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 (GENMASK(FAULT_MAX - 1, 0)) 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 & BIT(type)) 78 #endif 79 80 /* 81 * For mount options 82 */ 83 #define F2FS_MOUNT_DISABLE_ROLL_FORWARD 0x00000001 84 #define F2FS_MOUNT_DISCARD 0x00000002 85 #define F2FS_MOUNT_NOHEAP 0x00000004 86 #define F2FS_MOUNT_XATTR_USER 0x00000008 87 #define F2FS_MOUNT_POSIX_ACL 0x00000010 88 #define F2FS_MOUNT_DISABLE_EXT_IDENTIFY 0x00000020 89 #define F2FS_MOUNT_INLINE_XATTR 0x00000040 90 #define F2FS_MOUNT_INLINE_DATA 0x00000080 91 #define F2FS_MOUNT_INLINE_DENTRY 0x00000100 92 #define F2FS_MOUNT_FLUSH_MERGE 0x00000200 93 #define F2FS_MOUNT_NOBARRIER 0x00000400 94 #define F2FS_MOUNT_FASTBOOT 0x00000800 95 #define F2FS_MOUNT_READ_EXTENT_CACHE 0x00001000 96 #define F2FS_MOUNT_DATA_FLUSH 0x00002000 97 #define F2FS_MOUNT_FAULT_INJECTION 0x00004000 98 #define F2FS_MOUNT_USRQUOTA 0x00008000 99 #define F2FS_MOUNT_GRPQUOTA 0x00010000 100 #define F2FS_MOUNT_PRJQUOTA 0x00020000 101 #define F2FS_MOUNT_QUOTA 0x00040000 102 #define F2FS_MOUNT_INLINE_XATTR_SIZE 0x00080000 103 #define F2FS_MOUNT_RESERVE_ROOT 0x00100000 104 #define F2FS_MOUNT_DISABLE_CHECKPOINT 0x00200000 105 #define F2FS_MOUNT_NORECOVERY 0x00400000 106 #define F2FS_MOUNT_ATGC 0x00800000 107 #define F2FS_MOUNT_MERGE_CHECKPOINT 0x01000000 108 #define F2FS_MOUNT_GC_MERGE 0x02000000 109 #define F2FS_MOUNT_COMPRESS_CACHE 0x04000000 110 #define F2FS_MOUNT_AGE_EXTENT_CACHE 0x08000000 111 112 #define F2FS_OPTION(sbi) ((sbi)->mount_opt) 113 #define clear_opt(sbi, option) (F2FS_OPTION(sbi).opt &= ~F2FS_MOUNT_##option) 114 #define set_opt(sbi, option) (F2FS_OPTION(sbi).opt |= F2FS_MOUNT_##option) 115 #define test_opt(sbi, option) (F2FS_OPTION(sbi).opt & F2FS_MOUNT_##option) 116 117 #define ver_after(a, b) (typecheck(unsigned long long, a) && \ 118 typecheck(unsigned long long, b) && \ 119 ((long long)((a) - (b)) > 0)) 120 121 typedef u32 block_t; /* 122 * should not change u32, since it is the on-disk block 123 * address format, __le32. 124 */ 125 typedef u32 nid_t; 126 127 #define COMPRESS_EXT_NUM 16 128 129 /* 130 * An implementation of an rwsem that is explicitly unfair to readers. This 131 * prevents priority inversion when a low-priority reader acquires the read lock 132 * while sleeping on the write lock but the write lock is needed by 133 * higher-priority clients. 134 */ 135 136 struct f2fs_rwsem { 137 struct rw_semaphore internal_rwsem; 138 #ifdef CONFIG_F2FS_UNFAIR_RWSEM 139 wait_queue_head_t read_waiters; 140 #endif 141 }; 142 143 struct f2fs_mount_info { 144 unsigned int opt; 145 int write_io_size_bits; /* Write IO size bits */ 146 block_t root_reserved_blocks; /* root reserved blocks */ 147 kuid_t s_resuid; /* reserved blocks for uid */ 148 kgid_t s_resgid; /* reserved blocks for gid */ 149 int active_logs; /* # of active logs */ 150 int inline_xattr_size; /* inline xattr size */ 151 #ifdef CONFIG_F2FS_FAULT_INJECTION 152 struct f2fs_fault_info fault_info; /* For fault injection */ 153 #endif 154 #ifdef CONFIG_QUOTA 155 /* Names of quota files with journalled quota */ 156 char *s_qf_names[MAXQUOTAS]; 157 int s_jquota_fmt; /* Format of quota to use */ 158 #endif 159 /* For which write hints are passed down to block layer */ 160 int alloc_mode; /* segment allocation policy */ 161 int fsync_mode; /* fsync policy */ 162 int fs_mode; /* fs mode: LFS or ADAPTIVE */ 163 int bggc_mode; /* bggc mode: off, on or sync */ 164 int memory_mode; /* memory mode */ 165 int errors; /* errors parameter */ 166 int discard_unit; /* 167 * discard command's offset/size should 168 * be aligned to this unit: block, 169 * segment or section 170 */ 171 struct fscrypt_dummy_policy dummy_enc_policy; /* test dummy encryption */ 172 block_t unusable_cap_perc; /* percentage for cap */ 173 block_t unusable_cap; /* Amount of space allowed to be 174 * unusable when disabling checkpoint 175 */ 176 177 /* For compression */ 178 unsigned char compress_algorithm; /* algorithm type */ 179 unsigned char compress_log_size; /* cluster log size */ 180 unsigned char compress_level; /* compress level */ 181 bool compress_chksum; /* compressed data chksum */ 182 unsigned char compress_ext_cnt; /* extension count */ 183 unsigned char nocompress_ext_cnt; /* nocompress extension count */ 184 int compress_mode; /* compression mode */ 185 unsigned char extensions[COMPRESS_EXT_NUM][F2FS_EXTENSION_LEN]; /* extensions */ 186 unsigned char noextensions[COMPRESS_EXT_NUM][F2FS_EXTENSION_LEN]; /* extensions */ 187 }; 188 189 #define F2FS_FEATURE_ENCRYPT 0x00000001 190 #define F2FS_FEATURE_BLKZONED 0x00000002 191 #define F2FS_FEATURE_ATOMIC_WRITE 0x00000004 192 #define F2FS_FEATURE_EXTRA_ATTR 0x00000008 193 #define F2FS_FEATURE_PRJQUOTA 0x00000010 194 #define F2FS_FEATURE_INODE_CHKSUM 0x00000020 195 #define F2FS_FEATURE_FLEXIBLE_INLINE_XATTR 0x00000040 196 #define F2FS_FEATURE_QUOTA_INO 0x00000080 197 #define F2FS_FEATURE_INODE_CRTIME 0x00000100 198 #define F2FS_FEATURE_LOST_FOUND 0x00000200 199 #define F2FS_FEATURE_VERITY 0x00000400 200 #define F2FS_FEATURE_SB_CHKSUM 0x00000800 201 #define F2FS_FEATURE_CASEFOLD 0x00001000 202 #define F2FS_FEATURE_COMPRESSION 0x00002000 203 #define F2FS_FEATURE_RO 0x00004000 204 205 #define __F2FS_HAS_FEATURE(raw_super, mask) \ 206 ((raw_super->feature & cpu_to_le32(mask)) != 0) 207 #define F2FS_HAS_FEATURE(sbi, mask) __F2FS_HAS_FEATURE(sbi->raw_super, mask) 208 209 /* 210 * Default values for user and/or group using reserved blocks 211 */ 212 #define F2FS_DEF_RESUID 0 213 #define F2FS_DEF_RESGID 0 214 215 /* 216 * For checkpoint manager 217 */ 218 enum { 219 NAT_BITMAP, 220 SIT_BITMAP 221 }; 222 223 #define CP_UMOUNT 0x00000001 224 #define CP_FASTBOOT 0x00000002 225 #define CP_SYNC 0x00000004 226 #define CP_RECOVERY 0x00000008 227 #define CP_DISCARD 0x00000010 228 #define CP_TRIMMED 0x00000020 229 #define CP_PAUSE 0x00000040 230 #define CP_RESIZE 0x00000080 231 232 #define DEF_MAX_DISCARD_REQUEST 8 /* issue 8 discards per round */ 233 #define DEF_MIN_DISCARD_ISSUE_TIME 50 /* 50 ms, if exists */ 234 #define DEF_MID_DISCARD_ISSUE_TIME 500 /* 500 ms, if device busy */ 235 #define DEF_MAX_DISCARD_ISSUE_TIME 60000 /* 60 s, if no candidates */ 236 #define DEF_DISCARD_URGENT_UTIL 80 /* do more discard over 80% */ 237 #define DEF_CP_INTERVAL 60 /* 60 secs */ 238 #define DEF_IDLE_INTERVAL 5 /* 5 secs */ 239 #define DEF_DISABLE_INTERVAL 5 /* 5 secs */ 240 #define DEF_DISABLE_QUICK_INTERVAL 1 /* 1 secs */ 241 #define DEF_UMOUNT_DISCARD_TIMEOUT 5 /* 5 secs */ 242 243 struct cp_control { 244 int reason; 245 __u64 trim_start; 246 __u64 trim_end; 247 __u64 trim_minlen; 248 }; 249 250 /* 251 * indicate meta/data type 252 */ 253 enum { 254 META_CP, 255 META_NAT, 256 META_SIT, 257 META_SSA, 258 META_MAX, 259 META_POR, 260 DATA_GENERIC, /* check range only */ 261 DATA_GENERIC_ENHANCE, /* strong check on range and segment bitmap */ 262 DATA_GENERIC_ENHANCE_READ, /* 263 * strong check on range and segment 264 * bitmap but no warning due to race 265 * condition of read on truncated area 266 * by extent_cache 267 */ 268 DATA_GENERIC_ENHANCE_UPDATE, /* 269 * strong check on range and segment 270 * bitmap for update case 271 */ 272 META_GENERIC, 273 }; 274 275 /* for the list of ino */ 276 enum { 277 ORPHAN_INO, /* for orphan ino list */ 278 APPEND_INO, /* for append ino list */ 279 UPDATE_INO, /* for update ino list */ 280 TRANS_DIR_INO, /* for transactions dir ino list */ 281 FLUSH_INO, /* for multiple device flushing */ 282 MAX_INO_ENTRY, /* max. list */ 283 }; 284 285 struct ino_entry { 286 struct list_head list; /* list head */ 287 nid_t ino; /* inode number */ 288 unsigned int dirty_device; /* dirty device bitmap */ 289 }; 290 291 /* for the list of inodes to be GCed */ 292 struct inode_entry { 293 struct list_head list; /* list head */ 294 struct inode *inode; /* vfs inode pointer */ 295 }; 296 297 struct fsync_node_entry { 298 struct list_head list; /* list head */ 299 struct page *page; /* warm node page pointer */ 300 unsigned int seq_id; /* sequence id */ 301 }; 302 303 struct ckpt_req { 304 struct completion wait; /* completion for checkpoint done */ 305 struct llist_node llnode; /* llist_node to be linked in wait queue */ 306 int ret; /* return code of checkpoint */ 307 ktime_t queue_time; /* request queued time */ 308 }; 309 310 struct ckpt_req_control { 311 struct task_struct *f2fs_issue_ckpt; /* checkpoint task */ 312 int ckpt_thread_ioprio; /* checkpoint merge thread ioprio */ 313 wait_queue_head_t ckpt_wait_queue; /* waiting queue for wake-up */ 314 atomic_t issued_ckpt; /* # of actually issued ckpts */ 315 atomic_t total_ckpt; /* # of total ckpts */ 316 atomic_t queued_ckpt; /* # of queued ckpts */ 317 struct llist_head issue_list; /* list for command issue */ 318 spinlock_t stat_lock; /* lock for below checkpoint time stats */ 319 unsigned int cur_time; /* cur wait time in msec for currently issued checkpoint */ 320 unsigned int peak_time; /* peak wait time in msec until now */ 321 }; 322 323 /* for the bitmap indicate blocks to be discarded */ 324 struct discard_entry { 325 struct list_head list; /* list head */ 326 block_t start_blkaddr; /* start blockaddr of current segment */ 327 unsigned char discard_map[SIT_VBLOCK_MAP_SIZE]; /* segment discard bitmap */ 328 }; 329 330 /* minimum discard granularity, unit: block count */ 331 #define MIN_DISCARD_GRANULARITY 1 332 /* 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 struct discard_info di; /* discard info */ 358 struct list_head list; /* command list */ 359 struct completion wait; /* compleation */ 360 struct block_device *bdev; /* bdev */ 361 unsigned short ref; /* reference count */ 362 unsigned char state; /* state */ 363 unsigned char queued; /* queued discard */ 364 int error; /* bio error */ 365 spinlock_t lock; /* for state/bio_ref updating */ 366 unsigned short bio_ref; /* bio reference count */ 367 }; 368 369 enum { 370 DPOLICY_BG, 371 DPOLICY_FORCE, 372 DPOLICY_FSTRIM, 373 DPOLICY_UMOUNT, 374 MAX_DPOLICY, 375 }; 376 377 struct discard_policy { 378 int type; /* type of discard */ 379 unsigned int min_interval; /* used for candidates exist */ 380 unsigned int mid_interval; /* used for device busy */ 381 unsigned int max_interval; /* used for candidates not exist */ 382 unsigned int max_requests; /* # of discards issued per round */ 383 unsigned int io_aware_gran; /* minimum granularity discard not be aware of I/O */ 384 bool io_aware; /* issue discard in idle time */ 385 bool sync; /* submit discard with REQ_SYNC flag */ 386 bool ordered; /* issue discard by lba order */ 387 bool timeout; /* discard timeout for put_super */ 388 unsigned int granularity; /* discard granularity */ 389 }; 390 391 struct discard_cmd_control { 392 struct task_struct *f2fs_issue_discard; /* discard thread */ 393 struct list_head entry_list; /* 4KB discard entry list */ 394 struct list_head pend_list[MAX_PLIST_NUM];/* store pending entries */ 395 struct list_head wait_list; /* store on-flushing entries */ 396 struct list_head fstrim_list; /* in-flight discard from fstrim */ 397 wait_queue_head_t discard_wait_queue; /* waiting queue for wake-up */ 398 struct mutex cmd_lock; 399 unsigned int nr_discards; /* # of discards in the list */ 400 unsigned int max_discards; /* max. discards to be issued */ 401 unsigned int max_discard_request; /* max. discard request per round */ 402 unsigned int min_discard_issue_time; /* min. interval between discard issue */ 403 unsigned int mid_discard_issue_time; /* mid. interval between discard issue */ 404 unsigned int max_discard_issue_time; /* max. interval between discard issue */ 405 unsigned int discard_io_aware_gran; /* minimum discard granularity not be aware of I/O */ 406 unsigned int discard_urgent_util; /* utilization which issue discard proactively */ 407 unsigned int discard_granularity; /* discard granularity */ 408 unsigned int max_ordered_discard; /* maximum discard granularity issued by lba order */ 409 unsigned int undiscard_blks; /* # of undiscard blocks */ 410 unsigned int next_pos; /* next discard position */ 411 atomic_t issued_discard; /* # of issued discard */ 412 atomic_t queued_discard; /* # of queued discard */ 413 atomic_t discard_cmd_cnt; /* # of cached cmd count */ 414 struct rb_root_cached root; /* root of discard rb-tree */ 415 bool rbtree_check; /* config for consistence check */ 416 bool discard_wake; /* to wake up discard thread */ 417 }; 418 419 /* for the list of fsync inodes, used only during recovery */ 420 struct fsync_inode_entry { 421 struct list_head list; /* list head */ 422 struct inode *inode; /* vfs inode pointer */ 423 block_t blkaddr; /* block address locating the last fsync */ 424 block_t last_dentry; /* block address locating the last dentry */ 425 }; 426 427 #define nats_in_cursum(jnl) (le16_to_cpu((jnl)->n_nats)) 428 #define sits_in_cursum(jnl) (le16_to_cpu((jnl)->n_sits)) 429 430 #define nat_in_journal(jnl, i) ((jnl)->nat_j.entries[i].ne) 431 #define nid_in_journal(jnl, i) ((jnl)->nat_j.entries[i].nid) 432 #define sit_in_journal(jnl, i) ((jnl)->sit_j.entries[i].se) 433 #define segno_in_journal(jnl, i) ((jnl)->sit_j.entries[i].segno) 434 435 #define MAX_NAT_JENTRIES(jnl) (NAT_JOURNAL_ENTRIES - nats_in_cursum(jnl)) 436 #define MAX_SIT_JENTRIES(jnl) (SIT_JOURNAL_ENTRIES - sits_in_cursum(jnl)) 437 438 static inline int update_nats_in_cursum(struct f2fs_journal *journal, int i) 439 { 440 int before = nats_in_cursum(journal); 441 442 journal->n_nats = cpu_to_le16(before + i); 443 return before; 444 } 445 446 static inline int update_sits_in_cursum(struct f2fs_journal *journal, int i) 447 { 448 int before = sits_in_cursum(journal); 449 450 journal->n_sits = cpu_to_le16(before + i); 451 return before; 452 } 453 454 static inline bool __has_cursum_space(struct f2fs_journal *journal, 455 int size, int type) 456 { 457 if (type == NAT_JOURNAL) 458 return size <= MAX_NAT_JENTRIES(journal); 459 return size <= MAX_SIT_JENTRIES(journal); 460 } 461 462 /* for inline stuff */ 463 #define DEF_INLINE_RESERVED_SIZE 1 464 static inline int get_extra_isize(struct inode *inode); 465 static inline int get_inline_xattr_addrs(struct inode *inode); 466 #define MAX_INLINE_DATA(inode) (sizeof(__le32) * \ 467 (CUR_ADDRS_PER_INODE(inode) - \ 468 get_inline_xattr_addrs(inode) - \ 469 DEF_INLINE_RESERVED_SIZE)) 470 471 /* for inline dir */ 472 #define NR_INLINE_DENTRY(inode) (MAX_INLINE_DATA(inode) * BITS_PER_BYTE / \ 473 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \ 474 BITS_PER_BYTE + 1)) 475 #define INLINE_DENTRY_BITMAP_SIZE(inode) \ 476 DIV_ROUND_UP(NR_INLINE_DENTRY(inode), BITS_PER_BYTE) 477 #define INLINE_RESERVED_SIZE(inode) (MAX_INLINE_DATA(inode) - \ 478 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \ 479 NR_INLINE_DENTRY(inode) + \ 480 INLINE_DENTRY_BITMAP_SIZE(inode))) 481 482 /* 483 * For INODE and NODE manager 484 */ 485 /* for directory operations */ 486 487 struct f2fs_filename { 488 /* 489 * The filename the user specified. This is NULL for some 490 * filesystem-internal operations, e.g. converting an inline directory 491 * to a non-inline one, or roll-forward recovering an encrypted dentry. 492 */ 493 const struct qstr *usr_fname; 494 495 /* 496 * The on-disk filename. For encrypted directories, this is encrypted. 497 * This may be NULL for lookups in an encrypted dir without the key. 498 */ 499 struct fscrypt_str disk_name; 500 501 /* The dirhash of this filename */ 502 f2fs_hash_t hash; 503 504 #ifdef CONFIG_FS_ENCRYPTION 505 /* 506 * For lookups in encrypted directories: either the buffer backing 507 * disk_name, or a buffer that holds the decoded no-key name. 508 */ 509 struct fscrypt_str crypto_buf; 510 #endif 511 #if IS_ENABLED(CONFIG_UNICODE) 512 /* 513 * For casefolded directories: the casefolded name, but it's left NULL 514 * if the original name is not valid Unicode, if the original name is 515 * "." or "..", if the directory is both casefolded and encrypted and 516 * its encryption key is unavailable, or if the filesystem is doing an 517 * internal operation where usr_fname is also NULL. In all these cases 518 * we fall back to treating the name as an opaque byte sequence. 519 */ 520 struct fscrypt_str cf_name; 521 #endif 522 }; 523 524 struct f2fs_dentry_ptr { 525 struct inode *inode; 526 void *bitmap; 527 struct f2fs_dir_entry *dentry; 528 __u8 (*filename)[F2FS_SLOT_LEN]; 529 int max; 530 int nr_bitmap; 531 }; 532 533 static inline void make_dentry_ptr_block(struct inode *inode, 534 struct f2fs_dentry_ptr *d, struct f2fs_dentry_block *t) 535 { 536 d->inode = inode; 537 d->max = NR_DENTRY_IN_BLOCK; 538 d->nr_bitmap = SIZE_OF_DENTRY_BITMAP; 539 d->bitmap = t->dentry_bitmap; 540 d->dentry = t->dentry; 541 d->filename = t->filename; 542 } 543 544 static inline void make_dentry_ptr_inline(struct inode *inode, 545 struct f2fs_dentry_ptr *d, void *t) 546 { 547 int entry_cnt = NR_INLINE_DENTRY(inode); 548 int bitmap_size = INLINE_DENTRY_BITMAP_SIZE(inode); 549 int reserved_size = INLINE_RESERVED_SIZE(inode); 550 551 d->inode = inode; 552 d->max = entry_cnt; 553 d->nr_bitmap = bitmap_size; 554 d->bitmap = t; 555 d->dentry = t + bitmap_size + reserved_size; 556 d->filename = t + bitmap_size + reserved_size + 557 SIZE_OF_DIR_ENTRY * entry_cnt; 558 } 559 560 /* 561 * XATTR_NODE_OFFSET stores xattrs to one node block per file keeping -1 562 * as its node offset to distinguish from index node blocks. 563 * But some bits are used to mark the node block. 564 */ 565 #define XATTR_NODE_OFFSET ((((unsigned int)-1) << OFFSET_BIT_SHIFT) \ 566 >> OFFSET_BIT_SHIFT) 567 enum { 568 ALLOC_NODE, /* allocate a new node page if needed */ 569 LOOKUP_NODE, /* look up a node without readahead */ 570 LOOKUP_NODE_RA, /* 571 * look up a node with readahead called 572 * by get_data_block. 573 */ 574 }; 575 576 #define DEFAULT_RETRY_IO_COUNT 8 /* maximum retry read IO or flush count */ 577 578 /* congestion wait timeout value, default: 20ms */ 579 #define DEFAULT_IO_TIMEOUT (msecs_to_jiffies(20)) 580 581 /* maximum retry quota flush count */ 582 #define DEFAULT_RETRY_QUOTA_FLUSH_COUNT 8 583 584 /* maximum retry of EIO'ed page */ 585 #define MAX_RETRY_PAGE_EIO 100 586 587 #define F2FS_LINK_MAX 0xffffffff /* maximum link count per file */ 588 589 #define MAX_DIR_RA_PAGES 4 /* maximum ra pages of dir */ 590 591 /* dirty segments threshold for triggering CP */ 592 #define DEFAULT_DIRTY_THRESHOLD 4 593 594 #define RECOVERY_MAX_RA_BLOCKS BIO_MAX_VECS 595 #define RECOVERY_MIN_RA_BLOCKS 1 596 597 #define F2FS_ONSTACK_PAGES 16 /* nr of onstack pages */ 598 599 /* for in-memory extent cache entry */ 600 #define F2FS_MIN_EXTENT_LEN 64 /* minimum extent length */ 601 602 /* number of extent info in extent cache we try to shrink */ 603 #define READ_EXTENT_CACHE_SHRINK_NUMBER 128 604 605 /* number of age extent info in extent cache we try to shrink */ 606 #define AGE_EXTENT_CACHE_SHRINK_NUMBER 128 607 #define LAST_AGE_WEIGHT 30 608 #define SAME_AGE_REGION 1024 609 610 /* 611 * Define data block with age less than 1GB as hot data 612 * define data block with age less than 10GB but more than 1GB as warm data 613 */ 614 #define DEF_HOT_DATA_AGE_THRESHOLD 262144 615 #define DEF_WARM_DATA_AGE_THRESHOLD 2621440 616 617 /* extent cache type */ 618 enum extent_type { 619 EX_READ, 620 EX_BLOCK_AGE, 621 NR_EXTENT_CACHES, 622 }; 623 624 struct extent_info { 625 unsigned int fofs; /* start offset in a file */ 626 unsigned int len; /* length of the extent */ 627 union { 628 /* read extent_cache */ 629 struct { 630 /* start block address of the extent */ 631 block_t blk; 632 #ifdef CONFIG_F2FS_FS_COMPRESSION 633 /* physical extent length of compressed blocks */ 634 unsigned int c_len; 635 #endif 636 }; 637 /* block age extent_cache */ 638 struct { 639 /* block age of the extent */ 640 unsigned long long age; 641 /* last total blocks allocated */ 642 unsigned long long last_blocks; 643 }; 644 }; 645 }; 646 647 struct extent_node { 648 struct rb_node rb_node; /* rb node located in rb-tree */ 649 struct extent_info ei; /* extent info */ 650 struct list_head list; /* node in global extent list of sbi */ 651 struct extent_tree *et; /* extent tree pointer */ 652 }; 653 654 struct extent_tree { 655 nid_t ino; /* inode number */ 656 enum extent_type type; /* keep the extent tree type */ 657 struct rb_root_cached root; /* root of extent info rb-tree */ 658 struct extent_node *cached_en; /* recently accessed extent node */ 659 struct list_head list; /* to be used by sbi->zombie_list */ 660 rwlock_t lock; /* protect extent info rb-tree */ 661 atomic_t node_cnt; /* # of extent node in rb-tree*/ 662 bool largest_updated; /* largest extent updated */ 663 struct extent_info largest; /* largest cached extent for EX_READ */ 664 }; 665 666 struct extent_tree_info { 667 struct radix_tree_root extent_tree_root;/* cache extent cache entries */ 668 struct mutex extent_tree_lock; /* locking extent radix tree */ 669 struct list_head extent_list; /* lru list for shrinker */ 670 spinlock_t extent_lock; /* locking extent lru list */ 671 atomic_t total_ext_tree; /* extent tree count */ 672 struct list_head zombie_list; /* extent zombie tree list */ 673 atomic_t total_zombie_tree; /* extent zombie tree count */ 674 atomic_t total_ext_node; /* extent info count */ 675 }; 676 677 /* 678 * State of block returned by f2fs_map_blocks. 679 */ 680 #define F2FS_MAP_NEW (1U << 0) 681 #define F2FS_MAP_MAPPED (1U << 1) 682 #define F2FS_MAP_DELALLOC (1U << 2) 683 #define F2FS_MAP_FLAGS (F2FS_MAP_NEW | F2FS_MAP_MAPPED |\ 684 F2FS_MAP_DELALLOC) 685 686 struct f2fs_map_blocks { 687 struct block_device *m_bdev; /* for multi-device dio */ 688 block_t m_pblk; 689 block_t m_lblk; 690 unsigned int m_len; 691 unsigned int m_flags; 692 pgoff_t *m_next_pgofs; /* point next possible non-hole pgofs */ 693 pgoff_t *m_next_extent; /* point to next possible extent */ 694 int m_seg_type; 695 bool m_may_create; /* indicate it is from write path */ 696 bool m_multidev_dio; /* indicate it allows multi-device dio */ 697 }; 698 699 /* for flag in get_data_block */ 700 enum { 701 F2FS_GET_BLOCK_DEFAULT, 702 F2FS_GET_BLOCK_FIEMAP, 703 F2FS_GET_BLOCK_BMAP, 704 F2FS_GET_BLOCK_DIO, 705 F2FS_GET_BLOCK_PRE_DIO, 706 F2FS_GET_BLOCK_PRE_AIO, 707 F2FS_GET_BLOCK_PRECACHE, 708 }; 709 710 /* 711 * i_advise uses FADVISE_XXX_BIT. We can add additional hints later. 712 */ 713 #define FADVISE_COLD_BIT 0x01 714 #define FADVISE_LOST_PINO_BIT 0x02 715 #define FADVISE_ENCRYPT_BIT 0x04 716 #define FADVISE_ENC_NAME_BIT 0x08 717 #define FADVISE_KEEP_SIZE_BIT 0x10 718 #define FADVISE_HOT_BIT 0x20 719 #define FADVISE_VERITY_BIT 0x40 720 #define FADVISE_TRUNC_BIT 0x80 721 722 #define FADVISE_MODIFIABLE_BITS (FADVISE_COLD_BIT | FADVISE_HOT_BIT) 723 724 #define file_is_cold(inode) is_file(inode, FADVISE_COLD_BIT) 725 #define file_set_cold(inode) set_file(inode, FADVISE_COLD_BIT) 726 #define file_clear_cold(inode) clear_file(inode, FADVISE_COLD_BIT) 727 728 #define file_wrong_pino(inode) is_file(inode, FADVISE_LOST_PINO_BIT) 729 #define file_lost_pino(inode) set_file(inode, FADVISE_LOST_PINO_BIT) 730 #define file_got_pino(inode) clear_file(inode, FADVISE_LOST_PINO_BIT) 731 732 #define file_is_encrypt(inode) is_file(inode, FADVISE_ENCRYPT_BIT) 733 #define file_set_encrypt(inode) set_file(inode, FADVISE_ENCRYPT_BIT) 734 735 #define file_enc_name(inode) is_file(inode, FADVISE_ENC_NAME_BIT) 736 #define file_set_enc_name(inode) set_file(inode, FADVISE_ENC_NAME_BIT) 737 738 #define file_keep_isize(inode) is_file(inode, FADVISE_KEEP_SIZE_BIT) 739 #define file_set_keep_isize(inode) set_file(inode, FADVISE_KEEP_SIZE_BIT) 740 741 #define file_is_hot(inode) is_file(inode, FADVISE_HOT_BIT) 742 #define file_set_hot(inode) set_file(inode, FADVISE_HOT_BIT) 743 #define file_clear_hot(inode) clear_file(inode, FADVISE_HOT_BIT) 744 745 #define file_is_verity(inode) is_file(inode, FADVISE_VERITY_BIT) 746 #define file_set_verity(inode) set_file(inode, FADVISE_VERITY_BIT) 747 748 #define file_should_truncate(inode) is_file(inode, FADVISE_TRUNC_BIT) 749 #define file_need_truncate(inode) set_file(inode, FADVISE_TRUNC_BIT) 750 #define file_dont_truncate(inode) clear_file(inode, FADVISE_TRUNC_BIT) 751 752 #define DEF_DIR_LEVEL 0 753 754 enum { 755 GC_FAILURE_PIN, 756 MAX_GC_FAILURE 757 }; 758 759 /* used for f2fs_inode_info->flags */ 760 enum { 761 FI_NEW_INODE, /* indicate newly allocated inode */ 762 FI_DIRTY_INODE, /* indicate inode is dirty or not */ 763 FI_AUTO_RECOVER, /* indicate inode is recoverable */ 764 FI_DIRTY_DIR, /* indicate directory has dirty pages */ 765 FI_INC_LINK, /* need to increment i_nlink */ 766 FI_ACL_MODE, /* indicate acl mode */ 767 FI_NO_ALLOC, /* should not allocate any blocks */ 768 FI_FREE_NID, /* free allocated nide */ 769 FI_NO_EXTENT, /* not to use the extent cache */ 770 FI_INLINE_XATTR, /* used for inline xattr */ 771 FI_INLINE_DATA, /* used for inline data*/ 772 FI_INLINE_DENTRY, /* used for inline dentry */ 773 FI_APPEND_WRITE, /* inode has appended data */ 774 FI_UPDATE_WRITE, /* inode has in-place-update data */ 775 FI_NEED_IPU, /* used for ipu per file */ 776 FI_ATOMIC_FILE, /* indicate atomic file */ 777 FI_FIRST_BLOCK_WRITTEN, /* indicate #0 data block was written */ 778 FI_DROP_CACHE, /* drop dirty page cache */ 779 FI_DATA_EXIST, /* indicate data exists */ 780 FI_INLINE_DOTS, /* indicate inline dot dentries */ 781 FI_SKIP_WRITES, /* should skip data page writeback */ 782 FI_OPU_WRITE, /* used for opu per file */ 783 FI_DIRTY_FILE, /* indicate regular/symlink has dirty pages */ 784 FI_PREALLOCATED_ALL, /* all blocks for write were preallocated */ 785 FI_HOT_DATA, /* indicate file is hot */ 786 FI_EXTRA_ATTR, /* indicate file has extra attribute */ 787 FI_PROJ_INHERIT, /* indicate file inherits projectid */ 788 FI_PIN_FILE, /* indicate file should not be gced */ 789 FI_VERITY_IN_PROGRESS, /* building fs-verity Merkle tree */ 790 FI_COMPRESSED_FILE, /* indicate file's data can be compressed */ 791 FI_COMPRESS_CORRUPT, /* indicate compressed cluster is corrupted */ 792 FI_MMAP_FILE, /* indicate file was mmapped */ 793 FI_ENABLE_COMPRESS, /* enable compression in "user" compression mode */ 794 FI_COMPRESS_RELEASED, /* compressed blocks were released */ 795 FI_ALIGNED_WRITE, /* enable aligned write */ 796 FI_COW_FILE, /* indicate COW file */ 797 FI_ATOMIC_COMMITTED, /* indicate atomic commit completed except disk sync */ 798 FI_ATOMIC_REPLACE, /* indicate atomic replace */ 799 FI_MAX, /* max flag, never be used */ 800 }; 801 802 struct f2fs_inode_info { 803 struct inode vfs_inode; /* serve a vfs inode */ 804 unsigned long i_flags; /* keep an inode flags for ioctl */ 805 unsigned char i_advise; /* use to give file attribute hints */ 806 unsigned char i_dir_level; /* use for dentry level for large dir */ 807 unsigned int i_current_depth; /* only for directory depth */ 808 /* for gc failure statistic */ 809 unsigned int i_gc_failures[MAX_GC_FAILURE]; 810 unsigned int i_pino; /* parent inode number */ 811 umode_t i_acl_mode; /* keep file acl mode temporarily */ 812 813 /* Use below internally in f2fs*/ 814 unsigned long flags[BITS_TO_LONGS(FI_MAX)]; /* use to pass per-file flags */ 815 struct f2fs_rwsem i_sem; /* protect fi info */ 816 atomic_t dirty_pages; /* # of dirty pages */ 817 f2fs_hash_t chash; /* hash value of given file name */ 818 unsigned int clevel; /* maximum level of given file name */ 819 struct task_struct *task; /* lookup and create consistency */ 820 struct task_struct *cp_task; /* separate cp/wb IO stats*/ 821 struct task_struct *wb_task; /* indicate inode is in context of writeback */ 822 nid_t i_xattr_nid; /* node id that contains xattrs */ 823 loff_t last_disk_size; /* lastly written file size */ 824 spinlock_t i_size_lock; /* protect last_disk_size */ 825 826 #ifdef CONFIG_QUOTA 827 struct dquot *i_dquot[MAXQUOTAS]; 828 829 /* quota space reservation, managed internally by quota code */ 830 qsize_t i_reserved_quota; 831 #endif 832 struct list_head dirty_list; /* dirty list for dirs and files */ 833 struct list_head gdirty_list; /* linked in global dirty list */ 834 struct task_struct *atomic_write_task; /* store atomic write task */ 835 struct extent_tree *extent_tree[NR_EXTENT_CACHES]; 836 /* cached extent_tree entry */ 837 struct inode *cow_inode; /* copy-on-write inode for atomic write */ 838 839 /* avoid racing between foreground op and gc */ 840 struct f2fs_rwsem i_gc_rwsem[2]; 841 struct f2fs_rwsem i_xattr_sem; /* avoid racing between reading and changing EAs */ 842 843 int i_extra_isize; /* size of extra space located in i_addr */ 844 kprojid_t i_projid; /* id for project quota */ 845 int i_inline_xattr_size; /* inline xattr size */ 846 struct timespec64 i_crtime; /* inode creation time */ 847 struct timespec64 i_disk_time[3];/* inode disk times */ 848 849 /* for file compress */ 850 atomic_t i_compr_blocks; /* # of compressed blocks */ 851 unsigned char i_compress_algorithm; /* algorithm type */ 852 unsigned char i_log_cluster_size; /* log of cluster size */ 853 unsigned char i_compress_level; /* compress level (lz4hc,zstd) */ 854 unsigned char i_compress_flag; /* compress flag */ 855 unsigned int i_cluster_size; /* cluster size */ 856 857 unsigned int atomic_write_cnt; 858 loff_t original_i_size; /* original i_size before atomic write */ 859 }; 860 861 static inline void get_read_extent_info(struct extent_info *ext, 862 struct f2fs_extent *i_ext) 863 { 864 ext->fofs = le32_to_cpu(i_ext->fofs); 865 ext->blk = le32_to_cpu(i_ext->blk); 866 ext->len = le32_to_cpu(i_ext->len); 867 } 868 869 static inline void set_raw_read_extent(struct extent_info *ext, 870 struct f2fs_extent *i_ext) 871 { 872 i_ext->fofs = cpu_to_le32(ext->fofs); 873 i_ext->blk = cpu_to_le32(ext->blk); 874 i_ext->len = cpu_to_le32(ext->len); 875 } 876 877 static inline bool __is_discard_mergeable(struct discard_info *back, 878 struct discard_info *front, unsigned int max_len) 879 { 880 return (back->lstart + back->len == front->lstart) && 881 (back->len + front->len <= max_len); 882 } 883 884 static inline bool __is_discard_back_mergeable(struct discard_info *cur, 885 struct discard_info *back, unsigned int max_len) 886 { 887 return __is_discard_mergeable(back, cur, max_len); 888 } 889 890 static inline bool __is_discard_front_mergeable(struct discard_info *cur, 891 struct discard_info *front, unsigned int max_len) 892 { 893 return __is_discard_mergeable(cur, front, max_len); 894 } 895 896 /* 897 * For free nid management 898 */ 899 enum nid_state { 900 FREE_NID, /* newly added to free nid list */ 901 PREALLOC_NID, /* it is preallocated */ 902 MAX_NID_STATE, 903 }; 904 905 enum nat_state { 906 TOTAL_NAT, 907 DIRTY_NAT, 908 RECLAIMABLE_NAT, 909 MAX_NAT_STATE, 910 }; 911 912 struct f2fs_nm_info { 913 block_t nat_blkaddr; /* base disk address of NAT */ 914 nid_t max_nid; /* maximum possible node ids */ 915 nid_t available_nids; /* # of available node ids */ 916 nid_t next_scan_nid; /* the next nid to be scanned */ 917 nid_t max_rf_node_blocks; /* max # of nodes for recovery */ 918 unsigned int ram_thresh; /* control the memory footprint */ 919 unsigned int ra_nid_pages; /* # of nid pages to be readaheaded */ 920 unsigned int dirty_nats_ratio; /* control dirty nats ratio threshold */ 921 922 /* NAT cache management */ 923 struct radix_tree_root nat_root;/* root of the nat entry cache */ 924 struct radix_tree_root nat_set_root;/* root of the nat set cache */ 925 struct f2fs_rwsem nat_tree_lock; /* protect nat entry tree */ 926 struct list_head nat_entries; /* cached nat entry list (clean) */ 927 spinlock_t nat_list_lock; /* protect clean nat entry list */ 928 unsigned int nat_cnt[MAX_NAT_STATE]; /* the # of cached nat entries */ 929 unsigned int nat_blocks; /* # of nat blocks */ 930 931 /* free node ids management */ 932 struct radix_tree_root free_nid_root;/* root of the free_nid cache */ 933 struct list_head free_nid_list; /* list for free nids excluding preallocated nids */ 934 unsigned int nid_cnt[MAX_NID_STATE]; /* the number of free node id */ 935 spinlock_t nid_list_lock; /* protect nid lists ops */ 936 struct mutex build_lock; /* lock for build free nids */ 937 unsigned char **free_nid_bitmap; 938 unsigned char *nat_block_bitmap; 939 unsigned short *free_nid_count; /* free nid count of NAT block */ 940 941 /* for checkpoint */ 942 char *nat_bitmap; /* NAT bitmap pointer */ 943 944 unsigned int nat_bits_blocks; /* # of nat bits blocks */ 945 unsigned char *nat_bits; /* NAT bits blocks */ 946 unsigned char *full_nat_bits; /* full NAT pages */ 947 unsigned char *empty_nat_bits; /* empty NAT pages */ 948 #ifdef CONFIG_F2FS_CHECK_FS 949 char *nat_bitmap_mir; /* NAT bitmap mirror */ 950 #endif 951 int bitmap_size; /* bitmap size */ 952 }; 953 954 /* 955 * this structure is used as one of function parameters. 956 * all the information are dedicated to a given direct node block determined 957 * by the data offset in a file. 958 */ 959 struct dnode_of_data { 960 struct inode *inode; /* vfs inode pointer */ 961 struct page *inode_page; /* its inode page, NULL is possible */ 962 struct page *node_page; /* cached direct node page */ 963 nid_t nid; /* node id of the direct node block */ 964 unsigned int ofs_in_node; /* data offset in the node page */ 965 bool inode_page_locked; /* inode page is locked or not */ 966 bool node_changed; /* is node block changed */ 967 char cur_level; /* level of hole node page */ 968 char max_level; /* level of current page located */ 969 block_t data_blkaddr; /* block address of the node block */ 970 }; 971 972 static inline void set_new_dnode(struct dnode_of_data *dn, struct inode *inode, 973 struct page *ipage, struct page *npage, nid_t nid) 974 { 975 memset(dn, 0, sizeof(*dn)); 976 dn->inode = inode; 977 dn->inode_page = ipage; 978 dn->node_page = npage; 979 dn->nid = nid; 980 } 981 982 /* 983 * For SIT manager 984 * 985 * By default, there are 6 active log areas across the whole main area. 986 * When considering hot and cold data separation to reduce cleaning overhead, 987 * we split 3 for data logs and 3 for node logs as hot, warm, and cold types, 988 * respectively. 989 * In the current design, you should not change the numbers intentionally. 990 * Instead, as a mount option such as active_logs=x, you can use 2, 4, and 6 991 * logs individually according to the underlying devices. (default: 6) 992 * Just in case, on-disk layout covers maximum 16 logs that consist of 8 for 993 * data and 8 for node logs. 994 */ 995 #define NR_CURSEG_DATA_TYPE (3) 996 #define NR_CURSEG_NODE_TYPE (3) 997 #define NR_CURSEG_INMEM_TYPE (2) 998 #define NR_CURSEG_RO_TYPE (2) 999 #define NR_CURSEG_PERSIST_TYPE (NR_CURSEG_DATA_TYPE + NR_CURSEG_NODE_TYPE) 1000 #define NR_CURSEG_TYPE (NR_CURSEG_INMEM_TYPE + NR_CURSEG_PERSIST_TYPE) 1001 1002 enum { 1003 CURSEG_HOT_DATA = 0, /* directory entry blocks */ 1004 CURSEG_WARM_DATA, /* data blocks */ 1005 CURSEG_COLD_DATA, /* multimedia or GCed data blocks */ 1006 CURSEG_HOT_NODE, /* direct node blocks of directory files */ 1007 CURSEG_WARM_NODE, /* direct node blocks of normal files */ 1008 CURSEG_COLD_NODE, /* indirect node blocks */ 1009 NR_PERSISTENT_LOG, /* number of persistent log */ 1010 CURSEG_COLD_DATA_PINNED = NR_PERSISTENT_LOG, 1011 /* pinned file that needs consecutive block address */ 1012 CURSEG_ALL_DATA_ATGC, /* SSR alloctor in hot/warm/cold data area */ 1013 NO_CHECK_TYPE, /* number of persistent & inmem log */ 1014 }; 1015 1016 struct flush_cmd { 1017 struct completion wait; 1018 struct llist_node llnode; 1019 nid_t ino; 1020 int ret; 1021 }; 1022 1023 struct flush_cmd_control { 1024 struct task_struct *f2fs_issue_flush; /* flush thread */ 1025 wait_queue_head_t flush_wait_queue; /* waiting queue for wake-up */ 1026 atomic_t issued_flush; /* # of issued flushes */ 1027 atomic_t queued_flush; /* # of queued flushes */ 1028 struct llist_head issue_list; /* list for command issue */ 1029 struct llist_node *dispatch_list; /* list for command dispatch */ 1030 }; 1031 1032 struct f2fs_sm_info { 1033 struct sit_info *sit_info; /* whole segment information */ 1034 struct free_segmap_info *free_info; /* free segment information */ 1035 struct dirty_seglist_info *dirty_info; /* dirty segment information */ 1036 struct curseg_info *curseg_array; /* active segment information */ 1037 1038 struct f2fs_rwsem curseg_lock; /* for preventing curseg change */ 1039 1040 block_t seg0_blkaddr; /* block address of 0'th segment */ 1041 block_t main_blkaddr; /* start block address of main area */ 1042 block_t ssa_blkaddr; /* start block address of SSA area */ 1043 1044 unsigned int segment_count; /* total # of segments */ 1045 unsigned int main_segments; /* # of segments in main area */ 1046 unsigned int reserved_segments; /* # of reserved segments */ 1047 unsigned int additional_reserved_segments;/* reserved segs for IO align feature */ 1048 unsigned int ovp_segments; /* # of overprovision segments */ 1049 1050 /* a threshold to reclaim prefree segments */ 1051 unsigned int rec_prefree_segments; 1052 1053 struct list_head sit_entry_set; /* sit entry set list */ 1054 1055 unsigned int ipu_policy; /* in-place-update policy */ 1056 unsigned int min_ipu_util; /* in-place-update threshold */ 1057 unsigned int min_fsync_blocks; /* threshold for fsync */ 1058 unsigned int min_seq_blocks; /* threshold for sequential blocks */ 1059 unsigned int min_hot_blocks; /* threshold for hot block allocation */ 1060 unsigned int min_ssr_sections; /* threshold to trigger SSR allocation */ 1061 1062 /* for flush command control */ 1063 struct flush_cmd_control *fcc_info; 1064 1065 /* for discard command control */ 1066 struct discard_cmd_control *dcc_info; 1067 }; 1068 1069 /* 1070 * For superblock 1071 */ 1072 /* 1073 * COUNT_TYPE for monitoring 1074 * 1075 * f2fs monitors the number of several block types such as on-writeback, 1076 * dirty dentry blocks, dirty node blocks, and dirty meta blocks. 1077 */ 1078 #define WB_DATA_TYPE(p) (__is_cp_guaranteed(p) ? F2FS_WB_CP_DATA : F2FS_WB_DATA) 1079 enum count_type { 1080 F2FS_DIRTY_DENTS, 1081 F2FS_DIRTY_DATA, 1082 F2FS_DIRTY_QDATA, 1083 F2FS_DIRTY_NODES, 1084 F2FS_DIRTY_META, 1085 F2FS_DIRTY_IMETA, 1086 F2FS_WB_CP_DATA, 1087 F2FS_WB_DATA, 1088 F2FS_RD_DATA, 1089 F2FS_RD_NODE, 1090 F2FS_RD_META, 1091 F2FS_DIO_WRITE, 1092 F2FS_DIO_READ, 1093 NR_COUNT_TYPE, 1094 }; 1095 1096 /* 1097 * The below are the page types of bios used in submit_bio(). 1098 * The available types are: 1099 * DATA User data pages. It operates as async mode. 1100 * NODE Node pages. It operates as async mode. 1101 * META FS metadata pages such as SIT, NAT, CP. 1102 * NR_PAGE_TYPE The number of page types. 1103 * META_FLUSH Make sure the previous pages are written 1104 * with waiting the bio's completion 1105 * ... Only can be used with META. 1106 */ 1107 #define PAGE_TYPE_OF_BIO(type) ((type) > META ? META : (type)) 1108 enum page_type { 1109 DATA = 0, 1110 NODE = 1, /* should not change this */ 1111 META, 1112 NR_PAGE_TYPE, 1113 META_FLUSH, 1114 IPU, /* the below types are used by tracepoints only. */ 1115 OPU, 1116 }; 1117 1118 enum temp_type { 1119 HOT = 0, /* must be zero for meta bio */ 1120 WARM, 1121 COLD, 1122 NR_TEMP_TYPE, 1123 }; 1124 1125 enum need_lock_type { 1126 LOCK_REQ = 0, 1127 LOCK_DONE, 1128 LOCK_RETRY, 1129 }; 1130 1131 enum cp_reason_type { 1132 CP_NO_NEEDED, 1133 CP_NON_REGULAR, 1134 CP_COMPRESSED, 1135 CP_HARDLINK, 1136 CP_SB_NEED_CP, 1137 CP_WRONG_PINO, 1138 CP_NO_SPC_ROLL, 1139 CP_NODE_NEED_CP, 1140 CP_FASTBOOT_MODE, 1141 CP_SPEC_LOG_NUM, 1142 CP_RECOVER_DIR, 1143 }; 1144 1145 enum iostat_type { 1146 /* WRITE IO */ 1147 APP_DIRECT_IO, /* app direct write IOs */ 1148 APP_BUFFERED_IO, /* app buffered write IOs */ 1149 APP_WRITE_IO, /* app write IOs */ 1150 APP_MAPPED_IO, /* app mapped IOs */ 1151 APP_BUFFERED_CDATA_IO, /* app buffered write IOs on compressed file */ 1152 APP_MAPPED_CDATA_IO, /* app mapped write IOs on compressed file */ 1153 FS_DATA_IO, /* data IOs from kworker/fsync/reclaimer */ 1154 FS_CDATA_IO, /* data IOs from kworker/fsync/reclaimer on compressed file */ 1155 FS_NODE_IO, /* node IOs from kworker/fsync/reclaimer */ 1156 FS_META_IO, /* meta IOs from kworker/reclaimer */ 1157 FS_GC_DATA_IO, /* data IOs from forground gc */ 1158 FS_GC_NODE_IO, /* node IOs from forground gc */ 1159 FS_CP_DATA_IO, /* data IOs from checkpoint */ 1160 FS_CP_NODE_IO, /* node IOs from checkpoint */ 1161 FS_CP_META_IO, /* meta IOs from checkpoint */ 1162 1163 /* READ IO */ 1164 APP_DIRECT_READ_IO, /* app direct read IOs */ 1165 APP_BUFFERED_READ_IO, /* app buffered read IOs */ 1166 APP_READ_IO, /* app read IOs */ 1167 APP_MAPPED_READ_IO, /* app mapped read IOs */ 1168 APP_BUFFERED_CDATA_READ_IO, /* app buffered read IOs on compressed file */ 1169 APP_MAPPED_CDATA_READ_IO, /* app mapped read IOs on compressed file */ 1170 FS_DATA_READ_IO, /* data read IOs */ 1171 FS_GDATA_READ_IO, /* data read IOs from background gc */ 1172 FS_CDATA_READ_IO, /* compressed data read IOs */ 1173 FS_NODE_READ_IO, /* node read IOs */ 1174 FS_META_READ_IO, /* meta read IOs */ 1175 1176 /* other */ 1177 FS_DISCARD_IO, /* discard */ 1178 FS_FLUSH_IO, /* flush */ 1179 FS_ZONE_RESET_IO, /* zone reset */ 1180 NR_IO_TYPE, 1181 }; 1182 1183 struct f2fs_io_info { 1184 struct f2fs_sb_info *sbi; /* f2fs_sb_info pointer */ 1185 nid_t ino; /* inode number */ 1186 enum page_type type; /* contains DATA/NODE/META/META_FLUSH */ 1187 enum temp_type temp; /* contains HOT/WARM/COLD */ 1188 enum req_op op; /* contains REQ_OP_ */ 1189 blk_opf_t op_flags; /* req_flag_bits */ 1190 block_t new_blkaddr; /* new block address to be written */ 1191 block_t old_blkaddr; /* old block address before Cow */ 1192 struct page *page; /* page to be written */ 1193 struct page *encrypted_page; /* encrypted page */ 1194 struct page *compressed_page; /* compressed page */ 1195 struct list_head list; /* serialize IOs */ 1196 unsigned int compr_blocks; /* # of compressed block addresses */ 1197 unsigned int need_lock:8; /* indicate we need to lock cp_rwsem */ 1198 unsigned int version:8; /* version of the node */ 1199 unsigned int submitted:1; /* indicate IO submission */ 1200 unsigned int in_list:1; /* indicate fio is in io_list */ 1201 unsigned int is_por:1; /* indicate IO is from recovery or not */ 1202 unsigned int retry:1; /* need to reallocate block address */ 1203 unsigned int encrypted:1; /* indicate file is encrypted */ 1204 unsigned int post_read:1; /* require post read */ 1205 enum iostat_type io_type; /* io type */ 1206 struct writeback_control *io_wbc; /* writeback control */ 1207 struct bio **bio; /* bio for ipu */ 1208 sector_t *last_block; /* last block number in bio */ 1209 }; 1210 1211 struct bio_entry { 1212 struct bio *bio; 1213 struct list_head list; 1214 }; 1215 1216 #define is_read_io(rw) ((rw) == READ) 1217 struct f2fs_bio_info { 1218 struct f2fs_sb_info *sbi; /* f2fs superblock */ 1219 struct bio *bio; /* bios to merge */ 1220 sector_t last_block_in_bio; /* last block number */ 1221 struct f2fs_io_info fio; /* store buffered io info. */ 1222 #ifdef CONFIG_BLK_DEV_ZONED 1223 struct completion zone_wait; /* condition value for the previous open zone to close */ 1224 struct bio *zone_pending_bio; /* pending bio for the previous zone */ 1225 void *bi_private; /* previous bi_private for pending bio */ 1226 #endif 1227 struct f2fs_rwsem io_rwsem; /* blocking op for bio */ 1228 spinlock_t io_lock; /* serialize DATA/NODE IOs */ 1229 struct list_head io_list; /* track fios */ 1230 struct list_head bio_list; /* bio entry list head */ 1231 struct f2fs_rwsem bio_list_lock; /* lock to protect bio entry list */ 1232 }; 1233 1234 #define FDEV(i) (sbi->devs[i]) 1235 #define RDEV(i) (raw_super->devs[i]) 1236 struct f2fs_dev_info { 1237 struct block_device *bdev; 1238 char path[MAX_PATH_LEN]; 1239 unsigned int total_segments; 1240 block_t start_blk; 1241 block_t end_blk; 1242 #ifdef CONFIG_BLK_DEV_ZONED 1243 unsigned int nr_blkz; /* Total number of zones */ 1244 unsigned long *blkz_seq; /* Bitmap indicating sequential zones */ 1245 #endif 1246 }; 1247 1248 enum inode_type { 1249 DIR_INODE, /* for dirty dir inode */ 1250 FILE_INODE, /* for dirty regular/symlink inode */ 1251 DIRTY_META, /* for all dirtied inode metadata */ 1252 NR_INODE_TYPE, 1253 }; 1254 1255 /* for inner inode cache management */ 1256 struct inode_management { 1257 struct radix_tree_root ino_root; /* ino entry array */ 1258 spinlock_t ino_lock; /* for ino entry lock */ 1259 struct list_head ino_list; /* inode list head */ 1260 unsigned long ino_num; /* number of entries */ 1261 }; 1262 1263 /* for GC_AT */ 1264 struct atgc_management { 1265 bool atgc_enabled; /* ATGC is enabled or not */ 1266 struct rb_root_cached root; /* root of victim rb-tree */ 1267 struct list_head victim_list; /* linked with all victim entries */ 1268 unsigned int victim_count; /* victim count in rb-tree */ 1269 unsigned int candidate_ratio; /* candidate ratio */ 1270 unsigned int max_candidate_count; /* max candidate count */ 1271 unsigned int age_weight; /* age weight, vblock_weight = 100 - age_weight */ 1272 unsigned long long age_threshold; /* age threshold */ 1273 }; 1274 1275 struct f2fs_gc_control { 1276 unsigned int victim_segno; /* target victim segment number */ 1277 int init_gc_type; /* FG_GC or BG_GC */ 1278 bool no_bg_gc; /* check the space and stop bg_gc */ 1279 bool should_migrate_blocks; /* should migrate blocks */ 1280 bool err_gc_skipped; /* return EAGAIN if GC skipped */ 1281 unsigned int nr_free_secs; /* # of free sections to do GC */ 1282 }; 1283 1284 /* 1285 * For s_flag in struct f2fs_sb_info 1286 * Modification on enum should be synchronized with s_flag array 1287 */ 1288 enum { 1289 SBI_IS_DIRTY, /* dirty flag for checkpoint */ 1290 SBI_IS_CLOSE, /* specify unmounting */ 1291 SBI_NEED_FSCK, /* need fsck.f2fs to fix */ 1292 SBI_POR_DOING, /* recovery is doing or not */ 1293 SBI_NEED_SB_WRITE, /* need to recover superblock */ 1294 SBI_NEED_CP, /* need to checkpoint */ 1295 SBI_IS_SHUTDOWN, /* shutdown by ioctl */ 1296 SBI_IS_RECOVERED, /* recovered orphan/data */ 1297 SBI_CP_DISABLED, /* CP was disabled last mount */ 1298 SBI_CP_DISABLED_QUICK, /* CP was disabled quickly */ 1299 SBI_QUOTA_NEED_FLUSH, /* need to flush quota info in CP */ 1300 SBI_QUOTA_SKIP_FLUSH, /* skip flushing quota in current CP */ 1301 SBI_QUOTA_NEED_REPAIR, /* quota file may be corrupted */ 1302 SBI_IS_RESIZEFS, /* resizefs is in process */ 1303 SBI_IS_FREEZING, /* freezefs is in process */ 1304 SBI_IS_WRITABLE, /* remove ro mountoption transiently */ 1305 MAX_SBI_FLAG, 1306 }; 1307 1308 enum { 1309 CP_TIME, 1310 REQ_TIME, 1311 DISCARD_TIME, 1312 GC_TIME, 1313 DISABLE_TIME, 1314 UMOUNT_DISCARD_TIMEOUT, 1315 MAX_TIME, 1316 }; 1317 1318 /* Note that you need to keep synchronization with this gc_mode_names array */ 1319 enum { 1320 GC_NORMAL, 1321 GC_IDLE_CB, 1322 GC_IDLE_GREEDY, 1323 GC_IDLE_AT, 1324 GC_URGENT_HIGH, 1325 GC_URGENT_LOW, 1326 GC_URGENT_MID, 1327 MAX_GC_MODE, 1328 }; 1329 1330 enum { 1331 BGGC_MODE_ON, /* background gc is on */ 1332 BGGC_MODE_OFF, /* background gc is off */ 1333 BGGC_MODE_SYNC, /* 1334 * background gc is on, migrating blocks 1335 * like foreground gc 1336 */ 1337 }; 1338 1339 enum { 1340 FS_MODE_ADAPTIVE, /* use both lfs/ssr allocation */ 1341 FS_MODE_LFS, /* use lfs allocation only */ 1342 FS_MODE_FRAGMENT_SEG, /* segment fragmentation mode */ 1343 FS_MODE_FRAGMENT_BLK, /* block fragmentation mode */ 1344 }; 1345 1346 enum { 1347 ALLOC_MODE_DEFAULT, /* stay default */ 1348 ALLOC_MODE_REUSE, /* reuse segments as much as possible */ 1349 }; 1350 1351 enum fsync_mode { 1352 FSYNC_MODE_POSIX, /* fsync follows posix semantics */ 1353 FSYNC_MODE_STRICT, /* fsync behaves in line with ext4 */ 1354 FSYNC_MODE_NOBARRIER, /* fsync behaves nobarrier based on posix */ 1355 }; 1356 1357 enum { 1358 COMPR_MODE_FS, /* 1359 * automatically compress compression 1360 * enabled files 1361 */ 1362 COMPR_MODE_USER, /* 1363 * automatical compression is disabled. 1364 * user can control the file compression 1365 * using ioctls 1366 */ 1367 }; 1368 1369 enum { 1370 DISCARD_UNIT_BLOCK, /* basic discard unit is block */ 1371 DISCARD_UNIT_SEGMENT, /* basic discard unit is segment */ 1372 DISCARD_UNIT_SECTION, /* basic discard unit is section */ 1373 }; 1374 1375 enum { 1376 MEMORY_MODE_NORMAL, /* memory mode for normal devices */ 1377 MEMORY_MODE_LOW, /* memory mode for low memry devices */ 1378 }; 1379 1380 enum errors_option { 1381 MOUNT_ERRORS_READONLY, /* remount fs ro on errors */ 1382 MOUNT_ERRORS_CONTINUE, /* continue on errors */ 1383 MOUNT_ERRORS_PANIC, /* panic on errors */ 1384 }; 1385 1386 static inline int f2fs_test_bit(unsigned int nr, char *addr); 1387 static inline void f2fs_set_bit(unsigned int nr, char *addr); 1388 static inline void f2fs_clear_bit(unsigned int nr, char *addr); 1389 1390 /* 1391 * Layout of f2fs page.private: 1392 * 1393 * Layout A: lowest bit should be 1 1394 * | bit0 = 1 | bit1 | bit2 | ... | bit MAX | private data .... | 1395 * bit 0 PAGE_PRIVATE_NOT_POINTER 1396 * bit 1 PAGE_PRIVATE_DUMMY_WRITE 1397 * bit 2 PAGE_PRIVATE_ONGOING_MIGRATION 1398 * bit 3 PAGE_PRIVATE_INLINE_INODE 1399 * bit 4 PAGE_PRIVATE_REF_RESOURCE 1400 * bit 5- f2fs private data 1401 * 1402 * Layout B: lowest bit should be 0 1403 * page.private is a wrapped pointer. 1404 */ 1405 enum { 1406 PAGE_PRIVATE_NOT_POINTER, /* private contains non-pointer data */ 1407 PAGE_PRIVATE_DUMMY_WRITE, /* data page for padding aligned IO */ 1408 PAGE_PRIVATE_ONGOING_MIGRATION, /* data page which is on-going migrating */ 1409 PAGE_PRIVATE_INLINE_INODE, /* inode page contains inline data */ 1410 PAGE_PRIVATE_REF_RESOURCE, /* dirty page has referenced resources */ 1411 PAGE_PRIVATE_MAX 1412 }; 1413 1414 /* For compression */ 1415 enum compress_algorithm_type { 1416 COMPRESS_LZO, 1417 COMPRESS_LZ4, 1418 COMPRESS_ZSTD, 1419 COMPRESS_LZORLE, 1420 COMPRESS_MAX, 1421 }; 1422 1423 enum compress_flag { 1424 COMPRESS_CHKSUM, 1425 COMPRESS_MAX_FLAG, 1426 }; 1427 1428 #define COMPRESS_WATERMARK 20 1429 #define COMPRESS_PERCENT 20 1430 1431 #define COMPRESS_DATA_RESERVED_SIZE 4 1432 struct compress_data { 1433 __le32 clen; /* compressed data size */ 1434 __le32 chksum; /* compressed data chksum */ 1435 __le32 reserved[COMPRESS_DATA_RESERVED_SIZE]; /* reserved */ 1436 u8 cdata[]; /* compressed data */ 1437 }; 1438 1439 #define COMPRESS_HEADER_SIZE (sizeof(struct compress_data)) 1440 1441 #define F2FS_COMPRESSED_PAGE_MAGIC 0xF5F2C000 1442 1443 #define F2FS_ZSTD_DEFAULT_CLEVEL 1 1444 1445 #define COMPRESS_LEVEL_OFFSET 8 1446 1447 /* compress context */ 1448 struct compress_ctx { 1449 struct inode *inode; /* inode the context belong to */ 1450 pgoff_t cluster_idx; /* cluster index number */ 1451 unsigned int cluster_size; /* page count in cluster */ 1452 unsigned int log_cluster_size; /* log of cluster size */ 1453 struct page **rpages; /* pages store raw data in cluster */ 1454 unsigned int nr_rpages; /* total page number in rpages */ 1455 struct page **cpages; /* pages store compressed data in cluster */ 1456 unsigned int nr_cpages; /* total page number in cpages */ 1457 unsigned int valid_nr_cpages; /* valid page number in cpages */ 1458 void *rbuf; /* virtual mapped address on rpages */ 1459 struct compress_data *cbuf; /* virtual mapped address on cpages */ 1460 size_t rlen; /* valid data length in rbuf */ 1461 size_t clen; /* valid data length in cbuf */ 1462 void *private; /* payload buffer for specified compression algorithm */ 1463 void *private2; /* extra payload buffer */ 1464 }; 1465 1466 /* compress context for write IO path */ 1467 struct compress_io_ctx { 1468 u32 magic; /* magic number to indicate page is compressed */ 1469 struct inode *inode; /* inode the context belong to */ 1470 struct page **rpages; /* pages store raw data in cluster */ 1471 unsigned int nr_rpages; /* total page number in rpages */ 1472 atomic_t pending_pages; /* in-flight compressed page count */ 1473 }; 1474 1475 /* Context for decompressing one cluster on the read IO path */ 1476 struct decompress_io_ctx { 1477 u32 magic; /* magic number to indicate page is compressed */ 1478 struct inode *inode; /* inode the context belong to */ 1479 pgoff_t cluster_idx; /* cluster index number */ 1480 unsigned int cluster_size; /* page count in cluster */ 1481 unsigned int log_cluster_size; /* log of cluster size */ 1482 struct page **rpages; /* pages store raw data in cluster */ 1483 unsigned int nr_rpages; /* total page number in rpages */ 1484 struct page **cpages; /* pages store compressed data in cluster */ 1485 unsigned int nr_cpages; /* total page number in cpages */ 1486 struct page **tpages; /* temp pages to pad holes in cluster */ 1487 void *rbuf; /* virtual mapped address on rpages */ 1488 struct compress_data *cbuf; /* virtual mapped address on cpages */ 1489 size_t rlen; /* valid data length in rbuf */ 1490 size_t clen; /* valid data length in cbuf */ 1491 1492 /* 1493 * The number of compressed pages remaining to be read in this cluster. 1494 * This is initially nr_cpages. It is decremented by 1 each time a page 1495 * has been read (or failed to be read). When it reaches 0, the cluster 1496 * is decompressed (or an error is reported). 1497 * 1498 * If an error occurs before all the pages have been submitted for I/O, 1499 * then this will never reach 0. In this case the I/O submitter is 1500 * responsible for calling f2fs_decompress_end_io() instead. 1501 */ 1502 atomic_t remaining_pages; 1503 1504 /* 1505 * Number of references to this decompress_io_ctx. 1506 * 1507 * One reference is held for I/O completion. This reference is dropped 1508 * after the pagecache pages are updated and unlocked -- either after 1509 * decompression (and verity if enabled), or after an error. 1510 * 1511 * In addition, each compressed page holds a reference while it is in a 1512 * bio. These references are necessary prevent compressed pages from 1513 * being freed while they are still in a bio. 1514 */ 1515 refcount_t refcnt; 1516 1517 bool failed; /* IO error occurred before decompression? */ 1518 bool need_verity; /* need fs-verity verification after decompression? */ 1519 void *private; /* payload buffer for specified decompression algorithm */ 1520 void *private2; /* extra payload buffer */ 1521 struct work_struct verity_work; /* work to verify the decompressed pages */ 1522 struct work_struct free_work; /* work for late free this structure itself */ 1523 }; 1524 1525 #define NULL_CLUSTER ((unsigned int)(~0)) 1526 #define MIN_COMPRESS_LOG_SIZE 2 1527 #define MAX_COMPRESS_LOG_SIZE 8 1528 #define MAX_COMPRESS_WINDOW_SIZE(log_size) ((PAGE_SIZE) << (log_size)) 1529 1530 struct f2fs_sb_info { 1531 struct super_block *sb; /* pointer to VFS super block */ 1532 struct proc_dir_entry *s_proc; /* proc entry */ 1533 struct f2fs_super_block *raw_super; /* raw super block pointer */ 1534 struct f2fs_rwsem sb_lock; /* lock for raw super block */ 1535 int valid_super_block; /* valid super block no */ 1536 unsigned long s_flag; /* flags for sbi */ 1537 struct mutex writepages; /* mutex for writepages() */ 1538 1539 #ifdef CONFIG_BLK_DEV_ZONED 1540 unsigned int blocks_per_blkz; /* F2FS blocks per zone */ 1541 #endif 1542 1543 /* for node-related operations */ 1544 struct f2fs_nm_info *nm_info; /* node manager */ 1545 struct inode *node_inode; /* cache node blocks */ 1546 1547 /* for segment-related operations */ 1548 struct f2fs_sm_info *sm_info; /* segment manager */ 1549 1550 /* for bio operations */ 1551 struct f2fs_bio_info *write_io[NR_PAGE_TYPE]; /* for write bios */ 1552 /* keep migration IO order for LFS mode */ 1553 struct f2fs_rwsem io_order_lock; 1554 mempool_t *write_io_dummy; /* Dummy pages */ 1555 pgoff_t page_eio_ofs[NR_PAGE_TYPE]; /* EIO page offset */ 1556 int page_eio_cnt[NR_PAGE_TYPE]; /* EIO count */ 1557 1558 /* for checkpoint */ 1559 struct f2fs_checkpoint *ckpt; /* raw checkpoint pointer */ 1560 int cur_cp_pack; /* remain current cp pack */ 1561 spinlock_t cp_lock; /* for flag in ckpt */ 1562 struct inode *meta_inode; /* cache meta blocks */ 1563 struct f2fs_rwsem cp_global_sem; /* checkpoint procedure lock */ 1564 struct f2fs_rwsem cp_rwsem; /* blocking FS operations */ 1565 struct f2fs_rwsem node_write; /* locking node writes */ 1566 struct f2fs_rwsem node_change; /* locking node change */ 1567 wait_queue_head_t cp_wait; 1568 unsigned long last_time[MAX_TIME]; /* to store time in jiffies */ 1569 long interval_time[MAX_TIME]; /* to store thresholds */ 1570 struct ckpt_req_control cprc_info; /* for checkpoint request control */ 1571 1572 struct inode_management im[MAX_INO_ENTRY]; /* manage inode cache */ 1573 1574 spinlock_t fsync_node_lock; /* for node entry lock */ 1575 struct list_head fsync_node_list; /* node list head */ 1576 unsigned int fsync_seg_id; /* sequence id */ 1577 unsigned int fsync_node_num; /* number of node entries */ 1578 1579 /* for orphan inode, use 0'th array */ 1580 unsigned int max_orphans; /* max orphan inodes */ 1581 1582 /* for inode management */ 1583 struct list_head inode_list[NR_INODE_TYPE]; /* dirty inode list */ 1584 spinlock_t inode_lock[NR_INODE_TYPE]; /* for dirty inode list lock */ 1585 struct mutex flush_lock; /* for flush exclusion */ 1586 1587 /* for extent tree cache */ 1588 struct extent_tree_info extent_tree[NR_EXTENT_CACHES]; 1589 atomic64_t allocated_data_blocks; /* for block age extent_cache */ 1590 1591 /* The threshold used for hot and warm data seperation*/ 1592 unsigned int hot_data_age_threshold; 1593 unsigned int warm_data_age_threshold; 1594 unsigned int last_age_weight; 1595 1596 /* basic filesystem units */ 1597 unsigned int log_sectors_per_block; /* log2 sectors per block */ 1598 unsigned int log_blocksize; /* log2 block size */ 1599 unsigned int blocksize; /* block size */ 1600 unsigned int root_ino_num; /* root inode number*/ 1601 unsigned int node_ino_num; /* node inode number*/ 1602 unsigned int meta_ino_num; /* meta inode number*/ 1603 unsigned int log_blocks_per_seg; /* log2 blocks per segment */ 1604 unsigned int blocks_per_seg; /* blocks per segment */ 1605 unsigned int unusable_blocks_per_sec; /* unusable blocks per section */ 1606 unsigned int segs_per_sec; /* segments per section */ 1607 unsigned int secs_per_zone; /* sections per zone */ 1608 unsigned int total_sections; /* total section count */ 1609 unsigned int total_node_count; /* total node block count */ 1610 unsigned int total_valid_node_count; /* valid node block count */ 1611 int dir_level; /* directory level */ 1612 bool readdir_ra; /* readahead inode in readdir */ 1613 u64 max_io_bytes; /* max io bytes to merge IOs */ 1614 1615 block_t user_block_count; /* # of user blocks */ 1616 block_t total_valid_block_count; /* # of valid blocks */ 1617 block_t discard_blks; /* discard command candidats */ 1618 block_t last_valid_block_count; /* for recovery */ 1619 block_t reserved_blocks; /* configurable reserved blocks */ 1620 block_t current_reserved_blocks; /* current reserved blocks */ 1621 1622 /* Additional tracking for no checkpoint mode */ 1623 block_t unusable_block_count; /* # of blocks saved by last cp */ 1624 1625 unsigned int nquota_files; /* # of quota sysfile */ 1626 struct f2fs_rwsem quota_sem; /* blocking cp for flags */ 1627 1628 /* # of pages, see count_type */ 1629 atomic_t nr_pages[NR_COUNT_TYPE]; 1630 /* # of allocated blocks */ 1631 struct percpu_counter alloc_valid_block_count; 1632 /* # of node block writes as roll forward recovery */ 1633 struct percpu_counter rf_node_block_count; 1634 1635 /* writeback control */ 1636 atomic_t wb_sync_req[META]; /* count # of WB_SYNC threads */ 1637 1638 /* valid inode count */ 1639 struct percpu_counter total_valid_inode_count; 1640 1641 struct f2fs_mount_info mount_opt; /* mount options */ 1642 1643 /* for cleaning operations */ 1644 struct f2fs_rwsem gc_lock; /* 1645 * semaphore for GC, avoid 1646 * race between GC and GC or CP 1647 */ 1648 struct f2fs_gc_kthread *gc_thread; /* GC thread */ 1649 struct atgc_management am; /* atgc management */ 1650 unsigned int cur_victim_sec; /* current victim section num */ 1651 unsigned int gc_mode; /* current GC state */ 1652 unsigned int next_victim_seg[2]; /* next segment in victim section */ 1653 spinlock_t gc_remaining_trials_lock; 1654 /* remaining trial count for GC_URGENT_* and GC_IDLE_* */ 1655 unsigned int gc_remaining_trials; 1656 1657 /* for skip statistic */ 1658 unsigned long long skipped_gc_rwsem; /* FG_GC only */ 1659 1660 /* threshold for gc trials on pinned files */ 1661 u64 gc_pin_file_threshold; 1662 struct f2fs_rwsem pin_sem; 1663 1664 /* maximum # of trials to find a victim segment for SSR and GC */ 1665 unsigned int max_victim_search; 1666 /* migration granularity of garbage collection, unit: segment */ 1667 unsigned int migration_granularity; 1668 1669 /* 1670 * for stat information. 1671 * one is for the LFS mode, and the other is for the SSR mode. 1672 */ 1673 #ifdef CONFIG_F2FS_STAT_FS 1674 struct f2fs_stat_info *stat_info; /* FS status information */ 1675 atomic_t meta_count[META_MAX]; /* # of meta blocks */ 1676 unsigned int segment_count[2]; /* # of allocated segments */ 1677 unsigned int block_count[2]; /* # of allocated blocks */ 1678 atomic_t inplace_count; /* # of inplace update */ 1679 /* # of lookup extent cache */ 1680 atomic64_t total_hit_ext[NR_EXTENT_CACHES]; 1681 /* # of hit rbtree extent node */ 1682 atomic64_t read_hit_rbtree[NR_EXTENT_CACHES]; 1683 /* # of hit cached extent node */ 1684 atomic64_t read_hit_cached[NR_EXTENT_CACHES]; 1685 /* # of hit largest extent node in read extent cache */ 1686 atomic64_t read_hit_largest; 1687 atomic_t inline_xattr; /* # of inline_xattr inodes */ 1688 atomic_t inline_inode; /* # of inline_data inodes */ 1689 atomic_t inline_dir; /* # of inline_dentry inodes */ 1690 atomic_t compr_inode; /* # of compressed inodes */ 1691 atomic64_t compr_blocks; /* # of compressed blocks */ 1692 atomic_t swapfile_inode; /* # of swapfile inodes */ 1693 atomic_t atomic_files; /* # of opened atomic file */ 1694 atomic_t max_aw_cnt; /* max # of atomic writes */ 1695 unsigned int io_skip_bggc; /* skip background gc for in-flight IO */ 1696 unsigned int other_skip_bggc; /* skip background gc for other reasons */ 1697 unsigned int ndirty_inode[NR_INODE_TYPE]; /* # of dirty inodes */ 1698 #endif 1699 spinlock_t stat_lock; /* lock for stat operations */ 1700 1701 /* to attach REQ_META|REQ_FUA flags */ 1702 unsigned int data_io_flag; 1703 unsigned int node_io_flag; 1704 1705 /* For sysfs support */ 1706 struct kobject s_kobj; /* /sys/fs/f2fs/<devname> */ 1707 struct completion s_kobj_unregister; 1708 1709 struct kobject s_stat_kobj; /* /sys/fs/f2fs/<devname>/stat */ 1710 struct completion s_stat_kobj_unregister; 1711 1712 struct kobject s_feature_list_kobj; /* /sys/fs/f2fs/<devname>/feature_list */ 1713 struct completion s_feature_list_kobj_unregister; 1714 1715 /* For shrinker support */ 1716 struct list_head s_list; 1717 struct mutex umount_mutex; 1718 unsigned int shrinker_run_no; 1719 1720 /* For multi devices */ 1721 int s_ndevs; /* number of devices */ 1722 struct f2fs_dev_info *devs; /* for device list */ 1723 unsigned int dirty_device; /* for checkpoint data flush */ 1724 spinlock_t dev_lock; /* protect dirty_device */ 1725 bool aligned_blksize; /* all devices has the same logical blksize */ 1726 1727 /* For write statistics */ 1728 u64 sectors_written_start; 1729 u64 kbytes_written; 1730 1731 /* Reference to checksum algorithm driver via cryptoapi */ 1732 struct crypto_shash *s_chksum_driver; 1733 1734 /* Precomputed FS UUID checksum for seeding other checksums */ 1735 __u32 s_chksum_seed; 1736 1737 struct workqueue_struct *post_read_wq; /* post read workqueue */ 1738 1739 /* 1740 * If we are in irq context, let's update error information into 1741 * on-disk superblock in the work. 1742 */ 1743 struct work_struct s_error_work; 1744 unsigned char errors[MAX_F2FS_ERRORS]; /* error flags */ 1745 unsigned char stop_reason[MAX_STOP_REASON]; /* stop reason */ 1746 spinlock_t error_lock; /* protect errors/stop_reason array */ 1747 bool error_dirty; /* errors of sb is dirty */ 1748 1749 struct kmem_cache *inline_xattr_slab; /* inline xattr entry */ 1750 unsigned int inline_xattr_slab_size; /* default inline xattr slab size */ 1751 1752 /* For reclaimed segs statistics per each GC mode */ 1753 unsigned int gc_segment_mode; /* GC state for reclaimed segments */ 1754 unsigned int gc_reclaimed_segs[MAX_GC_MODE]; /* Reclaimed segs for each mode */ 1755 1756 unsigned long seq_file_ra_mul; /* multiplier for ra_pages of seq. files in fadvise */ 1757 1758 int max_fragment_chunk; /* max chunk size for block fragmentation mode */ 1759 int max_fragment_hole; /* max hole size for block fragmentation mode */ 1760 1761 /* For atomic write statistics */ 1762 atomic64_t current_atomic_write; 1763 s64 peak_atomic_write; 1764 u64 committed_atomic_block; 1765 u64 revoked_atomic_block; 1766 1767 #ifdef CONFIG_F2FS_FS_COMPRESSION 1768 struct kmem_cache *page_array_slab; /* page array entry */ 1769 unsigned int page_array_slab_size; /* default page array slab size */ 1770 1771 /* For runtime compression statistics */ 1772 u64 compr_written_block; 1773 u64 compr_saved_block; 1774 u32 compr_new_inode; 1775 1776 /* For compressed block cache */ 1777 struct inode *compress_inode; /* cache compressed blocks */ 1778 unsigned int compress_percent; /* cache page percentage */ 1779 unsigned int compress_watermark; /* cache page watermark */ 1780 atomic_t compress_page_hit; /* cache hit count */ 1781 #endif 1782 1783 #ifdef CONFIG_F2FS_IOSTAT 1784 /* For app/fs IO statistics */ 1785 spinlock_t iostat_lock; 1786 unsigned long long iostat_count[NR_IO_TYPE]; 1787 unsigned long long iostat_bytes[NR_IO_TYPE]; 1788 unsigned long long prev_iostat_bytes[NR_IO_TYPE]; 1789 bool iostat_enable; 1790 unsigned long iostat_next_period; 1791 unsigned int iostat_period_ms; 1792 1793 /* For io latency related statistics info in one iostat period */ 1794 spinlock_t iostat_lat_lock; 1795 struct iostat_lat_info *iostat_io_lat; 1796 #endif 1797 }; 1798 1799 #ifdef CONFIG_F2FS_FAULT_INJECTION 1800 #define time_to_inject(sbi, type) __time_to_inject(sbi, type, __func__, \ 1801 __builtin_return_address(0)) 1802 static inline bool __time_to_inject(struct f2fs_sb_info *sbi, int type, 1803 const char *func, const char *parent_func) 1804 { 1805 struct f2fs_fault_info *ffi = &F2FS_OPTION(sbi).fault_info; 1806 1807 if (!ffi->inject_rate) 1808 return false; 1809 1810 if (!IS_FAULT_SET(ffi, type)) 1811 return false; 1812 1813 atomic_inc(&ffi->inject_ops); 1814 if (atomic_read(&ffi->inject_ops) >= ffi->inject_rate) { 1815 atomic_set(&ffi->inject_ops, 0); 1816 printk_ratelimited("%sF2FS-fs (%s) : inject %s in %s of %pS\n", 1817 KERN_INFO, sbi->sb->s_id, f2fs_fault_name[type], 1818 func, parent_func); 1819 return true; 1820 } 1821 return false; 1822 } 1823 #else 1824 static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type) 1825 { 1826 return false; 1827 } 1828 #endif 1829 1830 /* 1831 * Test if the mounted volume is a multi-device volume. 1832 * - For a single regular disk volume, sbi->s_ndevs is 0. 1833 * - For a single zoned disk volume, sbi->s_ndevs is 1. 1834 * - For a multi-device volume, sbi->s_ndevs is always 2 or more. 1835 */ 1836 static inline bool f2fs_is_multi_device(struct f2fs_sb_info *sbi) 1837 { 1838 return sbi->s_ndevs > 1; 1839 } 1840 1841 static inline void f2fs_update_time(struct f2fs_sb_info *sbi, int type) 1842 { 1843 unsigned long now = jiffies; 1844 1845 sbi->last_time[type] = now; 1846 1847 /* DISCARD_TIME and GC_TIME are based on REQ_TIME */ 1848 if (type == REQ_TIME) { 1849 sbi->last_time[DISCARD_TIME] = now; 1850 sbi->last_time[GC_TIME] = now; 1851 } 1852 } 1853 1854 static inline bool f2fs_time_over(struct f2fs_sb_info *sbi, int type) 1855 { 1856 unsigned long interval = sbi->interval_time[type] * HZ; 1857 1858 return time_after(jiffies, sbi->last_time[type] + interval); 1859 } 1860 1861 static inline unsigned int f2fs_time_to_wait(struct f2fs_sb_info *sbi, 1862 int type) 1863 { 1864 unsigned long interval = sbi->interval_time[type] * HZ; 1865 unsigned int wait_ms = 0; 1866 long delta; 1867 1868 delta = (sbi->last_time[type] + interval) - jiffies; 1869 if (delta > 0) 1870 wait_ms = jiffies_to_msecs(delta); 1871 1872 return wait_ms; 1873 } 1874 1875 /* 1876 * Inline functions 1877 */ 1878 static inline u32 __f2fs_crc32(struct f2fs_sb_info *sbi, u32 crc, 1879 const void *address, unsigned int length) 1880 { 1881 struct { 1882 struct shash_desc shash; 1883 char ctx[4]; 1884 } desc; 1885 int err; 1886 1887 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver) != sizeof(desc.ctx)); 1888 1889 desc.shash.tfm = sbi->s_chksum_driver; 1890 *(u32 *)desc.ctx = crc; 1891 1892 err = crypto_shash_update(&desc.shash, address, length); 1893 BUG_ON(err); 1894 1895 return *(u32 *)desc.ctx; 1896 } 1897 1898 static inline u32 f2fs_crc32(struct f2fs_sb_info *sbi, const void *address, 1899 unsigned int length) 1900 { 1901 return __f2fs_crc32(sbi, F2FS_SUPER_MAGIC, address, length); 1902 } 1903 1904 static inline bool f2fs_crc_valid(struct f2fs_sb_info *sbi, __u32 blk_crc, 1905 void *buf, size_t buf_size) 1906 { 1907 return f2fs_crc32(sbi, buf, buf_size) == blk_crc; 1908 } 1909 1910 static inline u32 f2fs_chksum(struct f2fs_sb_info *sbi, u32 crc, 1911 const void *address, unsigned int length) 1912 { 1913 return __f2fs_crc32(sbi, crc, address, length); 1914 } 1915 1916 static inline struct f2fs_inode_info *F2FS_I(struct inode *inode) 1917 { 1918 return container_of(inode, struct f2fs_inode_info, vfs_inode); 1919 } 1920 1921 static inline struct f2fs_sb_info *F2FS_SB(struct super_block *sb) 1922 { 1923 return sb->s_fs_info; 1924 } 1925 1926 static inline struct f2fs_sb_info *F2FS_I_SB(struct inode *inode) 1927 { 1928 return F2FS_SB(inode->i_sb); 1929 } 1930 1931 static inline struct f2fs_sb_info *F2FS_M_SB(struct address_space *mapping) 1932 { 1933 return F2FS_I_SB(mapping->host); 1934 } 1935 1936 static inline struct f2fs_sb_info *F2FS_P_SB(struct page *page) 1937 { 1938 return F2FS_M_SB(page_file_mapping(page)); 1939 } 1940 1941 static inline struct f2fs_super_block *F2FS_RAW_SUPER(struct f2fs_sb_info *sbi) 1942 { 1943 return (struct f2fs_super_block *)(sbi->raw_super); 1944 } 1945 1946 static inline struct f2fs_checkpoint *F2FS_CKPT(struct f2fs_sb_info *sbi) 1947 { 1948 return (struct f2fs_checkpoint *)(sbi->ckpt); 1949 } 1950 1951 static inline struct f2fs_node *F2FS_NODE(struct page *page) 1952 { 1953 return (struct f2fs_node *)page_address(page); 1954 } 1955 1956 static inline struct f2fs_inode *F2FS_INODE(struct page *page) 1957 { 1958 return &((struct f2fs_node *)page_address(page))->i; 1959 } 1960 1961 static inline struct f2fs_nm_info *NM_I(struct f2fs_sb_info *sbi) 1962 { 1963 return (struct f2fs_nm_info *)(sbi->nm_info); 1964 } 1965 1966 static inline struct f2fs_sm_info *SM_I(struct f2fs_sb_info *sbi) 1967 { 1968 return (struct f2fs_sm_info *)(sbi->sm_info); 1969 } 1970 1971 static inline struct sit_info *SIT_I(struct f2fs_sb_info *sbi) 1972 { 1973 return (struct sit_info *)(SM_I(sbi)->sit_info); 1974 } 1975 1976 static inline struct free_segmap_info *FREE_I(struct f2fs_sb_info *sbi) 1977 { 1978 return (struct free_segmap_info *)(SM_I(sbi)->free_info); 1979 } 1980 1981 static inline struct dirty_seglist_info *DIRTY_I(struct f2fs_sb_info *sbi) 1982 { 1983 return (struct dirty_seglist_info *)(SM_I(sbi)->dirty_info); 1984 } 1985 1986 static inline struct address_space *META_MAPPING(struct f2fs_sb_info *sbi) 1987 { 1988 return sbi->meta_inode->i_mapping; 1989 } 1990 1991 static inline struct address_space *NODE_MAPPING(struct f2fs_sb_info *sbi) 1992 { 1993 return sbi->node_inode->i_mapping; 1994 } 1995 1996 static inline bool is_sbi_flag_set(struct f2fs_sb_info *sbi, unsigned int type) 1997 { 1998 return test_bit(type, &sbi->s_flag); 1999 } 2000 2001 static inline void set_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type) 2002 { 2003 set_bit(type, &sbi->s_flag); 2004 } 2005 2006 static inline void clear_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type) 2007 { 2008 clear_bit(type, &sbi->s_flag); 2009 } 2010 2011 static inline unsigned long long cur_cp_version(struct f2fs_checkpoint *cp) 2012 { 2013 return le64_to_cpu(cp->checkpoint_ver); 2014 } 2015 2016 static inline unsigned long f2fs_qf_ino(struct super_block *sb, int type) 2017 { 2018 if (type < F2FS_MAX_QUOTAS) 2019 return le32_to_cpu(F2FS_SB(sb)->raw_super->qf_ino[type]); 2020 return 0; 2021 } 2022 2023 static inline __u64 cur_cp_crc(struct f2fs_checkpoint *cp) 2024 { 2025 size_t crc_offset = le32_to_cpu(cp->checksum_offset); 2026 return le32_to_cpu(*((__le32 *)((unsigned char *)cp + crc_offset))); 2027 } 2028 2029 static inline bool __is_set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 2030 { 2031 unsigned int ckpt_flags = le32_to_cpu(cp->ckpt_flags); 2032 2033 return ckpt_flags & f; 2034 } 2035 2036 static inline bool is_set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 2037 { 2038 return __is_set_ckpt_flags(F2FS_CKPT(sbi), f); 2039 } 2040 2041 static inline void __set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 2042 { 2043 unsigned int ckpt_flags; 2044 2045 ckpt_flags = le32_to_cpu(cp->ckpt_flags); 2046 ckpt_flags |= f; 2047 cp->ckpt_flags = cpu_to_le32(ckpt_flags); 2048 } 2049 2050 static inline void set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 2051 { 2052 unsigned long flags; 2053 2054 spin_lock_irqsave(&sbi->cp_lock, flags); 2055 __set_ckpt_flags(F2FS_CKPT(sbi), f); 2056 spin_unlock_irqrestore(&sbi->cp_lock, flags); 2057 } 2058 2059 static inline void __clear_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 2060 { 2061 unsigned int ckpt_flags; 2062 2063 ckpt_flags = le32_to_cpu(cp->ckpt_flags); 2064 ckpt_flags &= (~f); 2065 cp->ckpt_flags = cpu_to_le32(ckpt_flags); 2066 } 2067 2068 static inline void clear_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 2069 { 2070 unsigned long flags; 2071 2072 spin_lock_irqsave(&sbi->cp_lock, flags); 2073 __clear_ckpt_flags(F2FS_CKPT(sbi), f); 2074 spin_unlock_irqrestore(&sbi->cp_lock, flags); 2075 } 2076 2077 #define init_f2fs_rwsem(sem) \ 2078 do { \ 2079 static struct lock_class_key __key; \ 2080 \ 2081 __init_f2fs_rwsem((sem), #sem, &__key); \ 2082 } while (0) 2083 2084 static inline void __init_f2fs_rwsem(struct f2fs_rwsem *sem, 2085 const char *sem_name, struct lock_class_key *key) 2086 { 2087 __init_rwsem(&sem->internal_rwsem, sem_name, key); 2088 #ifdef CONFIG_F2FS_UNFAIR_RWSEM 2089 init_waitqueue_head(&sem->read_waiters); 2090 #endif 2091 } 2092 2093 static inline int f2fs_rwsem_is_locked(struct f2fs_rwsem *sem) 2094 { 2095 return rwsem_is_locked(&sem->internal_rwsem); 2096 } 2097 2098 static inline int f2fs_rwsem_is_contended(struct f2fs_rwsem *sem) 2099 { 2100 return rwsem_is_contended(&sem->internal_rwsem); 2101 } 2102 2103 static inline void f2fs_down_read(struct f2fs_rwsem *sem) 2104 { 2105 #ifdef CONFIG_F2FS_UNFAIR_RWSEM 2106 wait_event(sem->read_waiters, down_read_trylock(&sem->internal_rwsem)); 2107 #else 2108 down_read(&sem->internal_rwsem); 2109 #endif 2110 } 2111 2112 static inline int f2fs_down_read_trylock(struct f2fs_rwsem *sem) 2113 { 2114 return down_read_trylock(&sem->internal_rwsem); 2115 } 2116 2117 #ifdef CONFIG_DEBUG_LOCK_ALLOC 2118 static inline void f2fs_down_read_nested(struct f2fs_rwsem *sem, int subclass) 2119 { 2120 down_read_nested(&sem->internal_rwsem, subclass); 2121 } 2122 #else 2123 #define f2fs_down_read_nested(sem, subclass) f2fs_down_read(sem) 2124 #endif 2125 2126 static inline void f2fs_up_read(struct f2fs_rwsem *sem) 2127 { 2128 up_read(&sem->internal_rwsem); 2129 } 2130 2131 static inline void f2fs_down_write(struct f2fs_rwsem *sem) 2132 { 2133 down_write(&sem->internal_rwsem); 2134 } 2135 2136 static inline int f2fs_down_write_trylock(struct f2fs_rwsem *sem) 2137 { 2138 return down_write_trylock(&sem->internal_rwsem); 2139 } 2140 2141 static inline void f2fs_up_write(struct f2fs_rwsem *sem) 2142 { 2143 up_write(&sem->internal_rwsem); 2144 #ifdef CONFIG_F2FS_UNFAIR_RWSEM 2145 wake_up_all(&sem->read_waiters); 2146 #endif 2147 } 2148 2149 static inline void f2fs_lock_op(struct f2fs_sb_info *sbi) 2150 { 2151 f2fs_down_read(&sbi->cp_rwsem); 2152 } 2153 2154 static inline int f2fs_trylock_op(struct f2fs_sb_info *sbi) 2155 { 2156 if (time_to_inject(sbi, FAULT_LOCK_OP)) 2157 return 0; 2158 return f2fs_down_read_trylock(&sbi->cp_rwsem); 2159 } 2160 2161 static inline void f2fs_unlock_op(struct f2fs_sb_info *sbi) 2162 { 2163 f2fs_up_read(&sbi->cp_rwsem); 2164 } 2165 2166 static inline void f2fs_lock_all(struct f2fs_sb_info *sbi) 2167 { 2168 f2fs_down_write(&sbi->cp_rwsem); 2169 } 2170 2171 static inline void f2fs_unlock_all(struct f2fs_sb_info *sbi) 2172 { 2173 f2fs_up_write(&sbi->cp_rwsem); 2174 } 2175 2176 static inline int __get_cp_reason(struct f2fs_sb_info *sbi) 2177 { 2178 int reason = CP_SYNC; 2179 2180 if (test_opt(sbi, FASTBOOT)) 2181 reason = CP_FASTBOOT; 2182 if (is_sbi_flag_set(sbi, SBI_IS_CLOSE)) 2183 reason = CP_UMOUNT; 2184 return reason; 2185 } 2186 2187 static inline bool __remain_node_summaries(int reason) 2188 { 2189 return (reason & (CP_UMOUNT | CP_FASTBOOT)); 2190 } 2191 2192 static inline bool __exist_node_summaries(struct f2fs_sb_info *sbi) 2193 { 2194 return (is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG) || 2195 is_set_ckpt_flags(sbi, CP_FASTBOOT_FLAG)); 2196 } 2197 2198 /* 2199 * Check whether the inode has blocks or not 2200 */ 2201 static inline int F2FS_HAS_BLOCKS(struct inode *inode) 2202 { 2203 block_t xattr_block = F2FS_I(inode)->i_xattr_nid ? 1 : 0; 2204 2205 return (inode->i_blocks >> F2FS_LOG_SECTORS_PER_BLOCK) > xattr_block; 2206 } 2207 2208 static inline bool f2fs_has_xattr_block(unsigned int ofs) 2209 { 2210 return ofs == XATTR_NODE_OFFSET; 2211 } 2212 2213 static inline bool __allow_reserved_blocks(struct f2fs_sb_info *sbi, 2214 struct inode *inode, bool cap) 2215 { 2216 if (!inode) 2217 return true; 2218 if (!test_opt(sbi, RESERVE_ROOT)) 2219 return false; 2220 if (IS_NOQUOTA(inode)) 2221 return true; 2222 if (uid_eq(F2FS_OPTION(sbi).s_resuid, current_fsuid())) 2223 return true; 2224 if (!gid_eq(F2FS_OPTION(sbi).s_resgid, GLOBAL_ROOT_GID) && 2225 in_group_p(F2FS_OPTION(sbi).s_resgid)) 2226 return true; 2227 if (cap && capable(CAP_SYS_RESOURCE)) 2228 return true; 2229 return false; 2230 } 2231 2232 static inline void f2fs_i_blocks_write(struct inode *, block_t, bool, bool); 2233 static inline int inc_valid_block_count(struct f2fs_sb_info *sbi, 2234 struct inode *inode, blkcnt_t *count) 2235 { 2236 blkcnt_t diff = 0, release = 0; 2237 block_t avail_user_block_count; 2238 int ret; 2239 2240 ret = dquot_reserve_block(inode, *count); 2241 if (ret) 2242 return ret; 2243 2244 if (time_to_inject(sbi, FAULT_BLOCK)) { 2245 release = *count; 2246 goto release_quota; 2247 } 2248 2249 /* 2250 * let's increase this in prior to actual block count change in order 2251 * for f2fs_sync_file to avoid data races when deciding checkpoint. 2252 */ 2253 percpu_counter_add(&sbi->alloc_valid_block_count, (*count)); 2254 2255 spin_lock(&sbi->stat_lock); 2256 sbi->total_valid_block_count += (block_t)(*count); 2257 avail_user_block_count = sbi->user_block_count - 2258 sbi->current_reserved_blocks; 2259 2260 if (!__allow_reserved_blocks(sbi, inode, true)) 2261 avail_user_block_count -= F2FS_OPTION(sbi).root_reserved_blocks; 2262 2263 if (F2FS_IO_ALIGNED(sbi)) 2264 avail_user_block_count -= sbi->blocks_per_seg * 2265 SM_I(sbi)->additional_reserved_segments; 2266 2267 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) { 2268 if (avail_user_block_count > sbi->unusable_block_count) 2269 avail_user_block_count -= sbi->unusable_block_count; 2270 else 2271 avail_user_block_count = 0; 2272 } 2273 if (unlikely(sbi->total_valid_block_count > avail_user_block_count)) { 2274 diff = sbi->total_valid_block_count - avail_user_block_count; 2275 if (diff > *count) 2276 diff = *count; 2277 *count -= diff; 2278 release = diff; 2279 sbi->total_valid_block_count -= diff; 2280 if (!*count) { 2281 spin_unlock(&sbi->stat_lock); 2282 goto enospc; 2283 } 2284 } 2285 spin_unlock(&sbi->stat_lock); 2286 2287 if (unlikely(release)) { 2288 percpu_counter_sub(&sbi->alloc_valid_block_count, release); 2289 dquot_release_reservation_block(inode, release); 2290 } 2291 f2fs_i_blocks_write(inode, *count, true, true); 2292 return 0; 2293 2294 enospc: 2295 percpu_counter_sub(&sbi->alloc_valid_block_count, release); 2296 release_quota: 2297 dquot_release_reservation_block(inode, release); 2298 return -ENOSPC; 2299 } 2300 2301 __printf(2, 3) 2302 void f2fs_printk(struct f2fs_sb_info *sbi, const char *fmt, ...); 2303 2304 #define f2fs_err(sbi, fmt, ...) \ 2305 f2fs_printk(sbi, KERN_ERR fmt, ##__VA_ARGS__) 2306 #define f2fs_warn(sbi, fmt, ...) \ 2307 f2fs_printk(sbi, KERN_WARNING fmt, ##__VA_ARGS__) 2308 #define f2fs_notice(sbi, fmt, ...) \ 2309 f2fs_printk(sbi, KERN_NOTICE fmt, ##__VA_ARGS__) 2310 #define f2fs_info(sbi, fmt, ...) \ 2311 f2fs_printk(sbi, KERN_INFO fmt, ##__VA_ARGS__) 2312 #define f2fs_debug(sbi, fmt, ...) \ 2313 f2fs_printk(sbi, KERN_DEBUG fmt, ##__VA_ARGS__) 2314 2315 #define PAGE_PRIVATE_GET_FUNC(name, flagname) \ 2316 static inline bool page_private_##name(struct page *page) \ 2317 { \ 2318 return PagePrivate(page) && \ 2319 test_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)) && \ 2320 test_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \ 2321 } 2322 2323 #define PAGE_PRIVATE_SET_FUNC(name, flagname) \ 2324 static inline void set_page_private_##name(struct page *page) \ 2325 { \ 2326 if (!PagePrivate(page)) \ 2327 attach_page_private(page, (void *)0); \ 2328 set_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)); \ 2329 set_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \ 2330 } 2331 2332 #define PAGE_PRIVATE_CLEAR_FUNC(name, flagname) \ 2333 static inline void clear_page_private_##name(struct page *page) \ 2334 { \ 2335 clear_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \ 2336 if (page_private(page) == BIT(PAGE_PRIVATE_NOT_POINTER)) \ 2337 detach_page_private(page); \ 2338 } 2339 2340 PAGE_PRIVATE_GET_FUNC(nonpointer, NOT_POINTER); 2341 PAGE_PRIVATE_GET_FUNC(inline, INLINE_INODE); 2342 PAGE_PRIVATE_GET_FUNC(gcing, ONGOING_MIGRATION); 2343 PAGE_PRIVATE_GET_FUNC(dummy, DUMMY_WRITE); 2344 2345 PAGE_PRIVATE_SET_FUNC(reference, REF_RESOURCE); 2346 PAGE_PRIVATE_SET_FUNC(inline, INLINE_INODE); 2347 PAGE_PRIVATE_SET_FUNC(gcing, ONGOING_MIGRATION); 2348 PAGE_PRIVATE_SET_FUNC(dummy, DUMMY_WRITE); 2349 2350 PAGE_PRIVATE_CLEAR_FUNC(reference, REF_RESOURCE); 2351 PAGE_PRIVATE_CLEAR_FUNC(inline, INLINE_INODE); 2352 PAGE_PRIVATE_CLEAR_FUNC(gcing, ONGOING_MIGRATION); 2353 PAGE_PRIVATE_CLEAR_FUNC(dummy, DUMMY_WRITE); 2354 2355 static inline unsigned long get_page_private_data(struct page *page) 2356 { 2357 unsigned long data = page_private(page); 2358 2359 if (!test_bit(PAGE_PRIVATE_NOT_POINTER, &data)) 2360 return 0; 2361 return data >> PAGE_PRIVATE_MAX; 2362 } 2363 2364 static inline void set_page_private_data(struct page *page, unsigned long data) 2365 { 2366 if (!PagePrivate(page)) 2367 attach_page_private(page, (void *)0); 2368 set_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)); 2369 page_private(page) |= data << PAGE_PRIVATE_MAX; 2370 } 2371 2372 static inline void clear_page_private_data(struct page *page) 2373 { 2374 page_private(page) &= GENMASK(PAGE_PRIVATE_MAX - 1, 0); 2375 if (page_private(page) == BIT(PAGE_PRIVATE_NOT_POINTER)) 2376 detach_page_private(page); 2377 } 2378 2379 static inline void clear_page_private_all(struct page *page) 2380 { 2381 clear_page_private_data(page); 2382 clear_page_private_reference(page); 2383 clear_page_private_gcing(page); 2384 clear_page_private_inline(page); 2385 2386 f2fs_bug_on(F2FS_P_SB(page), page_private(page)); 2387 } 2388 2389 static inline void dec_valid_block_count(struct f2fs_sb_info *sbi, 2390 struct inode *inode, 2391 block_t count) 2392 { 2393 blkcnt_t sectors = count << F2FS_LOG_SECTORS_PER_BLOCK; 2394 2395 spin_lock(&sbi->stat_lock); 2396 f2fs_bug_on(sbi, sbi->total_valid_block_count < (block_t) count); 2397 sbi->total_valid_block_count -= (block_t)count; 2398 if (sbi->reserved_blocks && 2399 sbi->current_reserved_blocks < sbi->reserved_blocks) 2400 sbi->current_reserved_blocks = min(sbi->reserved_blocks, 2401 sbi->current_reserved_blocks + count); 2402 spin_unlock(&sbi->stat_lock); 2403 if (unlikely(inode->i_blocks < sectors)) { 2404 f2fs_warn(sbi, "Inconsistent i_blocks, ino:%lu, iblocks:%llu, sectors:%llu", 2405 inode->i_ino, 2406 (unsigned long long)inode->i_blocks, 2407 (unsigned long long)sectors); 2408 set_sbi_flag(sbi, SBI_NEED_FSCK); 2409 return; 2410 } 2411 f2fs_i_blocks_write(inode, count, false, true); 2412 } 2413 2414 static inline void inc_page_count(struct f2fs_sb_info *sbi, int count_type) 2415 { 2416 atomic_inc(&sbi->nr_pages[count_type]); 2417 2418 if (count_type == F2FS_DIRTY_DENTS || 2419 count_type == F2FS_DIRTY_NODES || 2420 count_type == F2FS_DIRTY_META || 2421 count_type == F2FS_DIRTY_QDATA || 2422 count_type == F2FS_DIRTY_IMETA) 2423 set_sbi_flag(sbi, SBI_IS_DIRTY); 2424 } 2425 2426 static inline void inode_inc_dirty_pages(struct inode *inode) 2427 { 2428 atomic_inc(&F2FS_I(inode)->dirty_pages); 2429 inc_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ? 2430 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA); 2431 if (IS_NOQUOTA(inode)) 2432 inc_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA); 2433 } 2434 2435 static inline void dec_page_count(struct f2fs_sb_info *sbi, int count_type) 2436 { 2437 atomic_dec(&sbi->nr_pages[count_type]); 2438 } 2439 2440 static inline void inode_dec_dirty_pages(struct inode *inode) 2441 { 2442 if (!S_ISDIR(inode->i_mode) && !S_ISREG(inode->i_mode) && 2443 !S_ISLNK(inode->i_mode)) 2444 return; 2445 2446 atomic_dec(&F2FS_I(inode)->dirty_pages); 2447 dec_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ? 2448 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA); 2449 if (IS_NOQUOTA(inode)) 2450 dec_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA); 2451 } 2452 2453 static inline void inc_atomic_write_cnt(struct inode *inode) 2454 { 2455 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 2456 struct f2fs_inode_info *fi = F2FS_I(inode); 2457 u64 current_write; 2458 2459 fi->atomic_write_cnt++; 2460 atomic64_inc(&sbi->current_atomic_write); 2461 current_write = atomic64_read(&sbi->current_atomic_write); 2462 if (current_write > sbi->peak_atomic_write) 2463 sbi->peak_atomic_write = current_write; 2464 } 2465 2466 static inline void release_atomic_write_cnt(struct inode *inode) 2467 { 2468 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 2469 struct f2fs_inode_info *fi = F2FS_I(inode); 2470 2471 atomic64_sub(fi->atomic_write_cnt, &sbi->current_atomic_write); 2472 fi->atomic_write_cnt = 0; 2473 } 2474 2475 static inline s64 get_pages(struct f2fs_sb_info *sbi, int count_type) 2476 { 2477 return atomic_read(&sbi->nr_pages[count_type]); 2478 } 2479 2480 static inline int get_dirty_pages(struct inode *inode) 2481 { 2482 return atomic_read(&F2FS_I(inode)->dirty_pages); 2483 } 2484 2485 static inline int get_blocktype_secs(struct f2fs_sb_info *sbi, int block_type) 2486 { 2487 unsigned int pages_per_sec = sbi->segs_per_sec * sbi->blocks_per_seg; 2488 unsigned int segs = (get_pages(sbi, block_type) + pages_per_sec - 1) >> 2489 sbi->log_blocks_per_seg; 2490 2491 return segs / sbi->segs_per_sec; 2492 } 2493 2494 static inline block_t valid_user_blocks(struct f2fs_sb_info *sbi) 2495 { 2496 return sbi->total_valid_block_count; 2497 } 2498 2499 static inline block_t discard_blocks(struct f2fs_sb_info *sbi) 2500 { 2501 return sbi->discard_blks; 2502 } 2503 2504 static inline unsigned long __bitmap_size(struct f2fs_sb_info *sbi, int flag) 2505 { 2506 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi); 2507 2508 /* return NAT or SIT bitmap */ 2509 if (flag == NAT_BITMAP) 2510 return le32_to_cpu(ckpt->nat_ver_bitmap_bytesize); 2511 else if (flag == SIT_BITMAP) 2512 return le32_to_cpu(ckpt->sit_ver_bitmap_bytesize); 2513 2514 return 0; 2515 } 2516 2517 static inline block_t __cp_payload(struct f2fs_sb_info *sbi) 2518 { 2519 return le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_payload); 2520 } 2521 2522 static inline void *__bitmap_ptr(struct f2fs_sb_info *sbi, int flag) 2523 { 2524 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi); 2525 void *tmp_ptr = &ckpt->sit_nat_version_bitmap; 2526 int offset; 2527 2528 if (is_set_ckpt_flags(sbi, CP_LARGE_NAT_BITMAP_FLAG)) { 2529 offset = (flag == SIT_BITMAP) ? 2530 le32_to_cpu(ckpt->nat_ver_bitmap_bytesize) : 0; 2531 /* 2532 * if large_nat_bitmap feature is enabled, leave checksum 2533 * protection for all nat/sit bitmaps. 2534 */ 2535 return tmp_ptr + offset + sizeof(__le32); 2536 } 2537 2538 if (__cp_payload(sbi) > 0) { 2539 if (flag == NAT_BITMAP) 2540 return tmp_ptr; 2541 else 2542 return (unsigned char *)ckpt + F2FS_BLKSIZE; 2543 } else { 2544 offset = (flag == NAT_BITMAP) ? 2545 le32_to_cpu(ckpt->sit_ver_bitmap_bytesize) : 0; 2546 return tmp_ptr + offset; 2547 } 2548 } 2549 2550 static inline block_t __start_cp_addr(struct f2fs_sb_info *sbi) 2551 { 2552 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr); 2553 2554 if (sbi->cur_cp_pack == 2) 2555 start_addr += sbi->blocks_per_seg; 2556 return start_addr; 2557 } 2558 2559 static inline block_t __start_cp_next_addr(struct f2fs_sb_info *sbi) 2560 { 2561 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr); 2562 2563 if (sbi->cur_cp_pack == 1) 2564 start_addr += sbi->blocks_per_seg; 2565 return start_addr; 2566 } 2567 2568 static inline void __set_cp_next_pack(struct f2fs_sb_info *sbi) 2569 { 2570 sbi->cur_cp_pack = (sbi->cur_cp_pack == 1) ? 2 : 1; 2571 } 2572 2573 static inline block_t __start_sum_addr(struct f2fs_sb_info *sbi) 2574 { 2575 return le32_to_cpu(F2FS_CKPT(sbi)->cp_pack_start_sum); 2576 } 2577 2578 extern void f2fs_mark_inode_dirty_sync(struct inode *inode, bool sync); 2579 static inline int inc_valid_node_count(struct f2fs_sb_info *sbi, 2580 struct inode *inode, bool is_inode) 2581 { 2582 block_t valid_block_count; 2583 unsigned int valid_node_count, user_block_count; 2584 int err; 2585 2586 if (is_inode) { 2587 if (inode) { 2588 err = dquot_alloc_inode(inode); 2589 if (err) 2590 return err; 2591 } 2592 } else { 2593 err = dquot_reserve_block(inode, 1); 2594 if (err) 2595 return err; 2596 } 2597 2598 if (time_to_inject(sbi, FAULT_BLOCK)) 2599 goto enospc; 2600 2601 spin_lock(&sbi->stat_lock); 2602 2603 valid_block_count = sbi->total_valid_block_count + 2604 sbi->current_reserved_blocks + 1; 2605 2606 if (!__allow_reserved_blocks(sbi, inode, false)) 2607 valid_block_count += F2FS_OPTION(sbi).root_reserved_blocks; 2608 2609 if (F2FS_IO_ALIGNED(sbi)) 2610 valid_block_count += sbi->blocks_per_seg * 2611 SM_I(sbi)->additional_reserved_segments; 2612 2613 user_block_count = sbi->user_block_count; 2614 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) 2615 user_block_count -= sbi->unusable_block_count; 2616 2617 if (unlikely(valid_block_count > user_block_count)) { 2618 spin_unlock(&sbi->stat_lock); 2619 goto enospc; 2620 } 2621 2622 valid_node_count = sbi->total_valid_node_count + 1; 2623 if (unlikely(valid_node_count > sbi->total_node_count)) { 2624 spin_unlock(&sbi->stat_lock); 2625 goto enospc; 2626 } 2627 2628 sbi->total_valid_node_count++; 2629 sbi->total_valid_block_count++; 2630 spin_unlock(&sbi->stat_lock); 2631 2632 if (inode) { 2633 if (is_inode) 2634 f2fs_mark_inode_dirty_sync(inode, true); 2635 else 2636 f2fs_i_blocks_write(inode, 1, true, true); 2637 } 2638 2639 percpu_counter_inc(&sbi->alloc_valid_block_count); 2640 return 0; 2641 2642 enospc: 2643 if (is_inode) { 2644 if (inode) 2645 dquot_free_inode(inode); 2646 } else { 2647 dquot_release_reservation_block(inode, 1); 2648 } 2649 return -ENOSPC; 2650 } 2651 2652 static inline void dec_valid_node_count(struct f2fs_sb_info *sbi, 2653 struct inode *inode, bool is_inode) 2654 { 2655 spin_lock(&sbi->stat_lock); 2656 2657 if (unlikely(!sbi->total_valid_block_count || 2658 !sbi->total_valid_node_count)) { 2659 f2fs_warn(sbi, "dec_valid_node_count: inconsistent block counts, total_valid_block:%u, total_valid_node:%u", 2660 sbi->total_valid_block_count, 2661 sbi->total_valid_node_count); 2662 set_sbi_flag(sbi, SBI_NEED_FSCK); 2663 } else { 2664 sbi->total_valid_block_count--; 2665 sbi->total_valid_node_count--; 2666 } 2667 2668 if (sbi->reserved_blocks && 2669 sbi->current_reserved_blocks < sbi->reserved_blocks) 2670 sbi->current_reserved_blocks++; 2671 2672 spin_unlock(&sbi->stat_lock); 2673 2674 if (is_inode) { 2675 dquot_free_inode(inode); 2676 } else { 2677 if (unlikely(inode->i_blocks == 0)) { 2678 f2fs_warn(sbi, "dec_valid_node_count: inconsistent i_blocks, ino:%lu, iblocks:%llu", 2679 inode->i_ino, 2680 (unsigned long long)inode->i_blocks); 2681 set_sbi_flag(sbi, SBI_NEED_FSCK); 2682 return; 2683 } 2684 f2fs_i_blocks_write(inode, 1, false, true); 2685 } 2686 } 2687 2688 static inline unsigned int valid_node_count(struct f2fs_sb_info *sbi) 2689 { 2690 return sbi->total_valid_node_count; 2691 } 2692 2693 static inline void inc_valid_inode_count(struct f2fs_sb_info *sbi) 2694 { 2695 percpu_counter_inc(&sbi->total_valid_inode_count); 2696 } 2697 2698 static inline void dec_valid_inode_count(struct f2fs_sb_info *sbi) 2699 { 2700 percpu_counter_dec(&sbi->total_valid_inode_count); 2701 } 2702 2703 static inline s64 valid_inode_count(struct f2fs_sb_info *sbi) 2704 { 2705 return percpu_counter_sum_positive(&sbi->total_valid_inode_count); 2706 } 2707 2708 static inline struct page *f2fs_grab_cache_page(struct address_space *mapping, 2709 pgoff_t index, bool for_write) 2710 { 2711 struct page *page; 2712 unsigned int flags; 2713 2714 if (IS_ENABLED(CONFIG_F2FS_FAULT_INJECTION)) { 2715 if (!for_write) 2716 page = find_get_page_flags(mapping, index, 2717 FGP_LOCK | FGP_ACCESSED); 2718 else 2719 page = find_lock_page(mapping, index); 2720 if (page) 2721 return page; 2722 2723 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_ALLOC)) 2724 return NULL; 2725 } 2726 2727 if (!for_write) 2728 return grab_cache_page(mapping, index); 2729 2730 flags = memalloc_nofs_save(); 2731 page = grab_cache_page_write_begin(mapping, index); 2732 memalloc_nofs_restore(flags); 2733 2734 return page; 2735 } 2736 2737 static inline struct page *f2fs_pagecache_get_page( 2738 struct address_space *mapping, pgoff_t index, 2739 int fgp_flags, gfp_t gfp_mask) 2740 { 2741 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_GET)) 2742 return NULL; 2743 2744 return pagecache_get_page(mapping, index, fgp_flags, gfp_mask); 2745 } 2746 2747 static inline void f2fs_put_page(struct page *page, int unlock) 2748 { 2749 if (!page) 2750 return; 2751 2752 if (unlock) { 2753 f2fs_bug_on(F2FS_P_SB(page), !PageLocked(page)); 2754 unlock_page(page); 2755 } 2756 put_page(page); 2757 } 2758 2759 static inline void f2fs_put_dnode(struct dnode_of_data *dn) 2760 { 2761 if (dn->node_page) 2762 f2fs_put_page(dn->node_page, 1); 2763 if (dn->inode_page && dn->node_page != dn->inode_page) 2764 f2fs_put_page(dn->inode_page, 0); 2765 dn->node_page = NULL; 2766 dn->inode_page = NULL; 2767 } 2768 2769 static inline struct kmem_cache *f2fs_kmem_cache_create(const char *name, 2770 size_t size) 2771 { 2772 return kmem_cache_create(name, size, 0, SLAB_RECLAIM_ACCOUNT, NULL); 2773 } 2774 2775 static inline void *f2fs_kmem_cache_alloc_nofail(struct kmem_cache *cachep, 2776 gfp_t flags) 2777 { 2778 void *entry; 2779 2780 entry = kmem_cache_alloc(cachep, flags); 2781 if (!entry) 2782 entry = kmem_cache_alloc(cachep, flags | __GFP_NOFAIL); 2783 return entry; 2784 } 2785 2786 static inline void *f2fs_kmem_cache_alloc(struct kmem_cache *cachep, 2787 gfp_t flags, bool nofail, struct f2fs_sb_info *sbi) 2788 { 2789 if (nofail) 2790 return f2fs_kmem_cache_alloc_nofail(cachep, flags); 2791 2792 if (time_to_inject(sbi, FAULT_SLAB_ALLOC)) 2793 return NULL; 2794 2795 return kmem_cache_alloc(cachep, flags); 2796 } 2797 2798 static inline bool is_inflight_io(struct f2fs_sb_info *sbi, int type) 2799 { 2800 if (get_pages(sbi, F2FS_RD_DATA) || get_pages(sbi, F2FS_RD_NODE) || 2801 get_pages(sbi, F2FS_RD_META) || get_pages(sbi, F2FS_WB_DATA) || 2802 get_pages(sbi, F2FS_WB_CP_DATA) || 2803 get_pages(sbi, F2FS_DIO_READ) || 2804 get_pages(sbi, F2FS_DIO_WRITE)) 2805 return true; 2806 2807 if (type != DISCARD_TIME && SM_I(sbi) && SM_I(sbi)->dcc_info && 2808 atomic_read(&SM_I(sbi)->dcc_info->queued_discard)) 2809 return true; 2810 2811 if (SM_I(sbi) && SM_I(sbi)->fcc_info && 2812 atomic_read(&SM_I(sbi)->fcc_info->queued_flush)) 2813 return true; 2814 return false; 2815 } 2816 2817 static inline bool is_idle(struct f2fs_sb_info *sbi, int type) 2818 { 2819 if (sbi->gc_mode == GC_URGENT_HIGH) 2820 return true; 2821 2822 if (is_inflight_io(sbi, type)) 2823 return false; 2824 2825 if (sbi->gc_mode == GC_URGENT_MID) 2826 return true; 2827 2828 if (sbi->gc_mode == GC_URGENT_LOW && 2829 (type == DISCARD_TIME || type == GC_TIME)) 2830 return true; 2831 2832 return f2fs_time_over(sbi, type); 2833 } 2834 2835 static inline void f2fs_radix_tree_insert(struct radix_tree_root *root, 2836 unsigned long index, void *item) 2837 { 2838 while (radix_tree_insert(root, index, item)) 2839 cond_resched(); 2840 } 2841 2842 #define RAW_IS_INODE(p) ((p)->footer.nid == (p)->footer.ino) 2843 2844 static inline bool IS_INODE(struct page *page) 2845 { 2846 struct f2fs_node *p = F2FS_NODE(page); 2847 2848 return RAW_IS_INODE(p); 2849 } 2850 2851 static inline int offset_in_addr(struct f2fs_inode *i) 2852 { 2853 return (i->i_inline & F2FS_EXTRA_ATTR) ? 2854 (le16_to_cpu(i->i_extra_isize) / sizeof(__le32)) : 0; 2855 } 2856 2857 static inline __le32 *blkaddr_in_node(struct f2fs_node *node) 2858 { 2859 return RAW_IS_INODE(node) ? node->i.i_addr : node->dn.addr; 2860 } 2861 2862 static inline int f2fs_has_extra_attr(struct inode *inode); 2863 static inline block_t data_blkaddr(struct inode *inode, 2864 struct page *node_page, unsigned int offset) 2865 { 2866 struct f2fs_node *raw_node; 2867 __le32 *addr_array; 2868 int base = 0; 2869 bool is_inode = IS_INODE(node_page); 2870 2871 raw_node = F2FS_NODE(node_page); 2872 2873 if (is_inode) { 2874 if (!inode) 2875 /* from GC path only */ 2876 base = offset_in_addr(&raw_node->i); 2877 else if (f2fs_has_extra_attr(inode)) 2878 base = get_extra_isize(inode); 2879 } 2880 2881 addr_array = blkaddr_in_node(raw_node); 2882 return le32_to_cpu(addr_array[base + offset]); 2883 } 2884 2885 static inline block_t f2fs_data_blkaddr(struct dnode_of_data *dn) 2886 { 2887 return data_blkaddr(dn->inode, dn->node_page, dn->ofs_in_node); 2888 } 2889 2890 static inline int f2fs_test_bit(unsigned int nr, char *addr) 2891 { 2892 int mask; 2893 2894 addr += (nr >> 3); 2895 mask = BIT(7 - (nr & 0x07)); 2896 return mask & *addr; 2897 } 2898 2899 static inline void f2fs_set_bit(unsigned int nr, char *addr) 2900 { 2901 int mask; 2902 2903 addr += (nr >> 3); 2904 mask = BIT(7 - (nr & 0x07)); 2905 *addr |= mask; 2906 } 2907 2908 static inline void f2fs_clear_bit(unsigned int nr, char *addr) 2909 { 2910 int mask; 2911 2912 addr += (nr >> 3); 2913 mask = BIT(7 - (nr & 0x07)); 2914 *addr &= ~mask; 2915 } 2916 2917 static inline int f2fs_test_and_set_bit(unsigned int nr, char *addr) 2918 { 2919 int mask; 2920 int ret; 2921 2922 addr += (nr >> 3); 2923 mask = BIT(7 - (nr & 0x07)); 2924 ret = mask & *addr; 2925 *addr |= mask; 2926 return ret; 2927 } 2928 2929 static inline int f2fs_test_and_clear_bit(unsigned int nr, char *addr) 2930 { 2931 int mask; 2932 int ret; 2933 2934 addr += (nr >> 3); 2935 mask = BIT(7 - (nr & 0x07)); 2936 ret = mask & *addr; 2937 *addr &= ~mask; 2938 return ret; 2939 } 2940 2941 static inline void f2fs_change_bit(unsigned int nr, char *addr) 2942 { 2943 int mask; 2944 2945 addr += (nr >> 3); 2946 mask = BIT(7 - (nr & 0x07)); 2947 *addr ^= mask; 2948 } 2949 2950 /* 2951 * On-disk inode flags (f2fs_inode::i_flags) 2952 */ 2953 #define F2FS_COMPR_FL 0x00000004 /* Compress file */ 2954 #define F2FS_SYNC_FL 0x00000008 /* Synchronous updates */ 2955 #define F2FS_IMMUTABLE_FL 0x00000010 /* Immutable file */ 2956 #define F2FS_APPEND_FL 0x00000020 /* writes to file may only append */ 2957 #define F2FS_NODUMP_FL 0x00000040 /* do not dump file */ 2958 #define F2FS_NOATIME_FL 0x00000080 /* do not update atime */ 2959 #define F2FS_NOCOMP_FL 0x00000400 /* Don't compress */ 2960 #define F2FS_INDEX_FL 0x00001000 /* hash-indexed directory */ 2961 #define F2FS_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */ 2962 #define F2FS_PROJINHERIT_FL 0x20000000 /* Create with parents projid */ 2963 #define F2FS_CASEFOLD_FL 0x40000000 /* Casefolded file */ 2964 2965 #define F2FS_QUOTA_DEFAULT_FL (F2FS_NOATIME_FL | F2FS_IMMUTABLE_FL) 2966 2967 /* Flags that should be inherited by new inodes from their parent. */ 2968 #define F2FS_FL_INHERITED (F2FS_SYNC_FL | F2FS_NODUMP_FL | F2FS_NOATIME_FL | \ 2969 F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \ 2970 F2FS_CASEFOLD_FL) 2971 2972 /* Flags that are appropriate for regular files (all but dir-specific ones). */ 2973 #define F2FS_REG_FLMASK (~(F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \ 2974 F2FS_CASEFOLD_FL)) 2975 2976 /* Flags that are appropriate for non-directories/regular files. */ 2977 #define F2FS_OTHER_FLMASK (F2FS_NODUMP_FL | F2FS_NOATIME_FL) 2978 2979 static inline __u32 f2fs_mask_flags(umode_t mode, __u32 flags) 2980 { 2981 if (S_ISDIR(mode)) 2982 return flags; 2983 else if (S_ISREG(mode)) 2984 return flags & F2FS_REG_FLMASK; 2985 else 2986 return flags & F2FS_OTHER_FLMASK; 2987 } 2988 2989 static inline void __mark_inode_dirty_flag(struct inode *inode, 2990 int flag, bool set) 2991 { 2992 switch (flag) { 2993 case FI_INLINE_XATTR: 2994 case FI_INLINE_DATA: 2995 case FI_INLINE_DENTRY: 2996 case FI_NEW_INODE: 2997 if (set) 2998 return; 2999 fallthrough; 3000 case FI_DATA_EXIST: 3001 case FI_INLINE_DOTS: 3002 case FI_PIN_FILE: 3003 case FI_COMPRESS_RELEASED: 3004 f2fs_mark_inode_dirty_sync(inode, true); 3005 } 3006 } 3007 3008 static inline void set_inode_flag(struct inode *inode, int flag) 3009 { 3010 set_bit(flag, F2FS_I(inode)->flags); 3011 __mark_inode_dirty_flag(inode, flag, true); 3012 } 3013 3014 static inline int is_inode_flag_set(struct inode *inode, int flag) 3015 { 3016 return test_bit(flag, F2FS_I(inode)->flags); 3017 } 3018 3019 static inline void clear_inode_flag(struct inode *inode, int flag) 3020 { 3021 clear_bit(flag, F2FS_I(inode)->flags); 3022 __mark_inode_dirty_flag(inode, flag, false); 3023 } 3024 3025 static inline bool f2fs_verity_in_progress(struct inode *inode) 3026 { 3027 return IS_ENABLED(CONFIG_FS_VERITY) && 3028 is_inode_flag_set(inode, FI_VERITY_IN_PROGRESS); 3029 } 3030 3031 static inline void set_acl_inode(struct inode *inode, umode_t mode) 3032 { 3033 F2FS_I(inode)->i_acl_mode = mode; 3034 set_inode_flag(inode, FI_ACL_MODE); 3035 f2fs_mark_inode_dirty_sync(inode, false); 3036 } 3037 3038 static inline void f2fs_i_links_write(struct inode *inode, bool inc) 3039 { 3040 if (inc) 3041 inc_nlink(inode); 3042 else 3043 drop_nlink(inode); 3044 f2fs_mark_inode_dirty_sync(inode, true); 3045 } 3046 3047 static inline void f2fs_i_blocks_write(struct inode *inode, 3048 block_t diff, bool add, bool claim) 3049 { 3050 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE); 3051 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER); 3052 3053 /* add = 1, claim = 1 should be dquot_reserve_block in pair */ 3054 if (add) { 3055 if (claim) 3056 dquot_claim_block(inode, diff); 3057 else 3058 dquot_alloc_block_nofail(inode, diff); 3059 } else { 3060 dquot_free_block(inode, diff); 3061 } 3062 3063 f2fs_mark_inode_dirty_sync(inode, true); 3064 if (clean || recover) 3065 set_inode_flag(inode, FI_AUTO_RECOVER); 3066 } 3067 3068 static inline bool f2fs_is_atomic_file(struct inode *inode); 3069 3070 static inline void f2fs_i_size_write(struct inode *inode, loff_t i_size) 3071 { 3072 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE); 3073 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER); 3074 3075 if (i_size_read(inode) == i_size) 3076 return; 3077 3078 i_size_write(inode, i_size); 3079 3080 if (f2fs_is_atomic_file(inode)) 3081 return; 3082 3083 f2fs_mark_inode_dirty_sync(inode, true); 3084 if (clean || recover) 3085 set_inode_flag(inode, FI_AUTO_RECOVER); 3086 } 3087 3088 static inline void f2fs_i_depth_write(struct inode *inode, unsigned int depth) 3089 { 3090 F2FS_I(inode)->i_current_depth = depth; 3091 f2fs_mark_inode_dirty_sync(inode, true); 3092 } 3093 3094 static inline void f2fs_i_gc_failures_write(struct inode *inode, 3095 unsigned int count) 3096 { 3097 F2FS_I(inode)->i_gc_failures[GC_FAILURE_PIN] = count; 3098 f2fs_mark_inode_dirty_sync(inode, true); 3099 } 3100 3101 static inline void f2fs_i_xnid_write(struct inode *inode, nid_t xnid) 3102 { 3103 F2FS_I(inode)->i_xattr_nid = xnid; 3104 f2fs_mark_inode_dirty_sync(inode, true); 3105 } 3106 3107 static inline void f2fs_i_pino_write(struct inode *inode, nid_t pino) 3108 { 3109 F2FS_I(inode)->i_pino = pino; 3110 f2fs_mark_inode_dirty_sync(inode, true); 3111 } 3112 3113 static inline void get_inline_info(struct inode *inode, struct f2fs_inode *ri) 3114 { 3115 struct f2fs_inode_info *fi = F2FS_I(inode); 3116 3117 if (ri->i_inline & F2FS_INLINE_XATTR) 3118 set_bit(FI_INLINE_XATTR, fi->flags); 3119 if (ri->i_inline & F2FS_INLINE_DATA) 3120 set_bit(FI_INLINE_DATA, fi->flags); 3121 if (ri->i_inline & F2FS_INLINE_DENTRY) 3122 set_bit(FI_INLINE_DENTRY, fi->flags); 3123 if (ri->i_inline & F2FS_DATA_EXIST) 3124 set_bit(FI_DATA_EXIST, fi->flags); 3125 if (ri->i_inline & F2FS_INLINE_DOTS) 3126 set_bit(FI_INLINE_DOTS, fi->flags); 3127 if (ri->i_inline & F2FS_EXTRA_ATTR) 3128 set_bit(FI_EXTRA_ATTR, fi->flags); 3129 if (ri->i_inline & F2FS_PIN_FILE) 3130 set_bit(FI_PIN_FILE, fi->flags); 3131 if (ri->i_inline & F2FS_COMPRESS_RELEASED) 3132 set_bit(FI_COMPRESS_RELEASED, fi->flags); 3133 } 3134 3135 static inline void set_raw_inline(struct inode *inode, struct f2fs_inode *ri) 3136 { 3137 ri->i_inline = 0; 3138 3139 if (is_inode_flag_set(inode, FI_INLINE_XATTR)) 3140 ri->i_inline |= F2FS_INLINE_XATTR; 3141 if (is_inode_flag_set(inode, FI_INLINE_DATA)) 3142 ri->i_inline |= F2FS_INLINE_DATA; 3143 if (is_inode_flag_set(inode, FI_INLINE_DENTRY)) 3144 ri->i_inline |= F2FS_INLINE_DENTRY; 3145 if (is_inode_flag_set(inode, FI_DATA_EXIST)) 3146 ri->i_inline |= F2FS_DATA_EXIST; 3147 if (is_inode_flag_set(inode, FI_INLINE_DOTS)) 3148 ri->i_inline |= F2FS_INLINE_DOTS; 3149 if (is_inode_flag_set(inode, FI_EXTRA_ATTR)) 3150 ri->i_inline |= F2FS_EXTRA_ATTR; 3151 if (is_inode_flag_set(inode, FI_PIN_FILE)) 3152 ri->i_inline |= F2FS_PIN_FILE; 3153 if (is_inode_flag_set(inode, FI_COMPRESS_RELEASED)) 3154 ri->i_inline |= F2FS_COMPRESS_RELEASED; 3155 } 3156 3157 static inline int f2fs_has_extra_attr(struct inode *inode) 3158 { 3159 return is_inode_flag_set(inode, FI_EXTRA_ATTR); 3160 } 3161 3162 static inline int f2fs_has_inline_xattr(struct inode *inode) 3163 { 3164 return is_inode_flag_set(inode, FI_INLINE_XATTR); 3165 } 3166 3167 static inline int f2fs_compressed_file(struct inode *inode) 3168 { 3169 return S_ISREG(inode->i_mode) && 3170 is_inode_flag_set(inode, FI_COMPRESSED_FILE); 3171 } 3172 3173 static inline bool f2fs_need_compress_data(struct inode *inode) 3174 { 3175 int compress_mode = F2FS_OPTION(F2FS_I_SB(inode)).compress_mode; 3176 3177 if (!f2fs_compressed_file(inode)) 3178 return false; 3179 3180 if (compress_mode == COMPR_MODE_FS) 3181 return true; 3182 else if (compress_mode == COMPR_MODE_USER && 3183 is_inode_flag_set(inode, FI_ENABLE_COMPRESS)) 3184 return true; 3185 3186 return false; 3187 } 3188 3189 static inline unsigned int addrs_per_inode(struct inode *inode) 3190 { 3191 unsigned int addrs = CUR_ADDRS_PER_INODE(inode) - 3192 get_inline_xattr_addrs(inode); 3193 3194 if (!f2fs_compressed_file(inode)) 3195 return addrs; 3196 return ALIGN_DOWN(addrs, F2FS_I(inode)->i_cluster_size); 3197 } 3198 3199 static inline unsigned int addrs_per_block(struct inode *inode) 3200 { 3201 if (!f2fs_compressed_file(inode)) 3202 return DEF_ADDRS_PER_BLOCK; 3203 return ALIGN_DOWN(DEF_ADDRS_PER_BLOCK, F2FS_I(inode)->i_cluster_size); 3204 } 3205 3206 static inline void *inline_xattr_addr(struct inode *inode, struct page *page) 3207 { 3208 struct f2fs_inode *ri = F2FS_INODE(page); 3209 3210 return (void *)&(ri->i_addr[DEF_ADDRS_PER_INODE - 3211 get_inline_xattr_addrs(inode)]); 3212 } 3213 3214 static inline int inline_xattr_size(struct inode *inode) 3215 { 3216 if (f2fs_has_inline_xattr(inode)) 3217 return get_inline_xattr_addrs(inode) * sizeof(__le32); 3218 return 0; 3219 } 3220 3221 /* 3222 * Notice: check inline_data flag without inode page lock is unsafe. 3223 * It could change at any time by f2fs_convert_inline_page(). 3224 */ 3225 static inline int f2fs_has_inline_data(struct inode *inode) 3226 { 3227 return is_inode_flag_set(inode, FI_INLINE_DATA); 3228 } 3229 3230 static inline int f2fs_exist_data(struct inode *inode) 3231 { 3232 return is_inode_flag_set(inode, FI_DATA_EXIST); 3233 } 3234 3235 static inline int f2fs_has_inline_dots(struct inode *inode) 3236 { 3237 return is_inode_flag_set(inode, FI_INLINE_DOTS); 3238 } 3239 3240 static inline int f2fs_is_mmap_file(struct inode *inode) 3241 { 3242 return is_inode_flag_set(inode, FI_MMAP_FILE); 3243 } 3244 3245 static inline bool f2fs_is_pinned_file(struct inode *inode) 3246 { 3247 return is_inode_flag_set(inode, FI_PIN_FILE); 3248 } 3249 3250 static inline bool f2fs_is_atomic_file(struct inode *inode) 3251 { 3252 return is_inode_flag_set(inode, FI_ATOMIC_FILE); 3253 } 3254 3255 static inline bool f2fs_is_cow_file(struct inode *inode) 3256 { 3257 return is_inode_flag_set(inode, FI_COW_FILE); 3258 } 3259 3260 static inline bool f2fs_is_first_block_written(struct inode *inode) 3261 { 3262 return is_inode_flag_set(inode, FI_FIRST_BLOCK_WRITTEN); 3263 } 3264 3265 static inline bool f2fs_is_drop_cache(struct inode *inode) 3266 { 3267 return is_inode_flag_set(inode, FI_DROP_CACHE); 3268 } 3269 3270 static inline void *inline_data_addr(struct inode *inode, struct page *page) 3271 { 3272 struct f2fs_inode *ri = F2FS_INODE(page); 3273 int extra_size = get_extra_isize(inode); 3274 3275 return (void *)&(ri->i_addr[extra_size + DEF_INLINE_RESERVED_SIZE]); 3276 } 3277 3278 static inline int f2fs_has_inline_dentry(struct inode *inode) 3279 { 3280 return is_inode_flag_set(inode, FI_INLINE_DENTRY); 3281 } 3282 3283 static inline int is_file(struct inode *inode, int type) 3284 { 3285 return F2FS_I(inode)->i_advise & type; 3286 } 3287 3288 static inline void set_file(struct inode *inode, int type) 3289 { 3290 if (is_file(inode, type)) 3291 return; 3292 F2FS_I(inode)->i_advise |= type; 3293 f2fs_mark_inode_dirty_sync(inode, true); 3294 } 3295 3296 static inline void clear_file(struct inode *inode, int type) 3297 { 3298 if (!is_file(inode, type)) 3299 return; 3300 F2FS_I(inode)->i_advise &= ~type; 3301 f2fs_mark_inode_dirty_sync(inode, true); 3302 } 3303 3304 static inline bool f2fs_is_time_consistent(struct inode *inode) 3305 { 3306 if (!timespec64_equal(F2FS_I(inode)->i_disk_time, &inode->i_atime)) 3307 return false; 3308 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 1, &inode->i_ctime)) 3309 return false; 3310 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 2, &inode->i_mtime)) 3311 return false; 3312 return true; 3313 } 3314 3315 static inline bool f2fs_skip_inode_update(struct inode *inode, int dsync) 3316 { 3317 bool ret; 3318 3319 if (dsync) { 3320 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 3321 3322 spin_lock(&sbi->inode_lock[DIRTY_META]); 3323 ret = list_empty(&F2FS_I(inode)->gdirty_list); 3324 spin_unlock(&sbi->inode_lock[DIRTY_META]); 3325 return ret; 3326 } 3327 if (!is_inode_flag_set(inode, FI_AUTO_RECOVER) || 3328 file_keep_isize(inode) || 3329 i_size_read(inode) & ~PAGE_MASK) 3330 return false; 3331 3332 if (!f2fs_is_time_consistent(inode)) 3333 return false; 3334 3335 spin_lock(&F2FS_I(inode)->i_size_lock); 3336 ret = F2FS_I(inode)->last_disk_size == i_size_read(inode); 3337 spin_unlock(&F2FS_I(inode)->i_size_lock); 3338 3339 return ret; 3340 } 3341 3342 static inline bool f2fs_readonly(struct super_block *sb) 3343 { 3344 return sb_rdonly(sb); 3345 } 3346 3347 static inline bool f2fs_cp_error(struct f2fs_sb_info *sbi) 3348 { 3349 return is_set_ckpt_flags(sbi, CP_ERROR_FLAG); 3350 } 3351 3352 static inline bool is_dot_dotdot(const u8 *name, size_t len) 3353 { 3354 if (len == 1 && name[0] == '.') 3355 return true; 3356 3357 if (len == 2 && name[0] == '.' && name[1] == '.') 3358 return true; 3359 3360 return false; 3361 } 3362 3363 static inline void *f2fs_kmalloc(struct f2fs_sb_info *sbi, 3364 size_t size, gfp_t flags) 3365 { 3366 if (time_to_inject(sbi, FAULT_KMALLOC)) 3367 return NULL; 3368 3369 return kmalloc(size, flags); 3370 } 3371 3372 static inline void *f2fs_getname(struct f2fs_sb_info *sbi) 3373 { 3374 if (time_to_inject(sbi, FAULT_KMALLOC)) 3375 return NULL; 3376 3377 return __getname(); 3378 } 3379 3380 static inline void f2fs_putname(char *buf) 3381 { 3382 __putname(buf); 3383 } 3384 3385 static inline void *f2fs_kzalloc(struct f2fs_sb_info *sbi, 3386 size_t size, gfp_t flags) 3387 { 3388 return f2fs_kmalloc(sbi, size, flags | __GFP_ZERO); 3389 } 3390 3391 static inline void *f2fs_kvmalloc(struct f2fs_sb_info *sbi, 3392 size_t size, gfp_t flags) 3393 { 3394 if (time_to_inject(sbi, FAULT_KVMALLOC)) 3395 return NULL; 3396 3397 return kvmalloc(size, flags); 3398 } 3399 3400 static inline void *f2fs_kvzalloc(struct f2fs_sb_info *sbi, 3401 size_t size, gfp_t flags) 3402 { 3403 return f2fs_kvmalloc(sbi, size, flags | __GFP_ZERO); 3404 } 3405 3406 static inline int get_extra_isize(struct inode *inode) 3407 { 3408 return F2FS_I(inode)->i_extra_isize / sizeof(__le32); 3409 } 3410 3411 static inline int get_inline_xattr_addrs(struct inode *inode) 3412 { 3413 return F2FS_I(inode)->i_inline_xattr_size; 3414 } 3415 3416 #define f2fs_get_inode_mode(i) \ 3417 ((is_inode_flag_set(i, FI_ACL_MODE)) ? \ 3418 (F2FS_I(i)->i_acl_mode) : ((i)->i_mode)) 3419 3420 #define F2FS_MIN_EXTRA_ATTR_SIZE (sizeof(__le32)) 3421 3422 #define F2FS_TOTAL_EXTRA_ATTR_SIZE \ 3423 (offsetof(struct f2fs_inode, i_extra_end) - \ 3424 offsetof(struct f2fs_inode, i_extra_isize)) \ 3425 3426 #define F2FS_OLD_ATTRIBUTE_SIZE (offsetof(struct f2fs_inode, i_addr)) 3427 #define F2FS_FITS_IN_INODE(f2fs_inode, extra_isize, field) \ 3428 ((offsetof(typeof(*(f2fs_inode)), field) + \ 3429 sizeof((f2fs_inode)->field)) \ 3430 <= (F2FS_OLD_ATTRIBUTE_SIZE + (extra_isize))) \ 3431 3432 #define __is_large_section(sbi) ((sbi)->segs_per_sec > 1) 3433 3434 #define __is_meta_io(fio) (PAGE_TYPE_OF_BIO((fio)->type) == META) 3435 3436 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi, 3437 block_t blkaddr, int type); 3438 static inline void verify_blkaddr(struct f2fs_sb_info *sbi, 3439 block_t blkaddr, int type) 3440 { 3441 if (!f2fs_is_valid_blkaddr(sbi, blkaddr, type)) { 3442 f2fs_err(sbi, "invalid blkaddr: %u, type: %d, run fsck to fix.", 3443 blkaddr, type); 3444 f2fs_bug_on(sbi, 1); 3445 } 3446 } 3447 3448 static inline bool __is_valid_data_blkaddr(block_t blkaddr) 3449 { 3450 if (blkaddr == NEW_ADDR || blkaddr == NULL_ADDR || 3451 blkaddr == COMPRESS_ADDR) 3452 return false; 3453 return true; 3454 } 3455 3456 /* 3457 * file.c 3458 */ 3459 int f2fs_sync_file(struct file *file, loff_t start, loff_t end, int datasync); 3460 int f2fs_do_truncate_blocks(struct inode *inode, u64 from, bool lock); 3461 int f2fs_truncate_blocks(struct inode *inode, u64 from, bool lock); 3462 int f2fs_truncate(struct inode *inode); 3463 int f2fs_getattr(struct mnt_idmap *idmap, const struct path *path, 3464 struct kstat *stat, u32 request_mask, unsigned int flags); 3465 int f2fs_setattr(struct mnt_idmap *idmap, struct dentry *dentry, 3466 struct iattr *attr); 3467 int f2fs_truncate_hole(struct inode *inode, pgoff_t pg_start, pgoff_t pg_end); 3468 void f2fs_truncate_data_blocks_range(struct dnode_of_data *dn, int count); 3469 int f2fs_precache_extents(struct inode *inode); 3470 int f2fs_fileattr_get(struct dentry *dentry, struct fileattr *fa); 3471 int f2fs_fileattr_set(struct mnt_idmap *idmap, 3472 struct dentry *dentry, struct fileattr *fa); 3473 long f2fs_ioctl(struct file *filp, unsigned int cmd, unsigned long arg); 3474 long f2fs_compat_ioctl(struct file *file, unsigned int cmd, unsigned long arg); 3475 int f2fs_transfer_project_quota(struct inode *inode, kprojid_t kprojid); 3476 int f2fs_pin_file_control(struct inode *inode, bool inc); 3477 3478 /* 3479 * inode.c 3480 */ 3481 void f2fs_set_inode_flags(struct inode *inode); 3482 bool f2fs_inode_chksum_verify(struct f2fs_sb_info *sbi, struct page *page); 3483 void f2fs_inode_chksum_set(struct f2fs_sb_info *sbi, struct page *page); 3484 struct inode *f2fs_iget(struct super_block *sb, unsigned long ino); 3485 struct inode *f2fs_iget_retry(struct super_block *sb, unsigned long ino); 3486 int f2fs_try_to_free_nats(struct f2fs_sb_info *sbi, int nr_shrink); 3487 void f2fs_update_inode(struct inode *inode, struct page *node_page); 3488 void f2fs_update_inode_page(struct inode *inode); 3489 int f2fs_write_inode(struct inode *inode, struct writeback_control *wbc); 3490 void f2fs_evict_inode(struct inode *inode); 3491 void f2fs_handle_failed_inode(struct inode *inode); 3492 3493 /* 3494 * namei.c 3495 */ 3496 int f2fs_update_extension_list(struct f2fs_sb_info *sbi, const char *name, 3497 bool hot, bool set); 3498 struct dentry *f2fs_get_parent(struct dentry *child); 3499 int f2fs_get_tmpfile(struct mnt_idmap *idmap, struct inode *dir, 3500 struct inode **new_inode); 3501 3502 /* 3503 * dir.c 3504 */ 3505 int f2fs_init_casefolded_name(const struct inode *dir, 3506 struct f2fs_filename *fname); 3507 int f2fs_setup_filename(struct inode *dir, const struct qstr *iname, 3508 int lookup, struct f2fs_filename *fname); 3509 int f2fs_prepare_lookup(struct inode *dir, struct dentry *dentry, 3510 struct f2fs_filename *fname); 3511 void f2fs_free_filename(struct f2fs_filename *fname); 3512 struct f2fs_dir_entry *f2fs_find_target_dentry(const struct f2fs_dentry_ptr *d, 3513 const struct f2fs_filename *fname, int *max_slots); 3514 int f2fs_fill_dentries(struct dir_context *ctx, struct f2fs_dentry_ptr *d, 3515 unsigned int start_pos, struct fscrypt_str *fstr); 3516 void f2fs_do_make_empty_dir(struct inode *inode, struct inode *parent, 3517 struct f2fs_dentry_ptr *d); 3518 struct page *f2fs_init_inode_metadata(struct inode *inode, struct inode *dir, 3519 const struct f2fs_filename *fname, struct page *dpage); 3520 void f2fs_update_parent_metadata(struct inode *dir, struct inode *inode, 3521 unsigned int current_depth); 3522 int f2fs_room_for_filename(const void *bitmap, int slots, int max_slots); 3523 void f2fs_drop_nlink(struct inode *dir, struct inode *inode); 3524 struct f2fs_dir_entry *__f2fs_find_entry(struct inode *dir, 3525 const struct f2fs_filename *fname, 3526 struct page **res_page); 3527 struct f2fs_dir_entry *f2fs_find_entry(struct inode *dir, 3528 const struct qstr *child, struct page **res_page); 3529 struct f2fs_dir_entry *f2fs_parent_dir(struct inode *dir, struct page **p); 3530 ino_t f2fs_inode_by_name(struct inode *dir, const struct qstr *qstr, 3531 struct page **page); 3532 void f2fs_set_link(struct inode *dir, struct f2fs_dir_entry *de, 3533 struct page *page, struct inode *inode); 3534 bool f2fs_has_enough_room(struct inode *dir, struct page *ipage, 3535 const struct f2fs_filename *fname); 3536 void f2fs_update_dentry(nid_t ino, umode_t mode, struct f2fs_dentry_ptr *d, 3537 const struct fscrypt_str *name, f2fs_hash_t name_hash, 3538 unsigned int bit_pos); 3539 int f2fs_add_regular_entry(struct inode *dir, const struct f2fs_filename *fname, 3540 struct inode *inode, nid_t ino, umode_t mode); 3541 int f2fs_add_dentry(struct inode *dir, const struct f2fs_filename *fname, 3542 struct inode *inode, nid_t ino, umode_t mode); 3543 int f2fs_do_add_link(struct inode *dir, const struct qstr *name, 3544 struct inode *inode, nid_t ino, umode_t mode); 3545 void f2fs_delete_entry(struct f2fs_dir_entry *dentry, struct page *page, 3546 struct inode *dir, struct inode *inode); 3547 int f2fs_do_tmpfile(struct inode *inode, struct inode *dir); 3548 bool f2fs_empty_dir(struct inode *dir); 3549 3550 static inline int f2fs_add_link(struct dentry *dentry, struct inode *inode) 3551 { 3552 if (fscrypt_is_nokey_name(dentry)) 3553 return -ENOKEY; 3554 return f2fs_do_add_link(d_inode(dentry->d_parent), &dentry->d_name, 3555 inode, inode->i_ino, inode->i_mode); 3556 } 3557 3558 /* 3559 * super.c 3560 */ 3561 int f2fs_inode_dirtied(struct inode *inode, bool sync); 3562 void f2fs_inode_synced(struct inode *inode); 3563 int f2fs_dquot_initialize(struct inode *inode); 3564 int f2fs_enable_quota_files(struct f2fs_sb_info *sbi, bool rdonly); 3565 int f2fs_quota_sync(struct super_block *sb, int type); 3566 loff_t max_file_blocks(struct inode *inode); 3567 void f2fs_quota_off_umount(struct super_block *sb); 3568 void f2fs_save_errors(struct f2fs_sb_info *sbi, unsigned char flag); 3569 void f2fs_handle_critical_error(struct f2fs_sb_info *sbi, unsigned char reason, 3570 bool irq_context); 3571 void f2fs_handle_error(struct f2fs_sb_info *sbi, unsigned char error); 3572 void f2fs_handle_error_async(struct f2fs_sb_info *sbi, unsigned char error); 3573 int f2fs_commit_super(struct f2fs_sb_info *sbi, bool recover); 3574 int f2fs_sync_fs(struct super_block *sb, int sync); 3575 int f2fs_sanity_check_ckpt(struct f2fs_sb_info *sbi); 3576 3577 /* 3578 * hash.c 3579 */ 3580 void f2fs_hash_filename(const struct inode *dir, struct f2fs_filename *fname); 3581 3582 /* 3583 * node.c 3584 */ 3585 struct node_info; 3586 3587 int f2fs_check_nid_range(struct f2fs_sb_info *sbi, nid_t nid); 3588 bool f2fs_available_free_memory(struct f2fs_sb_info *sbi, int type); 3589 bool f2fs_in_warm_node_list(struct f2fs_sb_info *sbi, struct page *page); 3590 void f2fs_init_fsync_node_info(struct f2fs_sb_info *sbi); 3591 void f2fs_del_fsync_node_entry(struct f2fs_sb_info *sbi, struct page *page); 3592 void f2fs_reset_fsync_node_info(struct f2fs_sb_info *sbi); 3593 int f2fs_need_dentry_mark(struct f2fs_sb_info *sbi, nid_t nid); 3594 bool f2fs_is_checkpointed_node(struct f2fs_sb_info *sbi, nid_t nid); 3595 bool f2fs_need_inode_block_update(struct f2fs_sb_info *sbi, nid_t ino); 3596 int f2fs_get_node_info(struct f2fs_sb_info *sbi, nid_t nid, 3597 struct node_info *ni, bool checkpoint_context); 3598 pgoff_t f2fs_get_next_page_offset(struct dnode_of_data *dn, pgoff_t pgofs); 3599 int f2fs_get_dnode_of_data(struct dnode_of_data *dn, pgoff_t index, int mode); 3600 int f2fs_truncate_inode_blocks(struct inode *inode, pgoff_t from); 3601 int f2fs_truncate_xattr_node(struct inode *inode); 3602 int f2fs_wait_on_node_pages_writeback(struct f2fs_sb_info *sbi, 3603 unsigned int seq_id); 3604 bool f2fs_nat_bitmap_enabled(struct f2fs_sb_info *sbi); 3605 int f2fs_remove_inode_page(struct inode *inode); 3606 struct page *f2fs_new_inode_page(struct inode *inode); 3607 struct page *f2fs_new_node_page(struct dnode_of_data *dn, unsigned int ofs); 3608 void f2fs_ra_node_page(struct f2fs_sb_info *sbi, nid_t nid); 3609 struct page *f2fs_get_node_page(struct f2fs_sb_info *sbi, pgoff_t nid); 3610 struct page *f2fs_get_node_page_ra(struct page *parent, int start); 3611 int f2fs_move_node_page(struct page *node_page, int gc_type); 3612 void f2fs_flush_inline_data(struct f2fs_sb_info *sbi); 3613 int f2fs_fsync_node_pages(struct f2fs_sb_info *sbi, struct inode *inode, 3614 struct writeback_control *wbc, bool atomic, 3615 unsigned int *seq_id); 3616 int f2fs_sync_node_pages(struct f2fs_sb_info *sbi, 3617 struct writeback_control *wbc, 3618 bool do_balance, enum iostat_type io_type); 3619 int f2fs_build_free_nids(struct f2fs_sb_info *sbi, bool sync, bool mount); 3620 bool f2fs_alloc_nid(struct f2fs_sb_info *sbi, nid_t *nid); 3621 void f2fs_alloc_nid_done(struct f2fs_sb_info *sbi, nid_t nid); 3622 void f2fs_alloc_nid_failed(struct f2fs_sb_info *sbi, nid_t nid); 3623 int f2fs_try_to_free_nids(struct f2fs_sb_info *sbi, int nr_shrink); 3624 int f2fs_recover_inline_xattr(struct inode *inode, struct page *page); 3625 int f2fs_recover_xattr_data(struct inode *inode, struct page *page); 3626 int f2fs_recover_inode_page(struct f2fs_sb_info *sbi, struct page *page); 3627 int f2fs_restore_node_summary(struct f2fs_sb_info *sbi, 3628 unsigned int segno, struct f2fs_summary_block *sum); 3629 void f2fs_enable_nat_bits(struct f2fs_sb_info *sbi); 3630 int f2fs_flush_nat_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3631 int f2fs_build_node_manager(struct f2fs_sb_info *sbi); 3632 void f2fs_destroy_node_manager(struct f2fs_sb_info *sbi); 3633 int __init f2fs_create_node_manager_caches(void); 3634 void f2fs_destroy_node_manager_caches(void); 3635 3636 /* 3637 * segment.c 3638 */ 3639 bool f2fs_need_SSR(struct f2fs_sb_info *sbi); 3640 int f2fs_commit_atomic_write(struct inode *inode); 3641 void f2fs_abort_atomic_write(struct inode *inode, bool clean); 3642 void f2fs_balance_fs(struct f2fs_sb_info *sbi, bool need); 3643 void f2fs_balance_fs_bg(struct f2fs_sb_info *sbi, bool from_bg); 3644 int f2fs_issue_flush(struct f2fs_sb_info *sbi, nid_t ino); 3645 int f2fs_create_flush_cmd_control(struct f2fs_sb_info *sbi); 3646 int f2fs_flush_device_cache(struct f2fs_sb_info *sbi); 3647 void f2fs_destroy_flush_cmd_control(struct f2fs_sb_info *sbi, bool free); 3648 void f2fs_invalidate_blocks(struct f2fs_sb_info *sbi, block_t addr); 3649 bool f2fs_is_checkpointed_data(struct f2fs_sb_info *sbi, block_t blkaddr); 3650 int f2fs_start_discard_thread(struct f2fs_sb_info *sbi); 3651 void f2fs_drop_discard_cmd(struct f2fs_sb_info *sbi); 3652 void f2fs_stop_discard_thread(struct f2fs_sb_info *sbi); 3653 bool f2fs_issue_discard_timeout(struct f2fs_sb_info *sbi); 3654 void f2fs_clear_prefree_segments(struct f2fs_sb_info *sbi, 3655 struct cp_control *cpc); 3656 void f2fs_dirty_to_prefree(struct f2fs_sb_info *sbi); 3657 block_t f2fs_get_unusable_blocks(struct f2fs_sb_info *sbi); 3658 int f2fs_disable_cp_again(struct f2fs_sb_info *sbi, block_t unusable); 3659 void f2fs_release_discard_addrs(struct f2fs_sb_info *sbi); 3660 int f2fs_npages_for_summary_flush(struct f2fs_sb_info *sbi, bool for_ra); 3661 bool f2fs_segment_has_free_slot(struct f2fs_sb_info *sbi, int segno); 3662 void f2fs_init_inmem_curseg(struct f2fs_sb_info *sbi); 3663 void f2fs_save_inmem_curseg(struct f2fs_sb_info *sbi); 3664 void f2fs_restore_inmem_curseg(struct f2fs_sb_info *sbi); 3665 void f2fs_get_new_segment(struct f2fs_sb_info *sbi, 3666 unsigned int *newseg, bool new_sec, int dir); 3667 void f2fs_allocate_segment_for_resize(struct f2fs_sb_info *sbi, int type, 3668 unsigned int start, unsigned int end); 3669 void f2fs_allocate_new_section(struct f2fs_sb_info *sbi, int type, bool force); 3670 void f2fs_allocate_new_segments(struct f2fs_sb_info *sbi); 3671 int f2fs_trim_fs(struct f2fs_sb_info *sbi, struct fstrim_range *range); 3672 bool f2fs_exist_trim_candidates(struct f2fs_sb_info *sbi, 3673 struct cp_control *cpc); 3674 struct page *f2fs_get_sum_page(struct f2fs_sb_info *sbi, unsigned int segno); 3675 void f2fs_update_meta_page(struct f2fs_sb_info *sbi, void *src, 3676 block_t blk_addr); 3677 void f2fs_do_write_meta_page(struct f2fs_sb_info *sbi, struct page *page, 3678 enum iostat_type io_type); 3679 void f2fs_do_write_node_page(unsigned int nid, struct f2fs_io_info *fio); 3680 void f2fs_outplace_write_data(struct dnode_of_data *dn, 3681 struct f2fs_io_info *fio); 3682 int f2fs_inplace_write_data(struct f2fs_io_info *fio); 3683 void f2fs_do_replace_block(struct f2fs_sb_info *sbi, struct f2fs_summary *sum, 3684 block_t old_blkaddr, block_t new_blkaddr, 3685 bool recover_curseg, bool recover_newaddr, 3686 bool from_gc); 3687 void f2fs_replace_block(struct f2fs_sb_info *sbi, struct dnode_of_data *dn, 3688 block_t old_addr, block_t new_addr, 3689 unsigned char version, bool recover_curseg, 3690 bool recover_newaddr); 3691 void f2fs_allocate_data_block(struct f2fs_sb_info *sbi, struct page *page, 3692 block_t old_blkaddr, block_t *new_blkaddr, 3693 struct f2fs_summary *sum, int type, 3694 struct f2fs_io_info *fio); 3695 void f2fs_update_device_state(struct f2fs_sb_info *sbi, nid_t ino, 3696 block_t blkaddr, unsigned int blkcnt); 3697 void f2fs_wait_on_page_writeback(struct page *page, 3698 enum page_type type, bool ordered, bool locked); 3699 void f2fs_wait_on_block_writeback(struct inode *inode, block_t blkaddr); 3700 void f2fs_wait_on_block_writeback_range(struct inode *inode, block_t blkaddr, 3701 block_t len); 3702 void f2fs_write_data_summaries(struct f2fs_sb_info *sbi, block_t start_blk); 3703 void f2fs_write_node_summaries(struct f2fs_sb_info *sbi, block_t start_blk); 3704 int f2fs_lookup_journal_in_cursum(struct f2fs_journal *journal, int type, 3705 unsigned int val, int alloc); 3706 void f2fs_flush_sit_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3707 int f2fs_fix_curseg_write_pointer(struct f2fs_sb_info *sbi); 3708 int f2fs_check_write_pointer(struct f2fs_sb_info *sbi); 3709 int f2fs_build_segment_manager(struct f2fs_sb_info *sbi); 3710 void f2fs_destroy_segment_manager(struct f2fs_sb_info *sbi); 3711 int __init f2fs_create_segment_manager_caches(void); 3712 void f2fs_destroy_segment_manager_caches(void); 3713 int f2fs_rw_hint_to_seg_type(enum rw_hint hint); 3714 unsigned int f2fs_usable_segs_in_sec(struct f2fs_sb_info *sbi, 3715 unsigned int segno); 3716 unsigned int f2fs_usable_blks_in_seg(struct f2fs_sb_info *sbi, 3717 unsigned int segno); 3718 3719 #define DEF_FRAGMENT_SIZE 4 3720 #define MIN_FRAGMENT_SIZE 1 3721 #define MAX_FRAGMENT_SIZE 512 3722 3723 static inline bool f2fs_need_rand_seg(struct f2fs_sb_info *sbi) 3724 { 3725 return F2FS_OPTION(sbi).fs_mode == FS_MODE_FRAGMENT_SEG || 3726 F2FS_OPTION(sbi).fs_mode == FS_MODE_FRAGMENT_BLK; 3727 } 3728 3729 /* 3730 * checkpoint.c 3731 */ 3732 void f2fs_stop_checkpoint(struct f2fs_sb_info *sbi, bool end_io, 3733 unsigned char reason); 3734 void f2fs_flush_ckpt_thread(struct f2fs_sb_info *sbi); 3735 struct page *f2fs_grab_meta_page(struct f2fs_sb_info *sbi, pgoff_t index); 3736 struct page *f2fs_get_meta_page(struct f2fs_sb_info *sbi, pgoff_t index); 3737 struct page *f2fs_get_meta_page_retry(struct f2fs_sb_info *sbi, pgoff_t index); 3738 struct page *f2fs_get_tmp_page(struct f2fs_sb_info *sbi, pgoff_t index); 3739 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi, 3740 block_t blkaddr, int type); 3741 int f2fs_ra_meta_pages(struct f2fs_sb_info *sbi, block_t start, int nrpages, 3742 int type, bool sync); 3743 void f2fs_ra_meta_pages_cond(struct f2fs_sb_info *sbi, pgoff_t index, 3744 unsigned int ra_blocks); 3745 long f2fs_sync_meta_pages(struct f2fs_sb_info *sbi, enum page_type type, 3746 long nr_to_write, enum iostat_type io_type); 3747 void f2fs_add_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type); 3748 void f2fs_remove_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type); 3749 void f2fs_release_ino_entry(struct f2fs_sb_info *sbi, bool all); 3750 bool f2fs_exist_written_data(struct f2fs_sb_info *sbi, nid_t ino, int mode); 3751 void f2fs_set_dirty_device(struct f2fs_sb_info *sbi, nid_t ino, 3752 unsigned int devidx, int type); 3753 bool f2fs_is_dirty_device(struct f2fs_sb_info *sbi, nid_t ino, 3754 unsigned int devidx, int type); 3755 int f2fs_acquire_orphan_inode(struct f2fs_sb_info *sbi); 3756 void f2fs_release_orphan_inode(struct f2fs_sb_info *sbi); 3757 void f2fs_add_orphan_inode(struct inode *inode); 3758 void f2fs_remove_orphan_inode(struct f2fs_sb_info *sbi, nid_t ino); 3759 int f2fs_recover_orphan_inodes(struct f2fs_sb_info *sbi); 3760 int f2fs_get_valid_checkpoint(struct f2fs_sb_info *sbi); 3761 void f2fs_update_dirty_folio(struct inode *inode, struct folio *folio); 3762 void f2fs_remove_dirty_inode(struct inode *inode); 3763 int f2fs_sync_dirty_inodes(struct f2fs_sb_info *sbi, enum inode_type type, 3764 bool from_cp); 3765 void f2fs_wait_on_all_pages(struct f2fs_sb_info *sbi, int type); 3766 u64 f2fs_get_sectors_written(struct f2fs_sb_info *sbi); 3767 int f2fs_write_checkpoint(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3768 void f2fs_init_ino_entry_info(struct f2fs_sb_info *sbi); 3769 int __init f2fs_create_checkpoint_caches(void); 3770 void f2fs_destroy_checkpoint_caches(void); 3771 int f2fs_issue_checkpoint(struct f2fs_sb_info *sbi); 3772 int f2fs_start_ckpt_thread(struct f2fs_sb_info *sbi); 3773 void f2fs_stop_ckpt_thread(struct f2fs_sb_info *sbi); 3774 void f2fs_init_ckpt_req_control(struct f2fs_sb_info *sbi); 3775 3776 /* 3777 * data.c 3778 */ 3779 int __init f2fs_init_bioset(void); 3780 void f2fs_destroy_bioset(void); 3781 int f2fs_init_bio_entry_cache(void); 3782 void f2fs_destroy_bio_entry_cache(void); 3783 void f2fs_submit_read_bio(struct f2fs_sb_info *sbi, struct bio *bio, 3784 enum page_type type); 3785 int f2fs_init_write_merge_io(struct f2fs_sb_info *sbi); 3786 void f2fs_submit_merged_write(struct f2fs_sb_info *sbi, enum page_type type); 3787 void f2fs_submit_merged_write_cond(struct f2fs_sb_info *sbi, 3788 struct inode *inode, struct page *page, 3789 nid_t ino, enum page_type type); 3790 void f2fs_submit_merged_ipu_write(struct f2fs_sb_info *sbi, 3791 struct bio **bio, struct page *page); 3792 void f2fs_flush_merged_writes(struct f2fs_sb_info *sbi); 3793 int f2fs_submit_page_bio(struct f2fs_io_info *fio); 3794 int f2fs_merge_page_bio(struct f2fs_io_info *fio); 3795 void f2fs_submit_page_write(struct f2fs_io_info *fio); 3796 struct block_device *f2fs_target_device(struct f2fs_sb_info *sbi, 3797 block_t blk_addr, sector_t *sector); 3798 int f2fs_target_device_index(struct f2fs_sb_info *sbi, block_t blkaddr); 3799 void f2fs_set_data_blkaddr(struct dnode_of_data *dn); 3800 void f2fs_update_data_blkaddr(struct dnode_of_data *dn, block_t blkaddr); 3801 int f2fs_reserve_new_blocks(struct dnode_of_data *dn, blkcnt_t count); 3802 int f2fs_reserve_new_block(struct dnode_of_data *dn); 3803 int f2fs_get_block_locked(struct dnode_of_data *dn, pgoff_t index); 3804 int f2fs_reserve_block(struct dnode_of_data *dn, pgoff_t index); 3805 struct page *f2fs_get_read_data_page(struct inode *inode, pgoff_t index, 3806 blk_opf_t op_flags, bool for_write, pgoff_t *next_pgofs); 3807 struct page *f2fs_find_data_page(struct inode *inode, pgoff_t index, 3808 pgoff_t *next_pgofs); 3809 struct page *f2fs_get_lock_data_page(struct inode *inode, pgoff_t index, 3810 bool for_write); 3811 struct page *f2fs_get_new_data_page(struct inode *inode, 3812 struct page *ipage, pgoff_t index, bool new_i_size); 3813 int f2fs_do_write_data_page(struct f2fs_io_info *fio); 3814 int f2fs_map_blocks(struct inode *inode, struct f2fs_map_blocks *map, int flag); 3815 int f2fs_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, 3816 u64 start, u64 len); 3817 int f2fs_encrypt_one_page(struct f2fs_io_info *fio); 3818 bool f2fs_should_update_inplace(struct inode *inode, struct f2fs_io_info *fio); 3819 bool f2fs_should_update_outplace(struct inode *inode, struct f2fs_io_info *fio); 3820 int f2fs_write_single_data_page(struct page *page, int *submitted, 3821 struct bio **bio, sector_t *last_block, 3822 struct writeback_control *wbc, 3823 enum iostat_type io_type, 3824 int compr_blocks, bool allow_balance); 3825 void f2fs_write_failed(struct inode *inode, loff_t to); 3826 void f2fs_invalidate_folio(struct folio *folio, size_t offset, size_t length); 3827 bool f2fs_release_folio(struct folio *folio, gfp_t wait); 3828 bool f2fs_overwrite_io(struct inode *inode, loff_t pos, size_t len); 3829 void f2fs_clear_page_cache_dirty_tag(struct page *page); 3830 int f2fs_init_post_read_processing(void); 3831 void f2fs_destroy_post_read_processing(void); 3832 int f2fs_init_post_read_wq(struct f2fs_sb_info *sbi); 3833 void f2fs_destroy_post_read_wq(struct f2fs_sb_info *sbi); 3834 extern const struct iomap_ops f2fs_iomap_ops; 3835 3836 /* 3837 * gc.c 3838 */ 3839 int f2fs_start_gc_thread(struct f2fs_sb_info *sbi); 3840 void f2fs_stop_gc_thread(struct f2fs_sb_info *sbi); 3841 block_t f2fs_start_bidx_of_node(unsigned int node_ofs, struct inode *inode); 3842 int f2fs_gc(struct f2fs_sb_info *sbi, struct f2fs_gc_control *gc_control); 3843 void f2fs_build_gc_manager(struct f2fs_sb_info *sbi); 3844 int f2fs_resize_fs(struct file *filp, __u64 block_count); 3845 int __init f2fs_create_garbage_collection_cache(void); 3846 void f2fs_destroy_garbage_collection_cache(void); 3847 /* victim selection function for cleaning and SSR */ 3848 int f2fs_get_victim(struct f2fs_sb_info *sbi, unsigned int *result, 3849 int gc_type, int type, char alloc_mode, 3850 unsigned long long age); 3851 3852 /* 3853 * recovery.c 3854 */ 3855 int f2fs_recover_fsync_data(struct f2fs_sb_info *sbi, bool check_only); 3856 bool f2fs_space_for_roll_forward(struct f2fs_sb_info *sbi); 3857 int __init f2fs_create_recovery_cache(void); 3858 void f2fs_destroy_recovery_cache(void); 3859 3860 /* 3861 * debug.c 3862 */ 3863 #ifdef CONFIG_F2FS_STAT_FS 3864 struct f2fs_stat_info { 3865 struct list_head stat_list; 3866 struct f2fs_sb_info *sbi; 3867 int all_area_segs, sit_area_segs, nat_area_segs, ssa_area_segs; 3868 int main_area_segs, main_area_sections, main_area_zones; 3869 unsigned long long hit_cached[NR_EXTENT_CACHES]; 3870 unsigned long long hit_rbtree[NR_EXTENT_CACHES]; 3871 unsigned long long total_ext[NR_EXTENT_CACHES]; 3872 unsigned long long hit_total[NR_EXTENT_CACHES]; 3873 int ext_tree[NR_EXTENT_CACHES]; 3874 int zombie_tree[NR_EXTENT_CACHES]; 3875 int ext_node[NR_EXTENT_CACHES]; 3876 /* to count memory footprint */ 3877 unsigned long long ext_mem[NR_EXTENT_CACHES]; 3878 /* for read extent cache */ 3879 unsigned long long hit_largest; 3880 /* for block age extent cache */ 3881 unsigned long long allocated_data_blocks; 3882 int ndirty_node, ndirty_dent, ndirty_meta, ndirty_imeta; 3883 int ndirty_data, ndirty_qdata; 3884 unsigned int ndirty_dirs, ndirty_files, nquota_files, ndirty_all; 3885 int nats, dirty_nats, sits, dirty_sits; 3886 int free_nids, avail_nids, alloc_nids; 3887 int total_count, utilization; 3888 int bg_gc, nr_wb_cp_data, nr_wb_data; 3889 int nr_rd_data, nr_rd_node, nr_rd_meta; 3890 int nr_dio_read, nr_dio_write; 3891 unsigned int io_skip_bggc, other_skip_bggc; 3892 int nr_flushing, nr_flushed, flush_list_empty; 3893 int nr_discarding, nr_discarded; 3894 int nr_discard_cmd; 3895 unsigned int undiscard_blks; 3896 int nr_issued_ckpt, nr_total_ckpt, nr_queued_ckpt; 3897 unsigned int cur_ckpt_time, peak_ckpt_time; 3898 int inline_xattr, inline_inode, inline_dir, append, update, orphans; 3899 int compr_inode, swapfile_inode; 3900 unsigned long long compr_blocks; 3901 int aw_cnt, max_aw_cnt; 3902 unsigned int valid_count, valid_node_count, valid_inode_count, discard_blks; 3903 unsigned int bimodal, avg_vblocks; 3904 int util_free, util_valid, util_invalid; 3905 int rsvd_segs, overp_segs; 3906 int dirty_count, node_pages, meta_pages, compress_pages; 3907 int compress_page_hit; 3908 int prefree_count, call_count, cp_count, bg_cp_count; 3909 int tot_segs, node_segs, data_segs, free_segs, free_secs; 3910 int bg_node_segs, bg_data_segs; 3911 int tot_blks, data_blks, node_blks; 3912 int bg_data_blks, bg_node_blks; 3913 int curseg[NR_CURSEG_TYPE]; 3914 int cursec[NR_CURSEG_TYPE]; 3915 int curzone[NR_CURSEG_TYPE]; 3916 unsigned int dirty_seg[NR_CURSEG_TYPE]; 3917 unsigned int full_seg[NR_CURSEG_TYPE]; 3918 unsigned int valid_blks[NR_CURSEG_TYPE]; 3919 3920 unsigned int meta_count[META_MAX]; 3921 unsigned int segment_count[2]; 3922 unsigned int block_count[2]; 3923 unsigned int inplace_count; 3924 unsigned long long base_mem, cache_mem, page_mem; 3925 }; 3926 3927 static inline struct f2fs_stat_info *F2FS_STAT(struct f2fs_sb_info *sbi) 3928 { 3929 return (struct f2fs_stat_info *)sbi->stat_info; 3930 } 3931 3932 #define stat_inc_cp_count(si) ((si)->cp_count++) 3933 #define stat_inc_bg_cp_count(si) ((si)->bg_cp_count++) 3934 #define stat_inc_call_count(si) ((si)->call_count++) 3935 #define stat_inc_bggc_count(si) ((si)->bg_gc++) 3936 #define stat_io_skip_bggc_count(sbi) ((sbi)->io_skip_bggc++) 3937 #define stat_other_skip_bggc_count(sbi) ((sbi)->other_skip_bggc++) 3938 #define stat_inc_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]++) 3939 #define stat_dec_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]--) 3940 #define stat_inc_total_hit(sbi, type) (atomic64_inc(&(sbi)->total_hit_ext[type])) 3941 #define stat_inc_rbtree_node_hit(sbi, type) (atomic64_inc(&(sbi)->read_hit_rbtree[type])) 3942 #define stat_inc_largest_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_largest)) 3943 #define stat_inc_cached_node_hit(sbi, type) (atomic64_inc(&(sbi)->read_hit_cached[type])) 3944 #define stat_inc_inline_xattr(inode) \ 3945 do { \ 3946 if (f2fs_has_inline_xattr(inode)) \ 3947 (atomic_inc(&F2FS_I_SB(inode)->inline_xattr)); \ 3948 } while (0) 3949 #define stat_dec_inline_xattr(inode) \ 3950 do { \ 3951 if (f2fs_has_inline_xattr(inode)) \ 3952 (atomic_dec(&F2FS_I_SB(inode)->inline_xattr)); \ 3953 } while (0) 3954 #define stat_inc_inline_inode(inode) \ 3955 do { \ 3956 if (f2fs_has_inline_data(inode)) \ 3957 (atomic_inc(&F2FS_I_SB(inode)->inline_inode)); \ 3958 } while (0) 3959 #define stat_dec_inline_inode(inode) \ 3960 do { \ 3961 if (f2fs_has_inline_data(inode)) \ 3962 (atomic_dec(&F2FS_I_SB(inode)->inline_inode)); \ 3963 } while (0) 3964 #define stat_inc_inline_dir(inode) \ 3965 do { \ 3966 if (f2fs_has_inline_dentry(inode)) \ 3967 (atomic_inc(&F2FS_I_SB(inode)->inline_dir)); \ 3968 } while (0) 3969 #define stat_dec_inline_dir(inode) \ 3970 do { \ 3971 if (f2fs_has_inline_dentry(inode)) \ 3972 (atomic_dec(&F2FS_I_SB(inode)->inline_dir)); \ 3973 } while (0) 3974 #define stat_inc_compr_inode(inode) \ 3975 do { \ 3976 if (f2fs_compressed_file(inode)) \ 3977 (atomic_inc(&F2FS_I_SB(inode)->compr_inode)); \ 3978 } while (0) 3979 #define stat_dec_compr_inode(inode) \ 3980 do { \ 3981 if (f2fs_compressed_file(inode)) \ 3982 (atomic_dec(&F2FS_I_SB(inode)->compr_inode)); \ 3983 } while (0) 3984 #define stat_add_compr_blocks(inode, blocks) \ 3985 (atomic64_add(blocks, &F2FS_I_SB(inode)->compr_blocks)) 3986 #define stat_sub_compr_blocks(inode, blocks) \ 3987 (atomic64_sub(blocks, &F2FS_I_SB(inode)->compr_blocks)) 3988 #define stat_inc_swapfile_inode(inode) \ 3989 (atomic_inc(&F2FS_I_SB(inode)->swapfile_inode)) 3990 #define stat_dec_swapfile_inode(inode) \ 3991 (atomic_dec(&F2FS_I_SB(inode)->swapfile_inode)) 3992 #define stat_inc_atomic_inode(inode) \ 3993 (atomic_inc(&F2FS_I_SB(inode)->atomic_files)) 3994 #define stat_dec_atomic_inode(inode) \ 3995 (atomic_dec(&F2FS_I_SB(inode)->atomic_files)) 3996 #define stat_inc_meta_count(sbi, blkaddr) \ 3997 do { \ 3998 if (blkaddr < SIT_I(sbi)->sit_base_addr) \ 3999 atomic_inc(&(sbi)->meta_count[META_CP]); \ 4000 else if (blkaddr < NM_I(sbi)->nat_blkaddr) \ 4001 atomic_inc(&(sbi)->meta_count[META_SIT]); \ 4002 else if (blkaddr < SM_I(sbi)->ssa_blkaddr) \ 4003 atomic_inc(&(sbi)->meta_count[META_NAT]); \ 4004 else if (blkaddr < SM_I(sbi)->main_blkaddr) \ 4005 atomic_inc(&(sbi)->meta_count[META_SSA]); \ 4006 } while (0) 4007 #define stat_inc_seg_type(sbi, curseg) \ 4008 ((sbi)->segment_count[(curseg)->alloc_type]++) 4009 #define stat_inc_block_count(sbi, curseg) \ 4010 ((sbi)->block_count[(curseg)->alloc_type]++) 4011 #define stat_inc_inplace_blocks(sbi) \ 4012 (atomic_inc(&(sbi)->inplace_count)) 4013 #define stat_update_max_atomic_write(inode) \ 4014 do { \ 4015 int cur = atomic_read(&F2FS_I_SB(inode)->atomic_files); \ 4016 int max = atomic_read(&F2FS_I_SB(inode)->max_aw_cnt); \ 4017 if (cur > max) \ 4018 atomic_set(&F2FS_I_SB(inode)->max_aw_cnt, cur); \ 4019 } while (0) 4020 #define stat_inc_seg_count(sbi, type, gc_type) \ 4021 do { \ 4022 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 4023 si->tot_segs++; \ 4024 if ((type) == SUM_TYPE_DATA) { \ 4025 si->data_segs++; \ 4026 si->bg_data_segs += (gc_type == BG_GC) ? 1 : 0; \ 4027 } else { \ 4028 si->node_segs++; \ 4029 si->bg_node_segs += (gc_type == BG_GC) ? 1 : 0; \ 4030 } \ 4031 } while (0) 4032 4033 #define stat_inc_tot_blk_count(si, blks) \ 4034 ((si)->tot_blks += (blks)) 4035 4036 #define stat_inc_data_blk_count(sbi, blks, gc_type) \ 4037 do { \ 4038 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 4039 stat_inc_tot_blk_count(si, blks); \ 4040 si->data_blks += (blks); \ 4041 si->bg_data_blks += ((gc_type) == BG_GC) ? (blks) : 0; \ 4042 } while (0) 4043 4044 #define stat_inc_node_blk_count(sbi, blks, gc_type) \ 4045 do { \ 4046 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 4047 stat_inc_tot_blk_count(si, blks); \ 4048 si->node_blks += (blks); \ 4049 si->bg_node_blks += ((gc_type) == BG_GC) ? (blks) : 0; \ 4050 } while (0) 4051 4052 int f2fs_build_stats(struct f2fs_sb_info *sbi); 4053 void f2fs_destroy_stats(struct f2fs_sb_info *sbi); 4054 void __init f2fs_create_root_stats(void); 4055 void f2fs_destroy_root_stats(void); 4056 void f2fs_update_sit_info(struct f2fs_sb_info *sbi); 4057 #else 4058 #define stat_inc_cp_count(si) do { } while (0) 4059 #define stat_inc_bg_cp_count(si) do { } while (0) 4060 #define stat_inc_call_count(si) do { } while (0) 4061 #define stat_inc_bggc_count(si) do { } while (0) 4062 #define stat_io_skip_bggc_count(sbi) do { } while (0) 4063 #define stat_other_skip_bggc_count(sbi) do { } while (0) 4064 #define stat_inc_dirty_inode(sbi, type) do { } while (0) 4065 #define stat_dec_dirty_inode(sbi, type) do { } while (0) 4066 #define stat_inc_total_hit(sbi, type) do { } while (0) 4067 #define stat_inc_rbtree_node_hit(sbi, type) do { } while (0) 4068 #define stat_inc_largest_node_hit(sbi) do { } while (0) 4069 #define stat_inc_cached_node_hit(sbi, type) do { } while (0) 4070 #define stat_inc_inline_xattr(inode) do { } while (0) 4071 #define stat_dec_inline_xattr(inode) do { } while (0) 4072 #define stat_inc_inline_inode(inode) do { } while (0) 4073 #define stat_dec_inline_inode(inode) do { } while (0) 4074 #define stat_inc_inline_dir(inode) do { } while (0) 4075 #define stat_dec_inline_dir(inode) do { } while (0) 4076 #define stat_inc_compr_inode(inode) do { } while (0) 4077 #define stat_dec_compr_inode(inode) do { } while (0) 4078 #define stat_add_compr_blocks(inode, blocks) do { } while (0) 4079 #define stat_sub_compr_blocks(inode, blocks) do { } while (0) 4080 #define stat_inc_swapfile_inode(inode) do { } while (0) 4081 #define stat_dec_swapfile_inode(inode) do { } while (0) 4082 #define stat_inc_atomic_inode(inode) do { } while (0) 4083 #define stat_dec_atomic_inode(inode) do { } while (0) 4084 #define stat_update_max_atomic_write(inode) do { } while (0) 4085 #define stat_inc_meta_count(sbi, blkaddr) do { } while (0) 4086 #define stat_inc_seg_type(sbi, curseg) do { } while (0) 4087 #define stat_inc_block_count(sbi, curseg) do { } while (0) 4088 #define stat_inc_inplace_blocks(sbi) do { } while (0) 4089 #define stat_inc_seg_count(sbi, type, gc_type) do { } while (0) 4090 #define stat_inc_tot_blk_count(si, blks) do { } while (0) 4091 #define stat_inc_data_blk_count(sbi, blks, gc_type) do { } while (0) 4092 #define stat_inc_node_blk_count(sbi, blks, gc_type) do { } while (0) 4093 4094 static inline int f2fs_build_stats(struct f2fs_sb_info *sbi) { return 0; } 4095 static inline void f2fs_destroy_stats(struct f2fs_sb_info *sbi) { } 4096 static inline void __init f2fs_create_root_stats(void) { } 4097 static inline void f2fs_destroy_root_stats(void) { } 4098 static inline void f2fs_update_sit_info(struct f2fs_sb_info *sbi) {} 4099 #endif 4100 4101 extern const struct file_operations f2fs_dir_operations; 4102 extern const struct file_operations f2fs_file_operations; 4103 extern const struct inode_operations f2fs_file_inode_operations; 4104 extern const struct address_space_operations f2fs_dblock_aops; 4105 extern const struct address_space_operations f2fs_node_aops; 4106 extern const struct address_space_operations f2fs_meta_aops; 4107 extern const struct inode_operations f2fs_dir_inode_operations; 4108 extern const struct inode_operations f2fs_symlink_inode_operations; 4109 extern const struct inode_operations f2fs_encrypted_symlink_inode_operations; 4110 extern const struct inode_operations f2fs_special_inode_operations; 4111 extern struct kmem_cache *f2fs_inode_entry_slab; 4112 4113 /* 4114 * inline.c 4115 */ 4116 bool f2fs_may_inline_data(struct inode *inode); 4117 bool f2fs_sanity_check_inline_data(struct inode *inode); 4118 bool f2fs_may_inline_dentry(struct inode *inode); 4119 void f2fs_do_read_inline_data(struct page *page, struct page *ipage); 4120 void f2fs_truncate_inline_inode(struct inode *inode, 4121 struct page *ipage, u64 from); 4122 int f2fs_read_inline_data(struct inode *inode, struct page *page); 4123 int f2fs_convert_inline_page(struct dnode_of_data *dn, struct page *page); 4124 int f2fs_convert_inline_inode(struct inode *inode); 4125 int f2fs_try_convert_inline_dir(struct inode *dir, struct dentry *dentry); 4126 int f2fs_write_inline_data(struct inode *inode, struct page *page); 4127 int f2fs_recover_inline_data(struct inode *inode, struct page *npage); 4128 struct f2fs_dir_entry *f2fs_find_in_inline_dir(struct inode *dir, 4129 const struct f2fs_filename *fname, 4130 struct page **res_page); 4131 int f2fs_make_empty_inline_dir(struct inode *inode, struct inode *parent, 4132 struct page *ipage); 4133 int f2fs_add_inline_entry(struct inode *dir, const struct f2fs_filename *fname, 4134 struct inode *inode, nid_t ino, umode_t mode); 4135 void f2fs_delete_inline_entry(struct f2fs_dir_entry *dentry, 4136 struct page *page, struct inode *dir, 4137 struct inode *inode); 4138 bool f2fs_empty_inline_dir(struct inode *dir); 4139 int f2fs_read_inline_dir(struct file *file, struct dir_context *ctx, 4140 struct fscrypt_str *fstr); 4141 int f2fs_inline_data_fiemap(struct inode *inode, 4142 struct fiemap_extent_info *fieinfo, 4143 __u64 start, __u64 len); 4144 4145 /* 4146 * shrinker.c 4147 */ 4148 unsigned long f2fs_shrink_count(struct shrinker *shrink, 4149 struct shrink_control *sc); 4150 unsigned long f2fs_shrink_scan(struct shrinker *shrink, 4151 struct shrink_control *sc); 4152 void f2fs_join_shrinker(struct f2fs_sb_info *sbi); 4153 void f2fs_leave_shrinker(struct f2fs_sb_info *sbi); 4154 4155 /* 4156 * extent_cache.c 4157 */ 4158 bool sanity_check_extent_cache(struct inode *inode); 4159 void f2fs_init_extent_tree(struct inode *inode); 4160 void f2fs_drop_extent_tree(struct inode *inode); 4161 void f2fs_destroy_extent_node(struct inode *inode); 4162 void f2fs_destroy_extent_tree(struct inode *inode); 4163 void f2fs_init_extent_cache_info(struct f2fs_sb_info *sbi); 4164 int __init f2fs_create_extent_cache(void); 4165 void f2fs_destroy_extent_cache(void); 4166 4167 /* read extent cache ops */ 4168 void f2fs_init_read_extent_tree(struct inode *inode, struct page *ipage); 4169 bool f2fs_lookup_read_extent_cache(struct inode *inode, pgoff_t pgofs, 4170 struct extent_info *ei); 4171 bool f2fs_lookup_read_extent_cache_block(struct inode *inode, pgoff_t index, 4172 block_t *blkaddr); 4173 void f2fs_update_read_extent_cache(struct dnode_of_data *dn); 4174 void f2fs_update_read_extent_cache_range(struct dnode_of_data *dn, 4175 pgoff_t fofs, block_t blkaddr, unsigned int len); 4176 unsigned int f2fs_shrink_read_extent_tree(struct f2fs_sb_info *sbi, 4177 int nr_shrink); 4178 4179 /* block age extent cache ops */ 4180 void f2fs_init_age_extent_tree(struct inode *inode); 4181 bool f2fs_lookup_age_extent_cache(struct inode *inode, pgoff_t pgofs, 4182 struct extent_info *ei); 4183 void f2fs_update_age_extent_cache(struct dnode_of_data *dn); 4184 void f2fs_update_age_extent_cache_range(struct dnode_of_data *dn, 4185 pgoff_t fofs, unsigned int len); 4186 unsigned int f2fs_shrink_age_extent_tree(struct f2fs_sb_info *sbi, 4187 int nr_shrink); 4188 4189 /* 4190 * sysfs.c 4191 */ 4192 #define MIN_RA_MUL 2 4193 #define MAX_RA_MUL 256 4194 4195 int __init f2fs_init_sysfs(void); 4196 void f2fs_exit_sysfs(void); 4197 int f2fs_register_sysfs(struct f2fs_sb_info *sbi); 4198 void f2fs_unregister_sysfs(struct f2fs_sb_info *sbi); 4199 4200 /* verity.c */ 4201 extern const struct fsverity_operations f2fs_verityops; 4202 4203 /* 4204 * crypto support 4205 */ 4206 static inline bool f2fs_encrypted_file(struct inode *inode) 4207 { 4208 return IS_ENCRYPTED(inode) && S_ISREG(inode->i_mode); 4209 } 4210 4211 static inline void f2fs_set_encrypted_inode(struct inode *inode) 4212 { 4213 #ifdef CONFIG_FS_ENCRYPTION 4214 file_set_encrypt(inode); 4215 f2fs_set_inode_flags(inode); 4216 #endif 4217 } 4218 4219 /* 4220 * Returns true if the reads of the inode's data need to undergo some 4221 * postprocessing step, like decryption or authenticity verification. 4222 */ 4223 static inline bool f2fs_post_read_required(struct inode *inode) 4224 { 4225 return f2fs_encrypted_file(inode) || fsverity_active(inode) || 4226 f2fs_compressed_file(inode); 4227 } 4228 4229 /* 4230 * compress.c 4231 */ 4232 #ifdef CONFIG_F2FS_FS_COMPRESSION 4233 bool f2fs_is_compressed_page(struct page *page); 4234 struct page *f2fs_compress_control_page(struct page *page); 4235 int f2fs_prepare_compress_overwrite(struct inode *inode, 4236 struct page **pagep, pgoff_t index, void **fsdata); 4237 bool f2fs_compress_write_end(struct inode *inode, void *fsdata, 4238 pgoff_t index, unsigned copied); 4239 int f2fs_truncate_partial_cluster(struct inode *inode, u64 from, bool lock); 4240 void f2fs_compress_write_end_io(struct bio *bio, struct page *page); 4241 bool f2fs_is_compress_backend_ready(struct inode *inode); 4242 bool f2fs_is_compress_level_valid(int alg, int lvl); 4243 int __init f2fs_init_compress_mempool(void); 4244 void f2fs_destroy_compress_mempool(void); 4245 void f2fs_decompress_cluster(struct decompress_io_ctx *dic, bool in_task); 4246 void f2fs_end_read_compressed_page(struct page *page, bool failed, 4247 block_t blkaddr, bool in_task); 4248 bool f2fs_cluster_is_empty(struct compress_ctx *cc); 4249 bool f2fs_cluster_can_merge_page(struct compress_ctx *cc, pgoff_t index); 4250 bool f2fs_all_cluster_page_ready(struct compress_ctx *cc, struct page **pages, 4251 int index, int nr_pages, bool uptodate); 4252 bool f2fs_sanity_check_cluster(struct dnode_of_data *dn); 4253 void f2fs_compress_ctx_add_page(struct compress_ctx *cc, struct page *page); 4254 int f2fs_write_multi_pages(struct compress_ctx *cc, 4255 int *submitted, 4256 struct writeback_control *wbc, 4257 enum iostat_type io_type); 4258 int f2fs_is_compressed_cluster(struct inode *inode, pgoff_t index); 4259 void f2fs_update_read_extent_tree_range_compressed(struct inode *inode, 4260 pgoff_t fofs, block_t blkaddr, 4261 unsigned int llen, unsigned int c_len); 4262 int f2fs_read_multi_pages(struct compress_ctx *cc, struct bio **bio_ret, 4263 unsigned nr_pages, sector_t *last_block_in_bio, 4264 bool is_readahead, bool for_write); 4265 struct decompress_io_ctx *f2fs_alloc_dic(struct compress_ctx *cc); 4266 void f2fs_decompress_end_io(struct decompress_io_ctx *dic, bool failed, 4267 bool in_task); 4268 void f2fs_put_page_dic(struct page *page, bool in_task); 4269 unsigned int f2fs_cluster_blocks_are_contiguous(struct dnode_of_data *dn); 4270 int f2fs_init_compress_ctx(struct compress_ctx *cc); 4271 void f2fs_destroy_compress_ctx(struct compress_ctx *cc, bool reuse); 4272 void f2fs_init_compress_info(struct f2fs_sb_info *sbi); 4273 int f2fs_init_compress_inode(struct f2fs_sb_info *sbi); 4274 void f2fs_destroy_compress_inode(struct f2fs_sb_info *sbi); 4275 int f2fs_init_page_array_cache(struct f2fs_sb_info *sbi); 4276 void f2fs_destroy_page_array_cache(struct f2fs_sb_info *sbi); 4277 int __init f2fs_init_compress_cache(void); 4278 void f2fs_destroy_compress_cache(void); 4279 struct address_space *COMPRESS_MAPPING(struct f2fs_sb_info *sbi); 4280 void f2fs_invalidate_compress_page(struct f2fs_sb_info *sbi, block_t blkaddr); 4281 void f2fs_cache_compressed_page(struct f2fs_sb_info *sbi, struct page *page, 4282 nid_t ino, block_t blkaddr); 4283 bool f2fs_load_compressed_page(struct f2fs_sb_info *sbi, struct page *page, 4284 block_t blkaddr); 4285 void f2fs_invalidate_compress_pages(struct f2fs_sb_info *sbi, nid_t ino); 4286 #define inc_compr_inode_stat(inode) \ 4287 do { \ 4288 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); \ 4289 sbi->compr_new_inode++; \ 4290 } while (0) 4291 #define add_compr_block_stat(inode, blocks) \ 4292 do { \ 4293 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); \ 4294 int diff = F2FS_I(inode)->i_cluster_size - blocks; \ 4295 sbi->compr_written_block += blocks; \ 4296 sbi->compr_saved_block += diff; \ 4297 } while (0) 4298 #else 4299 static inline bool f2fs_is_compressed_page(struct page *page) { return false; } 4300 static inline bool f2fs_is_compress_backend_ready(struct inode *inode) 4301 { 4302 if (!f2fs_compressed_file(inode)) 4303 return true; 4304 /* not support compression */ 4305 return false; 4306 } 4307 static inline bool f2fs_is_compress_level_valid(int alg, int lvl) { return false; } 4308 static inline struct page *f2fs_compress_control_page(struct page *page) 4309 { 4310 WARN_ON_ONCE(1); 4311 return ERR_PTR(-EINVAL); 4312 } 4313 static inline int __init f2fs_init_compress_mempool(void) { return 0; } 4314 static inline void f2fs_destroy_compress_mempool(void) { } 4315 static inline void f2fs_decompress_cluster(struct decompress_io_ctx *dic, 4316 bool in_task) { } 4317 static inline void f2fs_end_read_compressed_page(struct page *page, 4318 bool failed, block_t blkaddr, bool in_task) 4319 { 4320 WARN_ON_ONCE(1); 4321 } 4322 static inline void f2fs_put_page_dic(struct page *page, bool in_task) 4323 { 4324 WARN_ON_ONCE(1); 4325 } 4326 static inline unsigned int f2fs_cluster_blocks_are_contiguous(struct dnode_of_data *dn) { return 0; } 4327 static inline bool f2fs_sanity_check_cluster(struct dnode_of_data *dn) { return false; } 4328 static inline int f2fs_init_compress_inode(struct f2fs_sb_info *sbi) { return 0; } 4329 static inline void f2fs_destroy_compress_inode(struct f2fs_sb_info *sbi) { } 4330 static inline int f2fs_init_page_array_cache(struct f2fs_sb_info *sbi) { return 0; } 4331 static inline void f2fs_destroy_page_array_cache(struct f2fs_sb_info *sbi) { } 4332 static inline int __init f2fs_init_compress_cache(void) { return 0; } 4333 static inline void f2fs_destroy_compress_cache(void) { } 4334 static inline void f2fs_invalidate_compress_page(struct f2fs_sb_info *sbi, 4335 block_t blkaddr) { } 4336 static inline void f2fs_cache_compressed_page(struct f2fs_sb_info *sbi, 4337 struct page *page, nid_t ino, block_t blkaddr) { } 4338 static inline bool f2fs_load_compressed_page(struct f2fs_sb_info *sbi, 4339 struct page *page, block_t blkaddr) { return false; } 4340 static inline void f2fs_invalidate_compress_pages(struct f2fs_sb_info *sbi, 4341 nid_t ino) { } 4342 #define inc_compr_inode_stat(inode) do { } while (0) 4343 static inline void f2fs_update_read_extent_tree_range_compressed( 4344 struct inode *inode, 4345 pgoff_t fofs, block_t blkaddr, 4346 unsigned int llen, unsigned int c_len) { } 4347 #endif 4348 4349 static inline int set_compress_context(struct inode *inode) 4350 { 4351 #ifdef CONFIG_F2FS_FS_COMPRESSION 4352 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 4353 4354 F2FS_I(inode)->i_compress_algorithm = 4355 F2FS_OPTION(sbi).compress_algorithm; 4356 F2FS_I(inode)->i_log_cluster_size = 4357 F2FS_OPTION(sbi).compress_log_size; 4358 F2FS_I(inode)->i_compress_flag = 4359 F2FS_OPTION(sbi).compress_chksum ? 4360 BIT(COMPRESS_CHKSUM) : 0; 4361 F2FS_I(inode)->i_cluster_size = 4362 BIT(F2FS_I(inode)->i_log_cluster_size); 4363 if ((F2FS_I(inode)->i_compress_algorithm == COMPRESS_LZ4 || 4364 F2FS_I(inode)->i_compress_algorithm == COMPRESS_ZSTD) && 4365 F2FS_OPTION(sbi).compress_level) 4366 F2FS_I(inode)->i_compress_level = 4367 F2FS_OPTION(sbi).compress_level; 4368 F2FS_I(inode)->i_flags |= F2FS_COMPR_FL; 4369 set_inode_flag(inode, FI_COMPRESSED_FILE); 4370 stat_inc_compr_inode(inode); 4371 inc_compr_inode_stat(inode); 4372 f2fs_mark_inode_dirty_sync(inode, true); 4373 return 0; 4374 #else 4375 return -EOPNOTSUPP; 4376 #endif 4377 } 4378 4379 static inline bool f2fs_disable_compressed_file(struct inode *inode) 4380 { 4381 struct f2fs_inode_info *fi = F2FS_I(inode); 4382 4383 if (!f2fs_compressed_file(inode)) 4384 return true; 4385 if (S_ISREG(inode->i_mode) && F2FS_HAS_BLOCKS(inode)) 4386 return false; 4387 4388 fi->i_flags &= ~F2FS_COMPR_FL; 4389 stat_dec_compr_inode(inode); 4390 clear_inode_flag(inode, FI_COMPRESSED_FILE); 4391 f2fs_mark_inode_dirty_sync(inode, true); 4392 return true; 4393 } 4394 4395 #define F2FS_FEATURE_FUNCS(name, flagname) \ 4396 static inline bool f2fs_sb_has_##name(struct f2fs_sb_info *sbi) \ 4397 { \ 4398 return F2FS_HAS_FEATURE(sbi, F2FS_FEATURE_##flagname); \ 4399 } 4400 4401 F2FS_FEATURE_FUNCS(encrypt, ENCRYPT); 4402 F2FS_FEATURE_FUNCS(blkzoned, BLKZONED); 4403 F2FS_FEATURE_FUNCS(extra_attr, EXTRA_ATTR); 4404 F2FS_FEATURE_FUNCS(project_quota, PRJQUOTA); 4405 F2FS_FEATURE_FUNCS(inode_chksum, INODE_CHKSUM); 4406 F2FS_FEATURE_FUNCS(flexible_inline_xattr, FLEXIBLE_INLINE_XATTR); 4407 F2FS_FEATURE_FUNCS(quota_ino, QUOTA_INO); 4408 F2FS_FEATURE_FUNCS(inode_crtime, INODE_CRTIME); 4409 F2FS_FEATURE_FUNCS(lost_found, LOST_FOUND); 4410 F2FS_FEATURE_FUNCS(verity, VERITY); 4411 F2FS_FEATURE_FUNCS(sb_chksum, SB_CHKSUM); 4412 F2FS_FEATURE_FUNCS(casefold, CASEFOLD); 4413 F2FS_FEATURE_FUNCS(compression, COMPRESSION); 4414 F2FS_FEATURE_FUNCS(readonly, RO); 4415 4416 #ifdef CONFIG_BLK_DEV_ZONED 4417 static inline bool f2fs_blkz_is_seq(struct f2fs_sb_info *sbi, int devi, 4418 block_t blkaddr) 4419 { 4420 unsigned int zno = blkaddr / sbi->blocks_per_blkz; 4421 4422 return test_bit(zno, FDEV(devi).blkz_seq); 4423 } 4424 #endif 4425 4426 static inline bool f2fs_hw_should_discard(struct f2fs_sb_info *sbi) 4427 { 4428 return f2fs_sb_has_blkzoned(sbi); 4429 } 4430 4431 static inline bool f2fs_bdev_support_discard(struct block_device *bdev) 4432 { 4433 return bdev_max_discard_sectors(bdev) || bdev_is_zoned(bdev); 4434 } 4435 4436 static inline bool f2fs_hw_support_discard(struct f2fs_sb_info *sbi) 4437 { 4438 int i; 4439 4440 if (!f2fs_is_multi_device(sbi)) 4441 return f2fs_bdev_support_discard(sbi->sb->s_bdev); 4442 4443 for (i = 0; i < sbi->s_ndevs; i++) 4444 if (f2fs_bdev_support_discard(FDEV(i).bdev)) 4445 return true; 4446 return false; 4447 } 4448 4449 static inline bool f2fs_realtime_discard_enable(struct f2fs_sb_info *sbi) 4450 { 4451 return (test_opt(sbi, DISCARD) && f2fs_hw_support_discard(sbi)) || 4452 f2fs_hw_should_discard(sbi); 4453 } 4454 4455 static inline bool f2fs_hw_is_readonly(struct f2fs_sb_info *sbi) 4456 { 4457 int i; 4458 4459 if (!f2fs_is_multi_device(sbi)) 4460 return bdev_read_only(sbi->sb->s_bdev); 4461 4462 for (i = 0; i < sbi->s_ndevs; i++) 4463 if (bdev_read_only(FDEV(i).bdev)) 4464 return true; 4465 return false; 4466 } 4467 4468 static inline bool f2fs_dev_is_readonly(struct f2fs_sb_info *sbi) 4469 { 4470 return f2fs_sb_has_readonly(sbi) || f2fs_hw_is_readonly(sbi); 4471 } 4472 4473 static inline bool f2fs_lfs_mode(struct f2fs_sb_info *sbi) 4474 { 4475 return F2FS_OPTION(sbi).fs_mode == FS_MODE_LFS; 4476 } 4477 4478 static inline bool f2fs_low_mem_mode(struct f2fs_sb_info *sbi) 4479 { 4480 return F2FS_OPTION(sbi).memory_mode == MEMORY_MODE_LOW; 4481 } 4482 4483 static inline bool f2fs_may_compress(struct inode *inode) 4484 { 4485 if (IS_SWAPFILE(inode) || f2fs_is_pinned_file(inode) || 4486 f2fs_is_atomic_file(inode) || f2fs_has_inline_data(inode)) 4487 return false; 4488 return S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode); 4489 } 4490 4491 static inline void f2fs_i_compr_blocks_update(struct inode *inode, 4492 u64 blocks, bool add) 4493 { 4494 struct f2fs_inode_info *fi = F2FS_I(inode); 4495 int diff = fi->i_cluster_size - blocks; 4496 4497 /* don't update i_compr_blocks if saved blocks were released */ 4498 if (!add && !atomic_read(&fi->i_compr_blocks)) 4499 return; 4500 4501 if (add) { 4502 atomic_add(diff, &fi->i_compr_blocks); 4503 stat_add_compr_blocks(inode, diff); 4504 } else { 4505 atomic_sub(diff, &fi->i_compr_blocks); 4506 stat_sub_compr_blocks(inode, diff); 4507 } 4508 f2fs_mark_inode_dirty_sync(inode, true); 4509 } 4510 4511 static inline bool f2fs_allow_multi_device_dio(struct f2fs_sb_info *sbi, 4512 int flag) 4513 { 4514 if (!f2fs_is_multi_device(sbi)) 4515 return false; 4516 if (flag != F2FS_GET_BLOCK_DIO) 4517 return false; 4518 return sbi->aligned_blksize; 4519 } 4520 4521 static inline bool f2fs_need_verity(const struct inode *inode, pgoff_t idx) 4522 { 4523 return fsverity_active(inode) && 4524 idx < DIV_ROUND_UP(inode->i_size, PAGE_SIZE); 4525 } 4526 4527 #ifdef CONFIG_F2FS_FAULT_INJECTION 4528 extern void f2fs_build_fault_attr(struct f2fs_sb_info *sbi, unsigned int rate, 4529 unsigned int type); 4530 #else 4531 #define f2fs_build_fault_attr(sbi, rate, type) do { } while (0) 4532 #endif 4533 4534 static inline bool is_journalled_quota(struct f2fs_sb_info *sbi) 4535 { 4536 #ifdef CONFIG_QUOTA 4537 if (f2fs_sb_has_quota_ino(sbi)) 4538 return true; 4539 if (F2FS_OPTION(sbi).s_qf_names[USRQUOTA] || 4540 F2FS_OPTION(sbi).s_qf_names[GRPQUOTA] || 4541 F2FS_OPTION(sbi).s_qf_names[PRJQUOTA]) 4542 return true; 4543 #endif 4544 return false; 4545 } 4546 4547 static inline bool f2fs_block_unit_discard(struct f2fs_sb_info *sbi) 4548 { 4549 return F2FS_OPTION(sbi).discard_unit == DISCARD_UNIT_BLOCK; 4550 } 4551 4552 static inline void f2fs_io_schedule_timeout(long timeout) 4553 { 4554 set_current_state(TASK_UNINTERRUPTIBLE); 4555 io_schedule_timeout(timeout); 4556 } 4557 4558 static inline void f2fs_handle_page_eio(struct f2fs_sb_info *sbi, pgoff_t ofs, 4559 enum page_type type) 4560 { 4561 if (unlikely(f2fs_cp_error(sbi))) 4562 return; 4563 4564 if (ofs == sbi->page_eio_ofs[type]) { 4565 if (sbi->page_eio_cnt[type]++ == MAX_RETRY_PAGE_EIO) 4566 set_ckpt_flags(sbi, CP_ERROR_FLAG); 4567 } else { 4568 sbi->page_eio_ofs[type] = ofs; 4569 sbi->page_eio_cnt[type] = 0; 4570 } 4571 } 4572 4573 static inline bool f2fs_is_readonly(struct f2fs_sb_info *sbi) 4574 { 4575 return f2fs_sb_has_readonly(sbi) || f2fs_readonly(sbi->sb); 4576 } 4577 4578 #define EFSBADCRC EBADMSG /* Bad CRC detected */ 4579 #define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */ 4580 4581 #endif /* _LINUX_F2FS_H */ 4582