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_perc; /* percentage for cap */ 143 block_t unusable_cap; /* Amount of space allowed to be 144 * unusable when disabling checkpoint 145 */ 146 147 /* For compression */ 148 unsigned char compress_algorithm; /* algorithm type */ 149 unsigned compress_log_size; /* cluster log size */ 150 unsigned char compress_ext_cnt; /* extension count */ 151 unsigned char extensions[COMPRESS_EXT_NUM][F2FS_EXTENSION_LEN]; /* extensions */ 152 }; 153 154 #define F2FS_FEATURE_ENCRYPT 0x0001 155 #define F2FS_FEATURE_BLKZONED 0x0002 156 #define F2FS_FEATURE_ATOMIC_WRITE 0x0004 157 #define F2FS_FEATURE_EXTRA_ATTR 0x0008 158 #define F2FS_FEATURE_PRJQUOTA 0x0010 159 #define F2FS_FEATURE_INODE_CHKSUM 0x0020 160 #define F2FS_FEATURE_FLEXIBLE_INLINE_XATTR 0x0040 161 #define F2FS_FEATURE_QUOTA_INO 0x0080 162 #define F2FS_FEATURE_INODE_CRTIME 0x0100 163 #define F2FS_FEATURE_LOST_FOUND 0x0200 164 #define F2FS_FEATURE_VERITY 0x0400 165 #define F2FS_FEATURE_SB_CHKSUM 0x0800 166 #define F2FS_FEATURE_CASEFOLD 0x1000 167 #define F2FS_FEATURE_COMPRESSION 0x2000 168 169 #define __F2FS_HAS_FEATURE(raw_super, mask) \ 170 ((raw_super->feature & cpu_to_le32(mask)) != 0) 171 #define F2FS_HAS_FEATURE(sbi, mask) __F2FS_HAS_FEATURE(sbi->raw_super, mask) 172 #define F2FS_SET_FEATURE(sbi, mask) \ 173 (sbi->raw_super->feature |= cpu_to_le32(mask)) 174 #define F2FS_CLEAR_FEATURE(sbi, mask) \ 175 (sbi->raw_super->feature &= ~cpu_to_le32(mask)) 176 177 /* 178 * Default values for user and/or group using reserved blocks 179 */ 180 #define F2FS_DEF_RESUID 0 181 #define F2FS_DEF_RESGID 0 182 183 /* 184 * For checkpoint manager 185 */ 186 enum { 187 NAT_BITMAP, 188 SIT_BITMAP 189 }; 190 191 #define CP_UMOUNT 0x00000001 192 #define CP_FASTBOOT 0x00000002 193 #define CP_SYNC 0x00000004 194 #define CP_RECOVERY 0x00000008 195 #define CP_DISCARD 0x00000010 196 #define CP_TRIMMED 0x00000020 197 #define CP_PAUSE 0x00000040 198 #define CP_RESIZE 0x00000080 199 200 #define MAX_DISCARD_BLOCKS(sbi) BLKS_PER_SEC(sbi) 201 #define DEF_MAX_DISCARD_REQUEST 8 /* issue 8 discards per round */ 202 #define DEF_MIN_DISCARD_ISSUE_TIME 50 /* 50 ms, if exists */ 203 #define DEF_MID_DISCARD_ISSUE_TIME 500 /* 500 ms, if device busy */ 204 #define DEF_MAX_DISCARD_ISSUE_TIME 60000 /* 60 s, if no candidates */ 205 #define DEF_DISCARD_URGENT_UTIL 80 /* do more discard over 80% */ 206 #define DEF_CP_INTERVAL 60 /* 60 secs */ 207 #define DEF_IDLE_INTERVAL 5 /* 5 secs */ 208 #define DEF_DISABLE_INTERVAL 5 /* 5 secs */ 209 #define DEF_DISABLE_QUICK_INTERVAL 1 /* 1 secs */ 210 #define DEF_UMOUNT_DISCARD_TIMEOUT 5 /* 5 secs */ 211 212 struct cp_control { 213 int reason; 214 __u64 trim_start; 215 __u64 trim_end; 216 __u64 trim_minlen; 217 }; 218 219 /* 220 * indicate meta/data type 221 */ 222 enum { 223 META_CP, 224 META_NAT, 225 META_SIT, 226 META_SSA, 227 META_MAX, 228 META_POR, 229 DATA_GENERIC, /* check range only */ 230 DATA_GENERIC_ENHANCE, /* strong check on range and segment bitmap */ 231 DATA_GENERIC_ENHANCE_READ, /* 232 * strong check on range and segment 233 * bitmap but no warning due to race 234 * condition of read on truncated area 235 * by extent_cache 236 */ 237 META_GENERIC, 238 }; 239 240 /* for the list of ino */ 241 enum { 242 ORPHAN_INO, /* for orphan ino list */ 243 APPEND_INO, /* for append ino list */ 244 UPDATE_INO, /* for update ino list */ 245 TRANS_DIR_INO, /* for trasactions dir ino list */ 246 FLUSH_INO, /* for multiple device flushing */ 247 MAX_INO_ENTRY, /* max. list */ 248 }; 249 250 struct ino_entry { 251 struct list_head list; /* list head */ 252 nid_t ino; /* inode number */ 253 unsigned int dirty_device; /* dirty device bitmap */ 254 }; 255 256 /* for the list of inodes to be GCed */ 257 struct inode_entry { 258 struct list_head list; /* list head */ 259 struct inode *inode; /* vfs inode pointer */ 260 }; 261 262 struct fsync_node_entry { 263 struct list_head list; /* list head */ 264 struct page *page; /* warm node page pointer */ 265 unsigned int seq_id; /* sequence id */ 266 }; 267 268 /* for the bitmap indicate blocks to be discarded */ 269 struct discard_entry { 270 struct list_head list; /* list head */ 271 block_t start_blkaddr; /* start blockaddr of current segment */ 272 unsigned char discard_map[SIT_VBLOCK_MAP_SIZE]; /* segment discard bitmap */ 273 }; 274 275 /* default discard granularity of inner discard thread, unit: block count */ 276 #define DEFAULT_DISCARD_GRANULARITY 16 277 278 /* max discard pend list number */ 279 #define MAX_PLIST_NUM 512 280 #define plist_idx(blk_num) ((blk_num) >= MAX_PLIST_NUM ? \ 281 (MAX_PLIST_NUM - 1) : ((blk_num) - 1)) 282 283 enum { 284 D_PREP, /* initial */ 285 D_PARTIAL, /* partially submitted */ 286 D_SUBMIT, /* all submitted */ 287 D_DONE, /* finished */ 288 }; 289 290 struct discard_info { 291 block_t lstart; /* logical start address */ 292 block_t len; /* length */ 293 block_t start; /* actual start address in dev */ 294 }; 295 296 struct discard_cmd { 297 struct rb_node rb_node; /* rb node located in rb-tree */ 298 union { 299 struct { 300 block_t lstart; /* logical start address */ 301 block_t len; /* length */ 302 block_t start; /* actual start address in dev */ 303 }; 304 struct discard_info di; /* discard info */ 305 306 }; 307 struct list_head list; /* command list */ 308 struct completion wait; /* compleation */ 309 struct block_device *bdev; /* bdev */ 310 unsigned short ref; /* reference count */ 311 unsigned char state; /* state */ 312 unsigned char queued; /* queued discard */ 313 int error; /* bio error */ 314 spinlock_t lock; /* for state/bio_ref updating */ 315 unsigned short bio_ref; /* bio reference count */ 316 }; 317 318 enum { 319 DPOLICY_BG, 320 DPOLICY_FORCE, 321 DPOLICY_FSTRIM, 322 DPOLICY_UMOUNT, 323 MAX_DPOLICY, 324 }; 325 326 struct discard_policy { 327 int type; /* type of discard */ 328 unsigned int min_interval; /* used for candidates exist */ 329 unsigned int mid_interval; /* used for device busy */ 330 unsigned int max_interval; /* used for candidates not exist */ 331 unsigned int max_requests; /* # of discards issued per round */ 332 unsigned int io_aware_gran; /* minimum granularity discard not be aware of I/O */ 333 bool io_aware; /* issue discard in idle time */ 334 bool sync; /* submit discard with REQ_SYNC flag */ 335 bool ordered; /* issue discard by lba order */ 336 bool timeout; /* discard timeout for put_super */ 337 unsigned int granularity; /* discard granularity */ 338 }; 339 340 struct discard_cmd_control { 341 struct task_struct *f2fs_issue_discard; /* discard thread */ 342 struct list_head entry_list; /* 4KB discard entry list */ 343 struct list_head pend_list[MAX_PLIST_NUM];/* store pending entries */ 344 struct list_head wait_list; /* store on-flushing entries */ 345 struct list_head fstrim_list; /* in-flight discard from fstrim */ 346 wait_queue_head_t discard_wait_queue; /* waiting queue for wake-up */ 347 unsigned int discard_wake; /* to wake up discard thread */ 348 struct mutex cmd_lock; 349 unsigned int nr_discards; /* # of discards in the list */ 350 unsigned int max_discards; /* max. discards to be issued */ 351 unsigned int discard_granularity; /* discard granularity */ 352 unsigned int undiscard_blks; /* # of undiscard blocks */ 353 unsigned int next_pos; /* next discard position */ 354 atomic_t issued_discard; /* # of issued discard */ 355 atomic_t queued_discard; /* # of queued discard */ 356 atomic_t discard_cmd_cnt; /* # of cached cmd count */ 357 struct rb_root_cached root; /* root of discard rb-tree */ 358 bool rbtree_check; /* config for consistence check */ 359 }; 360 361 /* for the list of fsync inodes, used only during recovery */ 362 struct fsync_inode_entry { 363 struct list_head list; /* list head */ 364 struct inode *inode; /* vfs inode pointer */ 365 block_t blkaddr; /* block address locating the last fsync */ 366 block_t last_dentry; /* block address locating the last dentry */ 367 }; 368 369 #define nats_in_cursum(jnl) (le16_to_cpu((jnl)->n_nats)) 370 #define sits_in_cursum(jnl) (le16_to_cpu((jnl)->n_sits)) 371 372 #define nat_in_journal(jnl, i) ((jnl)->nat_j.entries[i].ne) 373 #define nid_in_journal(jnl, i) ((jnl)->nat_j.entries[i].nid) 374 #define sit_in_journal(jnl, i) ((jnl)->sit_j.entries[i].se) 375 #define segno_in_journal(jnl, i) ((jnl)->sit_j.entries[i].segno) 376 377 #define MAX_NAT_JENTRIES(jnl) (NAT_JOURNAL_ENTRIES - nats_in_cursum(jnl)) 378 #define MAX_SIT_JENTRIES(jnl) (SIT_JOURNAL_ENTRIES - sits_in_cursum(jnl)) 379 380 static inline int update_nats_in_cursum(struct f2fs_journal *journal, int i) 381 { 382 int before = nats_in_cursum(journal); 383 384 journal->n_nats = cpu_to_le16(before + i); 385 return before; 386 } 387 388 static inline int update_sits_in_cursum(struct f2fs_journal *journal, int i) 389 { 390 int before = sits_in_cursum(journal); 391 392 journal->n_sits = cpu_to_le16(before + i); 393 return before; 394 } 395 396 static inline bool __has_cursum_space(struct f2fs_journal *journal, 397 int size, int type) 398 { 399 if (type == NAT_JOURNAL) 400 return size <= MAX_NAT_JENTRIES(journal); 401 return size <= MAX_SIT_JENTRIES(journal); 402 } 403 404 /* 405 * ioctl commands 406 */ 407 #define F2FS_IOC_GETFLAGS FS_IOC_GETFLAGS 408 #define F2FS_IOC_SETFLAGS FS_IOC_SETFLAGS 409 #define F2FS_IOC_GETVERSION FS_IOC_GETVERSION 410 411 #define F2FS_IOCTL_MAGIC 0xf5 412 #define F2FS_IOC_START_ATOMIC_WRITE _IO(F2FS_IOCTL_MAGIC, 1) 413 #define F2FS_IOC_COMMIT_ATOMIC_WRITE _IO(F2FS_IOCTL_MAGIC, 2) 414 #define F2FS_IOC_START_VOLATILE_WRITE _IO(F2FS_IOCTL_MAGIC, 3) 415 #define F2FS_IOC_RELEASE_VOLATILE_WRITE _IO(F2FS_IOCTL_MAGIC, 4) 416 #define F2FS_IOC_ABORT_VOLATILE_WRITE _IO(F2FS_IOCTL_MAGIC, 5) 417 #define F2FS_IOC_GARBAGE_COLLECT _IOW(F2FS_IOCTL_MAGIC, 6, __u32) 418 #define F2FS_IOC_WRITE_CHECKPOINT _IO(F2FS_IOCTL_MAGIC, 7) 419 #define F2FS_IOC_DEFRAGMENT _IOWR(F2FS_IOCTL_MAGIC, 8, \ 420 struct f2fs_defragment) 421 #define F2FS_IOC_MOVE_RANGE _IOWR(F2FS_IOCTL_MAGIC, 9, \ 422 struct f2fs_move_range) 423 #define F2FS_IOC_FLUSH_DEVICE _IOW(F2FS_IOCTL_MAGIC, 10, \ 424 struct f2fs_flush_device) 425 #define F2FS_IOC_GARBAGE_COLLECT_RANGE _IOW(F2FS_IOCTL_MAGIC, 11, \ 426 struct f2fs_gc_range) 427 #define F2FS_IOC_GET_FEATURES _IOR(F2FS_IOCTL_MAGIC, 12, __u32) 428 #define F2FS_IOC_SET_PIN_FILE _IOW(F2FS_IOCTL_MAGIC, 13, __u32) 429 #define F2FS_IOC_GET_PIN_FILE _IOR(F2FS_IOCTL_MAGIC, 14, __u32) 430 #define F2FS_IOC_PRECACHE_EXTENTS _IO(F2FS_IOCTL_MAGIC, 15) 431 #define F2FS_IOC_RESIZE_FS _IOW(F2FS_IOCTL_MAGIC, 16, __u64) 432 #define F2FS_IOC_GET_COMPRESS_BLOCKS _IOR(F2FS_IOCTL_MAGIC, 17, __u64) 433 #define F2FS_IOC_RELEASE_COMPRESS_BLOCKS \ 434 _IOR(F2FS_IOCTL_MAGIC, 18, __u64) 435 #define F2FS_IOC_RESERVE_COMPRESS_BLOCKS \ 436 _IOR(F2FS_IOCTL_MAGIC, 19, __u64) 437 438 #define F2FS_IOC_GET_VOLUME_NAME FS_IOC_GETFSLABEL 439 #define F2FS_IOC_SET_VOLUME_NAME FS_IOC_SETFSLABEL 440 441 #define F2FS_IOC_SET_ENCRYPTION_POLICY FS_IOC_SET_ENCRYPTION_POLICY 442 #define F2FS_IOC_GET_ENCRYPTION_POLICY FS_IOC_GET_ENCRYPTION_POLICY 443 #define F2FS_IOC_GET_ENCRYPTION_PWSALT FS_IOC_GET_ENCRYPTION_PWSALT 444 445 /* 446 * should be same as XFS_IOC_GOINGDOWN. 447 * Flags for going down operation used by FS_IOC_GOINGDOWN 448 */ 449 #define F2FS_IOC_SHUTDOWN _IOR('X', 125, __u32) /* Shutdown */ 450 #define F2FS_GOING_DOWN_FULLSYNC 0x0 /* going down with full sync */ 451 #define F2FS_GOING_DOWN_METASYNC 0x1 /* going down with metadata */ 452 #define F2FS_GOING_DOWN_NOSYNC 0x2 /* going down */ 453 #define F2FS_GOING_DOWN_METAFLUSH 0x3 /* going down with meta flush */ 454 #define F2FS_GOING_DOWN_NEED_FSCK 0x4 /* going down to trigger fsck */ 455 456 #if defined(__KERNEL__) && defined(CONFIG_COMPAT) 457 /* 458 * ioctl commands in 32 bit emulation 459 */ 460 #define F2FS_IOC32_GETFLAGS FS_IOC32_GETFLAGS 461 #define F2FS_IOC32_SETFLAGS FS_IOC32_SETFLAGS 462 #define F2FS_IOC32_GETVERSION FS_IOC32_GETVERSION 463 #endif 464 465 #define F2FS_IOC_FSGETXATTR FS_IOC_FSGETXATTR 466 #define F2FS_IOC_FSSETXATTR FS_IOC_FSSETXATTR 467 468 struct f2fs_gc_range { 469 u32 sync; 470 u64 start; 471 u64 len; 472 }; 473 474 struct f2fs_defragment { 475 u64 start; 476 u64 len; 477 }; 478 479 struct f2fs_move_range { 480 u32 dst_fd; /* destination fd */ 481 u64 pos_in; /* start position in src_fd */ 482 u64 pos_out; /* start position in dst_fd */ 483 u64 len; /* size to move */ 484 }; 485 486 struct f2fs_flush_device { 487 u32 dev_num; /* device number to flush */ 488 u32 segments; /* # of segments to flush */ 489 }; 490 491 /* for inline stuff */ 492 #define DEF_INLINE_RESERVED_SIZE 1 493 static inline int get_extra_isize(struct inode *inode); 494 static inline int get_inline_xattr_addrs(struct inode *inode); 495 #define MAX_INLINE_DATA(inode) (sizeof(__le32) * \ 496 (CUR_ADDRS_PER_INODE(inode) - \ 497 get_inline_xattr_addrs(inode) - \ 498 DEF_INLINE_RESERVED_SIZE)) 499 500 /* for inline dir */ 501 #define NR_INLINE_DENTRY(inode) (MAX_INLINE_DATA(inode) * BITS_PER_BYTE / \ 502 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \ 503 BITS_PER_BYTE + 1)) 504 #define INLINE_DENTRY_BITMAP_SIZE(inode) \ 505 DIV_ROUND_UP(NR_INLINE_DENTRY(inode), BITS_PER_BYTE) 506 #define INLINE_RESERVED_SIZE(inode) (MAX_INLINE_DATA(inode) - \ 507 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \ 508 NR_INLINE_DENTRY(inode) + \ 509 INLINE_DENTRY_BITMAP_SIZE(inode))) 510 511 /* 512 * For INODE and NODE manager 513 */ 514 /* for directory operations */ 515 516 struct f2fs_filename { 517 /* 518 * The filename the user specified. This is NULL for some 519 * filesystem-internal operations, e.g. converting an inline directory 520 * to a non-inline one, or roll-forward recovering an encrypted dentry. 521 */ 522 const struct qstr *usr_fname; 523 524 /* 525 * The on-disk filename. For encrypted directories, this is encrypted. 526 * This may be NULL for lookups in an encrypted dir without the key. 527 */ 528 struct fscrypt_str disk_name; 529 530 /* The dirhash of this filename */ 531 f2fs_hash_t hash; 532 533 #ifdef CONFIG_FS_ENCRYPTION 534 /* 535 * For lookups in encrypted directories: either the buffer backing 536 * disk_name, or a buffer that holds the decoded no-key name. 537 */ 538 struct fscrypt_str crypto_buf; 539 #endif 540 #ifdef CONFIG_UNICODE 541 /* 542 * For casefolded directories: the casefolded name, but it's left NULL 543 * if the original name is not valid Unicode or if the filesystem is 544 * doing an internal operation where usr_fname is also NULL. In these 545 * cases we fall back to treating the name as an opaque byte sequence. 546 */ 547 struct fscrypt_str cf_name; 548 #endif 549 }; 550 551 struct f2fs_dentry_ptr { 552 struct inode *inode; 553 void *bitmap; 554 struct f2fs_dir_entry *dentry; 555 __u8 (*filename)[F2FS_SLOT_LEN]; 556 int max; 557 int nr_bitmap; 558 }; 559 560 static inline void make_dentry_ptr_block(struct inode *inode, 561 struct f2fs_dentry_ptr *d, struct f2fs_dentry_block *t) 562 { 563 d->inode = inode; 564 d->max = NR_DENTRY_IN_BLOCK; 565 d->nr_bitmap = SIZE_OF_DENTRY_BITMAP; 566 d->bitmap = t->dentry_bitmap; 567 d->dentry = t->dentry; 568 d->filename = t->filename; 569 } 570 571 static inline void make_dentry_ptr_inline(struct inode *inode, 572 struct f2fs_dentry_ptr *d, void *t) 573 { 574 int entry_cnt = NR_INLINE_DENTRY(inode); 575 int bitmap_size = INLINE_DENTRY_BITMAP_SIZE(inode); 576 int reserved_size = INLINE_RESERVED_SIZE(inode); 577 578 d->inode = inode; 579 d->max = entry_cnt; 580 d->nr_bitmap = bitmap_size; 581 d->bitmap = t; 582 d->dentry = t + bitmap_size + reserved_size; 583 d->filename = t + bitmap_size + reserved_size + 584 SIZE_OF_DIR_ENTRY * entry_cnt; 585 } 586 587 /* 588 * XATTR_NODE_OFFSET stores xattrs to one node block per file keeping -1 589 * as its node offset to distinguish from index node blocks. 590 * But some bits are used to mark the node block. 591 */ 592 #define XATTR_NODE_OFFSET ((((unsigned int)-1) << OFFSET_BIT_SHIFT) \ 593 >> OFFSET_BIT_SHIFT) 594 enum { 595 ALLOC_NODE, /* allocate a new node page if needed */ 596 LOOKUP_NODE, /* look up a node without readahead */ 597 LOOKUP_NODE_RA, /* 598 * look up a node with readahead called 599 * by get_data_block. 600 */ 601 }; 602 603 #define DEFAULT_RETRY_IO_COUNT 8 /* maximum retry read IO count */ 604 605 /* congestion wait timeout value, default: 20ms */ 606 #define DEFAULT_IO_TIMEOUT (msecs_to_jiffies(20)) 607 608 /* maximum retry quota flush count */ 609 #define DEFAULT_RETRY_QUOTA_FLUSH_COUNT 8 610 611 #define F2FS_LINK_MAX 0xffffffff /* maximum link count per file */ 612 613 #define MAX_DIR_RA_PAGES 4 /* maximum ra pages of dir */ 614 615 /* for in-memory extent cache entry */ 616 #define F2FS_MIN_EXTENT_LEN 64 /* minimum extent length */ 617 618 /* number of extent info in extent cache we try to shrink */ 619 #define EXTENT_CACHE_SHRINK_NUMBER 128 620 621 struct rb_entry { 622 struct rb_node rb_node; /* rb node located in rb-tree */ 623 unsigned int ofs; /* start offset of the entry */ 624 unsigned int len; /* length of the entry */ 625 }; 626 627 struct extent_info { 628 unsigned int fofs; /* start offset in a file */ 629 unsigned int len; /* length of the extent */ 630 u32 blk; /* start block address of the extent */ 631 }; 632 633 struct extent_node { 634 struct rb_node rb_node; /* rb node located in rb-tree */ 635 struct extent_info ei; /* extent info */ 636 struct list_head list; /* node in global extent list of sbi */ 637 struct extent_tree *et; /* extent tree pointer */ 638 }; 639 640 struct extent_tree { 641 nid_t ino; /* inode number */ 642 struct rb_root_cached root; /* root of extent info rb-tree */ 643 struct extent_node *cached_en; /* recently accessed extent node */ 644 struct extent_info largest; /* largested extent info */ 645 struct list_head list; /* to be used by sbi->zombie_list */ 646 rwlock_t lock; /* protect extent info rb-tree */ 647 atomic_t node_cnt; /* # of extent node in rb-tree*/ 648 bool largest_updated; /* largest extent updated */ 649 }; 650 651 /* 652 * This structure is taken from ext4_map_blocks. 653 * 654 * Note that, however, f2fs uses NEW and MAPPED flags for f2fs_map_blocks(). 655 */ 656 #define F2FS_MAP_NEW (1 << BH_New) 657 #define F2FS_MAP_MAPPED (1 << BH_Mapped) 658 #define F2FS_MAP_UNWRITTEN (1 << BH_Unwritten) 659 #define F2FS_MAP_FLAGS (F2FS_MAP_NEW | F2FS_MAP_MAPPED |\ 660 F2FS_MAP_UNWRITTEN) 661 662 struct f2fs_map_blocks { 663 block_t m_pblk; 664 block_t m_lblk; 665 unsigned int m_len; 666 unsigned int m_flags; 667 pgoff_t *m_next_pgofs; /* point next possible non-hole pgofs */ 668 pgoff_t *m_next_extent; /* point to next possible extent */ 669 int m_seg_type; 670 bool m_may_create; /* indicate it is from write path */ 671 }; 672 673 /* for flag in get_data_block */ 674 enum { 675 F2FS_GET_BLOCK_DEFAULT, 676 F2FS_GET_BLOCK_FIEMAP, 677 F2FS_GET_BLOCK_BMAP, 678 F2FS_GET_BLOCK_DIO, 679 F2FS_GET_BLOCK_PRE_DIO, 680 F2FS_GET_BLOCK_PRE_AIO, 681 F2FS_GET_BLOCK_PRECACHE, 682 }; 683 684 /* 685 * i_advise uses FADVISE_XXX_BIT. We can add additional hints later. 686 */ 687 #define FADVISE_COLD_BIT 0x01 688 #define FADVISE_LOST_PINO_BIT 0x02 689 #define FADVISE_ENCRYPT_BIT 0x04 690 #define FADVISE_ENC_NAME_BIT 0x08 691 #define FADVISE_KEEP_SIZE_BIT 0x10 692 #define FADVISE_HOT_BIT 0x20 693 #define FADVISE_VERITY_BIT 0x40 694 695 #define FADVISE_MODIFIABLE_BITS (FADVISE_COLD_BIT | FADVISE_HOT_BIT) 696 697 #define file_is_cold(inode) is_file(inode, FADVISE_COLD_BIT) 698 #define file_wrong_pino(inode) is_file(inode, FADVISE_LOST_PINO_BIT) 699 #define file_set_cold(inode) set_file(inode, FADVISE_COLD_BIT) 700 #define file_lost_pino(inode) set_file(inode, FADVISE_LOST_PINO_BIT) 701 #define file_clear_cold(inode) clear_file(inode, FADVISE_COLD_BIT) 702 #define file_got_pino(inode) clear_file(inode, FADVISE_LOST_PINO_BIT) 703 #define file_is_encrypt(inode) is_file(inode, FADVISE_ENCRYPT_BIT) 704 #define file_set_encrypt(inode) set_file(inode, FADVISE_ENCRYPT_BIT) 705 #define file_clear_encrypt(inode) clear_file(inode, FADVISE_ENCRYPT_BIT) 706 #define file_enc_name(inode) is_file(inode, FADVISE_ENC_NAME_BIT) 707 #define file_set_enc_name(inode) set_file(inode, FADVISE_ENC_NAME_BIT) 708 #define file_keep_isize(inode) is_file(inode, FADVISE_KEEP_SIZE_BIT) 709 #define file_set_keep_isize(inode) set_file(inode, FADVISE_KEEP_SIZE_BIT) 710 #define file_is_hot(inode) is_file(inode, FADVISE_HOT_BIT) 711 #define file_set_hot(inode) set_file(inode, FADVISE_HOT_BIT) 712 #define file_clear_hot(inode) clear_file(inode, FADVISE_HOT_BIT) 713 #define file_is_verity(inode) is_file(inode, FADVISE_VERITY_BIT) 714 #define file_set_verity(inode) set_file(inode, FADVISE_VERITY_BIT) 715 716 #define DEF_DIR_LEVEL 0 717 718 enum { 719 GC_FAILURE_PIN, 720 GC_FAILURE_ATOMIC, 721 MAX_GC_FAILURE 722 }; 723 724 /* used for f2fs_inode_info->flags */ 725 enum { 726 FI_NEW_INODE, /* indicate newly allocated inode */ 727 FI_DIRTY_INODE, /* indicate inode is dirty or not */ 728 FI_AUTO_RECOVER, /* indicate inode is recoverable */ 729 FI_DIRTY_DIR, /* indicate directory has dirty pages */ 730 FI_INC_LINK, /* need to increment i_nlink */ 731 FI_ACL_MODE, /* indicate acl mode */ 732 FI_NO_ALLOC, /* should not allocate any blocks */ 733 FI_FREE_NID, /* free allocated nide */ 734 FI_NO_EXTENT, /* not to use the extent cache */ 735 FI_INLINE_XATTR, /* used for inline xattr */ 736 FI_INLINE_DATA, /* used for inline data*/ 737 FI_INLINE_DENTRY, /* used for inline dentry */ 738 FI_APPEND_WRITE, /* inode has appended data */ 739 FI_UPDATE_WRITE, /* inode has in-place-update data */ 740 FI_NEED_IPU, /* used for ipu per file */ 741 FI_ATOMIC_FILE, /* indicate atomic file */ 742 FI_ATOMIC_COMMIT, /* indicate the state of atomical committing */ 743 FI_VOLATILE_FILE, /* indicate volatile file */ 744 FI_FIRST_BLOCK_WRITTEN, /* indicate #0 data block was written */ 745 FI_DROP_CACHE, /* drop dirty page cache */ 746 FI_DATA_EXIST, /* indicate data exists */ 747 FI_INLINE_DOTS, /* indicate inline dot dentries */ 748 FI_DO_DEFRAG, /* indicate defragment is running */ 749 FI_DIRTY_FILE, /* indicate regular/symlink has dirty pages */ 750 FI_NO_PREALLOC, /* indicate skipped preallocated blocks */ 751 FI_HOT_DATA, /* indicate file is hot */ 752 FI_EXTRA_ATTR, /* indicate file has extra attribute */ 753 FI_PROJ_INHERIT, /* indicate file inherits projectid */ 754 FI_PIN_FILE, /* indicate file should not be gced */ 755 FI_ATOMIC_REVOKE_REQUEST, /* request to drop atomic data */ 756 FI_VERITY_IN_PROGRESS, /* building fs-verity Merkle tree */ 757 FI_COMPRESSED_FILE, /* indicate file's data can be compressed */ 758 FI_MMAP_FILE, /* indicate file was mmapped */ 759 FI_MAX, /* max flag, never be used */ 760 }; 761 762 struct f2fs_inode_info { 763 struct inode vfs_inode; /* serve a vfs inode */ 764 unsigned long i_flags; /* keep an inode flags for ioctl */ 765 unsigned char i_advise; /* use to give file attribute hints */ 766 unsigned char i_dir_level; /* use for dentry level for large dir */ 767 unsigned int i_current_depth; /* only for directory depth */ 768 /* for gc failure statistic */ 769 unsigned int i_gc_failures[MAX_GC_FAILURE]; 770 unsigned int i_pino; /* parent inode number */ 771 umode_t i_acl_mode; /* keep file acl mode temporarily */ 772 773 /* Use below internally in f2fs*/ 774 unsigned long flags[BITS_TO_LONGS(FI_MAX)]; /* use to pass per-file flags */ 775 struct rw_semaphore i_sem; /* protect fi info */ 776 atomic_t dirty_pages; /* # of dirty pages */ 777 f2fs_hash_t chash; /* hash value of given file name */ 778 unsigned int clevel; /* maximum level of given file name */ 779 struct task_struct *task; /* lookup and create consistency */ 780 struct task_struct *cp_task; /* separate cp/wb IO stats*/ 781 nid_t i_xattr_nid; /* node id that contains xattrs */ 782 loff_t last_disk_size; /* lastly written file size */ 783 spinlock_t i_size_lock; /* protect last_disk_size */ 784 785 #ifdef CONFIG_QUOTA 786 struct dquot *i_dquot[MAXQUOTAS]; 787 788 /* quota space reservation, managed internally by quota code */ 789 qsize_t i_reserved_quota; 790 #endif 791 struct list_head dirty_list; /* dirty list for dirs and files */ 792 struct list_head gdirty_list; /* linked in global dirty list */ 793 struct list_head inmem_ilist; /* list for inmem inodes */ 794 struct list_head inmem_pages; /* inmemory pages managed by f2fs */ 795 struct task_struct *inmem_task; /* store inmemory task */ 796 struct mutex inmem_lock; /* lock for inmemory pages */ 797 struct extent_tree *extent_tree; /* cached extent_tree entry */ 798 799 /* avoid racing between foreground op and gc */ 800 struct rw_semaphore i_gc_rwsem[2]; 801 struct rw_semaphore i_mmap_sem; 802 struct rw_semaphore i_xattr_sem; /* avoid racing between reading and changing EAs */ 803 804 int i_extra_isize; /* size of extra space located in i_addr */ 805 kprojid_t i_projid; /* id for project quota */ 806 int i_inline_xattr_size; /* inline xattr size */ 807 struct timespec64 i_crtime; /* inode creation time */ 808 struct timespec64 i_disk_time[4];/* inode disk times */ 809 810 /* for file compress */ 811 u64 i_compr_blocks; /* # of compressed blocks */ 812 unsigned char i_compress_algorithm; /* algorithm type */ 813 unsigned char i_log_cluster_size; /* log of cluster size */ 814 unsigned int i_cluster_size; /* cluster size */ 815 }; 816 817 static inline void get_extent_info(struct extent_info *ext, 818 struct f2fs_extent *i_ext) 819 { 820 ext->fofs = le32_to_cpu(i_ext->fofs); 821 ext->blk = le32_to_cpu(i_ext->blk); 822 ext->len = le32_to_cpu(i_ext->len); 823 } 824 825 static inline void set_raw_extent(struct extent_info *ext, 826 struct f2fs_extent *i_ext) 827 { 828 i_ext->fofs = cpu_to_le32(ext->fofs); 829 i_ext->blk = cpu_to_le32(ext->blk); 830 i_ext->len = cpu_to_le32(ext->len); 831 } 832 833 static inline void set_extent_info(struct extent_info *ei, unsigned int fofs, 834 u32 blk, unsigned int len) 835 { 836 ei->fofs = fofs; 837 ei->blk = blk; 838 ei->len = len; 839 } 840 841 static inline bool __is_discard_mergeable(struct discard_info *back, 842 struct discard_info *front, unsigned int max_len) 843 { 844 return (back->lstart + back->len == front->lstart) && 845 (back->len + front->len <= max_len); 846 } 847 848 static inline bool __is_discard_back_mergeable(struct discard_info *cur, 849 struct discard_info *back, unsigned int max_len) 850 { 851 return __is_discard_mergeable(back, cur, max_len); 852 } 853 854 static inline bool __is_discard_front_mergeable(struct discard_info *cur, 855 struct discard_info *front, unsigned int max_len) 856 { 857 return __is_discard_mergeable(cur, front, max_len); 858 } 859 860 static inline bool __is_extent_mergeable(struct extent_info *back, 861 struct extent_info *front) 862 { 863 return (back->fofs + back->len == front->fofs && 864 back->blk + back->len == front->blk); 865 } 866 867 static inline bool __is_back_mergeable(struct extent_info *cur, 868 struct extent_info *back) 869 { 870 return __is_extent_mergeable(back, cur); 871 } 872 873 static inline bool __is_front_mergeable(struct extent_info *cur, 874 struct extent_info *front) 875 { 876 return __is_extent_mergeable(cur, front); 877 } 878 879 extern void f2fs_mark_inode_dirty_sync(struct inode *inode, bool sync); 880 static inline void __try_update_largest_extent(struct extent_tree *et, 881 struct extent_node *en) 882 { 883 if (en->ei.len > et->largest.len) { 884 et->largest = en->ei; 885 et->largest_updated = true; 886 } 887 } 888 889 /* 890 * For free nid management 891 */ 892 enum nid_state { 893 FREE_NID, /* newly added to free nid list */ 894 PREALLOC_NID, /* it is preallocated */ 895 MAX_NID_STATE, 896 }; 897 898 struct f2fs_nm_info { 899 block_t nat_blkaddr; /* base disk address of NAT */ 900 nid_t max_nid; /* maximum possible node ids */ 901 nid_t available_nids; /* # of available node ids */ 902 nid_t next_scan_nid; /* the next nid to be scanned */ 903 unsigned int ram_thresh; /* control the memory footprint */ 904 unsigned int ra_nid_pages; /* # of nid pages to be readaheaded */ 905 unsigned int dirty_nats_ratio; /* control dirty nats ratio threshold */ 906 907 /* NAT cache management */ 908 struct radix_tree_root nat_root;/* root of the nat entry cache */ 909 struct radix_tree_root nat_set_root;/* root of the nat set cache */ 910 struct rw_semaphore nat_tree_lock; /* protect nat_tree_lock */ 911 struct list_head nat_entries; /* cached nat entry list (clean) */ 912 spinlock_t nat_list_lock; /* protect clean nat entry list */ 913 unsigned int nat_cnt; /* the # of cached nat entries */ 914 unsigned int dirty_nat_cnt; /* total num of nat entries in set */ 915 unsigned int nat_blocks; /* # of nat blocks */ 916 917 /* free node ids management */ 918 struct radix_tree_root free_nid_root;/* root of the free_nid cache */ 919 struct list_head free_nid_list; /* list for free nids excluding preallocated nids */ 920 unsigned int nid_cnt[MAX_NID_STATE]; /* the number of free node id */ 921 spinlock_t nid_list_lock; /* protect nid lists ops */ 922 struct mutex build_lock; /* lock for build free nids */ 923 unsigned char **free_nid_bitmap; 924 unsigned char *nat_block_bitmap; 925 unsigned short *free_nid_count; /* free nid count of NAT block */ 926 927 /* for checkpoint */ 928 char *nat_bitmap; /* NAT bitmap pointer */ 929 930 unsigned int nat_bits_blocks; /* # of nat bits blocks */ 931 unsigned char *nat_bits; /* NAT bits blocks */ 932 unsigned char *full_nat_bits; /* full NAT pages */ 933 unsigned char *empty_nat_bits; /* empty NAT pages */ 934 #ifdef CONFIG_F2FS_CHECK_FS 935 char *nat_bitmap_mir; /* NAT bitmap mirror */ 936 #endif 937 int bitmap_size; /* bitmap size */ 938 }; 939 940 /* 941 * this structure is used as one of function parameters. 942 * all the information are dedicated to a given direct node block determined 943 * by the data offset in a file. 944 */ 945 struct dnode_of_data { 946 struct inode *inode; /* vfs inode pointer */ 947 struct page *inode_page; /* its inode page, NULL is possible */ 948 struct page *node_page; /* cached direct node page */ 949 nid_t nid; /* node id of the direct node block */ 950 unsigned int ofs_in_node; /* data offset in the node page */ 951 bool inode_page_locked; /* inode page is locked or not */ 952 bool node_changed; /* is node block changed */ 953 char cur_level; /* level of hole node page */ 954 char max_level; /* level of current page located */ 955 block_t data_blkaddr; /* block address of the node block */ 956 }; 957 958 static inline void set_new_dnode(struct dnode_of_data *dn, struct inode *inode, 959 struct page *ipage, struct page *npage, nid_t nid) 960 { 961 memset(dn, 0, sizeof(*dn)); 962 dn->inode = inode; 963 dn->inode_page = ipage; 964 dn->node_page = npage; 965 dn->nid = nid; 966 } 967 968 /* 969 * For SIT manager 970 * 971 * By default, there are 6 active log areas across the whole main area. 972 * When considering hot and cold data separation to reduce cleaning overhead, 973 * we split 3 for data logs and 3 for node logs as hot, warm, and cold types, 974 * respectively. 975 * In the current design, you should not change the numbers intentionally. 976 * Instead, as a mount option such as active_logs=x, you can use 2, 4, and 6 977 * logs individually according to the underlying devices. (default: 6) 978 * Just in case, on-disk layout covers maximum 16 logs that consist of 8 for 979 * data and 8 for node logs. 980 */ 981 #define NR_CURSEG_DATA_TYPE (3) 982 #define NR_CURSEG_NODE_TYPE (3) 983 #define NR_CURSEG_TYPE (NR_CURSEG_DATA_TYPE + NR_CURSEG_NODE_TYPE) 984 985 enum { 986 CURSEG_HOT_DATA = 0, /* directory entry blocks */ 987 CURSEG_WARM_DATA, /* data blocks */ 988 CURSEG_COLD_DATA, /* multimedia or GCed data blocks */ 989 CURSEG_HOT_NODE, /* direct node blocks of directory files */ 990 CURSEG_WARM_NODE, /* direct node blocks of normal files */ 991 CURSEG_COLD_NODE, /* indirect node blocks */ 992 NO_CHECK_TYPE, 993 CURSEG_COLD_DATA_PINNED,/* cold data for pinned file */ 994 }; 995 996 struct flush_cmd { 997 struct completion wait; 998 struct llist_node llnode; 999 nid_t ino; 1000 int ret; 1001 }; 1002 1003 struct flush_cmd_control { 1004 struct task_struct *f2fs_issue_flush; /* flush thread */ 1005 wait_queue_head_t flush_wait_queue; /* waiting queue for wake-up */ 1006 atomic_t issued_flush; /* # of issued flushes */ 1007 atomic_t queued_flush; /* # of queued flushes */ 1008 struct llist_head issue_list; /* list for command issue */ 1009 struct llist_node *dispatch_list; /* list for command dispatch */ 1010 }; 1011 1012 struct f2fs_sm_info { 1013 struct sit_info *sit_info; /* whole segment information */ 1014 struct free_segmap_info *free_info; /* free segment information */ 1015 struct dirty_seglist_info *dirty_info; /* dirty segment information */ 1016 struct curseg_info *curseg_array; /* active segment information */ 1017 1018 struct rw_semaphore curseg_lock; /* for preventing curseg change */ 1019 1020 block_t seg0_blkaddr; /* block address of 0'th segment */ 1021 block_t main_blkaddr; /* start block address of main area */ 1022 block_t ssa_blkaddr; /* start block address of SSA area */ 1023 1024 unsigned int segment_count; /* total # of segments */ 1025 unsigned int main_segments; /* # of segments in main area */ 1026 unsigned int reserved_segments; /* # of reserved segments */ 1027 unsigned int ovp_segments; /* # of overprovision segments */ 1028 1029 /* a threshold to reclaim prefree segments */ 1030 unsigned int rec_prefree_segments; 1031 1032 /* for batched trimming */ 1033 unsigned int trim_sections; /* # of sections to trim */ 1034 1035 struct list_head sit_entry_set; /* sit entry set list */ 1036 1037 unsigned int ipu_policy; /* in-place-update policy */ 1038 unsigned int min_ipu_util; /* in-place-update threshold */ 1039 unsigned int min_fsync_blocks; /* threshold for fsync */ 1040 unsigned int min_seq_blocks; /* threshold for sequential blocks */ 1041 unsigned int min_hot_blocks; /* threshold for hot block allocation */ 1042 unsigned int min_ssr_sections; /* threshold to trigger SSR allocation */ 1043 1044 /* for flush command control */ 1045 struct flush_cmd_control *fcc_info; 1046 1047 /* for discard command control */ 1048 struct discard_cmd_control *dcc_info; 1049 }; 1050 1051 /* 1052 * For superblock 1053 */ 1054 /* 1055 * COUNT_TYPE for monitoring 1056 * 1057 * f2fs monitors the number of several block types such as on-writeback, 1058 * dirty dentry blocks, dirty node blocks, and dirty meta blocks. 1059 */ 1060 #define WB_DATA_TYPE(p) (__is_cp_guaranteed(p) ? F2FS_WB_CP_DATA : F2FS_WB_DATA) 1061 enum count_type { 1062 F2FS_DIRTY_DENTS, 1063 F2FS_DIRTY_DATA, 1064 F2FS_DIRTY_QDATA, 1065 F2FS_DIRTY_NODES, 1066 F2FS_DIRTY_META, 1067 F2FS_INMEM_PAGES, 1068 F2FS_DIRTY_IMETA, 1069 F2FS_WB_CP_DATA, 1070 F2FS_WB_DATA, 1071 F2FS_RD_DATA, 1072 F2FS_RD_NODE, 1073 F2FS_RD_META, 1074 F2FS_DIO_WRITE, 1075 F2FS_DIO_READ, 1076 NR_COUNT_TYPE, 1077 }; 1078 1079 /* 1080 * The below are the page types of bios used in submit_bio(). 1081 * The available types are: 1082 * DATA User data pages. It operates as async mode. 1083 * NODE Node pages. It operates as async mode. 1084 * META FS metadata pages such as SIT, NAT, CP. 1085 * NR_PAGE_TYPE The number of page types. 1086 * META_FLUSH Make sure the previous pages are written 1087 * with waiting the bio's completion 1088 * ... Only can be used with META. 1089 */ 1090 #define PAGE_TYPE_OF_BIO(type) ((type) > META ? META : (type)) 1091 enum page_type { 1092 DATA, 1093 NODE, 1094 META, 1095 NR_PAGE_TYPE, 1096 META_FLUSH, 1097 INMEM, /* the below types are used by tracepoints only. */ 1098 INMEM_DROP, 1099 INMEM_INVALIDATE, 1100 INMEM_REVOKE, 1101 IPU, 1102 OPU, 1103 }; 1104 1105 enum temp_type { 1106 HOT = 0, /* must be zero for meta bio */ 1107 WARM, 1108 COLD, 1109 NR_TEMP_TYPE, 1110 }; 1111 1112 enum need_lock_type { 1113 LOCK_REQ = 0, 1114 LOCK_DONE, 1115 LOCK_RETRY, 1116 }; 1117 1118 enum cp_reason_type { 1119 CP_NO_NEEDED, 1120 CP_NON_REGULAR, 1121 CP_COMPRESSED, 1122 CP_HARDLINK, 1123 CP_SB_NEED_CP, 1124 CP_WRONG_PINO, 1125 CP_NO_SPC_ROLL, 1126 CP_NODE_NEED_CP, 1127 CP_FASTBOOT_MODE, 1128 CP_SPEC_LOG_NUM, 1129 CP_RECOVER_DIR, 1130 }; 1131 1132 enum iostat_type { 1133 /* WRITE IO */ 1134 APP_DIRECT_IO, /* app direct write IOs */ 1135 APP_BUFFERED_IO, /* app buffered write IOs */ 1136 APP_WRITE_IO, /* app write IOs */ 1137 APP_MAPPED_IO, /* app mapped IOs */ 1138 FS_DATA_IO, /* data IOs from kworker/fsync/reclaimer */ 1139 FS_NODE_IO, /* node IOs from kworker/fsync/reclaimer */ 1140 FS_META_IO, /* meta IOs from kworker/reclaimer */ 1141 FS_GC_DATA_IO, /* data IOs from forground gc */ 1142 FS_GC_NODE_IO, /* node IOs from forground gc */ 1143 FS_CP_DATA_IO, /* data IOs from checkpoint */ 1144 FS_CP_NODE_IO, /* node IOs from checkpoint */ 1145 FS_CP_META_IO, /* meta IOs from checkpoint */ 1146 1147 /* READ IO */ 1148 APP_DIRECT_READ_IO, /* app direct read IOs */ 1149 APP_BUFFERED_READ_IO, /* app buffered read IOs */ 1150 APP_READ_IO, /* app read IOs */ 1151 APP_MAPPED_READ_IO, /* app mapped read IOs */ 1152 FS_DATA_READ_IO, /* data read IOs */ 1153 FS_GDATA_READ_IO, /* data read IOs from background gc */ 1154 FS_CDATA_READ_IO, /* compressed data read IOs */ 1155 FS_NODE_READ_IO, /* node read IOs */ 1156 FS_META_READ_IO, /* meta read IOs */ 1157 1158 /* other */ 1159 FS_DISCARD, /* discard */ 1160 NR_IO_TYPE, 1161 }; 1162 1163 struct f2fs_io_info { 1164 struct f2fs_sb_info *sbi; /* f2fs_sb_info pointer */ 1165 nid_t ino; /* inode number */ 1166 enum page_type type; /* contains DATA/NODE/META/META_FLUSH */ 1167 enum temp_type temp; /* contains HOT/WARM/COLD */ 1168 int op; /* contains REQ_OP_ */ 1169 int op_flags; /* req_flag_bits */ 1170 block_t new_blkaddr; /* new block address to be written */ 1171 block_t old_blkaddr; /* old block address before Cow */ 1172 struct page *page; /* page to be written */ 1173 struct page *encrypted_page; /* encrypted page */ 1174 struct page *compressed_page; /* compressed page */ 1175 struct list_head list; /* serialize IOs */ 1176 bool submitted; /* indicate IO submission */ 1177 int need_lock; /* indicate we need to lock cp_rwsem */ 1178 bool in_list; /* indicate fio is in io_list */ 1179 bool is_por; /* indicate IO is from recovery or not */ 1180 bool retry; /* need to reallocate block address */ 1181 int compr_blocks; /* # of compressed block addresses */ 1182 bool encrypted; /* indicate file is encrypted */ 1183 enum iostat_type io_type; /* io type */ 1184 struct writeback_control *io_wbc; /* writeback control */ 1185 struct bio **bio; /* bio for ipu */ 1186 sector_t *last_block; /* last block number in bio */ 1187 unsigned char version; /* version of the node */ 1188 }; 1189 1190 struct bio_entry { 1191 struct bio *bio; 1192 struct list_head list; 1193 }; 1194 1195 #define is_read_io(rw) ((rw) == READ) 1196 struct f2fs_bio_info { 1197 struct f2fs_sb_info *sbi; /* f2fs superblock */ 1198 struct bio *bio; /* bios to merge */ 1199 sector_t last_block_in_bio; /* last block number */ 1200 struct f2fs_io_info fio; /* store buffered io info. */ 1201 struct rw_semaphore io_rwsem; /* blocking op for bio */ 1202 spinlock_t io_lock; /* serialize DATA/NODE IOs */ 1203 struct list_head io_list; /* track fios */ 1204 struct list_head bio_list; /* bio entry list head */ 1205 struct rw_semaphore bio_list_lock; /* lock to protect bio entry list */ 1206 }; 1207 1208 #define FDEV(i) (sbi->devs[i]) 1209 #define RDEV(i) (raw_super->devs[i]) 1210 struct f2fs_dev_info { 1211 struct block_device *bdev; 1212 char path[MAX_PATH_LEN]; 1213 unsigned int total_segments; 1214 block_t start_blk; 1215 block_t end_blk; 1216 #ifdef CONFIG_BLK_DEV_ZONED 1217 unsigned int nr_blkz; /* Total number of zones */ 1218 unsigned long *blkz_seq; /* Bitmap indicating sequential zones */ 1219 #endif 1220 }; 1221 1222 enum inode_type { 1223 DIR_INODE, /* for dirty dir inode */ 1224 FILE_INODE, /* for dirty regular/symlink inode */ 1225 DIRTY_META, /* for all dirtied inode metadata */ 1226 ATOMIC_FILE, /* for all atomic files */ 1227 NR_INODE_TYPE, 1228 }; 1229 1230 /* for inner inode cache management */ 1231 struct inode_management { 1232 struct radix_tree_root ino_root; /* ino entry array */ 1233 spinlock_t ino_lock; /* for ino entry lock */ 1234 struct list_head ino_list; /* inode list head */ 1235 unsigned long ino_num; /* number of entries */ 1236 }; 1237 1238 /* For s_flag in struct f2fs_sb_info */ 1239 enum { 1240 SBI_IS_DIRTY, /* dirty flag for checkpoint */ 1241 SBI_IS_CLOSE, /* specify unmounting */ 1242 SBI_NEED_FSCK, /* need fsck.f2fs to fix */ 1243 SBI_POR_DOING, /* recovery is doing or not */ 1244 SBI_NEED_SB_WRITE, /* need to recover superblock */ 1245 SBI_NEED_CP, /* need to checkpoint */ 1246 SBI_IS_SHUTDOWN, /* shutdown by ioctl */ 1247 SBI_IS_RECOVERED, /* recovered orphan/data */ 1248 SBI_CP_DISABLED, /* CP was disabled last mount */ 1249 SBI_CP_DISABLED_QUICK, /* CP was disabled quickly */ 1250 SBI_QUOTA_NEED_FLUSH, /* need to flush quota info in CP */ 1251 SBI_QUOTA_SKIP_FLUSH, /* skip flushing quota in current CP */ 1252 SBI_QUOTA_NEED_REPAIR, /* quota file may be corrupted */ 1253 SBI_IS_RESIZEFS, /* resizefs is in process */ 1254 }; 1255 1256 enum { 1257 CP_TIME, 1258 REQ_TIME, 1259 DISCARD_TIME, 1260 GC_TIME, 1261 DISABLE_TIME, 1262 UMOUNT_DISCARD_TIMEOUT, 1263 MAX_TIME, 1264 }; 1265 1266 enum { 1267 GC_NORMAL, 1268 GC_IDLE_CB, 1269 GC_IDLE_GREEDY, 1270 GC_URGENT, 1271 }; 1272 1273 enum { 1274 BGGC_MODE_ON, /* background gc is on */ 1275 BGGC_MODE_OFF, /* background gc is off */ 1276 BGGC_MODE_SYNC, /* 1277 * background gc is on, migrating blocks 1278 * like foreground gc 1279 */ 1280 }; 1281 1282 enum { 1283 FS_MODE_ADAPTIVE, /* use both lfs/ssr allocation */ 1284 FS_MODE_LFS, /* use lfs allocation only */ 1285 }; 1286 1287 enum { 1288 WHINT_MODE_OFF, /* not pass down write hints */ 1289 WHINT_MODE_USER, /* try to pass down hints given by users */ 1290 WHINT_MODE_FS, /* pass down hints with F2FS policy */ 1291 }; 1292 1293 enum { 1294 ALLOC_MODE_DEFAULT, /* stay default */ 1295 ALLOC_MODE_REUSE, /* reuse segments as much as possible */ 1296 }; 1297 1298 enum fsync_mode { 1299 FSYNC_MODE_POSIX, /* fsync follows posix semantics */ 1300 FSYNC_MODE_STRICT, /* fsync behaves in line with ext4 */ 1301 FSYNC_MODE_NOBARRIER, /* fsync behaves nobarrier based on posix */ 1302 }; 1303 1304 /* 1305 * this value is set in page as a private data which indicate that 1306 * the page is atomically written, and it is in inmem_pages list. 1307 */ 1308 #define ATOMIC_WRITTEN_PAGE ((unsigned long)-1) 1309 #define DUMMY_WRITTEN_PAGE ((unsigned long)-2) 1310 1311 #define IS_ATOMIC_WRITTEN_PAGE(page) \ 1312 (page_private(page) == (unsigned long)ATOMIC_WRITTEN_PAGE) 1313 #define IS_DUMMY_WRITTEN_PAGE(page) \ 1314 (page_private(page) == (unsigned long)DUMMY_WRITTEN_PAGE) 1315 1316 #ifdef CONFIG_FS_ENCRYPTION 1317 #define DUMMY_ENCRYPTION_ENABLED(sbi) \ 1318 (unlikely(F2FS_OPTION(sbi).test_dummy_encryption)) 1319 #else 1320 #define DUMMY_ENCRYPTION_ENABLED(sbi) (0) 1321 #endif 1322 1323 /* For compression */ 1324 enum compress_algorithm_type { 1325 COMPRESS_LZO, 1326 COMPRESS_LZ4, 1327 COMPRESS_ZSTD, 1328 COMPRESS_LZORLE, 1329 COMPRESS_MAX, 1330 }; 1331 1332 #define COMPRESS_DATA_RESERVED_SIZE 5 1333 struct compress_data { 1334 __le32 clen; /* compressed data size */ 1335 __le32 reserved[COMPRESS_DATA_RESERVED_SIZE]; /* reserved */ 1336 u8 cdata[]; /* compressed data */ 1337 }; 1338 1339 #define COMPRESS_HEADER_SIZE (sizeof(struct compress_data)) 1340 1341 #define F2FS_COMPRESSED_PAGE_MAGIC 0xF5F2C000 1342 1343 /* compress context */ 1344 struct compress_ctx { 1345 struct inode *inode; /* inode the context belong to */ 1346 pgoff_t cluster_idx; /* cluster index number */ 1347 unsigned int cluster_size; /* page count in cluster */ 1348 unsigned int log_cluster_size; /* log of cluster size */ 1349 struct page **rpages; /* pages store raw data in cluster */ 1350 unsigned int nr_rpages; /* total page number in rpages */ 1351 struct page **cpages; /* pages store compressed data in cluster */ 1352 unsigned int nr_cpages; /* total page number in cpages */ 1353 void *rbuf; /* virtual mapped address on rpages */ 1354 struct compress_data *cbuf; /* virtual mapped address on cpages */ 1355 size_t rlen; /* valid data length in rbuf */ 1356 size_t clen; /* valid data length in cbuf */ 1357 void *private; /* payload buffer for specified compression algorithm */ 1358 void *private2; /* extra payload buffer */ 1359 }; 1360 1361 /* compress context for write IO path */ 1362 struct compress_io_ctx { 1363 u32 magic; /* magic number to indicate page is compressed */ 1364 struct inode *inode; /* inode the context belong to */ 1365 struct page **rpages; /* pages store raw data in cluster */ 1366 unsigned int nr_rpages; /* total page number in rpages */ 1367 refcount_t ref; /* referrence count of raw page */ 1368 }; 1369 1370 /* decompress io context for read IO path */ 1371 struct decompress_io_ctx { 1372 u32 magic; /* magic number to indicate page is compressed */ 1373 struct inode *inode; /* inode the context belong to */ 1374 pgoff_t cluster_idx; /* cluster index number */ 1375 unsigned int cluster_size; /* page count in cluster */ 1376 unsigned int log_cluster_size; /* log of cluster size */ 1377 struct page **rpages; /* pages store raw data in cluster */ 1378 unsigned int nr_rpages; /* total page number in rpages */ 1379 struct page **cpages; /* pages store compressed data in cluster */ 1380 unsigned int nr_cpages; /* total page number in cpages */ 1381 struct page **tpages; /* temp pages to pad holes in cluster */ 1382 void *rbuf; /* virtual mapped address on rpages */ 1383 struct compress_data *cbuf; /* virtual mapped address on cpages */ 1384 size_t rlen; /* valid data length in rbuf */ 1385 size_t clen; /* valid data length in cbuf */ 1386 refcount_t ref; /* referrence count of compressed page */ 1387 bool failed; /* indicate IO error during decompression */ 1388 void *private; /* payload buffer for specified decompression algorithm */ 1389 void *private2; /* extra payload buffer */ 1390 }; 1391 1392 #define NULL_CLUSTER ((unsigned int)(~0)) 1393 #define MIN_COMPRESS_LOG_SIZE 2 1394 #define MAX_COMPRESS_LOG_SIZE 8 1395 #define MAX_COMPRESS_WINDOW_SIZE ((PAGE_SIZE) << MAX_COMPRESS_LOG_SIZE) 1396 1397 struct f2fs_sb_info { 1398 struct super_block *sb; /* pointer to VFS super block */ 1399 struct proc_dir_entry *s_proc; /* proc entry */ 1400 struct f2fs_super_block *raw_super; /* raw super block pointer */ 1401 struct rw_semaphore sb_lock; /* lock for raw super block */ 1402 int valid_super_block; /* valid super block no */ 1403 unsigned long s_flag; /* flags for sbi */ 1404 struct mutex writepages; /* mutex for writepages() */ 1405 #ifdef CONFIG_UNICODE 1406 struct unicode_map *s_encoding; 1407 __u16 s_encoding_flags; 1408 #endif 1409 1410 #ifdef CONFIG_BLK_DEV_ZONED 1411 unsigned int blocks_per_blkz; /* F2FS blocks per zone */ 1412 unsigned int log_blocks_per_blkz; /* log2 F2FS blocks per zone */ 1413 #endif 1414 1415 /* for node-related operations */ 1416 struct f2fs_nm_info *nm_info; /* node manager */ 1417 struct inode *node_inode; /* cache node blocks */ 1418 1419 /* for segment-related operations */ 1420 struct f2fs_sm_info *sm_info; /* segment manager */ 1421 1422 /* for bio operations */ 1423 struct f2fs_bio_info *write_io[NR_PAGE_TYPE]; /* for write bios */ 1424 /* keep migration IO order for LFS mode */ 1425 struct rw_semaphore io_order_lock; 1426 mempool_t *write_io_dummy; /* Dummy pages */ 1427 1428 /* for checkpoint */ 1429 struct f2fs_checkpoint *ckpt; /* raw checkpoint pointer */ 1430 int cur_cp_pack; /* remain current cp pack */ 1431 spinlock_t cp_lock; /* for flag in ckpt */ 1432 struct inode *meta_inode; /* cache meta blocks */ 1433 struct mutex cp_mutex; /* checkpoint procedure lock */ 1434 struct rw_semaphore cp_rwsem; /* blocking FS operations */ 1435 struct rw_semaphore node_write; /* locking node writes */ 1436 struct rw_semaphore node_change; /* locking node change */ 1437 wait_queue_head_t cp_wait; 1438 unsigned long last_time[MAX_TIME]; /* to store time in jiffies */ 1439 long interval_time[MAX_TIME]; /* to store thresholds */ 1440 1441 struct inode_management im[MAX_INO_ENTRY]; /* manage inode cache */ 1442 1443 spinlock_t fsync_node_lock; /* for node entry lock */ 1444 struct list_head fsync_node_list; /* node list head */ 1445 unsigned int fsync_seg_id; /* sequence id */ 1446 unsigned int fsync_node_num; /* number of node entries */ 1447 1448 /* for orphan inode, use 0'th array */ 1449 unsigned int max_orphans; /* max orphan inodes */ 1450 1451 /* for inode management */ 1452 struct list_head inode_list[NR_INODE_TYPE]; /* dirty inode list */ 1453 spinlock_t inode_lock[NR_INODE_TYPE]; /* for dirty inode list lock */ 1454 struct mutex flush_lock; /* for flush exclusion */ 1455 1456 /* for extent tree cache */ 1457 struct radix_tree_root extent_tree_root;/* cache extent cache entries */ 1458 struct mutex extent_tree_lock; /* locking extent radix tree */ 1459 struct list_head extent_list; /* lru list for shrinker */ 1460 spinlock_t extent_lock; /* locking extent lru list */ 1461 atomic_t total_ext_tree; /* extent tree count */ 1462 struct list_head zombie_list; /* extent zombie tree list */ 1463 atomic_t total_zombie_tree; /* extent zombie tree count */ 1464 atomic_t total_ext_node; /* extent info count */ 1465 1466 /* basic filesystem units */ 1467 unsigned int log_sectors_per_block; /* log2 sectors per block */ 1468 unsigned int log_blocksize; /* log2 block size */ 1469 unsigned int blocksize; /* block size */ 1470 unsigned int root_ino_num; /* root inode number*/ 1471 unsigned int node_ino_num; /* node inode number*/ 1472 unsigned int meta_ino_num; /* meta inode number*/ 1473 unsigned int log_blocks_per_seg; /* log2 blocks per segment */ 1474 unsigned int blocks_per_seg; /* blocks per segment */ 1475 unsigned int segs_per_sec; /* segments per section */ 1476 unsigned int secs_per_zone; /* sections per zone */ 1477 unsigned int total_sections; /* total section count */ 1478 unsigned int total_node_count; /* total node block count */ 1479 unsigned int total_valid_node_count; /* valid node block count */ 1480 loff_t max_file_blocks; /* max block index of file */ 1481 int dir_level; /* directory level */ 1482 int readdir_ra; /* readahead inode in readdir */ 1483 1484 block_t user_block_count; /* # of user blocks */ 1485 block_t total_valid_block_count; /* # of valid blocks */ 1486 block_t discard_blks; /* discard command candidats */ 1487 block_t last_valid_block_count; /* for recovery */ 1488 block_t reserved_blocks; /* configurable reserved blocks */ 1489 block_t current_reserved_blocks; /* current reserved blocks */ 1490 1491 /* Additional tracking for no checkpoint mode */ 1492 block_t unusable_block_count; /* # of blocks saved by last cp */ 1493 1494 unsigned int nquota_files; /* # of quota sysfile */ 1495 struct rw_semaphore quota_sem; /* blocking cp for flags */ 1496 1497 /* # of pages, see count_type */ 1498 atomic_t nr_pages[NR_COUNT_TYPE]; 1499 /* # of allocated blocks */ 1500 struct percpu_counter alloc_valid_block_count; 1501 1502 /* writeback control */ 1503 atomic_t wb_sync_req[META]; /* count # of WB_SYNC threads */ 1504 1505 /* valid inode count */ 1506 struct percpu_counter total_valid_inode_count; 1507 1508 struct f2fs_mount_info mount_opt; /* mount options */ 1509 1510 /* for cleaning operations */ 1511 struct rw_semaphore gc_lock; /* 1512 * semaphore for GC, avoid 1513 * race between GC and GC or CP 1514 */ 1515 struct f2fs_gc_kthread *gc_thread; /* GC thread */ 1516 unsigned int cur_victim_sec; /* current victim section num */ 1517 unsigned int gc_mode; /* current GC state */ 1518 unsigned int next_victim_seg[2]; /* next segment in victim section */ 1519 /* for skip statistic */ 1520 unsigned int atomic_files; /* # of opened atomic file */ 1521 unsigned long long skipped_atomic_files[2]; /* FG_GC and BG_GC */ 1522 unsigned long long skipped_gc_rwsem; /* FG_GC only */ 1523 1524 /* threshold for gc trials on pinned files */ 1525 u64 gc_pin_file_threshold; 1526 struct rw_semaphore pin_sem; 1527 1528 /* maximum # of trials to find a victim segment for SSR and GC */ 1529 unsigned int max_victim_search; 1530 /* migration granularity of garbage collection, unit: segment */ 1531 unsigned int migration_granularity; 1532 1533 /* 1534 * for stat information. 1535 * one is for the LFS mode, and the other is for the SSR mode. 1536 */ 1537 #ifdef CONFIG_F2FS_STAT_FS 1538 struct f2fs_stat_info *stat_info; /* FS status information */ 1539 atomic_t meta_count[META_MAX]; /* # of meta blocks */ 1540 unsigned int segment_count[2]; /* # of allocated segments */ 1541 unsigned int block_count[2]; /* # of allocated blocks */ 1542 atomic_t inplace_count; /* # of inplace update */ 1543 atomic64_t total_hit_ext; /* # of lookup extent cache */ 1544 atomic64_t read_hit_rbtree; /* # of hit rbtree extent node */ 1545 atomic64_t read_hit_largest; /* # of hit largest extent node */ 1546 atomic64_t read_hit_cached; /* # of hit cached extent node */ 1547 atomic_t inline_xattr; /* # of inline_xattr inodes */ 1548 atomic_t inline_inode; /* # of inline_data inodes */ 1549 atomic_t inline_dir; /* # of inline_dentry inodes */ 1550 atomic_t compr_inode; /* # of compressed inodes */ 1551 atomic_t compr_blocks; /* # of compressed blocks */ 1552 atomic_t vw_cnt; /* # of volatile writes */ 1553 atomic_t max_aw_cnt; /* max # of atomic writes */ 1554 atomic_t max_vw_cnt; /* max # of volatile writes */ 1555 unsigned int io_skip_bggc; /* skip background gc for in-flight IO */ 1556 unsigned int other_skip_bggc; /* skip background gc for other reasons */ 1557 unsigned int ndirty_inode[NR_INODE_TYPE]; /* # of dirty inodes */ 1558 #endif 1559 spinlock_t stat_lock; /* lock for stat operations */ 1560 1561 /* For app/fs IO statistics */ 1562 spinlock_t iostat_lock; 1563 unsigned long long rw_iostat[NR_IO_TYPE]; 1564 unsigned long long prev_rw_iostat[NR_IO_TYPE]; 1565 bool iostat_enable; 1566 unsigned long iostat_next_period; 1567 unsigned int iostat_period_ms; 1568 1569 /* to attach REQ_META|REQ_FUA flags */ 1570 unsigned int data_io_flag; 1571 1572 /* For sysfs suppport */ 1573 struct kobject s_kobj; 1574 struct completion s_kobj_unregister; 1575 1576 /* For shrinker support */ 1577 struct list_head s_list; 1578 int s_ndevs; /* number of devices */ 1579 struct f2fs_dev_info *devs; /* for device list */ 1580 unsigned int dirty_device; /* for checkpoint data flush */ 1581 spinlock_t dev_lock; /* protect dirty_device */ 1582 struct mutex umount_mutex; 1583 unsigned int shrinker_run_no; 1584 1585 /* For write statistics */ 1586 u64 sectors_written_start; 1587 u64 kbytes_written; 1588 1589 /* Reference to checksum algorithm driver via cryptoapi */ 1590 struct crypto_shash *s_chksum_driver; 1591 1592 /* Precomputed FS UUID checksum for seeding other checksums */ 1593 __u32 s_chksum_seed; 1594 1595 struct workqueue_struct *post_read_wq; /* post read workqueue */ 1596 1597 struct kmem_cache *inline_xattr_slab; /* inline xattr entry */ 1598 unsigned int inline_xattr_slab_size; /* default inline xattr slab size */ 1599 }; 1600 1601 struct f2fs_private_dio { 1602 struct inode *inode; 1603 void *orig_private; 1604 bio_end_io_t *orig_end_io; 1605 bool write; 1606 }; 1607 1608 #ifdef CONFIG_F2FS_FAULT_INJECTION 1609 #define f2fs_show_injection_info(sbi, type) \ 1610 printk_ratelimited("%sF2FS-fs (%s) : inject %s in %s of %pS\n", \ 1611 KERN_INFO, sbi->sb->s_id, \ 1612 f2fs_fault_name[type], \ 1613 __func__, __builtin_return_address(0)) 1614 static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type) 1615 { 1616 struct f2fs_fault_info *ffi = &F2FS_OPTION(sbi).fault_info; 1617 1618 if (!ffi->inject_rate) 1619 return false; 1620 1621 if (!IS_FAULT_SET(ffi, type)) 1622 return false; 1623 1624 atomic_inc(&ffi->inject_ops); 1625 if (atomic_read(&ffi->inject_ops) >= ffi->inject_rate) { 1626 atomic_set(&ffi->inject_ops, 0); 1627 return true; 1628 } 1629 return false; 1630 } 1631 #else 1632 #define f2fs_show_injection_info(sbi, type) do { } while (0) 1633 static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type) 1634 { 1635 return false; 1636 } 1637 #endif 1638 1639 /* 1640 * Test if the mounted volume is a multi-device volume. 1641 * - For a single regular disk volume, sbi->s_ndevs is 0. 1642 * - For a single zoned disk volume, sbi->s_ndevs is 1. 1643 * - For a multi-device volume, sbi->s_ndevs is always 2 or more. 1644 */ 1645 static inline bool f2fs_is_multi_device(struct f2fs_sb_info *sbi) 1646 { 1647 return sbi->s_ndevs > 1; 1648 } 1649 1650 /* For write statistics. Suppose sector size is 512 bytes, 1651 * and the return value is in kbytes. s is of struct f2fs_sb_info. 1652 */ 1653 #define BD_PART_WRITTEN(s) \ 1654 (((u64)part_stat_read((s)->sb->s_bdev->bd_part, sectors[STAT_WRITE]) - \ 1655 (s)->sectors_written_start) >> 1) 1656 1657 static inline void f2fs_update_time(struct f2fs_sb_info *sbi, int type) 1658 { 1659 unsigned long now = jiffies; 1660 1661 sbi->last_time[type] = now; 1662 1663 /* DISCARD_TIME and GC_TIME are based on REQ_TIME */ 1664 if (type == REQ_TIME) { 1665 sbi->last_time[DISCARD_TIME] = now; 1666 sbi->last_time[GC_TIME] = now; 1667 } 1668 } 1669 1670 static inline bool f2fs_time_over(struct f2fs_sb_info *sbi, int type) 1671 { 1672 unsigned long interval = sbi->interval_time[type] * HZ; 1673 1674 return time_after(jiffies, sbi->last_time[type] + interval); 1675 } 1676 1677 static inline unsigned int f2fs_time_to_wait(struct f2fs_sb_info *sbi, 1678 int type) 1679 { 1680 unsigned long interval = sbi->interval_time[type] * HZ; 1681 unsigned int wait_ms = 0; 1682 long delta; 1683 1684 delta = (sbi->last_time[type] + interval) - jiffies; 1685 if (delta > 0) 1686 wait_ms = jiffies_to_msecs(delta); 1687 1688 return wait_ms; 1689 } 1690 1691 /* 1692 * Inline functions 1693 */ 1694 static inline u32 __f2fs_crc32(struct f2fs_sb_info *sbi, u32 crc, 1695 const void *address, unsigned int length) 1696 { 1697 struct { 1698 struct shash_desc shash; 1699 char ctx[4]; 1700 } desc; 1701 int err; 1702 1703 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver) != sizeof(desc.ctx)); 1704 1705 desc.shash.tfm = sbi->s_chksum_driver; 1706 *(u32 *)desc.ctx = crc; 1707 1708 err = crypto_shash_update(&desc.shash, address, length); 1709 BUG_ON(err); 1710 1711 return *(u32 *)desc.ctx; 1712 } 1713 1714 static inline u32 f2fs_crc32(struct f2fs_sb_info *sbi, const void *address, 1715 unsigned int length) 1716 { 1717 return __f2fs_crc32(sbi, F2FS_SUPER_MAGIC, address, length); 1718 } 1719 1720 static inline bool f2fs_crc_valid(struct f2fs_sb_info *sbi, __u32 blk_crc, 1721 void *buf, size_t buf_size) 1722 { 1723 return f2fs_crc32(sbi, buf, buf_size) == blk_crc; 1724 } 1725 1726 static inline u32 f2fs_chksum(struct f2fs_sb_info *sbi, u32 crc, 1727 const void *address, unsigned int length) 1728 { 1729 return __f2fs_crc32(sbi, crc, address, length); 1730 } 1731 1732 static inline struct f2fs_inode_info *F2FS_I(struct inode *inode) 1733 { 1734 return container_of(inode, struct f2fs_inode_info, vfs_inode); 1735 } 1736 1737 static inline struct f2fs_sb_info *F2FS_SB(struct super_block *sb) 1738 { 1739 return sb->s_fs_info; 1740 } 1741 1742 static inline struct f2fs_sb_info *F2FS_I_SB(struct inode *inode) 1743 { 1744 return F2FS_SB(inode->i_sb); 1745 } 1746 1747 static inline struct f2fs_sb_info *F2FS_M_SB(struct address_space *mapping) 1748 { 1749 return F2FS_I_SB(mapping->host); 1750 } 1751 1752 static inline struct f2fs_sb_info *F2FS_P_SB(struct page *page) 1753 { 1754 return F2FS_M_SB(page_file_mapping(page)); 1755 } 1756 1757 static inline struct f2fs_super_block *F2FS_RAW_SUPER(struct f2fs_sb_info *sbi) 1758 { 1759 return (struct f2fs_super_block *)(sbi->raw_super); 1760 } 1761 1762 static inline struct f2fs_checkpoint *F2FS_CKPT(struct f2fs_sb_info *sbi) 1763 { 1764 return (struct f2fs_checkpoint *)(sbi->ckpt); 1765 } 1766 1767 static inline struct f2fs_node *F2FS_NODE(struct page *page) 1768 { 1769 return (struct f2fs_node *)page_address(page); 1770 } 1771 1772 static inline struct f2fs_inode *F2FS_INODE(struct page *page) 1773 { 1774 return &((struct f2fs_node *)page_address(page))->i; 1775 } 1776 1777 static inline struct f2fs_nm_info *NM_I(struct f2fs_sb_info *sbi) 1778 { 1779 return (struct f2fs_nm_info *)(sbi->nm_info); 1780 } 1781 1782 static inline struct f2fs_sm_info *SM_I(struct f2fs_sb_info *sbi) 1783 { 1784 return (struct f2fs_sm_info *)(sbi->sm_info); 1785 } 1786 1787 static inline struct sit_info *SIT_I(struct f2fs_sb_info *sbi) 1788 { 1789 return (struct sit_info *)(SM_I(sbi)->sit_info); 1790 } 1791 1792 static inline struct free_segmap_info *FREE_I(struct f2fs_sb_info *sbi) 1793 { 1794 return (struct free_segmap_info *)(SM_I(sbi)->free_info); 1795 } 1796 1797 static inline struct dirty_seglist_info *DIRTY_I(struct f2fs_sb_info *sbi) 1798 { 1799 return (struct dirty_seglist_info *)(SM_I(sbi)->dirty_info); 1800 } 1801 1802 static inline struct address_space *META_MAPPING(struct f2fs_sb_info *sbi) 1803 { 1804 return sbi->meta_inode->i_mapping; 1805 } 1806 1807 static inline struct address_space *NODE_MAPPING(struct f2fs_sb_info *sbi) 1808 { 1809 return sbi->node_inode->i_mapping; 1810 } 1811 1812 static inline bool is_sbi_flag_set(struct f2fs_sb_info *sbi, unsigned int type) 1813 { 1814 return test_bit(type, &sbi->s_flag); 1815 } 1816 1817 static inline void set_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type) 1818 { 1819 set_bit(type, &sbi->s_flag); 1820 } 1821 1822 static inline void clear_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type) 1823 { 1824 clear_bit(type, &sbi->s_flag); 1825 } 1826 1827 static inline unsigned long long cur_cp_version(struct f2fs_checkpoint *cp) 1828 { 1829 return le64_to_cpu(cp->checkpoint_ver); 1830 } 1831 1832 static inline unsigned long f2fs_qf_ino(struct super_block *sb, int type) 1833 { 1834 if (type < F2FS_MAX_QUOTAS) 1835 return le32_to_cpu(F2FS_SB(sb)->raw_super->qf_ino[type]); 1836 return 0; 1837 } 1838 1839 static inline __u64 cur_cp_crc(struct f2fs_checkpoint *cp) 1840 { 1841 size_t crc_offset = le32_to_cpu(cp->checksum_offset); 1842 return le32_to_cpu(*((__le32 *)((unsigned char *)cp + crc_offset))); 1843 } 1844 1845 static inline bool __is_set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 1846 { 1847 unsigned int ckpt_flags = le32_to_cpu(cp->ckpt_flags); 1848 1849 return ckpt_flags & f; 1850 } 1851 1852 static inline bool is_set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 1853 { 1854 return __is_set_ckpt_flags(F2FS_CKPT(sbi), f); 1855 } 1856 1857 static inline void __set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 1858 { 1859 unsigned int ckpt_flags; 1860 1861 ckpt_flags = le32_to_cpu(cp->ckpt_flags); 1862 ckpt_flags |= f; 1863 cp->ckpt_flags = cpu_to_le32(ckpt_flags); 1864 } 1865 1866 static inline void set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 1867 { 1868 unsigned long flags; 1869 1870 spin_lock_irqsave(&sbi->cp_lock, flags); 1871 __set_ckpt_flags(F2FS_CKPT(sbi), f); 1872 spin_unlock_irqrestore(&sbi->cp_lock, flags); 1873 } 1874 1875 static inline void __clear_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 1876 { 1877 unsigned int ckpt_flags; 1878 1879 ckpt_flags = le32_to_cpu(cp->ckpt_flags); 1880 ckpt_flags &= (~f); 1881 cp->ckpt_flags = cpu_to_le32(ckpt_flags); 1882 } 1883 1884 static inline void clear_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 1885 { 1886 unsigned long flags; 1887 1888 spin_lock_irqsave(&sbi->cp_lock, flags); 1889 __clear_ckpt_flags(F2FS_CKPT(sbi), f); 1890 spin_unlock_irqrestore(&sbi->cp_lock, flags); 1891 } 1892 1893 static inline void disable_nat_bits(struct f2fs_sb_info *sbi, bool lock) 1894 { 1895 unsigned long flags; 1896 unsigned char *nat_bits; 1897 1898 /* 1899 * In order to re-enable nat_bits we need to call fsck.f2fs by 1900 * set_sbi_flag(sbi, SBI_NEED_FSCK). But it may give huge cost, 1901 * so let's rely on regular fsck or unclean shutdown. 1902 */ 1903 1904 if (lock) 1905 spin_lock_irqsave(&sbi->cp_lock, flags); 1906 __clear_ckpt_flags(F2FS_CKPT(sbi), CP_NAT_BITS_FLAG); 1907 nat_bits = NM_I(sbi)->nat_bits; 1908 NM_I(sbi)->nat_bits = NULL; 1909 if (lock) 1910 spin_unlock_irqrestore(&sbi->cp_lock, flags); 1911 1912 kvfree(nat_bits); 1913 } 1914 1915 static inline bool enabled_nat_bits(struct f2fs_sb_info *sbi, 1916 struct cp_control *cpc) 1917 { 1918 bool set = is_set_ckpt_flags(sbi, CP_NAT_BITS_FLAG); 1919 1920 return (cpc) ? (cpc->reason & CP_UMOUNT) && set : set; 1921 } 1922 1923 static inline void f2fs_lock_op(struct f2fs_sb_info *sbi) 1924 { 1925 down_read(&sbi->cp_rwsem); 1926 } 1927 1928 static inline int f2fs_trylock_op(struct f2fs_sb_info *sbi) 1929 { 1930 return down_read_trylock(&sbi->cp_rwsem); 1931 } 1932 1933 static inline void f2fs_unlock_op(struct f2fs_sb_info *sbi) 1934 { 1935 up_read(&sbi->cp_rwsem); 1936 } 1937 1938 static inline void f2fs_lock_all(struct f2fs_sb_info *sbi) 1939 { 1940 down_write(&sbi->cp_rwsem); 1941 } 1942 1943 static inline void f2fs_unlock_all(struct f2fs_sb_info *sbi) 1944 { 1945 up_write(&sbi->cp_rwsem); 1946 } 1947 1948 static inline int __get_cp_reason(struct f2fs_sb_info *sbi) 1949 { 1950 int reason = CP_SYNC; 1951 1952 if (test_opt(sbi, FASTBOOT)) 1953 reason = CP_FASTBOOT; 1954 if (is_sbi_flag_set(sbi, SBI_IS_CLOSE)) 1955 reason = CP_UMOUNT; 1956 return reason; 1957 } 1958 1959 static inline bool __remain_node_summaries(int reason) 1960 { 1961 return (reason & (CP_UMOUNT | CP_FASTBOOT)); 1962 } 1963 1964 static inline bool __exist_node_summaries(struct f2fs_sb_info *sbi) 1965 { 1966 return (is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG) || 1967 is_set_ckpt_flags(sbi, CP_FASTBOOT_FLAG)); 1968 } 1969 1970 /* 1971 * Check whether the inode has blocks or not 1972 */ 1973 static inline int F2FS_HAS_BLOCKS(struct inode *inode) 1974 { 1975 block_t xattr_block = F2FS_I(inode)->i_xattr_nid ? 1 : 0; 1976 1977 return (inode->i_blocks >> F2FS_LOG_SECTORS_PER_BLOCK) > xattr_block; 1978 } 1979 1980 static inline bool f2fs_has_xattr_block(unsigned int ofs) 1981 { 1982 return ofs == XATTR_NODE_OFFSET; 1983 } 1984 1985 static inline bool __allow_reserved_blocks(struct f2fs_sb_info *sbi, 1986 struct inode *inode, bool cap) 1987 { 1988 if (!inode) 1989 return true; 1990 if (!test_opt(sbi, RESERVE_ROOT)) 1991 return false; 1992 if (IS_NOQUOTA(inode)) 1993 return true; 1994 if (uid_eq(F2FS_OPTION(sbi).s_resuid, current_fsuid())) 1995 return true; 1996 if (!gid_eq(F2FS_OPTION(sbi).s_resgid, GLOBAL_ROOT_GID) && 1997 in_group_p(F2FS_OPTION(sbi).s_resgid)) 1998 return true; 1999 if (cap && capable(CAP_SYS_RESOURCE)) 2000 return true; 2001 return false; 2002 } 2003 2004 static inline void f2fs_i_blocks_write(struct inode *, block_t, bool, bool); 2005 static inline int inc_valid_block_count(struct f2fs_sb_info *sbi, 2006 struct inode *inode, blkcnt_t *count) 2007 { 2008 blkcnt_t diff = 0, release = 0; 2009 block_t avail_user_block_count; 2010 int ret; 2011 2012 ret = dquot_reserve_block(inode, *count); 2013 if (ret) 2014 return ret; 2015 2016 if (time_to_inject(sbi, FAULT_BLOCK)) { 2017 f2fs_show_injection_info(sbi, FAULT_BLOCK); 2018 release = *count; 2019 goto release_quota; 2020 } 2021 2022 /* 2023 * let's increase this in prior to actual block count change in order 2024 * for f2fs_sync_file to avoid data races when deciding checkpoint. 2025 */ 2026 percpu_counter_add(&sbi->alloc_valid_block_count, (*count)); 2027 2028 spin_lock(&sbi->stat_lock); 2029 sbi->total_valid_block_count += (block_t)(*count); 2030 avail_user_block_count = sbi->user_block_count - 2031 sbi->current_reserved_blocks; 2032 2033 if (!__allow_reserved_blocks(sbi, inode, true)) 2034 avail_user_block_count -= F2FS_OPTION(sbi).root_reserved_blocks; 2035 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) { 2036 if (avail_user_block_count > sbi->unusable_block_count) 2037 avail_user_block_count -= sbi->unusable_block_count; 2038 else 2039 avail_user_block_count = 0; 2040 } 2041 if (unlikely(sbi->total_valid_block_count > avail_user_block_count)) { 2042 diff = sbi->total_valid_block_count - avail_user_block_count; 2043 if (diff > *count) 2044 diff = *count; 2045 *count -= diff; 2046 release = diff; 2047 sbi->total_valid_block_count -= diff; 2048 if (!*count) { 2049 spin_unlock(&sbi->stat_lock); 2050 goto enospc; 2051 } 2052 } 2053 spin_unlock(&sbi->stat_lock); 2054 2055 if (unlikely(release)) { 2056 percpu_counter_sub(&sbi->alloc_valid_block_count, release); 2057 dquot_release_reservation_block(inode, release); 2058 } 2059 f2fs_i_blocks_write(inode, *count, true, true); 2060 return 0; 2061 2062 enospc: 2063 percpu_counter_sub(&sbi->alloc_valid_block_count, release); 2064 release_quota: 2065 dquot_release_reservation_block(inode, release); 2066 return -ENOSPC; 2067 } 2068 2069 __printf(2, 3) 2070 void f2fs_printk(struct f2fs_sb_info *sbi, const char *fmt, ...); 2071 2072 #define f2fs_err(sbi, fmt, ...) \ 2073 f2fs_printk(sbi, KERN_ERR fmt, ##__VA_ARGS__) 2074 #define f2fs_warn(sbi, fmt, ...) \ 2075 f2fs_printk(sbi, KERN_WARNING fmt, ##__VA_ARGS__) 2076 #define f2fs_notice(sbi, fmt, ...) \ 2077 f2fs_printk(sbi, KERN_NOTICE fmt, ##__VA_ARGS__) 2078 #define f2fs_info(sbi, fmt, ...) \ 2079 f2fs_printk(sbi, KERN_INFO fmt, ##__VA_ARGS__) 2080 #define f2fs_debug(sbi, fmt, ...) \ 2081 f2fs_printk(sbi, KERN_DEBUG fmt, ##__VA_ARGS__) 2082 2083 static inline void dec_valid_block_count(struct f2fs_sb_info *sbi, 2084 struct inode *inode, 2085 block_t count) 2086 { 2087 blkcnt_t sectors = count << F2FS_LOG_SECTORS_PER_BLOCK; 2088 2089 spin_lock(&sbi->stat_lock); 2090 f2fs_bug_on(sbi, sbi->total_valid_block_count < (block_t) count); 2091 sbi->total_valid_block_count -= (block_t)count; 2092 if (sbi->reserved_blocks && 2093 sbi->current_reserved_blocks < sbi->reserved_blocks) 2094 sbi->current_reserved_blocks = min(sbi->reserved_blocks, 2095 sbi->current_reserved_blocks + count); 2096 spin_unlock(&sbi->stat_lock); 2097 if (unlikely(inode->i_blocks < sectors)) { 2098 f2fs_warn(sbi, "Inconsistent i_blocks, ino:%lu, iblocks:%llu, sectors:%llu", 2099 inode->i_ino, 2100 (unsigned long long)inode->i_blocks, 2101 (unsigned long long)sectors); 2102 set_sbi_flag(sbi, SBI_NEED_FSCK); 2103 return; 2104 } 2105 f2fs_i_blocks_write(inode, count, false, true); 2106 } 2107 2108 static inline void inc_page_count(struct f2fs_sb_info *sbi, int count_type) 2109 { 2110 atomic_inc(&sbi->nr_pages[count_type]); 2111 2112 if (count_type == F2FS_DIRTY_DENTS || 2113 count_type == F2FS_DIRTY_NODES || 2114 count_type == F2FS_DIRTY_META || 2115 count_type == F2FS_DIRTY_QDATA || 2116 count_type == F2FS_DIRTY_IMETA) 2117 set_sbi_flag(sbi, SBI_IS_DIRTY); 2118 } 2119 2120 static inline void inode_inc_dirty_pages(struct inode *inode) 2121 { 2122 atomic_inc(&F2FS_I(inode)->dirty_pages); 2123 inc_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ? 2124 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA); 2125 if (IS_NOQUOTA(inode)) 2126 inc_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA); 2127 } 2128 2129 static inline void dec_page_count(struct f2fs_sb_info *sbi, int count_type) 2130 { 2131 atomic_dec(&sbi->nr_pages[count_type]); 2132 } 2133 2134 static inline void inode_dec_dirty_pages(struct inode *inode) 2135 { 2136 if (!S_ISDIR(inode->i_mode) && !S_ISREG(inode->i_mode) && 2137 !S_ISLNK(inode->i_mode)) 2138 return; 2139 2140 atomic_dec(&F2FS_I(inode)->dirty_pages); 2141 dec_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ? 2142 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA); 2143 if (IS_NOQUOTA(inode)) 2144 dec_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA); 2145 } 2146 2147 static inline s64 get_pages(struct f2fs_sb_info *sbi, int count_type) 2148 { 2149 return atomic_read(&sbi->nr_pages[count_type]); 2150 } 2151 2152 static inline int get_dirty_pages(struct inode *inode) 2153 { 2154 return atomic_read(&F2FS_I(inode)->dirty_pages); 2155 } 2156 2157 static inline int get_blocktype_secs(struct f2fs_sb_info *sbi, int block_type) 2158 { 2159 unsigned int pages_per_sec = sbi->segs_per_sec * sbi->blocks_per_seg; 2160 unsigned int segs = (get_pages(sbi, block_type) + pages_per_sec - 1) >> 2161 sbi->log_blocks_per_seg; 2162 2163 return segs / sbi->segs_per_sec; 2164 } 2165 2166 static inline block_t valid_user_blocks(struct f2fs_sb_info *sbi) 2167 { 2168 return sbi->total_valid_block_count; 2169 } 2170 2171 static inline block_t discard_blocks(struct f2fs_sb_info *sbi) 2172 { 2173 return sbi->discard_blks; 2174 } 2175 2176 static inline unsigned long __bitmap_size(struct f2fs_sb_info *sbi, int flag) 2177 { 2178 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi); 2179 2180 /* return NAT or SIT bitmap */ 2181 if (flag == NAT_BITMAP) 2182 return le32_to_cpu(ckpt->nat_ver_bitmap_bytesize); 2183 else if (flag == SIT_BITMAP) 2184 return le32_to_cpu(ckpt->sit_ver_bitmap_bytesize); 2185 2186 return 0; 2187 } 2188 2189 static inline block_t __cp_payload(struct f2fs_sb_info *sbi) 2190 { 2191 return le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_payload); 2192 } 2193 2194 static inline void *__bitmap_ptr(struct f2fs_sb_info *sbi, int flag) 2195 { 2196 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi); 2197 int offset; 2198 2199 if (is_set_ckpt_flags(sbi, CP_LARGE_NAT_BITMAP_FLAG)) { 2200 offset = (flag == SIT_BITMAP) ? 2201 le32_to_cpu(ckpt->nat_ver_bitmap_bytesize) : 0; 2202 /* 2203 * if large_nat_bitmap feature is enabled, leave checksum 2204 * protection for all nat/sit bitmaps. 2205 */ 2206 return &ckpt->sit_nat_version_bitmap + offset + sizeof(__le32); 2207 } 2208 2209 if (__cp_payload(sbi) > 0) { 2210 if (flag == NAT_BITMAP) 2211 return &ckpt->sit_nat_version_bitmap; 2212 else 2213 return (unsigned char *)ckpt + F2FS_BLKSIZE; 2214 } else { 2215 offset = (flag == NAT_BITMAP) ? 2216 le32_to_cpu(ckpt->sit_ver_bitmap_bytesize) : 0; 2217 return &ckpt->sit_nat_version_bitmap + offset; 2218 } 2219 } 2220 2221 static inline block_t __start_cp_addr(struct f2fs_sb_info *sbi) 2222 { 2223 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr); 2224 2225 if (sbi->cur_cp_pack == 2) 2226 start_addr += sbi->blocks_per_seg; 2227 return start_addr; 2228 } 2229 2230 static inline block_t __start_cp_next_addr(struct f2fs_sb_info *sbi) 2231 { 2232 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr); 2233 2234 if (sbi->cur_cp_pack == 1) 2235 start_addr += sbi->blocks_per_seg; 2236 return start_addr; 2237 } 2238 2239 static inline void __set_cp_next_pack(struct f2fs_sb_info *sbi) 2240 { 2241 sbi->cur_cp_pack = (sbi->cur_cp_pack == 1) ? 2 : 1; 2242 } 2243 2244 static inline block_t __start_sum_addr(struct f2fs_sb_info *sbi) 2245 { 2246 return le32_to_cpu(F2FS_CKPT(sbi)->cp_pack_start_sum); 2247 } 2248 2249 static inline int inc_valid_node_count(struct f2fs_sb_info *sbi, 2250 struct inode *inode, bool is_inode) 2251 { 2252 block_t valid_block_count; 2253 unsigned int valid_node_count, user_block_count; 2254 int err; 2255 2256 if (is_inode) { 2257 if (inode) { 2258 err = dquot_alloc_inode(inode); 2259 if (err) 2260 return err; 2261 } 2262 } else { 2263 err = dquot_reserve_block(inode, 1); 2264 if (err) 2265 return err; 2266 } 2267 2268 if (time_to_inject(sbi, FAULT_BLOCK)) { 2269 f2fs_show_injection_info(sbi, FAULT_BLOCK); 2270 goto enospc; 2271 } 2272 2273 spin_lock(&sbi->stat_lock); 2274 2275 valid_block_count = sbi->total_valid_block_count + 2276 sbi->current_reserved_blocks + 1; 2277 2278 if (!__allow_reserved_blocks(sbi, inode, false)) 2279 valid_block_count += F2FS_OPTION(sbi).root_reserved_blocks; 2280 user_block_count = sbi->user_block_count; 2281 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) 2282 user_block_count -= sbi->unusable_block_count; 2283 2284 if (unlikely(valid_block_count > user_block_count)) { 2285 spin_unlock(&sbi->stat_lock); 2286 goto enospc; 2287 } 2288 2289 valid_node_count = sbi->total_valid_node_count + 1; 2290 if (unlikely(valid_node_count > sbi->total_node_count)) { 2291 spin_unlock(&sbi->stat_lock); 2292 goto enospc; 2293 } 2294 2295 sbi->total_valid_node_count++; 2296 sbi->total_valid_block_count++; 2297 spin_unlock(&sbi->stat_lock); 2298 2299 if (inode) { 2300 if (is_inode) 2301 f2fs_mark_inode_dirty_sync(inode, true); 2302 else 2303 f2fs_i_blocks_write(inode, 1, true, true); 2304 } 2305 2306 percpu_counter_inc(&sbi->alloc_valid_block_count); 2307 return 0; 2308 2309 enospc: 2310 if (is_inode) { 2311 if (inode) 2312 dquot_free_inode(inode); 2313 } else { 2314 dquot_release_reservation_block(inode, 1); 2315 } 2316 return -ENOSPC; 2317 } 2318 2319 static inline void dec_valid_node_count(struct f2fs_sb_info *sbi, 2320 struct inode *inode, bool is_inode) 2321 { 2322 spin_lock(&sbi->stat_lock); 2323 2324 f2fs_bug_on(sbi, !sbi->total_valid_block_count); 2325 f2fs_bug_on(sbi, !sbi->total_valid_node_count); 2326 2327 sbi->total_valid_node_count--; 2328 sbi->total_valid_block_count--; 2329 if (sbi->reserved_blocks && 2330 sbi->current_reserved_blocks < sbi->reserved_blocks) 2331 sbi->current_reserved_blocks++; 2332 2333 spin_unlock(&sbi->stat_lock); 2334 2335 if (is_inode) { 2336 dquot_free_inode(inode); 2337 } else { 2338 if (unlikely(inode->i_blocks == 0)) { 2339 f2fs_warn(sbi, "dec_valid_node_count: inconsistent i_blocks, ino:%lu, iblocks:%llu", 2340 inode->i_ino, 2341 (unsigned long long)inode->i_blocks); 2342 set_sbi_flag(sbi, SBI_NEED_FSCK); 2343 return; 2344 } 2345 f2fs_i_blocks_write(inode, 1, false, true); 2346 } 2347 } 2348 2349 static inline unsigned int valid_node_count(struct f2fs_sb_info *sbi) 2350 { 2351 return sbi->total_valid_node_count; 2352 } 2353 2354 static inline void inc_valid_inode_count(struct f2fs_sb_info *sbi) 2355 { 2356 percpu_counter_inc(&sbi->total_valid_inode_count); 2357 } 2358 2359 static inline void dec_valid_inode_count(struct f2fs_sb_info *sbi) 2360 { 2361 percpu_counter_dec(&sbi->total_valid_inode_count); 2362 } 2363 2364 static inline s64 valid_inode_count(struct f2fs_sb_info *sbi) 2365 { 2366 return percpu_counter_sum_positive(&sbi->total_valid_inode_count); 2367 } 2368 2369 static inline struct page *f2fs_grab_cache_page(struct address_space *mapping, 2370 pgoff_t index, bool for_write) 2371 { 2372 struct page *page; 2373 2374 if (IS_ENABLED(CONFIG_F2FS_FAULT_INJECTION)) { 2375 if (!for_write) 2376 page = find_get_page_flags(mapping, index, 2377 FGP_LOCK | FGP_ACCESSED); 2378 else 2379 page = find_lock_page(mapping, index); 2380 if (page) 2381 return page; 2382 2383 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_ALLOC)) { 2384 f2fs_show_injection_info(F2FS_M_SB(mapping), 2385 FAULT_PAGE_ALLOC); 2386 return NULL; 2387 } 2388 } 2389 2390 if (!for_write) 2391 return grab_cache_page(mapping, index); 2392 return grab_cache_page_write_begin(mapping, index, AOP_FLAG_NOFS); 2393 } 2394 2395 static inline struct page *f2fs_pagecache_get_page( 2396 struct address_space *mapping, pgoff_t index, 2397 int fgp_flags, gfp_t gfp_mask) 2398 { 2399 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_GET)) { 2400 f2fs_show_injection_info(F2FS_M_SB(mapping), FAULT_PAGE_GET); 2401 return NULL; 2402 } 2403 2404 return pagecache_get_page(mapping, index, fgp_flags, gfp_mask); 2405 } 2406 2407 static inline void f2fs_copy_page(struct page *src, struct page *dst) 2408 { 2409 char *src_kaddr = kmap(src); 2410 char *dst_kaddr = kmap(dst); 2411 2412 memcpy(dst_kaddr, src_kaddr, PAGE_SIZE); 2413 kunmap(dst); 2414 kunmap(src); 2415 } 2416 2417 static inline void f2fs_put_page(struct page *page, int unlock) 2418 { 2419 if (!page) 2420 return; 2421 2422 if (unlock) { 2423 f2fs_bug_on(F2FS_P_SB(page), !PageLocked(page)); 2424 unlock_page(page); 2425 } 2426 put_page(page); 2427 } 2428 2429 static inline void f2fs_put_dnode(struct dnode_of_data *dn) 2430 { 2431 if (dn->node_page) 2432 f2fs_put_page(dn->node_page, 1); 2433 if (dn->inode_page && dn->node_page != dn->inode_page) 2434 f2fs_put_page(dn->inode_page, 0); 2435 dn->node_page = NULL; 2436 dn->inode_page = NULL; 2437 } 2438 2439 static inline struct kmem_cache *f2fs_kmem_cache_create(const char *name, 2440 size_t size) 2441 { 2442 return kmem_cache_create(name, size, 0, SLAB_RECLAIM_ACCOUNT, NULL); 2443 } 2444 2445 static inline void *f2fs_kmem_cache_alloc(struct kmem_cache *cachep, 2446 gfp_t flags) 2447 { 2448 void *entry; 2449 2450 entry = kmem_cache_alloc(cachep, flags); 2451 if (!entry) 2452 entry = kmem_cache_alloc(cachep, flags | __GFP_NOFAIL); 2453 return entry; 2454 } 2455 2456 static inline bool is_idle(struct f2fs_sb_info *sbi, int type) 2457 { 2458 if (sbi->gc_mode == GC_URGENT) 2459 return true; 2460 2461 if (get_pages(sbi, F2FS_RD_DATA) || get_pages(sbi, F2FS_RD_NODE) || 2462 get_pages(sbi, F2FS_RD_META) || get_pages(sbi, F2FS_WB_DATA) || 2463 get_pages(sbi, F2FS_WB_CP_DATA) || 2464 get_pages(sbi, F2FS_DIO_READ) || 2465 get_pages(sbi, F2FS_DIO_WRITE)) 2466 return false; 2467 2468 if (type != DISCARD_TIME && SM_I(sbi) && SM_I(sbi)->dcc_info && 2469 atomic_read(&SM_I(sbi)->dcc_info->queued_discard)) 2470 return false; 2471 2472 if (SM_I(sbi) && SM_I(sbi)->fcc_info && 2473 atomic_read(&SM_I(sbi)->fcc_info->queued_flush)) 2474 return false; 2475 2476 return f2fs_time_over(sbi, type); 2477 } 2478 2479 static inline void f2fs_radix_tree_insert(struct radix_tree_root *root, 2480 unsigned long index, void *item) 2481 { 2482 while (radix_tree_insert(root, index, item)) 2483 cond_resched(); 2484 } 2485 2486 #define RAW_IS_INODE(p) ((p)->footer.nid == (p)->footer.ino) 2487 2488 static inline bool IS_INODE(struct page *page) 2489 { 2490 struct f2fs_node *p = F2FS_NODE(page); 2491 2492 return RAW_IS_INODE(p); 2493 } 2494 2495 static inline int offset_in_addr(struct f2fs_inode *i) 2496 { 2497 return (i->i_inline & F2FS_EXTRA_ATTR) ? 2498 (le16_to_cpu(i->i_extra_isize) / sizeof(__le32)) : 0; 2499 } 2500 2501 static inline __le32 *blkaddr_in_node(struct f2fs_node *node) 2502 { 2503 return RAW_IS_INODE(node) ? node->i.i_addr : node->dn.addr; 2504 } 2505 2506 static inline int f2fs_has_extra_attr(struct inode *inode); 2507 static inline block_t data_blkaddr(struct inode *inode, 2508 struct page *node_page, unsigned int offset) 2509 { 2510 struct f2fs_node *raw_node; 2511 __le32 *addr_array; 2512 int base = 0; 2513 bool is_inode = IS_INODE(node_page); 2514 2515 raw_node = F2FS_NODE(node_page); 2516 2517 if (is_inode) { 2518 if (!inode) 2519 /* from GC path only */ 2520 base = offset_in_addr(&raw_node->i); 2521 else if (f2fs_has_extra_attr(inode)) 2522 base = get_extra_isize(inode); 2523 } 2524 2525 addr_array = blkaddr_in_node(raw_node); 2526 return le32_to_cpu(addr_array[base + offset]); 2527 } 2528 2529 static inline block_t f2fs_data_blkaddr(struct dnode_of_data *dn) 2530 { 2531 return data_blkaddr(dn->inode, dn->node_page, dn->ofs_in_node); 2532 } 2533 2534 static inline int f2fs_test_bit(unsigned int nr, char *addr) 2535 { 2536 int mask; 2537 2538 addr += (nr >> 3); 2539 mask = 1 << (7 - (nr & 0x07)); 2540 return mask & *addr; 2541 } 2542 2543 static inline void f2fs_set_bit(unsigned int nr, char *addr) 2544 { 2545 int mask; 2546 2547 addr += (nr >> 3); 2548 mask = 1 << (7 - (nr & 0x07)); 2549 *addr |= mask; 2550 } 2551 2552 static inline void f2fs_clear_bit(unsigned int nr, char *addr) 2553 { 2554 int mask; 2555 2556 addr += (nr >> 3); 2557 mask = 1 << (7 - (nr & 0x07)); 2558 *addr &= ~mask; 2559 } 2560 2561 static inline int f2fs_test_and_set_bit(unsigned int nr, char *addr) 2562 { 2563 int mask; 2564 int ret; 2565 2566 addr += (nr >> 3); 2567 mask = 1 << (7 - (nr & 0x07)); 2568 ret = mask & *addr; 2569 *addr |= mask; 2570 return ret; 2571 } 2572 2573 static inline int f2fs_test_and_clear_bit(unsigned int nr, char *addr) 2574 { 2575 int mask; 2576 int ret; 2577 2578 addr += (nr >> 3); 2579 mask = 1 << (7 - (nr & 0x07)); 2580 ret = mask & *addr; 2581 *addr &= ~mask; 2582 return ret; 2583 } 2584 2585 static inline void f2fs_change_bit(unsigned int nr, char *addr) 2586 { 2587 int mask; 2588 2589 addr += (nr >> 3); 2590 mask = 1 << (7 - (nr & 0x07)); 2591 *addr ^= mask; 2592 } 2593 2594 /* 2595 * On-disk inode flags (f2fs_inode::i_flags) 2596 */ 2597 #define F2FS_COMPR_FL 0x00000004 /* Compress file */ 2598 #define F2FS_SYNC_FL 0x00000008 /* Synchronous updates */ 2599 #define F2FS_IMMUTABLE_FL 0x00000010 /* Immutable file */ 2600 #define F2FS_APPEND_FL 0x00000020 /* writes to file may only append */ 2601 #define F2FS_NODUMP_FL 0x00000040 /* do not dump file */ 2602 #define F2FS_NOATIME_FL 0x00000080 /* do not update atime */ 2603 #define F2FS_NOCOMP_FL 0x00000400 /* Don't compress */ 2604 #define F2FS_INDEX_FL 0x00001000 /* hash-indexed directory */ 2605 #define F2FS_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */ 2606 #define F2FS_PROJINHERIT_FL 0x20000000 /* Create with parents projid */ 2607 #define F2FS_CASEFOLD_FL 0x40000000 /* Casefolded file */ 2608 2609 /* Flags that should be inherited by new inodes from their parent. */ 2610 #define F2FS_FL_INHERITED (F2FS_SYNC_FL | F2FS_NODUMP_FL | F2FS_NOATIME_FL | \ 2611 F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \ 2612 F2FS_CASEFOLD_FL | F2FS_COMPR_FL | F2FS_NOCOMP_FL) 2613 2614 /* Flags that are appropriate for regular files (all but dir-specific ones). */ 2615 #define F2FS_REG_FLMASK (~(F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \ 2616 F2FS_CASEFOLD_FL)) 2617 2618 /* Flags that are appropriate for non-directories/regular files. */ 2619 #define F2FS_OTHER_FLMASK (F2FS_NODUMP_FL | F2FS_NOATIME_FL) 2620 2621 static inline __u32 f2fs_mask_flags(umode_t mode, __u32 flags) 2622 { 2623 if (S_ISDIR(mode)) 2624 return flags; 2625 else if (S_ISREG(mode)) 2626 return flags & F2FS_REG_FLMASK; 2627 else 2628 return flags & F2FS_OTHER_FLMASK; 2629 } 2630 2631 static inline void __mark_inode_dirty_flag(struct inode *inode, 2632 int flag, bool set) 2633 { 2634 switch (flag) { 2635 case FI_INLINE_XATTR: 2636 case FI_INLINE_DATA: 2637 case FI_INLINE_DENTRY: 2638 case FI_NEW_INODE: 2639 if (set) 2640 return; 2641 /* fall through */ 2642 case FI_DATA_EXIST: 2643 case FI_INLINE_DOTS: 2644 case FI_PIN_FILE: 2645 f2fs_mark_inode_dirty_sync(inode, true); 2646 } 2647 } 2648 2649 static inline void set_inode_flag(struct inode *inode, int flag) 2650 { 2651 test_and_set_bit(flag, F2FS_I(inode)->flags); 2652 __mark_inode_dirty_flag(inode, flag, true); 2653 } 2654 2655 static inline int is_inode_flag_set(struct inode *inode, int flag) 2656 { 2657 return test_bit(flag, F2FS_I(inode)->flags); 2658 } 2659 2660 static inline void clear_inode_flag(struct inode *inode, int flag) 2661 { 2662 test_and_clear_bit(flag, F2FS_I(inode)->flags); 2663 __mark_inode_dirty_flag(inode, flag, false); 2664 } 2665 2666 static inline bool f2fs_verity_in_progress(struct inode *inode) 2667 { 2668 return IS_ENABLED(CONFIG_FS_VERITY) && 2669 is_inode_flag_set(inode, FI_VERITY_IN_PROGRESS); 2670 } 2671 2672 static inline void set_acl_inode(struct inode *inode, umode_t mode) 2673 { 2674 F2FS_I(inode)->i_acl_mode = mode; 2675 set_inode_flag(inode, FI_ACL_MODE); 2676 f2fs_mark_inode_dirty_sync(inode, false); 2677 } 2678 2679 static inline void f2fs_i_links_write(struct inode *inode, bool inc) 2680 { 2681 if (inc) 2682 inc_nlink(inode); 2683 else 2684 drop_nlink(inode); 2685 f2fs_mark_inode_dirty_sync(inode, true); 2686 } 2687 2688 static inline void f2fs_i_blocks_write(struct inode *inode, 2689 block_t diff, bool add, bool claim) 2690 { 2691 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE); 2692 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER); 2693 2694 /* add = 1, claim = 1 should be dquot_reserve_block in pair */ 2695 if (add) { 2696 if (claim) 2697 dquot_claim_block(inode, diff); 2698 else 2699 dquot_alloc_block_nofail(inode, diff); 2700 } else { 2701 dquot_free_block(inode, diff); 2702 } 2703 2704 f2fs_mark_inode_dirty_sync(inode, true); 2705 if (clean || recover) 2706 set_inode_flag(inode, FI_AUTO_RECOVER); 2707 } 2708 2709 static inline void f2fs_i_size_write(struct inode *inode, loff_t i_size) 2710 { 2711 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE); 2712 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER); 2713 2714 if (i_size_read(inode) == i_size) 2715 return; 2716 2717 i_size_write(inode, i_size); 2718 f2fs_mark_inode_dirty_sync(inode, true); 2719 if (clean || recover) 2720 set_inode_flag(inode, FI_AUTO_RECOVER); 2721 } 2722 2723 static inline void f2fs_i_depth_write(struct inode *inode, unsigned int depth) 2724 { 2725 F2FS_I(inode)->i_current_depth = depth; 2726 f2fs_mark_inode_dirty_sync(inode, true); 2727 } 2728 2729 static inline void f2fs_i_gc_failures_write(struct inode *inode, 2730 unsigned int count) 2731 { 2732 F2FS_I(inode)->i_gc_failures[GC_FAILURE_PIN] = count; 2733 f2fs_mark_inode_dirty_sync(inode, true); 2734 } 2735 2736 static inline void f2fs_i_xnid_write(struct inode *inode, nid_t xnid) 2737 { 2738 F2FS_I(inode)->i_xattr_nid = xnid; 2739 f2fs_mark_inode_dirty_sync(inode, true); 2740 } 2741 2742 static inline void f2fs_i_pino_write(struct inode *inode, nid_t pino) 2743 { 2744 F2FS_I(inode)->i_pino = pino; 2745 f2fs_mark_inode_dirty_sync(inode, true); 2746 } 2747 2748 static inline void get_inline_info(struct inode *inode, struct f2fs_inode *ri) 2749 { 2750 struct f2fs_inode_info *fi = F2FS_I(inode); 2751 2752 if (ri->i_inline & F2FS_INLINE_XATTR) 2753 set_bit(FI_INLINE_XATTR, fi->flags); 2754 if (ri->i_inline & F2FS_INLINE_DATA) 2755 set_bit(FI_INLINE_DATA, fi->flags); 2756 if (ri->i_inline & F2FS_INLINE_DENTRY) 2757 set_bit(FI_INLINE_DENTRY, fi->flags); 2758 if (ri->i_inline & F2FS_DATA_EXIST) 2759 set_bit(FI_DATA_EXIST, fi->flags); 2760 if (ri->i_inline & F2FS_INLINE_DOTS) 2761 set_bit(FI_INLINE_DOTS, fi->flags); 2762 if (ri->i_inline & F2FS_EXTRA_ATTR) 2763 set_bit(FI_EXTRA_ATTR, fi->flags); 2764 if (ri->i_inline & F2FS_PIN_FILE) 2765 set_bit(FI_PIN_FILE, fi->flags); 2766 } 2767 2768 static inline void set_raw_inline(struct inode *inode, struct f2fs_inode *ri) 2769 { 2770 ri->i_inline = 0; 2771 2772 if (is_inode_flag_set(inode, FI_INLINE_XATTR)) 2773 ri->i_inline |= F2FS_INLINE_XATTR; 2774 if (is_inode_flag_set(inode, FI_INLINE_DATA)) 2775 ri->i_inline |= F2FS_INLINE_DATA; 2776 if (is_inode_flag_set(inode, FI_INLINE_DENTRY)) 2777 ri->i_inline |= F2FS_INLINE_DENTRY; 2778 if (is_inode_flag_set(inode, FI_DATA_EXIST)) 2779 ri->i_inline |= F2FS_DATA_EXIST; 2780 if (is_inode_flag_set(inode, FI_INLINE_DOTS)) 2781 ri->i_inline |= F2FS_INLINE_DOTS; 2782 if (is_inode_flag_set(inode, FI_EXTRA_ATTR)) 2783 ri->i_inline |= F2FS_EXTRA_ATTR; 2784 if (is_inode_flag_set(inode, FI_PIN_FILE)) 2785 ri->i_inline |= F2FS_PIN_FILE; 2786 } 2787 2788 static inline int f2fs_has_extra_attr(struct inode *inode) 2789 { 2790 return is_inode_flag_set(inode, FI_EXTRA_ATTR); 2791 } 2792 2793 static inline int f2fs_has_inline_xattr(struct inode *inode) 2794 { 2795 return is_inode_flag_set(inode, FI_INLINE_XATTR); 2796 } 2797 2798 static inline int f2fs_compressed_file(struct inode *inode) 2799 { 2800 return S_ISREG(inode->i_mode) && 2801 is_inode_flag_set(inode, FI_COMPRESSED_FILE); 2802 } 2803 2804 static inline unsigned int addrs_per_inode(struct inode *inode) 2805 { 2806 unsigned int addrs = CUR_ADDRS_PER_INODE(inode) - 2807 get_inline_xattr_addrs(inode); 2808 2809 if (!f2fs_compressed_file(inode)) 2810 return addrs; 2811 return ALIGN_DOWN(addrs, F2FS_I(inode)->i_cluster_size); 2812 } 2813 2814 static inline unsigned int addrs_per_block(struct inode *inode) 2815 { 2816 if (!f2fs_compressed_file(inode)) 2817 return DEF_ADDRS_PER_BLOCK; 2818 return ALIGN_DOWN(DEF_ADDRS_PER_BLOCK, F2FS_I(inode)->i_cluster_size); 2819 } 2820 2821 static inline void *inline_xattr_addr(struct inode *inode, struct page *page) 2822 { 2823 struct f2fs_inode *ri = F2FS_INODE(page); 2824 2825 return (void *)&(ri->i_addr[DEF_ADDRS_PER_INODE - 2826 get_inline_xattr_addrs(inode)]); 2827 } 2828 2829 static inline int inline_xattr_size(struct inode *inode) 2830 { 2831 if (f2fs_has_inline_xattr(inode)) 2832 return get_inline_xattr_addrs(inode) * sizeof(__le32); 2833 return 0; 2834 } 2835 2836 static inline int f2fs_has_inline_data(struct inode *inode) 2837 { 2838 return is_inode_flag_set(inode, FI_INLINE_DATA); 2839 } 2840 2841 static inline int f2fs_exist_data(struct inode *inode) 2842 { 2843 return is_inode_flag_set(inode, FI_DATA_EXIST); 2844 } 2845 2846 static inline int f2fs_has_inline_dots(struct inode *inode) 2847 { 2848 return is_inode_flag_set(inode, FI_INLINE_DOTS); 2849 } 2850 2851 static inline int f2fs_is_mmap_file(struct inode *inode) 2852 { 2853 return is_inode_flag_set(inode, FI_MMAP_FILE); 2854 } 2855 2856 static inline bool f2fs_is_pinned_file(struct inode *inode) 2857 { 2858 return is_inode_flag_set(inode, FI_PIN_FILE); 2859 } 2860 2861 static inline bool f2fs_is_atomic_file(struct inode *inode) 2862 { 2863 return is_inode_flag_set(inode, FI_ATOMIC_FILE); 2864 } 2865 2866 static inline bool f2fs_is_commit_atomic_write(struct inode *inode) 2867 { 2868 return is_inode_flag_set(inode, FI_ATOMIC_COMMIT); 2869 } 2870 2871 static inline bool f2fs_is_volatile_file(struct inode *inode) 2872 { 2873 return is_inode_flag_set(inode, FI_VOLATILE_FILE); 2874 } 2875 2876 static inline bool f2fs_is_first_block_written(struct inode *inode) 2877 { 2878 return is_inode_flag_set(inode, FI_FIRST_BLOCK_WRITTEN); 2879 } 2880 2881 static inline bool f2fs_is_drop_cache(struct inode *inode) 2882 { 2883 return is_inode_flag_set(inode, FI_DROP_CACHE); 2884 } 2885 2886 static inline void *inline_data_addr(struct inode *inode, struct page *page) 2887 { 2888 struct f2fs_inode *ri = F2FS_INODE(page); 2889 int extra_size = get_extra_isize(inode); 2890 2891 return (void *)&(ri->i_addr[extra_size + DEF_INLINE_RESERVED_SIZE]); 2892 } 2893 2894 static inline int f2fs_has_inline_dentry(struct inode *inode) 2895 { 2896 return is_inode_flag_set(inode, FI_INLINE_DENTRY); 2897 } 2898 2899 static inline int is_file(struct inode *inode, int type) 2900 { 2901 return F2FS_I(inode)->i_advise & type; 2902 } 2903 2904 static inline void set_file(struct inode *inode, int type) 2905 { 2906 F2FS_I(inode)->i_advise |= type; 2907 f2fs_mark_inode_dirty_sync(inode, true); 2908 } 2909 2910 static inline void clear_file(struct inode *inode, int type) 2911 { 2912 F2FS_I(inode)->i_advise &= ~type; 2913 f2fs_mark_inode_dirty_sync(inode, true); 2914 } 2915 2916 static inline bool f2fs_is_time_consistent(struct inode *inode) 2917 { 2918 if (!timespec64_equal(F2FS_I(inode)->i_disk_time, &inode->i_atime)) 2919 return false; 2920 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 1, &inode->i_ctime)) 2921 return false; 2922 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 2, &inode->i_mtime)) 2923 return false; 2924 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 3, 2925 &F2FS_I(inode)->i_crtime)) 2926 return false; 2927 return true; 2928 } 2929 2930 static inline bool f2fs_skip_inode_update(struct inode *inode, int dsync) 2931 { 2932 bool ret; 2933 2934 if (dsync) { 2935 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 2936 2937 spin_lock(&sbi->inode_lock[DIRTY_META]); 2938 ret = list_empty(&F2FS_I(inode)->gdirty_list); 2939 spin_unlock(&sbi->inode_lock[DIRTY_META]); 2940 return ret; 2941 } 2942 if (!is_inode_flag_set(inode, FI_AUTO_RECOVER) || 2943 file_keep_isize(inode) || 2944 i_size_read(inode) & ~PAGE_MASK) 2945 return false; 2946 2947 if (!f2fs_is_time_consistent(inode)) 2948 return false; 2949 2950 spin_lock(&F2FS_I(inode)->i_size_lock); 2951 ret = F2FS_I(inode)->last_disk_size == i_size_read(inode); 2952 spin_unlock(&F2FS_I(inode)->i_size_lock); 2953 2954 return ret; 2955 } 2956 2957 static inline bool f2fs_readonly(struct super_block *sb) 2958 { 2959 return sb_rdonly(sb); 2960 } 2961 2962 static inline bool f2fs_cp_error(struct f2fs_sb_info *sbi) 2963 { 2964 return is_set_ckpt_flags(sbi, CP_ERROR_FLAG); 2965 } 2966 2967 static inline bool is_dot_dotdot(const u8 *name, size_t len) 2968 { 2969 if (len == 1 && name[0] == '.') 2970 return true; 2971 2972 if (len == 2 && name[0] == '.' && name[1] == '.') 2973 return true; 2974 2975 return false; 2976 } 2977 2978 static inline bool f2fs_may_extent_tree(struct inode *inode) 2979 { 2980 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 2981 2982 if (!test_opt(sbi, EXTENT_CACHE) || 2983 is_inode_flag_set(inode, FI_NO_EXTENT) || 2984 is_inode_flag_set(inode, FI_COMPRESSED_FILE)) 2985 return false; 2986 2987 /* 2988 * for recovered files during mount do not create extents 2989 * if shrinker is not registered. 2990 */ 2991 if (list_empty(&sbi->s_list)) 2992 return false; 2993 2994 return S_ISREG(inode->i_mode); 2995 } 2996 2997 static inline void *f2fs_kmalloc(struct f2fs_sb_info *sbi, 2998 size_t size, gfp_t flags) 2999 { 3000 void *ret; 3001 3002 if (time_to_inject(sbi, FAULT_KMALLOC)) { 3003 f2fs_show_injection_info(sbi, FAULT_KMALLOC); 3004 return NULL; 3005 } 3006 3007 ret = kmalloc(size, flags); 3008 if (ret) 3009 return ret; 3010 3011 return kvmalloc(size, flags); 3012 } 3013 3014 static inline void *f2fs_kzalloc(struct f2fs_sb_info *sbi, 3015 size_t size, gfp_t flags) 3016 { 3017 return f2fs_kmalloc(sbi, size, flags | __GFP_ZERO); 3018 } 3019 3020 static inline void *f2fs_kvmalloc(struct f2fs_sb_info *sbi, 3021 size_t size, gfp_t flags) 3022 { 3023 if (time_to_inject(sbi, FAULT_KVMALLOC)) { 3024 f2fs_show_injection_info(sbi, FAULT_KVMALLOC); 3025 return NULL; 3026 } 3027 3028 return kvmalloc(size, flags); 3029 } 3030 3031 static inline void *f2fs_kvzalloc(struct f2fs_sb_info *sbi, 3032 size_t size, gfp_t flags) 3033 { 3034 return f2fs_kvmalloc(sbi, size, flags | __GFP_ZERO); 3035 } 3036 3037 static inline int get_extra_isize(struct inode *inode) 3038 { 3039 return F2FS_I(inode)->i_extra_isize / sizeof(__le32); 3040 } 3041 3042 static inline int get_inline_xattr_addrs(struct inode *inode) 3043 { 3044 return F2FS_I(inode)->i_inline_xattr_size; 3045 } 3046 3047 #define f2fs_get_inode_mode(i) \ 3048 ((is_inode_flag_set(i, FI_ACL_MODE)) ? \ 3049 (F2FS_I(i)->i_acl_mode) : ((i)->i_mode)) 3050 3051 #define F2FS_TOTAL_EXTRA_ATTR_SIZE \ 3052 (offsetof(struct f2fs_inode, i_extra_end) - \ 3053 offsetof(struct f2fs_inode, i_extra_isize)) \ 3054 3055 #define F2FS_OLD_ATTRIBUTE_SIZE (offsetof(struct f2fs_inode, i_addr)) 3056 #define F2FS_FITS_IN_INODE(f2fs_inode, extra_isize, field) \ 3057 ((offsetof(typeof(*(f2fs_inode)), field) + \ 3058 sizeof((f2fs_inode)->field)) \ 3059 <= (F2FS_OLD_ATTRIBUTE_SIZE + (extra_isize))) \ 3060 3061 #define DEFAULT_IOSTAT_PERIOD_MS 3000 3062 #define MIN_IOSTAT_PERIOD_MS 100 3063 /* maximum period of iostat tracing is 1 day */ 3064 #define MAX_IOSTAT_PERIOD_MS 8640000 3065 3066 static inline void f2fs_reset_iostat(struct f2fs_sb_info *sbi) 3067 { 3068 int i; 3069 3070 spin_lock(&sbi->iostat_lock); 3071 for (i = 0; i < NR_IO_TYPE; i++) { 3072 sbi->rw_iostat[i] = 0; 3073 sbi->prev_rw_iostat[i] = 0; 3074 } 3075 spin_unlock(&sbi->iostat_lock); 3076 } 3077 3078 extern void f2fs_record_iostat(struct f2fs_sb_info *sbi); 3079 3080 static inline void f2fs_update_iostat(struct f2fs_sb_info *sbi, 3081 enum iostat_type type, unsigned long long io_bytes) 3082 { 3083 if (!sbi->iostat_enable) 3084 return; 3085 spin_lock(&sbi->iostat_lock); 3086 sbi->rw_iostat[type] += io_bytes; 3087 3088 if (type == APP_WRITE_IO || type == APP_DIRECT_IO) 3089 sbi->rw_iostat[APP_BUFFERED_IO] = 3090 sbi->rw_iostat[APP_WRITE_IO] - 3091 sbi->rw_iostat[APP_DIRECT_IO]; 3092 3093 if (type == APP_READ_IO || type == APP_DIRECT_READ_IO) 3094 sbi->rw_iostat[APP_BUFFERED_READ_IO] = 3095 sbi->rw_iostat[APP_READ_IO] - 3096 sbi->rw_iostat[APP_DIRECT_READ_IO]; 3097 spin_unlock(&sbi->iostat_lock); 3098 3099 f2fs_record_iostat(sbi); 3100 } 3101 3102 #define __is_large_section(sbi) ((sbi)->segs_per_sec > 1) 3103 3104 #define __is_meta_io(fio) (PAGE_TYPE_OF_BIO((fio)->type) == META) 3105 3106 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi, 3107 block_t blkaddr, int type); 3108 static inline void verify_blkaddr(struct f2fs_sb_info *sbi, 3109 block_t blkaddr, int type) 3110 { 3111 if (!f2fs_is_valid_blkaddr(sbi, blkaddr, type)) { 3112 f2fs_err(sbi, "invalid blkaddr: %u, type: %d, run fsck to fix.", 3113 blkaddr, type); 3114 f2fs_bug_on(sbi, 1); 3115 } 3116 } 3117 3118 static inline bool __is_valid_data_blkaddr(block_t blkaddr) 3119 { 3120 if (blkaddr == NEW_ADDR || blkaddr == NULL_ADDR || 3121 blkaddr == COMPRESS_ADDR) 3122 return false; 3123 return true; 3124 } 3125 3126 static inline void f2fs_set_page_private(struct page *page, 3127 unsigned long data) 3128 { 3129 if (PagePrivate(page)) 3130 return; 3131 3132 get_page(page); 3133 SetPagePrivate(page); 3134 set_page_private(page, data); 3135 } 3136 3137 static inline void f2fs_clear_page_private(struct page *page) 3138 { 3139 if (!PagePrivate(page)) 3140 return; 3141 3142 set_page_private(page, 0); 3143 ClearPagePrivate(page); 3144 f2fs_put_page(page, 0); 3145 } 3146 3147 /* 3148 * file.c 3149 */ 3150 int f2fs_sync_file(struct file *file, loff_t start, loff_t end, int datasync); 3151 void f2fs_truncate_data_blocks(struct dnode_of_data *dn); 3152 int f2fs_do_truncate_blocks(struct inode *inode, u64 from, bool lock); 3153 int f2fs_truncate_blocks(struct inode *inode, u64 from, bool lock); 3154 int f2fs_truncate(struct inode *inode); 3155 int f2fs_getattr(const struct path *path, struct kstat *stat, 3156 u32 request_mask, unsigned int flags); 3157 int f2fs_setattr(struct dentry *dentry, struct iattr *attr); 3158 int f2fs_truncate_hole(struct inode *inode, pgoff_t pg_start, pgoff_t pg_end); 3159 void f2fs_truncate_data_blocks_range(struct dnode_of_data *dn, int count); 3160 int f2fs_precache_extents(struct inode *inode); 3161 long f2fs_ioctl(struct file *filp, unsigned int cmd, unsigned long arg); 3162 long f2fs_compat_ioctl(struct file *file, unsigned int cmd, unsigned long arg); 3163 int f2fs_transfer_project_quota(struct inode *inode, kprojid_t kprojid); 3164 int f2fs_pin_file_control(struct inode *inode, bool inc); 3165 3166 /* 3167 * inode.c 3168 */ 3169 void f2fs_set_inode_flags(struct inode *inode); 3170 bool f2fs_inode_chksum_verify(struct f2fs_sb_info *sbi, struct page *page); 3171 void f2fs_inode_chksum_set(struct f2fs_sb_info *sbi, struct page *page); 3172 struct inode *f2fs_iget(struct super_block *sb, unsigned long ino); 3173 struct inode *f2fs_iget_retry(struct super_block *sb, unsigned long ino); 3174 int f2fs_try_to_free_nats(struct f2fs_sb_info *sbi, int nr_shrink); 3175 void f2fs_update_inode(struct inode *inode, struct page *node_page); 3176 void f2fs_update_inode_page(struct inode *inode); 3177 int f2fs_write_inode(struct inode *inode, struct writeback_control *wbc); 3178 void f2fs_evict_inode(struct inode *inode); 3179 void f2fs_handle_failed_inode(struct inode *inode); 3180 3181 /* 3182 * namei.c 3183 */ 3184 int f2fs_update_extension_list(struct f2fs_sb_info *sbi, const char *name, 3185 bool hot, bool set); 3186 struct dentry *f2fs_get_parent(struct dentry *child); 3187 3188 /* 3189 * dir.c 3190 */ 3191 unsigned char f2fs_get_de_type(struct f2fs_dir_entry *de); 3192 int f2fs_init_casefolded_name(const struct inode *dir, 3193 struct f2fs_filename *fname); 3194 int f2fs_setup_filename(struct inode *dir, const struct qstr *iname, 3195 int lookup, struct f2fs_filename *fname); 3196 int f2fs_prepare_lookup(struct inode *dir, struct dentry *dentry, 3197 struct f2fs_filename *fname); 3198 void f2fs_free_filename(struct f2fs_filename *fname); 3199 struct f2fs_dir_entry *f2fs_find_target_dentry(const struct f2fs_dentry_ptr *d, 3200 const struct f2fs_filename *fname, int *max_slots); 3201 int f2fs_fill_dentries(struct dir_context *ctx, struct f2fs_dentry_ptr *d, 3202 unsigned int start_pos, struct fscrypt_str *fstr); 3203 void f2fs_do_make_empty_dir(struct inode *inode, struct inode *parent, 3204 struct f2fs_dentry_ptr *d); 3205 struct page *f2fs_init_inode_metadata(struct inode *inode, struct inode *dir, 3206 const struct f2fs_filename *fname, struct page *dpage); 3207 void f2fs_update_parent_metadata(struct inode *dir, struct inode *inode, 3208 unsigned int current_depth); 3209 int f2fs_room_for_filename(const void *bitmap, int slots, int max_slots); 3210 void f2fs_drop_nlink(struct inode *dir, struct inode *inode); 3211 struct f2fs_dir_entry *__f2fs_find_entry(struct inode *dir, 3212 const struct f2fs_filename *fname, 3213 struct page **res_page); 3214 struct f2fs_dir_entry *f2fs_find_entry(struct inode *dir, 3215 const struct qstr *child, struct page **res_page); 3216 struct f2fs_dir_entry *f2fs_parent_dir(struct inode *dir, struct page **p); 3217 ino_t f2fs_inode_by_name(struct inode *dir, const struct qstr *qstr, 3218 struct page **page); 3219 void f2fs_set_link(struct inode *dir, struct f2fs_dir_entry *de, 3220 struct page *page, struct inode *inode); 3221 bool f2fs_has_enough_room(struct inode *dir, struct page *ipage, 3222 const struct f2fs_filename *fname); 3223 void f2fs_update_dentry(nid_t ino, umode_t mode, struct f2fs_dentry_ptr *d, 3224 const struct fscrypt_str *name, f2fs_hash_t name_hash, 3225 unsigned int bit_pos); 3226 int f2fs_add_regular_entry(struct inode *dir, const struct f2fs_filename *fname, 3227 struct inode *inode, nid_t ino, umode_t mode); 3228 int f2fs_add_dentry(struct inode *dir, const struct f2fs_filename *fname, 3229 struct inode *inode, nid_t ino, umode_t mode); 3230 int f2fs_do_add_link(struct inode *dir, const struct qstr *name, 3231 struct inode *inode, nid_t ino, umode_t mode); 3232 void f2fs_delete_entry(struct f2fs_dir_entry *dentry, struct page *page, 3233 struct inode *dir, struct inode *inode); 3234 int f2fs_do_tmpfile(struct inode *inode, struct inode *dir); 3235 bool f2fs_empty_dir(struct inode *dir); 3236 3237 static inline int f2fs_add_link(struct dentry *dentry, struct inode *inode) 3238 { 3239 return f2fs_do_add_link(d_inode(dentry->d_parent), &dentry->d_name, 3240 inode, inode->i_ino, inode->i_mode); 3241 } 3242 3243 /* 3244 * super.c 3245 */ 3246 int f2fs_inode_dirtied(struct inode *inode, bool sync); 3247 void f2fs_inode_synced(struct inode *inode); 3248 int f2fs_enable_quota_files(struct f2fs_sb_info *sbi, bool rdonly); 3249 int f2fs_quota_sync(struct super_block *sb, int type); 3250 void f2fs_quota_off_umount(struct super_block *sb); 3251 int f2fs_commit_super(struct f2fs_sb_info *sbi, bool recover); 3252 int f2fs_sync_fs(struct super_block *sb, int sync); 3253 int f2fs_sanity_check_ckpt(struct f2fs_sb_info *sbi); 3254 3255 /* 3256 * hash.c 3257 */ 3258 void f2fs_hash_filename(const struct inode *dir, struct f2fs_filename *fname); 3259 3260 /* 3261 * node.c 3262 */ 3263 struct dnode_of_data; 3264 struct node_info; 3265 3266 int f2fs_check_nid_range(struct f2fs_sb_info *sbi, nid_t nid); 3267 bool f2fs_available_free_memory(struct f2fs_sb_info *sbi, int type); 3268 bool f2fs_in_warm_node_list(struct f2fs_sb_info *sbi, struct page *page); 3269 void f2fs_init_fsync_node_info(struct f2fs_sb_info *sbi); 3270 void f2fs_del_fsync_node_entry(struct f2fs_sb_info *sbi, struct page *page); 3271 void f2fs_reset_fsync_node_info(struct f2fs_sb_info *sbi); 3272 int f2fs_need_dentry_mark(struct f2fs_sb_info *sbi, nid_t nid); 3273 bool f2fs_is_checkpointed_node(struct f2fs_sb_info *sbi, nid_t nid); 3274 bool f2fs_need_inode_block_update(struct f2fs_sb_info *sbi, nid_t ino); 3275 int f2fs_get_node_info(struct f2fs_sb_info *sbi, nid_t nid, 3276 struct node_info *ni); 3277 pgoff_t f2fs_get_next_page_offset(struct dnode_of_data *dn, pgoff_t pgofs); 3278 int f2fs_get_dnode_of_data(struct dnode_of_data *dn, pgoff_t index, int mode); 3279 int f2fs_truncate_inode_blocks(struct inode *inode, pgoff_t from); 3280 int f2fs_truncate_xattr_node(struct inode *inode); 3281 int f2fs_wait_on_node_pages_writeback(struct f2fs_sb_info *sbi, 3282 unsigned int seq_id); 3283 int f2fs_remove_inode_page(struct inode *inode); 3284 struct page *f2fs_new_inode_page(struct inode *inode); 3285 struct page *f2fs_new_node_page(struct dnode_of_data *dn, unsigned int ofs); 3286 void f2fs_ra_node_page(struct f2fs_sb_info *sbi, nid_t nid); 3287 struct page *f2fs_get_node_page(struct f2fs_sb_info *sbi, pgoff_t nid); 3288 struct page *f2fs_get_node_page_ra(struct page *parent, int start); 3289 int f2fs_move_node_page(struct page *node_page, int gc_type); 3290 int f2fs_flush_inline_data(struct f2fs_sb_info *sbi); 3291 int f2fs_fsync_node_pages(struct f2fs_sb_info *sbi, struct inode *inode, 3292 struct writeback_control *wbc, bool atomic, 3293 unsigned int *seq_id); 3294 int f2fs_sync_node_pages(struct f2fs_sb_info *sbi, 3295 struct writeback_control *wbc, 3296 bool do_balance, enum iostat_type io_type); 3297 int f2fs_build_free_nids(struct f2fs_sb_info *sbi, bool sync, bool mount); 3298 bool f2fs_alloc_nid(struct f2fs_sb_info *sbi, nid_t *nid); 3299 void f2fs_alloc_nid_done(struct f2fs_sb_info *sbi, nid_t nid); 3300 void f2fs_alloc_nid_failed(struct f2fs_sb_info *sbi, nid_t nid); 3301 int f2fs_try_to_free_nids(struct f2fs_sb_info *sbi, int nr_shrink); 3302 void f2fs_recover_inline_xattr(struct inode *inode, struct page *page); 3303 int f2fs_recover_xattr_data(struct inode *inode, struct page *page); 3304 int f2fs_recover_inode_page(struct f2fs_sb_info *sbi, struct page *page); 3305 int f2fs_restore_node_summary(struct f2fs_sb_info *sbi, 3306 unsigned int segno, struct f2fs_summary_block *sum); 3307 int f2fs_flush_nat_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3308 int f2fs_build_node_manager(struct f2fs_sb_info *sbi); 3309 void f2fs_destroy_node_manager(struct f2fs_sb_info *sbi); 3310 int __init f2fs_create_node_manager_caches(void); 3311 void f2fs_destroy_node_manager_caches(void); 3312 3313 /* 3314 * segment.c 3315 */ 3316 bool f2fs_need_SSR(struct f2fs_sb_info *sbi); 3317 void f2fs_register_inmem_page(struct inode *inode, struct page *page); 3318 void f2fs_drop_inmem_pages_all(struct f2fs_sb_info *sbi, bool gc_failure); 3319 void f2fs_drop_inmem_pages(struct inode *inode); 3320 void f2fs_drop_inmem_page(struct inode *inode, struct page *page); 3321 int f2fs_commit_inmem_pages(struct inode *inode); 3322 void f2fs_balance_fs(struct f2fs_sb_info *sbi, bool need); 3323 void f2fs_balance_fs_bg(struct f2fs_sb_info *sbi, bool from_bg); 3324 int f2fs_issue_flush(struct f2fs_sb_info *sbi, nid_t ino); 3325 int f2fs_create_flush_cmd_control(struct f2fs_sb_info *sbi); 3326 int f2fs_flush_device_cache(struct f2fs_sb_info *sbi); 3327 void f2fs_destroy_flush_cmd_control(struct f2fs_sb_info *sbi, bool free); 3328 void f2fs_invalidate_blocks(struct f2fs_sb_info *sbi, block_t addr); 3329 bool f2fs_is_checkpointed_data(struct f2fs_sb_info *sbi, block_t blkaddr); 3330 void f2fs_drop_discard_cmd(struct f2fs_sb_info *sbi); 3331 void f2fs_stop_discard_thread(struct f2fs_sb_info *sbi); 3332 bool f2fs_issue_discard_timeout(struct f2fs_sb_info *sbi); 3333 void f2fs_clear_prefree_segments(struct f2fs_sb_info *sbi, 3334 struct cp_control *cpc); 3335 void f2fs_dirty_to_prefree(struct f2fs_sb_info *sbi); 3336 block_t f2fs_get_unusable_blocks(struct f2fs_sb_info *sbi); 3337 int f2fs_disable_cp_again(struct f2fs_sb_info *sbi, block_t unusable); 3338 void f2fs_release_discard_addrs(struct f2fs_sb_info *sbi); 3339 int f2fs_npages_for_summary_flush(struct f2fs_sb_info *sbi, bool for_ra); 3340 void allocate_segment_for_resize(struct f2fs_sb_info *sbi, int type, 3341 unsigned int start, unsigned int end); 3342 void f2fs_allocate_new_segments(struct f2fs_sb_info *sbi, int type); 3343 int f2fs_trim_fs(struct f2fs_sb_info *sbi, struct fstrim_range *range); 3344 bool f2fs_exist_trim_candidates(struct f2fs_sb_info *sbi, 3345 struct cp_control *cpc); 3346 struct page *f2fs_get_sum_page(struct f2fs_sb_info *sbi, unsigned int segno); 3347 void f2fs_update_meta_page(struct f2fs_sb_info *sbi, void *src, 3348 block_t blk_addr); 3349 void f2fs_do_write_meta_page(struct f2fs_sb_info *sbi, struct page *page, 3350 enum iostat_type io_type); 3351 void f2fs_do_write_node_page(unsigned int nid, struct f2fs_io_info *fio); 3352 void f2fs_outplace_write_data(struct dnode_of_data *dn, 3353 struct f2fs_io_info *fio); 3354 int f2fs_inplace_write_data(struct f2fs_io_info *fio); 3355 void f2fs_do_replace_block(struct f2fs_sb_info *sbi, struct f2fs_summary *sum, 3356 block_t old_blkaddr, block_t new_blkaddr, 3357 bool recover_curseg, bool recover_newaddr); 3358 void f2fs_replace_block(struct f2fs_sb_info *sbi, struct dnode_of_data *dn, 3359 block_t old_addr, block_t new_addr, 3360 unsigned char version, bool recover_curseg, 3361 bool recover_newaddr); 3362 void f2fs_allocate_data_block(struct f2fs_sb_info *sbi, struct page *page, 3363 block_t old_blkaddr, block_t *new_blkaddr, 3364 struct f2fs_summary *sum, int type, 3365 struct f2fs_io_info *fio, bool add_list); 3366 void f2fs_wait_on_page_writeback(struct page *page, 3367 enum page_type type, bool ordered, bool locked); 3368 void f2fs_wait_on_block_writeback(struct inode *inode, block_t blkaddr); 3369 void f2fs_wait_on_block_writeback_range(struct inode *inode, block_t blkaddr, 3370 block_t len); 3371 void f2fs_write_data_summaries(struct f2fs_sb_info *sbi, block_t start_blk); 3372 void f2fs_write_node_summaries(struct f2fs_sb_info *sbi, block_t start_blk); 3373 int f2fs_lookup_journal_in_cursum(struct f2fs_journal *journal, int type, 3374 unsigned int val, int alloc); 3375 void f2fs_flush_sit_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3376 int f2fs_fix_curseg_write_pointer(struct f2fs_sb_info *sbi); 3377 int f2fs_check_write_pointer(struct f2fs_sb_info *sbi); 3378 int f2fs_build_segment_manager(struct f2fs_sb_info *sbi); 3379 void f2fs_destroy_segment_manager(struct f2fs_sb_info *sbi); 3380 int __init f2fs_create_segment_manager_caches(void); 3381 void f2fs_destroy_segment_manager_caches(void); 3382 int f2fs_rw_hint_to_seg_type(enum rw_hint hint); 3383 enum rw_hint f2fs_io_type_to_rw_hint(struct f2fs_sb_info *sbi, 3384 enum page_type type, enum temp_type temp); 3385 3386 /* 3387 * checkpoint.c 3388 */ 3389 void f2fs_stop_checkpoint(struct f2fs_sb_info *sbi, bool end_io); 3390 struct page *f2fs_grab_meta_page(struct f2fs_sb_info *sbi, pgoff_t index); 3391 struct page *f2fs_get_meta_page(struct f2fs_sb_info *sbi, pgoff_t index); 3392 struct page *f2fs_get_meta_page_nofail(struct f2fs_sb_info *sbi, pgoff_t index); 3393 struct page *f2fs_get_tmp_page(struct f2fs_sb_info *sbi, pgoff_t index); 3394 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi, 3395 block_t blkaddr, int type); 3396 int f2fs_ra_meta_pages(struct f2fs_sb_info *sbi, block_t start, int nrpages, 3397 int type, bool sync); 3398 void f2fs_ra_meta_pages_cond(struct f2fs_sb_info *sbi, pgoff_t index); 3399 long f2fs_sync_meta_pages(struct f2fs_sb_info *sbi, enum page_type type, 3400 long nr_to_write, enum iostat_type io_type); 3401 void f2fs_add_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type); 3402 void f2fs_remove_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type); 3403 void f2fs_release_ino_entry(struct f2fs_sb_info *sbi, bool all); 3404 bool f2fs_exist_written_data(struct f2fs_sb_info *sbi, nid_t ino, int mode); 3405 void f2fs_set_dirty_device(struct f2fs_sb_info *sbi, nid_t ino, 3406 unsigned int devidx, int type); 3407 bool f2fs_is_dirty_device(struct f2fs_sb_info *sbi, nid_t ino, 3408 unsigned int devidx, int type); 3409 int f2fs_sync_inode_meta(struct f2fs_sb_info *sbi); 3410 int f2fs_acquire_orphan_inode(struct f2fs_sb_info *sbi); 3411 void f2fs_release_orphan_inode(struct f2fs_sb_info *sbi); 3412 void f2fs_add_orphan_inode(struct inode *inode); 3413 void f2fs_remove_orphan_inode(struct f2fs_sb_info *sbi, nid_t ino); 3414 int f2fs_recover_orphan_inodes(struct f2fs_sb_info *sbi); 3415 int f2fs_get_valid_checkpoint(struct f2fs_sb_info *sbi); 3416 void f2fs_update_dirty_page(struct inode *inode, struct page *page); 3417 void f2fs_remove_dirty_inode(struct inode *inode); 3418 int f2fs_sync_dirty_inodes(struct f2fs_sb_info *sbi, enum inode_type type); 3419 void f2fs_wait_on_all_pages(struct f2fs_sb_info *sbi, int type); 3420 int f2fs_write_checkpoint(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3421 void f2fs_init_ino_entry_info(struct f2fs_sb_info *sbi); 3422 int __init f2fs_create_checkpoint_caches(void); 3423 void f2fs_destroy_checkpoint_caches(void); 3424 3425 /* 3426 * data.c 3427 */ 3428 int __init f2fs_init_bioset(void); 3429 void f2fs_destroy_bioset(void); 3430 struct bio *f2fs_bio_alloc(struct f2fs_sb_info *sbi, int npages, bool noio); 3431 int f2fs_init_bio_entry_cache(void); 3432 void f2fs_destroy_bio_entry_cache(void); 3433 void f2fs_submit_bio(struct f2fs_sb_info *sbi, 3434 struct bio *bio, enum page_type type); 3435 void f2fs_submit_merged_write(struct f2fs_sb_info *sbi, enum page_type type); 3436 void f2fs_submit_merged_write_cond(struct f2fs_sb_info *sbi, 3437 struct inode *inode, struct page *page, 3438 nid_t ino, enum page_type type); 3439 void f2fs_submit_merged_ipu_write(struct f2fs_sb_info *sbi, 3440 struct bio **bio, struct page *page); 3441 void f2fs_flush_merged_writes(struct f2fs_sb_info *sbi); 3442 int f2fs_submit_page_bio(struct f2fs_io_info *fio); 3443 int f2fs_merge_page_bio(struct f2fs_io_info *fio); 3444 void f2fs_submit_page_write(struct f2fs_io_info *fio); 3445 struct block_device *f2fs_target_device(struct f2fs_sb_info *sbi, 3446 block_t blk_addr, struct bio *bio); 3447 int f2fs_target_device_index(struct f2fs_sb_info *sbi, block_t blkaddr); 3448 void f2fs_set_data_blkaddr(struct dnode_of_data *dn); 3449 void f2fs_update_data_blkaddr(struct dnode_of_data *dn, block_t blkaddr); 3450 int f2fs_reserve_new_blocks(struct dnode_of_data *dn, blkcnt_t count); 3451 int f2fs_reserve_new_block(struct dnode_of_data *dn); 3452 int f2fs_get_block(struct dnode_of_data *dn, pgoff_t index); 3453 int f2fs_preallocate_blocks(struct kiocb *iocb, struct iov_iter *from); 3454 int f2fs_reserve_block(struct dnode_of_data *dn, pgoff_t index); 3455 int f2fs_mpage_readpages(struct address_space *mapping, 3456 struct list_head *pages, struct page *page, 3457 unsigned nr_pages, bool is_readahead); 3458 struct page *f2fs_get_read_data_page(struct inode *inode, pgoff_t index, 3459 int op_flags, bool for_write); 3460 struct page *f2fs_find_data_page(struct inode *inode, pgoff_t index); 3461 struct page *f2fs_get_lock_data_page(struct inode *inode, pgoff_t index, 3462 bool for_write); 3463 struct page *f2fs_get_new_data_page(struct inode *inode, 3464 struct page *ipage, pgoff_t index, bool new_i_size); 3465 int f2fs_do_write_data_page(struct f2fs_io_info *fio); 3466 void __do_map_lock(struct f2fs_sb_info *sbi, int flag, bool lock); 3467 int f2fs_map_blocks(struct inode *inode, struct f2fs_map_blocks *map, 3468 int create, int flag); 3469 int f2fs_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, 3470 u64 start, u64 len); 3471 int f2fs_encrypt_one_page(struct f2fs_io_info *fio); 3472 bool f2fs_should_update_inplace(struct inode *inode, struct f2fs_io_info *fio); 3473 bool f2fs_should_update_outplace(struct inode *inode, struct f2fs_io_info *fio); 3474 int f2fs_write_single_data_page(struct page *page, int *submitted, 3475 struct bio **bio, sector_t *last_block, 3476 struct writeback_control *wbc, 3477 enum iostat_type io_type, 3478 int compr_blocks); 3479 void f2fs_invalidate_page(struct page *page, unsigned int offset, 3480 unsigned int length); 3481 int f2fs_release_page(struct page *page, gfp_t wait); 3482 #ifdef CONFIG_MIGRATION 3483 int f2fs_migrate_page(struct address_space *mapping, struct page *newpage, 3484 struct page *page, enum migrate_mode mode); 3485 #endif 3486 bool f2fs_overwrite_io(struct inode *inode, loff_t pos, size_t len); 3487 void f2fs_clear_page_cache_dirty_tag(struct page *page); 3488 int f2fs_init_post_read_processing(void); 3489 void f2fs_destroy_post_read_processing(void); 3490 int f2fs_init_post_read_wq(struct f2fs_sb_info *sbi); 3491 void f2fs_destroy_post_read_wq(struct f2fs_sb_info *sbi); 3492 3493 /* 3494 * gc.c 3495 */ 3496 int f2fs_start_gc_thread(struct f2fs_sb_info *sbi); 3497 void f2fs_stop_gc_thread(struct f2fs_sb_info *sbi); 3498 block_t f2fs_start_bidx_of_node(unsigned int node_ofs, struct inode *inode); 3499 int f2fs_gc(struct f2fs_sb_info *sbi, bool sync, bool background, 3500 unsigned int segno); 3501 void f2fs_build_gc_manager(struct f2fs_sb_info *sbi); 3502 int f2fs_resize_fs(struct f2fs_sb_info *sbi, __u64 block_count); 3503 3504 /* 3505 * recovery.c 3506 */ 3507 int f2fs_recover_fsync_data(struct f2fs_sb_info *sbi, bool check_only); 3508 bool f2fs_space_for_roll_forward(struct f2fs_sb_info *sbi); 3509 3510 /* 3511 * debug.c 3512 */ 3513 #ifdef CONFIG_F2FS_STAT_FS 3514 struct f2fs_stat_info { 3515 struct list_head stat_list; 3516 struct f2fs_sb_info *sbi; 3517 int all_area_segs, sit_area_segs, nat_area_segs, ssa_area_segs; 3518 int main_area_segs, main_area_sections, main_area_zones; 3519 unsigned long long hit_largest, hit_cached, hit_rbtree; 3520 unsigned long long hit_total, total_ext; 3521 int ext_tree, zombie_tree, ext_node; 3522 int ndirty_node, ndirty_dent, ndirty_meta, ndirty_imeta; 3523 int ndirty_data, ndirty_qdata; 3524 int inmem_pages; 3525 unsigned int ndirty_dirs, ndirty_files, nquota_files, ndirty_all; 3526 int nats, dirty_nats, sits, dirty_sits; 3527 int free_nids, avail_nids, alloc_nids; 3528 int total_count, utilization; 3529 int bg_gc, nr_wb_cp_data, nr_wb_data; 3530 int nr_rd_data, nr_rd_node, nr_rd_meta; 3531 int nr_dio_read, nr_dio_write; 3532 unsigned int io_skip_bggc, other_skip_bggc; 3533 int nr_flushing, nr_flushed, flush_list_empty; 3534 int nr_discarding, nr_discarded; 3535 int nr_discard_cmd; 3536 unsigned int undiscard_blks; 3537 int inline_xattr, inline_inode, inline_dir, append, update, orphans; 3538 int compr_inode, compr_blocks; 3539 int aw_cnt, max_aw_cnt, vw_cnt, max_vw_cnt; 3540 unsigned int valid_count, valid_node_count, valid_inode_count, discard_blks; 3541 unsigned int bimodal, avg_vblocks; 3542 int util_free, util_valid, util_invalid; 3543 int rsvd_segs, overp_segs; 3544 int dirty_count, node_pages, meta_pages; 3545 int prefree_count, call_count, cp_count, bg_cp_count; 3546 int tot_segs, node_segs, data_segs, free_segs, free_secs; 3547 int bg_node_segs, bg_data_segs; 3548 int tot_blks, data_blks, node_blks; 3549 int bg_data_blks, bg_node_blks; 3550 unsigned long long skipped_atomic_files[2]; 3551 int curseg[NR_CURSEG_TYPE]; 3552 int cursec[NR_CURSEG_TYPE]; 3553 int curzone[NR_CURSEG_TYPE]; 3554 3555 unsigned int meta_count[META_MAX]; 3556 unsigned int segment_count[2]; 3557 unsigned int block_count[2]; 3558 unsigned int inplace_count; 3559 unsigned long long base_mem, cache_mem, page_mem; 3560 }; 3561 3562 static inline struct f2fs_stat_info *F2FS_STAT(struct f2fs_sb_info *sbi) 3563 { 3564 return (struct f2fs_stat_info *)sbi->stat_info; 3565 } 3566 3567 #define stat_inc_cp_count(si) ((si)->cp_count++) 3568 #define stat_inc_bg_cp_count(si) ((si)->bg_cp_count++) 3569 #define stat_inc_call_count(si) ((si)->call_count++) 3570 #define stat_inc_bggc_count(si) ((si)->bg_gc++) 3571 #define stat_io_skip_bggc_count(sbi) ((sbi)->io_skip_bggc++) 3572 #define stat_other_skip_bggc_count(sbi) ((sbi)->other_skip_bggc++) 3573 #define stat_inc_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]++) 3574 #define stat_dec_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]--) 3575 #define stat_inc_total_hit(sbi) (atomic64_inc(&(sbi)->total_hit_ext)) 3576 #define stat_inc_rbtree_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_rbtree)) 3577 #define stat_inc_largest_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_largest)) 3578 #define stat_inc_cached_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_cached)) 3579 #define stat_inc_inline_xattr(inode) \ 3580 do { \ 3581 if (f2fs_has_inline_xattr(inode)) \ 3582 (atomic_inc(&F2FS_I_SB(inode)->inline_xattr)); \ 3583 } while (0) 3584 #define stat_dec_inline_xattr(inode) \ 3585 do { \ 3586 if (f2fs_has_inline_xattr(inode)) \ 3587 (atomic_dec(&F2FS_I_SB(inode)->inline_xattr)); \ 3588 } while (0) 3589 #define stat_inc_inline_inode(inode) \ 3590 do { \ 3591 if (f2fs_has_inline_data(inode)) \ 3592 (atomic_inc(&F2FS_I_SB(inode)->inline_inode)); \ 3593 } while (0) 3594 #define stat_dec_inline_inode(inode) \ 3595 do { \ 3596 if (f2fs_has_inline_data(inode)) \ 3597 (atomic_dec(&F2FS_I_SB(inode)->inline_inode)); \ 3598 } while (0) 3599 #define stat_inc_inline_dir(inode) \ 3600 do { \ 3601 if (f2fs_has_inline_dentry(inode)) \ 3602 (atomic_inc(&F2FS_I_SB(inode)->inline_dir)); \ 3603 } while (0) 3604 #define stat_dec_inline_dir(inode) \ 3605 do { \ 3606 if (f2fs_has_inline_dentry(inode)) \ 3607 (atomic_dec(&F2FS_I_SB(inode)->inline_dir)); \ 3608 } while (0) 3609 #define stat_inc_compr_inode(inode) \ 3610 do { \ 3611 if (f2fs_compressed_file(inode)) \ 3612 (atomic_inc(&F2FS_I_SB(inode)->compr_inode)); \ 3613 } while (0) 3614 #define stat_dec_compr_inode(inode) \ 3615 do { \ 3616 if (f2fs_compressed_file(inode)) \ 3617 (atomic_dec(&F2FS_I_SB(inode)->compr_inode)); \ 3618 } while (0) 3619 #define stat_add_compr_blocks(inode, blocks) \ 3620 (atomic_add(blocks, &F2FS_I_SB(inode)->compr_blocks)) 3621 #define stat_sub_compr_blocks(inode, blocks) \ 3622 (atomic_sub(blocks, &F2FS_I_SB(inode)->compr_blocks)) 3623 #define stat_inc_meta_count(sbi, blkaddr) \ 3624 do { \ 3625 if (blkaddr < SIT_I(sbi)->sit_base_addr) \ 3626 atomic_inc(&(sbi)->meta_count[META_CP]); \ 3627 else if (blkaddr < NM_I(sbi)->nat_blkaddr) \ 3628 atomic_inc(&(sbi)->meta_count[META_SIT]); \ 3629 else if (blkaddr < SM_I(sbi)->ssa_blkaddr) \ 3630 atomic_inc(&(sbi)->meta_count[META_NAT]); \ 3631 else if (blkaddr < SM_I(sbi)->main_blkaddr) \ 3632 atomic_inc(&(sbi)->meta_count[META_SSA]); \ 3633 } while (0) 3634 #define stat_inc_seg_type(sbi, curseg) \ 3635 ((sbi)->segment_count[(curseg)->alloc_type]++) 3636 #define stat_inc_block_count(sbi, curseg) \ 3637 ((sbi)->block_count[(curseg)->alloc_type]++) 3638 #define stat_inc_inplace_blocks(sbi) \ 3639 (atomic_inc(&(sbi)->inplace_count)) 3640 #define stat_update_max_atomic_write(inode) \ 3641 do { \ 3642 int cur = F2FS_I_SB(inode)->atomic_files; \ 3643 int max = atomic_read(&F2FS_I_SB(inode)->max_aw_cnt); \ 3644 if (cur > max) \ 3645 atomic_set(&F2FS_I_SB(inode)->max_aw_cnt, cur); \ 3646 } while (0) 3647 #define stat_inc_volatile_write(inode) \ 3648 (atomic_inc(&F2FS_I_SB(inode)->vw_cnt)) 3649 #define stat_dec_volatile_write(inode) \ 3650 (atomic_dec(&F2FS_I_SB(inode)->vw_cnt)) 3651 #define stat_update_max_volatile_write(inode) \ 3652 do { \ 3653 int cur = atomic_read(&F2FS_I_SB(inode)->vw_cnt); \ 3654 int max = atomic_read(&F2FS_I_SB(inode)->max_vw_cnt); \ 3655 if (cur > max) \ 3656 atomic_set(&F2FS_I_SB(inode)->max_vw_cnt, cur); \ 3657 } while (0) 3658 #define stat_inc_seg_count(sbi, type, gc_type) \ 3659 do { \ 3660 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 3661 si->tot_segs++; \ 3662 if ((type) == SUM_TYPE_DATA) { \ 3663 si->data_segs++; \ 3664 si->bg_data_segs += (gc_type == BG_GC) ? 1 : 0; \ 3665 } else { \ 3666 si->node_segs++; \ 3667 si->bg_node_segs += (gc_type == BG_GC) ? 1 : 0; \ 3668 } \ 3669 } while (0) 3670 3671 #define stat_inc_tot_blk_count(si, blks) \ 3672 ((si)->tot_blks += (blks)) 3673 3674 #define stat_inc_data_blk_count(sbi, blks, gc_type) \ 3675 do { \ 3676 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 3677 stat_inc_tot_blk_count(si, blks); \ 3678 si->data_blks += (blks); \ 3679 si->bg_data_blks += ((gc_type) == BG_GC) ? (blks) : 0; \ 3680 } while (0) 3681 3682 #define stat_inc_node_blk_count(sbi, blks, gc_type) \ 3683 do { \ 3684 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 3685 stat_inc_tot_blk_count(si, blks); \ 3686 si->node_blks += (blks); \ 3687 si->bg_node_blks += ((gc_type) == BG_GC) ? (blks) : 0; \ 3688 } while (0) 3689 3690 int f2fs_build_stats(struct f2fs_sb_info *sbi); 3691 void f2fs_destroy_stats(struct f2fs_sb_info *sbi); 3692 void __init f2fs_create_root_stats(void); 3693 void f2fs_destroy_root_stats(void); 3694 void f2fs_update_sit_info(struct f2fs_sb_info *sbi); 3695 #else 3696 #define stat_inc_cp_count(si) do { } while (0) 3697 #define stat_inc_bg_cp_count(si) do { } while (0) 3698 #define stat_inc_call_count(si) do { } while (0) 3699 #define stat_inc_bggc_count(si) do { } while (0) 3700 #define stat_io_skip_bggc_count(sbi) do { } while (0) 3701 #define stat_other_skip_bggc_count(sbi) do { } while (0) 3702 #define stat_inc_dirty_inode(sbi, type) do { } while (0) 3703 #define stat_dec_dirty_inode(sbi, type) do { } while (0) 3704 #define stat_inc_total_hit(sbi) do { } while (0) 3705 #define stat_inc_rbtree_node_hit(sbi) do { } while (0) 3706 #define stat_inc_largest_node_hit(sbi) do { } while (0) 3707 #define stat_inc_cached_node_hit(sbi) do { } while (0) 3708 #define stat_inc_inline_xattr(inode) do { } while (0) 3709 #define stat_dec_inline_xattr(inode) do { } while (0) 3710 #define stat_inc_inline_inode(inode) do { } while (0) 3711 #define stat_dec_inline_inode(inode) do { } while (0) 3712 #define stat_inc_inline_dir(inode) do { } while (0) 3713 #define stat_dec_inline_dir(inode) do { } while (0) 3714 #define stat_inc_compr_inode(inode) do { } while (0) 3715 #define stat_dec_compr_inode(inode) do { } while (0) 3716 #define stat_add_compr_blocks(inode, blocks) do { } while (0) 3717 #define stat_sub_compr_blocks(inode, blocks) do { } while (0) 3718 #define stat_inc_atomic_write(inode) do { } while (0) 3719 #define stat_dec_atomic_write(inode) do { } while (0) 3720 #define stat_update_max_atomic_write(inode) do { } while (0) 3721 #define stat_inc_volatile_write(inode) do { } while (0) 3722 #define stat_dec_volatile_write(inode) do { } while (0) 3723 #define stat_update_max_volatile_write(inode) do { } while (0) 3724 #define stat_inc_meta_count(sbi, blkaddr) do { } while (0) 3725 #define stat_inc_seg_type(sbi, curseg) do { } while (0) 3726 #define stat_inc_block_count(sbi, curseg) do { } while (0) 3727 #define stat_inc_inplace_blocks(sbi) do { } while (0) 3728 #define stat_inc_seg_count(sbi, type, gc_type) do { } while (0) 3729 #define stat_inc_tot_blk_count(si, blks) do { } while (0) 3730 #define stat_inc_data_blk_count(sbi, blks, gc_type) do { } while (0) 3731 #define stat_inc_node_blk_count(sbi, blks, gc_type) do { } while (0) 3732 3733 static inline int f2fs_build_stats(struct f2fs_sb_info *sbi) { return 0; } 3734 static inline void f2fs_destroy_stats(struct f2fs_sb_info *sbi) { } 3735 static inline void __init f2fs_create_root_stats(void) { } 3736 static inline void f2fs_destroy_root_stats(void) { } 3737 static inline void f2fs_update_sit_info(struct f2fs_sb_info *sbi) {} 3738 #endif 3739 3740 extern const struct file_operations f2fs_dir_operations; 3741 #ifdef CONFIG_UNICODE 3742 extern const struct dentry_operations f2fs_dentry_ops; 3743 #endif 3744 extern const struct file_operations f2fs_file_operations; 3745 extern const struct inode_operations f2fs_file_inode_operations; 3746 extern const struct address_space_operations f2fs_dblock_aops; 3747 extern const struct address_space_operations f2fs_node_aops; 3748 extern const struct address_space_operations f2fs_meta_aops; 3749 extern const struct inode_operations f2fs_dir_inode_operations; 3750 extern const struct inode_operations f2fs_symlink_inode_operations; 3751 extern const struct inode_operations f2fs_encrypted_symlink_inode_operations; 3752 extern const struct inode_operations f2fs_special_inode_operations; 3753 extern struct kmem_cache *f2fs_inode_entry_slab; 3754 3755 /* 3756 * inline.c 3757 */ 3758 bool f2fs_may_inline_data(struct inode *inode); 3759 bool f2fs_may_inline_dentry(struct inode *inode); 3760 void f2fs_do_read_inline_data(struct page *page, struct page *ipage); 3761 void f2fs_truncate_inline_inode(struct inode *inode, 3762 struct page *ipage, u64 from); 3763 int f2fs_read_inline_data(struct inode *inode, struct page *page); 3764 int f2fs_convert_inline_page(struct dnode_of_data *dn, struct page *page); 3765 int f2fs_convert_inline_inode(struct inode *inode); 3766 int f2fs_try_convert_inline_dir(struct inode *dir, struct dentry *dentry); 3767 int f2fs_write_inline_data(struct inode *inode, struct page *page); 3768 bool f2fs_recover_inline_data(struct inode *inode, struct page *npage); 3769 struct f2fs_dir_entry *f2fs_find_in_inline_dir(struct inode *dir, 3770 const struct f2fs_filename *fname, 3771 struct page **res_page); 3772 int f2fs_make_empty_inline_dir(struct inode *inode, struct inode *parent, 3773 struct page *ipage); 3774 int f2fs_add_inline_entry(struct inode *dir, const struct f2fs_filename *fname, 3775 struct inode *inode, nid_t ino, umode_t mode); 3776 void f2fs_delete_inline_entry(struct f2fs_dir_entry *dentry, 3777 struct page *page, struct inode *dir, 3778 struct inode *inode); 3779 bool f2fs_empty_inline_dir(struct inode *dir); 3780 int f2fs_read_inline_dir(struct file *file, struct dir_context *ctx, 3781 struct fscrypt_str *fstr); 3782 int f2fs_inline_data_fiemap(struct inode *inode, 3783 struct fiemap_extent_info *fieinfo, 3784 __u64 start, __u64 len); 3785 3786 /* 3787 * shrinker.c 3788 */ 3789 unsigned long f2fs_shrink_count(struct shrinker *shrink, 3790 struct shrink_control *sc); 3791 unsigned long f2fs_shrink_scan(struct shrinker *shrink, 3792 struct shrink_control *sc); 3793 void f2fs_join_shrinker(struct f2fs_sb_info *sbi); 3794 void f2fs_leave_shrinker(struct f2fs_sb_info *sbi); 3795 3796 /* 3797 * extent_cache.c 3798 */ 3799 struct rb_entry *f2fs_lookup_rb_tree(struct rb_root_cached *root, 3800 struct rb_entry *cached_re, unsigned int ofs); 3801 struct rb_node **f2fs_lookup_rb_tree_for_insert(struct f2fs_sb_info *sbi, 3802 struct rb_root_cached *root, 3803 struct rb_node **parent, 3804 unsigned int ofs, bool *leftmost); 3805 struct rb_entry *f2fs_lookup_rb_tree_ret(struct rb_root_cached *root, 3806 struct rb_entry *cached_re, unsigned int ofs, 3807 struct rb_entry **prev_entry, struct rb_entry **next_entry, 3808 struct rb_node ***insert_p, struct rb_node **insert_parent, 3809 bool force, bool *leftmost); 3810 bool f2fs_check_rb_tree_consistence(struct f2fs_sb_info *sbi, 3811 struct rb_root_cached *root); 3812 unsigned int f2fs_shrink_extent_tree(struct f2fs_sb_info *sbi, int nr_shrink); 3813 bool f2fs_init_extent_tree(struct inode *inode, struct f2fs_extent *i_ext); 3814 void f2fs_drop_extent_tree(struct inode *inode); 3815 unsigned int f2fs_destroy_extent_node(struct inode *inode); 3816 void f2fs_destroy_extent_tree(struct inode *inode); 3817 bool f2fs_lookup_extent_cache(struct inode *inode, pgoff_t pgofs, 3818 struct extent_info *ei); 3819 void f2fs_update_extent_cache(struct dnode_of_data *dn); 3820 void f2fs_update_extent_cache_range(struct dnode_of_data *dn, 3821 pgoff_t fofs, block_t blkaddr, unsigned int len); 3822 void f2fs_init_extent_cache_info(struct f2fs_sb_info *sbi); 3823 int __init f2fs_create_extent_cache(void); 3824 void f2fs_destroy_extent_cache(void); 3825 3826 /* 3827 * sysfs.c 3828 */ 3829 int __init f2fs_init_sysfs(void); 3830 void f2fs_exit_sysfs(void); 3831 int f2fs_register_sysfs(struct f2fs_sb_info *sbi); 3832 void f2fs_unregister_sysfs(struct f2fs_sb_info *sbi); 3833 3834 /* verity.c */ 3835 extern const struct fsverity_operations f2fs_verityops; 3836 3837 /* 3838 * crypto support 3839 */ 3840 static inline bool f2fs_encrypted_file(struct inode *inode) 3841 { 3842 return IS_ENCRYPTED(inode) && S_ISREG(inode->i_mode); 3843 } 3844 3845 static inline void f2fs_set_encrypted_inode(struct inode *inode) 3846 { 3847 #ifdef CONFIG_FS_ENCRYPTION 3848 file_set_encrypt(inode); 3849 f2fs_set_inode_flags(inode); 3850 #endif 3851 } 3852 3853 /* 3854 * Returns true if the reads of the inode's data need to undergo some 3855 * postprocessing step, like decryption or authenticity verification. 3856 */ 3857 static inline bool f2fs_post_read_required(struct inode *inode) 3858 { 3859 return f2fs_encrypted_file(inode) || fsverity_active(inode) || 3860 f2fs_compressed_file(inode); 3861 } 3862 3863 /* 3864 * compress.c 3865 */ 3866 #ifdef CONFIG_F2FS_FS_COMPRESSION 3867 bool f2fs_is_compressed_page(struct page *page); 3868 struct page *f2fs_compress_control_page(struct page *page); 3869 int f2fs_prepare_compress_overwrite(struct inode *inode, 3870 struct page **pagep, pgoff_t index, void **fsdata); 3871 bool f2fs_compress_write_end(struct inode *inode, void *fsdata, 3872 pgoff_t index, unsigned copied); 3873 int f2fs_truncate_partial_cluster(struct inode *inode, u64 from, bool lock); 3874 void f2fs_compress_write_end_io(struct bio *bio, struct page *page); 3875 bool f2fs_is_compress_backend_ready(struct inode *inode); 3876 int f2fs_init_compress_mempool(void); 3877 void f2fs_destroy_compress_mempool(void); 3878 void f2fs_decompress_pages(struct bio *bio, struct page *page, bool verity); 3879 bool f2fs_cluster_is_empty(struct compress_ctx *cc); 3880 bool f2fs_cluster_can_merge_page(struct compress_ctx *cc, pgoff_t index); 3881 void f2fs_compress_ctx_add_page(struct compress_ctx *cc, struct page *page); 3882 int f2fs_write_multi_pages(struct compress_ctx *cc, 3883 int *submitted, 3884 struct writeback_control *wbc, 3885 enum iostat_type io_type); 3886 int f2fs_is_compressed_cluster(struct inode *inode, pgoff_t index); 3887 int f2fs_read_multi_pages(struct compress_ctx *cc, struct bio **bio_ret, 3888 unsigned nr_pages, sector_t *last_block_in_bio, 3889 bool is_readahead, bool for_write); 3890 struct decompress_io_ctx *f2fs_alloc_dic(struct compress_ctx *cc); 3891 void f2fs_free_dic(struct decompress_io_ctx *dic); 3892 void f2fs_decompress_end_io(struct page **rpages, 3893 unsigned int cluster_size, bool err, bool verity); 3894 int f2fs_init_compress_ctx(struct compress_ctx *cc); 3895 void f2fs_destroy_compress_ctx(struct compress_ctx *cc); 3896 void f2fs_init_compress_info(struct f2fs_sb_info *sbi); 3897 #else 3898 static inline bool f2fs_is_compressed_page(struct page *page) { return false; } 3899 static inline bool f2fs_is_compress_backend_ready(struct inode *inode) 3900 { 3901 if (!f2fs_compressed_file(inode)) 3902 return true; 3903 /* not support compression */ 3904 return false; 3905 } 3906 static inline struct page *f2fs_compress_control_page(struct page *page) 3907 { 3908 WARN_ON_ONCE(1); 3909 return ERR_PTR(-EINVAL); 3910 } 3911 static inline int f2fs_init_compress_mempool(void) { return 0; } 3912 static inline void f2fs_destroy_compress_mempool(void) { } 3913 #endif 3914 3915 static inline void set_compress_context(struct inode *inode) 3916 { 3917 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 3918 3919 F2FS_I(inode)->i_compress_algorithm = 3920 F2FS_OPTION(sbi).compress_algorithm; 3921 F2FS_I(inode)->i_log_cluster_size = 3922 F2FS_OPTION(sbi).compress_log_size; 3923 F2FS_I(inode)->i_cluster_size = 3924 1 << F2FS_I(inode)->i_log_cluster_size; 3925 F2FS_I(inode)->i_flags |= F2FS_COMPR_FL; 3926 set_inode_flag(inode, FI_COMPRESSED_FILE); 3927 stat_inc_compr_inode(inode); 3928 f2fs_mark_inode_dirty_sync(inode, true); 3929 } 3930 3931 static inline u64 f2fs_disable_compressed_file(struct inode *inode) 3932 { 3933 struct f2fs_inode_info *fi = F2FS_I(inode); 3934 3935 if (!f2fs_compressed_file(inode)) 3936 return 0; 3937 if (S_ISREG(inode->i_mode)) { 3938 if (get_dirty_pages(inode)) 3939 return 1; 3940 if (fi->i_compr_blocks) 3941 return fi->i_compr_blocks; 3942 } 3943 3944 fi->i_flags &= ~F2FS_COMPR_FL; 3945 stat_dec_compr_inode(inode); 3946 clear_inode_flag(inode, FI_COMPRESSED_FILE); 3947 f2fs_mark_inode_dirty_sync(inode, true); 3948 return 0; 3949 } 3950 3951 #define F2FS_FEATURE_FUNCS(name, flagname) \ 3952 static inline int f2fs_sb_has_##name(struct f2fs_sb_info *sbi) \ 3953 { \ 3954 return F2FS_HAS_FEATURE(sbi, F2FS_FEATURE_##flagname); \ 3955 } 3956 3957 F2FS_FEATURE_FUNCS(encrypt, ENCRYPT); 3958 F2FS_FEATURE_FUNCS(blkzoned, BLKZONED); 3959 F2FS_FEATURE_FUNCS(extra_attr, EXTRA_ATTR); 3960 F2FS_FEATURE_FUNCS(project_quota, PRJQUOTA); 3961 F2FS_FEATURE_FUNCS(inode_chksum, INODE_CHKSUM); 3962 F2FS_FEATURE_FUNCS(flexible_inline_xattr, FLEXIBLE_INLINE_XATTR); 3963 F2FS_FEATURE_FUNCS(quota_ino, QUOTA_INO); 3964 F2FS_FEATURE_FUNCS(inode_crtime, INODE_CRTIME); 3965 F2FS_FEATURE_FUNCS(lost_found, LOST_FOUND); 3966 F2FS_FEATURE_FUNCS(verity, VERITY); 3967 F2FS_FEATURE_FUNCS(sb_chksum, SB_CHKSUM); 3968 F2FS_FEATURE_FUNCS(casefold, CASEFOLD); 3969 F2FS_FEATURE_FUNCS(compression, COMPRESSION); 3970 3971 #ifdef CONFIG_BLK_DEV_ZONED 3972 static inline bool f2fs_blkz_is_seq(struct f2fs_sb_info *sbi, int devi, 3973 block_t blkaddr) 3974 { 3975 unsigned int zno = blkaddr >> sbi->log_blocks_per_blkz; 3976 3977 return test_bit(zno, FDEV(devi).blkz_seq); 3978 } 3979 #endif 3980 3981 static inline bool f2fs_hw_should_discard(struct f2fs_sb_info *sbi) 3982 { 3983 return f2fs_sb_has_blkzoned(sbi); 3984 } 3985 3986 static inline bool f2fs_bdev_support_discard(struct block_device *bdev) 3987 { 3988 return blk_queue_discard(bdev_get_queue(bdev)) || 3989 bdev_is_zoned(bdev); 3990 } 3991 3992 static inline bool f2fs_hw_support_discard(struct f2fs_sb_info *sbi) 3993 { 3994 int i; 3995 3996 if (!f2fs_is_multi_device(sbi)) 3997 return f2fs_bdev_support_discard(sbi->sb->s_bdev); 3998 3999 for (i = 0; i < sbi->s_ndevs; i++) 4000 if (f2fs_bdev_support_discard(FDEV(i).bdev)) 4001 return true; 4002 return false; 4003 } 4004 4005 static inline bool f2fs_realtime_discard_enable(struct f2fs_sb_info *sbi) 4006 { 4007 return (test_opt(sbi, DISCARD) && f2fs_hw_support_discard(sbi)) || 4008 f2fs_hw_should_discard(sbi); 4009 } 4010 4011 static inline bool f2fs_hw_is_readonly(struct f2fs_sb_info *sbi) 4012 { 4013 int i; 4014 4015 if (!f2fs_is_multi_device(sbi)) 4016 return bdev_read_only(sbi->sb->s_bdev); 4017 4018 for (i = 0; i < sbi->s_ndevs; i++) 4019 if (bdev_read_only(FDEV(i).bdev)) 4020 return true; 4021 return false; 4022 } 4023 4024 static inline bool f2fs_lfs_mode(struct f2fs_sb_info *sbi) 4025 { 4026 return F2FS_OPTION(sbi).fs_mode == FS_MODE_LFS; 4027 } 4028 4029 static inline bool f2fs_may_encrypt(struct inode *dir, struct inode *inode) 4030 { 4031 #ifdef CONFIG_FS_ENCRYPTION 4032 struct f2fs_sb_info *sbi = F2FS_I_SB(dir); 4033 umode_t mode = inode->i_mode; 4034 4035 /* 4036 * If the directory encrypted or dummy encryption enabled, 4037 * then we should encrypt the inode. 4038 */ 4039 if (IS_ENCRYPTED(dir) || DUMMY_ENCRYPTION_ENABLED(sbi)) 4040 return (S_ISREG(mode) || S_ISDIR(mode) || S_ISLNK(mode)); 4041 #endif 4042 return false; 4043 } 4044 4045 static inline bool f2fs_may_compress(struct inode *inode) 4046 { 4047 if (IS_SWAPFILE(inode) || f2fs_is_pinned_file(inode) || 4048 f2fs_is_atomic_file(inode) || 4049 f2fs_is_volatile_file(inode)) 4050 return false; 4051 return S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode); 4052 } 4053 4054 static inline void f2fs_i_compr_blocks_update(struct inode *inode, 4055 u64 blocks, bool add) 4056 { 4057 int diff = F2FS_I(inode)->i_cluster_size - blocks; 4058 4059 /* don't update i_compr_blocks if saved blocks were released */ 4060 if (!add && !F2FS_I(inode)->i_compr_blocks) 4061 return; 4062 4063 if (add) { 4064 F2FS_I(inode)->i_compr_blocks += diff; 4065 stat_add_compr_blocks(inode, diff); 4066 } else { 4067 F2FS_I(inode)->i_compr_blocks -= diff; 4068 stat_sub_compr_blocks(inode, diff); 4069 } 4070 f2fs_mark_inode_dirty_sync(inode, true); 4071 } 4072 4073 static inline int block_unaligned_IO(struct inode *inode, 4074 struct kiocb *iocb, struct iov_iter *iter) 4075 { 4076 unsigned int i_blkbits = READ_ONCE(inode->i_blkbits); 4077 unsigned int blocksize_mask = (1 << i_blkbits) - 1; 4078 loff_t offset = iocb->ki_pos; 4079 unsigned long align = offset | iov_iter_alignment(iter); 4080 4081 return align & blocksize_mask; 4082 } 4083 4084 static inline int allow_outplace_dio(struct inode *inode, 4085 struct kiocb *iocb, struct iov_iter *iter) 4086 { 4087 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 4088 int rw = iov_iter_rw(iter); 4089 4090 return (f2fs_lfs_mode(sbi) && (rw == WRITE) && 4091 !block_unaligned_IO(inode, iocb, iter)); 4092 } 4093 4094 static inline bool f2fs_force_buffered_io(struct inode *inode, 4095 struct kiocb *iocb, struct iov_iter *iter) 4096 { 4097 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 4098 int rw = iov_iter_rw(iter); 4099 4100 if (f2fs_post_read_required(inode)) 4101 return true; 4102 if (f2fs_is_multi_device(sbi)) 4103 return true; 4104 /* 4105 * for blkzoned device, fallback direct IO to buffered IO, so 4106 * all IOs can be serialized by log-structured write. 4107 */ 4108 if (f2fs_sb_has_blkzoned(sbi)) 4109 return true; 4110 if (f2fs_lfs_mode(sbi) && (rw == WRITE)) { 4111 if (block_unaligned_IO(inode, iocb, iter)) 4112 return true; 4113 if (F2FS_IO_ALIGNED(sbi)) 4114 return true; 4115 } 4116 if (is_sbi_flag_set(F2FS_I_SB(inode), SBI_CP_DISABLED) && 4117 !IS_SWAPFILE(inode)) 4118 return true; 4119 4120 return false; 4121 } 4122 4123 #ifdef CONFIG_F2FS_FAULT_INJECTION 4124 extern void f2fs_build_fault_attr(struct f2fs_sb_info *sbi, unsigned int rate, 4125 unsigned int type); 4126 #else 4127 #define f2fs_build_fault_attr(sbi, rate, type) do { } while (0) 4128 #endif 4129 4130 static inline bool is_journalled_quota(struct f2fs_sb_info *sbi) 4131 { 4132 #ifdef CONFIG_QUOTA 4133 if (f2fs_sb_has_quota_ino(sbi)) 4134 return true; 4135 if (F2FS_OPTION(sbi).s_qf_names[USRQUOTA] || 4136 F2FS_OPTION(sbi).s_qf_names[GRPQUOTA] || 4137 F2FS_OPTION(sbi).s_qf_names[PRJQUOTA]) 4138 return true; 4139 #endif 4140 return false; 4141 } 4142 4143 #define EFSBADCRC EBADMSG /* Bad CRC detected */ 4144 #define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */ 4145 4146 #endif /* _LINUX_F2FS_H */ 4147