1 /* SPDX-License-Identifier: GPL-2.0 */ 2 /* 3 * fs/f2fs/f2fs.h 4 * 5 * Copyright (c) 2012 Samsung Electronics Co., Ltd. 6 * http://www.samsung.com/ 7 */ 8 #ifndef _LINUX_F2FS_H 9 #define _LINUX_F2FS_H 10 11 #include <linux/uio.h> 12 #include <linux/types.h> 13 #include <linux/page-flags.h> 14 #include <linux/buffer_head.h> 15 #include <linux/slab.h> 16 #include <linux/crc32.h> 17 #include <linux/magic.h> 18 #include <linux/kobject.h> 19 #include <linux/sched.h> 20 #include <linux/cred.h> 21 #include <linux/sched/mm.h> 22 #include <linux/vmalloc.h> 23 #include <linux/bio.h> 24 #include <linux/blkdev.h> 25 #include <linux/quotaops.h> 26 #include <linux/part_stat.h> 27 #include <crypto/hash.h> 28 29 #include <linux/fscrypt.h> 30 #include <linux/fsverity.h> 31 32 struct pagevec; 33 34 #ifdef CONFIG_F2FS_CHECK_FS 35 #define f2fs_bug_on(sbi, condition) BUG_ON(condition) 36 #else 37 #define f2fs_bug_on(sbi, condition) \ 38 do { \ 39 if (WARN_ON(condition)) \ 40 set_sbi_flag(sbi, SBI_NEED_FSCK); \ 41 } while (0) 42 #endif 43 44 enum { 45 FAULT_KMALLOC, 46 FAULT_KVMALLOC, 47 FAULT_PAGE_ALLOC, 48 FAULT_PAGE_GET, 49 FAULT_ALLOC_BIO, /* it's obsolete due to bio_alloc() will never fail */ 50 FAULT_ALLOC_NID, 51 FAULT_ORPHAN, 52 FAULT_BLOCK, 53 FAULT_DIR_DEPTH, 54 FAULT_EVICT_INODE, 55 FAULT_TRUNCATE, 56 FAULT_READ_IO, 57 FAULT_CHECKPOINT, 58 FAULT_DISCARD, 59 FAULT_WRITE_IO, 60 FAULT_SLAB_ALLOC, 61 FAULT_DQUOT_INIT, 62 FAULT_LOCK_OP, 63 FAULT_BLKADDR, 64 FAULT_MAX, 65 }; 66 67 #ifdef CONFIG_F2FS_FAULT_INJECTION 68 #define F2FS_ALL_FAULT_TYPE (GENMASK(FAULT_MAX - 1, 0)) 69 70 struct f2fs_fault_info { 71 atomic_t inject_ops; 72 unsigned int inject_rate; 73 unsigned int inject_type; 74 }; 75 76 extern const char *f2fs_fault_name[FAULT_MAX]; 77 #define IS_FAULT_SET(fi, type) ((fi)->inject_type & BIT(type)) 78 #endif 79 80 /* 81 * For mount options 82 */ 83 #define F2FS_MOUNT_DISABLE_ROLL_FORWARD 0x00000001 84 #define F2FS_MOUNT_DISCARD 0x00000002 85 #define F2FS_MOUNT_NOHEAP 0x00000004 86 #define F2FS_MOUNT_XATTR_USER 0x00000008 87 #define F2FS_MOUNT_POSIX_ACL 0x00000010 88 #define F2FS_MOUNT_DISABLE_EXT_IDENTIFY 0x00000020 89 #define F2FS_MOUNT_INLINE_XATTR 0x00000040 90 #define F2FS_MOUNT_INLINE_DATA 0x00000080 91 #define F2FS_MOUNT_INLINE_DENTRY 0x00000100 92 #define F2FS_MOUNT_FLUSH_MERGE 0x00000200 93 #define F2FS_MOUNT_NOBARRIER 0x00000400 94 #define F2FS_MOUNT_FASTBOOT 0x00000800 95 #define F2FS_MOUNT_READ_EXTENT_CACHE 0x00001000 96 #define F2FS_MOUNT_DATA_FLUSH 0x00002000 97 #define F2FS_MOUNT_FAULT_INJECTION 0x00004000 98 #define F2FS_MOUNT_USRQUOTA 0x00008000 99 #define F2FS_MOUNT_GRPQUOTA 0x00010000 100 #define F2FS_MOUNT_PRJQUOTA 0x00020000 101 #define F2FS_MOUNT_QUOTA 0x00040000 102 #define F2FS_MOUNT_INLINE_XATTR_SIZE 0x00080000 103 #define F2FS_MOUNT_RESERVE_ROOT 0x00100000 104 #define F2FS_MOUNT_DISABLE_CHECKPOINT 0x00200000 105 #define F2FS_MOUNT_NORECOVERY 0x00400000 106 #define F2FS_MOUNT_ATGC 0x00800000 107 #define F2FS_MOUNT_MERGE_CHECKPOINT 0x01000000 108 #define F2FS_MOUNT_GC_MERGE 0x02000000 109 #define F2FS_MOUNT_COMPRESS_CACHE 0x04000000 110 #define F2FS_MOUNT_AGE_EXTENT_CACHE 0x08000000 111 112 #define F2FS_OPTION(sbi) ((sbi)->mount_opt) 113 #define clear_opt(sbi, option) (F2FS_OPTION(sbi).opt &= ~F2FS_MOUNT_##option) 114 #define set_opt(sbi, option) (F2FS_OPTION(sbi).opt |= F2FS_MOUNT_##option) 115 #define test_opt(sbi, option) (F2FS_OPTION(sbi).opt & F2FS_MOUNT_##option) 116 117 #define ver_after(a, b) (typecheck(unsigned long long, a) && \ 118 typecheck(unsigned long long, b) && \ 119 ((long long)((a) - (b)) > 0)) 120 121 typedef u32 block_t; /* 122 * should not change u32, since it is the on-disk block 123 * address format, __le32. 124 */ 125 typedef u32 nid_t; 126 127 #define COMPRESS_EXT_NUM 16 128 129 /* 130 * An implementation of an rwsem that is explicitly unfair to readers. This 131 * prevents priority inversion when a low-priority reader acquires the read lock 132 * while sleeping on the write lock but the write lock is needed by 133 * higher-priority clients. 134 */ 135 136 struct f2fs_rwsem { 137 struct rw_semaphore internal_rwsem; 138 #ifdef CONFIG_F2FS_UNFAIR_RWSEM 139 wait_queue_head_t read_waiters; 140 #endif 141 }; 142 143 struct f2fs_mount_info { 144 unsigned int opt; 145 int write_io_size_bits; /* Write IO size bits */ 146 block_t root_reserved_blocks; /* root reserved blocks */ 147 kuid_t s_resuid; /* reserved blocks for uid */ 148 kgid_t s_resgid; /* reserved blocks for gid */ 149 int active_logs; /* # of active logs */ 150 int inline_xattr_size; /* inline xattr size */ 151 #ifdef CONFIG_F2FS_FAULT_INJECTION 152 struct f2fs_fault_info fault_info; /* For fault injection */ 153 #endif 154 #ifdef CONFIG_QUOTA 155 /* Names of quota files with journalled quota */ 156 char *s_qf_names[MAXQUOTAS]; 157 int s_jquota_fmt; /* Format of quota to use */ 158 #endif 159 /* For which write hints are passed down to block layer */ 160 int alloc_mode; /* segment allocation policy */ 161 int fsync_mode; /* fsync policy */ 162 int fs_mode; /* fs mode: LFS or ADAPTIVE */ 163 int bggc_mode; /* bggc mode: off, on or sync */ 164 int memory_mode; /* memory mode */ 165 int errors; /* errors parameter */ 166 int discard_unit; /* 167 * discard command's offset/size should 168 * be aligned to this unit: block, 169 * segment or section 170 */ 171 struct fscrypt_dummy_policy dummy_enc_policy; /* test dummy encryption */ 172 block_t unusable_cap_perc; /* percentage for cap */ 173 block_t unusable_cap; /* Amount of space allowed to be 174 * unusable when disabling checkpoint 175 */ 176 177 /* For compression */ 178 unsigned char compress_algorithm; /* algorithm type */ 179 unsigned char compress_log_size; /* cluster log size */ 180 unsigned char compress_level; /* compress level */ 181 bool compress_chksum; /* compressed data chksum */ 182 unsigned char compress_ext_cnt; /* extension count */ 183 unsigned char nocompress_ext_cnt; /* nocompress extension count */ 184 int compress_mode; /* compression mode */ 185 unsigned char extensions[COMPRESS_EXT_NUM][F2FS_EXTENSION_LEN]; /* extensions */ 186 unsigned char noextensions[COMPRESS_EXT_NUM][F2FS_EXTENSION_LEN]; /* extensions */ 187 }; 188 189 #define F2FS_FEATURE_ENCRYPT 0x00000001 190 #define F2FS_FEATURE_BLKZONED 0x00000002 191 #define F2FS_FEATURE_ATOMIC_WRITE 0x00000004 192 #define F2FS_FEATURE_EXTRA_ATTR 0x00000008 193 #define F2FS_FEATURE_PRJQUOTA 0x00000010 194 #define F2FS_FEATURE_INODE_CHKSUM 0x00000020 195 #define F2FS_FEATURE_FLEXIBLE_INLINE_XATTR 0x00000040 196 #define F2FS_FEATURE_QUOTA_INO 0x00000080 197 #define F2FS_FEATURE_INODE_CRTIME 0x00000100 198 #define F2FS_FEATURE_LOST_FOUND 0x00000200 199 #define F2FS_FEATURE_VERITY 0x00000400 200 #define F2FS_FEATURE_SB_CHKSUM 0x00000800 201 #define F2FS_FEATURE_CASEFOLD 0x00001000 202 #define F2FS_FEATURE_COMPRESSION 0x00002000 203 #define F2FS_FEATURE_RO 0x00004000 204 205 #define __F2FS_HAS_FEATURE(raw_super, mask) \ 206 ((raw_super->feature & cpu_to_le32(mask)) != 0) 207 #define F2FS_HAS_FEATURE(sbi, mask) __F2FS_HAS_FEATURE(sbi->raw_super, mask) 208 209 /* 210 * Default values for user and/or group using reserved blocks 211 */ 212 #define F2FS_DEF_RESUID 0 213 #define F2FS_DEF_RESGID 0 214 215 /* 216 * For checkpoint manager 217 */ 218 enum { 219 NAT_BITMAP, 220 SIT_BITMAP 221 }; 222 223 #define CP_UMOUNT 0x00000001 224 #define CP_FASTBOOT 0x00000002 225 #define CP_SYNC 0x00000004 226 #define CP_RECOVERY 0x00000008 227 #define CP_DISCARD 0x00000010 228 #define CP_TRIMMED 0x00000020 229 #define CP_PAUSE 0x00000040 230 #define CP_RESIZE 0x00000080 231 232 #define DEF_MAX_DISCARD_REQUEST 8 /* issue 8 discards per round */ 233 #define DEF_MIN_DISCARD_ISSUE_TIME 50 /* 50 ms, if exists */ 234 #define DEF_MID_DISCARD_ISSUE_TIME 500 /* 500 ms, if device busy */ 235 #define DEF_MAX_DISCARD_ISSUE_TIME 60000 /* 60 s, if no candidates */ 236 #define DEF_DISCARD_URGENT_UTIL 80 /* do more discard over 80% */ 237 #define DEF_CP_INTERVAL 60 /* 60 secs */ 238 #define DEF_IDLE_INTERVAL 5 /* 5 secs */ 239 #define DEF_DISABLE_INTERVAL 5 /* 5 secs */ 240 #define DEF_DISABLE_QUICK_INTERVAL 1 /* 1 secs */ 241 #define DEF_UMOUNT_DISCARD_TIMEOUT 5 /* 5 secs */ 242 243 struct cp_control { 244 int reason; 245 __u64 trim_start; 246 __u64 trim_end; 247 __u64 trim_minlen; 248 }; 249 250 /* 251 * indicate meta/data type 252 */ 253 enum { 254 META_CP, 255 META_NAT, 256 META_SIT, 257 META_SSA, 258 META_MAX, 259 META_POR, 260 DATA_GENERIC, /* check range only */ 261 DATA_GENERIC_ENHANCE, /* strong check on range and segment bitmap */ 262 DATA_GENERIC_ENHANCE_READ, /* 263 * strong check on range and segment 264 * bitmap but no warning due to race 265 * condition of read on truncated area 266 * by extent_cache 267 */ 268 DATA_GENERIC_ENHANCE_UPDATE, /* 269 * strong check on range and segment 270 * bitmap for update case 271 */ 272 META_GENERIC, 273 }; 274 275 /* for the list of ino */ 276 enum { 277 ORPHAN_INO, /* for orphan ino list */ 278 APPEND_INO, /* for append ino list */ 279 UPDATE_INO, /* for update ino list */ 280 TRANS_DIR_INO, /* for transactions dir ino list */ 281 FLUSH_INO, /* for multiple device flushing */ 282 MAX_INO_ENTRY, /* max. list */ 283 }; 284 285 struct ino_entry { 286 struct list_head list; /* list head */ 287 nid_t ino; /* inode number */ 288 unsigned int dirty_device; /* dirty device bitmap */ 289 }; 290 291 /* for the list of inodes to be GCed */ 292 struct inode_entry { 293 struct list_head list; /* list head */ 294 struct inode *inode; /* vfs inode pointer */ 295 }; 296 297 struct fsync_node_entry { 298 struct list_head list; /* list head */ 299 struct page *page; /* warm node page pointer */ 300 unsigned int seq_id; /* sequence id */ 301 }; 302 303 struct ckpt_req { 304 struct completion wait; /* completion for checkpoint done */ 305 struct llist_node llnode; /* llist_node to be linked in wait queue */ 306 int ret; /* return code of checkpoint */ 307 ktime_t queue_time; /* request queued time */ 308 }; 309 310 struct ckpt_req_control { 311 struct task_struct *f2fs_issue_ckpt; /* checkpoint task */ 312 int ckpt_thread_ioprio; /* checkpoint merge thread ioprio */ 313 wait_queue_head_t ckpt_wait_queue; /* waiting queue for wake-up */ 314 atomic_t issued_ckpt; /* # of actually issued ckpts */ 315 atomic_t total_ckpt; /* # of total ckpts */ 316 atomic_t queued_ckpt; /* # of queued ckpts */ 317 struct llist_head issue_list; /* list for command issue */ 318 spinlock_t stat_lock; /* lock for below checkpoint time stats */ 319 unsigned int cur_time; /* cur wait time in msec for currently issued checkpoint */ 320 unsigned int peak_time; /* peak wait time in msec until now */ 321 }; 322 323 /* for the bitmap indicate blocks to be discarded */ 324 struct discard_entry { 325 struct list_head list; /* list head */ 326 block_t start_blkaddr; /* start blockaddr of current segment */ 327 unsigned char discard_map[SIT_VBLOCK_MAP_SIZE]; /* segment discard bitmap */ 328 }; 329 330 /* minimum discard granularity, unit: block count */ 331 #define MIN_DISCARD_GRANULARITY 1 332 /* default discard granularity of inner discard thread, unit: block count */ 333 #define DEFAULT_DISCARD_GRANULARITY 16 334 /* default maximum discard granularity of ordered discard, unit: block count */ 335 #define DEFAULT_MAX_ORDERED_DISCARD_GRANULARITY 16 336 337 /* max discard pend list number */ 338 #define MAX_PLIST_NUM 512 339 #define plist_idx(blk_num) ((blk_num) >= MAX_PLIST_NUM ? \ 340 (MAX_PLIST_NUM - 1) : ((blk_num) - 1)) 341 342 enum { 343 D_PREP, /* initial */ 344 D_PARTIAL, /* partially submitted */ 345 D_SUBMIT, /* all submitted */ 346 D_DONE, /* finished */ 347 }; 348 349 struct discard_info { 350 block_t lstart; /* logical start address */ 351 block_t len; /* length */ 352 block_t start; /* actual start address in dev */ 353 }; 354 355 struct discard_cmd { 356 struct rb_node rb_node; /* rb node located in rb-tree */ 357 struct discard_info di; /* discard info */ 358 struct list_head list; /* command list */ 359 struct completion wait; /* compleation */ 360 struct block_device *bdev; /* bdev */ 361 unsigned short ref; /* reference count */ 362 unsigned char state; /* state */ 363 unsigned char queued; /* queued discard */ 364 int error; /* bio error */ 365 spinlock_t lock; /* for state/bio_ref updating */ 366 unsigned short bio_ref; /* bio reference count */ 367 }; 368 369 enum { 370 DPOLICY_BG, 371 DPOLICY_FORCE, 372 DPOLICY_FSTRIM, 373 DPOLICY_UMOUNT, 374 MAX_DPOLICY, 375 }; 376 377 enum { 378 DPOLICY_IO_AWARE_DISABLE, /* force to not be aware of IO */ 379 DPOLICY_IO_AWARE_ENABLE, /* force to be aware of IO */ 380 DPOLICY_IO_AWARE_MAX, 381 }; 382 383 struct discard_policy { 384 int type; /* type of discard */ 385 unsigned int min_interval; /* used for candidates exist */ 386 unsigned int mid_interval; /* used for device busy */ 387 unsigned int max_interval; /* used for candidates not exist */ 388 unsigned int max_requests; /* # of discards issued per round */ 389 unsigned int io_aware_gran; /* minimum granularity discard not be aware of I/O */ 390 bool io_aware; /* issue discard in idle time */ 391 bool sync; /* submit discard with REQ_SYNC flag */ 392 bool ordered; /* issue discard by lba order */ 393 bool timeout; /* discard timeout for put_super */ 394 unsigned int granularity; /* discard granularity */ 395 }; 396 397 struct discard_cmd_control { 398 struct task_struct *f2fs_issue_discard; /* discard thread */ 399 struct list_head entry_list; /* 4KB discard entry list */ 400 struct list_head pend_list[MAX_PLIST_NUM];/* store pending entries */ 401 struct list_head wait_list; /* store on-flushing entries */ 402 struct list_head fstrim_list; /* in-flight discard from fstrim */ 403 wait_queue_head_t discard_wait_queue; /* waiting queue for wake-up */ 404 struct mutex cmd_lock; 405 unsigned int nr_discards; /* # of discards in the list */ 406 unsigned int max_discards; /* max. discards to be issued */ 407 unsigned int max_discard_request; /* max. discard request per round */ 408 unsigned int min_discard_issue_time; /* min. interval between discard issue */ 409 unsigned int mid_discard_issue_time; /* mid. interval between discard issue */ 410 unsigned int max_discard_issue_time; /* max. interval between discard issue */ 411 unsigned int discard_io_aware_gran; /* minimum discard granularity not be aware of I/O */ 412 unsigned int discard_urgent_util; /* utilization which issue discard proactively */ 413 unsigned int discard_granularity; /* discard granularity */ 414 unsigned int max_ordered_discard; /* maximum discard granularity issued by lba order */ 415 unsigned int discard_io_aware; /* io_aware policy */ 416 unsigned int undiscard_blks; /* # of undiscard blocks */ 417 unsigned int next_pos; /* next discard position */ 418 atomic_t issued_discard; /* # of issued discard */ 419 atomic_t queued_discard; /* # of queued discard */ 420 atomic_t discard_cmd_cnt; /* # of cached cmd count */ 421 struct rb_root_cached root; /* root of discard rb-tree */ 422 bool rbtree_check; /* config for consistence check */ 423 bool discard_wake; /* to wake up discard thread */ 424 }; 425 426 /* for the list of fsync inodes, used only during recovery */ 427 struct fsync_inode_entry { 428 struct list_head list; /* list head */ 429 struct inode *inode; /* vfs inode pointer */ 430 block_t blkaddr; /* block address locating the last fsync */ 431 block_t last_dentry; /* block address locating the last dentry */ 432 }; 433 434 #define nats_in_cursum(jnl) (le16_to_cpu((jnl)->n_nats)) 435 #define sits_in_cursum(jnl) (le16_to_cpu((jnl)->n_sits)) 436 437 #define nat_in_journal(jnl, i) ((jnl)->nat_j.entries[i].ne) 438 #define nid_in_journal(jnl, i) ((jnl)->nat_j.entries[i].nid) 439 #define sit_in_journal(jnl, i) ((jnl)->sit_j.entries[i].se) 440 #define segno_in_journal(jnl, i) ((jnl)->sit_j.entries[i].segno) 441 442 #define MAX_NAT_JENTRIES(jnl) (NAT_JOURNAL_ENTRIES - nats_in_cursum(jnl)) 443 #define MAX_SIT_JENTRIES(jnl) (SIT_JOURNAL_ENTRIES - sits_in_cursum(jnl)) 444 445 static inline int update_nats_in_cursum(struct f2fs_journal *journal, int i) 446 { 447 int before = nats_in_cursum(journal); 448 449 journal->n_nats = cpu_to_le16(before + i); 450 return before; 451 } 452 453 static inline int update_sits_in_cursum(struct f2fs_journal *journal, int i) 454 { 455 int before = sits_in_cursum(journal); 456 457 journal->n_sits = cpu_to_le16(before + i); 458 return before; 459 } 460 461 static inline bool __has_cursum_space(struct f2fs_journal *journal, 462 int size, int type) 463 { 464 if (type == NAT_JOURNAL) 465 return size <= MAX_NAT_JENTRIES(journal); 466 return size <= MAX_SIT_JENTRIES(journal); 467 } 468 469 /* for inline stuff */ 470 #define DEF_INLINE_RESERVED_SIZE 1 471 static inline int get_extra_isize(struct inode *inode); 472 static inline int get_inline_xattr_addrs(struct inode *inode); 473 #define MAX_INLINE_DATA(inode) (sizeof(__le32) * \ 474 (CUR_ADDRS_PER_INODE(inode) - \ 475 get_inline_xattr_addrs(inode) - \ 476 DEF_INLINE_RESERVED_SIZE)) 477 478 /* for inline dir */ 479 #define NR_INLINE_DENTRY(inode) (MAX_INLINE_DATA(inode) * BITS_PER_BYTE / \ 480 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \ 481 BITS_PER_BYTE + 1)) 482 #define INLINE_DENTRY_BITMAP_SIZE(inode) \ 483 DIV_ROUND_UP(NR_INLINE_DENTRY(inode), BITS_PER_BYTE) 484 #define INLINE_RESERVED_SIZE(inode) (MAX_INLINE_DATA(inode) - \ 485 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \ 486 NR_INLINE_DENTRY(inode) + \ 487 INLINE_DENTRY_BITMAP_SIZE(inode))) 488 489 /* 490 * For INODE and NODE manager 491 */ 492 /* for directory operations */ 493 494 struct f2fs_filename { 495 /* 496 * The filename the user specified. This is NULL for some 497 * filesystem-internal operations, e.g. converting an inline directory 498 * to a non-inline one, or roll-forward recovering an encrypted dentry. 499 */ 500 const struct qstr *usr_fname; 501 502 /* 503 * The on-disk filename. For encrypted directories, this is encrypted. 504 * This may be NULL for lookups in an encrypted dir without the key. 505 */ 506 struct fscrypt_str disk_name; 507 508 /* The dirhash of this filename */ 509 f2fs_hash_t hash; 510 511 #ifdef CONFIG_FS_ENCRYPTION 512 /* 513 * For lookups in encrypted directories: either the buffer backing 514 * disk_name, or a buffer that holds the decoded no-key name. 515 */ 516 struct fscrypt_str crypto_buf; 517 #endif 518 #if IS_ENABLED(CONFIG_UNICODE) 519 /* 520 * For casefolded directories: the casefolded name, but it's left NULL 521 * if the original name is not valid Unicode, if the original name is 522 * "." or "..", if the directory is both casefolded and encrypted and 523 * its encryption key is unavailable, or if the filesystem is doing an 524 * internal operation where usr_fname is also NULL. In all these cases 525 * we fall back to treating the name as an opaque byte sequence. 526 */ 527 struct fscrypt_str cf_name; 528 #endif 529 }; 530 531 struct f2fs_dentry_ptr { 532 struct inode *inode; 533 void *bitmap; 534 struct f2fs_dir_entry *dentry; 535 __u8 (*filename)[F2FS_SLOT_LEN]; 536 int max; 537 int nr_bitmap; 538 }; 539 540 static inline void make_dentry_ptr_block(struct inode *inode, 541 struct f2fs_dentry_ptr *d, struct f2fs_dentry_block *t) 542 { 543 d->inode = inode; 544 d->max = NR_DENTRY_IN_BLOCK; 545 d->nr_bitmap = SIZE_OF_DENTRY_BITMAP; 546 d->bitmap = t->dentry_bitmap; 547 d->dentry = t->dentry; 548 d->filename = t->filename; 549 } 550 551 static inline void make_dentry_ptr_inline(struct inode *inode, 552 struct f2fs_dentry_ptr *d, void *t) 553 { 554 int entry_cnt = NR_INLINE_DENTRY(inode); 555 int bitmap_size = INLINE_DENTRY_BITMAP_SIZE(inode); 556 int reserved_size = INLINE_RESERVED_SIZE(inode); 557 558 d->inode = inode; 559 d->max = entry_cnt; 560 d->nr_bitmap = bitmap_size; 561 d->bitmap = t; 562 d->dentry = t + bitmap_size + reserved_size; 563 d->filename = t + bitmap_size + reserved_size + 564 SIZE_OF_DIR_ENTRY * entry_cnt; 565 } 566 567 /* 568 * XATTR_NODE_OFFSET stores xattrs to one node block per file keeping -1 569 * as its node offset to distinguish from index node blocks. 570 * But some bits are used to mark the node block. 571 */ 572 #define XATTR_NODE_OFFSET ((((unsigned int)-1) << OFFSET_BIT_SHIFT) \ 573 >> OFFSET_BIT_SHIFT) 574 enum { 575 ALLOC_NODE, /* allocate a new node page if needed */ 576 LOOKUP_NODE, /* look up a node without readahead */ 577 LOOKUP_NODE_RA, /* 578 * look up a node with readahead called 579 * by get_data_block. 580 */ 581 }; 582 583 #define DEFAULT_RETRY_IO_COUNT 8 /* maximum retry read IO or flush count */ 584 585 /* congestion wait timeout value, default: 20ms */ 586 #define DEFAULT_IO_TIMEOUT (msecs_to_jiffies(20)) 587 588 /* maximum retry quota flush count */ 589 #define DEFAULT_RETRY_QUOTA_FLUSH_COUNT 8 590 591 /* maximum retry of EIO'ed page */ 592 #define MAX_RETRY_PAGE_EIO 100 593 594 #define F2FS_LINK_MAX 0xffffffff /* maximum link count per file */ 595 596 #define MAX_DIR_RA_PAGES 4 /* maximum ra pages of dir */ 597 598 /* dirty segments threshold for triggering CP */ 599 #define DEFAULT_DIRTY_THRESHOLD 4 600 601 #define RECOVERY_MAX_RA_BLOCKS BIO_MAX_VECS 602 #define RECOVERY_MIN_RA_BLOCKS 1 603 604 #define F2FS_ONSTACK_PAGES 16 /* nr of onstack pages */ 605 606 /* for in-memory extent cache entry */ 607 #define F2FS_MIN_EXTENT_LEN 64 /* minimum extent length */ 608 609 /* number of extent info in extent cache we try to shrink */ 610 #define READ_EXTENT_CACHE_SHRINK_NUMBER 128 611 612 /* number of age extent info in extent cache we try to shrink */ 613 #define AGE_EXTENT_CACHE_SHRINK_NUMBER 128 614 #define LAST_AGE_WEIGHT 30 615 #define SAME_AGE_REGION 1024 616 617 /* 618 * Define data block with age less than 1GB as hot data 619 * define data block with age less than 10GB but more than 1GB as warm data 620 */ 621 #define DEF_HOT_DATA_AGE_THRESHOLD 262144 622 #define DEF_WARM_DATA_AGE_THRESHOLD 2621440 623 624 /* extent cache type */ 625 enum extent_type { 626 EX_READ, 627 EX_BLOCK_AGE, 628 NR_EXTENT_CACHES, 629 }; 630 631 struct extent_info { 632 unsigned int fofs; /* start offset in a file */ 633 unsigned int len; /* length of the extent */ 634 union { 635 /* read extent_cache */ 636 struct { 637 /* start block address of the extent */ 638 block_t blk; 639 #ifdef CONFIG_F2FS_FS_COMPRESSION 640 /* physical extent length of compressed blocks */ 641 unsigned int c_len; 642 #endif 643 }; 644 /* block age extent_cache */ 645 struct { 646 /* block age of the extent */ 647 unsigned long long age; 648 /* last total blocks allocated */ 649 unsigned long long last_blocks; 650 }; 651 }; 652 }; 653 654 struct extent_node { 655 struct rb_node rb_node; /* rb node located in rb-tree */ 656 struct extent_info ei; /* extent info */ 657 struct list_head list; /* node in global extent list of sbi */ 658 struct extent_tree *et; /* extent tree pointer */ 659 }; 660 661 struct extent_tree { 662 nid_t ino; /* inode number */ 663 enum extent_type type; /* keep the extent tree type */ 664 struct rb_root_cached root; /* root of extent info rb-tree */ 665 struct extent_node *cached_en; /* recently accessed extent node */ 666 struct list_head list; /* to be used by sbi->zombie_list */ 667 rwlock_t lock; /* protect extent info rb-tree */ 668 atomic_t node_cnt; /* # of extent node in rb-tree*/ 669 bool largest_updated; /* largest extent updated */ 670 struct extent_info largest; /* largest cached extent for EX_READ */ 671 }; 672 673 struct extent_tree_info { 674 struct radix_tree_root extent_tree_root;/* cache extent cache entries */ 675 struct mutex extent_tree_lock; /* locking extent radix tree */ 676 struct list_head extent_list; /* lru list for shrinker */ 677 spinlock_t extent_lock; /* locking extent lru list */ 678 atomic_t total_ext_tree; /* extent tree count */ 679 struct list_head zombie_list; /* extent zombie tree list */ 680 atomic_t total_zombie_tree; /* extent zombie tree count */ 681 atomic_t total_ext_node; /* extent info count */ 682 }; 683 684 /* 685 * State of block returned by f2fs_map_blocks. 686 */ 687 #define F2FS_MAP_NEW (1U << 0) 688 #define F2FS_MAP_MAPPED (1U << 1) 689 #define F2FS_MAP_DELALLOC (1U << 2) 690 #define F2FS_MAP_FLAGS (F2FS_MAP_NEW | F2FS_MAP_MAPPED |\ 691 F2FS_MAP_DELALLOC) 692 693 struct f2fs_map_blocks { 694 struct block_device *m_bdev; /* for multi-device dio */ 695 block_t m_pblk; 696 block_t m_lblk; 697 unsigned int m_len; 698 unsigned int m_flags; 699 pgoff_t *m_next_pgofs; /* point next possible non-hole pgofs */ 700 pgoff_t *m_next_extent; /* point to next possible extent */ 701 int m_seg_type; 702 bool m_may_create; /* indicate it is from write path */ 703 bool m_multidev_dio; /* indicate it allows multi-device dio */ 704 }; 705 706 /* for flag in get_data_block */ 707 enum { 708 F2FS_GET_BLOCK_DEFAULT, 709 F2FS_GET_BLOCK_FIEMAP, 710 F2FS_GET_BLOCK_BMAP, 711 F2FS_GET_BLOCK_DIO, 712 F2FS_GET_BLOCK_PRE_DIO, 713 F2FS_GET_BLOCK_PRE_AIO, 714 F2FS_GET_BLOCK_PRECACHE, 715 }; 716 717 /* 718 * i_advise uses FADVISE_XXX_BIT. We can add additional hints later. 719 */ 720 #define FADVISE_COLD_BIT 0x01 721 #define FADVISE_LOST_PINO_BIT 0x02 722 #define FADVISE_ENCRYPT_BIT 0x04 723 #define FADVISE_ENC_NAME_BIT 0x08 724 #define FADVISE_KEEP_SIZE_BIT 0x10 725 #define FADVISE_HOT_BIT 0x20 726 #define FADVISE_VERITY_BIT 0x40 727 #define FADVISE_TRUNC_BIT 0x80 728 729 #define FADVISE_MODIFIABLE_BITS (FADVISE_COLD_BIT | FADVISE_HOT_BIT) 730 731 #define file_is_cold(inode) is_file(inode, FADVISE_COLD_BIT) 732 #define file_set_cold(inode) set_file(inode, FADVISE_COLD_BIT) 733 #define file_clear_cold(inode) clear_file(inode, FADVISE_COLD_BIT) 734 735 #define file_wrong_pino(inode) is_file(inode, FADVISE_LOST_PINO_BIT) 736 #define file_lost_pino(inode) set_file(inode, FADVISE_LOST_PINO_BIT) 737 #define file_got_pino(inode) clear_file(inode, FADVISE_LOST_PINO_BIT) 738 739 #define file_is_encrypt(inode) is_file(inode, FADVISE_ENCRYPT_BIT) 740 #define file_set_encrypt(inode) set_file(inode, FADVISE_ENCRYPT_BIT) 741 742 #define file_enc_name(inode) is_file(inode, FADVISE_ENC_NAME_BIT) 743 #define file_set_enc_name(inode) set_file(inode, FADVISE_ENC_NAME_BIT) 744 745 #define file_keep_isize(inode) is_file(inode, FADVISE_KEEP_SIZE_BIT) 746 #define file_set_keep_isize(inode) set_file(inode, FADVISE_KEEP_SIZE_BIT) 747 748 #define file_is_hot(inode) is_file(inode, FADVISE_HOT_BIT) 749 #define file_set_hot(inode) set_file(inode, FADVISE_HOT_BIT) 750 #define file_clear_hot(inode) clear_file(inode, FADVISE_HOT_BIT) 751 752 #define file_is_verity(inode) is_file(inode, FADVISE_VERITY_BIT) 753 #define file_set_verity(inode) set_file(inode, FADVISE_VERITY_BIT) 754 755 #define file_should_truncate(inode) is_file(inode, FADVISE_TRUNC_BIT) 756 #define file_need_truncate(inode) set_file(inode, FADVISE_TRUNC_BIT) 757 #define file_dont_truncate(inode) clear_file(inode, FADVISE_TRUNC_BIT) 758 759 #define DEF_DIR_LEVEL 0 760 761 enum { 762 GC_FAILURE_PIN, 763 MAX_GC_FAILURE 764 }; 765 766 /* used for f2fs_inode_info->flags */ 767 enum { 768 FI_NEW_INODE, /* indicate newly allocated inode */ 769 FI_DIRTY_INODE, /* indicate inode is dirty or not */ 770 FI_AUTO_RECOVER, /* indicate inode is recoverable */ 771 FI_DIRTY_DIR, /* indicate directory has dirty pages */ 772 FI_INC_LINK, /* need to increment i_nlink */ 773 FI_ACL_MODE, /* indicate acl mode */ 774 FI_NO_ALLOC, /* should not allocate any blocks */ 775 FI_FREE_NID, /* free allocated nide */ 776 FI_NO_EXTENT, /* not to use the extent cache */ 777 FI_INLINE_XATTR, /* used for inline xattr */ 778 FI_INLINE_DATA, /* used for inline data*/ 779 FI_INLINE_DENTRY, /* used for inline dentry */ 780 FI_APPEND_WRITE, /* inode has appended data */ 781 FI_UPDATE_WRITE, /* inode has in-place-update data */ 782 FI_NEED_IPU, /* used for ipu per file */ 783 FI_ATOMIC_FILE, /* indicate atomic file */ 784 FI_DATA_EXIST, /* indicate data exists */ 785 FI_INLINE_DOTS, /* indicate inline dot dentries */ 786 FI_SKIP_WRITES, /* should skip data page writeback */ 787 FI_OPU_WRITE, /* used for opu per file */ 788 FI_DIRTY_FILE, /* indicate regular/symlink has dirty pages */ 789 FI_PREALLOCATED_ALL, /* all blocks for write were preallocated */ 790 FI_HOT_DATA, /* indicate file is hot */ 791 FI_EXTRA_ATTR, /* indicate file has extra attribute */ 792 FI_PROJ_INHERIT, /* indicate file inherits projectid */ 793 FI_PIN_FILE, /* indicate file should not be gced */ 794 FI_VERITY_IN_PROGRESS, /* building fs-verity Merkle tree */ 795 FI_COMPRESSED_FILE, /* indicate file's data can be compressed */ 796 FI_COMPRESS_CORRUPT, /* indicate compressed cluster is corrupted */ 797 FI_MMAP_FILE, /* indicate file was mmapped */ 798 FI_ENABLE_COMPRESS, /* enable compression in "user" compression mode */ 799 FI_COMPRESS_RELEASED, /* compressed blocks were released */ 800 FI_ALIGNED_WRITE, /* enable aligned write */ 801 FI_COW_FILE, /* indicate COW file */ 802 FI_ATOMIC_COMMITTED, /* indicate atomic commit completed except disk sync */ 803 FI_ATOMIC_REPLACE, /* indicate atomic replace */ 804 FI_MAX, /* max flag, never be used */ 805 }; 806 807 struct f2fs_inode_info { 808 struct inode vfs_inode; /* serve a vfs inode */ 809 unsigned long i_flags; /* keep an inode flags for ioctl */ 810 unsigned char i_advise; /* use to give file attribute hints */ 811 unsigned char i_dir_level; /* use for dentry level for large dir */ 812 unsigned int i_current_depth; /* only for directory depth */ 813 /* for gc failure statistic */ 814 unsigned int i_gc_failures[MAX_GC_FAILURE]; 815 unsigned int i_pino; /* parent inode number */ 816 umode_t i_acl_mode; /* keep file acl mode temporarily */ 817 818 /* Use below internally in f2fs*/ 819 unsigned long flags[BITS_TO_LONGS(FI_MAX)]; /* use to pass per-file flags */ 820 struct f2fs_rwsem i_sem; /* protect fi info */ 821 atomic_t dirty_pages; /* # of dirty pages */ 822 f2fs_hash_t chash; /* hash value of given file name */ 823 unsigned int clevel; /* maximum level of given file name */ 824 struct task_struct *task; /* lookup and create consistency */ 825 struct task_struct *cp_task; /* separate cp/wb IO stats*/ 826 struct task_struct *wb_task; /* indicate inode is in context of writeback */ 827 nid_t i_xattr_nid; /* node id that contains xattrs */ 828 loff_t last_disk_size; /* lastly written file size */ 829 spinlock_t i_size_lock; /* protect last_disk_size */ 830 831 #ifdef CONFIG_QUOTA 832 struct dquot *i_dquot[MAXQUOTAS]; 833 834 /* quota space reservation, managed internally by quota code */ 835 qsize_t i_reserved_quota; 836 #endif 837 struct list_head dirty_list; /* dirty list for dirs and files */ 838 struct list_head gdirty_list; /* linked in global dirty list */ 839 struct task_struct *atomic_write_task; /* store atomic write task */ 840 struct extent_tree *extent_tree[NR_EXTENT_CACHES]; 841 /* cached extent_tree entry */ 842 struct inode *cow_inode; /* copy-on-write inode for atomic write */ 843 844 /* avoid racing between foreground op and gc */ 845 struct f2fs_rwsem i_gc_rwsem[2]; 846 struct f2fs_rwsem i_xattr_sem; /* avoid racing between reading and changing EAs */ 847 848 int i_extra_isize; /* size of extra space located in i_addr */ 849 kprojid_t i_projid; /* id for project quota */ 850 int i_inline_xattr_size; /* inline xattr size */ 851 struct timespec64 i_crtime; /* inode creation time */ 852 struct timespec64 i_disk_time[3];/* inode disk times */ 853 854 /* for file compress */ 855 atomic_t i_compr_blocks; /* # of compressed blocks */ 856 unsigned char i_compress_algorithm; /* algorithm type */ 857 unsigned char i_log_cluster_size; /* log of cluster size */ 858 unsigned char i_compress_level; /* compress level (lz4hc,zstd) */ 859 unsigned char i_compress_flag; /* compress flag */ 860 unsigned int i_cluster_size; /* cluster size */ 861 862 unsigned int atomic_write_cnt; 863 loff_t original_i_size; /* original i_size before atomic write */ 864 }; 865 866 static inline void get_read_extent_info(struct extent_info *ext, 867 struct f2fs_extent *i_ext) 868 { 869 ext->fofs = le32_to_cpu(i_ext->fofs); 870 ext->blk = le32_to_cpu(i_ext->blk); 871 ext->len = le32_to_cpu(i_ext->len); 872 } 873 874 static inline void set_raw_read_extent(struct extent_info *ext, 875 struct f2fs_extent *i_ext) 876 { 877 i_ext->fofs = cpu_to_le32(ext->fofs); 878 i_ext->blk = cpu_to_le32(ext->blk); 879 i_ext->len = cpu_to_le32(ext->len); 880 } 881 882 static inline bool __is_discard_mergeable(struct discard_info *back, 883 struct discard_info *front, unsigned int max_len) 884 { 885 return (back->lstart + back->len == front->lstart) && 886 (back->len + front->len <= max_len); 887 } 888 889 static inline bool __is_discard_back_mergeable(struct discard_info *cur, 890 struct discard_info *back, unsigned int max_len) 891 { 892 return __is_discard_mergeable(back, cur, max_len); 893 } 894 895 static inline bool __is_discard_front_mergeable(struct discard_info *cur, 896 struct discard_info *front, unsigned int max_len) 897 { 898 return __is_discard_mergeable(cur, front, max_len); 899 } 900 901 /* 902 * For free nid management 903 */ 904 enum nid_state { 905 FREE_NID, /* newly added to free nid list */ 906 PREALLOC_NID, /* it is preallocated */ 907 MAX_NID_STATE, 908 }; 909 910 enum nat_state { 911 TOTAL_NAT, 912 DIRTY_NAT, 913 RECLAIMABLE_NAT, 914 MAX_NAT_STATE, 915 }; 916 917 struct f2fs_nm_info { 918 block_t nat_blkaddr; /* base disk address of NAT */ 919 nid_t max_nid; /* maximum possible node ids */ 920 nid_t available_nids; /* # of available node ids */ 921 nid_t next_scan_nid; /* the next nid to be scanned */ 922 nid_t max_rf_node_blocks; /* max # of nodes for recovery */ 923 unsigned int ram_thresh; /* control the memory footprint */ 924 unsigned int ra_nid_pages; /* # of nid pages to be readaheaded */ 925 unsigned int dirty_nats_ratio; /* control dirty nats ratio threshold */ 926 927 /* NAT cache management */ 928 struct radix_tree_root nat_root;/* root of the nat entry cache */ 929 struct radix_tree_root nat_set_root;/* root of the nat set cache */ 930 struct f2fs_rwsem nat_tree_lock; /* protect nat entry tree */ 931 struct list_head nat_entries; /* cached nat entry list (clean) */ 932 spinlock_t nat_list_lock; /* protect clean nat entry list */ 933 unsigned int nat_cnt[MAX_NAT_STATE]; /* the # of cached nat entries */ 934 unsigned int nat_blocks; /* # of nat blocks */ 935 936 /* free node ids management */ 937 struct radix_tree_root free_nid_root;/* root of the free_nid cache */ 938 struct list_head free_nid_list; /* list for free nids excluding preallocated nids */ 939 unsigned int nid_cnt[MAX_NID_STATE]; /* the number of free node id */ 940 spinlock_t nid_list_lock; /* protect nid lists ops */ 941 struct mutex build_lock; /* lock for build free nids */ 942 unsigned char **free_nid_bitmap; 943 unsigned char *nat_block_bitmap; 944 unsigned short *free_nid_count; /* free nid count of NAT block */ 945 946 /* for checkpoint */ 947 char *nat_bitmap; /* NAT bitmap pointer */ 948 949 unsigned int nat_bits_blocks; /* # of nat bits blocks */ 950 unsigned char *nat_bits; /* NAT bits blocks */ 951 unsigned char *full_nat_bits; /* full NAT pages */ 952 unsigned char *empty_nat_bits; /* empty NAT pages */ 953 #ifdef CONFIG_F2FS_CHECK_FS 954 char *nat_bitmap_mir; /* NAT bitmap mirror */ 955 #endif 956 int bitmap_size; /* bitmap size */ 957 }; 958 959 /* 960 * this structure is used as one of function parameters. 961 * all the information are dedicated to a given direct node block determined 962 * by the data offset in a file. 963 */ 964 struct dnode_of_data { 965 struct inode *inode; /* vfs inode pointer */ 966 struct page *inode_page; /* its inode page, NULL is possible */ 967 struct page *node_page; /* cached direct node page */ 968 nid_t nid; /* node id of the direct node block */ 969 unsigned int ofs_in_node; /* data offset in the node page */ 970 bool inode_page_locked; /* inode page is locked or not */ 971 bool node_changed; /* is node block changed */ 972 char cur_level; /* level of hole node page */ 973 char max_level; /* level of current page located */ 974 block_t data_blkaddr; /* block address of the node block */ 975 }; 976 977 static inline void set_new_dnode(struct dnode_of_data *dn, struct inode *inode, 978 struct page *ipage, struct page *npage, nid_t nid) 979 { 980 memset(dn, 0, sizeof(*dn)); 981 dn->inode = inode; 982 dn->inode_page = ipage; 983 dn->node_page = npage; 984 dn->nid = nid; 985 } 986 987 /* 988 * For SIT manager 989 * 990 * By default, there are 6 active log areas across the whole main area. 991 * When considering hot and cold data separation to reduce cleaning overhead, 992 * we split 3 for data logs and 3 for node logs as hot, warm, and cold types, 993 * respectively. 994 * In the current design, you should not change the numbers intentionally. 995 * Instead, as a mount option such as active_logs=x, you can use 2, 4, and 6 996 * logs individually according to the underlying devices. (default: 6) 997 * Just in case, on-disk layout covers maximum 16 logs that consist of 8 for 998 * data and 8 for node logs. 999 */ 1000 #define NR_CURSEG_DATA_TYPE (3) 1001 #define NR_CURSEG_NODE_TYPE (3) 1002 #define NR_CURSEG_INMEM_TYPE (2) 1003 #define NR_CURSEG_RO_TYPE (2) 1004 #define NR_CURSEG_PERSIST_TYPE (NR_CURSEG_DATA_TYPE + NR_CURSEG_NODE_TYPE) 1005 #define NR_CURSEG_TYPE (NR_CURSEG_INMEM_TYPE + NR_CURSEG_PERSIST_TYPE) 1006 1007 enum { 1008 CURSEG_HOT_DATA = 0, /* directory entry blocks */ 1009 CURSEG_WARM_DATA, /* data blocks */ 1010 CURSEG_COLD_DATA, /* multimedia or GCed data blocks */ 1011 CURSEG_HOT_NODE, /* direct node blocks of directory files */ 1012 CURSEG_WARM_NODE, /* direct node blocks of normal files */ 1013 CURSEG_COLD_NODE, /* indirect node blocks */ 1014 NR_PERSISTENT_LOG, /* number of persistent log */ 1015 CURSEG_COLD_DATA_PINNED = NR_PERSISTENT_LOG, 1016 /* pinned file that needs consecutive block address */ 1017 CURSEG_ALL_DATA_ATGC, /* SSR alloctor in hot/warm/cold data area */ 1018 NO_CHECK_TYPE, /* number of persistent & inmem log */ 1019 }; 1020 1021 struct flush_cmd { 1022 struct completion wait; 1023 struct llist_node llnode; 1024 nid_t ino; 1025 int ret; 1026 }; 1027 1028 struct flush_cmd_control { 1029 struct task_struct *f2fs_issue_flush; /* flush thread */ 1030 wait_queue_head_t flush_wait_queue; /* waiting queue for wake-up */ 1031 atomic_t issued_flush; /* # of issued flushes */ 1032 atomic_t queued_flush; /* # of queued flushes */ 1033 struct llist_head issue_list; /* list for command issue */ 1034 struct llist_node *dispatch_list; /* list for command dispatch */ 1035 }; 1036 1037 struct f2fs_sm_info { 1038 struct sit_info *sit_info; /* whole segment information */ 1039 struct free_segmap_info *free_info; /* free segment information */ 1040 struct dirty_seglist_info *dirty_info; /* dirty segment information */ 1041 struct curseg_info *curseg_array; /* active segment information */ 1042 1043 struct f2fs_rwsem curseg_lock; /* for preventing curseg change */ 1044 1045 block_t seg0_blkaddr; /* block address of 0'th segment */ 1046 block_t main_blkaddr; /* start block address of main area */ 1047 block_t ssa_blkaddr; /* start block address of SSA area */ 1048 1049 unsigned int segment_count; /* total # of segments */ 1050 unsigned int main_segments; /* # of segments in main area */ 1051 unsigned int reserved_segments; /* # of reserved segments */ 1052 unsigned int additional_reserved_segments;/* reserved segs for IO align feature */ 1053 unsigned int ovp_segments; /* # of overprovision segments */ 1054 1055 /* a threshold to reclaim prefree segments */ 1056 unsigned int rec_prefree_segments; 1057 1058 struct list_head sit_entry_set; /* sit entry set list */ 1059 1060 unsigned int ipu_policy; /* in-place-update policy */ 1061 unsigned int min_ipu_util; /* in-place-update threshold */ 1062 unsigned int min_fsync_blocks; /* threshold for fsync */ 1063 unsigned int min_seq_blocks; /* threshold for sequential blocks */ 1064 unsigned int min_hot_blocks; /* threshold for hot block allocation */ 1065 unsigned int min_ssr_sections; /* threshold to trigger SSR allocation */ 1066 1067 /* for flush command control */ 1068 struct flush_cmd_control *fcc_info; 1069 1070 /* for discard command control */ 1071 struct discard_cmd_control *dcc_info; 1072 }; 1073 1074 /* 1075 * For superblock 1076 */ 1077 /* 1078 * COUNT_TYPE for monitoring 1079 * 1080 * f2fs monitors the number of several block types such as on-writeback, 1081 * dirty dentry blocks, dirty node blocks, and dirty meta blocks. 1082 */ 1083 #define WB_DATA_TYPE(p) (__is_cp_guaranteed(p) ? F2FS_WB_CP_DATA : F2FS_WB_DATA) 1084 enum count_type { 1085 F2FS_DIRTY_DENTS, 1086 F2FS_DIRTY_DATA, 1087 F2FS_DIRTY_QDATA, 1088 F2FS_DIRTY_NODES, 1089 F2FS_DIRTY_META, 1090 F2FS_DIRTY_IMETA, 1091 F2FS_WB_CP_DATA, 1092 F2FS_WB_DATA, 1093 F2FS_RD_DATA, 1094 F2FS_RD_NODE, 1095 F2FS_RD_META, 1096 F2FS_DIO_WRITE, 1097 F2FS_DIO_READ, 1098 NR_COUNT_TYPE, 1099 }; 1100 1101 /* 1102 * The below are the page types of bios used in submit_bio(). 1103 * The available types are: 1104 * DATA User data pages. It operates as async mode. 1105 * NODE Node pages. It operates as async mode. 1106 * META FS metadata pages such as SIT, NAT, CP. 1107 * NR_PAGE_TYPE The number of page types. 1108 * META_FLUSH Make sure the previous pages are written 1109 * with waiting the bio's completion 1110 * ... Only can be used with META. 1111 */ 1112 #define PAGE_TYPE_OF_BIO(type) ((type) > META ? META : (type)) 1113 enum page_type { 1114 DATA = 0, 1115 NODE = 1, /* should not change this */ 1116 META, 1117 NR_PAGE_TYPE, 1118 META_FLUSH, 1119 IPU, /* the below types are used by tracepoints only. */ 1120 OPU, 1121 }; 1122 1123 enum temp_type { 1124 HOT = 0, /* must be zero for meta bio */ 1125 WARM, 1126 COLD, 1127 NR_TEMP_TYPE, 1128 }; 1129 1130 enum need_lock_type { 1131 LOCK_REQ = 0, 1132 LOCK_DONE, 1133 LOCK_RETRY, 1134 }; 1135 1136 enum cp_reason_type { 1137 CP_NO_NEEDED, 1138 CP_NON_REGULAR, 1139 CP_COMPRESSED, 1140 CP_HARDLINK, 1141 CP_SB_NEED_CP, 1142 CP_WRONG_PINO, 1143 CP_NO_SPC_ROLL, 1144 CP_NODE_NEED_CP, 1145 CP_FASTBOOT_MODE, 1146 CP_SPEC_LOG_NUM, 1147 CP_RECOVER_DIR, 1148 }; 1149 1150 enum iostat_type { 1151 /* WRITE IO */ 1152 APP_DIRECT_IO, /* app direct write IOs */ 1153 APP_BUFFERED_IO, /* app buffered write IOs */ 1154 APP_WRITE_IO, /* app write IOs */ 1155 APP_MAPPED_IO, /* app mapped IOs */ 1156 APP_BUFFERED_CDATA_IO, /* app buffered write IOs on compressed file */ 1157 APP_MAPPED_CDATA_IO, /* app mapped write IOs on compressed file */ 1158 FS_DATA_IO, /* data IOs from kworker/fsync/reclaimer */ 1159 FS_CDATA_IO, /* data IOs from kworker/fsync/reclaimer on compressed file */ 1160 FS_NODE_IO, /* node IOs from kworker/fsync/reclaimer */ 1161 FS_META_IO, /* meta IOs from kworker/reclaimer */ 1162 FS_GC_DATA_IO, /* data IOs from forground gc */ 1163 FS_GC_NODE_IO, /* node IOs from forground gc */ 1164 FS_CP_DATA_IO, /* data IOs from checkpoint */ 1165 FS_CP_NODE_IO, /* node IOs from checkpoint */ 1166 FS_CP_META_IO, /* meta IOs from checkpoint */ 1167 1168 /* READ IO */ 1169 APP_DIRECT_READ_IO, /* app direct read IOs */ 1170 APP_BUFFERED_READ_IO, /* app buffered read IOs */ 1171 APP_READ_IO, /* app read IOs */ 1172 APP_MAPPED_READ_IO, /* app mapped read IOs */ 1173 APP_BUFFERED_CDATA_READ_IO, /* app buffered read IOs on compressed file */ 1174 APP_MAPPED_CDATA_READ_IO, /* app mapped read IOs on compressed file */ 1175 FS_DATA_READ_IO, /* data read IOs */ 1176 FS_GDATA_READ_IO, /* data read IOs from background gc */ 1177 FS_CDATA_READ_IO, /* compressed data read IOs */ 1178 FS_NODE_READ_IO, /* node read IOs */ 1179 FS_META_READ_IO, /* meta read IOs */ 1180 1181 /* other */ 1182 FS_DISCARD_IO, /* discard */ 1183 FS_FLUSH_IO, /* flush */ 1184 FS_ZONE_RESET_IO, /* zone reset */ 1185 NR_IO_TYPE, 1186 }; 1187 1188 struct f2fs_io_info { 1189 struct f2fs_sb_info *sbi; /* f2fs_sb_info pointer */ 1190 nid_t ino; /* inode number */ 1191 enum page_type type; /* contains DATA/NODE/META/META_FLUSH */ 1192 enum temp_type temp; /* contains HOT/WARM/COLD */ 1193 enum req_op op; /* contains REQ_OP_ */ 1194 blk_opf_t op_flags; /* req_flag_bits */ 1195 block_t new_blkaddr; /* new block address to be written */ 1196 block_t old_blkaddr; /* old block address before Cow */ 1197 struct page *page; /* page to be written */ 1198 struct page *encrypted_page; /* encrypted page */ 1199 struct page *compressed_page; /* compressed page */ 1200 struct list_head list; /* serialize IOs */ 1201 unsigned int compr_blocks; /* # of compressed block addresses */ 1202 unsigned int need_lock:8; /* indicate we need to lock cp_rwsem */ 1203 unsigned int version:8; /* version of the node */ 1204 unsigned int submitted:1; /* indicate IO submission */ 1205 unsigned int in_list:1; /* indicate fio is in io_list */ 1206 unsigned int is_por:1; /* indicate IO is from recovery or not */ 1207 unsigned int retry:1; /* need to reallocate block address */ 1208 unsigned int encrypted:1; /* indicate file is encrypted */ 1209 unsigned int post_read:1; /* require post read */ 1210 enum iostat_type io_type; /* io type */ 1211 struct writeback_control *io_wbc; /* writeback control */ 1212 struct bio **bio; /* bio for ipu */ 1213 sector_t *last_block; /* last block number in bio */ 1214 }; 1215 1216 struct bio_entry { 1217 struct bio *bio; 1218 struct list_head list; 1219 }; 1220 1221 #define is_read_io(rw) ((rw) == READ) 1222 struct f2fs_bio_info { 1223 struct f2fs_sb_info *sbi; /* f2fs superblock */ 1224 struct bio *bio; /* bios to merge */ 1225 sector_t last_block_in_bio; /* last block number */ 1226 struct f2fs_io_info fio; /* store buffered io info. */ 1227 #ifdef CONFIG_BLK_DEV_ZONED 1228 struct completion zone_wait; /* condition value for the previous open zone to close */ 1229 struct bio *zone_pending_bio; /* pending bio for the previous zone */ 1230 void *bi_private; /* previous bi_private for pending bio */ 1231 #endif 1232 struct f2fs_rwsem io_rwsem; /* blocking op for bio */ 1233 spinlock_t io_lock; /* serialize DATA/NODE IOs */ 1234 struct list_head io_list; /* track fios */ 1235 struct list_head bio_list; /* bio entry list head */ 1236 struct f2fs_rwsem bio_list_lock; /* lock to protect bio entry list */ 1237 }; 1238 1239 #define FDEV(i) (sbi->devs[i]) 1240 #define RDEV(i) (raw_super->devs[i]) 1241 struct f2fs_dev_info { 1242 struct bdev_handle *bdev_handle; 1243 struct block_device *bdev; 1244 char path[MAX_PATH_LEN]; 1245 unsigned int total_segments; 1246 block_t start_blk; 1247 block_t end_blk; 1248 #ifdef CONFIG_BLK_DEV_ZONED 1249 unsigned int nr_blkz; /* Total number of zones */ 1250 unsigned long *blkz_seq; /* Bitmap indicating sequential zones */ 1251 #endif 1252 }; 1253 1254 enum inode_type { 1255 DIR_INODE, /* for dirty dir inode */ 1256 FILE_INODE, /* for dirty regular/symlink inode */ 1257 DIRTY_META, /* for all dirtied inode metadata */ 1258 NR_INODE_TYPE, 1259 }; 1260 1261 /* for inner inode cache management */ 1262 struct inode_management { 1263 struct radix_tree_root ino_root; /* ino entry array */ 1264 spinlock_t ino_lock; /* for ino entry lock */ 1265 struct list_head ino_list; /* inode list head */ 1266 unsigned long ino_num; /* number of entries */ 1267 }; 1268 1269 /* for GC_AT */ 1270 struct atgc_management { 1271 bool atgc_enabled; /* ATGC is enabled or not */ 1272 struct rb_root_cached root; /* root of victim rb-tree */ 1273 struct list_head victim_list; /* linked with all victim entries */ 1274 unsigned int victim_count; /* victim count in rb-tree */ 1275 unsigned int candidate_ratio; /* candidate ratio */ 1276 unsigned int max_candidate_count; /* max candidate count */ 1277 unsigned int age_weight; /* age weight, vblock_weight = 100 - age_weight */ 1278 unsigned long long age_threshold; /* age threshold */ 1279 }; 1280 1281 struct f2fs_gc_control { 1282 unsigned int victim_segno; /* target victim segment number */ 1283 int init_gc_type; /* FG_GC or BG_GC */ 1284 bool no_bg_gc; /* check the space and stop bg_gc */ 1285 bool should_migrate_blocks; /* should migrate blocks */ 1286 bool err_gc_skipped; /* return EAGAIN if GC skipped */ 1287 unsigned int nr_free_secs; /* # of free sections to do GC */ 1288 }; 1289 1290 /* 1291 * For s_flag in struct f2fs_sb_info 1292 * Modification on enum should be synchronized with s_flag array 1293 */ 1294 enum { 1295 SBI_IS_DIRTY, /* dirty flag for checkpoint */ 1296 SBI_IS_CLOSE, /* specify unmounting */ 1297 SBI_NEED_FSCK, /* need fsck.f2fs to fix */ 1298 SBI_POR_DOING, /* recovery is doing or not */ 1299 SBI_NEED_SB_WRITE, /* need to recover superblock */ 1300 SBI_NEED_CP, /* need to checkpoint */ 1301 SBI_IS_SHUTDOWN, /* shutdown by ioctl */ 1302 SBI_IS_RECOVERED, /* recovered orphan/data */ 1303 SBI_CP_DISABLED, /* CP was disabled last mount */ 1304 SBI_CP_DISABLED_QUICK, /* CP was disabled quickly */ 1305 SBI_QUOTA_NEED_FLUSH, /* need to flush quota info in CP */ 1306 SBI_QUOTA_SKIP_FLUSH, /* skip flushing quota in current CP */ 1307 SBI_QUOTA_NEED_REPAIR, /* quota file may be corrupted */ 1308 SBI_IS_RESIZEFS, /* resizefs is in process */ 1309 SBI_IS_FREEZING, /* freezefs is in process */ 1310 SBI_IS_WRITABLE, /* remove ro mountoption transiently */ 1311 MAX_SBI_FLAG, 1312 }; 1313 1314 enum { 1315 CP_TIME, 1316 REQ_TIME, 1317 DISCARD_TIME, 1318 GC_TIME, 1319 DISABLE_TIME, 1320 UMOUNT_DISCARD_TIMEOUT, 1321 MAX_TIME, 1322 }; 1323 1324 /* Note that you need to keep synchronization with this gc_mode_names array */ 1325 enum { 1326 GC_NORMAL, 1327 GC_IDLE_CB, 1328 GC_IDLE_GREEDY, 1329 GC_IDLE_AT, 1330 GC_URGENT_HIGH, 1331 GC_URGENT_LOW, 1332 GC_URGENT_MID, 1333 MAX_GC_MODE, 1334 }; 1335 1336 enum { 1337 BGGC_MODE_ON, /* background gc is on */ 1338 BGGC_MODE_OFF, /* background gc is off */ 1339 BGGC_MODE_SYNC, /* 1340 * background gc is on, migrating blocks 1341 * like foreground gc 1342 */ 1343 }; 1344 1345 enum { 1346 FS_MODE_ADAPTIVE, /* use both lfs/ssr allocation */ 1347 FS_MODE_LFS, /* use lfs allocation only */ 1348 FS_MODE_FRAGMENT_SEG, /* segment fragmentation mode */ 1349 FS_MODE_FRAGMENT_BLK, /* block fragmentation mode */ 1350 }; 1351 1352 enum { 1353 ALLOC_MODE_DEFAULT, /* stay default */ 1354 ALLOC_MODE_REUSE, /* reuse segments as much as possible */ 1355 }; 1356 1357 enum fsync_mode { 1358 FSYNC_MODE_POSIX, /* fsync follows posix semantics */ 1359 FSYNC_MODE_STRICT, /* fsync behaves in line with ext4 */ 1360 FSYNC_MODE_NOBARRIER, /* fsync behaves nobarrier based on posix */ 1361 }; 1362 1363 enum { 1364 COMPR_MODE_FS, /* 1365 * automatically compress compression 1366 * enabled files 1367 */ 1368 COMPR_MODE_USER, /* 1369 * automatical compression is disabled. 1370 * user can control the file compression 1371 * using ioctls 1372 */ 1373 }; 1374 1375 enum { 1376 DISCARD_UNIT_BLOCK, /* basic discard unit is block */ 1377 DISCARD_UNIT_SEGMENT, /* basic discard unit is segment */ 1378 DISCARD_UNIT_SECTION, /* basic discard unit is section */ 1379 }; 1380 1381 enum { 1382 MEMORY_MODE_NORMAL, /* memory mode for normal devices */ 1383 MEMORY_MODE_LOW, /* memory mode for low memry devices */ 1384 }; 1385 1386 enum errors_option { 1387 MOUNT_ERRORS_READONLY, /* remount fs ro on errors */ 1388 MOUNT_ERRORS_CONTINUE, /* continue on errors */ 1389 MOUNT_ERRORS_PANIC, /* panic on errors */ 1390 }; 1391 1392 enum { 1393 BACKGROUND, 1394 FOREGROUND, 1395 MAX_CALL_TYPE, 1396 TOTAL_CALL = FOREGROUND, 1397 }; 1398 1399 static inline int f2fs_test_bit(unsigned int nr, char *addr); 1400 static inline void f2fs_set_bit(unsigned int nr, char *addr); 1401 static inline void f2fs_clear_bit(unsigned int nr, char *addr); 1402 1403 /* 1404 * Layout of f2fs page.private: 1405 * 1406 * Layout A: lowest bit should be 1 1407 * | bit0 = 1 | bit1 | bit2 | ... | bit MAX | private data .... | 1408 * bit 0 PAGE_PRIVATE_NOT_POINTER 1409 * bit 1 PAGE_PRIVATE_DUMMY_WRITE 1410 * bit 2 PAGE_PRIVATE_ONGOING_MIGRATION 1411 * bit 3 PAGE_PRIVATE_INLINE_INODE 1412 * bit 4 PAGE_PRIVATE_REF_RESOURCE 1413 * bit 5- f2fs private data 1414 * 1415 * Layout B: lowest bit should be 0 1416 * page.private is a wrapped pointer. 1417 */ 1418 enum { 1419 PAGE_PRIVATE_NOT_POINTER, /* private contains non-pointer data */ 1420 PAGE_PRIVATE_DUMMY_WRITE, /* data page for padding aligned IO */ 1421 PAGE_PRIVATE_ONGOING_MIGRATION, /* data page which is on-going migrating */ 1422 PAGE_PRIVATE_INLINE_INODE, /* inode page contains inline data */ 1423 PAGE_PRIVATE_REF_RESOURCE, /* dirty page has referenced resources */ 1424 PAGE_PRIVATE_MAX 1425 }; 1426 1427 /* For compression */ 1428 enum compress_algorithm_type { 1429 COMPRESS_LZO, 1430 COMPRESS_LZ4, 1431 COMPRESS_ZSTD, 1432 COMPRESS_LZORLE, 1433 COMPRESS_MAX, 1434 }; 1435 1436 enum compress_flag { 1437 COMPRESS_CHKSUM, 1438 COMPRESS_MAX_FLAG, 1439 }; 1440 1441 #define COMPRESS_WATERMARK 20 1442 #define COMPRESS_PERCENT 20 1443 1444 #define COMPRESS_DATA_RESERVED_SIZE 4 1445 struct compress_data { 1446 __le32 clen; /* compressed data size */ 1447 __le32 chksum; /* compressed data chksum */ 1448 __le32 reserved[COMPRESS_DATA_RESERVED_SIZE]; /* reserved */ 1449 u8 cdata[]; /* compressed data */ 1450 }; 1451 1452 #define COMPRESS_HEADER_SIZE (sizeof(struct compress_data)) 1453 1454 #define F2FS_COMPRESSED_PAGE_MAGIC 0xF5F2C000 1455 1456 #define F2FS_ZSTD_DEFAULT_CLEVEL 1 1457 1458 #define COMPRESS_LEVEL_OFFSET 8 1459 1460 /* compress context */ 1461 struct compress_ctx { 1462 struct inode *inode; /* inode the context belong to */ 1463 pgoff_t cluster_idx; /* cluster index number */ 1464 unsigned int cluster_size; /* page count in cluster */ 1465 unsigned int log_cluster_size; /* log of cluster size */ 1466 struct page **rpages; /* pages store raw data in cluster */ 1467 unsigned int nr_rpages; /* total page number in rpages */ 1468 struct page **cpages; /* pages store compressed data in cluster */ 1469 unsigned int nr_cpages; /* total page number in cpages */ 1470 unsigned int valid_nr_cpages; /* valid page number in cpages */ 1471 void *rbuf; /* virtual mapped address on rpages */ 1472 struct compress_data *cbuf; /* virtual mapped address on cpages */ 1473 size_t rlen; /* valid data length in rbuf */ 1474 size_t clen; /* valid data length in cbuf */ 1475 void *private; /* payload buffer for specified compression algorithm */ 1476 void *private2; /* extra payload buffer */ 1477 }; 1478 1479 /* compress context for write IO path */ 1480 struct compress_io_ctx { 1481 u32 magic; /* magic number to indicate page is compressed */ 1482 struct inode *inode; /* inode the context belong to */ 1483 struct page **rpages; /* pages store raw data in cluster */ 1484 unsigned int nr_rpages; /* total page number in rpages */ 1485 atomic_t pending_pages; /* in-flight compressed page count */ 1486 }; 1487 1488 /* Context for decompressing one cluster on the read IO path */ 1489 struct decompress_io_ctx { 1490 u32 magic; /* magic number to indicate page is compressed */ 1491 struct inode *inode; /* inode the context belong to */ 1492 pgoff_t cluster_idx; /* cluster index number */ 1493 unsigned int cluster_size; /* page count in cluster */ 1494 unsigned int log_cluster_size; /* log of cluster size */ 1495 struct page **rpages; /* pages store raw data in cluster */ 1496 unsigned int nr_rpages; /* total page number in rpages */ 1497 struct page **cpages; /* pages store compressed data in cluster */ 1498 unsigned int nr_cpages; /* total page number in cpages */ 1499 struct page **tpages; /* temp pages to pad holes in cluster */ 1500 void *rbuf; /* virtual mapped address on rpages */ 1501 struct compress_data *cbuf; /* virtual mapped address on cpages */ 1502 size_t rlen; /* valid data length in rbuf */ 1503 size_t clen; /* valid data length in cbuf */ 1504 1505 /* 1506 * The number of compressed pages remaining to be read in this cluster. 1507 * This is initially nr_cpages. It is decremented by 1 each time a page 1508 * has been read (or failed to be read). When it reaches 0, the cluster 1509 * is decompressed (or an error is reported). 1510 * 1511 * If an error occurs before all the pages have been submitted for I/O, 1512 * then this will never reach 0. In this case the I/O submitter is 1513 * responsible for calling f2fs_decompress_end_io() instead. 1514 */ 1515 atomic_t remaining_pages; 1516 1517 /* 1518 * Number of references to this decompress_io_ctx. 1519 * 1520 * One reference is held for I/O completion. This reference is dropped 1521 * after the pagecache pages are updated and unlocked -- either after 1522 * decompression (and verity if enabled), or after an error. 1523 * 1524 * In addition, each compressed page holds a reference while it is in a 1525 * bio. These references are necessary prevent compressed pages from 1526 * being freed while they are still in a bio. 1527 */ 1528 refcount_t refcnt; 1529 1530 bool failed; /* IO error occurred before decompression? */ 1531 bool need_verity; /* need fs-verity verification after decompression? */ 1532 void *private; /* payload buffer for specified decompression algorithm */ 1533 void *private2; /* extra payload buffer */ 1534 struct work_struct verity_work; /* work to verify the decompressed pages */ 1535 struct work_struct free_work; /* work for late free this structure itself */ 1536 }; 1537 1538 #define NULL_CLUSTER ((unsigned int)(~0)) 1539 #define MIN_COMPRESS_LOG_SIZE 2 1540 #define MAX_COMPRESS_LOG_SIZE 8 1541 #define MAX_COMPRESS_WINDOW_SIZE(log_size) ((PAGE_SIZE) << (log_size)) 1542 1543 struct f2fs_sb_info { 1544 struct super_block *sb; /* pointer to VFS super block */ 1545 struct proc_dir_entry *s_proc; /* proc entry */ 1546 struct f2fs_super_block *raw_super; /* raw super block pointer */ 1547 struct f2fs_rwsem sb_lock; /* lock for raw super block */ 1548 int valid_super_block; /* valid super block no */ 1549 unsigned long s_flag; /* flags for sbi */ 1550 struct mutex writepages; /* mutex for writepages() */ 1551 1552 #ifdef CONFIG_BLK_DEV_ZONED 1553 unsigned int blocks_per_blkz; /* F2FS blocks per zone */ 1554 #endif 1555 1556 /* for node-related operations */ 1557 struct f2fs_nm_info *nm_info; /* node manager */ 1558 struct inode *node_inode; /* cache node blocks */ 1559 1560 /* for segment-related operations */ 1561 struct f2fs_sm_info *sm_info; /* segment manager */ 1562 1563 /* for bio operations */ 1564 struct f2fs_bio_info *write_io[NR_PAGE_TYPE]; /* for write bios */ 1565 /* keep migration IO order for LFS mode */ 1566 struct f2fs_rwsem io_order_lock; 1567 mempool_t *write_io_dummy; /* Dummy pages */ 1568 pgoff_t page_eio_ofs[NR_PAGE_TYPE]; /* EIO page offset */ 1569 int page_eio_cnt[NR_PAGE_TYPE]; /* EIO count */ 1570 1571 /* for checkpoint */ 1572 struct f2fs_checkpoint *ckpt; /* raw checkpoint pointer */ 1573 int cur_cp_pack; /* remain current cp pack */ 1574 spinlock_t cp_lock; /* for flag in ckpt */ 1575 struct inode *meta_inode; /* cache meta blocks */ 1576 struct f2fs_rwsem cp_global_sem; /* checkpoint procedure lock */ 1577 struct f2fs_rwsem cp_rwsem; /* blocking FS operations */ 1578 struct f2fs_rwsem node_write; /* locking node writes */ 1579 struct f2fs_rwsem node_change; /* locking node change */ 1580 wait_queue_head_t cp_wait; 1581 unsigned long last_time[MAX_TIME]; /* to store time in jiffies */ 1582 long interval_time[MAX_TIME]; /* to store thresholds */ 1583 struct ckpt_req_control cprc_info; /* for checkpoint request control */ 1584 1585 struct inode_management im[MAX_INO_ENTRY]; /* manage inode cache */ 1586 1587 spinlock_t fsync_node_lock; /* for node entry lock */ 1588 struct list_head fsync_node_list; /* node list head */ 1589 unsigned int fsync_seg_id; /* sequence id */ 1590 unsigned int fsync_node_num; /* number of node entries */ 1591 1592 /* for orphan inode, use 0'th array */ 1593 unsigned int max_orphans; /* max orphan inodes */ 1594 1595 /* for inode management */ 1596 struct list_head inode_list[NR_INODE_TYPE]; /* dirty inode list */ 1597 spinlock_t inode_lock[NR_INODE_TYPE]; /* for dirty inode list lock */ 1598 struct mutex flush_lock; /* for flush exclusion */ 1599 1600 /* for extent tree cache */ 1601 struct extent_tree_info extent_tree[NR_EXTENT_CACHES]; 1602 atomic64_t allocated_data_blocks; /* for block age extent_cache */ 1603 1604 /* The threshold used for hot and warm data seperation*/ 1605 unsigned int hot_data_age_threshold; 1606 unsigned int warm_data_age_threshold; 1607 unsigned int last_age_weight; 1608 1609 /* basic filesystem units */ 1610 unsigned int log_sectors_per_block; /* log2 sectors per block */ 1611 unsigned int log_blocksize; /* log2 block size */ 1612 unsigned int blocksize; /* block size */ 1613 unsigned int root_ino_num; /* root inode number*/ 1614 unsigned int node_ino_num; /* node inode number*/ 1615 unsigned int meta_ino_num; /* meta inode number*/ 1616 unsigned int log_blocks_per_seg; /* log2 blocks per segment */ 1617 unsigned int blocks_per_seg; /* blocks per segment */ 1618 unsigned int unusable_blocks_per_sec; /* unusable blocks per section */ 1619 unsigned int segs_per_sec; /* segments per section */ 1620 unsigned int secs_per_zone; /* sections per zone */ 1621 unsigned int total_sections; /* total section count */ 1622 unsigned int total_node_count; /* total node block count */ 1623 unsigned int total_valid_node_count; /* valid node block count */ 1624 int dir_level; /* directory level */ 1625 bool readdir_ra; /* readahead inode in readdir */ 1626 u64 max_io_bytes; /* max io bytes to merge IOs */ 1627 1628 block_t user_block_count; /* # of user blocks */ 1629 block_t total_valid_block_count; /* # of valid blocks */ 1630 block_t discard_blks; /* discard command candidats */ 1631 block_t last_valid_block_count; /* for recovery */ 1632 block_t reserved_blocks; /* configurable reserved blocks */ 1633 block_t current_reserved_blocks; /* current reserved blocks */ 1634 1635 /* Additional tracking for no checkpoint mode */ 1636 block_t unusable_block_count; /* # of blocks saved by last cp */ 1637 1638 unsigned int nquota_files; /* # of quota sysfile */ 1639 struct f2fs_rwsem quota_sem; /* blocking cp for flags */ 1640 1641 /* # of pages, see count_type */ 1642 atomic_t nr_pages[NR_COUNT_TYPE]; 1643 /* # of allocated blocks */ 1644 struct percpu_counter alloc_valid_block_count; 1645 /* # of node block writes as roll forward recovery */ 1646 struct percpu_counter rf_node_block_count; 1647 1648 /* writeback control */ 1649 atomic_t wb_sync_req[META]; /* count # of WB_SYNC threads */ 1650 1651 /* valid inode count */ 1652 struct percpu_counter total_valid_inode_count; 1653 1654 struct f2fs_mount_info mount_opt; /* mount options */ 1655 1656 /* for cleaning operations */ 1657 struct f2fs_rwsem gc_lock; /* 1658 * semaphore for GC, avoid 1659 * race between GC and GC or CP 1660 */ 1661 struct f2fs_gc_kthread *gc_thread; /* GC thread */ 1662 struct atgc_management am; /* atgc management */ 1663 unsigned int cur_victim_sec; /* current victim section num */ 1664 unsigned int gc_mode; /* current GC state */ 1665 unsigned int next_victim_seg[2]; /* next segment in victim section */ 1666 spinlock_t gc_remaining_trials_lock; 1667 /* remaining trial count for GC_URGENT_* and GC_IDLE_* */ 1668 unsigned int gc_remaining_trials; 1669 1670 /* for skip statistic */ 1671 unsigned long long skipped_gc_rwsem; /* FG_GC only */ 1672 1673 /* threshold for gc trials on pinned files */ 1674 u64 gc_pin_file_threshold; 1675 struct f2fs_rwsem pin_sem; 1676 1677 /* maximum # of trials to find a victim segment for SSR and GC */ 1678 unsigned int max_victim_search; 1679 /* migration granularity of garbage collection, unit: segment */ 1680 unsigned int migration_granularity; 1681 1682 /* 1683 * for stat information. 1684 * one is for the LFS mode, and the other is for the SSR mode. 1685 */ 1686 #ifdef CONFIG_F2FS_STAT_FS 1687 struct f2fs_stat_info *stat_info; /* FS status information */ 1688 atomic_t meta_count[META_MAX]; /* # of meta blocks */ 1689 unsigned int segment_count[2]; /* # of allocated segments */ 1690 unsigned int block_count[2]; /* # of allocated blocks */ 1691 atomic_t inplace_count; /* # of inplace update */ 1692 /* # of lookup extent cache */ 1693 atomic64_t total_hit_ext[NR_EXTENT_CACHES]; 1694 /* # of hit rbtree extent node */ 1695 atomic64_t read_hit_rbtree[NR_EXTENT_CACHES]; 1696 /* # of hit cached extent node */ 1697 atomic64_t read_hit_cached[NR_EXTENT_CACHES]; 1698 /* # of hit largest extent node in read extent cache */ 1699 atomic64_t read_hit_largest; 1700 atomic_t inline_xattr; /* # of inline_xattr inodes */ 1701 atomic_t inline_inode; /* # of inline_data inodes */ 1702 atomic_t inline_dir; /* # of inline_dentry inodes */ 1703 atomic_t compr_inode; /* # of compressed inodes */ 1704 atomic64_t compr_blocks; /* # of compressed blocks */ 1705 atomic_t swapfile_inode; /* # of swapfile inodes */ 1706 atomic_t atomic_files; /* # of opened atomic file */ 1707 atomic_t max_aw_cnt; /* max # of atomic writes */ 1708 unsigned int io_skip_bggc; /* skip background gc for in-flight IO */ 1709 unsigned int other_skip_bggc; /* skip background gc for other reasons */ 1710 unsigned int ndirty_inode[NR_INODE_TYPE]; /* # of dirty inodes */ 1711 atomic_t cp_call_count[MAX_CALL_TYPE]; /* # of cp call */ 1712 #endif 1713 spinlock_t stat_lock; /* lock for stat operations */ 1714 1715 /* to attach REQ_META|REQ_FUA flags */ 1716 unsigned int data_io_flag; 1717 unsigned int node_io_flag; 1718 1719 /* For sysfs support */ 1720 struct kobject s_kobj; /* /sys/fs/f2fs/<devname> */ 1721 struct completion s_kobj_unregister; 1722 1723 struct kobject s_stat_kobj; /* /sys/fs/f2fs/<devname>/stat */ 1724 struct completion s_stat_kobj_unregister; 1725 1726 struct kobject s_feature_list_kobj; /* /sys/fs/f2fs/<devname>/feature_list */ 1727 struct completion s_feature_list_kobj_unregister; 1728 1729 /* For shrinker support */ 1730 struct list_head s_list; 1731 struct mutex umount_mutex; 1732 unsigned int shrinker_run_no; 1733 1734 /* For multi devices */ 1735 int s_ndevs; /* number of devices */ 1736 struct f2fs_dev_info *devs; /* for device list */ 1737 unsigned int dirty_device; /* for checkpoint data flush */ 1738 spinlock_t dev_lock; /* protect dirty_device */ 1739 bool aligned_blksize; /* all devices has the same logical blksize */ 1740 1741 /* For write statistics */ 1742 u64 sectors_written_start; 1743 u64 kbytes_written; 1744 1745 /* Reference to checksum algorithm driver via cryptoapi */ 1746 struct crypto_shash *s_chksum_driver; 1747 1748 /* Precomputed FS UUID checksum for seeding other checksums */ 1749 __u32 s_chksum_seed; 1750 1751 struct workqueue_struct *post_read_wq; /* post read workqueue */ 1752 1753 /* 1754 * If we are in irq context, let's update error information into 1755 * on-disk superblock in the work. 1756 */ 1757 struct work_struct s_error_work; 1758 unsigned char errors[MAX_F2FS_ERRORS]; /* error flags */ 1759 unsigned char stop_reason[MAX_STOP_REASON]; /* stop reason */ 1760 spinlock_t error_lock; /* protect errors/stop_reason array */ 1761 bool error_dirty; /* errors of sb is dirty */ 1762 1763 struct kmem_cache *inline_xattr_slab; /* inline xattr entry */ 1764 unsigned int inline_xattr_slab_size; /* default inline xattr slab size */ 1765 1766 /* For reclaimed segs statistics per each GC mode */ 1767 unsigned int gc_segment_mode; /* GC state for reclaimed segments */ 1768 unsigned int gc_reclaimed_segs[MAX_GC_MODE]; /* Reclaimed segs for each mode */ 1769 1770 unsigned long seq_file_ra_mul; /* multiplier for ra_pages of seq. files in fadvise */ 1771 1772 int max_fragment_chunk; /* max chunk size for block fragmentation mode */ 1773 int max_fragment_hole; /* max hole size for block fragmentation mode */ 1774 1775 /* For atomic write statistics */ 1776 atomic64_t current_atomic_write; 1777 s64 peak_atomic_write; 1778 u64 committed_atomic_block; 1779 u64 revoked_atomic_block; 1780 1781 #ifdef CONFIG_F2FS_FS_COMPRESSION 1782 struct kmem_cache *page_array_slab; /* page array entry */ 1783 unsigned int page_array_slab_size; /* default page array slab size */ 1784 1785 /* For runtime compression statistics */ 1786 u64 compr_written_block; 1787 u64 compr_saved_block; 1788 u32 compr_new_inode; 1789 1790 /* For compressed block cache */ 1791 struct inode *compress_inode; /* cache compressed blocks */ 1792 unsigned int compress_percent; /* cache page percentage */ 1793 unsigned int compress_watermark; /* cache page watermark */ 1794 atomic_t compress_page_hit; /* cache hit count */ 1795 #endif 1796 1797 #ifdef CONFIG_F2FS_IOSTAT 1798 /* For app/fs IO statistics */ 1799 spinlock_t iostat_lock; 1800 unsigned long long iostat_count[NR_IO_TYPE]; 1801 unsigned long long iostat_bytes[NR_IO_TYPE]; 1802 unsigned long long prev_iostat_bytes[NR_IO_TYPE]; 1803 bool iostat_enable; 1804 unsigned long iostat_next_period; 1805 unsigned int iostat_period_ms; 1806 1807 /* For io latency related statistics info in one iostat period */ 1808 spinlock_t iostat_lat_lock; 1809 struct iostat_lat_info *iostat_io_lat; 1810 #endif 1811 }; 1812 1813 #ifdef CONFIG_F2FS_FAULT_INJECTION 1814 #define time_to_inject(sbi, type) __time_to_inject(sbi, type, __func__, \ 1815 __builtin_return_address(0)) 1816 static inline bool __time_to_inject(struct f2fs_sb_info *sbi, int type, 1817 const char *func, const char *parent_func) 1818 { 1819 struct f2fs_fault_info *ffi = &F2FS_OPTION(sbi).fault_info; 1820 1821 if (!ffi->inject_rate) 1822 return false; 1823 1824 if (!IS_FAULT_SET(ffi, type)) 1825 return false; 1826 1827 atomic_inc(&ffi->inject_ops); 1828 if (atomic_read(&ffi->inject_ops) >= ffi->inject_rate) { 1829 atomic_set(&ffi->inject_ops, 0); 1830 printk_ratelimited("%sF2FS-fs (%s) : inject %s in %s of %pS\n", 1831 KERN_INFO, sbi->sb->s_id, f2fs_fault_name[type], 1832 func, parent_func); 1833 return true; 1834 } 1835 return false; 1836 } 1837 #else 1838 static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type) 1839 { 1840 return false; 1841 } 1842 #endif 1843 1844 /* 1845 * Test if the mounted volume is a multi-device volume. 1846 * - For a single regular disk volume, sbi->s_ndevs is 0. 1847 * - For a single zoned disk volume, sbi->s_ndevs is 1. 1848 * - For a multi-device volume, sbi->s_ndevs is always 2 or more. 1849 */ 1850 static inline bool f2fs_is_multi_device(struct f2fs_sb_info *sbi) 1851 { 1852 return sbi->s_ndevs > 1; 1853 } 1854 1855 static inline void f2fs_update_time(struct f2fs_sb_info *sbi, int type) 1856 { 1857 unsigned long now = jiffies; 1858 1859 sbi->last_time[type] = now; 1860 1861 /* DISCARD_TIME and GC_TIME are based on REQ_TIME */ 1862 if (type == REQ_TIME) { 1863 sbi->last_time[DISCARD_TIME] = now; 1864 sbi->last_time[GC_TIME] = now; 1865 } 1866 } 1867 1868 static inline bool f2fs_time_over(struct f2fs_sb_info *sbi, int type) 1869 { 1870 unsigned long interval = sbi->interval_time[type] * HZ; 1871 1872 return time_after(jiffies, sbi->last_time[type] + interval); 1873 } 1874 1875 static inline unsigned int f2fs_time_to_wait(struct f2fs_sb_info *sbi, 1876 int type) 1877 { 1878 unsigned long interval = sbi->interval_time[type] * HZ; 1879 unsigned int wait_ms = 0; 1880 long delta; 1881 1882 delta = (sbi->last_time[type] + interval) - jiffies; 1883 if (delta > 0) 1884 wait_ms = jiffies_to_msecs(delta); 1885 1886 return wait_ms; 1887 } 1888 1889 /* 1890 * Inline functions 1891 */ 1892 static inline u32 __f2fs_crc32(struct f2fs_sb_info *sbi, u32 crc, 1893 const void *address, unsigned int length) 1894 { 1895 struct { 1896 struct shash_desc shash; 1897 char ctx[4]; 1898 } desc; 1899 int err; 1900 1901 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver) != sizeof(desc.ctx)); 1902 1903 desc.shash.tfm = sbi->s_chksum_driver; 1904 *(u32 *)desc.ctx = crc; 1905 1906 err = crypto_shash_update(&desc.shash, address, length); 1907 BUG_ON(err); 1908 1909 return *(u32 *)desc.ctx; 1910 } 1911 1912 static inline u32 f2fs_crc32(struct f2fs_sb_info *sbi, const void *address, 1913 unsigned int length) 1914 { 1915 return __f2fs_crc32(sbi, F2FS_SUPER_MAGIC, address, length); 1916 } 1917 1918 static inline bool f2fs_crc_valid(struct f2fs_sb_info *sbi, __u32 blk_crc, 1919 void *buf, size_t buf_size) 1920 { 1921 return f2fs_crc32(sbi, buf, buf_size) == blk_crc; 1922 } 1923 1924 static inline u32 f2fs_chksum(struct f2fs_sb_info *sbi, u32 crc, 1925 const void *address, unsigned int length) 1926 { 1927 return __f2fs_crc32(sbi, crc, address, length); 1928 } 1929 1930 static inline struct f2fs_inode_info *F2FS_I(struct inode *inode) 1931 { 1932 return container_of(inode, struct f2fs_inode_info, vfs_inode); 1933 } 1934 1935 static inline struct f2fs_sb_info *F2FS_SB(struct super_block *sb) 1936 { 1937 return sb->s_fs_info; 1938 } 1939 1940 static inline struct f2fs_sb_info *F2FS_I_SB(struct inode *inode) 1941 { 1942 return F2FS_SB(inode->i_sb); 1943 } 1944 1945 static inline struct f2fs_sb_info *F2FS_M_SB(struct address_space *mapping) 1946 { 1947 return F2FS_I_SB(mapping->host); 1948 } 1949 1950 static inline struct f2fs_sb_info *F2FS_P_SB(struct page *page) 1951 { 1952 return F2FS_M_SB(page_file_mapping(page)); 1953 } 1954 1955 static inline struct f2fs_super_block *F2FS_RAW_SUPER(struct f2fs_sb_info *sbi) 1956 { 1957 return (struct f2fs_super_block *)(sbi->raw_super); 1958 } 1959 1960 static inline struct f2fs_checkpoint *F2FS_CKPT(struct f2fs_sb_info *sbi) 1961 { 1962 return (struct f2fs_checkpoint *)(sbi->ckpt); 1963 } 1964 1965 static inline struct f2fs_node *F2FS_NODE(struct page *page) 1966 { 1967 return (struct f2fs_node *)page_address(page); 1968 } 1969 1970 static inline struct f2fs_inode *F2FS_INODE(struct page *page) 1971 { 1972 return &((struct f2fs_node *)page_address(page))->i; 1973 } 1974 1975 static inline struct f2fs_nm_info *NM_I(struct f2fs_sb_info *sbi) 1976 { 1977 return (struct f2fs_nm_info *)(sbi->nm_info); 1978 } 1979 1980 static inline struct f2fs_sm_info *SM_I(struct f2fs_sb_info *sbi) 1981 { 1982 return (struct f2fs_sm_info *)(sbi->sm_info); 1983 } 1984 1985 static inline struct sit_info *SIT_I(struct f2fs_sb_info *sbi) 1986 { 1987 return (struct sit_info *)(SM_I(sbi)->sit_info); 1988 } 1989 1990 static inline struct free_segmap_info *FREE_I(struct f2fs_sb_info *sbi) 1991 { 1992 return (struct free_segmap_info *)(SM_I(sbi)->free_info); 1993 } 1994 1995 static inline struct dirty_seglist_info *DIRTY_I(struct f2fs_sb_info *sbi) 1996 { 1997 return (struct dirty_seglist_info *)(SM_I(sbi)->dirty_info); 1998 } 1999 2000 static inline struct address_space *META_MAPPING(struct f2fs_sb_info *sbi) 2001 { 2002 return sbi->meta_inode->i_mapping; 2003 } 2004 2005 static inline struct address_space *NODE_MAPPING(struct f2fs_sb_info *sbi) 2006 { 2007 return sbi->node_inode->i_mapping; 2008 } 2009 2010 static inline bool is_sbi_flag_set(struct f2fs_sb_info *sbi, unsigned int type) 2011 { 2012 return test_bit(type, &sbi->s_flag); 2013 } 2014 2015 static inline void set_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type) 2016 { 2017 set_bit(type, &sbi->s_flag); 2018 } 2019 2020 static inline void clear_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type) 2021 { 2022 clear_bit(type, &sbi->s_flag); 2023 } 2024 2025 static inline unsigned long long cur_cp_version(struct f2fs_checkpoint *cp) 2026 { 2027 return le64_to_cpu(cp->checkpoint_ver); 2028 } 2029 2030 static inline unsigned long f2fs_qf_ino(struct super_block *sb, int type) 2031 { 2032 if (type < F2FS_MAX_QUOTAS) 2033 return le32_to_cpu(F2FS_SB(sb)->raw_super->qf_ino[type]); 2034 return 0; 2035 } 2036 2037 static inline __u64 cur_cp_crc(struct f2fs_checkpoint *cp) 2038 { 2039 size_t crc_offset = le32_to_cpu(cp->checksum_offset); 2040 return le32_to_cpu(*((__le32 *)((unsigned char *)cp + crc_offset))); 2041 } 2042 2043 static inline bool __is_set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 2044 { 2045 unsigned int ckpt_flags = le32_to_cpu(cp->ckpt_flags); 2046 2047 return ckpt_flags & f; 2048 } 2049 2050 static inline bool is_set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 2051 { 2052 return __is_set_ckpt_flags(F2FS_CKPT(sbi), f); 2053 } 2054 2055 static inline void __set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 2056 { 2057 unsigned int ckpt_flags; 2058 2059 ckpt_flags = le32_to_cpu(cp->ckpt_flags); 2060 ckpt_flags |= f; 2061 cp->ckpt_flags = cpu_to_le32(ckpt_flags); 2062 } 2063 2064 static inline void set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 2065 { 2066 unsigned long flags; 2067 2068 spin_lock_irqsave(&sbi->cp_lock, flags); 2069 __set_ckpt_flags(F2FS_CKPT(sbi), f); 2070 spin_unlock_irqrestore(&sbi->cp_lock, flags); 2071 } 2072 2073 static inline void __clear_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 2074 { 2075 unsigned int ckpt_flags; 2076 2077 ckpt_flags = le32_to_cpu(cp->ckpt_flags); 2078 ckpt_flags &= (~f); 2079 cp->ckpt_flags = cpu_to_le32(ckpt_flags); 2080 } 2081 2082 static inline void clear_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 2083 { 2084 unsigned long flags; 2085 2086 spin_lock_irqsave(&sbi->cp_lock, flags); 2087 __clear_ckpt_flags(F2FS_CKPT(sbi), f); 2088 spin_unlock_irqrestore(&sbi->cp_lock, flags); 2089 } 2090 2091 #define init_f2fs_rwsem(sem) \ 2092 do { \ 2093 static struct lock_class_key __key; \ 2094 \ 2095 __init_f2fs_rwsem((sem), #sem, &__key); \ 2096 } while (0) 2097 2098 static inline void __init_f2fs_rwsem(struct f2fs_rwsem *sem, 2099 const char *sem_name, struct lock_class_key *key) 2100 { 2101 __init_rwsem(&sem->internal_rwsem, sem_name, key); 2102 #ifdef CONFIG_F2FS_UNFAIR_RWSEM 2103 init_waitqueue_head(&sem->read_waiters); 2104 #endif 2105 } 2106 2107 static inline int f2fs_rwsem_is_locked(struct f2fs_rwsem *sem) 2108 { 2109 return rwsem_is_locked(&sem->internal_rwsem); 2110 } 2111 2112 static inline int f2fs_rwsem_is_contended(struct f2fs_rwsem *sem) 2113 { 2114 return rwsem_is_contended(&sem->internal_rwsem); 2115 } 2116 2117 static inline void f2fs_down_read(struct f2fs_rwsem *sem) 2118 { 2119 #ifdef CONFIG_F2FS_UNFAIR_RWSEM 2120 wait_event(sem->read_waiters, down_read_trylock(&sem->internal_rwsem)); 2121 #else 2122 down_read(&sem->internal_rwsem); 2123 #endif 2124 } 2125 2126 static inline int f2fs_down_read_trylock(struct f2fs_rwsem *sem) 2127 { 2128 return down_read_trylock(&sem->internal_rwsem); 2129 } 2130 2131 static inline void f2fs_up_read(struct f2fs_rwsem *sem) 2132 { 2133 up_read(&sem->internal_rwsem); 2134 } 2135 2136 static inline void f2fs_down_write(struct f2fs_rwsem *sem) 2137 { 2138 down_write(&sem->internal_rwsem); 2139 } 2140 2141 #ifdef CONFIG_DEBUG_LOCK_ALLOC 2142 static inline void f2fs_down_read_nested(struct f2fs_rwsem *sem, int subclass) 2143 { 2144 down_read_nested(&sem->internal_rwsem, subclass); 2145 } 2146 2147 static inline void f2fs_down_write_nested(struct f2fs_rwsem *sem, int subclass) 2148 { 2149 down_write_nested(&sem->internal_rwsem, subclass); 2150 } 2151 #else 2152 #define f2fs_down_read_nested(sem, subclass) f2fs_down_read(sem) 2153 #define f2fs_down_write_nested(sem, subclass) f2fs_down_write(sem) 2154 #endif 2155 2156 static inline int f2fs_down_write_trylock(struct f2fs_rwsem *sem) 2157 { 2158 return down_write_trylock(&sem->internal_rwsem); 2159 } 2160 2161 static inline void f2fs_up_write(struct f2fs_rwsem *sem) 2162 { 2163 up_write(&sem->internal_rwsem); 2164 #ifdef CONFIG_F2FS_UNFAIR_RWSEM 2165 wake_up_all(&sem->read_waiters); 2166 #endif 2167 } 2168 2169 static inline void f2fs_lock_op(struct f2fs_sb_info *sbi) 2170 { 2171 f2fs_down_read(&sbi->cp_rwsem); 2172 } 2173 2174 static inline int f2fs_trylock_op(struct f2fs_sb_info *sbi) 2175 { 2176 if (time_to_inject(sbi, FAULT_LOCK_OP)) 2177 return 0; 2178 return f2fs_down_read_trylock(&sbi->cp_rwsem); 2179 } 2180 2181 static inline void f2fs_unlock_op(struct f2fs_sb_info *sbi) 2182 { 2183 f2fs_up_read(&sbi->cp_rwsem); 2184 } 2185 2186 static inline void f2fs_lock_all(struct f2fs_sb_info *sbi) 2187 { 2188 f2fs_down_write(&sbi->cp_rwsem); 2189 } 2190 2191 static inline void f2fs_unlock_all(struct f2fs_sb_info *sbi) 2192 { 2193 f2fs_up_write(&sbi->cp_rwsem); 2194 } 2195 2196 static inline int __get_cp_reason(struct f2fs_sb_info *sbi) 2197 { 2198 int reason = CP_SYNC; 2199 2200 if (test_opt(sbi, FASTBOOT)) 2201 reason = CP_FASTBOOT; 2202 if (is_sbi_flag_set(sbi, SBI_IS_CLOSE)) 2203 reason = CP_UMOUNT; 2204 return reason; 2205 } 2206 2207 static inline bool __remain_node_summaries(int reason) 2208 { 2209 return (reason & (CP_UMOUNT | CP_FASTBOOT)); 2210 } 2211 2212 static inline bool __exist_node_summaries(struct f2fs_sb_info *sbi) 2213 { 2214 return (is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG) || 2215 is_set_ckpt_flags(sbi, CP_FASTBOOT_FLAG)); 2216 } 2217 2218 /* 2219 * Check whether the inode has blocks or not 2220 */ 2221 static inline int F2FS_HAS_BLOCKS(struct inode *inode) 2222 { 2223 block_t xattr_block = F2FS_I(inode)->i_xattr_nid ? 1 : 0; 2224 2225 return (inode->i_blocks >> F2FS_LOG_SECTORS_PER_BLOCK) > xattr_block; 2226 } 2227 2228 static inline bool f2fs_has_xattr_block(unsigned int ofs) 2229 { 2230 return ofs == XATTR_NODE_OFFSET; 2231 } 2232 2233 static inline bool __allow_reserved_blocks(struct f2fs_sb_info *sbi, 2234 struct inode *inode, bool cap) 2235 { 2236 if (!inode) 2237 return true; 2238 if (!test_opt(sbi, RESERVE_ROOT)) 2239 return false; 2240 if (IS_NOQUOTA(inode)) 2241 return true; 2242 if (uid_eq(F2FS_OPTION(sbi).s_resuid, current_fsuid())) 2243 return true; 2244 if (!gid_eq(F2FS_OPTION(sbi).s_resgid, GLOBAL_ROOT_GID) && 2245 in_group_p(F2FS_OPTION(sbi).s_resgid)) 2246 return true; 2247 if (cap && capable(CAP_SYS_RESOURCE)) 2248 return true; 2249 return false; 2250 } 2251 2252 static inline void f2fs_i_blocks_write(struct inode *, block_t, bool, bool); 2253 static inline int inc_valid_block_count(struct f2fs_sb_info *sbi, 2254 struct inode *inode, blkcnt_t *count) 2255 { 2256 blkcnt_t diff = 0, release = 0; 2257 block_t avail_user_block_count; 2258 int ret; 2259 2260 ret = dquot_reserve_block(inode, *count); 2261 if (ret) 2262 return ret; 2263 2264 if (time_to_inject(sbi, FAULT_BLOCK)) { 2265 release = *count; 2266 goto release_quota; 2267 } 2268 2269 /* 2270 * let's increase this in prior to actual block count change in order 2271 * for f2fs_sync_file to avoid data races when deciding checkpoint. 2272 */ 2273 percpu_counter_add(&sbi->alloc_valid_block_count, (*count)); 2274 2275 spin_lock(&sbi->stat_lock); 2276 sbi->total_valid_block_count += (block_t)(*count); 2277 avail_user_block_count = sbi->user_block_count - 2278 sbi->current_reserved_blocks; 2279 2280 if (!__allow_reserved_blocks(sbi, inode, true)) 2281 avail_user_block_count -= F2FS_OPTION(sbi).root_reserved_blocks; 2282 2283 if (F2FS_IO_ALIGNED(sbi)) 2284 avail_user_block_count -= sbi->blocks_per_seg * 2285 SM_I(sbi)->additional_reserved_segments; 2286 2287 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) { 2288 if (avail_user_block_count > sbi->unusable_block_count) 2289 avail_user_block_count -= sbi->unusable_block_count; 2290 else 2291 avail_user_block_count = 0; 2292 } 2293 if (unlikely(sbi->total_valid_block_count > avail_user_block_count)) { 2294 diff = sbi->total_valid_block_count - avail_user_block_count; 2295 if (diff > *count) 2296 diff = *count; 2297 *count -= diff; 2298 release = diff; 2299 sbi->total_valid_block_count -= diff; 2300 if (!*count) { 2301 spin_unlock(&sbi->stat_lock); 2302 goto enospc; 2303 } 2304 } 2305 spin_unlock(&sbi->stat_lock); 2306 2307 if (unlikely(release)) { 2308 percpu_counter_sub(&sbi->alloc_valid_block_count, release); 2309 dquot_release_reservation_block(inode, release); 2310 } 2311 f2fs_i_blocks_write(inode, *count, true, true); 2312 return 0; 2313 2314 enospc: 2315 percpu_counter_sub(&sbi->alloc_valid_block_count, release); 2316 release_quota: 2317 dquot_release_reservation_block(inode, release); 2318 return -ENOSPC; 2319 } 2320 2321 __printf(2, 3) 2322 void f2fs_printk(struct f2fs_sb_info *sbi, const char *fmt, ...); 2323 2324 #define f2fs_err(sbi, fmt, ...) \ 2325 f2fs_printk(sbi, KERN_ERR fmt, ##__VA_ARGS__) 2326 #define f2fs_warn(sbi, fmt, ...) \ 2327 f2fs_printk(sbi, KERN_WARNING fmt, ##__VA_ARGS__) 2328 #define f2fs_notice(sbi, fmt, ...) \ 2329 f2fs_printk(sbi, KERN_NOTICE fmt, ##__VA_ARGS__) 2330 #define f2fs_info(sbi, fmt, ...) \ 2331 f2fs_printk(sbi, KERN_INFO fmt, ##__VA_ARGS__) 2332 #define f2fs_debug(sbi, fmt, ...) \ 2333 f2fs_printk(sbi, KERN_DEBUG fmt, ##__VA_ARGS__) 2334 2335 #define PAGE_PRIVATE_GET_FUNC(name, flagname) \ 2336 static inline bool page_private_##name(struct page *page) \ 2337 { \ 2338 return PagePrivate(page) && \ 2339 test_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)) && \ 2340 test_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \ 2341 } 2342 2343 #define PAGE_PRIVATE_SET_FUNC(name, flagname) \ 2344 static inline void set_page_private_##name(struct page *page) \ 2345 { \ 2346 if (!PagePrivate(page)) \ 2347 attach_page_private(page, (void *)0); \ 2348 set_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)); \ 2349 set_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \ 2350 } 2351 2352 #define PAGE_PRIVATE_CLEAR_FUNC(name, flagname) \ 2353 static inline void clear_page_private_##name(struct page *page) \ 2354 { \ 2355 clear_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \ 2356 if (page_private(page) == BIT(PAGE_PRIVATE_NOT_POINTER)) \ 2357 detach_page_private(page); \ 2358 } 2359 2360 PAGE_PRIVATE_GET_FUNC(nonpointer, NOT_POINTER); 2361 PAGE_PRIVATE_GET_FUNC(inline, INLINE_INODE); 2362 PAGE_PRIVATE_GET_FUNC(gcing, ONGOING_MIGRATION); 2363 PAGE_PRIVATE_GET_FUNC(dummy, DUMMY_WRITE); 2364 2365 PAGE_PRIVATE_SET_FUNC(reference, REF_RESOURCE); 2366 PAGE_PRIVATE_SET_FUNC(inline, INLINE_INODE); 2367 PAGE_PRIVATE_SET_FUNC(gcing, ONGOING_MIGRATION); 2368 PAGE_PRIVATE_SET_FUNC(dummy, DUMMY_WRITE); 2369 2370 PAGE_PRIVATE_CLEAR_FUNC(reference, REF_RESOURCE); 2371 PAGE_PRIVATE_CLEAR_FUNC(inline, INLINE_INODE); 2372 PAGE_PRIVATE_CLEAR_FUNC(gcing, ONGOING_MIGRATION); 2373 PAGE_PRIVATE_CLEAR_FUNC(dummy, DUMMY_WRITE); 2374 2375 static inline unsigned long get_page_private_data(struct page *page) 2376 { 2377 unsigned long data = page_private(page); 2378 2379 if (!test_bit(PAGE_PRIVATE_NOT_POINTER, &data)) 2380 return 0; 2381 return data >> PAGE_PRIVATE_MAX; 2382 } 2383 2384 static inline void set_page_private_data(struct page *page, unsigned long data) 2385 { 2386 if (!PagePrivate(page)) 2387 attach_page_private(page, (void *)0); 2388 set_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)); 2389 page_private(page) |= data << PAGE_PRIVATE_MAX; 2390 } 2391 2392 static inline void clear_page_private_data(struct page *page) 2393 { 2394 page_private(page) &= GENMASK(PAGE_PRIVATE_MAX - 1, 0); 2395 if (page_private(page) == BIT(PAGE_PRIVATE_NOT_POINTER)) 2396 detach_page_private(page); 2397 } 2398 2399 static inline void clear_page_private_all(struct page *page) 2400 { 2401 clear_page_private_data(page); 2402 clear_page_private_reference(page); 2403 clear_page_private_gcing(page); 2404 clear_page_private_inline(page); 2405 2406 f2fs_bug_on(F2FS_P_SB(page), page_private(page)); 2407 } 2408 2409 static inline void dec_valid_block_count(struct f2fs_sb_info *sbi, 2410 struct inode *inode, 2411 block_t count) 2412 { 2413 blkcnt_t sectors = count << F2FS_LOG_SECTORS_PER_BLOCK; 2414 2415 spin_lock(&sbi->stat_lock); 2416 f2fs_bug_on(sbi, sbi->total_valid_block_count < (block_t) count); 2417 sbi->total_valid_block_count -= (block_t)count; 2418 if (sbi->reserved_blocks && 2419 sbi->current_reserved_blocks < sbi->reserved_blocks) 2420 sbi->current_reserved_blocks = min(sbi->reserved_blocks, 2421 sbi->current_reserved_blocks + count); 2422 spin_unlock(&sbi->stat_lock); 2423 if (unlikely(inode->i_blocks < sectors)) { 2424 f2fs_warn(sbi, "Inconsistent i_blocks, ino:%lu, iblocks:%llu, sectors:%llu", 2425 inode->i_ino, 2426 (unsigned long long)inode->i_blocks, 2427 (unsigned long long)sectors); 2428 set_sbi_flag(sbi, SBI_NEED_FSCK); 2429 return; 2430 } 2431 f2fs_i_blocks_write(inode, count, false, true); 2432 } 2433 2434 static inline void inc_page_count(struct f2fs_sb_info *sbi, int count_type) 2435 { 2436 atomic_inc(&sbi->nr_pages[count_type]); 2437 2438 if (count_type == F2FS_DIRTY_DENTS || 2439 count_type == F2FS_DIRTY_NODES || 2440 count_type == F2FS_DIRTY_META || 2441 count_type == F2FS_DIRTY_QDATA || 2442 count_type == F2FS_DIRTY_IMETA) 2443 set_sbi_flag(sbi, SBI_IS_DIRTY); 2444 } 2445 2446 static inline void inode_inc_dirty_pages(struct inode *inode) 2447 { 2448 atomic_inc(&F2FS_I(inode)->dirty_pages); 2449 inc_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ? 2450 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA); 2451 if (IS_NOQUOTA(inode)) 2452 inc_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA); 2453 } 2454 2455 static inline void dec_page_count(struct f2fs_sb_info *sbi, int count_type) 2456 { 2457 atomic_dec(&sbi->nr_pages[count_type]); 2458 } 2459 2460 static inline void inode_dec_dirty_pages(struct inode *inode) 2461 { 2462 if (!S_ISDIR(inode->i_mode) && !S_ISREG(inode->i_mode) && 2463 !S_ISLNK(inode->i_mode)) 2464 return; 2465 2466 atomic_dec(&F2FS_I(inode)->dirty_pages); 2467 dec_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ? 2468 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA); 2469 if (IS_NOQUOTA(inode)) 2470 dec_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA); 2471 } 2472 2473 static inline void inc_atomic_write_cnt(struct inode *inode) 2474 { 2475 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 2476 struct f2fs_inode_info *fi = F2FS_I(inode); 2477 u64 current_write; 2478 2479 fi->atomic_write_cnt++; 2480 atomic64_inc(&sbi->current_atomic_write); 2481 current_write = atomic64_read(&sbi->current_atomic_write); 2482 if (current_write > sbi->peak_atomic_write) 2483 sbi->peak_atomic_write = current_write; 2484 } 2485 2486 static inline void release_atomic_write_cnt(struct inode *inode) 2487 { 2488 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 2489 struct f2fs_inode_info *fi = F2FS_I(inode); 2490 2491 atomic64_sub(fi->atomic_write_cnt, &sbi->current_atomic_write); 2492 fi->atomic_write_cnt = 0; 2493 } 2494 2495 static inline s64 get_pages(struct f2fs_sb_info *sbi, int count_type) 2496 { 2497 return atomic_read(&sbi->nr_pages[count_type]); 2498 } 2499 2500 static inline int get_dirty_pages(struct inode *inode) 2501 { 2502 return atomic_read(&F2FS_I(inode)->dirty_pages); 2503 } 2504 2505 static inline int get_blocktype_secs(struct f2fs_sb_info *sbi, int block_type) 2506 { 2507 unsigned int pages_per_sec = sbi->segs_per_sec * sbi->blocks_per_seg; 2508 unsigned int segs = (get_pages(sbi, block_type) + pages_per_sec - 1) >> 2509 sbi->log_blocks_per_seg; 2510 2511 return segs / sbi->segs_per_sec; 2512 } 2513 2514 static inline block_t valid_user_blocks(struct f2fs_sb_info *sbi) 2515 { 2516 return sbi->total_valid_block_count; 2517 } 2518 2519 static inline block_t discard_blocks(struct f2fs_sb_info *sbi) 2520 { 2521 return sbi->discard_blks; 2522 } 2523 2524 static inline unsigned long __bitmap_size(struct f2fs_sb_info *sbi, int flag) 2525 { 2526 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi); 2527 2528 /* return NAT or SIT bitmap */ 2529 if (flag == NAT_BITMAP) 2530 return le32_to_cpu(ckpt->nat_ver_bitmap_bytesize); 2531 else if (flag == SIT_BITMAP) 2532 return le32_to_cpu(ckpt->sit_ver_bitmap_bytesize); 2533 2534 return 0; 2535 } 2536 2537 static inline block_t __cp_payload(struct f2fs_sb_info *sbi) 2538 { 2539 return le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_payload); 2540 } 2541 2542 static inline void *__bitmap_ptr(struct f2fs_sb_info *sbi, int flag) 2543 { 2544 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi); 2545 void *tmp_ptr = &ckpt->sit_nat_version_bitmap; 2546 int offset; 2547 2548 if (is_set_ckpt_flags(sbi, CP_LARGE_NAT_BITMAP_FLAG)) { 2549 offset = (flag == SIT_BITMAP) ? 2550 le32_to_cpu(ckpt->nat_ver_bitmap_bytesize) : 0; 2551 /* 2552 * if large_nat_bitmap feature is enabled, leave checksum 2553 * protection for all nat/sit bitmaps. 2554 */ 2555 return tmp_ptr + offset + sizeof(__le32); 2556 } 2557 2558 if (__cp_payload(sbi) > 0) { 2559 if (flag == NAT_BITMAP) 2560 return tmp_ptr; 2561 else 2562 return (unsigned char *)ckpt + F2FS_BLKSIZE; 2563 } else { 2564 offset = (flag == NAT_BITMAP) ? 2565 le32_to_cpu(ckpt->sit_ver_bitmap_bytesize) : 0; 2566 return tmp_ptr + offset; 2567 } 2568 } 2569 2570 static inline block_t __start_cp_addr(struct f2fs_sb_info *sbi) 2571 { 2572 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr); 2573 2574 if (sbi->cur_cp_pack == 2) 2575 start_addr += sbi->blocks_per_seg; 2576 return start_addr; 2577 } 2578 2579 static inline block_t __start_cp_next_addr(struct f2fs_sb_info *sbi) 2580 { 2581 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr); 2582 2583 if (sbi->cur_cp_pack == 1) 2584 start_addr += sbi->blocks_per_seg; 2585 return start_addr; 2586 } 2587 2588 static inline void __set_cp_next_pack(struct f2fs_sb_info *sbi) 2589 { 2590 sbi->cur_cp_pack = (sbi->cur_cp_pack == 1) ? 2 : 1; 2591 } 2592 2593 static inline block_t __start_sum_addr(struct f2fs_sb_info *sbi) 2594 { 2595 return le32_to_cpu(F2FS_CKPT(sbi)->cp_pack_start_sum); 2596 } 2597 2598 extern void f2fs_mark_inode_dirty_sync(struct inode *inode, bool sync); 2599 static inline int inc_valid_node_count(struct f2fs_sb_info *sbi, 2600 struct inode *inode, bool is_inode) 2601 { 2602 block_t valid_block_count; 2603 unsigned int valid_node_count, user_block_count; 2604 int err; 2605 2606 if (is_inode) { 2607 if (inode) { 2608 err = dquot_alloc_inode(inode); 2609 if (err) 2610 return err; 2611 } 2612 } else { 2613 err = dquot_reserve_block(inode, 1); 2614 if (err) 2615 return err; 2616 } 2617 2618 if (time_to_inject(sbi, FAULT_BLOCK)) 2619 goto enospc; 2620 2621 spin_lock(&sbi->stat_lock); 2622 2623 valid_block_count = sbi->total_valid_block_count + 2624 sbi->current_reserved_blocks + 1; 2625 2626 if (!__allow_reserved_blocks(sbi, inode, false)) 2627 valid_block_count += F2FS_OPTION(sbi).root_reserved_blocks; 2628 2629 if (F2FS_IO_ALIGNED(sbi)) 2630 valid_block_count += sbi->blocks_per_seg * 2631 SM_I(sbi)->additional_reserved_segments; 2632 2633 user_block_count = sbi->user_block_count; 2634 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) 2635 user_block_count -= sbi->unusable_block_count; 2636 2637 if (unlikely(valid_block_count > user_block_count)) { 2638 spin_unlock(&sbi->stat_lock); 2639 goto enospc; 2640 } 2641 2642 valid_node_count = sbi->total_valid_node_count + 1; 2643 if (unlikely(valid_node_count > sbi->total_node_count)) { 2644 spin_unlock(&sbi->stat_lock); 2645 goto enospc; 2646 } 2647 2648 sbi->total_valid_node_count++; 2649 sbi->total_valid_block_count++; 2650 spin_unlock(&sbi->stat_lock); 2651 2652 if (inode) { 2653 if (is_inode) 2654 f2fs_mark_inode_dirty_sync(inode, true); 2655 else 2656 f2fs_i_blocks_write(inode, 1, true, true); 2657 } 2658 2659 percpu_counter_inc(&sbi->alloc_valid_block_count); 2660 return 0; 2661 2662 enospc: 2663 if (is_inode) { 2664 if (inode) 2665 dquot_free_inode(inode); 2666 } else { 2667 dquot_release_reservation_block(inode, 1); 2668 } 2669 return -ENOSPC; 2670 } 2671 2672 static inline void dec_valid_node_count(struct f2fs_sb_info *sbi, 2673 struct inode *inode, bool is_inode) 2674 { 2675 spin_lock(&sbi->stat_lock); 2676 2677 if (unlikely(!sbi->total_valid_block_count || 2678 !sbi->total_valid_node_count)) { 2679 f2fs_warn(sbi, "dec_valid_node_count: inconsistent block counts, total_valid_block:%u, total_valid_node:%u", 2680 sbi->total_valid_block_count, 2681 sbi->total_valid_node_count); 2682 set_sbi_flag(sbi, SBI_NEED_FSCK); 2683 } else { 2684 sbi->total_valid_block_count--; 2685 sbi->total_valid_node_count--; 2686 } 2687 2688 if (sbi->reserved_blocks && 2689 sbi->current_reserved_blocks < sbi->reserved_blocks) 2690 sbi->current_reserved_blocks++; 2691 2692 spin_unlock(&sbi->stat_lock); 2693 2694 if (is_inode) { 2695 dquot_free_inode(inode); 2696 } else { 2697 if (unlikely(inode->i_blocks == 0)) { 2698 f2fs_warn(sbi, "dec_valid_node_count: inconsistent i_blocks, ino:%lu, iblocks:%llu", 2699 inode->i_ino, 2700 (unsigned long long)inode->i_blocks); 2701 set_sbi_flag(sbi, SBI_NEED_FSCK); 2702 return; 2703 } 2704 f2fs_i_blocks_write(inode, 1, false, true); 2705 } 2706 } 2707 2708 static inline unsigned int valid_node_count(struct f2fs_sb_info *sbi) 2709 { 2710 return sbi->total_valid_node_count; 2711 } 2712 2713 static inline void inc_valid_inode_count(struct f2fs_sb_info *sbi) 2714 { 2715 percpu_counter_inc(&sbi->total_valid_inode_count); 2716 } 2717 2718 static inline void dec_valid_inode_count(struct f2fs_sb_info *sbi) 2719 { 2720 percpu_counter_dec(&sbi->total_valid_inode_count); 2721 } 2722 2723 static inline s64 valid_inode_count(struct f2fs_sb_info *sbi) 2724 { 2725 return percpu_counter_sum_positive(&sbi->total_valid_inode_count); 2726 } 2727 2728 static inline struct page *f2fs_grab_cache_page(struct address_space *mapping, 2729 pgoff_t index, bool for_write) 2730 { 2731 struct page *page; 2732 unsigned int flags; 2733 2734 if (IS_ENABLED(CONFIG_F2FS_FAULT_INJECTION)) { 2735 if (!for_write) 2736 page = find_get_page_flags(mapping, index, 2737 FGP_LOCK | FGP_ACCESSED); 2738 else 2739 page = find_lock_page(mapping, index); 2740 if (page) 2741 return page; 2742 2743 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_ALLOC)) 2744 return NULL; 2745 } 2746 2747 if (!for_write) 2748 return grab_cache_page(mapping, index); 2749 2750 flags = memalloc_nofs_save(); 2751 page = grab_cache_page_write_begin(mapping, index); 2752 memalloc_nofs_restore(flags); 2753 2754 return page; 2755 } 2756 2757 static inline struct page *f2fs_pagecache_get_page( 2758 struct address_space *mapping, pgoff_t index, 2759 fgf_t fgp_flags, gfp_t gfp_mask) 2760 { 2761 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_GET)) 2762 return NULL; 2763 2764 return pagecache_get_page(mapping, index, fgp_flags, gfp_mask); 2765 } 2766 2767 static inline void f2fs_put_page(struct page *page, int unlock) 2768 { 2769 if (!page) 2770 return; 2771 2772 if (unlock) { 2773 f2fs_bug_on(F2FS_P_SB(page), !PageLocked(page)); 2774 unlock_page(page); 2775 } 2776 put_page(page); 2777 } 2778 2779 static inline void f2fs_put_dnode(struct dnode_of_data *dn) 2780 { 2781 if (dn->node_page) 2782 f2fs_put_page(dn->node_page, 1); 2783 if (dn->inode_page && dn->node_page != dn->inode_page) 2784 f2fs_put_page(dn->inode_page, 0); 2785 dn->node_page = NULL; 2786 dn->inode_page = NULL; 2787 } 2788 2789 static inline struct kmem_cache *f2fs_kmem_cache_create(const char *name, 2790 size_t size) 2791 { 2792 return kmem_cache_create(name, size, 0, SLAB_RECLAIM_ACCOUNT, NULL); 2793 } 2794 2795 static inline void *f2fs_kmem_cache_alloc_nofail(struct kmem_cache *cachep, 2796 gfp_t flags) 2797 { 2798 void *entry; 2799 2800 entry = kmem_cache_alloc(cachep, flags); 2801 if (!entry) 2802 entry = kmem_cache_alloc(cachep, flags | __GFP_NOFAIL); 2803 return entry; 2804 } 2805 2806 static inline void *f2fs_kmem_cache_alloc(struct kmem_cache *cachep, 2807 gfp_t flags, bool nofail, struct f2fs_sb_info *sbi) 2808 { 2809 if (nofail) 2810 return f2fs_kmem_cache_alloc_nofail(cachep, flags); 2811 2812 if (time_to_inject(sbi, FAULT_SLAB_ALLOC)) 2813 return NULL; 2814 2815 return kmem_cache_alloc(cachep, flags); 2816 } 2817 2818 static inline bool is_inflight_io(struct f2fs_sb_info *sbi, int type) 2819 { 2820 if (get_pages(sbi, F2FS_RD_DATA) || get_pages(sbi, F2FS_RD_NODE) || 2821 get_pages(sbi, F2FS_RD_META) || get_pages(sbi, F2FS_WB_DATA) || 2822 get_pages(sbi, F2FS_WB_CP_DATA) || 2823 get_pages(sbi, F2FS_DIO_READ) || 2824 get_pages(sbi, F2FS_DIO_WRITE)) 2825 return true; 2826 2827 if (type != DISCARD_TIME && SM_I(sbi) && SM_I(sbi)->dcc_info && 2828 atomic_read(&SM_I(sbi)->dcc_info->queued_discard)) 2829 return true; 2830 2831 if (SM_I(sbi) && SM_I(sbi)->fcc_info && 2832 atomic_read(&SM_I(sbi)->fcc_info->queued_flush)) 2833 return true; 2834 return false; 2835 } 2836 2837 static inline bool is_idle(struct f2fs_sb_info *sbi, int type) 2838 { 2839 if (sbi->gc_mode == GC_URGENT_HIGH) 2840 return true; 2841 2842 if (is_inflight_io(sbi, type)) 2843 return false; 2844 2845 if (sbi->gc_mode == GC_URGENT_MID) 2846 return true; 2847 2848 if (sbi->gc_mode == GC_URGENT_LOW && 2849 (type == DISCARD_TIME || type == GC_TIME)) 2850 return true; 2851 2852 return f2fs_time_over(sbi, type); 2853 } 2854 2855 static inline void f2fs_radix_tree_insert(struct radix_tree_root *root, 2856 unsigned long index, void *item) 2857 { 2858 while (radix_tree_insert(root, index, item)) 2859 cond_resched(); 2860 } 2861 2862 #define RAW_IS_INODE(p) ((p)->footer.nid == (p)->footer.ino) 2863 2864 static inline bool IS_INODE(struct page *page) 2865 { 2866 struct f2fs_node *p = F2FS_NODE(page); 2867 2868 return RAW_IS_INODE(p); 2869 } 2870 2871 static inline int offset_in_addr(struct f2fs_inode *i) 2872 { 2873 return (i->i_inline & F2FS_EXTRA_ATTR) ? 2874 (le16_to_cpu(i->i_extra_isize) / sizeof(__le32)) : 0; 2875 } 2876 2877 static inline __le32 *blkaddr_in_node(struct f2fs_node *node) 2878 { 2879 return RAW_IS_INODE(node) ? node->i.i_addr : node->dn.addr; 2880 } 2881 2882 static inline int f2fs_has_extra_attr(struct inode *inode); 2883 static inline block_t data_blkaddr(struct inode *inode, 2884 struct page *node_page, unsigned int offset) 2885 { 2886 struct f2fs_node *raw_node; 2887 __le32 *addr_array; 2888 int base = 0; 2889 bool is_inode = IS_INODE(node_page); 2890 2891 raw_node = F2FS_NODE(node_page); 2892 2893 if (is_inode) { 2894 if (!inode) 2895 /* from GC path only */ 2896 base = offset_in_addr(&raw_node->i); 2897 else if (f2fs_has_extra_attr(inode)) 2898 base = get_extra_isize(inode); 2899 } 2900 2901 addr_array = blkaddr_in_node(raw_node); 2902 return le32_to_cpu(addr_array[base + offset]); 2903 } 2904 2905 static inline block_t f2fs_data_blkaddr(struct dnode_of_data *dn) 2906 { 2907 return data_blkaddr(dn->inode, dn->node_page, dn->ofs_in_node); 2908 } 2909 2910 static inline int f2fs_test_bit(unsigned int nr, char *addr) 2911 { 2912 int mask; 2913 2914 addr += (nr >> 3); 2915 mask = BIT(7 - (nr & 0x07)); 2916 return mask & *addr; 2917 } 2918 2919 static inline void f2fs_set_bit(unsigned int nr, char *addr) 2920 { 2921 int mask; 2922 2923 addr += (nr >> 3); 2924 mask = BIT(7 - (nr & 0x07)); 2925 *addr |= mask; 2926 } 2927 2928 static inline void f2fs_clear_bit(unsigned int nr, char *addr) 2929 { 2930 int mask; 2931 2932 addr += (nr >> 3); 2933 mask = BIT(7 - (nr & 0x07)); 2934 *addr &= ~mask; 2935 } 2936 2937 static inline int f2fs_test_and_set_bit(unsigned int nr, char *addr) 2938 { 2939 int mask; 2940 int ret; 2941 2942 addr += (nr >> 3); 2943 mask = BIT(7 - (nr & 0x07)); 2944 ret = mask & *addr; 2945 *addr |= mask; 2946 return ret; 2947 } 2948 2949 static inline int f2fs_test_and_clear_bit(unsigned int nr, char *addr) 2950 { 2951 int mask; 2952 int ret; 2953 2954 addr += (nr >> 3); 2955 mask = BIT(7 - (nr & 0x07)); 2956 ret = mask & *addr; 2957 *addr &= ~mask; 2958 return ret; 2959 } 2960 2961 static inline void f2fs_change_bit(unsigned int nr, char *addr) 2962 { 2963 int mask; 2964 2965 addr += (nr >> 3); 2966 mask = BIT(7 - (nr & 0x07)); 2967 *addr ^= mask; 2968 } 2969 2970 /* 2971 * On-disk inode flags (f2fs_inode::i_flags) 2972 */ 2973 #define F2FS_COMPR_FL 0x00000004 /* Compress file */ 2974 #define F2FS_SYNC_FL 0x00000008 /* Synchronous updates */ 2975 #define F2FS_IMMUTABLE_FL 0x00000010 /* Immutable file */ 2976 #define F2FS_APPEND_FL 0x00000020 /* writes to file may only append */ 2977 #define F2FS_NODUMP_FL 0x00000040 /* do not dump file */ 2978 #define F2FS_NOATIME_FL 0x00000080 /* do not update atime */ 2979 #define F2FS_NOCOMP_FL 0x00000400 /* Don't compress */ 2980 #define F2FS_INDEX_FL 0x00001000 /* hash-indexed directory */ 2981 #define F2FS_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */ 2982 #define F2FS_PROJINHERIT_FL 0x20000000 /* Create with parents projid */ 2983 #define F2FS_CASEFOLD_FL 0x40000000 /* Casefolded file */ 2984 2985 #define F2FS_QUOTA_DEFAULT_FL (F2FS_NOATIME_FL | F2FS_IMMUTABLE_FL) 2986 2987 /* Flags that should be inherited by new inodes from their parent. */ 2988 #define F2FS_FL_INHERITED (F2FS_SYNC_FL | F2FS_NODUMP_FL | F2FS_NOATIME_FL | \ 2989 F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \ 2990 F2FS_CASEFOLD_FL) 2991 2992 /* Flags that are appropriate for regular files (all but dir-specific ones). */ 2993 #define F2FS_REG_FLMASK (~(F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \ 2994 F2FS_CASEFOLD_FL)) 2995 2996 /* Flags that are appropriate for non-directories/regular files. */ 2997 #define F2FS_OTHER_FLMASK (F2FS_NODUMP_FL | F2FS_NOATIME_FL) 2998 2999 static inline __u32 f2fs_mask_flags(umode_t mode, __u32 flags) 3000 { 3001 if (S_ISDIR(mode)) 3002 return flags; 3003 else if (S_ISREG(mode)) 3004 return flags & F2FS_REG_FLMASK; 3005 else 3006 return flags & F2FS_OTHER_FLMASK; 3007 } 3008 3009 static inline void __mark_inode_dirty_flag(struct inode *inode, 3010 int flag, bool set) 3011 { 3012 switch (flag) { 3013 case FI_INLINE_XATTR: 3014 case FI_INLINE_DATA: 3015 case FI_INLINE_DENTRY: 3016 case FI_NEW_INODE: 3017 if (set) 3018 return; 3019 fallthrough; 3020 case FI_DATA_EXIST: 3021 case FI_INLINE_DOTS: 3022 case FI_PIN_FILE: 3023 case FI_COMPRESS_RELEASED: 3024 f2fs_mark_inode_dirty_sync(inode, true); 3025 } 3026 } 3027 3028 static inline void set_inode_flag(struct inode *inode, int flag) 3029 { 3030 set_bit(flag, F2FS_I(inode)->flags); 3031 __mark_inode_dirty_flag(inode, flag, true); 3032 } 3033 3034 static inline int is_inode_flag_set(struct inode *inode, int flag) 3035 { 3036 return test_bit(flag, F2FS_I(inode)->flags); 3037 } 3038 3039 static inline void clear_inode_flag(struct inode *inode, int flag) 3040 { 3041 clear_bit(flag, F2FS_I(inode)->flags); 3042 __mark_inode_dirty_flag(inode, flag, false); 3043 } 3044 3045 static inline bool f2fs_verity_in_progress(struct inode *inode) 3046 { 3047 return IS_ENABLED(CONFIG_FS_VERITY) && 3048 is_inode_flag_set(inode, FI_VERITY_IN_PROGRESS); 3049 } 3050 3051 static inline void set_acl_inode(struct inode *inode, umode_t mode) 3052 { 3053 F2FS_I(inode)->i_acl_mode = mode; 3054 set_inode_flag(inode, FI_ACL_MODE); 3055 f2fs_mark_inode_dirty_sync(inode, false); 3056 } 3057 3058 static inline void f2fs_i_links_write(struct inode *inode, bool inc) 3059 { 3060 if (inc) 3061 inc_nlink(inode); 3062 else 3063 drop_nlink(inode); 3064 f2fs_mark_inode_dirty_sync(inode, true); 3065 } 3066 3067 static inline void f2fs_i_blocks_write(struct inode *inode, 3068 block_t diff, bool add, bool claim) 3069 { 3070 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE); 3071 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER); 3072 3073 /* add = 1, claim = 1 should be dquot_reserve_block in pair */ 3074 if (add) { 3075 if (claim) 3076 dquot_claim_block(inode, diff); 3077 else 3078 dquot_alloc_block_nofail(inode, diff); 3079 } else { 3080 dquot_free_block(inode, diff); 3081 } 3082 3083 f2fs_mark_inode_dirty_sync(inode, true); 3084 if (clean || recover) 3085 set_inode_flag(inode, FI_AUTO_RECOVER); 3086 } 3087 3088 static inline bool f2fs_is_atomic_file(struct inode *inode); 3089 3090 static inline void f2fs_i_size_write(struct inode *inode, loff_t i_size) 3091 { 3092 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE); 3093 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER); 3094 3095 if (i_size_read(inode) == i_size) 3096 return; 3097 3098 i_size_write(inode, i_size); 3099 3100 if (f2fs_is_atomic_file(inode)) 3101 return; 3102 3103 f2fs_mark_inode_dirty_sync(inode, true); 3104 if (clean || recover) 3105 set_inode_flag(inode, FI_AUTO_RECOVER); 3106 } 3107 3108 static inline void f2fs_i_depth_write(struct inode *inode, unsigned int depth) 3109 { 3110 F2FS_I(inode)->i_current_depth = depth; 3111 f2fs_mark_inode_dirty_sync(inode, true); 3112 } 3113 3114 static inline void f2fs_i_gc_failures_write(struct inode *inode, 3115 unsigned int count) 3116 { 3117 F2FS_I(inode)->i_gc_failures[GC_FAILURE_PIN] = count; 3118 f2fs_mark_inode_dirty_sync(inode, true); 3119 } 3120 3121 static inline void f2fs_i_xnid_write(struct inode *inode, nid_t xnid) 3122 { 3123 F2FS_I(inode)->i_xattr_nid = xnid; 3124 f2fs_mark_inode_dirty_sync(inode, true); 3125 } 3126 3127 static inline void f2fs_i_pino_write(struct inode *inode, nid_t pino) 3128 { 3129 F2FS_I(inode)->i_pino = pino; 3130 f2fs_mark_inode_dirty_sync(inode, true); 3131 } 3132 3133 static inline void get_inline_info(struct inode *inode, struct f2fs_inode *ri) 3134 { 3135 struct f2fs_inode_info *fi = F2FS_I(inode); 3136 3137 if (ri->i_inline & F2FS_INLINE_XATTR) 3138 set_bit(FI_INLINE_XATTR, fi->flags); 3139 if (ri->i_inline & F2FS_INLINE_DATA) 3140 set_bit(FI_INLINE_DATA, fi->flags); 3141 if (ri->i_inline & F2FS_INLINE_DENTRY) 3142 set_bit(FI_INLINE_DENTRY, fi->flags); 3143 if (ri->i_inline & F2FS_DATA_EXIST) 3144 set_bit(FI_DATA_EXIST, fi->flags); 3145 if (ri->i_inline & F2FS_INLINE_DOTS) 3146 set_bit(FI_INLINE_DOTS, fi->flags); 3147 if (ri->i_inline & F2FS_EXTRA_ATTR) 3148 set_bit(FI_EXTRA_ATTR, fi->flags); 3149 if (ri->i_inline & F2FS_PIN_FILE) 3150 set_bit(FI_PIN_FILE, fi->flags); 3151 if (ri->i_inline & F2FS_COMPRESS_RELEASED) 3152 set_bit(FI_COMPRESS_RELEASED, fi->flags); 3153 } 3154 3155 static inline void set_raw_inline(struct inode *inode, struct f2fs_inode *ri) 3156 { 3157 ri->i_inline = 0; 3158 3159 if (is_inode_flag_set(inode, FI_INLINE_XATTR)) 3160 ri->i_inline |= F2FS_INLINE_XATTR; 3161 if (is_inode_flag_set(inode, FI_INLINE_DATA)) 3162 ri->i_inline |= F2FS_INLINE_DATA; 3163 if (is_inode_flag_set(inode, FI_INLINE_DENTRY)) 3164 ri->i_inline |= F2FS_INLINE_DENTRY; 3165 if (is_inode_flag_set(inode, FI_DATA_EXIST)) 3166 ri->i_inline |= F2FS_DATA_EXIST; 3167 if (is_inode_flag_set(inode, FI_INLINE_DOTS)) 3168 ri->i_inline |= F2FS_INLINE_DOTS; 3169 if (is_inode_flag_set(inode, FI_EXTRA_ATTR)) 3170 ri->i_inline |= F2FS_EXTRA_ATTR; 3171 if (is_inode_flag_set(inode, FI_PIN_FILE)) 3172 ri->i_inline |= F2FS_PIN_FILE; 3173 if (is_inode_flag_set(inode, FI_COMPRESS_RELEASED)) 3174 ri->i_inline |= F2FS_COMPRESS_RELEASED; 3175 } 3176 3177 static inline int f2fs_has_extra_attr(struct inode *inode) 3178 { 3179 return is_inode_flag_set(inode, FI_EXTRA_ATTR); 3180 } 3181 3182 static inline int f2fs_has_inline_xattr(struct inode *inode) 3183 { 3184 return is_inode_flag_set(inode, FI_INLINE_XATTR); 3185 } 3186 3187 static inline int f2fs_compressed_file(struct inode *inode) 3188 { 3189 return S_ISREG(inode->i_mode) && 3190 is_inode_flag_set(inode, FI_COMPRESSED_FILE); 3191 } 3192 3193 static inline bool f2fs_need_compress_data(struct inode *inode) 3194 { 3195 int compress_mode = F2FS_OPTION(F2FS_I_SB(inode)).compress_mode; 3196 3197 if (!f2fs_compressed_file(inode)) 3198 return false; 3199 3200 if (compress_mode == COMPR_MODE_FS) 3201 return true; 3202 else if (compress_mode == COMPR_MODE_USER && 3203 is_inode_flag_set(inode, FI_ENABLE_COMPRESS)) 3204 return true; 3205 3206 return false; 3207 } 3208 3209 static inline unsigned int addrs_per_inode(struct inode *inode) 3210 { 3211 unsigned int addrs = CUR_ADDRS_PER_INODE(inode) - 3212 get_inline_xattr_addrs(inode); 3213 3214 if (!f2fs_compressed_file(inode)) 3215 return addrs; 3216 return ALIGN_DOWN(addrs, F2FS_I(inode)->i_cluster_size); 3217 } 3218 3219 static inline unsigned int addrs_per_block(struct inode *inode) 3220 { 3221 if (!f2fs_compressed_file(inode)) 3222 return DEF_ADDRS_PER_BLOCK; 3223 return ALIGN_DOWN(DEF_ADDRS_PER_BLOCK, F2FS_I(inode)->i_cluster_size); 3224 } 3225 3226 static inline void *inline_xattr_addr(struct inode *inode, struct page *page) 3227 { 3228 struct f2fs_inode *ri = F2FS_INODE(page); 3229 3230 return (void *)&(ri->i_addr[DEF_ADDRS_PER_INODE - 3231 get_inline_xattr_addrs(inode)]); 3232 } 3233 3234 static inline int inline_xattr_size(struct inode *inode) 3235 { 3236 if (f2fs_has_inline_xattr(inode)) 3237 return get_inline_xattr_addrs(inode) * sizeof(__le32); 3238 return 0; 3239 } 3240 3241 /* 3242 * Notice: check inline_data flag without inode page lock is unsafe. 3243 * It could change at any time by f2fs_convert_inline_page(). 3244 */ 3245 static inline int f2fs_has_inline_data(struct inode *inode) 3246 { 3247 return is_inode_flag_set(inode, FI_INLINE_DATA); 3248 } 3249 3250 static inline int f2fs_exist_data(struct inode *inode) 3251 { 3252 return is_inode_flag_set(inode, FI_DATA_EXIST); 3253 } 3254 3255 static inline int f2fs_has_inline_dots(struct inode *inode) 3256 { 3257 return is_inode_flag_set(inode, FI_INLINE_DOTS); 3258 } 3259 3260 static inline int f2fs_is_mmap_file(struct inode *inode) 3261 { 3262 return is_inode_flag_set(inode, FI_MMAP_FILE); 3263 } 3264 3265 static inline bool f2fs_is_pinned_file(struct inode *inode) 3266 { 3267 return is_inode_flag_set(inode, FI_PIN_FILE); 3268 } 3269 3270 static inline bool f2fs_is_atomic_file(struct inode *inode) 3271 { 3272 return is_inode_flag_set(inode, FI_ATOMIC_FILE); 3273 } 3274 3275 static inline bool f2fs_is_cow_file(struct inode *inode) 3276 { 3277 return is_inode_flag_set(inode, FI_COW_FILE); 3278 } 3279 3280 static inline __le32 *get_dnode_addr(struct inode *inode, 3281 struct page *node_page); 3282 static inline void *inline_data_addr(struct inode *inode, struct page *page) 3283 { 3284 __le32 *addr = get_dnode_addr(inode, page); 3285 3286 return (void *)(addr + DEF_INLINE_RESERVED_SIZE); 3287 } 3288 3289 static inline int f2fs_has_inline_dentry(struct inode *inode) 3290 { 3291 return is_inode_flag_set(inode, FI_INLINE_DENTRY); 3292 } 3293 3294 static inline int is_file(struct inode *inode, int type) 3295 { 3296 return F2FS_I(inode)->i_advise & type; 3297 } 3298 3299 static inline void set_file(struct inode *inode, int type) 3300 { 3301 if (is_file(inode, type)) 3302 return; 3303 F2FS_I(inode)->i_advise |= type; 3304 f2fs_mark_inode_dirty_sync(inode, true); 3305 } 3306 3307 static inline void clear_file(struct inode *inode, int type) 3308 { 3309 if (!is_file(inode, type)) 3310 return; 3311 F2FS_I(inode)->i_advise &= ~type; 3312 f2fs_mark_inode_dirty_sync(inode, true); 3313 } 3314 3315 static inline bool f2fs_is_time_consistent(struct inode *inode) 3316 { 3317 struct timespec64 ts = inode_get_atime(inode); 3318 3319 if (!timespec64_equal(F2FS_I(inode)->i_disk_time, &ts)) 3320 return false; 3321 ts = inode_get_ctime(inode); 3322 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 1, &ts)) 3323 return false; 3324 ts = inode_get_mtime(inode); 3325 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 2, &ts)) 3326 return false; 3327 return true; 3328 } 3329 3330 static inline bool f2fs_skip_inode_update(struct inode *inode, int dsync) 3331 { 3332 bool ret; 3333 3334 if (dsync) { 3335 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 3336 3337 spin_lock(&sbi->inode_lock[DIRTY_META]); 3338 ret = list_empty(&F2FS_I(inode)->gdirty_list); 3339 spin_unlock(&sbi->inode_lock[DIRTY_META]); 3340 return ret; 3341 } 3342 if (!is_inode_flag_set(inode, FI_AUTO_RECOVER) || 3343 file_keep_isize(inode) || 3344 i_size_read(inode) & ~PAGE_MASK) 3345 return false; 3346 3347 if (!f2fs_is_time_consistent(inode)) 3348 return false; 3349 3350 spin_lock(&F2FS_I(inode)->i_size_lock); 3351 ret = F2FS_I(inode)->last_disk_size == i_size_read(inode); 3352 spin_unlock(&F2FS_I(inode)->i_size_lock); 3353 3354 return ret; 3355 } 3356 3357 static inline bool f2fs_readonly(struct super_block *sb) 3358 { 3359 return sb_rdonly(sb); 3360 } 3361 3362 static inline bool f2fs_cp_error(struct f2fs_sb_info *sbi) 3363 { 3364 return is_set_ckpt_flags(sbi, CP_ERROR_FLAG); 3365 } 3366 3367 static inline bool is_dot_dotdot(const u8 *name, size_t len) 3368 { 3369 if (len == 1 && name[0] == '.') 3370 return true; 3371 3372 if (len == 2 && name[0] == '.' && name[1] == '.') 3373 return true; 3374 3375 return false; 3376 } 3377 3378 static inline void *f2fs_kmalloc(struct f2fs_sb_info *sbi, 3379 size_t size, gfp_t flags) 3380 { 3381 if (time_to_inject(sbi, FAULT_KMALLOC)) 3382 return NULL; 3383 3384 return kmalloc(size, flags); 3385 } 3386 3387 static inline void *f2fs_getname(struct f2fs_sb_info *sbi) 3388 { 3389 if (time_to_inject(sbi, FAULT_KMALLOC)) 3390 return NULL; 3391 3392 return __getname(); 3393 } 3394 3395 static inline void f2fs_putname(char *buf) 3396 { 3397 __putname(buf); 3398 } 3399 3400 static inline void *f2fs_kzalloc(struct f2fs_sb_info *sbi, 3401 size_t size, gfp_t flags) 3402 { 3403 return f2fs_kmalloc(sbi, size, flags | __GFP_ZERO); 3404 } 3405 3406 static inline void *f2fs_kvmalloc(struct f2fs_sb_info *sbi, 3407 size_t size, gfp_t flags) 3408 { 3409 if (time_to_inject(sbi, FAULT_KVMALLOC)) 3410 return NULL; 3411 3412 return kvmalloc(size, flags); 3413 } 3414 3415 static inline void *f2fs_kvzalloc(struct f2fs_sb_info *sbi, 3416 size_t size, gfp_t flags) 3417 { 3418 return f2fs_kvmalloc(sbi, size, flags | __GFP_ZERO); 3419 } 3420 3421 static inline int get_extra_isize(struct inode *inode) 3422 { 3423 return F2FS_I(inode)->i_extra_isize / sizeof(__le32); 3424 } 3425 3426 static inline int get_inline_xattr_addrs(struct inode *inode) 3427 { 3428 return F2FS_I(inode)->i_inline_xattr_size; 3429 } 3430 3431 static inline __le32 *get_dnode_addr(struct inode *inode, 3432 struct page *node_page) 3433 { 3434 int base = 0; 3435 3436 if (IS_INODE(node_page) && f2fs_has_extra_attr(inode)) 3437 base = get_extra_isize(inode); 3438 3439 return blkaddr_in_node(F2FS_NODE(node_page)) + base; 3440 } 3441 3442 #define f2fs_get_inode_mode(i) \ 3443 ((is_inode_flag_set(i, FI_ACL_MODE)) ? \ 3444 (F2FS_I(i)->i_acl_mode) : ((i)->i_mode)) 3445 3446 #define F2FS_MIN_EXTRA_ATTR_SIZE (sizeof(__le32)) 3447 3448 #define F2FS_TOTAL_EXTRA_ATTR_SIZE \ 3449 (offsetof(struct f2fs_inode, i_extra_end) - \ 3450 offsetof(struct f2fs_inode, i_extra_isize)) \ 3451 3452 #define F2FS_OLD_ATTRIBUTE_SIZE (offsetof(struct f2fs_inode, i_addr)) 3453 #define F2FS_FITS_IN_INODE(f2fs_inode, extra_isize, field) \ 3454 ((offsetof(typeof(*(f2fs_inode)), field) + \ 3455 sizeof((f2fs_inode)->field)) \ 3456 <= (F2FS_OLD_ATTRIBUTE_SIZE + (extra_isize))) \ 3457 3458 #define __is_large_section(sbi) ((sbi)->segs_per_sec > 1) 3459 3460 #define __is_meta_io(fio) (PAGE_TYPE_OF_BIO((fio)->type) == META) 3461 3462 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi, 3463 block_t blkaddr, int type); 3464 static inline void verify_blkaddr(struct f2fs_sb_info *sbi, 3465 block_t blkaddr, int type) 3466 { 3467 if (!f2fs_is_valid_blkaddr(sbi, blkaddr, type)) { 3468 f2fs_err(sbi, "invalid blkaddr: %u, type: %d, run fsck to fix.", 3469 blkaddr, type); 3470 f2fs_bug_on(sbi, 1); 3471 } 3472 } 3473 3474 static inline bool __is_valid_data_blkaddr(block_t blkaddr) 3475 { 3476 if (blkaddr == NEW_ADDR || blkaddr == NULL_ADDR || 3477 blkaddr == COMPRESS_ADDR) 3478 return false; 3479 return true; 3480 } 3481 3482 /* 3483 * file.c 3484 */ 3485 int f2fs_sync_file(struct file *file, loff_t start, loff_t end, int datasync); 3486 int f2fs_do_truncate_blocks(struct inode *inode, u64 from, bool lock); 3487 int f2fs_truncate_blocks(struct inode *inode, u64 from, bool lock); 3488 int f2fs_truncate(struct inode *inode); 3489 int f2fs_getattr(struct mnt_idmap *idmap, const struct path *path, 3490 struct kstat *stat, u32 request_mask, unsigned int flags); 3491 int f2fs_setattr(struct mnt_idmap *idmap, struct dentry *dentry, 3492 struct iattr *attr); 3493 int f2fs_truncate_hole(struct inode *inode, pgoff_t pg_start, pgoff_t pg_end); 3494 void f2fs_truncate_data_blocks_range(struct dnode_of_data *dn, int count); 3495 int f2fs_precache_extents(struct inode *inode); 3496 int f2fs_fileattr_get(struct dentry *dentry, struct fileattr *fa); 3497 int f2fs_fileattr_set(struct mnt_idmap *idmap, 3498 struct dentry *dentry, struct fileattr *fa); 3499 long f2fs_ioctl(struct file *filp, unsigned int cmd, unsigned long arg); 3500 long f2fs_compat_ioctl(struct file *file, unsigned int cmd, unsigned long arg); 3501 int f2fs_transfer_project_quota(struct inode *inode, kprojid_t kprojid); 3502 int f2fs_pin_file_control(struct inode *inode, bool inc); 3503 3504 /* 3505 * inode.c 3506 */ 3507 void f2fs_set_inode_flags(struct inode *inode); 3508 bool f2fs_inode_chksum_verify(struct f2fs_sb_info *sbi, struct page *page); 3509 void f2fs_inode_chksum_set(struct f2fs_sb_info *sbi, struct page *page); 3510 struct inode *f2fs_iget(struct super_block *sb, unsigned long ino); 3511 struct inode *f2fs_iget_retry(struct super_block *sb, unsigned long ino); 3512 int f2fs_try_to_free_nats(struct f2fs_sb_info *sbi, int nr_shrink); 3513 void f2fs_update_inode(struct inode *inode, struct page *node_page); 3514 void f2fs_update_inode_page(struct inode *inode); 3515 int f2fs_write_inode(struct inode *inode, struct writeback_control *wbc); 3516 void f2fs_evict_inode(struct inode *inode); 3517 void f2fs_handle_failed_inode(struct inode *inode); 3518 3519 /* 3520 * namei.c 3521 */ 3522 int f2fs_update_extension_list(struct f2fs_sb_info *sbi, const char *name, 3523 bool hot, bool set); 3524 struct dentry *f2fs_get_parent(struct dentry *child); 3525 int f2fs_get_tmpfile(struct mnt_idmap *idmap, struct inode *dir, 3526 struct inode **new_inode); 3527 3528 /* 3529 * dir.c 3530 */ 3531 int f2fs_init_casefolded_name(const struct inode *dir, 3532 struct f2fs_filename *fname); 3533 int f2fs_setup_filename(struct inode *dir, const struct qstr *iname, 3534 int lookup, struct f2fs_filename *fname); 3535 int f2fs_prepare_lookup(struct inode *dir, struct dentry *dentry, 3536 struct f2fs_filename *fname); 3537 void f2fs_free_filename(struct f2fs_filename *fname); 3538 struct f2fs_dir_entry *f2fs_find_target_dentry(const struct f2fs_dentry_ptr *d, 3539 const struct f2fs_filename *fname, int *max_slots); 3540 int f2fs_fill_dentries(struct dir_context *ctx, struct f2fs_dentry_ptr *d, 3541 unsigned int start_pos, struct fscrypt_str *fstr); 3542 void f2fs_do_make_empty_dir(struct inode *inode, struct inode *parent, 3543 struct f2fs_dentry_ptr *d); 3544 struct page *f2fs_init_inode_metadata(struct inode *inode, struct inode *dir, 3545 const struct f2fs_filename *fname, struct page *dpage); 3546 void f2fs_update_parent_metadata(struct inode *dir, struct inode *inode, 3547 unsigned int current_depth); 3548 int f2fs_room_for_filename(const void *bitmap, int slots, int max_slots); 3549 void f2fs_drop_nlink(struct inode *dir, struct inode *inode); 3550 struct f2fs_dir_entry *__f2fs_find_entry(struct inode *dir, 3551 const struct f2fs_filename *fname, 3552 struct page **res_page); 3553 struct f2fs_dir_entry *f2fs_find_entry(struct inode *dir, 3554 const struct qstr *child, struct page **res_page); 3555 struct f2fs_dir_entry *f2fs_parent_dir(struct inode *dir, struct page **p); 3556 ino_t f2fs_inode_by_name(struct inode *dir, const struct qstr *qstr, 3557 struct page **page); 3558 void f2fs_set_link(struct inode *dir, struct f2fs_dir_entry *de, 3559 struct page *page, struct inode *inode); 3560 bool f2fs_has_enough_room(struct inode *dir, struct page *ipage, 3561 const struct f2fs_filename *fname); 3562 void f2fs_update_dentry(nid_t ino, umode_t mode, struct f2fs_dentry_ptr *d, 3563 const struct fscrypt_str *name, f2fs_hash_t name_hash, 3564 unsigned int bit_pos); 3565 int f2fs_add_regular_entry(struct inode *dir, const struct f2fs_filename *fname, 3566 struct inode *inode, nid_t ino, umode_t mode); 3567 int f2fs_add_dentry(struct inode *dir, const struct f2fs_filename *fname, 3568 struct inode *inode, nid_t ino, umode_t mode); 3569 int f2fs_do_add_link(struct inode *dir, const struct qstr *name, 3570 struct inode *inode, nid_t ino, umode_t mode); 3571 void f2fs_delete_entry(struct f2fs_dir_entry *dentry, struct page *page, 3572 struct inode *dir, struct inode *inode); 3573 int f2fs_do_tmpfile(struct inode *inode, struct inode *dir); 3574 bool f2fs_empty_dir(struct inode *dir); 3575 3576 static inline int f2fs_add_link(struct dentry *dentry, struct inode *inode) 3577 { 3578 if (fscrypt_is_nokey_name(dentry)) 3579 return -ENOKEY; 3580 return f2fs_do_add_link(d_inode(dentry->d_parent), &dentry->d_name, 3581 inode, inode->i_ino, inode->i_mode); 3582 } 3583 3584 /* 3585 * super.c 3586 */ 3587 int f2fs_inode_dirtied(struct inode *inode, bool sync); 3588 void f2fs_inode_synced(struct inode *inode); 3589 int f2fs_dquot_initialize(struct inode *inode); 3590 int f2fs_enable_quota_files(struct f2fs_sb_info *sbi, bool rdonly); 3591 int f2fs_quota_sync(struct super_block *sb, int type); 3592 loff_t max_file_blocks(struct inode *inode); 3593 void f2fs_quota_off_umount(struct super_block *sb); 3594 void f2fs_save_errors(struct f2fs_sb_info *sbi, unsigned char flag); 3595 void f2fs_handle_critical_error(struct f2fs_sb_info *sbi, unsigned char reason, 3596 bool irq_context); 3597 void f2fs_handle_error(struct f2fs_sb_info *sbi, unsigned char error); 3598 void f2fs_handle_error_async(struct f2fs_sb_info *sbi, unsigned char error); 3599 int f2fs_commit_super(struct f2fs_sb_info *sbi, bool recover); 3600 int f2fs_sync_fs(struct super_block *sb, int sync); 3601 int f2fs_sanity_check_ckpt(struct f2fs_sb_info *sbi); 3602 3603 /* 3604 * hash.c 3605 */ 3606 void f2fs_hash_filename(const struct inode *dir, struct f2fs_filename *fname); 3607 3608 /* 3609 * node.c 3610 */ 3611 struct node_info; 3612 3613 int f2fs_check_nid_range(struct f2fs_sb_info *sbi, nid_t nid); 3614 bool f2fs_available_free_memory(struct f2fs_sb_info *sbi, int type); 3615 bool f2fs_in_warm_node_list(struct f2fs_sb_info *sbi, struct page *page); 3616 void f2fs_init_fsync_node_info(struct f2fs_sb_info *sbi); 3617 void f2fs_del_fsync_node_entry(struct f2fs_sb_info *sbi, struct page *page); 3618 void f2fs_reset_fsync_node_info(struct f2fs_sb_info *sbi); 3619 int f2fs_need_dentry_mark(struct f2fs_sb_info *sbi, nid_t nid); 3620 bool f2fs_is_checkpointed_node(struct f2fs_sb_info *sbi, nid_t nid); 3621 bool f2fs_need_inode_block_update(struct f2fs_sb_info *sbi, nid_t ino); 3622 int f2fs_get_node_info(struct f2fs_sb_info *sbi, nid_t nid, 3623 struct node_info *ni, bool checkpoint_context); 3624 pgoff_t f2fs_get_next_page_offset(struct dnode_of_data *dn, pgoff_t pgofs); 3625 int f2fs_get_dnode_of_data(struct dnode_of_data *dn, pgoff_t index, int mode); 3626 int f2fs_truncate_inode_blocks(struct inode *inode, pgoff_t from); 3627 int f2fs_truncate_xattr_node(struct inode *inode); 3628 int f2fs_wait_on_node_pages_writeback(struct f2fs_sb_info *sbi, 3629 unsigned int seq_id); 3630 bool f2fs_nat_bitmap_enabled(struct f2fs_sb_info *sbi); 3631 int f2fs_remove_inode_page(struct inode *inode); 3632 struct page *f2fs_new_inode_page(struct inode *inode); 3633 struct page *f2fs_new_node_page(struct dnode_of_data *dn, unsigned int ofs); 3634 void f2fs_ra_node_page(struct f2fs_sb_info *sbi, nid_t nid); 3635 struct page *f2fs_get_node_page(struct f2fs_sb_info *sbi, pgoff_t nid); 3636 struct page *f2fs_get_node_page_ra(struct page *parent, int start); 3637 int f2fs_move_node_page(struct page *node_page, int gc_type); 3638 void f2fs_flush_inline_data(struct f2fs_sb_info *sbi); 3639 int f2fs_fsync_node_pages(struct f2fs_sb_info *sbi, struct inode *inode, 3640 struct writeback_control *wbc, bool atomic, 3641 unsigned int *seq_id); 3642 int f2fs_sync_node_pages(struct f2fs_sb_info *sbi, 3643 struct writeback_control *wbc, 3644 bool do_balance, enum iostat_type io_type); 3645 int f2fs_build_free_nids(struct f2fs_sb_info *sbi, bool sync, bool mount); 3646 bool f2fs_alloc_nid(struct f2fs_sb_info *sbi, nid_t *nid); 3647 void f2fs_alloc_nid_done(struct f2fs_sb_info *sbi, nid_t nid); 3648 void f2fs_alloc_nid_failed(struct f2fs_sb_info *sbi, nid_t nid); 3649 int f2fs_try_to_free_nids(struct f2fs_sb_info *sbi, int nr_shrink); 3650 int f2fs_recover_inline_xattr(struct inode *inode, struct page *page); 3651 int f2fs_recover_xattr_data(struct inode *inode, struct page *page); 3652 int f2fs_recover_inode_page(struct f2fs_sb_info *sbi, struct page *page); 3653 int f2fs_restore_node_summary(struct f2fs_sb_info *sbi, 3654 unsigned int segno, struct f2fs_summary_block *sum); 3655 void f2fs_enable_nat_bits(struct f2fs_sb_info *sbi); 3656 int f2fs_flush_nat_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3657 int f2fs_build_node_manager(struct f2fs_sb_info *sbi); 3658 void f2fs_destroy_node_manager(struct f2fs_sb_info *sbi); 3659 int __init f2fs_create_node_manager_caches(void); 3660 void f2fs_destroy_node_manager_caches(void); 3661 3662 /* 3663 * segment.c 3664 */ 3665 bool f2fs_need_SSR(struct f2fs_sb_info *sbi); 3666 int f2fs_commit_atomic_write(struct inode *inode); 3667 void f2fs_abort_atomic_write(struct inode *inode, bool clean); 3668 void f2fs_balance_fs(struct f2fs_sb_info *sbi, bool need); 3669 void f2fs_balance_fs_bg(struct f2fs_sb_info *sbi, bool from_bg); 3670 int f2fs_issue_flush(struct f2fs_sb_info *sbi, nid_t ino); 3671 int f2fs_create_flush_cmd_control(struct f2fs_sb_info *sbi); 3672 int f2fs_flush_device_cache(struct f2fs_sb_info *sbi); 3673 void f2fs_destroy_flush_cmd_control(struct f2fs_sb_info *sbi, bool free); 3674 void f2fs_invalidate_blocks(struct f2fs_sb_info *sbi, block_t addr); 3675 bool f2fs_is_checkpointed_data(struct f2fs_sb_info *sbi, block_t blkaddr); 3676 int f2fs_start_discard_thread(struct f2fs_sb_info *sbi); 3677 void f2fs_drop_discard_cmd(struct f2fs_sb_info *sbi); 3678 void f2fs_stop_discard_thread(struct f2fs_sb_info *sbi); 3679 bool f2fs_issue_discard_timeout(struct f2fs_sb_info *sbi); 3680 void f2fs_clear_prefree_segments(struct f2fs_sb_info *sbi, 3681 struct cp_control *cpc); 3682 void f2fs_dirty_to_prefree(struct f2fs_sb_info *sbi); 3683 block_t f2fs_get_unusable_blocks(struct f2fs_sb_info *sbi); 3684 int f2fs_disable_cp_again(struct f2fs_sb_info *sbi, block_t unusable); 3685 void f2fs_release_discard_addrs(struct f2fs_sb_info *sbi); 3686 int f2fs_npages_for_summary_flush(struct f2fs_sb_info *sbi, bool for_ra); 3687 bool f2fs_segment_has_free_slot(struct f2fs_sb_info *sbi, int segno); 3688 void f2fs_init_inmem_curseg(struct f2fs_sb_info *sbi); 3689 void f2fs_save_inmem_curseg(struct f2fs_sb_info *sbi); 3690 void f2fs_restore_inmem_curseg(struct f2fs_sb_info *sbi); 3691 void f2fs_get_new_segment(struct f2fs_sb_info *sbi, 3692 unsigned int *newseg, bool new_sec, int dir); 3693 void f2fs_allocate_segment_for_resize(struct f2fs_sb_info *sbi, int type, 3694 unsigned int start, unsigned int end); 3695 void f2fs_allocate_new_section(struct f2fs_sb_info *sbi, int type, bool force); 3696 void f2fs_allocate_new_segments(struct f2fs_sb_info *sbi); 3697 int f2fs_trim_fs(struct f2fs_sb_info *sbi, struct fstrim_range *range); 3698 bool f2fs_exist_trim_candidates(struct f2fs_sb_info *sbi, 3699 struct cp_control *cpc); 3700 struct page *f2fs_get_sum_page(struct f2fs_sb_info *sbi, unsigned int segno); 3701 void f2fs_update_meta_page(struct f2fs_sb_info *sbi, void *src, 3702 block_t blk_addr); 3703 void f2fs_do_write_meta_page(struct f2fs_sb_info *sbi, struct page *page, 3704 enum iostat_type io_type); 3705 void f2fs_do_write_node_page(unsigned int nid, struct f2fs_io_info *fio); 3706 void f2fs_outplace_write_data(struct dnode_of_data *dn, 3707 struct f2fs_io_info *fio); 3708 int f2fs_inplace_write_data(struct f2fs_io_info *fio); 3709 void f2fs_do_replace_block(struct f2fs_sb_info *sbi, struct f2fs_summary *sum, 3710 block_t old_blkaddr, block_t new_blkaddr, 3711 bool recover_curseg, bool recover_newaddr, 3712 bool from_gc); 3713 void f2fs_replace_block(struct f2fs_sb_info *sbi, struct dnode_of_data *dn, 3714 block_t old_addr, block_t new_addr, 3715 unsigned char version, bool recover_curseg, 3716 bool recover_newaddr); 3717 void f2fs_allocate_data_block(struct f2fs_sb_info *sbi, struct page *page, 3718 block_t old_blkaddr, block_t *new_blkaddr, 3719 struct f2fs_summary *sum, int type, 3720 struct f2fs_io_info *fio); 3721 void f2fs_update_device_state(struct f2fs_sb_info *sbi, nid_t ino, 3722 block_t blkaddr, unsigned int blkcnt); 3723 void f2fs_wait_on_page_writeback(struct page *page, 3724 enum page_type type, bool ordered, bool locked); 3725 void f2fs_wait_on_block_writeback(struct inode *inode, block_t blkaddr); 3726 void f2fs_wait_on_block_writeback_range(struct inode *inode, block_t blkaddr, 3727 block_t len); 3728 void f2fs_write_data_summaries(struct f2fs_sb_info *sbi, block_t start_blk); 3729 void f2fs_write_node_summaries(struct f2fs_sb_info *sbi, block_t start_blk); 3730 int f2fs_lookup_journal_in_cursum(struct f2fs_journal *journal, int type, 3731 unsigned int val, int alloc); 3732 void f2fs_flush_sit_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3733 int f2fs_fix_curseg_write_pointer(struct f2fs_sb_info *sbi); 3734 int f2fs_check_write_pointer(struct f2fs_sb_info *sbi); 3735 int f2fs_build_segment_manager(struct f2fs_sb_info *sbi); 3736 void f2fs_destroy_segment_manager(struct f2fs_sb_info *sbi); 3737 int __init f2fs_create_segment_manager_caches(void); 3738 void f2fs_destroy_segment_manager_caches(void); 3739 int f2fs_rw_hint_to_seg_type(enum rw_hint hint); 3740 unsigned int f2fs_usable_segs_in_sec(struct f2fs_sb_info *sbi, 3741 unsigned int segno); 3742 unsigned int f2fs_usable_blks_in_seg(struct f2fs_sb_info *sbi, 3743 unsigned int segno); 3744 3745 #define DEF_FRAGMENT_SIZE 4 3746 #define MIN_FRAGMENT_SIZE 1 3747 #define MAX_FRAGMENT_SIZE 512 3748 3749 static inline bool f2fs_need_rand_seg(struct f2fs_sb_info *sbi) 3750 { 3751 return F2FS_OPTION(sbi).fs_mode == FS_MODE_FRAGMENT_SEG || 3752 F2FS_OPTION(sbi).fs_mode == FS_MODE_FRAGMENT_BLK; 3753 } 3754 3755 /* 3756 * checkpoint.c 3757 */ 3758 void f2fs_stop_checkpoint(struct f2fs_sb_info *sbi, bool end_io, 3759 unsigned char reason); 3760 void f2fs_flush_ckpt_thread(struct f2fs_sb_info *sbi); 3761 struct page *f2fs_grab_meta_page(struct f2fs_sb_info *sbi, pgoff_t index); 3762 struct page *f2fs_get_meta_page(struct f2fs_sb_info *sbi, pgoff_t index); 3763 struct page *f2fs_get_meta_page_retry(struct f2fs_sb_info *sbi, pgoff_t index); 3764 struct page *f2fs_get_tmp_page(struct f2fs_sb_info *sbi, pgoff_t index); 3765 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi, 3766 block_t blkaddr, int type); 3767 int f2fs_ra_meta_pages(struct f2fs_sb_info *sbi, block_t start, int nrpages, 3768 int type, bool sync); 3769 void f2fs_ra_meta_pages_cond(struct f2fs_sb_info *sbi, pgoff_t index, 3770 unsigned int ra_blocks); 3771 long f2fs_sync_meta_pages(struct f2fs_sb_info *sbi, enum page_type type, 3772 long nr_to_write, enum iostat_type io_type); 3773 void f2fs_add_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type); 3774 void f2fs_remove_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type); 3775 void f2fs_release_ino_entry(struct f2fs_sb_info *sbi, bool all); 3776 bool f2fs_exist_written_data(struct f2fs_sb_info *sbi, nid_t ino, int mode); 3777 void f2fs_set_dirty_device(struct f2fs_sb_info *sbi, nid_t ino, 3778 unsigned int devidx, int type); 3779 bool f2fs_is_dirty_device(struct f2fs_sb_info *sbi, nid_t ino, 3780 unsigned int devidx, int type); 3781 int f2fs_acquire_orphan_inode(struct f2fs_sb_info *sbi); 3782 void f2fs_release_orphan_inode(struct f2fs_sb_info *sbi); 3783 void f2fs_add_orphan_inode(struct inode *inode); 3784 void f2fs_remove_orphan_inode(struct f2fs_sb_info *sbi, nid_t ino); 3785 int f2fs_recover_orphan_inodes(struct f2fs_sb_info *sbi); 3786 int f2fs_get_valid_checkpoint(struct f2fs_sb_info *sbi); 3787 void f2fs_update_dirty_folio(struct inode *inode, struct folio *folio); 3788 void f2fs_remove_dirty_inode(struct inode *inode); 3789 int f2fs_sync_dirty_inodes(struct f2fs_sb_info *sbi, enum inode_type type, 3790 bool from_cp); 3791 void f2fs_wait_on_all_pages(struct f2fs_sb_info *sbi, int type); 3792 u64 f2fs_get_sectors_written(struct f2fs_sb_info *sbi); 3793 int f2fs_write_checkpoint(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3794 void f2fs_init_ino_entry_info(struct f2fs_sb_info *sbi); 3795 int __init f2fs_create_checkpoint_caches(void); 3796 void f2fs_destroy_checkpoint_caches(void); 3797 int f2fs_issue_checkpoint(struct f2fs_sb_info *sbi); 3798 int f2fs_start_ckpt_thread(struct f2fs_sb_info *sbi); 3799 void f2fs_stop_ckpt_thread(struct f2fs_sb_info *sbi); 3800 void f2fs_init_ckpt_req_control(struct f2fs_sb_info *sbi); 3801 3802 /* 3803 * data.c 3804 */ 3805 int __init f2fs_init_bioset(void); 3806 void f2fs_destroy_bioset(void); 3807 int f2fs_init_bio_entry_cache(void); 3808 void f2fs_destroy_bio_entry_cache(void); 3809 void f2fs_submit_read_bio(struct f2fs_sb_info *sbi, struct bio *bio, 3810 enum page_type type); 3811 int f2fs_init_write_merge_io(struct f2fs_sb_info *sbi); 3812 void f2fs_submit_merged_write(struct f2fs_sb_info *sbi, enum page_type type); 3813 void f2fs_submit_merged_write_cond(struct f2fs_sb_info *sbi, 3814 struct inode *inode, struct page *page, 3815 nid_t ino, enum page_type type); 3816 void f2fs_submit_merged_ipu_write(struct f2fs_sb_info *sbi, 3817 struct bio **bio, struct page *page); 3818 void f2fs_flush_merged_writes(struct f2fs_sb_info *sbi); 3819 int f2fs_submit_page_bio(struct f2fs_io_info *fio); 3820 int f2fs_merge_page_bio(struct f2fs_io_info *fio); 3821 void f2fs_submit_page_write(struct f2fs_io_info *fio); 3822 struct block_device *f2fs_target_device(struct f2fs_sb_info *sbi, 3823 block_t blk_addr, sector_t *sector); 3824 int f2fs_target_device_index(struct f2fs_sb_info *sbi, block_t blkaddr); 3825 void f2fs_set_data_blkaddr(struct dnode_of_data *dn, block_t blkaddr); 3826 void f2fs_update_data_blkaddr(struct dnode_of_data *dn, block_t blkaddr); 3827 int f2fs_reserve_new_blocks(struct dnode_of_data *dn, blkcnt_t count); 3828 int f2fs_reserve_new_block(struct dnode_of_data *dn); 3829 int f2fs_get_block_locked(struct dnode_of_data *dn, pgoff_t index); 3830 int f2fs_reserve_block(struct dnode_of_data *dn, pgoff_t index); 3831 struct page *f2fs_get_read_data_page(struct inode *inode, pgoff_t index, 3832 blk_opf_t op_flags, bool for_write, pgoff_t *next_pgofs); 3833 struct page *f2fs_find_data_page(struct inode *inode, pgoff_t index, 3834 pgoff_t *next_pgofs); 3835 struct page *f2fs_get_lock_data_page(struct inode *inode, pgoff_t index, 3836 bool for_write); 3837 struct page *f2fs_get_new_data_page(struct inode *inode, 3838 struct page *ipage, pgoff_t index, bool new_i_size); 3839 int f2fs_do_write_data_page(struct f2fs_io_info *fio); 3840 int f2fs_map_blocks(struct inode *inode, struct f2fs_map_blocks *map, int flag); 3841 int f2fs_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, 3842 u64 start, u64 len); 3843 int f2fs_encrypt_one_page(struct f2fs_io_info *fio); 3844 bool f2fs_should_update_inplace(struct inode *inode, struct f2fs_io_info *fio); 3845 bool f2fs_should_update_outplace(struct inode *inode, struct f2fs_io_info *fio); 3846 int f2fs_write_single_data_page(struct page *page, int *submitted, 3847 struct bio **bio, sector_t *last_block, 3848 struct writeback_control *wbc, 3849 enum iostat_type io_type, 3850 int compr_blocks, bool allow_balance); 3851 void f2fs_write_failed(struct inode *inode, loff_t to); 3852 void f2fs_invalidate_folio(struct folio *folio, size_t offset, size_t length); 3853 bool f2fs_release_folio(struct folio *folio, gfp_t wait); 3854 bool f2fs_overwrite_io(struct inode *inode, loff_t pos, size_t len); 3855 void f2fs_clear_page_cache_dirty_tag(struct page *page); 3856 int f2fs_init_post_read_processing(void); 3857 void f2fs_destroy_post_read_processing(void); 3858 int f2fs_init_post_read_wq(struct f2fs_sb_info *sbi); 3859 void f2fs_destroy_post_read_wq(struct f2fs_sb_info *sbi); 3860 extern const struct iomap_ops f2fs_iomap_ops; 3861 3862 /* 3863 * gc.c 3864 */ 3865 int f2fs_start_gc_thread(struct f2fs_sb_info *sbi); 3866 void f2fs_stop_gc_thread(struct f2fs_sb_info *sbi); 3867 block_t f2fs_start_bidx_of_node(unsigned int node_ofs, struct inode *inode); 3868 int f2fs_gc(struct f2fs_sb_info *sbi, struct f2fs_gc_control *gc_control); 3869 void f2fs_build_gc_manager(struct f2fs_sb_info *sbi); 3870 int f2fs_resize_fs(struct file *filp, __u64 block_count); 3871 int __init f2fs_create_garbage_collection_cache(void); 3872 void f2fs_destroy_garbage_collection_cache(void); 3873 /* victim selection function for cleaning and SSR */ 3874 int f2fs_get_victim(struct f2fs_sb_info *sbi, unsigned int *result, 3875 int gc_type, int type, char alloc_mode, 3876 unsigned long long age); 3877 3878 /* 3879 * recovery.c 3880 */ 3881 int f2fs_recover_fsync_data(struct f2fs_sb_info *sbi, bool check_only); 3882 bool f2fs_space_for_roll_forward(struct f2fs_sb_info *sbi); 3883 int __init f2fs_create_recovery_cache(void); 3884 void f2fs_destroy_recovery_cache(void); 3885 3886 /* 3887 * debug.c 3888 */ 3889 #ifdef CONFIG_F2FS_STAT_FS 3890 struct f2fs_stat_info { 3891 struct list_head stat_list; 3892 struct f2fs_sb_info *sbi; 3893 int all_area_segs, sit_area_segs, nat_area_segs, ssa_area_segs; 3894 int main_area_segs, main_area_sections, main_area_zones; 3895 unsigned long long hit_cached[NR_EXTENT_CACHES]; 3896 unsigned long long hit_rbtree[NR_EXTENT_CACHES]; 3897 unsigned long long total_ext[NR_EXTENT_CACHES]; 3898 unsigned long long hit_total[NR_EXTENT_CACHES]; 3899 int ext_tree[NR_EXTENT_CACHES]; 3900 int zombie_tree[NR_EXTENT_CACHES]; 3901 int ext_node[NR_EXTENT_CACHES]; 3902 /* to count memory footprint */ 3903 unsigned long long ext_mem[NR_EXTENT_CACHES]; 3904 /* for read extent cache */ 3905 unsigned long long hit_largest; 3906 /* for block age extent cache */ 3907 unsigned long long allocated_data_blocks; 3908 int ndirty_node, ndirty_dent, ndirty_meta, ndirty_imeta; 3909 int ndirty_data, ndirty_qdata; 3910 unsigned int ndirty_dirs, ndirty_files, nquota_files, ndirty_all; 3911 int nats, dirty_nats, sits, dirty_sits; 3912 int free_nids, avail_nids, alloc_nids; 3913 int total_count, utilization; 3914 int nr_wb_cp_data, nr_wb_data; 3915 int nr_rd_data, nr_rd_node, nr_rd_meta; 3916 int nr_dio_read, nr_dio_write; 3917 unsigned int io_skip_bggc, other_skip_bggc; 3918 int nr_flushing, nr_flushed, flush_list_empty; 3919 int nr_discarding, nr_discarded; 3920 int nr_discard_cmd; 3921 unsigned int undiscard_blks; 3922 int nr_issued_ckpt, nr_total_ckpt, nr_queued_ckpt; 3923 unsigned int cur_ckpt_time, peak_ckpt_time; 3924 int inline_xattr, inline_inode, inline_dir, append, update, orphans; 3925 int compr_inode, swapfile_inode; 3926 unsigned long long compr_blocks; 3927 int aw_cnt, max_aw_cnt; 3928 unsigned int valid_count, valid_node_count, valid_inode_count, discard_blks; 3929 unsigned int bimodal, avg_vblocks; 3930 int util_free, util_valid, util_invalid; 3931 int rsvd_segs, overp_segs; 3932 int dirty_count, node_pages, meta_pages, compress_pages; 3933 int compress_page_hit; 3934 int prefree_count, free_segs, free_secs; 3935 int cp_call_count[MAX_CALL_TYPE], cp_count; 3936 int gc_call_count[MAX_CALL_TYPE]; 3937 int gc_segs[2][2]; 3938 int gc_secs[2][2]; 3939 int tot_blks, data_blks, node_blks; 3940 int bg_data_blks, bg_node_blks; 3941 int curseg[NR_CURSEG_TYPE]; 3942 int cursec[NR_CURSEG_TYPE]; 3943 int curzone[NR_CURSEG_TYPE]; 3944 unsigned int dirty_seg[NR_CURSEG_TYPE]; 3945 unsigned int full_seg[NR_CURSEG_TYPE]; 3946 unsigned int valid_blks[NR_CURSEG_TYPE]; 3947 3948 unsigned int meta_count[META_MAX]; 3949 unsigned int segment_count[2]; 3950 unsigned int block_count[2]; 3951 unsigned int inplace_count; 3952 unsigned long long base_mem, cache_mem, page_mem; 3953 }; 3954 3955 static inline struct f2fs_stat_info *F2FS_STAT(struct f2fs_sb_info *sbi) 3956 { 3957 return (struct f2fs_stat_info *)sbi->stat_info; 3958 } 3959 3960 #define stat_inc_cp_call_count(sbi, foreground) \ 3961 atomic_inc(&sbi->cp_call_count[(foreground)]) 3962 #define stat_inc_cp_count(si) (F2FS_STAT(sbi)->cp_count++) 3963 #define stat_io_skip_bggc_count(sbi) ((sbi)->io_skip_bggc++) 3964 #define stat_other_skip_bggc_count(sbi) ((sbi)->other_skip_bggc++) 3965 #define stat_inc_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]++) 3966 #define stat_dec_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]--) 3967 #define stat_inc_total_hit(sbi, type) (atomic64_inc(&(sbi)->total_hit_ext[type])) 3968 #define stat_inc_rbtree_node_hit(sbi, type) (atomic64_inc(&(sbi)->read_hit_rbtree[type])) 3969 #define stat_inc_largest_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_largest)) 3970 #define stat_inc_cached_node_hit(sbi, type) (atomic64_inc(&(sbi)->read_hit_cached[type])) 3971 #define stat_inc_inline_xattr(inode) \ 3972 do { \ 3973 if (f2fs_has_inline_xattr(inode)) \ 3974 (atomic_inc(&F2FS_I_SB(inode)->inline_xattr)); \ 3975 } while (0) 3976 #define stat_dec_inline_xattr(inode) \ 3977 do { \ 3978 if (f2fs_has_inline_xattr(inode)) \ 3979 (atomic_dec(&F2FS_I_SB(inode)->inline_xattr)); \ 3980 } while (0) 3981 #define stat_inc_inline_inode(inode) \ 3982 do { \ 3983 if (f2fs_has_inline_data(inode)) \ 3984 (atomic_inc(&F2FS_I_SB(inode)->inline_inode)); \ 3985 } while (0) 3986 #define stat_dec_inline_inode(inode) \ 3987 do { \ 3988 if (f2fs_has_inline_data(inode)) \ 3989 (atomic_dec(&F2FS_I_SB(inode)->inline_inode)); \ 3990 } while (0) 3991 #define stat_inc_inline_dir(inode) \ 3992 do { \ 3993 if (f2fs_has_inline_dentry(inode)) \ 3994 (atomic_inc(&F2FS_I_SB(inode)->inline_dir)); \ 3995 } while (0) 3996 #define stat_dec_inline_dir(inode) \ 3997 do { \ 3998 if (f2fs_has_inline_dentry(inode)) \ 3999 (atomic_dec(&F2FS_I_SB(inode)->inline_dir)); \ 4000 } while (0) 4001 #define stat_inc_compr_inode(inode) \ 4002 do { \ 4003 if (f2fs_compressed_file(inode)) \ 4004 (atomic_inc(&F2FS_I_SB(inode)->compr_inode)); \ 4005 } while (0) 4006 #define stat_dec_compr_inode(inode) \ 4007 do { \ 4008 if (f2fs_compressed_file(inode)) \ 4009 (atomic_dec(&F2FS_I_SB(inode)->compr_inode)); \ 4010 } while (0) 4011 #define stat_add_compr_blocks(inode, blocks) \ 4012 (atomic64_add(blocks, &F2FS_I_SB(inode)->compr_blocks)) 4013 #define stat_sub_compr_blocks(inode, blocks) \ 4014 (atomic64_sub(blocks, &F2FS_I_SB(inode)->compr_blocks)) 4015 #define stat_inc_swapfile_inode(inode) \ 4016 (atomic_inc(&F2FS_I_SB(inode)->swapfile_inode)) 4017 #define stat_dec_swapfile_inode(inode) \ 4018 (atomic_dec(&F2FS_I_SB(inode)->swapfile_inode)) 4019 #define stat_inc_atomic_inode(inode) \ 4020 (atomic_inc(&F2FS_I_SB(inode)->atomic_files)) 4021 #define stat_dec_atomic_inode(inode) \ 4022 (atomic_dec(&F2FS_I_SB(inode)->atomic_files)) 4023 #define stat_inc_meta_count(sbi, blkaddr) \ 4024 do { \ 4025 if (blkaddr < SIT_I(sbi)->sit_base_addr) \ 4026 atomic_inc(&(sbi)->meta_count[META_CP]); \ 4027 else if (blkaddr < NM_I(sbi)->nat_blkaddr) \ 4028 atomic_inc(&(sbi)->meta_count[META_SIT]); \ 4029 else if (blkaddr < SM_I(sbi)->ssa_blkaddr) \ 4030 atomic_inc(&(sbi)->meta_count[META_NAT]); \ 4031 else if (blkaddr < SM_I(sbi)->main_blkaddr) \ 4032 atomic_inc(&(sbi)->meta_count[META_SSA]); \ 4033 } while (0) 4034 #define stat_inc_seg_type(sbi, curseg) \ 4035 ((sbi)->segment_count[(curseg)->alloc_type]++) 4036 #define stat_inc_block_count(sbi, curseg) \ 4037 ((sbi)->block_count[(curseg)->alloc_type]++) 4038 #define stat_inc_inplace_blocks(sbi) \ 4039 (atomic_inc(&(sbi)->inplace_count)) 4040 #define stat_update_max_atomic_write(inode) \ 4041 do { \ 4042 int cur = atomic_read(&F2FS_I_SB(inode)->atomic_files); \ 4043 int max = atomic_read(&F2FS_I_SB(inode)->max_aw_cnt); \ 4044 if (cur > max) \ 4045 atomic_set(&F2FS_I_SB(inode)->max_aw_cnt, cur); \ 4046 } while (0) 4047 #define stat_inc_gc_call_count(sbi, foreground) \ 4048 (F2FS_STAT(sbi)->gc_call_count[(foreground)]++) 4049 #define stat_inc_gc_sec_count(sbi, type, gc_type) \ 4050 (F2FS_STAT(sbi)->gc_secs[(type)][(gc_type)]++) 4051 #define stat_inc_gc_seg_count(sbi, type, gc_type) \ 4052 (F2FS_STAT(sbi)->gc_segs[(type)][(gc_type)]++) 4053 4054 #define stat_inc_tot_blk_count(si, blks) \ 4055 ((si)->tot_blks += (blks)) 4056 4057 #define stat_inc_data_blk_count(sbi, blks, gc_type) \ 4058 do { \ 4059 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 4060 stat_inc_tot_blk_count(si, blks); \ 4061 si->data_blks += (blks); \ 4062 si->bg_data_blks += ((gc_type) == BG_GC) ? (blks) : 0; \ 4063 } while (0) 4064 4065 #define stat_inc_node_blk_count(sbi, blks, gc_type) \ 4066 do { \ 4067 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 4068 stat_inc_tot_blk_count(si, blks); \ 4069 si->node_blks += (blks); \ 4070 si->bg_node_blks += ((gc_type) == BG_GC) ? (blks) : 0; \ 4071 } while (0) 4072 4073 int f2fs_build_stats(struct f2fs_sb_info *sbi); 4074 void f2fs_destroy_stats(struct f2fs_sb_info *sbi); 4075 void __init f2fs_create_root_stats(void); 4076 void f2fs_destroy_root_stats(void); 4077 void f2fs_update_sit_info(struct f2fs_sb_info *sbi); 4078 #else 4079 #define stat_inc_cp_call_count(sbi, foreground) do { } while (0) 4080 #define stat_inc_cp_count(sbi) do { } while (0) 4081 #define stat_io_skip_bggc_count(sbi) do { } while (0) 4082 #define stat_other_skip_bggc_count(sbi) do { } while (0) 4083 #define stat_inc_dirty_inode(sbi, type) do { } while (0) 4084 #define stat_dec_dirty_inode(sbi, type) do { } while (0) 4085 #define stat_inc_total_hit(sbi, type) do { } while (0) 4086 #define stat_inc_rbtree_node_hit(sbi, type) do { } while (0) 4087 #define stat_inc_largest_node_hit(sbi) do { } while (0) 4088 #define stat_inc_cached_node_hit(sbi, type) do { } while (0) 4089 #define stat_inc_inline_xattr(inode) do { } while (0) 4090 #define stat_dec_inline_xattr(inode) do { } while (0) 4091 #define stat_inc_inline_inode(inode) do { } while (0) 4092 #define stat_dec_inline_inode(inode) do { } while (0) 4093 #define stat_inc_inline_dir(inode) do { } while (0) 4094 #define stat_dec_inline_dir(inode) do { } while (0) 4095 #define stat_inc_compr_inode(inode) do { } while (0) 4096 #define stat_dec_compr_inode(inode) do { } while (0) 4097 #define stat_add_compr_blocks(inode, blocks) do { } while (0) 4098 #define stat_sub_compr_blocks(inode, blocks) do { } while (0) 4099 #define stat_inc_swapfile_inode(inode) do { } while (0) 4100 #define stat_dec_swapfile_inode(inode) do { } while (0) 4101 #define stat_inc_atomic_inode(inode) do { } while (0) 4102 #define stat_dec_atomic_inode(inode) do { } while (0) 4103 #define stat_update_max_atomic_write(inode) do { } while (0) 4104 #define stat_inc_meta_count(sbi, blkaddr) do { } while (0) 4105 #define stat_inc_seg_type(sbi, curseg) do { } while (0) 4106 #define stat_inc_block_count(sbi, curseg) do { } while (0) 4107 #define stat_inc_inplace_blocks(sbi) do { } while (0) 4108 #define stat_inc_gc_call_count(sbi, foreground) do { } while (0) 4109 #define stat_inc_gc_sec_count(sbi, type, gc_type) do { } while (0) 4110 #define stat_inc_gc_seg_count(sbi, type, gc_type) do { } while (0) 4111 #define stat_inc_tot_blk_count(si, blks) do { } while (0) 4112 #define stat_inc_data_blk_count(sbi, blks, gc_type) do { } while (0) 4113 #define stat_inc_node_blk_count(sbi, blks, gc_type) do { } while (0) 4114 4115 static inline int f2fs_build_stats(struct f2fs_sb_info *sbi) { return 0; } 4116 static inline void f2fs_destroy_stats(struct f2fs_sb_info *sbi) { } 4117 static inline void __init f2fs_create_root_stats(void) { } 4118 static inline void f2fs_destroy_root_stats(void) { } 4119 static inline void f2fs_update_sit_info(struct f2fs_sb_info *sbi) {} 4120 #endif 4121 4122 extern const struct file_operations f2fs_dir_operations; 4123 extern const struct file_operations f2fs_file_operations; 4124 extern const struct inode_operations f2fs_file_inode_operations; 4125 extern const struct address_space_operations f2fs_dblock_aops; 4126 extern const struct address_space_operations f2fs_node_aops; 4127 extern const struct address_space_operations f2fs_meta_aops; 4128 extern const struct inode_operations f2fs_dir_inode_operations; 4129 extern const struct inode_operations f2fs_symlink_inode_operations; 4130 extern const struct inode_operations f2fs_encrypted_symlink_inode_operations; 4131 extern const struct inode_operations f2fs_special_inode_operations; 4132 extern struct kmem_cache *f2fs_inode_entry_slab; 4133 4134 /* 4135 * inline.c 4136 */ 4137 bool f2fs_may_inline_data(struct inode *inode); 4138 bool f2fs_sanity_check_inline_data(struct inode *inode); 4139 bool f2fs_may_inline_dentry(struct inode *inode); 4140 void f2fs_do_read_inline_data(struct page *page, struct page *ipage); 4141 void f2fs_truncate_inline_inode(struct inode *inode, 4142 struct page *ipage, u64 from); 4143 int f2fs_read_inline_data(struct inode *inode, struct page *page); 4144 int f2fs_convert_inline_page(struct dnode_of_data *dn, struct page *page); 4145 int f2fs_convert_inline_inode(struct inode *inode); 4146 int f2fs_try_convert_inline_dir(struct inode *dir, struct dentry *dentry); 4147 int f2fs_write_inline_data(struct inode *inode, struct page *page); 4148 int f2fs_recover_inline_data(struct inode *inode, struct page *npage); 4149 struct f2fs_dir_entry *f2fs_find_in_inline_dir(struct inode *dir, 4150 const struct f2fs_filename *fname, 4151 struct page **res_page); 4152 int f2fs_make_empty_inline_dir(struct inode *inode, struct inode *parent, 4153 struct page *ipage); 4154 int f2fs_add_inline_entry(struct inode *dir, const struct f2fs_filename *fname, 4155 struct inode *inode, nid_t ino, umode_t mode); 4156 void f2fs_delete_inline_entry(struct f2fs_dir_entry *dentry, 4157 struct page *page, struct inode *dir, 4158 struct inode *inode); 4159 bool f2fs_empty_inline_dir(struct inode *dir); 4160 int f2fs_read_inline_dir(struct file *file, struct dir_context *ctx, 4161 struct fscrypt_str *fstr); 4162 int f2fs_inline_data_fiemap(struct inode *inode, 4163 struct fiemap_extent_info *fieinfo, 4164 __u64 start, __u64 len); 4165 4166 /* 4167 * shrinker.c 4168 */ 4169 unsigned long f2fs_shrink_count(struct shrinker *shrink, 4170 struct shrink_control *sc); 4171 unsigned long f2fs_shrink_scan(struct shrinker *shrink, 4172 struct shrink_control *sc); 4173 void f2fs_join_shrinker(struct f2fs_sb_info *sbi); 4174 void f2fs_leave_shrinker(struct f2fs_sb_info *sbi); 4175 4176 /* 4177 * extent_cache.c 4178 */ 4179 bool sanity_check_extent_cache(struct inode *inode); 4180 void f2fs_init_extent_tree(struct inode *inode); 4181 void f2fs_drop_extent_tree(struct inode *inode); 4182 void f2fs_destroy_extent_node(struct inode *inode); 4183 void f2fs_destroy_extent_tree(struct inode *inode); 4184 void f2fs_init_extent_cache_info(struct f2fs_sb_info *sbi); 4185 int __init f2fs_create_extent_cache(void); 4186 void f2fs_destroy_extent_cache(void); 4187 4188 /* read extent cache ops */ 4189 void f2fs_init_read_extent_tree(struct inode *inode, struct page *ipage); 4190 bool f2fs_lookup_read_extent_cache(struct inode *inode, pgoff_t pgofs, 4191 struct extent_info *ei); 4192 bool f2fs_lookup_read_extent_cache_block(struct inode *inode, pgoff_t index, 4193 block_t *blkaddr); 4194 void f2fs_update_read_extent_cache(struct dnode_of_data *dn); 4195 void f2fs_update_read_extent_cache_range(struct dnode_of_data *dn, 4196 pgoff_t fofs, block_t blkaddr, unsigned int len); 4197 unsigned int f2fs_shrink_read_extent_tree(struct f2fs_sb_info *sbi, 4198 int nr_shrink); 4199 4200 /* block age extent cache ops */ 4201 void f2fs_init_age_extent_tree(struct inode *inode); 4202 bool f2fs_lookup_age_extent_cache(struct inode *inode, pgoff_t pgofs, 4203 struct extent_info *ei); 4204 void f2fs_update_age_extent_cache(struct dnode_of_data *dn); 4205 void f2fs_update_age_extent_cache_range(struct dnode_of_data *dn, 4206 pgoff_t fofs, unsigned int len); 4207 unsigned int f2fs_shrink_age_extent_tree(struct f2fs_sb_info *sbi, 4208 int nr_shrink); 4209 4210 /* 4211 * sysfs.c 4212 */ 4213 #define MIN_RA_MUL 2 4214 #define MAX_RA_MUL 256 4215 4216 int __init f2fs_init_sysfs(void); 4217 void f2fs_exit_sysfs(void); 4218 int f2fs_register_sysfs(struct f2fs_sb_info *sbi); 4219 void f2fs_unregister_sysfs(struct f2fs_sb_info *sbi); 4220 4221 /* verity.c */ 4222 extern const struct fsverity_operations f2fs_verityops; 4223 4224 /* 4225 * crypto support 4226 */ 4227 static inline bool f2fs_encrypted_file(struct inode *inode) 4228 { 4229 return IS_ENCRYPTED(inode) && S_ISREG(inode->i_mode); 4230 } 4231 4232 static inline void f2fs_set_encrypted_inode(struct inode *inode) 4233 { 4234 #ifdef CONFIG_FS_ENCRYPTION 4235 file_set_encrypt(inode); 4236 f2fs_set_inode_flags(inode); 4237 #endif 4238 } 4239 4240 /* 4241 * Returns true if the reads of the inode's data need to undergo some 4242 * postprocessing step, like decryption or authenticity verification. 4243 */ 4244 static inline bool f2fs_post_read_required(struct inode *inode) 4245 { 4246 return f2fs_encrypted_file(inode) || fsverity_active(inode) || 4247 f2fs_compressed_file(inode); 4248 } 4249 4250 /* 4251 * compress.c 4252 */ 4253 #ifdef CONFIG_F2FS_FS_COMPRESSION 4254 bool f2fs_is_compressed_page(struct page *page); 4255 struct page *f2fs_compress_control_page(struct page *page); 4256 int f2fs_prepare_compress_overwrite(struct inode *inode, 4257 struct page **pagep, pgoff_t index, void **fsdata); 4258 bool f2fs_compress_write_end(struct inode *inode, void *fsdata, 4259 pgoff_t index, unsigned copied); 4260 int f2fs_truncate_partial_cluster(struct inode *inode, u64 from, bool lock); 4261 void f2fs_compress_write_end_io(struct bio *bio, struct page *page); 4262 bool f2fs_is_compress_backend_ready(struct inode *inode); 4263 bool f2fs_is_compress_level_valid(int alg, int lvl); 4264 int __init f2fs_init_compress_mempool(void); 4265 void f2fs_destroy_compress_mempool(void); 4266 void f2fs_decompress_cluster(struct decompress_io_ctx *dic, bool in_task); 4267 void f2fs_end_read_compressed_page(struct page *page, bool failed, 4268 block_t blkaddr, bool in_task); 4269 bool f2fs_cluster_is_empty(struct compress_ctx *cc); 4270 bool f2fs_cluster_can_merge_page(struct compress_ctx *cc, pgoff_t index); 4271 bool f2fs_all_cluster_page_ready(struct compress_ctx *cc, struct page **pages, 4272 int index, int nr_pages, bool uptodate); 4273 bool f2fs_sanity_check_cluster(struct dnode_of_data *dn); 4274 void f2fs_compress_ctx_add_page(struct compress_ctx *cc, struct page *page); 4275 int f2fs_write_multi_pages(struct compress_ctx *cc, 4276 int *submitted, 4277 struct writeback_control *wbc, 4278 enum iostat_type io_type); 4279 int f2fs_is_compressed_cluster(struct inode *inode, pgoff_t index); 4280 void f2fs_update_read_extent_tree_range_compressed(struct inode *inode, 4281 pgoff_t fofs, block_t blkaddr, 4282 unsigned int llen, unsigned int c_len); 4283 int f2fs_read_multi_pages(struct compress_ctx *cc, struct bio **bio_ret, 4284 unsigned nr_pages, sector_t *last_block_in_bio, 4285 bool is_readahead, bool for_write); 4286 struct decompress_io_ctx *f2fs_alloc_dic(struct compress_ctx *cc); 4287 void f2fs_decompress_end_io(struct decompress_io_ctx *dic, bool failed, 4288 bool in_task); 4289 void f2fs_put_page_dic(struct page *page, bool in_task); 4290 unsigned int f2fs_cluster_blocks_are_contiguous(struct dnode_of_data *dn); 4291 int f2fs_init_compress_ctx(struct compress_ctx *cc); 4292 void f2fs_destroy_compress_ctx(struct compress_ctx *cc, bool reuse); 4293 void f2fs_init_compress_info(struct f2fs_sb_info *sbi); 4294 int f2fs_init_compress_inode(struct f2fs_sb_info *sbi); 4295 void f2fs_destroy_compress_inode(struct f2fs_sb_info *sbi); 4296 int f2fs_init_page_array_cache(struct f2fs_sb_info *sbi); 4297 void f2fs_destroy_page_array_cache(struct f2fs_sb_info *sbi); 4298 int __init f2fs_init_compress_cache(void); 4299 void f2fs_destroy_compress_cache(void); 4300 struct address_space *COMPRESS_MAPPING(struct f2fs_sb_info *sbi); 4301 void f2fs_invalidate_compress_page(struct f2fs_sb_info *sbi, block_t blkaddr); 4302 void f2fs_cache_compressed_page(struct f2fs_sb_info *sbi, struct page *page, 4303 nid_t ino, block_t blkaddr); 4304 bool f2fs_load_compressed_page(struct f2fs_sb_info *sbi, struct page *page, 4305 block_t blkaddr); 4306 void f2fs_invalidate_compress_pages(struct f2fs_sb_info *sbi, nid_t ino); 4307 #define inc_compr_inode_stat(inode) \ 4308 do { \ 4309 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); \ 4310 sbi->compr_new_inode++; \ 4311 } while (0) 4312 #define add_compr_block_stat(inode, blocks) \ 4313 do { \ 4314 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); \ 4315 int diff = F2FS_I(inode)->i_cluster_size - blocks; \ 4316 sbi->compr_written_block += blocks; \ 4317 sbi->compr_saved_block += diff; \ 4318 } while (0) 4319 #else 4320 static inline bool f2fs_is_compressed_page(struct page *page) { return false; } 4321 static inline bool f2fs_is_compress_backend_ready(struct inode *inode) 4322 { 4323 if (!f2fs_compressed_file(inode)) 4324 return true; 4325 /* not support compression */ 4326 return false; 4327 } 4328 static inline bool f2fs_is_compress_level_valid(int alg, int lvl) { return false; } 4329 static inline struct page *f2fs_compress_control_page(struct page *page) 4330 { 4331 WARN_ON_ONCE(1); 4332 return ERR_PTR(-EINVAL); 4333 } 4334 static inline int __init f2fs_init_compress_mempool(void) { return 0; } 4335 static inline void f2fs_destroy_compress_mempool(void) { } 4336 static inline void f2fs_decompress_cluster(struct decompress_io_ctx *dic, 4337 bool in_task) { } 4338 static inline void f2fs_end_read_compressed_page(struct page *page, 4339 bool failed, block_t blkaddr, bool in_task) 4340 { 4341 WARN_ON_ONCE(1); 4342 } 4343 static inline void f2fs_put_page_dic(struct page *page, bool in_task) 4344 { 4345 WARN_ON_ONCE(1); 4346 } 4347 static inline unsigned int f2fs_cluster_blocks_are_contiguous(struct dnode_of_data *dn) { return 0; } 4348 static inline bool f2fs_sanity_check_cluster(struct dnode_of_data *dn) { return false; } 4349 static inline int f2fs_init_compress_inode(struct f2fs_sb_info *sbi) { return 0; } 4350 static inline void f2fs_destroy_compress_inode(struct f2fs_sb_info *sbi) { } 4351 static inline int f2fs_init_page_array_cache(struct f2fs_sb_info *sbi) { return 0; } 4352 static inline void f2fs_destroy_page_array_cache(struct f2fs_sb_info *sbi) { } 4353 static inline int __init f2fs_init_compress_cache(void) { return 0; } 4354 static inline void f2fs_destroy_compress_cache(void) { } 4355 static inline void f2fs_invalidate_compress_page(struct f2fs_sb_info *sbi, 4356 block_t blkaddr) { } 4357 static inline void f2fs_cache_compressed_page(struct f2fs_sb_info *sbi, 4358 struct page *page, nid_t ino, block_t blkaddr) { } 4359 static inline bool f2fs_load_compressed_page(struct f2fs_sb_info *sbi, 4360 struct page *page, block_t blkaddr) { return false; } 4361 static inline void f2fs_invalidate_compress_pages(struct f2fs_sb_info *sbi, 4362 nid_t ino) { } 4363 #define inc_compr_inode_stat(inode) do { } while (0) 4364 static inline void f2fs_update_read_extent_tree_range_compressed( 4365 struct inode *inode, 4366 pgoff_t fofs, block_t blkaddr, 4367 unsigned int llen, unsigned int c_len) { } 4368 #endif 4369 4370 static inline int set_compress_context(struct inode *inode) 4371 { 4372 #ifdef CONFIG_F2FS_FS_COMPRESSION 4373 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 4374 4375 F2FS_I(inode)->i_compress_algorithm = 4376 F2FS_OPTION(sbi).compress_algorithm; 4377 F2FS_I(inode)->i_log_cluster_size = 4378 F2FS_OPTION(sbi).compress_log_size; 4379 F2FS_I(inode)->i_compress_flag = 4380 F2FS_OPTION(sbi).compress_chksum ? 4381 BIT(COMPRESS_CHKSUM) : 0; 4382 F2FS_I(inode)->i_cluster_size = 4383 BIT(F2FS_I(inode)->i_log_cluster_size); 4384 if ((F2FS_I(inode)->i_compress_algorithm == COMPRESS_LZ4 || 4385 F2FS_I(inode)->i_compress_algorithm == COMPRESS_ZSTD) && 4386 F2FS_OPTION(sbi).compress_level) 4387 F2FS_I(inode)->i_compress_level = 4388 F2FS_OPTION(sbi).compress_level; 4389 F2FS_I(inode)->i_flags |= F2FS_COMPR_FL; 4390 set_inode_flag(inode, FI_COMPRESSED_FILE); 4391 stat_inc_compr_inode(inode); 4392 inc_compr_inode_stat(inode); 4393 f2fs_mark_inode_dirty_sync(inode, true); 4394 return 0; 4395 #else 4396 return -EOPNOTSUPP; 4397 #endif 4398 } 4399 4400 static inline bool f2fs_disable_compressed_file(struct inode *inode) 4401 { 4402 struct f2fs_inode_info *fi = F2FS_I(inode); 4403 4404 if (!f2fs_compressed_file(inode)) 4405 return true; 4406 if (S_ISREG(inode->i_mode) && F2FS_HAS_BLOCKS(inode)) 4407 return false; 4408 4409 fi->i_flags &= ~F2FS_COMPR_FL; 4410 stat_dec_compr_inode(inode); 4411 clear_inode_flag(inode, FI_COMPRESSED_FILE); 4412 f2fs_mark_inode_dirty_sync(inode, true); 4413 return true; 4414 } 4415 4416 #define F2FS_FEATURE_FUNCS(name, flagname) \ 4417 static inline bool f2fs_sb_has_##name(struct f2fs_sb_info *sbi) \ 4418 { \ 4419 return F2FS_HAS_FEATURE(sbi, F2FS_FEATURE_##flagname); \ 4420 } 4421 4422 F2FS_FEATURE_FUNCS(encrypt, ENCRYPT); 4423 F2FS_FEATURE_FUNCS(blkzoned, BLKZONED); 4424 F2FS_FEATURE_FUNCS(extra_attr, EXTRA_ATTR); 4425 F2FS_FEATURE_FUNCS(project_quota, PRJQUOTA); 4426 F2FS_FEATURE_FUNCS(inode_chksum, INODE_CHKSUM); 4427 F2FS_FEATURE_FUNCS(flexible_inline_xattr, FLEXIBLE_INLINE_XATTR); 4428 F2FS_FEATURE_FUNCS(quota_ino, QUOTA_INO); 4429 F2FS_FEATURE_FUNCS(inode_crtime, INODE_CRTIME); 4430 F2FS_FEATURE_FUNCS(lost_found, LOST_FOUND); 4431 F2FS_FEATURE_FUNCS(verity, VERITY); 4432 F2FS_FEATURE_FUNCS(sb_chksum, SB_CHKSUM); 4433 F2FS_FEATURE_FUNCS(casefold, CASEFOLD); 4434 F2FS_FEATURE_FUNCS(compression, COMPRESSION); 4435 F2FS_FEATURE_FUNCS(readonly, RO); 4436 4437 #ifdef CONFIG_BLK_DEV_ZONED 4438 static inline bool f2fs_blkz_is_seq(struct f2fs_sb_info *sbi, int devi, 4439 block_t blkaddr) 4440 { 4441 unsigned int zno = blkaddr / sbi->blocks_per_blkz; 4442 4443 return test_bit(zno, FDEV(devi).blkz_seq); 4444 } 4445 #endif 4446 4447 static inline int f2fs_bdev_index(struct f2fs_sb_info *sbi, 4448 struct block_device *bdev) 4449 { 4450 int i; 4451 4452 if (!f2fs_is_multi_device(sbi)) 4453 return 0; 4454 4455 for (i = 0; i < sbi->s_ndevs; i++) 4456 if (FDEV(i).bdev == bdev) 4457 return i; 4458 4459 WARN_ON(1); 4460 return -1; 4461 } 4462 4463 static inline bool f2fs_hw_should_discard(struct f2fs_sb_info *sbi) 4464 { 4465 return f2fs_sb_has_blkzoned(sbi); 4466 } 4467 4468 static inline bool f2fs_bdev_support_discard(struct block_device *bdev) 4469 { 4470 return bdev_max_discard_sectors(bdev) || bdev_is_zoned(bdev); 4471 } 4472 4473 static inline bool f2fs_hw_support_discard(struct f2fs_sb_info *sbi) 4474 { 4475 int i; 4476 4477 if (!f2fs_is_multi_device(sbi)) 4478 return f2fs_bdev_support_discard(sbi->sb->s_bdev); 4479 4480 for (i = 0; i < sbi->s_ndevs; i++) 4481 if (f2fs_bdev_support_discard(FDEV(i).bdev)) 4482 return true; 4483 return false; 4484 } 4485 4486 static inline bool f2fs_realtime_discard_enable(struct f2fs_sb_info *sbi) 4487 { 4488 return (test_opt(sbi, DISCARD) && f2fs_hw_support_discard(sbi)) || 4489 f2fs_hw_should_discard(sbi); 4490 } 4491 4492 static inline bool f2fs_hw_is_readonly(struct f2fs_sb_info *sbi) 4493 { 4494 int i; 4495 4496 if (!f2fs_is_multi_device(sbi)) 4497 return bdev_read_only(sbi->sb->s_bdev); 4498 4499 for (i = 0; i < sbi->s_ndevs; i++) 4500 if (bdev_read_only(FDEV(i).bdev)) 4501 return true; 4502 return false; 4503 } 4504 4505 static inline bool f2fs_dev_is_readonly(struct f2fs_sb_info *sbi) 4506 { 4507 return f2fs_sb_has_readonly(sbi) || f2fs_hw_is_readonly(sbi); 4508 } 4509 4510 static inline bool f2fs_lfs_mode(struct f2fs_sb_info *sbi) 4511 { 4512 return F2FS_OPTION(sbi).fs_mode == FS_MODE_LFS; 4513 } 4514 4515 static inline bool f2fs_low_mem_mode(struct f2fs_sb_info *sbi) 4516 { 4517 return F2FS_OPTION(sbi).memory_mode == MEMORY_MODE_LOW; 4518 } 4519 4520 static inline bool f2fs_may_compress(struct inode *inode) 4521 { 4522 if (IS_SWAPFILE(inode) || f2fs_is_pinned_file(inode) || 4523 f2fs_is_atomic_file(inode) || f2fs_has_inline_data(inode) || 4524 f2fs_is_mmap_file(inode)) 4525 return false; 4526 return S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode); 4527 } 4528 4529 static inline void f2fs_i_compr_blocks_update(struct inode *inode, 4530 u64 blocks, bool add) 4531 { 4532 struct f2fs_inode_info *fi = F2FS_I(inode); 4533 int diff = fi->i_cluster_size - blocks; 4534 4535 /* don't update i_compr_blocks if saved blocks were released */ 4536 if (!add && !atomic_read(&fi->i_compr_blocks)) 4537 return; 4538 4539 if (add) { 4540 atomic_add(diff, &fi->i_compr_blocks); 4541 stat_add_compr_blocks(inode, diff); 4542 } else { 4543 atomic_sub(diff, &fi->i_compr_blocks); 4544 stat_sub_compr_blocks(inode, diff); 4545 } 4546 f2fs_mark_inode_dirty_sync(inode, true); 4547 } 4548 4549 static inline bool f2fs_allow_multi_device_dio(struct f2fs_sb_info *sbi, 4550 int flag) 4551 { 4552 if (!f2fs_is_multi_device(sbi)) 4553 return false; 4554 if (flag != F2FS_GET_BLOCK_DIO) 4555 return false; 4556 return sbi->aligned_blksize; 4557 } 4558 4559 static inline bool f2fs_need_verity(const struct inode *inode, pgoff_t idx) 4560 { 4561 return fsverity_active(inode) && 4562 idx < DIV_ROUND_UP(inode->i_size, PAGE_SIZE); 4563 } 4564 4565 #ifdef CONFIG_F2FS_FAULT_INJECTION 4566 extern void f2fs_build_fault_attr(struct f2fs_sb_info *sbi, unsigned int rate, 4567 unsigned int type); 4568 #else 4569 #define f2fs_build_fault_attr(sbi, rate, type) do { } while (0) 4570 #endif 4571 4572 static inline bool is_journalled_quota(struct f2fs_sb_info *sbi) 4573 { 4574 #ifdef CONFIG_QUOTA 4575 if (f2fs_sb_has_quota_ino(sbi)) 4576 return true; 4577 if (F2FS_OPTION(sbi).s_qf_names[USRQUOTA] || 4578 F2FS_OPTION(sbi).s_qf_names[GRPQUOTA] || 4579 F2FS_OPTION(sbi).s_qf_names[PRJQUOTA]) 4580 return true; 4581 #endif 4582 return false; 4583 } 4584 4585 static inline bool f2fs_block_unit_discard(struct f2fs_sb_info *sbi) 4586 { 4587 return F2FS_OPTION(sbi).discard_unit == DISCARD_UNIT_BLOCK; 4588 } 4589 4590 static inline void f2fs_io_schedule_timeout(long timeout) 4591 { 4592 set_current_state(TASK_UNINTERRUPTIBLE); 4593 io_schedule_timeout(timeout); 4594 } 4595 4596 static inline void f2fs_handle_page_eio(struct f2fs_sb_info *sbi, pgoff_t ofs, 4597 enum page_type type) 4598 { 4599 if (unlikely(f2fs_cp_error(sbi))) 4600 return; 4601 4602 if (ofs == sbi->page_eio_ofs[type]) { 4603 if (sbi->page_eio_cnt[type]++ == MAX_RETRY_PAGE_EIO) 4604 set_ckpt_flags(sbi, CP_ERROR_FLAG); 4605 } else { 4606 sbi->page_eio_ofs[type] = ofs; 4607 sbi->page_eio_cnt[type] = 0; 4608 } 4609 } 4610 4611 static inline bool f2fs_is_readonly(struct f2fs_sb_info *sbi) 4612 { 4613 return f2fs_sb_has_readonly(sbi) || f2fs_readonly(sbi->sb); 4614 } 4615 4616 static inline void f2fs_invalidate_internal_cache(struct f2fs_sb_info *sbi, 4617 block_t blkaddr) 4618 { 4619 invalidate_mapping_pages(META_MAPPING(sbi), blkaddr, blkaddr); 4620 f2fs_invalidate_compress_page(sbi, blkaddr); 4621 } 4622 4623 #define EFSBADCRC EBADMSG /* Bad CRC detected */ 4624 #define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */ 4625 4626 #endif /* _LINUX_F2FS_H */ 4627