1 // SPDX-License-Identifier: GPL-2.0 2 /* 3 * fs/f2fs/f2fs.h 4 * 5 * Copyright (c) 2012 Samsung Electronics Co., Ltd. 6 * http://www.samsung.com/ 7 */ 8 #ifndef _LINUX_F2FS_H 9 #define _LINUX_F2FS_H 10 11 #include <linux/uio.h> 12 #include <linux/types.h> 13 #include <linux/page-flags.h> 14 #include <linux/buffer_head.h> 15 #include <linux/slab.h> 16 #include <linux/crc32.h> 17 #include <linux/magic.h> 18 #include <linux/kobject.h> 19 #include <linux/sched.h> 20 #include <linux/cred.h> 21 #include <linux/vmalloc.h> 22 #include <linux/bio.h> 23 #include <linux/blkdev.h> 24 #include <linux/quotaops.h> 25 #include <linux/part_stat.h> 26 #include <crypto/hash.h> 27 28 #include <linux/fscrypt.h> 29 #include <linux/fsverity.h> 30 31 #ifdef CONFIG_F2FS_CHECK_FS 32 #define f2fs_bug_on(sbi, condition) BUG_ON(condition) 33 #else 34 #define f2fs_bug_on(sbi, condition) \ 35 do { \ 36 if (unlikely(condition)) { \ 37 WARN_ON(1); \ 38 set_sbi_flag(sbi, SBI_NEED_FSCK); \ 39 } \ 40 } while (0) 41 #endif 42 43 enum { 44 FAULT_KMALLOC, 45 FAULT_KVMALLOC, 46 FAULT_PAGE_ALLOC, 47 FAULT_PAGE_GET, 48 FAULT_ALLOC_BIO, 49 FAULT_ALLOC_NID, 50 FAULT_ORPHAN, 51 FAULT_BLOCK, 52 FAULT_DIR_DEPTH, 53 FAULT_EVICT_INODE, 54 FAULT_TRUNCATE, 55 FAULT_READ_IO, 56 FAULT_CHECKPOINT, 57 FAULT_DISCARD, 58 FAULT_WRITE_IO, 59 FAULT_MAX, 60 }; 61 62 #ifdef CONFIG_F2FS_FAULT_INJECTION 63 #define F2FS_ALL_FAULT_TYPE ((1 << FAULT_MAX) - 1) 64 65 struct f2fs_fault_info { 66 atomic_t inject_ops; 67 unsigned int inject_rate; 68 unsigned int inject_type; 69 }; 70 71 extern const char *f2fs_fault_name[FAULT_MAX]; 72 #define IS_FAULT_SET(fi, type) ((fi)->inject_type & (1 << (type))) 73 #endif 74 75 /* 76 * For mount options 77 */ 78 #define F2FS_MOUNT_DISABLE_ROLL_FORWARD 0x00000002 79 #define F2FS_MOUNT_DISCARD 0x00000004 80 #define F2FS_MOUNT_NOHEAP 0x00000008 81 #define F2FS_MOUNT_XATTR_USER 0x00000010 82 #define F2FS_MOUNT_POSIX_ACL 0x00000020 83 #define F2FS_MOUNT_DISABLE_EXT_IDENTIFY 0x00000040 84 #define F2FS_MOUNT_INLINE_XATTR 0x00000080 85 #define F2FS_MOUNT_INLINE_DATA 0x00000100 86 #define F2FS_MOUNT_INLINE_DENTRY 0x00000200 87 #define F2FS_MOUNT_FLUSH_MERGE 0x00000400 88 #define F2FS_MOUNT_NOBARRIER 0x00000800 89 #define F2FS_MOUNT_FASTBOOT 0x00001000 90 #define F2FS_MOUNT_EXTENT_CACHE 0x00002000 91 #define F2FS_MOUNT_DATA_FLUSH 0x00008000 92 #define F2FS_MOUNT_FAULT_INJECTION 0x00010000 93 #define F2FS_MOUNT_USRQUOTA 0x00080000 94 #define F2FS_MOUNT_GRPQUOTA 0x00100000 95 #define F2FS_MOUNT_PRJQUOTA 0x00200000 96 #define F2FS_MOUNT_QUOTA 0x00400000 97 #define F2FS_MOUNT_INLINE_XATTR_SIZE 0x00800000 98 #define F2FS_MOUNT_RESERVE_ROOT 0x01000000 99 #define F2FS_MOUNT_DISABLE_CHECKPOINT 0x02000000 100 #define F2FS_MOUNT_NORECOVERY 0x04000000 101 102 #define F2FS_OPTION(sbi) ((sbi)->mount_opt) 103 #define clear_opt(sbi, option) (F2FS_OPTION(sbi).opt &= ~F2FS_MOUNT_##option) 104 #define set_opt(sbi, option) (F2FS_OPTION(sbi).opt |= F2FS_MOUNT_##option) 105 #define test_opt(sbi, option) (F2FS_OPTION(sbi).opt & F2FS_MOUNT_##option) 106 107 #define ver_after(a, b) (typecheck(unsigned long long, a) && \ 108 typecheck(unsigned long long, b) && \ 109 ((long long)((a) - (b)) > 0)) 110 111 typedef u32 block_t; /* 112 * should not change u32, since it is the on-disk block 113 * address format, __le32. 114 */ 115 typedef u32 nid_t; 116 117 #define COMPRESS_EXT_NUM 16 118 119 struct f2fs_mount_info { 120 unsigned int opt; 121 int write_io_size_bits; /* Write IO size bits */ 122 block_t root_reserved_blocks; /* root reserved blocks */ 123 kuid_t s_resuid; /* reserved blocks for uid */ 124 kgid_t s_resgid; /* reserved blocks for gid */ 125 int active_logs; /* # of active logs */ 126 int inline_xattr_size; /* inline xattr size */ 127 #ifdef CONFIG_F2FS_FAULT_INJECTION 128 struct f2fs_fault_info fault_info; /* For fault injection */ 129 #endif 130 #ifdef CONFIG_QUOTA 131 /* Names of quota files with journalled quota */ 132 char *s_qf_names[MAXQUOTAS]; 133 int s_jquota_fmt; /* Format of quota to use */ 134 #endif 135 /* For which write hints are passed down to block layer */ 136 int whint_mode; 137 int alloc_mode; /* segment allocation policy */ 138 int fsync_mode; /* fsync policy */ 139 int fs_mode; /* fs mode: LFS or ADAPTIVE */ 140 int bggc_mode; /* bggc mode: off, on or sync */ 141 struct fscrypt_dummy_context dummy_enc_ctx; /* test dummy encryption */ 142 block_t unusable_cap; /* Amount of space allowed to be 143 * unusable when disabling checkpoint 144 */ 145 146 /* For compression */ 147 unsigned char compress_algorithm; /* algorithm type */ 148 unsigned compress_log_size; /* cluster log size */ 149 unsigned char compress_ext_cnt; /* extension count */ 150 unsigned char extensions[COMPRESS_EXT_NUM][F2FS_EXTENSION_LEN]; /* extensions */ 151 }; 152 153 #define F2FS_FEATURE_ENCRYPT 0x0001 154 #define F2FS_FEATURE_BLKZONED 0x0002 155 #define F2FS_FEATURE_ATOMIC_WRITE 0x0004 156 #define F2FS_FEATURE_EXTRA_ATTR 0x0008 157 #define F2FS_FEATURE_PRJQUOTA 0x0010 158 #define F2FS_FEATURE_INODE_CHKSUM 0x0020 159 #define F2FS_FEATURE_FLEXIBLE_INLINE_XATTR 0x0040 160 #define F2FS_FEATURE_QUOTA_INO 0x0080 161 #define F2FS_FEATURE_INODE_CRTIME 0x0100 162 #define F2FS_FEATURE_LOST_FOUND 0x0200 163 #define F2FS_FEATURE_VERITY 0x0400 164 #define F2FS_FEATURE_SB_CHKSUM 0x0800 165 #define F2FS_FEATURE_CASEFOLD 0x1000 166 #define F2FS_FEATURE_COMPRESSION 0x2000 167 168 #define __F2FS_HAS_FEATURE(raw_super, mask) \ 169 ((raw_super->feature & cpu_to_le32(mask)) != 0) 170 #define F2FS_HAS_FEATURE(sbi, mask) __F2FS_HAS_FEATURE(sbi->raw_super, mask) 171 #define F2FS_SET_FEATURE(sbi, mask) \ 172 (sbi->raw_super->feature |= cpu_to_le32(mask)) 173 #define F2FS_CLEAR_FEATURE(sbi, mask) \ 174 (sbi->raw_super->feature &= ~cpu_to_le32(mask)) 175 176 /* 177 * Default values for user and/or group using reserved blocks 178 */ 179 #define F2FS_DEF_RESUID 0 180 #define F2FS_DEF_RESGID 0 181 182 /* 183 * For checkpoint manager 184 */ 185 enum { 186 NAT_BITMAP, 187 SIT_BITMAP 188 }; 189 190 #define CP_UMOUNT 0x00000001 191 #define CP_FASTBOOT 0x00000002 192 #define CP_SYNC 0x00000004 193 #define CP_RECOVERY 0x00000008 194 #define CP_DISCARD 0x00000010 195 #define CP_TRIMMED 0x00000020 196 #define CP_PAUSE 0x00000040 197 198 #define MAX_DISCARD_BLOCKS(sbi) BLKS_PER_SEC(sbi) 199 #define DEF_MAX_DISCARD_REQUEST 8 /* issue 8 discards per round */ 200 #define DEF_MIN_DISCARD_ISSUE_TIME 50 /* 50 ms, if exists */ 201 #define DEF_MID_DISCARD_ISSUE_TIME 500 /* 500 ms, if device busy */ 202 #define DEF_MAX_DISCARD_ISSUE_TIME 60000 /* 60 s, if no candidates */ 203 #define DEF_DISCARD_URGENT_UTIL 80 /* do more discard over 80% */ 204 #define DEF_CP_INTERVAL 60 /* 60 secs */ 205 #define DEF_IDLE_INTERVAL 5 /* 5 secs */ 206 #define DEF_DISABLE_INTERVAL 5 /* 5 secs */ 207 #define DEF_DISABLE_QUICK_INTERVAL 1 /* 1 secs */ 208 #define DEF_UMOUNT_DISCARD_TIMEOUT 5 /* 5 secs */ 209 210 struct cp_control { 211 int reason; 212 __u64 trim_start; 213 __u64 trim_end; 214 __u64 trim_minlen; 215 }; 216 217 /* 218 * indicate meta/data type 219 */ 220 enum { 221 META_CP, 222 META_NAT, 223 META_SIT, 224 META_SSA, 225 META_MAX, 226 META_POR, 227 DATA_GENERIC, /* check range only */ 228 DATA_GENERIC_ENHANCE, /* strong check on range and segment bitmap */ 229 DATA_GENERIC_ENHANCE_READ, /* 230 * strong check on range and segment 231 * bitmap but no warning due to race 232 * condition of read on truncated area 233 * by extent_cache 234 */ 235 META_GENERIC, 236 }; 237 238 /* for the list of ino */ 239 enum { 240 ORPHAN_INO, /* for orphan ino list */ 241 APPEND_INO, /* for append ino list */ 242 UPDATE_INO, /* for update ino list */ 243 TRANS_DIR_INO, /* for trasactions dir ino list */ 244 FLUSH_INO, /* for multiple device flushing */ 245 MAX_INO_ENTRY, /* max. list */ 246 }; 247 248 struct ino_entry { 249 struct list_head list; /* list head */ 250 nid_t ino; /* inode number */ 251 unsigned int dirty_device; /* dirty device bitmap */ 252 }; 253 254 /* for the list of inodes to be GCed */ 255 struct inode_entry { 256 struct list_head list; /* list head */ 257 struct inode *inode; /* vfs inode pointer */ 258 }; 259 260 struct fsync_node_entry { 261 struct list_head list; /* list head */ 262 struct page *page; /* warm node page pointer */ 263 unsigned int seq_id; /* sequence id */ 264 }; 265 266 /* for the bitmap indicate blocks to be discarded */ 267 struct discard_entry { 268 struct list_head list; /* list head */ 269 block_t start_blkaddr; /* start blockaddr of current segment */ 270 unsigned char discard_map[SIT_VBLOCK_MAP_SIZE]; /* segment discard bitmap */ 271 }; 272 273 /* default discard granularity of inner discard thread, unit: block count */ 274 #define DEFAULT_DISCARD_GRANULARITY 16 275 276 /* max discard pend list number */ 277 #define MAX_PLIST_NUM 512 278 #define plist_idx(blk_num) ((blk_num) >= MAX_PLIST_NUM ? \ 279 (MAX_PLIST_NUM - 1) : ((blk_num) - 1)) 280 281 enum { 282 D_PREP, /* initial */ 283 D_PARTIAL, /* partially submitted */ 284 D_SUBMIT, /* all submitted */ 285 D_DONE, /* finished */ 286 }; 287 288 struct discard_info { 289 block_t lstart; /* logical start address */ 290 block_t len; /* length */ 291 block_t start; /* actual start address in dev */ 292 }; 293 294 struct discard_cmd { 295 struct rb_node rb_node; /* rb node located in rb-tree */ 296 union { 297 struct { 298 block_t lstart; /* logical start address */ 299 block_t len; /* length */ 300 block_t start; /* actual start address in dev */ 301 }; 302 struct discard_info di; /* discard info */ 303 304 }; 305 struct list_head list; /* command list */ 306 struct completion wait; /* compleation */ 307 struct block_device *bdev; /* bdev */ 308 unsigned short ref; /* reference count */ 309 unsigned char state; /* state */ 310 unsigned char queued; /* queued discard */ 311 int error; /* bio error */ 312 spinlock_t lock; /* for state/bio_ref updating */ 313 unsigned short bio_ref; /* bio reference count */ 314 }; 315 316 enum { 317 DPOLICY_BG, 318 DPOLICY_FORCE, 319 DPOLICY_FSTRIM, 320 DPOLICY_UMOUNT, 321 MAX_DPOLICY, 322 }; 323 324 struct discard_policy { 325 int type; /* type of discard */ 326 unsigned int min_interval; /* used for candidates exist */ 327 unsigned int mid_interval; /* used for device busy */ 328 unsigned int max_interval; /* used for candidates not exist */ 329 unsigned int max_requests; /* # of discards issued per round */ 330 unsigned int io_aware_gran; /* minimum granularity discard not be aware of I/O */ 331 bool io_aware; /* issue discard in idle time */ 332 bool sync; /* submit discard with REQ_SYNC flag */ 333 bool ordered; /* issue discard by lba order */ 334 bool timeout; /* discard timeout for put_super */ 335 unsigned int granularity; /* discard granularity */ 336 }; 337 338 struct discard_cmd_control { 339 struct task_struct *f2fs_issue_discard; /* discard thread */ 340 struct list_head entry_list; /* 4KB discard entry list */ 341 struct list_head pend_list[MAX_PLIST_NUM];/* store pending entries */ 342 struct list_head wait_list; /* store on-flushing entries */ 343 struct list_head fstrim_list; /* in-flight discard from fstrim */ 344 wait_queue_head_t discard_wait_queue; /* waiting queue for wake-up */ 345 unsigned int discard_wake; /* to wake up discard thread */ 346 struct mutex cmd_lock; 347 unsigned int nr_discards; /* # of discards in the list */ 348 unsigned int max_discards; /* max. discards to be issued */ 349 unsigned int discard_granularity; /* discard granularity */ 350 unsigned int undiscard_blks; /* # of undiscard blocks */ 351 unsigned int next_pos; /* next discard position */ 352 atomic_t issued_discard; /* # of issued discard */ 353 atomic_t queued_discard; /* # of queued discard */ 354 atomic_t discard_cmd_cnt; /* # of cached cmd count */ 355 struct rb_root_cached root; /* root of discard rb-tree */ 356 bool rbtree_check; /* config for consistence check */ 357 }; 358 359 /* for the list of fsync inodes, used only during recovery */ 360 struct fsync_inode_entry { 361 struct list_head list; /* list head */ 362 struct inode *inode; /* vfs inode pointer */ 363 block_t blkaddr; /* block address locating the last fsync */ 364 block_t last_dentry; /* block address locating the last dentry */ 365 }; 366 367 #define nats_in_cursum(jnl) (le16_to_cpu((jnl)->n_nats)) 368 #define sits_in_cursum(jnl) (le16_to_cpu((jnl)->n_sits)) 369 370 #define nat_in_journal(jnl, i) ((jnl)->nat_j.entries[i].ne) 371 #define nid_in_journal(jnl, i) ((jnl)->nat_j.entries[i].nid) 372 #define sit_in_journal(jnl, i) ((jnl)->sit_j.entries[i].se) 373 #define segno_in_journal(jnl, i) ((jnl)->sit_j.entries[i].segno) 374 375 #define MAX_NAT_JENTRIES(jnl) (NAT_JOURNAL_ENTRIES - nats_in_cursum(jnl)) 376 #define MAX_SIT_JENTRIES(jnl) (SIT_JOURNAL_ENTRIES - sits_in_cursum(jnl)) 377 378 static inline int update_nats_in_cursum(struct f2fs_journal *journal, int i) 379 { 380 int before = nats_in_cursum(journal); 381 382 journal->n_nats = cpu_to_le16(before + i); 383 return before; 384 } 385 386 static inline int update_sits_in_cursum(struct f2fs_journal *journal, int i) 387 { 388 int before = sits_in_cursum(journal); 389 390 journal->n_sits = cpu_to_le16(before + i); 391 return before; 392 } 393 394 static inline bool __has_cursum_space(struct f2fs_journal *journal, 395 int size, int type) 396 { 397 if (type == NAT_JOURNAL) 398 return size <= MAX_NAT_JENTRIES(journal); 399 return size <= MAX_SIT_JENTRIES(journal); 400 } 401 402 /* 403 * ioctl commands 404 */ 405 #define F2FS_IOC_GETFLAGS FS_IOC_GETFLAGS 406 #define F2FS_IOC_SETFLAGS FS_IOC_SETFLAGS 407 #define F2FS_IOC_GETVERSION FS_IOC_GETVERSION 408 409 #define F2FS_IOCTL_MAGIC 0xf5 410 #define F2FS_IOC_START_ATOMIC_WRITE _IO(F2FS_IOCTL_MAGIC, 1) 411 #define F2FS_IOC_COMMIT_ATOMIC_WRITE _IO(F2FS_IOCTL_MAGIC, 2) 412 #define F2FS_IOC_START_VOLATILE_WRITE _IO(F2FS_IOCTL_MAGIC, 3) 413 #define F2FS_IOC_RELEASE_VOLATILE_WRITE _IO(F2FS_IOCTL_MAGIC, 4) 414 #define F2FS_IOC_ABORT_VOLATILE_WRITE _IO(F2FS_IOCTL_MAGIC, 5) 415 #define F2FS_IOC_GARBAGE_COLLECT _IOW(F2FS_IOCTL_MAGIC, 6, __u32) 416 #define F2FS_IOC_WRITE_CHECKPOINT _IO(F2FS_IOCTL_MAGIC, 7) 417 #define F2FS_IOC_DEFRAGMENT _IOWR(F2FS_IOCTL_MAGIC, 8, \ 418 struct f2fs_defragment) 419 #define F2FS_IOC_MOVE_RANGE _IOWR(F2FS_IOCTL_MAGIC, 9, \ 420 struct f2fs_move_range) 421 #define F2FS_IOC_FLUSH_DEVICE _IOW(F2FS_IOCTL_MAGIC, 10, \ 422 struct f2fs_flush_device) 423 #define F2FS_IOC_GARBAGE_COLLECT_RANGE _IOW(F2FS_IOCTL_MAGIC, 11, \ 424 struct f2fs_gc_range) 425 #define F2FS_IOC_GET_FEATURES _IOR(F2FS_IOCTL_MAGIC, 12, __u32) 426 #define F2FS_IOC_SET_PIN_FILE _IOW(F2FS_IOCTL_MAGIC, 13, __u32) 427 #define F2FS_IOC_GET_PIN_FILE _IOR(F2FS_IOCTL_MAGIC, 14, __u32) 428 #define F2FS_IOC_PRECACHE_EXTENTS _IO(F2FS_IOCTL_MAGIC, 15) 429 #define F2FS_IOC_RESIZE_FS _IOW(F2FS_IOCTL_MAGIC, 16, __u64) 430 #define F2FS_IOC_GET_COMPRESS_BLOCKS _IOR(F2FS_IOCTL_MAGIC, 17, __u64) 431 432 #define F2FS_IOC_GET_VOLUME_NAME FS_IOC_GETFSLABEL 433 #define F2FS_IOC_SET_VOLUME_NAME FS_IOC_SETFSLABEL 434 435 #define F2FS_IOC_SET_ENCRYPTION_POLICY FS_IOC_SET_ENCRYPTION_POLICY 436 #define F2FS_IOC_GET_ENCRYPTION_POLICY FS_IOC_GET_ENCRYPTION_POLICY 437 #define F2FS_IOC_GET_ENCRYPTION_PWSALT FS_IOC_GET_ENCRYPTION_PWSALT 438 439 /* 440 * should be same as XFS_IOC_GOINGDOWN. 441 * Flags for going down operation used by FS_IOC_GOINGDOWN 442 */ 443 #define F2FS_IOC_SHUTDOWN _IOR('X', 125, __u32) /* Shutdown */ 444 #define F2FS_GOING_DOWN_FULLSYNC 0x0 /* going down with full sync */ 445 #define F2FS_GOING_DOWN_METASYNC 0x1 /* going down with metadata */ 446 #define F2FS_GOING_DOWN_NOSYNC 0x2 /* going down */ 447 #define F2FS_GOING_DOWN_METAFLUSH 0x3 /* going down with meta flush */ 448 #define F2FS_GOING_DOWN_NEED_FSCK 0x4 /* going down to trigger fsck */ 449 450 #if defined(__KERNEL__) && defined(CONFIG_COMPAT) 451 /* 452 * ioctl commands in 32 bit emulation 453 */ 454 #define F2FS_IOC32_GETFLAGS FS_IOC32_GETFLAGS 455 #define F2FS_IOC32_SETFLAGS FS_IOC32_SETFLAGS 456 #define F2FS_IOC32_GETVERSION FS_IOC32_GETVERSION 457 #endif 458 459 #define F2FS_IOC_FSGETXATTR FS_IOC_FSGETXATTR 460 #define F2FS_IOC_FSSETXATTR FS_IOC_FSSETXATTR 461 462 struct f2fs_gc_range { 463 u32 sync; 464 u64 start; 465 u64 len; 466 }; 467 468 struct f2fs_defragment { 469 u64 start; 470 u64 len; 471 }; 472 473 struct f2fs_move_range { 474 u32 dst_fd; /* destination fd */ 475 u64 pos_in; /* start position in src_fd */ 476 u64 pos_out; /* start position in dst_fd */ 477 u64 len; /* size to move */ 478 }; 479 480 struct f2fs_flush_device { 481 u32 dev_num; /* device number to flush */ 482 u32 segments; /* # of segments to flush */ 483 }; 484 485 /* for inline stuff */ 486 #define DEF_INLINE_RESERVED_SIZE 1 487 static inline int get_extra_isize(struct inode *inode); 488 static inline int get_inline_xattr_addrs(struct inode *inode); 489 #define MAX_INLINE_DATA(inode) (sizeof(__le32) * \ 490 (CUR_ADDRS_PER_INODE(inode) - \ 491 get_inline_xattr_addrs(inode) - \ 492 DEF_INLINE_RESERVED_SIZE)) 493 494 /* for inline dir */ 495 #define NR_INLINE_DENTRY(inode) (MAX_INLINE_DATA(inode) * BITS_PER_BYTE / \ 496 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \ 497 BITS_PER_BYTE + 1)) 498 #define INLINE_DENTRY_BITMAP_SIZE(inode) \ 499 DIV_ROUND_UP(NR_INLINE_DENTRY(inode), BITS_PER_BYTE) 500 #define INLINE_RESERVED_SIZE(inode) (MAX_INLINE_DATA(inode) - \ 501 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \ 502 NR_INLINE_DENTRY(inode) + \ 503 INLINE_DENTRY_BITMAP_SIZE(inode))) 504 505 /* 506 * For INODE and NODE manager 507 */ 508 /* for directory operations */ 509 struct f2fs_dentry_ptr { 510 struct inode *inode; 511 void *bitmap; 512 struct f2fs_dir_entry *dentry; 513 __u8 (*filename)[F2FS_SLOT_LEN]; 514 int max; 515 int nr_bitmap; 516 }; 517 518 static inline void make_dentry_ptr_block(struct inode *inode, 519 struct f2fs_dentry_ptr *d, struct f2fs_dentry_block *t) 520 { 521 d->inode = inode; 522 d->max = NR_DENTRY_IN_BLOCK; 523 d->nr_bitmap = SIZE_OF_DENTRY_BITMAP; 524 d->bitmap = t->dentry_bitmap; 525 d->dentry = t->dentry; 526 d->filename = t->filename; 527 } 528 529 static inline void make_dentry_ptr_inline(struct inode *inode, 530 struct f2fs_dentry_ptr *d, void *t) 531 { 532 int entry_cnt = NR_INLINE_DENTRY(inode); 533 int bitmap_size = INLINE_DENTRY_BITMAP_SIZE(inode); 534 int reserved_size = INLINE_RESERVED_SIZE(inode); 535 536 d->inode = inode; 537 d->max = entry_cnt; 538 d->nr_bitmap = bitmap_size; 539 d->bitmap = t; 540 d->dentry = t + bitmap_size + reserved_size; 541 d->filename = t + bitmap_size + reserved_size + 542 SIZE_OF_DIR_ENTRY * entry_cnt; 543 } 544 545 /* 546 * XATTR_NODE_OFFSET stores xattrs to one node block per file keeping -1 547 * as its node offset to distinguish from index node blocks. 548 * But some bits are used to mark the node block. 549 */ 550 #define XATTR_NODE_OFFSET ((((unsigned int)-1) << OFFSET_BIT_SHIFT) \ 551 >> OFFSET_BIT_SHIFT) 552 enum { 553 ALLOC_NODE, /* allocate a new node page if needed */ 554 LOOKUP_NODE, /* look up a node without readahead */ 555 LOOKUP_NODE_RA, /* 556 * look up a node with readahead called 557 * by get_data_block. 558 */ 559 }; 560 561 #define DEFAULT_RETRY_IO_COUNT 8 /* maximum retry read IO count */ 562 563 /* congestion wait timeout value, default: 20ms */ 564 #define DEFAULT_IO_TIMEOUT (msecs_to_jiffies(20)) 565 566 /* maximum retry quota flush count */ 567 #define DEFAULT_RETRY_QUOTA_FLUSH_COUNT 8 568 569 #define F2FS_LINK_MAX 0xffffffff /* maximum link count per file */ 570 571 #define MAX_DIR_RA_PAGES 4 /* maximum ra pages of dir */ 572 573 /* for in-memory extent cache entry */ 574 #define F2FS_MIN_EXTENT_LEN 64 /* minimum extent length */ 575 576 /* number of extent info in extent cache we try to shrink */ 577 #define EXTENT_CACHE_SHRINK_NUMBER 128 578 579 struct rb_entry { 580 struct rb_node rb_node; /* rb node located in rb-tree */ 581 unsigned int ofs; /* start offset of the entry */ 582 unsigned int len; /* length of the entry */ 583 }; 584 585 struct extent_info { 586 unsigned int fofs; /* start offset in a file */ 587 unsigned int len; /* length of the extent */ 588 u32 blk; /* start block address of the extent */ 589 }; 590 591 struct extent_node { 592 struct rb_node rb_node; /* rb node located in rb-tree */ 593 struct extent_info ei; /* extent info */ 594 struct list_head list; /* node in global extent list of sbi */ 595 struct extent_tree *et; /* extent tree pointer */ 596 }; 597 598 struct extent_tree { 599 nid_t ino; /* inode number */ 600 struct rb_root_cached root; /* root of extent info rb-tree */ 601 struct extent_node *cached_en; /* recently accessed extent node */ 602 struct extent_info largest; /* largested extent info */ 603 struct list_head list; /* to be used by sbi->zombie_list */ 604 rwlock_t lock; /* protect extent info rb-tree */ 605 atomic_t node_cnt; /* # of extent node in rb-tree*/ 606 bool largest_updated; /* largest extent updated */ 607 }; 608 609 /* 610 * This structure is taken from ext4_map_blocks. 611 * 612 * Note that, however, f2fs uses NEW and MAPPED flags for f2fs_map_blocks(). 613 */ 614 #define F2FS_MAP_NEW (1 << BH_New) 615 #define F2FS_MAP_MAPPED (1 << BH_Mapped) 616 #define F2FS_MAP_UNWRITTEN (1 << BH_Unwritten) 617 #define F2FS_MAP_FLAGS (F2FS_MAP_NEW | F2FS_MAP_MAPPED |\ 618 F2FS_MAP_UNWRITTEN) 619 620 struct f2fs_map_blocks { 621 block_t m_pblk; 622 block_t m_lblk; 623 unsigned int m_len; 624 unsigned int m_flags; 625 pgoff_t *m_next_pgofs; /* point next possible non-hole pgofs */ 626 pgoff_t *m_next_extent; /* point to next possible extent */ 627 int m_seg_type; 628 bool m_may_create; /* indicate it is from write path */ 629 }; 630 631 /* for flag in get_data_block */ 632 enum { 633 F2FS_GET_BLOCK_DEFAULT, 634 F2FS_GET_BLOCK_FIEMAP, 635 F2FS_GET_BLOCK_BMAP, 636 F2FS_GET_BLOCK_DIO, 637 F2FS_GET_BLOCK_PRE_DIO, 638 F2FS_GET_BLOCK_PRE_AIO, 639 F2FS_GET_BLOCK_PRECACHE, 640 }; 641 642 /* 643 * i_advise uses FADVISE_XXX_BIT. We can add additional hints later. 644 */ 645 #define FADVISE_COLD_BIT 0x01 646 #define FADVISE_LOST_PINO_BIT 0x02 647 #define FADVISE_ENCRYPT_BIT 0x04 648 #define FADVISE_ENC_NAME_BIT 0x08 649 #define FADVISE_KEEP_SIZE_BIT 0x10 650 #define FADVISE_HOT_BIT 0x20 651 #define FADVISE_VERITY_BIT 0x40 652 653 #define FADVISE_MODIFIABLE_BITS (FADVISE_COLD_BIT | FADVISE_HOT_BIT) 654 655 #define file_is_cold(inode) is_file(inode, FADVISE_COLD_BIT) 656 #define file_wrong_pino(inode) is_file(inode, FADVISE_LOST_PINO_BIT) 657 #define file_set_cold(inode) set_file(inode, FADVISE_COLD_BIT) 658 #define file_lost_pino(inode) set_file(inode, FADVISE_LOST_PINO_BIT) 659 #define file_clear_cold(inode) clear_file(inode, FADVISE_COLD_BIT) 660 #define file_got_pino(inode) clear_file(inode, FADVISE_LOST_PINO_BIT) 661 #define file_is_encrypt(inode) is_file(inode, FADVISE_ENCRYPT_BIT) 662 #define file_set_encrypt(inode) set_file(inode, FADVISE_ENCRYPT_BIT) 663 #define file_clear_encrypt(inode) clear_file(inode, FADVISE_ENCRYPT_BIT) 664 #define file_enc_name(inode) is_file(inode, FADVISE_ENC_NAME_BIT) 665 #define file_set_enc_name(inode) set_file(inode, FADVISE_ENC_NAME_BIT) 666 #define file_keep_isize(inode) is_file(inode, FADVISE_KEEP_SIZE_BIT) 667 #define file_set_keep_isize(inode) set_file(inode, FADVISE_KEEP_SIZE_BIT) 668 #define file_is_hot(inode) is_file(inode, FADVISE_HOT_BIT) 669 #define file_set_hot(inode) set_file(inode, FADVISE_HOT_BIT) 670 #define file_clear_hot(inode) clear_file(inode, FADVISE_HOT_BIT) 671 #define file_is_verity(inode) is_file(inode, FADVISE_VERITY_BIT) 672 #define file_set_verity(inode) set_file(inode, FADVISE_VERITY_BIT) 673 674 #define DEF_DIR_LEVEL 0 675 676 enum { 677 GC_FAILURE_PIN, 678 GC_FAILURE_ATOMIC, 679 MAX_GC_FAILURE 680 }; 681 682 /* used for f2fs_inode_info->flags */ 683 enum { 684 FI_NEW_INODE, /* indicate newly allocated inode */ 685 FI_DIRTY_INODE, /* indicate inode is dirty or not */ 686 FI_AUTO_RECOVER, /* indicate inode is recoverable */ 687 FI_DIRTY_DIR, /* indicate directory has dirty pages */ 688 FI_INC_LINK, /* need to increment i_nlink */ 689 FI_ACL_MODE, /* indicate acl mode */ 690 FI_NO_ALLOC, /* should not allocate any blocks */ 691 FI_FREE_NID, /* free allocated nide */ 692 FI_NO_EXTENT, /* not to use the extent cache */ 693 FI_INLINE_XATTR, /* used for inline xattr */ 694 FI_INLINE_DATA, /* used for inline data*/ 695 FI_INLINE_DENTRY, /* used for inline dentry */ 696 FI_APPEND_WRITE, /* inode has appended data */ 697 FI_UPDATE_WRITE, /* inode has in-place-update data */ 698 FI_NEED_IPU, /* used for ipu per file */ 699 FI_ATOMIC_FILE, /* indicate atomic file */ 700 FI_ATOMIC_COMMIT, /* indicate the state of atomical committing */ 701 FI_VOLATILE_FILE, /* indicate volatile file */ 702 FI_FIRST_BLOCK_WRITTEN, /* indicate #0 data block was written */ 703 FI_DROP_CACHE, /* drop dirty page cache */ 704 FI_DATA_EXIST, /* indicate data exists */ 705 FI_INLINE_DOTS, /* indicate inline dot dentries */ 706 FI_DO_DEFRAG, /* indicate defragment is running */ 707 FI_DIRTY_FILE, /* indicate regular/symlink has dirty pages */ 708 FI_NO_PREALLOC, /* indicate skipped preallocated blocks */ 709 FI_HOT_DATA, /* indicate file is hot */ 710 FI_EXTRA_ATTR, /* indicate file has extra attribute */ 711 FI_PROJ_INHERIT, /* indicate file inherits projectid */ 712 FI_PIN_FILE, /* indicate file should not be gced */ 713 FI_ATOMIC_REVOKE_REQUEST, /* request to drop atomic data */ 714 FI_VERITY_IN_PROGRESS, /* building fs-verity Merkle tree */ 715 FI_COMPRESSED_FILE, /* indicate file's data can be compressed */ 716 FI_MMAP_FILE, /* indicate file was mmapped */ 717 FI_MAX, /* max flag, never be used */ 718 }; 719 720 struct f2fs_inode_info { 721 struct inode vfs_inode; /* serve a vfs inode */ 722 unsigned long i_flags; /* keep an inode flags for ioctl */ 723 unsigned char i_advise; /* use to give file attribute hints */ 724 unsigned char i_dir_level; /* use for dentry level for large dir */ 725 unsigned int i_current_depth; /* only for directory depth */ 726 /* for gc failure statistic */ 727 unsigned int i_gc_failures[MAX_GC_FAILURE]; 728 unsigned int i_pino; /* parent inode number */ 729 umode_t i_acl_mode; /* keep file acl mode temporarily */ 730 731 /* Use below internally in f2fs*/ 732 unsigned long flags[BITS_TO_LONGS(FI_MAX)]; /* use to pass per-file flags */ 733 struct rw_semaphore i_sem; /* protect fi info */ 734 atomic_t dirty_pages; /* # of dirty pages */ 735 f2fs_hash_t chash; /* hash value of given file name */ 736 unsigned int clevel; /* maximum level of given file name */ 737 struct task_struct *task; /* lookup and create consistency */ 738 struct task_struct *cp_task; /* separate cp/wb IO stats*/ 739 nid_t i_xattr_nid; /* node id that contains xattrs */ 740 loff_t last_disk_size; /* lastly written file size */ 741 spinlock_t i_size_lock; /* protect last_disk_size */ 742 743 #ifdef CONFIG_QUOTA 744 struct dquot *i_dquot[MAXQUOTAS]; 745 746 /* quota space reservation, managed internally by quota code */ 747 qsize_t i_reserved_quota; 748 #endif 749 struct list_head dirty_list; /* dirty list for dirs and files */ 750 struct list_head gdirty_list; /* linked in global dirty list */ 751 struct list_head inmem_ilist; /* list for inmem inodes */ 752 struct list_head inmem_pages; /* inmemory pages managed by f2fs */ 753 struct task_struct *inmem_task; /* store inmemory task */ 754 struct mutex inmem_lock; /* lock for inmemory pages */ 755 struct extent_tree *extent_tree; /* cached extent_tree entry */ 756 757 /* avoid racing between foreground op and gc */ 758 struct rw_semaphore i_gc_rwsem[2]; 759 struct rw_semaphore i_mmap_sem; 760 struct rw_semaphore i_xattr_sem; /* avoid racing between reading and changing EAs */ 761 762 int i_extra_isize; /* size of extra space located in i_addr */ 763 kprojid_t i_projid; /* id for project quota */ 764 int i_inline_xattr_size; /* inline xattr size */ 765 struct timespec64 i_crtime; /* inode creation time */ 766 struct timespec64 i_disk_time[4];/* inode disk times */ 767 768 /* for file compress */ 769 u64 i_compr_blocks; /* # of compressed blocks */ 770 unsigned char i_compress_algorithm; /* algorithm type */ 771 unsigned char i_log_cluster_size; /* log of cluster size */ 772 unsigned int i_cluster_size; /* cluster size */ 773 }; 774 775 static inline void get_extent_info(struct extent_info *ext, 776 struct f2fs_extent *i_ext) 777 { 778 ext->fofs = le32_to_cpu(i_ext->fofs); 779 ext->blk = le32_to_cpu(i_ext->blk); 780 ext->len = le32_to_cpu(i_ext->len); 781 } 782 783 static inline void set_raw_extent(struct extent_info *ext, 784 struct f2fs_extent *i_ext) 785 { 786 i_ext->fofs = cpu_to_le32(ext->fofs); 787 i_ext->blk = cpu_to_le32(ext->blk); 788 i_ext->len = cpu_to_le32(ext->len); 789 } 790 791 static inline void set_extent_info(struct extent_info *ei, unsigned int fofs, 792 u32 blk, unsigned int len) 793 { 794 ei->fofs = fofs; 795 ei->blk = blk; 796 ei->len = len; 797 } 798 799 static inline bool __is_discard_mergeable(struct discard_info *back, 800 struct discard_info *front, unsigned int max_len) 801 { 802 return (back->lstart + back->len == front->lstart) && 803 (back->len + front->len <= max_len); 804 } 805 806 static inline bool __is_discard_back_mergeable(struct discard_info *cur, 807 struct discard_info *back, unsigned int max_len) 808 { 809 return __is_discard_mergeable(back, cur, max_len); 810 } 811 812 static inline bool __is_discard_front_mergeable(struct discard_info *cur, 813 struct discard_info *front, unsigned int max_len) 814 { 815 return __is_discard_mergeable(cur, front, max_len); 816 } 817 818 static inline bool __is_extent_mergeable(struct extent_info *back, 819 struct extent_info *front) 820 { 821 return (back->fofs + back->len == front->fofs && 822 back->blk + back->len == front->blk); 823 } 824 825 static inline bool __is_back_mergeable(struct extent_info *cur, 826 struct extent_info *back) 827 { 828 return __is_extent_mergeable(back, cur); 829 } 830 831 static inline bool __is_front_mergeable(struct extent_info *cur, 832 struct extent_info *front) 833 { 834 return __is_extent_mergeable(cur, front); 835 } 836 837 extern void f2fs_mark_inode_dirty_sync(struct inode *inode, bool sync); 838 static inline void __try_update_largest_extent(struct extent_tree *et, 839 struct extent_node *en) 840 { 841 if (en->ei.len > et->largest.len) { 842 et->largest = en->ei; 843 et->largest_updated = true; 844 } 845 } 846 847 /* 848 * For free nid management 849 */ 850 enum nid_state { 851 FREE_NID, /* newly added to free nid list */ 852 PREALLOC_NID, /* it is preallocated */ 853 MAX_NID_STATE, 854 }; 855 856 struct f2fs_nm_info { 857 block_t nat_blkaddr; /* base disk address of NAT */ 858 nid_t max_nid; /* maximum possible node ids */ 859 nid_t available_nids; /* # of available node ids */ 860 nid_t next_scan_nid; /* the next nid to be scanned */ 861 unsigned int ram_thresh; /* control the memory footprint */ 862 unsigned int ra_nid_pages; /* # of nid pages to be readaheaded */ 863 unsigned int dirty_nats_ratio; /* control dirty nats ratio threshold */ 864 865 /* NAT cache management */ 866 struct radix_tree_root nat_root;/* root of the nat entry cache */ 867 struct radix_tree_root nat_set_root;/* root of the nat set cache */ 868 struct rw_semaphore nat_tree_lock; /* protect nat_tree_lock */ 869 struct list_head nat_entries; /* cached nat entry list (clean) */ 870 spinlock_t nat_list_lock; /* protect clean nat entry list */ 871 unsigned int nat_cnt; /* the # of cached nat entries */ 872 unsigned int dirty_nat_cnt; /* total num of nat entries in set */ 873 unsigned int nat_blocks; /* # of nat blocks */ 874 875 /* free node ids management */ 876 struct radix_tree_root free_nid_root;/* root of the free_nid cache */ 877 struct list_head free_nid_list; /* list for free nids excluding preallocated nids */ 878 unsigned int nid_cnt[MAX_NID_STATE]; /* the number of free node id */ 879 spinlock_t nid_list_lock; /* protect nid lists ops */ 880 struct mutex build_lock; /* lock for build free nids */ 881 unsigned char **free_nid_bitmap; 882 unsigned char *nat_block_bitmap; 883 unsigned short *free_nid_count; /* free nid count of NAT block */ 884 885 /* for checkpoint */ 886 char *nat_bitmap; /* NAT bitmap pointer */ 887 888 unsigned int nat_bits_blocks; /* # of nat bits blocks */ 889 unsigned char *nat_bits; /* NAT bits blocks */ 890 unsigned char *full_nat_bits; /* full NAT pages */ 891 unsigned char *empty_nat_bits; /* empty NAT pages */ 892 #ifdef CONFIG_F2FS_CHECK_FS 893 char *nat_bitmap_mir; /* NAT bitmap mirror */ 894 #endif 895 int bitmap_size; /* bitmap size */ 896 }; 897 898 /* 899 * this structure is used as one of function parameters. 900 * all the information are dedicated to a given direct node block determined 901 * by the data offset in a file. 902 */ 903 struct dnode_of_data { 904 struct inode *inode; /* vfs inode pointer */ 905 struct page *inode_page; /* its inode page, NULL is possible */ 906 struct page *node_page; /* cached direct node page */ 907 nid_t nid; /* node id of the direct node block */ 908 unsigned int ofs_in_node; /* data offset in the node page */ 909 bool inode_page_locked; /* inode page is locked or not */ 910 bool node_changed; /* is node block changed */ 911 char cur_level; /* level of hole node page */ 912 char max_level; /* level of current page located */ 913 block_t data_blkaddr; /* block address of the node block */ 914 }; 915 916 static inline void set_new_dnode(struct dnode_of_data *dn, struct inode *inode, 917 struct page *ipage, struct page *npage, nid_t nid) 918 { 919 memset(dn, 0, sizeof(*dn)); 920 dn->inode = inode; 921 dn->inode_page = ipage; 922 dn->node_page = npage; 923 dn->nid = nid; 924 } 925 926 /* 927 * For SIT manager 928 * 929 * By default, there are 6 active log areas across the whole main area. 930 * When considering hot and cold data separation to reduce cleaning overhead, 931 * we split 3 for data logs and 3 for node logs as hot, warm, and cold types, 932 * respectively. 933 * In the current design, you should not change the numbers intentionally. 934 * Instead, as a mount option such as active_logs=x, you can use 2, 4, and 6 935 * logs individually according to the underlying devices. (default: 6) 936 * Just in case, on-disk layout covers maximum 16 logs that consist of 8 for 937 * data and 8 for node logs. 938 */ 939 #define NR_CURSEG_DATA_TYPE (3) 940 #define NR_CURSEG_NODE_TYPE (3) 941 #define NR_CURSEG_TYPE (NR_CURSEG_DATA_TYPE + NR_CURSEG_NODE_TYPE) 942 943 enum { 944 CURSEG_HOT_DATA = 0, /* directory entry blocks */ 945 CURSEG_WARM_DATA, /* data blocks */ 946 CURSEG_COLD_DATA, /* multimedia or GCed data blocks */ 947 CURSEG_HOT_NODE, /* direct node blocks of directory files */ 948 CURSEG_WARM_NODE, /* direct node blocks of normal files */ 949 CURSEG_COLD_NODE, /* indirect node blocks */ 950 NO_CHECK_TYPE, 951 CURSEG_COLD_DATA_PINNED,/* cold data for pinned file */ 952 }; 953 954 struct flush_cmd { 955 struct completion wait; 956 struct llist_node llnode; 957 nid_t ino; 958 int ret; 959 }; 960 961 struct flush_cmd_control { 962 struct task_struct *f2fs_issue_flush; /* flush thread */ 963 wait_queue_head_t flush_wait_queue; /* waiting queue for wake-up */ 964 atomic_t issued_flush; /* # of issued flushes */ 965 atomic_t queued_flush; /* # of queued flushes */ 966 struct llist_head issue_list; /* list for command issue */ 967 struct llist_node *dispatch_list; /* list for command dispatch */ 968 }; 969 970 struct f2fs_sm_info { 971 struct sit_info *sit_info; /* whole segment information */ 972 struct free_segmap_info *free_info; /* free segment information */ 973 struct dirty_seglist_info *dirty_info; /* dirty segment information */ 974 struct curseg_info *curseg_array; /* active segment information */ 975 976 struct rw_semaphore curseg_lock; /* for preventing curseg change */ 977 978 block_t seg0_blkaddr; /* block address of 0'th segment */ 979 block_t main_blkaddr; /* start block address of main area */ 980 block_t ssa_blkaddr; /* start block address of SSA area */ 981 982 unsigned int segment_count; /* total # of segments */ 983 unsigned int main_segments; /* # of segments in main area */ 984 unsigned int reserved_segments; /* # of reserved segments */ 985 unsigned int ovp_segments; /* # of overprovision segments */ 986 987 /* a threshold to reclaim prefree segments */ 988 unsigned int rec_prefree_segments; 989 990 /* for batched trimming */ 991 unsigned int trim_sections; /* # of sections to trim */ 992 993 struct list_head sit_entry_set; /* sit entry set list */ 994 995 unsigned int ipu_policy; /* in-place-update policy */ 996 unsigned int min_ipu_util; /* in-place-update threshold */ 997 unsigned int min_fsync_blocks; /* threshold for fsync */ 998 unsigned int min_seq_blocks; /* threshold for sequential blocks */ 999 unsigned int min_hot_blocks; /* threshold for hot block allocation */ 1000 unsigned int min_ssr_sections; /* threshold to trigger SSR allocation */ 1001 1002 /* for flush command control */ 1003 struct flush_cmd_control *fcc_info; 1004 1005 /* for discard command control */ 1006 struct discard_cmd_control *dcc_info; 1007 }; 1008 1009 /* 1010 * For superblock 1011 */ 1012 /* 1013 * COUNT_TYPE for monitoring 1014 * 1015 * f2fs monitors the number of several block types such as on-writeback, 1016 * dirty dentry blocks, dirty node blocks, and dirty meta blocks. 1017 */ 1018 #define WB_DATA_TYPE(p) (__is_cp_guaranteed(p) ? F2FS_WB_CP_DATA : F2FS_WB_DATA) 1019 enum count_type { 1020 F2FS_DIRTY_DENTS, 1021 F2FS_DIRTY_DATA, 1022 F2FS_DIRTY_QDATA, 1023 F2FS_DIRTY_NODES, 1024 F2FS_DIRTY_META, 1025 F2FS_INMEM_PAGES, 1026 F2FS_DIRTY_IMETA, 1027 F2FS_WB_CP_DATA, 1028 F2FS_WB_DATA, 1029 F2FS_RD_DATA, 1030 F2FS_RD_NODE, 1031 F2FS_RD_META, 1032 F2FS_DIO_WRITE, 1033 F2FS_DIO_READ, 1034 NR_COUNT_TYPE, 1035 }; 1036 1037 /* 1038 * The below are the page types of bios used in submit_bio(). 1039 * The available types are: 1040 * DATA User data pages. It operates as async mode. 1041 * NODE Node pages. It operates as async mode. 1042 * META FS metadata pages such as SIT, NAT, CP. 1043 * NR_PAGE_TYPE The number of page types. 1044 * META_FLUSH Make sure the previous pages are written 1045 * with waiting the bio's completion 1046 * ... Only can be used with META. 1047 */ 1048 #define PAGE_TYPE_OF_BIO(type) ((type) > META ? META : (type)) 1049 enum page_type { 1050 DATA, 1051 NODE, 1052 META, 1053 NR_PAGE_TYPE, 1054 META_FLUSH, 1055 INMEM, /* the below types are used by tracepoints only. */ 1056 INMEM_DROP, 1057 INMEM_INVALIDATE, 1058 INMEM_REVOKE, 1059 IPU, 1060 OPU, 1061 }; 1062 1063 enum temp_type { 1064 HOT = 0, /* must be zero for meta bio */ 1065 WARM, 1066 COLD, 1067 NR_TEMP_TYPE, 1068 }; 1069 1070 enum need_lock_type { 1071 LOCK_REQ = 0, 1072 LOCK_DONE, 1073 LOCK_RETRY, 1074 }; 1075 1076 enum cp_reason_type { 1077 CP_NO_NEEDED, 1078 CP_NON_REGULAR, 1079 CP_COMPRESSED, 1080 CP_HARDLINK, 1081 CP_SB_NEED_CP, 1082 CP_WRONG_PINO, 1083 CP_NO_SPC_ROLL, 1084 CP_NODE_NEED_CP, 1085 CP_FASTBOOT_MODE, 1086 CP_SPEC_LOG_NUM, 1087 CP_RECOVER_DIR, 1088 }; 1089 1090 enum iostat_type { 1091 APP_DIRECT_IO, /* app direct IOs */ 1092 APP_BUFFERED_IO, /* app buffered IOs */ 1093 APP_WRITE_IO, /* app write IOs */ 1094 APP_MAPPED_IO, /* app mapped IOs */ 1095 FS_DATA_IO, /* data IOs from kworker/fsync/reclaimer */ 1096 FS_NODE_IO, /* node IOs from kworker/fsync/reclaimer */ 1097 FS_META_IO, /* meta IOs from kworker/reclaimer */ 1098 FS_GC_DATA_IO, /* data IOs from forground gc */ 1099 FS_GC_NODE_IO, /* node IOs from forground gc */ 1100 FS_CP_DATA_IO, /* data IOs from checkpoint */ 1101 FS_CP_NODE_IO, /* node IOs from checkpoint */ 1102 FS_CP_META_IO, /* meta IOs from checkpoint */ 1103 FS_DISCARD, /* discard */ 1104 NR_IO_TYPE, 1105 }; 1106 1107 struct f2fs_io_info { 1108 struct f2fs_sb_info *sbi; /* f2fs_sb_info pointer */ 1109 nid_t ino; /* inode number */ 1110 enum page_type type; /* contains DATA/NODE/META/META_FLUSH */ 1111 enum temp_type temp; /* contains HOT/WARM/COLD */ 1112 int op; /* contains REQ_OP_ */ 1113 int op_flags; /* req_flag_bits */ 1114 block_t new_blkaddr; /* new block address to be written */ 1115 block_t old_blkaddr; /* old block address before Cow */ 1116 struct page *page; /* page to be written */ 1117 struct page *encrypted_page; /* encrypted page */ 1118 struct page *compressed_page; /* compressed page */ 1119 struct list_head list; /* serialize IOs */ 1120 bool submitted; /* indicate IO submission */ 1121 int need_lock; /* indicate we need to lock cp_rwsem */ 1122 bool in_list; /* indicate fio is in io_list */ 1123 bool is_por; /* indicate IO is from recovery or not */ 1124 bool retry; /* need to reallocate block address */ 1125 int compr_blocks; /* # of compressed block addresses */ 1126 bool encrypted; /* indicate file is encrypted */ 1127 enum iostat_type io_type; /* io type */ 1128 struct writeback_control *io_wbc; /* writeback control */ 1129 struct bio **bio; /* bio for ipu */ 1130 sector_t *last_block; /* last block number in bio */ 1131 unsigned char version; /* version of the node */ 1132 }; 1133 1134 struct bio_entry { 1135 struct bio *bio; 1136 struct list_head list; 1137 }; 1138 1139 #define is_read_io(rw) ((rw) == READ) 1140 struct f2fs_bio_info { 1141 struct f2fs_sb_info *sbi; /* f2fs superblock */ 1142 struct bio *bio; /* bios to merge */ 1143 sector_t last_block_in_bio; /* last block number */ 1144 struct f2fs_io_info fio; /* store buffered io info. */ 1145 struct rw_semaphore io_rwsem; /* blocking op for bio */ 1146 spinlock_t io_lock; /* serialize DATA/NODE IOs */ 1147 struct list_head io_list; /* track fios */ 1148 struct list_head bio_list; /* bio entry list head */ 1149 struct rw_semaphore bio_list_lock; /* lock to protect bio entry list */ 1150 }; 1151 1152 #define FDEV(i) (sbi->devs[i]) 1153 #define RDEV(i) (raw_super->devs[i]) 1154 struct f2fs_dev_info { 1155 struct block_device *bdev; 1156 char path[MAX_PATH_LEN]; 1157 unsigned int total_segments; 1158 block_t start_blk; 1159 block_t end_blk; 1160 #ifdef CONFIG_BLK_DEV_ZONED 1161 unsigned int nr_blkz; /* Total number of zones */ 1162 unsigned long *blkz_seq; /* Bitmap indicating sequential zones */ 1163 #endif 1164 }; 1165 1166 enum inode_type { 1167 DIR_INODE, /* for dirty dir inode */ 1168 FILE_INODE, /* for dirty regular/symlink inode */ 1169 DIRTY_META, /* for all dirtied inode metadata */ 1170 ATOMIC_FILE, /* for all atomic files */ 1171 NR_INODE_TYPE, 1172 }; 1173 1174 /* for inner inode cache management */ 1175 struct inode_management { 1176 struct radix_tree_root ino_root; /* ino entry array */ 1177 spinlock_t ino_lock; /* for ino entry lock */ 1178 struct list_head ino_list; /* inode list head */ 1179 unsigned long ino_num; /* number of entries */ 1180 }; 1181 1182 /* For s_flag in struct f2fs_sb_info */ 1183 enum { 1184 SBI_IS_DIRTY, /* dirty flag for checkpoint */ 1185 SBI_IS_CLOSE, /* specify unmounting */ 1186 SBI_NEED_FSCK, /* need fsck.f2fs to fix */ 1187 SBI_POR_DOING, /* recovery is doing or not */ 1188 SBI_NEED_SB_WRITE, /* need to recover superblock */ 1189 SBI_NEED_CP, /* need to checkpoint */ 1190 SBI_IS_SHUTDOWN, /* shutdown by ioctl */ 1191 SBI_IS_RECOVERED, /* recovered orphan/data */ 1192 SBI_CP_DISABLED, /* CP was disabled last mount */ 1193 SBI_CP_DISABLED_QUICK, /* CP was disabled quickly */ 1194 SBI_QUOTA_NEED_FLUSH, /* need to flush quota info in CP */ 1195 SBI_QUOTA_SKIP_FLUSH, /* skip flushing quota in current CP */ 1196 SBI_QUOTA_NEED_REPAIR, /* quota file may be corrupted */ 1197 SBI_IS_RESIZEFS, /* resizefs is in process */ 1198 }; 1199 1200 enum { 1201 CP_TIME, 1202 REQ_TIME, 1203 DISCARD_TIME, 1204 GC_TIME, 1205 DISABLE_TIME, 1206 UMOUNT_DISCARD_TIMEOUT, 1207 MAX_TIME, 1208 }; 1209 1210 enum { 1211 GC_NORMAL, 1212 GC_IDLE_CB, 1213 GC_IDLE_GREEDY, 1214 GC_URGENT, 1215 }; 1216 1217 enum { 1218 BGGC_MODE_ON, /* background gc is on */ 1219 BGGC_MODE_OFF, /* background gc is off */ 1220 BGGC_MODE_SYNC, /* 1221 * background gc is on, migrating blocks 1222 * like foreground gc 1223 */ 1224 }; 1225 1226 enum { 1227 FS_MODE_ADAPTIVE, /* use both lfs/ssr allocation */ 1228 FS_MODE_LFS, /* use lfs allocation only */ 1229 }; 1230 1231 enum { 1232 WHINT_MODE_OFF, /* not pass down write hints */ 1233 WHINT_MODE_USER, /* try to pass down hints given by users */ 1234 WHINT_MODE_FS, /* pass down hints with F2FS policy */ 1235 }; 1236 1237 enum { 1238 ALLOC_MODE_DEFAULT, /* stay default */ 1239 ALLOC_MODE_REUSE, /* reuse segments as much as possible */ 1240 }; 1241 1242 enum fsync_mode { 1243 FSYNC_MODE_POSIX, /* fsync follows posix semantics */ 1244 FSYNC_MODE_STRICT, /* fsync behaves in line with ext4 */ 1245 FSYNC_MODE_NOBARRIER, /* fsync behaves nobarrier based on posix */ 1246 }; 1247 1248 /* 1249 * this value is set in page as a private data which indicate that 1250 * the page is atomically written, and it is in inmem_pages list. 1251 */ 1252 #define ATOMIC_WRITTEN_PAGE ((unsigned long)-1) 1253 #define DUMMY_WRITTEN_PAGE ((unsigned long)-2) 1254 1255 #define IS_ATOMIC_WRITTEN_PAGE(page) \ 1256 (page_private(page) == (unsigned long)ATOMIC_WRITTEN_PAGE) 1257 #define IS_DUMMY_WRITTEN_PAGE(page) \ 1258 (page_private(page) == (unsigned long)DUMMY_WRITTEN_PAGE) 1259 1260 #ifdef CONFIG_FS_ENCRYPTION 1261 #define DUMMY_ENCRYPTION_ENABLED(sbi) \ 1262 (unlikely(F2FS_OPTION(sbi).dummy_enc_ctx.ctx != NULL)) 1263 #else 1264 #define DUMMY_ENCRYPTION_ENABLED(sbi) (0) 1265 #endif 1266 1267 /* For compression */ 1268 enum compress_algorithm_type { 1269 COMPRESS_LZO, 1270 COMPRESS_LZ4, 1271 COMPRESS_ZSTD, 1272 COMPRESS_MAX, 1273 }; 1274 1275 #define COMPRESS_DATA_RESERVED_SIZE 5 1276 struct compress_data { 1277 __le32 clen; /* compressed data size */ 1278 __le32 reserved[COMPRESS_DATA_RESERVED_SIZE]; /* reserved */ 1279 u8 cdata[]; /* compressed data */ 1280 }; 1281 1282 #define COMPRESS_HEADER_SIZE (sizeof(struct compress_data)) 1283 1284 #define F2FS_COMPRESSED_PAGE_MAGIC 0xF5F2C000 1285 1286 /* compress context */ 1287 struct compress_ctx { 1288 struct inode *inode; /* inode the context belong to */ 1289 pgoff_t cluster_idx; /* cluster index number */ 1290 unsigned int cluster_size; /* page count in cluster */ 1291 unsigned int log_cluster_size; /* log of cluster size */ 1292 struct page **rpages; /* pages store raw data in cluster */ 1293 unsigned int nr_rpages; /* total page number in rpages */ 1294 struct page **cpages; /* pages store compressed data in cluster */ 1295 unsigned int nr_cpages; /* total page number in cpages */ 1296 void *rbuf; /* virtual mapped address on rpages */ 1297 struct compress_data *cbuf; /* virtual mapped address on cpages */ 1298 size_t rlen; /* valid data length in rbuf */ 1299 size_t clen; /* valid data length in cbuf */ 1300 void *private; /* payload buffer for specified compression algorithm */ 1301 void *private2; /* extra payload buffer */ 1302 }; 1303 1304 /* compress context for write IO path */ 1305 struct compress_io_ctx { 1306 u32 magic; /* magic number to indicate page is compressed */ 1307 struct inode *inode; /* inode the context belong to */ 1308 struct page **rpages; /* pages store raw data in cluster */ 1309 unsigned int nr_rpages; /* total page number in rpages */ 1310 refcount_t ref; /* referrence count of raw page */ 1311 }; 1312 1313 /* decompress io context for read IO path */ 1314 struct decompress_io_ctx { 1315 u32 magic; /* magic number to indicate page is compressed */ 1316 struct inode *inode; /* inode the context belong to */ 1317 pgoff_t cluster_idx; /* cluster index number */ 1318 unsigned int cluster_size; /* page count in cluster */ 1319 unsigned int log_cluster_size; /* log of cluster size */ 1320 struct page **rpages; /* pages store raw data in cluster */ 1321 unsigned int nr_rpages; /* total page number in rpages */ 1322 struct page **cpages; /* pages store compressed data in cluster */ 1323 unsigned int nr_cpages; /* total page number in cpages */ 1324 struct page **tpages; /* temp pages to pad holes in cluster */ 1325 void *rbuf; /* virtual mapped address on rpages */ 1326 struct compress_data *cbuf; /* virtual mapped address on cpages */ 1327 size_t rlen; /* valid data length in rbuf */ 1328 size_t clen; /* valid data length in cbuf */ 1329 refcount_t ref; /* referrence count of compressed page */ 1330 bool failed; /* indicate IO error during decompression */ 1331 void *private; /* payload buffer for specified decompression algorithm */ 1332 void *private2; /* extra payload buffer */ 1333 }; 1334 1335 #define NULL_CLUSTER ((unsigned int)(~0)) 1336 #define MIN_COMPRESS_LOG_SIZE 2 1337 #define MAX_COMPRESS_LOG_SIZE 8 1338 #define MAX_COMPRESS_WINDOW_SIZE ((PAGE_SIZE) << MAX_COMPRESS_LOG_SIZE) 1339 1340 struct f2fs_sb_info { 1341 struct super_block *sb; /* pointer to VFS super block */ 1342 struct proc_dir_entry *s_proc; /* proc entry */ 1343 struct f2fs_super_block *raw_super; /* raw super block pointer */ 1344 struct rw_semaphore sb_lock; /* lock for raw super block */ 1345 int valid_super_block; /* valid super block no */ 1346 unsigned long s_flag; /* flags for sbi */ 1347 struct mutex writepages; /* mutex for writepages() */ 1348 #ifdef CONFIG_UNICODE 1349 struct unicode_map *s_encoding; 1350 __u16 s_encoding_flags; 1351 #endif 1352 1353 #ifdef CONFIG_BLK_DEV_ZONED 1354 unsigned int blocks_per_blkz; /* F2FS blocks per zone */ 1355 unsigned int log_blocks_per_blkz; /* log2 F2FS blocks per zone */ 1356 #endif 1357 1358 /* for node-related operations */ 1359 struct f2fs_nm_info *nm_info; /* node manager */ 1360 struct inode *node_inode; /* cache node blocks */ 1361 1362 /* for segment-related operations */ 1363 struct f2fs_sm_info *sm_info; /* segment manager */ 1364 1365 /* for bio operations */ 1366 struct f2fs_bio_info *write_io[NR_PAGE_TYPE]; /* for write bios */ 1367 /* keep migration IO order for LFS mode */ 1368 struct rw_semaphore io_order_lock; 1369 mempool_t *write_io_dummy; /* Dummy pages */ 1370 1371 /* for checkpoint */ 1372 struct f2fs_checkpoint *ckpt; /* raw checkpoint pointer */ 1373 int cur_cp_pack; /* remain current cp pack */ 1374 spinlock_t cp_lock; /* for flag in ckpt */ 1375 struct inode *meta_inode; /* cache meta blocks */ 1376 struct mutex cp_mutex; /* checkpoint procedure lock */ 1377 struct rw_semaphore cp_rwsem; /* blocking FS operations */ 1378 struct rw_semaphore node_write; /* locking node writes */ 1379 struct rw_semaphore node_change; /* locking node change */ 1380 wait_queue_head_t cp_wait; 1381 unsigned long last_time[MAX_TIME]; /* to store time in jiffies */ 1382 long interval_time[MAX_TIME]; /* to store thresholds */ 1383 1384 struct inode_management im[MAX_INO_ENTRY]; /* manage inode cache */ 1385 1386 spinlock_t fsync_node_lock; /* for node entry lock */ 1387 struct list_head fsync_node_list; /* node list head */ 1388 unsigned int fsync_seg_id; /* sequence id */ 1389 unsigned int fsync_node_num; /* number of node entries */ 1390 1391 /* for orphan inode, use 0'th array */ 1392 unsigned int max_orphans; /* max orphan inodes */ 1393 1394 /* for inode management */ 1395 struct list_head inode_list[NR_INODE_TYPE]; /* dirty inode list */ 1396 spinlock_t inode_lock[NR_INODE_TYPE]; /* for dirty inode list lock */ 1397 struct mutex flush_lock; /* for flush exclusion */ 1398 1399 /* for extent tree cache */ 1400 struct radix_tree_root extent_tree_root;/* cache extent cache entries */ 1401 struct mutex extent_tree_lock; /* locking extent radix tree */ 1402 struct list_head extent_list; /* lru list for shrinker */ 1403 spinlock_t extent_lock; /* locking extent lru list */ 1404 atomic_t total_ext_tree; /* extent tree count */ 1405 struct list_head zombie_list; /* extent zombie tree list */ 1406 atomic_t total_zombie_tree; /* extent zombie tree count */ 1407 atomic_t total_ext_node; /* extent info count */ 1408 1409 /* basic filesystem units */ 1410 unsigned int log_sectors_per_block; /* log2 sectors per block */ 1411 unsigned int log_blocksize; /* log2 block size */ 1412 unsigned int blocksize; /* block size */ 1413 unsigned int root_ino_num; /* root inode number*/ 1414 unsigned int node_ino_num; /* node inode number*/ 1415 unsigned int meta_ino_num; /* meta inode number*/ 1416 unsigned int log_blocks_per_seg; /* log2 blocks per segment */ 1417 unsigned int blocks_per_seg; /* blocks per segment */ 1418 unsigned int segs_per_sec; /* segments per section */ 1419 unsigned int secs_per_zone; /* sections per zone */ 1420 unsigned int total_sections; /* total section count */ 1421 struct mutex resize_mutex; /* for resize exclusion */ 1422 unsigned int total_node_count; /* total node block count */ 1423 unsigned int total_valid_node_count; /* valid node block count */ 1424 loff_t max_file_blocks; /* max block index of file */ 1425 int dir_level; /* directory level */ 1426 int readdir_ra; /* readahead inode in readdir */ 1427 1428 block_t user_block_count; /* # of user blocks */ 1429 block_t total_valid_block_count; /* # of valid blocks */ 1430 block_t discard_blks; /* discard command candidats */ 1431 block_t last_valid_block_count; /* for recovery */ 1432 block_t reserved_blocks; /* configurable reserved blocks */ 1433 block_t current_reserved_blocks; /* current reserved blocks */ 1434 1435 /* Additional tracking for no checkpoint mode */ 1436 block_t unusable_block_count; /* # of blocks saved by last cp */ 1437 1438 unsigned int nquota_files; /* # of quota sysfile */ 1439 struct rw_semaphore quota_sem; /* blocking cp for flags */ 1440 1441 /* # of pages, see count_type */ 1442 atomic_t nr_pages[NR_COUNT_TYPE]; 1443 /* # of allocated blocks */ 1444 struct percpu_counter alloc_valid_block_count; 1445 1446 /* writeback control */ 1447 atomic_t wb_sync_req[META]; /* count # of WB_SYNC threads */ 1448 1449 /* valid inode count */ 1450 struct percpu_counter total_valid_inode_count; 1451 1452 struct f2fs_mount_info mount_opt; /* mount options */ 1453 1454 /* for cleaning operations */ 1455 struct rw_semaphore gc_lock; /* 1456 * semaphore for GC, avoid 1457 * race between GC and GC or CP 1458 */ 1459 struct f2fs_gc_kthread *gc_thread; /* GC thread */ 1460 unsigned int cur_victim_sec; /* current victim section num */ 1461 unsigned int gc_mode; /* current GC state */ 1462 unsigned int next_victim_seg[2]; /* next segment in victim section */ 1463 /* for skip statistic */ 1464 unsigned int atomic_files; /* # of opened atomic file */ 1465 unsigned long long skipped_atomic_files[2]; /* FG_GC and BG_GC */ 1466 unsigned long long skipped_gc_rwsem; /* FG_GC only */ 1467 1468 /* threshold for gc trials on pinned files */ 1469 u64 gc_pin_file_threshold; 1470 struct rw_semaphore pin_sem; 1471 1472 /* maximum # of trials to find a victim segment for SSR and GC */ 1473 unsigned int max_victim_search; 1474 /* migration granularity of garbage collection, unit: segment */ 1475 unsigned int migration_granularity; 1476 1477 /* 1478 * for stat information. 1479 * one is for the LFS mode, and the other is for the SSR mode. 1480 */ 1481 #ifdef CONFIG_F2FS_STAT_FS 1482 struct f2fs_stat_info *stat_info; /* FS status information */ 1483 atomic_t meta_count[META_MAX]; /* # of meta blocks */ 1484 unsigned int segment_count[2]; /* # of allocated segments */ 1485 unsigned int block_count[2]; /* # of allocated blocks */ 1486 atomic_t inplace_count; /* # of inplace update */ 1487 atomic64_t total_hit_ext; /* # of lookup extent cache */ 1488 atomic64_t read_hit_rbtree; /* # of hit rbtree extent node */ 1489 atomic64_t read_hit_largest; /* # of hit largest extent node */ 1490 atomic64_t read_hit_cached; /* # of hit cached extent node */ 1491 atomic_t inline_xattr; /* # of inline_xattr inodes */ 1492 atomic_t inline_inode; /* # of inline_data inodes */ 1493 atomic_t inline_dir; /* # of inline_dentry inodes */ 1494 atomic_t compr_inode; /* # of compressed inodes */ 1495 atomic_t compr_blocks; /* # of compressed blocks */ 1496 atomic_t vw_cnt; /* # of volatile writes */ 1497 atomic_t max_aw_cnt; /* max # of atomic writes */ 1498 atomic_t max_vw_cnt; /* max # of volatile writes */ 1499 unsigned int io_skip_bggc; /* skip background gc for in-flight IO */ 1500 unsigned int other_skip_bggc; /* skip background gc for other reasons */ 1501 unsigned int ndirty_inode[NR_INODE_TYPE]; /* # of dirty inodes */ 1502 #endif 1503 spinlock_t stat_lock; /* lock for stat operations */ 1504 1505 /* For app/fs IO statistics */ 1506 spinlock_t iostat_lock; 1507 unsigned long long write_iostat[NR_IO_TYPE]; 1508 bool iostat_enable; 1509 1510 /* For sysfs suppport */ 1511 struct kobject s_kobj; 1512 struct completion s_kobj_unregister; 1513 1514 /* For shrinker support */ 1515 struct list_head s_list; 1516 int s_ndevs; /* number of devices */ 1517 struct f2fs_dev_info *devs; /* for device list */ 1518 unsigned int dirty_device; /* for checkpoint data flush */ 1519 spinlock_t dev_lock; /* protect dirty_device */ 1520 struct mutex umount_mutex; 1521 unsigned int shrinker_run_no; 1522 1523 /* For write statistics */ 1524 u64 sectors_written_start; 1525 u64 kbytes_written; 1526 1527 /* Reference to checksum algorithm driver via cryptoapi */ 1528 struct crypto_shash *s_chksum_driver; 1529 1530 /* Precomputed FS UUID checksum for seeding other checksums */ 1531 __u32 s_chksum_seed; 1532 1533 struct workqueue_struct *post_read_wq; /* post read workqueue */ 1534 1535 struct kmem_cache *inline_xattr_slab; /* inline xattr entry */ 1536 unsigned int inline_xattr_slab_size; /* default inline xattr slab size */ 1537 }; 1538 1539 struct f2fs_private_dio { 1540 struct inode *inode; 1541 void *orig_private; 1542 bio_end_io_t *orig_end_io; 1543 bool write; 1544 }; 1545 1546 #ifdef CONFIG_F2FS_FAULT_INJECTION 1547 #define f2fs_show_injection_info(sbi, type) \ 1548 printk_ratelimited("%sF2FS-fs (%s) : inject %s in %s of %pS\n", \ 1549 KERN_INFO, sbi->sb->s_id, \ 1550 f2fs_fault_name[type], \ 1551 __func__, __builtin_return_address(0)) 1552 static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type) 1553 { 1554 struct f2fs_fault_info *ffi = &F2FS_OPTION(sbi).fault_info; 1555 1556 if (!ffi->inject_rate) 1557 return false; 1558 1559 if (!IS_FAULT_SET(ffi, type)) 1560 return false; 1561 1562 atomic_inc(&ffi->inject_ops); 1563 if (atomic_read(&ffi->inject_ops) >= ffi->inject_rate) { 1564 atomic_set(&ffi->inject_ops, 0); 1565 return true; 1566 } 1567 return false; 1568 } 1569 #else 1570 #define f2fs_show_injection_info(sbi, type) do { } while (0) 1571 static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type) 1572 { 1573 return false; 1574 } 1575 #endif 1576 1577 /* 1578 * Test if the mounted volume is a multi-device volume. 1579 * - For a single regular disk volume, sbi->s_ndevs is 0. 1580 * - For a single zoned disk volume, sbi->s_ndevs is 1. 1581 * - For a multi-device volume, sbi->s_ndevs is always 2 or more. 1582 */ 1583 static inline bool f2fs_is_multi_device(struct f2fs_sb_info *sbi) 1584 { 1585 return sbi->s_ndevs > 1; 1586 } 1587 1588 /* For write statistics. Suppose sector size is 512 bytes, 1589 * and the return value is in kbytes. s is of struct f2fs_sb_info. 1590 */ 1591 #define BD_PART_WRITTEN(s) \ 1592 (((u64)part_stat_read((s)->sb->s_bdev->bd_part, sectors[STAT_WRITE]) - \ 1593 (s)->sectors_written_start) >> 1) 1594 1595 static inline void f2fs_update_time(struct f2fs_sb_info *sbi, int type) 1596 { 1597 unsigned long now = jiffies; 1598 1599 sbi->last_time[type] = now; 1600 1601 /* DISCARD_TIME and GC_TIME are based on REQ_TIME */ 1602 if (type == REQ_TIME) { 1603 sbi->last_time[DISCARD_TIME] = now; 1604 sbi->last_time[GC_TIME] = now; 1605 } 1606 } 1607 1608 static inline bool f2fs_time_over(struct f2fs_sb_info *sbi, int type) 1609 { 1610 unsigned long interval = sbi->interval_time[type] * HZ; 1611 1612 return time_after(jiffies, sbi->last_time[type] + interval); 1613 } 1614 1615 static inline unsigned int f2fs_time_to_wait(struct f2fs_sb_info *sbi, 1616 int type) 1617 { 1618 unsigned long interval = sbi->interval_time[type] * HZ; 1619 unsigned int wait_ms = 0; 1620 long delta; 1621 1622 delta = (sbi->last_time[type] + interval) - jiffies; 1623 if (delta > 0) 1624 wait_ms = jiffies_to_msecs(delta); 1625 1626 return wait_ms; 1627 } 1628 1629 /* 1630 * Inline functions 1631 */ 1632 static inline u32 __f2fs_crc32(struct f2fs_sb_info *sbi, u32 crc, 1633 const void *address, unsigned int length) 1634 { 1635 struct { 1636 struct shash_desc shash; 1637 char ctx[4]; 1638 } desc; 1639 int err; 1640 1641 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver) != sizeof(desc.ctx)); 1642 1643 desc.shash.tfm = sbi->s_chksum_driver; 1644 *(u32 *)desc.ctx = crc; 1645 1646 err = crypto_shash_update(&desc.shash, address, length); 1647 BUG_ON(err); 1648 1649 return *(u32 *)desc.ctx; 1650 } 1651 1652 static inline u32 f2fs_crc32(struct f2fs_sb_info *sbi, const void *address, 1653 unsigned int length) 1654 { 1655 return __f2fs_crc32(sbi, F2FS_SUPER_MAGIC, address, length); 1656 } 1657 1658 static inline bool f2fs_crc_valid(struct f2fs_sb_info *sbi, __u32 blk_crc, 1659 void *buf, size_t buf_size) 1660 { 1661 return f2fs_crc32(sbi, buf, buf_size) == blk_crc; 1662 } 1663 1664 static inline u32 f2fs_chksum(struct f2fs_sb_info *sbi, u32 crc, 1665 const void *address, unsigned int length) 1666 { 1667 return __f2fs_crc32(sbi, crc, address, length); 1668 } 1669 1670 static inline struct f2fs_inode_info *F2FS_I(struct inode *inode) 1671 { 1672 return container_of(inode, struct f2fs_inode_info, vfs_inode); 1673 } 1674 1675 static inline struct f2fs_sb_info *F2FS_SB(struct super_block *sb) 1676 { 1677 return sb->s_fs_info; 1678 } 1679 1680 static inline struct f2fs_sb_info *F2FS_I_SB(struct inode *inode) 1681 { 1682 return F2FS_SB(inode->i_sb); 1683 } 1684 1685 static inline struct f2fs_sb_info *F2FS_M_SB(struct address_space *mapping) 1686 { 1687 return F2FS_I_SB(mapping->host); 1688 } 1689 1690 static inline struct f2fs_sb_info *F2FS_P_SB(struct page *page) 1691 { 1692 return F2FS_M_SB(page_file_mapping(page)); 1693 } 1694 1695 static inline struct f2fs_super_block *F2FS_RAW_SUPER(struct f2fs_sb_info *sbi) 1696 { 1697 return (struct f2fs_super_block *)(sbi->raw_super); 1698 } 1699 1700 static inline struct f2fs_checkpoint *F2FS_CKPT(struct f2fs_sb_info *sbi) 1701 { 1702 return (struct f2fs_checkpoint *)(sbi->ckpt); 1703 } 1704 1705 static inline struct f2fs_node *F2FS_NODE(struct page *page) 1706 { 1707 return (struct f2fs_node *)page_address(page); 1708 } 1709 1710 static inline struct f2fs_inode *F2FS_INODE(struct page *page) 1711 { 1712 return &((struct f2fs_node *)page_address(page))->i; 1713 } 1714 1715 static inline struct f2fs_nm_info *NM_I(struct f2fs_sb_info *sbi) 1716 { 1717 return (struct f2fs_nm_info *)(sbi->nm_info); 1718 } 1719 1720 static inline struct f2fs_sm_info *SM_I(struct f2fs_sb_info *sbi) 1721 { 1722 return (struct f2fs_sm_info *)(sbi->sm_info); 1723 } 1724 1725 static inline struct sit_info *SIT_I(struct f2fs_sb_info *sbi) 1726 { 1727 return (struct sit_info *)(SM_I(sbi)->sit_info); 1728 } 1729 1730 static inline struct free_segmap_info *FREE_I(struct f2fs_sb_info *sbi) 1731 { 1732 return (struct free_segmap_info *)(SM_I(sbi)->free_info); 1733 } 1734 1735 static inline struct dirty_seglist_info *DIRTY_I(struct f2fs_sb_info *sbi) 1736 { 1737 return (struct dirty_seglist_info *)(SM_I(sbi)->dirty_info); 1738 } 1739 1740 static inline struct address_space *META_MAPPING(struct f2fs_sb_info *sbi) 1741 { 1742 return sbi->meta_inode->i_mapping; 1743 } 1744 1745 static inline struct address_space *NODE_MAPPING(struct f2fs_sb_info *sbi) 1746 { 1747 return sbi->node_inode->i_mapping; 1748 } 1749 1750 static inline bool is_sbi_flag_set(struct f2fs_sb_info *sbi, unsigned int type) 1751 { 1752 return test_bit(type, &sbi->s_flag); 1753 } 1754 1755 static inline void set_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type) 1756 { 1757 set_bit(type, &sbi->s_flag); 1758 } 1759 1760 static inline void clear_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type) 1761 { 1762 clear_bit(type, &sbi->s_flag); 1763 } 1764 1765 static inline unsigned long long cur_cp_version(struct f2fs_checkpoint *cp) 1766 { 1767 return le64_to_cpu(cp->checkpoint_ver); 1768 } 1769 1770 static inline unsigned long f2fs_qf_ino(struct super_block *sb, int type) 1771 { 1772 if (type < F2FS_MAX_QUOTAS) 1773 return le32_to_cpu(F2FS_SB(sb)->raw_super->qf_ino[type]); 1774 return 0; 1775 } 1776 1777 static inline __u64 cur_cp_crc(struct f2fs_checkpoint *cp) 1778 { 1779 size_t crc_offset = le32_to_cpu(cp->checksum_offset); 1780 return le32_to_cpu(*((__le32 *)((unsigned char *)cp + crc_offset))); 1781 } 1782 1783 static inline bool __is_set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 1784 { 1785 unsigned int ckpt_flags = le32_to_cpu(cp->ckpt_flags); 1786 1787 return ckpt_flags & f; 1788 } 1789 1790 static inline bool is_set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 1791 { 1792 return __is_set_ckpt_flags(F2FS_CKPT(sbi), f); 1793 } 1794 1795 static inline void __set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 1796 { 1797 unsigned int ckpt_flags; 1798 1799 ckpt_flags = le32_to_cpu(cp->ckpt_flags); 1800 ckpt_flags |= f; 1801 cp->ckpt_flags = cpu_to_le32(ckpt_flags); 1802 } 1803 1804 static inline void set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 1805 { 1806 unsigned long flags; 1807 1808 spin_lock_irqsave(&sbi->cp_lock, flags); 1809 __set_ckpt_flags(F2FS_CKPT(sbi), f); 1810 spin_unlock_irqrestore(&sbi->cp_lock, flags); 1811 } 1812 1813 static inline void __clear_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 1814 { 1815 unsigned int ckpt_flags; 1816 1817 ckpt_flags = le32_to_cpu(cp->ckpt_flags); 1818 ckpt_flags &= (~f); 1819 cp->ckpt_flags = cpu_to_le32(ckpt_flags); 1820 } 1821 1822 static inline void clear_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 1823 { 1824 unsigned long flags; 1825 1826 spin_lock_irqsave(&sbi->cp_lock, flags); 1827 __clear_ckpt_flags(F2FS_CKPT(sbi), f); 1828 spin_unlock_irqrestore(&sbi->cp_lock, flags); 1829 } 1830 1831 static inline void disable_nat_bits(struct f2fs_sb_info *sbi, bool lock) 1832 { 1833 unsigned long flags; 1834 unsigned char *nat_bits; 1835 1836 /* 1837 * In order to re-enable nat_bits we need to call fsck.f2fs by 1838 * set_sbi_flag(sbi, SBI_NEED_FSCK). But it may give huge cost, 1839 * so let's rely on regular fsck or unclean shutdown. 1840 */ 1841 1842 if (lock) 1843 spin_lock_irqsave(&sbi->cp_lock, flags); 1844 __clear_ckpt_flags(F2FS_CKPT(sbi), CP_NAT_BITS_FLAG); 1845 nat_bits = NM_I(sbi)->nat_bits; 1846 NM_I(sbi)->nat_bits = NULL; 1847 if (lock) 1848 spin_unlock_irqrestore(&sbi->cp_lock, flags); 1849 1850 kvfree(nat_bits); 1851 } 1852 1853 static inline bool enabled_nat_bits(struct f2fs_sb_info *sbi, 1854 struct cp_control *cpc) 1855 { 1856 bool set = is_set_ckpt_flags(sbi, CP_NAT_BITS_FLAG); 1857 1858 return (cpc) ? (cpc->reason & CP_UMOUNT) && set : set; 1859 } 1860 1861 static inline void f2fs_lock_op(struct f2fs_sb_info *sbi) 1862 { 1863 down_read(&sbi->cp_rwsem); 1864 } 1865 1866 static inline int f2fs_trylock_op(struct f2fs_sb_info *sbi) 1867 { 1868 return down_read_trylock(&sbi->cp_rwsem); 1869 } 1870 1871 static inline void f2fs_unlock_op(struct f2fs_sb_info *sbi) 1872 { 1873 up_read(&sbi->cp_rwsem); 1874 } 1875 1876 static inline void f2fs_lock_all(struct f2fs_sb_info *sbi) 1877 { 1878 down_write(&sbi->cp_rwsem); 1879 } 1880 1881 static inline void f2fs_unlock_all(struct f2fs_sb_info *sbi) 1882 { 1883 up_write(&sbi->cp_rwsem); 1884 } 1885 1886 static inline int __get_cp_reason(struct f2fs_sb_info *sbi) 1887 { 1888 int reason = CP_SYNC; 1889 1890 if (test_opt(sbi, FASTBOOT)) 1891 reason = CP_FASTBOOT; 1892 if (is_sbi_flag_set(sbi, SBI_IS_CLOSE)) 1893 reason = CP_UMOUNT; 1894 return reason; 1895 } 1896 1897 static inline bool __remain_node_summaries(int reason) 1898 { 1899 return (reason & (CP_UMOUNT | CP_FASTBOOT)); 1900 } 1901 1902 static inline bool __exist_node_summaries(struct f2fs_sb_info *sbi) 1903 { 1904 return (is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG) || 1905 is_set_ckpt_flags(sbi, CP_FASTBOOT_FLAG)); 1906 } 1907 1908 /* 1909 * Check whether the inode has blocks or not 1910 */ 1911 static inline int F2FS_HAS_BLOCKS(struct inode *inode) 1912 { 1913 block_t xattr_block = F2FS_I(inode)->i_xattr_nid ? 1 : 0; 1914 1915 return (inode->i_blocks >> F2FS_LOG_SECTORS_PER_BLOCK) > xattr_block; 1916 } 1917 1918 static inline bool f2fs_has_xattr_block(unsigned int ofs) 1919 { 1920 return ofs == XATTR_NODE_OFFSET; 1921 } 1922 1923 static inline bool __allow_reserved_blocks(struct f2fs_sb_info *sbi, 1924 struct inode *inode, bool cap) 1925 { 1926 if (!inode) 1927 return true; 1928 if (!test_opt(sbi, RESERVE_ROOT)) 1929 return false; 1930 if (IS_NOQUOTA(inode)) 1931 return true; 1932 if (uid_eq(F2FS_OPTION(sbi).s_resuid, current_fsuid())) 1933 return true; 1934 if (!gid_eq(F2FS_OPTION(sbi).s_resgid, GLOBAL_ROOT_GID) && 1935 in_group_p(F2FS_OPTION(sbi).s_resgid)) 1936 return true; 1937 if (cap && capable(CAP_SYS_RESOURCE)) 1938 return true; 1939 return false; 1940 } 1941 1942 static inline void f2fs_i_blocks_write(struct inode *, block_t, bool, bool); 1943 static inline int inc_valid_block_count(struct f2fs_sb_info *sbi, 1944 struct inode *inode, blkcnt_t *count) 1945 { 1946 blkcnt_t diff = 0, release = 0; 1947 block_t avail_user_block_count; 1948 int ret; 1949 1950 ret = dquot_reserve_block(inode, *count); 1951 if (ret) 1952 return ret; 1953 1954 if (time_to_inject(sbi, FAULT_BLOCK)) { 1955 f2fs_show_injection_info(sbi, FAULT_BLOCK); 1956 release = *count; 1957 goto release_quota; 1958 } 1959 1960 /* 1961 * let's increase this in prior to actual block count change in order 1962 * for f2fs_sync_file to avoid data races when deciding checkpoint. 1963 */ 1964 percpu_counter_add(&sbi->alloc_valid_block_count, (*count)); 1965 1966 spin_lock(&sbi->stat_lock); 1967 sbi->total_valid_block_count += (block_t)(*count); 1968 avail_user_block_count = sbi->user_block_count - 1969 sbi->current_reserved_blocks; 1970 1971 if (!__allow_reserved_blocks(sbi, inode, true)) 1972 avail_user_block_count -= F2FS_OPTION(sbi).root_reserved_blocks; 1973 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) { 1974 if (avail_user_block_count > sbi->unusable_block_count) 1975 avail_user_block_count -= sbi->unusable_block_count; 1976 else 1977 avail_user_block_count = 0; 1978 } 1979 if (unlikely(sbi->total_valid_block_count > avail_user_block_count)) { 1980 diff = sbi->total_valid_block_count - avail_user_block_count; 1981 if (diff > *count) 1982 diff = *count; 1983 *count -= diff; 1984 release = diff; 1985 sbi->total_valid_block_count -= diff; 1986 if (!*count) { 1987 spin_unlock(&sbi->stat_lock); 1988 goto enospc; 1989 } 1990 } 1991 spin_unlock(&sbi->stat_lock); 1992 1993 if (unlikely(release)) { 1994 percpu_counter_sub(&sbi->alloc_valid_block_count, release); 1995 dquot_release_reservation_block(inode, release); 1996 } 1997 f2fs_i_blocks_write(inode, *count, true, true); 1998 return 0; 1999 2000 enospc: 2001 percpu_counter_sub(&sbi->alloc_valid_block_count, release); 2002 release_quota: 2003 dquot_release_reservation_block(inode, release); 2004 return -ENOSPC; 2005 } 2006 2007 __printf(2, 3) 2008 void f2fs_printk(struct f2fs_sb_info *sbi, const char *fmt, ...); 2009 2010 #define f2fs_err(sbi, fmt, ...) \ 2011 f2fs_printk(sbi, KERN_ERR fmt, ##__VA_ARGS__) 2012 #define f2fs_warn(sbi, fmt, ...) \ 2013 f2fs_printk(sbi, KERN_WARNING fmt, ##__VA_ARGS__) 2014 #define f2fs_notice(sbi, fmt, ...) \ 2015 f2fs_printk(sbi, KERN_NOTICE fmt, ##__VA_ARGS__) 2016 #define f2fs_info(sbi, fmt, ...) \ 2017 f2fs_printk(sbi, KERN_INFO fmt, ##__VA_ARGS__) 2018 #define f2fs_debug(sbi, fmt, ...) \ 2019 f2fs_printk(sbi, KERN_DEBUG fmt, ##__VA_ARGS__) 2020 2021 static inline void dec_valid_block_count(struct f2fs_sb_info *sbi, 2022 struct inode *inode, 2023 block_t count) 2024 { 2025 blkcnt_t sectors = count << F2FS_LOG_SECTORS_PER_BLOCK; 2026 2027 spin_lock(&sbi->stat_lock); 2028 f2fs_bug_on(sbi, sbi->total_valid_block_count < (block_t) count); 2029 sbi->total_valid_block_count -= (block_t)count; 2030 if (sbi->reserved_blocks && 2031 sbi->current_reserved_blocks < sbi->reserved_blocks) 2032 sbi->current_reserved_blocks = min(sbi->reserved_blocks, 2033 sbi->current_reserved_blocks + count); 2034 spin_unlock(&sbi->stat_lock); 2035 if (unlikely(inode->i_blocks < sectors)) { 2036 f2fs_warn(sbi, "Inconsistent i_blocks, ino:%lu, iblocks:%llu, sectors:%llu", 2037 inode->i_ino, 2038 (unsigned long long)inode->i_blocks, 2039 (unsigned long long)sectors); 2040 set_sbi_flag(sbi, SBI_NEED_FSCK); 2041 return; 2042 } 2043 f2fs_i_blocks_write(inode, count, false, true); 2044 } 2045 2046 static inline void inc_page_count(struct f2fs_sb_info *sbi, int count_type) 2047 { 2048 atomic_inc(&sbi->nr_pages[count_type]); 2049 2050 if (count_type == F2FS_DIRTY_DENTS || 2051 count_type == F2FS_DIRTY_NODES || 2052 count_type == F2FS_DIRTY_META || 2053 count_type == F2FS_DIRTY_QDATA || 2054 count_type == F2FS_DIRTY_IMETA) 2055 set_sbi_flag(sbi, SBI_IS_DIRTY); 2056 } 2057 2058 static inline void inode_inc_dirty_pages(struct inode *inode) 2059 { 2060 atomic_inc(&F2FS_I(inode)->dirty_pages); 2061 inc_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ? 2062 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA); 2063 if (IS_NOQUOTA(inode)) 2064 inc_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA); 2065 } 2066 2067 static inline void dec_page_count(struct f2fs_sb_info *sbi, int count_type) 2068 { 2069 atomic_dec(&sbi->nr_pages[count_type]); 2070 } 2071 2072 static inline void inode_dec_dirty_pages(struct inode *inode) 2073 { 2074 if (!S_ISDIR(inode->i_mode) && !S_ISREG(inode->i_mode) && 2075 !S_ISLNK(inode->i_mode)) 2076 return; 2077 2078 atomic_dec(&F2FS_I(inode)->dirty_pages); 2079 dec_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ? 2080 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA); 2081 if (IS_NOQUOTA(inode)) 2082 dec_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA); 2083 } 2084 2085 static inline s64 get_pages(struct f2fs_sb_info *sbi, int count_type) 2086 { 2087 return atomic_read(&sbi->nr_pages[count_type]); 2088 } 2089 2090 static inline int get_dirty_pages(struct inode *inode) 2091 { 2092 return atomic_read(&F2FS_I(inode)->dirty_pages); 2093 } 2094 2095 static inline int get_blocktype_secs(struct f2fs_sb_info *sbi, int block_type) 2096 { 2097 unsigned int pages_per_sec = sbi->segs_per_sec * sbi->blocks_per_seg; 2098 unsigned int segs = (get_pages(sbi, block_type) + pages_per_sec - 1) >> 2099 sbi->log_blocks_per_seg; 2100 2101 return segs / sbi->segs_per_sec; 2102 } 2103 2104 static inline block_t valid_user_blocks(struct f2fs_sb_info *sbi) 2105 { 2106 return sbi->total_valid_block_count; 2107 } 2108 2109 static inline block_t discard_blocks(struct f2fs_sb_info *sbi) 2110 { 2111 return sbi->discard_blks; 2112 } 2113 2114 static inline unsigned long __bitmap_size(struct f2fs_sb_info *sbi, int flag) 2115 { 2116 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi); 2117 2118 /* return NAT or SIT bitmap */ 2119 if (flag == NAT_BITMAP) 2120 return le32_to_cpu(ckpt->nat_ver_bitmap_bytesize); 2121 else if (flag == SIT_BITMAP) 2122 return le32_to_cpu(ckpt->sit_ver_bitmap_bytesize); 2123 2124 return 0; 2125 } 2126 2127 static inline block_t __cp_payload(struct f2fs_sb_info *sbi) 2128 { 2129 return le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_payload); 2130 } 2131 2132 static inline void *__bitmap_ptr(struct f2fs_sb_info *sbi, int flag) 2133 { 2134 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi); 2135 int offset; 2136 2137 if (is_set_ckpt_flags(sbi, CP_LARGE_NAT_BITMAP_FLAG)) { 2138 offset = (flag == SIT_BITMAP) ? 2139 le32_to_cpu(ckpt->nat_ver_bitmap_bytesize) : 0; 2140 /* 2141 * if large_nat_bitmap feature is enabled, leave checksum 2142 * protection for all nat/sit bitmaps. 2143 */ 2144 return &ckpt->sit_nat_version_bitmap + offset + sizeof(__le32); 2145 } 2146 2147 if (__cp_payload(sbi) > 0) { 2148 if (flag == NAT_BITMAP) 2149 return &ckpt->sit_nat_version_bitmap; 2150 else 2151 return (unsigned char *)ckpt + F2FS_BLKSIZE; 2152 } else { 2153 offset = (flag == NAT_BITMAP) ? 2154 le32_to_cpu(ckpt->sit_ver_bitmap_bytesize) : 0; 2155 return &ckpt->sit_nat_version_bitmap + offset; 2156 } 2157 } 2158 2159 static inline block_t __start_cp_addr(struct f2fs_sb_info *sbi) 2160 { 2161 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr); 2162 2163 if (sbi->cur_cp_pack == 2) 2164 start_addr += sbi->blocks_per_seg; 2165 return start_addr; 2166 } 2167 2168 static inline block_t __start_cp_next_addr(struct f2fs_sb_info *sbi) 2169 { 2170 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr); 2171 2172 if (sbi->cur_cp_pack == 1) 2173 start_addr += sbi->blocks_per_seg; 2174 return start_addr; 2175 } 2176 2177 static inline void __set_cp_next_pack(struct f2fs_sb_info *sbi) 2178 { 2179 sbi->cur_cp_pack = (sbi->cur_cp_pack == 1) ? 2 : 1; 2180 } 2181 2182 static inline block_t __start_sum_addr(struct f2fs_sb_info *sbi) 2183 { 2184 return le32_to_cpu(F2FS_CKPT(sbi)->cp_pack_start_sum); 2185 } 2186 2187 static inline int inc_valid_node_count(struct f2fs_sb_info *sbi, 2188 struct inode *inode, bool is_inode) 2189 { 2190 block_t valid_block_count; 2191 unsigned int valid_node_count, user_block_count; 2192 int err; 2193 2194 if (is_inode) { 2195 if (inode) { 2196 err = dquot_alloc_inode(inode); 2197 if (err) 2198 return err; 2199 } 2200 } else { 2201 err = dquot_reserve_block(inode, 1); 2202 if (err) 2203 return err; 2204 } 2205 2206 if (time_to_inject(sbi, FAULT_BLOCK)) { 2207 f2fs_show_injection_info(sbi, FAULT_BLOCK); 2208 goto enospc; 2209 } 2210 2211 spin_lock(&sbi->stat_lock); 2212 2213 valid_block_count = sbi->total_valid_block_count + 2214 sbi->current_reserved_blocks + 1; 2215 2216 if (!__allow_reserved_blocks(sbi, inode, false)) 2217 valid_block_count += F2FS_OPTION(sbi).root_reserved_blocks; 2218 user_block_count = sbi->user_block_count; 2219 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) 2220 user_block_count -= sbi->unusable_block_count; 2221 2222 if (unlikely(valid_block_count > user_block_count)) { 2223 spin_unlock(&sbi->stat_lock); 2224 goto enospc; 2225 } 2226 2227 valid_node_count = sbi->total_valid_node_count + 1; 2228 if (unlikely(valid_node_count > sbi->total_node_count)) { 2229 spin_unlock(&sbi->stat_lock); 2230 goto enospc; 2231 } 2232 2233 sbi->total_valid_node_count++; 2234 sbi->total_valid_block_count++; 2235 spin_unlock(&sbi->stat_lock); 2236 2237 if (inode) { 2238 if (is_inode) 2239 f2fs_mark_inode_dirty_sync(inode, true); 2240 else 2241 f2fs_i_blocks_write(inode, 1, true, true); 2242 } 2243 2244 percpu_counter_inc(&sbi->alloc_valid_block_count); 2245 return 0; 2246 2247 enospc: 2248 if (is_inode) { 2249 if (inode) 2250 dquot_free_inode(inode); 2251 } else { 2252 dquot_release_reservation_block(inode, 1); 2253 } 2254 return -ENOSPC; 2255 } 2256 2257 static inline void dec_valid_node_count(struct f2fs_sb_info *sbi, 2258 struct inode *inode, bool is_inode) 2259 { 2260 spin_lock(&sbi->stat_lock); 2261 2262 f2fs_bug_on(sbi, !sbi->total_valid_block_count); 2263 f2fs_bug_on(sbi, !sbi->total_valid_node_count); 2264 2265 sbi->total_valid_node_count--; 2266 sbi->total_valid_block_count--; 2267 if (sbi->reserved_blocks && 2268 sbi->current_reserved_blocks < sbi->reserved_blocks) 2269 sbi->current_reserved_blocks++; 2270 2271 spin_unlock(&sbi->stat_lock); 2272 2273 if (is_inode) { 2274 dquot_free_inode(inode); 2275 } else { 2276 if (unlikely(inode->i_blocks == 0)) { 2277 f2fs_warn(sbi, "dec_valid_node_count: inconsistent i_blocks, ino:%lu, iblocks:%llu", 2278 inode->i_ino, 2279 (unsigned long long)inode->i_blocks); 2280 set_sbi_flag(sbi, SBI_NEED_FSCK); 2281 return; 2282 } 2283 f2fs_i_blocks_write(inode, 1, false, true); 2284 } 2285 } 2286 2287 static inline unsigned int valid_node_count(struct f2fs_sb_info *sbi) 2288 { 2289 return sbi->total_valid_node_count; 2290 } 2291 2292 static inline void inc_valid_inode_count(struct f2fs_sb_info *sbi) 2293 { 2294 percpu_counter_inc(&sbi->total_valid_inode_count); 2295 } 2296 2297 static inline void dec_valid_inode_count(struct f2fs_sb_info *sbi) 2298 { 2299 percpu_counter_dec(&sbi->total_valid_inode_count); 2300 } 2301 2302 static inline s64 valid_inode_count(struct f2fs_sb_info *sbi) 2303 { 2304 return percpu_counter_sum_positive(&sbi->total_valid_inode_count); 2305 } 2306 2307 static inline struct page *f2fs_grab_cache_page(struct address_space *mapping, 2308 pgoff_t index, bool for_write) 2309 { 2310 struct page *page; 2311 2312 if (IS_ENABLED(CONFIG_F2FS_FAULT_INJECTION)) { 2313 if (!for_write) 2314 page = find_get_page_flags(mapping, index, 2315 FGP_LOCK | FGP_ACCESSED); 2316 else 2317 page = find_lock_page(mapping, index); 2318 if (page) 2319 return page; 2320 2321 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_ALLOC)) { 2322 f2fs_show_injection_info(F2FS_M_SB(mapping), 2323 FAULT_PAGE_ALLOC); 2324 return NULL; 2325 } 2326 } 2327 2328 if (!for_write) 2329 return grab_cache_page(mapping, index); 2330 return grab_cache_page_write_begin(mapping, index, AOP_FLAG_NOFS); 2331 } 2332 2333 static inline struct page *f2fs_pagecache_get_page( 2334 struct address_space *mapping, pgoff_t index, 2335 int fgp_flags, gfp_t gfp_mask) 2336 { 2337 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_GET)) { 2338 f2fs_show_injection_info(F2FS_M_SB(mapping), FAULT_PAGE_GET); 2339 return NULL; 2340 } 2341 2342 return pagecache_get_page(mapping, index, fgp_flags, gfp_mask); 2343 } 2344 2345 static inline void f2fs_copy_page(struct page *src, struct page *dst) 2346 { 2347 char *src_kaddr = kmap(src); 2348 char *dst_kaddr = kmap(dst); 2349 2350 memcpy(dst_kaddr, src_kaddr, PAGE_SIZE); 2351 kunmap(dst); 2352 kunmap(src); 2353 } 2354 2355 static inline void f2fs_put_page(struct page *page, int unlock) 2356 { 2357 if (!page) 2358 return; 2359 2360 if (unlock) { 2361 f2fs_bug_on(F2FS_P_SB(page), !PageLocked(page)); 2362 unlock_page(page); 2363 } 2364 put_page(page); 2365 } 2366 2367 static inline void f2fs_put_dnode(struct dnode_of_data *dn) 2368 { 2369 if (dn->node_page) 2370 f2fs_put_page(dn->node_page, 1); 2371 if (dn->inode_page && dn->node_page != dn->inode_page) 2372 f2fs_put_page(dn->inode_page, 0); 2373 dn->node_page = NULL; 2374 dn->inode_page = NULL; 2375 } 2376 2377 static inline struct kmem_cache *f2fs_kmem_cache_create(const char *name, 2378 size_t size) 2379 { 2380 return kmem_cache_create(name, size, 0, SLAB_RECLAIM_ACCOUNT, NULL); 2381 } 2382 2383 static inline void *f2fs_kmem_cache_alloc(struct kmem_cache *cachep, 2384 gfp_t flags) 2385 { 2386 void *entry; 2387 2388 entry = kmem_cache_alloc(cachep, flags); 2389 if (!entry) 2390 entry = kmem_cache_alloc(cachep, flags | __GFP_NOFAIL); 2391 return entry; 2392 } 2393 2394 static inline bool is_idle(struct f2fs_sb_info *sbi, int type) 2395 { 2396 if (sbi->gc_mode == GC_URGENT) 2397 return true; 2398 2399 if (get_pages(sbi, F2FS_RD_DATA) || get_pages(sbi, F2FS_RD_NODE) || 2400 get_pages(sbi, F2FS_RD_META) || get_pages(sbi, F2FS_WB_DATA) || 2401 get_pages(sbi, F2FS_WB_CP_DATA) || 2402 get_pages(sbi, F2FS_DIO_READ) || 2403 get_pages(sbi, F2FS_DIO_WRITE)) 2404 return false; 2405 2406 if (type != DISCARD_TIME && SM_I(sbi) && SM_I(sbi)->dcc_info && 2407 atomic_read(&SM_I(sbi)->dcc_info->queued_discard)) 2408 return false; 2409 2410 if (SM_I(sbi) && SM_I(sbi)->fcc_info && 2411 atomic_read(&SM_I(sbi)->fcc_info->queued_flush)) 2412 return false; 2413 2414 return f2fs_time_over(sbi, type); 2415 } 2416 2417 static inline void f2fs_radix_tree_insert(struct radix_tree_root *root, 2418 unsigned long index, void *item) 2419 { 2420 while (radix_tree_insert(root, index, item)) 2421 cond_resched(); 2422 } 2423 2424 #define RAW_IS_INODE(p) ((p)->footer.nid == (p)->footer.ino) 2425 2426 static inline bool IS_INODE(struct page *page) 2427 { 2428 struct f2fs_node *p = F2FS_NODE(page); 2429 2430 return RAW_IS_INODE(p); 2431 } 2432 2433 static inline int offset_in_addr(struct f2fs_inode *i) 2434 { 2435 return (i->i_inline & F2FS_EXTRA_ATTR) ? 2436 (le16_to_cpu(i->i_extra_isize) / sizeof(__le32)) : 0; 2437 } 2438 2439 static inline __le32 *blkaddr_in_node(struct f2fs_node *node) 2440 { 2441 return RAW_IS_INODE(node) ? node->i.i_addr : node->dn.addr; 2442 } 2443 2444 static inline int f2fs_has_extra_attr(struct inode *inode); 2445 static inline block_t data_blkaddr(struct inode *inode, 2446 struct page *node_page, unsigned int offset) 2447 { 2448 struct f2fs_node *raw_node; 2449 __le32 *addr_array; 2450 int base = 0; 2451 bool is_inode = IS_INODE(node_page); 2452 2453 raw_node = F2FS_NODE(node_page); 2454 2455 if (is_inode) { 2456 if (!inode) 2457 /* from GC path only */ 2458 base = offset_in_addr(&raw_node->i); 2459 else if (f2fs_has_extra_attr(inode)) 2460 base = get_extra_isize(inode); 2461 } 2462 2463 addr_array = blkaddr_in_node(raw_node); 2464 return le32_to_cpu(addr_array[base + offset]); 2465 } 2466 2467 static inline block_t f2fs_data_blkaddr(struct dnode_of_data *dn) 2468 { 2469 return data_blkaddr(dn->inode, dn->node_page, dn->ofs_in_node); 2470 } 2471 2472 static inline int f2fs_test_bit(unsigned int nr, char *addr) 2473 { 2474 int mask; 2475 2476 addr += (nr >> 3); 2477 mask = 1 << (7 - (nr & 0x07)); 2478 return mask & *addr; 2479 } 2480 2481 static inline void f2fs_set_bit(unsigned int nr, char *addr) 2482 { 2483 int mask; 2484 2485 addr += (nr >> 3); 2486 mask = 1 << (7 - (nr & 0x07)); 2487 *addr |= mask; 2488 } 2489 2490 static inline void f2fs_clear_bit(unsigned int nr, char *addr) 2491 { 2492 int mask; 2493 2494 addr += (nr >> 3); 2495 mask = 1 << (7 - (nr & 0x07)); 2496 *addr &= ~mask; 2497 } 2498 2499 static inline int f2fs_test_and_set_bit(unsigned int nr, char *addr) 2500 { 2501 int mask; 2502 int ret; 2503 2504 addr += (nr >> 3); 2505 mask = 1 << (7 - (nr & 0x07)); 2506 ret = mask & *addr; 2507 *addr |= mask; 2508 return ret; 2509 } 2510 2511 static inline int f2fs_test_and_clear_bit(unsigned int nr, char *addr) 2512 { 2513 int mask; 2514 int ret; 2515 2516 addr += (nr >> 3); 2517 mask = 1 << (7 - (nr & 0x07)); 2518 ret = mask & *addr; 2519 *addr &= ~mask; 2520 return ret; 2521 } 2522 2523 static inline void f2fs_change_bit(unsigned int nr, char *addr) 2524 { 2525 int mask; 2526 2527 addr += (nr >> 3); 2528 mask = 1 << (7 - (nr & 0x07)); 2529 *addr ^= mask; 2530 } 2531 2532 /* 2533 * On-disk inode flags (f2fs_inode::i_flags) 2534 */ 2535 #define F2FS_COMPR_FL 0x00000004 /* Compress file */ 2536 #define F2FS_SYNC_FL 0x00000008 /* Synchronous updates */ 2537 #define F2FS_IMMUTABLE_FL 0x00000010 /* Immutable file */ 2538 #define F2FS_APPEND_FL 0x00000020 /* writes to file may only append */ 2539 #define F2FS_NODUMP_FL 0x00000040 /* do not dump file */ 2540 #define F2FS_NOATIME_FL 0x00000080 /* do not update atime */ 2541 #define F2FS_NOCOMP_FL 0x00000400 /* Don't compress */ 2542 #define F2FS_INDEX_FL 0x00001000 /* hash-indexed directory */ 2543 #define F2FS_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */ 2544 #define F2FS_PROJINHERIT_FL 0x20000000 /* Create with parents projid */ 2545 #define F2FS_CASEFOLD_FL 0x40000000 /* Casefolded file */ 2546 2547 /* Flags that should be inherited by new inodes from their parent. */ 2548 #define F2FS_FL_INHERITED (F2FS_SYNC_FL | F2FS_NODUMP_FL | F2FS_NOATIME_FL | \ 2549 F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \ 2550 F2FS_CASEFOLD_FL | F2FS_COMPR_FL | F2FS_NOCOMP_FL) 2551 2552 /* Flags that are appropriate for regular files (all but dir-specific ones). */ 2553 #define F2FS_REG_FLMASK (~(F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \ 2554 F2FS_CASEFOLD_FL)) 2555 2556 /* Flags that are appropriate for non-directories/regular files. */ 2557 #define F2FS_OTHER_FLMASK (F2FS_NODUMP_FL | F2FS_NOATIME_FL) 2558 2559 static inline __u32 f2fs_mask_flags(umode_t mode, __u32 flags) 2560 { 2561 if (S_ISDIR(mode)) 2562 return flags; 2563 else if (S_ISREG(mode)) 2564 return flags & F2FS_REG_FLMASK; 2565 else 2566 return flags & F2FS_OTHER_FLMASK; 2567 } 2568 2569 static inline void __mark_inode_dirty_flag(struct inode *inode, 2570 int flag, bool set) 2571 { 2572 switch (flag) { 2573 case FI_INLINE_XATTR: 2574 case FI_INLINE_DATA: 2575 case FI_INLINE_DENTRY: 2576 case FI_NEW_INODE: 2577 if (set) 2578 return; 2579 /* fall through */ 2580 case FI_DATA_EXIST: 2581 case FI_INLINE_DOTS: 2582 case FI_PIN_FILE: 2583 f2fs_mark_inode_dirty_sync(inode, true); 2584 } 2585 } 2586 2587 static inline void set_inode_flag(struct inode *inode, int flag) 2588 { 2589 test_and_set_bit(flag, F2FS_I(inode)->flags); 2590 __mark_inode_dirty_flag(inode, flag, true); 2591 } 2592 2593 static inline int is_inode_flag_set(struct inode *inode, int flag) 2594 { 2595 return test_bit(flag, F2FS_I(inode)->flags); 2596 } 2597 2598 static inline void clear_inode_flag(struct inode *inode, int flag) 2599 { 2600 test_and_clear_bit(flag, F2FS_I(inode)->flags); 2601 __mark_inode_dirty_flag(inode, flag, false); 2602 } 2603 2604 static inline bool f2fs_verity_in_progress(struct inode *inode) 2605 { 2606 return IS_ENABLED(CONFIG_FS_VERITY) && 2607 is_inode_flag_set(inode, FI_VERITY_IN_PROGRESS); 2608 } 2609 2610 static inline void set_acl_inode(struct inode *inode, umode_t mode) 2611 { 2612 F2FS_I(inode)->i_acl_mode = mode; 2613 set_inode_flag(inode, FI_ACL_MODE); 2614 f2fs_mark_inode_dirty_sync(inode, false); 2615 } 2616 2617 static inline void f2fs_i_links_write(struct inode *inode, bool inc) 2618 { 2619 if (inc) 2620 inc_nlink(inode); 2621 else 2622 drop_nlink(inode); 2623 f2fs_mark_inode_dirty_sync(inode, true); 2624 } 2625 2626 static inline void f2fs_i_blocks_write(struct inode *inode, 2627 block_t diff, bool add, bool claim) 2628 { 2629 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE); 2630 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER); 2631 2632 /* add = 1, claim = 1 should be dquot_reserve_block in pair */ 2633 if (add) { 2634 if (claim) 2635 dquot_claim_block(inode, diff); 2636 else 2637 dquot_alloc_block_nofail(inode, diff); 2638 } else { 2639 dquot_free_block(inode, diff); 2640 } 2641 2642 f2fs_mark_inode_dirty_sync(inode, true); 2643 if (clean || recover) 2644 set_inode_flag(inode, FI_AUTO_RECOVER); 2645 } 2646 2647 static inline void f2fs_i_size_write(struct inode *inode, loff_t i_size) 2648 { 2649 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE); 2650 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER); 2651 2652 if (i_size_read(inode) == i_size) 2653 return; 2654 2655 i_size_write(inode, i_size); 2656 f2fs_mark_inode_dirty_sync(inode, true); 2657 if (clean || recover) 2658 set_inode_flag(inode, FI_AUTO_RECOVER); 2659 } 2660 2661 static inline void f2fs_i_depth_write(struct inode *inode, unsigned int depth) 2662 { 2663 F2FS_I(inode)->i_current_depth = depth; 2664 f2fs_mark_inode_dirty_sync(inode, true); 2665 } 2666 2667 static inline void f2fs_i_gc_failures_write(struct inode *inode, 2668 unsigned int count) 2669 { 2670 F2FS_I(inode)->i_gc_failures[GC_FAILURE_PIN] = count; 2671 f2fs_mark_inode_dirty_sync(inode, true); 2672 } 2673 2674 static inline void f2fs_i_xnid_write(struct inode *inode, nid_t xnid) 2675 { 2676 F2FS_I(inode)->i_xattr_nid = xnid; 2677 f2fs_mark_inode_dirty_sync(inode, true); 2678 } 2679 2680 static inline void f2fs_i_pino_write(struct inode *inode, nid_t pino) 2681 { 2682 F2FS_I(inode)->i_pino = pino; 2683 f2fs_mark_inode_dirty_sync(inode, true); 2684 } 2685 2686 static inline void get_inline_info(struct inode *inode, struct f2fs_inode *ri) 2687 { 2688 struct f2fs_inode_info *fi = F2FS_I(inode); 2689 2690 if (ri->i_inline & F2FS_INLINE_XATTR) 2691 set_bit(FI_INLINE_XATTR, fi->flags); 2692 if (ri->i_inline & F2FS_INLINE_DATA) 2693 set_bit(FI_INLINE_DATA, fi->flags); 2694 if (ri->i_inline & F2FS_INLINE_DENTRY) 2695 set_bit(FI_INLINE_DENTRY, fi->flags); 2696 if (ri->i_inline & F2FS_DATA_EXIST) 2697 set_bit(FI_DATA_EXIST, fi->flags); 2698 if (ri->i_inline & F2FS_INLINE_DOTS) 2699 set_bit(FI_INLINE_DOTS, fi->flags); 2700 if (ri->i_inline & F2FS_EXTRA_ATTR) 2701 set_bit(FI_EXTRA_ATTR, fi->flags); 2702 if (ri->i_inline & F2FS_PIN_FILE) 2703 set_bit(FI_PIN_FILE, fi->flags); 2704 } 2705 2706 static inline void set_raw_inline(struct inode *inode, struct f2fs_inode *ri) 2707 { 2708 ri->i_inline = 0; 2709 2710 if (is_inode_flag_set(inode, FI_INLINE_XATTR)) 2711 ri->i_inline |= F2FS_INLINE_XATTR; 2712 if (is_inode_flag_set(inode, FI_INLINE_DATA)) 2713 ri->i_inline |= F2FS_INLINE_DATA; 2714 if (is_inode_flag_set(inode, FI_INLINE_DENTRY)) 2715 ri->i_inline |= F2FS_INLINE_DENTRY; 2716 if (is_inode_flag_set(inode, FI_DATA_EXIST)) 2717 ri->i_inline |= F2FS_DATA_EXIST; 2718 if (is_inode_flag_set(inode, FI_INLINE_DOTS)) 2719 ri->i_inline |= F2FS_INLINE_DOTS; 2720 if (is_inode_flag_set(inode, FI_EXTRA_ATTR)) 2721 ri->i_inline |= F2FS_EXTRA_ATTR; 2722 if (is_inode_flag_set(inode, FI_PIN_FILE)) 2723 ri->i_inline |= F2FS_PIN_FILE; 2724 } 2725 2726 static inline int f2fs_has_extra_attr(struct inode *inode) 2727 { 2728 return is_inode_flag_set(inode, FI_EXTRA_ATTR); 2729 } 2730 2731 static inline int f2fs_has_inline_xattr(struct inode *inode) 2732 { 2733 return is_inode_flag_set(inode, FI_INLINE_XATTR); 2734 } 2735 2736 static inline int f2fs_compressed_file(struct inode *inode) 2737 { 2738 return S_ISREG(inode->i_mode) && 2739 is_inode_flag_set(inode, FI_COMPRESSED_FILE); 2740 } 2741 2742 static inline unsigned int addrs_per_inode(struct inode *inode) 2743 { 2744 unsigned int addrs = CUR_ADDRS_PER_INODE(inode) - 2745 get_inline_xattr_addrs(inode); 2746 2747 if (!f2fs_compressed_file(inode)) 2748 return addrs; 2749 return ALIGN_DOWN(addrs, F2FS_I(inode)->i_cluster_size); 2750 } 2751 2752 static inline unsigned int addrs_per_block(struct inode *inode) 2753 { 2754 if (!f2fs_compressed_file(inode)) 2755 return DEF_ADDRS_PER_BLOCK; 2756 return ALIGN_DOWN(DEF_ADDRS_PER_BLOCK, F2FS_I(inode)->i_cluster_size); 2757 } 2758 2759 static inline void *inline_xattr_addr(struct inode *inode, struct page *page) 2760 { 2761 struct f2fs_inode *ri = F2FS_INODE(page); 2762 2763 return (void *)&(ri->i_addr[DEF_ADDRS_PER_INODE - 2764 get_inline_xattr_addrs(inode)]); 2765 } 2766 2767 static inline int inline_xattr_size(struct inode *inode) 2768 { 2769 if (f2fs_has_inline_xattr(inode)) 2770 return get_inline_xattr_addrs(inode) * sizeof(__le32); 2771 return 0; 2772 } 2773 2774 static inline int f2fs_has_inline_data(struct inode *inode) 2775 { 2776 return is_inode_flag_set(inode, FI_INLINE_DATA); 2777 } 2778 2779 static inline int f2fs_exist_data(struct inode *inode) 2780 { 2781 return is_inode_flag_set(inode, FI_DATA_EXIST); 2782 } 2783 2784 static inline int f2fs_has_inline_dots(struct inode *inode) 2785 { 2786 return is_inode_flag_set(inode, FI_INLINE_DOTS); 2787 } 2788 2789 static inline int f2fs_is_mmap_file(struct inode *inode) 2790 { 2791 return is_inode_flag_set(inode, FI_MMAP_FILE); 2792 } 2793 2794 static inline bool f2fs_is_pinned_file(struct inode *inode) 2795 { 2796 return is_inode_flag_set(inode, FI_PIN_FILE); 2797 } 2798 2799 static inline bool f2fs_is_atomic_file(struct inode *inode) 2800 { 2801 return is_inode_flag_set(inode, FI_ATOMIC_FILE); 2802 } 2803 2804 static inline bool f2fs_is_commit_atomic_write(struct inode *inode) 2805 { 2806 return is_inode_flag_set(inode, FI_ATOMIC_COMMIT); 2807 } 2808 2809 static inline bool f2fs_is_volatile_file(struct inode *inode) 2810 { 2811 return is_inode_flag_set(inode, FI_VOLATILE_FILE); 2812 } 2813 2814 static inline bool f2fs_is_first_block_written(struct inode *inode) 2815 { 2816 return is_inode_flag_set(inode, FI_FIRST_BLOCK_WRITTEN); 2817 } 2818 2819 static inline bool f2fs_is_drop_cache(struct inode *inode) 2820 { 2821 return is_inode_flag_set(inode, FI_DROP_CACHE); 2822 } 2823 2824 static inline void *inline_data_addr(struct inode *inode, struct page *page) 2825 { 2826 struct f2fs_inode *ri = F2FS_INODE(page); 2827 int extra_size = get_extra_isize(inode); 2828 2829 return (void *)&(ri->i_addr[extra_size + DEF_INLINE_RESERVED_SIZE]); 2830 } 2831 2832 static inline int f2fs_has_inline_dentry(struct inode *inode) 2833 { 2834 return is_inode_flag_set(inode, FI_INLINE_DENTRY); 2835 } 2836 2837 static inline int is_file(struct inode *inode, int type) 2838 { 2839 return F2FS_I(inode)->i_advise & type; 2840 } 2841 2842 static inline void set_file(struct inode *inode, int type) 2843 { 2844 F2FS_I(inode)->i_advise |= type; 2845 f2fs_mark_inode_dirty_sync(inode, true); 2846 } 2847 2848 static inline void clear_file(struct inode *inode, int type) 2849 { 2850 F2FS_I(inode)->i_advise &= ~type; 2851 f2fs_mark_inode_dirty_sync(inode, true); 2852 } 2853 2854 static inline bool f2fs_is_time_consistent(struct inode *inode) 2855 { 2856 if (!timespec64_equal(F2FS_I(inode)->i_disk_time, &inode->i_atime)) 2857 return false; 2858 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 1, &inode->i_ctime)) 2859 return false; 2860 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 2, &inode->i_mtime)) 2861 return false; 2862 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 3, 2863 &F2FS_I(inode)->i_crtime)) 2864 return false; 2865 return true; 2866 } 2867 2868 static inline bool f2fs_skip_inode_update(struct inode *inode, int dsync) 2869 { 2870 bool ret; 2871 2872 if (dsync) { 2873 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 2874 2875 spin_lock(&sbi->inode_lock[DIRTY_META]); 2876 ret = list_empty(&F2FS_I(inode)->gdirty_list); 2877 spin_unlock(&sbi->inode_lock[DIRTY_META]); 2878 return ret; 2879 } 2880 if (!is_inode_flag_set(inode, FI_AUTO_RECOVER) || 2881 file_keep_isize(inode) || 2882 i_size_read(inode) & ~PAGE_MASK) 2883 return false; 2884 2885 if (!f2fs_is_time_consistent(inode)) 2886 return false; 2887 2888 spin_lock(&F2FS_I(inode)->i_size_lock); 2889 ret = F2FS_I(inode)->last_disk_size == i_size_read(inode); 2890 spin_unlock(&F2FS_I(inode)->i_size_lock); 2891 2892 return ret; 2893 } 2894 2895 static inline bool f2fs_readonly(struct super_block *sb) 2896 { 2897 return sb_rdonly(sb); 2898 } 2899 2900 static inline bool f2fs_cp_error(struct f2fs_sb_info *sbi) 2901 { 2902 return is_set_ckpt_flags(sbi, CP_ERROR_FLAG); 2903 } 2904 2905 static inline bool is_dot_dotdot(const struct qstr *str) 2906 { 2907 if (str->len == 1 && str->name[0] == '.') 2908 return true; 2909 2910 if (str->len == 2 && str->name[0] == '.' && str->name[1] == '.') 2911 return true; 2912 2913 return false; 2914 } 2915 2916 static inline bool f2fs_may_extent_tree(struct inode *inode) 2917 { 2918 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 2919 2920 if (!test_opt(sbi, EXTENT_CACHE) || 2921 is_inode_flag_set(inode, FI_NO_EXTENT) || 2922 is_inode_flag_set(inode, FI_COMPRESSED_FILE)) 2923 return false; 2924 2925 /* 2926 * for recovered files during mount do not create extents 2927 * if shrinker is not registered. 2928 */ 2929 if (list_empty(&sbi->s_list)) 2930 return false; 2931 2932 return S_ISREG(inode->i_mode); 2933 } 2934 2935 static inline void *f2fs_kmalloc(struct f2fs_sb_info *sbi, 2936 size_t size, gfp_t flags) 2937 { 2938 void *ret; 2939 2940 if (time_to_inject(sbi, FAULT_KMALLOC)) { 2941 f2fs_show_injection_info(sbi, FAULT_KMALLOC); 2942 return NULL; 2943 } 2944 2945 ret = kmalloc(size, flags); 2946 if (ret) 2947 return ret; 2948 2949 return kvmalloc(size, flags); 2950 } 2951 2952 static inline void *f2fs_kzalloc(struct f2fs_sb_info *sbi, 2953 size_t size, gfp_t flags) 2954 { 2955 return f2fs_kmalloc(sbi, size, flags | __GFP_ZERO); 2956 } 2957 2958 static inline void *f2fs_kvmalloc(struct f2fs_sb_info *sbi, 2959 size_t size, gfp_t flags) 2960 { 2961 if (time_to_inject(sbi, FAULT_KVMALLOC)) { 2962 f2fs_show_injection_info(sbi, FAULT_KVMALLOC); 2963 return NULL; 2964 } 2965 2966 return kvmalloc(size, flags); 2967 } 2968 2969 static inline void *f2fs_kvzalloc(struct f2fs_sb_info *sbi, 2970 size_t size, gfp_t flags) 2971 { 2972 return f2fs_kvmalloc(sbi, size, flags | __GFP_ZERO); 2973 } 2974 2975 static inline int get_extra_isize(struct inode *inode) 2976 { 2977 return F2FS_I(inode)->i_extra_isize / sizeof(__le32); 2978 } 2979 2980 static inline int get_inline_xattr_addrs(struct inode *inode) 2981 { 2982 return F2FS_I(inode)->i_inline_xattr_size; 2983 } 2984 2985 #define f2fs_get_inode_mode(i) \ 2986 ((is_inode_flag_set(i, FI_ACL_MODE)) ? \ 2987 (F2FS_I(i)->i_acl_mode) : ((i)->i_mode)) 2988 2989 #define F2FS_TOTAL_EXTRA_ATTR_SIZE \ 2990 (offsetof(struct f2fs_inode, i_extra_end) - \ 2991 offsetof(struct f2fs_inode, i_extra_isize)) \ 2992 2993 #define F2FS_OLD_ATTRIBUTE_SIZE (offsetof(struct f2fs_inode, i_addr)) 2994 #define F2FS_FITS_IN_INODE(f2fs_inode, extra_isize, field) \ 2995 ((offsetof(typeof(*(f2fs_inode)), field) + \ 2996 sizeof((f2fs_inode)->field)) \ 2997 <= (F2FS_OLD_ATTRIBUTE_SIZE + (extra_isize))) \ 2998 2999 static inline void f2fs_reset_iostat(struct f2fs_sb_info *sbi) 3000 { 3001 int i; 3002 3003 spin_lock(&sbi->iostat_lock); 3004 for (i = 0; i < NR_IO_TYPE; i++) 3005 sbi->write_iostat[i] = 0; 3006 spin_unlock(&sbi->iostat_lock); 3007 } 3008 3009 static inline void f2fs_update_iostat(struct f2fs_sb_info *sbi, 3010 enum iostat_type type, unsigned long long io_bytes) 3011 { 3012 if (!sbi->iostat_enable) 3013 return; 3014 spin_lock(&sbi->iostat_lock); 3015 sbi->write_iostat[type] += io_bytes; 3016 3017 if (type == APP_WRITE_IO || type == APP_DIRECT_IO) 3018 sbi->write_iostat[APP_BUFFERED_IO] = 3019 sbi->write_iostat[APP_WRITE_IO] - 3020 sbi->write_iostat[APP_DIRECT_IO]; 3021 spin_unlock(&sbi->iostat_lock); 3022 } 3023 3024 #define __is_large_section(sbi) ((sbi)->segs_per_sec > 1) 3025 3026 #define __is_meta_io(fio) (PAGE_TYPE_OF_BIO((fio)->type) == META) 3027 3028 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi, 3029 block_t blkaddr, int type); 3030 static inline void verify_blkaddr(struct f2fs_sb_info *sbi, 3031 block_t blkaddr, int type) 3032 { 3033 if (!f2fs_is_valid_blkaddr(sbi, blkaddr, type)) { 3034 f2fs_err(sbi, "invalid blkaddr: %u, type: %d, run fsck to fix.", 3035 blkaddr, type); 3036 f2fs_bug_on(sbi, 1); 3037 } 3038 } 3039 3040 static inline bool __is_valid_data_blkaddr(block_t blkaddr) 3041 { 3042 if (blkaddr == NEW_ADDR || blkaddr == NULL_ADDR || 3043 blkaddr == COMPRESS_ADDR) 3044 return false; 3045 return true; 3046 } 3047 3048 static inline void f2fs_set_page_private(struct page *page, 3049 unsigned long data) 3050 { 3051 if (PagePrivate(page)) 3052 return; 3053 3054 attach_page_private(page, (void *)data); 3055 } 3056 3057 static inline void f2fs_clear_page_private(struct page *page) 3058 { 3059 detach_page_private(page); 3060 } 3061 3062 /* 3063 * file.c 3064 */ 3065 int f2fs_sync_file(struct file *file, loff_t start, loff_t end, int datasync); 3066 void f2fs_truncate_data_blocks(struct dnode_of_data *dn); 3067 int f2fs_truncate_blocks(struct inode *inode, u64 from, bool lock); 3068 int f2fs_truncate(struct inode *inode); 3069 int f2fs_getattr(const struct path *path, struct kstat *stat, 3070 u32 request_mask, unsigned int flags); 3071 int f2fs_setattr(struct dentry *dentry, struct iattr *attr); 3072 int f2fs_truncate_hole(struct inode *inode, pgoff_t pg_start, pgoff_t pg_end); 3073 void f2fs_truncate_data_blocks_range(struct dnode_of_data *dn, int count); 3074 int f2fs_precache_extents(struct inode *inode); 3075 long f2fs_ioctl(struct file *filp, unsigned int cmd, unsigned long arg); 3076 long f2fs_compat_ioctl(struct file *file, unsigned int cmd, unsigned long arg); 3077 int f2fs_transfer_project_quota(struct inode *inode, kprojid_t kprojid); 3078 int f2fs_pin_file_control(struct inode *inode, bool inc); 3079 3080 /* 3081 * inode.c 3082 */ 3083 void f2fs_set_inode_flags(struct inode *inode); 3084 bool f2fs_inode_chksum_verify(struct f2fs_sb_info *sbi, struct page *page); 3085 void f2fs_inode_chksum_set(struct f2fs_sb_info *sbi, struct page *page); 3086 struct inode *f2fs_iget(struct super_block *sb, unsigned long ino); 3087 struct inode *f2fs_iget_retry(struct super_block *sb, unsigned long ino); 3088 int f2fs_try_to_free_nats(struct f2fs_sb_info *sbi, int nr_shrink); 3089 void f2fs_update_inode(struct inode *inode, struct page *node_page); 3090 void f2fs_update_inode_page(struct inode *inode); 3091 int f2fs_write_inode(struct inode *inode, struct writeback_control *wbc); 3092 void f2fs_evict_inode(struct inode *inode); 3093 void f2fs_handle_failed_inode(struct inode *inode); 3094 3095 /* 3096 * namei.c 3097 */ 3098 int f2fs_update_extension_list(struct f2fs_sb_info *sbi, const char *name, 3099 bool hot, bool set); 3100 struct dentry *f2fs_get_parent(struct dentry *child); 3101 3102 extern int f2fs_ci_compare(const struct inode *parent, 3103 const struct qstr *name, 3104 const struct qstr *entry, 3105 bool quick); 3106 3107 /* 3108 * dir.c 3109 */ 3110 unsigned char f2fs_get_de_type(struct f2fs_dir_entry *de); 3111 struct f2fs_dir_entry *f2fs_find_target_dentry(struct fscrypt_name *fname, 3112 f2fs_hash_t namehash, int *max_slots, 3113 struct f2fs_dentry_ptr *d); 3114 int f2fs_fill_dentries(struct dir_context *ctx, struct f2fs_dentry_ptr *d, 3115 unsigned int start_pos, struct fscrypt_str *fstr); 3116 void f2fs_do_make_empty_dir(struct inode *inode, struct inode *parent, 3117 struct f2fs_dentry_ptr *d); 3118 struct page *f2fs_init_inode_metadata(struct inode *inode, struct inode *dir, 3119 const struct qstr *new_name, 3120 const struct qstr *orig_name, struct page *dpage); 3121 void f2fs_update_parent_metadata(struct inode *dir, struct inode *inode, 3122 unsigned int current_depth); 3123 int f2fs_room_for_filename(const void *bitmap, int slots, int max_slots); 3124 void f2fs_drop_nlink(struct inode *dir, struct inode *inode); 3125 struct f2fs_dir_entry *__f2fs_find_entry(struct inode *dir, 3126 struct fscrypt_name *fname, struct page **res_page); 3127 struct f2fs_dir_entry *f2fs_find_entry(struct inode *dir, 3128 const struct qstr *child, struct page **res_page); 3129 struct f2fs_dir_entry *f2fs_parent_dir(struct inode *dir, struct page **p); 3130 ino_t f2fs_inode_by_name(struct inode *dir, const struct qstr *qstr, 3131 struct page **page); 3132 void f2fs_set_link(struct inode *dir, struct f2fs_dir_entry *de, 3133 struct page *page, struct inode *inode); 3134 bool f2fs_has_enough_room(struct inode *dir, struct page *ipage, 3135 struct fscrypt_name *fname); 3136 void f2fs_update_dentry(nid_t ino, umode_t mode, struct f2fs_dentry_ptr *d, 3137 const struct qstr *name, f2fs_hash_t name_hash, 3138 unsigned int bit_pos); 3139 int f2fs_add_regular_entry(struct inode *dir, const struct qstr *new_name, 3140 const struct qstr *orig_name, 3141 struct inode *inode, nid_t ino, umode_t mode); 3142 int f2fs_add_dentry(struct inode *dir, struct fscrypt_name *fname, 3143 struct inode *inode, nid_t ino, umode_t mode); 3144 int f2fs_do_add_link(struct inode *dir, const struct qstr *name, 3145 struct inode *inode, nid_t ino, umode_t mode); 3146 void f2fs_delete_entry(struct f2fs_dir_entry *dentry, struct page *page, 3147 struct inode *dir, struct inode *inode); 3148 int f2fs_do_tmpfile(struct inode *inode, struct inode *dir); 3149 bool f2fs_empty_dir(struct inode *dir); 3150 3151 static inline int f2fs_add_link(struct dentry *dentry, struct inode *inode) 3152 { 3153 return f2fs_do_add_link(d_inode(dentry->d_parent), &dentry->d_name, 3154 inode, inode->i_ino, inode->i_mode); 3155 } 3156 3157 /* 3158 * super.c 3159 */ 3160 int f2fs_inode_dirtied(struct inode *inode, bool sync); 3161 void f2fs_inode_synced(struct inode *inode); 3162 int f2fs_enable_quota_files(struct f2fs_sb_info *sbi, bool rdonly); 3163 int f2fs_quota_sync(struct super_block *sb, int type); 3164 void f2fs_quota_off_umount(struct super_block *sb); 3165 int f2fs_commit_super(struct f2fs_sb_info *sbi, bool recover); 3166 int f2fs_sync_fs(struct super_block *sb, int sync); 3167 int f2fs_sanity_check_ckpt(struct f2fs_sb_info *sbi); 3168 3169 /* 3170 * hash.c 3171 */ 3172 f2fs_hash_t f2fs_dentry_hash(const struct inode *dir, 3173 const struct qstr *name_info, struct fscrypt_name *fname); 3174 3175 /* 3176 * node.c 3177 */ 3178 struct dnode_of_data; 3179 struct node_info; 3180 3181 int f2fs_check_nid_range(struct f2fs_sb_info *sbi, nid_t nid); 3182 bool f2fs_available_free_memory(struct f2fs_sb_info *sbi, int type); 3183 bool f2fs_in_warm_node_list(struct f2fs_sb_info *sbi, struct page *page); 3184 void f2fs_init_fsync_node_info(struct f2fs_sb_info *sbi); 3185 void f2fs_del_fsync_node_entry(struct f2fs_sb_info *sbi, struct page *page); 3186 void f2fs_reset_fsync_node_info(struct f2fs_sb_info *sbi); 3187 int f2fs_need_dentry_mark(struct f2fs_sb_info *sbi, nid_t nid); 3188 bool f2fs_is_checkpointed_node(struct f2fs_sb_info *sbi, nid_t nid); 3189 bool f2fs_need_inode_block_update(struct f2fs_sb_info *sbi, nid_t ino); 3190 int f2fs_get_node_info(struct f2fs_sb_info *sbi, nid_t nid, 3191 struct node_info *ni); 3192 pgoff_t f2fs_get_next_page_offset(struct dnode_of_data *dn, pgoff_t pgofs); 3193 int f2fs_get_dnode_of_data(struct dnode_of_data *dn, pgoff_t index, int mode); 3194 int f2fs_truncate_inode_blocks(struct inode *inode, pgoff_t from); 3195 int f2fs_truncate_xattr_node(struct inode *inode); 3196 int f2fs_wait_on_node_pages_writeback(struct f2fs_sb_info *sbi, 3197 unsigned int seq_id); 3198 int f2fs_remove_inode_page(struct inode *inode); 3199 struct page *f2fs_new_inode_page(struct inode *inode); 3200 struct page *f2fs_new_node_page(struct dnode_of_data *dn, unsigned int ofs); 3201 void f2fs_ra_node_page(struct f2fs_sb_info *sbi, nid_t nid); 3202 struct page *f2fs_get_node_page(struct f2fs_sb_info *sbi, pgoff_t nid); 3203 struct page *f2fs_get_node_page_ra(struct page *parent, int start); 3204 int f2fs_move_node_page(struct page *node_page, int gc_type); 3205 int f2fs_fsync_node_pages(struct f2fs_sb_info *sbi, struct inode *inode, 3206 struct writeback_control *wbc, bool atomic, 3207 unsigned int *seq_id); 3208 int f2fs_sync_node_pages(struct f2fs_sb_info *sbi, 3209 struct writeback_control *wbc, 3210 bool do_balance, enum iostat_type io_type); 3211 int f2fs_build_free_nids(struct f2fs_sb_info *sbi, bool sync, bool mount); 3212 bool f2fs_alloc_nid(struct f2fs_sb_info *sbi, nid_t *nid); 3213 void f2fs_alloc_nid_done(struct f2fs_sb_info *sbi, nid_t nid); 3214 void f2fs_alloc_nid_failed(struct f2fs_sb_info *sbi, nid_t nid); 3215 int f2fs_try_to_free_nids(struct f2fs_sb_info *sbi, int nr_shrink); 3216 void f2fs_recover_inline_xattr(struct inode *inode, struct page *page); 3217 int f2fs_recover_xattr_data(struct inode *inode, struct page *page); 3218 int f2fs_recover_inode_page(struct f2fs_sb_info *sbi, struct page *page); 3219 int f2fs_restore_node_summary(struct f2fs_sb_info *sbi, 3220 unsigned int segno, struct f2fs_summary_block *sum); 3221 int f2fs_flush_nat_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3222 int f2fs_build_node_manager(struct f2fs_sb_info *sbi); 3223 void f2fs_destroy_node_manager(struct f2fs_sb_info *sbi); 3224 int __init f2fs_create_node_manager_caches(void); 3225 void f2fs_destroy_node_manager_caches(void); 3226 3227 /* 3228 * segment.c 3229 */ 3230 bool f2fs_need_SSR(struct f2fs_sb_info *sbi); 3231 void f2fs_register_inmem_page(struct inode *inode, struct page *page); 3232 void f2fs_drop_inmem_pages_all(struct f2fs_sb_info *sbi, bool gc_failure); 3233 void f2fs_drop_inmem_pages(struct inode *inode); 3234 void f2fs_drop_inmem_page(struct inode *inode, struct page *page); 3235 int f2fs_commit_inmem_pages(struct inode *inode); 3236 void f2fs_balance_fs(struct f2fs_sb_info *sbi, bool need); 3237 void f2fs_balance_fs_bg(struct f2fs_sb_info *sbi, bool from_bg); 3238 int f2fs_issue_flush(struct f2fs_sb_info *sbi, nid_t ino); 3239 int f2fs_create_flush_cmd_control(struct f2fs_sb_info *sbi); 3240 int f2fs_flush_device_cache(struct f2fs_sb_info *sbi); 3241 void f2fs_destroy_flush_cmd_control(struct f2fs_sb_info *sbi, bool free); 3242 void f2fs_invalidate_blocks(struct f2fs_sb_info *sbi, block_t addr); 3243 bool f2fs_is_checkpointed_data(struct f2fs_sb_info *sbi, block_t blkaddr); 3244 void f2fs_drop_discard_cmd(struct f2fs_sb_info *sbi); 3245 void f2fs_stop_discard_thread(struct f2fs_sb_info *sbi); 3246 bool f2fs_issue_discard_timeout(struct f2fs_sb_info *sbi); 3247 void f2fs_clear_prefree_segments(struct f2fs_sb_info *sbi, 3248 struct cp_control *cpc); 3249 void f2fs_dirty_to_prefree(struct f2fs_sb_info *sbi); 3250 block_t f2fs_get_unusable_blocks(struct f2fs_sb_info *sbi); 3251 int f2fs_disable_cp_again(struct f2fs_sb_info *sbi, block_t unusable); 3252 void f2fs_release_discard_addrs(struct f2fs_sb_info *sbi); 3253 int f2fs_npages_for_summary_flush(struct f2fs_sb_info *sbi, bool for_ra); 3254 void allocate_segment_for_resize(struct f2fs_sb_info *sbi, int type, 3255 unsigned int start, unsigned int end); 3256 void f2fs_allocate_new_segments(struct f2fs_sb_info *sbi, int type); 3257 int f2fs_trim_fs(struct f2fs_sb_info *sbi, struct fstrim_range *range); 3258 bool f2fs_exist_trim_candidates(struct f2fs_sb_info *sbi, 3259 struct cp_control *cpc); 3260 struct page *f2fs_get_sum_page(struct f2fs_sb_info *sbi, unsigned int segno); 3261 void f2fs_update_meta_page(struct f2fs_sb_info *sbi, void *src, 3262 block_t blk_addr); 3263 void f2fs_do_write_meta_page(struct f2fs_sb_info *sbi, struct page *page, 3264 enum iostat_type io_type); 3265 void f2fs_do_write_node_page(unsigned int nid, struct f2fs_io_info *fio); 3266 void f2fs_outplace_write_data(struct dnode_of_data *dn, 3267 struct f2fs_io_info *fio); 3268 int f2fs_inplace_write_data(struct f2fs_io_info *fio); 3269 void f2fs_do_replace_block(struct f2fs_sb_info *sbi, struct f2fs_summary *sum, 3270 block_t old_blkaddr, block_t new_blkaddr, 3271 bool recover_curseg, bool recover_newaddr); 3272 void f2fs_replace_block(struct f2fs_sb_info *sbi, struct dnode_of_data *dn, 3273 block_t old_addr, block_t new_addr, 3274 unsigned char version, bool recover_curseg, 3275 bool recover_newaddr); 3276 void f2fs_allocate_data_block(struct f2fs_sb_info *sbi, struct page *page, 3277 block_t old_blkaddr, block_t *new_blkaddr, 3278 struct f2fs_summary *sum, int type, 3279 struct f2fs_io_info *fio, bool add_list); 3280 void f2fs_wait_on_page_writeback(struct page *page, 3281 enum page_type type, bool ordered, bool locked); 3282 void f2fs_wait_on_block_writeback(struct inode *inode, block_t blkaddr); 3283 void f2fs_wait_on_block_writeback_range(struct inode *inode, block_t blkaddr, 3284 block_t len); 3285 void f2fs_write_data_summaries(struct f2fs_sb_info *sbi, block_t start_blk); 3286 void f2fs_write_node_summaries(struct f2fs_sb_info *sbi, block_t start_blk); 3287 int f2fs_lookup_journal_in_cursum(struct f2fs_journal *journal, int type, 3288 unsigned int val, int alloc); 3289 void f2fs_flush_sit_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3290 int f2fs_fix_curseg_write_pointer(struct f2fs_sb_info *sbi); 3291 int f2fs_check_write_pointer(struct f2fs_sb_info *sbi); 3292 int f2fs_build_segment_manager(struct f2fs_sb_info *sbi); 3293 void f2fs_destroy_segment_manager(struct f2fs_sb_info *sbi); 3294 int __init f2fs_create_segment_manager_caches(void); 3295 void f2fs_destroy_segment_manager_caches(void); 3296 int f2fs_rw_hint_to_seg_type(enum rw_hint hint); 3297 enum rw_hint f2fs_io_type_to_rw_hint(struct f2fs_sb_info *sbi, 3298 enum page_type type, enum temp_type temp); 3299 3300 /* 3301 * checkpoint.c 3302 */ 3303 void f2fs_stop_checkpoint(struct f2fs_sb_info *sbi, bool end_io); 3304 struct page *f2fs_grab_meta_page(struct f2fs_sb_info *sbi, pgoff_t index); 3305 struct page *f2fs_get_meta_page(struct f2fs_sb_info *sbi, pgoff_t index); 3306 struct page *f2fs_get_meta_page_nofail(struct f2fs_sb_info *sbi, pgoff_t index); 3307 struct page *f2fs_get_tmp_page(struct f2fs_sb_info *sbi, pgoff_t index); 3308 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi, 3309 block_t blkaddr, int type); 3310 int f2fs_ra_meta_pages(struct f2fs_sb_info *sbi, block_t start, int nrpages, 3311 int type, bool sync); 3312 void f2fs_ra_meta_pages_cond(struct f2fs_sb_info *sbi, pgoff_t index); 3313 long f2fs_sync_meta_pages(struct f2fs_sb_info *sbi, enum page_type type, 3314 long nr_to_write, enum iostat_type io_type); 3315 void f2fs_add_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type); 3316 void f2fs_remove_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type); 3317 void f2fs_release_ino_entry(struct f2fs_sb_info *sbi, bool all); 3318 bool f2fs_exist_written_data(struct f2fs_sb_info *sbi, nid_t ino, int mode); 3319 void f2fs_set_dirty_device(struct f2fs_sb_info *sbi, nid_t ino, 3320 unsigned int devidx, int type); 3321 bool f2fs_is_dirty_device(struct f2fs_sb_info *sbi, nid_t ino, 3322 unsigned int devidx, int type); 3323 int f2fs_sync_inode_meta(struct f2fs_sb_info *sbi); 3324 int f2fs_acquire_orphan_inode(struct f2fs_sb_info *sbi); 3325 void f2fs_release_orphan_inode(struct f2fs_sb_info *sbi); 3326 void f2fs_add_orphan_inode(struct inode *inode); 3327 void f2fs_remove_orphan_inode(struct f2fs_sb_info *sbi, nid_t ino); 3328 int f2fs_recover_orphan_inodes(struct f2fs_sb_info *sbi); 3329 int f2fs_get_valid_checkpoint(struct f2fs_sb_info *sbi); 3330 void f2fs_update_dirty_page(struct inode *inode, struct page *page); 3331 void f2fs_remove_dirty_inode(struct inode *inode); 3332 int f2fs_sync_dirty_inodes(struct f2fs_sb_info *sbi, enum inode_type type); 3333 void f2fs_wait_on_all_pages(struct f2fs_sb_info *sbi, int type); 3334 int f2fs_write_checkpoint(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3335 void f2fs_init_ino_entry_info(struct f2fs_sb_info *sbi); 3336 int __init f2fs_create_checkpoint_caches(void); 3337 void f2fs_destroy_checkpoint_caches(void); 3338 3339 /* 3340 * data.c 3341 */ 3342 int __init f2fs_init_bioset(void); 3343 void f2fs_destroy_bioset(void); 3344 struct bio *f2fs_bio_alloc(struct f2fs_sb_info *sbi, int npages, bool noio); 3345 int f2fs_init_bio_entry_cache(void); 3346 void f2fs_destroy_bio_entry_cache(void); 3347 void f2fs_submit_bio(struct f2fs_sb_info *sbi, 3348 struct bio *bio, enum page_type type); 3349 void f2fs_submit_merged_write(struct f2fs_sb_info *sbi, enum page_type type); 3350 void f2fs_submit_merged_write_cond(struct f2fs_sb_info *sbi, 3351 struct inode *inode, struct page *page, 3352 nid_t ino, enum page_type type); 3353 void f2fs_submit_merged_ipu_write(struct f2fs_sb_info *sbi, 3354 struct bio **bio, struct page *page); 3355 void f2fs_flush_merged_writes(struct f2fs_sb_info *sbi); 3356 int f2fs_submit_page_bio(struct f2fs_io_info *fio); 3357 int f2fs_merge_page_bio(struct f2fs_io_info *fio); 3358 void f2fs_submit_page_write(struct f2fs_io_info *fio); 3359 struct block_device *f2fs_target_device(struct f2fs_sb_info *sbi, 3360 block_t blk_addr, struct bio *bio); 3361 int f2fs_target_device_index(struct f2fs_sb_info *sbi, block_t blkaddr); 3362 void f2fs_set_data_blkaddr(struct dnode_of_data *dn); 3363 void f2fs_update_data_blkaddr(struct dnode_of_data *dn, block_t blkaddr); 3364 int f2fs_reserve_new_blocks(struct dnode_of_data *dn, blkcnt_t count); 3365 int f2fs_reserve_new_block(struct dnode_of_data *dn); 3366 int f2fs_get_block(struct dnode_of_data *dn, pgoff_t index); 3367 int f2fs_preallocate_blocks(struct kiocb *iocb, struct iov_iter *from); 3368 int f2fs_reserve_block(struct dnode_of_data *dn, pgoff_t index); 3369 struct page *f2fs_get_read_data_page(struct inode *inode, pgoff_t index, 3370 int op_flags, bool for_write); 3371 struct page *f2fs_find_data_page(struct inode *inode, pgoff_t index); 3372 struct page *f2fs_get_lock_data_page(struct inode *inode, pgoff_t index, 3373 bool for_write); 3374 struct page *f2fs_get_new_data_page(struct inode *inode, 3375 struct page *ipage, pgoff_t index, bool new_i_size); 3376 int f2fs_do_write_data_page(struct f2fs_io_info *fio); 3377 void __do_map_lock(struct f2fs_sb_info *sbi, int flag, bool lock); 3378 int f2fs_map_blocks(struct inode *inode, struct f2fs_map_blocks *map, 3379 int create, int flag); 3380 int f2fs_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, 3381 u64 start, u64 len); 3382 int f2fs_encrypt_one_page(struct f2fs_io_info *fio); 3383 bool f2fs_should_update_inplace(struct inode *inode, struct f2fs_io_info *fio); 3384 bool f2fs_should_update_outplace(struct inode *inode, struct f2fs_io_info *fio); 3385 int f2fs_write_single_data_page(struct page *page, int *submitted, 3386 struct bio **bio, sector_t *last_block, 3387 struct writeback_control *wbc, 3388 enum iostat_type io_type, 3389 int compr_blocks); 3390 void f2fs_invalidate_page(struct page *page, unsigned int offset, 3391 unsigned int length); 3392 int f2fs_release_page(struct page *page, gfp_t wait); 3393 #ifdef CONFIG_MIGRATION 3394 int f2fs_migrate_page(struct address_space *mapping, struct page *newpage, 3395 struct page *page, enum migrate_mode mode); 3396 #endif 3397 bool f2fs_overwrite_io(struct inode *inode, loff_t pos, size_t len); 3398 void f2fs_clear_page_cache_dirty_tag(struct page *page); 3399 int f2fs_init_post_read_processing(void); 3400 void f2fs_destroy_post_read_processing(void); 3401 int f2fs_init_post_read_wq(struct f2fs_sb_info *sbi); 3402 void f2fs_destroy_post_read_wq(struct f2fs_sb_info *sbi); 3403 3404 /* 3405 * gc.c 3406 */ 3407 int f2fs_start_gc_thread(struct f2fs_sb_info *sbi); 3408 void f2fs_stop_gc_thread(struct f2fs_sb_info *sbi); 3409 block_t f2fs_start_bidx_of_node(unsigned int node_ofs, struct inode *inode); 3410 int f2fs_gc(struct f2fs_sb_info *sbi, bool sync, bool background, 3411 unsigned int segno); 3412 void f2fs_build_gc_manager(struct f2fs_sb_info *sbi); 3413 int f2fs_resize_fs(struct f2fs_sb_info *sbi, __u64 block_count); 3414 3415 /* 3416 * recovery.c 3417 */ 3418 int f2fs_recover_fsync_data(struct f2fs_sb_info *sbi, bool check_only); 3419 bool f2fs_space_for_roll_forward(struct f2fs_sb_info *sbi); 3420 3421 /* 3422 * debug.c 3423 */ 3424 #ifdef CONFIG_F2FS_STAT_FS 3425 struct f2fs_stat_info { 3426 struct list_head stat_list; 3427 struct f2fs_sb_info *sbi; 3428 int all_area_segs, sit_area_segs, nat_area_segs, ssa_area_segs; 3429 int main_area_segs, main_area_sections, main_area_zones; 3430 unsigned long long hit_largest, hit_cached, hit_rbtree; 3431 unsigned long long hit_total, total_ext; 3432 int ext_tree, zombie_tree, ext_node; 3433 int ndirty_node, ndirty_dent, ndirty_meta, ndirty_imeta; 3434 int ndirty_data, ndirty_qdata; 3435 int inmem_pages; 3436 unsigned int ndirty_dirs, ndirty_files, nquota_files, ndirty_all; 3437 int nats, dirty_nats, sits, dirty_sits; 3438 int free_nids, avail_nids, alloc_nids; 3439 int total_count, utilization; 3440 int bg_gc, nr_wb_cp_data, nr_wb_data; 3441 int nr_rd_data, nr_rd_node, nr_rd_meta; 3442 int nr_dio_read, nr_dio_write; 3443 unsigned int io_skip_bggc, other_skip_bggc; 3444 int nr_flushing, nr_flushed, flush_list_empty; 3445 int nr_discarding, nr_discarded; 3446 int nr_discard_cmd; 3447 unsigned int undiscard_blks; 3448 int inline_xattr, inline_inode, inline_dir, append, update, orphans; 3449 int compr_inode, compr_blocks; 3450 int aw_cnt, max_aw_cnt, vw_cnt, max_vw_cnt; 3451 unsigned int valid_count, valid_node_count, valid_inode_count, discard_blks; 3452 unsigned int bimodal, avg_vblocks; 3453 int util_free, util_valid, util_invalid; 3454 int rsvd_segs, overp_segs; 3455 int dirty_count, node_pages, meta_pages; 3456 int prefree_count, call_count, cp_count, bg_cp_count; 3457 int tot_segs, node_segs, data_segs, free_segs, free_secs; 3458 int bg_node_segs, bg_data_segs; 3459 int tot_blks, data_blks, node_blks; 3460 int bg_data_blks, bg_node_blks; 3461 unsigned long long skipped_atomic_files[2]; 3462 int curseg[NR_CURSEG_TYPE]; 3463 int cursec[NR_CURSEG_TYPE]; 3464 int curzone[NR_CURSEG_TYPE]; 3465 3466 unsigned int meta_count[META_MAX]; 3467 unsigned int segment_count[2]; 3468 unsigned int block_count[2]; 3469 unsigned int inplace_count; 3470 unsigned long long base_mem, cache_mem, page_mem; 3471 }; 3472 3473 static inline struct f2fs_stat_info *F2FS_STAT(struct f2fs_sb_info *sbi) 3474 { 3475 return (struct f2fs_stat_info *)sbi->stat_info; 3476 } 3477 3478 #define stat_inc_cp_count(si) ((si)->cp_count++) 3479 #define stat_inc_bg_cp_count(si) ((si)->bg_cp_count++) 3480 #define stat_inc_call_count(si) ((si)->call_count++) 3481 #define stat_inc_bggc_count(si) ((si)->bg_gc++) 3482 #define stat_io_skip_bggc_count(sbi) ((sbi)->io_skip_bggc++) 3483 #define stat_other_skip_bggc_count(sbi) ((sbi)->other_skip_bggc++) 3484 #define stat_inc_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]++) 3485 #define stat_dec_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]--) 3486 #define stat_inc_total_hit(sbi) (atomic64_inc(&(sbi)->total_hit_ext)) 3487 #define stat_inc_rbtree_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_rbtree)) 3488 #define stat_inc_largest_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_largest)) 3489 #define stat_inc_cached_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_cached)) 3490 #define stat_inc_inline_xattr(inode) \ 3491 do { \ 3492 if (f2fs_has_inline_xattr(inode)) \ 3493 (atomic_inc(&F2FS_I_SB(inode)->inline_xattr)); \ 3494 } while (0) 3495 #define stat_dec_inline_xattr(inode) \ 3496 do { \ 3497 if (f2fs_has_inline_xattr(inode)) \ 3498 (atomic_dec(&F2FS_I_SB(inode)->inline_xattr)); \ 3499 } while (0) 3500 #define stat_inc_inline_inode(inode) \ 3501 do { \ 3502 if (f2fs_has_inline_data(inode)) \ 3503 (atomic_inc(&F2FS_I_SB(inode)->inline_inode)); \ 3504 } while (0) 3505 #define stat_dec_inline_inode(inode) \ 3506 do { \ 3507 if (f2fs_has_inline_data(inode)) \ 3508 (atomic_dec(&F2FS_I_SB(inode)->inline_inode)); \ 3509 } while (0) 3510 #define stat_inc_inline_dir(inode) \ 3511 do { \ 3512 if (f2fs_has_inline_dentry(inode)) \ 3513 (atomic_inc(&F2FS_I_SB(inode)->inline_dir)); \ 3514 } while (0) 3515 #define stat_dec_inline_dir(inode) \ 3516 do { \ 3517 if (f2fs_has_inline_dentry(inode)) \ 3518 (atomic_dec(&F2FS_I_SB(inode)->inline_dir)); \ 3519 } while (0) 3520 #define stat_inc_compr_inode(inode) \ 3521 do { \ 3522 if (f2fs_compressed_file(inode)) \ 3523 (atomic_inc(&F2FS_I_SB(inode)->compr_inode)); \ 3524 } while (0) 3525 #define stat_dec_compr_inode(inode) \ 3526 do { \ 3527 if (f2fs_compressed_file(inode)) \ 3528 (atomic_dec(&F2FS_I_SB(inode)->compr_inode)); \ 3529 } while (0) 3530 #define stat_add_compr_blocks(inode, blocks) \ 3531 (atomic_add(blocks, &F2FS_I_SB(inode)->compr_blocks)) 3532 #define stat_sub_compr_blocks(inode, blocks) \ 3533 (atomic_sub(blocks, &F2FS_I_SB(inode)->compr_blocks)) 3534 #define stat_inc_meta_count(sbi, blkaddr) \ 3535 do { \ 3536 if (blkaddr < SIT_I(sbi)->sit_base_addr) \ 3537 atomic_inc(&(sbi)->meta_count[META_CP]); \ 3538 else if (blkaddr < NM_I(sbi)->nat_blkaddr) \ 3539 atomic_inc(&(sbi)->meta_count[META_SIT]); \ 3540 else if (blkaddr < SM_I(sbi)->ssa_blkaddr) \ 3541 atomic_inc(&(sbi)->meta_count[META_NAT]); \ 3542 else if (blkaddr < SM_I(sbi)->main_blkaddr) \ 3543 atomic_inc(&(sbi)->meta_count[META_SSA]); \ 3544 } while (0) 3545 #define stat_inc_seg_type(sbi, curseg) \ 3546 ((sbi)->segment_count[(curseg)->alloc_type]++) 3547 #define stat_inc_block_count(sbi, curseg) \ 3548 ((sbi)->block_count[(curseg)->alloc_type]++) 3549 #define stat_inc_inplace_blocks(sbi) \ 3550 (atomic_inc(&(sbi)->inplace_count)) 3551 #define stat_update_max_atomic_write(inode) \ 3552 do { \ 3553 int cur = F2FS_I_SB(inode)->atomic_files; \ 3554 int max = atomic_read(&F2FS_I_SB(inode)->max_aw_cnt); \ 3555 if (cur > max) \ 3556 atomic_set(&F2FS_I_SB(inode)->max_aw_cnt, cur); \ 3557 } while (0) 3558 #define stat_inc_volatile_write(inode) \ 3559 (atomic_inc(&F2FS_I_SB(inode)->vw_cnt)) 3560 #define stat_dec_volatile_write(inode) \ 3561 (atomic_dec(&F2FS_I_SB(inode)->vw_cnt)) 3562 #define stat_update_max_volatile_write(inode) \ 3563 do { \ 3564 int cur = atomic_read(&F2FS_I_SB(inode)->vw_cnt); \ 3565 int max = atomic_read(&F2FS_I_SB(inode)->max_vw_cnt); \ 3566 if (cur > max) \ 3567 atomic_set(&F2FS_I_SB(inode)->max_vw_cnt, cur); \ 3568 } while (0) 3569 #define stat_inc_seg_count(sbi, type, gc_type) \ 3570 do { \ 3571 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 3572 si->tot_segs++; \ 3573 if ((type) == SUM_TYPE_DATA) { \ 3574 si->data_segs++; \ 3575 si->bg_data_segs += (gc_type == BG_GC) ? 1 : 0; \ 3576 } else { \ 3577 si->node_segs++; \ 3578 si->bg_node_segs += (gc_type == BG_GC) ? 1 : 0; \ 3579 } \ 3580 } while (0) 3581 3582 #define stat_inc_tot_blk_count(si, blks) \ 3583 ((si)->tot_blks += (blks)) 3584 3585 #define stat_inc_data_blk_count(sbi, blks, gc_type) \ 3586 do { \ 3587 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 3588 stat_inc_tot_blk_count(si, blks); \ 3589 si->data_blks += (blks); \ 3590 si->bg_data_blks += ((gc_type) == BG_GC) ? (blks) : 0; \ 3591 } while (0) 3592 3593 #define stat_inc_node_blk_count(sbi, blks, gc_type) \ 3594 do { \ 3595 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 3596 stat_inc_tot_blk_count(si, blks); \ 3597 si->node_blks += (blks); \ 3598 si->bg_node_blks += ((gc_type) == BG_GC) ? (blks) : 0; \ 3599 } while (0) 3600 3601 int f2fs_build_stats(struct f2fs_sb_info *sbi); 3602 void f2fs_destroy_stats(struct f2fs_sb_info *sbi); 3603 void __init f2fs_create_root_stats(void); 3604 void f2fs_destroy_root_stats(void); 3605 void f2fs_update_sit_info(struct f2fs_sb_info *sbi); 3606 #else 3607 #define stat_inc_cp_count(si) do { } while (0) 3608 #define stat_inc_bg_cp_count(si) do { } while (0) 3609 #define stat_inc_call_count(si) do { } while (0) 3610 #define stat_inc_bggc_count(si) do { } while (0) 3611 #define stat_io_skip_bggc_count(sbi) do { } while (0) 3612 #define stat_other_skip_bggc_count(sbi) do { } while (0) 3613 #define stat_inc_dirty_inode(sbi, type) do { } while (0) 3614 #define stat_dec_dirty_inode(sbi, type) do { } while (0) 3615 #define stat_inc_total_hit(sbi) do { } while (0) 3616 #define stat_inc_rbtree_node_hit(sbi) do { } while (0) 3617 #define stat_inc_largest_node_hit(sbi) do { } while (0) 3618 #define stat_inc_cached_node_hit(sbi) do { } while (0) 3619 #define stat_inc_inline_xattr(inode) do { } while (0) 3620 #define stat_dec_inline_xattr(inode) do { } while (0) 3621 #define stat_inc_inline_inode(inode) do { } while (0) 3622 #define stat_dec_inline_inode(inode) do { } while (0) 3623 #define stat_inc_inline_dir(inode) do { } while (0) 3624 #define stat_dec_inline_dir(inode) do { } while (0) 3625 #define stat_inc_compr_inode(inode) do { } while (0) 3626 #define stat_dec_compr_inode(inode) do { } while (0) 3627 #define stat_add_compr_blocks(inode, blocks) do { } while (0) 3628 #define stat_sub_compr_blocks(inode, blocks) do { } while (0) 3629 #define stat_inc_atomic_write(inode) do { } while (0) 3630 #define stat_dec_atomic_write(inode) do { } while (0) 3631 #define stat_update_max_atomic_write(inode) do { } while (0) 3632 #define stat_inc_volatile_write(inode) do { } while (0) 3633 #define stat_dec_volatile_write(inode) do { } while (0) 3634 #define stat_update_max_volatile_write(inode) do { } while (0) 3635 #define stat_inc_meta_count(sbi, blkaddr) do { } while (0) 3636 #define stat_inc_seg_type(sbi, curseg) do { } while (0) 3637 #define stat_inc_block_count(sbi, curseg) do { } while (0) 3638 #define stat_inc_inplace_blocks(sbi) do { } while (0) 3639 #define stat_inc_seg_count(sbi, type, gc_type) do { } while (0) 3640 #define stat_inc_tot_blk_count(si, blks) do { } while (0) 3641 #define stat_inc_data_blk_count(sbi, blks, gc_type) do { } while (0) 3642 #define stat_inc_node_blk_count(sbi, blks, gc_type) do { } while (0) 3643 3644 static inline int f2fs_build_stats(struct f2fs_sb_info *sbi) { return 0; } 3645 static inline void f2fs_destroy_stats(struct f2fs_sb_info *sbi) { } 3646 static inline void __init f2fs_create_root_stats(void) { } 3647 static inline void f2fs_destroy_root_stats(void) { } 3648 static inline void update_sit_info(struct f2fs_sb_info *sbi) {} 3649 #endif 3650 3651 extern const struct file_operations f2fs_dir_operations; 3652 #ifdef CONFIG_UNICODE 3653 extern const struct dentry_operations f2fs_dentry_ops; 3654 #endif 3655 extern const struct file_operations f2fs_file_operations; 3656 extern const struct inode_operations f2fs_file_inode_operations; 3657 extern const struct address_space_operations f2fs_dblock_aops; 3658 extern const struct address_space_operations f2fs_node_aops; 3659 extern const struct address_space_operations f2fs_meta_aops; 3660 extern const struct inode_operations f2fs_dir_inode_operations; 3661 extern const struct inode_operations f2fs_symlink_inode_operations; 3662 extern const struct inode_operations f2fs_encrypted_symlink_inode_operations; 3663 extern const struct inode_operations f2fs_special_inode_operations; 3664 extern struct kmem_cache *f2fs_inode_entry_slab; 3665 3666 /* 3667 * inline.c 3668 */ 3669 bool f2fs_may_inline_data(struct inode *inode); 3670 bool f2fs_may_inline_dentry(struct inode *inode); 3671 void f2fs_do_read_inline_data(struct page *page, struct page *ipage); 3672 void f2fs_truncate_inline_inode(struct inode *inode, 3673 struct page *ipage, u64 from); 3674 int f2fs_read_inline_data(struct inode *inode, struct page *page); 3675 int f2fs_convert_inline_page(struct dnode_of_data *dn, struct page *page); 3676 int f2fs_convert_inline_inode(struct inode *inode); 3677 int f2fs_try_convert_inline_dir(struct inode *dir, struct dentry *dentry); 3678 int f2fs_write_inline_data(struct inode *inode, struct page *page); 3679 bool f2fs_recover_inline_data(struct inode *inode, struct page *npage); 3680 struct f2fs_dir_entry *f2fs_find_in_inline_dir(struct inode *dir, 3681 struct fscrypt_name *fname, struct page **res_page); 3682 int f2fs_make_empty_inline_dir(struct inode *inode, struct inode *parent, 3683 struct page *ipage); 3684 int f2fs_add_inline_entry(struct inode *dir, const struct qstr *new_name, 3685 const struct qstr *orig_name, 3686 struct inode *inode, nid_t ino, umode_t mode); 3687 void f2fs_delete_inline_entry(struct f2fs_dir_entry *dentry, 3688 struct page *page, struct inode *dir, 3689 struct inode *inode); 3690 bool f2fs_empty_inline_dir(struct inode *dir); 3691 int f2fs_read_inline_dir(struct file *file, struct dir_context *ctx, 3692 struct fscrypt_str *fstr); 3693 int f2fs_inline_data_fiemap(struct inode *inode, 3694 struct fiemap_extent_info *fieinfo, 3695 __u64 start, __u64 len); 3696 3697 /* 3698 * shrinker.c 3699 */ 3700 unsigned long f2fs_shrink_count(struct shrinker *shrink, 3701 struct shrink_control *sc); 3702 unsigned long f2fs_shrink_scan(struct shrinker *shrink, 3703 struct shrink_control *sc); 3704 void f2fs_join_shrinker(struct f2fs_sb_info *sbi); 3705 void f2fs_leave_shrinker(struct f2fs_sb_info *sbi); 3706 3707 /* 3708 * extent_cache.c 3709 */ 3710 struct rb_entry *f2fs_lookup_rb_tree(struct rb_root_cached *root, 3711 struct rb_entry *cached_re, unsigned int ofs); 3712 struct rb_node **f2fs_lookup_rb_tree_for_insert(struct f2fs_sb_info *sbi, 3713 struct rb_root_cached *root, 3714 struct rb_node **parent, 3715 unsigned int ofs, bool *leftmost); 3716 struct rb_entry *f2fs_lookup_rb_tree_ret(struct rb_root_cached *root, 3717 struct rb_entry *cached_re, unsigned int ofs, 3718 struct rb_entry **prev_entry, struct rb_entry **next_entry, 3719 struct rb_node ***insert_p, struct rb_node **insert_parent, 3720 bool force, bool *leftmost); 3721 bool f2fs_check_rb_tree_consistence(struct f2fs_sb_info *sbi, 3722 struct rb_root_cached *root); 3723 unsigned int f2fs_shrink_extent_tree(struct f2fs_sb_info *sbi, int nr_shrink); 3724 bool f2fs_init_extent_tree(struct inode *inode, struct f2fs_extent *i_ext); 3725 void f2fs_drop_extent_tree(struct inode *inode); 3726 unsigned int f2fs_destroy_extent_node(struct inode *inode); 3727 void f2fs_destroy_extent_tree(struct inode *inode); 3728 bool f2fs_lookup_extent_cache(struct inode *inode, pgoff_t pgofs, 3729 struct extent_info *ei); 3730 void f2fs_update_extent_cache(struct dnode_of_data *dn); 3731 void f2fs_update_extent_cache_range(struct dnode_of_data *dn, 3732 pgoff_t fofs, block_t blkaddr, unsigned int len); 3733 void f2fs_init_extent_cache_info(struct f2fs_sb_info *sbi); 3734 int __init f2fs_create_extent_cache(void); 3735 void f2fs_destroy_extent_cache(void); 3736 3737 /* 3738 * sysfs.c 3739 */ 3740 int __init f2fs_init_sysfs(void); 3741 void f2fs_exit_sysfs(void); 3742 int f2fs_register_sysfs(struct f2fs_sb_info *sbi); 3743 void f2fs_unregister_sysfs(struct f2fs_sb_info *sbi); 3744 3745 /* verity.c */ 3746 extern const struct fsverity_operations f2fs_verityops; 3747 3748 /* 3749 * crypto support 3750 */ 3751 static inline bool f2fs_encrypted_file(struct inode *inode) 3752 { 3753 return IS_ENCRYPTED(inode) && S_ISREG(inode->i_mode); 3754 } 3755 3756 static inline void f2fs_set_encrypted_inode(struct inode *inode) 3757 { 3758 #ifdef CONFIG_FS_ENCRYPTION 3759 file_set_encrypt(inode); 3760 f2fs_set_inode_flags(inode); 3761 #endif 3762 } 3763 3764 /* 3765 * Returns true if the reads of the inode's data need to undergo some 3766 * postprocessing step, like decryption or authenticity verification. 3767 */ 3768 static inline bool f2fs_post_read_required(struct inode *inode) 3769 { 3770 return f2fs_encrypted_file(inode) || fsverity_active(inode) || 3771 f2fs_compressed_file(inode); 3772 } 3773 3774 /* 3775 * compress.c 3776 */ 3777 #ifdef CONFIG_F2FS_FS_COMPRESSION 3778 bool f2fs_is_compressed_page(struct page *page); 3779 struct page *f2fs_compress_control_page(struct page *page); 3780 int f2fs_prepare_compress_overwrite(struct inode *inode, 3781 struct page **pagep, pgoff_t index, void **fsdata); 3782 bool f2fs_compress_write_end(struct inode *inode, void *fsdata, 3783 pgoff_t index, unsigned copied); 3784 void f2fs_compress_write_end_io(struct bio *bio, struct page *page); 3785 bool f2fs_is_compress_backend_ready(struct inode *inode); 3786 void f2fs_decompress_pages(struct bio *bio, struct page *page, bool verity); 3787 bool f2fs_cluster_is_empty(struct compress_ctx *cc); 3788 bool f2fs_cluster_can_merge_page(struct compress_ctx *cc, pgoff_t index); 3789 void f2fs_compress_ctx_add_page(struct compress_ctx *cc, struct page *page); 3790 int f2fs_write_multi_pages(struct compress_ctx *cc, 3791 int *submitted, 3792 struct writeback_control *wbc, 3793 enum iostat_type io_type); 3794 int f2fs_is_compressed_cluster(struct inode *inode, pgoff_t index); 3795 int f2fs_read_multi_pages(struct compress_ctx *cc, struct bio **bio_ret, 3796 unsigned nr_pages, sector_t *last_block_in_bio, 3797 bool is_readahead, bool for_write); 3798 struct decompress_io_ctx *f2fs_alloc_dic(struct compress_ctx *cc); 3799 void f2fs_free_dic(struct decompress_io_ctx *dic); 3800 void f2fs_decompress_end_io(struct page **rpages, 3801 unsigned int cluster_size, bool err, bool verity); 3802 int f2fs_init_compress_ctx(struct compress_ctx *cc); 3803 void f2fs_destroy_compress_ctx(struct compress_ctx *cc); 3804 void f2fs_init_compress_info(struct f2fs_sb_info *sbi); 3805 #else 3806 static inline bool f2fs_is_compressed_page(struct page *page) { return false; } 3807 static inline bool f2fs_is_compress_backend_ready(struct inode *inode) 3808 { 3809 if (!f2fs_compressed_file(inode)) 3810 return true; 3811 /* not support compression */ 3812 return false; 3813 } 3814 static inline struct page *f2fs_compress_control_page(struct page *page) 3815 { 3816 WARN_ON_ONCE(1); 3817 return ERR_PTR(-EINVAL); 3818 } 3819 #endif 3820 3821 static inline void set_compress_context(struct inode *inode) 3822 { 3823 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 3824 3825 F2FS_I(inode)->i_compress_algorithm = 3826 F2FS_OPTION(sbi).compress_algorithm; 3827 F2FS_I(inode)->i_log_cluster_size = 3828 F2FS_OPTION(sbi).compress_log_size; 3829 F2FS_I(inode)->i_cluster_size = 3830 1 << F2FS_I(inode)->i_log_cluster_size; 3831 F2FS_I(inode)->i_flags |= F2FS_COMPR_FL; 3832 set_inode_flag(inode, FI_COMPRESSED_FILE); 3833 stat_inc_compr_inode(inode); 3834 f2fs_mark_inode_dirty_sync(inode, true); 3835 } 3836 3837 static inline u64 f2fs_disable_compressed_file(struct inode *inode) 3838 { 3839 struct f2fs_inode_info *fi = F2FS_I(inode); 3840 3841 if (!f2fs_compressed_file(inode)) 3842 return 0; 3843 if (S_ISREG(inode->i_mode)) { 3844 if (get_dirty_pages(inode)) 3845 return 1; 3846 if (fi->i_compr_blocks) 3847 return fi->i_compr_blocks; 3848 } 3849 3850 fi->i_flags &= ~F2FS_COMPR_FL; 3851 stat_dec_compr_inode(inode); 3852 clear_inode_flag(inode, FI_COMPRESSED_FILE); 3853 f2fs_mark_inode_dirty_sync(inode, true); 3854 return 0; 3855 } 3856 3857 #define F2FS_FEATURE_FUNCS(name, flagname) \ 3858 static inline int f2fs_sb_has_##name(struct f2fs_sb_info *sbi) \ 3859 { \ 3860 return F2FS_HAS_FEATURE(sbi, F2FS_FEATURE_##flagname); \ 3861 } 3862 3863 F2FS_FEATURE_FUNCS(encrypt, ENCRYPT); 3864 F2FS_FEATURE_FUNCS(blkzoned, BLKZONED); 3865 F2FS_FEATURE_FUNCS(extra_attr, EXTRA_ATTR); 3866 F2FS_FEATURE_FUNCS(project_quota, PRJQUOTA); 3867 F2FS_FEATURE_FUNCS(inode_chksum, INODE_CHKSUM); 3868 F2FS_FEATURE_FUNCS(flexible_inline_xattr, FLEXIBLE_INLINE_XATTR); 3869 F2FS_FEATURE_FUNCS(quota_ino, QUOTA_INO); 3870 F2FS_FEATURE_FUNCS(inode_crtime, INODE_CRTIME); 3871 F2FS_FEATURE_FUNCS(lost_found, LOST_FOUND); 3872 F2FS_FEATURE_FUNCS(verity, VERITY); 3873 F2FS_FEATURE_FUNCS(sb_chksum, SB_CHKSUM); 3874 F2FS_FEATURE_FUNCS(casefold, CASEFOLD); 3875 F2FS_FEATURE_FUNCS(compression, COMPRESSION); 3876 3877 #ifdef CONFIG_BLK_DEV_ZONED 3878 static inline bool f2fs_blkz_is_seq(struct f2fs_sb_info *sbi, int devi, 3879 block_t blkaddr) 3880 { 3881 unsigned int zno = blkaddr >> sbi->log_blocks_per_blkz; 3882 3883 return test_bit(zno, FDEV(devi).blkz_seq); 3884 } 3885 #endif 3886 3887 static inline bool f2fs_hw_should_discard(struct f2fs_sb_info *sbi) 3888 { 3889 return f2fs_sb_has_blkzoned(sbi); 3890 } 3891 3892 static inline bool f2fs_bdev_support_discard(struct block_device *bdev) 3893 { 3894 return blk_queue_discard(bdev_get_queue(bdev)) || 3895 bdev_is_zoned(bdev); 3896 } 3897 3898 static inline bool f2fs_hw_support_discard(struct f2fs_sb_info *sbi) 3899 { 3900 int i; 3901 3902 if (!f2fs_is_multi_device(sbi)) 3903 return f2fs_bdev_support_discard(sbi->sb->s_bdev); 3904 3905 for (i = 0; i < sbi->s_ndevs; i++) 3906 if (f2fs_bdev_support_discard(FDEV(i).bdev)) 3907 return true; 3908 return false; 3909 } 3910 3911 static inline bool f2fs_realtime_discard_enable(struct f2fs_sb_info *sbi) 3912 { 3913 return (test_opt(sbi, DISCARD) && f2fs_hw_support_discard(sbi)) || 3914 f2fs_hw_should_discard(sbi); 3915 } 3916 3917 static inline bool f2fs_hw_is_readonly(struct f2fs_sb_info *sbi) 3918 { 3919 int i; 3920 3921 if (!f2fs_is_multi_device(sbi)) 3922 return bdev_read_only(sbi->sb->s_bdev); 3923 3924 for (i = 0; i < sbi->s_ndevs; i++) 3925 if (bdev_read_only(FDEV(i).bdev)) 3926 return true; 3927 return false; 3928 } 3929 3930 static inline bool f2fs_lfs_mode(struct f2fs_sb_info *sbi) 3931 { 3932 return F2FS_OPTION(sbi).fs_mode == FS_MODE_LFS; 3933 } 3934 3935 static inline bool f2fs_may_encrypt(struct inode *dir, struct inode *inode) 3936 { 3937 #ifdef CONFIG_FS_ENCRYPTION 3938 struct f2fs_sb_info *sbi = F2FS_I_SB(dir); 3939 umode_t mode = inode->i_mode; 3940 3941 /* 3942 * If the directory encrypted or dummy encryption enabled, 3943 * then we should encrypt the inode. 3944 */ 3945 if (IS_ENCRYPTED(dir) || DUMMY_ENCRYPTION_ENABLED(sbi)) 3946 return (S_ISREG(mode) || S_ISDIR(mode) || S_ISLNK(mode)); 3947 #endif 3948 return false; 3949 } 3950 3951 static inline bool f2fs_may_compress(struct inode *inode) 3952 { 3953 if (IS_SWAPFILE(inode) || f2fs_is_pinned_file(inode) || 3954 f2fs_is_atomic_file(inode) || 3955 f2fs_is_volatile_file(inode)) 3956 return false; 3957 return S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode); 3958 } 3959 3960 static inline void f2fs_i_compr_blocks_update(struct inode *inode, 3961 u64 blocks, bool add) 3962 { 3963 int diff = F2FS_I(inode)->i_cluster_size - blocks; 3964 3965 if (add) { 3966 F2FS_I(inode)->i_compr_blocks += diff; 3967 stat_add_compr_blocks(inode, diff); 3968 } else { 3969 F2FS_I(inode)->i_compr_blocks -= diff; 3970 stat_sub_compr_blocks(inode, diff); 3971 } 3972 f2fs_mark_inode_dirty_sync(inode, true); 3973 } 3974 3975 static inline int block_unaligned_IO(struct inode *inode, 3976 struct kiocb *iocb, struct iov_iter *iter) 3977 { 3978 unsigned int i_blkbits = READ_ONCE(inode->i_blkbits); 3979 unsigned int blocksize_mask = (1 << i_blkbits) - 1; 3980 loff_t offset = iocb->ki_pos; 3981 unsigned long align = offset | iov_iter_alignment(iter); 3982 3983 return align & blocksize_mask; 3984 } 3985 3986 static inline int allow_outplace_dio(struct inode *inode, 3987 struct kiocb *iocb, struct iov_iter *iter) 3988 { 3989 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 3990 int rw = iov_iter_rw(iter); 3991 3992 return (f2fs_lfs_mode(sbi) && (rw == WRITE) && 3993 !block_unaligned_IO(inode, iocb, iter)); 3994 } 3995 3996 static inline bool f2fs_force_buffered_io(struct inode *inode, 3997 struct kiocb *iocb, struct iov_iter *iter) 3998 { 3999 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 4000 int rw = iov_iter_rw(iter); 4001 4002 if (f2fs_post_read_required(inode)) 4003 return true; 4004 if (f2fs_is_multi_device(sbi)) 4005 return true; 4006 if (f2fs_compressed_file(inode)) 4007 return true; 4008 /* 4009 * for blkzoned device, fallback direct IO to buffered IO, so 4010 * all IOs can be serialized by log-structured write. 4011 */ 4012 if (f2fs_sb_has_blkzoned(sbi)) 4013 return true; 4014 if (f2fs_lfs_mode(sbi) && (rw == WRITE)) { 4015 if (block_unaligned_IO(inode, iocb, iter)) 4016 return true; 4017 if (F2FS_IO_ALIGNED(sbi)) 4018 return true; 4019 } 4020 if (is_sbi_flag_set(F2FS_I_SB(inode), SBI_CP_DISABLED) && 4021 !IS_SWAPFILE(inode)) 4022 return true; 4023 4024 return false; 4025 } 4026 4027 #ifdef CONFIG_F2FS_FAULT_INJECTION 4028 extern void f2fs_build_fault_attr(struct f2fs_sb_info *sbi, unsigned int rate, 4029 unsigned int type); 4030 #else 4031 #define f2fs_build_fault_attr(sbi, rate, type) do { } while (0) 4032 #endif 4033 4034 static inline bool is_journalled_quota(struct f2fs_sb_info *sbi) 4035 { 4036 #ifdef CONFIG_QUOTA 4037 if (f2fs_sb_has_quota_ino(sbi)) 4038 return true; 4039 if (F2FS_OPTION(sbi).s_qf_names[USRQUOTA] || 4040 F2FS_OPTION(sbi).s_qf_names[GRPQUOTA] || 4041 F2FS_OPTION(sbi).s_qf_names[PRJQUOTA]) 4042 return true; 4043 #endif 4044 return false; 4045 } 4046 4047 #define EFSBADCRC EBADMSG /* Bad CRC detected */ 4048 #define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */ 4049 4050 #endif /* _LINUX_F2FS_H */ 4051