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/mmzone.h> 14 #include <linux/page-flags.h> 15 #include <linux/slab.h> 16 #include <linux/crc32.h> 17 #include <linux/magic.h> 18 #include <linux/kobject.h> 19 #include <linux/sched.h> 20 #include <linux/cred.h> 21 #include <linux/sched/mm.h> 22 #include <linux/vmalloc.h> 23 #include <linux/bio.h> 24 #include <linux/blkdev.h> 25 #include <linux/quotaops.h> 26 #include <linux/part_stat.h> 27 #include <linux/rw_hint.h> 28 #include <linux/iomap.h> 29 30 #include <linux/fscrypt.h> 31 #include <linux/fsverity.h> 32 33 #ifdef CONFIG_F2FS_CHECK_FS 34 #define f2fs_bug_on(sbi, condition) BUG_ON(condition) 35 #else 36 #define f2fs_bug_on(sbi, condition) \ 37 do { \ 38 if (WARN_ON(condition)) \ 39 set_sbi_flag(sbi, SBI_NEED_FSCK); \ 40 } while (0) 41 #endif 42 43 enum { 44 FAULT_KMALLOC, 45 FAULT_KVMALLOC, 46 FAULT_PAGE_ALLOC, 47 FAULT_PAGE_GET, 48 FAULT_ALLOC_BIO, /* it's obsolete due to bio_alloc() will never fail */ 49 FAULT_ALLOC_NID, 50 FAULT_ORPHAN, 51 FAULT_BLOCK, 52 FAULT_DIR_DEPTH, 53 FAULT_EVICT_INODE, 54 FAULT_TRUNCATE, 55 FAULT_READ_IO, 56 FAULT_CHECKPOINT, 57 FAULT_DISCARD, /* it's obsolete due to __blkdev_issue_discard() will never fail */ 58 FAULT_WRITE_IO, 59 FAULT_SLAB_ALLOC, 60 FAULT_DQUOT_INIT, 61 FAULT_LOCK_OP, 62 FAULT_BLKADDR_VALIDITY, 63 FAULT_BLKADDR_CONSISTENCE, 64 FAULT_NO_SEGMENT, 65 FAULT_INCONSISTENT_FOOTER, 66 FAULT_ATOMIC_TIMEOUT, 67 FAULT_VMALLOC, 68 FAULT_LOCK_TIMEOUT, 69 FAULT_SKIP_WRITE, 70 FAULT_MAX, 71 }; 72 73 /* indicate which option to update */ 74 enum fault_option { 75 FAULT_RATE = 1, /* only update fault rate */ 76 FAULT_TYPE = 2, /* only update fault type */ 77 FAULT_TIMEOUT = 4, /* only update fault timeout type */ 78 FAULT_ALL = 8, /* reset all fault injection options/stats */ 79 }; 80 81 #ifdef CONFIG_F2FS_FAULT_INJECTION 82 struct f2fs_fault_info { 83 atomic_t inject_ops; 84 int inject_rate; 85 unsigned int inject_type; 86 /* Used to account total count of injection for each type */ 87 unsigned int inject_count[FAULT_MAX]; 88 unsigned int inject_lock_timeout; /* inject lock timeout */ 89 }; 90 91 extern const char *f2fs_fault_name[FAULT_MAX]; 92 #define IS_FAULT_SET(fi, type) ((fi)->inject_type & BIT(type)) 93 94 /* maximum retry count for injected failure */ 95 #define DEFAULT_FAILURE_RETRY_COUNT 8 96 #else 97 #define DEFAULT_FAILURE_RETRY_COUNT 1 98 #endif 99 100 enum { 101 REPORT_FAULT_NEED_FSCK, 102 REPORT_FAULT_STOP_CP, 103 REPORT_FAULT_MAX, 104 }; 105 106 void f2fs_fault_report(struct super_block *sb, unsigned int err_code, 107 const char *func, unsigned int data); 108 109 /* 110 * For mount options 111 */ 112 enum f2fs_mount_opt { 113 F2FS_MOUNT_DISABLE_ROLL_FORWARD, 114 F2FS_MOUNT_DISCARD, 115 F2FS_MOUNT_NOHEAP, 116 F2FS_MOUNT_XATTR_USER, 117 F2FS_MOUNT_POSIX_ACL, 118 F2FS_MOUNT_DISABLE_EXT_IDENTIFY, 119 F2FS_MOUNT_INLINE_XATTR, 120 F2FS_MOUNT_INLINE_DATA, 121 F2FS_MOUNT_INLINE_DENTRY, 122 F2FS_MOUNT_FLUSH_MERGE, 123 F2FS_MOUNT_NOBARRIER, 124 F2FS_MOUNT_FASTBOOT, 125 F2FS_MOUNT_READ_EXTENT_CACHE, 126 F2FS_MOUNT_DATA_FLUSH, 127 F2FS_MOUNT_FAULT_INJECTION, 128 F2FS_MOUNT_USRQUOTA, 129 F2FS_MOUNT_GRPQUOTA, 130 F2FS_MOUNT_PRJQUOTA, 131 F2FS_MOUNT_QUOTA, 132 F2FS_MOUNT_INLINE_XATTR_SIZE, 133 F2FS_MOUNT_RESERVE_ROOT, 134 F2FS_MOUNT_DISABLE_CHECKPOINT, 135 F2FS_MOUNT_NORECOVERY, 136 F2FS_MOUNT_ATGC, 137 F2FS_MOUNT_MERGE_CHECKPOINT, 138 F2FS_MOUNT_GC_MERGE, 139 F2FS_MOUNT_COMPRESS_CACHE, 140 F2FS_MOUNT_AGE_EXTENT_CACHE, 141 F2FS_MOUNT_NAT_BITS, 142 F2FS_MOUNT_INLINECRYPT, 143 /* 144 * Some f2fs environments expect to be able to pass the "lazytime" option 145 * string rather than using the MS_LAZYTIME flag, so this must remain. 146 */ 147 F2FS_MOUNT_LAZYTIME, 148 F2FS_MOUNT_RESERVE_NODE, 149 }; 150 151 #define F2FS_OPTION(sbi) ((sbi)->mount_opt) 152 #define clear_opt(sbi, option) \ 153 (F2FS_OPTION(sbi).opt &= ~BIT_ULL(F2FS_MOUNT_##option)) 154 #define set_opt(sbi, option) \ 155 (F2FS_OPTION(sbi).opt |= BIT_ULL(F2FS_MOUNT_##option)) 156 #define test_opt(sbi, option) \ 157 (F2FS_OPTION(sbi).opt & BIT_ULL(F2FS_MOUNT_##option)) 158 159 #define ver_after(a, b) (typecheck(unsigned long long, a) && \ 160 typecheck(unsigned long long, b) && \ 161 ((long long)((a) - (b)) > 0)) 162 163 typedef u32 block_t; /* 164 * should not change u32, since it is the on-disk block 165 * address format, __le32. 166 */ 167 typedef u32 nid_t; 168 169 #define COMPRESS_EXT_NUM 16 170 171 enum blkzone_allocation_policy { 172 BLKZONE_ALLOC_PRIOR_SEQ, /* Prioritize writing to sequential zones */ 173 BLKZONE_ALLOC_ONLY_SEQ, /* Only allow writing to sequential zones */ 174 BLKZONE_ALLOC_PRIOR_CONV, /* Prioritize writing to conventional zones */ 175 }; 176 177 enum bggc_io_aware_policy { 178 AWARE_ALL_IO, /* skip background GC if there is any kind of pending IO */ 179 AWARE_READ_IO, /* skip background GC if there is pending read IO */ 180 AWARE_NONE, /* don't aware IO for background GC */ 181 }; 182 183 enum device_allocation_policy { 184 ALLOCATE_FORWARD_NOHINT, 185 ALLOCATE_FORWARD_WITHIN_HINT, 186 ALLOCATE_FORWARD_FROM_HINT, 187 }; 188 189 enum f2fs_lock_name { 190 LOCK_NAME_NONE, 191 LOCK_NAME_CP_RWSEM, 192 LOCK_NAME_NODE_CHANGE, 193 LOCK_NAME_NODE_WRITE, 194 LOCK_NAME_GC_LOCK, 195 LOCK_NAME_CP_GLOBAL, 196 LOCK_NAME_IO_RWSEM, 197 LOCK_NAME_NAT_TREE_LOCK, 198 LOCK_NAME_MAX, 199 }; 200 201 enum f2fs_timeout_type { 202 TIMEOUT_TYPE_NONE, 203 TIMEOUT_TYPE_RUNNING, 204 TIMEOUT_TYPE_IO_SLEEP, 205 TIMEOUT_TYPE_NONIO_SLEEP, 206 TIMEOUT_TYPE_RUNNABLE, 207 TIMEOUT_TYPE_MAX, 208 }; 209 210 /* 211 * An implementation of an rwsem that is explicitly unfair to readers. This 212 * prevents priority inversion when a low-priority reader acquires the read lock 213 * while sleeping on the write lock but the write lock is needed by 214 * higher-priority clients. 215 */ 216 217 struct f2fs_rwsem { 218 struct f2fs_sb_info *sbi; 219 enum f2fs_lock_name name; 220 struct rw_semaphore internal_rwsem; 221 #ifdef CONFIG_F2FS_UNFAIR_RWSEM 222 wait_queue_head_t read_waiters; 223 #endif 224 }; 225 226 #include "cache.h" 227 228 struct f2fs_mount_info { 229 unsigned long long opt; 230 block_t root_reserved_blocks; /* root reserved blocks */ 231 block_t root_reserved_nodes; /* root reserved nodes */ 232 kuid_t s_resuid; /* reserved blocks for uid */ 233 kgid_t s_resgid; /* reserved blocks for gid */ 234 int active_logs; /* # of active logs */ 235 int inline_xattr_size; /* inline xattr size */ 236 #ifdef CONFIG_F2FS_FAULT_INJECTION 237 struct f2fs_fault_info fault_info; /* For fault injection */ 238 #endif 239 #ifdef CONFIG_QUOTA 240 /* Names of quota files with journalled quota */ 241 char *s_qf_names[MAXQUOTAS]; 242 int s_jquota_fmt; /* Format of quota to use */ 243 #endif 244 /* For which write hints are passed down to block layer */ 245 int alloc_mode; /* segment allocation policy */ 246 int fsync_mode; /* fsync policy */ 247 int fs_mode; /* fs mode: LFS or ADAPTIVE */ 248 int bggc_mode; /* bggc mode: off, on or sync */ 249 int memory_mode; /* memory mode */ 250 int errors; /* errors parameter */ 251 int discard_unit; /* 252 * discard command's offset/size should 253 * be aligned to this unit: block, 254 * segment or section 255 */ 256 struct fscrypt_dummy_policy dummy_enc_policy; /* test dummy encryption */ 257 block_t unusable_cap_perc; /* percentage for cap */ 258 block_t unusable_cap; /* Amount of space allowed to be 259 * unusable when disabling checkpoint 260 */ 261 unsigned int resizable_tail_secno; /* number of resizable tail sections */ 262 263 /* For compression */ 264 unsigned char compress_algorithm; /* algorithm type */ 265 unsigned char compress_log_size; /* cluster log size */ 266 unsigned char compress_level; /* compress level */ 267 bool compress_chksum; /* compressed data chksum */ 268 unsigned char compress_ext_cnt; /* extension count */ 269 unsigned char nocompress_ext_cnt; /* nocompress extension count */ 270 int compress_mode; /* compression mode */ 271 unsigned char extensions[COMPRESS_EXT_NUM][F2FS_EXTENSION_LEN]; /* extensions */ 272 unsigned char noextensions[COMPRESS_EXT_NUM][F2FS_EXTENSION_LEN]; /* extensions */ 273 unsigned int lookup_mode; 274 }; 275 276 #define F2FS_FEATURE_ENCRYPT 0x00000001 277 #define F2FS_FEATURE_BLKZONED 0x00000002 278 #define F2FS_FEATURE_ATOMIC_WRITE 0x00000004 279 #define F2FS_FEATURE_EXTRA_ATTR 0x00000008 280 #define F2FS_FEATURE_PRJQUOTA 0x00000010 281 #define F2FS_FEATURE_INODE_CHKSUM 0x00000020 282 #define F2FS_FEATURE_FLEXIBLE_INLINE_XATTR 0x00000040 283 #define F2FS_FEATURE_QUOTA_INO 0x00000080 284 #define F2FS_FEATURE_INODE_CRTIME 0x00000100 285 #define F2FS_FEATURE_LOST_FOUND 0x00000200 286 #define F2FS_FEATURE_VERITY 0x00000400 287 #define F2FS_FEATURE_SB_CHKSUM 0x00000800 288 #define F2FS_FEATURE_CASEFOLD 0x00001000 289 #define F2FS_FEATURE_COMPRESSION 0x00002000 290 #define F2FS_FEATURE_RO 0x00004000 291 #define F2FS_FEATURE_DEVICE_ALIAS 0x00008000 292 #define F2FS_FEATURE_PACKED_SSA 0x00010000 293 294 #define __F2FS_HAS_FEATURE(raw_super, mask) \ 295 ((raw_super->feature & cpu_to_le32(mask)) != 0) 296 #define F2FS_HAS_FEATURE(sbi, mask) __F2FS_HAS_FEATURE(sbi->raw_super, mask) 297 298 /* 299 * Default values for user and/or group using reserved blocks 300 */ 301 #define F2FS_DEF_RESUID 0 302 #define F2FS_DEF_RESGID 0 303 304 /* 305 * For checkpoint manager 306 */ 307 enum { 308 NAT_BITMAP, 309 SIT_BITMAP 310 }; 311 312 #define CP_UMOUNT 0x00000001 313 #define CP_FASTBOOT 0x00000002 314 #define CP_SYNC 0x00000004 315 #define CP_RECOVERY 0x00000008 316 #define CP_DISCARD 0x00000010 317 #define CP_TRIMMED 0x00000020 318 #define CP_PAUSE 0x00000040 319 #define CP_RESIZE 0x00000080 320 321 #define DEF_MAX_DISCARD_REQUEST 8 /* issue 8 discards per round */ 322 #define DEF_MIN_DISCARD_ISSUE_TIME 50 /* 50 ms, if exists */ 323 #define DEF_MID_DISCARD_ISSUE_TIME 500 /* 500 ms, if device busy */ 324 #define DEF_MAX_DISCARD_ISSUE_TIME 60000 /* 60 s, if no candidates */ 325 #define DEF_DISCARD_URGENT_UTIL 80 /* do more discard over 80% */ 326 #define DEF_CP_INTERVAL 60 /* 60 secs */ 327 #define DEF_IDLE_INTERVAL 5 /* 5 secs */ 328 #define DEF_DISABLE_INTERVAL 5 /* 5 secs */ 329 #define DEF_DISABLE_QUICK_INTERVAL 1 /* 1 secs */ 330 #define DEF_UMOUNT_DISCARD_TIMEOUT 5 /* 5 secs */ 331 332 enum cp_time { 333 CP_TIME_START, /* begin */ 334 CP_TIME_LOCK, /* after cp_global_sem */ 335 CP_TIME_OP_LOCK, /* after block_operation */ 336 CP_TIME_MERGE_WRITE, /* after flush DATA/NODE/META */ 337 CP_TIME_FLUSH_NAT, /* after flush nat */ 338 CP_TIME_FLUSH_SIT, /* after flush sit */ 339 CP_TIME_SYNC_META, /* after sync_meta_pages */ 340 CP_TIME_SYNC_CP_META, /* after sync cp meta pages */ 341 CP_TIME_WAIT_DIRTY_META,/* after wait on dirty meta */ 342 CP_TIME_WAIT_CP_DATA, /* after wait on cp data */ 343 CP_TIME_FLUSH_DEVICE, /* after flush device cache */ 344 CP_TIME_WAIT_LAST_CP, /* after wait on last cp pack */ 345 CP_TIME_END, /* after unblock_operation */ 346 CP_TIME_MAX, 347 }; 348 349 /* time cost stats of checkpoint */ 350 struct cp_stats { 351 ktime_t times[CP_TIME_MAX]; 352 }; 353 354 struct cp_control { 355 int reason; 356 __u64 trim_start; 357 __u64 trim_end; 358 __u64 trim_minlen; 359 struct cp_stats stats; 360 }; 361 362 enum f2fs_cp_phase { 363 CP_PHASE_START_BLOCK_OPS, 364 CP_PHASE_FINISH_BLOCK_OPS, 365 CP_PHASE_FINISH_CHECKPOINT, 366 }; 367 368 /* 369 * indicate meta/data type 370 */ 371 enum { 372 META_CP, 373 META_NAT, 374 META_SIT, 375 META_SSA, 376 META_MAX, 377 META_POR, 378 DATA_GENERIC, /* check range only */ 379 DATA_GENERIC_ENHANCE, /* strong check on range and segment bitmap */ 380 DATA_GENERIC_ENHANCE_READ, /* 381 * strong check on range and segment 382 * bitmap but no warning due to race 383 * condition of read on truncated area 384 * by extent_cache 385 */ 386 DATA_GENERIC_ENHANCE_UPDATE, /* 387 * strong check on range and segment 388 * bitmap for update case 389 */ 390 META_GENERIC, 391 }; 392 393 /* for the list of ino */ 394 enum { 395 ORPHAN_INO, /* for orphan ino list */ 396 FLUSH_INO, /* for multiple device flushing */ 397 APPEND_INO, /* for append ino list */ 398 UPDATE_INO, /* for update ino list */ 399 TRANS_DIR_INO, /* for transactions dir ino list */ 400 XATTR_DIR_INO, /* for xattr updated dir ino list */ 401 MAX_INO_ENTRY, /* max. list */ 402 }; 403 404 #define INO_BITS_PER_SLOT BITS_PER_XA_VALUE 405 #define INO_SLOT_INDEX(ino) ((ino) / INO_BITS_PER_SLOT) 406 #define INO_BIT_OFFSET(ino) ((ino) % INO_BITS_PER_SLOT) 407 408 struct ino_entry { 409 struct list_head list; /* list head */ 410 nid_t ino; /* inode number */ 411 unsigned int dirty_device; /* dirty device bitmap */ 412 }; 413 414 /* for the list of inodes to be GCed */ 415 struct inode_entry { 416 struct list_head list; /* list head */ 417 struct inode *inode; /* vfs inode pointer */ 418 }; 419 420 struct fsync_node_entry { 421 struct list_head list; /* list head */ 422 struct f2fs_cached_block *entry; /* warm node cache entry pointer */ 423 unsigned int seq_id; /* sequence id */ 424 }; 425 426 struct ckpt_req { 427 struct completion wait; /* completion for checkpoint done */ 428 struct llist_node llnode; /* llist_node to be linked in wait queue */ 429 int ret; /* return code of checkpoint */ 430 union { 431 ktime_t queue_time; /* request queued time */ 432 ktime_t delta_time; /* time in queue */ 433 }; 434 }; 435 436 struct ckpt_req_control { 437 struct task_struct *f2fs_issue_ckpt; /* checkpoint task */ 438 int ckpt_thread_ioprio; /* checkpoint merge thread ioprio */ 439 wait_queue_head_t ckpt_wait_queue; /* waiting queue for wake-up */ 440 atomic_t issued_ckpt; /* # of actually issued ckpts */ 441 atomic_t total_ckpt; /* # of total ckpts */ 442 atomic_t queued_ckpt; /* # of queued ckpts */ 443 struct llist_head issue_list; /* list for command issue */ 444 spinlock_t stat_lock; /* lock for below checkpoint time stats */ 445 unsigned int cur_time; /* cur wait time in msec for currently issued checkpoint */ 446 unsigned int peak_time; /* peak wait time in msec until now */ 447 }; 448 449 /* a time threshold that checkpoint was blocked for, unit: ms */ 450 #define CP_LONG_LATENCY_THRESHOLD 5000 451 452 /* for the bitmap indicate blocks to be discarded */ 453 struct discard_entry { 454 struct list_head list; /* list head */ 455 block_t start_blkaddr; /* start blockaddr of current segment */ 456 unsigned char discard_map[SIT_VBLOCK_MAP_SIZE]; /* segment discard bitmap */ 457 }; 458 459 /* minimum discard granularity, unit: block count */ 460 #define MIN_DISCARD_GRANULARITY 1 461 /* default discard granularity of inner discard thread, unit: block count */ 462 #define DEFAULT_DISCARD_GRANULARITY 16 463 /* default maximum discard granularity of ordered discard, unit: block count */ 464 #define DEFAULT_MAX_ORDERED_DISCARD_GRANULARITY 16 465 /* default interval of periodical discard submission */ 466 #define DEFAULT_DISCARD_INTERVAL (msecs_to_jiffies(20)) 467 468 /* max discard pend list number */ 469 #define MAX_PLIST_NUM 512 470 #define plist_idx(blk_num) ((blk_num) >= MAX_PLIST_NUM ? \ 471 (MAX_PLIST_NUM - 1) : ((blk_num) - 1)) 472 #define MAX_DISCARD_DROP_COUNT 512 473 474 enum { 475 D_PREP, /* initial */ 476 D_PARTIAL, /* partially submitted */ 477 D_SUBMIT, /* all submitted */ 478 D_DONE, /* finished */ 479 }; 480 481 struct discard_info { 482 block_t lstart; /* logical start address */ 483 block_t len; /* length */ 484 block_t start; /* actual start address in dev */ 485 }; 486 487 struct discard_cmd { 488 struct rb_node rb_node; /* rb node located in rb-tree */ 489 struct discard_info di; /* discard info */ 490 struct list_head list; /* command list */ 491 struct completion wait; /* completion */ 492 struct block_device *bdev; /* bdev */ 493 unsigned short ref; /* reference count */ 494 unsigned char state; /* state */ 495 unsigned char queued; /* queued discard */ 496 int error; /* bio error */ 497 spinlock_t lock; /* for state/bio_ref updating */ 498 unsigned short bio_ref; /* bio reference count */ 499 }; 500 501 enum { 502 DPOLICY_BG, 503 DPOLICY_FORCE, 504 DPOLICY_FSTRIM, 505 DPOLICY_UMOUNT, 506 MAX_DPOLICY, 507 }; 508 509 enum { 510 DPOLICY_IO_AWARE_DISABLE, /* force to not be aware of IO */ 511 DPOLICY_IO_AWARE_ENABLE, /* force to be aware of IO */ 512 DPOLICY_IO_AWARE_MAX, 513 }; 514 515 struct discard_policy { 516 int type; /* type of discard */ 517 unsigned int min_interval; /* used for candidates exist */ 518 unsigned int mid_interval; /* used for device busy */ 519 unsigned int max_interval; /* used for candidates not exist */ 520 unsigned int max_requests; /* # of discards issued per round */ 521 unsigned int io_aware_gran; /* minimum granularity discard not be aware of I/O */ 522 bool io_aware; /* issue discard in idle time */ 523 bool sync; /* submit discard with REQ_SYNC flag */ 524 bool ordered; /* issue discard by lba order */ 525 bool timeout; /* discard timeout for put_super */ 526 unsigned int granularity; /* discard granularity */ 527 }; 528 529 struct discard_cmd_control { 530 struct task_struct *f2fs_issue_discard; /* discard thread */ 531 struct list_head entry_list; /* 4KB discard entry list */ 532 struct list_head pend_list[MAX_PLIST_NUM];/* store pending entries */ 533 struct list_head wait_list; /* store on-flushing entries */ 534 struct list_head fstrim_list; /* in-flight discard from fstrim */ 535 wait_queue_head_t discard_wait_queue; /* waiting queue for wake-up */ 536 struct mutex cmd_lock; 537 unsigned int nr_discards; /* # of discards in the list */ 538 unsigned int max_discards; /* max. discards to be issued */ 539 unsigned int max_discard_request; /* max. discard request per round */ 540 unsigned int min_discard_issue_time; /* min. interval between discard issue */ 541 unsigned int mid_discard_issue_time; /* mid. interval between discard issue */ 542 unsigned int max_discard_issue_time; /* max. interval between discard issue */ 543 unsigned int discard_io_aware_gran; /* minimum discard granularity not be aware of I/O */ 544 unsigned int discard_urgent_util; /* utilization which issue discard proactively */ 545 unsigned int discard_granularity; /* discard granularity */ 546 unsigned int max_ordered_discard; /* maximum discard granularity issued by lba order */ 547 unsigned int discard_io_aware; /* io_aware policy */ 548 unsigned int undiscard_blks; /* # of undiscard blocks */ 549 unsigned int next_pos; /* next discard position */ 550 atomic_t issued_discard; /* # of issued discard */ 551 atomic_t queued_discard; /* # of queued discard */ 552 atomic_t discard_cmd_cnt; /* # of cached cmd count */ 553 struct rb_root_cached root; /* root of discard rb-tree */ 554 bool rbtree_check; /* config for consistence check */ 555 bool discard_wake; /* to wake up discard thread */ 556 }; 557 558 /* for the list of fsync inodes, used only during recovery */ 559 struct fsync_inode_entry { 560 struct list_head list; /* list head */ 561 struct inode *inode; /* vfs inode pointer */ 562 block_t blkaddr; /* block address locating the last fsync */ 563 block_t last_dentry; /* block address locating the last dentry */ 564 }; 565 566 #define nats_in_cursum(jnl) (le16_to_cpu((jnl)->n_nats)) 567 #define sits_in_cursum(jnl) (le16_to_cpu((jnl)->n_sits)) 568 569 #define nat_in_journal(jnl, i) \ 570 (((struct nat_journal_entry *)(jnl)->nat_j.entries)[i].ne) 571 #define nid_in_journal(jnl, i) \ 572 (((struct nat_journal_entry *)(jnl)->nat_j.entries)[i].nid) 573 #define sit_in_journal(jnl, i) \ 574 (((struct sit_journal_entry *)(jnl)->sit_j.entries)[i].se) 575 #define segno_in_journal(jnl, i) \ 576 (((struct sit_journal_entry *)(jnl)->sit_j.entries)[i].segno) 577 578 #define sum_entries(sum) ((struct f2fs_summary *)(sum)) 579 #define sum_journal(sbi, sum) \ 580 ((struct f2fs_journal *)((char *)(sum) + \ 581 ((sbi)->entries_in_sum * sizeof(struct f2fs_summary)))) 582 #define sum_footer(sbi, sum) \ 583 ((struct summary_footer *)((char *)(sum) + (sbi)->sum_blocksize - \ 584 sizeof(struct summary_footer))) 585 586 #define MAX_NAT_JENTRIES(sbi, jnl) ((sbi)->nat_journal_entries - nats_in_cursum(jnl)) 587 #define MAX_SIT_JENTRIES(sbi, jnl) ((sbi)->sit_journal_entries - sits_in_cursum(jnl)) 588 589 static inline int update_nats_in_cursum(struct f2fs_journal *journal, int i) 590 { 591 int before = nats_in_cursum(journal); 592 593 journal->n_nats = cpu_to_le16(before + i); 594 return before; 595 } 596 597 static inline int update_sits_in_cursum(struct f2fs_journal *journal, int i) 598 { 599 int before = sits_in_cursum(journal); 600 601 journal->n_sits = cpu_to_le16(before + i); 602 return before; 603 } 604 605 /* for inline stuff */ 606 #define DEF_INLINE_RESERVED_SIZE 1 607 static inline int get_extra_isize(struct inode *inode); 608 static inline int get_inline_xattr_addrs(struct inode *inode); 609 static inline unsigned int cur_addrs_per_inode(struct inode *inode); 610 #define MAX_INLINE_DATA(inode) (sizeof(__le32) * \ 611 (cur_addrs_per_inode(inode) - \ 612 get_inline_xattr_addrs(inode) - \ 613 DEF_INLINE_RESERVED_SIZE)) 614 615 /* for inline dir */ 616 #define NR_INLINE_DENTRY(inode) (MAX_INLINE_DATA(inode) * BITS_PER_BYTE / \ 617 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \ 618 BITS_PER_BYTE + 1)) 619 #define INLINE_DENTRY_BITMAP_SIZE(inode) \ 620 DIV_ROUND_UP(NR_INLINE_DENTRY(inode), BITS_PER_BYTE) 621 #define INLINE_RESERVED_SIZE(inode) (MAX_INLINE_DATA(inode) - \ 622 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \ 623 NR_INLINE_DENTRY(inode) + \ 624 INLINE_DENTRY_BITMAP_SIZE(inode))) 625 626 /* 627 * For INODE and NODE manager 628 */ 629 /* for directory operations */ 630 631 struct f2fs_filename { 632 /* 633 * The filename the user specified. This is NULL for some 634 * filesystem-internal operations, e.g. converting an inline directory 635 * to a non-inline one, or roll-forward recovering an encrypted dentry. 636 */ 637 const struct qstr *usr_fname; 638 639 /* 640 * The on-disk filename. For encrypted directories, this is encrypted. 641 * This may be NULL for lookups in an encrypted dir without the key. 642 */ 643 struct fscrypt_str disk_name; 644 645 /* The dirhash of this filename */ 646 f2fs_hash_t hash; 647 648 #ifdef CONFIG_FS_ENCRYPTION 649 /* 650 * For lookups in encrypted directories: either the buffer backing 651 * disk_name, or a buffer that holds the decoded no-key name. 652 */ 653 struct fscrypt_str crypto_buf; 654 #endif 655 #if IS_ENABLED(CONFIG_UNICODE) 656 /* 657 * For casefolded directories: the casefolded name, but it's left NULL 658 * if the original name is not valid Unicode, if the original name is 659 * "." or "..", if the directory is both casefolded and encrypted and 660 * its encryption key is unavailable, or if the filesystem is doing an 661 * internal operation where usr_fname is also NULL. In all these cases 662 * we fall back to treating the name as an opaque byte sequence. 663 */ 664 struct qstr cf_name; 665 #endif 666 }; 667 668 struct f2fs_dentry_ptr { 669 struct inode *inode; 670 void *bitmap; 671 struct f2fs_dir_entry *dentry; 672 __u8 (*filename)[F2FS_SLOT_LEN]; 673 int max; 674 int nr_bitmap; 675 }; 676 677 static inline void make_dentry_ptr_inline(struct inode *inode, 678 struct f2fs_dentry_ptr *d, void *t) 679 { 680 int entry_cnt = NR_INLINE_DENTRY(inode); 681 int bitmap_size = INLINE_DENTRY_BITMAP_SIZE(inode); 682 int reserved_size = INLINE_RESERVED_SIZE(inode); 683 684 d->inode = inode; 685 d->max = entry_cnt; 686 d->nr_bitmap = bitmap_size; 687 d->bitmap = t; 688 d->dentry = t + bitmap_size + reserved_size; 689 d->filename = t + bitmap_size + reserved_size + 690 SIZE_OF_DIR_ENTRY * entry_cnt; 691 } 692 693 /* 694 * XATTR_NODE_OFFSET stores xattrs to one node block per file keeping -1 695 * as its node offset to distinguish from index node blocks. 696 * But some bits are used to mark the node block. 697 */ 698 #define XATTR_NODE_OFFSET ((((unsigned int)-1) << OFFSET_BIT_SHIFT) \ 699 >> OFFSET_BIT_SHIFT) 700 enum { 701 ALLOC_NODE, /* allocate a new node page if needed */ 702 LOOKUP_NODE, /* look up a node without readahead */ 703 LOOKUP_NODE_RA, /* 704 * look up a node with readahead called 705 * by get_data_block. 706 */ 707 }; 708 709 #define DEFAULT_RETRY_IO_COUNT 8 /* maximum retry read IO or flush count */ 710 711 #define MAX_FLUSH_RETRY_COUNT 3 /* maximum flush retry count in f2fs_enable_checkpoint() */ 712 713 /* IO/non-IO congestion wait timeout value, default: 1 jiffies */ 714 #define DEFAULT_SCHEDULE_TIMEOUT 1 715 716 /* timeout value injected, default: 1000ms */ 717 #define DEFAULT_FAULT_TIMEOUT (msecs_to_jiffies(1000)) 718 719 /* maximum retry quota flush count */ 720 #define DEFAULT_RETRY_QUOTA_FLUSH_COUNT 8 721 722 /* maximum retry of EIO'ed page */ 723 #define MAX_RETRY_PAGE_EIO 100 724 725 #define F2FS_LINK_MAX 0xffffffff /* maximum link count per file */ 726 727 #define MAX_DIR_RA_PAGES 4 /* maximum ra pages of dir */ 728 729 /* dirty segments threshold for triggering CP */ 730 #define DEFAULT_DIRTY_THRESHOLD 4 731 732 #define RECOVERY_MAX_RA_BLOCKS BIO_MAX_VECS 733 #define RECOVERY_MIN_RA_BLOCKS 1 734 735 #define F2FS_ONSTACK_PAGES 16 /* nr of onstack pages */ 736 737 /* for in-memory extent cache entry */ 738 #define F2FS_MIN_EXTENT_LEN 64 /* minimum extent length */ 739 740 /* number of extent info in extent cache we try to shrink */ 741 #define READ_EXTENT_CACHE_SHRINK_NUMBER 128 742 743 /* number of age extent info in extent cache we try to shrink */ 744 #define AGE_EXTENT_CACHE_SHRINK_NUMBER 128 745 #define LAST_AGE_WEIGHT 30 746 #define SAME_AGE_REGION 1024 747 748 /* 749 * Define data block with age less than 1GB as hot data 750 * define data block with age less than 10GB but more than 1GB as warm data 751 */ 752 #define DEF_HOT_DATA_AGE_THRESHOLD 262144 753 #define DEF_WARM_DATA_AGE_THRESHOLD 2621440 754 755 /* default max read extent count per inode */ 756 #define DEF_MAX_READ_EXTENT_COUNT 10240 757 758 /* extent cache type */ 759 enum extent_type { 760 EX_READ, 761 EX_BLOCK_AGE, 762 NR_EXTENT_CACHES, 763 }; 764 765 /* 766 * Reserved value to mark invalid age extents, hence valid block range 767 * from 0 to ULLONG_MAX-1 768 */ 769 #define F2FS_EXTENT_AGE_INVALID ULLONG_MAX 770 771 struct extent_info { 772 unsigned int fofs; /* start offset in a file */ 773 unsigned int len; /* length of the extent */ 774 union { 775 /* read extent_cache */ 776 struct { 777 /* start block address of the extent */ 778 block_t blk; 779 #ifdef CONFIG_F2FS_FS_COMPRESSION 780 /* physical extent length of compressed blocks */ 781 unsigned int c_len; 782 #endif 783 }; 784 /* block age extent_cache */ 785 struct { 786 /* block age of the extent */ 787 unsigned long long age; 788 /* last total blocks allocated */ 789 unsigned long long last_blocks; 790 }; 791 }; 792 }; 793 794 struct extent_node { 795 struct rb_node rb_node; /* rb node located in rb-tree */ 796 struct extent_info ei; /* extent info */ 797 struct list_head list; /* node in global extent list of sbi */ 798 struct extent_tree *et; /* extent tree pointer */ 799 }; 800 801 struct extent_tree { 802 nid_t ino; /* inode number */ 803 enum extent_type type; /* keep the extent tree type */ 804 struct rb_root_cached root; /* root of extent info rb-tree */ 805 struct extent_node *cached_en; /* recently accessed extent node */ 806 struct list_head list; /* to be used by sbi->zombie_list */ 807 rwlock_t lock; /* protect extent info rb-tree */ 808 atomic_t node_cnt; /* # of extent node in rb-tree*/ 809 bool largest_updated; /* largest extent updated */ 810 struct extent_info largest; /* largest cached extent for EX_READ */ 811 }; 812 813 struct extent_tree_info { 814 struct radix_tree_root extent_tree_root;/* cache extent cache entries */ 815 struct mutex extent_tree_lock; /* locking extent radix tree */ 816 struct list_head extent_list; /* lru list for shrinker */ 817 spinlock_t extent_lock; /* locking extent lru list */ 818 atomic_t total_ext_tree; /* extent tree count */ 819 struct list_head zombie_list; /* extent zombie tree list */ 820 atomic_t total_zombie_tree; /* extent zombie tree count */ 821 atomic_t total_ext_node; /* extent info count */ 822 }; 823 824 /* 825 * State of block returned by f2fs_map_blocks. 826 */ 827 #define F2FS_MAP_NEW (1U << 0) 828 #define F2FS_MAP_MAPPED (1U << 1) 829 #define F2FS_MAP_DELALLOC (1U << 2) 830 #define F2FS_MAP_FLAGS (F2FS_MAP_NEW | F2FS_MAP_MAPPED |\ 831 F2FS_MAP_DELALLOC) 832 833 struct f2fs_map_blocks { 834 struct block_device *m_bdev; /* for multi-device dio */ 835 block_t m_pblk; 836 block_t m_lblk; 837 unsigned int m_len; 838 unsigned int m_flags; 839 unsigned long m_last_pblk; /* last allocated block, only used for DIO in LFS mode */ 840 pgoff_t *m_next_pgofs; /* point next possible non-hole pgofs */ 841 pgoff_t *m_next_extent; /* point to next possible extent */ 842 int m_seg_type; 843 bool m_may_create; /* indicate it is from write path */ 844 bool m_multidev_dio; /* indicate it allows multi-device dio */ 845 }; 846 847 /* for flag in get_data_block */ 848 enum { 849 F2FS_GET_BLOCK_DEFAULT, 850 F2FS_GET_BLOCK_FIEMAP, 851 F2FS_GET_BLOCK_BMAP, 852 F2FS_GET_BLOCK_DIO, 853 F2FS_GET_BLOCK_PRE_DIO, 854 F2FS_GET_BLOCK_PRE_AIO, 855 F2FS_GET_BLOCK_PRECACHE, 856 }; 857 858 /* 859 * i_advise uses FADVISE_XXX_BIT. We can add additional hints later. 860 */ 861 #define FADVISE_COLD_BIT 0x01 862 #define FADVISE_LOST_PINO_BIT 0x02 863 #define FADVISE_ENCRYPT_BIT 0x04 864 #define FADVISE_ENC_NAME_BIT 0x08 865 #define FADVISE_KEEP_SIZE_BIT 0x10 866 #define FADVISE_HOT_BIT 0x20 867 #define FADVISE_VERITY_BIT 0x40 868 #define FADVISE_TRUNC_BIT 0x80 869 870 #define FADVISE_MODIFIABLE_BITS (FADVISE_COLD_BIT | FADVISE_HOT_BIT) 871 872 #define file_is_cold(inode) is_file(inode, FADVISE_COLD_BIT) 873 #define file_set_cold(inode) set_file(inode, FADVISE_COLD_BIT) 874 #define file_clear_cold(inode) clear_file(inode, FADVISE_COLD_BIT) 875 876 #define file_wrong_pino(inode) is_file(inode, FADVISE_LOST_PINO_BIT) 877 #define file_lost_pino(inode) set_file(inode, FADVISE_LOST_PINO_BIT) 878 #define file_got_pino(inode) clear_file(inode, FADVISE_LOST_PINO_BIT) 879 880 #define file_is_encrypt(inode) is_file(inode, FADVISE_ENCRYPT_BIT) 881 #define file_set_encrypt(inode) set_file(inode, FADVISE_ENCRYPT_BIT) 882 883 #define file_enc_name(inode) is_file(inode, FADVISE_ENC_NAME_BIT) 884 #define file_set_enc_name(inode) set_file(inode, FADVISE_ENC_NAME_BIT) 885 886 #define file_keep_isize(inode) is_file(inode, FADVISE_KEEP_SIZE_BIT) 887 #define file_set_keep_isize(inode) set_file(inode, FADVISE_KEEP_SIZE_BIT) 888 889 #define file_is_hot(inode) is_file(inode, FADVISE_HOT_BIT) 890 #define file_set_hot(inode) set_file(inode, FADVISE_HOT_BIT) 891 #define file_clear_hot(inode) clear_file(inode, FADVISE_HOT_BIT) 892 893 #define file_is_verity(inode) is_file(inode, FADVISE_VERITY_BIT) 894 #define file_set_verity(inode) set_file(inode, FADVISE_VERITY_BIT) 895 896 #define file_should_truncate(inode) is_file(inode, FADVISE_TRUNC_BIT) 897 #define file_need_truncate(inode) set_file(inode, FADVISE_TRUNC_BIT) 898 #define file_dont_truncate(inode) clear_file(inode, FADVISE_TRUNC_BIT) 899 900 #define DEF_DIR_LEVEL 0 901 902 /* used for f2fs_inode_info->flags */ 903 enum { 904 FI_NEW_INODE, /* indicate newly allocated inode */ 905 FI_DIRTY_INODE, /* indicate inode is dirty or not */ 906 FI_AUTO_RECOVER, /* indicate inode is recoverable */ 907 FI_DIRTY_DIR, /* indicate directory has dirty pages */ 908 FI_INC_LINK, /* need to increment i_nlink */ 909 FI_ACL_MODE, /* indicate acl mode */ 910 FI_NO_ALLOC, /* should not allocate any blocks */ 911 FI_FREE_NID, /* free allocated nide */ 912 FI_NO_EXTENT, /* not to use the extent cache */ 913 FI_INLINE_XATTR, /* used for inline xattr */ 914 FI_INLINE_DATA, /* used for inline data*/ 915 FI_INLINE_DENTRY, /* used for inline dentry */ 916 FI_APPEND_WRITE, /* inode has appended data */ 917 FI_UPDATE_WRITE, /* inode has in-place-update data */ 918 FI_NEED_IPU, /* used for ipu per file */ 919 FI_ATOMIC_FILE, /* indicate atomic file */ 920 FI_DATA_EXIST, /* indicate data exists */ 921 FI_SKIP_WRITES, /* should skip data page writeback */ 922 FI_OPU_WRITE, /* used for opu per file */ 923 FI_DIRTY_FILE, /* indicate regular/symlink has dirty pages */ 924 FI_PREALLOCATED_ALL, /* all blocks for write were preallocated */ 925 FI_HOT_DATA, /* indicate file is hot */ 926 FI_EXTRA_ATTR, /* indicate file has extra attribute */ 927 FI_PROJ_INHERIT, /* indicate file inherits projectid */ 928 FI_PIN_FILE, /* indicate file should not be gced */ 929 FI_VERITY_IN_PROGRESS, /* building fs-verity Merkle tree */ 930 FI_COMPRESSED_FILE, /* indicate file's data can be compressed */ 931 FI_COMPRESS_CORRUPT, /* indicate compressed cluster is corrupted */ 932 FI_MMAP_FILE, /* indicate file was mmapped */ 933 FI_ENABLE_COMPRESS, /* enable compression in "user" compression mode */ 934 FI_COMPRESS_RELEASED, /* compressed blocks were released */ 935 FI_ALIGNED_WRITE, /* enable aligned write */ 936 FI_COW_FILE, /* indicate COW file */ 937 FI_ATOMIC_COMMITTED, /* indicate atomic commit completed except disk sync */ 938 FI_ATOMIC_DIRTIED, /* indicate atomic file is dirtied */ 939 FI_ATOMIC_REPLACE, /* indicate atomic replace */ 940 FI_OPENED_FILE, /* indicate file has been opened */ 941 FI_DONATE_FINISHED, /* indicate page donation of file has been finished */ 942 FI_MAX, /* max flag, never be used */ 943 }; 944 945 struct f2fs_inode_info { 946 struct inode vfs_inode; /* serve a vfs inode */ 947 unsigned long i_flags; /* keep an inode flags for ioctl */ 948 unsigned char i_advise; /* use to give file attribute hints */ 949 unsigned char i_dir_level; /* use for dentry level for large dir */ 950 union { 951 unsigned int i_current_depth; /* only for directory depth */ 952 unsigned short i_gc_failures; /* for gc failure statistic */ 953 }; 954 unsigned int i_pino; /* parent inode number */ 955 umode_t i_acl_mode; /* keep file acl mode temporarily */ 956 957 /* Use below internally in f2fs*/ 958 unsigned long flags[BITS_TO_LONGS(FI_MAX)]; /* use to pass per-file flags */ 959 unsigned int ioprio_hint; /* hint for IO priority */ 960 struct f2fs_rwsem i_sem; /* protect fi info */ 961 atomic_t dirty_pages; /* # of dirty pages */ 962 f2fs_hash_t chash; /* hash value of given file name */ 963 unsigned int clevel; /* maximum level of given file name */ 964 struct task_struct *task; /* lookup and create consistency */ 965 struct task_struct *cp_task; /* separate cp/wb IO stats*/ 966 struct task_struct *wb_task; /* indicate inode is in context of writeback */ 967 nid_t i_xattr_nid; /* node id that contains xattrs */ 968 loff_t last_disk_size; /* lastly written file size */ 969 spinlock_t i_size_lock; /* protect last_disk_size */ 970 971 #ifdef CONFIG_QUOTA 972 struct dquot __rcu *i_dquot[MAXQUOTAS]; 973 974 /* quota space reservation, managed internally by quota code */ 975 qsize_t i_reserved_quota; 976 #endif 977 struct list_head dirty_list; /* dirty list for dirs and files */ 978 struct list_head gdirty_list; /* linked in global dirty list */ 979 980 /* linked in global inode list for cache donation */ 981 struct list_head gdonate_list; 982 pgoff_t donate_start, donate_end; /* inclusive */ 983 atomic_t open_count; /* # of open files */ 984 985 struct task_struct *atomic_write_task; /* store atomic write task */ 986 struct extent_tree *extent_tree[NR_EXTENT_CACHES]; 987 /* cached extent_tree entry */ 988 union { 989 struct inode *cow_inode; /* copy-on-write inode for atomic write */ 990 struct inode *atomic_inode; 991 /* point to atomic_inode, available only for cow_inode */ 992 }; 993 994 /* avoid racing between foreground op and gc */ 995 struct f2fs_rwsem i_gc_rwsem[2]; 996 struct f2fs_rwsem i_xattr_sem; /* avoid racing between reading and changing EAs */ 997 998 int i_extra_isize; /* size of extra space located in i_addr */ 999 kprojid_t i_projid; /* id for project quota */ 1000 int i_inline_xattr_size; /* inline xattr size */ 1001 struct timespec64 i_crtime; /* inode creation time */ 1002 struct timespec64 i_disk_time[3];/* inode disk times */ 1003 1004 /* for file compress */ 1005 atomic_t i_compr_blocks; /* # of compressed blocks */ 1006 unsigned char i_compress_algorithm; /* algorithm type */ 1007 unsigned char i_log_cluster_size; /* log of cluster size */ 1008 unsigned char i_compress_level; /* compress level (lz4hc,zstd) */ 1009 unsigned char i_compress_flag; /* compress flag */ 1010 unsigned int i_cluster_size; /* cluster size */ 1011 atomic_t writeback; /* count # of writeback thread */ 1012 1013 unsigned int atomic_write_cnt; 1014 loff_t original_i_size; /* original i_size before atomic write */ 1015 #ifdef CONFIG_FS_ENCRYPTION 1016 struct fscrypt_inode_info *i_crypt_info; /* filesystem encryption info */ 1017 #endif 1018 }; 1019 1020 static inline void get_read_extent_info(struct extent_info *ext, 1021 struct f2fs_extent *i_ext) 1022 { 1023 ext->fofs = le32_to_cpu(i_ext->fofs); 1024 ext->blk = le32_to_cpu(i_ext->blk); 1025 ext->len = le32_to_cpu(i_ext->len); 1026 } 1027 1028 static inline void set_raw_read_extent(struct extent_info *ext, 1029 struct f2fs_extent *i_ext) 1030 { 1031 i_ext->fofs = cpu_to_le32(ext->fofs); 1032 i_ext->blk = cpu_to_le32(ext->blk); 1033 i_ext->len = cpu_to_le32(ext->len); 1034 } 1035 1036 static inline bool __is_discard_mergeable(struct discard_info *back, 1037 struct discard_info *front, unsigned int max_len) 1038 { 1039 return (back->lstart + back->len == front->lstart) && 1040 (back->len + front->len <= max_len); 1041 } 1042 1043 static inline bool __is_discard_back_mergeable(struct discard_info *cur, 1044 struct discard_info *back, unsigned int max_len) 1045 { 1046 return __is_discard_mergeable(back, cur, max_len); 1047 } 1048 1049 static inline bool __is_discard_front_mergeable(struct discard_info *cur, 1050 struct discard_info *front, unsigned int max_len) 1051 { 1052 return __is_discard_mergeable(cur, front, max_len); 1053 } 1054 1055 /* 1056 * For free nid management 1057 */ 1058 enum nid_state { 1059 FREE_NID, /* newly added to free nid list */ 1060 PREALLOC_NID, /* it is preallocated */ 1061 MAX_NID_STATE, 1062 }; 1063 1064 enum nat_state { 1065 TOTAL_NAT, 1066 DIRTY_NAT, 1067 RECLAIMABLE_NAT, 1068 MAX_NAT_STATE, 1069 }; 1070 1071 struct f2fs_nm_info { 1072 block_t nat_blkaddr; /* base disk address of NAT */ 1073 nid_t max_nid; /* maximum possible node ids */ 1074 nid_t available_nids; /* # of available node ids */ 1075 nid_t next_scan_nid; /* the next nid to be scanned */ 1076 nid_t max_rf_node_blocks; /* max # of nodes for recovery */ 1077 unsigned int ram_thresh; /* control the memory footprint */ 1078 unsigned int ra_nid_blocks; /* # of nid blocks to be readaheaded */ 1079 unsigned int dirty_nats_ratio; /* control dirty nats ratio threshold */ 1080 1081 /* NAT cache management */ 1082 struct radix_tree_root nat_root;/* root of the nat entry cache */ 1083 struct radix_tree_root nat_set_root;/* root of the nat set cache */ 1084 struct f2fs_rwsem nat_tree_lock; /* protect nat entry tree */ 1085 struct list_head nat_entries; /* cached nat entry list (clean) */ 1086 spinlock_t nat_list_lock; /* protect clean nat entry list */ 1087 unsigned int nat_cnt[MAX_NAT_STATE]; /* the # of cached nat entries */ 1088 unsigned int nat_blocks; /* # of nat blocks */ 1089 1090 /* free node ids management */ 1091 struct radix_tree_root free_nid_root;/* root of the free_nid cache */ 1092 struct list_head free_nid_list; /* list for free nids excluding preallocated nids */ 1093 unsigned int nid_cnt[MAX_NID_STATE]; /* the number of free node id */ 1094 spinlock_t nid_list_lock; /* protect nid lists ops */ 1095 struct mutex build_lock; /* lock for build free nids */ 1096 unsigned char **free_nid_bitmap; 1097 unsigned char *nat_block_bitmap; 1098 unsigned short *free_nid_count; /* free nid count of NAT block */ 1099 1100 /* for checkpoint */ 1101 char *nat_bitmap; /* NAT bitmap pointer */ 1102 1103 unsigned int nat_bits_blocks; /* # of nat bits blocks */ 1104 unsigned char *nat_bits; /* NAT bits blocks */ 1105 unsigned char *full_nat_bits; /* full NAT pages */ 1106 unsigned char *empty_nat_bits; /* empty NAT pages */ 1107 #ifdef CONFIG_F2FS_CHECK_FS 1108 char *nat_bitmap_mir; /* NAT bitmap mirror */ 1109 #endif 1110 int bitmap_size; /* bitmap size */ 1111 }; 1112 1113 /* 1114 * this structure is used as one of function parameters. 1115 * all the information are dedicated to a given direct node block determined 1116 * by the data offset in a file. 1117 */ 1118 struct dnode_of_data { 1119 struct inode *inode; /* vfs inode pointer */ 1120 struct f2fs_cached_block *inode_entry; /* generic cache inode entry */ 1121 struct f2fs_cached_block *node_entry; /* generic cache node entry */ 1122 nid_t nid; /* node id of the direct node block */ 1123 unsigned int ofs_in_node; /* data offset in the node page */ 1124 bool inode_entry_locked; /* inode entry is locked or not */ 1125 bool node_changed; /* is node block changed */ 1126 char cur_level; /* level of hole node page */ 1127 char max_level; /* level of current page located */ 1128 block_t data_blkaddr; /* block address of the node block */ 1129 }; 1130 1131 static inline void set_new_dnode(struct dnode_of_data *dn, struct inode *inode, 1132 struct f2fs_cached_block *ientry, struct f2fs_cached_block *nentry, nid_t nid) 1133 { 1134 memset(dn, 0, sizeof(*dn)); 1135 dn->inode = inode; 1136 dn->inode_entry = ientry; 1137 dn->node_entry = nentry; 1138 dn->nid = nid; 1139 } 1140 1141 /* 1142 * For SIT manager 1143 * 1144 * By default, there are 6 active log areas across the whole main area. 1145 * When considering hot and cold data separation to reduce cleaning overhead, 1146 * we split 3 for data logs and 3 for node logs as hot, warm, and cold types, 1147 * respectively. 1148 * In the current design, you should not change the numbers intentionally. 1149 * Instead, as a mount option such as active_logs=x, you can use 2, 4, and 6 1150 * logs individually according to the underlying devices. (default: 6) 1151 * Just in case, on-disk layout covers maximum 16 logs that consist of 8 for 1152 * data and 8 for node logs. 1153 */ 1154 #define NR_CURSEG_DATA_TYPE (3) 1155 #define NR_CURSEG_NODE_TYPE (3) 1156 #define NR_CURSEG_INMEM_TYPE (2) 1157 #define NR_CURSEG_RO_TYPE (2) 1158 #define NR_CURSEG_PERSIST_TYPE (NR_CURSEG_DATA_TYPE + NR_CURSEG_NODE_TYPE) 1159 #define NR_CURSEG_TYPE (NR_CURSEG_INMEM_TYPE + NR_CURSEG_PERSIST_TYPE) 1160 1161 enum log_type { 1162 CURSEG_HOT_DATA = 0, /* directory entry blocks */ 1163 CURSEG_WARM_DATA, /* data blocks */ 1164 CURSEG_COLD_DATA, /* multimedia or GCed data blocks */ 1165 CURSEG_HOT_NODE, /* direct node blocks of directory files */ 1166 CURSEG_WARM_NODE, /* direct node blocks of normal files */ 1167 CURSEG_COLD_NODE, /* indirect node blocks */ 1168 NR_PERSISTENT_LOG, /* number of persistent log */ 1169 CURSEG_COLD_DATA_PINNED = NR_PERSISTENT_LOG, 1170 /* pinned file that needs consecutive block address */ 1171 CURSEG_ALL_DATA_ATGC, /* SSR alloctor in hot/warm/cold data area */ 1172 NO_CHECK_TYPE, /* number of persistent & inmem log */ 1173 }; 1174 1175 struct flush_cmd { 1176 struct completion wait; 1177 struct llist_node llnode; 1178 nid_t ino; 1179 int ret; 1180 }; 1181 1182 struct flush_cmd_control { 1183 struct task_struct *f2fs_issue_flush; /* flush thread */ 1184 wait_queue_head_t flush_wait_queue; /* waiting queue for wake-up */ 1185 atomic_t issued_flush; /* # of issued flushes */ 1186 atomic_t queued_flush; /* # of queued flushes */ 1187 struct llist_head issue_list; /* list for command issue */ 1188 struct llist_node *dispatch_list; /* list for command dispatch */ 1189 }; 1190 1191 struct f2fs_sm_info { 1192 struct sit_info *sit_info; /* whole segment information */ 1193 struct free_segmap_info *free_info; /* free segment information */ 1194 struct dirty_seglist_info *dirty_info; /* dirty segment information */ 1195 struct curseg_info *curseg_array; /* active segment information */ 1196 1197 struct f2fs_rwsem curseg_lock; /* for preventing curseg change */ 1198 1199 block_t seg0_blkaddr; /* block address of 0'th segment */ 1200 block_t main_blkaddr; /* start block address of main area */ 1201 block_t ssa_blkaddr; /* start block address of SSA area */ 1202 1203 unsigned int segment_count; /* total # of segments */ 1204 unsigned int main_segments; /* # of segments in main area */ 1205 unsigned int reserved_segments; /* # of reserved segments */ 1206 unsigned int ovp_segments; /* # of overprovision segments */ 1207 1208 /* a threshold to reclaim prefree segments */ 1209 unsigned int rec_prefree_segments; 1210 1211 struct list_head sit_entry_set; /* sit entry set list */ 1212 1213 unsigned int ipu_policy; /* in-place-update policy */ 1214 unsigned int min_ipu_util; /* in-place-update threshold */ 1215 unsigned int min_fsync_blocks; /* threshold for fsync */ 1216 unsigned int min_seq_blocks; /* threshold for sequential blocks */ 1217 unsigned int min_hot_blocks; /* threshold for hot block allocation */ 1218 unsigned int min_ssr_sections; /* threshold to trigger SSR allocation */ 1219 1220 /* for flush command control */ 1221 struct flush_cmd_control *fcc_info; 1222 1223 /* for discard command control */ 1224 struct discard_cmd_control *dcc_info; 1225 }; 1226 1227 /* 1228 * For superblock 1229 */ 1230 /* 1231 * COUNT_TYPE for monitoring 1232 * 1233 * f2fs monitors the number of several block types such as on-writeback, 1234 * dirty dentry blocks, dirty node blocks, and dirty meta blocks. 1235 */ 1236 #define WB_DATA_TYPE(folio, f) \ 1237 (f || f2fs_is_cp_guaranteed(folio) ? F2FS_WB_CP_DATA : F2FS_WB_DATA) 1238 enum count_type { 1239 F2FS_DIRTY_DENTS, 1240 F2FS_DIRTY_DATA, 1241 F2FS_DIRTY_QDATA, 1242 F2FS_DIRTY_NODES, 1243 F2FS_DIRTY_META, 1244 F2FS_DIRTY_IMETA, 1245 F2FS_WB_CP_DATA, 1246 F2FS_WB_DATA, 1247 F2FS_RD_DATA, 1248 F2FS_RD_NODE, 1249 F2FS_RD_META, 1250 F2FS_DIO_WRITE, 1251 F2FS_DIO_READ, 1252 F2FS_SKIPPED_WRITE, /* skip or fail during f2fs_enable_checkpoint() */ 1253 NR_COUNT_TYPE, 1254 }; 1255 1256 /* 1257 * The below are the page types of bios used in submit_bio(). 1258 * The available types are: 1259 * DATA User data pages. It operates as async mode. 1260 * NODE Node pages. It operates as async mode. 1261 * META FS metadata pages such as SIT, NAT, CP. 1262 * NR_PAGE_TYPE The number of page types. 1263 * META_FLUSH Make sure the previous pages are written 1264 * with waiting the bio's completion 1265 * ... Only can be used with META. 1266 */ 1267 #define PAGE_TYPE_OF_BIO(type) ((type) > META ? META : (type)) 1268 #define PAGE_TYPE_ON_MAIN(type) ((type) == DATA || (type) == NODE) 1269 enum page_type { 1270 DATA = 0, 1271 NODE = 1, /* should not change this */ 1272 META, 1273 NR_PAGE_TYPE, 1274 META_FLUSH, 1275 IPU, /* the below types are used by tracepoints only. */ 1276 OPU, 1277 }; 1278 1279 enum temp_type { 1280 HOT = 0, /* must be zero for meta bio */ 1281 WARM, 1282 COLD, 1283 NR_TEMP_TYPE, 1284 }; 1285 1286 enum need_lock_type { 1287 LOCK_REQ = 0, 1288 LOCK_DONE, 1289 LOCK_RETRY, 1290 }; 1291 1292 enum cp_reason_type { 1293 CP_NO_NEEDED, 1294 CP_NON_REGULAR, 1295 CP_COMPRESSED, 1296 CP_HARDLINK, 1297 CP_SB_NEED_CP, 1298 CP_WRONG_PINO, 1299 CP_NO_SPC_ROLL, 1300 CP_NODE_NEED_CP, 1301 CP_FASTBOOT_MODE, 1302 CP_SPEC_LOG_NUM, 1303 CP_RECOVER_DIR, 1304 CP_XATTR_DIR, 1305 }; 1306 1307 enum iostat_type { 1308 /* WRITE IO */ 1309 APP_DIRECT_IO, /* app direct write IOs */ 1310 APP_BUFFERED_IO, /* app buffered write IOs */ 1311 APP_WRITE_IO, /* app write IOs */ 1312 APP_MAPPED_IO, /* app mapped IOs */ 1313 APP_BUFFERED_CDATA_IO, /* app buffered write IOs on compressed file */ 1314 APP_MAPPED_CDATA_IO, /* app mapped write IOs on compressed file */ 1315 FS_DATA_IO, /* data IOs from kworker/fsync/reclaimer */ 1316 FS_CDATA_IO, /* data IOs from kworker/fsync/reclaimer on compressed file */ 1317 FS_NODE_IO, /* node IOs from kworker/fsync/reclaimer */ 1318 FS_META_IO, /* meta IOs from kworker/reclaimer */ 1319 FS_GC_DATA_IO, /* data IOs from forground gc */ 1320 FS_GC_NODE_IO, /* node IOs from forground gc */ 1321 FS_CP_DATA_IO, /* data IOs from checkpoint */ 1322 FS_CP_NODE_IO, /* node IOs from checkpoint */ 1323 FS_CP_META_IO, /* meta IOs from checkpoint */ 1324 1325 /* READ IO */ 1326 APP_DIRECT_READ_IO, /* app direct read IOs */ 1327 APP_BUFFERED_READ_IO, /* app buffered read IOs */ 1328 APP_READ_IO, /* app read IOs */ 1329 APP_MAPPED_READ_IO, /* app mapped read IOs */ 1330 APP_BUFFERED_CDATA_READ_IO, /* app buffered read IOs on compressed file */ 1331 APP_MAPPED_CDATA_READ_IO, /* app mapped read IOs on compressed file */ 1332 FS_DATA_READ_IO, /* data read IOs */ 1333 FS_GDATA_READ_IO, /* data read IOs from background gc */ 1334 FS_CDATA_READ_IO, /* compressed data read IOs */ 1335 FS_NODE_READ_IO, /* node read IOs */ 1336 FS_META_READ_IO, /* meta read IOs */ 1337 1338 /* other */ 1339 FS_DISCARD_IO, /* discard */ 1340 FS_FLUSH_IO, /* flush */ 1341 FS_ZONE_RESET_IO, /* zone reset */ 1342 NR_IO_TYPE, 1343 }; 1344 1345 struct f2fs_io_info { 1346 struct f2fs_sb_info *sbi; /* f2fs_sb_info pointer */ 1347 nid_t ino; /* inode number */ 1348 enum page_type type; /* contains DATA/NODE/META/META_FLUSH */ 1349 enum temp_type temp; /* contains HOT/WARM/COLD */ 1350 enum req_op op; /* contains REQ_OP_ */ 1351 blk_opf_t op_flags; /* req_flag_bits */ 1352 block_t new_blkaddr; /* new block address to be written */ 1353 block_t old_blkaddr; /* old block address before Cow */ 1354 struct folio *folio; /* folio to be written */ 1355 struct page *encrypted_page; /* encrypted page */ 1356 struct page *compressed_page; /* compressed page */ 1357 struct list_head list; /* serialize IOs */ 1358 unsigned int compr_blocks; /* # of compressed block addresses */ 1359 unsigned int need_lock:8; /* indicate we need to lock cp_rwsem */ 1360 unsigned int version:8; /* version of the node */ 1361 unsigned int submitted:1; /* indicate IO submission */ 1362 unsigned int in_list:1; /* indicate fio is in io_list */ 1363 unsigned int is_por:1; /* indicate IO is from recovery or not */ 1364 unsigned int meta_gc:1; /* require meta inode GC */ 1365 unsigned int is_cache:1; /* indicate IO is from internal cache */ 1366 enum iostat_type io_type; /* io type */ 1367 struct writeback_control *io_wbc; /* writeback control */ 1368 struct f2fs_cached_block *cache_entry; 1369 struct bio **bio; /* bio for ipu */ 1370 sector_t *last_block; /* last block number in bio */ 1371 }; 1372 1373 struct bio_entry { 1374 struct bio *bio; 1375 struct list_head list; 1376 }; 1377 1378 #define is_read_io(rw) ((rw) == READ) 1379 struct f2fs_bio_info { 1380 struct f2fs_sb_info *sbi; /* f2fs superblock */ 1381 struct bio *bio; /* bios to merge */ 1382 sector_t last_block_in_bio; /* last block number */ 1383 struct f2fs_io_info fio; /* store buffered io info. */ 1384 #ifdef CONFIG_BLK_DEV_ZONED 1385 struct completion zone_wait; /* condition value for the previous open zone to close */ 1386 struct bio *zone_pending_bio; /* pending bio for the previous zone */ 1387 void *bi_private; /* previous bi_private for pending bio */ 1388 #endif 1389 struct f2fs_rwsem io_rwsem; /* blocking op for bio */ 1390 spinlock_t io_lock; /* serialize DATA/NODE IOs */ 1391 struct list_head io_list; /* track fios */ 1392 struct list_head bio_list; /* bio entry list head */ 1393 struct f2fs_rwsem bio_list_lock; /* lock to protect bio entry list */ 1394 }; 1395 1396 #define FDEV(i) (sbi->devs[i]) 1397 #define RDEV(i) (raw_super->devs[i]) 1398 struct f2fs_dev_info { 1399 struct file *bdev_file; 1400 struct block_device *bdev; 1401 char path[MAX_PATH_LEN + 1]; 1402 unsigned int total_segments; 1403 block_t start_blk; 1404 block_t end_blk; 1405 bool has_alias; 1406 bool is_reserving; 1407 #ifdef CONFIG_BLK_DEV_ZONED 1408 unsigned int nr_blkz; /* Total number of zones */ 1409 unsigned long *blkz_seq; /* Bitmap indicating sequential zones */ 1410 #endif 1411 }; 1412 1413 enum inode_type { 1414 DIR_INODE, /* for dirty dir inode */ 1415 FILE_INODE, /* for dirty regular/symlink inode */ 1416 DIRTY_META, /* for all dirtied inode metadata */ 1417 DONATE_INODE, /* for all inode to donate pages */ 1418 NR_INODE_TYPE, 1419 }; 1420 1421 /* for inner inode cache management */ 1422 struct inode_management { 1423 struct radix_tree_root ino_root; /* ino entry array */ 1424 spinlock_t ino_lock; /* for ino entry lock */ 1425 struct list_head ino_list; /* inode list head */ 1426 unsigned long ino_num; /* number of entries */ 1427 }; 1428 1429 /* for GC_AT */ 1430 struct atgc_management { 1431 bool atgc_enabled; /* ATGC is enabled or not */ 1432 struct rb_root_cached root; /* root of victim rb-tree */ 1433 struct list_head victim_list; /* linked with all victim entries */ 1434 unsigned int victim_count; /* victim count in rb-tree */ 1435 unsigned int candidate_ratio; /* candidate ratio */ 1436 unsigned int max_candidate_count; /* max candidate count */ 1437 unsigned int age_weight; /* age weight, vblock_weight = 100 - age_weight */ 1438 unsigned long long age_threshold; /* age threshold */ 1439 }; 1440 1441 struct f2fs_time_stat { 1442 unsigned long long total_time; /* total wall clock time */ 1443 #ifdef CONFIG_64BIT 1444 unsigned long long running_time; /* running time */ 1445 #endif 1446 #if defined(CONFIG_SCHED_INFO) && defined(CONFIG_SCHEDSTATS) 1447 unsigned long long runnable_time; /* runnable(including preempted) time */ 1448 #endif 1449 #ifdef CONFIG_TASK_DELAY_ACCT 1450 unsigned long long io_sleep_time; /* IO sleep time */ 1451 #endif 1452 }; 1453 1454 struct f2fs_lock_context { 1455 struct f2fs_time_stat ts; 1456 int orig_nice; 1457 int new_nice; 1458 bool lock_trace; 1459 bool need_restore; 1460 }; 1461 1462 struct f2fs_gc_control { 1463 unsigned int victim_segno; /* target victim segment number */ 1464 int init_gc_type; /* FG_GC or BG_GC */ 1465 bool no_bg_gc; /* check the space and stop bg_gc */ 1466 bool should_migrate_blocks; /* should migrate blocks */ 1467 bool err_gc_skipped; /* return EAGAIN if GC skipped */ 1468 bool one_time; /* require one time GC in one migration unit */ 1469 unsigned int nr_free_secs; /* # of free sections to do GC */ 1470 struct f2fs_lock_context lc; /* lock context for gc_lock */ 1471 }; 1472 1473 /* 1474 * For s_flag in struct f2fs_sb_info 1475 * Modification on enum should be synchronized with s_flag array 1476 */ 1477 enum { 1478 SBI_IS_DIRTY, /* dirty flag for checkpoint */ 1479 SBI_IS_CLOSE, /* specify unmounting */ 1480 SBI_NEED_FSCK, /* need fsck.f2fs to fix */ 1481 SBI_POR_DOING, /* recovery is doing or not */ 1482 SBI_NEED_SB_WRITE, /* need to recover superblock */ 1483 SBI_NEED_CP, /* need to checkpoint */ 1484 SBI_IS_SHUTDOWN, /* shutdown by ioctl */ 1485 SBI_IS_RECOVERED, /* recovered orphan/data */ 1486 SBI_CP_DISABLED, /* CP was disabled last mount */ 1487 SBI_CP_DISABLED_QUICK, /* CP was disabled quickly */ 1488 SBI_QUOTA_NEED_FLUSH, /* need to flush quota info in CP */ 1489 SBI_QUOTA_SKIP_FLUSH, /* skip flushing quota in current CP */ 1490 SBI_QUOTA_NEED_REPAIR, /* quota file may be corrupted */ 1491 SBI_IS_RESIZEFS, /* resizefs is in process */ 1492 SBI_IS_FREEZING, /* freezefs is in process */ 1493 SBI_IS_WRITABLE, /* remove ro mountoption transiently */ 1494 SBI_ENABLE_CHECKPOINT, /* indicate it's during f2fs_enable_checkpoint() */ 1495 MAX_SBI_FLAG, 1496 }; 1497 1498 enum { 1499 CP_TIME, 1500 REQ_TIME, 1501 DISCARD_TIME, 1502 GC_TIME, 1503 DISABLE_TIME, 1504 UMOUNT_DISCARD_TIMEOUT, 1505 MAX_TIME, 1506 }; 1507 1508 /* Note that you need to keep synchronization with this gc_mode_names array */ 1509 enum { 1510 GC_NORMAL, 1511 GC_IDLE_CB, 1512 GC_IDLE_GREEDY, 1513 GC_IDLE_AT, 1514 GC_URGENT_HIGH, 1515 GC_URGENT_LOW, 1516 GC_URGENT_MID, 1517 MAX_GC_MODE, 1518 }; 1519 1520 enum { 1521 BGGC_MODE_ON, /* background gc is on */ 1522 BGGC_MODE_OFF, /* background gc is off */ 1523 BGGC_MODE_SYNC, /* 1524 * background gc is on, migrating blocks 1525 * like foreground gc 1526 */ 1527 }; 1528 1529 enum { 1530 FS_MODE_ADAPTIVE, /* use both lfs/ssr allocation */ 1531 FS_MODE_LFS, /* use lfs allocation only */ 1532 FS_MODE_FRAGMENT_SEG, /* segment fragmentation mode */ 1533 FS_MODE_FRAGMENT_BLK, /* block fragmentation mode */ 1534 }; 1535 1536 enum { 1537 ALLOC_MODE_DEFAULT, /* stay default */ 1538 ALLOC_MODE_REUSE, /* reuse segments as much as possible */ 1539 }; 1540 1541 enum fsync_mode { 1542 FSYNC_MODE_POSIX, /* fsync follows posix semantics */ 1543 FSYNC_MODE_STRICT, /* fsync behaves in line with ext4 */ 1544 FSYNC_MODE_NOBARRIER, /* fsync behaves nobarrier based on posix */ 1545 }; 1546 1547 enum { 1548 COMPR_MODE_FS, /* 1549 * automatically compress compression 1550 * enabled files 1551 */ 1552 COMPR_MODE_USER, /* 1553 * automatical compression is disabled. 1554 * user can control the file compression 1555 * using ioctls 1556 */ 1557 }; 1558 1559 enum { 1560 DISCARD_UNIT_BLOCK, /* basic discard unit is block */ 1561 DISCARD_UNIT_SEGMENT, /* basic discard unit is segment */ 1562 DISCARD_UNIT_SECTION, /* basic discard unit is section */ 1563 }; 1564 1565 enum { 1566 MEMORY_MODE_NORMAL, /* memory mode for normal devices */ 1567 MEMORY_MODE_LOW, /* memory mode for low memory devices */ 1568 }; 1569 1570 enum errors_option { 1571 MOUNT_ERRORS_READONLY, /* remount fs ro on errors */ 1572 MOUNT_ERRORS_CONTINUE, /* continue on errors */ 1573 MOUNT_ERRORS_PANIC, /* panic on errors */ 1574 }; 1575 1576 enum { 1577 BACKGROUND, 1578 FOREGROUND, 1579 MAX_CALL_TYPE, 1580 TOTAL_CALL = FOREGROUND, 1581 }; 1582 1583 enum f2fs_lookup_mode { 1584 LOOKUP_PERF, 1585 LOOKUP_COMPAT, 1586 LOOKUP_AUTO, 1587 }; 1588 1589 /* For node type in __get_node_folio() */ 1590 enum node_type { 1591 NODE_TYPE_REGULAR, 1592 NODE_TYPE_INODE, 1593 NODE_TYPE_XATTR, 1594 NODE_TYPE_NON_INODE, 1595 NODE_TYPE_NON_IXNODE, /* non inode and xnode */ 1596 }; 1597 1598 /* a threshold of maximum elapsed time in critical region to print tracepoint */ 1599 #define MAX_LOCK_ELAPSED_TIME 500 1600 1601 #define F2FS_DEFAULT_TASK_PRIORITY (DEFAULT_PRIO) 1602 #define F2FS_CRITICAL_TASK_PRIORITY NICE_TO_PRIO(0) 1603 1604 static inline int f2fs_test_bit(unsigned int nr, char *addr); 1605 static inline void f2fs_set_bit(unsigned int nr, char *addr); 1606 static inline void f2fs_clear_bit(unsigned int nr, char *addr); 1607 1608 /* 1609 * Layout of f2fs folio->private: 1610 * 1611 * Layout A: lowest bit should be 1 1612 * | bit0 = 1 | bit1 | bit2 | ... | bit MAX | private data .... | 1613 * bit 0 F2FS_FOLIO_PRIVATE_NOT_POINTER 1614 * bit 1 F2FS_FOLIO_PRIVATE_ONGOING_MIGRATION 1615 * bit 2 F2FS_FOLIO_PRIVATE_REF_RESOURCE 1616 * bit 3 F2FS_FOLIO_PRIVATE_ATOMIC_WRITE 1617 * bit 4- f2fs private data 1618 * 1619 * Layout B: lowest bit should be 0 1620 * folio->private is a wrapped pointer. 1621 */ 1622 enum { 1623 F2FS_FOLIO_PRIVATE_NOT_POINTER, /* private contains non-pointer data */ 1624 F2FS_FOLIO_PRIVATE_ONGOING_MIGRATION, /* data page which is on-going migrating */ 1625 F2FS_FOLIO_PRIVATE_REF_RESOURCE, /* dirty page has referenced resources */ 1626 F2FS_FOLIO_PRIVATE_ATOMIC_WRITE, /* data page from atomic write path */ 1627 F2FS_FOLIO_PRIVATE_MAX 1628 }; 1629 1630 /* For compression */ 1631 enum compress_algorithm_type { 1632 COMPRESS_LZO, 1633 COMPRESS_LZ4, 1634 COMPRESS_ZSTD, 1635 COMPRESS_LZORLE, 1636 COMPRESS_MAX, 1637 }; 1638 1639 enum compress_flag { 1640 COMPRESS_CHKSUM, 1641 COMPRESS_MAX_FLAG, 1642 }; 1643 1644 #define COMPRESS_WATERMARK 20 1645 #define COMPRESS_PERCENT 20 1646 1647 #define COMPRESS_DATA_RESERVED_SIZE 4 1648 struct compress_data { 1649 __le32 clen; /* compressed data size */ 1650 __le32 chksum; /* compressed data checksum */ 1651 __le32 reserved[COMPRESS_DATA_RESERVED_SIZE]; /* reserved */ 1652 u8 cdata[]; /* compressed data */ 1653 }; 1654 1655 #define COMPRESS_HEADER_SIZE (sizeof(struct compress_data)) 1656 1657 #define F2FS_COMPRESSED_PAGE_MAGIC 0xF5F2C000 1658 1659 #define F2FS_ZSTD_DEFAULT_CLEVEL 1 1660 1661 #define COMPRESS_LEVEL_OFFSET 8 1662 1663 /* compress context */ 1664 struct compress_ctx { 1665 struct inode *inode; /* inode the context belong to */ 1666 pgoff_t cluster_idx; /* cluster index number */ 1667 unsigned int cluster_size; /* page count in cluster */ 1668 unsigned int log_cluster_size; /* log of cluster size */ 1669 struct page **rpages; /* pages store raw data in cluster */ 1670 unsigned int nr_rpages; /* total page number in rpages */ 1671 struct page **cpages; /* pages store compressed data in cluster */ 1672 unsigned int nr_cpages; /* total page number in cpages */ 1673 unsigned int valid_nr_cpages; /* valid page number in cpages */ 1674 void *rbuf; /* virtual mapped address on rpages */ 1675 struct compress_data *cbuf; /* virtual mapped address on cpages */ 1676 size_t rlen; /* valid data length in rbuf */ 1677 size_t clen; /* valid data length in cbuf */ 1678 void *private; /* payload buffer for specified compression algorithm */ 1679 void *private2; /* extra payload buffer */ 1680 struct fsverity_info *vi; /* verity info if needed */ 1681 }; 1682 1683 /* compress context for write IO path */ 1684 struct compress_io_ctx { 1685 u32 magic; /* magic number to indicate page is compressed */ 1686 struct inode *inode; /* inode the context belong to */ 1687 struct page **rpages; /* pages store raw data in cluster */ 1688 unsigned int nr_rpages; /* total page number in rpages */ 1689 atomic_t pending_pages; /* in-flight compressed page count */ 1690 }; 1691 1692 /* Context for decompressing one cluster on the read IO path */ 1693 struct decompress_io_ctx { 1694 u32 magic; /* magic number to indicate page is compressed */ 1695 struct inode *inode; /* inode the context belong to */ 1696 struct f2fs_sb_info *sbi; /* f2fs_sb_info pointer */ 1697 pgoff_t cluster_idx; /* cluster index number */ 1698 unsigned int cluster_size; /* page count in cluster */ 1699 unsigned int log_cluster_size; /* log of cluster size */ 1700 struct page **rpages; /* pages store raw data in cluster */ 1701 unsigned int nr_rpages; /* total page number in rpages */ 1702 struct page **cpages; /* pages store compressed data in cluster */ 1703 unsigned int nr_cpages; /* total page number in cpages */ 1704 struct page **tpages; /* temp pages to pad holes in cluster */ 1705 void *rbuf; /* virtual mapped address on rpages */ 1706 struct compress_data *cbuf; /* virtual mapped address on cpages */ 1707 size_t rlen; /* valid data length in rbuf */ 1708 size_t clen; /* valid data length in cbuf */ 1709 1710 /* 1711 * The number of compressed pages remaining to be read in this cluster. 1712 * This is initially nr_cpages. It is decremented by 1 each time a page 1713 * has been read (or failed to be read). When it reaches 0, the cluster 1714 * is decompressed (or an error is reported). 1715 * 1716 * If an error occurs before all the pages have been submitted for I/O, 1717 * then this will never reach 0. In this case the I/O submitter is 1718 * responsible for calling f2fs_decompress_end_io() instead. 1719 */ 1720 atomic_t remaining_pages; 1721 1722 /* 1723 * Number of references to this decompress_io_ctx. 1724 * 1725 * One reference is held for I/O completion. This reference is dropped 1726 * after the pagecache pages are updated and unlocked -- either after 1727 * decompression (and verity if enabled), or after an error. 1728 * 1729 * In addition, each compressed page holds a reference while it is in a 1730 * bio. These references are necessary prevent compressed pages from 1731 * being freed while they are still in a bio. 1732 */ 1733 refcount_t refcnt; 1734 1735 bool failed; /* IO error occurred before decompression? */ 1736 struct fsverity_info *vi; /* fs-verity context if needed */ 1737 unsigned char compress_algorithm; /* backup algorithm type */ 1738 void *private; /* payload buffer for specified decompression algorithm */ 1739 void *private2; /* extra payload buffer */ 1740 struct work_struct verity_work; /* work to verify the decompressed pages */ 1741 struct work_struct free_work; /* work for late free this structure itself */ 1742 }; 1743 1744 #define NULL_CLUSTER ((unsigned int)(~0)) 1745 #define MIN_COMPRESS_LOG_SIZE 2 1746 #define MAX_COMPRESS_LOG_SIZE 8 1747 #define MAX_COMPRESS_WINDOW_SIZE(log_size) ((PAGE_SIZE) << (log_size)) 1748 1749 struct f2fs_gc_kthread { 1750 struct task_struct *f2fs_gc_task; 1751 wait_queue_head_t gc_wait_queue_head; 1752 1753 /* for gc sleep time */ 1754 unsigned int urgent_sleep_time; 1755 unsigned int min_sleep_time; 1756 unsigned int max_sleep_time; 1757 unsigned int no_gc_sleep_time; 1758 1759 /* for changing gc mode */ 1760 bool gc_wake; 1761 1762 /* for GC_MERGE mount option */ 1763 wait_queue_head_t fggc_wq; /* 1764 * caller of f2fs_balance_fs() 1765 * will wait on this wait queue. 1766 */ 1767 1768 /* for gc control for zoned devices */ 1769 unsigned int no_zoned_gc_percent; 1770 unsigned int boost_zoned_gc_percent; 1771 unsigned int valid_thresh_ratio; 1772 unsigned int boost_gc_multiple; 1773 unsigned int boost_gc_greedy; 1774 }; 1775 1776 struct f2fs_bio { 1777 struct work_struct work; 1778 struct f2fs_cached_block *entry; 1779 unsigned int entry_cnt; 1780 struct bio bio; 1781 }; 1782 1783 struct f2fs_sb_info { 1784 struct super_block *sb; /* pointer to VFS super block */ 1785 struct proc_dir_entry *s_proc; /* proc entry */ 1786 struct f2fs_super_block *raw_super; /* raw super block pointer */ 1787 struct f2fs_rwsem sb_lock; /* lock for raw super block */ 1788 int valid_super_block; /* valid super block no */ 1789 unsigned long s_flag; /* flags for sbi */ 1790 struct mutex writepages; /* mutex for writepages() */ 1791 1792 #ifdef CONFIG_BLK_DEV_ZONED 1793 unsigned int blocks_per_blkz; /* F2FS blocks per zone */ 1794 unsigned int unusable_blocks_per_sec; /* unusable blocks per section */ 1795 unsigned int max_open_zones; /* max open zone resources of the zoned device */ 1796 /* For adjust the priority writing position of data in zone UFS */ 1797 unsigned int blkzone_alloc_policy; 1798 #endif 1799 1800 /* for node-related operations */ 1801 struct f2fs_nm_info *nm_info; /* node manager */ 1802 1803 /* for segment-related operations */ 1804 struct f2fs_sm_info *sm_info; /* segment manager */ 1805 1806 /* for bio operations */ 1807 /* Largest write bio size completed in atomic context (atc). */ 1808 u32 max_atc_write_bio_size; 1809 /* Largest write bio entry count completed in atomic context (atc). */ 1810 u32 max_atc_write_bio_entry_cnt; 1811 struct f2fs_bio_info *write_io[NR_PAGE_TYPE]; /* for write bios */ 1812 /* keep migration IO order for LFS mode */ 1813 struct f2fs_rwsem io_order_lock; 1814 pgoff_t page_eio_ofs[NR_PAGE_TYPE]; /* EIO page offset */ 1815 int page_eio_cnt[NR_PAGE_TYPE]; /* EIO count */ 1816 1817 /* for checkpoint */ 1818 struct f2fs_checkpoint *ckpt; /* raw checkpoint pointer */ 1819 int cur_cp_pack; /* remain current cp pack */ 1820 spinlock_t cp_lock; /* for flag in ckpt */ 1821 struct f2fs_rwsem cp_global_sem; /* checkpoint procedure lock */ 1822 struct f2fs_rwsem cp_rwsem; /* blocking FS operations */ 1823 struct f2fs_rwsem node_write; /* locking node writes */ 1824 struct f2fs_rwsem node_change; /* locking node change */ 1825 wait_queue_head_t cp_wait; 1826 unsigned long last_time[MAX_TIME]; /* to store time in jiffies */ 1827 long interval_time[MAX_TIME]; /* to store thresholds */ 1828 struct ckpt_req_control cprc_info; /* for checkpoint request control */ 1829 struct cp_stats cp_stats; /* for time stat of checkpoint */ 1830 1831 struct inode_management im[MAX_INO_ENTRY]; /* manage inode cache */ 1832 1833 spinlock_t fsync_node_lock; /* for node entry lock */ 1834 struct list_head fsync_node_list; /* node list head */ 1835 unsigned int fsync_seg_id; /* sequence id */ 1836 unsigned int fsync_node_num; /* number of node entries */ 1837 1838 /* for orphan inode, use 0'th array */ 1839 unsigned int max_orphans; /* max orphan inodes */ 1840 1841 /* for inode management */ 1842 struct list_head inode_list[NR_INODE_TYPE]; /* dirty inode list */ 1843 spinlock_t inode_lock[NR_INODE_TYPE]; /* for dirty inode list lock */ 1844 struct mutex flush_lock; /* for flush exclusion */ 1845 1846 /* for extent tree cache */ 1847 struct extent_tree_info extent_tree[NR_EXTENT_CACHES]; 1848 atomic64_t allocated_data_blocks; /* for block age extent_cache */ 1849 unsigned int max_read_extent_count; /* max read extent count per inode */ 1850 1851 /* The threshold used for hot and warm data seperation*/ 1852 unsigned int hot_data_age_threshold; 1853 unsigned int warm_data_age_threshold; 1854 unsigned int last_age_weight; 1855 1856 /* control donate caches */ 1857 unsigned int donate_files; 1858 1859 /* basic filesystem units */ 1860 unsigned int log_sectors_per_block; /* log2 sectors per block */ 1861 unsigned int log_blocksize; /* log2 block size */ 1862 unsigned int blocksize; /* block size */ 1863 unsigned int nat_entries_per_block; /* NAT entries in a block */ 1864 unsigned int addrs_per_inode; /* addresses in an inode block */ 1865 unsigned int addrs_per_block; /* addresses in a direct node block */ 1866 unsigned int nids_per_block; /* node IDs in an indirect node block */ 1867 unsigned int sit_entries_per_block; /* SIT entries in a block */ 1868 unsigned int orphans_per_block; /* orphan inodes in a block */ 1869 unsigned int dentries_per_block; /* dentries in a block */ 1870 unsigned int dentry_bitmap_size; /* dentry bitmap size in bytes */ 1871 unsigned int dentry_reserved_size; /* dentry reserved bytes */ 1872 unsigned int root_ino_num; /* root inode number*/ 1873 unsigned int node_ino_num; /* node inode number*/ 1874 unsigned int log_blocks_per_seg; /* log2 blocks per segment */ 1875 unsigned int blocks_per_seg; /* blocks per segment */ 1876 unsigned int segs_per_sec; /* segments per section */ 1877 unsigned int secs_per_zone; /* sections per zone */ 1878 unsigned int total_sections; /* total section count */ 1879 unsigned int total_node_count; /* total node block count */ 1880 unsigned int total_valid_node_count; /* valid node block count */ 1881 int dir_level; /* directory level */ 1882 bool readdir_ra; /* readahead inode in readdir */ 1883 unsigned int max_io_bytes; /* max io bytes to merge IOs */ 1884 1885 /* variable summary block units */ 1886 unsigned int sum_blocksize; /* sum block size */ 1887 unsigned int sums_per_block; /* sum block count per block */ 1888 unsigned int entries_in_sum; /* entry count in sum block */ 1889 unsigned int sum_entry_size; /* total entry size in sum block */ 1890 unsigned int sum_journal_size; /* journal size in sum block */ 1891 unsigned int nat_journal_entries; /* nat journal entry count in the journal */ 1892 unsigned int sit_journal_entries; /* sit journal entry count in the journal */ 1893 1894 block_t user_block_count; /* # of user blocks */ 1895 block_t total_valid_block_count; /* # of valid blocks */ 1896 block_t discard_blks; /* discard command candidats */ 1897 block_t last_valid_block_count; /* for recovery */ 1898 block_t reserved_blocks; /* configurable reserved blocks */ 1899 block_t current_reserved_blocks; /* current reserved blocks */ 1900 block_t alias_reserved_blocks; /* reserved blocks for device alias */ 1901 1902 /* Additional tracking for no checkpoint mode */ 1903 block_t unusable_block_count; /* # of blocks saved by last cp */ 1904 1905 unsigned int nquota_files; /* # of quota sysfile */ 1906 struct f2fs_rwsem quota_sem; /* blocking cp for flags */ 1907 struct task_struct *umount_lock_holder; /* s_umount lock holder */ 1908 1909 /* # of cache entries, see count_type */ 1910 atomic_t nr_caches[NR_COUNT_TYPE]; 1911 /* # of allocated blocks */ 1912 struct percpu_counter alloc_valid_block_count; 1913 /* # of node block writes as roll forward recovery */ 1914 struct percpu_counter rf_node_block_count; 1915 1916 /* writeback control */ 1917 atomic_t wb_sync_req[META]; /* count # of WB_SYNC threads */ 1918 1919 /* valid inode count */ 1920 struct percpu_counter total_valid_inode_count; 1921 1922 struct f2fs_mount_info mount_opt; /* mount options */ 1923 1924 /* for cleaning operations */ 1925 struct f2fs_rwsem gc_lock; /* 1926 * semaphore for GC, avoid 1927 * race between GC and GC or CP 1928 */ 1929 struct f2fs_gc_kthread gc_thread; /* GC thread */ 1930 struct atgc_management am; /* atgc management */ 1931 unsigned int cur_victim_sec; /* current victim section num */ 1932 unsigned int gc_mode; /* current GC state */ 1933 unsigned int next_victim_seg[2]; /* next segment in victim section */ 1934 spinlock_t gc_remaining_trials_lock; 1935 /* remaining trial count for GC_URGENT_* and GC_IDLE_* */ 1936 unsigned int gc_remaining_trials; 1937 1938 /* for skip statistic */ 1939 unsigned long long skipped_gc_rwsem; /* FG_GC only */ 1940 1941 /* free sections reserved for pinned file */ 1942 unsigned int reserved_pin_section; 1943 1944 /* threshold for gc trials on pinned files */ 1945 unsigned short gc_pin_file_threshold; 1946 struct f2fs_rwsem pin_sem; 1947 1948 /* maximum # of trials to find a victim segment for SSR and GC */ 1949 unsigned int max_victim_search; 1950 /* migration granularity of garbage collection, unit: segment */ 1951 unsigned int migration_granularity; 1952 /* migration window granularity of garbage collection, unit: segment */ 1953 unsigned int migration_window_granularity; 1954 1955 /* 1956 * for stat information. 1957 * one is for the LFS mode, and the other is for the SSR mode. 1958 */ 1959 #ifdef CONFIG_F2FS_STAT_FS 1960 struct f2fs_stat_info *stat_info; /* FS status information */ 1961 atomic_t meta_count[META_MAX]; /* # of meta blocks */ 1962 unsigned int segment_count[2]; /* # of allocated segments */ 1963 unsigned int block_count[2]; /* # of allocated blocks */ 1964 atomic_t inplace_count; /* # of inplace update */ 1965 /* # of lookup extent cache */ 1966 atomic64_t total_hit_ext[NR_EXTENT_CACHES]; 1967 /* # of hit rbtree extent node */ 1968 atomic64_t read_hit_rbtree[NR_EXTENT_CACHES]; 1969 /* # of hit cached extent node */ 1970 atomic64_t read_hit_cached[NR_EXTENT_CACHES]; 1971 /* # of hit largest extent node in read extent cache */ 1972 atomic64_t read_hit_largest; 1973 atomic_t inline_xattr; /* # of inline_xattr inodes */ 1974 atomic_t inline_inode; /* # of inline_data inodes */ 1975 atomic_t inline_dir; /* # of inline_dentry inodes */ 1976 atomic_t compr_inode; /* # of compressed inodes */ 1977 atomic64_t compr_blocks; /* # of compressed blocks */ 1978 atomic_t swapfile_inode; /* # of swapfile inodes */ 1979 atomic_t atomic_files; /* # of opened atomic file */ 1980 atomic_t max_aw_cnt; /* max # of atomic writes */ 1981 unsigned int io_skip_bggc; /* skip background gc for in-flight IO */ 1982 unsigned int other_skip_bggc; /* skip background gc for other reasons */ 1983 unsigned int ndirty_inode[NR_INODE_TYPE]; /* # of dirty inodes */ 1984 atomic_t cp_call_count[MAX_CALL_TYPE]; /* # of cp call */ 1985 #endif 1986 spinlock_t stat_lock; /* lock for stat operations */ 1987 1988 /* to attach REQ_META|REQ_FUA flags */ 1989 unsigned int data_io_flag; 1990 unsigned int node_io_flag; 1991 1992 /* For sysfs support */ 1993 struct kobject s_kobj; /* /sys/fs/f2fs/<devname> */ 1994 struct completion s_kobj_unregister; 1995 1996 struct kobject s_stat_kobj; /* /sys/fs/f2fs/<devname>/stat */ 1997 struct completion s_stat_kobj_unregister; 1998 1999 struct kobject s_feature_list_kobj; /* /sys/fs/f2fs/<devname>/feature_list */ 2000 struct completion s_feature_list_kobj_unregister; 2001 2002 /* For shrinker support */ 2003 struct list_head s_list; 2004 struct mutex umount_mutex; 2005 unsigned int shrinker_run_no; 2006 2007 /* For multi devices */ 2008 int s_ndevs; /* number of devices */ 2009 struct f2fs_dev_info *devs; /* for device list */ 2010 unsigned int dirty_device; /* for checkpoint data flush */ 2011 spinlock_t dev_lock; /* protect dirty_device */ 2012 bool aligned_blksize; /* all devices has the same logical blksize */ 2013 unsigned int first_seq_zone_segno; /* first segno in sequential zone */ 2014 unsigned int pinned_area_max_secno; /* upper bound section for pinned files */ 2015 unsigned int bggc_io_aware; /* For adjust the BG_GC priority when pending IO */ 2016 unsigned int allocate_section_hint; /* the boundary position between devices */ 2017 unsigned int allocate_section_policy; /* determine the section writing priority */ 2018 2019 /* For write statistics */ 2020 u64 sectors_written_start; 2021 u64 kbytes_written; 2022 2023 /* Precomputed FS UUID checksum for seeding other checksums */ 2024 __u32 s_chksum_seed; 2025 2026 struct workqueue_struct *wq; /* bio completion workqueue */ 2027 2028 struct workqueue_struct *evict_wq; /* inode eviction workqueue */ 2029 2030 /* 2031 * If we are in irq context, let's update error information into 2032 * on-disk superblock in the work. 2033 */ 2034 struct work_struct s_error_work; 2035 unsigned char errors[MAX_F2FS_ERRORS]; /* error flags */ 2036 unsigned char stop_reason[MAX_STOP_REASON]; /* stop reason */ 2037 spinlock_t error_lock; /* protect errors/stop_reason array */ 2038 bool error_dirty; /* errors of sb is dirty */ 2039 bool stop_reason_dirty; /* stop reason of sb is dirty */ 2040 2041 /* For reclaimed segs statistics per each GC mode */ 2042 unsigned int gc_segment_mode; /* GC state for reclaimed segments */ 2043 unsigned int gc_reclaimed_segs[MAX_GC_MODE]; /* Reclaimed segs for each mode */ 2044 2045 unsigned int seq_file_ra_mul; /* multiplier for ra_pages of seq. files in fadvise */ 2046 2047 int max_fragment_chunk; /* max chunk size for block fragmentation mode */ 2048 int max_fragment_hole; /* max hole size for block fragmentation mode */ 2049 2050 /* For atomic write statistics */ 2051 atomic64_t current_atomic_write; 2052 s64 peak_atomic_write; 2053 u64 committed_atomic_block; 2054 u64 revoked_atomic_block; 2055 2056 /* carve out reserved_blocks from total blocks */ 2057 bool carve_out; 2058 2059 /* max elapsed time threshold in critical region that lock covered */ 2060 unsigned long long max_lock_elapsed_time; 2061 2062 /* enable/disable to adjust task priority in critical region covered by lock */ 2063 unsigned int adjust_lock_priority; 2064 2065 /* adjust priority for task which is in critical region covered by lock */ 2066 unsigned int lock_duration_priority; 2067 2068 /* priority for critical task, e.g. f2fs_ckpt, f2fs_gc threads */ 2069 long critical_task_priority; 2070 2071 #ifdef CONFIG_F2FS_FS_COMPRESSION 2072 struct kmem_cache *page_array_slab; /* page array entry */ 2073 unsigned int page_array_slab_size; /* default page array slab size */ 2074 2075 /* For runtime compression statistics */ 2076 u64 compr_written_block; 2077 u64 compr_saved_block; 2078 u32 compr_new_inode; 2079 2080 /* For compressed block cache */ 2081 unsigned int compress_percent; /* cache page percentage */ 2082 unsigned int compress_watermark; /* cache page watermark */ 2083 atomic_t compress_page_hit; /* cache hit count */ 2084 #endif 2085 2086 #ifdef CONFIG_F2FS_IOSTAT 2087 /* For app/fs IO statistics */ 2088 spinlock_t iostat_lock; 2089 unsigned long long iostat_count[NR_IO_TYPE]; 2090 unsigned long long iostat_bytes[NR_IO_TYPE]; 2091 unsigned long long prev_iostat_bytes[NR_IO_TYPE]; 2092 unsigned long long iostat_read_folio_count[NR_PAGE_ORDERS]; 2093 unsigned long long prev_iostat_read_folio_count[NR_PAGE_ORDERS]; 2094 bool iostat_enable; 2095 unsigned long iostat_next_period; 2096 unsigned int iostat_period_ms; 2097 2098 /* For io latency related statistics info in one iostat period */ 2099 spinlock_t iostat_lat_lock; 2100 struct iostat_lat_info *iostat_io_lat; 2101 #endif 2102 #ifdef CONFIG_DEBUG_LOCK_ALLOC 2103 struct lock_class_key cp_global_sem_key; 2104 #endif 2105 2106 /* f2fs internal cache */ 2107 struct f2fs_cached_block_list meta_blocks; 2108 struct f2fs_cached_block_list node_blocks; 2109 struct f2fs_cached_block_list compress_blocks; 2110 2111 /* internal cache flush thread */ 2112 struct f2fs_cache_kthread cache_thread; 2113 }; 2114 2115 /* Definitions to access f2fs_sb_info */ 2116 #define SEGS_TO_BLKS(sbi, segs) \ 2117 ((segs) << (sbi)->log_blocks_per_seg) 2118 #define BLKS_TO_SEGS(sbi, blks) \ 2119 ((blks) >> (sbi)->log_blocks_per_seg) 2120 2121 #define BLKS_PER_SEG(sbi) ((sbi)->blocks_per_seg) 2122 #define BLKS_PER_SEC(sbi) (SEGS_TO_BLKS(sbi, (sbi)->segs_per_sec)) 2123 #define SEGS_PER_SEC(sbi) ((sbi)->segs_per_sec) 2124 2125 __printf(3, 4) 2126 void f2fs_printk(struct f2fs_sb_info *sbi, bool limit_rate, const char *fmt, ...); 2127 2128 #define f2fs_err(sbi, fmt, ...) \ 2129 f2fs_printk(sbi, false, KERN_ERR fmt, ##__VA_ARGS__) 2130 #define f2fs_warn(sbi, fmt, ...) \ 2131 f2fs_printk(sbi, false, KERN_WARNING fmt, ##__VA_ARGS__) 2132 #define f2fs_notice(sbi, fmt, ...) \ 2133 f2fs_printk(sbi, false, KERN_NOTICE fmt, ##__VA_ARGS__) 2134 #define f2fs_info(sbi, fmt, ...) \ 2135 f2fs_printk(sbi, false, KERN_INFO fmt, ##__VA_ARGS__) 2136 #define f2fs_debug(sbi, fmt, ...) \ 2137 f2fs_printk(sbi, false, KERN_DEBUG fmt, ##__VA_ARGS__) 2138 2139 #define f2fs_err_ratelimited(sbi, fmt, ...) \ 2140 f2fs_printk(sbi, true, KERN_ERR fmt, ##__VA_ARGS__) 2141 #define f2fs_warn_ratelimited(sbi, fmt, ...) \ 2142 f2fs_printk(sbi, true, KERN_WARNING fmt, ##__VA_ARGS__) 2143 #define f2fs_info_ratelimited(sbi, fmt, ...) \ 2144 f2fs_printk(sbi, true, KERN_INFO fmt, ##__VA_ARGS__) 2145 2146 #ifdef CONFIG_F2FS_FAULT_INJECTION 2147 #define time_to_inject(sbi, type) __time_to_inject(sbi, type, __func__, \ 2148 __builtin_return_address(0)) 2149 static inline bool __time_to_inject(struct f2fs_sb_info *sbi, int type, 2150 const char *func, const char *parent_func) 2151 { 2152 struct f2fs_fault_info *ffi = &F2FS_OPTION(sbi).fault_info; 2153 2154 if (!ffi->inject_rate) 2155 return false; 2156 2157 if (!IS_FAULT_SET(ffi, type)) 2158 return false; 2159 2160 atomic_inc(&ffi->inject_ops); 2161 if (atomic_read(&ffi->inject_ops) >= ffi->inject_rate) { 2162 atomic_set(&ffi->inject_ops, 0); 2163 ffi->inject_count[type]++; 2164 f2fs_info_ratelimited(sbi, "inject %s in %s of %pS", 2165 f2fs_fault_name[type], func, parent_func); 2166 return true; 2167 } 2168 return false; 2169 } 2170 #else 2171 static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type) 2172 { 2173 return false; 2174 } 2175 #endif 2176 2177 /* 2178 * Test if the mounted volume is a multi-device volume. 2179 * - For a single regular disk volume, sbi->s_ndevs is 0. 2180 * - For a single zoned disk volume, sbi->s_ndevs is 1. 2181 * - For a multi-device volume, sbi->s_ndevs is always 2 or more. 2182 */ 2183 static inline bool f2fs_is_multi_device(struct f2fs_sb_info *sbi) 2184 { 2185 return sbi->s_ndevs > 1; 2186 } 2187 2188 static inline void f2fs_update_time(struct f2fs_sb_info *sbi, int type) 2189 { 2190 unsigned long now = jiffies; 2191 2192 WRITE_ONCE(sbi->last_time[type], now); 2193 2194 /* DISCARD_TIME and GC_TIME are based on REQ_TIME */ 2195 if (type == REQ_TIME) { 2196 WRITE_ONCE(sbi->last_time[DISCARD_TIME], now); 2197 WRITE_ONCE(sbi->last_time[GC_TIME], now); 2198 } 2199 } 2200 2201 static inline bool f2fs_time_over(struct f2fs_sb_info *sbi, int type) 2202 { 2203 unsigned long interval = sbi->interval_time[type] * HZ; 2204 2205 return time_after(jiffies, READ_ONCE(sbi->last_time[type]) + interval); 2206 } 2207 2208 static inline unsigned int f2fs_time_to_wait(struct f2fs_sb_info *sbi, 2209 int type) 2210 { 2211 unsigned long interval = sbi->interval_time[type] * HZ; 2212 unsigned int wait_ms = 0; 2213 long delta; 2214 2215 delta = (READ_ONCE(sbi->last_time[type]) + interval) - jiffies; 2216 if (delta > 0) 2217 wait_ms = jiffies_to_msecs(delta); 2218 2219 return wait_ms; 2220 } 2221 2222 /* 2223 * Inline functions 2224 */ 2225 static inline u32 __f2fs_crc32(u32 crc, const void *address, 2226 unsigned int length) 2227 { 2228 return crc32(crc, address, length); 2229 } 2230 2231 static inline u32 f2fs_crc32(const void *address, unsigned int length) 2232 { 2233 return __f2fs_crc32(F2FS_SUPER_MAGIC, address, length); 2234 } 2235 2236 static inline u32 f2fs_chksum(u32 crc, const void *address, unsigned int length) 2237 { 2238 return __f2fs_crc32(crc, address, length); 2239 } 2240 2241 static inline struct f2fs_inode_info *F2FS_I(struct inode *inode) 2242 { 2243 return container_of(inode, struct f2fs_inode_info, vfs_inode); 2244 } 2245 2246 static inline struct f2fs_sb_info *F2FS_SB(struct super_block *sb) 2247 { 2248 return sb->s_fs_info; 2249 } 2250 2251 static inline struct f2fs_sb_info *F2FS_I_SB(struct inode *inode) 2252 { 2253 return F2FS_SB(inode->i_sb); 2254 } 2255 2256 static inline struct f2fs_sb_info *F2FS_M_SB(struct address_space *mapping) 2257 { 2258 return F2FS_I_SB(mapping->host); 2259 } 2260 2261 static inline struct f2fs_sb_info *F2FS_F_SB(const struct folio *folio) 2262 { 2263 return F2FS_M_SB(folio->mapping); 2264 } 2265 2266 #define SIT_ENTRY_PER_BLOCK(sbi) ((sbi)->sit_entries_per_block) 2267 #define NAT_ENTRY_PER_BLOCK(sbi) ((sbi)->nat_entries_per_block) 2268 #define DEF_ADDRS_PER_INODE(sbi) ((sbi)->addrs_per_inode) 2269 #define DEF_ADDRS_PER_BLOCK(sbi) ((sbi)->addrs_per_block) 2270 #define NIDS_PER_BLOCK(sbi) ((sbi)->nids_per_block) 2271 #define F2FS_ORPHANS_PER_BLOCK(sbi) ((sbi)->orphans_per_block) 2272 #define GET_ORPHAN_BLOCKS(sbi, n) DIV_ROUND_UP((n), \ 2273 F2FS_ORPHANS_PER_BLOCK(sbi)) 2274 #define CP_CHKSUM_OFFSET(sbi) (F2FS_BLKSIZE(sbi) - sizeof(__le32)) 2275 2276 #define NODE_DIR1_BLOCK(sbi) (DEF_ADDRS_PER_INODE(sbi) + 1) 2277 #define NODE_DIR2_BLOCK(sbi) (DEF_ADDRS_PER_INODE(sbi) + 2) 2278 #define NODE_IND1_BLOCK(sbi) (DEF_ADDRS_PER_INODE(sbi) + 3) 2279 #define NODE_IND2_BLOCK(sbi) (DEF_ADDRS_PER_INODE(sbi) + 4) 2280 #define NODE_DIND_BLOCK(sbi) (DEF_ADDRS_PER_INODE(sbi) + 5) 2281 2282 static inline struct f2fs_orphan_footer * 2283 f2fs_orphan_footer(void *orphan_block, struct f2fs_sb_info *sbi) 2284 { 2285 return (struct f2fs_orphan_footer *) 2286 ((char *)orphan_block + sbi->blocksize - 2287 sizeof(struct f2fs_orphan_footer)); 2288 } 2289 2290 static inline void make_dentry_ptr_block(struct inode *inode, 2291 struct f2fs_dentry_ptr *d, void *t) 2292 { 2293 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 2294 unsigned int entries = sbi->dentries_per_block; 2295 unsigned int bitmap_size = sbi->dentry_bitmap_size; 2296 unsigned int reserved_size = sbi->dentry_reserved_size; 2297 2298 d->inode = inode; 2299 d->max = entries; 2300 d->nr_bitmap = bitmap_size; 2301 d->bitmap = t; 2302 d->dentry = t + bitmap_size + reserved_size; 2303 d->filename = t + bitmap_size + reserved_size + 2304 SIZE_OF_DIR_ENTRY * entries; 2305 } 2306 2307 static inline struct f2fs_super_block *F2FS_RAW_SUPER(struct f2fs_sb_info *sbi) 2308 { 2309 return (struct f2fs_super_block *)(sbi->raw_super); 2310 } 2311 2312 static inline struct f2fs_super_block *F2FS_SUPER_BLOCK(struct folio *folio, 2313 pgoff_t index) 2314 { 2315 pgoff_t idx_in_folio = index % folio_nr_pages(folio); 2316 2317 return (struct f2fs_super_block *) 2318 (page_address(folio_page(folio, idx_in_folio)) + 2319 F2FS_SUPER_OFFSET); 2320 } 2321 2322 static inline struct f2fs_checkpoint *F2FS_CKPT(struct f2fs_sb_info *sbi) 2323 { 2324 return (struct f2fs_checkpoint *)(sbi->ckpt); 2325 } 2326 2327 static inline struct node_footer *F2FS_NODE_FOOTER(struct f2fs_sb_info *sbi, 2328 const struct f2fs_cached_block *entry) 2329 { 2330 return (struct node_footer *)(cache_address(entry) + 2331 F2FS_BLKSIZE(sbi) - sizeof(struct node_footer)); 2332 } 2333 2334 static inline struct f2fs_node *F2FS_NODE(const struct f2fs_cached_block *entry) 2335 { 2336 return (struct f2fs_node *)CACHED_NODE(entry); 2337 } 2338 2339 static inline struct f2fs_inode *F2FS_INODE(const struct f2fs_cached_block *entry) 2340 { 2341 return &CACHED_NODE(entry)->i; 2342 } 2343 2344 static inline __le32 *F2FS_INODE_NIDS(struct f2fs_sb_info *sbi, 2345 const struct f2fs_cached_block *entry) 2346 { 2347 return (__le32 *)(cache_address(entry) + 2348 F2FS_BLKSIZE(sbi) - 2349 sizeof(struct node_footer) - 2350 SIZE_OF_I_NID); 2351 } 2352 2353 static inline struct f2fs_nm_info *NM_I(struct f2fs_sb_info *sbi) 2354 { 2355 return (struct f2fs_nm_info *)(sbi->nm_info); 2356 } 2357 2358 static inline struct f2fs_sm_info *SM_I(struct f2fs_sb_info *sbi) 2359 { 2360 return (struct f2fs_sm_info *)(sbi->sm_info); 2361 } 2362 2363 static inline struct sit_info *SIT_I(struct f2fs_sb_info *sbi) 2364 { 2365 return (struct sit_info *)(SM_I(sbi)->sit_info); 2366 } 2367 2368 static inline struct free_segmap_info *FREE_I(struct f2fs_sb_info *sbi) 2369 { 2370 return (struct free_segmap_info *)(SM_I(sbi)->free_info); 2371 } 2372 2373 static inline struct dirty_seglist_info *DIRTY_I(struct f2fs_sb_info *sbi) 2374 { 2375 return (struct dirty_seglist_info *)(SM_I(sbi)->dirty_info); 2376 } 2377 2378 static inline bool f2fs_is_meta_cache(struct f2fs_cached_block *entry) 2379 { 2380 return entry->cache && entry->cache == META_CACHE(entry->cache->sbi); 2381 } 2382 2383 static inline bool f2fs_is_node_cache(struct f2fs_cached_block *entry) 2384 { 2385 return entry->cache && entry->cache == NODE_CACHE(entry->cache->sbi); 2386 } 2387 2388 static inline bool f2fs_is_compress_cache(struct f2fs_cached_block *entry) 2389 { 2390 return entry->cache && entry->cache == COMPRESS_CACHE(entry->cache->sbi); 2391 } 2392 2393 static inline struct f2fs_bio *F2FS_BIO(struct bio *bio) 2394 { 2395 return container_of(bio, struct f2fs_bio, bio); 2396 } 2397 2398 static inline bool f2fs_is_cache_bio(struct bio *bio) 2399 { 2400 return F2FS_BIO(bio)->entry != NULL; 2401 } 2402 2403 static inline bool is_sbi_flag_set(struct f2fs_sb_info *sbi, unsigned int type) 2404 { 2405 return test_bit(type, &sbi->s_flag); 2406 } 2407 2408 static inline void __set_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type) 2409 { 2410 set_bit(type, &sbi->s_flag); 2411 } 2412 2413 #define set_sbi_flag(sbi, type) \ 2414 do { \ 2415 __set_sbi_flag(sbi, type); \ 2416 if ((type) == SBI_NEED_FSCK) \ 2417 f2fs_fault_report(sbi->sb, REPORT_FAULT_NEED_FSCK, __func__, __LINE__); \ 2418 } while (0) 2419 2420 static inline void clear_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type) 2421 { 2422 clear_bit(type, &sbi->s_flag); 2423 } 2424 2425 static inline unsigned long long cur_cp_version(struct f2fs_checkpoint *cp) 2426 { 2427 return le64_to_cpu(cp->checkpoint_ver); 2428 } 2429 2430 static inline unsigned long f2fs_qf_ino(struct super_block *sb, int type) 2431 { 2432 if (type < F2FS_MAX_QUOTAS) 2433 return le32_to_cpu(F2FS_SB(sb)->raw_super->qf_ino[type]); 2434 return 0; 2435 } 2436 2437 static inline __u64 cur_cp_crc(struct f2fs_checkpoint *cp) 2438 { 2439 size_t crc_offset = le32_to_cpu(cp->checksum_offset); 2440 return le32_to_cpu(*((__le32 *)((unsigned char *)cp + crc_offset))); 2441 } 2442 2443 static inline bool __is_set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 2444 { 2445 unsigned int ckpt_flags = le32_to_cpu(cp->ckpt_flags); 2446 2447 return ckpt_flags & f; 2448 } 2449 2450 static inline bool is_set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 2451 { 2452 return __is_set_ckpt_flags(F2FS_CKPT(sbi), f); 2453 } 2454 2455 static inline void __set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 2456 { 2457 unsigned int ckpt_flags; 2458 2459 ckpt_flags = le32_to_cpu(cp->ckpt_flags); 2460 ckpt_flags |= f; 2461 cp->ckpt_flags = cpu_to_le32(ckpt_flags); 2462 } 2463 2464 static inline void set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 2465 { 2466 unsigned long flags; 2467 2468 spin_lock_irqsave(&sbi->cp_lock, flags); 2469 __set_ckpt_flags(F2FS_CKPT(sbi), f); 2470 spin_unlock_irqrestore(&sbi->cp_lock, flags); 2471 } 2472 2473 static inline void __clear_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 2474 { 2475 unsigned int ckpt_flags; 2476 2477 ckpt_flags = le32_to_cpu(cp->ckpt_flags); 2478 ckpt_flags &= (~f); 2479 cp->ckpt_flags = cpu_to_le32(ckpt_flags); 2480 } 2481 2482 static inline void clear_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 2483 { 2484 unsigned long flags; 2485 2486 spin_lock_irqsave(&sbi->cp_lock, flags); 2487 __clear_ckpt_flags(F2FS_CKPT(sbi), f); 2488 spin_unlock_irqrestore(&sbi->cp_lock, flags); 2489 } 2490 2491 #define init_f2fs_rwsem(sem) __init_f2fs_rwsem(sem, NULL, LOCK_NAME_NONE) 2492 #define init_f2fs_rwsem_trace __init_f2fs_rwsem 2493 2494 #define __init_f2fs_rwsem(sem, sbi, name) \ 2495 do { \ 2496 static struct lock_class_key __key; \ 2497 \ 2498 do_init_f2fs_rwsem((sem), #sem, &__key, sbi, name); \ 2499 } while (0) 2500 2501 static inline void do_init_f2fs_rwsem(struct f2fs_rwsem *sem, 2502 const char *sem_name, struct lock_class_key *key, 2503 struct f2fs_sb_info *sbi, enum f2fs_lock_name name) 2504 { 2505 sem->sbi = sbi; 2506 sem->name = name; 2507 __init_rwsem(&sem->internal_rwsem, sem_name, key); 2508 #ifdef CONFIG_F2FS_UNFAIR_RWSEM 2509 init_waitqueue_head(&sem->read_waiters); 2510 #endif 2511 } 2512 2513 static inline int f2fs_rwsem_is_locked(struct f2fs_rwsem *sem) 2514 { 2515 return rwsem_is_locked(&sem->internal_rwsem); 2516 } 2517 2518 static inline int f2fs_rwsem_is_contended(struct f2fs_rwsem *sem) 2519 { 2520 return rwsem_is_contended(&sem->internal_rwsem); 2521 } 2522 2523 static inline void f2fs_down_read(struct f2fs_rwsem *sem) 2524 { 2525 #ifdef CONFIG_F2FS_UNFAIR_RWSEM 2526 wait_event(sem->read_waiters, down_read_trylock(&sem->internal_rwsem)); 2527 #else 2528 down_read(&sem->internal_rwsem); 2529 #endif 2530 } 2531 2532 static inline int f2fs_down_read_trylock(struct f2fs_rwsem *sem) 2533 { 2534 return down_read_trylock(&sem->internal_rwsem); 2535 } 2536 2537 static inline void f2fs_up_read(struct f2fs_rwsem *sem) 2538 { 2539 up_read(&sem->internal_rwsem); 2540 } 2541 2542 static inline void f2fs_down_write(struct f2fs_rwsem *sem) 2543 { 2544 down_write(&sem->internal_rwsem); 2545 } 2546 2547 #ifdef CONFIG_DEBUG_LOCK_ALLOC 2548 static inline void f2fs_down_read_nested(struct f2fs_rwsem *sem, int subclass) 2549 { 2550 down_read_nested(&sem->internal_rwsem, subclass); 2551 } 2552 2553 static inline void f2fs_down_write_nested(struct f2fs_rwsem *sem, int subclass) 2554 { 2555 down_write_nested(&sem->internal_rwsem, subclass); 2556 } 2557 #else 2558 #define f2fs_down_read_nested(sem, subclass) f2fs_down_read(sem) 2559 #define f2fs_down_write_nested(sem, subclass) f2fs_down_write(sem) 2560 #endif 2561 2562 static inline int f2fs_down_write_trylock(struct f2fs_rwsem *sem) 2563 { 2564 return down_write_trylock(&sem->internal_rwsem); 2565 } 2566 2567 static inline void f2fs_up_write(struct f2fs_rwsem *sem) 2568 { 2569 up_write(&sem->internal_rwsem); 2570 #ifdef CONFIG_F2FS_UNFAIR_RWSEM 2571 wake_up_all(&sem->read_waiters); 2572 #endif 2573 } 2574 2575 void f2fs_down_read_trace(struct f2fs_rwsem *sem, struct f2fs_lock_context *lc); 2576 int f2fs_down_read_trylock_trace(struct f2fs_rwsem *sem, 2577 struct f2fs_lock_context *lc); 2578 void f2fs_up_read_trace(struct f2fs_rwsem *sem, struct f2fs_lock_context *lc); 2579 void f2fs_down_write_trace(struct f2fs_rwsem *sem, 2580 struct f2fs_lock_context *lc); 2581 int f2fs_down_write_trylock_trace(struct f2fs_rwsem *sem, 2582 struct f2fs_lock_context *lc); 2583 void f2fs_up_write_trace(struct f2fs_rwsem *sem, struct f2fs_lock_context *lc); 2584 2585 static inline void disable_nat_bits(struct f2fs_sb_info *sbi, bool lock) 2586 { 2587 unsigned long flags; 2588 unsigned char *nat_bits; 2589 2590 /* 2591 * In order to re-enable nat_bits we need to call fsck.f2fs by 2592 * set_sbi_flag(sbi, SBI_NEED_FSCK). But it may give huge cost, 2593 * so let's rely on regular fsck or unclean shutdown. 2594 */ 2595 2596 if (lock) 2597 spin_lock_irqsave(&sbi->cp_lock, flags); 2598 __clear_ckpt_flags(F2FS_CKPT(sbi), CP_NAT_BITS_FLAG); 2599 nat_bits = NM_I(sbi)->nat_bits; 2600 NM_I(sbi)->nat_bits = NULL; 2601 if (lock) 2602 spin_unlock_irqrestore(&sbi->cp_lock, flags); 2603 2604 kvfree(nat_bits); 2605 } 2606 2607 static inline bool enabled_nat_bits(struct f2fs_sb_info *sbi, 2608 struct cp_control *cpc) 2609 { 2610 bool set = is_set_ckpt_flags(sbi, CP_NAT_BITS_FLAG); 2611 2612 return (cpc) ? (cpc->reason & CP_UMOUNT) && set : set; 2613 } 2614 2615 static inline int __get_cp_reason(struct f2fs_sb_info *sbi) 2616 { 2617 int reason = CP_SYNC; 2618 2619 if (test_opt(sbi, FASTBOOT)) 2620 reason = CP_FASTBOOT; 2621 if (is_sbi_flag_set(sbi, SBI_IS_CLOSE)) 2622 reason = CP_UMOUNT; 2623 return reason; 2624 } 2625 2626 static inline bool __remain_node_summaries(int reason) 2627 { 2628 return (reason & (CP_UMOUNT | CP_FASTBOOT)); 2629 } 2630 2631 static inline bool __exist_node_summaries(struct f2fs_sb_info *sbi) 2632 { 2633 return (is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG) || 2634 is_set_ckpt_flags(sbi, CP_FASTBOOT_FLAG)); 2635 } 2636 2637 /* 2638 * Check whether the inode has blocks or not 2639 */ 2640 static inline int F2FS_HAS_BLOCKS(struct inode *inode) 2641 { 2642 block_t xattr_block = F2FS_I(inode)->i_xattr_nid ? 1 : 0; 2643 2644 return (inode->i_blocks >> 2645 F2FS_LOG_SECTORS_PER_BLOCK(F2FS_I_SB(inode))) > xattr_block; 2646 } 2647 2648 static inline bool f2fs_has_xattr_block(unsigned int ofs) 2649 { 2650 return ofs == XATTR_NODE_OFFSET; 2651 } 2652 2653 static inline bool __allow_reserved_root(struct f2fs_sb_info *sbi, 2654 struct inode *inode, bool cap) 2655 { 2656 if (!inode) 2657 return true; 2658 if (IS_NOQUOTA(inode)) 2659 return true; 2660 if (uid_eq(F2FS_OPTION(sbi).s_resuid, current_fsuid())) 2661 return true; 2662 if (!gid_eq(F2FS_OPTION(sbi).s_resgid, GLOBAL_ROOT_GID) && 2663 in_group_p(F2FS_OPTION(sbi).s_resgid)) 2664 return true; 2665 if (cap && capable(CAP_SYS_RESOURCE)) 2666 return true; 2667 return false; 2668 } 2669 2670 static inline unsigned int get_available_block_count(struct f2fs_sb_info *sbi, 2671 struct inode *inode, bool cap) 2672 { 2673 block_t avail_user_block_count; 2674 2675 avail_user_block_count = sbi->user_block_count - 2676 sbi->current_reserved_blocks - 2677 sbi->alias_reserved_blocks; 2678 2679 if (test_opt(sbi, RESERVE_ROOT) && !__allow_reserved_root(sbi, inode, cap)) 2680 avail_user_block_count -= F2FS_OPTION(sbi).root_reserved_blocks; 2681 2682 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) { 2683 if (avail_user_block_count > sbi->unusable_block_count) 2684 avail_user_block_count -= sbi->unusable_block_count; 2685 else 2686 avail_user_block_count = 0; 2687 } 2688 2689 return avail_user_block_count; 2690 } 2691 2692 static inline void f2fs_i_blocks_write(struct inode *, block_t, bool, bool); 2693 static inline int inc_valid_block_count(struct f2fs_sb_info *sbi, 2694 struct inode *inode, blkcnt_t *count, 2695 bool partial, bool alias_reserved) 2696 { 2697 long long diff = 0, release = 0; 2698 block_t avail_user_block_count; 2699 int ret; 2700 2701 ret = dquot_reserve_block(inode, *count); 2702 if (ret) 2703 return ret; 2704 2705 if (time_to_inject(sbi, FAULT_BLOCK)) { 2706 release = *count; 2707 goto release_quota; 2708 } 2709 2710 /* 2711 * let's increase this in prior to actual block count change in order 2712 * for f2fs_sync_file to avoid data races when deciding checkpoint. 2713 */ 2714 percpu_counter_add(&sbi->alloc_valid_block_count, (*count)); 2715 2716 spin_lock(&sbi->stat_lock); 2717 2718 if (alias_reserved) 2719 sbi->alias_reserved_blocks -= *count; 2720 2721 avail_user_block_count = get_available_block_count(sbi, inode, true); 2722 diff = (long long)sbi->total_valid_block_count + *count - 2723 avail_user_block_count; 2724 if (unlikely(diff > 0)) { 2725 if (alias_reserved) 2726 sbi->alias_reserved_blocks += *count; 2727 2728 if (!partial) { 2729 spin_unlock(&sbi->stat_lock); 2730 release = *count; 2731 goto enospc; 2732 } 2733 if (diff > *count) 2734 diff = *count; 2735 *count -= diff; 2736 release = diff; 2737 if (!*count) { 2738 spin_unlock(&sbi->stat_lock); 2739 goto enospc; 2740 } 2741 } 2742 sbi->total_valid_block_count += (block_t)(*count); 2743 2744 spin_unlock(&sbi->stat_lock); 2745 2746 if (unlikely(release)) { 2747 percpu_counter_sub(&sbi->alloc_valid_block_count, release); 2748 dquot_release_reservation_block(inode, release); 2749 } 2750 f2fs_i_blocks_write(inode, *count, true, true); 2751 return 0; 2752 2753 enospc: 2754 percpu_counter_sub(&sbi->alloc_valid_block_count, release); 2755 release_quota: 2756 dquot_release_reservation_block(inode, release); 2757 return -ENOSPC; 2758 } 2759 2760 #define F2FS_FOLIO_PRIVATE_GET_FUNC(name, flagname) \ 2761 static inline bool folio_test_f2fs_##name(const struct folio *folio) \ 2762 { \ 2763 unsigned long priv = (unsigned long)folio->private; \ 2764 unsigned long v = (1UL << F2FS_FOLIO_PRIVATE_NOT_POINTER) | \ 2765 (1UL << F2FS_FOLIO_PRIVATE_##flagname); \ 2766 return (priv & v) == v; \ 2767 } 2768 2769 #define F2FS_FOLIO_PRIVATE_SET_FUNC(name, flagname) \ 2770 static inline void folio_set_f2fs_##name(struct folio *folio) \ 2771 { \ 2772 unsigned long v = (1UL << F2FS_FOLIO_PRIVATE_NOT_POINTER) | \ 2773 (1UL << F2FS_FOLIO_PRIVATE_##flagname); \ 2774 if (!folio->private) \ 2775 folio_attach_private(folio, (void *)v); \ 2776 else { \ 2777 v |= (unsigned long)folio->private; \ 2778 folio->private = (void *)v; \ 2779 } \ 2780 } 2781 2782 #define F2FS_FOLIO_PRIVATE_CLEAR_FUNC(name, flagname) \ 2783 static inline void folio_clear_f2fs_##name(struct folio *folio) \ 2784 { \ 2785 unsigned long v = (unsigned long)folio->private; \ 2786 \ 2787 v &= ~(1UL << F2FS_FOLIO_PRIVATE_##flagname); \ 2788 if (v == (1UL << F2FS_FOLIO_PRIVATE_NOT_POINTER)) \ 2789 folio_detach_private(folio); \ 2790 else \ 2791 folio->private = (void *)v; \ 2792 } 2793 2794 F2FS_FOLIO_PRIVATE_GET_FUNC(nonpointer, NOT_POINTER); 2795 F2FS_FOLIO_PRIVATE_GET_FUNC(gcing, ONGOING_MIGRATION); 2796 F2FS_FOLIO_PRIVATE_GET_FUNC(atomic, ATOMIC_WRITE); 2797 2798 F2FS_FOLIO_PRIVATE_SET_FUNC(reference, REF_RESOURCE); 2799 F2FS_FOLIO_PRIVATE_SET_FUNC(gcing, ONGOING_MIGRATION); 2800 F2FS_FOLIO_PRIVATE_SET_FUNC(atomic, ATOMIC_WRITE); 2801 2802 F2FS_FOLIO_PRIVATE_CLEAR_FUNC(reference, REF_RESOURCE); 2803 F2FS_FOLIO_PRIVATE_CLEAR_FUNC(gcing, ONGOING_MIGRATION); 2804 F2FS_FOLIO_PRIVATE_CLEAR_FUNC(atomic, ATOMIC_WRITE); 2805 2806 static inline unsigned long folio_get_f2fs_data(struct folio *folio) 2807 { 2808 unsigned long data = (unsigned long)folio->private; 2809 2810 if (!test_bit(F2FS_FOLIO_PRIVATE_NOT_POINTER, &data)) 2811 return 0; 2812 return data >> F2FS_FOLIO_PRIVATE_MAX; 2813 } 2814 2815 static inline void folio_set_f2fs_data(struct folio *folio, unsigned long data) 2816 { 2817 data = (1UL << F2FS_FOLIO_PRIVATE_NOT_POINTER) | 2818 (data << F2FS_FOLIO_PRIVATE_MAX); 2819 2820 if (!folio_test_private(folio)) 2821 folio_attach_private(folio, (void *)data); 2822 else 2823 folio->private = (void *)((unsigned long)folio->private | data); 2824 } 2825 2826 static inline void dec_valid_block_count(struct f2fs_sb_info *sbi, 2827 struct inode *inode, 2828 block_t count) 2829 { 2830 blkcnt_t sectors = count << F2FS_LOG_SECTORS_PER_BLOCK(sbi); 2831 2832 spin_lock(&sbi->stat_lock); 2833 if (unlikely(sbi->total_valid_block_count < count)) { 2834 f2fs_warn(sbi, "Inconsistent total_valid_block_count:%u, ino:%llu, count:%u", 2835 sbi->total_valid_block_count, inode->i_ino, count); 2836 sbi->total_valid_block_count = 0; 2837 set_sbi_flag(sbi, SBI_NEED_FSCK); 2838 } else { 2839 sbi->total_valid_block_count -= count; 2840 } 2841 if (sbi->reserved_blocks && 2842 sbi->current_reserved_blocks < sbi->reserved_blocks) 2843 sbi->current_reserved_blocks = min(sbi->reserved_blocks, 2844 sbi->current_reserved_blocks + count); 2845 spin_unlock(&sbi->stat_lock); 2846 if (unlikely(inode->i_blocks < sectors)) { 2847 f2fs_warn(sbi, "Inconsistent i_blocks, ino:%llu, iblocks:%llu, sectors:%llu", 2848 inode->i_ino, 2849 (unsigned long long)inode->i_blocks, 2850 (unsigned long long)sectors); 2851 set_sbi_flag(sbi, SBI_NEED_FSCK); 2852 return; 2853 } 2854 f2fs_i_blocks_write(inode, count, false, true); 2855 } 2856 2857 static inline void inc_cache_count(struct f2fs_sb_info *sbi, int count_type) 2858 { 2859 atomic_inc(&sbi->nr_caches[count_type]); 2860 2861 if (count_type == F2FS_DIRTY_DENTS || 2862 count_type == F2FS_DIRTY_NODES || 2863 count_type == F2FS_DIRTY_META || 2864 count_type == F2FS_DIRTY_QDATA || 2865 count_type == F2FS_DIRTY_IMETA) 2866 set_sbi_flag(sbi, SBI_IS_DIRTY); 2867 } 2868 2869 static inline void inode_inc_dirty_pages(struct inode *inode) 2870 { 2871 atomic_inc(&F2FS_I(inode)->dirty_pages); 2872 inc_cache_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ? 2873 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA); 2874 if (IS_NOQUOTA(inode)) 2875 inc_cache_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA); 2876 } 2877 2878 static inline void dec_cache_count(struct f2fs_sb_info *sbi, int count_type) 2879 { 2880 atomic_dec(&sbi->nr_caches[count_type]); 2881 } 2882 2883 static inline void inode_dec_dirty_pages(struct inode *inode) 2884 { 2885 if (!S_ISDIR(inode->i_mode) && !S_ISREG(inode->i_mode) && 2886 !S_ISLNK(inode->i_mode)) 2887 return; 2888 2889 atomic_dec(&F2FS_I(inode)->dirty_pages); 2890 dec_cache_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ? 2891 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA); 2892 if (IS_NOQUOTA(inode)) 2893 dec_cache_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA); 2894 } 2895 2896 static inline void inc_atomic_write_cnt(struct inode *inode) 2897 { 2898 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 2899 struct f2fs_inode_info *fi = F2FS_I(inode); 2900 u64 current_write; 2901 2902 fi->atomic_write_cnt++; 2903 atomic64_inc(&sbi->current_atomic_write); 2904 current_write = atomic64_read(&sbi->current_atomic_write); 2905 if (current_write > sbi->peak_atomic_write) 2906 sbi->peak_atomic_write = current_write; 2907 } 2908 2909 static inline void release_atomic_write_cnt(struct inode *inode) 2910 { 2911 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 2912 struct f2fs_inode_info *fi = F2FS_I(inode); 2913 2914 atomic64_sub(fi->atomic_write_cnt, &sbi->current_atomic_write); 2915 fi->atomic_write_cnt = 0; 2916 } 2917 2918 static inline s64 get_nr_caches(struct f2fs_sb_info *sbi, int count_type) 2919 { 2920 return atomic_read(&sbi->nr_caches[count_type]); 2921 } 2922 2923 static inline int get_dirty_pages(struct inode *inode) 2924 { 2925 return atomic_read(&F2FS_I(inode)->dirty_pages); 2926 } 2927 2928 static inline int get_blocktype_secs(struct f2fs_sb_info *sbi, int block_type) 2929 { 2930 return div_u64(get_nr_caches(sbi, block_type) + BLKS_PER_SEC(sbi) - 1, 2931 BLKS_PER_SEC(sbi)); 2932 } 2933 2934 static inline block_t valid_user_blocks(struct f2fs_sb_info *sbi) 2935 { 2936 return sbi->total_valid_block_count; 2937 } 2938 2939 static inline block_t discard_blocks(struct f2fs_sb_info *sbi) 2940 { 2941 return sbi->discard_blks; 2942 } 2943 2944 static inline unsigned long __bitmap_size(struct f2fs_sb_info *sbi, int flag) 2945 { 2946 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi); 2947 2948 /* return NAT or SIT bitmap */ 2949 if (flag == NAT_BITMAP) 2950 return le32_to_cpu(ckpt->nat_ver_bitmap_bytesize); 2951 else if (flag == SIT_BITMAP) 2952 return le32_to_cpu(ckpt->sit_ver_bitmap_bytesize); 2953 2954 return 0; 2955 } 2956 2957 static inline block_t __cp_payload(struct f2fs_sb_info *sbi) 2958 { 2959 return le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_payload); 2960 } 2961 2962 static inline void *__bitmap_ptr(struct f2fs_sb_info *sbi, int flag) 2963 { 2964 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi); 2965 void *tmp_ptr = &ckpt->sit_nat_version_bitmap; 2966 int offset; 2967 2968 if (is_set_ckpt_flags(sbi, CP_LARGE_NAT_BITMAP_FLAG)) { 2969 offset = (flag == SIT_BITMAP) ? 2970 le32_to_cpu(ckpt->nat_ver_bitmap_bytesize) : 0; 2971 /* 2972 * if large_nat_bitmap feature is enabled, leave checksum 2973 * protection for all nat/sit bitmaps. 2974 */ 2975 return tmp_ptr + offset + sizeof(__le32); 2976 } 2977 2978 if (__cp_payload(sbi) > 0) { 2979 if (flag == NAT_BITMAP) 2980 return tmp_ptr; 2981 else 2982 return (unsigned char *)ckpt + F2FS_BLKSIZE(sbi); 2983 } else { 2984 offset = (flag == NAT_BITMAP) ? 2985 le32_to_cpu(ckpt->sit_ver_bitmap_bytesize) : 0; 2986 return tmp_ptr + offset; 2987 } 2988 } 2989 2990 static inline block_t __start_cp_addr(struct f2fs_sb_info *sbi) 2991 { 2992 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr); 2993 2994 if (sbi->cur_cp_pack == 2) 2995 start_addr += BLKS_PER_SEG(sbi); 2996 return start_addr; 2997 } 2998 2999 static inline block_t __start_cp_next_addr(struct f2fs_sb_info *sbi) 3000 { 3001 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr); 3002 3003 if (sbi->cur_cp_pack == 1) 3004 start_addr += BLKS_PER_SEG(sbi); 3005 return start_addr; 3006 } 3007 3008 static inline void __set_cp_next_pack(struct f2fs_sb_info *sbi) 3009 { 3010 sbi->cur_cp_pack = (sbi->cur_cp_pack == 1) ? 2 : 1; 3011 } 3012 3013 static inline block_t __start_sum_addr(struct f2fs_sb_info *sbi) 3014 { 3015 return le32_to_cpu(F2FS_CKPT(sbi)->cp_pack_start_sum); 3016 } 3017 3018 static inline bool __has_cursum_space(struct f2fs_sb_info *sbi, 3019 struct f2fs_journal *journal, unsigned int size, int type) 3020 { 3021 if (type == NAT_JOURNAL) 3022 return size <= MAX_NAT_JENTRIES(sbi, journal); 3023 return size <= MAX_SIT_JENTRIES(sbi, journal); 3024 } 3025 3026 extern void f2fs_mark_inode_dirty_sync(struct inode *inode, bool sync); 3027 static inline int inc_valid_node_count(struct f2fs_sb_info *sbi, 3028 struct inode *inode, bool is_inode) 3029 { 3030 block_t valid_block_count; 3031 unsigned int valid_node_count, avail_user_node_count; 3032 unsigned int avail_user_block_count; 3033 int err; 3034 3035 if (is_inode) { 3036 if (inode) { 3037 err = dquot_alloc_inode(inode); 3038 if (err) 3039 return err; 3040 } 3041 } else { 3042 err = dquot_reserve_block(inode, 1); 3043 if (err) 3044 return err; 3045 } 3046 3047 if (time_to_inject(sbi, FAULT_BLOCK)) 3048 goto enospc; 3049 3050 spin_lock(&sbi->stat_lock); 3051 3052 valid_block_count = sbi->total_valid_block_count + 1; 3053 avail_user_block_count = get_available_block_count(sbi, inode, 3054 test_opt(sbi, RESERVE_NODE)); 3055 3056 if (unlikely(valid_block_count > avail_user_block_count)) { 3057 spin_unlock(&sbi->stat_lock); 3058 goto enospc; 3059 } 3060 3061 avail_user_node_count = sbi->total_node_count - F2FS_RESERVED_NODE_NUM; 3062 if (test_opt(sbi, RESERVE_NODE) && 3063 !__allow_reserved_root(sbi, inode, true)) 3064 avail_user_node_count -= F2FS_OPTION(sbi).root_reserved_nodes; 3065 valid_node_count = sbi->total_valid_node_count + 1; 3066 if (unlikely(valid_node_count > avail_user_node_count)) { 3067 spin_unlock(&sbi->stat_lock); 3068 goto enospc; 3069 } 3070 3071 sbi->total_valid_node_count++; 3072 sbi->total_valid_block_count++; 3073 spin_unlock(&sbi->stat_lock); 3074 3075 if (inode) { 3076 if (is_inode) 3077 f2fs_mark_inode_dirty_sync(inode, true); 3078 else 3079 f2fs_i_blocks_write(inode, 1, true, true); 3080 } 3081 3082 percpu_counter_inc(&sbi->alloc_valid_block_count); 3083 return 0; 3084 3085 enospc: 3086 if (is_inode) { 3087 if (inode) 3088 dquot_free_inode(inode); 3089 } else { 3090 dquot_release_reservation_block(inode, 1); 3091 } 3092 return -ENOSPC; 3093 } 3094 3095 static inline void dec_valid_node_count(struct f2fs_sb_info *sbi, 3096 struct inode *inode, bool is_inode) 3097 { 3098 spin_lock(&sbi->stat_lock); 3099 3100 if (unlikely(!sbi->total_valid_block_count || 3101 !sbi->total_valid_node_count)) { 3102 f2fs_warn(sbi, "dec_valid_node_count: inconsistent block counts, total_valid_block:%u, total_valid_node:%u", 3103 sbi->total_valid_block_count, 3104 sbi->total_valid_node_count); 3105 set_sbi_flag(sbi, SBI_NEED_FSCK); 3106 } else { 3107 sbi->total_valid_block_count--; 3108 sbi->total_valid_node_count--; 3109 } 3110 3111 if (sbi->reserved_blocks && 3112 sbi->current_reserved_blocks < sbi->reserved_blocks) 3113 sbi->current_reserved_blocks++; 3114 3115 spin_unlock(&sbi->stat_lock); 3116 3117 if (is_inode) { 3118 dquot_free_inode(inode); 3119 } else { 3120 if (unlikely(inode->i_blocks == 0)) { 3121 f2fs_warn(sbi, "dec_valid_node_count: inconsistent i_blocks, ino:%llu, iblocks:%llu", 3122 inode->i_ino, 3123 (unsigned long long)inode->i_blocks); 3124 set_sbi_flag(sbi, SBI_NEED_FSCK); 3125 return; 3126 } 3127 f2fs_i_blocks_write(inode, 1, false, true); 3128 } 3129 } 3130 3131 static inline unsigned int valid_node_count(struct f2fs_sb_info *sbi) 3132 { 3133 return sbi->total_valid_node_count; 3134 } 3135 3136 static inline void inc_valid_inode_count(struct f2fs_sb_info *sbi) 3137 { 3138 percpu_counter_inc(&sbi->total_valid_inode_count); 3139 } 3140 3141 static inline void dec_valid_inode_count(struct f2fs_sb_info *sbi) 3142 { 3143 percpu_counter_dec(&sbi->total_valid_inode_count); 3144 } 3145 3146 static inline s64 valid_inode_count(struct f2fs_sb_info *sbi) 3147 { 3148 return percpu_counter_sum_positive(&sbi->total_valid_inode_count); 3149 } 3150 3151 static inline struct folio *f2fs_grab_cache_folio(struct address_space *mapping, 3152 pgoff_t index, bool for_write) 3153 { 3154 struct folio *folio; 3155 unsigned int flags; 3156 3157 if (IS_ENABLED(CONFIG_F2FS_FAULT_INJECTION)) { 3158 fgf_t fgf_flags; 3159 3160 if (!for_write) 3161 fgf_flags = FGP_LOCK | FGP_ACCESSED; 3162 else 3163 fgf_flags = FGP_LOCK; 3164 folio = __filemap_get_folio(mapping, index, fgf_flags, 0); 3165 if (!IS_ERR(folio)) 3166 return folio; 3167 3168 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_ALLOC)) 3169 return ERR_PTR(-ENOMEM); 3170 } 3171 3172 if (!for_write) 3173 return filemap_grab_folio(mapping, index); 3174 3175 flags = memalloc_nofs_save(); 3176 folio = __filemap_get_folio(mapping, index, FGP_WRITEBEGIN, 3177 mapping_gfp_mask(mapping)); 3178 memalloc_nofs_restore(flags); 3179 3180 return folio; 3181 } 3182 3183 static inline struct folio *f2fs_filemap_get_folio( 3184 struct address_space *mapping, pgoff_t index, 3185 fgf_t fgp_flags, gfp_t gfp_mask) 3186 { 3187 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_GET)) 3188 return ERR_PTR(-ENOMEM); 3189 3190 return __filemap_get_folio(mapping, index, fgp_flags, gfp_mask); 3191 } 3192 3193 static inline void f2fs_folio_put(struct folio *folio, bool unlock) 3194 { 3195 if (IS_ERR_OR_NULL(folio)) 3196 return; 3197 3198 if (unlock) { 3199 f2fs_bug_on(F2FS_F_SB(folio), !folio_test_locked(folio)); 3200 folio_unlock(folio); 3201 } 3202 folio_put(folio); 3203 } 3204 3205 static inline void f2fs_put_page(struct page *page, bool unlock) 3206 { 3207 if (!page) 3208 return; 3209 f2fs_folio_put(page_folio(page), unlock); 3210 } 3211 3212 #define F2FS_DENTRY_TAG_CACHE 1UL 3213 #define F2FS_DENTRY_TAG_MASK 1UL 3214 3215 static inline void *f2fs_cache_make_dentry_block(struct f2fs_cached_block *entry) 3216 { 3217 if (IS_ERR_OR_NULL(entry)) 3218 return entry; 3219 return (void *)((unsigned long)entry | F2FS_DENTRY_TAG_CACHE); 3220 } 3221 3222 static inline bool f2fs_dentry_is_cache(void *dentry_block) 3223 { 3224 return ((unsigned long)dentry_block & F2FS_DENTRY_TAG_MASK) == 3225 F2FS_DENTRY_TAG_CACHE; 3226 } 3227 3228 static inline struct f2fs_cached_block *f2fs_dentry_cache(void *dentry_block) 3229 { 3230 return (struct f2fs_cached_block *) 3231 ((unsigned long)dentry_block & ~F2FS_DENTRY_TAG_MASK); 3232 } 3233 3234 static inline struct folio *f2fs_dentry_folio(void *dentry_block) 3235 { 3236 return (struct folio *)dentry_block; 3237 } 3238 3239 static inline void f2fs_put_dentry_block(void *dentry_block, bool unlock) 3240 { 3241 if (IS_ERR_OR_NULL(dentry_block)) 3242 return; 3243 if (f2fs_dentry_is_cache(dentry_block)) 3244 f2fs_put_cache(f2fs_dentry_cache(dentry_block), unlock); 3245 else 3246 f2fs_folio_put(f2fs_dentry_folio(dentry_block), unlock); 3247 } 3248 3249 static inline void f2fs_put_dnode(struct dnode_of_data *dn) 3250 { 3251 if (dn->node_entry) 3252 f2fs_put_cache(dn->node_entry, true); 3253 if (dn->inode_entry && dn->node_entry != dn->inode_entry) 3254 f2fs_put_cache(dn->inode_entry, false); 3255 dn->node_entry = NULL; 3256 dn->inode_entry = NULL; 3257 } 3258 3259 static inline struct kmem_cache *f2fs_kmem_cache_create(const char *name, 3260 size_t size) 3261 { 3262 return kmem_cache_create(name, size, 0, SLAB_RECLAIM_ACCOUNT, NULL); 3263 } 3264 3265 static inline void *f2fs_kmem_cache_alloc_nofail(struct kmem_cache *cachep, 3266 gfp_t flags) 3267 { 3268 void *entry; 3269 3270 entry = kmem_cache_alloc(cachep, flags); 3271 if (!entry) 3272 entry = kmem_cache_alloc(cachep, flags | __GFP_NOFAIL); 3273 return entry; 3274 } 3275 3276 static inline void *f2fs_kmem_cache_alloc(struct kmem_cache *cachep, 3277 gfp_t flags, bool nofail, struct f2fs_sb_info *sbi) 3278 { 3279 if (nofail) 3280 return f2fs_kmem_cache_alloc_nofail(cachep, flags); 3281 3282 if (time_to_inject(sbi, FAULT_SLAB_ALLOC)) 3283 return NULL; 3284 3285 return kmem_cache_alloc(cachep, flags); 3286 } 3287 3288 static inline bool is_inflight_io(struct f2fs_sb_info *sbi, int type) 3289 { 3290 if (get_nr_caches(sbi, F2FS_RD_DATA) || 3291 get_nr_caches(sbi, F2FS_RD_NODE) || 3292 get_nr_caches(sbi, F2FS_RD_META) || 3293 get_nr_caches(sbi, F2FS_WB_DATA) || 3294 get_nr_caches(sbi, F2FS_WB_CP_DATA) || 3295 get_nr_caches(sbi, F2FS_DIO_READ) || 3296 get_nr_caches(sbi, F2FS_DIO_WRITE)) 3297 return true; 3298 3299 if (type != DISCARD_TIME && SM_I(sbi) && SM_I(sbi)->dcc_info && 3300 atomic_read(&SM_I(sbi)->dcc_info->queued_discard)) 3301 return true; 3302 3303 if (SM_I(sbi) && SM_I(sbi)->fcc_info && 3304 atomic_read(&SM_I(sbi)->fcc_info->queued_flush)) 3305 return true; 3306 return false; 3307 } 3308 3309 static inline bool is_inflight_read_io(struct f2fs_sb_info *sbi) 3310 { 3311 return get_nr_caches(sbi, F2FS_RD_DATA) || 3312 get_nr_caches(sbi, F2FS_DIO_READ); 3313 } 3314 3315 static inline bool is_idle(struct f2fs_sb_info *sbi, int type) 3316 { 3317 bool zoned_gc = (type == GC_TIME && 3318 F2FS_HAS_FEATURE(sbi, F2FS_FEATURE_BLKZONED)); 3319 3320 if (sbi->gc_mode == GC_URGENT_HIGH) 3321 return true; 3322 3323 if (sbi->bggc_io_aware == AWARE_READ_IO && is_inflight_read_io(sbi)) 3324 return false; 3325 if (sbi->bggc_io_aware == AWARE_ALL_IO && is_inflight_io(sbi, type)) 3326 return false; 3327 3328 if (sbi->gc_mode == GC_URGENT_MID) 3329 return true; 3330 3331 if (sbi->gc_mode == GC_URGENT_LOW && 3332 (type == DISCARD_TIME || type == GC_TIME)) 3333 return true; 3334 3335 if (zoned_gc) 3336 return true; 3337 3338 return f2fs_time_over(sbi, type); 3339 } 3340 3341 static inline void f2fs_radix_tree_insert(struct radix_tree_root *root, 3342 unsigned long index, void *item) 3343 { 3344 while (radix_tree_insert(root, index, item)) 3345 cond_resched(); 3346 } 3347 3348 static inline bool IS_INODE(struct f2fs_sb_info *sbi, const struct f2fs_cached_block *entry) 3349 { 3350 struct node_footer *footer = F2FS_NODE_FOOTER(sbi, entry); 3351 3352 return footer->nid == footer->ino; 3353 } 3354 3355 static inline int offset_in_addr(struct f2fs_inode *i) 3356 { 3357 return (i->i_inline & F2FS_EXTRA_ATTR) ? 3358 (le16_to_cpu(i->i_extra_isize) / sizeof(__le32)) : 0; 3359 } 3360 3361 static inline __le32 *blkaddr_in_node(struct f2fs_sb_info *sbi, 3362 const struct f2fs_cached_block *entry) 3363 { 3364 struct f2fs_node *node = F2FS_NODE(entry); 3365 3366 return IS_INODE(sbi, entry) ? node->i.i_addr : node->dn.addr; 3367 } 3368 3369 static inline int f2fs_has_extra_attr(struct inode *inode); 3370 3371 static inline unsigned int get_dnode_base(struct inode *inode, 3372 const struct f2fs_cached_block *entry) 3373 { 3374 if (!IS_INODE(inode ? F2FS_I_SB(inode) : entry->cache->sbi, entry)) 3375 return 0; 3376 3377 return inode ? get_extra_isize(inode) : 3378 offset_in_addr(&CACHED_NODE(entry)->i); 3379 } 3380 3381 static inline __le32 *get_dnode_addr(struct inode *inode, 3382 const struct f2fs_cached_block *entry) 3383 { 3384 struct f2fs_sb_info *sbi = inode ? F2FS_I_SB(inode) : entry->cache->sbi; 3385 3386 return blkaddr_in_node(sbi, entry) + 3387 get_dnode_base(inode, entry); 3388 } 3389 3390 static inline block_t data_blkaddr(struct inode *inode, 3391 const struct f2fs_cached_block *entry, 3392 unsigned int offset) 3393 { 3394 return le32_to_cpu(*(get_dnode_addr(inode, entry) + offset)); 3395 } 3396 3397 static inline block_t f2fs_data_blkaddr(struct dnode_of_data *dn) 3398 { 3399 return data_blkaddr(dn->inode, dn->node_entry, dn->ofs_in_node); 3400 } 3401 3402 static inline int f2fs_test_bit(unsigned int nr, char *addr) 3403 { 3404 int mask; 3405 3406 addr += (nr >> 3); 3407 mask = BIT(7 - (nr & 0x07)); 3408 return mask & *addr; 3409 } 3410 3411 static inline void f2fs_set_bit(unsigned int nr, char *addr) 3412 { 3413 int mask; 3414 3415 addr += (nr >> 3); 3416 mask = BIT(7 - (nr & 0x07)); 3417 *addr |= mask; 3418 } 3419 3420 static inline void f2fs_clear_bit(unsigned int nr, char *addr) 3421 { 3422 int mask; 3423 3424 addr += (nr >> 3); 3425 mask = BIT(7 - (nr & 0x07)); 3426 *addr &= ~mask; 3427 } 3428 3429 static inline int f2fs_test_and_set_bit(unsigned int nr, char *addr) 3430 { 3431 int mask; 3432 int ret; 3433 3434 addr += (nr >> 3); 3435 mask = BIT(7 - (nr & 0x07)); 3436 ret = mask & *addr; 3437 *addr |= mask; 3438 return ret; 3439 } 3440 3441 static inline int f2fs_test_and_clear_bit(unsigned int nr, char *addr) 3442 { 3443 int mask; 3444 int ret; 3445 3446 addr += (nr >> 3); 3447 mask = BIT(7 - (nr & 0x07)); 3448 ret = mask & *addr; 3449 *addr &= ~mask; 3450 return ret; 3451 } 3452 3453 static inline void f2fs_change_bit(unsigned int nr, char *addr) 3454 { 3455 int mask; 3456 3457 addr += (nr >> 3); 3458 mask = BIT(7 - (nr & 0x07)); 3459 *addr ^= mask; 3460 } 3461 3462 /* 3463 * On-disk inode flags (f2fs_inode::i_flags) 3464 */ 3465 #define F2FS_COMPR_FL 0x00000004 /* Compress file */ 3466 #define F2FS_SYNC_FL 0x00000008 /* Synchronous updates */ 3467 #define F2FS_IMMUTABLE_FL 0x00000010 /* Immutable file */ 3468 #define F2FS_APPEND_FL 0x00000020 /* writes to file may only append */ 3469 #define F2FS_NODUMP_FL 0x00000040 /* do not dump file */ 3470 #define F2FS_NOATIME_FL 0x00000080 /* do not update atime */ 3471 #define F2FS_NOCOMP_FL 0x00000400 /* Don't compress */ 3472 #define F2FS_INDEX_FL 0x00001000 /* hash-indexed directory */ 3473 #define F2FS_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */ 3474 #define F2FS_PROJINHERIT_FL 0x20000000 /* Create with parents projid */ 3475 #define F2FS_CASEFOLD_FL 0x40000000 /* Casefolded file */ 3476 #define F2FS_DEVICE_ALIAS_FL 0x80000000 /* File for aliasing a device */ 3477 3478 #define F2FS_QUOTA_DEFAULT_FL (F2FS_NOATIME_FL | F2FS_IMMUTABLE_FL) 3479 3480 /* Flags that should be inherited by new inodes from their parent. */ 3481 #define F2FS_FL_INHERITED (F2FS_SYNC_FL | F2FS_NODUMP_FL | F2FS_NOATIME_FL | \ 3482 F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \ 3483 F2FS_CASEFOLD_FL) 3484 3485 /* Flags that are appropriate for regular files (all but dir-specific ones). */ 3486 #define F2FS_REG_FLMASK (~(F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \ 3487 F2FS_CASEFOLD_FL)) 3488 3489 /* Flags that are appropriate for non-directories/regular files. */ 3490 #define F2FS_OTHER_FLMASK (F2FS_NODUMP_FL | F2FS_NOATIME_FL) 3491 3492 #define IS_DEVICE_ALIASING(inode) (F2FS_I(inode)->i_flags & F2FS_DEVICE_ALIAS_FL) 3493 3494 static inline __u32 f2fs_mask_flags(umode_t mode, __u32 flags) 3495 { 3496 if (S_ISDIR(mode)) 3497 return flags; 3498 else if (S_ISREG(mode)) 3499 return flags & F2FS_REG_FLMASK; 3500 else 3501 return flags & F2FS_OTHER_FLMASK; 3502 } 3503 3504 static inline void __mark_inode_dirty_flag(struct inode *inode, 3505 int flag, bool set) 3506 { 3507 switch (flag) { 3508 case FI_INLINE_XATTR: 3509 case FI_INLINE_DATA: 3510 case FI_INLINE_DENTRY: 3511 case FI_NEW_INODE: 3512 if (set) 3513 return; 3514 fallthrough; 3515 case FI_DATA_EXIST: 3516 case FI_PIN_FILE: 3517 case FI_COMPRESS_RELEASED: 3518 f2fs_mark_inode_dirty_sync(inode, true); 3519 } 3520 } 3521 3522 static inline void set_inode_flag(struct inode *inode, int flag) 3523 { 3524 set_bit(flag, F2FS_I(inode)->flags); 3525 __mark_inode_dirty_flag(inode, flag, true); 3526 } 3527 3528 static inline int is_inode_flag_set(struct inode *inode, int flag) 3529 { 3530 return test_bit(flag, F2FS_I(inode)->flags); 3531 } 3532 3533 static inline void clear_inode_flag(struct inode *inode, int flag) 3534 { 3535 clear_bit(flag, F2FS_I(inode)->flags); 3536 __mark_inode_dirty_flag(inode, flag, false); 3537 } 3538 3539 static inline bool f2fs_verity_in_progress(struct inode *inode) 3540 { 3541 return IS_ENABLED(CONFIG_FS_VERITY) && 3542 is_inode_flag_set(inode, FI_VERITY_IN_PROGRESS); 3543 } 3544 3545 static inline void set_acl_inode(struct inode *inode, umode_t mode) 3546 { 3547 F2FS_I(inode)->i_acl_mode = mode; 3548 set_inode_flag(inode, FI_ACL_MODE); 3549 f2fs_mark_inode_dirty_sync(inode, false); 3550 } 3551 3552 static inline void f2fs_i_links_write(struct inode *inode, bool inc) 3553 { 3554 if (inc) 3555 inc_nlink(inode); 3556 else 3557 drop_nlink(inode); 3558 f2fs_mark_inode_dirty_sync(inode, true); 3559 } 3560 3561 static inline void f2fs_i_blocks_write(struct inode *inode, 3562 block_t diff, bool add, bool claim) 3563 { 3564 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE); 3565 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER); 3566 3567 /* add = 1, claim = 1 should be dquot_reserve_block in pair */ 3568 if (add) { 3569 if (claim) 3570 dquot_claim_block(inode, diff); 3571 else 3572 dquot_alloc_block_nofail(inode, diff); 3573 } else { 3574 dquot_free_block(inode, diff); 3575 } 3576 3577 f2fs_mark_inode_dirty_sync(inode, true); 3578 if (clean || recover) 3579 set_inode_flag(inode, FI_AUTO_RECOVER); 3580 } 3581 3582 static inline bool f2fs_is_atomic_file(struct inode *inode); 3583 3584 static inline void f2fs_i_size_write(struct inode *inode, loff_t i_size) 3585 { 3586 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE); 3587 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER); 3588 3589 if (i_size_read(inode) == i_size) 3590 return; 3591 3592 i_size_write(inode, i_size); 3593 3594 if (f2fs_is_atomic_file(inode)) 3595 return; 3596 3597 f2fs_mark_inode_dirty_sync(inode, true); 3598 if (clean || recover) 3599 set_inode_flag(inode, FI_AUTO_RECOVER); 3600 } 3601 3602 static inline void f2fs_i_depth_write(struct inode *inode, unsigned int depth) 3603 { 3604 F2FS_I(inode)->i_current_depth = depth; 3605 f2fs_mark_inode_dirty_sync(inode, true); 3606 } 3607 3608 static inline void f2fs_i_gc_failures_write(struct inode *inode, 3609 unsigned int count) 3610 { 3611 F2FS_I(inode)->i_gc_failures = count; 3612 f2fs_mark_inode_dirty_sync(inode, true); 3613 } 3614 3615 static inline void f2fs_i_xnid_write(struct inode *inode, nid_t xnid) 3616 { 3617 F2FS_I(inode)->i_xattr_nid = xnid; 3618 f2fs_mark_inode_dirty_sync(inode, true); 3619 } 3620 3621 static inline void f2fs_i_pino_write(struct inode *inode, nid_t pino) 3622 { 3623 F2FS_I(inode)->i_pino = pino; 3624 f2fs_mark_inode_dirty_sync(inode, true); 3625 } 3626 3627 static inline void get_inline_info(struct inode *inode, struct f2fs_inode *ri) 3628 { 3629 struct f2fs_inode_info *fi = F2FS_I(inode); 3630 3631 if (ri->i_inline & F2FS_INLINE_XATTR) 3632 set_bit(FI_INLINE_XATTR, fi->flags); 3633 if (ri->i_inline & F2FS_INLINE_DATA) 3634 set_bit(FI_INLINE_DATA, fi->flags); 3635 if (ri->i_inline & F2FS_INLINE_DENTRY) 3636 set_bit(FI_INLINE_DENTRY, fi->flags); 3637 if (ri->i_inline & F2FS_DATA_EXIST) 3638 set_bit(FI_DATA_EXIST, fi->flags); 3639 if (ri->i_inline & F2FS_EXTRA_ATTR) 3640 set_bit(FI_EXTRA_ATTR, fi->flags); 3641 if (ri->i_inline & F2FS_PIN_FILE) 3642 set_bit(FI_PIN_FILE, fi->flags); 3643 if (ri->i_inline & F2FS_COMPRESS_RELEASED) 3644 set_bit(FI_COMPRESS_RELEASED, fi->flags); 3645 } 3646 3647 static inline void set_raw_inline(struct inode *inode, struct f2fs_inode *ri) 3648 { 3649 ri->i_inline = 0; 3650 3651 if (is_inode_flag_set(inode, FI_INLINE_XATTR)) 3652 ri->i_inline |= F2FS_INLINE_XATTR; 3653 if (is_inode_flag_set(inode, FI_INLINE_DATA)) 3654 ri->i_inline |= F2FS_INLINE_DATA; 3655 if (is_inode_flag_set(inode, FI_INLINE_DENTRY)) 3656 ri->i_inline |= F2FS_INLINE_DENTRY; 3657 if (is_inode_flag_set(inode, FI_DATA_EXIST)) 3658 ri->i_inline |= F2FS_DATA_EXIST; 3659 if (is_inode_flag_set(inode, FI_EXTRA_ATTR)) 3660 ri->i_inline |= F2FS_EXTRA_ATTR; 3661 if (is_inode_flag_set(inode, FI_PIN_FILE)) 3662 ri->i_inline |= F2FS_PIN_FILE; 3663 if (is_inode_flag_set(inode, FI_COMPRESS_RELEASED)) 3664 ri->i_inline |= F2FS_COMPRESS_RELEASED; 3665 } 3666 3667 static inline int f2fs_has_extra_attr(struct inode *inode) 3668 { 3669 return is_inode_flag_set(inode, FI_EXTRA_ATTR); 3670 } 3671 3672 static inline int f2fs_has_inline_xattr(struct inode *inode) 3673 { 3674 return is_inode_flag_set(inode, FI_INLINE_XATTR); 3675 } 3676 3677 static inline int f2fs_compressed_file(struct inode *inode) 3678 { 3679 return S_ISREG(inode->i_mode) && 3680 is_inode_flag_set(inode, FI_COMPRESSED_FILE); 3681 } 3682 3683 static inline bool f2fs_need_compress_data(struct inode *inode) 3684 { 3685 int compress_mode = F2FS_OPTION(F2FS_I_SB(inode)).compress_mode; 3686 3687 if (!f2fs_compressed_file(inode)) 3688 return false; 3689 3690 if (compress_mode == COMPR_MODE_FS) 3691 return true; 3692 else if (compress_mode == COMPR_MODE_USER && 3693 is_inode_flag_set(inode, FI_ENABLE_COMPRESS)) 3694 return true; 3695 3696 return false; 3697 } 3698 3699 static inline unsigned int addrs_per_page(struct inode *inode, 3700 bool is_inode) 3701 { 3702 unsigned int addrs = is_inode ? (cur_addrs_per_inode(inode) - 3703 get_inline_xattr_addrs(inode)) : 3704 DEF_ADDRS_PER_BLOCK(F2FS_I_SB(inode)); 3705 3706 if (f2fs_compressed_file(inode)) 3707 return ALIGN_DOWN(addrs, F2FS_I(inode)->i_cluster_size); 3708 return addrs; 3709 } 3710 3711 static inline unsigned int cur_addrs_per_inode(struct inode *inode) 3712 { 3713 return DEF_ADDRS_PER_INODE(F2FS_I_SB(inode)) - get_extra_isize(inode); 3714 } 3715 3716 static inline void *inline_xattr_addr(struct inode *inode, 3717 const struct f2fs_cached_block *entry) 3718 { 3719 struct f2fs_inode *ri = F2FS_INODE(entry); 3720 3721 return (void *)&(ri->i_addr[DEF_ADDRS_PER_INODE(F2FS_I_SB(inode)) - 3722 get_inline_xattr_addrs(inode)]); 3723 } 3724 3725 static inline int inline_xattr_size(struct inode *inode) 3726 { 3727 if (f2fs_has_inline_xattr(inode)) 3728 return get_inline_xattr_addrs(inode) * sizeof(__le32); 3729 return 0; 3730 } 3731 3732 /* 3733 * Notice: check inline_data flag without inode page lock is unsafe. 3734 * It could change at any time by f2fs_convert_inline_folio(). 3735 */ 3736 static inline int f2fs_has_inline_data(struct inode *inode) 3737 { 3738 return is_inode_flag_set(inode, FI_INLINE_DATA); 3739 } 3740 3741 static inline int f2fs_exist_data(struct inode *inode) 3742 { 3743 return is_inode_flag_set(inode, FI_DATA_EXIST); 3744 } 3745 3746 static inline int f2fs_is_mmap_file(struct inode *inode) 3747 { 3748 return is_inode_flag_set(inode, FI_MMAP_FILE); 3749 } 3750 3751 static inline bool f2fs_is_pinned_file(struct inode *inode) 3752 { 3753 return is_inode_flag_set(inode, FI_PIN_FILE); 3754 } 3755 3756 static inline bool f2fs_is_atomic_file(struct inode *inode) 3757 { 3758 return is_inode_flag_set(inode, FI_ATOMIC_FILE); 3759 } 3760 3761 static inline bool f2fs_is_cow_file(struct inode *inode) 3762 { 3763 return is_inode_flag_set(inode, FI_COW_FILE); 3764 } 3765 3766 static inline void *inline_data_addr(struct inode *inode, 3767 const struct f2fs_cached_block *entry) 3768 { 3769 __le32 *addr = get_dnode_addr(inode, entry); 3770 3771 return (void *)(addr + DEF_INLINE_RESERVED_SIZE); 3772 } 3773 3774 static inline int f2fs_has_inline_dentry(struct inode *inode) 3775 { 3776 return is_inode_flag_set(inode, FI_INLINE_DENTRY); 3777 } 3778 3779 static inline int is_file(struct inode *inode, int type) 3780 { 3781 return F2FS_I(inode)->i_advise & type; 3782 } 3783 3784 static inline void set_file(struct inode *inode, int type) 3785 { 3786 if (is_file(inode, type)) 3787 return; 3788 F2FS_I(inode)->i_advise |= type; 3789 f2fs_mark_inode_dirty_sync(inode, true); 3790 } 3791 3792 static inline void clear_file(struct inode *inode, int type) 3793 { 3794 if (!is_file(inode, type)) 3795 return; 3796 F2FS_I(inode)->i_advise &= ~type; 3797 f2fs_mark_inode_dirty_sync(inode, true); 3798 } 3799 3800 static inline bool f2fs_is_time_consistent(struct inode *inode) 3801 { 3802 struct timespec64 ts = inode_get_atime(inode); 3803 3804 if (!timespec64_equal(F2FS_I(inode)->i_disk_time, &ts)) 3805 return false; 3806 ts = inode_get_ctime(inode); 3807 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 1, &ts)) 3808 return false; 3809 ts = inode_get_mtime(inode); 3810 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 2, &ts)) 3811 return false; 3812 return true; 3813 } 3814 3815 static inline bool f2fs_skip_inode_update(struct inode *inode, int dsync) 3816 { 3817 bool ret; 3818 3819 if (dsync) { 3820 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 3821 3822 spin_lock(&sbi->inode_lock[DIRTY_META]); 3823 ret = list_empty(&F2FS_I(inode)->gdirty_list); 3824 spin_unlock(&sbi->inode_lock[DIRTY_META]); 3825 return ret; 3826 } 3827 if (!is_inode_flag_set(inode, FI_AUTO_RECOVER) || 3828 file_keep_isize(inode) || 3829 i_size_read(inode) & ~PAGE_MASK) 3830 return false; 3831 3832 if (!f2fs_is_time_consistent(inode)) 3833 return false; 3834 3835 spin_lock(&F2FS_I(inode)->i_size_lock); 3836 ret = F2FS_I(inode)->last_disk_size == i_size_read(inode); 3837 spin_unlock(&F2FS_I(inode)->i_size_lock); 3838 3839 return ret; 3840 } 3841 3842 static inline bool f2fs_readonly(struct super_block *sb) 3843 { 3844 return sb_rdonly(sb); 3845 } 3846 3847 static inline bool f2fs_cp_error(struct f2fs_sb_info *sbi) 3848 { 3849 return is_set_ckpt_flags(sbi, CP_ERROR_FLAG); 3850 } 3851 3852 static inline void *f2fs_kmalloc(struct f2fs_sb_info *sbi, 3853 size_t size, gfp_t flags) 3854 { 3855 if (time_to_inject(sbi, FAULT_KMALLOC)) 3856 return NULL; 3857 3858 return kmalloc(size, flags); 3859 } 3860 3861 static inline void *f2fs_getname(struct f2fs_sb_info *sbi) 3862 { 3863 if (time_to_inject(sbi, FAULT_KMALLOC)) 3864 return NULL; 3865 3866 return __getname(); 3867 } 3868 3869 static inline void f2fs_putname(char *buf) 3870 { 3871 __putname(buf); 3872 } 3873 3874 static inline void *f2fs_kzalloc(struct f2fs_sb_info *sbi, 3875 size_t size, gfp_t flags) 3876 { 3877 return f2fs_kmalloc(sbi, size, flags | __GFP_ZERO); 3878 } 3879 3880 static inline void *f2fs_kvmalloc(struct f2fs_sb_info *sbi, 3881 size_t size, gfp_t flags) 3882 { 3883 if (time_to_inject(sbi, FAULT_KVMALLOC)) 3884 return NULL; 3885 3886 return kvmalloc(size, flags); 3887 } 3888 3889 static inline void *f2fs_kvzalloc(struct f2fs_sb_info *sbi, 3890 size_t size, gfp_t flags) 3891 { 3892 return f2fs_kvmalloc(sbi, size, flags | __GFP_ZERO); 3893 } 3894 3895 static inline void *f2fs_vmalloc(struct f2fs_sb_info *sbi, size_t size) 3896 { 3897 if (time_to_inject(sbi, FAULT_VMALLOC)) 3898 return NULL; 3899 3900 return vmalloc(size); 3901 } 3902 3903 static inline int get_extra_isize(struct inode *inode) 3904 { 3905 return F2FS_I(inode)->i_extra_isize / sizeof(__le32); 3906 } 3907 3908 static inline int get_inline_xattr_addrs(struct inode *inode) 3909 { 3910 return F2FS_I(inode)->i_inline_xattr_size; 3911 } 3912 3913 #define f2fs_get_inode_mode(i) \ 3914 ((is_inode_flag_set(i, FI_ACL_MODE)) ? \ 3915 (F2FS_I(i)->i_acl_mode) : ((i)->i_mode)) 3916 3917 #define F2FS_MIN_EXTRA_ATTR_SIZE (sizeof(__le32)) 3918 3919 #define F2FS_TOTAL_EXTRA_ATTR_SIZE \ 3920 (offsetof(struct f2fs_inode, i_extra_end) - \ 3921 offsetof(struct f2fs_inode, i_extra_isize)) \ 3922 3923 #define F2FS_OLD_ATTRIBUTE_SIZE (offsetof(struct f2fs_inode, i_addr)) 3924 #define F2FS_FITS_IN_INODE(f2fs_inode, extra_isize, field) \ 3925 ((offsetof(typeof(*(f2fs_inode)), field) + \ 3926 sizeof((f2fs_inode)->field)) \ 3927 <= (F2FS_OLD_ATTRIBUTE_SIZE + (extra_isize))) \ 3928 3929 #define __is_large_section(sbi) (SEGS_PER_SEC(sbi) > 1) 3930 3931 #define __is_meta_io(fio) (PAGE_TYPE_OF_BIO((fio)->type) == META) 3932 3933 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi, 3934 block_t blkaddr, int type); 3935 static inline void verify_blkaddr(struct f2fs_sb_info *sbi, 3936 block_t blkaddr, int type) 3937 { 3938 if (!f2fs_is_valid_blkaddr(sbi, blkaddr, type)) 3939 f2fs_err(sbi, "invalid blkaddr: %u, type: %d, run fsck to fix.", 3940 blkaddr, type); 3941 } 3942 3943 static inline bool __is_valid_data_blkaddr(block_t blkaddr) 3944 { 3945 if (blkaddr == NEW_ADDR || blkaddr == NULL_ADDR || 3946 blkaddr == COMPRESS_ADDR) 3947 return false; 3948 return true; 3949 } 3950 3951 /* 3952 * file.c 3953 */ 3954 int f2fs_sync_file(struct file *file, loff_t start, loff_t end, int datasync); 3955 int f2fs_do_truncate_blocks(struct inode *inode, u64 from, bool lock); 3956 int f2fs_truncate_blocks(struct inode *inode, u64 from, bool lock); 3957 int f2fs_truncate(struct inode *inode); 3958 int f2fs_getattr(const struct mnt_idmap *idmap, const struct path *path, 3959 struct kstat *stat, u32 request_mask, unsigned int flags); 3960 int f2fs_setattr(const struct mnt_idmap *idmap, struct dentry *dentry, 3961 struct iattr *attr); 3962 int f2fs_truncate_hole(struct inode *inode, pgoff_t pg_start, pgoff_t pg_end); 3963 void f2fs_truncate_data_blocks_range(struct dnode_of_data *dn, int count); 3964 int f2fs_do_shutdown(struct f2fs_sb_info *sbi, unsigned int flag, 3965 bool readonly, bool need_lock); 3966 int f2fs_precache_extents(struct inode *inode); 3967 int f2fs_fileattr_get(struct dentry *dentry, struct file_kattr *fa); 3968 int f2fs_fileattr_set(const struct mnt_idmap *idmap, 3969 struct dentry *dentry, struct file_kattr *fa); 3970 long f2fs_ioctl(struct file *filp, unsigned int cmd, unsigned long arg); 3971 long f2fs_compat_ioctl(struct file *file, unsigned int cmd, unsigned long arg); 3972 int f2fs_transfer_project_quota(struct inode *inode, kprojid_t kprojid); 3973 int f2fs_pin_file_control(struct inode *inode, bool inc); 3974 3975 /* 3976 * inode.c 3977 */ 3978 void f2fs_set_inode_flags(struct inode *inode); 3979 bool f2fs_inode_chksum_verify(struct f2fs_sb_info *sbi, struct f2fs_cached_block *entry); 3980 void f2fs_inode_chksum_set(struct f2fs_sb_info *sbi, struct f2fs_cached_block *entry); 3981 struct inode *f2fs_iget(struct super_block *sb, unsigned long ino); 3982 struct inode *f2fs_iget_retry(struct super_block *sb, unsigned long ino); 3983 int f2fs_try_to_free_nats(struct f2fs_sb_info *sbi, int nr_shrink); 3984 void f2fs_update_inode(struct inode *inode, struct f2fs_cached_block *entry); 3985 void f2fs_update_inode_cache(struct inode *inode); 3986 int f2fs_write_inode(struct inode *inode, struct writeback_control *wbc); 3987 void f2fs_remove_donate_inode(struct inode *inode); 3988 void f2fs_evict_inode(struct inode *inode); 3989 void f2fs_handle_failed_inode(struct inode *inode, 3990 struct f2fs_lock_context *lc, bool add_orphan); 3991 int f2fs_init_evict_inode_work(void); 3992 void f2fs_destroy_evict_inode_work(void); 3993 3994 /* 3995 * namei.c 3996 */ 3997 int f2fs_update_extension_list(struct f2fs_sb_info *sbi, const char *name, 3998 bool hot, bool set); 3999 struct dentry *f2fs_get_parent(struct dentry *child); 4000 int f2fs_get_tmpfile(const struct mnt_idmap *idmap, struct inode *dir, 4001 struct inode **new_inode); 4002 4003 /* 4004 * dir.c 4005 */ 4006 #if IS_ENABLED(CONFIG_UNICODE) 4007 int f2fs_init_casefolded_name(const struct inode *dir, 4008 struct f2fs_filename *fname); 4009 void f2fs_free_casefolded_name(struct f2fs_filename *fname); 4010 #else 4011 static inline int f2fs_init_casefolded_name(const struct inode *dir, 4012 struct f2fs_filename *fname) 4013 { 4014 return 0; 4015 } 4016 4017 static inline void f2fs_free_casefolded_name(struct f2fs_filename *fname) 4018 { 4019 } 4020 #endif /* CONFIG_UNICODE */ 4021 4022 int f2fs_setup_filename(struct inode *dir, const struct qstr *iname, 4023 int lookup, struct f2fs_filename *fname); 4024 int f2fs_prepare_lookup(struct inode *dir, struct dentry *dentry, 4025 struct f2fs_filename *fname); 4026 void f2fs_free_filename(struct f2fs_filename *fname); 4027 struct f2fs_dir_entry *f2fs_find_target_dentry(const struct f2fs_dentry_ptr *d, 4028 const struct f2fs_filename *fname, int *max_slots, 4029 bool use_hash); 4030 int f2fs_fill_dentries(struct dir_context *ctx, struct f2fs_dentry_ptr *d, 4031 unsigned int start_pos, struct fscrypt_str *fstr); 4032 void f2fs_do_make_empty_dir(struct inode *inode, struct inode *parent, 4033 struct f2fs_dentry_ptr *d); 4034 struct f2fs_cached_block *f2fs_init_inode_metadata(struct inode *inode, struct inode *dir, 4035 const struct f2fs_filename *fname, struct f2fs_cached_block *dentry); 4036 void f2fs_update_parent_metadata(struct inode *dir, struct inode *inode, 4037 unsigned int current_depth); 4038 int f2fs_room_for_filename(const void *bitmap, int slots, int max_slots); 4039 void f2fs_drop_nlink(struct inode *dir, struct inode *inode); 4040 struct f2fs_dir_entry *__f2fs_find_entry(struct inode *dir, 4041 const struct f2fs_filename *fname, void **dentry_block); 4042 struct f2fs_dir_entry *f2fs_find_entry(struct inode *dir, 4043 const struct qstr *child, void **dentry_block); 4044 struct f2fs_dir_entry *f2fs_parent_dir(struct inode *dir, void **dentry_block); 4045 ino_t f2fs_inode_by_name(struct inode *dir, const struct qstr *qstr, 4046 void **dentry_block); 4047 void f2fs_set_link(struct inode *dir, struct f2fs_dir_entry *de, 4048 void *dentry_blk, struct inode *inode); 4049 bool f2fs_has_enough_room(struct inode *dir, struct f2fs_cached_block *ientry, 4050 const struct f2fs_filename *fname); 4051 void f2fs_update_dentry(nid_t ino, umode_t mode, struct f2fs_dentry_ptr *d, 4052 const struct fscrypt_str *name, f2fs_hash_t name_hash, 4053 unsigned int bit_pos); 4054 int f2fs_add_regular_entry(struct inode *dir, const struct f2fs_filename *fname, 4055 struct inode *inode, nid_t ino, umode_t mode); 4056 int f2fs_add_dentry(struct inode *dir, const struct f2fs_filename *fname, 4057 struct inode *inode, nid_t ino, umode_t mode); 4058 int f2fs_do_add_link(struct inode *dir, const struct qstr *name, 4059 struct inode *inode, nid_t ino, umode_t mode); 4060 void f2fs_delete_entry(struct f2fs_dir_entry *dentry, void *dentry_blk, 4061 struct inode *dir, struct inode *inode); 4062 int f2fs_do_tmpfile(struct inode *inode, struct inode *dir, 4063 struct f2fs_filename *fname); 4064 bool f2fs_empty_dir(struct inode *dir); 4065 4066 static inline int f2fs_add_link(struct dentry *dentry, struct inode *inode) 4067 { 4068 if (fscrypt_is_nokey_name(dentry)) 4069 return -ENOKEY; 4070 return f2fs_do_add_link(d_inode(dentry->d_parent), &dentry->d_name, 4071 inode, inode->i_ino, inode->i_mode); 4072 } 4073 4074 /* 4075 * super.c 4076 */ 4077 int f2fs_inode_dirtied(struct inode *inode, bool sync); 4078 void f2fs_inode_synced(struct inode *inode); 4079 int f2fs_dquot_initialize(struct inode *inode); 4080 int f2fs_enable_quota_files(struct f2fs_sb_info *sbi, bool rdonly); 4081 int f2fs_do_quota_sync(struct super_block *sb, int type); 4082 loff_t max_file_blocks(struct f2fs_sb_info *sbi, struct inode *inode); 4083 void f2fs_quota_off_umount(struct super_block *sb); 4084 void f2fs_save_errors(struct f2fs_sb_info *sbi, unsigned char flag); 4085 void f2fs_handle_error(struct f2fs_sb_info *sbi, unsigned char error); 4086 int f2fs_commit_super(struct f2fs_sb_info *sbi, bool recover); 4087 int f2fs_sync_fs(struct super_block *sb, int sync); 4088 int f2fs_sanity_check_ckpt(struct f2fs_sb_info *sbi); 4089 4090 /* 4091 * hash.c 4092 */ 4093 void f2fs_hash_filename(const struct inode *dir, struct f2fs_filename *fname); 4094 4095 /* 4096 * node.c 4097 */ 4098 struct node_info; 4099 enum node_type; 4100 4101 int f2fs_check_nid_range(struct f2fs_sb_info *sbi, nid_t nid); 4102 bool f2fs_available_free_memory(struct f2fs_sb_info *sbi, int type); 4103 bool f2fs_in_warm_node_list(struct f2fs_sb_info *sbi, struct f2fs_cached_block *entry); 4104 void f2fs_init_fsync_node_info(struct f2fs_sb_info *sbi); 4105 void f2fs_del_fsync_node_entry(struct f2fs_sb_info *sbi, struct f2fs_cached_block *entry); 4106 void f2fs_reset_fsync_node_info(struct f2fs_sb_info *sbi); 4107 bool f2fs_need_dentry_mark(struct f2fs_sb_info *sbi, nid_t nid); 4108 bool f2fs_is_checkpointed_node(struct f2fs_sb_info *sbi, nid_t nid); 4109 bool f2fs_need_inode_block_update(struct f2fs_sb_info *sbi, nid_t ino); 4110 int f2fs_get_node_info(struct f2fs_sb_info *sbi, nid_t nid, 4111 struct node_info *ni, bool checkpoint_context); 4112 pgoff_t f2fs_get_next_page_offset(struct dnode_of_data *dn, pgoff_t pgofs); 4113 int f2fs_get_dnode_of_data(struct dnode_of_data *dn, pgoff_t index, int mode); 4114 int f2fs_truncate_inode_blocks(struct inode *inode, pgoff_t from); 4115 int f2fs_truncate_xattr_node(struct inode *inode); 4116 int f2fs_wait_on_node_caches_writeback(struct f2fs_sb_info *sbi, 4117 unsigned int seq_id); 4118 int f2fs_write_node_caches(struct f2fs_sb_info *sbi, long *nr_written); 4119 int f2fs_remove_inode_cache(struct inode *inode); 4120 struct f2fs_cached_block *f2fs_new_inode_cache(struct inode *inode); 4121 struct f2fs_cached_block *f2fs_new_node_cache(struct dnode_of_data *dn, unsigned int ofs); 4122 void f2fs_ra_node_cache(struct f2fs_sb_info *sbi, nid_t nid); 4123 struct f2fs_cached_block *f2fs_get_node_cache(struct f2fs_sb_info *sbi, pgoff_t nid, 4124 enum node_type node_type); 4125 int f2fs_sanity_check_node_footer(struct f2fs_sb_info *sbi, 4126 struct f2fs_cached_block *entry, pgoff_t nid, 4127 enum node_type ntype, bool in_irq); 4128 struct f2fs_cached_block *f2fs_get_inode_cache(struct f2fs_sb_info *sbi, pgoff_t ino); 4129 struct f2fs_cached_block *f2fs_get_xnode_cache(struct f2fs_sb_info *sbi, pgoff_t xnid); 4130 int f2fs_write_node_cache(struct f2fs_cached_block *entry, int sync_mode, 4131 bool mark_dirty, enum iostat_type io_type); 4132 int f2fs_move_node_cache(struct f2fs_cached_block *entry, int gc_type); 4133 void f2fs_flush_inline_data(struct f2fs_sb_info *sbi); 4134 int f2fs_fsync_node_caches(struct f2fs_sb_info *sbi, struct inode *inode, 4135 bool atomic, unsigned int *seq_id); 4136 int f2fs_writeback_node_caches(struct f2fs_sb_info *sbi, long nr_to_write, 4137 bool sync, bool do_balance, enum iostat_type io_type, 4138 long *nr_written); 4139 int f2fs_build_free_nids(struct f2fs_sb_info *sbi, bool sync, bool mount); 4140 bool f2fs_alloc_nid(struct f2fs_sb_info *sbi, nid_t *nid); 4141 void f2fs_alloc_nid_done(struct f2fs_sb_info *sbi, nid_t nid); 4142 void f2fs_alloc_nid_failed(struct f2fs_sb_info *sbi, nid_t nid); 4143 int f2fs_try_to_free_nids(struct f2fs_sb_info *sbi, int nr_shrink); 4144 int f2fs_recover_inline_xattr(struct inode *inode, struct f2fs_cached_block *entry); 4145 int f2fs_recover_xattr_data(struct inode *inode, struct f2fs_cached_block *entry); 4146 int f2fs_recover_inode_cache(struct f2fs_sb_info *sbi, 4147 struct f2fs_cached_block *entry); 4148 int f2fs_restore_node_summary(struct f2fs_sb_info *sbi, 4149 unsigned int segno, struct f2fs_summary_block *sum); 4150 int f2fs_flush_nat_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc); 4151 int f2fs_build_node_manager(struct f2fs_sb_info *sbi); 4152 void f2fs_destroy_node_manager(struct f2fs_sb_info *sbi); 4153 int __init f2fs_create_node_manager_caches(void); 4154 void f2fs_destroy_node_manager_caches(void); 4155 4156 /* 4157 * segment.c 4158 */ 4159 bool f2fs_need_SSR(struct f2fs_sb_info *sbi); 4160 int f2fs_commit_atomic_write(struct inode *inode); 4161 void f2fs_abort_atomic_write(struct inode *inode, bool clean); 4162 void f2fs_balance_fs(struct f2fs_sb_info *sbi, bool need); 4163 void f2fs_balance_fs_bg(struct f2fs_sb_info *sbi, bool from_bg); 4164 int f2fs_issue_flush(struct f2fs_sb_info *sbi, nid_t ino); 4165 int f2fs_create_flush_cmd_control(struct f2fs_sb_info *sbi); 4166 int f2fs_flush_device_cache(struct f2fs_sb_info *sbi); 4167 void f2fs_destroy_flush_cmd_control(struct f2fs_sb_info *sbi, bool free); 4168 void f2fs_invalidate_blocks(struct f2fs_sb_info *sbi, block_t addr, 4169 unsigned int len); 4170 void f2fs_reserve_device_alias(struct f2fs_sb_info *sbi, block_t addr, 4171 unsigned int len); 4172 bool f2fs_is_checkpointed_data(struct f2fs_sb_info *sbi, block_t blkaddr); 4173 int f2fs_start_discard_thread(struct f2fs_sb_info *sbi); 4174 void f2fs_drop_discard_cmd(struct f2fs_sb_info *sbi); 4175 void f2fs_drop_discard_cmd_range(struct f2fs_sb_info *sbi, 4176 block_t start, block_t len); 4177 void f2fs_stop_discard_thread(struct f2fs_sb_info *sbi); 4178 bool f2fs_issue_discard_timeout(struct f2fs_sb_info *sbi, bool need_check); 4179 void f2fs_clear_prefree_segments(struct f2fs_sb_info *sbi, 4180 struct cp_control *cpc); 4181 void f2fs_dirty_to_prefree(struct f2fs_sb_info *sbi); 4182 block_t f2fs_get_unusable_blocks(struct f2fs_sb_info *sbi); 4183 int f2fs_disable_cp_again(struct f2fs_sb_info *sbi, block_t unusable); 4184 void f2fs_release_discard_addrs(struct f2fs_sb_info *sbi); 4185 int f2fs_nblocks_for_summary_flush(struct f2fs_sb_info *sbi, bool for_ra); 4186 bool f2fs_segment_has_free_slot(struct f2fs_sb_info *sbi, int segno); 4187 int f2fs_init_inmem_curseg(struct f2fs_sb_info *sbi); 4188 int f2fs_reinit_atgc_curseg(struct f2fs_sb_info *sbi); 4189 void f2fs_save_inmem_curseg(struct f2fs_sb_info *sbi); 4190 void f2fs_restore_inmem_curseg(struct f2fs_sb_info *sbi); 4191 int f2fs_allocate_segment_for_resize(struct f2fs_sb_info *sbi, int type, 4192 unsigned int start, unsigned int end); 4193 int f2fs_allocate_new_section(struct f2fs_sb_info *sbi, int type, bool force); 4194 int f2fs_allocate_pinning_section(struct f2fs_sb_info *sbi); 4195 int f2fs_allocate_new_segments(struct f2fs_sb_info *sbi); 4196 int f2fs_trim_fs(struct f2fs_sb_info *sbi, struct fstrim_range *range); 4197 bool f2fs_exist_trim_candidates(struct f2fs_sb_info *sbi, 4198 struct cp_control *cpc); 4199 struct f2fs_cached_block *f2fs_get_sum_cache(struct f2fs_sb_info *sbi, 4200 unsigned int segno); 4201 void f2fs_update_meta_block(struct f2fs_sb_info *sbi, void *src, 4202 block_t blk_addr); 4203 void f2fs_do_write_meta_cache(struct f2fs_sb_info *sbi, 4204 struct f2fs_cached_block *entry, 4205 enum iostat_type io_type); 4206 void f2fs_do_write_node_cache(unsigned int nid, struct f2fs_io_info *fio); 4207 void f2fs_outplace_write_data(struct dnode_of_data *dn, 4208 struct f2fs_io_info *fio); 4209 int f2fs_inplace_write_data(struct f2fs_io_info *fio); 4210 void f2fs_do_replace_block(struct f2fs_sb_info *sbi, struct f2fs_summary *sum, 4211 block_t old_blkaddr, block_t new_blkaddr, 4212 bool recover_curseg, bool recover_newaddr, 4213 bool from_gc); 4214 void f2fs_replace_block(struct f2fs_sb_info *sbi, struct dnode_of_data *dn, 4215 block_t old_addr, block_t new_addr, 4216 unsigned char version, bool recover_curseg, 4217 bool recover_newaddr); 4218 enum temp_type f2fs_get_segment_temp(struct f2fs_sb_info *sbi, 4219 enum log_type seg_type); 4220 int f2fs_allocate_data_block(struct f2fs_sb_info *sbi, 4221 block_t old_blkaddr, block_t *new_blkaddr, 4222 struct f2fs_summary *sum, int type, 4223 struct f2fs_io_info *fio); 4224 void f2fs_update_device_state(struct f2fs_sb_info *sbi, nid_t ino, 4225 block_t blkaddr, unsigned int blkcnt); 4226 void f2fs_folio_wait_writeback(struct folio *folio, bool ordered, bool locked); 4227 void f2fs_wait_on_block_writeback(struct inode *inode, block_t blkaddr); 4228 void f2fs_wait_on_block_writeback_range(struct inode *inode, block_t blkaddr, 4229 block_t len); 4230 void f2fs_write_data_summaries(struct f2fs_sb_info *sbi, block_t start_blk); 4231 void f2fs_write_node_summaries(struct f2fs_sb_info *sbi, block_t start_blk); 4232 int f2fs_lookup_journal_in_cursum(struct f2fs_sb_info *sbi, 4233 struct f2fs_journal *journal, int type, 4234 unsigned int val, int alloc); 4235 void f2fs_flush_sit_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc); 4236 int f2fs_check_and_fix_write_pointer(struct f2fs_sb_info *sbi); 4237 int f2fs_build_segment_manager(struct f2fs_sb_info *sbi); 4238 void f2fs_destroy_segment_manager(struct f2fs_sb_info *sbi); 4239 int __init f2fs_create_segment_manager_caches(void); 4240 void f2fs_destroy_segment_manager_caches(void); 4241 int f2fs_rw_hint_to_seg_type(struct f2fs_sb_info *sbi, enum rw_hint hint); 4242 enum rw_hint f2fs_io_type_to_rw_hint(struct f2fs_sb_info *sbi, 4243 enum page_type type, enum temp_type temp); 4244 u8 f2fs_io_type_to_write_stream(struct block_device *bdev, 4245 enum page_type type, enum temp_type temp); 4246 unsigned int f2fs_usable_segs_in_sec(struct f2fs_sb_info *sbi); 4247 unsigned int f2fs_usable_blks_in_seg(struct f2fs_sb_info *sbi, 4248 unsigned int segno); 4249 unsigned long long f2fs_get_section_mtime(struct f2fs_sb_info *sbi, 4250 unsigned int segno); 4251 4252 static inline struct inode *fio_inode(struct f2fs_io_info *fio) 4253 { 4254 return fio->folio->mapping->host; 4255 } 4256 4257 #define DEF_FRAGMENT_SIZE 4 4258 #define MIN_FRAGMENT_SIZE 1 4259 #define MAX_FRAGMENT_SIZE 512 4260 4261 static inline bool f2fs_need_rand_blk(struct f2fs_sb_info *sbi, 4262 enum log_type type) 4263 { 4264 if (type == CURSEG_COLD_DATA_PINNED) 4265 return false; 4266 return F2FS_OPTION(sbi).fs_mode == FS_MODE_FRAGMENT_BLK; 4267 } 4268 4269 static inline bool f2fs_need_rand_seg(struct f2fs_sb_info *sbi, 4270 enum log_type type) 4271 { 4272 if (type == CURSEG_COLD_DATA_PINNED) 4273 return false; 4274 return F2FS_OPTION(sbi).fs_mode == FS_MODE_FRAGMENT_SEG; 4275 } 4276 4277 static inline bool f2fs_need_rand_seg_blk(struct f2fs_sb_info *sbi, 4278 enum log_type type) 4279 { 4280 return f2fs_need_rand_blk(sbi, type) || f2fs_need_rand_seg(sbi, type); 4281 } 4282 4283 /* 4284 * checkpoint.c 4285 */ 4286 void f2fs_lock_op(struct f2fs_sb_info *sbi, struct f2fs_lock_context *lc); 4287 int f2fs_trylock_op(struct f2fs_sb_info *sbi, struct f2fs_lock_context *lc); 4288 void f2fs_unlock_op(struct f2fs_sb_info *sbi, struct f2fs_lock_context *lc); 4289 void f2fs_stop_checkpoint(struct f2fs_sb_info *sbi, bool end_io, 4290 unsigned char reason); 4291 void f2fs_flush_ckpt_thread(struct f2fs_sb_info *sbi); 4292 struct f2fs_cached_block *f2fs_grab_meta_cache(struct f2fs_sb_info *sbi, pgoff_t index); 4293 struct f2fs_cached_block *f2fs_get_meta_cache(struct f2fs_sb_info *sbi, pgoff_t index); 4294 struct f2fs_cached_block *f2fs_get_meta_cache_retry(struct f2fs_sb_info *sbi, pgoff_t index); 4295 struct f2fs_cached_block *f2fs_get_tmp_cache(struct f2fs_sb_info *sbi, pgoff_t index); 4296 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi, 4297 block_t blkaddr, int type); 4298 bool f2fs_is_valid_blkaddr_raw(struct f2fs_sb_info *sbi, 4299 block_t blkaddr, int type); 4300 int f2fs_ra_meta_caches(struct f2fs_sb_info *sbi, block_t start, int nrpages, 4301 int type, bool sync); 4302 void f2fs_ra_meta_caches_cond(struct f2fs_sb_info *sbi, pgoff_t index, 4303 unsigned int ra_blocks); 4304 void f2fs_write_meta_caches(struct f2fs_sb_info *sbi, long *nr_written); 4305 long f2fs_sync_meta_caches(struct f2fs_sb_info *sbi, long nr_to_write, 4306 bool sync, enum iostat_type io_type); 4307 void f2fs_add_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type); 4308 void f2fs_remove_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type); 4309 void f2fs_release_ino_entry(struct f2fs_sb_info *sbi, bool all); 4310 bool f2fs_exist_written_data(struct f2fs_sb_info *sbi, nid_t ino, int mode); 4311 void f2fs_set_dirty_device(struct f2fs_sb_info *sbi, nid_t ino, 4312 unsigned int devidx, int type); 4313 bool f2fs_is_dirty_device(struct f2fs_sb_info *sbi, nid_t ino, 4314 unsigned int devidx, int type); 4315 int f2fs_acquire_orphan_inode(struct f2fs_sb_info *sbi); 4316 void f2fs_release_orphan_inode(struct f2fs_sb_info *sbi); 4317 void f2fs_add_orphan_inode(struct inode *inode); 4318 void f2fs_remove_orphan_inode(struct f2fs_sb_info *sbi, nid_t ino); 4319 int f2fs_recover_orphan_inodes(struct f2fs_sb_info *sbi); 4320 int f2fs_get_valid_checkpoint(struct f2fs_sb_info *sbi); 4321 void f2fs_update_dirty_folio(struct inode *inode, struct folio *folio); 4322 void f2fs_remove_dirty_inode(struct inode *inode); 4323 int f2fs_sync_dirty_inodes(struct f2fs_sb_info *sbi, enum inode_type type, 4324 bool from_cp); 4325 void f2fs_sync_dirty_data(struct f2fs_sb_info *sbi, int type); 4326 u64 f2fs_get_sectors_written(struct f2fs_sb_info *sbi); 4327 int f2fs_write_checkpoint(struct f2fs_sb_info *sbi, struct cp_control *cpc); 4328 void f2fs_init_ino_entry_info(struct f2fs_sb_info *sbi); 4329 int __init f2fs_create_checkpoint_caches(void); 4330 void f2fs_destroy_checkpoint_caches(void); 4331 int f2fs_issue_checkpoint(struct f2fs_sb_info *sbi); 4332 int f2fs_start_ckpt_thread(struct f2fs_sb_info *sbi); 4333 void f2fs_stop_ckpt_thread(struct f2fs_sb_info *sbi); 4334 void f2fs_init_ckpt_req_control(struct f2fs_sb_info *sbi); 4335 4336 /* 4337 * data.c 4338 */ 4339 int __init f2fs_init_bioset(void); 4340 void f2fs_destroy_bioset(void); 4341 bool f2fs_is_cp_guaranteed(const struct folio *folio); 4342 int f2fs_init_bio_entry_cache(void); 4343 void f2fs_destroy_bio_entry_cache(void); 4344 void f2fs_submit_read_bio(struct f2fs_sb_info *sbi, struct bio *bio, 4345 enum page_type type); 4346 int f2fs_init_write_merge_io(struct f2fs_sb_info *sbi); 4347 void f2fs_submit_merged_write_cond(struct f2fs_sb_info *sbi, 4348 struct inode *inode, struct folio *folio); 4349 void f2fs_submit_merged_write_folio(struct f2fs_sb_info *sbi, 4350 struct folio *folio); 4351 bool f2fs_submit_merged_write_cache(struct f2fs_sb_info *sbi, 4352 struct f2fs_cached_block *entry, 4353 nid_t ino, enum page_type type); 4354 void f2fs_submit_merged_write(struct f2fs_sb_info *sbi, enum page_type type); 4355 void f2fs_submit_merged_ipu_write(struct f2fs_sb_info *sbi, 4356 struct bio **bio, struct folio *folio); 4357 void f2fs_submit_all_merged_ipu_writes(struct f2fs_sb_info *sbi); 4358 void f2fs_flush_merged_writes(struct f2fs_sb_info *sbi); 4359 int f2fs_submit_page_bio(struct f2fs_io_info *fio); 4360 int f2fs_merge_page_bio(struct f2fs_io_info *fio); 4361 void f2fs_submit_page_write(struct f2fs_io_info *fio); 4362 int f2fs_submit_cache_read(struct f2fs_io_info *fio); 4363 void f2fs_submit_cache_write(struct f2fs_io_info *fio); 4364 struct block_device *f2fs_target_device(struct f2fs_sb_info *sbi, 4365 block_t blk_addr, sector_t *sector); 4366 int f2fs_target_device_index(struct f2fs_sb_info *sbi, block_t blkaddr); 4367 void f2fs_set_data_blkaddr(struct dnode_of_data *dn, block_t blkaddr); 4368 void f2fs_update_data_blkaddr(struct dnode_of_data *dn, block_t blkaddr); 4369 int f2fs_reserve_new_blocks(struct dnode_of_data *dn, blkcnt_t count); 4370 int f2fs_reserve_new_block(struct dnode_of_data *dn); 4371 int f2fs_get_block_locked(struct dnode_of_data *dn, pgoff_t index); 4372 int f2fs_reserve_block(struct dnode_of_data *dn, pgoff_t index); 4373 struct folio *f2fs_get_read_data_folio(struct inode *inode, pgoff_t index, 4374 blk_opf_t op_flags, bool for_write, pgoff_t *next_pgofs); 4375 struct folio *f2fs_find_data_folio(struct inode *inode, pgoff_t index, 4376 pgoff_t *next_pgofs); 4377 struct folio *f2fs_get_lock_data_folio(struct inode *inode, pgoff_t index, 4378 bool for_write); 4379 struct folio *f2fs_get_new_data_folio(struct inode *inode, 4380 struct f2fs_cached_block *ientry, pgoff_t index, bool new_i_size); 4381 int f2fs_do_write_data_page(struct f2fs_io_info *fio); 4382 int f2fs_map_blocks(struct inode *inode, struct f2fs_map_blocks *map, int flag); 4383 int f2fs_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, 4384 u64 start, u64 len); 4385 bool f2fs_should_update_inplace(struct inode *inode, struct f2fs_io_info *fio); 4386 bool f2fs_should_update_outplace(struct inode *inode, struct f2fs_io_info *fio); 4387 int f2fs_write_single_data_page(struct folio *folio, int *submitted, 4388 struct bio **bio, sector_t *last_block, 4389 struct writeback_control *wbc, 4390 enum iostat_type io_type, 4391 int compr_blocks, bool allow_balance); 4392 void f2fs_write_failed(struct inode *inode, loff_t to); 4393 void f2fs_invalidate_folio(struct folio *folio, size_t offset, size_t length); 4394 bool f2fs_release_folio(struct folio *folio, gfp_t wait); 4395 bool f2fs_overwrite_io(struct inode *inode, loff_t pos, size_t len); 4396 void f2fs_clear_page_cache_dirty_tag(struct folio *folio); 4397 int f2fs_init_post_read_processing(void); 4398 void f2fs_destroy_post_read_processing(void); 4399 int f2fs_init_wq(struct f2fs_sb_info *sbi); 4400 void f2fs_destroy_wq(struct f2fs_sb_info *sbi); 4401 DECLARE_IOMAP_ITER_NEXT(f2fs_iomap_next); 4402 4403 /* 4404 * gc.c 4405 */ 4406 int f2fs_start_gc_thread(struct f2fs_sb_info *sbi); 4407 void f2fs_stop_gc_thread(struct f2fs_sb_info *sbi); 4408 block_t f2fs_start_bidx_of_node(unsigned int node_ofs, struct inode *inode); 4409 int f2fs_gc(struct f2fs_sb_info *sbi, struct f2fs_gc_control *gc_control); 4410 void f2fs_build_gc_manager(struct f2fs_sb_info *sbi); 4411 int f2fs_gc_range(struct f2fs_sb_info *sbi, 4412 unsigned int start_seg, unsigned int end_seg, 4413 bool dry_run, unsigned int dry_run_sections, bool lock); 4414 void f2fs_reset_gc_victim_resource(struct f2fs_sb_info *sbi, 4415 unsigned int start, unsigned int end); 4416 int f2fs_resize_fs(struct file *filp, __u64 block_count); 4417 int __init f2fs_create_garbage_collection_cache(void); 4418 void f2fs_destroy_garbage_collection_cache(void); 4419 /* victim selection function for cleaning and SSR */ 4420 int f2fs_get_victim(struct f2fs_sb_info *sbi, unsigned int *result, 4421 int gc_type, int type, char alloc_mode, 4422 unsigned long long age, bool one_time); 4423 4424 /* 4425 * recovery.c 4426 */ 4427 int f2fs_recover_fsync_data(struct f2fs_sb_info *sbi, bool check_only); 4428 bool f2fs_space_for_roll_forward(struct f2fs_sb_info *sbi); 4429 int __init f2fs_create_recovery_cache(void); 4430 void f2fs_destroy_recovery_cache(void); 4431 4432 /* 4433 * debug.c 4434 */ 4435 #ifdef CONFIG_F2FS_STAT_FS 4436 enum { 4437 DEVSTAT_INUSE, 4438 DEVSTAT_DIRTY, 4439 DEVSTAT_FULL, 4440 DEVSTAT_FREE, 4441 DEVSTAT_PREFREE, 4442 DEVSTAT_MAX, 4443 }; 4444 4445 struct f2fs_dev_stats { 4446 unsigned int devstats[2][DEVSTAT_MAX]; /* 0: segs, 1: secs */ 4447 }; 4448 4449 struct f2fs_stat_info { 4450 struct list_head stat_list; 4451 struct f2fs_sb_info *sbi; 4452 int all_area_segs, sit_area_segs, nat_area_segs, ssa_area_segs; 4453 int main_area_segs, main_area_sections, main_area_zones; 4454 unsigned long long hit_cached[NR_EXTENT_CACHES]; 4455 unsigned long long hit_rbtree[NR_EXTENT_CACHES]; 4456 unsigned long long total_ext[NR_EXTENT_CACHES]; 4457 unsigned long long hit_total[NR_EXTENT_CACHES]; 4458 int ext_tree[NR_EXTENT_CACHES]; 4459 int zombie_tree[NR_EXTENT_CACHES]; 4460 int ext_node[NR_EXTENT_CACHES]; 4461 /* to count memory footprint */ 4462 unsigned long long ext_mem[NR_EXTENT_CACHES]; 4463 /* for read extent cache */ 4464 unsigned long long hit_largest; 4465 /* for block age extent cache */ 4466 unsigned long long allocated_data_blocks; 4467 int ndirty_node, ndirty_dent, ndirty_meta, ndirty_imeta; 4468 int ndirty_data, ndirty_qdata; 4469 unsigned int ndirty_dirs, ndirty_files, ndirty_all; 4470 unsigned int nquota_files, ndonate_files; 4471 int nats, dirty_nats, sits, dirty_sits; 4472 int free_nids, avail_nids, alloc_nids; 4473 int total_count, utilization; 4474 int nr_wb_cp_data, nr_wb_data; 4475 int nr_rd_data, nr_rd_node, nr_rd_meta; 4476 int nr_dio_read, nr_dio_write; 4477 unsigned int io_skip_bggc, other_skip_bggc; 4478 int nr_flushing, nr_flushed, flush_list_empty; 4479 int nr_discarding, nr_discarded; 4480 int nr_discard_cmd; 4481 unsigned int undiscard_blks; 4482 int nr_issued_ckpt, nr_total_ckpt, nr_queued_ckpt; 4483 unsigned int cur_ckpt_time, peak_ckpt_time; 4484 int inline_xattr, inline_inode, inline_dir, append, update, orphans; 4485 int compr_inode, swapfile_inode; 4486 unsigned long long compr_blocks; 4487 int aw_cnt, max_aw_cnt; 4488 unsigned int valid_count, valid_node_count, valid_inode_count, discard_blks; 4489 unsigned int bimodal, avg_vblocks; 4490 int util_free, util_valid, util_invalid; 4491 int rsvd_segs, overp_segs; 4492 int dirty_count, node_caches, meta_caches, compress_pages; 4493 int compress_page_hit; 4494 int prefree_count, free_segs, free_secs; 4495 int cp_call_count[MAX_CALL_TYPE], cp_count; 4496 int gc_call_count[MAX_CALL_TYPE]; 4497 int gc_segs[2][2]; 4498 int gc_secs[2][2]; 4499 int tot_blks, data_blks, node_blks; 4500 int bg_data_blks, bg_node_blks; 4501 unsigned int defrag_blks; 4502 int blkoff[NR_CURSEG_TYPE]; 4503 int curseg[NR_CURSEG_TYPE]; 4504 int cursec[NR_CURSEG_TYPE]; 4505 int curzone[NR_CURSEG_TYPE]; 4506 unsigned int dirty_seg[NR_CURSEG_TYPE]; 4507 unsigned int full_seg[NR_CURSEG_TYPE]; 4508 unsigned int valid_blks[NR_CURSEG_TYPE]; 4509 4510 unsigned int meta_count[META_MAX]; 4511 unsigned int segment_count[2]; 4512 unsigned int block_count[2]; 4513 unsigned int inplace_count; 4514 unsigned long long base_mem, cache_mem, page_mem; 4515 unsigned long long cache_entry_mem[NR_CACHE_TYPES]; 4516 unsigned long long cache_data_mem[NR_CACHE_TYPES]; 4517 struct f2fs_dev_stats *dev_stats; 4518 }; 4519 4520 static inline struct f2fs_stat_info *F2FS_STAT(struct f2fs_sb_info *sbi) 4521 { 4522 return (struct f2fs_stat_info *)sbi->stat_info; 4523 } 4524 4525 #define stat_inc_cp_call_count(sbi, foreground) \ 4526 atomic_inc(&sbi->cp_call_count[(foreground)]) 4527 #define stat_inc_cp_count(sbi) (F2FS_STAT(sbi)->cp_count++) 4528 #define stat_io_skip_bggc_count(sbi) ((sbi)->io_skip_bggc++) 4529 #define stat_other_skip_bggc_count(sbi) ((sbi)->other_skip_bggc++) 4530 #define stat_inc_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]++) 4531 #define stat_dec_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]--) 4532 #define stat_inc_total_hit(sbi, type) (atomic64_inc(&(sbi)->total_hit_ext[type])) 4533 #define stat_inc_rbtree_node_hit(sbi, type) (atomic64_inc(&(sbi)->read_hit_rbtree[type])) 4534 #define stat_inc_largest_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_largest)) 4535 #define stat_inc_cached_node_hit(sbi, type) (atomic64_inc(&(sbi)->read_hit_cached[type])) 4536 #define stat_inc_inline_xattr(inode) \ 4537 do { \ 4538 if (f2fs_has_inline_xattr(inode)) \ 4539 (atomic_inc(&F2FS_I_SB(inode)->inline_xattr)); \ 4540 } while (0) 4541 #define stat_dec_inline_xattr(inode) \ 4542 do { \ 4543 if (f2fs_has_inline_xattr(inode)) \ 4544 (atomic_dec(&F2FS_I_SB(inode)->inline_xattr)); \ 4545 } while (0) 4546 #define stat_inc_inline_inode(inode) \ 4547 do { \ 4548 if (f2fs_has_inline_data(inode)) \ 4549 (atomic_inc(&F2FS_I_SB(inode)->inline_inode)); \ 4550 } while (0) 4551 #define stat_dec_inline_inode(inode) \ 4552 do { \ 4553 if (f2fs_has_inline_data(inode)) \ 4554 (atomic_dec(&F2FS_I_SB(inode)->inline_inode)); \ 4555 } while (0) 4556 #define stat_inc_inline_dir(inode) \ 4557 do { \ 4558 if (f2fs_has_inline_dentry(inode)) \ 4559 (atomic_inc(&F2FS_I_SB(inode)->inline_dir)); \ 4560 } while (0) 4561 #define stat_dec_inline_dir(inode) \ 4562 do { \ 4563 if (f2fs_has_inline_dentry(inode)) \ 4564 (atomic_dec(&F2FS_I_SB(inode)->inline_dir)); \ 4565 } while (0) 4566 #define stat_inc_compr_inode(inode) \ 4567 do { \ 4568 if (f2fs_compressed_file(inode)) \ 4569 (atomic_inc(&F2FS_I_SB(inode)->compr_inode)); \ 4570 } while (0) 4571 #define stat_dec_compr_inode(inode) \ 4572 do { \ 4573 if (f2fs_compressed_file(inode)) \ 4574 (atomic_dec(&F2FS_I_SB(inode)->compr_inode)); \ 4575 } while (0) 4576 #define stat_add_compr_blocks(inode, blocks) \ 4577 (atomic64_add(blocks, &F2FS_I_SB(inode)->compr_blocks)) 4578 #define stat_sub_compr_blocks(inode, blocks) \ 4579 (atomic64_sub(blocks, &F2FS_I_SB(inode)->compr_blocks)) 4580 #define stat_inc_swapfile_inode(inode) \ 4581 (atomic_inc(&F2FS_I_SB(inode)->swapfile_inode)) 4582 #define stat_dec_swapfile_inode(inode) \ 4583 (atomic_dec(&F2FS_I_SB(inode)->swapfile_inode)) 4584 #define stat_inc_atomic_inode(inode) \ 4585 (atomic_inc(&F2FS_I_SB(inode)->atomic_files)) 4586 #define stat_dec_atomic_inode(inode) \ 4587 (atomic_dec(&F2FS_I_SB(inode)->atomic_files)) 4588 #define stat_inc_meta_count(sbi, blkaddr) \ 4589 do { \ 4590 if (blkaddr < SIT_I(sbi)->sit_base_addr) \ 4591 atomic_inc(&(sbi)->meta_count[META_CP]); \ 4592 else if (blkaddr < NM_I(sbi)->nat_blkaddr) \ 4593 atomic_inc(&(sbi)->meta_count[META_SIT]); \ 4594 else if (blkaddr < SM_I(sbi)->ssa_blkaddr) \ 4595 atomic_inc(&(sbi)->meta_count[META_NAT]); \ 4596 else if (blkaddr < SM_I(sbi)->main_blkaddr) \ 4597 atomic_inc(&(sbi)->meta_count[META_SSA]); \ 4598 } while (0) 4599 #define stat_inc_seg_type(sbi, curseg) \ 4600 ((sbi)->segment_count[(curseg)->alloc_type]++) 4601 #define stat_inc_block_count(sbi, curseg) \ 4602 ((sbi)->block_count[(curseg)->alloc_type]++) 4603 #define stat_inc_inplace_blocks(sbi) \ 4604 (atomic_inc(&(sbi)->inplace_count)) 4605 #define stat_update_max_atomic_write(inode) \ 4606 do { \ 4607 int cur = atomic_read(&F2FS_I_SB(inode)->atomic_files); \ 4608 int max = atomic_read(&F2FS_I_SB(inode)->max_aw_cnt); \ 4609 if (cur > max) \ 4610 atomic_set(&F2FS_I_SB(inode)->max_aw_cnt, cur); \ 4611 } while (0) 4612 #define stat_inc_gc_call_count(sbi, foreground) \ 4613 (F2FS_STAT(sbi)->gc_call_count[(foreground)]++) 4614 #define stat_inc_gc_sec_count(sbi, type, gc_type) \ 4615 (F2FS_STAT(sbi)->gc_secs[(type)][(gc_type)]++) 4616 #define stat_inc_gc_seg_count(sbi, type, gc_type) \ 4617 (F2FS_STAT(sbi)->gc_segs[(type)][(gc_type)]++) 4618 4619 #define stat_inc_tot_blk_count(si, blks) \ 4620 ((si)->tot_blks += (blks)) 4621 4622 #define stat_inc_data_blk_count(sbi, blks, gc_type) \ 4623 do { \ 4624 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 4625 stat_inc_tot_blk_count(si, blks); \ 4626 si->data_blks += (blks); \ 4627 si->bg_data_blks += ((gc_type) == BG_GC) ? (blks) : 0; \ 4628 } while (0) 4629 4630 #define stat_inc_node_blk_count(sbi, blks, gc_type) \ 4631 do { \ 4632 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 4633 stat_inc_tot_blk_count(si, blks); \ 4634 si->node_blks += (blks); \ 4635 si->bg_node_blks += ((gc_type) == BG_GC) ? (blks) : 0; \ 4636 } while (0) 4637 4638 #define stat_inc_defrag_blk_count(sbi, blks) \ 4639 (F2FS_STAT(sbi)->defrag_blks += (blks)) 4640 4641 int f2fs_build_stats(struct f2fs_sb_info *sbi); 4642 void f2fs_destroy_stats(struct f2fs_sb_info *sbi); 4643 void __init f2fs_create_root_stats(void); 4644 void f2fs_destroy_root_stats(void); 4645 void f2fs_update_sit_info(struct f2fs_sb_info *sbi); 4646 #else 4647 #define stat_inc_cp_call_count(sbi, foreground) do { } while (0) 4648 #define stat_inc_cp_count(sbi) do { } while (0) 4649 #define stat_io_skip_bggc_count(sbi) do { } while (0) 4650 #define stat_other_skip_bggc_count(sbi) do { } while (0) 4651 #define stat_inc_dirty_inode(sbi, type) do { } while (0) 4652 #define stat_dec_dirty_inode(sbi, type) do { } while (0) 4653 #define stat_inc_total_hit(sbi, type) do { } while (0) 4654 #define stat_inc_rbtree_node_hit(sbi, type) do { } while (0) 4655 #define stat_inc_largest_node_hit(sbi) do { } while (0) 4656 #define stat_inc_cached_node_hit(sbi, type) do { } while (0) 4657 #define stat_inc_inline_xattr(inode) do { } while (0) 4658 #define stat_dec_inline_xattr(inode) do { } while (0) 4659 #define stat_inc_inline_inode(inode) do { } while (0) 4660 #define stat_dec_inline_inode(inode) do { } while (0) 4661 #define stat_inc_inline_dir(inode) do { } while (0) 4662 #define stat_dec_inline_dir(inode) do { } while (0) 4663 #define stat_inc_compr_inode(inode) do { } while (0) 4664 #define stat_dec_compr_inode(inode) do { } while (0) 4665 #define stat_add_compr_blocks(inode, blocks) do { } while (0) 4666 #define stat_sub_compr_blocks(inode, blocks) do { } while (0) 4667 #define stat_inc_swapfile_inode(inode) do { } while (0) 4668 #define stat_dec_swapfile_inode(inode) do { } while (0) 4669 #define stat_inc_atomic_inode(inode) do { } while (0) 4670 #define stat_dec_atomic_inode(inode) do { } while (0) 4671 #define stat_update_max_atomic_write(inode) do { } while (0) 4672 #define stat_inc_meta_count(sbi, blkaddr) do { } while (0) 4673 #define stat_inc_seg_type(sbi, curseg) do { } while (0) 4674 #define stat_inc_block_count(sbi, curseg) do { } while (0) 4675 #define stat_inc_inplace_blocks(sbi) do { } while (0) 4676 #define stat_inc_gc_call_count(sbi, foreground) do { } while (0) 4677 #define stat_inc_gc_sec_count(sbi, type, gc_type) do { } while (0) 4678 #define stat_inc_gc_seg_count(sbi, type, gc_type) do { } while (0) 4679 #define stat_inc_tot_blk_count(si, blks) do { } while (0) 4680 #define stat_inc_data_blk_count(sbi, blks, gc_type) do { } while (0) 4681 #define stat_inc_node_blk_count(sbi, blks, gc_type) do { } while (0) 4682 #define stat_inc_defrag_blk_count(sbi, blks) do { } while (0) 4683 4684 static inline int f2fs_build_stats(struct f2fs_sb_info *sbi) { return 0; } 4685 static inline void f2fs_destroy_stats(struct f2fs_sb_info *sbi) { } 4686 static inline void __init f2fs_create_root_stats(void) { } 4687 static inline void f2fs_destroy_root_stats(void) { } 4688 static inline void f2fs_update_sit_info(struct f2fs_sb_info *sbi) {} 4689 #endif 4690 4691 extern const struct file_operations f2fs_dir_operations; 4692 extern const struct file_operations f2fs_file_operations; 4693 extern const struct inode_operations f2fs_file_inode_operations; 4694 extern const struct address_space_operations f2fs_dblock_aops; 4695 extern const struct inode_operations f2fs_dir_inode_operations; 4696 extern const struct inode_operations f2fs_symlink_inode_operations; 4697 extern const struct inode_operations f2fs_encrypted_symlink_inode_operations; 4698 extern const struct inode_operations f2fs_special_inode_operations; 4699 extern struct kmem_cache *f2fs_inode_entry_slab; 4700 4701 /* 4702 * inline.c 4703 */ 4704 bool f2fs_may_inline_data(struct inode *inode); 4705 bool f2fs_sanity_check_inline_data(struct inode *inode, struct f2fs_cached_block *ientry); 4706 bool f2fs_may_inline_dentry(struct inode *inode); 4707 void f2fs_do_read_inline_data(struct folio *folio, struct f2fs_cached_block *ientry); 4708 void f2fs_truncate_inline_inode(struct inode *inode, struct f2fs_cached_block *ientry, 4709 u64 from); 4710 int f2fs_read_inline_data(struct inode *inode, struct folio *folio); 4711 int f2fs_convert_inline_folio(struct dnode_of_data *dn, struct folio *folio); 4712 int f2fs_convert_inline_inode(struct inode *inode); 4713 int f2fs_try_convert_inline_dir(struct inode *dir, struct dentry *dentry); 4714 int f2fs_write_inline_data(struct inode *inode, struct folio *folio); 4715 int f2fs_recover_inline_data(struct inode *inode, struct f2fs_cached_block *entry); 4716 struct f2fs_dir_entry *f2fs_find_in_inline_dir(struct inode *dir, 4717 const struct f2fs_filename *fname, void **dentry_block, 4718 bool use_hash); 4719 int f2fs_make_empty_inline_dir(struct inode *inode, struct inode *parent, 4720 struct f2fs_cached_block *ientry); 4721 int f2fs_add_inline_entry(struct inode *dir, const struct f2fs_filename *fname, 4722 struct inode *inode, nid_t ino, umode_t mode); 4723 void f2fs_delete_inline_entry(struct f2fs_dir_entry *dentry, 4724 struct f2fs_cached_block *ientry, struct inode *dir, 4725 struct inode *inode); 4726 bool f2fs_empty_inline_dir(struct inode *dir); 4727 int f2fs_read_inline_dir(struct file *file, struct dir_context *ctx, 4728 struct fscrypt_str *fstr); 4729 int f2fs_inline_data_fiemap(struct inode *inode, 4730 struct fiemap_extent_info *fieinfo, 4731 __u64 start, __u64 len); 4732 4733 /* 4734 * shrinker.c 4735 */ 4736 unsigned long f2fs_shrink_count(struct shrinker *shrink, 4737 struct shrink_control *sc); 4738 unsigned long f2fs_shrink_scan(struct shrinker *shrink, 4739 struct shrink_control *sc); 4740 unsigned int f2fs_donate_files(void); 4741 void f2fs_reclaim_caches(unsigned int reclaim_caches_kb); 4742 void f2fs_join_shrinker(struct f2fs_sb_info *sbi); 4743 void f2fs_leave_shrinker(struct f2fs_sb_info *sbi); 4744 4745 /* 4746 * extent_cache.c 4747 */ 4748 bool sanity_check_extent_cache(struct inode *inode, struct f2fs_cached_block *ientry); 4749 void f2fs_init_extent_tree(struct inode *inode); 4750 void f2fs_drop_extent_tree(struct inode *inode); 4751 void f2fs_destroy_extent_node(struct inode *inode); 4752 void f2fs_destroy_extent_tree(struct inode *inode); 4753 void f2fs_init_extent_cache_info(struct f2fs_sb_info *sbi); 4754 int __init f2fs_create_extent_cache(void); 4755 void f2fs_destroy_extent_cache(void); 4756 4757 /* read extent cache ops */ 4758 void f2fs_init_read_extent_tree(struct inode *inode, struct f2fs_cached_block *ientry); 4759 bool f2fs_lookup_read_extent_cache(struct inode *inode, pgoff_t pgofs, 4760 struct extent_info *ei); 4761 bool f2fs_lookup_read_extent_cache_block(struct inode *inode, pgoff_t index, 4762 block_t *blkaddr); 4763 void f2fs_update_read_extent_cache(struct dnode_of_data *dn); 4764 void f2fs_update_read_extent_cache_range(struct dnode_of_data *dn, 4765 pgoff_t fofs, block_t blkaddr, unsigned int len); 4766 unsigned int f2fs_shrink_read_extent_tree(struct f2fs_sb_info *sbi, 4767 int nr_shrink); 4768 4769 /* block age extent cache ops */ 4770 void f2fs_init_age_extent_tree(struct inode *inode); 4771 bool f2fs_lookup_age_extent_cache(struct inode *inode, pgoff_t pgofs, 4772 struct extent_info *ei); 4773 void f2fs_update_age_extent_cache(struct dnode_of_data *dn); 4774 void f2fs_update_age_extent_cache_range(struct dnode_of_data *dn, 4775 pgoff_t fofs, unsigned int len); 4776 unsigned int f2fs_shrink_age_extent_tree(struct f2fs_sb_info *sbi, 4777 int nr_shrink); 4778 4779 /* 4780 * sysfs.c 4781 */ 4782 #define MIN_RA_MUL 2 4783 #define MAX_RA_MUL 256 4784 4785 int __init f2fs_init_sysfs(void); 4786 void f2fs_exit_sysfs(void); 4787 int f2fs_register_sysfs(struct f2fs_sb_info *sbi); 4788 void f2fs_unregister_sysfs(struct f2fs_sb_info *sbi); 4789 4790 /* verity.c */ 4791 extern const struct fsverity_operations f2fs_verityops; 4792 4793 /* 4794 * crypto support 4795 */ 4796 static inline bool f2fs_encrypted_file(struct inode *inode) 4797 { 4798 return IS_ENCRYPTED(inode) && S_ISREG(inode->i_mode); 4799 } 4800 4801 static inline void f2fs_set_encrypted_inode(struct inode *inode) 4802 { 4803 #ifdef CONFIG_FS_ENCRYPTION 4804 file_set_encrypt(inode); 4805 f2fs_set_inode_flags(inode); 4806 #endif 4807 } 4808 4809 /* 4810 * Returns true if the reads of the inode's data need to undergo some 4811 * postprocessing step, like decryption or authenticity verification. 4812 */ 4813 static inline bool f2fs_post_read_required(struct inode *inode) 4814 { 4815 return f2fs_encrypted_file(inode) || fsverity_active(inode) || 4816 f2fs_compressed_file(inode); 4817 } 4818 4819 static inline bool f2fs_used_in_atomic_write(struct inode *inode) 4820 { 4821 return f2fs_is_atomic_file(inode) || f2fs_is_cow_file(inode); 4822 } 4823 4824 static inline bool f2fs_meta_inode_gc_required(struct inode *inode) 4825 { 4826 return f2fs_post_read_required(inode) || f2fs_used_in_atomic_write(inode); 4827 } 4828 4829 /* 4830 * compress.c 4831 */ 4832 #ifdef CONFIG_F2FS_FS_COMPRESSION 4833 enum cluster_check_type { 4834 CLUSTER_IS_COMPR, /* check only if compressed cluster */ 4835 CLUSTER_COMPR_BLKS, /* return # of compressed blocks in a cluster */ 4836 CLUSTER_RAW_BLKS /* return # of raw blocks in a cluster */ 4837 }; 4838 bool f2fs_is_compressed_page(struct folio *folio); 4839 struct folio *f2fs_compress_control_folio(struct folio *folio); 4840 int f2fs_prepare_compress_overwrite(struct inode *inode, 4841 struct page **pagep, pgoff_t index, void **fsdata); 4842 bool f2fs_compress_write_end(struct inode *inode, void *fsdata, 4843 pgoff_t index, unsigned copied); 4844 int f2fs_truncate_partial_cluster(struct inode *inode, u64 from, bool lock); 4845 void f2fs_compress_write_end_io(struct bio *bio, struct folio *folio); 4846 bool f2fs_is_compress_backend_ready(struct inode *inode); 4847 bool f2fs_is_compress_level_valid(int alg, int lvl); 4848 int __init f2fs_init_compress_mempool(void); 4849 void f2fs_destroy_compress_mempool(void); 4850 void f2fs_decompress_cluster(struct decompress_io_ctx *dic, bool in_task); 4851 void f2fs_end_read_compressed_page(struct folio *folio, bool failed, 4852 block_t blkaddr, bool in_task); 4853 bool f2fs_cluster_is_empty(struct compress_ctx *cc); 4854 bool f2fs_cluster_can_merge_page(struct compress_ctx *cc, pgoff_t index); 4855 bool f2fs_all_cluster_page_ready(struct compress_ctx *cc, struct page **pages, 4856 int index, int nr_pages, bool uptodate); 4857 bool f2fs_sanity_check_cluster(struct dnode_of_data *dn); 4858 void f2fs_compress_ctx_add_page(struct compress_ctx *cc, struct folio *folio); 4859 int f2fs_write_multi_pages(struct compress_ctx *cc, 4860 int *submitted, 4861 struct writeback_control *wbc, 4862 enum iostat_type io_type); 4863 int f2fs_is_compressed_cluster(struct inode *inode, pgoff_t index); 4864 bool f2fs_is_sparse_cluster(struct inode *inode, pgoff_t index); 4865 void f2fs_update_read_extent_tree_range_compressed(struct inode *inode, 4866 pgoff_t fofs, block_t blkaddr, 4867 unsigned int llen, unsigned int c_len); 4868 int f2fs_read_multi_pages(struct compress_ctx *cc, struct bio **bio_ret, 4869 unsigned nr_pages, sector_t *last_block_in_bio, 4870 struct readahead_control *rac, bool for_write); 4871 struct decompress_io_ctx *f2fs_alloc_dic(struct compress_ctx *cc); 4872 void f2fs_decompress_end_io(struct decompress_io_ctx *dic, bool failed, 4873 bool in_task); 4874 void f2fs_put_folio_dic(struct folio *folio, bool in_task); 4875 unsigned int f2fs_cluster_blocks_are_contiguous(struct dnode_of_data *dn, 4876 unsigned int ofs_in_node); 4877 int f2fs_init_compress_ctx(struct compress_ctx *cc); 4878 void f2fs_destroy_compress_ctx(struct compress_ctx *cc, bool reuse); 4879 void f2fs_init_compress_info(struct f2fs_sb_info *sbi); 4880 void f2fs_init_compress_cache_context(struct f2fs_sb_info *sbi); 4881 int f2fs_init_page_array_cache(struct f2fs_sb_info *sbi); 4882 void f2fs_destroy_page_array_cache(struct f2fs_sb_info *sbi); 4883 int __init f2fs_init_compress_cache(void); 4884 void f2fs_destroy_compress_cache(void); 4885 void f2fs_invalidate_compress_pages_range(struct f2fs_sb_info *sbi, 4886 block_t blkaddr, unsigned int len); 4887 bool f2fs_load_compressed_folio(struct f2fs_sb_info *sbi, struct folio *folio, 4888 block_t blkaddr); 4889 void f2fs_invalidate_compress_pages(struct f2fs_sb_info *sbi, nid_t ino); 4890 #define inc_compr_inode_stat(inode) \ 4891 do { \ 4892 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); \ 4893 sbi->compr_new_inode++; \ 4894 } while (0) 4895 #define add_compr_block_stat(inode, blocks) \ 4896 do { \ 4897 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); \ 4898 int diff = F2FS_I(inode)->i_cluster_size - blocks; \ 4899 sbi->compr_written_block += blocks; \ 4900 sbi->compr_saved_block += diff; \ 4901 } while (0) 4902 #else 4903 static inline bool f2fs_is_compressed_page(struct folio *folio) { return false; } 4904 static inline bool f2fs_is_compress_backend_ready(struct inode *inode) 4905 { 4906 if (!f2fs_compressed_file(inode)) 4907 return true; 4908 /* not support compression */ 4909 return false; 4910 } 4911 static inline bool f2fs_is_compress_level_valid(int alg, int lvl) { return false; } 4912 static inline struct folio *f2fs_compress_control_folio(struct folio *folio) 4913 { 4914 WARN_ON_ONCE(1); 4915 return ERR_PTR(-EINVAL); 4916 } 4917 static inline int __init f2fs_init_compress_mempool(void) { return 0; } 4918 static inline void f2fs_destroy_compress_mempool(void) { } 4919 static inline void f2fs_decompress_cluster(struct decompress_io_ctx *dic, 4920 bool in_task) { } 4921 static inline void f2fs_end_read_compressed_page(struct folio *folio, 4922 bool failed, block_t blkaddr, bool in_task) 4923 { 4924 WARN_ON_ONCE(1); 4925 } 4926 static inline void f2fs_put_folio_dic(struct folio *folio, bool in_task) 4927 { 4928 WARN_ON_ONCE(1); 4929 } 4930 static inline unsigned int f2fs_cluster_blocks_are_contiguous( 4931 struct dnode_of_data *dn, unsigned int ofs_in_node) { return 0; } 4932 static inline bool f2fs_sanity_check_cluster(struct dnode_of_data *dn) { return false; } 4933 static inline void f2fs_init_compress_cache_context(struct f2fs_sb_info *sbi) { } 4934 static inline int f2fs_init_page_array_cache(struct f2fs_sb_info *sbi) { return 0; } 4935 static inline void f2fs_destroy_page_array_cache(struct f2fs_sb_info *sbi) { } 4936 static inline int __init f2fs_init_compress_cache(void) { return 0; } 4937 static inline void f2fs_destroy_compress_cache(void) { } 4938 static inline void f2fs_invalidate_compress_pages_range(struct f2fs_sb_info *sbi, 4939 block_t blkaddr, unsigned int len) { } 4940 static inline bool f2fs_load_compressed_folio(struct f2fs_sb_info *sbi, 4941 struct folio *folio, block_t blkaddr) { return false; } 4942 static inline void f2fs_invalidate_compress_pages(struct f2fs_sb_info *sbi, 4943 nid_t ino) { } 4944 #define inc_compr_inode_stat(inode) do { } while (0) 4945 static inline int f2fs_is_compressed_cluster( 4946 struct inode *inode, 4947 pgoff_t index) { return 0; } 4948 static inline bool f2fs_is_sparse_cluster( 4949 struct inode *inode, 4950 pgoff_t index) { return true; } 4951 static inline void f2fs_update_read_extent_tree_range_compressed( 4952 struct inode *inode, 4953 pgoff_t fofs, block_t blkaddr, 4954 unsigned int llen, unsigned int c_len) { } 4955 #endif 4956 4957 static inline int set_compress_context(struct inode *inode) 4958 { 4959 #ifdef CONFIG_F2FS_FS_COMPRESSION 4960 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 4961 struct f2fs_inode_info *fi = F2FS_I(inode); 4962 4963 fi->i_compress_algorithm = F2FS_OPTION(sbi).compress_algorithm; 4964 fi->i_log_cluster_size = F2FS_OPTION(sbi).compress_log_size; 4965 fi->i_compress_flag = F2FS_OPTION(sbi).compress_chksum ? 4966 BIT(COMPRESS_CHKSUM) : 0; 4967 fi->i_cluster_size = BIT(fi->i_log_cluster_size); 4968 if ((fi->i_compress_algorithm == COMPRESS_LZ4 || 4969 fi->i_compress_algorithm == COMPRESS_ZSTD) && 4970 F2FS_OPTION(sbi).compress_level) 4971 fi->i_compress_level = F2FS_OPTION(sbi).compress_level; 4972 fi->i_flags |= F2FS_COMPR_FL; 4973 set_inode_flag(inode, FI_COMPRESSED_FILE); 4974 stat_inc_compr_inode(inode); 4975 inc_compr_inode_stat(inode); 4976 f2fs_mark_inode_dirty_sync(inode, true); 4977 return 0; 4978 #else 4979 return -EOPNOTSUPP; 4980 #endif 4981 } 4982 4983 static inline bool f2fs_disable_compressed_file(struct inode *inode) 4984 { 4985 struct f2fs_inode_info *fi = F2FS_I(inode); 4986 4987 f2fs_down_write(&fi->i_sem); 4988 4989 if (!f2fs_compressed_file(inode)) { 4990 f2fs_up_write(&fi->i_sem); 4991 return true; 4992 } 4993 if (f2fs_is_mmap_file(inode) || atomic_read(&fi->writeback) || 4994 (S_ISREG(inode->i_mode) && F2FS_HAS_BLOCKS(inode))) { 4995 f2fs_up_write(&fi->i_sem); 4996 return false; 4997 } 4998 4999 fi->i_flags &= ~F2FS_COMPR_FL; 5000 stat_dec_compr_inode(inode); 5001 clear_inode_flag(inode, FI_COMPRESSED_FILE); 5002 f2fs_mark_inode_dirty_sync(inode, true); 5003 5004 f2fs_up_write(&fi->i_sem); 5005 return true; 5006 } 5007 5008 #define F2FS_FEATURE_FUNCS(name, flagname) \ 5009 static inline bool f2fs_sb_has_##name(struct f2fs_sb_info *sbi) \ 5010 { \ 5011 return F2FS_HAS_FEATURE(sbi, F2FS_FEATURE_##flagname); \ 5012 } 5013 5014 F2FS_FEATURE_FUNCS(encrypt, ENCRYPT); 5015 F2FS_FEATURE_FUNCS(blkzoned, BLKZONED); 5016 F2FS_FEATURE_FUNCS(extra_attr, EXTRA_ATTR); 5017 F2FS_FEATURE_FUNCS(project_quota, PRJQUOTA); 5018 F2FS_FEATURE_FUNCS(inode_chksum, INODE_CHKSUM); 5019 F2FS_FEATURE_FUNCS(flexible_inline_xattr, FLEXIBLE_INLINE_XATTR); 5020 F2FS_FEATURE_FUNCS(quota_ino, QUOTA_INO); 5021 F2FS_FEATURE_FUNCS(inode_crtime, INODE_CRTIME); 5022 F2FS_FEATURE_FUNCS(lost_found, LOST_FOUND); 5023 F2FS_FEATURE_FUNCS(verity, VERITY); 5024 F2FS_FEATURE_FUNCS(sb_chksum, SB_CHKSUM); 5025 F2FS_FEATURE_FUNCS(casefold, CASEFOLD); 5026 F2FS_FEATURE_FUNCS(compression, COMPRESSION); 5027 F2FS_FEATURE_FUNCS(readonly, RO); 5028 F2FS_FEATURE_FUNCS(device_alias, DEVICE_ALIAS); 5029 F2FS_FEATURE_FUNCS(packed_ssa, PACKED_SSA); 5030 5031 #ifdef CONFIG_BLK_DEV_ZONED 5032 static inline bool f2fs_zone_is_seq(struct f2fs_sb_info *sbi, int devi, 5033 unsigned int zone) 5034 { 5035 return test_bit(zone, FDEV(devi).blkz_seq); 5036 } 5037 5038 static inline bool f2fs_blkz_is_seq(struct f2fs_sb_info *sbi, int devi, 5039 block_t blkaddr) 5040 { 5041 return f2fs_zone_is_seq(sbi, devi, blkaddr / sbi->blocks_per_blkz); 5042 } 5043 #endif 5044 5045 static inline int f2fs_bdev_index(struct f2fs_sb_info *sbi, 5046 struct block_device *bdev) 5047 { 5048 int i; 5049 5050 if (!f2fs_is_multi_device(sbi)) 5051 return 0; 5052 5053 for (i = 0; i < sbi->s_ndevs; i++) 5054 if (FDEV(i).bdev == bdev) 5055 return i; 5056 5057 WARN_ON(1); 5058 return -1; 5059 } 5060 5061 static inline bool f2fs_hw_should_discard(struct f2fs_sb_info *sbi) 5062 { 5063 return f2fs_sb_has_blkzoned(sbi); 5064 } 5065 5066 static inline bool f2fs_bdev_support_discard(struct block_device *bdev) 5067 { 5068 return bdev_max_discard_sectors(bdev) || bdev_is_zoned(bdev); 5069 } 5070 5071 static inline bool f2fs_hw_support_discard(struct f2fs_sb_info *sbi) 5072 { 5073 int i; 5074 5075 if (!f2fs_is_multi_device(sbi)) 5076 return f2fs_bdev_support_discard(sbi->sb->s_bdev); 5077 5078 for (i = 0; i < sbi->s_ndevs; i++) 5079 if (f2fs_bdev_support_discard(FDEV(i).bdev)) 5080 return true; 5081 return false; 5082 } 5083 5084 static inline unsigned int f2fs_hw_discard_granularity(struct f2fs_sb_info *sbi) 5085 { 5086 int i = 1; 5087 unsigned int discard_granularity = bdev_discard_granularity(sbi->sb->s_bdev); 5088 5089 if (f2fs_is_multi_device(sbi)) 5090 for (; i < sbi->s_ndevs && !bdev_is_zoned(FDEV(i).bdev); i++) 5091 discard_granularity = max_t(unsigned int, discard_granularity, 5092 bdev_discard_granularity(FDEV(i).bdev)); 5093 return discard_granularity; 5094 } 5095 5096 static inline bool f2fs_realtime_discard_enable(struct f2fs_sb_info *sbi) 5097 { 5098 return (test_opt(sbi, DISCARD) && f2fs_hw_support_discard(sbi)) || 5099 f2fs_hw_should_discard(sbi); 5100 } 5101 5102 static inline bool f2fs_hw_is_readonly(struct f2fs_sb_info *sbi) 5103 { 5104 int i; 5105 5106 if (!f2fs_is_multi_device(sbi)) 5107 return bdev_read_only(sbi->sb->s_bdev); 5108 5109 for (i = 0; i < sbi->s_ndevs; i++) 5110 if (bdev_read_only(FDEV(i).bdev)) 5111 return true; 5112 return false; 5113 } 5114 5115 static inline bool f2fs_dev_is_readonly(struct f2fs_sb_info *sbi) 5116 { 5117 return f2fs_sb_has_readonly(sbi) || f2fs_hw_is_readonly(sbi); 5118 } 5119 5120 static inline bool f2fs_lfs_mode(struct f2fs_sb_info *sbi) 5121 { 5122 return F2FS_OPTION(sbi).fs_mode == FS_MODE_LFS; 5123 } 5124 5125 static inline bool f2fs_is_sequential_zone_area(struct f2fs_sb_info *sbi, 5126 block_t blkaddr) 5127 { 5128 if (f2fs_sb_has_blkzoned(sbi)) { 5129 #ifdef CONFIG_BLK_DEV_ZONED 5130 int devi = f2fs_target_device_index(sbi, blkaddr); 5131 5132 if (!bdev_is_zoned(FDEV(devi).bdev)) 5133 return false; 5134 5135 if (f2fs_is_multi_device(sbi)) { 5136 if (blkaddr < FDEV(devi).start_blk || 5137 blkaddr > FDEV(devi).end_blk) { 5138 f2fs_err(sbi, "Invalid block %x", blkaddr); 5139 return false; 5140 } 5141 blkaddr -= FDEV(devi).start_blk; 5142 } 5143 5144 return f2fs_blkz_is_seq(sbi, devi, blkaddr); 5145 #else 5146 return false; 5147 #endif 5148 } 5149 return false; 5150 } 5151 5152 static inline bool f2fs_low_mem_mode(struct f2fs_sb_info *sbi) 5153 { 5154 return F2FS_OPTION(sbi).memory_mode == MEMORY_MODE_LOW; 5155 } 5156 5157 static inline bool f2fs_may_compress(struct inode *inode) 5158 { 5159 if (IS_SWAPFILE(inode) || f2fs_is_pinned_file(inode) || 5160 f2fs_is_atomic_file(inode) || f2fs_has_inline_data(inode) || 5161 f2fs_is_mmap_file(inode)) 5162 return false; 5163 return S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode); 5164 } 5165 5166 static inline void f2fs_i_compr_blocks_update(struct inode *inode, 5167 u64 blocks, bool add) 5168 { 5169 struct f2fs_inode_info *fi = F2FS_I(inode); 5170 int diff = fi->i_cluster_size - blocks; 5171 5172 /* don't update i_compr_blocks if saved blocks were released */ 5173 if (!add && !atomic_read(&fi->i_compr_blocks)) 5174 return; 5175 5176 if (add) { 5177 atomic_add(diff, &fi->i_compr_blocks); 5178 stat_add_compr_blocks(inode, diff); 5179 } else { 5180 atomic_sub(diff, &fi->i_compr_blocks); 5181 stat_sub_compr_blocks(inode, diff); 5182 } 5183 f2fs_mark_inode_dirty_sync(inode, true); 5184 } 5185 5186 static inline bool f2fs_allow_multi_device_dio(struct f2fs_sb_info *sbi, 5187 int flag) 5188 { 5189 if (!f2fs_is_multi_device(sbi)) 5190 return false; 5191 if (flag != F2FS_GET_BLOCK_DIO) 5192 return false; 5193 return sbi->aligned_blksize; 5194 } 5195 5196 #ifdef CONFIG_F2FS_FAULT_INJECTION 5197 extern int f2fs_build_fault_attr(struct f2fs_sb_info *sbi, unsigned long rate, 5198 unsigned long type, enum fault_option fo); 5199 extern void f2fs_simulate_lock_timeout(struct f2fs_sb_info *sbi); 5200 #else 5201 static inline int f2fs_build_fault_attr(struct f2fs_sb_info *sbi, 5202 unsigned long rate, unsigned long type, 5203 enum fault_option fo) 5204 { 5205 return 0; 5206 } 5207 static inline void f2fs_simulate_lock_timeout(struct f2fs_sb_info *sbi) 5208 { 5209 return; 5210 } 5211 #endif 5212 5213 static inline bool is_journalled_quota(struct f2fs_sb_info *sbi) 5214 { 5215 #ifdef CONFIG_QUOTA 5216 if (f2fs_sb_has_quota_ino(sbi)) 5217 return true; 5218 if (F2FS_OPTION(sbi).s_qf_names[USRQUOTA] || 5219 F2FS_OPTION(sbi).s_qf_names[GRPQUOTA] || 5220 F2FS_OPTION(sbi).s_qf_names[PRJQUOTA]) 5221 return true; 5222 #endif 5223 return false; 5224 } 5225 5226 static inline bool f2fs_quota_file(struct f2fs_sb_info *sbi, nid_t ino) 5227 { 5228 #ifdef CONFIG_QUOTA 5229 int i; 5230 5231 if (!f2fs_sb_has_quota_ino(sbi)) 5232 return false; 5233 5234 for (i = 0; i < MAXQUOTAS; i++) { 5235 if (f2fs_qf_ino(sbi->sb, i) == ino) 5236 return true; 5237 } 5238 #endif 5239 return false; 5240 } 5241 5242 static inline bool f2fs_block_unit_discard(struct f2fs_sb_info *sbi) 5243 { 5244 return F2FS_OPTION(sbi).discard_unit == DISCARD_UNIT_BLOCK; 5245 } 5246 5247 static inline void __f2fs_schedule_timeout(long timeout, bool io) 5248 { 5249 set_current_state(TASK_UNINTERRUPTIBLE); 5250 if (io) 5251 io_schedule_timeout(timeout); 5252 else 5253 schedule_timeout(timeout); 5254 } 5255 5256 #define f2fs_io_schedule_timeout(timeout) \ 5257 __f2fs_schedule_timeout(timeout, true) 5258 #define f2fs_schedule_timeout(timeout) \ 5259 __f2fs_schedule_timeout(timeout, false) 5260 5261 static inline void f2fs_schedule_timeout_killable(long timeout, bool io) 5262 { 5263 unsigned long last_time = jiffies + timeout; 5264 5265 while (jiffies < last_time) { 5266 if (fatal_signal_pending(current)) 5267 return; 5268 __f2fs_schedule_timeout(DEFAULT_SCHEDULE_TIMEOUT, io); 5269 } 5270 } 5271 5272 static inline void f2fs_handle_page_eio(struct f2fs_sb_info *sbi, 5273 pgoff_t ofs, enum page_type type) 5274 { 5275 if (unlikely(f2fs_cp_error(sbi))) 5276 return; 5277 5278 if (ofs == sbi->page_eio_ofs[type]) { 5279 if (sbi->page_eio_cnt[type]++ == MAX_RETRY_PAGE_EIO) { 5280 enum stop_cp_reason stop_reason; 5281 5282 switch (type) { 5283 case META: 5284 stop_reason = STOP_CP_REASON_READ_META; 5285 break; 5286 case NODE: 5287 stop_reason = STOP_CP_REASON_READ_NODE; 5288 break; 5289 case DATA: 5290 stop_reason = STOP_CP_REASON_READ_DATA; 5291 break; 5292 default: 5293 f2fs_bug_on(sbi, 1); 5294 return; 5295 } 5296 f2fs_stop_checkpoint(sbi, false, stop_reason); 5297 } 5298 } else { 5299 sbi->page_eio_ofs[type] = ofs; 5300 sbi->page_eio_cnt[type] = 0; 5301 } 5302 } 5303 5304 static inline bool f2fs_is_readonly(struct f2fs_sb_info *sbi) 5305 { 5306 return f2fs_sb_has_readonly(sbi) || f2fs_readonly(sbi->sb); 5307 } 5308 5309 static inline void f2fs_invalidate_internal_cache(struct f2fs_sb_info *sbi, 5310 block_t blkaddr, unsigned int len) 5311 { 5312 f2fs_truncate_meta_caches(sbi, blkaddr, len); 5313 f2fs_invalidate_compress_pages_range(sbi, blkaddr, len); 5314 } 5315 5316 #endif /* _LINUX_F2FS_H */ 5317