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 <crypto/hash.h> 26 27 #include <linux/fscrypt.h> 28 #include <linux/fsverity.h> 29 30 #ifdef CONFIG_F2FS_CHECK_FS 31 #define f2fs_bug_on(sbi, condition) BUG_ON(condition) 32 #else 33 #define f2fs_bug_on(sbi, condition) \ 34 do { \ 35 if (unlikely(condition)) { \ 36 WARN_ON(1); \ 37 set_sbi_flag(sbi, SBI_NEED_FSCK); \ 38 } \ 39 } while (0) 40 #endif 41 42 enum { 43 FAULT_KMALLOC, 44 FAULT_KVMALLOC, 45 FAULT_PAGE_ALLOC, 46 FAULT_PAGE_GET, 47 FAULT_ALLOC_BIO, 48 FAULT_ALLOC_NID, 49 FAULT_ORPHAN, 50 FAULT_BLOCK, 51 FAULT_DIR_DEPTH, 52 FAULT_EVICT_INODE, 53 FAULT_TRUNCATE, 54 FAULT_READ_IO, 55 FAULT_CHECKPOINT, 56 FAULT_DISCARD, 57 FAULT_WRITE_IO, 58 FAULT_MAX, 59 }; 60 61 #ifdef CONFIG_F2FS_FAULT_INJECTION 62 #define F2FS_ALL_FAULT_TYPE ((1 << FAULT_MAX) - 1) 63 64 struct f2fs_fault_info { 65 atomic_t inject_ops; 66 unsigned int inject_rate; 67 unsigned int inject_type; 68 }; 69 70 extern const char *f2fs_fault_name[FAULT_MAX]; 71 #define IS_FAULT_SET(fi, type) ((fi)->inject_type & (1 << (type))) 72 #endif 73 74 /* 75 * For mount options 76 */ 77 #define F2FS_MOUNT_BG_GC 0x00000001 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_FORCE_FG_GC 0x00004000 92 #define F2FS_MOUNT_DATA_FLUSH 0x00008000 93 #define F2FS_MOUNT_FAULT_INJECTION 0x00010000 94 #define F2FS_MOUNT_ADAPTIVE 0x00020000 95 #define F2FS_MOUNT_LFS 0x00040000 96 #define F2FS_MOUNT_USRQUOTA 0x00080000 97 #define F2FS_MOUNT_GRPQUOTA 0x00100000 98 #define F2FS_MOUNT_PRJQUOTA 0x00200000 99 #define F2FS_MOUNT_QUOTA 0x00400000 100 #define F2FS_MOUNT_INLINE_XATTR_SIZE 0x00800000 101 #define F2FS_MOUNT_RESERVE_ROOT 0x01000000 102 #define F2FS_MOUNT_DISABLE_CHECKPOINT 0x02000000 103 104 #define F2FS_OPTION(sbi) ((sbi)->mount_opt) 105 #define clear_opt(sbi, option) (F2FS_OPTION(sbi).opt &= ~F2FS_MOUNT_##option) 106 #define set_opt(sbi, option) (F2FS_OPTION(sbi).opt |= F2FS_MOUNT_##option) 107 #define test_opt(sbi, option) (F2FS_OPTION(sbi).opt & F2FS_MOUNT_##option) 108 109 #define ver_after(a, b) (typecheck(unsigned long long, a) && \ 110 typecheck(unsigned long long, b) && \ 111 ((long long)((a) - (b)) > 0)) 112 113 typedef u32 block_t; /* 114 * should not change u32, since it is the on-disk block 115 * address format, __le32. 116 */ 117 typedef u32 nid_t; 118 119 #define COMPRESS_EXT_NUM 16 120 121 struct f2fs_mount_info { 122 unsigned int opt; 123 int write_io_size_bits; /* Write IO size bits */ 124 block_t root_reserved_blocks; /* root reserved blocks */ 125 kuid_t s_resuid; /* reserved blocks for uid */ 126 kgid_t s_resgid; /* reserved blocks for gid */ 127 int active_logs; /* # of active logs */ 128 int inline_xattr_size; /* inline xattr size */ 129 #ifdef CONFIG_F2FS_FAULT_INJECTION 130 struct f2fs_fault_info fault_info; /* For fault injection */ 131 #endif 132 #ifdef CONFIG_QUOTA 133 /* Names of quota files with journalled quota */ 134 char *s_qf_names[MAXQUOTAS]; 135 int s_jquota_fmt; /* Format of quota to use */ 136 #endif 137 /* For which write hints are passed down to block layer */ 138 int whint_mode; 139 int alloc_mode; /* segment allocation policy */ 140 int fsync_mode; /* fsync policy */ 141 bool test_dummy_encryption; /* test dummy encryption */ 142 block_t unusable_cap; /* Amount of space allowed to be 143 * unusable when disabling checkpoint 144 */ 145 146 /* For compression */ 147 unsigned char compress_algorithm; /* algorithm type */ 148 unsigned compress_log_size; /* cluster log size */ 149 unsigned char compress_ext_cnt; /* extension count */ 150 unsigned char extensions[COMPRESS_EXT_NUM][F2FS_EXTENSION_LEN]; /* extensions */ 151 }; 152 153 #define F2FS_FEATURE_ENCRYPT 0x0001 154 #define F2FS_FEATURE_BLKZONED 0x0002 155 #define F2FS_FEATURE_ATOMIC_WRITE 0x0004 156 #define F2FS_FEATURE_EXTRA_ATTR 0x0008 157 #define F2FS_FEATURE_PRJQUOTA 0x0010 158 #define F2FS_FEATURE_INODE_CHKSUM 0x0020 159 #define F2FS_FEATURE_FLEXIBLE_INLINE_XATTR 0x0040 160 #define F2FS_FEATURE_QUOTA_INO 0x0080 161 #define F2FS_FEATURE_INODE_CRTIME 0x0100 162 #define F2FS_FEATURE_LOST_FOUND 0x0200 163 #define F2FS_FEATURE_VERITY 0x0400 164 #define F2FS_FEATURE_SB_CHKSUM 0x0800 165 #define F2FS_FEATURE_CASEFOLD 0x1000 166 #define F2FS_FEATURE_COMPRESSION 0x2000 167 168 #define __F2FS_HAS_FEATURE(raw_super, mask) \ 169 ((raw_super->feature & cpu_to_le32(mask)) != 0) 170 #define F2FS_HAS_FEATURE(sbi, mask) __F2FS_HAS_FEATURE(sbi->raw_super, mask) 171 #define F2FS_SET_FEATURE(sbi, mask) \ 172 (sbi->raw_super->feature |= cpu_to_le32(mask)) 173 #define F2FS_CLEAR_FEATURE(sbi, mask) \ 174 (sbi->raw_super->feature &= ~cpu_to_le32(mask)) 175 176 /* 177 * Default values for user and/or group using reserved blocks 178 */ 179 #define F2FS_DEF_RESUID 0 180 #define F2FS_DEF_RESGID 0 181 182 /* 183 * For checkpoint manager 184 */ 185 enum { 186 NAT_BITMAP, 187 SIT_BITMAP 188 }; 189 190 #define CP_UMOUNT 0x00000001 191 #define CP_FASTBOOT 0x00000002 192 #define CP_SYNC 0x00000004 193 #define CP_RECOVERY 0x00000008 194 #define CP_DISCARD 0x00000010 195 #define CP_TRIMMED 0x00000020 196 #define CP_PAUSE 0x00000040 197 198 #define MAX_DISCARD_BLOCKS(sbi) BLKS_PER_SEC(sbi) 199 #define DEF_MAX_DISCARD_REQUEST 8 /* issue 8 discards per round */ 200 #define DEF_MIN_DISCARD_ISSUE_TIME 50 /* 50 ms, if exists */ 201 #define DEF_MID_DISCARD_ISSUE_TIME 500 /* 500 ms, if device busy */ 202 #define DEF_MAX_DISCARD_ISSUE_TIME 60000 /* 60 s, if no candidates */ 203 #define DEF_DISCARD_URGENT_UTIL 80 /* do more discard over 80% */ 204 #define DEF_CP_INTERVAL 60 /* 60 secs */ 205 #define DEF_IDLE_INTERVAL 5 /* 5 secs */ 206 #define DEF_DISABLE_INTERVAL 5 /* 5 secs */ 207 #define DEF_DISABLE_QUICK_INTERVAL 1 /* 1 secs */ 208 #define DEF_UMOUNT_DISCARD_TIMEOUT 5 /* 5 secs */ 209 210 struct cp_control { 211 int reason; 212 __u64 trim_start; 213 __u64 trim_end; 214 __u64 trim_minlen; 215 }; 216 217 /* 218 * indicate meta/data type 219 */ 220 enum { 221 META_CP, 222 META_NAT, 223 META_SIT, 224 META_SSA, 225 META_MAX, 226 META_POR, 227 DATA_GENERIC, /* check range only */ 228 DATA_GENERIC_ENHANCE, /* strong check on range and segment bitmap */ 229 DATA_GENERIC_ENHANCE_READ, /* 230 * strong check on range and segment 231 * bitmap but no warning due to race 232 * condition of read on truncated area 233 * by extent_cache 234 */ 235 META_GENERIC, 236 }; 237 238 /* for the list of ino */ 239 enum { 240 ORPHAN_INO, /* for orphan ino list */ 241 APPEND_INO, /* for append ino list */ 242 UPDATE_INO, /* for update ino list */ 243 TRANS_DIR_INO, /* for trasactions dir ino list */ 244 FLUSH_INO, /* for multiple device flushing */ 245 MAX_INO_ENTRY, /* max. list */ 246 }; 247 248 struct ino_entry { 249 struct list_head list; /* list head */ 250 nid_t ino; /* inode number */ 251 unsigned int dirty_device; /* dirty device bitmap */ 252 }; 253 254 /* for the list of inodes to be GCed */ 255 struct inode_entry { 256 struct list_head list; /* list head */ 257 struct inode *inode; /* vfs inode pointer */ 258 }; 259 260 struct fsync_node_entry { 261 struct list_head list; /* list head */ 262 struct page *page; /* warm node page pointer */ 263 unsigned int seq_id; /* sequence id */ 264 }; 265 266 /* for the bitmap indicate blocks to be discarded */ 267 struct discard_entry { 268 struct list_head list; /* list head */ 269 block_t start_blkaddr; /* start blockaddr of current segment */ 270 unsigned char discard_map[SIT_VBLOCK_MAP_SIZE]; /* segment discard bitmap */ 271 }; 272 273 /* default discard granularity of inner discard thread, unit: block count */ 274 #define DEFAULT_DISCARD_GRANULARITY 16 275 276 /* max discard pend list number */ 277 #define MAX_PLIST_NUM 512 278 #define plist_idx(blk_num) ((blk_num) >= MAX_PLIST_NUM ? \ 279 (MAX_PLIST_NUM - 1) : ((blk_num) - 1)) 280 281 enum { 282 D_PREP, /* initial */ 283 D_PARTIAL, /* partially submitted */ 284 D_SUBMIT, /* all submitted */ 285 D_DONE, /* finished */ 286 }; 287 288 struct discard_info { 289 block_t lstart; /* logical start address */ 290 block_t len; /* length */ 291 block_t start; /* actual start address in dev */ 292 }; 293 294 struct discard_cmd { 295 struct rb_node rb_node; /* rb node located in rb-tree */ 296 union { 297 struct { 298 block_t lstart; /* logical start address */ 299 block_t len; /* length */ 300 block_t start; /* actual start address in dev */ 301 }; 302 struct discard_info di; /* discard info */ 303 304 }; 305 struct list_head list; /* command list */ 306 struct completion wait; /* compleation */ 307 struct block_device *bdev; /* bdev */ 308 unsigned short ref; /* reference count */ 309 unsigned char state; /* state */ 310 unsigned char queued; /* queued discard */ 311 int error; /* bio error */ 312 spinlock_t lock; /* for state/bio_ref updating */ 313 unsigned short bio_ref; /* bio reference count */ 314 }; 315 316 enum { 317 DPOLICY_BG, 318 DPOLICY_FORCE, 319 DPOLICY_FSTRIM, 320 DPOLICY_UMOUNT, 321 MAX_DPOLICY, 322 }; 323 324 struct discard_policy { 325 int type; /* type of discard */ 326 unsigned int min_interval; /* used for candidates exist */ 327 unsigned int mid_interval; /* used for device busy */ 328 unsigned int max_interval; /* used for candidates not exist */ 329 unsigned int max_requests; /* # of discards issued per round */ 330 unsigned int io_aware_gran; /* minimum granularity discard not be aware of I/O */ 331 bool io_aware; /* issue discard in idle time */ 332 bool sync; /* submit discard with REQ_SYNC flag */ 333 bool ordered; /* issue discard by lba order */ 334 unsigned int granularity; /* discard granularity */ 335 int timeout; /* discard timeout for put_super */ 336 }; 337 338 struct discard_cmd_control { 339 struct task_struct *f2fs_issue_discard; /* discard thread */ 340 struct list_head entry_list; /* 4KB discard entry list */ 341 struct list_head pend_list[MAX_PLIST_NUM];/* store pending entries */ 342 struct list_head wait_list; /* store on-flushing entries */ 343 struct list_head fstrim_list; /* in-flight discard from fstrim */ 344 wait_queue_head_t discard_wait_queue; /* waiting queue for wake-up */ 345 unsigned int discard_wake; /* to wake up discard thread */ 346 struct mutex cmd_lock; 347 unsigned int nr_discards; /* # of discards in the list */ 348 unsigned int max_discards; /* max. discards to be issued */ 349 unsigned int discard_granularity; /* discard granularity */ 350 unsigned int undiscard_blks; /* # of undiscard blocks */ 351 unsigned int next_pos; /* next discard position */ 352 atomic_t issued_discard; /* # of issued discard */ 353 atomic_t queued_discard; /* # of queued discard */ 354 atomic_t discard_cmd_cnt; /* # of cached cmd count */ 355 struct rb_root_cached root; /* root of discard rb-tree */ 356 bool rbtree_check; /* config for consistence check */ 357 }; 358 359 /* for the list of fsync inodes, used only during recovery */ 360 struct fsync_inode_entry { 361 struct list_head list; /* list head */ 362 struct inode *inode; /* vfs inode pointer */ 363 block_t blkaddr; /* block address locating the last fsync */ 364 block_t last_dentry; /* block address locating the last dentry */ 365 }; 366 367 #define nats_in_cursum(jnl) (le16_to_cpu((jnl)->n_nats)) 368 #define sits_in_cursum(jnl) (le16_to_cpu((jnl)->n_sits)) 369 370 #define nat_in_journal(jnl, i) ((jnl)->nat_j.entries[i].ne) 371 #define nid_in_journal(jnl, i) ((jnl)->nat_j.entries[i].nid) 372 #define sit_in_journal(jnl, i) ((jnl)->sit_j.entries[i].se) 373 #define segno_in_journal(jnl, i) ((jnl)->sit_j.entries[i].segno) 374 375 #define MAX_NAT_JENTRIES(jnl) (NAT_JOURNAL_ENTRIES - nats_in_cursum(jnl)) 376 #define MAX_SIT_JENTRIES(jnl) (SIT_JOURNAL_ENTRIES - sits_in_cursum(jnl)) 377 378 static inline int update_nats_in_cursum(struct f2fs_journal *journal, int i) 379 { 380 int before = nats_in_cursum(journal); 381 382 journal->n_nats = cpu_to_le16(before + i); 383 return before; 384 } 385 386 static inline int update_sits_in_cursum(struct f2fs_journal *journal, int i) 387 { 388 int before = sits_in_cursum(journal); 389 390 journal->n_sits = cpu_to_le16(before + i); 391 return before; 392 } 393 394 static inline bool __has_cursum_space(struct f2fs_journal *journal, 395 int size, int type) 396 { 397 if (type == NAT_JOURNAL) 398 return size <= MAX_NAT_JENTRIES(journal); 399 return size <= MAX_SIT_JENTRIES(journal); 400 } 401 402 /* 403 * ioctl commands 404 */ 405 #define F2FS_IOC_GETFLAGS FS_IOC_GETFLAGS 406 #define F2FS_IOC_SETFLAGS FS_IOC_SETFLAGS 407 #define F2FS_IOC_GETVERSION FS_IOC_GETVERSION 408 409 #define F2FS_IOCTL_MAGIC 0xf5 410 #define F2FS_IOC_START_ATOMIC_WRITE _IO(F2FS_IOCTL_MAGIC, 1) 411 #define F2FS_IOC_COMMIT_ATOMIC_WRITE _IO(F2FS_IOCTL_MAGIC, 2) 412 #define F2FS_IOC_START_VOLATILE_WRITE _IO(F2FS_IOCTL_MAGIC, 3) 413 #define F2FS_IOC_RELEASE_VOLATILE_WRITE _IO(F2FS_IOCTL_MAGIC, 4) 414 #define F2FS_IOC_ABORT_VOLATILE_WRITE _IO(F2FS_IOCTL_MAGIC, 5) 415 #define F2FS_IOC_GARBAGE_COLLECT _IOW(F2FS_IOCTL_MAGIC, 6, __u32) 416 #define F2FS_IOC_WRITE_CHECKPOINT _IO(F2FS_IOCTL_MAGIC, 7) 417 #define F2FS_IOC_DEFRAGMENT _IOWR(F2FS_IOCTL_MAGIC, 8, \ 418 struct f2fs_defragment) 419 #define F2FS_IOC_MOVE_RANGE _IOWR(F2FS_IOCTL_MAGIC, 9, \ 420 struct f2fs_move_range) 421 #define F2FS_IOC_FLUSH_DEVICE _IOW(F2FS_IOCTL_MAGIC, 10, \ 422 struct f2fs_flush_device) 423 #define F2FS_IOC_GARBAGE_COLLECT_RANGE _IOW(F2FS_IOCTL_MAGIC, 11, \ 424 struct f2fs_gc_range) 425 #define F2FS_IOC_GET_FEATURES _IOR(F2FS_IOCTL_MAGIC, 12, __u32) 426 #define F2FS_IOC_SET_PIN_FILE _IOW(F2FS_IOCTL_MAGIC, 13, __u32) 427 #define F2FS_IOC_GET_PIN_FILE _IOR(F2FS_IOCTL_MAGIC, 14, __u32) 428 #define F2FS_IOC_PRECACHE_EXTENTS _IO(F2FS_IOCTL_MAGIC, 15) 429 #define F2FS_IOC_RESIZE_FS _IOW(F2FS_IOCTL_MAGIC, 16, __u64) 430 431 #define F2FS_IOC_GET_VOLUME_NAME FS_IOC_GETFSLABEL 432 #define F2FS_IOC_SET_VOLUME_NAME FS_IOC_SETFSLABEL 433 434 #define F2FS_IOC_SET_ENCRYPTION_POLICY FS_IOC_SET_ENCRYPTION_POLICY 435 #define F2FS_IOC_GET_ENCRYPTION_POLICY FS_IOC_GET_ENCRYPTION_POLICY 436 #define F2FS_IOC_GET_ENCRYPTION_PWSALT FS_IOC_GET_ENCRYPTION_PWSALT 437 438 /* 439 * should be same as XFS_IOC_GOINGDOWN. 440 * Flags for going down operation used by FS_IOC_GOINGDOWN 441 */ 442 #define F2FS_IOC_SHUTDOWN _IOR('X', 125, __u32) /* Shutdown */ 443 #define F2FS_GOING_DOWN_FULLSYNC 0x0 /* going down with full sync */ 444 #define F2FS_GOING_DOWN_METASYNC 0x1 /* going down with metadata */ 445 #define F2FS_GOING_DOWN_NOSYNC 0x2 /* going down */ 446 #define F2FS_GOING_DOWN_METAFLUSH 0x3 /* going down with meta flush */ 447 #define F2FS_GOING_DOWN_NEED_FSCK 0x4 /* going down to trigger fsck */ 448 449 #if defined(__KERNEL__) && defined(CONFIG_COMPAT) 450 /* 451 * ioctl commands in 32 bit emulation 452 */ 453 #define F2FS_IOC32_GETFLAGS FS_IOC32_GETFLAGS 454 #define F2FS_IOC32_SETFLAGS FS_IOC32_SETFLAGS 455 #define F2FS_IOC32_GETVERSION FS_IOC32_GETVERSION 456 #endif 457 458 #define F2FS_IOC_FSGETXATTR FS_IOC_FSGETXATTR 459 #define F2FS_IOC_FSSETXATTR FS_IOC_FSSETXATTR 460 461 struct f2fs_gc_range { 462 u32 sync; 463 u64 start; 464 u64 len; 465 }; 466 467 struct f2fs_defragment { 468 u64 start; 469 u64 len; 470 }; 471 472 struct f2fs_move_range { 473 u32 dst_fd; /* destination fd */ 474 u64 pos_in; /* start position in src_fd */ 475 u64 pos_out; /* start position in dst_fd */ 476 u64 len; /* size to move */ 477 }; 478 479 struct f2fs_flush_device { 480 u32 dev_num; /* device number to flush */ 481 u32 segments; /* # of segments to flush */ 482 }; 483 484 /* for inline stuff */ 485 #define DEF_INLINE_RESERVED_SIZE 1 486 static inline int get_extra_isize(struct inode *inode); 487 static inline int get_inline_xattr_addrs(struct inode *inode); 488 #define MAX_INLINE_DATA(inode) (sizeof(__le32) * \ 489 (CUR_ADDRS_PER_INODE(inode) - \ 490 get_inline_xattr_addrs(inode) - \ 491 DEF_INLINE_RESERVED_SIZE)) 492 493 /* for inline dir */ 494 #define NR_INLINE_DENTRY(inode) (MAX_INLINE_DATA(inode) * BITS_PER_BYTE / \ 495 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \ 496 BITS_PER_BYTE + 1)) 497 #define INLINE_DENTRY_BITMAP_SIZE(inode) \ 498 DIV_ROUND_UP(NR_INLINE_DENTRY(inode), BITS_PER_BYTE) 499 #define INLINE_RESERVED_SIZE(inode) (MAX_INLINE_DATA(inode) - \ 500 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \ 501 NR_INLINE_DENTRY(inode) + \ 502 INLINE_DENTRY_BITMAP_SIZE(inode))) 503 504 /* 505 * For INODE and NODE manager 506 */ 507 /* for directory operations */ 508 struct f2fs_dentry_ptr { 509 struct inode *inode; 510 void *bitmap; 511 struct f2fs_dir_entry *dentry; 512 __u8 (*filename)[F2FS_SLOT_LEN]; 513 int max; 514 int nr_bitmap; 515 }; 516 517 static inline void make_dentry_ptr_block(struct inode *inode, 518 struct f2fs_dentry_ptr *d, struct f2fs_dentry_block *t) 519 { 520 d->inode = inode; 521 d->max = NR_DENTRY_IN_BLOCK; 522 d->nr_bitmap = SIZE_OF_DENTRY_BITMAP; 523 d->bitmap = t->dentry_bitmap; 524 d->dentry = t->dentry; 525 d->filename = t->filename; 526 } 527 528 static inline void make_dentry_ptr_inline(struct inode *inode, 529 struct f2fs_dentry_ptr *d, void *t) 530 { 531 int entry_cnt = NR_INLINE_DENTRY(inode); 532 int bitmap_size = INLINE_DENTRY_BITMAP_SIZE(inode); 533 int reserved_size = INLINE_RESERVED_SIZE(inode); 534 535 d->inode = inode; 536 d->max = entry_cnt; 537 d->nr_bitmap = bitmap_size; 538 d->bitmap = t; 539 d->dentry = t + bitmap_size + reserved_size; 540 d->filename = t + bitmap_size + reserved_size + 541 SIZE_OF_DIR_ENTRY * entry_cnt; 542 } 543 544 /* 545 * XATTR_NODE_OFFSET stores xattrs to one node block per file keeping -1 546 * as its node offset to distinguish from index node blocks. 547 * But some bits are used to mark the node block. 548 */ 549 #define XATTR_NODE_OFFSET ((((unsigned int)-1) << OFFSET_BIT_SHIFT) \ 550 >> OFFSET_BIT_SHIFT) 551 enum { 552 ALLOC_NODE, /* allocate a new node page if needed */ 553 LOOKUP_NODE, /* look up a node without readahead */ 554 LOOKUP_NODE_RA, /* 555 * look up a node with readahead called 556 * by get_data_block. 557 */ 558 }; 559 560 #define DEFAULT_RETRY_IO_COUNT 8 /* maximum retry read IO count */ 561 562 /* maximum retry quota flush count */ 563 #define DEFAULT_RETRY_QUOTA_FLUSH_COUNT 8 564 565 #define F2FS_LINK_MAX 0xffffffff /* maximum link count per file */ 566 567 #define MAX_DIR_RA_PAGES 4 /* maximum ra pages of dir */ 568 569 /* for in-memory extent cache entry */ 570 #define F2FS_MIN_EXTENT_LEN 64 /* minimum extent length */ 571 572 /* number of extent info in extent cache we try to shrink */ 573 #define EXTENT_CACHE_SHRINK_NUMBER 128 574 575 struct rb_entry { 576 struct rb_node rb_node; /* rb node located in rb-tree */ 577 unsigned int ofs; /* start offset of the entry */ 578 unsigned int len; /* length of the entry */ 579 }; 580 581 struct extent_info { 582 unsigned int fofs; /* start offset in a file */ 583 unsigned int len; /* length of the extent */ 584 u32 blk; /* start block address of the extent */ 585 }; 586 587 struct extent_node { 588 struct rb_node rb_node; /* rb node located in rb-tree */ 589 struct extent_info ei; /* extent info */ 590 struct list_head list; /* node in global extent list of sbi */ 591 struct extent_tree *et; /* extent tree pointer */ 592 }; 593 594 struct extent_tree { 595 nid_t ino; /* inode number */ 596 struct rb_root_cached root; /* root of extent info rb-tree */ 597 struct extent_node *cached_en; /* recently accessed extent node */ 598 struct extent_info largest; /* largested extent info */ 599 struct list_head list; /* to be used by sbi->zombie_list */ 600 rwlock_t lock; /* protect extent info rb-tree */ 601 atomic_t node_cnt; /* # of extent node in rb-tree*/ 602 bool largest_updated; /* largest extent updated */ 603 }; 604 605 /* 606 * This structure is taken from ext4_map_blocks. 607 * 608 * Note that, however, f2fs uses NEW and MAPPED flags for f2fs_map_blocks(). 609 */ 610 #define F2FS_MAP_NEW (1 << BH_New) 611 #define F2FS_MAP_MAPPED (1 << BH_Mapped) 612 #define F2FS_MAP_UNWRITTEN (1 << BH_Unwritten) 613 #define F2FS_MAP_FLAGS (F2FS_MAP_NEW | F2FS_MAP_MAPPED |\ 614 F2FS_MAP_UNWRITTEN) 615 616 struct f2fs_map_blocks { 617 block_t m_pblk; 618 block_t m_lblk; 619 unsigned int m_len; 620 unsigned int m_flags; 621 pgoff_t *m_next_pgofs; /* point next possible non-hole pgofs */ 622 pgoff_t *m_next_extent; /* point to next possible extent */ 623 int m_seg_type; 624 bool m_may_create; /* indicate it is from write path */ 625 }; 626 627 /* for flag in get_data_block */ 628 enum { 629 F2FS_GET_BLOCK_DEFAULT, 630 F2FS_GET_BLOCK_FIEMAP, 631 F2FS_GET_BLOCK_BMAP, 632 F2FS_GET_BLOCK_DIO, 633 F2FS_GET_BLOCK_PRE_DIO, 634 F2FS_GET_BLOCK_PRE_AIO, 635 F2FS_GET_BLOCK_PRECACHE, 636 }; 637 638 /* 639 * i_advise uses FADVISE_XXX_BIT. We can add additional hints later. 640 */ 641 #define FADVISE_COLD_BIT 0x01 642 #define FADVISE_LOST_PINO_BIT 0x02 643 #define FADVISE_ENCRYPT_BIT 0x04 644 #define FADVISE_ENC_NAME_BIT 0x08 645 #define FADVISE_KEEP_SIZE_BIT 0x10 646 #define FADVISE_HOT_BIT 0x20 647 #define FADVISE_VERITY_BIT 0x40 648 649 #define FADVISE_MODIFIABLE_BITS (FADVISE_COLD_BIT | FADVISE_HOT_BIT) 650 651 #define file_is_cold(inode) is_file(inode, FADVISE_COLD_BIT) 652 #define file_wrong_pino(inode) is_file(inode, FADVISE_LOST_PINO_BIT) 653 #define file_set_cold(inode) set_file(inode, FADVISE_COLD_BIT) 654 #define file_lost_pino(inode) set_file(inode, FADVISE_LOST_PINO_BIT) 655 #define file_clear_cold(inode) clear_file(inode, FADVISE_COLD_BIT) 656 #define file_got_pino(inode) clear_file(inode, FADVISE_LOST_PINO_BIT) 657 #define file_is_encrypt(inode) is_file(inode, FADVISE_ENCRYPT_BIT) 658 #define file_set_encrypt(inode) set_file(inode, FADVISE_ENCRYPT_BIT) 659 #define file_clear_encrypt(inode) clear_file(inode, FADVISE_ENCRYPT_BIT) 660 #define file_enc_name(inode) is_file(inode, FADVISE_ENC_NAME_BIT) 661 #define file_set_enc_name(inode) set_file(inode, FADVISE_ENC_NAME_BIT) 662 #define file_keep_isize(inode) is_file(inode, FADVISE_KEEP_SIZE_BIT) 663 #define file_set_keep_isize(inode) set_file(inode, FADVISE_KEEP_SIZE_BIT) 664 #define file_is_hot(inode) is_file(inode, FADVISE_HOT_BIT) 665 #define file_set_hot(inode) set_file(inode, FADVISE_HOT_BIT) 666 #define file_clear_hot(inode) clear_file(inode, FADVISE_HOT_BIT) 667 #define file_is_verity(inode) is_file(inode, FADVISE_VERITY_BIT) 668 #define file_set_verity(inode) set_file(inode, FADVISE_VERITY_BIT) 669 670 #define DEF_DIR_LEVEL 0 671 672 enum { 673 GC_FAILURE_PIN, 674 GC_FAILURE_ATOMIC, 675 MAX_GC_FAILURE 676 }; 677 678 struct f2fs_inode_info { 679 struct inode vfs_inode; /* serve a vfs inode */ 680 unsigned long i_flags; /* keep an inode flags for ioctl */ 681 unsigned char i_advise; /* use to give file attribute hints */ 682 unsigned char i_dir_level; /* use for dentry level for large dir */ 683 unsigned int i_current_depth; /* only for directory depth */ 684 /* for gc failure statistic */ 685 unsigned int i_gc_failures[MAX_GC_FAILURE]; 686 unsigned int i_pino; /* parent inode number */ 687 umode_t i_acl_mode; /* keep file acl mode temporarily */ 688 689 /* Use below internally in f2fs*/ 690 unsigned long flags; /* use to pass per-file flags */ 691 struct rw_semaphore i_sem; /* protect fi info */ 692 atomic_t dirty_pages; /* # of dirty pages */ 693 f2fs_hash_t chash; /* hash value of given file name */ 694 unsigned int clevel; /* maximum level of given file name */ 695 struct task_struct *task; /* lookup and create consistency */ 696 struct task_struct *cp_task; /* separate cp/wb IO stats*/ 697 nid_t i_xattr_nid; /* node id that contains xattrs */ 698 loff_t last_disk_size; /* lastly written file size */ 699 700 #ifdef CONFIG_QUOTA 701 struct dquot *i_dquot[MAXQUOTAS]; 702 703 /* quota space reservation, managed internally by quota code */ 704 qsize_t i_reserved_quota; 705 #endif 706 struct list_head dirty_list; /* dirty list for dirs and files */ 707 struct list_head gdirty_list; /* linked in global dirty list */ 708 struct list_head inmem_ilist; /* list for inmem inodes */ 709 struct list_head inmem_pages; /* inmemory pages managed by f2fs */ 710 struct task_struct *inmem_task; /* store inmemory task */ 711 struct mutex inmem_lock; /* lock for inmemory pages */ 712 struct extent_tree *extent_tree; /* cached extent_tree entry */ 713 714 /* avoid racing between foreground op and gc */ 715 struct rw_semaphore i_gc_rwsem[2]; 716 struct rw_semaphore i_mmap_sem; 717 struct rw_semaphore i_xattr_sem; /* avoid racing between reading and changing EAs */ 718 719 int i_extra_isize; /* size of extra space located in i_addr */ 720 kprojid_t i_projid; /* id for project quota */ 721 int i_inline_xattr_size; /* inline xattr size */ 722 struct timespec64 i_crtime; /* inode creation time */ 723 struct timespec64 i_disk_time[4];/* inode disk times */ 724 725 /* for file compress */ 726 u64 i_compr_blocks; /* # of compressed blocks */ 727 unsigned char i_compress_algorithm; /* algorithm type */ 728 unsigned char i_log_cluster_size; /* log of cluster size */ 729 unsigned int i_cluster_size; /* cluster size */ 730 }; 731 732 static inline void get_extent_info(struct extent_info *ext, 733 struct f2fs_extent *i_ext) 734 { 735 ext->fofs = le32_to_cpu(i_ext->fofs); 736 ext->blk = le32_to_cpu(i_ext->blk); 737 ext->len = le32_to_cpu(i_ext->len); 738 } 739 740 static inline void set_raw_extent(struct extent_info *ext, 741 struct f2fs_extent *i_ext) 742 { 743 i_ext->fofs = cpu_to_le32(ext->fofs); 744 i_ext->blk = cpu_to_le32(ext->blk); 745 i_ext->len = cpu_to_le32(ext->len); 746 } 747 748 static inline void set_extent_info(struct extent_info *ei, unsigned int fofs, 749 u32 blk, unsigned int len) 750 { 751 ei->fofs = fofs; 752 ei->blk = blk; 753 ei->len = len; 754 } 755 756 static inline bool __is_discard_mergeable(struct discard_info *back, 757 struct discard_info *front, unsigned int max_len) 758 { 759 return (back->lstart + back->len == front->lstart) && 760 (back->len + front->len <= max_len); 761 } 762 763 static inline bool __is_discard_back_mergeable(struct discard_info *cur, 764 struct discard_info *back, unsigned int max_len) 765 { 766 return __is_discard_mergeable(back, cur, max_len); 767 } 768 769 static inline bool __is_discard_front_mergeable(struct discard_info *cur, 770 struct discard_info *front, unsigned int max_len) 771 { 772 return __is_discard_mergeable(cur, front, max_len); 773 } 774 775 static inline bool __is_extent_mergeable(struct extent_info *back, 776 struct extent_info *front) 777 { 778 return (back->fofs + back->len == front->fofs && 779 back->blk + back->len == front->blk); 780 } 781 782 static inline bool __is_back_mergeable(struct extent_info *cur, 783 struct extent_info *back) 784 { 785 return __is_extent_mergeable(back, cur); 786 } 787 788 static inline bool __is_front_mergeable(struct extent_info *cur, 789 struct extent_info *front) 790 { 791 return __is_extent_mergeable(cur, front); 792 } 793 794 extern void f2fs_mark_inode_dirty_sync(struct inode *inode, bool sync); 795 static inline void __try_update_largest_extent(struct extent_tree *et, 796 struct extent_node *en) 797 { 798 if (en->ei.len > et->largest.len) { 799 et->largest = en->ei; 800 et->largest_updated = true; 801 } 802 } 803 804 /* 805 * For free nid management 806 */ 807 enum nid_state { 808 FREE_NID, /* newly added to free nid list */ 809 PREALLOC_NID, /* it is preallocated */ 810 MAX_NID_STATE, 811 }; 812 813 struct f2fs_nm_info { 814 block_t nat_blkaddr; /* base disk address of NAT */ 815 nid_t max_nid; /* maximum possible node ids */ 816 nid_t available_nids; /* # of available node ids */ 817 nid_t next_scan_nid; /* the next nid to be scanned */ 818 unsigned int ram_thresh; /* control the memory footprint */ 819 unsigned int ra_nid_pages; /* # of nid pages to be readaheaded */ 820 unsigned int dirty_nats_ratio; /* control dirty nats ratio threshold */ 821 822 /* NAT cache management */ 823 struct radix_tree_root nat_root;/* root of the nat entry cache */ 824 struct radix_tree_root nat_set_root;/* root of the nat set cache */ 825 struct rw_semaphore nat_tree_lock; /* protect nat_tree_lock */ 826 struct list_head nat_entries; /* cached nat entry list (clean) */ 827 spinlock_t nat_list_lock; /* protect clean nat entry list */ 828 unsigned int nat_cnt; /* the # of cached nat entries */ 829 unsigned int dirty_nat_cnt; /* total num of nat entries in set */ 830 unsigned int nat_blocks; /* # of nat blocks */ 831 832 /* free node ids management */ 833 struct radix_tree_root free_nid_root;/* root of the free_nid cache */ 834 struct list_head free_nid_list; /* list for free nids excluding preallocated nids */ 835 unsigned int nid_cnt[MAX_NID_STATE]; /* the number of free node id */ 836 spinlock_t nid_list_lock; /* protect nid lists ops */ 837 struct mutex build_lock; /* lock for build free nids */ 838 unsigned char **free_nid_bitmap; 839 unsigned char *nat_block_bitmap; 840 unsigned short *free_nid_count; /* free nid count of NAT block */ 841 842 /* for checkpoint */ 843 char *nat_bitmap; /* NAT bitmap pointer */ 844 845 unsigned int nat_bits_blocks; /* # of nat bits blocks */ 846 unsigned char *nat_bits; /* NAT bits blocks */ 847 unsigned char *full_nat_bits; /* full NAT pages */ 848 unsigned char *empty_nat_bits; /* empty NAT pages */ 849 #ifdef CONFIG_F2FS_CHECK_FS 850 char *nat_bitmap_mir; /* NAT bitmap mirror */ 851 #endif 852 int bitmap_size; /* bitmap size */ 853 }; 854 855 /* 856 * this structure is used as one of function parameters. 857 * all the information are dedicated to a given direct node block determined 858 * by the data offset in a file. 859 */ 860 struct dnode_of_data { 861 struct inode *inode; /* vfs inode pointer */ 862 struct page *inode_page; /* its inode page, NULL is possible */ 863 struct page *node_page; /* cached direct node page */ 864 nid_t nid; /* node id of the direct node block */ 865 unsigned int ofs_in_node; /* data offset in the node page */ 866 bool inode_page_locked; /* inode page is locked or not */ 867 bool node_changed; /* is node block changed */ 868 char cur_level; /* level of hole node page */ 869 char max_level; /* level of current page located */ 870 block_t data_blkaddr; /* block address of the node block */ 871 }; 872 873 static inline void set_new_dnode(struct dnode_of_data *dn, struct inode *inode, 874 struct page *ipage, struct page *npage, nid_t nid) 875 { 876 memset(dn, 0, sizeof(*dn)); 877 dn->inode = inode; 878 dn->inode_page = ipage; 879 dn->node_page = npage; 880 dn->nid = nid; 881 } 882 883 /* 884 * For SIT manager 885 * 886 * By default, there are 6 active log areas across the whole main area. 887 * When considering hot and cold data separation to reduce cleaning overhead, 888 * we split 3 for data logs and 3 for node logs as hot, warm, and cold types, 889 * respectively. 890 * In the current design, you should not change the numbers intentionally. 891 * Instead, as a mount option such as active_logs=x, you can use 2, 4, and 6 892 * logs individually according to the underlying devices. (default: 6) 893 * Just in case, on-disk layout covers maximum 16 logs that consist of 8 for 894 * data and 8 for node logs. 895 */ 896 #define NR_CURSEG_DATA_TYPE (3) 897 #define NR_CURSEG_NODE_TYPE (3) 898 #define NR_CURSEG_TYPE (NR_CURSEG_DATA_TYPE + NR_CURSEG_NODE_TYPE) 899 900 enum { 901 CURSEG_HOT_DATA = 0, /* directory entry blocks */ 902 CURSEG_WARM_DATA, /* data blocks */ 903 CURSEG_COLD_DATA, /* multimedia or GCed data blocks */ 904 CURSEG_HOT_NODE, /* direct node blocks of directory files */ 905 CURSEG_WARM_NODE, /* direct node blocks of normal files */ 906 CURSEG_COLD_NODE, /* indirect node blocks */ 907 NO_CHECK_TYPE, 908 CURSEG_COLD_DATA_PINNED,/* cold data for pinned file */ 909 }; 910 911 struct flush_cmd { 912 struct completion wait; 913 struct llist_node llnode; 914 nid_t ino; 915 int ret; 916 }; 917 918 struct flush_cmd_control { 919 struct task_struct *f2fs_issue_flush; /* flush thread */ 920 wait_queue_head_t flush_wait_queue; /* waiting queue for wake-up */ 921 atomic_t issued_flush; /* # of issued flushes */ 922 atomic_t queued_flush; /* # of queued flushes */ 923 struct llist_head issue_list; /* list for command issue */ 924 struct llist_node *dispatch_list; /* list for command dispatch */ 925 }; 926 927 struct f2fs_sm_info { 928 struct sit_info *sit_info; /* whole segment information */ 929 struct free_segmap_info *free_info; /* free segment information */ 930 struct dirty_seglist_info *dirty_info; /* dirty segment information */ 931 struct curseg_info *curseg_array; /* active segment information */ 932 933 struct rw_semaphore curseg_lock; /* for preventing curseg change */ 934 935 block_t seg0_blkaddr; /* block address of 0'th segment */ 936 block_t main_blkaddr; /* start block address of main area */ 937 block_t ssa_blkaddr; /* start block address of SSA area */ 938 939 unsigned int segment_count; /* total # of segments */ 940 unsigned int main_segments; /* # of segments in main area */ 941 unsigned int reserved_segments; /* # of reserved segments */ 942 unsigned int ovp_segments; /* # of overprovision segments */ 943 944 /* a threshold to reclaim prefree segments */ 945 unsigned int rec_prefree_segments; 946 947 /* for batched trimming */ 948 unsigned int trim_sections; /* # of sections to trim */ 949 950 struct list_head sit_entry_set; /* sit entry set list */ 951 952 unsigned int ipu_policy; /* in-place-update policy */ 953 unsigned int min_ipu_util; /* in-place-update threshold */ 954 unsigned int min_fsync_blocks; /* threshold for fsync */ 955 unsigned int min_seq_blocks; /* threshold for sequential blocks */ 956 unsigned int min_hot_blocks; /* threshold for hot block allocation */ 957 unsigned int min_ssr_sections; /* threshold to trigger SSR allocation */ 958 959 /* for flush command control */ 960 struct flush_cmd_control *fcc_info; 961 962 /* for discard command control */ 963 struct discard_cmd_control *dcc_info; 964 }; 965 966 /* 967 * For superblock 968 */ 969 /* 970 * COUNT_TYPE for monitoring 971 * 972 * f2fs monitors the number of several block types such as on-writeback, 973 * dirty dentry blocks, dirty node blocks, and dirty meta blocks. 974 */ 975 #define WB_DATA_TYPE(p) (__is_cp_guaranteed(p) ? F2FS_WB_CP_DATA : F2FS_WB_DATA) 976 enum count_type { 977 F2FS_DIRTY_DENTS, 978 F2FS_DIRTY_DATA, 979 F2FS_DIRTY_QDATA, 980 F2FS_DIRTY_NODES, 981 F2FS_DIRTY_META, 982 F2FS_INMEM_PAGES, 983 F2FS_DIRTY_IMETA, 984 F2FS_WB_CP_DATA, 985 F2FS_WB_DATA, 986 F2FS_RD_DATA, 987 F2FS_RD_NODE, 988 F2FS_RD_META, 989 F2FS_DIO_WRITE, 990 F2FS_DIO_READ, 991 NR_COUNT_TYPE, 992 }; 993 994 /* 995 * The below are the page types of bios used in submit_bio(). 996 * The available types are: 997 * DATA User data pages. It operates as async mode. 998 * NODE Node pages. It operates as async mode. 999 * META FS metadata pages such as SIT, NAT, CP. 1000 * NR_PAGE_TYPE The number of page types. 1001 * META_FLUSH Make sure the previous pages are written 1002 * with waiting the bio's completion 1003 * ... Only can be used with META. 1004 */ 1005 #define PAGE_TYPE_OF_BIO(type) ((type) > META ? META : (type)) 1006 enum page_type { 1007 DATA, 1008 NODE, 1009 META, 1010 NR_PAGE_TYPE, 1011 META_FLUSH, 1012 INMEM, /* the below types are used by tracepoints only. */ 1013 INMEM_DROP, 1014 INMEM_INVALIDATE, 1015 INMEM_REVOKE, 1016 IPU, 1017 OPU, 1018 }; 1019 1020 enum temp_type { 1021 HOT = 0, /* must be zero for meta bio */ 1022 WARM, 1023 COLD, 1024 NR_TEMP_TYPE, 1025 }; 1026 1027 enum need_lock_type { 1028 LOCK_REQ = 0, 1029 LOCK_DONE, 1030 LOCK_RETRY, 1031 }; 1032 1033 enum cp_reason_type { 1034 CP_NO_NEEDED, 1035 CP_NON_REGULAR, 1036 CP_COMPRESSED, 1037 CP_HARDLINK, 1038 CP_SB_NEED_CP, 1039 CP_WRONG_PINO, 1040 CP_NO_SPC_ROLL, 1041 CP_NODE_NEED_CP, 1042 CP_FASTBOOT_MODE, 1043 CP_SPEC_LOG_NUM, 1044 CP_RECOVER_DIR, 1045 }; 1046 1047 enum iostat_type { 1048 APP_DIRECT_IO, /* app direct IOs */ 1049 APP_BUFFERED_IO, /* app buffered IOs */ 1050 APP_WRITE_IO, /* app write IOs */ 1051 APP_MAPPED_IO, /* app mapped IOs */ 1052 FS_DATA_IO, /* data IOs from kworker/fsync/reclaimer */ 1053 FS_NODE_IO, /* node IOs from kworker/fsync/reclaimer */ 1054 FS_META_IO, /* meta IOs from kworker/reclaimer */ 1055 FS_GC_DATA_IO, /* data IOs from forground gc */ 1056 FS_GC_NODE_IO, /* node IOs from forground gc */ 1057 FS_CP_DATA_IO, /* data IOs from checkpoint */ 1058 FS_CP_NODE_IO, /* node IOs from checkpoint */ 1059 FS_CP_META_IO, /* meta IOs from checkpoint */ 1060 FS_DISCARD, /* discard */ 1061 NR_IO_TYPE, 1062 }; 1063 1064 struct f2fs_io_info { 1065 struct f2fs_sb_info *sbi; /* f2fs_sb_info pointer */ 1066 nid_t ino; /* inode number */ 1067 enum page_type type; /* contains DATA/NODE/META/META_FLUSH */ 1068 enum temp_type temp; /* contains HOT/WARM/COLD */ 1069 int op; /* contains REQ_OP_ */ 1070 int op_flags; /* req_flag_bits */ 1071 block_t new_blkaddr; /* new block address to be written */ 1072 block_t old_blkaddr; /* old block address before Cow */ 1073 struct page *page; /* page to be written */ 1074 struct page *encrypted_page; /* encrypted page */ 1075 struct page *compressed_page; /* compressed page */ 1076 struct list_head list; /* serialize IOs */ 1077 bool submitted; /* indicate IO submission */ 1078 int need_lock; /* indicate we need to lock cp_rwsem */ 1079 bool in_list; /* indicate fio is in io_list */ 1080 bool is_por; /* indicate IO is from recovery or not */ 1081 bool retry; /* need to reallocate block address */ 1082 int compr_blocks; /* # of compressed block addresses */ 1083 bool encrypted; /* indicate file is encrypted */ 1084 enum iostat_type io_type; /* io type */ 1085 struct writeback_control *io_wbc; /* writeback control */ 1086 struct bio **bio; /* bio for ipu */ 1087 sector_t *last_block; /* last block number in bio */ 1088 unsigned char version; /* version of the node */ 1089 }; 1090 1091 struct bio_entry { 1092 struct bio *bio; 1093 struct list_head list; 1094 }; 1095 1096 #define is_read_io(rw) ((rw) == READ) 1097 struct f2fs_bio_info { 1098 struct f2fs_sb_info *sbi; /* f2fs superblock */ 1099 struct bio *bio; /* bios to merge */ 1100 sector_t last_block_in_bio; /* last block number */ 1101 struct f2fs_io_info fio; /* store buffered io info. */ 1102 struct rw_semaphore io_rwsem; /* blocking op for bio */ 1103 spinlock_t io_lock; /* serialize DATA/NODE IOs */ 1104 struct list_head io_list; /* track fios */ 1105 struct list_head bio_list; /* bio entry list head */ 1106 struct rw_semaphore bio_list_lock; /* lock to protect bio entry list */ 1107 }; 1108 1109 #define FDEV(i) (sbi->devs[i]) 1110 #define RDEV(i) (raw_super->devs[i]) 1111 struct f2fs_dev_info { 1112 struct block_device *bdev; 1113 char path[MAX_PATH_LEN]; 1114 unsigned int total_segments; 1115 block_t start_blk; 1116 block_t end_blk; 1117 #ifdef CONFIG_BLK_DEV_ZONED 1118 unsigned int nr_blkz; /* Total number of zones */ 1119 unsigned long *blkz_seq; /* Bitmap indicating sequential zones */ 1120 #endif 1121 }; 1122 1123 enum inode_type { 1124 DIR_INODE, /* for dirty dir inode */ 1125 FILE_INODE, /* for dirty regular/symlink inode */ 1126 DIRTY_META, /* for all dirtied inode metadata */ 1127 ATOMIC_FILE, /* for all atomic files */ 1128 NR_INODE_TYPE, 1129 }; 1130 1131 /* for inner inode cache management */ 1132 struct inode_management { 1133 struct radix_tree_root ino_root; /* ino entry array */ 1134 spinlock_t ino_lock; /* for ino entry lock */ 1135 struct list_head ino_list; /* inode list head */ 1136 unsigned long ino_num; /* number of entries */ 1137 }; 1138 1139 /* For s_flag in struct f2fs_sb_info */ 1140 enum { 1141 SBI_IS_DIRTY, /* dirty flag for checkpoint */ 1142 SBI_IS_CLOSE, /* specify unmounting */ 1143 SBI_NEED_FSCK, /* need fsck.f2fs to fix */ 1144 SBI_POR_DOING, /* recovery is doing or not */ 1145 SBI_NEED_SB_WRITE, /* need to recover superblock */ 1146 SBI_NEED_CP, /* need to checkpoint */ 1147 SBI_IS_SHUTDOWN, /* shutdown by ioctl */ 1148 SBI_IS_RECOVERED, /* recovered orphan/data */ 1149 SBI_CP_DISABLED, /* CP was disabled last mount */ 1150 SBI_CP_DISABLED_QUICK, /* CP was disabled quickly */ 1151 SBI_QUOTA_NEED_FLUSH, /* need to flush quota info in CP */ 1152 SBI_QUOTA_SKIP_FLUSH, /* skip flushing quota in current CP */ 1153 SBI_QUOTA_NEED_REPAIR, /* quota file may be corrupted */ 1154 SBI_IS_RESIZEFS, /* resizefs is in process */ 1155 }; 1156 1157 enum { 1158 CP_TIME, 1159 REQ_TIME, 1160 DISCARD_TIME, 1161 GC_TIME, 1162 DISABLE_TIME, 1163 UMOUNT_DISCARD_TIMEOUT, 1164 MAX_TIME, 1165 }; 1166 1167 enum { 1168 GC_NORMAL, 1169 GC_IDLE_CB, 1170 GC_IDLE_GREEDY, 1171 GC_URGENT, 1172 }; 1173 1174 enum { 1175 WHINT_MODE_OFF, /* not pass down write hints */ 1176 WHINT_MODE_USER, /* try to pass down hints given by users */ 1177 WHINT_MODE_FS, /* pass down hints with F2FS policy */ 1178 }; 1179 1180 enum { 1181 ALLOC_MODE_DEFAULT, /* stay default */ 1182 ALLOC_MODE_REUSE, /* reuse segments as much as possible */ 1183 }; 1184 1185 enum fsync_mode { 1186 FSYNC_MODE_POSIX, /* fsync follows posix semantics */ 1187 FSYNC_MODE_STRICT, /* fsync behaves in line with ext4 */ 1188 FSYNC_MODE_NOBARRIER, /* fsync behaves nobarrier based on posix */ 1189 }; 1190 1191 /* 1192 * this value is set in page as a private data which indicate that 1193 * the page is atomically written, and it is in inmem_pages list. 1194 */ 1195 #define ATOMIC_WRITTEN_PAGE ((unsigned long)-1) 1196 #define DUMMY_WRITTEN_PAGE ((unsigned long)-2) 1197 1198 #define IS_ATOMIC_WRITTEN_PAGE(page) \ 1199 (page_private(page) == (unsigned long)ATOMIC_WRITTEN_PAGE) 1200 #define IS_DUMMY_WRITTEN_PAGE(page) \ 1201 (page_private(page) == (unsigned long)DUMMY_WRITTEN_PAGE) 1202 1203 #ifdef CONFIG_FS_ENCRYPTION 1204 #define DUMMY_ENCRYPTION_ENABLED(sbi) \ 1205 (unlikely(F2FS_OPTION(sbi).test_dummy_encryption)) 1206 #else 1207 #define DUMMY_ENCRYPTION_ENABLED(sbi) (0) 1208 #endif 1209 1210 /* For compression */ 1211 enum compress_algorithm_type { 1212 COMPRESS_LZO, 1213 COMPRESS_LZ4, 1214 COMPRESS_MAX, 1215 }; 1216 1217 #define COMPRESS_DATA_RESERVED_SIZE 4 1218 struct compress_data { 1219 __le32 clen; /* compressed data size */ 1220 __le32 chksum; /* checksum of compressed data */ 1221 __le32 reserved[COMPRESS_DATA_RESERVED_SIZE]; /* reserved */ 1222 u8 cdata[]; /* compressed data */ 1223 }; 1224 1225 #define COMPRESS_HEADER_SIZE (sizeof(struct compress_data)) 1226 1227 #define F2FS_COMPRESSED_PAGE_MAGIC 0xF5F2C000 1228 1229 /* compress context */ 1230 struct compress_ctx { 1231 struct inode *inode; /* inode the context belong to */ 1232 pgoff_t cluster_idx; /* cluster index number */ 1233 unsigned int cluster_size; /* page count in cluster */ 1234 unsigned int log_cluster_size; /* log of cluster size */ 1235 struct page **rpages; /* pages store raw data in cluster */ 1236 unsigned int nr_rpages; /* total page number in rpages */ 1237 struct page **cpages; /* pages store compressed data in cluster */ 1238 unsigned int nr_cpages; /* total page number in cpages */ 1239 void *rbuf; /* virtual mapped address on rpages */ 1240 struct compress_data *cbuf; /* virtual mapped address on cpages */ 1241 size_t rlen; /* valid data length in rbuf */ 1242 size_t clen; /* valid data length in cbuf */ 1243 void *private; /* payload buffer for specified compression algorithm */ 1244 }; 1245 1246 /* compress context for write IO path */ 1247 struct compress_io_ctx { 1248 u32 magic; /* magic number to indicate page is compressed */ 1249 struct inode *inode; /* inode the context belong to */ 1250 struct page **rpages; /* pages store raw data in cluster */ 1251 unsigned int nr_rpages; /* total page number in rpages */ 1252 refcount_t ref; /* referrence count of raw page */ 1253 }; 1254 1255 /* decompress io context for read IO path */ 1256 struct decompress_io_ctx { 1257 u32 magic; /* magic number to indicate page is compressed */ 1258 struct inode *inode; /* inode the context belong to */ 1259 pgoff_t cluster_idx; /* cluster index number */ 1260 unsigned int cluster_size; /* page count in cluster */ 1261 unsigned int log_cluster_size; /* log of cluster size */ 1262 struct page **rpages; /* pages store raw data in cluster */ 1263 unsigned int nr_rpages; /* total page number in rpages */ 1264 struct page **cpages; /* pages store compressed data in cluster */ 1265 unsigned int nr_cpages; /* total page number in cpages */ 1266 struct page **tpages; /* temp pages to pad holes in cluster */ 1267 void *rbuf; /* virtual mapped address on rpages */ 1268 struct compress_data *cbuf; /* virtual mapped address on cpages */ 1269 size_t rlen; /* valid data length in rbuf */ 1270 size_t clen; /* valid data length in cbuf */ 1271 refcount_t ref; /* referrence count of compressed page */ 1272 bool failed; /* indicate IO error during decompression */ 1273 }; 1274 1275 #define NULL_CLUSTER ((unsigned int)(~0)) 1276 #define MIN_COMPRESS_LOG_SIZE 2 1277 #define MAX_COMPRESS_LOG_SIZE 8 1278 1279 struct f2fs_sb_info { 1280 struct super_block *sb; /* pointer to VFS super block */ 1281 struct proc_dir_entry *s_proc; /* proc entry */ 1282 struct f2fs_super_block *raw_super; /* raw super block pointer */ 1283 struct rw_semaphore sb_lock; /* lock for raw super block */ 1284 int valid_super_block; /* valid super block no */ 1285 unsigned long s_flag; /* flags for sbi */ 1286 struct mutex writepages; /* mutex for writepages() */ 1287 #ifdef CONFIG_UNICODE 1288 struct unicode_map *s_encoding; 1289 __u16 s_encoding_flags; 1290 #endif 1291 1292 #ifdef CONFIG_BLK_DEV_ZONED 1293 unsigned int blocks_per_blkz; /* F2FS blocks per zone */ 1294 unsigned int log_blocks_per_blkz; /* log2 F2FS blocks per zone */ 1295 #endif 1296 1297 /* for node-related operations */ 1298 struct f2fs_nm_info *nm_info; /* node manager */ 1299 struct inode *node_inode; /* cache node blocks */ 1300 1301 /* for segment-related operations */ 1302 struct f2fs_sm_info *sm_info; /* segment manager */ 1303 1304 /* for bio operations */ 1305 struct f2fs_bio_info *write_io[NR_PAGE_TYPE]; /* for write bios */ 1306 /* keep migration IO order for LFS mode */ 1307 struct rw_semaphore io_order_lock; 1308 mempool_t *write_io_dummy; /* Dummy pages */ 1309 1310 /* for checkpoint */ 1311 struct f2fs_checkpoint *ckpt; /* raw checkpoint pointer */ 1312 int cur_cp_pack; /* remain current cp pack */ 1313 spinlock_t cp_lock; /* for flag in ckpt */ 1314 struct inode *meta_inode; /* cache meta blocks */ 1315 struct mutex cp_mutex; /* checkpoint procedure lock */ 1316 struct rw_semaphore cp_rwsem; /* blocking FS operations */ 1317 struct rw_semaphore node_write; /* locking node writes */ 1318 struct rw_semaphore node_change; /* locking node change */ 1319 wait_queue_head_t cp_wait; 1320 unsigned long last_time[MAX_TIME]; /* to store time in jiffies */ 1321 long interval_time[MAX_TIME]; /* to store thresholds */ 1322 1323 struct inode_management im[MAX_INO_ENTRY]; /* manage inode cache */ 1324 1325 spinlock_t fsync_node_lock; /* for node entry lock */ 1326 struct list_head fsync_node_list; /* node list head */ 1327 unsigned int fsync_seg_id; /* sequence id */ 1328 unsigned int fsync_node_num; /* number of node entries */ 1329 1330 /* for orphan inode, use 0'th array */ 1331 unsigned int max_orphans; /* max orphan inodes */ 1332 1333 /* for inode management */ 1334 struct list_head inode_list[NR_INODE_TYPE]; /* dirty inode list */ 1335 spinlock_t inode_lock[NR_INODE_TYPE]; /* for dirty inode list lock */ 1336 struct mutex flush_lock; /* for flush exclusion */ 1337 1338 /* for extent tree cache */ 1339 struct radix_tree_root extent_tree_root;/* cache extent cache entries */ 1340 struct mutex extent_tree_lock; /* locking extent radix tree */ 1341 struct list_head extent_list; /* lru list for shrinker */ 1342 spinlock_t extent_lock; /* locking extent lru list */ 1343 atomic_t total_ext_tree; /* extent tree count */ 1344 struct list_head zombie_list; /* extent zombie tree list */ 1345 atomic_t total_zombie_tree; /* extent zombie tree count */ 1346 atomic_t total_ext_node; /* extent info count */ 1347 1348 /* basic filesystem units */ 1349 unsigned int log_sectors_per_block; /* log2 sectors per block */ 1350 unsigned int log_blocksize; /* log2 block size */ 1351 unsigned int blocksize; /* block size */ 1352 unsigned int root_ino_num; /* root inode number*/ 1353 unsigned int node_ino_num; /* node inode number*/ 1354 unsigned int meta_ino_num; /* meta inode number*/ 1355 unsigned int log_blocks_per_seg; /* log2 blocks per segment */ 1356 unsigned int blocks_per_seg; /* blocks per segment */ 1357 unsigned int segs_per_sec; /* segments per section */ 1358 unsigned int secs_per_zone; /* sections per zone */ 1359 unsigned int total_sections; /* total section count */ 1360 struct mutex resize_mutex; /* for resize exclusion */ 1361 unsigned int total_node_count; /* total node block count */ 1362 unsigned int total_valid_node_count; /* valid node block count */ 1363 loff_t max_file_blocks; /* max block index of file */ 1364 int dir_level; /* directory level */ 1365 int readdir_ra; /* readahead inode in readdir */ 1366 1367 block_t user_block_count; /* # of user blocks */ 1368 block_t total_valid_block_count; /* # of valid blocks */ 1369 block_t discard_blks; /* discard command candidats */ 1370 block_t last_valid_block_count; /* for recovery */ 1371 block_t reserved_blocks; /* configurable reserved blocks */ 1372 block_t current_reserved_blocks; /* current reserved blocks */ 1373 1374 /* Additional tracking for no checkpoint mode */ 1375 block_t unusable_block_count; /* # of blocks saved by last cp */ 1376 1377 unsigned int nquota_files; /* # of quota sysfile */ 1378 struct rw_semaphore quota_sem; /* blocking cp for flags */ 1379 1380 /* # of pages, see count_type */ 1381 atomic_t nr_pages[NR_COUNT_TYPE]; 1382 /* # of allocated blocks */ 1383 struct percpu_counter alloc_valid_block_count; 1384 1385 /* writeback control */ 1386 atomic_t wb_sync_req[META]; /* count # of WB_SYNC threads */ 1387 1388 /* valid inode count */ 1389 struct percpu_counter total_valid_inode_count; 1390 1391 struct f2fs_mount_info mount_opt; /* mount options */ 1392 1393 /* for cleaning operations */ 1394 struct rw_semaphore gc_lock; /* 1395 * semaphore for GC, avoid 1396 * race between GC and GC or CP 1397 */ 1398 struct f2fs_gc_kthread *gc_thread; /* GC thread */ 1399 unsigned int cur_victim_sec; /* current victim section num */ 1400 unsigned int gc_mode; /* current GC state */ 1401 unsigned int next_victim_seg[2]; /* next segment in victim section */ 1402 /* for skip statistic */ 1403 unsigned int atomic_files; /* # of opened atomic file */ 1404 unsigned long long skipped_atomic_files[2]; /* FG_GC and BG_GC */ 1405 unsigned long long skipped_gc_rwsem; /* FG_GC only */ 1406 1407 /* threshold for gc trials on pinned files */ 1408 u64 gc_pin_file_threshold; 1409 struct rw_semaphore pin_sem; 1410 1411 /* maximum # of trials to find a victim segment for SSR and GC */ 1412 unsigned int max_victim_search; 1413 /* migration granularity of garbage collection, unit: segment */ 1414 unsigned int migration_granularity; 1415 1416 /* 1417 * for stat information. 1418 * one is for the LFS mode, and the other is for the SSR mode. 1419 */ 1420 #ifdef CONFIG_F2FS_STAT_FS 1421 struct f2fs_stat_info *stat_info; /* FS status information */ 1422 atomic_t meta_count[META_MAX]; /* # of meta blocks */ 1423 unsigned int segment_count[2]; /* # of allocated segments */ 1424 unsigned int block_count[2]; /* # of allocated blocks */ 1425 atomic_t inplace_count; /* # of inplace update */ 1426 atomic64_t total_hit_ext; /* # of lookup extent cache */ 1427 atomic64_t read_hit_rbtree; /* # of hit rbtree extent node */ 1428 atomic64_t read_hit_largest; /* # of hit largest extent node */ 1429 atomic64_t read_hit_cached; /* # of hit cached extent node */ 1430 atomic_t inline_xattr; /* # of inline_xattr inodes */ 1431 atomic_t inline_inode; /* # of inline_data inodes */ 1432 atomic_t inline_dir; /* # of inline_dentry inodes */ 1433 atomic_t compr_inode; /* # of compressed inodes */ 1434 atomic_t compr_blocks; /* # of compressed blocks */ 1435 atomic_t vw_cnt; /* # of volatile writes */ 1436 atomic_t max_aw_cnt; /* max # of atomic writes */ 1437 atomic_t max_vw_cnt; /* max # of volatile writes */ 1438 unsigned int io_skip_bggc; /* skip background gc for in-flight IO */ 1439 unsigned int other_skip_bggc; /* skip background gc for other reasons */ 1440 unsigned int ndirty_inode[NR_INODE_TYPE]; /* # of dirty inodes */ 1441 #endif 1442 spinlock_t stat_lock; /* lock for stat operations */ 1443 1444 /* For app/fs IO statistics */ 1445 spinlock_t iostat_lock; 1446 unsigned long long write_iostat[NR_IO_TYPE]; 1447 bool iostat_enable; 1448 1449 /* For sysfs suppport */ 1450 struct kobject s_kobj; 1451 struct completion s_kobj_unregister; 1452 1453 /* For shrinker support */ 1454 struct list_head s_list; 1455 int s_ndevs; /* number of devices */ 1456 struct f2fs_dev_info *devs; /* for device list */ 1457 unsigned int dirty_device; /* for checkpoint data flush */ 1458 spinlock_t dev_lock; /* protect dirty_device */ 1459 struct mutex umount_mutex; 1460 unsigned int shrinker_run_no; 1461 1462 /* For write statistics */ 1463 u64 sectors_written_start; 1464 u64 kbytes_written; 1465 1466 /* Reference to checksum algorithm driver via cryptoapi */ 1467 struct crypto_shash *s_chksum_driver; 1468 1469 /* Precomputed FS UUID checksum for seeding other checksums */ 1470 __u32 s_chksum_seed; 1471 1472 struct workqueue_struct *post_read_wq; /* post read workqueue */ 1473 }; 1474 1475 struct f2fs_private_dio { 1476 struct inode *inode; 1477 void *orig_private; 1478 bio_end_io_t *orig_end_io; 1479 bool write; 1480 }; 1481 1482 #ifdef CONFIG_F2FS_FAULT_INJECTION 1483 #define f2fs_show_injection_info(sbi, type) \ 1484 printk_ratelimited("%sF2FS-fs (%s) : inject %s in %s of %pS\n", \ 1485 KERN_INFO, sbi->sb->s_id, \ 1486 f2fs_fault_name[type], \ 1487 __func__, __builtin_return_address(0)) 1488 static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type) 1489 { 1490 struct f2fs_fault_info *ffi = &F2FS_OPTION(sbi).fault_info; 1491 1492 if (!ffi->inject_rate) 1493 return false; 1494 1495 if (!IS_FAULT_SET(ffi, type)) 1496 return false; 1497 1498 atomic_inc(&ffi->inject_ops); 1499 if (atomic_read(&ffi->inject_ops) >= ffi->inject_rate) { 1500 atomic_set(&ffi->inject_ops, 0); 1501 return true; 1502 } 1503 return false; 1504 } 1505 #else 1506 #define f2fs_show_injection_info(sbi, type) do { } while (0) 1507 static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type) 1508 { 1509 return false; 1510 } 1511 #endif 1512 1513 /* 1514 * Test if the mounted volume is a multi-device volume. 1515 * - For a single regular disk volume, sbi->s_ndevs is 0. 1516 * - For a single zoned disk volume, sbi->s_ndevs is 1. 1517 * - For a multi-device volume, sbi->s_ndevs is always 2 or more. 1518 */ 1519 static inline bool f2fs_is_multi_device(struct f2fs_sb_info *sbi) 1520 { 1521 return sbi->s_ndevs > 1; 1522 } 1523 1524 /* For write statistics. Suppose sector size is 512 bytes, 1525 * and the return value is in kbytes. s is of struct f2fs_sb_info. 1526 */ 1527 #define BD_PART_WRITTEN(s) \ 1528 (((u64)part_stat_read((s)->sb->s_bdev->bd_part, sectors[STAT_WRITE]) - \ 1529 (s)->sectors_written_start) >> 1) 1530 1531 static inline void f2fs_update_time(struct f2fs_sb_info *sbi, int type) 1532 { 1533 unsigned long now = jiffies; 1534 1535 sbi->last_time[type] = now; 1536 1537 /* DISCARD_TIME and GC_TIME are based on REQ_TIME */ 1538 if (type == REQ_TIME) { 1539 sbi->last_time[DISCARD_TIME] = now; 1540 sbi->last_time[GC_TIME] = now; 1541 } 1542 } 1543 1544 static inline bool f2fs_time_over(struct f2fs_sb_info *sbi, int type) 1545 { 1546 unsigned long interval = sbi->interval_time[type] * HZ; 1547 1548 return time_after(jiffies, sbi->last_time[type] + interval); 1549 } 1550 1551 static inline unsigned int f2fs_time_to_wait(struct f2fs_sb_info *sbi, 1552 int type) 1553 { 1554 unsigned long interval = sbi->interval_time[type] * HZ; 1555 unsigned int wait_ms = 0; 1556 long delta; 1557 1558 delta = (sbi->last_time[type] + interval) - jiffies; 1559 if (delta > 0) 1560 wait_ms = jiffies_to_msecs(delta); 1561 1562 return wait_ms; 1563 } 1564 1565 /* 1566 * Inline functions 1567 */ 1568 static inline u32 __f2fs_crc32(struct f2fs_sb_info *sbi, u32 crc, 1569 const void *address, unsigned int length) 1570 { 1571 struct { 1572 struct shash_desc shash; 1573 char ctx[4]; 1574 } desc; 1575 int err; 1576 1577 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver) != sizeof(desc.ctx)); 1578 1579 desc.shash.tfm = sbi->s_chksum_driver; 1580 *(u32 *)desc.ctx = crc; 1581 1582 err = crypto_shash_update(&desc.shash, address, length); 1583 BUG_ON(err); 1584 1585 return *(u32 *)desc.ctx; 1586 } 1587 1588 static inline u32 f2fs_crc32(struct f2fs_sb_info *sbi, const void *address, 1589 unsigned int length) 1590 { 1591 return __f2fs_crc32(sbi, F2FS_SUPER_MAGIC, address, length); 1592 } 1593 1594 static inline bool f2fs_crc_valid(struct f2fs_sb_info *sbi, __u32 blk_crc, 1595 void *buf, size_t buf_size) 1596 { 1597 return f2fs_crc32(sbi, buf, buf_size) == blk_crc; 1598 } 1599 1600 static inline u32 f2fs_chksum(struct f2fs_sb_info *sbi, u32 crc, 1601 const void *address, unsigned int length) 1602 { 1603 return __f2fs_crc32(sbi, crc, address, length); 1604 } 1605 1606 static inline struct f2fs_inode_info *F2FS_I(struct inode *inode) 1607 { 1608 return container_of(inode, struct f2fs_inode_info, vfs_inode); 1609 } 1610 1611 static inline struct f2fs_sb_info *F2FS_SB(struct super_block *sb) 1612 { 1613 return sb->s_fs_info; 1614 } 1615 1616 static inline struct f2fs_sb_info *F2FS_I_SB(struct inode *inode) 1617 { 1618 return F2FS_SB(inode->i_sb); 1619 } 1620 1621 static inline struct f2fs_sb_info *F2FS_M_SB(struct address_space *mapping) 1622 { 1623 return F2FS_I_SB(mapping->host); 1624 } 1625 1626 static inline struct f2fs_sb_info *F2FS_P_SB(struct page *page) 1627 { 1628 return F2FS_M_SB(page_file_mapping(page)); 1629 } 1630 1631 static inline struct f2fs_super_block *F2FS_RAW_SUPER(struct f2fs_sb_info *sbi) 1632 { 1633 return (struct f2fs_super_block *)(sbi->raw_super); 1634 } 1635 1636 static inline struct f2fs_checkpoint *F2FS_CKPT(struct f2fs_sb_info *sbi) 1637 { 1638 return (struct f2fs_checkpoint *)(sbi->ckpt); 1639 } 1640 1641 static inline struct f2fs_node *F2FS_NODE(struct page *page) 1642 { 1643 return (struct f2fs_node *)page_address(page); 1644 } 1645 1646 static inline struct f2fs_inode *F2FS_INODE(struct page *page) 1647 { 1648 return &((struct f2fs_node *)page_address(page))->i; 1649 } 1650 1651 static inline struct f2fs_nm_info *NM_I(struct f2fs_sb_info *sbi) 1652 { 1653 return (struct f2fs_nm_info *)(sbi->nm_info); 1654 } 1655 1656 static inline struct f2fs_sm_info *SM_I(struct f2fs_sb_info *sbi) 1657 { 1658 return (struct f2fs_sm_info *)(sbi->sm_info); 1659 } 1660 1661 static inline struct sit_info *SIT_I(struct f2fs_sb_info *sbi) 1662 { 1663 return (struct sit_info *)(SM_I(sbi)->sit_info); 1664 } 1665 1666 static inline struct free_segmap_info *FREE_I(struct f2fs_sb_info *sbi) 1667 { 1668 return (struct free_segmap_info *)(SM_I(sbi)->free_info); 1669 } 1670 1671 static inline struct dirty_seglist_info *DIRTY_I(struct f2fs_sb_info *sbi) 1672 { 1673 return (struct dirty_seglist_info *)(SM_I(sbi)->dirty_info); 1674 } 1675 1676 static inline struct address_space *META_MAPPING(struct f2fs_sb_info *sbi) 1677 { 1678 return sbi->meta_inode->i_mapping; 1679 } 1680 1681 static inline struct address_space *NODE_MAPPING(struct f2fs_sb_info *sbi) 1682 { 1683 return sbi->node_inode->i_mapping; 1684 } 1685 1686 static inline bool is_sbi_flag_set(struct f2fs_sb_info *sbi, unsigned int type) 1687 { 1688 return test_bit(type, &sbi->s_flag); 1689 } 1690 1691 static inline void set_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type) 1692 { 1693 set_bit(type, &sbi->s_flag); 1694 } 1695 1696 static inline void clear_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type) 1697 { 1698 clear_bit(type, &sbi->s_flag); 1699 } 1700 1701 static inline unsigned long long cur_cp_version(struct f2fs_checkpoint *cp) 1702 { 1703 return le64_to_cpu(cp->checkpoint_ver); 1704 } 1705 1706 static inline unsigned long f2fs_qf_ino(struct super_block *sb, int type) 1707 { 1708 if (type < F2FS_MAX_QUOTAS) 1709 return le32_to_cpu(F2FS_SB(sb)->raw_super->qf_ino[type]); 1710 return 0; 1711 } 1712 1713 static inline __u64 cur_cp_crc(struct f2fs_checkpoint *cp) 1714 { 1715 size_t crc_offset = le32_to_cpu(cp->checksum_offset); 1716 return le32_to_cpu(*((__le32 *)((unsigned char *)cp + crc_offset))); 1717 } 1718 1719 static inline bool __is_set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 1720 { 1721 unsigned int ckpt_flags = le32_to_cpu(cp->ckpt_flags); 1722 1723 return ckpt_flags & f; 1724 } 1725 1726 static inline bool is_set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 1727 { 1728 return __is_set_ckpt_flags(F2FS_CKPT(sbi), f); 1729 } 1730 1731 static inline void __set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 1732 { 1733 unsigned int ckpt_flags; 1734 1735 ckpt_flags = le32_to_cpu(cp->ckpt_flags); 1736 ckpt_flags |= f; 1737 cp->ckpt_flags = cpu_to_le32(ckpt_flags); 1738 } 1739 1740 static inline void set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 1741 { 1742 unsigned long flags; 1743 1744 spin_lock_irqsave(&sbi->cp_lock, flags); 1745 __set_ckpt_flags(F2FS_CKPT(sbi), f); 1746 spin_unlock_irqrestore(&sbi->cp_lock, flags); 1747 } 1748 1749 static inline void __clear_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 1750 { 1751 unsigned int ckpt_flags; 1752 1753 ckpt_flags = le32_to_cpu(cp->ckpt_flags); 1754 ckpt_flags &= (~f); 1755 cp->ckpt_flags = cpu_to_le32(ckpt_flags); 1756 } 1757 1758 static inline void clear_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 1759 { 1760 unsigned long flags; 1761 1762 spin_lock_irqsave(&sbi->cp_lock, flags); 1763 __clear_ckpt_flags(F2FS_CKPT(sbi), f); 1764 spin_unlock_irqrestore(&sbi->cp_lock, flags); 1765 } 1766 1767 static inline void disable_nat_bits(struct f2fs_sb_info *sbi, bool lock) 1768 { 1769 unsigned long flags; 1770 unsigned char *nat_bits; 1771 1772 /* 1773 * In order to re-enable nat_bits we need to call fsck.f2fs by 1774 * set_sbi_flag(sbi, SBI_NEED_FSCK). But it may give huge cost, 1775 * so let's rely on regular fsck or unclean shutdown. 1776 */ 1777 1778 if (lock) 1779 spin_lock_irqsave(&sbi->cp_lock, flags); 1780 __clear_ckpt_flags(F2FS_CKPT(sbi), CP_NAT_BITS_FLAG); 1781 nat_bits = NM_I(sbi)->nat_bits; 1782 NM_I(sbi)->nat_bits = NULL; 1783 if (lock) 1784 spin_unlock_irqrestore(&sbi->cp_lock, flags); 1785 1786 kvfree(nat_bits); 1787 } 1788 1789 static inline bool enabled_nat_bits(struct f2fs_sb_info *sbi, 1790 struct cp_control *cpc) 1791 { 1792 bool set = is_set_ckpt_flags(sbi, CP_NAT_BITS_FLAG); 1793 1794 return (cpc) ? (cpc->reason & CP_UMOUNT) && set : set; 1795 } 1796 1797 static inline void f2fs_lock_op(struct f2fs_sb_info *sbi) 1798 { 1799 down_read(&sbi->cp_rwsem); 1800 } 1801 1802 static inline int f2fs_trylock_op(struct f2fs_sb_info *sbi) 1803 { 1804 return down_read_trylock(&sbi->cp_rwsem); 1805 } 1806 1807 static inline void f2fs_unlock_op(struct f2fs_sb_info *sbi) 1808 { 1809 up_read(&sbi->cp_rwsem); 1810 } 1811 1812 static inline void f2fs_lock_all(struct f2fs_sb_info *sbi) 1813 { 1814 down_write(&sbi->cp_rwsem); 1815 } 1816 1817 static inline void f2fs_unlock_all(struct f2fs_sb_info *sbi) 1818 { 1819 up_write(&sbi->cp_rwsem); 1820 } 1821 1822 static inline int __get_cp_reason(struct f2fs_sb_info *sbi) 1823 { 1824 int reason = CP_SYNC; 1825 1826 if (test_opt(sbi, FASTBOOT)) 1827 reason = CP_FASTBOOT; 1828 if (is_sbi_flag_set(sbi, SBI_IS_CLOSE)) 1829 reason = CP_UMOUNT; 1830 return reason; 1831 } 1832 1833 static inline bool __remain_node_summaries(int reason) 1834 { 1835 return (reason & (CP_UMOUNT | CP_FASTBOOT)); 1836 } 1837 1838 static inline bool __exist_node_summaries(struct f2fs_sb_info *sbi) 1839 { 1840 return (is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG) || 1841 is_set_ckpt_flags(sbi, CP_FASTBOOT_FLAG)); 1842 } 1843 1844 /* 1845 * Check whether the inode has blocks or not 1846 */ 1847 static inline int F2FS_HAS_BLOCKS(struct inode *inode) 1848 { 1849 block_t xattr_block = F2FS_I(inode)->i_xattr_nid ? 1 : 0; 1850 1851 return (inode->i_blocks >> F2FS_LOG_SECTORS_PER_BLOCK) > xattr_block; 1852 } 1853 1854 static inline bool f2fs_has_xattr_block(unsigned int ofs) 1855 { 1856 return ofs == XATTR_NODE_OFFSET; 1857 } 1858 1859 static inline bool __allow_reserved_blocks(struct f2fs_sb_info *sbi, 1860 struct inode *inode, bool cap) 1861 { 1862 if (!inode) 1863 return true; 1864 if (!test_opt(sbi, RESERVE_ROOT)) 1865 return false; 1866 if (IS_NOQUOTA(inode)) 1867 return true; 1868 if (uid_eq(F2FS_OPTION(sbi).s_resuid, current_fsuid())) 1869 return true; 1870 if (!gid_eq(F2FS_OPTION(sbi).s_resgid, GLOBAL_ROOT_GID) && 1871 in_group_p(F2FS_OPTION(sbi).s_resgid)) 1872 return true; 1873 if (cap && capable(CAP_SYS_RESOURCE)) 1874 return true; 1875 return false; 1876 } 1877 1878 static inline void f2fs_i_blocks_write(struct inode *, block_t, bool, bool); 1879 static inline int inc_valid_block_count(struct f2fs_sb_info *sbi, 1880 struct inode *inode, blkcnt_t *count) 1881 { 1882 blkcnt_t diff = 0, release = 0; 1883 block_t avail_user_block_count; 1884 int ret; 1885 1886 ret = dquot_reserve_block(inode, *count); 1887 if (ret) 1888 return ret; 1889 1890 if (time_to_inject(sbi, FAULT_BLOCK)) { 1891 f2fs_show_injection_info(sbi, FAULT_BLOCK); 1892 release = *count; 1893 goto release_quota; 1894 } 1895 1896 /* 1897 * let's increase this in prior to actual block count change in order 1898 * for f2fs_sync_file to avoid data races when deciding checkpoint. 1899 */ 1900 percpu_counter_add(&sbi->alloc_valid_block_count, (*count)); 1901 1902 spin_lock(&sbi->stat_lock); 1903 sbi->total_valid_block_count += (block_t)(*count); 1904 avail_user_block_count = sbi->user_block_count - 1905 sbi->current_reserved_blocks; 1906 1907 if (!__allow_reserved_blocks(sbi, inode, true)) 1908 avail_user_block_count -= F2FS_OPTION(sbi).root_reserved_blocks; 1909 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) { 1910 if (avail_user_block_count > sbi->unusable_block_count) 1911 avail_user_block_count -= sbi->unusable_block_count; 1912 else 1913 avail_user_block_count = 0; 1914 } 1915 if (unlikely(sbi->total_valid_block_count > avail_user_block_count)) { 1916 diff = sbi->total_valid_block_count - avail_user_block_count; 1917 if (diff > *count) 1918 diff = *count; 1919 *count -= diff; 1920 release = diff; 1921 sbi->total_valid_block_count -= diff; 1922 if (!*count) { 1923 spin_unlock(&sbi->stat_lock); 1924 goto enospc; 1925 } 1926 } 1927 spin_unlock(&sbi->stat_lock); 1928 1929 if (unlikely(release)) { 1930 percpu_counter_sub(&sbi->alloc_valid_block_count, release); 1931 dquot_release_reservation_block(inode, release); 1932 } 1933 f2fs_i_blocks_write(inode, *count, true, true); 1934 return 0; 1935 1936 enospc: 1937 percpu_counter_sub(&sbi->alloc_valid_block_count, release); 1938 release_quota: 1939 dquot_release_reservation_block(inode, release); 1940 return -ENOSPC; 1941 } 1942 1943 __printf(2, 3) 1944 void f2fs_printk(struct f2fs_sb_info *sbi, const char *fmt, ...); 1945 1946 #define f2fs_err(sbi, fmt, ...) \ 1947 f2fs_printk(sbi, KERN_ERR fmt, ##__VA_ARGS__) 1948 #define f2fs_warn(sbi, fmt, ...) \ 1949 f2fs_printk(sbi, KERN_WARNING fmt, ##__VA_ARGS__) 1950 #define f2fs_notice(sbi, fmt, ...) \ 1951 f2fs_printk(sbi, KERN_NOTICE fmt, ##__VA_ARGS__) 1952 #define f2fs_info(sbi, fmt, ...) \ 1953 f2fs_printk(sbi, KERN_INFO fmt, ##__VA_ARGS__) 1954 #define f2fs_debug(sbi, fmt, ...) \ 1955 f2fs_printk(sbi, KERN_DEBUG fmt, ##__VA_ARGS__) 1956 1957 static inline void dec_valid_block_count(struct f2fs_sb_info *sbi, 1958 struct inode *inode, 1959 block_t count) 1960 { 1961 blkcnt_t sectors = count << F2FS_LOG_SECTORS_PER_BLOCK; 1962 1963 spin_lock(&sbi->stat_lock); 1964 f2fs_bug_on(sbi, sbi->total_valid_block_count < (block_t) count); 1965 sbi->total_valid_block_count -= (block_t)count; 1966 if (sbi->reserved_blocks && 1967 sbi->current_reserved_blocks < sbi->reserved_blocks) 1968 sbi->current_reserved_blocks = min(sbi->reserved_blocks, 1969 sbi->current_reserved_blocks + count); 1970 spin_unlock(&sbi->stat_lock); 1971 if (unlikely(inode->i_blocks < sectors)) { 1972 f2fs_warn(sbi, "Inconsistent i_blocks, ino:%lu, iblocks:%llu, sectors:%llu", 1973 inode->i_ino, 1974 (unsigned long long)inode->i_blocks, 1975 (unsigned long long)sectors); 1976 set_sbi_flag(sbi, SBI_NEED_FSCK); 1977 return; 1978 } 1979 f2fs_i_blocks_write(inode, count, false, true); 1980 } 1981 1982 static inline void inc_page_count(struct f2fs_sb_info *sbi, int count_type) 1983 { 1984 atomic_inc(&sbi->nr_pages[count_type]); 1985 1986 if (count_type == F2FS_DIRTY_DENTS || 1987 count_type == F2FS_DIRTY_NODES || 1988 count_type == F2FS_DIRTY_META || 1989 count_type == F2FS_DIRTY_QDATA || 1990 count_type == F2FS_DIRTY_IMETA) 1991 set_sbi_flag(sbi, SBI_IS_DIRTY); 1992 } 1993 1994 static inline void inode_inc_dirty_pages(struct inode *inode) 1995 { 1996 atomic_inc(&F2FS_I(inode)->dirty_pages); 1997 inc_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ? 1998 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA); 1999 if (IS_NOQUOTA(inode)) 2000 inc_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA); 2001 } 2002 2003 static inline void dec_page_count(struct f2fs_sb_info *sbi, int count_type) 2004 { 2005 atomic_dec(&sbi->nr_pages[count_type]); 2006 } 2007 2008 static inline void inode_dec_dirty_pages(struct inode *inode) 2009 { 2010 if (!S_ISDIR(inode->i_mode) && !S_ISREG(inode->i_mode) && 2011 !S_ISLNK(inode->i_mode)) 2012 return; 2013 2014 atomic_dec(&F2FS_I(inode)->dirty_pages); 2015 dec_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ? 2016 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA); 2017 if (IS_NOQUOTA(inode)) 2018 dec_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA); 2019 } 2020 2021 static inline s64 get_pages(struct f2fs_sb_info *sbi, int count_type) 2022 { 2023 return atomic_read(&sbi->nr_pages[count_type]); 2024 } 2025 2026 static inline int get_dirty_pages(struct inode *inode) 2027 { 2028 return atomic_read(&F2FS_I(inode)->dirty_pages); 2029 } 2030 2031 static inline int get_blocktype_secs(struct f2fs_sb_info *sbi, int block_type) 2032 { 2033 unsigned int pages_per_sec = sbi->segs_per_sec * sbi->blocks_per_seg; 2034 unsigned int segs = (get_pages(sbi, block_type) + pages_per_sec - 1) >> 2035 sbi->log_blocks_per_seg; 2036 2037 return segs / sbi->segs_per_sec; 2038 } 2039 2040 static inline block_t valid_user_blocks(struct f2fs_sb_info *sbi) 2041 { 2042 return sbi->total_valid_block_count; 2043 } 2044 2045 static inline block_t discard_blocks(struct f2fs_sb_info *sbi) 2046 { 2047 return sbi->discard_blks; 2048 } 2049 2050 static inline unsigned long __bitmap_size(struct f2fs_sb_info *sbi, int flag) 2051 { 2052 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi); 2053 2054 /* return NAT or SIT bitmap */ 2055 if (flag == NAT_BITMAP) 2056 return le32_to_cpu(ckpt->nat_ver_bitmap_bytesize); 2057 else if (flag == SIT_BITMAP) 2058 return le32_to_cpu(ckpt->sit_ver_bitmap_bytesize); 2059 2060 return 0; 2061 } 2062 2063 static inline block_t __cp_payload(struct f2fs_sb_info *sbi) 2064 { 2065 return le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_payload); 2066 } 2067 2068 static inline void *__bitmap_ptr(struct f2fs_sb_info *sbi, int flag) 2069 { 2070 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi); 2071 int offset; 2072 2073 if (is_set_ckpt_flags(sbi, CP_LARGE_NAT_BITMAP_FLAG)) { 2074 offset = (flag == SIT_BITMAP) ? 2075 le32_to_cpu(ckpt->nat_ver_bitmap_bytesize) : 0; 2076 /* 2077 * if large_nat_bitmap feature is enabled, leave checksum 2078 * protection for all nat/sit bitmaps. 2079 */ 2080 return &ckpt->sit_nat_version_bitmap + offset + sizeof(__le32); 2081 } 2082 2083 if (__cp_payload(sbi) > 0) { 2084 if (flag == NAT_BITMAP) 2085 return &ckpt->sit_nat_version_bitmap; 2086 else 2087 return (unsigned char *)ckpt + F2FS_BLKSIZE; 2088 } else { 2089 offset = (flag == NAT_BITMAP) ? 2090 le32_to_cpu(ckpt->sit_ver_bitmap_bytesize) : 0; 2091 return &ckpt->sit_nat_version_bitmap + offset; 2092 } 2093 } 2094 2095 static inline block_t __start_cp_addr(struct f2fs_sb_info *sbi) 2096 { 2097 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr); 2098 2099 if (sbi->cur_cp_pack == 2) 2100 start_addr += sbi->blocks_per_seg; 2101 return start_addr; 2102 } 2103 2104 static inline block_t __start_cp_next_addr(struct f2fs_sb_info *sbi) 2105 { 2106 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr); 2107 2108 if (sbi->cur_cp_pack == 1) 2109 start_addr += sbi->blocks_per_seg; 2110 return start_addr; 2111 } 2112 2113 static inline void __set_cp_next_pack(struct f2fs_sb_info *sbi) 2114 { 2115 sbi->cur_cp_pack = (sbi->cur_cp_pack == 1) ? 2 : 1; 2116 } 2117 2118 static inline block_t __start_sum_addr(struct f2fs_sb_info *sbi) 2119 { 2120 return le32_to_cpu(F2FS_CKPT(sbi)->cp_pack_start_sum); 2121 } 2122 2123 static inline int inc_valid_node_count(struct f2fs_sb_info *sbi, 2124 struct inode *inode, bool is_inode) 2125 { 2126 block_t valid_block_count; 2127 unsigned int valid_node_count, user_block_count; 2128 int err; 2129 2130 if (is_inode) { 2131 if (inode) { 2132 err = dquot_alloc_inode(inode); 2133 if (err) 2134 return err; 2135 } 2136 } else { 2137 err = dquot_reserve_block(inode, 1); 2138 if (err) 2139 return err; 2140 } 2141 2142 if (time_to_inject(sbi, FAULT_BLOCK)) { 2143 f2fs_show_injection_info(sbi, FAULT_BLOCK); 2144 goto enospc; 2145 } 2146 2147 spin_lock(&sbi->stat_lock); 2148 2149 valid_block_count = sbi->total_valid_block_count + 2150 sbi->current_reserved_blocks + 1; 2151 2152 if (!__allow_reserved_blocks(sbi, inode, false)) 2153 valid_block_count += F2FS_OPTION(sbi).root_reserved_blocks; 2154 user_block_count = sbi->user_block_count; 2155 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) 2156 user_block_count -= sbi->unusable_block_count; 2157 2158 if (unlikely(valid_block_count > user_block_count)) { 2159 spin_unlock(&sbi->stat_lock); 2160 goto enospc; 2161 } 2162 2163 valid_node_count = sbi->total_valid_node_count + 1; 2164 if (unlikely(valid_node_count > sbi->total_node_count)) { 2165 spin_unlock(&sbi->stat_lock); 2166 goto enospc; 2167 } 2168 2169 sbi->total_valid_node_count++; 2170 sbi->total_valid_block_count++; 2171 spin_unlock(&sbi->stat_lock); 2172 2173 if (inode) { 2174 if (is_inode) 2175 f2fs_mark_inode_dirty_sync(inode, true); 2176 else 2177 f2fs_i_blocks_write(inode, 1, true, true); 2178 } 2179 2180 percpu_counter_inc(&sbi->alloc_valid_block_count); 2181 return 0; 2182 2183 enospc: 2184 if (is_inode) { 2185 if (inode) 2186 dquot_free_inode(inode); 2187 } else { 2188 dquot_release_reservation_block(inode, 1); 2189 } 2190 return -ENOSPC; 2191 } 2192 2193 static inline void dec_valid_node_count(struct f2fs_sb_info *sbi, 2194 struct inode *inode, bool is_inode) 2195 { 2196 spin_lock(&sbi->stat_lock); 2197 2198 f2fs_bug_on(sbi, !sbi->total_valid_block_count); 2199 f2fs_bug_on(sbi, !sbi->total_valid_node_count); 2200 2201 sbi->total_valid_node_count--; 2202 sbi->total_valid_block_count--; 2203 if (sbi->reserved_blocks && 2204 sbi->current_reserved_blocks < sbi->reserved_blocks) 2205 sbi->current_reserved_blocks++; 2206 2207 spin_unlock(&sbi->stat_lock); 2208 2209 if (is_inode) { 2210 dquot_free_inode(inode); 2211 } else { 2212 if (unlikely(inode->i_blocks == 0)) { 2213 f2fs_warn(sbi, "Inconsistent i_blocks, ino:%lu, iblocks:%llu", 2214 inode->i_ino, 2215 (unsigned long long)inode->i_blocks); 2216 set_sbi_flag(sbi, SBI_NEED_FSCK); 2217 return; 2218 } 2219 f2fs_i_blocks_write(inode, 1, false, true); 2220 } 2221 } 2222 2223 static inline unsigned int valid_node_count(struct f2fs_sb_info *sbi) 2224 { 2225 return sbi->total_valid_node_count; 2226 } 2227 2228 static inline void inc_valid_inode_count(struct f2fs_sb_info *sbi) 2229 { 2230 percpu_counter_inc(&sbi->total_valid_inode_count); 2231 } 2232 2233 static inline void dec_valid_inode_count(struct f2fs_sb_info *sbi) 2234 { 2235 percpu_counter_dec(&sbi->total_valid_inode_count); 2236 } 2237 2238 static inline s64 valid_inode_count(struct f2fs_sb_info *sbi) 2239 { 2240 return percpu_counter_sum_positive(&sbi->total_valid_inode_count); 2241 } 2242 2243 static inline struct page *f2fs_grab_cache_page(struct address_space *mapping, 2244 pgoff_t index, bool for_write) 2245 { 2246 struct page *page; 2247 2248 if (IS_ENABLED(CONFIG_F2FS_FAULT_INJECTION)) { 2249 if (!for_write) 2250 page = find_get_page_flags(mapping, index, 2251 FGP_LOCK | FGP_ACCESSED); 2252 else 2253 page = find_lock_page(mapping, index); 2254 if (page) 2255 return page; 2256 2257 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_ALLOC)) { 2258 f2fs_show_injection_info(F2FS_M_SB(mapping), 2259 FAULT_PAGE_ALLOC); 2260 return NULL; 2261 } 2262 } 2263 2264 if (!for_write) 2265 return grab_cache_page(mapping, index); 2266 return grab_cache_page_write_begin(mapping, index, AOP_FLAG_NOFS); 2267 } 2268 2269 static inline struct page *f2fs_pagecache_get_page( 2270 struct address_space *mapping, pgoff_t index, 2271 int fgp_flags, gfp_t gfp_mask) 2272 { 2273 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_GET)) { 2274 f2fs_show_injection_info(F2FS_M_SB(mapping), FAULT_PAGE_GET); 2275 return NULL; 2276 } 2277 2278 return pagecache_get_page(mapping, index, fgp_flags, gfp_mask); 2279 } 2280 2281 static inline void f2fs_copy_page(struct page *src, struct page *dst) 2282 { 2283 char *src_kaddr = kmap(src); 2284 char *dst_kaddr = kmap(dst); 2285 2286 memcpy(dst_kaddr, src_kaddr, PAGE_SIZE); 2287 kunmap(dst); 2288 kunmap(src); 2289 } 2290 2291 static inline void f2fs_put_page(struct page *page, int unlock) 2292 { 2293 if (!page) 2294 return; 2295 2296 if (unlock) { 2297 f2fs_bug_on(F2FS_P_SB(page), !PageLocked(page)); 2298 unlock_page(page); 2299 } 2300 put_page(page); 2301 } 2302 2303 static inline void f2fs_put_dnode(struct dnode_of_data *dn) 2304 { 2305 if (dn->node_page) 2306 f2fs_put_page(dn->node_page, 1); 2307 if (dn->inode_page && dn->node_page != dn->inode_page) 2308 f2fs_put_page(dn->inode_page, 0); 2309 dn->node_page = NULL; 2310 dn->inode_page = NULL; 2311 } 2312 2313 static inline struct kmem_cache *f2fs_kmem_cache_create(const char *name, 2314 size_t size) 2315 { 2316 return kmem_cache_create(name, size, 0, SLAB_RECLAIM_ACCOUNT, NULL); 2317 } 2318 2319 static inline void *f2fs_kmem_cache_alloc(struct kmem_cache *cachep, 2320 gfp_t flags) 2321 { 2322 void *entry; 2323 2324 entry = kmem_cache_alloc(cachep, flags); 2325 if (!entry) 2326 entry = kmem_cache_alloc(cachep, flags | __GFP_NOFAIL); 2327 return entry; 2328 } 2329 2330 static inline bool is_idle(struct f2fs_sb_info *sbi, int type) 2331 { 2332 if (sbi->gc_mode == GC_URGENT) 2333 return true; 2334 2335 if (get_pages(sbi, F2FS_RD_DATA) || get_pages(sbi, F2FS_RD_NODE) || 2336 get_pages(sbi, F2FS_RD_META) || get_pages(sbi, F2FS_WB_DATA) || 2337 get_pages(sbi, F2FS_WB_CP_DATA) || 2338 get_pages(sbi, F2FS_DIO_READ) || 2339 get_pages(sbi, F2FS_DIO_WRITE)) 2340 return false; 2341 2342 if (type != DISCARD_TIME && SM_I(sbi) && SM_I(sbi)->dcc_info && 2343 atomic_read(&SM_I(sbi)->dcc_info->queued_discard)) 2344 return false; 2345 2346 if (SM_I(sbi) && SM_I(sbi)->fcc_info && 2347 atomic_read(&SM_I(sbi)->fcc_info->queued_flush)) 2348 return false; 2349 2350 return f2fs_time_over(sbi, type); 2351 } 2352 2353 static inline void f2fs_radix_tree_insert(struct radix_tree_root *root, 2354 unsigned long index, void *item) 2355 { 2356 while (radix_tree_insert(root, index, item)) 2357 cond_resched(); 2358 } 2359 2360 #define RAW_IS_INODE(p) ((p)->footer.nid == (p)->footer.ino) 2361 2362 static inline bool IS_INODE(struct page *page) 2363 { 2364 struct f2fs_node *p = F2FS_NODE(page); 2365 2366 return RAW_IS_INODE(p); 2367 } 2368 2369 static inline int offset_in_addr(struct f2fs_inode *i) 2370 { 2371 return (i->i_inline & F2FS_EXTRA_ATTR) ? 2372 (le16_to_cpu(i->i_extra_isize) / sizeof(__le32)) : 0; 2373 } 2374 2375 static inline __le32 *blkaddr_in_node(struct f2fs_node *node) 2376 { 2377 return RAW_IS_INODE(node) ? node->i.i_addr : node->dn.addr; 2378 } 2379 2380 static inline int f2fs_has_extra_attr(struct inode *inode); 2381 static inline block_t datablock_addr(struct inode *inode, 2382 struct page *node_page, unsigned int offset) 2383 { 2384 struct f2fs_node *raw_node; 2385 __le32 *addr_array; 2386 int base = 0; 2387 bool is_inode = IS_INODE(node_page); 2388 2389 raw_node = F2FS_NODE(node_page); 2390 2391 /* from GC path only */ 2392 if (is_inode) { 2393 if (!inode) 2394 base = offset_in_addr(&raw_node->i); 2395 else if (f2fs_has_extra_attr(inode)) 2396 base = get_extra_isize(inode); 2397 } 2398 2399 addr_array = blkaddr_in_node(raw_node); 2400 return le32_to_cpu(addr_array[base + offset]); 2401 } 2402 2403 static inline int f2fs_test_bit(unsigned int nr, char *addr) 2404 { 2405 int mask; 2406 2407 addr += (nr >> 3); 2408 mask = 1 << (7 - (nr & 0x07)); 2409 return mask & *addr; 2410 } 2411 2412 static inline void f2fs_set_bit(unsigned int nr, char *addr) 2413 { 2414 int mask; 2415 2416 addr += (nr >> 3); 2417 mask = 1 << (7 - (nr & 0x07)); 2418 *addr |= mask; 2419 } 2420 2421 static inline void f2fs_clear_bit(unsigned int nr, char *addr) 2422 { 2423 int mask; 2424 2425 addr += (nr >> 3); 2426 mask = 1 << (7 - (nr & 0x07)); 2427 *addr &= ~mask; 2428 } 2429 2430 static inline int f2fs_test_and_set_bit(unsigned int nr, char *addr) 2431 { 2432 int mask; 2433 int ret; 2434 2435 addr += (nr >> 3); 2436 mask = 1 << (7 - (nr & 0x07)); 2437 ret = mask & *addr; 2438 *addr |= mask; 2439 return ret; 2440 } 2441 2442 static inline int f2fs_test_and_clear_bit(unsigned int nr, char *addr) 2443 { 2444 int mask; 2445 int ret; 2446 2447 addr += (nr >> 3); 2448 mask = 1 << (7 - (nr & 0x07)); 2449 ret = mask & *addr; 2450 *addr &= ~mask; 2451 return ret; 2452 } 2453 2454 static inline void f2fs_change_bit(unsigned int nr, char *addr) 2455 { 2456 int mask; 2457 2458 addr += (nr >> 3); 2459 mask = 1 << (7 - (nr & 0x07)); 2460 *addr ^= mask; 2461 } 2462 2463 /* 2464 * On-disk inode flags (f2fs_inode::i_flags) 2465 */ 2466 #define F2FS_COMPR_FL 0x00000004 /* Compress file */ 2467 #define F2FS_SYNC_FL 0x00000008 /* Synchronous updates */ 2468 #define F2FS_IMMUTABLE_FL 0x00000010 /* Immutable file */ 2469 #define F2FS_APPEND_FL 0x00000020 /* writes to file may only append */ 2470 #define F2FS_NODUMP_FL 0x00000040 /* do not dump file */ 2471 #define F2FS_NOATIME_FL 0x00000080 /* do not update atime */ 2472 #define F2FS_NOCOMP_FL 0x00000400 /* Don't compress */ 2473 #define F2FS_INDEX_FL 0x00001000 /* hash-indexed directory */ 2474 #define F2FS_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */ 2475 #define F2FS_PROJINHERIT_FL 0x20000000 /* Create with parents projid */ 2476 #define F2FS_CASEFOLD_FL 0x40000000 /* Casefolded file */ 2477 2478 /* Flags that should be inherited by new inodes from their parent. */ 2479 #define F2FS_FL_INHERITED (F2FS_SYNC_FL | F2FS_NODUMP_FL | F2FS_NOATIME_FL | \ 2480 F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \ 2481 F2FS_CASEFOLD_FL | F2FS_COMPR_FL | F2FS_NOCOMP_FL) 2482 2483 /* Flags that are appropriate for regular files (all but dir-specific ones). */ 2484 #define F2FS_REG_FLMASK (~(F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \ 2485 F2FS_CASEFOLD_FL)) 2486 2487 /* Flags that are appropriate for non-directories/regular files. */ 2488 #define F2FS_OTHER_FLMASK (F2FS_NODUMP_FL | F2FS_NOATIME_FL) 2489 2490 static inline __u32 f2fs_mask_flags(umode_t mode, __u32 flags) 2491 { 2492 if (S_ISDIR(mode)) 2493 return flags; 2494 else if (S_ISREG(mode)) 2495 return flags & F2FS_REG_FLMASK; 2496 else 2497 return flags & F2FS_OTHER_FLMASK; 2498 } 2499 2500 /* used for f2fs_inode_info->flags */ 2501 enum { 2502 FI_NEW_INODE, /* indicate newly allocated inode */ 2503 FI_DIRTY_INODE, /* indicate inode is dirty or not */ 2504 FI_AUTO_RECOVER, /* indicate inode is recoverable */ 2505 FI_DIRTY_DIR, /* indicate directory has dirty pages */ 2506 FI_INC_LINK, /* need to increment i_nlink */ 2507 FI_ACL_MODE, /* indicate acl mode */ 2508 FI_NO_ALLOC, /* should not allocate any blocks */ 2509 FI_FREE_NID, /* free allocated nide */ 2510 FI_NO_EXTENT, /* not to use the extent cache */ 2511 FI_INLINE_XATTR, /* used for inline xattr */ 2512 FI_INLINE_DATA, /* used for inline data*/ 2513 FI_INLINE_DENTRY, /* used for inline dentry */ 2514 FI_APPEND_WRITE, /* inode has appended data */ 2515 FI_UPDATE_WRITE, /* inode has in-place-update data */ 2516 FI_NEED_IPU, /* used for ipu per file */ 2517 FI_ATOMIC_FILE, /* indicate atomic file */ 2518 FI_ATOMIC_COMMIT, /* indicate the state of atomical committing */ 2519 FI_VOLATILE_FILE, /* indicate volatile file */ 2520 FI_FIRST_BLOCK_WRITTEN, /* indicate #0 data block was written */ 2521 FI_DROP_CACHE, /* drop dirty page cache */ 2522 FI_DATA_EXIST, /* indicate data exists */ 2523 FI_INLINE_DOTS, /* indicate inline dot dentries */ 2524 FI_DO_DEFRAG, /* indicate defragment is running */ 2525 FI_DIRTY_FILE, /* indicate regular/symlink has dirty pages */ 2526 FI_NO_PREALLOC, /* indicate skipped preallocated blocks */ 2527 FI_HOT_DATA, /* indicate file is hot */ 2528 FI_EXTRA_ATTR, /* indicate file has extra attribute */ 2529 FI_PROJ_INHERIT, /* indicate file inherits projectid */ 2530 FI_PIN_FILE, /* indicate file should not be gced */ 2531 FI_ATOMIC_REVOKE_REQUEST, /* request to drop atomic data */ 2532 FI_VERITY_IN_PROGRESS, /* building fs-verity Merkle tree */ 2533 FI_COMPRESSED_FILE, /* indicate file's data can be compressed */ 2534 FI_MMAP_FILE, /* indicate file was mmapped */ 2535 }; 2536 2537 static inline void __mark_inode_dirty_flag(struct inode *inode, 2538 int flag, bool set) 2539 { 2540 switch (flag) { 2541 case FI_INLINE_XATTR: 2542 case FI_INLINE_DATA: 2543 case FI_INLINE_DENTRY: 2544 case FI_NEW_INODE: 2545 if (set) 2546 return; 2547 /* fall through */ 2548 case FI_DATA_EXIST: 2549 case FI_INLINE_DOTS: 2550 case FI_PIN_FILE: 2551 case FI_COMPRESSED_FILE: 2552 f2fs_mark_inode_dirty_sync(inode, true); 2553 } 2554 } 2555 2556 static inline void set_inode_flag(struct inode *inode, int flag) 2557 { 2558 if (!test_bit(flag, &F2FS_I(inode)->flags)) 2559 set_bit(flag, &F2FS_I(inode)->flags); 2560 __mark_inode_dirty_flag(inode, flag, true); 2561 } 2562 2563 static inline int is_inode_flag_set(struct inode *inode, int flag) 2564 { 2565 return test_bit(flag, &F2FS_I(inode)->flags); 2566 } 2567 2568 static inline void clear_inode_flag(struct inode *inode, int flag) 2569 { 2570 if (test_bit(flag, &F2FS_I(inode)->flags)) 2571 clear_bit(flag, &F2FS_I(inode)->flags); 2572 __mark_inode_dirty_flag(inode, flag, false); 2573 } 2574 2575 static inline bool f2fs_verity_in_progress(struct inode *inode) 2576 { 2577 return IS_ENABLED(CONFIG_FS_VERITY) && 2578 is_inode_flag_set(inode, FI_VERITY_IN_PROGRESS); 2579 } 2580 2581 static inline void set_acl_inode(struct inode *inode, umode_t mode) 2582 { 2583 F2FS_I(inode)->i_acl_mode = mode; 2584 set_inode_flag(inode, FI_ACL_MODE); 2585 f2fs_mark_inode_dirty_sync(inode, false); 2586 } 2587 2588 static inline void f2fs_i_links_write(struct inode *inode, bool inc) 2589 { 2590 if (inc) 2591 inc_nlink(inode); 2592 else 2593 drop_nlink(inode); 2594 f2fs_mark_inode_dirty_sync(inode, true); 2595 } 2596 2597 static inline void f2fs_i_blocks_write(struct inode *inode, 2598 block_t diff, bool add, bool claim) 2599 { 2600 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE); 2601 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER); 2602 2603 /* add = 1, claim = 1 should be dquot_reserve_block in pair */ 2604 if (add) { 2605 if (claim) 2606 dquot_claim_block(inode, diff); 2607 else 2608 dquot_alloc_block_nofail(inode, diff); 2609 } else { 2610 dquot_free_block(inode, diff); 2611 } 2612 2613 f2fs_mark_inode_dirty_sync(inode, true); 2614 if (clean || recover) 2615 set_inode_flag(inode, FI_AUTO_RECOVER); 2616 } 2617 2618 static inline void f2fs_i_size_write(struct inode *inode, loff_t i_size) 2619 { 2620 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE); 2621 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER); 2622 2623 if (i_size_read(inode) == i_size) 2624 return; 2625 2626 i_size_write(inode, i_size); 2627 f2fs_mark_inode_dirty_sync(inode, true); 2628 if (clean || recover) 2629 set_inode_flag(inode, FI_AUTO_RECOVER); 2630 } 2631 2632 static inline void f2fs_i_depth_write(struct inode *inode, unsigned int depth) 2633 { 2634 F2FS_I(inode)->i_current_depth = depth; 2635 f2fs_mark_inode_dirty_sync(inode, true); 2636 } 2637 2638 static inline void f2fs_i_gc_failures_write(struct inode *inode, 2639 unsigned int count) 2640 { 2641 F2FS_I(inode)->i_gc_failures[GC_FAILURE_PIN] = count; 2642 f2fs_mark_inode_dirty_sync(inode, true); 2643 } 2644 2645 static inline void f2fs_i_xnid_write(struct inode *inode, nid_t xnid) 2646 { 2647 F2FS_I(inode)->i_xattr_nid = xnid; 2648 f2fs_mark_inode_dirty_sync(inode, true); 2649 } 2650 2651 static inline void f2fs_i_pino_write(struct inode *inode, nid_t pino) 2652 { 2653 F2FS_I(inode)->i_pino = pino; 2654 f2fs_mark_inode_dirty_sync(inode, true); 2655 } 2656 2657 static inline void get_inline_info(struct inode *inode, struct f2fs_inode *ri) 2658 { 2659 struct f2fs_inode_info *fi = F2FS_I(inode); 2660 2661 if (ri->i_inline & F2FS_INLINE_XATTR) 2662 set_bit(FI_INLINE_XATTR, &fi->flags); 2663 if (ri->i_inline & F2FS_INLINE_DATA) 2664 set_bit(FI_INLINE_DATA, &fi->flags); 2665 if (ri->i_inline & F2FS_INLINE_DENTRY) 2666 set_bit(FI_INLINE_DENTRY, &fi->flags); 2667 if (ri->i_inline & F2FS_DATA_EXIST) 2668 set_bit(FI_DATA_EXIST, &fi->flags); 2669 if (ri->i_inline & F2FS_INLINE_DOTS) 2670 set_bit(FI_INLINE_DOTS, &fi->flags); 2671 if (ri->i_inline & F2FS_EXTRA_ATTR) 2672 set_bit(FI_EXTRA_ATTR, &fi->flags); 2673 if (ri->i_inline & F2FS_PIN_FILE) 2674 set_bit(FI_PIN_FILE, &fi->flags); 2675 } 2676 2677 static inline void set_raw_inline(struct inode *inode, struct f2fs_inode *ri) 2678 { 2679 ri->i_inline = 0; 2680 2681 if (is_inode_flag_set(inode, FI_INLINE_XATTR)) 2682 ri->i_inline |= F2FS_INLINE_XATTR; 2683 if (is_inode_flag_set(inode, FI_INLINE_DATA)) 2684 ri->i_inline |= F2FS_INLINE_DATA; 2685 if (is_inode_flag_set(inode, FI_INLINE_DENTRY)) 2686 ri->i_inline |= F2FS_INLINE_DENTRY; 2687 if (is_inode_flag_set(inode, FI_DATA_EXIST)) 2688 ri->i_inline |= F2FS_DATA_EXIST; 2689 if (is_inode_flag_set(inode, FI_INLINE_DOTS)) 2690 ri->i_inline |= F2FS_INLINE_DOTS; 2691 if (is_inode_flag_set(inode, FI_EXTRA_ATTR)) 2692 ri->i_inline |= F2FS_EXTRA_ATTR; 2693 if (is_inode_flag_set(inode, FI_PIN_FILE)) 2694 ri->i_inline |= F2FS_PIN_FILE; 2695 } 2696 2697 static inline int f2fs_has_extra_attr(struct inode *inode) 2698 { 2699 return is_inode_flag_set(inode, FI_EXTRA_ATTR); 2700 } 2701 2702 static inline int f2fs_has_inline_xattr(struct inode *inode) 2703 { 2704 return is_inode_flag_set(inode, FI_INLINE_XATTR); 2705 } 2706 2707 static inline int f2fs_compressed_file(struct inode *inode) 2708 { 2709 return S_ISREG(inode->i_mode) && 2710 is_inode_flag_set(inode, FI_COMPRESSED_FILE); 2711 } 2712 2713 static inline unsigned int addrs_per_inode(struct inode *inode) 2714 { 2715 unsigned int addrs = CUR_ADDRS_PER_INODE(inode) - 2716 get_inline_xattr_addrs(inode); 2717 2718 if (!f2fs_compressed_file(inode)) 2719 return addrs; 2720 return ALIGN_DOWN(addrs, F2FS_I(inode)->i_cluster_size); 2721 } 2722 2723 static inline unsigned int addrs_per_block(struct inode *inode) 2724 { 2725 if (!f2fs_compressed_file(inode)) 2726 return DEF_ADDRS_PER_BLOCK; 2727 return ALIGN_DOWN(DEF_ADDRS_PER_BLOCK, F2FS_I(inode)->i_cluster_size); 2728 } 2729 2730 static inline void *inline_xattr_addr(struct inode *inode, struct page *page) 2731 { 2732 struct f2fs_inode *ri = F2FS_INODE(page); 2733 2734 return (void *)&(ri->i_addr[DEF_ADDRS_PER_INODE - 2735 get_inline_xattr_addrs(inode)]); 2736 } 2737 2738 static inline int inline_xattr_size(struct inode *inode) 2739 { 2740 if (f2fs_has_inline_xattr(inode)) 2741 return get_inline_xattr_addrs(inode) * sizeof(__le32); 2742 return 0; 2743 } 2744 2745 static inline int f2fs_has_inline_data(struct inode *inode) 2746 { 2747 return is_inode_flag_set(inode, FI_INLINE_DATA); 2748 } 2749 2750 static inline int f2fs_exist_data(struct inode *inode) 2751 { 2752 return is_inode_flag_set(inode, FI_DATA_EXIST); 2753 } 2754 2755 static inline int f2fs_has_inline_dots(struct inode *inode) 2756 { 2757 return is_inode_flag_set(inode, FI_INLINE_DOTS); 2758 } 2759 2760 static inline int f2fs_is_mmap_file(struct inode *inode) 2761 { 2762 return is_inode_flag_set(inode, FI_MMAP_FILE); 2763 } 2764 2765 static inline bool f2fs_is_pinned_file(struct inode *inode) 2766 { 2767 return is_inode_flag_set(inode, FI_PIN_FILE); 2768 } 2769 2770 static inline bool f2fs_is_atomic_file(struct inode *inode) 2771 { 2772 return is_inode_flag_set(inode, FI_ATOMIC_FILE); 2773 } 2774 2775 static inline bool f2fs_is_commit_atomic_write(struct inode *inode) 2776 { 2777 return is_inode_flag_set(inode, FI_ATOMIC_COMMIT); 2778 } 2779 2780 static inline bool f2fs_is_volatile_file(struct inode *inode) 2781 { 2782 return is_inode_flag_set(inode, FI_VOLATILE_FILE); 2783 } 2784 2785 static inline bool f2fs_is_first_block_written(struct inode *inode) 2786 { 2787 return is_inode_flag_set(inode, FI_FIRST_BLOCK_WRITTEN); 2788 } 2789 2790 static inline bool f2fs_is_drop_cache(struct inode *inode) 2791 { 2792 return is_inode_flag_set(inode, FI_DROP_CACHE); 2793 } 2794 2795 static inline void *inline_data_addr(struct inode *inode, struct page *page) 2796 { 2797 struct f2fs_inode *ri = F2FS_INODE(page); 2798 int extra_size = get_extra_isize(inode); 2799 2800 return (void *)&(ri->i_addr[extra_size + DEF_INLINE_RESERVED_SIZE]); 2801 } 2802 2803 static inline int f2fs_has_inline_dentry(struct inode *inode) 2804 { 2805 return is_inode_flag_set(inode, FI_INLINE_DENTRY); 2806 } 2807 2808 static inline int is_file(struct inode *inode, int type) 2809 { 2810 return F2FS_I(inode)->i_advise & type; 2811 } 2812 2813 static inline void set_file(struct inode *inode, int type) 2814 { 2815 F2FS_I(inode)->i_advise |= type; 2816 f2fs_mark_inode_dirty_sync(inode, true); 2817 } 2818 2819 static inline void clear_file(struct inode *inode, int type) 2820 { 2821 F2FS_I(inode)->i_advise &= ~type; 2822 f2fs_mark_inode_dirty_sync(inode, true); 2823 } 2824 2825 static inline bool f2fs_is_time_consistent(struct inode *inode) 2826 { 2827 if (!timespec64_equal(F2FS_I(inode)->i_disk_time, &inode->i_atime)) 2828 return false; 2829 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 1, &inode->i_ctime)) 2830 return false; 2831 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 2, &inode->i_mtime)) 2832 return false; 2833 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 3, 2834 &F2FS_I(inode)->i_crtime)) 2835 return false; 2836 return true; 2837 } 2838 2839 static inline bool f2fs_skip_inode_update(struct inode *inode, int dsync) 2840 { 2841 bool ret; 2842 2843 if (dsync) { 2844 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 2845 2846 spin_lock(&sbi->inode_lock[DIRTY_META]); 2847 ret = list_empty(&F2FS_I(inode)->gdirty_list); 2848 spin_unlock(&sbi->inode_lock[DIRTY_META]); 2849 return ret; 2850 } 2851 if (!is_inode_flag_set(inode, FI_AUTO_RECOVER) || 2852 file_keep_isize(inode) || 2853 i_size_read(inode) & ~PAGE_MASK) 2854 return false; 2855 2856 if (!f2fs_is_time_consistent(inode)) 2857 return false; 2858 2859 down_read(&F2FS_I(inode)->i_sem); 2860 ret = F2FS_I(inode)->last_disk_size == i_size_read(inode); 2861 up_read(&F2FS_I(inode)->i_sem); 2862 2863 return ret; 2864 } 2865 2866 static inline bool f2fs_readonly(struct super_block *sb) 2867 { 2868 return sb_rdonly(sb); 2869 } 2870 2871 static inline bool f2fs_cp_error(struct f2fs_sb_info *sbi) 2872 { 2873 return is_set_ckpt_flags(sbi, CP_ERROR_FLAG); 2874 } 2875 2876 static inline bool is_dot_dotdot(const struct qstr *str) 2877 { 2878 if (str->len == 1 && str->name[0] == '.') 2879 return true; 2880 2881 if (str->len == 2 && str->name[0] == '.' && str->name[1] == '.') 2882 return true; 2883 2884 return false; 2885 } 2886 2887 static inline bool f2fs_may_extent_tree(struct inode *inode) 2888 { 2889 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 2890 2891 if (!test_opt(sbi, EXTENT_CACHE) || 2892 is_inode_flag_set(inode, FI_NO_EXTENT) || 2893 is_inode_flag_set(inode, FI_COMPRESSED_FILE)) 2894 return false; 2895 2896 /* 2897 * for recovered files during mount do not create extents 2898 * if shrinker is not registered. 2899 */ 2900 if (list_empty(&sbi->s_list)) 2901 return false; 2902 2903 return S_ISREG(inode->i_mode); 2904 } 2905 2906 static inline void *f2fs_kmalloc(struct f2fs_sb_info *sbi, 2907 size_t size, gfp_t flags) 2908 { 2909 void *ret; 2910 2911 if (time_to_inject(sbi, FAULT_KMALLOC)) { 2912 f2fs_show_injection_info(sbi, FAULT_KMALLOC); 2913 return NULL; 2914 } 2915 2916 ret = kmalloc(size, flags); 2917 if (ret) 2918 return ret; 2919 2920 return kvmalloc(size, flags); 2921 } 2922 2923 static inline void *f2fs_kzalloc(struct f2fs_sb_info *sbi, 2924 size_t size, gfp_t flags) 2925 { 2926 return f2fs_kmalloc(sbi, size, flags | __GFP_ZERO); 2927 } 2928 2929 static inline void *f2fs_kvmalloc(struct f2fs_sb_info *sbi, 2930 size_t size, gfp_t flags) 2931 { 2932 if (time_to_inject(sbi, FAULT_KVMALLOC)) { 2933 f2fs_show_injection_info(sbi, FAULT_KVMALLOC); 2934 return NULL; 2935 } 2936 2937 return kvmalloc(size, flags); 2938 } 2939 2940 static inline void *f2fs_kvzalloc(struct f2fs_sb_info *sbi, 2941 size_t size, gfp_t flags) 2942 { 2943 return f2fs_kvmalloc(sbi, size, flags | __GFP_ZERO); 2944 } 2945 2946 static inline int get_extra_isize(struct inode *inode) 2947 { 2948 return F2FS_I(inode)->i_extra_isize / sizeof(__le32); 2949 } 2950 2951 static inline int get_inline_xattr_addrs(struct inode *inode) 2952 { 2953 return F2FS_I(inode)->i_inline_xattr_size; 2954 } 2955 2956 #define f2fs_get_inode_mode(i) \ 2957 ((is_inode_flag_set(i, FI_ACL_MODE)) ? \ 2958 (F2FS_I(i)->i_acl_mode) : ((i)->i_mode)) 2959 2960 #define F2FS_TOTAL_EXTRA_ATTR_SIZE \ 2961 (offsetof(struct f2fs_inode, i_extra_end) - \ 2962 offsetof(struct f2fs_inode, i_extra_isize)) \ 2963 2964 #define F2FS_OLD_ATTRIBUTE_SIZE (offsetof(struct f2fs_inode, i_addr)) 2965 #define F2FS_FITS_IN_INODE(f2fs_inode, extra_isize, field) \ 2966 ((offsetof(typeof(*(f2fs_inode)), field) + \ 2967 sizeof((f2fs_inode)->field)) \ 2968 <= (F2FS_OLD_ATTRIBUTE_SIZE + (extra_isize))) \ 2969 2970 static inline void f2fs_reset_iostat(struct f2fs_sb_info *sbi) 2971 { 2972 int i; 2973 2974 spin_lock(&sbi->iostat_lock); 2975 for (i = 0; i < NR_IO_TYPE; i++) 2976 sbi->write_iostat[i] = 0; 2977 spin_unlock(&sbi->iostat_lock); 2978 } 2979 2980 static inline void f2fs_update_iostat(struct f2fs_sb_info *sbi, 2981 enum iostat_type type, unsigned long long io_bytes) 2982 { 2983 if (!sbi->iostat_enable) 2984 return; 2985 spin_lock(&sbi->iostat_lock); 2986 sbi->write_iostat[type] += io_bytes; 2987 2988 if (type == APP_WRITE_IO || type == APP_DIRECT_IO) 2989 sbi->write_iostat[APP_BUFFERED_IO] = 2990 sbi->write_iostat[APP_WRITE_IO] - 2991 sbi->write_iostat[APP_DIRECT_IO]; 2992 spin_unlock(&sbi->iostat_lock); 2993 } 2994 2995 #define __is_large_section(sbi) ((sbi)->segs_per_sec > 1) 2996 2997 #define __is_meta_io(fio) (PAGE_TYPE_OF_BIO((fio)->type) == META) 2998 2999 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi, 3000 block_t blkaddr, int type); 3001 static inline void verify_blkaddr(struct f2fs_sb_info *sbi, 3002 block_t blkaddr, int type) 3003 { 3004 if (!f2fs_is_valid_blkaddr(sbi, blkaddr, type)) { 3005 f2fs_err(sbi, "invalid blkaddr: %u, type: %d, run fsck to fix.", 3006 blkaddr, type); 3007 f2fs_bug_on(sbi, 1); 3008 } 3009 } 3010 3011 static inline bool __is_valid_data_blkaddr(block_t blkaddr) 3012 { 3013 if (blkaddr == NEW_ADDR || blkaddr == NULL_ADDR || 3014 blkaddr == COMPRESS_ADDR) 3015 return false; 3016 return true; 3017 } 3018 3019 static inline void f2fs_set_page_private(struct page *page, 3020 unsigned long data) 3021 { 3022 if (PagePrivate(page)) 3023 return; 3024 3025 get_page(page); 3026 SetPagePrivate(page); 3027 set_page_private(page, data); 3028 } 3029 3030 static inline void f2fs_clear_page_private(struct page *page) 3031 { 3032 if (!PagePrivate(page)) 3033 return; 3034 3035 set_page_private(page, 0); 3036 ClearPagePrivate(page); 3037 f2fs_put_page(page, 0); 3038 } 3039 3040 /* 3041 * file.c 3042 */ 3043 int f2fs_sync_file(struct file *file, loff_t start, loff_t end, int datasync); 3044 void f2fs_truncate_data_blocks(struct dnode_of_data *dn); 3045 int f2fs_truncate_blocks(struct inode *inode, u64 from, bool lock); 3046 int f2fs_truncate(struct inode *inode); 3047 int f2fs_getattr(const struct path *path, struct kstat *stat, 3048 u32 request_mask, unsigned int flags); 3049 int f2fs_setattr(struct dentry *dentry, struct iattr *attr); 3050 int f2fs_truncate_hole(struct inode *inode, pgoff_t pg_start, pgoff_t pg_end); 3051 void f2fs_truncate_data_blocks_range(struct dnode_of_data *dn, int count); 3052 int f2fs_precache_extents(struct inode *inode); 3053 long f2fs_ioctl(struct file *filp, unsigned int cmd, unsigned long arg); 3054 long f2fs_compat_ioctl(struct file *file, unsigned int cmd, unsigned long arg); 3055 int f2fs_transfer_project_quota(struct inode *inode, kprojid_t kprojid); 3056 int f2fs_pin_file_control(struct inode *inode, bool inc); 3057 3058 /* 3059 * inode.c 3060 */ 3061 void f2fs_set_inode_flags(struct inode *inode); 3062 bool f2fs_inode_chksum_verify(struct f2fs_sb_info *sbi, struct page *page); 3063 void f2fs_inode_chksum_set(struct f2fs_sb_info *sbi, struct page *page); 3064 struct inode *f2fs_iget(struct super_block *sb, unsigned long ino); 3065 struct inode *f2fs_iget_retry(struct super_block *sb, unsigned long ino); 3066 int f2fs_try_to_free_nats(struct f2fs_sb_info *sbi, int nr_shrink); 3067 void f2fs_update_inode(struct inode *inode, struct page *node_page); 3068 void f2fs_update_inode_page(struct inode *inode); 3069 int f2fs_write_inode(struct inode *inode, struct writeback_control *wbc); 3070 void f2fs_evict_inode(struct inode *inode); 3071 void f2fs_handle_failed_inode(struct inode *inode); 3072 3073 /* 3074 * namei.c 3075 */ 3076 int f2fs_update_extension_list(struct f2fs_sb_info *sbi, const char *name, 3077 bool hot, bool set); 3078 struct dentry *f2fs_get_parent(struct dentry *child); 3079 3080 extern int f2fs_ci_compare(const struct inode *parent, 3081 const struct qstr *name, 3082 const struct qstr *entry, 3083 bool quick); 3084 3085 /* 3086 * dir.c 3087 */ 3088 unsigned char f2fs_get_de_type(struct f2fs_dir_entry *de); 3089 struct f2fs_dir_entry *f2fs_find_target_dentry(struct fscrypt_name *fname, 3090 f2fs_hash_t namehash, int *max_slots, 3091 struct f2fs_dentry_ptr *d); 3092 int f2fs_fill_dentries(struct dir_context *ctx, struct f2fs_dentry_ptr *d, 3093 unsigned int start_pos, struct fscrypt_str *fstr); 3094 void f2fs_do_make_empty_dir(struct inode *inode, struct inode *parent, 3095 struct f2fs_dentry_ptr *d); 3096 struct page *f2fs_init_inode_metadata(struct inode *inode, struct inode *dir, 3097 const struct qstr *new_name, 3098 const struct qstr *orig_name, struct page *dpage); 3099 void f2fs_update_parent_metadata(struct inode *dir, struct inode *inode, 3100 unsigned int current_depth); 3101 int f2fs_room_for_filename(const void *bitmap, int slots, int max_slots); 3102 void f2fs_drop_nlink(struct inode *dir, struct inode *inode); 3103 struct f2fs_dir_entry *__f2fs_find_entry(struct inode *dir, 3104 struct fscrypt_name *fname, struct page **res_page); 3105 struct f2fs_dir_entry *f2fs_find_entry(struct inode *dir, 3106 const struct qstr *child, struct page **res_page); 3107 struct f2fs_dir_entry *f2fs_parent_dir(struct inode *dir, struct page **p); 3108 ino_t f2fs_inode_by_name(struct inode *dir, const struct qstr *qstr, 3109 struct page **page); 3110 void f2fs_set_link(struct inode *dir, struct f2fs_dir_entry *de, 3111 struct page *page, struct inode *inode); 3112 bool f2fs_has_enough_room(struct inode *dir, struct page *ipage, 3113 struct fscrypt_name *fname); 3114 void f2fs_update_dentry(nid_t ino, umode_t mode, struct f2fs_dentry_ptr *d, 3115 const struct qstr *name, f2fs_hash_t name_hash, 3116 unsigned int bit_pos); 3117 int f2fs_add_regular_entry(struct inode *dir, const struct qstr *new_name, 3118 const struct qstr *orig_name, 3119 struct inode *inode, nid_t ino, umode_t mode); 3120 int f2fs_add_dentry(struct inode *dir, struct fscrypt_name *fname, 3121 struct inode *inode, nid_t ino, umode_t mode); 3122 int f2fs_do_add_link(struct inode *dir, const struct qstr *name, 3123 struct inode *inode, nid_t ino, umode_t mode); 3124 void f2fs_delete_entry(struct f2fs_dir_entry *dentry, struct page *page, 3125 struct inode *dir, struct inode *inode); 3126 int f2fs_do_tmpfile(struct inode *inode, struct inode *dir); 3127 bool f2fs_empty_dir(struct inode *dir); 3128 3129 static inline int f2fs_add_link(struct dentry *dentry, struct inode *inode) 3130 { 3131 return f2fs_do_add_link(d_inode(dentry->d_parent), &dentry->d_name, 3132 inode, inode->i_ino, inode->i_mode); 3133 } 3134 3135 /* 3136 * super.c 3137 */ 3138 int f2fs_inode_dirtied(struct inode *inode, bool sync); 3139 void f2fs_inode_synced(struct inode *inode); 3140 int f2fs_enable_quota_files(struct f2fs_sb_info *sbi, bool rdonly); 3141 int f2fs_quota_sync(struct super_block *sb, int type); 3142 void f2fs_quota_off_umount(struct super_block *sb); 3143 int f2fs_commit_super(struct f2fs_sb_info *sbi, bool recover); 3144 int f2fs_sync_fs(struct super_block *sb, int sync); 3145 int f2fs_sanity_check_ckpt(struct f2fs_sb_info *sbi); 3146 3147 /* 3148 * hash.c 3149 */ 3150 f2fs_hash_t f2fs_dentry_hash(const struct inode *dir, 3151 const struct qstr *name_info, struct fscrypt_name *fname); 3152 3153 /* 3154 * node.c 3155 */ 3156 struct dnode_of_data; 3157 struct node_info; 3158 3159 int f2fs_check_nid_range(struct f2fs_sb_info *sbi, nid_t nid); 3160 bool f2fs_available_free_memory(struct f2fs_sb_info *sbi, int type); 3161 bool f2fs_in_warm_node_list(struct f2fs_sb_info *sbi, struct page *page); 3162 void f2fs_init_fsync_node_info(struct f2fs_sb_info *sbi); 3163 void f2fs_del_fsync_node_entry(struct f2fs_sb_info *sbi, struct page *page); 3164 void f2fs_reset_fsync_node_info(struct f2fs_sb_info *sbi); 3165 int f2fs_need_dentry_mark(struct f2fs_sb_info *sbi, nid_t nid); 3166 bool f2fs_is_checkpointed_node(struct f2fs_sb_info *sbi, nid_t nid); 3167 bool f2fs_need_inode_block_update(struct f2fs_sb_info *sbi, nid_t ino); 3168 int f2fs_get_node_info(struct f2fs_sb_info *sbi, nid_t nid, 3169 struct node_info *ni); 3170 pgoff_t f2fs_get_next_page_offset(struct dnode_of_data *dn, pgoff_t pgofs); 3171 int f2fs_get_dnode_of_data(struct dnode_of_data *dn, pgoff_t index, int mode); 3172 int f2fs_truncate_inode_blocks(struct inode *inode, pgoff_t from); 3173 int f2fs_truncate_xattr_node(struct inode *inode); 3174 int f2fs_wait_on_node_pages_writeback(struct f2fs_sb_info *sbi, 3175 unsigned int seq_id); 3176 int f2fs_remove_inode_page(struct inode *inode); 3177 struct page *f2fs_new_inode_page(struct inode *inode); 3178 struct page *f2fs_new_node_page(struct dnode_of_data *dn, unsigned int ofs); 3179 void f2fs_ra_node_page(struct f2fs_sb_info *sbi, nid_t nid); 3180 struct page *f2fs_get_node_page(struct f2fs_sb_info *sbi, pgoff_t nid); 3181 struct page *f2fs_get_node_page_ra(struct page *parent, int start); 3182 int f2fs_move_node_page(struct page *node_page, int gc_type); 3183 int f2fs_fsync_node_pages(struct f2fs_sb_info *sbi, struct inode *inode, 3184 struct writeback_control *wbc, bool atomic, 3185 unsigned int *seq_id); 3186 int f2fs_sync_node_pages(struct f2fs_sb_info *sbi, 3187 struct writeback_control *wbc, 3188 bool do_balance, enum iostat_type io_type); 3189 int f2fs_build_free_nids(struct f2fs_sb_info *sbi, bool sync, bool mount); 3190 bool f2fs_alloc_nid(struct f2fs_sb_info *sbi, nid_t *nid); 3191 void f2fs_alloc_nid_done(struct f2fs_sb_info *sbi, nid_t nid); 3192 void f2fs_alloc_nid_failed(struct f2fs_sb_info *sbi, nid_t nid); 3193 int f2fs_try_to_free_nids(struct f2fs_sb_info *sbi, int nr_shrink); 3194 void f2fs_recover_inline_xattr(struct inode *inode, struct page *page); 3195 int f2fs_recover_xattr_data(struct inode *inode, struct page *page); 3196 int f2fs_recover_inode_page(struct f2fs_sb_info *sbi, struct page *page); 3197 int f2fs_restore_node_summary(struct f2fs_sb_info *sbi, 3198 unsigned int segno, struct f2fs_summary_block *sum); 3199 int f2fs_flush_nat_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3200 int f2fs_build_node_manager(struct f2fs_sb_info *sbi); 3201 void f2fs_destroy_node_manager(struct f2fs_sb_info *sbi); 3202 int __init f2fs_create_node_manager_caches(void); 3203 void f2fs_destroy_node_manager_caches(void); 3204 3205 /* 3206 * segment.c 3207 */ 3208 bool f2fs_need_SSR(struct f2fs_sb_info *sbi); 3209 void f2fs_register_inmem_page(struct inode *inode, struct page *page); 3210 void f2fs_drop_inmem_pages_all(struct f2fs_sb_info *sbi, bool gc_failure); 3211 void f2fs_drop_inmem_pages(struct inode *inode); 3212 void f2fs_drop_inmem_page(struct inode *inode, struct page *page); 3213 int f2fs_commit_inmem_pages(struct inode *inode); 3214 void f2fs_balance_fs(struct f2fs_sb_info *sbi, bool need); 3215 void f2fs_balance_fs_bg(struct f2fs_sb_info *sbi); 3216 int f2fs_issue_flush(struct f2fs_sb_info *sbi, nid_t ino); 3217 int f2fs_create_flush_cmd_control(struct f2fs_sb_info *sbi); 3218 int f2fs_flush_device_cache(struct f2fs_sb_info *sbi); 3219 void f2fs_destroy_flush_cmd_control(struct f2fs_sb_info *sbi, bool free); 3220 void f2fs_invalidate_blocks(struct f2fs_sb_info *sbi, block_t addr); 3221 bool f2fs_is_checkpointed_data(struct f2fs_sb_info *sbi, block_t blkaddr); 3222 void f2fs_drop_discard_cmd(struct f2fs_sb_info *sbi); 3223 void f2fs_stop_discard_thread(struct f2fs_sb_info *sbi); 3224 bool f2fs_issue_discard_timeout(struct f2fs_sb_info *sbi); 3225 void f2fs_clear_prefree_segments(struct f2fs_sb_info *sbi, 3226 struct cp_control *cpc); 3227 void f2fs_dirty_to_prefree(struct f2fs_sb_info *sbi); 3228 block_t f2fs_get_unusable_blocks(struct f2fs_sb_info *sbi); 3229 int f2fs_disable_cp_again(struct f2fs_sb_info *sbi, block_t unusable); 3230 void f2fs_release_discard_addrs(struct f2fs_sb_info *sbi); 3231 int f2fs_npages_for_summary_flush(struct f2fs_sb_info *sbi, bool for_ra); 3232 void allocate_segment_for_resize(struct f2fs_sb_info *sbi, int type, 3233 unsigned int start, unsigned int end); 3234 void f2fs_allocate_new_segments(struct f2fs_sb_info *sbi, int type); 3235 int f2fs_trim_fs(struct f2fs_sb_info *sbi, struct fstrim_range *range); 3236 bool f2fs_exist_trim_candidates(struct f2fs_sb_info *sbi, 3237 struct cp_control *cpc); 3238 struct page *f2fs_get_sum_page(struct f2fs_sb_info *sbi, unsigned int segno); 3239 void f2fs_update_meta_page(struct f2fs_sb_info *sbi, void *src, 3240 block_t blk_addr); 3241 void f2fs_do_write_meta_page(struct f2fs_sb_info *sbi, struct page *page, 3242 enum iostat_type io_type); 3243 void f2fs_do_write_node_page(unsigned int nid, struct f2fs_io_info *fio); 3244 void f2fs_outplace_write_data(struct dnode_of_data *dn, 3245 struct f2fs_io_info *fio); 3246 int f2fs_inplace_write_data(struct f2fs_io_info *fio); 3247 void f2fs_do_replace_block(struct f2fs_sb_info *sbi, struct f2fs_summary *sum, 3248 block_t old_blkaddr, block_t new_blkaddr, 3249 bool recover_curseg, bool recover_newaddr); 3250 void f2fs_replace_block(struct f2fs_sb_info *sbi, struct dnode_of_data *dn, 3251 block_t old_addr, block_t new_addr, 3252 unsigned char version, bool recover_curseg, 3253 bool recover_newaddr); 3254 void f2fs_allocate_data_block(struct f2fs_sb_info *sbi, struct page *page, 3255 block_t old_blkaddr, block_t *new_blkaddr, 3256 struct f2fs_summary *sum, int type, 3257 struct f2fs_io_info *fio, bool add_list); 3258 void f2fs_wait_on_page_writeback(struct page *page, 3259 enum page_type type, bool ordered, bool locked); 3260 void f2fs_wait_on_block_writeback(struct inode *inode, block_t blkaddr); 3261 void f2fs_wait_on_block_writeback_range(struct inode *inode, block_t blkaddr, 3262 block_t len); 3263 void f2fs_write_data_summaries(struct f2fs_sb_info *sbi, block_t start_blk); 3264 void f2fs_write_node_summaries(struct f2fs_sb_info *sbi, block_t start_blk); 3265 int f2fs_lookup_journal_in_cursum(struct f2fs_journal *journal, int type, 3266 unsigned int val, int alloc); 3267 void f2fs_flush_sit_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3268 int f2fs_fix_curseg_write_pointer(struct f2fs_sb_info *sbi); 3269 int f2fs_check_write_pointer(struct f2fs_sb_info *sbi); 3270 int f2fs_build_segment_manager(struct f2fs_sb_info *sbi); 3271 void f2fs_destroy_segment_manager(struct f2fs_sb_info *sbi); 3272 int __init f2fs_create_segment_manager_caches(void); 3273 void f2fs_destroy_segment_manager_caches(void); 3274 int f2fs_rw_hint_to_seg_type(enum rw_hint hint); 3275 enum rw_hint f2fs_io_type_to_rw_hint(struct f2fs_sb_info *sbi, 3276 enum page_type type, enum temp_type temp); 3277 3278 /* 3279 * checkpoint.c 3280 */ 3281 void f2fs_stop_checkpoint(struct f2fs_sb_info *sbi, bool end_io); 3282 struct page *f2fs_grab_meta_page(struct f2fs_sb_info *sbi, pgoff_t index); 3283 struct page *f2fs_get_meta_page(struct f2fs_sb_info *sbi, pgoff_t index); 3284 struct page *f2fs_get_meta_page_nofail(struct f2fs_sb_info *sbi, pgoff_t index); 3285 struct page *f2fs_get_tmp_page(struct f2fs_sb_info *sbi, pgoff_t index); 3286 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi, 3287 block_t blkaddr, int type); 3288 int f2fs_ra_meta_pages(struct f2fs_sb_info *sbi, block_t start, int nrpages, 3289 int type, bool sync); 3290 void f2fs_ra_meta_pages_cond(struct f2fs_sb_info *sbi, pgoff_t index); 3291 long f2fs_sync_meta_pages(struct f2fs_sb_info *sbi, enum page_type type, 3292 long nr_to_write, enum iostat_type io_type); 3293 void f2fs_add_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type); 3294 void f2fs_remove_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type); 3295 void f2fs_release_ino_entry(struct f2fs_sb_info *sbi, bool all); 3296 bool f2fs_exist_written_data(struct f2fs_sb_info *sbi, nid_t ino, int mode); 3297 void f2fs_set_dirty_device(struct f2fs_sb_info *sbi, nid_t ino, 3298 unsigned int devidx, int type); 3299 bool f2fs_is_dirty_device(struct f2fs_sb_info *sbi, nid_t ino, 3300 unsigned int devidx, int type); 3301 int f2fs_sync_inode_meta(struct f2fs_sb_info *sbi); 3302 int f2fs_acquire_orphan_inode(struct f2fs_sb_info *sbi); 3303 void f2fs_release_orphan_inode(struct f2fs_sb_info *sbi); 3304 void f2fs_add_orphan_inode(struct inode *inode); 3305 void f2fs_remove_orphan_inode(struct f2fs_sb_info *sbi, nid_t ino); 3306 int f2fs_recover_orphan_inodes(struct f2fs_sb_info *sbi); 3307 int f2fs_get_valid_checkpoint(struct f2fs_sb_info *sbi); 3308 void f2fs_update_dirty_page(struct inode *inode, struct page *page); 3309 void f2fs_remove_dirty_inode(struct inode *inode); 3310 int f2fs_sync_dirty_inodes(struct f2fs_sb_info *sbi, enum inode_type type); 3311 void f2fs_wait_on_all_pages_writeback(struct f2fs_sb_info *sbi); 3312 int f2fs_write_checkpoint(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3313 void f2fs_init_ino_entry_info(struct f2fs_sb_info *sbi); 3314 int __init f2fs_create_checkpoint_caches(void); 3315 void f2fs_destroy_checkpoint_caches(void); 3316 3317 /* 3318 * data.c 3319 */ 3320 int __init f2fs_init_bioset(void); 3321 void f2fs_destroy_bioset(void); 3322 struct bio *f2fs_bio_alloc(struct f2fs_sb_info *sbi, int npages, bool no_fail); 3323 int f2fs_init_bio_entry_cache(void); 3324 void f2fs_destroy_bio_entry_cache(void); 3325 void f2fs_submit_bio(struct f2fs_sb_info *sbi, 3326 struct bio *bio, enum page_type type); 3327 void f2fs_submit_merged_write(struct f2fs_sb_info *sbi, enum page_type type); 3328 void f2fs_submit_merged_write_cond(struct f2fs_sb_info *sbi, 3329 struct inode *inode, struct page *page, 3330 nid_t ino, enum page_type type); 3331 void f2fs_submit_merged_ipu_write(struct f2fs_sb_info *sbi, 3332 struct bio **bio, struct page *page); 3333 void f2fs_flush_merged_writes(struct f2fs_sb_info *sbi); 3334 int f2fs_submit_page_bio(struct f2fs_io_info *fio); 3335 int f2fs_merge_page_bio(struct f2fs_io_info *fio); 3336 void f2fs_submit_page_write(struct f2fs_io_info *fio); 3337 struct block_device *f2fs_target_device(struct f2fs_sb_info *sbi, 3338 block_t blk_addr, struct bio *bio); 3339 int f2fs_target_device_index(struct f2fs_sb_info *sbi, block_t blkaddr); 3340 void f2fs_set_data_blkaddr(struct dnode_of_data *dn); 3341 void f2fs_update_data_blkaddr(struct dnode_of_data *dn, block_t blkaddr); 3342 int f2fs_reserve_new_blocks(struct dnode_of_data *dn, blkcnt_t count); 3343 int f2fs_reserve_new_block(struct dnode_of_data *dn); 3344 int f2fs_get_block(struct dnode_of_data *dn, pgoff_t index); 3345 int f2fs_preallocate_blocks(struct kiocb *iocb, struct iov_iter *from); 3346 int f2fs_reserve_block(struct dnode_of_data *dn, pgoff_t index); 3347 int f2fs_mpage_readpages(struct address_space *mapping, 3348 struct list_head *pages, struct page *page, 3349 unsigned nr_pages, bool is_readahead); 3350 struct page *f2fs_get_read_data_page(struct inode *inode, pgoff_t index, 3351 int op_flags, bool for_write); 3352 struct page *f2fs_find_data_page(struct inode *inode, pgoff_t index); 3353 struct page *f2fs_get_lock_data_page(struct inode *inode, pgoff_t index, 3354 bool for_write); 3355 struct page *f2fs_get_new_data_page(struct inode *inode, 3356 struct page *ipage, pgoff_t index, bool new_i_size); 3357 int f2fs_do_write_data_page(struct f2fs_io_info *fio); 3358 void __do_map_lock(struct f2fs_sb_info *sbi, int flag, bool lock); 3359 int f2fs_map_blocks(struct inode *inode, struct f2fs_map_blocks *map, 3360 int create, int flag); 3361 int f2fs_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, 3362 u64 start, u64 len); 3363 int f2fs_encrypt_one_page(struct f2fs_io_info *fio); 3364 bool f2fs_should_update_inplace(struct inode *inode, struct f2fs_io_info *fio); 3365 bool f2fs_should_update_outplace(struct inode *inode, struct f2fs_io_info *fio); 3366 int f2fs_write_single_data_page(struct page *page, int *submitted, 3367 struct bio **bio, sector_t *last_block, 3368 struct writeback_control *wbc, 3369 enum iostat_type io_type, 3370 int compr_blocks); 3371 void f2fs_invalidate_page(struct page *page, unsigned int offset, 3372 unsigned int length); 3373 int f2fs_release_page(struct page *page, gfp_t wait); 3374 #ifdef CONFIG_MIGRATION 3375 int f2fs_migrate_page(struct address_space *mapping, struct page *newpage, 3376 struct page *page, enum migrate_mode mode); 3377 #endif 3378 bool f2fs_overwrite_io(struct inode *inode, loff_t pos, size_t len); 3379 void f2fs_clear_page_cache_dirty_tag(struct page *page); 3380 int f2fs_init_post_read_processing(void); 3381 void f2fs_destroy_post_read_processing(void); 3382 int f2fs_init_post_read_wq(struct f2fs_sb_info *sbi); 3383 void f2fs_destroy_post_read_wq(struct f2fs_sb_info *sbi); 3384 3385 /* 3386 * gc.c 3387 */ 3388 int f2fs_start_gc_thread(struct f2fs_sb_info *sbi); 3389 void f2fs_stop_gc_thread(struct f2fs_sb_info *sbi); 3390 block_t f2fs_start_bidx_of_node(unsigned int node_ofs, struct inode *inode); 3391 int f2fs_gc(struct f2fs_sb_info *sbi, bool sync, bool background, 3392 unsigned int segno); 3393 void f2fs_build_gc_manager(struct f2fs_sb_info *sbi); 3394 int f2fs_resize_fs(struct f2fs_sb_info *sbi, __u64 block_count); 3395 3396 /* 3397 * recovery.c 3398 */ 3399 int f2fs_recover_fsync_data(struct f2fs_sb_info *sbi, bool check_only); 3400 bool f2fs_space_for_roll_forward(struct f2fs_sb_info *sbi); 3401 3402 /* 3403 * debug.c 3404 */ 3405 #ifdef CONFIG_F2FS_STAT_FS 3406 struct f2fs_stat_info { 3407 struct list_head stat_list; 3408 struct f2fs_sb_info *sbi; 3409 int all_area_segs, sit_area_segs, nat_area_segs, ssa_area_segs; 3410 int main_area_segs, main_area_sections, main_area_zones; 3411 unsigned long long hit_largest, hit_cached, hit_rbtree; 3412 unsigned long long hit_total, total_ext; 3413 int ext_tree, zombie_tree, ext_node; 3414 int ndirty_node, ndirty_dent, ndirty_meta, ndirty_imeta; 3415 int ndirty_data, ndirty_qdata; 3416 int inmem_pages; 3417 unsigned int ndirty_dirs, ndirty_files, nquota_files, ndirty_all; 3418 int nats, dirty_nats, sits, dirty_sits; 3419 int free_nids, avail_nids, alloc_nids; 3420 int total_count, utilization; 3421 int bg_gc, nr_wb_cp_data, nr_wb_data; 3422 int nr_rd_data, nr_rd_node, nr_rd_meta; 3423 int nr_dio_read, nr_dio_write; 3424 unsigned int io_skip_bggc, other_skip_bggc; 3425 int nr_flushing, nr_flushed, flush_list_empty; 3426 int nr_discarding, nr_discarded; 3427 int nr_discard_cmd; 3428 unsigned int undiscard_blks; 3429 int inline_xattr, inline_inode, inline_dir, append, update, orphans; 3430 int compr_inode, compr_blocks; 3431 int aw_cnt, max_aw_cnt, vw_cnt, max_vw_cnt; 3432 unsigned int valid_count, valid_node_count, valid_inode_count, discard_blks; 3433 unsigned int bimodal, avg_vblocks; 3434 int util_free, util_valid, util_invalid; 3435 int rsvd_segs, overp_segs; 3436 int dirty_count, node_pages, meta_pages; 3437 int prefree_count, call_count, cp_count, bg_cp_count; 3438 int tot_segs, node_segs, data_segs, free_segs, free_secs; 3439 int bg_node_segs, bg_data_segs; 3440 int tot_blks, data_blks, node_blks; 3441 int bg_data_blks, bg_node_blks; 3442 unsigned long long skipped_atomic_files[2]; 3443 int curseg[NR_CURSEG_TYPE]; 3444 int cursec[NR_CURSEG_TYPE]; 3445 int curzone[NR_CURSEG_TYPE]; 3446 3447 unsigned int meta_count[META_MAX]; 3448 unsigned int segment_count[2]; 3449 unsigned int block_count[2]; 3450 unsigned int inplace_count; 3451 unsigned long long base_mem, cache_mem, page_mem; 3452 }; 3453 3454 static inline struct f2fs_stat_info *F2FS_STAT(struct f2fs_sb_info *sbi) 3455 { 3456 return (struct f2fs_stat_info *)sbi->stat_info; 3457 } 3458 3459 #define stat_inc_cp_count(si) ((si)->cp_count++) 3460 #define stat_inc_bg_cp_count(si) ((si)->bg_cp_count++) 3461 #define stat_inc_call_count(si) ((si)->call_count++) 3462 #define stat_inc_bggc_count(si) ((si)->bg_gc++) 3463 #define stat_io_skip_bggc_count(sbi) ((sbi)->io_skip_bggc++) 3464 #define stat_other_skip_bggc_count(sbi) ((sbi)->other_skip_bggc++) 3465 #define stat_inc_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]++) 3466 #define stat_dec_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]--) 3467 #define stat_inc_total_hit(sbi) (atomic64_inc(&(sbi)->total_hit_ext)) 3468 #define stat_inc_rbtree_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_rbtree)) 3469 #define stat_inc_largest_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_largest)) 3470 #define stat_inc_cached_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_cached)) 3471 #define stat_inc_inline_xattr(inode) \ 3472 do { \ 3473 if (f2fs_has_inline_xattr(inode)) \ 3474 (atomic_inc(&F2FS_I_SB(inode)->inline_xattr)); \ 3475 } while (0) 3476 #define stat_dec_inline_xattr(inode) \ 3477 do { \ 3478 if (f2fs_has_inline_xattr(inode)) \ 3479 (atomic_dec(&F2FS_I_SB(inode)->inline_xattr)); \ 3480 } while (0) 3481 #define stat_inc_inline_inode(inode) \ 3482 do { \ 3483 if (f2fs_has_inline_data(inode)) \ 3484 (atomic_inc(&F2FS_I_SB(inode)->inline_inode)); \ 3485 } while (0) 3486 #define stat_dec_inline_inode(inode) \ 3487 do { \ 3488 if (f2fs_has_inline_data(inode)) \ 3489 (atomic_dec(&F2FS_I_SB(inode)->inline_inode)); \ 3490 } while (0) 3491 #define stat_inc_inline_dir(inode) \ 3492 do { \ 3493 if (f2fs_has_inline_dentry(inode)) \ 3494 (atomic_inc(&F2FS_I_SB(inode)->inline_dir)); \ 3495 } while (0) 3496 #define stat_dec_inline_dir(inode) \ 3497 do { \ 3498 if (f2fs_has_inline_dentry(inode)) \ 3499 (atomic_dec(&F2FS_I_SB(inode)->inline_dir)); \ 3500 } while (0) 3501 #define stat_inc_compr_inode(inode) \ 3502 do { \ 3503 if (f2fs_compressed_file(inode)) \ 3504 (atomic_inc(&F2FS_I_SB(inode)->compr_inode)); \ 3505 } while (0) 3506 #define stat_dec_compr_inode(inode) \ 3507 do { \ 3508 if (f2fs_compressed_file(inode)) \ 3509 (atomic_dec(&F2FS_I_SB(inode)->compr_inode)); \ 3510 } while (0) 3511 #define stat_add_compr_blocks(inode, blocks) \ 3512 (atomic_add(blocks, &F2FS_I_SB(inode)->compr_blocks)) 3513 #define stat_sub_compr_blocks(inode, blocks) \ 3514 (atomic_sub(blocks, &F2FS_I_SB(inode)->compr_blocks)) 3515 #define stat_inc_meta_count(sbi, blkaddr) \ 3516 do { \ 3517 if (blkaddr < SIT_I(sbi)->sit_base_addr) \ 3518 atomic_inc(&(sbi)->meta_count[META_CP]); \ 3519 else if (blkaddr < NM_I(sbi)->nat_blkaddr) \ 3520 atomic_inc(&(sbi)->meta_count[META_SIT]); \ 3521 else if (blkaddr < SM_I(sbi)->ssa_blkaddr) \ 3522 atomic_inc(&(sbi)->meta_count[META_NAT]); \ 3523 else if (blkaddr < SM_I(sbi)->main_blkaddr) \ 3524 atomic_inc(&(sbi)->meta_count[META_SSA]); \ 3525 } while (0) 3526 #define stat_inc_seg_type(sbi, curseg) \ 3527 ((sbi)->segment_count[(curseg)->alloc_type]++) 3528 #define stat_inc_block_count(sbi, curseg) \ 3529 ((sbi)->block_count[(curseg)->alloc_type]++) 3530 #define stat_inc_inplace_blocks(sbi) \ 3531 (atomic_inc(&(sbi)->inplace_count)) 3532 #define stat_update_max_atomic_write(inode) \ 3533 do { \ 3534 int cur = F2FS_I_SB(inode)->atomic_files; \ 3535 int max = atomic_read(&F2FS_I_SB(inode)->max_aw_cnt); \ 3536 if (cur > max) \ 3537 atomic_set(&F2FS_I_SB(inode)->max_aw_cnt, cur); \ 3538 } while (0) 3539 #define stat_inc_volatile_write(inode) \ 3540 (atomic_inc(&F2FS_I_SB(inode)->vw_cnt)) 3541 #define stat_dec_volatile_write(inode) \ 3542 (atomic_dec(&F2FS_I_SB(inode)->vw_cnt)) 3543 #define stat_update_max_volatile_write(inode) \ 3544 do { \ 3545 int cur = atomic_read(&F2FS_I_SB(inode)->vw_cnt); \ 3546 int max = atomic_read(&F2FS_I_SB(inode)->max_vw_cnt); \ 3547 if (cur > max) \ 3548 atomic_set(&F2FS_I_SB(inode)->max_vw_cnt, cur); \ 3549 } while (0) 3550 #define stat_inc_seg_count(sbi, type, gc_type) \ 3551 do { \ 3552 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 3553 si->tot_segs++; \ 3554 if ((type) == SUM_TYPE_DATA) { \ 3555 si->data_segs++; \ 3556 si->bg_data_segs += (gc_type == BG_GC) ? 1 : 0; \ 3557 } else { \ 3558 si->node_segs++; \ 3559 si->bg_node_segs += (gc_type == BG_GC) ? 1 : 0; \ 3560 } \ 3561 } while (0) 3562 3563 #define stat_inc_tot_blk_count(si, blks) \ 3564 ((si)->tot_blks += (blks)) 3565 3566 #define stat_inc_data_blk_count(sbi, blks, gc_type) \ 3567 do { \ 3568 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 3569 stat_inc_tot_blk_count(si, blks); \ 3570 si->data_blks += (blks); \ 3571 si->bg_data_blks += ((gc_type) == BG_GC) ? (blks) : 0; \ 3572 } while (0) 3573 3574 #define stat_inc_node_blk_count(sbi, blks, gc_type) \ 3575 do { \ 3576 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 3577 stat_inc_tot_blk_count(si, blks); \ 3578 si->node_blks += (blks); \ 3579 si->bg_node_blks += ((gc_type) == BG_GC) ? (blks) : 0; \ 3580 } while (0) 3581 3582 int f2fs_build_stats(struct f2fs_sb_info *sbi); 3583 void f2fs_destroy_stats(struct f2fs_sb_info *sbi); 3584 void __init f2fs_create_root_stats(void); 3585 void f2fs_destroy_root_stats(void); 3586 void f2fs_update_sit_info(struct f2fs_sb_info *sbi); 3587 #else 3588 #define stat_inc_cp_count(si) do { } while (0) 3589 #define stat_inc_bg_cp_count(si) do { } while (0) 3590 #define stat_inc_call_count(si) do { } while (0) 3591 #define stat_inc_bggc_count(si) do { } while (0) 3592 #define stat_io_skip_bggc_count(sbi) do { } while (0) 3593 #define stat_other_skip_bggc_count(sbi) do { } while (0) 3594 #define stat_inc_dirty_inode(sbi, type) do { } while (0) 3595 #define stat_dec_dirty_inode(sbi, type) do { } while (0) 3596 #define stat_inc_total_hit(sbi) do { } while (0) 3597 #define stat_inc_rbtree_node_hit(sbi) do { } while (0) 3598 #define stat_inc_largest_node_hit(sbi) do { } while (0) 3599 #define stat_inc_cached_node_hit(sbi) do { } while (0) 3600 #define stat_inc_inline_xattr(inode) do { } while (0) 3601 #define stat_dec_inline_xattr(inode) do { } while (0) 3602 #define stat_inc_inline_inode(inode) do { } while (0) 3603 #define stat_dec_inline_inode(inode) do { } while (0) 3604 #define stat_inc_inline_dir(inode) do { } while (0) 3605 #define stat_dec_inline_dir(inode) do { } while (0) 3606 #define stat_inc_compr_inode(inode) do { } while (0) 3607 #define stat_dec_compr_inode(inode) do { } while (0) 3608 #define stat_add_compr_blocks(inode, blocks) do { } while (0) 3609 #define stat_sub_compr_blocks(inode, blocks) do { } while (0) 3610 #define stat_inc_atomic_write(inode) do { } while (0) 3611 #define stat_dec_atomic_write(inode) do { } while (0) 3612 #define stat_update_max_atomic_write(inode) do { } while (0) 3613 #define stat_inc_volatile_write(inode) do { } while (0) 3614 #define stat_dec_volatile_write(inode) do { } while (0) 3615 #define stat_update_max_volatile_write(inode) do { } while (0) 3616 #define stat_inc_meta_count(sbi, blkaddr) do { } while (0) 3617 #define stat_inc_seg_type(sbi, curseg) do { } while (0) 3618 #define stat_inc_block_count(sbi, curseg) do { } while (0) 3619 #define stat_inc_inplace_blocks(sbi) do { } while (0) 3620 #define stat_inc_seg_count(sbi, type, gc_type) do { } while (0) 3621 #define stat_inc_tot_blk_count(si, blks) do { } while (0) 3622 #define stat_inc_data_blk_count(sbi, blks, gc_type) do { } while (0) 3623 #define stat_inc_node_blk_count(sbi, blks, gc_type) do { } while (0) 3624 3625 static inline int f2fs_build_stats(struct f2fs_sb_info *sbi) { return 0; } 3626 static inline void f2fs_destroy_stats(struct f2fs_sb_info *sbi) { } 3627 static inline void __init f2fs_create_root_stats(void) { } 3628 static inline void f2fs_destroy_root_stats(void) { } 3629 static inline void update_sit_info(struct f2fs_sb_info *sbi) {} 3630 #endif 3631 3632 extern const struct file_operations f2fs_dir_operations; 3633 #ifdef CONFIG_UNICODE 3634 extern const struct dentry_operations f2fs_dentry_ops; 3635 #endif 3636 extern const struct file_operations f2fs_file_operations; 3637 extern const struct inode_operations f2fs_file_inode_operations; 3638 extern const struct address_space_operations f2fs_dblock_aops; 3639 extern const struct address_space_operations f2fs_node_aops; 3640 extern const struct address_space_operations f2fs_meta_aops; 3641 extern const struct inode_operations f2fs_dir_inode_operations; 3642 extern const struct inode_operations f2fs_symlink_inode_operations; 3643 extern const struct inode_operations f2fs_encrypted_symlink_inode_operations; 3644 extern const struct inode_operations f2fs_special_inode_operations; 3645 extern struct kmem_cache *f2fs_inode_entry_slab; 3646 3647 /* 3648 * inline.c 3649 */ 3650 bool f2fs_may_inline_data(struct inode *inode); 3651 bool f2fs_may_inline_dentry(struct inode *inode); 3652 void f2fs_do_read_inline_data(struct page *page, struct page *ipage); 3653 void f2fs_truncate_inline_inode(struct inode *inode, 3654 struct page *ipage, u64 from); 3655 int f2fs_read_inline_data(struct inode *inode, struct page *page); 3656 int f2fs_convert_inline_page(struct dnode_of_data *dn, struct page *page); 3657 int f2fs_convert_inline_inode(struct inode *inode); 3658 int f2fs_try_convert_inline_dir(struct inode *dir, struct dentry *dentry); 3659 int f2fs_write_inline_data(struct inode *inode, struct page *page); 3660 bool f2fs_recover_inline_data(struct inode *inode, struct page *npage); 3661 struct f2fs_dir_entry *f2fs_find_in_inline_dir(struct inode *dir, 3662 struct fscrypt_name *fname, struct page **res_page); 3663 int f2fs_make_empty_inline_dir(struct inode *inode, struct inode *parent, 3664 struct page *ipage); 3665 int f2fs_add_inline_entry(struct inode *dir, const struct qstr *new_name, 3666 const struct qstr *orig_name, 3667 struct inode *inode, nid_t ino, umode_t mode); 3668 void f2fs_delete_inline_entry(struct f2fs_dir_entry *dentry, 3669 struct page *page, struct inode *dir, 3670 struct inode *inode); 3671 bool f2fs_empty_inline_dir(struct inode *dir); 3672 int f2fs_read_inline_dir(struct file *file, struct dir_context *ctx, 3673 struct fscrypt_str *fstr); 3674 int f2fs_inline_data_fiemap(struct inode *inode, 3675 struct fiemap_extent_info *fieinfo, 3676 __u64 start, __u64 len); 3677 3678 /* 3679 * shrinker.c 3680 */ 3681 unsigned long f2fs_shrink_count(struct shrinker *shrink, 3682 struct shrink_control *sc); 3683 unsigned long f2fs_shrink_scan(struct shrinker *shrink, 3684 struct shrink_control *sc); 3685 void f2fs_join_shrinker(struct f2fs_sb_info *sbi); 3686 void f2fs_leave_shrinker(struct f2fs_sb_info *sbi); 3687 3688 /* 3689 * extent_cache.c 3690 */ 3691 struct rb_entry *f2fs_lookup_rb_tree(struct rb_root_cached *root, 3692 struct rb_entry *cached_re, unsigned int ofs); 3693 struct rb_node **f2fs_lookup_rb_tree_for_insert(struct f2fs_sb_info *sbi, 3694 struct rb_root_cached *root, 3695 struct rb_node **parent, 3696 unsigned int ofs, bool *leftmost); 3697 struct rb_entry *f2fs_lookup_rb_tree_ret(struct rb_root_cached *root, 3698 struct rb_entry *cached_re, unsigned int ofs, 3699 struct rb_entry **prev_entry, struct rb_entry **next_entry, 3700 struct rb_node ***insert_p, struct rb_node **insert_parent, 3701 bool force, bool *leftmost); 3702 bool f2fs_check_rb_tree_consistence(struct f2fs_sb_info *sbi, 3703 struct rb_root_cached *root); 3704 unsigned int f2fs_shrink_extent_tree(struct f2fs_sb_info *sbi, int nr_shrink); 3705 bool f2fs_init_extent_tree(struct inode *inode, struct f2fs_extent *i_ext); 3706 void f2fs_drop_extent_tree(struct inode *inode); 3707 unsigned int f2fs_destroy_extent_node(struct inode *inode); 3708 void f2fs_destroy_extent_tree(struct inode *inode); 3709 bool f2fs_lookup_extent_cache(struct inode *inode, pgoff_t pgofs, 3710 struct extent_info *ei); 3711 void f2fs_update_extent_cache(struct dnode_of_data *dn); 3712 void f2fs_update_extent_cache_range(struct dnode_of_data *dn, 3713 pgoff_t fofs, block_t blkaddr, unsigned int len); 3714 void f2fs_init_extent_cache_info(struct f2fs_sb_info *sbi); 3715 int __init f2fs_create_extent_cache(void); 3716 void f2fs_destroy_extent_cache(void); 3717 3718 /* 3719 * sysfs.c 3720 */ 3721 int __init f2fs_init_sysfs(void); 3722 void f2fs_exit_sysfs(void); 3723 int f2fs_register_sysfs(struct f2fs_sb_info *sbi); 3724 void f2fs_unregister_sysfs(struct f2fs_sb_info *sbi); 3725 3726 /* verity.c */ 3727 extern const struct fsverity_operations f2fs_verityops; 3728 3729 /* 3730 * crypto support 3731 */ 3732 static inline bool f2fs_encrypted_file(struct inode *inode) 3733 { 3734 return IS_ENCRYPTED(inode) && S_ISREG(inode->i_mode); 3735 } 3736 3737 static inline void f2fs_set_encrypted_inode(struct inode *inode) 3738 { 3739 #ifdef CONFIG_FS_ENCRYPTION 3740 file_set_encrypt(inode); 3741 f2fs_set_inode_flags(inode); 3742 #endif 3743 } 3744 3745 /* 3746 * Returns true if the reads of the inode's data need to undergo some 3747 * postprocessing step, like decryption or authenticity verification. 3748 */ 3749 static inline bool f2fs_post_read_required(struct inode *inode) 3750 { 3751 return f2fs_encrypted_file(inode) || fsverity_active(inode) || 3752 f2fs_compressed_file(inode); 3753 } 3754 3755 /* 3756 * compress.c 3757 */ 3758 #ifdef CONFIG_F2FS_FS_COMPRESSION 3759 bool f2fs_is_compressed_page(struct page *page); 3760 struct page *f2fs_compress_control_page(struct page *page); 3761 int f2fs_prepare_compress_overwrite(struct inode *inode, 3762 struct page **pagep, pgoff_t index, void **fsdata); 3763 bool f2fs_compress_write_end(struct inode *inode, void *fsdata, 3764 pgoff_t index, unsigned copied); 3765 void f2fs_compress_write_end_io(struct bio *bio, struct page *page); 3766 bool f2fs_is_compress_backend_ready(struct inode *inode); 3767 void f2fs_decompress_pages(struct bio *bio, struct page *page, bool verity); 3768 bool f2fs_cluster_is_empty(struct compress_ctx *cc); 3769 bool f2fs_cluster_can_merge_page(struct compress_ctx *cc, pgoff_t index); 3770 void f2fs_compress_ctx_add_page(struct compress_ctx *cc, struct page *page); 3771 int f2fs_write_multi_pages(struct compress_ctx *cc, 3772 int *submitted, 3773 struct writeback_control *wbc, 3774 enum iostat_type io_type); 3775 int f2fs_is_compressed_cluster(struct inode *inode, pgoff_t index); 3776 int f2fs_read_multi_pages(struct compress_ctx *cc, struct bio **bio_ret, 3777 unsigned nr_pages, sector_t *last_block_in_bio, 3778 bool is_readahead); 3779 struct decompress_io_ctx *f2fs_alloc_dic(struct compress_ctx *cc); 3780 void f2fs_free_dic(struct decompress_io_ctx *dic); 3781 void f2fs_decompress_end_io(struct page **rpages, 3782 unsigned int cluster_size, bool err, bool verity); 3783 int f2fs_init_compress_ctx(struct compress_ctx *cc); 3784 void f2fs_destroy_compress_ctx(struct compress_ctx *cc); 3785 void f2fs_init_compress_info(struct f2fs_sb_info *sbi); 3786 #else 3787 static inline bool f2fs_is_compressed_page(struct page *page) { return false; } 3788 static inline bool f2fs_is_compress_backend_ready(struct inode *inode) 3789 { 3790 if (!f2fs_compressed_file(inode)) 3791 return true; 3792 /* not support compression */ 3793 return false; 3794 } 3795 static inline struct page *f2fs_compress_control_page(struct page *page) 3796 { 3797 WARN_ON_ONCE(1); 3798 return ERR_PTR(-EINVAL); 3799 } 3800 #endif 3801 3802 static inline void set_compress_context(struct inode *inode) 3803 { 3804 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 3805 3806 F2FS_I(inode)->i_compress_algorithm = 3807 F2FS_OPTION(sbi).compress_algorithm; 3808 F2FS_I(inode)->i_log_cluster_size = 3809 F2FS_OPTION(sbi).compress_log_size; 3810 F2FS_I(inode)->i_cluster_size = 3811 1 << F2FS_I(inode)->i_log_cluster_size; 3812 F2FS_I(inode)->i_flags |= F2FS_COMPR_FL; 3813 set_inode_flag(inode, FI_COMPRESSED_FILE); 3814 stat_inc_compr_inode(inode); 3815 } 3816 3817 static inline u64 f2fs_disable_compressed_file(struct inode *inode) 3818 { 3819 struct f2fs_inode_info *fi = F2FS_I(inode); 3820 3821 if (!f2fs_compressed_file(inode)) 3822 return 0; 3823 if (fi->i_compr_blocks) 3824 return fi->i_compr_blocks; 3825 3826 fi->i_flags &= ~F2FS_COMPR_FL; 3827 clear_inode_flag(inode, FI_COMPRESSED_FILE); 3828 stat_dec_compr_inode(inode); 3829 return 0; 3830 } 3831 3832 #define F2FS_FEATURE_FUNCS(name, flagname) \ 3833 static inline int f2fs_sb_has_##name(struct f2fs_sb_info *sbi) \ 3834 { \ 3835 return F2FS_HAS_FEATURE(sbi, F2FS_FEATURE_##flagname); \ 3836 } 3837 3838 F2FS_FEATURE_FUNCS(encrypt, ENCRYPT); 3839 F2FS_FEATURE_FUNCS(blkzoned, BLKZONED); 3840 F2FS_FEATURE_FUNCS(extra_attr, EXTRA_ATTR); 3841 F2FS_FEATURE_FUNCS(project_quota, PRJQUOTA); 3842 F2FS_FEATURE_FUNCS(inode_chksum, INODE_CHKSUM); 3843 F2FS_FEATURE_FUNCS(flexible_inline_xattr, FLEXIBLE_INLINE_XATTR); 3844 F2FS_FEATURE_FUNCS(quota_ino, QUOTA_INO); 3845 F2FS_FEATURE_FUNCS(inode_crtime, INODE_CRTIME); 3846 F2FS_FEATURE_FUNCS(lost_found, LOST_FOUND); 3847 F2FS_FEATURE_FUNCS(verity, VERITY); 3848 F2FS_FEATURE_FUNCS(sb_chksum, SB_CHKSUM); 3849 F2FS_FEATURE_FUNCS(casefold, CASEFOLD); 3850 F2FS_FEATURE_FUNCS(compression, COMPRESSION); 3851 3852 #ifdef CONFIG_BLK_DEV_ZONED 3853 static inline bool f2fs_blkz_is_seq(struct f2fs_sb_info *sbi, int devi, 3854 block_t blkaddr) 3855 { 3856 unsigned int zno = blkaddr >> sbi->log_blocks_per_blkz; 3857 3858 return test_bit(zno, FDEV(devi).blkz_seq); 3859 } 3860 #endif 3861 3862 static inline bool f2fs_hw_should_discard(struct f2fs_sb_info *sbi) 3863 { 3864 return f2fs_sb_has_blkzoned(sbi); 3865 } 3866 3867 static inline bool f2fs_bdev_support_discard(struct block_device *bdev) 3868 { 3869 return blk_queue_discard(bdev_get_queue(bdev)) || 3870 bdev_is_zoned(bdev); 3871 } 3872 3873 static inline bool f2fs_hw_support_discard(struct f2fs_sb_info *sbi) 3874 { 3875 int i; 3876 3877 if (!f2fs_is_multi_device(sbi)) 3878 return f2fs_bdev_support_discard(sbi->sb->s_bdev); 3879 3880 for (i = 0; i < sbi->s_ndevs; i++) 3881 if (f2fs_bdev_support_discard(FDEV(i).bdev)) 3882 return true; 3883 return false; 3884 } 3885 3886 static inline bool f2fs_realtime_discard_enable(struct f2fs_sb_info *sbi) 3887 { 3888 return (test_opt(sbi, DISCARD) && f2fs_hw_support_discard(sbi)) || 3889 f2fs_hw_should_discard(sbi); 3890 } 3891 3892 static inline bool f2fs_hw_is_readonly(struct f2fs_sb_info *sbi) 3893 { 3894 int i; 3895 3896 if (!f2fs_is_multi_device(sbi)) 3897 return bdev_read_only(sbi->sb->s_bdev); 3898 3899 for (i = 0; i < sbi->s_ndevs; i++) 3900 if (bdev_read_only(FDEV(i).bdev)) 3901 return true; 3902 return false; 3903 } 3904 3905 3906 static inline void set_opt_mode(struct f2fs_sb_info *sbi, unsigned int mt) 3907 { 3908 clear_opt(sbi, ADAPTIVE); 3909 clear_opt(sbi, LFS); 3910 3911 switch (mt) { 3912 case F2FS_MOUNT_ADAPTIVE: 3913 set_opt(sbi, ADAPTIVE); 3914 break; 3915 case F2FS_MOUNT_LFS: 3916 set_opt(sbi, LFS); 3917 break; 3918 } 3919 } 3920 3921 static inline bool f2fs_may_encrypt(struct inode *inode) 3922 { 3923 #ifdef CONFIG_FS_ENCRYPTION 3924 umode_t mode = inode->i_mode; 3925 3926 return (S_ISREG(mode) || S_ISDIR(mode) || S_ISLNK(mode)); 3927 #else 3928 return false; 3929 #endif 3930 } 3931 3932 static inline bool f2fs_may_compress(struct inode *inode) 3933 { 3934 if (IS_SWAPFILE(inode) || f2fs_is_pinned_file(inode) || 3935 f2fs_is_atomic_file(inode) || 3936 f2fs_is_volatile_file(inode)) 3937 return false; 3938 return S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode); 3939 } 3940 3941 static inline void f2fs_i_compr_blocks_update(struct inode *inode, 3942 u64 blocks, bool add) 3943 { 3944 int diff = F2FS_I(inode)->i_cluster_size - blocks; 3945 3946 if (add) { 3947 F2FS_I(inode)->i_compr_blocks += diff; 3948 stat_add_compr_blocks(inode, diff); 3949 } else { 3950 F2FS_I(inode)->i_compr_blocks -= diff; 3951 stat_sub_compr_blocks(inode, diff); 3952 } 3953 f2fs_mark_inode_dirty_sync(inode, true); 3954 } 3955 3956 static inline int block_unaligned_IO(struct inode *inode, 3957 struct kiocb *iocb, struct iov_iter *iter) 3958 { 3959 unsigned int i_blkbits = READ_ONCE(inode->i_blkbits); 3960 unsigned int blocksize_mask = (1 << i_blkbits) - 1; 3961 loff_t offset = iocb->ki_pos; 3962 unsigned long align = offset | iov_iter_alignment(iter); 3963 3964 return align & blocksize_mask; 3965 } 3966 3967 static inline int allow_outplace_dio(struct inode *inode, 3968 struct kiocb *iocb, struct iov_iter *iter) 3969 { 3970 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 3971 int rw = iov_iter_rw(iter); 3972 3973 return (test_opt(sbi, LFS) && (rw == WRITE) && 3974 !block_unaligned_IO(inode, iocb, iter)); 3975 } 3976 3977 static inline bool f2fs_force_buffered_io(struct inode *inode, 3978 struct kiocb *iocb, struct iov_iter *iter) 3979 { 3980 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 3981 int rw = iov_iter_rw(iter); 3982 3983 if (f2fs_post_read_required(inode)) 3984 return true; 3985 if (f2fs_is_multi_device(sbi)) 3986 return true; 3987 if (f2fs_compressed_file(inode)) 3988 return true; 3989 /* 3990 * for blkzoned device, fallback direct IO to buffered IO, so 3991 * all IOs can be serialized by log-structured write. 3992 */ 3993 if (f2fs_sb_has_blkzoned(sbi)) 3994 return true; 3995 if (test_opt(sbi, LFS) && (rw == WRITE)) { 3996 if (block_unaligned_IO(inode, iocb, iter)) 3997 return true; 3998 if (F2FS_IO_ALIGNED(sbi)) 3999 return true; 4000 } 4001 if (is_sbi_flag_set(F2FS_I_SB(inode), SBI_CP_DISABLED) && 4002 !IS_SWAPFILE(inode)) 4003 return true; 4004 4005 return false; 4006 } 4007 4008 #ifdef CONFIG_F2FS_FAULT_INJECTION 4009 extern void f2fs_build_fault_attr(struct f2fs_sb_info *sbi, unsigned int rate, 4010 unsigned int type); 4011 #else 4012 #define f2fs_build_fault_attr(sbi, rate, type) do { } while (0) 4013 #endif 4014 4015 static inline bool is_journalled_quota(struct f2fs_sb_info *sbi) 4016 { 4017 #ifdef CONFIG_QUOTA 4018 if (f2fs_sb_has_quota_ino(sbi)) 4019 return true; 4020 if (F2FS_OPTION(sbi).s_qf_names[USRQUOTA] || 4021 F2FS_OPTION(sbi).s_qf_names[GRPQUOTA] || 4022 F2FS_OPTION(sbi).s_qf_names[PRJQUOTA]) 4023 return true; 4024 #endif 4025 return false; 4026 } 4027 4028 #define EFSBADCRC EBADMSG /* Bad CRC detected */ 4029 #define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */ 4030 4031 #endif /* _LINUX_F2FS_H */ 4032