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 struct mutex cmd_lock; 406 unsigned int nr_discards; /* # of discards in the list */ 407 unsigned int max_discards; /* max. discards to be issued */ 408 unsigned int max_discard_request; /* max. discard request per round */ 409 unsigned int min_discard_issue_time; /* min. interval between discard issue */ 410 unsigned int mid_discard_issue_time; /* mid. interval between discard issue */ 411 unsigned int max_discard_issue_time; /* max. interval between discard issue */ 412 unsigned int discard_io_aware_gran; /* minimum discard granularity not be aware of I/O */ 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 bool discard_wake; /* to wake up discard thread */ 424 }; 425 426 /* for the list of fsync inodes, used only during recovery */ 427 struct fsync_inode_entry { 428 struct list_head list; /* list head */ 429 struct inode *inode; /* vfs inode pointer */ 430 block_t blkaddr; /* block address locating the last fsync */ 431 block_t last_dentry; /* block address locating the last dentry */ 432 }; 433 434 #define nats_in_cursum(jnl) (le16_to_cpu((jnl)->n_nats)) 435 #define sits_in_cursum(jnl) (le16_to_cpu((jnl)->n_sits)) 436 437 #define nat_in_journal(jnl, i) ((jnl)->nat_j.entries[i].ne) 438 #define nid_in_journal(jnl, i) ((jnl)->nat_j.entries[i].nid) 439 #define sit_in_journal(jnl, i) ((jnl)->sit_j.entries[i].se) 440 #define segno_in_journal(jnl, i) ((jnl)->sit_j.entries[i].segno) 441 442 #define MAX_NAT_JENTRIES(jnl) (NAT_JOURNAL_ENTRIES - nats_in_cursum(jnl)) 443 #define MAX_SIT_JENTRIES(jnl) (SIT_JOURNAL_ENTRIES - sits_in_cursum(jnl)) 444 445 static inline int update_nats_in_cursum(struct f2fs_journal *journal, int i) 446 { 447 int before = nats_in_cursum(journal); 448 449 journal->n_nats = cpu_to_le16(before + i); 450 return before; 451 } 452 453 static inline int update_sits_in_cursum(struct f2fs_journal *journal, int i) 454 { 455 int before = sits_in_cursum(journal); 456 457 journal->n_sits = cpu_to_le16(before + i); 458 return before; 459 } 460 461 static inline bool __has_cursum_space(struct f2fs_journal *journal, 462 int size, int type) 463 { 464 if (type == NAT_JOURNAL) 465 return size <= MAX_NAT_JENTRIES(journal); 466 return size <= MAX_SIT_JENTRIES(journal); 467 } 468 469 /* for inline stuff */ 470 #define DEF_INLINE_RESERVED_SIZE 1 471 static inline int get_extra_isize(struct inode *inode); 472 static inline int get_inline_xattr_addrs(struct inode *inode); 473 #define MAX_INLINE_DATA(inode) (sizeof(__le32) * \ 474 (CUR_ADDRS_PER_INODE(inode) - \ 475 get_inline_xattr_addrs(inode) - \ 476 DEF_INLINE_RESERVED_SIZE)) 477 478 /* for inline dir */ 479 #define NR_INLINE_DENTRY(inode) (MAX_INLINE_DATA(inode) * BITS_PER_BYTE / \ 480 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \ 481 BITS_PER_BYTE + 1)) 482 #define INLINE_DENTRY_BITMAP_SIZE(inode) \ 483 DIV_ROUND_UP(NR_INLINE_DENTRY(inode), BITS_PER_BYTE) 484 #define INLINE_RESERVED_SIZE(inode) (MAX_INLINE_DATA(inode) - \ 485 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \ 486 NR_INLINE_DENTRY(inode) + \ 487 INLINE_DENTRY_BITMAP_SIZE(inode))) 488 489 /* 490 * For INODE and NODE manager 491 */ 492 /* for directory operations */ 493 494 struct f2fs_filename { 495 /* 496 * The filename the user specified. This is NULL for some 497 * filesystem-internal operations, e.g. converting an inline directory 498 * to a non-inline one, or roll-forward recovering an encrypted dentry. 499 */ 500 const struct qstr *usr_fname; 501 502 /* 503 * The on-disk filename. For encrypted directories, this is encrypted. 504 * This may be NULL for lookups in an encrypted dir without the key. 505 */ 506 struct fscrypt_str disk_name; 507 508 /* The dirhash of this filename */ 509 f2fs_hash_t hash; 510 511 #ifdef CONFIG_FS_ENCRYPTION 512 /* 513 * For lookups in encrypted directories: either the buffer backing 514 * disk_name, or a buffer that holds the decoded no-key name. 515 */ 516 struct fscrypt_str crypto_buf; 517 #endif 518 #if IS_ENABLED(CONFIG_UNICODE) 519 /* 520 * For casefolded directories: the casefolded name, but it's left NULL 521 * if the original name is not valid Unicode, if the original name is 522 * "." or "..", if the directory is both casefolded and encrypted and 523 * its encryption key is unavailable, or if the filesystem is doing an 524 * internal operation where usr_fname is also NULL. In all these cases 525 * we fall back to treating the name as an opaque byte sequence. 526 */ 527 struct fscrypt_str cf_name; 528 #endif 529 }; 530 531 struct f2fs_dentry_ptr { 532 struct inode *inode; 533 void *bitmap; 534 struct f2fs_dir_entry *dentry; 535 __u8 (*filename)[F2FS_SLOT_LEN]; 536 int max; 537 int nr_bitmap; 538 }; 539 540 static inline void make_dentry_ptr_block(struct inode *inode, 541 struct f2fs_dentry_ptr *d, struct f2fs_dentry_block *t) 542 { 543 d->inode = inode; 544 d->max = NR_DENTRY_IN_BLOCK; 545 d->nr_bitmap = SIZE_OF_DENTRY_BITMAP; 546 d->bitmap = t->dentry_bitmap; 547 d->dentry = t->dentry; 548 d->filename = t->filename; 549 } 550 551 static inline void make_dentry_ptr_inline(struct inode *inode, 552 struct f2fs_dentry_ptr *d, void *t) 553 { 554 int entry_cnt = NR_INLINE_DENTRY(inode); 555 int bitmap_size = INLINE_DENTRY_BITMAP_SIZE(inode); 556 int reserved_size = INLINE_RESERVED_SIZE(inode); 557 558 d->inode = inode; 559 d->max = entry_cnt; 560 d->nr_bitmap = bitmap_size; 561 d->bitmap = t; 562 d->dentry = t + bitmap_size + reserved_size; 563 d->filename = t + bitmap_size + reserved_size + 564 SIZE_OF_DIR_ENTRY * entry_cnt; 565 } 566 567 /* 568 * XATTR_NODE_OFFSET stores xattrs to one node block per file keeping -1 569 * as its node offset to distinguish from index node blocks. 570 * But some bits are used to mark the node block. 571 */ 572 #define XATTR_NODE_OFFSET ((((unsigned int)-1) << OFFSET_BIT_SHIFT) \ 573 >> OFFSET_BIT_SHIFT) 574 enum { 575 ALLOC_NODE, /* allocate a new node page if needed */ 576 LOOKUP_NODE, /* look up a node without readahead */ 577 LOOKUP_NODE_RA, /* 578 * look up a node with readahead called 579 * by get_data_block. 580 */ 581 }; 582 583 #define DEFAULT_RETRY_IO_COUNT 8 /* maximum retry read IO or flush count */ 584 585 /* congestion wait timeout value, default: 20ms */ 586 #define DEFAULT_IO_TIMEOUT (msecs_to_jiffies(20)) 587 588 /* maximum retry quota flush count */ 589 #define DEFAULT_RETRY_QUOTA_FLUSH_COUNT 8 590 591 /* maximum retry of EIO'ed page */ 592 #define MAX_RETRY_PAGE_EIO 100 593 594 #define F2FS_LINK_MAX 0xffffffff /* maximum link count per file */ 595 596 #define MAX_DIR_RA_PAGES 4 /* maximum ra pages of dir */ 597 598 /* dirty segments threshold for triggering CP */ 599 #define DEFAULT_DIRTY_THRESHOLD 4 600 601 #define RECOVERY_MAX_RA_BLOCKS BIO_MAX_VECS 602 #define RECOVERY_MIN_RA_BLOCKS 1 603 604 #define F2FS_ONSTACK_PAGES 16 /* nr of onstack pages */ 605 606 /* for in-memory extent cache entry */ 607 #define F2FS_MIN_EXTENT_LEN 64 /* minimum extent length */ 608 609 /* number of extent info in extent cache we try to shrink */ 610 #define READ_EXTENT_CACHE_SHRINK_NUMBER 128 611 612 /* number of age extent info in extent cache we try to shrink */ 613 #define AGE_EXTENT_CACHE_SHRINK_NUMBER 128 614 #define LAST_AGE_WEIGHT 30 615 #define SAME_AGE_REGION 1024 616 617 /* 618 * Define data block with age less than 1GB as hot data 619 * define data block with age less than 10GB but more than 1GB as warm data 620 */ 621 #define DEF_HOT_DATA_AGE_THRESHOLD 262144 622 #define DEF_WARM_DATA_AGE_THRESHOLD 2621440 623 624 /* extent cache type */ 625 enum extent_type { 626 EX_READ, 627 EX_BLOCK_AGE, 628 NR_EXTENT_CACHES, 629 }; 630 631 struct rb_entry { 632 struct rb_node rb_node; /* rb node located in rb-tree */ 633 union { 634 struct { 635 unsigned int ofs; /* start offset of the entry */ 636 unsigned int len; /* length of the entry */ 637 }; 638 unsigned long long key; /* 64-bits key */ 639 } __packed; 640 }; 641 642 struct extent_info { 643 unsigned int fofs; /* start offset in a file */ 644 unsigned int len; /* length of the extent */ 645 union { 646 /* read extent_cache */ 647 struct { 648 /* start block address of the extent */ 649 block_t blk; 650 #ifdef CONFIG_F2FS_FS_COMPRESSION 651 /* physical extent length of compressed blocks */ 652 unsigned int c_len; 653 #endif 654 }; 655 /* block age extent_cache */ 656 struct { 657 /* block age of the extent */ 658 unsigned long long age; 659 /* last total blocks allocated */ 660 unsigned long long last_blocks; 661 }; 662 }; 663 }; 664 665 struct extent_node { 666 struct rb_node rb_node; /* rb node located in rb-tree */ 667 struct extent_info ei; /* extent info */ 668 struct list_head list; /* node in global extent list of sbi */ 669 struct extent_tree *et; /* extent tree pointer */ 670 }; 671 672 struct extent_tree { 673 nid_t ino; /* inode number */ 674 enum extent_type type; /* keep the extent tree type */ 675 struct rb_root_cached root; /* root of extent info rb-tree */ 676 struct extent_node *cached_en; /* recently accessed extent node */ 677 struct list_head list; /* to be used by sbi->zombie_list */ 678 rwlock_t lock; /* protect extent info rb-tree */ 679 atomic_t node_cnt; /* # of extent node in rb-tree*/ 680 bool largest_updated; /* largest extent updated */ 681 struct extent_info largest; /* largest cached extent for EX_READ */ 682 }; 683 684 struct extent_tree_info { 685 struct radix_tree_root extent_tree_root;/* cache extent cache entries */ 686 struct mutex extent_tree_lock; /* locking extent radix tree */ 687 struct list_head extent_list; /* lru list for shrinker */ 688 spinlock_t extent_lock; /* locking extent lru list */ 689 atomic_t total_ext_tree; /* extent tree count */ 690 struct list_head zombie_list; /* extent zombie tree list */ 691 atomic_t total_zombie_tree; /* extent zombie tree count */ 692 atomic_t total_ext_node; /* extent info count */ 693 }; 694 695 /* 696 * State of block returned by f2fs_map_blocks. 697 */ 698 #define F2FS_MAP_NEW (1U << 0) 699 #define F2FS_MAP_MAPPED (1U << 1) 700 #define F2FS_MAP_DELALLOC (1U << 2) 701 #define F2FS_MAP_FLAGS (F2FS_MAP_NEW | F2FS_MAP_MAPPED |\ 702 F2FS_MAP_DELALLOC) 703 704 struct f2fs_map_blocks { 705 struct block_device *m_bdev; /* for multi-device dio */ 706 block_t m_pblk; 707 block_t m_lblk; 708 unsigned int m_len; 709 unsigned int m_flags; 710 pgoff_t *m_next_pgofs; /* point next possible non-hole pgofs */ 711 pgoff_t *m_next_extent; /* point to next possible extent */ 712 int m_seg_type; 713 bool m_may_create; /* indicate it is from write path */ 714 bool m_multidev_dio; /* indicate it allows multi-device dio */ 715 }; 716 717 /* for flag in get_data_block */ 718 enum { 719 F2FS_GET_BLOCK_DEFAULT, 720 F2FS_GET_BLOCK_FIEMAP, 721 F2FS_GET_BLOCK_BMAP, 722 F2FS_GET_BLOCK_DIO, 723 F2FS_GET_BLOCK_PRE_DIO, 724 F2FS_GET_BLOCK_PRE_AIO, 725 F2FS_GET_BLOCK_PRECACHE, 726 }; 727 728 /* 729 * i_advise uses FADVISE_XXX_BIT. We can add additional hints later. 730 */ 731 #define FADVISE_COLD_BIT 0x01 732 #define FADVISE_LOST_PINO_BIT 0x02 733 #define FADVISE_ENCRYPT_BIT 0x04 734 #define FADVISE_ENC_NAME_BIT 0x08 735 #define FADVISE_KEEP_SIZE_BIT 0x10 736 #define FADVISE_HOT_BIT 0x20 737 #define FADVISE_VERITY_BIT 0x40 738 #define FADVISE_TRUNC_BIT 0x80 739 740 #define FADVISE_MODIFIABLE_BITS (FADVISE_COLD_BIT | FADVISE_HOT_BIT) 741 742 #define file_is_cold(inode) is_file(inode, FADVISE_COLD_BIT) 743 #define file_set_cold(inode) set_file(inode, FADVISE_COLD_BIT) 744 #define file_clear_cold(inode) clear_file(inode, FADVISE_COLD_BIT) 745 746 #define file_wrong_pino(inode) is_file(inode, FADVISE_LOST_PINO_BIT) 747 #define file_lost_pino(inode) set_file(inode, FADVISE_LOST_PINO_BIT) 748 #define file_got_pino(inode) clear_file(inode, FADVISE_LOST_PINO_BIT) 749 750 #define file_is_encrypt(inode) is_file(inode, FADVISE_ENCRYPT_BIT) 751 #define file_set_encrypt(inode) set_file(inode, FADVISE_ENCRYPT_BIT) 752 753 #define file_enc_name(inode) is_file(inode, FADVISE_ENC_NAME_BIT) 754 #define file_set_enc_name(inode) set_file(inode, FADVISE_ENC_NAME_BIT) 755 756 #define file_keep_isize(inode) is_file(inode, FADVISE_KEEP_SIZE_BIT) 757 #define file_set_keep_isize(inode) set_file(inode, FADVISE_KEEP_SIZE_BIT) 758 759 #define file_is_hot(inode) is_file(inode, FADVISE_HOT_BIT) 760 #define file_set_hot(inode) set_file(inode, FADVISE_HOT_BIT) 761 #define file_clear_hot(inode) clear_file(inode, FADVISE_HOT_BIT) 762 763 #define file_is_verity(inode) is_file(inode, FADVISE_VERITY_BIT) 764 #define file_set_verity(inode) set_file(inode, FADVISE_VERITY_BIT) 765 766 #define file_should_truncate(inode) is_file(inode, FADVISE_TRUNC_BIT) 767 #define file_need_truncate(inode) set_file(inode, FADVISE_TRUNC_BIT) 768 #define file_dont_truncate(inode) clear_file(inode, FADVISE_TRUNC_BIT) 769 770 #define DEF_DIR_LEVEL 0 771 772 enum { 773 GC_FAILURE_PIN, 774 MAX_GC_FAILURE 775 }; 776 777 /* used for f2fs_inode_info->flags */ 778 enum { 779 FI_NEW_INODE, /* indicate newly allocated inode */ 780 FI_DIRTY_INODE, /* indicate inode is dirty or not */ 781 FI_AUTO_RECOVER, /* indicate inode is recoverable */ 782 FI_DIRTY_DIR, /* indicate directory has dirty pages */ 783 FI_INC_LINK, /* need to increment i_nlink */ 784 FI_ACL_MODE, /* indicate acl mode */ 785 FI_NO_ALLOC, /* should not allocate any blocks */ 786 FI_FREE_NID, /* free allocated nide */ 787 FI_NO_EXTENT, /* not to use the extent cache */ 788 FI_INLINE_XATTR, /* used for inline xattr */ 789 FI_INLINE_DATA, /* used for inline data*/ 790 FI_INLINE_DENTRY, /* used for inline dentry */ 791 FI_APPEND_WRITE, /* inode has appended data */ 792 FI_UPDATE_WRITE, /* inode has in-place-update data */ 793 FI_NEED_IPU, /* used for ipu per file */ 794 FI_ATOMIC_FILE, /* indicate atomic file */ 795 FI_FIRST_BLOCK_WRITTEN, /* indicate #0 data block was written */ 796 FI_DROP_CACHE, /* drop dirty page cache */ 797 FI_DATA_EXIST, /* indicate data exists */ 798 FI_INLINE_DOTS, /* indicate inline dot dentries */ 799 FI_SKIP_WRITES, /* should skip data page writeback */ 800 FI_OPU_WRITE, /* used for opu per file */ 801 FI_DIRTY_FILE, /* indicate regular/symlink has dirty pages */ 802 FI_PREALLOCATED_ALL, /* all blocks for write were preallocated */ 803 FI_HOT_DATA, /* indicate file is hot */ 804 FI_EXTRA_ATTR, /* indicate file has extra attribute */ 805 FI_PROJ_INHERIT, /* indicate file inherits projectid */ 806 FI_PIN_FILE, /* indicate file should not be gced */ 807 FI_VERITY_IN_PROGRESS, /* building fs-verity Merkle tree */ 808 FI_COMPRESSED_FILE, /* indicate file's data can be compressed */ 809 FI_COMPRESS_CORRUPT, /* indicate compressed cluster is corrupted */ 810 FI_MMAP_FILE, /* indicate file was mmapped */ 811 FI_ENABLE_COMPRESS, /* enable compression in "user" compression mode */ 812 FI_COMPRESS_RELEASED, /* compressed blocks were released */ 813 FI_ALIGNED_WRITE, /* enable aligned write */ 814 FI_COW_FILE, /* indicate COW file */ 815 FI_ATOMIC_COMMITTED, /* indicate atomic commit completed except disk sync */ 816 FI_ATOMIC_REPLACE, /* indicate atomic replace */ 817 FI_MAX, /* max flag, never be used */ 818 }; 819 820 struct f2fs_inode_info { 821 struct inode vfs_inode; /* serve a vfs inode */ 822 unsigned long i_flags; /* keep an inode flags for ioctl */ 823 unsigned char i_advise; /* use to give file attribute hints */ 824 unsigned char i_dir_level; /* use for dentry level for large dir */ 825 unsigned int i_current_depth; /* only for directory depth */ 826 /* for gc failure statistic */ 827 unsigned int i_gc_failures[MAX_GC_FAILURE]; 828 unsigned int i_pino; /* parent inode number */ 829 umode_t i_acl_mode; /* keep file acl mode temporarily */ 830 831 /* Use below internally in f2fs*/ 832 unsigned long flags[BITS_TO_LONGS(FI_MAX)]; /* use to pass per-file flags */ 833 struct f2fs_rwsem i_sem; /* protect fi info */ 834 atomic_t dirty_pages; /* # of dirty pages */ 835 f2fs_hash_t chash; /* hash value of given file name */ 836 unsigned int clevel; /* maximum level of given file name */ 837 struct task_struct *task; /* lookup and create consistency */ 838 struct task_struct *cp_task; /* separate cp/wb IO stats*/ 839 struct task_struct *wb_task; /* indicate inode is in context of writeback */ 840 nid_t i_xattr_nid; /* node id that contains xattrs */ 841 loff_t last_disk_size; /* lastly written file size */ 842 spinlock_t i_size_lock; /* protect last_disk_size */ 843 844 #ifdef CONFIG_QUOTA 845 struct dquot *i_dquot[MAXQUOTAS]; 846 847 /* quota space reservation, managed internally by quota code */ 848 qsize_t i_reserved_quota; 849 #endif 850 struct list_head dirty_list; /* dirty list for dirs and files */ 851 struct list_head gdirty_list; /* linked in global dirty list */ 852 struct task_struct *atomic_write_task; /* store atomic write task */ 853 struct extent_tree *extent_tree[NR_EXTENT_CACHES]; 854 /* cached extent_tree entry */ 855 struct inode *cow_inode; /* copy-on-write inode for atomic write */ 856 857 /* avoid racing between foreground op and gc */ 858 struct f2fs_rwsem i_gc_rwsem[2]; 859 struct f2fs_rwsem i_xattr_sem; /* avoid racing between reading and changing EAs */ 860 861 int i_extra_isize; /* size of extra space located in i_addr */ 862 kprojid_t i_projid; /* id for project quota */ 863 int i_inline_xattr_size; /* inline xattr size */ 864 struct timespec64 i_crtime; /* inode creation time */ 865 struct timespec64 i_disk_time[4];/* inode disk times */ 866 867 /* for file compress */ 868 atomic_t i_compr_blocks; /* # of compressed blocks */ 869 unsigned char i_compress_algorithm; /* algorithm type */ 870 unsigned char i_log_cluster_size; /* log of cluster size */ 871 unsigned char i_compress_level; /* compress level (lz4hc,zstd) */ 872 unsigned char i_compress_flag; /* compress flag */ 873 unsigned int i_cluster_size; /* cluster size */ 874 875 unsigned int atomic_write_cnt; 876 loff_t original_i_size; /* original i_size before atomic write */ 877 }; 878 879 static inline void get_read_extent_info(struct extent_info *ext, 880 struct f2fs_extent *i_ext) 881 { 882 ext->fofs = le32_to_cpu(i_ext->fofs); 883 ext->blk = le32_to_cpu(i_ext->blk); 884 ext->len = le32_to_cpu(i_ext->len); 885 } 886 887 static inline void set_raw_read_extent(struct extent_info *ext, 888 struct f2fs_extent *i_ext) 889 { 890 i_ext->fofs = cpu_to_le32(ext->fofs); 891 i_ext->blk = cpu_to_le32(ext->blk); 892 i_ext->len = cpu_to_le32(ext->len); 893 } 894 895 static inline bool __is_discard_mergeable(struct discard_info *back, 896 struct discard_info *front, unsigned int max_len) 897 { 898 return (back->lstart + back->len == front->lstart) && 899 (back->len + front->len <= max_len); 900 } 901 902 static inline bool __is_discard_back_mergeable(struct discard_info *cur, 903 struct discard_info *back, unsigned int max_len) 904 { 905 return __is_discard_mergeable(back, cur, max_len); 906 } 907 908 static inline bool __is_discard_front_mergeable(struct discard_info *cur, 909 struct discard_info *front, unsigned int max_len) 910 { 911 return __is_discard_mergeable(cur, front, max_len); 912 } 913 914 /* 915 * For free nid management 916 */ 917 enum nid_state { 918 FREE_NID, /* newly added to free nid list */ 919 PREALLOC_NID, /* it is preallocated */ 920 MAX_NID_STATE, 921 }; 922 923 enum nat_state { 924 TOTAL_NAT, 925 DIRTY_NAT, 926 RECLAIMABLE_NAT, 927 MAX_NAT_STATE, 928 }; 929 930 struct f2fs_nm_info { 931 block_t nat_blkaddr; /* base disk address of NAT */ 932 nid_t max_nid; /* maximum possible node ids */ 933 nid_t available_nids; /* # of available node ids */ 934 nid_t next_scan_nid; /* the next nid to be scanned */ 935 nid_t max_rf_node_blocks; /* max # of nodes for recovery */ 936 unsigned int ram_thresh; /* control the memory footprint */ 937 unsigned int ra_nid_pages; /* # of nid pages to be readaheaded */ 938 unsigned int dirty_nats_ratio; /* control dirty nats ratio threshold */ 939 940 /* NAT cache management */ 941 struct radix_tree_root nat_root;/* root of the nat entry cache */ 942 struct radix_tree_root nat_set_root;/* root of the nat set cache */ 943 struct f2fs_rwsem nat_tree_lock; /* protect nat entry tree */ 944 struct list_head nat_entries; /* cached nat entry list (clean) */ 945 spinlock_t nat_list_lock; /* protect clean nat entry list */ 946 unsigned int nat_cnt[MAX_NAT_STATE]; /* the # of cached nat entries */ 947 unsigned int nat_blocks; /* # of nat blocks */ 948 949 /* free node ids management */ 950 struct radix_tree_root free_nid_root;/* root of the free_nid cache */ 951 struct list_head free_nid_list; /* list for free nids excluding preallocated nids */ 952 unsigned int nid_cnt[MAX_NID_STATE]; /* the number of free node id */ 953 spinlock_t nid_list_lock; /* protect nid lists ops */ 954 struct mutex build_lock; /* lock for build free nids */ 955 unsigned char **free_nid_bitmap; 956 unsigned char *nat_block_bitmap; 957 unsigned short *free_nid_count; /* free nid count of NAT block */ 958 959 /* for checkpoint */ 960 char *nat_bitmap; /* NAT bitmap pointer */ 961 962 unsigned int nat_bits_blocks; /* # of nat bits blocks */ 963 unsigned char *nat_bits; /* NAT bits blocks */ 964 unsigned char *full_nat_bits; /* full NAT pages */ 965 unsigned char *empty_nat_bits; /* empty NAT pages */ 966 #ifdef CONFIG_F2FS_CHECK_FS 967 char *nat_bitmap_mir; /* NAT bitmap mirror */ 968 #endif 969 int bitmap_size; /* bitmap size */ 970 }; 971 972 /* 973 * this structure is used as one of function parameters. 974 * all the information are dedicated to a given direct node block determined 975 * by the data offset in a file. 976 */ 977 struct dnode_of_data { 978 struct inode *inode; /* vfs inode pointer */ 979 struct page *inode_page; /* its inode page, NULL is possible */ 980 struct page *node_page; /* cached direct node page */ 981 nid_t nid; /* node id of the direct node block */ 982 unsigned int ofs_in_node; /* data offset in the node page */ 983 bool inode_page_locked; /* inode page is locked or not */ 984 bool node_changed; /* is node block changed */ 985 char cur_level; /* level of hole node page */ 986 char max_level; /* level of current page located */ 987 block_t data_blkaddr; /* block address of the node block */ 988 }; 989 990 static inline void set_new_dnode(struct dnode_of_data *dn, struct inode *inode, 991 struct page *ipage, struct page *npage, nid_t nid) 992 { 993 memset(dn, 0, sizeof(*dn)); 994 dn->inode = inode; 995 dn->inode_page = ipage; 996 dn->node_page = npage; 997 dn->nid = nid; 998 } 999 1000 /* 1001 * For SIT manager 1002 * 1003 * By default, there are 6 active log areas across the whole main area. 1004 * When considering hot and cold data separation to reduce cleaning overhead, 1005 * we split 3 for data logs and 3 for node logs as hot, warm, and cold types, 1006 * respectively. 1007 * In the current design, you should not change the numbers intentionally. 1008 * Instead, as a mount option such as active_logs=x, you can use 2, 4, and 6 1009 * logs individually according to the underlying devices. (default: 6) 1010 * Just in case, on-disk layout covers maximum 16 logs that consist of 8 for 1011 * data and 8 for node logs. 1012 */ 1013 #define NR_CURSEG_DATA_TYPE (3) 1014 #define NR_CURSEG_NODE_TYPE (3) 1015 #define NR_CURSEG_INMEM_TYPE (2) 1016 #define NR_CURSEG_RO_TYPE (2) 1017 #define NR_CURSEG_PERSIST_TYPE (NR_CURSEG_DATA_TYPE + NR_CURSEG_NODE_TYPE) 1018 #define NR_CURSEG_TYPE (NR_CURSEG_INMEM_TYPE + NR_CURSEG_PERSIST_TYPE) 1019 1020 enum { 1021 CURSEG_HOT_DATA = 0, /* directory entry blocks */ 1022 CURSEG_WARM_DATA, /* data blocks */ 1023 CURSEG_COLD_DATA, /* multimedia or GCed data blocks */ 1024 CURSEG_HOT_NODE, /* direct node blocks of directory files */ 1025 CURSEG_WARM_NODE, /* direct node blocks of normal files */ 1026 CURSEG_COLD_NODE, /* indirect node blocks */ 1027 NR_PERSISTENT_LOG, /* number of persistent log */ 1028 CURSEG_COLD_DATA_PINNED = NR_PERSISTENT_LOG, 1029 /* pinned file that needs consecutive block address */ 1030 CURSEG_ALL_DATA_ATGC, /* SSR alloctor in hot/warm/cold data area */ 1031 NO_CHECK_TYPE, /* number of persistent & inmem log */ 1032 }; 1033 1034 struct flush_cmd { 1035 struct completion wait; 1036 struct llist_node llnode; 1037 nid_t ino; 1038 int ret; 1039 }; 1040 1041 struct flush_cmd_control { 1042 struct task_struct *f2fs_issue_flush; /* flush thread */ 1043 wait_queue_head_t flush_wait_queue; /* waiting queue for wake-up */ 1044 atomic_t issued_flush; /* # of issued flushes */ 1045 atomic_t queued_flush; /* # of queued flushes */ 1046 struct llist_head issue_list; /* list for command issue */ 1047 struct llist_node *dispatch_list; /* list for command dispatch */ 1048 }; 1049 1050 struct f2fs_sm_info { 1051 struct sit_info *sit_info; /* whole segment information */ 1052 struct free_segmap_info *free_info; /* free segment information */ 1053 struct dirty_seglist_info *dirty_info; /* dirty segment information */ 1054 struct curseg_info *curseg_array; /* active segment information */ 1055 1056 struct f2fs_rwsem curseg_lock; /* for preventing curseg change */ 1057 1058 block_t seg0_blkaddr; /* block address of 0'th segment */ 1059 block_t main_blkaddr; /* start block address of main area */ 1060 block_t ssa_blkaddr; /* start block address of SSA area */ 1061 1062 unsigned int segment_count; /* total # of segments */ 1063 unsigned int main_segments; /* # of segments in main area */ 1064 unsigned int reserved_segments; /* # of reserved segments */ 1065 unsigned int additional_reserved_segments;/* reserved segs for IO align feature */ 1066 unsigned int ovp_segments; /* # of overprovision segments */ 1067 1068 /* a threshold to reclaim prefree segments */ 1069 unsigned int rec_prefree_segments; 1070 1071 struct list_head sit_entry_set; /* sit entry set list */ 1072 1073 unsigned int ipu_policy; /* in-place-update policy */ 1074 unsigned int min_ipu_util; /* in-place-update threshold */ 1075 unsigned int min_fsync_blocks; /* threshold for fsync */ 1076 unsigned int min_seq_blocks; /* threshold for sequential blocks */ 1077 unsigned int min_hot_blocks; /* threshold for hot block allocation */ 1078 unsigned int min_ssr_sections; /* threshold to trigger SSR allocation */ 1079 1080 /* for flush command control */ 1081 struct flush_cmd_control *fcc_info; 1082 1083 /* for discard command control */ 1084 struct discard_cmd_control *dcc_info; 1085 }; 1086 1087 /* 1088 * For superblock 1089 */ 1090 /* 1091 * COUNT_TYPE for monitoring 1092 * 1093 * f2fs monitors the number of several block types such as on-writeback, 1094 * dirty dentry blocks, dirty node blocks, and dirty meta blocks. 1095 */ 1096 #define WB_DATA_TYPE(p) (__is_cp_guaranteed(p) ? F2FS_WB_CP_DATA : F2FS_WB_DATA) 1097 enum count_type { 1098 F2FS_DIRTY_DENTS, 1099 F2FS_DIRTY_DATA, 1100 F2FS_DIRTY_QDATA, 1101 F2FS_DIRTY_NODES, 1102 F2FS_DIRTY_META, 1103 F2FS_DIRTY_IMETA, 1104 F2FS_WB_CP_DATA, 1105 F2FS_WB_DATA, 1106 F2FS_RD_DATA, 1107 F2FS_RD_NODE, 1108 F2FS_RD_META, 1109 F2FS_DIO_WRITE, 1110 F2FS_DIO_READ, 1111 NR_COUNT_TYPE, 1112 }; 1113 1114 /* 1115 * The below are the page types of bios used in submit_bio(). 1116 * The available types are: 1117 * DATA User data pages. It operates as async mode. 1118 * NODE Node pages. It operates as async mode. 1119 * META FS metadata pages such as SIT, NAT, CP. 1120 * NR_PAGE_TYPE The number of page types. 1121 * META_FLUSH Make sure the previous pages are written 1122 * with waiting the bio's completion 1123 * ... Only can be used with META. 1124 */ 1125 #define PAGE_TYPE_OF_BIO(type) ((type) > META ? META : (type)) 1126 enum page_type { 1127 DATA = 0, 1128 NODE = 1, /* should not change this */ 1129 META, 1130 NR_PAGE_TYPE, 1131 META_FLUSH, 1132 IPU, /* the below types are used by tracepoints only. */ 1133 OPU, 1134 }; 1135 1136 enum temp_type { 1137 HOT = 0, /* must be zero for meta bio */ 1138 WARM, 1139 COLD, 1140 NR_TEMP_TYPE, 1141 }; 1142 1143 enum need_lock_type { 1144 LOCK_REQ = 0, 1145 LOCK_DONE, 1146 LOCK_RETRY, 1147 }; 1148 1149 enum cp_reason_type { 1150 CP_NO_NEEDED, 1151 CP_NON_REGULAR, 1152 CP_COMPRESSED, 1153 CP_HARDLINK, 1154 CP_SB_NEED_CP, 1155 CP_WRONG_PINO, 1156 CP_NO_SPC_ROLL, 1157 CP_NODE_NEED_CP, 1158 CP_FASTBOOT_MODE, 1159 CP_SPEC_LOG_NUM, 1160 CP_RECOVER_DIR, 1161 }; 1162 1163 enum iostat_type { 1164 /* WRITE IO */ 1165 APP_DIRECT_IO, /* app direct write IOs */ 1166 APP_BUFFERED_IO, /* app buffered write IOs */ 1167 APP_WRITE_IO, /* app write IOs */ 1168 APP_MAPPED_IO, /* app mapped IOs */ 1169 APP_BUFFERED_CDATA_IO, /* app buffered write IOs on compressed file */ 1170 APP_MAPPED_CDATA_IO, /* app mapped write IOs on compressed file */ 1171 FS_DATA_IO, /* data IOs from kworker/fsync/reclaimer */ 1172 FS_CDATA_IO, /* data IOs from kworker/fsync/reclaimer on compressed file */ 1173 FS_NODE_IO, /* node IOs from kworker/fsync/reclaimer */ 1174 FS_META_IO, /* meta IOs from kworker/reclaimer */ 1175 FS_GC_DATA_IO, /* data IOs from forground gc */ 1176 FS_GC_NODE_IO, /* node IOs from forground gc */ 1177 FS_CP_DATA_IO, /* data IOs from checkpoint */ 1178 FS_CP_NODE_IO, /* node IOs from checkpoint */ 1179 FS_CP_META_IO, /* meta IOs from checkpoint */ 1180 1181 /* READ IO */ 1182 APP_DIRECT_READ_IO, /* app direct read IOs */ 1183 APP_BUFFERED_READ_IO, /* app buffered read IOs */ 1184 APP_READ_IO, /* app read IOs */ 1185 APP_MAPPED_READ_IO, /* app mapped read IOs */ 1186 APP_BUFFERED_CDATA_READ_IO, /* app buffered read IOs on compressed file */ 1187 APP_MAPPED_CDATA_READ_IO, /* app mapped read IOs on compressed file */ 1188 FS_DATA_READ_IO, /* data read IOs */ 1189 FS_GDATA_READ_IO, /* data read IOs from background gc */ 1190 FS_CDATA_READ_IO, /* compressed data read IOs */ 1191 FS_NODE_READ_IO, /* node read IOs */ 1192 FS_META_READ_IO, /* meta read IOs */ 1193 1194 /* other */ 1195 FS_DISCARD_IO, /* discard */ 1196 FS_FLUSH_IO, /* flush */ 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 unsigned int compr_blocks; /* # of compressed block addresses */ 1214 unsigned int need_lock:8; /* indicate we need to lock cp_rwsem */ 1215 unsigned int version:8; /* version of the node */ 1216 unsigned int submitted:1; /* indicate IO submission */ 1217 unsigned int in_list:1; /* indicate fio is in io_list */ 1218 unsigned int is_por:1; /* indicate IO is from recovery or not */ 1219 unsigned int retry:1; /* need to reallocate block address */ 1220 unsigned int encrypted:1; /* indicate file is encrypted */ 1221 unsigned int post_read:1; /* require post read */ 1222 enum iostat_type io_type; /* io type */ 1223 struct writeback_control *io_wbc; /* writeback control */ 1224 struct bio **bio; /* bio for ipu */ 1225 sector_t *last_block; /* last block number in bio */ 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 static inline int f2fs_test_bit(unsigned int nr, char *addr); 1388 static inline void f2fs_set_bit(unsigned int nr, char *addr); 1389 static inline void f2fs_clear_bit(unsigned int nr, char *addr); 1390 1391 /* 1392 * Layout of f2fs page.private: 1393 * 1394 * Layout A: lowest bit should be 1 1395 * | bit0 = 1 | bit1 | bit2 | ... | bit MAX | private data .... | 1396 * bit 0 PAGE_PRIVATE_NOT_POINTER 1397 * bit 1 PAGE_PRIVATE_DUMMY_WRITE 1398 * bit 2 PAGE_PRIVATE_ONGOING_MIGRATION 1399 * bit 3 PAGE_PRIVATE_INLINE_INODE 1400 * bit 4 PAGE_PRIVATE_REF_RESOURCE 1401 * bit 5- f2fs private data 1402 * 1403 * Layout B: lowest bit should be 0 1404 * page.private is a wrapped pointer. 1405 */ 1406 enum { 1407 PAGE_PRIVATE_NOT_POINTER, /* private contains non-pointer data */ 1408 PAGE_PRIVATE_DUMMY_WRITE, /* data page for padding aligned IO */ 1409 PAGE_PRIVATE_ONGOING_MIGRATION, /* data page which is on-going migrating */ 1410 PAGE_PRIVATE_INLINE_INODE, /* inode page contains inline data */ 1411 PAGE_PRIVATE_REF_RESOURCE, /* dirty page has referenced resources */ 1412 PAGE_PRIVATE_MAX 1413 }; 1414 1415 #define PAGE_PRIVATE_GET_FUNC(name, flagname) \ 1416 static inline bool page_private_##name(struct page *page) \ 1417 { \ 1418 return PagePrivate(page) && \ 1419 test_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)) && \ 1420 test_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \ 1421 } 1422 1423 #define PAGE_PRIVATE_SET_FUNC(name, flagname) \ 1424 static inline void set_page_private_##name(struct page *page) \ 1425 { \ 1426 if (!PagePrivate(page)) { \ 1427 get_page(page); \ 1428 SetPagePrivate(page); \ 1429 set_page_private(page, 0); \ 1430 } \ 1431 set_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)); \ 1432 set_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \ 1433 } 1434 1435 #define PAGE_PRIVATE_CLEAR_FUNC(name, flagname) \ 1436 static inline void clear_page_private_##name(struct page *page) \ 1437 { \ 1438 clear_bit(PAGE_PRIVATE_##flagname, &page_private(page)); \ 1439 if (page_private(page) == 1 << PAGE_PRIVATE_NOT_POINTER) { \ 1440 set_page_private(page, 0); \ 1441 if (PagePrivate(page)) { \ 1442 ClearPagePrivate(page); \ 1443 put_page(page); \ 1444 }\ 1445 } \ 1446 } 1447 1448 PAGE_PRIVATE_GET_FUNC(nonpointer, NOT_POINTER); 1449 PAGE_PRIVATE_GET_FUNC(inline, INLINE_INODE); 1450 PAGE_PRIVATE_GET_FUNC(gcing, ONGOING_MIGRATION); 1451 PAGE_PRIVATE_GET_FUNC(dummy, DUMMY_WRITE); 1452 1453 PAGE_PRIVATE_SET_FUNC(reference, REF_RESOURCE); 1454 PAGE_PRIVATE_SET_FUNC(inline, INLINE_INODE); 1455 PAGE_PRIVATE_SET_FUNC(gcing, ONGOING_MIGRATION); 1456 PAGE_PRIVATE_SET_FUNC(dummy, DUMMY_WRITE); 1457 1458 PAGE_PRIVATE_CLEAR_FUNC(reference, REF_RESOURCE); 1459 PAGE_PRIVATE_CLEAR_FUNC(inline, INLINE_INODE); 1460 PAGE_PRIVATE_CLEAR_FUNC(gcing, ONGOING_MIGRATION); 1461 PAGE_PRIVATE_CLEAR_FUNC(dummy, DUMMY_WRITE); 1462 1463 static inline unsigned long get_page_private_data(struct page *page) 1464 { 1465 unsigned long data = page_private(page); 1466 1467 if (!test_bit(PAGE_PRIVATE_NOT_POINTER, &data)) 1468 return 0; 1469 return data >> PAGE_PRIVATE_MAX; 1470 } 1471 1472 static inline void set_page_private_data(struct page *page, unsigned long data) 1473 { 1474 if (!PagePrivate(page)) { 1475 get_page(page); 1476 SetPagePrivate(page); 1477 set_page_private(page, 0); 1478 } 1479 set_bit(PAGE_PRIVATE_NOT_POINTER, &page_private(page)); 1480 page_private(page) |= data << PAGE_PRIVATE_MAX; 1481 } 1482 1483 static inline void clear_page_private_data(struct page *page) 1484 { 1485 page_private(page) &= (1 << PAGE_PRIVATE_MAX) - 1; 1486 if (page_private(page) == 1 << PAGE_PRIVATE_NOT_POINTER) { 1487 set_page_private(page, 0); 1488 if (PagePrivate(page)) { 1489 ClearPagePrivate(page); 1490 put_page(page); 1491 } 1492 } 1493 } 1494 1495 /* For compression */ 1496 enum compress_algorithm_type { 1497 COMPRESS_LZO, 1498 COMPRESS_LZ4, 1499 COMPRESS_ZSTD, 1500 COMPRESS_LZORLE, 1501 COMPRESS_MAX, 1502 }; 1503 1504 enum compress_flag { 1505 COMPRESS_CHKSUM, 1506 COMPRESS_MAX_FLAG, 1507 }; 1508 1509 #define COMPRESS_WATERMARK 20 1510 #define COMPRESS_PERCENT 20 1511 1512 #define COMPRESS_DATA_RESERVED_SIZE 4 1513 struct compress_data { 1514 __le32 clen; /* compressed data size */ 1515 __le32 chksum; /* compressed data chksum */ 1516 __le32 reserved[COMPRESS_DATA_RESERVED_SIZE]; /* reserved */ 1517 u8 cdata[]; /* compressed data */ 1518 }; 1519 1520 #define COMPRESS_HEADER_SIZE (sizeof(struct compress_data)) 1521 1522 #define F2FS_COMPRESSED_PAGE_MAGIC 0xF5F2C000 1523 1524 #define COMPRESS_LEVEL_OFFSET 8 1525 1526 /* compress context */ 1527 struct compress_ctx { 1528 struct inode *inode; /* inode the context belong to */ 1529 pgoff_t cluster_idx; /* cluster index number */ 1530 unsigned int cluster_size; /* page count in cluster */ 1531 unsigned int log_cluster_size; /* log of cluster size */ 1532 struct page **rpages; /* pages store raw data in cluster */ 1533 unsigned int nr_rpages; /* total page number in rpages */ 1534 struct page **cpages; /* pages store compressed data in cluster */ 1535 unsigned int nr_cpages; /* total page number in cpages */ 1536 unsigned int valid_nr_cpages; /* valid page number in cpages */ 1537 void *rbuf; /* virtual mapped address on rpages */ 1538 struct compress_data *cbuf; /* virtual mapped address on cpages */ 1539 size_t rlen; /* valid data length in rbuf */ 1540 size_t clen; /* valid data length in cbuf */ 1541 void *private; /* payload buffer for specified compression algorithm */ 1542 void *private2; /* extra payload buffer */ 1543 }; 1544 1545 /* compress context for write IO path */ 1546 struct compress_io_ctx { 1547 u32 magic; /* magic number to indicate page is compressed */ 1548 struct inode *inode; /* inode the context belong to */ 1549 struct page **rpages; /* pages store raw data in cluster */ 1550 unsigned int nr_rpages; /* total page number in rpages */ 1551 atomic_t pending_pages; /* in-flight compressed page count */ 1552 }; 1553 1554 /* Context for decompressing one cluster on the read IO path */ 1555 struct decompress_io_ctx { 1556 u32 magic; /* magic number to indicate page is compressed */ 1557 struct inode *inode; /* inode the context belong to */ 1558 pgoff_t cluster_idx; /* cluster index number */ 1559 unsigned int cluster_size; /* page count in cluster */ 1560 unsigned int log_cluster_size; /* log of cluster size */ 1561 struct page **rpages; /* pages store raw data in cluster */ 1562 unsigned int nr_rpages; /* total page number in rpages */ 1563 struct page **cpages; /* pages store compressed data in cluster */ 1564 unsigned int nr_cpages; /* total page number in cpages */ 1565 struct page **tpages; /* temp pages to pad holes in cluster */ 1566 void *rbuf; /* virtual mapped address on rpages */ 1567 struct compress_data *cbuf; /* virtual mapped address on cpages */ 1568 size_t rlen; /* valid data length in rbuf */ 1569 size_t clen; /* valid data length in cbuf */ 1570 1571 /* 1572 * The number of compressed pages remaining to be read in this cluster. 1573 * This is initially nr_cpages. It is decremented by 1 each time a page 1574 * has been read (or failed to be read). When it reaches 0, the cluster 1575 * is decompressed (or an error is reported). 1576 * 1577 * If an error occurs before all the pages have been submitted for I/O, 1578 * then this will never reach 0. In this case the I/O submitter is 1579 * responsible for calling f2fs_decompress_end_io() instead. 1580 */ 1581 atomic_t remaining_pages; 1582 1583 /* 1584 * Number of references to this decompress_io_ctx. 1585 * 1586 * One reference is held for I/O completion. This reference is dropped 1587 * after the pagecache pages are updated and unlocked -- either after 1588 * decompression (and verity if enabled), or after an error. 1589 * 1590 * In addition, each compressed page holds a reference while it is in a 1591 * bio. These references are necessary prevent compressed pages from 1592 * being freed while they are still in a bio. 1593 */ 1594 refcount_t refcnt; 1595 1596 bool failed; /* IO error occurred before decompression? */ 1597 bool need_verity; /* need fs-verity verification after decompression? */ 1598 void *private; /* payload buffer for specified decompression algorithm */ 1599 void *private2; /* extra payload buffer */ 1600 struct work_struct verity_work; /* work to verify the decompressed pages */ 1601 struct work_struct free_work; /* work for late free this structure itself */ 1602 }; 1603 1604 #define NULL_CLUSTER ((unsigned int)(~0)) 1605 #define MIN_COMPRESS_LOG_SIZE 2 1606 #define MAX_COMPRESS_LOG_SIZE 8 1607 #define MAX_COMPRESS_WINDOW_SIZE(log_size) ((PAGE_SIZE) << (log_size)) 1608 1609 struct f2fs_sb_info { 1610 struct super_block *sb; /* pointer to VFS super block */ 1611 struct proc_dir_entry *s_proc; /* proc entry */ 1612 struct f2fs_super_block *raw_super; /* raw super block pointer */ 1613 struct f2fs_rwsem sb_lock; /* lock for raw super block */ 1614 int valid_super_block; /* valid super block no */ 1615 unsigned long s_flag; /* flags for sbi */ 1616 struct mutex writepages; /* mutex for writepages() */ 1617 1618 #ifdef CONFIG_BLK_DEV_ZONED 1619 unsigned int blocks_per_blkz; /* F2FS blocks per zone */ 1620 unsigned int log_blocks_per_blkz; /* log2 F2FS blocks per zone */ 1621 #endif 1622 1623 /* for node-related operations */ 1624 struct f2fs_nm_info *nm_info; /* node manager */ 1625 struct inode *node_inode; /* cache node blocks */ 1626 1627 /* for segment-related operations */ 1628 struct f2fs_sm_info *sm_info; /* segment manager */ 1629 1630 /* for bio operations */ 1631 struct f2fs_bio_info *write_io[NR_PAGE_TYPE]; /* for write bios */ 1632 /* keep migration IO order for LFS mode */ 1633 struct f2fs_rwsem io_order_lock; 1634 mempool_t *write_io_dummy; /* Dummy pages */ 1635 pgoff_t page_eio_ofs[NR_PAGE_TYPE]; /* EIO page offset */ 1636 int page_eio_cnt[NR_PAGE_TYPE]; /* EIO count */ 1637 1638 /* for checkpoint */ 1639 struct f2fs_checkpoint *ckpt; /* raw checkpoint pointer */ 1640 int cur_cp_pack; /* remain current cp pack */ 1641 spinlock_t cp_lock; /* for flag in ckpt */ 1642 struct inode *meta_inode; /* cache meta blocks */ 1643 struct f2fs_rwsem cp_global_sem; /* checkpoint procedure lock */ 1644 struct f2fs_rwsem cp_rwsem; /* blocking FS operations */ 1645 struct f2fs_rwsem node_write; /* locking node writes */ 1646 struct f2fs_rwsem node_change; /* locking node change */ 1647 wait_queue_head_t cp_wait; 1648 unsigned long last_time[MAX_TIME]; /* to store time in jiffies */ 1649 long interval_time[MAX_TIME]; /* to store thresholds */ 1650 struct ckpt_req_control cprc_info; /* for checkpoint request control */ 1651 1652 struct inode_management im[MAX_INO_ENTRY]; /* manage inode cache */ 1653 1654 spinlock_t fsync_node_lock; /* for node entry lock */ 1655 struct list_head fsync_node_list; /* node list head */ 1656 unsigned int fsync_seg_id; /* sequence id */ 1657 unsigned int fsync_node_num; /* number of node entries */ 1658 1659 /* for orphan inode, use 0'th array */ 1660 unsigned int max_orphans; /* max orphan inodes */ 1661 1662 /* for inode management */ 1663 struct list_head inode_list[NR_INODE_TYPE]; /* dirty inode list */ 1664 spinlock_t inode_lock[NR_INODE_TYPE]; /* for dirty inode list lock */ 1665 struct mutex flush_lock; /* for flush exclusion */ 1666 1667 /* for extent tree cache */ 1668 struct extent_tree_info extent_tree[NR_EXTENT_CACHES]; 1669 atomic64_t allocated_data_blocks; /* for block age extent_cache */ 1670 1671 /* The threshold used for hot and warm data seperation*/ 1672 unsigned int hot_data_age_threshold; 1673 unsigned int warm_data_age_threshold; 1674 unsigned int last_age_weight; 1675 1676 /* basic filesystem units */ 1677 unsigned int log_sectors_per_block; /* log2 sectors per block */ 1678 unsigned int log_blocksize; /* log2 block size */ 1679 unsigned int blocksize; /* block size */ 1680 unsigned int root_ino_num; /* root inode number*/ 1681 unsigned int node_ino_num; /* node inode number*/ 1682 unsigned int meta_ino_num; /* meta inode number*/ 1683 unsigned int log_blocks_per_seg; /* log2 blocks per segment */ 1684 unsigned int blocks_per_seg; /* blocks per segment */ 1685 unsigned int unusable_blocks_per_sec; /* unusable blocks per section */ 1686 unsigned int segs_per_sec; /* segments per section */ 1687 unsigned int secs_per_zone; /* sections per zone */ 1688 unsigned int total_sections; /* total section count */ 1689 unsigned int total_node_count; /* total node block count */ 1690 unsigned int total_valid_node_count; /* valid node block count */ 1691 int dir_level; /* directory level */ 1692 bool readdir_ra; /* readahead inode in readdir */ 1693 u64 max_io_bytes; /* max io bytes to merge IOs */ 1694 1695 block_t user_block_count; /* # of user blocks */ 1696 block_t total_valid_block_count; /* # of valid blocks */ 1697 block_t discard_blks; /* discard command candidats */ 1698 block_t last_valid_block_count; /* for recovery */ 1699 block_t reserved_blocks; /* configurable reserved blocks */ 1700 block_t current_reserved_blocks; /* current reserved blocks */ 1701 1702 /* Additional tracking for no checkpoint mode */ 1703 block_t unusable_block_count; /* # of blocks saved by last cp */ 1704 1705 unsigned int nquota_files; /* # of quota sysfile */ 1706 struct f2fs_rwsem quota_sem; /* blocking cp for flags */ 1707 1708 /* # of pages, see count_type */ 1709 atomic_t nr_pages[NR_COUNT_TYPE]; 1710 /* # of allocated blocks */ 1711 struct percpu_counter alloc_valid_block_count; 1712 /* # of node block writes as roll forward recovery */ 1713 struct percpu_counter rf_node_block_count; 1714 1715 /* writeback control */ 1716 atomic_t wb_sync_req[META]; /* count # of WB_SYNC threads */ 1717 1718 /* valid inode count */ 1719 struct percpu_counter total_valid_inode_count; 1720 1721 struct f2fs_mount_info mount_opt; /* mount options */ 1722 1723 /* for cleaning operations */ 1724 struct f2fs_rwsem gc_lock; /* 1725 * semaphore for GC, avoid 1726 * race between GC and GC or CP 1727 */ 1728 struct f2fs_gc_kthread *gc_thread; /* GC thread */ 1729 struct atgc_management am; /* atgc management */ 1730 unsigned int cur_victim_sec; /* current victim section num */ 1731 unsigned int gc_mode; /* current GC state */ 1732 unsigned int next_victim_seg[2]; /* next segment in victim section */ 1733 spinlock_t gc_remaining_trials_lock; 1734 /* remaining trial count for GC_URGENT_* and GC_IDLE_* */ 1735 unsigned int gc_remaining_trials; 1736 1737 /* for skip statistic */ 1738 unsigned long long skipped_gc_rwsem; /* FG_GC only */ 1739 1740 /* threshold for gc trials on pinned files */ 1741 u64 gc_pin_file_threshold; 1742 struct f2fs_rwsem pin_sem; 1743 1744 /* maximum # of trials to find a victim segment for SSR and GC */ 1745 unsigned int max_victim_search; 1746 /* migration granularity of garbage collection, unit: segment */ 1747 unsigned int migration_granularity; 1748 1749 /* 1750 * for stat information. 1751 * one is for the LFS mode, and the other is for the SSR mode. 1752 */ 1753 #ifdef CONFIG_F2FS_STAT_FS 1754 struct f2fs_stat_info *stat_info; /* FS status information */ 1755 atomic_t meta_count[META_MAX]; /* # of meta blocks */ 1756 unsigned int segment_count[2]; /* # of allocated segments */ 1757 unsigned int block_count[2]; /* # of allocated blocks */ 1758 atomic_t inplace_count; /* # of inplace update */ 1759 /* # of lookup extent cache */ 1760 atomic64_t total_hit_ext[NR_EXTENT_CACHES]; 1761 /* # of hit rbtree extent node */ 1762 atomic64_t read_hit_rbtree[NR_EXTENT_CACHES]; 1763 /* # of hit cached extent node */ 1764 atomic64_t read_hit_cached[NR_EXTENT_CACHES]; 1765 /* # of hit largest extent node in read extent cache */ 1766 atomic64_t read_hit_largest; 1767 atomic_t inline_xattr; /* # of inline_xattr inodes */ 1768 atomic_t inline_inode; /* # of inline_data inodes */ 1769 atomic_t inline_dir; /* # of inline_dentry inodes */ 1770 atomic_t compr_inode; /* # of compressed inodes */ 1771 atomic64_t compr_blocks; /* # of compressed blocks */ 1772 atomic_t swapfile_inode; /* # of swapfile inodes */ 1773 atomic_t atomic_files; /* # of opened atomic file */ 1774 atomic_t max_aw_cnt; /* max # of atomic writes */ 1775 unsigned int io_skip_bggc; /* skip background gc for in-flight IO */ 1776 unsigned int other_skip_bggc; /* skip background gc for other reasons */ 1777 unsigned int ndirty_inode[NR_INODE_TYPE]; /* # of dirty inodes */ 1778 #endif 1779 spinlock_t stat_lock; /* lock for stat operations */ 1780 1781 /* to attach REQ_META|REQ_FUA flags */ 1782 unsigned int data_io_flag; 1783 unsigned int node_io_flag; 1784 1785 /* For sysfs support */ 1786 struct kobject s_kobj; /* /sys/fs/f2fs/<devname> */ 1787 struct completion s_kobj_unregister; 1788 1789 struct kobject s_stat_kobj; /* /sys/fs/f2fs/<devname>/stat */ 1790 struct completion s_stat_kobj_unregister; 1791 1792 struct kobject s_feature_list_kobj; /* /sys/fs/f2fs/<devname>/feature_list */ 1793 struct completion s_feature_list_kobj_unregister; 1794 1795 /* For shrinker support */ 1796 struct list_head s_list; 1797 struct mutex umount_mutex; 1798 unsigned int shrinker_run_no; 1799 1800 /* For multi devices */ 1801 int s_ndevs; /* number of devices */ 1802 struct f2fs_dev_info *devs; /* for device list */ 1803 unsigned int dirty_device; /* for checkpoint data flush */ 1804 spinlock_t dev_lock; /* protect dirty_device */ 1805 bool aligned_blksize; /* all devices has the same logical blksize */ 1806 1807 /* For write statistics */ 1808 u64 sectors_written_start; 1809 u64 kbytes_written; 1810 1811 /* Reference to checksum algorithm driver via cryptoapi */ 1812 struct crypto_shash *s_chksum_driver; 1813 1814 /* Precomputed FS UUID checksum for seeding other checksums */ 1815 __u32 s_chksum_seed; 1816 1817 struct workqueue_struct *post_read_wq; /* post read workqueue */ 1818 1819 unsigned char errors[MAX_F2FS_ERRORS]; /* error flags */ 1820 spinlock_t error_lock; /* protect errors array */ 1821 bool error_dirty; /* errors of sb is dirty */ 1822 1823 struct kmem_cache *inline_xattr_slab; /* inline xattr entry */ 1824 unsigned int inline_xattr_slab_size; /* default inline xattr slab size */ 1825 1826 /* For reclaimed segs statistics per each GC mode */ 1827 unsigned int gc_segment_mode; /* GC state for reclaimed segments */ 1828 unsigned int gc_reclaimed_segs[MAX_GC_MODE]; /* Reclaimed segs for each mode */ 1829 1830 unsigned long seq_file_ra_mul; /* multiplier for ra_pages of seq. files in fadvise */ 1831 1832 int max_fragment_chunk; /* max chunk size for block fragmentation mode */ 1833 int max_fragment_hole; /* max hole size for block fragmentation mode */ 1834 1835 /* For atomic write statistics */ 1836 atomic64_t current_atomic_write; 1837 s64 peak_atomic_write; 1838 u64 committed_atomic_block; 1839 u64 revoked_atomic_block; 1840 1841 #ifdef CONFIG_F2FS_FS_COMPRESSION 1842 struct kmem_cache *page_array_slab; /* page array entry */ 1843 unsigned int page_array_slab_size; /* default page array slab size */ 1844 1845 /* For runtime compression statistics */ 1846 u64 compr_written_block; 1847 u64 compr_saved_block; 1848 u32 compr_new_inode; 1849 1850 /* For compressed block cache */ 1851 struct inode *compress_inode; /* cache compressed blocks */ 1852 unsigned int compress_percent; /* cache page percentage */ 1853 unsigned int compress_watermark; /* cache page watermark */ 1854 atomic_t compress_page_hit; /* cache hit count */ 1855 #endif 1856 1857 #ifdef CONFIG_F2FS_IOSTAT 1858 /* For app/fs IO statistics */ 1859 spinlock_t iostat_lock; 1860 unsigned long long iostat_count[NR_IO_TYPE]; 1861 unsigned long long iostat_bytes[NR_IO_TYPE]; 1862 unsigned long long prev_iostat_bytes[NR_IO_TYPE]; 1863 bool iostat_enable; 1864 unsigned long iostat_next_period; 1865 unsigned int iostat_period_ms; 1866 1867 /* For io latency related statistics info in one iostat period */ 1868 spinlock_t iostat_lat_lock; 1869 struct iostat_lat_info *iostat_io_lat; 1870 #endif 1871 }; 1872 1873 #ifdef CONFIG_F2FS_FAULT_INJECTION 1874 #define time_to_inject(sbi, type) __time_to_inject(sbi, type, __func__, \ 1875 __builtin_return_address(0)) 1876 static inline bool __time_to_inject(struct f2fs_sb_info *sbi, int type, 1877 const char *func, const char *parent_func) 1878 { 1879 struct f2fs_fault_info *ffi = &F2FS_OPTION(sbi).fault_info; 1880 1881 if (!ffi->inject_rate) 1882 return false; 1883 1884 if (!IS_FAULT_SET(ffi, type)) 1885 return false; 1886 1887 atomic_inc(&ffi->inject_ops); 1888 if (atomic_read(&ffi->inject_ops) >= ffi->inject_rate) { 1889 atomic_set(&ffi->inject_ops, 0); 1890 printk_ratelimited("%sF2FS-fs (%s) : inject %s in %s of %pS\n", 1891 KERN_INFO, sbi->sb->s_id, f2fs_fault_name[type], 1892 func, parent_func); 1893 return true; 1894 } 1895 return false; 1896 } 1897 #else 1898 static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type) 1899 { 1900 return false; 1901 } 1902 #endif 1903 1904 /* 1905 * Test if the mounted volume is a multi-device volume. 1906 * - For a single regular disk volume, sbi->s_ndevs is 0. 1907 * - For a single zoned disk volume, sbi->s_ndevs is 1. 1908 * - For a multi-device volume, sbi->s_ndevs is always 2 or more. 1909 */ 1910 static inline bool f2fs_is_multi_device(struct f2fs_sb_info *sbi) 1911 { 1912 return sbi->s_ndevs > 1; 1913 } 1914 1915 static inline void f2fs_update_time(struct f2fs_sb_info *sbi, int type) 1916 { 1917 unsigned long now = jiffies; 1918 1919 sbi->last_time[type] = now; 1920 1921 /* DISCARD_TIME and GC_TIME are based on REQ_TIME */ 1922 if (type == REQ_TIME) { 1923 sbi->last_time[DISCARD_TIME] = now; 1924 sbi->last_time[GC_TIME] = now; 1925 } 1926 } 1927 1928 static inline bool f2fs_time_over(struct f2fs_sb_info *sbi, int type) 1929 { 1930 unsigned long interval = sbi->interval_time[type] * HZ; 1931 1932 return time_after(jiffies, sbi->last_time[type] + interval); 1933 } 1934 1935 static inline unsigned int f2fs_time_to_wait(struct f2fs_sb_info *sbi, 1936 int type) 1937 { 1938 unsigned long interval = sbi->interval_time[type] * HZ; 1939 unsigned int wait_ms = 0; 1940 long delta; 1941 1942 delta = (sbi->last_time[type] + interval) - jiffies; 1943 if (delta > 0) 1944 wait_ms = jiffies_to_msecs(delta); 1945 1946 return wait_ms; 1947 } 1948 1949 /* 1950 * Inline functions 1951 */ 1952 static inline u32 __f2fs_crc32(struct f2fs_sb_info *sbi, u32 crc, 1953 const void *address, unsigned int length) 1954 { 1955 struct { 1956 struct shash_desc shash; 1957 char ctx[4]; 1958 } desc; 1959 int err; 1960 1961 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver) != sizeof(desc.ctx)); 1962 1963 desc.shash.tfm = sbi->s_chksum_driver; 1964 *(u32 *)desc.ctx = crc; 1965 1966 err = crypto_shash_update(&desc.shash, address, length); 1967 BUG_ON(err); 1968 1969 return *(u32 *)desc.ctx; 1970 } 1971 1972 static inline u32 f2fs_crc32(struct f2fs_sb_info *sbi, const void *address, 1973 unsigned int length) 1974 { 1975 return __f2fs_crc32(sbi, F2FS_SUPER_MAGIC, address, length); 1976 } 1977 1978 static inline bool f2fs_crc_valid(struct f2fs_sb_info *sbi, __u32 blk_crc, 1979 void *buf, size_t buf_size) 1980 { 1981 return f2fs_crc32(sbi, buf, buf_size) == blk_crc; 1982 } 1983 1984 static inline u32 f2fs_chksum(struct f2fs_sb_info *sbi, u32 crc, 1985 const void *address, unsigned int length) 1986 { 1987 return __f2fs_crc32(sbi, crc, address, length); 1988 } 1989 1990 static inline struct f2fs_inode_info *F2FS_I(struct inode *inode) 1991 { 1992 return container_of(inode, struct f2fs_inode_info, vfs_inode); 1993 } 1994 1995 static inline struct f2fs_sb_info *F2FS_SB(struct super_block *sb) 1996 { 1997 return sb->s_fs_info; 1998 } 1999 2000 static inline struct f2fs_sb_info *F2FS_I_SB(struct inode *inode) 2001 { 2002 return F2FS_SB(inode->i_sb); 2003 } 2004 2005 static inline struct f2fs_sb_info *F2FS_M_SB(struct address_space *mapping) 2006 { 2007 return F2FS_I_SB(mapping->host); 2008 } 2009 2010 static inline struct f2fs_sb_info *F2FS_P_SB(struct page *page) 2011 { 2012 return F2FS_M_SB(page_file_mapping(page)); 2013 } 2014 2015 static inline struct f2fs_super_block *F2FS_RAW_SUPER(struct f2fs_sb_info *sbi) 2016 { 2017 return (struct f2fs_super_block *)(sbi->raw_super); 2018 } 2019 2020 static inline struct f2fs_checkpoint *F2FS_CKPT(struct f2fs_sb_info *sbi) 2021 { 2022 return (struct f2fs_checkpoint *)(sbi->ckpt); 2023 } 2024 2025 static inline struct f2fs_node *F2FS_NODE(struct page *page) 2026 { 2027 return (struct f2fs_node *)page_address(page); 2028 } 2029 2030 static inline struct f2fs_inode *F2FS_INODE(struct page *page) 2031 { 2032 return &((struct f2fs_node *)page_address(page))->i; 2033 } 2034 2035 static inline struct f2fs_nm_info *NM_I(struct f2fs_sb_info *sbi) 2036 { 2037 return (struct f2fs_nm_info *)(sbi->nm_info); 2038 } 2039 2040 static inline struct f2fs_sm_info *SM_I(struct f2fs_sb_info *sbi) 2041 { 2042 return (struct f2fs_sm_info *)(sbi->sm_info); 2043 } 2044 2045 static inline struct sit_info *SIT_I(struct f2fs_sb_info *sbi) 2046 { 2047 return (struct sit_info *)(SM_I(sbi)->sit_info); 2048 } 2049 2050 static inline struct free_segmap_info *FREE_I(struct f2fs_sb_info *sbi) 2051 { 2052 return (struct free_segmap_info *)(SM_I(sbi)->free_info); 2053 } 2054 2055 static inline struct dirty_seglist_info *DIRTY_I(struct f2fs_sb_info *sbi) 2056 { 2057 return (struct dirty_seglist_info *)(SM_I(sbi)->dirty_info); 2058 } 2059 2060 static inline struct address_space *META_MAPPING(struct f2fs_sb_info *sbi) 2061 { 2062 return sbi->meta_inode->i_mapping; 2063 } 2064 2065 static inline struct address_space *NODE_MAPPING(struct f2fs_sb_info *sbi) 2066 { 2067 return sbi->node_inode->i_mapping; 2068 } 2069 2070 static inline bool is_sbi_flag_set(struct f2fs_sb_info *sbi, unsigned int type) 2071 { 2072 return test_bit(type, &sbi->s_flag); 2073 } 2074 2075 static inline void set_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type) 2076 { 2077 set_bit(type, &sbi->s_flag); 2078 } 2079 2080 static inline void clear_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type) 2081 { 2082 clear_bit(type, &sbi->s_flag); 2083 } 2084 2085 static inline unsigned long long cur_cp_version(struct f2fs_checkpoint *cp) 2086 { 2087 return le64_to_cpu(cp->checkpoint_ver); 2088 } 2089 2090 static inline unsigned long f2fs_qf_ino(struct super_block *sb, int type) 2091 { 2092 if (type < F2FS_MAX_QUOTAS) 2093 return le32_to_cpu(F2FS_SB(sb)->raw_super->qf_ino[type]); 2094 return 0; 2095 } 2096 2097 static inline __u64 cur_cp_crc(struct f2fs_checkpoint *cp) 2098 { 2099 size_t crc_offset = le32_to_cpu(cp->checksum_offset); 2100 return le32_to_cpu(*((__le32 *)((unsigned char *)cp + crc_offset))); 2101 } 2102 2103 static inline bool __is_set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 2104 { 2105 unsigned int ckpt_flags = le32_to_cpu(cp->ckpt_flags); 2106 2107 return ckpt_flags & f; 2108 } 2109 2110 static inline bool is_set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 2111 { 2112 return __is_set_ckpt_flags(F2FS_CKPT(sbi), f); 2113 } 2114 2115 static inline void __set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 2116 { 2117 unsigned int ckpt_flags; 2118 2119 ckpt_flags = le32_to_cpu(cp->ckpt_flags); 2120 ckpt_flags |= f; 2121 cp->ckpt_flags = cpu_to_le32(ckpt_flags); 2122 } 2123 2124 static inline void set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 2125 { 2126 unsigned long flags; 2127 2128 spin_lock_irqsave(&sbi->cp_lock, flags); 2129 __set_ckpt_flags(F2FS_CKPT(sbi), f); 2130 spin_unlock_irqrestore(&sbi->cp_lock, flags); 2131 } 2132 2133 static inline void __clear_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 2134 { 2135 unsigned int ckpt_flags; 2136 2137 ckpt_flags = le32_to_cpu(cp->ckpt_flags); 2138 ckpt_flags &= (~f); 2139 cp->ckpt_flags = cpu_to_le32(ckpt_flags); 2140 } 2141 2142 static inline void clear_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 2143 { 2144 unsigned long flags; 2145 2146 spin_lock_irqsave(&sbi->cp_lock, flags); 2147 __clear_ckpt_flags(F2FS_CKPT(sbi), f); 2148 spin_unlock_irqrestore(&sbi->cp_lock, flags); 2149 } 2150 2151 #define init_f2fs_rwsem(sem) \ 2152 do { \ 2153 static struct lock_class_key __key; \ 2154 \ 2155 __init_f2fs_rwsem((sem), #sem, &__key); \ 2156 } while (0) 2157 2158 static inline void __init_f2fs_rwsem(struct f2fs_rwsem *sem, 2159 const char *sem_name, struct lock_class_key *key) 2160 { 2161 __init_rwsem(&sem->internal_rwsem, sem_name, key); 2162 #ifdef CONFIG_F2FS_UNFAIR_RWSEM 2163 init_waitqueue_head(&sem->read_waiters); 2164 #endif 2165 } 2166 2167 static inline int f2fs_rwsem_is_locked(struct f2fs_rwsem *sem) 2168 { 2169 return rwsem_is_locked(&sem->internal_rwsem); 2170 } 2171 2172 static inline int f2fs_rwsem_is_contended(struct f2fs_rwsem *sem) 2173 { 2174 return rwsem_is_contended(&sem->internal_rwsem); 2175 } 2176 2177 static inline void f2fs_down_read(struct f2fs_rwsem *sem) 2178 { 2179 #ifdef CONFIG_F2FS_UNFAIR_RWSEM 2180 wait_event(sem->read_waiters, down_read_trylock(&sem->internal_rwsem)); 2181 #else 2182 down_read(&sem->internal_rwsem); 2183 #endif 2184 } 2185 2186 static inline int f2fs_down_read_trylock(struct f2fs_rwsem *sem) 2187 { 2188 return down_read_trylock(&sem->internal_rwsem); 2189 } 2190 2191 #ifdef CONFIG_DEBUG_LOCK_ALLOC 2192 static inline void f2fs_down_read_nested(struct f2fs_rwsem *sem, int subclass) 2193 { 2194 down_read_nested(&sem->internal_rwsem, subclass); 2195 } 2196 #else 2197 #define f2fs_down_read_nested(sem, subclass) f2fs_down_read(sem) 2198 #endif 2199 2200 static inline void f2fs_up_read(struct f2fs_rwsem *sem) 2201 { 2202 up_read(&sem->internal_rwsem); 2203 } 2204 2205 static inline void f2fs_down_write(struct f2fs_rwsem *sem) 2206 { 2207 down_write(&sem->internal_rwsem); 2208 } 2209 2210 static inline int f2fs_down_write_trylock(struct f2fs_rwsem *sem) 2211 { 2212 return down_write_trylock(&sem->internal_rwsem); 2213 } 2214 2215 static inline void f2fs_up_write(struct f2fs_rwsem *sem) 2216 { 2217 up_write(&sem->internal_rwsem); 2218 #ifdef CONFIG_F2FS_UNFAIR_RWSEM 2219 wake_up_all(&sem->read_waiters); 2220 #endif 2221 } 2222 2223 static inline void f2fs_lock_op(struct f2fs_sb_info *sbi) 2224 { 2225 f2fs_down_read(&sbi->cp_rwsem); 2226 } 2227 2228 static inline int f2fs_trylock_op(struct f2fs_sb_info *sbi) 2229 { 2230 if (time_to_inject(sbi, FAULT_LOCK_OP)) 2231 return 0; 2232 return f2fs_down_read_trylock(&sbi->cp_rwsem); 2233 } 2234 2235 static inline void f2fs_unlock_op(struct f2fs_sb_info *sbi) 2236 { 2237 f2fs_up_read(&sbi->cp_rwsem); 2238 } 2239 2240 static inline void f2fs_lock_all(struct f2fs_sb_info *sbi) 2241 { 2242 f2fs_down_write(&sbi->cp_rwsem); 2243 } 2244 2245 static inline void f2fs_unlock_all(struct f2fs_sb_info *sbi) 2246 { 2247 f2fs_up_write(&sbi->cp_rwsem); 2248 } 2249 2250 static inline int __get_cp_reason(struct f2fs_sb_info *sbi) 2251 { 2252 int reason = CP_SYNC; 2253 2254 if (test_opt(sbi, FASTBOOT)) 2255 reason = CP_FASTBOOT; 2256 if (is_sbi_flag_set(sbi, SBI_IS_CLOSE)) 2257 reason = CP_UMOUNT; 2258 return reason; 2259 } 2260 2261 static inline bool __remain_node_summaries(int reason) 2262 { 2263 return (reason & (CP_UMOUNT | CP_FASTBOOT)); 2264 } 2265 2266 static inline bool __exist_node_summaries(struct f2fs_sb_info *sbi) 2267 { 2268 return (is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG) || 2269 is_set_ckpt_flags(sbi, CP_FASTBOOT_FLAG)); 2270 } 2271 2272 /* 2273 * Check whether the inode has blocks or not 2274 */ 2275 static inline int F2FS_HAS_BLOCKS(struct inode *inode) 2276 { 2277 block_t xattr_block = F2FS_I(inode)->i_xattr_nid ? 1 : 0; 2278 2279 return (inode->i_blocks >> F2FS_LOG_SECTORS_PER_BLOCK) > xattr_block; 2280 } 2281 2282 static inline bool f2fs_has_xattr_block(unsigned int ofs) 2283 { 2284 return ofs == XATTR_NODE_OFFSET; 2285 } 2286 2287 static inline bool __allow_reserved_blocks(struct f2fs_sb_info *sbi, 2288 struct inode *inode, bool cap) 2289 { 2290 if (!inode) 2291 return true; 2292 if (!test_opt(sbi, RESERVE_ROOT)) 2293 return false; 2294 if (IS_NOQUOTA(inode)) 2295 return true; 2296 if (uid_eq(F2FS_OPTION(sbi).s_resuid, current_fsuid())) 2297 return true; 2298 if (!gid_eq(F2FS_OPTION(sbi).s_resgid, GLOBAL_ROOT_GID) && 2299 in_group_p(F2FS_OPTION(sbi).s_resgid)) 2300 return true; 2301 if (cap && capable(CAP_SYS_RESOURCE)) 2302 return true; 2303 return false; 2304 } 2305 2306 static inline void f2fs_i_blocks_write(struct inode *, block_t, bool, bool); 2307 static inline int inc_valid_block_count(struct f2fs_sb_info *sbi, 2308 struct inode *inode, blkcnt_t *count) 2309 { 2310 blkcnt_t diff = 0, release = 0; 2311 block_t avail_user_block_count; 2312 int ret; 2313 2314 ret = dquot_reserve_block(inode, *count); 2315 if (ret) 2316 return ret; 2317 2318 if (time_to_inject(sbi, FAULT_BLOCK)) { 2319 release = *count; 2320 goto release_quota; 2321 } 2322 2323 /* 2324 * let's increase this in prior to actual block count change in order 2325 * for f2fs_sync_file to avoid data races when deciding checkpoint. 2326 */ 2327 percpu_counter_add(&sbi->alloc_valid_block_count, (*count)); 2328 2329 spin_lock(&sbi->stat_lock); 2330 sbi->total_valid_block_count += (block_t)(*count); 2331 avail_user_block_count = sbi->user_block_count - 2332 sbi->current_reserved_blocks; 2333 2334 if (!__allow_reserved_blocks(sbi, inode, true)) 2335 avail_user_block_count -= F2FS_OPTION(sbi).root_reserved_blocks; 2336 2337 if (F2FS_IO_ALIGNED(sbi)) 2338 avail_user_block_count -= sbi->blocks_per_seg * 2339 SM_I(sbi)->additional_reserved_segments; 2340 2341 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) { 2342 if (avail_user_block_count > sbi->unusable_block_count) 2343 avail_user_block_count -= sbi->unusable_block_count; 2344 else 2345 avail_user_block_count = 0; 2346 } 2347 if (unlikely(sbi->total_valid_block_count > avail_user_block_count)) { 2348 diff = sbi->total_valid_block_count - avail_user_block_count; 2349 if (diff > *count) 2350 diff = *count; 2351 *count -= diff; 2352 release = diff; 2353 sbi->total_valid_block_count -= diff; 2354 if (!*count) { 2355 spin_unlock(&sbi->stat_lock); 2356 goto enospc; 2357 } 2358 } 2359 spin_unlock(&sbi->stat_lock); 2360 2361 if (unlikely(release)) { 2362 percpu_counter_sub(&sbi->alloc_valid_block_count, release); 2363 dquot_release_reservation_block(inode, release); 2364 } 2365 f2fs_i_blocks_write(inode, *count, true, true); 2366 return 0; 2367 2368 enospc: 2369 percpu_counter_sub(&sbi->alloc_valid_block_count, release); 2370 release_quota: 2371 dquot_release_reservation_block(inode, release); 2372 return -ENOSPC; 2373 } 2374 2375 __printf(2, 3) 2376 void f2fs_printk(struct f2fs_sb_info *sbi, const char *fmt, ...); 2377 2378 #define f2fs_err(sbi, fmt, ...) \ 2379 f2fs_printk(sbi, KERN_ERR fmt, ##__VA_ARGS__) 2380 #define f2fs_warn(sbi, fmt, ...) \ 2381 f2fs_printk(sbi, KERN_WARNING fmt, ##__VA_ARGS__) 2382 #define f2fs_notice(sbi, fmt, ...) \ 2383 f2fs_printk(sbi, KERN_NOTICE fmt, ##__VA_ARGS__) 2384 #define f2fs_info(sbi, fmt, ...) \ 2385 f2fs_printk(sbi, KERN_INFO fmt, ##__VA_ARGS__) 2386 #define f2fs_debug(sbi, fmt, ...) \ 2387 f2fs_printk(sbi, KERN_DEBUG fmt, ##__VA_ARGS__) 2388 2389 static inline void dec_valid_block_count(struct f2fs_sb_info *sbi, 2390 struct inode *inode, 2391 block_t count) 2392 { 2393 blkcnt_t sectors = count << F2FS_LOG_SECTORS_PER_BLOCK; 2394 2395 spin_lock(&sbi->stat_lock); 2396 f2fs_bug_on(sbi, sbi->total_valid_block_count < (block_t) count); 2397 sbi->total_valid_block_count -= (block_t)count; 2398 if (sbi->reserved_blocks && 2399 sbi->current_reserved_blocks < sbi->reserved_blocks) 2400 sbi->current_reserved_blocks = min(sbi->reserved_blocks, 2401 sbi->current_reserved_blocks + count); 2402 spin_unlock(&sbi->stat_lock); 2403 if (unlikely(inode->i_blocks < sectors)) { 2404 f2fs_warn(sbi, "Inconsistent i_blocks, ino:%lu, iblocks:%llu, sectors:%llu", 2405 inode->i_ino, 2406 (unsigned long long)inode->i_blocks, 2407 (unsigned long long)sectors); 2408 set_sbi_flag(sbi, SBI_NEED_FSCK); 2409 return; 2410 } 2411 f2fs_i_blocks_write(inode, count, false, true); 2412 } 2413 2414 static inline void inc_page_count(struct f2fs_sb_info *sbi, int count_type) 2415 { 2416 atomic_inc(&sbi->nr_pages[count_type]); 2417 2418 if (count_type == F2FS_DIRTY_DENTS || 2419 count_type == F2FS_DIRTY_NODES || 2420 count_type == F2FS_DIRTY_META || 2421 count_type == F2FS_DIRTY_QDATA || 2422 count_type == F2FS_DIRTY_IMETA) 2423 set_sbi_flag(sbi, SBI_IS_DIRTY); 2424 } 2425 2426 static inline void inode_inc_dirty_pages(struct inode *inode) 2427 { 2428 atomic_inc(&F2FS_I(inode)->dirty_pages); 2429 inc_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ? 2430 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA); 2431 if (IS_NOQUOTA(inode)) 2432 inc_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA); 2433 } 2434 2435 static inline void dec_page_count(struct f2fs_sb_info *sbi, int count_type) 2436 { 2437 atomic_dec(&sbi->nr_pages[count_type]); 2438 } 2439 2440 static inline void inode_dec_dirty_pages(struct inode *inode) 2441 { 2442 if (!S_ISDIR(inode->i_mode) && !S_ISREG(inode->i_mode) && 2443 !S_ISLNK(inode->i_mode)) 2444 return; 2445 2446 atomic_dec(&F2FS_I(inode)->dirty_pages); 2447 dec_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ? 2448 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA); 2449 if (IS_NOQUOTA(inode)) 2450 dec_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA); 2451 } 2452 2453 static inline void inc_atomic_write_cnt(struct inode *inode) 2454 { 2455 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 2456 struct f2fs_inode_info *fi = F2FS_I(inode); 2457 u64 current_write; 2458 2459 fi->atomic_write_cnt++; 2460 atomic64_inc(&sbi->current_atomic_write); 2461 current_write = atomic64_read(&sbi->current_atomic_write); 2462 if (current_write > sbi->peak_atomic_write) 2463 sbi->peak_atomic_write = current_write; 2464 } 2465 2466 static inline void release_atomic_write_cnt(struct inode *inode) 2467 { 2468 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 2469 struct f2fs_inode_info *fi = F2FS_I(inode); 2470 2471 atomic64_sub(fi->atomic_write_cnt, &sbi->current_atomic_write); 2472 fi->atomic_write_cnt = 0; 2473 } 2474 2475 static inline s64 get_pages(struct f2fs_sb_info *sbi, int count_type) 2476 { 2477 return atomic_read(&sbi->nr_pages[count_type]); 2478 } 2479 2480 static inline int get_dirty_pages(struct inode *inode) 2481 { 2482 return atomic_read(&F2FS_I(inode)->dirty_pages); 2483 } 2484 2485 static inline int get_blocktype_secs(struct f2fs_sb_info *sbi, int block_type) 2486 { 2487 unsigned int pages_per_sec = sbi->segs_per_sec * sbi->blocks_per_seg; 2488 unsigned int segs = (get_pages(sbi, block_type) + pages_per_sec - 1) >> 2489 sbi->log_blocks_per_seg; 2490 2491 return segs / sbi->segs_per_sec; 2492 } 2493 2494 static inline block_t valid_user_blocks(struct f2fs_sb_info *sbi) 2495 { 2496 return sbi->total_valid_block_count; 2497 } 2498 2499 static inline block_t discard_blocks(struct f2fs_sb_info *sbi) 2500 { 2501 return sbi->discard_blks; 2502 } 2503 2504 static inline unsigned long __bitmap_size(struct f2fs_sb_info *sbi, int flag) 2505 { 2506 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi); 2507 2508 /* return NAT or SIT bitmap */ 2509 if (flag == NAT_BITMAP) 2510 return le32_to_cpu(ckpt->nat_ver_bitmap_bytesize); 2511 else if (flag == SIT_BITMAP) 2512 return le32_to_cpu(ckpt->sit_ver_bitmap_bytesize); 2513 2514 return 0; 2515 } 2516 2517 static inline block_t __cp_payload(struct f2fs_sb_info *sbi) 2518 { 2519 return le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_payload); 2520 } 2521 2522 static inline void *__bitmap_ptr(struct f2fs_sb_info *sbi, int flag) 2523 { 2524 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi); 2525 void *tmp_ptr = &ckpt->sit_nat_version_bitmap; 2526 int offset; 2527 2528 if (is_set_ckpt_flags(sbi, CP_LARGE_NAT_BITMAP_FLAG)) { 2529 offset = (flag == SIT_BITMAP) ? 2530 le32_to_cpu(ckpt->nat_ver_bitmap_bytesize) : 0; 2531 /* 2532 * if large_nat_bitmap feature is enabled, leave checksum 2533 * protection for all nat/sit bitmaps. 2534 */ 2535 return tmp_ptr + offset + sizeof(__le32); 2536 } 2537 2538 if (__cp_payload(sbi) > 0) { 2539 if (flag == NAT_BITMAP) 2540 return tmp_ptr; 2541 else 2542 return (unsigned char *)ckpt + F2FS_BLKSIZE; 2543 } else { 2544 offset = (flag == NAT_BITMAP) ? 2545 le32_to_cpu(ckpt->sit_ver_bitmap_bytesize) : 0; 2546 return tmp_ptr + offset; 2547 } 2548 } 2549 2550 static inline block_t __start_cp_addr(struct f2fs_sb_info *sbi) 2551 { 2552 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr); 2553 2554 if (sbi->cur_cp_pack == 2) 2555 start_addr += sbi->blocks_per_seg; 2556 return start_addr; 2557 } 2558 2559 static inline block_t __start_cp_next_addr(struct f2fs_sb_info *sbi) 2560 { 2561 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr); 2562 2563 if (sbi->cur_cp_pack == 1) 2564 start_addr += sbi->blocks_per_seg; 2565 return start_addr; 2566 } 2567 2568 static inline void __set_cp_next_pack(struct f2fs_sb_info *sbi) 2569 { 2570 sbi->cur_cp_pack = (sbi->cur_cp_pack == 1) ? 2 : 1; 2571 } 2572 2573 static inline block_t __start_sum_addr(struct f2fs_sb_info *sbi) 2574 { 2575 return le32_to_cpu(F2FS_CKPT(sbi)->cp_pack_start_sum); 2576 } 2577 2578 extern void f2fs_mark_inode_dirty_sync(struct inode *inode, bool sync); 2579 static inline int inc_valid_node_count(struct f2fs_sb_info *sbi, 2580 struct inode *inode, bool is_inode) 2581 { 2582 block_t valid_block_count; 2583 unsigned int valid_node_count, user_block_count; 2584 int err; 2585 2586 if (is_inode) { 2587 if (inode) { 2588 err = dquot_alloc_inode(inode); 2589 if (err) 2590 return err; 2591 } 2592 } else { 2593 err = dquot_reserve_block(inode, 1); 2594 if (err) 2595 return err; 2596 } 2597 2598 if (time_to_inject(sbi, FAULT_BLOCK)) 2599 goto enospc; 2600 2601 spin_lock(&sbi->stat_lock); 2602 2603 valid_block_count = sbi->total_valid_block_count + 2604 sbi->current_reserved_blocks + 1; 2605 2606 if (!__allow_reserved_blocks(sbi, inode, false)) 2607 valid_block_count += F2FS_OPTION(sbi).root_reserved_blocks; 2608 2609 if (F2FS_IO_ALIGNED(sbi)) 2610 valid_block_count += sbi->blocks_per_seg * 2611 SM_I(sbi)->additional_reserved_segments; 2612 2613 user_block_count = sbi->user_block_count; 2614 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) 2615 user_block_count -= sbi->unusable_block_count; 2616 2617 if (unlikely(valid_block_count > user_block_count)) { 2618 spin_unlock(&sbi->stat_lock); 2619 goto enospc; 2620 } 2621 2622 valid_node_count = sbi->total_valid_node_count + 1; 2623 if (unlikely(valid_node_count > sbi->total_node_count)) { 2624 spin_unlock(&sbi->stat_lock); 2625 goto enospc; 2626 } 2627 2628 sbi->total_valid_node_count++; 2629 sbi->total_valid_block_count++; 2630 spin_unlock(&sbi->stat_lock); 2631 2632 if (inode) { 2633 if (is_inode) 2634 f2fs_mark_inode_dirty_sync(inode, true); 2635 else 2636 f2fs_i_blocks_write(inode, 1, true, true); 2637 } 2638 2639 percpu_counter_inc(&sbi->alloc_valid_block_count); 2640 return 0; 2641 2642 enospc: 2643 if (is_inode) { 2644 if (inode) 2645 dquot_free_inode(inode); 2646 } else { 2647 dquot_release_reservation_block(inode, 1); 2648 } 2649 return -ENOSPC; 2650 } 2651 2652 static inline void dec_valid_node_count(struct f2fs_sb_info *sbi, 2653 struct inode *inode, bool is_inode) 2654 { 2655 spin_lock(&sbi->stat_lock); 2656 2657 if (unlikely(!sbi->total_valid_block_count || 2658 !sbi->total_valid_node_count)) { 2659 f2fs_warn(sbi, "dec_valid_node_count: inconsistent block counts, total_valid_block:%u, total_valid_node:%u", 2660 sbi->total_valid_block_count, 2661 sbi->total_valid_node_count); 2662 set_sbi_flag(sbi, SBI_NEED_FSCK); 2663 } else { 2664 sbi->total_valid_block_count--; 2665 sbi->total_valid_node_count--; 2666 } 2667 2668 if (sbi->reserved_blocks && 2669 sbi->current_reserved_blocks < sbi->reserved_blocks) 2670 sbi->current_reserved_blocks++; 2671 2672 spin_unlock(&sbi->stat_lock); 2673 2674 if (is_inode) { 2675 dquot_free_inode(inode); 2676 } else { 2677 if (unlikely(inode->i_blocks == 0)) { 2678 f2fs_warn(sbi, "dec_valid_node_count: inconsistent i_blocks, ino:%lu, iblocks:%llu", 2679 inode->i_ino, 2680 (unsigned long long)inode->i_blocks); 2681 set_sbi_flag(sbi, SBI_NEED_FSCK); 2682 return; 2683 } 2684 f2fs_i_blocks_write(inode, 1, false, true); 2685 } 2686 } 2687 2688 static inline unsigned int valid_node_count(struct f2fs_sb_info *sbi) 2689 { 2690 return sbi->total_valid_node_count; 2691 } 2692 2693 static inline void inc_valid_inode_count(struct f2fs_sb_info *sbi) 2694 { 2695 percpu_counter_inc(&sbi->total_valid_inode_count); 2696 } 2697 2698 static inline void dec_valid_inode_count(struct f2fs_sb_info *sbi) 2699 { 2700 percpu_counter_dec(&sbi->total_valid_inode_count); 2701 } 2702 2703 static inline s64 valid_inode_count(struct f2fs_sb_info *sbi) 2704 { 2705 return percpu_counter_sum_positive(&sbi->total_valid_inode_count); 2706 } 2707 2708 static inline struct page *f2fs_grab_cache_page(struct address_space *mapping, 2709 pgoff_t index, bool for_write) 2710 { 2711 struct page *page; 2712 unsigned int flags; 2713 2714 if (IS_ENABLED(CONFIG_F2FS_FAULT_INJECTION)) { 2715 if (!for_write) 2716 page = find_get_page_flags(mapping, index, 2717 FGP_LOCK | FGP_ACCESSED); 2718 else 2719 page = find_lock_page(mapping, index); 2720 if (page) 2721 return page; 2722 2723 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_ALLOC)) 2724 return NULL; 2725 } 2726 2727 if (!for_write) 2728 return grab_cache_page(mapping, index); 2729 2730 flags = memalloc_nofs_save(); 2731 page = grab_cache_page_write_begin(mapping, index); 2732 memalloc_nofs_restore(flags); 2733 2734 return page; 2735 } 2736 2737 static inline struct page *f2fs_pagecache_get_page( 2738 struct address_space *mapping, pgoff_t index, 2739 int fgp_flags, gfp_t gfp_mask) 2740 { 2741 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_GET)) 2742 return NULL; 2743 2744 return pagecache_get_page(mapping, index, fgp_flags, gfp_mask); 2745 } 2746 2747 static inline void f2fs_put_page(struct page *page, int unlock) 2748 { 2749 if (!page) 2750 return; 2751 2752 if (unlock) { 2753 f2fs_bug_on(F2FS_P_SB(page), !PageLocked(page)); 2754 unlock_page(page); 2755 } 2756 put_page(page); 2757 } 2758 2759 static inline void f2fs_put_dnode(struct dnode_of_data *dn) 2760 { 2761 if (dn->node_page) 2762 f2fs_put_page(dn->node_page, 1); 2763 if (dn->inode_page && dn->node_page != dn->inode_page) 2764 f2fs_put_page(dn->inode_page, 0); 2765 dn->node_page = NULL; 2766 dn->inode_page = NULL; 2767 } 2768 2769 static inline struct kmem_cache *f2fs_kmem_cache_create(const char *name, 2770 size_t size) 2771 { 2772 return kmem_cache_create(name, size, 0, SLAB_RECLAIM_ACCOUNT, NULL); 2773 } 2774 2775 static inline void *f2fs_kmem_cache_alloc_nofail(struct kmem_cache *cachep, 2776 gfp_t flags) 2777 { 2778 void *entry; 2779 2780 entry = kmem_cache_alloc(cachep, flags); 2781 if (!entry) 2782 entry = kmem_cache_alloc(cachep, flags | __GFP_NOFAIL); 2783 return entry; 2784 } 2785 2786 static inline void *f2fs_kmem_cache_alloc(struct kmem_cache *cachep, 2787 gfp_t flags, bool nofail, struct f2fs_sb_info *sbi) 2788 { 2789 if (nofail) 2790 return f2fs_kmem_cache_alloc_nofail(cachep, flags); 2791 2792 if (time_to_inject(sbi, FAULT_SLAB_ALLOC)) 2793 return NULL; 2794 2795 return kmem_cache_alloc(cachep, flags); 2796 } 2797 2798 static inline bool is_inflight_io(struct f2fs_sb_info *sbi, int type) 2799 { 2800 if (get_pages(sbi, F2FS_RD_DATA) || get_pages(sbi, F2FS_RD_NODE) || 2801 get_pages(sbi, F2FS_RD_META) || get_pages(sbi, F2FS_WB_DATA) || 2802 get_pages(sbi, F2FS_WB_CP_DATA) || 2803 get_pages(sbi, F2FS_DIO_READ) || 2804 get_pages(sbi, F2FS_DIO_WRITE)) 2805 return true; 2806 2807 if (type != DISCARD_TIME && SM_I(sbi) && SM_I(sbi)->dcc_info && 2808 atomic_read(&SM_I(sbi)->dcc_info->queued_discard)) 2809 return true; 2810 2811 if (SM_I(sbi) && SM_I(sbi)->fcc_info && 2812 atomic_read(&SM_I(sbi)->fcc_info->queued_flush)) 2813 return true; 2814 return false; 2815 } 2816 2817 static inline bool is_idle(struct f2fs_sb_info *sbi, int type) 2818 { 2819 if (sbi->gc_mode == GC_URGENT_HIGH) 2820 return true; 2821 2822 if (is_inflight_io(sbi, type)) 2823 return false; 2824 2825 if (sbi->gc_mode == GC_URGENT_MID) 2826 return true; 2827 2828 if (sbi->gc_mode == GC_URGENT_LOW && 2829 (type == DISCARD_TIME || type == GC_TIME)) 2830 return true; 2831 2832 return f2fs_time_over(sbi, type); 2833 } 2834 2835 static inline void f2fs_radix_tree_insert(struct radix_tree_root *root, 2836 unsigned long index, void *item) 2837 { 2838 while (radix_tree_insert(root, index, item)) 2839 cond_resched(); 2840 } 2841 2842 #define RAW_IS_INODE(p) ((p)->footer.nid == (p)->footer.ino) 2843 2844 static inline bool IS_INODE(struct page *page) 2845 { 2846 struct f2fs_node *p = F2FS_NODE(page); 2847 2848 return RAW_IS_INODE(p); 2849 } 2850 2851 static inline int offset_in_addr(struct f2fs_inode *i) 2852 { 2853 return (i->i_inline & F2FS_EXTRA_ATTR) ? 2854 (le16_to_cpu(i->i_extra_isize) / sizeof(__le32)) : 0; 2855 } 2856 2857 static inline __le32 *blkaddr_in_node(struct f2fs_node *node) 2858 { 2859 return RAW_IS_INODE(node) ? node->i.i_addr : node->dn.addr; 2860 } 2861 2862 static inline int f2fs_has_extra_attr(struct inode *inode); 2863 static inline block_t data_blkaddr(struct inode *inode, 2864 struct page *node_page, unsigned int offset) 2865 { 2866 struct f2fs_node *raw_node; 2867 __le32 *addr_array; 2868 int base = 0; 2869 bool is_inode = IS_INODE(node_page); 2870 2871 raw_node = F2FS_NODE(node_page); 2872 2873 if (is_inode) { 2874 if (!inode) 2875 /* from GC path only */ 2876 base = offset_in_addr(&raw_node->i); 2877 else if (f2fs_has_extra_attr(inode)) 2878 base = get_extra_isize(inode); 2879 } 2880 2881 addr_array = blkaddr_in_node(raw_node); 2882 return le32_to_cpu(addr_array[base + offset]); 2883 } 2884 2885 static inline block_t f2fs_data_blkaddr(struct dnode_of_data *dn) 2886 { 2887 return data_blkaddr(dn->inode, dn->node_page, dn->ofs_in_node); 2888 } 2889 2890 static inline int f2fs_test_bit(unsigned int nr, char *addr) 2891 { 2892 int mask; 2893 2894 addr += (nr >> 3); 2895 mask = 1 << (7 - (nr & 0x07)); 2896 return mask & *addr; 2897 } 2898 2899 static inline void f2fs_set_bit(unsigned int nr, char *addr) 2900 { 2901 int mask; 2902 2903 addr += (nr >> 3); 2904 mask = 1 << (7 - (nr & 0x07)); 2905 *addr |= mask; 2906 } 2907 2908 static inline void f2fs_clear_bit(unsigned int nr, char *addr) 2909 { 2910 int mask; 2911 2912 addr += (nr >> 3); 2913 mask = 1 << (7 - (nr & 0x07)); 2914 *addr &= ~mask; 2915 } 2916 2917 static inline int f2fs_test_and_set_bit(unsigned int nr, char *addr) 2918 { 2919 int mask; 2920 int ret; 2921 2922 addr += (nr >> 3); 2923 mask = 1 << (7 - (nr & 0x07)); 2924 ret = mask & *addr; 2925 *addr |= mask; 2926 return ret; 2927 } 2928 2929 static inline int f2fs_test_and_clear_bit(unsigned int nr, char *addr) 2930 { 2931 int mask; 2932 int ret; 2933 2934 addr += (nr >> 3); 2935 mask = 1 << (7 - (nr & 0x07)); 2936 ret = mask & *addr; 2937 *addr &= ~mask; 2938 return ret; 2939 } 2940 2941 static inline void f2fs_change_bit(unsigned int nr, char *addr) 2942 { 2943 int mask; 2944 2945 addr += (nr >> 3); 2946 mask = 1 << (7 - (nr & 0x07)); 2947 *addr ^= mask; 2948 } 2949 2950 /* 2951 * On-disk inode flags (f2fs_inode::i_flags) 2952 */ 2953 #define F2FS_COMPR_FL 0x00000004 /* Compress file */ 2954 #define F2FS_SYNC_FL 0x00000008 /* Synchronous updates */ 2955 #define F2FS_IMMUTABLE_FL 0x00000010 /* Immutable file */ 2956 #define F2FS_APPEND_FL 0x00000020 /* writes to file may only append */ 2957 #define F2FS_NODUMP_FL 0x00000040 /* do not dump file */ 2958 #define F2FS_NOATIME_FL 0x00000080 /* do not update atime */ 2959 #define F2FS_NOCOMP_FL 0x00000400 /* Don't compress */ 2960 #define F2FS_INDEX_FL 0x00001000 /* hash-indexed directory */ 2961 #define F2FS_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */ 2962 #define F2FS_PROJINHERIT_FL 0x20000000 /* Create with parents projid */ 2963 #define F2FS_CASEFOLD_FL 0x40000000 /* Casefolded file */ 2964 2965 /* Flags that should be inherited by new inodes from their parent. */ 2966 #define F2FS_FL_INHERITED (F2FS_SYNC_FL | F2FS_NODUMP_FL | F2FS_NOATIME_FL | \ 2967 F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \ 2968 F2FS_CASEFOLD_FL) 2969 2970 /* Flags that are appropriate for regular files (all but dir-specific ones). */ 2971 #define F2FS_REG_FLMASK (~(F2FS_DIRSYNC_FL | F2FS_PROJINHERIT_FL | \ 2972 F2FS_CASEFOLD_FL)) 2973 2974 /* Flags that are appropriate for non-directories/regular files. */ 2975 #define F2FS_OTHER_FLMASK (F2FS_NODUMP_FL | F2FS_NOATIME_FL) 2976 2977 static inline __u32 f2fs_mask_flags(umode_t mode, __u32 flags) 2978 { 2979 if (S_ISDIR(mode)) 2980 return flags; 2981 else if (S_ISREG(mode)) 2982 return flags & F2FS_REG_FLMASK; 2983 else 2984 return flags & F2FS_OTHER_FLMASK; 2985 } 2986 2987 static inline void __mark_inode_dirty_flag(struct inode *inode, 2988 int flag, bool set) 2989 { 2990 switch (flag) { 2991 case FI_INLINE_XATTR: 2992 case FI_INLINE_DATA: 2993 case FI_INLINE_DENTRY: 2994 case FI_NEW_INODE: 2995 if (set) 2996 return; 2997 fallthrough; 2998 case FI_DATA_EXIST: 2999 case FI_INLINE_DOTS: 3000 case FI_PIN_FILE: 3001 case FI_COMPRESS_RELEASED: 3002 f2fs_mark_inode_dirty_sync(inode, true); 3003 } 3004 } 3005 3006 static inline void set_inode_flag(struct inode *inode, int flag) 3007 { 3008 set_bit(flag, F2FS_I(inode)->flags); 3009 __mark_inode_dirty_flag(inode, flag, true); 3010 } 3011 3012 static inline int is_inode_flag_set(struct inode *inode, int flag) 3013 { 3014 return test_bit(flag, F2FS_I(inode)->flags); 3015 } 3016 3017 static inline void clear_inode_flag(struct inode *inode, int flag) 3018 { 3019 clear_bit(flag, F2FS_I(inode)->flags); 3020 __mark_inode_dirty_flag(inode, flag, false); 3021 } 3022 3023 static inline bool f2fs_verity_in_progress(struct inode *inode) 3024 { 3025 return IS_ENABLED(CONFIG_FS_VERITY) && 3026 is_inode_flag_set(inode, FI_VERITY_IN_PROGRESS); 3027 } 3028 3029 static inline void set_acl_inode(struct inode *inode, umode_t mode) 3030 { 3031 F2FS_I(inode)->i_acl_mode = mode; 3032 set_inode_flag(inode, FI_ACL_MODE); 3033 f2fs_mark_inode_dirty_sync(inode, false); 3034 } 3035 3036 static inline void f2fs_i_links_write(struct inode *inode, bool inc) 3037 { 3038 if (inc) 3039 inc_nlink(inode); 3040 else 3041 drop_nlink(inode); 3042 f2fs_mark_inode_dirty_sync(inode, true); 3043 } 3044 3045 static inline void f2fs_i_blocks_write(struct inode *inode, 3046 block_t diff, bool add, bool claim) 3047 { 3048 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE); 3049 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER); 3050 3051 /* add = 1, claim = 1 should be dquot_reserve_block in pair */ 3052 if (add) { 3053 if (claim) 3054 dquot_claim_block(inode, diff); 3055 else 3056 dquot_alloc_block_nofail(inode, diff); 3057 } else { 3058 dquot_free_block(inode, diff); 3059 } 3060 3061 f2fs_mark_inode_dirty_sync(inode, true); 3062 if (clean || recover) 3063 set_inode_flag(inode, FI_AUTO_RECOVER); 3064 } 3065 3066 static inline bool f2fs_is_atomic_file(struct inode *inode); 3067 3068 static inline void f2fs_i_size_write(struct inode *inode, loff_t i_size) 3069 { 3070 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE); 3071 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER); 3072 3073 if (i_size_read(inode) == i_size) 3074 return; 3075 3076 i_size_write(inode, i_size); 3077 3078 if (f2fs_is_atomic_file(inode)) 3079 return; 3080 3081 f2fs_mark_inode_dirty_sync(inode, true); 3082 if (clean || recover) 3083 set_inode_flag(inode, FI_AUTO_RECOVER); 3084 } 3085 3086 static inline void f2fs_i_depth_write(struct inode *inode, unsigned int depth) 3087 { 3088 F2FS_I(inode)->i_current_depth = depth; 3089 f2fs_mark_inode_dirty_sync(inode, true); 3090 } 3091 3092 static inline void f2fs_i_gc_failures_write(struct inode *inode, 3093 unsigned int count) 3094 { 3095 F2FS_I(inode)->i_gc_failures[GC_FAILURE_PIN] = count; 3096 f2fs_mark_inode_dirty_sync(inode, true); 3097 } 3098 3099 static inline void f2fs_i_xnid_write(struct inode *inode, nid_t xnid) 3100 { 3101 F2FS_I(inode)->i_xattr_nid = xnid; 3102 f2fs_mark_inode_dirty_sync(inode, true); 3103 } 3104 3105 static inline void f2fs_i_pino_write(struct inode *inode, nid_t pino) 3106 { 3107 F2FS_I(inode)->i_pino = pino; 3108 f2fs_mark_inode_dirty_sync(inode, true); 3109 } 3110 3111 static inline void get_inline_info(struct inode *inode, struct f2fs_inode *ri) 3112 { 3113 struct f2fs_inode_info *fi = F2FS_I(inode); 3114 3115 if (ri->i_inline & F2FS_INLINE_XATTR) 3116 set_bit(FI_INLINE_XATTR, fi->flags); 3117 if (ri->i_inline & F2FS_INLINE_DATA) 3118 set_bit(FI_INLINE_DATA, fi->flags); 3119 if (ri->i_inline & F2FS_INLINE_DENTRY) 3120 set_bit(FI_INLINE_DENTRY, fi->flags); 3121 if (ri->i_inline & F2FS_DATA_EXIST) 3122 set_bit(FI_DATA_EXIST, fi->flags); 3123 if (ri->i_inline & F2FS_INLINE_DOTS) 3124 set_bit(FI_INLINE_DOTS, fi->flags); 3125 if (ri->i_inline & F2FS_EXTRA_ATTR) 3126 set_bit(FI_EXTRA_ATTR, fi->flags); 3127 if (ri->i_inline & F2FS_PIN_FILE) 3128 set_bit(FI_PIN_FILE, fi->flags); 3129 if (ri->i_inline & F2FS_COMPRESS_RELEASED) 3130 set_bit(FI_COMPRESS_RELEASED, fi->flags); 3131 } 3132 3133 static inline void set_raw_inline(struct inode *inode, struct f2fs_inode *ri) 3134 { 3135 ri->i_inline = 0; 3136 3137 if (is_inode_flag_set(inode, FI_INLINE_XATTR)) 3138 ri->i_inline |= F2FS_INLINE_XATTR; 3139 if (is_inode_flag_set(inode, FI_INLINE_DATA)) 3140 ri->i_inline |= F2FS_INLINE_DATA; 3141 if (is_inode_flag_set(inode, FI_INLINE_DENTRY)) 3142 ri->i_inline |= F2FS_INLINE_DENTRY; 3143 if (is_inode_flag_set(inode, FI_DATA_EXIST)) 3144 ri->i_inline |= F2FS_DATA_EXIST; 3145 if (is_inode_flag_set(inode, FI_INLINE_DOTS)) 3146 ri->i_inline |= F2FS_INLINE_DOTS; 3147 if (is_inode_flag_set(inode, FI_EXTRA_ATTR)) 3148 ri->i_inline |= F2FS_EXTRA_ATTR; 3149 if (is_inode_flag_set(inode, FI_PIN_FILE)) 3150 ri->i_inline |= F2FS_PIN_FILE; 3151 if (is_inode_flag_set(inode, FI_COMPRESS_RELEASED)) 3152 ri->i_inline |= F2FS_COMPRESS_RELEASED; 3153 } 3154 3155 static inline int f2fs_has_extra_attr(struct inode *inode) 3156 { 3157 return is_inode_flag_set(inode, FI_EXTRA_ATTR); 3158 } 3159 3160 static inline int f2fs_has_inline_xattr(struct inode *inode) 3161 { 3162 return is_inode_flag_set(inode, FI_INLINE_XATTR); 3163 } 3164 3165 static inline int f2fs_compressed_file(struct inode *inode) 3166 { 3167 return S_ISREG(inode->i_mode) && 3168 is_inode_flag_set(inode, FI_COMPRESSED_FILE); 3169 } 3170 3171 static inline bool f2fs_need_compress_data(struct inode *inode) 3172 { 3173 int compress_mode = F2FS_OPTION(F2FS_I_SB(inode)).compress_mode; 3174 3175 if (!f2fs_compressed_file(inode)) 3176 return false; 3177 3178 if (compress_mode == COMPR_MODE_FS) 3179 return true; 3180 else if (compress_mode == COMPR_MODE_USER && 3181 is_inode_flag_set(inode, FI_ENABLE_COMPRESS)) 3182 return true; 3183 3184 return false; 3185 } 3186 3187 static inline unsigned int addrs_per_inode(struct inode *inode) 3188 { 3189 unsigned int addrs = CUR_ADDRS_PER_INODE(inode) - 3190 get_inline_xattr_addrs(inode); 3191 3192 if (!f2fs_compressed_file(inode)) 3193 return addrs; 3194 return ALIGN_DOWN(addrs, F2FS_I(inode)->i_cluster_size); 3195 } 3196 3197 static inline unsigned int addrs_per_block(struct inode *inode) 3198 { 3199 if (!f2fs_compressed_file(inode)) 3200 return DEF_ADDRS_PER_BLOCK; 3201 return ALIGN_DOWN(DEF_ADDRS_PER_BLOCK, F2FS_I(inode)->i_cluster_size); 3202 } 3203 3204 static inline void *inline_xattr_addr(struct inode *inode, struct page *page) 3205 { 3206 struct f2fs_inode *ri = F2FS_INODE(page); 3207 3208 return (void *)&(ri->i_addr[DEF_ADDRS_PER_INODE - 3209 get_inline_xattr_addrs(inode)]); 3210 } 3211 3212 static inline int inline_xattr_size(struct inode *inode) 3213 { 3214 if (f2fs_has_inline_xattr(inode)) 3215 return get_inline_xattr_addrs(inode) * sizeof(__le32); 3216 return 0; 3217 } 3218 3219 /* 3220 * Notice: check inline_data flag without inode page lock is unsafe. 3221 * It could change at any time by f2fs_convert_inline_page(). 3222 */ 3223 static inline int f2fs_has_inline_data(struct inode *inode) 3224 { 3225 return is_inode_flag_set(inode, FI_INLINE_DATA); 3226 } 3227 3228 static inline int f2fs_exist_data(struct inode *inode) 3229 { 3230 return is_inode_flag_set(inode, FI_DATA_EXIST); 3231 } 3232 3233 static inline int f2fs_has_inline_dots(struct inode *inode) 3234 { 3235 return is_inode_flag_set(inode, FI_INLINE_DOTS); 3236 } 3237 3238 static inline int f2fs_is_mmap_file(struct inode *inode) 3239 { 3240 return is_inode_flag_set(inode, FI_MMAP_FILE); 3241 } 3242 3243 static inline bool f2fs_is_pinned_file(struct inode *inode) 3244 { 3245 return is_inode_flag_set(inode, FI_PIN_FILE); 3246 } 3247 3248 static inline bool f2fs_is_atomic_file(struct inode *inode) 3249 { 3250 return is_inode_flag_set(inode, FI_ATOMIC_FILE); 3251 } 3252 3253 static inline bool f2fs_is_cow_file(struct inode *inode) 3254 { 3255 return is_inode_flag_set(inode, FI_COW_FILE); 3256 } 3257 3258 static inline bool f2fs_is_first_block_written(struct inode *inode) 3259 { 3260 return is_inode_flag_set(inode, FI_FIRST_BLOCK_WRITTEN); 3261 } 3262 3263 static inline bool f2fs_is_drop_cache(struct inode *inode) 3264 { 3265 return is_inode_flag_set(inode, FI_DROP_CACHE); 3266 } 3267 3268 static inline void *inline_data_addr(struct inode *inode, struct page *page) 3269 { 3270 struct f2fs_inode *ri = F2FS_INODE(page); 3271 int extra_size = get_extra_isize(inode); 3272 3273 return (void *)&(ri->i_addr[extra_size + DEF_INLINE_RESERVED_SIZE]); 3274 } 3275 3276 static inline int f2fs_has_inline_dentry(struct inode *inode) 3277 { 3278 return is_inode_flag_set(inode, FI_INLINE_DENTRY); 3279 } 3280 3281 static inline int is_file(struct inode *inode, int type) 3282 { 3283 return F2FS_I(inode)->i_advise & type; 3284 } 3285 3286 static inline void set_file(struct inode *inode, int type) 3287 { 3288 if (is_file(inode, type)) 3289 return; 3290 F2FS_I(inode)->i_advise |= type; 3291 f2fs_mark_inode_dirty_sync(inode, true); 3292 } 3293 3294 static inline void clear_file(struct inode *inode, int type) 3295 { 3296 if (!is_file(inode, type)) 3297 return; 3298 F2FS_I(inode)->i_advise &= ~type; 3299 f2fs_mark_inode_dirty_sync(inode, true); 3300 } 3301 3302 static inline bool f2fs_is_time_consistent(struct inode *inode) 3303 { 3304 if (!timespec64_equal(F2FS_I(inode)->i_disk_time, &inode->i_atime)) 3305 return false; 3306 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 1, &inode->i_ctime)) 3307 return false; 3308 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 2, &inode->i_mtime)) 3309 return false; 3310 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 3, 3311 &F2FS_I(inode)->i_crtime)) 3312 return false; 3313 return true; 3314 } 3315 3316 static inline bool f2fs_skip_inode_update(struct inode *inode, int dsync) 3317 { 3318 bool ret; 3319 3320 if (dsync) { 3321 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 3322 3323 spin_lock(&sbi->inode_lock[DIRTY_META]); 3324 ret = list_empty(&F2FS_I(inode)->gdirty_list); 3325 spin_unlock(&sbi->inode_lock[DIRTY_META]); 3326 return ret; 3327 } 3328 if (!is_inode_flag_set(inode, FI_AUTO_RECOVER) || 3329 file_keep_isize(inode) || 3330 i_size_read(inode) & ~PAGE_MASK) 3331 return false; 3332 3333 if (!f2fs_is_time_consistent(inode)) 3334 return false; 3335 3336 spin_lock(&F2FS_I(inode)->i_size_lock); 3337 ret = F2FS_I(inode)->last_disk_size == i_size_read(inode); 3338 spin_unlock(&F2FS_I(inode)->i_size_lock); 3339 3340 return ret; 3341 } 3342 3343 static inline bool f2fs_readonly(struct super_block *sb) 3344 { 3345 return sb_rdonly(sb); 3346 } 3347 3348 static inline bool f2fs_cp_error(struct f2fs_sb_info *sbi) 3349 { 3350 return is_set_ckpt_flags(sbi, CP_ERROR_FLAG); 3351 } 3352 3353 static inline bool is_dot_dotdot(const u8 *name, size_t len) 3354 { 3355 if (len == 1 && name[0] == '.') 3356 return true; 3357 3358 if (len == 2 && name[0] == '.' && name[1] == '.') 3359 return true; 3360 3361 return false; 3362 } 3363 3364 static inline void *f2fs_kmalloc(struct f2fs_sb_info *sbi, 3365 size_t size, gfp_t flags) 3366 { 3367 if (time_to_inject(sbi, FAULT_KMALLOC)) 3368 return NULL; 3369 3370 return kmalloc(size, flags); 3371 } 3372 3373 static inline void *f2fs_kzalloc(struct f2fs_sb_info *sbi, 3374 size_t size, gfp_t flags) 3375 { 3376 return f2fs_kmalloc(sbi, size, flags | __GFP_ZERO); 3377 } 3378 3379 static inline void *f2fs_kvmalloc(struct f2fs_sb_info *sbi, 3380 size_t size, gfp_t flags) 3381 { 3382 if (time_to_inject(sbi, FAULT_KVMALLOC)) 3383 return NULL; 3384 3385 return kvmalloc(size, flags); 3386 } 3387 3388 static inline void *f2fs_kvzalloc(struct f2fs_sb_info *sbi, 3389 size_t size, gfp_t flags) 3390 { 3391 return f2fs_kvmalloc(sbi, size, flags | __GFP_ZERO); 3392 } 3393 3394 static inline int get_extra_isize(struct inode *inode) 3395 { 3396 return F2FS_I(inode)->i_extra_isize / sizeof(__le32); 3397 } 3398 3399 static inline int get_inline_xattr_addrs(struct inode *inode) 3400 { 3401 return F2FS_I(inode)->i_inline_xattr_size; 3402 } 3403 3404 #define f2fs_get_inode_mode(i) \ 3405 ((is_inode_flag_set(i, FI_ACL_MODE)) ? \ 3406 (F2FS_I(i)->i_acl_mode) : ((i)->i_mode)) 3407 3408 #define F2FS_TOTAL_EXTRA_ATTR_SIZE \ 3409 (offsetof(struct f2fs_inode, i_extra_end) - \ 3410 offsetof(struct f2fs_inode, i_extra_isize)) \ 3411 3412 #define F2FS_OLD_ATTRIBUTE_SIZE (offsetof(struct f2fs_inode, i_addr)) 3413 #define F2FS_FITS_IN_INODE(f2fs_inode, extra_isize, field) \ 3414 ((offsetof(typeof(*(f2fs_inode)), field) + \ 3415 sizeof((f2fs_inode)->field)) \ 3416 <= (F2FS_OLD_ATTRIBUTE_SIZE + (extra_isize))) \ 3417 3418 #define __is_large_section(sbi) ((sbi)->segs_per_sec > 1) 3419 3420 #define __is_meta_io(fio) (PAGE_TYPE_OF_BIO((fio)->type) == META) 3421 3422 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi, 3423 block_t blkaddr, int type); 3424 static inline void verify_blkaddr(struct f2fs_sb_info *sbi, 3425 block_t blkaddr, int type) 3426 { 3427 if (!f2fs_is_valid_blkaddr(sbi, blkaddr, type)) { 3428 f2fs_err(sbi, "invalid blkaddr: %u, type: %d, run fsck to fix.", 3429 blkaddr, type); 3430 f2fs_bug_on(sbi, 1); 3431 } 3432 } 3433 3434 static inline bool __is_valid_data_blkaddr(block_t blkaddr) 3435 { 3436 if (blkaddr == NEW_ADDR || blkaddr == NULL_ADDR || 3437 blkaddr == COMPRESS_ADDR) 3438 return false; 3439 return true; 3440 } 3441 3442 /* 3443 * file.c 3444 */ 3445 int f2fs_sync_file(struct file *file, loff_t start, loff_t end, int datasync); 3446 void f2fs_truncate_data_blocks(struct dnode_of_data *dn); 3447 int f2fs_do_truncate_blocks(struct inode *inode, u64 from, bool lock); 3448 int f2fs_truncate_blocks(struct inode *inode, u64 from, bool lock); 3449 int f2fs_truncate(struct inode *inode); 3450 int f2fs_getattr(struct mnt_idmap *idmap, const struct path *path, 3451 struct kstat *stat, u32 request_mask, unsigned int flags); 3452 int f2fs_setattr(struct mnt_idmap *idmap, struct dentry *dentry, 3453 struct iattr *attr); 3454 int f2fs_truncate_hole(struct inode *inode, pgoff_t pg_start, pgoff_t pg_end); 3455 void f2fs_truncate_data_blocks_range(struct dnode_of_data *dn, int count); 3456 int f2fs_precache_extents(struct inode *inode); 3457 int f2fs_fileattr_get(struct dentry *dentry, struct fileattr *fa); 3458 int f2fs_fileattr_set(struct mnt_idmap *idmap, 3459 struct dentry *dentry, struct fileattr *fa); 3460 long f2fs_ioctl(struct file *filp, unsigned int cmd, unsigned long arg); 3461 long f2fs_compat_ioctl(struct file *file, unsigned int cmd, unsigned long arg); 3462 int f2fs_transfer_project_quota(struct inode *inode, kprojid_t kprojid); 3463 int f2fs_pin_file_control(struct inode *inode, bool inc); 3464 3465 /* 3466 * inode.c 3467 */ 3468 void f2fs_set_inode_flags(struct inode *inode); 3469 bool f2fs_inode_chksum_verify(struct f2fs_sb_info *sbi, struct page *page); 3470 void f2fs_inode_chksum_set(struct f2fs_sb_info *sbi, struct page *page); 3471 struct inode *f2fs_iget(struct super_block *sb, unsigned long ino); 3472 struct inode *f2fs_iget_retry(struct super_block *sb, unsigned long ino); 3473 int f2fs_try_to_free_nats(struct f2fs_sb_info *sbi, int nr_shrink); 3474 void f2fs_update_inode(struct inode *inode, struct page *node_page); 3475 void f2fs_update_inode_page(struct inode *inode); 3476 int f2fs_write_inode(struct inode *inode, struct writeback_control *wbc); 3477 void f2fs_evict_inode(struct inode *inode); 3478 void f2fs_handle_failed_inode(struct inode *inode); 3479 3480 /* 3481 * namei.c 3482 */ 3483 int f2fs_update_extension_list(struct f2fs_sb_info *sbi, const char *name, 3484 bool hot, bool set); 3485 struct dentry *f2fs_get_parent(struct dentry *child); 3486 int f2fs_get_tmpfile(struct mnt_idmap *idmap, struct inode *dir, 3487 struct inode **new_inode); 3488 3489 /* 3490 * dir.c 3491 */ 3492 unsigned char f2fs_get_de_type(struct f2fs_dir_entry *de); 3493 int f2fs_init_casefolded_name(const struct inode *dir, 3494 struct f2fs_filename *fname); 3495 int f2fs_setup_filename(struct inode *dir, const struct qstr *iname, 3496 int lookup, struct f2fs_filename *fname); 3497 int f2fs_prepare_lookup(struct inode *dir, struct dentry *dentry, 3498 struct f2fs_filename *fname); 3499 void f2fs_free_filename(struct f2fs_filename *fname); 3500 struct f2fs_dir_entry *f2fs_find_target_dentry(const struct f2fs_dentry_ptr *d, 3501 const struct f2fs_filename *fname, int *max_slots); 3502 int f2fs_fill_dentries(struct dir_context *ctx, struct f2fs_dentry_ptr *d, 3503 unsigned int start_pos, struct fscrypt_str *fstr); 3504 void f2fs_do_make_empty_dir(struct inode *inode, struct inode *parent, 3505 struct f2fs_dentry_ptr *d); 3506 struct page *f2fs_init_inode_metadata(struct inode *inode, struct inode *dir, 3507 const struct f2fs_filename *fname, struct page *dpage); 3508 void f2fs_update_parent_metadata(struct inode *dir, struct inode *inode, 3509 unsigned int current_depth); 3510 int f2fs_room_for_filename(const void *bitmap, int slots, int max_slots); 3511 void f2fs_drop_nlink(struct inode *dir, struct inode *inode); 3512 struct f2fs_dir_entry *__f2fs_find_entry(struct inode *dir, 3513 const struct f2fs_filename *fname, 3514 struct page **res_page); 3515 struct f2fs_dir_entry *f2fs_find_entry(struct inode *dir, 3516 const struct qstr *child, struct page **res_page); 3517 struct f2fs_dir_entry *f2fs_parent_dir(struct inode *dir, struct page **p); 3518 ino_t f2fs_inode_by_name(struct inode *dir, const struct qstr *qstr, 3519 struct page **page); 3520 void f2fs_set_link(struct inode *dir, struct f2fs_dir_entry *de, 3521 struct page *page, struct inode *inode); 3522 bool f2fs_has_enough_room(struct inode *dir, struct page *ipage, 3523 const struct f2fs_filename *fname); 3524 void f2fs_update_dentry(nid_t ino, umode_t mode, struct f2fs_dentry_ptr *d, 3525 const struct fscrypt_str *name, f2fs_hash_t name_hash, 3526 unsigned int bit_pos); 3527 int f2fs_add_regular_entry(struct inode *dir, const struct f2fs_filename *fname, 3528 struct inode *inode, nid_t ino, umode_t mode); 3529 int f2fs_add_dentry(struct inode *dir, const struct f2fs_filename *fname, 3530 struct inode *inode, nid_t ino, umode_t mode); 3531 int f2fs_do_add_link(struct inode *dir, const struct qstr *name, 3532 struct inode *inode, nid_t ino, umode_t mode); 3533 void f2fs_delete_entry(struct f2fs_dir_entry *dentry, struct page *page, 3534 struct inode *dir, struct inode *inode); 3535 int f2fs_do_tmpfile(struct inode *inode, struct inode *dir); 3536 bool f2fs_empty_dir(struct inode *dir); 3537 3538 static inline int f2fs_add_link(struct dentry *dentry, struct inode *inode) 3539 { 3540 if (fscrypt_is_nokey_name(dentry)) 3541 return -ENOKEY; 3542 return f2fs_do_add_link(d_inode(dentry->d_parent), &dentry->d_name, 3543 inode, inode->i_ino, inode->i_mode); 3544 } 3545 3546 /* 3547 * super.c 3548 */ 3549 int f2fs_inode_dirtied(struct inode *inode, bool sync); 3550 void f2fs_inode_synced(struct inode *inode); 3551 int f2fs_dquot_initialize(struct inode *inode); 3552 int f2fs_enable_quota_files(struct f2fs_sb_info *sbi, bool rdonly); 3553 int f2fs_quota_sync(struct super_block *sb, int type); 3554 loff_t max_file_blocks(struct inode *inode); 3555 void f2fs_quota_off_umount(struct super_block *sb); 3556 void f2fs_handle_stop(struct f2fs_sb_info *sbi, unsigned char reason); 3557 void f2fs_handle_error(struct f2fs_sb_info *sbi, unsigned char error); 3558 int f2fs_commit_super(struct f2fs_sb_info *sbi, bool recover); 3559 int f2fs_sync_fs(struct super_block *sb, int sync); 3560 int f2fs_sanity_check_ckpt(struct f2fs_sb_info *sbi); 3561 3562 /* 3563 * hash.c 3564 */ 3565 void f2fs_hash_filename(const struct inode *dir, struct f2fs_filename *fname); 3566 3567 /* 3568 * node.c 3569 */ 3570 struct node_info; 3571 3572 int f2fs_check_nid_range(struct f2fs_sb_info *sbi, nid_t nid); 3573 bool f2fs_available_free_memory(struct f2fs_sb_info *sbi, int type); 3574 bool f2fs_in_warm_node_list(struct f2fs_sb_info *sbi, struct page *page); 3575 void f2fs_init_fsync_node_info(struct f2fs_sb_info *sbi); 3576 void f2fs_del_fsync_node_entry(struct f2fs_sb_info *sbi, struct page *page); 3577 void f2fs_reset_fsync_node_info(struct f2fs_sb_info *sbi); 3578 int f2fs_need_dentry_mark(struct f2fs_sb_info *sbi, nid_t nid); 3579 bool f2fs_is_checkpointed_node(struct f2fs_sb_info *sbi, nid_t nid); 3580 bool f2fs_need_inode_block_update(struct f2fs_sb_info *sbi, nid_t ino); 3581 int f2fs_get_node_info(struct f2fs_sb_info *sbi, nid_t nid, 3582 struct node_info *ni, bool checkpoint_context); 3583 pgoff_t f2fs_get_next_page_offset(struct dnode_of_data *dn, pgoff_t pgofs); 3584 int f2fs_get_dnode_of_data(struct dnode_of_data *dn, pgoff_t index, int mode); 3585 int f2fs_truncate_inode_blocks(struct inode *inode, pgoff_t from); 3586 int f2fs_truncate_xattr_node(struct inode *inode); 3587 int f2fs_wait_on_node_pages_writeback(struct f2fs_sb_info *sbi, 3588 unsigned int seq_id); 3589 bool f2fs_nat_bitmap_enabled(struct f2fs_sb_info *sbi); 3590 int f2fs_remove_inode_page(struct inode *inode); 3591 struct page *f2fs_new_inode_page(struct inode *inode); 3592 struct page *f2fs_new_node_page(struct dnode_of_data *dn, unsigned int ofs); 3593 void f2fs_ra_node_page(struct f2fs_sb_info *sbi, nid_t nid); 3594 struct page *f2fs_get_node_page(struct f2fs_sb_info *sbi, pgoff_t nid); 3595 struct page *f2fs_get_node_page_ra(struct page *parent, int start); 3596 int f2fs_move_node_page(struct page *node_page, int gc_type); 3597 void f2fs_flush_inline_data(struct f2fs_sb_info *sbi); 3598 int f2fs_fsync_node_pages(struct f2fs_sb_info *sbi, struct inode *inode, 3599 struct writeback_control *wbc, bool atomic, 3600 unsigned int *seq_id); 3601 int f2fs_sync_node_pages(struct f2fs_sb_info *sbi, 3602 struct writeback_control *wbc, 3603 bool do_balance, enum iostat_type io_type); 3604 int f2fs_build_free_nids(struct f2fs_sb_info *sbi, bool sync, bool mount); 3605 bool f2fs_alloc_nid(struct f2fs_sb_info *sbi, nid_t *nid); 3606 void f2fs_alloc_nid_done(struct f2fs_sb_info *sbi, nid_t nid); 3607 void f2fs_alloc_nid_failed(struct f2fs_sb_info *sbi, nid_t nid); 3608 int f2fs_try_to_free_nids(struct f2fs_sb_info *sbi, int nr_shrink); 3609 int f2fs_recover_inline_xattr(struct inode *inode, struct page *page); 3610 int f2fs_recover_xattr_data(struct inode *inode, struct page *page); 3611 int f2fs_recover_inode_page(struct f2fs_sb_info *sbi, struct page *page); 3612 int f2fs_restore_node_summary(struct f2fs_sb_info *sbi, 3613 unsigned int segno, struct f2fs_summary_block *sum); 3614 void f2fs_enable_nat_bits(struct f2fs_sb_info *sbi); 3615 int f2fs_flush_nat_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3616 int f2fs_build_node_manager(struct f2fs_sb_info *sbi); 3617 void f2fs_destroy_node_manager(struct f2fs_sb_info *sbi); 3618 int __init f2fs_create_node_manager_caches(void); 3619 void f2fs_destroy_node_manager_caches(void); 3620 3621 /* 3622 * segment.c 3623 */ 3624 bool f2fs_need_SSR(struct f2fs_sb_info *sbi); 3625 int f2fs_commit_atomic_write(struct inode *inode); 3626 void f2fs_abort_atomic_write(struct inode *inode, bool clean); 3627 void f2fs_balance_fs(struct f2fs_sb_info *sbi, bool need); 3628 void f2fs_balance_fs_bg(struct f2fs_sb_info *sbi, bool from_bg); 3629 int f2fs_issue_flush(struct f2fs_sb_info *sbi, nid_t ino); 3630 int f2fs_create_flush_cmd_control(struct f2fs_sb_info *sbi); 3631 int f2fs_flush_device_cache(struct f2fs_sb_info *sbi); 3632 void f2fs_destroy_flush_cmd_control(struct f2fs_sb_info *sbi, bool free); 3633 void f2fs_invalidate_blocks(struct f2fs_sb_info *sbi, block_t addr); 3634 bool f2fs_is_checkpointed_data(struct f2fs_sb_info *sbi, block_t blkaddr); 3635 int f2fs_start_discard_thread(struct f2fs_sb_info *sbi); 3636 void f2fs_drop_discard_cmd(struct f2fs_sb_info *sbi); 3637 void f2fs_stop_discard_thread(struct f2fs_sb_info *sbi); 3638 bool f2fs_issue_discard_timeout(struct f2fs_sb_info *sbi); 3639 void f2fs_clear_prefree_segments(struct f2fs_sb_info *sbi, 3640 struct cp_control *cpc); 3641 void f2fs_dirty_to_prefree(struct f2fs_sb_info *sbi); 3642 block_t f2fs_get_unusable_blocks(struct f2fs_sb_info *sbi); 3643 int f2fs_disable_cp_again(struct f2fs_sb_info *sbi, block_t unusable); 3644 void f2fs_release_discard_addrs(struct f2fs_sb_info *sbi); 3645 int f2fs_npages_for_summary_flush(struct f2fs_sb_info *sbi, bool for_ra); 3646 bool f2fs_segment_has_free_slot(struct f2fs_sb_info *sbi, int segno); 3647 void f2fs_init_inmem_curseg(struct f2fs_sb_info *sbi); 3648 void f2fs_save_inmem_curseg(struct f2fs_sb_info *sbi); 3649 void f2fs_restore_inmem_curseg(struct f2fs_sb_info *sbi); 3650 void f2fs_get_new_segment(struct f2fs_sb_info *sbi, 3651 unsigned int *newseg, bool new_sec, int dir); 3652 void f2fs_allocate_segment_for_resize(struct f2fs_sb_info *sbi, int type, 3653 unsigned int start, unsigned int end); 3654 void f2fs_allocate_new_section(struct f2fs_sb_info *sbi, int type, bool force); 3655 void f2fs_allocate_new_segments(struct f2fs_sb_info *sbi); 3656 int f2fs_trim_fs(struct f2fs_sb_info *sbi, struct fstrim_range *range); 3657 bool f2fs_exist_trim_candidates(struct f2fs_sb_info *sbi, 3658 struct cp_control *cpc); 3659 struct page *f2fs_get_sum_page(struct f2fs_sb_info *sbi, unsigned int segno); 3660 void f2fs_update_meta_page(struct f2fs_sb_info *sbi, void *src, 3661 block_t blk_addr); 3662 void f2fs_do_write_meta_page(struct f2fs_sb_info *sbi, struct page *page, 3663 enum iostat_type io_type); 3664 void f2fs_do_write_node_page(unsigned int nid, struct f2fs_io_info *fio); 3665 void f2fs_outplace_write_data(struct dnode_of_data *dn, 3666 struct f2fs_io_info *fio); 3667 int f2fs_inplace_write_data(struct f2fs_io_info *fio); 3668 void f2fs_do_replace_block(struct f2fs_sb_info *sbi, struct f2fs_summary *sum, 3669 block_t old_blkaddr, block_t new_blkaddr, 3670 bool recover_curseg, bool recover_newaddr, 3671 bool from_gc); 3672 void f2fs_replace_block(struct f2fs_sb_info *sbi, struct dnode_of_data *dn, 3673 block_t old_addr, block_t new_addr, 3674 unsigned char version, bool recover_curseg, 3675 bool recover_newaddr); 3676 void f2fs_allocate_data_block(struct f2fs_sb_info *sbi, struct page *page, 3677 block_t old_blkaddr, block_t *new_blkaddr, 3678 struct f2fs_summary *sum, int type, 3679 struct f2fs_io_info *fio); 3680 void f2fs_update_device_state(struct f2fs_sb_info *sbi, nid_t ino, 3681 block_t blkaddr, unsigned int blkcnt); 3682 void f2fs_wait_on_page_writeback(struct page *page, 3683 enum page_type type, bool ordered, bool locked); 3684 void f2fs_wait_on_block_writeback(struct inode *inode, block_t blkaddr); 3685 void f2fs_wait_on_block_writeback_range(struct inode *inode, block_t blkaddr, 3686 block_t len); 3687 void f2fs_write_data_summaries(struct f2fs_sb_info *sbi, block_t start_blk); 3688 void f2fs_write_node_summaries(struct f2fs_sb_info *sbi, block_t start_blk); 3689 int f2fs_lookup_journal_in_cursum(struct f2fs_journal *journal, int type, 3690 unsigned int val, int alloc); 3691 void f2fs_flush_sit_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3692 int f2fs_fix_curseg_write_pointer(struct f2fs_sb_info *sbi); 3693 int f2fs_check_write_pointer(struct f2fs_sb_info *sbi); 3694 int f2fs_build_segment_manager(struct f2fs_sb_info *sbi); 3695 void f2fs_destroy_segment_manager(struct f2fs_sb_info *sbi); 3696 int __init f2fs_create_segment_manager_caches(void); 3697 void f2fs_destroy_segment_manager_caches(void); 3698 int f2fs_rw_hint_to_seg_type(enum rw_hint hint); 3699 unsigned int f2fs_usable_segs_in_sec(struct f2fs_sb_info *sbi, 3700 unsigned int segno); 3701 unsigned int f2fs_usable_blks_in_seg(struct f2fs_sb_info *sbi, 3702 unsigned int segno); 3703 3704 #define DEF_FRAGMENT_SIZE 4 3705 #define MIN_FRAGMENT_SIZE 1 3706 #define MAX_FRAGMENT_SIZE 512 3707 3708 static inline bool f2fs_need_rand_seg(struct f2fs_sb_info *sbi) 3709 { 3710 return F2FS_OPTION(sbi).fs_mode == FS_MODE_FRAGMENT_SEG || 3711 F2FS_OPTION(sbi).fs_mode == FS_MODE_FRAGMENT_BLK; 3712 } 3713 3714 /* 3715 * checkpoint.c 3716 */ 3717 void f2fs_stop_checkpoint(struct f2fs_sb_info *sbi, bool end_io, 3718 unsigned char reason); 3719 void f2fs_flush_ckpt_thread(struct f2fs_sb_info *sbi); 3720 struct page *f2fs_grab_meta_page(struct f2fs_sb_info *sbi, pgoff_t index); 3721 struct page *f2fs_get_meta_page(struct f2fs_sb_info *sbi, pgoff_t index); 3722 struct page *f2fs_get_meta_page_retry(struct f2fs_sb_info *sbi, pgoff_t index); 3723 struct page *f2fs_get_tmp_page(struct f2fs_sb_info *sbi, pgoff_t index); 3724 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi, 3725 block_t blkaddr, int type); 3726 int f2fs_ra_meta_pages(struct f2fs_sb_info *sbi, block_t start, int nrpages, 3727 int type, bool sync); 3728 void f2fs_ra_meta_pages_cond(struct f2fs_sb_info *sbi, pgoff_t index, 3729 unsigned int ra_blocks); 3730 long f2fs_sync_meta_pages(struct f2fs_sb_info *sbi, enum page_type type, 3731 long nr_to_write, enum iostat_type io_type); 3732 void f2fs_add_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type); 3733 void f2fs_remove_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type); 3734 void f2fs_release_ino_entry(struct f2fs_sb_info *sbi, bool all); 3735 bool f2fs_exist_written_data(struct f2fs_sb_info *sbi, nid_t ino, int mode); 3736 void f2fs_set_dirty_device(struct f2fs_sb_info *sbi, nid_t ino, 3737 unsigned int devidx, int type); 3738 bool f2fs_is_dirty_device(struct f2fs_sb_info *sbi, nid_t ino, 3739 unsigned int devidx, int type); 3740 int f2fs_sync_inode_meta(struct f2fs_sb_info *sbi); 3741 int f2fs_acquire_orphan_inode(struct f2fs_sb_info *sbi); 3742 void f2fs_release_orphan_inode(struct f2fs_sb_info *sbi); 3743 void f2fs_add_orphan_inode(struct inode *inode); 3744 void f2fs_remove_orphan_inode(struct f2fs_sb_info *sbi, nid_t ino); 3745 int f2fs_recover_orphan_inodes(struct f2fs_sb_info *sbi); 3746 int f2fs_get_valid_checkpoint(struct f2fs_sb_info *sbi); 3747 void f2fs_update_dirty_folio(struct inode *inode, struct folio *folio); 3748 void f2fs_remove_dirty_inode(struct inode *inode); 3749 int f2fs_sync_dirty_inodes(struct f2fs_sb_info *sbi, enum inode_type type, 3750 bool from_cp); 3751 void f2fs_wait_on_all_pages(struct f2fs_sb_info *sbi, int type); 3752 u64 f2fs_get_sectors_written(struct f2fs_sb_info *sbi); 3753 int f2fs_write_checkpoint(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3754 void f2fs_init_ino_entry_info(struct f2fs_sb_info *sbi); 3755 int __init f2fs_create_checkpoint_caches(void); 3756 void f2fs_destroy_checkpoint_caches(void); 3757 int f2fs_issue_checkpoint(struct f2fs_sb_info *sbi); 3758 int f2fs_start_ckpt_thread(struct f2fs_sb_info *sbi); 3759 void f2fs_stop_ckpt_thread(struct f2fs_sb_info *sbi); 3760 void f2fs_init_ckpt_req_control(struct f2fs_sb_info *sbi); 3761 3762 /* 3763 * data.c 3764 */ 3765 int __init f2fs_init_bioset(void); 3766 void f2fs_destroy_bioset(void); 3767 int f2fs_init_bio_entry_cache(void); 3768 void f2fs_destroy_bio_entry_cache(void); 3769 void f2fs_submit_read_bio(struct f2fs_sb_info *sbi, struct bio *bio, 3770 enum page_type type); 3771 int f2fs_init_write_merge_io(struct f2fs_sb_info *sbi); 3772 void f2fs_submit_merged_write(struct f2fs_sb_info *sbi, enum page_type type); 3773 void f2fs_submit_merged_write_cond(struct f2fs_sb_info *sbi, 3774 struct inode *inode, struct page *page, 3775 nid_t ino, enum page_type type); 3776 void f2fs_submit_merged_ipu_write(struct f2fs_sb_info *sbi, 3777 struct bio **bio, struct page *page); 3778 void f2fs_flush_merged_writes(struct f2fs_sb_info *sbi); 3779 int f2fs_submit_page_bio(struct f2fs_io_info *fio); 3780 int f2fs_merge_page_bio(struct f2fs_io_info *fio); 3781 void f2fs_submit_page_write(struct f2fs_io_info *fio); 3782 struct block_device *f2fs_target_device(struct f2fs_sb_info *sbi, 3783 block_t blk_addr, sector_t *sector); 3784 int f2fs_target_device_index(struct f2fs_sb_info *sbi, block_t blkaddr); 3785 void f2fs_set_data_blkaddr(struct dnode_of_data *dn); 3786 void f2fs_update_data_blkaddr(struct dnode_of_data *dn, block_t blkaddr); 3787 int f2fs_reserve_new_blocks(struct dnode_of_data *dn, blkcnt_t count); 3788 int f2fs_reserve_new_block(struct dnode_of_data *dn); 3789 int f2fs_get_block_locked(struct dnode_of_data *dn, pgoff_t index); 3790 int f2fs_reserve_block(struct dnode_of_data *dn, pgoff_t index); 3791 struct page *f2fs_get_read_data_page(struct inode *inode, pgoff_t index, 3792 blk_opf_t op_flags, bool for_write, pgoff_t *next_pgofs); 3793 struct page *f2fs_find_data_page(struct inode *inode, pgoff_t index, 3794 pgoff_t *next_pgofs); 3795 struct page *f2fs_get_lock_data_page(struct inode *inode, pgoff_t index, 3796 bool for_write); 3797 struct page *f2fs_get_new_data_page(struct inode *inode, 3798 struct page *ipage, pgoff_t index, bool new_i_size); 3799 int f2fs_do_write_data_page(struct f2fs_io_info *fio); 3800 int f2fs_map_blocks(struct inode *inode, struct f2fs_map_blocks *map, int flag); 3801 int f2fs_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, 3802 u64 start, u64 len); 3803 int f2fs_encrypt_one_page(struct f2fs_io_info *fio); 3804 bool f2fs_should_update_inplace(struct inode *inode, struct f2fs_io_info *fio); 3805 bool f2fs_should_update_outplace(struct inode *inode, struct f2fs_io_info *fio); 3806 int f2fs_write_single_data_page(struct page *page, int *submitted, 3807 struct bio **bio, sector_t *last_block, 3808 struct writeback_control *wbc, 3809 enum iostat_type io_type, 3810 int compr_blocks, bool allow_balance); 3811 void f2fs_write_failed(struct inode *inode, loff_t to); 3812 void f2fs_invalidate_folio(struct folio *folio, size_t offset, size_t length); 3813 bool f2fs_release_folio(struct folio *folio, gfp_t wait); 3814 bool f2fs_overwrite_io(struct inode *inode, loff_t pos, size_t len); 3815 void f2fs_clear_page_cache_dirty_tag(struct page *page); 3816 int f2fs_init_post_read_processing(void); 3817 void f2fs_destroy_post_read_processing(void); 3818 int f2fs_init_post_read_wq(struct f2fs_sb_info *sbi); 3819 void f2fs_destroy_post_read_wq(struct f2fs_sb_info *sbi); 3820 extern const struct iomap_ops f2fs_iomap_ops; 3821 3822 /* 3823 * gc.c 3824 */ 3825 int f2fs_start_gc_thread(struct f2fs_sb_info *sbi); 3826 void f2fs_stop_gc_thread(struct f2fs_sb_info *sbi); 3827 block_t f2fs_start_bidx_of_node(unsigned int node_ofs, struct inode *inode); 3828 int f2fs_gc(struct f2fs_sb_info *sbi, struct f2fs_gc_control *gc_control); 3829 void f2fs_build_gc_manager(struct f2fs_sb_info *sbi); 3830 int f2fs_resize_fs(struct f2fs_sb_info *sbi, __u64 block_count); 3831 int __init f2fs_create_garbage_collection_cache(void); 3832 void f2fs_destroy_garbage_collection_cache(void); 3833 3834 /* 3835 * recovery.c 3836 */ 3837 int f2fs_recover_fsync_data(struct f2fs_sb_info *sbi, bool check_only); 3838 bool f2fs_space_for_roll_forward(struct f2fs_sb_info *sbi); 3839 int __init f2fs_create_recovery_cache(void); 3840 void f2fs_destroy_recovery_cache(void); 3841 3842 /* 3843 * debug.c 3844 */ 3845 #ifdef CONFIG_F2FS_STAT_FS 3846 struct f2fs_stat_info { 3847 struct list_head stat_list; 3848 struct f2fs_sb_info *sbi; 3849 int all_area_segs, sit_area_segs, nat_area_segs, ssa_area_segs; 3850 int main_area_segs, main_area_sections, main_area_zones; 3851 unsigned long long hit_cached[NR_EXTENT_CACHES]; 3852 unsigned long long hit_rbtree[NR_EXTENT_CACHES]; 3853 unsigned long long total_ext[NR_EXTENT_CACHES]; 3854 unsigned long long hit_total[NR_EXTENT_CACHES]; 3855 int ext_tree[NR_EXTENT_CACHES]; 3856 int zombie_tree[NR_EXTENT_CACHES]; 3857 int ext_node[NR_EXTENT_CACHES]; 3858 /* to count memory footprint */ 3859 unsigned long long ext_mem[NR_EXTENT_CACHES]; 3860 /* for read extent cache */ 3861 unsigned long long hit_largest; 3862 /* for block age extent cache */ 3863 unsigned long long allocated_data_blocks; 3864 int ndirty_node, ndirty_dent, ndirty_meta, ndirty_imeta; 3865 int ndirty_data, ndirty_qdata; 3866 unsigned int ndirty_dirs, ndirty_files, nquota_files, ndirty_all; 3867 int nats, dirty_nats, sits, dirty_sits; 3868 int free_nids, avail_nids, alloc_nids; 3869 int total_count, utilization; 3870 int bg_gc, nr_wb_cp_data, nr_wb_data; 3871 int nr_rd_data, nr_rd_node, nr_rd_meta; 3872 int nr_dio_read, nr_dio_write; 3873 unsigned int io_skip_bggc, other_skip_bggc; 3874 int nr_flushing, nr_flushed, flush_list_empty; 3875 int nr_discarding, nr_discarded; 3876 int nr_discard_cmd; 3877 unsigned int undiscard_blks; 3878 int nr_issued_ckpt, nr_total_ckpt, nr_queued_ckpt; 3879 unsigned int cur_ckpt_time, peak_ckpt_time; 3880 int inline_xattr, inline_inode, inline_dir, append, update, orphans; 3881 int compr_inode, swapfile_inode; 3882 unsigned long long compr_blocks; 3883 int aw_cnt, max_aw_cnt; 3884 unsigned int valid_count, valid_node_count, valid_inode_count, discard_blks; 3885 unsigned int bimodal, avg_vblocks; 3886 int util_free, util_valid, util_invalid; 3887 int rsvd_segs, overp_segs; 3888 int dirty_count, node_pages, meta_pages, compress_pages; 3889 int compress_page_hit; 3890 int prefree_count, call_count, cp_count, bg_cp_count; 3891 int tot_segs, node_segs, data_segs, free_segs, free_secs; 3892 int bg_node_segs, bg_data_segs; 3893 int tot_blks, data_blks, node_blks; 3894 int bg_data_blks, bg_node_blks; 3895 int curseg[NR_CURSEG_TYPE]; 3896 int cursec[NR_CURSEG_TYPE]; 3897 int curzone[NR_CURSEG_TYPE]; 3898 unsigned int dirty_seg[NR_CURSEG_TYPE]; 3899 unsigned int full_seg[NR_CURSEG_TYPE]; 3900 unsigned int valid_blks[NR_CURSEG_TYPE]; 3901 3902 unsigned int meta_count[META_MAX]; 3903 unsigned int segment_count[2]; 3904 unsigned int block_count[2]; 3905 unsigned int inplace_count; 3906 unsigned long long base_mem, cache_mem, page_mem; 3907 }; 3908 3909 static inline struct f2fs_stat_info *F2FS_STAT(struct f2fs_sb_info *sbi) 3910 { 3911 return (struct f2fs_stat_info *)sbi->stat_info; 3912 } 3913 3914 #define stat_inc_cp_count(si) ((si)->cp_count++) 3915 #define stat_inc_bg_cp_count(si) ((si)->bg_cp_count++) 3916 #define stat_inc_call_count(si) ((si)->call_count++) 3917 #define stat_inc_bggc_count(si) ((si)->bg_gc++) 3918 #define stat_io_skip_bggc_count(sbi) ((sbi)->io_skip_bggc++) 3919 #define stat_other_skip_bggc_count(sbi) ((sbi)->other_skip_bggc++) 3920 #define stat_inc_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]++) 3921 #define stat_dec_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]--) 3922 #define stat_inc_total_hit(sbi, type) (atomic64_inc(&(sbi)->total_hit_ext[type])) 3923 #define stat_inc_rbtree_node_hit(sbi, type) (atomic64_inc(&(sbi)->read_hit_rbtree[type])) 3924 #define stat_inc_largest_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_largest)) 3925 #define stat_inc_cached_node_hit(sbi, type) (atomic64_inc(&(sbi)->read_hit_cached[type])) 3926 #define stat_inc_inline_xattr(inode) \ 3927 do { \ 3928 if (f2fs_has_inline_xattr(inode)) \ 3929 (atomic_inc(&F2FS_I_SB(inode)->inline_xattr)); \ 3930 } while (0) 3931 #define stat_dec_inline_xattr(inode) \ 3932 do { \ 3933 if (f2fs_has_inline_xattr(inode)) \ 3934 (atomic_dec(&F2FS_I_SB(inode)->inline_xattr)); \ 3935 } while (0) 3936 #define stat_inc_inline_inode(inode) \ 3937 do { \ 3938 if (f2fs_has_inline_data(inode)) \ 3939 (atomic_inc(&F2FS_I_SB(inode)->inline_inode)); \ 3940 } while (0) 3941 #define stat_dec_inline_inode(inode) \ 3942 do { \ 3943 if (f2fs_has_inline_data(inode)) \ 3944 (atomic_dec(&F2FS_I_SB(inode)->inline_inode)); \ 3945 } while (0) 3946 #define stat_inc_inline_dir(inode) \ 3947 do { \ 3948 if (f2fs_has_inline_dentry(inode)) \ 3949 (atomic_inc(&F2FS_I_SB(inode)->inline_dir)); \ 3950 } while (0) 3951 #define stat_dec_inline_dir(inode) \ 3952 do { \ 3953 if (f2fs_has_inline_dentry(inode)) \ 3954 (atomic_dec(&F2FS_I_SB(inode)->inline_dir)); \ 3955 } while (0) 3956 #define stat_inc_compr_inode(inode) \ 3957 do { \ 3958 if (f2fs_compressed_file(inode)) \ 3959 (atomic_inc(&F2FS_I_SB(inode)->compr_inode)); \ 3960 } while (0) 3961 #define stat_dec_compr_inode(inode) \ 3962 do { \ 3963 if (f2fs_compressed_file(inode)) \ 3964 (atomic_dec(&F2FS_I_SB(inode)->compr_inode)); \ 3965 } while (0) 3966 #define stat_add_compr_blocks(inode, blocks) \ 3967 (atomic64_add(blocks, &F2FS_I_SB(inode)->compr_blocks)) 3968 #define stat_sub_compr_blocks(inode, blocks) \ 3969 (atomic64_sub(blocks, &F2FS_I_SB(inode)->compr_blocks)) 3970 #define stat_inc_swapfile_inode(inode) \ 3971 (atomic_inc(&F2FS_I_SB(inode)->swapfile_inode)) 3972 #define stat_dec_swapfile_inode(inode) \ 3973 (atomic_dec(&F2FS_I_SB(inode)->swapfile_inode)) 3974 #define stat_inc_atomic_inode(inode) \ 3975 (atomic_inc(&F2FS_I_SB(inode)->atomic_files)) 3976 #define stat_dec_atomic_inode(inode) \ 3977 (atomic_dec(&F2FS_I_SB(inode)->atomic_files)) 3978 #define stat_inc_meta_count(sbi, blkaddr) \ 3979 do { \ 3980 if (blkaddr < SIT_I(sbi)->sit_base_addr) \ 3981 atomic_inc(&(sbi)->meta_count[META_CP]); \ 3982 else if (blkaddr < NM_I(sbi)->nat_blkaddr) \ 3983 atomic_inc(&(sbi)->meta_count[META_SIT]); \ 3984 else if (blkaddr < SM_I(sbi)->ssa_blkaddr) \ 3985 atomic_inc(&(sbi)->meta_count[META_NAT]); \ 3986 else if (blkaddr < SM_I(sbi)->main_blkaddr) \ 3987 atomic_inc(&(sbi)->meta_count[META_SSA]); \ 3988 } while (0) 3989 #define stat_inc_seg_type(sbi, curseg) \ 3990 ((sbi)->segment_count[(curseg)->alloc_type]++) 3991 #define stat_inc_block_count(sbi, curseg) \ 3992 ((sbi)->block_count[(curseg)->alloc_type]++) 3993 #define stat_inc_inplace_blocks(sbi) \ 3994 (atomic_inc(&(sbi)->inplace_count)) 3995 #define stat_update_max_atomic_write(inode) \ 3996 do { \ 3997 int cur = atomic_read(&F2FS_I_SB(inode)->atomic_files); \ 3998 int max = atomic_read(&F2FS_I_SB(inode)->max_aw_cnt); \ 3999 if (cur > max) \ 4000 atomic_set(&F2FS_I_SB(inode)->max_aw_cnt, cur); \ 4001 } while (0) 4002 #define stat_inc_seg_count(sbi, type, gc_type) \ 4003 do { \ 4004 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 4005 si->tot_segs++; \ 4006 if ((type) == SUM_TYPE_DATA) { \ 4007 si->data_segs++; \ 4008 si->bg_data_segs += (gc_type == BG_GC) ? 1 : 0; \ 4009 } else { \ 4010 si->node_segs++; \ 4011 si->bg_node_segs += (gc_type == BG_GC) ? 1 : 0; \ 4012 } \ 4013 } while (0) 4014 4015 #define stat_inc_tot_blk_count(si, blks) \ 4016 ((si)->tot_blks += (blks)) 4017 4018 #define stat_inc_data_blk_count(sbi, blks, gc_type) \ 4019 do { \ 4020 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 4021 stat_inc_tot_blk_count(si, blks); \ 4022 si->data_blks += (blks); \ 4023 si->bg_data_blks += ((gc_type) == BG_GC) ? (blks) : 0; \ 4024 } while (0) 4025 4026 #define stat_inc_node_blk_count(sbi, blks, gc_type) \ 4027 do { \ 4028 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 4029 stat_inc_tot_blk_count(si, blks); \ 4030 si->node_blks += (blks); \ 4031 si->bg_node_blks += ((gc_type) == BG_GC) ? (blks) : 0; \ 4032 } while (0) 4033 4034 int f2fs_build_stats(struct f2fs_sb_info *sbi); 4035 void f2fs_destroy_stats(struct f2fs_sb_info *sbi); 4036 void __init f2fs_create_root_stats(void); 4037 void f2fs_destroy_root_stats(void); 4038 void f2fs_update_sit_info(struct f2fs_sb_info *sbi); 4039 #else 4040 #define stat_inc_cp_count(si) do { } while (0) 4041 #define stat_inc_bg_cp_count(si) do { } while (0) 4042 #define stat_inc_call_count(si) do { } while (0) 4043 #define stat_inc_bggc_count(si) do { } while (0) 4044 #define stat_io_skip_bggc_count(sbi) do { } while (0) 4045 #define stat_other_skip_bggc_count(sbi) do { } while (0) 4046 #define stat_inc_dirty_inode(sbi, type) do { } while (0) 4047 #define stat_dec_dirty_inode(sbi, type) do { } while (0) 4048 #define stat_inc_total_hit(sbi, type) do { } while (0) 4049 #define stat_inc_rbtree_node_hit(sbi, type) do { } while (0) 4050 #define stat_inc_largest_node_hit(sbi) do { } while (0) 4051 #define stat_inc_cached_node_hit(sbi, type) do { } while (0) 4052 #define stat_inc_inline_xattr(inode) do { } while (0) 4053 #define stat_dec_inline_xattr(inode) do { } while (0) 4054 #define stat_inc_inline_inode(inode) do { } while (0) 4055 #define stat_dec_inline_inode(inode) do { } while (0) 4056 #define stat_inc_inline_dir(inode) do { } while (0) 4057 #define stat_dec_inline_dir(inode) do { } while (0) 4058 #define stat_inc_compr_inode(inode) do { } while (0) 4059 #define stat_dec_compr_inode(inode) do { } while (0) 4060 #define stat_add_compr_blocks(inode, blocks) do { } while (0) 4061 #define stat_sub_compr_blocks(inode, blocks) do { } while (0) 4062 #define stat_inc_swapfile_inode(inode) do { } while (0) 4063 #define stat_dec_swapfile_inode(inode) do { } while (0) 4064 #define stat_inc_atomic_inode(inode) do { } while (0) 4065 #define stat_dec_atomic_inode(inode) do { } while (0) 4066 #define stat_update_max_atomic_write(inode) do { } while (0) 4067 #define stat_inc_meta_count(sbi, blkaddr) do { } while (0) 4068 #define stat_inc_seg_type(sbi, curseg) do { } while (0) 4069 #define stat_inc_block_count(sbi, curseg) do { } while (0) 4070 #define stat_inc_inplace_blocks(sbi) do { } while (0) 4071 #define stat_inc_seg_count(sbi, type, gc_type) do { } while (0) 4072 #define stat_inc_tot_blk_count(si, blks) do { } while (0) 4073 #define stat_inc_data_blk_count(sbi, blks, gc_type) do { } while (0) 4074 #define stat_inc_node_blk_count(sbi, blks, gc_type) do { } while (0) 4075 4076 static inline int f2fs_build_stats(struct f2fs_sb_info *sbi) { return 0; } 4077 static inline void f2fs_destroy_stats(struct f2fs_sb_info *sbi) { } 4078 static inline void __init f2fs_create_root_stats(void) { } 4079 static inline void f2fs_destroy_root_stats(void) { } 4080 static inline void f2fs_update_sit_info(struct f2fs_sb_info *sbi) {} 4081 #endif 4082 4083 extern const struct file_operations f2fs_dir_operations; 4084 extern const struct file_operations f2fs_file_operations; 4085 extern const struct inode_operations f2fs_file_inode_operations; 4086 extern const struct address_space_operations f2fs_dblock_aops; 4087 extern const struct address_space_operations f2fs_node_aops; 4088 extern const struct address_space_operations f2fs_meta_aops; 4089 extern const struct inode_operations f2fs_dir_inode_operations; 4090 extern const struct inode_operations f2fs_symlink_inode_operations; 4091 extern const struct inode_operations f2fs_encrypted_symlink_inode_operations; 4092 extern const struct inode_operations f2fs_special_inode_operations; 4093 extern struct kmem_cache *f2fs_inode_entry_slab; 4094 4095 /* 4096 * inline.c 4097 */ 4098 bool f2fs_may_inline_data(struct inode *inode); 4099 bool f2fs_sanity_check_inline_data(struct inode *inode); 4100 bool f2fs_may_inline_dentry(struct inode *inode); 4101 void f2fs_do_read_inline_data(struct page *page, struct page *ipage); 4102 void f2fs_truncate_inline_inode(struct inode *inode, 4103 struct page *ipage, u64 from); 4104 int f2fs_read_inline_data(struct inode *inode, struct page *page); 4105 int f2fs_convert_inline_page(struct dnode_of_data *dn, struct page *page); 4106 int f2fs_convert_inline_inode(struct inode *inode); 4107 int f2fs_try_convert_inline_dir(struct inode *dir, struct dentry *dentry); 4108 int f2fs_write_inline_data(struct inode *inode, struct page *page); 4109 int f2fs_recover_inline_data(struct inode *inode, struct page *npage); 4110 struct f2fs_dir_entry *f2fs_find_in_inline_dir(struct inode *dir, 4111 const struct f2fs_filename *fname, 4112 struct page **res_page); 4113 int f2fs_make_empty_inline_dir(struct inode *inode, struct inode *parent, 4114 struct page *ipage); 4115 int f2fs_add_inline_entry(struct inode *dir, const struct f2fs_filename *fname, 4116 struct inode *inode, nid_t ino, umode_t mode); 4117 void f2fs_delete_inline_entry(struct f2fs_dir_entry *dentry, 4118 struct page *page, struct inode *dir, 4119 struct inode *inode); 4120 bool f2fs_empty_inline_dir(struct inode *dir); 4121 int f2fs_read_inline_dir(struct file *file, struct dir_context *ctx, 4122 struct fscrypt_str *fstr); 4123 int f2fs_inline_data_fiemap(struct inode *inode, 4124 struct fiemap_extent_info *fieinfo, 4125 __u64 start, __u64 len); 4126 4127 /* 4128 * shrinker.c 4129 */ 4130 unsigned long f2fs_shrink_count(struct shrinker *shrink, 4131 struct shrink_control *sc); 4132 unsigned long f2fs_shrink_scan(struct shrinker *shrink, 4133 struct shrink_control *sc); 4134 void f2fs_join_shrinker(struct f2fs_sb_info *sbi); 4135 void f2fs_leave_shrinker(struct f2fs_sb_info *sbi); 4136 4137 /* 4138 * extent_cache.c 4139 */ 4140 bool sanity_check_extent_cache(struct inode *inode); 4141 struct rb_entry *f2fs_lookup_rb_tree(struct rb_root_cached *root, 4142 struct rb_entry *cached_re, unsigned int ofs); 4143 struct rb_node **f2fs_lookup_rb_tree_ext(struct f2fs_sb_info *sbi, 4144 struct rb_root_cached *root, 4145 struct rb_node **parent, 4146 unsigned long long key, bool *left_most); 4147 struct rb_node **f2fs_lookup_rb_tree_for_insert(struct f2fs_sb_info *sbi, 4148 struct rb_root_cached *root, 4149 struct rb_node **parent, 4150 unsigned int ofs, bool *leftmost); 4151 struct rb_entry *f2fs_lookup_rb_tree_ret(struct rb_root_cached *root, 4152 struct rb_entry *cached_re, unsigned int ofs, 4153 struct rb_entry **prev_entry, struct rb_entry **next_entry, 4154 struct rb_node ***insert_p, struct rb_node **insert_parent, 4155 bool force, bool *leftmost); 4156 bool f2fs_check_rb_tree_consistence(struct f2fs_sb_info *sbi, 4157 struct rb_root_cached *root, bool check_key); 4158 void f2fs_init_extent_tree(struct inode *inode); 4159 void f2fs_drop_extent_tree(struct inode *inode); 4160 void f2fs_destroy_extent_node(struct inode *inode); 4161 void f2fs_destroy_extent_tree(struct inode *inode); 4162 void f2fs_init_extent_cache_info(struct f2fs_sb_info *sbi); 4163 int __init f2fs_create_extent_cache(void); 4164 void f2fs_destroy_extent_cache(void); 4165 4166 /* read extent cache ops */ 4167 void f2fs_init_read_extent_tree(struct inode *inode, struct page *ipage); 4168 bool f2fs_lookup_read_extent_cache(struct inode *inode, pgoff_t pgofs, 4169 struct extent_info *ei); 4170 bool f2fs_lookup_read_extent_cache_block(struct inode *inode, pgoff_t index, 4171 block_t *blkaddr); 4172 void f2fs_update_read_extent_cache(struct dnode_of_data *dn); 4173 void f2fs_update_read_extent_cache_range(struct dnode_of_data *dn, 4174 pgoff_t fofs, block_t blkaddr, unsigned int len); 4175 unsigned int f2fs_shrink_read_extent_tree(struct f2fs_sb_info *sbi, 4176 int nr_shrink); 4177 4178 /* block age extent cache ops */ 4179 void f2fs_init_age_extent_tree(struct inode *inode); 4180 bool f2fs_lookup_age_extent_cache(struct inode *inode, pgoff_t pgofs, 4181 struct extent_info *ei); 4182 void f2fs_update_age_extent_cache(struct dnode_of_data *dn); 4183 void f2fs_update_age_extent_cache_range(struct dnode_of_data *dn, 4184 pgoff_t fofs, unsigned int len); 4185 unsigned int f2fs_shrink_age_extent_tree(struct f2fs_sb_info *sbi, 4186 int nr_shrink); 4187 4188 /* 4189 * sysfs.c 4190 */ 4191 #define MIN_RA_MUL 2 4192 #define MAX_RA_MUL 256 4193 4194 int __init f2fs_init_sysfs(void); 4195 void f2fs_exit_sysfs(void); 4196 int f2fs_register_sysfs(struct f2fs_sb_info *sbi); 4197 void f2fs_unregister_sysfs(struct f2fs_sb_info *sbi); 4198 4199 /* verity.c */ 4200 extern const struct fsverity_operations f2fs_verityops; 4201 4202 /* 4203 * crypto support 4204 */ 4205 static inline bool f2fs_encrypted_file(struct inode *inode) 4206 { 4207 return IS_ENCRYPTED(inode) && S_ISREG(inode->i_mode); 4208 } 4209 4210 static inline void f2fs_set_encrypted_inode(struct inode *inode) 4211 { 4212 #ifdef CONFIG_FS_ENCRYPTION 4213 file_set_encrypt(inode); 4214 f2fs_set_inode_flags(inode); 4215 #endif 4216 } 4217 4218 /* 4219 * Returns true if the reads of the inode's data need to undergo some 4220 * postprocessing step, like decryption or authenticity verification. 4221 */ 4222 static inline bool f2fs_post_read_required(struct inode *inode) 4223 { 4224 return f2fs_encrypted_file(inode) || fsverity_active(inode) || 4225 f2fs_compressed_file(inode); 4226 } 4227 4228 /* 4229 * compress.c 4230 */ 4231 #ifdef CONFIG_F2FS_FS_COMPRESSION 4232 bool f2fs_is_compressed_page(struct page *page); 4233 struct page *f2fs_compress_control_page(struct page *page); 4234 int f2fs_prepare_compress_overwrite(struct inode *inode, 4235 struct page **pagep, pgoff_t index, void **fsdata); 4236 bool f2fs_compress_write_end(struct inode *inode, void *fsdata, 4237 pgoff_t index, unsigned copied); 4238 int f2fs_truncate_partial_cluster(struct inode *inode, u64 from, bool lock); 4239 void f2fs_compress_write_end_io(struct bio *bio, struct page *page); 4240 bool f2fs_is_compress_backend_ready(struct inode *inode); 4241 int __init f2fs_init_compress_mempool(void); 4242 void f2fs_destroy_compress_mempool(void); 4243 void f2fs_decompress_cluster(struct decompress_io_ctx *dic, bool in_task); 4244 void f2fs_end_read_compressed_page(struct page *page, bool failed, 4245 block_t blkaddr, bool in_task); 4246 bool f2fs_cluster_is_empty(struct compress_ctx *cc); 4247 bool f2fs_cluster_can_merge_page(struct compress_ctx *cc, pgoff_t index); 4248 bool f2fs_all_cluster_page_ready(struct compress_ctx *cc, struct page **pages, 4249 int index, int nr_pages, bool uptodate); 4250 bool f2fs_sanity_check_cluster(struct dnode_of_data *dn); 4251 void f2fs_compress_ctx_add_page(struct compress_ctx *cc, struct page *page); 4252 int f2fs_write_multi_pages(struct compress_ctx *cc, 4253 int *submitted, 4254 struct writeback_control *wbc, 4255 enum iostat_type io_type); 4256 int f2fs_is_compressed_cluster(struct inode *inode, pgoff_t index); 4257 void f2fs_update_read_extent_tree_range_compressed(struct inode *inode, 4258 pgoff_t fofs, block_t blkaddr, 4259 unsigned int llen, unsigned int c_len); 4260 int f2fs_read_multi_pages(struct compress_ctx *cc, struct bio **bio_ret, 4261 unsigned nr_pages, sector_t *last_block_in_bio, 4262 bool is_readahead, bool for_write); 4263 struct decompress_io_ctx *f2fs_alloc_dic(struct compress_ctx *cc); 4264 void f2fs_decompress_end_io(struct decompress_io_ctx *dic, bool failed, 4265 bool in_task); 4266 void f2fs_put_page_dic(struct page *page, bool in_task); 4267 unsigned int f2fs_cluster_blocks_are_contiguous(struct dnode_of_data *dn); 4268 int f2fs_init_compress_ctx(struct compress_ctx *cc); 4269 void f2fs_destroy_compress_ctx(struct compress_ctx *cc, bool reuse); 4270 void f2fs_init_compress_info(struct f2fs_sb_info *sbi); 4271 int f2fs_init_compress_inode(struct f2fs_sb_info *sbi); 4272 void f2fs_destroy_compress_inode(struct f2fs_sb_info *sbi); 4273 int f2fs_init_page_array_cache(struct f2fs_sb_info *sbi); 4274 void f2fs_destroy_page_array_cache(struct f2fs_sb_info *sbi); 4275 int __init f2fs_init_compress_cache(void); 4276 void f2fs_destroy_compress_cache(void); 4277 struct address_space *COMPRESS_MAPPING(struct f2fs_sb_info *sbi); 4278 void f2fs_invalidate_compress_page(struct f2fs_sb_info *sbi, block_t blkaddr); 4279 void f2fs_cache_compressed_page(struct f2fs_sb_info *sbi, struct page *page, 4280 nid_t ino, block_t blkaddr); 4281 bool f2fs_load_compressed_page(struct f2fs_sb_info *sbi, struct page *page, 4282 block_t blkaddr); 4283 void f2fs_invalidate_compress_pages(struct f2fs_sb_info *sbi, nid_t ino); 4284 #define inc_compr_inode_stat(inode) \ 4285 do { \ 4286 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); \ 4287 sbi->compr_new_inode++; \ 4288 } while (0) 4289 #define add_compr_block_stat(inode, blocks) \ 4290 do { \ 4291 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); \ 4292 int diff = F2FS_I(inode)->i_cluster_size - blocks; \ 4293 sbi->compr_written_block += blocks; \ 4294 sbi->compr_saved_block += diff; \ 4295 } while (0) 4296 #else 4297 static inline bool f2fs_is_compressed_page(struct page *page) { return false; } 4298 static inline bool f2fs_is_compress_backend_ready(struct inode *inode) 4299 { 4300 if (!f2fs_compressed_file(inode)) 4301 return true; 4302 /* not support compression */ 4303 return false; 4304 } 4305 static inline struct page *f2fs_compress_control_page(struct page *page) 4306 { 4307 WARN_ON_ONCE(1); 4308 return ERR_PTR(-EINVAL); 4309 } 4310 static inline int __init f2fs_init_compress_mempool(void) { return 0; } 4311 static inline void f2fs_destroy_compress_mempool(void) { } 4312 static inline void f2fs_decompress_cluster(struct decompress_io_ctx *dic, 4313 bool in_task) { } 4314 static inline void f2fs_end_read_compressed_page(struct page *page, 4315 bool failed, block_t blkaddr, bool in_task) 4316 { 4317 WARN_ON_ONCE(1); 4318 } 4319 static inline void f2fs_put_page_dic(struct page *page, bool in_task) 4320 { 4321 WARN_ON_ONCE(1); 4322 } 4323 static inline unsigned int f2fs_cluster_blocks_are_contiguous(struct dnode_of_data *dn) { return 0; } 4324 static inline bool f2fs_sanity_check_cluster(struct dnode_of_data *dn) { return false; } 4325 static inline int f2fs_init_compress_inode(struct f2fs_sb_info *sbi) { return 0; } 4326 static inline void f2fs_destroy_compress_inode(struct f2fs_sb_info *sbi) { } 4327 static inline int f2fs_init_page_array_cache(struct f2fs_sb_info *sbi) { return 0; } 4328 static inline void f2fs_destroy_page_array_cache(struct f2fs_sb_info *sbi) { } 4329 static inline int __init f2fs_init_compress_cache(void) { return 0; } 4330 static inline void f2fs_destroy_compress_cache(void) { } 4331 static inline void f2fs_invalidate_compress_page(struct f2fs_sb_info *sbi, 4332 block_t blkaddr) { } 4333 static inline void f2fs_cache_compressed_page(struct f2fs_sb_info *sbi, 4334 struct page *page, nid_t ino, block_t blkaddr) { } 4335 static inline bool f2fs_load_compressed_page(struct f2fs_sb_info *sbi, 4336 struct page *page, block_t blkaddr) { return false; } 4337 static inline void f2fs_invalidate_compress_pages(struct f2fs_sb_info *sbi, 4338 nid_t ino) { } 4339 #define inc_compr_inode_stat(inode) do { } while (0) 4340 static inline void f2fs_update_read_extent_tree_range_compressed( 4341 struct inode *inode, 4342 pgoff_t fofs, block_t blkaddr, 4343 unsigned int llen, unsigned int c_len) { } 4344 #endif 4345 4346 static inline int set_compress_context(struct inode *inode) 4347 { 4348 #ifdef CONFIG_F2FS_FS_COMPRESSION 4349 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 4350 4351 F2FS_I(inode)->i_compress_algorithm = 4352 F2FS_OPTION(sbi).compress_algorithm; 4353 F2FS_I(inode)->i_log_cluster_size = 4354 F2FS_OPTION(sbi).compress_log_size; 4355 F2FS_I(inode)->i_compress_flag = 4356 F2FS_OPTION(sbi).compress_chksum ? 4357 1 << COMPRESS_CHKSUM : 0; 4358 F2FS_I(inode)->i_cluster_size = 4359 1 << F2FS_I(inode)->i_log_cluster_size; 4360 if ((F2FS_I(inode)->i_compress_algorithm == COMPRESS_LZ4 || 4361 F2FS_I(inode)->i_compress_algorithm == COMPRESS_ZSTD) && 4362 F2FS_OPTION(sbi).compress_level) 4363 F2FS_I(inode)->i_compress_level = 4364 F2FS_OPTION(sbi).compress_level; 4365 F2FS_I(inode)->i_flags |= F2FS_COMPR_FL; 4366 set_inode_flag(inode, FI_COMPRESSED_FILE); 4367 stat_inc_compr_inode(inode); 4368 inc_compr_inode_stat(inode); 4369 f2fs_mark_inode_dirty_sync(inode, true); 4370 return 0; 4371 #else 4372 return -EOPNOTSUPP; 4373 #endif 4374 } 4375 4376 static inline bool f2fs_disable_compressed_file(struct inode *inode) 4377 { 4378 struct f2fs_inode_info *fi = F2FS_I(inode); 4379 4380 if (!f2fs_compressed_file(inode)) 4381 return true; 4382 if (S_ISREG(inode->i_mode) && F2FS_HAS_BLOCKS(inode)) 4383 return false; 4384 4385 fi->i_flags &= ~F2FS_COMPR_FL; 4386 stat_dec_compr_inode(inode); 4387 clear_inode_flag(inode, FI_COMPRESSED_FILE); 4388 f2fs_mark_inode_dirty_sync(inode, true); 4389 return true; 4390 } 4391 4392 #define F2FS_FEATURE_FUNCS(name, flagname) \ 4393 static inline bool f2fs_sb_has_##name(struct f2fs_sb_info *sbi) \ 4394 { \ 4395 return F2FS_HAS_FEATURE(sbi, F2FS_FEATURE_##flagname); \ 4396 } 4397 4398 F2FS_FEATURE_FUNCS(encrypt, ENCRYPT); 4399 F2FS_FEATURE_FUNCS(blkzoned, BLKZONED); 4400 F2FS_FEATURE_FUNCS(extra_attr, EXTRA_ATTR); 4401 F2FS_FEATURE_FUNCS(project_quota, PRJQUOTA); 4402 F2FS_FEATURE_FUNCS(inode_chksum, INODE_CHKSUM); 4403 F2FS_FEATURE_FUNCS(flexible_inline_xattr, FLEXIBLE_INLINE_XATTR); 4404 F2FS_FEATURE_FUNCS(quota_ino, QUOTA_INO); 4405 F2FS_FEATURE_FUNCS(inode_crtime, INODE_CRTIME); 4406 F2FS_FEATURE_FUNCS(lost_found, LOST_FOUND); 4407 F2FS_FEATURE_FUNCS(verity, VERITY); 4408 F2FS_FEATURE_FUNCS(sb_chksum, SB_CHKSUM); 4409 F2FS_FEATURE_FUNCS(casefold, CASEFOLD); 4410 F2FS_FEATURE_FUNCS(compression, COMPRESSION); 4411 F2FS_FEATURE_FUNCS(readonly, RO); 4412 4413 #ifdef CONFIG_BLK_DEV_ZONED 4414 static inline bool f2fs_blkz_is_seq(struct f2fs_sb_info *sbi, int devi, 4415 block_t blkaddr) 4416 { 4417 unsigned int zno = blkaddr >> sbi->log_blocks_per_blkz; 4418 4419 return test_bit(zno, FDEV(devi).blkz_seq); 4420 } 4421 #endif 4422 4423 static inline bool f2fs_hw_should_discard(struct f2fs_sb_info *sbi) 4424 { 4425 return f2fs_sb_has_blkzoned(sbi); 4426 } 4427 4428 static inline bool f2fs_bdev_support_discard(struct block_device *bdev) 4429 { 4430 return bdev_max_discard_sectors(bdev) || bdev_is_zoned(bdev); 4431 } 4432 4433 static inline bool f2fs_hw_support_discard(struct f2fs_sb_info *sbi) 4434 { 4435 int i; 4436 4437 if (!f2fs_is_multi_device(sbi)) 4438 return f2fs_bdev_support_discard(sbi->sb->s_bdev); 4439 4440 for (i = 0; i < sbi->s_ndevs; i++) 4441 if (f2fs_bdev_support_discard(FDEV(i).bdev)) 4442 return true; 4443 return false; 4444 } 4445 4446 static inline bool f2fs_realtime_discard_enable(struct f2fs_sb_info *sbi) 4447 { 4448 return (test_opt(sbi, DISCARD) && f2fs_hw_support_discard(sbi)) || 4449 f2fs_hw_should_discard(sbi); 4450 } 4451 4452 static inline bool f2fs_hw_is_readonly(struct f2fs_sb_info *sbi) 4453 { 4454 int i; 4455 4456 if (!f2fs_is_multi_device(sbi)) 4457 return bdev_read_only(sbi->sb->s_bdev); 4458 4459 for (i = 0; i < sbi->s_ndevs; i++) 4460 if (bdev_read_only(FDEV(i).bdev)) 4461 return true; 4462 return false; 4463 } 4464 4465 static inline bool f2fs_lfs_mode(struct f2fs_sb_info *sbi) 4466 { 4467 return F2FS_OPTION(sbi).fs_mode == FS_MODE_LFS; 4468 } 4469 4470 static inline bool f2fs_low_mem_mode(struct f2fs_sb_info *sbi) 4471 { 4472 return F2FS_OPTION(sbi).memory_mode == MEMORY_MODE_LOW; 4473 } 4474 4475 static inline bool f2fs_may_compress(struct inode *inode) 4476 { 4477 if (IS_SWAPFILE(inode) || f2fs_is_pinned_file(inode) || 4478 f2fs_is_atomic_file(inode) || f2fs_has_inline_data(inode)) 4479 return false; 4480 return S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode); 4481 } 4482 4483 static inline void f2fs_i_compr_blocks_update(struct inode *inode, 4484 u64 blocks, bool add) 4485 { 4486 struct f2fs_inode_info *fi = F2FS_I(inode); 4487 int diff = fi->i_cluster_size - blocks; 4488 4489 /* don't update i_compr_blocks if saved blocks were released */ 4490 if (!add && !atomic_read(&fi->i_compr_blocks)) 4491 return; 4492 4493 if (add) { 4494 atomic_add(diff, &fi->i_compr_blocks); 4495 stat_add_compr_blocks(inode, diff); 4496 } else { 4497 atomic_sub(diff, &fi->i_compr_blocks); 4498 stat_sub_compr_blocks(inode, diff); 4499 } 4500 f2fs_mark_inode_dirty_sync(inode, true); 4501 } 4502 4503 static inline bool f2fs_allow_multi_device_dio(struct f2fs_sb_info *sbi, 4504 int flag) 4505 { 4506 if (!f2fs_is_multi_device(sbi)) 4507 return false; 4508 if (flag != F2FS_GET_BLOCK_DIO) 4509 return false; 4510 return sbi->aligned_blksize; 4511 } 4512 4513 static inline bool f2fs_need_verity(const struct inode *inode, pgoff_t idx) 4514 { 4515 return fsverity_active(inode) && 4516 idx < DIV_ROUND_UP(inode->i_size, PAGE_SIZE); 4517 } 4518 4519 #ifdef CONFIG_F2FS_FAULT_INJECTION 4520 extern void f2fs_build_fault_attr(struct f2fs_sb_info *sbi, unsigned int rate, 4521 unsigned int type); 4522 #else 4523 #define f2fs_build_fault_attr(sbi, rate, type) do { } while (0) 4524 #endif 4525 4526 static inline bool is_journalled_quota(struct f2fs_sb_info *sbi) 4527 { 4528 #ifdef CONFIG_QUOTA 4529 if (f2fs_sb_has_quota_ino(sbi)) 4530 return true; 4531 if (F2FS_OPTION(sbi).s_qf_names[USRQUOTA] || 4532 F2FS_OPTION(sbi).s_qf_names[GRPQUOTA] || 4533 F2FS_OPTION(sbi).s_qf_names[PRJQUOTA]) 4534 return true; 4535 #endif 4536 return false; 4537 } 4538 4539 static inline bool f2fs_block_unit_discard(struct f2fs_sb_info *sbi) 4540 { 4541 return F2FS_OPTION(sbi).discard_unit == DISCARD_UNIT_BLOCK; 4542 } 4543 4544 static inline void f2fs_io_schedule_timeout(long timeout) 4545 { 4546 set_current_state(TASK_UNINTERRUPTIBLE); 4547 io_schedule_timeout(timeout); 4548 } 4549 4550 static inline void f2fs_handle_page_eio(struct f2fs_sb_info *sbi, pgoff_t ofs, 4551 enum page_type type) 4552 { 4553 if (unlikely(f2fs_cp_error(sbi))) 4554 return; 4555 4556 if (ofs == sbi->page_eio_ofs[type]) { 4557 if (sbi->page_eio_cnt[type]++ == MAX_RETRY_PAGE_EIO) 4558 set_ckpt_flags(sbi, CP_ERROR_FLAG); 4559 } else { 4560 sbi->page_eio_ofs[type] = ofs; 4561 sbi->page_eio_cnt[type] = 0; 4562 } 4563 } 4564 4565 static inline bool f2fs_is_readonly(struct f2fs_sb_info *sbi) 4566 { 4567 return f2fs_sb_has_readonly(sbi) || f2fs_readonly(sbi->sb); 4568 } 4569 4570 #define EFSBADCRC EBADMSG /* Bad CRC detected */ 4571 #define EFSCORRUPTED EUCLEAN /* Filesystem is corrupted */ 4572 4573 #endif /* _LINUX_F2FS_H */ 4574