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 int inject_rate; 73 unsigned int inject_type; 74 }; 75 76 extern const char *f2fs_fault_name[FAULT_MAX]; 77 #define IS_FAULT_SET(fi, type) ((fi)->inject_type & BIT(type)) 78 79 /* maximum retry count for injected failure */ 80 #define DEFAULT_FAILURE_RETRY_COUNT 8 81 #else 82 #define DEFAULT_FAILURE_RETRY_COUNT 1 83 #endif 84 85 /* 86 * For mount options 87 */ 88 #define F2FS_MOUNT_DISABLE_ROLL_FORWARD 0x00000001 89 #define F2FS_MOUNT_DISCARD 0x00000002 90 #define F2FS_MOUNT_NOHEAP 0x00000004 91 #define F2FS_MOUNT_XATTR_USER 0x00000008 92 #define F2FS_MOUNT_POSIX_ACL 0x00000010 93 #define F2FS_MOUNT_DISABLE_EXT_IDENTIFY 0x00000020 94 #define F2FS_MOUNT_INLINE_XATTR 0x00000040 95 #define F2FS_MOUNT_INLINE_DATA 0x00000080 96 #define F2FS_MOUNT_INLINE_DENTRY 0x00000100 97 #define F2FS_MOUNT_FLUSH_MERGE 0x00000200 98 #define F2FS_MOUNT_NOBARRIER 0x00000400 99 #define F2FS_MOUNT_FASTBOOT 0x00000800 100 #define F2FS_MOUNT_READ_EXTENT_CACHE 0x00001000 101 #define F2FS_MOUNT_DATA_FLUSH 0x00002000 102 #define F2FS_MOUNT_FAULT_INJECTION 0x00004000 103 #define F2FS_MOUNT_USRQUOTA 0x00008000 104 #define F2FS_MOUNT_GRPQUOTA 0x00010000 105 #define F2FS_MOUNT_PRJQUOTA 0x00020000 106 #define F2FS_MOUNT_QUOTA 0x00040000 107 #define F2FS_MOUNT_INLINE_XATTR_SIZE 0x00080000 108 #define F2FS_MOUNT_RESERVE_ROOT 0x00100000 109 #define F2FS_MOUNT_DISABLE_CHECKPOINT 0x00200000 110 #define F2FS_MOUNT_NORECOVERY 0x00400000 111 #define F2FS_MOUNT_ATGC 0x00800000 112 #define F2FS_MOUNT_MERGE_CHECKPOINT 0x01000000 113 #define F2FS_MOUNT_GC_MERGE 0x02000000 114 #define F2FS_MOUNT_COMPRESS_CACHE 0x04000000 115 #define F2FS_MOUNT_AGE_EXTENT_CACHE 0x08000000 116 117 #define F2FS_OPTION(sbi) ((sbi)->mount_opt) 118 #define clear_opt(sbi, option) (F2FS_OPTION(sbi).opt &= ~F2FS_MOUNT_##option) 119 #define set_opt(sbi, option) (F2FS_OPTION(sbi).opt |= F2FS_MOUNT_##option) 120 #define test_opt(sbi, option) (F2FS_OPTION(sbi).opt & F2FS_MOUNT_##option) 121 122 #define ver_after(a, b) (typecheck(unsigned long long, a) && \ 123 typecheck(unsigned long long, b) && \ 124 ((long long)((a) - (b)) > 0)) 125 126 typedef u32 block_t; /* 127 * should not change u32, since it is the on-disk block 128 * address format, __le32. 129 */ 130 typedef u32 nid_t; 131 132 #define COMPRESS_EXT_NUM 16 133 134 /* 135 * An implementation of an rwsem that is explicitly unfair to readers. This 136 * prevents priority inversion when a low-priority reader acquires the read lock 137 * while sleeping on the write lock but the write lock is needed by 138 * higher-priority clients. 139 */ 140 141 struct f2fs_rwsem { 142 struct rw_semaphore internal_rwsem; 143 #ifdef CONFIG_F2FS_UNFAIR_RWSEM 144 wait_queue_head_t read_waiters; 145 #endif 146 }; 147 148 struct f2fs_mount_info { 149 unsigned int opt; 150 block_t root_reserved_blocks; /* root reserved blocks */ 151 kuid_t s_resuid; /* reserved blocks for uid */ 152 kgid_t s_resgid; /* reserved blocks for gid */ 153 int active_logs; /* # of active logs */ 154 int inline_xattr_size; /* inline xattr size */ 155 #ifdef CONFIG_F2FS_FAULT_INJECTION 156 struct f2fs_fault_info fault_info; /* For fault injection */ 157 #endif 158 #ifdef CONFIG_QUOTA 159 /* Names of quota files with journalled quota */ 160 char *s_qf_names[MAXQUOTAS]; 161 int s_jquota_fmt; /* Format of quota to use */ 162 #endif 163 /* For which write hints are passed down to block layer */ 164 int alloc_mode; /* segment allocation policy */ 165 int fsync_mode; /* fsync policy */ 166 int fs_mode; /* fs mode: LFS or ADAPTIVE */ 167 int bggc_mode; /* bggc mode: off, on or sync */ 168 int memory_mode; /* memory mode */ 169 int errors; /* errors parameter */ 170 int discard_unit; /* 171 * discard command's offset/size should 172 * be aligned to this unit: block, 173 * segment or section 174 */ 175 struct fscrypt_dummy_policy dummy_enc_policy; /* test dummy encryption */ 176 block_t unusable_cap_perc; /* percentage for cap */ 177 block_t unusable_cap; /* Amount of space allowed to be 178 * unusable when disabling checkpoint 179 */ 180 181 /* For compression */ 182 unsigned char compress_algorithm; /* algorithm type */ 183 unsigned char compress_log_size; /* cluster log size */ 184 unsigned char compress_level; /* compress level */ 185 bool compress_chksum; /* compressed data chksum */ 186 unsigned char compress_ext_cnt; /* extension count */ 187 unsigned char nocompress_ext_cnt; /* nocompress extension count */ 188 int compress_mode; /* compression mode */ 189 unsigned char extensions[COMPRESS_EXT_NUM][F2FS_EXTENSION_LEN]; /* extensions */ 190 unsigned char noextensions[COMPRESS_EXT_NUM][F2FS_EXTENSION_LEN]; /* extensions */ 191 }; 192 193 #define F2FS_FEATURE_ENCRYPT 0x00000001 194 #define F2FS_FEATURE_BLKZONED 0x00000002 195 #define F2FS_FEATURE_ATOMIC_WRITE 0x00000004 196 #define F2FS_FEATURE_EXTRA_ATTR 0x00000008 197 #define F2FS_FEATURE_PRJQUOTA 0x00000010 198 #define F2FS_FEATURE_INODE_CHKSUM 0x00000020 199 #define F2FS_FEATURE_FLEXIBLE_INLINE_XATTR 0x00000040 200 #define F2FS_FEATURE_QUOTA_INO 0x00000080 201 #define F2FS_FEATURE_INODE_CRTIME 0x00000100 202 #define F2FS_FEATURE_LOST_FOUND 0x00000200 203 #define F2FS_FEATURE_VERITY 0x00000400 204 #define F2FS_FEATURE_SB_CHKSUM 0x00000800 205 #define F2FS_FEATURE_CASEFOLD 0x00001000 206 #define F2FS_FEATURE_COMPRESSION 0x00002000 207 #define F2FS_FEATURE_RO 0x00004000 208 209 #define __F2FS_HAS_FEATURE(raw_super, mask) \ 210 ((raw_super->feature & cpu_to_le32(mask)) != 0) 211 #define F2FS_HAS_FEATURE(sbi, mask) __F2FS_HAS_FEATURE(sbi->raw_super, mask) 212 213 /* 214 * Default values for user and/or group using reserved blocks 215 */ 216 #define F2FS_DEF_RESUID 0 217 #define F2FS_DEF_RESGID 0 218 219 /* 220 * For checkpoint manager 221 */ 222 enum { 223 NAT_BITMAP, 224 SIT_BITMAP 225 }; 226 227 #define CP_UMOUNT 0x00000001 228 #define CP_FASTBOOT 0x00000002 229 #define CP_SYNC 0x00000004 230 #define CP_RECOVERY 0x00000008 231 #define CP_DISCARD 0x00000010 232 #define CP_TRIMMED 0x00000020 233 #define CP_PAUSE 0x00000040 234 #define CP_RESIZE 0x00000080 235 236 #define DEF_MAX_DISCARD_REQUEST 8 /* issue 8 discards per round */ 237 #define DEF_MIN_DISCARD_ISSUE_TIME 50 /* 50 ms, if exists */ 238 #define DEF_MID_DISCARD_ISSUE_TIME 500 /* 500 ms, if device busy */ 239 #define DEF_MAX_DISCARD_ISSUE_TIME 60000 /* 60 s, if no candidates */ 240 #define DEF_DISCARD_URGENT_UTIL 80 /* do more discard over 80% */ 241 #define DEF_CP_INTERVAL 60 /* 60 secs */ 242 #define DEF_IDLE_INTERVAL 5 /* 5 secs */ 243 #define DEF_DISABLE_INTERVAL 5 /* 5 secs */ 244 #define DEF_DISABLE_QUICK_INTERVAL 1 /* 1 secs */ 245 #define DEF_UMOUNT_DISCARD_TIMEOUT 5 /* 5 secs */ 246 247 struct cp_control { 248 int reason; 249 __u64 trim_start; 250 __u64 trim_end; 251 __u64 trim_minlen; 252 }; 253 254 /* 255 * indicate meta/data type 256 */ 257 enum { 258 META_CP, 259 META_NAT, 260 META_SIT, 261 META_SSA, 262 META_MAX, 263 META_POR, 264 DATA_GENERIC, /* check range only */ 265 DATA_GENERIC_ENHANCE, /* strong check on range and segment bitmap */ 266 DATA_GENERIC_ENHANCE_READ, /* 267 * strong check on range and segment 268 * bitmap but no warning due to race 269 * condition of read on truncated area 270 * by extent_cache 271 */ 272 DATA_GENERIC_ENHANCE_UPDATE, /* 273 * strong check on range and segment 274 * bitmap for update case 275 */ 276 META_GENERIC, 277 }; 278 279 /* for the list of ino */ 280 enum { 281 ORPHAN_INO, /* for orphan ino list */ 282 APPEND_INO, /* for append ino list */ 283 UPDATE_INO, /* for update ino list */ 284 TRANS_DIR_INO, /* for transactions dir ino list */ 285 XATTR_DIR_INO, /* for xattr updated dir ino list */ 286 FLUSH_INO, /* for multiple device flushing */ 287 MAX_INO_ENTRY, /* max. list */ 288 }; 289 290 struct ino_entry { 291 struct list_head list; /* list head */ 292 nid_t ino; /* inode number */ 293 unsigned int dirty_device; /* dirty device bitmap */ 294 }; 295 296 /* for the list of inodes to be GCed */ 297 struct inode_entry { 298 struct list_head list; /* list head */ 299 struct inode *inode; /* vfs inode pointer */ 300 }; 301 302 struct fsync_node_entry { 303 struct list_head list; /* list head */ 304 struct page *page; /* warm node page pointer */ 305 unsigned int seq_id; /* sequence id */ 306 }; 307 308 struct ckpt_req { 309 struct completion wait; /* completion for checkpoint done */ 310 struct llist_node llnode; /* llist_node to be linked in wait queue */ 311 int ret; /* return code of checkpoint */ 312 ktime_t queue_time; /* request queued time */ 313 }; 314 315 struct ckpt_req_control { 316 struct task_struct *f2fs_issue_ckpt; /* checkpoint task */ 317 int ckpt_thread_ioprio; /* checkpoint merge thread ioprio */ 318 wait_queue_head_t ckpt_wait_queue; /* waiting queue for wake-up */ 319 atomic_t issued_ckpt; /* # of actually issued ckpts */ 320 atomic_t total_ckpt; /* # of total ckpts */ 321 atomic_t queued_ckpt; /* # of queued ckpts */ 322 struct llist_head issue_list; /* list for command issue */ 323 spinlock_t stat_lock; /* lock for below checkpoint time stats */ 324 unsigned int cur_time; /* cur wait time in msec for currently issued checkpoint */ 325 unsigned int peak_time; /* peak wait time in msec until now */ 326 }; 327 328 /* for the bitmap indicate blocks to be discarded */ 329 struct discard_entry { 330 struct list_head list; /* list head */ 331 block_t start_blkaddr; /* start blockaddr of current segment */ 332 unsigned char discard_map[SIT_VBLOCK_MAP_SIZE]; /* segment discard bitmap */ 333 }; 334 335 /* minimum discard granularity, unit: block count */ 336 #define MIN_DISCARD_GRANULARITY 1 337 /* default discard granularity of inner discard thread, unit: block count */ 338 #define DEFAULT_DISCARD_GRANULARITY 16 339 /* default maximum discard granularity of ordered discard, unit: block count */ 340 #define DEFAULT_MAX_ORDERED_DISCARD_GRANULARITY 16 341 342 /* max discard pend list number */ 343 #define MAX_PLIST_NUM 512 344 #define plist_idx(blk_num) ((blk_num) >= MAX_PLIST_NUM ? \ 345 (MAX_PLIST_NUM - 1) : ((blk_num) - 1)) 346 347 enum { 348 D_PREP, /* initial */ 349 D_PARTIAL, /* partially submitted */ 350 D_SUBMIT, /* all submitted */ 351 D_DONE, /* finished */ 352 }; 353 354 struct discard_info { 355 block_t lstart; /* logical start address */ 356 block_t len; /* length */ 357 block_t start; /* actual start address in dev */ 358 }; 359 360 struct discard_cmd { 361 struct rb_node rb_node; /* rb node located in rb-tree */ 362 struct discard_info di; /* discard info */ 363 struct list_head list; /* command list */ 364 struct completion wait; /* compleation */ 365 struct block_device *bdev; /* bdev */ 366 unsigned short ref; /* reference count */ 367 unsigned char state; /* state */ 368 unsigned char queued; /* queued discard */ 369 int error; /* bio error */ 370 spinlock_t lock; /* for state/bio_ref updating */ 371 unsigned short bio_ref; /* bio reference count */ 372 }; 373 374 enum { 375 DPOLICY_BG, 376 DPOLICY_FORCE, 377 DPOLICY_FSTRIM, 378 DPOLICY_UMOUNT, 379 MAX_DPOLICY, 380 }; 381 382 struct discard_policy { 383 int type; /* type of discard */ 384 unsigned int min_interval; /* used for candidates exist */ 385 unsigned int mid_interval; /* used for device busy */ 386 unsigned int max_interval; /* used for candidates not exist */ 387 unsigned int max_requests; /* # of discards issued per round */ 388 unsigned int io_aware_gran; /* minimum granularity discard not be aware of I/O */ 389 bool io_aware; /* issue discard in idle time */ 390 bool sync; /* submit discard with REQ_SYNC flag */ 391 bool ordered; /* issue discard by lba order */ 392 bool timeout; /* discard timeout for put_super */ 393 unsigned int granularity; /* discard granularity */ 394 }; 395 396 struct discard_cmd_control { 397 struct task_struct *f2fs_issue_discard; /* discard thread */ 398 struct list_head entry_list; /* 4KB discard entry list */ 399 struct list_head pend_list[MAX_PLIST_NUM];/* store pending entries */ 400 struct list_head wait_list; /* store on-flushing entries */ 401 struct list_head fstrim_list; /* in-flight discard from fstrim */ 402 wait_queue_head_t discard_wait_queue; /* waiting queue for wake-up */ 403 struct mutex cmd_lock; 404 unsigned int nr_discards; /* # of discards in the list */ 405 unsigned int max_discards; /* max. discards to be issued */ 406 unsigned int max_discard_request; /* max. discard request per round */ 407 unsigned int min_discard_issue_time; /* min. interval between discard issue */ 408 unsigned int mid_discard_issue_time; /* mid. interval between discard issue */ 409 unsigned int max_discard_issue_time; /* max. interval between discard issue */ 410 unsigned int discard_io_aware_gran; /* minimum discard granularity not be aware of I/O */ 411 unsigned int discard_urgent_util; /* utilization which issue discard proactively */ 412 unsigned int discard_granularity; /* discard granularity */ 413 unsigned int max_ordered_discard; /* maximum discard granularity issued by lba order */ 414 unsigned int undiscard_blks; /* # of undiscard blocks */ 415 unsigned int next_pos; /* next discard position */ 416 atomic_t issued_discard; /* # of issued discard */ 417 atomic_t queued_discard; /* # of queued discard */ 418 atomic_t discard_cmd_cnt; /* # of cached cmd count */ 419 struct rb_root_cached root; /* root of discard rb-tree */ 420 bool rbtree_check; /* config for consistence check */ 421 bool discard_wake; /* to wake up discard thread */ 422 }; 423 424 /* for the list of fsync inodes, used only during recovery */ 425 struct fsync_inode_entry { 426 struct list_head list; /* list head */ 427 struct inode *inode; /* vfs inode pointer */ 428 block_t blkaddr; /* block address locating the last fsync */ 429 block_t last_dentry; /* block address locating the last dentry */ 430 }; 431 432 #define nats_in_cursum(jnl) (le16_to_cpu((jnl)->n_nats)) 433 #define sits_in_cursum(jnl) (le16_to_cpu((jnl)->n_sits)) 434 435 #define nat_in_journal(jnl, i) ((jnl)->nat_j.entries[i].ne) 436 #define nid_in_journal(jnl, i) ((jnl)->nat_j.entries[i].nid) 437 #define sit_in_journal(jnl, i) ((jnl)->sit_j.entries[i].se) 438 #define segno_in_journal(jnl, i) ((jnl)->sit_j.entries[i].segno) 439 440 #define MAX_NAT_JENTRIES(jnl) (NAT_JOURNAL_ENTRIES - nats_in_cursum(jnl)) 441 #define MAX_SIT_JENTRIES(jnl) (SIT_JOURNAL_ENTRIES - sits_in_cursum(jnl)) 442 443 static inline int update_nats_in_cursum(struct f2fs_journal *journal, int i) 444 { 445 int before = nats_in_cursum(journal); 446 447 journal->n_nats = cpu_to_le16(before + i); 448 return before; 449 } 450 451 static inline int update_sits_in_cursum(struct f2fs_journal *journal, int i) 452 { 453 int before = sits_in_cursum(journal); 454 455 journal->n_sits = cpu_to_le16(before + i); 456 return before; 457 } 458 459 static inline bool __has_cursum_space(struct f2fs_journal *journal, 460 int size, int type) 461 { 462 if (type == NAT_JOURNAL) 463 return size <= MAX_NAT_JENTRIES(journal); 464 return size <= MAX_SIT_JENTRIES(journal); 465 } 466 467 /* for inline stuff */ 468 #define DEF_INLINE_RESERVED_SIZE 1 469 static inline int get_extra_isize(struct inode *inode); 470 static inline int get_inline_xattr_addrs(struct inode *inode); 471 #define MAX_INLINE_DATA(inode) (sizeof(__le32) * \ 472 (CUR_ADDRS_PER_INODE(inode) - \ 473 get_inline_xattr_addrs(inode) - \ 474 DEF_INLINE_RESERVED_SIZE)) 475 476 /* for inline dir */ 477 #define NR_INLINE_DENTRY(inode) (MAX_INLINE_DATA(inode) * BITS_PER_BYTE / \ 478 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \ 479 BITS_PER_BYTE + 1)) 480 #define INLINE_DENTRY_BITMAP_SIZE(inode) \ 481 DIV_ROUND_UP(NR_INLINE_DENTRY(inode), BITS_PER_BYTE) 482 #define INLINE_RESERVED_SIZE(inode) (MAX_INLINE_DATA(inode) - \ 483 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \ 484 NR_INLINE_DENTRY(inode) + \ 485 INLINE_DENTRY_BITMAP_SIZE(inode))) 486 487 /* 488 * For INODE and NODE manager 489 */ 490 /* for directory operations */ 491 492 struct f2fs_filename { 493 /* 494 * The filename the user specified. This is NULL for some 495 * filesystem-internal operations, e.g. converting an inline directory 496 * to a non-inline one, or roll-forward recovering an encrypted dentry. 497 */ 498 const struct qstr *usr_fname; 499 500 /* 501 * The on-disk filename. For encrypted directories, this is encrypted. 502 * This may be NULL for lookups in an encrypted dir without the key. 503 */ 504 struct fscrypt_str disk_name; 505 506 /* The dirhash of this filename */ 507 f2fs_hash_t hash; 508 509 #ifdef CONFIG_FS_ENCRYPTION 510 /* 511 * For lookups in encrypted directories: either the buffer backing 512 * disk_name, or a buffer that holds the decoded no-key name. 513 */ 514 struct fscrypt_str crypto_buf; 515 #endif 516 #if IS_ENABLED(CONFIG_UNICODE) 517 /* 518 * For casefolded directories: the casefolded name, but it's left NULL 519 * if the original name is not valid Unicode, if the original name is 520 * "." or "..", if the directory is both casefolded and encrypted and 521 * its encryption key is unavailable, or if the filesystem is doing an 522 * internal operation where usr_fname is also NULL. In all these cases 523 * we fall back to treating the name as an opaque byte sequence. 524 */ 525 struct fscrypt_str cf_name; 526 #endif 527 }; 528 529 struct f2fs_dentry_ptr { 530 struct inode *inode; 531 void *bitmap; 532 struct f2fs_dir_entry *dentry; 533 __u8 (*filename)[F2FS_SLOT_LEN]; 534 int max; 535 int nr_bitmap; 536 }; 537 538 static inline void make_dentry_ptr_block(struct inode *inode, 539 struct f2fs_dentry_ptr *d, struct f2fs_dentry_block *t) 540 { 541 d->inode = inode; 542 d->max = NR_DENTRY_IN_BLOCK; 543 d->nr_bitmap = SIZE_OF_DENTRY_BITMAP; 544 d->bitmap = t->dentry_bitmap; 545 d->dentry = t->dentry; 546 d->filename = t->filename; 547 } 548 549 static inline void make_dentry_ptr_inline(struct inode *inode, 550 struct f2fs_dentry_ptr *d, void *t) 551 { 552 int entry_cnt = NR_INLINE_DENTRY(inode); 553 int bitmap_size = INLINE_DENTRY_BITMAP_SIZE(inode); 554 int reserved_size = INLINE_RESERVED_SIZE(inode); 555 556 d->inode = inode; 557 d->max = entry_cnt; 558 d->nr_bitmap = bitmap_size; 559 d->bitmap = t; 560 d->dentry = t + bitmap_size + reserved_size; 561 d->filename = t + bitmap_size + reserved_size + 562 SIZE_OF_DIR_ENTRY * entry_cnt; 563 } 564 565 /* 566 * XATTR_NODE_OFFSET stores xattrs to one node block per file keeping -1 567 * as its node offset to distinguish from index node blocks. 568 * But some bits are used to mark the node block. 569 */ 570 #define XATTR_NODE_OFFSET ((((unsigned int)-1) << OFFSET_BIT_SHIFT) \ 571 >> OFFSET_BIT_SHIFT) 572 enum { 573 ALLOC_NODE, /* allocate a new node page if needed */ 574 LOOKUP_NODE, /* look up a node without readahead */ 575 LOOKUP_NODE_RA, /* 576 * look up a node with readahead called 577 * by get_data_block. 578 */ 579 }; 580 581 #define DEFAULT_RETRY_IO_COUNT 8 /* maximum retry read IO or flush count */ 582 583 /* congestion wait timeout value, default: 20ms */ 584 #define DEFAULT_IO_TIMEOUT (msecs_to_jiffies(20)) 585 586 /* maximum retry quota flush count */ 587 #define DEFAULT_RETRY_QUOTA_FLUSH_COUNT 8 588 589 /* maximum retry of EIO'ed page */ 590 #define MAX_RETRY_PAGE_EIO 100 591 592 #define F2FS_LINK_MAX 0xffffffff /* maximum link count per file */ 593 594 #define MAX_DIR_RA_PAGES 4 /* maximum ra pages of dir */ 595 596 /* dirty segments threshold for triggering CP */ 597 #define DEFAULT_DIRTY_THRESHOLD 4 598 599 #define RECOVERY_MAX_RA_BLOCKS BIO_MAX_VECS 600 #define RECOVERY_MIN_RA_BLOCKS 1 601 602 #define F2FS_ONSTACK_PAGES 16 /* nr of onstack pages */ 603 604 /* for in-memory extent cache entry */ 605 #define F2FS_MIN_EXTENT_LEN 64 /* minimum extent length */ 606 607 /* number of extent info in extent cache we try to shrink */ 608 #define READ_EXTENT_CACHE_SHRINK_NUMBER 128 609 610 /* number of age extent info in extent cache we try to shrink */ 611 #define AGE_EXTENT_CACHE_SHRINK_NUMBER 128 612 #define LAST_AGE_WEIGHT 30 613 #define SAME_AGE_REGION 1024 614 615 /* 616 * Define data block with age less than 1GB as hot data 617 * define data block with age less than 10GB but more than 1GB as warm data 618 */ 619 #define DEF_HOT_DATA_AGE_THRESHOLD 262144 620 #define DEF_WARM_DATA_AGE_THRESHOLD 2621440 621 622 /* extent cache type */ 623 enum extent_type { 624 EX_READ, 625 EX_BLOCK_AGE, 626 NR_EXTENT_CACHES, 627 }; 628 629 struct extent_info { 630 unsigned int fofs; /* start offset in a file */ 631 unsigned int len; /* length of the extent */ 632 union { 633 /* read extent_cache */ 634 struct { 635 /* start block address of the extent */ 636 block_t blk; 637 #ifdef CONFIG_F2FS_FS_COMPRESSION 638 /* physical extent length of compressed blocks */ 639 unsigned int c_len; 640 #endif 641 }; 642 /* block age extent_cache */ 643 struct { 644 /* block age of the extent */ 645 unsigned long long age; 646 /* last total blocks allocated */ 647 unsigned long long last_blocks; 648 }; 649 }; 650 }; 651 652 struct extent_node { 653 struct rb_node rb_node; /* rb node located in rb-tree */ 654 struct extent_info ei; /* extent info */ 655 struct list_head list; /* node in global extent list of sbi */ 656 struct extent_tree *et; /* extent tree pointer */ 657 }; 658 659 struct extent_tree { 660 nid_t ino; /* inode number */ 661 enum extent_type type; /* keep the extent tree type */ 662 struct rb_root_cached root; /* root of extent info rb-tree */ 663 struct extent_node *cached_en; /* recently accessed extent node */ 664 struct list_head list; /* to be used by sbi->zombie_list */ 665 rwlock_t lock; /* protect extent info rb-tree */ 666 atomic_t node_cnt; /* # of extent node in rb-tree*/ 667 bool largest_updated; /* largest extent updated */ 668 struct extent_info largest; /* largest cached extent for EX_READ */ 669 }; 670 671 struct extent_tree_info { 672 struct radix_tree_root extent_tree_root;/* cache extent cache entries */ 673 struct mutex extent_tree_lock; /* locking extent radix tree */ 674 struct list_head extent_list; /* lru list for shrinker */ 675 spinlock_t extent_lock; /* locking extent lru list */ 676 atomic_t total_ext_tree; /* extent tree count */ 677 struct list_head zombie_list; /* extent zombie tree list */ 678 atomic_t total_zombie_tree; /* extent zombie tree count */ 679 atomic_t total_ext_node; /* extent info count */ 680 }; 681 682 /* 683 * State of block returned by f2fs_map_blocks. 684 */ 685 #define F2FS_MAP_NEW (1U << 0) 686 #define F2FS_MAP_MAPPED (1U << 1) 687 #define F2FS_MAP_DELALLOC (1U << 2) 688 #define F2FS_MAP_FLAGS (F2FS_MAP_NEW | F2FS_MAP_MAPPED |\ 689 F2FS_MAP_DELALLOC) 690 691 struct f2fs_map_blocks { 692 struct block_device *m_bdev; /* for multi-device dio */ 693 block_t m_pblk; 694 block_t m_lblk; 695 unsigned int m_len; 696 unsigned int m_flags; 697 pgoff_t *m_next_pgofs; /* point next possible non-hole pgofs */ 698 pgoff_t *m_next_extent; /* point to next possible extent */ 699 int m_seg_type; 700 bool m_may_create; /* indicate it is from write path */ 701 bool m_multidev_dio; /* indicate it allows multi-device dio */ 702 }; 703 704 /* for flag in get_data_block */ 705 enum { 706 F2FS_GET_BLOCK_DEFAULT, 707 F2FS_GET_BLOCK_FIEMAP, 708 F2FS_GET_BLOCK_BMAP, 709 F2FS_GET_BLOCK_DIO, 710 F2FS_GET_BLOCK_PRE_DIO, 711 F2FS_GET_BLOCK_PRE_AIO, 712 F2FS_GET_BLOCK_PRECACHE, 713 }; 714 715 /* 716 * i_advise uses FADVISE_XXX_BIT. We can add additional hints later. 717 */ 718 #define FADVISE_COLD_BIT 0x01 719 #define FADVISE_LOST_PINO_BIT 0x02 720 #define FADVISE_ENCRYPT_BIT 0x04 721 #define FADVISE_ENC_NAME_BIT 0x08 722 #define FADVISE_KEEP_SIZE_BIT 0x10 723 #define FADVISE_HOT_BIT 0x20 724 #define FADVISE_VERITY_BIT 0x40 725 #define FADVISE_TRUNC_BIT 0x80 726 727 #define FADVISE_MODIFIABLE_BITS (FADVISE_COLD_BIT | FADVISE_HOT_BIT) 728 729 #define file_is_cold(inode) is_file(inode, FADVISE_COLD_BIT) 730 #define file_set_cold(inode) set_file(inode, FADVISE_COLD_BIT) 731 #define file_clear_cold(inode) clear_file(inode, FADVISE_COLD_BIT) 732 733 #define file_wrong_pino(inode) is_file(inode, FADVISE_LOST_PINO_BIT) 734 #define file_lost_pino(inode) set_file(inode, FADVISE_LOST_PINO_BIT) 735 #define file_got_pino(inode) clear_file(inode, FADVISE_LOST_PINO_BIT) 736 737 #define file_is_encrypt(inode) is_file(inode, FADVISE_ENCRYPT_BIT) 738 #define file_set_encrypt(inode) set_file(inode, FADVISE_ENCRYPT_BIT) 739 740 #define file_enc_name(inode) is_file(inode, FADVISE_ENC_NAME_BIT) 741 #define file_set_enc_name(inode) set_file(inode, FADVISE_ENC_NAME_BIT) 742 743 #define file_keep_isize(inode) is_file(inode, FADVISE_KEEP_SIZE_BIT) 744 #define file_set_keep_isize(inode) set_file(inode, FADVISE_KEEP_SIZE_BIT) 745 746 #define file_is_hot(inode) is_file(inode, FADVISE_HOT_BIT) 747 #define file_set_hot(inode) set_file(inode, FADVISE_HOT_BIT) 748 #define file_clear_hot(inode) clear_file(inode, FADVISE_HOT_BIT) 749 750 #define file_is_verity(inode) is_file(inode, FADVISE_VERITY_BIT) 751 #define file_set_verity(inode) set_file(inode, FADVISE_VERITY_BIT) 752 753 #define file_should_truncate(inode) is_file(inode, FADVISE_TRUNC_BIT) 754 #define file_need_truncate(inode) set_file(inode, FADVISE_TRUNC_BIT) 755 #define file_dont_truncate(inode) clear_file(inode, FADVISE_TRUNC_BIT) 756 757 #define DEF_DIR_LEVEL 0 758 759 enum { 760 GC_FAILURE_PIN, 761 MAX_GC_FAILURE 762 }; 763 764 /* used for f2fs_inode_info->flags */ 765 enum { 766 FI_NEW_INODE, /* indicate newly allocated inode */ 767 FI_DIRTY_INODE, /* indicate inode is dirty or not */ 768 FI_AUTO_RECOVER, /* indicate inode is recoverable */ 769 FI_DIRTY_DIR, /* indicate directory has dirty pages */ 770 FI_INC_LINK, /* need to increment i_nlink */ 771 FI_ACL_MODE, /* indicate acl mode */ 772 FI_NO_ALLOC, /* should not allocate any blocks */ 773 FI_FREE_NID, /* free allocated nide */ 774 FI_NO_EXTENT, /* not to use the extent cache */ 775 FI_INLINE_XATTR, /* used for inline xattr */ 776 FI_INLINE_DATA, /* used for inline data*/ 777 FI_INLINE_DENTRY, /* used for inline dentry */ 778 FI_APPEND_WRITE, /* inode has appended data */ 779 FI_UPDATE_WRITE, /* inode has in-place-update data */ 780 FI_NEED_IPU, /* used for ipu per file */ 781 FI_ATOMIC_FILE, /* indicate atomic file */ 782 FI_DATA_EXIST, /* indicate data exists */ 783 FI_SKIP_WRITES, /* should skip data page writeback */ 784 FI_OPU_WRITE, /* used for opu per file */ 785 FI_DIRTY_FILE, /* indicate regular/symlink has dirty pages */ 786 FI_PREALLOCATED_ALL, /* all blocks for write were preallocated */ 787 FI_HOT_DATA, /* indicate file is hot */ 788 FI_EXTRA_ATTR, /* indicate file has extra attribute */ 789 FI_PROJ_INHERIT, /* indicate file inherits projectid */ 790 FI_PIN_FILE, /* indicate file should not be gced */ 791 FI_VERITY_IN_PROGRESS, /* building fs-verity Merkle tree */ 792 FI_COMPRESSED_FILE, /* indicate file's data can be compressed */ 793 FI_COMPRESS_CORRUPT, /* indicate compressed cluster is corrupted */ 794 FI_MMAP_FILE, /* indicate file was mmapped */ 795 FI_ENABLE_COMPRESS, /* enable compression in "user" compression mode */ 796 FI_COMPRESS_RELEASED, /* compressed blocks were released */ 797 FI_ALIGNED_WRITE, /* enable aligned write */ 798 FI_COW_FILE, /* indicate COW file */ 799 FI_ATOMIC_COMMITTED, /* indicate atomic commit completed except disk sync */ 800 FI_ATOMIC_DIRTIED, /* indicate atomic file is dirtied */ 801 FI_ATOMIC_REPLACE, /* indicate atomic replace */ 802 FI_OPENED_FILE, /* indicate file has been opened */ 803 FI_MAX, /* max flag, never be used */ 804 }; 805 806 struct f2fs_inode_info { 807 struct inode vfs_inode; /* serve a vfs inode */ 808 unsigned long i_flags; /* keep an inode flags for ioctl */ 809 unsigned char i_advise; /* use to give file attribute hints */ 810 unsigned char i_dir_level; /* use for dentry level for large dir */ 811 unsigned int i_current_depth; /* only for directory depth */ 812 /* for gc failure statistic */ 813 unsigned int i_gc_failures[MAX_GC_FAILURE]; 814 unsigned int i_pino; /* parent inode number */ 815 umode_t i_acl_mode; /* keep file acl mode temporarily */ 816 817 /* Use below internally in f2fs*/ 818 unsigned long flags[BITS_TO_LONGS(FI_MAX)]; /* use to pass per-file flags */ 819 struct f2fs_rwsem i_sem; /* protect fi info */ 820 atomic_t dirty_pages; /* # of dirty pages */ 821 f2fs_hash_t chash; /* hash value of given file name */ 822 unsigned int clevel; /* maximum level of given file name */ 823 struct task_struct *task; /* lookup and create consistency */ 824 struct task_struct *cp_task; /* separate cp/wb IO stats*/ 825 struct task_struct *wb_task; /* indicate inode is in context of writeback */ 826 nid_t i_xattr_nid; /* node id that contains xattrs */ 827 loff_t last_disk_size; /* lastly written file size */ 828 spinlock_t i_size_lock; /* protect last_disk_size */ 829 830 #ifdef CONFIG_QUOTA 831 struct dquot __rcu *i_dquot[MAXQUOTAS]; 832 833 /* quota space reservation, managed internally by quota code */ 834 qsize_t i_reserved_quota; 835 #endif 836 struct list_head dirty_list; /* dirty list for dirs and files */ 837 struct list_head gdirty_list; /* linked in global dirty list */ 838 struct task_struct *atomic_write_task; /* store atomic write task */ 839 struct extent_tree *extent_tree[NR_EXTENT_CACHES]; 840 /* cached extent_tree entry */ 841 union { 842 struct inode *cow_inode; /* copy-on-write inode for atomic write */ 843 struct inode *atomic_inode; 844 /* point to atomic_inode, available only for cow_inode */ 845 }; 846 847 /* avoid racing between foreground op and gc */ 848 struct f2fs_rwsem i_gc_rwsem[2]; 849 struct f2fs_rwsem i_xattr_sem; /* avoid racing between reading and changing EAs */ 850 851 int i_extra_isize; /* size of extra space located in i_addr */ 852 kprojid_t i_projid; /* id for project quota */ 853 int i_inline_xattr_size; /* inline xattr size */ 854 struct timespec64 i_crtime; /* inode creation time */ 855 struct timespec64 i_disk_time[3];/* inode disk times */ 856 857 /* for file compress */ 858 atomic_t i_compr_blocks; /* # of compressed blocks */ 859 unsigned char i_compress_algorithm; /* algorithm type */ 860 unsigned char i_log_cluster_size; /* log of cluster size */ 861 unsigned char i_compress_level; /* compress level (lz4hc,zstd) */ 862 unsigned char i_compress_flag; /* compress flag */ 863 unsigned int i_cluster_size; /* cluster size */ 864 865 unsigned int atomic_write_cnt; 866 loff_t original_i_size; /* original i_size before atomic write */ 867 }; 868 869 static inline void get_read_extent_info(struct extent_info *ext, 870 struct f2fs_extent *i_ext) 871 { 872 ext->fofs = le32_to_cpu(i_ext->fofs); 873 ext->blk = le32_to_cpu(i_ext->blk); 874 ext->len = le32_to_cpu(i_ext->len); 875 } 876 877 static inline void set_raw_read_extent(struct extent_info *ext, 878 struct f2fs_extent *i_ext) 879 { 880 i_ext->fofs = cpu_to_le32(ext->fofs); 881 i_ext->blk = cpu_to_le32(ext->blk); 882 i_ext->len = cpu_to_le32(ext->len); 883 } 884 885 static inline bool __is_discard_mergeable(struct discard_info *back, 886 struct discard_info *front, unsigned int max_len) 887 { 888 return (back->lstart + back->len == front->lstart) && 889 (back->len + front->len <= max_len); 890 } 891 892 static inline bool __is_discard_back_mergeable(struct discard_info *cur, 893 struct discard_info *back, unsigned int max_len) 894 { 895 return __is_discard_mergeable(back, cur, max_len); 896 } 897 898 static inline bool __is_discard_front_mergeable(struct discard_info *cur, 899 struct discard_info *front, unsigned int max_len) 900 { 901 return __is_discard_mergeable(cur, front, max_len); 902 } 903 904 /* 905 * For free nid management 906 */ 907 enum nid_state { 908 FREE_NID, /* newly added to free nid list */ 909 PREALLOC_NID, /* it is preallocated */ 910 MAX_NID_STATE, 911 }; 912 913 enum nat_state { 914 TOTAL_NAT, 915 DIRTY_NAT, 916 RECLAIMABLE_NAT, 917 MAX_NAT_STATE, 918 }; 919 920 struct f2fs_nm_info { 921 block_t nat_blkaddr; /* base disk address of NAT */ 922 nid_t max_nid; /* maximum possible node ids */ 923 nid_t available_nids; /* # of available node ids */ 924 nid_t next_scan_nid; /* the next nid to be scanned */ 925 nid_t max_rf_node_blocks; /* max # of nodes for recovery */ 926 unsigned int ram_thresh; /* control the memory footprint */ 927 unsigned int ra_nid_pages; /* # of nid pages to be readaheaded */ 928 unsigned int dirty_nats_ratio; /* control dirty nats ratio threshold */ 929 930 /* NAT cache management */ 931 struct radix_tree_root nat_root;/* root of the nat entry cache */ 932 struct radix_tree_root nat_set_root;/* root of the nat set cache */ 933 struct f2fs_rwsem nat_tree_lock; /* protect nat entry tree */ 934 struct list_head nat_entries; /* cached nat entry list (clean) */ 935 spinlock_t nat_list_lock; /* protect clean nat entry list */ 936 unsigned int nat_cnt[MAX_NAT_STATE]; /* the # of cached nat entries */ 937 unsigned int nat_blocks; /* # of nat blocks */ 938 939 /* free node ids management */ 940 struct radix_tree_root free_nid_root;/* root of the free_nid cache */ 941 struct list_head free_nid_list; /* list for free nids excluding preallocated nids */ 942 unsigned int nid_cnt[MAX_NID_STATE]; /* the number of free node id */ 943 spinlock_t nid_list_lock; /* protect nid lists ops */ 944 struct mutex build_lock; /* lock for build free nids */ 945 unsigned char **free_nid_bitmap; 946 unsigned char *nat_block_bitmap; 947 unsigned short *free_nid_count; /* free nid count of NAT block */ 948 949 /* for checkpoint */ 950 char *nat_bitmap; /* NAT bitmap pointer */ 951 952 unsigned int nat_bits_blocks; /* # of nat bits blocks */ 953 unsigned char *nat_bits; /* NAT bits blocks */ 954 unsigned char *full_nat_bits; /* full NAT pages */ 955 unsigned char *empty_nat_bits; /* empty NAT pages */ 956 #ifdef CONFIG_F2FS_CHECK_FS 957 char *nat_bitmap_mir; /* NAT bitmap mirror */ 958 #endif 959 int bitmap_size; /* bitmap size */ 960 }; 961 962 /* 963 * this structure is used as one of function parameters. 964 * all the information are dedicated to a given direct node block determined 965 * by the data offset in a file. 966 */ 967 struct dnode_of_data { 968 struct inode *inode; /* vfs inode pointer */ 969 struct page *inode_page; /* its inode page, NULL is possible */ 970 struct page *node_page; /* cached direct node page */ 971 nid_t nid; /* node id of the direct node block */ 972 unsigned int ofs_in_node; /* data offset in the node page */ 973 bool inode_page_locked; /* inode page is locked or not */ 974 bool node_changed; /* is node block changed */ 975 char cur_level; /* level of hole node page */ 976 char max_level; /* level of current page located */ 977 block_t data_blkaddr; /* block address of the node block */ 978 }; 979 980 static inline void set_new_dnode(struct dnode_of_data *dn, struct inode *inode, 981 struct page *ipage, struct page *npage, nid_t nid) 982 { 983 memset(dn, 0, sizeof(*dn)); 984 dn->inode = inode; 985 dn->inode_page = ipage; 986 dn->node_page = npage; 987 dn->nid = nid; 988 } 989 990 /* 991 * For SIT manager 992 * 993 * By default, there are 6 active log areas across the whole main area. 994 * When considering hot and cold data separation to reduce cleaning overhead, 995 * we split 3 for data logs and 3 for node logs as hot, warm, and cold types, 996 * respectively. 997 * In the current design, you should not change the numbers intentionally. 998 * Instead, as a mount option such as active_logs=x, you can use 2, 4, and 6 999 * logs individually according to the underlying devices. (default: 6) 1000 * Just in case, on-disk layout covers maximum 16 logs that consist of 8 for 1001 * data and 8 for node logs. 1002 */ 1003 #define NR_CURSEG_DATA_TYPE (3) 1004 #define NR_CURSEG_NODE_TYPE (3) 1005 #define NR_CURSEG_INMEM_TYPE (2) 1006 #define NR_CURSEG_RO_TYPE (2) 1007 #define NR_CURSEG_PERSIST_TYPE (NR_CURSEG_DATA_TYPE + NR_CURSEG_NODE_TYPE) 1008 #define NR_CURSEG_TYPE (NR_CURSEG_INMEM_TYPE + NR_CURSEG_PERSIST_TYPE) 1009 1010 enum { 1011 CURSEG_HOT_DATA = 0, /* directory entry blocks */ 1012 CURSEG_WARM_DATA, /* data blocks */ 1013 CURSEG_COLD_DATA, /* multimedia or GCed data blocks */ 1014 CURSEG_HOT_NODE, /* direct node blocks of directory files */ 1015 CURSEG_WARM_NODE, /* direct node blocks of normal files */ 1016 CURSEG_COLD_NODE, /* indirect node blocks */ 1017 NR_PERSISTENT_LOG, /* number of persistent log */ 1018 CURSEG_COLD_DATA_PINNED = NR_PERSISTENT_LOG, 1019 /* pinned file that needs consecutive block address */ 1020 CURSEG_ALL_DATA_ATGC, /* SSR alloctor in hot/warm/cold data area */ 1021 NO_CHECK_TYPE, /* number of persistent & inmem log */ 1022 }; 1023 1024 struct flush_cmd { 1025 struct completion wait; 1026 struct llist_node llnode; 1027 nid_t ino; 1028 int ret; 1029 }; 1030 1031 struct flush_cmd_control { 1032 struct task_struct *f2fs_issue_flush; /* flush thread */ 1033 wait_queue_head_t flush_wait_queue; /* waiting queue for wake-up */ 1034 atomic_t issued_flush; /* # of issued flushes */ 1035 atomic_t queued_flush; /* # of queued flushes */ 1036 struct llist_head issue_list; /* list for command issue */ 1037 struct llist_node *dispatch_list; /* list for command dispatch */ 1038 }; 1039 1040 struct f2fs_sm_info { 1041 struct sit_info *sit_info; /* whole segment information */ 1042 struct free_segmap_info *free_info; /* free segment information */ 1043 struct dirty_seglist_info *dirty_info; /* dirty segment information */ 1044 struct curseg_info *curseg_array; /* active segment information */ 1045 1046 struct f2fs_rwsem curseg_lock; /* for preventing curseg change */ 1047 1048 block_t seg0_blkaddr; /* block address of 0'th segment */ 1049 block_t main_blkaddr; /* start block address of main area */ 1050 block_t ssa_blkaddr; /* start block address of SSA area */ 1051 1052 unsigned int segment_count; /* total # of segments */ 1053 unsigned int main_segments; /* # of segments in main area */ 1054 unsigned int reserved_segments; /* # of reserved segments */ 1055 unsigned int additional_reserved_segments;/* reserved segs for IO align feature */ 1056 unsigned int ovp_segments; /* # of overprovision segments */ 1057 1058 /* a threshold to reclaim prefree segments */ 1059 unsigned int rec_prefree_segments; 1060 1061 struct list_head sit_entry_set; /* sit entry set list */ 1062 1063 unsigned int ipu_policy; /* in-place-update policy */ 1064 unsigned int min_ipu_util; /* in-place-update threshold */ 1065 unsigned int min_fsync_blocks; /* threshold for fsync */ 1066 unsigned int min_seq_blocks; /* threshold for sequential blocks */ 1067 unsigned int min_hot_blocks; /* threshold for hot block allocation */ 1068 unsigned int min_ssr_sections; /* threshold to trigger SSR allocation */ 1069 1070 /* for flush command control */ 1071 struct flush_cmd_control *fcc_info; 1072 1073 /* for discard command control */ 1074 struct discard_cmd_control *dcc_info; 1075 }; 1076 1077 /* 1078 * For superblock 1079 */ 1080 /* 1081 * COUNT_TYPE for monitoring 1082 * 1083 * f2fs monitors the number of several block types such as on-writeback, 1084 * dirty dentry blocks, dirty node blocks, and dirty meta blocks. 1085 */ 1086 #define WB_DATA_TYPE(p, f) \ 1087 (f || f2fs_is_cp_guaranteed(p) ? F2FS_WB_CP_DATA : F2FS_WB_DATA) 1088 enum count_type { 1089 F2FS_DIRTY_DENTS, 1090 F2FS_DIRTY_DATA, 1091 F2FS_DIRTY_QDATA, 1092 F2FS_DIRTY_NODES, 1093 F2FS_DIRTY_META, 1094 F2FS_DIRTY_IMETA, 1095 F2FS_WB_CP_DATA, 1096 F2FS_WB_DATA, 1097 F2FS_RD_DATA, 1098 F2FS_RD_NODE, 1099 F2FS_RD_META, 1100 F2FS_DIO_WRITE, 1101 F2FS_DIO_READ, 1102 NR_COUNT_TYPE, 1103 }; 1104 1105 /* 1106 * The below are the page types of bios used in submit_bio(). 1107 * The available types are: 1108 * DATA User data pages. It operates as async mode. 1109 * NODE Node pages. It operates as async mode. 1110 * META FS metadata pages such as SIT, NAT, CP. 1111 * NR_PAGE_TYPE The number of page types. 1112 * META_FLUSH Make sure the previous pages are written 1113 * with waiting the bio's completion 1114 * ... Only can be used with META. 1115 */ 1116 #define PAGE_TYPE_OF_BIO(type) ((type) > META ? META : (type)) 1117 #define PAGE_TYPE_ON_MAIN(type) ((type) == DATA || (type) == NODE) 1118 enum page_type { 1119 DATA = 0, 1120 NODE = 1, /* should not change this */ 1121 META, 1122 NR_PAGE_TYPE, 1123 META_FLUSH, 1124 IPU, /* the below types are used by tracepoints only. */ 1125 OPU, 1126 }; 1127 1128 enum temp_type { 1129 HOT = 0, /* must be zero for meta bio */ 1130 WARM, 1131 COLD, 1132 NR_TEMP_TYPE, 1133 }; 1134 1135 enum need_lock_type { 1136 LOCK_REQ = 0, 1137 LOCK_DONE, 1138 LOCK_RETRY, 1139 }; 1140 1141 enum cp_reason_type { 1142 CP_NO_NEEDED, 1143 CP_NON_REGULAR, 1144 CP_COMPRESSED, 1145 CP_HARDLINK, 1146 CP_SB_NEED_CP, 1147 CP_WRONG_PINO, 1148 CP_NO_SPC_ROLL, 1149 CP_NODE_NEED_CP, 1150 CP_FASTBOOT_MODE, 1151 CP_SPEC_LOG_NUM, 1152 CP_RECOVER_DIR, 1153 CP_XATTR_DIR, 1154 }; 1155 1156 enum iostat_type { 1157 /* WRITE IO */ 1158 APP_DIRECT_IO, /* app direct write IOs */ 1159 APP_BUFFERED_IO, /* app buffered write IOs */ 1160 APP_WRITE_IO, /* app write IOs */ 1161 APP_MAPPED_IO, /* app mapped IOs */ 1162 APP_BUFFERED_CDATA_IO, /* app buffered write IOs on compressed file */ 1163 APP_MAPPED_CDATA_IO, /* app mapped write IOs on compressed file */ 1164 FS_DATA_IO, /* data IOs from kworker/fsync/reclaimer */ 1165 FS_CDATA_IO, /* data IOs from kworker/fsync/reclaimer on compressed file */ 1166 FS_NODE_IO, /* node IOs from kworker/fsync/reclaimer */ 1167 FS_META_IO, /* meta IOs from kworker/reclaimer */ 1168 FS_GC_DATA_IO, /* data IOs from forground gc */ 1169 FS_GC_NODE_IO, /* node IOs from forground gc */ 1170 FS_CP_DATA_IO, /* data IOs from checkpoint */ 1171 FS_CP_NODE_IO, /* node IOs from checkpoint */ 1172 FS_CP_META_IO, /* meta IOs from checkpoint */ 1173 1174 /* READ IO */ 1175 APP_DIRECT_READ_IO, /* app direct read IOs */ 1176 APP_BUFFERED_READ_IO, /* app buffered read IOs */ 1177 APP_READ_IO, /* app read IOs */ 1178 APP_MAPPED_READ_IO, /* app mapped read IOs */ 1179 APP_BUFFERED_CDATA_READ_IO, /* app buffered read IOs on compressed file */ 1180 APP_MAPPED_CDATA_READ_IO, /* app mapped read IOs on compressed file */ 1181 FS_DATA_READ_IO, /* data read IOs */ 1182 FS_GDATA_READ_IO, /* data read IOs from background gc */ 1183 FS_CDATA_READ_IO, /* compressed data read IOs */ 1184 FS_NODE_READ_IO, /* node read IOs */ 1185 FS_META_READ_IO, /* meta read IOs */ 1186 1187 /* other */ 1188 FS_DISCARD_IO, /* discard */ 1189 FS_FLUSH_IO, /* flush */ 1190 FS_ZONE_RESET_IO, /* zone reset */ 1191 NR_IO_TYPE, 1192 }; 1193 1194 struct f2fs_io_info { 1195 struct f2fs_sb_info *sbi; /* f2fs_sb_info pointer */ 1196 nid_t ino; /* inode number */ 1197 enum page_type type; /* contains DATA/NODE/META/META_FLUSH */ 1198 enum temp_type temp; /* contains HOT/WARM/COLD */ 1199 enum req_op op; /* contains REQ_OP_ */ 1200 blk_opf_t op_flags; /* req_flag_bits */ 1201 block_t new_blkaddr; /* new block address to be written */ 1202 block_t old_blkaddr; /* old block address before Cow */ 1203 struct page *page; /* page to be written */ 1204 struct page *encrypted_page; /* encrypted page */ 1205 struct page *compressed_page; /* compressed page */ 1206 struct list_head list; /* serialize IOs */ 1207 unsigned int compr_blocks; /* # of compressed block addresses */ 1208 unsigned int need_lock:8; /* indicate we need to lock cp_rwsem */ 1209 unsigned int version:8; /* version of the node */ 1210 unsigned int submitted:1; /* indicate IO submission */ 1211 unsigned int in_list:1; /* indicate fio is in io_list */ 1212 unsigned int is_por:1; /* indicate IO is from recovery or not */ 1213 unsigned int encrypted:1; /* indicate file is encrypted */ 1214 unsigned int meta_gc:1; /* require meta inode GC */ 1215 enum iostat_type io_type; /* io type */ 1216 struct writeback_control *io_wbc; /* writeback control */ 1217 struct bio **bio; /* bio for ipu */ 1218 sector_t *last_block; /* last block number in bio */ 1219 }; 1220 1221 struct bio_entry { 1222 struct bio *bio; 1223 struct list_head list; 1224 }; 1225 1226 #define is_read_io(rw) ((rw) == READ) 1227 struct f2fs_bio_info { 1228 struct f2fs_sb_info *sbi; /* f2fs superblock */ 1229 struct bio *bio; /* bios to merge */ 1230 sector_t last_block_in_bio; /* last block number */ 1231 struct f2fs_io_info fio; /* store buffered io info. */ 1232 #ifdef CONFIG_BLK_DEV_ZONED 1233 struct completion zone_wait; /* condition value for the previous open zone to close */ 1234 struct bio *zone_pending_bio; /* pending bio for the previous zone */ 1235 void *bi_private; /* previous bi_private for pending bio */ 1236 #endif 1237 struct f2fs_rwsem io_rwsem; /* blocking op for bio */ 1238 spinlock_t io_lock; /* serialize DATA/NODE IOs */ 1239 struct list_head io_list; /* track fios */ 1240 struct list_head bio_list; /* bio entry list head */ 1241 struct f2fs_rwsem bio_list_lock; /* lock to protect bio entry list */ 1242 }; 1243 1244 #define FDEV(i) (sbi->devs[i]) 1245 #define RDEV(i) (raw_super->devs[i]) 1246 struct f2fs_dev_info { 1247 struct block_device *bdev; 1248 char path[MAX_PATH_LEN]; 1249 unsigned int total_segments; 1250 block_t start_blk; 1251 block_t end_blk; 1252 #ifdef CONFIG_BLK_DEV_ZONED 1253 unsigned int nr_blkz; /* Total number of zones */ 1254 unsigned long *blkz_seq; /* Bitmap indicating sequential zones */ 1255 #endif 1256 }; 1257 1258 enum inode_type { 1259 DIR_INODE, /* for dirty dir inode */ 1260 FILE_INODE, /* for dirty regular/symlink inode */ 1261 DIRTY_META, /* for all dirtied inode metadata */ 1262 NR_INODE_TYPE, 1263 }; 1264 1265 /* for inner inode cache management */ 1266 struct inode_management { 1267 struct radix_tree_root ino_root; /* ino entry array */ 1268 spinlock_t ino_lock; /* for ino entry lock */ 1269 struct list_head ino_list; /* inode list head */ 1270 unsigned long ino_num; /* number of entries */ 1271 }; 1272 1273 /* for GC_AT */ 1274 struct atgc_management { 1275 bool atgc_enabled; /* ATGC is enabled or not */ 1276 struct rb_root_cached root; /* root of victim rb-tree */ 1277 struct list_head victim_list; /* linked with all victim entries */ 1278 unsigned int victim_count; /* victim count in rb-tree */ 1279 unsigned int candidate_ratio; /* candidate ratio */ 1280 unsigned int max_candidate_count; /* max candidate count */ 1281 unsigned int age_weight; /* age weight, vblock_weight = 100 - age_weight */ 1282 unsigned long long age_threshold; /* age threshold */ 1283 }; 1284 1285 struct f2fs_gc_control { 1286 unsigned int victim_segno; /* target victim segment number */ 1287 int init_gc_type; /* FG_GC or BG_GC */ 1288 bool no_bg_gc; /* check the space and stop bg_gc */ 1289 bool should_migrate_blocks; /* should migrate blocks */ 1290 bool err_gc_skipped; /* return EAGAIN if GC skipped */ 1291 unsigned int nr_free_secs; /* # of free sections to do GC */ 1292 }; 1293 1294 /* 1295 * For s_flag in struct f2fs_sb_info 1296 * Modification on enum should be synchronized with s_flag array 1297 */ 1298 enum { 1299 SBI_IS_DIRTY, /* dirty flag for checkpoint */ 1300 SBI_IS_CLOSE, /* specify unmounting */ 1301 SBI_NEED_FSCK, /* need fsck.f2fs to fix */ 1302 SBI_POR_DOING, /* recovery is doing or not */ 1303 SBI_NEED_SB_WRITE, /* need to recover superblock */ 1304 SBI_NEED_CP, /* need to checkpoint */ 1305 SBI_IS_SHUTDOWN, /* shutdown by ioctl */ 1306 SBI_IS_RECOVERED, /* recovered orphan/data */ 1307 SBI_CP_DISABLED, /* CP was disabled last mount */ 1308 SBI_CP_DISABLED_QUICK, /* CP was disabled quickly */ 1309 SBI_QUOTA_NEED_FLUSH, /* need to flush quota info in CP */ 1310 SBI_QUOTA_SKIP_FLUSH, /* skip flushing quota in current CP */ 1311 SBI_QUOTA_NEED_REPAIR, /* quota file may be corrupted */ 1312 SBI_IS_RESIZEFS, /* resizefs is in process */ 1313 SBI_IS_FREEZING, /* freezefs is in process */ 1314 SBI_IS_WRITABLE, /* remove ro mountoption transiently */ 1315 MAX_SBI_FLAG, 1316 }; 1317 1318 enum { 1319 CP_TIME, 1320 REQ_TIME, 1321 DISCARD_TIME, 1322 GC_TIME, 1323 DISABLE_TIME, 1324 UMOUNT_DISCARD_TIMEOUT, 1325 MAX_TIME, 1326 }; 1327 1328 /* Note that you need to keep synchronization with this gc_mode_names array */ 1329 enum { 1330 GC_NORMAL, 1331 GC_IDLE_CB, 1332 GC_IDLE_GREEDY, 1333 GC_IDLE_AT, 1334 GC_URGENT_HIGH, 1335 GC_URGENT_LOW, 1336 GC_URGENT_MID, 1337 MAX_GC_MODE, 1338 }; 1339 1340 enum { 1341 BGGC_MODE_ON, /* background gc is on */ 1342 BGGC_MODE_OFF, /* background gc is off */ 1343 BGGC_MODE_SYNC, /* 1344 * background gc is on, migrating blocks 1345 * like foreground gc 1346 */ 1347 }; 1348 1349 enum { 1350 FS_MODE_ADAPTIVE, /* use both lfs/ssr allocation */ 1351 FS_MODE_LFS, /* use lfs allocation only */ 1352 FS_MODE_FRAGMENT_SEG, /* segment fragmentation mode */ 1353 FS_MODE_FRAGMENT_BLK, /* block fragmentation mode */ 1354 }; 1355 1356 enum { 1357 ALLOC_MODE_DEFAULT, /* stay default */ 1358 ALLOC_MODE_REUSE, /* reuse segments as much as possible */ 1359 }; 1360 1361 enum fsync_mode { 1362 FSYNC_MODE_POSIX, /* fsync follows posix semantics */ 1363 FSYNC_MODE_STRICT, /* fsync behaves in line with ext4 */ 1364 FSYNC_MODE_NOBARRIER, /* fsync behaves nobarrier based on posix */ 1365 }; 1366 1367 enum { 1368 COMPR_MODE_FS, /* 1369 * automatically compress compression 1370 * enabled files 1371 */ 1372 COMPR_MODE_USER, /* 1373 * automatical compression is disabled. 1374 * user can control the file compression 1375 * using ioctls 1376 */ 1377 }; 1378 1379 enum { 1380 DISCARD_UNIT_BLOCK, /* basic discard unit is block */ 1381 DISCARD_UNIT_SEGMENT, /* basic discard unit is segment */ 1382 DISCARD_UNIT_SECTION, /* basic discard unit is section */ 1383 }; 1384 1385 enum { 1386 MEMORY_MODE_NORMAL, /* memory mode for normal devices */ 1387 MEMORY_MODE_LOW, /* memory mode for low memry devices */ 1388 }; 1389 1390 enum errors_option { 1391 MOUNT_ERRORS_READONLY, /* remount fs ro on errors */ 1392 MOUNT_ERRORS_CONTINUE, /* continue on errors */ 1393 MOUNT_ERRORS_PANIC, /* panic on errors */ 1394 }; 1395 1396 enum { 1397 BACKGROUND, 1398 FOREGROUND, 1399 MAX_CALL_TYPE, 1400 TOTAL_CALL = FOREGROUND, 1401 }; 1402 1403 static inline int f2fs_test_bit(unsigned int nr, char *addr); 1404 static inline void f2fs_set_bit(unsigned int nr, char *addr); 1405 static inline void f2fs_clear_bit(unsigned int nr, char *addr); 1406 1407 /* 1408 * Layout of f2fs page.private: 1409 * 1410 * Layout A: lowest bit should be 1 1411 * | bit0 = 1 | bit1 | bit2 | ... | bit MAX | private data .... | 1412 * bit 0 PAGE_PRIVATE_NOT_POINTER 1413 * bit 1 PAGE_PRIVATE_ONGOING_MIGRATION 1414 * bit 2 PAGE_PRIVATE_INLINE_INODE 1415 * bit 3 PAGE_PRIVATE_REF_RESOURCE 1416 * bit 4 PAGE_PRIVATE_ATOMIC_WRITE 1417 * bit 5- f2fs private data 1418 * 1419 * Layout B: lowest bit should be 0 1420 * page.private is a wrapped pointer. 1421 */ 1422 enum { 1423 PAGE_PRIVATE_NOT_POINTER, /* private contains non-pointer data */ 1424 PAGE_PRIVATE_ONGOING_MIGRATION, /* data page which is on-going migrating */ 1425 PAGE_PRIVATE_INLINE_INODE, /* inode page contains inline data */ 1426 PAGE_PRIVATE_REF_RESOURCE, /* dirty page has referenced resources */ 1427 PAGE_PRIVATE_ATOMIC_WRITE, /* data page from atomic write path */ 1428 PAGE_PRIVATE_MAX 1429 }; 1430 1431 /* For compression */ 1432 enum compress_algorithm_type { 1433 COMPRESS_LZO, 1434 COMPRESS_LZ4, 1435 COMPRESS_ZSTD, 1436 COMPRESS_LZORLE, 1437 COMPRESS_MAX, 1438 }; 1439 1440 enum compress_flag { 1441 COMPRESS_CHKSUM, 1442 COMPRESS_MAX_FLAG, 1443 }; 1444 1445 #define COMPRESS_WATERMARK 20 1446 #define COMPRESS_PERCENT 20 1447 1448 #define COMPRESS_DATA_RESERVED_SIZE 4 1449 struct compress_data { 1450 __le32 clen; /* compressed data size */ 1451 __le32 chksum; /* compressed data chksum */ 1452 __le32 reserved[COMPRESS_DATA_RESERVED_SIZE]; /* reserved */ 1453 u8 cdata[]; /* compressed data */ 1454 }; 1455 1456 #define COMPRESS_HEADER_SIZE (sizeof(struct compress_data)) 1457 1458 #define F2FS_COMPRESSED_PAGE_MAGIC 0xF5F2C000 1459 1460 #define F2FS_ZSTD_DEFAULT_CLEVEL 1 1461 1462 #define COMPRESS_LEVEL_OFFSET 8 1463 1464 /* compress context */ 1465 struct compress_ctx { 1466 struct inode *inode; /* inode the context belong to */ 1467 pgoff_t cluster_idx; /* cluster index number */ 1468 unsigned int cluster_size; /* page count in cluster */ 1469 unsigned int log_cluster_size; /* log of cluster size */ 1470 struct page **rpages; /* pages store raw data in cluster */ 1471 unsigned int nr_rpages; /* total page number in rpages */ 1472 struct page **cpages; /* pages store compressed data in cluster */ 1473 unsigned int nr_cpages; /* total page number in cpages */ 1474 unsigned int valid_nr_cpages; /* valid page number in cpages */ 1475 void *rbuf; /* virtual mapped address on rpages */ 1476 struct compress_data *cbuf; /* virtual mapped address on cpages */ 1477 size_t rlen; /* valid data length in rbuf */ 1478 size_t clen; /* valid data length in cbuf */ 1479 void *private; /* payload buffer for specified compression algorithm */ 1480 void *private2; /* extra payload buffer */ 1481 }; 1482 1483 /* compress context for write IO path */ 1484 struct compress_io_ctx { 1485 u32 magic; /* magic number to indicate page is compressed */ 1486 struct inode *inode; /* inode the context belong to */ 1487 struct page **rpages; /* pages store raw data in cluster */ 1488 unsigned int nr_rpages; /* total page number in rpages */ 1489 atomic_t pending_pages; /* in-flight compressed page count */ 1490 }; 1491 1492 /* Context for decompressing one cluster on the read IO path */ 1493 struct decompress_io_ctx { 1494 u32 magic; /* magic number to indicate page is compressed */ 1495 struct inode *inode; /* inode the context belong to */ 1496 pgoff_t cluster_idx; /* cluster index number */ 1497 unsigned int cluster_size; /* page count in cluster */ 1498 unsigned int log_cluster_size; /* log of cluster size */ 1499 struct page **rpages; /* pages store raw data in cluster */ 1500 unsigned int nr_rpages; /* total page number in rpages */ 1501 struct page **cpages; /* pages store compressed data in cluster */ 1502 unsigned int nr_cpages; /* total page number in cpages */ 1503 struct page **tpages; /* temp pages to pad holes in cluster */ 1504 void *rbuf; /* virtual mapped address on rpages */ 1505 struct compress_data *cbuf; /* virtual mapped address on cpages */ 1506 size_t rlen; /* valid data length in rbuf */ 1507 size_t clen; /* valid data length in cbuf */ 1508 1509 /* 1510 * The number of compressed pages remaining to be read in this cluster. 1511 * This is initially nr_cpages. It is decremented by 1 each time a page 1512 * has been read (or failed to be read). When it reaches 0, the cluster 1513 * is decompressed (or an error is reported). 1514 * 1515 * If an error occurs before all the pages have been submitted for I/O, 1516 * then this will never reach 0. In this case the I/O submitter is 1517 * responsible for calling f2fs_decompress_end_io() instead. 1518 */ 1519 atomic_t remaining_pages; 1520 1521 /* 1522 * Number of references to this decompress_io_ctx. 1523 * 1524 * One reference is held for I/O completion. This reference is dropped 1525 * after the pagecache pages are updated and unlocked -- either after 1526 * decompression (and verity if enabled), or after an error. 1527 * 1528 * In addition, each compressed page holds a reference while it is in a 1529 * bio. These references are necessary prevent compressed pages from 1530 * being freed while they are still in a bio. 1531 */ 1532 refcount_t refcnt; 1533 1534 bool failed; /* IO error occurred before decompression? */ 1535 bool need_verity; /* need fs-verity verification after decompression? */ 1536 void *private; /* payload buffer for specified decompression algorithm */ 1537 void *private2; /* extra payload buffer */ 1538 struct work_struct verity_work; /* work to verify the decompressed pages */ 1539 struct work_struct free_work; /* work for late free this structure itself */ 1540 }; 1541 1542 #define NULL_CLUSTER ((unsigned int)(~0)) 1543 #define MIN_COMPRESS_LOG_SIZE 2 1544 #define MAX_COMPRESS_LOG_SIZE 8 1545 #define MAX_COMPRESS_WINDOW_SIZE(log_size) ((PAGE_SIZE) << (log_size)) 1546 1547 struct f2fs_sb_info { 1548 struct super_block *sb; /* pointer to VFS super block */ 1549 struct proc_dir_entry *s_proc; /* proc entry */ 1550 struct f2fs_super_block *raw_super; /* raw super block pointer */ 1551 struct f2fs_rwsem sb_lock; /* lock for raw super block */ 1552 int valid_super_block; /* valid super block no */ 1553 unsigned long s_flag; /* flags for sbi */ 1554 struct mutex writepages; /* mutex for writepages() */ 1555 1556 #ifdef CONFIG_BLK_DEV_ZONED 1557 unsigned int blocks_per_blkz; /* F2FS blocks per zone */ 1558 #endif 1559 1560 /* for node-related operations */ 1561 struct f2fs_nm_info *nm_info; /* node manager */ 1562 struct inode *node_inode; /* cache node blocks */ 1563 1564 /* for segment-related operations */ 1565 struct f2fs_sm_info *sm_info; /* segment manager */ 1566 1567 /* for bio operations */ 1568 struct f2fs_bio_info *write_io[NR_PAGE_TYPE]; /* for write bios */ 1569 /* keep migration IO order for LFS mode */ 1570 struct f2fs_rwsem io_order_lock; 1571 pgoff_t page_eio_ofs[NR_PAGE_TYPE]; /* EIO page offset */ 1572 int page_eio_cnt[NR_PAGE_TYPE]; /* EIO count */ 1573 1574 /* for checkpoint */ 1575 struct f2fs_checkpoint *ckpt; /* raw checkpoint pointer */ 1576 int cur_cp_pack; /* remain current cp pack */ 1577 spinlock_t cp_lock; /* for flag in ckpt */ 1578 struct inode *meta_inode; /* cache meta blocks */ 1579 struct f2fs_rwsem cp_global_sem; /* checkpoint procedure lock */ 1580 struct f2fs_rwsem cp_rwsem; /* blocking FS operations */ 1581 struct f2fs_rwsem node_write; /* locking node writes */ 1582 struct f2fs_rwsem node_change; /* locking node change */ 1583 wait_queue_head_t cp_wait; 1584 unsigned long last_time[MAX_TIME]; /* to store time in jiffies */ 1585 long interval_time[MAX_TIME]; /* to store thresholds */ 1586 struct ckpt_req_control cprc_info; /* for checkpoint request control */ 1587 1588 struct inode_management im[MAX_INO_ENTRY]; /* manage inode cache */ 1589 1590 spinlock_t fsync_node_lock; /* for node entry lock */ 1591 struct list_head fsync_node_list; /* node list head */ 1592 unsigned int fsync_seg_id; /* sequence id */ 1593 unsigned int fsync_node_num; /* number of node entries */ 1594 1595 /* for orphan inode, use 0'th array */ 1596 unsigned int max_orphans; /* max orphan inodes */ 1597 1598 /* for inode management */ 1599 struct list_head inode_list[NR_INODE_TYPE]; /* dirty inode list */ 1600 spinlock_t inode_lock[NR_INODE_TYPE]; /* for dirty inode list lock */ 1601 struct mutex flush_lock; /* for flush exclusion */ 1602 1603 /* for extent tree cache */ 1604 struct extent_tree_info extent_tree[NR_EXTENT_CACHES]; 1605 atomic64_t allocated_data_blocks; /* for block age extent_cache */ 1606 1607 /* The threshold used for hot and warm data seperation*/ 1608 unsigned int hot_data_age_threshold; 1609 unsigned int warm_data_age_threshold; 1610 unsigned int last_age_weight; 1611 1612 /* basic filesystem units */ 1613 unsigned int log_sectors_per_block; /* log2 sectors per block */ 1614 unsigned int log_blocksize; /* log2 block size */ 1615 unsigned int blocksize; /* block size */ 1616 unsigned int root_ino_num; /* root inode number*/ 1617 unsigned int node_ino_num; /* node inode number*/ 1618 unsigned int meta_ino_num; /* meta inode number*/ 1619 unsigned int log_blocks_per_seg; /* log2 blocks per segment */ 1620 unsigned int blocks_per_seg; /* blocks per segment */ 1621 unsigned int unusable_blocks_per_sec; /* unusable blocks per section */ 1622 unsigned int segs_per_sec; /* segments per section */ 1623 unsigned int secs_per_zone; /* sections per zone */ 1624 unsigned int total_sections; /* total section count */ 1625 unsigned int total_node_count; /* total node block count */ 1626 unsigned int total_valid_node_count; /* valid node block count */ 1627 int dir_level; /* directory level */ 1628 bool readdir_ra; /* readahead inode in readdir */ 1629 u64 max_io_bytes; /* max io bytes to merge IOs */ 1630 1631 block_t user_block_count; /* # of user blocks */ 1632 block_t total_valid_block_count; /* # of valid blocks */ 1633 block_t discard_blks; /* discard command candidats */ 1634 block_t last_valid_block_count; /* for recovery */ 1635 block_t reserved_blocks; /* configurable reserved blocks */ 1636 block_t current_reserved_blocks; /* current reserved blocks */ 1637 1638 /* Additional tracking for no checkpoint mode */ 1639 block_t unusable_block_count; /* # of blocks saved by last cp */ 1640 1641 unsigned int nquota_files; /* # of quota sysfile */ 1642 struct f2fs_rwsem quota_sem; /* blocking cp for flags */ 1643 1644 /* # of pages, see count_type */ 1645 atomic_t nr_pages[NR_COUNT_TYPE]; 1646 /* # of allocated blocks */ 1647 struct percpu_counter alloc_valid_block_count; 1648 /* # of node block writes as roll forward recovery */ 1649 struct percpu_counter rf_node_block_count; 1650 1651 /* writeback control */ 1652 atomic_t wb_sync_req[META]; /* count # of WB_SYNC threads */ 1653 1654 /* valid inode count */ 1655 struct percpu_counter total_valid_inode_count; 1656 1657 struct f2fs_mount_info mount_opt; /* mount options */ 1658 1659 /* for cleaning operations */ 1660 struct f2fs_rwsem gc_lock; /* 1661 * semaphore for GC, avoid 1662 * race between GC and GC or CP 1663 */ 1664 struct f2fs_gc_kthread *gc_thread; /* GC thread */ 1665 struct atgc_management am; /* atgc management */ 1666 unsigned int cur_victim_sec; /* current victim section num */ 1667 unsigned int gc_mode; /* current GC state */ 1668 unsigned int next_victim_seg[2]; /* next segment in victim section */ 1669 spinlock_t gc_remaining_trials_lock; 1670 /* remaining trial count for GC_URGENT_* and GC_IDLE_* */ 1671 unsigned int gc_remaining_trials; 1672 1673 /* for skip statistic */ 1674 unsigned long long skipped_gc_rwsem; /* FG_GC only */ 1675 1676 /* threshold for gc trials on pinned files */ 1677 u64 gc_pin_file_threshold; 1678 struct f2fs_rwsem pin_sem; 1679 1680 /* maximum # of trials to find a victim segment for SSR and GC */ 1681 unsigned int max_victim_search; 1682 /* migration granularity of garbage collection, unit: segment */ 1683 unsigned int migration_granularity; 1684 1685 /* 1686 * for stat information. 1687 * one is for the LFS mode, and the other is for the SSR mode. 1688 */ 1689 #ifdef CONFIG_F2FS_STAT_FS 1690 struct f2fs_stat_info *stat_info; /* FS status information */ 1691 atomic_t meta_count[META_MAX]; /* # of meta blocks */ 1692 unsigned int segment_count[2]; /* # of allocated segments */ 1693 unsigned int block_count[2]; /* # of allocated blocks */ 1694 atomic_t inplace_count; /* # of inplace update */ 1695 /* # of lookup extent cache */ 1696 atomic64_t total_hit_ext[NR_EXTENT_CACHES]; 1697 /* # of hit rbtree extent node */ 1698 atomic64_t read_hit_rbtree[NR_EXTENT_CACHES]; 1699 /* # of hit cached extent node */ 1700 atomic64_t read_hit_cached[NR_EXTENT_CACHES]; 1701 /* # of hit largest extent node in read extent cache */ 1702 atomic64_t read_hit_largest; 1703 atomic_t inline_xattr; /* # of inline_xattr inodes */ 1704 atomic_t inline_inode; /* # of inline_data inodes */ 1705 atomic_t inline_dir; /* # of inline_dentry inodes */ 1706 atomic_t compr_inode; /* # of compressed inodes */ 1707 atomic64_t compr_blocks; /* # of compressed blocks */ 1708 atomic_t swapfile_inode; /* # of swapfile inodes */ 1709 atomic_t atomic_files; /* # of opened atomic file */ 1710 atomic_t max_aw_cnt; /* max # of atomic writes */ 1711 unsigned int io_skip_bggc; /* skip background gc for in-flight IO */ 1712 unsigned int other_skip_bggc; /* skip background gc for other reasons */ 1713 unsigned int ndirty_inode[NR_INODE_TYPE]; /* # of dirty inodes */ 1714 atomic_t cp_call_count[MAX_CALL_TYPE]; /* # of cp call */ 1715 #endif 1716 spinlock_t stat_lock; /* lock for stat operations */ 1717 1718 /* to attach REQ_META|REQ_FUA flags */ 1719 unsigned int data_io_flag; 1720 unsigned int node_io_flag; 1721 1722 /* For sysfs support */ 1723 struct kobject s_kobj; /* /sys/fs/f2fs/<devname> */ 1724 struct completion s_kobj_unregister; 1725 1726 struct kobject s_stat_kobj; /* /sys/fs/f2fs/<devname>/stat */ 1727 struct completion s_stat_kobj_unregister; 1728 1729 struct kobject s_feature_list_kobj; /* /sys/fs/f2fs/<devname>/feature_list */ 1730 struct completion s_feature_list_kobj_unregister; 1731 1732 /* For shrinker support */ 1733 struct list_head s_list; 1734 struct mutex umount_mutex; 1735 unsigned int shrinker_run_no; 1736 1737 /* For multi devices */ 1738 int s_ndevs; /* number of devices */ 1739 struct f2fs_dev_info *devs; /* for device list */ 1740 unsigned int dirty_device; /* for checkpoint data flush */ 1741 spinlock_t dev_lock; /* protect dirty_device */ 1742 bool aligned_blksize; /* all devices has the same logical blksize */ 1743 1744 /* For write statistics */ 1745 u64 sectors_written_start; 1746 u64 kbytes_written; 1747 1748 /* Reference to checksum algorithm driver via cryptoapi */ 1749 struct crypto_shash *s_chksum_driver; 1750 1751 /* Precomputed FS UUID checksum for seeding other checksums */ 1752 __u32 s_chksum_seed; 1753 1754 struct workqueue_struct *post_read_wq; /* post read workqueue */ 1755 1756 /* 1757 * If we are in irq context, let's update error information into 1758 * on-disk superblock in the work. 1759 */ 1760 struct work_struct s_error_work; 1761 unsigned char errors[MAX_F2FS_ERRORS]; /* error flags */ 1762 unsigned char stop_reason[MAX_STOP_REASON]; /* stop reason */ 1763 spinlock_t error_lock; /* protect errors/stop_reason array */ 1764 bool error_dirty; /* errors of sb is dirty */ 1765 1766 struct kmem_cache *inline_xattr_slab; /* inline xattr entry */ 1767 unsigned int inline_xattr_slab_size; /* default inline xattr slab size */ 1768 1769 /* For reclaimed segs statistics per each GC mode */ 1770 unsigned int gc_segment_mode; /* GC state for reclaimed segments */ 1771 unsigned int gc_reclaimed_segs[MAX_GC_MODE]; /* Reclaimed segs for each mode */ 1772 1773 unsigned long seq_file_ra_mul; /* multiplier for ra_pages of seq. files in fadvise */ 1774 1775 int max_fragment_chunk; /* max chunk size for block fragmentation mode */ 1776 int max_fragment_hole; /* max hole size for block fragmentation mode */ 1777 1778 /* For atomic write statistics */ 1779 atomic64_t current_atomic_write; 1780 s64 peak_atomic_write; 1781 u64 committed_atomic_block; 1782 u64 revoked_atomic_block; 1783 1784 #ifdef CONFIG_F2FS_FS_COMPRESSION 1785 struct kmem_cache *page_array_slab; /* page array entry */ 1786 unsigned int page_array_slab_size; /* default page array slab size */ 1787 1788 /* For runtime compression statistics */ 1789 u64 compr_written_block; 1790 u64 compr_saved_block; 1791 u32 compr_new_inode; 1792 1793 /* For compressed block cache */ 1794 struct inode *compress_inode; /* cache compressed blocks */ 1795 unsigned int compress_percent; /* cache page percentage */ 1796 unsigned int compress_watermark; /* cache page watermark */ 1797 atomic_t compress_page_hit; /* cache hit count */ 1798 #endif 1799 1800 #ifdef CONFIG_F2FS_IOSTAT 1801 /* For app/fs IO statistics */ 1802 spinlock_t iostat_lock; 1803 unsigned long long iostat_count[NR_IO_TYPE]; 1804 unsigned long long iostat_bytes[NR_IO_TYPE]; 1805 unsigned long long prev_iostat_bytes[NR_IO_TYPE]; 1806 bool iostat_enable; 1807 unsigned long iostat_next_period; 1808 unsigned int iostat_period_ms; 1809 1810 /* For io latency related statistics info in one iostat period */ 1811 spinlock_t iostat_lat_lock; 1812 struct iostat_lat_info *iostat_io_lat; 1813 #endif 1814 }; 1815 1816 /* Definitions to access f2fs_sb_info */ 1817 #define BLKS_PER_SEG(sbi) \ 1818 ((sbi)->blocks_per_seg) 1819 #define BLKS_PER_SEC(sbi) \ 1820 ((sbi)->segs_per_sec << (sbi)->log_blocks_per_seg) 1821 #define SEGS_PER_SEC(sbi) \ 1822 ((sbi)->segs_per_sec) 1823 1824 __printf(3, 4) 1825 void f2fs_printk(struct f2fs_sb_info *sbi, bool limit_rate, const char *fmt, ...); 1826 1827 #define f2fs_err(sbi, fmt, ...) \ 1828 f2fs_printk(sbi, false, KERN_ERR fmt, ##__VA_ARGS__) 1829 #define f2fs_warn(sbi, fmt, ...) \ 1830 f2fs_printk(sbi, false, KERN_WARNING fmt, ##__VA_ARGS__) 1831 #define f2fs_notice(sbi, fmt, ...) \ 1832 f2fs_printk(sbi, false, KERN_NOTICE fmt, ##__VA_ARGS__) 1833 #define f2fs_info(sbi, fmt, ...) \ 1834 f2fs_printk(sbi, false, KERN_INFO fmt, ##__VA_ARGS__) 1835 #define f2fs_debug(sbi, fmt, ...) \ 1836 f2fs_printk(sbi, false, KERN_DEBUG fmt, ##__VA_ARGS__) 1837 1838 #define f2fs_err_ratelimited(sbi, fmt, ...) \ 1839 f2fs_printk(sbi, true, KERN_ERR fmt, ##__VA_ARGS__) 1840 #define f2fs_warn_ratelimited(sbi, fmt, ...) \ 1841 f2fs_printk(sbi, true, KERN_WARNING fmt, ##__VA_ARGS__) 1842 #define f2fs_info_ratelimited(sbi, fmt, ...) \ 1843 f2fs_printk(sbi, true, KERN_INFO fmt, ##__VA_ARGS__) 1844 1845 #ifdef CONFIG_F2FS_FAULT_INJECTION 1846 #define time_to_inject(sbi, type) __time_to_inject(sbi, type, __func__, \ 1847 __builtin_return_address(0)) 1848 static inline bool __time_to_inject(struct f2fs_sb_info *sbi, int type, 1849 const char *func, const char *parent_func) 1850 { 1851 struct f2fs_fault_info *ffi = &F2FS_OPTION(sbi).fault_info; 1852 1853 if (!ffi->inject_rate) 1854 return false; 1855 1856 if (!IS_FAULT_SET(ffi, type)) 1857 return false; 1858 1859 atomic_inc(&ffi->inject_ops); 1860 if (atomic_read(&ffi->inject_ops) >= ffi->inject_rate) { 1861 atomic_set(&ffi->inject_ops, 0); 1862 f2fs_info_ratelimited(sbi, "inject %s in %s of %pS", 1863 f2fs_fault_name[type], func, parent_func); 1864 return true; 1865 } 1866 return false; 1867 } 1868 #else 1869 static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type) 1870 { 1871 return false; 1872 } 1873 #endif 1874 1875 /* 1876 * Test if the mounted volume is a multi-device volume. 1877 * - For a single regular disk volume, sbi->s_ndevs is 0. 1878 * - For a single zoned disk volume, sbi->s_ndevs is 1. 1879 * - For a multi-device volume, sbi->s_ndevs is always 2 or more. 1880 */ 1881 static inline bool f2fs_is_multi_device(struct f2fs_sb_info *sbi) 1882 { 1883 return sbi->s_ndevs > 1; 1884 } 1885 1886 static inline void f2fs_update_time(struct f2fs_sb_info *sbi, int type) 1887 { 1888 unsigned long now = jiffies; 1889 1890 sbi->last_time[type] = now; 1891 1892 /* DISCARD_TIME and GC_TIME are based on REQ_TIME */ 1893 if (type == REQ_TIME) { 1894 sbi->last_time[DISCARD_TIME] = now; 1895 sbi->last_time[GC_TIME] = now; 1896 } 1897 } 1898 1899 static inline bool f2fs_time_over(struct f2fs_sb_info *sbi, int type) 1900 { 1901 unsigned long interval = sbi->interval_time[type] * HZ; 1902 1903 return time_after(jiffies, sbi->last_time[type] + interval); 1904 } 1905 1906 static inline unsigned int f2fs_time_to_wait(struct f2fs_sb_info *sbi, 1907 int type) 1908 { 1909 unsigned long interval = sbi->interval_time[type] * HZ; 1910 unsigned int wait_ms = 0; 1911 long delta; 1912 1913 delta = (sbi->last_time[type] + interval) - jiffies; 1914 if (delta > 0) 1915 wait_ms = jiffies_to_msecs(delta); 1916 1917 return wait_ms; 1918 } 1919 1920 /* 1921 * Inline functions 1922 */ 1923 static inline u32 __f2fs_crc32(struct f2fs_sb_info *sbi, u32 crc, 1924 const void *address, unsigned int length) 1925 { 1926 struct { 1927 struct shash_desc shash; 1928 char ctx[4]; 1929 } desc; 1930 int err; 1931 1932 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver) != sizeof(desc.ctx)); 1933 1934 desc.shash.tfm = sbi->s_chksum_driver; 1935 *(u32 *)desc.ctx = crc; 1936 1937 err = crypto_shash_update(&desc.shash, address, length); 1938 BUG_ON(err); 1939 1940 return *(u32 *)desc.ctx; 1941 } 1942 1943 static inline u32 f2fs_crc32(struct f2fs_sb_info *sbi, const void *address, 1944 unsigned int length) 1945 { 1946 return __f2fs_crc32(sbi, F2FS_SUPER_MAGIC, address, length); 1947 } 1948 1949 static inline bool f2fs_crc_valid(struct f2fs_sb_info *sbi, __u32 blk_crc, 1950 void *buf, size_t buf_size) 1951 { 1952 return f2fs_crc32(sbi, buf, buf_size) == blk_crc; 1953 } 1954 1955 static inline u32 f2fs_chksum(struct f2fs_sb_info *sbi, u32 crc, 1956 const void *address, unsigned int length) 1957 { 1958 return __f2fs_crc32(sbi, crc, address, length); 1959 } 1960 1961 static inline struct f2fs_inode_info *F2FS_I(struct inode *inode) 1962 { 1963 return container_of(inode, struct f2fs_inode_info, vfs_inode); 1964 } 1965 1966 static inline struct f2fs_sb_info *F2FS_SB(struct super_block *sb) 1967 { 1968 return sb->s_fs_info; 1969 } 1970 1971 static inline struct f2fs_sb_info *F2FS_I_SB(struct inode *inode) 1972 { 1973 return F2FS_SB(inode->i_sb); 1974 } 1975 1976 static inline struct f2fs_sb_info *F2FS_M_SB(struct address_space *mapping) 1977 { 1978 return F2FS_I_SB(mapping->host); 1979 } 1980 1981 static inline struct f2fs_sb_info *F2FS_P_SB(struct page *page) 1982 { 1983 return F2FS_M_SB(page_file_mapping(page)); 1984 } 1985 1986 static inline struct f2fs_super_block *F2FS_RAW_SUPER(struct f2fs_sb_info *sbi) 1987 { 1988 return (struct f2fs_super_block *)(sbi->raw_super); 1989 } 1990 1991 static inline struct f2fs_checkpoint *F2FS_CKPT(struct f2fs_sb_info *sbi) 1992 { 1993 return (struct f2fs_checkpoint *)(sbi->ckpt); 1994 } 1995 1996 static inline struct f2fs_node *F2FS_NODE(struct page *page) 1997 { 1998 return (struct f2fs_node *)page_address(page); 1999 } 2000 2001 static inline struct f2fs_inode *F2FS_INODE(struct page *page) 2002 { 2003 return &((struct f2fs_node *)page_address(page))->i; 2004 } 2005 2006 static inline struct f2fs_nm_info *NM_I(struct f2fs_sb_info *sbi) 2007 { 2008 return (struct f2fs_nm_info *)(sbi->nm_info); 2009 } 2010 2011 static inline struct f2fs_sm_info *SM_I(struct f2fs_sb_info *sbi) 2012 { 2013 return (struct f2fs_sm_info *)(sbi->sm_info); 2014 } 2015 2016 static inline struct sit_info *SIT_I(struct f2fs_sb_info *sbi) 2017 { 2018 return (struct sit_info *)(SM_I(sbi)->sit_info); 2019 } 2020 2021 static inline struct free_segmap_info *FREE_I(struct f2fs_sb_info *sbi) 2022 { 2023 return (struct free_segmap_info *)(SM_I(sbi)->free_info); 2024 } 2025 2026 static inline struct dirty_seglist_info *DIRTY_I(struct f2fs_sb_info *sbi) 2027 { 2028 return (struct dirty_seglist_info *)(SM_I(sbi)->dirty_info); 2029 } 2030 2031 static inline struct address_space *META_MAPPING(struct f2fs_sb_info *sbi) 2032 { 2033 return sbi->meta_inode->i_mapping; 2034 } 2035 2036 static inline struct address_space *NODE_MAPPING(struct f2fs_sb_info *sbi) 2037 { 2038 return sbi->node_inode->i_mapping; 2039 } 2040 2041 static inline bool is_sbi_flag_set(struct f2fs_sb_info *sbi, unsigned int type) 2042 { 2043 return test_bit(type, &sbi->s_flag); 2044 } 2045 2046 static inline void set_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type) 2047 { 2048 set_bit(type, &sbi->s_flag); 2049 } 2050 2051 static inline void clear_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type) 2052 { 2053 clear_bit(type, &sbi->s_flag); 2054 } 2055 2056 static inline unsigned long long cur_cp_version(struct f2fs_checkpoint *cp) 2057 { 2058 return le64_to_cpu(cp->checkpoint_ver); 2059 } 2060 2061 static inline unsigned long f2fs_qf_ino(struct super_block *sb, int type) 2062 { 2063 if (type < F2FS_MAX_QUOTAS) 2064 return le32_to_cpu(F2FS_SB(sb)->raw_super->qf_ino[type]); 2065 return 0; 2066 } 2067 2068 static inline __u64 cur_cp_crc(struct f2fs_checkpoint *cp) 2069 { 2070 size_t crc_offset = le32_to_cpu(cp->checksum_offset); 2071 return le32_to_cpu(*((__le32 *)((unsigned char *)cp + crc_offset))); 2072 } 2073 2074 static inline bool __is_set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 2075 { 2076 unsigned int ckpt_flags = le32_to_cpu(cp->ckpt_flags); 2077 2078 return ckpt_flags & f; 2079 } 2080 2081 static inline bool is_set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 2082 { 2083 return __is_set_ckpt_flags(F2FS_CKPT(sbi), f); 2084 } 2085 2086 static inline void __set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 2087 { 2088 unsigned int ckpt_flags; 2089 2090 ckpt_flags = le32_to_cpu(cp->ckpt_flags); 2091 ckpt_flags |= f; 2092 cp->ckpt_flags = cpu_to_le32(ckpt_flags); 2093 } 2094 2095 static inline void set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 2096 { 2097 unsigned long flags; 2098 2099 spin_lock_irqsave(&sbi->cp_lock, flags); 2100 __set_ckpt_flags(F2FS_CKPT(sbi), f); 2101 spin_unlock_irqrestore(&sbi->cp_lock, flags); 2102 } 2103 2104 static inline void __clear_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 2105 { 2106 unsigned int ckpt_flags; 2107 2108 ckpt_flags = le32_to_cpu(cp->ckpt_flags); 2109 ckpt_flags &= (~f); 2110 cp->ckpt_flags = cpu_to_le32(ckpt_flags); 2111 } 2112 2113 static inline void clear_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 2114 { 2115 unsigned long flags; 2116 2117 spin_lock_irqsave(&sbi->cp_lock, flags); 2118 __clear_ckpt_flags(F2FS_CKPT(sbi), f); 2119 spin_unlock_irqrestore(&sbi->cp_lock, flags); 2120 } 2121 2122 #define init_f2fs_rwsem(sem) \ 2123 do { \ 2124 static struct lock_class_key __key; \ 2125 \ 2126 __init_f2fs_rwsem((sem), #sem, &__key); \ 2127 } while (0) 2128 2129 static inline void __init_f2fs_rwsem(struct f2fs_rwsem *sem, 2130 const char *sem_name, struct lock_class_key *key) 2131 { 2132 __init_rwsem(&sem->internal_rwsem, sem_name, key); 2133 #ifdef CONFIG_F2FS_UNFAIR_RWSEM 2134 init_waitqueue_head(&sem->read_waiters); 2135 #endif 2136 } 2137 2138 static inline int f2fs_rwsem_is_locked(struct f2fs_rwsem *sem) 2139 { 2140 return rwsem_is_locked(&sem->internal_rwsem); 2141 } 2142 2143 static inline int f2fs_rwsem_is_contended(struct f2fs_rwsem *sem) 2144 { 2145 return rwsem_is_contended(&sem->internal_rwsem); 2146 } 2147 2148 static inline void f2fs_down_read(struct f2fs_rwsem *sem) 2149 { 2150 #ifdef CONFIG_F2FS_UNFAIR_RWSEM 2151 wait_event(sem->read_waiters, down_read_trylock(&sem->internal_rwsem)); 2152 #else 2153 down_read(&sem->internal_rwsem); 2154 #endif 2155 } 2156 2157 static inline int f2fs_down_read_trylock(struct f2fs_rwsem *sem) 2158 { 2159 return down_read_trylock(&sem->internal_rwsem); 2160 } 2161 2162 static inline void f2fs_up_read(struct f2fs_rwsem *sem) 2163 { 2164 up_read(&sem->internal_rwsem); 2165 } 2166 2167 static inline void f2fs_down_write(struct f2fs_rwsem *sem) 2168 { 2169 down_write(&sem->internal_rwsem); 2170 } 2171 2172 #ifdef CONFIG_DEBUG_LOCK_ALLOC 2173 static inline void f2fs_down_read_nested(struct f2fs_rwsem *sem, int subclass) 2174 { 2175 down_read_nested(&sem->internal_rwsem, subclass); 2176 } 2177 2178 static inline void f2fs_down_write_nested(struct f2fs_rwsem *sem, int subclass) 2179 { 2180 down_write_nested(&sem->internal_rwsem, subclass); 2181 } 2182 #else 2183 #define f2fs_down_read_nested(sem, subclass) f2fs_down_read(sem) 2184 #define f2fs_down_write_nested(sem, subclass) f2fs_down_write(sem) 2185 #endif 2186 2187 static inline int f2fs_down_write_trylock(struct f2fs_rwsem *sem) 2188 { 2189 return down_write_trylock(&sem->internal_rwsem); 2190 } 2191 2192 static inline void f2fs_up_write(struct f2fs_rwsem *sem) 2193 { 2194 up_write(&sem->internal_rwsem); 2195 #ifdef CONFIG_F2FS_UNFAIR_RWSEM 2196 wake_up_all(&sem->read_waiters); 2197 #endif 2198 } 2199 2200 static inline void f2fs_lock_op(struct f2fs_sb_info *sbi) 2201 { 2202 f2fs_down_read(&sbi->cp_rwsem); 2203 } 2204 2205 static inline int f2fs_trylock_op(struct f2fs_sb_info *sbi) 2206 { 2207 if (time_to_inject(sbi, FAULT_LOCK_OP)) 2208 return 0; 2209 return f2fs_down_read_trylock(&sbi->cp_rwsem); 2210 } 2211 2212 static inline void f2fs_unlock_op(struct f2fs_sb_info *sbi) 2213 { 2214 f2fs_up_read(&sbi->cp_rwsem); 2215 } 2216 2217 static inline void f2fs_lock_all(struct f2fs_sb_info *sbi) 2218 { 2219 f2fs_down_write(&sbi->cp_rwsem); 2220 } 2221 2222 static inline void f2fs_unlock_all(struct f2fs_sb_info *sbi) 2223 { 2224 f2fs_up_write(&sbi->cp_rwsem); 2225 } 2226 2227 static inline int __get_cp_reason(struct f2fs_sb_info *sbi) 2228 { 2229 int reason = CP_SYNC; 2230 2231 if (test_opt(sbi, FASTBOOT)) 2232 reason = CP_FASTBOOT; 2233 if (is_sbi_flag_set(sbi, SBI_IS_CLOSE)) 2234 reason = CP_UMOUNT; 2235 return reason; 2236 } 2237 2238 static inline bool __remain_node_summaries(int reason) 2239 { 2240 return (reason & (CP_UMOUNT | CP_FASTBOOT)); 2241 } 2242 2243 static inline bool __exist_node_summaries(struct f2fs_sb_info *sbi) 2244 { 2245 return (is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG) || 2246 is_set_ckpt_flags(sbi, CP_FASTBOOT_FLAG)); 2247 } 2248 2249 /* 2250 * Check whether the inode has blocks or not 2251 */ 2252 static inline int F2FS_HAS_BLOCKS(struct inode *inode) 2253 { 2254 block_t xattr_block = F2FS_I(inode)->i_xattr_nid ? 1 : 0; 2255 2256 return (inode->i_blocks >> F2FS_LOG_SECTORS_PER_BLOCK) > xattr_block; 2257 } 2258 2259 static inline bool f2fs_has_xattr_block(unsigned int ofs) 2260 { 2261 return ofs == XATTR_NODE_OFFSET; 2262 } 2263 2264 static inline bool __allow_reserved_blocks(struct f2fs_sb_info *sbi, 2265 struct inode *inode, bool cap) 2266 { 2267 if (!inode) 2268 return true; 2269 if (!test_opt(sbi, RESERVE_ROOT)) 2270 return false; 2271 if (IS_NOQUOTA(inode)) 2272 return true; 2273 if (uid_eq(F2FS_OPTION(sbi).s_resuid, current_fsuid())) 2274 return true; 2275 if (!gid_eq(F2FS_OPTION(sbi).s_resgid, GLOBAL_ROOT_GID) && 2276 in_group_p(F2FS_OPTION(sbi).s_resgid)) 2277 return true; 2278 if (cap && capable(CAP_SYS_RESOURCE)) 2279 return true; 2280 return false; 2281 } 2282 2283 static inline unsigned int get_available_block_count(struct f2fs_sb_info *sbi, 2284 struct inode *inode, bool cap) 2285 { 2286 block_t avail_user_block_count; 2287 2288 avail_user_block_count = sbi->user_block_count - 2289 sbi->current_reserved_blocks; 2290 2291 if (!__allow_reserved_blocks(sbi, inode, cap)) 2292 avail_user_block_count -= F2FS_OPTION(sbi).root_reserved_blocks; 2293 2294 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) { 2295 if (avail_user_block_count > sbi->unusable_block_count) 2296 avail_user_block_count -= sbi->unusable_block_count; 2297 else 2298 avail_user_block_count = 0; 2299 } 2300 2301 return avail_user_block_count; 2302 } 2303 2304 static inline void f2fs_i_blocks_write(struct inode *, block_t, bool, bool); 2305 static inline int inc_valid_block_count(struct f2fs_sb_info *sbi, 2306 struct inode *inode, blkcnt_t *count, bool partial) 2307 { 2308 long long diff = 0, release = 0; 2309 block_t avail_user_block_count; 2310 int ret; 2311 2312 ret = dquot_reserve_block(inode, *count); 2313 if (ret) 2314 return ret; 2315 2316 if (time_to_inject(sbi, FAULT_BLOCK)) { 2317 release = *count; 2318 goto release_quota; 2319 } 2320 2321 /* 2322 * let's increase this in prior to actual block count change in order 2323 * for f2fs_sync_file to avoid data races when deciding checkpoint. 2324 */ 2325 percpu_counter_add(&sbi->alloc_valid_block_count, (*count)); 2326 2327 spin_lock(&sbi->stat_lock); 2328 2329 avail_user_block_count = get_available_block_count(sbi, inode, true); 2330 diff = (long long)sbi->total_valid_block_count + *count - 2331 avail_user_block_count; 2332 if (unlikely(diff > 0)) { 2333 if (!partial) { 2334 spin_unlock(&sbi->stat_lock); 2335 release = *count; 2336 goto enospc; 2337 } 2338 if (diff > *count) 2339 diff = *count; 2340 *count -= diff; 2341 release = diff; 2342 if (!*count) { 2343 spin_unlock(&sbi->stat_lock); 2344 goto enospc; 2345 } 2346 } 2347 sbi->total_valid_block_count += (block_t)(*count); 2348 2349 spin_unlock(&sbi->stat_lock); 2350 2351 if (unlikely(release)) { 2352 percpu_counter_sub(&sbi->alloc_valid_block_count, release); 2353 dquot_release_reservation_block(inode, release); 2354 } 2355 f2fs_i_blocks_write(inode, *count, true, true); 2356 return 0; 2357 2358 enospc: 2359 percpu_counter_sub(&sbi->alloc_valid_block_count, release); 2360 release_quota: 2361 dquot_release_reservation_block(inode, release); 2362 return -ENOSPC; 2363 } 2364 2365 #define PAGE_PRIVATE_GET_FUNC(name, flagname) \ 2366 static inline bool page_private_##name(struct page *page) \ 2367 { \ 2368 return PagePrivate(page) && \ 2369 test_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)) && \ 2370 test_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \ 2371 } 2372 2373 #define PAGE_PRIVATE_SET_FUNC(name, flagname) \ 2374 static inline void set_page_private_##name(struct page *page) \ 2375 { \ 2376 if (!PagePrivate(page)) \ 2377 attach_page_private(page, (void *)0); \ 2378 set_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)); \ 2379 set_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \ 2380 } 2381 2382 #define PAGE_PRIVATE_CLEAR_FUNC(name, flagname) \ 2383 static inline void clear_page_private_##name(struct page *page) \ 2384 { \ 2385 clear_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \ 2386 if (page_private(page) == BIT(PAGE_PRIVATE_NOT_POINTER)) \ 2387 detach_page_private(page); \ 2388 } 2389 2390 PAGE_PRIVATE_GET_FUNC(nonpointer, NOT_POINTER); 2391 PAGE_PRIVATE_GET_FUNC(inline, INLINE_INODE); 2392 PAGE_PRIVATE_GET_FUNC(gcing, ONGOING_MIGRATION); 2393 PAGE_PRIVATE_GET_FUNC(atomic, ATOMIC_WRITE); 2394 2395 PAGE_PRIVATE_SET_FUNC(reference, REF_RESOURCE); 2396 PAGE_PRIVATE_SET_FUNC(inline, INLINE_INODE); 2397 PAGE_PRIVATE_SET_FUNC(gcing, ONGOING_MIGRATION); 2398 PAGE_PRIVATE_SET_FUNC(atomic, ATOMIC_WRITE); 2399 2400 PAGE_PRIVATE_CLEAR_FUNC(reference, REF_RESOURCE); 2401 PAGE_PRIVATE_CLEAR_FUNC(inline, INLINE_INODE); 2402 PAGE_PRIVATE_CLEAR_FUNC(gcing, ONGOING_MIGRATION); 2403 PAGE_PRIVATE_CLEAR_FUNC(atomic, ATOMIC_WRITE); 2404 2405 static inline unsigned long get_page_private_data(struct page *page) 2406 { 2407 unsigned long data = page_private(page); 2408 2409 if (!test_bit(PAGE_PRIVATE_NOT_POINTER, &data)) 2410 return 0; 2411 return data >> PAGE_PRIVATE_MAX; 2412 } 2413 2414 static inline void set_page_private_data(struct page *page, unsigned long data) 2415 { 2416 if (!PagePrivate(page)) 2417 attach_page_private(page, (void *)0); 2418 set_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)); 2419 page_private(page) |= data << PAGE_PRIVATE_MAX; 2420 } 2421 2422 static inline void clear_page_private_data(struct page *page) 2423 { 2424 page_private(page) &= GENMASK(PAGE_PRIVATE_MAX - 1, 0); 2425 if (page_private(page) == BIT(PAGE_PRIVATE_NOT_POINTER)) 2426 detach_page_private(page); 2427 } 2428 2429 static inline void clear_page_private_all(struct page *page) 2430 { 2431 clear_page_private_data(page); 2432 clear_page_private_reference(page); 2433 clear_page_private_gcing(page); 2434 clear_page_private_inline(page); 2435 clear_page_private_atomic(page); 2436 2437 f2fs_bug_on(F2FS_P_SB(page), page_private(page)); 2438 } 2439 2440 static inline void dec_valid_block_count(struct f2fs_sb_info *sbi, 2441 struct inode *inode, 2442 block_t count) 2443 { 2444 blkcnt_t sectors = count << F2FS_LOG_SECTORS_PER_BLOCK; 2445 2446 spin_lock(&sbi->stat_lock); 2447 f2fs_bug_on(sbi, sbi->total_valid_block_count < (block_t) count); 2448 sbi->total_valid_block_count -= (block_t)count; 2449 if (sbi->reserved_blocks && 2450 sbi->current_reserved_blocks < sbi->reserved_blocks) 2451 sbi->current_reserved_blocks = min(sbi->reserved_blocks, 2452 sbi->current_reserved_blocks + count); 2453 spin_unlock(&sbi->stat_lock); 2454 if (unlikely(inode->i_blocks < sectors)) { 2455 f2fs_warn(sbi, "Inconsistent i_blocks, ino:%lu, iblocks:%llu, sectors:%llu", 2456 inode->i_ino, 2457 (unsigned long long)inode->i_blocks, 2458 (unsigned long long)sectors); 2459 set_sbi_flag(sbi, SBI_NEED_FSCK); 2460 return; 2461 } 2462 f2fs_i_blocks_write(inode, count, false, true); 2463 } 2464 2465 static inline void inc_page_count(struct f2fs_sb_info *sbi, int count_type) 2466 { 2467 atomic_inc(&sbi->nr_pages[count_type]); 2468 2469 if (count_type == F2FS_DIRTY_DENTS || 2470 count_type == F2FS_DIRTY_NODES || 2471 count_type == F2FS_DIRTY_META || 2472 count_type == F2FS_DIRTY_QDATA || 2473 count_type == F2FS_DIRTY_IMETA) 2474 set_sbi_flag(sbi, SBI_IS_DIRTY); 2475 } 2476 2477 static inline void inode_inc_dirty_pages(struct inode *inode) 2478 { 2479 atomic_inc(&F2FS_I(inode)->dirty_pages); 2480 inc_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ? 2481 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA); 2482 if (IS_NOQUOTA(inode)) 2483 inc_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA); 2484 } 2485 2486 static inline void dec_page_count(struct f2fs_sb_info *sbi, int count_type) 2487 { 2488 atomic_dec(&sbi->nr_pages[count_type]); 2489 } 2490 2491 static inline void inode_dec_dirty_pages(struct inode *inode) 2492 { 2493 if (!S_ISDIR(inode->i_mode) && !S_ISREG(inode->i_mode) && 2494 !S_ISLNK(inode->i_mode)) 2495 return; 2496 2497 atomic_dec(&F2FS_I(inode)->dirty_pages); 2498 dec_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ? 2499 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA); 2500 if (IS_NOQUOTA(inode)) 2501 dec_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA); 2502 } 2503 2504 static inline void inc_atomic_write_cnt(struct inode *inode) 2505 { 2506 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 2507 struct f2fs_inode_info *fi = F2FS_I(inode); 2508 u64 current_write; 2509 2510 fi->atomic_write_cnt++; 2511 atomic64_inc(&sbi->current_atomic_write); 2512 current_write = atomic64_read(&sbi->current_atomic_write); 2513 if (current_write > sbi->peak_atomic_write) 2514 sbi->peak_atomic_write = current_write; 2515 } 2516 2517 static inline void release_atomic_write_cnt(struct inode *inode) 2518 { 2519 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 2520 struct f2fs_inode_info *fi = F2FS_I(inode); 2521 2522 atomic64_sub(fi->atomic_write_cnt, &sbi->current_atomic_write); 2523 fi->atomic_write_cnt = 0; 2524 } 2525 2526 static inline s64 get_pages(struct f2fs_sb_info *sbi, int count_type) 2527 { 2528 return atomic_read(&sbi->nr_pages[count_type]); 2529 } 2530 2531 static inline int get_dirty_pages(struct inode *inode) 2532 { 2533 return atomic_read(&F2FS_I(inode)->dirty_pages); 2534 } 2535 2536 static inline int get_blocktype_secs(struct f2fs_sb_info *sbi, int block_type) 2537 { 2538 return div_u64(get_pages(sbi, block_type) + BLKS_PER_SEC(sbi) - 1, 2539 BLKS_PER_SEC(sbi)); 2540 } 2541 2542 static inline block_t valid_user_blocks(struct f2fs_sb_info *sbi) 2543 { 2544 return sbi->total_valid_block_count; 2545 } 2546 2547 static inline block_t discard_blocks(struct f2fs_sb_info *sbi) 2548 { 2549 return sbi->discard_blks; 2550 } 2551 2552 static inline unsigned long __bitmap_size(struct f2fs_sb_info *sbi, int flag) 2553 { 2554 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi); 2555 2556 /* return NAT or SIT bitmap */ 2557 if (flag == NAT_BITMAP) 2558 return le32_to_cpu(ckpt->nat_ver_bitmap_bytesize); 2559 else if (flag == SIT_BITMAP) 2560 return le32_to_cpu(ckpt->sit_ver_bitmap_bytesize); 2561 2562 return 0; 2563 } 2564 2565 static inline block_t __cp_payload(struct f2fs_sb_info *sbi) 2566 { 2567 return le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_payload); 2568 } 2569 2570 static inline void *__bitmap_ptr(struct f2fs_sb_info *sbi, int flag) 2571 { 2572 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi); 2573 void *tmp_ptr = &ckpt->sit_nat_version_bitmap; 2574 int offset; 2575 2576 if (is_set_ckpt_flags(sbi, CP_LARGE_NAT_BITMAP_FLAG)) { 2577 offset = (flag == SIT_BITMAP) ? 2578 le32_to_cpu(ckpt->nat_ver_bitmap_bytesize) : 0; 2579 /* 2580 * if large_nat_bitmap feature is enabled, leave checksum 2581 * protection for all nat/sit bitmaps. 2582 */ 2583 return tmp_ptr + offset + sizeof(__le32); 2584 } 2585 2586 if (__cp_payload(sbi) > 0) { 2587 if (flag == NAT_BITMAP) 2588 return tmp_ptr; 2589 else 2590 return (unsigned char *)ckpt + F2FS_BLKSIZE; 2591 } else { 2592 offset = (flag == NAT_BITMAP) ? 2593 le32_to_cpu(ckpt->sit_ver_bitmap_bytesize) : 0; 2594 return tmp_ptr + offset; 2595 } 2596 } 2597 2598 static inline block_t __start_cp_addr(struct f2fs_sb_info *sbi) 2599 { 2600 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr); 2601 2602 if (sbi->cur_cp_pack == 2) 2603 start_addr += BLKS_PER_SEG(sbi); 2604 return start_addr; 2605 } 2606 2607 static inline block_t __start_cp_next_addr(struct f2fs_sb_info *sbi) 2608 { 2609 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr); 2610 2611 if (sbi->cur_cp_pack == 1) 2612 start_addr += BLKS_PER_SEG(sbi); 2613 return start_addr; 2614 } 2615 2616 static inline void __set_cp_next_pack(struct f2fs_sb_info *sbi) 2617 { 2618 sbi->cur_cp_pack = (sbi->cur_cp_pack == 1) ? 2 : 1; 2619 } 2620 2621 static inline block_t __start_sum_addr(struct f2fs_sb_info *sbi) 2622 { 2623 return le32_to_cpu(F2FS_CKPT(sbi)->cp_pack_start_sum); 2624 } 2625 2626 extern void f2fs_mark_inode_dirty_sync(struct inode *inode, bool sync); 2627 static inline int inc_valid_node_count(struct f2fs_sb_info *sbi, 2628 struct inode *inode, bool is_inode) 2629 { 2630 block_t valid_block_count; 2631 unsigned int valid_node_count; 2632 unsigned int avail_user_block_count; 2633 int err; 2634 2635 if (is_inode) { 2636 if (inode) { 2637 err = dquot_alloc_inode(inode); 2638 if (err) 2639 return err; 2640 } 2641 } else { 2642 err = dquot_reserve_block(inode, 1); 2643 if (err) 2644 return err; 2645 } 2646 2647 if (time_to_inject(sbi, FAULT_BLOCK)) 2648 goto enospc; 2649 2650 spin_lock(&sbi->stat_lock); 2651 2652 valid_block_count = sbi->total_valid_block_count + 1; 2653 avail_user_block_count = get_available_block_count(sbi, inode, false); 2654 2655 if (unlikely(valid_block_count > avail_user_block_count)) { 2656 spin_unlock(&sbi->stat_lock); 2657 goto enospc; 2658 } 2659 2660 valid_node_count = sbi->total_valid_node_count + 1; 2661 if (unlikely(valid_node_count > sbi->total_node_count)) { 2662 spin_unlock(&sbi->stat_lock); 2663 goto enospc; 2664 } 2665 2666 sbi->total_valid_node_count++; 2667 sbi->total_valid_block_count++; 2668 spin_unlock(&sbi->stat_lock); 2669 2670 if (inode) { 2671 if (is_inode) 2672 f2fs_mark_inode_dirty_sync(inode, true); 2673 else 2674 f2fs_i_blocks_write(inode, 1, true, true); 2675 } 2676 2677 percpu_counter_inc(&sbi->alloc_valid_block_count); 2678 return 0; 2679 2680 enospc: 2681 if (is_inode) { 2682 if (inode) 2683 dquot_free_inode(inode); 2684 } else { 2685 dquot_release_reservation_block(inode, 1); 2686 } 2687 return -ENOSPC; 2688 } 2689 2690 static inline void dec_valid_node_count(struct f2fs_sb_info *sbi, 2691 struct inode *inode, bool is_inode) 2692 { 2693 spin_lock(&sbi->stat_lock); 2694 2695 if (unlikely(!sbi->total_valid_block_count || 2696 !sbi->total_valid_node_count)) { 2697 f2fs_warn(sbi, "dec_valid_node_count: inconsistent block counts, total_valid_block:%u, total_valid_node:%u", 2698 sbi->total_valid_block_count, 2699 sbi->total_valid_node_count); 2700 set_sbi_flag(sbi, SBI_NEED_FSCK); 2701 } else { 2702 sbi->total_valid_block_count--; 2703 sbi->total_valid_node_count--; 2704 } 2705 2706 if (sbi->reserved_blocks && 2707 sbi->current_reserved_blocks < sbi->reserved_blocks) 2708 sbi->current_reserved_blocks++; 2709 2710 spin_unlock(&sbi->stat_lock); 2711 2712 if (is_inode) { 2713 dquot_free_inode(inode); 2714 } else { 2715 if (unlikely(inode->i_blocks == 0)) { 2716 f2fs_warn(sbi, "dec_valid_node_count: inconsistent i_blocks, ino:%lu, iblocks:%llu", 2717 inode->i_ino, 2718 (unsigned long long)inode->i_blocks); 2719 set_sbi_flag(sbi, SBI_NEED_FSCK); 2720 return; 2721 } 2722 f2fs_i_blocks_write(inode, 1, false, true); 2723 } 2724 } 2725 2726 static inline unsigned int valid_node_count(struct f2fs_sb_info *sbi) 2727 { 2728 return sbi->total_valid_node_count; 2729 } 2730 2731 static inline void inc_valid_inode_count(struct f2fs_sb_info *sbi) 2732 { 2733 percpu_counter_inc(&sbi->total_valid_inode_count); 2734 } 2735 2736 static inline void dec_valid_inode_count(struct f2fs_sb_info *sbi) 2737 { 2738 percpu_counter_dec(&sbi->total_valid_inode_count); 2739 } 2740 2741 static inline s64 valid_inode_count(struct f2fs_sb_info *sbi) 2742 { 2743 return percpu_counter_sum_positive(&sbi->total_valid_inode_count); 2744 } 2745 2746 static inline struct page *f2fs_grab_cache_page(struct address_space *mapping, 2747 pgoff_t index, bool for_write) 2748 { 2749 struct page *page; 2750 unsigned int flags; 2751 2752 if (IS_ENABLED(CONFIG_F2FS_FAULT_INJECTION)) { 2753 if (!for_write) 2754 page = find_get_page_flags(mapping, index, 2755 FGP_LOCK | FGP_ACCESSED); 2756 else 2757 page = find_lock_page(mapping, index); 2758 if (page) 2759 return page; 2760 2761 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_ALLOC)) 2762 return NULL; 2763 } 2764 2765 if (!for_write) 2766 return grab_cache_page(mapping, index); 2767 2768 flags = memalloc_nofs_save(); 2769 page = grab_cache_page_write_begin(mapping, index); 2770 memalloc_nofs_restore(flags); 2771 2772 return page; 2773 } 2774 2775 static inline struct page *f2fs_pagecache_get_page( 2776 struct address_space *mapping, pgoff_t index, 2777 fgf_t fgp_flags, gfp_t gfp_mask) 2778 { 2779 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_GET)) 2780 return NULL; 2781 2782 return pagecache_get_page(mapping, index, fgp_flags, gfp_mask); 2783 } 2784 2785 static inline void f2fs_put_page(struct page *page, int unlock) 2786 { 2787 if (!page) 2788 return; 2789 2790 if (unlock) { 2791 f2fs_bug_on(F2FS_P_SB(page), !PageLocked(page)); 2792 unlock_page(page); 2793 } 2794 put_page(page); 2795 } 2796 2797 static inline void f2fs_put_dnode(struct dnode_of_data *dn) 2798 { 2799 if (dn->node_page) 2800 f2fs_put_page(dn->node_page, 1); 2801 if (dn->inode_page && dn->node_page != dn->inode_page) 2802 f2fs_put_page(dn->inode_page, 0); 2803 dn->node_page = NULL; 2804 dn->inode_page = NULL; 2805 } 2806 2807 static inline struct kmem_cache *f2fs_kmem_cache_create(const char *name, 2808 size_t size) 2809 { 2810 return kmem_cache_create(name, size, 0, SLAB_RECLAIM_ACCOUNT, NULL); 2811 } 2812 2813 static inline void *f2fs_kmem_cache_alloc_nofail(struct kmem_cache *cachep, 2814 gfp_t flags) 2815 { 2816 void *entry; 2817 2818 entry = kmem_cache_alloc(cachep, flags); 2819 if (!entry) 2820 entry = kmem_cache_alloc(cachep, flags | __GFP_NOFAIL); 2821 return entry; 2822 } 2823 2824 static inline void *f2fs_kmem_cache_alloc(struct kmem_cache *cachep, 2825 gfp_t flags, bool nofail, struct f2fs_sb_info *sbi) 2826 { 2827 if (nofail) 2828 return f2fs_kmem_cache_alloc_nofail(cachep, flags); 2829 2830 if (time_to_inject(sbi, FAULT_SLAB_ALLOC)) 2831 return NULL; 2832 2833 return kmem_cache_alloc(cachep, flags); 2834 } 2835 2836 static inline bool is_inflight_io(struct f2fs_sb_info *sbi, int type) 2837 { 2838 if (get_pages(sbi, F2FS_RD_DATA) || get_pages(sbi, F2FS_RD_NODE) || 2839 get_pages(sbi, F2FS_RD_META) || get_pages(sbi, F2FS_WB_DATA) || 2840 get_pages(sbi, F2FS_WB_CP_DATA) || 2841 get_pages(sbi, F2FS_DIO_READ) || 2842 get_pages(sbi, F2FS_DIO_WRITE)) 2843 return true; 2844 2845 if (type != DISCARD_TIME && SM_I(sbi) && SM_I(sbi)->dcc_info && 2846 atomic_read(&SM_I(sbi)->dcc_info->queued_discard)) 2847 return true; 2848 2849 if (SM_I(sbi) && SM_I(sbi)->fcc_info && 2850 atomic_read(&SM_I(sbi)->fcc_info->queued_flush)) 2851 return true; 2852 return false; 2853 } 2854 2855 static inline bool is_idle(struct f2fs_sb_info *sbi, int type) 2856 { 2857 if (sbi->gc_mode == GC_URGENT_HIGH) 2858 return true; 2859 2860 if (is_inflight_io(sbi, type)) 2861 return false; 2862 2863 if (sbi->gc_mode == GC_URGENT_MID) 2864 return true; 2865 2866 if (sbi->gc_mode == GC_URGENT_LOW && 2867 (type == DISCARD_TIME || type == GC_TIME)) 2868 return true; 2869 2870 return f2fs_time_over(sbi, type); 2871 } 2872 2873 static inline void f2fs_radix_tree_insert(struct radix_tree_root *root, 2874 unsigned long index, void *item) 2875 { 2876 while (radix_tree_insert(root, index, item)) 2877 cond_resched(); 2878 } 2879 2880 #define RAW_IS_INODE(p) ((p)->footer.nid == (p)->footer.ino) 2881 2882 static inline bool IS_INODE(struct page *page) 2883 { 2884 struct f2fs_node *p = F2FS_NODE(page); 2885 2886 return RAW_IS_INODE(p); 2887 } 2888 2889 static inline int offset_in_addr(struct f2fs_inode *i) 2890 { 2891 return (i->i_inline & F2FS_EXTRA_ATTR) ? 2892 (le16_to_cpu(i->i_extra_isize) / sizeof(__le32)) : 0; 2893 } 2894 2895 static inline __le32 *blkaddr_in_node(struct f2fs_node *node) 2896 { 2897 return RAW_IS_INODE(node) ? node->i.i_addr : node->dn.addr; 2898 } 2899 2900 static inline int f2fs_has_extra_attr(struct inode *inode); 2901 static inline block_t data_blkaddr(struct inode *inode, 2902 struct page *node_page, unsigned int offset) 2903 { 2904 struct f2fs_node *raw_node; 2905 __le32 *addr_array; 2906 int base = 0; 2907 bool is_inode = IS_INODE(node_page); 2908 2909 raw_node = F2FS_NODE(node_page); 2910 2911 if (is_inode) { 2912 if (!inode) 2913 /* from GC path only */ 2914 base = offset_in_addr(&raw_node->i); 2915 else if (f2fs_has_extra_attr(inode)) 2916 base = get_extra_isize(inode); 2917 } 2918 2919 addr_array = blkaddr_in_node(raw_node); 2920 return le32_to_cpu(addr_array[base + offset]); 2921 } 2922 2923 static inline block_t f2fs_data_blkaddr(struct dnode_of_data *dn) 2924 { 2925 return data_blkaddr(dn->inode, dn->node_page, dn->ofs_in_node); 2926 } 2927 2928 static inline int f2fs_test_bit(unsigned int nr, char *addr) 2929 { 2930 int mask; 2931 2932 addr += (nr >> 3); 2933 mask = BIT(7 - (nr & 0x07)); 2934 return mask & *addr; 2935 } 2936 2937 static inline void f2fs_set_bit(unsigned int nr, char *addr) 2938 { 2939 int mask; 2940 2941 addr += (nr >> 3); 2942 mask = BIT(7 - (nr & 0x07)); 2943 *addr |= mask; 2944 } 2945 2946 static inline void f2fs_clear_bit(unsigned int nr, char *addr) 2947 { 2948 int mask; 2949 2950 addr += (nr >> 3); 2951 mask = BIT(7 - (nr & 0x07)); 2952 *addr &= ~mask; 2953 } 2954 2955 static inline int f2fs_test_and_set_bit(unsigned int nr, char *addr) 2956 { 2957 int mask; 2958 int ret; 2959 2960 addr += (nr >> 3); 2961 mask = BIT(7 - (nr & 0x07)); 2962 ret = mask & *addr; 2963 *addr |= mask; 2964 return ret; 2965 } 2966 2967 static inline int f2fs_test_and_clear_bit(unsigned int nr, char *addr) 2968 { 2969 int mask; 2970 int ret; 2971 2972 addr += (nr >> 3); 2973 mask = BIT(7 - (nr & 0x07)); 2974 ret = mask & *addr; 2975 *addr &= ~mask; 2976 return ret; 2977 } 2978 2979 static inline void f2fs_change_bit(unsigned int nr, char *addr) 2980 { 2981 int mask; 2982 2983 addr += (nr >> 3); 2984 mask = BIT(7 - (nr & 0x07)); 2985 *addr ^= mask; 2986 } 2987 2988 /* 2989 * On-disk inode flags (f2fs_inode::i_flags) 2990 */ 2991 #define F2FS_COMPR_FL 0x00000004 /* Compress file */ 2992 #define F2FS_SYNC_FL 0x00000008 /* Synchronous updates */ 2993 #define F2FS_IMMUTABLE_FL 0x00000010 /* Immutable file */ 2994 #define F2FS_APPEND_FL 0x00000020 /* writes to file may only append */ 2995 #define F2FS_NODUMP_FL 0x00000040 /* do not dump file */ 2996 #define F2FS_NOATIME_FL 0x00000080 /* do not update atime */ 2997 #define F2FS_NOCOMP_FL 0x00000400 /* Don't compress */ 2998 #define F2FS_INDEX_FL 0x00001000 /* hash-indexed directory */ 2999 #define F2FS_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */ 3000 #define F2FS_PROJINHERIT_FL 0x20000000 /* Create with parents projid */ 3001 #define F2FS_CASEFOLD_FL 0x40000000 /* Casefolded file */ 3002 3003 #define F2FS_QUOTA_DEFAULT_FL (F2FS_NOATIME_FL | F2FS_IMMUTABLE_FL) 3004 3005 /* Flags that should be inherited by new inodes from their parent. */ 3006 #define F2FS_FL_INHERITED (F2FS_SYNC_FL | F2FS_NODUMP_FL | F2FS_NOATIME_FL | \ 3007 F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \ 3008 F2FS_CASEFOLD_FL) 3009 3010 /* Flags that are appropriate for regular files (all but dir-specific ones). */ 3011 #define F2FS_REG_FLMASK (~(F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \ 3012 F2FS_CASEFOLD_FL)) 3013 3014 /* Flags that are appropriate for non-directories/regular files. */ 3015 #define F2FS_OTHER_FLMASK (F2FS_NODUMP_FL | F2FS_NOATIME_FL) 3016 3017 static inline __u32 f2fs_mask_flags(umode_t mode, __u32 flags) 3018 { 3019 if (S_ISDIR(mode)) 3020 return flags; 3021 else if (S_ISREG(mode)) 3022 return flags & F2FS_REG_FLMASK; 3023 else 3024 return flags & F2FS_OTHER_FLMASK; 3025 } 3026 3027 static inline void __mark_inode_dirty_flag(struct inode *inode, 3028 int flag, bool set) 3029 { 3030 switch (flag) { 3031 case FI_INLINE_XATTR: 3032 case FI_INLINE_DATA: 3033 case FI_INLINE_DENTRY: 3034 case FI_NEW_INODE: 3035 if (set) 3036 return; 3037 fallthrough; 3038 case FI_DATA_EXIST: 3039 case FI_PIN_FILE: 3040 case FI_COMPRESS_RELEASED: 3041 f2fs_mark_inode_dirty_sync(inode, true); 3042 } 3043 } 3044 3045 static inline void set_inode_flag(struct inode *inode, int flag) 3046 { 3047 set_bit(flag, F2FS_I(inode)->flags); 3048 __mark_inode_dirty_flag(inode, flag, true); 3049 } 3050 3051 static inline int is_inode_flag_set(struct inode *inode, int flag) 3052 { 3053 return test_bit(flag, F2FS_I(inode)->flags); 3054 } 3055 3056 static inline void clear_inode_flag(struct inode *inode, int flag) 3057 { 3058 clear_bit(flag, F2FS_I(inode)->flags); 3059 __mark_inode_dirty_flag(inode, flag, false); 3060 } 3061 3062 static inline bool f2fs_verity_in_progress(struct inode *inode) 3063 { 3064 return IS_ENABLED(CONFIG_FS_VERITY) && 3065 is_inode_flag_set(inode, FI_VERITY_IN_PROGRESS); 3066 } 3067 3068 static inline void set_acl_inode(struct inode *inode, umode_t mode) 3069 { 3070 F2FS_I(inode)->i_acl_mode = mode; 3071 set_inode_flag(inode, FI_ACL_MODE); 3072 f2fs_mark_inode_dirty_sync(inode, false); 3073 } 3074 3075 static inline void f2fs_i_links_write(struct inode *inode, bool inc) 3076 { 3077 if (inc) 3078 inc_nlink(inode); 3079 else 3080 drop_nlink(inode); 3081 f2fs_mark_inode_dirty_sync(inode, true); 3082 } 3083 3084 static inline void f2fs_i_blocks_write(struct inode *inode, 3085 block_t diff, bool add, bool claim) 3086 { 3087 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE); 3088 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER); 3089 3090 /* add = 1, claim = 1 should be dquot_reserve_block in pair */ 3091 if (add) { 3092 if (claim) 3093 dquot_claim_block(inode, diff); 3094 else 3095 dquot_alloc_block_nofail(inode, diff); 3096 } else { 3097 dquot_free_block(inode, diff); 3098 } 3099 3100 f2fs_mark_inode_dirty_sync(inode, true); 3101 if (clean || recover) 3102 set_inode_flag(inode, FI_AUTO_RECOVER); 3103 } 3104 3105 static inline bool f2fs_is_atomic_file(struct inode *inode); 3106 3107 static inline void f2fs_i_size_write(struct inode *inode, loff_t i_size) 3108 { 3109 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE); 3110 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER); 3111 3112 if (i_size_read(inode) == i_size) 3113 return; 3114 3115 i_size_write(inode, i_size); 3116 3117 if (f2fs_is_atomic_file(inode)) 3118 return; 3119 3120 f2fs_mark_inode_dirty_sync(inode, true); 3121 if (clean || recover) 3122 set_inode_flag(inode, FI_AUTO_RECOVER); 3123 } 3124 3125 static inline void f2fs_i_depth_write(struct inode *inode, unsigned int depth) 3126 { 3127 F2FS_I(inode)->i_current_depth = depth; 3128 f2fs_mark_inode_dirty_sync(inode, true); 3129 } 3130 3131 static inline void f2fs_i_gc_failures_write(struct inode *inode, 3132 unsigned int count) 3133 { 3134 F2FS_I(inode)->i_gc_failures[GC_FAILURE_PIN] = count; 3135 f2fs_mark_inode_dirty_sync(inode, true); 3136 } 3137 3138 static inline void f2fs_i_xnid_write(struct inode *inode, nid_t xnid) 3139 { 3140 F2FS_I(inode)->i_xattr_nid = xnid; 3141 f2fs_mark_inode_dirty_sync(inode, true); 3142 } 3143 3144 static inline void f2fs_i_pino_write(struct inode *inode, nid_t pino) 3145 { 3146 F2FS_I(inode)->i_pino = pino; 3147 f2fs_mark_inode_dirty_sync(inode, true); 3148 } 3149 3150 static inline void get_inline_info(struct inode *inode, struct f2fs_inode *ri) 3151 { 3152 struct f2fs_inode_info *fi = F2FS_I(inode); 3153 3154 if (ri->i_inline & F2FS_INLINE_XATTR) 3155 set_bit(FI_INLINE_XATTR, fi->flags); 3156 if (ri->i_inline & F2FS_INLINE_DATA) 3157 set_bit(FI_INLINE_DATA, fi->flags); 3158 if (ri->i_inline & F2FS_INLINE_DENTRY) 3159 set_bit(FI_INLINE_DENTRY, fi->flags); 3160 if (ri->i_inline & F2FS_DATA_EXIST) 3161 set_bit(FI_DATA_EXIST, fi->flags); 3162 if (ri->i_inline & F2FS_EXTRA_ATTR) 3163 set_bit(FI_EXTRA_ATTR, fi->flags); 3164 if (ri->i_inline & F2FS_PIN_FILE) 3165 set_bit(FI_PIN_FILE, fi->flags); 3166 if (ri->i_inline & F2FS_COMPRESS_RELEASED) 3167 set_bit(FI_COMPRESS_RELEASED, fi->flags); 3168 } 3169 3170 static inline void set_raw_inline(struct inode *inode, struct f2fs_inode *ri) 3171 { 3172 ri->i_inline = 0; 3173 3174 if (is_inode_flag_set(inode, FI_INLINE_XATTR)) 3175 ri->i_inline |= F2FS_INLINE_XATTR; 3176 if (is_inode_flag_set(inode, FI_INLINE_DATA)) 3177 ri->i_inline |= F2FS_INLINE_DATA; 3178 if (is_inode_flag_set(inode, FI_INLINE_DENTRY)) 3179 ri->i_inline |= F2FS_INLINE_DENTRY; 3180 if (is_inode_flag_set(inode, FI_DATA_EXIST)) 3181 ri->i_inline |= F2FS_DATA_EXIST; 3182 if (is_inode_flag_set(inode, FI_EXTRA_ATTR)) 3183 ri->i_inline |= F2FS_EXTRA_ATTR; 3184 if (is_inode_flag_set(inode, FI_PIN_FILE)) 3185 ri->i_inline |= F2FS_PIN_FILE; 3186 if (is_inode_flag_set(inode, FI_COMPRESS_RELEASED)) 3187 ri->i_inline |= F2FS_COMPRESS_RELEASED; 3188 } 3189 3190 static inline int f2fs_has_extra_attr(struct inode *inode) 3191 { 3192 return is_inode_flag_set(inode, FI_EXTRA_ATTR); 3193 } 3194 3195 static inline int f2fs_has_inline_xattr(struct inode *inode) 3196 { 3197 return is_inode_flag_set(inode, FI_INLINE_XATTR); 3198 } 3199 3200 static inline int f2fs_compressed_file(struct inode *inode) 3201 { 3202 return S_ISREG(inode->i_mode) && 3203 is_inode_flag_set(inode, FI_COMPRESSED_FILE); 3204 } 3205 3206 static inline bool f2fs_need_compress_data(struct inode *inode) 3207 { 3208 int compress_mode = F2FS_OPTION(F2FS_I_SB(inode)).compress_mode; 3209 3210 if (!f2fs_compressed_file(inode)) 3211 return false; 3212 3213 if (compress_mode == COMPR_MODE_FS) 3214 return true; 3215 else if (compress_mode == COMPR_MODE_USER && 3216 is_inode_flag_set(inode, FI_ENABLE_COMPRESS)) 3217 return true; 3218 3219 return false; 3220 } 3221 3222 static inline unsigned int addrs_per_inode(struct inode *inode) 3223 { 3224 unsigned int addrs = CUR_ADDRS_PER_INODE(inode) - 3225 get_inline_xattr_addrs(inode); 3226 3227 if (!f2fs_compressed_file(inode)) 3228 return addrs; 3229 return ALIGN_DOWN(addrs, F2FS_I(inode)->i_cluster_size); 3230 } 3231 3232 static inline unsigned int addrs_per_block(struct inode *inode) 3233 { 3234 if (!f2fs_compressed_file(inode)) 3235 return DEF_ADDRS_PER_BLOCK; 3236 return ALIGN_DOWN(DEF_ADDRS_PER_BLOCK, F2FS_I(inode)->i_cluster_size); 3237 } 3238 3239 static inline void *inline_xattr_addr(struct inode *inode, struct page *page) 3240 { 3241 struct f2fs_inode *ri = F2FS_INODE(page); 3242 3243 return (void *)&(ri->i_addr[DEF_ADDRS_PER_INODE - 3244 get_inline_xattr_addrs(inode)]); 3245 } 3246 3247 static inline int inline_xattr_size(struct inode *inode) 3248 { 3249 if (f2fs_has_inline_xattr(inode)) 3250 return get_inline_xattr_addrs(inode) * sizeof(__le32); 3251 return 0; 3252 } 3253 3254 /* 3255 * Notice: check inline_data flag without inode page lock is unsafe. 3256 * It could change at any time by f2fs_convert_inline_page(). 3257 */ 3258 static inline int f2fs_has_inline_data(struct inode *inode) 3259 { 3260 return is_inode_flag_set(inode, FI_INLINE_DATA); 3261 } 3262 3263 static inline int f2fs_exist_data(struct inode *inode) 3264 { 3265 return is_inode_flag_set(inode, FI_DATA_EXIST); 3266 } 3267 3268 static inline int f2fs_is_mmap_file(struct inode *inode) 3269 { 3270 return is_inode_flag_set(inode, FI_MMAP_FILE); 3271 } 3272 3273 static inline bool f2fs_is_pinned_file(struct inode *inode) 3274 { 3275 return is_inode_flag_set(inode, FI_PIN_FILE); 3276 } 3277 3278 static inline bool f2fs_is_atomic_file(struct inode *inode) 3279 { 3280 return is_inode_flag_set(inode, FI_ATOMIC_FILE); 3281 } 3282 3283 static inline bool f2fs_is_cow_file(struct inode *inode) 3284 { 3285 return is_inode_flag_set(inode, FI_COW_FILE); 3286 } 3287 3288 static inline __le32 *get_dnode_addr(struct inode *inode, 3289 struct page *node_page); 3290 static inline void *inline_data_addr(struct inode *inode, struct page *page) 3291 { 3292 __le32 *addr = get_dnode_addr(inode, page); 3293 3294 return (void *)(addr + DEF_INLINE_RESERVED_SIZE); 3295 } 3296 3297 static inline int f2fs_has_inline_dentry(struct inode *inode) 3298 { 3299 return is_inode_flag_set(inode, FI_INLINE_DENTRY); 3300 } 3301 3302 static inline int is_file(struct inode *inode, int type) 3303 { 3304 return F2FS_I(inode)->i_advise & type; 3305 } 3306 3307 static inline void set_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 void clear_file(struct inode *inode, int type) 3316 { 3317 if (!is_file(inode, type)) 3318 return; 3319 F2FS_I(inode)->i_advise &= ~type; 3320 f2fs_mark_inode_dirty_sync(inode, true); 3321 } 3322 3323 static inline bool f2fs_is_time_consistent(struct inode *inode) 3324 { 3325 struct timespec64 ctime = inode_get_ctime(inode); 3326 3327 if (!timespec64_equal(F2FS_I(inode)->i_disk_time, &inode->i_atime)) 3328 return false; 3329 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 1, &ctime)) 3330 return false; 3331 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 2, &inode->i_mtime)) 3332 return false; 3333 return true; 3334 } 3335 3336 static inline bool f2fs_skip_inode_update(struct inode *inode, int dsync) 3337 { 3338 bool ret; 3339 3340 if (dsync) { 3341 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 3342 3343 spin_lock(&sbi->inode_lock[DIRTY_META]); 3344 ret = list_empty(&F2FS_I(inode)->gdirty_list); 3345 spin_unlock(&sbi->inode_lock[DIRTY_META]); 3346 return ret; 3347 } 3348 if (!is_inode_flag_set(inode, FI_AUTO_RECOVER) || 3349 file_keep_isize(inode) || 3350 i_size_read(inode) & ~PAGE_MASK) 3351 return false; 3352 3353 if (!f2fs_is_time_consistent(inode)) 3354 return false; 3355 3356 spin_lock(&F2FS_I(inode)->i_size_lock); 3357 ret = F2FS_I(inode)->last_disk_size == i_size_read(inode); 3358 spin_unlock(&F2FS_I(inode)->i_size_lock); 3359 3360 return ret; 3361 } 3362 3363 static inline bool f2fs_readonly(struct super_block *sb) 3364 { 3365 return sb_rdonly(sb); 3366 } 3367 3368 static inline bool f2fs_cp_error(struct f2fs_sb_info *sbi) 3369 { 3370 return is_set_ckpt_flags(sbi, CP_ERROR_FLAG); 3371 } 3372 3373 static inline void *f2fs_kmalloc(struct f2fs_sb_info *sbi, 3374 size_t size, gfp_t flags) 3375 { 3376 if (time_to_inject(sbi, FAULT_KMALLOC)) 3377 return NULL; 3378 3379 return kmalloc(size, flags); 3380 } 3381 3382 static inline void *f2fs_getname(struct f2fs_sb_info *sbi) 3383 { 3384 if (time_to_inject(sbi, FAULT_KMALLOC)) 3385 return NULL; 3386 3387 return __getname(); 3388 } 3389 3390 static inline void f2fs_putname(char *buf) 3391 { 3392 __putname(buf); 3393 } 3394 3395 static inline void *f2fs_kzalloc(struct f2fs_sb_info *sbi, 3396 size_t size, gfp_t flags) 3397 { 3398 return f2fs_kmalloc(sbi, size, flags | __GFP_ZERO); 3399 } 3400 3401 static inline void *f2fs_kvmalloc(struct f2fs_sb_info *sbi, 3402 size_t size, gfp_t flags) 3403 { 3404 if (time_to_inject(sbi, FAULT_KVMALLOC)) 3405 return NULL; 3406 3407 return kvmalloc(size, flags); 3408 } 3409 3410 static inline void *f2fs_kvzalloc(struct f2fs_sb_info *sbi, 3411 size_t size, gfp_t flags) 3412 { 3413 return f2fs_kvmalloc(sbi, size, flags | __GFP_ZERO); 3414 } 3415 3416 static inline int get_extra_isize(struct inode *inode) 3417 { 3418 return F2FS_I(inode)->i_extra_isize / sizeof(__le32); 3419 } 3420 3421 static inline int get_inline_xattr_addrs(struct inode *inode) 3422 { 3423 return F2FS_I(inode)->i_inline_xattr_size; 3424 } 3425 3426 static inline __le32 *get_dnode_addr(struct inode *inode, 3427 struct page *node_page) 3428 { 3429 int base = 0; 3430 3431 if (IS_INODE(node_page) && f2fs_has_extra_attr(inode)) 3432 base = get_extra_isize(inode); 3433 3434 return blkaddr_in_node(F2FS_NODE(node_page)) + base; 3435 } 3436 3437 #define f2fs_get_inode_mode(i) \ 3438 ((is_inode_flag_set(i, FI_ACL_MODE)) ? \ 3439 (F2FS_I(i)->i_acl_mode) : ((i)->i_mode)) 3440 3441 #define F2FS_MIN_EXTRA_ATTR_SIZE (sizeof(__le32)) 3442 3443 #define F2FS_TOTAL_EXTRA_ATTR_SIZE \ 3444 (offsetof(struct f2fs_inode, i_extra_end) - \ 3445 offsetof(struct f2fs_inode, i_extra_isize)) \ 3446 3447 #define F2FS_OLD_ATTRIBUTE_SIZE (offsetof(struct f2fs_inode, i_addr)) 3448 #define F2FS_FITS_IN_INODE(f2fs_inode, extra_isize, field) \ 3449 ((offsetof(typeof(*(f2fs_inode)), field) + \ 3450 sizeof((f2fs_inode)->field)) \ 3451 <= (F2FS_OLD_ATTRIBUTE_SIZE + (extra_isize))) \ 3452 3453 #define __is_large_section(sbi) (SEGS_PER_SEC(sbi) > 1) 3454 3455 #define __is_meta_io(fio) (PAGE_TYPE_OF_BIO((fio)->type) == META) 3456 3457 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi, 3458 block_t blkaddr, int type); 3459 static inline void verify_blkaddr(struct f2fs_sb_info *sbi, 3460 block_t blkaddr, int type) 3461 { 3462 if (!f2fs_is_valid_blkaddr(sbi, blkaddr, type)) 3463 f2fs_err(sbi, "invalid blkaddr: %u, type: %d, run fsck to fix.", 3464 blkaddr, type); 3465 } 3466 3467 static inline bool __is_valid_data_blkaddr(block_t blkaddr) 3468 { 3469 if (blkaddr == NEW_ADDR || blkaddr == NULL_ADDR || 3470 blkaddr == COMPRESS_ADDR) 3471 return false; 3472 return true; 3473 } 3474 3475 /* 3476 * file.c 3477 */ 3478 int f2fs_sync_file(struct file *file, loff_t start, loff_t end, int datasync); 3479 int f2fs_do_truncate_blocks(struct inode *inode, u64 from, bool lock); 3480 int f2fs_truncate_blocks(struct inode *inode, u64 from, bool lock); 3481 int f2fs_truncate(struct inode *inode); 3482 int f2fs_getattr(struct mnt_idmap *idmap, const struct path *path, 3483 struct kstat *stat, u32 request_mask, unsigned int flags); 3484 int f2fs_setattr(struct mnt_idmap *idmap, struct dentry *dentry, 3485 struct iattr *attr); 3486 int f2fs_truncate_hole(struct inode *inode, pgoff_t pg_start, pgoff_t pg_end); 3487 void f2fs_truncate_data_blocks_range(struct dnode_of_data *dn, int count); 3488 int f2fs_do_shutdown(struct f2fs_sb_info *sbi, unsigned int flag, 3489 bool readonly, bool need_lock); 3490 int f2fs_precache_extents(struct inode *inode); 3491 int f2fs_fileattr_get(struct dentry *dentry, struct fileattr *fa); 3492 int f2fs_fileattr_set(struct mnt_idmap *idmap, 3493 struct dentry *dentry, struct fileattr *fa); 3494 long f2fs_ioctl(struct file *filp, unsigned int cmd, unsigned long arg); 3495 long f2fs_compat_ioctl(struct file *file, unsigned int cmd, unsigned long arg); 3496 int f2fs_transfer_project_quota(struct inode *inode, kprojid_t kprojid); 3497 int f2fs_pin_file_control(struct inode *inode, bool inc); 3498 3499 /* 3500 * inode.c 3501 */ 3502 void f2fs_set_inode_flags(struct inode *inode); 3503 bool f2fs_inode_chksum_verify(struct f2fs_sb_info *sbi, struct page *page); 3504 void f2fs_inode_chksum_set(struct f2fs_sb_info *sbi, struct page *page); 3505 struct inode *f2fs_iget(struct super_block *sb, unsigned long ino); 3506 struct inode *f2fs_iget_retry(struct super_block *sb, unsigned long ino); 3507 int f2fs_try_to_free_nats(struct f2fs_sb_info *sbi, int nr_shrink); 3508 void f2fs_update_inode(struct inode *inode, struct page *node_page); 3509 void f2fs_update_inode_page(struct inode *inode); 3510 int f2fs_write_inode(struct inode *inode, struct writeback_control *wbc); 3511 void f2fs_evict_inode(struct inode *inode); 3512 void f2fs_handle_failed_inode(struct inode *inode); 3513 3514 /* 3515 * namei.c 3516 */ 3517 int f2fs_update_extension_list(struct f2fs_sb_info *sbi, const char *name, 3518 bool hot, bool set); 3519 struct dentry *f2fs_get_parent(struct dentry *child); 3520 int f2fs_get_tmpfile(struct mnt_idmap *idmap, struct inode *dir, 3521 struct inode **new_inode); 3522 3523 /* 3524 * dir.c 3525 */ 3526 int f2fs_init_casefolded_name(const struct inode *dir, 3527 struct f2fs_filename *fname); 3528 int f2fs_setup_filename(struct inode *dir, const struct qstr *iname, 3529 int lookup, struct f2fs_filename *fname); 3530 int f2fs_prepare_lookup(struct inode *dir, struct dentry *dentry, 3531 struct f2fs_filename *fname); 3532 void f2fs_free_filename(struct f2fs_filename *fname); 3533 struct f2fs_dir_entry *f2fs_find_target_dentry(const struct f2fs_dentry_ptr *d, 3534 const struct f2fs_filename *fname, int *max_slots); 3535 int f2fs_fill_dentries(struct dir_context *ctx, struct f2fs_dentry_ptr *d, 3536 unsigned int start_pos, struct fscrypt_str *fstr); 3537 void f2fs_do_make_empty_dir(struct inode *inode, struct inode *parent, 3538 struct f2fs_dentry_ptr *d); 3539 struct page *f2fs_init_inode_metadata(struct inode *inode, struct inode *dir, 3540 const struct f2fs_filename *fname, struct page *dpage); 3541 void f2fs_update_parent_metadata(struct inode *dir, struct inode *inode, 3542 unsigned int current_depth); 3543 int f2fs_room_for_filename(const void *bitmap, int slots, int max_slots); 3544 void f2fs_drop_nlink(struct inode *dir, struct inode *inode); 3545 struct f2fs_dir_entry *__f2fs_find_entry(struct inode *dir, 3546 const struct f2fs_filename *fname, 3547 struct page **res_page); 3548 struct f2fs_dir_entry *f2fs_find_entry(struct inode *dir, 3549 const struct qstr *child, struct page **res_page); 3550 struct f2fs_dir_entry *f2fs_parent_dir(struct inode *dir, struct page **p); 3551 ino_t f2fs_inode_by_name(struct inode *dir, const struct qstr *qstr, 3552 struct page **page); 3553 void f2fs_set_link(struct inode *dir, struct f2fs_dir_entry *de, 3554 struct page *page, struct inode *inode); 3555 bool f2fs_has_enough_room(struct inode *dir, struct page *ipage, 3556 const struct f2fs_filename *fname); 3557 void f2fs_update_dentry(nid_t ino, umode_t mode, struct f2fs_dentry_ptr *d, 3558 const struct fscrypt_str *name, f2fs_hash_t name_hash, 3559 unsigned int bit_pos); 3560 int f2fs_add_regular_entry(struct inode *dir, const struct f2fs_filename *fname, 3561 struct inode *inode, nid_t ino, umode_t mode); 3562 int f2fs_add_dentry(struct inode *dir, const struct f2fs_filename *fname, 3563 struct inode *inode, nid_t ino, umode_t mode); 3564 int f2fs_do_add_link(struct inode *dir, const struct qstr *name, 3565 struct inode *inode, nid_t ino, umode_t mode); 3566 void f2fs_delete_entry(struct f2fs_dir_entry *dentry, struct page *page, 3567 struct inode *dir, struct inode *inode); 3568 int f2fs_do_tmpfile(struct inode *inode, struct inode *dir, 3569 struct f2fs_filename *fname); 3570 bool f2fs_empty_dir(struct inode *dir); 3571 3572 static inline int f2fs_add_link(struct dentry *dentry, struct inode *inode) 3573 { 3574 if (fscrypt_is_nokey_name(dentry)) 3575 return -ENOKEY; 3576 return f2fs_do_add_link(d_inode(dentry->d_parent), &dentry->d_name, 3577 inode, inode->i_ino, inode->i_mode); 3578 } 3579 3580 /* 3581 * super.c 3582 */ 3583 int f2fs_inode_dirtied(struct inode *inode, bool sync); 3584 void f2fs_inode_synced(struct inode *inode); 3585 int f2fs_dquot_initialize(struct inode *inode); 3586 int f2fs_enable_quota_files(struct f2fs_sb_info *sbi, bool rdonly); 3587 int f2fs_quota_sync(struct super_block *sb, int type); 3588 loff_t max_file_blocks(struct inode *inode); 3589 void f2fs_quota_off_umount(struct super_block *sb); 3590 void f2fs_save_errors(struct f2fs_sb_info *sbi, unsigned char flag); 3591 void f2fs_handle_critical_error(struct f2fs_sb_info *sbi, unsigned char reason); 3592 void f2fs_handle_error(struct f2fs_sb_info *sbi, unsigned char error); 3593 void f2fs_handle_error_async(struct f2fs_sb_info *sbi, unsigned char error); 3594 int f2fs_commit_super(struct f2fs_sb_info *sbi, bool recover); 3595 int f2fs_sync_fs(struct super_block *sb, int sync); 3596 int f2fs_sanity_check_ckpt(struct f2fs_sb_info *sbi); 3597 3598 /* 3599 * hash.c 3600 */ 3601 void f2fs_hash_filename(const struct inode *dir, struct f2fs_filename *fname); 3602 3603 /* 3604 * node.c 3605 */ 3606 struct node_info; 3607 3608 int f2fs_check_nid_range(struct f2fs_sb_info *sbi, nid_t nid); 3609 bool f2fs_available_free_memory(struct f2fs_sb_info *sbi, int type); 3610 bool f2fs_in_warm_node_list(struct f2fs_sb_info *sbi, struct page *page); 3611 void f2fs_init_fsync_node_info(struct f2fs_sb_info *sbi); 3612 void f2fs_del_fsync_node_entry(struct f2fs_sb_info *sbi, struct page *page); 3613 void f2fs_reset_fsync_node_info(struct f2fs_sb_info *sbi); 3614 int f2fs_need_dentry_mark(struct f2fs_sb_info *sbi, nid_t nid); 3615 bool f2fs_is_checkpointed_node(struct f2fs_sb_info *sbi, nid_t nid); 3616 bool f2fs_need_inode_block_update(struct f2fs_sb_info *sbi, nid_t ino); 3617 int f2fs_get_node_info(struct f2fs_sb_info *sbi, nid_t nid, 3618 struct node_info *ni, bool checkpoint_context); 3619 pgoff_t f2fs_get_next_page_offset(struct dnode_of_data *dn, pgoff_t pgofs); 3620 int f2fs_get_dnode_of_data(struct dnode_of_data *dn, pgoff_t index, int mode); 3621 int f2fs_truncate_inode_blocks(struct inode *inode, pgoff_t from); 3622 int f2fs_truncate_xattr_node(struct inode *inode); 3623 int f2fs_wait_on_node_pages_writeback(struct f2fs_sb_info *sbi, 3624 unsigned int seq_id); 3625 bool f2fs_nat_bitmap_enabled(struct f2fs_sb_info *sbi); 3626 int f2fs_remove_inode_page(struct inode *inode); 3627 struct page *f2fs_new_inode_page(struct inode *inode); 3628 struct page *f2fs_new_node_page(struct dnode_of_data *dn, unsigned int ofs); 3629 void f2fs_ra_node_page(struct f2fs_sb_info *sbi, nid_t nid); 3630 struct page *f2fs_get_node_page(struct f2fs_sb_info *sbi, pgoff_t nid); 3631 struct page *f2fs_get_node_page_ra(struct page *parent, int start); 3632 int f2fs_move_node_page(struct page *node_page, int gc_type); 3633 void f2fs_flush_inline_data(struct f2fs_sb_info *sbi); 3634 int f2fs_fsync_node_pages(struct f2fs_sb_info *sbi, struct inode *inode, 3635 struct writeback_control *wbc, bool atomic, 3636 unsigned int *seq_id); 3637 int f2fs_sync_node_pages(struct f2fs_sb_info *sbi, 3638 struct writeback_control *wbc, 3639 bool do_balance, enum iostat_type io_type); 3640 int f2fs_build_free_nids(struct f2fs_sb_info *sbi, bool sync, bool mount); 3641 bool f2fs_alloc_nid(struct f2fs_sb_info *sbi, nid_t *nid); 3642 void f2fs_alloc_nid_done(struct f2fs_sb_info *sbi, nid_t nid); 3643 void f2fs_alloc_nid_failed(struct f2fs_sb_info *sbi, nid_t nid); 3644 int f2fs_try_to_free_nids(struct f2fs_sb_info *sbi, int nr_shrink); 3645 int f2fs_recover_inline_xattr(struct inode *inode, struct page *page); 3646 int f2fs_recover_xattr_data(struct inode *inode, struct page *page); 3647 int f2fs_recover_inode_page(struct f2fs_sb_info *sbi, struct page *page); 3648 int f2fs_restore_node_summary(struct f2fs_sb_info *sbi, 3649 unsigned int segno, struct f2fs_summary_block *sum); 3650 void f2fs_enable_nat_bits(struct f2fs_sb_info *sbi); 3651 int f2fs_flush_nat_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3652 int f2fs_build_node_manager(struct f2fs_sb_info *sbi); 3653 void f2fs_destroy_node_manager(struct f2fs_sb_info *sbi); 3654 int __init f2fs_create_node_manager_caches(void); 3655 void f2fs_destroy_node_manager_caches(void); 3656 3657 /* 3658 * segment.c 3659 */ 3660 bool f2fs_need_SSR(struct f2fs_sb_info *sbi); 3661 int f2fs_commit_atomic_write(struct inode *inode); 3662 void f2fs_abort_atomic_write(struct inode *inode, bool clean); 3663 void f2fs_balance_fs(struct f2fs_sb_info *sbi, bool need); 3664 void f2fs_balance_fs_bg(struct f2fs_sb_info *sbi, bool from_bg); 3665 int f2fs_issue_flush(struct f2fs_sb_info *sbi, nid_t ino); 3666 int f2fs_create_flush_cmd_control(struct f2fs_sb_info *sbi); 3667 int f2fs_flush_device_cache(struct f2fs_sb_info *sbi); 3668 void f2fs_destroy_flush_cmd_control(struct f2fs_sb_info *sbi, bool free); 3669 void f2fs_invalidate_blocks(struct f2fs_sb_info *sbi, block_t addr); 3670 bool f2fs_is_checkpointed_data(struct f2fs_sb_info *sbi, block_t blkaddr); 3671 int f2fs_start_discard_thread(struct f2fs_sb_info *sbi); 3672 void f2fs_drop_discard_cmd(struct f2fs_sb_info *sbi); 3673 void f2fs_stop_discard_thread(struct f2fs_sb_info *sbi); 3674 bool f2fs_issue_discard_timeout(struct f2fs_sb_info *sbi); 3675 void f2fs_clear_prefree_segments(struct f2fs_sb_info *sbi, 3676 struct cp_control *cpc); 3677 void f2fs_dirty_to_prefree(struct f2fs_sb_info *sbi); 3678 block_t f2fs_get_unusable_blocks(struct f2fs_sb_info *sbi); 3679 int f2fs_disable_cp_again(struct f2fs_sb_info *sbi, block_t unusable); 3680 void f2fs_release_discard_addrs(struct f2fs_sb_info *sbi); 3681 int f2fs_npages_for_summary_flush(struct f2fs_sb_info *sbi, bool for_ra); 3682 bool f2fs_segment_has_free_slot(struct f2fs_sb_info *sbi, int segno); 3683 void f2fs_init_inmem_curseg(struct f2fs_sb_info *sbi); 3684 void f2fs_save_inmem_curseg(struct f2fs_sb_info *sbi); 3685 void f2fs_restore_inmem_curseg(struct f2fs_sb_info *sbi); 3686 void f2fs_get_new_segment(struct f2fs_sb_info *sbi, 3687 unsigned int *newseg, bool new_sec, int dir); 3688 void f2fs_allocate_segment_for_resize(struct f2fs_sb_info *sbi, int type, 3689 unsigned int start, unsigned int end); 3690 int f2fs_allocate_new_section(struct f2fs_sb_info *sbi, int type, bool force); 3691 int f2fs_allocate_pinning_section(struct f2fs_sb_info *sbi); 3692 void f2fs_allocate_new_segments(struct f2fs_sb_info *sbi); 3693 int f2fs_trim_fs(struct f2fs_sb_info *sbi, struct fstrim_range *range); 3694 bool f2fs_exist_trim_candidates(struct f2fs_sb_info *sbi, 3695 struct cp_control *cpc); 3696 struct page *f2fs_get_sum_page(struct f2fs_sb_info *sbi, unsigned int segno); 3697 void f2fs_update_meta_page(struct f2fs_sb_info *sbi, void *src, 3698 block_t blk_addr); 3699 void f2fs_do_write_meta_page(struct f2fs_sb_info *sbi, struct page *page, 3700 enum iostat_type io_type); 3701 void f2fs_do_write_node_page(unsigned int nid, struct f2fs_io_info *fio); 3702 void f2fs_outplace_write_data(struct dnode_of_data *dn, 3703 struct f2fs_io_info *fio); 3704 int f2fs_inplace_write_data(struct f2fs_io_info *fio); 3705 void f2fs_do_replace_block(struct f2fs_sb_info *sbi, struct f2fs_summary *sum, 3706 block_t old_blkaddr, block_t new_blkaddr, 3707 bool recover_curseg, bool recover_newaddr, 3708 bool from_gc); 3709 void f2fs_replace_block(struct f2fs_sb_info *sbi, struct dnode_of_data *dn, 3710 block_t old_addr, block_t new_addr, 3711 unsigned char version, bool recover_curseg, 3712 bool recover_newaddr); 3713 void f2fs_allocate_data_block(struct f2fs_sb_info *sbi, struct page *page, 3714 block_t old_blkaddr, block_t *new_blkaddr, 3715 struct f2fs_summary *sum, int type, 3716 struct f2fs_io_info *fio); 3717 void f2fs_update_device_state(struct f2fs_sb_info *sbi, nid_t ino, 3718 block_t blkaddr, unsigned int blkcnt); 3719 void f2fs_wait_on_page_writeback(struct page *page, 3720 enum page_type type, bool ordered, bool locked); 3721 void f2fs_wait_on_block_writeback(struct inode *inode, block_t blkaddr); 3722 void f2fs_wait_on_block_writeback_range(struct inode *inode, block_t blkaddr, 3723 block_t len); 3724 void f2fs_write_data_summaries(struct f2fs_sb_info *sbi, block_t start_blk); 3725 void f2fs_write_node_summaries(struct f2fs_sb_info *sbi, block_t start_blk); 3726 int f2fs_lookup_journal_in_cursum(struct f2fs_journal *journal, int type, 3727 unsigned int val, int alloc); 3728 void f2fs_flush_sit_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3729 int f2fs_fix_curseg_write_pointer(struct f2fs_sb_info *sbi); 3730 int f2fs_check_write_pointer(struct f2fs_sb_info *sbi); 3731 int f2fs_build_segment_manager(struct f2fs_sb_info *sbi); 3732 void f2fs_destroy_segment_manager(struct f2fs_sb_info *sbi); 3733 int __init f2fs_create_segment_manager_caches(void); 3734 void f2fs_destroy_segment_manager_caches(void); 3735 int f2fs_rw_hint_to_seg_type(enum rw_hint hint); 3736 unsigned int f2fs_usable_segs_in_sec(struct f2fs_sb_info *sbi, 3737 unsigned int segno); 3738 unsigned int f2fs_usable_blks_in_seg(struct f2fs_sb_info *sbi, 3739 unsigned int segno); 3740 3741 #define DEF_FRAGMENT_SIZE 4 3742 #define MIN_FRAGMENT_SIZE 1 3743 #define MAX_FRAGMENT_SIZE 512 3744 3745 static inline bool f2fs_need_rand_seg(struct f2fs_sb_info *sbi) 3746 { 3747 return F2FS_OPTION(sbi).fs_mode == FS_MODE_FRAGMENT_SEG || 3748 F2FS_OPTION(sbi).fs_mode == FS_MODE_FRAGMENT_BLK; 3749 } 3750 3751 /* 3752 * checkpoint.c 3753 */ 3754 void f2fs_stop_checkpoint(struct f2fs_sb_info *sbi, bool end_io, 3755 unsigned char reason); 3756 void f2fs_flush_ckpt_thread(struct f2fs_sb_info *sbi); 3757 struct page *f2fs_grab_meta_page(struct f2fs_sb_info *sbi, pgoff_t index); 3758 struct page *f2fs_get_meta_page(struct f2fs_sb_info *sbi, pgoff_t index); 3759 struct page *f2fs_get_meta_page_retry(struct f2fs_sb_info *sbi, pgoff_t index); 3760 struct page *f2fs_get_tmp_page(struct f2fs_sb_info *sbi, pgoff_t index); 3761 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi, 3762 block_t blkaddr, int type); 3763 int f2fs_ra_meta_pages(struct f2fs_sb_info *sbi, block_t start, int nrpages, 3764 int type, bool sync); 3765 void f2fs_ra_meta_pages_cond(struct f2fs_sb_info *sbi, pgoff_t index, 3766 unsigned int ra_blocks); 3767 long f2fs_sync_meta_pages(struct f2fs_sb_info *sbi, enum page_type type, 3768 long nr_to_write, enum iostat_type io_type); 3769 void f2fs_add_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type); 3770 void f2fs_remove_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type); 3771 void f2fs_release_ino_entry(struct f2fs_sb_info *sbi, bool all); 3772 bool f2fs_exist_written_data(struct f2fs_sb_info *sbi, nid_t ino, int mode); 3773 void f2fs_set_dirty_device(struct f2fs_sb_info *sbi, nid_t ino, 3774 unsigned int devidx, int type); 3775 bool f2fs_is_dirty_device(struct f2fs_sb_info *sbi, nid_t ino, 3776 unsigned int devidx, int type); 3777 int f2fs_acquire_orphan_inode(struct f2fs_sb_info *sbi); 3778 void f2fs_release_orphan_inode(struct f2fs_sb_info *sbi); 3779 void f2fs_add_orphan_inode(struct inode *inode); 3780 void f2fs_remove_orphan_inode(struct f2fs_sb_info *sbi, nid_t ino); 3781 int f2fs_recover_orphan_inodes(struct f2fs_sb_info *sbi); 3782 int f2fs_get_valid_checkpoint(struct f2fs_sb_info *sbi); 3783 void f2fs_update_dirty_folio(struct inode *inode, struct folio *folio); 3784 void f2fs_remove_dirty_inode(struct inode *inode); 3785 int f2fs_sync_dirty_inodes(struct f2fs_sb_info *sbi, enum inode_type type, 3786 bool from_cp); 3787 void f2fs_wait_on_all_pages(struct f2fs_sb_info *sbi, int type); 3788 u64 f2fs_get_sectors_written(struct f2fs_sb_info *sbi); 3789 int f2fs_write_checkpoint(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3790 void f2fs_init_ino_entry_info(struct f2fs_sb_info *sbi); 3791 int __init f2fs_create_checkpoint_caches(void); 3792 void f2fs_destroy_checkpoint_caches(void); 3793 int f2fs_issue_checkpoint(struct f2fs_sb_info *sbi); 3794 int f2fs_start_ckpt_thread(struct f2fs_sb_info *sbi); 3795 void f2fs_stop_ckpt_thread(struct f2fs_sb_info *sbi); 3796 void f2fs_init_ckpt_req_control(struct f2fs_sb_info *sbi); 3797 3798 /* 3799 * data.c 3800 */ 3801 int __init f2fs_init_bioset(void); 3802 void f2fs_destroy_bioset(void); 3803 bool f2fs_is_cp_guaranteed(struct page *page); 3804 int f2fs_init_bio_entry_cache(void); 3805 void f2fs_destroy_bio_entry_cache(void); 3806 void f2fs_submit_read_bio(struct f2fs_sb_info *sbi, struct bio *bio, 3807 enum page_type type); 3808 int f2fs_init_write_merge_io(struct f2fs_sb_info *sbi); 3809 void f2fs_submit_merged_write(struct f2fs_sb_info *sbi, enum page_type type); 3810 void f2fs_submit_merged_write_cond(struct f2fs_sb_info *sbi, 3811 struct inode *inode, struct page *page, 3812 nid_t ino, enum page_type type); 3813 void f2fs_submit_merged_ipu_write(struct f2fs_sb_info *sbi, 3814 struct bio **bio, struct page *page); 3815 void f2fs_flush_merged_writes(struct f2fs_sb_info *sbi); 3816 int f2fs_submit_page_bio(struct f2fs_io_info *fio); 3817 int f2fs_merge_page_bio(struct f2fs_io_info *fio); 3818 void f2fs_submit_page_write(struct f2fs_io_info *fio); 3819 struct block_device *f2fs_target_device(struct f2fs_sb_info *sbi, 3820 block_t blk_addr, sector_t *sector); 3821 int f2fs_target_device_index(struct f2fs_sb_info *sbi, block_t blkaddr); 3822 void f2fs_set_data_blkaddr(struct dnode_of_data *dn, block_t blkaddr); 3823 void f2fs_update_data_blkaddr(struct dnode_of_data *dn, block_t blkaddr); 3824 int f2fs_reserve_new_blocks(struct dnode_of_data *dn, blkcnt_t count); 3825 int f2fs_reserve_new_block(struct dnode_of_data *dn); 3826 int f2fs_get_block_locked(struct dnode_of_data *dn, pgoff_t index); 3827 int f2fs_reserve_block(struct dnode_of_data *dn, pgoff_t index); 3828 struct page *f2fs_get_read_data_page(struct inode *inode, pgoff_t index, 3829 blk_opf_t op_flags, bool for_write, pgoff_t *next_pgofs); 3830 struct page *f2fs_find_data_page(struct inode *inode, pgoff_t index, 3831 pgoff_t *next_pgofs); 3832 struct page *f2fs_get_lock_data_page(struct inode *inode, pgoff_t index, 3833 bool for_write); 3834 struct page *f2fs_get_new_data_page(struct inode *inode, 3835 struct page *ipage, pgoff_t index, bool new_i_size); 3836 int f2fs_do_write_data_page(struct f2fs_io_info *fio); 3837 int f2fs_map_blocks(struct inode *inode, struct f2fs_map_blocks *map, int flag); 3838 int f2fs_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, 3839 u64 start, u64 len); 3840 int f2fs_encrypt_one_page(struct f2fs_io_info *fio); 3841 bool f2fs_should_update_inplace(struct inode *inode, struct f2fs_io_info *fio); 3842 bool f2fs_should_update_outplace(struct inode *inode, struct f2fs_io_info *fio); 3843 int f2fs_write_single_data_page(struct page *page, int *submitted, 3844 struct bio **bio, sector_t *last_block, 3845 struct writeback_control *wbc, 3846 enum iostat_type io_type, 3847 int compr_blocks, bool allow_balance); 3848 void f2fs_write_failed(struct inode *inode, loff_t to); 3849 void f2fs_invalidate_folio(struct folio *folio, size_t offset, size_t length); 3850 bool f2fs_release_folio(struct folio *folio, gfp_t wait); 3851 bool f2fs_overwrite_io(struct inode *inode, loff_t pos, size_t len); 3852 void f2fs_clear_page_cache_dirty_tag(struct page *page); 3853 int f2fs_init_post_read_processing(void); 3854 void f2fs_destroy_post_read_processing(void); 3855 int f2fs_init_post_read_wq(struct f2fs_sb_info *sbi); 3856 void f2fs_destroy_post_read_wq(struct f2fs_sb_info *sbi); 3857 extern const struct iomap_ops f2fs_iomap_ops; 3858 3859 /* 3860 * gc.c 3861 */ 3862 int f2fs_start_gc_thread(struct f2fs_sb_info *sbi); 3863 void f2fs_stop_gc_thread(struct f2fs_sb_info *sbi); 3864 block_t f2fs_start_bidx_of_node(unsigned int node_ofs, struct inode *inode); 3865 int f2fs_gc(struct f2fs_sb_info *sbi, struct f2fs_gc_control *gc_control); 3866 void f2fs_build_gc_manager(struct f2fs_sb_info *sbi); 3867 int f2fs_gc_range(struct f2fs_sb_info *sbi, 3868 unsigned int start_seg, unsigned int end_seg, 3869 bool dry_run, unsigned int dry_run_sections); 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, struct page *ipage); 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, struct page *ipage); 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 static inline bool f2fs_used_in_atomic_write(struct inode *inode) 4251 { 4252 return f2fs_is_atomic_file(inode) || f2fs_is_cow_file(inode); 4253 } 4254 4255 static inline bool f2fs_meta_inode_gc_required(struct inode *inode) 4256 { 4257 return f2fs_post_read_required(inode) || f2fs_used_in_atomic_write(inode); 4258 } 4259 4260 /* 4261 * compress.c 4262 */ 4263 #ifdef CONFIG_F2FS_FS_COMPRESSION 4264 enum cluster_check_type { 4265 CLUSTER_IS_COMPR, /* check only if compressed cluster */ 4266 CLUSTER_COMPR_BLKS, /* return # of compressed blocks in a cluster */ 4267 CLUSTER_RAW_BLKS /* return # of raw blocks in a cluster */ 4268 }; 4269 bool f2fs_is_compressed_page(struct page *page); 4270 struct page *f2fs_compress_control_page(struct page *page); 4271 int f2fs_prepare_compress_overwrite(struct inode *inode, 4272 struct page **pagep, pgoff_t index, void **fsdata); 4273 bool f2fs_compress_write_end(struct inode *inode, void *fsdata, 4274 pgoff_t index, unsigned copied); 4275 int f2fs_truncate_partial_cluster(struct inode *inode, u64 from, bool lock); 4276 void f2fs_compress_write_end_io(struct bio *bio, struct page *page); 4277 bool f2fs_is_compress_backend_ready(struct inode *inode); 4278 bool f2fs_is_compress_level_valid(int alg, int lvl); 4279 int __init f2fs_init_compress_mempool(void); 4280 void f2fs_destroy_compress_mempool(void); 4281 void f2fs_decompress_cluster(struct decompress_io_ctx *dic, bool in_task); 4282 void f2fs_end_read_compressed_page(struct page *page, bool failed, 4283 block_t blkaddr, bool in_task); 4284 bool f2fs_cluster_is_empty(struct compress_ctx *cc); 4285 bool f2fs_cluster_can_merge_page(struct compress_ctx *cc, pgoff_t index); 4286 bool f2fs_all_cluster_page_ready(struct compress_ctx *cc, struct page **pages, 4287 int index, int nr_pages, bool uptodate); 4288 bool f2fs_sanity_check_cluster(struct dnode_of_data *dn); 4289 void f2fs_compress_ctx_add_page(struct compress_ctx *cc, struct page *page); 4290 int f2fs_write_multi_pages(struct compress_ctx *cc, 4291 int *submitted, 4292 struct writeback_control *wbc, 4293 enum iostat_type io_type); 4294 int f2fs_is_compressed_cluster(struct inode *inode, pgoff_t index); 4295 bool f2fs_is_sparse_cluster(struct inode *inode, pgoff_t index); 4296 void f2fs_update_read_extent_tree_range_compressed(struct inode *inode, 4297 pgoff_t fofs, block_t blkaddr, 4298 unsigned int llen, unsigned int c_len); 4299 int f2fs_read_multi_pages(struct compress_ctx *cc, struct bio **bio_ret, 4300 unsigned nr_pages, sector_t *last_block_in_bio, 4301 bool is_readahead, bool for_write); 4302 struct decompress_io_ctx *f2fs_alloc_dic(struct compress_ctx *cc); 4303 void f2fs_decompress_end_io(struct decompress_io_ctx *dic, bool failed, 4304 bool in_task); 4305 void f2fs_put_page_dic(struct page *page, bool in_task); 4306 unsigned int f2fs_cluster_blocks_are_contiguous(struct dnode_of_data *dn, 4307 unsigned int ofs_in_node); 4308 int f2fs_init_compress_ctx(struct compress_ctx *cc); 4309 void f2fs_destroy_compress_ctx(struct compress_ctx *cc, bool reuse); 4310 void f2fs_init_compress_info(struct f2fs_sb_info *sbi); 4311 int f2fs_init_compress_inode(struct f2fs_sb_info *sbi); 4312 void f2fs_destroy_compress_inode(struct f2fs_sb_info *sbi); 4313 int f2fs_init_page_array_cache(struct f2fs_sb_info *sbi); 4314 void f2fs_destroy_page_array_cache(struct f2fs_sb_info *sbi); 4315 int __init f2fs_init_compress_cache(void); 4316 void f2fs_destroy_compress_cache(void); 4317 struct address_space *COMPRESS_MAPPING(struct f2fs_sb_info *sbi); 4318 void f2fs_invalidate_compress_page(struct f2fs_sb_info *sbi, block_t blkaddr); 4319 void f2fs_cache_compressed_page(struct f2fs_sb_info *sbi, struct page *page, 4320 nid_t ino, block_t blkaddr); 4321 bool f2fs_load_compressed_page(struct f2fs_sb_info *sbi, struct page *page, 4322 block_t blkaddr); 4323 void f2fs_invalidate_compress_pages(struct f2fs_sb_info *sbi, nid_t ino); 4324 #define inc_compr_inode_stat(inode) \ 4325 do { \ 4326 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); \ 4327 sbi->compr_new_inode++; \ 4328 } while (0) 4329 #define add_compr_block_stat(inode, blocks) \ 4330 do { \ 4331 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); \ 4332 int diff = F2FS_I(inode)->i_cluster_size - blocks; \ 4333 sbi->compr_written_block += blocks; \ 4334 sbi->compr_saved_block += diff; \ 4335 } while (0) 4336 #else 4337 static inline bool f2fs_is_compressed_page(struct page *page) { return false; } 4338 static inline bool f2fs_is_compress_backend_ready(struct inode *inode) 4339 { 4340 if (!f2fs_compressed_file(inode)) 4341 return true; 4342 /* not support compression */ 4343 return false; 4344 } 4345 static inline bool f2fs_is_compress_level_valid(int alg, int lvl) { return false; } 4346 static inline struct page *f2fs_compress_control_page(struct page *page) 4347 { 4348 WARN_ON_ONCE(1); 4349 return ERR_PTR(-EINVAL); 4350 } 4351 static inline int __init f2fs_init_compress_mempool(void) { return 0; } 4352 static inline void f2fs_destroy_compress_mempool(void) { } 4353 static inline void f2fs_decompress_cluster(struct decompress_io_ctx *dic, 4354 bool in_task) { } 4355 static inline void f2fs_end_read_compressed_page(struct page *page, 4356 bool failed, block_t blkaddr, bool in_task) 4357 { 4358 WARN_ON_ONCE(1); 4359 } 4360 static inline void f2fs_put_page_dic(struct page *page, bool in_task) 4361 { 4362 WARN_ON_ONCE(1); 4363 } 4364 static inline unsigned int f2fs_cluster_blocks_are_contiguous( 4365 struct dnode_of_data *dn, unsigned int ofs_in_node) { return 0; } 4366 static inline bool f2fs_sanity_check_cluster(struct dnode_of_data *dn) { return false; } 4367 static inline int f2fs_init_compress_inode(struct f2fs_sb_info *sbi) { return 0; } 4368 static inline void f2fs_destroy_compress_inode(struct f2fs_sb_info *sbi) { } 4369 static inline int f2fs_init_page_array_cache(struct f2fs_sb_info *sbi) { return 0; } 4370 static inline void f2fs_destroy_page_array_cache(struct f2fs_sb_info *sbi) { } 4371 static inline int __init f2fs_init_compress_cache(void) { return 0; } 4372 static inline void f2fs_destroy_compress_cache(void) { } 4373 static inline void f2fs_invalidate_compress_page(struct f2fs_sb_info *sbi, 4374 block_t blkaddr) { } 4375 static inline void f2fs_cache_compressed_page(struct f2fs_sb_info *sbi, 4376 struct page *page, nid_t ino, block_t blkaddr) { } 4377 static inline bool f2fs_load_compressed_page(struct f2fs_sb_info *sbi, 4378 struct page *page, block_t blkaddr) { return false; } 4379 static inline void f2fs_invalidate_compress_pages(struct f2fs_sb_info *sbi, 4380 nid_t ino) { } 4381 #define inc_compr_inode_stat(inode) do { } while (0) 4382 static inline int f2fs_is_compressed_cluster( 4383 struct inode *inode, 4384 pgoff_t index) { return 0; } 4385 static inline bool f2fs_is_sparse_cluster( 4386 struct inode *inode, 4387 pgoff_t index) { return true; } 4388 static inline void f2fs_update_read_extent_tree_range_compressed( 4389 struct inode *inode, 4390 pgoff_t fofs, block_t blkaddr, 4391 unsigned int llen, unsigned int c_len) { } 4392 #endif 4393 4394 static inline int set_compress_context(struct inode *inode) 4395 { 4396 #ifdef CONFIG_F2FS_FS_COMPRESSION 4397 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 4398 4399 F2FS_I(inode)->i_compress_algorithm = 4400 F2FS_OPTION(sbi).compress_algorithm; 4401 F2FS_I(inode)->i_log_cluster_size = 4402 F2FS_OPTION(sbi).compress_log_size; 4403 F2FS_I(inode)->i_compress_flag = 4404 F2FS_OPTION(sbi).compress_chksum ? 4405 BIT(COMPRESS_CHKSUM) : 0; 4406 F2FS_I(inode)->i_cluster_size = 4407 BIT(F2FS_I(inode)->i_log_cluster_size); 4408 if ((F2FS_I(inode)->i_compress_algorithm == COMPRESS_LZ4 || 4409 F2FS_I(inode)->i_compress_algorithm == COMPRESS_ZSTD) && 4410 F2FS_OPTION(sbi).compress_level) 4411 F2FS_I(inode)->i_compress_level = 4412 F2FS_OPTION(sbi).compress_level; 4413 F2FS_I(inode)->i_flags |= F2FS_COMPR_FL; 4414 set_inode_flag(inode, FI_COMPRESSED_FILE); 4415 stat_inc_compr_inode(inode); 4416 inc_compr_inode_stat(inode); 4417 f2fs_mark_inode_dirty_sync(inode, true); 4418 return 0; 4419 #else 4420 return -EOPNOTSUPP; 4421 #endif 4422 } 4423 4424 static inline bool f2fs_disable_compressed_file(struct inode *inode) 4425 { 4426 struct f2fs_inode_info *fi = F2FS_I(inode); 4427 4428 f2fs_down_write(&F2FS_I(inode)->i_sem); 4429 4430 if (!f2fs_compressed_file(inode)) { 4431 f2fs_up_write(&F2FS_I(inode)->i_sem); 4432 return true; 4433 } 4434 if (f2fs_is_mmap_file(inode) || 4435 (S_ISREG(inode->i_mode) && F2FS_HAS_BLOCKS(inode))) { 4436 f2fs_up_write(&F2FS_I(inode)->i_sem); 4437 return false; 4438 } 4439 4440 fi->i_flags &= ~F2FS_COMPR_FL; 4441 stat_dec_compr_inode(inode); 4442 clear_inode_flag(inode, FI_COMPRESSED_FILE); 4443 f2fs_mark_inode_dirty_sync(inode, true); 4444 4445 f2fs_up_write(&F2FS_I(inode)->i_sem); 4446 return true; 4447 } 4448 4449 #define F2FS_FEATURE_FUNCS(name, flagname) \ 4450 static inline bool f2fs_sb_has_##name(struct f2fs_sb_info *sbi) \ 4451 { \ 4452 return F2FS_HAS_FEATURE(sbi, F2FS_FEATURE_##flagname); \ 4453 } 4454 4455 F2FS_FEATURE_FUNCS(encrypt, ENCRYPT); 4456 F2FS_FEATURE_FUNCS(blkzoned, BLKZONED); 4457 F2FS_FEATURE_FUNCS(extra_attr, EXTRA_ATTR); 4458 F2FS_FEATURE_FUNCS(project_quota, PRJQUOTA); 4459 F2FS_FEATURE_FUNCS(inode_chksum, INODE_CHKSUM); 4460 F2FS_FEATURE_FUNCS(flexible_inline_xattr, FLEXIBLE_INLINE_XATTR); 4461 F2FS_FEATURE_FUNCS(quota_ino, QUOTA_INO); 4462 F2FS_FEATURE_FUNCS(inode_crtime, INODE_CRTIME); 4463 F2FS_FEATURE_FUNCS(lost_found, LOST_FOUND); 4464 F2FS_FEATURE_FUNCS(verity, VERITY); 4465 F2FS_FEATURE_FUNCS(sb_chksum, SB_CHKSUM); 4466 F2FS_FEATURE_FUNCS(casefold, CASEFOLD); 4467 F2FS_FEATURE_FUNCS(compression, COMPRESSION); 4468 F2FS_FEATURE_FUNCS(readonly, RO); 4469 4470 #ifdef CONFIG_BLK_DEV_ZONED 4471 static inline bool f2fs_blkz_is_seq(struct f2fs_sb_info *sbi, int devi, 4472 block_t blkaddr) 4473 { 4474 unsigned int zno = blkaddr / sbi->blocks_per_blkz; 4475 4476 return test_bit(zno, FDEV(devi).blkz_seq); 4477 } 4478 #endif 4479 4480 static inline int f2fs_bdev_index(struct f2fs_sb_info *sbi, 4481 struct block_device *bdev) 4482 { 4483 int i; 4484 4485 if (!f2fs_is_multi_device(sbi)) 4486 return 0; 4487 4488 for (i = 0; i < sbi->s_ndevs; i++) 4489 if (FDEV(i).bdev == bdev) 4490 return i; 4491 4492 WARN_ON(1); 4493 return -1; 4494 } 4495 4496 static inline bool f2fs_hw_should_discard(struct f2fs_sb_info *sbi) 4497 { 4498 return f2fs_sb_has_blkzoned(sbi); 4499 } 4500 4501 static inline bool f2fs_bdev_support_discard(struct block_device *bdev) 4502 { 4503 return bdev_max_discard_sectors(bdev) || bdev_is_zoned(bdev); 4504 } 4505 4506 static inline bool f2fs_hw_support_discard(struct f2fs_sb_info *sbi) 4507 { 4508 int i; 4509 4510 if (!f2fs_is_multi_device(sbi)) 4511 return f2fs_bdev_support_discard(sbi->sb->s_bdev); 4512 4513 for (i = 0; i < sbi->s_ndevs; i++) 4514 if (f2fs_bdev_support_discard(FDEV(i).bdev)) 4515 return true; 4516 return false; 4517 } 4518 4519 static inline bool f2fs_realtime_discard_enable(struct f2fs_sb_info *sbi) 4520 { 4521 return (test_opt(sbi, DISCARD) && f2fs_hw_support_discard(sbi)) || 4522 f2fs_hw_should_discard(sbi); 4523 } 4524 4525 static inline bool f2fs_hw_is_readonly(struct f2fs_sb_info *sbi) 4526 { 4527 int i; 4528 4529 if (!f2fs_is_multi_device(sbi)) 4530 return bdev_read_only(sbi->sb->s_bdev); 4531 4532 for (i = 0; i < sbi->s_ndevs; i++) 4533 if (bdev_read_only(FDEV(i).bdev)) 4534 return true; 4535 return false; 4536 } 4537 4538 static inline bool f2fs_dev_is_readonly(struct f2fs_sb_info *sbi) 4539 { 4540 return f2fs_sb_has_readonly(sbi) || f2fs_hw_is_readonly(sbi); 4541 } 4542 4543 static inline bool f2fs_lfs_mode(struct f2fs_sb_info *sbi) 4544 { 4545 return F2FS_OPTION(sbi).fs_mode == FS_MODE_LFS; 4546 } 4547 4548 static inline bool f2fs_valid_pinned_area(struct f2fs_sb_info *sbi, 4549 block_t blkaddr) 4550 { 4551 if (f2fs_sb_has_blkzoned(sbi)) { 4552 int devi = f2fs_target_device_index(sbi, blkaddr); 4553 4554 return !bdev_is_zoned(FDEV(devi).bdev); 4555 } 4556 return true; 4557 } 4558 4559 static inline bool f2fs_low_mem_mode(struct f2fs_sb_info *sbi) 4560 { 4561 return F2FS_OPTION(sbi).memory_mode == MEMORY_MODE_LOW; 4562 } 4563 4564 static inline bool f2fs_may_compress(struct inode *inode) 4565 { 4566 if (IS_SWAPFILE(inode) || f2fs_is_pinned_file(inode) || 4567 f2fs_is_atomic_file(inode) || f2fs_has_inline_data(inode) || 4568 f2fs_is_mmap_file(inode)) 4569 return false; 4570 return S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode); 4571 } 4572 4573 static inline void f2fs_i_compr_blocks_update(struct inode *inode, 4574 u64 blocks, bool add) 4575 { 4576 struct f2fs_inode_info *fi = F2FS_I(inode); 4577 int diff = fi->i_cluster_size - blocks; 4578 4579 /* don't update i_compr_blocks if saved blocks were released */ 4580 if (!add && !atomic_read(&fi->i_compr_blocks)) 4581 return; 4582 4583 if (add) { 4584 atomic_add(diff, &fi->i_compr_blocks); 4585 stat_add_compr_blocks(inode, diff); 4586 } else { 4587 atomic_sub(diff, &fi->i_compr_blocks); 4588 stat_sub_compr_blocks(inode, diff); 4589 } 4590 f2fs_mark_inode_dirty_sync(inode, true); 4591 } 4592 4593 static inline bool f2fs_allow_multi_device_dio(struct f2fs_sb_info *sbi, 4594 int flag) 4595 { 4596 if (!f2fs_is_multi_device(sbi)) 4597 return false; 4598 if (flag != F2FS_GET_BLOCK_DIO) 4599 return false; 4600 return sbi->aligned_blksize; 4601 } 4602 4603 static inline bool f2fs_need_verity(const struct inode *inode, pgoff_t idx) 4604 { 4605 return fsverity_active(inode) && 4606 idx < DIV_ROUND_UP(inode->i_size, PAGE_SIZE); 4607 } 4608 4609 #ifdef CONFIG_F2FS_FAULT_INJECTION 4610 extern int f2fs_build_fault_attr(struct f2fs_sb_info *sbi, unsigned long rate, 4611 unsigned long type); 4612 #else 4613 static inline int f2fs_build_fault_attr(struct f2fs_sb_info *sbi, 4614 unsigned long rate, unsigned long type) 4615 { 4616 return 0; 4617 } 4618 #endif 4619 4620 static inline bool is_journalled_quota(struct f2fs_sb_info *sbi) 4621 { 4622 #ifdef CONFIG_QUOTA 4623 if (f2fs_sb_has_quota_ino(sbi)) 4624 return true; 4625 if (F2FS_OPTION(sbi).s_qf_names[USRQUOTA] || 4626 F2FS_OPTION(sbi).s_qf_names[GRPQUOTA] || 4627 F2FS_OPTION(sbi).s_qf_names[PRJQUOTA]) 4628 return true; 4629 #endif 4630 return false; 4631 } 4632 4633 static inline bool f2fs_block_unit_discard(struct f2fs_sb_info *sbi) 4634 { 4635 return F2FS_OPTION(sbi).discard_unit == DISCARD_UNIT_BLOCK; 4636 } 4637 4638 static inline void f2fs_io_schedule_timeout(long timeout) 4639 { 4640 set_current_state(TASK_UNINTERRUPTIBLE); 4641 io_schedule_timeout(timeout); 4642 } 4643 4644 static inline void f2fs_handle_page_eio(struct f2fs_sb_info *sbi, pgoff_t ofs, 4645 enum page_type type) 4646 { 4647 if (unlikely(f2fs_cp_error(sbi))) 4648 return; 4649 4650 if (ofs == sbi->page_eio_ofs[type]) { 4651 if (sbi->page_eio_cnt[type]++ == MAX_RETRY_PAGE_EIO) 4652 set_ckpt_flags(sbi, CP_ERROR_FLAG); 4653 } else { 4654 sbi->page_eio_ofs[type] = ofs; 4655 sbi->page_eio_cnt[type] = 0; 4656 } 4657 } 4658 4659 static inline bool f2fs_is_readonly(struct f2fs_sb_info *sbi) 4660 { 4661 return f2fs_sb_has_readonly(sbi) || f2fs_readonly(sbi->sb); 4662 } 4663 4664 static inline void f2fs_truncate_meta_inode_pages(struct f2fs_sb_info *sbi, 4665 block_t blkaddr, unsigned int cnt) 4666 { 4667 bool need_submit = false; 4668 int i = 0; 4669 4670 do { 4671 struct page *page; 4672 4673 page = find_get_page(META_MAPPING(sbi), blkaddr + i); 4674 if (page) { 4675 if (PageWriteback(page)) 4676 need_submit = true; 4677 f2fs_put_page(page, 0); 4678 } 4679 } while (++i < cnt && !need_submit); 4680 4681 if (need_submit) 4682 f2fs_submit_merged_write_cond(sbi, sbi->meta_inode, 4683 NULL, 0, DATA); 4684 4685 truncate_inode_pages_range(META_MAPPING(sbi), 4686 F2FS_BLK_TO_BYTES((loff_t)blkaddr), 4687 F2FS_BLK_END_BYTES((loff_t)(blkaddr + cnt - 1))); 4688 } 4689 4690 static inline void f2fs_invalidate_internal_cache(struct f2fs_sb_info *sbi, 4691 block_t blkaddr) 4692 { 4693 f2fs_truncate_meta_inode_pages(sbi, blkaddr, 1); 4694 f2fs_invalidate_compress_page(sbi, blkaddr); 4695 } 4696 4697 #define EFSBADCRC EBADMSG /* Bad CRC detected */ 4698 #define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */ 4699 4700 #endif /* _LINUX_F2FS_H */ 4701