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 (WARN_ON(condition)) \ 37 set_sbi_flag(sbi, SBI_NEED_FSCK); \ 38 } while (0) 39 #endif 40 41 enum { 42 FAULT_KMALLOC, 43 FAULT_KVMALLOC, 44 FAULT_PAGE_ALLOC, 45 FAULT_PAGE_GET, 46 FAULT_ALLOC_BIO, /* it's obsolete due to bio_alloc() will never fail */ 47 FAULT_ALLOC_NID, 48 FAULT_ORPHAN, 49 FAULT_BLOCK, 50 FAULT_DIR_DEPTH, 51 FAULT_EVICT_INODE, 52 FAULT_TRUNCATE, 53 FAULT_READ_IO, 54 FAULT_CHECKPOINT, 55 FAULT_DISCARD, 56 FAULT_WRITE_IO, 57 FAULT_SLAB_ALLOC, 58 FAULT_MAX, 59 }; 60 61 #ifdef CONFIG_F2FS_FAULT_INJECTION 62 #define F2FS_ALL_FAULT_TYPE ((1 << FAULT_MAX) - 1) 63 64 struct f2fs_fault_info { 65 atomic_t inject_ops; 66 unsigned int inject_rate; 67 unsigned int inject_type; 68 }; 69 70 extern const char *f2fs_fault_name[FAULT_MAX]; 71 #define IS_FAULT_SET(fi, type) ((fi)->inject_type & (1 << (type))) 72 #endif 73 74 /* 75 * For mount options 76 */ 77 #define F2FS_MOUNT_DISABLE_ROLL_FORWARD 0x00000002 78 #define F2FS_MOUNT_DISCARD 0x00000004 79 #define F2FS_MOUNT_NOHEAP 0x00000008 80 #define F2FS_MOUNT_XATTR_USER 0x00000010 81 #define F2FS_MOUNT_POSIX_ACL 0x00000020 82 #define F2FS_MOUNT_DISABLE_EXT_IDENTIFY 0x00000040 83 #define F2FS_MOUNT_INLINE_XATTR 0x00000080 84 #define F2FS_MOUNT_INLINE_DATA 0x00000100 85 #define F2FS_MOUNT_INLINE_DENTRY 0x00000200 86 #define F2FS_MOUNT_FLUSH_MERGE 0x00000400 87 #define F2FS_MOUNT_NOBARRIER 0x00000800 88 #define F2FS_MOUNT_FASTBOOT 0x00001000 89 #define F2FS_MOUNT_EXTENT_CACHE 0x00002000 90 #define F2FS_MOUNT_DATA_FLUSH 0x00008000 91 #define F2FS_MOUNT_FAULT_INJECTION 0x00010000 92 #define F2FS_MOUNT_USRQUOTA 0x00080000 93 #define F2FS_MOUNT_GRPQUOTA 0x00100000 94 #define F2FS_MOUNT_PRJQUOTA 0x00200000 95 #define F2FS_MOUNT_QUOTA 0x00400000 96 #define F2FS_MOUNT_INLINE_XATTR_SIZE 0x00800000 97 #define F2FS_MOUNT_RESERVE_ROOT 0x01000000 98 #define F2FS_MOUNT_DISABLE_CHECKPOINT 0x02000000 99 #define F2FS_MOUNT_NORECOVERY 0x04000000 100 #define F2FS_MOUNT_ATGC 0x08000000 101 #define F2FS_MOUNT_MERGE_CHECKPOINT 0x10000000 102 #define F2FS_MOUNT_GC_MERGE 0x20000000 103 #define F2FS_MOUNT_COMPRESS_CACHE 0x40000000 104 105 #define F2FS_OPTION(sbi) ((sbi)->mount_opt) 106 #define clear_opt(sbi, option) (F2FS_OPTION(sbi).opt &= ~F2FS_MOUNT_##option) 107 #define set_opt(sbi, option) (F2FS_OPTION(sbi).opt |= F2FS_MOUNT_##option) 108 #define test_opt(sbi, option) (F2FS_OPTION(sbi).opt & F2FS_MOUNT_##option) 109 110 #define ver_after(a, b) (typecheck(unsigned long long, a) && \ 111 typecheck(unsigned long long, b) && \ 112 ((long long)((a) - (b)) > 0)) 113 114 typedef u32 block_t; /* 115 * should not change u32, since it is the on-disk block 116 * address format, __le32. 117 */ 118 typedef u32 nid_t; 119 120 #define COMPRESS_EXT_NUM 16 121 122 struct f2fs_mount_info { 123 unsigned int opt; 124 int write_io_size_bits; /* Write IO size bits */ 125 block_t root_reserved_blocks; /* root reserved blocks */ 126 kuid_t s_resuid; /* reserved blocks for uid */ 127 kgid_t s_resgid; /* reserved blocks for gid */ 128 int active_logs; /* # of active logs */ 129 int inline_xattr_size; /* inline xattr size */ 130 #ifdef CONFIG_F2FS_FAULT_INJECTION 131 struct f2fs_fault_info fault_info; /* For fault injection */ 132 #endif 133 #ifdef CONFIG_QUOTA 134 /* Names of quota files with journalled quota */ 135 char *s_qf_names[MAXQUOTAS]; 136 int s_jquota_fmt; /* Format of quota to use */ 137 #endif 138 /* For which write hints are passed down to block layer */ 139 int whint_mode; 140 int alloc_mode; /* segment allocation policy */ 141 int fsync_mode; /* fsync policy */ 142 int fs_mode; /* fs mode: LFS or ADAPTIVE */ 143 int bggc_mode; /* bggc mode: off, on or sync */ 144 int discard_unit; /* 145 * discard command's offset/size should 146 * be aligned to this unit: block, 147 * segment or section 148 */ 149 struct fscrypt_dummy_policy dummy_enc_policy; /* test dummy encryption */ 150 block_t unusable_cap_perc; /* percentage for cap */ 151 block_t unusable_cap; /* Amount of space allowed to be 152 * unusable when disabling checkpoint 153 */ 154 155 /* For compression */ 156 unsigned char compress_algorithm; /* algorithm type */ 157 unsigned char compress_log_size; /* cluster log size */ 158 unsigned char compress_level; /* compress level */ 159 bool compress_chksum; /* compressed data chksum */ 160 unsigned char compress_ext_cnt; /* extension count */ 161 unsigned char nocompress_ext_cnt; /* nocompress extension count */ 162 int compress_mode; /* compression mode */ 163 unsigned char extensions[COMPRESS_EXT_NUM][F2FS_EXTENSION_LEN]; /* extensions */ 164 unsigned char noextensions[COMPRESS_EXT_NUM][F2FS_EXTENSION_LEN]; /* extensions */ 165 }; 166 167 #define F2FS_FEATURE_ENCRYPT 0x0001 168 #define F2FS_FEATURE_BLKZONED 0x0002 169 #define F2FS_FEATURE_ATOMIC_WRITE 0x0004 170 #define F2FS_FEATURE_EXTRA_ATTR 0x0008 171 #define F2FS_FEATURE_PRJQUOTA 0x0010 172 #define F2FS_FEATURE_INODE_CHKSUM 0x0020 173 #define F2FS_FEATURE_FLEXIBLE_INLINE_XATTR 0x0040 174 #define F2FS_FEATURE_QUOTA_INO 0x0080 175 #define F2FS_FEATURE_INODE_CRTIME 0x0100 176 #define F2FS_FEATURE_LOST_FOUND 0x0200 177 #define F2FS_FEATURE_VERITY 0x0400 178 #define F2FS_FEATURE_SB_CHKSUM 0x0800 179 #define F2FS_FEATURE_CASEFOLD 0x1000 180 #define F2FS_FEATURE_COMPRESSION 0x2000 181 #define F2FS_FEATURE_RO 0x4000 182 183 #define __F2FS_HAS_FEATURE(raw_super, mask) \ 184 ((raw_super->feature & cpu_to_le32(mask)) != 0) 185 #define F2FS_HAS_FEATURE(sbi, mask) __F2FS_HAS_FEATURE(sbi->raw_super, mask) 186 #define F2FS_SET_FEATURE(sbi, mask) \ 187 (sbi->raw_super->feature |= cpu_to_le32(mask)) 188 #define F2FS_CLEAR_FEATURE(sbi, mask) \ 189 (sbi->raw_super->feature &= ~cpu_to_le32(mask)) 190 191 /* 192 * Default values for user and/or group using reserved blocks 193 */ 194 #define F2FS_DEF_RESUID 0 195 #define F2FS_DEF_RESGID 0 196 197 /* 198 * For checkpoint manager 199 */ 200 enum { 201 NAT_BITMAP, 202 SIT_BITMAP 203 }; 204 205 #define CP_UMOUNT 0x00000001 206 #define CP_FASTBOOT 0x00000002 207 #define CP_SYNC 0x00000004 208 #define CP_RECOVERY 0x00000008 209 #define CP_DISCARD 0x00000010 210 #define CP_TRIMMED 0x00000020 211 #define CP_PAUSE 0x00000040 212 #define CP_RESIZE 0x00000080 213 214 #define MAX_DISCARD_BLOCKS(sbi) BLKS_PER_SEC(sbi) 215 #define DEF_MAX_DISCARD_REQUEST 8 /* issue 8 discards per round */ 216 #define DEF_MIN_DISCARD_ISSUE_TIME 50 /* 50 ms, if exists */ 217 #define DEF_MID_DISCARD_ISSUE_TIME 500 /* 500 ms, if device busy */ 218 #define DEF_MAX_DISCARD_ISSUE_TIME 60000 /* 60 s, if no candidates */ 219 #define DEF_DISCARD_URGENT_UTIL 80 /* do more discard over 80% */ 220 #define DEF_CP_INTERVAL 60 /* 60 secs */ 221 #define DEF_IDLE_INTERVAL 5 /* 5 secs */ 222 #define DEF_DISABLE_INTERVAL 5 /* 5 secs */ 223 #define DEF_DISABLE_QUICK_INTERVAL 1 /* 1 secs */ 224 #define DEF_UMOUNT_DISCARD_TIMEOUT 5 /* 5 secs */ 225 226 struct cp_control { 227 int reason; 228 __u64 trim_start; 229 __u64 trim_end; 230 __u64 trim_minlen; 231 }; 232 233 /* 234 * indicate meta/data type 235 */ 236 enum { 237 META_CP, 238 META_NAT, 239 META_SIT, 240 META_SSA, 241 META_MAX, 242 META_POR, 243 DATA_GENERIC, /* check range only */ 244 DATA_GENERIC_ENHANCE, /* strong check on range and segment bitmap */ 245 DATA_GENERIC_ENHANCE_READ, /* 246 * strong check on range and segment 247 * bitmap but no warning due to race 248 * condition of read on truncated area 249 * by extent_cache 250 */ 251 META_GENERIC, 252 }; 253 254 /* for the list of ino */ 255 enum { 256 ORPHAN_INO, /* for orphan ino list */ 257 APPEND_INO, /* for append ino list */ 258 UPDATE_INO, /* for update ino list */ 259 TRANS_DIR_INO, /* for trasactions dir ino list */ 260 FLUSH_INO, /* for multiple device flushing */ 261 MAX_INO_ENTRY, /* max. list */ 262 }; 263 264 struct ino_entry { 265 struct list_head list; /* list head */ 266 nid_t ino; /* inode number */ 267 unsigned int dirty_device; /* dirty device bitmap */ 268 }; 269 270 /* for the list of inodes to be GCed */ 271 struct inode_entry { 272 struct list_head list; /* list head */ 273 struct inode *inode; /* vfs inode pointer */ 274 }; 275 276 struct fsync_node_entry { 277 struct list_head list; /* list head */ 278 struct page *page; /* warm node page pointer */ 279 unsigned int seq_id; /* sequence id */ 280 }; 281 282 struct ckpt_req { 283 struct completion wait; /* completion for checkpoint done */ 284 struct llist_node llnode; /* llist_node to be linked in wait queue */ 285 int ret; /* return code of checkpoint */ 286 ktime_t queue_time; /* request queued time */ 287 }; 288 289 struct ckpt_req_control { 290 struct task_struct *f2fs_issue_ckpt; /* checkpoint task */ 291 int ckpt_thread_ioprio; /* checkpoint merge thread ioprio */ 292 wait_queue_head_t ckpt_wait_queue; /* waiting queue for wake-up */ 293 atomic_t issued_ckpt; /* # of actually issued ckpts */ 294 atomic_t total_ckpt; /* # of total ckpts */ 295 atomic_t queued_ckpt; /* # of queued ckpts */ 296 struct llist_head issue_list; /* list for command issue */ 297 spinlock_t stat_lock; /* lock for below checkpoint time stats */ 298 unsigned int cur_time; /* cur wait time in msec for currently issued checkpoint */ 299 unsigned int peak_time; /* peak wait time in msec until now */ 300 }; 301 302 /* for the bitmap indicate blocks to be discarded */ 303 struct discard_entry { 304 struct list_head list; /* list head */ 305 block_t start_blkaddr; /* start blockaddr of current segment */ 306 unsigned char discard_map[SIT_VBLOCK_MAP_SIZE]; /* segment discard bitmap */ 307 }; 308 309 /* default discard granularity of inner discard thread, unit: block count */ 310 #define DEFAULT_DISCARD_GRANULARITY 16 311 312 /* max discard pend list number */ 313 #define MAX_PLIST_NUM 512 314 #define plist_idx(blk_num) ((blk_num) >= MAX_PLIST_NUM ? \ 315 (MAX_PLIST_NUM - 1) : ((blk_num) - 1)) 316 317 enum { 318 D_PREP, /* initial */ 319 D_PARTIAL, /* partially submitted */ 320 D_SUBMIT, /* all submitted */ 321 D_DONE, /* finished */ 322 }; 323 324 struct discard_info { 325 block_t lstart; /* logical start address */ 326 block_t len; /* length */ 327 block_t start; /* actual start address in dev */ 328 }; 329 330 struct discard_cmd { 331 struct rb_node rb_node; /* rb node located in rb-tree */ 332 union { 333 struct { 334 block_t lstart; /* logical start address */ 335 block_t len; /* length */ 336 block_t start; /* actual start address in dev */ 337 }; 338 struct discard_info di; /* discard info */ 339 340 }; 341 struct list_head list; /* command list */ 342 struct completion wait; /* compleation */ 343 struct block_device *bdev; /* bdev */ 344 unsigned short ref; /* reference count */ 345 unsigned char state; /* state */ 346 unsigned char queued; /* queued discard */ 347 int error; /* bio error */ 348 spinlock_t lock; /* for state/bio_ref updating */ 349 unsigned short bio_ref; /* bio reference count */ 350 }; 351 352 enum { 353 DPOLICY_BG, 354 DPOLICY_FORCE, 355 DPOLICY_FSTRIM, 356 DPOLICY_UMOUNT, 357 MAX_DPOLICY, 358 }; 359 360 struct discard_policy { 361 int type; /* type of discard */ 362 unsigned int min_interval; /* used for candidates exist */ 363 unsigned int mid_interval; /* used for device busy */ 364 unsigned int max_interval; /* used for candidates not exist */ 365 unsigned int max_requests; /* # of discards issued per round */ 366 unsigned int io_aware_gran; /* minimum granularity discard not be aware of I/O */ 367 bool io_aware; /* issue discard in idle time */ 368 bool sync; /* submit discard with REQ_SYNC flag */ 369 bool ordered; /* issue discard by lba order */ 370 bool timeout; /* discard timeout for put_super */ 371 unsigned int granularity; /* discard granularity */ 372 }; 373 374 struct discard_cmd_control { 375 struct task_struct *f2fs_issue_discard; /* discard thread */ 376 struct list_head entry_list; /* 4KB discard entry list */ 377 struct list_head pend_list[MAX_PLIST_NUM];/* store pending entries */ 378 struct list_head wait_list; /* store on-flushing entries */ 379 struct list_head fstrim_list; /* in-flight discard from fstrim */ 380 wait_queue_head_t discard_wait_queue; /* waiting queue for wake-up */ 381 unsigned int discard_wake; /* to wake up discard thread */ 382 struct mutex cmd_lock; 383 unsigned int nr_discards; /* # of discards in the list */ 384 unsigned int max_discards; /* max. discards to be issued */ 385 unsigned int discard_granularity; /* discard granularity */ 386 unsigned int undiscard_blks; /* # of undiscard blocks */ 387 unsigned int next_pos; /* next discard position */ 388 atomic_t issued_discard; /* # of issued discard */ 389 atomic_t queued_discard; /* # of queued discard */ 390 atomic_t discard_cmd_cnt; /* # of cached cmd count */ 391 struct rb_root_cached root; /* root of discard rb-tree */ 392 bool rbtree_check; /* config for consistence check */ 393 }; 394 395 /* for the list of fsync inodes, used only during recovery */ 396 struct fsync_inode_entry { 397 struct list_head list; /* list head */ 398 struct inode *inode; /* vfs inode pointer */ 399 block_t blkaddr; /* block address locating the last fsync */ 400 block_t last_dentry; /* block address locating the last dentry */ 401 }; 402 403 #define nats_in_cursum(jnl) (le16_to_cpu((jnl)->n_nats)) 404 #define sits_in_cursum(jnl) (le16_to_cpu((jnl)->n_sits)) 405 406 #define nat_in_journal(jnl, i) ((jnl)->nat_j.entries[i].ne) 407 #define nid_in_journal(jnl, i) ((jnl)->nat_j.entries[i].nid) 408 #define sit_in_journal(jnl, i) ((jnl)->sit_j.entries[i].se) 409 #define segno_in_journal(jnl, i) ((jnl)->sit_j.entries[i].segno) 410 411 #define MAX_NAT_JENTRIES(jnl) (NAT_JOURNAL_ENTRIES - nats_in_cursum(jnl)) 412 #define MAX_SIT_JENTRIES(jnl) (SIT_JOURNAL_ENTRIES - sits_in_cursum(jnl)) 413 414 static inline int update_nats_in_cursum(struct f2fs_journal *journal, int i) 415 { 416 int before = nats_in_cursum(journal); 417 418 journal->n_nats = cpu_to_le16(before + i); 419 return before; 420 } 421 422 static inline int update_sits_in_cursum(struct f2fs_journal *journal, int i) 423 { 424 int before = sits_in_cursum(journal); 425 426 journal->n_sits = cpu_to_le16(before + i); 427 return before; 428 } 429 430 static inline bool __has_cursum_space(struct f2fs_journal *journal, 431 int size, int type) 432 { 433 if (type == NAT_JOURNAL) 434 return size <= MAX_NAT_JENTRIES(journal); 435 return size <= MAX_SIT_JENTRIES(journal); 436 } 437 438 /* for inline stuff */ 439 #define DEF_INLINE_RESERVED_SIZE 1 440 static inline int get_extra_isize(struct inode *inode); 441 static inline int get_inline_xattr_addrs(struct inode *inode); 442 #define MAX_INLINE_DATA(inode) (sizeof(__le32) * \ 443 (CUR_ADDRS_PER_INODE(inode) - \ 444 get_inline_xattr_addrs(inode) - \ 445 DEF_INLINE_RESERVED_SIZE)) 446 447 /* for inline dir */ 448 #define NR_INLINE_DENTRY(inode) (MAX_INLINE_DATA(inode) * BITS_PER_BYTE / \ 449 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \ 450 BITS_PER_BYTE + 1)) 451 #define INLINE_DENTRY_BITMAP_SIZE(inode) \ 452 DIV_ROUND_UP(NR_INLINE_DENTRY(inode), BITS_PER_BYTE) 453 #define INLINE_RESERVED_SIZE(inode) (MAX_INLINE_DATA(inode) - \ 454 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \ 455 NR_INLINE_DENTRY(inode) + \ 456 INLINE_DENTRY_BITMAP_SIZE(inode))) 457 458 /* 459 * For INODE and NODE manager 460 */ 461 /* for directory operations */ 462 463 struct f2fs_filename { 464 /* 465 * The filename the user specified. This is NULL for some 466 * filesystem-internal operations, e.g. converting an inline directory 467 * to a non-inline one, or roll-forward recovering an encrypted dentry. 468 */ 469 const struct qstr *usr_fname; 470 471 /* 472 * The on-disk filename. For encrypted directories, this is encrypted. 473 * This may be NULL for lookups in an encrypted dir without the key. 474 */ 475 struct fscrypt_str disk_name; 476 477 /* The dirhash of this filename */ 478 f2fs_hash_t hash; 479 480 #ifdef CONFIG_FS_ENCRYPTION 481 /* 482 * For lookups in encrypted directories: either the buffer backing 483 * disk_name, or a buffer that holds the decoded no-key name. 484 */ 485 struct fscrypt_str crypto_buf; 486 #endif 487 #ifdef CONFIG_UNICODE 488 /* 489 * For casefolded directories: the casefolded name, but it's left NULL 490 * if the original name is not valid Unicode, if the directory is both 491 * casefolded and encrypted and its encryption key is unavailable, or if 492 * the filesystem is doing an internal operation where usr_fname is also 493 * NULL. In all these cases we fall back to treating the name as an 494 * opaque byte sequence. 495 */ 496 struct fscrypt_str cf_name; 497 #endif 498 }; 499 500 struct f2fs_dentry_ptr { 501 struct inode *inode; 502 void *bitmap; 503 struct f2fs_dir_entry *dentry; 504 __u8 (*filename)[F2FS_SLOT_LEN]; 505 int max; 506 int nr_bitmap; 507 }; 508 509 static inline void make_dentry_ptr_block(struct inode *inode, 510 struct f2fs_dentry_ptr *d, struct f2fs_dentry_block *t) 511 { 512 d->inode = inode; 513 d->max = NR_DENTRY_IN_BLOCK; 514 d->nr_bitmap = SIZE_OF_DENTRY_BITMAP; 515 d->bitmap = t->dentry_bitmap; 516 d->dentry = t->dentry; 517 d->filename = t->filename; 518 } 519 520 static inline void make_dentry_ptr_inline(struct inode *inode, 521 struct f2fs_dentry_ptr *d, void *t) 522 { 523 int entry_cnt = NR_INLINE_DENTRY(inode); 524 int bitmap_size = INLINE_DENTRY_BITMAP_SIZE(inode); 525 int reserved_size = INLINE_RESERVED_SIZE(inode); 526 527 d->inode = inode; 528 d->max = entry_cnt; 529 d->nr_bitmap = bitmap_size; 530 d->bitmap = t; 531 d->dentry = t + bitmap_size + reserved_size; 532 d->filename = t + bitmap_size + reserved_size + 533 SIZE_OF_DIR_ENTRY * entry_cnt; 534 } 535 536 /* 537 * XATTR_NODE_OFFSET stores xattrs to one node block per file keeping -1 538 * as its node offset to distinguish from index node blocks. 539 * But some bits are used to mark the node block. 540 */ 541 #define XATTR_NODE_OFFSET ((((unsigned int)-1) << OFFSET_BIT_SHIFT) \ 542 >> OFFSET_BIT_SHIFT) 543 enum { 544 ALLOC_NODE, /* allocate a new node page if needed */ 545 LOOKUP_NODE, /* look up a node without readahead */ 546 LOOKUP_NODE_RA, /* 547 * look up a node with readahead called 548 * by get_data_block. 549 */ 550 }; 551 552 #define DEFAULT_RETRY_IO_COUNT 8 /* maximum retry read IO or flush count */ 553 554 /* congestion wait timeout value, default: 20ms */ 555 #define DEFAULT_IO_TIMEOUT (msecs_to_jiffies(20)) 556 557 /* maximum retry quota flush count */ 558 #define DEFAULT_RETRY_QUOTA_FLUSH_COUNT 8 559 560 #define F2FS_LINK_MAX 0xffffffff /* maximum link count per file */ 561 562 #define MAX_DIR_RA_PAGES 4 /* maximum ra pages of dir */ 563 564 /* dirty segments threshold for triggering CP */ 565 #define DEFAULT_DIRTY_THRESHOLD 4 566 567 /* for in-memory extent cache entry */ 568 #define F2FS_MIN_EXTENT_LEN 64 /* minimum extent length */ 569 570 /* number of extent info in extent cache we try to shrink */ 571 #define EXTENT_CACHE_SHRINK_NUMBER 128 572 573 struct rb_entry { 574 struct rb_node rb_node; /* rb node located in rb-tree */ 575 union { 576 struct { 577 unsigned int ofs; /* start offset of the entry */ 578 unsigned int len; /* length of the entry */ 579 }; 580 unsigned long long key; /* 64-bits key */ 581 } __packed; 582 }; 583 584 struct extent_info { 585 unsigned int fofs; /* start offset in a file */ 586 unsigned int len; /* length of the extent */ 587 u32 blk; /* start block address of the extent */ 588 #ifdef CONFIG_F2FS_FS_COMPRESSION 589 unsigned int c_len; /* physical extent length of compressed blocks */ 590 #endif 591 }; 592 593 struct extent_node { 594 struct rb_node rb_node; /* rb node located in rb-tree */ 595 struct extent_info ei; /* extent info */ 596 struct list_head list; /* node in global extent list of sbi */ 597 struct extent_tree *et; /* extent tree pointer */ 598 }; 599 600 struct extent_tree { 601 nid_t ino; /* inode number */ 602 struct rb_root_cached root; /* root of extent info rb-tree */ 603 struct extent_node *cached_en; /* recently accessed extent node */ 604 struct extent_info largest; /* largested extent info */ 605 struct list_head list; /* to be used by sbi->zombie_list */ 606 rwlock_t lock; /* protect extent info rb-tree */ 607 atomic_t node_cnt; /* # of extent node in rb-tree*/ 608 bool largest_updated; /* largest extent updated */ 609 }; 610 611 /* 612 * This structure is taken from ext4_map_blocks. 613 * 614 * Note that, however, f2fs uses NEW and MAPPED flags for f2fs_map_blocks(). 615 */ 616 #define F2FS_MAP_NEW (1 << BH_New) 617 #define F2FS_MAP_MAPPED (1 << BH_Mapped) 618 #define F2FS_MAP_UNWRITTEN (1 << BH_Unwritten) 619 #define F2FS_MAP_FLAGS (F2FS_MAP_NEW | F2FS_MAP_MAPPED |\ 620 F2FS_MAP_UNWRITTEN) 621 622 struct f2fs_map_blocks { 623 block_t m_pblk; 624 block_t m_lblk; 625 unsigned int m_len; 626 unsigned int m_flags; 627 pgoff_t *m_next_pgofs; /* point next possible non-hole pgofs */ 628 pgoff_t *m_next_extent; /* point to next possible extent */ 629 int m_seg_type; 630 bool m_may_create; /* indicate it is from write path */ 631 }; 632 633 /* for flag in get_data_block */ 634 enum { 635 F2FS_GET_BLOCK_DEFAULT, 636 F2FS_GET_BLOCK_FIEMAP, 637 F2FS_GET_BLOCK_BMAP, 638 F2FS_GET_BLOCK_DIO, 639 F2FS_GET_BLOCK_PRE_DIO, 640 F2FS_GET_BLOCK_PRE_AIO, 641 F2FS_GET_BLOCK_PRECACHE, 642 }; 643 644 /* 645 * i_advise uses FADVISE_XXX_BIT. We can add additional hints later. 646 */ 647 #define FADVISE_COLD_BIT 0x01 648 #define FADVISE_LOST_PINO_BIT 0x02 649 #define FADVISE_ENCRYPT_BIT 0x04 650 #define FADVISE_ENC_NAME_BIT 0x08 651 #define FADVISE_KEEP_SIZE_BIT 0x10 652 #define FADVISE_HOT_BIT 0x20 653 #define FADVISE_VERITY_BIT 0x40 654 655 #define FADVISE_MODIFIABLE_BITS (FADVISE_COLD_BIT | FADVISE_HOT_BIT) 656 657 #define file_is_cold(inode) is_file(inode, FADVISE_COLD_BIT) 658 #define file_set_cold(inode) set_file(inode, FADVISE_COLD_BIT) 659 #define file_clear_cold(inode) clear_file(inode, FADVISE_COLD_BIT) 660 661 #define file_wrong_pino(inode) is_file(inode, FADVISE_LOST_PINO_BIT) 662 #define file_lost_pino(inode) set_file(inode, FADVISE_LOST_PINO_BIT) 663 #define file_got_pino(inode) clear_file(inode, FADVISE_LOST_PINO_BIT) 664 665 #define file_is_encrypt(inode) is_file(inode, FADVISE_ENCRYPT_BIT) 666 #define file_set_encrypt(inode) set_file(inode, FADVISE_ENCRYPT_BIT) 667 668 #define file_enc_name(inode) is_file(inode, FADVISE_ENC_NAME_BIT) 669 #define file_set_enc_name(inode) set_file(inode, FADVISE_ENC_NAME_BIT) 670 671 #define file_keep_isize(inode) is_file(inode, FADVISE_KEEP_SIZE_BIT) 672 #define file_set_keep_isize(inode) set_file(inode, FADVISE_KEEP_SIZE_BIT) 673 674 #define file_is_hot(inode) is_file(inode, FADVISE_HOT_BIT) 675 #define file_set_hot(inode) set_file(inode, FADVISE_HOT_BIT) 676 #define file_clear_hot(inode) clear_file(inode, FADVISE_HOT_BIT) 677 678 #define file_is_verity(inode) is_file(inode, FADVISE_VERITY_BIT) 679 #define file_set_verity(inode) set_file(inode, FADVISE_VERITY_BIT) 680 681 #define DEF_DIR_LEVEL 0 682 683 enum { 684 GC_FAILURE_PIN, 685 GC_FAILURE_ATOMIC, 686 MAX_GC_FAILURE 687 }; 688 689 /* used for f2fs_inode_info->flags */ 690 enum { 691 FI_NEW_INODE, /* indicate newly allocated inode */ 692 FI_DIRTY_INODE, /* indicate inode is dirty or not */ 693 FI_AUTO_RECOVER, /* indicate inode is recoverable */ 694 FI_DIRTY_DIR, /* indicate directory has dirty pages */ 695 FI_INC_LINK, /* need to increment i_nlink */ 696 FI_ACL_MODE, /* indicate acl mode */ 697 FI_NO_ALLOC, /* should not allocate any blocks */ 698 FI_FREE_NID, /* free allocated nide */ 699 FI_NO_EXTENT, /* not to use the extent cache */ 700 FI_INLINE_XATTR, /* used for inline xattr */ 701 FI_INLINE_DATA, /* used for inline data*/ 702 FI_INLINE_DENTRY, /* used for inline dentry */ 703 FI_APPEND_WRITE, /* inode has appended data */ 704 FI_UPDATE_WRITE, /* inode has in-place-update data */ 705 FI_NEED_IPU, /* used for ipu per file */ 706 FI_ATOMIC_FILE, /* indicate atomic file */ 707 FI_ATOMIC_COMMIT, /* indicate the state of atomical committing */ 708 FI_VOLATILE_FILE, /* indicate volatile file */ 709 FI_FIRST_BLOCK_WRITTEN, /* indicate #0 data block was written */ 710 FI_DROP_CACHE, /* drop dirty page cache */ 711 FI_DATA_EXIST, /* indicate data exists */ 712 FI_INLINE_DOTS, /* indicate inline dot dentries */ 713 FI_DO_DEFRAG, /* indicate defragment is running */ 714 FI_DIRTY_FILE, /* indicate regular/symlink has dirty pages */ 715 FI_NO_PREALLOC, /* indicate skipped preallocated blocks */ 716 FI_HOT_DATA, /* indicate file is hot */ 717 FI_EXTRA_ATTR, /* indicate file has extra attribute */ 718 FI_PROJ_INHERIT, /* indicate file inherits projectid */ 719 FI_PIN_FILE, /* indicate file should not be gced */ 720 FI_ATOMIC_REVOKE_REQUEST, /* request to drop atomic data */ 721 FI_VERITY_IN_PROGRESS, /* building fs-verity Merkle tree */ 722 FI_COMPRESSED_FILE, /* indicate file's data can be compressed */ 723 FI_COMPRESS_CORRUPT, /* indicate compressed cluster is corrupted */ 724 FI_MMAP_FILE, /* indicate file was mmapped */ 725 FI_ENABLE_COMPRESS, /* enable compression in "user" compression mode */ 726 FI_COMPRESS_RELEASED, /* compressed blocks were released */ 727 FI_ALIGNED_WRITE, /* enable aligned write */ 728 FI_MAX, /* max flag, never be used */ 729 }; 730 731 struct f2fs_inode_info { 732 struct inode vfs_inode; /* serve a vfs inode */ 733 unsigned long i_flags; /* keep an inode flags for ioctl */ 734 unsigned char i_advise; /* use to give file attribute hints */ 735 unsigned char i_dir_level; /* use for dentry level for large dir */ 736 unsigned int i_current_depth; /* only for directory depth */ 737 /* for gc failure statistic */ 738 unsigned int i_gc_failures[MAX_GC_FAILURE]; 739 unsigned int i_pino; /* parent inode number */ 740 umode_t i_acl_mode; /* keep file acl mode temporarily */ 741 742 /* Use below internally in f2fs*/ 743 unsigned long flags[BITS_TO_LONGS(FI_MAX)]; /* use to pass per-file flags */ 744 struct rw_semaphore i_sem; /* protect fi info */ 745 atomic_t dirty_pages; /* # of dirty pages */ 746 f2fs_hash_t chash; /* hash value of given file name */ 747 unsigned int clevel; /* maximum level of given file name */ 748 struct task_struct *task; /* lookup and create consistency */ 749 struct task_struct *cp_task; /* separate cp/wb IO stats*/ 750 nid_t i_xattr_nid; /* node id that contains xattrs */ 751 loff_t last_disk_size; /* lastly written file size */ 752 spinlock_t i_size_lock; /* protect last_disk_size */ 753 754 #ifdef CONFIG_QUOTA 755 struct dquot *i_dquot[MAXQUOTAS]; 756 757 /* quota space reservation, managed internally by quota code */ 758 qsize_t i_reserved_quota; 759 #endif 760 struct list_head dirty_list; /* dirty list for dirs and files */ 761 struct list_head gdirty_list; /* linked in global dirty list */ 762 struct list_head inmem_ilist; /* list for inmem inodes */ 763 struct list_head inmem_pages; /* inmemory pages managed by f2fs */ 764 struct task_struct *inmem_task; /* store inmemory task */ 765 struct mutex inmem_lock; /* lock for inmemory pages */ 766 struct extent_tree *extent_tree; /* cached extent_tree entry */ 767 768 /* avoid racing between foreground op and gc */ 769 struct rw_semaphore i_gc_rwsem[2]; 770 struct rw_semaphore i_xattr_sem; /* avoid racing between reading and changing EAs */ 771 772 int i_extra_isize; /* size of extra space located in i_addr */ 773 kprojid_t i_projid; /* id for project quota */ 774 int i_inline_xattr_size; /* inline xattr size */ 775 struct timespec64 i_crtime; /* inode creation time */ 776 struct timespec64 i_disk_time[4];/* inode disk times */ 777 778 /* for file compress */ 779 atomic_t i_compr_blocks; /* # of compressed blocks */ 780 unsigned char i_compress_algorithm; /* algorithm type */ 781 unsigned char i_log_cluster_size; /* log of cluster size */ 782 unsigned char i_compress_level; /* compress level (lz4hc,zstd) */ 783 unsigned short i_compress_flag; /* compress flag */ 784 unsigned int i_cluster_size; /* cluster size */ 785 }; 786 787 static inline void get_extent_info(struct extent_info *ext, 788 struct f2fs_extent *i_ext) 789 { 790 ext->fofs = le32_to_cpu(i_ext->fofs); 791 ext->blk = le32_to_cpu(i_ext->blk); 792 ext->len = le32_to_cpu(i_ext->len); 793 } 794 795 static inline void set_raw_extent(struct extent_info *ext, 796 struct f2fs_extent *i_ext) 797 { 798 i_ext->fofs = cpu_to_le32(ext->fofs); 799 i_ext->blk = cpu_to_le32(ext->blk); 800 i_ext->len = cpu_to_le32(ext->len); 801 } 802 803 static inline void set_extent_info(struct extent_info *ei, unsigned int fofs, 804 u32 blk, unsigned int len) 805 { 806 ei->fofs = fofs; 807 ei->blk = blk; 808 ei->len = len; 809 #ifdef CONFIG_F2FS_FS_COMPRESSION 810 ei->c_len = 0; 811 #endif 812 } 813 814 static inline bool __is_discard_mergeable(struct discard_info *back, 815 struct discard_info *front, unsigned int max_len) 816 { 817 return (back->lstart + back->len == front->lstart) && 818 (back->len + front->len <= max_len); 819 } 820 821 static inline bool __is_discard_back_mergeable(struct discard_info *cur, 822 struct discard_info *back, unsigned int max_len) 823 { 824 return __is_discard_mergeable(back, cur, max_len); 825 } 826 827 static inline bool __is_discard_front_mergeable(struct discard_info *cur, 828 struct discard_info *front, unsigned int max_len) 829 { 830 return __is_discard_mergeable(cur, front, max_len); 831 } 832 833 static inline bool __is_extent_mergeable(struct extent_info *back, 834 struct extent_info *front) 835 { 836 #ifdef CONFIG_F2FS_FS_COMPRESSION 837 if (back->c_len && back->len != back->c_len) 838 return false; 839 if (front->c_len && front->len != front->c_len) 840 return false; 841 #endif 842 return (back->fofs + back->len == front->fofs && 843 back->blk + back->len == front->blk); 844 } 845 846 static inline bool __is_back_mergeable(struct extent_info *cur, 847 struct extent_info *back) 848 { 849 return __is_extent_mergeable(back, cur); 850 } 851 852 static inline bool __is_front_mergeable(struct extent_info *cur, 853 struct extent_info *front) 854 { 855 return __is_extent_mergeable(cur, front); 856 } 857 858 extern void f2fs_mark_inode_dirty_sync(struct inode *inode, bool sync); 859 static inline void __try_update_largest_extent(struct extent_tree *et, 860 struct extent_node *en) 861 { 862 if (en->ei.len > et->largest.len) { 863 et->largest = en->ei; 864 et->largest_updated = true; 865 } 866 } 867 868 /* 869 * For free nid management 870 */ 871 enum nid_state { 872 FREE_NID, /* newly added to free nid list */ 873 PREALLOC_NID, /* it is preallocated */ 874 MAX_NID_STATE, 875 }; 876 877 enum nat_state { 878 TOTAL_NAT, 879 DIRTY_NAT, 880 RECLAIMABLE_NAT, 881 MAX_NAT_STATE, 882 }; 883 884 struct f2fs_nm_info { 885 block_t nat_blkaddr; /* base disk address of NAT */ 886 nid_t max_nid; /* maximum possible node ids */ 887 nid_t available_nids; /* # of available node ids */ 888 nid_t next_scan_nid; /* the next nid to be scanned */ 889 unsigned int ram_thresh; /* control the memory footprint */ 890 unsigned int ra_nid_pages; /* # of nid pages to be readaheaded */ 891 unsigned int dirty_nats_ratio; /* control dirty nats ratio threshold */ 892 893 /* NAT cache management */ 894 struct radix_tree_root nat_root;/* root of the nat entry cache */ 895 struct radix_tree_root nat_set_root;/* root of the nat set cache */ 896 struct rw_semaphore nat_tree_lock; /* protect nat entry tree */ 897 struct list_head nat_entries; /* cached nat entry list (clean) */ 898 spinlock_t nat_list_lock; /* protect clean nat entry list */ 899 unsigned int nat_cnt[MAX_NAT_STATE]; /* the # of cached nat entries */ 900 unsigned int nat_blocks; /* # of nat blocks */ 901 902 /* free node ids management */ 903 struct radix_tree_root free_nid_root;/* root of the free_nid cache */ 904 struct list_head free_nid_list; /* list for free nids excluding preallocated nids */ 905 unsigned int nid_cnt[MAX_NID_STATE]; /* the number of free node id */ 906 spinlock_t nid_list_lock; /* protect nid lists ops */ 907 struct mutex build_lock; /* lock for build free nids */ 908 unsigned char **free_nid_bitmap; 909 unsigned char *nat_block_bitmap; 910 unsigned short *free_nid_count; /* free nid count of NAT block */ 911 912 /* for checkpoint */ 913 char *nat_bitmap; /* NAT bitmap pointer */ 914 915 unsigned int nat_bits_blocks; /* # of nat bits blocks */ 916 unsigned char *nat_bits; /* NAT bits blocks */ 917 unsigned char *full_nat_bits; /* full NAT pages */ 918 unsigned char *empty_nat_bits; /* empty NAT pages */ 919 #ifdef CONFIG_F2FS_CHECK_FS 920 char *nat_bitmap_mir; /* NAT bitmap mirror */ 921 #endif 922 int bitmap_size; /* bitmap size */ 923 }; 924 925 /* 926 * this structure is used as one of function parameters. 927 * all the information are dedicated to a given direct node block determined 928 * by the data offset in a file. 929 */ 930 struct dnode_of_data { 931 struct inode *inode; /* vfs inode pointer */ 932 struct page *inode_page; /* its inode page, NULL is possible */ 933 struct page *node_page; /* cached direct node page */ 934 nid_t nid; /* node id of the direct node block */ 935 unsigned int ofs_in_node; /* data offset in the node page */ 936 bool inode_page_locked; /* inode page is locked or not */ 937 bool node_changed; /* is node block changed */ 938 char cur_level; /* level of hole node page */ 939 char max_level; /* level of current page located */ 940 block_t data_blkaddr; /* block address of the node block */ 941 }; 942 943 static inline void set_new_dnode(struct dnode_of_data *dn, struct inode *inode, 944 struct page *ipage, struct page *npage, nid_t nid) 945 { 946 memset(dn, 0, sizeof(*dn)); 947 dn->inode = inode; 948 dn->inode_page = ipage; 949 dn->node_page = npage; 950 dn->nid = nid; 951 } 952 953 /* 954 * For SIT manager 955 * 956 * By default, there are 6 active log areas across the whole main area. 957 * When considering hot and cold data separation to reduce cleaning overhead, 958 * we split 3 for data logs and 3 for node logs as hot, warm, and cold types, 959 * respectively. 960 * In the current design, you should not change the numbers intentionally. 961 * Instead, as a mount option such as active_logs=x, you can use 2, 4, and 6 962 * logs individually according to the underlying devices. (default: 6) 963 * Just in case, on-disk layout covers maximum 16 logs that consist of 8 for 964 * data and 8 for node logs. 965 */ 966 #define NR_CURSEG_DATA_TYPE (3) 967 #define NR_CURSEG_NODE_TYPE (3) 968 #define NR_CURSEG_INMEM_TYPE (2) 969 #define NR_CURSEG_RO_TYPE (2) 970 #define NR_CURSEG_PERSIST_TYPE (NR_CURSEG_DATA_TYPE + NR_CURSEG_NODE_TYPE) 971 #define NR_CURSEG_TYPE (NR_CURSEG_INMEM_TYPE + NR_CURSEG_PERSIST_TYPE) 972 973 enum { 974 CURSEG_HOT_DATA = 0, /* directory entry blocks */ 975 CURSEG_WARM_DATA, /* data blocks */ 976 CURSEG_COLD_DATA, /* multimedia or GCed data blocks */ 977 CURSEG_HOT_NODE, /* direct node blocks of directory files */ 978 CURSEG_WARM_NODE, /* direct node blocks of normal files */ 979 CURSEG_COLD_NODE, /* indirect node blocks */ 980 NR_PERSISTENT_LOG, /* number of persistent log */ 981 CURSEG_COLD_DATA_PINNED = NR_PERSISTENT_LOG, 982 /* pinned file that needs consecutive block address */ 983 CURSEG_ALL_DATA_ATGC, /* SSR alloctor in hot/warm/cold data area */ 984 NO_CHECK_TYPE, /* number of persistent & inmem log */ 985 }; 986 987 struct flush_cmd { 988 struct completion wait; 989 struct llist_node llnode; 990 nid_t ino; 991 int ret; 992 }; 993 994 struct flush_cmd_control { 995 struct task_struct *f2fs_issue_flush; /* flush thread */ 996 wait_queue_head_t flush_wait_queue; /* waiting queue for wake-up */ 997 atomic_t issued_flush; /* # of issued flushes */ 998 atomic_t queued_flush; /* # of queued flushes */ 999 struct llist_head issue_list; /* list for command issue */ 1000 struct llist_node *dispatch_list; /* list for command dispatch */ 1001 }; 1002 1003 struct f2fs_sm_info { 1004 struct sit_info *sit_info; /* whole segment information */ 1005 struct free_segmap_info *free_info; /* free segment information */ 1006 struct dirty_seglist_info *dirty_info; /* dirty segment information */ 1007 struct curseg_info *curseg_array; /* active segment information */ 1008 1009 struct rw_semaphore curseg_lock; /* for preventing curseg change */ 1010 1011 block_t seg0_blkaddr; /* block address of 0'th segment */ 1012 block_t main_blkaddr; /* start block address of main area */ 1013 block_t ssa_blkaddr; /* start block address of SSA area */ 1014 1015 unsigned int segment_count; /* total # of segments */ 1016 unsigned int main_segments; /* # of segments in main area */ 1017 unsigned int reserved_segments; /* # of reserved segments */ 1018 unsigned int ovp_segments; /* # of overprovision segments */ 1019 1020 /* a threshold to reclaim prefree segments */ 1021 unsigned int rec_prefree_segments; 1022 1023 /* for batched trimming */ 1024 unsigned int trim_sections; /* # of sections to trim */ 1025 1026 struct list_head sit_entry_set; /* sit entry set list */ 1027 1028 unsigned int ipu_policy; /* in-place-update policy */ 1029 unsigned int min_ipu_util; /* in-place-update threshold */ 1030 unsigned int min_fsync_blocks; /* threshold for fsync */ 1031 unsigned int min_seq_blocks; /* threshold for sequential blocks */ 1032 unsigned int min_hot_blocks; /* threshold for hot block allocation */ 1033 unsigned int min_ssr_sections; /* threshold to trigger SSR allocation */ 1034 1035 /* for flush command control */ 1036 struct flush_cmd_control *fcc_info; 1037 1038 /* for discard command control */ 1039 struct discard_cmd_control *dcc_info; 1040 }; 1041 1042 /* 1043 * For superblock 1044 */ 1045 /* 1046 * COUNT_TYPE for monitoring 1047 * 1048 * f2fs monitors the number of several block types such as on-writeback, 1049 * dirty dentry blocks, dirty node blocks, and dirty meta blocks. 1050 */ 1051 #define WB_DATA_TYPE(p) (__is_cp_guaranteed(p) ? F2FS_WB_CP_DATA : F2FS_WB_DATA) 1052 enum count_type { 1053 F2FS_DIRTY_DENTS, 1054 F2FS_DIRTY_DATA, 1055 F2FS_DIRTY_QDATA, 1056 F2FS_DIRTY_NODES, 1057 F2FS_DIRTY_META, 1058 F2FS_INMEM_PAGES, 1059 F2FS_DIRTY_IMETA, 1060 F2FS_WB_CP_DATA, 1061 F2FS_WB_DATA, 1062 F2FS_RD_DATA, 1063 F2FS_RD_NODE, 1064 F2FS_RD_META, 1065 F2FS_DIO_WRITE, 1066 F2FS_DIO_READ, 1067 NR_COUNT_TYPE, 1068 }; 1069 1070 /* 1071 * The below are the page types of bios used in submit_bio(). 1072 * The available types are: 1073 * DATA User data pages. It operates as async mode. 1074 * NODE Node pages. It operates as async mode. 1075 * META FS metadata pages such as SIT, NAT, CP. 1076 * NR_PAGE_TYPE The number of page types. 1077 * META_FLUSH Make sure the previous pages are written 1078 * with waiting the bio's completion 1079 * ... Only can be used with META. 1080 */ 1081 #define PAGE_TYPE_OF_BIO(type) ((type) > META ? META : (type)) 1082 enum page_type { 1083 DATA, 1084 NODE, 1085 META, 1086 NR_PAGE_TYPE, 1087 META_FLUSH, 1088 INMEM, /* the below types are used by tracepoints only. */ 1089 INMEM_DROP, 1090 INMEM_INVALIDATE, 1091 INMEM_REVOKE, 1092 IPU, 1093 OPU, 1094 }; 1095 1096 enum temp_type { 1097 HOT = 0, /* must be zero for meta bio */ 1098 WARM, 1099 COLD, 1100 NR_TEMP_TYPE, 1101 }; 1102 1103 enum need_lock_type { 1104 LOCK_REQ = 0, 1105 LOCK_DONE, 1106 LOCK_RETRY, 1107 }; 1108 1109 enum cp_reason_type { 1110 CP_NO_NEEDED, 1111 CP_NON_REGULAR, 1112 CP_COMPRESSED, 1113 CP_HARDLINK, 1114 CP_SB_NEED_CP, 1115 CP_WRONG_PINO, 1116 CP_NO_SPC_ROLL, 1117 CP_NODE_NEED_CP, 1118 CP_FASTBOOT_MODE, 1119 CP_SPEC_LOG_NUM, 1120 CP_RECOVER_DIR, 1121 }; 1122 1123 enum iostat_type { 1124 /* WRITE IO */ 1125 APP_DIRECT_IO, /* app direct write IOs */ 1126 APP_BUFFERED_IO, /* app buffered write IOs */ 1127 APP_WRITE_IO, /* app write IOs */ 1128 APP_MAPPED_IO, /* app mapped IOs */ 1129 FS_DATA_IO, /* data IOs from kworker/fsync/reclaimer */ 1130 FS_NODE_IO, /* node IOs from kworker/fsync/reclaimer */ 1131 FS_META_IO, /* meta IOs from kworker/reclaimer */ 1132 FS_GC_DATA_IO, /* data IOs from forground gc */ 1133 FS_GC_NODE_IO, /* node IOs from forground gc */ 1134 FS_CP_DATA_IO, /* data IOs from checkpoint */ 1135 FS_CP_NODE_IO, /* node IOs from checkpoint */ 1136 FS_CP_META_IO, /* meta IOs from checkpoint */ 1137 1138 /* READ IO */ 1139 APP_DIRECT_READ_IO, /* app direct read IOs */ 1140 APP_BUFFERED_READ_IO, /* app buffered read IOs */ 1141 APP_READ_IO, /* app read IOs */ 1142 APP_MAPPED_READ_IO, /* app mapped read IOs */ 1143 FS_DATA_READ_IO, /* data read IOs */ 1144 FS_GDATA_READ_IO, /* data read IOs from background gc */ 1145 FS_CDATA_READ_IO, /* compressed data read IOs */ 1146 FS_NODE_READ_IO, /* node read IOs */ 1147 FS_META_READ_IO, /* meta read IOs */ 1148 1149 /* other */ 1150 FS_DISCARD, /* discard */ 1151 NR_IO_TYPE, 1152 }; 1153 1154 struct f2fs_io_info { 1155 struct f2fs_sb_info *sbi; /* f2fs_sb_info pointer */ 1156 nid_t ino; /* inode number */ 1157 enum page_type type; /* contains DATA/NODE/META/META_FLUSH */ 1158 enum temp_type temp; /* contains HOT/WARM/COLD */ 1159 int op; /* contains REQ_OP_ */ 1160 int op_flags; /* req_flag_bits */ 1161 block_t new_blkaddr; /* new block address to be written */ 1162 block_t old_blkaddr; /* old block address before Cow */ 1163 struct page *page; /* page to be written */ 1164 struct page *encrypted_page; /* encrypted page */ 1165 struct page *compressed_page; /* compressed page */ 1166 struct list_head list; /* serialize IOs */ 1167 bool submitted; /* indicate IO submission */ 1168 int need_lock; /* indicate we need to lock cp_rwsem */ 1169 bool in_list; /* indicate fio is in io_list */ 1170 bool is_por; /* indicate IO is from recovery or not */ 1171 bool retry; /* need to reallocate block address */ 1172 int compr_blocks; /* # of compressed block addresses */ 1173 bool encrypted; /* indicate file is encrypted */ 1174 enum iostat_type io_type; /* io type */ 1175 struct writeback_control *io_wbc; /* writeback control */ 1176 struct bio **bio; /* bio for ipu */ 1177 sector_t *last_block; /* last block number in bio */ 1178 unsigned char version; /* version of the node */ 1179 }; 1180 1181 struct bio_entry { 1182 struct bio *bio; 1183 struct list_head list; 1184 }; 1185 1186 #define is_read_io(rw) ((rw) == READ) 1187 struct f2fs_bio_info { 1188 struct f2fs_sb_info *sbi; /* f2fs superblock */ 1189 struct bio *bio; /* bios to merge */ 1190 sector_t last_block_in_bio; /* last block number */ 1191 struct f2fs_io_info fio; /* store buffered io info. */ 1192 struct rw_semaphore io_rwsem; /* blocking op for bio */ 1193 spinlock_t io_lock; /* serialize DATA/NODE IOs */ 1194 struct list_head io_list; /* track fios */ 1195 struct list_head bio_list; /* bio entry list head */ 1196 struct rw_semaphore bio_list_lock; /* lock to protect bio entry list */ 1197 }; 1198 1199 #define FDEV(i) (sbi->devs[i]) 1200 #define RDEV(i) (raw_super->devs[i]) 1201 struct f2fs_dev_info { 1202 struct block_device *bdev; 1203 char path[MAX_PATH_LEN]; 1204 unsigned int total_segments; 1205 block_t start_blk; 1206 block_t end_blk; 1207 #ifdef CONFIG_BLK_DEV_ZONED 1208 unsigned int nr_blkz; /* Total number of zones */ 1209 unsigned long *blkz_seq; /* Bitmap indicating sequential zones */ 1210 block_t *zone_capacity_blocks; /* Array of zone capacity in blks */ 1211 #endif 1212 }; 1213 1214 enum inode_type { 1215 DIR_INODE, /* for dirty dir inode */ 1216 FILE_INODE, /* for dirty regular/symlink inode */ 1217 DIRTY_META, /* for all dirtied inode metadata */ 1218 ATOMIC_FILE, /* for all atomic files */ 1219 NR_INODE_TYPE, 1220 }; 1221 1222 /* for inner inode cache management */ 1223 struct inode_management { 1224 struct radix_tree_root ino_root; /* ino entry array */ 1225 spinlock_t ino_lock; /* for ino entry lock */ 1226 struct list_head ino_list; /* inode list head */ 1227 unsigned long ino_num; /* number of entries */ 1228 }; 1229 1230 /* for GC_AT */ 1231 struct atgc_management { 1232 bool atgc_enabled; /* ATGC is enabled or not */ 1233 struct rb_root_cached root; /* root of victim rb-tree */ 1234 struct list_head victim_list; /* linked with all victim entries */ 1235 unsigned int victim_count; /* victim count in rb-tree */ 1236 unsigned int candidate_ratio; /* candidate ratio */ 1237 unsigned int max_candidate_count; /* max candidate count */ 1238 unsigned int age_weight; /* age weight, vblock_weight = 100 - age_weight */ 1239 unsigned long long age_threshold; /* age threshold */ 1240 }; 1241 1242 /* For s_flag in struct f2fs_sb_info */ 1243 enum { 1244 SBI_IS_DIRTY, /* dirty flag for checkpoint */ 1245 SBI_IS_CLOSE, /* specify unmounting */ 1246 SBI_NEED_FSCK, /* need fsck.f2fs to fix */ 1247 SBI_POR_DOING, /* recovery is doing or not */ 1248 SBI_NEED_SB_WRITE, /* need to recover superblock */ 1249 SBI_NEED_CP, /* need to checkpoint */ 1250 SBI_IS_SHUTDOWN, /* shutdown by ioctl */ 1251 SBI_IS_RECOVERED, /* recovered orphan/data */ 1252 SBI_CP_DISABLED, /* CP was disabled last mount */ 1253 SBI_CP_DISABLED_QUICK, /* CP was disabled quickly */ 1254 SBI_QUOTA_NEED_FLUSH, /* need to flush quota info in CP */ 1255 SBI_QUOTA_SKIP_FLUSH, /* skip flushing quota in current CP */ 1256 SBI_QUOTA_NEED_REPAIR, /* quota file may be corrupted */ 1257 SBI_IS_RESIZEFS, /* resizefs is in process */ 1258 }; 1259 1260 enum { 1261 CP_TIME, 1262 REQ_TIME, 1263 DISCARD_TIME, 1264 GC_TIME, 1265 DISABLE_TIME, 1266 UMOUNT_DISCARD_TIMEOUT, 1267 MAX_TIME, 1268 }; 1269 1270 enum { 1271 GC_NORMAL, 1272 GC_IDLE_CB, 1273 GC_IDLE_GREEDY, 1274 GC_IDLE_AT, 1275 GC_URGENT_HIGH, 1276 GC_URGENT_LOW, 1277 MAX_GC_MODE, 1278 }; 1279 1280 enum { 1281 BGGC_MODE_ON, /* background gc is on */ 1282 BGGC_MODE_OFF, /* background gc is off */ 1283 BGGC_MODE_SYNC, /* 1284 * background gc is on, migrating blocks 1285 * like foreground gc 1286 */ 1287 }; 1288 1289 enum { 1290 FS_MODE_ADAPTIVE, /* use both lfs/ssr allocation */ 1291 FS_MODE_LFS, /* use lfs allocation only */ 1292 FS_MODE_FRAGMENT_SEG, /* segment fragmentation mode */ 1293 FS_MODE_FRAGMENT_BLK, /* block fragmentation mode */ 1294 }; 1295 1296 enum { 1297 WHINT_MODE_OFF, /* not pass down write hints */ 1298 WHINT_MODE_USER, /* try to pass down hints given by users */ 1299 WHINT_MODE_FS, /* pass down hints with F2FS policy */ 1300 }; 1301 1302 enum { 1303 ALLOC_MODE_DEFAULT, /* stay default */ 1304 ALLOC_MODE_REUSE, /* reuse segments as much as possible */ 1305 }; 1306 1307 enum fsync_mode { 1308 FSYNC_MODE_POSIX, /* fsync follows posix semantics */ 1309 FSYNC_MODE_STRICT, /* fsync behaves in line with ext4 */ 1310 FSYNC_MODE_NOBARRIER, /* fsync behaves nobarrier based on posix */ 1311 }; 1312 1313 enum { 1314 COMPR_MODE_FS, /* 1315 * automatically compress compression 1316 * enabled files 1317 */ 1318 COMPR_MODE_USER, /* 1319 * automatical compression is disabled. 1320 * user can control the file compression 1321 * using ioctls 1322 */ 1323 }; 1324 1325 enum { 1326 DISCARD_UNIT_BLOCK, /* basic discard unit is block */ 1327 DISCARD_UNIT_SEGMENT, /* basic discard unit is segment */ 1328 DISCARD_UNIT_SECTION, /* basic discard unit is section */ 1329 }; 1330 1331 static inline int f2fs_test_bit(unsigned int nr, char *addr); 1332 static inline void f2fs_set_bit(unsigned int nr, char *addr); 1333 static inline void f2fs_clear_bit(unsigned int nr, char *addr); 1334 1335 /* 1336 * Layout of f2fs page.private: 1337 * 1338 * Layout A: lowest bit should be 1 1339 * | bit0 = 1 | bit1 | bit2 | ... | bit MAX | private data .... | 1340 * bit 0 PAGE_PRIVATE_NOT_POINTER 1341 * bit 1 PAGE_PRIVATE_ATOMIC_WRITE 1342 * bit 2 PAGE_PRIVATE_DUMMY_WRITE 1343 * bit 3 PAGE_PRIVATE_ONGOING_MIGRATION 1344 * bit 4 PAGE_PRIVATE_INLINE_INODE 1345 * bit 5 PAGE_PRIVATE_REF_RESOURCE 1346 * bit 6- f2fs private data 1347 * 1348 * Layout B: lowest bit should be 0 1349 * page.private is a wrapped pointer. 1350 */ 1351 enum { 1352 PAGE_PRIVATE_NOT_POINTER, /* private contains non-pointer data */ 1353 PAGE_PRIVATE_ATOMIC_WRITE, /* data page from atomic write path */ 1354 PAGE_PRIVATE_DUMMY_WRITE, /* data page for padding aligned IO */ 1355 PAGE_PRIVATE_ONGOING_MIGRATION, /* data page which is on-going migrating */ 1356 PAGE_PRIVATE_INLINE_INODE, /* inode page contains inline data */ 1357 PAGE_PRIVATE_REF_RESOURCE, /* dirty page has referenced resources */ 1358 PAGE_PRIVATE_MAX 1359 }; 1360 1361 #define PAGE_PRIVATE_GET_FUNC(name, flagname) \ 1362 static inline bool page_private_##name(struct page *page) \ 1363 { \ 1364 return PagePrivate(page) && \ 1365 test_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)) && \ 1366 test_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \ 1367 } 1368 1369 #define PAGE_PRIVATE_SET_FUNC(name, flagname) \ 1370 static inline void set_page_private_##name(struct page *page) \ 1371 { \ 1372 if (!PagePrivate(page)) { \ 1373 get_page(page); \ 1374 SetPagePrivate(page); \ 1375 set_page_private(page, 0); \ 1376 } \ 1377 set_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)); \ 1378 set_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \ 1379 } 1380 1381 #define PAGE_PRIVATE_CLEAR_FUNC(name, flagname) \ 1382 static inline void clear_page_private_##name(struct page *page) \ 1383 { \ 1384 clear_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \ 1385 if (page_private(page) == 1 << PAGE_PRIVATE_NOT_POINTER) { \ 1386 set_page_private(page, 0); \ 1387 if (PagePrivate(page)) { \ 1388 ClearPagePrivate(page); \ 1389 put_page(page); \ 1390 }\ 1391 } \ 1392 } 1393 1394 PAGE_PRIVATE_GET_FUNC(nonpointer, NOT_POINTER); 1395 PAGE_PRIVATE_GET_FUNC(reference, REF_RESOURCE); 1396 PAGE_PRIVATE_GET_FUNC(inline, INLINE_INODE); 1397 PAGE_PRIVATE_GET_FUNC(gcing, ONGOING_MIGRATION); 1398 PAGE_PRIVATE_GET_FUNC(atomic, ATOMIC_WRITE); 1399 PAGE_PRIVATE_GET_FUNC(dummy, DUMMY_WRITE); 1400 1401 PAGE_PRIVATE_SET_FUNC(reference, REF_RESOURCE); 1402 PAGE_PRIVATE_SET_FUNC(inline, INLINE_INODE); 1403 PAGE_PRIVATE_SET_FUNC(gcing, ONGOING_MIGRATION); 1404 PAGE_PRIVATE_SET_FUNC(atomic, ATOMIC_WRITE); 1405 PAGE_PRIVATE_SET_FUNC(dummy, DUMMY_WRITE); 1406 1407 PAGE_PRIVATE_CLEAR_FUNC(reference, REF_RESOURCE); 1408 PAGE_PRIVATE_CLEAR_FUNC(inline, INLINE_INODE); 1409 PAGE_PRIVATE_CLEAR_FUNC(gcing, ONGOING_MIGRATION); 1410 PAGE_PRIVATE_CLEAR_FUNC(atomic, ATOMIC_WRITE); 1411 PAGE_PRIVATE_CLEAR_FUNC(dummy, DUMMY_WRITE); 1412 1413 static inline unsigned long get_page_private_data(struct page *page) 1414 { 1415 unsigned long data = page_private(page); 1416 1417 if (!test_bit(PAGE_PRIVATE_NOT_POINTER, &data)) 1418 return 0; 1419 return data >> PAGE_PRIVATE_MAX; 1420 } 1421 1422 static inline void set_page_private_data(struct page *page, unsigned long data) 1423 { 1424 if (!PagePrivate(page)) { 1425 get_page(page); 1426 SetPagePrivate(page); 1427 set_page_private(page, 0); 1428 } 1429 set_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)); 1430 page_private(page) |= data << PAGE_PRIVATE_MAX; 1431 } 1432 1433 static inline void clear_page_private_data(struct page *page) 1434 { 1435 page_private(page) &= (1 << PAGE_PRIVATE_MAX) - 1; 1436 if (page_private(page) == 1 << PAGE_PRIVATE_NOT_POINTER) { 1437 set_page_private(page, 0); 1438 if (PagePrivate(page)) { 1439 ClearPagePrivate(page); 1440 put_page(page); 1441 } 1442 } 1443 } 1444 1445 /* For compression */ 1446 enum compress_algorithm_type { 1447 COMPRESS_LZO, 1448 COMPRESS_LZ4, 1449 COMPRESS_ZSTD, 1450 COMPRESS_LZORLE, 1451 COMPRESS_MAX, 1452 }; 1453 1454 enum compress_flag { 1455 COMPRESS_CHKSUM, 1456 COMPRESS_MAX_FLAG, 1457 }; 1458 1459 #define COMPRESS_WATERMARK 20 1460 #define COMPRESS_PERCENT 20 1461 1462 #define COMPRESS_DATA_RESERVED_SIZE 4 1463 struct compress_data { 1464 __le32 clen; /* compressed data size */ 1465 __le32 chksum; /* compressed data chksum */ 1466 __le32 reserved[COMPRESS_DATA_RESERVED_SIZE]; /* reserved */ 1467 u8 cdata[]; /* compressed data */ 1468 }; 1469 1470 #define COMPRESS_HEADER_SIZE (sizeof(struct compress_data)) 1471 1472 #define F2FS_COMPRESSED_PAGE_MAGIC 0xF5F2C000 1473 1474 #define COMPRESS_LEVEL_OFFSET 8 1475 1476 /* compress context */ 1477 struct compress_ctx { 1478 struct inode *inode; /* inode the context belong to */ 1479 pgoff_t cluster_idx; /* cluster index number */ 1480 unsigned int cluster_size; /* page count in cluster */ 1481 unsigned int log_cluster_size; /* log of cluster size */ 1482 struct page **rpages; /* pages store raw data in cluster */ 1483 unsigned int nr_rpages; /* total page number in rpages */ 1484 struct page **cpages; /* pages store compressed data in cluster */ 1485 unsigned int nr_cpages; /* total page number in cpages */ 1486 void *rbuf; /* virtual mapped address on rpages */ 1487 struct compress_data *cbuf; /* virtual mapped address on cpages */ 1488 size_t rlen; /* valid data length in rbuf */ 1489 size_t clen; /* valid data length in cbuf */ 1490 void *private; /* payload buffer for specified compression algorithm */ 1491 void *private2; /* extra payload buffer */ 1492 }; 1493 1494 /* compress context for write IO path */ 1495 struct compress_io_ctx { 1496 u32 magic; /* magic number to indicate page is compressed */ 1497 struct inode *inode; /* inode the context belong to */ 1498 struct page **rpages; /* pages store raw data in cluster */ 1499 unsigned int nr_rpages; /* total page number in rpages */ 1500 atomic_t pending_pages; /* in-flight compressed page count */ 1501 }; 1502 1503 /* Context for decompressing one cluster on the read IO path */ 1504 struct decompress_io_ctx { 1505 u32 magic; /* magic number to indicate page is compressed */ 1506 struct inode *inode; /* inode the context belong to */ 1507 pgoff_t cluster_idx; /* cluster index number */ 1508 unsigned int cluster_size; /* page count in cluster */ 1509 unsigned int log_cluster_size; /* log of cluster size */ 1510 struct page **rpages; /* pages store raw data in cluster */ 1511 unsigned int nr_rpages; /* total page number in rpages */ 1512 struct page **cpages; /* pages store compressed data in cluster */ 1513 unsigned int nr_cpages; /* total page number in cpages */ 1514 struct page **tpages; /* temp pages to pad holes in cluster */ 1515 void *rbuf; /* virtual mapped address on rpages */ 1516 struct compress_data *cbuf; /* virtual mapped address on cpages */ 1517 size_t rlen; /* valid data length in rbuf */ 1518 size_t clen; /* valid data length in cbuf */ 1519 1520 /* 1521 * The number of compressed pages remaining to be read in this cluster. 1522 * This is initially nr_cpages. It is decremented by 1 each time a page 1523 * has been read (or failed to be read). When it reaches 0, the cluster 1524 * is decompressed (or an error is reported). 1525 * 1526 * If an error occurs before all the pages have been submitted for I/O, 1527 * then this will never reach 0. In this case the I/O submitter is 1528 * responsible for calling f2fs_decompress_end_io() instead. 1529 */ 1530 atomic_t remaining_pages; 1531 1532 /* 1533 * Number of references to this decompress_io_ctx. 1534 * 1535 * One reference is held for I/O completion. This reference is dropped 1536 * after the pagecache pages are updated and unlocked -- either after 1537 * decompression (and verity if enabled), or after an error. 1538 * 1539 * In addition, each compressed page holds a reference while it is in a 1540 * bio. These references are necessary prevent compressed pages from 1541 * being freed while they are still in a bio. 1542 */ 1543 refcount_t refcnt; 1544 1545 bool failed; /* IO error occurred before decompression? */ 1546 bool need_verity; /* need fs-verity verification after decompression? */ 1547 void *private; /* payload buffer for specified decompression algorithm */ 1548 void *private2; /* extra payload buffer */ 1549 struct work_struct verity_work; /* work to verify the decompressed pages */ 1550 }; 1551 1552 #define NULL_CLUSTER ((unsigned int)(~0)) 1553 #define MIN_COMPRESS_LOG_SIZE 2 1554 #define MAX_COMPRESS_LOG_SIZE 8 1555 #define MAX_COMPRESS_WINDOW_SIZE(log_size) ((PAGE_SIZE) << (log_size)) 1556 1557 struct f2fs_sb_info { 1558 struct super_block *sb; /* pointer to VFS super block */ 1559 struct proc_dir_entry *s_proc; /* proc entry */ 1560 struct f2fs_super_block *raw_super; /* raw super block pointer */ 1561 struct rw_semaphore sb_lock; /* lock for raw super block */ 1562 int valid_super_block; /* valid super block no */ 1563 unsigned long s_flag; /* flags for sbi */ 1564 struct mutex writepages; /* mutex for writepages() */ 1565 1566 #ifdef CONFIG_BLK_DEV_ZONED 1567 unsigned int blocks_per_blkz; /* F2FS blocks per zone */ 1568 unsigned int log_blocks_per_blkz; /* log2 F2FS blocks per zone */ 1569 #endif 1570 1571 /* for node-related operations */ 1572 struct f2fs_nm_info *nm_info; /* node manager */ 1573 struct inode *node_inode; /* cache node blocks */ 1574 1575 /* for segment-related operations */ 1576 struct f2fs_sm_info *sm_info; /* segment manager */ 1577 1578 /* for bio operations */ 1579 struct f2fs_bio_info *write_io[NR_PAGE_TYPE]; /* for write bios */ 1580 /* keep migration IO order for LFS mode */ 1581 struct rw_semaphore io_order_lock; 1582 mempool_t *write_io_dummy; /* Dummy pages */ 1583 1584 /* for checkpoint */ 1585 struct f2fs_checkpoint *ckpt; /* raw checkpoint pointer */ 1586 int cur_cp_pack; /* remain current cp pack */ 1587 spinlock_t cp_lock; /* for flag in ckpt */ 1588 struct inode *meta_inode; /* cache meta blocks */ 1589 struct rw_semaphore cp_global_sem; /* checkpoint procedure lock */ 1590 struct rw_semaphore cp_rwsem; /* blocking FS operations */ 1591 struct rw_semaphore node_write; /* locking node writes */ 1592 struct rw_semaphore node_change; /* locking node change */ 1593 wait_queue_head_t cp_wait; 1594 unsigned long last_time[MAX_TIME]; /* to store time in jiffies */ 1595 long interval_time[MAX_TIME]; /* to store thresholds */ 1596 struct ckpt_req_control cprc_info; /* for checkpoint request control */ 1597 1598 struct inode_management im[MAX_INO_ENTRY]; /* manage inode cache */ 1599 1600 spinlock_t fsync_node_lock; /* for node entry lock */ 1601 struct list_head fsync_node_list; /* node list head */ 1602 unsigned int fsync_seg_id; /* sequence id */ 1603 unsigned int fsync_node_num; /* number of node entries */ 1604 1605 /* for orphan inode, use 0'th array */ 1606 unsigned int max_orphans; /* max orphan inodes */ 1607 1608 /* for inode management */ 1609 struct list_head inode_list[NR_INODE_TYPE]; /* dirty inode list */ 1610 spinlock_t inode_lock[NR_INODE_TYPE]; /* for dirty inode list lock */ 1611 struct mutex flush_lock; /* for flush exclusion */ 1612 1613 /* for extent tree cache */ 1614 struct radix_tree_root extent_tree_root;/* cache extent cache entries */ 1615 struct mutex extent_tree_lock; /* locking extent radix tree */ 1616 struct list_head extent_list; /* lru list for shrinker */ 1617 spinlock_t extent_lock; /* locking extent lru list */ 1618 atomic_t total_ext_tree; /* extent tree count */ 1619 struct list_head zombie_list; /* extent zombie tree list */ 1620 atomic_t total_zombie_tree; /* extent zombie tree count */ 1621 atomic_t total_ext_node; /* extent info count */ 1622 1623 /* basic filesystem units */ 1624 unsigned int log_sectors_per_block; /* log2 sectors per block */ 1625 unsigned int log_blocksize; /* log2 block size */ 1626 unsigned int blocksize; /* block size */ 1627 unsigned int root_ino_num; /* root inode number*/ 1628 unsigned int node_ino_num; /* node inode number*/ 1629 unsigned int meta_ino_num; /* meta inode number*/ 1630 unsigned int log_blocks_per_seg; /* log2 blocks per segment */ 1631 unsigned int blocks_per_seg; /* blocks per segment */ 1632 unsigned int segs_per_sec; /* segments per section */ 1633 unsigned int secs_per_zone; /* sections per zone */ 1634 unsigned int total_sections; /* total section count */ 1635 unsigned int total_node_count; /* total node block count */ 1636 unsigned int total_valid_node_count; /* valid node block count */ 1637 int dir_level; /* directory level */ 1638 int readdir_ra; /* readahead inode in readdir */ 1639 u64 max_io_bytes; /* max io bytes to merge IOs */ 1640 1641 block_t user_block_count; /* # of user blocks */ 1642 block_t total_valid_block_count; /* # of valid blocks */ 1643 block_t discard_blks; /* discard command candidats */ 1644 block_t last_valid_block_count; /* for recovery */ 1645 block_t reserved_blocks; /* configurable reserved blocks */ 1646 block_t current_reserved_blocks; /* current reserved blocks */ 1647 1648 /* Additional tracking for no checkpoint mode */ 1649 block_t unusable_block_count; /* # of blocks saved by last cp */ 1650 1651 unsigned int nquota_files; /* # of quota sysfile */ 1652 struct rw_semaphore quota_sem; /* blocking cp for flags */ 1653 1654 /* # of pages, see count_type */ 1655 atomic_t nr_pages[NR_COUNT_TYPE]; 1656 /* # of allocated blocks */ 1657 struct percpu_counter alloc_valid_block_count; 1658 1659 /* writeback control */ 1660 atomic_t wb_sync_req[META]; /* count # of WB_SYNC threads */ 1661 1662 /* valid inode count */ 1663 struct percpu_counter total_valid_inode_count; 1664 1665 struct f2fs_mount_info mount_opt; /* mount options */ 1666 1667 /* for cleaning operations */ 1668 struct rw_semaphore gc_lock; /* 1669 * semaphore for GC, avoid 1670 * race between GC and GC or CP 1671 */ 1672 struct f2fs_gc_kthread *gc_thread; /* GC thread */ 1673 struct atgc_management am; /* atgc management */ 1674 unsigned int cur_victim_sec; /* current victim section num */ 1675 unsigned int gc_mode; /* current GC state */ 1676 unsigned int next_victim_seg[2]; /* next segment in victim section */ 1677 1678 /* for skip statistic */ 1679 unsigned int atomic_files; /* # of opened atomic file */ 1680 unsigned long long skipped_atomic_files[2]; /* FG_GC and BG_GC */ 1681 unsigned long long skipped_gc_rwsem; /* FG_GC only */ 1682 1683 /* threshold for gc trials on pinned files */ 1684 u64 gc_pin_file_threshold; 1685 struct rw_semaphore pin_sem; 1686 1687 /* maximum # of trials to find a victim segment for SSR and GC */ 1688 unsigned int max_victim_search; 1689 /* migration granularity of garbage collection, unit: segment */ 1690 unsigned int migration_granularity; 1691 1692 /* 1693 * for stat information. 1694 * one is for the LFS mode, and the other is for the SSR mode. 1695 */ 1696 #ifdef CONFIG_F2FS_STAT_FS 1697 struct f2fs_stat_info *stat_info; /* FS status information */ 1698 atomic_t meta_count[META_MAX]; /* # of meta blocks */ 1699 unsigned int segment_count[2]; /* # of allocated segments */ 1700 unsigned int block_count[2]; /* # of allocated blocks */ 1701 atomic_t inplace_count; /* # of inplace update */ 1702 atomic64_t total_hit_ext; /* # of lookup extent cache */ 1703 atomic64_t read_hit_rbtree; /* # of hit rbtree extent node */ 1704 atomic64_t read_hit_largest; /* # of hit largest extent node */ 1705 atomic64_t read_hit_cached; /* # of hit cached extent node */ 1706 atomic_t inline_xattr; /* # of inline_xattr inodes */ 1707 atomic_t inline_inode; /* # of inline_data inodes */ 1708 atomic_t inline_dir; /* # of inline_dentry inodes */ 1709 atomic_t compr_inode; /* # of compressed inodes */ 1710 atomic64_t compr_blocks; /* # of compressed blocks */ 1711 atomic_t vw_cnt; /* # of volatile writes */ 1712 atomic_t max_aw_cnt; /* max # of atomic writes */ 1713 atomic_t max_vw_cnt; /* max # of volatile writes */ 1714 unsigned int io_skip_bggc; /* skip background gc for in-flight IO */ 1715 unsigned int other_skip_bggc; /* skip background gc for other reasons */ 1716 unsigned int ndirty_inode[NR_INODE_TYPE]; /* # of dirty inodes */ 1717 #endif 1718 spinlock_t stat_lock; /* lock for stat operations */ 1719 1720 /* to attach REQ_META|REQ_FUA flags */ 1721 unsigned int data_io_flag; 1722 unsigned int node_io_flag; 1723 1724 /* For sysfs suppport */ 1725 struct kobject s_kobj; /* /sys/fs/f2fs/<devname> */ 1726 struct completion s_kobj_unregister; 1727 1728 struct kobject s_stat_kobj; /* /sys/fs/f2fs/<devname>/stat */ 1729 struct completion s_stat_kobj_unregister; 1730 1731 struct kobject s_feature_list_kobj; /* /sys/fs/f2fs/<devname>/feature_list */ 1732 struct completion s_feature_list_kobj_unregister; 1733 1734 /* For shrinker support */ 1735 struct list_head s_list; 1736 int s_ndevs; /* number of devices */ 1737 struct f2fs_dev_info *devs; /* for device list */ 1738 unsigned int dirty_device; /* for checkpoint data flush */ 1739 spinlock_t dev_lock; /* protect dirty_device */ 1740 struct mutex umount_mutex; 1741 unsigned int shrinker_run_no; 1742 1743 /* For write statistics */ 1744 u64 sectors_written_start; 1745 u64 kbytes_written; 1746 1747 /* Reference to checksum algorithm driver via cryptoapi */ 1748 struct crypto_shash *s_chksum_driver; 1749 1750 /* Precomputed FS UUID checksum for seeding other checksums */ 1751 __u32 s_chksum_seed; 1752 1753 struct workqueue_struct *post_read_wq; /* post read workqueue */ 1754 1755 struct kmem_cache *inline_xattr_slab; /* inline xattr entry */ 1756 unsigned int inline_xattr_slab_size; /* default inline xattr slab size */ 1757 1758 /* For reclaimed segs statistics per each GC mode */ 1759 unsigned int gc_segment_mode; /* GC state for reclaimed segments */ 1760 unsigned int gc_reclaimed_segs[MAX_GC_MODE]; /* Reclaimed segs for each mode */ 1761 1762 unsigned long seq_file_ra_mul; /* multiplier for ra_pages of seq. files in fadvise */ 1763 1764 int max_fragment_chunk; /* max chunk size for block fragmentation mode */ 1765 int max_fragment_hole; /* max hole size for block fragmentation mode */ 1766 1767 #ifdef CONFIG_F2FS_FS_COMPRESSION 1768 struct kmem_cache *page_array_slab; /* page array entry */ 1769 unsigned int page_array_slab_size; /* default page array slab size */ 1770 1771 /* For runtime compression statistics */ 1772 u64 compr_written_block; 1773 u64 compr_saved_block; 1774 u32 compr_new_inode; 1775 1776 /* For compressed block cache */ 1777 struct inode *compress_inode; /* cache compressed blocks */ 1778 unsigned int compress_percent; /* cache page percentage */ 1779 unsigned int compress_watermark; /* cache page watermark */ 1780 atomic_t compress_page_hit; /* cache hit count */ 1781 #endif 1782 1783 #ifdef CONFIG_F2FS_IOSTAT 1784 /* For app/fs IO statistics */ 1785 spinlock_t iostat_lock; 1786 unsigned long long rw_iostat[NR_IO_TYPE]; 1787 unsigned long long prev_rw_iostat[NR_IO_TYPE]; 1788 bool iostat_enable; 1789 unsigned long iostat_next_period; 1790 unsigned int iostat_period_ms; 1791 1792 /* For io latency related statistics info in one iostat period */ 1793 spinlock_t iostat_lat_lock; 1794 struct iostat_lat_info *iostat_io_lat; 1795 #endif 1796 }; 1797 1798 struct f2fs_private_dio { 1799 struct inode *inode; 1800 void *orig_private; 1801 bio_end_io_t *orig_end_io; 1802 bool write; 1803 }; 1804 1805 #ifdef CONFIG_F2FS_FAULT_INJECTION 1806 #define f2fs_show_injection_info(sbi, type) \ 1807 printk_ratelimited("%sF2FS-fs (%s) : inject %s in %s of %pS\n", \ 1808 KERN_INFO, sbi->sb->s_id, \ 1809 f2fs_fault_name[type], \ 1810 __func__, __builtin_return_address(0)) 1811 static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type) 1812 { 1813 struct f2fs_fault_info *ffi = &F2FS_OPTION(sbi).fault_info; 1814 1815 if (!ffi->inject_rate) 1816 return false; 1817 1818 if (!IS_FAULT_SET(ffi, type)) 1819 return false; 1820 1821 atomic_inc(&ffi->inject_ops); 1822 if (atomic_read(&ffi->inject_ops) >= ffi->inject_rate) { 1823 atomic_set(&ffi->inject_ops, 0); 1824 return true; 1825 } 1826 return false; 1827 } 1828 #else 1829 #define f2fs_show_injection_info(sbi, type) do { } while (0) 1830 static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type) 1831 { 1832 return false; 1833 } 1834 #endif 1835 1836 /* 1837 * Test if the mounted volume is a multi-device volume. 1838 * - For a single regular disk volume, sbi->s_ndevs is 0. 1839 * - For a single zoned disk volume, sbi->s_ndevs is 1. 1840 * - For a multi-device volume, sbi->s_ndevs is always 2 or more. 1841 */ 1842 static inline bool f2fs_is_multi_device(struct f2fs_sb_info *sbi) 1843 { 1844 return sbi->s_ndevs > 1; 1845 } 1846 1847 static inline void f2fs_update_time(struct f2fs_sb_info *sbi, int type) 1848 { 1849 unsigned long now = jiffies; 1850 1851 sbi->last_time[type] = now; 1852 1853 /* DISCARD_TIME and GC_TIME are based on REQ_TIME */ 1854 if (type == REQ_TIME) { 1855 sbi->last_time[DISCARD_TIME] = now; 1856 sbi->last_time[GC_TIME] = now; 1857 } 1858 } 1859 1860 static inline bool f2fs_time_over(struct f2fs_sb_info *sbi, int type) 1861 { 1862 unsigned long interval = sbi->interval_time[type] * HZ; 1863 1864 return time_after(jiffies, sbi->last_time[type] + interval); 1865 } 1866 1867 static inline unsigned int f2fs_time_to_wait(struct f2fs_sb_info *sbi, 1868 int type) 1869 { 1870 unsigned long interval = sbi->interval_time[type] * HZ; 1871 unsigned int wait_ms = 0; 1872 long delta; 1873 1874 delta = (sbi->last_time[type] + interval) - jiffies; 1875 if (delta > 0) 1876 wait_ms = jiffies_to_msecs(delta); 1877 1878 return wait_ms; 1879 } 1880 1881 /* 1882 * Inline functions 1883 */ 1884 static inline u32 __f2fs_crc32(struct f2fs_sb_info *sbi, u32 crc, 1885 const void *address, unsigned int length) 1886 { 1887 struct { 1888 struct shash_desc shash; 1889 char ctx[4]; 1890 } desc; 1891 int err; 1892 1893 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver) != sizeof(desc.ctx)); 1894 1895 desc.shash.tfm = sbi->s_chksum_driver; 1896 *(u32 *)desc.ctx = crc; 1897 1898 err = crypto_shash_update(&desc.shash, address, length); 1899 BUG_ON(err); 1900 1901 return *(u32 *)desc.ctx; 1902 } 1903 1904 static inline u32 f2fs_crc32(struct f2fs_sb_info *sbi, const void *address, 1905 unsigned int length) 1906 { 1907 return __f2fs_crc32(sbi, F2FS_SUPER_MAGIC, address, length); 1908 } 1909 1910 static inline bool f2fs_crc_valid(struct f2fs_sb_info *sbi, __u32 blk_crc, 1911 void *buf, size_t buf_size) 1912 { 1913 return f2fs_crc32(sbi, buf, buf_size) == blk_crc; 1914 } 1915 1916 static inline u32 f2fs_chksum(struct f2fs_sb_info *sbi, u32 crc, 1917 const void *address, unsigned int length) 1918 { 1919 return __f2fs_crc32(sbi, crc, address, length); 1920 } 1921 1922 static inline struct f2fs_inode_info *F2FS_I(struct inode *inode) 1923 { 1924 return container_of(inode, struct f2fs_inode_info, vfs_inode); 1925 } 1926 1927 static inline struct f2fs_sb_info *F2FS_SB(struct super_block *sb) 1928 { 1929 return sb->s_fs_info; 1930 } 1931 1932 static inline struct f2fs_sb_info *F2FS_I_SB(struct inode *inode) 1933 { 1934 return F2FS_SB(inode->i_sb); 1935 } 1936 1937 static inline struct f2fs_sb_info *F2FS_M_SB(struct address_space *mapping) 1938 { 1939 return F2FS_I_SB(mapping->host); 1940 } 1941 1942 static inline struct f2fs_sb_info *F2FS_P_SB(struct page *page) 1943 { 1944 return F2FS_M_SB(page_file_mapping(page)); 1945 } 1946 1947 static inline struct f2fs_super_block *F2FS_RAW_SUPER(struct f2fs_sb_info *sbi) 1948 { 1949 return (struct f2fs_super_block *)(sbi->raw_super); 1950 } 1951 1952 static inline struct f2fs_checkpoint *F2FS_CKPT(struct f2fs_sb_info *sbi) 1953 { 1954 return (struct f2fs_checkpoint *)(sbi->ckpt); 1955 } 1956 1957 static inline struct f2fs_node *F2FS_NODE(struct page *page) 1958 { 1959 return (struct f2fs_node *)page_address(page); 1960 } 1961 1962 static inline struct f2fs_inode *F2FS_INODE(struct page *page) 1963 { 1964 return &((struct f2fs_node *)page_address(page))->i; 1965 } 1966 1967 static inline struct f2fs_nm_info *NM_I(struct f2fs_sb_info *sbi) 1968 { 1969 return (struct f2fs_nm_info *)(sbi->nm_info); 1970 } 1971 1972 static inline struct f2fs_sm_info *SM_I(struct f2fs_sb_info *sbi) 1973 { 1974 return (struct f2fs_sm_info *)(sbi->sm_info); 1975 } 1976 1977 static inline struct sit_info *SIT_I(struct f2fs_sb_info *sbi) 1978 { 1979 return (struct sit_info *)(SM_I(sbi)->sit_info); 1980 } 1981 1982 static inline struct free_segmap_info *FREE_I(struct f2fs_sb_info *sbi) 1983 { 1984 return (struct free_segmap_info *)(SM_I(sbi)->free_info); 1985 } 1986 1987 static inline struct dirty_seglist_info *DIRTY_I(struct f2fs_sb_info *sbi) 1988 { 1989 return (struct dirty_seglist_info *)(SM_I(sbi)->dirty_info); 1990 } 1991 1992 static inline struct address_space *META_MAPPING(struct f2fs_sb_info *sbi) 1993 { 1994 return sbi->meta_inode->i_mapping; 1995 } 1996 1997 static inline struct address_space *NODE_MAPPING(struct f2fs_sb_info *sbi) 1998 { 1999 return sbi->node_inode->i_mapping; 2000 } 2001 2002 static inline bool is_sbi_flag_set(struct f2fs_sb_info *sbi, unsigned int type) 2003 { 2004 return test_bit(type, &sbi->s_flag); 2005 } 2006 2007 static inline void set_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type) 2008 { 2009 set_bit(type, &sbi->s_flag); 2010 } 2011 2012 static inline void clear_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type) 2013 { 2014 clear_bit(type, &sbi->s_flag); 2015 } 2016 2017 static inline unsigned long long cur_cp_version(struct f2fs_checkpoint *cp) 2018 { 2019 return le64_to_cpu(cp->checkpoint_ver); 2020 } 2021 2022 static inline unsigned long f2fs_qf_ino(struct super_block *sb, int type) 2023 { 2024 if (type < F2FS_MAX_QUOTAS) 2025 return le32_to_cpu(F2FS_SB(sb)->raw_super->qf_ino[type]); 2026 return 0; 2027 } 2028 2029 static inline __u64 cur_cp_crc(struct f2fs_checkpoint *cp) 2030 { 2031 size_t crc_offset = le32_to_cpu(cp->checksum_offset); 2032 return le32_to_cpu(*((__le32 *)((unsigned char *)cp + crc_offset))); 2033 } 2034 2035 static inline bool __is_set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 2036 { 2037 unsigned int ckpt_flags = le32_to_cpu(cp->ckpt_flags); 2038 2039 return ckpt_flags & f; 2040 } 2041 2042 static inline bool is_set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 2043 { 2044 return __is_set_ckpt_flags(F2FS_CKPT(sbi), f); 2045 } 2046 2047 static inline void __set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 2048 { 2049 unsigned int ckpt_flags; 2050 2051 ckpt_flags = le32_to_cpu(cp->ckpt_flags); 2052 ckpt_flags |= f; 2053 cp->ckpt_flags = cpu_to_le32(ckpt_flags); 2054 } 2055 2056 static inline void set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 2057 { 2058 unsigned long flags; 2059 2060 spin_lock_irqsave(&sbi->cp_lock, flags); 2061 __set_ckpt_flags(F2FS_CKPT(sbi), f); 2062 spin_unlock_irqrestore(&sbi->cp_lock, flags); 2063 } 2064 2065 static inline void __clear_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 2066 { 2067 unsigned int ckpt_flags; 2068 2069 ckpt_flags = le32_to_cpu(cp->ckpt_flags); 2070 ckpt_flags &= (~f); 2071 cp->ckpt_flags = cpu_to_le32(ckpt_flags); 2072 } 2073 2074 static inline void clear_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 2075 { 2076 unsigned long flags; 2077 2078 spin_lock_irqsave(&sbi->cp_lock, flags); 2079 __clear_ckpt_flags(F2FS_CKPT(sbi), f); 2080 spin_unlock_irqrestore(&sbi->cp_lock, flags); 2081 } 2082 2083 static inline void f2fs_lock_op(struct f2fs_sb_info *sbi) 2084 { 2085 down_read(&sbi->cp_rwsem); 2086 } 2087 2088 static inline int f2fs_trylock_op(struct f2fs_sb_info *sbi) 2089 { 2090 return down_read_trylock(&sbi->cp_rwsem); 2091 } 2092 2093 static inline void f2fs_unlock_op(struct f2fs_sb_info *sbi) 2094 { 2095 up_read(&sbi->cp_rwsem); 2096 } 2097 2098 static inline void f2fs_lock_all(struct f2fs_sb_info *sbi) 2099 { 2100 down_write(&sbi->cp_rwsem); 2101 } 2102 2103 static inline void f2fs_unlock_all(struct f2fs_sb_info *sbi) 2104 { 2105 up_write(&sbi->cp_rwsem); 2106 } 2107 2108 static inline int __get_cp_reason(struct f2fs_sb_info *sbi) 2109 { 2110 int reason = CP_SYNC; 2111 2112 if (test_opt(sbi, FASTBOOT)) 2113 reason = CP_FASTBOOT; 2114 if (is_sbi_flag_set(sbi, SBI_IS_CLOSE)) 2115 reason = CP_UMOUNT; 2116 return reason; 2117 } 2118 2119 static inline bool __remain_node_summaries(int reason) 2120 { 2121 return (reason & (CP_UMOUNT | CP_FASTBOOT)); 2122 } 2123 2124 static inline bool __exist_node_summaries(struct f2fs_sb_info *sbi) 2125 { 2126 return (is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG) || 2127 is_set_ckpt_flags(sbi, CP_FASTBOOT_FLAG)); 2128 } 2129 2130 /* 2131 * Check whether the inode has blocks or not 2132 */ 2133 static inline int F2FS_HAS_BLOCKS(struct inode *inode) 2134 { 2135 block_t xattr_block = F2FS_I(inode)->i_xattr_nid ? 1 : 0; 2136 2137 return (inode->i_blocks >> F2FS_LOG_SECTORS_PER_BLOCK) > xattr_block; 2138 } 2139 2140 static inline bool f2fs_has_xattr_block(unsigned int ofs) 2141 { 2142 return ofs == XATTR_NODE_OFFSET; 2143 } 2144 2145 static inline bool __allow_reserved_blocks(struct f2fs_sb_info *sbi, 2146 struct inode *inode, bool cap) 2147 { 2148 if (!inode) 2149 return true; 2150 if (!test_opt(sbi, RESERVE_ROOT)) 2151 return false; 2152 if (IS_NOQUOTA(inode)) 2153 return true; 2154 if (uid_eq(F2FS_OPTION(sbi).s_resuid, current_fsuid())) 2155 return true; 2156 if (!gid_eq(F2FS_OPTION(sbi).s_resgid, GLOBAL_ROOT_GID) && 2157 in_group_p(F2FS_OPTION(sbi).s_resgid)) 2158 return true; 2159 if (cap && capable(CAP_SYS_RESOURCE)) 2160 return true; 2161 return false; 2162 } 2163 2164 static inline void f2fs_i_blocks_write(struct inode *, block_t, bool, bool); 2165 static inline int inc_valid_block_count(struct f2fs_sb_info *sbi, 2166 struct inode *inode, blkcnt_t *count) 2167 { 2168 blkcnt_t diff = 0, release = 0; 2169 block_t avail_user_block_count; 2170 int ret; 2171 2172 ret = dquot_reserve_block(inode, *count); 2173 if (ret) 2174 return ret; 2175 2176 if (time_to_inject(sbi, FAULT_BLOCK)) { 2177 f2fs_show_injection_info(sbi, FAULT_BLOCK); 2178 release = *count; 2179 goto release_quota; 2180 } 2181 2182 /* 2183 * let's increase this in prior to actual block count change in order 2184 * for f2fs_sync_file to avoid data races when deciding checkpoint. 2185 */ 2186 percpu_counter_add(&sbi->alloc_valid_block_count, (*count)); 2187 2188 spin_lock(&sbi->stat_lock); 2189 sbi->total_valid_block_count += (block_t)(*count); 2190 avail_user_block_count = sbi->user_block_count - 2191 sbi->current_reserved_blocks; 2192 2193 if (!__allow_reserved_blocks(sbi, inode, true)) 2194 avail_user_block_count -= F2FS_OPTION(sbi).root_reserved_blocks; 2195 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) { 2196 if (avail_user_block_count > sbi->unusable_block_count) 2197 avail_user_block_count -= sbi->unusable_block_count; 2198 else 2199 avail_user_block_count = 0; 2200 } 2201 if (unlikely(sbi->total_valid_block_count > avail_user_block_count)) { 2202 diff = sbi->total_valid_block_count - avail_user_block_count; 2203 if (diff > *count) 2204 diff = *count; 2205 *count -= diff; 2206 release = diff; 2207 sbi->total_valid_block_count -= diff; 2208 if (!*count) { 2209 spin_unlock(&sbi->stat_lock); 2210 goto enospc; 2211 } 2212 } 2213 spin_unlock(&sbi->stat_lock); 2214 2215 if (unlikely(release)) { 2216 percpu_counter_sub(&sbi->alloc_valid_block_count, release); 2217 dquot_release_reservation_block(inode, release); 2218 } 2219 f2fs_i_blocks_write(inode, *count, true, true); 2220 return 0; 2221 2222 enospc: 2223 percpu_counter_sub(&sbi->alloc_valid_block_count, release); 2224 release_quota: 2225 dquot_release_reservation_block(inode, release); 2226 return -ENOSPC; 2227 } 2228 2229 __printf(2, 3) 2230 void f2fs_printk(struct f2fs_sb_info *sbi, const char *fmt, ...); 2231 2232 #define f2fs_err(sbi, fmt, ...) \ 2233 f2fs_printk(sbi, KERN_ERR fmt, ##__VA_ARGS__) 2234 #define f2fs_warn(sbi, fmt, ...) \ 2235 f2fs_printk(sbi, KERN_WARNING fmt, ##__VA_ARGS__) 2236 #define f2fs_notice(sbi, fmt, ...) \ 2237 f2fs_printk(sbi, KERN_NOTICE fmt, ##__VA_ARGS__) 2238 #define f2fs_info(sbi, fmt, ...) \ 2239 f2fs_printk(sbi, KERN_INFO fmt, ##__VA_ARGS__) 2240 #define f2fs_debug(sbi, fmt, ...) \ 2241 f2fs_printk(sbi, KERN_DEBUG fmt, ##__VA_ARGS__) 2242 2243 static inline void dec_valid_block_count(struct f2fs_sb_info *sbi, 2244 struct inode *inode, 2245 block_t count) 2246 { 2247 blkcnt_t sectors = count << F2FS_LOG_SECTORS_PER_BLOCK; 2248 2249 spin_lock(&sbi->stat_lock); 2250 f2fs_bug_on(sbi, sbi->total_valid_block_count < (block_t) count); 2251 sbi->total_valid_block_count -= (block_t)count; 2252 if (sbi->reserved_blocks && 2253 sbi->current_reserved_blocks < sbi->reserved_blocks) 2254 sbi->current_reserved_blocks = min(sbi->reserved_blocks, 2255 sbi->current_reserved_blocks + count); 2256 spin_unlock(&sbi->stat_lock); 2257 if (unlikely(inode->i_blocks < sectors)) { 2258 f2fs_warn(sbi, "Inconsistent i_blocks, ino:%lu, iblocks:%llu, sectors:%llu", 2259 inode->i_ino, 2260 (unsigned long long)inode->i_blocks, 2261 (unsigned long long)sectors); 2262 set_sbi_flag(sbi, SBI_NEED_FSCK); 2263 return; 2264 } 2265 f2fs_i_blocks_write(inode, count, false, true); 2266 } 2267 2268 static inline void inc_page_count(struct f2fs_sb_info *sbi, int count_type) 2269 { 2270 atomic_inc(&sbi->nr_pages[count_type]); 2271 2272 if (count_type == F2FS_DIRTY_DENTS || 2273 count_type == F2FS_DIRTY_NODES || 2274 count_type == F2FS_DIRTY_META || 2275 count_type == F2FS_DIRTY_QDATA || 2276 count_type == F2FS_DIRTY_IMETA) 2277 set_sbi_flag(sbi, SBI_IS_DIRTY); 2278 } 2279 2280 static inline void inode_inc_dirty_pages(struct inode *inode) 2281 { 2282 atomic_inc(&F2FS_I(inode)->dirty_pages); 2283 inc_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ? 2284 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA); 2285 if (IS_NOQUOTA(inode)) 2286 inc_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA); 2287 } 2288 2289 static inline void dec_page_count(struct f2fs_sb_info *sbi, int count_type) 2290 { 2291 atomic_dec(&sbi->nr_pages[count_type]); 2292 } 2293 2294 static inline void inode_dec_dirty_pages(struct inode *inode) 2295 { 2296 if (!S_ISDIR(inode->i_mode) && !S_ISREG(inode->i_mode) && 2297 !S_ISLNK(inode->i_mode)) 2298 return; 2299 2300 atomic_dec(&F2FS_I(inode)->dirty_pages); 2301 dec_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ? 2302 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA); 2303 if (IS_NOQUOTA(inode)) 2304 dec_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA); 2305 } 2306 2307 static inline s64 get_pages(struct f2fs_sb_info *sbi, int count_type) 2308 { 2309 return atomic_read(&sbi->nr_pages[count_type]); 2310 } 2311 2312 static inline int get_dirty_pages(struct inode *inode) 2313 { 2314 return atomic_read(&F2FS_I(inode)->dirty_pages); 2315 } 2316 2317 static inline int get_blocktype_secs(struct f2fs_sb_info *sbi, int block_type) 2318 { 2319 unsigned int pages_per_sec = sbi->segs_per_sec * sbi->blocks_per_seg; 2320 unsigned int segs = (get_pages(sbi, block_type) + pages_per_sec - 1) >> 2321 sbi->log_blocks_per_seg; 2322 2323 return segs / sbi->segs_per_sec; 2324 } 2325 2326 static inline block_t valid_user_blocks(struct f2fs_sb_info *sbi) 2327 { 2328 return sbi->total_valid_block_count; 2329 } 2330 2331 static inline block_t discard_blocks(struct f2fs_sb_info *sbi) 2332 { 2333 return sbi->discard_blks; 2334 } 2335 2336 static inline unsigned long __bitmap_size(struct f2fs_sb_info *sbi, int flag) 2337 { 2338 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi); 2339 2340 /* return NAT or SIT bitmap */ 2341 if (flag == NAT_BITMAP) 2342 return le32_to_cpu(ckpt->nat_ver_bitmap_bytesize); 2343 else if (flag == SIT_BITMAP) 2344 return le32_to_cpu(ckpt->sit_ver_bitmap_bytesize); 2345 2346 return 0; 2347 } 2348 2349 static inline block_t __cp_payload(struct f2fs_sb_info *sbi) 2350 { 2351 return le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_payload); 2352 } 2353 2354 static inline void *__bitmap_ptr(struct f2fs_sb_info *sbi, int flag) 2355 { 2356 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi); 2357 void *tmp_ptr = &ckpt->sit_nat_version_bitmap; 2358 int offset; 2359 2360 if (is_set_ckpt_flags(sbi, CP_LARGE_NAT_BITMAP_FLAG)) { 2361 offset = (flag == SIT_BITMAP) ? 2362 le32_to_cpu(ckpt->nat_ver_bitmap_bytesize) : 0; 2363 /* 2364 * if large_nat_bitmap feature is enabled, leave checksum 2365 * protection for all nat/sit bitmaps. 2366 */ 2367 return tmp_ptr + offset + sizeof(__le32); 2368 } 2369 2370 if (__cp_payload(sbi) > 0) { 2371 if (flag == NAT_BITMAP) 2372 return &ckpt->sit_nat_version_bitmap; 2373 else 2374 return (unsigned char *)ckpt + F2FS_BLKSIZE; 2375 } else { 2376 offset = (flag == NAT_BITMAP) ? 2377 le32_to_cpu(ckpt->sit_ver_bitmap_bytesize) : 0; 2378 return tmp_ptr + offset; 2379 } 2380 } 2381 2382 static inline block_t __start_cp_addr(struct f2fs_sb_info *sbi) 2383 { 2384 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr); 2385 2386 if (sbi->cur_cp_pack == 2) 2387 start_addr += sbi->blocks_per_seg; 2388 return start_addr; 2389 } 2390 2391 static inline block_t __start_cp_next_addr(struct f2fs_sb_info *sbi) 2392 { 2393 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr); 2394 2395 if (sbi->cur_cp_pack == 1) 2396 start_addr += sbi->blocks_per_seg; 2397 return start_addr; 2398 } 2399 2400 static inline void __set_cp_next_pack(struct f2fs_sb_info *sbi) 2401 { 2402 sbi->cur_cp_pack = (sbi->cur_cp_pack == 1) ? 2 : 1; 2403 } 2404 2405 static inline block_t __start_sum_addr(struct f2fs_sb_info *sbi) 2406 { 2407 return le32_to_cpu(F2FS_CKPT(sbi)->cp_pack_start_sum); 2408 } 2409 2410 static inline int inc_valid_node_count(struct f2fs_sb_info *sbi, 2411 struct inode *inode, bool is_inode) 2412 { 2413 block_t valid_block_count; 2414 unsigned int valid_node_count, user_block_count; 2415 int err; 2416 2417 if (is_inode) { 2418 if (inode) { 2419 err = dquot_alloc_inode(inode); 2420 if (err) 2421 return err; 2422 } 2423 } else { 2424 err = dquot_reserve_block(inode, 1); 2425 if (err) 2426 return err; 2427 } 2428 2429 if (time_to_inject(sbi, FAULT_BLOCK)) { 2430 f2fs_show_injection_info(sbi, FAULT_BLOCK); 2431 goto enospc; 2432 } 2433 2434 spin_lock(&sbi->stat_lock); 2435 2436 valid_block_count = sbi->total_valid_block_count + 2437 sbi->current_reserved_blocks + 1; 2438 2439 if (!__allow_reserved_blocks(sbi, inode, false)) 2440 valid_block_count += F2FS_OPTION(sbi).root_reserved_blocks; 2441 user_block_count = sbi->user_block_count; 2442 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) 2443 user_block_count -= sbi->unusable_block_count; 2444 2445 if (unlikely(valid_block_count > user_block_count)) { 2446 spin_unlock(&sbi->stat_lock); 2447 goto enospc; 2448 } 2449 2450 valid_node_count = sbi->total_valid_node_count + 1; 2451 if (unlikely(valid_node_count > sbi->total_node_count)) { 2452 spin_unlock(&sbi->stat_lock); 2453 goto enospc; 2454 } 2455 2456 sbi->total_valid_node_count++; 2457 sbi->total_valid_block_count++; 2458 spin_unlock(&sbi->stat_lock); 2459 2460 if (inode) { 2461 if (is_inode) 2462 f2fs_mark_inode_dirty_sync(inode, true); 2463 else 2464 f2fs_i_blocks_write(inode, 1, true, true); 2465 } 2466 2467 percpu_counter_inc(&sbi->alloc_valid_block_count); 2468 return 0; 2469 2470 enospc: 2471 if (is_inode) { 2472 if (inode) 2473 dquot_free_inode(inode); 2474 } else { 2475 dquot_release_reservation_block(inode, 1); 2476 } 2477 return -ENOSPC; 2478 } 2479 2480 static inline void dec_valid_node_count(struct f2fs_sb_info *sbi, 2481 struct inode *inode, bool is_inode) 2482 { 2483 spin_lock(&sbi->stat_lock); 2484 2485 f2fs_bug_on(sbi, !sbi->total_valid_block_count); 2486 f2fs_bug_on(sbi, !sbi->total_valid_node_count); 2487 2488 sbi->total_valid_node_count--; 2489 sbi->total_valid_block_count--; 2490 if (sbi->reserved_blocks && 2491 sbi->current_reserved_blocks < sbi->reserved_blocks) 2492 sbi->current_reserved_blocks++; 2493 2494 spin_unlock(&sbi->stat_lock); 2495 2496 if (is_inode) { 2497 dquot_free_inode(inode); 2498 } else { 2499 if (unlikely(inode->i_blocks == 0)) { 2500 f2fs_warn(sbi, "dec_valid_node_count: inconsistent i_blocks, ino:%lu, iblocks:%llu", 2501 inode->i_ino, 2502 (unsigned long long)inode->i_blocks); 2503 set_sbi_flag(sbi, SBI_NEED_FSCK); 2504 return; 2505 } 2506 f2fs_i_blocks_write(inode, 1, false, true); 2507 } 2508 } 2509 2510 static inline unsigned int valid_node_count(struct f2fs_sb_info *sbi) 2511 { 2512 return sbi->total_valid_node_count; 2513 } 2514 2515 static inline void inc_valid_inode_count(struct f2fs_sb_info *sbi) 2516 { 2517 percpu_counter_inc(&sbi->total_valid_inode_count); 2518 } 2519 2520 static inline void dec_valid_inode_count(struct f2fs_sb_info *sbi) 2521 { 2522 percpu_counter_dec(&sbi->total_valid_inode_count); 2523 } 2524 2525 static inline s64 valid_inode_count(struct f2fs_sb_info *sbi) 2526 { 2527 return percpu_counter_sum_positive(&sbi->total_valid_inode_count); 2528 } 2529 2530 static inline struct page *f2fs_grab_cache_page(struct address_space *mapping, 2531 pgoff_t index, bool for_write) 2532 { 2533 struct page *page; 2534 2535 if (IS_ENABLED(CONFIG_F2FS_FAULT_INJECTION)) { 2536 if (!for_write) 2537 page = find_get_page_flags(mapping, index, 2538 FGP_LOCK | FGP_ACCESSED); 2539 else 2540 page = find_lock_page(mapping, index); 2541 if (page) 2542 return page; 2543 2544 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_ALLOC)) { 2545 f2fs_show_injection_info(F2FS_M_SB(mapping), 2546 FAULT_PAGE_ALLOC); 2547 return NULL; 2548 } 2549 } 2550 2551 if (!for_write) 2552 return grab_cache_page(mapping, index); 2553 return grab_cache_page_write_begin(mapping, index, AOP_FLAG_NOFS); 2554 } 2555 2556 static inline struct page *f2fs_pagecache_get_page( 2557 struct address_space *mapping, pgoff_t index, 2558 int fgp_flags, gfp_t gfp_mask) 2559 { 2560 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_GET)) { 2561 f2fs_show_injection_info(F2FS_M_SB(mapping), FAULT_PAGE_GET); 2562 return NULL; 2563 } 2564 2565 return pagecache_get_page(mapping, index, fgp_flags, gfp_mask); 2566 } 2567 2568 static inline void f2fs_copy_page(struct page *src, struct page *dst) 2569 { 2570 char *src_kaddr = kmap(src); 2571 char *dst_kaddr = kmap(dst); 2572 2573 memcpy(dst_kaddr, src_kaddr, PAGE_SIZE); 2574 kunmap(dst); 2575 kunmap(src); 2576 } 2577 2578 static inline void f2fs_put_page(struct page *page, int unlock) 2579 { 2580 if (!page) 2581 return; 2582 2583 if (unlock) { 2584 f2fs_bug_on(F2FS_P_SB(page), !PageLocked(page)); 2585 unlock_page(page); 2586 } 2587 put_page(page); 2588 } 2589 2590 static inline void f2fs_put_dnode(struct dnode_of_data *dn) 2591 { 2592 if (dn->node_page) 2593 f2fs_put_page(dn->node_page, 1); 2594 if (dn->inode_page && dn->node_page != dn->inode_page) 2595 f2fs_put_page(dn->inode_page, 0); 2596 dn->node_page = NULL; 2597 dn->inode_page = NULL; 2598 } 2599 2600 static inline struct kmem_cache *f2fs_kmem_cache_create(const char *name, 2601 size_t size) 2602 { 2603 return kmem_cache_create(name, size, 0, SLAB_RECLAIM_ACCOUNT, NULL); 2604 } 2605 2606 static inline void *f2fs_kmem_cache_alloc_nofail(struct kmem_cache *cachep, 2607 gfp_t flags) 2608 { 2609 void *entry; 2610 2611 entry = kmem_cache_alloc(cachep, flags); 2612 if (!entry) 2613 entry = kmem_cache_alloc(cachep, flags | __GFP_NOFAIL); 2614 return entry; 2615 } 2616 2617 static inline void *f2fs_kmem_cache_alloc(struct kmem_cache *cachep, 2618 gfp_t flags, bool nofail, struct f2fs_sb_info *sbi) 2619 { 2620 if (nofail) 2621 return f2fs_kmem_cache_alloc_nofail(cachep, flags); 2622 2623 if (time_to_inject(sbi, FAULT_SLAB_ALLOC)) { 2624 f2fs_show_injection_info(sbi, FAULT_SLAB_ALLOC); 2625 return NULL; 2626 } 2627 2628 return kmem_cache_alloc(cachep, flags); 2629 } 2630 2631 static inline bool is_inflight_io(struct f2fs_sb_info *sbi, int type) 2632 { 2633 if (get_pages(sbi, F2FS_RD_DATA) || get_pages(sbi, F2FS_RD_NODE) || 2634 get_pages(sbi, F2FS_RD_META) || get_pages(sbi, F2FS_WB_DATA) || 2635 get_pages(sbi, F2FS_WB_CP_DATA) || 2636 get_pages(sbi, F2FS_DIO_READ) || 2637 get_pages(sbi, F2FS_DIO_WRITE)) 2638 return true; 2639 2640 if (type != DISCARD_TIME && SM_I(sbi) && SM_I(sbi)->dcc_info && 2641 atomic_read(&SM_I(sbi)->dcc_info->queued_discard)) 2642 return true; 2643 2644 if (SM_I(sbi) && SM_I(sbi)->fcc_info && 2645 atomic_read(&SM_I(sbi)->fcc_info->queued_flush)) 2646 return true; 2647 return false; 2648 } 2649 2650 static inline bool is_idle(struct f2fs_sb_info *sbi, int type) 2651 { 2652 if (sbi->gc_mode == GC_URGENT_HIGH) 2653 return true; 2654 2655 if (is_inflight_io(sbi, type)) 2656 return false; 2657 2658 if (sbi->gc_mode == GC_URGENT_LOW && 2659 (type == DISCARD_TIME || type == GC_TIME)) 2660 return true; 2661 2662 return f2fs_time_over(sbi, type); 2663 } 2664 2665 static inline void f2fs_radix_tree_insert(struct radix_tree_root *root, 2666 unsigned long index, void *item) 2667 { 2668 while (radix_tree_insert(root, index, item)) 2669 cond_resched(); 2670 } 2671 2672 #define RAW_IS_INODE(p) ((p)->footer.nid == (p)->footer.ino) 2673 2674 static inline bool IS_INODE(struct page *page) 2675 { 2676 struct f2fs_node *p = F2FS_NODE(page); 2677 2678 return RAW_IS_INODE(p); 2679 } 2680 2681 static inline int offset_in_addr(struct f2fs_inode *i) 2682 { 2683 return (i->i_inline & F2FS_EXTRA_ATTR) ? 2684 (le16_to_cpu(i->i_extra_isize) / sizeof(__le32)) : 0; 2685 } 2686 2687 static inline __le32 *blkaddr_in_node(struct f2fs_node *node) 2688 { 2689 return RAW_IS_INODE(node) ? node->i.i_addr : node->dn.addr; 2690 } 2691 2692 static inline int f2fs_has_extra_attr(struct inode *inode); 2693 static inline block_t data_blkaddr(struct inode *inode, 2694 struct page *node_page, unsigned int offset) 2695 { 2696 struct f2fs_node *raw_node; 2697 __le32 *addr_array; 2698 int base = 0; 2699 bool is_inode = IS_INODE(node_page); 2700 2701 raw_node = F2FS_NODE(node_page); 2702 2703 if (is_inode) { 2704 if (!inode) 2705 /* from GC path only */ 2706 base = offset_in_addr(&raw_node->i); 2707 else if (f2fs_has_extra_attr(inode)) 2708 base = get_extra_isize(inode); 2709 } 2710 2711 addr_array = blkaddr_in_node(raw_node); 2712 return le32_to_cpu(addr_array[base + offset]); 2713 } 2714 2715 static inline block_t f2fs_data_blkaddr(struct dnode_of_data *dn) 2716 { 2717 return data_blkaddr(dn->inode, dn->node_page, dn->ofs_in_node); 2718 } 2719 2720 static inline int f2fs_test_bit(unsigned int nr, char *addr) 2721 { 2722 int mask; 2723 2724 addr += (nr >> 3); 2725 mask = 1 << (7 - (nr & 0x07)); 2726 return mask & *addr; 2727 } 2728 2729 static inline void f2fs_set_bit(unsigned int nr, char *addr) 2730 { 2731 int mask; 2732 2733 addr += (nr >> 3); 2734 mask = 1 << (7 - (nr & 0x07)); 2735 *addr |= mask; 2736 } 2737 2738 static inline void f2fs_clear_bit(unsigned int nr, char *addr) 2739 { 2740 int mask; 2741 2742 addr += (nr >> 3); 2743 mask = 1 << (7 - (nr & 0x07)); 2744 *addr &= ~mask; 2745 } 2746 2747 static inline int f2fs_test_and_set_bit(unsigned int nr, char *addr) 2748 { 2749 int mask; 2750 int ret; 2751 2752 addr += (nr >> 3); 2753 mask = 1 << (7 - (nr & 0x07)); 2754 ret = mask & *addr; 2755 *addr |= mask; 2756 return ret; 2757 } 2758 2759 static inline int f2fs_test_and_clear_bit(unsigned int nr, char *addr) 2760 { 2761 int mask; 2762 int ret; 2763 2764 addr += (nr >> 3); 2765 mask = 1 << (7 - (nr & 0x07)); 2766 ret = mask & *addr; 2767 *addr &= ~mask; 2768 return ret; 2769 } 2770 2771 static inline void f2fs_change_bit(unsigned int nr, char *addr) 2772 { 2773 int mask; 2774 2775 addr += (nr >> 3); 2776 mask = 1 << (7 - (nr & 0x07)); 2777 *addr ^= mask; 2778 } 2779 2780 /* 2781 * On-disk inode flags (f2fs_inode::i_flags) 2782 */ 2783 #define F2FS_COMPR_FL 0x00000004 /* Compress file */ 2784 #define F2FS_SYNC_FL 0x00000008 /* Synchronous updates */ 2785 #define F2FS_IMMUTABLE_FL 0x00000010 /* Immutable file */ 2786 #define F2FS_APPEND_FL 0x00000020 /* writes to file may only append */ 2787 #define F2FS_NODUMP_FL 0x00000040 /* do not dump file */ 2788 #define F2FS_NOATIME_FL 0x00000080 /* do not update atime */ 2789 #define F2FS_NOCOMP_FL 0x00000400 /* Don't compress */ 2790 #define F2FS_INDEX_FL 0x00001000 /* hash-indexed directory */ 2791 #define F2FS_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */ 2792 #define F2FS_PROJINHERIT_FL 0x20000000 /* Create with parents projid */ 2793 #define F2FS_CASEFOLD_FL 0x40000000 /* Casefolded file */ 2794 2795 /* Flags that should be inherited by new inodes from their parent. */ 2796 #define F2FS_FL_INHERITED (F2FS_SYNC_FL | F2FS_NODUMP_FL | F2FS_NOATIME_FL | \ 2797 F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \ 2798 F2FS_CASEFOLD_FL | F2FS_COMPR_FL | F2FS_NOCOMP_FL) 2799 2800 /* Flags that are appropriate for regular files (all but dir-specific ones). */ 2801 #define F2FS_REG_FLMASK (~(F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \ 2802 F2FS_CASEFOLD_FL)) 2803 2804 /* Flags that are appropriate for non-directories/regular files. */ 2805 #define F2FS_OTHER_FLMASK (F2FS_NODUMP_FL | F2FS_NOATIME_FL) 2806 2807 static inline __u32 f2fs_mask_flags(umode_t mode, __u32 flags) 2808 { 2809 if (S_ISDIR(mode)) 2810 return flags; 2811 else if (S_ISREG(mode)) 2812 return flags & F2FS_REG_FLMASK; 2813 else 2814 return flags & F2FS_OTHER_FLMASK; 2815 } 2816 2817 static inline void __mark_inode_dirty_flag(struct inode *inode, 2818 int flag, bool set) 2819 { 2820 switch (flag) { 2821 case FI_INLINE_XATTR: 2822 case FI_INLINE_DATA: 2823 case FI_INLINE_DENTRY: 2824 case FI_NEW_INODE: 2825 if (set) 2826 return; 2827 fallthrough; 2828 case FI_DATA_EXIST: 2829 case FI_INLINE_DOTS: 2830 case FI_PIN_FILE: 2831 case FI_COMPRESS_RELEASED: 2832 f2fs_mark_inode_dirty_sync(inode, true); 2833 } 2834 } 2835 2836 static inline void set_inode_flag(struct inode *inode, int flag) 2837 { 2838 set_bit(flag, F2FS_I(inode)->flags); 2839 __mark_inode_dirty_flag(inode, flag, true); 2840 } 2841 2842 static inline int is_inode_flag_set(struct inode *inode, int flag) 2843 { 2844 return test_bit(flag, F2FS_I(inode)->flags); 2845 } 2846 2847 static inline void clear_inode_flag(struct inode *inode, int flag) 2848 { 2849 clear_bit(flag, F2FS_I(inode)->flags); 2850 __mark_inode_dirty_flag(inode, flag, false); 2851 } 2852 2853 static inline bool f2fs_verity_in_progress(struct inode *inode) 2854 { 2855 return IS_ENABLED(CONFIG_FS_VERITY) && 2856 is_inode_flag_set(inode, FI_VERITY_IN_PROGRESS); 2857 } 2858 2859 static inline void set_acl_inode(struct inode *inode, umode_t mode) 2860 { 2861 F2FS_I(inode)->i_acl_mode = mode; 2862 set_inode_flag(inode, FI_ACL_MODE); 2863 f2fs_mark_inode_dirty_sync(inode, false); 2864 } 2865 2866 static inline void f2fs_i_links_write(struct inode *inode, bool inc) 2867 { 2868 if (inc) 2869 inc_nlink(inode); 2870 else 2871 drop_nlink(inode); 2872 f2fs_mark_inode_dirty_sync(inode, true); 2873 } 2874 2875 static inline void f2fs_i_blocks_write(struct inode *inode, 2876 block_t diff, bool add, bool claim) 2877 { 2878 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE); 2879 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER); 2880 2881 /* add = 1, claim = 1 should be dquot_reserve_block in pair */ 2882 if (add) { 2883 if (claim) 2884 dquot_claim_block(inode, diff); 2885 else 2886 dquot_alloc_block_nofail(inode, diff); 2887 } else { 2888 dquot_free_block(inode, diff); 2889 } 2890 2891 f2fs_mark_inode_dirty_sync(inode, true); 2892 if (clean || recover) 2893 set_inode_flag(inode, FI_AUTO_RECOVER); 2894 } 2895 2896 static inline void f2fs_i_size_write(struct inode *inode, loff_t i_size) 2897 { 2898 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE); 2899 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER); 2900 2901 if (i_size_read(inode) == i_size) 2902 return; 2903 2904 i_size_write(inode, i_size); 2905 f2fs_mark_inode_dirty_sync(inode, true); 2906 if (clean || recover) 2907 set_inode_flag(inode, FI_AUTO_RECOVER); 2908 } 2909 2910 static inline void f2fs_i_depth_write(struct inode *inode, unsigned int depth) 2911 { 2912 F2FS_I(inode)->i_current_depth = depth; 2913 f2fs_mark_inode_dirty_sync(inode, true); 2914 } 2915 2916 static inline void f2fs_i_gc_failures_write(struct inode *inode, 2917 unsigned int count) 2918 { 2919 F2FS_I(inode)->i_gc_failures[GC_FAILURE_PIN] = count; 2920 f2fs_mark_inode_dirty_sync(inode, true); 2921 } 2922 2923 static inline void f2fs_i_xnid_write(struct inode *inode, nid_t xnid) 2924 { 2925 F2FS_I(inode)->i_xattr_nid = xnid; 2926 f2fs_mark_inode_dirty_sync(inode, true); 2927 } 2928 2929 static inline void f2fs_i_pino_write(struct inode *inode, nid_t pino) 2930 { 2931 F2FS_I(inode)->i_pino = pino; 2932 f2fs_mark_inode_dirty_sync(inode, true); 2933 } 2934 2935 static inline void get_inline_info(struct inode *inode, struct f2fs_inode *ri) 2936 { 2937 struct f2fs_inode_info *fi = F2FS_I(inode); 2938 2939 if (ri->i_inline & F2FS_INLINE_XATTR) 2940 set_bit(FI_INLINE_XATTR, fi->flags); 2941 if (ri->i_inline & F2FS_INLINE_DATA) 2942 set_bit(FI_INLINE_DATA, fi->flags); 2943 if (ri->i_inline & F2FS_INLINE_DENTRY) 2944 set_bit(FI_INLINE_DENTRY, fi->flags); 2945 if (ri->i_inline & F2FS_DATA_EXIST) 2946 set_bit(FI_DATA_EXIST, fi->flags); 2947 if (ri->i_inline & F2FS_INLINE_DOTS) 2948 set_bit(FI_INLINE_DOTS, fi->flags); 2949 if (ri->i_inline & F2FS_EXTRA_ATTR) 2950 set_bit(FI_EXTRA_ATTR, fi->flags); 2951 if (ri->i_inline & F2FS_PIN_FILE) 2952 set_bit(FI_PIN_FILE, fi->flags); 2953 if (ri->i_inline & F2FS_COMPRESS_RELEASED) 2954 set_bit(FI_COMPRESS_RELEASED, fi->flags); 2955 } 2956 2957 static inline void set_raw_inline(struct inode *inode, struct f2fs_inode *ri) 2958 { 2959 ri->i_inline = 0; 2960 2961 if (is_inode_flag_set(inode, FI_INLINE_XATTR)) 2962 ri->i_inline |= F2FS_INLINE_XATTR; 2963 if (is_inode_flag_set(inode, FI_INLINE_DATA)) 2964 ri->i_inline |= F2FS_INLINE_DATA; 2965 if (is_inode_flag_set(inode, FI_INLINE_DENTRY)) 2966 ri->i_inline |= F2FS_INLINE_DENTRY; 2967 if (is_inode_flag_set(inode, FI_DATA_EXIST)) 2968 ri->i_inline |= F2FS_DATA_EXIST; 2969 if (is_inode_flag_set(inode, FI_INLINE_DOTS)) 2970 ri->i_inline |= F2FS_INLINE_DOTS; 2971 if (is_inode_flag_set(inode, FI_EXTRA_ATTR)) 2972 ri->i_inline |= F2FS_EXTRA_ATTR; 2973 if (is_inode_flag_set(inode, FI_PIN_FILE)) 2974 ri->i_inline |= F2FS_PIN_FILE; 2975 if (is_inode_flag_set(inode, FI_COMPRESS_RELEASED)) 2976 ri->i_inline |= F2FS_COMPRESS_RELEASED; 2977 } 2978 2979 static inline int f2fs_has_extra_attr(struct inode *inode) 2980 { 2981 return is_inode_flag_set(inode, FI_EXTRA_ATTR); 2982 } 2983 2984 static inline int f2fs_has_inline_xattr(struct inode *inode) 2985 { 2986 return is_inode_flag_set(inode, FI_INLINE_XATTR); 2987 } 2988 2989 static inline int f2fs_compressed_file(struct inode *inode) 2990 { 2991 return S_ISREG(inode->i_mode) && 2992 is_inode_flag_set(inode, FI_COMPRESSED_FILE); 2993 } 2994 2995 static inline bool f2fs_need_compress_data(struct inode *inode) 2996 { 2997 int compress_mode = F2FS_OPTION(F2FS_I_SB(inode)).compress_mode; 2998 2999 if (!f2fs_compressed_file(inode)) 3000 return false; 3001 3002 if (compress_mode == COMPR_MODE_FS) 3003 return true; 3004 else if (compress_mode == COMPR_MODE_USER && 3005 is_inode_flag_set(inode, FI_ENABLE_COMPRESS)) 3006 return true; 3007 3008 return false; 3009 } 3010 3011 static inline unsigned int addrs_per_inode(struct inode *inode) 3012 { 3013 unsigned int addrs = CUR_ADDRS_PER_INODE(inode) - 3014 get_inline_xattr_addrs(inode); 3015 3016 if (!f2fs_compressed_file(inode)) 3017 return addrs; 3018 return ALIGN_DOWN(addrs, F2FS_I(inode)->i_cluster_size); 3019 } 3020 3021 static inline unsigned int addrs_per_block(struct inode *inode) 3022 { 3023 if (!f2fs_compressed_file(inode)) 3024 return DEF_ADDRS_PER_BLOCK; 3025 return ALIGN_DOWN(DEF_ADDRS_PER_BLOCK, F2FS_I(inode)->i_cluster_size); 3026 } 3027 3028 static inline void *inline_xattr_addr(struct inode *inode, struct page *page) 3029 { 3030 struct f2fs_inode *ri = F2FS_INODE(page); 3031 3032 return (void *)&(ri->i_addr[DEF_ADDRS_PER_INODE - 3033 get_inline_xattr_addrs(inode)]); 3034 } 3035 3036 static inline int inline_xattr_size(struct inode *inode) 3037 { 3038 if (f2fs_has_inline_xattr(inode)) 3039 return get_inline_xattr_addrs(inode) * sizeof(__le32); 3040 return 0; 3041 } 3042 3043 static inline int f2fs_has_inline_data(struct inode *inode) 3044 { 3045 return is_inode_flag_set(inode, FI_INLINE_DATA); 3046 } 3047 3048 static inline int f2fs_exist_data(struct inode *inode) 3049 { 3050 return is_inode_flag_set(inode, FI_DATA_EXIST); 3051 } 3052 3053 static inline int f2fs_has_inline_dots(struct inode *inode) 3054 { 3055 return is_inode_flag_set(inode, FI_INLINE_DOTS); 3056 } 3057 3058 static inline int f2fs_is_mmap_file(struct inode *inode) 3059 { 3060 return is_inode_flag_set(inode, FI_MMAP_FILE); 3061 } 3062 3063 static inline bool f2fs_is_pinned_file(struct inode *inode) 3064 { 3065 return is_inode_flag_set(inode, FI_PIN_FILE); 3066 } 3067 3068 static inline bool f2fs_is_atomic_file(struct inode *inode) 3069 { 3070 return is_inode_flag_set(inode, FI_ATOMIC_FILE); 3071 } 3072 3073 static inline bool f2fs_is_commit_atomic_write(struct inode *inode) 3074 { 3075 return is_inode_flag_set(inode, FI_ATOMIC_COMMIT); 3076 } 3077 3078 static inline bool f2fs_is_volatile_file(struct inode *inode) 3079 { 3080 return is_inode_flag_set(inode, FI_VOLATILE_FILE); 3081 } 3082 3083 static inline bool f2fs_is_first_block_written(struct inode *inode) 3084 { 3085 return is_inode_flag_set(inode, FI_FIRST_BLOCK_WRITTEN); 3086 } 3087 3088 static inline bool f2fs_is_drop_cache(struct inode *inode) 3089 { 3090 return is_inode_flag_set(inode, FI_DROP_CACHE); 3091 } 3092 3093 static inline void *inline_data_addr(struct inode *inode, struct page *page) 3094 { 3095 struct f2fs_inode *ri = F2FS_INODE(page); 3096 int extra_size = get_extra_isize(inode); 3097 3098 return (void *)&(ri->i_addr[extra_size + DEF_INLINE_RESERVED_SIZE]); 3099 } 3100 3101 static inline int f2fs_has_inline_dentry(struct inode *inode) 3102 { 3103 return is_inode_flag_set(inode, FI_INLINE_DENTRY); 3104 } 3105 3106 static inline int is_file(struct inode *inode, int type) 3107 { 3108 return F2FS_I(inode)->i_advise & type; 3109 } 3110 3111 static inline void set_file(struct inode *inode, int type) 3112 { 3113 F2FS_I(inode)->i_advise |= type; 3114 f2fs_mark_inode_dirty_sync(inode, true); 3115 } 3116 3117 static inline void clear_file(struct inode *inode, int type) 3118 { 3119 F2FS_I(inode)->i_advise &= ~type; 3120 f2fs_mark_inode_dirty_sync(inode, true); 3121 } 3122 3123 static inline bool f2fs_is_time_consistent(struct inode *inode) 3124 { 3125 if (!timespec64_equal(F2FS_I(inode)->i_disk_time, &inode->i_atime)) 3126 return false; 3127 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 1, &inode->i_ctime)) 3128 return false; 3129 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 2, &inode->i_mtime)) 3130 return false; 3131 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 3, 3132 &F2FS_I(inode)->i_crtime)) 3133 return false; 3134 return true; 3135 } 3136 3137 static inline bool f2fs_skip_inode_update(struct inode *inode, int dsync) 3138 { 3139 bool ret; 3140 3141 if (dsync) { 3142 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 3143 3144 spin_lock(&sbi->inode_lock[DIRTY_META]); 3145 ret = list_empty(&F2FS_I(inode)->gdirty_list); 3146 spin_unlock(&sbi->inode_lock[DIRTY_META]); 3147 return ret; 3148 } 3149 if (!is_inode_flag_set(inode, FI_AUTO_RECOVER) || 3150 file_keep_isize(inode) || 3151 i_size_read(inode) & ~PAGE_MASK) 3152 return false; 3153 3154 if (!f2fs_is_time_consistent(inode)) 3155 return false; 3156 3157 spin_lock(&F2FS_I(inode)->i_size_lock); 3158 ret = F2FS_I(inode)->last_disk_size == i_size_read(inode); 3159 spin_unlock(&F2FS_I(inode)->i_size_lock); 3160 3161 return ret; 3162 } 3163 3164 static inline bool f2fs_readonly(struct super_block *sb) 3165 { 3166 return sb_rdonly(sb); 3167 } 3168 3169 static inline bool f2fs_cp_error(struct f2fs_sb_info *sbi) 3170 { 3171 return is_set_ckpt_flags(sbi, CP_ERROR_FLAG); 3172 } 3173 3174 static inline bool is_dot_dotdot(const u8 *name, size_t len) 3175 { 3176 if (len == 1 && name[0] == '.') 3177 return true; 3178 3179 if (len == 2 && name[0] == '.' && name[1] == '.') 3180 return true; 3181 3182 return false; 3183 } 3184 3185 static inline void *f2fs_kmalloc(struct f2fs_sb_info *sbi, 3186 size_t size, gfp_t flags) 3187 { 3188 if (time_to_inject(sbi, FAULT_KMALLOC)) { 3189 f2fs_show_injection_info(sbi, FAULT_KMALLOC); 3190 return NULL; 3191 } 3192 3193 return kmalloc(size, flags); 3194 } 3195 3196 static inline void *f2fs_kzalloc(struct f2fs_sb_info *sbi, 3197 size_t size, gfp_t flags) 3198 { 3199 return f2fs_kmalloc(sbi, size, flags | __GFP_ZERO); 3200 } 3201 3202 static inline void *f2fs_kvmalloc(struct f2fs_sb_info *sbi, 3203 size_t size, gfp_t flags) 3204 { 3205 if (time_to_inject(sbi, FAULT_KVMALLOC)) { 3206 f2fs_show_injection_info(sbi, FAULT_KVMALLOC); 3207 return NULL; 3208 } 3209 3210 return kvmalloc(size, flags); 3211 } 3212 3213 static inline void *f2fs_kvzalloc(struct f2fs_sb_info *sbi, 3214 size_t size, gfp_t flags) 3215 { 3216 return f2fs_kvmalloc(sbi, size, flags | __GFP_ZERO); 3217 } 3218 3219 static inline int get_extra_isize(struct inode *inode) 3220 { 3221 return F2FS_I(inode)->i_extra_isize / sizeof(__le32); 3222 } 3223 3224 static inline int get_inline_xattr_addrs(struct inode *inode) 3225 { 3226 return F2FS_I(inode)->i_inline_xattr_size; 3227 } 3228 3229 #define f2fs_get_inode_mode(i) \ 3230 ((is_inode_flag_set(i, FI_ACL_MODE)) ? \ 3231 (F2FS_I(i)->i_acl_mode) : ((i)->i_mode)) 3232 3233 #define F2FS_TOTAL_EXTRA_ATTR_SIZE \ 3234 (offsetof(struct f2fs_inode, i_extra_end) - \ 3235 offsetof(struct f2fs_inode, i_extra_isize)) \ 3236 3237 #define F2FS_OLD_ATTRIBUTE_SIZE (offsetof(struct f2fs_inode, i_addr)) 3238 #define F2FS_FITS_IN_INODE(f2fs_inode, extra_isize, field) \ 3239 ((offsetof(typeof(*(f2fs_inode)), field) + \ 3240 sizeof((f2fs_inode)->field)) \ 3241 <= (F2FS_OLD_ATTRIBUTE_SIZE + (extra_isize))) \ 3242 3243 #define __is_large_section(sbi) ((sbi)->segs_per_sec > 1) 3244 3245 #define __is_meta_io(fio) (PAGE_TYPE_OF_BIO((fio)->type) == META) 3246 3247 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi, 3248 block_t blkaddr, int type); 3249 static inline void verify_blkaddr(struct f2fs_sb_info *sbi, 3250 block_t blkaddr, int type) 3251 { 3252 if (!f2fs_is_valid_blkaddr(sbi, blkaddr, type)) { 3253 f2fs_err(sbi, "invalid blkaddr: %u, type: %d, run fsck to fix.", 3254 blkaddr, type); 3255 f2fs_bug_on(sbi, 1); 3256 } 3257 } 3258 3259 static inline bool __is_valid_data_blkaddr(block_t blkaddr) 3260 { 3261 if (blkaddr == NEW_ADDR || blkaddr == NULL_ADDR || 3262 blkaddr == COMPRESS_ADDR) 3263 return false; 3264 return true; 3265 } 3266 3267 /* 3268 * file.c 3269 */ 3270 int f2fs_sync_file(struct file *file, loff_t start, loff_t end, int datasync); 3271 void f2fs_truncate_data_blocks(struct dnode_of_data *dn); 3272 int f2fs_do_truncate_blocks(struct inode *inode, u64 from, bool lock); 3273 int f2fs_truncate_blocks(struct inode *inode, u64 from, bool lock); 3274 int f2fs_truncate(struct inode *inode); 3275 int f2fs_getattr(struct user_namespace *mnt_userns, const struct path *path, 3276 struct kstat *stat, u32 request_mask, unsigned int flags); 3277 int f2fs_setattr(struct user_namespace *mnt_userns, struct dentry *dentry, 3278 struct iattr *attr); 3279 int f2fs_truncate_hole(struct inode *inode, pgoff_t pg_start, pgoff_t pg_end); 3280 void f2fs_truncate_data_blocks_range(struct dnode_of_data *dn, int count); 3281 int f2fs_precache_extents(struct inode *inode); 3282 int f2fs_fileattr_get(struct dentry *dentry, struct fileattr *fa); 3283 int f2fs_fileattr_set(struct user_namespace *mnt_userns, 3284 struct dentry *dentry, struct fileattr *fa); 3285 long f2fs_ioctl(struct file *filp, unsigned int cmd, unsigned long arg); 3286 long f2fs_compat_ioctl(struct file *file, unsigned int cmd, unsigned long arg); 3287 int f2fs_transfer_project_quota(struct inode *inode, kprojid_t kprojid); 3288 int f2fs_pin_file_control(struct inode *inode, bool inc); 3289 3290 /* 3291 * inode.c 3292 */ 3293 void f2fs_set_inode_flags(struct inode *inode); 3294 bool f2fs_inode_chksum_verify(struct f2fs_sb_info *sbi, struct page *page); 3295 void f2fs_inode_chksum_set(struct f2fs_sb_info *sbi, struct page *page); 3296 struct inode *f2fs_iget(struct super_block *sb, unsigned long ino); 3297 struct inode *f2fs_iget_retry(struct super_block *sb, unsigned long ino); 3298 int f2fs_try_to_free_nats(struct f2fs_sb_info *sbi, int nr_shrink); 3299 void f2fs_update_inode(struct inode *inode, struct page *node_page); 3300 void f2fs_update_inode_page(struct inode *inode); 3301 int f2fs_write_inode(struct inode *inode, struct writeback_control *wbc); 3302 void f2fs_evict_inode(struct inode *inode); 3303 void f2fs_handle_failed_inode(struct inode *inode); 3304 3305 /* 3306 * namei.c 3307 */ 3308 int f2fs_update_extension_list(struct f2fs_sb_info *sbi, const char *name, 3309 bool hot, bool set); 3310 struct dentry *f2fs_get_parent(struct dentry *child); 3311 3312 /* 3313 * dir.c 3314 */ 3315 unsigned char f2fs_get_de_type(struct f2fs_dir_entry *de); 3316 int f2fs_init_casefolded_name(const struct inode *dir, 3317 struct f2fs_filename *fname); 3318 int f2fs_setup_filename(struct inode *dir, const struct qstr *iname, 3319 int lookup, struct f2fs_filename *fname); 3320 int f2fs_prepare_lookup(struct inode *dir, struct dentry *dentry, 3321 struct f2fs_filename *fname); 3322 void f2fs_free_filename(struct f2fs_filename *fname); 3323 struct f2fs_dir_entry *f2fs_find_target_dentry(const struct f2fs_dentry_ptr *d, 3324 const struct f2fs_filename *fname, int *max_slots); 3325 int f2fs_fill_dentries(struct dir_context *ctx, struct f2fs_dentry_ptr *d, 3326 unsigned int start_pos, struct fscrypt_str *fstr); 3327 void f2fs_do_make_empty_dir(struct inode *inode, struct inode *parent, 3328 struct f2fs_dentry_ptr *d); 3329 struct page *f2fs_init_inode_metadata(struct inode *inode, struct inode *dir, 3330 const struct f2fs_filename *fname, struct page *dpage); 3331 void f2fs_update_parent_metadata(struct inode *dir, struct inode *inode, 3332 unsigned int current_depth); 3333 int f2fs_room_for_filename(const void *bitmap, int slots, int max_slots); 3334 void f2fs_drop_nlink(struct inode *dir, struct inode *inode); 3335 struct f2fs_dir_entry *__f2fs_find_entry(struct inode *dir, 3336 const struct f2fs_filename *fname, 3337 struct page **res_page); 3338 struct f2fs_dir_entry *f2fs_find_entry(struct inode *dir, 3339 const struct qstr *child, struct page **res_page); 3340 struct f2fs_dir_entry *f2fs_parent_dir(struct inode *dir, struct page **p); 3341 ino_t f2fs_inode_by_name(struct inode *dir, const struct qstr *qstr, 3342 struct page **page); 3343 void f2fs_set_link(struct inode *dir, struct f2fs_dir_entry *de, 3344 struct page *page, struct inode *inode); 3345 bool f2fs_has_enough_room(struct inode *dir, struct page *ipage, 3346 const struct f2fs_filename *fname); 3347 void f2fs_update_dentry(nid_t ino, umode_t mode, struct f2fs_dentry_ptr *d, 3348 const struct fscrypt_str *name, f2fs_hash_t name_hash, 3349 unsigned int bit_pos); 3350 int f2fs_add_regular_entry(struct inode *dir, const struct f2fs_filename *fname, 3351 struct inode *inode, nid_t ino, umode_t mode); 3352 int f2fs_add_dentry(struct inode *dir, const struct f2fs_filename *fname, 3353 struct inode *inode, nid_t ino, umode_t mode); 3354 int f2fs_do_add_link(struct inode *dir, const struct qstr *name, 3355 struct inode *inode, nid_t ino, umode_t mode); 3356 void f2fs_delete_entry(struct f2fs_dir_entry *dentry, struct page *page, 3357 struct inode *dir, struct inode *inode); 3358 int f2fs_do_tmpfile(struct inode *inode, struct inode *dir); 3359 bool f2fs_empty_dir(struct inode *dir); 3360 3361 static inline int f2fs_add_link(struct dentry *dentry, struct inode *inode) 3362 { 3363 if (fscrypt_is_nokey_name(dentry)) 3364 return -ENOKEY; 3365 return f2fs_do_add_link(d_inode(dentry->d_parent), &dentry->d_name, 3366 inode, inode->i_ino, inode->i_mode); 3367 } 3368 3369 /* 3370 * super.c 3371 */ 3372 int f2fs_inode_dirtied(struct inode *inode, bool sync); 3373 void f2fs_inode_synced(struct inode *inode); 3374 int f2fs_enable_quota_files(struct f2fs_sb_info *sbi, bool rdonly); 3375 int f2fs_quota_sync(struct super_block *sb, int type); 3376 loff_t max_file_blocks(struct inode *inode); 3377 void f2fs_quota_off_umount(struct super_block *sb); 3378 int f2fs_commit_super(struct f2fs_sb_info *sbi, bool recover); 3379 int f2fs_sync_fs(struct super_block *sb, int sync); 3380 int f2fs_sanity_check_ckpt(struct f2fs_sb_info *sbi); 3381 3382 /* 3383 * hash.c 3384 */ 3385 void f2fs_hash_filename(const struct inode *dir, struct f2fs_filename *fname); 3386 3387 /* 3388 * node.c 3389 */ 3390 struct node_info; 3391 3392 int f2fs_check_nid_range(struct f2fs_sb_info *sbi, nid_t nid); 3393 bool f2fs_available_free_memory(struct f2fs_sb_info *sbi, int type); 3394 bool f2fs_in_warm_node_list(struct f2fs_sb_info *sbi, struct page *page); 3395 void f2fs_init_fsync_node_info(struct f2fs_sb_info *sbi); 3396 void f2fs_del_fsync_node_entry(struct f2fs_sb_info *sbi, struct page *page); 3397 void f2fs_reset_fsync_node_info(struct f2fs_sb_info *sbi); 3398 int f2fs_need_dentry_mark(struct f2fs_sb_info *sbi, nid_t nid); 3399 bool f2fs_is_checkpointed_node(struct f2fs_sb_info *sbi, nid_t nid); 3400 bool f2fs_need_inode_block_update(struct f2fs_sb_info *sbi, nid_t ino); 3401 int f2fs_get_node_info(struct f2fs_sb_info *sbi, nid_t nid, 3402 struct node_info *ni); 3403 pgoff_t f2fs_get_next_page_offset(struct dnode_of_data *dn, pgoff_t pgofs); 3404 int f2fs_get_dnode_of_data(struct dnode_of_data *dn, pgoff_t index, int mode); 3405 int f2fs_truncate_inode_blocks(struct inode *inode, pgoff_t from); 3406 int f2fs_truncate_xattr_node(struct inode *inode); 3407 int f2fs_wait_on_node_pages_writeback(struct f2fs_sb_info *sbi, 3408 unsigned int seq_id); 3409 bool f2fs_nat_bitmap_enabled(struct f2fs_sb_info *sbi); 3410 int f2fs_remove_inode_page(struct inode *inode); 3411 struct page *f2fs_new_inode_page(struct inode *inode); 3412 struct page *f2fs_new_node_page(struct dnode_of_data *dn, unsigned int ofs); 3413 void f2fs_ra_node_page(struct f2fs_sb_info *sbi, nid_t nid); 3414 struct page *f2fs_get_node_page(struct f2fs_sb_info *sbi, pgoff_t nid); 3415 struct page *f2fs_get_node_page_ra(struct page *parent, int start); 3416 int f2fs_move_node_page(struct page *node_page, int gc_type); 3417 void f2fs_flush_inline_data(struct f2fs_sb_info *sbi); 3418 int f2fs_fsync_node_pages(struct f2fs_sb_info *sbi, struct inode *inode, 3419 struct writeback_control *wbc, bool atomic, 3420 unsigned int *seq_id); 3421 int f2fs_sync_node_pages(struct f2fs_sb_info *sbi, 3422 struct writeback_control *wbc, 3423 bool do_balance, enum iostat_type io_type); 3424 int f2fs_build_free_nids(struct f2fs_sb_info *sbi, bool sync, bool mount); 3425 bool f2fs_alloc_nid(struct f2fs_sb_info *sbi, nid_t *nid); 3426 void f2fs_alloc_nid_done(struct f2fs_sb_info *sbi, nid_t nid); 3427 void f2fs_alloc_nid_failed(struct f2fs_sb_info *sbi, nid_t nid); 3428 int f2fs_try_to_free_nids(struct f2fs_sb_info *sbi, int nr_shrink); 3429 int f2fs_recover_inline_xattr(struct inode *inode, struct page *page); 3430 int f2fs_recover_xattr_data(struct inode *inode, struct page *page); 3431 int f2fs_recover_inode_page(struct f2fs_sb_info *sbi, struct page *page); 3432 int f2fs_restore_node_summary(struct f2fs_sb_info *sbi, 3433 unsigned int segno, struct f2fs_summary_block *sum); 3434 void f2fs_enable_nat_bits(struct f2fs_sb_info *sbi); 3435 int f2fs_flush_nat_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3436 int f2fs_build_node_manager(struct f2fs_sb_info *sbi); 3437 void f2fs_destroy_node_manager(struct f2fs_sb_info *sbi); 3438 int __init f2fs_create_node_manager_caches(void); 3439 void f2fs_destroy_node_manager_caches(void); 3440 3441 /* 3442 * segment.c 3443 */ 3444 bool f2fs_need_SSR(struct f2fs_sb_info *sbi); 3445 void f2fs_register_inmem_page(struct inode *inode, struct page *page); 3446 void f2fs_drop_inmem_pages_all(struct f2fs_sb_info *sbi, bool gc_failure); 3447 void f2fs_drop_inmem_pages(struct inode *inode); 3448 void f2fs_drop_inmem_page(struct inode *inode, struct page *page); 3449 int f2fs_commit_inmem_pages(struct inode *inode); 3450 void f2fs_balance_fs(struct f2fs_sb_info *sbi, bool need); 3451 void f2fs_balance_fs_bg(struct f2fs_sb_info *sbi, bool from_bg); 3452 int f2fs_issue_flush(struct f2fs_sb_info *sbi, nid_t ino); 3453 int f2fs_create_flush_cmd_control(struct f2fs_sb_info *sbi); 3454 int f2fs_flush_device_cache(struct f2fs_sb_info *sbi); 3455 void f2fs_destroy_flush_cmd_control(struct f2fs_sb_info *sbi, bool free); 3456 void f2fs_invalidate_blocks(struct f2fs_sb_info *sbi, block_t addr); 3457 bool f2fs_is_checkpointed_data(struct f2fs_sb_info *sbi, block_t blkaddr); 3458 int f2fs_start_discard_thread(struct f2fs_sb_info *sbi); 3459 void f2fs_drop_discard_cmd(struct f2fs_sb_info *sbi); 3460 void f2fs_stop_discard_thread(struct f2fs_sb_info *sbi); 3461 bool f2fs_issue_discard_timeout(struct f2fs_sb_info *sbi); 3462 void f2fs_clear_prefree_segments(struct f2fs_sb_info *sbi, 3463 struct cp_control *cpc); 3464 void f2fs_dirty_to_prefree(struct f2fs_sb_info *sbi); 3465 block_t f2fs_get_unusable_blocks(struct f2fs_sb_info *sbi); 3466 int f2fs_disable_cp_again(struct f2fs_sb_info *sbi, block_t unusable); 3467 void f2fs_release_discard_addrs(struct f2fs_sb_info *sbi); 3468 int f2fs_npages_for_summary_flush(struct f2fs_sb_info *sbi, bool for_ra); 3469 bool f2fs_segment_has_free_slot(struct f2fs_sb_info *sbi, int segno); 3470 void f2fs_init_inmem_curseg(struct f2fs_sb_info *sbi); 3471 void f2fs_save_inmem_curseg(struct f2fs_sb_info *sbi); 3472 void f2fs_restore_inmem_curseg(struct f2fs_sb_info *sbi); 3473 void f2fs_get_new_segment(struct f2fs_sb_info *sbi, 3474 unsigned int *newseg, bool new_sec, int dir); 3475 void f2fs_allocate_segment_for_resize(struct f2fs_sb_info *sbi, int type, 3476 unsigned int start, unsigned int end); 3477 void f2fs_allocate_new_section(struct f2fs_sb_info *sbi, int type, bool force); 3478 void f2fs_allocate_new_segments(struct f2fs_sb_info *sbi); 3479 int f2fs_trim_fs(struct f2fs_sb_info *sbi, struct fstrim_range *range); 3480 bool f2fs_exist_trim_candidates(struct f2fs_sb_info *sbi, 3481 struct cp_control *cpc); 3482 struct page *f2fs_get_sum_page(struct f2fs_sb_info *sbi, unsigned int segno); 3483 void f2fs_update_meta_page(struct f2fs_sb_info *sbi, void *src, 3484 block_t blk_addr); 3485 void f2fs_do_write_meta_page(struct f2fs_sb_info *sbi, struct page *page, 3486 enum iostat_type io_type); 3487 void f2fs_do_write_node_page(unsigned int nid, struct f2fs_io_info *fio); 3488 void f2fs_outplace_write_data(struct dnode_of_data *dn, 3489 struct f2fs_io_info *fio); 3490 int f2fs_inplace_write_data(struct f2fs_io_info *fio); 3491 void f2fs_do_replace_block(struct f2fs_sb_info *sbi, struct f2fs_summary *sum, 3492 block_t old_blkaddr, block_t new_blkaddr, 3493 bool recover_curseg, bool recover_newaddr, 3494 bool from_gc); 3495 void f2fs_replace_block(struct f2fs_sb_info *sbi, struct dnode_of_data *dn, 3496 block_t old_addr, block_t new_addr, 3497 unsigned char version, bool recover_curseg, 3498 bool recover_newaddr); 3499 void f2fs_allocate_data_block(struct f2fs_sb_info *sbi, struct page *page, 3500 block_t old_blkaddr, block_t *new_blkaddr, 3501 struct f2fs_summary *sum, int type, 3502 struct f2fs_io_info *fio); 3503 void f2fs_wait_on_page_writeback(struct page *page, 3504 enum page_type type, bool ordered, bool locked); 3505 void f2fs_wait_on_block_writeback(struct inode *inode, block_t blkaddr); 3506 void f2fs_wait_on_block_writeback_range(struct inode *inode, block_t blkaddr, 3507 block_t len); 3508 void f2fs_write_data_summaries(struct f2fs_sb_info *sbi, block_t start_blk); 3509 void f2fs_write_node_summaries(struct f2fs_sb_info *sbi, block_t start_blk); 3510 int f2fs_lookup_journal_in_cursum(struct f2fs_journal *journal, int type, 3511 unsigned int val, int alloc); 3512 void f2fs_flush_sit_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3513 int f2fs_fix_curseg_write_pointer(struct f2fs_sb_info *sbi); 3514 int f2fs_check_write_pointer(struct f2fs_sb_info *sbi); 3515 int f2fs_build_segment_manager(struct f2fs_sb_info *sbi); 3516 void f2fs_destroy_segment_manager(struct f2fs_sb_info *sbi); 3517 int __init f2fs_create_segment_manager_caches(void); 3518 void f2fs_destroy_segment_manager_caches(void); 3519 int f2fs_rw_hint_to_seg_type(enum rw_hint hint); 3520 enum rw_hint f2fs_io_type_to_rw_hint(struct f2fs_sb_info *sbi, 3521 enum page_type type, enum temp_type temp); 3522 unsigned int f2fs_usable_segs_in_sec(struct f2fs_sb_info *sbi, 3523 unsigned int segno); 3524 unsigned int f2fs_usable_blks_in_seg(struct f2fs_sb_info *sbi, 3525 unsigned int segno); 3526 3527 #define DEF_FRAGMENT_SIZE 4 3528 #define MIN_FRAGMENT_SIZE 1 3529 #define MAX_FRAGMENT_SIZE 512 3530 3531 static inline bool f2fs_need_rand_seg(struct f2fs_sb_info *sbi) 3532 { 3533 return F2FS_OPTION(sbi).fs_mode == FS_MODE_FRAGMENT_SEG || 3534 F2FS_OPTION(sbi).fs_mode == FS_MODE_FRAGMENT_BLK; 3535 } 3536 3537 /* 3538 * checkpoint.c 3539 */ 3540 void f2fs_stop_checkpoint(struct f2fs_sb_info *sbi, bool end_io); 3541 struct page *f2fs_grab_meta_page(struct f2fs_sb_info *sbi, pgoff_t index); 3542 struct page *f2fs_get_meta_page(struct f2fs_sb_info *sbi, pgoff_t index); 3543 struct page *f2fs_get_meta_page_retry(struct f2fs_sb_info *sbi, pgoff_t index); 3544 struct page *f2fs_get_tmp_page(struct f2fs_sb_info *sbi, pgoff_t index); 3545 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi, 3546 block_t blkaddr, int type); 3547 int f2fs_ra_meta_pages(struct f2fs_sb_info *sbi, block_t start, int nrpages, 3548 int type, bool sync); 3549 void f2fs_ra_meta_pages_cond(struct f2fs_sb_info *sbi, pgoff_t index); 3550 long f2fs_sync_meta_pages(struct f2fs_sb_info *sbi, enum page_type type, 3551 long nr_to_write, enum iostat_type io_type); 3552 void f2fs_add_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type); 3553 void f2fs_remove_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type); 3554 void f2fs_release_ino_entry(struct f2fs_sb_info *sbi, bool all); 3555 bool f2fs_exist_written_data(struct f2fs_sb_info *sbi, nid_t ino, int mode); 3556 void f2fs_set_dirty_device(struct f2fs_sb_info *sbi, nid_t ino, 3557 unsigned int devidx, int type); 3558 bool f2fs_is_dirty_device(struct f2fs_sb_info *sbi, nid_t ino, 3559 unsigned int devidx, int type); 3560 int f2fs_sync_inode_meta(struct f2fs_sb_info *sbi); 3561 int f2fs_acquire_orphan_inode(struct f2fs_sb_info *sbi); 3562 void f2fs_release_orphan_inode(struct f2fs_sb_info *sbi); 3563 void f2fs_add_orphan_inode(struct inode *inode); 3564 void f2fs_remove_orphan_inode(struct f2fs_sb_info *sbi, nid_t ino); 3565 int f2fs_recover_orphan_inodes(struct f2fs_sb_info *sbi); 3566 int f2fs_get_valid_checkpoint(struct f2fs_sb_info *sbi); 3567 void f2fs_update_dirty_page(struct inode *inode, struct page *page); 3568 void f2fs_remove_dirty_inode(struct inode *inode); 3569 int f2fs_sync_dirty_inodes(struct f2fs_sb_info *sbi, enum inode_type type); 3570 void f2fs_wait_on_all_pages(struct f2fs_sb_info *sbi, int type); 3571 u64 f2fs_get_sectors_written(struct f2fs_sb_info *sbi); 3572 int f2fs_write_checkpoint(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3573 void f2fs_init_ino_entry_info(struct f2fs_sb_info *sbi); 3574 int __init f2fs_create_checkpoint_caches(void); 3575 void f2fs_destroy_checkpoint_caches(void); 3576 int f2fs_issue_checkpoint(struct f2fs_sb_info *sbi); 3577 int f2fs_start_ckpt_thread(struct f2fs_sb_info *sbi); 3578 void f2fs_stop_ckpt_thread(struct f2fs_sb_info *sbi); 3579 void f2fs_init_ckpt_req_control(struct f2fs_sb_info *sbi); 3580 3581 /* 3582 * data.c 3583 */ 3584 int __init f2fs_init_bioset(void); 3585 void f2fs_destroy_bioset(void); 3586 int f2fs_init_bio_entry_cache(void); 3587 void f2fs_destroy_bio_entry_cache(void); 3588 void f2fs_submit_bio(struct f2fs_sb_info *sbi, 3589 struct bio *bio, enum page_type type); 3590 void f2fs_submit_merged_write(struct f2fs_sb_info *sbi, enum page_type type); 3591 void f2fs_submit_merged_write_cond(struct f2fs_sb_info *sbi, 3592 struct inode *inode, struct page *page, 3593 nid_t ino, enum page_type type); 3594 void f2fs_submit_merged_ipu_write(struct f2fs_sb_info *sbi, 3595 struct bio **bio, struct page *page); 3596 void f2fs_flush_merged_writes(struct f2fs_sb_info *sbi); 3597 int f2fs_submit_page_bio(struct f2fs_io_info *fio); 3598 int f2fs_merge_page_bio(struct f2fs_io_info *fio); 3599 void f2fs_submit_page_write(struct f2fs_io_info *fio); 3600 struct block_device *f2fs_target_device(struct f2fs_sb_info *sbi, 3601 block_t blk_addr, struct bio *bio); 3602 int f2fs_target_device_index(struct f2fs_sb_info *sbi, block_t blkaddr); 3603 void f2fs_set_data_blkaddr(struct dnode_of_data *dn); 3604 void f2fs_update_data_blkaddr(struct dnode_of_data *dn, block_t blkaddr); 3605 int f2fs_reserve_new_blocks(struct dnode_of_data *dn, blkcnt_t count); 3606 int f2fs_reserve_new_block(struct dnode_of_data *dn); 3607 int f2fs_get_block(struct dnode_of_data *dn, pgoff_t index); 3608 int f2fs_preallocate_blocks(struct kiocb *iocb, struct iov_iter *from); 3609 int f2fs_reserve_block(struct dnode_of_data *dn, pgoff_t index); 3610 struct page *f2fs_get_read_data_page(struct inode *inode, pgoff_t index, 3611 int op_flags, bool for_write); 3612 struct page *f2fs_find_data_page(struct inode *inode, pgoff_t index); 3613 struct page *f2fs_get_lock_data_page(struct inode *inode, pgoff_t index, 3614 bool for_write); 3615 struct page *f2fs_get_new_data_page(struct inode *inode, 3616 struct page *ipage, pgoff_t index, bool new_i_size); 3617 int f2fs_do_write_data_page(struct f2fs_io_info *fio); 3618 void f2fs_do_map_lock(struct f2fs_sb_info *sbi, int flag, bool lock); 3619 int f2fs_map_blocks(struct inode *inode, struct f2fs_map_blocks *map, 3620 int create, int flag); 3621 int f2fs_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, 3622 u64 start, u64 len); 3623 int f2fs_encrypt_one_page(struct f2fs_io_info *fio); 3624 bool f2fs_should_update_inplace(struct inode *inode, struct f2fs_io_info *fio); 3625 bool f2fs_should_update_outplace(struct inode *inode, struct f2fs_io_info *fio); 3626 int f2fs_write_single_data_page(struct page *page, int *submitted, 3627 struct bio **bio, sector_t *last_block, 3628 struct writeback_control *wbc, 3629 enum iostat_type io_type, 3630 int compr_blocks, bool allow_balance); 3631 void f2fs_invalidate_page(struct page *page, unsigned int offset, 3632 unsigned int length); 3633 int f2fs_release_page(struct page *page, gfp_t wait); 3634 #ifdef CONFIG_MIGRATION 3635 int f2fs_migrate_page(struct address_space *mapping, struct page *newpage, 3636 struct page *page, enum migrate_mode mode); 3637 #endif 3638 bool f2fs_overwrite_io(struct inode *inode, loff_t pos, size_t len); 3639 void f2fs_clear_page_cache_dirty_tag(struct page *page); 3640 int f2fs_init_post_read_processing(void); 3641 void f2fs_destroy_post_read_processing(void); 3642 int f2fs_init_post_read_wq(struct f2fs_sb_info *sbi); 3643 void f2fs_destroy_post_read_wq(struct f2fs_sb_info *sbi); 3644 3645 /* 3646 * gc.c 3647 */ 3648 int f2fs_start_gc_thread(struct f2fs_sb_info *sbi); 3649 void f2fs_stop_gc_thread(struct f2fs_sb_info *sbi); 3650 block_t f2fs_start_bidx_of_node(unsigned int node_ofs, struct inode *inode); 3651 int f2fs_gc(struct f2fs_sb_info *sbi, bool sync, bool background, bool force, 3652 unsigned int segno); 3653 void f2fs_build_gc_manager(struct f2fs_sb_info *sbi); 3654 int f2fs_resize_fs(struct f2fs_sb_info *sbi, __u64 block_count); 3655 int __init f2fs_create_garbage_collection_cache(void); 3656 void f2fs_destroy_garbage_collection_cache(void); 3657 3658 /* 3659 * recovery.c 3660 */ 3661 int f2fs_recover_fsync_data(struct f2fs_sb_info *sbi, bool check_only); 3662 bool f2fs_space_for_roll_forward(struct f2fs_sb_info *sbi); 3663 int __init f2fs_create_recovery_cache(void); 3664 void f2fs_destroy_recovery_cache(void); 3665 3666 /* 3667 * debug.c 3668 */ 3669 #ifdef CONFIG_F2FS_STAT_FS 3670 struct f2fs_stat_info { 3671 struct list_head stat_list; 3672 struct f2fs_sb_info *sbi; 3673 int all_area_segs, sit_area_segs, nat_area_segs, ssa_area_segs; 3674 int main_area_segs, main_area_sections, main_area_zones; 3675 unsigned long long hit_largest, hit_cached, hit_rbtree; 3676 unsigned long long hit_total, total_ext; 3677 int ext_tree, zombie_tree, ext_node; 3678 int ndirty_node, ndirty_dent, ndirty_meta, ndirty_imeta; 3679 int ndirty_data, ndirty_qdata; 3680 int inmem_pages; 3681 unsigned int ndirty_dirs, ndirty_files, nquota_files, ndirty_all; 3682 int nats, dirty_nats, sits, dirty_sits; 3683 int free_nids, avail_nids, alloc_nids; 3684 int total_count, utilization; 3685 int bg_gc, nr_wb_cp_data, nr_wb_data; 3686 int nr_rd_data, nr_rd_node, nr_rd_meta; 3687 int nr_dio_read, nr_dio_write; 3688 unsigned int io_skip_bggc, other_skip_bggc; 3689 int nr_flushing, nr_flushed, flush_list_empty; 3690 int nr_discarding, nr_discarded; 3691 int nr_discard_cmd; 3692 unsigned int undiscard_blks; 3693 int nr_issued_ckpt, nr_total_ckpt, nr_queued_ckpt; 3694 unsigned int cur_ckpt_time, peak_ckpt_time; 3695 int inline_xattr, inline_inode, inline_dir, append, update, orphans; 3696 int compr_inode; 3697 unsigned long long compr_blocks; 3698 int aw_cnt, max_aw_cnt, vw_cnt, max_vw_cnt; 3699 unsigned int valid_count, valid_node_count, valid_inode_count, discard_blks; 3700 unsigned int bimodal, avg_vblocks; 3701 int util_free, util_valid, util_invalid; 3702 int rsvd_segs, overp_segs; 3703 int dirty_count, node_pages, meta_pages, compress_pages; 3704 int compress_page_hit; 3705 int prefree_count, call_count, cp_count, bg_cp_count; 3706 int tot_segs, node_segs, data_segs, free_segs, free_secs; 3707 int bg_node_segs, bg_data_segs; 3708 int tot_blks, data_blks, node_blks; 3709 int bg_data_blks, bg_node_blks; 3710 unsigned long long skipped_atomic_files[2]; 3711 int curseg[NR_CURSEG_TYPE]; 3712 int cursec[NR_CURSEG_TYPE]; 3713 int curzone[NR_CURSEG_TYPE]; 3714 unsigned int dirty_seg[NR_CURSEG_TYPE]; 3715 unsigned int full_seg[NR_CURSEG_TYPE]; 3716 unsigned int valid_blks[NR_CURSEG_TYPE]; 3717 3718 unsigned int meta_count[META_MAX]; 3719 unsigned int segment_count[2]; 3720 unsigned int block_count[2]; 3721 unsigned int inplace_count; 3722 unsigned long long base_mem, cache_mem, page_mem; 3723 }; 3724 3725 static inline struct f2fs_stat_info *F2FS_STAT(struct f2fs_sb_info *sbi) 3726 { 3727 return (struct f2fs_stat_info *)sbi->stat_info; 3728 } 3729 3730 #define stat_inc_cp_count(si) ((si)->cp_count++) 3731 #define stat_inc_bg_cp_count(si) ((si)->bg_cp_count++) 3732 #define stat_inc_call_count(si) ((si)->call_count++) 3733 #define stat_inc_bggc_count(si) ((si)->bg_gc++) 3734 #define stat_io_skip_bggc_count(sbi) ((sbi)->io_skip_bggc++) 3735 #define stat_other_skip_bggc_count(sbi) ((sbi)->other_skip_bggc++) 3736 #define stat_inc_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]++) 3737 #define stat_dec_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]--) 3738 #define stat_inc_total_hit(sbi) (atomic64_inc(&(sbi)->total_hit_ext)) 3739 #define stat_inc_rbtree_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_rbtree)) 3740 #define stat_inc_largest_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_largest)) 3741 #define stat_inc_cached_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_cached)) 3742 #define stat_inc_inline_xattr(inode) \ 3743 do { \ 3744 if (f2fs_has_inline_xattr(inode)) \ 3745 (atomic_inc(&F2FS_I_SB(inode)->inline_xattr)); \ 3746 } while (0) 3747 #define stat_dec_inline_xattr(inode) \ 3748 do { \ 3749 if (f2fs_has_inline_xattr(inode)) \ 3750 (atomic_dec(&F2FS_I_SB(inode)->inline_xattr)); \ 3751 } while (0) 3752 #define stat_inc_inline_inode(inode) \ 3753 do { \ 3754 if (f2fs_has_inline_data(inode)) \ 3755 (atomic_inc(&F2FS_I_SB(inode)->inline_inode)); \ 3756 } while (0) 3757 #define stat_dec_inline_inode(inode) \ 3758 do { \ 3759 if (f2fs_has_inline_data(inode)) \ 3760 (atomic_dec(&F2FS_I_SB(inode)->inline_inode)); \ 3761 } while (0) 3762 #define stat_inc_inline_dir(inode) \ 3763 do { \ 3764 if (f2fs_has_inline_dentry(inode)) \ 3765 (atomic_inc(&F2FS_I_SB(inode)->inline_dir)); \ 3766 } while (0) 3767 #define stat_dec_inline_dir(inode) \ 3768 do { \ 3769 if (f2fs_has_inline_dentry(inode)) \ 3770 (atomic_dec(&F2FS_I_SB(inode)->inline_dir)); \ 3771 } while (0) 3772 #define stat_inc_compr_inode(inode) \ 3773 do { \ 3774 if (f2fs_compressed_file(inode)) \ 3775 (atomic_inc(&F2FS_I_SB(inode)->compr_inode)); \ 3776 } while (0) 3777 #define stat_dec_compr_inode(inode) \ 3778 do { \ 3779 if (f2fs_compressed_file(inode)) \ 3780 (atomic_dec(&F2FS_I_SB(inode)->compr_inode)); \ 3781 } while (0) 3782 #define stat_add_compr_blocks(inode, blocks) \ 3783 (atomic64_add(blocks, &F2FS_I_SB(inode)->compr_blocks)) 3784 #define stat_sub_compr_blocks(inode, blocks) \ 3785 (atomic64_sub(blocks, &F2FS_I_SB(inode)->compr_blocks)) 3786 #define stat_inc_meta_count(sbi, blkaddr) \ 3787 do { \ 3788 if (blkaddr < SIT_I(sbi)->sit_base_addr) \ 3789 atomic_inc(&(sbi)->meta_count[META_CP]); \ 3790 else if (blkaddr < NM_I(sbi)->nat_blkaddr) \ 3791 atomic_inc(&(sbi)->meta_count[META_SIT]); \ 3792 else if (blkaddr < SM_I(sbi)->ssa_blkaddr) \ 3793 atomic_inc(&(sbi)->meta_count[META_NAT]); \ 3794 else if (blkaddr < SM_I(sbi)->main_blkaddr) \ 3795 atomic_inc(&(sbi)->meta_count[META_SSA]); \ 3796 } while (0) 3797 #define stat_inc_seg_type(sbi, curseg) \ 3798 ((sbi)->segment_count[(curseg)->alloc_type]++) 3799 #define stat_inc_block_count(sbi, curseg) \ 3800 ((sbi)->block_count[(curseg)->alloc_type]++) 3801 #define stat_inc_inplace_blocks(sbi) \ 3802 (atomic_inc(&(sbi)->inplace_count)) 3803 #define stat_update_max_atomic_write(inode) \ 3804 do { \ 3805 int cur = F2FS_I_SB(inode)->atomic_files; \ 3806 int max = atomic_read(&F2FS_I_SB(inode)->max_aw_cnt); \ 3807 if (cur > max) \ 3808 atomic_set(&F2FS_I_SB(inode)->max_aw_cnt, cur); \ 3809 } while (0) 3810 #define stat_inc_volatile_write(inode) \ 3811 (atomic_inc(&F2FS_I_SB(inode)->vw_cnt)) 3812 #define stat_dec_volatile_write(inode) \ 3813 (atomic_dec(&F2FS_I_SB(inode)->vw_cnt)) 3814 #define stat_update_max_volatile_write(inode) \ 3815 do { \ 3816 int cur = atomic_read(&F2FS_I_SB(inode)->vw_cnt); \ 3817 int max = atomic_read(&F2FS_I_SB(inode)->max_vw_cnt); \ 3818 if (cur > max) \ 3819 atomic_set(&F2FS_I_SB(inode)->max_vw_cnt, cur); \ 3820 } while (0) 3821 #define stat_inc_seg_count(sbi, type, gc_type) \ 3822 do { \ 3823 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 3824 si->tot_segs++; \ 3825 if ((type) == SUM_TYPE_DATA) { \ 3826 si->data_segs++; \ 3827 si->bg_data_segs += (gc_type == BG_GC) ? 1 : 0; \ 3828 } else { \ 3829 si->node_segs++; \ 3830 si->bg_node_segs += (gc_type == BG_GC) ? 1 : 0; \ 3831 } \ 3832 } while (0) 3833 3834 #define stat_inc_tot_blk_count(si, blks) \ 3835 ((si)->tot_blks += (blks)) 3836 3837 #define stat_inc_data_blk_count(sbi, blks, gc_type) \ 3838 do { \ 3839 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 3840 stat_inc_tot_blk_count(si, blks); \ 3841 si->data_blks += (blks); \ 3842 si->bg_data_blks += ((gc_type) == BG_GC) ? (blks) : 0; \ 3843 } while (0) 3844 3845 #define stat_inc_node_blk_count(sbi, blks, gc_type) \ 3846 do { \ 3847 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 3848 stat_inc_tot_blk_count(si, blks); \ 3849 si->node_blks += (blks); \ 3850 si->bg_node_blks += ((gc_type) == BG_GC) ? (blks) : 0; \ 3851 } while (0) 3852 3853 int f2fs_build_stats(struct f2fs_sb_info *sbi); 3854 void f2fs_destroy_stats(struct f2fs_sb_info *sbi); 3855 void __init f2fs_create_root_stats(void); 3856 void f2fs_destroy_root_stats(void); 3857 void f2fs_update_sit_info(struct f2fs_sb_info *sbi); 3858 #else 3859 #define stat_inc_cp_count(si) do { } while (0) 3860 #define stat_inc_bg_cp_count(si) do { } while (0) 3861 #define stat_inc_call_count(si) do { } while (0) 3862 #define stat_inc_bggc_count(si) do { } while (0) 3863 #define stat_io_skip_bggc_count(sbi) do { } while (0) 3864 #define stat_other_skip_bggc_count(sbi) do { } while (0) 3865 #define stat_inc_dirty_inode(sbi, type) do { } while (0) 3866 #define stat_dec_dirty_inode(sbi, type) do { } while (0) 3867 #define stat_inc_total_hit(sbi) do { } while (0) 3868 #define stat_inc_rbtree_node_hit(sbi) do { } while (0) 3869 #define stat_inc_largest_node_hit(sbi) do { } while (0) 3870 #define stat_inc_cached_node_hit(sbi) do { } while (0) 3871 #define stat_inc_inline_xattr(inode) do { } while (0) 3872 #define stat_dec_inline_xattr(inode) do { } while (0) 3873 #define stat_inc_inline_inode(inode) do { } while (0) 3874 #define stat_dec_inline_inode(inode) do { } while (0) 3875 #define stat_inc_inline_dir(inode) do { } while (0) 3876 #define stat_dec_inline_dir(inode) do { } while (0) 3877 #define stat_inc_compr_inode(inode) do { } while (0) 3878 #define stat_dec_compr_inode(inode) do { } while (0) 3879 #define stat_add_compr_blocks(inode, blocks) do { } while (0) 3880 #define stat_sub_compr_blocks(inode, blocks) do { } while (0) 3881 #define stat_update_max_atomic_write(inode) do { } while (0) 3882 #define stat_inc_volatile_write(inode) do { } while (0) 3883 #define stat_dec_volatile_write(inode) do { } while (0) 3884 #define stat_update_max_volatile_write(inode) do { } while (0) 3885 #define stat_inc_meta_count(sbi, blkaddr) do { } while (0) 3886 #define stat_inc_seg_type(sbi, curseg) do { } while (0) 3887 #define stat_inc_block_count(sbi, curseg) do { } while (0) 3888 #define stat_inc_inplace_blocks(sbi) do { } while (0) 3889 #define stat_inc_seg_count(sbi, type, gc_type) do { } while (0) 3890 #define stat_inc_tot_blk_count(si, blks) do { } while (0) 3891 #define stat_inc_data_blk_count(sbi, blks, gc_type) do { } while (0) 3892 #define stat_inc_node_blk_count(sbi, blks, gc_type) do { } while (0) 3893 3894 static inline int f2fs_build_stats(struct f2fs_sb_info *sbi) { return 0; } 3895 static inline void f2fs_destroy_stats(struct f2fs_sb_info *sbi) { } 3896 static inline void __init f2fs_create_root_stats(void) { } 3897 static inline void f2fs_destroy_root_stats(void) { } 3898 static inline void f2fs_update_sit_info(struct f2fs_sb_info *sbi) {} 3899 #endif 3900 3901 extern const struct file_operations f2fs_dir_operations; 3902 extern const struct file_operations f2fs_file_operations; 3903 extern const struct inode_operations f2fs_file_inode_operations; 3904 extern const struct address_space_operations f2fs_dblock_aops; 3905 extern const struct address_space_operations f2fs_node_aops; 3906 extern const struct address_space_operations f2fs_meta_aops; 3907 extern const struct inode_operations f2fs_dir_inode_operations; 3908 extern const struct inode_operations f2fs_symlink_inode_operations; 3909 extern const struct inode_operations f2fs_encrypted_symlink_inode_operations; 3910 extern const struct inode_operations f2fs_special_inode_operations; 3911 extern struct kmem_cache *f2fs_inode_entry_slab; 3912 3913 /* 3914 * inline.c 3915 */ 3916 bool f2fs_may_inline_data(struct inode *inode); 3917 bool f2fs_may_inline_dentry(struct inode *inode); 3918 void f2fs_do_read_inline_data(struct page *page, struct page *ipage); 3919 void f2fs_truncate_inline_inode(struct inode *inode, 3920 struct page *ipage, u64 from); 3921 int f2fs_read_inline_data(struct inode *inode, struct page *page); 3922 int f2fs_convert_inline_page(struct dnode_of_data *dn, struct page *page); 3923 int f2fs_convert_inline_inode(struct inode *inode); 3924 int f2fs_try_convert_inline_dir(struct inode *dir, struct dentry *dentry); 3925 int f2fs_write_inline_data(struct inode *inode, struct page *page); 3926 int f2fs_recover_inline_data(struct inode *inode, struct page *npage); 3927 struct f2fs_dir_entry *f2fs_find_in_inline_dir(struct inode *dir, 3928 const struct f2fs_filename *fname, 3929 struct page **res_page); 3930 int f2fs_make_empty_inline_dir(struct inode *inode, struct inode *parent, 3931 struct page *ipage); 3932 int f2fs_add_inline_entry(struct inode *dir, const struct f2fs_filename *fname, 3933 struct inode *inode, nid_t ino, umode_t mode); 3934 void f2fs_delete_inline_entry(struct f2fs_dir_entry *dentry, 3935 struct page *page, struct inode *dir, 3936 struct inode *inode); 3937 bool f2fs_empty_inline_dir(struct inode *dir); 3938 int f2fs_read_inline_dir(struct file *file, struct dir_context *ctx, 3939 struct fscrypt_str *fstr); 3940 int f2fs_inline_data_fiemap(struct inode *inode, 3941 struct fiemap_extent_info *fieinfo, 3942 __u64 start, __u64 len); 3943 3944 /* 3945 * shrinker.c 3946 */ 3947 unsigned long f2fs_shrink_count(struct shrinker *shrink, 3948 struct shrink_control *sc); 3949 unsigned long f2fs_shrink_scan(struct shrinker *shrink, 3950 struct shrink_control *sc); 3951 void f2fs_join_shrinker(struct f2fs_sb_info *sbi); 3952 void f2fs_leave_shrinker(struct f2fs_sb_info *sbi); 3953 3954 /* 3955 * extent_cache.c 3956 */ 3957 struct rb_entry *f2fs_lookup_rb_tree(struct rb_root_cached *root, 3958 struct rb_entry *cached_re, unsigned int ofs); 3959 struct rb_node **f2fs_lookup_rb_tree_ext(struct f2fs_sb_info *sbi, 3960 struct rb_root_cached *root, 3961 struct rb_node **parent, 3962 unsigned long long key, bool *left_most); 3963 struct rb_node **f2fs_lookup_rb_tree_for_insert(struct f2fs_sb_info *sbi, 3964 struct rb_root_cached *root, 3965 struct rb_node **parent, 3966 unsigned int ofs, bool *leftmost); 3967 struct rb_entry *f2fs_lookup_rb_tree_ret(struct rb_root_cached *root, 3968 struct rb_entry *cached_re, unsigned int ofs, 3969 struct rb_entry **prev_entry, struct rb_entry **next_entry, 3970 struct rb_node ***insert_p, struct rb_node **insert_parent, 3971 bool force, bool *leftmost); 3972 bool f2fs_check_rb_tree_consistence(struct f2fs_sb_info *sbi, 3973 struct rb_root_cached *root, bool check_key); 3974 unsigned int f2fs_shrink_extent_tree(struct f2fs_sb_info *sbi, int nr_shrink); 3975 void f2fs_init_extent_tree(struct inode *inode, struct page *ipage); 3976 void f2fs_drop_extent_tree(struct inode *inode); 3977 unsigned int f2fs_destroy_extent_node(struct inode *inode); 3978 void f2fs_destroy_extent_tree(struct inode *inode); 3979 bool f2fs_lookup_extent_cache(struct inode *inode, pgoff_t pgofs, 3980 struct extent_info *ei); 3981 void f2fs_update_extent_cache(struct dnode_of_data *dn); 3982 void f2fs_update_extent_cache_range(struct dnode_of_data *dn, 3983 pgoff_t fofs, block_t blkaddr, unsigned int len); 3984 void f2fs_init_extent_cache_info(struct f2fs_sb_info *sbi); 3985 int __init f2fs_create_extent_cache(void); 3986 void f2fs_destroy_extent_cache(void); 3987 3988 /* 3989 * sysfs.c 3990 */ 3991 #define MIN_RA_MUL 2 3992 #define MAX_RA_MUL 256 3993 3994 int __init f2fs_init_sysfs(void); 3995 void f2fs_exit_sysfs(void); 3996 int f2fs_register_sysfs(struct f2fs_sb_info *sbi); 3997 void f2fs_unregister_sysfs(struct f2fs_sb_info *sbi); 3998 3999 /* verity.c */ 4000 extern const struct fsverity_operations f2fs_verityops; 4001 4002 /* 4003 * crypto support 4004 */ 4005 static inline bool f2fs_encrypted_file(struct inode *inode) 4006 { 4007 return IS_ENCRYPTED(inode) && S_ISREG(inode->i_mode); 4008 } 4009 4010 static inline void f2fs_set_encrypted_inode(struct inode *inode) 4011 { 4012 #ifdef CONFIG_FS_ENCRYPTION 4013 file_set_encrypt(inode); 4014 f2fs_set_inode_flags(inode); 4015 #endif 4016 } 4017 4018 /* 4019 * Returns true if the reads of the inode's data need to undergo some 4020 * postprocessing step, like decryption or authenticity verification. 4021 */ 4022 static inline bool f2fs_post_read_required(struct inode *inode) 4023 { 4024 return f2fs_encrypted_file(inode) || fsverity_active(inode) || 4025 f2fs_compressed_file(inode); 4026 } 4027 4028 /* 4029 * compress.c 4030 */ 4031 #ifdef CONFIG_F2FS_FS_COMPRESSION 4032 bool f2fs_is_compressed_page(struct page *page); 4033 struct page *f2fs_compress_control_page(struct page *page); 4034 int f2fs_prepare_compress_overwrite(struct inode *inode, 4035 struct page **pagep, pgoff_t index, void **fsdata); 4036 bool f2fs_compress_write_end(struct inode *inode, void *fsdata, 4037 pgoff_t index, unsigned copied); 4038 int f2fs_truncate_partial_cluster(struct inode *inode, u64 from, bool lock); 4039 void f2fs_compress_write_end_io(struct bio *bio, struct page *page); 4040 bool f2fs_is_compress_backend_ready(struct inode *inode); 4041 int f2fs_init_compress_mempool(void); 4042 void f2fs_destroy_compress_mempool(void); 4043 void f2fs_decompress_cluster(struct decompress_io_ctx *dic); 4044 void f2fs_end_read_compressed_page(struct page *page, bool failed, 4045 block_t blkaddr); 4046 bool f2fs_cluster_is_empty(struct compress_ctx *cc); 4047 bool f2fs_cluster_can_merge_page(struct compress_ctx *cc, pgoff_t index); 4048 bool f2fs_sanity_check_cluster(struct dnode_of_data *dn); 4049 void f2fs_compress_ctx_add_page(struct compress_ctx *cc, struct page *page); 4050 int f2fs_write_multi_pages(struct compress_ctx *cc, 4051 int *submitted, 4052 struct writeback_control *wbc, 4053 enum iostat_type io_type); 4054 int f2fs_is_compressed_cluster(struct inode *inode, pgoff_t index); 4055 void f2fs_update_extent_tree_range_compressed(struct inode *inode, 4056 pgoff_t fofs, block_t blkaddr, unsigned int llen, 4057 unsigned int c_len); 4058 int f2fs_read_multi_pages(struct compress_ctx *cc, struct bio **bio_ret, 4059 unsigned nr_pages, sector_t *last_block_in_bio, 4060 bool is_readahead, bool for_write); 4061 struct decompress_io_ctx *f2fs_alloc_dic(struct compress_ctx *cc); 4062 void f2fs_decompress_end_io(struct decompress_io_ctx *dic, bool failed); 4063 void f2fs_put_page_dic(struct page *page); 4064 unsigned int f2fs_cluster_blocks_are_contiguous(struct dnode_of_data *dn); 4065 int f2fs_init_compress_ctx(struct compress_ctx *cc); 4066 void f2fs_destroy_compress_ctx(struct compress_ctx *cc, bool reuse); 4067 void f2fs_init_compress_info(struct f2fs_sb_info *sbi); 4068 int f2fs_init_compress_inode(struct f2fs_sb_info *sbi); 4069 void f2fs_destroy_compress_inode(struct f2fs_sb_info *sbi); 4070 int f2fs_init_page_array_cache(struct f2fs_sb_info *sbi); 4071 void f2fs_destroy_page_array_cache(struct f2fs_sb_info *sbi); 4072 int __init f2fs_init_compress_cache(void); 4073 void f2fs_destroy_compress_cache(void); 4074 struct address_space *COMPRESS_MAPPING(struct f2fs_sb_info *sbi); 4075 void f2fs_invalidate_compress_page(struct f2fs_sb_info *sbi, block_t blkaddr); 4076 void f2fs_cache_compressed_page(struct f2fs_sb_info *sbi, struct page *page, 4077 nid_t ino, block_t blkaddr); 4078 bool f2fs_load_compressed_page(struct f2fs_sb_info *sbi, struct page *page, 4079 block_t blkaddr); 4080 void f2fs_invalidate_compress_pages(struct f2fs_sb_info *sbi, nid_t ino); 4081 #define inc_compr_inode_stat(inode) \ 4082 do { \ 4083 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); \ 4084 sbi->compr_new_inode++; \ 4085 } while (0) 4086 #define add_compr_block_stat(inode, blocks) \ 4087 do { \ 4088 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); \ 4089 int diff = F2FS_I(inode)->i_cluster_size - blocks; \ 4090 sbi->compr_written_block += blocks; \ 4091 sbi->compr_saved_block += diff; \ 4092 } while (0) 4093 #else 4094 static inline bool f2fs_is_compressed_page(struct page *page) { return false; } 4095 static inline bool f2fs_is_compress_backend_ready(struct inode *inode) 4096 { 4097 if (!f2fs_compressed_file(inode)) 4098 return true; 4099 /* not support compression */ 4100 return false; 4101 } 4102 static inline struct page *f2fs_compress_control_page(struct page *page) 4103 { 4104 WARN_ON_ONCE(1); 4105 return ERR_PTR(-EINVAL); 4106 } 4107 static inline int f2fs_init_compress_mempool(void) { return 0; } 4108 static inline void f2fs_destroy_compress_mempool(void) { } 4109 static inline void f2fs_decompress_cluster(struct decompress_io_ctx *dic) { } 4110 static inline void f2fs_end_read_compressed_page(struct page *page, 4111 bool failed, block_t blkaddr) 4112 { 4113 WARN_ON_ONCE(1); 4114 } 4115 static inline void f2fs_put_page_dic(struct page *page) 4116 { 4117 WARN_ON_ONCE(1); 4118 } 4119 static inline unsigned int f2fs_cluster_blocks_are_contiguous(struct dnode_of_data *dn) { return 0; } 4120 static inline bool f2fs_sanity_check_cluster(struct dnode_of_data *dn) { return false; } 4121 static inline int f2fs_init_compress_inode(struct f2fs_sb_info *sbi) { return 0; } 4122 static inline void f2fs_destroy_compress_inode(struct f2fs_sb_info *sbi) { } 4123 static inline int f2fs_init_page_array_cache(struct f2fs_sb_info *sbi) { return 0; } 4124 static inline void f2fs_destroy_page_array_cache(struct f2fs_sb_info *sbi) { } 4125 static inline int __init f2fs_init_compress_cache(void) { return 0; } 4126 static inline void f2fs_destroy_compress_cache(void) { } 4127 static inline void f2fs_invalidate_compress_page(struct f2fs_sb_info *sbi, 4128 block_t blkaddr) { } 4129 static inline void f2fs_cache_compressed_page(struct f2fs_sb_info *sbi, 4130 struct page *page, nid_t ino, block_t blkaddr) { } 4131 static inline bool f2fs_load_compressed_page(struct f2fs_sb_info *sbi, 4132 struct page *page, block_t blkaddr) { return false; } 4133 static inline void f2fs_invalidate_compress_pages(struct f2fs_sb_info *sbi, 4134 nid_t ino) { } 4135 #define inc_compr_inode_stat(inode) do { } while (0) 4136 static inline void f2fs_update_extent_tree_range_compressed(struct inode *inode, 4137 pgoff_t fofs, block_t blkaddr, unsigned int llen, 4138 unsigned int c_len) { } 4139 #endif 4140 4141 static inline void set_compress_context(struct inode *inode) 4142 { 4143 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 4144 4145 F2FS_I(inode)->i_compress_algorithm = 4146 F2FS_OPTION(sbi).compress_algorithm; 4147 F2FS_I(inode)->i_log_cluster_size = 4148 F2FS_OPTION(sbi).compress_log_size; 4149 F2FS_I(inode)->i_compress_flag = 4150 F2FS_OPTION(sbi).compress_chksum ? 4151 1 << COMPRESS_CHKSUM : 0; 4152 F2FS_I(inode)->i_cluster_size = 4153 1 << F2FS_I(inode)->i_log_cluster_size; 4154 if ((F2FS_I(inode)->i_compress_algorithm == COMPRESS_LZ4 || 4155 F2FS_I(inode)->i_compress_algorithm == COMPRESS_ZSTD) && 4156 F2FS_OPTION(sbi).compress_level) 4157 F2FS_I(inode)->i_compress_flag |= 4158 F2FS_OPTION(sbi).compress_level << 4159 COMPRESS_LEVEL_OFFSET; 4160 F2FS_I(inode)->i_flags |= F2FS_COMPR_FL; 4161 set_inode_flag(inode, FI_COMPRESSED_FILE); 4162 stat_inc_compr_inode(inode); 4163 inc_compr_inode_stat(inode); 4164 f2fs_mark_inode_dirty_sync(inode, true); 4165 } 4166 4167 static inline bool f2fs_disable_compressed_file(struct inode *inode) 4168 { 4169 struct f2fs_inode_info *fi = F2FS_I(inode); 4170 4171 if (!f2fs_compressed_file(inode)) 4172 return true; 4173 if (S_ISREG(inode->i_mode) && 4174 (get_dirty_pages(inode) || atomic_read(&fi->i_compr_blocks))) 4175 return false; 4176 4177 fi->i_flags &= ~F2FS_COMPR_FL; 4178 stat_dec_compr_inode(inode); 4179 clear_inode_flag(inode, FI_COMPRESSED_FILE); 4180 f2fs_mark_inode_dirty_sync(inode, true); 4181 return true; 4182 } 4183 4184 #define F2FS_FEATURE_FUNCS(name, flagname) \ 4185 static inline int f2fs_sb_has_##name(struct f2fs_sb_info *sbi) \ 4186 { \ 4187 return F2FS_HAS_FEATURE(sbi, F2FS_FEATURE_##flagname); \ 4188 } 4189 4190 F2FS_FEATURE_FUNCS(encrypt, ENCRYPT); 4191 F2FS_FEATURE_FUNCS(blkzoned, BLKZONED); 4192 F2FS_FEATURE_FUNCS(extra_attr, EXTRA_ATTR); 4193 F2FS_FEATURE_FUNCS(project_quota, PRJQUOTA); 4194 F2FS_FEATURE_FUNCS(inode_chksum, INODE_CHKSUM); 4195 F2FS_FEATURE_FUNCS(flexible_inline_xattr, FLEXIBLE_INLINE_XATTR); 4196 F2FS_FEATURE_FUNCS(quota_ino, QUOTA_INO); 4197 F2FS_FEATURE_FUNCS(inode_crtime, INODE_CRTIME); 4198 F2FS_FEATURE_FUNCS(lost_found, LOST_FOUND); 4199 F2FS_FEATURE_FUNCS(verity, VERITY); 4200 F2FS_FEATURE_FUNCS(sb_chksum, SB_CHKSUM); 4201 F2FS_FEATURE_FUNCS(casefold, CASEFOLD); 4202 F2FS_FEATURE_FUNCS(compression, COMPRESSION); 4203 F2FS_FEATURE_FUNCS(readonly, RO); 4204 4205 static inline bool f2fs_may_extent_tree(struct inode *inode) 4206 { 4207 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 4208 4209 if (!test_opt(sbi, EXTENT_CACHE) || 4210 is_inode_flag_set(inode, FI_NO_EXTENT) || 4211 (is_inode_flag_set(inode, FI_COMPRESSED_FILE) && 4212 !f2fs_sb_has_readonly(sbi))) 4213 return false; 4214 4215 /* 4216 * for recovered files during mount do not create extents 4217 * if shrinker is not registered. 4218 */ 4219 if (list_empty(&sbi->s_list)) 4220 return false; 4221 4222 return S_ISREG(inode->i_mode); 4223 } 4224 4225 #ifdef CONFIG_BLK_DEV_ZONED 4226 static inline bool f2fs_blkz_is_seq(struct f2fs_sb_info *sbi, int devi, 4227 block_t blkaddr) 4228 { 4229 unsigned int zno = blkaddr >> sbi->log_blocks_per_blkz; 4230 4231 return test_bit(zno, FDEV(devi).blkz_seq); 4232 } 4233 #endif 4234 4235 static inline bool f2fs_hw_should_discard(struct f2fs_sb_info *sbi) 4236 { 4237 return f2fs_sb_has_blkzoned(sbi); 4238 } 4239 4240 static inline bool f2fs_bdev_support_discard(struct block_device *bdev) 4241 { 4242 return blk_queue_discard(bdev_get_queue(bdev)) || 4243 bdev_is_zoned(bdev); 4244 } 4245 4246 static inline bool f2fs_hw_support_discard(struct f2fs_sb_info *sbi) 4247 { 4248 int i; 4249 4250 if (!f2fs_is_multi_device(sbi)) 4251 return f2fs_bdev_support_discard(sbi->sb->s_bdev); 4252 4253 for (i = 0; i < sbi->s_ndevs; i++) 4254 if (f2fs_bdev_support_discard(FDEV(i).bdev)) 4255 return true; 4256 return false; 4257 } 4258 4259 static inline bool f2fs_realtime_discard_enable(struct f2fs_sb_info *sbi) 4260 { 4261 return (test_opt(sbi, DISCARD) && f2fs_hw_support_discard(sbi)) || 4262 f2fs_hw_should_discard(sbi); 4263 } 4264 4265 static inline bool f2fs_hw_is_readonly(struct f2fs_sb_info *sbi) 4266 { 4267 int i; 4268 4269 if (!f2fs_is_multi_device(sbi)) 4270 return bdev_read_only(sbi->sb->s_bdev); 4271 4272 for (i = 0; i < sbi->s_ndevs; i++) 4273 if (bdev_read_only(FDEV(i).bdev)) 4274 return true; 4275 return false; 4276 } 4277 4278 static inline bool f2fs_lfs_mode(struct f2fs_sb_info *sbi) 4279 { 4280 return F2FS_OPTION(sbi).fs_mode == FS_MODE_LFS; 4281 } 4282 4283 static inline bool f2fs_may_compress(struct inode *inode) 4284 { 4285 if (IS_SWAPFILE(inode) || f2fs_is_pinned_file(inode) || 4286 f2fs_is_atomic_file(inode) || 4287 f2fs_is_volatile_file(inode)) 4288 return false; 4289 return S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode); 4290 } 4291 4292 static inline void f2fs_i_compr_blocks_update(struct inode *inode, 4293 u64 blocks, bool add) 4294 { 4295 int diff = F2FS_I(inode)->i_cluster_size - blocks; 4296 struct f2fs_inode_info *fi = F2FS_I(inode); 4297 4298 /* don't update i_compr_blocks if saved blocks were released */ 4299 if (!add && !atomic_read(&fi->i_compr_blocks)) 4300 return; 4301 4302 if (add) { 4303 atomic_add(diff, &fi->i_compr_blocks); 4304 stat_add_compr_blocks(inode, diff); 4305 } else { 4306 atomic_sub(diff, &fi->i_compr_blocks); 4307 stat_sub_compr_blocks(inode, diff); 4308 } 4309 f2fs_mark_inode_dirty_sync(inode, true); 4310 } 4311 4312 static inline int block_unaligned_IO(struct inode *inode, 4313 struct kiocb *iocb, struct iov_iter *iter) 4314 { 4315 unsigned int i_blkbits = READ_ONCE(inode->i_blkbits); 4316 unsigned int blocksize_mask = (1 << i_blkbits) - 1; 4317 loff_t offset = iocb->ki_pos; 4318 unsigned long align = offset | iov_iter_alignment(iter); 4319 4320 return align & blocksize_mask; 4321 } 4322 4323 static inline bool f2fs_force_buffered_io(struct inode *inode, 4324 struct kiocb *iocb, struct iov_iter *iter) 4325 { 4326 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 4327 int rw = iov_iter_rw(iter); 4328 4329 if (f2fs_post_read_required(inode)) 4330 return true; 4331 if (f2fs_is_multi_device(sbi)) 4332 return true; 4333 /* 4334 * for blkzoned device, fallback direct IO to buffered IO, so 4335 * all IOs can be serialized by log-structured write. 4336 */ 4337 if (f2fs_sb_has_blkzoned(sbi)) 4338 return true; 4339 if (f2fs_lfs_mode(sbi) && (rw == WRITE)) { 4340 if (block_unaligned_IO(inode, iocb, iter)) 4341 return true; 4342 if (F2FS_IO_ALIGNED(sbi)) 4343 return true; 4344 } 4345 if (is_sbi_flag_set(F2FS_I_SB(inode), SBI_CP_DISABLED)) 4346 return true; 4347 4348 return false; 4349 } 4350 4351 static inline bool f2fs_need_verity(const struct inode *inode, pgoff_t idx) 4352 { 4353 return fsverity_active(inode) && 4354 idx < DIV_ROUND_UP(inode->i_size, PAGE_SIZE); 4355 } 4356 4357 #ifdef CONFIG_F2FS_FAULT_INJECTION 4358 extern void f2fs_build_fault_attr(struct f2fs_sb_info *sbi, unsigned int rate, 4359 unsigned int type); 4360 #else 4361 #define f2fs_build_fault_attr(sbi, rate, type) do { } while (0) 4362 #endif 4363 4364 static inline bool is_journalled_quota(struct f2fs_sb_info *sbi) 4365 { 4366 #ifdef CONFIG_QUOTA 4367 if (f2fs_sb_has_quota_ino(sbi)) 4368 return true; 4369 if (F2FS_OPTION(sbi).s_qf_names[USRQUOTA] || 4370 F2FS_OPTION(sbi).s_qf_names[GRPQUOTA] || 4371 F2FS_OPTION(sbi).s_qf_names[PRJQUOTA]) 4372 return true; 4373 #endif 4374 return false; 4375 } 4376 4377 static inline bool f2fs_block_unit_discard(struct f2fs_sb_info *sbi) 4378 { 4379 return F2FS_OPTION(sbi).discard_unit == DISCARD_UNIT_BLOCK; 4380 } 4381 4382 #define EFSBADCRC EBADMSG /* Bad CRC detected */ 4383 #define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */ 4384 4385 #endif /* _LINUX_F2FS_H */ 4386