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 29 #ifdef CONFIG_F2FS_CHECK_FS 30 #define f2fs_bug_on(sbi, condition) BUG_ON(condition) 31 #else 32 #define f2fs_bug_on(sbi, condition) \ 33 do { \ 34 if (unlikely(condition)) { \ 35 WARN_ON(1); \ 36 set_sbi_flag(sbi, SBI_NEED_FSCK); \ 37 } \ 38 } while (0) 39 #endif 40 41 enum { 42 FAULT_KMALLOC, 43 FAULT_KVMALLOC, 44 FAULT_PAGE_ALLOC, 45 FAULT_PAGE_GET, 46 FAULT_ALLOC_BIO, 47 FAULT_ALLOC_NID, 48 FAULT_ORPHAN, 49 FAULT_BLOCK, 50 FAULT_DIR_DEPTH, 51 FAULT_EVICT_INODE, 52 FAULT_TRUNCATE, 53 FAULT_READ_IO, 54 FAULT_CHECKPOINT, 55 FAULT_DISCARD, 56 FAULT_WRITE_IO, 57 FAULT_MAX, 58 }; 59 60 #ifdef CONFIG_F2FS_FAULT_INJECTION 61 #define F2FS_ALL_FAULT_TYPE ((1 << FAULT_MAX) - 1) 62 63 struct f2fs_fault_info { 64 atomic_t inject_ops; 65 unsigned int inject_rate; 66 unsigned int inject_type; 67 }; 68 69 extern const char *f2fs_fault_name[FAULT_MAX]; 70 #define IS_FAULT_SET(fi, type) ((fi)->inject_type & (1 << (type))) 71 #endif 72 73 /* 74 * For mount options 75 */ 76 #define F2FS_MOUNT_BG_GC 0x00000001 77 #define F2FS_MOUNT_DISABLE_ROLL_FORWARD 0x00000002 78 #define F2FS_MOUNT_DISCARD 0x00000004 79 #define F2FS_MOUNT_NOHEAP 0x00000008 80 #define F2FS_MOUNT_XATTR_USER 0x00000010 81 #define F2FS_MOUNT_POSIX_ACL 0x00000020 82 #define F2FS_MOUNT_DISABLE_EXT_IDENTIFY 0x00000040 83 #define F2FS_MOUNT_INLINE_XATTR 0x00000080 84 #define F2FS_MOUNT_INLINE_DATA 0x00000100 85 #define F2FS_MOUNT_INLINE_DENTRY 0x00000200 86 #define F2FS_MOUNT_FLUSH_MERGE 0x00000400 87 #define F2FS_MOUNT_NOBARRIER 0x00000800 88 #define F2FS_MOUNT_FASTBOOT 0x00001000 89 #define F2FS_MOUNT_EXTENT_CACHE 0x00002000 90 #define F2FS_MOUNT_FORCE_FG_GC 0x00004000 91 #define F2FS_MOUNT_DATA_FLUSH 0x00008000 92 #define F2FS_MOUNT_FAULT_INJECTION 0x00010000 93 #define F2FS_MOUNT_ADAPTIVE 0x00020000 94 #define F2FS_MOUNT_LFS 0x00040000 95 #define F2FS_MOUNT_USRQUOTA 0x00080000 96 #define F2FS_MOUNT_GRPQUOTA 0x00100000 97 #define F2FS_MOUNT_PRJQUOTA 0x00200000 98 #define F2FS_MOUNT_QUOTA 0x00400000 99 #define F2FS_MOUNT_INLINE_XATTR_SIZE 0x00800000 100 #define F2FS_MOUNT_RESERVE_ROOT 0x01000000 101 #define F2FS_MOUNT_DISABLE_CHECKPOINT 0x02000000 102 103 #define F2FS_OPTION(sbi) ((sbi)->mount_opt) 104 #define clear_opt(sbi, option) (F2FS_OPTION(sbi).opt &= ~F2FS_MOUNT_##option) 105 #define set_opt(sbi, option) (F2FS_OPTION(sbi).opt |= F2FS_MOUNT_##option) 106 #define test_opt(sbi, option) (F2FS_OPTION(sbi).opt & F2FS_MOUNT_##option) 107 108 #define ver_after(a, b) (typecheck(unsigned long long, a) && \ 109 typecheck(unsigned long long, b) && \ 110 ((long long)((a) - (b)) > 0)) 111 112 typedef u32 block_t; /* 113 * should not change u32, since it is the on-disk block 114 * address format, __le32. 115 */ 116 typedef u32 nid_t; 117 118 struct f2fs_mount_info { 119 unsigned int opt; 120 int write_io_size_bits; /* Write IO size bits */ 121 block_t root_reserved_blocks; /* root reserved blocks */ 122 kuid_t s_resuid; /* reserved blocks for uid */ 123 kgid_t s_resgid; /* reserved blocks for gid */ 124 int active_logs; /* # of active logs */ 125 int inline_xattr_size; /* inline xattr size */ 126 #ifdef CONFIG_F2FS_FAULT_INJECTION 127 struct f2fs_fault_info fault_info; /* For fault injection */ 128 #endif 129 #ifdef CONFIG_QUOTA 130 /* Names of quota files with journalled quota */ 131 char *s_qf_names[MAXQUOTAS]; 132 int s_jquota_fmt; /* Format of quota to use */ 133 #endif 134 /* For which write hints are passed down to block layer */ 135 int whint_mode; 136 int alloc_mode; /* segment allocation policy */ 137 int fsync_mode; /* fsync policy */ 138 bool test_dummy_encryption; /* test dummy encryption */ 139 }; 140 141 #define F2FS_FEATURE_ENCRYPT 0x0001 142 #define F2FS_FEATURE_BLKZONED 0x0002 143 #define F2FS_FEATURE_ATOMIC_WRITE 0x0004 144 #define F2FS_FEATURE_EXTRA_ATTR 0x0008 145 #define F2FS_FEATURE_PRJQUOTA 0x0010 146 #define F2FS_FEATURE_INODE_CHKSUM 0x0020 147 #define F2FS_FEATURE_FLEXIBLE_INLINE_XATTR 0x0040 148 #define F2FS_FEATURE_QUOTA_INO 0x0080 149 #define F2FS_FEATURE_INODE_CRTIME 0x0100 150 #define F2FS_FEATURE_LOST_FOUND 0x0200 151 #define F2FS_FEATURE_VERITY 0x0400 /* reserved */ 152 #define F2FS_FEATURE_SB_CHKSUM 0x0800 153 154 #define __F2FS_HAS_FEATURE(raw_super, mask) \ 155 ((raw_super->feature & cpu_to_le32(mask)) != 0) 156 #define F2FS_HAS_FEATURE(sbi, mask) __F2FS_HAS_FEATURE(sbi->raw_super, mask) 157 #define F2FS_SET_FEATURE(sbi, mask) \ 158 (sbi->raw_super->feature |= cpu_to_le32(mask)) 159 #define F2FS_CLEAR_FEATURE(sbi, mask) \ 160 (sbi->raw_super->feature &= ~cpu_to_le32(mask)) 161 162 /* 163 * Default values for user and/or group using reserved blocks 164 */ 165 #define F2FS_DEF_RESUID 0 166 #define F2FS_DEF_RESGID 0 167 168 /* 169 * For checkpoint manager 170 */ 171 enum { 172 NAT_BITMAP, 173 SIT_BITMAP 174 }; 175 176 #define CP_UMOUNT 0x00000001 177 #define CP_FASTBOOT 0x00000002 178 #define CP_SYNC 0x00000004 179 #define CP_RECOVERY 0x00000008 180 #define CP_DISCARD 0x00000010 181 #define CP_TRIMMED 0x00000020 182 #define CP_PAUSE 0x00000040 183 184 #define MAX_DISCARD_BLOCKS(sbi) BLKS_PER_SEC(sbi) 185 #define DEF_MAX_DISCARD_REQUEST 8 /* issue 8 discards per round */ 186 #define DEF_MIN_DISCARD_ISSUE_TIME 50 /* 50 ms, if exists */ 187 #define DEF_MID_DISCARD_ISSUE_TIME 500 /* 500 ms, if device busy */ 188 #define DEF_MAX_DISCARD_ISSUE_TIME 60000 /* 60 s, if no candidates */ 189 #define DEF_DISCARD_URGENT_UTIL 80 /* do more discard over 80% */ 190 #define DEF_CP_INTERVAL 60 /* 60 secs */ 191 #define DEF_IDLE_INTERVAL 5 /* 5 secs */ 192 #define DEF_DISABLE_INTERVAL 5 /* 5 secs */ 193 #define DEF_DISABLE_QUICK_INTERVAL 1 /* 1 secs */ 194 #define DEF_UMOUNT_DISCARD_TIMEOUT 5 /* 5 secs */ 195 196 struct cp_control { 197 int reason; 198 __u64 trim_start; 199 __u64 trim_end; 200 __u64 trim_minlen; 201 }; 202 203 /* 204 * indicate meta/data type 205 */ 206 enum { 207 META_CP, 208 META_NAT, 209 META_SIT, 210 META_SSA, 211 META_MAX, 212 META_POR, 213 DATA_GENERIC, 214 META_GENERIC, 215 }; 216 217 /* for the list of ino */ 218 enum { 219 ORPHAN_INO, /* for orphan ino list */ 220 APPEND_INO, /* for append ino list */ 221 UPDATE_INO, /* for update ino list */ 222 TRANS_DIR_INO, /* for trasactions dir ino list */ 223 FLUSH_INO, /* for multiple device flushing */ 224 MAX_INO_ENTRY, /* max. list */ 225 }; 226 227 struct ino_entry { 228 struct list_head list; /* list head */ 229 nid_t ino; /* inode number */ 230 unsigned int dirty_device; /* dirty device bitmap */ 231 }; 232 233 /* for the list of inodes to be GCed */ 234 struct inode_entry { 235 struct list_head list; /* list head */ 236 struct inode *inode; /* vfs inode pointer */ 237 }; 238 239 struct fsync_node_entry { 240 struct list_head list; /* list head */ 241 struct page *page; /* warm node page pointer */ 242 unsigned int seq_id; /* sequence id */ 243 }; 244 245 /* for the bitmap indicate blocks to be discarded */ 246 struct discard_entry { 247 struct list_head list; /* list head */ 248 block_t start_blkaddr; /* start blockaddr of current segment */ 249 unsigned char discard_map[SIT_VBLOCK_MAP_SIZE]; /* segment discard bitmap */ 250 }; 251 252 /* default discard granularity of inner discard thread, unit: block count */ 253 #define DEFAULT_DISCARD_GRANULARITY 16 254 255 /* max discard pend list number */ 256 #define MAX_PLIST_NUM 512 257 #define plist_idx(blk_num) ((blk_num) >= MAX_PLIST_NUM ? \ 258 (MAX_PLIST_NUM - 1) : ((blk_num) - 1)) 259 260 enum { 261 D_PREP, /* initial */ 262 D_PARTIAL, /* partially submitted */ 263 D_SUBMIT, /* all submitted */ 264 D_DONE, /* finished */ 265 }; 266 267 struct discard_info { 268 block_t lstart; /* logical start address */ 269 block_t len; /* length */ 270 block_t start; /* actual start address in dev */ 271 }; 272 273 struct discard_cmd { 274 struct rb_node rb_node; /* rb node located in rb-tree */ 275 union { 276 struct { 277 block_t lstart; /* logical start address */ 278 block_t len; /* length */ 279 block_t start; /* actual start address in dev */ 280 }; 281 struct discard_info di; /* discard info */ 282 283 }; 284 struct list_head list; /* command list */ 285 struct completion wait; /* compleation */ 286 struct block_device *bdev; /* bdev */ 287 unsigned short ref; /* reference count */ 288 unsigned char state; /* state */ 289 unsigned char queued; /* queued discard */ 290 int error; /* bio error */ 291 spinlock_t lock; /* for state/bio_ref updating */ 292 unsigned short bio_ref; /* bio reference count */ 293 }; 294 295 enum { 296 DPOLICY_BG, 297 DPOLICY_FORCE, 298 DPOLICY_FSTRIM, 299 DPOLICY_UMOUNT, 300 MAX_DPOLICY, 301 }; 302 303 struct discard_policy { 304 int type; /* type of discard */ 305 unsigned int min_interval; /* used for candidates exist */ 306 unsigned int mid_interval; /* used for device busy */ 307 unsigned int max_interval; /* used for candidates not exist */ 308 unsigned int max_requests; /* # of discards issued per round */ 309 unsigned int io_aware_gran; /* minimum granularity discard not be aware of I/O */ 310 bool io_aware; /* issue discard in idle time */ 311 bool sync; /* submit discard with REQ_SYNC flag */ 312 bool ordered; /* issue discard by lba order */ 313 unsigned int granularity; /* discard granularity */ 314 int timeout; /* discard timeout for put_super */ 315 }; 316 317 struct discard_cmd_control { 318 struct task_struct *f2fs_issue_discard; /* discard thread */ 319 struct list_head entry_list; /* 4KB discard entry list */ 320 struct list_head pend_list[MAX_PLIST_NUM];/* store pending entries */ 321 struct list_head wait_list; /* store on-flushing entries */ 322 struct list_head fstrim_list; /* in-flight discard from fstrim */ 323 wait_queue_head_t discard_wait_queue; /* waiting queue for wake-up */ 324 unsigned int discard_wake; /* to wake up discard thread */ 325 struct mutex cmd_lock; 326 unsigned int nr_discards; /* # of discards in the list */ 327 unsigned int max_discards; /* max. discards to be issued */ 328 unsigned int discard_granularity; /* discard granularity */ 329 unsigned int undiscard_blks; /* # of undiscard blocks */ 330 unsigned int next_pos; /* next discard position */ 331 atomic_t issued_discard; /* # of issued discard */ 332 atomic_t queued_discard; /* # of queued discard */ 333 atomic_t discard_cmd_cnt; /* # of cached cmd count */ 334 struct rb_root_cached root; /* root of discard rb-tree */ 335 bool rbtree_check; /* config for consistence check */ 336 }; 337 338 /* for the list of fsync inodes, used only during recovery */ 339 struct fsync_inode_entry { 340 struct list_head list; /* list head */ 341 struct inode *inode; /* vfs inode pointer */ 342 block_t blkaddr; /* block address locating the last fsync */ 343 block_t last_dentry; /* block address locating the last dentry */ 344 }; 345 346 #define nats_in_cursum(jnl) (le16_to_cpu((jnl)->n_nats)) 347 #define sits_in_cursum(jnl) (le16_to_cpu((jnl)->n_sits)) 348 349 #define nat_in_journal(jnl, i) ((jnl)->nat_j.entries[i].ne) 350 #define nid_in_journal(jnl, i) ((jnl)->nat_j.entries[i].nid) 351 #define sit_in_journal(jnl, i) ((jnl)->sit_j.entries[i].se) 352 #define segno_in_journal(jnl, i) ((jnl)->sit_j.entries[i].segno) 353 354 #define MAX_NAT_JENTRIES(jnl) (NAT_JOURNAL_ENTRIES - nats_in_cursum(jnl)) 355 #define MAX_SIT_JENTRIES(jnl) (SIT_JOURNAL_ENTRIES - sits_in_cursum(jnl)) 356 357 static inline int update_nats_in_cursum(struct f2fs_journal *journal, int i) 358 { 359 int before = nats_in_cursum(journal); 360 361 journal->n_nats = cpu_to_le16(before + i); 362 return before; 363 } 364 365 static inline int update_sits_in_cursum(struct f2fs_journal *journal, int i) 366 { 367 int before = sits_in_cursum(journal); 368 369 journal->n_sits = cpu_to_le16(before + i); 370 return before; 371 } 372 373 static inline bool __has_cursum_space(struct f2fs_journal *journal, 374 int size, int type) 375 { 376 if (type == NAT_JOURNAL) 377 return size <= MAX_NAT_JENTRIES(journal); 378 return size <= MAX_SIT_JENTRIES(journal); 379 } 380 381 /* 382 * ioctl commands 383 */ 384 #define F2FS_IOC_GETFLAGS FS_IOC_GETFLAGS 385 #define F2FS_IOC_SETFLAGS FS_IOC_SETFLAGS 386 #define F2FS_IOC_GETVERSION FS_IOC_GETVERSION 387 388 #define F2FS_IOCTL_MAGIC 0xf5 389 #define F2FS_IOC_START_ATOMIC_WRITE _IO(F2FS_IOCTL_MAGIC, 1) 390 #define F2FS_IOC_COMMIT_ATOMIC_WRITE _IO(F2FS_IOCTL_MAGIC, 2) 391 #define F2FS_IOC_START_VOLATILE_WRITE _IO(F2FS_IOCTL_MAGIC, 3) 392 #define F2FS_IOC_RELEASE_VOLATILE_WRITE _IO(F2FS_IOCTL_MAGIC, 4) 393 #define F2FS_IOC_ABORT_VOLATILE_WRITE _IO(F2FS_IOCTL_MAGIC, 5) 394 #define F2FS_IOC_GARBAGE_COLLECT _IOW(F2FS_IOCTL_MAGIC, 6, __u32) 395 #define F2FS_IOC_WRITE_CHECKPOINT _IO(F2FS_IOCTL_MAGIC, 7) 396 #define F2FS_IOC_DEFRAGMENT _IOWR(F2FS_IOCTL_MAGIC, 8, \ 397 struct f2fs_defragment) 398 #define F2FS_IOC_MOVE_RANGE _IOWR(F2FS_IOCTL_MAGIC, 9, \ 399 struct f2fs_move_range) 400 #define F2FS_IOC_FLUSH_DEVICE _IOW(F2FS_IOCTL_MAGIC, 10, \ 401 struct f2fs_flush_device) 402 #define F2FS_IOC_GARBAGE_COLLECT_RANGE _IOW(F2FS_IOCTL_MAGIC, 11, \ 403 struct f2fs_gc_range) 404 #define F2FS_IOC_GET_FEATURES _IOR(F2FS_IOCTL_MAGIC, 12, __u32) 405 #define F2FS_IOC_SET_PIN_FILE _IOW(F2FS_IOCTL_MAGIC, 13, __u32) 406 #define F2FS_IOC_GET_PIN_FILE _IOR(F2FS_IOCTL_MAGIC, 14, __u32) 407 #define F2FS_IOC_PRECACHE_EXTENTS _IO(F2FS_IOCTL_MAGIC, 15) 408 409 #define F2FS_IOC_SET_ENCRYPTION_POLICY FS_IOC_SET_ENCRYPTION_POLICY 410 #define F2FS_IOC_GET_ENCRYPTION_POLICY FS_IOC_GET_ENCRYPTION_POLICY 411 #define F2FS_IOC_GET_ENCRYPTION_PWSALT FS_IOC_GET_ENCRYPTION_PWSALT 412 413 /* 414 * should be same as XFS_IOC_GOINGDOWN. 415 * Flags for going down operation used by FS_IOC_GOINGDOWN 416 */ 417 #define F2FS_IOC_SHUTDOWN _IOR('X', 125, __u32) /* Shutdown */ 418 #define F2FS_GOING_DOWN_FULLSYNC 0x0 /* going down with full sync */ 419 #define F2FS_GOING_DOWN_METASYNC 0x1 /* going down with metadata */ 420 #define F2FS_GOING_DOWN_NOSYNC 0x2 /* going down */ 421 #define F2FS_GOING_DOWN_METAFLUSH 0x3 /* going down with meta flush */ 422 #define F2FS_GOING_DOWN_NEED_FSCK 0x4 /* going down to trigger fsck */ 423 424 #if defined(__KERNEL__) && defined(CONFIG_COMPAT) 425 /* 426 * ioctl commands in 32 bit emulation 427 */ 428 #define F2FS_IOC32_GETFLAGS FS_IOC32_GETFLAGS 429 #define F2FS_IOC32_SETFLAGS FS_IOC32_SETFLAGS 430 #define F2FS_IOC32_GETVERSION FS_IOC32_GETVERSION 431 #endif 432 433 #define F2FS_IOC_FSGETXATTR FS_IOC_FSGETXATTR 434 #define F2FS_IOC_FSSETXATTR FS_IOC_FSSETXATTR 435 436 struct f2fs_gc_range { 437 u32 sync; 438 u64 start; 439 u64 len; 440 }; 441 442 struct f2fs_defragment { 443 u64 start; 444 u64 len; 445 }; 446 447 struct f2fs_move_range { 448 u32 dst_fd; /* destination fd */ 449 u64 pos_in; /* start position in src_fd */ 450 u64 pos_out; /* start position in dst_fd */ 451 u64 len; /* size to move */ 452 }; 453 454 struct f2fs_flush_device { 455 u32 dev_num; /* device number to flush */ 456 u32 segments; /* # of segments to flush */ 457 }; 458 459 /* for inline stuff */ 460 #define DEF_INLINE_RESERVED_SIZE 1 461 static inline int get_extra_isize(struct inode *inode); 462 static inline int get_inline_xattr_addrs(struct inode *inode); 463 #define MAX_INLINE_DATA(inode) (sizeof(__le32) * \ 464 (CUR_ADDRS_PER_INODE(inode) - \ 465 get_inline_xattr_addrs(inode) - \ 466 DEF_INLINE_RESERVED_SIZE)) 467 468 /* for inline dir */ 469 #define NR_INLINE_DENTRY(inode) (MAX_INLINE_DATA(inode) * BITS_PER_BYTE / \ 470 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \ 471 BITS_PER_BYTE + 1)) 472 #define INLINE_DENTRY_BITMAP_SIZE(inode) ((NR_INLINE_DENTRY(inode) + \ 473 BITS_PER_BYTE - 1) / BITS_PER_BYTE) 474 #define INLINE_RESERVED_SIZE(inode) (MAX_INLINE_DATA(inode) - \ 475 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \ 476 NR_INLINE_DENTRY(inode) + \ 477 INLINE_DENTRY_BITMAP_SIZE(inode))) 478 479 /* 480 * For INODE and NODE manager 481 */ 482 /* for directory operations */ 483 struct f2fs_dentry_ptr { 484 struct inode *inode; 485 void *bitmap; 486 struct f2fs_dir_entry *dentry; 487 __u8 (*filename)[F2FS_SLOT_LEN]; 488 int max; 489 int nr_bitmap; 490 }; 491 492 static inline void make_dentry_ptr_block(struct inode *inode, 493 struct f2fs_dentry_ptr *d, struct f2fs_dentry_block *t) 494 { 495 d->inode = inode; 496 d->max = NR_DENTRY_IN_BLOCK; 497 d->nr_bitmap = SIZE_OF_DENTRY_BITMAP; 498 d->bitmap = t->dentry_bitmap; 499 d->dentry = t->dentry; 500 d->filename = t->filename; 501 } 502 503 static inline void make_dentry_ptr_inline(struct inode *inode, 504 struct f2fs_dentry_ptr *d, void *t) 505 { 506 int entry_cnt = NR_INLINE_DENTRY(inode); 507 int bitmap_size = INLINE_DENTRY_BITMAP_SIZE(inode); 508 int reserved_size = INLINE_RESERVED_SIZE(inode); 509 510 d->inode = inode; 511 d->max = entry_cnt; 512 d->nr_bitmap = bitmap_size; 513 d->bitmap = t; 514 d->dentry = t + bitmap_size + reserved_size; 515 d->filename = t + bitmap_size + reserved_size + 516 SIZE_OF_DIR_ENTRY * entry_cnt; 517 } 518 519 /* 520 * XATTR_NODE_OFFSET stores xattrs to one node block per file keeping -1 521 * as its node offset to distinguish from index node blocks. 522 * But some bits are used to mark the node block. 523 */ 524 #define XATTR_NODE_OFFSET ((((unsigned int)-1) << OFFSET_BIT_SHIFT) \ 525 >> OFFSET_BIT_SHIFT) 526 enum { 527 ALLOC_NODE, /* allocate a new node page if needed */ 528 LOOKUP_NODE, /* look up a node without readahead */ 529 LOOKUP_NODE_RA, /* 530 * look up a node with readahead called 531 * by get_data_block. 532 */ 533 }; 534 535 #define DEFAULT_RETRY_IO_COUNT 8 /* maximum retry read IO count */ 536 537 /* maximum retry quota flush count */ 538 #define DEFAULT_RETRY_QUOTA_FLUSH_COUNT 8 539 540 #define F2FS_LINK_MAX 0xffffffff /* maximum link count per file */ 541 542 #define MAX_DIR_RA_PAGES 4 /* maximum ra pages of dir */ 543 544 /* for in-memory extent cache entry */ 545 #define F2FS_MIN_EXTENT_LEN 64 /* minimum extent length */ 546 547 /* number of extent info in extent cache we try to shrink */ 548 #define EXTENT_CACHE_SHRINK_NUMBER 128 549 550 struct rb_entry { 551 struct rb_node rb_node; /* rb node located in rb-tree */ 552 unsigned int ofs; /* start offset of the entry */ 553 unsigned int len; /* length of the entry */ 554 }; 555 556 struct extent_info { 557 unsigned int fofs; /* start offset in a file */ 558 unsigned int len; /* length of the extent */ 559 u32 blk; /* start block address of the extent */ 560 }; 561 562 struct extent_node { 563 struct rb_node rb_node; /* rb node located in rb-tree */ 564 struct extent_info ei; /* extent info */ 565 struct list_head list; /* node in global extent list of sbi */ 566 struct extent_tree *et; /* extent tree pointer */ 567 }; 568 569 struct extent_tree { 570 nid_t ino; /* inode number */ 571 struct rb_root_cached root; /* root of extent info rb-tree */ 572 struct extent_node *cached_en; /* recently accessed extent node */ 573 struct extent_info largest; /* largested extent info */ 574 struct list_head list; /* to be used by sbi->zombie_list */ 575 rwlock_t lock; /* protect extent info rb-tree */ 576 atomic_t node_cnt; /* # of extent node in rb-tree*/ 577 bool largest_updated; /* largest extent updated */ 578 }; 579 580 /* 581 * This structure is taken from ext4_map_blocks. 582 * 583 * Note that, however, f2fs uses NEW and MAPPED flags for f2fs_map_blocks(). 584 */ 585 #define F2FS_MAP_NEW (1 << BH_New) 586 #define F2FS_MAP_MAPPED (1 << BH_Mapped) 587 #define F2FS_MAP_UNWRITTEN (1 << BH_Unwritten) 588 #define F2FS_MAP_FLAGS (F2FS_MAP_NEW | F2FS_MAP_MAPPED |\ 589 F2FS_MAP_UNWRITTEN) 590 591 struct f2fs_map_blocks { 592 block_t m_pblk; 593 block_t m_lblk; 594 unsigned int m_len; 595 unsigned int m_flags; 596 pgoff_t *m_next_pgofs; /* point next possible non-hole pgofs */ 597 pgoff_t *m_next_extent; /* point to next possible extent */ 598 int m_seg_type; 599 bool m_may_create; /* indicate it is from write path */ 600 }; 601 602 /* for flag in get_data_block */ 603 enum { 604 F2FS_GET_BLOCK_DEFAULT, 605 F2FS_GET_BLOCK_FIEMAP, 606 F2FS_GET_BLOCK_BMAP, 607 F2FS_GET_BLOCK_DIO, 608 F2FS_GET_BLOCK_PRE_DIO, 609 F2FS_GET_BLOCK_PRE_AIO, 610 F2FS_GET_BLOCK_PRECACHE, 611 }; 612 613 /* 614 * i_advise uses FADVISE_XXX_BIT. We can add additional hints later. 615 */ 616 #define FADVISE_COLD_BIT 0x01 617 #define FADVISE_LOST_PINO_BIT 0x02 618 #define FADVISE_ENCRYPT_BIT 0x04 619 #define FADVISE_ENC_NAME_BIT 0x08 620 #define FADVISE_KEEP_SIZE_BIT 0x10 621 #define FADVISE_HOT_BIT 0x20 622 #define FADVISE_VERITY_BIT 0x40 /* reserved */ 623 624 #define FADVISE_MODIFIABLE_BITS (FADVISE_COLD_BIT | FADVISE_HOT_BIT) 625 626 #define file_is_cold(inode) is_file(inode, FADVISE_COLD_BIT) 627 #define file_wrong_pino(inode) is_file(inode, FADVISE_LOST_PINO_BIT) 628 #define file_set_cold(inode) set_file(inode, FADVISE_COLD_BIT) 629 #define file_lost_pino(inode) set_file(inode, FADVISE_LOST_PINO_BIT) 630 #define file_clear_cold(inode) clear_file(inode, FADVISE_COLD_BIT) 631 #define file_got_pino(inode) clear_file(inode, FADVISE_LOST_PINO_BIT) 632 #define file_is_encrypt(inode) is_file(inode, FADVISE_ENCRYPT_BIT) 633 #define file_set_encrypt(inode) set_file(inode, FADVISE_ENCRYPT_BIT) 634 #define file_clear_encrypt(inode) clear_file(inode, FADVISE_ENCRYPT_BIT) 635 #define file_enc_name(inode) is_file(inode, FADVISE_ENC_NAME_BIT) 636 #define file_set_enc_name(inode) set_file(inode, FADVISE_ENC_NAME_BIT) 637 #define file_keep_isize(inode) is_file(inode, FADVISE_KEEP_SIZE_BIT) 638 #define file_set_keep_isize(inode) set_file(inode, FADVISE_KEEP_SIZE_BIT) 639 #define file_is_hot(inode) is_file(inode, FADVISE_HOT_BIT) 640 #define file_set_hot(inode) set_file(inode, FADVISE_HOT_BIT) 641 #define file_clear_hot(inode) clear_file(inode, FADVISE_HOT_BIT) 642 643 #define DEF_DIR_LEVEL 0 644 645 enum { 646 GC_FAILURE_PIN, 647 GC_FAILURE_ATOMIC, 648 MAX_GC_FAILURE 649 }; 650 651 struct f2fs_inode_info { 652 struct inode vfs_inode; /* serve a vfs inode */ 653 unsigned long i_flags; /* keep an inode flags for ioctl */ 654 unsigned char i_advise; /* use to give file attribute hints */ 655 unsigned char i_dir_level; /* use for dentry level for large dir */ 656 unsigned int i_current_depth; /* only for directory depth */ 657 /* for gc failure statistic */ 658 unsigned int i_gc_failures[MAX_GC_FAILURE]; 659 unsigned int i_pino; /* parent inode number */ 660 umode_t i_acl_mode; /* keep file acl mode temporarily */ 661 662 /* Use below internally in f2fs*/ 663 unsigned long flags; /* use to pass per-file flags */ 664 struct rw_semaphore i_sem; /* protect fi info */ 665 atomic_t dirty_pages; /* # of dirty pages */ 666 f2fs_hash_t chash; /* hash value of given file name */ 667 unsigned int clevel; /* maximum level of given file name */ 668 struct task_struct *task; /* lookup and create consistency */ 669 struct task_struct *cp_task; /* separate cp/wb IO stats*/ 670 nid_t i_xattr_nid; /* node id that contains xattrs */ 671 loff_t last_disk_size; /* lastly written file size */ 672 673 #ifdef CONFIG_QUOTA 674 struct dquot *i_dquot[MAXQUOTAS]; 675 676 /* quota space reservation, managed internally by quota code */ 677 qsize_t i_reserved_quota; 678 #endif 679 struct list_head dirty_list; /* dirty list for dirs and files */ 680 struct list_head gdirty_list; /* linked in global dirty list */ 681 struct list_head inmem_ilist; /* list for inmem inodes */ 682 struct list_head inmem_pages; /* inmemory pages managed by f2fs */ 683 struct task_struct *inmem_task; /* store inmemory task */ 684 struct mutex inmem_lock; /* lock for inmemory pages */ 685 struct extent_tree *extent_tree; /* cached extent_tree entry */ 686 687 /* avoid racing between foreground op and gc */ 688 struct rw_semaphore i_gc_rwsem[2]; 689 struct rw_semaphore i_mmap_sem; 690 struct rw_semaphore i_xattr_sem; /* avoid racing between reading and changing EAs */ 691 692 int i_extra_isize; /* size of extra space located in i_addr */ 693 kprojid_t i_projid; /* id for project quota */ 694 int i_inline_xattr_size; /* inline xattr size */ 695 struct timespec64 i_crtime; /* inode creation time */ 696 struct timespec64 i_disk_time[4];/* inode disk times */ 697 }; 698 699 static inline void get_extent_info(struct extent_info *ext, 700 struct f2fs_extent *i_ext) 701 { 702 ext->fofs = le32_to_cpu(i_ext->fofs); 703 ext->blk = le32_to_cpu(i_ext->blk); 704 ext->len = le32_to_cpu(i_ext->len); 705 } 706 707 static inline void set_raw_extent(struct extent_info *ext, 708 struct f2fs_extent *i_ext) 709 { 710 i_ext->fofs = cpu_to_le32(ext->fofs); 711 i_ext->blk = cpu_to_le32(ext->blk); 712 i_ext->len = cpu_to_le32(ext->len); 713 } 714 715 static inline void set_extent_info(struct extent_info *ei, unsigned int fofs, 716 u32 blk, unsigned int len) 717 { 718 ei->fofs = fofs; 719 ei->blk = blk; 720 ei->len = len; 721 } 722 723 static inline bool __is_discard_mergeable(struct discard_info *back, 724 struct discard_info *front, unsigned int max_len) 725 { 726 return (back->lstart + back->len == front->lstart) && 727 (back->len + front->len <= max_len); 728 } 729 730 static inline bool __is_discard_back_mergeable(struct discard_info *cur, 731 struct discard_info *back, unsigned int max_len) 732 { 733 return __is_discard_mergeable(back, cur, max_len); 734 } 735 736 static inline bool __is_discard_front_mergeable(struct discard_info *cur, 737 struct discard_info *front, unsigned int max_len) 738 { 739 return __is_discard_mergeable(cur, front, max_len); 740 } 741 742 static inline bool __is_extent_mergeable(struct extent_info *back, 743 struct extent_info *front) 744 { 745 return (back->fofs + back->len == front->fofs && 746 back->blk + back->len == front->blk); 747 } 748 749 static inline bool __is_back_mergeable(struct extent_info *cur, 750 struct extent_info *back) 751 { 752 return __is_extent_mergeable(back, cur); 753 } 754 755 static inline bool __is_front_mergeable(struct extent_info *cur, 756 struct extent_info *front) 757 { 758 return __is_extent_mergeable(cur, front); 759 } 760 761 extern void f2fs_mark_inode_dirty_sync(struct inode *inode, bool sync); 762 static inline void __try_update_largest_extent(struct extent_tree *et, 763 struct extent_node *en) 764 { 765 if (en->ei.len > et->largest.len) { 766 et->largest = en->ei; 767 et->largest_updated = true; 768 } 769 } 770 771 /* 772 * For free nid management 773 */ 774 enum nid_state { 775 FREE_NID, /* newly added to free nid list */ 776 PREALLOC_NID, /* it is preallocated */ 777 MAX_NID_STATE, 778 }; 779 780 struct f2fs_nm_info { 781 block_t nat_blkaddr; /* base disk address of NAT */ 782 nid_t max_nid; /* maximum possible node ids */ 783 nid_t available_nids; /* # of available node ids */ 784 nid_t next_scan_nid; /* the next nid to be scanned */ 785 unsigned int ram_thresh; /* control the memory footprint */ 786 unsigned int ra_nid_pages; /* # of nid pages to be readaheaded */ 787 unsigned int dirty_nats_ratio; /* control dirty nats ratio threshold */ 788 789 /* NAT cache management */ 790 struct radix_tree_root nat_root;/* root of the nat entry cache */ 791 struct radix_tree_root nat_set_root;/* root of the nat set cache */ 792 struct rw_semaphore nat_tree_lock; /* protect nat_tree_lock */ 793 struct list_head nat_entries; /* cached nat entry list (clean) */ 794 spinlock_t nat_list_lock; /* protect clean nat entry list */ 795 unsigned int nat_cnt; /* the # of cached nat entries */ 796 unsigned int dirty_nat_cnt; /* total num of nat entries in set */ 797 unsigned int nat_blocks; /* # of nat blocks */ 798 799 /* free node ids management */ 800 struct radix_tree_root free_nid_root;/* root of the free_nid cache */ 801 struct list_head free_nid_list; /* list for free nids excluding preallocated nids */ 802 unsigned int nid_cnt[MAX_NID_STATE]; /* the number of free node id */ 803 spinlock_t nid_list_lock; /* protect nid lists ops */ 804 struct mutex build_lock; /* lock for build free nids */ 805 unsigned char **free_nid_bitmap; 806 unsigned char *nat_block_bitmap; 807 unsigned short *free_nid_count; /* free nid count of NAT block */ 808 809 /* for checkpoint */ 810 char *nat_bitmap; /* NAT bitmap pointer */ 811 812 unsigned int nat_bits_blocks; /* # of nat bits blocks */ 813 unsigned char *nat_bits; /* NAT bits blocks */ 814 unsigned char *full_nat_bits; /* full NAT pages */ 815 unsigned char *empty_nat_bits; /* empty NAT pages */ 816 #ifdef CONFIG_F2FS_CHECK_FS 817 char *nat_bitmap_mir; /* NAT bitmap mirror */ 818 #endif 819 int bitmap_size; /* bitmap size */ 820 }; 821 822 /* 823 * this structure is used as one of function parameters. 824 * all the information are dedicated to a given direct node block determined 825 * by the data offset in a file. 826 */ 827 struct dnode_of_data { 828 struct inode *inode; /* vfs inode pointer */ 829 struct page *inode_page; /* its inode page, NULL is possible */ 830 struct page *node_page; /* cached direct node page */ 831 nid_t nid; /* node id of the direct node block */ 832 unsigned int ofs_in_node; /* data offset in the node page */ 833 bool inode_page_locked; /* inode page is locked or not */ 834 bool node_changed; /* is node block changed */ 835 char cur_level; /* level of hole node page */ 836 char max_level; /* level of current page located */ 837 block_t data_blkaddr; /* block address of the node block */ 838 }; 839 840 static inline void set_new_dnode(struct dnode_of_data *dn, struct inode *inode, 841 struct page *ipage, struct page *npage, nid_t nid) 842 { 843 memset(dn, 0, sizeof(*dn)); 844 dn->inode = inode; 845 dn->inode_page = ipage; 846 dn->node_page = npage; 847 dn->nid = nid; 848 } 849 850 /* 851 * For SIT manager 852 * 853 * By default, there are 6 active log areas across the whole main area. 854 * When considering hot and cold data separation to reduce cleaning overhead, 855 * we split 3 for data logs and 3 for node logs as hot, warm, and cold types, 856 * respectively. 857 * In the current design, you should not change the numbers intentionally. 858 * Instead, as a mount option such as active_logs=x, you can use 2, 4, and 6 859 * logs individually according to the underlying devices. (default: 6) 860 * Just in case, on-disk layout covers maximum 16 logs that consist of 8 for 861 * data and 8 for node logs. 862 */ 863 #define NR_CURSEG_DATA_TYPE (3) 864 #define NR_CURSEG_NODE_TYPE (3) 865 #define NR_CURSEG_TYPE (NR_CURSEG_DATA_TYPE + NR_CURSEG_NODE_TYPE) 866 867 enum { 868 CURSEG_HOT_DATA = 0, /* directory entry blocks */ 869 CURSEG_WARM_DATA, /* data blocks */ 870 CURSEG_COLD_DATA, /* multimedia or GCed data blocks */ 871 CURSEG_HOT_NODE, /* direct node blocks of directory files */ 872 CURSEG_WARM_NODE, /* direct node blocks of normal files */ 873 CURSEG_COLD_NODE, /* indirect node blocks */ 874 NO_CHECK_TYPE, 875 }; 876 877 struct flush_cmd { 878 struct completion wait; 879 struct llist_node llnode; 880 nid_t ino; 881 int ret; 882 }; 883 884 struct flush_cmd_control { 885 struct task_struct *f2fs_issue_flush; /* flush thread */ 886 wait_queue_head_t flush_wait_queue; /* waiting queue for wake-up */ 887 atomic_t issued_flush; /* # of issued flushes */ 888 atomic_t queued_flush; /* # of queued flushes */ 889 struct llist_head issue_list; /* list for command issue */ 890 struct llist_node *dispatch_list; /* list for command dispatch */ 891 }; 892 893 struct f2fs_sm_info { 894 struct sit_info *sit_info; /* whole segment information */ 895 struct free_segmap_info *free_info; /* free segment information */ 896 struct dirty_seglist_info *dirty_info; /* dirty segment information */ 897 struct curseg_info *curseg_array; /* active segment information */ 898 899 struct rw_semaphore curseg_lock; /* for preventing curseg change */ 900 901 block_t seg0_blkaddr; /* block address of 0'th segment */ 902 block_t main_blkaddr; /* start block address of main area */ 903 block_t ssa_blkaddr; /* start block address of SSA area */ 904 905 unsigned int segment_count; /* total # of segments */ 906 unsigned int main_segments; /* # of segments in main area */ 907 unsigned int reserved_segments; /* # of reserved segments */ 908 unsigned int ovp_segments; /* # of overprovision segments */ 909 910 /* a threshold to reclaim prefree segments */ 911 unsigned int rec_prefree_segments; 912 913 /* for batched trimming */ 914 unsigned int trim_sections; /* # of sections to trim */ 915 916 struct list_head sit_entry_set; /* sit entry set list */ 917 918 unsigned int ipu_policy; /* in-place-update policy */ 919 unsigned int min_ipu_util; /* in-place-update threshold */ 920 unsigned int min_fsync_blocks; /* threshold for fsync */ 921 unsigned int min_seq_blocks; /* threshold for sequential blocks */ 922 unsigned int min_hot_blocks; /* threshold for hot block allocation */ 923 unsigned int min_ssr_sections; /* threshold to trigger SSR allocation */ 924 925 /* for flush command control */ 926 struct flush_cmd_control *fcc_info; 927 928 /* for discard command control */ 929 struct discard_cmd_control *dcc_info; 930 }; 931 932 /* 933 * For superblock 934 */ 935 /* 936 * COUNT_TYPE for monitoring 937 * 938 * f2fs monitors the number of several block types such as on-writeback, 939 * dirty dentry blocks, dirty node blocks, and dirty meta blocks. 940 */ 941 #define WB_DATA_TYPE(p) (__is_cp_guaranteed(p) ? F2FS_WB_CP_DATA : F2FS_WB_DATA) 942 enum count_type { 943 F2FS_DIRTY_DENTS, 944 F2FS_DIRTY_DATA, 945 F2FS_DIRTY_QDATA, 946 F2FS_DIRTY_NODES, 947 F2FS_DIRTY_META, 948 F2FS_INMEM_PAGES, 949 F2FS_DIRTY_IMETA, 950 F2FS_WB_CP_DATA, 951 F2FS_WB_DATA, 952 F2FS_RD_DATA, 953 F2FS_RD_NODE, 954 F2FS_RD_META, 955 F2FS_DIO_WRITE, 956 F2FS_DIO_READ, 957 NR_COUNT_TYPE, 958 }; 959 960 /* 961 * The below are the page types of bios used in submit_bio(). 962 * The available types are: 963 * DATA User data pages. It operates as async mode. 964 * NODE Node pages. It operates as async mode. 965 * META FS metadata pages such as SIT, NAT, CP. 966 * NR_PAGE_TYPE The number of page types. 967 * META_FLUSH Make sure the previous pages are written 968 * with waiting the bio's completion 969 * ... Only can be used with META. 970 */ 971 #define PAGE_TYPE_OF_BIO(type) ((type) > META ? META : (type)) 972 enum page_type { 973 DATA, 974 NODE, 975 META, 976 NR_PAGE_TYPE, 977 META_FLUSH, 978 INMEM, /* the below types are used by tracepoints only. */ 979 INMEM_DROP, 980 INMEM_INVALIDATE, 981 INMEM_REVOKE, 982 IPU, 983 OPU, 984 }; 985 986 enum temp_type { 987 HOT = 0, /* must be zero for meta bio */ 988 WARM, 989 COLD, 990 NR_TEMP_TYPE, 991 }; 992 993 enum need_lock_type { 994 LOCK_REQ = 0, 995 LOCK_DONE, 996 LOCK_RETRY, 997 }; 998 999 enum cp_reason_type { 1000 CP_NO_NEEDED, 1001 CP_NON_REGULAR, 1002 CP_HARDLINK, 1003 CP_SB_NEED_CP, 1004 CP_WRONG_PINO, 1005 CP_NO_SPC_ROLL, 1006 CP_NODE_NEED_CP, 1007 CP_FASTBOOT_MODE, 1008 CP_SPEC_LOG_NUM, 1009 CP_RECOVER_DIR, 1010 }; 1011 1012 enum iostat_type { 1013 APP_DIRECT_IO, /* app direct IOs */ 1014 APP_BUFFERED_IO, /* app buffered IOs */ 1015 APP_WRITE_IO, /* app write IOs */ 1016 APP_MAPPED_IO, /* app mapped IOs */ 1017 FS_DATA_IO, /* data IOs from kworker/fsync/reclaimer */ 1018 FS_NODE_IO, /* node IOs from kworker/fsync/reclaimer */ 1019 FS_META_IO, /* meta IOs from kworker/reclaimer */ 1020 FS_GC_DATA_IO, /* data IOs from forground gc */ 1021 FS_GC_NODE_IO, /* node IOs from forground gc */ 1022 FS_CP_DATA_IO, /* data IOs from checkpoint */ 1023 FS_CP_NODE_IO, /* node IOs from checkpoint */ 1024 FS_CP_META_IO, /* meta IOs from checkpoint */ 1025 FS_DISCARD, /* discard */ 1026 NR_IO_TYPE, 1027 }; 1028 1029 struct f2fs_io_info { 1030 struct f2fs_sb_info *sbi; /* f2fs_sb_info pointer */ 1031 nid_t ino; /* inode number */ 1032 enum page_type type; /* contains DATA/NODE/META/META_FLUSH */ 1033 enum temp_type temp; /* contains HOT/WARM/COLD */ 1034 int op; /* contains REQ_OP_ */ 1035 int op_flags; /* req_flag_bits */ 1036 block_t new_blkaddr; /* new block address to be written */ 1037 block_t old_blkaddr; /* old block address before Cow */ 1038 struct page *page; /* page to be written */ 1039 struct page *encrypted_page; /* encrypted page */ 1040 struct list_head list; /* serialize IOs */ 1041 bool submitted; /* indicate IO submission */ 1042 int need_lock; /* indicate we need to lock cp_rwsem */ 1043 bool in_list; /* indicate fio is in io_list */ 1044 bool is_meta; /* indicate borrow meta inode mapping or not */ 1045 bool retry; /* need to reallocate block address */ 1046 enum iostat_type io_type; /* io type */ 1047 struct writeback_control *io_wbc; /* writeback control */ 1048 unsigned char version; /* version of the node */ 1049 }; 1050 1051 #define is_read_io(rw) ((rw) == READ) 1052 struct f2fs_bio_info { 1053 struct f2fs_sb_info *sbi; /* f2fs superblock */ 1054 struct bio *bio; /* bios to merge */ 1055 sector_t last_block_in_bio; /* last block number */ 1056 struct f2fs_io_info fio; /* store buffered io info. */ 1057 struct rw_semaphore io_rwsem; /* blocking op for bio */ 1058 spinlock_t io_lock; /* serialize DATA/NODE IOs */ 1059 struct list_head io_list; /* track fios */ 1060 }; 1061 1062 #define FDEV(i) (sbi->devs[i]) 1063 #define RDEV(i) (raw_super->devs[i]) 1064 struct f2fs_dev_info { 1065 struct block_device *bdev; 1066 char path[MAX_PATH_LEN]; 1067 unsigned int total_segments; 1068 block_t start_blk; 1069 block_t end_blk; 1070 #ifdef CONFIG_BLK_DEV_ZONED 1071 unsigned int nr_blkz; /* Total number of zones */ 1072 unsigned long *blkz_seq; /* Bitmap indicating sequential zones */ 1073 #endif 1074 }; 1075 1076 enum inode_type { 1077 DIR_INODE, /* for dirty dir inode */ 1078 FILE_INODE, /* for dirty regular/symlink inode */ 1079 DIRTY_META, /* for all dirtied inode metadata */ 1080 ATOMIC_FILE, /* for all atomic files */ 1081 NR_INODE_TYPE, 1082 }; 1083 1084 /* for inner inode cache management */ 1085 struct inode_management { 1086 struct radix_tree_root ino_root; /* ino entry array */ 1087 spinlock_t ino_lock; /* for ino entry lock */ 1088 struct list_head ino_list; /* inode list head */ 1089 unsigned long ino_num; /* number of entries */ 1090 }; 1091 1092 /* For s_flag in struct f2fs_sb_info */ 1093 enum { 1094 SBI_IS_DIRTY, /* dirty flag for checkpoint */ 1095 SBI_IS_CLOSE, /* specify unmounting */ 1096 SBI_NEED_FSCK, /* need fsck.f2fs to fix */ 1097 SBI_POR_DOING, /* recovery is doing or not */ 1098 SBI_NEED_SB_WRITE, /* need to recover superblock */ 1099 SBI_NEED_CP, /* need to checkpoint */ 1100 SBI_IS_SHUTDOWN, /* shutdown by ioctl */ 1101 SBI_IS_RECOVERED, /* recovered orphan/data */ 1102 SBI_CP_DISABLED, /* CP was disabled last mount */ 1103 SBI_CP_DISABLED_QUICK, /* CP was disabled quickly */ 1104 SBI_QUOTA_NEED_FLUSH, /* need to flush quota info in CP */ 1105 SBI_QUOTA_SKIP_FLUSH, /* skip flushing quota in current CP */ 1106 SBI_QUOTA_NEED_REPAIR, /* quota file may be corrupted */ 1107 }; 1108 1109 enum { 1110 CP_TIME, 1111 REQ_TIME, 1112 DISCARD_TIME, 1113 GC_TIME, 1114 DISABLE_TIME, 1115 UMOUNT_DISCARD_TIMEOUT, 1116 MAX_TIME, 1117 }; 1118 1119 enum { 1120 GC_NORMAL, 1121 GC_IDLE_CB, 1122 GC_IDLE_GREEDY, 1123 GC_URGENT, 1124 }; 1125 1126 enum { 1127 WHINT_MODE_OFF, /* not pass down write hints */ 1128 WHINT_MODE_USER, /* try to pass down hints given by users */ 1129 WHINT_MODE_FS, /* pass down hints with F2FS policy */ 1130 }; 1131 1132 enum { 1133 ALLOC_MODE_DEFAULT, /* stay default */ 1134 ALLOC_MODE_REUSE, /* reuse segments as much as possible */ 1135 }; 1136 1137 enum fsync_mode { 1138 FSYNC_MODE_POSIX, /* fsync follows posix semantics */ 1139 FSYNC_MODE_STRICT, /* fsync behaves in line with ext4 */ 1140 FSYNC_MODE_NOBARRIER, /* fsync behaves nobarrier based on posix */ 1141 }; 1142 1143 #ifdef CONFIG_FS_ENCRYPTION 1144 #define DUMMY_ENCRYPTION_ENABLED(sbi) \ 1145 (unlikely(F2FS_OPTION(sbi).test_dummy_encryption)) 1146 #else 1147 #define DUMMY_ENCRYPTION_ENABLED(sbi) (0) 1148 #endif 1149 1150 struct f2fs_sb_info { 1151 struct super_block *sb; /* pointer to VFS super block */ 1152 struct proc_dir_entry *s_proc; /* proc entry */ 1153 struct f2fs_super_block *raw_super; /* raw super block pointer */ 1154 struct rw_semaphore sb_lock; /* lock for raw super block */ 1155 int valid_super_block; /* valid super block no */ 1156 unsigned long s_flag; /* flags for sbi */ 1157 struct mutex writepages; /* mutex for writepages() */ 1158 1159 #ifdef CONFIG_BLK_DEV_ZONED 1160 unsigned int blocks_per_blkz; /* F2FS blocks per zone */ 1161 unsigned int log_blocks_per_blkz; /* log2 F2FS blocks per zone */ 1162 #endif 1163 1164 /* for node-related operations */ 1165 struct f2fs_nm_info *nm_info; /* node manager */ 1166 struct inode *node_inode; /* cache node blocks */ 1167 1168 /* for segment-related operations */ 1169 struct f2fs_sm_info *sm_info; /* segment manager */ 1170 1171 /* for bio operations */ 1172 struct f2fs_bio_info *write_io[NR_PAGE_TYPE]; /* for write bios */ 1173 /* keep migration IO order for LFS mode */ 1174 struct rw_semaphore io_order_lock; 1175 mempool_t *write_io_dummy; /* Dummy pages */ 1176 1177 /* for checkpoint */ 1178 struct f2fs_checkpoint *ckpt; /* raw checkpoint pointer */ 1179 int cur_cp_pack; /* remain current cp pack */ 1180 spinlock_t cp_lock; /* for flag in ckpt */ 1181 struct inode *meta_inode; /* cache meta blocks */ 1182 struct mutex cp_mutex; /* checkpoint procedure lock */ 1183 struct rw_semaphore cp_rwsem; /* blocking FS operations */ 1184 struct rw_semaphore node_write; /* locking node writes */ 1185 struct rw_semaphore node_change; /* locking node change */ 1186 wait_queue_head_t cp_wait; 1187 unsigned long last_time[MAX_TIME]; /* to store time in jiffies */ 1188 long interval_time[MAX_TIME]; /* to store thresholds */ 1189 1190 struct inode_management im[MAX_INO_ENTRY]; /* manage inode cache */ 1191 1192 spinlock_t fsync_node_lock; /* for node entry lock */ 1193 struct list_head fsync_node_list; /* node list head */ 1194 unsigned int fsync_seg_id; /* sequence id */ 1195 unsigned int fsync_node_num; /* number of node entries */ 1196 1197 /* for orphan inode, use 0'th array */ 1198 unsigned int max_orphans; /* max orphan inodes */ 1199 1200 /* for inode management */ 1201 struct list_head inode_list[NR_INODE_TYPE]; /* dirty inode list */ 1202 spinlock_t inode_lock[NR_INODE_TYPE]; /* for dirty inode list lock */ 1203 1204 /* for extent tree cache */ 1205 struct radix_tree_root extent_tree_root;/* cache extent cache entries */ 1206 struct mutex extent_tree_lock; /* locking extent radix tree */ 1207 struct list_head extent_list; /* lru list for shrinker */ 1208 spinlock_t extent_lock; /* locking extent lru list */ 1209 atomic_t total_ext_tree; /* extent tree count */ 1210 struct list_head zombie_list; /* extent zombie tree list */ 1211 atomic_t total_zombie_tree; /* extent zombie tree count */ 1212 atomic_t total_ext_node; /* extent info count */ 1213 1214 /* basic filesystem units */ 1215 unsigned int log_sectors_per_block; /* log2 sectors per block */ 1216 unsigned int log_blocksize; /* log2 block size */ 1217 unsigned int blocksize; /* block size */ 1218 unsigned int root_ino_num; /* root inode number*/ 1219 unsigned int node_ino_num; /* node inode number*/ 1220 unsigned int meta_ino_num; /* meta inode number*/ 1221 unsigned int log_blocks_per_seg; /* log2 blocks per segment */ 1222 unsigned int blocks_per_seg; /* blocks per segment */ 1223 unsigned int segs_per_sec; /* segments per section */ 1224 unsigned int secs_per_zone; /* sections per zone */ 1225 unsigned int total_sections; /* total section count */ 1226 unsigned int total_node_count; /* total node block count */ 1227 unsigned int total_valid_node_count; /* valid node block count */ 1228 loff_t max_file_blocks; /* max block index of file */ 1229 int dir_level; /* directory level */ 1230 int readdir_ra; /* readahead inode in readdir */ 1231 1232 block_t user_block_count; /* # of user blocks */ 1233 block_t total_valid_block_count; /* # of valid blocks */ 1234 block_t discard_blks; /* discard command candidats */ 1235 block_t last_valid_block_count; /* for recovery */ 1236 block_t reserved_blocks; /* configurable reserved blocks */ 1237 block_t current_reserved_blocks; /* current reserved blocks */ 1238 1239 /* Additional tracking for no checkpoint mode */ 1240 block_t unusable_block_count; /* # of blocks saved by last cp */ 1241 1242 unsigned int nquota_files; /* # of quota sysfile */ 1243 1244 /* # of pages, see count_type */ 1245 atomic_t nr_pages[NR_COUNT_TYPE]; 1246 /* # of allocated blocks */ 1247 struct percpu_counter alloc_valid_block_count; 1248 1249 /* writeback control */ 1250 atomic_t wb_sync_req[META]; /* count # of WB_SYNC threads */ 1251 1252 /* valid inode count */ 1253 struct percpu_counter total_valid_inode_count; 1254 1255 struct f2fs_mount_info mount_opt; /* mount options */ 1256 1257 /* for cleaning operations */ 1258 struct mutex gc_mutex; /* mutex for GC */ 1259 struct f2fs_gc_kthread *gc_thread; /* GC thread */ 1260 unsigned int cur_victim_sec; /* current victim section num */ 1261 unsigned int gc_mode; /* current GC state */ 1262 unsigned int next_victim_seg[2]; /* next segment in victim section */ 1263 /* for skip statistic */ 1264 unsigned long long skipped_atomic_files[2]; /* FG_GC and BG_GC */ 1265 unsigned long long skipped_gc_rwsem; /* FG_GC only */ 1266 1267 /* threshold for gc trials on pinned files */ 1268 u64 gc_pin_file_threshold; 1269 1270 /* maximum # of trials to find a victim segment for SSR and GC */ 1271 unsigned int max_victim_search; 1272 /* migration granularity of garbage collection, unit: segment */ 1273 unsigned int migration_granularity; 1274 1275 /* 1276 * for stat information. 1277 * one is for the LFS mode, and the other is for the SSR mode. 1278 */ 1279 #ifdef CONFIG_F2FS_STAT_FS 1280 struct f2fs_stat_info *stat_info; /* FS status information */ 1281 atomic_t meta_count[META_MAX]; /* # of meta blocks */ 1282 unsigned int segment_count[2]; /* # of allocated segments */ 1283 unsigned int block_count[2]; /* # of allocated blocks */ 1284 atomic_t inplace_count; /* # of inplace update */ 1285 atomic64_t total_hit_ext; /* # of lookup extent cache */ 1286 atomic64_t read_hit_rbtree; /* # of hit rbtree extent node */ 1287 atomic64_t read_hit_largest; /* # of hit largest extent node */ 1288 atomic64_t read_hit_cached; /* # of hit cached extent node */ 1289 atomic_t inline_xattr; /* # of inline_xattr inodes */ 1290 atomic_t inline_inode; /* # of inline_data inodes */ 1291 atomic_t inline_dir; /* # of inline_dentry inodes */ 1292 atomic_t aw_cnt; /* # of atomic writes */ 1293 atomic_t vw_cnt; /* # of volatile writes */ 1294 atomic_t max_aw_cnt; /* max # of atomic writes */ 1295 atomic_t max_vw_cnt; /* max # of volatile writes */ 1296 int bg_gc; /* background gc calls */ 1297 unsigned int io_skip_bggc; /* skip background gc for in-flight IO */ 1298 unsigned int other_skip_bggc; /* skip background gc for other reasons */ 1299 unsigned int ndirty_inode[NR_INODE_TYPE]; /* # of dirty inodes */ 1300 #endif 1301 spinlock_t stat_lock; /* lock for stat operations */ 1302 1303 /* For app/fs IO statistics */ 1304 spinlock_t iostat_lock; 1305 unsigned long long write_iostat[NR_IO_TYPE]; 1306 bool iostat_enable; 1307 1308 /* For sysfs suppport */ 1309 struct kobject s_kobj; 1310 struct completion s_kobj_unregister; 1311 1312 /* For shrinker support */ 1313 struct list_head s_list; 1314 int s_ndevs; /* number of devices */ 1315 struct f2fs_dev_info *devs; /* for device list */ 1316 unsigned int dirty_device; /* for checkpoint data flush */ 1317 spinlock_t dev_lock; /* protect dirty_device */ 1318 struct mutex umount_mutex; 1319 unsigned int shrinker_run_no; 1320 1321 /* For write statistics */ 1322 u64 sectors_written_start; 1323 u64 kbytes_written; 1324 1325 /* Reference to checksum algorithm driver via cryptoapi */ 1326 struct crypto_shash *s_chksum_driver; 1327 1328 /* Precomputed FS UUID checksum for seeding other checksums */ 1329 __u32 s_chksum_seed; 1330 }; 1331 1332 struct f2fs_private_dio { 1333 struct inode *inode; 1334 void *orig_private; 1335 bio_end_io_t *orig_end_io; 1336 bool write; 1337 }; 1338 1339 #ifdef CONFIG_F2FS_FAULT_INJECTION 1340 #define f2fs_show_injection_info(type) \ 1341 printk_ratelimited("%sF2FS-fs : inject %s in %s of %pF\n", \ 1342 KERN_INFO, f2fs_fault_name[type], \ 1343 __func__, __builtin_return_address(0)) 1344 static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type) 1345 { 1346 struct f2fs_fault_info *ffi = &F2FS_OPTION(sbi).fault_info; 1347 1348 if (!ffi->inject_rate) 1349 return false; 1350 1351 if (!IS_FAULT_SET(ffi, type)) 1352 return false; 1353 1354 atomic_inc(&ffi->inject_ops); 1355 if (atomic_read(&ffi->inject_ops) >= ffi->inject_rate) { 1356 atomic_set(&ffi->inject_ops, 0); 1357 return true; 1358 } 1359 return false; 1360 } 1361 #else 1362 #define f2fs_show_injection_info(type) do { } while (0) 1363 static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type) 1364 { 1365 return false; 1366 } 1367 #endif 1368 1369 /* 1370 * Test if the mounted volume is a multi-device volume. 1371 * - For a single regular disk volume, sbi->s_ndevs is 0. 1372 * - For a single zoned disk volume, sbi->s_ndevs is 1. 1373 * - For a multi-device volume, sbi->s_ndevs is always 2 or more. 1374 */ 1375 static inline bool f2fs_is_multi_device(struct f2fs_sb_info *sbi) 1376 { 1377 return sbi->s_ndevs > 1; 1378 } 1379 1380 /* For write statistics. Suppose sector size is 512 bytes, 1381 * and the return value is in kbytes. s is of struct f2fs_sb_info. 1382 */ 1383 #define BD_PART_WRITTEN(s) \ 1384 (((u64)part_stat_read((s)->sb->s_bdev->bd_part, sectors[STAT_WRITE]) - \ 1385 (s)->sectors_written_start) >> 1) 1386 1387 static inline void f2fs_update_time(struct f2fs_sb_info *sbi, int type) 1388 { 1389 unsigned long now = jiffies; 1390 1391 sbi->last_time[type] = now; 1392 1393 /* DISCARD_TIME and GC_TIME are based on REQ_TIME */ 1394 if (type == REQ_TIME) { 1395 sbi->last_time[DISCARD_TIME] = now; 1396 sbi->last_time[GC_TIME] = now; 1397 } 1398 } 1399 1400 static inline bool f2fs_time_over(struct f2fs_sb_info *sbi, int type) 1401 { 1402 unsigned long interval = sbi->interval_time[type] * HZ; 1403 1404 return time_after(jiffies, sbi->last_time[type] + interval); 1405 } 1406 1407 static inline unsigned int f2fs_time_to_wait(struct f2fs_sb_info *sbi, 1408 int type) 1409 { 1410 unsigned long interval = sbi->interval_time[type] * HZ; 1411 unsigned int wait_ms = 0; 1412 long delta; 1413 1414 delta = (sbi->last_time[type] + interval) - jiffies; 1415 if (delta > 0) 1416 wait_ms = jiffies_to_msecs(delta); 1417 1418 return wait_ms; 1419 } 1420 1421 /* 1422 * Inline functions 1423 */ 1424 static inline u32 __f2fs_crc32(struct f2fs_sb_info *sbi, u32 crc, 1425 const void *address, unsigned int length) 1426 { 1427 struct { 1428 struct shash_desc shash; 1429 char ctx[4]; 1430 } desc; 1431 int err; 1432 1433 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver) != sizeof(desc.ctx)); 1434 1435 desc.shash.tfm = sbi->s_chksum_driver; 1436 desc.shash.flags = 0; 1437 *(u32 *)desc.ctx = crc; 1438 1439 err = crypto_shash_update(&desc.shash, address, length); 1440 BUG_ON(err); 1441 1442 return *(u32 *)desc.ctx; 1443 } 1444 1445 static inline u32 f2fs_crc32(struct f2fs_sb_info *sbi, const void *address, 1446 unsigned int length) 1447 { 1448 return __f2fs_crc32(sbi, F2FS_SUPER_MAGIC, address, length); 1449 } 1450 1451 static inline bool f2fs_crc_valid(struct f2fs_sb_info *sbi, __u32 blk_crc, 1452 void *buf, size_t buf_size) 1453 { 1454 return f2fs_crc32(sbi, buf, buf_size) == blk_crc; 1455 } 1456 1457 static inline u32 f2fs_chksum(struct f2fs_sb_info *sbi, u32 crc, 1458 const void *address, unsigned int length) 1459 { 1460 return __f2fs_crc32(sbi, crc, address, length); 1461 } 1462 1463 static inline struct f2fs_inode_info *F2FS_I(struct inode *inode) 1464 { 1465 return container_of(inode, struct f2fs_inode_info, vfs_inode); 1466 } 1467 1468 static inline struct f2fs_sb_info *F2FS_SB(struct super_block *sb) 1469 { 1470 return sb->s_fs_info; 1471 } 1472 1473 static inline struct f2fs_sb_info *F2FS_I_SB(struct inode *inode) 1474 { 1475 return F2FS_SB(inode->i_sb); 1476 } 1477 1478 static inline struct f2fs_sb_info *F2FS_M_SB(struct address_space *mapping) 1479 { 1480 return F2FS_I_SB(mapping->host); 1481 } 1482 1483 static inline struct f2fs_sb_info *F2FS_P_SB(struct page *page) 1484 { 1485 return F2FS_M_SB(page->mapping); 1486 } 1487 1488 static inline struct f2fs_super_block *F2FS_RAW_SUPER(struct f2fs_sb_info *sbi) 1489 { 1490 return (struct f2fs_super_block *)(sbi->raw_super); 1491 } 1492 1493 static inline struct f2fs_checkpoint *F2FS_CKPT(struct f2fs_sb_info *sbi) 1494 { 1495 return (struct f2fs_checkpoint *)(sbi->ckpt); 1496 } 1497 1498 static inline struct f2fs_node *F2FS_NODE(struct page *page) 1499 { 1500 return (struct f2fs_node *)page_address(page); 1501 } 1502 1503 static inline struct f2fs_inode *F2FS_INODE(struct page *page) 1504 { 1505 return &((struct f2fs_node *)page_address(page))->i; 1506 } 1507 1508 static inline struct f2fs_nm_info *NM_I(struct f2fs_sb_info *sbi) 1509 { 1510 return (struct f2fs_nm_info *)(sbi->nm_info); 1511 } 1512 1513 static inline struct f2fs_sm_info *SM_I(struct f2fs_sb_info *sbi) 1514 { 1515 return (struct f2fs_sm_info *)(sbi->sm_info); 1516 } 1517 1518 static inline struct sit_info *SIT_I(struct f2fs_sb_info *sbi) 1519 { 1520 return (struct sit_info *)(SM_I(sbi)->sit_info); 1521 } 1522 1523 static inline struct free_segmap_info *FREE_I(struct f2fs_sb_info *sbi) 1524 { 1525 return (struct free_segmap_info *)(SM_I(sbi)->free_info); 1526 } 1527 1528 static inline struct dirty_seglist_info *DIRTY_I(struct f2fs_sb_info *sbi) 1529 { 1530 return (struct dirty_seglist_info *)(SM_I(sbi)->dirty_info); 1531 } 1532 1533 static inline struct address_space *META_MAPPING(struct f2fs_sb_info *sbi) 1534 { 1535 return sbi->meta_inode->i_mapping; 1536 } 1537 1538 static inline struct address_space *NODE_MAPPING(struct f2fs_sb_info *sbi) 1539 { 1540 return sbi->node_inode->i_mapping; 1541 } 1542 1543 static inline bool is_sbi_flag_set(struct f2fs_sb_info *sbi, unsigned int type) 1544 { 1545 return test_bit(type, &sbi->s_flag); 1546 } 1547 1548 static inline void set_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type) 1549 { 1550 set_bit(type, &sbi->s_flag); 1551 } 1552 1553 static inline void clear_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type) 1554 { 1555 clear_bit(type, &sbi->s_flag); 1556 } 1557 1558 static inline unsigned long long cur_cp_version(struct f2fs_checkpoint *cp) 1559 { 1560 return le64_to_cpu(cp->checkpoint_ver); 1561 } 1562 1563 static inline unsigned long f2fs_qf_ino(struct super_block *sb, int type) 1564 { 1565 if (type < F2FS_MAX_QUOTAS) 1566 return le32_to_cpu(F2FS_SB(sb)->raw_super->qf_ino[type]); 1567 return 0; 1568 } 1569 1570 static inline __u64 cur_cp_crc(struct f2fs_checkpoint *cp) 1571 { 1572 size_t crc_offset = le32_to_cpu(cp->checksum_offset); 1573 return le32_to_cpu(*((__le32 *)((unsigned char *)cp + crc_offset))); 1574 } 1575 1576 static inline bool __is_set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 1577 { 1578 unsigned int ckpt_flags = le32_to_cpu(cp->ckpt_flags); 1579 1580 return ckpt_flags & f; 1581 } 1582 1583 static inline bool is_set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 1584 { 1585 return __is_set_ckpt_flags(F2FS_CKPT(sbi), f); 1586 } 1587 1588 static inline void __set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 1589 { 1590 unsigned int ckpt_flags; 1591 1592 ckpt_flags = le32_to_cpu(cp->ckpt_flags); 1593 ckpt_flags |= f; 1594 cp->ckpt_flags = cpu_to_le32(ckpt_flags); 1595 } 1596 1597 static inline void set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 1598 { 1599 unsigned long flags; 1600 1601 spin_lock_irqsave(&sbi->cp_lock, flags); 1602 __set_ckpt_flags(F2FS_CKPT(sbi), f); 1603 spin_unlock_irqrestore(&sbi->cp_lock, flags); 1604 } 1605 1606 static inline void __clear_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 1607 { 1608 unsigned int ckpt_flags; 1609 1610 ckpt_flags = le32_to_cpu(cp->ckpt_flags); 1611 ckpt_flags &= (~f); 1612 cp->ckpt_flags = cpu_to_le32(ckpt_flags); 1613 } 1614 1615 static inline void clear_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 1616 { 1617 unsigned long flags; 1618 1619 spin_lock_irqsave(&sbi->cp_lock, flags); 1620 __clear_ckpt_flags(F2FS_CKPT(sbi), f); 1621 spin_unlock_irqrestore(&sbi->cp_lock, flags); 1622 } 1623 1624 static inline void disable_nat_bits(struct f2fs_sb_info *sbi, bool lock) 1625 { 1626 unsigned long flags; 1627 1628 /* 1629 * In order to re-enable nat_bits we need to call fsck.f2fs by 1630 * set_sbi_flag(sbi, SBI_NEED_FSCK). But it may give huge cost, 1631 * so let's rely on regular fsck or unclean shutdown. 1632 */ 1633 1634 if (lock) 1635 spin_lock_irqsave(&sbi->cp_lock, flags); 1636 __clear_ckpt_flags(F2FS_CKPT(sbi), CP_NAT_BITS_FLAG); 1637 kvfree(NM_I(sbi)->nat_bits); 1638 NM_I(sbi)->nat_bits = NULL; 1639 if (lock) 1640 spin_unlock_irqrestore(&sbi->cp_lock, flags); 1641 } 1642 1643 static inline bool enabled_nat_bits(struct f2fs_sb_info *sbi, 1644 struct cp_control *cpc) 1645 { 1646 bool set = is_set_ckpt_flags(sbi, CP_NAT_BITS_FLAG); 1647 1648 return (cpc) ? (cpc->reason & CP_UMOUNT) && set : set; 1649 } 1650 1651 static inline void f2fs_lock_op(struct f2fs_sb_info *sbi) 1652 { 1653 down_read(&sbi->cp_rwsem); 1654 } 1655 1656 static inline int f2fs_trylock_op(struct f2fs_sb_info *sbi) 1657 { 1658 return down_read_trylock(&sbi->cp_rwsem); 1659 } 1660 1661 static inline void f2fs_unlock_op(struct f2fs_sb_info *sbi) 1662 { 1663 up_read(&sbi->cp_rwsem); 1664 } 1665 1666 static inline void f2fs_lock_all(struct f2fs_sb_info *sbi) 1667 { 1668 down_write(&sbi->cp_rwsem); 1669 } 1670 1671 static inline void f2fs_unlock_all(struct f2fs_sb_info *sbi) 1672 { 1673 up_write(&sbi->cp_rwsem); 1674 } 1675 1676 static inline int __get_cp_reason(struct f2fs_sb_info *sbi) 1677 { 1678 int reason = CP_SYNC; 1679 1680 if (test_opt(sbi, FASTBOOT)) 1681 reason = CP_FASTBOOT; 1682 if (is_sbi_flag_set(sbi, SBI_IS_CLOSE)) 1683 reason = CP_UMOUNT; 1684 return reason; 1685 } 1686 1687 static inline bool __remain_node_summaries(int reason) 1688 { 1689 return (reason & (CP_UMOUNT | CP_FASTBOOT)); 1690 } 1691 1692 static inline bool __exist_node_summaries(struct f2fs_sb_info *sbi) 1693 { 1694 return (is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG) || 1695 is_set_ckpt_flags(sbi, CP_FASTBOOT_FLAG)); 1696 } 1697 1698 /* 1699 * Check whether the inode has blocks or not 1700 */ 1701 static inline int F2FS_HAS_BLOCKS(struct inode *inode) 1702 { 1703 block_t xattr_block = F2FS_I(inode)->i_xattr_nid ? 1 : 0; 1704 1705 return (inode->i_blocks >> F2FS_LOG_SECTORS_PER_BLOCK) > xattr_block; 1706 } 1707 1708 static inline bool f2fs_has_xattr_block(unsigned int ofs) 1709 { 1710 return ofs == XATTR_NODE_OFFSET; 1711 } 1712 1713 static inline bool __allow_reserved_blocks(struct f2fs_sb_info *sbi, 1714 struct inode *inode, bool cap) 1715 { 1716 if (!inode) 1717 return true; 1718 if (!test_opt(sbi, RESERVE_ROOT)) 1719 return false; 1720 if (IS_NOQUOTA(inode)) 1721 return true; 1722 if (uid_eq(F2FS_OPTION(sbi).s_resuid, current_fsuid())) 1723 return true; 1724 if (!gid_eq(F2FS_OPTION(sbi).s_resgid, GLOBAL_ROOT_GID) && 1725 in_group_p(F2FS_OPTION(sbi).s_resgid)) 1726 return true; 1727 if (cap && capable(CAP_SYS_RESOURCE)) 1728 return true; 1729 return false; 1730 } 1731 1732 static inline void f2fs_i_blocks_write(struct inode *, block_t, bool, bool); 1733 static inline int inc_valid_block_count(struct f2fs_sb_info *sbi, 1734 struct inode *inode, blkcnt_t *count) 1735 { 1736 blkcnt_t diff = 0, release = 0; 1737 block_t avail_user_block_count; 1738 int ret; 1739 1740 ret = dquot_reserve_block(inode, *count); 1741 if (ret) 1742 return ret; 1743 1744 if (time_to_inject(sbi, FAULT_BLOCK)) { 1745 f2fs_show_injection_info(FAULT_BLOCK); 1746 release = *count; 1747 goto enospc; 1748 } 1749 1750 /* 1751 * let's increase this in prior to actual block count change in order 1752 * for f2fs_sync_file to avoid data races when deciding checkpoint. 1753 */ 1754 percpu_counter_add(&sbi->alloc_valid_block_count, (*count)); 1755 1756 spin_lock(&sbi->stat_lock); 1757 sbi->total_valid_block_count += (block_t)(*count); 1758 avail_user_block_count = sbi->user_block_count - 1759 sbi->current_reserved_blocks; 1760 1761 if (!__allow_reserved_blocks(sbi, inode, true)) 1762 avail_user_block_count -= F2FS_OPTION(sbi).root_reserved_blocks; 1763 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) 1764 avail_user_block_count -= sbi->unusable_block_count; 1765 if (unlikely(sbi->total_valid_block_count > avail_user_block_count)) { 1766 diff = sbi->total_valid_block_count - avail_user_block_count; 1767 if (diff > *count) 1768 diff = *count; 1769 *count -= diff; 1770 release = diff; 1771 sbi->total_valid_block_count -= diff; 1772 if (!*count) { 1773 spin_unlock(&sbi->stat_lock); 1774 goto enospc; 1775 } 1776 } 1777 spin_unlock(&sbi->stat_lock); 1778 1779 if (unlikely(release)) { 1780 percpu_counter_sub(&sbi->alloc_valid_block_count, release); 1781 dquot_release_reservation_block(inode, release); 1782 } 1783 f2fs_i_blocks_write(inode, *count, true, true); 1784 return 0; 1785 1786 enospc: 1787 percpu_counter_sub(&sbi->alloc_valid_block_count, release); 1788 dquot_release_reservation_block(inode, release); 1789 return -ENOSPC; 1790 } 1791 1792 void f2fs_msg(struct super_block *sb, const char *level, const char *fmt, ...); 1793 static inline void dec_valid_block_count(struct f2fs_sb_info *sbi, 1794 struct inode *inode, 1795 block_t count) 1796 { 1797 blkcnt_t sectors = count << F2FS_LOG_SECTORS_PER_BLOCK; 1798 1799 spin_lock(&sbi->stat_lock); 1800 f2fs_bug_on(sbi, sbi->total_valid_block_count < (block_t) count); 1801 sbi->total_valid_block_count -= (block_t)count; 1802 if (sbi->reserved_blocks && 1803 sbi->current_reserved_blocks < sbi->reserved_blocks) 1804 sbi->current_reserved_blocks = min(sbi->reserved_blocks, 1805 sbi->current_reserved_blocks + count); 1806 spin_unlock(&sbi->stat_lock); 1807 if (unlikely(inode->i_blocks < sectors)) { 1808 f2fs_msg(sbi->sb, KERN_WARNING, 1809 "Inconsistent i_blocks, ino:%lu, iblocks:%llu, sectors:%llu", 1810 inode->i_ino, 1811 (unsigned long long)inode->i_blocks, 1812 (unsigned long long)sectors); 1813 set_sbi_flag(sbi, SBI_NEED_FSCK); 1814 return; 1815 } 1816 f2fs_i_blocks_write(inode, count, false, true); 1817 } 1818 1819 static inline void inc_page_count(struct f2fs_sb_info *sbi, int count_type) 1820 { 1821 atomic_inc(&sbi->nr_pages[count_type]); 1822 1823 if (count_type == F2FS_DIRTY_DENTS || 1824 count_type == F2FS_DIRTY_NODES || 1825 count_type == F2FS_DIRTY_META || 1826 count_type == F2FS_DIRTY_QDATA || 1827 count_type == F2FS_DIRTY_IMETA) 1828 set_sbi_flag(sbi, SBI_IS_DIRTY); 1829 } 1830 1831 static inline void inode_inc_dirty_pages(struct inode *inode) 1832 { 1833 atomic_inc(&F2FS_I(inode)->dirty_pages); 1834 inc_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ? 1835 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA); 1836 if (IS_NOQUOTA(inode)) 1837 inc_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA); 1838 } 1839 1840 static inline void dec_page_count(struct f2fs_sb_info *sbi, int count_type) 1841 { 1842 atomic_dec(&sbi->nr_pages[count_type]); 1843 } 1844 1845 static inline void inode_dec_dirty_pages(struct inode *inode) 1846 { 1847 if (!S_ISDIR(inode->i_mode) && !S_ISREG(inode->i_mode) && 1848 !S_ISLNK(inode->i_mode)) 1849 return; 1850 1851 atomic_dec(&F2FS_I(inode)->dirty_pages); 1852 dec_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ? 1853 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA); 1854 if (IS_NOQUOTA(inode)) 1855 dec_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA); 1856 } 1857 1858 static inline s64 get_pages(struct f2fs_sb_info *sbi, int count_type) 1859 { 1860 return atomic_read(&sbi->nr_pages[count_type]); 1861 } 1862 1863 static inline int get_dirty_pages(struct inode *inode) 1864 { 1865 return atomic_read(&F2FS_I(inode)->dirty_pages); 1866 } 1867 1868 static inline int get_blocktype_secs(struct f2fs_sb_info *sbi, int block_type) 1869 { 1870 unsigned int pages_per_sec = sbi->segs_per_sec * sbi->blocks_per_seg; 1871 unsigned int segs = (get_pages(sbi, block_type) + pages_per_sec - 1) >> 1872 sbi->log_blocks_per_seg; 1873 1874 return segs / sbi->segs_per_sec; 1875 } 1876 1877 static inline block_t valid_user_blocks(struct f2fs_sb_info *sbi) 1878 { 1879 return sbi->total_valid_block_count; 1880 } 1881 1882 static inline block_t discard_blocks(struct f2fs_sb_info *sbi) 1883 { 1884 return sbi->discard_blks; 1885 } 1886 1887 static inline unsigned long __bitmap_size(struct f2fs_sb_info *sbi, int flag) 1888 { 1889 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi); 1890 1891 /* return NAT or SIT bitmap */ 1892 if (flag == NAT_BITMAP) 1893 return le32_to_cpu(ckpt->nat_ver_bitmap_bytesize); 1894 else if (flag == SIT_BITMAP) 1895 return le32_to_cpu(ckpt->sit_ver_bitmap_bytesize); 1896 1897 return 0; 1898 } 1899 1900 static inline block_t __cp_payload(struct f2fs_sb_info *sbi) 1901 { 1902 return le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_payload); 1903 } 1904 1905 static inline void *__bitmap_ptr(struct f2fs_sb_info *sbi, int flag) 1906 { 1907 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi); 1908 int offset; 1909 1910 if (is_set_ckpt_flags(sbi, CP_LARGE_NAT_BITMAP_FLAG)) { 1911 offset = (flag == SIT_BITMAP) ? 1912 le32_to_cpu(ckpt->nat_ver_bitmap_bytesize) : 0; 1913 return &ckpt->sit_nat_version_bitmap + offset; 1914 } 1915 1916 if (__cp_payload(sbi) > 0) { 1917 if (flag == NAT_BITMAP) 1918 return &ckpt->sit_nat_version_bitmap; 1919 else 1920 return (unsigned char *)ckpt + F2FS_BLKSIZE; 1921 } else { 1922 offset = (flag == NAT_BITMAP) ? 1923 le32_to_cpu(ckpt->sit_ver_bitmap_bytesize) : 0; 1924 return &ckpt->sit_nat_version_bitmap + offset; 1925 } 1926 } 1927 1928 static inline block_t __start_cp_addr(struct f2fs_sb_info *sbi) 1929 { 1930 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr); 1931 1932 if (sbi->cur_cp_pack == 2) 1933 start_addr += sbi->blocks_per_seg; 1934 return start_addr; 1935 } 1936 1937 static inline block_t __start_cp_next_addr(struct f2fs_sb_info *sbi) 1938 { 1939 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr); 1940 1941 if (sbi->cur_cp_pack == 1) 1942 start_addr += sbi->blocks_per_seg; 1943 return start_addr; 1944 } 1945 1946 static inline void __set_cp_next_pack(struct f2fs_sb_info *sbi) 1947 { 1948 sbi->cur_cp_pack = (sbi->cur_cp_pack == 1) ? 2 : 1; 1949 } 1950 1951 static inline block_t __start_sum_addr(struct f2fs_sb_info *sbi) 1952 { 1953 return le32_to_cpu(F2FS_CKPT(sbi)->cp_pack_start_sum); 1954 } 1955 1956 static inline int inc_valid_node_count(struct f2fs_sb_info *sbi, 1957 struct inode *inode, bool is_inode) 1958 { 1959 block_t valid_block_count; 1960 unsigned int valid_node_count; 1961 int err; 1962 1963 if (is_inode) { 1964 if (inode) { 1965 err = dquot_alloc_inode(inode); 1966 if (err) 1967 return err; 1968 } 1969 } else { 1970 err = dquot_reserve_block(inode, 1); 1971 if (err) 1972 return err; 1973 } 1974 1975 if (time_to_inject(sbi, FAULT_BLOCK)) { 1976 f2fs_show_injection_info(FAULT_BLOCK); 1977 goto enospc; 1978 } 1979 1980 spin_lock(&sbi->stat_lock); 1981 1982 valid_block_count = sbi->total_valid_block_count + 1983 sbi->current_reserved_blocks + 1; 1984 1985 if (!__allow_reserved_blocks(sbi, inode, false)) 1986 valid_block_count += F2FS_OPTION(sbi).root_reserved_blocks; 1987 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) 1988 valid_block_count += sbi->unusable_block_count; 1989 1990 if (unlikely(valid_block_count > sbi->user_block_count)) { 1991 spin_unlock(&sbi->stat_lock); 1992 goto enospc; 1993 } 1994 1995 valid_node_count = sbi->total_valid_node_count + 1; 1996 if (unlikely(valid_node_count > sbi->total_node_count)) { 1997 spin_unlock(&sbi->stat_lock); 1998 goto enospc; 1999 } 2000 2001 sbi->total_valid_node_count++; 2002 sbi->total_valid_block_count++; 2003 spin_unlock(&sbi->stat_lock); 2004 2005 if (inode) { 2006 if (is_inode) 2007 f2fs_mark_inode_dirty_sync(inode, true); 2008 else 2009 f2fs_i_blocks_write(inode, 1, true, true); 2010 } 2011 2012 percpu_counter_inc(&sbi->alloc_valid_block_count); 2013 return 0; 2014 2015 enospc: 2016 if (is_inode) { 2017 if (inode) 2018 dquot_free_inode(inode); 2019 } else { 2020 dquot_release_reservation_block(inode, 1); 2021 } 2022 return -ENOSPC; 2023 } 2024 2025 static inline void dec_valid_node_count(struct f2fs_sb_info *sbi, 2026 struct inode *inode, bool is_inode) 2027 { 2028 spin_lock(&sbi->stat_lock); 2029 2030 f2fs_bug_on(sbi, !sbi->total_valid_block_count); 2031 f2fs_bug_on(sbi, !sbi->total_valid_node_count); 2032 2033 sbi->total_valid_node_count--; 2034 sbi->total_valid_block_count--; 2035 if (sbi->reserved_blocks && 2036 sbi->current_reserved_blocks < sbi->reserved_blocks) 2037 sbi->current_reserved_blocks++; 2038 2039 spin_unlock(&sbi->stat_lock); 2040 2041 if (is_inode) { 2042 dquot_free_inode(inode); 2043 } else { 2044 if (unlikely(inode->i_blocks == 0)) { 2045 f2fs_msg(sbi->sb, KERN_WARNING, 2046 "Inconsistent i_blocks, ino:%lu, iblocks:%llu", 2047 inode->i_ino, 2048 (unsigned long long)inode->i_blocks); 2049 set_sbi_flag(sbi, SBI_NEED_FSCK); 2050 return; 2051 } 2052 f2fs_i_blocks_write(inode, 1, false, true); 2053 } 2054 } 2055 2056 static inline unsigned int valid_node_count(struct f2fs_sb_info *sbi) 2057 { 2058 return sbi->total_valid_node_count; 2059 } 2060 2061 static inline void inc_valid_inode_count(struct f2fs_sb_info *sbi) 2062 { 2063 percpu_counter_inc(&sbi->total_valid_inode_count); 2064 } 2065 2066 static inline void dec_valid_inode_count(struct f2fs_sb_info *sbi) 2067 { 2068 percpu_counter_dec(&sbi->total_valid_inode_count); 2069 } 2070 2071 static inline s64 valid_inode_count(struct f2fs_sb_info *sbi) 2072 { 2073 return percpu_counter_sum_positive(&sbi->total_valid_inode_count); 2074 } 2075 2076 static inline struct page *f2fs_grab_cache_page(struct address_space *mapping, 2077 pgoff_t index, bool for_write) 2078 { 2079 struct page *page; 2080 2081 if (IS_ENABLED(CONFIG_F2FS_FAULT_INJECTION)) { 2082 if (!for_write) 2083 page = find_get_page_flags(mapping, index, 2084 FGP_LOCK | FGP_ACCESSED); 2085 else 2086 page = find_lock_page(mapping, index); 2087 if (page) 2088 return page; 2089 2090 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_ALLOC)) { 2091 f2fs_show_injection_info(FAULT_PAGE_ALLOC); 2092 return NULL; 2093 } 2094 } 2095 2096 if (!for_write) 2097 return grab_cache_page(mapping, index); 2098 return grab_cache_page_write_begin(mapping, index, AOP_FLAG_NOFS); 2099 } 2100 2101 static inline struct page *f2fs_pagecache_get_page( 2102 struct address_space *mapping, pgoff_t index, 2103 int fgp_flags, gfp_t gfp_mask) 2104 { 2105 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_GET)) { 2106 f2fs_show_injection_info(FAULT_PAGE_GET); 2107 return NULL; 2108 } 2109 2110 return pagecache_get_page(mapping, index, fgp_flags, gfp_mask); 2111 } 2112 2113 static inline void f2fs_copy_page(struct page *src, struct page *dst) 2114 { 2115 char *src_kaddr = kmap(src); 2116 char *dst_kaddr = kmap(dst); 2117 2118 memcpy(dst_kaddr, src_kaddr, PAGE_SIZE); 2119 kunmap(dst); 2120 kunmap(src); 2121 } 2122 2123 static inline void f2fs_put_page(struct page *page, int unlock) 2124 { 2125 if (!page) 2126 return; 2127 2128 if (unlock) { 2129 f2fs_bug_on(F2FS_P_SB(page), !PageLocked(page)); 2130 unlock_page(page); 2131 } 2132 put_page(page); 2133 } 2134 2135 static inline void f2fs_put_dnode(struct dnode_of_data *dn) 2136 { 2137 if (dn->node_page) 2138 f2fs_put_page(dn->node_page, 1); 2139 if (dn->inode_page && dn->node_page != dn->inode_page) 2140 f2fs_put_page(dn->inode_page, 0); 2141 dn->node_page = NULL; 2142 dn->inode_page = NULL; 2143 } 2144 2145 static inline struct kmem_cache *f2fs_kmem_cache_create(const char *name, 2146 size_t size) 2147 { 2148 return kmem_cache_create(name, size, 0, SLAB_RECLAIM_ACCOUNT, NULL); 2149 } 2150 2151 static inline void *f2fs_kmem_cache_alloc(struct kmem_cache *cachep, 2152 gfp_t flags) 2153 { 2154 void *entry; 2155 2156 entry = kmem_cache_alloc(cachep, flags); 2157 if (!entry) 2158 entry = kmem_cache_alloc(cachep, flags | __GFP_NOFAIL); 2159 return entry; 2160 } 2161 2162 static inline struct bio *f2fs_bio_alloc(struct f2fs_sb_info *sbi, 2163 int npages, bool no_fail) 2164 { 2165 struct bio *bio; 2166 2167 if (no_fail) { 2168 /* No failure on bio allocation */ 2169 bio = bio_alloc(GFP_NOIO, npages); 2170 if (!bio) 2171 bio = bio_alloc(GFP_NOIO | __GFP_NOFAIL, npages); 2172 return bio; 2173 } 2174 if (time_to_inject(sbi, FAULT_ALLOC_BIO)) { 2175 f2fs_show_injection_info(FAULT_ALLOC_BIO); 2176 return NULL; 2177 } 2178 2179 return bio_alloc(GFP_KERNEL, npages); 2180 } 2181 2182 static inline bool is_idle(struct f2fs_sb_info *sbi, int type) 2183 { 2184 if (get_pages(sbi, F2FS_RD_DATA) || get_pages(sbi, F2FS_RD_NODE) || 2185 get_pages(sbi, F2FS_RD_META) || get_pages(sbi, F2FS_WB_DATA) || 2186 get_pages(sbi, F2FS_WB_CP_DATA) || 2187 get_pages(sbi, F2FS_DIO_READ) || 2188 get_pages(sbi, F2FS_DIO_WRITE)) 2189 return false; 2190 2191 if (SM_I(sbi) && SM_I(sbi)->dcc_info && 2192 atomic_read(&SM_I(sbi)->dcc_info->queued_discard)) 2193 return false; 2194 2195 if (SM_I(sbi) && SM_I(sbi)->fcc_info && 2196 atomic_read(&SM_I(sbi)->fcc_info->queued_flush)) 2197 return false; 2198 2199 return f2fs_time_over(sbi, type); 2200 } 2201 2202 static inline void f2fs_radix_tree_insert(struct radix_tree_root *root, 2203 unsigned long index, void *item) 2204 { 2205 while (radix_tree_insert(root, index, item)) 2206 cond_resched(); 2207 } 2208 2209 #define RAW_IS_INODE(p) ((p)->footer.nid == (p)->footer.ino) 2210 2211 static inline bool IS_INODE(struct page *page) 2212 { 2213 struct f2fs_node *p = F2FS_NODE(page); 2214 2215 return RAW_IS_INODE(p); 2216 } 2217 2218 static inline int offset_in_addr(struct f2fs_inode *i) 2219 { 2220 return (i->i_inline & F2FS_EXTRA_ATTR) ? 2221 (le16_to_cpu(i->i_extra_isize) / sizeof(__le32)) : 0; 2222 } 2223 2224 static inline __le32 *blkaddr_in_node(struct f2fs_node *node) 2225 { 2226 return RAW_IS_INODE(node) ? node->i.i_addr : node->dn.addr; 2227 } 2228 2229 static inline int f2fs_has_extra_attr(struct inode *inode); 2230 static inline block_t datablock_addr(struct inode *inode, 2231 struct page *node_page, unsigned int offset) 2232 { 2233 struct f2fs_node *raw_node; 2234 __le32 *addr_array; 2235 int base = 0; 2236 bool is_inode = IS_INODE(node_page); 2237 2238 raw_node = F2FS_NODE(node_page); 2239 2240 /* from GC path only */ 2241 if (is_inode) { 2242 if (!inode) 2243 base = offset_in_addr(&raw_node->i); 2244 else if (f2fs_has_extra_attr(inode)) 2245 base = get_extra_isize(inode); 2246 } 2247 2248 addr_array = blkaddr_in_node(raw_node); 2249 return le32_to_cpu(addr_array[base + offset]); 2250 } 2251 2252 static inline int f2fs_test_bit(unsigned int nr, char *addr) 2253 { 2254 int mask; 2255 2256 addr += (nr >> 3); 2257 mask = 1 << (7 - (nr & 0x07)); 2258 return mask & *addr; 2259 } 2260 2261 static inline void f2fs_set_bit(unsigned int nr, char *addr) 2262 { 2263 int mask; 2264 2265 addr += (nr >> 3); 2266 mask = 1 << (7 - (nr & 0x07)); 2267 *addr |= mask; 2268 } 2269 2270 static inline void f2fs_clear_bit(unsigned int nr, char *addr) 2271 { 2272 int mask; 2273 2274 addr += (nr >> 3); 2275 mask = 1 << (7 - (nr & 0x07)); 2276 *addr &= ~mask; 2277 } 2278 2279 static inline int f2fs_test_and_set_bit(unsigned int nr, char *addr) 2280 { 2281 int mask; 2282 int ret; 2283 2284 addr += (nr >> 3); 2285 mask = 1 << (7 - (nr & 0x07)); 2286 ret = mask & *addr; 2287 *addr |= mask; 2288 return ret; 2289 } 2290 2291 static inline int f2fs_test_and_clear_bit(unsigned int nr, char *addr) 2292 { 2293 int mask; 2294 int ret; 2295 2296 addr += (nr >> 3); 2297 mask = 1 << (7 - (nr & 0x07)); 2298 ret = mask & *addr; 2299 *addr &= ~mask; 2300 return ret; 2301 } 2302 2303 static inline void f2fs_change_bit(unsigned int nr, char *addr) 2304 { 2305 int mask; 2306 2307 addr += (nr >> 3); 2308 mask = 1 << (7 - (nr & 0x07)); 2309 *addr ^= mask; 2310 } 2311 2312 /* 2313 * Inode flags 2314 */ 2315 #define F2FS_SECRM_FL 0x00000001 /* Secure deletion */ 2316 #define F2FS_UNRM_FL 0x00000002 /* Undelete */ 2317 #define F2FS_COMPR_FL 0x00000004 /* Compress file */ 2318 #define F2FS_SYNC_FL 0x00000008 /* Synchronous updates */ 2319 #define F2FS_IMMUTABLE_FL 0x00000010 /* Immutable file */ 2320 #define F2FS_APPEND_FL 0x00000020 /* writes to file may only append */ 2321 #define F2FS_NODUMP_FL 0x00000040 /* do not dump file */ 2322 #define F2FS_NOATIME_FL 0x00000080 /* do not update atime */ 2323 /* Reserved for compression usage... */ 2324 #define F2FS_DIRTY_FL 0x00000100 2325 #define F2FS_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */ 2326 #define F2FS_NOCOMPR_FL 0x00000400 /* Don't compress */ 2327 #define F2FS_ENCRYPT_FL 0x00000800 /* encrypted file */ 2328 /* End compression flags --- maybe not all used */ 2329 #define F2FS_INDEX_FL 0x00001000 /* hash-indexed directory */ 2330 #define F2FS_IMAGIC_FL 0x00002000 /* AFS directory */ 2331 #define F2FS_JOURNAL_DATA_FL 0x00004000 /* file data should be journaled */ 2332 #define F2FS_NOTAIL_FL 0x00008000 /* file tail should not be merged */ 2333 #define F2FS_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */ 2334 #define F2FS_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/ 2335 #define F2FS_HUGE_FILE_FL 0x00040000 /* Set to each huge file */ 2336 #define F2FS_EXTENTS_FL 0x00080000 /* Inode uses extents */ 2337 #define F2FS_EA_INODE_FL 0x00200000 /* Inode used for large EA */ 2338 #define F2FS_EOFBLOCKS_FL 0x00400000 /* Blocks allocated beyond EOF */ 2339 #define F2FS_NOCOW_FL 0x00800000 /* Do not cow file */ 2340 #define F2FS_INLINE_DATA_FL 0x10000000 /* Inode has inline data. */ 2341 #define F2FS_PROJINHERIT_FL 0x20000000 /* Create with parents projid */ 2342 #define F2FS_RESERVED_FL 0x80000000 /* reserved for ext4 lib */ 2343 2344 #define F2FS_FL_USER_VISIBLE 0x30CBDFFF /* User visible flags */ 2345 #define F2FS_FL_USER_MODIFIABLE 0x204BC0FF /* User modifiable flags */ 2346 2347 /* Flags we can manipulate with through F2FS_IOC_FSSETXATTR */ 2348 #define F2FS_FL_XFLAG_VISIBLE (F2FS_SYNC_FL | \ 2349 F2FS_IMMUTABLE_FL | \ 2350 F2FS_APPEND_FL | \ 2351 F2FS_NODUMP_FL | \ 2352 F2FS_NOATIME_FL | \ 2353 F2FS_PROJINHERIT_FL) 2354 2355 /* Flags that should be inherited by new inodes from their parent. */ 2356 #define F2FS_FL_INHERITED (F2FS_SECRM_FL | F2FS_UNRM_FL | F2FS_COMPR_FL |\ 2357 F2FS_SYNC_FL | F2FS_NODUMP_FL | F2FS_NOATIME_FL |\ 2358 F2FS_NOCOMPR_FL | F2FS_JOURNAL_DATA_FL |\ 2359 F2FS_NOTAIL_FL | F2FS_DIRSYNC_FL |\ 2360 F2FS_PROJINHERIT_FL) 2361 2362 /* Flags that are appropriate for regular files (all but dir-specific ones). */ 2363 #define F2FS_REG_FLMASK (~(F2FS_DIRSYNC_FL | F2FS_TOPDIR_FL)) 2364 2365 /* Flags that are appropriate for non-directories/regular files. */ 2366 #define F2FS_OTHER_FLMASK (F2FS_NODUMP_FL | F2FS_NOATIME_FL) 2367 2368 static inline __u32 f2fs_mask_flags(umode_t mode, __u32 flags) 2369 { 2370 if (S_ISDIR(mode)) 2371 return flags; 2372 else if (S_ISREG(mode)) 2373 return flags & F2FS_REG_FLMASK; 2374 else 2375 return flags & F2FS_OTHER_FLMASK; 2376 } 2377 2378 /* used for f2fs_inode_info->flags */ 2379 enum { 2380 FI_NEW_INODE, /* indicate newly allocated inode */ 2381 FI_DIRTY_INODE, /* indicate inode is dirty or not */ 2382 FI_AUTO_RECOVER, /* indicate inode is recoverable */ 2383 FI_DIRTY_DIR, /* indicate directory has dirty pages */ 2384 FI_INC_LINK, /* need to increment i_nlink */ 2385 FI_ACL_MODE, /* indicate acl mode */ 2386 FI_NO_ALLOC, /* should not allocate any blocks */ 2387 FI_FREE_NID, /* free allocated nide */ 2388 FI_NO_EXTENT, /* not to use the extent cache */ 2389 FI_INLINE_XATTR, /* used for inline xattr */ 2390 FI_INLINE_DATA, /* used for inline data*/ 2391 FI_INLINE_DENTRY, /* used for inline dentry */ 2392 FI_APPEND_WRITE, /* inode has appended data */ 2393 FI_UPDATE_WRITE, /* inode has in-place-update data */ 2394 FI_NEED_IPU, /* used for ipu per file */ 2395 FI_ATOMIC_FILE, /* indicate atomic file */ 2396 FI_ATOMIC_COMMIT, /* indicate the state of atomical committing */ 2397 FI_VOLATILE_FILE, /* indicate volatile file */ 2398 FI_FIRST_BLOCK_WRITTEN, /* indicate #0 data block was written */ 2399 FI_DROP_CACHE, /* drop dirty page cache */ 2400 FI_DATA_EXIST, /* indicate data exists */ 2401 FI_INLINE_DOTS, /* indicate inline dot dentries */ 2402 FI_DO_DEFRAG, /* indicate defragment is running */ 2403 FI_DIRTY_FILE, /* indicate regular/symlink has dirty pages */ 2404 FI_NO_PREALLOC, /* indicate skipped preallocated blocks */ 2405 FI_HOT_DATA, /* indicate file is hot */ 2406 FI_EXTRA_ATTR, /* indicate file has extra attribute */ 2407 FI_PROJ_INHERIT, /* indicate file inherits projectid */ 2408 FI_PIN_FILE, /* indicate file should not be gced */ 2409 FI_ATOMIC_REVOKE_REQUEST, /* request to drop atomic data */ 2410 }; 2411 2412 static inline void __mark_inode_dirty_flag(struct inode *inode, 2413 int flag, bool set) 2414 { 2415 switch (flag) { 2416 case FI_INLINE_XATTR: 2417 case FI_INLINE_DATA: 2418 case FI_INLINE_DENTRY: 2419 case FI_NEW_INODE: 2420 if (set) 2421 return; 2422 /* fall through */ 2423 case FI_DATA_EXIST: 2424 case FI_INLINE_DOTS: 2425 case FI_PIN_FILE: 2426 f2fs_mark_inode_dirty_sync(inode, true); 2427 } 2428 } 2429 2430 static inline void set_inode_flag(struct inode *inode, int flag) 2431 { 2432 if (!test_bit(flag, &F2FS_I(inode)->flags)) 2433 set_bit(flag, &F2FS_I(inode)->flags); 2434 __mark_inode_dirty_flag(inode, flag, true); 2435 } 2436 2437 static inline int is_inode_flag_set(struct inode *inode, int flag) 2438 { 2439 return test_bit(flag, &F2FS_I(inode)->flags); 2440 } 2441 2442 static inline void clear_inode_flag(struct inode *inode, int flag) 2443 { 2444 if (test_bit(flag, &F2FS_I(inode)->flags)) 2445 clear_bit(flag, &F2FS_I(inode)->flags); 2446 __mark_inode_dirty_flag(inode, flag, false); 2447 } 2448 2449 static inline void set_acl_inode(struct inode *inode, umode_t mode) 2450 { 2451 F2FS_I(inode)->i_acl_mode = mode; 2452 set_inode_flag(inode, FI_ACL_MODE); 2453 f2fs_mark_inode_dirty_sync(inode, false); 2454 } 2455 2456 static inline void f2fs_i_links_write(struct inode *inode, bool inc) 2457 { 2458 if (inc) 2459 inc_nlink(inode); 2460 else 2461 drop_nlink(inode); 2462 f2fs_mark_inode_dirty_sync(inode, true); 2463 } 2464 2465 static inline void f2fs_i_blocks_write(struct inode *inode, 2466 block_t diff, bool add, bool claim) 2467 { 2468 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE); 2469 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER); 2470 2471 /* add = 1, claim = 1 should be dquot_reserve_block in pair */ 2472 if (add) { 2473 if (claim) 2474 dquot_claim_block(inode, diff); 2475 else 2476 dquot_alloc_block_nofail(inode, diff); 2477 } else { 2478 dquot_free_block(inode, diff); 2479 } 2480 2481 f2fs_mark_inode_dirty_sync(inode, true); 2482 if (clean || recover) 2483 set_inode_flag(inode, FI_AUTO_RECOVER); 2484 } 2485 2486 static inline void f2fs_i_size_write(struct inode *inode, loff_t i_size) 2487 { 2488 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE); 2489 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER); 2490 2491 if (i_size_read(inode) == i_size) 2492 return; 2493 2494 i_size_write(inode, i_size); 2495 f2fs_mark_inode_dirty_sync(inode, true); 2496 if (clean || recover) 2497 set_inode_flag(inode, FI_AUTO_RECOVER); 2498 } 2499 2500 static inline void f2fs_i_depth_write(struct inode *inode, unsigned int depth) 2501 { 2502 F2FS_I(inode)->i_current_depth = depth; 2503 f2fs_mark_inode_dirty_sync(inode, true); 2504 } 2505 2506 static inline void f2fs_i_gc_failures_write(struct inode *inode, 2507 unsigned int count) 2508 { 2509 F2FS_I(inode)->i_gc_failures[GC_FAILURE_PIN] = count; 2510 f2fs_mark_inode_dirty_sync(inode, true); 2511 } 2512 2513 static inline void f2fs_i_xnid_write(struct inode *inode, nid_t xnid) 2514 { 2515 F2FS_I(inode)->i_xattr_nid = xnid; 2516 f2fs_mark_inode_dirty_sync(inode, true); 2517 } 2518 2519 static inline void f2fs_i_pino_write(struct inode *inode, nid_t pino) 2520 { 2521 F2FS_I(inode)->i_pino = pino; 2522 f2fs_mark_inode_dirty_sync(inode, true); 2523 } 2524 2525 static inline void get_inline_info(struct inode *inode, struct f2fs_inode *ri) 2526 { 2527 struct f2fs_inode_info *fi = F2FS_I(inode); 2528 2529 if (ri->i_inline & F2FS_INLINE_XATTR) 2530 set_bit(FI_INLINE_XATTR, &fi->flags); 2531 if (ri->i_inline & F2FS_INLINE_DATA) 2532 set_bit(FI_INLINE_DATA, &fi->flags); 2533 if (ri->i_inline & F2FS_INLINE_DENTRY) 2534 set_bit(FI_INLINE_DENTRY, &fi->flags); 2535 if (ri->i_inline & F2FS_DATA_EXIST) 2536 set_bit(FI_DATA_EXIST, &fi->flags); 2537 if (ri->i_inline & F2FS_INLINE_DOTS) 2538 set_bit(FI_INLINE_DOTS, &fi->flags); 2539 if (ri->i_inline & F2FS_EXTRA_ATTR) 2540 set_bit(FI_EXTRA_ATTR, &fi->flags); 2541 if (ri->i_inline & F2FS_PIN_FILE) 2542 set_bit(FI_PIN_FILE, &fi->flags); 2543 } 2544 2545 static inline void set_raw_inline(struct inode *inode, struct f2fs_inode *ri) 2546 { 2547 ri->i_inline = 0; 2548 2549 if (is_inode_flag_set(inode, FI_INLINE_XATTR)) 2550 ri->i_inline |= F2FS_INLINE_XATTR; 2551 if (is_inode_flag_set(inode, FI_INLINE_DATA)) 2552 ri->i_inline |= F2FS_INLINE_DATA; 2553 if (is_inode_flag_set(inode, FI_INLINE_DENTRY)) 2554 ri->i_inline |= F2FS_INLINE_DENTRY; 2555 if (is_inode_flag_set(inode, FI_DATA_EXIST)) 2556 ri->i_inline |= F2FS_DATA_EXIST; 2557 if (is_inode_flag_set(inode, FI_INLINE_DOTS)) 2558 ri->i_inline |= F2FS_INLINE_DOTS; 2559 if (is_inode_flag_set(inode, FI_EXTRA_ATTR)) 2560 ri->i_inline |= F2FS_EXTRA_ATTR; 2561 if (is_inode_flag_set(inode, FI_PIN_FILE)) 2562 ri->i_inline |= F2FS_PIN_FILE; 2563 } 2564 2565 static inline int f2fs_has_extra_attr(struct inode *inode) 2566 { 2567 return is_inode_flag_set(inode, FI_EXTRA_ATTR); 2568 } 2569 2570 static inline int f2fs_has_inline_xattr(struct inode *inode) 2571 { 2572 return is_inode_flag_set(inode, FI_INLINE_XATTR); 2573 } 2574 2575 static inline unsigned int addrs_per_inode(struct inode *inode) 2576 { 2577 return CUR_ADDRS_PER_INODE(inode) - get_inline_xattr_addrs(inode); 2578 } 2579 2580 static inline void *inline_xattr_addr(struct inode *inode, struct page *page) 2581 { 2582 struct f2fs_inode *ri = F2FS_INODE(page); 2583 2584 return (void *)&(ri->i_addr[DEF_ADDRS_PER_INODE - 2585 get_inline_xattr_addrs(inode)]); 2586 } 2587 2588 static inline int inline_xattr_size(struct inode *inode) 2589 { 2590 if (f2fs_has_inline_xattr(inode)) 2591 return get_inline_xattr_addrs(inode) * sizeof(__le32); 2592 return 0; 2593 } 2594 2595 static inline int f2fs_has_inline_data(struct inode *inode) 2596 { 2597 return is_inode_flag_set(inode, FI_INLINE_DATA); 2598 } 2599 2600 static inline int f2fs_exist_data(struct inode *inode) 2601 { 2602 return is_inode_flag_set(inode, FI_DATA_EXIST); 2603 } 2604 2605 static inline int f2fs_has_inline_dots(struct inode *inode) 2606 { 2607 return is_inode_flag_set(inode, FI_INLINE_DOTS); 2608 } 2609 2610 static inline bool f2fs_is_pinned_file(struct inode *inode) 2611 { 2612 return is_inode_flag_set(inode, FI_PIN_FILE); 2613 } 2614 2615 static inline bool f2fs_is_atomic_file(struct inode *inode) 2616 { 2617 return is_inode_flag_set(inode, FI_ATOMIC_FILE); 2618 } 2619 2620 static inline bool f2fs_is_commit_atomic_write(struct inode *inode) 2621 { 2622 return is_inode_flag_set(inode, FI_ATOMIC_COMMIT); 2623 } 2624 2625 static inline bool f2fs_is_volatile_file(struct inode *inode) 2626 { 2627 return is_inode_flag_set(inode, FI_VOLATILE_FILE); 2628 } 2629 2630 static inline bool f2fs_is_first_block_written(struct inode *inode) 2631 { 2632 return is_inode_flag_set(inode, FI_FIRST_BLOCK_WRITTEN); 2633 } 2634 2635 static inline bool f2fs_is_drop_cache(struct inode *inode) 2636 { 2637 return is_inode_flag_set(inode, FI_DROP_CACHE); 2638 } 2639 2640 static inline void *inline_data_addr(struct inode *inode, struct page *page) 2641 { 2642 struct f2fs_inode *ri = F2FS_INODE(page); 2643 int extra_size = get_extra_isize(inode); 2644 2645 return (void *)&(ri->i_addr[extra_size + DEF_INLINE_RESERVED_SIZE]); 2646 } 2647 2648 static inline int f2fs_has_inline_dentry(struct inode *inode) 2649 { 2650 return is_inode_flag_set(inode, FI_INLINE_DENTRY); 2651 } 2652 2653 static inline int is_file(struct inode *inode, int type) 2654 { 2655 return F2FS_I(inode)->i_advise & type; 2656 } 2657 2658 static inline void set_file(struct inode *inode, int type) 2659 { 2660 F2FS_I(inode)->i_advise |= type; 2661 f2fs_mark_inode_dirty_sync(inode, true); 2662 } 2663 2664 static inline void clear_file(struct inode *inode, int type) 2665 { 2666 F2FS_I(inode)->i_advise &= ~type; 2667 f2fs_mark_inode_dirty_sync(inode, true); 2668 } 2669 2670 static inline bool f2fs_skip_inode_update(struct inode *inode, int dsync) 2671 { 2672 bool ret; 2673 2674 if (dsync) { 2675 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 2676 2677 spin_lock(&sbi->inode_lock[DIRTY_META]); 2678 ret = list_empty(&F2FS_I(inode)->gdirty_list); 2679 spin_unlock(&sbi->inode_lock[DIRTY_META]); 2680 return ret; 2681 } 2682 if (!is_inode_flag_set(inode, FI_AUTO_RECOVER) || 2683 file_keep_isize(inode) || 2684 i_size_read(inode) & ~PAGE_MASK) 2685 return false; 2686 2687 if (!timespec64_equal(F2FS_I(inode)->i_disk_time, &inode->i_atime)) 2688 return false; 2689 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 1, &inode->i_ctime)) 2690 return false; 2691 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 2, &inode->i_mtime)) 2692 return false; 2693 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 3, 2694 &F2FS_I(inode)->i_crtime)) 2695 return false; 2696 2697 down_read(&F2FS_I(inode)->i_sem); 2698 ret = F2FS_I(inode)->last_disk_size == i_size_read(inode); 2699 up_read(&F2FS_I(inode)->i_sem); 2700 2701 return ret; 2702 } 2703 2704 static inline bool f2fs_readonly(struct super_block *sb) 2705 { 2706 return sb_rdonly(sb); 2707 } 2708 2709 static inline bool f2fs_cp_error(struct f2fs_sb_info *sbi) 2710 { 2711 return is_set_ckpt_flags(sbi, CP_ERROR_FLAG); 2712 } 2713 2714 static inline bool is_dot_dotdot(const struct qstr *str) 2715 { 2716 if (str->len == 1 && str->name[0] == '.') 2717 return true; 2718 2719 if (str->len == 2 && str->name[0] == '.' && str->name[1] == '.') 2720 return true; 2721 2722 return false; 2723 } 2724 2725 static inline bool f2fs_may_extent_tree(struct inode *inode) 2726 { 2727 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 2728 2729 if (!test_opt(sbi, EXTENT_CACHE) || 2730 is_inode_flag_set(inode, FI_NO_EXTENT)) 2731 return false; 2732 2733 /* 2734 * for recovered files during mount do not create extents 2735 * if shrinker is not registered. 2736 */ 2737 if (list_empty(&sbi->s_list)) 2738 return false; 2739 2740 return S_ISREG(inode->i_mode); 2741 } 2742 2743 static inline void *f2fs_kmalloc(struct f2fs_sb_info *sbi, 2744 size_t size, gfp_t flags) 2745 { 2746 void *ret; 2747 2748 if (time_to_inject(sbi, FAULT_KMALLOC)) { 2749 f2fs_show_injection_info(FAULT_KMALLOC); 2750 return NULL; 2751 } 2752 2753 ret = kmalloc(size, flags); 2754 if (ret) 2755 return ret; 2756 2757 return kvmalloc(size, flags); 2758 } 2759 2760 static inline void *f2fs_kzalloc(struct f2fs_sb_info *sbi, 2761 size_t size, gfp_t flags) 2762 { 2763 return f2fs_kmalloc(sbi, size, flags | __GFP_ZERO); 2764 } 2765 2766 static inline void *f2fs_kvmalloc(struct f2fs_sb_info *sbi, 2767 size_t size, gfp_t flags) 2768 { 2769 if (time_to_inject(sbi, FAULT_KVMALLOC)) { 2770 f2fs_show_injection_info(FAULT_KVMALLOC); 2771 return NULL; 2772 } 2773 2774 return kvmalloc(size, flags); 2775 } 2776 2777 static inline void *f2fs_kvzalloc(struct f2fs_sb_info *sbi, 2778 size_t size, gfp_t flags) 2779 { 2780 return f2fs_kvmalloc(sbi, size, flags | __GFP_ZERO); 2781 } 2782 2783 static inline int get_extra_isize(struct inode *inode) 2784 { 2785 return F2FS_I(inode)->i_extra_isize / sizeof(__le32); 2786 } 2787 2788 static inline int get_inline_xattr_addrs(struct inode *inode) 2789 { 2790 return F2FS_I(inode)->i_inline_xattr_size; 2791 } 2792 2793 #define f2fs_get_inode_mode(i) \ 2794 ((is_inode_flag_set(i, FI_ACL_MODE)) ? \ 2795 (F2FS_I(i)->i_acl_mode) : ((i)->i_mode)) 2796 2797 #define F2FS_TOTAL_EXTRA_ATTR_SIZE \ 2798 (offsetof(struct f2fs_inode, i_extra_end) - \ 2799 offsetof(struct f2fs_inode, i_extra_isize)) \ 2800 2801 #define F2FS_OLD_ATTRIBUTE_SIZE (offsetof(struct f2fs_inode, i_addr)) 2802 #define F2FS_FITS_IN_INODE(f2fs_inode, extra_isize, field) \ 2803 ((offsetof(typeof(*(f2fs_inode)), field) + \ 2804 sizeof((f2fs_inode)->field)) \ 2805 <= (F2FS_OLD_ATTRIBUTE_SIZE + (extra_isize))) \ 2806 2807 static inline void f2fs_reset_iostat(struct f2fs_sb_info *sbi) 2808 { 2809 int i; 2810 2811 spin_lock(&sbi->iostat_lock); 2812 for (i = 0; i < NR_IO_TYPE; i++) 2813 sbi->write_iostat[i] = 0; 2814 spin_unlock(&sbi->iostat_lock); 2815 } 2816 2817 static inline void f2fs_update_iostat(struct f2fs_sb_info *sbi, 2818 enum iostat_type type, unsigned long long io_bytes) 2819 { 2820 if (!sbi->iostat_enable) 2821 return; 2822 spin_lock(&sbi->iostat_lock); 2823 sbi->write_iostat[type] += io_bytes; 2824 2825 if (type == APP_WRITE_IO || type == APP_DIRECT_IO) 2826 sbi->write_iostat[APP_BUFFERED_IO] = 2827 sbi->write_iostat[APP_WRITE_IO] - 2828 sbi->write_iostat[APP_DIRECT_IO]; 2829 spin_unlock(&sbi->iostat_lock); 2830 } 2831 2832 #define __is_large_section(sbi) ((sbi)->segs_per_sec > 1) 2833 2834 #define __is_meta_io(fio) (PAGE_TYPE_OF_BIO((fio)->type) == META && \ 2835 (!is_read_io((fio)->op) || (fio)->is_meta)) 2836 2837 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi, 2838 block_t blkaddr, int type); 2839 static inline void verify_blkaddr(struct f2fs_sb_info *sbi, 2840 block_t blkaddr, int type) 2841 { 2842 if (!f2fs_is_valid_blkaddr(sbi, blkaddr, type)) { 2843 f2fs_msg(sbi->sb, KERN_ERR, 2844 "invalid blkaddr: %u, type: %d, run fsck to fix.", 2845 blkaddr, type); 2846 f2fs_bug_on(sbi, 1); 2847 } 2848 } 2849 2850 static inline bool __is_valid_data_blkaddr(block_t blkaddr) 2851 { 2852 if (blkaddr == NEW_ADDR || blkaddr == NULL_ADDR) 2853 return false; 2854 return true; 2855 } 2856 2857 static inline bool is_valid_data_blkaddr(struct f2fs_sb_info *sbi, 2858 block_t blkaddr) 2859 { 2860 if (!__is_valid_data_blkaddr(blkaddr)) 2861 return false; 2862 verify_blkaddr(sbi, blkaddr, DATA_GENERIC); 2863 return true; 2864 } 2865 2866 static inline void f2fs_set_page_private(struct page *page, 2867 unsigned long data) 2868 { 2869 if (PagePrivate(page)) 2870 return; 2871 2872 get_page(page); 2873 SetPagePrivate(page); 2874 set_page_private(page, data); 2875 } 2876 2877 static inline void f2fs_clear_page_private(struct page *page) 2878 { 2879 if (!PagePrivate(page)) 2880 return; 2881 2882 set_page_private(page, 0); 2883 ClearPagePrivate(page); 2884 f2fs_put_page(page, 0); 2885 } 2886 2887 /* 2888 * file.c 2889 */ 2890 int f2fs_sync_file(struct file *file, loff_t start, loff_t end, int datasync); 2891 void f2fs_truncate_data_blocks(struct dnode_of_data *dn); 2892 int f2fs_truncate_blocks(struct inode *inode, u64 from, bool lock); 2893 int f2fs_truncate(struct inode *inode); 2894 int f2fs_getattr(const struct path *path, struct kstat *stat, 2895 u32 request_mask, unsigned int flags); 2896 int f2fs_setattr(struct dentry *dentry, struct iattr *attr); 2897 int f2fs_truncate_hole(struct inode *inode, pgoff_t pg_start, pgoff_t pg_end); 2898 void f2fs_truncate_data_blocks_range(struct dnode_of_data *dn, int count); 2899 int f2fs_precache_extents(struct inode *inode); 2900 long f2fs_ioctl(struct file *filp, unsigned int cmd, unsigned long arg); 2901 long f2fs_compat_ioctl(struct file *file, unsigned int cmd, unsigned long arg); 2902 int f2fs_transfer_project_quota(struct inode *inode, kprojid_t kprojid); 2903 int f2fs_pin_file_control(struct inode *inode, bool inc); 2904 2905 /* 2906 * inode.c 2907 */ 2908 void f2fs_set_inode_flags(struct inode *inode); 2909 bool f2fs_inode_chksum_verify(struct f2fs_sb_info *sbi, struct page *page); 2910 void f2fs_inode_chksum_set(struct f2fs_sb_info *sbi, struct page *page); 2911 struct inode *f2fs_iget(struct super_block *sb, unsigned long ino); 2912 struct inode *f2fs_iget_retry(struct super_block *sb, unsigned long ino); 2913 int f2fs_try_to_free_nats(struct f2fs_sb_info *sbi, int nr_shrink); 2914 void f2fs_update_inode(struct inode *inode, struct page *node_page); 2915 void f2fs_update_inode_page(struct inode *inode); 2916 int f2fs_write_inode(struct inode *inode, struct writeback_control *wbc); 2917 void f2fs_evict_inode(struct inode *inode); 2918 void f2fs_handle_failed_inode(struct inode *inode); 2919 2920 /* 2921 * namei.c 2922 */ 2923 int f2fs_update_extension_list(struct f2fs_sb_info *sbi, const char *name, 2924 bool hot, bool set); 2925 struct dentry *f2fs_get_parent(struct dentry *child); 2926 2927 /* 2928 * dir.c 2929 */ 2930 unsigned char f2fs_get_de_type(struct f2fs_dir_entry *de); 2931 struct f2fs_dir_entry *f2fs_find_target_dentry(struct fscrypt_name *fname, 2932 f2fs_hash_t namehash, int *max_slots, 2933 struct f2fs_dentry_ptr *d); 2934 int f2fs_fill_dentries(struct dir_context *ctx, struct f2fs_dentry_ptr *d, 2935 unsigned int start_pos, struct fscrypt_str *fstr); 2936 void f2fs_do_make_empty_dir(struct inode *inode, struct inode *parent, 2937 struct f2fs_dentry_ptr *d); 2938 struct page *f2fs_init_inode_metadata(struct inode *inode, struct inode *dir, 2939 const struct qstr *new_name, 2940 const struct qstr *orig_name, struct page *dpage); 2941 void f2fs_update_parent_metadata(struct inode *dir, struct inode *inode, 2942 unsigned int current_depth); 2943 int f2fs_room_for_filename(const void *bitmap, int slots, int max_slots); 2944 void f2fs_drop_nlink(struct inode *dir, struct inode *inode); 2945 struct f2fs_dir_entry *__f2fs_find_entry(struct inode *dir, 2946 struct fscrypt_name *fname, struct page **res_page); 2947 struct f2fs_dir_entry *f2fs_find_entry(struct inode *dir, 2948 const struct qstr *child, struct page **res_page); 2949 struct f2fs_dir_entry *f2fs_parent_dir(struct inode *dir, struct page **p); 2950 ino_t f2fs_inode_by_name(struct inode *dir, const struct qstr *qstr, 2951 struct page **page); 2952 void f2fs_set_link(struct inode *dir, struct f2fs_dir_entry *de, 2953 struct page *page, struct inode *inode); 2954 void f2fs_update_dentry(nid_t ino, umode_t mode, struct f2fs_dentry_ptr *d, 2955 const struct qstr *name, f2fs_hash_t name_hash, 2956 unsigned int bit_pos); 2957 int f2fs_add_regular_entry(struct inode *dir, const struct qstr *new_name, 2958 const struct qstr *orig_name, 2959 struct inode *inode, nid_t ino, umode_t mode); 2960 int f2fs_add_dentry(struct inode *dir, struct fscrypt_name *fname, 2961 struct inode *inode, nid_t ino, umode_t mode); 2962 int f2fs_do_add_link(struct inode *dir, const struct qstr *name, 2963 struct inode *inode, nid_t ino, umode_t mode); 2964 void f2fs_delete_entry(struct f2fs_dir_entry *dentry, struct page *page, 2965 struct inode *dir, struct inode *inode); 2966 int f2fs_do_tmpfile(struct inode *inode, struct inode *dir); 2967 bool f2fs_empty_dir(struct inode *dir); 2968 2969 static inline int f2fs_add_link(struct dentry *dentry, struct inode *inode) 2970 { 2971 return f2fs_do_add_link(d_inode(dentry->d_parent), &dentry->d_name, 2972 inode, inode->i_ino, inode->i_mode); 2973 } 2974 2975 /* 2976 * super.c 2977 */ 2978 int f2fs_inode_dirtied(struct inode *inode, bool sync); 2979 void f2fs_inode_synced(struct inode *inode); 2980 int f2fs_enable_quota_files(struct f2fs_sb_info *sbi, bool rdonly); 2981 int f2fs_quota_sync(struct super_block *sb, int type); 2982 void f2fs_quota_off_umount(struct super_block *sb); 2983 int f2fs_commit_super(struct f2fs_sb_info *sbi, bool recover); 2984 int f2fs_sync_fs(struct super_block *sb, int sync); 2985 extern __printf(3, 4) 2986 void f2fs_msg(struct super_block *sb, const char *level, const char *fmt, ...); 2987 int f2fs_sanity_check_ckpt(struct f2fs_sb_info *sbi); 2988 2989 /* 2990 * hash.c 2991 */ 2992 f2fs_hash_t f2fs_dentry_hash(const struct qstr *name_info, 2993 struct fscrypt_name *fname); 2994 2995 /* 2996 * node.c 2997 */ 2998 struct dnode_of_data; 2999 struct node_info; 3000 3001 int f2fs_check_nid_range(struct f2fs_sb_info *sbi, nid_t nid); 3002 bool f2fs_available_free_memory(struct f2fs_sb_info *sbi, int type); 3003 bool f2fs_in_warm_node_list(struct f2fs_sb_info *sbi, struct page *page); 3004 void f2fs_init_fsync_node_info(struct f2fs_sb_info *sbi); 3005 void f2fs_del_fsync_node_entry(struct f2fs_sb_info *sbi, struct page *page); 3006 void f2fs_reset_fsync_node_info(struct f2fs_sb_info *sbi); 3007 int f2fs_need_dentry_mark(struct f2fs_sb_info *sbi, nid_t nid); 3008 bool f2fs_is_checkpointed_node(struct f2fs_sb_info *sbi, nid_t nid); 3009 bool f2fs_need_inode_block_update(struct f2fs_sb_info *sbi, nid_t ino); 3010 int f2fs_get_node_info(struct f2fs_sb_info *sbi, nid_t nid, 3011 struct node_info *ni); 3012 pgoff_t f2fs_get_next_page_offset(struct dnode_of_data *dn, pgoff_t pgofs); 3013 int f2fs_get_dnode_of_data(struct dnode_of_data *dn, pgoff_t index, int mode); 3014 int f2fs_truncate_inode_blocks(struct inode *inode, pgoff_t from); 3015 int f2fs_truncate_xattr_node(struct inode *inode); 3016 int f2fs_wait_on_node_pages_writeback(struct f2fs_sb_info *sbi, 3017 unsigned int seq_id); 3018 int f2fs_remove_inode_page(struct inode *inode); 3019 struct page *f2fs_new_inode_page(struct inode *inode); 3020 struct page *f2fs_new_node_page(struct dnode_of_data *dn, unsigned int ofs); 3021 void f2fs_ra_node_page(struct f2fs_sb_info *sbi, nid_t nid); 3022 struct page *f2fs_get_node_page(struct f2fs_sb_info *sbi, pgoff_t nid); 3023 struct page *f2fs_get_node_page_ra(struct page *parent, int start); 3024 int f2fs_move_node_page(struct page *node_page, int gc_type); 3025 int f2fs_fsync_node_pages(struct f2fs_sb_info *sbi, struct inode *inode, 3026 struct writeback_control *wbc, bool atomic, 3027 unsigned int *seq_id); 3028 int f2fs_sync_node_pages(struct f2fs_sb_info *sbi, 3029 struct writeback_control *wbc, 3030 bool do_balance, enum iostat_type io_type); 3031 int f2fs_build_free_nids(struct f2fs_sb_info *sbi, bool sync, bool mount); 3032 bool f2fs_alloc_nid(struct f2fs_sb_info *sbi, nid_t *nid); 3033 void f2fs_alloc_nid_done(struct f2fs_sb_info *sbi, nid_t nid); 3034 void f2fs_alloc_nid_failed(struct f2fs_sb_info *sbi, nid_t nid); 3035 int f2fs_try_to_free_nids(struct f2fs_sb_info *sbi, int nr_shrink); 3036 void f2fs_recover_inline_xattr(struct inode *inode, struct page *page); 3037 int f2fs_recover_xattr_data(struct inode *inode, struct page *page); 3038 int f2fs_recover_inode_page(struct f2fs_sb_info *sbi, struct page *page); 3039 int f2fs_restore_node_summary(struct f2fs_sb_info *sbi, 3040 unsigned int segno, struct f2fs_summary_block *sum); 3041 int f2fs_flush_nat_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3042 int f2fs_build_node_manager(struct f2fs_sb_info *sbi); 3043 void f2fs_destroy_node_manager(struct f2fs_sb_info *sbi); 3044 int __init f2fs_create_node_manager_caches(void); 3045 void f2fs_destroy_node_manager_caches(void); 3046 3047 /* 3048 * segment.c 3049 */ 3050 bool f2fs_need_SSR(struct f2fs_sb_info *sbi); 3051 void f2fs_register_inmem_page(struct inode *inode, struct page *page); 3052 void f2fs_drop_inmem_pages_all(struct f2fs_sb_info *sbi, bool gc_failure); 3053 void f2fs_drop_inmem_pages(struct inode *inode); 3054 void f2fs_drop_inmem_page(struct inode *inode, struct page *page); 3055 int f2fs_commit_inmem_pages(struct inode *inode); 3056 void f2fs_balance_fs(struct f2fs_sb_info *sbi, bool need); 3057 void f2fs_balance_fs_bg(struct f2fs_sb_info *sbi); 3058 int f2fs_issue_flush(struct f2fs_sb_info *sbi, nid_t ino); 3059 int f2fs_create_flush_cmd_control(struct f2fs_sb_info *sbi); 3060 int f2fs_flush_device_cache(struct f2fs_sb_info *sbi); 3061 void f2fs_destroy_flush_cmd_control(struct f2fs_sb_info *sbi, bool free); 3062 void f2fs_invalidate_blocks(struct f2fs_sb_info *sbi, block_t addr); 3063 bool f2fs_is_checkpointed_data(struct f2fs_sb_info *sbi, block_t blkaddr); 3064 void f2fs_drop_discard_cmd(struct f2fs_sb_info *sbi); 3065 void f2fs_stop_discard_thread(struct f2fs_sb_info *sbi); 3066 bool f2fs_issue_discard_timeout(struct f2fs_sb_info *sbi); 3067 void f2fs_clear_prefree_segments(struct f2fs_sb_info *sbi, 3068 struct cp_control *cpc); 3069 void f2fs_dirty_to_prefree(struct f2fs_sb_info *sbi); 3070 int f2fs_disable_cp_again(struct f2fs_sb_info *sbi); 3071 void f2fs_release_discard_addrs(struct f2fs_sb_info *sbi); 3072 int f2fs_npages_for_summary_flush(struct f2fs_sb_info *sbi, bool for_ra); 3073 void f2fs_allocate_new_segments(struct f2fs_sb_info *sbi); 3074 int f2fs_trim_fs(struct f2fs_sb_info *sbi, struct fstrim_range *range); 3075 bool f2fs_exist_trim_candidates(struct f2fs_sb_info *sbi, 3076 struct cp_control *cpc); 3077 struct page *f2fs_get_sum_page(struct f2fs_sb_info *sbi, unsigned int segno); 3078 void f2fs_update_meta_page(struct f2fs_sb_info *sbi, void *src, 3079 block_t blk_addr); 3080 void f2fs_do_write_meta_page(struct f2fs_sb_info *sbi, struct page *page, 3081 enum iostat_type io_type); 3082 void f2fs_do_write_node_page(unsigned int nid, struct f2fs_io_info *fio); 3083 void f2fs_outplace_write_data(struct dnode_of_data *dn, 3084 struct f2fs_io_info *fio); 3085 int f2fs_inplace_write_data(struct f2fs_io_info *fio); 3086 void f2fs_do_replace_block(struct f2fs_sb_info *sbi, struct f2fs_summary *sum, 3087 block_t old_blkaddr, block_t new_blkaddr, 3088 bool recover_curseg, bool recover_newaddr); 3089 void f2fs_replace_block(struct f2fs_sb_info *sbi, struct dnode_of_data *dn, 3090 block_t old_addr, block_t new_addr, 3091 unsigned char version, bool recover_curseg, 3092 bool recover_newaddr); 3093 void f2fs_allocate_data_block(struct f2fs_sb_info *sbi, struct page *page, 3094 block_t old_blkaddr, block_t *new_blkaddr, 3095 struct f2fs_summary *sum, int type, 3096 struct f2fs_io_info *fio, bool add_list); 3097 void f2fs_wait_on_page_writeback(struct page *page, 3098 enum page_type type, bool ordered, bool locked); 3099 void f2fs_wait_on_block_writeback(struct inode *inode, block_t blkaddr); 3100 void f2fs_wait_on_block_writeback_range(struct inode *inode, block_t blkaddr, 3101 block_t len); 3102 void f2fs_write_data_summaries(struct f2fs_sb_info *sbi, block_t start_blk); 3103 void f2fs_write_node_summaries(struct f2fs_sb_info *sbi, block_t start_blk); 3104 int f2fs_lookup_journal_in_cursum(struct f2fs_journal *journal, int type, 3105 unsigned int val, int alloc); 3106 void f2fs_flush_sit_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3107 int f2fs_build_segment_manager(struct f2fs_sb_info *sbi); 3108 void f2fs_destroy_segment_manager(struct f2fs_sb_info *sbi); 3109 int __init f2fs_create_segment_manager_caches(void); 3110 void f2fs_destroy_segment_manager_caches(void); 3111 int f2fs_rw_hint_to_seg_type(enum rw_hint hint); 3112 enum rw_hint f2fs_io_type_to_rw_hint(struct f2fs_sb_info *sbi, 3113 enum page_type type, enum temp_type temp); 3114 3115 /* 3116 * checkpoint.c 3117 */ 3118 void f2fs_stop_checkpoint(struct f2fs_sb_info *sbi, bool end_io); 3119 struct page *f2fs_grab_meta_page(struct f2fs_sb_info *sbi, pgoff_t index); 3120 struct page *f2fs_get_meta_page(struct f2fs_sb_info *sbi, pgoff_t index); 3121 struct page *f2fs_get_meta_page_nofail(struct f2fs_sb_info *sbi, pgoff_t index); 3122 struct page *f2fs_get_tmp_page(struct f2fs_sb_info *sbi, pgoff_t index); 3123 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi, 3124 block_t blkaddr, int type); 3125 int f2fs_ra_meta_pages(struct f2fs_sb_info *sbi, block_t start, int nrpages, 3126 int type, bool sync); 3127 void f2fs_ra_meta_pages_cond(struct f2fs_sb_info *sbi, pgoff_t index); 3128 long f2fs_sync_meta_pages(struct f2fs_sb_info *sbi, enum page_type type, 3129 long nr_to_write, enum iostat_type io_type); 3130 void f2fs_add_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type); 3131 void f2fs_remove_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type); 3132 void f2fs_release_ino_entry(struct f2fs_sb_info *sbi, bool all); 3133 bool f2fs_exist_written_data(struct f2fs_sb_info *sbi, nid_t ino, int mode); 3134 void f2fs_set_dirty_device(struct f2fs_sb_info *sbi, nid_t ino, 3135 unsigned int devidx, int type); 3136 bool f2fs_is_dirty_device(struct f2fs_sb_info *sbi, nid_t ino, 3137 unsigned int devidx, int type); 3138 int f2fs_sync_inode_meta(struct f2fs_sb_info *sbi); 3139 int f2fs_acquire_orphan_inode(struct f2fs_sb_info *sbi); 3140 void f2fs_release_orphan_inode(struct f2fs_sb_info *sbi); 3141 void f2fs_add_orphan_inode(struct inode *inode); 3142 void f2fs_remove_orphan_inode(struct f2fs_sb_info *sbi, nid_t ino); 3143 int f2fs_recover_orphan_inodes(struct f2fs_sb_info *sbi); 3144 int f2fs_get_valid_checkpoint(struct f2fs_sb_info *sbi); 3145 void f2fs_update_dirty_page(struct inode *inode, struct page *page); 3146 void f2fs_remove_dirty_inode(struct inode *inode); 3147 int f2fs_sync_dirty_inodes(struct f2fs_sb_info *sbi, enum inode_type type); 3148 void f2fs_wait_on_all_pages_writeback(struct f2fs_sb_info *sbi); 3149 int f2fs_write_checkpoint(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3150 void f2fs_init_ino_entry_info(struct f2fs_sb_info *sbi); 3151 int __init f2fs_create_checkpoint_caches(void); 3152 void f2fs_destroy_checkpoint_caches(void); 3153 3154 /* 3155 * data.c 3156 */ 3157 int f2fs_init_post_read_processing(void); 3158 void f2fs_destroy_post_read_processing(void); 3159 void f2fs_submit_merged_write(struct f2fs_sb_info *sbi, enum page_type type); 3160 void f2fs_submit_merged_write_cond(struct f2fs_sb_info *sbi, 3161 struct inode *inode, struct page *page, 3162 nid_t ino, enum page_type type); 3163 void f2fs_flush_merged_writes(struct f2fs_sb_info *sbi); 3164 int f2fs_submit_page_bio(struct f2fs_io_info *fio); 3165 void f2fs_submit_page_write(struct f2fs_io_info *fio); 3166 struct block_device *f2fs_target_device(struct f2fs_sb_info *sbi, 3167 block_t blk_addr, struct bio *bio); 3168 int f2fs_target_device_index(struct f2fs_sb_info *sbi, block_t blkaddr); 3169 void f2fs_set_data_blkaddr(struct dnode_of_data *dn); 3170 void f2fs_update_data_blkaddr(struct dnode_of_data *dn, block_t blkaddr); 3171 int f2fs_reserve_new_blocks(struct dnode_of_data *dn, blkcnt_t count); 3172 int f2fs_reserve_new_block(struct dnode_of_data *dn); 3173 int f2fs_get_block(struct dnode_of_data *dn, pgoff_t index); 3174 int f2fs_preallocate_blocks(struct kiocb *iocb, struct iov_iter *from); 3175 int f2fs_reserve_block(struct dnode_of_data *dn, pgoff_t index); 3176 struct page *f2fs_get_read_data_page(struct inode *inode, pgoff_t index, 3177 int op_flags, bool for_write); 3178 struct page *f2fs_find_data_page(struct inode *inode, pgoff_t index); 3179 struct page *f2fs_get_lock_data_page(struct inode *inode, pgoff_t index, 3180 bool for_write); 3181 struct page *f2fs_get_new_data_page(struct inode *inode, 3182 struct page *ipage, pgoff_t index, bool new_i_size); 3183 int f2fs_do_write_data_page(struct f2fs_io_info *fio); 3184 void __do_map_lock(struct f2fs_sb_info *sbi, int flag, bool lock); 3185 int f2fs_map_blocks(struct inode *inode, struct f2fs_map_blocks *map, 3186 int create, int flag); 3187 int f2fs_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, 3188 u64 start, u64 len); 3189 bool f2fs_should_update_inplace(struct inode *inode, struct f2fs_io_info *fio); 3190 bool f2fs_should_update_outplace(struct inode *inode, struct f2fs_io_info *fio); 3191 void f2fs_invalidate_page(struct page *page, unsigned int offset, 3192 unsigned int length); 3193 int f2fs_release_page(struct page *page, gfp_t wait); 3194 #ifdef CONFIG_MIGRATION 3195 int f2fs_migrate_page(struct address_space *mapping, struct page *newpage, 3196 struct page *page, enum migrate_mode mode); 3197 #endif 3198 bool f2fs_overwrite_io(struct inode *inode, loff_t pos, size_t len); 3199 void f2fs_clear_page_cache_dirty_tag(struct page *page); 3200 3201 /* 3202 * gc.c 3203 */ 3204 int f2fs_start_gc_thread(struct f2fs_sb_info *sbi); 3205 void f2fs_stop_gc_thread(struct f2fs_sb_info *sbi); 3206 block_t f2fs_start_bidx_of_node(unsigned int node_ofs, struct inode *inode); 3207 int f2fs_gc(struct f2fs_sb_info *sbi, bool sync, bool background, 3208 unsigned int segno); 3209 void f2fs_build_gc_manager(struct f2fs_sb_info *sbi); 3210 3211 /* 3212 * recovery.c 3213 */ 3214 int f2fs_recover_fsync_data(struct f2fs_sb_info *sbi, bool check_only); 3215 bool f2fs_space_for_roll_forward(struct f2fs_sb_info *sbi); 3216 3217 /* 3218 * debug.c 3219 */ 3220 #ifdef CONFIG_F2FS_STAT_FS 3221 struct f2fs_stat_info { 3222 struct list_head stat_list; 3223 struct f2fs_sb_info *sbi; 3224 int all_area_segs, sit_area_segs, nat_area_segs, ssa_area_segs; 3225 int main_area_segs, main_area_sections, main_area_zones; 3226 unsigned long long hit_largest, hit_cached, hit_rbtree; 3227 unsigned long long hit_total, total_ext; 3228 int ext_tree, zombie_tree, ext_node; 3229 int ndirty_node, ndirty_dent, ndirty_meta, ndirty_imeta; 3230 int ndirty_data, ndirty_qdata; 3231 int inmem_pages; 3232 unsigned int ndirty_dirs, ndirty_files, nquota_files, ndirty_all; 3233 int nats, dirty_nats, sits, dirty_sits; 3234 int free_nids, avail_nids, alloc_nids; 3235 int total_count, utilization; 3236 int bg_gc, nr_wb_cp_data, nr_wb_data; 3237 int nr_rd_data, nr_rd_node, nr_rd_meta; 3238 int nr_dio_read, nr_dio_write; 3239 unsigned int io_skip_bggc, other_skip_bggc; 3240 int nr_flushing, nr_flushed, flush_list_empty; 3241 int nr_discarding, nr_discarded; 3242 int nr_discard_cmd; 3243 unsigned int undiscard_blks; 3244 int inline_xattr, inline_inode, inline_dir, append, update, orphans; 3245 int aw_cnt, max_aw_cnt, vw_cnt, max_vw_cnt; 3246 unsigned int valid_count, valid_node_count, valid_inode_count, discard_blks; 3247 unsigned int bimodal, avg_vblocks; 3248 int util_free, util_valid, util_invalid; 3249 int rsvd_segs, overp_segs; 3250 int dirty_count, node_pages, meta_pages; 3251 int prefree_count, call_count, cp_count, bg_cp_count; 3252 int tot_segs, node_segs, data_segs, free_segs, free_secs; 3253 int bg_node_segs, bg_data_segs; 3254 int tot_blks, data_blks, node_blks; 3255 int bg_data_blks, bg_node_blks; 3256 unsigned long long skipped_atomic_files[2]; 3257 int curseg[NR_CURSEG_TYPE]; 3258 int cursec[NR_CURSEG_TYPE]; 3259 int curzone[NR_CURSEG_TYPE]; 3260 3261 unsigned int meta_count[META_MAX]; 3262 unsigned int segment_count[2]; 3263 unsigned int block_count[2]; 3264 unsigned int inplace_count; 3265 unsigned long long base_mem, cache_mem, page_mem; 3266 }; 3267 3268 static inline struct f2fs_stat_info *F2FS_STAT(struct f2fs_sb_info *sbi) 3269 { 3270 return (struct f2fs_stat_info *)sbi->stat_info; 3271 } 3272 3273 #define stat_inc_cp_count(si) ((si)->cp_count++) 3274 #define stat_inc_bg_cp_count(si) ((si)->bg_cp_count++) 3275 #define stat_inc_call_count(si) ((si)->call_count++) 3276 #define stat_inc_bggc_count(sbi) ((sbi)->bg_gc++) 3277 #define stat_io_skip_bggc_count(sbi) ((sbi)->io_skip_bggc++) 3278 #define stat_other_skip_bggc_count(sbi) ((sbi)->other_skip_bggc++) 3279 #define stat_inc_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]++) 3280 #define stat_dec_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]--) 3281 #define stat_inc_total_hit(sbi) (atomic64_inc(&(sbi)->total_hit_ext)) 3282 #define stat_inc_rbtree_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_rbtree)) 3283 #define stat_inc_largest_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_largest)) 3284 #define stat_inc_cached_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_cached)) 3285 #define stat_inc_inline_xattr(inode) \ 3286 do { \ 3287 if (f2fs_has_inline_xattr(inode)) \ 3288 (atomic_inc(&F2FS_I_SB(inode)->inline_xattr)); \ 3289 } while (0) 3290 #define stat_dec_inline_xattr(inode) \ 3291 do { \ 3292 if (f2fs_has_inline_xattr(inode)) \ 3293 (atomic_dec(&F2FS_I_SB(inode)->inline_xattr)); \ 3294 } while (0) 3295 #define stat_inc_inline_inode(inode) \ 3296 do { \ 3297 if (f2fs_has_inline_data(inode)) \ 3298 (atomic_inc(&F2FS_I_SB(inode)->inline_inode)); \ 3299 } while (0) 3300 #define stat_dec_inline_inode(inode) \ 3301 do { \ 3302 if (f2fs_has_inline_data(inode)) \ 3303 (atomic_dec(&F2FS_I_SB(inode)->inline_inode)); \ 3304 } while (0) 3305 #define stat_inc_inline_dir(inode) \ 3306 do { \ 3307 if (f2fs_has_inline_dentry(inode)) \ 3308 (atomic_inc(&F2FS_I_SB(inode)->inline_dir)); \ 3309 } while (0) 3310 #define stat_dec_inline_dir(inode) \ 3311 do { \ 3312 if (f2fs_has_inline_dentry(inode)) \ 3313 (atomic_dec(&F2FS_I_SB(inode)->inline_dir)); \ 3314 } while (0) 3315 #define stat_inc_meta_count(sbi, blkaddr) \ 3316 do { \ 3317 if (blkaddr < SIT_I(sbi)->sit_base_addr) \ 3318 atomic_inc(&(sbi)->meta_count[META_CP]); \ 3319 else if (blkaddr < NM_I(sbi)->nat_blkaddr) \ 3320 atomic_inc(&(sbi)->meta_count[META_SIT]); \ 3321 else if (blkaddr < SM_I(sbi)->ssa_blkaddr) \ 3322 atomic_inc(&(sbi)->meta_count[META_NAT]); \ 3323 else if (blkaddr < SM_I(sbi)->main_blkaddr) \ 3324 atomic_inc(&(sbi)->meta_count[META_SSA]); \ 3325 } while (0) 3326 #define stat_inc_seg_type(sbi, curseg) \ 3327 ((sbi)->segment_count[(curseg)->alloc_type]++) 3328 #define stat_inc_block_count(sbi, curseg) \ 3329 ((sbi)->block_count[(curseg)->alloc_type]++) 3330 #define stat_inc_inplace_blocks(sbi) \ 3331 (atomic_inc(&(sbi)->inplace_count)) 3332 #define stat_inc_atomic_write(inode) \ 3333 (atomic_inc(&F2FS_I_SB(inode)->aw_cnt)) 3334 #define stat_dec_atomic_write(inode) \ 3335 (atomic_dec(&F2FS_I_SB(inode)->aw_cnt)) 3336 #define stat_update_max_atomic_write(inode) \ 3337 do { \ 3338 int cur = atomic_read(&F2FS_I_SB(inode)->aw_cnt); \ 3339 int max = atomic_read(&F2FS_I_SB(inode)->max_aw_cnt); \ 3340 if (cur > max) \ 3341 atomic_set(&F2FS_I_SB(inode)->max_aw_cnt, cur); \ 3342 } while (0) 3343 #define stat_inc_volatile_write(inode) \ 3344 (atomic_inc(&F2FS_I_SB(inode)->vw_cnt)) 3345 #define stat_dec_volatile_write(inode) \ 3346 (atomic_dec(&F2FS_I_SB(inode)->vw_cnt)) 3347 #define stat_update_max_volatile_write(inode) \ 3348 do { \ 3349 int cur = atomic_read(&F2FS_I_SB(inode)->vw_cnt); \ 3350 int max = atomic_read(&F2FS_I_SB(inode)->max_vw_cnt); \ 3351 if (cur > max) \ 3352 atomic_set(&F2FS_I_SB(inode)->max_vw_cnt, cur); \ 3353 } while (0) 3354 #define stat_inc_seg_count(sbi, type, gc_type) \ 3355 do { \ 3356 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 3357 si->tot_segs++; \ 3358 if ((type) == SUM_TYPE_DATA) { \ 3359 si->data_segs++; \ 3360 si->bg_data_segs += (gc_type == BG_GC) ? 1 : 0; \ 3361 } else { \ 3362 si->node_segs++; \ 3363 si->bg_node_segs += (gc_type == BG_GC) ? 1 : 0; \ 3364 } \ 3365 } while (0) 3366 3367 #define stat_inc_tot_blk_count(si, blks) \ 3368 ((si)->tot_blks += (blks)) 3369 3370 #define stat_inc_data_blk_count(sbi, blks, gc_type) \ 3371 do { \ 3372 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 3373 stat_inc_tot_blk_count(si, blks); \ 3374 si->data_blks += (blks); \ 3375 si->bg_data_blks += ((gc_type) == BG_GC) ? (blks) : 0; \ 3376 } while (0) 3377 3378 #define stat_inc_node_blk_count(sbi, blks, gc_type) \ 3379 do { \ 3380 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 3381 stat_inc_tot_blk_count(si, blks); \ 3382 si->node_blks += (blks); \ 3383 si->bg_node_blks += ((gc_type) == BG_GC) ? (blks) : 0; \ 3384 } while (0) 3385 3386 int f2fs_build_stats(struct f2fs_sb_info *sbi); 3387 void f2fs_destroy_stats(struct f2fs_sb_info *sbi); 3388 void __init f2fs_create_root_stats(void); 3389 void f2fs_destroy_root_stats(void); 3390 #else 3391 #define stat_inc_cp_count(si) do { } while (0) 3392 #define stat_inc_bg_cp_count(si) do { } while (0) 3393 #define stat_inc_call_count(si) do { } while (0) 3394 #define stat_inc_bggc_count(si) do { } while (0) 3395 #define stat_io_skip_bggc_count(sbi) do { } while (0) 3396 #define stat_other_skip_bggc_count(sbi) do { } while (0) 3397 #define stat_inc_dirty_inode(sbi, type) do { } while (0) 3398 #define stat_dec_dirty_inode(sbi, type) do { } while (0) 3399 #define stat_inc_total_hit(sb) do { } while (0) 3400 #define stat_inc_rbtree_node_hit(sb) do { } while (0) 3401 #define stat_inc_largest_node_hit(sbi) do { } while (0) 3402 #define stat_inc_cached_node_hit(sbi) do { } while (0) 3403 #define stat_inc_inline_xattr(inode) do { } while (0) 3404 #define stat_dec_inline_xattr(inode) do { } while (0) 3405 #define stat_inc_inline_inode(inode) do { } while (0) 3406 #define stat_dec_inline_inode(inode) do { } while (0) 3407 #define stat_inc_inline_dir(inode) do { } while (0) 3408 #define stat_dec_inline_dir(inode) do { } while (0) 3409 #define stat_inc_atomic_write(inode) do { } while (0) 3410 #define stat_dec_atomic_write(inode) do { } while (0) 3411 #define stat_update_max_atomic_write(inode) do { } while (0) 3412 #define stat_inc_volatile_write(inode) do { } while (0) 3413 #define stat_dec_volatile_write(inode) do { } while (0) 3414 #define stat_update_max_volatile_write(inode) do { } while (0) 3415 #define stat_inc_meta_count(sbi, blkaddr) do { } while (0) 3416 #define stat_inc_seg_type(sbi, curseg) do { } while (0) 3417 #define stat_inc_block_count(sbi, curseg) do { } while (0) 3418 #define stat_inc_inplace_blocks(sbi) do { } while (0) 3419 #define stat_inc_seg_count(sbi, type, gc_type) do { } while (0) 3420 #define stat_inc_tot_blk_count(si, blks) do { } while (0) 3421 #define stat_inc_data_blk_count(sbi, blks, gc_type) do { } while (0) 3422 #define stat_inc_node_blk_count(sbi, blks, gc_type) do { } while (0) 3423 3424 static inline int f2fs_build_stats(struct f2fs_sb_info *sbi) { return 0; } 3425 static inline void f2fs_destroy_stats(struct f2fs_sb_info *sbi) { } 3426 static inline void __init f2fs_create_root_stats(void) { } 3427 static inline void f2fs_destroy_root_stats(void) { } 3428 #endif 3429 3430 extern const struct file_operations f2fs_dir_operations; 3431 extern const struct file_operations f2fs_file_operations; 3432 extern const struct inode_operations f2fs_file_inode_operations; 3433 extern const struct address_space_operations f2fs_dblock_aops; 3434 extern const struct address_space_operations f2fs_node_aops; 3435 extern const struct address_space_operations f2fs_meta_aops; 3436 extern const struct inode_operations f2fs_dir_inode_operations; 3437 extern const struct inode_operations f2fs_symlink_inode_operations; 3438 extern const struct inode_operations f2fs_encrypted_symlink_inode_operations; 3439 extern const struct inode_operations f2fs_special_inode_operations; 3440 extern struct kmem_cache *f2fs_inode_entry_slab; 3441 3442 /* 3443 * inline.c 3444 */ 3445 bool f2fs_may_inline_data(struct inode *inode); 3446 bool f2fs_may_inline_dentry(struct inode *inode); 3447 void f2fs_do_read_inline_data(struct page *page, struct page *ipage); 3448 void f2fs_truncate_inline_inode(struct inode *inode, 3449 struct page *ipage, u64 from); 3450 int f2fs_read_inline_data(struct inode *inode, struct page *page); 3451 int f2fs_convert_inline_page(struct dnode_of_data *dn, struct page *page); 3452 int f2fs_convert_inline_inode(struct inode *inode); 3453 int f2fs_write_inline_data(struct inode *inode, struct page *page); 3454 bool f2fs_recover_inline_data(struct inode *inode, struct page *npage); 3455 struct f2fs_dir_entry *f2fs_find_in_inline_dir(struct inode *dir, 3456 struct fscrypt_name *fname, struct page **res_page); 3457 int f2fs_make_empty_inline_dir(struct inode *inode, struct inode *parent, 3458 struct page *ipage); 3459 int f2fs_add_inline_entry(struct inode *dir, const struct qstr *new_name, 3460 const struct qstr *orig_name, 3461 struct inode *inode, nid_t ino, umode_t mode); 3462 void f2fs_delete_inline_entry(struct f2fs_dir_entry *dentry, 3463 struct page *page, struct inode *dir, 3464 struct inode *inode); 3465 bool f2fs_empty_inline_dir(struct inode *dir); 3466 int f2fs_read_inline_dir(struct file *file, struct dir_context *ctx, 3467 struct fscrypt_str *fstr); 3468 int f2fs_inline_data_fiemap(struct inode *inode, 3469 struct fiemap_extent_info *fieinfo, 3470 __u64 start, __u64 len); 3471 3472 /* 3473 * shrinker.c 3474 */ 3475 unsigned long f2fs_shrink_count(struct shrinker *shrink, 3476 struct shrink_control *sc); 3477 unsigned long f2fs_shrink_scan(struct shrinker *shrink, 3478 struct shrink_control *sc); 3479 void f2fs_join_shrinker(struct f2fs_sb_info *sbi); 3480 void f2fs_leave_shrinker(struct f2fs_sb_info *sbi); 3481 3482 /* 3483 * extent_cache.c 3484 */ 3485 struct rb_entry *f2fs_lookup_rb_tree(struct rb_root_cached *root, 3486 struct rb_entry *cached_re, unsigned int ofs); 3487 struct rb_node **f2fs_lookup_rb_tree_for_insert(struct f2fs_sb_info *sbi, 3488 struct rb_root_cached *root, 3489 struct rb_node **parent, 3490 unsigned int ofs, bool *leftmost); 3491 struct rb_entry *f2fs_lookup_rb_tree_ret(struct rb_root_cached *root, 3492 struct rb_entry *cached_re, unsigned int ofs, 3493 struct rb_entry **prev_entry, struct rb_entry **next_entry, 3494 struct rb_node ***insert_p, struct rb_node **insert_parent, 3495 bool force, bool *leftmost); 3496 bool f2fs_check_rb_tree_consistence(struct f2fs_sb_info *sbi, 3497 struct rb_root_cached *root); 3498 unsigned int f2fs_shrink_extent_tree(struct f2fs_sb_info *sbi, int nr_shrink); 3499 bool f2fs_init_extent_tree(struct inode *inode, struct f2fs_extent *i_ext); 3500 void f2fs_drop_extent_tree(struct inode *inode); 3501 unsigned int f2fs_destroy_extent_node(struct inode *inode); 3502 void f2fs_destroy_extent_tree(struct inode *inode); 3503 bool f2fs_lookup_extent_cache(struct inode *inode, pgoff_t pgofs, 3504 struct extent_info *ei); 3505 void f2fs_update_extent_cache(struct dnode_of_data *dn); 3506 void f2fs_update_extent_cache_range(struct dnode_of_data *dn, 3507 pgoff_t fofs, block_t blkaddr, unsigned int len); 3508 void f2fs_init_extent_cache_info(struct f2fs_sb_info *sbi); 3509 int __init f2fs_create_extent_cache(void); 3510 void f2fs_destroy_extent_cache(void); 3511 3512 /* 3513 * sysfs.c 3514 */ 3515 int __init f2fs_init_sysfs(void); 3516 void f2fs_exit_sysfs(void); 3517 int f2fs_register_sysfs(struct f2fs_sb_info *sbi); 3518 void f2fs_unregister_sysfs(struct f2fs_sb_info *sbi); 3519 3520 /* 3521 * crypto support 3522 */ 3523 static inline bool f2fs_encrypted_file(struct inode *inode) 3524 { 3525 return IS_ENCRYPTED(inode) && S_ISREG(inode->i_mode); 3526 } 3527 3528 static inline void f2fs_set_encrypted_inode(struct inode *inode) 3529 { 3530 #ifdef CONFIG_FS_ENCRYPTION 3531 file_set_encrypt(inode); 3532 f2fs_set_inode_flags(inode); 3533 #endif 3534 } 3535 3536 /* 3537 * Returns true if the reads of the inode's data need to undergo some 3538 * postprocessing step, like decryption or authenticity verification. 3539 */ 3540 static inline bool f2fs_post_read_required(struct inode *inode) 3541 { 3542 return f2fs_encrypted_file(inode); 3543 } 3544 3545 #define F2FS_FEATURE_FUNCS(name, flagname) \ 3546 static inline int f2fs_sb_has_##name(struct f2fs_sb_info *sbi) \ 3547 { \ 3548 return F2FS_HAS_FEATURE(sbi, F2FS_FEATURE_##flagname); \ 3549 } 3550 3551 F2FS_FEATURE_FUNCS(encrypt, ENCRYPT); 3552 F2FS_FEATURE_FUNCS(blkzoned, BLKZONED); 3553 F2FS_FEATURE_FUNCS(extra_attr, EXTRA_ATTR); 3554 F2FS_FEATURE_FUNCS(project_quota, PRJQUOTA); 3555 F2FS_FEATURE_FUNCS(inode_chksum, INODE_CHKSUM); 3556 F2FS_FEATURE_FUNCS(flexible_inline_xattr, FLEXIBLE_INLINE_XATTR); 3557 F2FS_FEATURE_FUNCS(quota_ino, QUOTA_INO); 3558 F2FS_FEATURE_FUNCS(inode_crtime, INODE_CRTIME); 3559 F2FS_FEATURE_FUNCS(lost_found, LOST_FOUND); 3560 F2FS_FEATURE_FUNCS(sb_chksum, SB_CHKSUM); 3561 3562 #ifdef CONFIG_BLK_DEV_ZONED 3563 static inline bool f2fs_blkz_is_seq(struct f2fs_sb_info *sbi, int devi, 3564 block_t blkaddr) 3565 { 3566 unsigned int zno = blkaddr >> sbi->log_blocks_per_blkz; 3567 3568 return test_bit(zno, FDEV(devi).blkz_seq); 3569 } 3570 #endif 3571 3572 static inline bool f2fs_hw_should_discard(struct f2fs_sb_info *sbi) 3573 { 3574 return f2fs_sb_has_blkzoned(sbi); 3575 } 3576 3577 static inline bool f2fs_bdev_support_discard(struct block_device *bdev) 3578 { 3579 return blk_queue_discard(bdev_get_queue(bdev)) || 3580 bdev_is_zoned(bdev); 3581 } 3582 3583 static inline bool f2fs_hw_support_discard(struct f2fs_sb_info *sbi) 3584 { 3585 int i; 3586 3587 if (!f2fs_is_multi_device(sbi)) 3588 return f2fs_bdev_support_discard(sbi->sb->s_bdev); 3589 3590 for (i = 0; i < sbi->s_ndevs; i++) 3591 if (f2fs_bdev_support_discard(FDEV(i).bdev)) 3592 return true; 3593 return false; 3594 } 3595 3596 static inline bool f2fs_realtime_discard_enable(struct f2fs_sb_info *sbi) 3597 { 3598 return (test_opt(sbi, DISCARD) && f2fs_hw_support_discard(sbi)) || 3599 f2fs_hw_should_discard(sbi); 3600 } 3601 3602 static inline void set_opt_mode(struct f2fs_sb_info *sbi, unsigned int mt) 3603 { 3604 clear_opt(sbi, ADAPTIVE); 3605 clear_opt(sbi, LFS); 3606 3607 switch (mt) { 3608 case F2FS_MOUNT_ADAPTIVE: 3609 set_opt(sbi, ADAPTIVE); 3610 break; 3611 case F2FS_MOUNT_LFS: 3612 set_opt(sbi, LFS); 3613 break; 3614 } 3615 } 3616 3617 static inline bool f2fs_may_encrypt(struct inode *inode) 3618 { 3619 #ifdef CONFIG_FS_ENCRYPTION 3620 umode_t mode = inode->i_mode; 3621 3622 return (S_ISREG(mode) || S_ISDIR(mode) || S_ISLNK(mode)); 3623 #else 3624 return false; 3625 #endif 3626 } 3627 3628 static inline int block_unaligned_IO(struct inode *inode, 3629 struct kiocb *iocb, struct iov_iter *iter) 3630 { 3631 unsigned int i_blkbits = READ_ONCE(inode->i_blkbits); 3632 unsigned int blocksize_mask = (1 << i_blkbits) - 1; 3633 loff_t offset = iocb->ki_pos; 3634 unsigned long align = offset | iov_iter_alignment(iter); 3635 3636 return align & blocksize_mask; 3637 } 3638 3639 static inline int allow_outplace_dio(struct inode *inode, 3640 struct kiocb *iocb, struct iov_iter *iter) 3641 { 3642 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 3643 int rw = iov_iter_rw(iter); 3644 3645 return (test_opt(sbi, LFS) && (rw == WRITE) && 3646 !block_unaligned_IO(inode, iocb, iter)); 3647 } 3648 3649 static inline bool f2fs_force_buffered_io(struct inode *inode, 3650 struct kiocb *iocb, struct iov_iter *iter) 3651 { 3652 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 3653 int rw = iov_iter_rw(iter); 3654 3655 if (f2fs_post_read_required(inode)) 3656 return true; 3657 if (f2fs_is_multi_device(sbi)) 3658 return true; 3659 /* 3660 * for blkzoned device, fallback direct IO to buffered IO, so 3661 * all IOs can be serialized by log-structured write. 3662 */ 3663 if (f2fs_sb_has_blkzoned(sbi)) 3664 return true; 3665 if (test_opt(sbi, LFS) && (rw == WRITE) && 3666 block_unaligned_IO(inode, iocb, iter)) 3667 return true; 3668 if (is_sbi_flag_set(F2FS_I_SB(inode), SBI_CP_DISABLED)) 3669 return true; 3670 3671 return false; 3672 } 3673 3674 #ifdef CONFIG_F2FS_FAULT_INJECTION 3675 extern void f2fs_build_fault_attr(struct f2fs_sb_info *sbi, unsigned int rate, 3676 unsigned int type); 3677 #else 3678 #define f2fs_build_fault_attr(sbi, rate, type) do { } while (0) 3679 #endif 3680 3681 static inline bool is_journalled_quota(struct f2fs_sb_info *sbi) 3682 { 3683 #ifdef CONFIG_QUOTA 3684 if (f2fs_sb_has_quota_ino(sbi)) 3685 return true; 3686 if (F2FS_OPTION(sbi).s_qf_names[USRQUOTA] || 3687 F2FS_OPTION(sbi).s_qf_names[GRPQUOTA] || 3688 F2FS_OPTION(sbi).s_qf_names[PRJQUOTA]) 3689 return true; 3690 #endif 3691 return false; 3692 } 3693 3694 #endif /* _LINUX_F2FS_H */ 3695