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 <linux/rw_hint.h> 28 #include <crypto/hash.h> 29 30 #include <linux/fscrypt.h> 31 #include <linux/fsverity.h> 32 33 struct pagevec; 34 35 #ifdef CONFIG_F2FS_CHECK_FS 36 #define f2fs_bug_on(sbi, condition) BUG_ON(condition) 37 #else 38 #define f2fs_bug_on(sbi, condition) \ 39 do { \ 40 if (WARN_ON(condition)) \ 41 set_sbi_flag(sbi, SBI_NEED_FSCK); \ 42 } while (0) 43 #endif 44 45 enum { 46 FAULT_KMALLOC, 47 FAULT_KVMALLOC, 48 FAULT_PAGE_ALLOC, 49 FAULT_PAGE_GET, 50 FAULT_ALLOC_BIO, /* it's obsolete due to bio_alloc() will never fail */ 51 FAULT_ALLOC_NID, 52 FAULT_ORPHAN, 53 FAULT_BLOCK, 54 FAULT_DIR_DEPTH, 55 FAULT_EVICT_INODE, 56 FAULT_TRUNCATE, 57 FAULT_READ_IO, 58 FAULT_CHECKPOINT, 59 FAULT_DISCARD, 60 FAULT_WRITE_IO, 61 FAULT_SLAB_ALLOC, 62 FAULT_DQUOT_INIT, 63 FAULT_LOCK_OP, 64 FAULT_BLKADDR_VALIDITY, 65 FAULT_BLKADDR_CONSISTENCE, 66 FAULT_NO_SEGMENT, 67 FAULT_MAX, 68 }; 69 70 #ifdef CONFIG_F2FS_FAULT_INJECTION 71 #define F2FS_ALL_FAULT_TYPE (GENMASK(FAULT_MAX - 1, 0)) 72 73 struct f2fs_fault_info { 74 atomic_t inject_ops; 75 unsigned int inject_rate; 76 unsigned int inject_type; 77 }; 78 79 extern const char *f2fs_fault_name[FAULT_MAX]; 80 #define IS_FAULT_SET(fi, type) ((fi)->inject_type & BIT(type)) 81 82 /* maximum retry count for injected failure */ 83 #define DEFAULT_FAILURE_RETRY_COUNT 8 84 #else 85 #define DEFAULT_FAILURE_RETRY_COUNT 1 86 #endif 87 88 /* 89 * For mount options 90 */ 91 #define F2FS_MOUNT_DISABLE_ROLL_FORWARD 0x00000001 92 #define F2FS_MOUNT_DISCARD 0x00000002 93 #define F2FS_MOUNT_NOHEAP 0x00000004 94 #define F2FS_MOUNT_XATTR_USER 0x00000008 95 #define F2FS_MOUNT_POSIX_ACL 0x00000010 96 #define F2FS_MOUNT_DISABLE_EXT_IDENTIFY 0x00000020 97 #define F2FS_MOUNT_INLINE_XATTR 0x00000040 98 #define F2FS_MOUNT_INLINE_DATA 0x00000080 99 #define F2FS_MOUNT_INLINE_DENTRY 0x00000100 100 #define F2FS_MOUNT_FLUSH_MERGE 0x00000200 101 #define F2FS_MOUNT_NOBARRIER 0x00000400 102 #define F2FS_MOUNT_FASTBOOT 0x00000800 103 #define F2FS_MOUNT_READ_EXTENT_CACHE 0x00001000 104 #define F2FS_MOUNT_DATA_FLUSH 0x00002000 105 #define F2FS_MOUNT_FAULT_INJECTION 0x00004000 106 #define F2FS_MOUNT_USRQUOTA 0x00008000 107 #define F2FS_MOUNT_GRPQUOTA 0x00010000 108 #define F2FS_MOUNT_PRJQUOTA 0x00020000 109 #define F2FS_MOUNT_QUOTA 0x00040000 110 #define F2FS_MOUNT_INLINE_XATTR_SIZE 0x00080000 111 #define F2FS_MOUNT_RESERVE_ROOT 0x00100000 112 #define F2FS_MOUNT_DISABLE_CHECKPOINT 0x00200000 113 #define F2FS_MOUNT_NORECOVERY 0x00400000 114 #define F2FS_MOUNT_ATGC 0x00800000 115 #define F2FS_MOUNT_MERGE_CHECKPOINT 0x01000000 116 #define F2FS_MOUNT_GC_MERGE 0x02000000 117 #define F2FS_MOUNT_COMPRESS_CACHE 0x04000000 118 #define F2FS_MOUNT_AGE_EXTENT_CACHE 0x08000000 119 120 #define F2FS_OPTION(sbi) ((sbi)->mount_opt) 121 #define clear_opt(sbi, option) (F2FS_OPTION(sbi).opt &= ~F2FS_MOUNT_##option) 122 #define set_opt(sbi, option) (F2FS_OPTION(sbi).opt |= F2FS_MOUNT_##option) 123 #define test_opt(sbi, option) (F2FS_OPTION(sbi).opt & F2FS_MOUNT_##option) 124 125 #define ver_after(a, b) (typecheck(unsigned long long, a) && \ 126 typecheck(unsigned long long, b) && \ 127 ((long long)((a) - (b)) > 0)) 128 129 typedef u32 block_t; /* 130 * should not change u32, since it is the on-disk block 131 * address format, __le32. 132 */ 133 typedef u32 nid_t; 134 135 #define COMPRESS_EXT_NUM 16 136 137 /* 138 * An implementation of an rwsem that is explicitly unfair to readers. This 139 * prevents priority inversion when a low-priority reader acquires the read lock 140 * while sleeping on the write lock but the write lock is needed by 141 * higher-priority clients. 142 */ 143 144 struct f2fs_rwsem { 145 struct rw_semaphore internal_rwsem; 146 #ifdef CONFIG_F2FS_UNFAIR_RWSEM 147 wait_queue_head_t read_waiters; 148 #endif 149 }; 150 151 struct f2fs_mount_info { 152 unsigned int opt; 153 block_t root_reserved_blocks; /* root reserved blocks */ 154 kuid_t s_resuid; /* reserved blocks for uid */ 155 kgid_t s_resgid; /* reserved blocks for gid */ 156 int active_logs; /* # of active logs */ 157 int inline_xattr_size; /* inline xattr size */ 158 #ifdef CONFIG_F2FS_FAULT_INJECTION 159 struct f2fs_fault_info fault_info; /* For fault injection */ 160 #endif 161 #ifdef CONFIG_QUOTA 162 /* Names of quota files with journalled quota */ 163 char *s_qf_names[MAXQUOTAS]; 164 int s_jquota_fmt; /* Format of quota to use */ 165 #endif 166 /* For which write hints are passed down to block layer */ 167 int alloc_mode; /* segment allocation policy */ 168 int fsync_mode; /* fsync policy */ 169 int fs_mode; /* fs mode: LFS or ADAPTIVE */ 170 int bggc_mode; /* bggc mode: off, on or sync */ 171 int memory_mode; /* memory mode */ 172 int errors; /* errors parameter */ 173 int discard_unit; /* 174 * discard command's offset/size should 175 * be aligned to this unit: block, 176 * segment or section 177 */ 178 struct fscrypt_dummy_policy dummy_enc_policy; /* test dummy encryption */ 179 block_t unusable_cap_perc; /* percentage for cap */ 180 block_t unusable_cap; /* Amount of space allowed to be 181 * unusable when disabling checkpoint 182 */ 183 184 /* For compression */ 185 unsigned char compress_algorithm; /* algorithm type */ 186 unsigned char compress_log_size; /* cluster log size */ 187 unsigned char compress_level; /* compress level */ 188 bool compress_chksum; /* compressed data chksum */ 189 unsigned char compress_ext_cnt; /* extension count */ 190 unsigned char nocompress_ext_cnt; /* nocompress extension count */ 191 int compress_mode; /* compression mode */ 192 unsigned char extensions[COMPRESS_EXT_NUM][F2FS_EXTENSION_LEN]; /* extensions */ 193 unsigned char noextensions[COMPRESS_EXT_NUM][F2FS_EXTENSION_LEN]; /* extensions */ 194 }; 195 196 #define F2FS_FEATURE_ENCRYPT 0x00000001 197 #define F2FS_FEATURE_BLKZONED 0x00000002 198 #define F2FS_FEATURE_ATOMIC_WRITE 0x00000004 199 #define F2FS_FEATURE_EXTRA_ATTR 0x00000008 200 #define F2FS_FEATURE_PRJQUOTA 0x00000010 201 #define F2FS_FEATURE_INODE_CHKSUM 0x00000020 202 #define F2FS_FEATURE_FLEXIBLE_INLINE_XATTR 0x00000040 203 #define F2FS_FEATURE_QUOTA_INO 0x00000080 204 #define F2FS_FEATURE_INODE_CRTIME 0x00000100 205 #define F2FS_FEATURE_LOST_FOUND 0x00000200 206 #define F2FS_FEATURE_VERITY 0x00000400 207 #define F2FS_FEATURE_SB_CHKSUM 0x00000800 208 #define F2FS_FEATURE_CASEFOLD 0x00001000 209 #define F2FS_FEATURE_COMPRESSION 0x00002000 210 #define F2FS_FEATURE_RO 0x00004000 211 212 #define __F2FS_HAS_FEATURE(raw_super, mask) \ 213 ((raw_super->feature & cpu_to_le32(mask)) != 0) 214 #define F2FS_HAS_FEATURE(sbi, mask) __F2FS_HAS_FEATURE(sbi->raw_super, mask) 215 216 /* 217 * Default values for user and/or group using reserved blocks 218 */ 219 #define F2FS_DEF_RESUID 0 220 #define F2FS_DEF_RESGID 0 221 222 /* 223 * For checkpoint manager 224 */ 225 enum { 226 NAT_BITMAP, 227 SIT_BITMAP 228 }; 229 230 #define CP_UMOUNT 0x00000001 231 #define CP_FASTBOOT 0x00000002 232 #define CP_SYNC 0x00000004 233 #define CP_RECOVERY 0x00000008 234 #define CP_DISCARD 0x00000010 235 #define CP_TRIMMED 0x00000020 236 #define CP_PAUSE 0x00000040 237 #define CP_RESIZE 0x00000080 238 239 #define DEF_MAX_DISCARD_REQUEST 8 /* issue 8 discards per round */ 240 #define DEF_MIN_DISCARD_ISSUE_TIME 50 /* 50 ms, if exists */ 241 #define DEF_MID_DISCARD_ISSUE_TIME 500 /* 500 ms, if device busy */ 242 #define DEF_MAX_DISCARD_ISSUE_TIME 60000 /* 60 s, if no candidates */ 243 #define DEF_DISCARD_URGENT_UTIL 80 /* do more discard over 80% */ 244 #define DEF_CP_INTERVAL 60 /* 60 secs */ 245 #define DEF_IDLE_INTERVAL 5 /* 5 secs */ 246 #define DEF_DISABLE_INTERVAL 5 /* 5 secs */ 247 #define DEF_DISABLE_QUICK_INTERVAL 1 /* 1 secs */ 248 #define DEF_UMOUNT_DISCARD_TIMEOUT 5 /* 5 secs */ 249 250 struct cp_control { 251 int reason; 252 __u64 trim_start; 253 __u64 trim_end; 254 __u64 trim_minlen; 255 }; 256 257 /* 258 * indicate meta/data type 259 */ 260 enum { 261 META_CP, 262 META_NAT, 263 META_SIT, 264 META_SSA, 265 META_MAX, 266 META_POR, 267 DATA_GENERIC, /* check range only */ 268 DATA_GENERIC_ENHANCE, /* strong check on range and segment bitmap */ 269 DATA_GENERIC_ENHANCE_READ, /* 270 * strong check on range and segment 271 * bitmap but no warning due to race 272 * condition of read on truncated area 273 * by extent_cache 274 */ 275 DATA_GENERIC_ENHANCE_UPDATE, /* 276 * strong check on range and segment 277 * bitmap for update case 278 */ 279 META_GENERIC, 280 }; 281 282 /* for the list of ino */ 283 enum { 284 ORPHAN_INO, /* for orphan ino list */ 285 APPEND_INO, /* for append ino list */ 286 UPDATE_INO, /* for update ino list */ 287 TRANS_DIR_INO, /* for transactions dir ino list */ 288 FLUSH_INO, /* for multiple device flushing */ 289 MAX_INO_ENTRY, /* max. list */ 290 }; 291 292 struct ino_entry { 293 struct list_head list; /* list head */ 294 nid_t ino; /* inode number */ 295 unsigned int dirty_device; /* dirty device bitmap */ 296 }; 297 298 /* for the list of inodes to be GCed */ 299 struct inode_entry { 300 struct list_head list; /* list head */ 301 struct inode *inode; /* vfs inode pointer */ 302 }; 303 304 struct fsync_node_entry { 305 struct list_head list; /* list head */ 306 struct page *page; /* warm node page pointer */ 307 unsigned int seq_id; /* sequence id */ 308 }; 309 310 struct ckpt_req { 311 struct completion wait; /* completion for checkpoint done */ 312 struct llist_node llnode; /* llist_node to be linked in wait queue */ 313 int ret; /* return code of checkpoint */ 314 ktime_t queue_time; /* request queued time */ 315 }; 316 317 struct ckpt_req_control { 318 struct task_struct *f2fs_issue_ckpt; /* checkpoint task */ 319 int ckpt_thread_ioprio; /* checkpoint merge thread ioprio */ 320 wait_queue_head_t ckpt_wait_queue; /* waiting queue for wake-up */ 321 atomic_t issued_ckpt; /* # of actually issued ckpts */ 322 atomic_t total_ckpt; /* # of total ckpts */ 323 atomic_t queued_ckpt; /* # of queued ckpts */ 324 struct llist_head issue_list; /* list for command issue */ 325 spinlock_t stat_lock; /* lock for below checkpoint time stats */ 326 unsigned int cur_time; /* cur wait time in msec for currently issued checkpoint */ 327 unsigned int peak_time; /* peak wait time in msec until now */ 328 }; 329 330 /* for the bitmap indicate blocks to be discarded */ 331 struct discard_entry { 332 struct list_head list; /* list head */ 333 block_t start_blkaddr; /* start blockaddr of current segment */ 334 unsigned char discard_map[SIT_VBLOCK_MAP_SIZE]; /* segment discard bitmap */ 335 }; 336 337 /* minimum discard granularity, unit: block count */ 338 #define MIN_DISCARD_GRANULARITY 1 339 /* default discard granularity of inner discard thread, unit: block count */ 340 #define DEFAULT_DISCARD_GRANULARITY 16 341 /* default maximum discard granularity of ordered discard, unit: block count */ 342 #define DEFAULT_MAX_ORDERED_DISCARD_GRANULARITY 16 343 344 /* max discard pend list number */ 345 #define MAX_PLIST_NUM 512 346 #define plist_idx(blk_num) ((blk_num) >= MAX_PLIST_NUM ? \ 347 (MAX_PLIST_NUM - 1) : ((blk_num) - 1)) 348 349 enum { 350 D_PREP, /* initial */ 351 D_PARTIAL, /* partially submitted */ 352 D_SUBMIT, /* all submitted */ 353 D_DONE, /* finished */ 354 }; 355 356 struct discard_info { 357 block_t lstart; /* logical start address */ 358 block_t len; /* length */ 359 block_t start; /* actual start address in dev */ 360 }; 361 362 struct discard_cmd { 363 struct rb_node rb_node; /* rb node located in rb-tree */ 364 struct discard_info di; /* discard info */ 365 struct list_head list; /* command list */ 366 struct completion wait; /* compleation */ 367 struct block_device *bdev; /* bdev */ 368 unsigned short ref; /* reference count */ 369 unsigned char state; /* state */ 370 unsigned char queued; /* queued discard */ 371 int error; /* bio error */ 372 spinlock_t lock; /* for state/bio_ref updating */ 373 unsigned short bio_ref; /* bio reference count */ 374 }; 375 376 enum { 377 DPOLICY_BG, 378 DPOLICY_FORCE, 379 DPOLICY_FSTRIM, 380 DPOLICY_UMOUNT, 381 MAX_DPOLICY, 382 }; 383 384 enum { 385 DPOLICY_IO_AWARE_DISABLE, /* force to not be aware of IO */ 386 DPOLICY_IO_AWARE_ENABLE, /* force to be aware of IO */ 387 DPOLICY_IO_AWARE_MAX, 388 }; 389 390 struct discard_policy { 391 int type; /* type of discard */ 392 unsigned int min_interval; /* used for candidates exist */ 393 unsigned int mid_interval; /* used for device busy */ 394 unsigned int max_interval; /* used for candidates not exist */ 395 unsigned int max_requests; /* # of discards issued per round */ 396 unsigned int io_aware_gran; /* minimum granularity discard not be aware of I/O */ 397 bool io_aware; /* issue discard in idle time */ 398 bool sync; /* submit discard with REQ_SYNC flag */ 399 bool ordered; /* issue discard by lba order */ 400 bool timeout; /* discard timeout for put_super */ 401 unsigned int granularity; /* discard granularity */ 402 }; 403 404 struct discard_cmd_control { 405 struct task_struct *f2fs_issue_discard; /* discard thread */ 406 struct list_head entry_list; /* 4KB discard entry list */ 407 struct list_head pend_list[MAX_PLIST_NUM];/* store pending entries */ 408 struct list_head wait_list; /* store on-flushing entries */ 409 struct list_head fstrim_list; /* in-flight discard from fstrim */ 410 wait_queue_head_t discard_wait_queue; /* waiting queue for wake-up */ 411 struct mutex cmd_lock; 412 unsigned int nr_discards; /* # of discards in the list */ 413 unsigned int max_discards; /* max. discards to be issued */ 414 unsigned int max_discard_request; /* max. discard request per round */ 415 unsigned int min_discard_issue_time; /* min. interval between discard issue */ 416 unsigned int mid_discard_issue_time; /* mid. interval between discard issue */ 417 unsigned int max_discard_issue_time; /* max. interval between discard issue */ 418 unsigned int discard_io_aware_gran; /* minimum discard granularity not be aware of I/O */ 419 unsigned int discard_urgent_util; /* utilization which issue discard proactively */ 420 unsigned int discard_granularity; /* discard granularity */ 421 unsigned int max_ordered_discard; /* maximum discard granularity issued by lba order */ 422 unsigned int discard_io_aware; /* io_aware policy */ 423 unsigned int undiscard_blks; /* # of undiscard blocks */ 424 unsigned int next_pos; /* next discard position */ 425 atomic_t issued_discard; /* # of issued discard */ 426 atomic_t queued_discard; /* # of queued discard */ 427 atomic_t discard_cmd_cnt; /* # of cached cmd count */ 428 struct rb_root_cached root; /* root of discard rb-tree */ 429 bool rbtree_check; /* config for consistence check */ 430 bool discard_wake; /* to wake up discard thread */ 431 }; 432 433 /* for the list of fsync inodes, used only during recovery */ 434 struct fsync_inode_entry { 435 struct list_head list; /* list head */ 436 struct inode *inode; /* vfs inode pointer */ 437 block_t blkaddr; /* block address locating the last fsync */ 438 block_t last_dentry; /* block address locating the last dentry */ 439 }; 440 441 #define nats_in_cursum(jnl) (le16_to_cpu((jnl)->n_nats)) 442 #define sits_in_cursum(jnl) (le16_to_cpu((jnl)->n_sits)) 443 444 #define nat_in_journal(jnl, i) ((jnl)->nat_j.entries[i].ne) 445 #define nid_in_journal(jnl, i) ((jnl)->nat_j.entries[i].nid) 446 #define sit_in_journal(jnl, i) ((jnl)->sit_j.entries[i].se) 447 #define segno_in_journal(jnl, i) ((jnl)->sit_j.entries[i].segno) 448 449 #define MAX_NAT_JENTRIES(jnl) (NAT_JOURNAL_ENTRIES - nats_in_cursum(jnl)) 450 #define MAX_SIT_JENTRIES(jnl) (SIT_JOURNAL_ENTRIES - sits_in_cursum(jnl)) 451 452 static inline int update_nats_in_cursum(struct f2fs_journal *journal, int i) 453 { 454 int before = nats_in_cursum(journal); 455 456 journal->n_nats = cpu_to_le16(before + i); 457 return before; 458 } 459 460 static inline int update_sits_in_cursum(struct f2fs_journal *journal, int i) 461 { 462 int before = sits_in_cursum(journal); 463 464 journal->n_sits = cpu_to_le16(before + i); 465 return before; 466 } 467 468 static inline bool __has_cursum_space(struct f2fs_journal *journal, 469 int size, int type) 470 { 471 if (type == NAT_JOURNAL) 472 return size <= MAX_NAT_JENTRIES(journal); 473 return size <= MAX_SIT_JENTRIES(journal); 474 } 475 476 /* for inline stuff */ 477 #define DEF_INLINE_RESERVED_SIZE 1 478 static inline int get_extra_isize(struct inode *inode); 479 static inline int get_inline_xattr_addrs(struct inode *inode); 480 #define MAX_INLINE_DATA(inode) (sizeof(__le32) * \ 481 (CUR_ADDRS_PER_INODE(inode) - \ 482 get_inline_xattr_addrs(inode) - \ 483 DEF_INLINE_RESERVED_SIZE)) 484 485 /* for inline dir */ 486 #define NR_INLINE_DENTRY(inode) (MAX_INLINE_DATA(inode) * BITS_PER_BYTE / \ 487 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \ 488 BITS_PER_BYTE + 1)) 489 #define INLINE_DENTRY_BITMAP_SIZE(inode) \ 490 DIV_ROUND_UP(NR_INLINE_DENTRY(inode), BITS_PER_BYTE) 491 #define INLINE_RESERVED_SIZE(inode) (MAX_INLINE_DATA(inode) - \ 492 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \ 493 NR_INLINE_DENTRY(inode) + \ 494 INLINE_DENTRY_BITMAP_SIZE(inode))) 495 496 /* 497 * For INODE and NODE manager 498 */ 499 /* for directory operations */ 500 501 struct f2fs_filename { 502 /* 503 * The filename the user specified. This is NULL for some 504 * filesystem-internal operations, e.g. converting an inline directory 505 * to a non-inline one, or roll-forward recovering an encrypted dentry. 506 */ 507 const struct qstr *usr_fname; 508 509 /* 510 * The on-disk filename. For encrypted directories, this is encrypted. 511 * This may be NULL for lookups in an encrypted dir without the key. 512 */ 513 struct fscrypt_str disk_name; 514 515 /* The dirhash of this filename */ 516 f2fs_hash_t hash; 517 518 #ifdef CONFIG_FS_ENCRYPTION 519 /* 520 * For lookups in encrypted directories: either the buffer backing 521 * disk_name, or a buffer that holds the decoded no-key name. 522 */ 523 struct fscrypt_str crypto_buf; 524 #endif 525 #if IS_ENABLED(CONFIG_UNICODE) 526 /* 527 * For casefolded directories: the casefolded name, but it's left NULL 528 * if the original name is not valid Unicode, if the original name is 529 * "." or "..", if the directory is both casefolded and encrypted and 530 * its encryption key is unavailable, or if the filesystem is doing an 531 * internal operation where usr_fname is also NULL. In all these cases 532 * we fall back to treating the name as an opaque byte sequence. 533 */ 534 struct fscrypt_str cf_name; 535 #endif 536 }; 537 538 struct f2fs_dentry_ptr { 539 struct inode *inode; 540 void *bitmap; 541 struct f2fs_dir_entry *dentry; 542 __u8 (*filename)[F2FS_SLOT_LEN]; 543 int max; 544 int nr_bitmap; 545 }; 546 547 static inline void make_dentry_ptr_block(struct inode *inode, 548 struct f2fs_dentry_ptr *d, struct f2fs_dentry_block *t) 549 { 550 d->inode = inode; 551 d->max = NR_DENTRY_IN_BLOCK; 552 d->nr_bitmap = SIZE_OF_DENTRY_BITMAP; 553 d->bitmap = t->dentry_bitmap; 554 d->dentry = t->dentry; 555 d->filename = t->filename; 556 } 557 558 static inline void make_dentry_ptr_inline(struct inode *inode, 559 struct f2fs_dentry_ptr *d, void *t) 560 { 561 int entry_cnt = NR_INLINE_DENTRY(inode); 562 int bitmap_size = INLINE_DENTRY_BITMAP_SIZE(inode); 563 int reserved_size = INLINE_RESERVED_SIZE(inode); 564 565 d->inode = inode; 566 d->max = entry_cnt; 567 d->nr_bitmap = bitmap_size; 568 d->bitmap = t; 569 d->dentry = t + bitmap_size + reserved_size; 570 d->filename = t + bitmap_size + reserved_size + 571 SIZE_OF_DIR_ENTRY * entry_cnt; 572 } 573 574 /* 575 * XATTR_NODE_OFFSET stores xattrs to one node block per file keeping -1 576 * as its node offset to distinguish from index node blocks. 577 * But some bits are used to mark the node block. 578 */ 579 #define XATTR_NODE_OFFSET ((((unsigned int)-1) << OFFSET_BIT_SHIFT) \ 580 >> OFFSET_BIT_SHIFT) 581 enum { 582 ALLOC_NODE, /* allocate a new node page if needed */ 583 LOOKUP_NODE, /* look up a node without readahead */ 584 LOOKUP_NODE_RA, /* 585 * look up a node with readahead called 586 * by get_data_block. 587 */ 588 }; 589 590 #define DEFAULT_RETRY_IO_COUNT 8 /* maximum retry read IO or flush count */ 591 592 /* congestion wait timeout value, default: 20ms */ 593 #define DEFAULT_IO_TIMEOUT (msecs_to_jiffies(20)) 594 595 /* maximum retry quota flush count */ 596 #define DEFAULT_RETRY_QUOTA_FLUSH_COUNT 8 597 598 /* maximum retry of EIO'ed page */ 599 #define MAX_RETRY_PAGE_EIO 100 600 601 #define F2FS_LINK_MAX 0xffffffff /* maximum link count per file */ 602 603 #define MAX_DIR_RA_PAGES 4 /* maximum ra pages of dir */ 604 605 /* dirty segments threshold for triggering CP */ 606 #define DEFAULT_DIRTY_THRESHOLD 4 607 608 #define RECOVERY_MAX_RA_BLOCKS BIO_MAX_VECS 609 #define RECOVERY_MIN_RA_BLOCKS 1 610 611 #define F2FS_ONSTACK_PAGES 16 /* nr of onstack pages */ 612 613 /* for in-memory extent cache entry */ 614 #define F2FS_MIN_EXTENT_LEN 64 /* minimum extent length */ 615 616 /* number of extent info in extent cache we try to shrink */ 617 #define READ_EXTENT_CACHE_SHRINK_NUMBER 128 618 619 /* number of age extent info in extent cache we try to shrink */ 620 #define AGE_EXTENT_CACHE_SHRINK_NUMBER 128 621 #define LAST_AGE_WEIGHT 30 622 #define SAME_AGE_REGION 1024 623 624 /* 625 * Define data block with age less than 1GB as hot data 626 * define data block with age less than 10GB but more than 1GB as warm data 627 */ 628 #define DEF_HOT_DATA_AGE_THRESHOLD 262144 629 #define DEF_WARM_DATA_AGE_THRESHOLD 2621440 630 631 /* extent cache type */ 632 enum extent_type { 633 EX_READ, 634 EX_BLOCK_AGE, 635 NR_EXTENT_CACHES, 636 }; 637 638 struct extent_info { 639 unsigned int fofs; /* start offset in a file */ 640 unsigned int len; /* length of the extent */ 641 union { 642 /* read extent_cache */ 643 struct { 644 /* start block address of the extent */ 645 block_t blk; 646 #ifdef CONFIG_F2FS_FS_COMPRESSION 647 /* physical extent length of compressed blocks */ 648 unsigned int c_len; 649 #endif 650 }; 651 /* block age extent_cache */ 652 struct { 653 /* block age of the extent */ 654 unsigned long long age; 655 /* last total blocks allocated */ 656 unsigned long long last_blocks; 657 }; 658 }; 659 }; 660 661 struct extent_node { 662 struct rb_node rb_node; /* rb node located in rb-tree */ 663 struct extent_info ei; /* extent info */ 664 struct list_head list; /* node in global extent list of sbi */ 665 struct extent_tree *et; /* extent tree pointer */ 666 }; 667 668 struct extent_tree { 669 nid_t ino; /* inode number */ 670 enum extent_type type; /* keep the extent tree type */ 671 struct rb_root_cached root; /* root of extent info rb-tree */ 672 struct extent_node *cached_en; /* recently accessed extent node */ 673 struct list_head list; /* to be used by sbi->zombie_list */ 674 rwlock_t lock; /* protect extent info rb-tree */ 675 atomic_t node_cnt; /* # of extent node in rb-tree*/ 676 bool largest_updated; /* largest extent updated */ 677 struct extent_info largest; /* largest cached extent for EX_READ */ 678 }; 679 680 struct extent_tree_info { 681 struct radix_tree_root extent_tree_root;/* cache extent cache entries */ 682 struct mutex extent_tree_lock; /* locking extent radix tree */ 683 struct list_head extent_list; /* lru list for shrinker */ 684 spinlock_t extent_lock; /* locking extent lru list */ 685 atomic_t total_ext_tree; /* extent tree count */ 686 struct list_head zombie_list; /* extent zombie tree list */ 687 atomic_t total_zombie_tree; /* extent zombie tree count */ 688 atomic_t total_ext_node; /* extent info count */ 689 }; 690 691 /* 692 * State of block returned by f2fs_map_blocks. 693 */ 694 #define F2FS_MAP_NEW (1U << 0) 695 #define F2FS_MAP_MAPPED (1U << 1) 696 #define F2FS_MAP_DELALLOC (1U << 2) 697 #define F2FS_MAP_FLAGS (F2FS_MAP_NEW | F2FS_MAP_MAPPED |\ 698 F2FS_MAP_DELALLOC) 699 700 struct f2fs_map_blocks { 701 struct block_device *m_bdev; /* for multi-device dio */ 702 block_t m_pblk; 703 block_t m_lblk; 704 unsigned int m_len; 705 unsigned int m_flags; 706 pgoff_t *m_next_pgofs; /* point next possible non-hole pgofs */ 707 pgoff_t *m_next_extent; /* point to next possible extent */ 708 int m_seg_type; 709 bool m_may_create; /* indicate it is from write path */ 710 bool m_multidev_dio; /* indicate it allows multi-device dio */ 711 }; 712 713 /* for flag in get_data_block */ 714 enum { 715 F2FS_GET_BLOCK_DEFAULT, 716 F2FS_GET_BLOCK_FIEMAP, 717 F2FS_GET_BLOCK_BMAP, 718 F2FS_GET_BLOCK_DIO, 719 F2FS_GET_BLOCK_PRE_DIO, 720 F2FS_GET_BLOCK_PRE_AIO, 721 F2FS_GET_BLOCK_PRECACHE, 722 }; 723 724 /* 725 * i_advise uses FADVISE_XXX_BIT. We can add additional hints later. 726 */ 727 #define FADVISE_COLD_BIT 0x01 728 #define FADVISE_LOST_PINO_BIT 0x02 729 #define FADVISE_ENCRYPT_BIT 0x04 730 #define FADVISE_ENC_NAME_BIT 0x08 731 #define FADVISE_KEEP_SIZE_BIT 0x10 732 #define FADVISE_HOT_BIT 0x20 733 #define FADVISE_VERITY_BIT 0x40 734 #define FADVISE_TRUNC_BIT 0x80 735 736 #define FADVISE_MODIFIABLE_BITS (FADVISE_COLD_BIT | FADVISE_HOT_BIT) 737 738 #define file_is_cold(inode) is_file(inode, FADVISE_COLD_BIT) 739 #define file_set_cold(inode) set_file(inode, FADVISE_COLD_BIT) 740 #define file_clear_cold(inode) clear_file(inode, FADVISE_COLD_BIT) 741 742 #define file_wrong_pino(inode) is_file(inode, FADVISE_LOST_PINO_BIT) 743 #define file_lost_pino(inode) set_file(inode, FADVISE_LOST_PINO_BIT) 744 #define file_got_pino(inode) clear_file(inode, FADVISE_LOST_PINO_BIT) 745 746 #define file_is_encrypt(inode) is_file(inode, FADVISE_ENCRYPT_BIT) 747 #define file_set_encrypt(inode) set_file(inode, FADVISE_ENCRYPT_BIT) 748 749 #define file_enc_name(inode) is_file(inode, FADVISE_ENC_NAME_BIT) 750 #define file_set_enc_name(inode) set_file(inode, FADVISE_ENC_NAME_BIT) 751 752 #define file_keep_isize(inode) is_file(inode, FADVISE_KEEP_SIZE_BIT) 753 #define file_set_keep_isize(inode) set_file(inode, FADVISE_KEEP_SIZE_BIT) 754 755 #define file_is_hot(inode) is_file(inode, FADVISE_HOT_BIT) 756 #define file_set_hot(inode) set_file(inode, FADVISE_HOT_BIT) 757 #define file_clear_hot(inode) clear_file(inode, FADVISE_HOT_BIT) 758 759 #define file_is_verity(inode) is_file(inode, FADVISE_VERITY_BIT) 760 #define file_set_verity(inode) set_file(inode, FADVISE_VERITY_BIT) 761 762 #define file_should_truncate(inode) is_file(inode, FADVISE_TRUNC_BIT) 763 #define file_need_truncate(inode) set_file(inode, FADVISE_TRUNC_BIT) 764 #define file_dont_truncate(inode) clear_file(inode, FADVISE_TRUNC_BIT) 765 766 #define DEF_DIR_LEVEL 0 767 768 enum { 769 GC_FAILURE_PIN, 770 MAX_GC_FAILURE 771 }; 772 773 /* used for f2fs_inode_info->flags */ 774 enum { 775 FI_NEW_INODE, /* indicate newly allocated inode */ 776 FI_DIRTY_INODE, /* indicate inode is dirty or not */ 777 FI_AUTO_RECOVER, /* indicate inode is recoverable */ 778 FI_DIRTY_DIR, /* indicate directory has dirty pages */ 779 FI_INC_LINK, /* need to increment i_nlink */ 780 FI_ACL_MODE, /* indicate acl mode */ 781 FI_NO_ALLOC, /* should not allocate any blocks */ 782 FI_FREE_NID, /* free allocated nide */ 783 FI_NO_EXTENT, /* not to use the extent cache */ 784 FI_INLINE_XATTR, /* used for inline xattr */ 785 FI_INLINE_DATA, /* used for inline data*/ 786 FI_INLINE_DENTRY, /* used for inline dentry */ 787 FI_APPEND_WRITE, /* inode has appended data */ 788 FI_UPDATE_WRITE, /* inode has in-place-update data */ 789 FI_NEED_IPU, /* used for ipu per file */ 790 FI_ATOMIC_FILE, /* indicate atomic file */ 791 FI_DATA_EXIST, /* indicate data exists */ 792 FI_INLINE_DOTS, /* indicate inline dot dentries */ 793 FI_SKIP_WRITES, /* should skip data page writeback */ 794 FI_OPU_WRITE, /* used for opu per file */ 795 FI_DIRTY_FILE, /* indicate regular/symlink has dirty pages */ 796 FI_PREALLOCATED_ALL, /* all blocks for write were preallocated */ 797 FI_HOT_DATA, /* indicate file is hot */ 798 FI_EXTRA_ATTR, /* indicate file has extra attribute */ 799 FI_PROJ_INHERIT, /* indicate file inherits projectid */ 800 FI_PIN_FILE, /* indicate file should not be gced */ 801 FI_VERITY_IN_PROGRESS, /* building fs-verity Merkle tree */ 802 FI_COMPRESSED_FILE, /* indicate file's data can be compressed */ 803 FI_COMPRESS_CORRUPT, /* indicate compressed cluster is corrupted */ 804 FI_MMAP_FILE, /* indicate file was mmapped */ 805 FI_ENABLE_COMPRESS, /* enable compression in "user" compression mode */ 806 FI_COMPRESS_RELEASED, /* compressed blocks were released */ 807 FI_ALIGNED_WRITE, /* enable aligned write */ 808 FI_COW_FILE, /* indicate COW file */ 809 FI_ATOMIC_COMMITTED, /* indicate atomic commit completed except disk sync */ 810 FI_ATOMIC_REPLACE, /* indicate atomic replace */ 811 FI_MAX, /* max flag, never be used */ 812 }; 813 814 struct f2fs_inode_info { 815 struct inode vfs_inode; /* serve a vfs inode */ 816 unsigned long i_flags; /* keep an inode flags for ioctl */ 817 unsigned char i_advise; /* use to give file attribute hints */ 818 unsigned char i_dir_level; /* use for dentry level for large dir */ 819 unsigned int i_current_depth; /* only for directory depth */ 820 /* for gc failure statistic */ 821 unsigned int i_gc_failures[MAX_GC_FAILURE]; 822 unsigned int i_pino; /* parent inode number */ 823 umode_t i_acl_mode; /* keep file acl mode temporarily */ 824 825 /* Use below internally in f2fs*/ 826 unsigned long flags[BITS_TO_LONGS(FI_MAX)]; /* use to pass per-file flags */ 827 struct f2fs_rwsem i_sem; /* protect fi info */ 828 atomic_t dirty_pages; /* # of dirty pages */ 829 f2fs_hash_t chash; /* hash value of given file name */ 830 unsigned int clevel; /* maximum level of given file name */ 831 struct task_struct *task; /* lookup and create consistency */ 832 struct task_struct *cp_task; /* separate cp/wb IO stats*/ 833 struct task_struct *wb_task; /* indicate inode is in context of writeback */ 834 nid_t i_xattr_nid; /* node id that contains xattrs */ 835 loff_t last_disk_size; /* lastly written file size */ 836 spinlock_t i_size_lock; /* protect last_disk_size */ 837 838 #ifdef CONFIG_QUOTA 839 struct dquot __rcu *i_dquot[MAXQUOTAS]; 840 841 /* quota space reservation, managed internally by quota code */ 842 qsize_t i_reserved_quota; 843 #endif 844 struct list_head dirty_list; /* dirty list for dirs and files */ 845 struct list_head gdirty_list; /* linked in global dirty list */ 846 struct task_struct *atomic_write_task; /* store atomic write task */ 847 struct extent_tree *extent_tree[NR_EXTENT_CACHES]; 848 /* cached extent_tree entry */ 849 struct inode *cow_inode; /* copy-on-write inode for atomic write */ 850 851 /* avoid racing between foreground op and gc */ 852 struct f2fs_rwsem i_gc_rwsem[2]; 853 struct f2fs_rwsem i_xattr_sem; /* avoid racing between reading and changing EAs */ 854 855 int i_extra_isize; /* size of extra space located in i_addr */ 856 kprojid_t i_projid; /* id for project quota */ 857 int i_inline_xattr_size; /* inline xattr size */ 858 struct timespec64 i_crtime; /* inode creation time */ 859 struct timespec64 i_disk_time[3];/* inode disk times */ 860 861 /* for file compress */ 862 atomic_t i_compr_blocks; /* # of compressed blocks */ 863 unsigned char i_compress_algorithm; /* algorithm type */ 864 unsigned char i_log_cluster_size; /* log of cluster size */ 865 unsigned char i_compress_level; /* compress level (lz4hc,zstd) */ 866 unsigned char i_compress_flag; /* compress flag */ 867 unsigned int i_cluster_size; /* cluster size */ 868 869 unsigned int atomic_write_cnt; 870 loff_t original_i_size; /* original i_size before atomic write */ 871 }; 872 873 static inline void get_read_extent_info(struct extent_info *ext, 874 struct f2fs_extent *i_ext) 875 { 876 ext->fofs = le32_to_cpu(i_ext->fofs); 877 ext->blk = le32_to_cpu(i_ext->blk); 878 ext->len = le32_to_cpu(i_ext->len); 879 } 880 881 static inline void set_raw_read_extent(struct extent_info *ext, 882 struct f2fs_extent *i_ext) 883 { 884 i_ext->fofs = cpu_to_le32(ext->fofs); 885 i_ext->blk = cpu_to_le32(ext->blk); 886 i_ext->len = cpu_to_le32(ext->len); 887 } 888 889 static inline bool __is_discard_mergeable(struct discard_info *back, 890 struct discard_info *front, unsigned int max_len) 891 { 892 return (back->lstart + back->len == front->lstart) && 893 (back->len + front->len <= max_len); 894 } 895 896 static inline bool __is_discard_back_mergeable(struct discard_info *cur, 897 struct discard_info *back, unsigned int max_len) 898 { 899 return __is_discard_mergeable(back, cur, max_len); 900 } 901 902 static inline bool __is_discard_front_mergeable(struct discard_info *cur, 903 struct discard_info *front, unsigned int max_len) 904 { 905 return __is_discard_mergeable(cur, front, max_len); 906 } 907 908 /* 909 * For free nid management 910 */ 911 enum nid_state { 912 FREE_NID, /* newly added to free nid list */ 913 PREALLOC_NID, /* it is preallocated */ 914 MAX_NID_STATE, 915 }; 916 917 enum nat_state { 918 TOTAL_NAT, 919 DIRTY_NAT, 920 RECLAIMABLE_NAT, 921 MAX_NAT_STATE, 922 }; 923 924 struct f2fs_nm_info { 925 block_t nat_blkaddr; /* base disk address of NAT */ 926 nid_t max_nid; /* maximum possible node ids */ 927 nid_t available_nids; /* # of available node ids */ 928 nid_t next_scan_nid; /* the next nid to be scanned */ 929 nid_t max_rf_node_blocks; /* max # of nodes for recovery */ 930 unsigned int ram_thresh; /* control the memory footprint */ 931 unsigned int ra_nid_pages; /* # of nid pages to be readaheaded */ 932 unsigned int dirty_nats_ratio; /* control dirty nats ratio threshold */ 933 934 /* NAT cache management */ 935 struct radix_tree_root nat_root;/* root of the nat entry cache */ 936 struct radix_tree_root nat_set_root;/* root of the nat set cache */ 937 struct f2fs_rwsem nat_tree_lock; /* protect nat entry tree */ 938 struct list_head nat_entries; /* cached nat entry list (clean) */ 939 spinlock_t nat_list_lock; /* protect clean nat entry list */ 940 unsigned int nat_cnt[MAX_NAT_STATE]; /* the # of cached nat entries */ 941 unsigned int nat_blocks; /* # of nat blocks */ 942 943 /* free node ids management */ 944 struct radix_tree_root free_nid_root;/* root of the free_nid cache */ 945 struct list_head free_nid_list; /* list for free nids excluding preallocated nids */ 946 unsigned int nid_cnt[MAX_NID_STATE]; /* the number of free node id */ 947 spinlock_t nid_list_lock; /* protect nid lists ops */ 948 struct mutex build_lock; /* lock for build free nids */ 949 unsigned char **free_nid_bitmap; 950 unsigned char *nat_block_bitmap; 951 unsigned short *free_nid_count; /* free nid count of NAT block */ 952 953 /* for checkpoint */ 954 char *nat_bitmap; /* NAT bitmap pointer */ 955 956 unsigned int nat_bits_blocks; /* # of nat bits blocks */ 957 unsigned char *nat_bits; /* NAT bits blocks */ 958 unsigned char *full_nat_bits; /* full NAT pages */ 959 unsigned char *empty_nat_bits; /* empty NAT pages */ 960 #ifdef CONFIG_F2FS_CHECK_FS 961 char *nat_bitmap_mir; /* NAT bitmap mirror */ 962 #endif 963 int bitmap_size; /* bitmap size */ 964 }; 965 966 /* 967 * this structure is used as one of function parameters. 968 * all the information are dedicated to a given direct node block determined 969 * by the data offset in a file. 970 */ 971 struct dnode_of_data { 972 struct inode *inode; /* vfs inode pointer */ 973 struct page *inode_page; /* its inode page, NULL is possible */ 974 struct page *node_page; /* cached direct node page */ 975 nid_t nid; /* node id of the direct node block */ 976 unsigned int ofs_in_node; /* data offset in the node page */ 977 bool inode_page_locked; /* inode page is locked or not */ 978 bool node_changed; /* is node block changed */ 979 char cur_level; /* level of hole node page */ 980 char max_level; /* level of current page located */ 981 block_t data_blkaddr; /* block address of the node block */ 982 }; 983 984 static inline void set_new_dnode(struct dnode_of_data *dn, struct inode *inode, 985 struct page *ipage, struct page *npage, nid_t nid) 986 { 987 memset(dn, 0, sizeof(*dn)); 988 dn->inode = inode; 989 dn->inode_page = ipage; 990 dn->node_page = npage; 991 dn->nid = nid; 992 } 993 994 /* 995 * For SIT manager 996 * 997 * By default, there are 6 active log areas across the whole main area. 998 * When considering hot and cold data separation to reduce cleaning overhead, 999 * we split 3 for data logs and 3 for node logs as hot, warm, and cold types, 1000 * respectively. 1001 * In the current design, you should not change the numbers intentionally. 1002 * Instead, as a mount option such as active_logs=x, you can use 2, 4, and 6 1003 * logs individually according to the underlying devices. (default: 6) 1004 * Just in case, on-disk layout covers maximum 16 logs that consist of 8 for 1005 * data and 8 for node logs. 1006 */ 1007 #define NR_CURSEG_DATA_TYPE (3) 1008 #define NR_CURSEG_NODE_TYPE (3) 1009 #define NR_CURSEG_INMEM_TYPE (2) 1010 #define NR_CURSEG_RO_TYPE (2) 1011 #define NR_CURSEG_PERSIST_TYPE (NR_CURSEG_DATA_TYPE + NR_CURSEG_NODE_TYPE) 1012 #define NR_CURSEG_TYPE (NR_CURSEG_INMEM_TYPE + NR_CURSEG_PERSIST_TYPE) 1013 1014 enum { 1015 CURSEG_HOT_DATA = 0, /* directory entry blocks */ 1016 CURSEG_WARM_DATA, /* data blocks */ 1017 CURSEG_COLD_DATA, /* multimedia or GCed data blocks */ 1018 CURSEG_HOT_NODE, /* direct node blocks of directory files */ 1019 CURSEG_WARM_NODE, /* direct node blocks of normal files */ 1020 CURSEG_COLD_NODE, /* indirect node blocks */ 1021 NR_PERSISTENT_LOG, /* number of persistent log */ 1022 CURSEG_COLD_DATA_PINNED = NR_PERSISTENT_LOG, 1023 /* pinned file that needs consecutive block address */ 1024 CURSEG_ALL_DATA_ATGC, /* SSR alloctor in hot/warm/cold data area */ 1025 NO_CHECK_TYPE, /* number of persistent & inmem log */ 1026 }; 1027 1028 struct flush_cmd { 1029 struct completion wait; 1030 struct llist_node llnode; 1031 nid_t ino; 1032 int ret; 1033 }; 1034 1035 struct flush_cmd_control { 1036 struct task_struct *f2fs_issue_flush; /* flush thread */ 1037 wait_queue_head_t flush_wait_queue; /* waiting queue for wake-up */ 1038 atomic_t issued_flush; /* # of issued flushes */ 1039 atomic_t queued_flush; /* # of queued flushes */ 1040 struct llist_head issue_list; /* list for command issue */ 1041 struct llist_node *dispatch_list; /* list for command dispatch */ 1042 }; 1043 1044 struct f2fs_sm_info { 1045 struct sit_info *sit_info; /* whole segment information */ 1046 struct free_segmap_info *free_info; /* free segment information */ 1047 struct dirty_seglist_info *dirty_info; /* dirty segment information */ 1048 struct curseg_info *curseg_array; /* active segment information */ 1049 1050 struct f2fs_rwsem curseg_lock; /* for preventing curseg change */ 1051 1052 block_t seg0_blkaddr; /* block address of 0'th segment */ 1053 block_t main_blkaddr; /* start block address of main area */ 1054 block_t ssa_blkaddr; /* start block address of SSA area */ 1055 1056 unsigned int segment_count; /* total # of segments */ 1057 unsigned int main_segments; /* # of segments in main area */ 1058 unsigned int reserved_segments; /* # of reserved segments */ 1059 unsigned int additional_reserved_segments;/* reserved segs for IO align feature */ 1060 unsigned int ovp_segments; /* # of overprovision segments */ 1061 1062 /* a threshold to reclaim prefree segments */ 1063 unsigned int rec_prefree_segments; 1064 1065 struct list_head sit_entry_set; /* sit entry set list */ 1066 1067 unsigned int ipu_policy; /* in-place-update policy */ 1068 unsigned int min_ipu_util; /* in-place-update threshold */ 1069 unsigned int min_fsync_blocks; /* threshold for fsync */ 1070 unsigned int min_seq_blocks; /* threshold for sequential blocks */ 1071 unsigned int min_hot_blocks; /* threshold for hot block allocation */ 1072 unsigned int min_ssr_sections; /* threshold to trigger SSR allocation */ 1073 1074 /* for flush command control */ 1075 struct flush_cmd_control *fcc_info; 1076 1077 /* for discard command control */ 1078 struct discard_cmd_control *dcc_info; 1079 }; 1080 1081 /* 1082 * For superblock 1083 */ 1084 /* 1085 * COUNT_TYPE for monitoring 1086 * 1087 * f2fs monitors the number of several block types such as on-writeback, 1088 * dirty dentry blocks, dirty node blocks, and dirty meta blocks. 1089 */ 1090 #define WB_DATA_TYPE(p, f) \ 1091 (f || f2fs_is_cp_guaranteed(p) ? F2FS_WB_CP_DATA : F2FS_WB_DATA) 1092 enum count_type { 1093 F2FS_DIRTY_DENTS, 1094 F2FS_DIRTY_DATA, 1095 F2FS_DIRTY_QDATA, 1096 F2FS_DIRTY_NODES, 1097 F2FS_DIRTY_META, 1098 F2FS_DIRTY_IMETA, 1099 F2FS_WB_CP_DATA, 1100 F2FS_WB_DATA, 1101 F2FS_RD_DATA, 1102 F2FS_RD_NODE, 1103 F2FS_RD_META, 1104 F2FS_DIO_WRITE, 1105 F2FS_DIO_READ, 1106 NR_COUNT_TYPE, 1107 }; 1108 1109 /* 1110 * The below are the page types of bios used in submit_bio(). 1111 * The available types are: 1112 * DATA User data pages. It operates as async mode. 1113 * NODE Node pages. It operates as async mode. 1114 * META FS metadata pages such as SIT, NAT, CP. 1115 * NR_PAGE_TYPE The number of page types. 1116 * META_FLUSH Make sure the previous pages are written 1117 * with waiting the bio's completion 1118 * ... Only can be used with META. 1119 */ 1120 #define PAGE_TYPE_OF_BIO(type) ((type) > META ? META : (type)) 1121 #define PAGE_TYPE_ON_MAIN(type) ((type) == DATA || (type) == NODE) 1122 enum page_type { 1123 DATA = 0, 1124 NODE = 1, /* should not change this */ 1125 META, 1126 NR_PAGE_TYPE, 1127 META_FLUSH, 1128 IPU, /* the below types are used by tracepoints only. */ 1129 OPU, 1130 }; 1131 1132 enum temp_type { 1133 HOT = 0, /* must be zero for meta bio */ 1134 WARM, 1135 COLD, 1136 NR_TEMP_TYPE, 1137 }; 1138 1139 enum need_lock_type { 1140 LOCK_REQ = 0, 1141 LOCK_DONE, 1142 LOCK_RETRY, 1143 }; 1144 1145 enum cp_reason_type { 1146 CP_NO_NEEDED, 1147 CP_NON_REGULAR, 1148 CP_COMPRESSED, 1149 CP_HARDLINK, 1150 CP_SB_NEED_CP, 1151 CP_WRONG_PINO, 1152 CP_NO_SPC_ROLL, 1153 CP_NODE_NEED_CP, 1154 CP_FASTBOOT_MODE, 1155 CP_SPEC_LOG_NUM, 1156 CP_RECOVER_DIR, 1157 }; 1158 1159 enum iostat_type { 1160 /* WRITE IO */ 1161 APP_DIRECT_IO, /* app direct write IOs */ 1162 APP_BUFFERED_IO, /* app buffered write IOs */ 1163 APP_WRITE_IO, /* app write IOs */ 1164 APP_MAPPED_IO, /* app mapped IOs */ 1165 APP_BUFFERED_CDATA_IO, /* app buffered write IOs on compressed file */ 1166 APP_MAPPED_CDATA_IO, /* app mapped write IOs on compressed file */ 1167 FS_DATA_IO, /* data IOs from kworker/fsync/reclaimer */ 1168 FS_CDATA_IO, /* data IOs from kworker/fsync/reclaimer on compressed file */ 1169 FS_NODE_IO, /* node IOs from kworker/fsync/reclaimer */ 1170 FS_META_IO, /* meta IOs from kworker/reclaimer */ 1171 FS_GC_DATA_IO, /* data IOs from forground gc */ 1172 FS_GC_NODE_IO, /* node IOs from forground gc */ 1173 FS_CP_DATA_IO, /* data IOs from checkpoint */ 1174 FS_CP_NODE_IO, /* node IOs from checkpoint */ 1175 FS_CP_META_IO, /* meta IOs from checkpoint */ 1176 1177 /* READ IO */ 1178 APP_DIRECT_READ_IO, /* app direct read IOs */ 1179 APP_BUFFERED_READ_IO, /* app buffered read IOs */ 1180 APP_READ_IO, /* app read IOs */ 1181 APP_MAPPED_READ_IO, /* app mapped read IOs */ 1182 APP_BUFFERED_CDATA_READ_IO, /* app buffered read IOs on compressed file */ 1183 APP_MAPPED_CDATA_READ_IO, /* app mapped read IOs on compressed file */ 1184 FS_DATA_READ_IO, /* data read IOs */ 1185 FS_GDATA_READ_IO, /* data read IOs from background gc */ 1186 FS_CDATA_READ_IO, /* compressed data read IOs */ 1187 FS_NODE_READ_IO, /* node read IOs */ 1188 FS_META_READ_IO, /* meta read IOs */ 1189 1190 /* other */ 1191 FS_DISCARD_IO, /* discard */ 1192 FS_FLUSH_IO, /* flush */ 1193 FS_ZONE_RESET_IO, /* zone reset */ 1194 NR_IO_TYPE, 1195 }; 1196 1197 struct f2fs_io_info { 1198 struct f2fs_sb_info *sbi; /* f2fs_sb_info pointer */ 1199 nid_t ino; /* inode number */ 1200 enum page_type type; /* contains DATA/NODE/META/META_FLUSH */ 1201 enum temp_type temp; /* contains HOT/WARM/COLD */ 1202 enum req_op op; /* contains REQ_OP_ */ 1203 blk_opf_t op_flags; /* req_flag_bits */ 1204 block_t new_blkaddr; /* new block address to be written */ 1205 block_t old_blkaddr; /* old block address before Cow */ 1206 struct page *page; /* page to be written */ 1207 struct page *encrypted_page; /* encrypted page */ 1208 struct page *compressed_page; /* compressed page */ 1209 struct list_head list; /* serialize IOs */ 1210 unsigned int compr_blocks; /* # of compressed block addresses */ 1211 unsigned int need_lock:8; /* indicate we need to lock cp_rwsem */ 1212 unsigned int version:8; /* version of the node */ 1213 unsigned int submitted:1; /* indicate IO submission */ 1214 unsigned int in_list:1; /* indicate fio is in io_list */ 1215 unsigned int is_por:1; /* indicate IO is from recovery or not */ 1216 unsigned int encrypted:1; /* indicate file is encrypted */ 1217 unsigned int post_read:1; /* require post read */ 1218 enum iostat_type io_type; /* io type */ 1219 struct writeback_control *io_wbc; /* writeback control */ 1220 struct bio **bio; /* bio for ipu */ 1221 sector_t *last_block; /* last block number in bio */ 1222 }; 1223 1224 struct bio_entry { 1225 struct bio *bio; 1226 struct list_head list; 1227 }; 1228 1229 #define is_read_io(rw) ((rw) == READ) 1230 struct f2fs_bio_info { 1231 struct f2fs_sb_info *sbi; /* f2fs superblock */ 1232 struct bio *bio; /* bios to merge */ 1233 sector_t last_block_in_bio; /* last block number */ 1234 struct f2fs_io_info fio; /* store buffered io info. */ 1235 #ifdef CONFIG_BLK_DEV_ZONED 1236 struct completion zone_wait; /* condition value for the previous open zone to close */ 1237 struct bio *zone_pending_bio; /* pending bio for the previous zone */ 1238 void *bi_private; /* previous bi_private for pending bio */ 1239 #endif 1240 struct f2fs_rwsem io_rwsem; /* blocking op for bio */ 1241 spinlock_t io_lock; /* serialize DATA/NODE IOs */ 1242 struct list_head io_list; /* track fios */ 1243 struct list_head bio_list; /* bio entry list head */ 1244 struct f2fs_rwsem bio_list_lock; /* lock to protect bio entry list */ 1245 }; 1246 1247 #define FDEV(i) (sbi->devs[i]) 1248 #define RDEV(i) (raw_super->devs[i]) 1249 struct f2fs_dev_info { 1250 struct file *bdev_file; 1251 struct block_device *bdev; 1252 char path[MAX_PATH_LEN]; 1253 unsigned int total_segments; 1254 block_t start_blk; 1255 block_t end_blk; 1256 #ifdef CONFIG_BLK_DEV_ZONED 1257 unsigned int nr_blkz; /* Total number of zones */ 1258 unsigned long *blkz_seq; /* Bitmap indicating sequential zones */ 1259 #endif 1260 }; 1261 1262 enum inode_type { 1263 DIR_INODE, /* for dirty dir inode */ 1264 FILE_INODE, /* for dirty regular/symlink inode */ 1265 DIRTY_META, /* for all dirtied inode metadata */ 1266 NR_INODE_TYPE, 1267 }; 1268 1269 /* for inner inode cache management */ 1270 struct inode_management { 1271 struct radix_tree_root ino_root; /* ino entry array */ 1272 spinlock_t ino_lock; /* for ino entry lock */ 1273 struct list_head ino_list; /* inode list head */ 1274 unsigned long ino_num; /* number of entries */ 1275 }; 1276 1277 /* for GC_AT */ 1278 struct atgc_management { 1279 bool atgc_enabled; /* ATGC is enabled or not */ 1280 struct rb_root_cached root; /* root of victim rb-tree */ 1281 struct list_head victim_list; /* linked with all victim entries */ 1282 unsigned int victim_count; /* victim count in rb-tree */ 1283 unsigned int candidate_ratio; /* candidate ratio */ 1284 unsigned int max_candidate_count; /* max candidate count */ 1285 unsigned int age_weight; /* age weight, vblock_weight = 100 - age_weight */ 1286 unsigned long long age_threshold; /* age threshold */ 1287 }; 1288 1289 struct f2fs_gc_control { 1290 unsigned int victim_segno; /* target victim segment number */ 1291 int init_gc_type; /* FG_GC or BG_GC */ 1292 bool no_bg_gc; /* check the space and stop bg_gc */ 1293 bool should_migrate_blocks; /* should migrate blocks */ 1294 bool err_gc_skipped; /* return EAGAIN if GC skipped */ 1295 unsigned int nr_free_secs; /* # of free sections to do GC */ 1296 }; 1297 1298 /* 1299 * For s_flag in struct f2fs_sb_info 1300 * Modification on enum should be synchronized with s_flag array 1301 */ 1302 enum { 1303 SBI_IS_DIRTY, /* dirty flag for checkpoint */ 1304 SBI_IS_CLOSE, /* specify unmounting */ 1305 SBI_NEED_FSCK, /* need fsck.f2fs to fix */ 1306 SBI_POR_DOING, /* recovery is doing or not */ 1307 SBI_NEED_SB_WRITE, /* need to recover superblock */ 1308 SBI_NEED_CP, /* need to checkpoint */ 1309 SBI_IS_SHUTDOWN, /* shutdown by ioctl */ 1310 SBI_IS_RECOVERED, /* recovered orphan/data */ 1311 SBI_CP_DISABLED, /* CP was disabled last mount */ 1312 SBI_CP_DISABLED_QUICK, /* CP was disabled quickly */ 1313 SBI_QUOTA_NEED_FLUSH, /* need to flush quota info in CP */ 1314 SBI_QUOTA_SKIP_FLUSH, /* skip flushing quota in current CP */ 1315 SBI_QUOTA_NEED_REPAIR, /* quota file may be corrupted */ 1316 SBI_IS_RESIZEFS, /* resizefs is in process */ 1317 SBI_IS_FREEZING, /* freezefs is in process */ 1318 SBI_IS_WRITABLE, /* remove ro mountoption transiently */ 1319 MAX_SBI_FLAG, 1320 }; 1321 1322 enum { 1323 CP_TIME, 1324 REQ_TIME, 1325 DISCARD_TIME, 1326 GC_TIME, 1327 DISABLE_TIME, 1328 UMOUNT_DISCARD_TIMEOUT, 1329 MAX_TIME, 1330 }; 1331 1332 /* Note that you need to keep synchronization with this gc_mode_names array */ 1333 enum { 1334 GC_NORMAL, 1335 GC_IDLE_CB, 1336 GC_IDLE_GREEDY, 1337 GC_IDLE_AT, 1338 GC_URGENT_HIGH, 1339 GC_URGENT_LOW, 1340 GC_URGENT_MID, 1341 MAX_GC_MODE, 1342 }; 1343 1344 enum { 1345 BGGC_MODE_ON, /* background gc is on */ 1346 BGGC_MODE_OFF, /* background gc is off */ 1347 BGGC_MODE_SYNC, /* 1348 * background gc is on, migrating blocks 1349 * like foreground gc 1350 */ 1351 }; 1352 1353 enum { 1354 FS_MODE_ADAPTIVE, /* use both lfs/ssr allocation */ 1355 FS_MODE_LFS, /* use lfs allocation only */ 1356 FS_MODE_FRAGMENT_SEG, /* segment fragmentation mode */ 1357 FS_MODE_FRAGMENT_BLK, /* block fragmentation mode */ 1358 }; 1359 1360 enum { 1361 ALLOC_MODE_DEFAULT, /* stay default */ 1362 ALLOC_MODE_REUSE, /* reuse segments as much as possible */ 1363 }; 1364 1365 enum fsync_mode { 1366 FSYNC_MODE_POSIX, /* fsync follows posix semantics */ 1367 FSYNC_MODE_STRICT, /* fsync behaves in line with ext4 */ 1368 FSYNC_MODE_NOBARRIER, /* fsync behaves nobarrier based on posix */ 1369 }; 1370 1371 enum { 1372 COMPR_MODE_FS, /* 1373 * automatically compress compression 1374 * enabled files 1375 */ 1376 COMPR_MODE_USER, /* 1377 * automatical compression is disabled. 1378 * user can control the file compression 1379 * using ioctls 1380 */ 1381 }; 1382 1383 enum { 1384 DISCARD_UNIT_BLOCK, /* basic discard unit is block */ 1385 DISCARD_UNIT_SEGMENT, /* basic discard unit is segment */ 1386 DISCARD_UNIT_SECTION, /* basic discard unit is section */ 1387 }; 1388 1389 enum { 1390 MEMORY_MODE_NORMAL, /* memory mode for normal devices */ 1391 MEMORY_MODE_LOW, /* memory mode for low memry devices */ 1392 }; 1393 1394 enum errors_option { 1395 MOUNT_ERRORS_READONLY, /* remount fs ro on errors */ 1396 MOUNT_ERRORS_CONTINUE, /* continue on errors */ 1397 MOUNT_ERRORS_PANIC, /* panic on errors */ 1398 }; 1399 1400 enum { 1401 BACKGROUND, 1402 FOREGROUND, 1403 MAX_CALL_TYPE, 1404 TOTAL_CALL = FOREGROUND, 1405 }; 1406 1407 static inline int f2fs_test_bit(unsigned int nr, char *addr); 1408 static inline void f2fs_set_bit(unsigned int nr, char *addr); 1409 static inline void f2fs_clear_bit(unsigned int nr, char *addr); 1410 1411 /* 1412 * Layout of f2fs page.private: 1413 * 1414 * Layout A: lowest bit should be 1 1415 * | bit0 = 1 | bit1 | bit2 | ... | bit MAX | private data .... | 1416 * bit 0 PAGE_PRIVATE_NOT_POINTER 1417 * bit 1 PAGE_PRIVATE_ONGOING_MIGRATION 1418 * bit 2 PAGE_PRIVATE_INLINE_INODE 1419 * bit 3 PAGE_PRIVATE_REF_RESOURCE 1420 * bit 4- f2fs private data 1421 * 1422 * Layout B: lowest bit should be 0 1423 * page.private is a wrapped pointer. 1424 */ 1425 enum { 1426 PAGE_PRIVATE_NOT_POINTER, /* private contains non-pointer data */ 1427 PAGE_PRIVATE_ONGOING_MIGRATION, /* data page which is on-going migrating */ 1428 PAGE_PRIVATE_INLINE_INODE, /* inode page contains inline data */ 1429 PAGE_PRIVATE_REF_RESOURCE, /* dirty page has referenced resources */ 1430 PAGE_PRIVATE_MAX 1431 }; 1432 1433 /* For compression */ 1434 enum compress_algorithm_type { 1435 COMPRESS_LZO, 1436 COMPRESS_LZ4, 1437 COMPRESS_ZSTD, 1438 COMPRESS_LZORLE, 1439 COMPRESS_MAX, 1440 }; 1441 1442 enum compress_flag { 1443 COMPRESS_CHKSUM, 1444 COMPRESS_MAX_FLAG, 1445 }; 1446 1447 #define COMPRESS_WATERMARK 20 1448 #define COMPRESS_PERCENT 20 1449 1450 #define COMPRESS_DATA_RESERVED_SIZE 4 1451 struct compress_data { 1452 __le32 clen; /* compressed data size */ 1453 __le32 chksum; /* compressed data chksum */ 1454 __le32 reserved[COMPRESS_DATA_RESERVED_SIZE]; /* reserved */ 1455 u8 cdata[]; /* compressed data */ 1456 }; 1457 1458 #define COMPRESS_HEADER_SIZE (sizeof(struct compress_data)) 1459 1460 #define F2FS_COMPRESSED_PAGE_MAGIC 0xF5F2C000 1461 1462 #define F2FS_ZSTD_DEFAULT_CLEVEL 1 1463 1464 #define COMPRESS_LEVEL_OFFSET 8 1465 1466 /* compress context */ 1467 struct compress_ctx { 1468 struct inode *inode; /* inode the context belong to */ 1469 pgoff_t cluster_idx; /* cluster index number */ 1470 unsigned int cluster_size; /* page count in cluster */ 1471 unsigned int log_cluster_size; /* log of cluster size */ 1472 struct page **rpages; /* pages store raw data in cluster */ 1473 unsigned int nr_rpages; /* total page number in rpages */ 1474 struct page **cpages; /* pages store compressed data in cluster */ 1475 unsigned int nr_cpages; /* total page number in cpages */ 1476 unsigned int valid_nr_cpages; /* valid page number in cpages */ 1477 void *rbuf; /* virtual mapped address on rpages */ 1478 struct compress_data *cbuf; /* virtual mapped address on cpages */ 1479 size_t rlen; /* valid data length in rbuf */ 1480 size_t clen; /* valid data length in cbuf */ 1481 void *private; /* payload buffer for specified compression algorithm */ 1482 void *private2; /* extra payload buffer */ 1483 }; 1484 1485 /* compress context for write IO path */ 1486 struct compress_io_ctx { 1487 u32 magic; /* magic number to indicate page is compressed */ 1488 struct inode *inode; /* inode the context belong to */ 1489 struct page **rpages; /* pages store raw data in cluster */ 1490 unsigned int nr_rpages; /* total page number in rpages */ 1491 atomic_t pending_pages; /* in-flight compressed page count */ 1492 }; 1493 1494 /* Context for decompressing one cluster on the read IO path */ 1495 struct decompress_io_ctx { 1496 u32 magic; /* magic number to indicate page is compressed */ 1497 struct inode *inode; /* inode the context belong to */ 1498 pgoff_t cluster_idx; /* cluster index number */ 1499 unsigned int cluster_size; /* page count in cluster */ 1500 unsigned int log_cluster_size; /* log of cluster size */ 1501 struct page **rpages; /* pages store raw data in cluster */ 1502 unsigned int nr_rpages; /* total page number in rpages */ 1503 struct page **cpages; /* pages store compressed data in cluster */ 1504 unsigned int nr_cpages; /* total page number in cpages */ 1505 struct page **tpages; /* temp pages to pad holes in cluster */ 1506 void *rbuf; /* virtual mapped address on rpages */ 1507 struct compress_data *cbuf; /* virtual mapped address on cpages */ 1508 size_t rlen; /* valid data length in rbuf */ 1509 size_t clen; /* valid data length in cbuf */ 1510 1511 /* 1512 * The number of compressed pages remaining to be read in this cluster. 1513 * This is initially nr_cpages. It is decremented by 1 each time a page 1514 * has been read (or failed to be read). When it reaches 0, the cluster 1515 * is decompressed (or an error is reported). 1516 * 1517 * If an error occurs before all the pages have been submitted for I/O, 1518 * then this will never reach 0. In this case the I/O submitter is 1519 * responsible for calling f2fs_decompress_end_io() instead. 1520 */ 1521 atomic_t remaining_pages; 1522 1523 /* 1524 * Number of references to this decompress_io_ctx. 1525 * 1526 * One reference is held for I/O completion. This reference is dropped 1527 * after the pagecache pages are updated and unlocked -- either after 1528 * decompression (and verity if enabled), or after an error. 1529 * 1530 * In addition, each compressed page holds a reference while it is in a 1531 * bio. These references are necessary prevent compressed pages from 1532 * being freed while they are still in a bio. 1533 */ 1534 refcount_t refcnt; 1535 1536 bool failed; /* IO error occurred before decompression? */ 1537 bool need_verity; /* need fs-verity verification after decompression? */ 1538 void *private; /* payload buffer for specified decompression algorithm */ 1539 void *private2; /* extra payload buffer */ 1540 struct work_struct verity_work; /* work to verify the decompressed pages */ 1541 struct work_struct free_work; /* work for late free this structure itself */ 1542 }; 1543 1544 #define NULL_CLUSTER ((unsigned int)(~0)) 1545 #define MIN_COMPRESS_LOG_SIZE 2 1546 #define MAX_COMPRESS_LOG_SIZE 8 1547 #define MAX_COMPRESS_WINDOW_SIZE(log_size) ((PAGE_SIZE) << (log_size)) 1548 1549 struct f2fs_sb_info { 1550 struct super_block *sb; /* pointer to VFS super block */ 1551 struct proc_dir_entry *s_proc; /* proc entry */ 1552 struct f2fs_super_block *raw_super; /* raw super block pointer */ 1553 struct f2fs_rwsem sb_lock; /* lock for raw super block */ 1554 int valid_super_block; /* valid super block no */ 1555 unsigned long s_flag; /* flags for sbi */ 1556 struct mutex writepages; /* mutex for writepages() */ 1557 1558 #ifdef CONFIG_BLK_DEV_ZONED 1559 unsigned int blocks_per_blkz; /* F2FS blocks per zone */ 1560 #endif 1561 1562 /* for node-related operations */ 1563 struct f2fs_nm_info *nm_info; /* node manager */ 1564 struct inode *node_inode; /* cache node blocks */ 1565 1566 /* for segment-related operations */ 1567 struct f2fs_sm_info *sm_info; /* segment manager */ 1568 1569 /* for bio operations */ 1570 struct f2fs_bio_info *write_io[NR_PAGE_TYPE]; /* for write bios */ 1571 /* keep migration IO order for LFS mode */ 1572 struct f2fs_rwsem io_order_lock; 1573 pgoff_t page_eio_ofs[NR_PAGE_TYPE]; /* EIO page offset */ 1574 int page_eio_cnt[NR_PAGE_TYPE]; /* EIO count */ 1575 1576 /* for checkpoint */ 1577 struct f2fs_checkpoint *ckpt; /* raw checkpoint pointer */ 1578 int cur_cp_pack; /* remain current cp pack */ 1579 spinlock_t cp_lock; /* for flag in ckpt */ 1580 struct inode *meta_inode; /* cache meta blocks */ 1581 struct f2fs_rwsem cp_global_sem; /* checkpoint procedure lock */ 1582 struct f2fs_rwsem cp_rwsem; /* blocking FS operations */ 1583 struct f2fs_rwsem node_write; /* locking node writes */ 1584 struct f2fs_rwsem node_change; /* locking node change */ 1585 wait_queue_head_t cp_wait; 1586 unsigned long last_time[MAX_TIME]; /* to store time in jiffies */ 1587 long interval_time[MAX_TIME]; /* to store thresholds */ 1588 struct ckpt_req_control cprc_info; /* for checkpoint request control */ 1589 1590 struct inode_management im[MAX_INO_ENTRY]; /* manage inode cache */ 1591 1592 spinlock_t fsync_node_lock; /* for node entry lock */ 1593 struct list_head fsync_node_list; /* node list head */ 1594 unsigned int fsync_seg_id; /* sequence id */ 1595 unsigned int fsync_node_num; /* number of node entries */ 1596 1597 /* for orphan inode, use 0'th array */ 1598 unsigned int max_orphans; /* max orphan inodes */ 1599 1600 /* for inode management */ 1601 struct list_head inode_list[NR_INODE_TYPE]; /* dirty inode list */ 1602 spinlock_t inode_lock[NR_INODE_TYPE]; /* for dirty inode list lock */ 1603 struct mutex flush_lock; /* for flush exclusion */ 1604 1605 /* for extent tree cache */ 1606 struct extent_tree_info extent_tree[NR_EXTENT_CACHES]; 1607 atomic64_t allocated_data_blocks; /* for block age extent_cache */ 1608 1609 /* The threshold used for hot and warm data seperation*/ 1610 unsigned int hot_data_age_threshold; 1611 unsigned int warm_data_age_threshold; 1612 unsigned int last_age_weight; 1613 1614 /* basic filesystem units */ 1615 unsigned int log_sectors_per_block; /* log2 sectors per block */ 1616 unsigned int log_blocksize; /* log2 block size */ 1617 unsigned int blocksize; /* block size */ 1618 unsigned int root_ino_num; /* root inode number*/ 1619 unsigned int node_ino_num; /* node inode number*/ 1620 unsigned int meta_ino_num; /* meta inode number*/ 1621 unsigned int log_blocks_per_seg; /* log2 blocks per segment */ 1622 unsigned int blocks_per_seg; /* blocks per segment */ 1623 unsigned int unusable_blocks_per_sec; /* unusable blocks per section */ 1624 unsigned int segs_per_sec; /* segments per section */ 1625 unsigned int secs_per_zone; /* sections per zone */ 1626 unsigned int total_sections; /* total section count */ 1627 unsigned int total_node_count; /* total node block count */ 1628 unsigned int total_valid_node_count; /* valid node block count */ 1629 int dir_level; /* directory level */ 1630 bool readdir_ra; /* readahead inode in readdir */ 1631 u64 max_io_bytes; /* max io bytes to merge IOs */ 1632 1633 block_t user_block_count; /* # of user blocks */ 1634 block_t total_valid_block_count; /* # of valid blocks */ 1635 block_t discard_blks; /* discard command candidats */ 1636 block_t last_valid_block_count; /* for recovery */ 1637 block_t reserved_blocks; /* configurable reserved blocks */ 1638 block_t current_reserved_blocks; /* current reserved blocks */ 1639 1640 /* Additional tracking for no checkpoint mode */ 1641 block_t unusable_block_count; /* # of blocks saved by last cp */ 1642 1643 unsigned int nquota_files; /* # of quota sysfile */ 1644 struct f2fs_rwsem quota_sem; /* blocking cp for flags */ 1645 1646 /* # of pages, see count_type */ 1647 atomic_t nr_pages[NR_COUNT_TYPE]; 1648 /* # of allocated blocks */ 1649 struct percpu_counter alloc_valid_block_count; 1650 /* # of node block writes as roll forward recovery */ 1651 struct percpu_counter rf_node_block_count; 1652 1653 /* writeback control */ 1654 atomic_t wb_sync_req[META]; /* count # of WB_SYNC threads */ 1655 1656 /* valid inode count */ 1657 struct percpu_counter total_valid_inode_count; 1658 1659 struct f2fs_mount_info mount_opt; /* mount options */ 1660 1661 /* for cleaning operations */ 1662 struct f2fs_rwsem gc_lock; /* 1663 * semaphore for GC, avoid 1664 * race between GC and GC or CP 1665 */ 1666 struct f2fs_gc_kthread *gc_thread; /* GC thread */ 1667 struct atgc_management am; /* atgc management */ 1668 unsigned int cur_victim_sec; /* current victim section num */ 1669 unsigned int gc_mode; /* current GC state */ 1670 unsigned int next_victim_seg[2]; /* next segment in victim section */ 1671 spinlock_t gc_remaining_trials_lock; 1672 /* remaining trial count for GC_URGENT_* and GC_IDLE_* */ 1673 unsigned int gc_remaining_trials; 1674 1675 /* for skip statistic */ 1676 unsigned long long skipped_gc_rwsem; /* FG_GC only */ 1677 1678 /* threshold for gc trials on pinned files */ 1679 u64 gc_pin_file_threshold; 1680 struct f2fs_rwsem pin_sem; 1681 1682 /* maximum # of trials to find a victim segment for SSR and GC */ 1683 unsigned int max_victim_search; 1684 /* migration granularity of garbage collection, unit: segment */ 1685 unsigned int migration_granularity; 1686 1687 /* 1688 * for stat information. 1689 * one is for the LFS mode, and the other is for the SSR mode. 1690 */ 1691 #ifdef CONFIG_F2FS_STAT_FS 1692 struct f2fs_stat_info *stat_info; /* FS status information */ 1693 atomic_t meta_count[META_MAX]; /* # of meta blocks */ 1694 unsigned int segment_count[2]; /* # of allocated segments */ 1695 unsigned int block_count[2]; /* # of allocated blocks */ 1696 atomic_t inplace_count; /* # of inplace update */ 1697 /* # of lookup extent cache */ 1698 atomic64_t total_hit_ext[NR_EXTENT_CACHES]; 1699 /* # of hit rbtree extent node */ 1700 atomic64_t read_hit_rbtree[NR_EXTENT_CACHES]; 1701 /* # of hit cached extent node */ 1702 atomic64_t read_hit_cached[NR_EXTENT_CACHES]; 1703 /* # of hit largest extent node in read extent cache */ 1704 atomic64_t read_hit_largest; 1705 atomic_t inline_xattr; /* # of inline_xattr inodes */ 1706 atomic_t inline_inode; /* # of inline_data inodes */ 1707 atomic_t inline_dir; /* # of inline_dentry inodes */ 1708 atomic_t compr_inode; /* # of compressed inodes */ 1709 atomic64_t compr_blocks; /* # of compressed blocks */ 1710 atomic_t swapfile_inode; /* # of swapfile inodes */ 1711 atomic_t atomic_files; /* # of opened atomic file */ 1712 atomic_t max_aw_cnt; /* max # of atomic writes */ 1713 unsigned int io_skip_bggc; /* skip background gc for in-flight IO */ 1714 unsigned int other_skip_bggc; /* skip background gc for other reasons */ 1715 unsigned int ndirty_inode[NR_INODE_TYPE]; /* # of dirty inodes */ 1716 atomic_t cp_call_count[MAX_CALL_TYPE]; /* # of cp call */ 1717 #endif 1718 spinlock_t stat_lock; /* lock for stat operations */ 1719 1720 /* to attach REQ_META|REQ_FUA flags */ 1721 unsigned int data_io_flag; 1722 unsigned int node_io_flag; 1723 1724 /* For sysfs support */ 1725 struct kobject s_kobj; /* /sys/fs/f2fs/<devname> */ 1726 struct completion s_kobj_unregister; 1727 1728 struct kobject s_stat_kobj; /* /sys/fs/f2fs/<devname>/stat */ 1729 struct completion s_stat_kobj_unregister; 1730 1731 struct kobject s_feature_list_kobj; /* /sys/fs/f2fs/<devname>/feature_list */ 1732 struct completion s_feature_list_kobj_unregister; 1733 1734 /* For shrinker support */ 1735 struct list_head s_list; 1736 struct mutex umount_mutex; 1737 unsigned int shrinker_run_no; 1738 1739 /* For multi devices */ 1740 int s_ndevs; /* number of devices */ 1741 struct f2fs_dev_info *devs; /* for device list */ 1742 unsigned int dirty_device; /* for checkpoint data flush */ 1743 spinlock_t dev_lock; /* protect dirty_device */ 1744 bool aligned_blksize; /* all devices has the same logical blksize */ 1745 1746 /* For write statistics */ 1747 u64 sectors_written_start; 1748 u64 kbytes_written; 1749 1750 /* Reference to checksum algorithm driver via cryptoapi */ 1751 struct crypto_shash *s_chksum_driver; 1752 1753 /* Precomputed FS UUID checksum for seeding other checksums */ 1754 __u32 s_chksum_seed; 1755 1756 struct workqueue_struct *post_read_wq; /* post read workqueue */ 1757 1758 /* 1759 * If we are in irq context, let's update error information into 1760 * on-disk superblock in the work. 1761 */ 1762 struct work_struct s_error_work; 1763 unsigned char errors[MAX_F2FS_ERRORS]; /* error flags */ 1764 unsigned char stop_reason[MAX_STOP_REASON]; /* stop reason */ 1765 spinlock_t error_lock; /* protect errors/stop_reason array */ 1766 bool error_dirty; /* errors of sb is dirty */ 1767 1768 struct kmem_cache *inline_xattr_slab; /* inline xattr entry */ 1769 unsigned int inline_xattr_slab_size; /* default inline xattr slab size */ 1770 1771 /* For reclaimed segs statistics per each GC mode */ 1772 unsigned int gc_segment_mode; /* GC state for reclaimed segments */ 1773 unsigned int gc_reclaimed_segs[MAX_GC_MODE]; /* Reclaimed segs for each mode */ 1774 1775 unsigned long seq_file_ra_mul; /* multiplier for ra_pages of seq. files in fadvise */ 1776 1777 int max_fragment_chunk; /* max chunk size for block fragmentation mode */ 1778 int max_fragment_hole; /* max hole size for block fragmentation mode */ 1779 1780 /* For atomic write statistics */ 1781 atomic64_t current_atomic_write; 1782 s64 peak_atomic_write; 1783 u64 committed_atomic_block; 1784 u64 revoked_atomic_block; 1785 1786 #ifdef CONFIG_F2FS_FS_COMPRESSION 1787 struct kmem_cache *page_array_slab; /* page array entry */ 1788 unsigned int page_array_slab_size; /* default page array slab size */ 1789 1790 /* For runtime compression statistics */ 1791 u64 compr_written_block; 1792 u64 compr_saved_block; 1793 u32 compr_new_inode; 1794 1795 /* For compressed block cache */ 1796 struct inode *compress_inode; /* cache compressed blocks */ 1797 unsigned int compress_percent; /* cache page percentage */ 1798 unsigned int compress_watermark; /* cache page watermark */ 1799 atomic_t compress_page_hit; /* cache hit count */ 1800 #endif 1801 1802 #ifdef CONFIG_F2FS_IOSTAT 1803 /* For app/fs IO statistics */ 1804 spinlock_t iostat_lock; 1805 unsigned long long iostat_count[NR_IO_TYPE]; 1806 unsigned long long iostat_bytes[NR_IO_TYPE]; 1807 unsigned long long prev_iostat_bytes[NR_IO_TYPE]; 1808 bool iostat_enable; 1809 unsigned long iostat_next_period; 1810 unsigned int iostat_period_ms; 1811 1812 /* For io latency related statistics info in one iostat period */ 1813 spinlock_t iostat_lat_lock; 1814 struct iostat_lat_info *iostat_io_lat; 1815 #endif 1816 }; 1817 1818 /* Definitions to access f2fs_sb_info */ 1819 #define SEGS_TO_BLKS(sbi, segs) \ 1820 ((segs) << (sbi)->log_blocks_per_seg) 1821 #define BLKS_TO_SEGS(sbi, blks) \ 1822 ((blks) >> (sbi)->log_blocks_per_seg) 1823 1824 #define BLKS_PER_SEG(sbi) ((sbi)->blocks_per_seg) 1825 #define BLKS_PER_SEC(sbi) (SEGS_TO_BLKS(sbi, (sbi)->segs_per_sec)) 1826 #define SEGS_PER_SEC(sbi) ((sbi)->segs_per_sec) 1827 1828 __printf(3, 4) 1829 void f2fs_printk(struct f2fs_sb_info *sbi, bool limit_rate, const char *fmt, ...); 1830 1831 #define f2fs_err(sbi, fmt, ...) \ 1832 f2fs_printk(sbi, false, KERN_ERR fmt, ##__VA_ARGS__) 1833 #define f2fs_warn(sbi, fmt, ...) \ 1834 f2fs_printk(sbi, false, KERN_WARNING fmt, ##__VA_ARGS__) 1835 #define f2fs_notice(sbi, fmt, ...) \ 1836 f2fs_printk(sbi, false, KERN_NOTICE fmt, ##__VA_ARGS__) 1837 #define f2fs_info(sbi, fmt, ...) \ 1838 f2fs_printk(sbi, false, KERN_INFO fmt, ##__VA_ARGS__) 1839 #define f2fs_debug(sbi, fmt, ...) \ 1840 f2fs_printk(sbi, false, KERN_DEBUG fmt, ##__VA_ARGS__) 1841 1842 #define f2fs_err_ratelimited(sbi, fmt, ...) \ 1843 f2fs_printk(sbi, true, KERN_ERR fmt, ##__VA_ARGS__) 1844 #define f2fs_warn_ratelimited(sbi, fmt, ...) \ 1845 f2fs_printk(sbi, true, KERN_WARNING fmt, ##__VA_ARGS__) 1846 #define f2fs_info_ratelimited(sbi, fmt, ...) \ 1847 f2fs_printk(sbi, true, KERN_INFO fmt, ##__VA_ARGS__) 1848 1849 #ifdef CONFIG_F2FS_FAULT_INJECTION 1850 #define time_to_inject(sbi, type) __time_to_inject(sbi, type, __func__, \ 1851 __builtin_return_address(0)) 1852 static inline bool __time_to_inject(struct f2fs_sb_info *sbi, int type, 1853 const char *func, const char *parent_func) 1854 { 1855 struct f2fs_fault_info *ffi = &F2FS_OPTION(sbi).fault_info; 1856 1857 if (!ffi->inject_rate) 1858 return false; 1859 1860 if (!IS_FAULT_SET(ffi, type)) 1861 return false; 1862 1863 atomic_inc(&ffi->inject_ops); 1864 if (atomic_read(&ffi->inject_ops) >= ffi->inject_rate) { 1865 atomic_set(&ffi->inject_ops, 0); 1866 f2fs_info_ratelimited(sbi, "inject %s in %s of %pS", 1867 f2fs_fault_name[type], func, parent_func); 1868 return true; 1869 } 1870 return false; 1871 } 1872 #else 1873 static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type) 1874 { 1875 return false; 1876 } 1877 #endif 1878 1879 /* 1880 * Test if the mounted volume is a multi-device volume. 1881 * - For a single regular disk volume, sbi->s_ndevs is 0. 1882 * - For a single zoned disk volume, sbi->s_ndevs is 1. 1883 * - For a multi-device volume, sbi->s_ndevs is always 2 or more. 1884 */ 1885 static inline bool f2fs_is_multi_device(struct f2fs_sb_info *sbi) 1886 { 1887 return sbi->s_ndevs > 1; 1888 } 1889 1890 static inline void f2fs_update_time(struct f2fs_sb_info *sbi, int type) 1891 { 1892 unsigned long now = jiffies; 1893 1894 sbi->last_time[type] = now; 1895 1896 /* DISCARD_TIME and GC_TIME are based on REQ_TIME */ 1897 if (type == REQ_TIME) { 1898 sbi->last_time[DISCARD_TIME] = now; 1899 sbi->last_time[GC_TIME] = now; 1900 } 1901 } 1902 1903 static inline bool f2fs_time_over(struct f2fs_sb_info *sbi, int type) 1904 { 1905 unsigned long interval = sbi->interval_time[type] * HZ; 1906 1907 return time_after(jiffies, sbi->last_time[type] + interval); 1908 } 1909 1910 static inline unsigned int f2fs_time_to_wait(struct f2fs_sb_info *sbi, 1911 int type) 1912 { 1913 unsigned long interval = sbi->interval_time[type] * HZ; 1914 unsigned int wait_ms = 0; 1915 long delta; 1916 1917 delta = (sbi->last_time[type] + interval) - jiffies; 1918 if (delta > 0) 1919 wait_ms = jiffies_to_msecs(delta); 1920 1921 return wait_ms; 1922 } 1923 1924 /* 1925 * Inline functions 1926 */ 1927 static inline u32 __f2fs_crc32(struct f2fs_sb_info *sbi, u32 crc, 1928 const void *address, unsigned int length) 1929 { 1930 struct { 1931 struct shash_desc shash; 1932 char ctx[4]; 1933 } desc; 1934 int err; 1935 1936 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver) != sizeof(desc.ctx)); 1937 1938 desc.shash.tfm = sbi->s_chksum_driver; 1939 *(u32 *)desc.ctx = crc; 1940 1941 err = crypto_shash_update(&desc.shash, address, length); 1942 BUG_ON(err); 1943 1944 return *(u32 *)desc.ctx; 1945 } 1946 1947 static inline u32 f2fs_crc32(struct f2fs_sb_info *sbi, const void *address, 1948 unsigned int length) 1949 { 1950 return __f2fs_crc32(sbi, F2FS_SUPER_MAGIC, address, length); 1951 } 1952 1953 static inline bool f2fs_crc_valid(struct f2fs_sb_info *sbi, __u32 blk_crc, 1954 void *buf, size_t buf_size) 1955 { 1956 return f2fs_crc32(sbi, buf, buf_size) == blk_crc; 1957 } 1958 1959 static inline u32 f2fs_chksum(struct f2fs_sb_info *sbi, u32 crc, 1960 const void *address, unsigned int length) 1961 { 1962 return __f2fs_crc32(sbi, crc, address, length); 1963 } 1964 1965 static inline struct f2fs_inode_info *F2FS_I(struct inode *inode) 1966 { 1967 return container_of(inode, struct f2fs_inode_info, vfs_inode); 1968 } 1969 1970 static inline struct f2fs_sb_info *F2FS_SB(struct super_block *sb) 1971 { 1972 return sb->s_fs_info; 1973 } 1974 1975 static inline struct f2fs_sb_info *F2FS_I_SB(struct inode *inode) 1976 { 1977 return F2FS_SB(inode->i_sb); 1978 } 1979 1980 static inline struct f2fs_sb_info *F2FS_M_SB(struct address_space *mapping) 1981 { 1982 return F2FS_I_SB(mapping->host); 1983 } 1984 1985 static inline struct f2fs_sb_info *F2FS_P_SB(struct page *page) 1986 { 1987 return F2FS_M_SB(page_file_mapping(page)); 1988 } 1989 1990 static inline struct f2fs_super_block *F2FS_RAW_SUPER(struct f2fs_sb_info *sbi) 1991 { 1992 return (struct f2fs_super_block *)(sbi->raw_super); 1993 } 1994 1995 static inline struct f2fs_checkpoint *F2FS_CKPT(struct f2fs_sb_info *sbi) 1996 { 1997 return (struct f2fs_checkpoint *)(sbi->ckpt); 1998 } 1999 2000 static inline struct f2fs_node *F2FS_NODE(struct page *page) 2001 { 2002 return (struct f2fs_node *)page_address(page); 2003 } 2004 2005 static inline struct f2fs_inode *F2FS_INODE(struct page *page) 2006 { 2007 return &((struct f2fs_node *)page_address(page))->i; 2008 } 2009 2010 static inline struct f2fs_nm_info *NM_I(struct f2fs_sb_info *sbi) 2011 { 2012 return (struct f2fs_nm_info *)(sbi->nm_info); 2013 } 2014 2015 static inline struct f2fs_sm_info *SM_I(struct f2fs_sb_info *sbi) 2016 { 2017 return (struct f2fs_sm_info *)(sbi->sm_info); 2018 } 2019 2020 static inline struct sit_info *SIT_I(struct f2fs_sb_info *sbi) 2021 { 2022 return (struct sit_info *)(SM_I(sbi)->sit_info); 2023 } 2024 2025 static inline struct free_segmap_info *FREE_I(struct f2fs_sb_info *sbi) 2026 { 2027 return (struct free_segmap_info *)(SM_I(sbi)->free_info); 2028 } 2029 2030 static inline struct dirty_seglist_info *DIRTY_I(struct f2fs_sb_info *sbi) 2031 { 2032 return (struct dirty_seglist_info *)(SM_I(sbi)->dirty_info); 2033 } 2034 2035 static inline struct address_space *META_MAPPING(struct f2fs_sb_info *sbi) 2036 { 2037 return sbi->meta_inode->i_mapping; 2038 } 2039 2040 static inline struct address_space *NODE_MAPPING(struct f2fs_sb_info *sbi) 2041 { 2042 return sbi->node_inode->i_mapping; 2043 } 2044 2045 static inline bool is_sbi_flag_set(struct f2fs_sb_info *sbi, unsigned int type) 2046 { 2047 return test_bit(type, &sbi->s_flag); 2048 } 2049 2050 static inline void set_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type) 2051 { 2052 set_bit(type, &sbi->s_flag); 2053 } 2054 2055 static inline void clear_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type) 2056 { 2057 clear_bit(type, &sbi->s_flag); 2058 } 2059 2060 static inline unsigned long long cur_cp_version(struct f2fs_checkpoint *cp) 2061 { 2062 return le64_to_cpu(cp->checkpoint_ver); 2063 } 2064 2065 static inline unsigned long f2fs_qf_ino(struct super_block *sb, int type) 2066 { 2067 if (type < F2FS_MAX_QUOTAS) 2068 return le32_to_cpu(F2FS_SB(sb)->raw_super->qf_ino[type]); 2069 return 0; 2070 } 2071 2072 static inline __u64 cur_cp_crc(struct f2fs_checkpoint *cp) 2073 { 2074 size_t crc_offset = le32_to_cpu(cp->checksum_offset); 2075 return le32_to_cpu(*((__le32 *)((unsigned char *)cp + crc_offset))); 2076 } 2077 2078 static inline bool __is_set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 2079 { 2080 unsigned int ckpt_flags = le32_to_cpu(cp->ckpt_flags); 2081 2082 return ckpt_flags & f; 2083 } 2084 2085 static inline bool is_set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 2086 { 2087 return __is_set_ckpt_flags(F2FS_CKPT(sbi), f); 2088 } 2089 2090 static inline void __set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 2091 { 2092 unsigned int ckpt_flags; 2093 2094 ckpt_flags = le32_to_cpu(cp->ckpt_flags); 2095 ckpt_flags |= f; 2096 cp->ckpt_flags = cpu_to_le32(ckpt_flags); 2097 } 2098 2099 static inline void set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 2100 { 2101 unsigned long flags; 2102 2103 spin_lock_irqsave(&sbi->cp_lock, flags); 2104 __set_ckpt_flags(F2FS_CKPT(sbi), f); 2105 spin_unlock_irqrestore(&sbi->cp_lock, flags); 2106 } 2107 2108 static inline void __clear_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 2109 { 2110 unsigned int ckpt_flags; 2111 2112 ckpt_flags = le32_to_cpu(cp->ckpt_flags); 2113 ckpt_flags &= (~f); 2114 cp->ckpt_flags = cpu_to_le32(ckpt_flags); 2115 } 2116 2117 static inline void clear_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 2118 { 2119 unsigned long flags; 2120 2121 spin_lock_irqsave(&sbi->cp_lock, flags); 2122 __clear_ckpt_flags(F2FS_CKPT(sbi), f); 2123 spin_unlock_irqrestore(&sbi->cp_lock, flags); 2124 } 2125 2126 #define init_f2fs_rwsem(sem) \ 2127 do { \ 2128 static struct lock_class_key __key; \ 2129 \ 2130 __init_f2fs_rwsem((sem), #sem, &__key); \ 2131 } while (0) 2132 2133 static inline void __init_f2fs_rwsem(struct f2fs_rwsem *sem, 2134 const char *sem_name, struct lock_class_key *key) 2135 { 2136 __init_rwsem(&sem->internal_rwsem, sem_name, key); 2137 #ifdef CONFIG_F2FS_UNFAIR_RWSEM 2138 init_waitqueue_head(&sem->read_waiters); 2139 #endif 2140 } 2141 2142 static inline int f2fs_rwsem_is_locked(struct f2fs_rwsem *sem) 2143 { 2144 return rwsem_is_locked(&sem->internal_rwsem); 2145 } 2146 2147 static inline int f2fs_rwsem_is_contended(struct f2fs_rwsem *sem) 2148 { 2149 return rwsem_is_contended(&sem->internal_rwsem); 2150 } 2151 2152 static inline void f2fs_down_read(struct f2fs_rwsem *sem) 2153 { 2154 #ifdef CONFIG_F2FS_UNFAIR_RWSEM 2155 wait_event(sem->read_waiters, down_read_trylock(&sem->internal_rwsem)); 2156 #else 2157 down_read(&sem->internal_rwsem); 2158 #endif 2159 } 2160 2161 static inline int f2fs_down_read_trylock(struct f2fs_rwsem *sem) 2162 { 2163 return down_read_trylock(&sem->internal_rwsem); 2164 } 2165 2166 static inline void f2fs_up_read(struct f2fs_rwsem *sem) 2167 { 2168 up_read(&sem->internal_rwsem); 2169 } 2170 2171 static inline void f2fs_down_write(struct f2fs_rwsem *sem) 2172 { 2173 down_write(&sem->internal_rwsem); 2174 } 2175 2176 #ifdef CONFIG_DEBUG_LOCK_ALLOC 2177 static inline void f2fs_down_read_nested(struct f2fs_rwsem *sem, int subclass) 2178 { 2179 down_read_nested(&sem->internal_rwsem, subclass); 2180 } 2181 2182 static inline void f2fs_down_write_nested(struct f2fs_rwsem *sem, int subclass) 2183 { 2184 down_write_nested(&sem->internal_rwsem, subclass); 2185 } 2186 #else 2187 #define f2fs_down_read_nested(sem, subclass) f2fs_down_read(sem) 2188 #define f2fs_down_write_nested(sem, subclass) f2fs_down_write(sem) 2189 #endif 2190 2191 static inline int f2fs_down_write_trylock(struct f2fs_rwsem *sem) 2192 { 2193 return down_write_trylock(&sem->internal_rwsem); 2194 } 2195 2196 static inline void f2fs_up_write(struct f2fs_rwsem *sem) 2197 { 2198 up_write(&sem->internal_rwsem); 2199 #ifdef CONFIG_F2FS_UNFAIR_RWSEM 2200 wake_up_all(&sem->read_waiters); 2201 #endif 2202 } 2203 2204 static inline void f2fs_lock_op(struct f2fs_sb_info *sbi) 2205 { 2206 f2fs_down_read(&sbi->cp_rwsem); 2207 } 2208 2209 static inline int f2fs_trylock_op(struct f2fs_sb_info *sbi) 2210 { 2211 if (time_to_inject(sbi, FAULT_LOCK_OP)) 2212 return 0; 2213 return f2fs_down_read_trylock(&sbi->cp_rwsem); 2214 } 2215 2216 static inline void f2fs_unlock_op(struct f2fs_sb_info *sbi) 2217 { 2218 f2fs_up_read(&sbi->cp_rwsem); 2219 } 2220 2221 static inline void f2fs_lock_all(struct f2fs_sb_info *sbi) 2222 { 2223 f2fs_down_write(&sbi->cp_rwsem); 2224 } 2225 2226 static inline void f2fs_unlock_all(struct f2fs_sb_info *sbi) 2227 { 2228 f2fs_up_write(&sbi->cp_rwsem); 2229 } 2230 2231 static inline int __get_cp_reason(struct f2fs_sb_info *sbi) 2232 { 2233 int reason = CP_SYNC; 2234 2235 if (test_opt(sbi, FASTBOOT)) 2236 reason = CP_FASTBOOT; 2237 if (is_sbi_flag_set(sbi, SBI_IS_CLOSE)) 2238 reason = CP_UMOUNT; 2239 return reason; 2240 } 2241 2242 static inline bool __remain_node_summaries(int reason) 2243 { 2244 return (reason & (CP_UMOUNT | CP_FASTBOOT)); 2245 } 2246 2247 static inline bool __exist_node_summaries(struct f2fs_sb_info *sbi) 2248 { 2249 return (is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG) || 2250 is_set_ckpt_flags(sbi, CP_FASTBOOT_FLAG)); 2251 } 2252 2253 /* 2254 * Check whether the inode has blocks or not 2255 */ 2256 static inline int F2FS_HAS_BLOCKS(struct inode *inode) 2257 { 2258 block_t xattr_block = F2FS_I(inode)->i_xattr_nid ? 1 : 0; 2259 2260 return (inode->i_blocks >> F2FS_LOG_SECTORS_PER_BLOCK) > xattr_block; 2261 } 2262 2263 static inline bool f2fs_has_xattr_block(unsigned int ofs) 2264 { 2265 return ofs == XATTR_NODE_OFFSET; 2266 } 2267 2268 static inline bool __allow_reserved_blocks(struct f2fs_sb_info *sbi, 2269 struct inode *inode, bool cap) 2270 { 2271 if (!inode) 2272 return true; 2273 if (!test_opt(sbi, RESERVE_ROOT)) 2274 return false; 2275 if (IS_NOQUOTA(inode)) 2276 return true; 2277 if (uid_eq(F2FS_OPTION(sbi).s_resuid, current_fsuid())) 2278 return true; 2279 if (!gid_eq(F2FS_OPTION(sbi).s_resgid, GLOBAL_ROOT_GID) && 2280 in_group_p(F2FS_OPTION(sbi).s_resgid)) 2281 return true; 2282 if (cap && capable(CAP_SYS_RESOURCE)) 2283 return true; 2284 return false; 2285 } 2286 2287 static inline unsigned int get_available_block_count(struct f2fs_sb_info *sbi, 2288 struct inode *inode, bool cap) 2289 { 2290 block_t avail_user_block_count; 2291 2292 avail_user_block_count = sbi->user_block_count - 2293 sbi->current_reserved_blocks; 2294 2295 if (!__allow_reserved_blocks(sbi, inode, cap)) 2296 avail_user_block_count -= F2FS_OPTION(sbi).root_reserved_blocks; 2297 2298 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) { 2299 if (avail_user_block_count > sbi->unusable_block_count) 2300 avail_user_block_count -= sbi->unusable_block_count; 2301 else 2302 avail_user_block_count = 0; 2303 } 2304 2305 return avail_user_block_count; 2306 } 2307 2308 static inline void f2fs_i_blocks_write(struct inode *, block_t, bool, bool); 2309 static inline int inc_valid_block_count(struct f2fs_sb_info *sbi, 2310 struct inode *inode, blkcnt_t *count, bool partial) 2311 { 2312 blkcnt_t diff = 0, release = 0; 2313 block_t avail_user_block_count; 2314 int ret; 2315 2316 ret = dquot_reserve_block(inode, *count); 2317 if (ret) 2318 return ret; 2319 2320 if (time_to_inject(sbi, FAULT_BLOCK)) { 2321 release = *count; 2322 goto release_quota; 2323 } 2324 2325 /* 2326 * let's increase this in prior to actual block count change in order 2327 * for f2fs_sync_file to avoid data races when deciding checkpoint. 2328 */ 2329 percpu_counter_add(&sbi->alloc_valid_block_count, (*count)); 2330 2331 spin_lock(&sbi->stat_lock); 2332 sbi->total_valid_block_count += (block_t)(*count); 2333 avail_user_block_count = get_available_block_count(sbi, inode, true); 2334 2335 if (unlikely(sbi->total_valid_block_count > avail_user_block_count)) { 2336 if (!partial) { 2337 spin_unlock(&sbi->stat_lock); 2338 goto enospc; 2339 } 2340 2341 diff = sbi->total_valid_block_count - avail_user_block_count; 2342 if (diff > *count) 2343 diff = *count; 2344 *count -= diff; 2345 release = diff; 2346 sbi->total_valid_block_count -= diff; 2347 if (!*count) { 2348 spin_unlock(&sbi->stat_lock); 2349 goto enospc; 2350 } 2351 } 2352 spin_unlock(&sbi->stat_lock); 2353 2354 if (unlikely(release)) { 2355 percpu_counter_sub(&sbi->alloc_valid_block_count, release); 2356 dquot_release_reservation_block(inode, release); 2357 } 2358 f2fs_i_blocks_write(inode, *count, true, true); 2359 return 0; 2360 2361 enospc: 2362 percpu_counter_sub(&sbi->alloc_valid_block_count, release); 2363 release_quota: 2364 dquot_release_reservation_block(inode, release); 2365 return -ENOSPC; 2366 } 2367 2368 #define PAGE_PRIVATE_GET_FUNC(name, flagname) \ 2369 static inline bool page_private_##name(struct page *page) \ 2370 { \ 2371 return PagePrivate(page) && \ 2372 test_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)) && \ 2373 test_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \ 2374 } 2375 2376 #define PAGE_PRIVATE_SET_FUNC(name, flagname) \ 2377 static inline void set_page_private_##name(struct page *page) \ 2378 { \ 2379 if (!PagePrivate(page)) \ 2380 attach_page_private(page, (void *)0); \ 2381 set_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)); \ 2382 set_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \ 2383 } 2384 2385 #define PAGE_PRIVATE_CLEAR_FUNC(name, flagname) \ 2386 static inline void clear_page_private_##name(struct page *page) \ 2387 { \ 2388 clear_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \ 2389 if (page_private(page) == BIT(PAGE_PRIVATE_NOT_POINTER)) \ 2390 detach_page_private(page); \ 2391 } 2392 2393 PAGE_PRIVATE_GET_FUNC(nonpointer, NOT_POINTER); 2394 PAGE_PRIVATE_GET_FUNC(inline, INLINE_INODE); 2395 PAGE_PRIVATE_GET_FUNC(gcing, ONGOING_MIGRATION); 2396 2397 PAGE_PRIVATE_SET_FUNC(reference, REF_RESOURCE); 2398 PAGE_PRIVATE_SET_FUNC(inline, INLINE_INODE); 2399 PAGE_PRIVATE_SET_FUNC(gcing, ONGOING_MIGRATION); 2400 2401 PAGE_PRIVATE_CLEAR_FUNC(reference, REF_RESOURCE); 2402 PAGE_PRIVATE_CLEAR_FUNC(inline, INLINE_INODE); 2403 PAGE_PRIVATE_CLEAR_FUNC(gcing, ONGOING_MIGRATION); 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 2436 f2fs_bug_on(F2FS_P_SB(page), page_private(page)); 2437 } 2438 2439 static inline void dec_valid_block_count(struct f2fs_sb_info *sbi, 2440 struct inode *inode, 2441 block_t count) 2442 { 2443 blkcnt_t sectors = count << F2FS_LOG_SECTORS_PER_BLOCK; 2444 2445 spin_lock(&sbi->stat_lock); 2446 f2fs_bug_on(sbi, sbi->total_valid_block_count < (block_t) count); 2447 sbi->total_valid_block_count -= (block_t)count; 2448 if (sbi->reserved_blocks && 2449 sbi->current_reserved_blocks < sbi->reserved_blocks) 2450 sbi->current_reserved_blocks = min(sbi->reserved_blocks, 2451 sbi->current_reserved_blocks + count); 2452 spin_unlock(&sbi->stat_lock); 2453 if (unlikely(inode->i_blocks < sectors)) { 2454 f2fs_warn(sbi, "Inconsistent i_blocks, ino:%lu, iblocks:%llu, sectors:%llu", 2455 inode->i_ino, 2456 (unsigned long long)inode->i_blocks, 2457 (unsigned long long)sectors); 2458 set_sbi_flag(sbi, SBI_NEED_FSCK); 2459 return; 2460 } 2461 f2fs_i_blocks_write(inode, count, false, true); 2462 } 2463 2464 static inline void inc_page_count(struct f2fs_sb_info *sbi, int count_type) 2465 { 2466 atomic_inc(&sbi->nr_pages[count_type]); 2467 2468 if (count_type == F2FS_DIRTY_DENTS || 2469 count_type == F2FS_DIRTY_NODES || 2470 count_type == F2FS_DIRTY_META || 2471 count_type == F2FS_DIRTY_QDATA || 2472 count_type == F2FS_DIRTY_IMETA) 2473 set_sbi_flag(sbi, SBI_IS_DIRTY); 2474 } 2475 2476 static inline void inode_inc_dirty_pages(struct inode *inode) 2477 { 2478 atomic_inc(&F2FS_I(inode)->dirty_pages); 2479 inc_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ? 2480 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA); 2481 if (IS_NOQUOTA(inode)) 2482 inc_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA); 2483 } 2484 2485 static inline void dec_page_count(struct f2fs_sb_info *sbi, int count_type) 2486 { 2487 atomic_dec(&sbi->nr_pages[count_type]); 2488 } 2489 2490 static inline void inode_dec_dirty_pages(struct inode *inode) 2491 { 2492 if (!S_ISDIR(inode->i_mode) && !S_ISREG(inode->i_mode) && 2493 !S_ISLNK(inode->i_mode)) 2494 return; 2495 2496 atomic_dec(&F2FS_I(inode)->dirty_pages); 2497 dec_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ? 2498 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA); 2499 if (IS_NOQUOTA(inode)) 2500 dec_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA); 2501 } 2502 2503 static inline void inc_atomic_write_cnt(struct inode *inode) 2504 { 2505 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 2506 struct f2fs_inode_info *fi = F2FS_I(inode); 2507 u64 current_write; 2508 2509 fi->atomic_write_cnt++; 2510 atomic64_inc(&sbi->current_atomic_write); 2511 current_write = atomic64_read(&sbi->current_atomic_write); 2512 if (current_write > sbi->peak_atomic_write) 2513 sbi->peak_atomic_write = current_write; 2514 } 2515 2516 static inline void release_atomic_write_cnt(struct inode *inode) 2517 { 2518 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 2519 struct f2fs_inode_info *fi = F2FS_I(inode); 2520 2521 atomic64_sub(fi->atomic_write_cnt, &sbi->current_atomic_write); 2522 fi->atomic_write_cnt = 0; 2523 } 2524 2525 static inline s64 get_pages(struct f2fs_sb_info *sbi, int count_type) 2526 { 2527 return atomic_read(&sbi->nr_pages[count_type]); 2528 } 2529 2530 static inline int get_dirty_pages(struct inode *inode) 2531 { 2532 return atomic_read(&F2FS_I(inode)->dirty_pages); 2533 } 2534 2535 static inline int get_blocktype_secs(struct f2fs_sb_info *sbi, int block_type) 2536 { 2537 return div_u64(get_pages(sbi, block_type) + BLKS_PER_SEC(sbi) - 1, 2538 BLKS_PER_SEC(sbi)); 2539 } 2540 2541 static inline block_t valid_user_blocks(struct f2fs_sb_info *sbi) 2542 { 2543 return sbi->total_valid_block_count; 2544 } 2545 2546 static inline block_t discard_blocks(struct f2fs_sb_info *sbi) 2547 { 2548 return sbi->discard_blks; 2549 } 2550 2551 static inline unsigned long __bitmap_size(struct f2fs_sb_info *sbi, int flag) 2552 { 2553 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi); 2554 2555 /* return NAT or SIT bitmap */ 2556 if (flag == NAT_BITMAP) 2557 return le32_to_cpu(ckpt->nat_ver_bitmap_bytesize); 2558 else if (flag == SIT_BITMAP) 2559 return le32_to_cpu(ckpt->sit_ver_bitmap_bytesize); 2560 2561 return 0; 2562 } 2563 2564 static inline block_t __cp_payload(struct f2fs_sb_info *sbi) 2565 { 2566 return le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_payload); 2567 } 2568 2569 static inline void *__bitmap_ptr(struct f2fs_sb_info *sbi, int flag) 2570 { 2571 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi); 2572 void *tmp_ptr = &ckpt->sit_nat_version_bitmap; 2573 int offset; 2574 2575 if (is_set_ckpt_flags(sbi, CP_LARGE_NAT_BITMAP_FLAG)) { 2576 offset = (flag == SIT_BITMAP) ? 2577 le32_to_cpu(ckpt->nat_ver_bitmap_bytesize) : 0; 2578 /* 2579 * if large_nat_bitmap feature is enabled, leave checksum 2580 * protection for all nat/sit bitmaps. 2581 */ 2582 return tmp_ptr + offset + sizeof(__le32); 2583 } 2584 2585 if (__cp_payload(sbi) > 0) { 2586 if (flag == NAT_BITMAP) 2587 return tmp_ptr; 2588 else 2589 return (unsigned char *)ckpt + F2FS_BLKSIZE; 2590 } else { 2591 offset = (flag == NAT_BITMAP) ? 2592 le32_to_cpu(ckpt->sit_ver_bitmap_bytesize) : 0; 2593 return tmp_ptr + offset; 2594 } 2595 } 2596 2597 static inline block_t __start_cp_addr(struct f2fs_sb_info *sbi) 2598 { 2599 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr); 2600 2601 if (sbi->cur_cp_pack == 2) 2602 start_addr += BLKS_PER_SEG(sbi); 2603 return start_addr; 2604 } 2605 2606 static inline block_t __start_cp_next_addr(struct f2fs_sb_info *sbi) 2607 { 2608 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr); 2609 2610 if (sbi->cur_cp_pack == 1) 2611 start_addr += BLKS_PER_SEG(sbi); 2612 return start_addr; 2613 } 2614 2615 static inline void __set_cp_next_pack(struct f2fs_sb_info *sbi) 2616 { 2617 sbi->cur_cp_pack = (sbi->cur_cp_pack == 1) ? 2 : 1; 2618 } 2619 2620 static inline block_t __start_sum_addr(struct f2fs_sb_info *sbi) 2621 { 2622 return le32_to_cpu(F2FS_CKPT(sbi)->cp_pack_start_sum); 2623 } 2624 2625 extern void f2fs_mark_inode_dirty_sync(struct inode *inode, bool sync); 2626 static inline int inc_valid_node_count(struct f2fs_sb_info *sbi, 2627 struct inode *inode, bool is_inode) 2628 { 2629 block_t valid_block_count; 2630 unsigned int valid_node_count; 2631 unsigned int avail_user_block_count; 2632 int err; 2633 2634 if (is_inode) { 2635 if (inode) { 2636 err = dquot_alloc_inode(inode); 2637 if (err) 2638 return err; 2639 } 2640 } else { 2641 err = dquot_reserve_block(inode, 1); 2642 if (err) 2643 return err; 2644 } 2645 2646 if (time_to_inject(sbi, FAULT_BLOCK)) 2647 goto enospc; 2648 2649 spin_lock(&sbi->stat_lock); 2650 2651 valid_block_count = sbi->total_valid_block_count + 1; 2652 avail_user_block_count = get_available_block_count(sbi, inode, false); 2653 2654 if (unlikely(valid_block_count > avail_user_block_count)) { 2655 spin_unlock(&sbi->stat_lock); 2656 goto enospc; 2657 } 2658 2659 valid_node_count = sbi->total_valid_node_count + 1; 2660 if (unlikely(valid_node_count > sbi->total_node_count)) { 2661 spin_unlock(&sbi->stat_lock); 2662 goto enospc; 2663 } 2664 2665 sbi->total_valid_node_count++; 2666 sbi->total_valid_block_count++; 2667 spin_unlock(&sbi->stat_lock); 2668 2669 if (inode) { 2670 if (is_inode) 2671 f2fs_mark_inode_dirty_sync(inode, true); 2672 else 2673 f2fs_i_blocks_write(inode, 1, true, true); 2674 } 2675 2676 percpu_counter_inc(&sbi->alloc_valid_block_count); 2677 return 0; 2678 2679 enospc: 2680 if (is_inode) { 2681 if (inode) 2682 dquot_free_inode(inode); 2683 } else { 2684 dquot_release_reservation_block(inode, 1); 2685 } 2686 return -ENOSPC; 2687 } 2688 2689 static inline void dec_valid_node_count(struct f2fs_sb_info *sbi, 2690 struct inode *inode, bool is_inode) 2691 { 2692 spin_lock(&sbi->stat_lock); 2693 2694 if (unlikely(!sbi->total_valid_block_count || 2695 !sbi->total_valid_node_count)) { 2696 f2fs_warn(sbi, "dec_valid_node_count: inconsistent block counts, total_valid_block:%u, total_valid_node:%u", 2697 sbi->total_valid_block_count, 2698 sbi->total_valid_node_count); 2699 set_sbi_flag(sbi, SBI_NEED_FSCK); 2700 } else { 2701 sbi->total_valid_block_count--; 2702 sbi->total_valid_node_count--; 2703 } 2704 2705 if (sbi->reserved_blocks && 2706 sbi->current_reserved_blocks < sbi->reserved_blocks) 2707 sbi->current_reserved_blocks++; 2708 2709 spin_unlock(&sbi->stat_lock); 2710 2711 if (is_inode) { 2712 dquot_free_inode(inode); 2713 } else { 2714 if (unlikely(inode->i_blocks == 0)) { 2715 f2fs_warn(sbi, "dec_valid_node_count: inconsistent i_blocks, ino:%lu, iblocks:%llu", 2716 inode->i_ino, 2717 (unsigned long long)inode->i_blocks); 2718 set_sbi_flag(sbi, SBI_NEED_FSCK); 2719 return; 2720 } 2721 f2fs_i_blocks_write(inode, 1, false, true); 2722 } 2723 } 2724 2725 static inline unsigned int valid_node_count(struct f2fs_sb_info *sbi) 2726 { 2727 return sbi->total_valid_node_count; 2728 } 2729 2730 static inline void inc_valid_inode_count(struct f2fs_sb_info *sbi) 2731 { 2732 percpu_counter_inc(&sbi->total_valid_inode_count); 2733 } 2734 2735 static inline void dec_valid_inode_count(struct f2fs_sb_info *sbi) 2736 { 2737 percpu_counter_dec(&sbi->total_valid_inode_count); 2738 } 2739 2740 static inline s64 valid_inode_count(struct f2fs_sb_info *sbi) 2741 { 2742 return percpu_counter_sum_positive(&sbi->total_valid_inode_count); 2743 } 2744 2745 static inline struct page *f2fs_grab_cache_page(struct address_space *mapping, 2746 pgoff_t index, bool for_write) 2747 { 2748 struct page *page; 2749 unsigned int flags; 2750 2751 if (IS_ENABLED(CONFIG_F2FS_FAULT_INJECTION)) { 2752 if (!for_write) 2753 page = find_get_page_flags(mapping, index, 2754 FGP_LOCK | FGP_ACCESSED); 2755 else 2756 page = find_lock_page(mapping, index); 2757 if (page) 2758 return page; 2759 2760 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_ALLOC)) 2761 return NULL; 2762 } 2763 2764 if (!for_write) 2765 return grab_cache_page(mapping, index); 2766 2767 flags = memalloc_nofs_save(); 2768 page = grab_cache_page_write_begin(mapping, index); 2769 memalloc_nofs_restore(flags); 2770 2771 return page; 2772 } 2773 2774 static inline struct page *f2fs_pagecache_get_page( 2775 struct address_space *mapping, pgoff_t index, 2776 fgf_t fgp_flags, gfp_t gfp_mask) 2777 { 2778 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_GET)) 2779 return NULL; 2780 2781 return pagecache_get_page(mapping, index, fgp_flags, gfp_mask); 2782 } 2783 2784 static inline void f2fs_put_page(struct page *page, int unlock) 2785 { 2786 if (!page) 2787 return; 2788 2789 if (unlock) { 2790 f2fs_bug_on(F2FS_P_SB(page), !PageLocked(page)); 2791 unlock_page(page); 2792 } 2793 put_page(page); 2794 } 2795 2796 static inline void f2fs_put_dnode(struct dnode_of_data *dn) 2797 { 2798 if (dn->node_page) 2799 f2fs_put_page(dn->node_page, 1); 2800 if (dn->inode_page && dn->node_page != dn->inode_page) 2801 f2fs_put_page(dn->inode_page, 0); 2802 dn->node_page = NULL; 2803 dn->inode_page = NULL; 2804 } 2805 2806 static inline struct kmem_cache *f2fs_kmem_cache_create(const char *name, 2807 size_t size) 2808 { 2809 return kmem_cache_create(name, size, 0, SLAB_RECLAIM_ACCOUNT, NULL); 2810 } 2811 2812 static inline void *f2fs_kmem_cache_alloc_nofail(struct kmem_cache *cachep, 2813 gfp_t flags) 2814 { 2815 void *entry; 2816 2817 entry = kmem_cache_alloc(cachep, flags); 2818 if (!entry) 2819 entry = kmem_cache_alloc(cachep, flags | __GFP_NOFAIL); 2820 return entry; 2821 } 2822 2823 static inline void *f2fs_kmem_cache_alloc(struct kmem_cache *cachep, 2824 gfp_t flags, bool nofail, struct f2fs_sb_info *sbi) 2825 { 2826 if (nofail) 2827 return f2fs_kmem_cache_alloc_nofail(cachep, flags); 2828 2829 if (time_to_inject(sbi, FAULT_SLAB_ALLOC)) 2830 return NULL; 2831 2832 return kmem_cache_alloc(cachep, flags); 2833 } 2834 2835 static inline bool is_inflight_io(struct f2fs_sb_info *sbi, int type) 2836 { 2837 if (get_pages(sbi, F2FS_RD_DATA) || get_pages(sbi, F2FS_RD_NODE) || 2838 get_pages(sbi, F2FS_RD_META) || get_pages(sbi, F2FS_WB_DATA) || 2839 get_pages(sbi, F2FS_WB_CP_DATA) || 2840 get_pages(sbi, F2FS_DIO_READ) || 2841 get_pages(sbi, F2FS_DIO_WRITE)) 2842 return true; 2843 2844 if (type != DISCARD_TIME && SM_I(sbi) && SM_I(sbi)->dcc_info && 2845 atomic_read(&SM_I(sbi)->dcc_info->queued_discard)) 2846 return true; 2847 2848 if (SM_I(sbi) && SM_I(sbi)->fcc_info && 2849 atomic_read(&SM_I(sbi)->fcc_info->queued_flush)) 2850 return true; 2851 return false; 2852 } 2853 2854 static inline bool is_idle(struct f2fs_sb_info *sbi, int type) 2855 { 2856 if (sbi->gc_mode == GC_URGENT_HIGH) 2857 return true; 2858 2859 if (is_inflight_io(sbi, type)) 2860 return false; 2861 2862 if (sbi->gc_mode == GC_URGENT_MID) 2863 return true; 2864 2865 if (sbi->gc_mode == GC_URGENT_LOW && 2866 (type == DISCARD_TIME || type == GC_TIME)) 2867 return true; 2868 2869 return f2fs_time_over(sbi, type); 2870 } 2871 2872 static inline void f2fs_radix_tree_insert(struct radix_tree_root *root, 2873 unsigned long index, void *item) 2874 { 2875 while (radix_tree_insert(root, index, item)) 2876 cond_resched(); 2877 } 2878 2879 #define RAW_IS_INODE(p) ((p)->footer.nid == (p)->footer.ino) 2880 2881 static inline bool IS_INODE(struct page *page) 2882 { 2883 struct f2fs_node *p = F2FS_NODE(page); 2884 2885 return RAW_IS_INODE(p); 2886 } 2887 2888 static inline int offset_in_addr(struct f2fs_inode *i) 2889 { 2890 return (i->i_inline & F2FS_EXTRA_ATTR) ? 2891 (le16_to_cpu(i->i_extra_isize) / sizeof(__le32)) : 0; 2892 } 2893 2894 static inline __le32 *blkaddr_in_node(struct f2fs_node *node) 2895 { 2896 return RAW_IS_INODE(node) ? node->i.i_addr : node->dn.addr; 2897 } 2898 2899 static inline int f2fs_has_extra_attr(struct inode *inode); 2900 static inline block_t data_blkaddr(struct inode *inode, 2901 struct page *node_page, unsigned int offset) 2902 { 2903 struct f2fs_node *raw_node; 2904 __le32 *addr_array; 2905 int base = 0; 2906 bool is_inode = IS_INODE(node_page); 2907 2908 raw_node = F2FS_NODE(node_page); 2909 2910 if (is_inode) { 2911 if (!inode) 2912 /* from GC path only */ 2913 base = offset_in_addr(&raw_node->i); 2914 else if (f2fs_has_extra_attr(inode)) 2915 base = get_extra_isize(inode); 2916 } 2917 2918 addr_array = blkaddr_in_node(raw_node); 2919 return le32_to_cpu(addr_array[base + offset]); 2920 } 2921 2922 static inline block_t f2fs_data_blkaddr(struct dnode_of_data *dn) 2923 { 2924 return data_blkaddr(dn->inode, dn->node_page, dn->ofs_in_node); 2925 } 2926 2927 static inline int f2fs_test_bit(unsigned int nr, char *addr) 2928 { 2929 int mask; 2930 2931 addr += (nr >> 3); 2932 mask = BIT(7 - (nr & 0x07)); 2933 return mask & *addr; 2934 } 2935 2936 static inline void f2fs_set_bit(unsigned int nr, char *addr) 2937 { 2938 int mask; 2939 2940 addr += (nr >> 3); 2941 mask = BIT(7 - (nr & 0x07)); 2942 *addr |= mask; 2943 } 2944 2945 static inline void f2fs_clear_bit(unsigned int nr, char *addr) 2946 { 2947 int mask; 2948 2949 addr += (nr >> 3); 2950 mask = BIT(7 - (nr & 0x07)); 2951 *addr &= ~mask; 2952 } 2953 2954 static inline int f2fs_test_and_set_bit(unsigned int nr, char *addr) 2955 { 2956 int mask; 2957 int ret; 2958 2959 addr += (nr >> 3); 2960 mask = BIT(7 - (nr & 0x07)); 2961 ret = mask & *addr; 2962 *addr |= mask; 2963 return ret; 2964 } 2965 2966 static inline int f2fs_test_and_clear_bit(unsigned int nr, char *addr) 2967 { 2968 int mask; 2969 int ret; 2970 2971 addr += (nr >> 3); 2972 mask = BIT(7 - (nr & 0x07)); 2973 ret = mask & *addr; 2974 *addr &= ~mask; 2975 return ret; 2976 } 2977 2978 static inline void f2fs_change_bit(unsigned int nr, char *addr) 2979 { 2980 int mask; 2981 2982 addr += (nr >> 3); 2983 mask = BIT(7 - (nr & 0x07)); 2984 *addr ^= mask; 2985 } 2986 2987 /* 2988 * On-disk inode flags (f2fs_inode::i_flags) 2989 */ 2990 #define F2FS_COMPR_FL 0x00000004 /* Compress file */ 2991 #define F2FS_SYNC_FL 0x00000008 /* Synchronous updates */ 2992 #define F2FS_IMMUTABLE_FL 0x00000010 /* Immutable file */ 2993 #define F2FS_APPEND_FL 0x00000020 /* writes to file may only append */ 2994 #define F2FS_NODUMP_FL 0x00000040 /* do not dump file */ 2995 #define F2FS_NOATIME_FL 0x00000080 /* do not update atime */ 2996 #define F2FS_NOCOMP_FL 0x00000400 /* Don't compress */ 2997 #define F2FS_INDEX_FL 0x00001000 /* hash-indexed directory */ 2998 #define F2FS_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */ 2999 #define F2FS_PROJINHERIT_FL 0x20000000 /* Create with parents projid */ 3000 #define F2FS_CASEFOLD_FL 0x40000000 /* Casefolded file */ 3001 3002 #define F2FS_QUOTA_DEFAULT_FL (F2FS_NOATIME_FL | F2FS_IMMUTABLE_FL) 3003 3004 /* Flags that should be inherited by new inodes from their parent. */ 3005 #define F2FS_FL_INHERITED (F2FS_SYNC_FL | F2FS_NODUMP_FL | F2FS_NOATIME_FL | \ 3006 F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \ 3007 F2FS_CASEFOLD_FL) 3008 3009 /* Flags that are appropriate for regular files (all but dir-specific ones). */ 3010 #define F2FS_REG_FLMASK (~(F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \ 3011 F2FS_CASEFOLD_FL)) 3012 3013 /* Flags that are appropriate for non-directories/regular files. */ 3014 #define F2FS_OTHER_FLMASK (F2FS_NODUMP_FL | F2FS_NOATIME_FL) 3015 3016 static inline __u32 f2fs_mask_flags(umode_t mode, __u32 flags) 3017 { 3018 if (S_ISDIR(mode)) 3019 return flags; 3020 else if (S_ISREG(mode)) 3021 return flags & F2FS_REG_FLMASK; 3022 else 3023 return flags & F2FS_OTHER_FLMASK; 3024 } 3025 3026 static inline void __mark_inode_dirty_flag(struct inode *inode, 3027 int flag, bool set) 3028 { 3029 switch (flag) { 3030 case FI_INLINE_XATTR: 3031 case FI_INLINE_DATA: 3032 case FI_INLINE_DENTRY: 3033 case FI_NEW_INODE: 3034 if (set) 3035 return; 3036 fallthrough; 3037 case FI_DATA_EXIST: 3038 case FI_INLINE_DOTS: 3039 case FI_PIN_FILE: 3040 case FI_COMPRESS_RELEASED: 3041 case FI_ATOMIC_COMMITTED: 3042 f2fs_mark_inode_dirty_sync(inode, true); 3043 } 3044 } 3045 3046 static inline void set_inode_flag(struct inode *inode, int flag) 3047 { 3048 set_bit(flag, F2FS_I(inode)->flags); 3049 __mark_inode_dirty_flag(inode, flag, true); 3050 } 3051 3052 static inline int is_inode_flag_set(struct inode *inode, int flag) 3053 { 3054 return test_bit(flag, F2FS_I(inode)->flags); 3055 } 3056 3057 static inline void clear_inode_flag(struct inode *inode, int flag) 3058 { 3059 clear_bit(flag, F2FS_I(inode)->flags); 3060 __mark_inode_dirty_flag(inode, flag, false); 3061 } 3062 3063 static inline bool f2fs_verity_in_progress(struct inode *inode) 3064 { 3065 return IS_ENABLED(CONFIG_FS_VERITY) && 3066 is_inode_flag_set(inode, FI_VERITY_IN_PROGRESS); 3067 } 3068 3069 static inline void set_acl_inode(struct inode *inode, umode_t mode) 3070 { 3071 F2FS_I(inode)->i_acl_mode = mode; 3072 set_inode_flag(inode, FI_ACL_MODE); 3073 f2fs_mark_inode_dirty_sync(inode, false); 3074 } 3075 3076 static inline void f2fs_i_links_write(struct inode *inode, bool inc) 3077 { 3078 if (inc) 3079 inc_nlink(inode); 3080 else 3081 drop_nlink(inode); 3082 f2fs_mark_inode_dirty_sync(inode, true); 3083 } 3084 3085 static inline void f2fs_i_blocks_write(struct inode *inode, 3086 block_t diff, bool add, bool claim) 3087 { 3088 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE); 3089 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER); 3090 3091 /* add = 1, claim = 1 should be dquot_reserve_block in pair */ 3092 if (add) { 3093 if (claim) 3094 dquot_claim_block(inode, diff); 3095 else 3096 dquot_alloc_block_nofail(inode, diff); 3097 } else { 3098 dquot_free_block(inode, diff); 3099 } 3100 3101 f2fs_mark_inode_dirty_sync(inode, true); 3102 if (clean || recover) 3103 set_inode_flag(inode, FI_AUTO_RECOVER); 3104 } 3105 3106 static inline bool f2fs_is_atomic_file(struct inode *inode); 3107 3108 static inline void f2fs_i_size_write(struct inode *inode, loff_t i_size) 3109 { 3110 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE); 3111 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER); 3112 3113 if (i_size_read(inode) == i_size) 3114 return; 3115 3116 i_size_write(inode, i_size); 3117 3118 if (f2fs_is_atomic_file(inode)) 3119 return; 3120 3121 f2fs_mark_inode_dirty_sync(inode, true); 3122 if (clean || recover) 3123 set_inode_flag(inode, FI_AUTO_RECOVER); 3124 } 3125 3126 static inline void f2fs_i_depth_write(struct inode *inode, unsigned int depth) 3127 { 3128 F2FS_I(inode)->i_current_depth = depth; 3129 f2fs_mark_inode_dirty_sync(inode, true); 3130 } 3131 3132 static inline void f2fs_i_gc_failures_write(struct inode *inode, 3133 unsigned int count) 3134 { 3135 F2FS_I(inode)->i_gc_failures[GC_FAILURE_PIN] = count; 3136 f2fs_mark_inode_dirty_sync(inode, true); 3137 } 3138 3139 static inline void f2fs_i_xnid_write(struct inode *inode, nid_t xnid) 3140 { 3141 F2FS_I(inode)->i_xattr_nid = xnid; 3142 f2fs_mark_inode_dirty_sync(inode, true); 3143 } 3144 3145 static inline void f2fs_i_pino_write(struct inode *inode, nid_t pino) 3146 { 3147 F2FS_I(inode)->i_pino = pino; 3148 f2fs_mark_inode_dirty_sync(inode, true); 3149 } 3150 3151 static inline void get_inline_info(struct inode *inode, struct f2fs_inode *ri) 3152 { 3153 struct f2fs_inode_info *fi = F2FS_I(inode); 3154 3155 if (ri->i_inline & F2FS_INLINE_XATTR) 3156 set_bit(FI_INLINE_XATTR, fi->flags); 3157 if (ri->i_inline & F2FS_INLINE_DATA) 3158 set_bit(FI_INLINE_DATA, fi->flags); 3159 if (ri->i_inline & F2FS_INLINE_DENTRY) 3160 set_bit(FI_INLINE_DENTRY, fi->flags); 3161 if (ri->i_inline & F2FS_DATA_EXIST) 3162 set_bit(FI_DATA_EXIST, fi->flags); 3163 if (ri->i_inline & F2FS_INLINE_DOTS) 3164 set_bit(FI_INLINE_DOTS, fi->flags); 3165 if (ri->i_inline & F2FS_EXTRA_ATTR) 3166 set_bit(FI_EXTRA_ATTR, fi->flags); 3167 if (ri->i_inline & F2FS_PIN_FILE) 3168 set_bit(FI_PIN_FILE, fi->flags); 3169 if (ri->i_inline & F2FS_COMPRESS_RELEASED) 3170 set_bit(FI_COMPRESS_RELEASED, fi->flags); 3171 } 3172 3173 static inline void set_raw_inline(struct inode *inode, struct f2fs_inode *ri) 3174 { 3175 ri->i_inline = 0; 3176 3177 if (is_inode_flag_set(inode, FI_INLINE_XATTR)) 3178 ri->i_inline |= F2FS_INLINE_XATTR; 3179 if (is_inode_flag_set(inode, FI_INLINE_DATA)) 3180 ri->i_inline |= F2FS_INLINE_DATA; 3181 if (is_inode_flag_set(inode, FI_INLINE_DENTRY)) 3182 ri->i_inline |= F2FS_INLINE_DENTRY; 3183 if (is_inode_flag_set(inode, FI_DATA_EXIST)) 3184 ri->i_inline |= F2FS_DATA_EXIST; 3185 if (is_inode_flag_set(inode, FI_INLINE_DOTS)) 3186 ri->i_inline |= F2FS_INLINE_DOTS; 3187 if (is_inode_flag_set(inode, FI_EXTRA_ATTR)) 3188 ri->i_inline |= F2FS_EXTRA_ATTR; 3189 if (is_inode_flag_set(inode, FI_PIN_FILE)) 3190 ri->i_inline |= F2FS_PIN_FILE; 3191 if (is_inode_flag_set(inode, FI_COMPRESS_RELEASED)) 3192 ri->i_inline |= F2FS_COMPRESS_RELEASED; 3193 } 3194 3195 static inline int f2fs_has_extra_attr(struct inode *inode) 3196 { 3197 return is_inode_flag_set(inode, FI_EXTRA_ATTR); 3198 } 3199 3200 static inline int f2fs_has_inline_xattr(struct inode *inode) 3201 { 3202 return is_inode_flag_set(inode, FI_INLINE_XATTR); 3203 } 3204 3205 static inline int f2fs_compressed_file(struct inode *inode) 3206 { 3207 return S_ISREG(inode->i_mode) && 3208 is_inode_flag_set(inode, FI_COMPRESSED_FILE); 3209 } 3210 3211 static inline bool f2fs_need_compress_data(struct inode *inode) 3212 { 3213 int compress_mode = F2FS_OPTION(F2FS_I_SB(inode)).compress_mode; 3214 3215 if (!f2fs_compressed_file(inode)) 3216 return false; 3217 3218 if (compress_mode == COMPR_MODE_FS) 3219 return true; 3220 else if (compress_mode == COMPR_MODE_USER && 3221 is_inode_flag_set(inode, FI_ENABLE_COMPRESS)) 3222 return true; 3223 3224 return false; 3225 } 3226 3227 static inline unsigned int addrs_per_inode(struct inode *inode) 3228 { 3229 unsigned int addrs = CUR_ADDRS_PER_INODE(inode) - 3230 get_inline_xattr_addrs(inode); 3231 3232 if (!f2fs_compressed_file(inode)) 3233 return addrs; 3234 return ALIGN_DOWN(addrs, F2FS_I(inode)->i_cluster_size); 3235 } 3236 3237 static inline unsigned int addrs_per_block(struct inode *inode) 3238 { 3239 if (!f2fs_compressed_file(inode)) 3240 return DEF_ADDRS_PER_BLOCK; 3241 return ALIGN_DOWN(DEF_ADDRS_PER_BLOCK, F2FS_I(inode)->i_cluster_size); 3242 } 3243 3244 static inline void *inline_xattr_addr(struct inode *inode, struct page *page) 3245 { 3246 struct f2fs_inode *ri = F2FS_INODE(page); 3247 3248 return (void *)&(ri->i_addr[DEF_ADDRS_PER_INODE - 3249 get_inline_xattr_addrs(inode)]); 3250 } 3251 3252 static inline int inline_xattr_size(struct inode *inode) 3253 { 3254 if (f2fs_has_inline_xattr(inode)) 3255 return get_inline_xattr_addrs(inode) * sizeof(__le32); 3256 return 0; 3257 } 3258 3259 /* 3260 * Notice: check inline_data flag without inode page lock is unsafe. 3261 * It could change at any time by f2fs_convert_inline_page(). 3262 */ 3263 static inline int f2fs_has_inline_data(struct inode *inode) 3264 { 3265 return is_inode_flag_set(inode, FI_INLINE_DATA); 3266 } 3267 3268 static inline int f2fs_exist_data(struct inode *inode) 3269 { 3270 return is_inode_flag_set(inode, FI_DATA_EXIST); 3271 } 3272 3273 static inline int f2fs_has_inline_dots(struct inode *inode) 3274 { 3275 return is_inode_flag_set(inode, FI_INLINE_DOTS); 3276 } 3277 3278 static inline int f2fs_is_mmap_file(struct inode *inode) 3279 { 3280 return is_inode_flag_set(inode, FI_MMAP_FILE); 3281 } 3282 3283 static inline bool f2fs_is_pinned_file(struct inode *inode) 3284 { 3285 return is_inode_flag_set(inode, FI_PIN_FILE); 3286 } 3287 3288 static inline bool f2fs_is_atomic_file(struct inode *inode) 3289 { 3290 return is_inode_flag_set(inode, FI_ATOMIC_FILE); 3291 } 3292 3293 static inline bool f2fs_is_cow_file(struct inode *inode) 3294 { 3295 return is_inode_flag_set(inode, FI_COW_FILE); 3296 } 3297 3298 static inline __le32 *get_dnode_addr(struct inode *inode, 3299 struct page *node_page); 3300 static inline void *inline_data_addr(struct inode *inode, struct page *page) 3301 { 3302 __le32 *addr = get_dnode_addr(inode, page); 3303 3304 return (void *)(addr + DEF_INLINE_RESERVED_SIZE); 3305 } 3306 3307 static inline int f2fs_has_inline_dentry(struct inode *inode) 3308 { 3309 return is_inode_flag_set(inode, FI_INLINE_DENTRY); 3310 } 3311 3312 static inline int is_file(struct inode *inode, int type) 3313 { 3314 return F2FS_I(inode)->i_advise & type; 3315 } 3316 3317 static inline void set_file(struct inode *inode, int type) 3318 { 3319 if (is_file(inode, type)) 3320 return; 3321 F2FS_I(inode)->i_advise |= type; 3322 f2fs_mark_inode_dirty_sync(inode, true); 3323 } 3324 3325 static inline void clear_file(struct inode *inode, int type) 3326 { 3327 if (!is_file(inode, type)) 3328 return; 3329 F2FS_I(inode)->i_advise &= ~type; 3330 f2fs_mark_inode_dirty_sync(inode, true); 3331 } 3332 3333 static inline bool f2fs_is_time_consistent(struct inode *inode) 3334 { 3335 struct timespec64 ts = inode_get_atime(inode); 3336 3337 if (!timespec64_equal(F2FS_I(inode)->i_disk_time, &ts)) 3338 return false; 3339 ts = inode_get_ctime(inode); 3340 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 1, &ts)) 3341 return false; 3342 ts = inode_get_mtime(inode); 3343 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 2, &ts)) 3344 return false; 3345 return true; 3346 } 3347 3348 static inline bool f2fs_skip_inode_update(struct inode *inode, int dsync) 3349 { 3350 bool ret; 3351 3352 if (dsync) { 3353 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 3354 3355 spin_lock(&sbi->inode_lock[DIRTY_META]); 3356 ret = list_empty(&F2FS_I(inode)->gdirty_list); 3357 spin_unlock(&sbi->inode_lock[DIRTY_META]); 3358 return ret; 3359 } 3360 if (!is_inode_flag_set(inode, FI_AUTO_RECOVER) || 3361 file_keep_isize(inode) || 3362 i_size_read(inode) & ~PAGE_MASK) 3363 return false; 3364 3365 if (!f2fs_is_time_consistent(inode)) 3366 return false; 3367 3368 spin_lock(&F2FS_I(inode)->i_size_lock); 3369 ret = F2FS_I(inode)->last_disk_size == i_size_read(inode); 3370 spin_unlock(&F2FS_I(inode)->i_size_lock); 3371 3372 return ret; 3373 } 3374 3375 static inline bool f2fs_readonly(struct super_block *sb) 3376 { 3377 return sb_rdonly(sb); 3378 } 3379 3380 static inline bool f2fs_cp_error(struct f2fs_sb_info *sbi) 3381 { 3382 return is_set_ckpt_flags(sbi, CP_ERROR_FLAG); 3383 } 3384 3385 static inline void *f2fs_kmalloc(struct f2fs_sb_info *sbi, 3386 size_t size, gfp_t flags) 3387 { 3388 if (time_to_inject(sbi, FAULT_KMALLOC)) 3389 return NULL; 3390 3391 return kmalloc(size, flags); 3392 } 3393 3394 static inline void *f2fs_getname(struct f2fs_sb_info *sbi) 3395 { 3396 if (time_to_inject(sbi, FAULT_KMALLOC)) 3397 return NULL; 3398 3399 return __getname(); 3400 } 3401 3402 static inline void f2fs_putname(char *buf) 3403 { 3404 __putname(buf); 3405 } 3406 3407 static inline void *f2fs_kzalloc(struct f2fs_sb_info *sbi, 3408 size_t size, gfp_t flags) 3409 { 3410 return f2fs_kmalloc(sbi, size, flags | __GFP_ZERO); 3411 } 3412 3413 static inline void *f2fs_kvmalloc(struct f2fs_sb_info *sbi, 3414 size_t size, gfp_t flags) 3415 { 3416 if (time_to_inject(sbi, FAULT_KVMALLOC)) 3417 return NULL; 3418 3419 return kvmalloc(size, flags); 3420 } 3421 3422 static inline void *f2fs_kvzalloc(struct f2fs_sb_info *sbi, 3423 size_t size, gfp_t flags) 3424 { 3425 return f2fs_kvmalloc(sbi, size, flags | __GFP_ZERO); 3426 } 3427 3428 static inline int get_extra_isize(struct inode *inode) 3429 { 3430 return F2FS_I(inode)->i_extra_isize / sizeof(__le32); 3431 } 3432 3433 static inline int get_inline_xattr_addrs(struct inode *inode) 3434 { 3435 return F2FS_I(inode)->i_inline_xattr_size; 3436 } 3437 3438 static inline __le32 *get_dnode_addr(struct inode *inode, 3439 struct page *node_page) 3440 { 3441 int base = 0; 3442 3443 if (IS_INODE(node_page) && f2fs_has_extra_attr(inode)) 3444 base = get_extra_isize(inode); 3445 3446 return blkaddr_in_node(F2FS_NODE(node_page)) + base; 3447 } 3448 3449 #define f2fs_get_inode_mode(i) \ 3450 ((is_inode_flag_set(i, FI_ACL_MODE)) ? \ 3451 (F2FS_I(i)->i_acl_mode) : ((i)->i_mode)) 3452 3453 #define F2FS_MIN_EXTRA_ATTR_SIZE (sizeof(__le32)) 3454 3455 #define F2FS_TOTAL_EXTRA_ATTR_SIZE \ 3456 (offsetof(struct f2fs_inode, i_extra_end) - \ 3457 offsetof(struct f2fs_inode, i_extra_isize)) \ 3458 3459 #define F2FS_OLD_ATTRIBUTE_SIZE (offsetof(struct f2fs_inode, i_addr)) 3460 #define F2FS_FITS_IN_INODE(f2fs_inode, extra_isize, field) \ 3461 ((offsetof(typeof(*(f2fs_inode)), field) + \ 3462 sizeof((f2fs_inode)->field)) \ 3463 <= (F2FS_OLD_ATTRIBUTE_SIZE + (extra_isize))) \ 3464 3465 #define __is_large_section(sbi) (SEGS_PER_SEC(sbi) > 1) 3466 3467 #define __is_meta_io(fio) (PAGE_TYPE_OF_BIO((fio)->type) == META) 3468 3469 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi, 3470 block_t blkaddr, int type); 3471 static inline void verify_blkaddr(struct f2fs_sb_info *sbi, 3472 block_t blkaddr, int type) 3473 { 3474 if (!f2fs_is_valid_blkaddr(sbi, blkaddr, type)) 3475 f2fs_err(sbi, "invalid blkaddr: %u, type: %d, run fsck to fix.", 3476 blkaddr, type); 3477 } 3478 3479 static inline bool __is_valid_data_blkaddr(block_t blkaddr) 3480 { 3481 if (blkaddr == NEW_ADDR || blkaddr == NULL_ADDR || 3482 blkaddr == COMPRESS_ADDR) 3483 return false; 3484 return true; 3485 } 3486 3487 /* 3488 * file.c 3489 */ 3490 int f2fs_sync_file(struct file *file, loff_t start, loff_t end, int datasync); 3491 int f2fs_do_truncate_blocks(struct inode *inode, u64 from, bool lock); 3492 int f2fs_truncate_blocks(struct inode *inode, u64 from, bool lock); 3493 int f2fs_truncate(struct inode *inode); 3494 int f2fs_getattr(struct mnt_idmap *idmap, const struct path *path, 3495 struct kstat *stat, u32 request_mask, unsigned int flags); 3496 int f2fs_setattr(struct mnt_idmap *idmap, struct dentry *dentry, 3497 struct iattr *attr); 3498 int f2fs_truncate_hole(struct inode *inode, pgoff_t pg_start, pgoff_t pg_end); 3499 void f2fs_truncate_data_blocks_range(struct dnode_of_data *dn, int count); 3500 int f2fs_precache_extents(struct inode *inode); 3501 int f2fs_fileattr_get(struct dentry *dentry, struct fileattr *fa); 3502 int f2fs_fileattr_set(struct mnt_idmap *idmap, 3503 struct dentry *dentry, struct fileattr *fa); 3504 long f2fs_ioctl(struct file *filp, unsigned int cmd, unsigned long arg); 3505 long f2fs_compat_ioctl(struct file *file, unsigned int cmd, unsigned long arg); 3506 int f2fs_transfer_project_quota(struct inode *inode, kprojid_t kprojid); 3507 int f2fs_pin_file_control(struct inode *inode, bool inc); 3508 3509 /* 3510 * inode.c 3511 */ 3512 void f2fs_set_inode_flags(struct inode *inode); 3513 bool f2fs_inode_chksum_verify(struct f2fs_sb_info *sbi, struct page *page); 3514 void f2fs_inode_chksum_set(struct f2fs_sb_info *sbi, struct page *page); 3515 struct inode *f2fs_iget(struct super_block *sb, unsigned long ino); 3516 struct inode *f2fs_iget_retry(struct super_block *sb, unsigned long ino); 3517 int f2fs_try_to_free_nats(struct f2fs_sb_info *sbi, int nr_shrink); 3518 void f2fs_update_inode(struct inode *inode, struct page *node_page); 3519 void f2fs_update_inode_page(struct inode *inode); 3520 int f2fs_write_inode(struct inode *inode, struct writeback_control *wbc); 3521 void f2fs_evict_inode(struct inode *inode); 3522 void f2fs_handle_failed_inode(struct inode *inode); 3523 3524 /* 3525 * namei.c 3526 */ 3527 int f2fs_update_extension_list(struct f2fs_sb_info *sbi, const char *name, 3528 bool hot, bool set); 3529 struct dentry *f2fs_get_parent(struct dentry *child); 3530 int f2fs_get_tmpfile(struct mnt_idmap *idmap, struct inode *dir, 3531 struct inode **new_inode); 3532 3533 /* 3534 * dir.c 3535 */ 3536 int f2fs_init_casefolded_name(const struct inode *dir, 3537 struct f2fs_filename *fname); 3538 int f2fs_setup_filename(struct inode *dir, const struct qstr *iname, 3539 int lookup, struct f2fs_filename *fname); 3540 int f2fs_prepare_lookup(struct inode *dir, struct dentry *dentry, 3541 struct f2fs_filename *fname); 3542 void f2fs_free_filename(struct f2fs_filename *fname); 3543 struct f2fs_dir_entry *f2fs_find_target_dentry(const struct f2fs_dentry_ptr *d, 3544 const struct f2fs_filename *fname, int *max_slots); 3545 int f2fs_fill_dentries(struct dir_context *ctx, struct f2fs_dentry_ptr *d, 3546 unsigned int start_pos, struct fscrypt_str *fstr); 3547 void f2fs_do_make_empty_dir(struct inode *inode, struct inode *parent, 3548 struct f2fs_dentry_ptr *d); 3549 struct page *f2fs_init_inode_metadata(struct inode *inode, struct inode *dir, 3550 const struct f2fs_filename *fname, struct page *dpage); 3551 void f2fs_update_parent_metadata(struct inode *dir, struct inode *inode, 3552 unsigned int current_depth); 3553 int f2fs_room_for_filename(const void *bitmap, int slots, int max_slots); 3554 void f2fs_drop_nlink(struct inode *dir, struct inode *inode); 3555 struct f2fs_dir_entry *__f2fs_find_entry(struct inode *dir, 3556 const struct f2fs_filename *fname, 3557 struct page **res_page); 3558 struct f2fs_dir_entry *f2fs_find_entry(struct inode *dir, 3559 const struct qstr *child, struct page **res_page); 3560 struct f2fs_dir_entry *f2fs_parent_dir(struct inode *dir, struct page **p); 3561 ino_t f2fs_inode_by_name(struct inode *dir, const struct qstr *qstr, 3562 struct page **page); 3563 void f2fs_set_link(struct inode *dir, struct f2fs_dir_entry *de, 3564 struct page *page, struct inode *inode); 3565 bool f2fs_has_enough_room(struct inode *dir, struct page *ipage, 3566 const struct f2fs_filename *fname); 3567 void f2fs_update_dentry(nid_t ino, umode_t mode, struct f2fs_dentry_ptr *d, 3568 const struct fscrypt_str *name, f2fs_hash_t name_hash, 3569 unsigned int bit_pos); 3570 int f2fs_add_regular_entry(struct inode *dir, const struct f2fs_filename *fname, 3571 struct inode *inode, nid_t ino, umode_t mode); 3572 int f2fs_add_dentry(struct inode *dir, const struct f2fs_filename *fname, 3573 struct inode *inode, nid_t ino, umode_t mode); 3574 int f2fs_do_add_link(struct inode *dir, const struct qstr *name, 3575 struct inode *inode, nid_t ino, umode_t mode); 3576 void f2fs_delete_entry(struct f2fs_dir_entry *dentry, struct page *page, 3577 struct inode *dir, struct inode *inode); 3578 int f2fs_do_tmpfile(struct inode *inode, struct inode *dir, 3579 struct f2fs_filename *fname); 3580 bool f2fs_empty_dir(struct inode *dir); 3581 3582 static inline int f2fs_add_link(struct dentry *dentry, struct inode *inode) 3583 { 3584 if (fscrypt_is_nokey_name(dentry)) 3585 return -ENOKEY; 3586 return f2fs_do_add_link(d_inode(dentry->d_parent), &dentry->d_name, 3587 inode, inode->i_ino, inode->i_mode); 3588 } 3589 3590 /* 3591 * super.c 3592 */ 3593 int f2fs_inode_dirtied(struct inode *inode, bool sync); 3594 void f2fs_inode_synced(struct inode *inode); 3595 int f2fs_dquot_initialize(struct inode *inode); 3596 int f2fs_enable_quota_files(struct f2fs_sb_info *sbi, bool rdonly); 3597 int f2fs_quota_sync(struct super_block *sb, int type); 3598 loff_t max_file_blocks(struct inode *inode); 3599 void f2fs_quota_off_umount(struct super_block *sb); 3600 void f2fs_save_errors(struct f2fs_sb_info *sbi, unsigned char flag); 3601 void f2fs_handle_critical_error(struct f2fs_sb_info *sbi, unsigned char reason, 3602 bool irq_context); 3603 void f2fs_handle_error(struct f2fs_sb_info *sbi, unsigned char error); 3604 void f2fs_handle_error_async(struct f2fs_sb_info *sbi, unsigned char error); 3605 int f2fs_commit_super(struct f2fs_sb_info *sbi, bool recover); 3606 int f2fs_sync_fs(struct super_block *sb, int sync); 3607 int f2fs_sanity_check_ckpt(struct f2fs_sb_info *sbi); 3608 3609 /* 3610 * hash.c 3611 */ 3612 void f2fs_hash_filename(const struct inode *dir, struct f2fs_filename *fname); 3613 3614 /* 3615 * node.c 3616 */ 3617 struct node_info; 3618 3619 int f2fs_check_nid_range(struct f2fs_sb_info *sbi, nid_t nid); 3620 bool f2fs_available_free_memory(struct f2fs_sb_info *sbi, int type); 3621 bool f2fs_in_warm_node_list(struct f2fs_sb_info *sbi, struct page *page); 3622 void f2fs_init_fsync_node_info(struct f2fs_sb_info *sbi); 3623 void f2fs_del_fsync_node_entry(struct f2fs_sb_info *sbi, struct page *page); 3624 void f2fs_reset_fsync_node_info(struct f2fs_sb_info *sbi); 3625 int f2fs_need_dentry_mark(struct f2fs_sb_info *sbi, nid_t nid); 3626 bool f2fs_is_checkpointed_node(struct f2fs_sb_info *sbi, nid_t nid); 3627 bool f2fs_need_inode_block_update(struct f2fs_sb_info *sbi, nid_t ino); 3628 int f2fs_get_node_info(struct f2fs_sb_info *sbi, nid_t nid, 3629 struct node_info *ni, bool checkpoint_context); 3630 pgoff_t f2fs_get_next_page_offset(struct dnode_of_data *dn, pgoff_t pgofs); 3631 int f2fs_get_dnode_of_data(struct dnode_of_data *dn, pgoff_t index, int mode); 3632 int f2fs_truncate_inode_blocks(struct inode *inode, pgoff_t from); 3633 int f2fs_truncate_xattr_node(struct inode *inode); 3634 int f2fs_wait_on_node_pages_writeback(struct f2fs_sb_info *sbi, 3635 unsigned int seq_id); 3636 bool f2fs_nat_bitmap_enabled(struct f2fs_sb_info *sbi); 3637 int f2fs_remove_inode_page(struct inode *inode); 3638 struct page *f2fs_new_inode_page(struct inode *inode); 3639 struct page *f2fs_new_node_page(struct dnode_of_data *dn, unsigned int ofs); 3640 void f2fs_ra_node_page(struct f2fs_sb_info *sbi, nid_t nid); 3641 struct page *f2fs_get_node_page(struct f2fs_sb_info *sbi, pgoff_t nid); 3642 struct page *f2fs_get_node_page_ra(struct page *parent, int start); 3643 int f2fs_move_node_page(struct page *node_page, int gc_type); 3644 void f2fs_flush_inline_data(struct f2fs_sb_info *sbi); 3645 int f2fs_fsync_node_pages(struct f2fs_sb_info *sbi, struct inode *inode, 3646 struct writeback_control *wbc, bool atomic, 3647 unsigned int *seq_id); 3648 int f2fs_sync_node_pages(struct f2fs_sb_info *sbi, 3649 struct writeback_control *wbc, 3650 bool do_balance, enum iostat_type io_type); 3651 int f2fs_build_free_nids(struct f2fs_sb_info *sbi, bool sync, bool mount); 3652 bool f2fs_alloc_nid(struct f2fs_sb_info *sbi, nid_t *nid); 3653 void f2fs_alloc_nid_done(struct f2fs_sb_info *sbi, nid_t nid); 3654 void f2fs_alloc_nid_failed(struct f2fs_sb_info *sbi, nid_t nid); 3655 int f2fs_try_to_free_nids(struct f2fs_sb_info *sbi, int nr_shrink); 3656 int f2fs_recover_inline_xattr(struct inode *inode, struct page *page); 3657 int f2fs_recover_xattr_data(struct inode *inode, struct page *page); 3658 int f2fs_recover_inode_page(struct f2fs_sb_info *sbi, struct page *page); 3659 int f2fs_restore_node_summary(struct f2fs_sb_info *sbi, 3660 unsigned int segno, struct f2fs_summary_block *sum); 3661 void f2fs_enable_nat_bits(struct f2fs_sb_info *sbi); 3662 int f2fs_flush_nat_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3663 int f2fs_build_node_manager(struct f2fs_sb_info *sbi); 3664 void f2fs_destroy_node_manager(struct f2fs_sb_info *sbi); 3665 int __init f2fs_create_node_manager_caches(void); 3666 void f2fs_destroy_node_manager_caches(void); 3667 3668 /* 3669 * segment.c 3670 */ 3671 bool f2fs_need_SSR(struct f2fs_sb_info *sbi); 3672 int f2fs_commit_atomic_write(struct inode *inode); 3673 void f2fs_abort_atomic_write(struct inode *inode, bool clean); 3674 void f2fs_balance_fs(struct f2fs_sb_info *sbi, bool need); 3675 void f2fs_balance_fs_bg(struct f2fs_sb_info *sbi, bool from_bg); 3676 int f2fs_issue_flush(struct f2fs_sb_info *sbi, nid_t ino); 3677 int f2fs_create_flush_cmd_control(struct f2fs_sb_info *sbi); 3678 int f2fs_flush_device_cache(struct f2fs_sb_info *sbi); 3679 void f2fs_destroy_flush_cmd_control(struct f2fs_sb_info *sbi, bool free); 3680 void f2fs_invalidate_blocks(struct f2fs_sb_info *sbi, block_t addr); 3681 bool f2fs_is_checkpointed_data(struct f2fs_sb_info *sbi, block_t blkaddr); 3682 int f2fs_start_discard_thread(struct f2fs_sb_info *sbi); 3683 void f2fs_drop_discard_cmd(struct f2fs_sb_info *sbi); 3684 void f2fs_stop_discard_thread(struct f2fs_sb_info *sbi); 3685 bool f2fs_issue_discard_timeout(struct f2fs_sb_info *sbi); 3686 void f2fs_clear_prefree_segments(struct f2fs_sb_info *sbi, 3687 struct cp_control *cpc); 3688 void f2fs_dirty_to_prefree(struct f2fs_sb_info *sbi); 3689 block_t f2fs_get_unusable_blocks(struct f2fs_sb_info *sbi); 3690 int f2fs_disable_cp_again(struct f2fs_sb_info *sbi, block_t unusable); 3691 void f2fs_release_discard_addrs(struct f2fs_sb_info *sbi); 3692 int f2fs_npages_for_summary_flush(struct f2fs_sb_info *sbi, bool for_ra); 3693 bool f2fs_segment_has_free_slot(struct f2fs_sb_info *sbi, int segno); 3694 int f2fs_init_inmem_curseg(struct f2fs_sb_info *sbi); 3695 void f2fs_save_inmem_curseg(struct f2fs_sb_info *sbi); 3696 void f2fs_restore_inmem_curseg(struct f2fs_sb_info *sbi); 3697 int f2fs_allocate_segment_for_resize(struct f2fs_sb_info *sbi, int type, 3698 unsigned int start, unsigned int end); 3699 int f2fs_allocate_new_section(struct f2fs_sb_info *sbi, int type, bool force); 3700 int f2fs_allocate_pinning_section(struct f2fs_sb_info *sbi); 3701 int f2fs_allocate_new_segments(struct f2fs_sb_info *sbi); 3702 int f2fs_trim_fs(struct f2fs_sb_info *sbi, struct fstrim_range *range); 3703 bool f2fs_exist_trim_candidates(struct f2fs_sb_info *sbi, 3704 struct cp_control *cpc); 3705 struct page *f2fs_get_sum_page(struct f2fs_sb_info *sbi, unsigned int segno); 3706 void f2fs_update_meta_page(struct f2fs_sb_info *sbi, void *src, 3707 block_t blk_addr); 3708 void f2fs_do_write_meta_page(struct f2fs_sb_info *sbi, struct page *page, 3709 enum iostat_type io_type); 3710 void f2fs_do_write_node_page(unsigned int nid, struct f2fs_io_info *fio); 3711 void f2fs_outplace_write_data(struct dnode_of_data *dn, 3712 struct f2fs_io_info *fio); 3713 int f2fs_inplace_write_data(struct f2fs_io_info *fio); 3714 void f2fs_do_replace_block(struct f2fs_sb_info *sbi, struct f2fs_summary *sum, 3715 block_t old_blkaddr, block_t new_blkaddr, 3716 bool recover_curseg, bool recover_newaddr, 3717 bool from_gc); 3718 void f2fs_replace_block(struct f2fs_sb_info *sbi, struct dnode_of_data *dn, 3719 block_t old_addr, block_t new_addr, 3720 unsigned char version, bool recover_curseg, 3721 bool recover_newaddr); 3722 int f2fs_allocate_data_block(struct f2fs_sb_info *sbi, struct page *page, 3723 block_t old_blkaddr, block_t *new_blkaddr, 3724 struct f2fs_summary *sum, int type, 3725 struct f2fs_io_info *fio); 3726 void f2fs_update_device_state(struct f2fs_sb_info *sbi, nid_t ino, 3727 block_t blkaddr, unsigned int blkcnt); 3728 void f2fs_wait_on_page_writeback(struct page *page, 3729 enum page_type type, bool ordered, bool locked); 3730 void f2fs_wait_on_block_writeback(struct inode *inode, block_t blkaddr); 3731 void f2fs_wait_on_block_writeback_range(struct inode *inode, block_t blkaddr, 3732 block_t len); 3733 void f2fs_write_data_summaries(struct f2fs_sb_info *sbi, block_t start_blk); 3734 void f2fs_write_node_summaries(struct f2fs_sb_info *sbi, block_t start_blk); 3735 int f2fs_lookup_journal_in_cursum(struct f2fs_journal *journal, int type, 3736 unsigned int val, int alloc); 3737 void f2fs_flush_sit_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3738 int f2fs_fix_curseg_write_pointer(struct f2fs_sb_info *sbi); 3739 int f2fs_check_write_pointer(struct f2fs_sb_info *sbi); 3740 int f2fs_build_segment_manager(struct f2fs_sb_info *sbi); 3741 void f2fs_destroy_segment_manager(struct f2fs_sb_info *sbi); 3742 int __init f2fs_create_segment_manager_caches(void); 3743 void f2fs_destroy_segment_manager_caches(void); 3744 int f2fs_rw_hint_to_seg_type(enum rw_hint hint); 3745 unsigned int f2fs_usable_segs_in_sec(struct f2fs_sb_info *sbi, 3746 unsigned int segno); 3747 unsigned int f2fs_usable_blks_in_seg(struct f2fs_sb_info *sbi, 3748 unsigned int segno); 3749 3750 #define DEF_FRAGMENT_SIZE 4 3751 #define MIN_FRAGMENT_SIZE 1 3752 #define MAX_FRAGMENT_SIZE 512 3753 3754 static inline bool f2fs_need_rand_seg(struct f2fs_sb_info *sbi) 3755 { 3756 return F2FS_OPTION(sbi).fs_mode == FS_MODE_FRAGMENT_SEG || 3757 F2FS_OPTION(sbi).fs_mode == FS_MODE_FRAGMENT_BLK; 3758 } 3759 3760 /* 3761 * checkpoint.c 3762 */ 3763 void f2fs_stop_checkpoint(struct f2fs_sb_info *sbi, bool end_io, 3764 unsigned char reason); 3765 void f2fs_flush_ckpt_thread(struct f2fs_sb_info *sbi); 3766 struct page *f2fs_grab_meta_page(struct f2fs_sb_info *sbi, pgoff_t index); 3767 struct page *f2fs_get_meta_page(struct f2fs_sb_info *sbi, pgoff_t index); 3768 struct page *f2fs_get_meta_page_retry(struct f2fs_sb_info *sbi, pgoff_t index); 3769 struct page *f2fs_get_tmp_page(struct f2fs_sb_info *sbi, pgoff_t index); 3770 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi, 3771 block_t blkaddr, int type); 3772 bool f2fs_is_valid_blkaddr_raw(struct f2fs_sb_info *sbi, 3773 block_t blkaddr, int type); 3774 int f2fs_ra_meta_pages(struct f2fs_sb_info *sbi, block_t start, int nrpages, 3775 int type, bool sync); 3776 void f2fs_ra_meta_pages_cond(struct f2fs_sb_info *sbi, pgoff_t index, 3777 unsigned int ra_blocks); 3778 long f2fs_sync_meta_pages(struct f2fs_sb_info *sbi, enum page_type type, 3779 long nr_to_write, enum iostat_type io_type); 3780 void f2fs_add_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type); 3781 void f2fs_remove_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type); 3782 void f2fs_release_ino_entry(struct f2fs_sb_info *sbi, bool all); 3783 bool f2fs_exist_written_data(struct f2fs_sb_info *sbi, nid_t ino, int mode); 3784 void f2fs_set_dirty_device(struct f2fs_sb_info *sbi, nid_t ino, 3785 unsigned int devidx, int type); 3786 bool f2fs_is_dirty_device(struct f2fs_sb_info *sbi, nid_t ino, 3787 unsigned int devidx, int type); 3788 int f2fs_acquire_orphan_inode(struct f2fs_sb_info *sbi); 3789 void f2fs_release_orphan_inode(struct f2fs_sb_info *sbi); 3790 void f2fs_add_orphan_inode(struct inode *inode); 3791 void f2fs_remove_orphan_inode(struct f2fs_sb_info *sbi, nid_t ino); 3792 int f2fs_recover_orphan_inodes(struct f2fs_sb_info *sbi); 3793 int f2fs_get_valid_checkpoint(struct f2fs_sb_info *sbi); 3794 void f2fs_update_dirty_folio(struct inode *inode, struct folio *folio); 3795 void f2fs_remove_dirty_inode(struct inode *inode); 3796 int f2fs_sync_dirty_inodes(struct f2fs_sb_info *sbi, enum inode_type type, 3797 bool from_cp); 3798 void f2fs_wait_on_all_pages(struct f2fs_sb_info *sbi, int type); 3799 u64 f2fs_get_sectors_written(struct f2fs_sb_info *sbi); 3800 int f2fs_write_checkpoint(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3801 void f2fs_init_ino_entry_info(struct f2fs_sb_info *sbi); 3802 int __init f2fs_create_checkpoint_caches(void); 3803 void f2fs_destroy_checkpoint_caches(void); 3804 int f2fs_issue_checkpoint(struct f2fs_sb_info *sbi); 3805 int f2fs_start_ckpt_thread(struct f2fs_sb_info *sbi); 3806 void f2fs_stop_ckpt_thread(struct f2fs_sb_info *sbi); 3807 void f2fs_init_ckpt_req_control(struct f2fs_sb_info *sbi); 3808 3809 /* 3810 * data.c 3811 */ 3812 int __init f2fs_init_bioset(void); 3813 void f2fs_destroy_bioset(void); 3814 bool f2fs_is_cp_guaranteed(struct page *page); 3815 int f2fs_init_bio_entry_cache(void); 3816 void f2fs_destroy_bio_entry_cache(void); 3817 void f2fs_submit_read_bio(struct f2fs_sb_info *sbi, struct bio *bio, 3818 enum page_type type); 3819 int f2fs_init_write_merge_io(struct f2fs_sb_info *sbi); 3820 void f2fs_submit_merged_write(struct f2fs_sb_info *sbi, enum page_type type); 3821 void f2fs_submit_merged_write_cond(struct f2fs_sb_info *sbi, 3822 struct inode *inode, struct page *page, 3823 nid_t ino, enum page_type type); 3824 void f2fs_submit_merged_ipu_write(struct f2fs_sb_info *sbi, 3825 struct bio **bio, struct page *page); 3826 void f2fs_flush_merged_writes(struct f2fs_sb_info *sbi); 3827 int f2fs_submit_page_bio(struct f2fs_io_info *fio); 3828 int f2fs_merge_page_bio(struct f2fs_io_info *fio); 3829 void f2fs_submit_page_write(struct f2fs_io_info *fio); 3830 struct block_device *f2fs_target_device(struct f2fs_sb_info *sbi, 3831 block_t blk_addr, sector_t *sector); 3832 int f2fs_target_device_index(struct f2fs_sb_info *sbi, block_t blkaddr); 3833 void f2fs_set_data_blkaddr(struct dnode_of_data *dn, block_t blkaddr); 3834 void f2fs_update_data_blkaddr(struct dnode_of_data *dn, block_t blkaddr); 3835 int f2fs_reserve_new_blocks(struct dnode_of_data *dn, blkcnt_t count); 3836 int f2fs_reserve_new_block(struct dnode_of_data *dn); 3837 int f2fs_get_block_locked(struct dnode_of_data *dn, pgoff_t index); 3838 int f2fs_reserve_block(struct dnode_of_data *dn, pgoff_t index); 3839 struct page *f2fs_get_read_data_page(struct inode *inode, pgoff_t index, 3840 blk_opf_t op_flags, bool for_write, pgoff_t *next_pgofs); 3841 struct page *f2fs_find_data_page(struct inode *inode, pgoff_t index, 3842 pgoff_t *next_pgofs); 3843 struct page *f2fs_get_lock_data_page(struct inode *inode, pgoff_t index, 3844 bool for_write); 3845 struct page *f2fs_get_new_data_page(struct inode *inode, 3846 struct page *ipage, pgoff_t index, bool new_i_size); 3847 int f2fs_do_write_data_page(struct f2fs_io_info *fio); 3848 int f2fs_map_blocks(struct inode *inode, struct f2fs_map_blocks *map, int flag); 3849 int f2fs_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, 3850 u64 start, u64 len); 3851 int f2fs_encrypt_one_page(struct f2fs_io_info *fio); 3852 bool f2fs_should_update_inplace(struct inode *inode, struct f2fs_io_info *fio); 3853 bool f2fs_should_update_outplace(struct inode *inode, struct f2fs_io_info *fio); 3854 int f2fs_write_single_data_page(struct page *page, int *submitted, 3855 struct bio **bio, sector_t *last_block, 3856 struct writeback_control *wbc, 3857 enum iostat_type io_type, 3858 int compr_blocks, bool allow_balance); 3859 void f2fs_write_failed(struct inode *inode, loff_t to); 3860 void f2fs_invalidate_folio(struct folio *folio, size_t offset, size_t length); 3861 bool f2fs_release_folio(struct folio *folio, gfp_t wait); 3862 bool f2fs_overwrite_io(struct inode *inode, loff_t pos, size_t len); 3863 void f2fs_clear_page_cache_dirty_tag(struct page *page); 3864 int f2fs_init_post_read_processing(void); 3865 void f2fs_destroy_post_read_processing(void); 3866 int f2fs_init_post_read_wq(struct f2fs_sb_info *sbi); 3867 void f2fs_destroy_post_read_wq(struct f2fs_sb_info *sbi); 3868 extern const struct iomap_ops f2fs_iomap_ops; 3869 3870 /* 3871 * gc.c 3872 */ 3873 int f2fs_start_gc_thread(struct f2fs_sb_info *sbi); 3874 void f2fs_stop_gc_thread(struct f2fs_sb_info *sbi); 3875 block_t f2fs_start_bidx_of_node(unsigned int node_ofs, struct inode *inode); 3876 int f2fs_gc(struct f2fs_sb_info *sbi, struct f2fs_gc_control *gc_control); 3877 void f2fs_build_gc_manager(struct f2fs_sb_info *sbi); 3878 int f2fs_gc_range(struct f2fs_sb_info *sbi, 3879 unsigned int start_seg, unsigned int end_seg, 3880 bool dry_run, unsigned int dry_run_sections); 3881 int f2fs_resize_fs(struct file *filp, __u64 block_count); 3882 int __init f2fs_create_garbage_collection_cache(void); 3883 void f2fs_destroy_garbage_collection_cache(void); 3884 /* victim selection function for cleaning and SSR */ 3885 int f2fs_get_victim(struct f2fs_sb_info *sbi, unsigned int *result, 3886 int gc_type, int type, char alloc_mode, 3887 unsigned long long age); 3888 3889 /* 3890 * recovery.c 3891 */ 3892 int f2fs_recover_fsync_data(struct f2fs_sb_info *sbi, bool check_only); 3893 bool f2fs_space_for_roll_forward(struct f2fs_sb_info *sbi); 3894 int __init f2fs_create_recovery_cache(void); 3895 void f2fs_destroy_recovery_cache(void); 3896 3897 /* 3898 * debug.c 3899 */ 3900 #ifdef CONFIG_F2FS_STAT_FS 3901 struct f2fs_stat_info { 3902 struct list_head stat_list; 3903 struct f2fs_sb_info *sbi; 3904 int all_area_segs, sit_area_segs, nat_area_segs, ssa_area_segs; 3905 int main_area_segs, main_area_sections, main_area_zones; 3906 unsigned long long hit_cached[NR_EXTENT_CACHES]; 3907 unsigned long long hit_rbtree[NR_EXTENT_CACHES]; 3908 unsigned long long total_ext[NR_EXTENT_CACHES]; 3909 unsigned long long hit_total[NR_EXTENT_CACHES]; 3910 int ext_tree[NR_EXTENT_CACHES]; 3911 int zombie_tree[NR_EXTENT_CACHES]; 3912 int ext_node[NR_EXTENT_CACHES]; 3913 /* to count memory footprint */ 3914 unsigned long long ext_mem[NR_EXTENT_CACHES]; 3915 /* for read extent cache */ 3916 unsigned long long hit_largest; 3917 /* for block age extent cache */ 3918 unsigned long long allocated_data_blocks; 3919 int ndirty_node, ndirty_dent, ndirty_meta, ndirty_imeta; 3920 int ndirty_data, ndirty_qdata; 3921 unsigned int ndirty_dirs, ndirty_files, nquota_files, ndirty_all; 3922 int nats, dirty_nats, sits, dirty_sits; 3923 int free_nids, avail_nids, alloc_nids; 3924 int total_count, utilization; 3925 int nr_wb_cp_data, nr_wb_data; 3926 int nr_rd_data, nr_rd_node, nr_rd_meta; 3927 int nr_dio_read, nr_dio_write; 3928 unsigned int io_skip_bggc, other_skip_bggc; 3929 int nr_flushing, nr_flushed, flush_list_empty; 3930 int nr_discarding, nr_discarded; 3931 int nr_discard_cmd; 3932 unsigned int undiscard_blks; 3933 int nr_issued_ckpt, nr_total_ckpt, nr_queued_ckpt; 3934 unsigned int cur_ckpt_time, peak_ckpt_time; 3935 int inline_xattr, inline_inode, inline_dir, append, update, orphans; 3936 int compr_inode, swapfile_inode; 3937 unsigned long long compr_blocks; 3938 int aw_cnt, max_aw_cnt; 3939 unsigned int valid_count, valid_node_count, valid_inode_count, discard_blks; 3940 unsigned int bimodal, avg_vblocks; 3941 int util_free, util_valid, util_invalid; 3942 int rsvd_segs, overp_segs; 3943 int dirty_count, node_pages, meta_pages, compress_pages; 3944 int compress_page_hit; 3945 int prefree_count, free_segs, free_secs; 3946 int cp_call_count[MAX_CALL_TYPE], cp_count; 3947 int gc_call_count[MAX_CALL_TYPE]; 3948 int gc_segs[2][2]; 3949 int gc_secs[2][2]; 3950 int tot_blks, data_blks, node_blks; 3951 int bg_data_blks, bg_node_blks; 3952 int curseg[NR_CURSEG_TYPE]; 3953 int cursec[NR_CURSEG_TYPE]; 3954 int curzone[NR_CURSEG_TYPE]; 3955 unsigned int dirty_seg[NR_CURSEG_TYPE]; 3956 unsigned int full_seg[NR_CURSEG_TYPE]; 3957 unsigned int valid_blks[NR_CURSEG_TYPE]; 3958 3959 unsigned int meta_count[META_MAX]; 3960 unsigned int segment_count[2]; 3961 unsigned int block_count[2]; 3962 unsigned int inplace_count; 3963 unsigned long long base_mem, cache_mem, page_mem; 3964 }; 3965 3966 static inline struct f2fs_stat_info *F2FS_STAT(struct f2fs_sb_info *sbi) 3967 { 3968 return (struct f2fs_stat_info *)sbi->stat_info; 3969 } 3970 3971 #define stat_inc_cp_call_count(sbi, foreground) \ 3972 atomic_inc(&sbi->cp_call_count[(foreground)]) 3973 #define stat_inc_cp_count(si) (F2FS_STAT(sbi)->cp_count++) 3974 #define stat_io_skip_bggc_count(sbi) ((sbi)->io_skip_bggc++) 3975 #define stat_other_skip_bggc_count(sbi) ((sbi)->other_skip_bggc++) 3976 #define stat_inc_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]++) 3977 #define stat_dec_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]--) 3978 #define stat_inc_total_hit(sbi, type) (atomic64_inc(&(sbi)->total_hit_ext[type])) 3979 #define stat_inc_rbtree_node_hit(sbi, type) (atomic64_inc(&(sbi)->read_hit_rbtree[type])) 3980 #define stat_inc_largest_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_largest)) 3981 #define stat_inc_cached_node_hit(sbi, type) (atomic64_inc(&(sbi)->read_hit_cached[type])) 3982 #define stat_inc_inline_xattr(inode) \ 3983 do { \ 3984 if (f2fs_has_inline_xattr(inode)) \ 3985 (atomic_inc(&F2FS_I_SB(inode)->inline_xattr)); \ 3986 } while (0) 3987 #define stat_dec_inline_xattr(inode) \ 3988 do { \ 3989 if (f2fs_has_inline_xattr(inode)) \ 3990 (atomic_dec(&F2FS_I_SB(inode)->inline_xattr)); \ 3991 } while (0) 3992 #define stat_inc_inline_inode(inode) \ 3993 do { \ 3994 if (f2fs_has_inline_data(inode)) \ 3995 (atomic_inc(&F2FS_I_SB(inode)->inline_inode)); \ 3996 } while (0) 3997 #define stat_dec_inline_inode(inode) \ 3998 do { \ 3999 if (f2fs_has_inline_data(inode)) \ 4000 (atomic_dec(&F2FS_I_SB(inode)->inline_inode)); \ 4001 } while (0) 4002 #define stat_inc_inline_dir(inode) \ 4003 do { \ 4004 if (f2fs_has_inline_dentry(inode)) \ 4005 (atomic_inc(&F2FS_I_SB(inode)->inline_dir)); \ 4006 } while (0) 4007 #define stat_dec_inline_dir(inode) \ 4008 do { \ 4009 if (f2fs_has_inline_dentry(inode)) \ 4010 (atomic_dec(&F2FS_I_SB(inode)->inline_dir)); \ 4011 } while (0) 4012 #define stat_inc_compr_inode(inode) \ 4013 do { \ 4014 if (f2fs_compressed_file(inode)) \ 4015 (atomic_inc(&F2FS_I_SB(inode)->compr_inode)); \ 4016 } while (0) 4017 #define stat_dec_compr_inode(inode) \ 4018 do { \ 4019 if (f2fs_compressed_file(inode)) \ 4020 (atomic_dec(&F2FS_I_SB(inode)->compr_inode)); \ 4021 } while (0) 4022 #define stat_add_compr_blocks(inode, blocks) \ 4023 (atomic64_add(blocks, &F2FS_I_SB(inode)->compr_blocks)) 4024 #define stat_sub_compr_blocks(inode, blocks) \ 4025 (atomic64_sub(blocks, &F2FS_I_SB(inode)->compr_blocks)) 4026 #define stat_inc_swapfile_inode(inode) \ 4027 (atomic_inc(&F2FS_I_SB(inode)->swapfile_inode)) 4028 #define stat_dec_swapfile_inode(inode) \ 4029 (atomic_dec(&F2FS_I_SB(inode)->swapfile_inode)) 4030 #define stat_inc_atomic_inode(inode) \ 4031 (atomic_inc(&F2FS_I_SB(inode)->atomic_files)) 4032 #define stat_dec_atomic_inode(inode) \ 4033 (atomic_dec(&F2FS_I_SB(inode)->atomic_files)) 4034 #define stat_inc_meta_count(sbi, blkaddr) \ 4035 do { \ 4036 if (blkaddr < SIT_I(sbi)->sit_base_addr) \ 4037 atomic_inc(&(sbi)->meta_count[META_CP]); \ 4038 else if (blkaddr < NM_I(sbi)->nat_blkaddr) \ 4039 atomic_inc(&(sbi)->meta_count[META_SIT]); \ 4040 else if (blkaddr < SM_I(sbi)->ssa_blkaddr) \ 4041 atomic_inc(&(sbi)->meta_count[META_NAT]); \ 4042 else if (blkaddr < SM_I(sbi)->main_blkaddr) \ 4043 atomic_inc(&(sbi)->meta_count[META_SSA]); \ 4044 } while (0) 4045 #define stat_inc_seg_type(sbi, curseg) \ 4046 ((sbi)->segment_count[(curseg)->alloc_type]++) 4047 #define stat_inc_block_count(sbi, curseg) \ 4048 ((sbi)->block_count[(curseg)->alloc_type]++) 4049 #define stat_inc_inplace_blocks(sbi) \ 4050 (atomic_inc(&(sbi)->inplace_count)) 4051 #define stat_update_max_atomic_write(inode) \ 4052 do { \ 4053 int cur = atomic_read(&F2FS_I_SB(inode)->atomic_files); \ 4054 int max = atomic_read(&F2FS_I_SB(inode)->max_aw_cnt); \ 4055 if (cur > max) \ 4056 atomic_set(&F2FS_I_SB(inode)->max_aw_cnt, cur); \ 4057 } while (0) 4058 #define stat_inc_gc_call_count(sbi, foreground) \ 4059 (F2FS_STAT(sbi)->gc_call_count[(foreground)]++) 4060 #define stat_inc_gc_sec_count(sbi, type, gc_type) \ 4061 (F2FS_STAT(sbi)->gc_secs[(type)][(gc_type)]++) 4062 #define stat_inc_gc_seg_count(sbi, type, gc_type) \ 4063 (F2FS_STAT(sbi)->gc_segs[(type)][(gc_type)]++) 4064 4065 #define stat_inc_tot_blk_count(si, blks) \ 4066 ((si)->tot_blks += (blks)) 4067 4068 #define stat_inc_data_blk_count(sbi, blks, gc_type) \ 4069 do { \ 4070 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 4071 stat_inc_tot_blk_count(si, blks); \ 4072 si->data_blks += (blks); \ 4073 si->bg_data_blks += ((gc_type) == BG_GC) ? (blks) : 0; \ 4074 } while (0) 4075 4076 #define stat_inc_node_blk_count(sbi, blks, gc_type) \ 4077 do { \ 4078 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 4079 stat_inc_tot_blk_count(si, blks); \ 4080 si->node_blks += (blks); \ 4081 si->bg_node_blks += ((gc_type) == BG_GC) ? (blks) : 0; \ 4082 } while (0) 4083 4084 int f2fs_build_stats(struct f2fs_sb_info *sbi); 4085 void f2fs_destroy_stats(struct f2fs_sb_info *sbi); 4086 void __init f2fs_create_root_stats(void); 4087 void f2fs_destroy_root_stats(void); 4088 void f2fs_update_sit_info(struct f2fs_sb_info *sbi); 4089 #else 4090 #define stat_inc_cp_call_count(sbi, foreground) do { } while (0) 4091 #define stat_inc_cp_count(sbi) do { } while (0) 4092 #define stat_io_skip_bggc_count(sbi) do { } while (0) 4093 #define stat_other_skip_bggc_count(sbi) do { } while (0) 4094 #define stat_inc_dirty_inode(sbi, type) do { } while (0) 4095 #define stat_dec_dirty_inode(sbi, type) do { } while (0) 4096 #define stat_inc_total_hit(sbi, type) do { } while (0) 4097 #define stat_inc_rbtree_node_hit(sbi, type) do { } while (0) 4098 #define stat_inc_largest_node_hit(sbi) do { } while (0) 4099 #define stat_inc_cached_node_hit(sbi, type) do { } while (0) 4100 #define stat_inc_inline_xattr(inode) do { } while (0) 4101 #define stat_dec_inline_xattr(inode) do { } while (0) 4102 #define stat_inc_inline_inode(inode) do { } while (0) 4103 #define stat_dec_inline_inode(inode) do { } while (0) 4104 #define stat_inc_inline_dir(inode) do { } while (0) 4105 #define stat_dec_inline_dir(inode) do { } while (0) 4106 #define stat_inc_compr_inode(inode) do { } while (0) 4107 #define stat_dec_compr_inode(inode) do { } while (0) 4108 #define stat_add_compr_blocks(inode, blocks) do { } while (0) 4109 #define stat_sub_compr_blocks(inode, blocks) do { } while (0) 4110 #define stat_inc_swapfile_inode(inode) do { } while (0) 4111 #define stat_dec_swapfile_inode(inode) do { } while (0) 4112 #define stat_inc_atomic_inode(inode) do { } while (0) 4113 #define stat_dec_atomic_inode(inode) do { } while (0) 4114 #define stat_update_max_atomic_write(inode) do { } while (0) 4115 #define stat_inc_meta_count(sbi, blkaddr) do { } while (0) 4116 #define stat_inc_seg_type(sbi, curseg) do { } while (0) 4117 #define stat_inc_block_count(sbi, curseg) do { } while (0) 4118 #define stat_inc_inplace_blocks(sbi) do { } while (0) 4119 #define stat_inc_gc_call_count(sbi, foreground) do { } while (0) 4120 #define stat_inc_gc_sec_count(sbi, type, gc_type) do { } while (0) 4121 #define stat_inc_gc_seg_count(sbi, type, gc_type) do { } while (0) 4122 #define stat_inc_tot_blk_count(si, blks) do { } while (0) 4123 #define stat_inc_data_blk_count(sbi, blks, gc_type) do { } while (0) 4124 #define stat_inc_node_blk_count(sbi, blks, gc_type) do { } while (0) 4125 4126 static inline int f2fs_build_stats(struct f2fs_sb_info *sbi) { return 0; } 4127 static inline void f2fs_destroy_stats(struct f2fs_sb_info *sbi) { } 4128 static inline void __init f2fs_create_root_stats(void) { } 4129 static inline void f2fs_destroy_root_stats(void) { } 4130 static inline void f2fs_update_sit_info(struct f2fs_sb_info *sbi) {} 4131 #endif 4132 4133 extern const struct file_operations f2fs_dir_operations; 4134 extern const struct file_operations f2fs_file_operations; 4135 extern const struct inode_operations f2fs_file_inode_operations; 4136 extern const struct address_space_operations f2fs_dblock_aops; 4137 extern const struct address_space_operations f2fs_node_aops; 4138 extern const struct address_space_operations f2fs_meta_aops; 4139 extern const struct inode_operations f2fs_dir_inode_operations; 4140 extern const struct inode_operations f2fs_symlink_inode_operations; 4141 extern const struct inode_operations f2fs_encrypted_symlink_inode_operations; 4142 extern const struct inode_operations f2fs_special_inode_operations; 4143 extern struct kmem_cache *f2fs_inode_entry_slab; 4144 4145 /* 4146 * inline.c 4147 */ 4148 bool f2fs_may_inline_data(struct inode *inode); 4149 bool f2fs_sanity_check_inline_data(struct inode *inode); 4150 bool f2fs_may_inline_dentry(struct inode *inode); 4151 void f2fs_do_read_inline_data(struct page *page, struct page *ipage); 4152 void f2fs_truncate_inline_inode(struct inode *inode, 4153 struct page *ipage, u64 from); 4154 int f2fs_read_inline_data(struct inode *inode, struct page *page); 4155 int f2fs_convert_inline_page(struct dnode_of_data *dn, struct page *page); 4156 int f2fs_convert_inline_inode(struct inode *inode); 4157 int f2fs_try_convert_inline_dir(struct inode *dir, struct dentry *dentry); 4158 int f2fs_write_inline_data(struct inode *inode, struct page *page); 4159 int f2fs_recover_inline_data(struct inode *inode, struct page *npage); 4160 struct f2fs_dir_entry *f2fs_find_in_inline_dir(struct inode *dir, 4161 const struct f2fs_filename *fname, 4162 struct page **res_page); 4163 int f2fs_make_empty_inline_dir(struct inode *inode, struct inode *parent, 4164 struct page *ipage); 4165 int f2fs_add_inline_entry(struct inode *dir, const struct f2fs_filename *fname, 4166 struct inode *inode, nid_t ino, umode_t mode); 4167 void f2fs_delete_inline_entry(struct f2fs_dir_entry *dentry, 4168 struct page *page, struct inode *dir, 4169 struct inode *inode); 4170 bool f2fs_empty_inline_dir(struct inode *dir); 4171 int f2fs_read_inline_dir(struct file *file, struct dir_context *ctx, 4172 struct fscrypt_str *fstr); 4173 int f2fs_inline_data_fiemap(struct inode *inode, 4174 struct fiemap_extent_info *fieinfo, 4175 __u64 start, __u64 len); 4176 4177 /* 4178 * shrinker.c 4179 */ 4180 unsigned long f2fs_shrink_count(struct shrinker *shrink, 4181 struct shrink_control *sc); 4182 unsigned long f2fs_shrink_scan(struct shrinker *shrink, 4183 struct shrink_control *sc); 4184 void f2fs_join_shrinker(struct f2fs_sb_info *sbi); 4185 void f2fs_leave_shrinker(struct f2fs_sb_info *sbi); 4186 4187 /* 4188 * extent_cache.c 4189 */ 4190 bool sanity_check_extent_cache(struct inode *inode); 4191 void f2fs_init_extent_tree(struct inode *inode); 4192 void f2fs_drop_extent_tree(struct inode *inode); 4193 void f2fs_destroy_extent_node(struct inode *inode); 4194 void f2fs_destroy_extent_tree(struct inode *inode); 4195 void f2fs_init_extent_cache_info(struct f2fs_sb_info *sbi); 4196 int __init f2fs_create_extent_cache(void); 4197 void f2fs_destroy_extent_cache(void); 4198 4199 /* read extent cache ops */ 4200 void f2fs_init_read_extent_tree(struct inode *inode, struct page *ipage); 4201 bool f2fs_lookup_read_extent_cache(struct inode *inode, pgoff_t pgofs, 4202 struct extent_info *ei); 4203 bool f2fs_lookup_read_extent_cache_block(struct inode *inode, pgoff_t index, 4204 block_t *blkaddr); 4205 void f2fs_update_read_extent_cache(struct dnode_of_data *dn); 4206 void f2fs_update_read_extent_cache_range(struct dnode_of_data *dn, 4207 pgoff_t fofs, block_t blkaddr, unsigned int len); 4208 unsigned int f2fs_shrink_read_extent_tree(struct f2fs_sb_info *sbi, 4209 int nr_shrink); 4210 4211 /* block age extent cache ops */ 4212 void f2fs_init_age_extent_tree(struct inode *inode); 4213 bool f2fs_lookup_age_extent_cache(struct inode *inode, pgoff_t pgofs, 4214 struct extent_info *ei); 4215 void f2fs_update_age_extent_cache(struct dnode_of_data *dn); 4216 void f2fs_update_age_extent_cache_range(struct dnode_of_data *dn, 4217 pgoff_t fofs, unsigned int len); 4218 unsigned int f2fs_shrink_age_extent_tree(struct f2fs_sb_info *sbi, 4219 int nr_shrink); 4220 4221 /* 4222 * sysfs.c 4223 */ 4224 #define MIN_RA_MUL 2 4225 #define MAX_RA_MUL 256 4226 4227 int __init f2fs_init_sysfs(void); 4228 void f2fs_exit_sysfs(void); 4229 int f2fs_register_sysfs(struct f2fs_sb_info *sbi); 4230 void f2fs_unregister_sysfs(struct f2fs_sb_info *sbi); 4231 4232 /* verity.c */ 4233 extern const struct fsverity_operations f2fs_verityops; 4234 4235 /* 4236 * crypto support 4237 */ 4238 static inline bool f2fs_encrypted_file(struct inode *inode) 4239 { 4240 return IS_ENCRYPTED(inode) && S_ISREG(inode->i_mode); 4241 } 4242 4243 static inline void f2fs_set_encrypted_inode(struct inode *inode) 4244 { 4245 #ifdef CONFIG_FS_ENCRYPTION 4246 file_set_encrypt(inode); 4247 f2fs_set_inode_flags(inode); 4248 #endif 4249 } 4250 4251 /* 4252 * Returns true if the reads of the inode's data need to undergo some 4253 * postprocessing step, like decryption or authenticity verification. 4254 */ 4255 static inline bool f2fs_post_read_required(struct inode *inode) 4256 { 4257 return f2fs_encrypted_file(inode) || fsverity_active(inode) || 4258 f2fs_compressed_file(inode); 4259 } 4260 4261 /* 4262 * compress.c 4263 */ 4264 #ifdef CONFIG_F2FS_FS_COMPRESSION 4265 bool f2fs_is_compressed_page(struct page *page); 4266 struct page *f2fs_compress_control_page(struct page *page); 4267 int f2fs_prepare_compress_overwrite(struct inode *inode, 4268 struct page **pagep, pgoff_t index, void **fsdata); 4269 bool f2fs_compress_write_end(struct inode *inode, void *fsdata, 4270 pgoff_t index, unsigned copied); 4271 int f2fs_truncate_partial_cluster(struct inode *inode, u64 from, bool lock); 4272 void f2fs_compress_write_end_io(struct bio *bio, struct page *page); 4273 bool f2fs_is_compress_backend_ready(struct inode *inode); 4274 bool f2fs_is_compress_level_valid(int alg, int lvl); 4275 int __init f2fs_init_compress_mempool(void); 4276 void f2fs_destroy_compress_mempool(void); 4277 void f2fs_decompress_cluster(struct decompress_io_ctx *dic, bool in_task); 4278 void f2fs_end_read_compressed_page(struct page *page, bool failed, 4279 block_t blkaddr, bool in_task); 4280 bool f2fs_cluster_is_empty(struct compress_ctx *cc); 4281 bool f2fs_cluster_can_merge_page(struct compress_ctx *cc, pgoff_t index); 4282 bool f2fs_all_cluster_page_ready(struct compress_ctx *cc, struct page **pages, 4283 int index, int nr_pages, bool uptodate); 4284 bool f2fs_sanity_check_cluster(struct dnode_of_data *dn); 4285 void f2fs_compress_ctx_add_page(struct compress_ctx *cc, struct page *page); 4286 int f2fs_write_multi_pages(struct compress_ctx *cc, 4287 int *submitted, 4288 struct writeback_control *wbc, 4289 enum iostat_type io_type); 4290 int f2fs_is_compressed_cluster(struct inode *inode, pgoff_t index); 4291 void f2fs_update_read_extent_tree_range_compressed(struct inode *inode, 4292 pgoff_t fofs, block_t blkaddr, 4293 unsigned int llen, unsigned int c_len); 4294 int f2fs_read_multi_pages(struct compress_ctx *cc, struct bio **bio_ret, 4295 unsigned nr_pages, sector_t *last_block_in_bio, 4296 bool is_readahead, bool for_write); 4297 struct decompress_io_ctx *f2fs_alloc_dic(struct compress_ctx *cc); 4298 void f2fs_decompress_end_io(struct decompress_io_ctx *dic, bool failed, 4299 bool in_task); 4300 void f2fs_put_page_dic(struct page *page, bool in_task); 4301 unsigned int f2fs_cluster_blocks_are_contiguous(struct dnode_of_data *dn, 4302 unsigned int ofs_in_node); 4303 int f2fs_init_compress_ctx(struct compress_ctx *cc); 4304 void f2fs_destroy_compress_ctx(struct compress_ctx *cc, bool reuse); 4305 void f2fs_init_compress_info(struct f2fs_sb_info *sbi); 4306 int f2fs_init_compress_inode(struct f2fs_sb_info *sbi); 4307 void f2fs_destroy_compress_inode(struct f2fs_sb_info *sbi); 4308 int f2fs_init_page_array_cache(struct f2fs_sb_info *sbi); 4309 void f2fs_destroy_page_array_cache(struct f2fs_sb_info *sbi); 4310 int __init f2fs_init_compress_cache(void); 4311 void f2fs_destroy_compress_cache(void); 4312 struct address_space *COMPRESS_MAPPING(struct f2fs_sb_info *sbi); 4313 void f2fs_invalidate_compress_page(struct f2fs_sb_info *sbi, block_t blkaddr); 4314 void f2fs_cache_compressed_page(struct f2fs_sb_info *sbi, struct page *page, 4315 nid_t ino, block_t blkaddr); 4316 bool f2fs_load_compressed_page(struct f2fs_sb_info *sbi, struct page *page, 4317 block_t blkaddr); 4318 void f2fs_invalidate_compress_pages(struct f2fs_sb_info *sbi, nid_t ino); 4319 #define inc_compr_inode_stat(inode) \ 4320 do { \ 4321 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); \ 4322 sbi->compr_new_inode++; \ 4323 } while (0) 4324 #define add_compr_block_stat(inode, blocks) \ 4325 do { \ 4326 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); \ 4327 int diff = F2FS_I(inode)->i_cluster_size - blocks; \ 4328 sbi->compr_written_block += blocks; \ 4329 sbi->compr_saved_block += diff; \ 4330 } while (0) 4331 #else 4332 static inline bool f2fs_is_compressed_page(struct page *page) { return false; } 4333 static inline bool f2fs_is_compress_backend_ready(struct inode *inode) 4334 { 4335 if (!f2fs_compressed_file(inode)) 4336 return true; 4337 /* not support compression */ 4338 return false; 4339 } 4340 static inline bool f2fs_is_compress_level_valid(int alg, int lvl) { return false; } 4341 static inline struct page *f2fs_compress_control_page(struct page *page) 4342 { 4343 WARN_ON_ONCE(1); 4344 return ERR_PTR(-EINVAL); 4345 } 4346 static inline int __init f2fs_init_compress_mempool(void) { return 0; } 4347 static inline void f2fs_destroy_compress_mempool(void) { } 4348 static inline void f2fs_decompress_cluster(struct decompress_io_ctx *dic, 4349 bool in_task) { } 4350 static inline void f2fs_end_read_compressed_page(struct page *page, 4351 bool failed, block_t blkaddr, bool in_task) 4352 { 4353 WARN_ON_ONCE(1); 4354 } 4355 static inline void f2fs_put_page_dic(struct page *page, bool in_task) 4356 { 4357 WARN_ON_ONCE(1); 4358 } 4359 static inline unsigned int f2fs_cluster_blocks_are_contiguous( 4360 struct dnode_of_data *dn, unsigned int ofs_in_node) { return 0; } 4361 static inline bool f2fs_sanity_check_cluster(struct dnode_of_data *dn) { return false; } 4362 static inline int f2fs_init_compress_inode(struct f2fs_sb_info *sbi) { return 0; } 4363 static inline void f2fs_destroy_compress_inode(struct f2fs_sb_info *sbi) { } 4364 static inline int f2fs_init_page_array_cache(struct f2fs_sb_info *sbi) { return 0; } 4365 static inline void f2fs_destroy_page_array_cache(struct f2fs_sb_info *sbi) { } 4366 static inline int __init f2fs_init_compress_cache(void) { return 0; } 4367 static inline void f2fs_destroy_compress_cache(void) { } 4368 static inline void f2fs_invalidate_compress_page(struct f2fs_sb_info *sbi, 4369 block_t blkaddr) { } 4370 static inline void f2fs_cache_compressed_page(struct f2fs_sb_info *sbi, 4371 struct page *page, nid_t ino, block_t blkaddr) { } 4372 static inline bool f2fs_load_compressed_page(struct f2fs_sb_info *sbi, 4373 struct page *page, block_t blkaddr) { return false; } 4374 static inline void f2fs_invalidate_compress_pages(struct f2fs_sb_info *sbi, 4375 nid_t ino) { } 4376 #define inc_compr_inode_stat(inode) do { } while (0) 4377 static inline void f2fs_update_read_extent_tree_range_compressed( 4378 struct inode *inode, 4379 pgoff_t fofs, block_t blkaddr, 4380 unsigned int llen, unsigned int c_len) { } 4381 #endif 4382 4383 static inline int set_compress_context(struct inode *inode) 4384 { 4385 #ifdef CONFIG_F2FS_FS_COMPRESSION 4386 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 4387 4388 F2FS_I(inode)->i_compress_algorithm = 4389 F2FS_OPTION(sbi).compress_algorithm; 4390 F2FS_I(inode)->i_log_cluster_size = 4391 F2FS_OPTION(sbi).compress_log_size; 4392 F2FS_I(inode)->i_compress_flag = 4393 F2FS_OPTION(sbi).compress_chksum ? 4394 BIT(COMPRESS_CHKSUM) : 0; 4395 F2FS_I(inode)->i_cluster_size = 4396 BIT(F2FS_I(inode)->i_log_cluster_size); 4397 if ((F2FS_I(inode)->i_compress_algorithm == COMPRESS_LZ4 || 4398 F2FS_I(inode)->i_compress_algorithm == COMPRESS_ZSTD) && 4399 F2FS_OPTION(sbi).compress_level) 4400 F2FS_I(inode)->i_compress_level = 4401 F2FS_OPTION(sbi).compress_level; 4402 F2FS_I(inode)->i_flags |= F2FS_COMPR_FL; 4403 set_inode_flag(inode, FI_COMPRESSED_FILE); 4404 stat_inc_compr_inode(inode); 4405 inc_compr_inode_stat(inode); 4406 f2fs_mark_inode_dirty_sync(inode, true); 4407 return 0; 4408 #else 4409 return -EOPNOTSUPP; 4410 #endif 4411 } 4412 4413 static inline bool f2fs_disable_compressed_file(struct inode *inode) 4414 { 4415 struct f2fs_inode_info *fi = F2FS_I(inode); 4416 4417 f2fs_down_write(&F2FS_I(inode)->i_sem); 4418 4419 if (!f2fs_compressed_file(inode)) { 4420 f2fs_up_write(&F2FS_I(inode)->i_sem); 4421 return true; 4422 } 4423 if (f2fs_is_mmap_file(inode) || 4424 (S_ISREG(inode->i_mode) && F2FS_HAS_BLOCKS(inode))) { 4425 f2fs_up_write(&F2FS_I(inode)->i_sem); 4426 return false; 4427 } 4428 4429 fi->i_flags &= ~F2FS_COMPR_FL; 4430 stat_dec_compr_inode(inode); 4431 clear_inode_flag(inode, FI_COMPRESSED_FILE); 4432 f2fs_mark_inode_dirty_sync(inode, true); 4433 4434 f2fs_up_write(&F2FS_I(inode)->i_sem); 4435 return true; 4436 } 4437 4438 #define F2FS_FEATURE_FUNCS(name, flagname) \ 4439 static inline bool f2fs_sb_has_##name(struct f2fs_sb_info *sbi) \ 4440 { \ 4441 return F2FS_HAS_FEATURE(sbi, F2FS_FEATURE_##flagname); \ 4442 } 4443 4444 F2FS_FEATURE_FUNCS(encrypt, ENCRYPT); 4445 F2FS_FEATURE_FUNCS(blkzoned, BLKZONED); 4446 F2FS_FEATURE_FUNCS(extra_attr, EXTRA_ATTR); 4447 F2FS_FEATURE_FUNCS(project_quota, PRJQUOTA); 4448 F2FS_FEATURE_FUNCS(inode_chksum, INODE_CHKSUM); 4449 F2FS_FEATURE_FUNCS(flexible_inline_xattr, FLEXIBLE_INLINE_XATTR); 4450 F2FS_FEATURE_FUNCS(quota_ino, QUOTA_INO); 4451 F2FS_FEATURE_FUNCS(inode_crtime, INODE_CRTIME); 4452 F2FS_FEATURE_FUNCS(lost_found, LOST_FOUND); 4453 F2FS_FEATURE_FUNCS(verity, VERITY); 4454 F2FS_FEATURE_FUNCS(sb_chksum, SB_CHKSUM); 4455 F2FS_FEATURE_FUNCS(casefold, CASEFOLD); 4456 F2FS_FEATURE_FUNCS(compression, COMPRESSION); 4457 F2FS_FEATURE_FUNCS(readonly, RO); 4458 4459 #ifdef CONFIG_BLK_DEV_ZONED 4460 static inline bool f2fs_blkz_is_seq(struct f2fs_sb_info *sbi, int devi, 4461 block_t blkaddr) 4462 { 4463 unsigned int zno = blkaddr / sbi->blocks_per_blkz; 4464 4465 return test_bit(zno, FDEV(devi).blkz_seq); 4466 } 4467 #endif 4468 4469 static inline int f2fs_bdev_index(struct f2fs_sb_info *sbi, 4470 struct block_device *bdev) 4471 { 4472 int i; 4473 4474 if (!f2fs_is_multi_device(sbi)) 4475 return 0; 4476 4477 for (i = 0; i < sbi->s_ndevs; i++) 4478 if (FDEV(i).bdev == bdev) 4479 return i; 4480 4481 WARN_ON(1); 4482 return -1; 4483 } 4484 4485 static inline bool f2fs_hw_should_discard(struct f2fs_sb_info *sbi) 4486 { 4487 return f2fs_sb_has_blkzoned(sbi); 4488 } 4489 4490 static inline bool f2fs_bdev_support_discard(struct block_device *bdev) 4491 { 4492 return bdev_max_discard_sectors(bdev) || bdev_is_zoned(bdev); 4493 } 4494 4495 static inline bool f2fs_hw_support_discard(struct f2fs_sb_info *sbi) 4496 { 4497 int i; 4498 4499 if (!f2fs_is_multi_device(sbi)) 4500 return f2fs_bdev_support_discard(sbi->sb->s_bdev); 4501 4502 for (i = 0; i < sbi->s_ndevs; i++) 4503 if (f2fs_bdev_support_discard(FDEV(i).bdev)) 4504 return true; 4505 return false; 4506 } 4507 4508 static inline bool f2fs_realtime_discard_enable(struct f2fs_sb_info *sbi) 4509 { 4510 return (test_opt(sbi, DISCARD) && f2fs_hw_support_discard(sbi)) || 4511 f2fs_hw_should_discard(sbi); 4512 } 4513 4514 static inline bool f2fs_hw_is_readonly(struct f2fs_sb_info *sbi) 4515 { 4516 int i; 4517 4518 if (!f2fs_is_multi_device(sbi)) 4519 return bdev_read_only(sbi->sb->s_bdev); 4520 4521 for (i = 0; i < sbi->s_ndevs; i++) 4522 if (bdev_read_only(FDEV(i).bdev)) 4523 return true; 4524 return false; 4525 } 4526 4527 static inline bool f2fs_dev_is_readonly(struct f2fs_sb_info *sbi) 4528 { 4529 return f2fs_sb_has_readonly(sbi) || f2fs_hw_is_readonly(sbi); 4530 } 4531 4532 static inline bool f2fs_lfs_mode(struct f2fs_sb_info *sbi) 4533 { 4534 return F2FS_OPTION(sbi).fs_mode == FS_MODE_LFS; 4535 } 4536 4537 static inline bool f2fs_valid_pinned_area(struct f2fs_sb_info *sbi, 4538 block_t blkaddr) 4539 { 4540 if (f2fs_sb_has_blkzoned(sbi)) { 4541 int devi = f2fs_target_device_index(sbi, blkaddr); 4542 4543 return !bdev_is_zoned(FDEV(devi).bdev); 4544 } 4545 return true; 4546 } 4547 4548 static inline bool f2fs_low_mem_mode(struct f2fs_sb_info *sbi) 4549 { 4550 return F2FS_OPTION(sbi).memory_mode == MEMORY_MODE_LOW; 4551 } 4552 4553 static inline bool f2fs_may_compress(struct inode *inode) 4554 { 4555 if (IS_SWAPFILE(inode) || f2fs_is_pinned_file(inode) || 4556 f2fs_is_atomic_file(inode) || f2fs_has_inline_data(inode) || 4557 f2fs_is_mmap_file(inode)) 4558 return false; 4559 return S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode); 4560 } 4561 4562 static inline void f2fs_i_compr_blocks_update(struct inode *inode, 4563 u64 blocks, bool add) 4564 { 4565 struct f2fs_inode_info *fi = F2FS_I(inode); 4566 int diff = fi->i_cluster_size - blocks; 4567 4568 /* don't update i_compr_blocks if saved blocks were released */ 4569 if (!add && !atomic_read(&fi->i_compr_blocks)) 4570 return; 4571 4572 if (add) { 4573 atomic_add(diff, &fi->i_compr_blocks); 4574 stat_add_compr_blocks(inode, diff); 4575 } else { 4576 atomic_sub(diff, &fi->i_compr_blocks); 4577 stat_sub_compr_blocks(inode, diff); 4578 } 4579 f2fs_mark_inode_dirty_sync(inode, true); 4580 } 4581 4582 static inline bool f2fs_allow_multi_device_dio(struct f2fs_sb_info *sbi, 4583 int flag) 4584 { 4585 if (!f2fs_is_multi_device(sbi)) 4586 return false; 4587 if (flag != F2FS_GET_BLOCK_DIO) 4588 return false; 4589 return sbi->aligned_blksize; 4590 } 4591 4592 static inline bool f2fs_need_verity(const struct inode *inode, pgoff_t idx) 4593 { 4594 return fsverity_active(inode) && 4595 idx < DIV_ROUND_UP(inode->i_size, PAGE_SIZE); 4596 } 4597 4598 #ifdef CONFIG_F2FS_FAULT_INJECTION 4599 extern void f2fs_build_fault_attr(struct f2fs_sb_info *sbi, unsigned int rate, 4600 unsigned int type); 4601 #else 4602 #define f2fs_build_fault_attr(sbi, rate, type) do { } while (0) 4603 #endif 4604 4605 static inline bool is_journalled_quota(struct f2fs_sb_info *sbi) 4606 { 4607 #ifdef CONFIG_QUOTA 4608 if (f2fs_sb_has_quota_ino(sbi)) 4609 return true; 4610 if (F2FS_OPTION(sbi).s_qf_names[USRQUOTA] || 4611 F2FS_OPTION(sbi).s_qf_names[GRPQUOTA] || 4612 F2FS_OPTION(sbi).s_qf_names[PRJQUOTA]) 4613 return true; 4614 #endif 4615 return false; 4616 } 4617 4618 static inline bool f2fs_block_unit_discard(struct f2fs_sb_info *sbi) 4619 { 4620 return F2FS_OPTION(sbi).discard_unit == DISCARD_UNIT_BLOCK; 4621 } 4622 4623 static inline void f2fs_io_schedule_timeout(long timeout) 4624 { 4625 set_current_state(TASK_UNINTERRUPTIBLE); 4626 io_schedule_timeout(timeout); 4627 } 4628 4629 static inline void f2fs_handle_page_eio(struct f2fs_sb_info *sbi, pgoff_t ofs, 4630 enum page_type type) 4631 { 4632 if (unlikely(f2fs_cp_error(sbi))) 4633 return; 4634 4635 if (ofs == sbi->page_eio_ofs[type]) { 4636 if (sbi->page_eio_cnt[type]++ == MAX_RETRY_PAGE_EIO) 4637 set_ckpt_flags(sbi, CP_ERROR_FLAG); 4638 } else { 4639 sbi->page_eio_ofs[type] = ofs; 4640 sbi->page_eio_cnt[type] = 0; 4641 } 4642 } 4643 4644 static inline bool f2fs_is_readonly(struct f2fs_sb_info *sbi) 4645 { 4646 return f2fs_sb_has_readonly(sbi) || f2fs_readonly(sbi->sb); 4647 } 4648 4649 static inline void f2fs_truncate_meta_inode_pages(struct f2fs_sb_info *sbi, 4650 block_t blkaddr, unsigned int cnt) 4651 { 4652 bool need_submit = false; 4653 int i = 0; 4654 4655 do { 4656 struct page *page; 4657 4658 page = find_get_page(META_MAPPING(sbi), blkaddr + i); 4659 if (page) { 4660 if (PageWriteback(page)) 4661 need_submit = true; 4662 f2fs_put_page(page, 0); 4663 } 4664 } while (++i < cnt && !need_submit); 4665 4666 if (need_submit) 4667 f2fs_submit_merged_write_cond(sbi, sbi->meta_inode, 4668 NULL, 0, DATA); 4669 4670 truncate_inode_pages_range(META_MAPPING(sbi), 4671 F2FS_BLK_TO_BYTES((loff_t)blkaddr), 4672 F2FS_BLK_END_BYTES((loff_t)(blkaddr + cnt - 1))); 4673 } 4674 4675 static inline void f2fs_invalidate_internal_cache(struct f2fs_sb_info *sbi, 4676 block_t blkaddr) 4677 { 4678 f2fs_truncate_meta_inode_pages(sbi, blkaddr, 1); 4679 f2fs_invalidate_compress_page(sbi, blkaddr); 4680 } 4681 4682 #define EFSBADCRC EBADMSG /* Bad CRC detected */ 4683 #define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */ 4684 4685 #endif /* _LINUX_F2FS_H */ 4686