3974 3975 static inline bool f2fs_hw_should_discard(struct f2fs_sb_info *sbi) 3976 { 3977 return f2fs_sb_has_blkzoned(sbi); 3978 } 3979 3980 static inline bool f2fs_bdev_support_discard(struct block_device *bdev) 3981 { 3982 return blk_queue_discard(bdev_get_queue(bdev)) || 3983 bdev_is_zoned(bdev); 3984 } 3985 3986 static inline bool f2fs_hw_support_discard(struct f2fs_sb_info *sbi) 3987 { 3988 int i; 3989 3990 if (!f2fs_is_multi_device(sbi)) 3991 return f2fs_bdev_support_discard(sbi->sb->s_bdev); 3992 3993 for (i = 0; i < sbi->s_ndevs; i++) 3994 if (f2fs_bdev_support_discard(FDEV(i).bdev)) 3995 return true; 3996 return false; 3997 } 3998 3999 static inline bool f2fs_realtime_discard_enable(struct f2fs_sb_info *sbi) 4000 { 4001 return (test_opt(sbi, DISCARD) && f2fs_hw_support_discard(sbi)) || 4002 f2fs_hw_should_discard(sbi); 4003 } 4004 4005 static inline bool f2fs_hw_is_readonly(struct f2fs_sb_info *sbi) 4006 { 4007 int i; 4008 4009 if (!f2fs_is_multi_device(sbi)) 4010 return bdev_read_only(sbi->sb->s_bdev); 4011 4012 for (i = 0; i < sbi->s_ndevs; i++) 4013 if (bdev_read_only(FDEV(i).bdev)) 4014 return true; 4015 return false; 4016 } 4017 4018 static inline bool f2fs_lfs_mode(struct f2fs_sb_info *sbi) 4019 { 4020 return F2FS_OPTION(sbi).fs_mode == FS_MODE_LFS; 4021 } 4022 4023 static inline bool f2fs_may_encrypt(struct inode *dir, struct inode *inode) 4024 { 4025 #ifdef CONFIG_FS_ENCRYPTION 4026 struct f2fs_sb_info *sbi = F2FS_I_SB(dir); 4027 umode_t mode = inode->i_mode; 4028 4029 /* 4030 * If the directory encrypted or dummy encryption enabled, 4031 * then we should encrypt the inode. 4032 */ 4033 if (IS_ENCRYPTED(dir) || DUMMY_ENCRYPTION_ENABLED(sbi)) 4034 return (S_ISREG(mode) || S_ISDIR(mode) || S_ISLNK(mode)); 4035 #endif 4036 return false; 4037 } 4038 4039 static inline bool f2fs_may_compress(struct inode *inode) 4040 { 4041 if (IS_SWAPFILE(inode) || f2fs_is_pinned_file(inode) || 4042 f2fs_is_atomic_file(inode) || 4043 f2fs_is_volatile_file(inode)) 4044 return false; 4045 return S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode); 4046 } 4047 4048 static inline void f2fs_i_compr_blocks_update(struct inode *inode, 4049 u64 blocks, bool add) 4050 { 4051 int diff = F2FS_I(inode)->i_cluster_size - blocks; 4052 4053 /* don't update i_compr_blocks if saved blocks were released */ 4054 if (!add && !F2FS_I(inode)->i_compr_blocks) 4055 return; 4056 4057 if (add) { 4058 F2FS_I(inode)->i_compr_blocks += diff; 4059 stat_add_compr_blocks(inode, diff); 4060 } else { 4061 F2FS_I(inode)->i_compr_blocks -= diff; 4062 stat_sub_compr_blocks(inode, diff); 4063 } 4064 f2fs_mark_inode_dirty_sync(inode, true); 4065 } 4066 4067 static inline int block_unaligned_IO(struct inode *inode, 4068 struct kiocb *iocb, struct iov_iter *iter) 4069 { 4070 unsigned int i_blkbits = READ_ONCE(inode->i_blkbits); 4071 unsigned int blocksize_mask = (1 << i_blkbits) - 1; 4072 loff_t offset = iocb->ki_pos; 4073 unsigned long align = offset | iov_iter_alignment(iter); 4074 4075 return align & blocksize_mask; 4076 } 4077 4078 static inline int allow_outplace_dio(struct inode *inode, 4079 struct kiocb *iocb, struct iov_iter *iter) 4080 { 4081 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 4082 int rw = iov_iter_rw(iter); 4083 4084 return (f2fs_lfs_mode(sbi) && (rw == WRITE) && 4085 !block_unaligned_IO(inode, iocb, iter)); 4086 } 4087 4088 static inline bool f2fs_force_buffered_io(struct inode *inode, 4089 struct kiocb *iocb, struct iov_iter *iter) 4090 { 4091 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 4092 int rw = iov_iter_rw(iter); 4093 4094 if (f2fs_post_read_required(inode)) 4095 return true; 4096 if (f2fs_is_multi_device(sbi)) 4097 return true; 4098 /* 4099 * for blkzoned device, fallback direct IO to buffered IO, so 4100 * all IOs can be serialized by log-structured write. 4101 */ 4102 if (f2fs_sb_has_blkzoned(sbi)) 4103 return true; 4104 if (f2fs_lfs_mode(sbi) && (rw == WRITE)) { 4105 if (block_unaligned_IO(inode, iocb, iter)) 4106 return true; 4107 if (F2FS_IO_ALIGNED(sbi)) 4108 return true; 4109 } 4110 if (is_sbi_flag_set(F2FS_I_SB(inode), SBI_CP_DISABLED) && 4111 !IS_SWAPFILE(inode)) 4112 return true; 4113 4114 return false; 4115 } 4116 4117 #ifdef CONFIG_F2FS_FAULT_INJECTION 4118 extern void f2fs_build_fault_attr(struct f2fs_sb_info *sbi, unsigned int rate, 4119 unsigned int type); 4120 #else 4121 #define f2fs_build_fault_attr(sbi, rate, type) do { } while (0) 4122 #endif 4123 4124 static inline bool is_journalled_quota(struct f2fs_sb_info *sbi) 4125 { 4126 #ifdef CONFIG_QUOTA 4127 if (f2fs_sb_has_quota_ino(sbi)) 4128 return true; 4129 if (F2FS_OPTION(sbi).s_qf_names[USRQUOTA] || 4130 F2FS_OPTION(sbi).s_qf_names[GRPQUOTA] || 4131 F2FS_OPTION(sbi).s_qf_names[PRJQUOTA]) 4132 return true; 4133 #endif 4134 return false; 4135 } 4136 4137 #define EFSBADCRC EBADMSG /* Bad CRC detected */ 4138 #define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */ 4139 4140 #endif /* _LINUX_F2FS_H */ 4141