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 79 /* maximum retry count for injected failure */ 80 #define DEFAULT_FAILURE_RETRY_COUNT 8 81 #else 82 #define DEFAULT_FAILURE_RETRY_COUNT 1 83 #endif 84 85 /* 86 * For mount options 87 */ 88 #define F2FS_MOUNT_DISABLE_ROLL_FORWARD 0x00000001 89 #define F2FS_MOUNT_DISCARD 0x00000002 90 #define F2FS_MOUNT_NOHEAP 0x00000004 91 #define F2FS_MOUNT_XATTR_USER 0x00000008 92 #define F2FS_MOUNT_POSIX_ACL 0x00000010 93 #define F2FS_MOUNT_DISABLE_EXT_IDENTIFY 0x00000020 94 #define F2FS_MOUNT_INLINE_XATTR 0x00000040 95 #define F2FS_MOUNT_INLINE_DATA 0x00000080 96 #define F2FS_MOUNT_INLINE_DENTRY 0x00000100 97 #define F2FS_MOUNT_FLUSH_MERGE 0x00000200 98 #define F2FS_MOUNT_NOBARRIER 0x00000400 99 #define F2FS_MOUNT_FASTBOOT 0x00000800 100 #define F2FS_MOUNT_READ_EXTENT_CACHE 0x00001000 101 #define F2FS_MOUNT_DATA_FLUSH 0x00002000 102 #define F2FS_MOUNT_FAULT_INJECTION 0x00004000 103 #define F2FS_MOUNT_USRQUOTA 0x00008000 104 #define F2FS_MOUNT_GRPQUOTA 0x00010000 105 #define F2FS_MOUNT_PRJQUOTA 0x00020000 106 #define F2FS_MOUNT_QUOTA 0x00040000 107 #define F2FS_MOUNT_INLINE_XATTR_SIZE 0x00080000 108 #define F2FS_MOUNT_RESERVE_ROOT 0x00100000 109 #define F2FS_MOUNT_DISABLE_CHECKPOINT 0x00200000 110 #define F2FS_MOUNT_NORECOVERY 0x00400000 111 #define F2FS_MOUNT_ATGC 0x00800000 112 #define F2FS_MOUNT_MERGE_CHECKPOINT 0x01000000 113 #define F2FS_MOUNT_GC_MERGE 0x02000000 114 #define F2FS_MOUNT_COMPRESS_CACHE 0x04000000 115 #define F2FS_MOUNT_AGE_EXTENT_CACHE 0x08000000 116 117 #define F2FS_OPTION(sbi) ((sbi)->mount_opt) 118 #define clear_opt(sbi, option) (F2FS_OPTION(sbi).opt &= ~F2FS_MOUNT_##option) 119 #define set_opt(sbi, option) (F2FS_OPTION(sbi).opt |= F2FS_MOUNT_##option) 120 #define test_opt(sbi, option) (F2FS_OPTION(sbi).opt & F2FS_MOUNT_##option) 121 122 #define ver_after(a, b) (typecheck(unsigned long long, a) && \ 123 typecheck(unsigned long long, b) && \ 124 ((long long)((a) - (b)) > 0)) 125 126 typedef u32 block_t; /* 127 * should not change u32, since it is the on-disk block 128 * address format, __le32. 129 */ 130 typedef u32 nid_t; 131 132 #define COMPRESS_EXT_NUM 16 133 134 /* 135 * An implementation of an rwsem that is explicitly unfair to readers. This 136 * prevents priority inversion when a low-priority reader acquires the read lock 137 * while sleeping on the write lock but the write lock is needed by 138 * higher-priority clients. 139 */ 140 141 struct f2fs_rwsem { 142 struct rw_semaphore internal_rwsem; 143 #ifdef CONFIG_F2FS_UNFAIR_RWSEM 144 wait_queue_head_t read_waiters; 145 #endif 146 }; 147 148 struct f2fs_mount_info { 149 unsigned int opt; 150 int write_io_size_bits; /* Write IO size bits */ 151 block_t root_reserved_blocks; /* root reserved blocks */ 152 kuid_t s_resuid; /* reserved blocks for uid */ 153 kgid_t s_resgid; /* reserved blocks for gid */ 154 int active_logs; /* # of active logs */ 155 int inline_xattr_size; /* inline xattr size */ 156 #ifdef CONFIG_F2FS_FAULT_INJECTION 157 struct f2fs_fault_info fault_info; /* For fault injection */ 158 #endif 159 #ifdef CONFIG_QUOTA 160 /* Names of quota files with journalled quota */ 161 char *s_qf_names[MAXQUOTAS]; 162 int s_jquota_fmt; /* Format of quota to use */ 163 #endif 164 /* For which write hints are passed down to block layer */ 165 int alloc_mode; /* segment allocation policy */ 166 int fsync_mode; /* fsync policy */ 167 int fs_mode; /* fs mode: LFS or ADAPTIVE */ 168 int bggc_mode; /* bggc mode: off, on or sync */ 169 int memory_mode; /* memory mode */ 170 int errors; /* errors parameter */ 171 int discard_unit; /* 172 * discard command's offset/size should 173 * be aligned to this unit: block, 174 * segment or section 175 */ 176 struct fscrypt_dummy_policy dummy_enc_policy; /* test dummy encryption */ 177 block_t unusable_cap_perc; /* percentage for cap */ 178 block_t unusable_cap; /* Amount of space allowed to be 179 * unusable when disabling checkpoint 180 */ 181 182 /* For compression */ 183 unsigned char compress_algorithm; /* algorithm type */ 184 unsigned char compress_log_size; /* cluster log size */ 185 unsigned char compress_level; /* compress level */ 186 bool compress_chksum; /* compressed data chksum */ 187 unsigned char compress_ext_cnt; /* extension count */ 188 unsigned char nocompress_ext_cnt; /* nocompress extension count */ 189 int compress_mode; /* compression mode */ 190 unsigned char extensions[COMPRESS_EXT_NUM][F2FS_EXTENSION_LEN]; /* extensions */ 191 unsigned char noextensions[COMPRESS_EXT_NUM][F2FS_EXTENSION_LEN]; /* extensions */ 192 }; 193 194 #define F2FS_FEATURE_ENCRYPT 0x00000001 195 #define F2FS_FEATURE_BLKZONED 0x00000002 196 #define F2FS_FEATURE_ATOMIC_WRITE 0x00000004 197 #define F2FS_FEATURE_EXTRA_ATTR 0x00000008 198 #define F2FS_FEATURE_PRJQUOTA 0x00000010 199 #define F2FS_FEATURE_INODE_CHKSUM 0x00000020 200 #define F2FS_FEATURE_FLEXIBLE_INLINE_XATTR 0x00000040 201 #define F2FS_FEATURE_QUOTA_INO 0x00000080 202 #define F2FS_FEATURE_INODE_CRTIME 0x00000100 203 #define F2FS_FEATURE_LOST_FOUND 0x00000200 204 #define F2FS_FEATURE_VERITY 0x00000400 205 #define F2FS_FEATURE_SB_CHKSUM 0x00000800 206 #define F2FS_FEATURE_CASEFOLD 0x00001000 207 #define F2FS_FEATURE_COMPRESSION 0x00002000 208 #define F2FS_FEATURE_RO 0x00004000 209 210 #define __F2FS_HAS_FEATURE(raw_super, mask) \ 211 ((raw_super->feature & cpu_to_le32(mask)) != 0) 212 #define F2FS_HAS_FEATURE(sbi, mask) __F2FS_HAS_FEATURE(sbi->raw_super, mask) 213 214 /* 215 * Default values for user and/or group using reserved blocks 216 */ 217 #define F2FS_DEF_RESUID 0 218 #define F2FS_DEF_RESGID 0 219 220 /* 221 * For checkpoint manager 222 */ 223 enum { 224 NAT_BITMAP, 225 SIT_BITMAP 226 }; 227 228 #define CP_UMOUNT 0x00000001 229 #define CP_FASTBOOT 0x00000002 230 #define CP_SYNC 0x00000004 231 #define CP_RECOVERY 0x00000008 232 #define CP_DISCARD 0x00000010 233 #define CP_TRIMMED 0x00000020 234 #define CP_PAUSE 0x00000040 235 #define CP_RESIZE 0x00000080 236 237 #define DEF_MAX_DISCARD_REQUEST 8 /* issue 8 discards per round */ 238 #define DEF_MIN_DISCARD_ISSUE_TIME 50 /* 50 ms, if exists */ 239 #define DEF_MID_DISCARD_ISSUE_TIME 500 /* 500 ms, if device busy */ 240 #define DEF_MAX_DISCARD_ISSUE_TIME 60000 /* 60 s, if no candidates */ 241 #define DEF_DISCARD_URGENT_UTIL 80 /* do more discard over 80% */ 242 #define DEF_CP_INTERVAL 60 /* 60 secs */ 243 #define DEF_IDLE_INTERVAL 5 /* 5 secs */ 244 #define DEF_DISABLE_INTERVAL 5 /* 5 secs */ 245 #define DEF_DISABLE_QUICK_INTERVAL 1 /* 1 secs */ 246 #define DEF_UMOUNT_DISCARD_TIMEOUT 5 /* 5 secs */ 247 248 struct cp_control { 249 int reason; 250 __u64 trim_start; 251 __u64 trim_end; 252 __u64 trim_minlen; 253 }; 254 255 /* 256 * indicate meta/data type 257 */ 258 enum { 259 META_CP, 260 META_NAT, 261 META_SIT, 262 META_SSA, 263 META_MAX, 264 META_POR, 265 DATA_GENERIC, /* check range only */ 266 DATA_GENERIC_ENHANCE, /* strong check on range and segment bitmap */ 267 DATA_GENERIC_ENHANCE_READ, /* 268 * strong check on range and segment 269 * bitmap but no warning due to race 270 * condition of read on truncated area 271 * by extent_cache 272 */ 273 DATA_GENERIC_ENHANCE_UPDATE, /* 274 * strong check on range and segment 275 * bitmap for update case 276 */ 277 META_GENERIC, 278 }; 279 280 /* for the list of ino */ 281 enum { 282 ORPHAN_INO, /* for orphan ino list */ 283 APPEND_INO, /* for append ino list */ 284 UPDATE_INO, /* for update ino list */ 285 TRANS_DIR_INO, /* for transactions dir ino list */ 286 FLUSH_INO, /* for multiple device flushing */ 287 MAX_INO_ENTRY, /* max. list */ 288 }; 289 290 struct ino_entry { 291 struct list_head list; /* list head */ 292 nid_t ino; /* inode number */ 293 unsigned int dirty_device; /* dirty device bitmap */ 294 }; 295 296 /* for the list of inodes to be GCed */ 297 struct inode_entry { 298 struct list_head list; /* list head */ 299 struct inode *inode; /* vfs inode pointer */ 300 }; 301 302 struct fsync_node_entry { 303 struct list_head list; /* list head */ 304 struct page *page; /* warm node page pointer */ 305 unsigned int seq_id; /* sequence id */ 306 }; 307 308 struct ckpt_req { 309 struct completion wait; /* completion for checkpoint done */ 310 struct llist_node llnode; /* llist_node to be linked in wait queue */ 311 int ret; /* return code of checkpoint */ 312 ktime_t queue_time; /* request queued time */ 313 }; 314 315 struct ckpt_req_control { 316 struct task_struct *f2fs_issue_ckpt; /* checkpoint task */ 317 int ckpt_thread_ioprio; /* checkpoint merge thread ioprio */ 318 wait_queue_head_t ckpt_wait_queue; /* waiting queue for wake-up */ 319 atomic_t issued_ckpt; /* # of actually issued ckpts */ 320 atomic_t total_ckpt; /* # of total ckpts */ 321 atomic_t queued_ckpt; /* # of queued ckpts */ 322 struct llist_head issue_list; /* list for command issue */ 323 spinlock_t stat_lock; /* lock for below checkpoint time stats */ 324 unsigned int cur_time; /* cur wait time in msec for currently issued checkpoint */ 325 unsigned int peak_time; /* peak wait time in msec until now */ 326 }; 327 328 /* for the bitmap indicate blocks to be discarded */ 329 struct discard_entry { 330 struct list_head list; /* list head */ 331 block_t start_blkaddr; /* start blockaddr of current segment */ 332 unsigned char discard_map[SIT_VBLOCK_MAP_SIZE]; /* segment discard bitmap */ 333 }; 334 335 /* minimum discard granularity, unit: block count */ 336 #define MIN_DISCARD_GRANULARITY 1 337 /* default discard granularity of inner discard thread, unit: block count */ 338 #define DEFAULT_DISCARD_GRANULARITY 16 339 /* default maximum discard granularity of ordered discard, unit: block count */ 340 #define DEFAULT_MAX_ORDERED_DISCARD_GRANULARITY 16 341 342 /* max discard pend list number */ 343 #define MAX_PLIST_NUM 512 344 #define plist_idx(blk_num) ((blk_num) >= MAX_PLIST_NUM ? \ 345 (MAX_PLIST_NUM - 1) : ((blk_num) - 1)) 346 347 enum { 348 D_PREP, /* initial */ 349 D_PARTIAL, /* partially submitted */ 350 D_SUBMIT, /* all submitted */ 351 D_DONE, /* finished */ 352 }; 353 354 struct discard_info { 355 block_t lstart; /* logical start address */ 356 block_t len; /* length */ 357 block_t start; /* actual start address in dev */ 358 }; 359 360 struct discard_cmd { 361 struct rb_node rb_node; /* rb node located in rb-tree */ 362 struct discard_info di; /* discard info */ 363 struct list_head list; /* command list */ 364 struct completion wait; /* compleation */ 365 struct block_device *bdev; /* bdev */ 366 unsigned short ref; /* reference count */ 367 unsigned char state; /* state */ 368 unsigned char queued; /* queued discard */ 369 int error; /* bio error */ 370 spinlock_t lock; /* for state/bio_ref updating */ 371 unsigned short bio_ref; /* bio reference count */ 372 }; 373 374 enum { 375 DPOLICY_BG, 376 DPOLICY_FORCE, 377 DPOLICY_FSTRIM, 378 DPOLICY_UMOUNT, 379 MAX_DPOLICY, 380 }; 381 382 struct discard_policy { 383 int type; /* type of discard */ 384 unsigned int min_interval; /* used for candidates exist */ 385 unsigned int mid_interval; /* used for device busy */ 386 unsigned int max_interval; /* used for candidates not exist */ 387 unsigned int max_requests; /* # of discards issued per round */ 388 unsigned int io_aware_gran; /* minimum granularity discard not be aware of I/O */ 389 bool io_aware; /* issue discard in idle time */ 390 bool sync; /* submit discard with REQ_SYNC flag */ 391 bool ordered; /* issue discard by lba order */ 392 bool timeout; /* discard timeout for put_super */ 393 unsigned int granularity; /* discard granularity */ 394 }; 395 396 struct discard_cmd_control { 397 struct task_struct *f2fs_issue_discard; /* discard thread */ 398 struct list_head entry_list; /* 4KB discard entry list */ 399 struct list_head pend_list[MAX_PLIST_NUM];/* store pending entries */ 400 struct list_head wait_list; /* store on-flushing entries */ 401 struct list_head fstrim_list; /* in-flight discard from fstrim */ 402 wait_queue_head_t discard_wait_queue; /* waiting queue for wake-up */ 403 struct mutex cmd_lock; 404 unsigned int nr_discards; /* # of discards in the list */ 405 unsigned int max_discards; /* max. discards to be issued */ 406 unsigned int max_discard_request; /* max. discard request per round */ 407 unsigned int min_discard_issue_time; /* min. interval between discard issue */ 408 unsigned int mid_discard_issue_time; /* mid. interval between discard issue */ 409 unsigned int max_discard_issue_time; /* max. interval between discard issue */ 410 unsigned int discard_io_aware_gran; /* minimum discard granularity not be aware of I/O */ 411 unsigned int discard_urgent_util; /* utilization which issue discard proactively */ 412 unsigned int discard_granularity; /* discard granularity */ 413 unsigned int max_ordered_discard; /* maximum discard granularity issued by lba order */ 414 unsigned int undiscard_blks; /* # of undiscard blocks */ 415 unsigned int next_pos; /* next discard position */ 416 atomic_t issued_discard; /* # of issued discard */ 417 atomic_t queued_discard; /* # of queued discard */ 418 atomic_t discard_cmd_cnt; /* # of cached cmd count */ 419 struct rb_root_cached root; /* root of discard rb-tree */ 420 bool rbtree_check; /* config for consistence check */ 421 bool discard_wake; /* to wake up discard thread */ 422 }; 423 424 /* for the list of fsync inodes, used only during recovery */ 425 struct fsync_inode_entry { 426 struct list_head list; /* list head */ 427 struct inode *inode; /* vfs inode pointer */ 428 block_t blkaddr; /* block address locating the last fsync */ 429 block_t last_dentry; /* block address locating the last dentry */ 430 }; 431 432 #define nats_in_cursum(jnl) (le16_to_cpu((jnl)->n_nats)) 433 #define sits_in_cursum(jnl) (le16_to_cpu((jnl)->n_sits)) 434 435 #define nat_in_journal(jnl, i) ((jnl)->nat_j.entries[i].ne) 436 #define nid_in_journal(jnl, i) ((jnl)->nat_j.entries[i].nid) 437 #define sit_in_journal(jnl, i) ((jnl)->sit_j.entries[i].se) 438 #define segno_in_journal(jnl, i) ((jnl)->sit_j.entries[i].segno) 439 440 #define MAX_NAT_JENTRIES(jnl) (NAT_JOURNAL_ENTRIES - nats_in_cursum(jnl)) 441 #define MAX_SIT_JENTRIES(jnl) (SIT_JOURNAL_ENTRIES - sits_in_cursum(jnl)) 442 443 static inline int update_nats_in_cursum(struct f2fs_journal *journal, int i) 444 { 445 int before = nats_in_cursum(journal); 446 447 journal->n_nats = cpu_to_le16(before + i); 448 return before; 449 } 450 451 static inline int update_sits_in_cursum(struct f2fs_journal *journal, int i) 452 { 453 int before = sits_in_cursum(journal); 454 455 journal->n_sits = cpu_to_le16(before + i); 456 return before; 457 } 458 459 static inline bool __has_cursum_space(struct f2fs_journal *journal, 460 int size, int type) 461 { 462 if (type == NAT_JOURNAL) 463 return size <= MAX_NAT_JENTRIES(journal); 464 return size <= MAX_SIT_JENTRIES(journal); 465 } 466 467 /* for inline stuff */ 468 #define DEF_INLINE_RESERVED_SIZE 1 469 static inline int get_extra_isize(struct inode *inode); 470 static inline int get_inline_xattr_addrs(struct inode *inode); 471 #define MAX_INLINE_DATA(inode) (sizeof(__le32) * \ 472 (CUR_ADDRS_PER_INODE(inode) - \ 473 get_inline_xattr_addrs(inode) - \ 474 DEF_INLINE_RESERVED_SIZE)) 475 476 /* for inline dir */ 477 #define NR_INLINE_DENTRY(inode) (MAX_INLINE_DATA(inode) * BITS_PER_BYTE / \ 478 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \ 479 BITS_PER_BYTE + 1)) 480 #define INLINE_DENTRY_BITMAP_SIZE(inode) \ 481 DIV_ROUND_UP(NR_INLINE_DENTRY(inode), BITS_PER_BYTE) 482 #define INLINE_RESERVED_SIZE(inode) (MAX_INLINE_DATA(inode) - \ 483 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \ 484 NR_INLINE_DENTRY(inode) + \ 485 INLINE_DENTRY_BITMAP_SIZE(inode))) 486 487 /* 488 * For INODE and NODE manager 489 */ 490 /* for directory operations */ 491 492 struct f2fs_filename { 493 /* 494 * The filename the user specified. This is NULL for some 495 * filesystem-internal operations, e.g. converting an inline directory 496 * to a non-inline one, or roll-forward recovering an encrypted dentry. 497 */ 498 const struct qstr *usr_fname; 499 500 /* 501 * The on-disk filename. For encrypted directories, this is encrypted. 502 * This may be NULL for lookups in an encrypted dir without the key. 503 */ 504 struct fscrypt_str disk_name; 505 506 /* The dirhash of this filename */ 507 f2fs_hash_t hash; 508 509 #ifdef CONFIG_FS_ENCRYPTION 510 /* 511 * For lookups in encrypted directories: either the buffer backing 512 * disk_name, or a buffer that holds the decoded no-key name. 513 */ 514 struct fscrypt_str crypto_buf; 515 #endif 516 #if IS_ENABLED(CONFIG_UNICODE) 517 /* 518 * For casefolded directories: the casefolded name, but it's left NULL 519 * if the original name is not valid Unicode, if the original name is 520 * "." or "..", if the directory is both casefolded and encrypted and 521 * its encryption key is unavailable, or if the filesystem is doing an 522 * internal operation where usr_fname is also NULL. In all these cases 523 * we fall back to treating the name as an opaque byte sequence. 524 */ 525 struct fscrypt_str cf_name; 526 #endif 527 }; 528 529 struct f2fs_dentry_ptr { 530 struct inode *inode; 531 void *bitmap; 532 struct f2fs_dir_entry *dentry; 533 __u8 (*filename)[F2FS_SLOT_LEN]; 534 int max; 535 int nr_bitmap; 536 }; 537 538 static inline void make_dentry_ptr_block(struct inode *inode, 539 struct f2fs_dentry_ptr *d, struct f2fs_dentry_block *t) 540 { 541 d->inode = inode; 542 d->max = NR_DENTRY_IN_BLOCK; 543 d->nr_bitmap = SIZE_OF_DENTRY_BITMAP; 544 d->bitmap = t->dentry_bitmap; 545 d->dentry = t->dentry; 546 d->filename = t->filename; 547 } 548 549 static inline void make_dentry_ptr_inline(struct inode *inode, 550 struct f2fs_dentry_ptr *d, void *t) 551 { 552 int entry_cnt = NR_INLINE_DENTRY(inode); 553 int bitmap_size = INLINE_DENTRY_BITMAP_SIZE(inode); 554 int reserved_size = INLINE_RESERVED_SIZE(inode); 555 556 d->inode = inode; 557 d->max = entry_cnt; 558 d->nr_bitmap = bitmap_size; 559 d->bitmap = t; 560 d->dentry = t + bitmap_size + reserved_size; 561 d->filename = t + bitmap_size + reserved_size + 562 SIZE_OF_DIR_ENTRY * entry_cnt; 563 } 564 565 /* 566 * XATTR_NODE_OFFSET stores xattrs to one node block per file keeping -1 567 * as its node offset to distinguish from index node blocks. 568 * But some bits are used to mark the node block. 569 */ 570 #define XATTR_NODE_OFFSET ((((unsigned int)-1) << OFFSET_BIT_SHIFT) \ 571 >> OFFSET_BIT_SHIFT) 572 enum { 573 ALLOC_NODE, /* allocate a new node page if needed */ 574 LOOKUP_NODE, /* look up a node without readahead */ 575 LOOKUP_NODE_RA, /* 576 * look up a node with readahead called 577 * by get_data_block. 578 */ 579 }; 580 581 #define DEFAULT_RETRY_IO_COUNT 8 /* maximum retry read IO or flush count */ 582 583 /* congestion wait timeout value, default: 20ms */ 584 #define DEFAULT_IO_TIMEOUT (msecs_to_jiffies(20)) 585 586 /* maximum retry quota flush count */ 587 #define DEFAULT_RETRY_QUOTA_FLUSH_COUNT 8 588 589 /* maximum retry of EIO'ed page */ 590 #define MAX_RETRY_PAGE_EIO 100 591 592 #define F2FS_LINK_MAX 0xffffffff /* maximum link count per file */ 593 594 #define MAX_DIR_RA_PAGES 4 /* maximum ra pages of dir */ 595 596 /* dirty segments threshold for triggering CP */ 597 #define DEFAULT_DIRTY_THRESHOLD 4 598 599 #define RECOVERY_MAX_RA_BLOCKS BIO_MAX_VECS 600 #define RECOVERY_MIN_RA_BLOCKS 1 601 602 #define F2FS_ONSTACK_PAGES 16 /* nr of onstack pages */ 603 604 /* for in-memory extent cache entry */ 605 #define F2FS_MIN_EXTENT_LEN 64 /* minimum extent length */ 606 607 /* number of extent info in extent cache we try to shrink */ 608 #define READ_EXTENT_CACHE_SHRINK_NUMBER 128 609 610 /* number of age extent info in extent cache we try to shrink */ 611 #define AGE_EXTENT_CACHE_SHRINK_NUMBER 128 612 #define LAST_AGE_WEIGHT 30 613 #define SAME_AGE_REGION 1024 614 615 /* 616 * Define data block with age less than 1GB as hot data 617 * define data block with age less than 10GB but more than 1GB as warm data 618 */ 619 #define DEF_HOT_DATA_AGE_THRESHOLD 262144 620 #define DEF_WARM_DATA_AGE_THRESHOLD 2621440 621 622 /* extent cache type */ 623 enum extent_type { 624 EX_READ, 625 EX_BLOCK_AGE, 626 NR_EXTENT_CACHES, 627 }; 628 629 struct extent_info { 630 unsigned int fofs; /* start offset in a file */ 631 unsigned int len; /* length of the extent */ 632 union { 633 /* read extent_cache */ 634 struct { 635 /* start block address of the extent */ 636 block_t blk; 637 #ifdef CONFIG_F2FS_FS_COMPRESSION 638 /* physical extent length of compressed blocks */ 639 unsigned int c_len; 640 #endif 641 }; 642 /* block age extent_cache */ 643 struct { 644 /* block age of the extent */ 645 unsigned long long age; 646 /* last total blocks allocated */ 647 unsigned long long last_blocks; 648 }; 649 }; 650 }; 651 652 struct extent_node { 653 struct rb_node rb_node; /* rb node located in rb-tree */ 654 struct extent_info ei; /* extent info */ 655 struct list_head list; /* node in global extent list of sbi */ 656 struct extent_tree *et; /* extent tree pointer */ 657 }; 658 659 struct extent_tree { 660 nid_t ino; /* inode number */ 661 enum extent_type type; /* keep the extent tree type */ 662 struct rb_root_cached root; /* root of extent info rb-tree */ 663 struct extent_node *cached_en; /* recently accessed extent node */ 664 struct list_head list; /* to be used by sbi->zombie_list */ 665 rwlock_t lock; /* protect extent info rb-tree */ 666 atomic_t node_cnt; /* # of extent node in rb-tree*/ 667 bool largest_updated; /* largest extent updated */ 668 struct extent_info largest; /* largest cached extent for EX_READ */ 669 }; 670 671 struct extent_tree_info { 672 struct radix_tree_root extent_tree_root;/* cache extent cache entries */ 673 struct mutex extent_tree_lock; /* locking extent radix tree */ 674 struct list_head extent_list; /* lru list for shrinker */ 675 spinlock_t extent_lock; /* locking extent lru list */ 676 atomic_t total_ext_tree; /* extent tree count */ 677 struct list_head zombie_list; /* extent zombie tree list */ 678 atomic_t total_zombie_tree; /* extent zombie tree count */ 679 atomic_t total_ext_node; /* extent info count */ 680 }; 681 682 /* 683 * State of block returned by f2fs_map_blocks. 684 */ 685 #define F2FS_MAP_NEW (1U << 0) 686 #define F2FS_MAP_MAPPED (1U << 1) 687 #define F2FS_MAP_DELALLOC (1U << 2) 688 #define F2FS_MAP_FLAGS (F2FS_MAP_NEW | F2FS_MAP_MAPPED |\ 689 F2FS_MAP_DELALLOC) 690 691 struct f2fs_map_blocks { 692 struct block_device *m_bdev; /* for multi-device dio */ 693 block_t m_pblk; 694 block_t m_lblk; 695 unsigned int m_len; 696 unsigned int m_flags; 697 pgoff_t *m_next_pgofs; /* point next possible non-hole pgofs */ 698 pgoff_t *m_next_extent; /* point to next possible extent */ 699 int m_seg_type; 700 bool m_may_create; /* indicate it is from write path */ 701 bool m_multidev_dio; /* indicate it allows multi-device dio */ 702 }; 703 704 /* for flag in get_data_block */ 705 enum { 706 F2FS_GET_BLOCK_DEFAULT, 707 F2FS_GET_BLOCK_FIEMAP, 708 F2FS_GET_BLOCK_BMAP, 709 F2FS_GET_BLOCK_DIO, 710 F2FS_GET_BLOCK_PRE_DIO, 711 F2FS_GET_BLOCK_PRE_AIO, 712 F2FS_GET_BLOCK_PRECACHE, 713 }; 714 715 /* 716 * i_advise uses FADVISE_XXX_BIT. We can add additional hints later. 717 */ 718 #define FADVISE_COLD_BIT 0x01 719 #define FADVISE_LOST_PINO_BIT 0x02 720 #define FADVISE_ENCRYPT_BIT 0x04 721 #define FADVISE_ENC_NAME_BIT 0x08 722 #define FADVISE_KEEP_SIZE_BIT 0x10 723 #define FADVISE_HOT_BIT 0x20 724 #define FADVISE_VERITY_BIT 0x40 725 #define FADVISE_TRUNC_BIT 0x80 726 727 #define FADVISE_MODIFIABLE_BITS (FADVISE_COLD_BIT | FADVISE_HOT_BIT) 728 729 #define file_is_cold(inode) is_file(inode, FADVISE_COLD_BIT) 730 #define file_set_cold(inode) set_file(inode, FADVISE_COLD_BIT) 731 #define file_clear_cold(inode) clear_file(inode, FADVISE_COLD_BIT) 732 733 #define file_wrong_pino(inode) is_file(inode, FADVISE_LOST_PINO_BIT) 734 #define file_lost_pino(inode) set_file(inode, FADVISE_LOST_PINO_BIT) 735 #define file_got_pino(inode) clear_file(inode, FADVISE_LOST_PINO_BIT) 736 737 #define file_is_encrypt(inode) is_file(inode, FADVISE_ENCRYPT_BIT) 738 #define file_set_encrypt(inode) set_file(inode, FADVISE_ENCRYPT_BIT) 739 740 #define file_enc_name(inode) is_file(inode, FADVISE_ENC_NAME_BIT) 741 #define file_set_enc_name(inode) set_file(inode, FADVISE_ENC_NAME_BIT) 742 743 #define file_keep_isize(inode) is_file(inode, FADVISE_KEEP_SIZE_BIT) 744 #define file_set_keep_isize(inode) set_file(inode, FADVISE_KEEP_SIZE_BIT) 745 746 #define file_is_hot(inode) is_file(inode, FADVISE_HOT_BIT) 747 #define file_set_hot(inode) set_file(inode, FADVISE_HOT_BIT) 748 #define file_clear_hot(inode) clear_file(inode, FADVISE_HOT_BIT) 749 750 #define file_is_verity(inode) is_file(inode, FADVISE_VERITY_BIT) 751 #define file_set_verity(inode) set_file(inode, FADVISE_VERITY_BIT) 752 753 #define file_should_truncate(inode) is_file(inode, FADVISE_TRUNC_BIT) 754 #define file_need_truncate(inode) set_file(inode, FADVISE_TRUNC_BIT) 755 #define file_dont_truncate(inode) clear_file(inode, FADVISE_TRUNC_BIT) 756 757 #define DEF_DIR_LEVEL 0 758 759 enum { 760 GC_FAILURE_PIN, 761 MAX_GC_FAILURE 762 }; 763 764 /* used for f2fs_inode_info->flags */ 765 enum { 766 FI_NEW_INODE, /* indicate newly allocated inode */ 767 FI_DIRTY_INODE, /* indicate inode is dirty or not */ 768 FI_AUTO_RECOVER, /* indicate inode is recoverable */ 769 FI_DIRTY_DIR, /* indicate directory has dirty pages */ 770 FI_INC_LINK, /* need to increment i_nlink */ 771 FI_ACL_MODE, /* indicate acl mode */ 772 FI_NO_ALLOC, /* should not allocate any blocks */ 773 FI_FREE_NID, /* free allocated nide */ 774 FI_NO_EXTENT, /* not to use the extent cache */ 775 FI_INLINE_XATTR, /* used for inline xattr */ 776 FI_INLINE_DATA, /* used for inline data*/ 777 FI_INLINE_DENTRY, /* used for inline dentry */ 778 FI_APPEND_WRITE, /* inode has appended data */ 779 FI_UPDATE_WRITE, /* inode has in-place-update data */ 780 FI_NEED_IPU, /* used for ipu per file */ 781 FI_ATOMIC_FILE, /* indicate atomic file */ 782 FI_DATA_EXIST, /* indicate data exists */ 783 FI_INLINE_DOTS, /* indicate inline dot dentries */ 784 FI_SKIP_WRITES, /* should skip data page writeback */ 785 FI_OPU_WRITE, /* used for opu per file */ 786 FI_DIRTY_FILE, /* indicate regular/symlink has dirty pages */ 787 FI_PREALLOCATED_ALL, /* all blocks for write were preallocated */ 788 FI_HOT_DATA, /* indicate file is hot */ 789 FI_EXTRA_ATTR, /* indicate file has extra attribute */ 790 FI_PROJ_INHERIT, /* indicate file inherits projectid */ 791 FI_PIN_FILE, /* indicate file should not be gced */ 792 FI_VERITY_IN_PROGRESS, /* building fs-verity Merkle tree */ 793 FI_COMPRESSED_FILE, /* indicate file's data can be compressed */ 794 FI_COMPRESS_CORRUPT, /* indicate compressed cluster is corrupted */ 795 FI_MMAP_FILE, /* indicate file was mmapped */ 796 FI_ENABLE_COMPRESS, /* enable compression in "user" compression mode */ 797 FI_COMPRESS_RELEASED, /* compressed blocks were released */ 798 FI_ALIGNED_WRITE, /* enable aligned write */ 799 FI_COW_FILE, /* indicate COW file */ 800 FI_ATOMIC_COMMITTED, /* indicate atomic commit completed except disk sync */ 801 FI_ATOMIC_REPLACE, /* indicate atomic replace */ 802 FI_MAX, /* max flag, never be used */ 803 }; 804 805 struct f2fs_inode_info { 806 struct inode vfs_inode; /* serve a vfs inode */ 807 unsigned long i_flags; /* keep an inode flags for ioctl */ 808 unsigned char i_advise; /* use to give file attribute hints */ 809 unsigned char i_dir_level; /* use for dentry level for large dir */ 810 unsigned int i_current_depth; /* only for directory depth */ 811 /* for gc failure statistic */ 812 unsigned int i_gc_failures[MAX_GC_FAILURE]; 813 unsigned int i_pino; /* parent inode number */ 814 umode_t i_acl_mode; /* keep file acl mode temporarily */ 815 816 /* Use below internally in f2fs*/ 817 unsigned long flags[BITS_TO_LONGS(FI_MAX)]; /* use to pass per-file flags */ 818 struct f2fs_rwsem i_sem; /* protect fi info */ 819 atomic_t dirty_pages; /* # of dirty pages */ 820 f2fs_hash_t chash; /* hash value of given file name */ 821 unsigned int clevel; /* maximum level of given file name */ 822 struct task_struct *task; /* lookup and create consistency */ 823 struct task_struct *cp_task; /* separate cp/wb IO stats*/ 824 struct task_struct *wb_task; /* indicate inode is in context of writeback */ 825 nid_t i_xattr_nid; /* node id that contains xattrs */ 826 loff_t last_disk_size; /* lastly written file size */ 827 spinlock_t i_size_lock; /* protect last_disk_size */ 828 829 #ifdef CONFIG_QUOTA 830 struct dquot __rcu *i_dquot[MAXQUOTAS]; 831 832 /* quota space reservation, managed internally by quota code */ 833 qsize_t i_reserved_quota; 834 #endif 835 struct list_head dirty_list; /* dirty list for dirs and files */ 836 struct list_head gdirty_list; /* linked in global dirty list */ 837 struct task_struct *atomic_write_task; /* store atomic write task */ 838 struct extent_tree *extent_tree[NR_EXTENT_CACHES]; 839 /* cached extent_tree entry */ 840 struct inode *cow_inode; /* copy-on-write inode for atomic write */ 841 842 /* avoid racing between foreground op and gc */ 843 struct f2fs_rwsem i_gc_rwsem[2]; 844 struct f2fs_rwsem i_xattr_sem; /* avoid racing between reading and changing EAs */ 845 846 int i_extra_isize; /* size of extra space located in i_addr */ 847 kprojid_t i_projid; /* id for project quota */ 848 int i_inline_xattr_size; /* inline xattr size */ 849 struct timespec64 i_crtime; /* inode creation time */ 850 struct timespec64 i_disk_time[3];/* inode disk times */ 851 852 /* for file compress */ 853 atomic_t i_compr_blocks; /* # of compressed blocks */ 854 unsigned char i_compress_algorithm; /* algorithm type */ 855 unsigned char i_log_cluster_size; /* log of cluster size */ 856 unsigned char i_compress_level; /* compress level (lz4hc,zstd) */ 857 unsigned char i_compress_flag; /* compress flag */ 858 unsigned int i_cluster_size; /* cluster size */ 859 860 unsigned int atomic_write_cnt; 861 loff_t original_i_size; /* original i_size before atomic write */ 862 }; 863 864 static inline void get_read_extent_info(struct extent_info *ext, 865 struct f2fs_extent *i_ext) 866 { 867 ext->fofs = le32_to_cpu(i_ext->fofs); 868 ext->blk = le32_to_cpu(i_ext->blk); 869 ext->len = le32_to_cpu(i_ext->len); 870 } 871 872 static inline void set_raw_read_extent(struct extent_info *ext, 873 struct f2fs_extent *i_ext) 874 { 875 i_ext->fofs = cpu_to_le32(ext->fofs); 876 i_ext->blk = cpu_to_le32(ext->blk); 877 i_ext->len = cpu_to_le32(ext->len); 878 } 879 880 static inline bool __is_discard_mergeable(struct discard_info *back, 881 struct discard_info *front, unsigned int max_len) 882 { 883 return (back->lstart + back->len == front->lstart) && 884 (back->len + front->len <= max_len); 885 } 886 887 static inline bool __is_discard_back_mergeable(struct discard_info *cur, 888 struct discard_info *back, unsigned int max_len) 889 { 890 return __is_discard_mergeable(back, cur, max_len); 891 } 892 893 static inline bool __is_discard_front_mergeable(struct discard_info *cur, 894 struct discard_info *front, unsigned int max_len) 895 { 896 return __is_discard_mergeable(cur, front, max_len); 897 } 898 899 /* 900 * For free nid management 901 */ 902 enum nid_state { 903 FREE_NID, /* newly added to free nid list */ 904 PREALLOC_NID, /* it is preallocated */ 905 MAX_NID_STATE, 906 }; 907 908 enum nat_state { 909 TOTAL_NAT, 910 DIRTY_NAT, 911 RECLAIMABLE_NAT, 912 MAX_NAT_STATE, 913 }; 914 915 struct f2fs_nm_info { 916 block_t nat_blkaddr; /* base disk address of NAT */ 917 nid_t max_nid; /* maximum possible node ids */ 918 nid_t available_nids; /* # of available node ids */ 919 nid_t next_scan_nid; /* the next nid to be scanned */ 920 nid_t max_rf_node_blocks; /* max # of nodes for recovery */ 921 unsigned int ram_thresh; /* control the memory footprint */ 922 unsigned int ra_nid_pages; /* # of nid pages to be readaheaded */ 923 unsigned int dirty_nats_ratio; /* control dirty nats ratio threshold */ 924 925 /* NAT cache management */ 926 struct radix_tree_root nat_root;/* root of the nat entry cache */ 927 struct radix_tree_root nat_set_root;/* root of the nat set cache */ 928 struct f2fs_rwsem nat_tree_lock; /* protect nat entry tree */ 929 struct list_head nat_entries; /* cached nat entry list (clean) */ 930 spinlock_t nat_list_lock; /* protect clean nat entry list */ 931 unsigned int nat_cnt[MAX_NAT_STATE]; /* the # of cached nat entries */ 932 unsigned int nat_blocks; /* # of nat blocks */ 933 934 /* free node ids management */ 935 struct radix_tree_root free_nid_root;/* root of the free_nid cache */ 936 struct list_head free_nid_list; /* list for free nids excluding preallocated nids */ 937 unsigned int nid_cnt[MAX_NID_STATE]; /* the number of free node id */ 938 spinlock_t nid_list_lock; /* protect nid lists ops */ 939 struct mutex build_lock; /* lock for build free nids */ 940 unsigned char **free_nid_bitmap; 941 unsigned char *nat_block_bitmap; 942 unsigned short *free_nid_count; /* free nid count of NAT block */ 943 944 /* for checkpoint */ 945 char *nat_bitmap; /* NAT bitmap pointer */ 946 947 unsigned int nat_bits_blocks; /* # of nat bits blocks */ 948 unsigned char *nat_bits; /* NAT bits blocks */ 949 unsigned char *full_nat_bits; /* full NAT pages */ 950 unsigned char *empty_nat_bits; /* empty NAT pages */ 951 #ifdef CONFIG_F2FS_CHECK_FS 952 char *nat_bitmap_mir; /* NAT bitmap mirror */ 953 #endif 954 int bitmap_size; /* bitmap size */ 955 }; 956 957 /* 958 * this structure is used as one of function parameters. 959 * all the information are dedicated to a given direct node block determined 960 * by the data offset in a file. 961 */ 962 struct dnode_of_data { 963 struct inode *inode; /* vfs inode pointer */ 964 struct page *inode_page; /* its inode page, NULL is possible */ 965 struct page *node_page; /* cached direct node page */ 966 nid_t nid; /* node id of the direct node block */ 967 unsigned int ofs_in_node; /* data offset in the node page */ 968 bool inode_page_locked; /* inode page is locked or not */ 969 bool node_changed; /* is node block changed */ 970 char cur_level; /* level of hole node page */ 971 char max_level; /* level of current page located */ 972 block_t data_blkaddr; /* block address of the node block */ 973 }; 974 975 static inline void set_new_dnode(struct dnode_of_data *dn, struct inode *inode, 976 struct page *ipage, struct page *npage, nid_t nid) 977 { 978 memset(dn, 0, sizeof(*dn)); 979 dn->inode = inode; 980 dn->inode_page = ipage; 981 dn->node_page = npage; 982 dn->nid = nid; 983 } 984 985 /* 986 * For SIT manager 987 * 988 * By default, there are 6 active log areas across the whole main area. 989 * When considering hot and cold data separation to reduce cleaning overhead, 990 * we split 3 for data logs and 3 for node logs as hot, warm, and cold types, 991 * respectively. 992 * In the current design, you should not change the numbers intentionally. 993 * Instead, as a mount option such as active_logs=x, you can use 2, 4, and 6 994 * logs individually according to the underlying devices. (default: 6) 995 * Just in case, on-disk layout covers maximum 16 logs that consist of 8 for 996 * data and 8 for node logs. 997 */ 998 #define NR_CURSEG_DATA_TYPE (3) 999 #define NR_CURSEG_NODE_TYPE (3) 1000 #define NR_CURSEG_INMEM_TYPE (2) 1001 #define NR_CURSEG_RO_TYPE (2) 1002 #define NR_CURSEG_PERSIST_TYPE (NR_CURSEG_DATA_TYPE + NR_CURSEG_NODE_TYPE) 1003 #define NR_CURSEG_TYPE (NR_CURSEG_INMEM_TYPE + NR_CURSEG_PERSIST_TYPE) 1004 1005 enum { 1006 CURSEG_HOT_DATA = 0, /* directory entry blocks */ 1007 CURSEG_WARM_DATA, /* data blocks */ 1008 CURSEG_COLD_DATA, /* multimedia or GCed data blocks */ 1009 CURSEG_HOT_NODE, /* direct node blocks of directory files */ 1010 CURSEG_WARM_NODE, /* direct node blocks of normal files */ 1011 CURSEG_COLD_NODE, /* indirect node blocks */ 1012 NR_PERSISTENT_LOG, /* number of persistent log */ 1013 CURSEG_COLD_DATA_PINNED = NR_PERSISTENT_LOG, 1014 /* pinned file that needs consecutive block address */ 1015 CURSEG_ALL_DATA_ATGC, /* SSR alloctor in hot/warm/cold data area */ 1016 NO_CHECK_TYPE, /* number of persistent & inmem log */ 1017 }; 1018 1019 struct flush_cmd { 1020 struct completion wait; 1021 struct llist_node llnode; 1022 nid_t ino; 1023 int ret; 1024 }; 1025 1026 struct flush_cmd_control { 1027 struct task_struct *f2fs_issue_flush; /* flush thread */ 1028 wait_queue_head_t flush_wait_queue; /* waiting queue for wake-up */ 1029 atomic_t issued_flush; /* # of issued flushes */ 1030 atomic_t queued_flush; /* # of queued flushes */ 1031 struct llist_head issue_list; /* list for command issue */ 1032 struct llist_node *dispatch_list; /* list for command dispatch */ 1033 }; 1034 1035 struct f2fs_sm_info { 1036 struct sit_info *sit_info; /* whole segment information */ 1037 struct free_segmap_info *free_info; /* free segment information */ 1038 struct dirty_seglist_info *dirty_info; /* dirty segment information */ 1039 struct curseg_info *curseg_array; /* active segment information */ 1040 1041 struct f2fs_rwsem curseg_lock; /* for preventing curseg change */ 1042 1043 block_t seg0_blkaddr; /* block address of 0'th segment */ 1044 block_t main_blkaddr; /* start block address of main area */ 1045 block_t ssa_blkaddr; /* start block address of SSA area */ 1046 1047 unsigned int segment_count; /* total # of segments */ 1048 unsigned int main_segments; /* # of segments in main area */ 1049 unsigned int reserved_segments; /* # of reserved segments */ 1050 unsigned int additional_reserved_segments;/* reserved segs for IO align feature */ 1051 unsigned int ovp_segments; /* # of overprovision segments */ 1052 1053 /* a threshold to reclaim prefree segments */ 1054 unsigned int rec_prefree_segments; 1055 1056 struct list_head sit_entry_set; /* sit entry set list */ 1057 1058 unsigned int ipu_policy; /* in-place-update policy */ 1059 unsigned int min_ipu_util; /* in-place-update threshold */ 1060 unsigned int min_fsync_blocks; /* threshold for fsync */ 1061 unsigned int min_seq_blocks; /* threshold for sequential blocks */ 1062 unsigned int min_hot_blocks; /* threshold for hot block allocation */ 1063 unsigned int min_ssr_sections; /* threshold to trigger SSR allocation */ 1064 1065 /* for flush command control */ 1066 struct flush_cmd_control *fcc_info; 1067 1068 /* for discard command control */ 1069 struct discard_cmd_control *dcc_info; 1070 }; 1071 1072 /* 1073 * For superblock 1074 */ 1075 /* 1076 * COUNT_TYPE for monitoring 1077 * 1078 * f2fs monitors the number of several block types such as on-writeback, 1079 * dirty dentry blocks, dirty node blocks, and dirty meta blocks. 1080 */ 1081 #define WB_DATA_TYPE(p, f) \ 1082 (f || f2fs_is_cp_guaranteed(p) ? F2FS_WB_CP_DATA : F2FS_WB_DATA) 1083 enum count_type { 1084 F2FS_DIRTY_DENTS, 1085 F2FS_DIRTY_DATA, 1086 F2FS_DIRTY_QDATA, 1087 F2FS_DIRTY_NODES, 1088 F2FS_DIRTY_META, 1089 F2FS_DIRTY_IMETA, 1090 F2FS_WB_CP_DATA, 1091 F2FS_WB_DATA, 1092 F2FS_RD_DATA, 1093 F2FS_RD_NODE, 1094 F2FS_RD_META, 1095 F2FS_DIO_WRITE, 1096 F2FS_DIO_READ, 1097 NR_COUNT_TYPE, 1098 }; 1099 1100 /* 1101 * The below are the page types of bios used in submit_bio(). 1102 * The available types are: 1103 * DATA User data pages. It operates as async mode. 1104 * NODE Node pages. It operates as async mode. 1105 * META FS metadata pages such as SIT, NAT, CP. 1106 * NR_PAGE_TYPE The number of page types. 1107 * META_FLUSH Make sure the previous pages are written 1108 * with waiting the bio's completion 1109 * ... Only can be used with META. 1110 */ 1111 #define PAGE_TYPE_OF_BIO(type) ((type) > META ? META : (type)) 1112 enum page_type { 1113 DATA = 0, 1114 NODE = 1, /* should not change this */ 1115 META, 1116 NR_PAGE_TYPE, 1117 META_FLUSH, 1118 IPU, /* the below types are used by tracepoints only. */ 1119 OPU, 1120 }; 1121 1122 enum temp_type { 1123 HOT = 0, /* must be zero for meta bio */ 1124 WARM, 1125 COLD, 1126 NR_TEMP_TYPE, 1127 }; 1128 1129 enum need_lock_type { 1130 LOCK_REQ = 0, 1131 LOCK_DONE, 1132 LOCK_RETRY, 1133 }; 1134 1135 enum cp_reason_type { 1136 CP_NO_NEEDED, 1137 CP_NON_REGULAR, 1138 CP_COMPRESSED, 1139 CP_HARDLINK, 1140 CP_SB_NEED_CP, 1141 CP_WRONG_PINO, 1142 CP_NO_SPC_ROLL, 1143 CP_NODE_NEED_CP, 1144 CP_FASTBOOT_MODE, 1145 CP_SPEC_LOG_NUM, 1146 CP_RECOVER_DIR, 1147 }; 1148 1149 enum iostat_type { 1150 /* WRITE IO */ 1151 APP_DIRECT_IO, /* app direct write IOs */ 1152 APP_BUFFERED_IO, /* app buffered write IOs */ 1153 APP_WRITE_IO, /* app write IOs */ 1154 APP_MAPPED_IO, /* app mapped IOs */ 1155 APP_BUFFERED_CDATA_IO, /* app buffered write IOs on compressed file */ 1156 APP_MAPPED_CDATA_IO, /* app mapped write IOs on compressed file */ 1157 FS_DATA_IO, /* data IOs from kworker/fsync/reclaimer */ 1158 FS_CDATA_IO, /* data IOs from kworker/fsync/reclaimer on compressed file */ 1159 FS_NODE_IO, /* node IOs from kworker/fsync/reclaimer */ 1160 FS_META_IO, /* meta IOs from kworker/reclaimer */ 1161 FS_GC_DATA_IO, /* data IOs from forground gc */ 1162 FS_GC_NODE_IO, /* node IOs from forground gc */ 1163 FS_CP_DATA_IO, /* data IOs from checkpoint */ 1164 FS_CP_NODE_IO, /* node IOs from checkpoint */ 1165 FS_CP_META_IO, /* meta IOs from checkpoint */ 1166 1167 /* READ IO */ 1168 APP_DIRECT_READ_IO, /* app direct read IOs */ 1169 APP_BUFFERED_READ_IO, /* app buffered read IOs */ 1170 APP_READ_IO, /* app read IOs */ 1171 APP_MAPPED_READ_IO, /* app mapped read IOs */ 1172 APP_BUFFERED_CDATA_READ_IO, /* app buffered read IOs on compressed file */ 1173 APP_MAPPED_CDATA_READ_IO, /* app mapped read IOs on compressed file */ 1174 FS_DATA_READ_IO, /* data read IOs */ 1175 FS_GDATA_READ_IO, /* data read IOs from background gc */ 1176 FS_CDATA_READ_IO, /* compressed data read IOs */ 1177 FS_NODE_READ_IO, /* node read IOs */ 1178 FS_META_READ_IO, /* meta read IOs */ 1179 1180 /* other */ 1181 FS_DISCARD_IO, /* discard */ 1182 FS_FLUSH_IO, /* flush */ 1183 FS_ZONE_RESET_IO, /* zone reset */ 1184 NR_IO_TYPE, 1185 }; 1186 1187 struct f2fs_io_info { 1188 struct f2fs_sb_info *sbi; /* f2fs_sb_info pointer */ 1189 nid_t ino; /* inode number */ 1190 enum page_type type; /* contains DATA/NODE/META/META_FLUSH */ 1191 enum temp_type temp; /* contains HOT/WARM/COLD */ 1192 enum req_op op; /* contains REQ_OP_ */ 1193 blk_opf_t op_flags; /* req_flag_bits */ 1194 block_t new_blkaddr; /* new block address to be written */ 1195 block_t old_blkaddr; /* old block address before Cow */ 1196 struct page *page; /* page to be written */ 1197 struct page *encrypted_page; /* encrypted page */ 1198 struct page *compressed_page; /* compressed page */ 1199 struct list_head list; /* serialize IOs */ 1200 unsigned int compr_blocks; /* # of compressed block addresses */ 1201 unsigned int need_lock:8; /* indicate we need to lock cp_rwsem */ 1202 unsigned int version:8; /* version of the node */ 1203 unsigned int submitted:1; /* indicate IO submission */ 1204 unsigned int in_list:1; /* indicate fio is in io_list */ 1205 unsigned int is_por:1; /* indicate IO is from recovery or not */ 1206 unsigned int retry:1; /* need to reallocate block address */ 1207 unsigned int encrypted:1; /* indicate file is encrypted */ 1208 unsigned int post_read:1; /* require post read */ 1209 enum iostat_type io_type; /* io type */ 1210 struct writeback_control *io_wbc; /* writeback control */ 1211 struct bio **bio; /* bio for ipu */ 1212 sector_t *last_block; /* last block number in bio */ 1213 }; 1214 1215 struct bio_entry { 1216 struct bio *bio; 1217 struct list_head list; 1218 }; 1219 1220 #define is_read_io(rw) ((rw) == READ) 1221 struct f2fs_bio_info { 1222 struct f2fs_sb_info *sbi; /* f2fs superblock */ 1223 struct bio *bio; /* bios to merge */ 1224 sector_t last_block_in_bio; /* last block number */ 1225 struct f2fs_io_info fio; /* store buffered io info. */ 1226 #ifdef CONFIG_BLK_DEV_ZONED 1227 struct completion zone_wait; /* condition value for the previous open zone to close */ 1228 struct bio *zone_pending_bio; /* pending bio for the previous zone */ 1229 void *bi_private; /* previous bi_private for pending bio */ 1230 #endif 1231 struct f2fs_rwsem io_rwsem; /* blocking op for bio */ 1232 spinlock_t io_lock; /* serialize DATA/NODE IOs */ 1233 struct list_head io_list; /* track fios */ 1234 struct list_head bio_list; /* bio entry list head */ 1235 struct f2fs_rwsem bio_list_lock; /* lock to protect bio entry list */ 1236 }; 1237 1238 #define FDEV(i) (sbi->devs[i]) 1239 #define RDEV(i) (raw_super->devs[i]) 1240 struct f2fs_dev_info { 1241 struct block_device *bdev; 1242 char path[MAX_PATH_LEN]; 1243 unsigned int total_segments; 1244 block_t start_blk; 1245 block_t end_blk; 1246 #ifdef CONFIG_BLK_DEV_ZONED 1247 unsigned int nr_blkz; /* Total number of zones */ 1248 unsigned long *blkz_seq; /* Bitmap indicating sequential zones */ 1249 #endif 1250 }; 1251 1252 enum inode_type { 1253 DIR_INODE, /* for dirty dir inode */ 1254 FILE_INODE, /* for dirty regular/symlink inode */ 1255 DIRTY_META, /* for all dirtied inode metadata */ 1256 NR_INODE_TYPE, 1257 }; 1258 1259 /* for inner inode cache management */ 1260 struct inode_management { 1261 struct radix_tree_root ino_root; /* ino entry array */ 1262 spinlock_t ino_lock; /* for ino entry lock */ 1263 struct list_head ino_list; /* inode list head */ 1264 unsigned long ino_num; /* number of entries */ 1265 }; 1266 1267 /* for GC_AT */ 1268 struct atgc_management { 1269 bool atgc_enabled; /* ATGC is enabled or not */ 1270 struct rb_root_cached root; /* root of victim rb-tree */ 1271 struct list_head victim_list; /* linked with all victim entries */ 1272 unsigned int victim_count; /* victim count in rb-tree */ 1273 unsigned int candidate_ratio; /* candidate ratio */ 1274 unsigned int max_candidate_count; /* max candidate count */ 1275 unsigned int age_weight; /* age weight, vblock_weight = 100 - age_weight */ 1276 unsigned long long age_threshold; /* age threshold */ 1277 }; 1278 1279 struct f2fs_gc_control { 1280 unsigned int victim_segno; /* target victim segment number */ 1281 int init_gc_type; /* FG_GC or BG_GC */ 1282 bool no_bg_gc; /* check the space and stop bg_gc */ 1283 bool should_migrate_blocks; /* should migrate blocks */ 1284 bool err_gc_skipped; /* return EAGAIN if GC skipped */ 1285 unsigned int nr_free_secs; /* # of free sections to do GC */ 1286 }; 1287 1288 /* 1289 * For s_flag in struct f2fs_sb_info 1290 * Modification on enum should be synchronized with s_flag array 1291 */ 1292 enum { 1293 SBI_IS_DIRTY, /* dirty flag for checkpoint */ 1294 SBI_IS_CLOSE, /* specify unmounting */ 1295 SBI_NEED_FSCK, /* need fsck.f2fs to fix */ 1296 SBI_POR_DOING, /* recovery is doing or not */ 1297 SBI_NEED_SB_WRITE, /* need to recover superblock */ 1298 SBI_NEED_CP, /* need to checkpoint */ 1299 SBI_IS_SHUTDOWN, /* shutdown by ioctl */ 1300 SBI_IS_RECOVERED, /* recovered orphan/data */ 1301 SBI_CP_DISABLED, /* CP was disabled last mount */ 1302 SBI_CP_DISABLED_QUICK, /* CP was disabled quickly */ 1303 SBI_QUOTA_NEED_FLUSH, /* need to flush quota info in CP */ 1304 SBI_QUOTA_SKIP_FLUSH, /* skip flushing quota in current CP */ 1305 SBI_QUOTA_NEED_REPAIR, /* quota file may be corrupted */ 1306 SBI_IS_RESIZEFS, /* resizefs is in process */ 1307 SBI_IS_FREEZING, /* freezefs is in process */ 1308 SBI_IS_WRITABLE, /* remove ro mountoption transiently */ 1309 MAX_SBI_FLAG, 1310 }; 1311 1312 enum { 1313 CP_TIME, 1314 REQ_TIME, 1315 DISCARD_TIME, 1316 GC_TIME, 1317 DISABLE_TIME, 1318 UMOUNT_DISCARD_TIMEOUT, 1319 MAX_TIME, 1320 }; 1321 1322 /* Note that you need to keep synchronization with this gc_mode_names array */ 1323 enum { 1324 GC_NORMAL, 1325 GC_IDLE_CB, 1326 GC_IDLE_GREEDY, 1327 GC_IDLE_AT, 1328 GC_URGENT_HIGH, 1329 GC_URGENT_LOW, 1330 GC_URGENT_MID, 1331 MAX_GC_MODE, 1332 }; 1333 1334 enum { 1335 BGGC_MODE_ON, /* background gc is on */ 1336 BGGC_MODE_OFF, /* background gc is off */ 1337 BGGC_MODE_SYNC, /* 1338 * background gc is on, migrating blocks 1339 * like foreground gc 1340 */ 1341 }; 1342 1343 enum { 1344 FS_MODE_ADAPTIVE, /* use both lfs/ssr allocation */ 1345 FS_MODE_LFS, /* use lfs allocation only */ 1346 FS_MODE_FRAGMENT_SEG, /* segment fragmentation mode */ 1347 FS_MODE_FRAGMENT_BLK, /* block fragmentation mode */ 1348 }; 1349 1350 enum { 1351 ALLOC_MODE_DEFAULT, /* stay default */ 1352 ALLOC_MODE_REUSE, /* reuse segments as much as possible */ 1353 }; 1354 1355 enum fsync_mode { 1356 FSYNC_MODE_POSIX, /* fsync follows posix semantics */ 1357 FSYNC_MODE_STRICT, /* fsync behaves in line with ext4 */ 1358 FSYNC_MODE_NOBARRIER, /* fsync behaves nobarrier based on posix */ 1359 }; 1360 1361 enum { 1362 COMPR_MODE_FS, /* 1363 * automatically compress compression 1364 * enabled files 1365 */ 1366 COMPR_MODE_USER, /* 1367 * automatical compression is disabled. 1368 * user can control the file compression 1369 * using ioctls 1370 */ 1371 }; 1372 1373 enum { 1374 DISCARD_UNIT_BLOCK, /* basic discard unit is block */ 1375 DISCARD_UNIT_SEGMENT, /* basic discard unit is segment */ 1376 DISCARD_UNIT_SECTION, /* basic discard unit is section */ 1377 }; 1378 1379 enum { 1380 MEMORY_MODE_NORMAL, /* memory mode for normal devices */ 1381 MEMORY_MODE_LOW, /* memory mode for low memry devices */ 1382 }; 1383 1384 enum errors_option { 1385 MOUNT_ERRORS_READONLY, /* remount fs ro on errors */ 1386 MOUNT_ERRORS_CONTINUE, /* continue on errors */ 1387 MOUNT_ERRORS_PANIC, /* panic on errors */ 1388 }; 1389 1390 enum { 1391 BACKGROUND, 1392 FOREGROUND, 1393 MAX_CALL_TYPE, 1394 TOTAL_CALL = FOREGROUND, 1395 }; 1396 1397 static inline int f2fs_test_bit(unsigned int nr, char *addr); 1398 static inline void f2fs_set_bit(unsigned int nr, char *addr); 1399 static inline void f2fs_clear_bit(unsigned int nr, char *addr); 1400 1401 /* 1402 * Layout of f2fs page.private: 1403 * 1404 * Layout A: lowest bit should be 1 1405 * | bit0 = 1 | bit1 | bit2 | ... | bit MAX | private data .... | 1406 * bit 0 PAGE_PRIVATE_NOT_POINTER 1407 * bit 1 PAGE_PRIVATE_DUMMY_WRITE 1408 * bit 2 PAGE_PRIVATE_ONGOING_MIGRATION 1409 * bit 3 PAGE_PRIVATE_INLINE_INODE 1410 * bit 4 PAGE_PRIVATE_REF_RESOURCE 1411 * bit 5- f2fs private data 1412 * 1413 * Layout B: lowest bit should be 0 1414 * page.private is a wrapped pointer. 1415 */ 1416 enum { 1417 PAGE_PRIVATE_NOT_POINTER, /* private contains non-pointer data */ 1418 PAGE_PRIVATE_DUMMY_WRITE, /* data page for padding aligned IO */ 1419 PAGE_PRIVATE_ONGOING_MIGRATION, /* data page which is on-going migrating */ 1420 PAGE_PRIVATE_INLINE_INODE, /* inode page contains inline data */ 1421 PAGE_PRIVATE_REF_RESOURCE, /* dirty page has referenced resources */ 1422 PAGE_PRIVATE_MAX 1423 }; 1424 1425 /* For compression */ 1426 enum compress_algorithm_type { 1427 COMPRESS_LZO, 1428 COMPRESS_LZ4, 1429 COMPRESS_ZSTD, 1430 COMPRESS_LZORLE, 1431 COMPRESS_MAX, 1432 }; 1433 1434 enum compress_flag { 1435 COMPRESS_CHKSUM, 1436 COMPRESS_MAX_FLAG, 1437 }; 1438 1439 #define COMPRESS_WATERMARK 20 1440 #define COMPRESS_PERCENT 20 1441 1442 #define COMPRESS_DATA_RESERVED_SIZE 4 1443 struct compress_data { 1444 __le32 clen; /* compressed data size */ 1445 __le32 chksum; /* compressed data chksum */ 1446 __le32 reserved[COMPRESS_DATA_RESERVED_SIZE]; /* reserved */ 1447 u8 cdata[]; /* compressed data */ 1448 }; 1449 1450 #define COMPRESS_HEADER_SIZE (sizeof(struct compress_data)) 1451 1452 #define F2FS_COMPRESSED_PAGE_MAGIC 0xF5F2C000 1453 1454 #define F2FS_ZSTD_DEFAULT_CLEVEL 1 1455 1456 #define COMPRESS_LEVEL_OFFSET 8 1457 1458 /* compress context */ 1459 struct compress_ctx { 1460 struct inode *inode; /* inode the context belong to */ 1461 pgoff_t cluster_idx; /* cluster index number */ 1462 unsigned int cluster_size; /* page count in cluster */ 1463 unsigned int log_cluster_size; /* log of cluster size */ 1464 struct page **rpages; /* pages store raw data in cluster */ 1465 unsigned int nr_rpages; /* total page number in rpages */ 1466 struct page **cpages; /* pages store compressed data in cluster */ 1467 unsigned int nr_cpages; /* total page number in cpages */ 1468 unsigned int valid_nr_cpages; /* valid page number in cpages */ 1469 void *rbuf; /* virtual mapped address on rpages */ 1470 struct compress_data *cbuf; /* virtual mapped address on cpages */ 1471 size_t rlen; /* valid data length in rbuf */ 1472 size_t clen; /* valid data length in cbuf */ 1473 void *private; /* payload buffer for specified compression algorithm */ 1474 void *private2; /* extra payload buffer */ 1475 }; 1476 1477 /* compress context for write IO path */ 1478 struct compress_io_ctx { 1479 u32 magic; /* magic number to indicate page is compressed */ 1480 struct inode *inode; /* inode the context belong to */ 1481 struct page **rpages; /* pages store raw data in cluster */ 1482 unsigned int nr_rpages; /* total page number in rpages */ 1483 atomic_t pending_pages; /* in-flight compressed page count */ 1484 }; 1485 1486 /* Context for decompressing one cluster on the read IO path */ 1487 struct decompress_io_ctx { 1488 u32 magic; /* magic number to indicate page is compressed */ 1489 struct inode *inode; /* inode the context belong to */ 1490 pgoff_t cluster_idx; /* cluster index number */ 1491 unsigned int cluster_size; /* page count in cluster */ 1492 unsigned int log_cluster_size; /* log of cluster size */ 1493 struct page **rpages; /* pages store raw data in cluster */ 1494 unsigned int nr_rpages; /* total page number in rpages */ 1495 struct page **cpages; /* pages store compressed data in cluster */ 1496 unsigned int nr_cpages; /* total page number in cpages */ 1497 struct page **tpages; /* temp pages to pad holes in cluster */ 1498 void *rbuf; /* virtual mapped address on rpages */ 1499 struct compress_data *cbuf; /* virtual mapped address on cpages */ 1500 size_t rlen; /* valid data length in rbuf */ 1501 size_t clen; /* valid data length in cbuf */ 1502 1503 /* 1504 * The number of compressed pages remaining to be read in this cluster. 1505 * This is initially nr_cpages. It is decremented by 1 each time a page 1506 * has been read (or failed to be read). When it reaches 0, the cluster 1507 * is decompressed (or an error is reported). 1508 * 1509 * If an error occurs before all the pages have been submitted for I/O, 1510 * then this will never reach 0. In this case the I/O submitter is 1511 * responsible for calling f2fs_decompress_end_io() instead. 1512 */ 1513 atomic_t remaining_pages; 1514 1515 /* 1516 * Number of references to this decompress_io_ctx. 1517 * 1518 * One reference is held for I/O completion. This reference is dropped 1519 * after the pagecache pages are updated and unlocked -- either after 1520 * decompression (and verity if enabled), or after an error. 1521 * 1522 * In addition, each compressed page holds a reference while it is in a 1523 * bio. These references are necessary prevent compressed pages from 1524 * being freed while they are still in a bio. 1525 */ 1526 refcount_t refcnt; 1527 1528 bool failed; /* IO error occurred before decompression? */ 1529 bool need_verity; /* need fs-verity verification after decompression? */ 1530 void *private; /* payload buffer for specified decompression algorithm */ 1531 void *private2; /* extra payload buffer */ 1532 struct work_struct verity_work; /* work to verify the decompressed pages */ 1533 struct work_struct free_work; /* work for late free this structure itself */ 1534 }; 1535 1536 #define NULL_CLUSTER ((unsigned int)(~0)) 1537 #define MIN_COMPRESS_LOG_SIZE 2 1538 #define MAX_COMPRESS_LOG_SIZE 8 1539 #define MAX_COMPRESS_WINDOW_SIZE(log_size) ((PAGE_SIZE) << (log_size)) 1540 1541 struct f2fs_sb_info { 1542 struct super_block *sb; /* pointer to VFS super block */ 1543 struct proc_dir_entry *s_proc; /* proc entry */ 1544 struct f2fs_super_block *raw_super; /* raw super block pointer */ 1545 struct f2fs_rwsem sb_lock; /* lock for raw super block */ 1546 int valid_super_block; /* valid super block no */ 1547 unsigned long s_flag; /* flags for sbi */ 1548 struct mutex writepages; /* mutex for writepages() */ 1549 1550 #ifdef CONFIG_BLK_DEV_ZONED 1551 unsigned int blocks_per_blkz; /* F2FS blocks per zone */ 1552 #endif 1553 1554 /* for node-related operations */ 1555 struct f2fs_nm_info *nm_info; /* node manager */ 1556 struct inode *node_inode; /* cache node blocks */ 1557 1558 /* for segment-related operations */ 1559 struct f2fs_sm_info *sm_info; /* segment manager */ 1560 1561 /* for bio operations */ 1562 struct f2fs_bio_info *write_io[NR_PAGE_TYPE]; /* for write bios */ 1563 /* keep migration IO order for LFS mode */ 1564 struct f2fs_rwsem io_order_lock; 1565 mempool_t *write_io_dummy; /* Dummy pages */ 1566 pgoff_t page_eio_ofs[NR_PAGE_TYPE]; /* EIO page offset */ 1567 int page_eio_cnt[NR_PAGE_TYPE]; /* EIO count */ 1568 1569 /* for checkpoint */ 1570 struct f2fs_checkpoint *ckpt; /* raw checkpoint pointer */ 1571 int cur_cp_pack; /* remain current cp pack */ 1572 spinlock_t cp_lock; /* for flag in ckpt */ 1573 struct inode *meta_inode; /* cache meta blocks */ 1574 struct f2fs_rwsem cp_global_sem; /* checkpoint procedure lock */ 1575 struct f2fs_rwsem cp_rwsem; /* blocking FS operations */ 1576 struct f2fs_rwsem node_write; /* locking node writes */ 1577 struct f2fs_rwsem node_change; /* locking node change */ 1578 wait_queue_head_t cp_wait; 1579 unsigned long last_time[MAX_TIME]; /* to store time in jiffies */ 1580 long interval_time[MAX_TIME]; /* to store thresholds */ 1581 struct ckpt_req_control cprc_info; /* for checkpoint request control */ 1582 1583 struct inode_management im[MAX_INO_ENTRY]; /* manage inode cache */ 1584 1585 spinlock_t fsync_node_lock; /* for node entry lock */ 1586 struct list_head fsync_node_list; /* node list head */ 1587 unsigned int fsync_seg_id; /* sequence id */ 1588 unsigned int fsync_node_num; /* number of node entries */ 1589 1590 /* for orphan inode, use 0'th array */ 1591 unsigned int max_orphans; /* max orphan inodes */ 1592 1593 /* for inode management */ 1594 struct list_head inode_list[NR_INODE_TYPE]; /* dirty inode list */ 1595 spinlock_t inode_lock[NR_INODE_TYPE]; /* for dirty inode list lock */ 1596 struct mutex flush_lock; /* for flush exclusion */ 1597 1598 /* for extent tree cache */ 1599 struct extent_tree_info extent_tree[NR_EXTENT_CACHES]; 1600 atomic64_t allocated_data_blocks; /* for block age extent_cache */ 1601 1602 /* The threshold used for hot and warm data seperation*/ 1603 unsigned int hot_data_age_threshold; 1604 unsigned int warm_data_age_threshold; 1605 unsigned int last_age_weight; 1606 1607 /* basic filesystem units */ 1608 unsigned int log_sectors_per_block; /* log2 sectors per block */ 1609 unsigned int log_blocksize; /* log2 block size */ 1610 unsigned int blocksize; /* block size */ 1611 unsigned int root_ino_num; /* root inode number*/ 1612 unsigned int node_ino_num; /* node inode number*/ 1613 unsigned int meta_ino_num; /* meta inode number*/ 1614 unsigned int log_blocks_per_seg; /* log2 blocks per segment */ 1615 unsigned int blocks_per_seg; /* blocks per segment */ 1616 unsigned int unusable_blocks_per_sec; /* unusable blocks per section */ 1617 unsigned int segs_per_sec; /* segments per section */ 1618 unsigned int secs_per_zone; /* sections per zone */ 1619 unsigned int total_sections; /* total section count */ 1620 unsigned int total_node_count; /* total node block count */ 1621 unsigned int total_valid_node_count; /* valid node block count */ 1622 int dir_level; /* directory level */ 1623 bool readdir_ra; /* readahead inode in readdir */ 1624 u64 max_io_bytes; /* max io bytes to merge IOs */ 1625 1626 block_t user_block_count; /* # of user blocks */ 1627 block_t total_valid_block_count; /* # of valid blocks */ 1628 block_t discard_blks; /* discard command candidats */ 1629 block_t last_valid_block_count; /* for recovery */ 1630 block_t reserved_blocks; /* configurable reserved blocks */ 1631 block_t current_reserved_blocks; /* current reserved blocks */ 1632 1633 /* Additional tracking for no checkpoint mode */ 1634 block_t unusable_block_count; /* # of blocks saved by last cp */ 1635 1636 unsigned int nquota_files; /* # of quota sysfile */ 1637 struct f2fs_rwsem quota_sem; /* blocking cp for flags */ 1638 1639 /* # of pages, see count_type */ 1640 atomic_t nr_pages[NR_COUNT_TYPE]; 1641 /* # of allocated blocks */ 1642 struct percpu_counter alloc_valid_block_count; 1643 /* # of node block writes as roll forward recovery */ 1644 struct percpu_counter rf_node_block_count; 1645 1646 /* writeback control */ 1647 atomic_t wb_sync_req[META]; /* count # of WB_SYNC threads */ 1648 1649 /* valid inode count */ 1650 struct percpu_counter total_valid_inode_count; 1651 1652 struct f2fs_mount_info mount_opt; /* mount options */ 1653 1654 /* for cleaning operations */ 1655 struct f2fs_rwsem gc_lock; /* 1656 * semaphore for GC, avoid 1657 * race between GC and GC or CP 1658 */ 1659 struct f2fs_gc_kthread *gc_thread; /* GC thread */ 1660 struct atgc_management am; /* atgc management */ 1661 unsigned int cur_victim_sec; /* current victim section num */ 1662 unsigned int gc_mode; /* current GC state */ 1663 unsigned int next_victim_seg[2]; /* next segment in victim section */ 1664 spinlock_t gc_remaining_trials_lock; 1665 /* remaining trial count for GC_URGENT_* and GC_IDLE_* */ 1666 unsigned int gc_remaining_trials; 1667 1668 /* for skip statistic */ 1669 unsigned long long skipped_gc_rwsem; /* FG_GC only */ 1670 1671 /* threshold for gc trials on pinned files */ 1672 u64 gc_pin_file_threshold; 1673 struct f2fs_rwsem pin_sem; 1674 1675 /* maximum # of trials to find a victim segment for SSR and GC */ 1676 unsigned int max_victim_search; 1677 /* migration granularity of garbage collection, unit: segment */ 1678 unsigned int migration_granularity; 1679 1680 /* 1681 * for stat information. 1682 * one is for the LFS mode, and the other is for the SSR mode. 1683 */ 1684 #ifdef CONFIG_F2FS_STAT_FS 1685 struct f2fs_stat_info *stat_info; /* FS status information */ 1686 atomic_t meta_count[META_MAX]; /* # of meta blocks */ 1687 unsigned int segment_count[2]; /* # of allocated segments */ 1688 unsigned int block_count[2]; /* # of allocated blocks */ 1689 atomic_t inplace_count; /* # of inplace update */ 1690 /* # of lookup extent cache */ 1691 atomic64_t total_hit_ext[NR_EXTENT_CACHES]; 1692 /* # of hit rbtree extent node */ 1693 atomic64_t read_hit_rbtree[NR_EXTENT_CACHES]; 1694 /* # of hit cached extent node */ 1695 atomic64_t read_hit_cached[NR_EXTENT_CACHES]; 1696 /* # of hit largest extent node in read extent cache */ 1697 atomic64_t read_hit_largest; 1698 atomic_t inline_xattr; /* # of inline_xattr inodes */ 1699 atomic_t inline_inode; /* # of inline_data inodes */ 1700 atomic_t inline_dir; /* # of inline_dentry inodes */ 1701 atomic_t compr_inode; /* # of compressed inodes */ 1702 atomic64_t compr_blocks; /* # of compressed blocks */ 1703 atomic_t swapfile_inode; /* # of swapfile inodes */ 1704 atomic_t atomic_files; /* # of opened atomic file */ 1705 atomic_t max_aw_cnt; /* max # of atomic writes */ 1706 unsigned int io_skip_bggc; /* skip background gc for in-flight IO */ 1707 unsigned int other_skip_bggc; /* skip background gc for other reasons */ 1708 unsigned int ndirty_inode[NR_INODE_TYPE]; /* # of dirty inodes */ 1709 atomic_t cp_call_count[MAX_CALL_TYPE]; /* # of cp call */ 1710 #endif 1711 spinlock_t stat_lock; /* lock for stat operations */ 1712 1713 /* to attach REQ_META|REQ_FUA flags */ 1714 unsigned int data_io_flag; 1715 unsigned int node_io_flag; 1716 1717 /* For sysfs support */ 1718 struct kobject s_kobj; /* /sys/fs/f2fs/<devname> */ 1719 struct completion s_kobj_unregister; 1720 1721 struct kobject s_stat_kobj; /* /sys/fs/f2fs/<devname>/stat */ 1722 struct completion s_stat_kobj_unregister; 1723 1724 struct kobject s_feature_list_kobj; /* /sys/fs/f2fs/<devname>/feature_list */ 1725 struct completion s_feature_list_kobj_unregister; 1726 1727 /* For shrinker support */ 1728 struct list_head s_list; 1729 struct mutex umount_mutex; 1730 unsigned int shrinker_run_no; 1731 1732 /* For multi devices */ 1733 int s_ndevs; /* number of devices */ 1734 struct f2fs_dev_info *devs; /* for device list */ 1735 unsigned int dirty_device; /* for checkpoint data flush */ 1736 spinlock_t dev_lock; /* protect dirty_device */ 1737 bool aligned_blksize; /* all devices has the same logical blksize */ 1738 1739 /* For write statistics */ 1740 u64 sectors_written_start; 1741 u64 kbytes_written; 1742 1743 /* Reference to checksum algorithm driver via cryptoapi */ 1744 struct crypto_shash *s_chksum_driver; 1745 1746 /* Precomputed FS UUID checksum for seeding other checksums */ 1747 __u32 s_chksum_seed; 1748 1749 struct workqueue_struct *post_read_wq; /* post read workqueue */ 1750 1751 /* 1752 * If we are in irq context, let's update error information into 1753 * on-disk superblock in the work. 1754 */ 1755 struct work_struct s_error_work; 1756 unsigned char errors[MAX_F2FS_ERRORS]; /* error flags */ 1757 unsigned char stop_reason[MAX_STOP_REASON]; /* stop reason */ 1758 spinlock_t error_lock; /* protect errors/stop_reason array */ 1759 bool error_dirty; /* errors of sb is dirty */ 1760 1761 struct kmem_cache *inline_xattr_slab; /* inline xattr entry */ 1762 unsigned int inline_xattr_slab_size; /* default inline xattr slab size */ 1763 1764 /* For reclaimed segs statistics per each GC mode */ 1765 unsigned int gc_segment_mode; /* GC state for reclaimed segments */ 1766 unsigned int gc_reclaimed_segs[MAX_GC_MODE]; /* Reclaimed segs for each mode */ 1767 1768 unsigned long seq_file_ra_mul; /* multiplier for ra_pages of seq. files in fadvise */ 1769 1770 int max_fragment_chunk; /* max chunk size for block fragmentation mode */ 1771 int max_fragment_hole; /* max hole size for block fragmentation mode */ 1772 1773 /* For atomic write statistics */ 1774 atomic64_t current_atomic_write; 1775 s64 peak_atomic_write; 1776 u64 committed_atomic_block; 1777 u64 revoked_atomic_block; 1778 1779 #ifdef CONFIG_F2FS_FS_COMPRESSION 1780 struct kmem_cache *page_array_slab; /* page array entry */ 1781 unsigned int page_array_slab_size; /* default page array slab size */ 1782 1783 /* For runtime compression statistics */ 1784 u64 compr_written_block; 1785 u64 compr_saved_block; 1786 u32 compr_new_inode; 1787 1788 /* For compressed block cache */ 1789 struct inode *compress_inode; /* cache compressed blocks */ 1790 unsigned int compress_percent; /* cache page percentage */ 1791 unsigned int compress_watermark; /* cache page watermark */ 1792 atomic_t compress_page_hit; /* cache hit count */ 1793 #endif 1794 1795 #ifdef CONFIG_F2FS_IOSTAT 1796 /* For app/fs IO statistics */ 1797 spinlock_t iostat_lock; 1798 unsigned long long iostat_count[NR_IO_TYPE]; 1799 unsigned long long iostat_bytes[NR_IO_TYPE]; 1800 unsigned long long prev_iostat_bytes[NR_IO_TYPE]; 1801 bool iostat_enable; 1802 unsigned long iostat_next_period; 1803 unsigned int iostat_period_ms; 1804 1805 /* For io latency related statistics info in one iostat period */ 1806 spinlock_t iostat_lat_lock; 1807 struct iostat_lat_info *iostat_io_lat; 1808 #endif 1809 }; 1810 1811 /* Definitions to access f2fs_sb_info */ 1812 #define BLKS_PER_SEG(sbi) \ 1813 ((sbi)->blocks_per_seg) 1814 #define BLKS_PER_SEC(sbi) \ 1815 ((sbi)->segs_per_sec << (sbi)->log_blocks_per_seg) 1816 #define SEGS_PER_SEC(sbi) \ 1817 ((sbi)->segs_per_sec) 1818 1819 __printf(3, 4) 1820 void f2fs_printk(struct f2fs_sb_info *sbi, bool limit_rate, const char *fmt, ...); 1821 1822 #define f2fs_err(sbi, fmt, ...) \ 1823 f2fs_printk(sbi, false, KERN_ERR fmt, ##__VA_ARGS__) 1824 #define f2fs_warn(sbi, fmt, ...) \ 1825 f2fs_printk(sbi, false, KERN_WARNING fmt, ##__VA_ARGS__) 1826 #define f2fs_notice(sbi, fmt, ...) \ 1827 f2fs_printk(sbi, false, KERN_NOTICE fmt, ##__VA_ARGS__) 1828 #define f2fs_info(sbi, fmt, ...) \ 1829 f2fs_printk(sbi, false, KERN_INFO fmt, ##__VA_ARGS__) 1830 #define f2fs_debug(sbi, fmt, ...) \ 1831 f2fs_printk(sbi, false, KERN_DEBUG fmt, ##__VA_ARGS__) 1832 1833 #define f2fs_err_ratelimited(sbi, fmt, ...) \ 1834 f2fs_printk(sbi, true, KERN_ERR fmt, ##__VA_ARGS__) 1835 #define f2fs_warn_ratelimited(sbi, fmt, ...) \ 1836 f2fs_printk(sbi, true, KERN_WARNING fmt, ##__VA_ARGS__) 1837 #define f2fs_info_ratelimited(sbi, fmt, ...) \ 1838 f2fs_printk(sbi, true, KERN_INFO fmt, ##__VA_ARGS__) 1839 1840 #ifdef CONFIG_F2FS_FAULT_INJECTION 1841 #define time_to_inject(sbi, type) __time_to_inject(sbi, type, __func__, \ 1842 __builtin_return_address(0)) 1843 static inline bool __time_to_inject(struct f2fs_sb_info *sbi, int type, 1844 const char *func, const char *parent_func) 1845 { 1846 struct f2fs_fault_info *ffi = &F2FS_OPTION(sbi).fault_info; 1847 1848 if (!ffi->inject_rate) 1849 return false; 1850 1851 if (!IS_FAULT_SET(ffi, type)) 1852 return false; 1853 1854 atomic_inc(&ffi->inject_ops); 1855 if (atomic_read(&ffi->inject_ops) >= ffi->inject_rate) { 1856 atomic_set(&ffi->inject_ops, 0); 1857 f2fs_info_ratelimited(sbi, "inject %s in %s of %pS", 1858 f2fs_fault_name[type], func, parent_func); 1859 return true; 1860 } 1861 return false; 1862 } 1863 #else 1864 static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type) 1865 { 1866 return false; 1867 } 1868 #endif 1869 1870 /* 1871 * Test if the mounted volume is a multi-device volume. 1872 * - For a single regular disk volume, sbi->s_ndevs is 0. 1873 * - For a single zoned disk volume, sbi->s_ndevs is 1. 1874 * - For a multi-device volume, sbi->s_ndevs is always 2 or more. 1875 */ 1876 static inline bool f2fs_is_multi_device(struct f2fs_sb_info *sbi) 1877 { 1878 return sbi->s_ndevs > 1; 1879 } 1880 1881 static inline void f2fs_update_time(struct f2fs_sb_info *sbi, int type) 1882 { 1883 unsigned long now = jiffies; 1884 1885 sbi->last_time[type] = now; 1886 1887 /* DISCARD_TIME and GC_TIME are based on REQ_TIME */ 1888 if (type == REQ_TIME) { 1889 sbi->last_time[DISCARD_TIME] = now; 1890 sbi->last_time[GC_TIME] = now; 1891 } 1892 } 1893 1894 static inline bool f2fs_time_over(struct f2fs_sb_info *sbi, int type) 1895 { 1896 unsigned long interval = sbi->interval_time[type] * HZ; 1897 1898 return time_after(jiffies, sbi->last_time[type] + interval); 1899 } 1900 1901 static inline unsigned int f2fs_time_to_wait(struct f2fs_sb_info *sbi, 1902 int type) 1903 { 1904 unsigned long interval = sbi->interval_time[type] * HZ; 1905 unsigned int wait_ms = 0; 1906 long delta; 1907 1908 delta = (sbi->last_time[type] + interval) - jiffies; 1909 if (delta > 0) 1910 wait_ms = jiffies_to_msecs(delta); 1911 1912 return wait_ms; 1913 } 1914 1915 /* 1916 * Inline functions 1917 */ 1918 static inline u32 __f2fs_crc32(struct f2fs_sb_info *sbi, u32 crc, 1919 const void *address, unsigned int length) 1920 { 1921 struct { 1922 struct shash_desc shash; 1923 char ctx[4]; 1924 } desc; 1925 int err; 1926 1927 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver) != sizeof(desc.ctx)); 1928 1929 desc.shash.tfm = sbi->s_chksum_driver; 1930 *(u32 *)desc.ctx = crc; 1931 1932 err = crypto_shash_update(&desc.shash, address, length); 1933 BUG_ON(err); 1934 1935 return *(u32 *)desc.ctx; 1936 } 1937 1938 static inline u32 f2fs_crc32(struct f2fs_sb_info *sbi, const void *address, 1939 unsigned int length) 1940 { 1941 return __f2fs_crc32(sbi, F2FS_SUPER_MAGIC, address, length); 1942 } 1943 1944 static inline bool f2fs_crc_valid(struct f2fs_sb_info *sbi, __u32 blk_crc, 1945 void *buf, size_t buf_size) 1946 { 1947 return f2fs_crc32(sbi, buf, buf_size) == blk_crc; 1948 } 1949 1950 static inline u32 f2fs_chksum(struct f2fs_sb_info *sbi, u32 crc, 1951 const void *address, unsigned int length) 1952 { 1953 return __f2fs_crc32(sbi, crc, address, length); 1954 } 1955 1956 static inline struct f2fs_inode_info *F2FS_I(struct inode *inode) 1957 { 1958 return container_of(inode, struct f2fs_inode_info, vfs_inode); 1959 } 1960 1961 static inline struct f2fs_sb_info *F2FS_SB(struct super_block *sb) 1962 { 1963 return sb->s_fs_info; 1964 } 1965 1966 static inline struct f2fs_sb_info *F2FS_I_SB(struct inode *inode) 1967 { 1968 return F2FS_SB(inode->i_sb); 1969 } 1970 1971 static inline struct f2fs_sb_info *F2FS_M_SB(struct address_space *mapping) 1972 { 1973 return F2FS_I_SB(mapping->host); 1974 } 1975 1976 static inline struct f2fs_sb_info *F2FS_P_SB(struct page *page) 1977 { 1978 return F2FS_M_SB(page_file_mapping(page)); 1979 } 1980 1981 static inline struct f2fs_super_block *F2FS_RAW_SUPER(struct f2fs_sb_info *sbi) 1982 { 1983 return (struct f2fs_super_block *)(sbi->raw_super); 1984 } 1985 1986 static inline struct f2fs_checkpoint *F2FS_CKPT(struct f2fs_sb_info *sbi) 1987 { 1988 return (struct f2fs_checkpoint *)(sbi->ckpt); 1989 } 1990 1991 static inline struct f2fs_node *F2FS_NODE(struct page *page) 1992 { 1993 return (struct f2fs_node *)page_address(page); 1994 } 1995 1996 static inline struct f2fs_inode *F2FS_INODE(struct page *page) 1997 { 1998 return &((struct f2fs_node *)page_address(page))->i; 1999 } 2000 2001 static inline struct f2fs_nm_info *NM_I(struct f2fs_sb_info *sbi) 2002 { 2003 return (struct f2fs_nm_info *)(sbi->nm_info); 2004 } 2005 2006 static inline struct f2fs_sm_info *SM_I(struct f2fs_sb_info *sbi) 2007 { 2008 return (struct f2fs_sm_info *)(sbi->sm_info); 2009 } 2010 2011 static inline struct sit_info *SIT_I(struct f2fs_sb_info *sbi) 2012 { 2013 return (struct sit_info *)(SM_I(sbi)->sit_info); 2014 } 2015 2016 static inline struct free_segmap_info *FREE_I(struct f2fs_sb_info *sbi) 2017 { 2018 return (struct free_segmap_info *)(SM_I(sbi)->free_info); 2019 } 2020 2021 static inline struct dirty_seglist_info *DIRTY_I(struct f2fs_sb_info *sbi) 2022 { 2023 return (struct dirty_seglist_info *)(SM_I(sbi)->dirty_info); 2024 } 2025 2026 static inline struct address_space *META_MAPPING(struct f2fs_sb_info *sbi) 2027 { 2028 return sbi->meta_inode->i_mapping; 2029 } 2030 2031 static inline struct address_space *NODE_MAPPING(struct f2fs_sb_info *sbi) 2032 { 2033 return sbi->node_inode->i_mapping; 2034 } 2035 2036 static inline bool is_sbi_flag_set(struct f2fs_sb_info *sbi, unsigned int type) 2037 { 2038 return test_bit(type, &sbi->s_flag); 2039 } 2040 2041 static inline void set_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type) 2042 { 2043 set_bit(type, &sbi->s_flag); 2044 } 2045 2046 static inline void clear_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type) 2047 { 2048 clear_bit(type, &sbi->s_flag); 2049 } 2050 2051 static inline unsigned long long cur_cp_version(struct f2fs_checkpoint *cp) 2052 { 2053 return le64_to_cpu(cp->checkpoint_ver); 2054 } 2055 2056 static inline unsigned long f2fs_qf_ino(struct super_block *sb, int type) 2057 { 2058 if (type < F2FS_MAX_QUOTAS) 2059 return le32_to_cpu(F2FS_SB(sb)->raw_super->qf_ino[type]); 2060 return 0; 2061 } 2062 2063 static inline __u64 cur_cp_crc(struct f2fs_checkpoint *cp) 2064 { 2065 size_t crc_offset = le32_to_cpu(cp->checksum_offset); 2066 return le32_to_cpu(*((__le32 *)((unsigned char *)cp + crc_offset))); 2067 } 2068 2069 static inline bool __is_set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 2070 { 2071 unsigned int ckpt_flags = le32_to_cpu(cp->ckpt_flags); 2072 2073 return ckpt_flags & f; 2074 } 2075 2076 static inline bool is_set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 2077 { 2078 return __is_set_ckpt_flags(F2FS_CKPT(sbi), f); 2079 } 2080 2081 static inline void __set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 2082 { 2083 unsigned int ckpt_flags; 2084 2085 ckpt_flags = le32_to_cpu(cp->ckpt_flags); 2086 ckpt_flags |= f; 2087 cp->ckpt_flags = cpu_to_le32(ckpt_flags); 2088 } 2089 2090 static inline void set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 2091 { 2092 unsigned long flags; 2093 2094 spin_lock_irqsave(&sbi->cp_lock, flags); 2095 __set_ckpt_flags(F2FS_CKPT(sbi), f); 2096 spin_unlock_irqrestore(&sbi->cp_lock, flags); 2097 } 2098 2099 static inline void __clear_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 2100 { 2101 unsigned int ckpt_flags; 2102 2103 ckpt_flags = le32_to_cpu(cp->ckpt_flags); 2104 ckpt_flags &= (~f); 2105 cp->ckpt_flags = cpu_to_le32(ckpt_flags); 2106 } 2107 2108 static inline void clear_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 2109 { 2110 unsigned long flags; 2111 2112 spin_lock_irqsave(&sbi->cp_lock, flags); 2113 __clear_ckpt_flags(F2FS_CKPT(sbi), f); 2114 spin_unlock_irqrestore(&sbi->cp_lock, flags); 2115 } 2116 2117 #define init_f2fs_rwsem(sem) \ 2118 do { \ 2119 static struct lock_class_key __key; \ 2120 \ 2121 __init_f2fs_rwsem((sem), #sem, &__key); \ 2122 } while (0) 2123 2124 static inline void __init_f2fs_rwsem(struct f2fs_rwsem *sem, 2125 const char *sem_name, struct lock_class_key *key) 2126 { 2127 __init_rwsem(&sem->internal_rwsem, sem_name, key); 2128 #ifdef CONFIG_F2FS_UNFAIR_RWSEM 2129 init_waitqueue_head(&sem->read_waiters); 2130 #endif 2131 } 2132 2133 static inline int f2fs_rwsem_is_locked(struct f2fs_rwsem *sem) 2134 { 2135 return rwsem_is_locked(&sem->internal_rwsem); 2136 } 2137 2138 static inline int f2fs_rwsem_is_contended(struct f2fs_rwsem *sem) 2139 { 2140 return rwsem_is_contended(&sem->internal_rwsem); 2141 } 2142 2143 static inline void f2fs_down_read(struct f2fs_rwsem *sem) 2144 { 2145 #ifdef CONFIG_F2FS_UNFAIR_RWSEM 2146 wait_event(sem->read_waiters, down_read_trylock(&sem->internal_rwsem)); 2147 #else 2148 down_read(&sem->internal_rwsem); 2149 #endif 2150 } 2151 2152 static inline int f2fs_down_read_trylock(struct f2fs_rwsem *sem) 2153 { 2154 return down_read_trylock(&sem->internal_rwsem); 2155 } 2156 2157 static inline void f2fs_up_read(struct f2fs_rwsem *sem) 2158 { 2159 up_read(&sem->internal_rwsem); 2160 } 2161 2162 static inline void f2fs_down_write(struct f2fs_rwsem *sem) 2163 { 2164 down_write(&sem->internal_rwsem); 2165 } 2166 2167 #ifdef CONFIG_DEBUG_LOCK_ALLOC 2168 static inline void f2fs_down_read_nested(struct f2fs_rwsem *sem, int subclass) 2169 { 2170 down_read_nested(&sem->internal_rwsem, subclass); 2171 } 2172 2173 static inline void f2fs_down_write_nested(struct f2fs_rwsem *sem, int subclass) 2174 { 2175 down_write_nested(&sem->internal_rwsem, subclass); 2176 } 2177 #else 2178 #define f2fs_down_read_nested(sem, subclass) f2fs_down_read(sem) 2179 #define f2fs_down_write_nested(sem, subclass) f2fs_down_write(sem) 2180 #endif 2181 2182 static inline int f2fs_down_write_trylock(struct f2fs_rwsem *sem) 2183 { 2184 return down_write_trylock(&sem->internal_rwsem); 2185 } 2186 2187 static inline void f2fs_up_write(struct f2fs_rwsem *sem) 2188 { 2189 up_write(&sem->internal_rwsem); 2190 #ifdef CONFIG_F2FS_UNFAIR_RWSEM 2191 wake_up_all(&sem->read_waiters); 2192 #endif 2193 } 2194 2195 static inline void f2fs_lock_op(struct f2fs_sb_info *sbi) 2196 { 2197 f2fs_down_read(&sbi->cp_rwsem); 2198 } 2199 2200 static inline int f2fs_trylock_op(struct f2fs_sb_info *sbi) 2201 { 2202 if (time_to_inject(sbi, FAULT_LOCK_OP)) 2203 return 0; 2204 return f2fs_down_read_trylock(&sbi->cp_rwsem); 2205 } 2206 2207 static inline void f2fs_unlock_op(struct f2fs_sb_info *sbi) 2208 { 2209 f2fs_up_read(&sbi->cp_rwsem); 2210 } 2211 2212 static inline void f2fs_lock_all(struct f2fs_sb_info *sbi) 2213 { 2214 f2fs_down_write(&sbi->cp_rwsem); 2215 } 2216 2217 static inline void f2fs_unlock_all(struct f2fs_sb_info *sbi) 2218 { 2219 f2fs_up_write(&sbi->cp_rwsem); 2220 } 2221 2222 static inline int __get_cp_reason(struct f2fs_sb_info *sbi) 2223 { 2224 int reason = CP_SYNC; 2225 2226 if (test_opt(sbi, FASTBOOT)) 2227 reason = CP_FASTBOOT; 2228 if (is_sbi_flag_set(sbi, SBI_IS_CLOSE)) 2229 reason = CP_UMOUNT; 2230 return reason; 2231 } 2232 2233 static inline bool __remain_node_summaries(int reason) 2234 { 2235 return (reason & (CP_UMOUNT | CP_FASTBOOT)); 2236 } 2237 2238 static inline bool __exist_node_summaries(struct f2fs_sb_info *sbi) 2239 { 2240 return (is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG) || 2241 is_set_ckpt_flags(sbi, CP_FASTBOOT_FLAG)); 2242 } 2243 2244 /* 2245 * Check whether the inode has blocks or not 2246 */ 2247 static inline int F2FS_HAS_BLOCKS(struct inode *inode) 2248 { 2249 block_t xattr_block = F2FS_I(inode)->i_xattr_nid ? 1 : 0; 2250 2251 return (inode->i_blocks >> F2FS_LOG_SECTORS_PER_BLOCK) > xattr_block; 2252 } 2253 2254 static inline bool f2fs_has_xattr_block(unsigned int ofs) 2255 { 2256 return ofs == XATTR_NODE_OFFSET; 2257 } 2258 2259 static inline bool __allow_reserved_blocks(struct f2fs_sb_info *sbi, 2260 struct inode *inode, bool cap) 2261 { 2262 if (!inode) 2263 return true; 2264 if (!test_opt(sbi, RESERVE_ROOT)) 2265 return false; 2266 if (IS_NOQUOTA(inode)) 2267 return true; 2268 if (uid_eq(F2FS_OPTION(sbi).s_resuid, current_fsuid())) 2269 return true; 2270 if (!gid_eq(F2FS_OPTION(sbi).s_resgid, GLOBAL_ROOT_GID) && 2271 in_group_p(F2FS_OPTION(sbi).s_resgid)) 2272 return true; 2273 if (cap && capable(CAP_SYS_RESOURCE)) 2274 return true; 2275 return false; 2276 } 2277 2278 static inline void f2fs_i_blocks_write(struct inode *, block_t, bool, bool); 2279 static inline int inc_valid_block_count(struct f2fs_sb_info *sbi, 2280 struct inode *inode, blkcnt_t *count, bool partial) 2281 { 2282 blkcnt_t diff = 0, release = 0; 2283 block_t avail_user_block_count; 2284 int ret; 2285 2286 ret = dquot_reserve_block(inode, *count); 2287 if (ret) 2288 return ret; 2289 2290 if (time_to_inject(sbi, FAULT_BLOCK)) { 2291 release = *count; 2292 goto release_quota; 2293 } 2294 2295 /* 2296 * let's increase this in prior to actual block count change in order 2297 * for f2fs_sync_file to avoid data races when deciding checkpoint. 2298 */ 2299 percpu_counter_add(&sbi->alloc_valid_block_count, (*count)); 2300 2301 spin_lock(&sbi->stat_lock); 2302 sbi->total_valid_block_count += (block_t)(*count); 2303 avail_user_block_count = sbi->user_block_count - 2304 sbi->current_reserved_blocks; 2305 2306 if (!__allow_reserved_blocks(sbi, inode, true)) 2307 avail_user_block_count -= F2FS_OPTION(sbi).root_reserved_blocks; 2308 2309 if (F2FS_IO_ALIGNED(sbi)) 2310 avail_user_block_count -= sbi->blocks_per_seg * 2311 SM_I(sbi)->additional_reserved_segments; 2312 2313 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) { 2314 if (avail_user_block_count > sbi->unusable_block_count) 2315 avail_user_block_count -= sbi->unusable_block_count; 2316 else 2317 avail_user_block_count = 0; 2318 } 2319 if (unlikely(sbi->total_valid_block_count > avail_user_block_count)) { 2320 if (!partial) { 2321 spin_unlock(&sbi->stat_lock); 2322 goto enospc; 2323 } 2324 2325 diff = sbi->total_valid_block_count - avail_user_block_count; 2326 if (diff > *count) 2327 diff = *count; 2328 *count -= diff; 2329 release = diff; 2330 sbi->total_valid_block_count -= diff; 2331 if (!*count) { 2332 spin_unlock(&sbi->stat_lock); 2333 goto enospc; 2334 } 2335 } 2336 spin_unlock(&sbi->stat_lock); 2337 2338 if (unlikely(release)) { 2339 percpu_counter_sub(&sbi->alloc_valid_block_count, release); 2340 dquot_release_reservation_block(inode, release); 2341 } 2342 f2fs_i_blocks_write(inode, *count, true, true); 2343 return 0; 2344 2345 enospc: 2346 percpu_counter_sub(&sbi->alloc_valid_block_count, release); 2347 release_quota: 2348 dquot_release_reservation_block(inode, release); 2349 return -ENOSPC; 2350 } 2351 2352 #define PAGE_PRIVATE_GET_FUNC(name, flagname) \ 2353 static inline bool page_private_##name(struct page *page) \ 2354 { \ 2355 return PagePrivate(page) && \ 2356 test_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)) && \ 2357 test_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \ 2358 } 2359 2360 #define PAGE_PRIVATE_SET_FUNC(name, flagname) \ 2361 static inline void set_page_private_##name(struct page *page) \ 2362 { \ 2363 if (!PagePrivate(page)) \ 2364 attach_page_private(page, (void *)0); \ 2365 set_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)); \ 2366 set_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \ 2367 } 2368 2369 #define PAGE_PRIVATE_CLEAR_FUNC(name, flagname) \ 2370 static inline void clear_page_private_##name(struct page *page) \ 2371 { \ 2372 clear_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \ 2373 if (page_private(page) == BIT(PAGE_PRIVATE_NOT_POINTER)) \ 2374 detach_page_private(page); \ 2375 } 2376 2377 PAGE_PRIVATE_GET_FUNC(nonpointer, NOT_POINTER); 2378 PAGE_PRIVATE_GET_FUNC(inline, INLINE_INODE); 2379 PAGE_PRIVATE_GET_FUNC(gcing, ONGOING_MIGRATION); 2380 PAGE_PRIVATE_GET_FUNC(dummy, DUMMY_WRITE); 2381 2382 PAGE_PRIVATE_SET_FUNC(reference, REF_RESOURCE); 2383 PAGE_PRIVATE_SET_FUNC(inline, INLINE_INODE); 2384 PAGE_PRIVATE_SET_FUNC(gcing, ONGOING_MIGRATION); 2385 PAGE_PRIVATE_SET_FUNC(dummy, DUMMY_WRITE); 2386 2387 PAGE_PRIVATE_CLEAR_FUNC(reference, REF_RESOURCE); 2388 PAGE_PRIVATE_CLEAR_FUNC(inline, INLINE_INODE); 2389 PAGE_PRIVATE_CLEAR_FUNC(gcing, ONGOING_MIGRATION); 2390 PAGE_PRIVATE_CLEAR_FUNC(dummy, DUMMY_WRITE); 2391 2392 static inline unsigned long get_page_private_data(struct page *page) 2393 { 2394 unsigned long data = page_private(page); 2395 2396 if (!test_bit(PAGE_PRIVATE_NOT_POINTER, &data)) 2397 return 0; 2398 return data >> PAGE_PRIVATE_MAX; 2399 } 2400 2401 static inline void set_page_private_data(struct page *page, unsigned long data) 2402 { 2403 if (!PagePrivate(page)) 2404 attach_page_private(page, (void *)0); 2405 set_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)); 2406 page_private(page) |= data << PAGE_PRIVATE_MAX; 2407 } 2408 2409 static inline void clear_page_private_data(struct page *page) 2410 { 2411 page_private(page) &= GENMASK(PAGE_PRIVATE_MAX - 1, 0); 2412 if (page_private(page) == BIT(PAGE_PRIVATE_NOT_POINTER)) 2413 detach_page_private(page); 2414 } 2415 2416 static inline void clear_page_private_all(struct page *page) 2417 { 2418 clear_page_private_data(page); 2419 clear_page_private_reference(page); 2420 clear_page_private_gcing(page); 2421 clear_page_private_inline(page); 2422 2423 f2fs_bug_on(F2FS_P_SB(page), page_private(page)); 2424 } 2425 2426 static inline void dec_valid_block_count(struct f2fs_sb_info *sbi, 2427 struct inode *inode, 2428 block_t count) 2429 { 2430 blkcnt_t sectors = count << F2FS_LOG_SECTORS_PER_BLOCK; 2431 2432 spin_lock(&sbi->stat_lock); 2433 f2fs_bug_on(sbi, sbi->total_valid_block_count < (block_t) count); 2434 sbi->total_valid_block_count -= (block_t)count; 2435 if (sbi->reserved_blocks && 2436 sbi->current_reserved_blocks < sbi->reserved_blocks) 2437 sbi->current_reserved_blocks = min(sbi->reserved_blocks, 2438 sbi->current_reserved_blocks + count); 2439 spin_unlock(&sbi->stat_lock); 2440 if (unlikely(inode->i_blocks < sectors)) { 2441 f2fs_warn(sbi, "Inconsistent i_blocks, ino:%lu, iblocks:%llu, sectors:%llu", 2442 inode->i_ino, 2443 (unsigned long long)inode->i_blocks, 2444 (unsigned long long)sectors); 2445 set_sbi_flag(sbi, SBI_NEED_FSCK); 2446 return; 2447 } 2448 f2fs_i_blocks_write(inode, count, false, true); 2449 } 2450 2451 static inline void inc_page_count(struct f2fs_sb_info *sbi, int count_type) 2452 { 2453 atomic_inc(&sbi->nr_pages[count_type]); 2454 2455 if (count_type == F2FS_DIRTY_DENTS || 2456 count_type == F2FS_DIRTY_NODES || 2457 count_type == F2FS_DIRTY_META || 2458 count_type == F2FS_DIRTY_QDATA || 2459 count_type == F2FS_DIRTY_IMETA) 2460 set_sbi_flag(sbi, SBI_IS_DIRTY); 2461 } 2462 2463 static inline void inode_inc_dirty_pages(struct inode *inode) 2464 { 2465 atomic_inc(&F2FS_I(inode)->dirty_pages); 2466 inc_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ? 2467 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA); 2468 if (IS_NOQUOTA(inode)) 2469 inc_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA); 2470 } 2471 2472 static inline void dec_page_count(struct f2fs_sb_info *sbi, int count_type) 2473 { 2474 atomic_dec(&sbi->nr_pages[count_type]); 2475 } 2476 2477 static inline void inode_dec_dirty_pages(struct inode *inode) 2478 { 2479 if (!S_ISDIR(inode->i_mode) && !S_ISREG(inode->i_mode) && 2480 !S_ISLNK(inode->i_mode)) 2481 return; 2482 2483 atomic_dec(&F2FS_I(inode)->dirty_pages); 2484 dec_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ? 2485 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA); 2486 if (IS_NOQUOTA(inode)) 2487 dec_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA); 2488 } 2489 2490 static inline void inc_atomic_write_cnt(struct inode *inode) 2491 { 2492 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 2493 struct f2fs_inode_info *fi = F2FS_I(inode); 2494 u64 current_write; 2495 2496 fi->atomic_write_cnt++; 2497 atomic64_inc(&sbi->current_atomic_write); 2498 current_write = atomic64_read(&sbi->current_atomic_write); 2499 if (current_write > sbi->peak_atomic_write) 2500 sbi->peak_atomic_write = current_write; 2501 } 2502 2503 static inline void release_atomic_write_cnt(struct inode *inode) 2504 { 2505 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 2506 struct f2fs_inode_info *fi = F2FS_I(inode); 2507 2508 atomic64_sub(fi->atomic_write_cnt, &sbi->current_atomic_write); 2509 fi->atomic_write_cnt = 0; 2510 } 2511 2512 static inline s64 get_pages(struct f2fs_sb_info *sbi, int count_type) 2513 { 2514 return atomic_read(&sbi->nr_pages[count_type]); 2515 } 2516 2517 static inline int get_dirty_pages(struct inode *inode) 2518 { 2519 return atomic_read(&F2FS_I(inode)->dirty_pages); 2520 } 2521 2522 static inline int get_blocktype_secs(struct f2fs_sb_info *sbi, int block_type) 2523 { 2524 return div_u64(get_pages(sbi, block_type) + BLKS_PER_SEC(sbi) - 1, 2525 BLKS_PER_SEC(sbi)); 2526 } 2527 2528 static inline block_t valid_user_blocks(struct f2fs_sb_info *sbi) 2529 { 2530 return sbi->total_valid_block_count; 2531 } 2532 2533 static inline block_t discard_blocks(struct f2fs_sb_info *sbi) 2534 { 2535 return sbi->discard_blks; 2536 } 2537 2538 static inline unsigned long __bitmap_size(struct f2fs_sb_info *sbi, int flag) 2539 { 2540 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi); 2541 2542 /* return NAT or SIT bitmap */ 2543 if (flag == NAT_BITMAP) 2544 return le32_to_cpu(ckpt->nat_ver_bitmap_bytesize); 2545 else if (flag == SIT_BITMAP) 2546 return le32_to_cpu(ckpt->sit_ver_bitmap_bytesize); 2547 2548 return 0; 2549 } 2550 2551 static inline block_t __cp_payload(struct f2fs_sb_info *sbi) 2552 { 2553 return le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_payload); 2554 } 2555 2556 static inline void *__bitmap_ptr(struct f2fs_sb_info *sbi, int flag) 2557 { 2558 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi); 2559 void *tmp_ptr = &ckpt->sit_nat_version_bitmap; 2560 int offset; 2561 2562 if (is_set_ckpt_flags(sbi, CP_LARGE_NAT_BITMAP_FLAG)) { 2563 offset = (flag == SIT_BITMAP) ? 2564 le32_to_cpu(ckpt->nat_ver_bitmap_bytesize) : 0; 2565 /* 2566 * if large_nat_bitmap feature is enabled, leave checksum 2567 * protection for all nat/sit bitmaps. 2568 */ 2569 return tmp_ptr + offset + sizeof(__le32); 2570 } 2571 2572 if (__cp_payload(sbi) > 0) { 2573 if (flag == NAT_BITMAP) 2574 return tmp_ptr; 2575 else 2576 return (unsigned char *)ckpt + F2FS_BLKSIZE; 2577 } else { 2578 offset = (flag == NAT_BITMAP) ? 2579 le32_to_cpu(ckpt->sit_ver_bitmap_bytesize) : 0; 2580 return tmp_ptr + offset; 2581 } 2582 } 2583 2584 static inline block_t __start_cp_addr(struct f2fs_sb_info *sbi) 2585 { 2586 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr); 2587 2588 if (sbi->cur_cp_pack == 2) 2589 start_addr += BLKS_PER_SEG(sbi); 2590 return start_addr; 2591 } 2592 2593 static inline block_t __start_cp_next_addr(struct f2fs_sb_info *sbi) 2594 { 2595 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr); 2596 2597 if (sbi->cur_cp_pack == 1) 2598 start_addr += BLKS_PER_SEG(sbi); 2599 return start_addr; 2600 } 2601 2602 static inline void __set_cp_next_pack(struct f2fs_sb_info *sbi) 2603 { 2604 sbi->cur_cp_pack = (sbi->cur_cp_pack == 1) ? 2 : 1; 2605 } 2606 2607 static inline block_t __start_sum_addr(struct f2fs_sb_info *sbi) 2608 { 2609 return le32_to_cpu(F2FS_CKPT(sbi)->cp_pack_start_sum); 2610 } 2611 2612 extern void f2fs_mark_inode_dirty_sync(struct inode *inode, bool sync); 2613 static inline int inc_valid_node_count(struct f2fs_sb_info *sbi, 2614 struct inode *inode, bool is_inode) 2615 { 2616 block_t valid_block_count; 2617 unsigned int valid_node_count, user_block_count; 2618 int err; 2619 2620 if (is_inode) { 2621 if (inode) { 2622 err = dquot_alloc_inode(inode); 2623 if (err) 2624 return err; 2625 } 2626 } else { 2627 err = dquot_reserve_block(inode, 1); 2628 if (err) 2629 return err; 2630 } 2631 2632 if (time_to_inject(sbi, FAULT_BLOCK)) 2633 goto enospc; 2634 2635 spin_lock(&sbi->stat_lock); 2636 2637 valid_block_count = sbi->total_valid_block_count + 2638 sbi->current_reserved_blocks + 1; 2639 2640 if (!__allow_reserved_blocks(sbi, inode, false)) 2641 valid_block_count += F2FS_OPTION(sbi).root_reserved_blocks; 2642 2643 if (F2FS_IO_ALIGNED(sbi)) 2644 valid_block_count += sbi->blocks_per_seg * 2645 SM_I(sbi)->additional_reserved_segments; 2646 2647 user_block_count = sbi->user_block_count; 2648 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) 2649 user_block_count -= sbi->unusable_block_count; 2650 2651 if (unlikely(valid_block_count > user_block_count)) { 2652 spin_unlock(&sbi->stat_lock); 2653 goto enospc; 2654 } 2655 2656 valid_node_count = sbi->total_valid_node_count + 1; 2657 if (unlikely(valid_node_count > sbi->total_node_count)) { 2658 spin_unlock(&sbi->stat_lock); 2659 goto enospc; 2660 } 2661 2662 sbi->total_valid_node_count++; 2663 sbi->total_valid_block_count++; 2664 spin_unlock(&sbi->stat_lock); 2665 2666 if (inode) { 2667 if (is_inode) 2668 f2fs_mark_inode_dirty_sync(inode, true); 2669 else 2670 f2fs_i_blocks_write(inode, 1, true, true); 2671 } 2672 2673 percpu_counter_inc(&sbi->alloc_valid_block_count); 2674 return 0; 2675 2676 enospc: 2677 if (is_inode) { 2678 if (inode) 2679 dquot_free_inode(inode); 2680 } else { 2681 dquot_release_reservation_block(inode, 1); 2682 } 2683 return -ENOSPC; 2684 } 2685 2686 static inline void dec_valid_node_count(struct f2fs_sb_info *sbi, 2687 struct inode *inode, bool is_inode) 2688 { 2689 spin_lock(&sbi->stat_lock); 2690 2691 if (unlikely(!sbi->total_valid_block_count || 2692 !sbi->total_valid_node_count)) { 2693 f2fs_warn(sbi, "dec_valid_node_count: inconsistent block counts, total_valid_block:%u, total_valid_node:%u", 2694 sbi->total_valid_block_count, 2695 sbi->total_valid_node_count); 2696 set_sbi_flag(sbi, SBI_NEED_FSCK); 2697 } else { 2698 sbi->total_valid_block_count--; 2699 sbi->total_valid_node_count--; 2700 } 2701 2702 if (sbi->reserved_blocks && 2703 sbi->current_reserved_blocks < sbi->reserved_blocks) 2704 sbi->current_reserved_blocks++; 2705 2706 spin_unlock(&sbi->stat_lock); 2707 2708 if (is_inode) { 2709 dquot_free_inode(inode); 2710 } else { 2711 if (unlikely(inode->i_blocks == 0)) { 2712 f2fs_warn(sbi, "dec_valid_node_count: inconsistent i_blocks, ino:%lu, iblocks:%llu", 2713 inode->i_ino, 2714 (unsigned long long)inode->i_blocks); 2715 set_sbi_flag(sbi, SBI_NEED_FSCK); 2716 return; 2717 } 2718 f2fs_i_blocks_write(inode, 1, false, true); 2719 } 2720 } 2721 2722 static inline unsigned int valid_node_count(struct f2fs_sb_info *sbi) 2723 { 2724 return sbi->total_valid_node_count; 2725 } 2726 2727 static inline void inc_valid_inode_count(struct f2fs_sb_info *sbi) 2728 { 2729 percpu_counter_inc(&sbi->total_valid_inode_count); 2730 } 2731 2732 static inline void dec_valid_inode_count(struct f2fs_sb_info *sbi) 2733 { 2734 percpu_counter_dec(&sbi->total_valid_inode_count); 2735 } 2736 2737 static inline s64 valid_inode_count(struct f2fs_sb_info *sbi) 2738 { 2739 return percpu_counter_sum_positive(&sbi->total_valid_inode_count); 2740 } 2741 2742 static inline struct page *f2fs_grab_cache_page(struct address_space *mapping, 2743 pgoff_t index, bool for_write) 2744 { 2745 struct page *page; 2746 unsigned int flags; 2747 2748 if (IS_ENABLED(CONFIG_F2FS_FAULT_INJECTION)) { 2749 if (!for_write) 2750 page = find_get_page_flags(mapping, index, 2751 FGP_LOCK | FGP_ACCESSED); 2752 else 2753 page = find_lock_page(mapping, index); 2754 if (page) 2755 return page; 2756 2757 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_ALLOC)) 2758 return NULL; 2759 } 2760 2761 if (!for_write) 2762 return grab_cache_page(mapping, index); 2763 2764 flags = memalloc_nofs_save(); 2765 page = grab_cache_page_write_begin(mapping, index); 2766 memalloc_nofs_restore(flags); 2767 2768 return page; 2769 } 2770 2771 static inline struct page *f2fs_pagecache_get_page( 2772 struct address_space *mapping, pgoff_t index, 2773 fgf_t fgp_flags, gfp_t gfp_mask) 2774 { 2775 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_GET)) 2776 return NULL; 2777 2778 return pagecache_get_page(mapping, index, fgp_flags, gfp_mask); 2779 } 2780 2781 static inline void f2fs_put_page(struct page *page, int unlock) 2782 { 2783 if (!page) 2784 return; 2785 2786 if (unlock) { 2787 f2fs_bug_on(F2FS_P_SB(page), !PageLocked(page)); 2788 unlock_page(page); 2789 } 2790 put_page(page); 2791 } 2792 2793 static inline void f2fs_put_dnode(struct dnode_of_data *dn) 2794 { 2795 if (dn->node_page) 2796 f2fs_put_page(dn->node_page, 1); 2797 if (dn->inode_page && dn->node_page != dn->inode_page) 2798 f2fs_put_page(dn->inode_page, 0); 2799 dn->node_page = NULL; 2800 dn->inode_page = NULL; 2801 } 2802 2803 static inline struct kmem_cache *f2fs_kmem_cache_create(const char *name, 2804 size_t size) 2805 { 2806 return kmem_cache_create(name, size, 0, SLAB_RECLAIM_ACCOUNT, NULL); 2807 } 2808 2809 static inline void *f2fs_kmem_cache_alloc_nofail(struct kmem_cache *cachep, 2810 gfp_t flags) 2811 { 2812 void *entry; 2813 2814 entry = kmem_cache_alloc(cachep, flags); 2815 if (!entry) 2816 entry = kmem_cache_alloc(cachep, flags | __GFP_NOFAIL); 2817 return entry; 2818 } 2819 2820 static inline void *f2fs_kmem_cache_alloc(struct kmem_cache *cachep, 2821 gfp_t flags, bool nofail, struct f2fs_sb_info *sbi) 2822 { 2823 if (nofail) 2824 return f2fs_kmem_cache_alloc_nofail(cachep, flags); 2825 2826 if (time_to_inject(sbi, FAULT_SLAB_ALLOC)) 2827 return NULL; 2828 2829 return kmem_cache_alloc(cachep, flags); 2830 } 2831 2832 static inline bool is_inflight_io(struct f2fs_sb_info *sbi, int type) 2833 { 2834 if (get_pages(sbi, F2FS_RD_DATA) || get_pages(sbi, F2FS_RD_NODE) || 2835 get_pages(sbi, F2FS_RD_META) || get_pages(sbi, F2FS_WB_DATA) || 2836 get_pages(sbi, F2FS_WB_CP_DATA) || 2837 get_pages(sbi, F2FS_DIO_READ) || 2838 get_pages(sbi, F2FS_DIO_WRITE)) 2839 return true; 2840 2841 if (type != DISCARD_TIME && SM_I(sbi) && SM_I(sbi)->dcc_info && 2842 atomic_read(&SM_I(sbi)->dcc_info->queued_discard)) 2843 return true; 2844 2845 if (SM_I(sbi) && SM_I(sbi)->fcc_info && 2846 atomic_read(&SM_I(sbi)->fcc_info->queued_flush)) 2847 return true; 2848 return false; 2849 } 2850 2851 static inline bool is_idle(struct f2fs_sb_info *sbi, int type) 2852 { 2853 if (sbi->gc_mode == GC_URGENT_HIGH) 2854 return true; 2855 2856 if (is_inflight_io(sbi, type)) 2857 return false; 2858 2859 if (sbi->gc_mode == GC_URGENT_MID) 2860 return true; 2861 2862 if (sbi->gc_mode == GC_URGENT_LOW && 2863 (type == DISCARD_TIME || type == GC_TIME)) 2864 return true; 2865 2866 return f2fs_time_over(sbi, type); 2867 } 2868 2869 static inline void f2fs_radix_tree_insert(struct radix_tree_root *root, 2870 unsigned long index, void *item) 2871 { 2872 while (radix_tree_insert(root, index, item)) 2873 cond_resched(); 2874 } 2875 2876 #define RAW_IS_INODE(p) ((p)->footer.nid == (p)->footer.ino) 2877 2878 static inline bool IS_INODE(struct page *page) 2879 { 2880 struct f2fs_node *p = F2FS_NODE(page); 2881 2882 return RAW_IS_INODE(p); 2883 } 2884 2885 static inline int offset_in_addr(struct f2fs_inode *i) 2886 { 2887 return (i->i_inline & F2FS_EXTRA_ATTR) ? 2888 (le16_to_cpu(i->i_extra_isize) / sizeof(__le32)) : 0; 2889 } 2890 2891 static inline __le32 *blkaddr_in_node(struct f2fs_node *node) 2892 { 2893 return RAW_IS_INODE(node) ? node->i.i_addr : node->dn.addr; 2894 } 2895 2896 static inline int f2fs_has_extra_attr(struct inode *inode); 2897 static inline block_t data_blkaddr(struct inode *inode, 2898 struct page *node_page, unsigned int offset) 2899 { 2900 struct f2fs_node *raw_node; 2901 __le32 *addr_array; 2902 int base = 0; 2903 bool is_inode = IS_INODE(node_page); 2904 2905 raw_node = F2FS_NODE(node_page); 2906 2907 if (is_inode) { 2908 if (!inode) 2909 /* from GC path only */ 2910 base = offset_in_addr(&raw_node->i); 2911 else if (f2fs_has_extra_attr(inode)) 2912 base = get_extra_isize(inode); 2913 } 2914 2915 addr_array = blkaddr_in_node(raw_node); 2916 return le32_to_cpu(addr_array[base + offset]); 2917 } 2918 2919 static inline block_t f2fs_data_blkaddr(struct dnode_of_data *dn) 2920 { 2921 return data_blkaddr(dn->inode, dn->node_page, dn->ofs_in_node); 2922 } 2923 2924 static inline int f2fs_test_bit(unsigned int nr, char *addr) 2925 { 2926 int mask; 2927 2928 addr += (nr >> 3); 2929 mask = BIT(7 - (nr & 0x07)); 2930 return mask & *addr; 2931 } 2932 2933 static inline void f2fs_set_bit(unsigned int nr, char *addr) 2934 { 2935 int mask; 2936 2937 addr += (nr >> 3); 2938 mask = BIT(7 - (nr & 0x07)); 2939 *addr |= mask; 2940 } 2941 2942 static inline void f2fs_clear_bit(unsigned int nr, char *addr) 2943 { 2944 int mask; 2945 2946 addr += (nr >> 3); 2947 mask = BIT(7 - (nr & 0x07)); 2948 *addr &= ~mask; 2949 } 2950 2951 static inline int f2fs_test_and_set_bit(unsigned int nr, char *addr) 2952 { 2953 int mask; 2954 int ret; 2955 2956 addr += (nr >> 3); 2957 mask = BIT(7 - (nr & 0x07)); 2958 ret = mask & *addr; 2959 *addr |= mask; 2960 return ret; 2961 } 2962 2963 static inline int f2fs_test_and_clear_bit(unsigned int nr, char *addr) 2964 { 2965 int mask; 2966 int ret; 2967 2968 addr += (nr >> 3); 2969 mask = BIT(7 - (nr & 0x07)); 2970 ret = mask & *addr; 2971 *addr &= ~mask; 2972 return ret; 2973 } 2974 2975 static inline void f2fs_change_bit(unsigned int nr, char *addr) 2976 { 2977 int mask; 2978 2979 addr += (nr >> 3); 2980 mask = BIT(7 - (nr & 0x07)); 2981 *addr ^= mask; 2982 } 2983 2984 /* 2985 * On-disk inode flags (f2fs_inode::i_flags) 2986 */ 2987 #define F2FS_COMPR_FL 0x00000004 /* Compress file */ 2988 #define F2FS_SYNC_FL 0x00000008 /* Synchronous updates */ 2989 #define F2FS_IMMUTABLE_FL 0x00000010 /* Immutable file */ 2990 #define F2FS_APPEND_FL 0x00000020 /* writes to file may only append */ 2991 #define F2FS_NODUMP_FL 0x00000040 /* do not dump file */ 2992 #define F2FS_NOATIME_FL 0x00000080 /* do not update atime */ 2993 #define F2FS_NOCOMP_FL 0x00000400 /* Don't compress */ 2994 #define F2FS_INDEX_FL 0x00001000 /* hash-indexed directory */ 2995 #define F2FS_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */ 2996 #define F2FS_PROJINHERIT_FL 0x20000000 /* Create with parents projid */ 2997 #define F2FS_CASEFOLD_FL 0x40000000 /* Casefolded file */ 2998 2999 #define F2FS_QUOTA_DEFAULT_FL (F2FS_NOATIME_FL | F2FS_IMMUTABLE_FL) 3000 3001 /* Flags that should be inherited by new inodes from their parent. */ 3002 #define F2FS_FL_INHERITED (F2FS_SYNC_FL | F2FS_NODUMP_FL | F2FS_NOATIME_FL | \ 3003 F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \ 3004 F2FS_CASEFOLD_FL) 3005 3006 /* Flags that are appropriate for regular files (all but dir-specific ones). */ 3007 #define F2FS_REG_FLMASK (~(F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \ 3008 F2FS_CASEFOLD_FL)) 3009 3010 /* Flags that are appropriate for non-directories/regular files. */ 3011 #define F2FS_OTHER_FLMASK (F2FS_NODUMP_FL | F2FS_NOATIME_FL) 3012 3013 static inline __u32 f2fs_mask_flags(umode_t mode, __u32 flags) 3014 { 3015 if (S_ISDIR(mode)) 3016 return flags; 3017 else if (S_ISREG(mode)) 3018 return flags & F2FS_REG_FLMASK; 3019 else 3020 return flags & F2FS_OTHER_FLMASK; 3021 } 3022 3023 static inline void __mark_inode_dirty_flag(struct inode *inode, 3024 int flag, bool set) 3025 { 3026 switch (flag) { 3027 case FI_INLINE_XATTR: 3028 case FI_INLINE_DATA: 3029 case FI_INLINE_DENTRY: 3030 case FI_NEW_INODE: 3031 if (set) 3032 return; 3033 fallthrough; 3034 case FI_DATA_EXIST: 3035 case FI_INLINE_DOTS: 3036 case FI_PIN_FILE: 3037 case FI_COMPRESS_RELEASED: 3038 case FI_ATOMIC_COMMITTED: 3039 f2fs_mark_inode_dirty_sync(inode, true); 3040 } 3041 } 3042 3043 static inline void set_inode_flag(struct inode *inode, int flag) 3044 { 3045 set_bit(flag, F2FS_I(inode)->flags); 3046 __mark_inode_dirty_flag(inode, flag, true); 3047 } 3048 3049 static inline int is_inode_flag_set(struct inode *inode, int flag) 3050 { 3051 return test_bit(flag, F2FS_I(inode)->flags); 3052 } 3053 3054 static inline void clear_inode_flag(struct inode *inode, int flag) 3055 { 3056 clear_bit(flag, F2FS_I(inode)->flags); 3057 __mark_inode_dirty_flag(inode, flag, false); 3058 } 3059 3060 static inline bool f2fs_verity_in_progress(struct inode *inode) 3061 { 3062 return IS_ENABLED(CONFIG_FS_VERITY) && 3063 is_inode_flag_set(inode, FI_VERITY_IN_PROGRESS); 3064 } 3065 3066 static inline void set_acl_inode(struct inode *inode, umode_t mode) 3067 { 3068 F2FS_I(inode)->i_acl_mode = mode; 3069 set_inode_flag(inode, FI_ACL_MODE); 3070 f2fs_mark_inode_dirty_sync(inode, false); 3071 } 3072 3073 static inline void f2fs_i_links_write(struct inode *inode, bool inc) 3074 { 3075 if (inc) 3076 inc_nlink(inode); 3077 else 3078 drop_nlink(inode); 3079 f2fs_mark_inode_dirty_sync(inode, true); 3080 } 3081 3082 static inline void f2fs_i_blocks_write(struct inode *inode, 3083 block_t diff, bool add, bool claim) 3084 { 3085 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE); 3086 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER); 3087 3088 /* add = 1, claim = 1 should be dquot_reserve_block in pair */ 3089 if (add) { 3090 if (claim) 3091 dquot_claim_block(inode, diff); 3092 else 3093 dquot_alloc_block_nofail(inode, diff); 3094 } else { 3095 dquot_free_block(inode, diff); 3096 } 3097 3098 f2fs_mark_inode_dirty_sync(inode, true); 3099 if (clean || recover) 3100 set_inode_flag(inode, FI_AUTO_RECOVER); 3101 } 3102 3103 static inline bool f2fs_is_atomic_file(struct inode *inode); 3104 3105 static inline void f2fs_i_size_write(struct inode *inode, loff_t i_size) 3106 { 3107 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE); 3108 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER); 3109 3110 if (i_size_read(inode) == i_size) 3111 return; 3112 3113 i_size_write(inode, i_size); 3114 3115 if (f2fs_is_atomic_file(inode)) 3116 return; 3117 3118 f2fs_mark_inode_dirty_sync(inode, true); 3119 if (clean || recover) 3120 set_inode_flag(inode, FI_AUTO_RECOVER); 3121 } 3122 3123 static inline void f2fs_i_depth_write(struct inode *inode, unsigned int depth) 3124 { 3125 F2FS_I(inode)->i_current_depth = depth; 3126 f2fs_mark_inode_dirty_sync(inode, true); 3127 } 3128 3129 static inline void f2fs_i_gc_failures_write(struct inode *inode, 3130 unsigned int count) 3131 { 3132 F2FS_I(inode)->i_gc_failures[GC_FAILURE_PIN] = count; 3133 f2fs_mark_inode_dirty_sync(inode, true); 3134 } 3135 3136 static inline void f2fs_i_xnid_write(struct inode *inode, nid_t xnid) 3137 { 3138 F2FS_I(inode)->i_xattr_nid = xnid; 3139 f2fs_mark_inode_dirty_sync(inode, true); 3140 } 3141 3142 static inline void f2fs_i_pino_write(struct inode *inode, nid_t pino) 3143 { 3144 F2FS_I(inode)->i_pino = pino; 3145 f2fs_mark_inode_dirty_sync(inode, true); 3146 } 3147 3148 static inline void get_inline_info(struct inode *inode, struct f2fs_inode *ri) 3149 { 3150 struct f2fs_inode_info *fi = F2FS_I(inode); 3151 3152 if (ri->i_inline & F2FS_INLINE_XATTR) 3153 set_bit(FI_INLINE_XATTR, fi->flags); 3154 if (ri->i_inline & F2FS_INLINE_DATA) 3155 set_bit(FI_INLINE_DATA, fi->flags); 3156 if (ri->i_inline & F2FS_INLINE_DENTRY) 3157 set_bit(FI_INLINE_DENTRY, fi->flags); 3158 if (ri->i_inline & F2FS_DATA_EXIST) 3159 set_bit(FI_DATA_EXIST, fi->flags); 3160 if (ri->i_inline & F2FS_INLINE_DOTS) 3161 set_bit(FI_INLINE_DOTS, fi->flags); 3162 if (ri->i_inline & F2FS_EXTRA_ATTR) 3163 set_bit(FI_EXTRA_ATTR, fi->flags); 3164 if (ri->i_inline & F2FS_PIN_FILE) 3165 set_bit(FI_PIN_FILE, fi->flags); 3166 if (ri->i_inline & F2FS_COMPRESS_RELEASED) 3167 set_bit(FI_COMPRESS_RELEASED, fi->flags); 3168 } 3169 3170 static inline void set_raw_inline(struct inode *inode, struct f2fs_inode *ri) 3171 { 3172 ri->i_inline = 0; 3173 3174 if (is_inode_flag_set(inode, FI_INLINE_XATTR)) 3175 ri->i_inline |= F2FS_INLINE_XATTR; 3176 if (is_inode_flag_set(inode, FI_INLINE_DATA)) 3177 ri->i_inline |= F2FS_INLINE_DATA; 3178 if (is_inode_flag_set(inode, FI_INLINE_DENTRY)) 3179 ri->i_inline |= F2FS_INLINE_DENTRY; 3180 if (is_inode_flag_set(inode, FI_DATA_EXIST)) 3181 ri->i_inline |= F2FS_DATA_EXIST; 3182 if (is_inode_flag_set(inode, FI_INLINE_DOTS)) 3183 ri->i_inline |= F2FS_INLINE_DOTS; 3184 if (is_inode_flag_set(inode, FI_EXTRA_ATTR)) 3185 ri->i_inline |= F2FS_EXTRA_ATTR; 3186 if (is_inode_flag_set(inode, FI_PIN_FILE)) 3187 ri->i_inline |= F2FS_PIN_FILE; 3188 if (is_inode_flag_set(inode, FI_COMPRESS_RELEASED)) 3189 ri->i_inline |= F2FS_COMPRESS_RELEASED; 3190 } 3191 3192 static inline int f2fs_has_extra_attr(struct inode *inode) 3193 { 3194 return is_inode_flag_set(inode, FI_EXTRA_ATTR); 3195 } 3196 3197 static inline int f2fs_has_inline_xattr(struct inode *inode) 3198 { 3199 return is_inode_flag_set(inode, FI_INLINE_XATTR); 3200 } 3201 3202 static inline int f2fs_compressed_file(struct inode *inode) 3203 { 3204 return S_ISREG(inode->i_mode) && 3205 is_inode_flag_set(inode, FI_COMPRESSED_FILE); 3206 } 3207 3208 static inline bool f2fs_need_compress_data(struct inode *inode) 3209 { 3210 int compress_mode = F2FS_OPTION(F2FS_I_SB(inode)).compress_mode; 3211 3212 if (!f2fs_compressed_file(inode)) 3213 return false; 3214 3215 if (compress_mode == COMPR_MODE_FS) 3216 return true; 3217 else if (compress_mode == COMPR_MODE_USER && 3218 is_inode_flag_set(inode, FI_ENABLE_COMPRESS)) 3219 return true; 3220 3221 return false; 3222 } 3223 3224 static inline unsigned int addrs_per_inode(struct inode *inode) 3225 { 3226 unsigned int addrs = CUR_ADDRS_PER_INODE(inode) - 3227 get_inline_xattr_addrs(inode); 3228 3229 if (!f2fs_compressed_file(inode)) 3230 return addrs; 3231 return ALIGN_DOWN(addrs, F2FS_I(inode)->i_cluster_size); 3232 } 3233 3234 static inline unsigned int addrs_per_block(struct inode *inode) 3235 { 3236 if (!f2fs_compressed_file(inode)) 3237 return DEF_ADDRS_PER_BLOCK; 3238 return ALIGN_DOWN(DEF_ADDRS_PER_BLOCK, F2FS_I(inode)->i_cluster_size); 3239 } 3240 3241 static inline void *inline_xattr_addr(struct inode *inode, struct page *page) 3242 { 3243 struct f2fs_inode *ri = F2FS_INODE(page); 3244 3245 return (void *)&(ri->i_addr[DEF_ADDRS_PER_INODE - 3246 get_inline_xattr_addrs(inode)]); 3247 } 3248 3249 static inline int inline_xattr_size(struct inode *inode) 3250 { 3251 if (f2fs_has_inline_xattr(inode)) 3252 return get_inline_xattr_addrs(inode) * sizeof(__le32); 3253 return 0; 3254 } 3255 3256 /* 3257 * Notice: check inline_data flag without inode page lock is unsafe. 3258 * It could change at any time by f2fs_convert_inline_page(). 3259 */ 3260 static inline int f2fs_has_inline_data(struct inode *inode) 3261 { 3262 return is_inode_flag_set(inode, FI_INLINE_DATA); 3263 } 3264 3265 static inline int f2fs_exist_data(struct inode *inode) 3266 { 3267 return is_inode_flag_set(inode, FI_DATA_EXIST); 3268 } 3269 3270 static inline int f2fs_has_inline_dots(struct inode *inode) 3271 { 3272 return is_inode_flag_set(inode, FI_INLINE_DOTS); 3273 } 3274 3275 static inline int f2fs_is_mmap_file(struct inode *inode) 3276 { 3277 return is_inode_flag_set(inode, FI_MMAP_FILE); 3278 } 3279 3280 static inline bool f2fs_is_pinned_file(struct inode *inode) 3281 { 3282 return is_inode_flag_set(inode, FI_PIN_FILE); 3283 } 3284 3285 static inline bool f2fs_is_atomic_file(struct inode *inode) 3286 { 3287 return is_inode_flag_set(inode, FI_ATOMIC_FILE); 3288 } 3289 3290 static inline bool f2fs_is_cow_file(struct inode *inode) 3291 { 3292 return is_inode_flag_set(inode, FI_COW_FILE); 3293 } 3294 3295 static inline __le32 *get_dnode_addr(struct inode *inode, 3296 struct page *node_page); 3297 static inline void *inline_data_addr(struct inode *inode, struct page *page) 3298 { 3299 __le32 *addr = get_dnode_addr(inode, page); 3300 3301 return (void *)(addr + DEF_INLINE_RESERVED_SIZE); 3302 } 3303 3304 static inline int f2fs_has_inline_dentry(struct inode *inode) 3305 { 3306 return is_inode_flag_set(inode, FI_INLINE_DENTRY); 3307 } 3308 3309 static inline int is_file(struct inode *inode, int type) 3310 { 3311 return F2FS_I(inode)->i_advise & type; 3312 } 3313 3314 static inline void set_file(struct inode *inode, int type) 3315 { 3316 if (is_file(inode, type)) 3317 return; 3318 F2FS_I(inode)->i_advise |= type; 3319 f2fs_mark_inode_dirty_sync(inode, true); 3320 } 3321 3322 static inline void clear_file(struct inode *inode, int type) 3323 { 3324 if (!is_file(inode, type)) 3325 return; 3326 F2FS_I(inode)->i_advise &= ~type; 3327 f2fs_mark_inode_dirty_sync(inode, true); 3328 } 3329 3330 static inline bool f2fs_is_time_consistent(struct inode *inode) 3331 { 3332 struct timespec64 ctime = inode_get_ctime(inode); 3333 3334 if (!timespec64_equal(F2FS_I(inode)->i_disk_time, &inode->i_atime)) 3335 return false; 3336 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 1, &ctime)) 3337 return false; 3338 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 2, &inode->i_mtime)) 3339 return false; 3340 return true; 3341 } 3342 3343 static inline bool f2fs_skip_inode_update(struct inode *inode, int dsync) 3344 { 3345 bool ret; 3346 3347 if (dsync) { 3348 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 3349 3350 spin_lock(&sbi->inode_lock[DIRTY_META]); 3351 ret = list_empty(&F2FS_I(inode)->gdirty_list); 3352 spin_unlock(&sbi->inode_lock[DIRTY_META]); 3353 return ret; 3354 } 3355 if (!is_inode_flag_set(inode, FI_AUTO_RECOVER) || 3356 file_keep_isize(inode) || 3357 i_size_read(inode) & ~PAGE_MASK) 3358 return false; 3359 3360 if (!f2fs_is_time_consistent(inode)) 3361 return false; 3362 3363 spin_lock(&F2FS_I(inode)->i_size_lock); 3364 ret = F2FS_I(inode)->last_disk_size == i_size_read(inode); 3365 spin_unlock(&F2FS_I(inode)->i_size_lock); 3366 3367 return ret; 3368 } 3369 3370 static inline bool f2fs_readonly(struct super_block *sb) 3371 { 3372 return sb_rdonly(sb); 3373 } 3374 3375 static inline bool f2fs_cp_error(struct f2fs_sb_info *sbi) 3376 { 3377 return is_set_ckpt_flags(sbi, CP_ERROR_FLAG); 3378 } 3379 3380 static inline bool is_dot_dotdot(const u8 *name, size_t len) 3381 { 3382 if (len == 1 && name[0] == '.') 3383 return true; 3384 3385 if (len == 2 && name[0] == '.' && name[1] == '.') 3386 return true; 3387 3388 return false; 3389 } 3390 3391 static inline void *f2fs_kmalloc(struct f2fs_sb_info *sbi, 3392 size_t size, gfp_t flags) 3393 { 3394 if (time_to_inject(sbi, FAULT_KMALLOC)) 3395 return NULL; 3396 3397 return kmalloc(size, flags); 3398 } 3399 3400 static inline void *f2fs_getname(struct f2fs_sb_info *sbi) 3401 { 3402 if (time_to_inject(sbi, FAULT_KMALLOC)) 3403 return NULL; 3404 3405 return __getname(); 3406 } 3407 3408 static inline void f2fs_putname(char *buf) 3409 { 3410 __putname(buf); 3411 } 3412 3413 static inline void *f2fs_kzalloc(struct f2fs_sb_info *sbi, 3414 size_t size, gfp_t flags) 3415 { 3416 return f2fs_kmalloc(sbi, size, flags | __GFP_ZERO); 3417 } 3418 3419 static inline void *f2fs_kvmalloc(struct f2fs_sb_info *sbi, 3420 size_t size, gfp_t flags) 3421 { 3422 if (time_to_inject(sbi, FAULT_KVMALLOC)) 3423 return NULL; 3424 3425 return kvmalloc(size, flags); 3426 } 3427 3428 static inline void *f2fs_kvzalloc(struct f2fs_sb_info *sbi, 3429 size_t size, gfp_t flags) 3430 { 3431 return f2fs_kvmalloc(sbi, size, flags | __GFP_ZERO); 3432 } 3433 3434 static inline int get_extra_isize(struct inode *inode) 3435 { 3436 return F2FS_I(inode)->i_extra_isize / sizeof(__le32); 3437 } 3438 3439 static inline int get_inline_xattr_addrs(struct inode *inode) 3440 { 3441 return F2FS_I(inode)->i_inline_xattr_size; 3442 } 3443 3444 static inline __le32 *get_dnode_addr(struct inode *inode, 3445 struct page *node_page) 3446 { 3447 int base = 0; 3448 3449 if (IS_INODE(node_page) && f2fs_has_extra_attr(inode)) 3450 base = get_extra_isize(inode); 3451 3452 return blkaddr_in_node(F2FS_NODE(node_page)) + base; 3453 } 3454 3455 #define f2fs_get_inode_mode(i) \ 3456 ((is_inode_flag_set(i, FI_ACL_MODE)) ? \ 3457 (F2FS_I(i)->i_acl_mode) : ((i)->i_mode)) 3458 3459 #define F2FS_MIN_EXTRA_ATTR_SIZE (sizeof(__le32)) 3460 3461 #define F2FS_TOTAL_EXTRA_ATTR_SIZE \ 3462 (offsetof(struct f2fs_inode, i_extra_end) - \ 3463 offsetof(struct f2fs_inode, i_extra_isize)) \ 3464 3465 #define F2FS_OLD_ATTRIBUTE_SIZE (offsetof(struct f2fs_inode, i_addr)) 3466 #define F2FS_FITS_IN_INODE(f2fs_inode, extra_isize, field) \ 3467 ((offsetof(typeof(*(f2fs_inode)), field) + \ 3468 sizeof((f2fs_inode)->field)) \ 3469 <= (F2FS_OLD_ATTRIBUTE_SIZE + (extra_isize))) \ 3470 3471 #define __is_large_section(sbi) (SEGS_PER_SEC(sbi) > 1) 3472 3473 #define __is_meta_io(fio) (PAGE_TYPE_OF_BIO((fio)->type) == META) 3474 3475 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi, 3476 block_t blkaddr, int type); 3477 static inline void verify_blkaddr(struct f2fs_sb_info *sbi, 3478 block_t blkaddr, int type) 3479 { 3480 if (!f2fs_is_valid_blkaddr(sbi, blkaddr, type)) 3481 f2fs_err(sbi, "invalid blkaddr: %u, type: %d, run fsck to fix.", 3482 blkaddr, type); 3483 } 3484 3485 static inline bool __is_valid_data_blkaddr(block_t blkaddr) 3486 { 3487 if (blkaddr == NEW_ADDR || blkaddr == NULL_ADDR || 3488 blkaddr == COMPRESS_ADDR) 3489 return false; 3490 return true; 3491 } 3492 3493 /* 3494 * file.c 3495 */ 3496 int f2fs_sync_file(struct file *file, loff_t start, loff_t end, int datasync); 3497 int f2fs_do_truncate_blocks(struct inode *inode, u64 from, bool lock); 3498 int f2fs_truncate_blocks(struct inode *inode, u64 from, bool lock); 3499 int f2fs_truncate(struct inode *inode); 3500 int f2fs_getattr(struct mnt_idmap *idmap, const struct path *path, 3501 struct kstat *stat, u32 request_mask, unsigned int flags); 3502 int f2fs_setattr(struct mnt_idmap *idmap, struct dentry *dentry, 3503 struct iattr *attr); 3504 int f2fs_truncate_hole(struct inode *inode, pgoff_t pg_start, pgoff_t pg_end); 3505 void f2fs_truncate_data_blocks_range(struct dnode_of_data *dn, int count); 3506 int f2fs_precache_extents(struct inode *inode); 3507 int f2fs_fileattr_get(struct dentry *dentry, struct fileattr *fa); 3508 int f2fs_fileattr_set(struct mnt_idmap *idmap, 3509 struct dentry *dentry, struct fileattr *fa); 3510 long f2fs_ioctl(struct file *filp, unsigned int cmd, unsigned long arg); 3511 long f2fs_compat_ioctl(struct file *file, unsigned int cmd, unsigned long arg); 3512 int f2fs_transfer_project_quota(struct inode *inode, kprojid_t kprojid); 3513 int f2fs_pin_file_control(struct inode *inode, bool inc); 3514 3515 /* 3516 * inode.c 3517 */ 3518 void f2fs_set_inode_flags(struct inode *inode); 3519 bool f2fs_inode_chksum_verify(struct f2fs_sb_info *sbi, struct page *page); 3520 void f2fs_inode_chksum_set(struct f2fs_sb_info *sbi, struct page *page); 3521 struct inode *f2fs_iget(struct super_block *sb, unsigned long ino); 3522 struct inode *f2fs_iget_retry(struct super_block *sb, unsigned long ino); 3523 int f2fs_try_to_free_nats(struct f2fs_sb_info *sbi, int nr_shrink); 3524 void f2fs_update_inode(struct inode *inode, struct page *node_page); 3525 void f2fs_update_inode_page(struct inode *inode); 3526 int f2fs_write_inode(struct inode *inode, struct writeback_control *wbc); 3527 void f2fs_evict_inode(struct inode *inode); 3528 void f2fs_handle_failed_inode(struct inode *inode); 3529 3530 /* 3531 * namei.c 3532 */ 3533 int f2fs_update_extension_list(struct f2fs_sb_info *sbi, const char *name, 3534 bool hot, bool set); 3535 struct dentry *f2fs_get_parent(struct dentry *child); 3536 int f2fs_get_tmpfile(struct mnt_idmap *idmap, struct inode *dir, 3537 struct inode **new_inode); 3538 3539 /* 3540 * dir.c 3541 */ 3542 int f2fs_init_casefolded_name(const struct inode *dir, 3543 struct f2fs_filename *fname); 3544 int f2fs_setup_filename(struct inode *dir, const struct qstr *iname, 3545 int lookup, struct f2fs_filename *fname); 3546 int f2fs_prepare_lookup(struct inode *dir, struct dentry *dentry, 3547 struct f2fs_filename *fname); 3548 void f2fs_free_filename(struct f2fs_filename *fname); 3549 struct f2fs_dir_entry *f2fs_find_target_dentry(const struct f2fs_dentry_ptr *d, 3550 const struct f2fs_filename *fname, int *max_slots); 3551 int f2fs_fill_dentries(struct dir_context *ctx, struct f2fs_dentry_ptr *d, 3552 unsigned int start_pos, struct fscrypt_str *fstr); 3553 void f2fs_do_make_empty_dir(struct inode *inode, struct inode *parent, 3554 struct f2fs_dentry_ptr *d); 3555 struct page *f2fs_init_inode_metadata(struct inode *inode, struct inode *dir, 3556 const struct f2fs_filename *fname, struct page *dpage); 3557 void f2fs_update_parent_metadata(struct inode *dir, struct inode *inode, 3558 unsigned int current_depth); 3559 int f2fs_room_for_filename(const void *bitmap, int slots, int max_slots); 3560 void f2fs_drop_nlink(struct inode *dir, struct inode *inode); 3561 struct f2fs_dir_entry *__f2fs_find_entry(struct inode *dir, 3562 const struct f2fs_filename *fname, 3563 struct page **res_page); 3564 struct f2fs_dir_entry *f2fs_find_entry(struct inode *dir, 3565 const struct qstr *child, struct page **res_page); 3566 struct f2fs_dir_entry *f2fs_parent_dir(struct inode *dir, struct page **p); 3567 ino_t f2fs_inode_by_name(struct inode *dir, const struct qstr *qstr, 3568 struct page **page); 3569 void f2fs_set_link(struct inode *dir, struct f2fs_dir_entry *de, 3570 struct page *page, struct inode *inode); 3571 bool f2fs_has_enough_room(struct inode *dir, struct page *ipage, 3572 const struct f2fs_filename *fname); 3573 void f2fs_update_dentry(nid_t ino, umode_t mode, struct f2fs_dentry_ptr *d, 3574 const struct fscrypt_str *name, f2fs_hash_t name_hash, 3575 unsigned int bit_pos); 3576 int f2fs_add_regular_entry(struct inode *dir, const struct f2fs_filename *fname, 3577 struct inode *inode, nid_t ino, umode_t mode); 3578 int f2fs_add_dentry(struct inode *dir, const struct f2fs_filename *fname, 3579 struct inode *inode, nid_t ino, umode_t mode); 3580 int f2fs_do_add_link(struct inode *dir, const struct qstr *name, 3581 struct inode *inode, nid_t ino, umode_t mode); 3582 void f2fs_delete_entry(struct f2fs_dir_entry *dentry, struct page *page, 3583 struct inode *dir, struct inode *inode); 3584 int f2fs_do_tmpfile(struct inode *inode, struct inode *dir, 3585 struct f2fs_filename *fname); 3586 bool f2fs_empty_dir(struct inode *dir); 3587 3588 static inline int f2fs_add_link(struct dentry *dentry, struct inode *inode) 3589 { 3590 if (fscrypt_is_nokey_name(dentry)) 3591 return -ENOKEY; 3592 return f2fs_do_add_link(d_inode(dentry->d_parent), &dentry->d_name, 3593 inode, inode->i_ino, inode->i_mode); 3594 } 3595 3596 /* 3597 * super.c 3598 */ 3599 int f2fs_inode_dirtied(struct inode *inode, bool sync); 3600 void f2fs_inode_synced(struct inode *inode); 3601 int f2fs_dquot_initialize(struct inode *inode); 3602 int f2fs_enable_quota_files(struct f2fs_sb_info *sbi, bool rdonly); 3603 int f2fs_quota_sync(struct super_block *sb, int type); 3604 loff_t max_file_blocks(struct inode *inode); 3605 void f2fs_quota_off_umount(struct super_block *sb); 3606 void f2fs_save_errors(struct f2fs_sb_info *sbi, unsigned char flag); 3607 void f2fs_handle_critical_error(struct f2fs_sb_info *sbi, unsigned char reason, 3608 bool irq_context); 3609 void f2fs_handle_error(struct f2fs_sb_info *sbi, unsigned char error); 3610 void f2fs_handle_error_async(struct f2fs_sb_info *sbi, unsigned char error); 3611 int f2fs_commit_super(struct f2fs_sb_info *sbi, bool recover); 3612 int f2fs_sync_fs(struct super_block *sb, int sync); 3613 int f2fs_sanity_check_ckpt(struct f2fs_sb_info *sbi); 3614 3615 /* 3616 * hash.c 3617 */ 3618 void f2fs_hash_filename(const struct inode *dir, struct f2fs_filename *fname); 3619 3620 /* 3621 * node.c 3622 */ 3623 struct node_info; 3624 3625 int f2fs_check_nid_range(struct f2fs_sb_info *sbi, nid_t nid); 3626 bool f2fs_available_free_memory(struct f2fs_sb_info *sbi, int type); 3627 bool f2fs_in_warm_node_list(struct f2fs_sb_info *sbi, struct page *page); 3628 void f2fs_init_fsync_node_info(struct f2fs_sb_info *sbi); 3629 void f2fs_del_fsync_node_entry(struct f2fs_sb_info *sbi, struct page *page); 3630 void f2fs_reset_fsync_node_info(struct f2fs_sb_info *sbi); 3631 int f2fs_need_dentry_mark(struct f2fs_sb_info *sbi, nid_t nid); 3632 bool f2fs_is_checkpointed_node(struct f2fs_sb_info *sbi, nid_t nid); 3633 bool f2fs_need_inode_block_update(struct f2fs_sb_info *sbi, nid_t ino); 3634 int f2fs_get_node_info(struct f2fs_sb_info *sbi, nid_t nid, 3635 struct node_info *ni, bool checkpoint_context); 3636 pgoff_t f2fs_get_next_page_offset(struct dnode_of_data *dn, pgoff_t pgofs); 3637 int f2fs_get_dnode_of_data(struct dnode_of_data *dn, pgoff_t index, int mode); 3638 int f2fs_truncate_inode_blocks(struct inode *inode, pgoff_t from); 3639 int f2fs_truncate_xattr_node(struct inode *inode); 3640 int f2fs_wait_on_node_pages_writeback(struct f2fs_sb_info *sbi, 3641 unsigned int seq_id); 3642 bool f2fs_nat_bitmap_enabled(struct f2fs_sb_info *sbi); 3643 int f2fs_remove_inode_page(struct inode *inode); 3644 struct page *f2fs_new_inode_page(struct inode *inode); 3645 struct page *f2fs_new_node_page(struct dnode_of_data *dn, unsigned int ofs); 3646 void f2fs_ra_node_page(struct f2fs_sb_info *sbi, nid_t nid); 3647 struct page *f2fs_get_node_page(struct f2fs_sb_info *sbi, pgoff_t nid); 3648 struct page *f2fs_get_node_page_ra(struct page *parent, int start); 3649 int f2fs_move_node_page(struct page *node_page, int gc_type); 3650 void f2fs_flush_inline_data(struct f2fs_sb_info *sbi); 3651 int f2fs_fsync_node_pages(struct f2fs_sb_info *sbi, struct inode *inode, 3652 struct writeback_control *wbc, bool atomic, 3653 unsigned int *seq_id); 3654 int f2fs_sync_node_pages(struct f2fs_sb_info *sbi, 3655 struct writeback_control *wbc, 3656 bool do_balance, enum iostat_type io_type); 3657 int f2fs_build_free_nids(struct f2fs_sb_info *sbi, bool sync, bool mount); 3658 bool f2fs_alloc_nid(struct f2fs_sb_info *sbi, nid_t *nid); 3659 void f2fs_alloc_nid_done(struct f2fs_sb_info *sbi, nid_t nid); 3660 void f2fs_alloc_nid_failed(struct f2fs_sb_info *sbi, nid_t nid); 3661 int f2fs_try_to_free_nids(struct f2fs_sb_info *sbi, int nr_shrink); 3662 int f2fs_recover_inline_xattr(struct inode *inode, struct page *page); 3663 int f2fs_recover_xattr_data(struct inode *inode, struct page *page); 3664 int f2fs_recover_inode_page(struct f2fs_sb_info *sbi, struct page *page); 3665 int f2fs_restore_node_summary(struct f2fs_sb_info *sbi, 3666 unsigned int segno, struct f2fs_summary_block *sum); 3667 void f2fs_enable_nat_bits(struct f2fs_sb_info *sbi); 3668 int f2fs_flush_nat_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3669 int f2fs_build_node_manager(struct f2fs_sb_info *sbi); 3670 void f2fs_destroy_node_manager(struct f2fs_sb_info *sbi); 3671 int __init f2fs_create_node_manager_caches(void); 3672 void f2fs_destroy_node_manager_caches(void); 3673 3674 /* 3675 * segment.c 3676 */ 3677 bool f2fs_need_SSR(struct f2fs_sb_info *sbi); 3678 int f2fs_commit_atomic_write(struct inode *inode); 3679 void f2fs_abort_atomic_write(struct inode *inode, bool clean); 3680 void f2fs_balance_fs(struct f2fs_sb_info *sbi, bool need); 3681 void f2fs_balance_fs_bg(struct f2fs_sb_info *sbi, bool from_bg); 3682 int f2fs_issue_flush(struct f2fs_sb_info *sbi, nid_t ino); 3683 int f2fs_create_flush_cmd_control(struct f2fs_sb_info *sbi); 3684 int f2fs_flush_device_cache(struct f2fs_sb_info *sbi); 3685 void f2fs_destroy_flush_cmd_control(struct f2fs_sb_info *sbi, bool free); 3686 void f2fs_invalidate_blocks(struct f2fs_sb_info *sbi, block_t addr); 3687 bool f2fs_is_checkpointed_data(struct f2fs_sb_info *sbi, block_t blkaddr); 3688 int f2fs_start_discard_thread(struct f2fs_sb_info *sbi); 3689 void f2fs_drop_discard_cmd(struct f2fs_sb_info *sbi); 3690 void f2fs_stop_discard_thread(struct f2fs_sb_info *sbi); 3691 bool f2fs_issue_discard_timeout(struct f2fs_sb_info *sbi); 3692 void f2fs_clear_prefree_segments(struct f2fs_sb_info *sbi, 3693 struct cp_control *cpc); 3694 void f2fs_dirty_to_prefree(struct f2fs_sb_info *sbi); 3695 block_t f2fs_get_unusable_blocks(struct f2fs_sb_info *sbi); 3696 int f2fs_disable_cp_again(struct f2fs_sb_info *sbi, block_t unusable); 3697 void f2fs_release_discard_addrs(struct f2fs_sb_info *sbi); 3698 int f2fs_npages_for_summary_flush(struct f2fs_sb_info *sbi, bool for_ra); 3699 bool f2fs_segment_has_free_slot(struct f2fs_sb_info *sbi, int segno); 3700 void f2fs_init_inmem_curseg(struct f2fs_sb_info *sbi); 3701 void f2fs_save_inmem_curseg(struct f2fs_sb_info *sbi); 3702 void f2fs_restore_inmem_curseg(struct f2fs_sb_info *sbi); 3703 void f2fs_get_new_segment(struct f2fs_sb_info *sbi, 3704 unsigned int *newseg, bool new_sec, int dir); 3705 void f2fs_allocate_segment_for_resize(struct f2fs_sb_info *sbi, int type, 3706 unsigned int start, unsigned int end); 3707 int f2fs_allocate_new_section(struct f2fs_sb_info *sbi, int type, bool force); 3708 int f2fs_allocate_pinning_section(struct f2fs_sb_info *sbi); 3709 void f2fs_allocate_new_segments(struct f2fs_sb_info *sbi); 3710 int f2fs_trim_fs(struct f2fs_sb_info *sbi, struct fstrim_range *range); 3711 bool f2fs_exist_trim_candidates(struct f2fs_sb_info *sbi, 3712 struct cp_control *cpc); 3713 struct page *f2fs_get_sum_page(struct f2fs_sb_info *sbi, unsigned int segno); 3714 void f2fs_update_meta_page(struct f2fs_sb_info *sbi, void *src, 3715 block_t blk_addr); 3716 void f2fs_do_write_meta_page(struct f2fs_sb_info *sbi, struct page *page, 3717 enum iostat_type io_type); 3718 void f2fs_do_write_node_page(unsigned int nid, struct f2fs_io_info *fio); 3719 void f2fs_outplace_write_data(struct dnode_of_data *dn, 3720 struct f2fs_io_info *fio); 3721 int f2fs_inplace_write_data(struct f2fs_io_info *fio); 3722 void f2fs_do_replace_block(struct f2fs_sb_info *sbi, struct f2fs_summary *sum, 3723 block_t old_blkaddr, block_t new_blkaddr, 3724 bool recover_curseg, bool recover_newaddr, 3725 bool from_gc); 3726 void f2fs_replace_block(struct f2fs_sb_info *sbi, struct dnode_of_data *dn, 3727 block_t old_addr, block_t new_addr, 3728 unsigned char version, bool recover_curseg, 3729 bool recover_newaddr); 3730 void f2fs_allocate_data_block(struct f2fs_sb_info *sbi, struct page *page, 3731 block_t old_blkaddr, block_t *new_blkaddr, 3732 struct f2fs_summary *sum, int type, 3733 struct f2fs_io_info *fio); 3734 void f2fs_update_device_state(struct f2fs_sb_info *sbi, nid_t ino, 3735 block_t blkaddr, unsigned int blkcnt); 3736 void f2fs_wait_on_page_writeback(struct page *page, 3737 enum page_type type, bool ordered, bool locked); 3738 void f2fs_wait_on_block_writeback(struct inode *inode, block_t blkaddr); 3739 void f2fs_wait_on_block_writeback_range(struct inode *inode, block_t blkaddr, 3740 block_t len); 3741 void f2fs_write_data_summaries(struct f2fs_sb_info *sbi, block_t start_blk); 3742 void f2fs_write_node_summaries(struct f2fs_sb_info *sbi, block_t start_blk); 3743 int f2fs_lookup_journal_in_cursum(struct f2fs_journal *journal, int type, 3744 unsigned int val, int alloc); 3745 void f2fs_flush_sit_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3746 int f2fs_fix_curseg_write_pointer(struct f2fs_sb_info *sbi); 3747 int f2fs_check_write_pointer(struct f2fs_sb_info *sbi); 3748 int f2fs_build_segment_manager(struct f2fs_sb_info *sbi); 3749 void f2fs_destroy_segment_manager(struct f2fs_sb_info *sbi); 3750 int __init f2fs_create_segment_manager_caches(void); 3751 void f2fs_destroy_segment_manager_caches(void); 3752 int f2fs_rw_hint_to_seg_type(enum rw_hint hint); 3753 unsigned int f2fs_usable_segs_in_sec(struct f2fs_sb_info *sbi, 3754 unsigned int segno); 3755 unsigned int f2fs_usable_blks_in_seg(struct f2fs_sb_info *sbi, 3756 unsigned int segno); 3757 3758 #define DEF_FRAGMENT_SIZE 4 3759 #define MIN_FRAGMENT_SIZE 1 3760 #define MAX_FRAGMENT_SIZE 512 3761 3762 static inline bool f2fs_need_rand_seg(struct f2fs_sb_info *sbi) 3763 { 3764 return F2FS_OPTION(sbi).fs_mode == FS_MODE_FRAGMENT_SEG || 3765 F2FS_OPTION(sbi).fs_mode == FS_MODE_FRAGMENT_BLK; 3766 } 3767 3768 /* 3769 * checkpoint.c 3770 */ 3771 void f2fs_stop_checkpoint(struct f2fs_sb_info *sbi, bool end_io, 3772 unsigned char reason); 3773 void f2fs_flush_ckpt_thread(struct f2fs_sb_info *sbi); 3774 struct page *f2fs_grab_meta_page(struct f2fs_sb_info *sbi, pgoff_t index); 3775 struct page *f2fs_get_meta_page(struct f2fs_sb_info *sbi, pgoff_t index); 3776 struct page *f2fs_get_meta_page_retry(struct f2fs_sb_info *sbi, pgoff_t index); 3777 struct page *f2fs_get_tmp_page(struct f2fs_sb_info *sbi, pgoff_t index); 3778 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi, 3779 block_t blkaddr, int type); 3780 int f2fs_ra_meta_pages(struct f2fs_sb_info *sbi, block_t start, int nrpages, 3781 int type, bool sync); 3782 void f2fs_ra_meta_pages_cond(struct f2fs_sb_info *sbi, pgoff_t index, 3783 unsigned int ra_blocks); 3784 long f2fs_sync_meta_pages(struct f2fs_sb_info *sbi, enum page_type type, 3785 long nr_to_write, enum iostat_type io_type); 3786 void f2fs_add_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type); 3787 void f2fs_remove_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type); 3788 void f2fs_release_ino_entry(struct f2fs_sb_info *sbi, bool all); 3789 bool f2fs_exist_written_data(struct f2fs_sb_info *sbi, nid_t ino, int mode); 3790 void f2fs_set_dirty_device(struct f2fs_sb_info *sbi, nid_t ino, 3791 unsigned int devidx, int type); 3792 bool f2fs_is_dirty_device(struct f2fs_sb_info *sbi, nid_t ino, 3793 unsigned int devidx, int type); 3794 int f2fs_acquire_orphan_inode(struct f2fs_sb_info *sbi); 3795 void f2fs_release_orphan_inode(struct f2fs_sb_info *sbi); 3796 void f2fs_add_orphan_inode(struct inode *inode); 3797 void f2fs_remove_orphan_inode(struct f2fs_sb_info *sbi, nid_t ino); 3798 int f2fs_recover_orphan_inodes(struct f2fs_sb_info *sbi); 3799 int f2fs_get_valid_checkpoint(struct f2fs_sb_info *sbi); 3800 void f2fs_update_dirty_folio(struct inode *inode, struct folio *folio); 3801 void f2fs_remove_dirty_inode(struct inode *inode); 3802 int f2fs_sync_dirty_inodes(struct f2fs_sb_info *sbi, enum inode_type type, 3803 bool from_cp); 3804 void f2fs_wait_on_all_pages(struct f2fs_sb_info *sbi, int type); 3805 u64 f2fs_get_sectors_written(struct f2fs_sb_info *sbi); 3806 int f2fs_write_checkpoint(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3807 void f2fs_init_ino_entry_info(struct f2fs_sb_info *sbi); 3808 int __init f2fs_create_checkpoint_caches(void); 3809 void f2fs_destroy_checkpoint_caches(void); 3810 int f2fs_issue_checkpoint(struct f2fs_sb_info *sbi); 3811 int f2fs_start_ckpt_thread(struct f2fs_sb_info *sbi); 3812 void f2fs_stop_ckpt_thread(struct f2fs_sb_info *sbi); 3813 void f2fs_init_ckpt_req_control(struct f2fs_sb_info *sbi); 3814 3815 /* 3816 * data.c 3817 */ 3818 int __init f2fs_init_bioset(void); 3819 void f2fs_destroy_bioset(void); 3820 bool f2fs_is_cp_guaranteed(struct page *page); 3821 int f2fs_init_bio_entry_cache(void); 3822 void f2fs_destroy_bio_entry_cache(void); 3823 void f2fs_submit_read_bio(struct f2fs_sb_info *sbi, struct bio *bio, 3824 enum page_type type); 3825 int f2fs_init_write_merge_io(struct f2fs_sb_info *sbi); 3826 void f2fs_submit_merged_write(struct f2fs_sb_info *sbi, enum page_type type); 3827 void f2fs_submit_merged_write_cond(struct f2fs_sb_info *sbi, 3828 struct inode *inode, struct page *page, 3829 nid_t ino, enum page_type type); 3830 void f2fs_submit_merged_ipu_write(struct f2fs_sb_info *sbi, 3831 struct bio **bio, struct page *page); 3832 void f2fs_flush_merged_writes(struct f2fs_sb_info *sbi); 3833 int f2fs_submit_page_bio(struct f2fs_io_info *fio); 3834 int f2fs_merge_page_bio(struct f2fs_io_info *fio); 3835 void f2fs_submit_page_write(struct f2fs_io_info *fio); 3836 struct block_device *f2fs_target_device(struct f2fs_sb_info *sbi, 3837 block_t blk_addr, sector_t *sector); 3838 int f2fs_target_device_index(struct f2fs_sb_info *sbi, block_t blkaddr); 3839 void f2fs_set_data_blkaddr(struct dnode_of_data *dn, block_t blkaddr); 3840 void f2fs_update_data_blkaddr(struct dnode_of_data *dn, block_t blkaddr); 3841 int f2fs_reserve_new_blocks(struct dnode_of_data *dn, blkcnt_t count); 3842 int f2fs_reserve_new_block(struct dnode_of_data *dn); 3843 int f2fs_get_block_locked(struct dnode_of_data *dn, pgoff_t index); 3844 int f2fs_reserve_block(struct dnode_of_data *dn, pgoff_t index); 3845 struct page *f2fs_get_read_data_page(struct inode *inode, pgoff_t index, 3846 blk_opf_t op_flags, bool for_write, pgoff_t *next_pgofs); 3847 struct page *f2fs_find_data_page(struct inode *inode, pgoff_t index, 3848 pgoff_t *next_pgofs); 3849 struct page *f2fs_get_lock_data_page(struct inode *inode, pgoff_t index, 3850 bool for_write); 3851 struct page *f2fs_get_new_data_page(struct inode *inode, 3852 struct page *ipage, pgoff_t index, bool new_i_size); 3853 int f2fs_do_write_data_page(struct f2fs_io_info *fio); 3854 int f2fs_map_blocks(struct inode *inode, struct f2fs_map_blocks *map, int flag); 3855 int f2fs_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, 3856 u64 start, u64 len); 3857 int f2fs_encrypt_one_page(struct f2fs_io_info *fio); 3858 bool f2fs_should_update_inplace(struct inode *inode, struct f2fs_io_info *fio); 3859 bool f2fs_should_update_outplace(struct inode *inode, struct f2fs_io_info *fio); 3860 int f2fs_write_single_data_page(struct page *page, int *submitted, 3861 struct bio **bio, sector_t *last_block, 3862 struct writeback_control *wbc, 3863 enum iostat_type io_type, 3864 int compr_blocks, bool allow_balance); 3865 void f2fs_write_failed(struct inode *inode, loff_t to); 3866 void f2fs_invalidate_folio(struct folio *folio, size_t offset, size_t length); 3867 bool f2fs_release_folio(struct folio *folio, gfp_t wait); 3868 bool f2fs_overwrite_io(struct inode *inode, loff_t pos, size_t len); 3869 void f2fs_clear_page_cache_dirty_tag(struct page *page); 3870 int f2fs_init_post_read_processing(void); 3871 void f2fs_destroy_post_read_processing(void); 3872 int f2fs_init_post_read_wq(struct f2fs_sb_info *sbi); 3873 void f2fs_destroy_post_read_wq(struct f2fs_sb_info *sbi); 3874 extern const struct iomap_ops f2fs_iomap_ops; 3875 3876 /* 3877 * gc.c 3878 */ 3879 int f2fs_start_gc_thread(struct f2fs_sb_info *sbi); 3880 void f2fs_stop_gc_thread(struct f2fs_sb_info *sbi); 3881 block_t f2fs_start_bidx_of_node(unsigned int node_ofs, struct inode *inode); 3882 int f2fs_gc(struct f2fs_sb_info *sbi, struct f2fs_gc_control *gc_control); 3883 void f2fs_build_gc_manager(struct f2fs_sb_info *sbi); 3884 int f2fs_gc_range(struct f2fs_sb_info *sbi, 3885 unsigned int start_seg, unsigned int end_seg, 3886 bool dry_run, unsigned int dry_run_sections); 3887 int f2fs_resize_fs(struct file *filp, __u64 block_count); 3888 int __init f2fs_create_garbage_collection_cache(void); 3889 void f2fs_destroy_garbage_collection_cache(void); 3890 /* victim selection function for cleaning and SSR */ 3891 int f2fs_get_victim(struct f2fs_sb_info *sbi, unsigned int *result, 3892 int gc_type, int type, char alloc_mode, 3893 unsigned long long age); 3894 3895 /* 3896 * recovery.c 3897 */ 3898 int f2fs_recover_fsync_data(struct f2fs_sb_info *sbi, bool check_only); 3899 bool f2fs_space_for_roll_forward(struct f2fs_sb_info *sbi); 3900 int __init f2fs_create_recovery_cache(void); 3901 void f2fs_destroy_recovery_cache(void); 3902 3903 /* 3904 * debug.c 3905 */ 3906 #ifdef CONFIG_F2FS_STAT_FS 3907 struct f2fs_stat_info { 3908 struct list_head stat_list; 3909 struct f2fs_sb_info *sbi; 3910 int all_area_segs, sit_area_segs, nat_area_segs, ssa_area_segs; 3911 int main_area_segs, main_area_sections, main_area_zones; 3912 unsigned long long hit_cached[NR_EXTENT_CACHES]; 3913 unsigned long long hit_rbtree[NR_EXTENT_CACHES]; 3914 unsigned long long total_ext[NR_EXTENT_CACHES]; 3915 unsigned long long hit_total[NR_EXTENT_CACHES]; 3916 int ext_tree[NR_EXTENT_CACHES]; 3917 int zombie_tree[NR_EXTENT_CACHES]; 3918 int ext_node[NR_EXTENT_CACHES]; 3919 /* to count memory footprint */ 3920 unsigned long long ext_mem[NR_EXTENT_CACHES]; 3921 /* for read extent cache */ 3922 unsigned long long hit_largest; 3923 /* for block age extent cache */ 3924 unsigned long long allocated_data_blocks; 3925 int ndirty_node, ndirty_dent, ndirty_meta, ndirty_imeta; 3926 int ndirty_data, ndirty_qdata; 3927 unsigned int ndirty_dirs, ndirty_files, nquota_files, ndirty_all; 3928 int nats, dirty_nats, sits, dirty_sits; 3929 int free_nids, avail_nids, alloc_nids; 3930 int total_count, utilization; 3931 int nr_wb_cp_data, nr_wb_data; 3932 int nr_rd_data, nr_rd_node, nr_rd_meta; 3933 int nr_dio_read, nr_dio_write; 3934 unsigned int io_skip_bggc, other_skip_bggc; 3935 int nr_flushing, nr_flushed, flush_list_empty; 3936 int nr_discarding, nr_discarded; 3937 int nr_discard_cmd; 3938 unsigned int undiscard_blks; 3939 int nr_issued_ckpt, nr_total_ckpt, nr_queued_ckpt; 3940 unsigned int cur_ckpt_time, peak_ckpt_time; 3941 int inline_xattr, inline_inode, inline_dir, append, update, orphans; 3942 int compr_inode, swapfile_inode; 3943 unsigned long long compr_blocks; 3944 int aw_cnt, max_aw_cnt; 3945 unsigned int valid_count, valid_node_count, valid_inode_count, discard_blks; 3946 unsigned int bimodal, avg_vblocks; 3947 int util_free, util_valid, util_invalid; 3948 int rsvd_segs, overp_segs; 3949 int dirty_count, node_pages, meta_pages, compress_pages; 3950 int compress_page_hit; 3951 int prefree_count, free_segs, free_secs; 3952 int cp_call_count[MAX_CALL_TYPE], cp_count; 3953 int gc_call_count[MAX_CALL_TYPE]; 3954 int gc_segs[2][2]; 3955 int gc_secs[2][2]; 3956 int tot_blks, data_blks, node_blks; 3957 int bg_data_blks, bg_node_blks; 3958 int curseg[NR_CURSEG_TYPE]; 3959 int cursec[NR_CURSEG_TYPE]; 3960 int curzone[NR_CURSEG_TYPE]; 3961 unsigned int dirty_seg[NR_CURSEG_TYPE]; 3962 unsigned int full_seg[NR_CURSEG_TYPE]; 3963 unsigned int valid_blks[NR_CURSEG_TYPE]; 3964 3965 unsigned int meta_count[META_MAX]; 3966 unsigned int segment_count[2]; 3967 unsigned int block_count[2]; 3968 unsigned int inplace_count; 3969 unsigned long long base_mem, cache_mem, page_mem; 3970 }; 3971 3972 static inline struct f2fs_stat_info *F2FS_STAT(struct f2fs_sb_info *sbi) 3973 { 3974 return (struct f2fs_stat_info *)sbi->stat_info; 3975 } 3976 3977 #define stat_inc_cp_call_count(sbi, foreground) \ 3978 atomic_inc(&sbi->cp_call_count[(foreground)]) 3979 #define stat_inc_cp_count(si) (F2FS_STAT(sbi)->cp_count++) 3980 #define stat_io_skip_bggc_count(sbi) ((sbi)->io_skip_bggc++) 3981 #define stat_other_skip_bggc_count(sbi) ((sbi)->other_skip_bggc++) 3982 #define stat_inc_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]++) 3983 #define stat_dec_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]--) 3984 #define stat_inc_total_hit(sbi, type) (atomic64_inc(&(sbi)->total_hit_ext[type])) 3985 #define stat_inc_rbtree_node_hit(sbi, type) (atomic64_inc(&(sbi)->read_hit_rbtree[type])) 3986 #define stat_inc_largest_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_largest)) 3987 #define stat_inc_cached_node_hit(sbi, type) (atomic64_inc(&(sbi)->read_hit_cached[type])) 3988 #define stat_inc_inline_xattr(inode) \ 3989 do { \ 3990 if (f2fs_has_inline_xattr(inode)) \ 3991 (atomic_inc(&F2FS_I_SB(inode)->inline_xattr)); \ 3992 } while (0) 3993 #define stat_dec_inline_xattr(inode) \ 3994 do { \ 3995 if (f2fs_has_inline_xattr(inode)) \ 3996 (atomic_dec(&F2FS_I_SB(inode)->inline_xattr)); \ 3997 } while (0) 3998 #define stat_inc_inline_inode(inode) \ 3999 do { \ 4000 if (f2fs_has_inline_data(inode)) \ 4001 (atomic_inc(&F2FS_I_SB(inode)->inline_inode)); \ 4002 } while (0) 4003 #define stat_dec_inline_inode(inode) \ 4004 do { \ 4005 if (f2fs_has_inline_data(inode)) \ 4006 (atomic_dec(&F2FS_I_SB(inode)->inline_inode)); \ 4007 } while (0) 4008 #define stat_inc_inline_dir(inode) \ 4009 do { \ 4010 if (f2fs_has_inline_dentry(inode)) \ 4011 (atomic_inc(&F2FS_I_SB(inode)->inline_dir)); \ 4012 } while (0) 4013 #define stat_dec_inline_dir(inode) \ 4014 do { \ 4015 if (f2fs_has_inline_dentry(inode)) \ 4016 (atomic_dec(&F2FS_I_SB(inode)->inline_dir)); \ 4017 } while (0) 4018 #define stat_inc_compr_inode(inode) \ 4019 do { \ 4020 if (f2fs_compressed_file(inode)) \ 4021 (atomic_inc(&F2FS_I_SB(inode)->compr_inode)); \ 4022 } while (0) 4023 #define stat_dec_compr_inode(inode) \ 4024 do { \ 4025 if (f2fs_compressed_file(inode)) \ 4026 (atomic_dec(&F2FS_I_SB(inode)->compr_inode)); \ 4027 } while (0) 4028 #define stat_add_compr_blocks(inode, blocks) \ 4029 (atomic64_add(blocks, &F2FS_I_SB(inode)->compr_blocks)) 4030 #define stat_sub_compr_blocks(inode, blocks) \ 4031 (atomic64_sub(blocks, &F2FS_I_SB(inode)->compr_blocks)) 4032 #define stat_inc_swapfile_inode(inode) \ 4033 (atomic_inc(&F2FS_I_SB(inode)->swapfile_inode)) 4034 #define stat_dec_swapfile_inode(inode) \ 4035 (atomic_dec(&F2FS_I_SB(inode)->swapfile_inode)) 4036 #define stat_inc_atomic_inode(inode) \ 4037 (atomic_inc(&F2FS_I_SB(inode)->atomic_files)) 4038 #define stat_dec_atomic_inode(inode) \ 4039 (atomic_dec(&F2FS_I_SB(inode)->atomic_files)) 4040 #define stat_inc_meta_count(sbi, blkaddr) \ 4041 do { \ 4042 if (blkaddr < SIT_I(sbi)->sit_base_addr) \ 4043 atomic_inc(&(sbi)->meta_count[META_CP]); \ 4044 else if (blkaddr < NM_I(sbi)->nat_blkaddr) \ 4045 atomic_inc(&(sbi)->meta_count[META_SIT]); \ 4046 else if (blkaddr < SM_I(sbi)->ssa_blkaddr) \ 4047 atomic_inc(&(sbi)->meta_count[META_NAT]); \ 4048 else if (blkaddr < SM_I(sbi)->main_blkaddr) \ 4049 atomic_inc(&(sbi)->meta_count[META_SSA]); \ 4050 } while (0) 4051 #define stat_inc_seg_type(sbi, curseg) \ 4052 ((sbi)->segment_count[(curseg)->alloc_type]++) 4053 #define stat_inc_block_count(sbi, curseg) \ 4054 ((sbi)->block_count[(curseg)->alloc_type]++) 4055 #define stat_inc_inplace_blocks(sbi) \ 4056 (atomic_inc(&(sbi)->inplace_count)) 4057 #define stat_update_max_atomic_write(inode) \ 4058 do { \ 4059 int cur = atomic_read(&F2FS_I_SB(inode)->atomic_files); \ 4060 int max = atomic_read(&F2FS_I_SB(inode)->max_aw_cnt); \ 4061 if (cur > max) \ 4062 atomic_set(&F2FS_I_SB(inode)->max_aw_cnt, cur); \ 4063 } while (0) 4064 #define stat_inc_gc_call_count(sbi, foreground) \ 4065 (F2FS_STAT(sbi)->gc_call_count[(foreground)]++) 4066 #define stat_inc_gc_sec_count(sbi, type, gc_type) \ 4067 (F2FS_STAT(sbi)->gc_secs[(type)][(gc_type)]++) 4068 #define stat_inc_gc_seg_count(sbi, type, gc_type) \ 4069 (F2FS_STAT(sbi)->gc_segs[(type)][(gc_type)]++) 4070 4071 #define stat_inc_tot_blk_count(si, blks) \ 4072 ((si)->tot_blks += (blks)) 4073 4074 #define stat_inc_data_blk_count(sbi, blks, gc_type) \ 4075 do { \ 4076 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 4077 stat_inc_tot_blk_count(si, blks); \ 4078 si->data_blks += (blks); \ 4079 si->bg_data_blks += ((gc_type) == BG_GC) ? (blks) : 0; \ 4080 } while (0) 4081 4082 #define stat_inc_node_blk_count(sbi, blks, gc_type) \ 4083 do { \ 4084 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 4085 stat_inc_tot_blk_count(si, blks); \ 4086 si->node_blks += (blks); \ 4087 si->bg_node_blks += ((gc_type) == BG_GC) ? (blks) : 0; \ 4088 } while (0) 4089 4090 int f2fs_build_stats(struct f2fs_sb_info *sbi); 4091 void f2fs_destroy_stats(struct f2fs_sb_info *sbi); 4092 void __init f2fs_create_root_stats(void); 4093 void f2fs_destroy_root_stats(void); 4094 void f2fs_update_sit_info(struct f2fs_sb_info *sbi); 4095 #else 4096 #define stat_inc_cp_call_count(sbi, foreground) do { } while (0) 4097 #define stat_inc_cp_count(sbi) do { } while (0) 4098 #define stat_io_skip_bggc_count(sbi) do { } while (0) 4099 #define stat_other_skip_bggc_count(sbi) do { } while (0) 4100 #define stat_inc_dirty_inode(sbi, type) do { } while (0) 4101 #define stat_dec_dirty_inode(sbi, type) do { } while (0) 4102 #define stat_inc_total_hit(sbi, type) do { } while (0) 4103 #define stat_inc_rbtree_node_hit(sbi, type) do { } while (0) 4104 #define stat_inc_largest_node_hit(sbi) do { } while (0) 4105 #define stat_inc_cached_node_hit(sbi, type) do { } while (0) 4106 #define stat_inc_inline_xattr(inode) do { } while (0) 4107 #define stat_dec_inline_xattr(inode) do { } while (0) 4108 #define stat_inc_inline_inode(inode) do { } while (0) 4109 #define stat_dec_inline_inode(inode) do { } while (0) 4110 #define stat_inc_inline_dir(inode) do { } while (0) 4111 #define stat_dec_inline_dir(inode) do { } while (0) 4112 #define stat_inc_compr_inode(inode) do { } while (0) 4113 #define stat_dec_compr_inode(inode) do { } while (0) 4114 #define stat_add_compr_blocks(inode, blocks) do { } while (0) 4115 #define stat_sub_compr_blocks(inode, blocks) do { } while (0) 4116 #define stat_inc_swapfile_inode(inode) do { } while (0) 4117 #define stat_dec_swapfile_inode(inode) do { } while (0) 4118 #define stat_inc_atomic_inode(inode) do { } while (0) 4119 #define stat_dec_atomic_inode(inode) do { } while (0) 4120 #define stat_update_max_atomic_write(inode) do { } while (0) 4121 #define stat_inc_meta_count(sbi, blkaddr) do { } while (0) 4122 #define stat_inc_seg_type(sbi, curseg) do { } while (0) 4123 #define stat_inc_block_count(sbi, curseg) do { } while (0) 4124 #define stat_inc_inplace_blocks(sbi) do { } while (0) 4125 #define stat_inc_gc_call_count(sbi, foreground) do { } while (0) 4126 #define stat_inc_gc_sec_count(sbi, type, gc_type) do { } while (0) 4127 #define stat_inc_gc_seg_count(sbi, type, gc_type) do { } while (0) 4128 #define stat_inc_tot_blk_count(si, blks) do { } while (0) 4129 #define stat_inc_data_blk_count(sbi, blks, gc_type) do { } while (0) 4130 #define stat_inc_node_blk_count(sbi, blks, gc_type) do { } while (0) 4131 4132 static inline int f2fs_build_stats(struct f2fs_sb_info *sbi) { return 0; } 4133 static inline void f2fs_destroy_stats(struct f2fs_sb_info *sbi) { } 4134 static inline void __init f2fs_create_root_stats(void) { } 4135 static inline void f2fs_destroy_root_stats(void) { } 4136 static inline void f2fs_update_sit_info(struct f2fs_sb_info *sbi) {} 4137 #endif 4138 4139 extern const struct file_operations f2fs_dir_operations; 4140 extern const struct file_operations f2fs_file_operations; 4141 extern const struct inode_operations f2fs_file_inode_operations; 4142 extern const struct address_space_operations f2fs_dblock_aops; 4143 extern const struct address_space_operations f2fs_node_aops; 4144 extern const struct address_space_operations f2fs_meta_aops; 4145 extern const struct inode_operations f2fs_dir_inode_operations; 4146 extern const struct inode_operations f2fs_symlink_inode_operations; 4147 extern const struct inode_operations f2fs_encrypted_symlink_inode_operations; 4148 extern const struct inode_operations f2fs_special_inode_operations; 4149 extern struct kmem_cache *f2fs_inode_entry_slab; 4150 4151 /* 4152 * inline.c 4153 */ 4154 bool f2fs_may_inline_data(struct inode *inode); 4155 bool f2fs_sanity_check_inline_data(struct inode *inode); 4156 bool f2fs_may_inline_dentry(struct inode *inode); 4157 void f2fs_do_read_inline_data(struct page *page, struct page *ipage); 4158 void f2fs_truncate_inline_inode(struct inode *inode, 4159 struct page *ipage, u64 from); 4160 int f2fs_read_inline_data(struct inode *inode, struct page *page); 4161 int f2fs_convert_inline_page(struct dnode_of_data *dn, struct page *page); 4162 int f2fs_convert_inline_inode(struct inode *inode); 4163 int f2fs_try_convert_inline_dir(struct inode *dir, struct dentry *dentry); 4164 int f2fs_write_inline_data(struct inode *inode, struct page *page); 4165 int f2fs_recover_inline_data(struct inode *inode, struct page *npage); 4166 struct f2fs_dir_entry *f2fs_find_in_inline_dir(struct inode *dir, 4167 const struct f2fs_filename *fname, 4168 struct page **res_page); 4169 int f2fs_make_empty_inline_dir(struct inode *inode, struct inode *parent, 4170 struct page *ipage); 4171 int f2fs_add_inline_entry(struct inode *dir, const struct f2fs_filename *fname, 4172 struct inode *inode, nid_t ino, umode_t mode); 4173 void f2fs_delete_inline_entry(struct f2fs_dir_entry *dentry, 4174 struct page *page, struct inode *dir, 4175 struct inode *inode); 4176 bool f2fs_empty_inline_dir(struct inode *dir); 4177 int f2fs_read_inline_dir(struct file *file, struct dir_context *ctx, 4178 struct fscrypt_str *fstr); 4179 int f2fs_inline_data_fiemap(struct inode *inode, 4180 struct fiemap_extent_info *fieinfo, 4181 __u64 start, __u64 len); 4182 4183 /* 4184 * shrinker.c 4185 */ 4186 unsigned long f2fs_shrink_count(struct shrinker *shrink, 4187 struct shrink_control *sc); 4188 unsigned long f2fs_shrink_scan(struct shrinker *shrink, 4189 struct shrink_control *sc); 4190 void f2fs_join_shrinker(struct f2fs_sb_info *sbi); 4191 void f2fs_leave_shrinker(struct f2fs_sb_info *sbi); 4192 4193 /* 4194 * extent_cache.c 4195 */ 4196 bool sanity_check_extent_cache(struct inode *inode); 4197 void f2fs_init_extent_tree(struct inode *inode); 4198 void f2fs_drop_extent_tree(struct inode *inode); 4199 void f2fs_destroy_extent_node(struct inode *inode); 4200 void f2fs_destroy_extent_tree(struct inode *inode); 4201 void f2fs_init_extent_cache_info(struct f2fs_sb_info *sbi); 4202 int __init f2fs_create_extent_cache(void); 4203 void f2fs_destroy_extent_cache(void); 4204 4205 /* read extent cache ops */ 4206 void f2fs_init_read_extent_tree(struct inode *inode, struct page *ipage); 4207 bool f2fs_lookup_read_extent_cache(struct inode *inode, pgoff_t pgofs, 4208 struct extent_info *ei); 4209 bool f2fs_lookup_read_extent_cache_block(struct inode *inode, pgoff_t index, 4210 block_t *blkaddr); 4211 void f2fs_update_read_extent_cache(struct dnode_of_data *dn); 4212 void f2fs_update_read_extent_cache_range(struct dnode_of_data *dn, 4213 pgoff_t fofs, block_t blkaddr, unsigned int len); 4214 unsigned int f2fs_shrink_read_extent_tree(struct f2fs_sb_info *sbi, 4215 int nr_shrink); 4216 4217 /* block age extent cache ops */ 4218 void f2fs_init_age_extent_tree(struct inode *inode); 4219 bool f2fs_lookup_age_extent_cache(struct inode *inode, pgoff_t pgofs, 4220 struct extent_info *ei); 4221 void f2fs_update_age_extent_cache(struct dnode_of_data *dn); 4222 void f2fs_update_age_extent_cache_range(struct dnode_of_data *dn, 4223 pgoff_t fofs, unsigned int len); 4224 unsigned int f2fs_shrink_age_extent_tree(struct f2fs_sb_info *sbi, 4225 int nr_shrink); 4226 4227 /* 4228 * sysfs.c 4229 */ 4230 #define MIN_RA_MUL 2 4231 #define MAX_RA_MUL 256 4232 4233 int __init f2fs_init_sysfs(void); 4234 void f2fs_exit_sysfs(void); 4235 int f2fs_register_sysfs(struct f2fs_sb_info *sbi); 4236 void f2fs_unregister_sysfs(struct f2fs_sb_info *sbi); 4237 4238 /* verity.c */ 4239 extern const struct fsverity_operations f2fs_verityops; 4240 4241 /* 4242 * crypto support 4243 */ 4244 static inline bool f2fs_encrypted_file(struct inode *inode) 4245 { 4246 return IS_ENCRYPTED(inode) && S_ISREG(inode->i_mode); 4247 } 4248 4249 static inline void f2fs_set_encrypted_inode(struct inode *inode) 4250 { 4251 #ifdef CONFIG_FS_ENCRYPTION 4252 file_set_encrypt(inode); 4253 f2fs_set_inode_flags(inode); 4254 #endif 4255 } 4256 4257 /* 4258 * Returns true if the reads of the inode's data need to undergo some 4259 * postprocessing step, like decryption or authenticity verification. 4260 */ 4261 static inline bool f2fs_post_read_required(struct inode *inode) 4262 { 4263 return f2fs_encrypted_file(inode) || fsverity_active(inode) || 4264 f2fs_compressed_file(inode); 4265 } 4266 4267 /* 4268 * compress.c 4269 */ 4270 #ifdef CONFIG_F2FS_FS_COMPRESSION 4271 bool f2fs_is_compressed_page(struct page *page); 4272 struct page *f2fs_compress_control_page(struct page *page); 4273 int f2fs_prepare_compress_overwrite(struct inode *inode, 4274 struct page **pagep, pgoff_t index, void **fsdata); 4275 bool f2fs_compress_write_end(struct inode *inode, void *fsdata, 4276 pgoff_t index, unsigned copied); 4277 int f2fs_truncate_partial_cluster(struct inode *inode, u64 from, bool lock); 4278 void f2fs_compress_write_end_io(struct bio *bio, struct page *page); 4279 bool f2fs_is_compress_backend_ready(struct inode *inode); 4280 bool f2fs_is_compress_level_valid(int alg, int lvl); 4281 int __init f2fs_init_compress_mempool(void); 4282 void f2fs_destroy_compress_mempool(void); 4283 void f2fs_decompress_cluster(struct decompress_io_ctx *dic, bool in_task); 4284 void f2fs_end_read_compressed_page(struct page *page, bool failed, 4285 block_t blkaddr, bool in_task); 4286 bool f2fs_cluster_is_empty(struct compress_ctx *cc); 4287 bool f2fs_cluster_can_merge_page(struct compress_ctx *cc, pgoff_t index); 4288 bool f2fs_all_cluster_page_ready(struct compress_ctx *cc, struct page **pages, 4289 int index, int nr_pages, bool uptodate); 4290 bool f2fs_sanity_check_cluster(struct dnode_of_data *dn); 4291 void f2fs_compress_ctx_add_page(struct compress_ctx *cc, struct page *page); 4292 int f2fs_write_multi_pages(struct compress_ctx *cc, 4293 int *submitted, 4294 struct writeback_control *wbc, 4295 enum iostat_type io_type); 4296 int f2fs_is_compressed_cluster(struct inode *inode, pgoff_t index); 4297 void f2fs_update_read_extent_tree_range_compressed(struct inode *inode, 4298 pgoff_t fofs, block_t blkaddr, 4299 unsigned int llen, unsigned int c_len); 4300 int f2fs_read_multi_pages(struct compress_ctx *cc, struct bio **bio_ret, 4301 unsigned nr_pages, sector_t *last_block_in_bio, 4302 bool is_readahead, bool for_write); 4303 struct decompress_io_ctx *f2fs_alloc_dic(struct compress_ctx *cc); 4304 void f2fs_decompress_end_io(struct decompress_io_ctx *dic, bool failed, 4305 bool in_task); 4306 void f2fs_put_page_dic(struct page *page, bool in_task); 4307 unsigned int f2fs_cluster_blocks_are_contiguous(struct dnode_of_data *dn, 4308 unsigned int ofs_in_node); 4309 int f2fs_init_compress_ctx(struct compress_ctx *cc); 4310 void f2fs_destroy_compress_ctx(struct compress_ctx *cc, bool reuse); 4311 void f2fs_init_compress_info(struct f2fs_sb_info *sbi); 4312 int f2fs_init_compress_inode(struct f2fs_sb_info *sbi); 4313 void f2fs_destroy_compress_inode(struct f2fs_sb_info *sbi); 4314 int f2fs_init_page_array_cache(struct f2fs_sb_info *sbi); 4315 void f2fs_destroy_page_array_cache(struct f2fs_sb_info *sbi); 4316 int __init f2fs_init_compress_cache(void); 4317 void f2fs_destroy_compress_cache(void); 4318 struct address_space *COMPRESS_MAPPING(struct f2fs_sb_info *sbi); 4319 void f2fs_invalidate_compress_page(struct f2fs_sb_info *sbi, block_t blkaddr); 4320 void f2fs_cache_compressed_page(struct f2fs_sb_info *sbi, struct page *page, 4321 nid_t ino, block_t blkaddr); 4322 bool f2fs_load_compressed_page(struct f2fs_sb_info *sbi, struct page *page, 4323 block_t blkaddr); 4324 void f2fs_invalidate_compress_pages(struct f2fs_sb_info *sbi, nid_t ino); 4325 #define inc_compr_inode_stat(inode) \ 4326 do { \ 4327 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); \ 4328 sbi->compr_new_inode++; \ 4329 } while (0) 4330 #define add_compr_block_stat(inode, blocks) \ 4331 do { \ 4332 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); \ 4333 int diff = F2FS_I(inode)->i_cluster_size - blocks; \ 4334 sbi->compr_written_block += blocks; \ 4335 sbi->compr_saved_block += diff; \ 4336 } while (0) 4337 #else 4338 static inline bool f2fs_is_compressed_page(struct page *page) { return false; } 4339 static inline bool f2fs_is_compress_backend_ready(struct inode *inode) 4340 { 4341 if (!f2fs_compressed_file(inode)) 4342 return true; 4343 /* not support compression */ 4344 return false; 4345 } 4346 static inline bool f2fs_is_compress_level_valid(int alg, int lvl) { return false; } 4347 static inline struct page *f2fs_compress_control_page(struct page *page) 4348 { 4349 WARN_ON_ONCE(1); 4350 return ERR_PTR(-EINVAL); 4351 } 4352 static inline int __init f2fs_init_compress_mempool(void) { return 0; } 4353 static inline void f2fs_destroy_compress_mempool(void) { } 4354 static inline void f2fs_decompress_cluster(struct decompress_io_ctx *dic, 4355 bool in_task) { } 4356 static inline void f2fs_end_read_compressed_page(struct page *page, 4357 bool failed, block_t blkaddr, bool in_task) 4358 { 4359 WARN_ON_ONCE(1); 4360 } 4361 static inline void f2fs_put_page_dic(struct page *page, bool in_task) 4362 { 4363 WARN_ON_ONCE(1); 4364 } 4365 static inline unsigned int f2fs_cluster_blocks_are_contiguous( 4366 struct dnode_of_data *dn, unsigned int ofs_in_node) { return 0; } 4367 static inline bool f2fs_sanity_check_cluster(struct dnode_of_data *dn) { return false; } 4368 static inline int f2fs_init_compress_inode(struct f2fs_sb_info *sbi) { return 0; } 4369 static inline void f2fs_destroy_compress_inode(struct f2fs_sb_info *sbi) { } 4370 static inline int f2fs_init_page_array_cache(struct f2fs_sb_info *sbi) { return 0; } 4371 static inline void f2fs_destroy_page_array_cache(struct f2fs_sb_info *sbi) { } 4372 static inline int __init f2fs_init_compress_cache(void) { return 0; } 4373 static inline void f2fs_destroy_compress_cache(void) { } 4374 static inline void f2fs_invalidate_compress_page(struct f2fs_sb_info *sbi, 4375 block_t blkaddr) { } 4376 static inline void f2fs_cache_compressed_page(struct f2fs_sb_info *sbi, 4377 struct page *page, nid_t ino, block_t blkaddr) { } 4378 static inline bool f2fs_load_compressed_page(struct f2fs_sb_info *sbi, 4379 struct page *page, block_t blkaddr) { return false; } 4380 static inline void f2fs_invalidate_compress_pages(struct f2fs_sb_info *sbi, 4381 nid_t ino) { } 4382 #define inc_compr_inode_stat(inode) do { } while (0) 4383 static inline void f2fs_update_read_extent_tree_range_compressed( 4384 struct inode *inode, 4385 pgoff_t fofs, block_t blkaddr, 4386 unsigned int llen, unsigned int c_len) { } 4387 #endif 4388 4389 static inline int set_compress_context(struct inode *inode) 4390 { 4391 #ifdef CONFIG_F2FS_FS_COMPRESSION 4392 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 4393 4394 F2FS_I(inode)->i_compress_algorithm = 4395 F2FS_OPTION(sbi).compress_algorithm; 4396 F2FS_I(inode)->i_log_cluster_size = 4397 F2FS_OPTION(sbi).compress_log_size; 4398 F2FS_I(inode)->i_compress_flag = 4399 F2FS_OPTION(sbi).compress_chksum ? 4400 BIT(COMPRESS_CHKSUM) : 0; 4401 F2FS_I(inode)->i_cluster_size = 4402 BIT(F2FS_I(inode)->i_log_cluster_size); 4403 if ((F2FS_I(inode)->i_compress_algorithm == COMPRESS_LZ4 || 4404 F2FS_I(inode)->i_compress_algorithm == COMPRESS_ZSTD) && 4405 F2FS_OPTION(sbi).compress_level) 4406 F2FS_I(inode)->i_compress_level = 4407 F2FS_OPTION(sbi).compress_level; 4408 F2FS_I(inode)->i_flags |= F2FS_COMPR_FL; 4409 set_inode_flag(inode, FI_COMPRESSED_FILE); 4410 stat_inc_compr_inode(inode); 4411 inc_compr_inode_stat(inode); 4412 f2fs_mark_inode_dirty_sync(inode, true); 4413 return 0; 4414 #else 4415 return -EOPNOTSUPP; 4416 #endif 4417 } 4418 4419 static inline bool f2fs_disable_compressed_file(struct inode *inode) 4420 { 4421 struct f2fs_inode_info *fi = F2FS_I(inode); 4422 4423 f2fs_down_write(&F2FS_I(inode)->i_sem); 4424 4425 if (!f2fs_compressed_file(inode)) { 4426 f2fs_up_write(&F2FS_I(inode)->i_sem); 4427 return true; 4428 } 4429 if (f2fs_is_mmap_file(inode) || 4430 (S_ISREG(inode->i_mode) && F2FS_HAS_BLOCKS(inode))) { 4431 f2fs_up_write(&F2FS_I(inode)->i_sem); 4432 return false; 4433 } 4434 4435 fi->i_flags &= ~F2FS_COMPR_FL; 4436 stat_dec_compr_inode(inode); 4437 clear_inode_flag(inode, FI_COMPRESSED_FILE); 4438 f2fs_mark_inode_dirty_sync(inode, true); 4439 4440 f2fs_up_write(&F2FS_I(inode)->i_sem); 4441 return true; 4442 } 4443 4444 #define F2FS_FEATURE_FUNCS(name, flagname) \ 4445 static inline bool f2fs_sb_has_##name(struct f2fs_sb_info *sbi) \ 4446 { \ 4447 return F2FS_HAS_FEATURE(sbi, F2FS_FEATURE_##flagname); \ 4448 } 4449 4450 F2FS_FEATURE_FUNCS(encrypt, ENCRYPT); 4451 F2FS_FEATURE_FUNCS(blkzoned, BLKZONED); 4452 F2FS_FEATURE_FUNCS(extra_attr, EXTRA_ATTR); 4453 F2FS_FEATURE_FUNCS(project_quota, PRJQUOTA); 4454 F2FS_FEATURE_FUNCS(inode_chksum, INODE_CHKSUM); 4455 F2FS_FEATURE_FUNCS(flexible_inline_xattr, FLEXIBLE_INLINE_XATTR); 4456 F2FS_FEATURE_FUNCS(quota_ino, QUOTA_INO); 4457 F2FS_FEATURE_FUNCS(inode_crtime, INODE_CRTIME); 4458 F2FS_FEATURE_FUNCS(lost_found, LOST_FOUND); 4459 F2FS_FEATURE_FUNCS(verity, VERITY); 4460 F2FS_FEATURE_FUNCS(sb_chksum, SB_CHKSUM); 4461 F2FS_FEATURE_FUNCS(casefold, CASEFOLD); 4462 F2FS_FEATURE_FUNCS(compression, COMPRESSION); 4463 F2FS_FEATURE_FUNCS(readonly, RO); 4464 4465 #ifdef CONFIG_BLK_DEV_ZONED 4466 static inline bool f2fs_blkz_is_seq(struct f2fs_sb_info *sbi, int devi, 4467 block_t blkaddr) 4468 { 4469 unsigned int zno = blkaddr / sbi->blocks_per_blkz; 4470 4471 return test_bit(zno, FDEV(devi).blkz_seq); 4472 } 4473 #endif 4474 4475 static inline int f2fs_bdev_index(struct f2fs_sb_info *sbi, 4476 struct block_device *bdev) 4477 { 4478 int i; 4479 4480 if (!f2fs_is_multi_device(sbi)) 4481 return 0; 4482 4483 for (i = 0; i < sbi->s_ndevs; i++) 4484 if (FDEV(i).bdev == bdev) 4485 return i; 4486 4487 WARN_ON(1); 4488 return -1; 4489 } 4490 4491 static inline bool f2fs_hw_should_discard(struct f2fs_sb_info *sbi) 4492 { 4493 return f2fs_sb_has_blkzoned(sbi); 4494 } 4495 4496 static inline bool f2fs_bdev_support_discard(struct block_device *bdev) 4497 { 4498 return bdev_max_discard_sectors(bdev) || bdev_is_zoned(bdev); 4499 } 4500 4501 static inline bool f2fs_hw_support_discard(struct f2fs_sb_info *sbi) 4502 { 4503 int i; 4504 4505 if (!f2fs_is_multi_device(sbi)) 4506 return f2fs_bdev_support_discard(sbi->sb->s_bdev); 4507 4508 for (i = 0; i < sbi->s_ndevs; i++) 4509 if (f2fs_bdev_support_discard(FDEV(i).bdev)) 4510 return true; 4511 return false; 4512 } 4513 4514 static inline bool f2fs_realtime_discard_enable(struct f2fs_sb_info *sbi) 4515 { 4516 return (test_opt(sbi, DISCARD) && f2fs_hw_support_discard(sbi)) || 4517 f2fs_hw_should_discard(sbi); 4518 } 4519 4520 static inline bool f2fs_hw_is_readonly(struct f2fs_sb_info *sbi) 4521 { 4522 int i; 4523 4524 if (!f2fs_is_multi_device(sbi)) 4525 return bdev_read_only(sbi->sb->s_bdev); 4526 4527 for (i = 0; i < sbi->s_ndevs; i++) 4528 if (bdev_read_only(FDEV(i).bdev)) 4529 return true; 4530 return false; 4531 } 4532 4533 static inline bool f2fs_dev_is_readonly(struct f2fs_sb_info *sbi) 4534 { 4535 return f2fs_sb_has_readonly(sbi) || f2fs_hw_is_readonly(sbi); 4536 } 4537 4538 static inline bool f2fs_lfs_mode(struct f2fs_sb_info *sbi) 4539 { 4540 return F2FS_OPTION(sbi).fs_mode == FS_MODE_LFS; 4541 } 4542 4543 static inline bool f2fs_valid_pinned_area(struct f2fs_sb_info *sbi, 4544 block_t blkaddr) 4545 { 4546 if (f2fs_sb_has_blkzoned(sbi)) { 4547 int devi = f2fs_target_device_index(sbi, blkaddr); 4548 4549 return !bdev_is_zoned(FDEV(devi).bdev); 4550 } 4551 return true; 4552 } 4553 4554 static inline bool f2fs_low_mem_mode(struct f2fs_sb_info *sbi) 4555 { 4556 return F2FS_OPTION(sbi).memory_mode == MEMORY_MODE_LOW; 4557 } 4558 4559 static inline bool f2fs_may_compress(struct inode *inode) 4560 { 4561 if (IS_SWAPFILE(inode) || f2fs_is_pinned_file(inode) || 4562 f2fs_is_atomic_file(inode) || f2fs_has_inline_data(inode) || 4563 f2fs_is_mmap_file(inode)) 4564 return false; 4565 return S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode); 4566 } 4567 4568 static inline void f2fs_i_compr_blocks_update(struct inode *inode, 4569 u64 blocks, bool add) 4570 { 4571 struct f2fs_inode_info *fi = F2FS_I(inode); 4572 int diff = fi->i_cluster_size - blocks; 4573 4574 /* don't update i_compr_blocks if saved blocks were released */ 4575 if (!add && !atomic_read(&fi->i_compr_blocks)) 4576 return; 4577 4578 if (add) { 4579 atomic_add(diff, &fi->i_compr_blocks); 4580 stat_add_compr_blocks(inode, diff); 4581 } else { 4582 atomic_sub(diff, &fi->i_compr_blocks); 4583 stat_sub_compr_blocks(inode, diff); 4584 } 4585 f2fs_mark_inode_dirty_sync(inode, true); 4586 } 4587 4588 static inline bool f2fs_allow_multi_device_dio(struct f2fs_sb_info *sbi, 4589 int flag) 4590 { 4591 if (!f2fs_is_multi_device(sbi)) 4592 return false; 4593 if (flag != F2FS_GET_BLOCK_DIO) 4594 return false; 4595 return sbi->aligned_blksize; 4596 } 4597 4598 static inline bool f2fs_need_verity(const struct inode *inode, pgoff_t idx) 4599 { 4600 return fsverity_active(inode) && 4601 idx < DIV_ROUND_UP(inode->i_size, PAGE_SIZE); 4602 } 4603 4604 #ifdef CONFIG_F2FS_FAULT_INJECTION 4605 extern void f2fs_build_fault_attr(struct f2fs_sb_info *sbi, unsigned int rate, 4606 unsigned int type); 4607 #else 4608 #define f2fs_build_fault_attr(sbi, rate, type) do { } while (0) 4609 #endif 4610 4611 static inline bool is_journalled_quota(struct f2fs_sb_info *sbi) 4612 { 4613 #ifdef CONFIG_QUOTA 4614 if (f2fs_sb_has_quota_ino(sbi)) 4615 return true; 4616 if (F2FS_OPTION(sbi).s_qf_names[USRQUOTA] || 4617 F2FS_OPTION(sbi).s_qf_names[GRPQUOTA] || 4618 F2FS_OPTION(sbi).s_qf_names[PRJQUOTA]) 4619 return true; 4620 #endif 4621 return false; 4622 } 4623 4624 static inline bool f2fs_block_unit_discard(struct f2fs_sb_info *sbi) 4625 { 4626 return F2FS_OPTION(sbi).discard_unit == DISCARD_UNIT_BLOCK; 4627 } 4628 4629 static inline void f2fs_io_schedule_timeout(long timeout) 4630 { 4631 set_current_state(TASK_UNINTERRUPTIBLE); 4632 io_schedule_timeout(timeout); 4633 } 4634 4635 static inline void f2fs_handle_page_eio(struct f2fs_sb_info *sbi, pgoff_t ofs, 4636 enum page_type type) 4637 { 4638 if (unlikely(f2fs_cp_error(sbi))) 4639 return; 4640 4641 if (ofs == sbi->page_eio_ofs[type]) { 4642 if (sbi->page_eio_cnt[type]++ == MAX_RETRY_PAGE_EIO) 4643 set_ckpt_flags(sbi, CP_ERROR_FLAG); 4644 } else { 4645 sbi->page_eio_ofs[type] = ofs; 4646 sbi->page_eio_cnt[type] = 0; 4647 } 4648 } 4649 4650 static inline bool f2fs_is_readonly(struct f2fs_sb_info *sbi) 4651 { 4652 return f2fs_sb_has_readonly(sbi) || f2fs_readonly(sbi->sb); 4653 } 4654 4655 static inline void f2fs_truncate_meta_inode_pages(struct f2fs_sb_info *sbi, 4656 block_t blkaddr, unsigned int cnt) 4657 { 4658 bool need_submit = false; 4659 int i = 0; 4660 4661 do { 4662 struct page *page; 4663 4664 page = find_get_page(META_MAPPING(sbi), blkaddr + i); 4665 if (page) { 4666 if (PageWriteback(page)) 4667 need_submit = true; 4668 f2fs_put_page(page, 0); 4669 } 4670 } while (++i < cnt && !need_submit); 4671 4672 if (need_submit) 4673 f2fs_submit_merged_write_cond(sbi, sbi->meta_inode, 4674 NULL, 0, DATA); 4675 4676 truncate_inode_pages_range(META_MAPPING(sbi), 4677 F2FS_BLK_TO_BYTES((loff_t)blkaddr), 4678 F2FS_BLK_END_BYTES((loff_t)(blkaddr + cnt - 1))); 4679 } 4680 4681 static inline void f2fs_invalidate_internal_cache(struct f2fs_sb_info *sbi, 4682 block_t blkaddr) 4683 { 4684 f2fs_truncate_meta_inode_pages(sbi, blkaddr, 1); 4685 f2fs_invalidate_compress_page(sbi, blkaddr); 4686 } 4687 4688 #define EFSBADCRC EBADMSG /* Bad CRC detected */ 4689 #define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */ 4690 4691 #endif /* _LINUX_F2FS_H */ 4692