1 /* SPDX-License-Identifier: GPL-2.0 */ 2 /* 3 * fs/f2fs/f2fs.h 4 * 5 * Copyright (c) 2012 Samsung Electronics Co., Ltd. 6 * http://www.samsung.com/ 7 */ 8 #ifndef _LINUX_F2FS_H 9 #define _LINUX_F2FS_H 10 11 #include <linux/uio.h> 12 #include <linux/types.h> 13 #include <linux/page-flags.h> 14 #include <linux/buffer_head.h> 15 #include <linux/slab.h> 16 #include <linux/crc32.h> 17 #include <linux/magic.h> 18 #include <linux/kobject.h> 19 #include <linux/sched.h> 20 #include <linux/cred.h> 21 #include <linux/sched/mm.h> 22 #include <linux/vmalloc.h> 23 #include <linux/bio.h> 24 #include <linux/blkdev.h> 25 #include <linux/quotaops.h> 26 #include <linux/part_stat.h> 27 #include <crypto/hash.h> 28 29 #include <linux/fscrypt.h> 30 #include <linux/fsverity.h> 31 32 struct pagevec; 33 34 #ifdef CONFIG_F2FS_CHECK_FS 35 #define f2fs_bug_on(sbi, condition) BUG_ON(condition) 36 #else 37 #define f2fs_bug_on(sbi, condition) \ 38 do { \ 39 if (WARN_ON(condition)) \ 40 set_sbi_flag(sbi, SBI_NEED_FSCK); \ 41 } while (0) 42 #endif 43 44 enum { 45 FAULT_KMALLOC, 46 FAULT_KVMALLOC, 47 FAULT_PAGE_ALLOC, 48 FAULT_PAGE_GET, 49 FAULT_ALLOC_BIO, /* it's obsolete due to bio_alloc() will never fail */ 50 FAULT_ALLOC_NID, 51 FAULT_ORPHAN, 52 FAULT_BLOCK, 53 FAULT_DIR_DEPTH, 54 FAULT_EVICT_INODE, 55 FAULT_TRUNCATE, 56 FAULT_READ_IO, 57 FAULT_CHECKPOINT, 58 FAULT_DISCARD, 59 FAULT_WRITE_IO, 60 FAULT_SLAB_ALLOC, 61 FAULT_DQUOT_INIT, 62 FAULT_LOCK_OP, 63 FAULT_BLKADDR, 64 FAULT_MAX, 65 }; 66 67 #ifdef CONFIG_F2FS_FAULT_INJECTION 68 #define F2FS_ALL_FAULT_TYPE ((1 << FAULT_MAX) - 1) 69 70 struct f2fs_fault_info { 71 atomic_t inject_ops; 72 unsigned int inject_rate; 73 unsigned int inject_type; 74 }; 75 76 extern const char *f2fs_fault_name[FAULT_MAX]; 77 #define IS_FAULT_SET(fi, type) ((fi)->inject_type & (1 << (type))) 78 #endif 79 80 /* 81 * For mount options 82 */ 83 #define F2FS_MOUNT_DISABLE_ROLL_FORWARD 0x00000002 84 #define F2FS_MOUNT_DISCARD 0x00000004 85 #define F2FS_MOUNT_NOHEAP 0x00000008 86 #define F2FS_MOUNT_XATTR_USER 0x00000010 87 #define F2FS_MOUNT_POSIX_ACL 0x00000020 88 #define F2FS_MOUNT_DISABLE_EXT_IDENTIFY 0x00000040 89 #define F2FS_MOUNT_INLINE_XATTR 0x00000080 90 #define F2FS_MOUNT_INLINE_DATA 0x00000100 91 #define F2FS_MOUNT_INLINE_DENTRY 0x00000200 92 #define F2FS_MOUNT_FLUSH_MERGE 0x00000400 93 #define F2FS_MOUNT_NOBARRIER 0x00000800 94 #define F2FS_MOUNT_FASTBOOT 0x00001000 95 #define F2FS_MOUNT_READ_EXTENT_CACHE 0x00002000 96 #define F2FS_MOUNT_DATA_FLUSH 0x00008000 97 #define F2FS_MOUNT_FAULT_INJECTION 0x00010000 98 #define F2FS_MOUNT_USRQUOTA 0x00080000 99 #define F2FS_MOUNT_GRPQUOTA 0x00100000 100 #define F2FS_MOUNT_PRJQUOTA 0x00200000 101 #define F2FS_MOUNT_QUOTA 0x00400000 102 #define F2FS_MOUNT_INLINE_XATTR_SIZE 0x00800000 103 #define F2FS_MOUNT_RESERVE_ROOT 0x01000000 104 #define F2FS_MOUNT_DISABLE_CHECKPOINT 0x02000000 105 #define F2FS_MOUNT_NORECOVERY 0x04000000 106 #define F2FS_MOUNT_ATGC 0x08000000 107 #define F2FS_MOUNT_MERGE_CHECKPOINT 0x10000000 108 #define F2FS_MOUNT_GC_MERGE 0x20000000 109 #define F2FS_MOUNT_COMPRESS_CACHE 0x40000000 110 #define F2FS_MOUNT_AGE_EXTENT_CACHE 0x80000000 111 112 #define F2FS_OPTION(sbi) ((sbi)->mount_opt) 113 #define clear_opt(sbi, option) (F2FS_OPTION(sbi).opt &= ~F2FS_MOUNT_##option) 114 #define set_opt(sbi, option) (F2FS_OPTION(sbi).opt |= F2FS_MOUNT_##option) 115 #define test_opt(sbi, option) (F2FS_OPTION(sbi).opt & F2FS_MOUNT_##option) 116 117 #define ver_after(a, b) (typecheck(unsigned long long, a) && \ 118 typecheck(unsigned long long, b) && \ 119 ((long long)((a) - (b)) > 0)) 120 121 typedef u32 block_t; /* 122 * should not change u32, since it is the on-disk block 123 * address format, __le32. 124 */ 125 typedef u32 nid_t; 126 127 #define COMPRESS_EXT_NUM 16 128 129 /* 130 * An implementation of an rwsem that is explicitly unfair to readers. This 131 * prevents priority inversion when a low-priority reader acquires the read lock 132 * while sleeping on the write lock but the write lock is needed by 133 * higher-priority clients. 134 */ 135 136 struct f2fs_rwsem { 137 struct rw_semaphore internal_rwsem; 138 #ifdef CONFIG_F2FS_UNFAIR_RWSEM 139 wait_queue_head_t read_waiters; 140 #endif 141 }; 142 143 struct f2fs_mount_info { 144 unsigned int opt; 145 int write_io_size_bits; /* Write IO size bits */ 146 block_t root_reserved_blocks; /* root reserved blocks */ 147 kuid_t s_resuid; /* reserved blocks for uid */ 148 kgid_t s_resgid; /* reserved blocks for gid */ 149 int active_logs; /* # of active logs */ 150 int inline_xattr_size; /* inline xattr size */ 151 #ifdef CONFIG_F2FS_FAULT_INJECTION 152 struct f2fs_fault_info fault_info; /* For fault injection */ 153 #endif 154 #ifdef CONFIG_QUOTA 155 /* Names of quota files with journalled quota */ 156 char *s_qf_names[MAXQUOTAS]; 157 int s_jquota_fmt; /* Format of quota to use */ 158 #endif 159 /* For which write hints are passed down to block layer */ 160 int alloc_mode; /* segment allocation policy */ 161 int fsync_mode; /* fsync policy */ 162 int fs_mode; /* fs mode: LFS or ADAPTIVE */ 163 int bggc_mode; /* bggc mode: off, on or sync */ 164 int memory_mode; /* memory mode */ 165 int discard_unit; /* 166 * discard command's offset/size should 167 * be aligned to this unit: block, 168 * segment or section 169 */ 170 struct fscrypt_dummy_policy dummy_enc_policy; /* test dummy encryption */ 171 block_t unusable_cap_perc; /* percentage for cap */ 172 block_t unusable_cap; /* Amount of space allowed to be 173 * unusable when disabling checkpoint 174 */ 175 176 /* For compression */ 177 unsigned char compress_algorithm; /* algorithm type */ 178 unsigned char compress_log_size; /* cluster log size */ 179 unsigned char compress_level; /* compress level */ 180 bool compress_chksum; /* compressed data chksum */ 181 unsigned char compress_ext_cnt; /* extension count */ 182 unsigned char nocompress_ext_cnt; /* nocompress extension count */ 183 int compress_mode; /* compression mode */ 184 unsigned char extensions[COMPRESS_EXT_NUM][F2FS_EXTENSION_LEN]; /* extensions */ 185 unsigned char noextensions[COMPRESS_EXT_NUM][F2FS_EXTENSION_LEN]; /* extensions */ 186 }; 187 188 #define F2FS_FEATURE_ENCRYPT 0x0001 189 #define F2FS_FEATURE_BLKZONED 0x0002 190 #define F2FS_FEATURE_ATOMIC_WRITE 0x0004 191 #define F2FS_FEATURE_EXTRA_ATTR 0x0008 192 #define F2FS_FEATURE_PRJQUOTA 0x0010 193 #define F2FS_FEATURE_INODE_CHKSUM 0x0020 194 #define F2FS_FEATURE_FLEXIBLE_INLINE_XATTR 0x0040 195 #define F2FS_FEATURE_QUOTA_INO 0x0080 196 #define F2FS_FEATURE_INODE_CRTIME 0x0100 197 #define F2FS_FEATURE_LOST_FOUND 0x0200 198 #define F2FS_FEATURE_VERITY 0x0400 199 #define F2FS_FEATURE_SB_CHKSUM 0x0800 200 #define F2FS_FEATURE_CASEFOLD 0x1000 201 #define F2FS_FEATURE_COMPRESSION 0x2000 202 #define F2FS_FEATURE_RO 0x4000 203 204 #define __F2FS_HAS_FEATURE(raw_super, mask) \ 205 ((raw_super->feature & cpu_to_le32(mask)) != 0) 206 #define F2FS_HAS_FEATURE(sbi, mask) __F2FS_HAS_FEATURE(sbi->raw_super, mask) 207 208 /* 209 * Default values for user and/or group using reserved blocks 210 */ 211 #define F2FS_DEF_RESUID 0 212 #define F2FS_DEF_RESGID 0 213 214 /* 215 * For checkpoint manager 216 */ 217 enum { 218 NAT_BITMAP, 219 SIT_BITMAP 220 }; 221 222 #define CP_UMOUNT 0x00000001 223 #define CP_FASTBOOT 0x00000002 224 #define CP_SYNC 0x00000004 225 #define CP_RECOVERY 0x00000008 226 #define CP_DISCARD 0x00000010 227 #define CP_TRIMMED 0x00000020 228 #define CP_PAUSE 0x00000040 229 #define CP_RESIZE 0x00000080 230 231 #define DEF_MAX_DISCARD_REQUEST 8 /* issue 8 discards per round */ 232 #define DEF_MIN_DISCARD_ISSUE_TIME 50 /* 50 ms, if exists */ 233 #define DEF_MID_DISCARD_ISSUE_TIME 500 /* 500 ms, if device busy */ 234 #define DEF_MAX_DISCARD_ISSUE_TIME 60000 /* 60 s, if no candidates */ 235 #define DEF_DISCARD_URGENT_UTIL 80 /* do more discard over 80% */ 236 #define DEF_CP_INTERVAL 60 /* 60 secs */ 237 #define DEF_IDLE_INTERVAL 5 /* 5 secs */ 238 #define DEF_DISABLE_INTERVAL 5 /* 5 secs */ 239 #define DEF_DISABLE_QUICK_INTERVAL 1 /* 1 secs */ 240 #define DEF_UMOUNT_DISCARD_TIMEOUT 5 /* 5 secs */ 241 242 struct cp_control { 243 int reason; 244 __u64 trim_start; 245 __u64 trim_end; 246 __u64 trim_minlen; 247 }; 248 249 /* 250 * indicate meta/data type 251 */ 252 enum { 253 META_CP, 254 META_NAT, 255 META_SIT, 256 META_SSA, 257 META_MAX, 258 META_POR, 259 DATA_GENERIC, /* check range only */ 260 DATA_GENERIC_ENHANCE, /* strong check on range and segment bitmap */ 261 DATA_GENERIC_ENHANCE_READ, /* 262 * strong check on range and segment 263 * bitmap but no warning due to race 264 * condition of read on truncated area 265 * by extent_cache 266 */ 267 DATA_GENERIC_ENHANCE_UPDATE, /* 268 * strong check on range and segment 269 * bitmap for update case 270 */ 271 META_GENERIC, 272 }; 273 274 /* for the list of ino */ 275 enum { 276 ORPHAN_INO, /* for orphan ino list */ 277 APPEND_INO, /* for append ino list */ 278 UPDATE_INO, /* for update ino list */ 279 TRANS_DIR_INO, /* for transactions dir ino list */ 280 FLUSH_INO, /* for multiple device flushing */ 281 MAX_INO_ENTRY, /* max. list */ 282 }; 283 284 struct ino_entry { 285 struct list_head list; /* list head */ 286 nid_t ino; /* inode number */ 287 unsigned int dirty_device; /* dirty device bitmap */ 288 }; 289 290 /* for the list of inodes to be GCed */ 291 struct inode_entry { 292 struct list_head list; /* list head */ 293 struct inode *inode; /* vfs inode pointer */ 294 }; 295 296 struct fsync_node_entry { 297 struct list_head list; /* list head */ 298 struct page *page; /* warm node page pointer */ 299 unsigned int seq_id; /* sequence id */ 300 }; 301 302 struct ckpt_req { 303 struct completion wait; /* completion for checkpoint done */ 304 struct llist_node llnode; /* llist_node to be linked in wait queue */ 305 int ret; /* return code of checkpoint */ 306 ktime_t queue_time; /* request queued time */ 307 }; 308 309 struct ckpt_req_control { 310 struct task_struct *f2fs_issue_ckpt; /* checkpoint task */ 311 int ckpt_thread_ioprio; /* checkpoint merge thread ioprio */ 312 wait_queue_head_t ckpt_wait_queue; /* waiting queue for wake-up */ 313 atomic_t issued_ckpt; /* # of actually issued ckpts */ 314 atomic_t total_ckpt; /* # of total ckpts */ 315 atomic_t queued_ckpt; /* # of queued ckpts */ 316 struct llist_head issue_list; /* list for command issue */ 317 spinlock_t stat_lock; /* lock for below checkpoint time stats */ 318 unsigned int cur_time; /* cur wait time in msec for currently issued checkpoint */ 319 unsigned int peak_time; /* peak wait time in msec until now */ 320 }; 321 322 /* for the bitmap indicate blocks to be discarded */ 323 struct discard_entry { 324 struct list_head list; /* list head */ 325 block_t start_blkaddr; /* start blockaddr of current segment */ 326 unsigned char discard_map[SIT_VBLOCK_MAP_SIZE]; /* segment discard bitmap */ 327 }; 328 329 /* minimum discard granularity, unit: block count */ 330 #define MIN_DISCARD_GRANULARITY 1 331 /* default discard granularity of inner discard thread, unit: block count */ 332 #define DEFAULT_DISCARD_GRANULARITY 16 333 /* default maximum discard granularity of ordered discard, unit: block count */ 334 #define DEFAULT_MAX_ORDERED_DISCARD_GRANULARITY 16 335 336 /* max discard pend list number */ 337 #define MAX_PLIST_NUM 512 338 #define plist_idx(blk_num) ((blk_num) >= MAX_PLIST_NUM ? \ 339 (MAX_PLIST_NUM - 1) : ((blk_num) - 1)) 340 341 enum { 342 D_PREP, /* initial */ 343 D_PARTIAL, /* partially submitted */ 344 D_SUBMIT, /* all submitted */ 345 D_DONE, /* finished */ 346 }; 347 348 struct discard_info { 349 block_t lstart; /* logical start address */ 350 block_t len; /* length */ 351 block_t start; /* actual start address in dev */ 352 }; 353 354 struct discard_cmd { 355 struct rb_node rb_node; /* rb node located in rb-tree */ 356 union { 357 struct { 358 block_t lstart; /* logical start address */ 359 block_t len; /* length */ 360 block_t start; /* actual start address in dev */ 361 }; 362 struct discard_info di; /* discard info */ 363 364 }; 365 struct list_head list; /* command list */ 366 struct completion wait; /* compleation */ 367 struct block_device *bdev; /* bdev */ 368 unsigned short ref; /* reference count */ 369 unsigned char state; /* state */ 370 unsigned char queued; /* queued discard */ 371 int error; /* bio error */ 372 spinlock_t lock; /* for state/bio_ref updating */ 373 unsigned short bio_ref; /* bio reference count */ 374 }; 375 376 enum { 377 DPOLICY_BG, 378 DPOLICY_FORCE, 379 DPOLICY_FSTRIM, 380 DPOLICY_UMOUNT, 381 MAX_DPOLICY, 382 }; 383 384 struct discard_policy { 385 int type; /* type of discard */ 386 unsigned int min_interval; /* used for candidates exist */ 387 unsigned int mid_interval; /* used for device busy */ 388 unsigned int max_interval; /* used for candidates not exist */ 389 unsigned int max_requests; /* # of discards issued per round */ 390 unsigned int io_aware_gran; /* minimum granularity discard not be aware of I/O */ 391 bool io_aware; /* issue discard in idle time */ 392 bool sync; /* submit discard with REQ_SYNC flag */ 393 bool ordered; /* issue discard by lba order */ 394 bool timeout; /* discard timeout for put_super */ 395 unsigned int granularity; /* discard granularity */ 396 }; 397 398 struct discard_cmd_control { 399 struct task_struct *f2fs_issue_discard; /* discard thread */ 400 struct list_head entry_list; /* 4KB discard entry list */ 401 struct list_head pend_list[MAX_PLIST_NUM];/* store pending entries */ 402 struct list_head wait_list; /* store on-flushing entries */ 403 struct list_head fstrim_list; /* in-flight discard from fstrim */ 404 wait_queue_head_t discard_wait_queue; /* waiting queue for wake-up */ 405 unsigned int discard_wake; /* to wake up discard thread */ 406 struct mutex cmd_lock; 407 unsigned int nr_discards; /* # of discards in the list */ 408 unsigned int max_discards; /* max. discards to be issued */ 409 unsigned int max_discard_request; /* max. discard request per round */ 410 unsigned int min_discard_issue_time; /* min. interval between discard issue */ 411 unsigned int mid_discard_issue_time; /* mid. interval between discard issue */ 412 unsigned int max_discard_issue_time; /* max. interval between discard issue */ 413 unsigned int discard_urgent_util; /* utilization which issue discard proactively */ 414 unsigned int discard_granularity; /* discard granularity */ 415 unsigned int max_ordered_discard; /* maximum discard granularity issued by lba order */ 416 unsigned int undiscard_blks; /* # of undiscard blocks */ 417 unsigned int next_pos; /* next discard position */ 418 atomic_t issued_discard; /* # of issued discard */ 419 atomic_t queued_discard; /* # of queued discard */ 420 atomic_t discard_cmd_cnt; /* # of cached cmd count */ 421 struct rb_root_cached root; /* root of discard rb-tree */ 422 bool rbtree_check; /* config for consistence check */ 423 }; 424 425 /* for the list of fsync inodes, used only during recovery */ 426 struct fsync_inode_entry { 427 struct list_head list; /* list head */ 428 struct inode *inode; /* vfs inode pointer */ 429 block_t blkaddr; /* block address locating the last fsync */ 430 block_t last_dentry; /* block address locating the last dentry */ 431 }; 432 433 #define nats_in_cursum(jnl) (le16_to_cpu((jnl)->n_nats)) 434 #define sits_in_cursum(jnl) (le16_to_cpu((jnl)->n_sits)) 435 436 #define nat_in_journal(jnl, i) ((jnl)->nat_j.entries[i].ne) 437 #define nid_in_journal(jnl, i) ((jnl)->nat_j.entries[i].nid) 438 #define sit_in_journal(jnl, i) ((jnl)->sit_j.entries[i].se) 439 #define segno_in_journal(jnl, i) ((jnl)->sit_j.entries[i].segno) 440 441 #define MAX_NAT_JENTRIES(jnl) (NAT_JOURNAL_ENTRIES - nats_in_cursum(jnl)) 442 #define MAX_SIT_JENTRIES(jnl) (SIT_JOURNAL_ENTRIES - sits_in_cursum(jnl)) 443 444 static inline int update_nats_in_cursum(struct f2fs_journal *journal, int i) 445 { 446 int before = nats_in_cursum(journal); 447 448 journal->n_nats = cpu_to_le16(before + i); 449 return before; 450 } 451 452 static inline int update_sits_in_cursum(struct f2fs_journal *journal, int i) 453 { 454 int before = sits_in_cursum(journal); 455 456 journal->n_sits = cpu_to_le16(before + i); 457 return before; 458 } 459 460 static inline bool __has_cursum_space(struct f2fs_journal *journal, 461 int size, int type) 462 { 463 if (type == NAT_JOURNAL) 464 return size <= MAX_NAT_JENTRIES(journal); 465 return size <= MAX_SIT_JENTRIES(journal); 466 } 467 468 /* for inline stuff */ 469 #define DEF_INLINE_RESERVED_SIZE 1 470 static inline int get_extra_isize(struct inode *inode); 471 static inline int get_inline_xattr_addrs(struct inode *inode); 472 #define MAX_INLINE_DATA(inode) (sizeof(__le32) * \ 473 (CUR_ADDRS_PER_INODE(inode) - \ 474 get_inline_xattr_addrs(inode) - \ 475 DEF_INLINE_RESERVED_SIZE)) 476 477 /* for inline dir */ 478 #define NR_INLINE_DENTRY(inode) (MAX_INLINE_DATA(inode) * BITS_PER_BYTE / \ 479 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \ 480 BITS_PER_BYTE + 1)) 481 #define INLINE_DENTRY_BITMAP_SIZE(inode) \ 482 DIV_ROUND_UP(NR_INLINE_DENTRY(inode), BITS_PER_BYTE) 483 #define INLINE_RESERVED_SIZE(inode) (MAX_INLINE_DATA(inode) - \ 484 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \ 485 NR_INLINE_DENTRY(inode) + \ 486 INLINE_DENTRY_BITMAP_SIZE(inode))) 487 488 /* 489 * For INODE and NODE manager 490 */ 491 /* for directory operations */ 492 493 struct f2fs_filename { 494 /* 495 * The filename the user specified. This is NULL for some 496 * filesystem-internal operations, e.g. converting an inline directory 497 * to a non-inline one, or roll-forward recovering an encrypted dentry. 498 */ 499 const struct qstr *usr_fname; 500 501 /* 502 * The on-disk filename. For encrypted directories, this is encrypted. 503 * This may be NULL for lookups in an encrypted dir without the key. 504 */ 505 struct fscrypt_str disk_name; 506 507 /* The dirhash of this filename */ 508 f2fs_hash_t hash; 509 510 #ifdef CONFIG_FS_ENCRYPTION 511 /* 512 * For lookups in encrypted directories: either the buffer backing 513 * disk_name, or a buffer that holds the decoded no-key name. 514 */ 515 struct fscrypt_str crypto_buf; 516 #endif 517 #if IS_ENABLED(CONFIG_UNICODE) 518 /* 519 * For casefolded directories: the casefolded name, but it's left NULL 520 * if the original name is not valid Unicode, if the original name is 521 * "." or "..", if the directory is both casefolded and encrypted and 522 * its encryption key is unavailable, or if the filesystem is doing an 523 * internal operation where usr_fname is also NULL. In all these cases 524 * we fall back to treating the name as an opaque byte sequence. 525 */ 526 struct fscrypt_str cf_name; 527 #endif 528 }; 529 530 struct f2fs_dentry_ptr { 531 struct inode *inode; 532 void *bitmap; 533 struct f2fs_dir_entry *dentry; 534 __u8 (*filename)[F2FS_SLOT_LEN]; 535 int max; 536 int nr_bitmap; 537 }; 538 539 static inline void make_dentry_ptr_block(struct inode *inode, 540 struct f2fs_dentry_ptr *d, struct f2fs_dentry_block *t) 541 { 542 d->inode = inode; 543 d->max = NR_DENTRY_IN_BLOCK; 544 d->nr_bitmap = SIZE_OF_DENTRY_BITMAP; 545 d->bitmap = t->dentry_bitmap; 546 d->dentry = t->dentry; 547 d->filename = t->filename; 548 } 549 550 static inline void make_dentry_ptr_inline(struct inode *inode, 551 struct f2fs_dentry_ptr *d, void *t) 552 { 553 int entry_cnt = NR_INLINE_DENTRY(inode); 554 int bitmap_size = INLINE_DENTRY_BITMAP_SIZE(inode); 555 int reserved_size = INLINE_RESERVED_SIZE(inode); 556 557 d->inode = inode; 558 d->max = entry_cnt; 559 d->nr_bitmap = bitmap_size; 560 d->bitmap = t; 561 d->dentry = t + bitmap_size + reserved_size; 562 d->filename = t + bitmap_size + reserved_size + 563 SIZE_OF_DIR_ENTRY * entry_cnt; 564 } 565 566 /* 567 * XATTR_NODE_OFFSET stores xattrs to one node block per file keeping -1 568 * as its node offset to distinguish from index node blocks. 569 * But some bits are used to mark the node block. 570 */ 571 #define XATTR_NODE_OFFSET ((((unsigned int)-1) << OFFSET_BIT_SHIFT) \ 572 >> OFFSET_BIT_SHIFT) 573 enum { 574 ALLOC_NODE, /* allocate a new node page if needed */ 575 LOOKUP_NODE, /* look up a node without readahead */ 576 LOOKUP_NODE_RA, /* 577 * look up a node with readahead called 578 * by get_data_block. 579 */ 580 }; 581 582 #define DEFAULT_RETRY_IO_COUNT 8 /* maximum retry read IO or flush count */ 583 584 /* congestion wait timeout value, default: 20ms */ 585 #define DEFAULT_IO_TIMEOUT (msecs_to_jiffies(20)) 586 587 /* maximum retry quota flush count */ 588 #define DEFAULT_RETRY_QUOTA_FLUSH_COUNT 8 589 590 /* maximum retry of EIO'ed page */ 591 #define MAX_RETRY_PAGE_EIO 100 592 593 #define F2FS_LINK_MAX 0xffffffff /* maximum link count per file */ 594 595 #define MAX_DIR_RA_PAGES 4 /* maximum ra pages of dir */ 596 597 /* dirty segments threshold for triggering CP */ 598 #define DEFAULT_DIRTY_THRESHOLD 4 599 600 #define RECOVERY_MAX_RA_BLOCKS BIO_MAX_VECS 601 #define RECOVERY_MIN_RA_BLOCKS 1 602 603 #define F2FS_ONSTACK_PAGES 16 /* nr of onstack pages */ 604 605 /* for in-memory extent cache entry */ 606 #define F2FS_MIN_EXTENT_LEN 64 /* minimum extent length */ 607 608 /* number of extent info in extent cache we try to shrink */ 609 #define READ_EXTENT_CACHE_SHRINK_NUMBER 128 610 611 /* number of age extent info in extent cache we try to shrink */ 612 #define AGE_EXTENT_CACHE_SHRINK_NUMBER 128 613 #define LAST_AGE_WEIGHT 30 614 #define SAME_AGE_REGION 1024 615 616 /* 617 * Define data block with age less than 1GB as hot data 618 * define data block with age less than 10GB but more than 1GB as warm data 619 */ 620 #define DEF_HOT_DATA_AGE_THRESHOLD 262144 621 #define DEF_WARM_DATA_AGE_THRESHOLD 2621440 622 623 /* extent cache type */ 624 enum extent_type { 625 EX_READ, 626 EX_BLOCK_AGE, 627 NR_EXTENT_CACHES, 628 }; 629 630 struct rb_entry { 631 struct rb_node rb_node; /* rb node located in rb-tree */ 632 union { 633 struct { 634 unsigned int ofs; /* start offset of the entry */ 635 unsigned int len; /* length of the entry */ 636 }; 637 unsigned long long key; /* 64-bits key */ 638 } __packed; 639 }; 640 641 struct extent_info { 642 unsigned int fofs; /* start offset in a file */ 643 unsigned int len; /* length of the extent */ 644 union { 645 /* read extent_cache */ 646 struct { 647 /* start block address of the extent */ 648 block_t blk; 649 #ifdef CONFIG_F2FS_FS_COMPRESSION 650 /* physical extent length of compressed blocks */ 651 unsigned int c_len; 652 #endif 653 }; 654 /* block age extent_cache */ 655 struct { 656 /* block age of the extent */ 657 unsigned long long age; 658 /* last total blocks allocated */ 659 unsigned long long last_blocks; 660 }; 661 }; 662 }; 663 664 struct extent_node { 665 struct rb_node rb_node; /* rb node located in rb-tree */ 666 struct extent_info ei; /* extent info */ 667 struct list_head list; /* node in global extent list of sbi */ 668 struct extent_tree *et; /* extent tree pointer */ 669 }; 670 671 struct extent_tree { 672 nid_t ino; /* inode number */ 673 enum extent_type type; /* keep the extent tree type */ 674 struct rb_root_cached root; /* root of extent info rb-tree */ 675 struct extent_node *cached_en; /* recently accessed extent node */ 676 struct list_head list; /* to be used by sbi->zombie_list */ 677 rwlock_t lock; /* protect extent info rb-tree */ 678 atomic_t node_cnt; /* # of extent node in rb-tree*/ 679 bool largest_updated; /* largest extent updated */ 680 struct extent_info largest; /* largest cached extent for EX_READ */ 681 }; 682 683 struct extent_tree_info { 684 struct radix_tree_root extent_tree_root;/* cache extent cache entries */ 685 struct mutex extent_tree_lock; /* locking extent radix tree */ 686 struct list_head extent_list; /* lru list for shrinker */ 687 spinlock_t extent_lock; /* locking extent lru list */ 688 atomic_t total_ext_tree; /* extent tree count */ 689 struct list_head zombie_list; /* extent zombie tree list */ 690 atomic_t total_zombie_tree; /* extent zombie tree count */ 691 atomic_t total_ext_node; /* extent info count */ 692 }; 693 694 /* 695 * This structure is taken from ext4_map_blocks. 696 * 697 * Note that, however, f2fs uses NEW and MAPPED flags for f2fs_map_blocks(). 698 */ 699 #define F2FS_MAP_NEW (1 << BH_New) 700 #define F2FS_MAP_MAPPED (1 << BH_Mapped) 701 #define F2FS_MAP_UNWRITTEN (1 << BH_Unwritten) 702 #define F2FS_MAP_FLAGS (F2FS_MAP_NEW | F2FS_MAP_MAPPED |\ 703 F2FS_MAP_UNWRITTEN) 704 705 struct f2fs_map_blocks { 706 struct block_device *m_bdev; /* for multi-device dio */ 707 block_t m_pblk; 708 block_t m_lblk; 709 unsigned int m_len; 710 unsigned int m_flags; 711 pgoff_t *m_next_pgofs; /* point next possible non-hole pgofs */ 712 pgoff_t *m_next_extent; /* point to next possible extent */ 713 int m_seg_type; 714 bool m_may_create; /* indicate it is from write path */ 715 bool m_multidev_dio; /* indicate it allows multi-device dio */ 716 }; 717 718 /* for flag in get_data_block */ 719 enum { 720 F2FS_GET_BLOCK_DEFAULT, 721 F2FS_GET_BLOCK_FIEMAP, 722 F2FS_GET_BLOCK_BMAP, 723 F2FS_GET_BLOCK_DIO, 724 F2FS_GET_BLOCK_PRE_DIO, 725 F2FS_GET_BLOCK_PRE_AIO, 726 F2FS_GET_BLOCK_PRECACHE, 727 }; 728 729 /* 730 * i_advise uses FADVISE_XXX_BIT. We can add additional hints later. 731 */ 732 #define FADVISE_COLD_BIT 0x01 733 #define FADVISE_LOST_PINO_BIT 0x02 734 #define FADVISE_ENCRYPT_BIT 0x04 735 #define FADVISE_ENC_NAME_BIT 0x08 736 #define FADVISE_KEEP_SIZE_BIT 0x10 737 #define FADVISE_HOT_BIT 0x20 738 #define FADVISE_VERITY_BIT 0x40 739 #define FADVISE_TRUNC_BIT 0x80 740 741 #define FADVISE_MODIFIABLE_BITS (FADVISE_COLD_BIT | FADVISE_HOT_BIT) 742 743 #define file_is_cold(inode) is_file(inode, FADVISE_COLD_BIT) 744 #define file_set_cold(inode) set_file(inode, FADVISE_COLD_BIT) 745 #define file_clear_cold(inode) clear_file(inode, FADVISE_COLD_BIT) 746 747 #define file_wrong_pino(inode) is_file(inode, FADVISE_LOST_PINO_BIT) 748 #define file_lost_pino(inode) set_file(inode, FADVISE_LOST_PINO_BIT) 749 #define file_got_pino(inode) clear_file(inode, FADVISE_LOST_PINO_BIT) 750 751 #define file_is_encrypt(inode) is_file(inode, FADVISE_ENCRYPT_BIT) 752 #define file_set_encrypt(inode) set_file(inode, FADVISE_ENCRYPT_BIT) 753 754 #define file_enc_name(inode) is_file(inode, FADVISE_ENC_NAME_BIT) 755 #define file_set_enc_name(inode) set_file(inode, FADVISE_ENC_NAME_BIT) 756 757 #define file_keep_isize(inode) is_file(inode, FADVISE_KEEP_SIZE_BIT) 758 #define file_set_keep_isize(inode) set_file(inode, FADVISE_KEEP_SIZE_BIT) 759 760 #define file_is_hot(inode) is_file(inode, FADVISE_HOT_BIT) 761 #define file_set_hot(inode) set_file(inode, FADVISE_HOT_BIT) 762 #define file_clear_hot(inode) clear_file(inode, FADVISE_HOT_BIT) 763 764 #define file_is_verity(inode) is_file(inode, FADVISE_VERITY_BIT) 765 #define file_set_verity(inode) set_file(inode, FADVISE_VERITY_BIT) 766 767 #define file_should_truncate(inode) is_file(inode, FADVISE_TRUNC_BIT) 768 #define file_need_truncate(inode) set_file(inode, FADVISE_TRUNC_BIT) 769 #define file_dont_truncate(inode) clear_file(inode, FADVISE_TRUNC_BIT) 770 771 #define DEF_DIR_LEVEL 0 772 773 enum { 774 GC_FAILURE_PIN, 775 MAX_GC_FAILURE 776 }; 777 778 /* used for f2fs_inode_info->flags */ 779 enum { 780 FI_NEW_INODE, /* indicate newly allocated inode */ 781 FI_DIRTY_INODE, /* indicate inode is dirty or not */ 782 FI_AUTO_RECOVER, /* indicate inode is recoverable */ 783 FI_DIRTY_DIR, /* indicate directory has dirty pages */ 784 FI_INC_LINK, /* need to increment i_nlink */ 785 FI_ACL_MODE, /* indicate acl mode */ 786 FI_NO_ALLOC, /* should not allocate any blocks */ 787 FI_FREE_NID, /* free allocated nide */ 788 FI_NO_EXTENT, /* not to use the extent cache */ 789 FI_INLINE_XATTR, /* used for inline xattr */ 790 FI_INLINE_DATA, /* used for inline data*/ 791 FI_INLINE_DENTRY, /* used for inline dentry */ 792 FI_APPEND_WRITE, /* inode has appended data */ 793 FI_UPDATE_WRITE, /* inode has in-place-update data */ 794 FI_NEED_IPU, /* used for ipu per file */ 795 FI_ATOMIC_FILE, /* indicate atomic file */ 796 FI_FIRST_BLOCK_WRITTEN, /* indicate #0 data block was written */ 797 FI_DROP_CACHE, /* drop dirty page cache */ 798 FI_DATA_EXIST, /* indicate data exists */ 799 FI_INLINE_DOTS, /* indicate inline dot dentries */ 800 FI_SKIP_WRITES, /* should skip data page writeback */ 801 FI_OPU_WRITE, /* used for opu per file */ 802 FI_DIRTY_FILE, /* indicate regular/symlink has dirty pages */ 803 FI_PREALLOCATED_ALL, /* all blocks for write were preallocated */ 804 FI_HOT_DATA, /* indicate file is hot */ 805 FI_EXTRA_ATTR, /* indicate file has extra attribute */ 806 FI_PROJ_INHERIT, /* indicate file inherits projectid */ 807 FI_PIN_FILE, /* indicate file should not be gced */ 808 FI_VERITY_IN_PROGRESS, /* building fs-verity Merkle tree */ 809 FI_COMPRESSED_FILE, /* indicate file's data can be compressed */ 810 FI_COMPRESS_CORRUPT, /* indicate compressed cluster is corrupted */ 811 FI_MMAP_FILE, /* indicate file was mmapped */ 812 FI_ENABLE_COMPRESS, /* enable compression in "user" compression mode */ 813 FI_COMPRESS_RELEASED, /* compressed blocks were released */ 814 FI_ALIGNED_WRITE, /* enable aligned write */ 815 FI_COW_FILE, /* indicate COW file */ 816 FI_ATOMIC_COMMITTED, /* indicate atomic commit completed except disk sync */ 817 FI_ATOMIC_REPLACE, /* indicate atomic replace */ 818 FI_MAX, /* max flag, never be used */ 819 }; 820 821 struct f2fs_inode_info { 822 struct inode vfs_inode; /* serve a vfs inode */ 823 unsigned long i_flags; /* keep an inode flags for ioctl */ 824 unsigned char i_advise; /* use to give file attribute hints */ 825 unsigned char i_dir_level; /* use for dentry level for large dir */ 826 unsigned int i_current_depth; /* only for directory depth */ 827 /* for gc failure statistic */ 828 unsigned int i_gc_failures[MAX_GC_FAILURE]; 829 unsigned int i_pino; /* parent inode number */ 830 umode_t i_acl_mode; /* keep file acl mode temporarily */ 831 832 /* Use below internally in f2fs*/ 833 unsigned long flags[BITS_TO_LONGS(FI_MAX)]; /* use to pass per-file flags */ 834 struct f2fs_rwsem i_sem; /* protect fi info */ 835 atomic_t dirty_pages; /* # of dirty pages */ 836 f2fs_hash_t chash; /* hash value of given file name */ 837 unsigned int clevel; /* maximum level of given file name */ 838 struct task_struct *task; /* lookup and create consistency */ 839 struct task_struct *cp_task; /* separate cp/wb IO stats*/ 840 struct task_struct *wb_task; /* indicate inode is in context of writeback */ 841 nid_t i_xattr_nid; /* node id that contains xattrs */ 842 loff_t last_disk_size; /* lastly written file size */ 843 spinlock_t i_size_lock; /* protect last_disk_size */ 844 845 #ifdef CONFIG_QUOTA 846 struct dquot *i_dquot[MAXQUOTAS]; 847 848 /* quota space reservation, managed internally by quota code */ 849 qsize_t i_reserved_quota; 850 #endif 851 struct list_head dirty_list; /* dirty list for dirs and files */ 852 struct list_head gdirty_list; /* linked in global dirty list */ 853 struct task_struct *atomic_write_task; /* store atomic write task */ 854 struct extent_tree *extent_tree[NR_EXTENT_CACHES]; 855 /* cached extent_tree entry */ 856 struct inode *cow_inode; /* copy-on-write inode for atomic write */ 857 858 /* avoid racing between foreground op and gc */ 859 struct f2fs_rwsem i_gc_rwsem[2]; 860 struct f2fs_rwsem i_xattr_sem; /* avoid racing between reading and changing EAs */ 861 862 int i_extra_isize; /* size of extra space located in i_addr */ 863 kprojid_t i_projid; /* id for project quota */ 864 int i_inline_xattr_size; /* inline xattr size */ 865 struct timespec64 i_crtime; /* inode creation time */ 866 struct timespec64 i_disk_time[4];/* inode disk times */ 867 868 /* for file compress */ 869 atomic_t i_compr_blocks; /* # of compressed blocks */ 870 unsigned char i_compress_algorithm; /* algorithm type */ 871 unsigned char i_log_cluster_size; /* log of cluster size */ 872 unsigned char i_compress_level; /* compress level (lz4hc,zstd) */ 873 unsigned short i_compress_flag; /* compress flag */ 874 unsigned int i_cluster_size; /* cluster size */ 875 876 unsigned int atomic_write_cnt; 877 loff_t original_i_size; /* original i_size before atomic write */ 878 }; 879 880 static inline void get_read_extent_info(struct extent_info *ext, 881 struct f2fs_extent *i_ext) 882 { 883 ext->fofs = le32_to_cpu(i_ext->fofs); 884 ext->blk = le32_to_cpu(i_ext->blk); 885 ext->len = le32_to_cpu(i_ext->len); 886 } 887 888 static inline void set_raw_read_extent(struct extent_info *ext, 889 struct f2fs_extent *i_ext) 890 { 891 i_ext->fofs = cpu_to_le32(ext->fofs); 892 i_ext->blk = cpu_to_le32(ext->blk); 893 i_ext->len = cpu_to_le32(ext->len); 894 } 895 896 static inline bool __is_discard_mergeable(struct discard_info *back, 897 struct discard_info *front, unsigned int max_len) 898 { 899 return (back->lstart + back->len == front->lstart) && 900 (back->len + front->len <= max_len); 901 } 902 903 static inline bool __is_discard_back_mergeable(struct discard_info *cur, 904 struct discard_info *back, unsigned int max_len) 905 { 906 return __is_discard_mergeable(back, cur, max_len); 907 } 908 909 static inline bool __is_discard_front_mergeable(struct discard_info *cur, 910 struct discard_info *front, unsigned int max_len) 911 { 912 return __is_discard_mergeable(cur, front, max_len); 913 } 914 915 /* 916 * For free nid management 917 */ 918 enum nid_state { 919 FREE_NID, /* newly added to free nid list */ 920 PREALLOC_NID, /* it is preallocated */ 921 MAX_NID_STATE, 922 }; 923 924 enum nat_state { 925 TOTAL_NAT, 926 DIRTY_NAT, 927 RECLAIMABLE_NAT, 928 MAX_NAT_STATE, 929 }; 930 931 struct f2fs_nm_info { 932 block_t nat_blkaddr; /* base disk address of NAT */ 933 nid_t max_nid; /* maximum possible node ids */ 934 nid_t available_nids; /* # of available node ids */ 935 nid_t next_scan_nid; /* the next nid to be scanned */ 936 nid_t max_rf_node_blocks; /* max # of nodes for recovery */ 937 unsigned int ram_thresh; /* control the memory footprint */ 938 unsigned int ra_nid_pages; /* # of nid pages to be readaheaded */ 939 unsigned int dirty_nats_ratio; /* control dirty nats ratio threshold */ 940 941 /* NAT cache management */ 942 struct radix_tree_root nat_root;/* root of the nat entry cache */ 943 struct radix_tree_root nat_set_root;/* root of the nat set cache */ 944 struct f2fs_rwsem nat_tree_lock; /* protect nat entry tree */ 945 struct list_head nat_entries; /* cached nat entry list (clean) */ 946 spinlock_t nat_list_lock; /* protect clean nat entry list */ 947 unsigned int nat_cnt[MAX_NAT_STATE]; /* the # of cached nat entries */ 948 unsigned int nat_blocks; /* # of nat blocks */ 949 950 /* free node ids management */ 951 struct radix_tree_root free_nid_root;/* root of the free_nid cache */ 952 struct list_head free_nid_list; /* list for free nids excluding preallocated nids */ 953 unsigned int nid_cnt[MAX_NID_STATE]; /* the number of free node id */ 954 spinlock_t nid_list_lock; /* protect nid lists ops */ 955 struct mutex build_lock; /* lock for build free nids */ 956 unsigned char **free_nid_bitmap; 957 unsigned char *nat_block_bitmap; 958 unsigned short *free_nid_count; /* free nid count of NAT block */ 959 960 /* for checkpoint */ 961 char *nat_bitmap; /* NAT bitmap pointer */ 962 963 unsigned int nat_bits_blocks; /* # of nat bits blocks */ 964 unsigned char *nat_bits; /* NAT bits blocks */ 965 unsigned char *full_nat_bits; /* full NAT pages */ 966 unsigned char *empty_nat_bits; /* empty NAT pages */ 967 #ifdef CONFIG_F2FS_CHECK_FS 968 char *nat_bitmap_mir; /* NAT bitmap mirror */ 969 #endif 970 int bitmap_size; /* bitmap size */ 971 }; 972 973 /* 974 * this structure is used as one of function parameters. 975 * all the information are dedicated to a given direct node block determined 976 * by the data offset in a file. 977 */ 978 struct dnode_of_data { 979 struct inode *inode; /* vfs inode pointer */ 980 struct page *inode_page; /* its inode page, NULL is possible */ 981 struct page *node_page; /* cached direct node page */ 982 nid_t nid; /* node id of the direct node block */ 983 unsigned int ofs_in_node; /* data offset in the node page */ 984 bool inode_page_locked; /* inode page is locked or not */ 985 bool node_changed; /* is node block changed */ 986 char cur_level; /* level of hole node page */ 987 char max_level; /* level of current page located */ 988 block_t data_blkaddr; /* block address of the node block */ 989 }; 990 991 static inline void set_new_dnode(struct dnode_of_data *dn, struct inode *inode, 992 struct page *ipage, struct page *npage, nid_t nid) 993 { 994 memset(dn, 0, sizeof(*dn)); 995 dn->inode = inode; 996 dn->inode_page = ipage; 997 dn->node_page = npage; 998 dn->nid = nid; 999 } 1000 1001 /* 1002 * For SIT manager 1003 * 1004 * By default, there are 6 active log areas across the whole main area. 1005 * When considering hot and cold data separation to reduce cleaning overhead, 1006 * we split 3 for data logs and 3 for node logs as hot, warm, and cold types, 1007 * respectively. 1008 * In the current design, you should not change the numbers intentionally. 1009 * Instead, as a mount option such as active_logs=x, you can use 2, 4, and 6 1010 * logs individually according to the underlying devices. (default: 6) 1011 * Just in case, on-disk layout covers maximum 16 logs that consist of 8 for 1012 * data and 8 for node logs. 1013 */ 1014 #define NR_CURSEG_DATA_TYPE (3) 1015 #define NR_CURSEG_NODE_TYPE (3) 1016 #define NR_CURSEG_INMEM_TYPE (2) 1017 #define NR_CURSEG_RO_TYPE (2) 1018 #define NR_CURSEG_PERSIST_TYPE (NR_CURSEG_DATA_TYPE + NR_CURSEG_NODE_TYPE) 1019 #define NR_CURSEG_TYPE (NR_CURSEG_INMEM_TYPE + NR_CURSEG_PERSIST_TYPE) 1020 1021 enum { 1022 CURSEG_HOT_DATA = 0, /* directory entry blocks */ 1023 CURSEG_WARM_DATA, /* data blocks */ 1024 CURSEG_COLD_DATA, /* multimedia or GCed data blocks */ 1025 CURSEG_HOT_NODE, /* direct node blocks of directory files */ 1026 CURSEG_WARM_NODE, /* direct node blocks of normal files */ 1027 CURSEG_COLD_NODE, /* indirect node blocks */ 1028 NR_PERSISTENT_LOG, /* number of persistent log */ 1029 CURSEG_COLD_DATA_PINNED = NR_PERSISTENT_LOG, 1030 /* pinned file that needs consecutive block address */ 1031 CURSEG_ALL_DATA_ATGC, /* SSR alloctor in hot/warm/cold data area */ 1032 NO_CHECK_TYPE, /* number of persistent & inmem log */ 1033 }; 1034 1035 struct flush_cmd { 1036 struct completion wait; 1037 struct llist_node llnode; 1038 nid_t ino; 1039 int ret; 1040 }; 1041 1042 struct flush_cmd_control { 1043 struct task_struct *f2fs_issue_flush; /* flush thread */ 1044 wait_queue_head_t flush_wait_queue; /* waiting queue for wake-up */ 1045 atomic_t issued_flush; /* # of issued flushes */ 1046 atomic_t queued_flush; /* # of queued flushes */ 1047 struct llist_head issue_list; /* list for command issue */ 1048 struct llist_node *dispatch_list; /* list for command dispatch */ 1049 }; 1050 1051 struct f2fs_sm_info { 1052 struct sit_info *sit_info; /* whole segment information */ 1053 struct free_segmap_info *free_info; /* free segment information */ 1054 struct dirty_seglist_info *dirty_info; /* dirty segment information */ 1055 struct curseg_info *curseg_array; /* active segment information */ 1056 1057 struct f2fs_rwsem curseg_lock; /* for preventing curseg change */ 1058 1059 block_t seg0_blkaddr; /* block address of 0'th segment */ 1060 block_t main_blkaddr; /* start block address of main area */ 1061 block_t ssa_blkaddr; /* start block address of SSA area */ 1062 1063 unsigned int segment_count; /* total # of segments */ 1064 unsigned int main_segments; /* # of segments in main area */ 1065 unsigned int reserved_segments; /* # of reserved segments */ 1066 unsigned int additional_reserved_segments;/* reserved segs for IO align feature */ 1067 unsigned int ovp_segments; /* # of overprovision segments */ 1068 1069 /* a threshold to reclaim prefree segments */ 1070 unsigned int rec_prefree_segments; 1071 1072 struct list_head sit_entry_set; /* sit entry set list */ 1073 1074 unsigned int ipu_policy; /* in-place-update policy */ 1075 unsigned int min_ipu_util; /* in-place-update threshold */ 1076 unsigned int min_fsync_blocks; /* threshold for fsync */ 1077 unsigned int min_seq_blocks; /* threshold for sequential blocks */ 1078 unsigned int min_hot_blocks; /* threshold for hot block allocation */ 1079 unsigned int min_ssr_sections; /* threshold to trigger SSR allocation */ 1080 1081 /* for flush command control */ 1082 struct flush_cmd_control *fcc_info; 1083 1084 /* for discard command control */ 1085 struct discard_cmd_control *dcc_info; 1086 }; 1087 1088 /* 1089 * For superblock 1090 */ 1091 /* 1092 * COUNT_TYPE for monitoring 1093 * 1094 * f2fs monitors the number of several block types such as on-writeback, 1095 * dirty dentry blocks, dirty node blocks, and dirty meta blocks. 1096 */ 1097 #define WB_DATA_TYPE(p) (__is_cp_guaranteed(p) ? F2FS_WB_CP_DATA : F2FS_WB_DATA) 1098 enum count_type { 1099 F2FS_DIRTY_DENTS, 1100 F2FS_DIRTY_DATA, 1101 F2FS_DIRTY_QDATA, 1102 F2FS_DIRTY_NODES, 1103 F2FS_DIRTY_META, 1104 F2FS_DIRTY_IMETA, 1105 F2FS_WB_CP_DATA, 1106 F2FS_WB_DATA, 1107 F2FS_RD_DATA, 1108 F2FS_RD_NODE, 1109 F2FS_RD_META, 1110 F2FS_DIO_WRITE, 1111 F2FS_DIO_READ, 1112 NR_COUNT_TYPE, 1113 }; 1114 1115 /* 1116 * The below are the page types of bios used in submit_bio(). 1117 * The available types are: 1118 * DATA User data pages. It operates as async mode. 1119 * NODE Node pages. It operates as async mode. 1120 * META FS metadata pages such as SIT, NAT, CP. 1121 * NR_PAGE_TYPE The number of page types. 1122 * META_FLUSH Make sure the previous pages are written 1123 * with waiting the bio's completion 1124 * ... Only can be used with META. 1125 */ 1126 #define PAGE_TYPE_OF_BIO(type) ((type) > META ? META : (type)) 1127 enum page_type { 1128 DATA = 0, 1129 NODE = 1, /* should not change this */ 1130 META, 1131 NR_PAGE_TYPE, 1132 META_FLUSH, 1133 IPU, /* the below types are used by tracepoints only. */ 1134 OPU, 1135 }; 1136 1137 enum temp_type { 1138 HOT = 0, /* must be zero for meta bio */ 1139 WARM, 1140 COLD, 1141 NR_TEMP_TYPE, 1142 }; 1143 1144 enum need_lock_type { 1145 LOCK_REQ = 0, 1146 LOCK_DONE, 1147 LOCK_RETRY, 1148 }; 1149 1150 enum cp_reason_type { 1151 CP_NO_NEEDED, 1152 CP_NON_REGULAR, 1153 CP_COMPRESSED, 1154 CP_HARDLINK, 1155 CP_SB_NEED_CP, 1156 CP_WRONG_PINO, 1157 CP_NO_SPC_ROLL, 1158 CP_NODE_NEED_CP, 1159 CP_FASTBOOT_MODE, 1160 CP_SPEC_LOG_NUM, 1161 CP_RECOVER_DIR, 1162 }; 1163 1164 enum iostat_type { 1165 /* WRITE IO */ 1166 APP_DIRECT_IO, /* app direct write IOs */ 1167 APP_BUFFERED_IO, /* app buffered write IOs */ 1168 APP_WRITE_IO, /* app write IOs */ 1169 APP_MAPPED_IO, /* app mapped IOs */ 1170 APP_BUFFERED_CDATA_IO, /* app buffered write IOs on compressed file */ 1171 APP_MAPPED_CDATA_IO, /* app mapped write IOs on compressed file */ 1172 FS_DATA_IO, /* data IOs from kworker/fsync/reclaimer */ 1173 FS_CDATA_IO, /* data IOs from kworker/fsync/reclaimer on compressed file */ 1174 FS_NODE_IO, /* node IOs from kworker/fsync/reclaimer */ 1175 FS_META_IO, /* meta IOs from kworker/reclaimer */ 1176 FS_GC_DATA_IO, /* data IOs from forground gc */ 1177 FS_GC_NODE_IO, /* node IOs from forground gc */ 1178 FS_CP_DATA_IO, /* data IOs from checkpoint */ 1179 FS_CP_NODE_IO, /* node IOs from checkpoint */ 1180 FS_CP_META_IO, /* meta IOs from checkpoint */ 1181 1182 /* READ IO */ 1183 APP_DIRECT_READ_IO, /* app direct read IOs */ 1184 APP_BUFFERED_READ_IO, /* app buffered read IOs */ 1185 APP_READ_IO, /* app read IOs */ 1186 APP_MAPPED_READ_IO, /* app mapped read IOs */ 1187 APP_BUFFERED_CDATA_READ_IO, /* app buffered read IOs on compressed file */ 1188 APP_MAPPED_CDATA_READ_IO, /* app mapped read IOs on compressed file */ 1189 FS_DATA_READ_IO, /* data read IOs */ 1190 FS_GDATA_READ_IO, /* data read IOs from background gc */ 1191 FS_CDATA_READ_IO, /* compressed data read IOs */ 1192 FS_NODE_READ_IO, /* node read IOs */ 1193 FS_META_READ_IO, /* meta read IOs */ 1194 1195 /* other */ 1196 FS_DISCARD, /* discard */ 1197 NR_IO_TYPE, 1198 }; 1199 1200 struct f2fs_io_info { 1201 struct f2fs_sb_info *sbi; /* f2fs_sb_info pointer */ 1202 nid_t ino; /* inode number */ 1203 enum page_type type; /* contains DATA/NODE/META/META_FLUSH */ 1204 enum temp_type temp; /* contains HOT/WARM/COLD */ 1205 enum req_op op; /* contains REQ_OP_ */ 1206 blk_opf_t op_flags; /* req_flag_bits */ 1207 block_t new_blkaddr; /* new block address to be written */ 1208 block_t old_blkaddr; /* old block address before Cow */ 1209 struct page *page; /* page to be written */ 1210 struct page *encrypted_page; /* encrypted page */ 1211 struct page *compressed_page; /* compressed page */ 1212 struct list_head list; /* serialize IOs */ 1213 bool submitted; /* indicate IO submission */ 1214 int need_lock; /* indicate we need to lock cp_rwsem */ 1215 bool in_list; /* indicate fio is in io_list */ 1216 bool is_por; /* indicate IO is from recovery or not */ 1217 bool retry; /* need to reallocate block address */ 1218 int compr_blocks; /* # of compressed block addresses */ 1219 bool encrypted; /* indicate file is encrypted */ 1220 bool post_read; /* require post read */ 1221 enum iostat_type io_type; /* io type */ 1222 struct writeback_control *io_wbc; /* writeback control */ 1223 struct bio **bio; /* bio for ipu */ 1224 sector_t *last_block; /* last block number in bio */ 1225 unsigned char version; /* version of the node */ 1226 }; 1227 1228 struct bio_entry { 1229 struct bio *bio; 1230 struct list_head list; 1231 }; 1232 1233 #define is_read_io(rw) ((rw) == READ) 1234 struct f2fs_bio_info { 1235 struct f2fs_sb_info *sbi; /* f2fs superblock */ 1236 struct bio *bio; /* bios to merge */ 1237 sector_t last_block_in_bio; /* last block number */ 1238 struct f2fs_io_info fio; /* store buffered io info. */ 1239 struct f2fs_rwsem io_rwsem; /* blocking op for bio */ 1240 spinlock_t io_lock; /* serialize DATA/NODE IOs */ 1241 struct list_head io_list; /* track fios */ 1242 struct list_head bio_list; /* bio entry list head */ 1243 struct f2fs_rwsem bio_list_lock; /* lock to protect bio entry list */ 1244 }; 1245 1246 #define FDEV(i) (sbi->devs[i]) 1247 #define RDEV(i) (raw_super->devs[i]) 1248 struct f2fs_dev_info { 1249 struct block_device *bdev; 1250 char path[MAX_PATH_LEN]; 1251 unsigned int total_segments; 1252 block_t start_blk; 1253 block_t end_blk; 1254 #ifdef CONFIG_BLK_DEV_ZONED 1255 unsigned int nr_blkz; /* Total number of zones */ 1256 unsigned long *blkz_seq; /* Bitmap indicating sequential zones */ 1257 #endif 1258 }; 1259 1260 enum inode_type { 1261 DIR_INODE, /* for dirty dir inode */ 1262 FILE_INODE, /* for dirty regular/symlink inode */ 1263 DIRTY_META, /* for all dirtied inode metadata */ 1264 NR_INODE_TYPE, 1265 }; 1266 1267 /* for inner inode cache management */ 1268 struct inode_management { 1269 struct radix_tree_root ino_root; /* ino entry array */ 1270 spinlock_t ino_lock; /* for ino entry lock */ 1271 struct list_head ino_list; /* inode list head */ 1272 unsigned long ino_num; /* number of entries */ 1273 }; 1274 1275 /* for GC_AT */ 1276 struct atgc_management { 1277 bool atgc_enabled; /* ATGC is enabled or not */ 1278 struct rb_root_cached root; /* root of victim rb-tree */ 1279 struct list_head victim_list; /* linked with all victim entries */ 1280 unsigned int victim_count; /* victim count in rb-tree */ 1281 unsigned int candidate_ratio; /* candidate ratio */ 1282 unsigned int max_candidate_count; /* max candidate count */ 1283 unsigned int age_weight; /* age weight, vblock_weight = 100 - age_weight */ 1284 unsigned long long age_threshold; /* age threshold */ 1285 }; 1286 1287 struct f2fs_gc_control { 1288 unsigned int victim_segno; /* target victim segment number */ 1289 int init_gc_type; /* FG_GC or BG_GC */ 1290 bool no_bg_gc; /* check the space and stop bg_gc */ 1291 bool should_migrate_blocks; /* should migrate blocks */ 1292 bool err_gc_skipped; /* return EAGAIN if GC skipped */ 1293 unsigned int nr_free_secs; /* # of free sections to do GC */ 1294 }; 1295 1296 /* For s_flag in struct f2fs_sb_info */ 1297 enum { 1298 SBI_IS_DIRTY, /* dirty flag for checkpoint */ 1299 SBI_IS_CLOSE, /* specify unmounting */ 1300 SBI_NEED_FSCK, /* need fsck.f2fs to fix */ 1301 SBI_POR_DOING, /* recovery is doing or not */ 1302 SBI_NEED_SB_WRITE, /* need to recover superblock */ 1303 SBI_NEED_CP, /* need to checkpoint */ 1304 SBI_IS_SHUTDOWN, /* shutdown by ioctl */ 1305 SBI_IS_RECOVERED, /* recovered orphan/data */ 1306 SBI_CP_DISABLED, /* CP was disabled last mount */ 1307 SBI_CP_DISABLED_QUICK, /* CP was disabled quickly */ 1308 SBI_QUOTA_NEED_FLUSH, /* need to flush quota info in CP */ 1309 SBI_QUOTA_SKIP_FLUSH, /* skip flushing quota in current CP */ 1310 SBI_QUOTA_NEED_REPAIR, /* quota file may be corrupted */ 1311 SBI_IS_RESIZEFS, /* resizefs is in process */ 1312 SBI_IS_FREEZING, /* freezefs is in process */ 1313 }; 1314 1315 enum { 1316 CP_TIME, 1317 REQ_TIME, 1318 DISCARD_TIME, 1319 GC_TIME, 1320 DISABLE_TIME, 1321 UMOUNT_DISCARD_TIMEOUT, 1322 MAX_TIME, 1323 }; 1324 1325 /* Note that you need to keep synchronization with this gc_mode_names array */ 1326 enum { 1327 GC_NORMAL, 1328 GC_IDLE_CB, 1329 GC_IDLE_GREEDY, 1330 GC_IDLE_AT, 1331 GC_URGENT_HIGH, 1332 GC_URGENT_LOW, 1333 GC_URGENT_MID, 1334 MAX_GC_MODE, 1335 }; 1336 1337 enum { 1338 BGGC_MODE_ON, /* background gc is on */ 1339 BGGC_MODE_OFF, /* background gc is off */ 1340 BGGC_MODE_SYNC, /* 1341 * background gc is on, migrating blocks 1342 * like foreground gc 1343 */ 1344 }; 1345 1346 enum { 1347 FS_MODE_ADAPTIVE, /* use both lfs/ssr allocation */ 1348 FS_MODE_LFS, /* use lfs allocation only */ 1349 FS_MODE_FRAGMENT_SEG, /* segment fragmentation mode */ 1350 FS_MODE_FRAGMENT_BLK, /* block fragmentation mode */ 1351 }; 1352 1353 enum { 1354 ALLOC_MODE_DEFAULT, /* stay default */ 1355 ALLOC_MODE_REUSE, /* reuse segments as much as possible */ 1356 }; 1357 1358 enum fsync_mode { 1359 FSYNC_MODE_POSIX, /* fsync follows posix semantics */ 1360 FSYNC_MODE_STRICT, /* fsync behaves in line with ext4 */ 1361 FSYNC_MODE_NOBARRIER, /* fsync behaves nobarrier based on posix */ 1362 }; 1363 1364 enum { 1365 COMPR_MODE_FS, /* 1366 * automatically compress compression 1367 * enabled files 1368 */ 1369 COMPR_MODE_USER, /* 1370 * automatical compression is disabled. 1371 * user can control the file compression 1372 * using ioctls 1373 */ 1374 }; 1375 1376 enum { 1377 DISCARD_UNIT_BLOCK, /* basic discard unit is block */ 1378 DISCARD_UNIT_SEGMENT, /* basic discard unit is segment */ 1379 DISCARD_UNIT_SECTION, /* basic discard unit is section */ 1380 }; 1381 1382 enum { 1383 MEMORY_MODE_NORMAL, /* memory mode for normal devices */ 1384 MEMORY_MODE_LOW, /* memory mode for low memry devices */ 1385 }; 1386 1387 1388 1389 static inline int f2fs_test_bit(unsigned int nr, char *addr); 1390 static inline void f2fs_set_bit(unsigned int nr, char *addr); 1391 static inline void f2fs_clear_bit(unsigned int nr, char *addr); 1392 1393 /* 1394 * Layout of f2fs page.private: 1395 * 1396 * Layout A: lowest bit should be 1 1397 * | bit0 = 1 | bit1 | bit2 | ... | bit MAX | private data .... | 1398 * bit 0 PAGE_PRIVATE_NOT_POINTER 1399 * bit 1 PAGE_PRIVATE_DUMMY_WRITE 1400 * bit 2 PAGE_PRIVATE_ONGOING_MIGRATION 1401 * bit 3 PAGE_PRIVATE_INLINE_INODE 1402 * bit 4 PAGE_PRIVATE_REF_RESOURCE 1403 * bit 5- f2fs private data 1404 * 1405 * Layout B: lowest bit should be 0 1406 * page.private is a wrapped pointer. 1407 */ 1408 enum { 1409 PAGE_PRIVATE_NOT_POINTER, /* private contains non-pointer data */ 1410 PAGE_PRIVATE_DUMMY_WRITE, /* data page for padding aligned IO */ 1411 PAGE_PRIVATE_ONGOING_MIGRATION, /* data page which is on-going migrating */ 1412 PAGE_PRIVATE_INLINE_INODE, /* inode page contains inline data */ 1413 PAGE_PRIVATE_REF_RESOURCE, /* dirty page has referenced resources */ 1414 PAGE_PRIVATE_MAX 1415 }; 1416 1417 #define PAGE_PRIVATE_GET_FUNC(name, flagname) \ 1418 static inline bool page_private_##name(struct page *page) \ 1419 { \ 1420 return PagePrivate(page) && \ 1421 test_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)) && \ 1422 test_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \ 1423 } 1424 1425 #define PAGE_PRIVATE_SET_FUNC(name, flagname) \ 1426 static inline void set_page_private_##name(struct page *page) \ 1427 { \ 1428 if (!PagePrivate(page)) { \ 1429 get_page(page); \ 1430 SetPagePrivate(page); \ 1431 set_page_private(page, 0); \ 1432 } \ 1433 set_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)); \ 1434 set_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \ 1435 } 1436 1437 #define PAGE_PRIVATE_CLEAR_FUNC(name, flagname) \ 1438 static inline void clear_page_private_##name(struct page *page) \ 1439 { \ 1440 clear_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \ 1441 if (page_private(page) == 1 << PAGE_PRIVATE_NOT_POINTER) { \ 1442 set_page_private(page, 0); \ 1443 if (PagePrivate(page)) { \ 1444 ClearPagePrivate(page); \ 1445 put_page(page); \ 1446 }\ 1447 } \ 1448 } 1449 1450 PAGE_PRIVATE_GET_FUNC(nonpointer, NOT_POINTER); 1451 PAGE_PRIVATE_GET_FUNC(reference, REF_RESOURCE); 1452 PAGE_PRIVATE_GET_FUNC(inline, INLINE_INODE); 1453 PAGE_PRIVATE_GET_FUNC(gcing, ONGOING_MIGRATION); 1454 PAGE_PRIVATE_GET_FUNC(dummy, DUMMY_WRITE); 1455 1456 PAGE_PRIVATE_SET_FUNC(reference, REF_RESOURCE); 1457 PAGE_PRIVATE_SET_FUNC(inline, INLINE_INODE); 1458 PAGE_PRIVATE_SET_FUNC(gcing, ONGOING_MIGRATION); 1459 PAGE_PRIVATE_SET_FUNC(dummy, DUMMY_WRITE); 1460 1461 PAGE_PRIVATE_CLEAR_FUNC(reference, REF_RESOURCE); 1462 PAGE_PRIVATE_CLEAR_FUNC(inline, INLINE_INODE); 1463 PAGE_PRIVATE_CLEAR_FUNC(gcing, ONGOING_MIGRATION); 1464 PAGE_PRIVATE_CLEAR_FUNC(dummy, DUMMY_WRITE); 1465 1466 static inline unsigned long get_page_private_data(struct page *page) 1467 { 1468 unsigned long data = page_private(page); 1469 1470 if (!test_bit(PAGE_PRIVATE_NOT_POINTER, &data)) 1471 return 0; 1472 return data >> PAGE_PRIVATE_MAX; 1473 } 1474 1475 static inline void set_page_private_data(struct page *page, unsigned long data) 1476 { 1477 if (!PagePrivate(page)) { 1478 get_page(page); 1479 SetPagePrivate(page); 1480 set_page_private(page, 0); 1481 } 1482 set_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)); 1483 page_private(page) |= data << PAGE_PRIVATE_MAX; 1484 } 1485 1486 static inline void clear_page_private_data(struct page *page) 1487 { 1488 page_private(page) &= (1 << PAGE_PRIVATE_MAX) - 1; 1489 if (page_private(page) == 1 << PAGE_PRIVATE_NOT_POINTER) { 1490 set_page_private(page, 0); 1491 if (PagePrivate(page)) { 1492 ClearPagePrivate(page); 1493 put_page(page); 1494 } 1495 } 1496 } 1497 1498 /* For compression */ 1499 enum compress_algorithm_type { 1500 COMPRESS_LZO, 1501 COMPRESS_LZ4, 1502 COMPRESS_ZSTD, 1503 COMPRESS_LZORLE, 1504 COMPRESS_MAX, 1505 }; 1506 1507 enum compress_flag { 1508 COMPRESS_CHKSUM, 1509 COMPRESS_MAX_FLAG, 1510 }; 1511 1512 #define COMPRESS_WATERMARK 20 1513 #define COMPRESS_PERCENT 20 1514 1515 #define COMPRESS_DATA_RESERVED_SIZE 4 1516 struct compress_data { 1517 __le32 clen; /* compressed data size */ 1518 __le32 chksum; /* compressed data chksum */ 1519 __le32 reserved[COMPRESS_DATA_RESERVED_SIZE]; /* reserved */ 1520 u8 cdata[]; /* compressed data */ 1521 }; 1522 1523 #define COMPRESS_HEADER_SIZE (sizeof(struct compress_data)) 1524 1525 #define F2FS_COMPRESSED_PAGE_MAGIC 0xF5F2C000 1526 1527 #define COMPRESS_LEVEL_OFFSET 8 1528 1529 /* compress context */ 1530 struct compress_ctx { 1531 struct inode *inode; /* inode the context belong to */ 1532 pgoff_t cluster_idx; /* cluster index number */ 1533 unsigned int cluster_size; /* page count in cluster */ 1534 unsigned int log_cluster_size; /* log of cluster size */ 1535 struct page **rpages; /* pages store raw data in cluster */ 1536 unsigned int nr_rpages; /* total page number in rpages */ 1537 struct page **cpages; /* pages store compressed data in cluster */ 1538 unsigned int nr_cpages; /* total page number in cpages */ 1539 unsigned int valid_nr_cpages; /* valid page number in cpages */ 1540 void *rbuf; /* virtual mapped address on rpages */ 1541 struct compress_data *cbuf; /* virtual mapped address on cpages */ 1542 size_t rlen; /* valid data length in rbuf */ 1543 size_t clen; /* valid data length in cbuf */ 1544 void *private; /* payload buffer for specified compression algorithm */ 1545 void *private2; /* extra payload buffer */ 1546 }; 1547 1548 /* compress context for write IO path */ 1549 struct compress_io_ctx { 1550 u32 magic; /* magic number to indicate page is compressed */ 1551 struct inode *inode; /* inode the context belong to */ 1552 struct page **rpages; /* pages store raw data in cluster */ 1553 unsigned int nr_rpages; /* total page number in rpages */ 1554 atomic_t pending_pages; /* in-flight compressed page count */ 1555 }; 1556 1557 /* Context for decompressing one cluster on the read IO path */ 1558 struct decompress_io_ctx { 1559 u32 magic; /* magic number to indicate page is compressed */ 1560 struct inode *inode; /* inode the context belong to */ 1561 pgoff_t cluster_idx; /* cluster index number */ 1562 unsigned int cluster_size; /* page count in cluster */ 1563 unsigned int log_cluster_size; /* log of cluster size */ 1564 struct page **rpages; /* pages store raw data in cluster */ 1565 unsigned int nr_rpages; /* total page number in rpages */ 1566 struct page **cpages; /* pages store compressed data in cluster */ 1567 unsigned int nr_cpages; /* total page number in cpages */ 1568 struct page **tpages; /* temp pages to pad holes in cluster */ 1569 void *rbuf; /* virtual mapped address on rpages */ 1570 struct compress_data *cbuf; /* virtual mapped address on cpages */ 1571 size_t rlen; /* valid data length in rbuf */ 1572 size_t clen; /* valid data length in cbuf */ 1573 1574 /* 1575 * The number of compressed pages remaining to be read in this cluster. 1576 * This is initially nr_cpages. It is decremented by 1 each time a page 1577 * has been read (or failed to be read). When it reaches 0, the cluster 1578 * is decompressed (or an error is reported). 1579 * 1580 * If an error occurs before all the pages have been submitted for I/O, 1581 * then this will never reach 0. In this case the I/O submitter is 1582 * responsible for calling f2fs_decompress_end_io() instead. 1583 */ 1584 atomic_t remaining_pages; 1585 1586 /* 1587 * Number of references to this decompress_io_ctx. 1588 * 1589 * One reference is held for I/O completion. This reference is dropped 1590 * after the pagecache pages are updated and unlocked -- either after 1591 * decompression (and verity if enabled), or after an error. 1592 * 1593 * In addition, each compressed page holds a reference while it is in a 1594 * bio. These references are necessary prevent compressed pages from 1595 * being freed while they are still in a bio. 1596 */ 1597 refcount_t refcnt; 1598 1599 bool failed; /* IO error occurred before decompression? */ 1600 bool need_verity; /* need fs-verity verification after decompression? */ 1601 void *private; /* payload buffer for specified decompression algorithm */ 1602 void *private2; /* extra payload buffer */ 1603 struct work_struct verity_work; /* work to verify the decompressed pages */ 1604 struct work_struct free_work; /* work for late free this structure itself */ 1605 }; 1606 1607 #define NULL_CLUSTER ((unsigned int)(~0)) 1608 #define MIN_COMPRESS_LOG_SIZE 2 1609 #define MAX_COMPRESS_LOG_SIZE 8 1610 #define MAX_COMPRESS_WINDOW_SIZE(log_size) ((PAGE_SIZE) << (log_size)) 1611 1612 struct f2fs_sb_info { 1613 struct super_block *sb; /* pointer to VFS super block */ 1614 struct proc_dir_entry *s_proc; /* proc entry */ 1615 struct f2fs_super_block *raw_super; /* raw super block pointer */ 1616 struct f2fs_rwsem sb_lock; /* lock for raw super block */ 1617 int valid_super_block; /* valid super block no */ 1618 unsigned long s_flag; /* flags for sbi */ 1619 struct mutex writepages; /* mutex for writepages() */ 1620 1621 #ifdef CONFIG_BLK_DEV_ZONED 1622 unsigned int blocks_per_blkz; /* F2FS blocks per zone */ 1623 unsigned int log_blocks_per_blkz; /* log2 F2FS blocks per zone */ 1624 #endif 1625 1626 /* for node-related operations */ 1627 struct f2fs_nm_info *nm_info; /* node manager */ 1628 struct inode *node_inode; /* cache node blocks */ 1629 1630 /* for segment-related operations */ 1631 struct f2fs_sm_info *sm_info; /* segment manager */ 1632 1633 /* for bio operations */ 1634 struct f2fs_bio_info *write_io[NR_PAGE_TYPE]; /* for write bios */ 1635 /* keep migration IO order for LFS mode */ 1636 struct f2fs_rwsem io_order_lock; 1637 mempool_t *write_io_dummy; /* Dummy pages */ 1638 pgoff_t page_eio_ofs[NR_PAGE_TYPE]; /* EIO page offset */ 1639 int page_eio_cnt[NR_PAGE_TYPE]; /* EIO count */ 1640 1641 /* for checkpoint */ 1642 struct f2fs_checkpoint *ckpt; /* raw checkpoint pointer */ 1643 int cur_cp_pack; /* remain current cp pack */ 1644 spinlock_t cp_lock; /* for flag in ckpt */ 1645 struct inode *meta_inode; /* cache meta blocks */ 1646 struct f2fs_rwsem cp_global_sem; /* checkpoint procedure lock */ 1647 struct f2fs_rwsem cp_rwsem; /* blocking FS operations */ 1648 struct f2fs_rwsem node_write; /* locking node writes */ 1649 struct f2fs_rwsem node_change; /* locking node change */ 1650 wait_queue_head_t cp_wait; 1651 unsigned long last_time[MAX_TIME]; /* to store time in jiffies */ 1652 long interval_time[MAX_TIME]; /* to store thresholds */ 1653 struct ckpt_req_control cprc_info; /* for checkpoint request control */ 1654 1655 struct inode_management im[MAX_INO_ENTRY]; /* manage inode cache */ 1656 1657 spinlock_t fsync_node_lock; /* for node entry lock */ 1658 struct list_head fsync_node_list; /* node list head */ 1659 unsigned int fsync_seg_id; /* sequence id */ 1660 unsigned int fsync_node_num; /* number of node entries */ 1661 1662 /* for orphan inode, use 0'th array */ 1663 unsigned int max_orphans; /* max orphan inodes */ 1664 1665 /* for inode management */ 1666 struct list_head inode_list[NR_INODE_TYPE]; /* dirty inode list */ 1667 spinlock_t inode_lock[NR_INODE_TYPE]; /* for dirty inode list lock */ 1668 struct mutex flush_lock; /* for flush exclusion */ 1669 1670 /* for extent tree cache */ 1671 struct extent_tree_info extent_tree[NR_EXTENT_CACHES]; 1672 atomic64_t allocated_data_blocks; /* for block age extent_cache */ 1673 1674 /* The threshold used for hot and warm data seperation*/ 1675 unsigned int hot_data_age_threshold; 1676 unsigned int warm_data_age_threshold; 1677 1678 /* basic filesystem units */ 1679 unsigned int log_sectors_per_block; /* log2 sectors per block */ 1680 unsigned int log_blocksize; /* log2 block size */ 1681 unsigned int blocksize; /* block size */ 1682 unsigned int root_ino_num; /* root inode number*/ 1683 unsigned int node_ino_num; /* node inode number*/ 1684 unsigned int meta_ino_num; /* meta inode number*/ 1685 unsigned int log_blocks_per_seg; /* log2 blocks per segment */ 1686 unsigned int blocks_per_seg; /* blocks per segment */ 1687 unsigned int unusable_blocks_per_sec; /* unusable blocks per section */ 1688 unsigned int segs_per_sec; /* segments per section */ 1689 unsigned int secs_per_zone; /* sections per zone */ 1690 unsigned int total_sections; /* total section count */ 1691 unsigned int total_node_count; /* total node block count */ 1692 unsigned int total_valid_node_count; /* valid node block count */ 1693 int dir_level; /* directory level */ 1694 bool readdir_ra; /* readahead inode in readdir */ 1695 u64 max_io_bytes; /* max io bytes to merge IOs */ 1696 1697 block_t user_block_count; /* # of user blocks */ 1698 block_t total_valid_block_count; /* # of valid blocks */ 1699 block_t discard_blks; /* discard command candidats */ 1700 block_t last_valid_block_count; /* for recovery */ 1701 block_t reserved_blocks; /* configurable reserved blocks */ 1702 block_t current_reserved_blocks; /* current reserved blocks */ 1703 1704 /* Additional tracking for no checkpoint mode */ 1705 block_t unusable_block_count; /* # of blocks saved by last cp */ 1706 1707 unsigned int nquota_files; /* # of quota sysfile */ 1708 struct f2fs_rwsem quota_sem; /* blocking cp for flags */ 1709 1710 /* # of pages, see count_type */ 1711 atomic_t nr_pages[NR_COUNT_TYPE]; 1712 /* # of allocated blocks */ 1713 struct percpu_counter alloc_valid_block_count; 1714 /* # of node block writes as roll forward recovery */ 1715 struct percpu_counter rf_node_block_count; 1716 1717 /* writeback control */ 1718 atomic_t wb_sync_req[META]; /* count # of WB_SYNC threads */ 1719 1720 /* valid inode count */ 1721 struct percpu_counter total_valid_inode_count; 1722 1723 struct f2fs_mount_info mount_opt; /* mount options */ 1724 1725 /* for cleaning operations */ 1726 struct f2fs_rwsem gc_lock; /* 1727 * semaphore for GC, avoid 1728 * race between GC and GC or CP 1729 */ 1730 struct f2fs_gc_kthread *gc_thread; /* GC thread */ 1731 struct atgc_management am; /* atgc management */ 1732 unsigned int cur_victim_sec; /* current victim section num */ 1733 unsigned int gc_mode; /* current GC state */ 1734 unsigned int next_victim_seg[2]; /* next segment in victim section */ 1735 spinlock_t gc_remaining_trials_lock; 1736 /* remaining trial count for GC_URGENT_* and GC_IDLE_* */ 1737 unsigned int gc_remaining_trials; 1738 1739 /* for skip statistic */ 1740 unsigned long long skipped_gc_rwsem; /* FG_GC only */ 1741 1742 /* threshold for gc trials on pinned files */ 1743 u64 gc_pin_file_threshold; 1744 struct f2fs_rwsem pin_sem; 1745 1746 /* maximum # of trials to find a victim segment for SSR and GC */ 1747 unsigned int max_victim_search; 1748 /* migration granularity of garbage collection, unit: segment */ 1749 unsigned int migration_granularity; 1750 1751 /* 1752 * for stat information. 1753 * one is for the LFS mode, and the other is for the SSR mode. 1754 */ 1755 #ifdef CONFIG_F2FS_STAT_FS 1756 struct f2fs_stat_info *stat_info; /* FS status information */ 1757 atomic_t meta_count[META_MAX]; /* # of meta blocks */ 1758 unsigned int segment_count[2]; /* # of allocated segments */ 1759 unsigned int block_count[2]; /* # of allocated blocks */ 1760 atomic_t inplace_count; /* # of inplace update */ 1761 /* # of lookup extent cache */ 1762 atomic64_t total_hit_ext[NR_EXTENT_CACHES]; 1763 /* # of hit rbtree extent node */ 1764 atomic64_t read_hit_rbtree[NR_EXTENT_CACHES]; 1765 /* # of hit cached extent node */ 1766 atomic64_t read_hit_cached[NR_EXTENT_CACHES]; 1767 /* # of hit largest extent node in read extent cache */ 1768 atomic64_t read_hit_largest; 1769 atomic_t inline_xattr; /* # of inline_xattr inodes */ 1770 atomic_t inline_inode; /* # of inline_data inodes */ 1771 atomic_t inline_dir; /* # of inline_dentry inodes */ 1772 atomic_t compr_inode; /* # of compressed inodes */ 1773 atomic64_t compr_blocks; /* # of compressed blocks */ 1774 atomic_t swapfile_inode; /* # of swapfile inodes */ 1775 atomic_t atomic_files; /* # of opened atomic file */ 1776 atomic_t max_aw_cnt; /* max # of atomic writes */ 1777 unsigned int io_skip_bggc; /* skip background gc for in-flight IO */ 1778 unsigned int other_skip_bggc; /* skip background gc for other reasons */ 1779 unsigned int ndirty_inode[NR_INODE_TYPE]; /* # of dirty inodes */ 1780 #endif 1781 spinlock_t stat_lock; /* lock for stat operations */ 1782 1783 /* to attach REQ_META|REQ_FUA flags */ 1784 unsigned int data_io_flag; 1785 unsigned int node_io_flag; 1786 1787 /* For sysfs support */ 1788 struct kobject s_kobj; /* /sys/fs/f2fs/<devname> */ 1789 struct completion s_kobj_unregister; 1790 1791 struct kobject s_stat_kobj; /* /sys/fs/f2fs/<devname>/stat */ 1792 struct completion s_stat_kobj_unregister; 1793 1794 struct kobject s_feature_list_kobj; /* /sys/fs/f2fs/<devname>/feature_list */ 1795 struct completion s_feature_list_kobj_unregister; 1796 1797 /* For shrinker support */ 1798 struct list_head s_list; 1799 struct mutex umount_mutex; 1800 unsigned int shrinker_run_no; 1801 1802 /* For multi devices */ 1803 int s_ndevs; /* number of devices */ 1804 struct f2fs_dev_info *devs; /* for device list */ 1805 unsigned int dirty_device; /* for checkpoint data flush */ 1806 spinlock_t dev_lock; /* protect dirty_device */ 1807 bool aligned_blksize; /* all devices has the same logical blksize */ 1808 1809 /* For write statistics */ 1810 u64 sectors_written_start; 1811 u64 kbytes_written; 1812 1813 /* Reference to checksum algorithm driver via cryptoapi */ 1814 struct crypto_shash *s_chksum_driver; 1815 1816 /* Precomputed FS UUID checksum for seeding other checksums */ 1817 __u32 s_chksum_seed; 1818 1819 struct workqueue_struct *post_read_wq; /* post read workqueue */ 1820 1821 unsigned char errors[MAX_F2FS_ERRORS]; /* error flags */ 1822 spinlock_t error_lock; /* protect errors array */ 1823 bool error_dirty; /* errors of sb is dirty */ 1824 1825 struct kmem_cache *inline_xattr_slab; /* inline xattr entry */ 1826 unsigned int inline_xattr_slab_size; /* default inline xattr slab size */ 1827 1828 /* For reclaimed segs statistics per each GC mode */ 1829 unsigned int gc_segment_mode; /* GC state for reclaimed segments */ 1830 unsigned int gc_reclaimed_segs[MAX_GC_MODE]; /* Reclaimed segs for each mode */ 1831 1832 unsigned long seq_file_ra_mul; /* multiplier for ra_pages of seq. files in fadvise */ 1833 1834 int max_fragment_chunk; /* max chunk size for block fragmentation mode */ 1835 int max_fragment_hole; /* max hole size for block fragmentation mode */ 1836 1837 /* For atomic write statistics */ 1838 atomic64_t current_atomic_write; 1839 s64 peak_atomic_write; 1840 u64 committed_atomic_block; 1841 u64 revoked_atomic_block; 1842 1843 #ifdef CONFIG_F2FS_FS_COMPRESSION 1844 struct kmem_cache *page_array_slab; /* page array entry */ 1845 unsigned int page_array_slab_size; /* default page array slab size */ 1846 1847 /* For runtime compression statistics */ 1848 u64 compr_written_block; 1849 u64 compr_saved_block; 1850 u32 compr_new_inode; 1851 1852 /* For compressed block cache */ 1853 struct inode *compress_inode; /* cache compressed blocks */ 1854 unsigned int compress_percent; /* cache page percentage */ 1855 unsigned int compress_watermark; /* cache page watermark */ 1856 atomic_t compress_page_hit; /* cache hit count */ 1857 #endif 1858 1859 #ifdef CONFIG_F2FS_IOSTAT 1860 /* For app/fs IO statistics */ 1861 spinlock_t iostat_lock; 1862 unsigned long long rw_iostat[NR_IO_TYPE]; 1863 unsigned long long prev_rw_iostat[NR_IO_TYPE]; 1864 bool iostat_enable; 1865 unsigned long iostat_next_period; 1866 unsigned int iostat_period_ms; 1867 1868 /* For io latency related statistics info in one iostat period */ 1869 spinlock_t iostat_lat_lock; 1870 struct iostat_lat_info *iostat_io_lat; 1871 #endif 1872 }; 1873 1874 #ifdef CONFIG_F2FS_FAULT_INJECTION 1875 #define f2fs_show_injection_info(sbi, type) \ 1876 printk_ratelimited("%sF2FS-fs (%s) : inject %s in %s of %pS\n", \ 1877 KERN_INFO, sbi->sb->s_id, \ 1878 f2fs_fault_name[type], \ 1879 __func__, __builtin_return_address(0)) 1880 static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type) 1881 { 1882 struct f2fs_fault_info *ffi = &F2FS_OPTION(sbi).fault_info; 1883 1884 if (!ffi->inject_rate) 1885 return false; 1886 1887 if (!IS_FAULT_SET(ffi, type)) 1888 return false; 1889 1890 atomic_inc(&ffi->inject_ops); 1891 if (atomic_read(&ffi->inject_ops) >= ffi->inject_rate) { 1892 atomic_set(&ffi->inject_ops, 0); 1893 return true; 1894 } 1895 return false; 1896 } 1897 #else 1898 #define f2fs_show_injection_info(sbi, type) do { } while (0) 1899 static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type) 1900 { 1901 return false; 1902 } 1903 #endif 1904 1905 /* 1906 * Test if the mounted volume is a multi-device volume. 1907 * - For a single regular disk volume, sbi->s_ndevs is 0. 1908 * - For a single zoned disk volume, sbi->s_ndevs is 1. 1909 * - For a multi-device volume, sbi->s_ndevs is always 2 or more. 1910 */ 1911 static inline bool f2fs_is_multi_device(struct f2fs_sb_info *sbi) 1912 { 1913 return sbi->s_ndevs > 1; 1914 } 1915 1916 static inline void f2fs_update_time(struct f2fs_sb_info *sbi, int type) 1917 { 1918 unsigned long now = jiffies; 1919 1920 sbi->last_time[type] = now; 1921 1922 /* DISCARD_TIME and GC_TIME are based on REQ_TIME */ 1923 if (type == REQ_TIME) { 1924 sbi->last_time[DISCARD_TIME] = now; 1925 sbi->last_time[GC_TIME] = now; 1926 } 1927 } 1928 1929 static inline bool f2fs_time_over(struct f2fs_sb_info *sbi, int type) 1930 { 1931 unsigned long interval = sbi->interval_time[type] * HZ; 1932 1933 return time_after(jiffies, sbi->last_time[type] + interval); 1934 } 1935 1936 static inline unsigned int f2fs_time_to_wait(struct f2fs_sb_info *sbi, 1937 int type) 1938 { 1939 unsigned long interval = sbi->interval_time[type] * HZ; 1940 unsigned int wait_ms = 0; 1941 long delta; 1942 1943 delta = (sbi->last_time[type] + interval) - jiffies; 1944 if (delta > 0) 1945 wait_ms = jiffies_to_msecs(delta); 1946 1947 return wait_ms; 1948 } 1949 1950 /* 1951 * Inline functions 1952 */ 1953 static inline u32 __f2fs_crc32(struct f2fs_sb_info *sbi, u32 crc, 1954 const void *address, unsigned int length) 1955 { 1956 struct { 1957 struct shash_desc shash; 1958 char ctx[4]; 1959 } desc; 1960 int err; 1961 1962 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver) != sizeof(desc.ctx)); 1963 1964 desc.shash.tfm = sbi->s_chksum_driver; 1965 *(u32 *)desc.ctx = crc; 1966 1967 err = crypto_shash_update(&desc.shash, address, length); 1968 BUG_ON(err); 1969 1970 return *(u32 *)desc.ctx; 1971 } 1972 1973 static inline u32 f2fs_crc32(struct f2fs_sb_info *sbi, const void *address, 1974 unsigned int length) 1975 { 1976 return __f2fs_crc32(sbi, F2FS_SUPER_MAGIC, address, length); 1977 } 1978 1979 static inline bool f2fs_crc_valid(struct f2fs_sb_info *sbi, __u32 blk_crc, 1980 void *buf, size_t buf_size) 1981 { 1982 return f2fs_crc32(sbi, buf, buf_size) == blk_crc; 1983 } 1984 1985 static inline u32 f2fs_chksum(struct f2fs_sb_info *sbi, u32 crc, 1986 const void *address, unsigned int length) 1987 { 1988 return __f2fs_crc32(sbi, crc, address, length); 1989 } 1990 1991 static inline struct f2fs_inode_info *F2FS_I(struct inode *inode) 1992 { 1993 return container_of(inode, struct f2fs_inode_info, vfs_inode); 1994 } 1995 1996 static inline struct f2fs_sb_info *F2FS_SB(struct super_block *sb) 1997 { 1998 return sb->s_fs_info; 1999 } 2000 2001 static inline struct f2fs_sb_info *F2FS_I_SB(struct inode *inode) 2002 { 2003 return F2FS_SB(inode->i_sb); 2004 } 2005 2006 static inline struct f2fs_sb_info *F2FS_M_SB(struct address_space *mapping) 2007 { 2008 return F2FS_I_SB(mapping->host); 2009 } 2010 2011 static inline struct f2fs_sb_info *F2FS_P_SB(struct page *page) 2012 { 2013 return F2FS_M_SB(page_file_mapping(page)); 2014 } 2015 2016 static inline struct f2fs_super_block *F2FS_RAW_SUPER(struct f2fs_sb_info *sbi) 2017 { 2018 return (struct f2fs_super_block *)(sbi->raw_super); 2019 } 2020 2021 static inline struct f2fs_checkpoint *F2FS_CKPT(struct f2fs_sb_info *sbi) 2022 { 2023 return (struct f2fs_checkpoint *)(sbi->ckpt); 2024 } 2025 2026 static inline struct f2fs_node *F2FS_NODE(struct page *page) 2027 { 2028 return (struct f2fs_node *)page_address(page); 2029 } 2030 2031 static inline struct f2fs_inode *F2FS_INODE(struct page *page) 2032 { 2033 return &((struct f2fs_node *)page_address(page))->i; 2034 } 2035 2036 static inline struct f2fs_nm_info *NM_I(struct f2fs_sb_info *sbi) 2037 { 2038 return (struct f2fs_nm_info *)(sbi->nm_info); 2039 } 2040 2041 static inline struct f2fs_sm_info *SM_I(struct f2fs_sb_info *sbi) 2042 { 2043 return (struct f2fs_sm_info *)(sbi->sm_info); 2044 } 2045 2046 static inline struct sit_info *SIT_I(struct f2fs_sb_info *sbi) 2047 { 2048 return (struct sit_info *)(SM_I(sbi)->sit_info); 2049 } 2050 2051 static inline struct free_segmap_info *FREE_I(struct f2fs_sb_info *sbi) 2052 { 2053 return (struct free_segmap_info *)(SM_I(sbi)->free_info); 2054 } 2055 2056 static inline struct dirty_seglist_info *DIRTY_I(struct f2fs_sb_info *sbi) 2057 { 2058 return (struct dirty_seglist_info *)(SM_I(sbi)->dirty_info); 2059 } 2060 2061 static inline struct address_space *META_MAPPING(struct f2fs_sb_info *sbi) 2062 { 2063 return sbi->meta_inode->i_mapping; 2064 } 2065 2066 static inline struct address_space *NODE_MAPPING(struct f2fs_sb_info *sbi) 2067 { 2068 return sbi->node_inode->i_mapping; 2069 } 2070 2071 static inline bool is_sbi_flag_set(struct f2fs_sb_info *sbi, unsigned int type) 2072 { 2073 return test_bit(type, &sbi->s_flag); 2074 } 2075 2076 static inline void set_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type) 2077 { 2078 set_bit(type, &sbi->s_flag); 2079 } 2080 2081 static inline void clear_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type) 2082 { 2083 clear_bit(type, &sbi->s_flag); 2084 } 2085 2086 static inline unsigned long long cur_cp_version(struct f2fs_checkpoint *cp) 2087 { 2088 return le64_to_cpu(cp->checkpoint_ver); 2089 } 2090 2091 static inline unsigned long f2fs_qf_ino(struct super_block *sb, int type) 2092 { 2093 if (type < F2FS_MAX_QUOTAS) 2094 return le32_to_cpu(F2FS_SB(sb)->raw_super->qf_ino[type]); 2095 return 0; 2096 } 2097 2098 static inline __u64 cur_cp_crc(struct f2fs_checkpoint *cp) 2099 { 2100 size_t crc_offset = le32_to_cpu(cp->checksum_offset); 2101 return le32_to_cpu(*((__le32 *)((unsigned char *)cp + crc_offset))); 2102 } 2103 2104 static inline bool __is_set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 2105 { 2106 unsigned int ckpt_flags = le32_to_cpu(cp->ckpt_flags); 2107 2108 return ckpt_flags & f; 2109 } 2110 2111 static inline bool is_set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 2112 { 2113 return __is_set_ckpt_flags(F2FS_CKPT(sbi), f); 2114 } 2115 2116 static inline void __set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 2117 { 2118 unsigned int ckpt_flags; 2119 2120 ckpt_flags = le32_to_cpu(cp->ckpt_flags); 2121 ckpt_flags |= f; 2122 cp->ckpt_flags = cpu_to_le32(ckpt_flags); 2123 } 2124 2125 static inline void set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 2126 { 2127 unsigned long flags; 2128 2129 spin_lock_irqsave(&sbi->cp_lock, flags); 2130 __set_ckpt_flags(F2FS_CKPT(sbi), f); 2131 spin_unlock_irqrestore(&sbi->cp_lock, flags); 2132 } 2133 2134 static inline void __clear_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 2135 { 2136 unsigned int ckpt_flags; 2137 2138 ckpt_flags = le32_to_cpu(cp->ckpt_flags); 2139 ckpt_flags &= (~f); 2140 cp->ckpt_flags = cpu_to_le32(ckpt_flags); 2141 } 2142 2143 static inline void clear_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 2144 { 2145 unsigned long flags; 2146 2147 spin_lock_irqsave(&sbi->cp_lock, flags); 2148 __clear_ckpt_flags(F2FS_CKPT(sbi), f); 2149 spin_unlock_irqrestore(&sbi->cp_lock, flags); 2150 } 2151 2152 #define init_f2fs_rwsem(sem) \ 2153 do { \ 2154 static struct lock_class_key __key; \ 2155 \ 2156 __init_f2fs_rwsem((sem), #sem, &__key); \ 2157 } while (0) 2158 2159 static inline void __init_f2fs_rwsem(struct f2fs_rwsem *sem, 2160 const char *sem_name, struct lock_class_key *key) 2161 { 2162 __init_rwsem(&sem->internal_rwsem, sem_name, key); 2163 #ifdef CONFIG_F2FS_UNFAIR_RWSEM 2164 init_waitqueue_head(&sem->read_waiters); 2165 #endif 2166 } 2167 2168 static inline int f2fs_rwsem_is_locked(struct f2fs_rwsem *sem) 2169 { 2170 return rwsem_is_locked(&sem->internal_rwsem); 2171 } 2172 2173 static inline int f2fs_rwsem_is_contended(struct f2fs_rwsem *sem) 2174 { 2175 return rwsem_is_contended(&sem->internal_rwsem); 2176 } 2177 2178 static inline void f2fs_down_read(struct f2fs_rwsem *sem) 2179 { 2180 #ifdef CONFIG_F2FS_UNFAIR_RWSEM 2181 wait_event(sem->read_waiters, down_read_trylock(&sem->internal_rwsem)); 2182 #else 2183 down_read(&sem->internal_rwsem); 2184 #endif 2185 } 2186 2187 static inline int f2fs_down_read_trylock(struct f2fs_rwsem *sem) 2188 { 2189 return down_read_trylock(&sem->internal_rwsem); 2190 } 2191 2192 #ifdef CONFIG_DEBUG_LOCK_ALLOC 2193 static inline void f2fs_down_read_nested(struct f2fs_rwsem *sem, int subclass) 2194 { 2195 down_read_nested(&sem->internal_rwsem, subclass); 2196 } 2197 #else 2198 #define f2fs_down_read_nested(sem, subclass) f2fs_down_read(sem) 2199 #endif 2200 2201 static inline void f2fs_up_read(struct f2fs_rwsem *sem) 2202 { 2203 up_read(&sem->internal_rwsem); 2204 } 2205 2206 static inline void f2fs_down_write(struct f2fs_rwsem *sem) 2207 { 2208 down_write(&sem->internal_rwsem); 2209 } 2210 2211 static inline int f2fs_down_write_trylock(struct f2fs_rwsem *sem) 2212 { 2213 return down_write_trylock(&sem->internal_rwsem); 2214 } 2215 2216 static inline void f2fs_up_write(struct f2fs_rwsem *sem) 2217 { 2218 up_write(&sem->internal_rwsem); 2219 #ifdef CONFIG_F2FS_UNFAIR_RWSEM 2220 wake_up_all(&sem->read_waiters); 2221 #endif 2222 } 2223 2224 static inline void f2fs_lock_op(struct f2fs_sb_info *sbi) 2225 { 2226 f2fs_down_read(&sbi->cp_rwsem); 2227 } 2228 2229 static inline int f2fs_trylock_op(struct f2fs_sb_info *sbi) 2230 { 2231 if (time_to_inject(sbi, FAULT_LOCK_OP)) { 2232 f2fs_show_injection_info(sbi, FAULT_LOCK_OP); 2233 return 0; 2234 } 2235 return f2fs_down_read_trylock(&sbi->cp_rwsem); 2236 } 2237 2238 static inline void f2fs_unlock_op(struct f2fs_sb_info *sbi) 2239 { 2240 f2fs_up_read(&sbi->cp_rwsem); 2241 } 2242 2243 static inline void f2fs_lock_all(struct f2fs_sb_info *sbi) 2244 { 2245 f2fs_down_write(&sbi->cp_rwsem); 2246 } 2247 2248 static inline void f2fs_unlock_all(struct f2fs_sb_info *sbi) 2249 { 2250 f2fs_up_write(&sbi->cp_rwsem); 2251 } 2252 2253 static inline int __get_cp_reason(struct f2fs_sb_info *sbi) 2254 { 2255 int reason = CP_SYNC; 2256 2257 if (test_opt(sbi, FASTBOOT)) 2258 reason = CP_FASTBOOT; 2259 if (is_sbi_flag_set(sbi, SBI_IS_CLOSE)) 2260 reason = CP_UMOUNT; 2261 return reason; 2262 } 2263 2264 static inline bool __remain_node_summaries(int reason) 2265 { 2266 return (reason & (CP_UMOUNT | CP_FASTBOOT)); 2267 } 2268 2269 static inline bool __exist_node_summaries(struct f2fs_sb_info *sbi) 2270 { 2271 return (is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG) || 2272 is_set_ckpt_flags(sbi, CP_FASTBOOT_FLAG)); 2273 } 2274 2275 /* 2276 * Check whether the inode has blocks or not 2277 */ 2278 static inline int F2FS_HAS_BLOCKS(struct inode *inode) 2279 { 2280 block_t xattr_block = F2FS_I(inode)->i_xattr_nid ? 1 : 0; 2281 2282 return (inode->i_blocks >> F2FS_LOG_SECTORS_PER_BLOCK) > xattr_block; 2283 } 2284 2285 static inline bool f2fs_has_xattr_block(unsigned int ofs) 2286 { 2287 return ofs == XATTR_NODE_OFFSET; 2288 } 2289 2290 static inline bool __allow_reserved_blocks(struct f2fs_sb_info *sbi, 2291 struct inode *inode, bool cap) 2292 { 2293 if (!inode) 2294 return true; 2295 if (!test_opt(sbi, RESERVE_ROOT)) 2296 return false; 2297 if (IS_NOQUOTA(inode)) 2298 return true; 2299 if (uid_eq(F2FS_OPTION(sbi).s_resuid, current_fsuid())) 2300 return true; 2301 if (!gid_eq(F2FS_OPTION(sbi).s_resgid, GLOBAL_ROOT_GID) && 2302 in_group_p(F2FS_OPTION(sbi).s_resgid)) 2303 return true; 2304 if (cap && capable(CAP_SYS_RESOURCE)) 2305 return true; 2306 return false; 2307 } 2308 2309 static inline void f2fs_i_blocks_write(struct inode *, block_t, bool, bool); 2310 static inline int inc_valid_block_count(struct f2fs_sb_info *sbi, 2311 struct inode *inode, blkcnt_t *count) 2312 { 2313 blkcnt_t diff = 0, release = 0; 2314 block_t avail_user_block_count; 2315 int ret; 2316 2317 ret = dquot_reserve_block(inode, *count); 2318 if (ret) 2319 return ret; 2320 2321 if (time_to_inject(sbi, FAULT_BLOCK)) { 2322 f2fs_show_injection_info(sbi, FAULT_BLOCK); 2323 release = *count; 2324 goto release_quota; 2325 } 2326 2327 /* 2328 * let's increase this in prior to actual block count change in order 2329 * for f2fs_sync_file to avoid data races when deciding checkpoint. 2330 */ 2331 percpu_counter_add(&sbi->alloc_valid_block_count, (*count)); 2332 2333 spin_lock(&sbi->stat_lock); 2334 sbi->total_valid_block_count += (block_t)(*count); 2335 avail_user_block_count = sbi->user_block_count - 2336 sbi->current_reserved_blocks; 2337 2338 if (!__allow_reserved_blocks(sbi, inode, true)) 2339 avail_user_block_count -= F2FS_OPTION(sbi).root_reserved_blocks; 2340 2341 if (F2FS_IO_ALIGNED(sbi)) 2342 avail_user_block_count -= sbi->blocks_per_seg * 2343 SM_I(sbi)->additional_reserved_segments; 2344 2345 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) { 2346 if (avail_user_block_count > sbi->unusable_block_count) 2347 avail_user_block_count -= sbi->unusable_block_count; 2348 else 2349 avail_user_block_count = 0; 2350 } 2351 if (unlikely(sbi->total_valid_block_count > avail_user_block_count)) { 2352 diff = sbi->total_valid_block_count - avail_user_block_count; 2353 if (diff > *count) 2354 diff = *count; 2355 *count -= diff; 2356 release = diff; 2357 sbi->total_valid_block_count -= diff; 2358 if (!*count) { 2359 spin_unlock(&sbi->stat_lock); 2360 goto enospc; 2361 } 2362 } 2363 spin_unlock(&sbi->stat_lock); 2364 2365 if (unlikely(release)) { 2366 percpu_counter_sub(&sbi->alloc_valid_block_count, release); 2367 dquot_release_reservation_block(inode, release); 2368 } 2369 f2fs_i_blocks_write(inode, *count, true, true); 2370 return 0; 2371 2372 enospc: 2373 percpu_counter_sub(&sbi->alloc_valid_block_count, release); 2374 release_quota: 2375 dquot_release_reservation_block(inode, release); 2376 return -ENOSPC; 2377 } 2378 2379 __printf(2, 3) 2380 void f2fs_printk(struct f2fs_sb_info *sbi, const char *fmt, ...); 2381 2382 #define f2fs_err(sbi, fmt, ...) \ 2383 f2fs_printk(sbi, KERN_ERR fmt, ##__VA_ARGS__) 2384 #define f2fs_warn(sbi, fmt, ...) \ 2385 f2fs_printk(sbi, KERN_WARNING fmt, ##__VA_ARGS__) 2386 #define f2fs_notice(sbi, fmt, ...) \ 2387 f2fs_printk(sbi, KERN_NOTICE fmt, ##__VA_ARGS__) 2388 #define f2fs_info(sbi, fmt, ...) \ 2389 f2fs_printk(sbi, KERN_INFO fmt, ##__VA_ARGS__) 2390 #define f2fs_debug(sbi, fmt, ...) \ 2391 f2fs_printk(sbi, KERN_DEBUG fmt, ##__VA_ARGS__) 2392 2393 static inline void dec_valid_block_count(struct f2fs_sb_info *sbi, 2394 struct inode *inode, 2395 block_t count) 2396 { 2397 blkcnt_t sectors = count << F2FS_LOG_SECTORS_PER_BLOCK; 2398 2399 spin_lock(&sbi->stat_lock); 2400 f2fs_bug_on(sbi, sbi->total_valid_block_count < (block_t) count); 2401 sbi->total_valid_block_count -= (block_t)count; 2402 if (sbi->reserved_blocks && 2403 sbi->current_reserved_blocks < sbi->reserved_blocks) 2404 sbi->current_reserved_blocks = min(sbi->reserved_blocks, 2405 sbi->current_reserved_blocks + count); 2406 spin_unlock(&sbi->stat_lock); 2407 if (unlikely(inode->i_blocks < sectors)) { 2408 f2fs_warn(sbi, "Inconsistent i_blocks, ino:%lu, iblocks:%llu, sectors:%llu", 2409 inode->i_ino, 2410 (unsigned long long)inode->i_blocks, 2411 (unsigned long long)sectors); 2412 set_sbi_flag(sbi, SBI_NEED_FSCK); 2413 return; 2414 } 2415 f2fs_i_blocks_write(inode, count, false, true); 2416 } 2417 2418 static inline void inc_page_count(struct f2fs_sb_info *sbi, int count_type) 2419 { 2420 atomic_inc(&sbi->nr_pages[count_type]); 2421 2422 if (count_type == F2FS_DIRTY_DENTS || 2423 count_type == F2FS_DIRTY_NODES || 2424 count_type == F2FS_DIRTY_META || 2425 count_type == F2FS_DIRTY_QDATA || 2426 count_type == F2FS_DIRTY_IMETA) 2427 set_sbi_flag(sbi, SBI_IS_DIRTY); 2428 } 2429 2430 static inline void inode_inc_dirty_pages(struct inode *inode) 2431 { 2432 atomic_inc(&F2FS_I(inode)->dirty_pages); 2433 inc_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ? 2434 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA); 2435 if (IS_NOQUOTA(inode)) 2436 inc_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA); 2437 } 2438 2439 static inline void dec_page_count(struct f2fs_sb_info *sbi, int count_type) 2440 { 2441 atomic_dec(&sbi->nr_pages[count_type]); 2442 } 2443 2444 static inline void inode_dec_dirty_pages(struct inode *inode) 2445 { 2446 if (!S_ISDIR(inode->i_mode) && !S_ISREG(inode->i_mode) && 2447 !S_ISLNK(inode->i_mode)) 2448 return; 2449 2450 atomic_dec(&F2FS_I(inode)->dirty_pages); 2451 dec_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ? 2452 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA); 2453 if (IS_NOQUOTA(inode)) 2454 dec_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA); 2455 } 2456 2457 static inline void inc_atomic_write_cnt(struct inode *inode) 2458 { 2459 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 2460 struct f2fs_inode_info *fi = F2FS_I(inode); 2461 u64 current_write; 2462 2463 fi->atomic_write_cnt++; 2464 atomic64_inc(&sbi->current_atomic_write); 2465 current_write = atomic64_read(&sbi->current_atomic_write); 2466 if (current_write > sbi->peak_atomic_write) 2467 sbi->peak_atomic_write = current_write; 2468 } 2469 2470 static inline void release_atomic_write_cnt(struct inode *inode) 2471 { 2472 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 2473 struct f2fs_inode_info *fi = F2FS_I(inode); 2474 2475 atomic64_sub(fi->atomic_write_cnt, &sbi->current_atomic_write); 2476 fi->atomic_write_cnt = 0; 2477 } 2478 2479 static inline s64 get_pages(struct f2fs_sb_info *sbi, int count_type) 2480 { 2481 return atomic_read(&sbi->nr_pages[count_type]); 2482 } 2483 2484 static inline int get_dirty_pages(struct inode *inode) 2485 { 2486 return atomic_read(&F2FS_I(inode)->dirty_pages); 2487 } 2488 2489 static inline int get_blocktype_secs(struct f2fs_sb_info *sbi, int block_type) 2490 { 2491 unsigned int pages_per_sec = sbi->segs_per_sec * sbi->blocks_per_seg; 2492 unsigned int segs = (get_pages(sbi, block_type) + pages_per_sec - 1) >> 2493 sbi->log_blocks_per_seg; 2494 2495 return segs / sbi->segs_per_sec; 2496 } 2497 2498 static inline block_t valid_user_blocks(struct f2fs_sb_info *sbi) 2499 { 2500 return sbi->total_valid_block_count; 2501 } 2502 2503 static inline block_t discard_blocks(struct f2fs_sb_info *sbi) 2504 { 2505 return sbi->discard_blks; 2506 } 2507 2508 static inline unsigned long __bitmap_size(struct f2fs_sb_info *sbi, int flag) 2509 { 2510 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi); 2511 2512 /* return NAT or SIT bitmap */ 2513 if (flag == NAT_BITMAP) 2514 return le32_to_cpu(ckpt->nat_ver_bitmap_bytesize); 2515 else if (flag == SIT_BITMAP) 2516 return le32_to_cpu(ckpt->sit_ver_bitmap_bytesize); 2517 2518 return 0; 2519 } 2520 2521 static inline block_t __cp_payload(struct f2fs_sb_info *sbi) 2522 { 2523 return le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_payload); 2524 } 2525 2526 static inline void *__bitmap_ptr(struct f2fs_sb_info *sbi, int flag) 2527 { 2528 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi); 2529 void *tmp_ptr = &ckpt->sit_nat_version_bitmap; 2530 int offset; 2531 2532 if (is_set_ckpt_flags(sbi, CP_LARGE_NAT_BITMAP_FLAG)) { 2533 offset = (flag == SIT_BITMAP) ? 2534 le32_to_cpu(ckpt->nat_ver_bitmap_bytesize) : 0; 2535 /* 2536 * if large_nat_bitmap feature is enabled, leave checksum 2537 * protection for all nat/sit bitmaps. 2538 */ 2539 return tmp_ptr + offset + sizeof(__le32); 2540 } 2541 2542 if (__cp_payload(sbi) > 0) { 2543 if (flag == NAT_BITMAP) 2544 return tmp_ptr; 2545 else 2546 return (unsigned char *)ckpt + F2FS_BLKSIZE; 2547 } else { 2548 offset = (flag == NAT_BITMAP) ? 2549 le32_to_cpu(ckpt->sit_ver_bitmap_bytesize) : 0; 2550 return tmp_ptr + offset; 2551 } 2552 } 2553 2554 static inline block_t __start_cp_addr(struct f2fs_sb_info *sbi) 2555 { 2556 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr); 2557 2558 if (sbi->cur_cp_pack == 2) 2559 start_addr += sbi->blocks_per_seg; 2560 return start_addr; 2561 } 2562 2563 static inline block_t __start_cp_next_addr(struct f2fs_sb_info *sbi) 2564 { 2565 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr); 2566 2567 if (sbi->cur_cp_pack == 1) 2568 start_addr += sbi->blocks_per_seg; 2569 return start_addr; 2570 } 2571 2572 static inline void __set_cp_next_pack(struct f2fs_sb_info *sbi) 2573 { 2574 sbi->cur_cp_pack = (sbi->cur_cp_pack == 1) ? 2 : 1; 2575 } 2576 2577 static inline block_t __start_sum_addr(struct f2fs_sb_info *sbi) 2578 { 2579 return le32_to_cpu(F2FS_CKPT(sbi)->cp_pack_start_sum); 2580 } 2581 2582 extern void f2fs_mark_inode_dirty_sync(struct inode *inode, bool sync); 2583 static inline int inc_valid_node_count(struct f2fs_sb_info *sbi, 2584 struct inode *inode, bool is_inode) 2585 { 2586 block_t valid_block_count; 2587 unsigned int valid_node_count, user_block_count; 2588 int err; 2589 2590 if (is_inode) { 2591 if (inode) { 2592 err = dquot_alloc_inode(inode); 2593 if (err) 2594 return err; 2595 } 2596 } else { 2597 err = dquot_reserve_block(inode, 1); 2598 if (err) 2599 return err; 2600 } 2601 2602 if (time_to_inject(sbi, FAULT_BLOCK)) { 2603 f2fs_show_injection_info(sbi, FAULT_BLOCK); 2604 goto enospc; 2605 } 2606 2607 spin_lock(&sbi->stat_lock); 2608 2609 valid_block_count = sbi->total_valid_block_count + 2610 sbi->current_reserved_blocks + 1; 2611 2612 if (!__allow_reserved_blocks(sbi, inode, false)) 2613 valid_block_count += F2FS_OPTION(sbi).root_reserved_blocks; 2614 2615 if (F2FS_IO_ALIGNED(sbi)) 2616 valid_block_count += sbi->blocks_per_seg * 2617 SM_I(sbi)->additional_reserved_segments; 2618 2619 user_block_count = sbi->user_block_count; 2620 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) 2621 user_block_count -= sbi->unusable_block_count; 2622 2623 if (unlikely(valid_block_count > user_block_count)) { 2624 spin_unlock(&sbi->stat_lock); 2625 goto enospc; 2626 } 2627 2628 valid_node_count = sbi->total_valid_node_count + 1; 2629 if (unlikely(valid_node_count > sbi->total_node_count)) { 2630 spin_unlock(&sbi->stat_lock); 2631 goto enospc; 2632 } 2633 2634 sbi->total_valid_node_count++; 2635 sbi->total_valid_block_count++; 2636 spin_unlock(&sbi->stat_lock); 2637 2638 if (inode) { 2639 if (is_inode) 2640 f2fs_mark_inode_dirty_sync(inode, true); 2641 else 2642 f2fs_i_blocks_write(inode, 1, true, true); 2643 } 2644 2645 percpu_counter_inc(&sbi->alloc_valid_block_count); 2646 return 0; 2647 2648 enospc: 2649 if (is_inode) { 2650 if (inode) 2651 dquot_free_inode(inode); 2652 } else { 2653 dquot_release_reservation_block(inode, 1); 2654 } 2655 return -ENOSPC; 2656 } 2657 2658 static inline void dec_valid_node_count(struct f2fs_sb_info *sbi, 2659 struct inode *inode, bool is_inode) 2660 { 2661 spin_lock(&sbi->stat_lock); 2662 2663 if (unlikely(!sbi->total_valid_block_count || 2664 !sbi->total_valid_node_count)) { 2665 f2fs_warn(sbi, "dec_valid_node_count: inconsistent block counts, total_valid_block:%u, total_valid_node:%u", 2666 sbi->total_valid_block_count, 2667 sbi->total_valid_node_count); 2668 set_sbi_flag(sbi, SBI_NEED_FSCK); 2669 } else { 2670 sbi->total_valid_block_count--; 2671 sbi->total_valid_node_count--; 2672 } 2673 2674 if (sbi->reserved_blocks && 2675 sbi->current_reserved_blocks < sbi->reserved_blocks) 2676 sbi->current_reserved_blocks++; 2677 2678 spin_unlock(&sbi->stat_lock); 2679 2680 if (is_inode) { 2681 dquot_free_inode(inode); 2682 } else { 2683 if (unlikely(inode->i_blocks == 0)) { 2684 f2fs_warn(sbi, "dec_valid_node_count: inconsistent i_blocks, ino:%lu, iblocks:%llu", 2685 inode->i_ino, 2686 (unsigned long long)inode->i_blocks); 2687 set_sbi_flag(sbi, SBI_NEED_FSCK); 2688 return; 2689 } 2690 f2fs_i_blocks_write(inode, 1, false, true); 2691 } 2692 } 2693 2694 static inline unsigned int valid_node_count(struct f2fs_sb_info *sbi) 2695 { 2696 return sbi->total_valid_node_count; 2697 } 2698 2699 static inline void inc_valid_inode_count(struct f2fs_sb_info *sbi) 2700 { 2701 percpu_counter_inc(&sbi->total_valid_inode_count); 2702 } 2703 2704 static inline void dec_valid_inode_count(struct f2fs_sb_info *sbi) 2705 { 2706 percpu_counter_dec(&sbi->total_valid_inode_count); 2707 } 2708 2709 static inline s64 valid_inode_count(struct f2fs_sb_info *sbi) 2710 { 2711 return percpu_counter_sum_positive(&sbi->total_valid_inode_count); 2712 } 2713 2714 static inline struct page *f2fs_grab_cache_page(struct address_space *mapping, 2715 pgoff_t index, bool for_write) 2716 { 2717 struct page *page; 2718 unsigned int flags; 2719 2720 if (IS_ENABLED(CONFIG_F2FS_FAULT_INJECTION)) { 2721 if (!for_write) 2722 page = find_get_page_flags(mapping, index, 2723 FGP_LOCK | FGP_ACCESSED); 2724 else 2725 page = find_lock_page(mapping, index); 2726 if (page) 2727 return page; 2728 2729 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_ALLOC)) { 2730 f2fs_show_injection_info(F2FS_M_SB(mapping), 2731 FAULT_PAGE_ALLOC); 2732 return NULL; 2733 } 2734 } 2735 2736 if (!for_write) 2737 return grab_cache_page(mapping, index); 2738 2739 flags = memalloc_nofs_save(); 2740 page = grab_cache_page_write_begin(mapping, index); 2741 memalloc_nofs_restore(flags); 2742 2743 return page; 2744 } 2745 2746 static inline struct page *f2fs_pagecache_get_page( 2747 struct address_space *mapping, pgoff_t index, 2748 int fgp_flags, gfp_t gfp_mask) 2749 { 2750 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_GET)) { 2751 f2fs_show_injection_info(F2FS_M_SB(mapping), FAULT_PAGE_GET); 2752 return NULL; 2753 } 2754 2755 return pagecache_get_page(mapping, index, fgp_flags, gfp_mask); 2756 } 2757 2758 static inline void f2fs_put_page(struct page *page, int unlock) 2759 { 2760 if (!page) 2761 return; 2762 2763 if (unlock) { 2764 f2fs_bug_on(F2FS_P_SB(page), !PageLocked(page)); 2765 unlock_page(page); 2766 } 2767 put_page(page); 2768 } 2769 2770 static inline void f2fs_put_dnode(struct dnode_of_data *dn) 2771 { 2772 if (dn->node_page) 2773 f2fs_put_page(dn->node_page, 1); 2774 if (dn->inode_page && dn->node_page != dn->inode_page) 2775 f2fs_put_page(dn->inode_page, 0); 2776 dn->node_page = NULL; 2777 dn->inode_page = NULL; 2778 } 2779 2780 static inline struct kmem_cache *f2fs_kmem_cache_create(const char *name, 2781 size_t size) 2782 { 2783 return kmem_cache_create(name, size, 0, SLAB_RECLAIM_ACCOUNT, NULL); 2784 } 2785 2786 static inline void *f2fs_kmem_cache_alloc_nofail(struct kmem_cache *cachep, 2787 gfp_t flags) 2788 { 2789 void *entry; 2790 2791 entry = kmem_cache_alloc(cachep, flags); 2792 if (!entry) 2793 entry = kmem_cache_alloc(cachep, flags | __GFP_NOFAIL); 2794 return entry; 2795 } 2796 2797 static inline void *f2fs_kmem_cache_alloc(struct kmem_cache *cachep, 2798 gfp_t flags, bool nofail, struct f2fs_sb_info *sbi) 2799 { 2800 if (nofail) 2801 return f2fs_kmem_cache_alloc_nofail(cachep, flags); 2802 2803 if (time_to_inject(sbi, FAULT_SLAB_ALLOC)) { 2804 f2fs_show_injection_info(sbi, FAULT_SLAB_ALLOC); 2805 return NULL; 2806 } 2807 2808 return kmem_cache_alloc(cachep, flags); 2809 } 2810 2811 static inline bool is_inflight_io(struct f2fs_sb_info *sbi, int type) 2812 { 2813 if (get_pages(sbi, F2FS_RD_DATA) || get_pages(sbi, F2FS_RD_NODE) || 2814 get_pages(sbi, F2FS_RD_META) || get_pages(sbi, F2FS_WB_DATA) || 2815 get_pages(sbi, F2FS_WB_CP_DATA) || 2816 get_pages(sbi, F2FS_DIO_READ) || 2817 get_pages(sbi, F2FS_DIO_WRITE)) 2818 return true; 2819 2820 if (type != DISCARD_TIME && SM_I(sbi) && SM_I(sbi)->dcc_info && 2821 atomic_read(&SM_I(sbi)->dcc_info->queued_discard)) 2822 return true; 2823 2824 if (SM_I(sbi) && SM_I(sbi)->fcc_info && 2825 atomic_read(&SM_I(sbi)->fcc_info->queued_flush)) 2826 return true; 2827 return false; 2828 } 2829 2830 static inline bool is_idle(struct f2fs_sb_info *sbi, int type) 2831 { 2832 if (sbi->gc_mode == GC_URGENT_HIGH) 2833 return true; 2834 2835 if (is_inflight_io(sbi, type)) 2836 return false; 2837 2838 if (sbi->gc_mode == GC_URGENT_MID) 2839 return true; 2840 2841 if (sbi->gc_mode == GC_URGENT_LOW && 2842 (type == DISCARD_TIME || type == GC_TIME)) 2843 return true; 2844 2845 return f2fs_time_over(sbi, type); 2846 } 2847 2848 static inline void f2fs_radix_tree_insert(struct radix_tree_root *root, 2849 unsigned long index, void *item) 2850 { 2851 while (radix_tree_insert(root, index, item)) 2852 cond_resched(); 2853 } 2854 2855 #define RAW_IS_INODE(p) ((p)->footer.nid == (p)->footer.ino) 2856 2857 static inline bool IS_INODE(struct page *page) 2858 { 2859 struct f2fs_node *p = F2FS_NODE(page); 2860 2861 return RAW_IS_INODE(p); 2862 } 2863 2864 static inline int offset_in_addr(struct f2fs_inode *i) 2865 { 2866 return (i->i_inline & F2FS_EXTRA_ATTR) ? 2867 (le16_to_cpu(i->i_extra_isize) / sizeof(__le32)) : 0; 2868 } 2869 2870 static inline __le32 *blkaddr_in_node(struct f2fs_node *node) 2871 { 2872 return RAW_IS_INODE(node) ? node->i.i_addr : node->dn.addr; 2873 } 2874 2875 static inline int f2fs_has_extra_attr(struct inode *inode); 2876 static inline block_t data_blkaddr(struct inode *inode, 2877 struct page *node_page, unsigned int offset) 2878 { 2879 struct f2fs_node *raw_node; 2880 __le32 *addr_array; 2881 int base = 0; 2882 bool is_inode = IS_INODE(node_page); 2883 2884 raw_node = F2FS_NODE(node_page); 2885 2886 if (is_inode) { 2887 if (!inode) 2888 /* from GC path only */ 2889 base = offset_in_addr(&raw_node->i); 2890 else if (f2fs_has_extra_attr(inode)) 2891 base = get_extra_isize(inode); 2892 } 2893 2894 addr_array = blkaddr_in_node(raw_node); 2895 return le32_to_cpu(addr_array[base + offset]); 2896 } 2897 2898 static inline block_t f2fs_data_blkaddr(struct dnode_of_data *dn) 2899 { 2900 return data_blkaddr(dn->inode, dn->node_page, dn->ofs_in_node); 2901 } 2902 2903 static inline int f2fs_test_bit(unsigned int nr, char *addr) 2904 { 2905 int mask; 2906 2907 addr += (nr >> 3); 2908 mask = 1 << (7 - (nr & 0x07)); 2909 return mask & *addr; 2910 } 2911 2912 static inline void f2fs_set_bit(unsigned int nr, char *addr) 2913 { 2914 int mask; 2915 2916 addr += (nr >> 3); 2917 mask = 1 << (7 - (nr & 0x07)); 2918 *addr |= mask; 2919 } 2920 2921 static inline void f2fs_clear_bit(unsigned int nr, char *addr) 2922 { 2923 int mask; 2924 2925 addr += (nr >> 3); 2926 mask = 1 << (7 - (nr & 0x07)); 2927 *addr &= ~mask; 2928 } 2929 2930 static inline int f2fs_test_and_set_bit(unsigned int nr, char *addr) 2931 { 2932 int mask; 2933 int ret; 2934 2935 addr += (nr >> 3); 2936 mask = 1 << (7 - (nr & 0x07)); 2937 ret = mask & *addr; 2938 *addr |= mask; 2939 return ret; 2940 } 2941 2942 static inline int f2fs_test_and_clear_bit(unsigned int nr, char *addr) 2943 { 2944 int mask; 2945 int ret; 2946 2947 addr += (nr >> 3); 2948 mask = 1 << (7 - (nr & 0x07)); 2949 ret = mask & *addr; 2950 *addr &= ~mask; 2951 return ret; 2952 } 2953 2954 static inline void f2fs_change_bit(unsigned int nr, char *addr) 2955 { 2956 int mask; 2957 2958 addr += (nr >> 3); 2959 mask = 1 << (7 - (nr & 0x07)); 2960 *addr ^= mask; 2961 } 2962 2963 /* 2964 * On-disk inode flags (f2fs_inode::i_flags) 2965 */ 2966 #define F2FS_COMPR_FL 0x00000004 /* Compress file */ 2967 #define F2FS_SYNC_FL 0x00000008 /* Synchronous updates */ 2968 #define F2FS_IMMUTABLE_FL 0x00000010 /* Immutable file */ 2969 #define F2FS_APPEND_FL 0x00000020 /* writes to file may only append */ 2970 #define F2FS_NODUMP_FL 0x00000040 /* do not dump file */ 2971 #define F2FS_NOATIME_FL 0x00000080 /* do not update atime */ 2972 #define F2FS_NOCOMP_FL 0x00000400 /* Don't compress */ 2973 #define F2FS_INDEX_FL 0x00001000 /* hash-indexed directory */ 2974 #define F2FS_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */ 2975 #define F2FS_PROJINHERIT_FL 0x20000000 /* Create with parents projid */ 2976 #define F2FS_CASEFOLD_FL 0x40000000 /* Casefolded file */ 2977 2978 /* Flags that should be inherited by new inodes from their parent. */ 2979 #define F2FS_FL_INHERITED (F2FS_SYNC_FL | F2FS_NODUMP_FL | F2FS_NOATIME_FL | \ 2980 F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \ 2981 F2FS_CASEFOLD_FL) 2982 2983 /* Flags that are appropriate for regular files (all but dir-specific ones). */ 2984 #define F2FS_REG_FLMASK (~(F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \ 2985 F2FS_CASEFOLD_FL)) 2986 2987 /* Flags that are appropriate for non-directories/regular files. */ 2988 #define F2FS_OTHER_FLMASK (F2FS_NODUMP_FL | F2FS_NOATIME_FL) 2989 2990 static inline __u32 f2fs_mask_flags(umode_t mode, __u32 flags) 2991 { 2992 if (S_ISDIR(mode)) 2993 return flags; 2994 else if (S_ISREG(mode)) 2995 return flags & F2FS_REG_FLMASK; 2996 else 2997 return flags & F2FS_OTHER_FLMASK; 2998 } 2999 3000 static inline void __mark_inode_dirty_flag(struct inode *inode, 3001 int flag, bool set) 3002 { 3003 switch (flag) { 3004 case FI_INLINE_XATTR: 3005 case FI_INLINE_DATA: 3006 case FI_INLINE_DENTRY: 3007 case FI_NEW_INODE: 3008 if (set) 3009 return; 3010 fallthrough; 3011 case FI_DATA_EXIST: 3012 case FI_INLINE_DOTS: 3013 case FI_PIN_FILE: 3014 case FI_COMPRESS_RELEASED: 3015 f2fs_mark_inode_dirty_sync(inode, true); 3016 } 3017 } 3018 3019 static inline void set_inode_flag(struct inode *inode, int flag) 3020 { 3021 set_bit(flag, F2FS_I(inode)->flags); 3022 __mark_inode_dirty_flag(inode, flag, true); 3023 } 3024 3025 static inline int is_inode_flag_set(struct inode *inode, int flag) 3026 { 3027 return test_bit(flag, F2FS_I(inode)->flags); 3028 } 3029 3030 static inline void clear_inode_flag(struct inode *inode, int flag) 3031 { 3032 clear_bit(flag, F2FS_I(inode)->flags); 3033 __mark_inode_dirty_flag(inode, flag, false); 3034 } 3035 3036 static inline bool f2fs_verity_in_progress(struct inode *inode) 3037 { 3038 return IS_ENABLED(CONFIG_FS_VERITY) && 3039 is_inode_flag_set(inode, FI_VERITY_IN_PROGRESS); 3040 } 3041 3042 static inline void set_acl_inode(struct inode *inode, umode_t mode) 3043 { 3044 F2FS_I(inode)->i_acl_mode = mode; 3045 set_inode_flag(inode, FI_ACL_MODE); 3046 f2fs_mark_inode_dirty_sync(inode, false); 3047 } 3048 3049 static inline void f2fs_i_links_write(struct inode *inode, bool inc) 3050 { 3051 if (inc) 3052 inc_nlink(inode); 3053 else 3054 drop_nlink(inode); 3055 f2fs_mark_inode_dirty_sync(inode, true); 3056 } 3057 3058 static inline void f2fs_i_blocks_write(struct inode *inode, 3059 block_t diff, bool add, bool claim) 3060 { 3061 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE); 3062 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER); 3063 3064 /* add = 1, claim = 1 should be dquot_reserve_block in pair */ 3065 if (add) { 3066 if (claim) 3067 dquot_claim_block(inode, diff); 3068 else 3069 dquot_alloc_block_nofail(inode, diff); 3070 } else { 3071 dquot_free_block(inode, diff); 3072 } 3073 3074 f2fs_mark_inode_dirty_sync(inode, true); 3075 if (clean || recover) 3076 set_inode_flag(inode, FI_AUTO_RECOVER); 3077 } 3078 3079 static inline bool f2fs_is_atomic_file(struct inode *inode); 3080 3081 static inline void f2fs_i_size_write(struct inode *inode, loff_t i_size) 3082 { 3083 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE); 3084 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER); 3085 3086 if (i_size_read(inode) == i_size) 3087 return; 3088 3089 i_size_write(inode, i_size); 3090 3091 if (f2fs_is_atomic_file(inode)) 3092 return; 3093 3094 f2fs_mark_inode_dirty_sync(inode, true); 3095 if (clean || recover) 3096 set_inode_flag(inode, FI_AUTO_RECOVER); 3097 } 3098 3099 static inline void f2fs_i_depth_write(struct inode *inode, unsigned int depth) 3100 { 3101 F2FS_I(inode)->i_current_depth = depth; 3102 f2fs_mark_inode_dirty_sync(inode, true); 3103 } 3104 3105 static inline void f2fs_i_gc_failures_write(struct inode *inode, 3106 unsigned int count) 3107 { 3108 F2FS_I(inode)->i_gc_failures[GC_FAILURE_PIN] = count; 3109 f2fs_mark_inode_dirty_sync(inode, true); 3110 } 3111 3112 static inline void f2fs_i_xnid_write(struct inode *inode, nid_t xnid) 3113 { 3114 F2FS_I(inode)->i_xattr_nid = xnid; 3115 f2fs_mark_inode_dirty_sync(inode, true); 3116 } 3117 3118 static inline void f2fs_i_pino_write(struct inode *inode, nid_t pino) 3119 { 3120 F2FS_I(inode)->i_pino = pino; 3121 f2fs_mark_inode_dirty_sync(inode, true); 3122 } 3123 3124 static inline void get_inline_info(struct inode *inode, struct f2fs_inode *ri) 3125 { 3126 struct f2fs_inode_info *fi = F2FS_I(inode); 3127 3128 if (ri->i_inline & F2FS_INLINE_XATTR) 3129 set_bit(FI_INLINE_XATTR, fi->flags); 3130 if (ri->i_inline & F2FS_INLINE_DATA) 3131 set_bit(FI_INLINE_DATA, fi->flags); 3132 if (ri->i_inline & F2FS_INLINE_DENTRY) 3133 set_bit(FI_INLINE_DENTRY, fi->flags); 3134 if (ri->i_inline & F2FS_DATA_EXIST) 3135 set_bit(FI_DATA_EXIST, fi->flags); 3136 if (ri->i_inline & F2FS_INLINE_DOTS) 3137 set_bit(FI_INLINE_DOTS, fi->flags); 3138 if (ri->i_inline & F2FS_EXTRA_ATTR) 3139 set_bit(FI_EXTRA_ATTR, fi->flags); 3140 if (ri->i_inline & F2FS_PIN_FILE) 3141 set_bit(FI_PIN_FILE, fi->flags); 3142 if (ri->i_inline & F2FS_COMPRESS_RELEASED) 3143 set_bit(FI_COMPRESS_RELEASED, fi->flags); 3144 } 3145 3146 static inline void set_raw_inline(struct inode *inode, struct f2fs_inode *ri) 3147 { 3148 ri->i_inline = 0; 3149 3150 if (is_inode_flag_set(inode, FI_INLINE_XATTR)) 3151 ri->i_inline |= F2FS_INLINE_XATTR; 3152 if (is_inode_flag_set(inode, FI_INLINE_DATA)) 3153 ri->i_inline |= F2FS_INLINE_DATA; 3154 if (is_inode_flag_set(inode, FI_INLINE_DENTRY)) 3155 ri->i_inline |= F2FS_INLINE_DENTRY; 3156 if (is_inode_flag_set(inode, FI_DATA_EXIST)) 3157 ri->i_inline |= F2FS_DATA_EXIST; 3158 if (is_inode_flag_set(inode, FI_INLINE_DOTS)) 3159 ri->i_inline |= F2FS_INLINE_DOTS; 3160 if (is_inode_flag_set(inode, FI_EXTRA_ATTR)) 3161 ri->i_inline |= F2FS_EXTRA_ATTR; 3162 if (is_inode_flag_set(inode, FI_PIN_FILE)) 3163 ri->i_inline |= F2FS_PIN_FILE; 3164 if (is_inode_flag_set(inode, FI_COMPRESS_RELEASED)) 3165 ri->i_inline |= F2FS_COMPRESS_RELEASED; 3166 } 3167 3168 static inline int f2fs_has_extra_attr(struct inode *inode) 3169 { 3170 return is_inode_flag_set(inode, FI_EXTRA_ATTR); 3171 } 3172 3173 static inline int f2fs_has_inline_xattr(struct inode *inode) 3174 { 3175 return is_inode_flag_set(inode, FI_INLINE_XATTR); 3176 } 3177 3178 static inline int f2fs_compressed_file(struct inode *inode) 3179 { 3180 return S_ISREG(inode->i_mode) && 3181 is_inode_flag_set(inode, FI_COMPRESSED_FILE); 3182 } 3183 3184 static inline bool f2fs_need_compress_data(struct inode *inode) 3185 { 3186 int compress_mode = F2FS_OPTION(F2FS_I_SB(inode)).compress_mode; 3187 3188 if (!f2fs_compressed_file(inode)) 3189 return false; 3190 3191 if (compress_mode == COMPR_MODE_FS) 3192 return true; 3193 else if (compress_mode == COMPR_MODE_USER && 3194 is_inode_flag_set(inode, FI_ENABLE_COMPRESS)) 3195 return true; 3196 3197 return false; 3198 } 3199 3200 static inline unsigned int addrs_per_inode(struct inode *inode) 3201 { 3202 unsigned int addrs = CUR_ADDRS_PER_INODE(inode) - 3203 get_inline_xattr_addrs(inode); 3204 3205 if (!f2fs_compressed_file(inode)) 3206 return addrs; 3207 return ALIGN_DOWN(addrs, F2FS_I(inode)->i_cluster_size); 3208 } 3209 3210 static inline unsigned int addrs_per_block(struct inode *inode) 3211 { 3212 if (!f2fs_compressed_file(inode)) 3213 return DEF_ADDRS_PER_BLOCK; 3214 return ALIGN_DOWN(DEF_ADDRS_PER_BLOCK, F2FS_I(inode)->i_cluster_size); 3215 } 3216 3217 static inline void *inline_xattr_addr(struct inode *inode, struct page *page) 3218 { 3219 struct f2fs_inode *ri = F2FS_INODE(page); 3220 3221 return (void *)&(ri->i_addr[DEF_ADDRS_PER_INODE - 3222 get_inline_xattr_addrs(inode)]); 3223 } 3224 3225 static inline int inline_xattr_size(struct inode *inode) 3226 { 3227 if (f2fs_has_inline_xattr(inode)) 3228 return get_inline_xattr_addrs(inode) * sizeof(__le32); 3229 return 0; 3230 } 3231 3232 /* 3233 * Notice: check inline_data flag without inode page lock is unsafe. 3234 * It could change at any time by f2fs_convert_inline_page(). 3235 */ 3236 static inline int f2fs_has_inline_data(struct inode *inode) 3237 { 3238 return is_inode_flag_set(inode, FI_INLINE_DATA); 3239 } 3240 3241 static inline int f2fs_exist_data(struct inode *inode) 3242 { 3243 return is_inode_flag_set(inode, FI_DATA_EXIST); 3244 } 3245 3246 static inline int f2fs_has_inline_dots(struct inode *inode) 3247 { 3248 return is_inode_flag_set(inode, FI_INLINE_DOTS); 3249 } 3250 3251 static inline int f2fs_is_mmap_file(struct inode *inode) 3252 { 3253 return is_inode_flag_set(inode, FI_MMAP_FILE); 3254 } 3255 3256 static inline bool f2fs_is_pinned_file(struct inode *inode) 3257 { 3258 return is_inode_flag_set(inode, FI_PIN_FILE); 3259 } 3260 3261 static inline bool f2fs_is_atomic_file(struct inode *inode) 3262 { 3263 return is_inode_flag_set(inode, FI_ATOMIC_FILE); 3264 } 3265 3266 static inline bool f2fs_is_cow_file(struct inode *inode) 3267 { 3268 return is_inode_flag_set(inode, FI_COW_FILE); 3269 } 3270 3271 static inline bool f2fs_is_first_block_written(struct inode *inode) 3272 { 3273 return is_inode_flag_set(inode, FI_FIRST_BLOCK_WRITTEN); 3274 } 3275 3276 static inline bool f2fs_is_drop_cache(struct inode *inode) 3277 { 3278 return is_inode_flag_set(inode, FI_DROP_CACHE); 3279 } 3280 3281 static inline void *inline_data_addr(struct inode *inode, struct page *page) 3282 { 3283 struct f2fs_inode *ri = F2FS_INODE(page); 3284 int extra_size = get_extra_isize(inode); 3285 3286 return (void *)&(ri->i_addr[extra_size + DEF_INLINE_RESERVED_SIZE]); 3287 } 3288 3289 static inline int f2fs_has_inline_dentry(struct inode *inode) 3290 { 3291 return is_inode_flag_set(inode, FI_INLINE_DENTRY); 3292 } 3293 3294 static inline int is_file(struct inode *inode, int type) 3295 { 3296 return F2FS_I(inode)->i_advise & type; 3297 } 3298 3299 static inline void set_file(struct inode *inode, int type) 3300 { 3301 if (is_file(inode, type)) 3302 return; 3303 F2FS_I(inode)->i_advise |= type; 3304 f2fs_mark_inode_dirty_sync(inode, true); 3305 } 3306 3307 static inline void clear_file(struct inode *inode, int type) 3308 { 3309 if (!is_file(inode, type)) 3310 return; 3311 F2FS_I(inode)->i_advise &= ~type; 3312 f2fs_mark_inode_dirty_sync(inode, true); 3313 } 3314 3315 static inline bool f2fs_is_time_consistent(struct inode *inode) 3316 { 3317 if (!timespec64_equal(F2FS_I(inode)->i_disk_time, &inode->i_atime)) 3318 return false; 3319 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 1, &inode->i_ctime)) 3320 return false; 3321 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 2, &inode->i_mtime)) 3322 return false; 3323 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 3, 3324 &F2FS_I(inode)->i_crtime)) 3325 return false; 3326 return true; 3327 } 3328 3329 static inline bool f2fs_skip_inode_update(struct inode *inode, int dsync) 3330 { 3331 bool ret; 3332 3333 if (dsync) { 3334 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 3335 3336 spin_lock(&sbi->inode_lock[DIRTY_META]); 3337 ret = list_empty(&F2FS_I(inode)->gdirty_list); 3338 spin_unlock(&sbi->inode_lock[DIRTY_META]); 3339 return ret; 3340 } 3341 if (!is_inode_flag_set(inode, FI_AUTO_RECOVER) || 3342 file_keep_isize(inode) || 3343 i_size_read(inode) & ~PAGE_MASK) 3344 return false; 3345 3346 if (!f2fs_is_time_consistent(inode)) 3347 return false; 3348 3349 spin_lock(&F2FS_I(inode)->i_size_lock); 3350 ret = F2FS_I(inode)->last_disk_size == i_size_read(inode); 3351 spin_unlock(&F2FS_I(inode)->i_size_lock); 3352 3353 return ret; 3354 } 3355 3356 static inline bool f2fs_readonly(struct super_block *sb) 3357 { 3358 return sb_rdonly(sb); 3359 } 3360 3361 static inline bool f2fs_cp_error(struct f2fs_sb_info *sbi) 3362 { 3363 return is_set_ckpt_flags(sbi, CP_ERROR_FLAG); 3364 } 3365 3366 static inline bool is_dot_dotdot(const u8 *name, size_t len) 3367 { 3368 if (len == 1 && name[0] == '.') 3369 return true; 3370 3371 if (len == 2 && name[0] == '.' && name[1] == '.') 3372 return true; 3373 3374 return false; 3375 } 3376 3377 static inline void *f2fs_kmalloc(struct f2fs_sb_info *sbi, 3378 size_t size, gfp_t flags) 3379 { 3380 if (time_to_inject(sbi, FAULT_KMALLOC)) { 3381 f2fs_show_injection_info(sbi, FAULT_KMALLOC); 3382 return NULL; 3383 } 3384 3385 return kmalloc(size, flags); 3386 } 3387 3388 static inline void *f2fs_kzalloc(struct f2fs_sb_info *sbi, 3389 size_t size, gfp_t flags) 3390 { 3391 return f2fs_kmalloc(sbi, size, flags | __GFP_ZERO); 3392 } 3393 3394 static inline void *f2fs_kvmalloc(struct f2fs_sb_info *sbi, 3395 size_t size, gfp_t flags) 3396 { 3397 if (time_to_inject(sbi, FAULT_KVMALLOC)) { 3398 f2fs_show_injection_info(sbi, FAULT_KVMALLOC); 3399 return NULL; 3400 } 3401 3402 return kvmalloc(size, flags); 3403 } 3404 3405 static inline void *f2fs_kvzalloc(struct f2fs_sb_info *sbi, 3406 size_t size, gfp_t flags) 3407 { 3408 return f2fs_kvmalloc(sbi, size, flags | __GFP_ZERO); 3409 } 3410 3411 static inline int get_extra_isize(struct inode *inode) 3412 { 3413 return F2FS_I(inode)->i_extra_isize / sizeof(__le32); 3414 } 3415 3416 static inline int get_inline_xattr_addrs(struct inode *inode) 3417 { 3418 return F2FS_I(inode)->i_inline_xattr_size; 3419 } 3420 3421 #define f2fs_get_inode_mode(i) \ 3422 ((is_inode_flag_set(i, FI_ACL_MODE)) ? \ 3423 (F2FS_I(i)->i_acl_mode) : ((i)->i_mode)) 3424 3425 #define F2FS_TOTAL_EXTRA_ATTR_SIZE \ 3426 (offsetof(struct f2fs_inode, i_extra_end) - \ 3427 offsetof(struct f2fs_inode, i_extra_isize)) \ 3428 3429 #define F2FS_OLD_ATTRIBUTE_SIZE (offsetof(struct f2fs_inode, i_addr)) 3430 #define F2FS_FITS_IN_INODE(f2fs_inode, extra_isize, field) \ 3431 ((offsetof(typeof(*(f2fs_inode)), field) + \ 3432 sizeof((f2fs_inode)->field)) \ 3433 <= (F2FS_OLD_ATTRIBUTE_SIZE + (extra_isize))) \ 3434 3435 #define __is_large_section(sbi) ((sbi)->segs_per_sec > 1) 3436 3437 #define __is_meta_io(fio) (PAGE_TYPE_OF_BIO((fio)->type) == META) 3438 3439 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi, 3440 block_t blkaddr, int type); 3441 static inline void verify_blkaddr(struct f2fs_sb_info *sbi, 3442 block_t blkaddr, int type) 3443 { 3444 if (!f2fs_is_valid_blkaddr(sbi, blkaddr, type)) { 3445 f2fs_err(sbi, "invalid blkaddr: %u, type: %d, run fsck to fix.", 3446 blkaddr, type); 3447 f2fs_bug_on(sbi, 1); 3448 } 3449 } 3450 3451 static inline bool __is_valid_data_blkaddr(block_t blkaddr) 3452 { 3453 if (blkaddr == NEW_ADDR || blkaddr == NULL_ADDR || 3454 blkaddr == COMPRESS_ADDR) 3455 return false; 3456 return true; 3457 } 3458 3459 /* 3460 * file.c 3461 */ 3462 int f2fs_sync_file(struct file *file, loff_t start, loff_t end, int datasync); 3463 void f2fs_truncate_data_blocks(struct dnode_of_data *dn); 3464 int f2fs_do_truncate_blocks(struct inode *inode, u64 from, bool lock); 3465 int f2fs_truncate_blocks(struct inode *inode, u64 from, bool lock); 3466 int f2fs_truncate(struct inode *inode); 3467 int f2fs_getattr(struct user_namespace *mnt_userns, const struct path *path, 3468 struct kstat *stat, u32 request_mask, unsigned int flags); 3469 int f2fs_setattr(struct user_namespace *mnt_userns, struct dentry *dentry, 3470 struct iattr *attr); 3471 int f2fs_truncate_hole(struct inode *inode, pgoff_t pg_start, pgoff_t pg_end); 3472 void f2fs_truncate_data_blocks_range(struct dnode_of_data *dn, int count); 3473 int f2fs_precache_extents(struct inode *inode); 3474 int f2fs_fileattr_get(struct dentry *dentry, struct fileattr *fa); 3475 int f2fs_fileattr_set(struct user_namespace *mnt_userns, 3476 struct dentry *dentry, struct fileattr *fa); 3477 long f2fs_ioctl(struct file *filp, unsigned int cmd, unsigned long arg); 3478 long f2fs_compat_ioctl(struct file *file, unsigned int cmd, unsigned long arg); 3479 int f2fs_transfer_project_quota(struct inode *inode, kprojid_t kprojid); 3480 int f2fs_pin_file_control(struct inode *inode, bool inc); 3481 3482 /* 3483 * inode.c 3484 */ 3485 void f2fs_set_inode_flags(struct inode *inode); 3486 bool f2fs_inode_chksum_verify(struct f2fs_sb_info *sbi, struct page *page); 3487 void f2fs_inode_chksum_set(struct f2fs_sb_info *sbi, struct page *page); 3488 struct inode *f2fs_iget(struct super_block *sb, unsigned long ino); 3489 struct inode *f2fs_iget_retry(struct super_block *sb, unsigned long ino); 3490 int f2fs_try_to_free_nats(struct f2fs_sb_info *sbi, int nr_shrink); 3491 void f2fs_update_inode(struct inode *inode, struct page *node_page); 3492 void f2fs_update_inode_page(struct inode *inode); 3493 int f2fs_write_inode(struct inode *inode, struct writeback_control *wbc); 3494 void f2fs_evict_inode(struct inode *inode); 3495 void f2fs_handle_failed_inode(struct inode *inode); 3496 3497 /* 3498 * namei.c 3499 */ 3500 int f2fs_update_extension_list(struct f2fs_sb_info *sbi, const char *name, 3501 bool hot, bool set); 3502 struct dentry *f2fs_get_parent(struct dentry *child); 3503 int f2fs_get_tmpfile(struct user_namespace *mnt_userns, struct inode *dir, 3504 struct inode **new_inode); 3505 3506 /* 3507 * dir.c 3508 */ 3509 unsigned char f2fs_get_de_type(struct f2fs_dir_entry *de); 3510 int f2fs_init_casefolded_name(const struct inode *dir, 3511 struct f2fs_filename *fname); 3512 int f2fs_setup_filename(struct inode *dir, const struct qstr *iname, 3513 int lookup, struct f2fs_filename *fname); 3514 int f2fs_prepare_lookup(struct inode *dir, struct dentry *dentry, 3515 struct f2fs_filename *fname); 3516 void f2fs_free_filename(struct f2fs_filename *fname); 3517 struct f2fs_dir_entry *f2fs_find_target_dentry(const struct f2fs_dentry_ptr *d, 3518 const struct f2fs_filename *fname, int *max_slots); 3519 int f2fs_fill_dentries(struct dir_context *ctx, struct f2fs_dentry_ptr *d, 3520 unsigned int start_pos, struct fscrypt_str *fstr); 3521 void f2fs_do_make_empty_dir(struct inode *inode, struct inode *parent, 3522 struct f2fs_dentry_ptr *d); 3523 struct page *f2fs_init_inode_metadata(struct inode *inode, struct inode *dir, 3524 const struct f2fs_filename *fname, struct page *dpage); 3525 void f2fs_update_parent_metadata(struct inode *dir, struct inode *inode, 3526 unsigned int current_depth); 3527 int f2fs_room_for_filename(const void *bitmap, int slots, int max_slots); 3528 void f2fs_drop_nlink(struct inode *dir, struct inode *inode); 3529 struct f2fs_dir_entry *__f2fs_find_entry(struct inode *dir, 3530 const struct f2fs_filename *fname, 3531 struct page **res_page); 3532 struct f2fs_dir_entry *f2fs_find_entry(struct inode *dir, 3533 const struct qstr *child, struct page **res_page); 3534 struct f2fs_dir_entry *f2fs_parent_dir(struct inode *dir, struct page **p); 3535 ino_t f2fs_inode_by_name(struct inode *dir, const struct qstr *qstr, 3536 struct page **page); 3537 void f2fs_set_link(struct inode *dir, struct f2fs_dir_entry *de, 3538 struct page *page, struct inode *inode); 3539 bool f2fs_has_enough_room(struct inode *dir, struct page *ipage, 3540 const struct f2fs_filename *fname); 3541 void f2fs_update_dentry(nid_t ino, umode_t mode, struct f2fs_dentry_ptr *d, 3542 const struct fscrypt_str *name, f2fs_hash_t name_hash, 3543 unsigned int bit_pos); 3544 int f2fs_add_regular_entry(struct inode *dir, const struct f2fs_filename *fname, 3545 struct inode *inode, nid_t ino, umode_t mode); 3546 int f2fs_add_dentry(struct inode *dir, const struct f2fs_filename *fname, 3547 struct inode *inode, nid_t ino, umode_t mode); 3548 int f2fs_do_add_link(struct inode *dir, const struct qstr *name, 3549 struct inode *inode, nid_t ino, umode_t mode); 3550 void f2fs_delete_entry(struct f2fs_dir_entry *dentry, struct page *page, 3551 struct inode *dir, struct inode *inode); 3552 int f2fs_do_tmpfile(struct inode *inode, struct inode *dir); 3553 bool f2fs_empty_dir(struct inode *dir); 3554 3555 static inline int f2fs_add_link(struct dentry *dentry, struct inode *inode) 3556 { 3557 if (fscrypt_is_nokey_name(dentry)) 3558 return -ENOKEY; 3559 return f2fs_do_add_link(d_inode(dentry->d_parent), &dentry->d_name, 3560 inode, inode->i_ino, inode->i_mode); 3561 } 3562 3563 /* 3564 * super.c 3565 */ 3566 int f2fs_inode_dirtied(struct inode *inode, bool sync); 3567 void f2fs_inode_synced(struct inode *inode); 3568 int f2fs_dquot_initialize(struct inode *inode); 3569 int f2fs_enable_quota_files(struct f2fs_sb_info *sbi, bool rdonly); 3570 int f2fs_quota_sync(struct super_block *sb, int type); 3571 loff_t max_file_blocks(struct inode *inode); 3572 void f2fs_quota_off_umount(struct super_block *sb); 3573 void f2fs_handle_stop(struct f2fs_sb_info *sbi, unsigned char reason); 3574 void f2fs_handle_error(struct f2fs_sb_info *sbi, unsigned char error); 3575 int f2fs_commit_super(struct f2fs_sb_info *sbi, bool recover); 3576 int f2fs_sync_fs(struct super_block *sb, int sync); 3577 int f2fs_sanity_check_ckpt(struct f2fs_sb_info *sbi); 3578 3579 /* 3580 * hash.c 3581 */ 3582 void f2fs_hash_filename(const struct inode *dir, struct f2fs_filename *fname); 3583 3584 /* 3585 * node.c 3586 */ 3587 struct node_info; 3588 3589 int f2fs_check_nid_range(struct f2fs_sb_info *sbi, nid_t nid); 3590 bool f2fs_available_free_memory(struct f2fs_sb_info *sbi, int type); 3591 bool f2fs_in_warm_node_list(struct f2fs_sb_info *sbi, struct page *page); 3592 void f2fs_init_fsync_node_info(struct f2fs_sb_info *sbi); 3593 void f2fs_del_fsync_node_entry(struct f2fs_sb_info *sbi, struct page *page); 3594 void f2fs_reset_fsync_node_info(struct f2fs_sb_info *sbi); 3595 int f2fs_need_dentry_mark(struct f2fs_sb_info *sbi, nid_t nid); 3596 bool f2fs_is_checkpointed_node(struct f2fs_sb_info *sbi, nid_t nid); 3597 bool f2fs_need_inode_block_update(struct f2fs_sb_info *sbi, nid_t ino); 3598 int f2fs_get_node_info(struct f2fs_sb_info *sbi, nid_t nid, 3599 struct node_info *ni, bool checkpoint_context); 3600 pgoff_t f2fs_get_next_page_offset(struct dnode_of_data *dn, pgoff_t pgofs); 3601 int f2fs_get_dnode_of_data(struct dnode_of_data *dn, pgoff_t index, int mode); 3602 int f2fs_truncate_inode_blocks(struct inode *inode, pgoff_t from); 3603 int f2fs_truncate_xattr_node(struct inode *inode); 3604 int f2fs_wait_on_node_pages_writeback(struct f2fs_sb_info *sbi, 3605 unsigned int seq_id); 3606 bool f2fs_nat_bitmap_enabled(struct f2fs_sb_info *sbi); 3607 int f2fs_remove_inode_page(struct inode *inode); 3608 struct page *f2fs_new_inode_page(struct inode *inode); 3609 struct page *f2fs_new_node_page(struct dnode_of_data *dn, unsigned int ofs); 3610 void f2fs_ra_node_page(struct f2fs_sb_info *sbi, nid_t nid); 3611 struct page *f2fs_get_node_page(struct f2fs_sb_info *sbi, pgoff_t nid); 3612 struct page *f2fs_get_node_page_ra(struct page *parent, int start); 3613 int f2fs_move_node_page(struct page *node_page, int gc_type); 3614 void f2fs_flush_inline_data(struct f2fs_sb_info *sbi); 3615 int f2fs_fsync_node_pages(struct f2fs_sb_info *sbi, struct inode *inode, 3616 struct writeback_control *wbc, bool atomic, 3617 unsigned int *seq_id); 3618 int f2fs_sync_node_pages(struct f2fs_sb_info *sbi, 3619 struct writeback_control *wbc, 3620 bool do_balance, enum iostat_type io_type); 3621 int f2fs_build_free_nids(struct f2fs_sb_info *sbi, bool sync, bool mount); 3622 bool f2fs_alloc_nid(struct f2fs_sb_info *sbi, nid_t *nid); 3623 void f2fs_alloc_nid_done(struct f2fs_sb_info *sbi, nid_t nid); 3624 void f2fs_alloc_nid_failed(struct f2fs_sb_info *sbi, nid_t nid); 3625 int f2fs_try_to_free_nids(struct f2fs_sb_info *sbi, int nr_shrink); 3626 int f2fs_recover_inline_xattr(struct inode *inode, struct page *page); 3627 int f2fs_recover_xattr_data(struct inode *inode, struct page *page); 3628 int f2fs_recover_inode_page(struct f2fs_sb_info *sbi, struct page *page); 3629 int f2fs_restore_node_summary(struct f2fs_sb_info *sbi, 3630 unsigned int segno, struct f2fs_summary_block *sum); 3631 void f2fs_enable_nat_bits(struct f2fs_sb_info *sbi); 3632 int f2fs_flush_nat_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3633 int f2fs_build_node_manager(struct f2fs_sb_info *sbi); 3634 void f2fs_destroy_node_manager(struct f2fs_sb_info *sbi); 3635 int __init f2fs_create_node_manager_caches(void); 3636 void f2fs_destroy_node_manager_caches(void); 3637 3638 /* 3639 * segment.c 3640 */ 3641 bool f2fs_need_SSR(struct f2fs_sb_info *sbi); 3642 int f2fs_commit_atomic_write(struct inode *inode); 3643 void f2fs_abort_atomic_write(struct inode *inode, bool clean); 3644 void f2fs_balance_fs(struct f2fs_sb_info *sbi, bool need); 3645 void f2fs_balance_fs_bg(struct f2fs_sb_info *sbi, bool from_bg); 3646 int f2fs_issue_flush(struct f2fs_sb_info *sbi, nid_t ino); 3647 int f2fs_create_flush_cmd_control(struct f2fs_sb_info *sbi); 3648 int f2fs_flush_device_cache(struct f2fs_sb_info *sbi); 3649 void f2fs_destroy_flush_cmd_control(struct f2fs_sb_info *sbi, bool free); 3650 void f2fs_invalidate_blocks(struct f2fs_sb_info *sbi, block_t addr); 3651 bool f2fs_is_checkpointed_data(struct f2fs_sb_info *sbi, block_t blkaddr); 3652 int f2fs_start_discard_thread(struct f2fs_sb_info *sbi); 3653 void f2fs_drop_discard_cmd(struct f2fs_sb_info *sbi); 3654 void f2fs_stop_discard_thread(struct f2fs_sb_info *sbi); 3655 bool f2fs_issue_discard_timeout(struct f2fs_sb_info *sbi); 3656 void f2fs_clear_prefree_segments(struct f2fs_sb_info *sbi, 3657 struct cp_control *cpc); 3658 void f2fs_dirty_to_prefree(struct f2fs_sb_info *sbi); 3659 block_t f2fs_get_unusable_blocks(struct f2fs_sb_info *sbi); 3660 int f2fs_disable_cp_again(struct f2fs_sb_info *sbi, block_t unusable); 3661 void f2fs_release_discard_addrs(struct f2fs_sb_info *sbi); 3662 int f2fs_npages_for_summary_flush(struct f2fs_sb_info *sbi, bool for_ra); 3663 bool f2fs_segment_has_free_slot(struct f2fs_sb_info *sbi, int segno); 3664 void f2fs_init_inmem_curseg(struct f2fs_sb_info *sbi); 3665 void f2fs_save_inmem_curseg(struct f2fs_sb_info *sbi); 3666 void f2fs_restore_inmem_curseg(struct f2fs_sb_info *sbi); 3667 void f2fs_get_new_segment(struct f2fs_sb_info *sbi, 3668 unsigned int *newseg, bool new_sec, int dir); 3669 void f2fs_allocate_segment_for_resize(struct f2fs_sb_info *sbi, int type, 3670 unsigned int start, unsigned int end); 3671 void f2fs_allocate_new_section(struct f2fs_sb_info *sbi, int type, bool force); 3672 void f2fs_allocate_new_segments(struct f2fs_sb_info *sbi); 3673 int f2fs_trim_fs(struct f2fs_sb_info *sbi, struct fstrim_range *range); 3674 bool f2fs_exist_trim_candidates(struct f2fs_sb_info *sbi, 3675 struct cp_control *cpc); 3676 struct page *f2fs_get_sum_page(struct f2fs_sb_info *sbi, unsigned int segno); 3677 void f2fs_update_meta_page(struct f2fs_sb_info *sbi, void *src, 3678 block_t blk_addr); 3679 void f2fs_do_write_meta_page(struct f2fs_sb_info *sbi, struct page *page, 3680 enum iostat_type io_type); 3681 void f2fs_do_write_node_page(unsigned int nid, struct f2fs_io_info *fio); 3682 void f2fs_outplace_write_data(struct dnode_of_data *dn, 3683 struct f2fs_io_info *fio); 3684 int f2fs_inplace_write_data(struct f2fs_io_info *fio); 3685 void f2fs_do_replace_block(struct f2fs_sb_info *sbi, struct f2fs_summary *sum, 3686 block_t old_blkaddr, block_t new_blkaddr, 3687 bool recover_curseg, bool recover_newaddr, 3688 bool from_gc); 3689 void f2fs_replace_block(struct f2fs_sb_info *sbi, struct dnode_of_data *dn, 3690 block_t old_addr, block_t new_addr, 3691 unsigned char version, bool recover_curseg, 3692 bool recover_newaddr); 3693 void f2fs_allocate_data_block(struct f2fs_sb_info *sbi, struct page *page, 3694 block_t old_blkaddr, block_t *new_blkaddr, 3695 struct f2fs_summary *sum, int type, 3696 struct f2fs_io_info *fio); 3697 void f2fs_update_device_state(struct f2fs_sb_info *sbi, nid_t ino, 3698 block_t blkaddr, unsigned int blkcnt); 3699 void f2fs_wait_on_page_writeback(struct page *page, 3700 enum page_type type, bool ordered, bool locked); 3701 void f2fs_wait_on_block_writeback(struct inode *inode, block_t blkaddr); 3702 void f2fs_wait_on_block_writeback_range(struct inode *inode, block_t blkaddr, 3703 block_t len); 3704 void f2fs_write_data_summaries(struct f2fs_sb_info *sbi, block_t start_blk); 3705 void f2fs_write_node_summaries(struct f2fs_sb_info *sbi, block_t start_blk); 3706 int f2fs_lookup_journal_in_cursum(struct f2fs_journal *journal, int type, 3707 unsigned int val, int alloc); 3708 void f2fs_flush_sit_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3709 int f2fs_fix_curseg_write_pointer(struct f2fs_sb_info *sbi); 3710 int f2fs_check_write_pointer(struct f2fs_sb_info *sbi); 3711 int f2fs_build_segment_manager(struct f2fs_sb_info *sbi); 3712 void f2fs_destroy_segment_manager(struct f2fs_sb_info *sbi); 3713 int __init f2fs_create_segment_manager_caches(void); 3714 void f2fs_destroy_segment_manager_caches(void); 3715 int f2fs_rw_hint_to_seg_type(enum rw_hint hint); 3716 unsigned int f2fs_usable_segs_in_sec(struct f2fs_sb_info *sbi, 3717 unsigned int segno); 3718 unsigned int f2fs_usable_blks_in_seg(struct f2fs_sb_info *sbi, 3719 unsigned int segno); 3720 3721 #define DEF_FRAGMENT_SIZE 4 3722 #define MIN_FRAGMENT_SIZE 1 3723 #define MAX_FRAGMENT_SIZE 512 3724 3725 static inline bool f2fs_need_rand_seg(struct f2fs_sb_info *sbi) 3726 { 3727 return F2FS_OPTION(sbi).fs_mode == FS_MODE_FRAGMENT_SEG || 3728 F2FS_OPTION(sbi).fs_mode == FS_MODE_FRAGMENT_BLK; 3729 } 3730 3731 /* 3732 * checkpoint.c 3733 */ 3734 void f2fs_stop_checkpoint(struct f2fs_sb_info *sbi, bool end_io, 3735 unsigned char reason); 3736 void f2fs_flush_ckpt_thread(struct f2fs_sb_info *sbi); 3737 struct page *f2fs_grab_meta_page(struct f2fs_sb_info *sbi, pgoff_t index); 3738 struct page *f2fs_get_meta_page(struct f2fs_sb_info *sbi, pgoff_t index); 3739 struct page *f2fs_get_meta_page_retry(struct f2fs_sb_info *sbi, pgoff_t index); 3740 struct page *f2fs_get_tmp_page(struct f2fs_sb_info *sbi, pgoff_t index); 3741 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi, 3742 block_t blkaddr, int type); 3743 int f2fs_ra_meta_pages(struct f2fs_sb_info *sbi, block_t start, int nrpages, 3744 int type, bool sync); 3745 void f2fs_ra_meta_pages_cond(struct f2fs_sb_info *sbi, pgoff_t index, 3746 unsigned int ra_blocks); 3747 long f2fs_sync_meta_pages(struct f2fs_sb_info *sbi, enum page_type type, 3748 long nr_to_write, enum iostat_type io_type); 3749 void f2fs_add_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type); 3750 void f2fs_remove_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type); 3751 void f2fs_release_ino_entry(struct f2fs_sb_info *sbi, bool all); 3752 bool f2fs_exist_written_data(struct f2fs_sb_info *sbi, nid_t ino, int mode); 3753 void f2fs_set_dirty_device(struct f2fs_sb_info *sbi, nid_t ino, 3754 unsigned int devidx, int type); 3755 bool f2fs_is_dirty_device(struct f2fs_sb_info *sbi, nid_t ino, 3756 unsigned int devidx, int type); 3757 int f2fs_sync_inode_meta(struct f2fs_sb_info *sbi); 3758 int f2fs_acquire_orphan_inode(struct f2fs_sb_info *sbi); 3759 void f2fs_release_orphan_inode(struct f2fs_sb_info *sbi); 3760 void f2fs_add_orphan_inode(struct inode *inode); 3761 void f2fs_remove_orphan_inode(struct f2fs_sb_info *sbi, nid_t ino); 3762 int f2fs_recover_orphan_inodes(struct f2fs_sb_info *sbi); 3763 int f2fs_get_valid_checkpoint(struct f2fs_sb_info *sbi); 3764 void f2fs_update_dirty_folio(struct inode *inode, struct folio *folio); 3765 void f2fs_remove_dirty_inode(struct inode *inode); 3766 int f2fs_sync_dirty_inodes(struct f2fs_sb_info *sbi, enum inode_type type, 3767 bool from_cp); 3768 void f2fs_wait_on_all_pages(struct f2fs_sb_info *sbi, int type); 3769 u64 f2fs_get_sectors_written(struct f2fs_sb_info *sbi); 3770 int f2fs_write_checkpoint(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3771 void f2fs_init_ino_entry_info(struct f2fs_sb_info *sbi); 3772 int __init f2fs_create_checkpoint_caches(void); 3773 void f2fs_destroy_checkpoint_caches(void); 3774 int f2fs_issue_checkpoint(struct f2fs_sb_info *sbi); 3775 int f2fs_start_ckpt_thread(struct f2fs_sb_info *sbi); 3776 void f2fs_stop_ckpt_thread(struct f2fs_sb_info *sbi); 3777 void f2fs_init_ckpt_req_control(struct f2fs_sb_info *sbi); 3778 3779 /* 3780 * data.c 3781 */ 3782 int __init f2fs_init_bioset(void); 3783 void f2fs_destroy_bioset(void); 3784 int f2fs_init_bio_entry_cache(void); 3785 void f2fs_destroy_bio_entry_cache(void); 3786 void f2fs_submit_bio(struct f2fs_sb_info *sbi, 3787 struct bio *bio, enum page_type type); 3788 int f2fs_init_write_merge_io(struct f2fs_sb_info *sbi); 3789 void f2fs_submit_merged_write(struct f2fs_sb_info *sbi, enum page_type type); 3790 void f2fs_submit_merged_write_cond(struct f2fs_sb_info *sbi, 3791 struct inode *inode, struct page *page, 3792 nid_t ino, enum page_type type); 3793 void f2fs_submit_merged_ipu_write(struct f2fs_sb_info *sbi, 3794 struct bio **bio, struct page *page); 3795 void f2fs_flush_merged_writes(struct f2fs_sb_info *sbi); 3796 int f2fs_submit_page_bio(struct f2fs_io_info *fio); 3797 int f2fs_merge_page_bio(struct f2fs_io_info *fio); 3798 void f2fs_submit_page_write(struct f2fs_io_info *fio); 3799 struct block_device *f2fs_target_device(struct f2fs_sb_info *sbi, 3800 block_t blk_addr, sector_t *sector); 3801 int f2fs_target_device_index(struct f2fs_sb_info *sbi, block_t blkaddr); 3802 void f2fs_set_data_blkaddr(struct dnode_of_data *dn); 3803 void f2fs_update_data_blkaddr(struct dnode_of_data *dn, block_t blkaddr); 3804 int f2fs_reserve_new_blocks(struct dnode_of_data *dn, blkcnt_t count); 3805 int f2fs_reserve_new_block(struct dnode_of_data *dn); 3806 int f2fs_get_block(struct dnode_of_data *dn, pgoff_t index); 3807 int f2fs_reserve_block(struct dnode_of_data *dn, pgoff_t index); 3808 struct page *f2fs_get_read_data_page(struct inode *inode, pgoff_t index, 3809 blk_opf_t op_flags, bool for_write, pgoff_t *next_pgofs); 3810 struct page *f2fs_find_data_page(struct inode *inode, pgoff_t index, 3811 pgoff_t *next_pgofs); 3812 struct page *f2fs_get_lock_data_page(struct inode *inode, pgoff_t index, 3813 bool for_write); 3814 struct page *f2fs_get_new_data_page(struct inode *inode, 3815 struct page *ipage, pgoff_t index, bool new_i_size); 3816 int f2fs_do_write_data_page(struct f2fs_io_info *fio); 3817 void f2fs_do_map_lock(struct f2fs_sb_info *sbi, int flag, bool lock); 3818 int f2fs_map_blocks(struct inode *inode, struct f2fs_map_blocks *map, 3819 int create, int flag); 3820 int f2fs_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, 3821 u64 start, u64 len); 3822 int f2fs_encrypt_one_page(struct f2fs_io_info *fio); 3823 bool f2fs_should_update_inplace(struct inode *inode, struct f2fs_io_info *fio); 3824 bool f2fs_should_update_outplace(struct inode *inode, struct f2fs_io_info *fio); 3825 int f2fs_write_single_data_page(struct page *page, int *submitted, 3826 struct bio **bio, sector_t *last_block, 3827 struct writeback_control *wbc, 3828 enum iostat_type io_type, 3829 int compr_blocks, bool allow_balance); 3830 void f2fs_write_failed(struct inode *inode, loff_t to); 3831 void f2fs_invalidate_folio(struct folio *folio, size_t offset, size_t length); 3832 bool f2fs_release_folio(struct folio *folio, gfp_t wait); 3833 bool f2fs_overwrite_io(struct inode *inode, loff_t pos, size_t len); 3834 void f2fs_clear_page_cache_dirty_tag(struct page *page); 3835 int f2fs_init_post_read_processing(void); 3836 void f2fs_destroy_post_read_processing(void); 3837 int f2fs_init_post_read_wq(struct f2fs_sb_info *sbi); 3838 void f2fs_destroy_post_read_wq(struct f2fs_sb_info *sbi); 3839 extern const struct iomap_ops f2fs_iomap_ops; 3840 3841 /* 3842 * gc.c 3843 */ 3844 int f2fs_start_gc_thread(struct f2fs_sb_info *sbi); 3845 void f2fs_stop_gc_thread(struct f2fs_sb_info *sbi); 3846 block_t f2fs_start_bidx_of_node(unsigned int node_ofs, struct inode *inode); 3847 int f2fs_gc(struct f2fs_sb_info *sbi, struct f2fs_gc_control *gc_control); 3848 void f2fs_build_gc_manager(struct f2fs_sb_info *sbi); 3849 int f2fs_resize_fs(struct f2fs_sb_info *sbi, __u64 block_count); 3850 int __init f2fs_create_garbage_collection_cache(void); 3851 void f2fs_destroy_garbage_collection_cache(void); 3852 3853 /* 3854 * recovery.c 3855 */ 3856 int f2fs_recover_fsync_data(struct f2fs_sb_info *sbi, bool check_only); 3857 bool f2fs_space_for_roll_forward(struct f2fs_sb_info *sbi); 3858 int __init f2fs_create_recovery_cache(void); 3859 void f2fs_destroy_recovery_cache(void); 3860 3861 /* 3862 * debug.c 3863 */ 3864 #ifdef CONFIG_F2FS_STAT_FS 3865 struct f2fs_stat_info { 3866 struct list_head stat_list; 3867 struct f2fs_sb_info *sbi; 3868 int all_area_segs, sit_area_segs, nat_area_segs, ssa_area_segs; 3869 int main_area_segs, main_area_sections, main_area_zones; 3870 unsigned long long hit_cached[NR_EXTENT_CACHES]; 3871 unsigned long long hit_rbtree[NR_EXTENT_CACHES]; 3872 unsigned long long total_ext[NR_EXTENT_CACHES]; 3873 unsigned long long hit_total[NR_EXTENT_CACHES]; 3874 int ext_tree[NR_EXTENT_CACHES]; 3875 int zombie_tree[NR_EXTENT_CACHES]; 3876 int ext_node[NR_EXTENT_CACHES]; 3877 /* to count memory footprint */ 3878 unsigned long long ext_mem[NR_EXTENT_CACHES]; 3879 /* for read extent cache */ 3880 unsigned long long hit_largest; 3881 /* for block age extent cache */ 3882 unsigned long long allocated_data_blocks; 3883 int ndirty_node, ndirty_dent, ndirty_meta, ndirty_imeta; 3884 int ndirty_data, ndirty_qdata; 3885 unsigned int ndirty_dirs, ndirty_files, nquota_files, ndirty_all; 3886 int nats, dirty_nats, sits, dirty_sits; 3887 int free_nids, avail_nids, alloc_nids; 3888 int total_count, utilization; 3889 int bg_gc, nr_wb_cp_data, nr_wb_data; 3890 int nr_rd_data, nr_rd_node, nr_rd_meta; 3891 int nr_dio_read, nr_dio_write; 3892 unsigned int io_skip_bggc, other_skip_bggc; 3893 int nr_flushing, nr_flushed, flush_list_empty; 3894 int nr_discarding, nr_discarded; 3895 int nr_discard_cmd; 3896 unsigned int undiscard_blks; 3897 int nr_issued_ckpt, nr_total_ckpt, nr_queued_ckpt; 3898 unsigned int cur_ckpt_time, peak_ckpt_time; 3899 int inline_xattr, inline_inode, inline_dir, append, update, orphans; 3900 int compr_inode, swapfile_inode; 3901 unsigned long long compr_blocks; 3902 int aw_cnt, max_aw_cnt; 3903 unsigned int valid_count, valid_node_count, valid_inode_count, discard_blks; 3904 unsigned int bimodal, avg_vblocks; 3905 int util_free, util_valid, util_invalid; 3906 int rsvd_segs, overp_segs; 3907 int dirty_count, node_pages, meta_pages, compress_pages; 3908 int compress_page_hit; 3909 int prefree_count, call_count, cp_count, bg_cp_count; 3910 int tot_segs, node_segs, data_segs, free_segs, free_secs; 3911 int bg_node_segs, bg_data_segs; 3912 int tot_blks, data_blks, node_blks; 3913 int bg_data_blks, bg_node_blks; 3914 int curseg[NR_CURSEG_TYPE]; 3915 int cursec[NR_CURSEG_TYPE]; 3916 int curzone[NR_CURSEG_TYPE]; 3917 unsigned int dirty_seg[NR_CURSEG_TYPE]; 3918 unsigned int full_seg[NR_CURSEG_TYPE]; 3919 unsigned int valid_blks[NR_CURSEG_TYPE]; 3920 3921 unsigned int meta_count[META_MAX]; 3922 unsigned int segment_count[2]; 3923 unsigned int block_count[2]; 3924 unsigned int inplace_count; 3925 unsigned long long base_mem, cache_mem, page_mem; 3926 }; 3927 3928 static inline struct f2fs_stat_info *F2FS_STAT(struct f2fs_sb_info *sbi) 3929 { 3930 return (struct f2fs_stat_info *)sbi->stat_info; 3931 } 3932 3933 #define stat_inc_cp_count(si) ((si)->cp_count++) 3934 #define stat_inc_bg_cp_count(si) ((si)->bg_cp_count++) 3935 #define stat_inc_call_count(si) ((si)->call_count++) 3936 #define stat_inc_bggc_count(si) ((si)->bg_gc++) 3937 #define stat_io_skip_bggc_count(sbi) ((sbi)->io_skip_bggc++) 3938 #define stat_other_skip_bggc_count(sbi) ((sbi)->other_skip_bggc++) 3939 #define stat_inc_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]++) 3940 #define stat_dec_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]--) 3941 #define stat_inc_total_hit(sbi, type) (atomic64_inc(&(sbi)->total_hit_ext[type])) 3942 #define stat_inc_rbtree_node_hit(sbi, type) (atomic64_inc(&(sbi)->read_hit_rbtree[type])) 3943 #define stat_inc_largest_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_largest)) 3944 #define stat_inc_cached_node_hit(sbi, type) (atomic64_inc(&(sbi)->read_hit_cached[type])) 3945 #define stat_inc_inline_xattr(inode) \ 3946 do { \ 3947 if (f2fs_has_inline_xattr(inode)) \ 3948 (atomic_inc(&F2FS_I_SB(inode)->inline_xattr)); \ 3949 } while (0) 3950 #define stat_dec_inline_xattr(inode) \ 3951 do { \ 3952 if (f2fs_has_inline_xattr(inode)) \ 3953 (atomic_dec(&F2FS_I_SB(inode)->inline_xattr)); \ 3954 } while (0) 3955 #define stat_inc_inline_inode(inode) \ 3956 do { \ 3957 if (f2fs_has_inline_data(inode)) \ 3958 (atomic_inc(&F2FS_I_SB(inode)->inline_inode)); \ 3959 } while (0) 3960 #define stat_dec_inline_inode(inode) \ 3961 do { \ 3962 if (f2fs_has_inline_data(inode)) \ 3963 (atomic_dec(&F2FS_I_SB(inode)->inline_inode)); \ 3964 } while (0) 3965 #define stat_inc_inline_dir(inode) \ 3966 do { \ 3967 if (f2fs_has_inline_dentry(inode)) \ 3968 (atomic_inc(&F2FS_I_SB(inode)->inline_dir)); \ 3969 } while (0) 3970 #define stat_dec_inline_dir(inode) \ 3971 do { \ 3972 if (f2fs_has_inline_dentry(inode)) \ 3973 (atomic_dec(&F2FS_I_SB(inode)->inline_dir)); \ 3974 } while (0) 3975 #define stat_inc_compr_inode(inode) \ 3976 do { \ 3977 if (f2fs_compressed_file(inode)) \ 3978 (atomic_inc(&F2FS_I_SB(inode)->compr_inode)); \ 3979 } while (0) 3980 #define stat_dec_compr_inode(inode) \ 3981 do { \ 3982 if (f2fs_compressed_file(inode)) \ 3983 (atomic_dec(&F2FS_I_SB(inode)->compr_inode)); \ 3984 } while (0) 3985 #define stat_add_compr_blocks(inode, blocks) \ 3986 (atomic64_add(blocks, &F2FS_I_SB(inode)->compr_blocks)) 3987 #define stat_sub_compr_blocks(inode, blocks) \ 3988 (atomic64_sub(blocks, &F2FS_I_SB(inode)->compr_blocks)) 3989 #define stat_inc_swapfile_inode(inode) \ 3990 (atomic_inc(&F2FS_I_SB(inode)->swapfile_inode)) 3991 #define stat_dec_swapfile_inode(inode) \ 3992 (atomic_dec(&F2FS_I_SB(inode)->swapfile_inode)) 3993 #define stat_inc_atomic_inode(inode) \ 3994 (atomic_inc(&F2FS_I_SB(inode)->atomic_files)) 3995 #define stat_dec_atomic_inode(inode) \ 3996 (atomic_dec(&F2FS_I_SB(inode)->atomic_files)) 3997 #define stat_inc_meta_count(sbi, blkaddr) \ 3998 do { \ 3999 if (blkaddr < SIT_I(sbi)->sit_base_addr) \ 4000 atomic_inc(&(sbi)->meta_count[META_CP]); \ 4001 else if (blkaddr < NM_I(sbi)->nat_blkaddr) \ 4002 atomic_inc(&(sbi)->meta_count[META_SIT]); \ 4003 else if (blkaddr < SM_I(sbi)->ssa_blkaddr) \ 4004 atomic_inc(&(sbi)->meta_count[META_NAT]); \ 4005 else if (blkaddr < SM_I(sbi)->main_blkaddr) \ 4006 atomic_inc(&(sbi)->meta_count[META_SSA]); \ 4007 } while (0) 4008 #define stat_inc_seg_type(sbi, curseg) \ 4009 ((sbi)->segment_count[(curseg)->alloc_type]++) 4010 #define stat_inc_block_count(sbi, curseg) \ 4011 ((sbi)->block_count[(curseg)->alloc_type]++) 4012 #define stat_inc_inplace_blocks(sbi) \ 4013 (atomic_inc(&(sbi)->inplace_count)) 4014 #define stat_update_max_atomic_write(inode) \ 4015 do { \ 4016 int cur = atomic_read(&F2FS_I_SB(inode)->atomic_files); \ 4017 int max = atomic_read(&F2FS_I_SB(inode)->max_aw_cnt); \ 4018 if (cur > max) \ 4019 atomic_set(&F2FS_I_SB(inode)->max_aw_cnt, cur); \ 4020 } while (0) 4021 #define stat_inc_seg_count(sbi, type, gc_type) \ 4022 do { \ 4023 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 4024 si->tot_segs++; \ 4025 if ((type) == SUM_TYPE_DATA) { \ 4026 si->data_segs++; \ 4027 si->bg_data_segs += (gc_type == BG_GC) ? 1 : 0; \ 4028 } else { \ 4029 si->node_segs++; \ 4030 si->bg_node_segs += (gc_type == BG_GC) ? 1 : 0; \ 4031 } \ 4032 } while (0) 4033 4034 #define stat_inc_tot_blk_count(si, blks) \ 4035 ((si)->tot_blks += (blks)) 4036 4037 #define stat_inc_data_blk_count(sbi, blks, gc_type) \ 4038 do { \ 4039 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 4040 stat_inc_tot_blk_count(si, blks); \ 4041 si->data_blks += (blks); \ 4042 si->bg_data_blks += ((gc_type) == BG_GC) ? (blks) : 0; \ 4043 } while (0) 4044 4045 #define stat_inc_node_blk_count(sbi, blks, gc_type) \ 4046 do { \ 4047 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 4048 stat_inc_tot_blk_count(si, blks); \ 4049 si->node_blks += (blks); \ 4050 si->bg_node_blks += ((gc_type) == BG_GC) ? (blks) : 0; \ 4051 } while (0) 4052 4053 int f2fs_build_stats(struct f2fs_sb_info *sbi); 4054 void f2fs_destroy_stats(struct f2fs_sb_info *sbi); 4055 void __init f2fs_create_root_stats(void); 4056 void f2fs_destroy_root_stats(void); 4057 void f2fs_update_sit_info(struct f2fs_sb_info *sbi); 4058 #else 4059 #define stat_inc_cp_count(si) do { } while (0) 4060 #define stat_inc_bg_cp_count(si) do { } while (0) 4061 #define stat_inc_call_count(si) do { } while (0) 4062 #define stat_inc_bggc_count(si) do { } while (0) 4063 #define stat_io_skip_bggc_count(sbi) do { } while (0) 4064 #define stat_other_skip_bggc_count(sbi) do { } while (0) 4065 #define stat_inc_dirty_inode(sbi, type) do { } while (0) 4066 #define stat_dec_dirty_inode(sbi, type) do { } while (0) 4067 #define stat_inc_total_hit(sbi, type) do { } while (0) 4068 #define stat_inc_rbtree_node_hit(sbi, type) do { } while (0) 4069 #define stat_inc_largest_node_hit(sbi) do { } while (0) 4070 #define stat_inc_cached_node_hit(sbi, type) do { } while (0) 4071 #define stat_inc_inline_xattr(inode) do { } while (0) 4072 #define stat_dec_inline_xattr(inode) do { } while (0) 4073 #define stat_inc_inline_inode(inode) do { } while (0) 4074 #define stat_dec_inline_inode(inode) do { } while (0) 4075 #define stat_inc_inline_dir(inode) do { } while (0) 4076 #define stat_dec_inline_dir(inode) do { } while (0) 4077 #define stat_inc_compr_inode(inode) do { } while (0) 4078 #define stat_dec_compr_inode(inode) do { } while (0) 4079 #define stat_add_compr_blocks(inode, blocks) do { } while (0) 4080 #define stat_sub_compr_blocks(inode, blocks) do { } while (0) 4081 #define stat_inc_swapfile_inode(inode) do { } while (0) 4082 #define stat_dec_swapfile_inode(inode) do { } while (0) 4083 #define stat_inc_atomic_inode(inode) do { } while (0) 4084 #define stat_dec_atomic_inode(inode) do { } while (0) 4085 #define stat_update_max_atomic_write(inode) do { } while (0) 4086 #define stat_inc_meta_count(sbi, blkaddr) do { } while (0) 4087 #define stat_inc_seg_type(sbi, curseg) do { } while (0) 4088 #define stat_inc_block_count(sbi, curseg) do { } while (0) 4089 #define stat_inc_inplace_blocks(sbi) do { } while (0) 4090 #define stat_inc_seg_count(sbi, type, gc_type) do { } while (0) 4091 #define stat_inc_tot_blk_count(si, blks) do { } while (0) 4092 #define stat_inc_data_blk_count(sbi, blks, gc_type) do { } while (0) 4093 #define stat_inc_node_blk_count(sbi, blks, gc_type) do { } while (0) 4094 4095 static inline int f2fs_build_stats(struct f2fs_sb_info *sbi) { return 0; } 4096 static inline void f2fs_destroy_stats(struct f2fs_sb_info *sbi) { } 4097 static inline void __init f2fs_create_root_stats(void) { } 4098 static inline void f2fs_destroy_root_stats(void) { } 4099 static inline void f2fs_update_sit_info(struct f2fs_sb_info *sbi) {} 4100 #endif 4101 4102 extern const struct file_operations f2fs_dir_operations; 4103 extern const struct file_operations f2fs_file_operations; 4104 extern const struct inode_operations f2fs_file_inode_operations; 4105 extern const struct address_space_operations f2fs_dblock_aops; 4106 extern const struct address_space_operations f2fs_node_aops; 4107 extern const struct address_space_operations f2fs_meta_aops; 4108 extern const struct inode_operations f2fs_dir_inode_operations; 4109 extern const struct inode_operations f2fs_symlink_inode_operations; 4110 extern const struct inode_operations f2fs_encrypted_symlink_inode_operations; 4111 extern const struct inode_operations f2fs_special_inode_operations; 4112 extern struct kmem_cache *f2fs_inode_entry_slab; 4113 4114 /* 4115 * inline.c 4116 */ 4117 bool f2fs_may_inline_data(struct inode *inode); 4118 bool f2fs_sanity_check_inline_data(struct inode *inode); 4119 bool f2fs_may_inline_dentry(struct inode *inode); 4120 void f2fs_do_read_inline_data(struct page *page, struct page *ipage); 4121 void f2fs_truncate_inline_inode(struct inode *inode, 4122 struct page *ipage, u64 from); 4123 int f2fs_read_inline_data(struct inode *inode, struct page *page); 4124 int f2fs_convert_inline_page(struct dnode_of_data *dn, struct page *page); 4125 int f2fs_convert_inline_inode(struct inode *inode); 4126 int f2fs_try_convert_inline_dir(struct inode *dir, struct dentry *dentry); 4127 int f2fs_write_inline_data(struct inode *inode, struct page *page); 4128 int f2fs_recover_inline_data(struct inode *inode, struct page *npage); 4129 struct f2fs_dir_entry *f2fs_find_in_inline_dir(struct inode *dir, 4130 const struct f2fs_filename *fname, 4131 struct page **res_page); 4132 int f2fs_make_empty_inline_dir(struct inode *inode, struct inode *parent, 4133 struct page *ipage); 4134 int f2fs_add_inline_entry(struct inode *dir, const struct f2fs_filename *fname, 4135 struct inode *inode, nid_t ino, umode_t mode); 4136 void f2fs_delete_inline_entry(struct f2fs_dir_entry *dentry, 4137 struct page *page, struct inode *dir, 4138 struct inode *inode); 4139 bool f2fs_empty_inline_dir(struct inode *dir); 4140 int f2fs_read_inline_dir(struct file *file, struct dir_context *ctx, 4141 struct fscrypt_str *fstr); 4142 int f2fs_inline_data_fiemap(struct inode *inode, 4143 struct fiemap_extent_info *fieinfo, 4144 __u64 start, __u64 len); 4145 4146 /* 4147 * shrinker.c 4148 */ 4149 unsigned long f2fs_shrink_count(struct shrinker *shrink, 4150 struct shrink_control *sc); 4151 unsigned long f2fs_shrink_scan(struct shrinker *shrink, 4152 struct shrink_control *sc); 4153 void f2fs_join_shrinker(struct f2fs_sb_info *sbi); 4154 void f2fs_leave_shrinker(struct f2fs_sb_info *sbi); 4155 4156 /* 4157 * extent_cache.c 4158 */ 4159 struct rb_entry *f2fs_lookup_rb_tree(struct rb_root_cached *root, 4160 struct rb_entry *cached_re, unsigned int ofs); 4161 struct rb_node **f2fs_lookup_rb_tree_ext(struct f2fs_sb_info *sbi, 4162 struct rb_root_cached *root, 4163 struct rb_node **parent, 4164 unsigned long long key, bool *left_most); 4165 struct rb_node **f2fs_lookup_rb_tree_for_insert(struct f2fs_sb_info *sbi, 4166 struct rb_root_cached *root, 4167 struct rb_node **parent, 4168 unsigned int ofs, bool *leftmost); 4169 struct rb_entry *f2fs_lookup_rb_tree_ret(struct rb_root_cached *root, 4170 struct rb_entry *cached_re, unsigned int ofs, 4171 struct rb_entry **prev_entry, struct rb_entry **next_entry, 4172 struct rb_node ***insert_p, struct rb_node **insert_parent, 4173 bool force, bool *leftmost); 4174 bool f2fs_check_rb_tree_consistence(struct f2fs_sb_info *sbi, 4175 struct rb_root_cached *root, bool check_key); 4176 void f2fs_init_extent_tree(struct inode *inode); 4177 void f2fs_drop_extent_tree(struct inode *inode); 4178 void f2fs_destroy_extent_node(struct inode *inode); 4179 void f2fs_destroy_extent_tree(struct inode *inode); 4180 void f2fs_init_extent_cache_info(struct f2fs_sb_info *sbi); 4181 int __init f2fs_create_extent_cache(void); 4182 void f2fs_destroy_extent_cache(void); 4183 4184 /* read extent cache ops */ 4185 void f2fs_init_read_extent_tree(struct inode *inode, struct page *ipage); 4186 bool f2fs_lookup_read_extent_cache(struct inode *inode, pgoff_t pgofs, 4187 struct extent_info *ei); 4188 void f2fs_update_read_extent_cache(struct dnode_of_data *dn); 4189 void f2fs_update_read_extent_cache_range(struct dnode_of_data *dn, 4190 pgoff_t fofs, block_t blkaddr, unsigned int len); 4191 unsigned int f2fs_shrink_read_extent_tree(struct f2fs_sb_info *sbi, 4192 int nr_shrink); 4193 4194 /* block age extent cache ops */ 4195 void f2fs_init_age_extent_tree(struct inode *inode); 4196 bool f2fs_lookup_age_extent_cache(struct inode *inode, pgoff_t pgofs, 4197 struct extent_info *ei); 4198 void f2fs_update_age_extent_cache(struct dnode_of_data *dn); 4199 void f2fs_update_age_extent_cache_range(struct dnode_of_data *dn, 4200 pgoff_t fofs, unsigned int len); 4201 unsigned int f2fs_shrink_age_extent_tree(struct f2fs_sb_info *sbi, 4202 int nr_shrink); 4203 4204 /* 4205 * sysfs.c 4206 */ 4207 #define MIN_RA_MUL 2 4208 #define MAX_RA_MUL 256 4209 4210 int __init f2fs_init_sysfs(void); 4211 void f2fs_exit_sysfs(void); 4212 int f2fs_register_sysfs(struct f2fs_sb_info *sbi); 4213 void f2fs_unregister_sysfs(struct f2fs_sb_info *sbi); 4214 4215 /* verity.c */ 4216 extern const struct fsverity_operations f2fs_verityops; 4217 4218 /* 4219 * crypto support 4220 */ 4221 static inline bool f2fs_encrypted_file(struct inode *inode) 4222 { 4223 return IS_ENCRYPTED(inode) && S_ISREG(inode->i_mode); 4224 } 4225 4226 static inline void f2fs_set_encrypted_inode(struct inode *inode) 4227 { 4228 #ifdef CONFIG_FS_ENCRYPTION 4229 file_set_encrypt(inode); 4230 f2fs_set_inode_flags(inode); 4231 #endif 4232 } 4233 4234 /* 4235 * Returns true if the reads of the inode's data need to undergo some 4236 * postprocessing step, like decryption or authenticity verification. 4237 */ 4238 static inline bool f2fs_post_read_required(struct inode *inode) 4239 { 4240 return f2fs_encrypted_file(inode) || fsverity_active(inode) || 4241 f2fs_compressed_file(inode); 4242 } 4243 4244 /* 4245 * compress.c 4246 */ 4247 #ifdef CONFIG_F2FS_FS_COMPRESSION 4248 bool f2fs_is_compressed_page(struct page *page); 4249 struct page *f2fs_compress_control_page(struct page *page); 4250 int f2fs_prepare_compress_overwrite(struct inode *inode, 4251 struct page **pagep, pgoff_t index, void **fsdata); 4252 bool f2fs_compress_write_end(struct inode *inode, void *fsdata, 4253 pgoff_t index, unsigned copied); 4254 int f2fs_truncate_partial_cluster(struct inode *inode, u64 from, bool lock); 4255 void f2fs_compress_write_end_io(struct bio *bio, struct page *page); 4256 bool f2fs_is_compress_backend_ready(struct inode *inode); 4257 int f2fs_init_compress_mempool(void); 4258 void f2fs_destroy_compress_mempool(void); 4259 void f2fs_decompress_cluster(struct decompress_io_ctx *dic, bool in_task); 4260 void f2fs_end_read_compressed_page(struct page *page, bool failed, 4261 block_t blkaddr, bool in_task); 4262 bool f2fs_cluster_is_empty(struct compress_ctx *cc); 4263 bool f2fs_cluster_can_merge_page(struct compress_ctx *cc, pgoff_t index); 4264 bool f2fs_all_cluster_page_ready(struct compress_ctx *cc, struct page **pages, 4265 int index, int nr_pages, bool uptodate); 4266 bool f2fs_sanity_check_cluster(struct dnode_of_data *dn); 4267 void f2fs_compress_ctx_add_page(struct compress_ctx *cc, struct page *page); 4268 int f2fs_write_multi_pages(struct compress_ctx *cc, 4269 int *submitted, 4270 struct writeback_control *wbc, 4271 enum iostat_type io_type); 4272 int f2fs_is_compressed_cluster(struct inode *inode, pgoff_t index); 4273 void f2fs_update_read_extent_tree_range_compressed(struct inode *inode, 4274 pgoff_t fofs, block_t blkaddr, 4275 unsigned int llen, unsigned int c_len); 4276 int f2fs_read_multi_pages(struct compress_ctx *cc, struct bio **bio_ret, 4277 unsigned nr_pages, sector_t *last_block_in_bio, 4278 bool is_readahead, bool for_write); 4279 struct decompress_io_ctx *f2fs_alloc_dic(struct compress_ctx *cc); 4280 void f2fs_decompress_end_io(struct decompress_io_ctx *dic, bool failed, 4281 bool in_task); 4282 void f2fs_put_page_dic(struct page *page, bool in_task); 4283 unsigned int f2fs_cluster_blocks_are_contiguous(struct dnode_of_data *dn); 4284 int f2fs_init_compress_ctx(struct compress_ctx *cc); 4285 void f2fs_destroy_compress_ctx(struct compress_ctx *cc, bool reuse); 4286 void f2fs_init_compress_info(struct f2fs_sb_info *sbi); 4287 int f2fs_init_compress_inode(struct f2fs_sb_info *sbi); 4288 void f2fs_destroy_compress_inode(struct f2fs_sb_info *sbi); 4289 int f2fs_init_page_array_cache(struct f2fs_sb_info *sbi); 4290 void f2fs_destroy_page_array_cache(struct f2fs_sb_info *sbi); 4291 int __init f2fs_init_compress_cache(void); 4292 void f2fs_destroy_compress_cache(void); 4293 struct address_space *COMPRESS_MAPPING(struct f2fs_sb_info *sbi); 4294 void f2fs_invalidate_compress_page(struct f2fs_sb_info *sbi, block_t blkaddr); 4295 void f2fs_cache_compressed_page(struct f2fs_sb_info *sbi, struct page *page, 4296 nid_t ino, block_t blkaddr); 4297 bool f2fs_load_compressed_page(struct f2fs_sb_info *sbi, struct page *page, 4298 block_t blkaddr); 4299 void f2fs_invalidate_compress_pages(struct f2fs_sb_info *sbi, nid_t ino); 4300 #define inc_compr_inode_stat(inode) \ 4301 do { \ 4302 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); \ 4303 sbi->compr_new_inode++; \ 4304 } while (0) 4305 #define add_compr_block_stat(inode, blocks) \ 4306 do { \ 4307 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); \ 4308 int diff = F2FS_I(inode)->i_cluster_size - blocks; \ 4309 sbi->compr_written_block += blocks; \ 4310 sbi->compr_saved_block += diff; \ 4311 } while (0) 4312 #else 4313 static inline bool f2fs_is_compressed_page(struct page *page) { return false; } 4314 static inline bool f2fs_is_compress_backend_ready(struct inode *inode) 4315 { 4316 if (!f2fs_compressed_file(inode)) 4317 return true; 4318 /* not support compression */ 4319 return false; 4320 } 4321 static inline struct page *f2fs_compress_control_page(struct page *page) 4322 { 4323 WARN_ON_ONCE(1); 4324 return ERR_PTR(-EINVAL); 4325 } 4326 static inline int f2fs_init_compress_mempool(void) { return 0; } 4327 static inline void f2fs_destroy_compress_mempool(void) { } 4328 static inline void f2fs_decompress_cluster(struct decompress_io_ctx *dic, 4329 bool in_task) { } 4330 static inline void f2fs_end_read_compressed_page(struct page *page, 4331 bool failed, block_t blkaddr, bool in_task) 4332 { 4333 WARN_ON_ONCE(1); 4334 } 4335 static inline void f2fs_put_page_dic(struct page *page, bool in_task) 4336 { 4337 WARN_ON_ONCE(1); 4338 } 4339 static inline unsigned int f2fs_cluster_blocks_are_contiguous(struct dnode_of_data *dn) { return 0; } 4340 static inline bool f2fs_sanity_check_cluster(struct dnode_of_data *dn) { return false; } 4341 static inline int f2fs_init_compress_inode(struct f2fs_sb_info *sbi) { return 0; } 4342 static inline void f2fs_destroy_compress_inode(struct f2fs_sb_info *sbi) { } 4343 static inline int f2fs_init_page_array_cache(struct f2fs_sb_info *sbi) { return 0; } 4344 static inline void f2fs_destroy_page_array_cache(struct f2fs_sb_info *sbi) { } 4345 static inline int __init f2fs_init_compress_cache(void) { return 0; } 4346 static inline void f2fs_destroy_compress_cache(void) { } 4347 static inline void f2fs_invalidate_compress_page(struct f2fs_sb_info *sbi, 4348 block_t blkaddr) { } 4349 static inline void f2fs_cache_compressed_page(struct f2fs_sb_info *sbi, 4350 struct page *page, nid_t ino, block_t blkaddr) { } 4351 static inline bool f2fs_load_compressed_page(struct f2fs_sb_info *sbi, 4352 struct page *page, block_t blkaddr) { return false; } 4353 static inline void f2fs_invalidate_compress_pages(struct f2fs_sb_info *sbi, 4354 nid_t ino) { } 4355 #define inc_compr_inode_stat(inode) do { } while (0) 4356 static inline void f2fs_update_read_extent_tree_range_compressed( 4357 struct inode *inode, 4358 pgoff_t fofs, block_t blkaddr, 4359 unsigned int llen, unsigned int c_len) { } 4360 #endif 4361 4362 static inline int set_compress_context(struct inode *inode) 4363 { 4364 #ifdef CONFIG_F2FS_FS_COMPRESSION 4365 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 4366 4367 F2FS_I(inode)->i_compress_algorithm = 4368 F2FS_OPTION(sbi).compress_algorithm; 4369 F2FS_I(inode)->i_log_cluster_size = 4370 F2FS_OPTION(sbi).compress_log_size; 4371 F2FS_I(inode)->i_compress_flag = 4372 F2FS_OPTION(sbi).compress_chksum ? 4373 1 << COMPRESS_CHKSUM : 0; 4374 F2FS_I(inode)->i_cluster_size = 4375 1 << F2FS_I(inode)->i_log_cluster_size; 4376 if ((F2FS_I(inode)->i_compress_algorithm == COMPRESS_LZ4 || 4377 F2FS_I(inode)->i_compress_algorithm == COMPRESS_ZSTD) && 4378 F2FS_OPTION(sbi).compress_level) 4379 F2FS_I(inode)->i_compress_flag |= 4380 F2FS_OPTION(sbi).compress_level << 4381 COMPRESS_LEVEL_OFFSET; 4382 F2FS_I(inode)->i_flags |= F2FS_COMPR_FL; 4383 set_inode_flag(inode, FI_COMPRESSED_FILE); 4384 stat_inc_compr_inode(inode); 4385 inc_compr_inode_stat(inode); 4386 f2fs_mark_inode_dirty_sync(inode, true); 4387 return 0; 4388 #else 4389 return -EOPNOTSUPP; 4390 #endif 4391 } 4392 4393 static inline bool f2fs_disable_compressed_file(struct inode *inode) 4394 { 4395 struct f2fs_inode_info *fi = F2FS_I(inode); 4396 4397 if (!f2fs_compressed_file(inode)) 4398 return true; 4399 if (S_ISREG(inode->i_mode) && F2FS_HAS_BLOCKS(inode)) 4400 return false; 4401 4402 fi->i_flags &= ~F2FS_COMPR_FL; 4403 stat_dec_compr_inode(inode); 4404 clear_inode_flag(inode, FI_COMPRESSED_FILE); 4405 f2fs_mark_inode_dirty_sync(inode, true); 4406 return true; 4407 } 4408 4409 #define F2FS_FEATURE_FUNCS(name, flagname) \ 4410 static inline bool f2fs_sb_has_##name(struct f2fs_sb_info *sbi) \ 4411 { \ 4412 return F2FS_HAS_FEATURE(sbi, F2FS_FEATURE_##flagname); \ 4413 } 4414 4415 F2FS_FEATURE_FUNCS(encrypt, ENCRYPT); 4416 F2FS_FEATURE_FUNCS(blkzoned, BLKZONED); 4417 F2FS_FEATURE_FUNCS(extra_attr, EXTRA_ATTR); 4418 F2FS_FEATURE_FUNCS(project_quota, PRJQUOTA); 4419 F2FS_FEATURE_FUNCS(inode_chksum, INODE_CHKSUM); 4420 F2FS_FEATURE_FUNCS(flexible_inline_xattr, FLEXIBLE_INLINE_XATTR); 4421 F2FS_FEATURE_FUNCS(quota_ino, QUOTA_INO); 4422 F2FS_FEATURE_FUNCS(inode_crtime, INODE_CRTIME); 4423 F2FS_FEATURE_FUNCS(lost_found, LOST_FOUND); 4424 F2FS_FEATURE_FUNCS(verity, VERITY); 4425 F2FS_FEATURE_FUNCS(sb_chksum, SB_CHKSUM); 4426 F2FS_FEATURE_FUNCS(casefold, CASEFOLD); 4427 F2FS_FEATURE_FUNCS(compression, COMPRESSION); 4428 F2FS_FEATURE_FUNCS(readonly, RO); 4429 4430 #ifdef CONFIG_BLK_DEV_ZONED 4431 static inline bool f2fs_blkz_is_seq(struct f2fs_sb_info *sbi, int devi, 4432 block_t blkaddr) 4433 { 4434 unsigned int zno = blkaddr >> sbi->log_blocks_per_blkz; 4435 4436 return test_bit(zno, FDEV(devi).blkz_seq); 4437 } 4438 #endif 4439 4440 static inline bool f2fs_hw_should_discard(struct f2fs_sb_info *sbi) 4441 { 4442 return f2fs_sb_has_blkzoned(sbi); 4443 } 4444 4445 static inline bool f2fs_bdev_support_discard(struct block_device *bdev) 4446 { 4447 return bdev_max_discard_sectors(bdev) || bdev_is_zoned(bdev); 4448 } 4449 4450 static inline bool f2fs_hw_support_discard(struct f2fs_sb_info *sbi) 4451 { 4452 int i; 4453 4454 if (!f2fs_is_multi_device(sbi)) 4455 return f2fs_bdev_support_discard(sbi->sb->s_bdev); 4456 4457 for (i = 0; i < sbi->s_ndevs; i++) 4458 if (f2fs_bdev_support_discard(FDEV(i).bdev)) 4459 return true; 4460 return false; 4461 } 4462 4463 static inline bool f2fs_realtime_discard_enable(struct f2fs_sb_info *sbi) 4464 { 4465 return (test_opt(sbi, DISCARD) && f2fs_hw_support_discard(sbi)) || 4466 f2fs_hw_should_discard(sbi); 4467 } 4468 4469 static inline bool f2fs_hw_is_readonly(struct f2fs_sb_info *sbi) 4470 { 4471 int i; 4472 4473 if (!f2fs_is_multi_device(sbi)) 4474 return bdev_read_only(sbi->sb->s_bdev); 4475 4476 for (i = 0; i < sbi->s_ndevs; i++) 4477 if (bdev_read_only(FDEV(i).bdev)) 4478 return true; 4479 return false; 4480 } 4481 4482 static inline bool f2fs_lfs_mode(struct f2fs_sb_info *sbi) 4483 { 4484 return F2FS_OPTION(sbi).fs_mode == FS_MODE_LFS; 4485 } 4486 4487 static inline bool f2fs_low_mem_mode(struct f2fs_sb_info *sbi) 4488 { 4489 return F2FS_OPTION(sbi).memory_mode == MEMORY_MODE_LOW; 4490 } 4491 4492 static inline bool f2fs_may_compress(struct inode *inode) 4493 { 4494 if (IS_SWAPFILE(inode) || f2fs_is_pinned_file(inode) || 4495 f2fs_is_atomic_file(inode) || f2fs_has_inline_data(inode)) 4496 return false; 4497 return S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode); 4498 } 4499 4500 static inline void f2fs_i_compr_blocks_update(struct inode *inode, 4501 u64 blocks, bool add) 4502 { 4503 struct f2fs_inode_info *fi = F2FS_I(inode); 4504 int diff = fi->i_cluster_size - blocks; 4505 4506 /* don't update i_compr_blocks if saved blocks were released */ 4507 if (!add && !atomic_read(&fi->i_compr_blocks)) 4508 return; 4509 4510 if (add) { 4511 atomic_add(diff, &fi->i_compr_blocks); 4512 stat_add_compr_blocks(inode, diff); 4513 } else { 4514 atomic_sub(diff, &fi->i_compr_blocks); 4515 stat_sub_compr_blocks(inode, diff); 4516 } 4517 f2fs_mark_inode_dirty_sync(inode, true); 4518 } 4519 4520 static inline bool f2fs_allow_multi_device_dio(struct f2fs_sb_info *sbi, 4521 int flag) 4522 { 4523 if (!f2fs_is_multi_device(sbi)) 4524 return false; 4525 if (flag != F2FS_GET_BLOCK_DIO) 4526 return false; 4527 return sbi->aligned_blksize; 4528 } 4529 4530 static inline bool f2fs_need_verity(const struct inode *inode, pgoff_t idx) 4531 { 4532 return fsverity_active(inode) && 4533 idx < DIV_ROUND_UP(inode->i_size, PAGE_SIZE); 4534 } 4535 4536 #ifdef CONFIG_F2FS_FAULT_INJECTION 4537 extern void f2fs_build_fault_attr(struct f2fs_sb_info *sbi, unsigned int rate, 4538 unsigned int type); 4539 #else 4540 #define f2fs_build_fault_attr(sbi, rate, type) do { } while (0) 4541 #endif 4542 4543 static inline bool is_journalled_quota(struct f2fs_sb_info *sbi) 4544 { 4545 #ifdef CONFIG_QUOTA 4546 if (f2fs_sb_has_quota_ino(sbi)) 4547 return true; 4548 if (F2FS_OPTION(sbi).s_qf_names[USRQUOTA] || 4549 F2FS_OPTION(sbi).s_qf_names[GRPQUOTA] || 4550 F2FS_OPTION(sbi).s_qf_names[PRJQUOTA]) 4551 return true; 4552 #endif 4553 return false; 4554 } 4555 4556 static inline bool f2fs_block_unit_discard(struct f2fs_sb_info *sbi) 4557 { 4558 return F2FS_OPTION(sbi).discard_unit == DISCARD_UNIT_BLOCK; 4559 } 4560 4561 static inline void f2fs_io_schedule_timeout(long timeout) 4562 { 4563 set_current_state(TASK_UNINTERRUPTIBLE); 4564 io_schedule_timeout(timeout); 4565 } 4566 4567 static inline void f2fs_handle_page_eio(struct f2fs_sb_info *sbi, pgoff_t ofs, 4568 enum page_type type) 4569 { 4570 if (unlikely(f2fs_cp_error(sbi))) 4571 return; 4572 4573 if (ofs == sbi->page_eio_ofs[type]) { 4574 if (sbi->page_eio_cnt[type]++ == MAX_RETRY_PAGE_EIO) 4575 set_ckpt_flags(sbi, CP_ERROR_FLAG); 4576 } else { 4577 sbi->page_eio_ofs[type] = ofs; 4578 sbi->page_eio_cnt[type] = 0; 4579 } 4580 } 4581 4582 static inline bool f2fs_is_readonly(struct f2fs_sb_info *sbi) 4583 { 4584 return f2fs_sb_has_readonly(sbi) || f2fs_readonly(sbi->sb); 4585 } 4586 4587 #define EFSBADCRC EBADMSG /* Bad CRC detected */ 4588 #define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */ 4589 4590 #endif /* _LINUX_F2FS_H */ 4591