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/sched/mm.h> 22 #include <linux/vmalloc.h> 23 #include <linux/bio.h> 24 #include <linux/blkdev.h> 25 #include <linux/quotaops.h> 26 #include <linux/part_stat.h> 27 #include <crypto/hash.h> 28 29 #include <linux/fscrypt.h> 30 #include <linux/fsverity.h> 31 32 struct pagevec; 33 34 #ifdef CONFIG_F2FS_CHECK_FS 35 #define f2fs_bug_on(sbi, condition) BUG_ON(condition) 36 #else 37 #define f2fs_bug_on(sbi, condition) \ 38 do { \ 39 if (WARN_ON(condition)) \ 40 set_sbi_flag(sbi, SBI_NEED_FSCK); \ 41 } while (0) 42 #endif 43 44 enum { 45 FAULT_KMALLOC, 46 FAULT_KVMALLOC, 47 FAULT_PAGE_ALLOC, 48 FAULT_PAGE_GET, 49 FAULT_ALLOC_BIO, /* it's obsolete due to bio_alloc() will never fail */ 50 FAULT_ALLOC_NID, 51 FAULT_ORPHAN, 52 FAULT_BLOCK, 53 FAULT_DIR_DEPTH, 54 FAULT_EVICT_INODE, 55 FAULT_TRUNCATE, 56 FAULT_READ_IO, 57 FAULT_CHECKPOINT, 58 FAULT_DISCARD, 59 FAULT_WRITE_IO, 60 FAULT_SLAB_ALLOC, 61 FAULT_DQUOT_INIT, 62 FAULT_LOCK_OP, 63 FAULT_BLKADDR, 64 FAULT_MAX, 65 }; 66 67 #ifdef CONFIG_F2FS_FAULT_INJECTION 68 #define F2FS_ALL_FAULT_TYPE ((1 << FAULT_MAX) - 1) 69 70 struct f2fs_fault_info { 71 atomic_t inject_ops; 72 unsigned int inject_rate; 73 unsigned int inject_type; 74 }; 75 76 extern const char *f2fs_fault_name[FAULT_MAX]; 77 #define IS_FAULT_SET(fi, type) ((fi)->inject_type & (1 << (type))) 78 #endif 79 80 /* 81 * For mount options 82 */ 83 #define F2FS_MOUNT_DISABLE_ROLL_FORWARD 0x00000002 84 #define F2FS_MOUNT_DISCARD 0x00000004 85 #define F2FS_MOUNT_NOHEAP 0x00000008 86 #define F2FS_MOUNT_XATTR_USER 0x00000010 87 #define F2FS_MOUNT_POSIX_ACL 0x00000020 88 #define F2FS_MOUNT_DISABLE_EXT_IDENTIFY 0x00000040 89 #define F2FS_MOUNT_INLINE_XATTR 0x00000080 90 #define F2FS_MOUNT_INLINE_DATA 0x00000100 91 #define F2FS_MOUNT_INLINE_DENTRY 0x00000200 92 #define F2FS_MOUNT_FLUSH_MERGE 0x00000400 93 #define F2FS_MOUNT_NOBARRIER 0x00000800 94 #define F2FS_MOUNT_FASTBOOT 0x00001000 95 #define F2FS_MOUNT_EXTENT_CACHE 0x00002000 96 #define F2FS_MOUNT_DATA_FLUSH 0x00008000 97 #define F2FS_MOUNT_FAULT_INJECTION 0x00010000 98 #define F2FS_MOUNT_USRQUOTA 0x00080000 99 #define F2FS_MOUNT_GRPQUOTA 0x00100000 100 #define F2FS_MOUNT_PRJQUOTA 0x00200000 101 #define F2FS_MOUNT_QUOTA 0x00400000 102 #define F2FS_MOUNT_INLINE_XATTR_SIZE 0x00800000 103 #define F2FS_MOUNT_RESERVE_ROOT 0x01000000 104 #define F2FS_MOUNT_DISABLE_CHECKPOINT 0x02000000 105 #define F2FS_MOUNT_NORECOVERY 0x04000000 106 #define F2FS_MOUNT_ATGC 0x08000000 107 #define F2FS_MOUNT_MERGE_CHECKPOINT 0x10000000 108 #define F2FS_MOUNT_GC_MERGE 0x20000000 109 #define F2FS_MOUNT_COMPRESS_CACHE 0x40000000 110 111 #define F2FS_OPTION(sbi) ((sbi)->mount_opt) 112 #define clear_opt(sbi, option) (F2FS_OPTION(sbi).opt &= ~F2FS_MOUNT_##option) 113 #define set_opt(sbi, option) (F2FS_OPTION(sbi).opt |= F2FS_MOUNT_##option) 114 #define test_opt(sbi, option) (F2FS_OPTION(sbi).opt & F2FS_MOUNT_##option) 115 116 #define ver_after(a, b) (typecheck(unsigned long long, a) && \ 117 typecheck(unsigned long long, b) && \ 118 ((long long)((a) - (b)) > 0)) 119 120 typedef u32 block_t; /* 121 * should not change u32, since it is the on-disk block 122 * address format, __le32. 123 */ 124 typedef u32 nid_t; 125 126 #define COMPRESS_EXT_NUM 16 127 128 /* 129 * An implementation of an rwsem that is explicitly unfair to readers. This 130 * prevents priority inversion when a low-priority reader acquires the read lock 131 * while sleeping on the write lock but the write lock is needed by 132 * higher-priority clients. 133 */ 134 135 struct f2fs_rwsem { 136 struct rw_semaphore internal_rwsem; 137 #ifdef CONFIG_F2FS_UNFAIR_RWSEM 138 wait_queue_head_t read_waiters; 139 #endif 140 }; 141 142 struct f2fs_mount_info { 143 unsigned int opt; 144 int write_io_size_bits; /* Write IO size bits */ 145 block_t root_reserved_blocks; /* root reserved blocks */ 146 kuid_t s_resuid; /* reserved blocks for uid */ 147 kgid_t s_resgid; /* reserved blocks for gid */ 148 int active_logs; /* # of active logs */ 149 int inline_xattr_size; /* inline xattr size */ 150 #ifdef CONFIG_F2FS_FAULT_INJECTION 151 struct f2fs_fault_info fault_info; /* For fault injection */ 152 #endif 153 #ifdef CONFIG_QUOTA 154 /* Names of quota files with journalled quota */ 155 char *s_qf_names[MAXQUOTAS]; 156 int s_jquota_fmt; /* Format of quota to use */ 157 #endif 158 /* For which write hints are passed down to block layer */ 159 int alloc_mode; /* segment allocation policy */ 160 int fsync_mode; /* fsync policy */ 161 int fs_mode; /* fs mode: LFS or ADAPTIVE */ 162 int bggc_mode; /* bggc mode: off, on or sync */ 163 int memory_mode; /* memory mode */ 164 int discard_unit; /* 165 * discard command's offset/size should 166 * be aligned to this unit: block, 167 * segment or section 168 */ 169 struct fscrypt_dummy_policy dummy_enc_policy; /* test dummy encryption */ 170 block_t unusable_cap_perc; /* percentage for cap */ 171 block_t unusable_cap; /* Amount of space allowed to be 172 * unusable when disabling checkpoint 173 */ 174 175 /* For compression */ 176 unsigned char compress_algorithm; /* algorithm type */ 177 unsigned char compress_log_size; /* cluster log size */ 178 unsigned char compress_level; /* compress level */ 179 bool compress_chksum; /* compressed data chksum */ 180 unsigned char compress_ext_cnt; /* extension count */ 181 unsigned char nocompress_ext_cnt; /* nocompress extension count */ 182 int compress_mode; /* compression mode */ 183 unsigned char extensions[COMPRESS_EXT_NUM][F2FS_EXTENSION_LEN]; /* extensions */ 184 unsigned char noextensions[COMPRESS_EXT_NUM][F2FS_EXTENSION_LEN]; /* extensions */ 185 }; 186 187 #define F2FS_FEATURE_ENCRYPT 0x0001 188 #define F2FS_FEATURE_BLKZONED 0x0002 189 #define F2FS_FEATURE_ATOMIC_WRITE 0x0004 190 #define F2FS_FEATURE_EXTRA_ATTR 0x0008 191 #define F2FS_FEATURE_PRJQUOTA 0x0010 192 #define F2FS_FEATURE_INODE_CHKSUM 0x0020 193 #define F2FS_FEATURE_FLEXIBLE_INLINE_XATTR 0x0040 194 #define F2FS_FEATURE_QUOTA_INO 0x0080 195 #define F2FS_FEATURE_INODE_CRTIME 0x0100 196 #define F2FS_FEATURE_LOST_FOUND 0x0200 197 #define F2FS_FEATURE_VERITY 0x0400 198 #define F2FS_FEATURE_SB_CHKSUM 0x0800 199 #define F2FS_FEATURE_CASEFOLD 0x1000 200 #define F2FS_FEATURE_COMPRESSION 0x2000 201 #define F2FS_FEATURE_RO 0x4000 202 203 #define __F2FS_HAS_FEATURE(raw_super, mask) \ 204 ((raw_super->feature & cpu_to_le32(mask)) != 0) 205 #define F2FS_HAS_FEATURE(sbi, mask) __F2FS_HAS_FEATURE(sbi->raw_super, mask) 206 #define F2FS_SET_FEATURE(sbi, mask) \ 207 (sbi->raw_super->feature |= cpu_to_le32(mask)) 208 #define F2FS_CLEAR_FEATURE(sbi, mask) \ 209 (sbi->raw_super->feature &= ~cpu_to_le32(mask)) 210 211 /* 212 * Default values for user and/or group using reserved blocks 213 */ 214 #define F2FS_DEF_RESUID 0 215 #define F2FS_DEF_RESGID 0 216 217 /* 218 * For checkpoint manager 219 */ 220 enum { 221 NAT_BITMAP, 222 SIT_BITMAP 223 }; 224 225 #define CP_UMOUNT 0x00000001 226 #define CP_FASTBOOT 0x00000002 227 #define CP_SYNC 0x00000004 228 #define CP_RECOVERY 0x00000008 229 #define CP_DISCARD 0x00000010 230 #define CP_TRIMMED 0x00000020 231 #define CP_PAUSE 0x00000040 232 #define CP_RESIZE 0x00000080 233 234 #define DEF_MAX_DISCARD_REQUEST 8 /* issue 8 discards per round */ 235 #define DEF_MIN_DISCARD_ISSUE_TIME 50 /* 50 ms, if exists */ 236 #define DEF_MID_DISCARD_ISSUE_TIME 500 /* 500 ms, if device busy */ 237 #define DEF_MAX_DISCARD_ISSUE_TIME 60000 /* 60 s, if no candidates */ 238 #define DEF_DISCARD_URGENT_UTIL 80 /* do more discard over 80% */ 239 #define DEF_CP_INTERVAL 60 /* 60 secs */ 240 #define DEF_IDLE_INTERVAL 5 /* 5 secs */ 241 #define DEF_DISABLE_INTERVAL 5 /* 5 secs */ 242 #define DEF_DISABLE_QUICK_INTERVAL 1 /* 1 secs */ 243 #define DEF_UMOUNT_DISCARD_TIMEOUT 5 /* 5 secs */ 244 245 struct cp_control { 246 int reason; 247 __u64 trim_start; 248 __u64 trim_end; 249 __u64 trim_minlen; 250 }; 251 252 /* 253 * indicate meta/data type 254 */ 255 enum { 256 META_CP, 257 META_NAT, 258 META_SIT, 259 META_SSA, 260 META_MAX, 261 META_POR, 262 DATA_GENERIC, /* check range only */ 263 DATA_GENERIC_ENHANCE, /* strong check on range and segment bitmap */ 264 DATA_GENERIC_ENHANCE_READ, /* 265 * strong check on range and segment 266 * bitmap but no warning due to race 267 * condition of read on truncated area 268 * by extent_cache 269 */ 270 DATA_GENERIC_ENHANCE_UPDATE, /* 271 * strong check on range and segment 272 * bitmap for update case 273 */ 274 META_GENERIC, 275 }; 276 277 /* for the list of ino */ 278 enum { 279 ORPHAN_INO, /* for orphan ino list */ 280 APPEND_INO, /* for append ino list */ 281 UPDATE_INO, /* for update ino list */ 282 TRANS_DIR_INO, /* for transactions dir ino list */ 283 FLUSH_INO, /* for multiple device flushing */ 284 MAX_INO_ENTRY, /* max. list */ 285 }; 286 287 struct ino_entry { 288 struct list_head list; /* list head */ 289 nid_t ino; /* inode number */ 290 unsigned int dirty_device; /* dirty device bitmap */ 291 }; 292 293 /* for the list of inodes to be GCed */ 294 struct inode_entry { 295 struct list_head list; /* list head */ 296 struct inode *inode; /* vfs inode pointer */ 297 }; 298 299 struct fsync_node_entry { 300 struct list_head list; /* list head */ 301 struct page *page; /* warm node page pointer */ 302 unsigned int seq_id; /* sequence id */ 303 }; 304 305 struct ckpt_req { 306 struct completion wait; /* completion for checkpoint done */ 307 struct llist_node llnode; /* llist_node to be linked in wait queue */ 308 int ret; /* return code of checkpoint */ 309 ktime_t queue_time; /* request queued time */ 310 }; 311 312 struct ckpt_req_control { 313 struct task_struct *f2fs_issue_ckpt; /* checkpoint task */ 314 int ckpt_thread_ioprio; /* checkpoint merge thread ioprio */ 315 wait_queue_head_t ckpt_wait_queue; /* waiting queue for wake-up */ 316 atomic_t issued_ckpt; /* # of actually issued ckpts */ 317 atomic_t total_ckpt; /* # of total ckpts */ 318 atomic_t queued_ckpt; /* # of queued ckpts */ 319 struct llist_head issue_list; /* list for command issue */ 320 spinlock_t stat_lock; /* lock for below checkpoint time stats */ 321 unsigned int cur_time; /* cur wait time in msec for currently issued checkpoint */ 322 unsigned int peak_time; /* peak wait time in msec until now */ 323 }; 324 325 /* for the bitmap indicate blocks to be discarded */ 326 struct discard_entry { 327 struct list_head list; /* list head */ 328 block_t start_blkaddr; /* start blockaddr of current segment */ 329 unsigned char discard_map[SIT_VBLOCK_MAP_SIZE]; /* segment discard bitmap */ 330 }; 331 332 /* default discard granularity of inner discard thread, unit: block count */ 333 #define DEFAULT_DISCARD_GRANULARITY 16 334 /* default maximum discard granularity of ordered discard, unit: block count */ 335 #define DEFAULT_MAX_ORDERED_DISCARD_GRANULARITY 16 336 337 /* max discard pend list number */ 338 #define MAX_PLIST_NUM 512 339 #define plist_idx(blk_num) ((blk_num) >= MAX_PLIST_NUM ? \ 340 (MAX_PLIST_NUM - 1) : ((blk_num) - 1)) 341 342 enum { 343 D_PREP, /* initial */ 344 D_PARTIAL, /* partially submitted */ 345 D_SUBMIT, /* all submitted */ 346 D_DONE, /* finished */ 347 }; 348 349 struct discard_info { 350 block_t lstart; /* logical start address */ 351 block_t len; /* length */ 352 block_t start; /* actual start address in dev */ 353 }; 354 355 struct discard_cmd { 356 struct rb_node rb_node; /* rb node located in rb-tree */ 357 union { 358 struct { 359 block_t lstart; /* logical start address */ 360 block_t len; /* length */ 361 block_t start; /* actual start address in dev */ 362 }; 363 struct discard_info di; /* discard info */ 364 365 }; 366 struct list_head list; /* command list */ 367 struct completion wait; /* compleation */ 368 struct block_device *bdev; /* bdev */ 369 unsigned short ref; /* reference count */ 370 unsigned char state; /* state */ 371 unsigned char queued; /* queued discard */ 372 int error; /* bio error */ 373 spinlock_t lock; /* for state/bio_ref updating */ 374 unsigned short bio_ref; /* bio reference count */ 375 }; 376 377 enum { 378 DPOLICY_BG, 379 DPOLICY_FORCE, 380 DPOLICY_FSTRIM, 381 DPOLICY_UMOUNT, 382 MAX_DPOLICY, 383 }; 384 385 struct discard_policy { 386 int type; /* type of discard */ 387 unsigned int min_interval; /* used for candidates exist */ 388 unsigned int mid_interval; /* used for device busy */ 389 unsigned int max_interval; /* used for candidates not exist */ 390 unsigned int max_requests; /* # of discards issued per round */ 391 unsigned int io_aware_gran; /* minimum granularity discard not be aware of I/O */ 392 bool io_aware; /* issue discard in idle time */ 393 bool sync; /* submit discard with REQ_SYNC flag */ 394 bool ordered; /* issue discard by lba order */ 395 bool timeout; /* discard timeout for put_super */ 396 unsigned int granularity; /* discard granularity */ 397 }; 398 399 struct discard_cmd_control { 400 struct task_struct *f2fs_issue_discard; /* discard thread */ 401 struct list_head entry_list; /* 4KB discard entry list */ 402 struct list_head pend_list[MAX_PLIST_NUM];/* store pending entries */ 403 struct list_head wait_list; /* store on-flushing entries */ 404 struct list_head fstrim_list; /* in-flight discard from fstrim */ 405 wait_queue_head_t discard_wait_queue; /* waiting queue for wake-up */ 406 unsigned int discard_wake; /* to wake up discard thread */ 407 struct mutex cmd_lock; 408 unsigned int nr_discards; /* # of discards in the list */ 409 unsigned int max_discards; /* max. discards to be issued */ 410 unsigned int max_discard_request; /* max. discard request per round */ 411 unsigned int min_discard_issue_time; /* min. interval between discard issue */ 412 unsigned int mid_discard_issue_time; /* mid. interval between discard issue */ 413 unsigned int max_discard_issue_time; /* max. interval between discard issue */ 414 unsigned int discard_granularity; /* discard granularity */ 415 unsigned int max_ordered_discard; /* maximum discard granularity issued by lba order */ 416 unsigned int undiscard_blks; /* # of undiscard blocks */ 417 unsigned int next_pos; /* next discard position */ 418 atomic_t issued_discard; /* # of issued discard */ 419 atomic_t queued_discard; /* # of queued discard */ 420 atomic_t discard_cmd_cnt; /* # of cached cmd count */ 421 struct rb_root_cached root; /* root of discard rb-tree */ 422 bool rbtree_check; /* config for consistence check */ 423 }; 424 425 /* for the list of fsync inodes, used only during recovery */ 426 struct fsync_inode_entry { 427 struct list_head list; /* list head */ 428 struct inode *inode; /* vfs inode pointer */ 429 block_t blkaddr; /* block address locating the last fsync */ 430 block_t last_dentry; /* block address locating the last dentry */ 431 }; 432 433 #define nats_in_cursum(jnl) (le16_to_cpu((jnl)->n_nats)) 434 #define sits_in_cursum(jnl) (le16_to_cpu((jnl)->n_sits)) 435 436 #define nat_in_journal(jnl, i) ((jnl)->nat_j.entries[i].ne) 437 #define nid_in_journal(jnl, i) ((jnl)->nat_j.entries[i].nid) 438 #define sit_in_journal(jnl, i) ((jnl)->sit_j.entries[i].se) 439 #define segno_in_journal(jnl, i) ((jnl)->sit_j.entries[i].segno) 440 441 #define MAX_NAT_JENTRIES(jnl) (NAT_JOURNAL_ENTRIES - nats_in_cursum(jnl)) 442 #define MAX_SIT_JENTRIES(jnl) (SIT_JOURNAL_ENTRIES - sits_in_cursum(jnl)) 443 444 static inline int update_nats_in_cursum(struct f2fs_journal *journal, int i) 445 { 446 int before = nats_in_cursum(journal); 447 448 journal->n_nats = cpu_to_le16(before + i); 449 return before; 450 } 451 452 static inline int update_sits_in_cursum(struct f2fs_journal *journal, int i) 453 { 454 int before = sits_in_cursum(journal); 455 456 journal->n_sits = cpu_to_le16(before + i); 457 return before; 458 } 459 460 static inline bool __has_cursum_space(struct f2fs_journal *journal, 461 int size, int type) 462 { 463 if (type == NAT_JOURNAL) 464 return size <= MAX_NAT_JENTRIES(journal); 465 return size <= MAX_SIT_JENTRIES(journal); 466 } 467 468 /* for inline stuff */ 469 #define DEF_INLINE_RESERVED_SIZE 1 470 static inline int get_extra_isize(struct inode *inode); 471 static inline int get_inline_xattr_addrs(struct inode *inode); 472 #define MAX_INLINE_DATA(inode) (sizeof(__le32) * \ 473 (CUR_ADDRS_PER_INODE(inode) - \ 474 get_inline_xattr_addrs(inode) - \ 475 DEF_INLINE_RESERVED_SIZE)) 476 477 /* for inline dir */ 478 #define NR_INLINE_DENTRY(inode) (MAX_INLINE_DATA(inode) * BITS_PER_BYTE / \ 479 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \ 480 BITS_PER_BYTE + 1)) 481 #define INLINE_DENTRY_BITMAP_SIZE(inode) \ 482 DIV_ROUND_UP(NR_INLINE_DENTRY(inode), BITS_PER_BYTE) 483 #define INLINE_RESERVED_SIZE(inode) (MAX_INLINE_DATA(inode) - \ 484 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \ 485 NR_INLINE_DENTRY(inode) + \ 486 INLINE_DENTRY_BITMAP_SIZE(inode))) 487 488 /* 489 * For INODE and NODE manager 490 */ 491 /* for directory operations */ 492 493 struct f2fs_filename { 494 /* 495 * The filename the user specified. This is NULL for some 496 * filesystem-internal operations, e.g. converting an inline directory 497 * to a non-inline one, or roll-forward recovering an encrypted dentry. 498 */ 499 const struct qstr *usr_fname; 500 501 /* 502 * The on-disk filename. For encrypted directories, this is encrypted. 503 * This may be NULL for lookups in an encrypted dir without the key. 504 */ 505 struct fscrypt_str disk_name; 506 507 /* The dirhash of this filename */ 508 f2fs_hash_t hash; 509 510 #ifdef CONFIG_FS_ENCRYPTION 511 /* 512 * For lookups in encrypted directories: either the buffer backing 513 * disk_name, or a buffer that holds the decoded no-key name. 514 */ 515 struct fscrypt_str crypto_buf; 516 #endif 517 #if IS_ENABLED(CONFIG_UNICODE) 518 /* 519 * For casefolded directories: the casefolded name, but it's left NULL 520 * if the original name is not valid Unicode, if the original name is 521 * "." or "..", if the directory is both casefolded and encrypted and 522 * its encryption key is unavailable, or if the filesystem is doing an 523 * internal operation where usr_fname is also NULL. In all these cases 524 * we fall back to treating the name as an opaque byte sequence. 525 */ 526 struct fscrypt_str cf_name; 527 #endif 528 }; 529 530 struct f2fs_dentry_ptr { 531 struct inode *inode; 532 void *bitmap; 533 struct f2fs_dir_entry *dentry; 534 __u8 (*filename)[F2FS_SLOT_LEN]; 535 int max; 536 int nr_bitmap; 537 }; 538 539 static inline void make_dentry_ptr_block(struct inode *inode, 540 struct f2fs_dentry_ptr *d, struct f2fs_dentry_block *t) 541 { 542 d->inode = inode; 543 d->max = NR_DENTRY_IN_BLOCK; 544 d->nr_bitmap = SIZE_OF_DENTRY_BITMAP; 545 d->bitmap = t->dentry_bitmap; 546 d->dentry = t->dentry; 547 d->filename = t->filename; 548 } 549 550 static inline void make_dentry_ptr_inline(struct inode *inode, 551 struct f2fs_dentry_ptr *d, void *t) 552 { 553 int entry_cnt = NR_INLINE_DENTRY(inode); 554 int bitmap_size = INLINE_DENTRY_BITMAP_SIZE(inode); 555 int reserved_size = INLINE_RESERVED_SIZE(inode); 556 557 d->inode = inode; 558 d->max = entry_cnt; 559 d->nr_bitmap = bitmap_size; 560 d->bitmap = t; 561 d->dentry = t + bitmap_size + reserved_size; 562 d->filename = t + bitmap_size + reserved_size + 563 SIZE_OF_DIR_ENTRY * entry_cnt; 564 } 565 566 /* 567 * XATTR_NODE_OFFSET stores xattrs to one node block per file keeping -1 568 * as its node offset to distinguish from index node blocks. 569 * But some bits are used to mark the node block. 570 */ 571 #define XATTR_NODE_OFFSET ((((unsigned int)-1) << OFFSET_BIT_SHIFT) \ 572 >> OFFSET_BIT_SHIFT) 573 enum { 574 ALLOC_NODE, /* allocate a new node page if needed */ 575 LOOKUP_NODE, /* look up a node without readahead */ 576 LOOKUP_NODE_RA, /* 577 * look up a node with readahead called 578 * by get_data_block. 579 */ 580 }; 581 582 #define DEFAULT_RETRY_IO_COUNT 8 /* maximum retry read IO or flush count */ 583 584 /* congestion wait timeout value, default: 20ms */ 585 #define DEFAULT_IO_TIMEOUT (msecs_to_jiffies(20)) 586 587 /* maximum retry quota flush count */ 588 #define DEFAULT_RETRY_QUOTA_FLUSH_COUNT 8 589 590 /* maximum retry of EIO'ed page */ 591 #define MAX_RETRY_PAGE_EIO 100 592 593 #define F2FS_LINK_MAX 0xffffffff /* maximum link count per file */ 594 595 #define MAX_DIR_RA_PAGES 4 /* maximum ra pages of dir */ 596 597 /* dirty segments threshold for triggering CP */ 598 #define DEFAULT_DIRTY_THRESHOLD 4 599 600 /* for in-memory extent cache entry */ 601 #define F2FS_MIN_EXTENT_LEN 64 /* minimum extent length */ 602 603 /* number of extent info in extent cache we try to shrink */ 604 #define EXTENT_CACHE_SHRINK_NUMBER 128 605 606 #define RECOVERY_MAX_RA_BLOCKS BIO_MAX_VECS 607 #define RECOVERY_MIN_RA_BLOCKS 1 608 609 #define F2FS_ONSTACK_PAGES 16 /* nr of onstack pages */ 610 611 struct rb_entry { 612 struct rb_node rb_node; /* rb node located in rb-tree */ 613 union { 614 struct { 615 unsigned int ofs; /* start offset of the entry */ 616 unsigned int len; /* length of the entry */ 617 }; 618 unsigned long long key; /* 64-bits key */ 619 } __packed; 620 }; 621 622 struct extent_info { 623 unsigned int fofs; /* start offset in a file */ 624 unsigned int len; /* length of the extent */ 625 u32 blk; /* start block address of the extent */ 626 #ifdef CONFIG_F2FS_FS_COMPRESSION 627 unsigned int c_len; /* physical extent length of compressed blocks */ 628 #endif 629 }; 630 631 struct extent_node { 632 struct rb_node rb_node; /* rb node located in rb-tree */ 633 struct extent_info ei; /* extent info */ 634 struct list_head list; /* node in global extent list of sbi */ 635 struct extent_tree *et; /* extent tree pointer */ 636 }; 637 638 struct extent_tree { 639 nid_t ino; /* inode number */ 640 struct rb_root_cached root; /* root of extent info rb-tree */ 641 struct extent_node *cached_en; /* recently accessed extent node */ 642 struct extent_info largest; /* largested extent info */ 643 struct list_head list; /* to be used by sbi->zombie_list */ 644 rwlock_t lock; /* protect extent info rb-tree */ 645 atomic_t node_cnt; /* # of extent node in rb-tree*/ 646 bool largest_updated; /* largest extent updated */ 647 }; 648 649 /* 650 * This structure is taken from ext4_map_blocks. 651 * 652 * Note that, however, f2fs uses NEW and MAPPED flags for f2fs_map_blocks(). 653 */ 654 #define F2FS_MAP_NEW (1 << BH_New) 655 #define F2FS_MAP_MAPPED (1 << BH_Mapped) 656 #define F2FS_MAP_UNWRITTEN (1 << BH_Unwritten) 657 #define F2FS_MAP_FLAGS (F2FS_MAP_NEW | F2FS_MAP_MAPPED |\ 658 F2FS_MAP_UNWRITTEN) 659 660 struct f2fs_map_blocks { 661 struct block_device *m_bdev; /* for multi-device dio */ 662 block_t m_pblk; 663 block_t m_lblk; 664 unsigned int m_len; 665 unsigned int m_flags; 666 pgoff_t *m_next_pgofs; /* point next possible non-hole pgofs */ 667 pgoff_t *m_next_extent; /* point to next possible extent */ 668 int m_seg_type; 669 bool m_may_create; /* indicate it is from write path */ 670 bool m_multidev_dio; /* indicate it allows multi-device dio */ 671 }; 672 673 /* for flag in get_data_block */ 674 enum { 675 F2FS_GET_BLOCK_DEFAULT, 676 F2FS_GET_BLOCK_FIEMAP, 677 F2FS_GET_BLOCK_BMAP, 678 F2FS_GET_BLOCK_DIO, 679 F2FS_GET_BLOCK_PRE_DIO, 680 F2FS_GET_BLOCK_PRE_AIO, 681 F2FS_GET_BLOCK_PRECACHE, 682 }; 683 684 /* 685 * i_advise uses FADVISE_XXX_BIT. We can add additional hints later. 686 */ 687 #define FADVISE_COLD_BIT 0x01 688 #define FADVISE_LOST_PINO_BIT 0x02 689 #define FADVISE_ENCRYPT_BIT 0x04 690 #define FADVISE_ENC_NAME_BIT 0x08 691 #define FADVISE_KEEP_SIZE_BIT 0x10 692 #define FADVISE_HOT_BIT 0x20 693 #define FADVISE_VERITY_BIT 0x40 694 #define FADVISE_TRUNC_BIT 0x80 695 696 #define FADVISE_MODIFIABLE_BITS (FADVISE_COLD_BIT | FADVISE_HOT_BIT) 697 698 #define file_is_cold(inode) is_file(inode, FADVISE_COLD_BIT) 699 #define file_set_cold(inode) set_file(inode, FADVISE_COLD_BIT) 700 #define file_clear_cold(inode) clear_file(inode, FADVISE_COLD_BIT) 701 702 #define file_wrong_pino(inode) is_file(inode, FADVISE_LOST_PINO_BIT) 703 #define file_lost_pino(inode) set_file(inode, FADVISE_LOST_PINO_BIT) 704 #define file_got_pino(inode) clear_file(inode, FADVISE_LOST_PINO_BIT) 705 706 #define file_is_encrypt(inode) is_file(inode, FADVISE_ENCRYPT_BIT) 707 #define file_set_encrypt(inode) set_file(inode, FADVISE_ENCRYPT_BIT) 708 709 #define file_enc_name(inode) is_file(inode, FADVISE_ENC_NAME_BIT) 710 #define file_set_enc_name(inode) set_file(inode, FADVISE_ENC_NAME_BIT) 711 712 #define file_keep_isize(inode) is_file(inode, FADVISE_KEEP_SIZE_BIT) 713 #define file_set_keep_isize(inode) set_file(inode, FADVISE_KEEP_SIZE_BIT) 714 715 #define file_is_hot(inode) is_file(inode, FADVISE_HOT_BIT) 716 #define file_set_hot(inode) set_file(inode, FADVISE_HOT_BIT) 717 #define file_clear_hot(inode) clear_file(inode, FADVISE_HOT_BIT) 718 719 #define file_is_verity(inode) is_file(inode, FADVISE_VERITY_BIT) 720 #define file_set_verity(inode) set_file(inode, FADVISE_VERITY_BIT) 721 722 #define file_should_truncate(inode) is_file(inode, FADVISE_TRUNC_BIT) 723 #define file_need_truncate(inode) set_file(inode, FADVISE_TRUNC_BIT) 724 #define file_dont_truncate(inode) clear_file(inode, FADVISE_TRUNC_BIT) 725 726 #define DEF_DIR_LEVEL 0 727 728 enum { 729 GC_FAILURE_PIN, 730 MAX_GC_FAILURE 731 }; 732 733 /* used for f2fs_inode_info->flags */ 734 enum { 735 FI_NEW_INODE, /* indicate newly allocated inode */ 736 FI_DIRTY_INODE, /* indicate inode is dirty or not */ 737 FI_AUTO_RECOVER, /* indicate inode is recoverable */ 738 FI_DIRTY_DIR, /* indicate directory has dirty pages */ 739 FI_INC_LINK, /* need to increment i_nlink */ 740 FI_ACL_MODE, /* indicate acl mode */ 741 FI_NO_ALLOC, /* should not allocate any blocks */ 742 FI_FREE_NID, /* free allocated nide */ 743 FI_NO_EXTENT, /* not to use the extent cache */ 744 FI_INLINE_XATTR, /* used for inline xattr */ 745 FI_INLINE_DATA, /* used for inline data*/ 746 FI_INLINE_DENTRY, /* used for inline dentry */ 747 FI_APPEND_WRITE, /* inode has appended data */ 748 FI_UPDATE_WRITE, /* inode has in-place-update data */ 749 FI_NEED_IPU, /* used for ipu per file */ 750 FI_ATOMIC_FILE, /* indicate atomic file */ 751 FI_FIRST_BLOCK_WRITTEN, /* indicate #0 data block was written */ 752 FI_DROP_CACHE, /* drop dirty page cache */ 753 FI_DATA_EXIST, /* indicate data exists */ 754 FI_INLINE_DOTS, /* indicate inline dot dentries */ 755 FI_SKIP_WRITES, /* should skip data page writeback */ 756 FI_OPU_WRITE, /* used for opu per file */ 757 FI_DIRTY_FILE, /* indicate regular/symlink has dirty pages */ 758 FI_PREALLOCATED_ALL, /* all blocks for write were preallocated */ 759 FI_HOT_DATA, /* indicate file is hot */ 760 FI_EXTRA_ATTR, /* indicate file has extra attribute */ 761 FI_PROJ_INHERIT, /* indicate file inherits projectid */ 762 FI_PIN_FILE, /* indicate file should not be gced */ 763 FI_VERITY_IN_PROGRESS, /* building fs-verity Merkle tree */ 764 FI_COMPRESSED_FILE, /* indicate file's data can be compressed */ 765 FI_COMPRESS_CORRUPT, /* indicate compressed cluster is corrupted */ 766 FI_MMAP_FILE, /* indicate file was mmapped */ 767 FI_ENABLE_COMPRESS, /* enable compression in "user" compression mode */ 768 FI_COMPRESS_RELEASED, /* compressed blocks were released */ 769 FI_ALIGNED_WRITE, /* enable aligned write */ 770 FI_COW_FILE, /* indicate COW file */ 771 FI_ATOMIC_COMMITTED, /* indicate atomic commit completed except disk sync */ 772 FI_MAX, /* max flag, never be used */ 773 }; 774 775 struct f2fs_inode_info { 776 struct inode vfs_inode; /* serve a vfs inode */ 777 unsigned long i_flags; /* keep an inode flags for ioctl */ 778 unsigned char i_advise; /* use to give file attribute hints */ 779 unsigned char i_dir_level; /* use for dentry level for large dir */ 780 unsigned int i_current_depth; /* only for directory depth */ 781 /* for gc failure statistic */ 782 unsigned int i_gc_failures[MAX_GC_FAILURE]; 783 unsigned int i_pino; /* parent inode number */ 784 umode_t i_acl_mode; /* keep file acl mode temporarily */ 785 786 /* Use below internally in f2fs*/ 787 unsigned long flags[BITS_TO_LONGS(FI_MAX)]; /* use to pass per-file flags */ 788 struct f2fs_rwsem i_sem; /* protect fi info */ 789 atomic_t dirty_pages; /* # of dirty pages */ 790 f2fs_hash_t chash; /* hash value of given file name */ 791 unsigned int clevel; /* maximum level of given file name */ 792 struct task_struct *task; /* lookup and create consistency */ 793 struct task_struct *cp_task; /* separate cp/wb IO stats*/ 794 struct task_struct *wb_task; /* indicate inode is in context of writeback */ 795 nid_t i_xattr_nid; /* node id that contains xattrs */ 796 loff_t last_disk_size; /* lastly written file size */ 797 spinlock_t i_size_lock; /* protect last_disk_size */ 798 799 #ifdef CONFIG_QUOTA 800 struct dquot *i_dquot[MAXQUOTAS]; 801 802 /* quota space reservation, managed internally by quota code */ 803 qsize_t i_reserved_quota; 804 #endif 805 struct list_head dirty_list; /* dirty list for dirs and files */ 806 struct list_head gdirty_list; /* linked in global dirty list */ 807 struct task_struct *atomic_write_task; /* store atomic write task */ 808 struct extent_tree *extent_tree; /* cached extent_tree entry */ 809 struct inode *cow_inode; /* copy-on-write inode for atomic write */ 810 811 /* avoid racing between foreground op and gc */ 812 struct f2fs_rwsem i_gc_rwsem[2]; 813 struct f2fs_rwsem i_xattr_sem; /* avoid racing between reading and changing EAs */ 814 815 int i_extra_isize; /* size of extra space located in i_addr */ 816 kprojid_t i_projid; /* id for project quota */ 817 int i_inline_xattr_size; /* inline xattr size */ 818 struct timespec64 i_crtime; /* inode creation time */ 819 struct timespec64 i_disk_time[4];/* inode disk times */ 820 821 /* for file compress */ 822 atomic_t i_compr_blocks; /* # of compressed blocks */ 823 unsigned char i_compress_algorithm; /* algorithm type */ 824 unsigned char i_log_cluster_size; /* log of cluster size */ 825 unsigned char i_compress_level; /* compress level (lz4hc,zstd) */ 826 unsigned short i_compress_flag; /* compress flag */ 827 unsigned int i_cluster_size; /* cluster size */ 828 829 unsigned int atomic_write_cnt; 830 loff_t original_i_size; /* original i_size before atomic write */ 831 }; 832 833 static inline void get_extent_info(struct extent_info *ext, 834 struct f2fs_extent *i_ext) 835 { 836 ext->fofs = le32_to_cpu(i_ext->fofs); 837 ext->blk = le32_to_cpu(i_ext->blk); 838 ext->len = le32_to_cpu(i_ext->len); 839 } 840 841 static inline void set_raw_extent(struct extent_info *ext, 842 struct f2fs_extent *i_ext) 843 { 844 i_ext->fofs = cpu_to_le32(ext->fofs); 845 i_ext->blk = cpu_to_le32(ext->blk); 846 i_ext->len = cpu_to_le32(ext->len); 847 } 848 849 static inline void set_extent_info(struct extent_info *ei, unsigned int fofs, 850 u32 blk, unsigned int len) 851 { 852 ei->fofs = fofs; 853 ei->blk = blk; 854 ei->len = len; 855 #ifdef CONFIG_F2FS_FS_COMPRESSION 856 ei->c_len = 0; 857 #endif 858 } 859 860 static inline bool __is_discard_mergeable(struct discard_info *back, 861 struct discard_info *front, unsigned int max_len) 862 { 863 return (back->lstart + back->len == front->lstart) && 864 (back->len + front->len <= max_len); 865 } 866 867 static inline bool __is_discard_back_mergeable(struct discard_info *cur, 868 struct discard_info *back, unsigned int max_len) 869 { 870 return __is_discard_mergeable(back, cur, max_len); 871 } 872 873 static inline bool __is_discard_front_mergeable(struct discard_info *cur, 874 struct discard_info *front, unsigned int max_len) 875 { 876 return __is_discard_mergeable(cur, front, max_len); 877 } 878 879 static inline bool __is_extent_mergeable(struct extent_info *back, 880 struct extent_info *front) 881 { 882 #ifdef CONFIG_F2FS_FS_COMPRESSION 883 if (back->c_len && back->len != back->c_len) 884 return false; 885 if (front->c_len && front->len != front->c_len) 886 return false; 887 #endif 888 return (back->fofs + back->len == front->fofs && 889 back->blk + back->len == front->blk); 890 } 891 892 static inline bool __is_back_mergeable(struct extent_info *cur, 893 struct extent_info *back) 894 { 895 return __is_extent_mergeable(back, cur); 896 } 897 898 static inline bool __is_front_mergeable(struct extent_info *cur, 899 struct extent_info *front) 900 { 901 return __is_extent_mergeable(cur, front); 902 } 903 904 extern void f2fs_mark_inode_dirty_sync(struct inode *inode, bool sync); 905 static inline void __try_update_largest_extent(struct extent_tree *et, 906 struct extent_node *en) 907 { 908 if (en->ei.len > et->largest.len) { 909 et->largest = en->ei; 910 et->largest_updated = true; 911 } 912 } 913 914 /* 915 * For free nid management 916 */ 917 enum nid_state { 918 FREE_NID, /* newly added to free nid list */ 919 PREALLOC_NID, /* it is preallocated */ 920 MAX_NID_STATE, 921 }; 922 923 enum nat_state { 924 TOTAL_NAT, 925 DIRTY_NAT, 926 RECLAIMABLE_NAT, 927 MAX_NAT_STATE, 928 }; 929 930 struct f2fs_nm_info { 931 block_t nat_blkaddr; /* base disk address of NAT */ 932 nid_t max_nid; /* maximum possible node ids */ 933 nid_t available_nids; /* # of available node ids */ 934 nid_t next_scan_nid; /* the next nid to be scanned */ 935 nid_t max_rf_node_blocks; /* max # of nodes for recovery */ 936 unsigned int ram_thresh; /* control the memory footprint */ 937 unsigned int ra_nid_pages; /* # of nid pages to be readaheaded */ 938 unsigned int dirty_nats_ratio; /* control dirty nats ratio threshold */ 939 940 /* NAT cache management */ 941 struct radix_tree_root nat_root;/* root of the nat entry cache */ 942 struct radix_tree_root nat_set_root;/* root of the nat set cache */ 943 struct f2fs_rwsem nat_tree_lock; /* protect nat entry tree */ 944 struct list_head nat_entries; /* cached nat entry list (clean) */ 945 spinlock_t nat_list_lock; /* protect clean nat entry list */ 946 unsigned int nat_cnt[MAX_NAT_STATE]; /* the # of cached nat entries */ 947 unsigned int nat_blocks; /* # of nat blocks */ 948 949 /* free node ids management */ 950 struct radix_tree_root free_nid_root;/* root of the free_nid cache */ 951 struct list_head free_nid_list; /* list for free nids excluding preallocated nids */ 952 unsigned int nid_cnt[MAX_NID_STATE]; /* the number of free node id */ 953 spinlock_t nid_list_lock; /* protect nid lists ops */ 954 struct mutex build_lock; /* lock for build free nids */ 955 unsigned char **free_nid_bitmap; 956 unsigned char *nat_block_bitmap; 957 unsigned short *free_nid_count; /* free nid count of NAT block */ 958 959 /* for checkpoint */ 960 char *nat_bitmap; /* NAT bitmap pointer */ 961 962 unsigned int nat_bits_blocks; /* # of nat bits blocks */ 963 unsigned char *nat_bits; /* NAT bits blocks */ 964 unsigned char *full_nat_bits; /* full NAT pages */ 965 unsigned char *empty_nat_bits; /* empty NAT pages */ 966 #ifdef CONFIG_F2FS_CHECK_FS 967 char *nat_bitmap_mir; /* NAT bitmap mirror */ 968 #endif 969 int bitmap_size; /* bitmap size */ 970 }; 971 972 /* 973 * this structure is used as one of function parameters. 974 * all the information are dedicated to a given direct node block determined 975 * by the data offset in a file. 976 */ 977 struct dnode_of_data { 978 struct inode *inode; /* vfs inode pointer */ 979 struct page *inode_page; /* its inode page, NULL is possible */ 980 struct page *node_page; /* cached direct node page */ 981 nid_t nid; /* node id of the direct node block */ 982 unsigned int ofs_in_node; /* data offset in the node page */ 983 bool inode_page_locked; /* inode page is locked or not */ 984 bool node_changed; /* is node block changed */ 985 char cur_level; /* level of hole node page */ 986 char max_level; /* level of current page located */ 987 block_t data_blkaddr; /* block address of the node block */ 988 }; 989 990 static inline void set_new_dnode(struct dnode_of_data *dn, struct inode *inode, 991 struct page *ipage, struct page *npage, nid_t nid) 992 { 993 memset(dn, 0, sizeof(*dn)); 994 dn->inode = inode; 995 dn->inode_page = ipage; 996 dn->node_page = npage; 997 dn->nid = nid; 998 } 999 1000 /* 1001 * For SIT manager 1002 * 1003 * By default, there are 6 active log areas across the whole main area. 1004 * When considering hot and cold data separation to reduce cleaning overhead, 1005 * we split 3 for data logs and 3 for node logs as hot, warm, and cold types, 1006 * respectively. 1007 * In the current design, you should not change the numbers intentionally. 1008 * Instead, as a mount option such as active_logs=x, you can use 2, 4, and 6 1009 * logs individually according to the underlying devices. (default: 6) 1010 * Just in case, on-disk layout covers maximum 16 logs that consist of 8 for 1011 * data and 8 for node logs. 1012 */ 1013 #define NR_CURSEG_DATA_TYPE (3) 1014 #define NR_CURSEG_NODE_TYPE (3) 1015 #define NR_CURSEG_INMEM_TYPE (2) 1016 #define NR_CURSEG_RO_TYPE (2) 1017 #define NR_CURSEG_PERSIST_TYPE (NR_CURSEG_DATA_TYPE + NR_CURSEG_NODE_TYPE) 1018 #define NR_CURSEG_TYPE (NR_CURSEG_INMEM_TYPE + NR_CURSEG_PERSIST_TYPE) 1019 1020 enum { 1021 CURSEG_HOT_DATA = 0, /* directory entry blocks */ 1022 CURSEG_WARM_DATA, /* data blocks */ 1023 CURSEG_COLD_DATA, /* multimedia or GCed data blocks */ 1024 CURSEG_HOT_NODE, /* direct node blocks of directory files */ 1025 CURSEG_WARM_NODE, /* direct node blocks of normal files */ 1026 CURSEG_COLD_NODE, /* indirect node blocks */ 1027 NR_PERSISTENT_LOG, /* number of persistent log */ 1028 CURSEG_COLD_DATA_PINNED = NR_PERSISTENT_LOG, 1029 /* pinned file that needs consecutive block address */ 1030 CURSEG_ALL_DATA_ATGC, /* SSR alloctor in hot/warm/cold data area */ 1031 NO_CHECK_TYPE, /* number of persistent & inmem log */ 1032 }; 1033 1034 struct flush_cmd { 1035 struct completion wait; 1036 struct llist_node llnode; 1037 nid_t ino; 1038 int ret; 1039 }; 1040 1041 struct flush_cmd_control { 1042 struct task_struct *f2fs_issue_flush; /* flush thread */ 1043 wait_queue_head_t flush_wait_queue; /* waiting queue for wake-up */ 1044 atomic_t issued_flush; /* # of issued flushes */ 1045 atomic_t queued_flush; /* # of queued flushes */ 1046 struct llist_head issue_list; /* list for command issue */ 1047 struct llist_node *dispatch_list; /* list for command dispatch */ 1048 }; 1049 1050 struct f2fs_sm_info { 1051 struct sit_info *sit_info; /* whole segment information */ 1052 struct free_segmap_info *free_info; /* free segment information */ 1053 struct dirty_seglist_info *dirty_info; /* dirty segment information */ 1054 struct curseg_info *curseg_array; /* active segment information */ 1055 1056 struct f2fs_rwsem curseg_lock; /* for preventing curseg change */ 1057 1058 block_t seg0_blkaddr; /* block address of 0'th segment */ 1059 block_t main_blkaddr; /* start block address of main area */ 1060 block_t ssa_blkaddr; /* start block address of SSA area */ 1061 1062 unsigned int segment_count; /* total # of segments */ 1063 unsigned int main_segments; /* # of segments in main area */ 1064 unsigned int reserved_segments; /* # of reserved segments */ 1065 unsigned int additional_reserved_segments;/* reserved segs for IO align feature */ 1066 unsigned int ovp_segments; /* # of overprovision segments */ 1067 1068 /* a threshold to reclaim prefree segments */ 1069 unsigned int rec_prefree_segments; 1070 1071 struct list_head sit_entry_set; /* sit entry set list */ 1072 1073 unsigned int ipu_policy; /* in-place-update policy */ 1074 unsigned int min_ipu_util; /* in-place-update threshold */ 1075 unsigned int min_fsync_blocks; /* threshold for fsync */ 1076 unsigned int min_seq_blocks; /* threshold for sequential blocks */ 1077 unsigned int min_hot_blocks; /* threshold for hot block allocation */ 1078 unsigned int min_ssr_sections; /* threshold to trigger SSR allocation */ 1079 1080 /* for flush command control */ 1081 struct flush_cmd_control *fcc_info; 1082 1083 /* for discard command control */ 1084 struct discard_cmd_control *dcc_info; 1085 }; 1086 1087 /* 1088 * For superblock 1089 */ 1090 /* 1091 * COUNT_TYPE for monitoring 1092 * 1093 * f2fs monitors the number of several block types such as on-writeback, 1094 * dirty dentry blocks, dirty node blocks, and dirty meta blocks. 1095 */ 1096 #define WB_DATA_TYPE(p) (__is_cp_guaranteed(p) ? F2FS_WB_CP_DATA : F2FS_WB_DATA) 1097 enum count_type { 1098 F2FS_DIRTY_DENTS, 1099 F2FS_DIRTY_DATA, 1100 F2FS_DIRTY_QDATA, 1101 F2FS_DIRTY_NODES, 1102 F2FS_DIRTY_META, 1103 F2FS_DIRTY_IMETA, 1104 F2FS_WB_CP_DATA, 1105 F2FS_WB_DATA, 1106 F2FS_RD_DATA, 1107 F2FS_RD_NODE, 1108 F2FS_RD_META, 1109 F2FS_DIO_WRITE, 1110 F2FS_DIO_READ, 1111 NR_COUNT_TYPE, 1112 }; 1113 1114 /* 1115 * The below are the page types of bios used in submit_bio(). 1116 * The available types are: 1117 * DATA User data pages. It operates as async mode. 1118 * NODE Node pages. It operates as async mode. 1119 * META FS metadata pages such as SIT, NAT, CP. 1120 * NR_PAGE_TYPE The number of page types. 1121 * META_FLUSH Make sure the previous pages are written 1122 * with waiting the bio's completion 1123 * ... Only can be used with META. 1124 */ 1125 #define PAGE_TYPE_OF_BIO(type) ((type) > META ? META : (type)) 1126 enum page_type { 1127 DATA = 0, 1128 NODE = 1, /* should not change this */ 1129 META, 1130 NR_PAGE_TYPE, 1131 META_FLUSH, 1132 IPU, /* the below types are used by tracepoints only. */ 1133 OPU, 1134 }; 1135 1136 enum temp_type { 1137 HOT = 0, /* must be zero for meta bio */ 1138 WARM, 1139 COLD, 1140 NR_TEMP_TYPE, 1141 }; 1142 1143 enum need_lock_type { 1144 LOCK_REQ = 0, 1145 LOCK_DONE, 1146 LOCK_RETRY, 1147 }; 1148 1149 enum cp_reason_type { 1150 CP_NO_NEEDED, 1151 CP_NON_REGULAR, 1152 CP_COMPRESSED, 1153 CP_HARDLINK, 1154 CP_SB_NEED_CP, 1155 CP_WRONG_PINO, 1156 CP_NO_SPC_ROLL, 1157 CP_NODE_NEED_CP, 1158 CP_FASTBOOT_MODE, 1159 CP_SPEC_LOG_NUM, 1160 CP_RECOVER_DIR, 1161 }; 1162 1163 enum iostat_type { 1164 /* WRITE IO */ 1165 APP_DIRECT_IO, /* app direct write IOs */ 1166 APP_BUFFERED_IO, /* app buffered write IOs */ 1167 APP_WRITE_IO, /* app write IOs */ 1168 APP_MAPPED_IO, /* app mapped IOs */ 1169 APP_BUFFERED_CDATA_IO, /* app buffered write IOs on compressed file */ 1170 APP_MAPPED_CDATA_IO, /* app mapped write IOs on compressed file */ 1171 FS_DATA_IO, /* data IOs from kworker/fsync/reclaimer */ 1172 FS_CDATA_IO, /* data IOs from kworker/fsync/reclaimer on compressed file */ 1173 FS_NODE_IO, /* node IOs from kworker/fsync/reclaimer */ 1174 FS_META_IO, /* meta IOs from kworker/reclaimer */ 1175 FS_GC_DATA_IO, /* data IOs from forground gc */ 1176 FS_GC_NODE_IO, /* node IOs from forground gc */ 1177 FS_CP_DATA_IO, /* data IOs from checkpoint */ 1178 FS_CP_NODE_IO, /* node IOs from checkpoint */ 1179 FS_CP_META_IO, /* meta IOs from checkpoint */ 1180 1181 /* READ IO */ 1182 APP_DIRECT_READ_IO, /* app direct read IOs */ 1183 APP_BUFFERED_READ_IO, /* app buffered read IOs */ 1184 APP_READ_IO, /* app read IOs */ 1185 APP_MAPPED_READ_IO, /* app mapped read IOs */ 1186 APP_BUFFERED_CDATA_READ_IO, /* app buffered read IOs on compressed file */ 1187 APP_MAPPED_CDATA_READ_IO, /* app mapped read IOs on compressed file */ 1188 FS_DATA_READ_IO, /* data read IOs */ 1189 FS_GDATA_READ_IO, /* data read IOs from background gc */ 1190 FS_CDATA_READ_IO, /* compressed data read IOs */ 1191 FS_NODE_READ_IO, /* node read IOs */ 1192 FS_META_READ_IO, /* meta read IOs */ 1193 1194 /* other */ 1195 FS_DISCARD, /* discard */ 1196 NR_IO_TYPE, 1197 }; 1198 1199 struct f2fs_io_info { 1200 struct f2fs_sb_info *sbi; /* f2fs_sb_info pointer */ 1201 nid_t ino; /* inode number */ 1202 enum page_type type; /* contains DATA/NODE/META/META_FLUSH */ 1203 enum temp_type temp; /* contains HOT/WARM/COLD */ 1204 enum req_op op; /* contains REQ_OP_ */ 1205 blk_opf_t op_flags; /* req_flag_bits */ 1206 block_t new_blkaddr; /* new block address to be written */ 1207 block_t old_blkaddr; /* old block address before Cow */ 1208 struct page *page; /* page to be written */ 1209 struct page *encrypted_page; /* encrypted page */ 1210 struct page *compressed_page; /* compressed page */ 1211 struct list_head list; /* serialize IOs */ 1212 bool submitted; /* indicate IO submission */ 1213 int need_lock; /* indicate we need to lock cp_rwsem */ 1214 bool in_list; /* indicate fio is in io_list */ 1215 bool is_por; /* indicate IO is from recovery or not */ 1216 bool retry; /* need to reallocate block address */ 1217 int compr_blocks; /* # of compressed block addresses */ 1218 bool encrypted; /* indicate file is encrypted */ 1219 bool post_read; /* require post read */ 1220 enum iostat_type io_type; /* io type */ 1221 struct writeback_control *io_wbc; /* writeback control */ 1222 struct bio **bio; /* bio for ipu */ 1223 sector_t *last_block; /* last block number in bio */ 1224 unsigned char version; /* version of the node */ 1225 }; 1226 1227 struct bio_entry { 1228 struct bio *bio; 1229 struct list_head list; 1230 }; 1231 1232 #define is_read_io(rw) ((rw) == READ) 1233 struct f2fs_bio_info { 1234 struct f2fs_sb_info *sbi; /* f2fs superblock */ 1235 struct bio *bio; /* bios to merge */ 1236 sector_t last_block_in_bio; /* last block number */ 1237 struct f2fs_io_info fio; /* store buffered io info. */ 1238 struct f2fs_rwsem io_rwsem; /* blocking op for bio */ 1239 spinlock_t io_lock; /* serialize DATA/NODE IOs */ 1240 struct list_head io_list; /* track fios */ 1241 struct list_head bio_list; /* bio entry list head */ 1242 struct f2fs_rwsem bio_list_lock; /* lock to protect bio entry list */ 1243 }; 1244 1245 #define FDEV(i) (sbi->devs[i]) 1246 #define RDEV(i) (raw_super->devs[i]) 1247 struct f2fs_dev_info { 1248 struct block_device *bdev; 1249 char path[MAX_PATH_LEN]; 1250 unsigned int total_segments; 1251 block_t start_blk; 1252 block_t end_blk; 1253 #ifdef CONFIG_BLK_DEV_ZONED 1254 unsigned int nr_blkz; /* Total number of zones */ 1255 unsigned long *blkz_seq; /* Bitmap indicating sequential zones */ 1256 #endif 1257 }; 1258 1259 enum inode_type { 1260 DIR_INODE, /* for dirty dir inode */ 1261 FILE_INODE, /* for dirty regular/symlink inode */ 1262 DIRTY_META, /* for all dirtied inode metadata */ 1263 NR_INODE_TYPE, 1264 }; 1265 1266 /* for inner inode cache management */ 1267 struct inode_management { 1268 struct radix_tree_root ino_root; /* ino entry array */ 1269 spinlock_t ino_lock; /* for ino entry lock */ 1270 struct list_head ino_list; /* inode list head */ 1271 unsigned long ino_num; /* number of entries */ 1272 }; 1273 1274 /* for GC_AT */ 1275 struct atgc_management { 1276 bool atgc_enabled; /* ATGC is enabled or not */ 1277 struct rb_root_cached root; /* root of victim rb-tree */ 1278 struct list_head victim_list; /* linked with all victim entries */ 1279 unsigned int victim_count; /* victim count in rb-tree */ 1280 unsigned int candidate_ratio; /* candidate ratio */ 1281 unsigned int max_candidate_count; /* max candidate count */ 1282 unsigned int age_weight; /* age weight, vblock_weight = 100 - age_weight */ 1283 unsigned long long age_threshold; /* age threshold */ 1284 }; 1285 1286 struct f2fs_gc_control { 1287 unsigned int victim_segno; /* target victim segment number */ 1288 int init_gc_type; /* FG_GC or BG_GC */ 1289 bool no_bg_gc; /* check the space and stop bg_gc */ 1290 bool should_migrate_blocks; /* should migrate blocks */ 1291 bool err_gc_skipped; /* return EAGAIN if GC skipped */ 1292 unsigned int nr_free_secs; /* # of free sections to do GC */ 1293 }; 1294 1295 /* For s_flag in struct f2fs_sb_info */ 1296 enum { 1297 SBI_IS_DIRTY, /* dirty flag for checkpoint */ 1298 SBI_IS_CLOSE, /* specify unmounting */ 1299 SBI_NEED_FSCK, /* need fsck.f2fs to fix */ 1300 SBI_POR_DOING, /* recovery is doing or not */ 1301 SBI_NEED_SB_WRITE, /* need to recover superblock */ 1302 SBI_NEED_CP, /* need to checkpoint */ 1303 SBI_IS_SHUTDOWN, /* shutdown by ioctl */ 1304 SBI_IS_RECOVERED, /* recovered orphan/data */ 1305 SBI_CP_DISABLED, /* CP was disabled last mount */ 1306 SBI_CP_DISABLED_QUICK, /* CP was disabled quickly */ 1307 SBI_QUOTA_NEED_FLUSH, /* need to flush quota info in CP */ 1308 SBI_QUOTA_SKIP_FLUSH, /* skip flushing quota in current CP */ 1309 SBI_QUOTA_NEED_REPAIR, /* quota file may be corrupted */ 1310 SBI_IS_RESIZEFS, /* resizefs is in process */ 1311 SBI_IS_FREEZING, /* freezefs is in process */ 1312 }; 1313 1314 enum { 1315 CP_TIME, 1316 REQ_TIME, 1317 DISCARD_TIME, 1318 GC_TIME, 1319 DISABLE_TIME, 1320 UMOUNT_DISCARD_TIMEOUT, 1321 MAX_TIME, 1322 }; 1323 1324 /* Note that you need to keep synchronization with this gc_mode_names array */ 1325 enum { 1326 GC_NORMAL, 1327 GC_IDLE_CB, 1328 GC_IDLE_GREEDY, 1329 GC_IDLE_AT, 1330 GC_URGENT_HIGH, 1331 GC_URGENT_LOW, 1332 GC_URGENT_MID, 1333 MAX_GC_MODE, 1334 }; 1335 1336 enum { 1337 BGGC_MODE_ON, /* background gc is on */ 1338 BGGC_MODE_OFF, /* background gc is off */ 1339 BGGC_MODE_SYNC, /* 1340 * background gc is on, migrating blocks 1341 * like foreground gc 1342 */ 1343 }; 1344 1345 enum { 1346 FS_MODE_ADAPTIVE, /* use both lfs/ssr allocation */ 1347 FS_MODE_LFS, /* use lfs allocation only */ 1348 FS_MODE_FRAGMENT_SEG, /* segment fragmentation mode */ 1349 FS_MODE_FRAGMENT_BLK, /* block fragmentation mode */ 1350 }; 1351 1352 enum { 1353 ALLOC_MODE_DEFAULT, /* stay default */ 1354 ALLOC_MODE_REUSE, /* reuse segments as much as possible */ 1355 }; 1356 1357 enum fsync_mode { 1358 FSYNC_MODE_POSIX, /* fsync follows posix semantics */ 1359 FSYNC_MODE_STRICT, /* fsync behaves in line with ext4 */ 1360 FSYNC_MODE_NOBARRIER, /* fsync behaves nobarrier based on posix */ 1361 }; 1362 1363 enum { 1364 COMPR_MODE_FS, /* 1365 * automatically compress compression 1366 * enabled files 1367 */ 1368 COMPR_MODE_USER, /* 1369 * automatical compression is disabled. 1370 * user can control the file compression 1371 * using ioctls 1372 */ 1373 }; 1374 1375 enum { 1376 DISCARD_UNIT_BLOCK, /* basic discard unit is block */ 1377 DISCARD_UNIT_SEGMENT, /* basic discard unit is segment */ 1378 DISCARD_UNIT_SECTION, /* basic discard unit is section */ 1379 }; 1380 1381 enum { 1382 MEMORY_MODE_NORMAL, /* memory mode for normal devices */ 1383 MEMORY_MODE_LOW, /* memory mode for low memry devices */ 1384 }; 1385 1386 1387 1388 static inline int f2fs_test_bit(unsigned int nr, char *addr); 1389 static inline void f2fs_set_bit(unsigned int nr, char *addr); 1390 static inline void f2fs_clear_bit(unsigned int nr, char *addr); 1391 1392 /* 1393 * Layout of f2fs page.private: 1394 * 1395 * Layout A: lowest bit should be 1 1396 * | bit0 = 1 | bit1 | bit2 | ... | bit MAX | private data .... | 1397 * bit 0 PAGE_PRIVATE_NOT_POINTER 1398 * bit 1 PAGE_PRIVATE_ATOMIC_WRITE 1399 * bit 2 PAGE_PRIVATE_DUMMY_WRITE 1400 * bit 3 PAGE_PRIVATE_ONGOING_MIGRATION 1401 * bit 4 PAGE_PRIVATE_INLINE_INODE 1402 * bit 5 PAGE_PRIVATE_REF_RESOURCE 1403 * bit 6- f2fs private data 1404 * 1405 * Layout B: lowest bit should be 0 1406 * page.private is a wrapped pointer. 1407 */ 1408 enum { 1409 PAGE_PRIVATE_NOT_POINTER, /* private contains non-pointer data */ 1410 PAGE_PRIVATE_ATOMIC_WRITE, /* data page from atomic write path */ 1411 PAGE_PRIVATE_DUMMY_WRITE, /* data page for padding aligned IO */ 1412 PAGE_PRIVATE_ONGOING_MIGRATION, /* data page which is on-going migrating */ 1413 PAGE_PRIVATE_INLINE_INODE, /* inode page contains inline data */ 1414 PAGE_PRIVATE_REF_RESOURCE, /* dirty page has referenced resources */ 1415 PAGE_PRIVATE_MAX 1416 }; 1417 1418 #define PAGE_PRIVATE_GET_FUNC(name, flagname) \ 1419 static inline bool page_private_##name(struct page *page) \ 1420 { \ 1421 return PagePrivate(page) && \ 1422 test_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)) && \ 1423 test_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \ 1424 } 1425 1426 #define PAGE_PRIVATE_SET_FUNC(name, flagname) \ 1427 static inline void set_page_private_##name(struct page *page) \ 1428 { \ 1429 if (!PagePrivate(page)) { \ 1430 get_page(page); \ 1431 SetPagePrivate(page); \ 1432 set_page_private(page, 0); \ 1433 } \ 1434 set_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)); \ 1435 set_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \ 1436 } 1437 1438 #define PAGE_PRIVATE_CLEAR_FUNC(name, flagname) \ 1439 static inline void clear_page_private_##name(struct page *page) \ 1440 { \ 1441 clear_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \ 1442 if (page_private(page) == 1 << PAGE_PRIVATE_NOT_POINTER) { \ 1443 set_page_private(page, 0); \ 1444 if (PagePrivate(page)) { \ 1445 ClearPagePrivate(page); \ 1446 put_page(page); \ 1447 }\ 1448 } \ 1449 } 1450 1451 PAGE_PRIVATE_GET_FUNC(nonpointer, NOT_POINTER); 1452 PAGE_PRIVATE_GET_FUNC(reference, REF_RESOURCE); 1453 PAGE_PRIVATE_GET_FUNC(inline, INLINE_INODE); 1454 PAGE_PRIVATE_GET_FUNC(gcing, ONGOING_MIGRATION); 1455 PAGE_PRIVATE_GET_FUNC(atomic, ATOMIC_WRITE); 1456 PAGE_PRIVATE_GET_FUNC(dummy, DUMMY_WRITE); 1457 1458 PAGE_PRIVATE_SET_FUNC(reference, REF_RESOURCE); 1459 PAGE_PRIVATE_SET_FUNC(inline, INLINE_INODE); 1460 PAGE_PRIVATE_SET_FUNC(gcing, ONGOING_MIGRATION); 1461 PAGE_PRIVATE_SET_FUNC(atomic, ATOMIC_WRITE); 1462 PAGE_PRIVATE_SET_FUNC(dummy, DUMMY_WRITE); 1463 1464 PAGE_PRIVATE_CLEAR_FUNC(reference, REF_RESOURCE); 1465 PAGE_PRIVATE_CLEAR_FUNC(inline, INLINE_INODE); 1466 PAGE_PRIVATE_CLEAR_FUNC(gcing, ONGOING_MIGRATION); 1467 PAGE_PRIVATE_CLEAR_FUNC(atomic, ATOMIC_WRITE); 1468 PAGE_PRIVATE_CLEAR_FUNC(dummy, DUMMY_WRITE); 1469 1470 static inline unsigned long get_page_private_data(struct page *page) 1471 { 1472 unsigned long data = page_private(page); 1473 1474 if (!test_bit(PAGE_PRIVATE_NOT_POINTER, &data)) 1475 return 0; 1476 return data >> PAGE_PRIVATE_MAX; 1477 } 1478 1479 static inline void set_page_private_data(struct page *page, unsigned long data) 1480 { 1481 if (!PagePrivate(page)) { 1482 get_page(page); 1483 SetPagePrivate(page); 1484 set_page_private(page, 0); 1485 } 1486 set_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)); 1487 page_private(page) |= data << PAGE_PRIVATE_MAX; 1488 } 1489 1490 static inline void clear_page_private_data(struct page *page) 1491 { 1492 page_private(page) &= (1 << PAGE_PRIVATE_MAX) - 1; 1493 if (page_private(page) == 1 << PAGE_PRIVATE_NOT_POINTER) { 1494 set_page_private(page, 0); 1495 if (PagePrivate(page)) { 1496 ClearPagePrivate(page); 1497 put_page(page); 1498 } 1499 } 1500 } 1501 1502 /* For compression */ 1503 enum compress_algorithm_type { 1504 COMPRESS_LZO, 1505 COMPRESS_LZ4, 1506 COMPRESS_ZSTD, 1507 COMPRESS_LZORLE, 1508 COMPRESS_MAX, 1509 }; 1510 1511 enum compress_flag { 1512 COMPRESS_CHKSUM, 1513 COMPRESS_MAX_FLAG, 1514 }; 1515 1516 #define COMPRESS_WATERMARK 20 1517 #define COMPRESS_PERCENT 20 1518 1519 #define COMPRESS_DATA_RESERVED_SIZE 4 1520 struct compress_data { 1521 __le32 clen; /* compressed data size */ 1522 __le32 chksum; /* compressed data chksum */ 1523 __le32 reserved[COMPRESS_DATA_RESERVED_SIZE]; /* reserved */ 1524 u8 cdata[]; /* compressed data */ 1525 }; 1526 1527 #define COMPRESS_HEADER_SIZE (sizeof(struct compress_data)) 1528 1529 #define F2FS_COMPRESSED_PAGE_MAGIC 0xF5F2C000 1530 1531 #define COMPRESS_LEVEL_OFFSET 8 1532 1533 /* compress context */ 1534 struct compress_ctx { 1535 struct inode *inode; /* inode the context belong to */ 1536 pgoff_t cluster_idx; /* cluster index number */ 1537 unsigned int cluster_size; /* page count in cluster */ 1538 unsigned int log_cluster_size; /* log of cluster size */ 1539 struct page **rpages; /* pages store raw data in cluster */ 1540 unsigned int nr_rpages; /* total page number in rpages */ 1541 struct page **cpages; /* pages store compressed data in cluster */ 1542 unsigned int nr_cpages; /* total page number in cpages */ 1543 unsigned int valid_nr_cpages; /* valid page number in cpages */ 1544 void *rbuf; /* virtual mapped address on rpages */ 1545 struct compress_data *cbuf; /* virtual mapped address on cpages */ 1546 size_t rlen; /* valid data length in rbuf */ 1547 size_t clen; /* valid data length in cbuf */ 1548 void *private; /* payload buffer for specified compression algorithm */ 1549 void *private2; /* extra payload buffer */ 1550 }; 1551 1552 /* compress context for write IO path */ 1553 struct compress_io_ctx { 1554 u32 magic; /* magic number to indicate page is compressed */ 1555 struct inode *inode; /* inode the context belong to */ 1556 struct page **rpages; /* pages store raw data in cluster */ 1557 unsigned int nr_rpages; /* total page number in rpages */ 1558 atomic_t pending_pages; /* in-flight compressed page count */ 1559 }; 1560 1561 /* Context for decompressing one cluster on the read IO path */ 1562 struct decompress_io_ctx { 1563 u32 magic; /* magic number to indicate page is compressed */ 1564 struct inode *inode; /* inode the context belong to */ 1565 pgoff_t cluster_idx; /* cluster index number */ 1566 unsigned int cluster_size; /* page count in cluster */ 1567 unsigned int log_cluster_size; /* log of cluster size */ 1568 struct page **rpages; /* pages store raw data in cluster */ 1569 unsigned int nr_rpages; /* total page number in rpages */ 1570 struct page **cpages; /* pages store compressed data in cluster */ 1571 unsigned int nr_cpages; /* total page number in cpages */ 1572 struct page **tpages; /* temp pages to pad holes in cluster */ 1573 void *rbuf; /* virtual mapped address on rpages */ 1574 struct compress_data *cbuf; /* virtual mapped address on cpages */ 1575 size_t rlen; /* valid data length in rbuf */ 1576 size_t clen; /* valid data length in cbuf */ 1577 1578 /* 1579 * The number of compressed pages remaining to be read in this cluster. 1580 * This is initially nr_cpages. It is decremented by 1 each time a page 1581 * has been read (or failed to be read). When it reaches 0, the cluster 1582 * is decompressed (or an error is reported). 1583 * 1584 * If an error occurs before all the pages have been submitted for I/O, 1585 * then this will never reach 0. In this case the I/O submitter is 1586 * responsible for calling f2fs_decompress_end_io() instead. 1587 */ 1588 atomic_t remaining_pages; 1589 1590 /* 1591 * Number of references to this decompress_io_ctx. 1592 * 1593 * One reference is held for I/O completion. This reference is dropped 1594 * after the pagecache pages are updated and unlocked -- either after 1595 * decompression (and verity if enabled), or after an error. 1596 * 1597 * In addition, each compressed page holds a reference while it is in a 1598 * bio. These references are necessary prevent compressed pages from 1599 * being freed while they are still in a bio. 1600 */ 1601 refcount_t refcnt; 1602 1603 bool failed; /* IO error occurred before decompression? */ 1604 bool need_verity; /* need fs-verity verification after decompression? */ 1605 void *private; /* payload buffer for specified decompression algorithm */ 1606 void *private2; /* extra payload buffer */ 1607 struct work_struct verity_work; /* work to verify the decompressed pages */ 1608 struct work_struct free_work; /* work for late free this structure itself */ 1609 }; 1610 1611 #define NULL_CLUSTER ((unsigned int)(~0)) 1612 #define MIN_COMPRESS_LOG_SIZE 2 1613 #define MAX_COMPRESS_LOG_SIZE 8 1614 #define MAX_COMPRESS_WINDOW_SIZE(log_size) ((PAGE_SIZE) << (log_size)) 1615 1616 struct f2fs_sb_info { 1617 struct super_block *sb; /* pointer to VFS super block */ 1618 struct proc_dir_entry *s_proc; /* proc entry */ 1619 struct f2fs_super_block *raw_super; /* raw super block pointer */ 1620 struct f2fs_rwsem sb_lock; /* lock for raw super block */ 1621 int valid_super_block; /* valid super block no */ 1622 unsigned long s_flag; /* flags for sbi */ 1623 struct mutex writepages; /* mutex for writepages() */ 1624 1625 #ifdef CONFIG_BLK_DEV_ZONED 1626 unsigned int blocks_per_blkz; /* F2FS blocks per zone */ 1627 unsigned int log_blocks_per_blkz; /* log2 F2FS blocks per zone */ 1628 #endif 1629 1630 /* for node-related operations */ 1631 struct f2fs_nm_info *nm_info; /* node manager */ 1632 struct inode *node_inode; /* cache node blocks */ 1633 1634 /* for segment-related operations */ 1635 struct f2fs_sm_info *sm_info; /* segment manager */ 1636 1637 /* for bio operations */ 1638 struct f2fs_bio_info *write_io[NR_PAGE_TYPE]; /* for write bios */ 1639 /* keep migration IO order for LFS mode */ 1640 struct f2fs_rwsem io_order_lock; 1641 mempool_t *write_io_dummy; /* Dummy pages */ 1642 pgoff_t page_eio_ofs[NR_PAGE_TYPE]; /* EIO page offset */ 1643 int page_eio_cnt[NR_PAGE_TYPE]; /* EIO count */ 1644 1645 /* for checkpoint */ 1646 struct f2fs_checkpoint *ckpt; /* raw checkpoint pointer */ 1647 int cur_cp_pack; /* remain current cp pack */ 1648 spinlock_t cp_lock; /* for flag in ckpt */ 1649 struct inode *meta_inode; /* cache meta blocks */ 1650 struct f2fs_rwsem cp_global_sem; /* checkpoint procedure lock */ 1651 struct f2fs_rwsem cp_rwsem; /* blocking FS operations */ 1652 struct f2fs_rwsem node_write; /* locking node writes */ 1653 struct f2fs_rwsem node_change; /* locking node change */ 1654 wait_queue_head_t cp_wait; 1655 unsigned long last_time[MAX_TIME]; /* to store time in jiffies */ 1656 long interval_time[MAX_TIME]; /* to store thresholds */ 1657 struct ckpt_req_control cprc_info; /* for checkpoint request control */ 1658 1659 struct inode_management im[MAX_INO_ENTRY]; /* manage inode cache */ 1660 1661 spinlock_t fsync_node_lock; /* for node entry lock */ 1662 struct list_head fsync_node_list; /* node list head */ 1663 unsigned int fsync_seg_id; /* sequence id */ 1664 unsigned int fsync_node_num; /* number of node entries */ 1665 1666 /* for orphan inode, use 0'th array */ 1667 unsigned int max_orphans; /* max orphan inodes */ 1668 1669 /* for inode management */ 1670 struct list_head inode_list[NR_INODE_TYPE]; /* dirty inode list */ 1671 spinlock_t inode_lock[NR_INODE_TYPE]; /* for dirty inode list lock */ 1672 struct mutex flush_lock; /* for flush exclusion */ 1673 1674 /* for extent tree cache */ 1675 struct radix_tree_root extent_tree_root;/* cache extent cache entries */ 1676 struct mutex extent_tree_lock; /* locking extent radix tree */ 1677 struct list_head extent_list; /* lru list for shrinker */ 1678 spinlock_t extent_lock; /* locking extent lru list */ 1679 atomic_t total_ext_tree; /* extent tree count */ 1680 struct list_head zombie_list; /* extent zombie tree list */ 1681 atomic_t total_zombie_tree; /* extent zombie tree count */ 1682 atomic_t total_ext_node; /* extent info count */ 1683 1684 /* basic filesystem units */ 1685 unsigned int log_sectors_per_block; /* log2 sectors per block */ 1686 unsigned int log_blocksize; /* log2 block size */ 1687 unsigned int blocksize; /* block size */ 1688 unsigned int root_ino_num; /* root inode number*/ 1689 unsigned int node_ino_num; /* node inode number*/ 1690 unsigned int meta_ino_num; /* meta inode number*/ 1691 unsigned int log_blocks_per_seg; /* log2 blocks per segment */ 1692 unsigned int blocks_per_seg; /* blocks per segment */ 1693 unsigned int unusable_blocks_per_sec; /* unusable blocks per section */ 1694 unsigned int segs_per_sec; /* segments per section */ 1695 unsigned int secs_per_zone; /* sections per zone */ 1696 unsigned int total_sections; /* total section count */ 1697 unsigned int total_node_count; /* total node block count */ 1698 unsigned int total_valid_node_count; /* valid node block count */ 1699 int dir_level; /* directory level */ 1700 int readdir_ra; /* readahead inode in readdir */ 1701 u64 max_io_bytes; /* max io bytes to merge IOs */ 1702 1703 block_t user_block_count; /* # of user blocks */ 1704 block_t total_valid_block_count; /* # of valid blocks */ 1705 block_t discard_blks; /* discard command candidats */ 1706 block_t last_valid_block_count; /* for recovery */ 1707 block_t reserved_blocks; /* configurable reserved blocks */ 1708 block_t current_reserved_blocks; /* current reserved blocks */ 1709 1710 /* Additional tracking for no checkpoint mode */ 1711 block_t unusable_block_count; /* # of blocks saved by last cp */ 1712 1713 unsigned int nquota_files; /* # of quota sysfile */ 1714 struct f2fs_rwsem quota_sem; /* blocking cp for flags */ 1715 1716 /* # of pages, see count_type */ 1717 atomic_t nr_pages[NR_COUNT_TYPE]; 1718 /* # of allocated blocks */ 1719 struct percpu_counter alloc_valid_block_count; 1720 /* # of node block writes as roll forward recovery */ 1721 struct percpu_counter rf_node_block_count; 1722 1723 /* writeback control */ 1724 atomic_t wb_sync_req[META]; /* count # of WB_SYNC threads */ 1725 1726 /* valid inode count */ 1727 struct percpu_counter total_valid_inode_count; 1728 1729 struct f2fs_mount_info mount_opt; /* mount options */ 1730 1731 /* for cleaning operations */ 1732 struct f2fs_rwsem gc_lock; /* 1733 * semaphore for GC, avoid 1734 * race between GC and GC or CP 1735 */ 1736 struct f2fs_gc_kthread *gc_thread; /* GC thread */ 1737 struct atgc_management am; /* atgc management */ 1738 unsigned int cur_victim_sec; /* current victim section num */ 1739 unsigned int gc_mode; /* current GC state */ 1740 unsigned int next_victim_seg[2]; /* next segment in victim section */ 1741 spinlock_t gc_remaining_trials_lock; 1742 /* remaining trial count for GC_URGENT_* and GC_IDLE_* */ 1743 unsigned int gc_remaining_trials; 1744 1745 /* for skip statistic */ 1746 unsigned long long skipped_gc_rwsem; /* FG_GC only */ 1747 1748 /* threshold for gc trials on pinned files */ 1749 u64 gc_pin_file_threshold; 1750 struct f2fs_rwsem pin_sem; 1751 1752 /* maximum # of trials to find a victim segment for SSR and GC */ 1753 unsigned int max_victim_search; 1754 /* migration granularity of garbage collection, unit: segment */ 1755 unsigned int migration_granularity; 1756 1757 /* 1758 * for stat information. 1759 * one is for the LFS mode, and the other is for the SSR mode. 1760 */ 1761 #ifdef CONFIG_F2FS_STAT_FS 1762 struct f2fs_stat_info *stat_info; /* FS status information */ 1763 atomic_t meta_count[META_MAX]; /* # of meta blocks */ 1764 unsigned int segment_count[2]; /* # of allocated segments */ 1765 unsigned int block_count[2]; /* # of allocated blocks */ 1766 atomic_t inplace_count; /* # of inplace update */ 1767 atomic64_t total_hit_ext; /* # of lookup extent cache */ 1768 atomic64_t read_hit_rbtree; /* # of hit rbtree extent node */ 1769 atomic64_t read_hit_largest; /* # of hit largest extent node */ 1770 atomic64_t read_hit_cached; /* # of hit cached extent node */ 1771 atomic_t inline_xattr; /* # of inline_xattr inodes */ 1772 atomic_t inline_inode; /* # of inline_data inodes */ 1773 atomic_t inline_dir; /* # of inline_dentry inodes */ 1774 atomic_t compr_inode; /* # of compressed inodes */ 1775 atomic64_t compr_blocks; /* # of compressed blocks */ 1776 atomic_t swapfile_inode; /* # of swapfile inodes */ 1777 atomic_t atomic_files; /* # of opened atomic file */ 1778 atomic_t max_aw_cnt; /* max # of atomic writes */ 1779 unsigned int io_skip_bggc; /* skip background gc for in-flight IO */ 1780 unsigned int other_skip_bggc; /* skip background gc for other reasons */ 1781 unsigned int ndirty_inode[NR_INODE_TYPE]; /* # of dirty inodes */ 1782 #endif 1783 spinlock_t stat_lock; /* lock for stat operations */ 1784 1785 /* to attach REQ_META|REQ_FUA flags */ 1786 unsigned int data_io_flag; 1787 unsigned int node_io_flag; 1788 1789 /* For sysfs support */ 1790 struct kobject s_kobj; /* /sys/fs/f2fs/<devname> */ 1791 struct completion s_kobj_unregister; 1792 1793 struct kobject s_stat_kobj; /* /sys/fs/f2fs/<devname>/stat */ 1794 struct completion s_stat_kobj_unregister; 1795 1796 struct kobject s_feature_list_kobj; /* /sys/fs/f2fs/<devname>/feature_list */ 1797 struct completion s_feature_list_kobj_unregister; 1798 1799 /* For shrinker support */ 1800 struct list_head s_list; 1801 struct mutex umount_mutex; 1802 unsigned int shrinker_run_no; 1803 1804 /* For multi devices */ 1805 int s_ndevs; /* number of devices */ 1806 struct f2fs_dev_info *devs; /* for device list */ 1807 unsigned int dirty_device; /* for checkpoint data flush */ 1808 spinlock_t dev_lock; /* protect dirty_device */ 1809 bool aligned_blksize; /* all devices has the same logical blksize */ 1810 1811 /* For write statistics */ 1812 u64 sectors_written_start; 1813 u64 kbytes_written; 1814 1815 /* Reference to checksum algorithm driver via cryptoapi */ 1816 struct crypto_shash *s_chksum_driver; 1817 1818 /* Precomputed FS UUID checksum for seeding other checksums */ 1819 __u32 s_chksum_seed; 1820 1821 struct workqueue_struct *post_read_wq; /* post read workqueue */ 1822 1823 unsigned char errors[MAX_F2FS_ERRORS]; /* error flags */ 1824 spinlock_t error_lock; /* protect errors array */ 1825 bool error_dirty; /* errors of sb is dirty */ 1826 1827 struct kmem_cache *inline_xattr_slab; /* inline xattr entry */ 1828 unsigned int inline_xattr_slab_size; /* default inline xattr slab size */ 1829 1830 /* For reclaimed segs statistics per each GC mode */ 1831 unsigned int gc_segment_mode; /* GC state for reclaimed segments */ 1832 unsigned int gc_reclaimed_segs[MAX_GC_MODE]; /* Reclaimed segs for each mode */ 1833 1834 unsigned long seq_file_ra_mul; /* multiplier for ra_pages of seq. files in fadvise */ 1835 1836 int max_fragment_chunk; /* max chunk size for block fragmentation mode */ 1837 int max_fragment_hole; /* max hole size for block fragmentation mode */ 1838 1839 /* For atomic write statistics */ 1840 atomic64_t current_atomic_write; 1841 s64 peak_atomic_write; 1842 u64 committed_atomic_block; 1843 u64 revoked_atomic_block; 1844 1845 #ifdef CONFIG_F2FS_FS_COMPRESSION 1846 struct kmem_cache *page_array_slab; /* page array entry */ 1847 unsigned int page_array_slab_size; /* default page array slab size */ 1848 1849 /* For runtime compression statistics */ 1850 u64 compr_written_block; 1851 u64 compr_saved_block; 1852 u32 compr_new_inode; 1853 1854 /* For compressed block cache */ 1855 struct inode *compress_inode; /* cache compressed blocks */ 1856 unsigned int compress_percent; /* cache page percentage */ 1857 unsigned int compress_watermark; /* cache page watermark */ 1858 atomic_t compress_page_hit; /* cache hit count */ 1859 #endif 1860 1861 #ifdef CONFIG_F2FS_IOSTAT 1862 /* For app/fs IO statistics */ 1863 spinlock_t iostat_lock; 1864 unsigned long long rw_iostat[NR_IO_TYPE]; 1865 unsigned long long prev_rw_iostat[NR_IO_TYPE]; 1866 bool iostat_enable; 1867 unsigned long iostat_next_period; 1868 unsigned int iostat_period_ms; 1869 1870 /* For io latency related statistics info in one iostat period */ 1871 spinlock_t iostat_lat_lock; 1872 struct iostat_lat_info *iostat_io_lat; 1873 #endif 1874 }; 1875 1876 #ifdef CONFIG_F2FS_FAULT_INJECTION 1877 #define f2fs_show_injection_info(sbi, type) \ 1878 printk_ratelimited("%sF2FS-fs (%s) : inject %s in %s of %pS\n", \ 1879 KERN_INFO, sbi->sb->s_id, \ 1880 f2fs_fault_name[type], \ 1881 __func__, __builtin_return_address(0)) 1882 static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type) 1883 { 1884 struct f2fs_fault_info *ffi = &F2FS_OPTION(sbi).fault_info; 1885 1886 if (!ffi->inject_rate) 1887 return false; 1888 1889 if (!IS_FAULT_SET(ffi, type)) 1890 return false; 1891 1892 atomic_inc(&ffi->inject_ops); 1893 if (atomic_read(&ffi->inject_ops) >= ffi->inject_rate) { 1894 atomic_set(&ffi->inject_ops, 0); 1895 return true; 1896 } 1897 return false; 1898 } 1899 #else 1900 #define f2fs_show_injection_info(sbi, type) do { } while (0) 1901 static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type) 1902 { 1903 return false; 1904 } 1905 #endif 1906 1907 /* 1908 * Test if the mounted volume is a multi-device volume. 1909 * - For a single regular disk volume, sbi->s_ndevs is 0. 1910 * - For a single zoned disk volume, sbi->s_ndevs is 1. 1911 * - For a multi-device volume, sbi->s_ndevs is always 2 or more. 1912 */ 1913 static inline bool f2fs_is_multi_device(struct f2fs_sb_info *sbi) 1914 { 1915 return sbi->s_ndevs > 1; 1916 } 1917 1918 static inline void f2fs_update_time(struct f2fs_sb_info *sbi, int type) 1919 { 1920 unsigned long now = jiffies; 1921 1922 sbi->last_time[type] = now; 1923 1924 /* DISCARD_TIME and GC_TIME are based on REQ_TIME */ 1925 if (type == REQ_TIME) { 1926 sbi->last_time[DISCARD_TIME] = now; 1927 sbi->last_time[GC_TIME] = now; 1928 } 1929 } 1930 1931 static inline bool f2fs_time_over(struct f2fs_sb_info *sbi, int type) 1932 { 1933 unsigned long interval = sbi->interval_time[type] * HZ; 1934 1935 return time_after(jiffies, sbi->last_time[type] + interval); 1936 } 1937 1938 static inline unsigned int f2fs_time_to_wait(struct f2fs_sb_info *sbi, 1939 int type) 1940 { 1941 unsigned long interval = sbi->interval_time[type] * HZ; 1942 unsigned int wait_ms = 0; 1943 long delta; 1944 1945 delta = (sbi->last_time[type] + interval) - jiffies; 1946 if (delta > 0) 1947 wait_ms = jiffies_to_msecs(delta); 1948 1949 return wait_ms; 1950 } 1951 1952 /* 1953 * Inline functions 1954 */ 1955 static inline u32 __f2fs_crc32(struct f2fs_sb_info *sbi, u32 crc, 1956 const void *address, unsigned int length) 1957 { 1958 struct { 1959 struct shash_desc shash; 1960 char ctx[4]; 1961 } desc; 1962 int err; 1963 1964 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver) != sizeof(desc.ctx)); 1965 1966 desc.shash.tfm = sbi->s_chksum_driver; 1967 *(u32 *)desc.ctx = crc; 1968 1969 err = crypto_shash_update(&desc.shash, address, length); 1970 BUG_ON(err); 1971 1972 return *(u32 *)desc.ctx; 1973 } 1974 1975 static inline u32 f2fs_crc32(struct f2fs_sb_info *sbi, const void *address, 1976 unsigned int length) 1977 { 1978 return __f2fs_crc32(sbi, F2FS_SUPER_MAGIC, address, length); 1979 } 1980 1981 static inline bool f2fs_crc_valid(struct f2fs_sb_info *sbi, __u32 blk_crc, 1982 void *buf, size_t buf_size) 1983 { 1984 return f2fs_crc32(sbi, buf, buf_size) == blk_crc; 1985 } 1986 1987 static inline u32 f2fs_chksum(struct f2fs_sb_info *sbi, u32 crc, 1988 const void *address, unsigned int length) 1989 { 1990 return __f2fs_crc32(sbi, crc, address, length); 1991 } 1992 1993 static inline struct f2fs_inode_info *F2FS_I(struct inode *inode) 1994 { 1995 return container_of(inode, struct f2fs_inode_info, vfs_inode); 1996 } 1997 1998 static inline struct f2fs_sb_info *F2FS_SB(struct super_block *sb) 1999 { 2000 return sb->s_fs_info; 2001 } 2002 2003 static inline struct f2fs_sb_info *F2FS_I_SB(struct inode *inode) 2004 { 2005 return F2FS_SB(inode->i_sb); 2006 } 2007 2008 static inline struct f2fs_sb_info *F2FS_M_SB(struct address_space *mapping) 2009 { 2010 return F2FS_I_SB(mapping->host); 2011 } 2012 2013 static inline struct f2fs_sb_info *F2FS_P_SB(struct page *page) 2014 { 2015 return F2FS_M_SB(page_file_mapping(page)); 2016 } 2017 2018 static inline struct f2fs_super_block *F2FS_RAW_SUPER(struct f2fs_sb_info *sbi) 2019 { 2020 return (struct f2fs_super_block *)(sbi->raw_super); 2021 } 2022 2023 static inline struct f2fs_checkpoint *F2FS_CKPT(struct f2fs_sb_info *sbi) 2024 { 2025 return (struct f2fs_checkpoint *)(sbi->ckpt); 2026 } 2027 2028 static inline struct f2fs_node *F2FS_NODE(struct page *page) 2029 { 2030 return (struct f2fs_node *)page_address(page); 2031 } 2032 2033 static inline struct f2fs_inode *F2FS_INODE(struct page *page) 2034 { 2035 return &((struct f2fs_node *)page_address(page))->i; 2036 } 2037 2038 static inline struct f2fs_nm_info *NM_I(struct f2fs_sb_info *sbi) 2039 { 2040 return (struct f2fs_nm_info *)(sbi->nm_info); 2041 } 2042 2043 static inline struct f2fs_sm_info *SM_I(struct f2fs_sb_info *sbi) 2044 { 2045 return (struct f2fs_sm_info *)(sbi->sm_info); 2046 } 2047 2048 static inline struct sit_info *SIT_I(struct f2fs_sb_info *sbi) 2049 { 2050 return (struct sit_info *)(SM_I(sbi)->sit_info); 2051 } 2052 2053 static inline struct free_segmap_info *FREE_I(struct f2fs_sb_info *sbi) 2054 { 2055 return (struct free_segmap_info *)(SM_I(sbi)->free_info); 2056 } 2057 2058 static inline struct dirty_seglist_info *DIRTY_I(struct f2fs_sb_info *sbi) 2059 { 2060 return (struct dirty_seglist_info *)(SM_I(sbi)->dirty_info); 2061 } 2062 2063 static inline struct address_space *META_MAPPING(struct f2fs_sb_info *sbi) 2064 { 2065 return sbi->meta_inode->i_mapping; 2066 } 2067 2068 static inline struct address_space *NODE_MAPPING(struct f2fs_sb_info *sbi) 2069 { 2070 return sbi->node_inode->i_mapping; 2071 } 2072 2073 static inline bool is_sbi_flag_set(struct f2fs_sb_info *sbi, unsigned int type) 2074 { 2075 return test_bit(type, &sbi->s_flag); 2076 } 2077 2078 static inline void set_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type) 2079 { 2080 set_bit(type, &sbi->s_flag); 2081 } 2082 2083 static inline void clear_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type) 2084 { 2085 clear_bit(type, &sbi->s_flag); 2086 } 2087 2088 static inline unsigned long long cur_cp_version(struct f2fs_checkpoint *cp) 2089 { 2090 return le64_to_cpu(cp->checkpoint_ver); 2091 } 2092 2093 static inline unsigned long f2fs_qf_ino(struct super_block *sb, int type) 2094 { 2095 if (type < F2FS_MAX_QUOTAS) 2096 return le32_to_cpu(F2FS_SB(sb)->raw_super->qf_ino[type]); 2097 return 0; 2098 } 2099 2100 static inline __u64 cur_cp_crc(struct f2fs_checkpoint *cp) 2101 { 2102 size_t crc_offset = le32_to_cpu(cp->checksum_offset); 2103 return le32_to_cpu(*((__le32 *)((unsigned char *)cp + crc_offset))); 2104 } 2105 2106 static inline bool __is_set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 2107 { 2108 unsigned int ckpt_flags = le32_to_cpu(cp->ckpt_flags); 2109 2110 return ckpt_flags & f; 2111 } 2112 2113 static inline bool is_set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 2114 { 2115 return __is_set_ckpt_flags(F2FS_CKPT(sbi), f); 2116 } 2117 2118 static inline void __set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 2119 { 2120 unsigned int ckpt_flags; 2121 2122 ckpt_flags = le32_to_cpu(cp->ckpt_flags); 2123 ckpt_flags |= f; 2124 cp->ckpt_flags = cpu_to_le32(ckpt_flags); 2125 } 2126 2127 static inline void set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 2128 { 2129 unsigned long flags; 2130 2131 spin_lock_irqsave(&sbi->cp_lock, flags); 2132 __set_ckpt_flags(F2FS_CKPT(sbi), f); 2133 spin_unlock_irqrestore(&sbi->cp_lock, flags); 2134 } 2135 2136 static inline void __clear_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 2137 { 2138 unsigned int ckpt_flags; 2139 2140 ckpt_flags = le32_to_cpu(cp->ckpt_flags); 2141 ckpt_flags &= (~f); 2142 cp->ckpt_flags = cpu_to_le32(ckpt_flags); 2143 } 2144 2145 static inline void clear_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 2146 { 2147 unsigned long flags; 2148 2149 spin_lock_irqsave(&sbi->cp_lock, flags); 2150 __clear_ckpt_flags(F2FS_CKPT(sbi), f); 2151 spin_unlock_irqrestore(&sbi->cp_lock, flags); 2152 } 2153 2154 #define init_f2fs_rwsem(sem) \ 2155 do { \ 2156 static struct lock_class_key __key; \ 2157 \ 2158 __init_f2fs_rwsem((sem), #sem, &__key); \ 2159 } while (0) 2160 2161 static inline void __init_f2fs_rwsem(struct f2fs_rwsem *sem, 2162 const char *sem_name, struct lock_class_key *key) 2163 { 2164 __init_rwsem(&sem->internal_rwsem, sem_name, key); 2165 #ifdef CONFIG_F2FS_UNFAIR_RWSEM 2166 init_waitqueue_head(&sem->read_waiters); 2167 #endif 2168 } 2169 2170 static inline int f2fs_rwsem_is_locked(struct f2fs_rwsem *sem) 2171 { 2172 return rwsem_is_locked(&sem->internal_rwsem); 2173 } 2174 2175 static inline int f2fs_rwsem_is_contended(struct f2fs_rwsem *sem) 2176 { 2177 return rwsem_is_contended(&sem->internal_rwsem); 2178 } 2179 2180 static inline void f2fs_down_read(struct f2fs_rwsem *sem) 2181 { 2182 #ifdef CONFIG_F2FS_UNFAIR_RWSEM 2183 wait_event(sem->read_waiters, down_read_trylock(&sem->internal_rwsem)); 2184 #else 2185 down_read(&sem->internal_rwsem); 2186 #endif 2187 } 2188 2189 static inline int f2fs_down_read_trylock(struct f2fs_rwsem *sem) 2190 { 2191 return down_read_trylock(&sem->internal_rwsem); 2192 } 2193 2194 #ifdef CONFIG_DEBUG_LOCK_ALLOC 2195 static inline void f2fs_down_read_nested(struct f2fs_rwsem *sem, int subclass) 2196 { 2197 down_read_nested(&sem->internal_rwsem, subclass); 2198 } 2199 #else 2200 #define f2fs_down_read_nested(sem, subclass) f2fs_down_read(sem) 2201 #endif 2202 2203 static inline void f2fs_up_read(struct f2fs_rwsem *sem) 2204 { 2205 up_read(&sem->internal_rwsem); 2206 } 2207 2208 static inline void f2fs_down_write(struct f2fs_rwsem *sem) 2209 { 2210 down_write(&sem->internal_rwsem); 2211 } 2212 2213 static inline int f2fs_down_write_trylock(struct f2fs_rwsem *sem) 2214 { 2215 return down_write_trylock(&sem->internal_rwsem); 2216 } 2217 2218 static inline void f2fs_up_write(struct f2fs_rwsem *sem) 2219 { 2220 up_write(&sem->internal_rwsem); 2221 #ifdef CONFIG_F2FS_UNFAIR_RWSEM 2222 wake_up_all(&sem->read_waiters); 2223 #endif 2224 } 2225 2226 static inline void f2fs_lock_op(struct f2fs_sb_info *sbi) 2227 { 2228 f2fs_down_read(&sbi->cp_rwsem); 2229 } 2230 2231 static inline int f2fs_trylock_op(struct f2fs_sb_info *sbi) 2232 { 2233 if (time_to_inject(sbi, FAULT_LOCK_OP)) { 2234 f2fs_show_injection_info(sbi, FAULT_LOCK_OP); 2235 return 0; 2236 } 2237 return f2fs_down_read_trylock(&sbi->cp_rwsem); 2238 } 2239 2240 static inline void f2fs_unlock_op(struct f2fs_sb_info *sbi) 2241 { 2242 f2fs_up_read(&sbi->cp_rwsem); 2243 } 2244 2245 static inline void f2fs_lock_all(struct f2fs_sb_info *sbi) 2246 { 2247 f2fs_down_write(&sbi->cp_rwsem); 2248 } 2249 2250 static inline void f2fs_unlock_all(struct f2fs_sb_info *sbi) 2251 { 2252 f2fs_up_write(&sbi->cp_rwsem); 2253 } 2254 2255 static inline int __get_cp_reason(struct f2fs_sb_info *sbi) 2256 { 2257 int reason = CP_SYNC; 2258 2259 if (test_opt(sbi, FASTBOOT)) 2260 reason = CP_FASTBOOT; 2261 if (is_sbi_flag_set(sbi, SBI_IS_CLOSE)) 2262 reason = CP_UMOUNT; 2263 return reason; 2264 } 2265 2266 static inline bool __remain_node_summaries(int reason) 2267 { 2268 return (reason & (CP_UMOUNT | CP_FASTBOOT)); 2269 } 2270 2271 static inline bool __exist_node_summaries(struct f2fs_sb_info *sbi) 2272 { 2273 return (is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG) || 2274 is_set_ckpt_flags(sbi, CP_FASTBOOT_FLAG)); 2275 } 2276 2277 /* 2278 * Check whether the inode has blocks or not 2279 */ 2280 static inline int F2FS_HAS_BLOCKS(struct inode *inode) 2281 { 2282 block_t xattr_block = F2FS_I(inode)->i_xattr_nid ? 1 : 0; 2283 2284 return (inode->i_blocks >> F2FS_LOG_SECTORS_PER_BLOCK) > xattr_block; 2285 } 2286 2287 static inline bool f2fs_has_xattr_block(unsigned int ofs) 2288 { 2289 return ofs == XATTR_NODE_OFFSET; 2290 } 2291 2292 static inline bool __allow_reserved_blocks(struct f2fs_sb_info *sbi, 2293 struct inode *inode, bool cap) 2294 { 2295 if (!inode) 2296 return true; 2297 if (!test_opt(sbi, RESERVE_ROOT)) 2298 return false; 2299 if (IS_NOQUOTA(inode)) 2300 return true; 2301 if (uid_eq(F2FS_OPTION(sbi).s_resuid, current_fsuid())) 2302 return true; 2303 if (!gid_eq(F2FS_OPTION(sbi).s_resgid, GLOBAL_ROOT_GID) && 2304 in_group_p(F2FS_OPTION(sbi).s_resgid)) 2305 return true; 2306 if (cap && capable(CAP_SYS_RESOURCE)) 2307 return true; 2308 return false; 2309 } 2310 2311 static inline void f2fs_i_blocks_write(struct inode *, block_t, bool, bool); 2312 static inline int inc_valid_block_count(struct f2fs_sb_info *sbi, 2313 struct inode *inode, blkcnt_t *count) 2314 { 2315 blkcnt_t diff = 0, release = 0; 2316 block_t avail_user_block_count; 2317 int ret; 2318 2319 ret = dquot_reserve_block(inode, *count); 2320 if (ret) 2321 return ret; 2322 2323 if (time_to_inject(sbi, FAULT_BLOCK)) { 2324 f2fs_show_injection_info(sbi, FAULT_BLOCK); 2325 release = *count; 2326 goto release_quota; 2327 } 2328 2329 /* 2330 * let's increase this in prior to actual block count change in order 2331 * for f2fs_sync_file to avoid data races when deciding checkpoint. 2332 */ 2333 percpu_counter_add(&sbi->alloc_valid_block_count, (*count)); 2334 2335 spin_lock(&sbi->stat_lock); 2336 sbi->total_valid_block_count += (block_t)(*count); 2337 avail_user_block_count = sbi->user_block_count - 2338 sbi->current_reserved_blocks; 2339 2340 if (!__allow_reserved_blocks(sbi, inode, true)) 2341 avail_user_block_count -= F2FS_OPTION(sbi).root_reserved_blocks; 2342 2343 if (F2FS_IO_ALIGNED(sbi)) 2344 avail_user_block_count -= sbi->blocks_per_seg * 2345 SM_I(sbi)->additional_reserved_segments; 2346 2347 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) { 2348 if (avail_user_block_count > sbi->unusable_block_count) 2349 avail_user_block_count -= sbi->unusable_block_count; 2350 else 2351 avail_user_block_count = 0; 2352 } 2353 if (unlikely(sbi->total_valid_block_count > avail_user_block_count)) { 2354 diff = sbi->total_valid_block_count - avail_user_block_count; 2355 if (diff > *count) 2356 diff = *count; 2357 *count -= diff; 2358 release = diff; 2359 sbi->total_valid_block_count -= diff; 2360 if (!*count) { 2361 spin_unlock(&sbi->stat_lock); 2362 goto enospc; 2363 } 2364 } 2365 spin_unlock(&sbi->stat_lock); 2366 2367 if (unlikely(release)) { 2368 percpu_counter_sub(&sbi->alloc_valid_block_count, release); 2369 dquot_release_reservation_block(inode, release); 2370 } 2371 f2fs_i_blocks_write(inode, *count, true, true); 2372 return 0; 2373 2374 enospc: 2375 percpu_counter_sub(&sbi->alloc_valid_block_count, release); 2376 release_quota: 2377 dquot_release_reservation_block(inode, release); 2378 return -ENOSPC; 2379 } 2380 2381 __printf(2, 3) 2382 void f2fs_printk(struct f2fs_sb_info *sbi, const char *fmt, ...); 2383 2384 #define f2fs_err(sbi, fmt, ...) \ 2385 f2fs_printk(sbi, KERN_ERR fmt, ##__VA_ARGS__) 2386 #define f2fs_warn(sbi, fmt, ...) \ 2387 f2fs_printk(sbi, KERN_WARNING fmt, ##__VA_ARGS__) 2388 #define f2fs_notice(sbi, fmt, ...) \ 2389 f2fs_printk(sbi, KERN_NOTICE fmt, ##__VA_ARGS__) 2390 #define f2fs_info(sbi, fmt, ...) \ 2391 f2fs_printk(sbi, KERN_INFO fmt, ##__VA_ARGS__) 2392 #define f2fs_debug(sbi, fmt, ...) \ 2393 f2fs_printk(sbi, KERN_DEBUG fmt, ##__VA_ARGS__) 2394 2395 static inline void dec_valid_block_count(struct f2fs_sb_info *sbi, 2396 struct inode *inode, 2397 block_t count) 2398 { 2399 blkcnt_t sectors = count << F2FS_LOG_SECTORS_PER_BLOCK; 2400 2401 spin_lock(&sbi->stat_lock); 2402 f2fs_bug_on(sbi, sbi->total_valid_block_count < (block_t) count); 2403 sbi->total_valid_block_count -= (block_t)count; 2404 if (sbi->reserved_blocks && 2405 sbi->current_reserved_blocks < sbi->reserved_blocks) 2406 sbi->current_reserved_blocks = min(sbi->reserved_blocks, 2407 sbi->current_reserved_blocks + count); 2408 spin_unlock(&sbi->stat_lock); 2409 if (unlikely(inode->i_blocks < sectors)) { 2410 f2fs_warn(sbi, "Inconsistent i_blocks, ino:%lu, iblocks:%llu, sectors:%llu", 2411 inode->i_ino, 2412 (unsigned long long)inode->i_blocks, 2413 (unsigned long long)sectors); 2414 set_sbi_flag(sbi, SBI_NEED_FSCK); 2415 return; 2416 } 2417 f2fs_i_blocks_write(inode, count, false, true); 2418 } 2419 2420 static inline void inc_page_count(struct f2fs_sb_info *sbi, int count_type) 2421 { 2422 atomic_inc(&sbi->nr_pages[count_type]); 2423 2424 if (count_type == F2FS_DIRTY_DENTS || 2425 count_type == F2FS_DIRTY_NODES || 2426 count_type == F2FS_DIRTY_META || 2427 count_type == F2FS_DIRTY_QDATA || 2428 count_type == F2FS_DIRTY_IMETA) 2429 set_sbi_flag(sbi, SBI_IS_DIRTY); 2430 } 2431 2432 static inline void inode_inc_dirty_pages(struct inode *inode) 2433 { 2434 atomic_inc(&F2FS_I(inode)->dirty_pages); 2435 inc_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ? 2436 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA); 2437 if (IS_NOQUOTA(inode)) 2438 inc_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA); 2439 } 2440 2441 static inline void dec_page_count(struct f2fs_sb_info *sbi, int count_type) 2442 { 2443 atomic_dec(&sbi->nr_pages[count_type]); 2444 } 2445 2446 static inline void inode_dec_dirty_pages(struct inode *inode) 2447 { 2448 if (!S_ISDIR(inode->i_mode) && !S_ISREG(inode->i_mode) && 2449 !S_ISLNK(inode->i_mode)) 2450 return; 2451 2452 atomic_dec(&F2FS_I(inode)->dirty_pages); 2453 dec_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ? 2454 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA); 2455 if (IS_NOQUOTA(inode)) 2456 dec_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA); 2457 } 2458 2459 static inline void inc_atomic_write_cnt(struct inode *inode) 2460 { 2461 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 2462 struct f2fs_inode_info *fi = F2FS_I(inode); 2463 u64 current_write; 2464 2465 fi->atomic_write_cnt++; 2466 atomic64_inc(&sbi->current_atomic_write); 2467 current_write = atomic64_read(&sbi->current_atomic_write); 2468 if (current_write > sbi->peak_atomic_write) 2469 sbi->peak_atomic_write = current_write; 2470 } 2471 2472 static inline void release_atomic_write_cnt(struct inode *inode) 2473 { 2474 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 2475 struct f2fs_inode_info *fi = F2FS_I(inode); 2476 2477 atomic64_sub(fi->atomic_write_cnt, &sbi->current_atomic_write); 2478 fi->atomic_write_cnt = 0; 2479 } 2480 2481 static inline s64 get_pages(struct f2fs_sb_info *sbi, int count_type) 2482 { 2483 return atomic_read(&sbi->nr_pages[count_type]); 2484 } 2485 2486 static inline int get_dirty_pages(struct inode *inode) 2487 { 2488 return atomic_read(&F2FS_I(inode)->dirty_pages); 2489 } 2490 2491 static inline int get_blocktype_secs(struct f2fs_sb_info *sbi, int block_type) 2492 { 2493 unsigned int pages_per_sec = sbi->segs_per_sec * sbi->blocks_per_seg; 2494 unsigned int segs = (get_pages(sbi, block_type) + pages_per_sec - 1) >> 2495 sbi->log_blocks_per_seg; 2496 2497 return segs / sbi->segs_per_sec; 2498 } 2499 2500 static inline block_t valid_user_blocks(struct f2fs_sb_info *sbi) 2501 { 2502 return sbi->total_valid_block_count; 2503 } 2504 2505 static inline block_t discard_blocks(struct f2fs_sb_info *sbi) 2506 { 2507 return sbi->discard_blks; 2508 } 2509 2510 static inline unsigned long __bitmap_size(struct f2fs_sb_info *sbi, int flag) 2511 { 2512 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi); 2513 2514 /* return NAT or SIT bitmap */ 2515 if (flag == NAT_BITMAP) 2516 return le32_to_cpu(ckpt->nat_ver_bitmap_bytesize); 2517 else if (flag == SIT_BITMAP) 2518 return le32_to_cpu(ckpt->sit_ver_bitmap_bytesize); 2519 2520 return 0; 2521 } 2522 2523 static inline block_t __cp_payload(struct f2fs_sb_info *sbi) 2524 { 2525 return le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_payload); 2526 } 2527 2528 static inline void *__bitmap_ptr(struct f2fs_sb_info *sbi, int flag) 2529 { 2530 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi); 2531 void *tmp_ptr = &ckpt->sit_nat_version_bitmap; 2532 int offset; 2533 2534 if (is_set_ckpt_flags(sbi, CP_LARGE_NAT_BITMAP_FLAG)) { 2535 offset = (flag == SIT_BITMAP) ? 2536 le32_to_cpu(ckpt->nat_ver_bitmap_bytesize) : 0; 2537 /* 2538 * if large_nat_bitmap feature is enabled, leave checksum 2539 * protection for all nat/sit bitmaps. 2540 */ 2541 return tmp_ptr + offset + sizeof(__le32); 2542 } 2543 2544 if (__cp_payload(sbi) > 0) { 2545 if (flag == NAT_BITMAP) 2546 return tmp_ptr; 2547 else 2548 return (unsigned char *)ckpt + F2FS_BLKSIZE; 2549 } else { 2550 offset = (flag == NAT_BITMAP) ? 2551 le32_to_cpu(ckpt->sit_ver_bitmap_bytesize) : 0; 2552 return tmp_ptr + offset; 2553 } 2554 } 2555 2556 static inline block_t __start_cp_addr(struct f2fs_sb_info *sbi) 2557 { 2558 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr); 2559 2560 if (sbi->cur_cp_pack == 2) 2561 start_addr += sbi->blocks_per_seg; 2562 return start_addr; 2563 } 2564 2565 static inline block_t __start_cp_next_addr(struct f2fs_sb_info *sbi) 2566 { 2567 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr); 2568 2569 if (sbi->cur_cp_pack == 1) 2570 start_addr += sbi->blocks_per_seg; 2571 return start_addr; 2572 } 2573 2574 static inline void __set_cp_next_pack(struct f2fs_sb_info *sbi) 2575 { 2576 sbi->cur_cp_pack = (sbi->cur_cp_pack == 1) ? 2 : 1; 2577 } 2578 2579 static inline block_t __start_sum_addr(struct f2fs_sb_info *sbi) 2580 { 2581 return le32_to_cpu(F2FS_CKPT(sbi)->cp_pack_start_sum); 2582 } 2583 2584 static inline int inc_valid_node_count(struct f2fs_sb_info *sbi, 2585 struct inode *inode, bool is_inode) 2586 { 2587 block_t valid_block_count; 2588 unsigned int valid_node_count, user_block_count; 2589 int err; 2590 2591 if (is_inode) { 2592 if (inode) { 2593 err = dquot_alloc_inode(inode); 2594 if (err) 2595 return err; 2596 } 2597 } else { 2598 err = dquot_reserve_block(inode, 1); 2599 if (err) 2600 return err; 2601 } 2602 2603 if (time_to_inject(sbi, FAULT_BLOCK)) { 2604 f2fs_show_injection_info(sbi, FAULT_BLOCK); 2605 goto enospc; 2606 } 2607 2608 spin_lock(&sbi->stat_lock); 2609 2610 valid_block_count = sbi->total_valid_block_count + 2611 sbi->current_reserved_blocks + 1; 2612 2613 if (!__allow_reserved_blocks(sbi, inode, false)) 2614 valid_block_count += F2FS_OPTION(sbi).root_reserved_blocks; 2615 2616 if (F2FS_IO_ALIGNED(sbi)) 2617 valid_block_count += sbi->blocks_per_seg * 2618 SM_I(sbi)->additional_reserved_segments; 2619 2620 user_block_count = sbi->user_block_count; 2621 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) 2622 user_block_count -= sbi->unusable_block_count; 2623 2624 if (unlikely(valid_block_count > user_block_count)) { 2625 spin_unlock(&sbi->stat_lock); 2626 goto enospc; 2627 } 2628 2629 valid_node_count = sbi->total_valid_node_count + 1; 2630 if (unlikely(valid_node_count > sbi->total_node_count)) { 2631 spin_unlock(&sbi->stat_lock); 2632 goto enospc; 2633 } 2634 2635 sbi->total_valid_node_count++; 2636 sbi->total_valid_block_count++; 2637 spin_unlock(&sbi->stat_lock); 2638 2639 if (inode) { 2640 if (is_inode) 2641 f2fs_mark_inode_dirty_sync(inode, true); 2642 else 2643 f2fs_i_blocks_write(inode, 1, true, true); 2644 } 2645 2646 percpu_counter_inc(&sbi->alloc_valid_block_count); 2647 return 0; 2648 2649 enospc: 2650 if (is_inode) { 2651 if (inode) 2652 dquot_free_inode(inode); 2653 } else { 2654 dquot_release_reservation_block(inode, 1); 2655 } 2656 return -ENOSPC; 2657 } 2658 2659 static inline void dec_valid_node_count(struct f2fs_sb_info *sbi, 2660 struct inode *inode, bool is_inode) 2661 { 2662 spin_lock(&sbi->stat_lock); 2663 2664 if (unlikely(!sbi->total_valid_block_count || 2665 !sbi->total_valid_node_count)) { 2666 f2fs_warn(sbi, "dec_valid_node_count: inconsistent block counts, total_valid_block:%u, total_valid_node:%u", 2667 sbi->total_valid_block_count, 2668 sbi->total_valid_node_count); 2669 set_sbi_flag(sbi, SBI_NEED_FSCK); 2670 } else { 2671 sbi->total_valid_block_count--; 2672 sbi->total_valid_node_count--; 2673 } 2674 2675 if (sbi->reserved_blocks && 2676 sbi->current_reserved_blocks < sbi->reserved_blocks) 2677 sbi->current_reserved_blocks++; 2678 2679 spin_unlock(&sbi->stat_lock); 2680 2681 if (is_inode) { 2682 dquot_free_inode(inode); 2683 } else { 2684 if (unlikely(inode->i_blocks == 0)) { 2685 f2fs_warn(sbi, "dec_valid_node_count: inconsistent i_blocks, ino:%lu, iblocks:%llu", 2686 inode->i_ino, 2687 (unsigned long long)inode->i_blocks); 2688 set_sbi_flag(sbi, SBI_NEED_FSCK); 2689 return; 2690 } 2691 f2fs_i_blocks_write(inode, 1, false, true); 2692 } 2693 } 2694 2695 static inline unsigned int valid_node_count(struct f2fs_sb_info *sbi) 2696 { 2697 return sbi->total_valid_node_count; 2698 } 2699 2700 static inline void inc_valid_inode_count(struct f2fs_sb_info *sbi) 2701 { 2702 percpu_counter_inc(&sbi->total_valid_inode_count); 2703 } 2704 2705 static inline void dec_valid_inode_count(struct f2fs_sb_info *sbi) 2706 { 2707 percpu_counter_dec(&sbi->total_valid_inode_count); 2708 } 2709 2710 static inline s64 valid_inode_count(struct f2fs_sb_info *sbi) 2711 { 2712 return percpu_counter_sum_positive(&sbi->total_valid_inode_count); 2713 } 2714 2715 static inline struct page *f2fs_grab_cache_page(struct address_space *mapping, 2716 pgoff_t index, bool for_write) 2717 { 2718 struct page *page; 2719 unsigned int flags; 2720 2721 if (IS_ENABLED(CONFIG_F2FS_FAULT_INJECTION)) { 2722 if (!for_write) 2723 page = find_get_page_flags(mapping, index, 2724 FGP_LOCK | FGP_ACCESSED); 2725 else 2726 page = find_lock_page(mapping, index); 2727 if (page) 2728 return page; 2729 2730 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_ALLOC)) { 2731 f2fs_show_injection_info(F2FS_M_SB(mapping), 2732 FAULT_PAGE_ALLOC); 2733 return NULL; 2734 } 2735 } 2736 2737 if (!for_write) 2738 return grab_cache_page(mapping, index); 2739 2740 flags = memalloc_nofs_save(); 2741 page = grab_cache_page_write_begin(mapping, index); 2742 memalloc_nofs_restore(flags); 2743 2744 return page; 2745 } 2746 2747 static inline struct page *f2fs_pagecache_get_page( 2748 struct address_space *mapping, pgoff_t index, 2749 int fgp_flags, gfp_t gfp_mask) 2750 { 2751 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_GET)) { 2752 f2fs_show_injection_info(F2FS_M_SB(mapping), FAULT_PAGE_GET); 2753 return NULL; 2754 } 2755 2756 return pagecache_get_page(mapping, index, fgp_flags, gfp_mask); 2757 } 2758 2759 static inline void f2fs_put_page(struct page *page, int unlock) 2760 { 2761 if (!page) 2762 return; 2763 2764 if (unlock) { 2765 f2fs_bug_on(F2FS_P_SB(page), !PageLocked(page)); 2766 unlock_page(page); 2767 } 2768 put_page(page); 2769 } 2770 2771 static inline void f2fs_put_dnode(struct dnode_of_data *dn) 2772 { 2773 if (dn->node_page) 2774 f2fs_put_page(dn->node_page, 1); 2775 if (dn->inode_page && dn->node_page != dn->inode_page) 2776 f2fs_put_page(dn->inode_page, 0); 2777 dn->node_page = NULL; 2778 dn->inode_page = NULL; 2779 } 2780 2781 static inline struct kmem_cache *f2fs_kmem_cache_create(const char *name, 2782 size_t size) 2783 { 2784 return kmem_cache_create(name, size, 0, SLAB_RECLAIM_ACCOUNT, NULL); 2785 } 2786 2787 static inline void *f2fs_kmem_cache_alloc_nofail(struct kmem_cache *cachep, 2788 gfp_t flags) 2789 { 2790 void *entry; 2791 2792 entry = kmem_cache_alloc(cachep, flags); 2793 if (!entry) 2794 entry = kmem_cache_alloc(cachep, flags | __GFP_NOFAIL); 2795 return entry; 2796 } 2797 2798 static inline void *f2fs_kmem_cache_alloc(struct kmem_cache *cachep, 2799 gfp_t flags, bool nofail, struct f2fs_sb_info *sbi) 2800 { 2801 if (nofail) 2802 return f2fs_kmem_cache_alloc_nofail(cachep, flags); 2803 2804 if (time_to_inject(sbi, FAULT_SLAB_ALLOC)) { 2805 f2fs_show_injection_info(sbi, FAULT_SLAB_ALLOC); 2806 return NULL; 2807 } 2808 2809 return kmem_cache_alloc(cachep, flags); 2810 } 2811 2812 static inline bool is_inflight_io(struct f2fs_sb_info *sbi, int type) 2813 { 2814 if (get_pages(sbi, F2FS_RD_DATA) || get_pages(sbi, F2FS_RD_NODE) || 2815 get_pages(sbi, F2FS_RD_META) || get_pages(sbi, F2FS_WB_DATA) || 2816 get_pages(sbi, F2FS_WB_CP_DATA) || 2817 get_pages(sbi, F2FS_DIO_READ) || 2818 get_pages(sbi, F2FS_DIO_WRITE)) 2819 return true; 2820 2821 if (type != DISCARD_TIME && SM_I(sbi) && SM_I(sbi)->dcc_info && 2822 atomic_read(&SM_I(sbi)->dcc_info->queued_discard)) 2823 return true; 2824 2825 if (SM_I(sbi) && SM_I(sbi)->fcc_info && 2826 atomic_read(&SM_I(sbi)->fcc_info->queued_flush)) 2827 return true; 2828 return false; 2829 } 2830 2831 static inline bool is_idle(struct f2fs_sb_info *sbi, int type) 2832 { 2833 if (sbi->gc_mode == GC_URGENT_HIGH) 2834 return true; 2835 2836 if (is_inflight_io(sbi, type)) 2837 return false; 2838 2839 if (sbi->gc_mode == GC_URGENT_MID) 2840 return true; 2841 2842 if (sbi->gc_mode == GC_URGENT_LOW && 2843 (type == DISCARD_TIME || type == GC_TIME)) 2844 return true; 2845 2846 return f2fs_time_over(sbi, type); 2847 } 2848 2849 static inline void f2fs_radix_tree_insert(struct radix_tree_root *root, 2850 unsigned long index, void *item) 2851 { 2852 while (radix_tree_insert(root, index, item)) 2853 cond_resched(); 2854 } 2855 2856 #define RAW_IS_INODE(p) ((p)->footer.nid == (p)->footer.ino) 2857 2858 static inline bool IS_INODE(struct page *page) 2859 { 2860 struct f2fs_node *p = F2FS_NODE(page); 2861 2862 return RAW_IS_INODE(p); 2863 } 2864 2865 static inline int offset_in_addr(struct f2fs_inode *i) 2866 { 2867 return (i->i_inline & F2FS_EXTRA_ATTR) ? 2868 (le16_to_cpu(i->i_extra_isize) / sizeof(__le32)) : 0; 2869 } 2870 2871 static inline __le32 *blkaddr_in_node(struct f2fs_node *node) 2872 { 2873 return RAW_IS_INODE(node) ? node->i.i_addr : node->dn.addr; 2874 } 2875 2876 static inline int f2fs_has_extra_attr(struct inode *inode); 2877 static inline block_t data_blkaddr(struct inode *inode, 2878 struct page *node_page, unsigned int offset) 2879 { 2880 struct f2fs_node *raw_node; 2881 __le32 *addr_array; 2882 int base = 0; 2883 bool is_inode = IS_INODE(node_page); 2884 2885 raw_node = F2FS_NODE(node_page); 2886 2887 if (is_inode) { 2888 if (!inode) 2889 /* from GC path only */ 2890 base = offset_in_addr(&raw_node->i); 2891 else if (f2fs_has_extra_attr(inode)) 2892 base = get_extra_isize(inode); 2893 } 2894 2895 addr_array = blkaddr_in_node(raw_node); 2896 return le32_to_cpu(addr_array[base + offset]); 2897 } 2898 2899 static inline block_t f2fs_data_blkaddr(struct dnode_of_data *dn) 2900 { 2901 return data_blkaddr(dn->inode, dn->node_page, dn->ofs_in_node); 2902 } 2903 2904 static inline int f2fs_test_bit(unsigned int nr, char *addr) 2905 { 2906 int mask; 2907 2908 addr += (nr >> 3); 2909 mask = 1 << (7 - (nr & 0x07)); 2910 return mask & *addr; 2911 } 2912 2913 static inline void f2fs_set_bit(unsigned int nr, char *addr) 2914 { 2915 int mask; 2916 2917 addr += (nr >> 3); 2918 mask = 1 << (7 - (nr & 0x07)); 2919 *addr |= mask; 2920 } 2921 2922 static inline void f2fs_clear_bit(unsigned int nr, char *addr) 2923 { 2924 int mask; 2925 2926 addr += (nr >> 3); 2927 mask = 1 << (7 - (nr & 0x07)); 2928 *addr &= ~mask; 2929 } 2930 2931 static inline int f2fs_test_and_set_bit(unsigned int nr, char *addr) 2932 { 2933 int mask; 2934 int ret; 2935 2936 addr += (nr >> 3); 2937 mask = 1 << (7 - (nr & 0x07)); 2938 ret = mask & *addr; 2939 *addr |= mask; 2940 return ret; 2941 } 2942 2943 static inline int f2fs_test_and_clear_bit(unsigned int nr, char *addr) 2944 { 2945 int mask; 2946 int ret; 2947 2948 addr += (nr >> 3); 2949 mask = 1 << (7 - (nr & 0x07)); 2950 ret = mask & *addr; 2951 *addr &= ~mask; 2952 return ret; 2953 } 2954 2955 static inline void f2fs_change_bit(unsigned int nr, char *addr) 2956 { 2957 int mask; 2958 2959 addr += (nr >> 3); 2960 mask = 1 << (7 - (nr & 0x07)); 2961 *addr ^= mask; 2962 } 2963 2964 /* 2965 * On-disk inode flags (f2fs_inode::i_flags) 2966 */ 2967 #define F2FS_COMPR_FL 0x00000004 /* Compress file */ 2968 #define F2FS_SYNC_FL 0x00000008 /* Synchronous updates */ 2969 #define F2FS_IMMUTABLE_FL 0x00000010 /* Immutable file */ 2970 #define F2FS_APPEND_FL 0x00000020 /* writes to file may only append */ 2971 #define F2FS_NODUMP_FL 0x00000040 /* do not dump file */ 2972 #define F2FS_NOATIME_FL 0x00000080 /* do not update atime */ 2973 #define F2FS_NOCOMP_FL 0x00000400 /* Don't compress */ 2974 #define F2FS_INDEX_FL 0x00001000 /* hash-indexed directory */ 2975 #define F2FS_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */ 2976 #define F2FS_PROJINHERIT_FL 0x20000000 /* Create with parents projid */ 2977 #define F2FS_CASEFOLD_FL 0x40000000 /* Casefolded file */ 2978 2979 /* Flags that should be inherited by new inodes from their parent. */ 2980 #define F2FS_FL_INHERITED (F2FS_SYNC_FL | F2FS_NODUMP_FL | F2FS_NOATIME_FL | \ 2981 F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \ 2982 F2FS_CASEFOLD_FL | F2FS_COMPR_FL | F2FS_NOCOMP_FL) 2983 2984 /* Flags that are appropriate for regular files (all but dir-specific ones). */ 2985 #define F2FS_REG_FLMASK (~(F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \ 2986 F2FS_CASEFOLD_FL)) 2987 2988 /* Flags that are appropriate for non-directories/regular files. */ 2989 #define F2FS_OTHER_FLMASK (F2FS_NODUMP_FL | F2FS_NOATIME_FL) 2990 2991 static inline __u32 f2fs_mask_flags(umode_t mode, __u32 flags) 2992 { 2993 if (S_ISDIR(mode)) 2994 return flags; 2995 else if (S_ISREG(mode)) 2996 return flags & F2FS_REG_FLMASK; 2997 else 2998 return flags & F2FS_OTHER_FLMASK; 2999 } 3000 3001 static inline void __mark_inode_dirty_flag(struct inode *inode, 3002 int flag, bool set) 3003 { 3004 switch (flag) { 3005 case FI_INLINE_XATTR: 3006 case FI_INLINE_DATA: 3007 case FI_INLINE_DENTRY: 3008 case FI_NEW_INODE: 3009 if (set) 3010 return; 3011 fallthrough; 3012 case FI_DATA_EXIST: 3013 case FI_INLINE_DOTS: 3014 case FI_PIN_FILE: 3015 case FI_COMPRESS_RELEASED: 3016 f2fs_mark_inode_dirty_sync(inode, true); 3017 } 3018 } 3019 3020 static inline void set_inode_flag(struct inode *inode, int flag) 3021 { 3022 set_bit(flag, F2FS_I(inode)->flags); 3023 __mark_inode_dirty_flag(inode, flag, true); 3024 } 3025 3026 static inline int is_inode_flag_set(struct inode *inode, int flag) 3027 { 3028 return test_bit(flag, F2FS_I(inode)->flags); 3029 } 3030 3031 static inline void clear_inode_flag(struct inode *inode, int flag) 3032 { 3033 clear_bit(flag, F2FS_I(inode)->flags); 3034 __mark_inode_dirty_flag(inode, flag, false); 3035 } 3036 3037 static inline bool f2fs_verity_in_progress(struct inode *inode) 3038 { 3039 return IS_ENABLED(CONFIG_FS_VERITY) && 3040 is_inode_flag_set(inode, FI_VERITY_IN_PROGRESS); 3041 } 3042 3043 static inline void set_acl_inode(struct inode *inode, umode_t mode) 3044 { 3045 F2FS_I(inode)->i_acl_mode = mode; 3046 set_inode_flag(inode, FI_ACL_MODE); 3047 f2fs_mark_inode_dirty_sync(inode, false); 3048 } 3049 3050 static inline void f2fs_i_links_write(struct inode *inode, bool inc) 3051 { 3052 if (inc) 3053 inc_nlink(inode); 3054 else 3055 drop_nlink(inode); 3056 f2fs_mark_inode_dirty_sync(inode, true); 3057 } 3058 3059 static inline void f2fs_i_blocks_write(struct inode *inode, 3060 block_t diff, bool add, bool claim) 3061 { 3062 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE); 3063 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER); 3064 3065 /* add = 1, claim = 1 should be dquot_reserve_block in pair */ 3066 if (add) { 3067 if (claim) 3068 dquot_claim_block(inode, diff); 3069 else 3070 dquot_alloc_block_nofail(inode, diff); 3071 } else { 3072 dquot_free_block(inode, diff); 3073 } 3074 3075 f2fs_mark_inode_dirty_sync(inode, true); 3076 if (clean || recover) 3077 set_inode_flag(inode, FI_AUTO_RECOVER); 3078 } 3079 3080 static inline bool f2fs_is_atomic_file(struct inode *inode); 3081 3082 static inline void f2fs_i_size_write(struct inode *inode, loff_t i_size) 3083 { 3084 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE); 3085 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER); 3086 3087 if (i_size_read(inode) == i_size) 3088 return; 3089 3090 i_size_write(inode, i_size); 3091 3092 if (f2fs_is_atomic_file(inode)) 3093 return; 3094 3095 f2fs_mark_inode_dirty_sync(inode, true); 3096 if (clean || recover) 3097 set_inode_flag(inode, FI_AUTO_RECOVER); 3098 } 3099 3100 static inline void f2fs_i_depth_write(struct inode *inode, unsigned int depth) 3101 { 3102 F2FS_I(inode)->i_current_depth = depth; 3103 f2fs_mark_inode_dirty_sync(inode, true); 3104 } 3105 3106 static inline void f2fs_i_gc_failures_write(struct inode *inode, 3107 unsigned int count) 3108 { 3109 F2FS_I(inode)->i_gc_failures[GC_FAILURE_PIN] = count; 3110 f2fs_mark_inode_dirty_sync(inode, true); 3111 } 3112 3113 static inline void f2fs_i_xnid_write(struct inode *inode, nid_t xnid) 3114 { 3115 F2FS_I(inode)->i_xattr_nid = xnid; 3116 f2fs_mark_inode_dirty_sync(inode, true); 3117 } 3118 3119 static inline void f2fs_i_pino_write(struct inode *inode, nid_t pino) 3120 { 3121 F2FS_I(inode)->i_pino = pino; 3122 f2fs_mark_inode_dirty_sync(inode, true); 3123 } 3124 3125 static inline void get_inline_info(struct inode *inode, struct f2fs_inode *ri) 3126 { 3127 struct f2fs_inode_info *fi = F2FS_I(inode); 3128 3129 if (ri->i_inline & F2FS_INLINE_XATTR) 3130 set_bit(FI_INLINE_XATTR, fi->flags); 3131 if (ri->i_inline & F2FS_INLINE_DATA) 3132 set_bit(FI_INLINE_DATA, fi->flags); 3133 if (ri->i_inline & F2FS_INLINE_DENTRY) 3134 set_bit(FI_INLINE_DENTRY, fi->flags); 3135 if (ri->i_inline & F2FS_DATA_EXIST) 3136 set_bit(FI_DATA_EXIST, fi->flags); 3137 if (ri->i_inline & F2FS_INLINE_DOTS) 3138 set_bit(FI_INLINE_DOTS, fi->flags); 3139 if (ri->i_inline & F2FS_EXTRA_ATTR) 3140 set_bit(FI_EXTRA_ATTR, fi->flags); 3141 if (ri->i_inline & F2FS_PIN_FILE) 3142 set_bit(FI_PIN_FILE, fi->flags); 3143 if (ri->i_inline & F2FS_COMPRESS_RELEASED) 3144 set_bit(FI_COMPRESS_RELEASED, fi->flags); 3145 } 3146 3147 static inline void set_raw_inline(struct inode *inode, struct f2fs_inode *ri) 3148 { 3149 ri->i_inline = 0; 3150 3151 if (is_inode_flag_set(inode, FI_INLINE_XATTR)) 3152 ri->i_inline |= F2FS_INLINE_XATTR; 3153 if (is_inode_flag_set(inode, FI_INLINE_DATA)) 3154 ri->i_inline |= F2FS_INLINE_DATA; 3155 if (is_inode_flag_set(inode, FI_INLINE_DENTRY)) 3156 ri->i_inline |= F2FS_INLINE_DENTRY; 3157 if (is_inode_flag_set(inode, FI_DATA_EXIST)) 3158 ri->i_inline |= F2FS_DATA_EXIST; 3159 if (is_inode_flag_set(inode, FI_INLINE_DOTS)) 3160 ri->i_inline |= F2FS_INLINE_DOTS; 3161 if (is_inode_flag_set(inode, FI_EXTRA_ATTR)) 3162 ri->i_inline |= F2FS_EXTRA_ATTR; 3163 if (is_inode_flag_set(inode, FI_PIN_FILE)) 3164 ri->i_inline |= F2FS_PIN_FILE; 3165 if (is_inode_flag_set(inode, FI_COMPRESS_RELEASED)) 3166 ri->i_inline |= F2FS_COMPRESS_RELEASED; 3167 } 3168 3169 static inline int f2fs_has_extra_attr(struct inode *inode) 3170 { 3171 return is_inode_flag_set(inode, FI_EXTRA_ATTR); 3172 } 3173 3174 static inline int f2fs_has_inline_xattr(struct inode *inode) 3175 { 3176 return is_inode_flag_set(inode, FI_INLINE_XATTR); 3177 } 3178 3179 static inline int f2fs_compressed_file(struct inode *inode) 3180 { 3181 return S_ISREG(inode->i_mode) && 3182 is_inode_flag_set(inode, FI_COMPRESSED_FILE); 3183 } 3184 3185 static inline bool f2fs_need_compress_data(struct inode *inode) 3186 { 3187 int compress_mode = F2FS_OPTION(F2FS_I_SB(inode)).compress_mode; 3188 3189 if (!f2fs_compressed_file(inode)) 3190 return false; 3191 3192 if (compress_mode == COMPR_MODE_FS) 3193 return true; 3194 else if (compress_mode == COMPR_MODE_USER && 3195 is_inode_flag_set(inode, FI_ENABLE_COMPRESS)) 3196 return true; 3197 3198 return false; 3199 } 3200 3201 static inline unsigned int addrs_per_inode(struct inode *inode) 3202 { 3203 unsigned int addrs = CUR_ADDRS_PER_INODE(inode) - 3204 get_inline_xattr_addrs(inode); 3205 3206 if (!f2fs_compressed_file(inode)) 3207 return addrs; 3208 return ALIGN_DOWN(addrs, F2FS_I(inode)->i_cluster_size); 3209 } 3210 3211 static inline unsigned int addrs_per_block(struct inode *inode) 3212 { 3213 if (!f2fs_compressed_file(inode)) 3214 return DEF_ADDRS_PER_BLOCK; 3215 return ALIGN_DOWN(DEF_ADDRS_PER_BLOCK, F2FS_I(inode)->i_cluster_size); 3216 } 3217 3218 static inline void *inline_xattr_addr(struct inode *inode, struct page *page) 3219 { 3220 struct f2fs_inode *ri = F2FS_INODE(page); 3221 3222 return (void *)&(ri->i_addr[DEF_ADDRS_PER_INODE - 3223 get_inline_xattr_addrs(inode)]); 3224 } 3225 3226 static inline int inline_xattr_size(struct inode *inode) 3227 { 3228 if (f2fs_has_inline_xattr(inode)) 3229 return get_inline_xattr_addrs(inode) * sizeof(__le32); 3230 return 0; 3231 } 3232 3233 /* 3234 * Notice: check inline_data flag without inode page lock is unsafe. 3235 * It could change at any time by f2fs_convert_inline_page(). 3236 */ 3237 static inline int f2fs_has_inline_data(struct inode *inode) 3238 { 3239 return is_inode_flag_set(inode, FI_INLINE_DATA); 3240 } 3241 3242 static inline int f2fs_exist_data(struct inode *inode) 3243 { 3244 return is_inode_flag_set(inode, FI_DATA_EXIST); 3245 } 3246 3247 static inline int f2fs_has_inline_dots(struct inode *inode) 3248 { 3249 return is_inode_flag_set(inode, FI_INLINE_DOTS); 3250 } 3251 3252 static inline int f2fs_is_mmap_file(struct inode *inode) 3253 { 3254 return is_inode_flag_set(inode, FI_MMAP_FILE); 3255 } 3256 3257 static inline bool f2fs_is_pinned_file(struct inode *inode) 3258 { 3259 return is_inode_flag_set(inode, FI_PIN_FILE); 3260 } 3261 3262 static inline bool f2fs_is_atomic_file(struct inode *inode) 3263 { 3264 return is_inode_flag_set(inode, FI_ATOMIC_FILE); 3265 } 3266 3267 static inline bool f2fs_is_cow_file(struct inode *inode) 3268 { 3269 return is_inode_flag_set(inode, FI_COW_FILE); 3270 } 3271 3272 static inline bool f2fs_is_first_block_written(struct inode *inode) 3273 { 3274 return is_inode_flag_set(inode, FI_FIRST_BLOCK_WRITTEN); 3275 } 3276 3277 static inline bool f2fs_is_drop_cache(struct inode *inode) 3278 { 3279 return is_inode_flag_set(inode, FI_DROP_CACHE); 3280 } 3281 3282 static inline void *inline_data_addr(struct inode *inode, struct page *page) 3283 { 3284 struct f2fs_inode *ri = F2FS_INODE(page); 3285 int extra_size = get_extra_isize(inode); 3286 3287 return (void *)&(ri->i_addr[extra_size + DEF_INLINE_RESERVED_SIZE]); 3288 } 3289 3290 static inline int f2fs_has_inline_dentry(struct inode *inode) 3291 { 3292 return is_inode_flag_set(inode, FI_INLINE_DENTRY); 3293 } 3294 3295 static inline int is_file(struct inode *inode, int type) 3296 { 3297 return F2FS_I(inode)->i_advise & type; 3298 } 3299 3300 static inline void set_file(struct inode *inode, int type) 3301 { 3302 if (is_file(inode, type)) 3303 return; 3304 F2FS_I(inode)->i_advise |= type; 3305 f2fs_mark_inode_dirty_sync(inode, true); 3306 } 3307 3308 static inline void clear_file(struct inode *inode, int type) 3309 { 3310 if (!is_file(inode, type)) 3311 return; 3312 F2FS_I(inode)->i_advise &= ~type; 3313 f2fs_mark_inode_dirty_sync(inode, true); 3314 } 3315 3316 static inline bool f2fs_is_time_consistent(struct inode *inode) 3317 { 3318 if (!timespec64_equal(F2FS_I(inode)->i_disk_time, &inode->i_atime)) 3319 return false; 3320 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 1, &inode->i_ctime)) 3321 return false; 3322 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 2, &inode->i_mtime)) 3323 return false; 3324 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 3, 3325 &F2FS_I(inode)->i_crtime)) 3326 return false; 3327 return true; 3328 } 3329 3330 static inline bool f2fs_skip_inode_update(struct inode *inode, int dsync) 3331 { 3332 bool ret; 3333 3334 if (dsync) { 3335 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 3336 3337 spin_lock(&sbi->inode_lock[DIRTY_META]); 3338 ret = list_empty(&F2FS_I(inode)->gdirty_list); 3339 spin_unlock(&sbi->inode_lock[DIRTY_META]); 3340 return ret; 3341 } 3342 if (!is_inode_flag_set(inode, FI_AUTO_RECOVER) || 3343 file_keep_isize(inode) || 3344 i_size_read(inode) & ~PAGE_MASK) 3345 return false; 3346 3347 if (!f2fs_is_time_consistent(inode)) 3348 return false; 3349 3350 spin_lock(&F2FS_I(inode)->i_size_lock); 3351 ret = F2FS_I(inode)->last_disk_size == i_size_read(inode); 3352 spin_unlock(&F2FS_I(inode)->i_size_lock); 3353 3354 return ret; 3355 } 3356 3357 static inline bool f2fs_readonly(struct super_block *sb) 3358 { 3359 return sb_rdonly(sb); 3360 } 3361 3362 static inline bool f2fs_cp_error(struct f2fs_sb_info *sbi) 3363 { 3364 return is_set_ckpt_flags(sbi, CP_ERROR_FLAG); 3365 } 3366 3367 static inline bool is_dot_dotdot(const u8 *name, size_t len) 3368 { 3369 if (len == 1 && name[0] == '.') 3370 return true; 3371 3372 if (len == 2 && name[0] == '.' && name[1] == '.') 3373 return true; 3374 3375 return false; 3376 } 3377 3378 static inline void *f2fs_kmalloc(struct f2fs_sb_info *sbi, 3379 size_t size, gfp_t flags) 3380 { 3381 if (time_to_inject(sbi, FAULT_KMALLOC)) { 3382 f2fs_show_injection_info(sbi, FAULT_KMALLOC); 3383 return NULL; 3384 } 3385 3386 return kmalloc(size, flags); 3387 } 3388 3389 static inline void *f2fs_kzalloc(struct f2fs_sb_info *sbi, 3390 size_t size, gfp_t flags) 3391 { 3392 return f2fs_kmalloc(sbi, size, flags | __GFP_ZERO); 3393 } 3394 3395 static inline void *f2fs_kvmalloc(struct f2fs_sb_info *sbi, 3396 size_t size, gfp_t flags) 3397 { 3398 if (time_to_inject(sbi, FAULT_KVMALLOC)) { 3399 f2fs_show_injection_info(sbi, FAULT_KVMALLOC); 3400 return NULL; 3401 } 3402 3403 return kvmalloc(size, flags); 3404 } 3405 3406 static inline void *f2fs_kvzalloc(struct f2fs_sb_info *sbi, 3407 size_t size, gfp_t flags) 3408 { 3409 return f2fs_kvmalloc(sbi, size, flags | __GFP_ZERO); 3410 } 3411 3412 static inline int get_extra_isize(struct inode *inode) 3413 { 3414 return F2FS_I(inode)->i_extra_isize / sizeof(__le32); 3415 } 3416 3417 static inline int get_inline_xattr_addrs(struct inode *inode) 3418 { 3419 return F2FS_I(inode)->i_inline_xattr_size; 3420 } 3421 3422 #define f2fs_get_inode_mode(i) \ 3423 ((is_inode_flag_set(i, FI_ACL_MODE)) ? \ 3424 (F2FS_I(i)->i_acl_mode) : ((i)->i_mode)) 3425 3426 #define F2FS_TOTAL_EXTRA_ATTR_SIZE \ 3427 (offsetof(struct f2fs_inode, i_extra_end) - \ 3428 offsetof(struct f2fs_inode, i_extra_isize)) \ 3429 3430 #define F2FS_OLD_ATTRIBUTE_SIZE (offsetof(struct f2fs_inode, i_addr)) 3431 #define F2FS_FITS_IN_INODE(f2fs_inode, extra_isize, field) \ 3432 ((offsetof(typeof(*(f2fs_inode)), field) + \ 3433 sizeof((f2fs_inode)->field)) \ 3434 <= (F2FS_OLD_ATTRIBUTE_SIZE + (extra_isize))) \ 3435 3436 #define __is_large_section(sbi) ((sbi)->segs_per_sec > 1) 3437 3438 #define __is_meta_io(fio) (PAGE_TYPE_OF_BIO((fio)->type) == META) 3439 3440 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi, 3441 block_t blkaddr, int type); 3442 static inline void verify_blkaddr(struct f2fs_sb_info *sbi, 3443 block_t blkaddr, int type) 3444 { 3445 if (!f2fs_is_valid_blkaddr(sbi, blkaddr, type)) { 3446 f2fs_err(sbi, "invalid blkaddr: %u, type: %d, run fsck to fix.", 3447 blkaddr, type); 3448 f2fs_bug_on(sbi, 1); 3449 } 3450 } 3451 3452 static inline bool __is_valid_data_blkaddr(block_t blkaddr) 3453 { 3454 if (blkaddr == NEW_ADDR || blkaddr == NULL_ADDR || 3455 blkaddr == COMPRESS_ADDR) 3456 return false; 3457 return true; 3458 } 3459 3460 /* 3461 * file.c 3462 */ 3463 int f2fs_sync_file(struct file *file, loff_t start, loff_t end, int datasync); 3464 void f2fs_truncate_data_blocks(struct dnode_of_data *dn); 3465 int f2fs_do_truncate_blocks(struct inode *inode, u64 from, bool lock); 3466 int f2fs_truncate_blocks(struct inode *inode, u64 from, bool lock); 3467 int f2fs_truncate(struct inode *inode); 3468 int f2fs_getattr(struct user_namespace *mnt_userns, const struct path *path, 3469 struct kstat *stat, u32 request_mask, unsigned int flags); 3470 int f2fs_setattr(struct user_namespace *mnt_userns, struct dentry *dentry, 3471 struct iattr *attr); 3472 int f2fs_truncate_hole(struct inode *inode, pgoff_t pg_start, pgoff_t pg_end); 3473 void f2fs_truncate_data_blocks_range(struct dnode_of_data *dn, int count); 3474 int f2fs_precache_extents(struct inode *inode); 3475 int f2fs_fileattr_get(struct dentry *dentry, struct fileattr *fa); 3476 int f2fs_fileattr_set(struct user_namespace *mnt_userns, 3477 struct dentry *dentry, struct fileattr *fa); 3478 long f2fs_ioctl(struct file *filp, unsigned int cmd, unsigned long arg); 3479 long f2fs_compat_ioctl(struct file *file, unsigned int cmd, unsigned long arg); 3480 int f2fs_transfer_project_quota(struct inode *inode, kprojid_t kprojid); 3481 int f2fs_pin_file_control(struct inode *inode, bool inc); 3482 3483 /* 3484 * inode.c 3485 */ 3486 void f2fs_set_inode_flags(struct inode *inode); 3487 bool f2fs_inode_chksum_verify(struct f2fs_sb_info *sbi, struct page *page); 3488 void f2fs_inode_chksum_set(struct f2fs_sb_info *sbi, struct page *page); 3489 struct inode *f2fs_iget(struct super_block *sb, unsigned long ino); 3490 struct inode *f2fs_iget_retry(struct super_block *sb, unsigned long ino); 3491 int f2fs_try_to_free_nats(struct f2fs_sb_info *sbi, int nr_shrink); 3492 void f2fs_update_inode(struct inode *inode, struct page *node_page); 3493 void f2fs_update_inode_page(struct inode *inode); 3494 int f2fs_write_inode(struct inode *inode, struct writeback_control *wbc); 3495 void f2fs_evict_inode(struct inode *inode); 3496 void f2fs_handle_failed_inode(struct inode *inode); 3497 3498 /* 3499 * namei.c 3500 */ 3501 int f2fs_update_extension_list(struct f2fs_sb_info *sbi, const char *name, 3502 bool hot, bool set); 3503 struct dentry *f2fs_get_parent(struct dentry *child); 3504 int f2fs_get_tmpfile(struct user_namespace *mnt_userns, struct inode *dir, 3505 struct inode **new_inode); 3506 3507 /* 3508 * dir.c 3509 */ 3510 unsigned char f2fs_get_de_type(struct f2fs_dir_entry *de); 3511 int f2fs_init_casefolded_name(const struct inode *dir, 3512 struct f2fs_filename *fname); 3513 int f2fs_setup_filename(struct inode *dir, const struct qstr *iname, 3514 int lookup, struct f2fs_filename *fname); 3515 int f2fs_prepare_lookup(struct inode *dir, struct dentry *dentry, 3516 struct f2fs_filename *fname); 3517 void f2fs_free_filename(struct f2fs_filename *fname); 3518 struct f2fs_dir_entry *f2fs_find_target_dentry(const struct f2fs_dentry_ptr *d, 3519 const struct f2fs_filename *fname, int *max_slots); 3520 int f2fs_fill_dentries(struct dir_context *ctx, struct f2fs_dentry_ptr *d, 3521 unsigned int start_pos, struct fscrypt_str *fstr); 3522 void f2fs_do_make_empty_dir(struct inode *inode, struct inode *parent, 3523 struct f2fs_dentry_ptr *d); 3524 struct page *f2fs_init_inode_metadata(struct inode *inode, struct inode *dir, 3525 const struct f2fs_filename *fname, struct page *dpage); 3526 void f2fs_update_parent_metadata(struct inode *dir, struct inode *inode, 3527 unsigned int current_depth); 3528 int f2fs_room_for_filename(const void *bitmap, int slots, int max_slots); 3529 void f2fs_drop_nlink(struct inode *dir, struct inode *inode); 3530 struct f2fs_dir_entry *__f2fs_find_entry(struct inode *dir, 3531 const struct f2fs_filename *fname, 3532 struct page **res_page); 3533 struct f2fs_dir_entry *f2fs_find_entry(struct inode *dir, 3534 const struct qstr *child, struct page **res_page); 3535 struct f2fs_dir_entry *f2fs_parent_dir(struct inode *dir, struct page **p); 3536 ino_t f2fs_inode_by_name(struct inode *dir, const struct qstr *qstr, 3537 struct page **page); 3538 void f2fs_set_link(struct inode *dir, struct f2fs_dir_entry *de, 3539 struct page *page, struct inode *inode); 3540 bool f2fs_has_enough_room(struct inode *dir, struct page *ipage, 3541 const struct f2fs_filename *fname); 3542 void f2fs_update_dentry(nid_t ino, umode_t mode, struct f2fs_dentry_ptr *d, 3543 const struct fscrypt_str *name, f2fs_hash_t name_hash, 3544 unsigned int bit_pos); 3545 int f2fs_add_regular_entry(struct inode *dir, const struct f2fs_filename *fname, 3546 struct inode *inode, nid_t ino, umode_t mode); 3547 int f2fs_add_dentry(struct inode *dir, const struct f2fs_filename *fname, 3548 struct inode *inode, nid_t ino, umode_t mode); 3549 int f2fs_do_add_link(struct inode *dir, const struct qstr *name, 3550 struct inode *inode, nid_t ino, umode_t mode); 3551 void f2fs_delete_entry(struct f2fs_dir_entry *dentry, struct page *page, 3552 struct inode *dir, struct inode *inode); 3553 int f2fs_do_tmpfile(struct inode *inode, struct inode *dir); 3554 bool f2fs_empty_dir(struct inode *dir); 3555 3556 static inline int f2fs_add_link(struct dentry *dentry, struct inode *inode) 3557 { 3558 if (fscrypt_is_nokey_name(dentry)) 3559 return -ENOKEY; 3560 return f2fs_do_add_link(d_inode(dentry->d_parent), &dentry->d_name, 3561 inode, inode->i_ino, inode->i_mode); 3562 } 3563 3564 /* 3565 * super.c 3566 */ 3567 int f2fs_inode_dirtied(struct inode *inode, bool sync); 3568 void f2fs_inode_synced(struct inode *inode); 3569 int f2fs_dquot_initialize(struct inode *inode); 3570 int f2fs_enable_quota_files(struct f2fs_sb_info *sbi, bool rdonly); 3571 int f2fs_quota_sync(struct super_block *sb, int type); 3572 loff_t max_file_blocks(struct inode *inode); 3573 void f2fs_quota_off_umount(struct super_block *sb); 3574 void f2fs_handle_stop(struct f2fs_sb_info *sbi, unsigned char reason); 3575 void f2fs_handle_error(struct f2fs_sb_info *sbi, unsigned char error); 3576 int f2fs_commit_super(struct f2fs_sb_info *sbi, bool recover); 3577 int f2fs_sync_fs(struct super_block *sb, int sync); 3578 int f2fs_sanity_check_ckpt(struct f2fs_sb_info *sbi); 3579 3580 /* 3581 * hash.c 3582 */ 3583 void f2fs_hash_filename(const struct inode *dir, struct f2fs_filename *fname); 3584 3585 /* 3586 * node.c 3587 */ 3588 struct node_info; 3589 3590 int f2fs_check_nid_range(struct f2fs_sb_info *sbi, nid_t nid); 3591 bool f2fs_available_free_memory(struct f2fs_sb_info *sbi, int type); 3592 bool f2fs_in_warm_node_list(struct f2fs_sb_info *sbi, struct page *page); 3593 void f2fs_init_fsync_node_info(struct f2fs_sb_info *sbi); 3594 void f2fs_del_fsync_node_entry(struct f2fs_sb_info *sbi, struct page *page); 3595 void f2fs_reset_fsync_node_info(struct f2fs_sb_info *sbi); 3596 int f2fs_need_dentry_mark(struct f2fs_sb_info *sbi, nid_t nid); 3597 bool f2fs_is_checkpointed_node(struct f2fs_sb_info *sbi, nid_t nid); 3598 bool f2fs_need_inode_block_update(struct f2fs_sb_info *sbi, nid_t ino); 3599 int f2fs_get_node_info(struct f2fs_sb_info *sbi, nid_t nid, 3600 struct node_info *ni, bool checkpoint_context); 3601 pgoff_t f2fs_get_next_page_offset(struct dnode_of_data *dn, pgoff_t pgofs); 3602 int f2fs_get_dnode_of_data(struct dnode_of_data *dn, pgoff_t index, int mode); 3603 int f2fs_truncate_inode_blocks(struct inode *inode, pgoff_t from); 3604 int f2fs_truncate_xattr_node(struct inode *inode); 3605 int f2fs_wait_on_node_pages_writeback(struct f2fs_sb_info *sbi, 3606 unsigned int seq_id); 3607 bool f2fs_nat_bitmap_enabled(struct f2fs_sb_info *sbi); 3608 int f2fs_remove_inode_page(struct inode *inode); 3609 struct page *f2fs_new_inode_page(struct inode *inode); 3610 struct page *f2fs_new_node_page(struct dnode_of_data *dn, unsigned int ofs); 3611 void f2fs_ra_node_page(struct f2fs_sb_info *sbi, nid_t nid); 3612 struct page *f2fs_get_node_page(struct f2fs_sb_info *sbi, pgoff_t nid); 3613 struct page *f2fs_get_node_page_ra(struct page *parent, int start); 3614 int f2fs_move_node_page(struct page *node_page, int gc_type); 3615 void f2fs_flush_inline_data(struct f2fs_sb_info *sbi); 3616 int f2fs_fsync_node_pages(struct f2fs_sb_info *sbi, struct inode *inode, 3617 struct writeback_control *wbc, bool atomic, 3618 unsigned int *seq_id); 3619 int f2fs_sync_node_pages(struct f2fs_sb_info *sbi, 3620 struct writeback_control *wbc, 3621 bool do_balance, enum iostat_type io_type); 3622 int f2fs_build_free_nids(struct f2fs_sb_info *sbi, bool sync, bool mount); 3623 bool f2fs_alloc_nid(struct f2fs_sb_info *sbi, nid_t *nid); 3624 void f2fs_alloc_nid_done(struct f2fs_sb_info *sbi, nid_t nid); 3625 void f2fs_alloc_nid_failed(struct f2fs_sb_info *sbi, nid_t nid); 3626 int f2fs_try_to_free_nids(struct f2fs_sb_info *sbi, int nr_shrink); 3627 int f2fs_recover_inline_xattr(struct inode *inode, struct page *page); 3628 int f2fs_recover_xattr_data(struct inode *inode, struct page *page); 3629 int f2fs_recover_inode_page(struct f2fs_sb_info *sbi, struct page *page); 3630 int f2fs_restore_node_summary(struct f2fs_sb_info *sbi, 3631 unsigned int segno, struct f2fs_summary_block *sum); 3632 void f2fs_enable_nat_bits(struct f2fs_sb_info *sbi); 3633 int f2fs_flush_nat_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3634 int f2fs_build_node_manager(struct f2fs_sb_info *sbi); 3635 void f2fs_destroy_node_manager(struct f2fs_sb_info *sbi); 3636 int __init f2fs_create_node_manager_caches(void); 3637 void f2fs_destroy_node_manager_caches(void); 3638 3639 /* 3640 * segment.c 3641 */ 3642 bool f2fs_need_SSR(struct f2fs_sb_info *sbi); 3643 int f2fs_commit_atomic_write(struct inode *inode); 3644 void f2fs_abort_atomic_write(struct inode *inode, bool clean); 3645 void f2fs_balance_fs(struct f2fs_sb_info *sbi, bool need); 3646 void f2fs_balance_fs_bg(struct f2fs_sb_info *sbi, bool from_bg); 3647 int f2fs_issue_flush(struct f2fs_sb_info *sbi, nid_t ino); 3648 int f2fs_create_flush_cmd_control(struct f2fs_sb_info *sbi); 3649 int f2fs_flush_device_cache(struct f2fs_sb_info *sbi); 3650 void f2fs_destroy_flush_cmd_control(struct f2fs_sb_info *sbi, bool free); 3651 void f2fs_invalidate_blocks(struct f2fs_sb_info *sbi, block_t addr); 3652 bool f2fs_is_checkpointed_data(struct f2fs_sb_info *sbi, block_t blkaddr); 3653 int f2fs_start_discard_thread(struct f2fs_sb_info *sbi); 3654 void f2fs_drop_discard_cmd(struct f2fs_sb_info *sbi); 3655 void f2fs_stop_discard_thread(struct f2fs_sb_info *sbi); 3656 bool f2fs_issue_discard_timeout(struct f2fs_sb_info *sbi); 3657 void f2fs_clear_prefree_segments(struct f2fs_sb_info *sbi, 3658 struct cp_control *cpc); 3659 void f2fs_dirty_to_prefree(struct f2fs_sb_info *sbi); 3660 block_t f2fs_get_unusable_blocks(struct f2fs_sb_info *sbi); 3661 int f2fs_disable_cp_again(struct f2fs_sb_info *sbi, block_t unusable); 3662 void f2fs_release_discard_addrs(struct f2fs_sb_info *sbi); 3663 int f2fs_npages_for_summary_flush(struct f2fs_sb_info *sbi, bool for_ra); 3664 bool f2fs_segment_has_free_slot(struct f2fs_sb_info *sbi, int segno); 3665 void f2fs_init_inmem_curseg(struct f2fs_sb_info *sbi); 3666 void f2fs_save_inmem_curseg(struct f2fs_sb_info *sbi); 3667 void f2fs_restore_inmem_curseg(struct f2fs_sb_info *sbi); 3668 void f2fs_get_new_segment(struct f2fs_sb_info *sbi, 3669 unsigned int *newseg, bool new_sec, int dir); 3670 void f2fs_allocate_segment_for_resize(struct f2fs_sb_info *sbi, int type, 3671 unsigned int start, unsigned int end); 3672 void f2fs_allocate_new_section(struct f2fs_sb_info *sbi, int type, bool force); 3673 void f2fs_allocate_new_segments(struct f2fs_sb_info *sbi); 3674 int f2fs_trim_fs(struct f2fs_sb_info *sbi, struct fstrim_range *range); 3675 bool f2fs_exist_trim_candidates(struct f2fs_sb_info *sbi, 3676 struct cp_control *cpc); 3677 struct page *f2fs_get_sum_page(struct f2fs_sb_info *sbi, unsigned int segno); 3678 void f2fs_update_meta_page(struct f2fs_sb_info *sbi, void *src, 3679 block_t blk_addr); 3680 void f2fs_do_write_meta_page(struct f2fs_sb_info *sbi, struct page *page, 3681 enum iostat_type io_type); 3682 void f2fs_do_write_node_page(unsigned int nid, struct f2fs_io_info *fio); 3683 void f2fs_outplace_write_data(struct dnode_of_data *dn, 3684 struct f2fs_io_info *fio); 3685 int f2fs_inplace_write_data(struct f2fs_io_info *fio); 3686 void f2fs_do_replace_block(struct f2fs_sb_info *sbi, struct f2fs_summary *sum, 3687 block_t old_blkaddr, block_t new_blkaddr, 3688 bool recover_curseg, bool recover_newaddr, 3689 bool from_gc); 3690 void f2fs_replace_block(struct f2fs_sb_info *sbi, struct dnode_of_data *dn, 3691 block_t old_addr, block_t new_addr, 3692 unsigned char version, bool recover_curseg, 3693 bool recover_newaddr); 3694 void f2fs_allocate_data_block(struct f2fs_sb_info *sbi, struct page *page, 3695 block_t old_blkaddr, block_t *new_blkaddr, 3696 struct f2fs_summary *sum, int type, 3697 struct f2fs_io_info *fio); 3698 void f2fs_update_device_state(struct f2fs_sb_info *sbi, nid_t ino, 3699 block_t blkaddr, unsigned int blkcnt); 3700 void f2fs_wait_on_page_writeback(struct page *page, 3701 enum page_type type, bool ordered, bool locked); 3702 void f2fs_wait_on_block_writeback(struct inode *inode, block_t blkaddr); 3703 void f2fs_wait_on_block_writeback_range(struct inode *inode, block_t blkaddr, 3704 block_t len); 3705 void f2fs_write_data_summaries(struct f2fs_sb_info *sbi, block_t start_blk); 3706 void f2fs_write_node_summaries(struct f2fs_sb_info *sbi, block_t start_blk); 3707 int f2fs_lookup_journal_in_cursum(struct f2fs_journal *journal, int type, 3708 unsigned int val, int alloc); 3709 void f2fs_flush_sit_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3710 int f2fs_fix_curseg_write_pointer(struct f2fs_sb_info *sbi); 3711 int f2fs_check_write_pointer(struct f2fs_sb_info *sbi); 3712 int f2fs_build_segment_manager(struct f2fs_sb_info *sbi); 3713 void f2fs_destroy_segment_manager(struct f2fs_sb_info *sbi); 3714 int __init f2fs_create_segment_manager_caches(void); 3715 void f2fs_destroy_segment_manager_caches(void); 3716 int f2fs_rw_hint_to_seg_type(enum rw_hint hint); 3717 unsigned int f2fs_usable_segs_in_sec(struct f2fs_sb_info *sbi, 3718 unsigned int segno); 3719 unsigned int f2fs_usable_blks_in_seg(struct f2fs_sb_info *sbi, 3720 unsigned int segno); 3721 3722 #define DEF_FRAGMENT_SIZE 4 3723 #define MIN_FRAGMENT_SIZE 1 3724 #define MAX_FRAGMENT_SIZE 512 3725 3726 static inline bool f2fs_need_rand_seg(struct f2fs_sb_info *sbi) 3727 { 3728 return F2FS_OPTION(sbi).fs_mode == FS_MODE_FRAGMENT_SEG || 3729 F2FS_OPTION(sbi).fs_mode == FS_MODE_FRAGMENT_BLK; 3730 } 3731 3732 /* 3733 * checkpoint.c 3734 */ 3735 void f2fs_stop_checkpoint(struct f2fs_sb_info *sbi, bool end_io, 3736 unsigned char reason); 3737 void f2fs_flush_ckpt_thread(struct f2fs_sb_info *sbi); 3738 struct page *f2fs_grab_meta_page(struct f2fs_sb_info *sbi, pgoff_t index); 3739 struct page *f2fs_get_meta_page(struct f2fs_sb_info *sbi, pgoff_t index); 3740 struct page *f2fs_get_meta_page_retry(struct f2fs_sb_info *sbi, pgoff_t index); 3741 struct page *f2fs_get_tmp_page(struct f2fs_sb_info *sbi, pgoff_t index); 3742 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi, 3743 block_t blkaddr, int type); 3744 int f2fs_ra_meta_pages(struct f2fs_sb_info *sbi, block_t start, int nrpages, 3745 int type, bool sync); 3746 void f2fs_ra_meta_pages_cond(struct f2fs_sb_info *sbi, pgoff_t index, 3747 unsigned int ra_blocks); 3748 long f2fs_sync_meta_pages(struct f2fs_sb_info *sbi, enum page_type type, 3749 long nr_to_write, enum iostat_type io_type); 3750 void f2fs_add_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type); 3751 void f2fs_remove_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type); 3752 void f2fs_release_ino_entry(struct f2fs_sb_info *sbi, bool all); 3753 bool f2fs_exist_written_data(struct f2fs_sb_info *sbi, nid_t ino, int mode); 3754 void f2fs_set_dirty_device(struct f2fs_sb_info *sbi, nid_t ino, 3755 unsigned int devidx, int type); 3756 bool f2fs_is_dirty_device(struct f2fs_sb_info *sbi, nid_t ino, 3757 unsigned int devidx, int type); 3758 int f2fs_sync_inode_meta(struct f2fs_sb_info *sbi); 3759 int f2fs_acquire_orphan_inode(struct f2fs_sb_info *sbi); 3760 void f2fs_release_orphan_inode(struct f2fs_sb_info *sbi); 3761 void f2fs_add_orphan_inode(struct inode *inode); 3762 void f2fs_remove_orphan_inode(struct f2fs_sb_info *sbi, nid_t ino); 3763 int f2fs_recover_orphan_inodes(struct f2fs_sb_info *sbi); 3764 int f2fs_get_valid_checkpoint(struct f2fs_sb_info *sbi); 3765 void f2fs_update_dirty_folio(struct inode *inode, struct folio *folio); 3766 void f2fs_remove_dirty_inode(struct inode *inode); 3767 int f2fs_sync_dirty_inodes(struct f2fs_sb_info *sbi, enum inode_type type, 3768 bool from_cp); 3769 void f2fs_wait_on_all_pages(struct f2fs_sb_info *sbi, int type); 3770 u64 f2fs_get_sectors_written(struct f2fs_sb_info *sbi); 3771 int f2fs_write_checkpoint(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3772 void f2fs_init_ino_entry_info(struct f2fs_sb_info *sbi); 3773 int __init f2fs_create_checkpoint_caches(void); 3774 void f2fs_destroy_checkpoint_caches(void); 3775 int f2fs_issue_checkpoint(struct f2fs_sb_info *sbi); 3776 int f2fs_start_ckpt_thread(struct f2fs_sb_info *sbi); 3777 void f2fs_stop_ckpt_thread(struct f2fs_sb_info *sbi); 3778 void f2fs_init_ckpt_req_control(struct f2fs_sb_info *sbi); 3779 3780 /* 3781 * data.c 3782 */ 3783 int __init f2fs_init_bioset(void); 3784 void f2fs_destroy_bioset(void); 3785 int f2fs_init_bio_entry_cache(void); 3786 void f2fs_destroy_bio_entry_cache(void); 3787 void f2fs_submit_bio(struct f2fs_sb_info *sbi, 3788 struct bio *bio, enum page_type type); 3789 int f2fs_init_write_merge_io(struct f2fs_sb_info *sbi); 3790 void f2fs_submit_merged_write(struct f2fs_sb_info *sbi, enum page_type type); 3791 void f2fs_submit_merged_write_cond(struct f2fs_sb_info *sbi, 3792 struct inode *inode, struct page *page, 3793 nid_t ino, enum page_type type); 3794 void f2fs_submit_merged_ipu_write(struct f2fs_sb_info *sbi, 3795 struct bio **bio, struct page *page); 3796 void f2fs_flush_merged_writes(struct f2fs_sb_info *sbi); 3797 int f2fs_submit_page_bio(struct f2fs_io_info *fio); 3798 int f2fs_merge_page_bio(struct f2fs_io_info *fio); 3799 void f2fs_submit_page_write(struct f2fs_io_info *fio); 3800 struct block_device *f2fs_target_device(struct f2fs_sb_info *sbi, 3801 block_t blk_addr, sector_t *sector); 3802 int f2fs_target_device_index(struct f2fs_sb_info *sbi, block_t blkaddr); 3803 void f2fs_set_data_blkaddr(struct dnode_of_data *dn); 3804 void f2fs_update_data_blkaddr(struct dnode_of_data *dn, block_t blkaddr); 3805 int f2fs_reserve_new_blocks(struct dnode_of_data *dn, blkcnt_t count); 3806 int f2fs_reserve_new_block(struct dnode_of_data *dn); 3807 int f2fs_get_block(struct dnode_of_data *dn, pgoff_t index); 3808 int f2fs_reserve_block(struct dnode_of_data *dn, pgoff_t index); 3809 struct page *f2fs_get_read_data_page(struct inode *inode, pgoff_t index, 3810 blk_opf_t op_flags, bool for_write); 3811 struct page *f2fs_find_data_page(struct inode *inode, pgoff_t index); 3812 struct page *f2fs_get_lock_data_page(struct inode *inode, pgoff_t index, 3813 bool for_write); 3814 struct page *f2fs_get_new_data_page(struct inode *inode, 3815 struct page *ipage, pgoff_t index, bool new_i_size); 3816 int f2fs_do_write_data_page(struct f2fs_io_info *fio); 3817 void f2fs_do_map_lock(struct f2fs_sb_info *sbi, int flag, bool lock); 3818 int f2fs_map_blocks(struct inode *inode, struct f2fs_map_blocks *map, 3819 int create, int flag); 3820 int f2fs_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, 3821 u64 start, u64 len); 3822 int f2fs_encrypt_one_page(struct f2fs_io_info *fio); 3823 bool f2fs_should_update_inplace(struct inode *inode, struct f2fs_io_info *fio); 3824 bool f2fs_should_update_outplace(struct inode *inode, struct f2fs_io_info *fio); 3825 int f2fs_write_single_data_page(struct page *page, int *submitted, 3826 struct bio **bio, sector_t *last_block, 3827 struct writeback_control *wbc, 3828 enum iostat_type io_type, 3829 int compr_blocks, bool allow_balance); 3830 void f2fs_write_failed(struct inode *inode, loff_t to); 3831 void f2fs_invalidate_folio(struct folio *folio, size_t offset, size_t length); 3832 bool f2fs_release_folio(struct folio *folio, gfp_t wait); 3833 bool f2fs_overwrite_io(struct inode *inode, loff_t pos, size_t len); 3834 void f2fs_clear_page_cache_dirty_tag(struct page *page); 3835 int f2fs_init_post_read_processing(void); 3836 void f2fs_destroy_post_read_processing(void); 3837 int f2fs_init_post_read_wq(struct f2fs_sb_info *sbi); 3838 void f2fs_destroy_post_read_wq(struct f2fs_sb_info *sbi); 3839 extern const struct iomap_ops f2fs_iomap_ops; 3840 3841 /* 3842 * gc.c 3843 */ 3844 int f2fs_start_gc_thread(struct f2fs_sb_info *sbi); 3845 void f2fs_stop_gc_thread(struct f2fs_sb_info *sbi); 3846 block_t f2fs_start_bidx_of_node(unsigned int node_ofs, struct inode *inode); 3847 int f2fs_gc(struct f2fs_sb_info *sbi, struct f2fs_gc_control *gc_control); 3848 void f2fs_build_gc_manager(struct f2fs_sb_info *sbi); 3849 int f2fs_resize_fs(struct f2fs_sb_info *sbi, __u64 block_count); 3850 int __init f2fs_create_garbage_collection_cache(void); 3851 void f2fs_destroy_garbage_collection_cache(void); 3852 3853 /* 3854 * recovery.c 3855 */ 3856 int f2fs_recover_fsync_data(struct f2fs_sb_info *sbi, bool check_only); 3857 bool f2fs_space_for_roll_forward(struct f2fs_sb_info *sbi); 3858 int __init f2fs_create_recovery_cache(void); 3859 void f2fs_destroy_recovery_cache(void); 3860 3861 /* 3862 * debug.c 3863 */ 3864 #ifdef CONFIG_F2FS_STAT_FS 3865 struct f2fs_stat_info { 3866 struct list_head stat_list; 3867 struct f2fs_sb_info *sbi; 3868 int all_area_segs, sit_area_segs, nat_area_segs, ssa_area_segs; 3869 int main_area_segs, main_area_sections, main_area_zones; 3870 unsigned long long hit_largest, hit_cached, hit_rbtree; 3871 unsigned long long hit_total, total_ext; 3872 int ext_tree, zombie_tree, ext_node; 3873 int ndirty_node, ndirty_dent, ndirty_meta, ndirty_imeta; 3874 int ndirty_data, ndirty_qdata; 3875 unsigned int ndirty_dirs, ndirty_files, nquota_files, ndirty_all; 3876 int nats, dirty_nats, sits, dirty_sits; 3877 int free_nids, avail_nids, alloc_nids; 3878 int total_count, utilization; 3879 int bg_gc, nr_wb_cp_data, nr_wb_data; 3880 int nr_rd_data, nr_rd_node, nr_rd_meta; 3881 int nr_dio_read, nr_dio_write; 3882 unsigned int io_skip_bggc, other_skip_bggc; 3883 int nr_flushing, nr_flushed, flush_list_empty; 3884 int nr_discarding, nr_discarded; 3885 int nr_discard_cmd; 3886 unsigned int undiscard_blks; 3887 int nr_issued_ckpt, nr_total_ckpt, nr_queued_ckpt; 3888 unsigned int cur_ckpt_time, peak_ckpt_time; 3889 int inline_xattr, inline_inode, inline_dir, append, update, orphans; 3890 int compr_inode, swapfile_inode; 3891 unsigned long long compr_blocks; 3892 int aw_cnt, max_aw_cnt; 3893 unsigned int valid_count, valid_node_count, valid_inode_count, discard_blks; 3894 unsigned int bimodal, avg_vblocks; 3895 int util_free, util_valid, util_invalid; 3896 int rsvd_segs, overp_segs; 3897 int dirty_count, node_pages, meta_pages, compress_pages; 3898 int compress_page_hit; 3899 int prefree_count, call_count, cp_count, bg_cp_count; 3900 int tot_segs, node_segs, data_segs, free_segs, free_secs; 3901 int bg_node_segs, bg_data_segs; 3902 int tot_blks, data_blks, node_blks; 3903 int bg_data_blks, bg_node_blks; 3904 int curseg[NR_CURSEG_TYPE]; 3905 int cursec[NR_CURSEG_TYPE]; 3906 int curzone[NR_CURSEG_TYPE]; 3907 unsigned int dirty_seg[NR_CURSEG_TYPE]; 3908 unsigned int full_seg[NR_CURSEG_TYPE]; 3909 unsigned int valid_blks[NR_CURSEG_TYPE]; 3910 3911 unsigned int meta_count[META_MAX]; 3912 unsigned int segment_count[2]; 3913 unsigned int block_count[2]; 3914 unsigned int inplace_count; 3915 unsigned long long base_mem, cache_mem, page_mem; 3916 }; 3917 3918 static inline struct f2fs_stat_info *F2FS_STAT(struct f2fs_sb_info *sbi) 3919 { 3920 return (struct f2fs_stat_info *)sbi->stat_info; 3921 } 3922 3923 #define stat_inc_cp_count(si) ((si)->cp_count++) 3924 #define stat_inc_bg_cp_count(si) ((si)->bg_cp_count++) 3925 #define stat_inc_call_count(si) ((si)->call_count++) 3926 #define stat_inc_bggc_count(si) ((si)->bg_gc++) 3927 #define stat_io_skip_bggc_count(sbi) ((sbi)->io_skip_bggc++) 3928 #define stat_other_skip_bggc_count(sbi) ((sbi)->other_skip_bggc++) 3929 #define stat_inc_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]++) 3930 #define stat_dec_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]--) 3931 #define stat_inc_total_hit(sbi) (atomic64_inc(&(sbi)->total_hit_ext)) 3932 #define stat_inc_rbtree_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_rbtree)) 3933 #define stat_inc_largest_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_largest)) 3934 #define stat_inc_cached_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_cached)) 3935 #define stat_inc_inline_xattr(inode) \ 3936 do { \ 3937 if (f2fs_has_inline_xattr(inode)) \ 3938 (atomic_inc(&F2FS_I_SB(inode)->inline_xattr)); \ 3939 } while (0) 3940 #define stat_dec_inline_xattr(inode) \ 3941 do { \ 3942 if (f2fs_has_inline_xattr(inode)) \ 3943 (atomic_dec(&F2FS_I_SB(inode)->inline_xattr)); \ 3944 } while (0) 3945 #define stat_inc_inline_inode(inode) \ 3946 do { \ 3947 if (f2fs_has_inline_data(inode)) \ 3948 (atomic_inc(&F2FS_I_SB(inode)->inline_inode)); \ 3949 } while (0) 3950 #define stat_dec_inline_inode(inode) \ 3951 do { \ 3952 if (f2fs_has_inline_data(inode)) \ 3953 (atomic_dec(&F2FS_I_SB(inode)->inline_inode)); \ 3954 } while (0) 3955 #define stat_inc_inline_dir(inode) \ 3956 do { \ 3957 if (f2fs_has_inline_dentry(inode)) \ 3958 (atomic_inc(&F2FS_I_SB(inode)->inline_dir)); \ 3959 } while (0) 3960 #define stat_dec_inline_dir(inode) \ 3961 do { \ 3962 if (f2fs_has_inline_dentry(inode)) \ 3963 (atomic_dec(&F2FS_I_SB(inode)->inline_dir)); \ 3964 } while (0) 3965 #define stat_inc_compr_inode(inode) \ 3966 do { \ 3967 if (f2fs_compressed_file(inode)) \ 3968 (atomic_inc(&F2FS_I_SB(inode)->compr_inode)); \ 3969 } while (0) 3970 #define stat_dec_compr_inode(inode) \ 3971 do { \ 3972 if (f2fs_compressed_file(inode)) \ 3973 (atomic_dec(&F2FS_I_SB(inode)->compr_inode)); \ 3974 } while (0) 3975 #define stat_add_compr_blocks(inode, blocks) \ 3976 (atomic64_add(blocks, &F2FS_I_SB(inode)->compr_blocks)) 3977 #define stat_sub_compr_blocks(inode, blocks) \ 3978 (atomic64_sub(blocks, &F2FS_I_SB(inode)->compr_blocks)) 3979 #define stat_inc_swapfile_inode(inode) \ 3980 (atomic_inc(&F2FS_I_SB(inode)->swapfile_inode)) 3981 #define stat_dec_swapfile_inode(inode) \ 3982 (atomic_dec(&F2FS_I_SB(inode)->swapfile_inode)) 3983 #define stat_inc_atomic_inode(inode) \ 3984 (atomic_inc(&F2FS_I_SB(inode)->atomic_files)) 3985 #define stat_dec_atomic_inode(inode) \ 3986 (atomic_dec(&F2FS_I_SB(inode)->atomic_files)) 3987 #define stat_inc_meta_count(sbi, blkaddr) \ 3988 do { \ 3989 if (blkaddr < SIT_I(sbi)->sit_base_addr) \ 3990 atomic_inc(&(sbi)->meta_count[META_CP]); \ 3991 else if (blkaddr < NM_I(sbi)->nat_blkaddr) \ 3992 atomic_inc(&(sbi)->meta_count[META_SIT]); \ 3993 else if (blkaddr < SM_I(sbi)->ssa_blkaddr) \ 3994 atomic_inc(&(sbi)->meta_count[META_NAT]); \ 3995 else if (blkaddr < SM_I(sbi)->main_blkaddr) \ 3996 atomic_inc(&(sbi)->meta_count[META_SSA]); \ 3997 } while (0) 3998 #define stat_inc_seg_type(sbi, curseg) \ 3999 ((sbi)->segment_count[(curseg)->alloc_type]++) 4000 #define stat_inc_block_count(sbi, curseg) \ 4001 ((sbi)->block_count[(curseg)->alloc_type]++) 4002 #define stat_inc_inplace_blocks(sbi) \ 4003 (atomic_inc(&(sbi)->inplace_count)) 4004 #define stat_update_max_atomic_write(inode) \ 4005 do { \ 4006 int cur = atomic_read(&F2FS_I_SB(inode)->atomic_files); \ 4007 int max = atomic_read(&F2FS_I_SB(inode)->max_aw_cnt); \ 4008 if (cur > max) \ 4009 atomic_set(&F2FS_I_SB(inode)->max_aw_cnt, cur); \ 4010 } while (0) 4011 #define stat_inc_seg_count(sbi, type, gc_type) \ 4012 do { \ 4013 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 4014 si->tot_segs++; \ 4015 if ((type) == SUM_TYPE_DATA) { \ 4016 si->data_segs++; \ 4017 si->bg_data_segs += (gc_type == BG_GC) ? 1 : 0; \ 4018 } else { \ 4019 si->node_segs++; \ 4020 si->bg_node_segs += (gc_type == BG_GC) ? 1 : 0; \ 4021 } \ 4022 } while (0) 4023 4024 #define stat_inc_tot_blk_count(si, blks) \ 4025 ((si)->tot_blks += (blks)) 4026 4027 #define stat_inc_data_blk_count(sbi, blks, gc_type) \ 4028 do { \ 4029 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 4030 stat_inc_tot_blk_count(si, blks); \ 4031 si->data_blks += (blks); \ 4032 si->bg_data_blks += ((gc_type) == BG_GC) ? (blks) : 0; \ 4033 } while (0) 4034 4035 #define stat_inc_node_blk_count(sbi, blks, gc_type) \ 4036 do { \ 4037 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 4038 stat_inc_tot_blk_count(si, blks); \ 4039 si->node_blks += (blks); \ 4040 si->bg_node_blks += ((gc_type) == BG_GC) ? (blks) : 0; \ 4041 } while (0) 4042 4043 int f2fs_build_stats(struct f2fs_sb_info *sbi); 4044 void f2fs_destroy_stats(struct f2fs_sb_info *sbi); 4045 void __init f2fs_create_root_stats(void); 4046 void f2fs_destroy_root_stats(void); 4047 void f2fs_update_sit_info(struct f2fs_sb_info *sbi); 4048 #else 4049 #define stat_inc_cp_count(si) do { } while (0) 4050 #define stat_inc_bg_cp_count(si) do { } while (0) 4051 #define stat_inc_call_count(si) do { } while (0) 4052 #define stat_inc_bggc_count(si) do { } while (0) 4053 #define stat_io_skip_bggc_count(sbi) do { } while (0) 4054 #define stat_other_skip_bggc_count(sbi) do { } while (0) 4055 #define stat_inc_dirty_inode(sbi, type) do { } while (0) 4056 #define stat_dec_dirty_inode(sbi, type) do { } while (0) 4057 #define stat_inc_total_hit(sbi) do { } while (0) 4058 #define stat_inc_rbtree_node_hit(sbi) do { } while (0) 4059 #define stat_inc_largest_node_hit(sbi) do { } while (0) 4060 #define stat_inc_cached_node_hit(sbi) do { } while (0) 4061 #define stat_inc_inline_xattr(inode) do { } while (0) 4062 #define stat_dec_inline_xattr(inode) do { } while (0) 4063 #define stat_inc_inline_inode(inode) do { } while (0) 4064 #define stat_dec_inline_inode(inode) do { } while (0) 4065 #define stat_inc_inline_dir(inode) do { } while (0) 4066 #define stat_dec_inline_dir(inode) do { } while (0) 4067 #define stat_inc_compr_inode(inode) do { } while (0) 4068 #define stat_dec_compr_inode(inode) do { } while (0) 4069 #define stat_add_compr_blocks(inode, blocks) do { } while (0) 4070 #define stat_sub_compr_blocks(inode, blocks) do { } while (0) 4071 #define stat_inc_swapfile_inode(inode) do { } while (0) 4072 #define stat_dec_swapfile_inode(inode) do { } while (0) 4073 #define stat_inc_atomic_inode(inode) do { } while (0) 4074 #define stat_dec_atomic_inode(inode) do { } while (0) 4075 #define stat_update_max_atomic_write(inode) do { } while (0) 4076 #define stat_inc_meta_count(sbi, blkaddr) do { } while (0) 4077 #define stat_inc_seg_type(sbi, curseg) do { } while (0) 4078 #define stat_inc_block_count(sbi, curseg) do { } while (0) 4079 #define stat_inc_inplace_blocks(sbi) do { } while (0) 4080 #define stat_inc_seg_count(sbi, type, gc_type) do { } while (0) 4081 #define stat_inc_tot_blk_count(si, blks) do { } while (0) 4082 #define stat_inc_data_blk_count(sbi, blks, gc_type) do { } while (0) 4083 #define stat_inc_node_blk_count(sbi, blks, gc_type) do { } while (0) 4084 4085 static inline int f2fs_build_stats(struct f2fs_sb_info *sbi) { return 0; } 4086 static inline void f2fs_destroy_stats(struct f2fs_sb_info *sbi) { } 4087 static inline void __init f2fs_create_root_stats(void) { } 4088 static inline void f2fs_destroy_root_stats(void) { } 4089 static inline void f2fs_update_sit_info(struct f2fs_sb_info *sbi) {} 4090 #endif 4091 4092 extern const struct file_operations f2fs_dir_operations; 4093 extern const struct file_operations f2fs_file_operations; 4094 extern const struct inode_operations f2fs_file_inode_operations; 4095 extern const struct address_space_operations f2fs_dblock_aops; 4096 extern const struct address_space_operations f2fs_node_aops; 4097 extern const struct address_space_operations f2fs_meta_aops; 4098 extern const struct inode_operations f2fs_dir_inode_operations; 4099 extern const struct inode_operations f2fs_symlink_inode_operations; 4100 extern const struct inode_operations f2fs_encrypted_symlink_inode_operations; 4101 extern const struct inode_operations f2fs_special_inode_operations; 4102 extern struct kmem_cache *f2fs_inode_entry_slab; 4103 4104 /* 4105 * inline.c 4106 */ 4107 bool f2fs_may_inline_data(struct inode *inode); 4108 bool f2fs_sanity_check_inline_data(struct inode *inode); 4109 bool f2fs_may_inline_dentry(struct inode *inode); 4110 void f2fs_do_read_inline_data(struct page *page, struct page *ipage); 4111 void f2fs_truncate_inline_inode(struct inode *inode, 4112 struct page *ipage, u64 from); 4113 int f2fs_read_inline_data(struct inode *inode, struct page *page); 4114 int f2fs_convert_inline_page(struct dnode_of_data *dn, struct page *page); 4115 int f2fs_convert_inline_inode(struct inode *inode); 4116 int f2fs_try_convert_inline_dir(struct inode *dir, struct dentry *dentry); 4117 int f2fs_write_inline_data(struct inode *inode, struct page *page); 4118 int f2fs_recover_inline_data(struct inode *inode, struct page *npage); 4119 struct f2fs_dir_entry *f2fs_find_in_inline_dir(struct inode *dir, 4120 const struct f2fs_filename *fname, 4121 struct page **res_page); 4122 int f2fs_make_empty_inline_dir(struct inode *inode, struct inode *parent, 4123 struct page *ipage); 4124 int f2fs_add_inline_entry(struct inode *dir, const struct f2fs_filename *fname, 4125 struct inode *inode, nid_t ino, umode_t mode); 4126 void f2fs_delete_inline_entry(struct f2fs_dir_entry *dentry, 4127 struct page *page, struct inode *dir, 4128 struct inode *inode); 4129 bool f2fs_empty_inline_dir(struct inode *dir); 4130 int f2fs_read_inline_dir(struct file *file, struct dir_context *ctx, 4131 struct fscrypt_str *fstr); 4132 int f2fs_inline_data_fiemap(struct inode *inode, 4133 struct fiemap_extent_info *fieinfo, 4134 __u64 start, __u64 len); 4135 4136 /* 4137 * shrinker.c 4138 */ 4139 unsigned long f2fs_shrink_count(struct shrinker *shrink, 4140 struct shrink_control *sc); 4141 unsigned long f2fs_shrink_scan(struct shrinker *shrink, 4142 struct shrink_control *sc); 4143 void f2fs_join_shrinker(struct f2fs_sb_info *sbi); 4144 void f2fs_leave_shrinker(struct f2fs_sb_info *sbi); 4145 4146 /* 4147 * extent_cache.c 4148 */ 4149 struct rb_entry *f2fs_lookup_rb_tree(struct rb_root_cached *root, 4150 struct rb_entry *cached_re, unsigned int ofs); 4151 struct rb_node **f2fs_lookup_rb_tree_ext(struct f2fs_sb_info *sbi, 4152 struct rb_root_cached *root, 4153 struct rb_node **parent, 4154 unsigned long long key, bool *left_most); 4155 struct rb_node **f2fs_lookup_rb_tree_for_insert(struct f2fs_sb_info *sbi, 4156 struct rb_root_cached *root, 4157 struct rb_node **parent, 4158 unsigned int ofs, bool *leftmost); 4159 struct rb_entry *f2fs_lookup_rb_tree_ret(struct rb_root_cached *root, 4160 struct rb_entry *cached_re, unsigned int ofs, 4161 struct rb_entry **prev_entry, struct rb_entry **next_entry, 4162 struct rb_node ***insert_p, struct rb_node **insert_parent, 4163 bool force, bool *leftmost); 4164 bool f2fs_check_rb_tree_consistence(struct f2fs_sb_info *sbi, 4165 struct rb_root_cached *root, bool check_key); 4166 unsigned int f2fs_shrink_extent_tree(struct f2fs_sb_info *sbi, int nr_shrink); 4167 void f2fs_init_extent_tree(struct inode *inode, struct page *ipage); 4168 void f2fs_drop_extent_tree(struct inode *inode); 4169 unsigned int f2fs_destroy_extent_node(struct inode *inode); 4170 void f2fs_destroy_extent_tree(struct inode *inode); 4171 bool f2fs_lookup_extent_cache(struct inode *inode, pgoff_t pgofs, 4172 struct extent_info *ei); 4173 void f2fs_update_extent_cache(struct dnode_of_data *dn); 4174 void f2fs_update_extent_cache_range(struct dnode_of_data *dn, 4175 pgoff_t fofs, block_t blkaddr, unsigned int len); 4176 void f2fs_init_extent_cache_info(struct f2fs_sb_info *sbi); 4177 int __init f2fs_create_extent_cache(void); 4178 void f2fs_destroy_extent_cache(void); 4179 4180 /* 4181 * sysfs.c 4182 */ 4183 #define MIN_RA_MUL 2 4184 #define MAX_RA_MUL 256 4185 4186 int __init f2fs_init_sysfs(void); 4187 void f2fs_exit_sysfs(void); 4188 int f2fs_register_sysfs(struct f2fs_sb_info *sbi); 4189 void f2fs_unregister_sysfs(struct f2fs_sb_info *sbi); 4190 4191 /* verity.c */ 4192 extern const struct fsverity_operations f2fs_verityops; 4193 4194 /* 4195 * crypto support 4196 */ 4197 static inline bool f2fs_encrypted_file(struct inode *inode) 4198 { 4199 return IS_ENCRYPTED(inode) && S_ISREG(inode->i_mode); 4200 } 4201 4202 static inline void f2fs_set_encrypted_inode(struct inode *inode) 4203 { 4204 #ifdef CONFIG_FS_ENCRYPTION 4205 file_set_encrypt(inode); 4206 f2fs_set_inode_flags(inode); 4207 #endif 4208 } 4209 4210 /* 4211 * Returns true if the reads of the inode's data need to undergo some 4212 * postprocessing step, like decryption or authenticity verification. 4213 */ 4214 static inline bool f2fs_post_read_required(struct inode *inode) 4215 { 4216 return f2fs_encrypted_file(inode) || fsverity_active(inode) || 4217 f2fs_compressed_file(inode); 4218 } 4219 4220 /* 4221 * compress.c 4222 */ 4223 #ifdef CONFIG_F2FS_FS_COMPRESSION 4224 bool f2fs_is_compressed_page(struct page *page); 4225 struct page *f2fs_compress_control_page(struct page *page); 4226 int f2fs_prepare_compress_overwrite(struct inode *inode, 4227 struct page **pagep, pgoff_t index, void **fsdata); 4228 bool f2fs_compress_write_end(struct inode *inode, void *fsdata, 4229 pgoff_t index, unsigned copied); 4230 int f2fs_truncate_partial_cluster(struct inode *inode, u64 from, bool lock); 4231 void f2fs_compress_write_end_io(struct bio *bio, struct page *page); 4232 bool f2fs_is_compress_backend_ready(struct inode *inode); 4233 int f2fs_init_compress_mempool(void); 4234 void f2fs_destroy_compress_mempool(void); 4235 void f2fs_decompress_cluster(struct decompress_io_ctx *dic, bool in_task); 4236 void f2fs_end_read_compressed_page(struct page *page, bool failed, 4237 block_t blkaddr, bool in_task); 4238 bool f2fs_cluster_is_empty(struct compress_ctx *cc); 4239 bool f2fs_cluster_can_merge_page(struct compress_ctx *cc, pgoff_t index); 4240 bool f2fs_all_cluster_page_ready(struct compress_ctx *cc, struct page **pages, 4241 int index, int nr_pages, bool uptodate); 4242 bool f2fs_sanity_check_cluster(struct dnode_of_data *dn); 4243 void f2fs_compress_ctx_add_page(struct compress_ctx *cc, struct page *page); 4244 int f2fs_write_multi_pages(struct compress_ctx *cc, 4245 int *submitted, 4246 struct writeback_control *wbc, 4247 enum iostat_type io_type); 4248 int f2fs_is_compressed_cluster(struct inode *inode, pgoff_t index); 4249 void f2fs_update_extent_tree_range_compressed(struct inode *inode, 4250 pgoff_t fofs, block_t blkaddr, unsigned int llen, 4251 unsigned int c_len); 4252 int f2fs_read_multi_pages(struct compress_ctx *cc, struct bio **bio_ret, 4253 unsigned nr_pages, sector_t *last_block_in_bio, 4254 bool is_readahead, bool for_write); 4255 struct decompress_io_ctx *f2fs_alloc_dic(struct compress_ctx *cc); 4256 void f2fs_decompress_end_io(struct decompress_io_ctx *dic, bool failed, 4257 bool in_task); 4258 void f2fs_put_page_dic(struct page *page, bool in_task); 4259 unsigned int f2fs_cluster_blocks_are_contiguous(struct dnode_of_data *dn); 4260 int f2fs_init_compress_ctx(struct compress_ctx *cc); 4261 void f2fs_destroy_compress_ctx(struct compress_ctx *cc, bool reuse); 4262 void f2fs_init_compress_info(struct f2fs_sb_info *sbi); 4263 int f2fs_init_compress_inode(struct f2fs_sb_info *sbi); 4264 void f2fs_destroy_compress_inode(struct f2fs_sb_info *sbi); 4265 int f2fs_init_page_array_cache(struct f2fs_sb_info *sbi); 4266 void f2fs_destroy_page_array_cache(struct f2fs_sb_info *sbi); 4267 int __init f2fs_init_compress_cache(void); 4268 void f2fs_destroy_compress_cache(void); 4269 struct address_space *COMPRESS_MAPPING(struct f2fs_sb_info *sbi); 4270 void f2fs_invalidate_compress_page(struct f2fs_sb_info *sbi, block_t blkaddr); 4271 void f2fs_cache_compressed_page(struct f2fs_sb_info *sbi, struct page *page, 4272 nid_t ino, block_t blkaddr); 4273 bool f2fs_load_compressed_page(struct f2fs_sb_info *sbi, struct page *page, 4274 block_t blkaddr); 4275 void f2fs_invalidate_compress_pages(struct f2fs_sb_info *sbi, nid_t ino); 4276 #define inc_compr_inode_stat(inode) \ 4277 do { \ 4278 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); \ 4279 sbi->compr_new_inode++; \ 4280 } while (0) 4281 #define add_compr_block_stat(inode, blocks) \ 4282 do { \ 4283 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); \ 4284 int diff = F2FS_I(inode)->i_cluster_size - blocks; \ 4285 sbi->compr_written_block += blocks; \ 4286 sbi->compr_saved_block += diff; \ 4287 } while (0) 4288 #else 4289 static inline bool f2fs_is_compressed_page(struct page *page) { return false; } 4290 static inline bool f2fs_is_compress_backend_ready(struct inode *inode) 4291 { 4292 if (!f2fs_compressed_file(inode)) 4293 return true; 4294 /* not support compression */ 4295 return false; 4296 } 4297 static inline struct page *f2fs_compress_control_page(struct page *page) 4298 { 4299 WARN_ON_ONCE(1); 4300 return ERR_PTR(-EINVAL); 4301 } 4302 static inline int f2fs_init_compress_mempool(void) { return 0; } 4303 static inline void f2fs_destroy_compress_mempool(void) { } 4304 static inline void f2fs_decompress_cluster(struct decompress_io_ctx *dic, 4305 bool in_task) { } 4306 static inline void f2fs_end_read_compressed_page(struct page *page, 4307 bool failed, block_t blkaddr, bool in_task) 4308 { 4309 WARN_ON_ONCE(1); 4310 } 4311 static inline void f2fs_put_page_dic(struct page *page, bool in_task) 4312 { 4313 WARN_ON_ONCE(1); 4314 } 4315 static inline unsigned int f2fs_cluster_blocks_are_contiguous(struct dnode_of_data *dn) { return 0; } 4316 static inline bool f2fs_sanity_check_cluster(struct dnode_of_data *dn) { return false; } 4317 static inline int f2fs_init_compress_inode(struct f2fs_sb_info *sbi) { return 0; } 4318 static inline void f2fs_destroy_compress_inode(struct f2fs_sb_info *sbi) { } 4319 static inline int f2fs_init_page_array_cache(struct f2fs_sb_info *sbi) { return 0; } 4320 static inline void f2fs_destroy_page_array_cache(struct f2fs_sb_info *sbi) { } 4321 static inline int __init f2fs_init_compress_cache(void) { return 0; } 4322 static inline void f2fs_destroy_compress_cache(void) { } 4323 static inline void f2fs_invalidate_compress_page(struct f2fs_sb_info *sbi, 4324 block_t blkaddr) { } 4325 static inline void f2fs_cache_compressed_page(struct f2fs_sb_info *sbi, 4326 struct page *page, nid_t ino, block_t blkaddr) { } 4327 static inline bool f2fs_load_compressed_page(struct f2fs_sb_info *sbi, 4328 struct page *page, block_t blkaddr) { return false; } 4329 static inline void f2fs_invalidate_compress_pages(struct f2fs_sb_info *sbi, 4330 nid_t ino) { } 4331 #define inc_compr_inode_stat(inode) do { } while (0) 4332 static inline void f2fs_update_extent_tree_range_compressed(struct inode *inode, 4333 pgoff_t fofs, block_t blkaddr, unsigned int llen, 4334 unsigned int c_len) { } 4335 #endif 4336 4337 static inline int set_compress_context(struct inode *inode) 4338 { 4339 #ifdef CONFIG_F2FS_FS_COMPRESSION 4340 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 4341 4342 F2FS_I(inode)->i_compress_algorithm = 4343 F2FS_OPTION(sbi).compress_algorithm; 4344 F2FS_I(inode)->i_log_cluster_size = 4345 F2FS_OPTION(sbi).compress_log_size; 4346 F2FS_I(inode)->i_compress_flag = 4347 F2FS_OPTION(sbi).compress_chksum ? 4348 1 << COMPRESS_CHKSUM : 0; 4349 F2FS_I(inode)->i_cluster_size = 4350 1 << F2FS_I(inode)->i_log_cluster_size; 4351 if ((F2FS_I(inode)->i_compress_algorithm == COMPRESS_LZ4 || 4352 F2FS_I(inode)->i_compress_algorithm == COMPRESS_ZSTD) && 4353 F2FS_OPTION(sbi).compress_level) 4354 F2FS_I(inode)->i_compress_flag |= 4355 F2FS_OPTION(sbi).compress_level << 4356 COMPRESS_LEVEL_OFFSET; 4357 F2FS_I(inode)->i_flags |= F2FS_COMPR_FL; 4358 set_inode_flag(inode, FI_COMPRESSED_FILE); 4359 stat_inc_compr_inode(inode); 4360 inc_compr_inode_stat(inode); 4361 f2fs_mark_inode_dirty_sync(inode, true); 4362 return 0; 4363 #else 4364 return -EOPNOTSUPP; 4365 #endif 4366 } 4367 4368 static inline bool f2fs_disable_compressed_file(struct inode *inode) 4369 { 4370 struct f2fs_inode_info *fi = F2FS_I(inode); 4371 4372 if (!f2fs_compressed_file(inode)) 4373 return true; 4374 if (S_ISREG(inode->i_mode) && F2FS_HAS_BLOCKS(inode)) 4375 return false; 4376 4377 fi->i_flags &= ~F2FS_COMPR_FL; 4378 stat_dec_compr_inode(inode); 4379 clear_inode_flag(inode, FI_COMPRESSED_FILE); 4380 f2fs_mark_inode_dirty_sync(inode, true); 4381 return true; 4382 } 4383 4384 #define F2FS_FEATURE_FUNCS(name, flagname) \ 4385 static inline int f2fs_sb_has_##name(struct f2fs_sb_info *sbi) \ 4386 { \ 4387 return F2FS_HAS_FEATURE(sbi, F2FS_FEATURE_##flagname); \ 4388 } 4389 4390 F2FS_FEATURE_FUNCS(encrypt, ENCRYPT); 4391 F2FS_FEATURE_FUNCS(blkzoned, BLKZONED); 4392 F2FS_FEATURE_FUNCS(extra_attr, EXTRA_ATTR); 4393 F2FS_FEATURE_FUNCS(project_quota, PRJQUOTA); 4394 F2FS_FEATURE_FUNCS(inode_chksum, INODE_CHKSUM); 4395 F2FS_FEATURE_FUNCS(flexible_inline_xattr, FLEXIBLE_INLINE_XATTR); 4396 F2FS_FEATURE_FUNCS(quota_ino, QUOTA_INO); 4397 F2FS_FEATURE_FUNCS(inode_crtime, INODE_CRTIME); 4398 F2FS_FEATURE_FUNCS(lost_found, LOST_FOUND); 4399 F2FS_FEATURE_FUNCS(verity, VERITY); 4400 F2FS_FEATURE_FUNCS(sb_chksum, SB_CHKSUM); 4401 F2FS_FEATURE_FUNCS(casefold, CASEFOLD); 4402 F2FS_FEATURE_FUNCS(compression, COMPRESSION); 4403 F2FS_FEATURE_FUNCS(readonly, RO); 4404 4405 static inline bool f2fs_may_extent_tree(struct inode *inode) 4406 { 4407 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 4408 4409 if (!test_opt(sbi, EXTENT_CACHE) || 4410 is_inode_flag_set(inode, FI_NO_EXTENT) || 4411 (is_inode_flag_set(inode, FI_COMPRESSED_FILE) && 4412 !f2fs_sb_has_readonly(sbi))) 4413 return false; 4414 4415 /* 4416 * for recovered files during mount do not create extents 4417 * if shrinker is not registered. 4418 */ 4419 if (list_empty(&sbi->s_list)) 4420 return false; 4421 4422 return S_ISREG(inode->i_mode); 4423 } 4424 4425 #ifdef CONFIG_BLK_DEV_ZONED 4426 static inline bool f2fs_blkz_is_seq(struct f2fs_sb_info *sbi, int devi, 4427 block_t blkaddr) 4428 { 4429 unsigned int zno = blkaddr >> sbi->log_blocks_per_blkz; 4430 4431 return test_bit(zno, FDEV(devi).blkz_seq); 4432 } 4433 #endif 4434 4435 static inline bool f2fs_hw_should_discard(struct f2fs_sb_info *sbi) 4436 { 4437 return f2fs_sb_has_blkzoned(sbi); 4438 } 4439 4440 static inline bool f2fs_bdev_support_discard(struct block_device *bdev) 4441 { 4442 return bdev_max_discard_sectors(bdev) || bdev_is_zoned(bdev); 4443 } 4444 4445 static inline bool f2fs_hw_support_discard(struct f2fs_sb_info *sbi) 4446 { 4447 int i; 4448 4449 if (!f2fs_is_multi_device(sbi)) 4450 return f2fs_bdev_support_discard(sbi->sb->s_bdev); 4451 4452 for (i = 0; i < sbi->s_ndevs; i++) 4453 if (f2fs_bdev_support_discard(FDEV(i).bdev)) 4454 return true; 4455 return false; 4456 } 4457 4458 static inline bool f2fs_realtime_discard_enable(struct f2fs_sb_info *sbi) 4459 { 4460 return (test_opt(sbi, DISCARD) && f2fs_hw_support_discard(sbi)) || 4461 f2fs_hw_should_discard(sbi); 4462 } 4463 4464 static inline bool f2fs_hw_is_readonly(struct f2fs_sb_info *sbi) 4465 { 4466 int i; 4467 4468 if (!f2fs_is_multi_device(sbi)) 4469 return bdev_read_only(sbi->sb->s_bdev); 4470 4471 for (i = 0; i < sbi->s_ndevs; i++) 4472 if (bdev_read_only(FDEV(i).bdev)) 4473 return true; 4474 return false; 4475 } 4476 4477 static inline bool f2fs_lfs_mode(struct f2fs_sb_info *sbi) 4478 { 4479 return F2FS_OPTION(sbi).fs_mode == FS_MODE_LFS; 4480 } 4481 4482 static inline bool f2fs_low_mem_mode(struct f2fs_sb_info *sbi) 4483 { 4484 return F2FS_OPTION(sbi).memory_mode == MEMORY_MODE_LOW; 4485 } 4486 4487 static inline bool f2fs_may_compress(struct inode *inode) 4488 { 4489 if (IS_SWAPFILE(inode) || f2fs_is_pinned_file(inode) || 4490 f2fs_is_atomic_file(inode) || f2fs_has_inline_data(inode)) 4491 return false; 4492 return S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode); 4493 } 4494 4495 static inline void f2fs_i_compr_blocks_update(struct inode *inode, 4496 u64 blocks, bool add) 4497 { 4498 struct f2fs_inode_info *fi = F2FS_I(inode); 4499 int diff = fi->i_cluster_size - blocks; 4500 4501 /* don't update i_compr_blocks if saved blocks were released */ 4502 if (!add && !atomic_read(&fi->i_compr_blocks)) 4503 return; 4504 4505 if (add) { 4506 atomic_add(diff, &fi->i_compr_blocks); 4507 stat_add_compr_blocks(inode, diff); 4508 } else { 4509 atomic_sub(diff, &fi->i_compr_blocks); 4510 stat_sub_compr_blocks(inode, diff); 4511 } 4512 f2fs_mark_inode_dirty_sync(inode, true); 4513 } 4514 4515 static inline bool f2fs_allow_multi_device_dio(struct f2fs_sb_info *sbi, 4516 int flag) 4517 { 4518 if (!f2fs_is_multi_device(sbi)) 4519 return false; 4520 if (flag != F2FS_GET_BLOCK_DIO) 4521 return false; 4522 return sbi->aligned_blksize; 4523 } 4524 4525 static inline bool f2fs_need_verity(const struct inode *inode, pgoff_t idx) 4526 { 4527 return fsverity_active(inode) && 4528 idx < DIV_ROUND_UP(inode->i_size, PAGE_SIZE); 4529 } 4530 4531 #ifdef CONFIG_F2FS_FAULT_INJECTION 4532 extern void f2fs_build_fault_attr(struct f2fs_sb_info *sbi, unsigned int rate, 4533 unsigned int type); 4534 #else 4535 #define f2fs_build_fault_attr(sbi, rate, type) do { } while (0) 4536 #endif 4537 4538 static inline bool is_journalled_quota(struct f2fs_sb_info *sbi) 4539 { 4540 #ifdef CONFIG_QUOTA 4541 if (f2fs_sb_has_quota_ino(sbi)) 4542 return true; 4543 if (F2FS_OPTION(sbi).s_qf_names[USRQUOTA] || 4544 F2FS_OPTION(sbi).s_qf_names[GRPQUOTA] || 4545 F2FS_OPTION(sbi).s_qf_names[PRJQUOTA]) 4546 return true; 4547 #endif 4548 return false; 4549 } 4550 4551 static inline bool f2fs_block_unit_discard(struct f2fs_sb_info *sbi) 4552 { 4553 return F2FS_OPTION(sbi).discard_unit == DISCARD_UNIT_BLOCK; 4554 } 4555 4556 static inline void f2fs_io_schedule_timeout(long timeout) 4557 { 4558 set_current_state(TASK_UNINTERRUPTIBLE); 4559 io_schedule_timeout(timeout); 4560 } 4561 4562 static inline void f2fs_handle_page_eio(struct f2fs_sb_info *sbi, pgoff_t ofs, 4563 enum page_type type) 4564 { 4565 if (unlikely(f2fs_cp_error(sbi))) 4566 return; 4567 4568 if (ofs == sbi->page_eio_ofs[type]) { 4569 if (sbi->page_eio_cnt[type]++ == MAX_RETRY_PAGE_EIO) 4570 set_ckpt_flags(sbi, CP_ERROR_FLAG); 4571 } else { 4572 sbi->page_eio_ofs[type] = ofs; 4573 sbi->page_eio_cnt[type] = 0; 4574 } 4575 } 4576 4577 #define EFSBADCRC EBADMSG /* Bad CRC detected */ 4578 #define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */ 4579 4580 #endif /* _LINUX_F2FS_H */ 4581