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/vmalloc.h> 22 #include <linux/bio.h> 23 #include <linux/blkdev.h> 24 #include <linux/quotaops.h> 25 #include <crypto/hash.h> 26 27 #include <linux/fscrypt.h> 28 29 #ifdef CONFIG_F2FS_CHECK_FS 30 #define f2fs_bug_on(sbi, condition) BUG_ON(condition) 31 #else 32 #define f2fs_bug_on(sbi, condition) \ 33 do { \ 34 if (unlikely(condition)) { \ 35 WARN_ON(1); \ 36 set_sbi_flag(sbi, SBI_NEED_FSCK); \ 37 } \ 38 } while (0) 39 #endif 40 41 enum { 42 FAULT_KMALLOC, 43 FAULT_KVMALLOC, 44 FAULT_PAGE_ALLOC, 45 FAULT_PAGE_GET, 46 FAULT_ALLOC_BIO, 47 FAULT_ALLOC_NID, 48 FAULT_ORPHAN, 49 FAULT_BLOCK, 50 FAULT_DIR_DEPTH, 51 FAULT_EVICT_INODE, 52 FAULT_TRUNCATE, 53 FAULT_READ_IO, 54 FAULT_CHECKPOINT, 55 FAULT_DISCARD, 56 FAULT_WRITE_IO, 57 FAULT_MAX, 58 }; 59 60 #ifdef CONFIG_F2FS_FAULT_INJECTION 61 #define F2FS_ALL_FAULT_TYPE ((1 << FAULT_MAX) - 1) 62 63 struct f2fs_fault_info { 64 atomic_t inject_ops; 65 unsigned int inject_rate; 66 unsigned int inject_type; 67 }; 68 69 extern const char *f2fs_fault_name[FAULT_MAX]; 70 #define IS_FAULT_SET(fi, type) ((fi)->inject_type & (1 << (type))) 71 #endif 72 73 /* 74 * For mount options 75 */ 76 #define F2FS_MOUNT_BG_GC 0x00000001 77 #define F2FS_MOUNT_DISABLE_ROLL_FORWARD 0x00000002 78 #define F2FS_MOUNT_DISCARD 0x00000004 79 #define F2FS_MOUNT_NOHEAP 0x00000008 80 #define F2FS_MOUNT_XATTR_USER 0x00000010 81 #define F2FS_MOUNT_POSIX_ACL 0x00000020 82 #define F2FS_MOUNT_DISABLE_EXT_IDENTIFY 0x00000040 83 #define F2FS_MOUNT_INLINE_XATTR 0x00000080 84 #define F2FS_MOUNT_INLINE_DATA 0x00000100 85 #define F2FS_MOUNT_INLINE_DENTRY 0x00000200 86 #define F2FS_MOUNT_FLUSH_MERGE 0x00000400 87 #define F2FS_MOUNT_NOBARRIER 0x00000800 88 #define F2FS_MOUNT_FASTBOOT 0x00001000 89 #define F2FS_MOUNT_EXTENT_CACHE 0x00002000 90 #define F2FS_MOUNT_FORCE_FG_GC 0x00004000 91 #define F2FS_MOUNT_DATA_FLUSH 0x00008000 92 #define F2FS_MOUNT_FAULT_INJECTION 0x00010000 93 #define F2FS_MOUNT_ADAPTIVE 0x00020000 94 #define F2FS_MOUNT_LFS 0x00040000 95 #define F2FS_MOUNT_USRQUOTA 0x00080000 96 #define F2FS_MOUNT_GRPQUOTA 0x00100000 97 #define F2FS_MOUNT_PRJQUOTA 0x00200000 98 #define F2FS_MOUNT_QUOTA 0x00400000 99 #define F2FS_MOUNT_INLINE_XATTR_SIZE 0x00800000 100 #define F2FS_MOUNT_RESERVE_ROOT 0x01000000 101 #define F2FS_MOUNT_DISABLE_CHECKPOINT 0x02000000 102 103 #define F2FS_OPTION(sbi) ((sbi)->mount_opt) 104 #define clear_opt(sbi, option) (F2FS_OPTION(sbi).opt &= ~F2FS_MOUNT_##option) 105 #define set_opt(sbi, option) (F2FS_OPTION(sbi).opt |= F2FS_MOUNT_##option) 106 #define test_opt(sbi, option) (F2FS_OPTION(sbi).opt & F2FS_MOUNT_##option) 107 108 #define ver_after(a, b) (typecheck(unsigned long long, a) && \ 109 typecheck(unsigned long long, b) && \ 110 ((long long)((a) - (b)) > 0)) 111 112 typedef u32 block_t; /* 113 * should not change u32, since it is the on-disk block 114 * address format, __le32. 115 */ 116 typedef u32 nid_t; 117 118 struct f2fs_mount_info { 119 unsigned int opt; 120 int write_io_size_bits; /* Write IO size bits */ 121 block_t root_reserved_blocks; /* root reserved blocks */ 122 kuid_t s_resuid; /* reserved blocks for uid */ 123 kgid_t s_resgid; /* reserved blocks for gid */ 124 int active_logs; /* # of active logs */ 125 int inline_xattr_size; /* inline xattr size */ 126 #ifdef CONFIG_F2FS_FAULT_INJECTION 127 struct f2fs_fault_info fault_info; /* For fault injection */ 128 #endif 129 #ifdef CONFIG_QUOTA 130 /* Names of quota files with journalled quota */ 131 char *s_qf_names[MAXQUOTAS]; 132 int s_jquota_fmt; /* Format of quota to use */ 133 #endif 134 /* For which write hints are passed down to block layer */ 135 int whint_mode; 136 int alloc_mode; /* segment allocation policy */ 137 int fsync_mode; /* fsync policy */ 138 bool test_dummy_encryption; /* test dummy encryption */ 139 }; 140 141 #define F2FS_FEATURE_ENCRYPT 0x0001 142 #define F2FS_FEATURE_BLKZONED 0x0002 143 #define F2FS_FEATURE_ATOMIC_WRITE 0x0004 144 #define F2FS_FEATURE_EXTRA_ATTR 0x0008 145 #define F2FS_FEATURE_PRJQUOTA 0x0010 146 #define F2FS_FEATURE_INODE_CHKSUM 0x0020 147 #define F2FS_FEATURE_FLEXIBLE_INLINE_XATTR 0x0040 148 #define F2FS_FEATURE_QUOTA_INO 0x0080 149 #define F2FS_FEATURE_INODE_CRTIME 0x0100 150 #define F2FS_FEATURE_LOST_FOUND 0x0200 151 #define F2FS_FEATURE_VERITY 0x0400 /* reserved */ 152 #define F2FS_FEATURE_SB_CHKSUM 0x0800 153 154 #define __F2FS_HAS_FEATURE(raw_super, mask) \ 155 ((raw_super->feature & cpu_to_le32(mask)) != 0) 156 #define F2FS_HAS_FEATURE(sbi, mask) __F2FS_HAS_FEATURE(sbi->raw_super, mask) 157 #define F2FS_SET_FEATURE(sbi, mask) \ 158 (sbi->raw_super->feature |= cpu_to_le32(mask)) 159 #define F2FS_CLEAR_FEATURE(sbi, mask) \ 160 (sbi->raw_super->feature &= ~cpu_to_le32(mask)) 161 162 /* 163 * Default values for user and/or group using reserved blocks 164 */ 165 #define F2FS_DEF_RESUID 0 166 #define F2FS_DEF_RESGID 0 167 168 /* 169 * For checkpoint manager 170 */ 171 enum { 172 NAT_BITMAP, 173 SIT_BITMAP 174 }; 175 176 #define CP_UMOUNT 0x00000001 177 #define CP_FASTBOOT 0x00000002 178 #define CP_SYNC 0x00000004 179 #define CP_RECOVERY 0x00000008 180 #define CP_DISCARD 0x00000010 181 #define CP_TRIMMED 0x00000020 182 #define CP_PAUSE 0x00000040 183 184 #define MAX_DISCARD_BLOCKS(sbi) BLKS_PER_SEC(sbi) 185 #define DEF_MAX_DISCARD_REQUEST 8 /* issue 8 discards per round */ 186 #define DEF_MIN_DISCARD_ISSUE_TIME 50 /* 50 ms, if exists */ 187 #define DEF_MID_DISCARD_ISSUE_TIME 500 /* 500 ms, if device busy */ 188 #define DEF_MAX_DISCARD_ISSUE_TIME 60000 /* 60 s, if no candidates */ 189 #define DEF_DISCARD_URGENT_UTIL 80 /* do more discard over 80% */ 190 #define DEF_CP_INTERVAL 60 /* 60 secs */ 191 #define DEF_IDLE_INTERVAL 5 /* 5 secs */ 192 #define DEF_DISABLE_INTERVAL 5 /* 5 secs */ 193 #define DEF_DISABLE_QUICK_INTERVAL 1 /* 1 secs */ 194 #define DEF_UMOUNT_DISCARD_TIMEOUT 5 /* 5 secs */ 195 196 struct cp_control { 197 int reason; 198 __u64 trim_start; 199 __u64 trim_end; 200 __u64 trim_minlen; 201 }; 202 203 /* 204 * indicate meta/data type 205 */ 206 enum { 207 META_CP, 208 META_NAT, 209 META_SIT, 210 META_SSA, 211 META_MAX, 212 META_POR, 213 DATA_GENERIC, /* check range only */ 214 DATA_GENERIC_ENHANCE, /* strong check on range and segment bitmap */ 215 DATA_GENERIC_ENHANCE_READ, /* 216 * strong check on range and segment 217 * bitmap but no warning due to race 218 * condition of read on truncated area 219 * by extent_cache 220 */ 221 META_GENERIC, 222 }; 223 224 /* for the list of ino */ 225 enum { 226 ORPHAN_INO, /* for orphan ino list */ 227 APPEND_INO, /* for append ino list */ 228 UPDATE_INO, /* for update ino list */ 229 TRANS_DIR_INO, /* for trasactions dir ino list */ 230 FLUSH_INO, /* for multiple device flushing */ 231 MAX_INO_ENTRY, /* max. list */ 232 }; 233 234 struct ino_entry { 235 struct list_head list; /* list head */ 236 nid_t ino; /* inode number */ 237 unsigned int dirty_device; /* dirty device bitmap */ 238 }; 239 240 /* for the list of inodes to be GCed */ 241 struct inode_entry { 242 struct list_head list; /* list head */ 243 struct inode *inode; /* vfs inode pointer */ 244 }; 245 246 struct fsync_node_entry { 247 struct list_head list; /* list head */ 248 struct page *page; /* warm node page pointer */ 249 unsigned int seq_id; /* sequence id */ 250 }; 251 252 /* for the bitmap indicate blocks to be discarded */ 253 struct discard_entry { 254 struct list_head list; /* list head */ 255 block_t start_blkaddr; /* start blockaddr of current segment */ 256 unsigned char discard_map[SIT_VBLOCK_MAP_SIZE]; /* segment discard bitmap */ 257 }; 258 259 /* default discard granularity of inner discard thread, unit: block count */ 260 #define DEFAULT_DISCARD_GRANULARITY 16 261 262 /* max discard pend list number */ 263 #define MAX_PLIST_NUM 512 264 #define plist_idx(blk_num) ((blk_num) >= MAX_PLIST_NUM ? \ 265 (MAX_PLIST_NUM - 1) : ((blk_num) - 1)) 266 267 enum { 268 D_PREP, /* initial */ 269 D_PARTIAL, /* partially submitted */ 270 D_SUBMIT, /* all submitted */ 271 D_DONE, /* finished */ 272 }; 273 274 struct discard_info { 275 block_t lstart; /* logical start address */ 276 block_t len; /* length */ 277 block_t start; /* actual start address in dev */ 278 }; 279 280 struct discard_cmd { 281 struct rb_node rb_node; /* rb node located in rb-tree */ 282 union { 283 struct { 284 block_t lstart; /* logical start address */ 285 block_t len; /* length */ 286 block_t start; /* actual start address in dev */ 287 }; 288 struct discard_info di; /* discard info */ 289 290 }; 291 struct list_head list; /* command list */ 292 struct completion wait; /* compleation */ 293 struct block_device *bdev; /* bdev */ 294 unsigned short ref; /* reference count */ 295 unsigned char state; /* state */ 296 unsigned char queued; /* queued discard */ 297 int error; /* bio error */ 298 spinlock_t lock; /* for state/bio_ref updating */ 299 unsigned short bio_ref; /* bio reference count */ 300 }; 301 302 enum { 303 DPOLICY_BG, 304 DPOLICY_FORCE, 305 DPOLICY_FSTRIM, 306 DPOLICY_UMOUNT, 307 MAX_DPOLICY, 308 }; 309 310 struct discard_policy { 311 int type; /* type of discard */ 312 unsigned int min_interval; /* used for candidates exist */ 313 unsigned int mid_interval; /* used for device busy */ 314 unsigned int max_interval; /* used for candidates not exist */ 315 unsigned int max_requests; /* # of discards issued per round */ 316 unsigned int io_aware_gran; /* minimum granularity discard not be aware of I/O */ 317 bool io_aware; /* issue discard in idle time */ 318 bool sync; /* submit discard with REQ_SYNC flag */ 319 bool ordered; /* issue discard by lba order */ 320 unsigned int granularity; /* discard granularity */ 321 int timeout; /* discard timeout for put_super */ 322 }; 323 324 struct discard_cmd_control { 325 struct task_struct *f2fs_issue_discard; /* discard thread */ 326 struct list_head entry_list; /* 4KB discard entry list */ 327 struct list_head pend_list[MAX_PLIST_NUM];/* store pending entries */ 328 struct list_head wait_list; /* store on-flushing entries */ 329 struct list_head fstrim_list; /* in-flight discard from fstrim */ 330 wait_queue_head_t discard_wait_queue; /* waiting queue for wake-up */ 331 unsigned int discard_wake; /* to wake up discard thread */ 332 struct mutex cmd_lock; 333 unsigned int nr_discards; /* # of discards in the list */ 334 unsigned int max_discards; /* max. discards to be issued */ 335 unsigned int discard_granularity; /* discard granularity */ 336 unsigned int undiscard_blks; /* # of undiscard blocks */ 337 unsigned int next_pos; /* next discard position */ 338 atomic_t issued_discard; /* # of issued discard */ 339 atomic_t queued_discard; /* # of queued discard */ 340 atomic_t discard_cmd_cnt; /* # of cached cmd count */ 341 struct rb_root_cached root; /* root of discard rb-tree */ 342 bool rbtree_check; /* config for consistence check */ 343 }; 344 345 /* for the list of fsync inodes, used only during recovery */ 346 struct fsync_inode_entry { 347 struct list_head list; /* list head */ 348 struct inode *inode; /* vfs inode pointer */ 349 block_t blkaddr; /* block address locating the last fsync */ 350 block_t last_dentry; /* block address locating the last dentry */ 351 }; 352 353 #define nats_in_cursum(jnl) (le16_to_cpu((jnl)->n_nats)) 354 #define sits_in_cursum(jnl) (le16_to_cpu((jnl)->n_sits)) 355 356 #define nat_in_journal(jnl, i) ((jnl)->nat_j.entries[i].ne) 357 #define nid_in_journal(jnl, i) ((jnl)->nat_j.entries[i].nid) 358 #define sit_in_journal(jnl, i) ((jnl)->sit_j.entries[i].se) 359 #define segno_in_journal(jnl, i) ((jnl)->sit_j.entries[i].segno) 360 361 #define MAX_NAT_JENTRIES(jnl) (NAT_JOURNAL_ENTRIES - nats_in_cursum(jnl)) 362 #define MAX_SIT_JENTRIES(jnl) (SIT_JOURNAL_ENTRIES - sits_in_cursum(jnl)) 363 364 static inline int update_nats_in_cursum(struct f2fs_journal *journal, int i) 365 { 366 int before = nats_in_cursum(journal); 367 368 journal->n_nats = cpu_to_le16(before + i); 369 return before; 370 } 371 372 static inline int update_sits_in_cursum(struct f2fs_journal *journal, int i) 373 { 374 int before = sits_in_cursum(journal); 375 376 journal->n_sits = cpu_to_le16(before + i); 377 return before; 378 } 379 380 static inline bool __has_cursum_space(struct f2fs_journal *journal, 381 int size, int type) 382 { 383 if (type == NAT_JOURNAL) 384 return size <= MAX_NAT_JENTRIES(journal); 385 return size <= MAX_SIT_JENTRIES(journal); 386 } 387 388 /* 389 * ioctl commands 390 */ 391 #define F2FS_IOC_GETFLAGS FS_IOC_GETFLAGS 392 #define F2FS_IOC_SETFLAGS FS_IOC_SETFLAGS 393 #define F2FS_IOC_GETVERSION FS_IOC_GETVERSION 394 395 #define F2FS_IOCTL_MAGIC 0xf5 396 #define F2FS_IOC_START_ATOMIC_WRITE _IO(F2FS_IOCTL_MAGIC, 1) 397 #define F2FS_IOC_COMMIT_ATOMIC_WRITE _IO(F2FS_IOCTL_MAGIC, 2) 398 #define F2FS_IOC_START_VOLATILE_WRITE _IO(F2FS_IOCTL_MAGIC, 3) 399 #define F2FS_IOC_RELEASE_VOLATILE_WRITE _IO(F2FS_IOCTL_MAGIC, 4) 400 #define F2FS_IOC_ABORT_VOLATILE_WRITE _IO(F2FS_IOCTL_MAGIC, 5) 401 #define F2FS_IOC_GARBAGE_COLLECT _IOW(F2FS_IOCTL_MAGIC, 6, __u32) 402 #define F2FS_IOC_WRITE_CHECKPOINT _IO(F2FS_IOCTL_MAGIC, 7) 403 #define F2FS_IOC_DEFRAGMENT _IOWR(F2FS_IOCTL_MAGIC, 8, \ 404 struct f2fs_defragment) 405 #define F2FS_IOC_MOVE_RANGE _IOWR(F2FS_IOCTL_MAGIC, 9, \ 406 struct f2fs_move_range) 407 #define F2FS_IOC_FLUSH_DEVICE _IOW(F2FS_IOCTL_MAGIC, 10, \ 408 struct f2fs_flush_device) 409 #define F2FS_IOC_GARBAGE_COLLECT_RANGE _IOW(F2FS_IOCTL_MAGIC, 11, \ 410 struct f2fs_gc_range) 411 #define F2FS_IOC_GET_FEATURES _IOR(F2FS_IOCTL_MAGIC, 12, __u32) 412 #define F2FS_IOC_SET_PIN_FILE _IOW(F2FS_IOCTL_MAGIC, 13, __u32) 413 #define F2FS_IOC_GET_PIN_FILE _IOR(F2FS_IOCTL_MAGIC, 14, __u32) 414 #define F2FS_IOC_PRECACHE_EXTENTS _IO(F2FS_IOCTL_MAGIC, 15) 415 416 #define F2FS_IOC_SET_ENCRYPTION_POLICY FS_IOC_SET_ENCRYPTION_POLICY 417 #define F2FS_IOC_GET_ENCRYPTION_POLICY FS_IOC_GET_ENCRYPTION_POLICY 418 #define F2FS_IOC_GET_ENCRYPTION_PWSALT FS_IOC_GET_ENCRYPTION_PWSALT 419 420 /* 421 * should be same as XFS_IOC_GOINGDOWN. 422 * Flags for going down operation used by FS_IOC_GOINGDOWN 423 */ 424 #define F2FS_IOC_SHUTDOWN _IOR('X', 125, __u32) /* Shutdown */ 425 #define F2FS_GOING_DOWN_FULLSYNC 0x0 /* going down with full sync */ 426 #define F2FS_GOING_DOWN_METASYNC 0x1 /* going down with metadata */ 427 #define F2FS_GOING_DOWN_NOSYNC 0x2 /* going down */ 428 #define F2FS_GOING_DOWN_METAFLUSH 0x3 /* going down with meta flush */ 429 #define F2FS_GOING_DOWN_NEED_FSCK 0x4 /* going down to trigger fsck */ 430 431 #if defined(__KERNEL__) && defined(CONFIG_COMPAT) 432 /* 433 * ioctl commands in 32 bit emulation 434 */ 435 #define F2FS_IOC32_GETFLAGS FS_IOC32_GETFLAGS 436 #define F2FS_IOC32_SETFLAGS FS_IOC32_SETFLAGS 437 #define F2FS_IOC32_GETVERSION FS_IOC32_GETVERSION 438 #endif 439 440 #define F2FS_IOC_FSGETXATTR FS_IOC_FSGETXATTR 441 #define F2FS_IOC_FSSETXATTR FS_IOC_FSSETXATTR 442 443 struct f2fs_gc_range { 444 u32 sync; 445 u64 start; 446 u64 len; 447 }; 448 449 struct f2fs_defragment { 450 u64 start; 451 u64 len; 452 }; 453 454 struct f2fs_move_range { 455 u32 dst_fd; /* destination fd */ 456 u64 pos_in; /* start position in src_fd */ 457 u64 pos_out; /* start position in dst_fd */ 458 u64 len; /* size to move */ 459 }; 460 461 struct f2fs_flush_device { 462 u32 dev_num; /* device number to flush */ 463 u32 segments; /* # of segments to flush */ 464 }; 465 466 /* for inline stuff */ 467 #define DEF_INLINE_RESERVED_SIZE 1 468 static inline int get_extra_isize(struct inode *inode); 469 static inline int get_inline_xattr_addrs(struct inode *inode); 470 #define MAX_INLINE_DATA(inode) (sizeof(__le32) * \ 471 (CUR_ADDRS_PER_INODE(inode) - \ 472 get_inline_xattr_addrs(inode) - \ 473 DEF_INLINE_RESERVED_SIZE)) 474 475 /* for inline dir */ 476 #define NR_INLINE_DENTRY(inode) (MAX_INLINE_DATA(inode) * BITS_PER_BYTE / \ 477 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \ 478 BITS_PER_BYTE + 1)) 479 #define INLINE_DENTRY_BITMAP_SIZE(inode) ((NR_INLINE_DENTRY(inode) + \ 480 BITS_PER_BYTE - 1) / BITS_PER_BYTE) 481 #define INLINE_RESERVED_SIZE(inode) (MAX_INLINE_DATA(inode) - \ 482 ((SIZE_OF_DIR_ENTRY + F2FS_SLOT_LEN) * \ 483 NR_INLINE_DENTRY(inode) + \ 484 INLINE_DENTRY_BITMAP_SIZE(inode))) 485 486 /* 487 * For INODE and NODE manager 488 */ 489 /* for directory operations */ 490 struct f2fs_dentry_ptr { 491 struct inode *inode; 492 void *bitmap; 493 struct f2fs_dir_entry *dentry; 494 __u8 (*filename)[F2FS_SLOT_LEN]; 495 int max; 496 int nr_bitmap; 497 }; 498 499 static inline void make_dentry_ptr_block(struct inode *inode, 500 struct f2fs_dentry_ptr *d, struct f2fs_dentry_block *t) 501 { 502 d->inode = inode; 503 d->max = NR_DENTRY_IN_BLOCK; 504 d->nr_bitmap = SIZE_OF_DENTRY_BITMAP; 505 d->bitmap = t->dentry_bitmap; 506 d->dentry = t->dentry; 507 d->filename = t->filename; 508 } 509 510 static inline void make_dentry_ptr_inline(struct inode *inode, 511 struct f2fs_dentry_ptr *d, void *t) 512 { 513 int entry_cnt = NR_INLINE_DENTRY(inode); 514 int bitmap_size = INLINE_DENTRY_BITMAP_SIZE(inode); 515 int reserved_size = INLINE_RESERVED_SIZE(inode); 516 517 d->inode = inode; 518 d->max = entry_cnt; 519 d->nr_bitmap = bitmap_size; 520 d->bitmap = t; 521 d->dentry = t + bitmap_size + reserved_size; 522 d->filename = t + bitmap_size + reserved_size + 523 SIZE_OF_DIR_ENTRY * entry_cnt; 524 } 525 526 /* 527 * XATTR_NODE_OFFSET stores xattrs to one node block per file keeping -1 528 * as its node offset to distinguish from index node blocks. 529 * But some bits are used to mark the node block. 530 */ 531 #define XATTR_NODE_OFFSET ((((unsigned int)-1) << OFFSET_BIT_SHIFT) \ 532 >> OFFSET_BIT_SHIFT) 533 enum { 534 ALLOC_NODE, /* allocate a new node page if needed */ 535 LOOKUP_NODE, /* look up a node without readahead */ 536 LOOKUP_NODE_RA, /* 537 * look up a node with readahead called 538 * by get_data_block. 539 */ 540 }; 541 542 #define DEFAULT_RETRY_IO_COUNT 8 /* maximum retry read IO count */ 543 544 /* maximum retry quota flush count */ 545 #define DEFAULT_RETRY_QUOTA_FLUSH_COUNT 8 546 547 #define F2FS_LINK_MAX 0xffffffff /* maximum link count per file */ 548 549 #define MAX_DIR_RA_PAGES 4 /* maximum ra pages of dir */ 550 551 /* for in-memory extent cache entry */ 552 #define F2FS_MIN_EXTENT_LEN 64 /* minimum extent length */ 553 554 /* number of extent info in extent cache we try to shrink */ 555 #define EXTENT_CACHE_SHRINK_NUMBER 128 556 557 struct rb_entry { 558 struct rb_node rb_node; /* rb node located in rb-tree */ 559 unsigned int ofs; /* start offset of the entry */ 560 unsigned int len; /* length of the entry */ 561 }; 562 563 struct extent_info { 564 unsigned int fofs; /* start offset in a file */ 565 unsigned int len; /* length of the extent */ 566 u32 blk; /* start block address of the extent */ 567 }; 568 569 struct extent_node { 570 struct rb_node rb_node; /* rb node located in rb-tree */ 571 struct extent_info ei; /* extent info */ 572 struct list_head list; /* node in global extent list of sbi */ 573 struct extent_tree *et; /* extent tree pointer */ 574 }; 575 576 struct extent_tree { 577 nid_t ino; /* inode number */ 578 struct rb_root_cached root; /* root of extent info rb-tree */ 579 struct extent_node *cached_en; /* recently accessed extent node */ 580 struct extent_info largest; /* largested extent info */ 581 struct list_head list; /* to be used by sbi->zombie_list */ 582 rwlock_t lock; /* protect extent info rb-tree */ 583 atomic_t node_cnt; /* # of extent node in rb-tree*/ 584 bool largest_updated; /* largest extent updated */ 585 }; 586 587 /* 588 * This structure is taken from ext4_map_blocks. 589 * 590 * Note that, however, f2fs uses NEW and MAPPED flags for f2fs_map_blocks(). 591 */ 592 #define F2FS_MAP_NEW (1 << BH_New) 593 #define F2FS_MAP_MAPPED (1 << BH_Mapped) 594 #define F2FS_MAP_UNWRITTEN (1 << BH_Unwritten) 595 #define F2FS_MAP_FLAGS (F2FS_MAP_NEW | F2FS_MAP_MAPPED |\ 596 F2FS_MAP_UNWRITTEN) 597 598 struct f2fs_map_blocks { 599 block_t m_pblk; 600 block_t m_lblk; 601 unsigned int m_len; 602 unsigned int m_flags; 603 pgoff_t *m_next_pgofs; /* point next possible non-hole pgofs */ 604 pgoff_t *m_next_extent; /* point to next possible extent */ 605 int m_seg_type; 606 bool m_may_create; /* indicate it is from write path */ 607 }; 608 609 /* for flag in get_data_block */ 610 enum { 611 F2FS_GET_BLOCK_DEFAULT, 612 F2FS_GET_BLOCK_FIEMAP, 613 F2FS_GET_BLOCK_BMAP, 614 F2FS_GET_BLOCK_DIO, 615 F2FS_GET_BLOCK_PRE_DIO, 616 F2FS_GET_BLOCK_PRE_AIO, 617 F2FS_GET_BLOCK_PRECACHE, 618 }; 619 620 /* 621 * i_advise uses FADVISE_XXX_BIT. We can add additional hints later. 622 */ 623 #define FADVISE_COLD_BIT 0x01 624 #define FADVISE_LOST_PINO_BIT 0x02 625 #define FADVISE_ENCRYPT_BIT 0x04 626 #define FADVISE_ENC_NAME_BIT 0x08 627 #define FADVISE_KEEP_SIZE_BIT 0x10 628 #define FADVISE_HOT_BIT 0x20 629 #define FADVISE_VERITY_BIT 0x40 /* reserved */ 630 631 #define FADVISE_MODIFIABLE_BITS (FADVISE_COLD_BIT | FADVISE_HOT_BIT) 632 633 #define file_is_cold(inode) is_file(inode, FADVISE_COLD_BIT) 634 #define file_wrong_pino(inode) is_file(inode, FADVISE_LOST_PINO_BIT) 635 #define file_set_cold(inode) set_file(inode, FADVISE_COLD_BIT) 636 #define file_lost_pino(inode) set_file(inode, FADVISE_LOST_PINO_BIT) 637 #define file_clear_cold(inode) clear_file(inode, FADVISE_COLD_BIT) 638 #define file_got_pino(inode) clear_file(inode, FADVISE_LOST_PINO_BIT) 639 #define file_is_encrypt(inode) is_file(inode, FADVISE_ENCRYPT_BIT) 640 #define file_set_encrypt(inode) set_file(inode, FADVISE_ENCRYPT_BIT) 641 #define file_clear_encrypt(inode) clear_file(inode, FADVISE_ENCRYPT_BIT) 642 #define file_enc_name(inode) is_file(inode, FADVISE_ENC_NAME_BIT) 643 #define file_set_enc_name(inode) set_file(inode, FADVISE_ENC_NAME_BIT) 644 #define file_keep_isize(inode) is_file(inode, FADVISE_KEEP_SIZE_BIT) 645 #define file_set_keep_isize(inode) set_file(inode, FADVISE_KEEP_SIZE_BIT) 646 #define file_is_hot(inode) is_file(inode, FADVISE_HOT_BIT) 647 #define file_set_hot(inode) set_file(inode, FADVISE_HOT_BIT) 648 #define file_clear_hot(inode) clear_file(inode, FADVISE_HOT_BIT) 649 650 #define DEF_DIR_LEVEL 0 651 652 enum { 653 GC_FAILURE_PIN, 654 GC_FAILURE_ATOMIC, 655 MAX_GC_FAILURE 656 }; 657 658 struct f2fs_inode_info { 659 struct inode vfs_inode; /* serve a vfs inode */ 660 unsigned long i_flags; /* keep an inode flags for ioctl */ 661 unsigned char i_advise; /* use to give file attribute hints */ 662 unsigned char i_dir_level; /* use for dentry level for large dir */ 663 unsigned int i_current_depth; /* only for directory depth */ 664 /* for gc failure statistic */ 665 unsigned int i_gc_failures[MAX_GC_FAILURE]; 666 unsigned int i_pino; /* parent inode number */ 667 umode_t i_acl_mode; /* keep file acl mode temporarily */ 668 669 /* Use below internally in f2fs*/ 670 unsigned long flags; /* use to pass per-file flags */ 671 struct rw_semaphore i_sem; /* protect fi info */ 672 atomic_t dirty_pages; /* # of dirty pages */ 673 f2fs_hash_t chash; /* hash value of given file name */ 674 unsigned int clevel; /* maximum level of given file name */ 675 struct task_struct *task; /* lookup and create consistency */ 676 struct task_struct *cp_task; /* separate cp/wb IO stats*/ 677 nid_t i_xattr_nid; /* node id that contains xattrs */ 678 loff_t last_disk_size; /* lastly written file size */ 679 680 #ifdef CONFIG_QUOTA 681 struct dquot *i_dquot[MAXQUOTAS]; 682 683 /* quota space reservation, managed internally by quota code */ 684 qsize_t i_reserved_quota; 685 #endif 686 struct list_head dirty_list; /* dirty list for dirs and files */ 687 struct list_head gdirty_list; /* linked in global dirty list */ 688 struct list_head inmem_ilist; /* list for inmem inodes */ 689 struct list_head inmem_pages; /* inmemory pages managed by f2fs */ 690 struct task_struct *inmem_task; /* store inmemory task */ 691 struct mutex inmem_lock; /* lock for inmemory pages */ 692 struct extent_tree *extent_tree; /* cached extent_tree entry */ 693 694 /* avoid racing between foreground op and gc */ 695 struct rw_semaphore i_gc_rwsem[2]; 696 struct rw_semaphore i_mmap_sem; 697 struct rw_semaphore i_xattr_sem; /* avoid racing between reading and changing EAs */ 698 699 int i_extra_isize; /* size of extra space located in i_addr */ 700 kprojid_t i_projid; /* id for project quota */ 701 int i_inline_xattr_size; /* inline xattr size */ 702 struct timespec64 i_crtime; /* inode creation time */ 703 struct timespec64 i_disk_time[4];/* inode disk times */ 704 }; 705 706 static inline void get_extent_info(struct extent_info *ext, 707 struct f2fs_extent *i_ext) 708 { 709 ext->fofs = le32_to_cpu(i_ext->fofs); 710 ext->blk = le32_to_cpu(i_ext->blk); 711 ext->len = le32_to_cpu(i_ext->len); 712 } 713 714 static inline void set_raw_extent(struct extent_info *ext, 715 struct f2fs_extent *i_ext) 716 { 717 i_ext->fofs = cpu_to_le32(ext->fofs); 718 i_ext->blk = cpu_to_le32(ext->blk); 719 i_ext->len = cpu_to_le32(ext->len); 720 } 721 722 static inline void set_extent_info(struct extent_info *ei, unsigned int fofs, 723 u32 blk, unsigned int len) 724 { 725 ei->fofs = fofs; 726 ei->blk = blk; 727 ei->len = len; 728 } 729 730 static inline bool __is_discard_mergeable(struct discard_info *back, 731 struct discard_info *front, unsigned int max_len) 732 { 733 return (back->lstart + back->len == front->lstart) && 734 (back->len + front->len <= max_len); 735 } 736 737 static inline bool __is_discard_back_mergeable(struct discard_info *cur, 738 struct discard_info *back, unsigned int max_len) 739 { 740 return __is_discard_mergeable(back, cur, max_len); 741 } 742 743 static inline bool __is_discard_front_mergeable(struct discard_info *cur, 744 struct discard_info *front, unsigned int max_len) 745 { 746 return __is_discard_mergeable(cur, front, max_len); 747 } 748 749 static inline bool __is_extent_mergeable(struct extent_info *back, 750 struct extent_info *front) 751 { 752 return (back->fofs + back->len == front->fofs && 753 back->blk + back->len == front->blk); 754 } 755 756 static inline bool __is_back_mergeable(struct extent_info *cur, 757 struct extent_info *back) 758 { 759 return __is_extent_mergeable(back, cur); 760 } 761 762 static inline bool __is_front_mergeable(struct extent_info *cur, 763 struct extent_info *front) 764 { 765 return __is_extent_mergeable(cur, front); 766 } 767 768 extern void f2fs_mark_inode_dirty_sync(struct inode *inode, bool sync); 769 static inline void __try_update_largest_extent(struct extent_tree *et, 770 struct extent_node *en) 771 { 772 if (en->ei.len > et->largest.len) { 773 et->largest = en->ei; 774 et->largest_updated = true; 775 } 776 } 777 778 /* 779 * For free nid management 780 */ 781 enum nid_state { 782 FREE_NID, /* newly added to free nid list */ 783 PREALLOC_NID, /* it is preallocated */ 784 MAX_NID_STATE, 785 }; 786 787 struct f2fs_nm_info { 788 block_t nat_blkaddr; /* base disk address of NAT */ 789 nid_t max_nid; /* maximum possible node ids */ 790 nid_t available_nids; /* # of available node ids */ 791 nid_t next_scan_nid; /* the next nid to be scanned */ 792 unsigned int ram_thresh; /* control the memory footprint */ 793 unsigned int ra_nid_pages; /* # of nid pages to be readaheaded */ 794 unsigned int dirty_nats_ratio; /* control dirty nats ratio threshold */ 795 796 /* NAT cache management */ 797 struct radix_tree_root nat_root;/* root of the nat entry cache */ 798 struct radix_tree_root nat_set_root;/* root of the nat set cache */ 799 struct rw_semaphore nat_tree_lock; /* protect nat_tree_lock */ 800 struct list_head nat_entries; /* cached nat entry list (clean) */ 801 spinlock_t nat_list_lock; /* protect clean nat entry list */ 802 unsigned int nat_cnt; /* the # of cached nat entries */ 803 unsigned int dirty_nat_cnt; /* total num of nat entries in set */ 804 unsigned int nat_blocks; /* # of nat blocks */ 805 806 /* free node ids management */ 807 struct radix_tree_root free_nid_root;/* root of the free_nid cache */ 808 struct list_head free_nid_list; /* list for free nids excluding preallocated nids */ 809 unsigned int nid_cnt[MAX_NID_STATE]; /* the number of free node id */ 810 spinlock_t nid_list_lock; /* protect nid lists ops */ 811 struct mutex build_lock; /* lock for build free nids */ 812 unsigned char **free_nid_bitmap; 813 unsigned char *nat_block_bitmap; 814 unsigned short *free_nid_count; /* free nid count of NAT block */ 815 816 /* for checkpoint */ 817 char *nat_bitmap; /* NAT bitmap pointer */ 818 819 unsigned int nat_bits_blocks; /* # of nat bits blocks */ 820 unsigned char *nat_bits; /* NAT bits blocks */ 821 unsigned char *full_nat_bits; /* full NAT pages */ 822 unsigned char *empty_nat_bits; /* empty NAT pages */ 823 #ifdef CONFIG_F2FS_CHECK_FS 824 char *nat_bitmap_mir; /* NAT bitmap mirror */ 825 #endif 826 int bitmap_size; /* bitmap size */ 827 }; 828 829 /* 830 * this structure is used as one of function parameters. 831 * all the information are dedicated to a given direct node block determined 832 * by the data offset in a file. 833 */ 834 struct dnode_of_data { 835 struct inode *inode; /* vfs inode pointer */ 836 struct page *inode_page; /* its inode page, NULL is possible */ 837 struct page *node_page; /* cached direct node page */ 838 nid_t nid; /* node id of the direct node block */ 839 unsigned int ofs_in_node; /* data offset in the node page */ 840 bool inode_page_locked; /* inode page is locked or not */ 841 bool node_changed; /* is node block changed */ 842 char cur_level; /* level of hole node page */ 843 char max_level; /* level of current page located */ 844 block_t data_blkaddr; /* block address of the node block */ 845 }; 846 847 static inline void set_new_dnode(struct dnode_of_data *dn, struct inode *inode, 848 struct page *ipage, struct page *npage, nid_t nid) 849 { 850 memset(dn, 0, sizeof(*dn)); 851 dn->inode = inode; 852 dn->inode_page = ipage; 853 dn->node_page = npage; 854 dn->nid = nid; 855 } 856 857 /* 858 * For SIT manager 859 * 860 * By default, there are 6 active log areas across the whole main area. 861 * When considering hot and cold data separation to reduce cleaning overhead, 862 * we split 3 for data logs and 3 for node logs as hot, warm, and cold types, 863 * respectively. 864 * In the current design, you should not change the numbers intentionally. 865 * Instead, as a mount option such as active_logs=x, you can use 2, 4, and 6 866 * logs individually according to the underlying devices. (default: 6) 867 * Just in case, on-disk layout covers maximum 16 logs that consist of 8 for 868 * data and 8 for node logs. 869 */ 870 #define NR_CURSEG_DATA_TYPE (3) 871 #define NR_CURSEG_NODE_TYPE (3) 872 #define NR_CURSEG_TYPE (NR_CURSEG_DATA_TYPE + NR_CURSEG_NODE_TYPE) 873 874 enum { 875 CURSEG_HOT_DATA = 0, /* directory entry blocks */ 876 CURSEG_WARM_DATA, /* data blocks */ 877 CURSEG_COLD_DATA, /* multimedia or GCed data blocks */ 878 CURSEG_HOT_NODE, /* direct node blocks of directory files */ 879 CURSEG_WARM_NODE, /* direct node blocks of normal files */ 880 CURSEG_COLD_NODE, /* indirect node blocks */ 881 NO_CHECK_TYPE, 882 }; 883 884 struct flush_cmd { 885 struct completion wait; 886 struct llist_node llnode; 887 nid_t ino; 888 int ret; 889 }; 890 891 struct flush_cmd_control { 892 struct task_struct *f2fs_issue_flush; /* flush thread */ 893 wait_queue_head_t flush_wait_queue; /* waiting queue for wake-up */ 894 atomic_t issued_flush; /* # of issued flushes */ 895 atomic_t queued_flush; /* # of queued flushes */ 896 struct llist_head issue_list; /* list for command issue */ 897 struct llist_node *dispatch_list; /* list for command dispatch */ 898 }; 899 900 struct f2fs_sm_info { 901 struct sit_info *sit_info; /* whole segment information */ 902 struct free_segmap_info *free_info; /* free segment information */ 903 struct dirty_seglist_info *dirty_info; /* dirty segment information */ 904 struct curseg_info *curseg_array; /* active segment information */ 905 906 struct rw_semaphore curseg_lock; /* for preventing curseg change */ 907 908 block_t seg0_blkaddr; /* block address of 0'th segment */ 909 block_t main_blkaddr; /* start block address of main area */ 910 block_t ssa_blkaddr; /* start block address of SSA area */ 911 912 unsigned int segment_count; /* total # of segments */ 913 unsigned int main_segments; /* # of segments in main area */ 914 unsigned int reserved_segments; /* # of reserved segments */ 915 unsigned int ovp_segments; /* # of overprovision segments */ 916 917 /* a threshold to reclaim prefree segments */ 918 unsigned int rec_prefree_segments; 919 920 /* for batched trimming */ 921 unsigned int trim_sections; /* # of sections to trim */ 922 923 struct list_head sit_entry_set; /* sit entry set list */ 924 925 unsigned int ipu_policy; /* in-place-update policy */ 926 unsigned int min_ipu_util; /* in-place-update threshold */ 927 unsigned int min_fsync_blocks; /* threshold for fsync */ 928 unsigned int min_seq_blocks; /* threshold for sequential blocks */ 929 unsigned int min_hot_blocks; /* threshold for hot block allocation */ 930 unsigned int min_ssr_sections; /* threshold to trigger SSR allocation */ 931 932 /* for flush command control */ 933 struct flush_cmd_control *fcc_info; 934 935 /* for discard command control */ 936 struct discard_cmd_control *dcc_info; 937 }; 938 939 /* 940 * For superblock 941 */ 942 /* 943 * COUNT_TYPE for monitoring 944 * 945 * f2fs monitors the number of several block types such as on-writeback, 946 * dirty dentry blocks, dirty node blocks, and dirty meta blocks. 947 */ 948 #define WB_DATA_TYPE(p) (__is_cp_guaranteed(p) ? F2FS_WB_CP_DATA : F2FS_WB_DATA) 949 enum count_type { 950 F2FS_DIRTY_DENTS, 951 F2FS_DIRTY_DATA, 952 F2FS_DIRTY_QDATA, 953 F2FS_DIRTY_NODES, 954 F2FS_DIRTY_META, 955 F2FS_INMEM_PAGES, 956 F2FS_DIRTY_IMETA, 957 F2FS_WB_CP_DATA, 958 F2FS_WB_DATA, 959 F2FS_RD_DATA, 960 F2FS_RD_NODE, 961 F2FS_RD_META, 962 F2FS_DIO_WRITE, 963 F2FS_DIO_READ, 964 NR_COUNT_TYPE, 965 }; 966 967 /* 968 * The below are the page types of bios used in submit_bio(). 969 * The available types are: 970 * DATA User data pages. It operates as async mode. 971 * NODE Node pages. It operates as async mode. 972 * META FS metadata pages such as SIT, NAT, CP. 973 * NR_PAGE_TYPE The number of page types. 974 * META_FLUSH Make sure the previous pages are written 975 * with waiting the bio's completion 976 * ... Only can be used with META. 977 */ 978 #define PAGE_TYPE_OF_BIO(type) ((type) > META ? META : (type)) 979 enum page_type { 980 DATA, 981 NODE, 982 META, 983 NR_PAGE_TYPE, 984 META_FLUSH, 985 INMEM, /* the below types are used by tracepoints only. */ 986 INMEM_DROP, 987 INMEM_INVALIDATE, 988 INMEM_REVOKE, 989 IPU, 990 OPU, 991 }; 992 993 enum temp_type { 994 HOT = 0, /* must be zero for meta bio */ 995 WARM, 996 COLD, 997 NR_TEMP_TYPE, 998 }; 999 1000 enum need_lock_type { 1001 LOCK_REQ = 0, 1002 LOCK_DONE, 1003 LOCK_RETRY, 1004 }; 1005 1006 enum cp_reason_type { 1007 CP_NO_NEEDED, 1008 CP_NON_REGULAR, 1009 CP_HARDLINK, 1010 CP_SB_NEED_CP, 1011 CP_WRONG_PINO, 1012 CP_NO_SPC_ROLL, 1013 CP_NODE_NEED_CP, 1014 CP_FASTBOOT_MODE, 1015 CP_SPEC_LOG_NUM, 1016 CP_RECOVER_DIR, 1017 }; 1018 1019 enum iostat_type { 1020 APP_DIRECT_IO, /* app direct IOs */ 1021 APP_BUFFERED_IO, /* app buffered IOs */ 1022 APP_WRITE_IO, /* app write IOs */ 1023 APP_MAPPED_IO, /* app mapped IOs */ 1024 FS_DATA_IO, /* data IOs from kworker/fsync/reclaimer */ 1025 FS_NODE_IO, /* node IOs from kworker/fsync/reclaimer */ 1026 FS_META_IO, /* meta IOs from kworker/reclaimer */ 1027 FS_GC_DATA_IO, /* data IOs from forground gc */ 1028 FS_GC_NODE_IO, /* node IOs from forground gc */ 1029 FS_CP_DATA_IO, /* data IOs from checkpoint */ 1030 FS_CP_NODE_IO, /* node IOs from checkpoint */ 1031 FS_CP_META_IO, /* meta IOs from checkpoint */ 1032 FS_DISCARD, /* discard */ 1033 NR_IO_TYPE, 1034 }; 1035 1036 struct f2fs_io_info { 1037 struct f2fs_sb_info *sbi; /* f2fs_sb_info pointer */ 1038 nid_t ino; /* inode number */ 1039 enum page_type type; /* contains DATA/NODE/META/META_FLUSH */ 1040 enum temp_type temp; /* contains HOT/WARM/COLD */ 1041 int op; /* contains REQ_OP_ */ 1042 int op_flags; /* req_flag_bits */ 1043 block_t new_blkaddr; /* new block address to be written */ 1044 block_t old_blkaddr; /* old block address before Cow */ 1045 struct page *page; /* page to be written */ 1046 struct page *encrypted_page; /* encrypted page */ 1047 struct list_head list; /* serialize IOs */ 1048 bool submitted; /* indicate IO submission */ 1049 int need_lock; /* indicate we need to lock cp_rwsem */ 1050 bool in_list; /* indicate fio is in io_list */ 1051 bool is_por; /* indicate IO is from recovery or not */ 1052 bool retry; /* need to reallocate block address */ 1053 enum iostat_type io_type; /* io type */ 1054 struct writeback_control *io_wbc; /* writeback control */ 1055 unsigned char version; /* version of the node */ 1056 }; 1057 1058 #define is_read_io(rw) ((rw) == READ) 1059 struct f2fs_bio_info { 1060 struct f2fs_sb_info *sbi; /* f2fs superblock */ 1061 struct bio *bio; /* bios to merge */ 1062 sector_t last_block_in_bio; /* last block number */ 1063 struct f2fs_io_info fio; /* store buffered io info. */ 1064 struct rw_semaphore io_rwsem; /* blocking op for bio */ 1065 spinlock_t io_lock; /* serialize DATA/NODE IOs */ 1066 struct list_head io_list; /* track fios */ 1067 }; 1068 1069 #define FDEV(i) (sbi->devs[i]) 1070 #define RDEV(i) (raw_super->devs[i]) 1071 struct f2fs_dev_info { 1072 struct block_device *bdev; 1073 char path[MAX_PATH_LEN]; 1074 unsigned int total_segments; 1075 block_t start_blk; 1076 block_t end_blk; 1077 #ifdef CONFIG_BLK_DEV_ZONED 1078 unsigned int nr_blkz; /* Total number of zones */ 1079 unsigned long *blkz_seq; /* Bitmap indicating sequential zones */ 1080 #endif 1081 }; 1082 1083 enum inode_type { 1084 DIR_INODE, /* for dirty dir inode */ 1085 FILE_INODE, /* for dirty regular/symlink inode */ 1086 DIRTY_META, /* for all dirtied inode metadata */ 1087 ATOMIC_FILE, /* for all atomic files */ 1088 NR_INODE_TYPE, 1089 }; 1090 1091 /* for inner inode cache management */ 1092 struct inode_management { 1093 struct radix_tree_root ino_root; /* ino entry array */ 1094 spinlock_t ino_lock; /* for ino entry lock */ 1095 struct list_head ino_list; /* inode list head */ 1096 unsigned long ino_num; /* number of entries */ 1097 }; 1098 1099 /* For s_flag in struct f2fs_sb_info */ 1100 enum { 1101 SBI_IS_DIRTY, /* dirty flag for checkpoint */ 1102 SBI_IS_CLOSE, /* specify unmounting */ 1103 SBI_NEED_FSCK, /* need fsck.f2fs to fix */ 1104 SBI_POR_DOING, /* recovery is doing or not */ 1105 SBI_NEED_SB_WRITE, /* need to recover superblock */ 1106 SBI_NEED_CP, /* need to checkpoint */ 1107 SBI_IS_SHUTDOWN, /* shutdown by ioctl */ 1108 SBI_IS_RECOVERED, /* recovered orphan/data */ 1109 SBI_CP_DISABLED, /* CP was disabled last mount */ 1110 SBI_CP_DISABLED_QUICK, /* CP was disabled quickly */ 1111 SBI_QUOTA_NEED_FLUSH, /* need to flush quota info in CP */ 1112 SBI_QUOTA_SKIP_FLUSH, /* skip flushing quota in current CP */ 1113 SBI_QUOTA_NEED_REPAIR, /* quota file may be corrupted */ 1114 }; 1115 1116 enum { 1117 CP_TIME, 1118 REQ_TIME, 1119 DISCARD_TIME, 1120 GC_TIME, 1121 DISABLE_TIME, 1122 UMOUNT_DISCARD_TIMEOUT, 1123 MAX_TIME, 1124 }; 1125 1126 enum { 1127 GC_NORMAL, 1128 GC_IDLE_CB, 1129 GC_IDLE_GREEDY, 1130 GC_URGENT, 1131 }; 1132 1133 enum { 1134 WHINT_MODE_OFF, /* not pass down write hints */ 1135 WHINT_MODE_USER, /* try to pass down hints given by users */ 1136 WHINT_MODE_FS, /* pass down hints with F2FS policy */ 1137 }; 1138 1139 enum { 1140 ALLOC_MODE_DEFAULT, /* stay default */ 1141 ALLOC_MODE_REUSE, /* reuse segments as much as possible */ 1142 }; 1143 1144 enum fsync_mode { 1145 FSYNC_MODE_POSIX, /* fsync follows posix semantics */ 1146 FSYNC_MODE_STRICT, /* fsync behaves in line with ext4 */ 1147 FSYNC_MODE_NOBARRIER, /* fsync behaves nobarrier based on posix */ 1148 }; 1149 1150 #ifdef CONFIG_FS_ENCRYPTION 1151 #define DUMMY_ENCRYPTION_ENABLED(sbi) \ 1152 (unlikely(F2FS_OPTION(sbi).test_dummy_encryption)) 1153 #else 1154 #define DUMMY_ENCRYPTION_ENABLED(sbi) (0) 1155 #endif 1156 1157 struct f2fs_sb_info { 1158 struct super_block *sb; /* pointer to VFS super block */ 1159 struct proc_dir_entry *s_proc; /* proc entry */ 1160 struct f2fs_super_block *raw_super; /* raw super block pointer */ 1161 struct rw_semaphore sb_lock; /* lock for raw super block */ 1162 int valid_super_block; /* valid super block no */ 1163 unsigned long s_flag; /* flags for sbi */ 1164 struct mutex writepages; /* mutex for writepages() */ 1165 1166 #ifdef CONFIG_BLK_DEV_ZONED 1167 unsigned int blocks_per_blkz; /* F2FS blocks per zone */ 1168 unsigned int log_blocks_per_blkz; /* log2 F2FS blocks per zone */ 1169 #endif 1170 1171 /* for node-related operations */ 1172 struct f2fs_nm_info *nm_info; /* node manager */ 1173 struct inode *node_inode; /* cache node blocks */ 1174 1175 /* for segment-related operations */ 1176 struct f2fs_sm_info *sm_info; /* segment manager */ 1177 1178 /* for bio operations */ 1179 struct f2fs_bio_info *write_io[NR_PAGE_TYPE]; /* for write bios */ 1180 /* keep migration IO order for LFS mode */ 1181 struct rw_semaphore io_order_lock; 1182 mempool_t *write_io_dummy; /* Dummy pages */ 1183 1184 /* for checkpoint */ 1185 struct f2fs_checkpoint *ckpt; /* raw checkpoint pointer */ 1186 int cur_cp_pack; /* remain current cp pack */ 1187 spinlock_t cp_lock; /* for flag in ckpt */ 1188 struct inode *meta_inode; /* cache meta blocks */ 1189 struct mutex cp_mutex; /* checkpoint procedure lock */ 1190 struct rw_semaphore cp_rwsem; /* blocking FS operations */ 1191 struct rw_semaphore node_write; /* locking node writes */ 1192 struct rw_semaphore node_change; /* locking node change */ 1193 wait_queue_head_t cp_wait; 1194 unsigned long last_time[MAX_TIME]; /* to store time in jiffies */ 1195 long interval_time[MAX_TIME]; /* to store thresholds */ 1196 1197 struct inode_management im[MAX_INO_ENTRY]; /* manage inode cache */ 1198 1199 spinlock_t fsync_node_lock; /* for node entry lock */ 1200 struct list_head fsync_node_list; /* node list head */ 1201 unsigned int fsync_seg_id; /* sequence id */ 1202 unsigned int fsync_node_num; /* number of node entries */ 1203 1204 /* for orphan inode, use 0'th array */ 1205 unsigned int max_orphans; /* max orphan inodes */ 1206 1207 /* for inode management */ 1208 struct list_head inode_list[NR_INODE_TYPE]; /* dirty inode list */ 1209 spinlock_t inode_lock[NR_INODE_TYPE]; /* for dirty inode list lock */ 1210 1211 /* for extent tree cache */ 1212 struct radix_tree_root extent_tree_root;/* cache extent cache entries */ 1213 struct mutex extent_tree_lock; /* locking extent radix tree */ 1214 struct list_head extent_list; /* lru list for shrinker */ 1215 spinlock_t extent_lock; /* locking extent lru list */ 1216 atomic_t total_ext_tree; /* extent tree count */ 1217 struct list_head zombie_list; /* extent zombie tree list */ 1218 atomic_t total_zombie_tree; /* extent zombie tree count */ 1219 atomic_t total_ext_node; /* extent info count */ 1220 1221 /* basic filesystem units */ 1222 unsigned int log_sectors_per_block; /* log2 sectors per block */ 1223 unsigned int log_blocksize; /* log2 block size */ 1224 unsigned int blocksize; /* block size */ 1225 unsigned int root_ino_num; /* root inode number*/ 1226 unsigned int node_ino_num; /* node inode number*/ 1227 unsigned int meta_ino_num; /* meta inode number*/ 1228 unsigned int log_blocks_per_seg; /* log2 blocks per segment */ 1229 unsigned int blocks_per_seg; /* blocks per segment */ 1230 unsigned int segs_per_sec; /* segments per section */ 1231 unsigned int secs_per_zone; /* sections per zone */ 1232 unsigned int total_sections; /* total section count */ 1233 unsigned int total_node_count; /* total node block count */ 1234 unsigned int total_valid_node_count; /* valid node block count */ 1235 loff_t max_file_blocks; /* max block index of file */ 1236 int dir_level; /* directory level */ 1237 int readdir_ra; /* readahead inode in readdir */ 1238 1239 block_t user_block_count; /* # of user blocks */ 1240 block_t total_valid_block_count; /* # of valid blocks */ 1241 block_t discard_blks; /* discard command candidats */ 1242 block_t last_valid_block_count; /* for recovery */ 1243 block_t reserved_blocks; /* configurable reserved blocks */ 1244 block_t current_reserved_blocks; /* current reserved blocks */ 1245 1246 /* Additional tracking for no checkpoint mode */ 1247 block_t unusable_block_count; /* # of blocks saved by last cp */ 1248 1249 unsigned int nquota_files; /* # of quota sysfile */ 1250 1251 /* # of pages, see count_type */ 1252 atomic_t nr_pages[NR_COUNT_TYPE]; 1253 /* # of allocated blocks */ 1254 struct percpu_counter alloc_valid_block_count; 1255 1256 /* writeback control */ 1257 atomic_t wb_sync_req[META]; /* count # of WB_SYNC threads */ 1258 1259 /* valid inode count */ 1260 struct percpu_counter total_valid_inode_count; 1261 1262 struct f2fs_mount_info mount_opt; /* mount options */ 1263 1264 /* for cleaning operations */ 1265 struct mutex gc_mutex; /* mutex for GC */ 1266 struct f2fs_gc_kthread *gc_thread; /* GC thread */ 1267 unsigned int cur_victim_sec; /* current victim section num */ 1268 unsigned int gc_mode; /* current GC state */ 1269 unsigned int next_victim_seg[2]; /* next segment in victim section */ 1270 /* for skip statistic */ 1271 unsigned long long skipped_atomic_files[2]; /* FG_GC and BG_GC */ 1272 unsigned long long skipped_gc_rwsem; /* FG_GC only */ 1273 1274 /* threshold for gc trials on pinned files */ 1275 u64 gc_pin_file_threshold; 1276 1277 /* maximum # of trials to find a victim segment for SSR and GC */ 1278 unsigned int max_victim_search; 1279 /* migration granularity of garbage collection, unit: segment */ 1280 unsigned int migration_granularity; 1281 1282 /* 1283 * for stat information. 1284 * one is for the LFS mode, and the other is for the SSR mode. 1285 */ 1286 #ifdef CONFIG_F2FS_STAT_FS 1287 struct f2fs_stat_info *stat_info; /* FS status information */ 1288 atomic_t meta_count[META_MAX]; /* # of meta blocks */ 1289 unsigned int segment_count[2]; /* # of allocated segments */ 1290 unsigned int block_count[2]; /* # of allocated blocks */ 1291 atomic_t inplace_count; /* # of inplace update */ 1292 atomic64_t total_hit_ext; /* # of lookup extent cache */ 1293 atomic64_t read_hit_rbtree; /* # of hit rbtree extent node */ 1294 atomic64_t read_hit_largest; /* # of hit largest extent node */ 1295 atomic64_t read_hit_cached; /* # of hit cached extent node */ 1296 atomic_t inline_xattr; /* # of inline_xattr inodes */ 1297 atomic_t inline_inode; /* # of inline_data inodes */ 1298 atomic_t inline_dir; /* # of inline_dentry inodes */ 1299 atomic_t aw_cnt; /* # of atomic writes */ 1300 atomic_t vw_cnt; /* # of volatile writes */ 1301 atomic_t max_aw_cnt; /* max # of atomic writes */ 1302 atomic_t max_vw_cnt; /* max # of volatile writes */ 1303 int bg_gc; /* background gc calls */ 1304 unsigned int io_skip_bggc; /* skip background gc for in-flight IO */ 1305 unsigned int other_skip_bggc; /* skip background gc for other reasons */ 1306 unsigned int ndirty_inode[NR_INODE_TYPE]; /* # of dirty inodes */ 1307 #endif 1308 spinlock_t stat_lock; /* lock for stat operations */ 1309 1310 /* For app/fs IO statistics */ 1311 spinlock_t iostat_lock; 1312 unsigned long long write_iostat[NR_IO_TYPE]; 1313 bool iostat_enable; 1314 1315 /* For sysfs suppport */ 1316 struct kobject s_kobj; 1317 struct completion s_kobj_unregister; 1318 1319 /* For shrinker support */ 1320 struct list_head s_list; 1321 int s_ndevs; /* number of devices */ 1322 struct f2fs_dev_info *devs; /* for device list */ 1323 unsigned int dirty_device; /* for checkpoint data flush */ 1324 spinlock_t dev_lock; /* protect dirty_device */ 1325 struct mutex umount_mutex; 1326 unsigned int shrinker_run_no; 1327 1328 /* For write statistics */ 1329 u64 sectors_written_start; 1330 u64 kbytes_written; 1331 1332 /* Reference to checksum algorithm driver via cryptoapi */ 1333 struct crypto_shash *s_chksum_driver; 1334 1335 /* Precomputed FS UUID checksum for seeding other checksums */ 1336 __u32 s_chksum_seed; 1337 }; 1338 1339 struct f2fs_private_dio { 1340 struct inode *inode; 1341 void *orig_private; 1342 bio_end_io_t *orig_end_io; 1343 bool write; 1344 }; 1345 1346 #ifdef CONFIG_F2FS_FAULT_INJECTION 1347 #define f2fs_show_injection_info(type) \ 1348 printk_ratelimited("%sF2FS-fs : inject %s in %s of %pS\n", \ 1349 KERN_INFO, f2fs_fault_name[type], \ 1350 __func__, __builtin_return_address(0)) 1351 static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type) 1352 { 1353 struct f2fs_fault_info *ffi = &F2FS_OPTION(sbi).fault_info; 1354 1355 if (!ffi->inject_rate) 1356 return false; 1357 1358 if (!IS_FAULT_SET(ffi, type)) 1359 return false; 1360 1361 atomic_inc(&ffi->inject_ops); 1362 if (atomic_read(&ffi->inject_ops) >= ffi->inject_rate) { 1363 atomic_set(&ffi->inject_ops, 0); 1364 return true; 1365 } 1366 return false; 1367 } 1368 #else 1369 #define f2fs_show_injection_info(type) do { } while (0) 1370 static inline bool time_to_inject(struct f2fs_sb_info *sbi, int type) 1371 { 1372 return false; 1373 } 1374 #endif 1375 1376 /* 1377 * Test if the mounted volume is a multi-device volume. 1378 * - For a single regular disk volume, sbi->s_ndevs is 0. 1379 * - For a single zoned disk volume, sbi->s_ndevs is 1. 1380 * - For a multi-device volume, sbi->s_ndevs is always 2 or more. 1381 */ 1382 static inline bool f2fs_is_multi_device(struct f2fs_sb_info *sbi) 1383 { 1384 return sbi->s_ndevs > 1; 1385 } 1386 1387 /* For write statistics. Suppose sector size is 512 bytes, 1388 * and the return value is in kbytes. s is of struct f2fs_sb_info. 1389 */ 1390 #define BD_PART_WRITTEN(s) \ 1391 (((u64)part_stat_read((s)->sb->s_bdev->bd_part, sectors[STAT_WRITE]) - \ 1392 (s)->sectors_written_start) >> 1) 1393 1394 static inline void f2fs_update_time(struct f2fs_sb_info *sbi, int type) 1395 { 1396 unsigned long now = jiffies; 1397 1398 sbi->last_time[type] = now; 1399 1400 /* DISCARD_TIME and GC_TIME are based on REQ_TIME */ 1401 if (type == REQ_TIME) { 1402 sbi->last_time[DISCARD_TIME] = now; 1403 sbi->last_time[GC_TIME] = now; 1404 } 1405 } 1406 1407 static inline bool f2fs_time_over(struct f2fs_sb_info *sbi, int type) 1408 { 1409 unsigned long interval = sbi->interval_time[type] * HZ; 1410 1411 return time_after(jiffies, sbi->last_time[type] + interval); 1412 } 1413 1414 static inline unsigned int f2fs_time_to_wait(struct f2fs_sb_info *sbi, 1415 int type) 1416 { 1417 unsigned long interval = sbi->interval_time[type] * HZ; 1418 unsigned int wait_ms = 0; 1419 long delta; 1420 1421 delta = (sbi->last_time[type] + interval) - jiffies; 1422 if (delta > 0) 1423 wait_ms = jiffies_to_msecs(delta); 1424 1425 return wait_ms; 1426 } 1427 1428 /* 1429 * Inline functions 1430 */ 1431 static inline u32 __f2fs_crc32(struct f2fs_sb_info *sbi, u32 crc, 1432 const void *address, unsigned int length) 1433 { 1434 struct { 1435 struct shash_desc shash; 1436 char ctx[4]; 1437 } desc; 1438 int err; 1439 1440 BUG_ON(crypto_shash_descsize(sbi->s_chksum_driver) != sizeof(desc.ctx)); 1441 1442 desc.shash.tfm = sbi->s_chksum_driver; 1443 *(u32 *)desc.ctx = crc; 1444 1445 err = crypto_shash_update(&desc.shash, address, length); 1446 BUG_ON(err); 1447 1448 return *(u32 *)desc.ctx; 1449 } 1450 1451 static inline u32 f2fs_crc32(struct f2fs_sb_info *sbi, const void *address, 1452 unsigned int length) 1453 { 1454 return __f2fs_crc32(sbi, F2FS_SUPER_MAGIC, address, length); 1455 } 1456 1457 static inline bool f2fs_crc_valid(struct f2fs_sb_info *sbi, __u32 blk_crc, 1458 void *buf, size_t buf_size) 1459 { 1460 return f2fs_crc32(sbi, buf, buf_size) == blk_crc; 1461 } 1462 1463 static inline u32 f2fs_chksum(struct f2fs_sb_info *sbi, u32 crc, 1464 const void *address, unsigned int length) 1465 { 1466 return __f2fs_crc32(sbi, crc, address, length); 1467 } 1468 1469 static inline struct f2fs_inode_info *F2FS_I(struct inode *inode) 1470 { 1471 return container_of(inode, struct f2fs_inode_info, vfs_inode); 1472 } 1473 1474 static inline struct f2fs_sb_info *F2FS_SB(struct super_block *sb) 1475 { 1476 return sb->s_fs_info; 1477 } 1478 1479 static inline struct f2fs_sb_info *F2FS_I_SB(struct inode *inode) 1480 { 1481 return F2FS_SB(inode->i_sb); 1482 } 1483 1484 static inline struct f2fs_sb_info *F2FS_M_SB(struct address_space *mapping) 1485 { 1486 return F2FS_I_SB(mapping->host); 1487 } 1488 1489 static inline struct f2fs_sb_info *F2FS_P_SB(struct page *page) 1490 { 1491 return F2FS_M_SB(page->mapping); 1492 } 1493 1494 static inline struct f2fs_super_block *F2FS_RAW_SUPER(struct f2fs_sb_info *sbi) 1495 { 1496 return (struct f2fs_super_block *)(sbi->raw_super); 1497 } 1498 1499 static inline struct f2fs_checkpoint *F2FS_CKPT(struct f2fs_sb_info *sbi) 1500 { 1501 return (struct f2fs_checkpoint *)(sbi->ckpt); 1502 } 1503 1504 static inline struct f2fs_node *F2FS_NODE(struct page *page) 1505 { 1506 return (struct f2fs_node *)page_address(page); 1507 } 1508 1509 static inline struct f2fs_inode *F2FS_INODE(struct page *page) 1510 { 1511 return &((struct f2fs_node *)page_address(page))->i; 1512 } 1513 1514 static inline struct f2fs_nm_info *NM_I(struct f2fs_sb_info *sbi) 1515 { 1516 return (struct f2fs_nm_info *)(sbi->nm_info); 1517 } 1518 1519 static inline struct f2fs_sm_info *SM_I(struct f2fs_sb_info *sbi) 1520 { 1521 return (struct f2fs_sm_info *)(sbi->sm_info); 1522 } 1523 1524 static inline struct sit_info *SIT_I(struct f2fs_sb_info *sbi) 1525 { 1526 return (struct sit_info *)(SM_I(sbi)->sit_info); 1527 } 1528 1529 static inline struct free_segmap_info *FREE_I(struct f2fs_sb_info *sbi) 1530 { 1531 return (struct free_segmap_info *)(SM_I(sbi)->free_info); 1532 } 1533 1534 static inline struct dirty_seglist_info *DIRTY_I(struct f2fs_sb_info *sbi) 1535 { 1536 return (struct dirty_seglist_info *)(SM_I(sbi)->dirty_info); 1537 } 1538 1539 static inline struct address_space *META_MAPPING(struct f2fs_sb_info *sbi) 1540 { 1541 return sbi->meta_inode->i_mapping; 1542 } 1543 1544 static inline struct address_space *NODE_MAPPING(struct f2fs_sb_info *sbi) 1545 { 1546 return sbi->node_inode->i_mapping; 1547 } 1548 1549 static inline bool is_sbi_flag_set(struct f2fs_sb_info *sbi, unsigned int type) 1550 { 1551 return test_bit(type, &sbi->s_flag); 1552 } 1553 1554 static inline void set_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type) 1555 { 1556 set_bit(type, &sbi->s_flag); 1557 } 1558 1559 static inline void clear_sbi_flag(struct f2fs_sb_info *sbi, unsigned int type) 1560 { 1561 clear_bit(type, &sbi->s_flag); 1562 } 1563 1564 static inline unsigned long long cur_cp_version(struct f2fs_checkpoint *cp) 1565 { 1566 return le64_to_cpu(cp->checkpoint_ver); 1567 } 1568 1569 static inline unsigned long f2fs_qf_ino(struct super_block *sb, int type) 1570 { 1571 if (type < F2FS_MAX_QUOTAS) 1572 return le32_to_cpu(F2FS_SB(sb)->raw_super->qf_ino[type]); 1573 return 0; 1574 } 1575 1576 static inline __u64 cur_cp_crc(struct f2fs_checkpoint *cp) 1577 { 1578 size_t crc_offset = le32_to_cpu(cp->checksum_offset); 1579 return le32_to_cpu(*((__le32 *)((unsigned char *)cp + crc_offset))); 1580 } 1581 1582 static inline bool __is_set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 1583 { 1584 unsigned int ckpt_flags = le32_to_cpu(cp->ckpt_flags); 1585 1586 return ckpt_flags & f; 1587 } 1588 1589 static inline bool is_set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 1590 { 1591 return __is_set_ckpt_flags(F2FS_CKPT(sbi), f); 1592 } 1593 1594 static inline void __set_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 1595 { 1596 unsigned int ckpt_flags; 1597 1598 ckpt_flags = le32_to_cpu(cp->ckpt_flags); 1599 ckpt_flags |= f; 1600 cp->ckpt_flags = cpu_to_le32(ckpt_flags); 1601 } 1602 1603 static inline void set_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 1604 { 1605 unsigned long flags; 1606 1607 spin_lock_irqsave(&sbi->cp_lock, flags); 1608 __set_ckpt_flags(F2FS_CKPT(sbi), f); 1609 spin_unlock_irqrestore(&sbi->cp_lock, flags); 1610 } 1611 1612 static inline void __clear_ckpt_flags(struct f2fs_checkpoint *cp, unsigned int f) 1613 { 1614 unsigned int ckpt_flags; 1615 1616 ckpt_flags = le32_to_cpu(cp->ckpt_flags); 1617 ckpt_flags &= (~f); 1618 cp->ckpt_flags = cpu_to_le32(ckpt_flags); 1619 } 1620 1621 static inline void clear_ckpt_flags(struct f2fs_sb_info *sbi, unsigned int f) 1622 { 1623 unsigned long flags; 1624 1625 spin_lock_irqsave(&sbi->cp_lock, flags); 1626 __clear_ckpt_flags(F2FS_CKPT(sbi), f); 1627 spin_unlock_irqrestore(&sbi->cp_lock, flags); 1628 } 1629 1630 static inline void disable_nat_bits(struct f2fs_sb_info *sbi, bool lock) 1631 { 1632 unsigned long flags; 1633 1634 /* 1635 * In order to re-enable nat_bits we need to call fsck.f2fs by 1636 * set_sbi_flag(sbi, SBI_NEED_FSCK). But it may give huge cost, 1637 * so let's rely on regular fsck or unclean shutdown. 1638 */ 1639 1640 if (lock) 1641 spin_lock_irqsave(&sbi->cp_lock, flags); 1642 __clear_ckpt_flags(F2FS_CKPT(sbi), CP_NAT_BITS_FLAG); 1643 kvfree(NM_I(sbi)->nat_bits); 1644 NM_I(sbi)->nat_bits = NULL; 1645 if (lock) 1646 spin_unlock_irqrestore(&sbi->cp_lock, flags); 1647 } 1648 1649 static inline bool enabled_nat_bits(struct f2fs_sb_info *sbi, 1650 struct cp_control *cpc) 1651 { 1652 bool set = is_set_ckpt_flags(sbi, CP_NAT_BITS_FLAG); 1653 1654 return (cpc) ? (cpc->reason & CP_UMOUNT) && set : set; 1655 } 1656 1657 static inline void f2fs_lock_op(struct f2fs_sb_info *sbi) 1658 { 1659 down_read(&sbi->cp_rwsem); 1660 } 1661 1662 static inline int f2fs_trylock_op(struct f2fs_sb_info *sbi) 1663 { 1664 return down_read_trylock(&sbi->cp_rwsem); 1665 } 1666 1667 static inline void f2fs_unlock_op(struct f2fs_sb_info *sbi) 1668 { 1669 up_read(&sbi->cp_rwsem); 1670 } 1671 1672 static inline void f2fs_lock_all(struct f2fs_sb_info *sbi) 1673 { 1674 down_write(&sbi->cp_rwsem); 1675 } 1676 1677 static inline void f2fs_unlock_all(struct f2fs_sb_info *sbi) 1678 { 1679 up_write(&sbi->cp_rwsem); 1680 } 1681 1682 static inline int __get_cp_reason(struct f2fs_sb_info *sbi) 1683 { 1684 int reason = CP_SYNC; 1685 1686 if (test_opt(sbi, FASTBOOT)) 1687 reason = CP_FASTBOOT; 1688 if (is_sbi_flag_set(sbi, SBI_IS_CLOSE)) 1689 reason = CP_UMOUNT; 1690 return reason; 1691 } 1692 1693 static inline bool __remain_node_summaries(int reason) 1694 { 1695 return (reason & (CP_UMOUNT | CP_FASTBOOT)); 1696 } 1697 1698 static inline bool __exist_node_summaries(struct f2fs_sb_info *sbi) 1699 { 1700 return (is_set_ckpt_flags(sbi, CP_UMOUNT_FLAG) || 1701 is_set_ckpt_flags(sbi, CP_FASTBOOT_FLAG)); 1702 } 1703 1704 /* 1705 * Check whether the inode has blocks or not 1706 */ 1707 static inline int F2FS_HAS_BLOCKS(struct inode *inode) 1708 { 1709 block_t xattr_block = F2FS_I(inode)->i_xattr_nid ? 1 : 0; 1710 1711 return (inode->i_blocks >> F2FS_LOG_SECTORS_PER_BLOCK) > xattr_block; 1712 } 1713 1714 static inline bool f2fs_has_xattr_block(unsigned int ofs) 1715 { 1716 return ofs == XATTR_NODE_OFFSET; 1717 } 1718 1719 static inline bool __allow_reserved_blocks(struct f2fs_sb_info *sbi, 1720 struct inode *inode, bool cap) 1721 { 1722 if (!inode) 1723 return true; 1724 if (!test_opt(sbi, RESERVE_ROOT)) 1725 return false; 1726 if (IS_NOQUOTA(inode)) 1727 return true; 1728 if (uid_eq(F2FS_OPTION(sbi).s_resuid, current_fsuid())) 1729 return true; 1730 if (!gid_eq(F2FS_OPTION(sbi).s_resgid, GLOBAL_ROOT_GID) && 1731 in_group_p(F2FS_OPTION(sbi).s_resgid)) 1732 return true; 1733 if (cap && capable(CAP_SYS_RESOURCE)) 1734 return true; 1735 return false; 1736 } 1737 1738 static inline void f2fs_i_blocks_write(struct inode *, block_t, bool, bool); 1739 static inline int inc_valid_block_count(struct f2fs_sb_info *sbi, 1740 struct inode *inode, blkcnt_t *count) 1741 { 1742 blkcnt_t diff = 0, release = 0; 1743 block_t avail_user_block_count; 1744 int ret; 1745 1746 ret = dquot_reserve_block(inode, *count); 1747 if (ret) 1748 return ret; 1749 1750 if (time_to_inject(sbi, FAULT_BLOCK)) { 1751 f2fs_show_injection_info(FAULT_BLOCK); 1752 release = *count; 1753 goto enospc; 1754 } 1755 1756 /* 1757 * let's increase this in prior to actual block count change in order 1758 * for f2fs_sync_file to avoid data races when deciding checkpoint. 1759 */ 1760 percpu_counter_add(&sbi->alloc_valid_block_count, (*count)); 1761 1762 spin_lock(&sbi->stat_lock); 1763 sbi->total_valid_block_count += (block_t)(*count); 1764 avail_user_block_count = sbi->user_block_count - 1765 sbi->current_reserved_blocks; 1766 1767 if (!__allow_reserved_blocks(sbi, inode, true)) 1768 avail_user_block_count -= F2FS_OPTION(sbi).root_reserved_blocks; 1769 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) 1770 avail_user_block_count -= sbi->unusable_block_count; 1771 if (unlikely(sbi->total_valid_block_count > avail_user_block_count)) { 1772 diff = sbi->total_valid_block_count - avail_user_block_count; 1773 if (diff > *count) 1774 diff = *count; 1775 *count -= diff; 1776 release = diff; 1777 sbi->total_valid_block_count -= diff; 1778 if (!*count) { 1779 spin_unlock(&sbi->stat_lock); 1780 goto enospc; 1781 } 1782 } 1783 spin_unlock(&sbi->stat_lock); 1784 1785 if (unlikely(release)) { 1786 percpu_counter_sub(&sbi->alloc_valid_block_count, release); 1787 dquot_release_reservation_block(inode, release); 1788 } 1789 f2fs_i_blocks_write(inode, *count, true, true); 1790 return 0; 1791 1792 enospc: 1793 percpu_counter_sub(&sbi->alloc_valid_block_count, release); 1794 dquot_release_reservation_block(inode, release); 1795 return -ENOSPC; 1796 } 1797 1798 void f2fs_msg(struct super_block *sb, const char *level, const char *fmt, ...); 1799 static inline void dec_valid_block_count(struct f2fs_sb_info *sbi, 1800 struct inode *inode, 1801 block_t count) 1802 { 1803 blkcnt_t sectors = count << F2FS_LOG_SECTORS_PER_BLOCK; 1804 1805 spin_lock(&sbi->stat_lock); 1806 f2fs_bug_on(sbi, sbi->total_valid_block_count < (block_t) count); 1807 sbi->total_valid_block_count -= (block_t)count; 1808 if (sbi->reserved_blocks && 1809 sbi->current_reserved_blocks < sbi->reserved_blocks) 1810 sbi->current_reserved_blocks = min(sbi->reserved_blocks, 1811 sbi->current_reserved_blocks + count); 1812 spin_unlock(&sbi->stat_lock); 1813 if (unlikely(inode->i_blocks < sectors)) { 1814 f2fs_msg(sbi->sb, KERN_WARNING, 1815 "Inconsistent i_blocks, ino:%lu, iblocks:%llu, sectors:%llu", 1816 inode->i_ino, 1817 (unsigned long long)inode->i_blocks, 1818 (unsigned long long)sectors); 1819 set_sbi_flag(sbi, SBI_NEED_FSCK); 1820 return; 1821 } 1822 f2fs_i_blocks_write(inode, count, false, true); 1823 } 1824 1825 static inline void inc_page_count(struct f2fs_sb_info *sbi, int count_type) 1826 { 1827 atomic_inc(&sbi->nr_pages[count_type]); 1828 1829 if (count_type == F2FS_DIRTY_DENTS || 1830 count_type == F2FS_DIRTY_NODES || 1831 count_type == F2FS_DIRTY_META || 1832 count_type == F2FS_DIRTY_QDATA || 1833 count_type == F2FS_DIRTY_IMETA) 1834 set_sbi_flag(sbi, SBI_IS_DIRTY); 1835 } 1836 1837 static inline void inode_inc_dirty_pages(struct inode *inode) 1838 { 1839 atomic_inc(&F2FS_I(inode)->dirty_pages); 1840 inc_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ? 1841 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA); 1842 if (IS_NOQUOTA(inode)) 1843 inc_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA); 1844 } 1845 1846 static inline void dec_page_count(struct f2fs_sb_info *sbi, int count_type) 1847 { 1848 atomic_dec(&sbi->nr_pages[count_type]); 1849 } 1850 1851 static inline void inode_dec_dirty_pages(struct inode *inode) 1852 { 1853 if (!S_ISDIR(inode->i_mode) && !S_ISREG(inode->i_mode) && 1854 !S_ISLNK(inode->i_mode)) 1855 return; 1856 1857 atomic_dec(&F2FS_I(inode)->dirty_pages); 1858 dec_page_count(F2FS_I_SB(inode), S_ISDIR(inode->i_mode) ? 1859 F2FS_DIRTY_DENTS : F2FS_DIRTY_DATA); 1860 if (IS_NOQUOTA(inode)) 1861 dec_page_count(F2FS_I_SB(inode), F2FS_DIRTY_QDATA); 1862 } 1863 1864 static inline s64 get_pages(struct f2fs_sb_info *sbi, int count_type) 1865 { 1866 return atomic_read(&sbi->nr_pages[count_type]); 1867 } 1868 1869 static inline int get_dirty_pages(struct inode *inode) 1870 { 1871 return atomic_read(&F2FS_I(inode)->dirty_pages); 1872 } 1873 1874 static inline int get_blocktype_secs(struct f2fs_sb_info *sbi, int block_type) 1875 { 1876 unsigned int pages_per_sec = sbi->segs_per_sec * sbi->blocks_per_seg; 1877 unsigned int segs = (get_pages(sbi, block_type) + pages_per_sec - 1) >> 1878 sbi->log_blocks_per_seg; 1879 1880 return segs / sbi->segs_per_sec; 1881 } 1882 1883 static inline block_t valid_user_blocks(struct f2fs_sb_info *sbi) 1884 { 1885 return sbi->total_valid_block_count; 1886 } 1887 1888 static inline block_t discard_blocks(struct f2fs_sb_info *sbi) 1889 { 1890 return sbi->discard_blks; 1891 } 1892 1893 static inline unsigned long __bitmap_size(struct f2fs_sb_info *sbi, int flag) 1894 { 1895 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi); 1896 1897 /* return NAT or SIT bitmap */ 1898 if (flag == NAT_BITMAP) 1899 return le32_to_cpu(ckpt->nat_ver_bitmap_bytesize); 1900 else if (flag == SIT_BITMAP) 1901 return le32_to_cpu(ckpt->sit_ver_bitmap_bytesize); 1902 1903 return 0; 1904 } 1905 1906 static inline block_t __cp_payload(struct f2fs_sb_info *sbi) 1907 { 1908 return le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_payload); 1909 } 1910 1911 static inline void *__bitmap_ptr(struct f2fs_sb_info *sbi, int flag) 1912 { 1913 struct f2fs_checkpoint *ckpt = F2FS_CKPT(sbi); 1914 int offset; 1915 1916 if (is_set_ckpt_flags(sbi, CP_LARGE_NAT_BITMAP_FLAG)) { 1917 offset = (flag == SIT_BITMAP) ? 1918 le32_to_cpu(ckpt->nat_ver_bitmap_bytesize) : 0; 1919 /* 1920 * if large_nat_bitmap feature is enabled, leave checksum 1921 * protection for all nat/sit bitmaps. 1922 */ 1923 return &ckpt->sit_nat_version_bitmap + offset + sizeof(__le32); 1924 } 1925 1926 if (__cp_payload(sbi) > 0) { 1927 if (flag == NAT_BITMAP) 1928 return &ckpt->sit_nat_version_bitmap; 1929 else 1930 return (unsigned char *)ckpt + F2FS_BLKSIZE; 1931 } else { 1932 offset = (flag == NAT_BITMAP) ? 1933 le32_to_cpu(ckpt->sit_ver_bitmap_bytesize) : 0; 1934 return &ckpt->sit_nat_version_bitmap + offset; 1935 } 1936 } 1937 1938 static inline block_t __start_cp_addr(struct f2fs_sb_info *sbi) 1939 { 1940 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr); 1941 1942 if (sbi->cur_cp_pack == 2) 1943 start_addr += sbi->blocks_per_seg; 1944 return start_addr; 1945 } 1946 1947 static inline block_t __start_cp_next_addr(struct f2fs_sb_info *sbi) 1948 { 1949 block_t start_addr = le32_to_cpu(F2FS_RAW_SUPER(sbi)->cp_blkaddr); 1950 1951 if (sbi->cur_cp_pack == 1) 1952 start_addr += sbi->blocks_per_seg; 1953 return start_addr; 1954 } 1955 1956 static inline void __set_cp_next_pack(struct f2fs_sb_info *sbi) 1957 { 1958 sbi->cur_cp_pack = (sbi->cur_cp_pack == 1) ? 2 : 1; 1959 } 1960 1961 static inline block_t __start_sum_addr(struct f2fs_sb_info *sbi) 1962 { 1963 return le32_to_cpu(F2FS_CKPT(sbi)->cp_pack_start_sum); 1964 } 1965 1966 static inline int inc_valid_node_count(struct f2fs_sb_info *sbi, 1967 struct inode *inode, bool is_inode) 1968 { 1969 block_t valid_block_count; 1970 unsigned int valid_node_count; 1971 int err; 1972 1973 if (is_inode) { 1974 if (inode) { 1975 err = dquot_alloc_inode(inode); 1976 if (err) 1977 return err; 1978 } 1979 } else { 1980 err = dquot_reserve_block(inode, 1); 1981 if (err) 1982 return err; 1983 } 1984 1985 if (time_to_inject(sbi, FAULT_BLOCK)) { 1986 f2fs_show_injection_info(FAULT_BLOCK); 1987 goto enospc; 1988 } 1989 1990 spin_lock(&sbi->stat_lock); 1991 1992 valid_block_count = sbi->total_valid_block_count + 1993 sbi->current_reserved_blocks + 1; 1994 1995 if (!__allow_reserved_blocks(sbi, inode, false)) 1996 valid_block_count += F2FS_OPTION(sbi).root_reserved_blocks; 1997 if (unlikely(is_sbi_flag_set(sbi, SBI_CP_DISABLED))) 1998 valid_block_count += sbi->unusable_block_count; 1999 2000 if (unlikely(valid_block_count > sbi->user_block_count)) { 2001 spin_unlock(&sbi->stat_lock); 2002 goto enospc; 2003 } 2004 2005 valid_node_count = sbi->total_valid_node_count + 1; 2006 if (unlikely(valid_node_count > sbi->total_node_count)) { 2007 spin_unlock(&sbi->stat_lock); 2008 goto enospc; 2009 } 2010 2011 sbi->total_valid_node_count++; 2012 sbi->total_valid_block_count++; 2013 spin_unlock(&sbi->stat_lock); 2014 2015 if (inode) { 2016 if (is_inode) 2017 f2fs_mark_inode_dirty_sync(inode, true); 2018 else 2019 f2fs_i_blocks_write(inode, 1, true, true); 2020 } 2021 2022 percpu_counter_inc(&sbi->alloc_valid_block_count); 2023 return 0; 2024 2025 enospc: 2026 if (is_inode) { 2027 if (inode) 2028 dquot_free_inode(inode); 2029 } else { 2030 dquot_release_reservation_block(inode, 1); 2031 } 2032 return -ENOSPC; 2033 } 2034 2035 static inline void dec_valid_node_count(struct f2fs_sb_info *sbi, 2036 struct inode *inode, bool is_inode) 2037 { 2038 spin_lock(&sbi->stat_lock); 2039 2040 f2fs_bug_on(sbi, !sbi->total_valid_block_count); 2041 f2fs_bug_on(sbi, !sbi->total_valid_node_count); 2042 2043 sbi->total_valid_node_count--; 2044 sbi->total_valid_block_count--; 2045 if (sbi->reserved_blocks && 2046 sbi->current_reserved_blocks < sbi->reserved_blocks) 2047 sbi->current_reserved_blocks++; 2048 2049 spin_unlock(&sbi->stat_lock); 2050 2051 if (is_inode) { 2052 dquot_free_inode(inode); 2053 } else { 2054 if (unlikely(inode->i_blocks == 0)) { 2055 f2fs_msg(sbi->sb, KERN_WARNING, 2056 "Inconsistent i_blocks, ino:%lu, iblocks:%llu", 2057 inode->i_ino, 2058 (unsigned long long)inode->i_blocks); 2059 set_sbi_flag(sbi, SBI_NEED_FSCK); 2060 return; 2061 } 2062 f2fs_i_blocks_write(inode, 1, false, true); 2063 } 2064 } 2065 2066 static inline unsigned int valid_node_count(struct f2fs_sb_info *sbi) 2067 { 2068 return sbi->total_valid_node_count; 2069 } 2070 2071 static inline void inc_valid_inode_count(struct f2fs_sb_info *sbi) 2072 { 2073 percpu_counter_inc(&sbi->total_valid_inode_count); 2074 } 2075 2076 static inline void dec_valid_inode_count(struct f2fs_sb_info *sbi) 2077 { 2078 percpu_counter_dec(&sbi->total_valid_inode_count); 2079 } 2080 2081 static inline s64 valid_inode_count(struct f2fs_sb_info *sbi) 2082 { 2083 return percpu_counter_sum_positive(&sbi->total_valid_inode_count); 2084 } 2085 2086 static inline struct page *f2fs_grab_cache_page(struct address_space *mapping, 2087 pgoff_t index, bool for_write) 2088 { 2089 struct page *page; 2090 2091 if (IS_ENABLED(CONFIG_F2FS_FAULT_INJECTION)) { 2092 if (!for_write) 2093 page = find_get_page_flags(mapping, index, 2094 FGP_LOCK | FGP_ACCESSED); 2095 else 2096 page = find_lock_page(mapping, index); 2097 if (page) 2098 return page; 2099 2100 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_ALLOC)) { 2101 f2fs_show_injection_info(FAULT_PAGE_ALLOC); 2102 return NULL; 2103 } 2104 } 2105 2106 if (!for_write) 2107 return grab_cache_page(mapping, index); 2108 return grab_cache_page_write_begin(mapping, index, AOP_FLAG_NOFS); 2109 } 2110 2111 static inline struct page *f2fs_pagecache_get_page( 2112 struct address_space *mapping, pgoff_t index, 2113 int fgp_flags, gfp_t gfp_mask) 2114 { 2115 if (time_to_inject(F2FS_M_SB(mapping), FAULT_PAGE_GET)) { 2116 f2fs_show_injection_info(FAULT_PAGE_GET); 2117 return NULL; 2118 } 2119 2120 return pagecache_get_page(mapping, index, fgp_flags, gfp_mask); 2121 } 2122 2123 static inline void f2fs_copy_page(struct page *src, struct page *dst) 2124 { 2125 char *src_kaddr = kmap(src); 2126 char *dst_kaddr = kmap(dst); 2127 2128 memcpy(dst_kaddr, src_kaddr, PAGE_SIZE); 2129 kunmap(dst); 2130 kunmap(src); 2131 } 2132 2133 static inline void f2fs_put_page(struct page *page, int unlock) 2134 { 2135 if (!page) 2136 return; 2137 2138 if (unlock) { 2139 f2fs_bug_on(F2FS_P_SB(page), !PageLocked(page)); 2140 unlock_page(page); 2141 } 2142 put_page(page); 2143 } 2144 2145 static inline void f2fs_put_dnode(struct dnode_of_data *dn) 2146 { 2147 if (dn->node_page) 2148 f2fs_put_page(dn->node_page, 1); 2149 if (dn->inode_page && dn->node_page != dn->inode_page) 2150 f2fs_put_page(dn->inode_page, 0); 2151 dn->node_page = NULL; 2152 dn->inode_page = NULL; 2153 } 2154 2155 static inline struct kmem_cache *f2fs_kmem_cache_create(const char *name, 2156 size_t size) 2157 { 2158 return kmem_cache_create(name, size, 0, SLAB_RECLAIM_ACCOUNT, NULL); 2159 } 2160 2161 static inline void *f2fs_kmem_cache_alloc(struct kmem_cache *cachep, 2162 gfp_t flags) 2163 { 2164 void *entry; 2165 2166 entry = kmem_cache_alloc(cachep, flags); 2167 if (!entry) 2168 entry = kmem_cache_alloc(cachep, flags | __GFP_NOFAIL); 2169 return entry; 2170 } 2171 2172 static inline struct bio *f2fs_bio_alloc(struct f2fs_sb_info *sbi, 2173 int npages, bool no_fail) 2174 { 2175 struct bio *bio; 2176 2177 if (no_fail) { 2178 /* No failure on bio allocation */ 2179 bio = bio_alloc(GFP_NOIO, npages); 2180 if (!bio) 2181 bio = bio_alloc(GFP_NOIO | __GFP_NOFAIL, npages); 2182 return bio; 2183 } 2184 if (time_to_inject(sbi, FAULT_ALLOC_BIO)) { 2185 f2fs_show_injection_info(FAULT_ALLOC_BIO); 2186 return NULL; 2187 } 2188 2189 return bio_alloc(GFP_KERNEL, npages); 2190 } 2191 2192 static inline bool is_idle(struct f2fs_sb_info *sbi, int type) 2193 { 2194 if (get_pages(sbi, F2FS_RD_DATA) || get_pages(sbi, F2FS_RD_NODE) || 2195 get_pages(sbi, F2FS_RD_META) || get_pages(sbi, F2FS_WB_DATA) || 2196 get_pages(sbi, F2FS_WB_CP_DATA) || 2197 get_pages(sbi, F2FS_DIO_READ) || 2198 get_pages(sbi, F2FS_DIO_WRITE)) 2199 return false; 2200 2201 if (SM_I(sbi) && SM_I(sbi)->dcc_info && 2202 atomic_read(&SM_I(sbi)->dcc_info->queued_discard)) 2203 return false; 2204 2205 if (SM_I(sbi) && SM_I(sbi)->fcc_info && 2206 atomic_read(&SM_I(sbi)->fcc_info->queued_flush)) 2207 return false; 2208 2209 return f2fs_time_over(sbi, type); 2210 } 2211 2212 static inline void f2fs_radix_tree_insert(struct radix_tree_root *root, 2213 unsigned long index, void *item) 2214 { 2215 while (radix_tree_insert(root, index, item)) 2216 cond_resched(); 2217 } 2218 2219 #define RAW_IS_INODE(p) ((p)->footer.nid == (p)->footer.ino) 2220 2221 static inline bool IS_INODE(struct page *page) 2222 { 2223 struct f2fs_node *p = F2FS_NODE(page); 2224 2225 return RAW_IS_INODE(p); 2226 } 2227 2228 static inline int offset_in_addr(struct f2fs_inode *i) 2229 { 2230 return (i->i_inline & F2FS_EXTRA_ATTR) ? 2231 (le16_to_cpu(i->i_extra_isize) / sizeof(__le32)) : 0; 2232 } 2233 2234 static inline __le32 *blkaddr_in_node(struct f2fs_node *node) 2235 { 2236 return RAW_IS_INODE(node) ? node->i.i_addr : node->dn.addr; 2237 } 2238 2239 static inline int f2fs_has_extra_attr(struct inode *inode); 2240 static inline block_t datablock_addr(struct inode *inode, 2241 struct page *node_page, unsigned int offset) 2242 { 2243 struct f2fs_node *raw_node; 2244 __le32 *addr_array; 2245 int base = 0; 2246 bool is_inode = IS_INODE(node_page); 2247 2248 raw_node = F2FS_NODE(node_page); 2249 2250 /* from GC path only */ 2251 if (is_inode) { 2252 if (!inode) 2253 base = offset_in_addr(&raw_node->i); 2254 else if (f2fs_has_extra_attr(inode)) 2255 base = get_extra_isize(inode); 2256 } 2257 2258 addr_array = blkaddr_in_node(raw_node); 2259 return le32_to_cpu(addr_array[base + offset]); 2260 } 2261 2262 static inline int f2fs_test_bit(unsigned int nr, char *addr) 2263 { 2264 int mask; 2265 2266 addr += (nr >> 3); 2267 mask = 1 << (7 - (nr & 0x07)); 2268 return mask & *addr; 2269 } 2270 2271 static inline void f2fs_set_bit(unsigned int nr, char *addr) 2272 { 2273 int mask; 2274 2275 addr += (nr >> 3); 2276 mask = 1 << (7 - (nr & 0x07)); 2277 *addr |= mask; 2278 } 2279 2280 static inline void f2fs_clear_bit(unsigned int nr, char *addr) 2281 { 2282 int mask; 2283 2284 addr += (nr >> 3); 2285 mask = 1 << (7 - (nr & 0x07)); 2286 *addr &= ~mask; 2287 } 2288 2289 static inline int f2fs_test_and_set_bit(unsigned int nr, char *addr) 2290 { 2291 int mask; 2292 int ret; 2293 2294 addr += (nr >> 3); 2295 mask = 1 << (7 - (nr & 0x07)); 2296 ret = mask & *addr; 2297 *addr |= mask; 2298 return ret; 2299 } 2300 2301 static inline int f2fs_test_and_clear_bit(unsigned int nr, char *addr) 2302 { 2303 int mask; 2304 int ret; 2305 2306 addr += (nr >> 3); 2307 mask = 1 << (7 - (nr & 0x07)); 2308 ret = mask & *addr; 2309 *addr &= ~mask; 2310 return ret; 2311 } 2312 2313 static inline void f2fs_change_bit(unsigned int nr, char *addr) 2314 { 2315 int mask; 2316 2317 addr += (nr >> 3); 2318 mask = 1 << (7 - (nr & 0x07)); 2319 *addr ^= mask; 2320 } 2321 2322 /* 2323 * Inode flags 2324 */ 2325 #define F2FS_SECRM_FL 0x00000001 /* Secure deletion */ 2326 #define F2FS_UNRM_FL 0x00000002 /* Undelete */ 2327 #define F2FS_COMPR_FL 0x00000004 /* Compress file */ 2328 #define F2FS_SYNC_FL 0x00000008 /* Synchronous updates */ 2329 #define F2FS_IMMUTABLE_FL 0x00000010 /* Immutable file */ 2330 #define F2FS_APPEND_FL 0x00000020 /* writes to file may only append */ 2331 #define F2FS_NODUMP_FL 0x00000040 /* do not dump file */ 2332 #define F2FS_NOATIME_FL 0x00000080 /* do not update atime */ 2333 /* Reserved for compression usage... */ 2334 #define F2FS_DIRTY_FL 0x00000100 2335 #define F2FS_COMPRBLK_FL 0x00000200 /* One or more compressed clusters */ 2336 #define F2FS_NOCOMPR_FL 0x00000400 /* Don't compress */ 2337 #define F2FS_ENCRYPT_FL 0x00000800 /* encrypted file */ 2338 /* End compression flags --- maybe not all used */ 2339 #define F2FS_INDEX_FL 0x00001000 /* hash-indexed directory */ 2340 #define F2FS_IMAGIC_FL 0x00002000 /* AFS directory */ 2341 #define F2FS_JOURNAL_DATA_FL 0x00004000 /* file data should be journaled */ 2342 #define F2FS_NOTAIL_FL 0x00008000 /* file tail should not be merged */ 2343 #define F2FS_DIRSYNC_FL 0x00010000 /* dirsync behaviour (directories only) */ 2344 #define F2FS_TOPDIR_FL 0x00020000 /* Top of directory hierarchies*/ 2345 #define F2FS_HUGE_FILE_FL 0x00040000 /* Set to each huge file */ 2346 #define F2FS_EXTENTS_FL 0x00080000 /* Inode uses extents */ 2347 #define F2FS_EA_INODE_FL 0x00200000 /* Inode used for large EA */ 2348 #define F2FS_EOFBLOCKS_FL 0x00400000 /* Blocks allocated beyond EOF */ 2349 #define F2FS_NOCOW_FL 0x00800000 /* Do not cow file */ 2350 #define F2FS_INLINE_DATA_FL 0x10000000 /* Inode has inline data. */ 2351 #define F2FS_PROJINHERIT_FL 0x20000000 /* Create with parents projid */ 2352 #define F2FS_RESERVED_FL 0x80000000 /* reserved for ext4 lib */ 2353 2354 #define F2FS_FL_USER_VISIBLE 0x30CBDFFF /* User visible flags */ 2355 #define F2FS_FL_USER_MODIFIABLE 0x204BC0FF /* User modifiable flags */ 2356 2357 /* Flags we can manipulate with through F2FS_IOC_FSSETXATTR */ 2358 #define F2FS_FL_XFLAG_VISIBLE (F2FS_SYNC_FL | \ 2359 F2FS_IMMUTABLE_FL | \ 2360 F2FS_APPEND_FL | \ 2361 F2FS_NODUMP_FL | \ 2362 F2FS_NOATIME_FL | \ 2363 F2FS_PROJINHERIT_FL) 2364 2365 /* Flags that should be inherited by new inodes from their parent. */ 2366 #define F2FS_FL_INHERITED (F2FS_SECRM_FL | F2FS_UNRM_FL | F2FS_COMPR_FL |\ 2367 F2FS_SYNC_FL | F2FS_NODUMP_FL | F2FS_NOATIME_FL |\ 2368 F2FS_NOCOMPR_FL | F2FS_JOURNAL_DATA_FL |\ 2369 F2FS_NOTAIL_FL | F2FS_DIRSYNC_FL |\ 2370 F2FS_PROJINHERIT_FL) 2371 2372 /* Flags that are appropriate for regular files (all but dir-specific ones). */ 2373 #define F2FS_REG_FLMASK (~(F2FS_DIRSYNC_FL | F2FS_TOPDIR_FL)) 2374 2375 /* Flags that are appropriate for non-directories/regular files. */ 2376 #define F2FS_OTHER_FLMASK (F2FS_NODUMP_FL | F2FS_NOATIME_FL) 2377 2378 static inline __u32 f2fs_mask_flags(umode_t mode, __u32 flags) 2379 { 2380 if (S_ISDIR(mode)) 2381 return flags; 2382 else if (S_ISREG(mode)) 2383 return flags & F2FS_REG_FLMASK; 2384 else 2385 return flags & F2FS_OTHER_FLMASK; 2386 } 2387 2388 /* used for f2fs_inode_info->flags */ 2389 enum { 2390 FI_NEW_INODE, /* indicate newly allocated inode */ 2391 FI_DIRTY_INODE, /* indicate inode is dirty or not */ 2392 FI_AUTO_RECOVER, /* indicate inode is recoverable */ 2393 FI_DIRTY_DIR, /* indicate directory has dirty pages */ 2394 FI_INC_LINK, /* need to increment i_nlink */ 2395 FI_ACL_MODE, /* indicate acl mode */ 2396 FI_NO_ALLOC, /* should not allocate any blocks */ 2397 FI_FREE_NID, /* free allocated nide */ 2398 FI_NO_EXTENT, /* not to use the extent cache */ 2399 FI_INLINE_XATTR, /* used for inline xattr */ 2400 FI_INLINE_DATA, /* used for inline data*/ 2401 FI_INLINE_DENTRY, /* used for inline dentry */ 2402 FI_APPEND_WRITE, /* inode has appended data */ 2403 FI_UPDATE_WRITE, /* inode has in-place-update data */ 2404 FI_NEED_IPU, /* used for ipu per file */ 2405 FI_ATOMIC_FILE, /* indicate atomic file */ 2406 FI_ATOMIC_COMMIT, /* indicate the state of atomical committing */ 2407 FI_VOLATILE_FILE, /* indicate volatile file */ 2408 FI_FIRST_BLOCK_WRITTEN, /* indicate #0 data block was written */ 2409 FI_DROP_CACHE, /* drop dirty page cache */ 2410 FI_DATA_EXIST, /* indicate data exists */ 2411 FI_INLINE_DOTS, /* indicate inline dot dentries */ 2412 FI_DO_DEFRAG, /* indicate defragment is running */ 2413 FI_DIRTY_FILE, /* indicate regular/symlink has dirty pages */ 2414 FI_NO_PREALLOC, /* indicate skipped preallocated blocks */ 2415 FI_HOT_DATA, /* indicate file is hot */ 2416 FI_EXTRA_ATTR, /* indicate file has extra attribute */ 2417 FI_PROJ_INHERIT, /* indicate file inherits projectid */ 2418 FI_PIN_FILE, /* indicate file should not be gced */ 2419 FI_ATOMIC_REVOKE_REQUEST, /* request to drop atomic data */ 2420 }; 2421 2422 static inline void __mark_inode_dirty_flag(struct inode *inode, 2423 int flag, bool set) 2424 { 2425 switch (flag) { 2426 case FI_INLINE_XATTR: 2427 case FI_INLINE_DATA: 2428 case FI_INLINE_DENTRY: 2429 case FI_NEW_INODE: 2430 if (set) 2431 return; 2432 /* fall through */ 2433 case FI_DATA_EXIST: 2434 case FI_INLINE_DOTS: 2435 case FI_PIN_FILE: 2436 f2fs_mark_inode_dirty_sync(inode, true); 2437 } 2438 } 2439 2440 static inline void set_inode_flag(struct inode *inode, int flag) 2441 { 2442 if (!test_bit(flag, &F2FS_I(inode)->flags)) 2443 set_bit(flag, &F2FS_I(inode)->flags); 2444 __mark_inode_dirty_flag(inode, flag, true); 2445 } 2446 2447 static inline int is_inode_flag_set(struct inode *inode, int flag) 2448 { 2449 return test_bit(flag, &F2FS_I(inode)->flags); 2450 } 2451 2452 static inline void clear_inode_flag(struct inode *inode, int flag) 2453 { 2454 if (test_bit(flag, &F2FS_I(inode)->flags)) 2455 clear_bit(flag, &F2FS_I(inode)->flags); 2456 __mark_inode_dirty_flag(inode, flag, false); 2457 } 2458 2459 static inline void set_acl_inode(struct inode *inode, umode_t mode) 2460 { 2461 F2FS_I(inode)->i_acl_mode = mode; 2462 set_inode_flag(inode, FI_ACL_MODE); 2463 f2fs_mark_inode_dirty_sync(inode, false); 2464 } 2465 2466 static inline void f2fs_i_links_write(struct inode *inode, bool inc) 2467 { 2468 if (inc) 2469 inc_nlink(inode); 2470 else 2471 drop_nlink(inode); 2472 f2fs_mark_inode_dirty_sync(inode, true); 2473 } 2474 2475 static inline void f2fs_i_blocks_write(struct inode *inode, 2476 block_t diff, bool add, bool claim) 2477 { 2478 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE); 2479 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER); 2480 2481 /* add = 1, claim = 1 should be dquot_reserve_block in pair */ 2482 if (add) { 2483 if (claim) 2484 dquot_claim_block(inode, diff); 2485 else 2486 dquot_alloc_block_nofail(inode, diff); 2487 } else { 2488 dquot_free_block(inode, diff); 2489 } 2490 2491 f2fs_mark_inode_dirty_sync(inode, true); 2492 if (clean || recover) 2493 set_inode_flag(inode, FI_AUTO_RECOVER); 2494 } 2495 2496 static inline void f2fs_i_size_write(struct inode *inode, loff_t i_size) 2497 { 2498 bool clean = !is_inode_flag_set(inode, FI_DIRTY_INODE); 2499 bool recover = is_inode_flag_set(inode, FI_AUTO_RECOVER); 2500 2501 if (i_size_read(inode) == i_size) 2502 return; 2503 2504 i_size_write(inode, i_size); 2505 f2fs_mark_inode_dirty_sync(inode, true); 2506 if (clean || recover) 2507 set_inode_flag(inode, FI_AUTO_RECOVER); 2508 } 2509 2510 static inline void f2fs_i_depth_write(struct inode *inode, unsigned int depth) 2511 { 2512 F2FS_I(inode)->i_current_depth = depth; 2513 f2fs_mark_inode_dirty_sync(inode, true); 2514 } 2515 2516 static inline void f2fs_i_gc_failures_write(struct inode *inode, 2517 unsigned int count) 2518 { 2519 F2FS_I(inode)->i_gc_failures[GC_FAILURE_PIN] = count; 2520 f2fs_mark_inode_dirty_sync(inode, true); 2521 } 2522 2523 static inline void f2fs_i_xnid_write(struct inode *inode, nid_t xnid) 2524 { 2525 F2FS_I(inode)->i_xattr_nid = xnid; 2526 f2fs_mark_inode_dirty_sync(inode, true); 2527 } 2528 2529 static inline void f2fs_i_pino_write(struct inode *inode, nid_t pino) 2530 { 2531 F2FS_I(inode)->i_pino = pino; 2532 f2fs_mark_inode_dirty_sync(inode, true); 2533 } 2534 2535 static inline void get_inline_info(struct inode *inode, struct f2fs_inode *ri) 2536 { 2537 struct f2fs_inode_info *fi = F2FS_I(inode); 2538 2539 if (ri->i_inline & F2FS_INLINE_XATTR) 2540 set_bit(FI_INLINE_XATTR, &fi->flags); 2541 if (ri->i_inline & F2FS_INLINE_DATA) 2542 set_bit(FI_INLINE_DATA, &fi->flags); 2543 if (ri->i_inline & F2FS_INLINE_DENTRY) 2544 set_bit(FI_INLINE_DENTRY, &fi->flags); 2545 if (ri->i_inline & F2FS_DATA_EXIST) 2546 set_bit(FI_DATA_EXIST, &fi->flags); 2547 if (ri->i_inline & F2FS_INLINE_DOTS) 2548 set_bit(FI_INLINE_DOTS, &fi->flags); 2549 if (ri->i_inline & F2FS_EXTRA_ATTR) 2550 set_bit(FI_EXTRA_ATTR, &fi->flags); 2551 if (ri->i_inline & F2FS_PIN_FILE) 2552 set_bit(FI_PIN_FILE, &fi->flags); 2553 } 2554 2555 static inline void set_raw_inline(struct inode *inode, struct f2fs_inode *ri) 2556 { 2557 ri->i_inline = 0; 2558 2559 if (is_inode_flag_set(inode, FI_INLINE_XATTR)) 2560 ri->i_inline |= F2FS_INLINE_XATTR; 2561 if (is_inode_flag_set(inode, FI_INLINE_DATA)) 2562 ri->i_inline |= F2FS_INLINE_DATA; 2563 if (is_inode_flag_set(inode, FI_INLINE_DENTRY)) 2564 ri->i_inline |= F2FS_INLINE_DENTRY; 2565 if (is_inode_flag_set(inode, FI_DATA_EXIST)) 2566 ri->i_inline |= F2FS_DATA_EXIST; 2567 if (is_inode_flag_set(inode, FI_INLINE_DOTS)) 2568 ri->i_inline |= F2FS_INLINE_DOTS; 2569 if (is_inode_flag_set(inode, FI_EXTRA_ATTR)) 2570 ri->i_inline |= F2FS_EXTRA_ATTR; 2571 if (is_inode_flag_set(inode, FI_PIN_FILE)) 2572 ri->i_inline |= F2FS_PIN_FILE; 2573 } 2574 2575 static inline int f2fs_has_extra_attr(struct inode *inode) 2576 { 2577 return is_inode_flag_set(inode, FI_EXTRA_ATTR); 2578 } 2579 2580 static inline int f2fs_has_inline_xattr(struct inode *inode) 2581 { 2582 return is_inode_flag_set(inode, FI_INLINE_XATTR); 2583 } 2584 2585 static inline unsigned int addrs_per_inode(struct inode *inode) 2586 { 2587 unsigned int addrs = CUR_ADDRS_PER_INODE(inode) - 2588 get_inline_xattr_addrs(inode); 2589 return ALIGN_DOWN(addrs, 1); 2590 } 2591 2592 static inline unsigned int addrs_per_block(struct inode *inode) 2593 { 2594 return ALIGN_DOWN(DEF_ADDRS_PER_BLOCK, 1); 2595 } 2596 2597 static inline void *inline_xattr_addr(struct inode *inode, struct page *page) 2598 { 2599 struct f2fs_inode *ri = F2FS_INODE(page); 2600 2601 return (void *)&(ri->i_addr[DEF_ADDRS_PER_INODE - 2602 get_inline_xattr_addrs(inode)]); 2603 } 2604 2605 static inline int inline_xattr_size(struct inode *inode) 2606 { 2607 if (f2fs_has_inline_xattr(inode)) 2608 return get_inline_xattr_addrs(inode) * sizeof(__le32); 2609 return 0; 2610 } 2611 2612 static inline int f2fs_has_inline_data(struct inode *inode) 2613 { 2614 return is_inode_flag_set(inode, FI_INLINE_DATA); 2615 } 2616 2617 static inline int f2fs_exist_data(struct inode *inode) 2618 { 2619 return is_inode_flag_set(inode, FI_DATA_EXIST); 2620 } 2621 2622 static inline int f2fs_has_inline_dots(struct inode *inode) 2623 { 2624 return is_inode_flag_set(inode, FI_INLINE_DOTS); 2625 } 2626 2627 static inline bool f2fs_is_pinned_file(struct inode *inode) 2628 { 2629 return is_inode_flag_set(inode, FI_PIN_FILE); 2630 } 2631 2632 static inline bool f2fs_is_atomic_file(struct inode *inode) 2633 { 2634 return is_inode_flag_set(inode, FI_ATOMIC_FILE); 2635 } 2636 2637 static inline bool f2fs_is_commit_atomic_write(struct inode *inode) 2638 { 2639 return is_inode_flag_set(inode, FI_ATOMIC_COMMIT); 2640 } 2641 2642 static inline bool f2fs_is_volatile_file(struct inode *inode) 2643 { 2644 return is_inode_flag_set(inode, FI_VOLATILE_FILE); 2645 } 2646 2647 static inline bool f2fs_is_first_block_written(struct inode *inode) 2648 { 2649 return is_inode_flag_set(inode, FI_FIRST_BLOCK_WRITTEN); 2650 } 2651 2652 static inline bool f2fs_is_drop_cache(struct inode *inode) 2653 { 2654 return is_inode_flag_set(inode, FI_DROP_CACHE); 2655 } 2656 2657 static inline void *inline_data_addr(struct inode *inode, struct page *page) 2658 { 2659 struct f2fs_inode *ri = F2FS_INODE(page); 2660 int extra_size = get_extra_isize(inode); 2661 2662 return (void *)&(ri->i_addr[extra_size + DEF_INLINE_RESERVED_SIZE]); 2663 } 2664 2665 static inline int f2fs_has_inline_dentry(struct inode *inode) 2666 { 2667 return is_inode_flag_set(inode, FI_INLINE_DENTRY); 2668 } 2669 2670 static inline int is_file(struct inode *inode, int type) 2671 { 2672 return F2FS_I(inode)->i_advise & type; 2673 } 2674 2675 static inline void set_file(struct inode *inode, int type) 2676 { 2677 F2FS_I(inode)->i_advise |= type; 2678 f2fs_mark_inode_dirty_sync(inode, true); 2679 } 2680 2681 static inline void clear_file(struct inode *inode, int type) 2682 { 2683 F2FS_I(inode)->i_advise &= ~type; 2684 f2fs_mark_inode_dirty_sync(inode, true); 2685 } 2686 2687 static inline bool f2fs_skip_inode_update(struct inode *inode, int dsync) 2688 { 2689 bool ret; 2690 2691 if (dsync) { 2692 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 2693 2694 spin_lock(&sbi->inode_lock[DIRTY_META]); 2695 ret = list_empty(&F2FS_I(inode)->gdirty_list); 2696 spin_unlock(&sbi->inode_lock[DIRTY_META]); 2697 return ret; 2698 } 2699 if (!is_inode_flag_set(inode, FI_AUTO_RECOVER) || 2700 file_keep_isize(inode) || 2701 i_size_read(inode) & ~PAGE_MASK) 2702 return false; 2703 2704 if (!timespec64_equal(F2FS_I(inode)->i_disk_time, &inode->i_atime)) 2705 return false; 2706 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 1, &inode->i_ctime)) 2707 return false; 2708 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 2, &inode->i_mtime)) 2709 return false; 2710 if (!timespec64_equal(F2FS_I(inode)->i_disk_time + 3, 2711 &F2FS_I(inode)->i_crtime)) 2712 return false; 2713 2714 down_read(&F2FS_I(inode)->i_sem); 2715 ret = F2FS_I(inode)->last_disk_size == i_size_read(inode); 2716 up_read(&F2FS_I(inode)->i_sem); 2717 2718 return ret; 2719 } 2720 2721 static inline bool f2fs_readonly(struct super_block *sb) 2722 { 2723 return sb_rdonly(sb); 2724 } 2725 2726 static inline bool f2fs_cp_error(struct f2fs_sb_info *sbi) 2727 { 2728 return is_set_ckpt_flags(sbi, CP_ERROR_FLAG); 2729 } 2730 2731 static inline bool is_dot_dotdot(const struct qstr *str) 2732 { 2733 if (str->len == 1 && str->name[0] == '.') 2734 return true; 2735 2736 if (str->len == 2 && str->name[0] == '.' && str->name[1] == '.') 2737 return true; 2738 2739 return false; 2740 } 2741 2742 static inline bool f2fs_may_extent_tree(struct inode *inode) 2743 { 2744 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 2745 2746 if (!test_opt(sbi, EXTENT_CACHE) || 2747 is_inode_flag_set(inode, FI_NO_EXTENT)) 2748 return false; 2749 2750 /* 2751 * for recovered files during mount do not create extents 2752 * if shrinker is not registered. 2753 */ 2754 if (list_empty(&sbi->s_list)) 2755 return false; 2756 2757 return S_ISREG(inode->i_mode); 2758 } 2759 2760 static inline void *f2fs_kmalloc(struct f2fs_sb_info *sbi, 2761 size_t size, gfp_t flags) 2762 { 2763 void *ret; 2764 2765 if (time_to_inject(sbi, FAULT_KMALLOC)) { 2766 f2fs_show_injection_info(FAULT_KMALLOC); 2767 return NULL; 2768 } 2769 2770 ret = kmalloc(size, flags); 2771 if (ret) 2772 return ret; 2773 2774 return kvmalloc(size, flags); 2775 } 2776 2777 static inline void *f2fs_kzalloc(struct f2fs_sb_info *sbi, 2778 size_t size, gfp_t flags) 2779 { 2780 return f2fs_kmalloc(sbi, size, flags | __GFP_ZERO); 2781 } 2782 2783 static inline void *f2fs_kvmalloc(struct f2fs_sb_info *sbi, 2784 size_t size, gfp_t flags) 2785 { 2786 if (time_to_inject(sbi, FAULT_KVMALLOC)) { 2787 f2fs_show_injection_info(FAULT_KVMALLOC); 2788 return NULL; 2789 } 2790 2791 return kvmalloc(size, flags); 2792 } 2793 2794 static inline void *f2fs_kvzalloc(struct f2fs_sb_info *sbi, 2795 size_t size, gfp_t flags) 2796 { 2797 return f2fs_kvmalloc(sbi, size, flags | __GFP_ZERO); 2798 } 2799 2800 static inline int get_extra_isize(struct inode *inode) 2801 { 2802 return F2FS_I(inode)->i_extra_isize / sizeof(__le32); 2803 } 2804 2805 static inline int get_inline_xattr_addrs(struct inode *inode) 2806 { 2807 return F2FS_I(inode)->i_inline_xattr_size; 2808 } 2809 2810 #define f2fs_get_inode_mode(i) \ 2811 ((is_inode_flag_set(i, FI_ACL_MODE)) ? \ 2812 (F2FS_I(i)->i_acl_mode) : ((i)->i_mode)) 2813 2814 #define F2FS_TOTAL_EXTRA_ATTR_SIZE \ 2815 (offsetof(struct f2fs_inode, i_extra_end) - \ 2816 offsetof(struct f2fs_inode, i_extra_isize)) \ 2817 2818 #define F2FS_OLD_ATTRIBUTE_SIZE (offsetof(struct f2fs_inode, i_addr)) 2819 #define F2FS_FITS_IN_INODE(f2fs_inode, extra_isize, field) \ 2820 ((offsetof(typeof(*(f2fs_inode)), field) + \ 2821 sizeof((f2fs_inode)->field)) \ 2822 <= (F2FS_OLD_ATTRIBUTE_SIZE + (extra_isize))) \ 2823 2824 static inline void f2fs_reset_iostat(struct f2fs_sb_info *sbi) 2825 { 2826 int i; 2827 2828 spin_lock(&sbi->iostat_lock); 2829 for (i = 0; i < NR_IO_TYPE; i++) 2830 sbi->write_iostat[i] = 0; 2831 spin_unlock(&sbi->iostat_lock); 2832 } 2833 2834 static inline void f2fs_update_iostat(struct f2fs_sb_info *sbi, 2835 enum iostat_type type, unsigned long long io_bytes) 2836 { 2837 if (!sbi->iostat_enable) 2838 return; 2839 spin_lock(&sbi->iostat_lock); 2840 sbi->write_iostat[type] += io_bytes; 2841 2842 if (type == APP_WRITE_IO || type == APP_DIRECT_IO) 2843 sbi->write_iostat[APP_BUFFERED_IO] = 2844 sbi->write_iostat[APP_WRITE_IO] - 2845 sbi->write_iostat[APP_DIRECT_IO]; 2846 spin_unlock(&sbi->iostat_lock); 2847 } 2848 2849 #define __is_large_section(sbi) ((sbi)->segs_per_sec > 1) 2850 2851 #define __is_meta_io(fio) (PAGE_TYPE_OF_BIO((fio)->type) == META) 2852 2853 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi, 2854 block_t blkaddr, int type); 2855 static inline void verify_blkaddr(struct f2fs_sb_info *sbi, 2856 block_t blkaddr, int type) 2857 { 2858 if (!f2fs_is_valid_blkaddr(sbi, blkaddr, type)) { 2859 f2fs_msg(sbi->sb, KERN_ERR, 2860 "invalid blkaddr: %u, type: %d, run fsck to fix.", 2861 blkaddr, type); 2862 f2fs_bug_on(sbi, 1); 2863 } 2864 } 2865 2866 static inline bool __is_valid_data_blkaddr(block_t blkaddr) 2867 { 2868 if (blkaddr == NEW_ADDR || blkaddr == NULL_ADDR) 2869 return false; 2870 return true; 2871 } 2872 2873 static inline void f2fs_set_page_private(struct page *page, 2874 unsigned long data) 2875 { 2876 if (PagePrivate(page)) 2877 return; 2878 2879 get_page(page); 2880 SetPagePrivate(page); 2881 set_page_private(page, data); 2882 } 2883 2884 static inline void f2fs_clear_page_private(struct page *page) 2885 { 2886 if (!PagePrivate(page)) 2887 return; 2888 2889 set_page_private(page, 0); 2890 ClearPagePrivate(page); 2891 f2fs_put_page(page, 0); 2892 } 2893 2894 /* 2895 * file.c 2896 */ 2897 int f2fs_sync_file(struct file *file, loff_t start, loff_t end, int datasync); 2898 void f2fs_truncate_data_blocks(struct dnode_of_data *dn); 2899 int f2fs_truncate_blocks(struct inode *inode, u64 from, bool lock); 2900 int f2fs_truncate(struct inode *inode); 2901 int f2fs_getattr(const struct path *path, struct kstat *stat, 2902 u32 request_mask, unsigned int flags); 2903 int f2fs_setattr(struct dentry *dentry, struct iattr *attr); 2904 int f2fs_truncate_hole(struct inode *inode, pgoff_t pg_start, pgoff_t pg_end); 2905 void f2fs_truncate_data_blocks_range(struct dnode_of_data *dn, int count); 2906 int f2fs_precache_extents(struct inode *inode); 2907 long f2fs_ioctl(struct file *filp, unsigned int cmd, unsigned long arg); 2908 long f2fs_compat_ioctl(struct file *file, unsigned int cmd, unsigned long arg); 2909 int f2fs_transfer_project_quota(struct inode *inode, kprojid_t kprojid); 2910 int f2fs_pin_file_control(struct inode *inode, bool inc); 2911 2912 /* 2913 * inode.c 2914 */ 2915 void f2fs_set_inode_flags(struct inode *inode); 2916 bool f2fs_inode_chksum_verify(struct f2fs_sb_info *sbi, struct page *page); 2917 void f2fs_inode_chksum_set(struct f2fs_sb_info *sbi, struct page *page); 2918 struct inode *f2fs_iget(struct super_block *sb, unsigned long ino); 2919 struct inode *f2fs_iget_retry(struct super_block *sb, unsigned long ino); 2920 int f2fs_try_to_free_nats(struct f2fs_sb_info *sbi, int nr_shrink); 2921 void f2fs_update_inode(struct inode *inode, struct page *node_page); 2922 void f2fs_update_inode_page(struct inode *inode); 2923 int f2fs_write_inode(struct inode *inode, struct writeback_control *wbc); 2924 void f2fs_evict_inode(struct inode *inode); 2925 void f2fs_handle_failed_inode(struct inode *inode); 2926 2927 /* 2928 * namei.c 2929 */ 2930 int f2fs_update_extension_list(struct f2fs_sb_info *sbi, const char *name, 2931 bool hot, bool set); 2932 struct dentry *f2fs_get_parent(struct dentry *child); 2933 2934 /* 2935 * dir.c 2936 */ 2937 unsigned char f2fs_get_de_type(struct f2fs_dir_entry *de); 2938 struct f2fs_dir_entry *f2fs_find_target_dentry(struct fscrypt_name *fname, 2939 f2fs_hash_t namehash, int *max_slots, 2940 struct f2fs_dentry_ptr *d); 2941 int f2fs_fill_dentries(struct dir_context *ctx, struct f2fs_dentry_ptr *d, 2942 unsigned int start_pos, struct fscrypt_str *fstr); 2943 void f2fs_do_make_empty_dir(struct inode *inode, struct inode *parent, 2944 struct f2fs_dentry_ptr *d); 2945 struct page *f2fs_init_inode_metadata(struct inode *inode, struct inode *dir, 2946 const struct qstr *new_name, 2947 const struct qstr *orig_name, struct page *dpage); 2948 void f2fs_update_parent_metadata(struct inode *dir, struct inode *inode, 2949 unsigned int current_depth); 2950 int f2fs_room_for_filename(const void *bitmap, int slots, int max_slots); 2951 void f2fs_drop_nlink(struct inode *dir, struct inode *inode); 2952 struct f2fs_dir_entry *__f2fs_find_entry(struct inode *dir, 2953 struct fscrypt_name *fname, struct page **res_page); 2954 struct f2fs_dir_entry *f2fs_find_entry(struct inode *dir, 2955 const struct qstr *child, struct page **res_page); 2956 struct f2fs_dir_entry *f2fs_parent_dir(struct inode *dir, struct page **p); 2957 ino_t f2fs_inode_by_name(struct inode *dir, const struct qstr *qstr, 2958 struct page **page); 2959 void f2fs_set_link(struct inode *dir, struct f2fs_dir_entry *de, 2960 struct page *page, struct inode *inode); 2961 void f2fs_update_dentry(nid_t ino, umode_t mode, struct f2fs_dentry_ptr *d, 2962 const struct qstr *name, f2fs_hash_t name_hash, 2963 unsigned int bit_pos); 2964 int f2fs_add_regular_entry(struct inode *dir, const struct qstr *new_name, 2965 const struct qstr *orig_name, 2966 struct inode *inode, nid_t ino, umode_t mode); 2967 int f2fs_add_dentry(struct inode *dir, struct fscrypt_name *fname, 2968 struct inode *inode, nid_t ino, umode_t mode); 2969 int f2fs_do_add_link(struct inode *dir, const struct qstr *name, 2970 struct inode *inode, nid_t ino, umode_t mode); 2971 void f2fs_delete_entry(struct f2fs_dir_entry *dentry, struct page *page, 2972 struct inode *dir, struct inode *inode); 2973 int f2fs_do_tmpfile(struct inode *inode, struct inode *dir); 2974 bool f2fs_empty_dir(struct inode *dir); 2975 2976 static inline int f2fs_add_link(struct dentry *dentry, struct inode *inode) 2977 { 2978 return f2fs_do_add_link(d_inode(dentry->d_parent), &dentry->d_name, 2979 inode, inode->i_ino, inode->i_mode); 2980 } 2981 2982 /* 2983 * super.c 2984 */ 2985 int f2fs_inode_dirtied(struct inode *inode, bool sync); 2986 void f2fs_inode_synced(struct inode *inode); 2987 int f2fs_enable_quota_files(struct f2fs_sb_info *sbi, bool rdonly); 2988 int f2fs_quota_sync(struct super_block *sb, int type); 2989 void f2fs_quota_off_umount(struct super_block *sb); 2990 int f2fs_commit_super(struct f2fs_sb_info *sbi, bool recover); 2991 int f2fs_sync_fs(struct super_block *sb, int sync); 2992 extern __printf(3, 4) 2993 void f2fs_msg(struct super_block *sb, const char *level, const char *fmt, ...); 2994 int f2fs_sanity_check_ckpt(struct f2fs_sb_info *sbi); 2995 2996 /* 2997 * hash.c 2998 */ 2999 f2fs_hash_t f2fs_dentry_hash(const struct qstr *name_info, 3000 struct fscrypt_name *fname); 3001 3002 /* 3003 * node.c 3004 */ 3005 struct dnode_of_data; 3006 struct node_info; 3007 3008 int f2fs_check_nid_range(struct f2fs_sb_info *sbi, nid_t nid); 3009 bool f2fs_available_free_memory(struct f2fs_sb_info *sbi, int type); 3010 bool f2fs_in_warm_node_list(struct f2fs_sb_info *sbi, struct page *page); 3011 void f2fs_init_fsync_node_info(struct f2fs_sb_info *sbi); 3012 void f2fs_del_fsync_node_entry(struct f2fs_sb_info *sbi, struct page *page); 3013 void f2fs_reset_fsync_node_info(struct f2fs_sb_info *sbi); 3014 int f2fs_need_dentry_mark(struct f2fs_sb_info *sbi, nid_t nid); 3015 bool f2fs_is_checkpointed_node(struct f2fs_sb_info *sbi, nid_t nid); 3016 bool f2fs_need_inode_block_update(struct f2fs_sb_info *sbi, nid_t ino); 3017 int f2fs_get_node_info(struct f2fs_sb_info *sbi, nid_t nid, 3018 struct node_info *ni); 3019 pgoff_t f2fs_get_next_page_offset(struct dnode_of_data *dn, pgoff_t pgofs); 3020 int f2fs_get_dnode_of_data(struct dnode_of_data *dn, pgoff_t index, int mode); 3021 int f2fs_truncate_inode_blocks(struct inode *inode, pgoff_t from); 3022 int f2fs_truncate_xattr_node(struct inode *inode); 3023 int f2fs_wait_on_node_pages_writeback(struct f2fs_sb_info *sbi, 3024 unsigned int seq_id); 3025 int f2fs_remove_inode_page(struct inode *inode); 3026 struct page *f2fs_new_inode_page(struct inode *inode); 3027 struct page *f2fs_new_node_page(struct dnode_of_data *dn, unsigned int ofs); 3028 void f2fs_ra_node_page(struct f2fs_sb_info *sbi, nid_t nid); 3029 struct page *f2fs_get_node_page(struct f2fs_sb_info *sbi, pgoff_t nid); 3030 struct page *f2fs_get_node_page_ra(struct page *parent, int start); 3031 int f2fs_move_node_page(struct page *node_page, int gc_type); 3032 int f2fs_fsync_node_pages(struct f2fs_sb_info *sbi, struct inode *inode, 3033 struct writeback_control *wbc, bool atomic, 3034 unsigned int *seq_id); 3035 int f2fs_sync_node_pages(struct f2fs_sb_info *sbi, 3036 struct writeback_control *wbc, 3037 bool do_balance, enum iostat_type io_type); 3038 int f2fs_build_free_nids(struct f2fs_sb_info *sbi, bool sync, bool mount); 3039 bool f2fs_alloc_nid(struct f2fs_sb_info *sbi, nid_t *nid); 3040 void f2fs_alloc_nid_done(struct f2fs_sb_info *sbi, nid_t nid); 3041 void f2fs_alloc_nid_failed(struct f2fs_sb_info *sbi, nid_t nid); 3042 int f2fs_try_to_free_nids(struct f2fs_sb_info *sbi, int nr_shrink); 3043 void f2fs_recover_inline_xattr(struct inode *inode, struct page *page); 3044 int f2fs_recover_xattr_data(struct inode *inode, struct page *page); 3045 int f2fs_recover_inode_page(struct f2fs_sb_info *sbi, struct page *page); 3046 int f2fs_restore_node_summary(struct f2fs_sb_info *sbi, 3047 unsigned int segno, struct f2fs_summary_block *sum); 3048 int f2fs_flush_nat_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3049 int f2fs_build_node_manager(struct f2fs_sb_info *sbi); 3050 void f2fs_destroy_node_manager(struct f2fs_sb_info *sbi); 3051 int __init f2fs_create_node_manager_caches(void); 3052 void f2fs_destroy_node_manager_caches(void); 3053 3054 /* 3055 * segment.c 3056 */ 3057 bool f2fs_need_SSR(struct f2fs_sb_info *sbi); 3058 void f2fs_register_inmem_page(struct inode *inode, struct page *page); 3059 void f2fs_drop_inmem_pages_all(struct f2fs_sb_info *sbi, bool gc_failure); 3060 void f2fs_drop_inmem_pages(struct inode *inode); 3061 void f2fs_drop_inmem_page(struct inode *inode, struct page *page); 3062 int f2fs_commit_inmem_pages(struct inode *inode); 3063 void f2fs_balance_fs(struct f2fs_sb_info *sbi, bool need); 3064 void f2fs_balance_fs_bg(struct f2fs_sb_info *sbi); 3065 int f2fs_issue_flush(struct f2fs_sb_info *sbi, nid_t ino); 3066 int f2fs_create_flush_cmd_control(struct f2fs_sb_info *sbi); 3067 int f2fs_flush_device_cache(struct f2fs_sb_info *sbi); 3068 void f2fs_destroy_flush_cmd_control(struct f2fs_sb_info *sbi, bool free); 3069 void f2fs_invalidate_blocks(struct f2fs_sb_info *sbi, block_t addr); 3070 bool f2fs_is_checkpointed_data(struct f2fs_sb_info *sbi, block_t blkaddr); 3071 void f2fs_drop_discard_cmd(struct f2fs_sb_info *sbi); 3072 void f2fs_stop_discard_thread(struct f2fs_sb_info *sbi); 3073 bool f2fs_issue_discard_timeout(struct f2fs_sb_info *sbi); 3074 void f2fs_clear_prefree_segments(struct f2fs_sb_info *sbi, 3075 struct cp_control *cpc); 3076 void f2fs_dirty_to_prefree(struct f2fs_sb_info *sbi); 3077 int f2fs_disable_cp_again(struct f2fs_sb_info *sbi); 3078 void f2fs_release_discard_addrs(struct f2fs_sb_info *sbi); 3079 int f2fs_npages_for_summary_flush(struct f2fs_sb_info *sbi, bool for_ra); 3080 void f2fs_allocate_new_segments(struct f2fs_sb_info *sbi); 3081 int f2fs_trim_fs(struct f2fs_sb_info *sbi, struct fstrim_range *range); 3082 bool f2fs_exist_trim_candidates(struct f2fs_sb_info *sbi, 3083 struct cp_control *cpc); 3084 struct page *f2fs_get_sum_page(struct f2fs_sb_info *sbi, unsigned int segno); 3085 void f2fs_update_meta_page(struct f2fs_sb_info *sbi, void *src, 3086 block_t blk_addr); 3087 void f2fs_do_write_meta_page(struct f2fs_sb_info *sbi, struct page *page, 3088 enum iostat_type io_type); 3089 void f2fs_do_write_node_page(unsigned int nid, struct f2fs_io_info *fio); 3090 void f2fs_outplace_write_data(struct dnode_of_data *dn, 3091 struct f2fs_io_info *fio); 3092 int f2fs_inplace_write_data(struct f2fs_io_info *fio); 3093 void f2fs_do_replace_block(struct f2fs_sb_info *sbi, struct f2fs_summary *sum, 3094 block_t old_blkaddr, block_t new_blkaddr, 3095 bool recover_curseg, bool recover_newaddr); 3096 void f2fs_replace_block(struct f2fs_sb_info *sbi, struct dnode_of_data *dn, 3097 block_t old_addr, block_t new_addr, 3098 unsigned char version, bool recover_curseg, 3099 bool recover_newaddr); 3100 void f2fs_allocate_data_block(struct f2fs_sb_info *sbi, struct page *page, 3101 block_t old_blkaddr, block_t *new_blkaddr, 3102 struct f2fs_summary *sum, int type, 3103 struct f2fs_io_info *fio, bool add_list); 3104 void f2fs_wait_on_page_writeback(struct page *page, 3105 enum page_type type, bool ordered, bool locked); 3106 void f2fs_wait_on_block_writeback(struct inode *inode, block_t blkaddr); 3107 void f2fs_wait_on_block_writeback_range(struct inode *inode, block_t blkaddr, 3108 block_t len); 3109 void f2fs_write_data_summaries(struct f2fs_sb_info *sbi, block_t start_blk); 3110 void f2fs_write_node_summaries(struct f2fs_sb_info *sbi, block_t start_blk); 3111 int f2fs_lookup_journal_in_cursum(struct f2fs_journal *journal, int type, 3112 unsigned int val, int alloc); 3113 void f2fs_flush_sit_entries(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3114 int f2fs_build_segment_manager(struct f2fs_sb_info *sbi); 3115 void f2fs_destroy_segment_manager(struct f2fs_sb_info *sbi); 3116 int __init f2fs_create_segment_manager_caches(void); 3117 void f2fs_destroy_segment_manager_caches(void); 3118 int f2fs_rw_hint_to_seg_type(enum rw_hint hint); 3119 enum rw_hint f2fs_io_type_to_rw_hint(struct f2fs_sb_info *sbi, 3120 enum page_type type, enum temp_type temp); 3121 3122 /* 3123 * checkpoint.c 3124 */ 3125 void f2fs_stop_checkpoint(struct f2fs_sb_info *sbi, bool end_io); 3126 struct page *f2fs_grab_meta_page(struct f2fs_sb_info *sbi, pgoff_t index); 3127 struct page *f2fs_get_meta_page(struct f2fs_sb_info *sbi, pgoff_t index); 3128 struct page *f2fs_get_meta_page_nofail(struct f2fs_sb_info *sbi, pgoff_t index); 3129 struct page *f2fs_get_tmp_page(struct f2fs_sb_info *sbi, pgoff_t index); 3130 bool f2fs_is_valid_blkaddr(struct f2fs_sb_info *sbi, 3131 block_t blkaddr, int type); 3132 int f2fs_ra_meta_pages(struct f2fs_sb_info *sbi, block_t start, int nrpages, 3133 int type, bool sync); 3134 void f2fs_ra_meta_pages_cond(struct f2fs_sb_info *sbi, pgoff_t index); 3135 long f2fs_sync_meta_pages(struct f2fs_sb_info *sbi, enum page_type type, 3136 long nr_to_write, enum iostat_type io_type); 3137 void f2fs_add_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type); 3138 void f2fs_remove_ino_entry(struct f2fs_sb_info *sbi, nid_t ino, int type); 3139 void f2fs_release_ino_entry(struct f2fs_sb_info *sbi, bool all); 3140 bool f2fs_exist_written_data(struct f2fs_sb_info *sbi, nid_t ino, int mode); 3141 void f2fs_set_dirty_device(struct f2fs_sb_info *sbi, nid_t ino, 3142 unsigned int devidx, int type); 3143 bool f2fs_is_dirty_device(struct f2fs_sb_info *sbi, nid_t ino, 3144 unsigned int devidx, int type); 3145 int f2fs_sync_inode_meta(struct f2fs_sb_info *sbi); 3146 int f2fs_acquire_orphan_inode(struct f2fs_sb_info *sbi); 3147 void f2fs_release_orphan_inode(struct f2fs_sb_info *sbi); 3148 void f2fs_add_orphan_inode(struct inode *inode); 3149 void f2fs_remove_orphan_inode(struct f2fs_sb_info *sbi, nid_t ino); 3150 int f2fs_recover_orphan_inodes(struct f2fs_sb_info *sbi); 3151 int f2fs_get_valid_checkpoint(struct f2fs_sb_info *sbi); 3152 void f2fs_update_dirty_page(struct inode *inode, struct page *page); 3153 void f2fs_remove_dirty_inode(struct inode *inode); 3154 int f2fs_sync_dirty_inodes(struct f2fs_sb_info *sbi, enum inode_type type); 3155 void f2fs_wait_on_all_pages_writeback(struct f2fs_sb_info *sbi); 3156 int f2fs_write_checkpoint(struct f2fs_sb_info *sbi, struct cp_control *cpc); 3157 void f2fs_init_ino_entry_info(struct f2fs_sb_info *sbi); 3158 int __init f2fs_create_checkpoint_caches(void); 3159 void f2fs_destroy_checkpoint_caches(void); 3160 3161 /* 3162 * data.c 3163 */ 3164 int f2fs_init_post_read_processing(void); 3165 void f2fs_destroy_post_read_processing(void); 3166 void f2fs_submit_merged_write(struct f2fs_sb_info *sbi, enum page_type type); 3167 void f2fs_submit_merged_write_cond(struct f2fs_sb_info *sbi, 3168 struct inode *inode, struct page *page, 3169 nid_t ino, enum page_type type); 3170 void f2fs_flush_merged_writes(struct f2fs_sb_info *sbi); 3171 int f2fs_submit_page_bio(struct f2fs_io_info *fio); 3172 void f2fs_submit_page_write(struct f2fs_io_info *fio); 3173 struct block_device *f2fs_target_device(struct f2fs_sb_info *sbi, 3174 block_t blk_addr, struct bio *bio); 3175 int f2fs_target_device_index(struct f2fs_sb_info *sbi, block_t blkaddr); 3176 void f2fs_set_data_blkaddr(struct dnode_of_data *dn); 3177 void f2fs_update_data_blkaddr(struct dnode_of_data *dn, block_t blkaddr); 3178 int f2fs_reserve_new_blocks(struct dnode_of_data *dn, blkcnt_t count); 3179 int f2fs_reserve_new_block(struct dnode_of_data *dn); 3180 int f2fs_get_block(struct dnode_of_data *dn, pgoff_t index); 3181 int f2fs_preallocate_blocks(struct kiocb *iocb, struct iov_iter *from); 3182 int f2fs_reserve_block(struct dnode_of_data *dn, pgoff_t index); 3183 struct page *f2fs_get_read_data_page(struct inode *inode, pgoff_t index, 3184 int op_flags, bool for_write); 3185 struct page *f2fs_find_data_page(struct inode *inode, pgoff_t index); 3186 struct page *f2fs_get_lock_data_page(struct inode *inode, pgoff_t index, 3187 bool for_write); 3188 struct page *f2fs_get_new_data_page(struct inode *inode, 3189 struct page *ipage, pgoff_t index, bool new_i_size); 3190 int f2fs_do_write_data_page(struct f2fs_io_info *fio); 3191 void __do_map_lock(struct f2fs_sb_info *sbi, int flag, bool lock); 3192 int f2fs_map_blocks(struct inode *inode, struct f2fs_map_blocks *map, 3193 int create, int flag); 3194 int f2fs_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, 3195 u64 start, u64 len); 3196 bool f2fs_should_update_inplace(struct inode *inode, struct f2fs_io_info *fio); 3197 bool f2fs_should_update_outplace(struct inode *inode, struct f2fs_io_info *fio); 3198 void f2fs_invalidate_page(struct page *page, unsigned int offset, 3199 unsigned int length); 3200 int f2fs_release_page(struct page *page, gfp_t wait); 3201 #ifdef CONFIG_MIGRATION 3202 int f2fs_migrate_page(struct address_space *mapping, struct page *newpage, 3203 struct page *page, enum migrate_mode mode); 3204 #endif 3205 bool f2fs_overwrite_io(struct inode *inode, loff_t pos, size_t len); 3206 void f2fs_clear_page_cache_dirty_tag(struct page *page); 3207 3208 /* 3209 * gc.c 3210 */ 3211 int f2fs_start_gc_thread(struct f2fs_sb_info *sbi); 3212 void f2fs_stop_gc_thread(struct f2fs_sb_info *sbi); 3213 block_t f2fs_start_bidx_of_node(unsigned int node_ofs, struct inode *inode); 3214 int f2fs_gc(struct f2fs_sb_info *sbi, bool sync, bool background, 3215 unsigned int segno); 3216 void f2fs_build_gc_manager(struct f2fs_sb_info *sbi); 3217 3218 /* 3219 * recovery.c 3220 */ 3221 int f2fs_recover_fsync_data(struct f2fs_sb_info *sbi, bool check_only); 3222 bool f2fs_space_for_roll_forward(struct f2fs_sb_info *sbi); 3223 3224 /* 3225 * debug.c 3226 */ 3227 #ifdef CONFIG_F2FS_STAT_FS 3228 struct f2fs_stat_info { 3229 struct list_head stat_list; 3230 struct f2fs_sb_info *sbi; 3231 int all_area_segs, sit_area_segs, nat_area_segs, ssa_area_segs; 3232 int main_area_segs, main_area_sections, main_area_zones; 3233 unsigned long long hit_largest, hit_cached, hit_rbtree; 3234 unsigned long long hit_total, total_ext; 3235 int ext_tree, zombie_tree, ext_node; 3236 int ndirty_node, ndirty_dent, ndirty_meta, ndirty_imeta; 3237 int ndirty_data, ndirty_qdata; 3238 int inmem_pages; 3239 unsigned int ndirty_dirs, ndirty_files, nquota_files, ndirty_all; 3240 int nats, dirty_nats, sits, dirty_sits; 3241 int free_nids, avail_nids, alloc_nids; 3242 int total_count, utilization; 3243 int bg_gc, nr_wb_cp_data, nr_wb_data; 3244 int nr_rd_data, nr_rd_node, nr_rd_meta; 3245 int nr_dio_read, nr_dio_write; 3246 unsigned int io_skip_bggc, other_skip_bggc; 3247 int nr_flushing, nr_flushed, flush_list_empty; 3248 int nr_discarding, nr_discarded; 3249 int nr_discard_cmd; 3250 unsigned int undiscard_blks; 3251 int inline_xattr, inline_inode, inline_dir, append, update, orphans; 3252 int aw_cnt, max_aw_cnt, vw_cnt, max_vw_cnt; 3253 unsigned int valid_count, valid_node_count, valid_inode_count, discard_blks; 3254 unsigned int bimodal, avg_vblocks; 3255 int util_free, util_valid, util_invalid; 3256 int rsvd_segs, overp_segs; 3257 int dirty_count, node_pages, meta_pages; 3258 int prefree_count, call_count, cp_count, bg_cp_count; 3259 int tot_segs, node_segs, data_segs, free_segs, free_secs; 3260 int bg_node_segs, bg_data_segs; 3261 int tot_blks, data_blks, node_blks; 3262 int bg_data_blks, bg_node_blks; 3263 unsigned long long skipped_atomic_files[2]; 3264 int curseg[NR_CURSEG_TYPE]; 3265 int cursec[NR_CURSEG_TYPE]; 3266 int curzone[NR_CURSEG_TYPE]; 3267 3268 unsigned int meta_count[META_MAX]; 3269 unsigned int segment_count[2]; 3270 unsigned int block_count[2]; 3271 unsigned int inplace_count; 3272 unsigned long long base_mem, cache_mem, page_mem; 3273 }; 3274 3275 static inline struct f2fs_stat_info *F2FS_STAT(struct f2fs_sb_info *sbi) 3276 { 3277 return (struct f2fs_stat_info *)sbi->stat_info; 3278 } 3279 3280 #define stat_inc_cp_count(si) ((si)->cp_count++) 3281 #define stat_inc_bg_cp_count(si) ((si)->bg_cp_count++) 3282 #define stat_inc_call_count(si) ((si)->call_count++) 3283 #define stat_inc_bggc_count(sbi) ((sbi)->bg_gc++) 3284 #define stat_io_skip_bggc_count(sbi) ((sbi)->io_skip_bggc++) 3285 #define stat_other_skip_bggc_count(sbi) ((sbi)->other_skip_bggc++) 3286 #define stat_inc_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]++) 3287 #define stat_dec_dirty_inode(sbi, type) ((sbi)->ndirty_inode[type]--) 3288 #define stat_inc_total_hit(sbi) (atomic64_inc(&(sbi)->total_hit_ext)) 3289 #define stat_inc_rbtree_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_rbtree)) 3290 #define stat_inc_largest_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_largest)) 3291 #define stat_inc_cached_node_hit(sbi) (atomic64_inc(&(sbi)->read_hit_cached)) 3292 #define stat_inc_inline_xattr(inode) \ 3293 do { \ 3294 if (f2fs_has_inline_xattr(inode)) \ 3295 (atomic_inc(&F2FS_I_SB(inode)->inline_xattr)); \ 3296 } while (0) 3297 #define stat_dec_inline_xattr(inode) \ 3298 do { \ 3299 if (f2fs_has_inline_xattr(inode)) \ 3300 (atomic_dec(&F2FS_I_SB(inode)->inline_xattr)); \ 3301 } while (0) 3302 #define stat_inc_inline_inode(inode) \ 3303 do { \ 3304 if (f2fs_has_inline_data(inode)) \ 3305 (atomic_inc(&F2FS_I_SB(inode)->inline_inode)); \ 3306 } while (0) 3307 #define stat_dec_inline_inode(inode) \ 3308 do { \ 3309 if (f2fs_has_inline_data(inode)) \ 3310 (atomic_dec(&F2FS_I_SB(inode)->inline_inode)); \ 3311 } while (0) 3312 #define stat_inc_inline_dir(inode) \ 3313 do { \ 3314 if (f2fs_has_inline_dentry(inode)) \ 3315 (atomic_inc(&F2FS_I_SB(inode)->inline_dir)); \ 3316 } while (0) 3317 #define stat_dec_inline_dir(inode) \ 3318 do { \ 3319 if (f2fs_has_inline_dentry(inode)) \ 3320 (atomic_dec(&F2FS_I_SB(inode)->inline_dir)); \ 3321 } while (0) 3322 #define stat_inc_meta_count(sbi, blkaddr) \ 3323 do { \ 3324 if (blkaddr < SIT_I(sbi)->sit_base_addr) \ 3325 atomic_inc(&(sbi)->meta_count[META_CP]); \ 3326 else if (blkaddr < NM_I(sbi)->nat_blkaddr) \ 3327 atomic_inc(&(sbi)->meta_count[META_SIT]); \ 3328 else if (blkaddr < SM_I(sbi)->ssa_blkaddr) \ 3329 atomic_inc(&(sbi)->meta_count[META_NAT]); \ 3330 else if (blkaddr < SM_I(sbi)->main_blkaddr) \ 3331 atomic_inc(&(sbi)->meta_count[META_SSA]); \ 3332 } while (0) 3333 #define stat_inc_seg_type(sbi, curseg) \ 3334 ((sbi)->segment_count[(curseg)->alloc_type]++) 3335 #define stat_inc_block_count(sbi, curseg) \ 3336 ((sbi)->block_count[(curseg)->alloc_type]++) 3337 #define stat_inc_inplace_blocks(sbi) \ 3338 (atomic_inc(&(sbi)->inplace_count)) 3339 #define stat_inc_atomic_write(inode) \ 3340 (atomic_inc(&F2FS_I_SB(inode)->aw_cnt)) 3341 #define stat_dec_atomic_write(inode) \ 3342 (atomic_dec(&F2FS_I_SB(inode)->aw_cnt)) 3343 #define stat_update_max_atomic_write(inode) \ 3344 do { \ 3345 int cur = atomic_read(&F2FS_I_SB(inode)->aw_cnt); \ 3346 int max = atomic_read(&F2FS_I_SB(inode)->max_aw_cnt); \ 3347 if (cur > max) \ 3348 atomic_set(&F2FS_I_SB(inode)->max_aw_cnt, cur); \ 3349 } while (0) 3350 #define stat_inc_volatile_write(inode) \ 3351 (atomic_inc(&F2FS_I_SB(inode)->vw_cnt)) 3352 #define stat_dec_volatile_write(inode) \ 3353 (atomic_dec(&F2FS_I_SB(inode)->vw_cnt)) 3354 #define stat_update_max_volatile_write(inode) \ 3355 do { \ 3356 int cur = atomic_read(&F2FS_I_SB(inode)->vw_cnt); \ 3357 int max = atomic_read(&F2FS_I_SB(inode)->max_vw_cnt); \ 3358 if (cur > max) \ 3359 atomic_set(&F2FS_I_SB(inode)->max_vw_cnt, cur); \ 3360 } while (0) 3361 #define stat_inc_seg_count(sbi, type, gc_type) \ 3362 do { \ 3363 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 3364 si->tot_segs++; \ 3365 if ((type) == SUM_TYPE_DATA) { \ 3366 si->data_segs++; \ 3367 si->bg_data_segs += (gc_type == BG_GC) ? 1 : 0; \ 3368 } else { \ 3369 si->node_segs++; \ 3370 si->bg_node_segs += (gc_type == BG_GC) ? 1 : 0; \ 3371 } \ 3372 } while (0) 3373 3374 #define stat_inc_tot_blk_count(si, blks) \ 3375 ((si)->tot_blks += (blks)) 3376 3377 #define stat_inc_data_blk_count(sbi, blks, gc_type) \ 3378 do { \ 3379 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 3380 stat_inc_tot_blk_count(si, blks); \ 3381 si->data_blks += (blks); \ 3382 si->bg_data_blks += ((gc_type) == BG_GC) ? (blks) : 0; \ 3383 } while (0) 3384 3385 #define stat_inc_node_blk_count(sbi, blks, gc_type) \ 3386 do { \ 3387 struct f2fs_stat_info *si = F2FS_STAT(sbi); \ 3388 stat_inc_tot_blk_count(si, blks); \ 3389 si->node_blks += (blks); \ 3390 si->bg_node_blks += ((gc_type) == BG_GC) ? (blks) : 0; \ 3391 } while (0) 3392 3393 int f2fs_build_stats(struct f2fs_sb_info *sbi); 3394 void f2fs_destroy_stats(struct f2fs_sb_info *sbi); 3395 void __init f2fs_create_root_stats(void); 3396 void f2fs_destroy_root_stats(void); 3397 #else 3398 #define stat_inc_cp_count(si) do { } while (0) 3399 #define stat_inc_bg_cp_count(si) do { } while (0) 3400 #define stat_inc_call_count(si) do { } while (0) 3401 #define stat_inc_bggc_count(si) do { } while (0) 3402 #define stat_io_skip_bggc_count(sbi) do { } while (0) 3403 #define stat_other_skip_bggc_count(sbi) do { } while (0) 3404 #define stat_inc_dirty_inode(sbi, type) do { } while (0) 3405 #define stat_dec_dirty_inode(sbi, type) do { } while (0) 3406 #define stat_inc_total_hit(sb) do { } while (0) 3407 #define stat_inc_rbtree_node_hit(sb) do { } while (0) 3408 #define stat_inc_largest_node_hit(sbi) do { } while (0) 3409 #define stat_inc_cached_node_hit(sbi) do { } while (0) 3410 #define stat_inc_inline_xattr(inode) do { } while (0) 3411 #define stat_dec_inline_xattr(inode) do { } while (0) 3412 #define stat_inc_inline_inode(inode) do { } while (0) 3413 #define stat_dec_inline_inode(inode) do { } while (0) 3414 #define stat_inc_inline_dir(inode) do { } while (0) 3415 #define stat_dec_inline_dir(inode) do { } while (0) 3416 #define stat_inc_atomic_write(inode) do { } while (0) 3417 #define stat_dec_atomic_write(inode) do { } while (0) 3418 #define stat_update_max_atomic_write(inode) do { } while (0) 3419 #define stat_inc_volatile_write(inode) do { } while (0) 3420 #define stat_dec_volatile_write(inode) do { } while (0) 3421 #define stat_update_max_volatile_write(inode) do { } while (0) 3422 #define stat_inc_meta_count(sbi, blkaddr) do { } while (0) 3423 #define stat_inc_seg_type(sbi, curseg) do { } while (0) 3424 #define stat_inc_block_count(sbi, curseg) do { } while (0) 3425 #define stat_inc_inplace_blocks(sbi) do { } while (0) 3426 #define stat_inc_seg_count(sbi, type, gc_type) do { } while (0) 3427 #define stat_inc_tot_blk_count(si, blks) do { } while (0) 3428 #define stat_inc_data_blk_count(sbi, blks, gc_type) do { } while (0) 3429 #define stat_inc_node_blk_count(sbi, blks, gc_type) do { } while (0) 3430 3431 static inline int f2fs_build_stats(struct f2fs_sb_info *sbi) { return 0; } 3432 static inline void f2fs_destroy_stats(struct f2fs_sb_info *sbi) { } 3433 static inline void __init f2fs_create_root_stats(void) { } 3434 static inline void f2fs_destroy_root_stats(void) { } 3435 #endif 3436 3437 extern const struct file_operations f2fs_dir_operations; 3438 extern const struct file_operations f2fs_file_operations; 3439 extern const struct inode_operations f2fs_file_inode_operations; 3440 extern const struct address_space_operations f2fs_dblock_aops; 3441 extern const struct address_space_operations f2fs_node_aops; 3442 extern const struct address_space_operations f2fs_meta_aops; 3443 extern const struct inode_operations f2fs_dir_inode_operations; 3444 extern const struct inode_operations f2fs_symlink_inode_operations; 3445 extern const struct inode_operations f2fs_encrypted_symlink_inode_operations; 3446 extern const struct inode_operations f2fs_special_inode_operations; 3447 extern struct kmem_cache *f2fs_inode_entry_slab; 3448 3449 /* 3450 * inline.c 3451 */ 3452 bool f2fs_may_inline_data(struct inode *inode); 3453 bool f2fs_may_inline_dentry(struct inode *inode); 3454 void f2fs_do_read_inline_data(struct page *page, struct page *ipage); 3455 void f2fs_truncate_inline_inode(struct inode *inode, 3456 struct page *ipage, u64 from); 3457 int f2fs_read_inline_data(struct inode *inode, struct page *page); 3458 int f2fs_convert_inline_page(struct dnode_of_data *dn, struct page *page); 3459 int f2fs_convert_inline_inode(struct inode *inode); 3460 int f2fs_write_inline_data(struct inode *inode, struct page *page); 3461 bool f2fs_recover_inline_data(struct inode *inode, struct page *npage); 3462 struct f2fs_dir_entry *f2fs_find_in_inline_dir(struct inode *dir, 3463 struct fscrypt_name *fname, struct page **res_page); 3464 int f2fs_make_empty_inline_dir(struct inode *inode, struct inode *parent, 3465 struct page *ipage); 3466 int f2fs_add_inline_entry(struct inode *dir, const struct qstr *new_name, 3467 const struct qstr *orig_name, 3468 struct inode *inode, nid_t ino, umode_t mode); 3469 void f2fs_delete_inline_entry(struct f2fs_dir_entry *dentry, 3470 struct page *page, struct inode *dir, 3471 struct inode *inode); 3472 bool f2fs_empty_inline_dir(struct inode *dir); 3473 int f2fs_read_inline_dir(struct file *file, struct dir_context *ctx, 3474 struct fscrypt_str *fstr); 3475 int f2fs_inline_data_fiemap(struct inode *inode, 3476 struct fiemap_extent_info *fieinfo, 3477 __u64 start, __u64 len); 3478 3479 /* 3480 * shrinker.c 3481 */ 3482 unsigned long f2fs_shrink_count(struct shrinker *shrink, 3483 struct shrink_control *sc); 3484 unsigned long f2fs_shrink_scan(struct shrinker *shrink, 3485 struct shrink_control *sc); 3486 void f2fs_join_shrinker(struct f2fs_sb_info *sbi); 3487 void f2fs_leave_shrinker(struct f2fs_sb_info *sbi); 3488 3489 /* 3490 * extent_cache.c 3491 */ 3492 struct rb_entry *f2fs_lookup_rb_tree(struct rb_root_cached *root, 3493 struct rb_entry *cached_re, unsigned int ofs); 3494 struct rb_node **f2fs_lookup_rb_tree_for_insert(struct f2fs_sb_info *sbi, 3495 struct rb_root_cached *root, 3496 struct rb_node **parent, 3497 unsigned int ofs, bool *leftmost); 3498 struct rb_entry *f2fs_lookup_rb_tree_ret(struct rb_root_cached *root, 3499 struct rb_entry *cached_re, unsigned int ofs, 3500 struct rb_entry **prev_entry, struct rb_entry **next_entry, 3501 struct rb_node ***insert_p, struct rb_node **insert_parent, 3502 bool force, bool *leftmost); 3503 bool f2fs_check_rb_tree_consistence(struct f2fs_sb_info *sbi, 3504 struct rb_root_cached *root); 3505 unsigned int f2fs_shrink_extent_tree(struct f2fs_sb_info *sbi, int nr_shrink); 3506 bool f2fs_init_extent_tree(struct inode *inode, struct f2fs_extent *i_ext); 3507 void f2fs_drop_extent_tree(struct inode *inode); 3508 unsigned int f2fs_destroy_extent_node(struct inode *inode); 3509 void f2fs_destroy_extent_tree(struct inode *inode); 3510 bool f2fs_lookup_extent_cache(struct inode *inode, pgoff_t pgofs, 3511 struct extent_info *ei); 3512 void f2fs_update_extent_cache(struct dnode_of_data *dn); 3513 void f2fs_update_extent_cache_range(struct dnode_of_data *dn, 3514 pgoff_t fofs, block_t blkaddr, unsigned int len); 3515 void f2fs_init_extent_cache_info(struct f2fs_sb_info *sbi); 3516 int __init f2fs_create_extent_cache(void); 3517 void f2fs_destroy_extent_cache(void); 3518 3519 /* 3520 * sysfs.c 3521 */ 3522 int __init f2fs_init_sysfs(void); 3523 void f2fs_exit_sysfs(void); 3524 int f2fs_register_sysfs(struct f2fs_sb_info *sbi); 3525 void f2fs_unregister_sysfs(struct f2fs_sb_info *sbi); 3526 3527 /* 3528 * crypto support 3529 */ 3530 static inline bool f2fs_encrypted_file(struct inode *inode) 3531 { 3532 return IS_ENCRYPTED(inode) && S_ISREG(inode->i_mode); 3533 } 3534 3535 static inline void f2fs_set_encrypted_inode(struct inode *inode) 3536 { 3537 #ifdef CONFIG_FS_ENCRYPTION 3538 file_set_encrypt(inode); 3539 f2fs_set_inode_flags(inode); 3540 #endif 3541 } 3542 3543 /* 3544 * Returns true if the reads of the inode's data need to undergo some 3545 * postprocessing step, like decryption or authenticity verification. 3546 */ 3547 static inline bool f2fs_post_read_required(struct inode *inode) 3548 { 3549 return f2fs_encrypted_file(inode); 3550 } 3551 3552 #define F2FS_FEATURE_FUNCS(name, flagname) \ 3553 static inline int f2fs_sb_has_##name(struct f2fs_sb_info *sbi) \ 3554 { \ 3555 return F2FS_HAS_FEATURE(sbi, F2FS_FEATURE_##flagname); \ 3556 } 3557 3558 F2FS_FEATURE_FUNCS(encrypt, ENCRYPT); 3559 F2FS_FEATURE_FUNCS(blkzoned, BLKZONED); 3560 F2FS_FEATURE_FUNCS(extra_attr, EXTRA_ATTR); 3561 F2FS_FEATURE_FUNCS(project_quota, PRJQUOTA); 3562 F2FS_FEATURE_FUNCS(inode_chksum, INODE_CHKSUM); 3563 F2FS_FEATURE_FUNCS(flexible_inline_xattr, FLEXIBLE_INLINE_XATTR); 3564 F2FS_FEATURE_FUNCS(quota_ino, QUOTA_INO); 3565 F2FS_FEATURE_FUNCS(inode_crtime, INODE_CRTIME); 3566 F2FS_FEATURE_FUNCS(lost_found, LOST_FOUND); 3567 F2FS_FEATURE_FUNCS(sb_chksum, SB_CHKSUM); 3568 3569 #ifdef CONFIG_BLK_DEV_ZONED 3570 static inline bool f2fs_blkz_is_seq(struct f2fs_sb_info *sbi, int devi, 3571 block_t blkaddr) 3572 { 3573 unsigned int zno = blkaddr >> sbi->log_blocks_per_blkz; 3574 3575 return test_bit(zno, FDEV(devi).blkz_seq); 3576 } 3577 #endif 3578 3579 static inline bool f2fs_hw_should_discard(struct f2fs_sb_info *sbi) 3580 { 3581 return f2fs_sb_has_blkzoned(sbi); 3582 } 3583 3584 static inline bool f2fs_bdev_support_discard(struct block_device *bdev) 3585 { 3586 return blk_queue_discard(bdev_get_queue(bdev)) || 3587 bdev_is_zoned(bdev); 3588 } 3589 3590 static inline bool f2fs_hw_support_discard(struct f2fs_sb_info *sbi) 3591 { 3592 int i; 3593 3594 if (!f2fs_is_multi_device(sbi)) 3595 return f2fs_bdev_support_discard(sbi->sb->s_bdev); 3596 3597 for (i = 0; i < sbi->s_ndevs; i++) 3598 if (f2fs_bdev_support_discard(FDEV(i).bdev)) 3599 return true; 3600 return false; 3601 } 3602 3603 static inline bool f2fs_realtime_discard_enable(struct f2fs_sb_info *sbi) 3604 { 3605 return (test_opt(sbi, DISCARD) && f2fs_hw_support_discard(sbi)) || 3606 f2fs_hw_should_discard(sbi); 3607 } 3608 3609 static inline bool f2fs_hw_is_readonly(struct f2fs_sb_info *sbi) 3610 { 3611 int i; 3612 3613 if (!f2fs_is_multi_device(sbi)) 3614 return bdev_read_only(sbi->sb->s_bdev); 3615 3616 for (i = 0; i < sbi->s_ndevs; i++) 3617 if (bdev_read_only(FDEV(i).bdev)) 3618 return true; 3619 return false; 3620 } 3621 3622 3623 static inline void set_opt_mode(struct f2fs_sb_info *sbi, unsigned int mt) 3624 { 3625 clear_opt(sbi, ADAPTIVE); 3626 clear_opt(sbi, LFS); 3627 3628 switch (mt) { 3629 case F2FS_MOUNT_ADAPTIVE: 3630 set_opt(sbi, ADAPTIVE); 3631 break; 3632 case F2FS_MOUNT_LFS: 3633 set_opt(sbi, LFS); 3634 break; 3635 } 3636 } 3637 3638 static inline bool f2fs_may_encrypt(struct inode *inode) 3639 { 3640 #ifdef CONFIG_FS_ENCRYPTION 3641 umode_t mode = inode->i_mode; 3642 3643 return (S_ISREG(mode) || S_ISDIR(mode) || S_ISLNK(mode)); 3644 #else 3645 return false; 3646 #endif 3647 } 3648 3649 static inline int block_unaligned_IO(struct inode *inode, 3650 struct kiocb *iocb, struct iov_iter *iter) 3651 { 3652 unsigned int i_blkbits = READ_ONCE(inode->i_blkbits); 3653 unsigned int blocksize_mask = (1 << i_blkbits) - 1; 3654 loff_t offset = iocb->ki_pos; 3655 unsigned long align = offset | iov_iter_alignment(iter); 3656 3657 return align & blocksize_mask; 3658 } 3659 3660 static inline int allow_outplace_dio(struct inode *inode, 3661 struct kiocb *iocb, struct iov_iter *iter) 3662 { 3663 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 3664 int rw = iov_iter_rw(iter); 3665 3666 return (test_opt(sbi, LFS) && (rw == WRITE) && 3667 !block_unaligned_IO(inode, iocb, iter)); 3668 } 3669 3670 static inline bool f2fs_force_buffered_io(struct inode *inode, 3671 struct kiocb *iocb, struct iov_iter *iter) 3672 { 3673 struct f2fs_sb_info *sbi = F2FS_I_SB(inode); 3674 int rw = iov_iter_rw(iter); 3675 3676 if (f2fs_post_read_required(inode)) 3677 return true; 3678 if (f2fs_is_multi_device(sbi)) 3679 return true; 3680 /* 3681 * for blkzoned device, fallback direct IO to buffered IO, so 3682 * all IOs can be serialized by log-structured write. 3683 */ 3684 if (f2fs_sb_has_blkzoned(sbi)) 3685 return true; 3686 if (test_opt(sbi, LFS) && (rw == WRITE) && 3687 block_unaligned_IO(inode, iocb, iter)) 3688 return true; 3689 if (is_sbi_flag_set(F2FS_I_SB(inode), SBI_CP_DISABLED)) 3690 return true; 3691 3692 return false; 3693 } 3694 3695 #ifdef CONFIG_F2FS_FAULT_INJECTION 3696 extern void f2fs_build_fault_attr(struct f2fs_sb_info *sbi, unsigned int rate, 3697 unsigned int type); 3698 #else 3699 #define f2fs_build_fault_attr(sbi, rate, type) do { } while (0) 3700 #endif 3701 3702 static inline bool is_journalled_quota(struct f2fs_sb_info *sbi) 3703 { 3704 #ifdef CONFIG_QUOTA 3705 if (f2fs_sb_has_quota_ino(sbi)) 3706 return true; 3707 if (F2FS_OPTION(sbi).s_qf_names[USRQUOTA] || 3708 F2FS_OPTION(sbi).s_qf_names[GRPQUOTA] || 3709 F2FS_OPTION(sbi).s_qf_names[PRJQUOTA]) 3710 return true; 3711 #endif 3712 return false; 3713 } 3714 3715 #endif /* _LINUX_F2FS_H */ 3716