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/slab.h> 15 #include <linux/crc32.h> 16 #include <linux/magic.h> 17 #include <linux/kobject.h> 18 #include <linux/sched.h> 19 #include <linux/cred.h> 20 #include <linux/sched/mm.h> 21 #include <linux/vmalloc.h> 22 #include <linux/bio.h> 23 #include <linux/blkdev.h> 24 #include <linux/quotaops.h> 25 #include <linux/part_stat.h> 26 #include <linux/rw_hint.h> 27 28 #include <linux/fscrypt.h> 29 #include <linux/fsverity.h> 30 31 #ifdef CONFIG_F2FS_CHECK_FS 32 #define f2fs_bug_on(sbi, condition) BUG_ON(condition) 33 #else 34 #define f2fs_bug_on(sbi, condition) \ 35 do { \ 36 if (WARN_ON(condition)) \ 37 set_sbi_flag(sbi, SBI_NEED_FSCK); \ 38 } while (0) 39 #endif 40 41 enum { 42 FAULT_KMALLOC, 43 FAULT_KVMALLOC, 44 FAULT_PAGE_ALLOC, 45 FAULT_PAGE_GET, 46 FAULT_ALLOC_BIO, /* it's obsolete due to bio_alloc() will never fail */ 47 FAULT_ALLOC_NID, 48 FAULT_ORPHAN, 49 FAULT_BLOCK, 50 FAULT_DIR_DEPTH, 51 FAULT_EVICT_INODE, 52 FAULT_TRUNCATE, 53 FAULT_READ_IO, 54 FAULT_CHECKPOINT, 55 FAULT_DISCARD, /* it's obsolete due to __blkdev_issue_discard() will never fail */ 56 FAULT_WRITE_IO, 57 FAULT_SLAB_ALLOC, 58 FAULT_DQUOT_INIT, 59 FAULT_LOCK_OP, 60 FAULT_BLKADDR_VALIDITY, 61 FAULT_BLKADDR_CONSISTENCE, 62 FAULT_NO_SEGMENT, 63 FAULT_INCONSISTENT_FOOTER, 64 FAULT_ATOMIC_TIMEOUT, 65 FAULT_VMALLOC, 66 FAULT_LOCK_TIMEOUT, 67 FAULT_SKIP_WRITE, 68 FAULT_MAX, 69 }; 70 71 /* indicate which option to update */ 72 enum fault_option { 73 FAULT_RATE = 1, /* only update fault rate */ 74 FAULT_TYPE = 2, /* only update fault type */ 75 FAULT_TIMEOUT = 4, /* only update fault timeout type */ 76 FAULT_ALL = 8, /* reset all fault injection options/stats */ 77 }; 78 79 #ifdef CONFIG_F2FS_FAULT_INJECTION 80 struct f2fs_fault_info { 81 atomic_t inject_ops; 82 int inject_rate; 83 unsigned int inject_type; 84 /* Used to account total count of injection for each type */ 85 unsigned int inject_count[FAULT_MAX]; 86 unsigned int inject_lock_timeout; /* inject lock timeout */ 87 }; 88 89 extern const char *f2fs_fault_name[FAULT_MAX]; 90 #define IS_FAULT_SET(fi, type) ((fi)->inject_type & BIT(type)) 91 92 /* maximum retry count for injected failure */ 93 #define DEFAULT_FAILURE_RETRY_COUNT 8 94 #else 95 #define DEFAULT_FAILURE_RETRY_COUNT 1 96 #endif 97 98 /* 99 * For mount options 100 */ 101 enum f2fs_mount_opt { 102 F2FS_MOUNT_DISABLE_ROLL_FORWARD, 103 F2FS_MOUNT_DISCARD, 104 F2FS_MOUNT_NOHEAP, 105 F2FS_MOUNT_XATTR_USER, 106 F2FS_MOUNT_POSIX_ACL, 107 F2FS_MOUNT_DISABLE_EXT_IDENTIFY, 108 F2FS_MOUNT_INLINE_XATTR, 109 F2FS_MOUNT_INLINE_DATA, 110 F2FS_MOUNT_INLINE_DENTRY, 111 F2FS_MOUNT_FLUSH_MERGE, 112 F2FS_MOUNT_NOBARRIER, 113 F2FS_MOUNT_FASTBOOT, 114 F2FS_MOUNT_READ_EXTENT_CACHE, 115 F2FS_MOUNT_DATA_FLUSH, 116 F2FS_MOUNT_FAULT_INJECTION, 117 F2FS_MOUNT_USRQUOTA, 118 F2FS_MOUNT_GRPQUOTA, 119 F2FS_MOUNT_PRJQUOTA, 120 F2FS_MOUNT_QUOTA, 121 F2FS_MOUNT_INLINE_XATTR_SIZE, 122 F2FS_MOUNT_RESERVE_ROOT, 123 F2FS_MOUNT_DISABLE_CHECKPOINT, 124 F2FS_MOUNT_NORECOVERY, 125 F2FS_MOUNT_ATGC, 126 F2FS_MOUNT_MERGE_CHECKPOINT, 127 F2FS_MOUNT_GC_MERGE, 128 F2FS_MOUNT_COMPRESS_CACHE, 129 F2FS_MOUNT_AGE_EXTENT_CACHE, 130 F2FS_MOUNT_NAT_BITS, 131 F2FS_MOUNT_INLINECRYPT, 132 /* 133 * Some f2fs environments expect to be able to pass the "lazytime" option 134 * string rather than using the MS_LAZYTIME flag, so this must remain. 135 */ 136 F2FS_MOUNT_LAZYTIME, 137 F2FS_MOUNT_RESERVE_NODE, 138 }; 139 140 #define F2FS_OPTION(sbi) ((sbi)->mount_opt) 141 #define clear_opt(sbi, option) \ 142 (F2FS_OPTION(sbi).opt &= ~BIT(F2FS_MOUNT_##option)) 143 #define set_opt(sbi, option) \ 144 (F2FS_OPTION(sbi).opt |= BIT(F2FS_MOUNT_##option)) 145 #define test_opt(sbi, option) \ 146 (F2FS_OPTION(sbi).opt & BIT(F2FS_MOUNT_##option)) 147 148 #define ver_after(a, b) (typecheck(unsigned long long, a) && \ 149 typecheck(unsigned long long, b) && \ 150 ((long long)((a) - (b)) > 0)) 151 152 typedef u32 block_t; /* 153 * should not change u32, since it is the on-disk block 154 * address format, __le32. 155 */ 156 typedef u32 nid_t; 157 158 #define COMPRESS_EXT_NUM 16 159 160 enum blkzone_allocation_policy { 161 BLKZONE_ALLOC_PRIOR_SEQ, /* Prioritize writing to sequential zones */ 162 BLKZONE_ALLOC_ONLY_SEQ, /* Only allow writing to sequential zones */ 163 BLKZONE_ALLOC_PRIOR_CONV, /* Prioritize writing to conventional zones */ 164 }; 165 166 enum bggc_io_aware_policy { 167 AWARE_ALL_IO, /* skip background GC if there is any kind of pending IO */ 168 AWARE_READ_IO, /* skip background GC if there is pending read IO */ 169 AWARE_NONE, /* don't aware IO for background GC */ 170 }; 171 172 enum device_allocation_policy { 173 ALLOCATE_FORWARD_NOHINT, 174 ALLOCATE_FORWARD_WITHIN_HINT, 175 ALLOCATE_FORWARD_FROM_HINT, 176 }; 177 178 enum f2fs_lock_name { 179 LOCK_NAME_NONE, 180 LOCK_NAME_CP_RWSEM, 181 LOCK_NAME_NODE_CHANGE, 182 LOCK_NAME_NODE_WRITE, 183 LOCK_NAME_GC_LOCK, 184 LOCK_NAME_CP_GLOBAL, 185 LOCK_NAME_IO_RWSEM, 186 LOCK_NAME_MAX, 187 }; 188 189 enum f2fs_timeout_type { 190 TIMEOUT_TYPE_NONE, 191 TIMEOUT_TYPE_RUNNING, 192 TIMEOUT_TYPE_IO_SLEEP, 193 TIMEOUT_TYPE_NONIO_SLEEP, 194 TIMEOUT_TYPE_RUNNABLE, 195 TIMEOUT_TYPE_MAX, 196 }; 197 198 /* 199 * An implementation of an rwsem that is explicitly unfair to readers. This 200 * prevents priority inversion when a low-priority reader acquires the read lock 201 * while sleeping on the write lock but the write lock is needed by 202 * higher-priority clients. 203 */ 204 205 struct f2fs_rwsem { 206 struct f2fs_sb_info *sbi; 207 enum f2fs_lock_name name; 208 struct rw_semaphore internal_rwsem; 209 #ifdef CONFIG_F2FS_UNFAIR_RWSEM 210 wait_queue_head_t read_waiters; 211 #endif 212 }; 213 214 struct f2fs_mount_info { 215 unsigned long long opt; 216 block_t root_reserved_blocks; /* root reserved blocks */ 217 block_t root_reserved_nodes; /* root reserved nodes */ 218 kuid_t s_resuid; /* reserved blocks for uid */ 219 kgid_t s_resgid; /* reserved blocks for gid */ 220 int active_logs; /* # of active logs */ 221 int inline_xattr_size; /* inline xattr size */ 222 #ifdef CONFIG_F2FS_FAULT_INJECTION 223 struct f2fs_fault_info fault_info; /* For fault injection */ 224 #endif 225 #ifdef CONFIG_QUOTA 226 /* Names of quota files with journalled quota */ 227 char *s_qf_names[MAXQUOTAS]; 228 int s_jquota_fmt; /* Format of quota to use */ 229 #endif 230 /* For which write hints are passed down to block layer */ 231 int alloc_mode; /* segment allocation policy */ 232 int fsync_mode; /* fsync policy */ 233 int fs_mode; /* fs mode: LFS or ADAPTIVE */ 234 int bggc_mode; /* bggc mode: off, on or sync */ 235 int memory_mode; /* memory mode */ 236 int errors; /* errors parameter */ 237 int discard_unit; /* 238 * discard command's offset/size should 239 * be aligned to this unit: block, 240 * segment or section 241 */ 242 struct fscrypt_dummy_policy dummy_enc_policy; /* test dummy encryption */ 243 block_t unusable_cap_perc; /* percentage for cap */ 244 block_t unusable_cap; /* Amount of space allowed to be 245 * unusable when disabling checkpoint 246 */ 247 248 /* For compression */ 249 unsigned char compress_algorithm; /* algorithm type */ 250 unsigned char compress_log_size; /* cluster log size */ 251 unsigned char compress_level; /* compress level */ 252 bool compress_chksum; /* compressed data chksum */ 253 unsigned char compress_ext_cnt; /* extension count */ 254 unsigned char nocompress_ext_cnt; /* nocompress extension count */ 255 int compress_mode; /* compression mode */ 256 unsigned char extensions[COMPRESS_EXT_NUM][F2FS_EXTENSION_LEN]; /* extensions */ 257 unsigned char noextensions[COMPRESS_EXT_NUM][F2FS_EXTENSION_LEN]; /* extensions */ 258 unsigned int lookup_mode; 259 }; 260 261 #define F2FS_FEATURE_ENCRYPT 0x00000001 262 #define F2FS_FEATURE_BLKZONED 0x00000002 263 #define F2FS_FEATURE_ATOMIC_WRITE 0x00000004 264 #define F2FS_FEATURE_EXTRA_ATTR 0x00000008 265 #define F2FS_FEATURE_PRJQUOTA 0x00000010 266 #define F2FS_FEATURE_INODE_CHKSUM 0x00000020 267 #define F2FS_FEATURE_FLEXIBLE_INLINE_XATTR 0x00000040 268 #define F2FS_FEATURE_QUOTA_INO 0x00000080 269 #define F2FS_FEATURE_INODE_CRTIME 0x00000100 270 #define F2FS_FEATURE_LOST_FOUND 0x00000200 271 #define F2FS_FEATURE_VERITY 0x00000400 272 #define F2FS_FEATURE_SB_CHKSUM 0x00000800 273 #define F2FS_FEATURE_CASEFOLD 0x00001000 274 #define F2FS_FEATURE_COMPRESSION 0x00002000 275 #define F2FS_FEATURE_RO 0x00004000 276 #define F2FS_FEATURE_DEVICE_ALIAS 0x00008000 277 #define F2FS_FEATURE_PACKED_SSA 0x00010000 278 279 #define __F2FS_HAS_FEATURE(raw_super, mask) \ 280 ((raw_super->feature & cpu_to_le32(mask)) != 0) 281 #define F2FS_HAS_FEATURE(sbi, mask) __F2FS_HAS_FEATURE(sbi->raw_super, mask) 282 283 /* 284 * Default values for user and/or group using reserved blocks 285 */ 286 #define F2FS_DEF_RESUID 0 287 #define F2FS_DEF_RESGID 0 288 289 /* 290 * For checkpoint manager 291 */ 292 enum { 293 NAT_BITMAP, 294 SIT_BITMAP 295 }; 296 297 #define CP_UMOUNT 0x00000001 298 #define CP_FASTBOOT 0x00000002 299 #define CP_SYNC 0x00000004 300 #define CP_RECOVERY 0x00000008 301 #define CP_DISCARD 0x00000010 302 #define CP_TRIMMED 0x00000020 303 #define CP_PAUSE 0x00000040 304 #define CP_RESIZE 0x00000080 305 306 #define DEF_MAX_DISCARD_REQUEST 8 /* issue 8 discards per round */ 307 #define DEF_MIN_DISCARD_ISSUE_TIME 50 /* 50 ms, if exists */ 308 #define DEF_MID_DISCARD_ISSUE_TIME 500 /* 500 ms, if device busy */ 309 #define DEF_MAX_DISCARD_ISSUE_TIME 60000 /* 60 s, if no candidates */ 310 #define DEF_DISCARD_URGENT_UTIL 80 /* do more discard over 80% */ 311 #define DEF_CP_INTERVAL 60 /* 60 secs */ 312 #define DEF_IDLE_INTERVAL 5 /* 5 secs */ 313 #define DEF_DISABLE_INTERVAL 5 /* 5 secs */ 314 #define DEF_DISABLE_QUICK_INTERVAL 1 /* 1 secs */ 315 #define DEF_UMOUNT_DISCARD_TIMEOUT 5 /* 5 secs */ 316 317 enum cp_time { 318 CP_TIME_START, /* begin */ 319 CP_TIME_LOCK, /* after cp_global_sem */ 320 CP_TIME_OP_LOCK, /* after block_operation */ 321 CP_TIME_MERGE_WRITE, /* after flush DATA/NODE/META */ 322 CP_TIME_FLUSH_NAT, /* after flush nat */ 323 CP_TIME_FLUSH_SIT, /* after flush sit */ 324 CP_TIME_SYNC_META, /* after sync_meta_pages */ 325 CP_TIME_SYNC_CP_META, /* after sync cp meta pages */ 326 CP_TIME_WAIT_DIRTY_META,/* after wait on dirty meta */ 327 CP_TIME_WAIT_CP_DATA, /* after wait on cp data */ 328 CP_TIME_FLUSH_DEVICE, /* after flush device cache */ 329 CP_TIME_WAIT_LAST_CP, /* after wait on last cp pack */ 330 CP_TIME_END, /* after unblock_operation */ 331 CP_TIME_MAX, 332 }; 333 334 /* time cost stats of checkpoint */ 335 struct cp_stats { 336 ktime_t times[CP_TIME_MAX]; 337 }; 338 339 struct cp_control { 340 int reason; 341 __u64 trim_start; 342 __u64 trim_end; 343 __u64 trim_minlen; 344 struct cp_stats stats; 345 }; 346 347 enum f2fs_cp_phase { 348 CP_PHASE_START_BLOCK_OPS, 349 CP_PHASE_FINISH_BLOCK_OPS, 350 CP_PHASE_FINISH_CHECKPOINT, 351 }; 352 353 /* 354 * indicate meta/data type 355 */ 356 enum { 357 META_CP, 358 META_NAT, 359 META_SIT, 360 META_SSA, 361 META_MAX, 362 META_POR, 363 DATA_GENERIC, /* check range only */ 364 DATA_GENERIC_ENHANCE, /* strong check on range and segment bitmap */ 365 DATA_GENERIC_ENHANCE_READ, /* 366 * strong check on range and segment 367 * bitmap but no warning due to race 368 * condition of read on truncated area 369 * by extent_cache 370 */ 371 DATA_GENERIC_ENHANCE_UPDATE, /* 372 * strong check on range and segment 373 * bitmap for update case 374 */ 375 META_GENERIC, 376 }; 377 378 /* for the list of ino */ 379 enum { 380 ORPHAN_INO, /* for orphan ino list */ 381 APPEND_INO, /* for append ino list */ 382 UPDATE_INO, /* for update ino list */ 383 TRANS_DIR_INO, /* for transactions dir ino list */ 384 XATTR_DIR_INO, /* for xattr updated dir ino list */ 385 FLUSH_INO, /* for multiple device flushing */ 386 MAX_INO_ENTRY, /* max. list */ 387 }; 388 389 struct ino_entry { 390 struct list_head list; /* list head */ 391 nid_t ino; /* inode number */ 392 unsigned int dirty_device; /* dirty device bitmap */ 393 }; 394 395 /* for the list of inodes to be GCed */ 396 struct inode_entry { 397 struct list_head list; /* list head */ 398 struct inode *inode; /* vfs inode pointer */ 399 }; 400 401 struct fsync_node_entry { 402 struct list_head list; /* list head */ 403 struct folio *folio; /* warm node folio pointer */ 404 unsigned int seq_id; /* sequence id */ 405 }; 406 407 struct ckpt_req { 408 struct completion wait; /* completion for checkpoint done */ 409 struct llist_node llnode; /* llist_node to be linked in wait queue */ 410 int ret; /* return code of checkpoint */ 411 union { 412 ktime_t queue_time; /* request queued time */ 413 ktime_t delta_time; /* time in queue */ 414 }; 415 }; 416 417 struct ckpt_req_control { 418 struct task_struct *f2fs_issue_ckpt; /* checkpoint task */ 419 int ckpt_thread_ioprio; /* checkpoint merge thread ioprio */ 420 wait_queue_head_t ckpt_wait_queue; /* waiting queue for wake-up */ 421 atomic_t issued_ckpt; /* # of actually issued ckpts */ 422 atomic_t total_ckpt; /* # of total ckpts */ 423 atomic_t queued_ckpt; /* # of queued ckpts */ 424 struct llist_head issue_list; /* list for command issue */ 425 spinlock_t stat_lock; /* lock for below checkpoint time stats */ 426 unsigned int cur_time; /* cur wait time in msec for currently issued checkpoint */ 427 unsigned int peak_time; /* peak wait time in msec until now */ 428 }; 429 430 /* a time threshold that checkpoint was blocked for, unit: ms */ 431 #define CP_LONG_LATENCY_THRESHOLD 5000 432 433 /* for the bitmap indicate blocks to be discarded */ 434 struct discard_entry { 435 struct list_head list; /* list head */ 436 block_t start_blkaddr; /* start blockaddr of current segment */ 437 unsigned char discard_map[SIT_VBLOCK_MAP_SIZE]; /* segment discard bitmap */ 438 }; 439 440 /* minimum discard granularity, unit: block count */ 441 #define MIN_DISCARD_GRANULARITY 1 442 /* default discard granularity of inner discard thread, unit: block count */ 443 #define DEFAULT_DISCARD_GRANULARITY 16 444 /* default maximum discard granularity of ordered discard, unit: block count */ 445 #define DEFAULT_MAX_ORDERED_DISCARD_GRANULARITY 16 446 /* default interval of periodical discard submission */ 447 #define DEFAULT_DISCARD_INTERVAL (msecs_to_jiffies(20)) 448 449 /* max discard pend list number */ 450 #define MAX_PLIST_NUM 512 451 #define plist_idx(blk_num) ((blk_num) >= MAX_PLIST_NUM ? \ 452 (MAX_PLIST_NUM - 1) : ((blk_num) - 1)) 453 454 enum { 455 D_PREP, /* initial */ 456 D_PARTIAL, /* partially submitted */ 457 D_SUBMIT, /* all submitted */ 458 D_DONE, /* finished */ 459 }; 460 461 struct discard_info { 462 block_t lstart; /* logical start address */ 463 block_t len; /* length */ 464 block_t start; /* actual start address in dev */ 465 }; 466 467 struct discard_cmd { 468 struct rb_node rb_node; /* rb node located in rb-tree */ 469 struct discard_info di; /* discard info */ 470 struct list_head list; /* command list */ 471 struct completion wait; /* completion */ 472 struct block_device *bdev; /* bdev */ 473 unsigned short ref; /* reference count */ 474 unsigned char state; /* state */ 475 unsigned char queued; /* queued discard */ 476 int error; /* bio error */ 477 spinlock_t lock; /* for state/bio_ref updating */ 478 unsigned short bio_ref; /* bio reference count */ 479 }; 480 481 enum { 482 DPOLICY_BG, 483 DPOLICY_FORCE, 484 DPOLICY_FSTRIM, 485 DPOLICY_UMOUNT, 486 MAX_DPOLICY, 487 }; 488 489 enum { 490 DPOLICY_IO_AWARE_DISABLE, /* force to not be aware of IO */ 491 DPOLICY_IO_AWARE_ENABLE, /* force to be aware of IO */ 492 DPOLICY_IO_AWARE_MAX, 493 }; 494 495 struct discard_policy { 496 int type; /* type of discard */ 497 unsigned int min_interval; /* used for candidates exist */ 498 unsigned int mid_interval; /* used for device busy */ 499 unsigned int max_interval; /* used for candidates not exist */ 500 unsigned int max_requests; /* # of discards issued per round */ 501 unsigned int io_aware_gran; /* minimum granularity discard not be aware of I/O */ 502 bool io_aware; /* issue discard in idle time */ 503 bool sync; /* submit discard with REQ_SYNC flag */ 504 bool ordered; /* issue discard by lba order */ 505 bool timeout; /* discard timeout for put_super */ 506 unsigned int granularity; /* discard granularity */ 507 }; 508 509 struct discard_cmd_control { 510 struct task_struct *f2fs_issue_discard; /* discard thread */ 511 struct list_head entry_list; /* 4KB discard entry list */ 512 struct list_head pend_list[MAX_PLIST_NUM];/* store pending entries */ 513 struct list_head wait_list; /* store on-flushing entries */ 514 struct list_head fstrim_list; /* in-flight discard from fstrim */ 515 wait_queue_head_t discard_wait_queue; /* waiting queue for wake-up */ 516 struct mutex cmd_lock; 517 unsigned int nr_discards; /* # of discards in the list */ 518 unsigned int max_discards; /* max. discards to be issued */ 519 unsigned int max_discard_request; /* max. discard request per round */ 520 unsigned int min_discard_issue_time; /* min. interval between discard issue */ 521 unsigned int mid_discard_issue_time; /* mid. interval between discard issue */ 522 unsigned int max_discard_issue_time; /* max. interval between discard issue */ 523 unsigned int discard_io_aware_gran; /* minimum discard granularity not be aware of I/O */ 524 unsigned int discard_urgent_util; /* utilization which issue discard proactively */ 525 unsigned int discard_granularity; /* discard granularity */ 526 unsigned int max_ordered_discard; /* maximum discard granularity issued by lba order */ 527 unsigned int discard_io_aware; /* io_aware policy */ 528 unsigned int undiscard_blks; /* # of undiscard blocks */ 529 unsigned int next_pos; /* next discard position */ 530 atomic_t issued_discard; /* # of issued discard */ 531 atomic_t queued_discard; /* # of queued discard */ 532 atomic_t discard_cmd_cnt; /* # of cached cmd count */ 533 struct rb_root_cached root; /* root of discard rb-tree */ 534 bool rbtree_check; /* config for consistence check */ 535 bool discard_wake; /* to wake up discard thread */ 536 }; 537 538 /* for the list of fsync inodes, used only during recovery */ 539 struct fsync_inode_entry { 540 struct list_head list; /* list head */ 541 struct inode *inode; /* vfs inode pointer */ 542 block_t blkaddr; /* block address locating the last fsync */ 543 block_t last_dentry; /* block address locating the last dentry */ 544 }; 545 546 #define nats_in_cursum(jnl) (le16_to_cpu((jnl)->n_nats)) 547 #define sits_in_cursum(jnl) (le16_to_cpu((jnl)->n_sits)) 548 549 #define nat_in_journal(jnl, i) \ 550 (((struct nat_journal_entry *)(jnl)->nat_j.entries)[i].ne) 551 #define nid_in_journal(jnl, i) \ 552 (((struct nat_journal_entry *)(jnl)->nat_j.entries)[i].nid) 553 #define sit_in_journal(jnl, i) \ 554 (((struct sit_journal_entry *)(jnl)->sit_j.entries)[i].se) 555 #define segno_in_journal(jnl, i) \ 556 (((struct sit_journal_entry *)(jnl)->sit_j.entries)[i].segno) 557 558 #define sum_entries(sum) ((struct f2fs_summary *)(sum)) 559 #define sum_journal(sbi, sum) \ 560 ((struct f2fs_journal *)((char *)(sum) + \ 561 ((sbi)->entries_in_sum * sizeof(struct f2fs_summary)))) 562 #define sum_footer(sbi, sum) \ 563 ((struct summary_footer *)((char *)(sum) + (sbi)->sum_blocksize - \ 564 sizeof(struct summary_footer))) 565 566 #define MAX_NAT_JENTRIES(sbi, jnl) ((sbi)->nat_journal_entries - nats_in_cursum(jnl)) 567 #define MAX_SIT_JENTRIES(sbi, jnl) ((sbi)->sit_journal_entries - sits_in_cursum(jnl)) 568 569 static inline int update_nats_in_cursum(struct f2fs_journal *journal, int i) 570 { 571 int before = nats_in_cursum(journal); 572 573 journal->n_nats = cpu_to_le16(before + i); 574 return before; 575 } 576 577 static inline int update_sits_in_cursum(struct f2fs_journal *journal, int i) 578 { 579 int before = sits_in_cursum(journal); 580 581 journal->n_sits = cpu_to_le16(before + i); 582 return before; 583 } 584 585 /* for inline stuff */ 586 #define DEF_INLINE_RESERVED_SIZE 1 587 static inline int get_extra_isize(struct inode *inode); 588 static inline int get_inline_xattr_addrs(struct inode *inode); 589 #define MAX_INLINE_DATA(inode) (sizeof(__le32) * \ 590 (CUR_ADDRS_PER_INODE(inode) - \ 591 get_inline_xattr_addrs(inode) - \ 592 DEF_INLINE_RESERVED_SIZE)) 593 594 /* for inline dir */ 595 #define NR_INLINE_DENTRY(inode) (MAX_INLINE_DATA(inode) * BITS_PER_BYTE / \ 596 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \ 597 BITS_PER_BYTE + 1)) 598 #define INLINE_DENTRY_BITMAP_SIZE(inode) \ 599 DIV_ROUND_UP(NR_INLINE_DENTRY(inode), BITS_PER_BYTE) 600 #define INLINE_RESERVED_SIZE(inode) (MAX_INLINE_DATA(inode) - \ 601 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \ 602 NR_INLINE_DENTRY(inode) + \ 603 INLINE_DENTRY_BITMAP_SIZE(inode))) 604 605 /* 606 * For INODE and NODE manager 607 */ 608 /* for directory operations */ 609 610 struct f2fs_filename { 611 /* 612 * The filename the user specified. This is NULL for some 613 * filesystem-internal operations, e.g. converting an inline directory 614 * to a non-inline one, or roll-forward recovering an encrypted dentry. 615 */ 616 const struct qstr *usr_fname; 617 618 /* 619 * The on-disk filename. For encrypted directories, this is encrypted. 620 * This may be NULL for lookups in an encrypted dir without the key. 621 */ 622 struct fscrypt_str disk_name; 623 624 /* The dirhash of this filename */ 625 f2fs_hash_t hash; 626 627 #ifdef CONFIG_FS_ENCRYPTION 628 /* 629 * For lookups in encrypted directories: either the buffer backing 630 * disk_name, or a buffer that holds the decoded no-key name. 631 */ 632 struct fscrypt_str crypto_buf; 633 #endif 634 #if IS_ENABLED(CONFIG_UNICODE) 635 /* 636 * For casefolded directories: the casefolded name, but it's left NULL 637 * if the original name is not valid Unicode, if the original name is 638 * "." or "..", if the directory is both casefolded and encrypted and 639 * its encryption key is unavailable, or if the filesystem is doing an 640 * internal operation where usr_fname is also NULL. In all these cases 641 * we fall back to treating the name as an opaque byte sequence. 642 */ 643 struct qstr cf_name; 644 #endif 645 }; 646 647 struct f2fs_dentry_ptr { 648 struct inode *inode; 649 void *bitmap; 650 struct f2fs_dir_entry *dentry; 651 __u8 (*filename)[F2FS_SLOT_LEN]; 652 int max; 653 int nr_bitmap; 654 }; 655 656 static inline void make_dentry_ptr_block(struct inode *inode, 657 struct f2fs_dentry_ptr *d, struct f2fs_dentry_block *t) 658 { 659 d->inode = inode; 660 d->max = NR_DENTRY_IN_BLOCK; 661 d->nr_bitmap = SIZE_OF_DENTRY_BITMAP; 662 d->bitmap = t->dentry_bitmap; 663 d->dentry = t->dentry; 664 d->filename = t->filename; 665 } 666 667 static inline void make_dentry_ptr_inline(struct inode *inode, 668 struct f2fs_dentry_ptr *d, void *t) 669 { 670 int entry_cnt = NR_INLINE_DENTRY(inode); 671 int bitmap_size = INLINE_DENTRY_BITMAP_SIZE(inode); 672 int reserved_size = INLINE_RESERVED_SIZE(inode); 673 674 d->inode = inode; 675 d->max = entry_cnt; 676 d->nr_bitmap = bitmap_size; 677 d->bitmap = t; 678 d->dentry = t + bitmap_size + reserved_size; 679 d->filename = t + bitmap_size + reserved_size + 680 SIZE_OF_DIR_ENTRY * entry_cnt; 681 } 682 683 /* 684 * XATTR_NODE_OFFSET stores xattrs to one node block per file keeping -1 685 * as its node offset to distinguish from index node blocks. 686 * But some bits are used to mark the node block. 687 */ 688 #define XATTR_NODE_OFFSET ((((unsigned int)-1) << OFFSET_BIT_SHIFT) \ 689 >> OFFSET_BIT_SHIFT) 690 enum { 691 ALLOC_NODE, /* allocate a new node page if needed */ 692 LOOKUP_NODE, /* look up a node without readahead */ 693 LOOKUP_NODE_RA, /* 694 * look up a node with readahead called 695 * by get_data_block. 696 */ 697 }; 698 699 #define DEFAULT_RETRY_IO_COUNT 8 /* maximum retry read IO or flush count */ 700 701 #define MAX_FLUSH_RETRY_COUNT 3 /* maximum flush retry count in f2fs_enable_checkpoint() */ 702 703 /* IO/non-IO congestion wait timeout value, default: 1 jiffies */ 704 #define DEFAULT_SCHEDULE_TIMEOUT 1 705 706 /* timeout value injected, default: 1000ms */ 707 #define DEFAULT_FAULT_TIMEOUT (msecs_to_jiffies(1000)) 708 709 /* maximum retry quota flush count */ 710 #define DEFAULT_RETRY_QUOTA_FLUSH_COUNT 8 711 712 /* maximum retry of EIO'ed page */ 713 #define MAX_RETRY_PAGE_EIO 100 714 715 #define F2FS_LINK_MAX 0xffffffff /* maximum link count per file */ 716 717 #define MAX_DIR_RA_PAGES 4 /* maximum ra pages of dir */ 718 719 /* dirty segments threshold for triggering CP */ 720 #define DEFAULT_DIRTY_THRESHOLD 4 721 722 #define RECOVERY_MAX_RA_BLOCKS BIO_MAX_VECS 723 #define RECOVERY_MIN_RA_BLOCKS 1 724 725 #define F2FS_ONSTACK_PAGES 16 /* nr of onstack pages */ 726 727 /* for in-memory extent cache entry */ 728 #define F2FS_MIN_EXTENT_LEN 64 /* minimum extent length */ 729 730 /* number of extent info in extent cache we try to shrink */ 731 #define READ_EXTENT_CACHE_SHRINK_NUMBER 128 732 733 /* number of age extent info in extent cache we try to shrink */ 734 #define AGE_EXTENT_CACHE_SHRINK_NUMBER 128 735 #define LAST_AGE_WEIGHT 30 736 #define SAME_AGE_REGION 1024 737 738 /* 739 * Define data block with age less than 1GB as hot data 740 * define data block with age less than 10GB but more than 1GB as warm data 741 */ 742 #define DEF_HOT_DATA_AGE_THRESHOLD 262144 743 #define DEF_WARM_DATA_AGE_THRESHOLD 2621440 744 745 /* default max read extent count per inode */ 746 #define DEF_MAX_READ_EXTENT_COUNT 10240 747 748 /* extent cache type */ 749 enum extent_type { 750 EX_READ, 751 EX_BLOCK_AGE, 752 NR_EXTENT_CACHES, 753 }; 754 755 /* 756 * Reserved value to mark invalid age extents, hence valid block range 757 * from 0 to ULLONG_MAX-1 758 */ 759 #define F2FS_EXTENT_AGE_INVALID ULLONG_MAX 760 761 struct extent_info { 762 unsigned int fofs; /* start offset in a file */ 763 unsigned int len; /* length of the extent */ 764 union { 765 /* read extent_cache */ 766 struct { 767 /* start block address of the extent */ 768 block_t blk; 769 #ifdef CONFIG_F2FS_FS_COMPRESSION 770 /* physical extent length of compressed blocks */ 771 unsigned int c_len; 772 #endif 773 }; 774 /* block age extent_cache */ 775 struct { 776 /* block age of the extent */ 777 unsigned long long age; 778 /* last total blocks allocated */ 779 unsigned long long last_blocks; 780 }; 781 }; 782 }; 783 784 struct extent_node { 785 struct rb_node rb_node; /* rb node located in rb-tree */ 786 struct extent_info ei; /* extent info */ 787 struct list_head list; /* node in global extent list of sbi */ 788 struct extent_tree *et; /* extent tree pointer */ 789 }; 790 791 struct extent_tree { 792 nid_t ino; /* inode number */ 793 enum extent_type type; /* keep the extent tree type */ 794 struct rb_root_cached root; /* root of extent info rb-tree */ 795 struct extent_node *cached_en; /* recently accessed extent node */ 796 struct list_head list; /* to be used by sbi->zombie_list */ 797 rwlock_t lock; /* protect extent info rb-tree */ 798 atomic_t node_cnt; /* # of extent node in rb-tree*/ 799 bool largest_updated; /* largest extent updated */ 800 struct extent_info largest; /* largest cached extent for EX_READ */ 801 }; 802 803 struct extent_tree_info { 804 struct radix_tree_root extent_tree_root;/* cache extent cache entries */ 805 struct mutex extent_tree_lock; /* locking extent radix tree */ 806 struct list_head extent_list; /* lru list for shrinker */ 807 spinlock_t extent_lock; /* locking extent lru list */ 808 atomic_t total_ext_tree; /* extent tree count */ 809 struct list_head zombie_list; /* extent zombie tree list */ 810 atomic_t total_zombie_tree; /* extent zombie tree count */ 811 atomic_t total_ext_node; /* extent info count */ 812 }; 813 814 /* 815 * State of block returned by f2fs_map_blocks. 816 */ 817 #define F2FS_MAP_NEW (1U << 0) 818 #define F2FS_MAP_MAPPED (1U << 1) 819 #define F2FS_MAP_DELALLOC (1U << 2) 820 #define F2FS_MAP_FLAGS (F2FS_MAP_NEW | F2FS_MAP_MAPPED |\ 821 F2FS_MAP_DELALLOC) 822 823 struct f2fs_map_blocks { 824 struct block_device *m_bdev; /* for multi-device dio */ 825 block_t m_pblk; 826 block_t m_lblk; 827 unsigned int m_len; 828 unsigned int m_flags; 829 unsigned long m_last_pblk; /* last allocated block, only used for DIO in LFS mode */ 830 pgoff_t *m_next_pgofs; /* point next possible non-hole pgofs */ 831 pgoff_t *m_next_extent; /* point to next possible extent */ 832 int m_seg_type; 833 bool m_may_create; /* indicate it is from write path */ 834 bool m_multidev_dio; /* indicate it allows multi-device dio */ 835 }; 836 837 /* for flag in get_data_block */ 838 enum { 839 F2FS_GET_BLOCK_DEFAULT, 840 F2FS_GET_BLOCK_FIEMAP, 841 F2FS_GET_BLOCK_BMAP, 842 F2FS_GET_BLOCK_DIO, 843 F2FS_GET_BLOCK_PRE_DIO, 844 F2FS_GET_BLOCK_PRE_AIO, 845 F2FS_GET_BLOCK_PRECACHE, 846 }; 847 848 /* 849 * i_advise uses FADVISE_XXX_BIT. We can add additional hints later. 850 */ 851 #define FADVISE_COLD_BIT 0x01 852 #define FADVISE_LOST_PINO_BIT 0x02 853 #define FADVISE_ENCRYPT_BIT 0x04 854 #define FADVISE_ENC_NAME_BIT 0x08 855 #define FADVISE_KEEP_SIZE_BIT 0x10 856 #define FADVISE_HOT_BIT 0x20 857 #define FADVISE_VERITY_BIT 0x40 858 #define FADVISE_TRUNC_BIT 0x80 859 860 #define FADVISE_MODIFIABLE_BITS (FADVISE_COLD_BIT | FADVISE_HOT_BIT) 861 862 #define file_is_cold(inode) is_file(inode, FADVISE_COLD_BIT) 863 #define file_set_cold(inode) set_file(inode, FADVISE_COLD_BIT) 864 #define file_clear_cold(inode) clear_file(inode, FADVISE_COLD_BIT) 865 866 #define file_wrong_pino(inode) is_file(inode, FADVISE_LOST_PINO_BIT) 867 #define file_lost_pino(inode) set_file(inode, FADVISE_LOST_PINO_BIT) 868 #define file_got_pino(inode) clear_file(inode, FADVISE_LOST_PINO_BIT) 869 870 #define file_is_encrypt(inode) is_file(inode, FADVISE_ENCRYPT_BIT) 871 #define file_set_encrypt(inode) set_file(inode, FADVISE_ENCRYPT_BIT) 872 873 #define file_enc_name(inode) is_file(inode, FADVISE_ENC_NAME_BIT) 874 #define file_set_enc_name(inode) set_file(inode, FADVISE_ENC_NAME_BIT) 875 876 #define file_keep_isize(inode) is_file(inode, FADVISE_KEEP_SIZE_BIT) 877 #define file_set_keep_isize(inode) set_file(inode, FADVISE_KEEP_SIZE_BIT) 878 879 #define file_is_hot(inode) is_file(inode, FADVISE_HOT_BIT) 880 #define file_set_hot(inode) set_file(inode, FADVISE_HOT_BIT) 881 #define file_clear_hot(inode) clear_file(inode, FADVISE_HOT_BIT) 882 883 #define file_is_verity(inode) is_file(inode, FADVISE_VERITY_BIT) 884 #define file_set_verity(inode) set_file(inode, FADVISE_VERITY_BIT) 885 886 #define file_should_truncate(inode) is_file(inode, FADVISE_TRUNC_BIT) 887 #define file_need_truncate(inode) set_file(inode, FADVISE_TRUNC_BIT) 888 #define file_dont_truncate(inode) clear_file(inode, FADVISE_TRUNC_BIT) 889 890 #define DEF_DIR_LEVEL 0 891 892 /* used for f2fs_inode_info->flags */ 893 enum { 894 FI_NEW_INODE, /* indicate newly allocated inode */ 895 FI_DIRTY_INODE, /* indicate inode is dirty or not */ 896 FI_AUTO_RECOVER, /* indicate inode is recoverable */ 897 FI_DIRTY_DIR, /* indicate directory has dirty pages */ 898 FI_INC_LINK, /* need to increment i_nlink */ 899 FI_ACL_MODE, /* indicate acl mode */ 900 FI_NO_ALLOC, /* should not allocate any blocks */ 901 FI_FREE_NID, /* free allocated nide */ 902 FI_NO_EXTENT, /* not to use the extent cache */ 903 FI_INLINE_XATTR, /* used for inline xattr */ 904 FI_INLINE_DATA, /* used for inline data*/ 905 FI_INLINE_DENTRY, /* used for inline dentry */ 906 FI_APPEND_WRITE, /* inode has appended data */ 907 FI_UPDATE_WRITE, /* inode has in-place-update data */ 908 FI_NEED_IPU, /* used for ipu per file */ 909 FI_ATOMIC_FILE, /* indicate atomic file */ 910 FI_DATA_EXIST, /* indicate data exists */ 911 FI_SKIP_WRITES, /* should skip data page writeback */ 912 FI_OPU_WRITE, /* used for opu per file */ 913 FI_DIRTY_FILE, /* indicate regular/symlink has dirty pages */ 914 FI_PREALLOCATED_ALL, /* all blocks for write were preallocated */ 915 FI_HOT_DATA, /* indicate file is hot */ 916 FI_EXTRA_ATTR, /* indicate file has extra attribute */ 917 FI_PROJ_INHERIT, /* indicate file inherits projectid */ 918 FI_PIN_FILE, /* indicate file should not be gced */ 919 FI_VERITY_IN_PROGRESS, /* building fs-verity Merkle tree */ 920 FI_COMPRESSED_FILE, /* indicate file's data can be compressed */ 921 FI_COMPRESS_CORRUPT, /* indicate compressed cluster is corrupted */ 922 FI_MMAP_FILE, /* indicate file was mmapped */ 923 FI_ENABLE_COMPRESS, /* enable compression in "user" compression mode */ 924 FI_COMPRESS_RELEASED, /* compressed blocks were released */ 925 FI_ALIGNED_WRITE, /* enable aligned write */ 926 FI_COW_FILE, /* indicate COW file */ 927 FI_ATOMIC_COMMITTED, /* indicate atomic commit completed except disk sync */ 928 FI_ATOMIC_DIRTIED, /* indicate atomic file is dirtied */ 929 FI_ATOMIC_REPLACE, /* indicate atomic replace */ 930 FI_OPENED_FILE, /* indicate file has been opened */ 931 FI_DONATE_FINISHED, /* indicate page donation of file has been finished */ 932 FI_MAX, /* max flag, never be used */ 933 }; 934 935 struct f2fs_inode_info { 936 struct inode vfs_inode; /* serve a vfs inode */ 937 unsigned long i_flags; /* keep an inode flags for ioctl */ 938 unsigned char i_advise; /* use to give file attribute hints */ 939 unsigned char i_dir_level; /* use for dentry level for large dir */ 940 union { 941 unsigned int i_current_depth; /* only for directory depth */ 942 unsigned short i_gc_failures; /* for gc failure statistic */ 943 }; 944 unsigned int i_pino; /* parent inode number */ 945 umode_t i_acl_mode; /* keep file acl mode temporarily */ 946 947 /* Use below internally in f2fs*/ 948 unsigned long flags[BITS_TO_LONGS(FI_MAX)]; /* use to pass per-file flags */ 949 unsigned int ioprio_hint; /* hint for IO priority */ 950 struct f2fs_rwsem i_sem; /* protect fi info */ 951 atomic_t dirty_pages; /* # of dirty pages */ 952 f2fs_hash_t chash; /* hash value of given file name */ 953 unsigned int clevel; /* maximum level of given file name */ 954 struct task_struct *task; /* lookup and create consistency */ 955 struct task_struct *cp_task; /* separate cp/wb IO stats*/ 956 struct task_struct *wb_task; /* indicate inode is in context of writeback */ 957 nid_t i_xattr_nid; /* node id that contains xattrs */ 958 loff_t last_disk_size; /* lastly written file size */ 959 spinlock_t i_size_lock; /* protect last_disk_size */ 960 961 #ifdef CONFIG_QUOTA 962 struct dquot __rcu *i_dquot[MAXQUOTAS]; 963 964 /* quota space reservation, managed internally by quota code */ 965 qsize_t i_reserved_quota; 966 #endif 967 struct list_head dirty_list; /* dirty list for dirs and files */ 968 struct list_head gdirty_list; /* linked in global dirty list */ 969 970 /* linked in global inode list for cache donation */ 971 struct list_head gdonate_list; 972 pgoff_t donate_start, donate_end; /* inclusive */ 973 atomic_t open_count; /* # of open files */ 974 975 struct task_struct *atomic_write_task; /* store atomic write task */ 976 struct extent_tree *extent_tree[NR_EXTENT_CACHES]; 977 /* cached extent_tree entry */ 978 union { 979 struct inode *cow_inode; /* copy-on-write inode for atomic write */ 980 struct inode *atomic_inode; 981 /* point to atomic_inode, available only for cow_inode */ 982 }; 983 984 /* avoid racing between foreground op and gc */ 985 struct f2fs_rwsem i_gc_rwsem[2]; 986 struct f2fs_rwsem i_xattr_sem; /* avoid racing between reading and changing EAs */ 987 988 int i_extra_isize; /* size of extra space located in i_addr */ 989 kprojid_t i_projid; /* id for project quota */ 990 int i_inline_xattr_size; /* inline xattr size */ 991 struct timespec64 i_crtime; /* inode creation time */ 992 struct timespec64 i_disk_time[3];/* inode disk times */ 993 994 /* for file compress */ 995 atomic_t i_compr_blocks; /* # of compressed blocks */ 996 unsigned char i_compress_algorithm; /* algorithm type */ 997 unsigned char i_log_cluster_size; /* log of cluster size */ 998 unsigned char i_compress_level; /* compress level (lz4hc,zstd) */ 999 unsigned char i_compress_flag; /* compress flag */ 1000 unsigned int i_cluster_size; /* cluster size */ 1001 atomic_t writeback; /* count # of writeback thread */ 1002 1003 unsigned int atomic_write_cnt; 1004 loff_t original_i_size; /* original i_size before atomic write */ 1005 #ifdef CONFIG_FS_ENCRYPTION 1006 struct fscrypt_inode_info *i_crypt_info; /* filesystem encryption info */ 1007 #endif 1008 }; 1009 1010 static inline void get_read_extent_info(struct extent_info *ext, 1011 struct f2fs_extent *i_ext) 1012 { 1013 ext->fofs = le32_to_cpu(i_ext->fofs); 1014 ext->blk = le32_to_cpu(i_ext->blk); 1015 ext->len = le32_to_cpu(i_ext->len); 1016 } 1017 1018 static inline void set_raw_read_extent(struct extent_info *ext, 1019 struct f2fs_extent *i_ext) 1020 { 1021 i_ext->fofs = cpu_to_le32(ext->fofs); 1022 i_ext->blk = cpu_to_le32(ext->blk); 1023 i_ext->len = cpu_to_le32(ext->len); 1024 } 1025 1026 static inline bool __is_discard_mergeable(struct discard_info *back, 1027 struct discard_info *front, unsigned int max_len) 1028 { 1029 return (back->lstart + back->len == front->lstart) && 1030 (back->len + front->len <= max_len); 1031 } 1032 1033 static inline bool __is_discard_back_mergeable(struct discard_info *cur, 1034 struct discard_info *back, unsigned int max_len) 1035 { 1036 return __is_discard_mergeable(back, cur, max_len); 1037 } 1038 1039 static inline bool __is_discard_front_mergeable(struct discard_info *cur, 1040 struct discard_info *front, unsigned int max_len) 1041 { 1042 return __is_discard_mergeable(cur, front, max_len); 1043 } 1044 1045 /* 1046 * For free nid management 1047 */ 1048 enum nid_state { 1049 FREE_NID, /* newly added to free nid list */ 1050 PREALLOC_NID, /* it is preallocated */ 1051 MAX_NID_STATE, 1052 }; 1053 1054 enum nat_state { 1055 TOTAL_NAT, 1056 DIRTY_NAT, 1057 RECLAIMABLE_NAT, 1058 MAX_NAT_STATE, 1059 }; 1060 1061 struct f2fs_nm_info { 1062 block_t nat_blkaddr; /* base disk address of NAT */ 1063 nid_t max_nid; /* maximum possible node ids */ 1064 nid_t available_nids; /* # of available node ids */ 1065 nid_t next_scan_nid; /* the next nid to be scanned */ 1066 nid_t max_rf_node_blocks; /* max # of nodes for recovery */ 1067 unsigned int ram_thresh; /* control the memory footprint */ 1068 unsigned int ra_nid_pages; /* # of nid pages to be readaheaded */ 1069 unsigned int dirty_nats_ratio; /* control dirty nats ratio threshold */ 1070 1071 /* NAT cache management */ 1072 struct radix_tree_root nat_root;/* root of the nat entry cache */ 1073 struct radix_tree_root nat_set_root;/* root of the nat set cache */ 1074 struct f2fs_rwsem nat_tree_lock; /* protect nat entry tree */ 1075 struct list_head nat_entries; /* cached nat entry list (clean) */ 1076 spinlock_t nat_list_lock; /* protect clean nat entry list */ 1077 unsigned int nat_cnt[MAX_NAT_STATE]; /* the # of cached nat entries */ 1078 unsigned int nat_blocks; /* # of nat blocks */ 1079 1080 /* free node ids management */ 1081 struct radix_tree_root free_nid_root;/* root of the free_nid cache */ 1082 struct list_head free_nid_list; /* list for free nids excluding preallocated nids */ 1083 unsigned int nid_cnt[MAX_NID_STATE]; /* the number of free node id */ 1084 spinlock_t nid_list_lock; /* protect nid lists ops */ 1085 struct mutex build_lock; /* lock for build free nids */ 1086 unsigned char **free_nid_bitmap; 1087 unsigned char *nat_block_bitmap; 1088 unsigned short *free_nid_count; /* free nid count of NAT block */ 1089 1090 /* for checkpoint */ 1091 char *nat_bitmap; /* NAT bitmap pointer */ 1092 1093 unsigned int nat_bits_blocks; /* # of nat bits blocks */ 1094 unsigned char *nat_bits; /* NAT bits blocks */ 1095 unsigned char *full_nat_bits; /* full NAT pages */ 1096 unsigned char *empty_nat_bits; /* empty NAT pages */ 1097 #ifdef CONFIG_F2FS_CHECK_FS 1098 char *nat_bitmap_mir; /* NAT bitmap mirror */ 1099 #endif 1100 int bitmap_size; /* bitmap size */ 1101 }; 1102 1103 /* 1104 * this structure is used as one of function parameters. 1105 * all the information are dedicated to a given direct node block determined 1106 * by the data offset in a file. 1107 */ 1108 struct dnode_of_data { 1109 struct inode *inode; /* vfs inode pointer */ 1110 struct folio *inode_folio; /* its inode folio, NULL is possible */ 1111 struct folio *node_folio; /* cached direct node folio */ 1112 nid_t nid; /* node id of the direct node block */ 1113 unsigned int ofs_in_node; /* data offset in the node page */ 1114 bool inode_folio_locked; /* inode folio is locked or not */ 1115 bool node_changed; /* is node block changed */ 1116 char cur_level; /* level of hole node page */ 1117 char max_level; /* level of current page located */ 1118 block_t data_blkaddr; /* block address of the node block */ 1119 }; 1120 1121 static inline void set_new_dnode(struct dnode_of_data *dn, struct inode *inode, 1122 struct folio *ifolio, struct folio *nfolio, nid_t nid) 1123 { 1124 memset(dn, 0, sizeof(*dn)); 1125 dn->inode = inode; 1126 dn->inode_folio = ifolio; 1127 dn->node_folio = nfolio; 1128 dn->nid = nid; 1129 } 1130 1131 /* 1132 * For SIT manager 1133 * 1134 * By default, there are 6 active log areas across the whole main area. 1135 * When considering hot and cold data separation to reduce cleaning overhead, 1136 * we split 3 for data logs and 3 for node logs as hot, warm, and cold types, 1137 * respectively. 1138 * In the current design, you should not change the numbers intentionally. 1139 * Instead, as a mount option such as active_logs=x, you can use 2, 4, and 6 1140 * logs individually according to the underlying devices. (default: 6) 1141 * Just in case, on-disk layout covers maximum 16 logs that consist of 8 for 1142 * data and 8 for node logs. 1143 */ 1144 #define NR_CURSEG_DATA_TYPE (3) 1145 #define NR_CURSEG_NODE_TYPE (3) 1146 #define NR_CURSEG_INMEM_TYPE (2) 1147 #define NR_CURSEG_RO_TYPE (2) 1148 #define NR_CURSEG_PERSIST_TYPE (NR_CURSEG_DATA_TYPE + NR_CURSEG_NODE_TYPE) 1149 #define NR_CURSEG_TYPE (NR_CURSEG_INMEM_TYPE + NR_CURSEG_PERSIST_TYPE) 1150 1151 enum log_type { 1152 CURSEG_HOT_DATA = 0, /* directory entry blocks */ 1153 CURSEG_WARM_DATA, /* data blocks */ 1154 CURSEG_COLD_DATA, /* multimedia or GCed data blocks */ 1155 CURSEG_HOT_NODE, /* direct node blocks of directory files */ 1156 CURSEG_WARM_NODE, /* direct node blocks of normal files */ 1157 CURSEG_COLD_NODE, /* indirect node blocks */ 1158 NR_PERSISTENT_LOG, /* number of persistent log */ 1159 CURSEG_COLD_DATA_PINNED = NR_PERSISTENT_LOG, 1160 /* pinned file that needs consecutive block address */ 1161 CURSEG_ALL_DATA_ATGC, /* SSR alloctor in hot/warm/cold data area */ 1162 NO_CHECK_TYPE, /* number of persistent & inmem log */ 1163 }; 1164 1165 struct flush_cmd { 1166 struct completion wait; 1167 struct llist_node llnode; 1168 nid_t ino; 1169 int ret; 1170 }; 1171 1172 struct flush_cmd_control { 1173 struct task_struct *f2fs_issue_flush; /* flush thread */ 1174 wait_queue_head_t flush_wait_queue; /* waiting queue for wake-up */ 1175 atomic_t issued_flush; /* # of issued flushes */ 1176 atomic_t queued_flush; /* # of queued flushes */ 1177 struct llist_head issue_list; /* list for command issue */ 1178 struct llist_node *dispatch_list; /* list for command dispatch */ 1179 }; 1180 1181 struct f2fs_sm_info { 1182 struct sit_info *sit_info; /* whole segment information */ 1183 struct free_segmap_info *free_info; /* free segment information */ 1184 struct dirty_seglist_info *dirty_info; /* dirty segment information */ 1185 struct curseg_info *curseg_array; /* active segment information */ 1186 1187 struct f2fs_rwsem curseg_lock; /* for preventing curseg change */ 1188 1189 block_t seg0_blkaddr; /* block address of 0'th segment */ 1190 block_t main_blkaddr; /* start block address of main area */ 1191 block_t ssa_blkaddr; /* start block address of SSA area */ 1192 1193 unsigned int segment_count; /* total # of segments */ 1194 unsigned int main_segments; /* # of segments in main area */ 1195 unsigned int reserved_segments; /* # of reserved segments */ 1196 unsigned int ovp_segments; /* # of overprovision segments */ 1197 1198 /* a threshold to reclaim prefree segments */ 1199 unsigned int rec_prefree_segments; 1200 1201 struct list_head sit_entry_set; /* sit entry set list */ 1202 1203 unsigned int ipu_policy; /* in-place-update policy */ 1204 unsigned int min_ipu_util; /* in-place-update threshold */ 1205 unsigned int min_fsync_blocks; /* threshold for fsync */ 1206 unsigned int min_seq_blocks; /* threshold for sequential blocks */ 1207 unsigned int min_hot_blocks; /* threshold for hot block allocation */ 1208 unsigned int min_ssr_sections; /* threshold to trigger SSR allocation */ 1209 1210 /* for flush command control */ 1211 struct flush_cmd_control *fcc_info; 1212 1213 /* for discard command control */ 1214 struct discard_cmd_control *dcc_info; 1215 }; 1216 1217 /* 1218 * For superblock 1219 */ 1220 /* 1221 * COUNT_TYPE for monitoring 1222 * 1223 * f2fs monitors the number of several block types such as on-writeback, 1224 * dirty dentry blocks, dirty node blocks, and dirty meta blocks. 1225 */ 1226 #define WB_DATA_TYPE(folio, f) \ 1227 (f || f2fs_is_cp_guaranteed(folio) ? F2FS_WB_CP_DATA : F2FS_WB_DATA) 1228 enum count_type { 1229 F2FS_DIRTY_DENTS, 1230 F2FS_DIRTY_DATA, 1231 F2FS_DIRTY_QDATA, 1232 F2FS_DIRTY_NODES, 1233 F2FS_DIRTY_META, 1234 F2FS_DIRTY_IMETA, 1235 F2FS_WB_CP_DATA, 1236 F2FS_WB_DATA, 1237 F2FS_RD_DATA, 1238 F2FS_RD_NODE, 1239 F2FS_RD_META, 1240 F2FS_DIO_WRITE, 1241 F2FS_DIO_READ, 1242 F2FS_SKIPPED_WRITE, /* skip or fail during f2fs_enable_checkpoint() */ 1243 NR_COUNT_TYPE, 1244 }; 1245 1246 /* 1247 * The below are the page types of bios used in submit_bio(). 1248 * The available types are: 1249 * DATA User data pages. It operates as async mode. 1250 * NODE Node pages. It operates as async mode. 1251 * META FS metadata pages such as SIT, NAT, CP. 1252 * NR_PAGE_TYPE The number of page types. 1253 * META_FLUSH Make sure the previous pages are written 1254 * with waiting the bio's completion 1255 * ... Only can be used with META. 1256 */ 1257 #define PAGE_TYPE_OF_BIO(type) ((type) > META ? META : (type)) 1258 #define PAGE_TYPE_ON_MAIN(type) ((type) == DATA || (type) == NODE) 1259 enum page_type { 1260 DATA = 0, 1261 NODE = 1, /* should not change this */ 1262 META, 1263 NR_PAGE_TYPE, 1264 META_FLUSH, 1265 IPU, /* the below types are used by tracepoints only. */ 1266 OPU, 1267 }; 1268 1269 enum temp_type { 1270 HOT = 0, /* must be zero for meta bio */ 1271 WARM, 1272 COLD, 1273 NR_TEMP_TYPE, 1274 }; 1275 1276 enum need_lock_type { 1277 LOCK_REQ = 0, 1278 LOCK_DONE, 1279 LOCK_RETRY, 1280 }; 1281 1282 enum cp_reason_type { 1283 CP_NO_NEEDED, 1284 CP_NON_REGULAR, 1285 CP_COMPRESSED, 1286 CP_HARDLINK, 1287 CP_SB_NEED_CP, 1288 CP_WRONG_PINO, 1289 CP_NO_SPC_ROLL, 1290 CP_NODE_NEED_CP, 1291 CP_FASTBOOT_MODE, 1292 CP_SPEC_LOG_NUM, 1293 CP_RECOVER_DIR, 1294 CP_XATTR_DIR, 1295 }; 1296 1297 enum iostat_type { 1298 /* WRITE IO */ 1299 APP_DIRECT_IO, /* app direct write IOs */ 1300 APP_BUFFERED_IO, /* app buffered write IOs */ 1301 APP_WRITE_IO, /* app write IOs */ 1302 APP_MAPPED_IO, /* app mapped IOs */ 1303 APP_BUFFERED_CDATA_IO, /* app buffered write IOs on compressed file */ 1304 APP_MAPPED_CDATA_IO, /* app mapped write IOs on compressed file */ 1305 FS_DATA_IO, /* data IOs from kworker/fsync/reclaimer */ 1306 FS_CDATA_IO, /* data IOs from kworker/fsync/reclaimer on compressed file */ 1307 FS_NODE_IO, /* node IOs from kworker/fsync/reclaimer */ 1308 FS_META_IO, /* meta IOs from kworker/reclaimer */ 1309 FS_GC_DATA_IO, /* data IOs from forground gc */ 1310 FS_GC_NODE_IO, /* node IOs from forground gc */ 1311 FS_CP_DATA_IO, /* data IOs from checkpoint */ 1312 FS_CP_NODE_IO, /* node IOs from checkpoint */ 1313 FS_CP_META_IO, /* meta IOs from checkpoint */ 1314 1315 /* READ IO */ 1316 APP_DIRECT_READ_IO, /* app direct read IOs */ 1317 APP_BUFFERED_READ_IO, /* app buffered read IOs */ 1318 APP_READ_IO, /* app read IOs */ 1319 APP_MAPPED_READ_IO, /* app mapped read IOs */ 1320 APP_BUFFERED_CDATA_READ_IO, /* app buffered read IOs on compressed file */ 1321 APP_MAPPED_CDATA_READ_IO, /* app mapped read IOs on compressed file */ 1322 FS_DATA_READ_IO, /* data read IOs */ 1323 FS_GDATA_READ_IO, /* data read IOs from background gc */ 1324 FS_CDATA_READ_IO, /* compressed data read IOs */ 1325 FS_NODE_READ_IO, /* node read IOs */ 1326 FS_META_READ_IO, /* meta read IOs */ 1327 1328 /* other */ 1329 FS_DISCARD_IO, /* discard */ 1330 FS_FLUSH_IO, /* flush */ 1331 FS_ZONE_RESET_IO, /* zone reset */ 1332 NR_IO_TYPE, 1333 }; 1334 1335 struct f2fs_io_info { 1336 struct f2fs_sb_info *sbi; /* f2fs_sb_info pointer */ 1337 nid_t ino; /* inode number */ 1338 enum page_type type; /* contains DATA/NODE/META/META_FLUSH */ 1339 enum temp_type temp; /* contains HOT/WARM/COLD */ 1340 enum req_op op; /* contains REQ_OP_ */ 1341 blk_opf_t op_flags; /* req_flag_bits */ 1342 block_t new_blkaddr; /* new block address to be written */ 1343 block_t old_blkaddr; /* old block address before Cow */ 1344 union { 1345 struct page *page; /* page to be written */ 1346 struct folio *folio; 1347 }; 1348 struct page *encrypted_page; /* encrypted page */ 1349 struct page *compressed_page; /* compressed page */ 1350 struct list_head list; /* serialize IOs */ 1351 unsigned int compr_blocks; /* # of compressed block addresses */ 1352 unsigned int need_lock:8; /* indicate we need to lock cp_rwsem */ 1353 unsigned int version:8; /* version of the node */ 1354 unsigned int submitted:1; /* indicate IO submission */ 1355 unsigned int in_list:1; /* indicate fio is in io_list */ 1356 unsigned int is_por:1; /* indicate IO is from recovery or not */ 1357 unsigned int encrypted:1; /* indicate file is encrypted */ 1358 unsigned int meta_gc:1; /* require meta inode GC */ 1359 enum iostat_type io_type; /* io type */ 1360 struct writeback_control *io_wbc; /* writeback control */ 1361 struct bio **bio; /* bio for ipu */ 1362 sector_t *last_block; /* last block number in bio */ 1363 }; 1364 1365 struct bio_entry { 1366 struct bio *bio; 1367 struct list_head list; 1368 }; 1369 1370 #define is_read_io(rw) ((rw) == READ) 1371 struct f2fs_bio_info { 1372 struct f2fs_sb_info *sbi; /* f2fs superblock */ 1373 struct bio *bio; /* bios to merge */ 1374 sector_t last_block_in_bio; /* last block number */ 1375 struct f2fs_io_info fio; /* store buffered io info. */ 1376 #ifdef CONFIG_BLK_DEV_ZONED 1377 struct completion zone_wait; /* condition value for the previous open zone to close */ 1378 struct bio *zone_pending_bio; /* pending bio for the previous zone */ 1379 void *bi_private; /* previous bi_private for pending bio */ 1380 #endif 1381 struct f2fs_rwsem io_rwsem; /* blocking op for bio */ 1382 spinlock_t io_lock; /* serialize DATA/NODE IOs */ 1383 struct list_head io_list; /* track fios */ 1384 struct list_head bio_list; /* bio entry list head */ 1385 struct f2fs_rwsem bio_list_lock; /* lock to protect bio entry list */ 1386 }; 1387 1388 #define FDEV(i) (sbi->devs[i]) 1389 #define RDEV(i) (raw_super->devs[i]) 1390 struct f2fs_dev_info { 1391 struct file *bdev_file; 1392 struct block_device *bdev; 1393 char path[MAX_PATH_LEN + 1]; 1394 unsigned int total_segments; 1395 block_t start_blk; 1396 block_t end_blk; 1397 #ifdef CONFIG_BLK_DEV_ZONED 1398 unsigned int nr_blkz; /* Total number of zones */ 1399 unsigned long *blkz_seq; /* Bitmap indicating sequential zones */ 1400 #endif 1401 }; 1402 1403 enum inode_type { 1404 DIR_INODE, /* for dirty dir inode */ 1405 FILE_INODE, /* for dirty regular/symlink inode */ 1406 DIRTY_META, /* for all dirtied inode metadata */ 1407 DONATE_INODE, /* for all inode to donate pages */ 1408 NR_INODE_TYPE, 1409 }; 1410 1411 /* for inner inode cache management */ 1412 struct inode_management { 1413 struct radix_tree_root ino_root; /* ino entry array */ 1414 spinlock_t ino_lock; /* for ino entry lock */ 1415 struct list_head ino_list; /* inode list head */ 1416 unsigned long ino_num; /* number of entries */ 1417 }; 1418 1419 /* for GC_AT */ 1420 struct atgc_management { 1421 bool atgc_enabled; /* ATGC is enabled or not */ 1422 struct rb_root_cached root; /* root of victim rb-tree */ 1423 struct list_head victim_list; /* linked with all victim entries */ 1424 unsigned int victim_count; /* victim count in rb-tree */ 1425 unsigned int candidate_ratio; /* candidate ratio */ 1426 unsigned int max_candidate_count; /* max candidate count */ 1427 unsigned int age_weight; /* age weight, vblock_weight = 100 - age_weight */ 1428 unsigned long long age_threshold; /* age threshold */ 1429 }; 1430 1431 struct f2fs_time_stat { 1432 unsigned long long total_time; /* total wall clock time */ 1433 #ifdef CONFIG_64BIT 1434 unsigned long long running_time; /* running time */ 1435 #endif 1436 #if defined(CONFIG_SCHED_INFO) && defined(CONFIG_SCHEDSTATS) 1437 unsigned long long runnable_time; /* runnable(including preempted) time */ 1438 #endif 1439 #ifdef CONFIG_TASK_DELAY_ACCT 1440 unsigned long long io_sleep_time; /* IO sleep time */ 1441 #endif 1442 }; 1443 1444 struct f2fs_lock_context { 1445 struct f2fs_time_stat ts; 1446 int orig_nice; 1447 int new_nice; 1448 bool lock_trace; 1449 bool need_restore; 1450 }; 1451 1452 struct f2fs_gc_control { 1453 unsigned int victim_segno; /* target victim segment number */ 1454 int init_gc_type; /* FG_GC or BG_GC */ 1455 bool no_bg_gc; /* check the space and stop bg_gc */ 1456 bool should_migrate_blocks; /* should migrate blocks */ 1457 bool err_gc_skipped; /* return EAGAIN if GC skipped */ 1458 bool one_time; /* require one time GC in one migration unit */ 1459 unsigned int nr_free_secs; /* # of free sections to do GC */ 1460 struct f2fs_lock_context lc; /* lock context for gc_lock */ 1461 }; 1462 1463 /* 1464 * For s_flag in struct f2fs_sb_info 1465 * Modification on enum should be synchronized with s_flag array 1466 */ 1467 enum { 1468 SBI_IS_DIRTY, /* dirty flag for checkpoint */ 1469 SBI_IS_CLOSE, /* specify unmounting */ 1470 SBI_NEED_FSCK, /* need fsck.f2fs to fix */ 1471 SBI_POR_DOING, /* recovery is doing or not */ 1472 SBI_NEED_SB_WRITE, /* need to recover superblock */ 1473 SBI_NEED_CP, /* need to checkpoint */ 1474 SBI_IS_SHUTDOWN, /* shutdown by ioctl */ 1475 SBI_IS_RECOVERED, /* recovered orphan/data */ 1476 SBI_CP_DISABLED, /* CP was disabled last mount */ 1477 SBI_CP_DISABLED_QUICK, /* CP was disabled quickly */ 1478 SBI_QUOTA_NEED_FLUSH, /* need to flush quota info in CP */ 1479 SBI_QUOTA_SKIP_FLUSH, /* skip flushing quota in current CP */ 1480 SBI_QUOTA_NEED_REPAIR, /* quota file may be corrupted */ 1481 SBI_IS_RESIZEFS, /* resizefs is in process */ 1482 SBI_IS_FREEZING, /* freezefs is in process */ 1483 SBI_IS_WRITABLE, /* remove ro mountoption transiently */ 1484 SBI_ENABLE_CHECKPOINT, /* indicate it's during f2fs_enable_checkpoint() */ 1485 MAX_SBI_FLAG, 1486 }; 1487 1488 enum { 1489 CP_TIME, 1490 REQ_TIME, 1491 DISCARD_TIME, 1492 GC_TIME, 1493 DISABLE_TIME, 1494 UMOUNT_DISCARD_TIMEOUT, 1495 MAX_TIME, 1496 }; 1497 1498 /* Note that you need to keep synchronization with this gc_mode_names array */ 1499 enum { 1500 GC_NORMAL, 1501 GC_IDLE_CB, 1502 GC_IDLE_GREEDY, 1503 GC_IDLE_AT, 1504 GC_URGENT_HIGH, 1505 GC_URGENT_LOW, 1506 GC_URGENT_MID, 1507 MAX_GC_MODE, 1508 }; 1509 1510 enum { 1511 BGGC_MODE_ON, /* background gc is on */ 1512 BGGC_MODE_OFF, /* background gc is off */ 1513 BGGC_MODE_SYNC, /* 1514 * background gc is on, migrating blocks 1515 * like foreground gc 1516 */ 1517 }; 1518 1519 enum { 1520 FS_MODE_ADAPTIVE, /* use both lfs/ssr allocation */ 1521 FS_MODE_LFS, /* use lfs allocation only */ 1522 FS_MODE_FRAGMENT_SEG, /* segment fragmentation mode */ 1523 FS_MODE_FRAGMENT_BLK, /* block fragmentation mode */ 1524 }; 1525 1526 enum { 1527 ALLOC_MODE_DEFAULT, /* stay default */ 1528 ALLOC_MODE_REUSE, /* reuse segments as much as possible */ 1529 }; 1530 1531 enum fsync_mode { 1532 FSYNC_MODE_POSIX, /* fsync follows posix semantics */ 1533 FSYNC_MODE_STRICT, /* fsync behaves in line with ext4 */ 1534 FSYNC_MODE_NOBARRIER, /* fsync behaves nobarrier based on posix */ 1535 }; 1536 1537 enum { 1538 COMPR_MODE_FS, /* 1539 * automatically compress compression 1540 * enabled files 1541 */ 1542 COMPR_MODE_USER, /* 1543 * automatical compression is disabled. 1544 * user can control the file compression 1545 * using ioctls 1546 */ 1547 }; 1548 1549 enum { 1550 DISCARD_UNIT_BLOCK, /* basic discard unit is block */ 1551 DISCARD_UNIT_SEGMENT, /* basic discard unit is segment */ 1552 DISCARD_UNIT_SECTION, /* basic discard unit is section */ 1553 }; 1554 1555 enum { 1556 MEMORY_MODE_NORMAL, /* memory mode for normal devices */ 1557 MEMORY_MODE_LOW, /* memory mode for low memory devices */ 1558 }; 1559 1560 enum errors_option { 1561 MOUNT_ERRORS_READONLY, /* remount fs ro on errors */ 1562 MOUNT_ERRORS_CONTINUE, /* continue on errors */ 1563 MOUNT_ERRORS_PANIC, /* panic on errors */ 1564 }; 1565 1566 enum { 1567 BACKGROUND, 1568 FOREGROUND, 1569 MAX_CALL_TYPE, 1570 TOTAL_CALL = FOREGROUND, 1571 }; 1572 1573 enum f2fs_lookup_mode { 1574 LOOKUP_PERF, 1575 LOOKUP_COMPAT, 1576 LOOKUP_AUTO, 1577 }; 1578 1579 /* For node type in __get_node_folio() */ 1580 enum node_type { 1581 NODE_TYPE_REGULAR, 1582 NODE_TYPE_INODE, 1583 NODE_TYPE_XATTR, 1584 NODE_TYPE_NON_INODE, 1585 }; 1586 1587 /* a threshold of maximum elapsed time in critical region to print tracepoint */ 1588 #define MAX_LOCK_ELAPSED_TIME 500 1589 1590 #define F2FS_DEFAULT_TASK_PRIORITY (DEFAULT_PRIO) 1591 #define F2FS_CRITICAL_TASK_PRIORITY NICE_TO_PRIO(0) 1592 1593 static inline int f2fs_test_bit(unsigned int nr, char *addr); 1594 static inline void f2fs_set_bit(unsigned int nr, char *addr); 1595 static inline void f2fs_clear_bit(unsigned int nr, char *addr); 1596 1597 /* 1598 * Layout of f2fs page.private: 1599 * 1600 * Layout A: lowest bit should be 1 1601 * | bit0 = 1 | bit1 | bit2 | ... | bit MAX | private data .... | 1602 * bit 0 PAGE_PRIVATE_NOT_POINTER 1603 * bit 1 PAGE_PRIVATE_ONGOING_MIGRATION 1604 * bit 2 PAGE_PRIVATE_INLINE_INODE 1605 * bit 3 PAGE_PRIVATE_REF_RESOURCE 1606 * bit 4 PAGE_PRIVATE_ATOMIC_WRITE 1607 * bit 5- f2fs private data 1608 * 1609 * Layout B: lowest bit should be 0 1610 * page.private is a wrapped pointer. 1611 */ 1612 enum { 1613 PAGE_PRIVATE_NOT_POINTER, /* private contains non-pointer data */ 1614 PAGE_PRIVATE_ONGOING_MIGRATION, /* data page which is on-going migrating */ 1615 PAGE_PRIVATE_INLINE_INODE, /* inode page contains inline data */ 1616 PAGE_PRIVATE_REF_RESOURCE, /* dirty page has referenced resources */ 1617 PAGE_PRIVATE_ATOMIC_WRITE, /* data page from atomic write path */ 1618 PAGE_PRIVATE_MAX 1619 }; 1620 1621 /* For compression */ 1622 enum compress_algorithm_type { 1623 COMPRESS_LZO, 1624 COMPRESS_LZ4, 1625 COMPRESS_ZSTD, 1626 COMPRESS_LZORLE, 1627 COMPRESS_MAX, 1628 }; 1629 1630 enum compress_flag { 1631 COMPRESS_CHKSUM, 1632 COMPRESS_MAX_FLAG, 1633 }; 1634 1635 #define COMPRESS_WATERMARK 20 1636 #define COMPRESS_PERCENT 20 1637 1638 #define COMPRESS_DATA_RESERVED_SIZE 4 1639 struct compress_data { 1640 __le32 clen; /* compressed data size */ 1641 __le32 chksum; /* compressed data checksum */ 1642 __le32 reserved[COMPRESS_DATA_RESERVED_SIZE]; /* reserved */ 1643 u8 cdata[]; /* compressed data */ 1644 }; 1645 1646 #define COMPRESS_HEADER_SIZE (sizeof(struct compress_data)) 1647 1648 #define F2FS_COMPRESSED_PAGE_MAGIC 0xF5F2C000 1649 1650 #define F2FS_ZSTD_DEFAULT_CLEVEL 1 1651 1652 #define COMPRESS_LEVEL_OFFSET 8 1653 1654 /* compress context */ 1655 struct compress_ctx { 1656 struct inode *inode; /* inode the context belong to */ 1657 pgoff_t cluster_idx; /* cluster index number */ 1658 unsigned int cluster_size; /* page count in cluster */ 1659 unsigned int log_cluster_size; /* log of cluster size */ 1660 struct page **rpages; /* pages store raw data in cluster */ 1661 unsigned int nr_rpages; /* total page number in rpages */ 1662 struct page **cpages; /* pages store compressed data in cluster */ 1663 unsigned int nr_cpages; /* total page number in cpages */ 1664 unsigned int valid_nr_cpages; /* valid page number in cpages */ 1665 void *rbuf; /* virtual mapped address on rpages */ 1666 struct compress_data *cbuf; /* virtual mapped address on cpages */ 1667 size_t rlen; /* valid data length in rbuf */ 1668 size_t clen; /* valid data length in cbuf */ 1669 void *private; /* payload buffer for specified compression algorithm */ 1670 void *private2; /* extra payload buffer */ 1671 struct fsverity_info *vi; /* verity info if needed */ 1672 }; 1673 1674 /* compress context for write IO path */ 1675 struct compress_io_ctx { 1676 u32 magic; /* magic number to indicate page is compressed */ 1677 struct inode *inode; /* inode the context belong to */ 1678 struct page **rpages; /* pages store raw data in cluster */ 1679 unsigned int nr_rpages; /* total page number in rpages */ 1680 atomic_t pending_pages; /* in-flight compressed page count */ 1681 }; 1682 1683 /* Context for decompressing one cluster on the read IO path */ 1684 struct decompress_io_ctx { 1685 u32 magic; /* magic number to indicate page is compressed */ 1686 struct inode *inode; /* inode the context belong to */ 1687 struct f2fs_sb_info *sbi; /* f2fs_sb_info pointer */ 1688 pgoff_t cluster_idx; /* cluster index number */ 1689 unsigned int cluster_size; /* page count in cluster */ 1690 unsigned int log_cluster_size; /* log of cluster size */ 1691 struct page **rpages; /* pages store raw data in cluster */ 1692 unsigned int nr_rpages; /* total page number in rpages */ 1693 struct page **cpages; /* pages store compressed data in cluster */ 1694 unsigned int nr_cpages; /* total page number in cpages */ 1695 struct page **tpages; /* temp pages to pad holes in cluster */ 1696 void *rbuf; /* virtual mapped address on rpages */ 1697 struct compress_data *cbuf; /* virtual mapped address on cpages */ 1698 size_t rlen; /* valid data length in rbuf */ 1699 size_t clen; /* valid data length in cbuf */ 1700 1701 /* 1702 * The number of compressed pages remaining to be read in this cluster. 1703 * This is initially nr_cpages. It is decremented by 1 each time a page 1704 * has been read (or failed to be read). When it reaches 0, the cluster 1705 * is decompressed (or an error is reported). 1706 * 1707 * If an error occurs before all the pages have been submitted for I/O, 1708 * then this will never reach 0. In this case the I/O submitter is 1709 * responsible for calling f2fs_decompress_end_io() instead. 1710 */ 1711 atomic_t remaining_pages; 1712 1713 /* 1714 * Number of references to this decompress_io_ctx. 1715 * 1716 * One reference is held for I/O completion. This reference is dropped 1717 * after the pagecache pages are updated and unlocked -- either after 1718 * decompression (and verity if enabled), or after an error. 1719 * 1720 * In addition, each compressed page holds a reference while it is in a 1721 * bio. These references are necessary prevent compressed pages from 1722 * being freed while they are still in a bio. 1723 */ 1724 refcount_t refcnt; 1725 1726 bool failed; /* IO error occurred before decompression? */ 1727 struct fsverity_info *vi; /* fs-verity context if needed */ 1728 unsigned char compress_algorithm; /* backup algorithm type */ 1729 void *private; /* payload buffer for specified decompression algorithm */ 1730 void *private2; /* extra payload buffer */ 1731 struct work_struct verity_work; /* work to verify the decompressed pages */ 1732 struct work_struct free_work; /* work for late free this structure itself */ 1733 }; 1734 1735 #define NULL_CLUSTER ((unsigned int)(~0)) 1736 #define MIN_COMPRESS_LOG_SIZE 2 1737 #define MAX_COMPRESS_LOG_SIZE 8 1738 #define MAX_COMPRESS_WINDOW_SIZE(log_size) ((PAGE_SIZE) << (log_size)) 1739 1740 struct f2fs_sb_info { 1741 struct super_block *sb; /* pointer to VFS super block */ 1742 struct proc_dir_entry *s_proc; /* proc entry */ 1743 struct f2fs_super_block *raw_super; /* raw super block pointer */ 1744 struct f2fs_rwsem sb_lock; /* lock for raw super block */ 1745 int valid_super_block; /* valid super block no */ 1746 unsigned long s_flag; /* flags for sbi */ 1747 struct mutex writepages; /* mutex for writepages() */ 1748 1749 #ifdef CONFIG_BLK_DEV_ZONED 1750 unsigned int blocks_per_blkz; /* F2FS blocks per zone */ 1751 unsigned int unusable_blocks_per_sec; /* unusable blocks per section */ 1752 unsigned int max_open_zones; /* max open zone resources of the zoned device */ 1753 /* For adjust the priority writing position of data in zone UFS */ 1754 unsigned int blkzone_alloc_policy; 1755 #endif 1756 1757 /* for node-related operations */ 1758 struct f2fs_nm_info *nm_info; /* node manager */ 1759 struct inode *node_inode; /* cache node blocks */ 1760 1761 /* for segment-related operations */ 1762 struct f2fs_sm_info *sm_info; /* segment manager */ 1763 1764 /* for bio operations */ 1765 struct f2fs_bio_info *write_io[NR_PAGE_TYPE]; /* for write bios */ 1766 /* keep migration IO order for LFS mode */ 1767 struct f2fs_rwsem io_order_lock; 1768 pgoff_t page_eio_ofs[NR_PAGE_TYPE]; /* EIO page offset */ 1769 int page_eio_cnt[NR_PAGE_TYPE]; /* EIO count */ 1770 1771 /* for checkpoint */ 1772 struct f2fs_checkpoint *ckpt; /* raw checkpoint pointer */ 1773 int cur_cp_pack; /* remain current cp pack */ 1774 spinlock_t cp_lock; /* for flag in ckpt */ 1775 struct inode *meta_inode; /* cache meta blocks */ 1776 struct f2fs_rwsem cp_global_sem; /* checkpoint procedure lock */ 1777 struct f2fs_rwsem cp_rwsem; /* blocking FS operations */ 1778 struct f2fs_rwsem node_write; /* locking node writes */ 1779 struct f2fs_rwsem node_change; /* locking node change */ 1780 wait_queue_head_t cp_wait; 1781 unsigned long last_time[MAX_TIME]; /* to store time in jiffies */ 1782 long interval_time[MAX_TIME]; /* to store thresholds */ 1783 struct ckpt_req_control cprc_info; /* for checkpoint request control */ 1784 struct cp_stats cp_stats; /* for time stat of checkpoint */ 1785 1786 struct inode_management im[MAX_INO_ENTRY]; /* manage inode cache */ 1787 1788 spinlock_t fsync_node_lock; /* for node entry lock */ 1789 struct list_head fsync_node_list; /* node list head */ 1790 unsigned int fsync_seg_id; /* sequence id */ 1791 unsigned int fsync_node_num; /* number of node entries */ 1792 1793 /* for orphan inode, use 0'th array */ 1794 unsigned int max_orphans; /* max orphan inodes */ 1795 1796 /* for inode management */ 1797 struct list_head inode_list[NR_INODE_TYPE]; /* dirty inode list */ 1798 spinlock_t inode_lock[NR_INODE_TYPE]; /* for dirty inode list lock */ 1799 struct mutex flush_lock; /* for flush exclusion */ 1800 1801 /* for extent tree cache */ 1802 struct extent_tree_info extent_tree[NR_EXTENT_CACHES]; 1803 atomic64_t allocated_data_blocks; /* for block age extent_cache */ 1804 unsigned int max_read_extent_count; /* max read extent count per inode */ 1805 1806 /* The threshold used for hot and warm data seperation*/ 1807 unsigned int hot_data_age_threshold; 1808 unsigned int warm_data_age_threshold; 1809 unsigned int last_age_weight; 1810 1811 /* control donate caches */ 1812 unsigned int donate_files; 1813 1814 /* basic filesystem units */ 1815 unsigned int log_sectors_per_block; /* log2 sectors per block */ 1816 unsigned int log_blocksize; /* log2 block size */ 1817 unsigned int blocksize; /* block size */ 1818 unsigned int root_ino_num; /* root inode number*/ 1819 unsigned int node_ino_num; /* node inode number*/ 1820 unsigned int meta_ino_num; /* meta inode number*/ 1821 unsigned int log_blocks_per_seg; /* log2 blocks per segment */ 1822 unsigned int blocks_per_seg; /* blocks per segment */ 1823 unsigned int segs_per_sec; /* segments per section */ 1824 unsigned int secs_per_zone; /* sections per zone */ 1825 unsigned int total_sections; /* total section count */ 1826 unsigned int total_node_count; /* total node block count */ 1827 unsigned int total_valid_node_count; /* valid node block count */ 1828 int dir_level; /* directory level */ 1829 bool readdir_ra; /* readahead inode in readdir */ 1830 unsigned int max_io_bytes; /* max io bytes to merge IOs */ 1831 1832 /* variable summary block units */ 1833 unsigned int sum_blocksize; /* sum block size */ 1834 unsigned int sums_per_block; /* sum block count per block */ 1835 unsigned int entries_in_sum; /* entry count in sum block */ 1836 unsigned int sum_entry_size; /* total entry size in sum block */ 1837 unsigned int sum_journal_size; /* journal size in sum block */ 1838 unsigned int nat_journal_entries; /* nat journal entry count in the journal */ 1839 unsigned int sit_journal_entries; /* sit journal entry count in the journal */ 1840 1841 block_t user_block_count; /* # of user blocks */ 1842 block_t total_valid_block_count; /* # of valid blocks */ 1843 block_t discard_blks; /* discard command candidats */ 1844 block_t last_valid_block_count; /* for recovery */ 1845 block_t reserved_blocks; /* configurable reserved blocks */ 1846 block_t current_reserved_blocks; /* current reserved blocks */ 1847 1848 /* Additional tracking for no checkpoint mode */ 1849 block_t unusable_block_count; /* # of blocks saved by last cp */ 1850 1851 unsigned int nquota_files; /* # of quota sysfile */ 1852 struct f2fs_rwsem quota_sem; /* blocking cp for flags */ 1853 struct task_struct *umount_lock_holder; /* s_umount lock holder */ 1854 1855 /* # of pages, see count_type */ 1856 atomic_t nr_pages[NR_COUNT_TYPE]; 1857 /* # of allocated blocks */ 1858 struct percpu_counter alloc_valid_block_count; 1859 /* # of node block writes as roll forward recovery */ 1860 struct percpu_counter rf_node_block_count; 1861 1862 /* writeback control */ 1863 atomic_t wb_sync_req[META]; /* count # of WB_SYNC threads */ 1864 1865 /* valid inode count */ 1866 struct percpu_counter total_valid_inode_count; 1867 1868 struct f2fs_mount_info mount_opt; /* mount options */ 1869 1870 /* for cleaning operations */ 1871 struct f2fs_rwsem gc_lock; /* 1872 * semaphore for GC, avoid 1873 * race between GC and GC or CP 1874 */ 1875 struct f2fs_gc_kthread *gc_thread; /* GC thread */ 1876 struct atgc_management am; /* atgc management */ 1877 unsigned int cur_victim_sec; /* current victim section num */ 1878 unsigned int gc_mode; /* current GC state */ 1879 unsigned int next_victim_seg[2]; /* next segment in victim section */ 1880 spinlock_t gc_remaining_trials_lock; 1881 /* remaining trial count for GC_URGENT_* and GC_IDLE_* */ 1882 unsigned int gc_remaining_trials; 1883 1884 /* for skip statistic */ 1885 unsigned long long skipped_gc_rwsem; /* FG_GC only */ 1886 1887 /* free sections reserved for pinned file */ 1888 unsigned int reserved_pin_section; 1889 1890 /* threshold for gc trials on pinned files */ 1891 unsigned short gc_pin_file_threshold; 1892 struct f2fs_rwsem pin_sem; 1893 1894 /* maximum # of trials to find a victim segment for SSR and GC */ 1895 unsigned int max_victim_search; 1896 /* migration granularity of garbage collection, unit: segment */ 1897 unsigned int migration_granularity; 1898 /* migration window granularity of garbage collection, unit: segment */ 1899 unsigned int migration_window_granularity; 1900 1901 /* 1902 * for stat information. 1903 * one is for the LFS mode, and the other is for the SSR mode. 1904 */ 1905 #ifdef CONFIG_F2FS_STAT_FS 1906 struct f2fs_stat_info *stat_info; /* FS status information */ 1907 atomic_t meta_count[META_MAX]; /* # of meta blocks */ 1908 unsigned int segment_count[2]; /* # of allocated segments */ 1909 unsigned int block_count[2]; /* # of allocated blocks */ 1910 atomic_t inplace_count; /* # of inplace update */ 1911 /* # of lookup extent cache */ 1912 atomic64_t total_hit_ext[NR_EXTENT_CACHES]; 1913 /* # of hit rbtree extent node */ 1914 atomic64_t read_hit_rbtree[NR_EXTENT_CACHES]; 1915 /* # of hit cached extent node */ 1916 atomic64_t read_hit_cached[NR_EXTENT_CACHES]; 1917 /* # of hit largest extent node in read extent cache */ 1918 atomic64_t read_hit_largest; 1919 atomic_t inline_xattr; /* # of inline_xattr inodes */ 1920 atomic_t inline_inode; /* # of inline_data inodes */ 1921 atomic_t inline_dir; /* # of inline_dentry inodes */ 1922 atomic_t compr_inode; /* # of compressed inodes */ 1923 atomic64_t compr_blocks; /* # of compressed blocks */ 1924 atomic_t swapfile_inode; /* # of swapfile inodes */ 1925 atomic_t atomic_files; /* # of opened atomic file */ 1926 atomic_t max_aw_cnt; /* max # of atomic writes */ 1927 unsigned int io_skip_bggc; /* skip background gc for in-flight IO */ 1928 unsigned int other_skip_bggc; /* skip background gc for other reasons */ 1929 unsigned int ndirty_inode[NR_INODE_TYPE]; /* # of dirty inodes */ 1930 atomic_t cp_call_count[MAX_CALL_TYPE]; /* # of cp call */ 1931 #endif 1932 spinlock_t stat_lock; /* lock for stat operations */ 1933 1934 /* to attach REQ_META|REQ_FUA flags */ 1935 unsigned int data_io_flag; 1936 unsigned int node_io_flag; 1937 1938 /* For sysfs support */ 1939 struct kobject s_kobj; /* /sys/fs/f2fs/<devname> */ 1940 struct completion s_kobj_unregister; 1941 1942 struct kobject s_stat_kobj; /* /sys/fs/f2fs/<devname>/stat */ 1943 struct completion s_stat_kobj_unregister; 1944 1945 struct kobject s_feature_list_kobj; /* /sys/fs/f2fs/<devname>/feature_list */ 1946 struct completion s_feature_list_kobj_unregister; 1947 1948 /* For shrinker support */ 1949 struct list_head s_list; 1950 struct mutex umount_mutex; 1951 unsigned int shrinker_run_no; 1952 1953 /* For multi devices */ 1954 int s_ndevs; /* number of devices */ 1955 struct f2fs_dev_info *devs; /* for device list */ 1956 unsigned int dirty_device; /* for checkpoint data flush */ 1957 spinlock_t dev_lock; /* protect dirty_device */ 1958 bool aligned_blksize; /* all devices has the same logical blksize */ 1959 unsigned int first_seq_zone_segno; /* first segno in sequential zone */ 1960 unsigned int bggc_io_aware; /* For adjust the BG_GC priority when pending IO */ 1961 unsigned int allocate_section_hint; /* the boundary position between devices */ 1962 unsigned int allocate_section_policy; /* determine the section writing priority */ 1963 1964 /* For write statistics */ 1965 u64 sectors_written_start; 1966 u64 kbytes_written; 1967 1968 /* Precomputed FS UUID checksum for seeding other checksums */ 1969 __u32 s_chksum_seed; 1970 1971 struct workqueue_struct *post_read_wq; /* post read workqueue */ 1972 1973 /* 1974 * If we are in irq context, let's update error information into 1975 * on-disk superblock in the work. 1976 */ 1977 struct work_struct s_error_work; 1978 unsigned char errors[MAX_F2FS_ERRORS]; /* error flags */ 1979 unsigned char stop_reason[MAX_STOP_REASON]; /* stop reason */ 1980 spinlock_t error_lock; /* protect errors/stop_reason array */ 1981 bool error_dirty; /* errors of sb is dirty */ 1982 1983 /* For reclaimed segs statistics per each GC mode */ 1984 unsigned int gc_segment_mode; /* GC state for reclaimed segments */ 1985 unsigned int gc_reclaimed_segs[MAX_GC_MODE]; /* Reclaimed segs for each mode */ 1986 1987 unsigned int seq_file_ra_mul; /* multiplier for ra_pages of seq. files in fadvise */ 1988 1989 int max_fragment_chunk; /* max chunk size for block fragmentation mode */ 1990 int max_fragment_hole; /* max hole size for block fragmentation mode */ 1991 1992 /* For atomic write statistics */ 1993 atomic64_t current_atomic_write; 1994 s64 peak_atomic_write; 1995 u64 committed_atomic_block; 1996 u64 revoked_atomic_block; 1997 1998 /* carve out reserved_blocks from total blocks */ 1999 bool carve_out; 2000 2001 /* max elapsed time threshold in critical region that lock covered */ 2002 unsigned long long max_lock_elapsed_time; 2003 2004 /* enable/disable to adjust task priority in critical region covered by lock */ 2005 unsigned int adjust_lock_priority; 2006 2007 /* adjust priority for task which is in critical region covered by lock */ 2008 unsigned int lock_duration_priority; 2009 2010 /* priority for critical task, e.g. f2fs_ckpt, f2fs_gc threads */ 2011 long critical_task_priority; 2012 2013 #ifdef CONFIG_F2FS_FS_COMPRESSION 2014 struct kmem_cache *page_array_slab; /* page array entry */ 2015 unsigned int page_array_slab_size; /* default page array slab size */ 2016 2017 /* For runtime compression statistics */ 2018 u64 compr_written_block; 2019 u64 compr_saved_block; 2020 u32 compr_new_inode; 2021 2022 /* For compressed block cache */ 2023 struct inode *compress_inode; /* cache compressed blocks */ 2024 unsigned int compress_percent; /* cache page percentage */ 2025 unsigned int compress_watermark; /* cache page watermark */ 2026 atomic_t compress_page_hit; /* cache hit count */ 2027 #endif 2028 2029 #ifdef CONFIG_F2FS_IOSTAT 2030 /* For app/fs IO statistics */ 2031 spinlock_t iostat_lock; 2032 unsigned long long iostat_count[NR_IO_TYPE]; 2033 unsigned long long iostat_bytes[NR_IO_TYPE]; 2034 unsigned long long prev_iostat_bytes[NR_IO_TYPE]; 2035 bool iostat_enable; 2036 unsigned long iostat_next_period; 2037 unsigned int iostat_period_ms; 2038 2039 /* For io latency related statistics info in one iostat period */ 2040 spinlock_t iostat_lat_lock; 2041 struct iostat_lat_info *iostat_io_lat; 2042 #endif 2043 }; 2044 2045 /* Definitions to access f2fs_sb_info */ 2046 #define SEGS_TO_BLKS(sbi, segs) \ 2047 ((segs) << (sbi)->log_blocks_per_seg) 2048 #define BLKS_TO_SEGS(sbi, blks) \ 2049 ((blks) >> (sbi)->log_blocks_per_seg) 2050 2051 #define BLKS_PER_SEG(sbi) ((sbi)->blocks_per_seg) 2052 #define BLKS_PER_SEC(sbi) (SEGS_TO_BLKS(sbi, (sbi)->segs_per_sec)) 2053 #define SEGS_PER_SEC(sbi) ((sbi)->segs_per_sec) 2054 2055 __printf(3, 4) 2056 void f2fs_printk(struct f2fs_sb_info *sbi, bool limit_rate, const char *fmt, ...); 2057 2058 #define f2fs_err(sbi, fmt, ...) \ 2059 f2fs_printk(sbi, false, KERN_ERR fmt, ##__VA_ARGS__) 2060 #define f2fs_warn(sbi, fmt, ...) \ 2061 f2fs_printk(sbi, false, KERN_WARNING fmt, ##__VA_ARGS__) 2062 #define f2fs_notice(sbi, fmt, ...) \ 2063 f2fs_printk(sbi, false, KERN_NOTICE fmt, ##__VA_ARGS__) 2064 #define f2fs_info(sbi, fmt, ...) \ 2065 f2fs_printk(sbi, false, KERN_INFO fmt, ##__VA_ARGS__) 2066 #define f2fs_debug(sbi, fmt, ...) \ 2067 f2fs_printk(sbi, false, KERN_DEBUG fmt, ##__VA_ARGS__) 2068 2069 #define f2fs_err_ratelimited(sbi, fmt, ...) \ 2070 f2fs_printk(sbi, true, KERN_ERR fmt, ##__VA_ARGS__) 2071 #define f2fs_warn_ratelimited(sbi, fmt, ...) \ 2072 f2fs_printk(sbi, true, KERN_WARNING fmt, ##__VA_ARGS__) 2073 #define f2fs_info_ratelimited(sbi, fmt, ...) \ 2074 f2fs_printk(sbi, true, KERN_INFO fmt, ##__VA_ARGS__) 2075 2076 #ifdef CONFIG_F2FS_FAULT_INJECTION 2077 #define time_to_inject(sbi, type) __time_to_inject(sbi, type, __func__, \ 2078 __builtin_return_address(0)) 2079 static inline bool __time_to_inject(struct f2fs_sb_info *sbi, int type, 2080 const char *func, const char *parent_func) 2081 { 2082 struct f2fs_fault_info *ffi = &F2FS_OPTION(sbi).fault_info; 2083 2084 if (!ffi->inject_rate) 2085 return false; 2086 2087 if (!IS_FAULT_SET(ffi, type)) 2088 return false; 2089 2090 atomic_inc(&ffi->inject_ops); 2091 if (atomic_read(&ffi->inject_ops) >= ffi->inject_rate) { 2092 atomic_set(&ffi->inject_ops, 0); 2093 ffi->inject_count[type]++; 2094 f2fs_info_ratelimited(sbi, "inject %s in %s of %pS", 2095 f2fs_fault_name[type], func, parent_func); 2096 return true; 2097 } 2098 return false; 2099 } 2100 #else 2101 static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type) 2102 { 2103 return false; 2104 } 2105 #endif 2106 2107 /* 2108 * Test if the mounted volume is a multi-device volume. 2109 * - For a single regular disk volume, sbi->s_ndevs is 0. 2110 * - For a single zoned disk volume, sbi->s_ndevs is 1. 2111 * - For a multi-device volume, sbi->s_ndevs is always 2 or more. 2112 */ 2113 static inline bool f2fs_is_multi_device(struct f2fs_sb_info *sbi) 2114 { 2115 return sbi->s_ndevs > 1; 2116 } 2117 2118 static inline void f2fs_update_time(struct f2fs_sb_info *sbi, int type) 2119 { 2120 unsigned long now = jiffies; 2121 2122 sbi->last_time[type] = now; 2123 2124 /* DISCARD_TIME and GC_TIME are based on REQ_TIME */ 2125 if (type == REQ_TIME) { 2126 sbi->last_time[DISCARD_TIME] = now; 2127 sbi->last_time[GC_TIME] = now; 2128 } 2129 } 2130 2131 static inline bool f2fs_time_over(struct f2fs_sb_info *sbi, int type) 2132 { 2133 unsigned long interval = sbi->interval_time[type] * HZ; 2134 2135 return time_after(jiffies, sbi->last_time[type] + interval); 2136 } 2137 2138 static inline unsigned int f2fs_time_to_wait(struct f2fs_sb_info *sbi, 2139 int type) 2140 { 2141 unsigned long interval = sbi->interval_time[type] * HZ; 2142 unsigned int wait_ms = 0; 2143 long delta; 2144 2145 delta = (sbi->last_time[type] + interval) - jiffies; 2146 if (delta > 0) 2147 wait_ms = jiffies_to_msecs(delta); 2148 2149 return wait_ms; 2150 } 2151 2152 /* 2153 * Inline functions 2154 */ 2155 static inline u32 __f2fs_crc32(u32 crc, const void *address, 2156 unsigned int length) 2157 { 2158 return crc32(crc, address, length); 2159 } 2160 2161 static inline u32 f2fs_crc32(const void *address, unsigned int length) 2162 { 2163 return __f2fs_crc32(F2FS_SUPER_MAGIC, address, length); 2164 } 2165 2166 static inline u32 f2fs_chksum(u32 crc, const void *address, unsigned int length) 2167 { 2168 return __f2fs_crc32(crc, address, length); 2169 } 2170 2171 static inline struct f2fs_inode_info *F2FS_I(struct inode *inode) 2172 { 2173 return container_of(inode, struct f2fs_inode_info, vfs_inode); 2174 } 2175 2176 static inline struct f2fs_sb_info *F2FS_SB(struct super_block *sb) 2177 { 2178 return sb->s_fs_info; 2179 } 2180 2181 static inline struct f2fs_sb_info *F2FS_I_SB(struct inode *inode) 2182 { 2183 return F2FS_SB(inode->i_sb); 2184 } 2185 2186 static inline struct f2fs_sb_info *F2FS_M_SB(struct address_space *mapping) 2187 { 2188 return F2FS_I_SB(mapping->host); 2189 } 2190 2191 static inline struct f2fs_sb_info *F2FS_F_SB(const struct folio *folio) 2192 { 2193 return F2FS_M_SB(folio->mapping); 2194 } 2195 2196 static inline struct f2fs_super_block *F2FS_RAW_SUPER(struct f2fs_sb_info *sbi) 2197 { 2198 return (struct f2fs_super_block *)(sbi->raw_super); 2199 } 2200 2201 static inline struct f2fs_super_block *F2FS_SUPER_BLOCK(struct folio *folio, 2202 pgoff_t index) 2203 { 2204 pgoff_t idx_in_folio = index % folio_nr_pages(folio); 2205 2206 return (struct f2fs_super_block *) 2207 (page_address(folio_page(folio, idx_in_folio)) + 2208 F2FS_SUPER_OFFSET); 2209 } 2210 2211 static inline struct f2fs_checkpoint *F2FS_CKPT(struct f2fs_sb_info *sbi) 2212 { 2213 return (struct f2fs_checkpoint *)(sbi->ckpt); 2214 } 2215 2216 static inline struct f2fs_node *F2FS_NODE(const struct folio *folio) 2217 { 2218 return (struct f2fs_node *)folio_address(folio); 2219 } 2220 2221 static inline struct f2fs_inode *F2FS_INODE(const struct folio *folio) 2222 { 2223 return &((struct f2fs_node *)folio_address(folio))->i; 2224 } 2225 2226 static inline struct f2fs_nm_info *NM_I(struct f2fs_sb_info *sbi) 2227 { 2228 return (struct f2fs_nm_info *)(sbi->nm_info); 2229 } 2230 2231 static inline struct f2fs_sm_info *SM_I(struct f2fs_sb_info *sbi) 2232 { 2233 return (struct f2fs_sm_info *)(sbi->sm_info); 2234 } 2235 2236 static inline struct sit_info *SIT_I(struct f2fs_sb_info *sbi) 2237 { 2238 return (struct sit_info *)(SM_I(sbi)->sit_info); 2239 } 2240 2241 static inline struct free_segmap_info *FREE_I(struct f2fs_sb_info *sbi) 2242 { 2243 return (struct free_segmap_info *)(SM_I(sbi)->free_info); 2244 } 2245 2246 static inline struct dirty_seglist_info *DIRTY_I(struct f2fs_sb_info *sbi) 2247 { 2248 return (struct dirty_seglist_info *)(SM_I(sbi)->dirty_info); 2249 } 2250 2251 static inline struct address_space *META_MAPPING(struct f2fs_sb_info *sbi) 2252 { 2253 return sbi->meta_inode->i_mapping; 2254 } 2255 2256 static inline struct address_space *NODE_MAPPING(struct f2fs_sb_info *sbi) 2257 { 2258 return sbi->node_inode->i_mapping; 2259 } 2260 2261 static inline bool is_meta_folio(struct folio *folio) 2262 { 2263 return folio->mapping == META_MAPPING(F2FS_F_SB(folio)); 2264 } 2265 2266 static inline bool is_node_folio(struct folio *folio) 2267 { 2268 return folio->mapping == NODE_MAPPING(F2FS_F_SB(folio)); 2269 } 2270 2271 static inline bool is_sbi_flag_set(struct f2fs_sb_info *sbi, unsigned int type) 2272 { 2273 return test_bit(type, &sbi->s_flag); 2274 } 2275 2276 static inline void set_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type) 2277 { 2278 set_bit(type, &sbi->s_flag); 2279 } 2280 2281 static inline void clear_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type) 2282 { 2283 clear_bit(type, &sbi->s_flag); 2284 } 2285 2286 static inline unsigned long long cur_cp_version(struct f2fs_checkpoint *cp) 2287 { 2288 return le64_to_cpu(cp->checkpoint_ver); 2289 } 2290 2291 static inline unsigned long f2fs_qf_ino(struct super_block *sb, int type) 2292 { 2293 if (type < F2FS_MAX_QUOTAS) 2294 return le32_to_cpu(F2FS_SB(sb)->raw_super->qf_ino[type]); 2295 return 0; 2296 } 2297 2298 static inline __u64 cur_cp_crc(struct f2fs_checkpoint *cp) 2299 { 2300 size_t crc_offset = le32_to_cpu(cp->checksum_offset); 2301 return le32_to_cpu(*((__le32 *)((unsigned char *)cp + crc_offset))); 2302 } 2303 2304 static inline bool __is_set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 2305 { 2306 unsigned int ckpt_flags = le32_to_cpu(cp->ckpt_flags); 2307 2308 return ckpt_flags & f; 2309 } 2310 2311 static inline bool is_set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 2312 { 2313 return __is_set_ckpt_flags(F2FS_CKPT(sbi), f); 2314 } 2315 2316 static inline void __set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 2317 { 2318 unsigned int ckpt_flags; 2319 2320 ckpt_flags = le32_to_cpu(cp->ckpt_flags); 2321 ckpt_flags |= f; 2322 cp->ckpt_flags = cpu_to_le32(ckpt_flags); 2323 } 2324 2325 static inline void set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 2326 { 2327 unsigned long flags; 2328 2329 spin_lock_irqsave(&sbi->cp_lock, flags); 2330 __set_ckpt_flags(F2FS_CKPT(sbi), f); 2331 spin_unlock_irqrestore(&sbi->cp_lock, flags); 2332 } 2333 2334 static inline void __clear_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 2335 { 2336 unsigned int ckpt_flags; 2337 2338 ckpt_flags = le32_to_cpu(cp->ckpt_flags); 2339 ckpt_flags &= (~f); 2340 cp->ckpt_flags = cpu_to_le32(ckpt_flags); 2341 } 2342 2343 static inline void clear_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 2344 { 2345 unsigned long flags; 2346 2347 spin_lock_irqsave(&sbi->cp_lock, flags); 2348 __clear_ckpt_flags(F2FS_CKPT(sbi), f); 2349 spin_unlock_irqrestore(&sbi->cp_lock, flags); 2350 } 2351 2352 #define init_f2fs_rwsem(sem) __init_f2fs_rwsem(sem, NULL, LOCK_NAME_NONE) 2353 #define init_f2fs_rwsem_trace __init_f2fs_rwsem 2354 2355 #define __init_f2fs_rwsem(sem, sbi, name) \ 2356 do { \ 2357 static struct lock_class_key __key; \ 2358 \ 2359 do_init_f2fs_rwsem((sem), #sem, &__key, sbi, name); \ 2360 } while (0) 2361 2362 static inline void do_init_f2fs_rwsem(struct f2fs_rwsem *sem, 2363 const char *sem_name, struct lock_class_key *key, 2364 struct f2fs_sb_info *sbi, enum f2fs_lock_name name) 2365 { 2366 sem->sbi = sbi; 2367 sem->name = name; 2368 __init_rwsem(&sem->internal_rwsem, sem_name, key); 2369 #ifdef CONFIG_F2FS_UNFAIR_RWSEM 2370 init_waitqueue_head(&sem->read_waiters); 2371 #endif 2372 } 2373 2374 static inline int f2fs_rwsem_is_locked(struct f2fs_rwsem *sem) 2375 { 2376 return rwsem_is_locked(&sem->internal_rwsem); 2377 } 2378 2379 static inline int f2fs_rwsem_is_contended(struct f2fs_rwsem *sem) 2380 { 2381 return rwsem_is_contended(&sem->internal_rwsem); 2382 } 2383 2384 static inline void f2fs_down_read(struct f2fs_rwsem *sem) 2385 { 2386 #ifdef CONFIG_F2FS_UNFAIR_RWSEM 2387 wait_event(sem->read_waiters, down_read_trylock(&sem->internal_rwsem)); 2388 #else 2389 down_read(&sem->internal_rwsem); 2390 #endif 2391 } 2392 2393 static inline int f2fs_down_read_trylock(struct f2fs_rwsem *sem) 2394 { 2395 return down_read_trylock(&sem->internal_rwsem); 2396 } 2397 2398 static inline void f2fs_up_read(struct f2fs_rwsem *sem) 2399 { 2400 up_read(&sem->internal_rwsem); 2401 } 2402 2403 static inline void f2fs_down_write(struct f2fs_rwsem *sem) 2404 { 2405 down_write(&sem->internal_rwsem); 2406 } 2407 2408 #ifdef CONFIG_DEBUG_LOCK_ALLOC 2409 static inline void f2fs_down_read_nested(struct f2fs_rwsem *sem, int subclass) 2410 { 2411 down_read_nested(&sem->internal_rwsem, subclass); 2412 } 2413 2414 static inline void f2fs_down_write_nested(struct f2fs_rwsem *sem, int subclass) 2415 { 2416 down_write_nested(&sem->internal_rwsem, subclass); 2417 } 2418 #else 2419 #define f2fs_down_read_nested(sem, subclass) f2fs_down_read(sem) 2420 #define f2fs_down_write_nested(sem, subclass) f2fs_down_write(sem) 2421 #endif 2422 2423 static inline int f2fs_down_write_trylock(struct f2fs_rwsem *sem) 2424 { 2425 return down_write_trylock(&sem->internal_rwsem); 2426 } 2427 2428 static inline void f2fs_up_write(struct f2fs_rwsem *sem) 2429 { 2430 up_write(&sem->internal_rwsem); 2431 #ifdef CONFIG_F2FS_UNFAIR_RWSEM 2432 wake_up_all(&sem->read_waiters); 2433 #endif 2434 } 2435 2436 void f2fs_down_read_trace(struct f2fs_rwsem *sem, struct f2fs_lock_context *lc); 2437 int f2fs_down_read_trylock_trace(struct f2fs_rwsem *sem, 2438 struct f2fs_lock_context *lc); 2439 void f2fs_up_read_trace(struct f2fs_rwsem *sem, struct f2fs_lock_context *lc); 2440 void f2fs_down_write_trace(struct f2fs_rwsem *sem, 2441 struct f2fs_lock_context *lc); 2442 int f2fs_down_write_trylock_trace(struct f2fs_rwsem *sem, 2443 struct f2fs_lock_context *lc); 2444 void f2fs_up_write_trace(struct f2fs_rwsem *sem, struct f2fs_lock_context *lc); 2445 2446 static inline void disable_nat_bits(struct f2fs_sb_info *sbi, bool lock) 2447 { 2448 unsigned long flags; 2449 unsigned char *nat_bits; 2450 2451 /* 2452 * In order to re-enable nat_bits we need to call fsck.f2fs by 2453 * set_sbi_flag(sbi, SBI_NEED_FSCK). But it may give huge cost, 2454 * so let's rely on regular fsck or unclean shutdown. 2455 */ 2456 2457 if (lock) 2458 spin_lock_irqsave(&sbi->cp_lock, flags); 2459 __clear_ckpt_flags(F2FS_CKPT(sbi), CP_NAT_BITS_FLAG); 2460 nat_bits = NM_I(sbi)->nat_bits; 2461 NM_I(sbi)->nat_bits = NULL; 2462 if (lock) 2463 spin_unlock_irqrestore(&sbi->cp_lock, flags); 2464 2465 kvfree(nat_bits); 2466 } 2467 2468 static inline bool enabled_nat_bits(struct f2fs_sb_info *sbi, 2469 struct cp_control *cpc) 2470 { 2471 bool set = is_set_ckpt_flags(sbi, CP_NAT_BITS_FLAG); 2472 2473 return (cpc) ? (cpc->reason & CP_UMOUNT) && set : set; 2474 } 2475 2476 static inline int __get_cp_reason(struct f2fs_sb_info *sbi) 2477 { 2478 int reason = CP_SYNC; 2479 2480 if (test_opt(sbi, FASTBOOT)) 2481 reason = CP_FASTBOOT; 2482 if (is_sbi_flag_set(sbi, SBI_IS_CLOSE)) 2483 reason = CP_UMOUNT; 2484 return reason; 2485 } 2486 2487 static inline bool __remain_node_summaries(int reason) 2488 { 2489 return (reason & (CP_UMOUNT | CP_FASTBOOT)); 2490 } 2491 2492 static inline bool __exist_node_summaries(struct f2fs_sb_info *sbi) 2493 { 2494 return (is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG) || 2495 is_set_ckpt_flags(sbi, CP_FASTBOOT_FLAG)); 2496 } 2497 2498 /* 2499 * Check whether the inode has blocks or not 2500 */ 2501 static inline int F2FS_HAS_BLOCKS(struct inode *inode) 2502 { 2503 block_t xattr_block = F2FS_I(inode)->i_xattr_nid ? 1 : 0; 2504 2505 return (inode->i_blocks >> F2FS_LOG_SECTORS_PER_BLOCK) > xattr_block; 2506 } 2507 2508 static inline bool f2fs_has_xattr_block(unsigned int ofs) 2509 { 2510 return ofs == XATTR_NODE_OFFSET; 2511 } 2512 2513 static inline bool __allow_reserved_root(struct f2fs_sb_info *sbi, 2514 struct inode *inode, bool cap) 2515 { 2516 if (!inode) 2517 return true; 2518 if (IS_NOQUOTA(inode)) 2519 return true; 2520 if (uid_eq(F2FS_OPTION(sbi).s_resuid, current_fsuid())) 2521 return true; 2522 if (!gid_eq(F2FS_OPTION(sbi).s_resgid, GLOBAL_ROOT_GID) && 2523 in_group_p(F2FS_OPTION(sbi).s_resgid)) 2524 return true; 2525 if (cap && capable(CAP_SYS_RESOURCE)) 2526 return true; 2527 return false; 2528 } 2529 2530 static inline unsigned int get_available_block_count(struct f2fs_sb_info *sbi, 2531 struct inode *inode, bool cap) 2532 { 2533 block_t avail_user_block_count; 2534 2535 avail_user_block_count = sbi->user_block_count - 2536 sbi->current_reserved_blocks; 2537 2538 if (test_opt(sbi, RESERVE_ROOT) && !__allow_reserved_root(sbi, inode, cap)) 2539 avail_user_block_count -= F2FS_OPTION(sbi).root_reserved_blocks; 2540 2541 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) { 2542 if (avail_user_block_count > sbi->unusable_block_count) 2543 avail_user_block_count -= sbi->unusable_block_count; 2544 else 2545 avail_user_block_count = 0; 2546 } 2547 2548 return avail_user_block_count; 2549 } 2550 2551 static inline void f2fs_i_blocks_write(struct inode *, block_t, bool, bool); 2552 static inline int inc_valid_block_count(struct f2fs_sb_info *sbi, 2553 struct inode *inode, blkcnt_t *count, bool partial) 2554 { 2555 long long diff = 0, release = 0; 2556 block_t avail_user_block_count; 2557 int ret; 2558 2559 ret = dquot_reserve_block(inode, *count); 2560 if (ret) 2561 return ret; 2562 2563 if (time_to_inject(sbi, FAULT_BLOCK)) { 2564 release = *count; 2565 goto release_quota; 2566 } 2567 2568 /* 2569 * let's increase this in prior to actual block count change in order 2570 * for f2fs_sync_file to avoid data races when deciding checkpoint. 2571 */ 2572 percpu_counter_add(&sbi->alloc_valid_block_count, (*count)); 2573 2574 spin_lock(&sbi->stat_lock); 2575 2576 avail_user_block_count = get_available_block_count(sbi, inode, true); 2577 diff = (long long)sbi->total_valid_block_count + *count - 2578 avail_user_block_count; 2579 if (unlikely(diff > 0)) { 2580 if (!partial) { 2581 spin_unlock(&sbi->stat_lock); 2582 release = *count; 2583 goto enospc; 2584 } 2585 if (diff > *count) 2586 diff = *count; 2587 *count -= diff; 2588 release = diff; 2589 if (!*count) { 2590 spin_unlock(&sbi->stat_lock); 2591 goto enospc; 2592 } 2593 } 2594 sbi->total_valid_block_count += (block_t)(*count); 2595 2596 spin_unlock(&sbi->stat_lock); 2597 2598 if (unlikely(release)) { 2599 percpu_counter_sub(&sbi->alloc_valid_block_count, release); 2600 dquot_release_reservation_block(inode, release); 2601 } 2602 f2fs_i_blocks_write(inode, *count, true, true); 2603 return 0; 2604 2605 enospc: 2606 percpu_counter_sub(&sbi->alloc_valid_block_count, release); 2607 release_quota: 2608 dquot_release_reservation_block(inode, release); 2609 return -ENOSPC; 2610 } 2611 2612 #define PAGE_PRIVATE_GET_FUNC(name, flagname) \ 2613 static inline bool folio_test_f2fs_##name(const struct folio *folio) \ 2614 { \ 2615 unsigned long priv = (unsigned long)folio->private; \ 2616 unsigned long v = (1UL << PAGE_PRIVATE_NOT_POINTER) | \ 2617 (1UL << PAGE_PRIVATE_##flagname); \ 2618 return (priv & v) == v; \ 2619 } \ 2620 static inline bool page_private_##name(struct page *page) \ 2621 { \ 2622 return PagePrivate(page) && \ 2623 test_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)) && \ 2624 test_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \ 2625 } 2626 2627 #define PAGE_PRIVATE_SET_FUNC(name, flagname) \ 2628 static inline void folio_set_f2fs_##name(struct folio *folio) \ 2629 { \ 2630 unsigned long v = (1UL << PAGE_PRIVATE_NOT_POINTER) | \ 2631 (1UL << PAGE_PRIVATE_##flagname); \ 2632 if (!folio->private) \ 2633 folio_attach_private(folio, (void *)v); \ 2634 else { \ 2635 v |= (unsigned long)folio->private; \ 2636 folio->private = (void *)v; \ 2637 } \ 2638 } \ 2639 static inline void set_page_private_##name(struct page *page) \ 2640 { \ 2641 if (!PagePrivate(page)) \ 2642 attach_page_private(page, (void *)0); \ 2643 set_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)); \ 2644 set_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \ 2645 } 2646 2647 #define PAGE_PRIVATE_CLEAR_FUNC(name, flagname) \ 2648 static inline void folio_clear_f2fs_##name(struct folio *folio) \ 2649 { \ 2650 unsigned long v = (unsigned long)folio->private; \ 2651 \ 2652 v &= ~(1UL << PAGE_PRIVATE_##flagname); \ 2653 if (v == (1UL << PAGE_PRIVATE_NOT_POINTER)) \ 2654 folio_detach_private(folio); \ 2655 else \ 2656 folio->private = (void *)v; \ 2657 } \ 2658 static inline void clear_page_private_##name(struct page *page) \ 2659 { \ 2660 clear_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \ 2661 if (page_private(page) == BIT(PAGE_PRIVATE_NOT_POINTER)) \ 2662 detach_page_private(page); \ 2663 } 2664 2665 PAGE_PRIVATE_GET_FUNC(nonpointer, NOT_POINTER); 2666 PAGE_PRIVATE_GET_FUNC(inline, INLINE_INODE); 2667 PAGE_PRIVATE_GET_FUNC(gcing, ONGOING_MIGRATION); 2668 PAGE_PRIVATE_GET_FUNC(atomic, ATOMIC_WRITE); 2669 2670 PAGE_PRIVATE_SET_FUNC(reference, REF_RESOURCE); 2671 PAGE_PRIVATE_SET_FUNC(inline, INLINE_INODE); 2672 PAGE_PRIVATE_SET_FUNC(gcing, ONGOING_MIGRATION); 2673 PAGE_PRIVATE_SET_FUNC(atomic, ATOMIC_WRITE); 2674 2675 PAGE_PRIVATE_CLEAR_FUNC(reference, REF_RESOURCE); 2676 PAGE_PRIVATE_CLEAR_FUNC(inline, INLINE_INODE); 2677 PAGE_PRIVATE_CLEAR_FUNC(gcing, ONGOING_MIGRATION); 2678 PAGE_PRIVATE_CLEAR_FUNC(atomic, ATOMIC_WRITE); 2679 2680 static inline unsigned long folio_get_f2fs_data(struct folio *folio) 2681 { 2682 unsigned long data = (unsigned long)folio->private; 2683 2684 if (!test_bit(PAGE_PRIVATE_NOT_POINTER, &data)) 2685 return 0; 2686 return data >> PAGE_PRIVATE_MAX; 2687 } 2688 2689 static inline void folio_set_f2fs_data(struct folio *folio, unsigned long data) 2690 { 2691 data = (1UL << PAGE_PRIVATE_NOT_POINTER) | (data << PAGE_PRIVATE_MAX); 2692 2693 if (!folio_test_private(folio)) 2694 folio_attach_private(folio, (void *)data); 2695 else 2696 folio->private = (void *)((unsigned long)folio->private | data); 2697 } 2698 2699 static inline void dec_valid_block_count(struct f2fs_sb_info *sbi, 2700 struct inode *inode, 2701 block_t count) 2702 { 2703 blkcnt_t sectors = count << F2FS_LOG_SECTORS_PER_BLOCK; 2704 2705 spin_lock(&sbi->stat_lock); 2706 if (unlikely(sbi->total_valid_block_count < count)) { 2707 f2fs_warn(sbi, "Inconsistent total_valid_block_count:%u, ino:%lu, count:%u", 2708 sbi->total_valid_block_count, inode->i_ino, count); 2709 sbi->total_valid_block_count = 0; 2710 set_sbi_flag(sbi, SBI_NEED_FSCK); 2711 } else { 2712 sbi->total_valid_block_count -= count; 2713 } 2714 if (sbi->reserved_blocks && 2715 sbi->current_reserved_blocks < sbi->reserved_blocks) 2716 sbi->current_reserved_blocks = min(sbi->reserved_blocks, 2717 sbi->current_reserved_blocks + count); 2718 spin_unlock(&sbi->stat_lock); 2719 if (unlikely(inode->i_blocks < sectors)) { 2720 f2fs_warn(sbi, "Inconsistent i_blocks, ino:%lu, iblocks:%llu, sectors:%llu", 2721 inode->i_ino, 2722 (unsigned long long)inode->i_blocks, 2723 (unsigned long long)sectors); 2724 set_sbi_flag(sbi, SBI_NEED_FSCK); 2725 return; 2726 } 2727 f2fs_i_blocks_write(inode, count, false, true); 2728 } 2729 2730 static inline void inc_page_count(struct f2fs_sb_info *sbi, int count_type) 2731 { 2732 atomic_inc(&sbi->nr_pages[count_type]); 2733 2734 if (count_type == F2FS_DIRTY_DENTS || 2735 count_type == F2FS_DIRTY_NODES || 2736 count_type == F2FS_DIRTY_META || 2737 count_type == F2FS_DIRTY_QDATA || 2738 count_type == F2FS_DIRTY_IMETA) 2739 set_sbi_flag(sbi, SBI_IS_DIRTY); 2740 } 2741 2742 static inline void inode_inc_dirty_pages(struct inode *inode) 2743 { 2744 atomic_inc(&F2FS_I(inode)->dirty_pages); 2745 inc_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ? 2746 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA); 2747 if (IS_NOQUOTA(inode)) 2748 inc_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA); 2749 } 2750 2751 static inline void dec_page_count(struct f2fs_sb_info *sbi, int count_type) 2752 { 2753 atomic_dec(&sbi->nr_pages[count_type]); 2754 } 2755 2756 static inline void inode_dec_dirty_pages(struct inode *inode) 2757 { 2758 if (!S_ISDIR(inode->i_mode) && !S_ISREG(inode->i_mode) && 2759 !S_ISLNK(inode->i_mode)) 2760 return; 2761 2762 atomic_dec(&F2FS_I(inode)->dirty_pages); 2763 dec_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ? 2764 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA); 2765 if (IS_NOQUOTA(inode)) 2766 dec_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA); 2767 } 2768 2769 static inline void inc_atomic_write_cnt(struct inode *inode) 2770 { 2771 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 2772 struct f2fs_inode_info *fi = F2FS_I(inode); 2773 u64 current_write; 2774 2775 fi->atomic_write_cnt++; 2776 atomic64_inc(&sbi->current_atomic_write); 2777 current_write = atomic64_read(&sbi->current_atomic_write); 2778 if (current_write > sbi->peak_atomic_write) 2779 sbi->peak_atomic_write = current_write; 2780 } 2781 2782 static inline void release_atomic_write_cnt(struct inode *inode) 2783 { 2784 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 2785 struct f2fs_inode_info *fi = F2FS_I(inode); 2786 2787 atomic64_sub(fi->atomic_write_cnt, &sbi->current_atomic_write); 2788 fi->atomic_write_cnt = 0; 2789 } 2790 2791 static inline s64 get_pages(struct f2fs_sb_info *sbi, int count_type) 2792 { 2793 return atomic_read(&sbi->nr_pages[count_type]); 2794 } 2795 2796 static inline int get_dirty_pages(struct inode *inode) 2797 { 2798 return atomic_read(&F2FS_I(inode)->dirty_pages); 2799 } 2800 2801 static inline int get_blocktype_secs(struct f2fs_sb_info *sbi, int block_type) 2802 { 2803 return div_u64(get_pages(sbi, block_type) + BLKS_PER_SEC(sbi) - 1, 2804 BLKS_PER_SEC(sbi)); 2805 } 2806 2807 static inline block_t valid_user_blocks(struct f2fs_sb_info *sbi) 2808 { 2809 return sbi->total_valid_block_count; 2810 } 2811 2812 static inline block_t discard_blocks(struct f2fs_sb_info *sbi) 2813 { 2814 return sbi->discard_blks; 2815 } 2816 2817 static inline unsigned long __bitmap_size(struct f2fs_sb_info *sbi, int flag) 2818 { 2819 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi); 2820 2821 /* return NAT or SIT bitmap */ 2822 if (flag == NAT_BITMAP) 2823 return le32_to_cpu(ckpt->nat_ver_bitmap_bytesize); 2824 else if (flag == SIT_BITMAP) 2825 return le32_to_cpu(ckpt->sit_ver_bitmap_bytesize); 2826 2827 return 0; 2828 } 2829 2830 static inline block_t __cp_payload(struct f2fs_sb_info *sbi) 2831 { 2832 return le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_payload); 2833 } 2834 2835 static inline void *__bitmap_ptr(struct f2fs_sb_info *sbi, int flag) 2836 { 2837 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi); 2838 void *tmp_ptr = &ckpt->sit_nat_version_bitmap; 2839 int offset; 2840 2841 if (is_set_ckpt_flags(sbi, CP_LARGE_NAT_BITMAP_FLAG)) { 2842 offset = (flag == SIT_BITMAP) ? 2843 le32_to_cpu(ckpt->nat_ver_bitmap_bytesize) : 0; 2844 /* 2845 * if large_nat_bitmap feature is enabled, leave checksum 2846 * protection for all nat/sit bitmaps. 2847 */ 2848 return tmp_ptr + offset + sizeof(__le32); 2849 } 2850 2851 if (__cp_payload(sbi) > 0) { 2852 if (flag == NAT_BITMAP) 2853 return tmp_ptr; 2854 else 2855 return (unsigned char *)ckpt + F2FS_BLKSIZE; 2856 } else { 2857 offset = (flag == NAT_BITMAP) ? 2858 le32_to_cpu(ckpt->sit_ver_bitmap_bytesize) : 0; 2859 return tmp_ptr + offset; 2860 } 2861 } 2862 2863 static inline block_t __start_cp_addr(struct f2fs_sb_info *sbi) 2864 { 2865 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr); 2866 2867 if (sbi->cur_cp_pack == 2) 2868 start_addr += BLKS_PER_SEG(sbi); 2869 return start_addr; 2870 } 2871 2872 static inline block_t __start_cp_next_addr(struct f2fs_sb_info *sbi) 2873 { 2874 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr); 2875 2876 if (sbi->cur_cp_pack == 1) 2877 start_addr += BLKS_PER_SEG(sbi); 2878 return start_addr; 2879 } 2880 2881 static inline void __set_cp_next_pack(struct f2fs_sb_info *sbi) 2882 { 2883 sbi->cur_cp_pack = (sbi->cur_cp_pack == 1) ? 2 : 1; 2884 } 2885 2886 static inline block_t __start_sum_addr(struct f2fs_sb_info *sbi) 2887 { 2888 return le32_to_cpu(F2FS_CKPT(sbi)->cp_pack_start_sum); 2889 } 2890 2891 static inline bool __has_cursum_space(struct f2fs_sb_info *sbi, 2892 struct f2fs_journal *journal, unsigned int size, int type) 2893 { 2894 if (type == NAT_JOURNAL) 2895 return size <= MAX_NAT_JENTRIES(sbi, journal); 2896 return size <= MAX_SIT_JENTRIES(sbi, journal); 2897 } 2898 2899 extern void f2fs_mark_inode_dirty_sync(struct inode *inode, bool sync); 2900 static inline int inc_valid_node_count(struct f2fs_sb_info *sbi, 2901 struct inode *inode, bool is_inode) 2902 { 2903 block_t valid_block_count; 2904 unsigned int valid_node_count, avail_user_node_count; 2905 unsigned int avail_user_block_count; 2906 int err; 2907 2908 if (is_inode) { 2909 if (inode) { 2910 err = dquot_alloc_inode(inode); 2911 if (err) 2912 return err; 2913 } 2914 } else { 2915 err = dquot_reserve_block(inode, 1); 2916 if (err) 2917 return err; 2918 } 2919 2920 if (time_to_inject(sbi, FAULT_BLOCK)) 2921 goto enospc; 2922 2923 spin_lock(&sbi->stat_lock); 2924 2925 valid_block_count = sbi->total_valid_block_count + 1; 2926 avail_user_block_count = get_available_block_count(sbi, inode, 2927 test_opt(sbi, RESERVE_NODE)); 2928 2929 if (unlikely(valid_block_count > avail_user_block_count)) { 2930 spin_unlock(&sbi->stat_lock); 2931 goto enospc; 2932 } 2933 2934 avail_user_node_count = sbi->total_node_count - F2FS_RESERVED_NODE_NUM; 2935 if (test_opt(sbi, RESERVE_NODE) && 2936 !__allow_reserved_root(sbi, inode, true)) 2937 avail_user_node_count -= F2FS_OPTION(sbi).root_reserved_nodes; 2938 valid_node_count = sbi->total_valid_node_count + 1; 2939 if (unlikely(valid_node_count > avail_user_node_count)) { 2940 spin_unlock(&sbi->stat_lock); 2941 goto enospc; 2942 } 2943 2944 sbi->total_valid_node_count++; 2945 sbi->total_valid_block_count++; 2946 spin_unlock(&sbi->stat_lock); 2947 2948 if (inode) { 2949 if (is_inode) 2950 f2fs_mark_inode_dirty_sync(inode, true); 2951 else 2952 f2fs_i_blocks_write(inode, 1, true, true); 2953 } 2954 2955 percpu_counter_inc(&sbi->alloc_valid_block_count); 2956 return 0; 2957 2958 enospc: 2959 if (is_inode) { 2960 if (inode) 2961 dquot_free_inode(inode); 2962 } else { 2963 dquot_release_reservation_block(inode, 1); 2964 } 2965 return -ENOSPC; 2966 } 2967 2968 static inline void dec_valid_node_count(struct f2fs_sb_info *sbi, 2969 struct inode *inode, bool is_inode) 2970 { 2971 spin_lock(&sbi->stat_lock); 2972 2973 if (unlikely(!sbi->total_valid_block_count || 2974 !sbi->total_valid_node_count)) { 2975 f2fs_warn(sbi, "dec_valid_node_count: inconsistent block counts, total_valid_block:%u, total_valid_node:%u", 2976 sbi->total_valid_block_count, 2977 sbi->total_valid_node_count); 2978 set_sbi_flag(sbi, SBI_NEED_FSCK); 2979 } else { 2980 sbi->total_valid_block_count--; 2981 sbi->total_valid_node_count--; 2982 } 2983 2984 if (sbi->reserved_blocks && 2985 sbi->current_reserved_blocks < sbi->reserved_blocks) 2986 sbi->current_reserved_blocks++; 2987 2988 spin_unlock(&sbi->stat_lock); 2989 2990 if (is_inode) { 2991 dquot_free_inode(inode); 2992 } else { 2993 if (unlikely(inode->i_blocks == 0)) { 2994 f2fs_warn(sbi, "dec_valid_node_count: inconsistent i_blocks, ino:%lu, iblocks:%llu", 2995 inode->i_ino, 2996 (unsigned long long)inode->i_blocks); 2997 set_sbi_flag(sbi, SBI_NEED_FSCK); 2998 return; 2999 } 3000 f2fs_i_blocks_write(inode, 1, false, true); 3001 } 3002 } 3003 3004 static inline unsigned int valid_node_count(struct f2fs_sb_info *sbi) 3005 { 3006 return sbi->total_valid_node_count; 3007 } 3008 3009 static inline void inc_valid_inode_count(struct f2fs_sb_info *sbi) 3010 { 3011 percpu_counter_inc(&sbi->total_valid_inode_count); 3012 } 3013 3014 static inline void dec_valid_inode_count(struct f2fs_sb_info *sbi) 3015 { 3016 percpu_counter_dec(&sbi->total_valid_inode_count); 3017 } 3018 3019 static inline s64 valid_inode_count(struct f2fs_sb_info *sbi) 3020 { 3021 return percpu_counter_sum_positive(&sbi->total_valid_inode_count); 3022 } 3023 3024 static inline struct folio *f2fs_grab_cache_folio(struct address_space *mapping, 3025 pgoff_t index, bool for_write) 3026 { 3027 struct folio *folio; 3028 unsigned int flags; 3029 3030 if (IS_ENABLED(CONFIG_F2FS_FAULT_INJECTION)) { 3031 fgf_t fgf_flags; 3032 3033 if (!for_write) 3034 fgf_flags = FGP_LOCK | FGP_ACCESSED; 3035 else 3036 fgf_flags = FGP_LOCK; 3037 folio = __filemap_get_folio(mapping, index, fgf_flags, 0); 3038 if (!IS_ERR(folio)) 3039 return folio; 3040 3041 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_ALLOC)) 3042 return ERR_PTR(-ENOMEM); 3043 } 3044 3045 if (!for_write) 3046 return filemap_grab_folio(mapping, index); 3047 3048 flags = memalloc_nofs_save(); 3049 folio = __filemap_get_folio(mapping, index, FGP_WRITEBEGIN, 3050 mapping_gfp_mask(mapping)); 3051 memalloc_nofs_restore(flags); 3052 3053 return folio; 3054 } 3055 3056 static inline struct folio *f2fs_filemap_get_folio( 3057 struct address_space *mapping, pgoff_t index, 3058 fgf_t fgp_flags, gfp_t gfp_mask) 3059 { 3060 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_GET)) 3061 return ERR_PTR(-ENOMEM); 3062 3063 return __filemap_get_folio(mapping, index, fgp_flags, gfp_mask); 3064 } 3065 3066 static inline void f2fs_folio_put(struct folio *folio, bool unlock) 3067 { 3068 if (IS_ERR_OR_NULL(folio)) 3069 return; 3070 3071 if (unlock) { 3072 f2fs_bug_on(F2FS_F_SB(folio), !folio_test_locked(folio)); 3073 folio_unlock(folio); 3074 } 3075 folio_put(folio); 3076 } 3077 3078 static inline void f2fs_put_page(struct page *page, bool unlock) 3079 { 3080 if (!page) 3081 return; 3082 f2fs_folio_put(page_folio(page), unlock); 3083 } 3084 3085 static inline void f2fs_put_dnode(struct dnode_of_data *dn) 3086 { 3087 if (dn->node_folio) 3088 f2fs_folio_put(dn->node_folio, true); 3089 if (dn->inode_folio && dn->node_folio != dn->inode_folio) 3090 f2fs_folio_put(dn->inode_folio, false); 3091 dn->node_folio = NULL; 3092 dn->inode_folio = NULL; 3093 } 3094 3095 static inline struct kmem_cache *f2fs_kmem_cache_create(const char *name, 3096 size_t size) 3097 { 3098 return kmem_cache_create(name, size, 0, SLAB_RECLAIM_ACCOUNT, NULL); 3099 } 3100 3101 static inline void *f2fs_kmem_cache_alloc_nofail(struct kmem_cache *cachep, 3102 gfp_t flags) 3103 { 3104 void *entry; 3105 3106 entry = kmem_cache_alloc(cachep, flags); 3107 if (!entry) 3108 entry = kmem_cache_alloc(cachep, flags | __GFP_NOFAIL); 3109 return entry; 3110 } 3111 3112 static inline void *f2fs_kmem_cache_alloc(struct kmem_cache *cachep, 3113 gfp_t flags, bool nofail, struct f2fs_sb_info *sbi) 3114 { 3115 if (nofail) 3116 return f2fs_kmem_cache_alloc_nofail(cachep, flags); 3117 3118 if (time_to_inject(sbi, FAULT_SLAB_ALLOC)) 3119 return NULL; 3120 3121 return kmem_cache_alloc(cachep, flags); 3122 } 3123 3124 static inline bool is_inflight_io(struct f2fs_sb_info *sbi, int type) 3125 { 3126 if (get_pages(sbi, F2FS_RD_DATA) || get_pages(sbi, F2FS_RD_NODE) || 3127 get_pages(sbi, F2FS_RD_META) || get_pages(sbi, F2FS_WB_DATA) || 3128 get_pages(sbi, F2FS_WB_CP_DATA) || 3129 get_pages(sbi, F2FS_DIO_READ) || 3130 get_pages(sbi, F2FS_DIO_WRITE)) 3131 return true; 3132 3133 if (type != DISCARD_TIME && SM_I(sbi) && SM_I(sbi)->dcc_info && 3134 atomic_read(&SM_I(sbi)->dcc_info->queued_discard)) 3135 return true; 3136 3137 if (SM_I(sbi) && SM_I(sbi)->fcc_info && 3138 atomic_read(&SM_I(sbi)->fcc_info->queued_flush)) 3139 return true; 3140 return false; 3141 } 3142 3143 static inline bool is_inflight_read_io(struct f2fs_sb_info *sbi) 3144 { 3145 return get_pages(sbi, F2FS_RD_DATA) || get_pages(sbi, F2FS_DIO_READ); 3146 } 3147 3148 static inline bool is_idle(struct f2fs_sb_info *sbi, int type) 3149 { 3150 bool zoned_gc = (type == GC_TIME && 3151 F2FS_HAS_FEATURE(sbi, F2FS_FEATURE_BLKZONED)); 3152 3153 if (sbi->gc_mode == GC_URGENT_HIGH) 3154 return true; 3155 3156 if (sbi->bggc_io_aware == AWARE_READ_IO && is_inflight_read_io(sbi)) 3157 return false; 3158 if (sbi->bggc_io_aware == AWARE_ALL_IO && is_inflight_io(sbi, type)) 3159 return false; 3160 3161 if (sbi->gc_mode == GC_URGENT_MID) 3162 return true; 3163 3164 if (sbi->gc_mode == GC_URGENT_LOW && 3165 (type == DISCARD_TIME || type == GC_TIME)) 3166 return true; 3167 3168 if (zoned_gc) 3169 return true; 3170 3171 return f2fs_time_over(sbi, type); 3172 } 3173 3174 static inline void f2fs_radix_tree_insert(struct radix_tree_root *root, 3175 unsigned long index, void *item) 3176 { 3177 while (radix_tree_insert(root, index, item)) 3178 cond_resched(); 3179 } 3180 3181 #define RAW_IS_INODE(p) ((p)->footer.nid == (p)->footer.ino) 3182 3183 static inline bool IS_INODE(const struct folio *folio) 3184 { 3185 struct f2fs_node *p = F2FS_NODE(folio); 3186 3187 return RAW_IS_INODE(p); 3188 } 3189 3190 static inline int offset_in_addr(struct f2fs_inode *i) 3191 { 3192 return (i->i_inline & F2FS_EXTRA_ATTR) ? 3193 (le16_to_cpu(i->i_extra_isize) / sizeof(__le32)) : 0; 3194 } 3195 3196 static inline __le32 *blkaddr_in_node(struct f2fs_node *node) 3197 { 3198 return RAW_IS_INODE(node) ? node->i.i_addr : node->dn.addr; 3199 } 3200 3201 static inline int f2fs_has_extra_attr(struct inode *inode); 3202 static inline unsigned int get_dnode_base(struct inode *inode, 3203 struct folio *node_folio) 3204 { 3205 if (!IS_INODE(node_folio)) 3206 return 0; 3207 3208 return inode ? get_extra_isize(inode) : 3209 offset_in_addr(&F2FS_NODE(node_folio)->i); 3210 } 3211 3212 static inline __le32 *get_dnode_addr(struct inode *inode, 3213 struct folio *node_folio) 3214 { 3215 return blkaddr_in_node(F2FS_NODE(node_folio)) + 3216 get_dnode_base(inode, node_folio); 3217 } 3218 3219 static inline block_t data_blkaddr(struct inode *inode, 3220 struct folio *node_folio, unsigned int offset) 3221 { 3222 return le32_to_cpu(*(get_dnode_addr(inode, node_folio) + offset)); 3223 } 3224 3225 static inline block_t f2fs_data_blkaddr(struct dnode_of_data *dn) 3226 { 3227 return data_blkaddr(dn->inode, dn->node_folio, dn->ofs_in_node); 3228 } 3229 3230 static inline int f2fs_test_bit(unsigned int nr, char *addr) 3231 { 3232 int mask; 3233 3234 addr += (nr >> 3); 3235 mask = BIT(7 - (nr & 0x07)); 3236 return mask & *addr; 3237 } 3238 3239 static inline void f2fs_set_bit(unsigned int nr, char *addr) 3240 { 3241 int mask; 3242 3243 addr += (nr >> 3); 3244 mask = BIT(7 - (nr & 0x07)); 3245 *addr |= mask; 3246 } 3247 3248 static inline void f2fs_clear_bit(unsigned int nr, char *addr) 3249 { 3250 int mask; 3251 3252 addr += (nr >> 3); 3253 mask = BIT(7 - (nr & 0x07)); 3254 *addr &= ~mask; 3255 } 3256 3257 static inline int f2fs_test_and_set_bit(unsigned int nr, char *addr) 3258 { 3259 int mask; 3260 int ret; 3261 3262 addr += (nr >> 3); 3263 mask = BIT(7 - (nr & 0x07)); 3264 ret = mask & *addr; 3265 *addr |= mask; 3266 return ret; 3267 } 3268 3269 static inline int f2fs_test_and_clear_bit(unsigned int nr, char *addr) 3270 { 3271 int mask; 3272 int ret; 3273 3274 addr += (nr >> 3); 3275 mask = BIT(7 - (nr & 0x07)); 3276 ret = mask & *addr; 3277 *addr &= ~mask; 3278 return ret; 3279 } 3280 3281 static inline void f2fs_change_bit(unsigned int nr, char *addr) 3282 { 3283 int mask; 3284 3285 addr += (nr >> 3); 3286 mask = BIT(7 - (nr & 0x07)); 3287 *addr ^= mask; 3288 } 3289 3290 /* 3291 * On-disk inode flags (f2fs_inode::i_flags) 3292 */ 3293 #define F2FS_COMPR_FL 0x00000004 /* Compress file */ 3294 #define F2FS_SYNC_FL 0x00000008 /* Synchronous updates */ 3295 #define F2FS_IMMUTABLE_FL 0x00000010 /* Immutable file */ 3296 #define F2FS_APPEND_FL 0x00000020 /* writes to file may only append */ 3297 #define F2FS_NODUMP_FL 0x00000040 /* do not dump file */ 3298 #define F2FS_NOATIME_FL 0x00000080 /* do not update atime */ 3299 #define F2FS_NOCOMP_FL 0x00000400 /* Don't compress */ 3300 #define F2FS_INDEX_FL 0x00001000 /* hash-indexed directory */ 3301 #define F2FS_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */ 3302 #define F2FS_PROJINHERIT_FL 0x20000000 /* Create with parents projid */ 3303 #define F2FS_CASEFOLD_FL 0x40000000 /* Casefolded file */ 3304 #define F2FS_DEVICE_ALIAS_FL 0x80000000 /* File for aliasing a device */ 3305 3306 #define F2FS_QUOTA_DEFAULT_FL (F2FS_NOATIME_FL | F2FS_IMMUTABLE_FL) 3307 3308 /* Flags that should be inherited by new inodes from their parent. */ 3309 #define F2FS_FL_INHERITED (F2FS_SYNC_FL | F2FS_NODUMP_FL | F2FS_NOATIME_FL | \ 3310 F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \ 3311 F2FS_CASEFOLD_FL) 3312 3313 /* Flags that are appropriate for regular files (all but dir-specific ones). */ 3314 #define F2FS_REG_FLMASK (~(F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \ 3315 F2FS_CASEFOLD_FL)) 3316 3317 /* Flags that are appropriate for non-directories/regular files. */ 3318 #define F2FS_OTHER_FLMASK (F2FS_NODUMP_FL | F2FS_NOATIME_FL) 3319 3320 #define IS_DEVICE_ALIASING(inode) (F2FS_I(inode)->i_flags & F2FS_DEVICE_ALIAS_FL) 3321 3322 static inline __u32 f2fs_mask_flags(umode_t mode, __u32 flags) 3323 { 3324 if (S_ISDIR(mode)) 3325 return flags; 3326 else if (S_ISREG(mode)) 3327 return flags & F2FS_REG_FLMASK; 3328 else 3329 return flags & F2FS_OTHER_FLMASK; 3330 } 3331 3332 static inline void __mark_inode_dirty_flag(struct inode *inode, 3333 int flag, bool set) 3334 { 3335 switch (flag) { 3336 case FI_INLINE_XATTR: 3337 case FI_INLINE_DATA: 3338 case FI_INLINE_DENTRY: 3339 case FI_NEW_INODE: 3340 if (set) 3341 return; 3342 fallthrough; 3343 case FI_DATA_EXIST: 3344 case FI_PIN_FILE: 3345 case FI_COMPRESS_RELEASED: 3346 f2fs_mark_inode_dirty_sync(inode, true); 3347 } 3348 } 3349 3350 static inline void set_inode_flag(struct inode *inode, int flag) 3351 { 3352 set_bit(flag, F2FS_I(inode)->flags); 3353 __mark_inode_dirty_flag(inode, flag, true); 3354 } 3355 3356 static inline int is_inode_flag_set(struct inode *inode, int flag) 3357 { 3358 return test_bit(flag, F2FS_I(inode)->flags); 3359 } 3360 3361 static inline void clear_inode_flag(struct inode *inode, int flag) 3362 { 3363 clear_bit(flag, F2FS_I(inode)->flags); 3364 __mark_inode_dirty_flag(inode, flag, false); 3365 } 3366 3367 static inline bool f2fs_verity_in_progress(struct inode *inode) 3368 { 3369 return IS_ENABLED(CONFIG_FS_VERITY) && 3370 is_inode_flag_set(inode, FI_VERITY_IN_PROGRESS); 3371 } 3372 3373 static inline void set_acl_inode(struct inode *inode, umode_t mode) 3374 { 3375 F2FS_I(inode)->i_acl_mode = mode; 3376 set_inode_flag(inode, FI_ACL_MODE); 3377 f2fs_mark_inode_dirty_sync(inode, false); 3378 } 3379 3380 static inline void f2fs_i_links_write(struct inode *inode, bool inc) 3381 { 3382 if (inc) 3383 inc_nlink(inode); 3384 else 3385 drop_nlink(inode); 3386 f2fs_mark_inode_dirty_sync(inode, true); 3387 } 3388 3389 static inline void f2fs_i_blocks_write(struct inode *inode, 3390 block_t diff, bool add, bool claim) 3391 { 3392 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE); 3393 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER); 3394 3395 /* add = 1, claim = 1 should be dquot_reserve_block in pair */ 3396 if (add) { 3397 if (claim) 3398 dquot_claim_block(inode, diff); 3399 else 3400 dquot_alloc_block_nofail(inode, diff); 3401 } else { 3402 dquot_free_block(inode, diff); 3403 } 3404 3405 f2fs_mark_inode_dirty_sync(inode, true); 3406 if (clean || recover) 3407 set_inode_flag(inode, FI_AUTO_RECOVER); 3408 } 3409 3410 static inline bool f2fs_is_atomic_file(struct inode *inode); 3411 3412 static inline void f2fs_i_size_write(struct inode *inode, loff_t i_size) 3413 { 3414 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE); 3415 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER); 3416 3417 if (i_size_read(inode) == i_size) 3418 return; 3419 3420 i_size_write(inode, i_size); 3421 3422 if (f2fs_is_atomic_file(inode)) 3423 return; 3424 3425 f2fs_mark_inode_dirty_sync(inode, true); 3426 if (clean || recover) 3427 set_inode_flag(inode, FI_AUTO_RECOVER); 3428 } 3429 3430 static inline void f2fs_i_depth_write(struct inode *inode, unsigned int depth) 3431 { 3432 F2FS_I(inode)->i_current_depth = depth; 3433 f2fs_mark_inode_dirty_sync(inode, true); 3434 } 3435 3436 static inline void f2fs_i_gc_failures_write(struct inode *inode, 3437 unsigned int count) 3438 { 3439 F2FS_I(inode)->i_gc_failures = count; 3440 f2fs_mark_inode_dirty_sync(inode, true); 3441 } 3442 3443 static inline void f2fs_i_xnid_write(struct inode *inode, nid_t xnid) 3444 { 3445 F2FS_I(inode)->i_xattr_nid = xnid; 3446 f2fs_mark_inode_dirty_sync(inode, true); 3447 } 3448 3449 static inline void f2fs_i_pino_write(struct inode *inode, nid_t pino) 3450 { 3451 F2FS_I(inode)->i_pino = pino; 3452 f2fs_mark_inode_dirty_sync(inode, true); 3453 } 3454 3455 static inline void get_inline_info(struct inode *inode, struct f2fs_inode *ri) 3456 { 3457 struct f2fs_inode_info *fi = F2FS_I(inode); 3458 3459 if (ri->i_inline & F2FS_INLINE_XATTR) 3460 set_bit(FI_INLINE_XATTR, fi->flags); 3461 if (ri->i_inline & F2FS_INLINE_DATA) 3462 set_bit(FI_INLINE_DATA, fi->flags); 3463 if (ri->i_inline & F2FS_INLINE_DENTRY) 3464 set_bit(FI_INLINE_DENTRY, fi->flags); 3465 if (ri->i_inline & F2FS_DATA_EXIST) 3466 set_bit(FI_DATA_EXIST, fi->flags); 3467 if (ri->i_inline & F2FS_EXTRA_ATTR) 3468 set_bit(FI_EXTRA_ATTR, fi->flags); 3469 if (ri->i_inline & F2FS_PIN_FILE) 3470 set_bit(FI_PIN_FILE, fi->flags); 3471 if (ri->i_inline & F2FS_COMPRESS_RELEASED) 3472 set_bit(FI_COMPRESS_RELEASED, fi->flags); 3473 } 3474 3475 static inline void set_raw_inline(struct inode *inode, struct f2fs_inode *ri) 3476 { 3477 ri->i_inline = 0; 3478 3479 if (is_inode_flag_set(inode, FI_INLINE_XATTR)) 3480 ri->i_inline |= F2FS_INLINE_XATTR; 3481 if (is_inode_flag_set(inode, FI_INLINE_DATA)) 3482 ri->i_inline |= F2FS_INLINE_DATA; 3483 if (is_inode_flag_set(inode, FI_INLINE_DENTRY)) 3484 ri->i_inline |= F2FS_INLINE_DENTRY; 3485 if (is_inode_flag_set(inode, FI_DATA_EXIST)) 3486 ri->i_inline |= F2FS_DATA_EXIST; 3487 if (is_inode_flag_set(inode, FI_EXTRA_ATTR)) 3488 ri->i_inline |= F2FS_EXTRA_ATTR; 3489 if (is_inode_flag_set(inode, FI_PIN_FILE)) 3490 ri->i_inline |= F2FS_PIN_FILE; 3491 if (is_inode_flag_set(inode, FI_COMPRESS_RELEASED)) 3492 ri->i_inline |= F2FS_COMPRESS_RELEASED; 3493 } 3494 3495 static inline int f2fs_has_extra_attr(struct inode *inode) 3496 { 3497 return is_inode_flag_set(inode, FI_EXTRA_ATTR); 3498 } 3499 3500 static inline int f2fs_has_inline_xattr(struct inode *inode) 3501 { 3502 return is_inode_flag_set(inode, FI_INLINE_XATTR); 3503 } 3504 3505 static inline int f2fs_compressed_file(struct inode *inode) 3506 { 3507 return S_ISREG(inode->i_mode) && 3508 is_inode_flag_set(inode, FI_COMPRESSED_FILE); 3509 } 3510 3511 static inline bool f2fs_need_compress_data(struct inode *inode) 3512 { 3513 int compress_mode = F2FS_OPTION(F2FS_I_SB(inode)).compress_mode; 3514 3515 if (!f2fs_compressed_file(inode)) 3516 return false; 3517 3518 if (compress_mode == COMPR_MODE_FS) 3519 return true; 3520 else if (compress_mode == COMPR_MODE_USER && 3521 is_inode_flag_set(inode, FI_ENABLE_COMPRESS)) 3522 return true; 3523 3524 return false; 3525 } 3526 3527 static inline unsigned int addrs_per_page(struct inode *inode, 3528 bool is_inode) 3529 { 3530 unsigned int addrs = is_inode ? (CUR_ADDRS_PER_INODE(inode) - 3531 get_inline_xattr_addrs(inode)) : DEF_ADDRS_PER_BLOCK; 3532 3533 if (f2fs_compressed_file(inode)) 3534 return ALIGN_DOWN(addrs, F2FS_I(inode)->i_cluster_size); 3535 return addrs; 3536 } 3537 3538 static inline 3539 void *inline_xattr_addr(struct inode *inode, const struct folio *folio) 3540 { 3541 struct f2fs_inode *ri = F2FS_INODE(folio); 3542 3543 return (void *)&(ri->i_addr[DEF_ADDRS_PER_INODE - 3544 get_inline_xattr_addrs(inode)]); 3545 } 3546 3547 static inline int inline_xattr_size(struct inode *inode) 3548 { 3549 if (f2fs_has_inline_xattr(inode)) 3550 return get_inline_xattr_addrs(inode) * sizeof(__le32); 3551 return 0; 3552 } 3553 3554 /* 3555 * Notice: check inline_data flag without inode page lock is unsafe. 3556 * It could change at any time by f2fs_convert_inline_folio(). 3557 */ 3558 static inline int f2fs_has_inline_data(struct inode *inode) 3559 { 3560 return is_inode_flag_set(inode, FI_INLINE_DATA); 3561 } 3562 3563 static inline int f2fs_exist_data(struct inode *inode) 3564 { 3565 return is_inode_flag_set(inode, FI_DATA_EXIST); 3566 } 3567 3568 static inline int f2fs_is_mmap_file(struct inode *inode) 3569 { 3570 return is_inode_flag_set(inode, FI_MMAP_FILE); 3571 } 3572 3573 static inline bool f2fs_is_pinned_file(struct inode *inode) 3574 { 3575 return is_inode_flag_set(inode, FI_PIN_FILE); 3576 } 3577 3578 static inline bool f2fs_is_atomic_file(struct inode *inode) 3579 { 3580 return is_inode_flag_set(inode, FI_ATOMIC_FILE); 3581 } 3582 3583 static inline bool f2fs_is_cow_file(struct inode *inode) 3584 { 3585 return is_inode_flag_set(inode, FI_COW_FILE); 3586 } 3587 3588 static inline void *inline_data_addr(struct inode *inode, struct folio *folio) 3589 { 3590 __le32 *addr = get_dnode_addr(inode, folio); 3591 3592 return (void *)(addr + DEF_INLINE_RESERVED_SIZE); 3593 } 3594 3595 static inline int f2fs_has_inline_dentry(struct inode *inode) 3596 { 3597 return is_inode_flag_set(inode, FI_INLINE_DENTRY); 3598 } 3599 3600 static inline int is_file(struct inode *inode, int type) 3601 { 3602 return F2FS_I(inode)->i_advise & type; 3603 } 3604 3605 static inline void set_file(struct inode *inode, int type) 3606 { 3607 if (is_file(inode, type)) 3608 return; 3609 F2FS_I(inode)->i_advise |= type; 3610 f2fs_mark_inode_dirty_sync(inode, true); 3611 } 3612 3613 static inline void clear_file(struct inode *inode, int type) 3614 { 3615 if (!is_file(inode, type)) 3616 return; 3617 F2FS_I(inode)->i_advise &= ~type; 3618 f2fs_mark_inode_dirty_sync(inode, true); 3619 } 3620 3621 static inline bool f2fs_is_time_consistent(struct inode *inode) 3622 { 3623 struct timespec64 ts = inode_get_atime(inode); 3624 3625 if (!timespec64_equal(F2FS_I(inode)->i_disk_time, &ts)) 3626 return false; 3627 ts = inode_get_ctime(inode); 3628 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 1, &ts)) 3629 return false; 3630 ts = inode_get_mtime(inode); 3631 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 2, &ts)) 3632 return false; 3633 return true; 3634 } 3635 3636 static inline bool f2fs_skip_inode_update(struct inode *inode, int dsync) 3637 { 3638 bool ret; 3639 3640 if (dsync) { 3641 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 3642 3643 spin_lock(&sbi->inode_lock[DIRTY_META]); 3644 ret = list_empty(&F2FS_I(inode)->gdirty_list); 3645 spin_unlock(&sbi->inode_lock[DIRTY_META]); 3646 return ret; 3647 } 3648 if (!is_inode_flag_set(inode, FI_AUTO_RECOVER) || 3649 file_keep_isize(inode) || 3650 i_size_read(inode) & ~PAGE_MASK) 3651 return false; 3652 3653 if (!f2fs_is_time_consistent(inode)) 3654 return false; 3655 3656 spin_lock(&F2FS_I(inode)->i_size_lock); 3657 ret = F2FS_I(inode)->last_disk_size == i_size_read(inode); 3658 spin_unlock(&F2FS_I(inode)->i_size_lock); 3659 3660 return ret; 3661 } 3662 3663 static inline bool f2fs_readonly(struct super_block *sb) 3664 { 3665 return sb_rdonly(sb); 3666 } 3667 3668 static inline bool f2fs_cp_error(struct f2fs_sb_info *sbi) 3669 { 3670 return is_set_ckpt_flags(sbi, CP_ERROR_FLAG); 3671 } 3672 3673 static inline void *f2fs_kmalloc(struct f2fs_sb_info *sbi, 3674 size_t size, gfp_t flags) 3675 { 3676 if (time_to_inject(sbi, FAULT_KMALLOC)) 3677 return NULL; 3678 3679 return kmalloc(size, flags); 3680 } 3681 3682 static inline void *f2fs_getname(struct f2fs_sb_info *sbi) 3683 { 3684 if (time_to_inject(sbi, FAULT_KMALLOC)) 3685 return NULL; 3686 3687 return __getname(); 3688 } 3689 3690 static inline void f2fs_putname(char *buf) 3691 { 3692 __putname(buf); 3693 } 3694 3695 static inline void *f2fs_kzalloc(struct f2fs_sb_info *sbi, 3696 size_t size, gfp_t flags) 3697 { 3698 return f2fs_kmalloc(sbi, size, flags | __GFP_ZERO); 3699 } 3700 3701 static inline void *f2fs_kvmalloc(struct f2fs_sb_info *sbi, 3702 size_t size, gfp_t flags) 3703 { 3704 if (time_to_inject(sbi, FAULT_KVMALLOC)) 3705 return NULL; 3706 3707 return kvmalloc(size, flags); 3708 } 3709 3710 static inline void *f2fs_kvzalloc(struct f2fs_sb_info *sbi, 3711 size_t size, gfp_t flags) 3712 { 3713 return f2fs_kvmalloc(sbi, size, flags | __GFP_ZERO); 3714 } 3715 3716 static inline void *f2fs_vmalloc(struct f2fs_sb_info *sbi, size_t size) 3717 { 3718 if (time_to_inject(sbi, FAULT_VMALLOC)) 3719 return NULL; 3720 3721 return vmalloc(size); 3722 } 3723 3724 static inline int get_extra_isize(struct inode *inode) 3725 { 3726 return F2FS_I(inode)->i_extra_isize / sizeof(__le32); 3727 } 3728 3729 static inline int get_inline_xattr_addrs(struct inode *inode) 3730 { 3731 return F2FS_I(inode)->i_inline_xattr_size; 3732 } 3733 3734 #define f2fs_get_inode_mode(i) \ 3735 ((is_inode_flag_set(i, FI_ACL_MODE)) ? \ 3736 (F2FS_I(i)->i_acl_mode) : ((i)->i_mode)) 3737 3738 #define F2FS_MIN_EXTRA_ATTR_SIZE (sizeof(__le32)) 3739 3740 #define F2FS_TOTAL_EXTRA_ATTR_SIZE \ 3741 (offsetof(struct f2fs_inode, i_extra_end) - \ 3742 offsetof(struct f2fs_inode, i_extra_isize)) \ 3743 3744 #define F2FS_OLD_ATTRIBUTE_SIZE (offsetof(struct f2fs_inode, i_addr)) 3745 #define F2FS_FITS_IN_INODE(f2fs_inode, extra_isize, field) \ 3746 ((offsetof(typeof(*(f2fs_inode)), field) + \ 3747 sizeof((f2fs_inode)->field)) \ 3748 <= (F2FS_OLD_ATTRIBUTE_SIZE + (extra_isize))) \ 3749 3750 #define __is_large_section(sbi) (SEGS_PER_SEC(sbi) > 1) 3751 3752 #define __is_meta_io(fio) (PAGE_TYPE_OF_BIO((fio)->type) == META) 3753 3754 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi, 3755 block_t blkaddr, int type); 3756 static inline void verify_blkaddr(struct f2fs_sb_info *sbi, 3757 block_t blkaddr, int type) 3758 { 3759 if (!f2fs_is_valid_blkaddr(sbi, blkaddr, type)) 3760 f2fs_err(sbi, "invalid blkaddr: %u, type: %d, run fsck to fix.", 3761 blkaddr, type); 3762 } 3763 3764 static inline bool __is_valid_data_blkaddr(block_t blkaddr) 3765 { 3766 if (blkaddr == NEW_ADDR || blkaddr == NULL_ADDR || 3767 blkaddr == COMPRESS_ADDR) 3768 return false; 3769 return true; 3770 } 3771 3772 /* 3773 * file.c 3774 */ 3775 int f2fs_sync_file(struct file *file, loff_t start, loff_t end, int datasync); 3776 int f2fs_do_truncate_blocks(struct inode *inode, u64 from, bool lock); 3777 int f2fs_truncate_blocks(struct inode *inode, u64 from, bool lock); 3778 int f2fs_truncate(struct inode *inode); 3779 int f2fs_getattr(struct mnt_idmap *idmap, const struct path *path, 3780 struct kstat *stat, u32 request_mask, unsigned int flags); 3781 int f2fs_setattr(struct mnt_idmap *idmap, struct dentry *dentry, 3782 struct iattr *attr); 3783 int f2fs_truncate_hole(struct inode *inode, pgoff_t pg_start, pgoff_t pg_end); 3784 void f2fs_truncate_data_blocks_range(struct dnode_of_data *dn, int count); 3785 int f2fs_do_shutdown(struct f2fs_sb_info *sbi, unsigned int flag, 3786 bool readonly, bool need_lock); 3787 int f2fs_precache_extents(struct inode *inode); 3788 int f2fs_fileattr_get(struct dentry *dentry, struct file_kattr *fa); 3789 int f2fs_fileattr_set(struct mnt_idmap *idmap, 3790 struct dentry *dentry, struct file_kattr *fa); 3791 long f2fs_ioctl(struct file *filp, unsigned int cmd, unsigned long arg); 3792 long f2fs_compat_ioctl(struct file *file, unsigned int cmd, unsigned long arg); 3793 int f2fs_transfer_project_quota(struct inode *inode, kprojid_t kprojid); 3794 int f2fs_pin_file_control(struct inode *inode, bool inc); 3795 3796 /* 3797 * inode.c 3798 */ 3799 void f2fs_set_inode_flags(struct inode *inode); 3800 bool f2fs_inode_chksum_verify(struct f2fs_sb_info *sbi, struct folio *folio); 3801 void f2fs_inode_chksum_set(struct f2fs_sb_info *sbi, struct folio *folio); 3802 struct inode *f2fs_iget(struct super_block *sb, unsigned long ino); 3803 struct inode *f2fs_iget_retry(struct super_block *sb, unsigned long ino); 3804 int f2fs_try_to_free_nats(struct f2fs_sb_info *sbi, int nr_shrink); 3805 void f2fs_update_inode(struct inode *inode, struct folio *node_folio); 3806 void f2fs_update_inode_page(struct inode *inode); 3807 int f2fs_write_inode(struct inode *inode, struct writeback_control *wbc); 3808 void f2fs_remove_donate_inode(struct inode *inode); 3809 void f2fs_evict_inode(struct inode *inode); 3810 void f2fs_handle_failed_inode(struct inode *inode, struct f2fs_lock_context *lc); 3811 3812 /* 3813 * namei.c 3814 */ 3815 int f2fs_update_extension_list(struct f2fs_sb_info *sbi, const char *name, 3816 bool hot, bool set); 3817 struct dentry *f2fs_get_parent(struct dentry *child); 3818 int f2fs_get_tmpfile(struct mnt_idmap *idmap, struct inode *dir, 3819 struct inode **new_inode); 3820 3821 /* 3822 * dir.c 3823 */ 3824 #if IS_ENABLED(CONFIG_UNICODE) 3825 int f2fs_init_casefolded_name(const struct inode *dir, 3826 struct f2fs_filename *fname); 3827 void f2fs_free_casefolded_name(struct f2fs_filename *fname); 3828 #else 3829 static inline int f2fs_init_casefolded_name(const struct inode *dir, 3830 struct f2fs_filename *fname) 3831 { 3832 return 0; 3833 } 3834 3835 static inline void f2fs_free_casefolded_name(struct f2fs_filename *fname) 3836 { 3837 } 3838 #endif /* CONFIG_UNICODE */ 3839 3840 int f2fs_setup_filename(struct inode *dir, const struct qstr *iname, 3841 int lookup, struct f2fs_filename *fname); 3842 int f2fs_prepare_lookup(struct inode *dir, struct dentry *dentry, 3843 struct f2fs_filename *fname); 3844 void f2fs_free_filename(struct f2fs_filename *fname); 3845 struct f2fs_dir_entry *f2fs_find_target_dentry(const struct f2fs_dentry_ptr *d, 3846 const struct f2fs_filename *fname, int *max_slots, 3847 bool use_hash); 3848 int f2fs_fill_dentries(struct dir_context *ctx, struct f2fs_dentry_ptr *d, 3849 unsigned int start_pos, struct fscrypt_str *fstr); 3850 void f2fs_do_make_empty_dir(struct inode *inode, struct inode *parent, 3851 struct f2fs_dentry_ptr *d); 3852 struct folio *f2fs_init_inode_metadata(struct inode *inode, struct inode *dir, 3853 const struct f2fs_filename *fname, struct folio *dfolio); 3854 void f2fs_update_parent_metadata(struct inode *dir, struct inode *inode, 3855 unsigned int current_depth); 3856 int f2fs_room_for_filename(const void *bitmap, int slots, int max_slots); 3857 void f2fs_drop_nlink(struct inode *dir, struct inode *inode); 3858 struct f2fs_dir_entry *__f2fs_find_entry(struct inode *dir, 3859 const struct f2fs_filename *fname, struct folio **res_folio); 3860 struct f2fs_dir_entry *f2fs_find_entry(struct inode *dir, 3861 const struct qstr *child, struct folio **res_folio); 3862 struct f2fs_dir_entry *f2fs_parent_dir(struct inode *dir, struct folio **f); 3863 ino_t f2fs_inode_by_name(struct inode *dir, const struct qstr *qstr, 3864 struct folio **folio); 3865 void f2fs_set_link(struct inode *dir, struct f2fs_dir_entry *de, 3866 struct folio *folio, struct inode *inode); 3867 bool f2fs_has_enough_room(struct inode *dir, struct folio *ifolio, 3868 const struct f2fs_filename *fname); 3869 void f2fs_update_dentry(nid_t ino, umode_t mode, struct f2fs_dentry_ptr *d, 3870 const struct fscrypt_str *name, f2fs_hash_t name_hash, 3871 unsigned int bit_pos); 3872 int f2fs_add_regular_entry(struct inode *dir, const struct f2fs_filename *fname, 3873 struct inode *inode, nid_t ino, umode_t mode); 3874 int f2fs_add_dentry(struct inode *dir, const struct f2fs_filename *fname, 3875 struct inode *inode, nid_t ino, umode_t mode); 3876 int f2fs_do_add_link(struct inode *dir, const struct qstr *name, 3877 struct inode *inode, nid_t ino, umode_t mode); 3878 void f2fs_delete_entry(struct f2fs_dir_entry *dentry, struct folio *folio, 3879 struct inode *dir, struct inode *inode); 3880 int f2fs_do_tmpfile(struct inode *inode, struct inode *dir, 3881 struct f2fs_filename *fname); 3882 bool f2fs_empty_dir(struct inode *dir); 3883 3884 static inline int f2fs_add_link(struct dentry *dentry, struct inode *inode) 3885 { 3886 if (fscrypt_is_nokey_name(dentry)) 3887 return -ENOKEY; 3888 return f2fs_do_add_link(d_inode(dentry->d_parent), &dentry->d_name, 3889 inode, inode->i_ino, inode->i_mode); 3890 } 3891 3892 /* 3893 * super.c 3894 */ 3895 int f2fs_inode_dirtied(struct inode *inode, bool sync); 3896 void f2fs_inode_synced(struct inode *inode); 3897 int f2fs_dquot_initialize(struct inode *inode); 3898 int f2fs_enable_quota_files(struct f2fs_sb_info *sbi, bool rdonly); 3899 int f2fs_do_quota_sync(struct super_block *sb, int type); 3900 loff_t max_file_blocks(struct inode *inode); 3901 void f2fs_quota_off_umount(struct super_block *sb); 3902 void f2fs_save_errors(struct f2fs_sb_info *sbi, unsigned char flag); 3903 void f2fs_handle_critical_error(struct f2fs_sb_info *sbi, unsigned char reason); 3904 void f2fs_handle_error(struct f2fs_sb_info *sbi, unsigned char error); 3905 int f2fs_commit_super(struct f2fs_sb_info *sbi, bool recover); 3906 int f2fs_sync_fs(struct super_block *sb, int sync); 3907 int f2fs_sanity_check_ckpt(struct f2fs_sb_info *sbi); 3908 3909 /* 3910 * hash.c 3911 */ 3912 void f2fs_hash_filename(const struct inode *dir, struct f2fs_filename *fname); 3913 3914 /* 3915 * node.c 3916 */ 3917 struct node_info; 3918 enum node_type; 3919 3920 int f2fs_check_nid_range(struct f2fs_sb_info *sbi, nid_t nid); 3921 bool f2fs_available_free_memory(struct f2fs_sb_info *sbi, int type); 3922 bool f2fs_in_warm_node_list(struct f2fs_sb_info *sbi, struct folio *folio); 3923 void f2fs_init_fsync_node_info(struct f2fs_sb_info *sbi); 3924 void f2fs_del_fsync_node_entry(struct f2fs_sb_info *sbi, struct folio *folio); 3925 void f2fs_reset_fsync_node_info(struct f2fs_sb_info *sbi); 3926 int f2fs_need_dentry_mark(struct f2fs_sb_info *sbi, nid_t nid); 3927 bool f2fs_is_checkpointed_node(struct f2fs_sb_info *sbi, nid_t nid); 3928 bool f2fs_need_inode_block_update(struct f2fs_sb_info *sbi, nid_t ino); 3929 int f2fs_get_node_info(struct f2fs_sb_info *sbi, nid_t nid, 3930 struct node_info *ni, bool checkpoint_context); 3931 pgoff_t f2fs_get_next_page_offset(struct dnode_of_data *dn, pgoff_t pgofs); 3932 int f2fs_get_dnode_of_data(struct dnode_of_data *dn, pgoff_t index, int mode); 3933 int f2fs_truncate_inode_blocks(struct inode *inode, pgoff_t from); 3934 int f2fs_truncate_xattr_node(struct inode *inode); 3935 int f2fs_wait_on_node_pages_writeback(struct f2fs_sb_info *sbi, 3936 unsigned int seq_id); 3937 int f2fs_remove_inode_page(struct inode *inode); 3938 struct folio *f2fs_new_inode_folio(struct inode *inode); 3939 struct folio *f2fs_new_node_folio(struct dnode_of_data *dn, unsigned int ofs); 3940 void f2fs_ra_node_page(struct f2fs_sb_info *sbi, nid_t nid); 3941 struct folio *f2fs_get_node_folio(struct f2fs_sb_info *sbi, pgoff_t nid, 3942 enum node_type node_type); 3943 int f2fs_sanity_check_node_footer(struct f2fs_sb_info *sbi, 3944 struct folio *folio, pgoff_t nid, 3945 enum node_type ntype, bool in_irq); 3946 struct folio *f2fs_get_inode_folio(struct f2fs_sb_info *sbi, pgoff_t ino); 3947 struct folio *f2fs_get_xnode_folio(struct f2fs_sb_info *sbi, pgoff_t xnid); 3948 int f2fs_move_node_folio(struct folio *node_folio, int gc_type); 3949 void f2fs_flush_inline_data(struct f2fs_sb_info *sbi); 3950 int f2fs_fsync_node_pages(struct f2fs_sb_info *sbi, struct inode *inode, 3951 struct writeback_control *wbc, bool atomic, 3952 unsigned int *seq_id); 3953 int f2fs_sync_node_pages(struct f2fs_sb_info *sbi, 3954 struct writeback_control *wbc, 3955 bool do_balance, enum iostat_type io_type); 3956 int f2fs_build_free_nids(struct f2fs_sb_info *sbi, bool sync, bool mount); 3957 bool f2fs_alloc_nid(struct f2fs_sb_info *sbi, nid_t *nid); 3958 void f2fs_alloc_nid_done(struct f2fs_sb_info *sbi, nid_t nid); 3959 void f2fs_alloc_nid_failed(struct f2fs_sb_info *sbi, nid_t nid); 3960 int f2fs_try_to_free_nids(struct f2fs_sb_info *sbi, int nr_shrink); 3961 int f2fs_recover_inline_xattr(struct inode *inode, struct folio *folio); 3962 int f2fs_recover_xattr_data(struct inode *inode, struct folio *folio); 3963 int f2fs_recover_inode_page(struct f2fs_sb_info *sbi, struct folio *folio); 3964 int f2fs_restore_node_summary(struct f2fs_sb_info *sbi, 3965 unsigned int segno, struct f2fs_summary_block *sum); 3966 int f2fs_flush_nat_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3967 int f2fs_build_node_manager(struct f2fs_sb_info *sbi); 3968 void f2fs_destroy_node_manager(struct f2fs_sb_info *sbi); 3969 int __init f2fs_create_node_manager_caches(void); 3970 void f2fs_destroy_node_manager_caches(void); 3971 3972 /* 3973 * segment.c 3974 */ 3975 bool f2fs_need_SSR(struct f2fs_sb_info *sbi); 3976 int f2fs_commit_atomic_write(struct inode *inode); 3977 void f2fs_abort_atomic_write(struct inode *inode, bool clean); 3978 void f2fs_balance_fs(struct f2fs_sb_info *sbi, bool need); 3979 void f2fs_balance_fs_bg(struct f2fs_sb_info *sbi, bool from_bg); 3980 int f2fs_issue_flush(struct f2fs_sb_info *sbi, nid_t ino); 3981 int f2fs_create_flush_cmd_control(struct f2fs_sb_info *sbi); 3982 int f2fs_flush_device_cache(struct f2fs_sb_info *sbi); 3983 void f2fs_destroy_flush_cmd_control(struct f2fs_sb_info *sbi, bool free); 3984 void f2fs_invalidate_blocks(struct f2fs_sb_info *sbi, block_t addr, 3985 unsigned int len); 3986 bool f2fs_is_checkpointed_data(struct f2fs_sb_info *sbi, block_t blkaddr); 3987 int f2fs_start_discard_thread(struct f2fs_sb_info *sbi); 3988 void f2fs_drop_discard_cmd(struct f2fs_sb_info *sbi); 3989 void f2fs_stop_discard_thread(struct f2fs_sb_info *sbi); 3990 bool f2fs_issue_discard_timeout(struct f2fs_sb_info *sbi); 3991 void f2fs_clear_prefree_segments(struct f2fs_sb_info *sbi, 3992 struct cp_control *cpc); 3993 void f2fs_dirty_to_prefree(struct f2fs_sb_info *sbi); 3994 block_t f2fs_get_unusable_blocks(struct f2fs_sb_info *sbi); 3995 int f2fs_disable_cp_again(struct f2fs_sb_info *sbi, block_t unusable); 3996 void f2fs_release_discard_addrs(struct f2fs_sb_info *sbi); 3997 int f2fs_npages_for_summary_flush(struct f2fs_sb_info *sbi, bool for_ra); 3998 bool f2fs_segment_has_free_slot(struct f2fs_sb_info *sbi, int segno); 3999 int f2fs_init_inmem_curseg(struct f2fs_sb_info *sbi); 4000 int f2fs_reinit_atgc_curseg(struct f2fs_sb_info *sbi); 4001 void f2fs_save_inmem_curseg(struct f2fs_sb_info *sbi); 4002 void f2fs_restore_inmem_curseg(struct f2fs_sb_info *sbi); 4003 int f2fs_allocate_segment_for_resize(struct f2fs_sb_info *sbi, int type, 4004 unsigned int start, unsigned int end); 4005 int f2fs_allocate_new_section(struct f2fs_sb_info *sbi, int type, bool force); 4006 int f2fs_allocate_pinning_section(struct f2fs_sb_info *sbi); 4007 int f2fs_allocate_new_segments(struct f2fs_sb_info *sbi); 4008 int f2fs_trim_fs(struct f2fs_sb_info *sbi, struct fstrim_range *range); 4009 bool f2fs_exist_trim_candidates(struct f2fs_sb_info *sbi, 4010 struct cp_control *cpc); 4011 struct folio *f2fs_get_sum_folio(struct f2fs_sb_info *sbi, unsigned int segno); 4012 void f2fs_update_meta_page(struct f2fs_sb_info *sbi, void *src, 4013 block_t blk_addr); 4014 void f2fs_do_write_meta_page(struct f2fs_sb_info *sbi, struct folio *folio, 4015 enum iostat_type io_type); 4016 void f2fs_do_write_node_page(unsigned int nid, struct f2fs_io_info *fio); 4017 void f2fs_outplace_write_data(struct dnode_of_data *dn, 4018 struct f2fs_io_info *fio); 4019 int f2fs_inplace_write_data(struct f2fs_io_info *fio); 4020 void f2fs_do_replace_block(struct f2fs_sb_info *sbi, struct f2fs_summary *sum, 4021 block_t old_blkaddr, block_t new_blkaddr, 4022 bool recover_curseg, bool recover_newaddr, 4023 bool from_gc); 4024 void f2fs_replace_block(struct f2fs_sb_info *sbi, struct dnode_of_data *dn, 4025 block_t old_addr, block_t new_addr, 4026 unsigned char version, bool recover_curseg, 4027 bool recover_newaddr); 4028 enum temp_type f2fs_get_segment_temp(struct f2fs_sb_info *sbi, 4029 enum log_type seg_type); 4030 int f2fs_allocate_data_block(struct f2fs_sb_info *sbi, struct folio *folio, 4031 block_t old_blkaddr, block_t *new_blkaddr, 4032 struct f2fs_summary *sum, int type, 4033 struct f2fs_io_info *fio); 4034 void f2fs_update_device_state(struct f2fs_sb_info *sbi, nid_t ino, 4035 block_t blkaddr, unsigned int blkcnt); 4036 void f2fs_folio_wait_writeback(struct folio *folio, enum page_type type, 4037 bool ordered, bool locked); 4038 #define f2fs_wait_on_page_writeback(page, type, ordered, locked) \ 4039 f2fs_folio_wait_writeback(page_folio(page), type, ordered, locked) 4040 void f2fs_wait_on_block_writeback(struct inode *inode, block_t blkaddr); 4041 void f2fs_wait_on_block_writeback_range(struct inode *inode, block_t blkaddr, 4042 block_t len); 4043 void f2fs_write_data_summaries(struct f2fs_sb_info *sbi, block_t start_blk); 4044 void f2fs_write_node_summaries(struct f2fs_sb_info *sbi, block_t start_blk); 4045 int f2fs_lookup_journal_in_cursum(struct f2fs_sb_info *sbi, 4046 struct f2fs_journal *journal, int type, 4047 unsigned int val, int alloc); 4048 void f2fs_flush_sit_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc); 4049 int f2fs_check_and_fix_write_pointer(struct f2fs_sb_info *sbi); 4050 int f2fs_build_segment_manager(struct f2fs_sb_info *sbi); 4051 void f2fs_destroy_segment_manager(struct f2fs_sb_info *sbi); 4052 int __init f2fs_create_segment_manager_caches(void); 4053 void f2fs_destroy_segment_manager_caches(void); 4054 int f2fs_rw_hint_to_seg_type(struct f2fs_sb_info *sbi, enum rw_hint hint); 4055 enum rw_hint f2fs_io_type_to_rw_hint(struct f2fs_sb_info *sbi, 4056 enum page_type type, enum temp_type temp); 4057 unsigned int f2fs_usable_segs_in_sec(struct f2fs_sb_info *sbi); 4058 unsigned int f2fs_usable_blks_in_seg(struct f2fs_sb_info *sbi, 4059 unsigned int segno); 4060 unsigned long long f2fs_get_section_mtime(struct f2fs_sb_info *sbi, 4061 unsigned int segno); 4062 4063 static inline struct inode *fio_inode(struct f2fs_io_info *fio) 4064 { 4065 return fio->folio->mapping->host; 4066 } 4067 4068 #define DEF_FRAGMENT_SIZE 4 4069 #define MIN_FRAGMENT_SIZE 1 4070 #define MAX_FRAGMENT_SIZE 512 4071 4072 static inline bool f2fs_need_rand_seg(struct f2fs_sb_info *sbi) 4073 { 4074 return F2FS_OPTION(sbi).fs_mode == FS_MODE_FRAGMENT_SEG || 4075 F2FS_OPTION(sbi).fs_mode == FS_MODE_FRAGMENT_BLK; 4076 } 4077 4078 /* 4079 * checkpoint.c 4080 */ 4081 void f2fs_lock_op(struct f2fs_sb_info *sbi, struct f2fs_lock_context *lc); 4082 int f2fs_trylock_op(struct f2fs_sb_info *sbi, struct f2fs_lock_context *lc); 4083 void f2fs_unlock_op(struct f2fs_sb_info *sbi, struct f2fs_lock_context *lc); 4084 void f2fs_stop_checkpoint(struct f2fs_sb_info *sbi, bool end_io, 4085 unsigned char reason); 4086 void f2fs_flush_ckpt_thread(struct f2fs_sb_info *sbi); 4087 struct folio *f2fs_grab_meta_folio(struct f2fs_sb_info *sbi, pgoff_t index); 4088 struct folio *f2fs_get_meta_folio(struct f2fs_sb_info *sbi, pgoff_t index); 4089 struct folio *f2fs_get_meta_folio_retry(struct f2fs_sb_info *sbi, pgoff_t index); 4090 struct folio *f2fs_get_tmp_folio(struct f2fs_sb_info *sbi, pgoff_t index); 4091 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi, 4092 block_t blkaddr, int type); 4093 bool f2fs_is_valid_blkaddr_raw(struct f2fs_sb_info *sbi, 4094 block_t blkaddr, int type); 4095 int f2fs_ra_meta_pages(struct f2fs_sb_info *sbi, block_t start, int nrpages, 4096 int type, bool sync); 4097 void f2fs_ra_meta_pages_cond(struct f2fs_sb_info *sbi, pgoff_t index, 4098 unsigned int ra_blocks); 4099 long f2fs_sync_meta_pages(struct f2fs_sb_info *sbi, long nr_to_write, 4100 enum iostat_type io_type); 4101 void f2fs_add_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type); 4102 void f2fs_remove_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type); 4103 void f2fs_release_ino_entry(struct f2fs_sb_info *sbi, bool all); 4104 bool f2fs_exist_written_data(struct f2fs_sb_info *sbi, nid_t ino, int mode); 4105 void f2fs_set_dirty_device(struct f2fs_sb_info *sbi, nid_t ino, 4106 unsigned int devidx, int type); 4107 bool f2fs_is_dirty_device(struct f2fs_sb_info *sbi, nid_t ino, 4108 unsigned int devidx, int type); 4109 int f2fs_acquire_orphan_inode(struct f2fs_sb_info *sbi); 4110 void f2fs_release_orphan_inode(struct f2fs_sb_info *sbi); 4111 void f2fs_add_orphan_inode(struct inode *inode); 4112 void f2fs_remove_orphan_inode(struct f2fs_sb_info *sbi, nid_t ino); 4113 int f2fs_recover_orphan_inodes(struct f2fs_sb_info *sbi); 4114 int f2fs_get_valid_checkpoint(struct f2fs_sb_info *sbi); 4115 void f2fs_update_dirty_folio(struct inode *inode, struct folio *folio); 4116 void f2fs_remove_dirty_inode(struct inode *inode); 4117 int f2fs_sync_dirty_inodes(struct f2fs_sb_info *sbi, enum inode_type type, 4118 bool from_cp); 4119 void f2fs_wait_on_all_pages(struct f2fs_sb_info *sbi, int type); 4120 u64 f2fs_get_sectors_written(struct f2fs_sb_info *sbi); 4121 int f2fs_write_checkpoint(struct f2fs_sb_info *sbi, struct cp_control *cpc); 4122 void f2fs_init_ino_entry_info(struct f2fs_sb_info *sbi); 4123 int __init f2fs_create_checkpoint_caches(void); 4124 void f2fs_destroy_checkpoint_caches(void); 4125 int f2fs_issue_checkpoint(struct f2fs_sb_info *sbi); 4126 int f2fs_start_ckpt_thread(struct f2fs_sb_info *sbi); 4127 void f2fs_stop_ckpt_thread(struct f2fs_sb_info *sbi); 4128 void f2fs_init_ckpt_req_control(struct f2fs_sb_info *sbi); 4129 4130 /* 4131 * data.c 4132 */ 4133 int __init f2fs_init_bioset(void); 4134 void f2fs_destroy_bioset(void); 4135 bool f2fs_is_cp_guaranteed(const struct folio *folio); 4136 int f2fs_init_bio_entry_cache(void); 4137 void f2fs_destroy_bio_entry_cache(void); 4138 void f2fs_submit_read_bio(struct f2fs_sb_info *sbi, struct bio *bio, 4139 enum page_type type); 4140 int f2fs_init_write_merge_io(struct f2fs_sb_info *sbi); 4141 void f2fs_submit_merged_write(struct f2fs_sb_info *sbi, enum page_type type); 4142 void f2fs_submit_merged_write_cond(struct f2fs_sb_info *sbi, 4143 struct inode *inode, struct folio *folio, 4144 nid_t ino, enum page_type type); 4145 void f2fs_submit_merged_write_folio(struct f2fs_sb_info *sbi, 4146 struct folio *folio, enum page_type type); 4147 void f2fs_submit_merged_ipu_write(struct f2fs_sb_info *sbi, 4148 struct bio **bio, struct folio *folio); 4149 void f2fs_flush_merged_writes(struct f2fs_sb_info *sbi); 4150 int f2fs_submit_page_bio(struct f2fs_io_info *fio); 4151 int f2fs_merge_page_bio(struct f2fs_io_info *fio); 4152 void f2fs_submit_page_write(struct f2fs_io_info *fio); 4153 struct block_device *f2fs_target_device(struct f2fs_sb_info *sbi, 4154 block_t blk_addr, sector_t *sector); 4155 int f2fs_target_device_index(struct f2fs_sb_info *sbi, block_t blkaddr); 4156 void f2fs_set_data_blkaddr(struct dnode_of_data *dn, block_t blkaddr); 4157 void f2fs_update_data_blkaddr(struct dnode_of_data *dn, block_t blkaddr); 4158 int f2fs_reserve_new_blocks(struct dnode_of_data *dn, blkcnt_t count); 4159 int f2fs_reserve_new_block(struct dnode_of_data *dn); 4160 int f2fs_get_block_locked(struct dnode_of_data *dn, pgoff_t index); 4161 int f2fs_reserve_block(struct dnode_of_data *dn, pgoff_t index); 4162 struct folio *f2fs_get_read_data_folio(struct inode *inode, pgoff_t index, 4163 blk_opf_t op_flags, bool for_write, pgoff_t *next_pgofs); 4164 struct folio *f2fs_find_data_folio(struct inode *inode, pgoff_t index, 4165 pgoff_t *next_pgofs); 4166 struct folio *f2fs_get_lock_data_folio(struct inode *inode, pgoff_t index, 4167 bool for_write); 4168 struct folio *f2fs_get_new_data_folio(struct inode *inode, 4169 struct folio *ifolio, pgoff_t index, bool new_i_size); 4170 int f2fs_do_write_data_page(struct f2fs_io_info *fio); 4171 int f2fs_map_blocks(struct inode *inode, struct f2fs_map_blocks *map, int flag); 4172 int f2fs_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, 4173 u64 start, u64 len); 4174 int f2fs_encrypt_one_page(struct f2fs_io_info *fio); 4175 bool f2fs_should_update_inplace(struct inode *inode, struct f2fs_io_info *fio); 4176 bool f2fs_should_update_outplace(struct inode *inode, struct f2fs_io_info *fio); 4177 int f2fs_write_single_data_page(struct folio *folio, int *submitted, 4178 struct bio **bio, sector_t *last_block, 4179 struct writeback_control *wbc, 4180 enum iostat_type io_type, 4181 int compr_blocks, bool allow_balance); 4182 void f2fs_write_failed(struct inode *inode, loff_t to); 4183 void f2fs_invalidate_folio(struct folio *folio, size_t offset, size_t length); 4184 bool f2fs_release_folio(struct folio *folio, gfp_t wait); 4185 bool f2fs_overwrite_io(struct inode *inode, loff_t pos, size_t len); 4186 void f2fs_clear_page_cache_dirty_tag(struct folio *folio); 4187 int f2fs_init_post_read_processing(void); 4188 void f2fs_destroy_post_read_processing(void); 4189 int f2fs_init_post_read_wq(struct f2fs_sb_info *sbi); 4190 void f2fs_destroy_post_read_wq(struct f2fs_sb_info *sbi); 4191 extern const struct iomap_ops f2fs_iomap_ops; 4192 4193 /* 4194 * gc.c 4195 */ 4196 int f2fs_start_gc_thread(struct f2fs_sb_info *sbi); 4197 void f2fs_stop_gc_thread(struct f2fs_sb_info *sbi); 4198 block_t f2fs_start_bidx_of_node(unsigned int node_ofs, struct inode *inode); 4199 int f2fs_gc(struct f2fs_sb_info *sbi, struct f2fs_gc_control *gc_control); 4200 void f2fs_build_gc_manager(struct f2fs_sb_info *sbi); 4201 int f2fs_gc_range(struct f2fs_sb_info *sbi, 4202 unsigned int start_seg, unsigned int end_seg, 4203 bool dry_run, unsigned int dry_run_sections); 4204 int f2fs_resize_fs(struct file *filp, __u64 block_count); 4205 int __init f2fs_create_garbage_collection_cache(void); 4206 void f2fs_destroy_garbage_collection_cache(void); 4207 /* victim selection function for cleaning and SSR */ 4208 int f2fs_get_victim(struct f2fs_sb_info *sbi, unsigned int *result, 4209 int gc_type, int type, char alloc_mode, 4210 unsigned long long age, bool one_time); 4211 4212 /* 4213 * recovery.c 4214 */ 4215 int f2fs_recover_fsync_data(struct f2fs_sb_info *sbi, bool check_only); 4216 bool f2fs_space_for_roll_forward(struct f2fs_sb_info *sbi); 4217 int __init f2fs_create_recovery_cache(void); 4218 void f2fs_destroy_recovery_cache(void); 4219 4220 /* 4221 * debug.c 4222 */ 4223 #ifdef CONFIG_F2FS_STAT_FS 4224 enum { 4225 DEVSTAT_INUSE, 4226 DEVSTAT_DIRTY, 4227 DEVSTAT_FULL, 4228 DEVSTAT_FREE, 4229 DEVSTAT_PREFREE, 4230 DEVSTAT_MAX, 4231 }; 4232 4233 struct f2fs_dev_stats { 4234 unsigned int devstats[2][DEVSTAT_MAX]; /* 0: segs, 1: secs */ 4235 }; 4236 4237 struct f2fs_stat_info { 4238 struct list_head stat_list; 4239 struct f2fs_sb_info *sbi; 4240 int all_area_segs, sit_area_segs, nat_area_segs, ssa_area_segs; 4241 int main_area_segs, main_area_sections, main_area_zones; 4242 unsigned long long hit_cached[NR_EXTENT_CACHES]; 4243 unsigned long long hit_rbtree[NR_EXTENT_CACHES]; 4244 unsigned long long total_ext[NR_EXTENT_CACHES]; 4245 unsigned long long hit_total[NR_EXTENT_CACHES]; 4246 int ext_tree[NR_EXTENT_CACHES]; 4247 int zombie_tree[NR_EXTENT_CACHES]; 4248 int ext_node[NR_EXTENT_CACHES]; 4249 /* to count memory footprint */ 4250 unsigned long long ext_mem[NR_EXTENT_CACHES]; 4251 /* for read extent cache */ 4252 unsigned long long hit_largest; 4253 /* for block age extent cache */ 4254 unsigned long long allocated_data_blocks; 4255 int ndirty_node, ndirty_dent, ndirty_meta, ndirty_imeta; 4256 int ndirty_data, ndirty_qdata; 4257 unsigned int ndirty_dirs, ndirty_files, ndirty_all; 4258 unsigned int nquota_files, ndonate_files; 4259 int nats, dirty_nats, sits, dirty_sits; 4260 int free_nids, avail_nids, alloc_nids; 4261 int total_count, utilization; 4262 int nr_wb_cp_data, nr_wb_data; 4263 int nr_rd_data, nr_rd_node, nr_rd_meta; 4264 int nr_dio_read, nr_dio_write; 4265 unsigned int io_skip_bggc, other_skip_bggc; 4266 int nr_flushing, nr_flushed, flush_list_empty; 4267 int nr_discarding, nr_discarded; 4268 int nr_discard_cmd; 4269 unsigned int undiscard_blks; 4270 int nr_issued_ckpt, nr_total_ckpt, nr_queued_ckpt; 4271 unsigned int cur_ckpt_time, peak_ckpt_time; 4272 int inline_xattr, inline_inode, inline_dir, append, update, orphans; 4273 int compr_inode, swapfile_inode; 4274 unsigned long long compr_blocks; 4275 int aw_cnt, max_aw_cnt; 4276 unsigned int valid_count, valid_node_count, valid_inode_count, discard_blks; 4277 unsigned int bimodal, avg_vblocks; 4278 int util_free, util_valid, util_invalid; 4279 int rsvd_segs, overp_segs; 4280 int dirty_count, node_pages, meta_pages, compress_pages; 4281 int compress_page_hit; 4282 int prefree_count, free_segs, free_secs; 4283 int cp_call_count[MAX_CALL_TYPE], cp_count; 4284 int gc_call_count[MAX_CALL_TYPE]; 4285 int gc_segs[2][2]; 4286 int gc_secs[2][2]; 4287 int tot_blks, data_blks, node_blks; 4288 int bg_data_blks, bg_node_blks; 4289 int blkoff[NR_CURSEG_TYPE]; 4290 int curseg[NR_CURSEG_TYPE]; 4291 int cursec[NR_CURSEG_TYPE]; 4292 int curzone[NR_CURSEG_TYPE]; 4293 unsigned int dirty_seg[NR_CURSEG_TYPE]; 4294 unsigned int full_seg[NR_CURSEG_TYPE]; 4295 unsigned int valid_blks[NR_CURSEG_TYPE]; 4296 4297 unsigned int meta_count[META_MAX]; 4298 unsigned int segment_count[2]; 4299 unsigned int block_count[2]; 4300 unsigned int inplace_count; 4301 unsigned long long base_mem, cache_mem, page_mem; 4302 struct f2fs_dev_stats *dev_stats; 4303 }; 4304 4305 static inline struct f2fs_stat_info *F2FS_STAT(struct f2fs_sb_info *sbi) 4306 { 4307 return (struct f2fs_stat_info *)sbi->stat_info; 4308 } 4309 4310 #define stat_inc_cp_call_count(sbi, foreground) \ 4311 atomic_inc(&sbi->cp_call_count[(foreground)]) 4312 #define stat_inc_cp_count(sbi) (F2FS_STAT(sbi)->cp_count++) 4313 #define stat_io_skip_bggc_count(sbi) ((sbi)->io_skip_bggc++) 4314 #define stat_other_skip_bggc_count(sbi) ((sbi)->other_skip_bggc++) 4315 #define stat_inc_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]++) 4316 #define stat_dec_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]--) 4317 #define stat_inc_total_hit(sbi, type) (atomic64_inc(&(sbi)->total_hit_ext[type])) 4318 #define stat_inc_rbtree_node_hit(sbi, type) (atomic64_inc(&(sbi)->read_hit_rbtree[type])) 4319 #define stat_inc_largest_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_largest)) 4320 #define stat_inc_cached_node_hit(sbi, type) (atomic64_inc(&(sbi)->read_hit_cached[type])) 4321 #define stat_inc_inline_xattr(inode) \ 4322 do { \ 4323 if (f2fs_has_inline_xattr(inode)) \ 4324 (atomic_inc(&F2FS_I_SB(inode)->inline_xattr)); \ 4325 } while (0) 4326 #define stat_dec_inline_xattr(inode) \ 4327 do { \ 4328 if (f2fs_has_inline_xattr(inode)) \ 4329 (atomic_dec(&F2FS_I_SB(inode)->inline_xattr)); \ 4330 } while (0) 4331 #define stat_inc_inline_inode(inode) \ 4332 do { \ 4333 if (f2fs_has_inline_data(inode)) \ 4334 (atomic_inc(&F2FS_I_SB(inode)->inline_inode)); \ 4335 } while (0) 4336 #define stat_dec_inline_inode(inode) \ 4337 do { \ 4338 if (f2fs_has_inline_data(inode)) \ 4339 (atomic_dec(&F2FS_I_SB(inode)->inline_inode)); \ 4340 } while (0) 4341 #define stat_inc_inline_dir(inode) \ 4342 do { \ 4343 if (f2fs_has_inline_dentry(inode)) \ 4344 (atomic_inc(&F2FS_I_SB(inode)->inline_dir)); \ 4345 } while (0) 4346 #define stat_dec_inline_dir(inode) \ 4347 do { \ 4348 if (f2fs_has_inline_dentry(inode)) \ 4349 (atomic_dec(&F2FS_I_SB(inode)->inline_dir)); \ 4350 } while (0) 4351 #define stat_inc_compr_inode(inode) \ 4352 do { \ 4353 if (f2fs_compressed_file(inode)) \ 4354 (atomic_inc(&F2FS_I_SB(inode)->compr_inode)); \ 4355 } while (0) 4356 #define stat_dec_compr_inode(inode) \ 4357 do { \ 4358 if (f2fs_compressed_file(inode)) \ 4359 (atomic_dec(&F2FS_I_SB(inode)->compr_inode)); \ 4360 } while (0) 4361 #define stat_add_compr_blocks(inode, blocks) \ 4362 (atomic64_add(blocks, &F2FS_I_SB(inode)->compr_blocks)) 4363 #define stat_sub_compr_blocks(inode, blocks) \ 4364 (atomic64_sub(blocks, &F2FS_I_SB(inode)->compr_blocks)) 4365 #define stat_inc_swapfile_inode(inode) \ 4366 (atomic_inc(&F2FS_I_SB(inode)->swapfile_inode)) 4367 #define stat_dec_swapfile_inode(inode) \ 4368 (atomic_dec(&F2FS_I_SB(inode)->swapfile_inode)) 4369 #define stat_inc_atomic_inode(inode) \ 4370 (atomic_inc(&F2FS_I_SB(inode)->atomic_files)) 4371 #define stat_dec_atomic_inode(inode) \ 4372 (atomic_dec(&F2FS_I_SB(inode)->atomic_files)) 4373 #define stat_inc_meta_count(sbi, blkaddr) \ 4374 do { \ 4375 if (blkaddr < SIT_I(sbi)->sit_base_addr) \ 4376 atomic_inc(&(sbi)->meta_count[META_CP]); \ 4377 else if (blkaddr < NM_I(sbi)->nat_blkaddr) \ 4378 atomic_inc(&(sbi)->meta_count[META_SIT]); \ 4379 else if (blkaddr < SM_I(sbi)->ssa_blkaddr) \ 4380 atomic_inc(&(sbi)->meta_count[META_NAT]); \ 4381 else if (blkaddr < SM_I(sbi)->main_blkaddr) \ 4382 atomic_inc(&(sbi)->meta_count[META_SSA]); \ 4383 } while (0) 4384 #define stat_inc_seg_type(sbi, curseg) \ 4385 ((sbi)->segment_count[(curseg)->alloc_type]++) 4386 #define stat_inc_block_count(sbi, curseg) \ 4387 ((sbi)->block_count[(curseg)->alloc_type]++) 4388 #define stat_inc_inplace_blocks(sbi) \ 4389 (atomic_inc(&(sbi)->inplace_count)) 4390 #define stat_update_max_atomic_write(inode) \ 4391 do { \ 4392 int cur = atomic_read(&F2FS_I_SB(inode)->atomic_files); \ 4393 int max = atomic_read(&F2FS_I_SB(inode)->max_aw_cnt); \ 4394 if (cur > max) \ 4395 atomic_set(&F2FS_I_SB(inode)->max_aw_cnt, cur); \ 4396 } while (0) 4397 #define stat_inc_gc_call_count(sbi, foreground) \ 4398 (F2FS_STAT(sbi)->gc_call_count[(foreground)]++) 4399 #define stat_inc_gc_sec_count(sbi, type, gc_type) \ 4400 (F2FS_STAT(sbi)->gc_secs[(type)][(gc_type)]++) 4401 #define stat_inc_gc_seg_count(sbi, type, gc_type) \ 4402 (F2FS_STAT(sbi)->gc_segs[(type)][(gc_type)]++) 4403 4404 #define stat_inc_tot_blk_count(si, blks) \ 4405 ((si)->tot_blks += (blks)) 4406 4407 #define stat_inc_data_blk_count(sbi, blks, gc_type) \ 4408 do { \ 4409 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 4410 stat_inc_tot_blk_count(si, blks); \ 4411 si->data_blks += (blks); \ 4412 si->bg_data_blks += ((gc_type) == BG_GC) ? (blks) : 0; \ 4413 } while (0) 4414 4415 #define stat_inc_node_blk_count(sbi, blks, gc_type) \ 4416 do { \ 4417 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 4418 stat_inc_tot_blk_count(si, blks); \ 4419 si->node_blks += (blks); \ 4420 si->bg_node_blks += ((gc_type) == BG_GC) ? (blks) : 0; \ 4421 } while (0) 4422 4423 int f2fs_build_stats(struct f2fs_sb_info *sbi); 4424 void f2fs_destroy_stats(struct f2fs_sb_info *sbi); 4425 void __init f2fs_create_root_stats(void); 4426 void f2fs_destroy_root_stats(void); 4427 void f2fs_update_sit_info(struct f2fs_sb_info *sbi); 4428 #else 4429 #define stat_inc_cp_call_count(sbi, foreground) do { } while (0) 4430 #define stat_inc_cp_count(sbi) do { } while (0) 4431 #define stat_io_skip_bggc_count(sbi) do { } while (0) 4432 #define stat_other_skip_bggc_count(sbi) do { } while (0) 4433 #define stat_inc_dirty_inode(sbi, type) do { } while (0) 4434 #define stat_dec_dirty_inode(sbi, type) do { } while (0) 4435 #define stat_inc_total_hit(sbi, type) do { } while (0) 4436 #define stat_inc_rbtree_node_hit(sbi, type) do { } while (0) 4437 #define stat_inc_largest_node_hit(sbi) do { } while (0) 4438 #define stat_inc_cached_node_hit(sbi, type) do { } while (0) 4439 #define stat_inc_inline_xattr(inode) do { } while (0) 4440 #define stat_dec_inline_xattr(inode) do { } while (0) 4441 #define stat_inc_inline_inode(inode) do { } while (0) 4442 #define stat_dec_inline_inode(inode) do { } while (0) 4443 #define stat_inc_inline_dir(inode) do { } while (0) 4444 #define stat_dec_inline_dir(inode) do { } while (0) 4445 #define stat_inc_compr_inode(inode) do { } while (0) 4446 #define stat_dec_compr_inode(inode) do { } while (0) 4447 #define stat_add_compr_blocks(inode, blocks) do { } while (0) 4448 #define stat_sub_compr_blocks(inode, blocks) do { } while (0) 4449 #define stat_inc_swapfile_inode(inode) do { } while (0) 4450 #define stat_dec_swapfile_inode(inode) do { } while (0) 4451 #define stat_inc_atomic_inode(inode) do { } while (0) 4452 #define stat_dec_atomic_inode(inode) do { } while (0) 4453 #define stat_update_max_atomic_write(inode) do { } while (0) 4454 #define stat_inc_meta_count(sbi, blkaddr) do { } while (0) 4455 #define stat_inc_seg_type(sbi, curseg) do { } while (0) 4456 #define stat_inc_block_count(sbi, curseg) do { } while (0) 4457 #define stat_inc_inplace_blocks(sbi) do { } while (0) 4458 #define stat_inc_gc_call_count(sbi, foreground) do { } while (0) 4459 #define stat_inc_gc_sec_count(sbi, type, gc_type) do { } while (0) 4460 #define stat_inc_gc_seg_count(sbi, type, gc_type) do { } while (0) 4461 #define stat_inc_tot_blk_count(si, blks) do { } while (0) 4462 #define stat_inc_data_blk_count(sbi, blks, gc_type) do { } while (0) 4463 #define stat_inc_node_blk_count(sbi, blks, gc_type) do { } while (0) 4464 4465 static inline int f2fs_build_stats(struct f2fs_sb_info *sbi) { return 0; } 4466 static inline void f2fs_destroy_stats(struct f2fs_sb_info *sbi) { } 4467 static inline void __init f2fs_create_root_stats(void) { } 4468 static inline void f2fs_destroy_root_stats(void) { } 4469 static inline void f2fs_update_sit_info(struct f2fs_sb_info *sbi) {} 4470 #endif 4471 4472 extern const struct file_operations f2fs_dir_operations; 4473 extern const struct file_operations f2fs_file_operations; 4474 extern const struct inode_operations f2fs_file_inode_operations; 4475 extern const struct address_space_operations f2fs_dblock_aops; 4476 extern const struct address_space_operations f2fs_node_aops; 4477 extern const struct address_space_operations f2fs_meta_aops; 4478 extern const struct inode_operations f2fs_dir_inode_operations; 4479 extern const struct inode_operations f2fs_symlink_inode_operations; 4480 extern const struct inode_operations f2fs_encrypted_symlink_inode_operations; 4481 extern const struct inode_operations f2fs_special_inode_operations; 4482 extern struct kmem_cache *f2fs_inode_entry_slab; 4483 4484 /* 4485 * inline.c 4486 */ 4487 bool f2fs_may_inline_data(struct inode *inode); 4488 bool f2fs_sanity_check_inline_data(struct inode *inode, struct folio *ifolio); 4489 bool f2fs_may_inline_dentry(struct inode *inode); 4490 void f2fs_do_read_inline_data(struct folio *folio, struct folio *ifolio); 4491 void f2fs_truncate_inline_inode(struct inode *inode, struct folio *ifolio, 4492 u64 from); 4493 int f2fs_read_inline_data(struct inode *inode, struct folio *folio); 4494 int f2fs_convert_inline_folio(struct dnode_of_data *dn, struct folio *folio); 4495 int f2fs_convert_inline_inode(struct inode *inode); 4496 int f2fs_try_convert_inline_dir(struct inode *dir, struct dentry *dentry); 4497 int f2fs_write_inline_data(struct inode *inode, struct folio *folio); 4498 int f2fs_recover_inline_data(struct inode *inode, struct folio *nfolio); 4499 struct f2fs_dir_entry *f2fs_find_in_inline_dir(struct inode *dir, 4500 const struct f2fs_filename *fname, struct folio **res_folio, 4501 bool use_hash); 4502 int f2fs_make_empty_inline_dir(struct inode *inode, struct inode *parent, 4503 struct folio *ifolio); 4504 int f2fs_add_inline_entry(struct inode *dir, const struct f2fs_filename *fname, 4505 struct inode *inode, nid_t ino, umode_t mode); 4506 void f2fs_delete_inline_entry(struct f2fs_dir_entry *dentry, 4507 struct folio *folio, struct inode *dir, struct inode *inode); 4508 bool f2fs_empty_inline_dir(struct inode *dir); 4509 int f2fs_read_inline_dir(struct file *file, struct dir_context *ctx, 4510 struct fscrypt_str *fstr); 4511 int f2fs_inline_data_fiemap(struct inode *inode, 4512 struct fiemap_extent_info *fieinfo, 4513 __u64 start, __u64 len); 4514 4515 /* 4516 * shrinker.c 4517 */ 4518 unsigned long f2fs_shrink_count(struct shrinker *shrink, 4519 struct shrink_control *sc); 4520 unsigned long f2fs_shrink_scan(struct shrinker *shrink, 4521 struct shrink_control *sc); 4522 unsigned int f2fs_donate_files(void); 4523 void f2fs_reclaim_caches(unsigned int reclaim_caches_kb); 4524 void f2fs_join_shrinker(struct f2fs_sb_info *sbi); 4525 void f2fs_leave_shrinker(struct f2fs_sb_info *sbi); 4526 4527 /* 4528 * extent_cache.c 4529 */ 4530 bool sanity_check_extent_cache(struct inode *inode, struct folio *ifolio); 4531 void f2fs_init_extent_tree(struct inode *inode); 4532 void f2fs_drop_extent_tree(struct inode *inode); 4533 void f2fs_destroy_extent_node(struct inode *inode); 4534 void f2fs_destroy_extent_tree(struct inode *inode); 4535 void f2fs_init_extent_cache_info(struct f2fs_sb_info *sbi); 4536 int __init f2fs_create_extent_cache(void); 4537 void f2fs_destroy_extent_cache(void); 4538 4539 /* read extent cache ops */ 4540 void f2fs_init_read_extent_tree(struct inode *inode, struct folio *ifolio); 4541 bool f2fs_lookup_read_extent_cache(struct inode *inode, pgoff_t pgofs, 4542 struct extent_info *ei); 4543 bool f2fs_lookup_read_extent_cache_block(struct inode *inode, pgoff_t index, 4544 block_t *blkaddr); 4545 void f2fs_update_read_extent_cache(struct dnode_of_data *dn); 4546 void f2fs_update_read_extent_cache_range(struct dnode_of_data *dn, 4547 pgoff_t fofs, block_t blkaddr, unsigned int len); 4548 unsigned int f2fs_shrink_read_extent_tree(struct f2fs_sb_info *sbi, 4549 int nr_shrink); 4550 4551 /* block age extent cache ops */ 4552 void f2fs_init_age_extent_tree(struct inode *inode); 4553 bool f2fs_lookup_age_extent_cache(struct inode *inode, pgoff_t pgofs, 4554 struct extent_info *ei); 4555 void f2fs_update_age_extent_cache(struct dnode_of_data *dn); 4556 void f2fs_update_age_extent_cache_range(struct dnode_of_data *dn, 4557 pgoff_t fofs, unsigned int len); 4558 unsigned int f2fs_shrink_age_extent_tree(struct f2fs_sb_info *sbi, 4559 int nr_shrink); 4560 4561 /* 4562 * sysfs.c 4563 */ 4564 #define MIN_RA_MUL 2 4565 #define MAX_RA_MUL 256 4566 4567 int __init f2fs_init_sysfs(void); 4568 void f2fs_exit_sysfs(void); 4569 int f2fs_register_sysfs(struct f2fs_sb_info *sbi); 4570 void f2fs_unregister_sysfs(struct f2fs_sb_info *sbi); 4571 4572 /* verity.c */ 4573 extern const struct fsverity_operations f2fs_verityops; 4574 4575 /* 4576 * crypto support 4577 */ 4578 static inline bool f2fs_encrypted_file(struct inode *inode) 4579 { 4580 return IS_ENCRYPTED(inode) && S_ISREG(inode->i_mode); 4581 } 4582 4583 static inline void f2fs_set_encrypted_inode(struct inode *inode) 4584 { 4585 #ifdef CONFIG_FS_ENCRYPTION 4586 file_set_encrypt(inode); 4587 f2fs_set_inode_flags(inode); 4588 #endif 4589 } 4590 4591 /* 4592 * Returns true if the reads of the inode's data need to undergo some 4593 * postprocessing step, like decryption or authenticity verification. 4594 */ 4595 static inline bool f2fs_post_read_required(struct inode *inode) 4596 { 4597 return f2fs_encrypted_file(inode) || fsverity_active(inode) || 4598 f2fs_compressed_file(inode); 4599 } 4600 4601 static inline bool f2fs_used_in_atomic_write(struct inode *inode) 4602 { 4603 return f2fs_is_atomic_file(inode) || f2fs_is_cow_file(inode); 4604 } 4605 4606 static inline bool f2fs_meta_inode_gc_required(struct inode *inode) 4607 { 4608 return f2fs_post_read_required(inode) || f2fs_used_in_atomic_write(inode); 4609 } 4610 4611 /* 4612 * compress.c 4613 */ 4614 #ifdef CONFIG_F2FS_FS_COMPRESSION 4615 enum cluster_check_type { 4616 CLUSTER_IS_COMPR, /* check only if compressed cluster */ 4617 CLUSTER_COMPR_BLKS, /* return # of compressed blocks in a cluster */ 4618 CLUSTER_RAW_BLKS /* return # of raw blocks in a cluster */ 4619 }; 4620 bool f2fs_is_compressed_page(struct folio *folio); 4621 struct folio *f2fs_compress_control_folio(struct folio *folio); 4622 int f2fs_prepare_compress_overwrite(struct inode *inode, 4623 struct page **pagep, pgoff_t index, void **fsdata); 4624 bool f2fs_compress_write_end(struct inode *inode, void *fsdata, 4625 pgoff_t index, unsigned copied); 4626 int f2fs_truncate_partial_cluster(struct inode *inode, u64 from, bool lock); 4627 void f2fs_compress_write_end_io(struct bio *bio, struct folio *folio); 4628 bool f2fs_is_compress_backend_ready(struct inode *inode); 4629 bool f2fs_is_compress_level_valid(int alg, int lvl); 4630 int __init f2fs_init_compress_mempool(void); 4631 void f2fs_destroy_compress_mempool(void); 4632 void f2fs_decompress_cluster(struct decompress_io_ctx *dic, bool in_task); 4633 void f2fs_end_read_compressed_page(struct folio *folio, bool failed, 4634 block_t blkaddr, bool in_task); 4635 bool f2fs_cluster_is_empty(struct compress_ctx *cc); 4636 bool f2fs_cluster_can_merge_page(struct compress_ctx *cc, pgoff_t index); 4637 bool f2fs_all_cluster_page_ready(struct compress_ctx *cc, struct page **pages, 4638 int index, int nr_pages, bool uptodate); 4639 bool f2fs_sanity_check_cluster(struct dnode_of_data *dn); 4640 void f2fs_compress_ctx_add_page(struct compress_ctx *cc, struct folio *folio); 4641 int f2fs_write_multi_pages(struct compress_ctx *cc, 4642 int *submitted, 4643 struct writeback_control *wbc, 4644 enum iostat_type io_type); 4645 int f2fs_is_compressed_cluster(struct inode *inode, pgoff_t index); 4646 bool f2fs_is_sparse_cluster(struct inode *inode, pgoff_t index); 4647 void f2fs_update_read_extent_tree_range_compressed(struct inode *inode, 4648 pgoff_t fofs, block_t blkaddr, 4649 unsigned int llen, unsigned int c_len); 4650 int f2fs_read_multi_pages(struct compress_ctx *cc, struct bio **bio_ret, 4651 unsigned nr_pages, sector_t *last_block_in_bio, 4652 struct readahead_control *rac, bool for_write); 4653 struct decompress_io_ctx *f2fs_alloc_dic(struct compress_ctx *cc); 4654 void f2fs_decompress_end_io(struct decompress_io_ctx *dic, bool failed, 4655 bool in_task); 4656 void f2fs_put_folio_dic(struct folio *folio, bool in_task); 4657 unsigned int f2fs_cluster_blocks_are_contiguous(struct dnode_of_data *dn, 4658 unsigned int ofs_in_node); 4659 int f2fs_init_compress_ctx(struct compress_ctx *cc); 4660 void f2fs_destroy_compress_ctx(struct compress_ctx *cc, bool reuse); 4661 void f2fs_init_compress_info(struct f2fs_sb_info *sbi); 4662 int f2fs_init_compress_inode(struct f2fs_sb_info *sbi); 4663 void f2fs_destroy_compress_inode(struct f2fs_sb_info *sbi); 4664 int f2fs_init_page_array_cache(struct f2fs_sb_info *sbi); 4665 void f2fs_destroy_page_array_cache(struct f2fs_sb_info *sbi); 4666 int __init f2fs_init_compress_cache(void); 4667 void f2fs_destroy_compress_cache(void); 4668 struct address_space *COMPRESS_MAPPING(struct f2fs_sb_info *sbi); 4669 void f2fs_invalidate_compress_pages_range(struct f2fs_sb_info *sbi, 4670 block_t blkaddr, unsigned int len); 4671 bool f2fs_load_compressed_folio(struct f2fs_sb_info *sbi, struct folio *folio, 4672 block_t blkaddr); 4673 void f2fs_invalidate_compress_pages(struct f2fs_sb_info *sbi, nid_t ino); 4674 #define inc_compr_inode_stat(inode) \ 4675 do { \ 4676 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); \ 4677 sbi->compr_new_inode++; \ 4678 } while (0) 4679 #define add_compr_block_stat(inode, blocks) \ 4680 do { \ 4681 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); \ 4682 int diff = F2FS_I(inode)->i_cluster_size - blocks; \ 4683 sbi->compr_written_block += blocks; \ 4684 sbi->compr_saved_block += diff; \ 4685 } while (0) 4686 #else 4687 static inline bool f2fs_is_compressed_page(struct folio *folio) { return false; } 4688 static inline bool f2fs_is_compress_backend_ready(struct inode *inode) 4689 { 4690 if (!f2fs_compressed_file(inode)) 4691 return true; 4692 /* not support compression */ 4693 return false; 4694 } 4695 static inline bool f2fs_is_compress_level_valid(int alg, int lvl) { return false; } 4696 static inline struct folio *f2fs_compress_control_folio(struct folio *folio) 4697 { 4698 WARN_ON_ONCE(1); 4699 return ERR_PTR(-EINVAL); 4700 } 4701 static inline int __init f2fs_init_compress_mempool(void) { return 0; } 4702 static inline void f2fs_destroy_compress_mempool(void) { } 4703 static inline void f2fs_decompress_cluster(struct decompress_io_ctx *dic, 4704 bool in_task) { } 4705 static inline void f2fs_end_read_compressed_page(struct folio *folio, 4706 bool failed, block_t blkaddr, bool in_task) 4707 { 4708 WARN_ON_ONCE(1); 4709 } 4710 static inline void f2fs_put_folio_dic(struct folio *folio, bool in_task) 4711 { 4712 WARN_ON_ONCE(1); 4713 } 4714 static inline unsigned int f2fs_cluster_blocks_are_contiguous( 4715 struct dnode_of_data *dn, unsigned int ofs_in_node) { return 0; } 4716 static inline bool f2fs_sanity_check_cluster(struct dnode_of_data *dn) { return false; } 4717 static inline int f2fs_init_compress_inode(struct f2fs_sb_info *sbi) { return 0; } 4718 static inline void f2fs_destroy_compress_inode(struct f2fs_sb_info *sbi) { } 4719 static inline int f2fs_init_page_array_cache(struct f2fs_sb_info *sbi) { return 0; } 4720 static inline void f2fs_destroy_page_array_cache(struct f2fs_sb_info *sbi) { } 4721 static inline int __init f2fs_init_compress_cache(void) { return 0; } 4722 static inline void f2fs_destroy_compress_cache(void) { } 4723 static inline void f2fs_invalidate_compress_pages_range(struct f2fs_sb_info *sbi, 4724 block_t blkaddr, unsigned int len) { } 4725 static inline bool f2fs_load_compressed_folio(struct f2fs_sb_info *sbi, 4726 struct folio *folio, block_t blkaddr) { return false; } 4727 static inline void f2fs_invalidate_compress_pages(struct f2fs_sb_info *sbi, 4728 nid_t ino) { } 4729 #define inc_compr_inode_stat(inode) do { } while (0) 4730 static inline int f2fs_is_compressed_cluster( 4731 struct inode *inode, 4732 pgoff_t index) { return 0; } 4733 static inline bool f2fs_is_sparse_cluster( 4734 struct inode *inode, 4735 pgoff_t index) { return true; } 4736 static inline void f2fs_update_read_extent_tree_range_compressed( 4737 struct inode *inode, 4738 pgoff_t fofs, block_t blkaddr, 4739 unsigned int llen, unsigned int c_len) { } 4740 #endif 4741 4742 static inline int set_compress_context(struct inode *inode) 4743 { 4744 #ifdef CONFIG_F2FS_FS_COMPRESSION 4745 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 4746 struct f2fs_inode_info *fi = F2FS_I(inode); 4747 4748 fi->i_compress_algorithm = F2FS_OPTION(sbi).compress_algorithm; 4749 fi->i_log_cluster_size = F2FS_OPTION(sbi).compress_log_size; 4750 fi->i_compress_flag = F2FS_OPTION(sbi).compress_chksum ? 4751 BIT(COMPRESS_CHKSUM) : 0; 4752 fi->i_cluster_size = BIT(fi->i_log_cluster_size); 4753 if ((fi->i_compress_algorithm == COMPRESS_LZ4 || 4754 fi->i_compress_algorithm == COMPRESS_ZSTD) && 4755 F2FS_OPTION(sbi).compress_level) 4756 fi->i_compress_level = F2FS_OPTION(sbi).compress_level; 4757 fi->i_flags |= F2FS_COMPR_FL; 4758 set_inode_flag(inode, FI_COMPRESSED_FILE); 4759 stat_inc_compr_inode(inode); 4760 inc_compr_inode_stat(inode); 4761 f2fs_mark_inode_dirty_sync(inode, true); 4762 return 0; 4763 #else 4764 return -EOPNOTSUPP; 4765 #endif 4766 } 4767 4768 static inline bool f2fs_disable_compressed_file(struct inode *inode) 4769 { 4770 struct f2fs_inode_info *fi = F2FS_I(inode); 4771 4772 f2fs_down_write(&fi->i_sem); 4773 4774 if (!f2fs_compressed_file(inode)) { 4775 f2fs_up_write(&fi->i_sem); 4776 return true; 4777 } 4778 if (f2fs_is_mmap_file(inode) || atomic_read(&fi->writeback) || 4779 (S_ISREG(inode->i_mode) && F2FS_HAS_BLOCKS(inode))) { 4780 f2fs_up_write(&fi->i_sem); 4781 return false; 4782 } 4783 4784 fi->i_flags &= ~F2FS_COMPR_FL; 4785 stat_dec_compr_inode(inode); 4786 clear_inode_flag(inode, FI_COMPRESSED_FILE); 4787 f2fs_mark_inode_dirty_sync(inode, true); 4788 4789 f2fs_up_write(&fi->i_sem); 4790 return true; 4791 } 4792 4793 #define F2FS_FEATURE_FUNCS(name, flagname) \ 4794 static inline bool f2fs_sb_has_##name(struct f2fs_sb_info *sbi) \ 4795 { \ 4796 return F2FS_HAS_FEATURE(sbi, F2FS_FEATURE_##flagname); \ 4797 } 4798 4799 F2FS_FEATURE_FUNCS(encrypt, ENCRYPT); 4800 F2FS_FEATURE_FUNCS(blkzoned, BLKZONED); 4801 F2FS_FEATURE_FUNCS(extra_attr, EXTRA_ATTR); 4802 F2FS_FEATURE_FUNCS(project_quota, PRJQUOTA); 4803 F2FS_FEATURE_FUNCS(inode_chksum, INODE_CHKSUM); 4804 F2FS_FEATURE_FUNCS(flexible_inline_xattr, FLEXIBLE_INLINE_XATTR); 4805 F2FS_FEATURE_FUNCS(quota_ino, QUOTA_INO); 4806 F2FS_FEATURE_FUNCS(inode_crtime, INODE_CRTIME); 4807 F2FS_FEATURE_FUNCS(lost_found, LOST_FOUND); 4808 F2FS_FEATURE_FUNCS(verity, VERITY); 4809 F2FS_FEATURE_FUNCS(sb_chksum, SB_CHKSUM); 4810 F2FS_FEATURE_FUNCS(casefold, CASEFOLD); 4811 F2FS_FEATURE_FUNCS(compression, COMPRESSION); 4812 F2FS_FEATURE_FUNCS(readonly, RO); 4813 F2FS_FEATURE_FUNCS(device_alias, DEVICE_ALIAS); 4814 F2FS_FEATURE_FUNCS(packed_ssa, PACKED_SSA); 4815 4816 #ifdef CONFIG_BLK_DEV_ZONED 4817 static inline bool f2fs_zone_is_seq(struct f2fs_sb_info *sbi, int devi, 4818 unsigned int zone) 4819 { 4820 return test_bit(zone, FDEV(devi).blkz_seq); 4821 } 4822 4823 static inline bool f2fs_blkz_is_seq(struct f2fs_sb_info *sbi, int devi, 4824 block_t blkaddr) 4825 { 4826 return f2fs_zone_is_seq(sbi, devi, blkaddr / sbi->blocks_per_blkz); 4827 } 4828 #endif 4829 4830 static inline int f2fs_bdev_index(struct f2fs_sb_info *sbi, 4831 struct block_device *bdev) 4832 { 4833 int i; 4834 4835 if (!f2fs_is_multi_device(sbi)) 4836 return 0; 4837 4838 for (i = 0; i < sbi->s_ndevs; i++) 4839 if (FDEV(i).bdev == bdev) 4840 return i; 4841 4842 WARN_ON(1); 4843 return -1; 4844 } 4845 4846 static inline bool f2fs_hw_should_discard(struct f2fs_sb_info *sbi) 4847 { 4848 return f2fs_sb_has_blkzoned(sbi); 4849 } 4850 4851 static inline bool f2fs_bdev_support_discard(struct block_device *bdev) 4852 { 4853 return bdev_max_discard_sectors(bdev) || bdev_is_zoned(bdev); 4854 } 4855 4856 static inline bool f2fs_hw_support_discard(struct f2fs_sb_info *sbi) 4857 { 4858 int i; 4859 4860 if (!f2fs_is_multi_device(sbi)) 4861 return f2fs_bdev_support_discard(sbi->sb->s_bdev); 4862 4863 for (i = 0; i < sbi->s_ndevs; i++) 4864 if (f2fs_bdev_support_discard(FDEV(i).bdev)) 4865 return true; 4866 return false; 4867 } 4868 4869 static inline unsigned int f2fs_hw_discard_granularity(struct f2fs_sb_info *sbi) 4870 { 4871 int i = 1; 4872 unsigned int discard_granularity = bdev_discard_granularity(sbi->sb->s_bdev); 4873 4874 if (f2fs_is_multi_device(sbi)) 4875 for (; i < sbi->s_ndevs && !bdev_is_zoned(FDEV(i).bdev); i++) 4876 discard_granularity = max_t(unsigned int, discard_granularity, 4877 bdev_discard_granularity(FDEV(i).bdev)); 4878 return discard_granularity; 4879 } 4880 4881 static inline bool f2fs_realtime_discard_enable(struct f2fs_sb_info *sbi) 4882 { 4883 return (test_opt(sbi, DISCARD) && f2fs_hw_support_discard(sbi)) || 4884 f2fs_hw_should_discard(sbi); 4885 } 4886 4887 static inline bool f2fs_hw_is_readonly(struct f2fs_sb_info *sbi) 4888 { 4889 int i; 4890 4891 if (!f2fs_is_multi_device(sbi)) 4892 return bdev_read_only(sbi->sb->s_bdev); 4893 4894 for (i = 0; i < sbi->s_ndevs; i++) 4895 if (bdev_read_only(FDEV(i).bdev)) 4896 return true; 4897 return false; 4898 } 4899 4900 static inline bool f2fs_dev_is_readonly(struct f2fs_sb_info *sbi) 4901 { 4902 return f2fs_sb_has_readonly(sbi) || f2fs_hw_is_readonly(sbi); 4903 } 4904 4905 static inline bool f2fs_lfs_mode(struct f2fs_sb_info *sbi) 4906 { 4907 return F2FS_OPTION(sbi).fs_mode == FS_MODE_LFS; 4908 } 4909 4910 static inline bool f2fs_is_sequential_zone_area(struct f2fs_sb_info *sbi, 4911 block_t blkaddr) 4912 { 4913 if (f2fs_sb_has_blkzoned(sbi)) { 4914 #ifdef CONFIG_BLK_DEV_ZONED 4915 int devi = f2fs_target_device_index(sbi, blkaddr); 4916 4917 if (!bdev_is_zoned(FDEV(devi).bdev)) 4918 return false; 4919 4920 if (f2fs_is_multi_device(sbi)) { 4921 if (blkaddr < FDEV(devi).start_blk || 4922 blkaddr > FDEV(devi).end_blk) { 4923 f2fs_err(sbi, "Invalid block %x", blkaddr); 4924 return false; 4925 } 4926 blkaddr -= FDEV(devi).start_blk; 4927 } 4928 4929 return f2fs_blkz_is_seq(sbi, devi, blkaddr); 4930 #else 4931 return false; 4932 #endif 4933 } 4934 return false; 4935 } 4936 4937 static inline bool f2fs_low_mem_mode(struct f2fs_sb_info *sbi) 4938 { 4939 return F2FS_OPTION(sbi).memory_mode == MEMORY_MODE_LOW; 4940 } 4941 4942 static inline bool f2fs_may_compress(struct inode *inode) 4943 { 4944 if (IS_SWAPFILE(inode) || f2fs_is_pinned_file(inode) || 4945 f2fs_is_atomic_file(inode) || f2fs_has_inline_data(inode) || 4946 f2fs_is_mmap_file(inode)) 4947 return false; 4948 return S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode); 4949 } 4950 4951 static inline void f2fs_i_compr_blocks_update(struct inode *inode, 4952 u64 blocks, bool add) 4953 { 4954 struct f2fs_inode_info *fi = F2FS_I(inode); 4955 int diff = fi->i_cluster_size - blocks; 4956 4957 /* don't update i_compr_blocks if saved blocks were released */ 4958 if (!add && !atomic_read(&fi->i_compr_blocks)) 4959 return; 4960 4961 if (add) { 4962 atomic_add(diff, &fi->i_compr_blocks); 4963 stat_add_compr_blocks(inode, diff); 4964 } else { 4965 atomic_sub(diff, &fi->i_compr_blocks); 4966 stat_sub_compr_blocks(inode, diff); 4967 } 4968 f2fs_mark_inode_dirty_sync(inode, true); 4969 } 4970 4971 static inline bool f2fs_allow_multi_device_dio(struct f2fs_sb_info *sbi, 4972 int flag) 4973 { 4974 if (!f2fs_is_multi_device(sbi)) 4975 return false; 4976 if (flag != F2FS_GET_BLOCK_DIO) 4977 return false; 4978 return sbi->aligned_blksize; 4979 } 4980 4981 #ifdef CONFIG_F2FS_FAULT_INJECTION 4982 extern int f2fs_build_fault_attr(struct f2fs_sb_info *sbi, unsigned long rate, 4983 unsigned long type, enum fault_option fo); 4984 extern void f2fs_simulate_lock_timeout(struct f2fs_sb_info *sbi); 4985 #else 4986 static inline int f2fs_build_fault_attr(struct f2fs_sb_info *sbi, 4987 unsigned long rate, unsigned long type, 4988 enum fault_option fo) 4989 { 4990 return 0; 4991 } 4992 static inline void f2fs_simulate_lock_timeout(struct f2fs_sb_info *sbi) 4993 { 4994 return; 4995 } 4996 #endif 4997 4998 static inline bool is_journalled_quota(struct f2fs_sb_info *sbi) 4999 { 5000 #ifdef CONFIG_QUOTA 5001 if (f2fs_sb_has_quota_ino(sbi)) 5002 return true; 5003 if (F2FS_OPTION(sbi).s_qf_names[USRQUOTA] || 5004 F2FS_OPTION(sbi).s_qf_names[GRPQUOTA] || 5005 F2FS_OPTION(sbi).s_qf_names[PRJQUOTA]) 5006 return true; 5007 #endif 5008 return false; 5009 } 5010 5011 static inline bool f2fs_quota_file(struct f2fs_sb_info *sbi, nid_t ino) 5012 { 5013 #ifdef CONFIG_QUOTA 5014 int i; 5015 5016 if (!f2fs_sb_has_quota_ino(sbi)) 5017 return false; 5018 5019 for (i = 0; i < MAXQUOTAS; i++) { 5020 if (f2fs_qf_ino(sbi->sb, i) == ino) 5021 return true; 5022 } 5023 #endif 5024 return false; 5025 } 5026 5027 static inline bool f2fs_block_unit_discard(struct f2fs_sb_info *sbi) 5028 { 5029 return F2FS_OPTION(sbi).discard_unit == DISCARD_UNIT_BLOCK; 5030 } 5031 5032 static inline void __f2fs_schedule_timeout(long timeout, bool io) 5033 { 5034 set_current_state(TASK_UNINTERRUPTIBLE); 5035 if (io) 5036 io_schedule_timeout(timeout); 5037 else 5038 schedule_timeout(timeout); 5039 } 5040 5041 #define f2fs_io_schedule_timeout(timeout) \ 5042 __f2fs_schedule_timeout(timeout, true) 5043 #define f2fs_schedule_timeout(timeout) \ 5044 __f2fs_schedule_timeout(timeout, false) 5045 5046 static inline void f2fs_schedule_timeout_killable(long timeout, bool io) 5047 { 5048 unsigned long last_time = jiffies + timeout; 5049 5050 while (jiffies < last_time) { 5051 if (fatal_signal_pending(current)) 5052 return; 5053 __f2fs_schedule_timeout(DEFAULT_SCHEDULE_TIMEOUT, io); 5054 } 5055 } 5056 5057 static inline void f2fs_handle_page_eio(struct f2fs_sb_info *sbi, 5058 struct folio *folio, enum page_type type) 5059 { 5060 pgoff_t ofs = folio->index; 5061 5062 if (unlikely(f2fs_cp_error(sbi))) 5063 return; 5064 5065 if (ofs == sbi->page_eio_ofs[type]) { 5066 if (sbi->page_eio_cnt[type]++ == MAX_RETRY_PAGE_EIO) 5067 set_ckpt_flags(sbi, CP_ERROR_FLAG); 5068 } else { 5069 sbi->page_eio_ofs[type] = ofs; 5070 sbi->page_eio_cnt[type] = 0; 5071 } 5072 } 5073 5074 static inline bool f2fs_is_readonly(struct f2fs_sb_info *sbi) 5075 { 5076 return f2fs_sb_has_readonly(sbi) || f2fs_readonly(sbi->sb); 5077 } 5078 5079 static inline void f2fs_truncate_meta_inode_pages(struct f2fs_sb_info *sbi, 5080 block_t blkaddr, unsigned int cnt) 5081 { 5082 bool need_submit = false; 5083 int i = 0; 5084 5085 do { 5086 struct folio *folio; 5087 5088 folio = filemap_get_folio(META_MAPPING(sbi), blkaddr + i); 5089 if (!IS_ERR(folio)) { 5090 if (folio_test_writeback(folio)) 5091 need_submit = true; 5092 f2fs_folio_put(folio, false); 5093 } 5094 } while (++i < cnt && !need_submit); 5095 5096 if (need_submit) 5097 f2fs_submit_merged_write_cond(sbi, sbi->meta_inode, 5098 NULL, 0, DATA); 5099 5100 truncate_inode_pages_range(META_MAPPING(sbi), 5101 F2FS_BLK_TO_BYTES((loff_t)blkaddr), 5102 F2FS_BLK_END_BYTES((loff_t)(blkaddr + cnt - 1))); 5103 } 5104 5105 static inline void f2fs_invalidate_internal_cache(struct f2fs_sb_info *sbi, 5106 block_t blkaddr, unsigned int len) 5107 { 5108 f2fs_truncate_meta_inode_pages(sbi, blkaddr, len); 5109 f2fs_invalidate_compress_pages_range(sbi, blkaddr, len); 5110 } 5111 5112 #endif /* _LINUX_F2FS_H */ 5113