1 #ifndef _LINUX_FS_H 2 #define _LINUX_FS_H 3 4 #include <linux/linkage.h> 5 #include <linux/wait.h> 6 #include <linux/kdev_t.h> 7 #include <linux/dcache.h> 8 #include <linux/path.h> 9 #include <linux/stat.h> 10 #include <linux/cache.h> 11 #include <linux/list.h> 12 #include <linux/list_lru.h> 13 #include <linux/llist.h> 14 #include <linux/radix-tree.h> 15 #include <linux/rbtree.h> 16 #include <linux/init.h> 17 #include <linux/pid.h> 18 #include <linux/bug.h> 19 #include <linux/mutex.h> 20 #include <linux/rwsem.h> 21 #include <linux/capability.h> 22 #include <linux/semaphore.h> 23 #include <linux/fiemap.h> 24 #include <linux/rculist_bl.h> 25 #include <linux/atomic.h> 26 #include <linux/shrinker.h> 27 #include <linux/migrate_mode.h> 28 #include <linux/uidgid.h> 29 #include <linux/lockdep.h> 30 #include <linux/percpu-rwsem.h> 31 #include <linux/workqueue.h> 32 #include <linux/delayed_call.h> 33 34 #include <asm/byteorder.h> 35 #include <uapi/linux/fs.h> 36 37 struct backing_dev_info; 38 struct bdi_writeback; 39 struct bio; 40 struct export_operations; 41 struct hd_geometry; 42 struct iovec; 43 struct kiocb; 44 struct kobject; 45 struct pipe_inode_info; 46 struct poll_table_struct; 47 struct kstatfs; 48 struct vm_area_struct; 49 struct vfsmount; 50 struct cred; 51 struct swap_info_struct; 52 struct seq_file; 53 struct workqueue_struct; 54 struct iov_iter; 55 struct fscrypt_info; 56 struct fscrypt_operations; 57 58 extern void __init inode_init(void); 59 extern void __init inode_init_early(void); 60 extern void __init files_init(void); 61 extern void __init files_maxfiles_init(void); 62 63 extern struct files_stat_struct files_stat; 64 extern unsigned long get_max_files(void); 65 extern unsigned int sysctl_nr_open; 66 extern struct inodes_stat_t inodes_stat; 67 extern int leases_enable, lease_break_time; 68 extern int sysctl_protected_symlinks; 69 extern int sysctl_protected_hardlinks; 70 71 struct buffer_head; 72 typedef int (get_block_t)(struct inode *inode, sector_t iblock, 73 struct buffer_head *bh_result, int create); 74 typedef int (dio_iodone_t)(struct kiocb *iocb, loff_t offset, 75 ssize_t bytes, void *private); 76 77 #define MAY_EXEC 0x00000001 78 #define MAY_WRITE 0x00000002 79 #define MAY_READ 0x00000004 80 #define MAY_APPEND 0x00000008 81 #define MAY_ACCESS 0x00000010 82 #define MAY_OPEN 0x00000020 83 #define MAY_CHDIR 0x00000040 84 /* called from RCU mode, don't block */ 85 #define MAY_NOT_BLOCK 0x00000080 86 87 /* 88 * flags in file.f_mode. Note that FMODE_READ and FMODE_WRITE must correspond 89 * to O_WRONLY and O_RDWR via the strange trick in __dentry_open() 90 */ 91 92 /* file is open for reading */ 93 #define FMODE_READ ((__force fmode_t)0x1) 94 /* file is open for writing */ 95 #define FMODE_WRITE ((__force fmode_t)0x2) 96 /* file is seekable */ 97 #define FMODE_LSEEK ((__force fmode_t)0x4) 98 /* file can be accessed using pread */ 99 #define FMODE_PREAD ((__force fmode_t)0x8) 100 /* file can be accessed using pwrite */ 101 #define FMODE_PWRITE ((__force fmode_t)0x10) 102 /* File is opened for execution with sys_execve / sys_uselib */ 103 #define FMODE_EXEC ((__force fmode_t)0x20) 104 /* File is opened with O_NDELAY (only set for block devices) */ 105 #define FMODE_NDELAY ((__force fmode_t)0x40) 106 /* File is opened with O_EXCL (only set for block devices) */ 107 #define FMODE_EXCL ((__force fmode_t)0x80) 108 /* File is opened using open(.., 3, ..) and is writeable only for ioctls 109 (specialy hack for floppy.c) */ 110 #define FMODE_WRITE_IOCTL ((__force fmode_t)0x100) 111 /* 32bit hashes as llseek() offset (for directories) */ 112 #define FMODE_32BITHASH ((__force fmode_t)0x200) 113 /* 64bit hashes as llseek() offset (for directories) */ 114 #define FMODE_64BITHASH ((__force fmode_t)0x400) 115 116 /* 117 * Don't update ctime and mtime. 118 * 119 * Currently a special hack for the XFS open_by_handle ioctl, but we'll 120 * hopefully graduate it to a proper O_CMTIME flag supported by open(2) soon. 121 */ 122 #define FMODE_NOCMTIME ((__force fmode_t)0x800) 123 124 /* Expect random access pattern */ 125 #define FMODE_RANDOM ((__force fmode_t)0x1000) 126 127 /* File is huge (eg. /dev/kmem): treat loff_t as unsigned */ 128 #define FMODE_UNSIGNED_OFFSET ((__force fmode_t)0x2000) 129 130 /* File is opened with O_PATH; almost nothing can be done with it */ 131 #define FMODE_PATH ((__force fmode_t)0x4000) 132 133 /* File needs atomic accesses to f_pos */ 134 #define FMODE_ATOMIC_POS ((__force fmode_t)0x8000) 135 /* Write access to underlying fs */ 136 #define FMODE_WRITER ((__force fmode_t)0x10000) 137 /* Has read method(s) */ 138 #define FMODE_CAN_READ ((__force fmode_t)0x20000) 139 /* Has write method(s) */ 140 #define FMODE_CAN_WRITE ((__force fmode_t)0x40000) 141 142 /* File was opened by fanotify and shouldn't generate fanotify events */ 143 #define FMODE_NONOTIFY ((__force fmode_t)0x4000000) 144 145 /* 146 * Flag for rw_copy_check_uvector and compat_rw_copy_check_uvector 147 * that indicates that they should check the contents of the iovec are 148 * valid, but not check the memory that the iovec elements 149 * points too. 150 */ 151 #define CHECK_IOVEC_ONLY -1 152 153 /* 154 * Attribute flags. These should be or-ed together to figure out what 155 * has been changed! 156 */ 157 #define ATTR_MODE (1 << 0) 158 #define ATTR_UID (1 << 1) 159 #define ATTR_GID (1 << 2) 160 #define ATTR_SIZE (1 << 3) 161 #define ATTR_ATIME (1 << 4) 162 #define ATTR_MTIME (1 << 5) 163 #define ATTR_CTIME (1 << 6) 164 #define ATTR_ATIME_SET (1 << 7) 165 #define ATTR_MTIME_SET (1 << 8) 166 #define ATTR_FORCE (1 << 9) /* Not a change, but a change it */ 167 #define ATTR_ATTR_FLAG (1 << 10) 168 #define ATTR_KILL_SUID (1 << 11) 169 #define ATTR_KILL_SGID (1 << 12) 170 #define ATTR_FILE (1 << 13) 171 #define ATTR_KILL_PRIV (1 << 14) 172 #define ATTR_OPEN (1 << 15) /* Truncating from open(O_TRUNC) */ 173 #define ATTR_TIMES_SET (1 << 16) 174 #define ATTR_TOUCH (1 << 17) 175 176 /* 177 * Whiteout is represented by a char device. The following constants define the 178 * mode and device number to use. 179 */ 180 #define WHITEOUT_MODE 0 181 #define WHITEOUT_DEV 0 182 183 /* 184 * This is the Inode Attributes structure, used for notify_change(). It 185 * uses the above definitions as flags, to know which values have changed. 186 * Also, in this manner, a Filesystem can look at only the values it cares 187 * about. Basically, these are the attributes that the VFS layer can 188 * request to change from the FS layer. 189 * 190 * Derek Atkins <warlord@MIT.EDU> 94-10-20 191 */ 192 struct iattr { 193 unsigned int ia_valid; 194 umode_t ia_mode; 195 kuid_t ia_uid; 196 kgid_t ia_gid; 197 loff_t ia_size; 198 struct timespec ia_atime; 199 struct timespec ia_mtime; 200 struct timespec ia_ctime; 201 202 /* 203 * Not an attribute, but an auxiliary info for filesystems wanting to 204 * implement an ftruncate() like method. NOTE: filesystem should 205 * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL). 206 */ 207 struct file *ia_file; 208 }; 209 210 /* 211 * Includes for diskquotas. 212 */ 213 #include <linux/quota.h> 214 215 /* 216 * Maximum number of layers of fs stack. Needs to be limited to 217 * prevent kernel stack overflow 218 */ 219 #define FILESYSTEM_MAX_STACK_DEPTH 2 220 221 /** 222 * enum positive_aop_returns - aop return codes with specific semantics 223 * 224 * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has 225 * completed, that the page is still locked, and 226 * should be considered active. The VM uses this hint 227 * to return the page to the active list -- it won't 228 * be a candidate for writeback again in the near 229 * future. Other callers must be careful to unlock 230 * the page if they get this return. Returned by 231 * writepage(); 232 * 233 * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has 234 * unlocked it and the page might have been truncated. 235 * The caller should back up to acquiring a new page and 236 * trying again. The aop will be taking reasonable 237 * precautions not to livelock. If the caller held a page 238 * reference, it should drop it before retrying. Returned 239 * by readpage(). 240 * 241 * address_space_operation functions return these large constants to indicate 242 * special semantics to the caller. These are much larger than the bytes in a 243 * page to allow for functions that return the number of bytes operated on in a 244 * given page. 245 */ 246 247 enum positive_aop_returns { 248 AOP_WRITEPAGE_ACTIVATE = 0x80000, 249 AOP_TRUNCATED_PAGE = 0x80001, 250 }; 251 252 #define AOP_FLAG_CONT_EXPAND 0x0001 /* called from cont_expand */ 253 #define AOP_FLAG_NOFS 0x0002 /* used by filesystem to direct 254 * helper code (eg buffer layer) 255 * to clear GFP_FS from alloc */ 256 257 /* 258 * oh the beauties of C type declarations. 259 */ 260 struct page; 261 struct address_space; 262 struct writeback_control; 263 264 #define IOCB_EVENTFD (1 << 0) 265 #define IOCB_APPEND (1 << 1) 266 #define IOCB_DIRECT (1 << 2) 267 #define IOCB_HIPRI (1 << 3) 268 #define IOCB_DSYNC (1 << 4) 269 #define IOCB_SYNC (1 << 5) 270 #define IOCB_WRITE (1 << 6) 271 272 struct kiocb { 273 struct file *ki_filp; 274 loff_t ki_pos; 275 void (*ki_complete)(struct kiocb *iocb, long ret, long ret2); 276 void *private; 277 int ki_flags; 278 }; 279 280 static inline bool is_sync_kiocb(struct kiocb *kiocb) 281 { 282 return kiocb->ki_complete == NULL; 283 } 284 285 static inline int iocb_flags(struct file *file); 286 287 static inline void init_sync_kiocb(struct kiocb *kiocb, struct file *filp) 288 { 289 *kiocb = (struct kiocb) { 290 .ki_filp = filp, 291 .ki_flags = iocb_flags(filp), 292 }; 293 } 294 295 /* 296 * "descriptor" for what we're up to with a read. 297 * This allows us to use the same read code yet 298 * have multiple different users of the data that 299 * we read from a file. 300 * 301 * The simplest case just copies the data to user 302 * mode. 303 */ 304 typedef struct { 305 size_t written; 306 size_t count; 307 union { 308 char __user *buf; 309 void *data; 310 } arg; 311 int error; 312 } read_descriptor_t; 313 314 typedef int (*read_actor_t)(read_descriptor_t *, struct page *, 315 unsigned long, unsigned long); 316 317 struct address_space_operations { 318 int (*writepage)(struct page *page, struct writeback_control *wbc); 319 int (*readpage)(struct file *, struct page *); 320 321 /* Write back some dirty pages from this mapping. */ 322 int (*writepages)(struct address_space *, struct writeback_control *); 323 324 /* Set a page dirty. Return true if this dirtied it */ 325 int (*set_page_dirty)(struct page *page); 326 327 int (*readpages)(struct file *filp, struct address_space *mapping, 328 struct list_head *pages, unsigned nr_pages); 329 330 int (*write_begin)(struct file *, struct address_space *mapping, 331 loff_t pos, unsigned len, unsigned flags, 332 struct page **pagep, void **fsdata); 333 int (*write_end)(struct file *, struct address_space *mapping, 334 loff_t pos, unsigned len, unsigned copied, 335 struct page *page, void *fsdata); 336 337 /* Unfortunately this kludge is needed for FIBMAP. Don't use it */ 338 sector_t (*bmap)(struct address_space *, sector_t); 339 void (*invalidatepage) (struct page *, unsigned int, unsigned int); 340 int (*releasepage) (struct page *, gfp_t); 341 void (*freepage)(struct page *); 342 ssize_t (*direct_IO)(struct kiocb *, struct iov_iter *iter); 343 /* 344 * migrate the contents of a page to the specified target. If 345 * migrate_mode is MIGRATE_ASYNC, it must not block. 346 */ 347 int (*migratepage) (struct address_space *, 348 struct page *, struct page *, enum migrate_mode); 349 bool (*isolate_page)(struct page *, isolate_mode_t); 350 void (*putback_page)(struct page *); 351 int (*launder_page) (struct page *); 352 int (*is_partially_uptodate) (struct page *, unsigned long, 353 unsigned long); 354 void (*is_dirty_writeback) (struct page *, bool *, bool *); 355 int (*error_remove_page)(struct address_space *, struct page *); 356 357 /* swapfile support */ 358 int (*swap_activate)(struct swap_info_struct *sis, struct file *file, 359 sector_t *span); 360 void (*swap_deactivate)(struct file *file); 361 }; 362 363 extern const struct address_space_operations empty_aops; 364 365 /* 366 * pagecache_write_begin/pagecache_write_end must be used by general code 367 * to write into the pagecache. 368 */ 369 int pagecache_write_begin(struct file *, struct address_space *mapping, 370 loff_t pos, unsigned len, unsigned flags, 371 struct page **pagep, void **fsdata); 372 373 int pagecache_write_end(struct file *, struct address_space *mapping, 374 loff_t pos, unsigned len, unsigned copied, 375 struct page *page, void *fsdata); 376 377 struct address_space { 378 struct inode *host; /* owner: inode, block_device */ 379 struct radix_tree_root page_tree; /* radix tree of all pages */ 380 spinlock_t tree_lock; /* and lock protecting it */ 381 atomic_t i_mmap_writable;/* count VM_SHARED mappings */ 382 struct rb_root i_mmap; /* tree of private and shared mappings */ 383 struct rw_semaphore i_mmap_rwsem; /* protect tree, count, list */ 384 /* Protected by tree_lock together with the radix tree */ 385 unsigned long nrpages; /* number of total pages */ 386 /* number of shadow or DAX exceptional entries */ 387 unsigned long nrexceptional; 388 pgoff_t writeback_index;/* writeback starts here */ 389 const struct address_space_operations *a_ops; /* methods */ 390 unsigned long flags; /* error bits */ 391 spinlock_t private_lock; /* for use by the address_space */ 392 gfp_t gfp_mask; /* implicit gfp mask for allocations */ 393 struct list_head private_list; /* ditto */ 394 void *private_data; /* ditto */ 395 } __attribute__((aligned(sizeof(long)))); 396 /* 397 * On most architectures that alignment is already the case; but 398 * must be enforced here for CRIS, to let the least significant bit 399 * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON. 400 */ 401 struct request_queue; 402 403 struct block_device { 404 dev_t bd_dev; /* not a kdev_t - it's a search key */ 405 int bd_openers; 406 struct inode * bd_inode; /* will die */ 407 struct super_block * bd_super; 408 struct mutex bd_mutex; /* open/close mutex */ 409 void * bd_claiming; 410 void * bd_holder; 411 int bd_holders; 412 bool bd_write_holder; 413 #ifdef CONFIG_SYSFS 414 struct list_head bd_holder_disks; 415 #endif 416 struct block_device * bd_contains; 417 unsigned bd_block_size; 418 struct hd_struct * bd_part; 419 /* number of times partitions within this device have been opened. */ 420 unsigned bd_part_count; 421 int bd_invalidated; 422 struct gendisk * bd_disk; 423 struct request_queue * bd_queue; 424 struct backing_dev_info *bd_bdi; 425 struct list_head bd_list; 426 /* 427 * Private data. You must have bd_claim'ed the block_device 428 * to use this. NOTE: bd_claim allows an owner to claim 429 * the same device multiple times, the owner must take special 430 * care to not mess up bd_private for that case. 431 */ 432 unsigned long bd_private; 433 434 /* The counter of freeze processes */ 435 int bd_fsfreeze_count; 436 /* Mutex for freeze */ 437 struct mutex bd_fsfreeze_mutex; 438 }; 439 440 /* 441 * Radix-tree tags, for tagging dirty and writeback pages within the pagecache 442 * radix trees 443 */ 444 #define PAGECACHE_TAG_DIRTY 0 445 #define PAGECACHE_TAG_WRITEBACK 1 446 #define PAGECACHE_TAG_TOWRITE 2 447 448 int mapping_tagged(struct address_space *mapping, int tag); 449 450 static inline void i_mmap_lock_write(struct address_space *mapping) 451 { 452 down_write(&mapping->i_mmap_rwsem); 453 } 454 455 static inline void i_mmap_unlock_write(struct address_space *mapping) 456 { 457 up_write(&mapping->i_mmap_rwsem); 458 } 459 460 static inline void i_mmap_lock_read(struct address_space *mapping) 461 { 462 down_read(&mapping->i_mmap_rwsem); 463 } 464 465 static inline void i_mmap_unlock_read(struct address_space *mapping) 466 { 467 up_read(&mapping->i_mmap_rwsem); 468 } 469 470 /* 471 * Might pages of this file be mapped into userspace? 472 */ 473 static inline int mapping_mapped(struct address_space *mapping) 474 { 475 return !RB_EMPTY_ROOT(&mapping->i_mmap); 476 } 477 478 /* 479 * Might pages of this file have been modified in userspace? 480 * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap_pgoff 481 * marks vma as VM_SHARED if it is shared, and the file was opened for 482 * writing i.e. vma may be mprotected writable even if now readonly. 483 * 484 * If i_mmap_writable is negative, no new writable mappings are allowed. You 485 * can only deny writable mappings, if none exists right now. 486 */ 487 static inline int mapping_writably_mapped(struct address_space *mapping) 488 { 489 return atomic_read(&mapping->i_mmap_writable) > 0; 490 } 491 492 static inline int mapping_map_writable(struct address_space *mapping) 493 { 494 return atomic_inc_unless_negative(&mapping->i_mmap_writable) ? 495 0 : -EPERM; 496 } 497 498 static inline void mapping_unmap_writable(struct address_space *mapping) 499 { 500 atomic_dec(&mapping->i_mmap_writable); 501 } 502 503 static inline int mapping_deny_writable(struct address_space *mapping) 504 { 505 return atomic_dec_unless_positive(&mapping->i_mmap_writable) ? 506 0 : -EBUSY; 507 } 508 509 static inline void mapping_allow_writable(struct address_space *mapping) 510 { 511 atomic_inc(&mapping->i_mmap_writable); 512 } 513 514 /* 515 * Use sequence counter to get consistent i_size on 32-bit processors. 516 */ 517 #if BITS_PER_LONG==32 && defined(CONFIG_SMP) 518 #include <linux/seqlock.h> 519 #define __NEED_I_SIZE_ORDERED 520 #define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount) 521 #else 522 #define i_size_ordered_init(inode) do { } while (0) 523 #endif 524 525 struct posix_acl; 526 #define ACL_NOT_CACHED ((void *)(-1)) 527 #define ACL_DONT_CACHE ((void *)(-3)) 528 529 static inline struct posix_acl * 530 uncached_acl_sentinel(struct task_struct *task) 531 { 532 return (void *)task + 1; 533 } 534 535 static inline bool 536 is_uncached_acl(struct posix_acl *acl) 537 { 538 return (long)acl & 1; 539 } 540 541 #define IOP_FASTPERM 0x0001 542 #define IOP_LOOKUP 0x0002 543 #define IOP_NOFOLLOW 0x0004 544 #define IOP_XATTR 0x0008 545 #define IOP_DEFAULT_READLINK 0x0010 546 547 struct fsnotify_mark_connector; 548 549 /* 550 * Keep mostly read-only and often accessed (especially for 551 * the RCU path lookup and 'stat' data) fields at the beginning 552 * of the 'struct inode' 553 */ 554 struct inode { 555 umode_t i_mode; 556 unsigned short i_opflags; 557 kuid_t i_uid; 558 kgid_t i_gid; 559 unsigned int i_flags; 560 561 #ifdef CONFIG_FS_POSIX_ACL 562 struct posix_acl *i_acl; 563 struct posix_acl *i_default_acl; 564 #endif 565 566 const struct inode_operations *i_op; 567 struct super_block *i_sb; 568 struct address_space *i_mapping; 569 570 #ifdef CONFIG_SECURITY 571 void *i_security; 572 #endif 573 574 /* Stat data, not accessed from path walking */ 575 unsigned long i_ino; 576 /* 577 * Filesystems may only read i_nlink directly. They shall use the 578 * following functions for modification: 579 * 580 * (set|clear|inc|drop)_nlink 581 * inode_(inc|dec)_link_count 582 */ 583 union { 584 const unsigned int i_nlink; 585 unsigned int __i_nlink; 586 }; 587 dev_t i_rdev; 588 loff_t i_size; 589 struct timespec i_atime; 590 struct timespec i_mtime; 591 struct timespec i_ctime; 592 spinlock_t i_lock; /* i_blocks, i_bytes, maybe i_size */ 593 unsigned short i_bytes; 594 unsigned int i_blkbits; 595 blkcnt_t i_blocks; 596 597 #ifdef __NEED_I_SIZE_ORDERED 598 seqcount_t i_size_seqcount; 599 #endif 600 601 /* Misc */ 602 unsigned long i_state; 603 struct rw_semaphore i_rwsem; 604 605 unsigned long dirtied_when; /* jiffies of first dirtying */ 606 unsigned long dirtied_time_when; 607 608 struct hlist_node i_hash; 609 struct list_head i_io_list; /* backing dev IO list */ 610 #ifdef CONFIG_CGROUP_WRITEBACK 611 struct bdi_writeback *i_wb; /* the associated cgroup wb */ 612 613 /* foreign inode detection, see wbc_detach_inode() */ 614 int i_wb_frn_winner; 615 u16 i_wb_frn_avg_time; 616 u16 i_wb_frn_history; 617 #endif 618 struct list_head i_lru; /* inode LRU list */ 619 struct list_head i_sb_list; 620 struct list_head i_wb_list; /* backing dev writeback list */ 621 union { 622 struct hlist_head i_dentry; 623 struct rcu_head i_rcu; 624 }; 625 u64 i_version; 626 atomic_t i_count; 627 atomic_t i_dio_count; 628 atomic_t i_writecount; 629 #ifdef CONFIG_IMA 630 atomic_t i_readcount; /* struct files open RO */ 631 #endif 632 const struct file_operations *i_fop; /* former ->i_op->default_file_ops */ 633 struct file_lock_context *i_flctx; 634 struct address_space i_data; 635 struct list_head i_devices; 636 union { 637 struct pipe_inode_info *i_pipe; 638 struct block_device *i_bdev; 639 struct cdev *i_cdev; 640 char *i_link; 641 unsigned i_dir_seq; 642 }; 643 644 __u32 i_generation; 645 646 #ifdef CONFIG_FSNOTIFY 647 __u32 i_fsnotify_mask; /* all events this inode cares about */ 648 struct fsnotify_mark_connector __rcu *i_fsnotify_marks; 649 #endif 650 651 #if IS_ENABLED(CONFIG_FS_ENCRYPTION) 652 struct fscrypt_info *i_crypt_info; 653 #endif 654 655 void *i_private; /* fs or device private pointer */ 656 }; 657 658 static inline unsigned int i_blocksize(const struct inode *node) 659 { 660 return (1 << node->i_blkbits); 661 } 662 663 static inline int inode_unhashed(struct inode *inode) 664 { 665 return hlist_unhashed(&inode->i_hash); 666 } 667 668 /* 669 * inode->i_mutex nesting subclasses for the lock validator: 670 * 671 * 0: the object of the current VFS operation 672 * 1: parent 673 * 2: child/target 674 * 3: xattr 675 * 4: second non-directory 676 * 5: second parent (when locking independent directories in rename) 677 * 678 * I_MUTEX_NONDIR2 is for certain operations (such as rename) which lock two 679 * non-directories at once. 680 * 681 * The locking order between these classes is 682 * parent[2] -> child -> grandchild -> normal -> xattr -> second non-directory 683 */ 684 enum inode_i_mutex_lock_class 685 { 686 I_MUTEX_NORMAL, 687 I_MUTEX_PARENT, 688 I_MUTEX_CHILD, 689 I_MUTEX_XATTR, 690 I_MUTEX_NONDIR2, 691 I_MUTEX_PARENT2, 692 }; 693 694 static inline void inode_lock(struct inode *inode) 695 { 696 down_write(&inode->i_rwsem); 697 } 698 699 static inline void inode_unlock(struct inode *inode) 700 { 701 up_write(&inode->i_rwsem); 702 } 703 704 static inline void inode_lock_shared(struct inode *inode) 705 { 706 down_read(&inode->i_rwsem); 707 } 708 709 static inline void inode_unlock_shared(struct inode *inode) 710 { 711 up_read(&inode->i_rwsem); 712 } 713 714 static inline int inode_trylock(struct inode *inode) 715 { 716 return down_write_trylock(&inode->i_rwsem); 717 } 718 719 static inline int inode_trylock_shared(struct inode *inode) 720 { 721 return down_read_trylock(&inode->i_rwsem); 722 } 723 724 static inline int inode_is_locked(struct inode *inode) 725 { 726 return rwsem_is_locked(&inode->i_rwsem); 727 } 728 729 static inline void inode_lock_nested(struct inode *inode, unsigned subclass) 730 { 731 down_write_nested(&inode->i_rwsem, subclass); 732 } 733 734 void lock_two_nondirectories(struct inode *, struct inode*); 735 void unlock_two_nondirectories(struct inode *, struct inode*); 736 737 /* 738 * NOTE: in a 32bit arch with a preemptable kernel and 739 * an UP compile the i_size_read/write must be atomic 740 * with respect to the local cpu (unlike with preempt disabled), 741 * but they don't need to be atomic with respect to other cpus like in 742 * true SMP (so they need either to either locally disable irq around 743 * the read or for example on x86 they can be still implemented as a 744 * cmpxchg8b without the need of the lock prefix). For SMP compiles 745 * and 64bit archs it makes no difference if preempt is enabled or not. 746 */ 747 static inline loff_t i_size_read(const struct inode *inode) 748 { 749 #if BITS_PER_LONG==32 && defined(CONFIG_SMP) 750 loff_t i_size; 751 unsigned int seq; 752 753 do { 754 seq = read_seqcount_begin(&inode->i_size_seqcount); 755 i_size = inode->i_size; 756 } while (read_seqcount_retry(&inode->i_size_seqcount, seq)); 757 return i_size; 758 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT) 759 loff_t i_size; 760 761 preempt_disable(); 762 i_size = inode->i_size; 763 preempt_enable(); 764 return i_size; 765 #else 766 return inode->i_size; 767 #endif 768 } 769 770 /* 771 * NOTE: unlike i_size_read(), i_size_write() does need locking around it 772 * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount 773 * can be lost, resulting in subsequent i_size_read() calls spinning forever. 774 */ 775 static inline void i_size_write(struct inode *inode, loff_t i_size) 776 { 777 #if BITS_PER_LONG==32 && defined(CONFIG_SMP) 778 preempt_disable(); 779 write_seqcount_begin(&inode->i_size_seqcount); 780 inode->i_size = i_size; 781 write_seqcount_end(&inode->i_size_seqcount); 782 preempt_enable(); 783 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT) 784 preempt_disable(); 785 inode->i_size = i_size; 786 preempt_enable(); 787 #else 788 inode->i_size = i_size; 789 #endif 790 } 791 792 static inline unsigned iminor(const struct inode *inode) 793 { 794 return MINOR(inode->i_rdev); 795 } 796 797 static inline unsigned imajor(const struct inode *inode) 798 { 799 return MAJOR(inode->i_rdev); 800 } 801 802 extern struct block_device *I_BDEV(struct inode *inode); 803 804 struct fown_struct { 805 rwlock_t lock; /* protects pid, uid, euid fields */ 806 struct pid *pid; /* pid or -pgrp where SIGIO should be sent */ 807 enum pid_type pid_type; /* Kind of process group SIGIO should be sent to */ 808 kuid_t uid, euid; /* uid/euid of process setting the owner */ 809 int signum; /* posix.1b rt signal to be delivered on IO */ 810 }; 811 812 /* 813 * Track a single file's readahead state 814 */ 815 struct file_ra_state { 816 pgoff_t start; /* where readahead started */ 817 unsigned int size; /* # of readahead pages */ 818 unsigned int async_size; /* do asynchronous readahead when 819 there are only # of pages ahead */ 820 821 unsigned int ra_pages; /* Maximum readahead window */ 822 unsigned int mmap_miss; /* Cache miss stat for mmap accesses */ 823 loff_t prev_pos; /* Cache last read() position */ 824 }; 825 826 /* 827 * Check if @index falls in the readahead windows. 828 */ 829 static inline int ra_has_index(struct file_ra_state *ra, pgoff_t index) 830 { 831 return (index >= ra->start && 832 index < ra->start + ra->size); 833 } 834 835 struct file { 836 union { 837 struct llist_node fu_llist; 838 struct rcu_head fu_rcuhead; 839 } f_u; 840 struct path f_path; 841 struct inode *f_inode; /* cached value */ 842 const struct file_operations *f_op; 843 844 /* 845 * Protects f_ep_links, f_flags. 846 * Must not be taken from IRQ context. 847 */ 848 spinlock_t f_lock; 849 atomic_long_t f_count; 850 unsigned int f_flags; 851 fmode_t f_mode; 852 struct mutex f_pos_lock; 853 loff_t f_pos; 854 struct fown_struct f_owner; 855 const struct cred *f_cred; 856 struct file_ra_state f_ra; 857 858 u64 f_version; 859 #ifdef CONFIG_SECURITY 860 void *f_security; 861 #endif 862 /* needed for tty driver, and maybe others */ 863 void *private_data; 864 865 #ifdef CONFIG_EPOLL 866 /* Used by fs/eventpoll.c to link all the hooks to this file */ 867 struct list_head f_ep_links; 868 struct list_head f_tfile_llink; 869 #endif /* #ifdef CONFIG_EPOLL */ 870 struct address_space *f_mapping; 871 } __attribute__((aligned(4))); /* lest something weird decides that 2 is OK */ 872 873 struct file_handle { 874 __u32 handle_bytes; 875 int handle_type; 876 /* file identifier */ 877 unsigned char f_handle[0]; 878 }; 879 880 static inline struct file *get_file(struct file *f) 881 { 882 atomic_long_inc(&f->f_count); 883 return f; 884 } 885 #define get_file_rcu(x) atomic_long_inc_not_zero(&(x)->f_count) 886 #define fput_atomic(x) atomic_long_add_unless(&(x)->f_count, -1, 1) 887 #define file_count(x) atomic_long_read(&(x)->f_count) 888 889 #define MAX_NON_LFS ((1UL<<31) - 1) 890 891 /* Page cache limit. The filesystems should put that into their s_maxbytes 892 limits, otherwise bad things can happen in VM. */ 893 #if BITS_PER_LONG==32 894 #define MAX_LFS_FILESIZE (((loff_t)PAGE_SIZE << (BITS_PER_LONG-1))-1) 895 #elif BITS_PER_LONG==64 896 #define MAX_LFS_FILESIZE ((loff_t)0x7fffffffffffffffLL) 897 #endif 898 899 #define FL_POSIX 1 900 #define FL_FLOCK 2 901 #define FL_DELEG 4 /* NFSv4 delegation */ 902 #define FL_ACCESS 8 /* not trying to lock, just looking */ 903 #define FL_EXISTS 16 /* when unlocking, test for existence */ 904 #define FL_LEASE 32 /* lease held on this file */ 905 #define FL_CLOSE 64 /* unlock on close */ 906 #define FL_SLEEP 128 /* A blocking lock */ 907 #define FL_DOWNGRADE_PENDING 256 /* Lease is being downgraded */ 908 #define FL_UNLOCK_PENDING 512 /* Lease is being broken */ 909 #define FL_OFDLCK 1024 /* lock is "owned" by struct file */ 910 #define FL_LAYOUT 2048 /* outstanding pNFS layout */ 911 912 #define FL_CLOSE_POSIX (FL_POSIX | FL_CLOSE) 913 914 /* 915 * Special return value from posix_lock_file() and vfs_lock_file() for 916 * asynchronous locking. 917 */ 918 #define FILE_LOCK_DEFERRED 1 919 920 /* legacy typedef, should eventually be removed */ 921 typedef void *fl_owner_t; 922 923 struct file_lock; 924 925 struct file_lock_operations { 926 void (*fl_copy_lock)(struct file_lock *, struct file_lock *); 927 void (*fl_release_private)(struct file_lock *); 928 }; 929 930 struct lock_manager_operations { 931 int (*lm_compare_owner)(struct file_lock *, struct file_lock *); 932 unsigned long (*lm_owner_key)(struct file_lock *); 933 fl_owner_t (*lm_get_owner)(fl_owner_t); 934 void (*lm_put_owner)(fl_owner_t); 935 void (*lm_notify)(struct file_lock *); /* unblock callback */ 936 int (*lm_grant)(struct file_lock *, int); 937 bool (*lm_break)(struct file_lock *); 938 int (*lm_change)(struct file_lock *, int, struct list_head *); 939 void (*lm_setup)(struct file_lock *, void **); 940 }; 941 942 struct lock_manager { 943 struct list_head list; 944 /* 945 * NFSv4 and up also want opens blocked during the grace period; 946 * NLM doesn't care: 947 */ 948 bool block_opens; 949 }; 950 951 struct net; 952 void locks_start_grace(struct net *, struct lock_manager *); 953 void locks_end_grace(struct lock_manager *); 954 int locks_in_grace(struct net *); 955 int opens_in_grace(struct net *); 956 957 /* that will die - we need it for nfs_lock_info */ 958 #include <linux/nfs_fs_i.h> 959 960 /* 961 * struct file_lock represents a generic "file lock". It's used to represent 962 * POSIX byte range locks, BSD (flock) locks, and leases. It's important to 963 * note that the same struct is used to represent both a request for a lock and 964 * the lock itself, but the same object is never used for both. 965 * 966 * FIXME: should we create a separate "struct lock_request" to help distinguish 967 * these two uses? 968 * 969 * The varous i_flctx lists are ordered by: 970 * 971 * 1) lock owner 972 * 2) lock range start 973 * 3) lock range end 974 * 975 * Obviously, the last two criteria only matter for POSIX locks. 976 */ 977 struct file_lock { 978 struct file_lock *fl_next; /* singly linked list for this inode */ 979 struct list_head fl_list; /* link into file_lock_context */ 980 struct hlist_node fl_link; /* node in global lists */ 981 struct list_head fl_block; /* circular list of blocked processes */ 982 fl_owner_t fl_owner; 983 unsigned int fl_flags; 984 unsigned char fl_type; 985 unsigned int fl_pid; 986 int fl_link_cpu; /* what cpu's list is this on? */ 987 struct pid *fl_nspid; 988 wait_queue_head_t fl_wait; 989 struct file *fl_file; 990 loff_t fl_start; 991 loff_t fl_end; 992 993 struct fasync_struct * fl_fasync; /* for lease break notifications */ 994 /* for lease breaks: */ 995 unsigned long fl_break_time; 996 unsigned long fl_downgrade_time; 997 998 const struct file_lock_operations *fl_ops; /* Callbacks for filesystems */ 999 const struct lock_manager_operations *fl_lmops; /* Callbacks for lockmanagers */ 1000 union { 1001 struct nfs_lock_info nfs_fl; 1002 struct nfs4_lock_info nfs4_fl; 1003 struct { 1004 struct list_head link; /* link in AFS vnode's pending_locks list */ 1005 int state; /* state of grant or error if -ve */ 1006 } afs; 1007 } fl_u; 1008 }; 1009 1010 struct file_lock_context { 1011 spinlock_t flc_lock; 1012 struct list_head flc_flock; 1013 struct list_head flc_posix; 1014 struct list_head flc_lease; 1015 }; 1016 1017 /* The following constant reflects the upper bound of the file/locking space */ 1018 #ifndef OFFSET_MAX 1019 #define INT_LIMIT(x) (~((x)1 << (sizeof(x)*8 - 1))) 1020 #define OFFSET_MAX INT_LIMIT(loff_t) 1021 #define OFFT_OFFSET_MAX INT_LIMIT(off_t) 1022 #endif 1023 1024 #include <linux/fcntl.h> 1025 1026 extern void send_sigio(struct fown_struct *fown, int fd, int band); 1027 1028 /* 1029 * Return the inode to use for locking 1030 * 1031 * For overlayfs this should be the overlay inode, not the real inode returned 1032 * by file_inode(). For any other fs file_inode(filp) and locks_inode(filp) are 1033 * equal. 1034 */ 1035 static inline struct inode *locks_inode(const struct file *f) 1036 { 1037 return f->f_path.dentry->d_inode; 1038 } 1039 1040 #ifdef CONFIG_FILE_LOCKING 1041 extern int fcntl_getlk(struct file *, unsigned int, struct flock __user *); 1042 extern int fcntl_setlk(unsigned int, struct file *, unsigned int, 1043 struct flock __user *); 1044 1045 #if BITS_PER_LONG == 32 1046 extern int fcntl_getlk64(struct file *, unsigned int, struct flock64 __user *); 1047 extern int fcntl_setlk64(unsigned int, struct file *, unsigned int, 1048 struct flock64 __user *); 1049 #endif 1050 1051 extern int fcntl_setlease(unsigned int fd, struct file *filp, long arg); 1052 extern int fcntl_getlease(struct file *filp); 1053 1054 /* fs/locks.c */ 1055 void locks_free_lock_context(struct inode *inode); 1056 void locks_free_lock(struct file_lock *fl); 1057 extern void locks_init_lock(struct file_lock *); 1058 extern struct file_lock * locks_alloc_lock(void); 1059 extern void locks_copy_lock(struct file_lock *, struct file_lock *); 1060 extern void locks_copy_conflock(struct file_lock *, struct file_lock *); 1061 extern void locks_remove_posix(struct file *, fl_owner_t); 1062 extern void locks_remove_file(struct file *); 1063 extern void locks_release_private(struct file_lock *); 1064 extern void posix_test_lock(struct file *, struct file_lock *); 1065 extern int posix_lock_file(struct file *, struct file_lock *, struct file_lock *); 1066 extern int posix_unblock_lock(struct file_lock *); 1067 extern int vfs_test_lock(struct file *, struct file_lock *); 1068 extern int vfs_lock_file(struct file *, unsigned int, struct file_lock *, struct file_lock *); 1069 extern int vfs_cancel_lock(struct file *filp, struct file_lock *fl); 1070 extern int locks_lock_inode_wait(struct inode *inode, struct file_lock *fl); 1071 extern int __break_lease(struct inode *inode, unsigned int flags, unsigned int type); 1072 extern void lease_get_mtime(struct inode *, struct timespec *time); 1073 extern int generic_setlease(struct file *, long, struct file_lock **, void **priv); 1074 extern int vfs_setlease(struct file *, long, struct file_lock **, void **); 1075 extern int lease_modify(struct file_lock *, int, struct list_head *); 1076 struct files_struct; 1077 extern void show_fd_locks(struct seq_file *f, 1078 struct file *filp, struct files_struct *files); 1079 #else /* !CONFIG_FILE_LOCKING */ 1080 static inline int fcntl_getlk(struct file *file, unsigned int cmd, 1081 struct flock __user *user) 1082 { 1083 return -EINVAL; 1084 } 1085 1086 static inline int fcntl_setlk(unsigned int fd, struct file *file, 1087 unsigned int cmd, struct flock __user *user) 1088 { 1089 return -EACCES; 1090 } 1091 1092 #if BITS_PER_LONG == 32 1093 static inline int fcntl_getlk64(struct file *file, unsigned int cmd, 1094 struct flock64 __user *user) 1095 { 1096 return -EINVAL; 1097 } 1098 1099 static inline int fcntl_setlk64(unsigned int fd, struct file *file, 1100 unsigned int cmd, struct flock64 __user *user) 1101 { 1102 return -EACCES; 1103 } 1104 #endif 1105 static inline int fcntl_setlease(unsigned int fd, struct file *filp, long arg) 1106 { 1107 return -EINVAL; 1108 } 1109 1110 static inline int fcntl_getlease(struct file *filp) 1111 { 1112 return F_UNLCK; 1113 } 1114 1115 static inline void 1116 locks_free_lock_context(struct inode *inode) 1117 { 1118 } 1119 1120 static inline void locks_init_lock(struct file_lock *fl) 1121 { 1122 return; 1123 } 1124 1125 static inline void locks_copy_conflock(struct file_lock *new, struct file_lock *fl) 1126 { 1127 return; 1128 } 1129 1130 static inline void locks_copy_lock(struct file_lock *new, struct file_lock *fl) 1131 { 1132 return; 1133 } 1134 1135 static inline void locks_remove_posix(struct file *filp, fl_owner_t owner) 1136 { 1137 return; 1138 } 1139 1140 static inline void locks_remove_file(struct file *filp) 1141 { 1142 return; 1143 } 1144 1145 static inline void posix_test_lock(struct file *filp, struct file_lock *fl) 1146 { 1147 return; 1148 } 1149 1150 static inline int posix_lock_file(struct file *filp, struct file_lock *fl, 1151 struct file_lock *conflock) 1152 { 1153 return -ENOLCK; 1154 } 1155 1156 static inline int posix_unblock_lock(struct file_lock *waiter) 1157 { 1158 return -ENOENT; 1159 } 1160 1161 static inline int vfs_test_lock(struct file *filp, struct file_lock *fl) 1162 { 1163 return 0; 1164 } 1165 1166 static inline int vfs_lock_file(struct file *filp, unsigned int cmd, 1167 struct file_lock *fl, struct file_lock *conf) 1168 { 1169 return -ENOLCK; 1170 } 1171 1172 static inline int vfs_cancel_lock(struct file *filp, struct file_lock *fl) 1173 { 1174 return 0; 1175 } 1176 1177 static inline int locks_lock_inode_wait(struct inode *inode, struct file_lock *fl) 1178 { 1179 return -ENOLCK; 1180 } 1181 1182 static inline int __break_lease(struct inode *inode, unsigned int mode, unsigned int type) 1183 { 1184 return 0; 1185 } 1186 1187 static inline void lease_get_mtime(struct inode *inode, struct timespec *time) 1188 { 1189 return; 1190 } 1191 1192 static inline int generic_setlease(struct file *filp, long arg, 1193 struct file_lock **flp, void **priv) 1194 { 1195 return -EINVAL; 1196 } 1197 1198 static inline int vfs_setlease(struct file *filp, long arg, 1199 struct file_lock **lease, void **priv) 1200 { 1201 return -EINVAL; 1202 } 1203 1204 static inline int lease_modify(struct file_lock *fl, int arg, 1205 struct list_head *dispose) 1206 { 1207 return -EINVAL; 1208 } 1209 1210 struct files_struct; 1211 static inline void show_fd_locks(struct seq_file *f, 1212 struct file *filp, struct files_struct *files) {} 1213 #endif /* !CONFIG_FILE_LOCKING */ 1214 1215 static inline struct inode *file_inode(const struct file *f) 1216 { 1217 return f->f_inode; 1218 } 1219 1220 static inline struct dentry *file_dentry(const struct file *file) 1221 { 1222 return d_real(file->f_path.dentry, file_inode(file), 0); 1223 } 1224 1225 static inline int locks_lock_file_wait(struct file *filp, struct file_lock *fl) 1226 { 1227 return locks_lock_inode_wait(locks_inode(filp), fl); 1228 } 1229 1230 struct fasync_struct { 1231 spinlock_t fa_lock; 1232 int magic; 1233 int fa_fd; 1234 struct fasync_struct *fa_next; /* singly linked list */ 1235 struct file *fa_file; 1236 struct rcu_head fa_rcu; 1237 }; 1238 1239 #define FASYNC_MAGIC 0x4601 1240 1241 /* SMP safe fasync helpers: */ 1242 extern int fasync_helper(int, struct file *, int, struct fasync_struct **); 1243 extern struct fasync_struct *fasync_insert_entry(int, struct file *, struct fasync_struct **, struct fasync_struct *); 1244 extern int fasync_remove_entry(struct file *, struct fasync_struct **); 1245 extern struct fasync_struct *fasync_alloc(void); 1246 extern void fasync_free(struct fasync_struct *); 1247 1248 /* can be called from interrupts */ 1249 extern void kill_fasync(struct fasync_struct **, int, int); 1250 1251 extern void __f_setown(struct file *filp, struct pid *, enum pid_type, int force); 1252 extern void f_setown(struct file *filp, unsigned long arg, int force); 1253 extern void f_delown(struct file *filp); 1254 extern pid_t f_getown(struct file *filp); 1255 extern int send_sigurg(struct fown_struct *fown); 1256 1257 struct mm_struct; 1258 1259 /* 1260 * Umount options 1261 */ 1262 1263 #define MNT_FORCE 0x00000001 /* Attempt to forcibily umount */ 1264 #define MNT_DETACH 0x00000002 /* Just detach from the tree */ 1265 #define MNT_EXPIRE 0x00000004 /* Mark for expiry */ 1266 #define UMOUNT_NOFOLLOW 0x00000008 /* Don't follow symlink on umount */ 1267 #define UMOUNT_UNUSED 0x80000000 /* Flag guaranteed to be unused */ 1268 1269 /* sb->s_iflags */ 1270 #define SB_I_CGROUPWB 0x00000001 /* cgroup-aware writeback enabled */ 1271 #define SB_I_NOEXEC 0x00000002 /* Ignore executables on this fs */ 1272 #define SB_I_NODEV 0x00000004 /* Ignore devices on this fs */ 1273 1274 /* sb->s_iflags to limit user namespace mounts */ 1275 #define SB_I_USERNS_VISIBLE 0x00000010 /* fstype already mounted */ 1276 1277 /* Possible states of 'frozen' field */ 1278 enum { 1279 SB_UNFROZEN = 0, /* FS is unfrozen */ 1280 SB_FREEZE_WRITE = 1, /* Writes, dir ops, ioctls frozen */ 1281 SB_FREEZE_PAGEFAULT = 2, /* Page faults stopped as well */ 1282 SB_FREEZE_FS = 3, /* For internal FS use (e.g. to stop 1283 * internal threads if needed) */ 1284 SB_FREEZE_COMPLETE = 4, /* ->freeze_fs finished successfully */ 1285 }; 1286 1287 #define SB_FREEZE_LEVELS (SB_FREEZE_COMPLETE - 1) 1288 1289 struct sb_writers { 1290 int frozen; /* Is sb frozen? */ 1291 wait_queue_head_t wait_unfrozen; /* for get_super_thawed() */ 1292 struct percpu_rw_semaphore rw_sem[SB_FREEZE_LEVELS]; 1293 }; 1294 1295 struct super_block { 1296 struct list_head s_list; /* Keep this first */ 1297 dev_t s_dev; /* search index; _not_ kdev_t */ 1298 unsigned char s_blocksize_bits; 1299 unsigned long s_blocksize; 1300 loff_t s_maxbytes; /* Max file size */ 1301 struct file_system_type *s_type; 1302 const struct super_operations *s_op; 1303 const struct dquot_operations *dq_op; 1304 const struct quotactl_ops *s_qcop; 1305 const struct export_operations *s_export_op; 1306 unsigned long s_flags; 1307 unsigned long s_iflags; /* internal SB_I_* flags */ 1308 unsigned long s_magic; 1309 struct dentry *s_root; 1310 struct rw_semaphore s_umount; 1311 int s_count; 1312 atomic_t s_active; 1313 #ifdef CONFIG_SECURITY 1314 void *s_security; 1315 #endif 1316 const struct xattr_handler **s_xattr; 1317 1318 const struct fscrypt_operations *s_cop; 1319 1320 struct hlist_bl_head s_anon; /* anonymous dentries for (nfs) exporting */ 1321 struct list_head s_mounts; /* list of mounts; _not_ for fs use */ 1322 struct block_device *s_bdev; 1323 struct backing_dev_info *s_bdi; 1324 struct mtd_info *s_mtd; 1325 struct hlist_node s_instances; 1326 unsigned int s_quota_types; /* Bitmask of supported quota types */ 1327 struct quota_info s_dquot; /* Diskquota specific options */ 1328 1329 struct sb_writers s_writers; 1330 1331 char s_id[32]; /* Informational name */ 1332 u8 s_uuid[16]; /* UUID */ 1333 1334 void *s_fs_info; /* Filesystem private info */ 1335 unsigned int s_max_links; 1336 fmode_t s_mode; 1337 1338 /* Granularity of c/m/atime in ns. 1339 Cannot be worse than a second */ 1340 u32 s_time_gran; 1341 1342 /* 1343 * The next field is for VFS *only*. No filesystems have any business 1344 * even looking at it. You had been warned. 1345 */ 1346 struct mutex s_vfs_rename_mutex; /* Kludge */ 1347 1348 /* 1349 * Filesystem subtype. If non-empty the filesystem type field 1350 * in /proc/mounts will be "type.subtype" 1351 */ 1352 char *s_subtype; 1353 1354 /* 1355 * Saved mount options for lazy filesystems using 1356 * generic_show_options() 1357 */ 1358 char __rcu *s_options; 1359 const struct dentry_operations *s_d_op; /* default d_op for dentries */ 1360 1361 /* 1362 * Saved pool identifier for cleancache (-1 means none) 1363 */ 1364 int cleancache_poolid; 1365 1366 struct shrinker s_shrink; /* per-sb shrinker handle */ 1367 1368 /* Number of inodes with nlink == 0 but still referenced */ 1369 atomic_long_t s_remove_count; 1370 1371 /* Being remounted read-only */ 1372 int s_readonly_remount; 1373 1374 /* AIO completions deferred from interrupt context */ 1375 struct workqueue_struct *s_dio_done_wq; 1376 struct hlist_head s_pins; 1377 1378 /* 1379 * Owning user namespace and default context in which to 1380 * interpret filesystem uids, gids, quotas, device nodes, 1381 * xattrs and security labels. 1382 */ 1383 struct user_namespace *s_user_ns; 1384 1385 /* 1386 * Keep the lru lists last in the structure so they always sit on their 1387 * own individual cachelines. 1388 */ 1389 struct list_lru s_dentry_lru ____cacheline_aligned_in_smp; 1390 struct list_lru s_inode_lru ____cacheline_aligned_in_smp; 1391 struct rcu_head rcu; 1392 struct work_struct destroy_work; 1393 1394 struct mutex s_sync_lock; /* sync serialisation lock */ 1395 1396 /* 1397 * Indicates how deep in a filesystem stack this SB is 1398 */ 1399 int s_stack_depth; 1400 1401 /* s_inode_list_lock protects s_inodes */ 1402 spinlock_t s_inode_list_lock ____cacheline_aligned_in_smp; 1403 struct list_head s_inodes; /* all inodes */ 1404 1405 spinlock_t s_inode_wblist_lock; 1406 struct list_head s_inodes_wb; /* writeback inodes */ 1407 }; 1408 1409 /* Helper functions so that in most cases filesystems will 1410 * not need to deal directly with kuid_t and kgid_t and can 1411 * instead deal with the raw numeric values that are stored 1412 * in the filesystem. 1413 */ 1414 static inline uid_t i_uid_read(const struct inode *inode) 1415 { 1416 return from_kuid(inode->i_sb->s_user_ns, inode->i_uid); 1417 } 1418 1419 static inline gid_t i_gid_read(const struct inode *inode) 1420 { 1421 return from_kgid(inode->i_sb->s_user_ns, inode->i_gid); 1422 } 1423 1424 static inline void i_uid_write(struct inode *inode, uid_t uid) 1425 { 1426 inode->i_uid = make_kuid(inode->i_sb->s_user_ns, uid); 1427 } 1428 1429 static inline void i_gid_write(struct inode *inode, gid_t gid) 1430 { 1431 inode->i_gid = make_kgid(inode->i_sb->s_user_ns, gid); 1432 } 1433 1434 extern struct timespec current_time(struct inode *inode); 1435 1436 /* 1437 * Snapshotting support. 1438 */ 1439 1440 void __sb_end_write(struct super_block *sb, int level); 1441 int __sb_start_write(struct super_block *sb, int level, bool wait); 1442 1443 #define __sb_writers_acquired(sb, lev) \ 1444 percpu_rwsem_acquire(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_) 1445 #define __sb_writers_release(sb, lev) \ 1446 percpu_rwsem_release(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_) 1447 1448 /** 1449 * sb_end_write - drop write access to a superblock 1450 * @sb: the super we wrote to 1451 * 1452 * Decrement number of writers to the filesystem. Wake up possible waiters 1453 * wanting to freeze the filesystem. 1454 */ 1455 static inline void sb_end_write(struct super_block *sb) 1456 { 1457 __sb_end_write(sb, SB_FREEZE_WRITE); 1458 } 1459 1460 /** 1461 * sb_end_pagefault - drop write access to a superblock from a page fault 1462 * @sb: the super we wrote to 1463 * 1464 * Decrement number of processes handling write page fault to the filesystem. 1465 * Wake up possible waiters wanting to freeze the filesystem. 1466 */ 1467 static inline void sb_end_pagefault(struct super_block *sb) 1468 { 1469 __sb_end_write(sb, SB_FREEZE_PAGEFAULT); 1470 } 1471 1472 /** 1473 * sb_end_intwrite - drop write access to a superblock for internal fs purposes 1474 * @sb: the super we wrote to 1475 * 1476 * Decrement fs-internal number of writers to the filesystem. Wake up possible 1477 * waiters wanting to freeze the filesystem. 1478 */ 1479 static inline void sb_end_intwrite(struct super_block *sb) 1480 { 1481 __sb_end_write(sb, SB_FREEZE_FS); 1482 } 1483 1484 /** 1485 * sb_start_write - get write access to a superblock 1486 * @sb: the super we write to 1487 * 1488 * When a process wants to write data or metadata to a file system (i.e. dirty 1489 * a page or an inode), it should embed the operation in a sb_start_write() - 1490 * sb_end_write() pair to get exclusion against file system freezing. This 1491 * function increments number of writers preventing freezing. If the file 1492 * system is already frozen, the function waits until the file system is 1493 * thawed. 1494 * 1495 * Since freeze protection behaves as a lock, users have to preserve 1496 * ordering of freeze protection and other filesystem locks. Generally, 1497 * freeze protection should be the outermost lock. In particular, we have: 1498 * 1499 * sb_start_write 1500 * -> i_mutex (write path, truncate, directory ops, ...) 1501 * -> s_umount (freeze_super, thaw_super) 1502 */ 1503 static inline void sb_start_write(struct super_block *sb) 1504 { 1505 __sb_start_write(sb, SB_FREEZE_WRITE, true); 1506 } 1507 1508 static inline int sb_start_write_trylock(struct super_block *sb) 1509 { 1510 return __sb_start_write(sb, SB_FREEZE_WRITE, false); 1511 } 1512 1513 /** 1514 * sb_start_pagefault - get write access to a superblock from a page fault 1515 * @sb: the super we write to 1516 * 1517 * When a process starts handling write page fault, it should embed the 1518 * operation into sb_start_pagefault() - sb_end_pagefault() pair to get 1519 * exclusion against file system freezing. This is needed since the page fault 1520 * is going to dirty a page. This function increments number of running page 1521 * faults preventing freezing. If the file system is already frozen, the 1522 * function waits until the file system is thawed. 1523 * 1524 * Since page fault freeze protection behaves as a lock, users have to preserve 1525 * ordering of freeze protection and other filesystem locks. It is advised to 1526 * put sb_start_pagefault() close to mmap_sem in lock ordering. Page fault 1527 * handling code implies lock dependency: 1528 * 1529 * mmap_sem 1530 * -> sb_start_pagefault 1531 */ 1532 static inline void sb_start_pagefault(struct super_block *sb) 1533 { 1534 __sb_start_write(sb, SB_FREEZE_PAGEFAULT, true); 1535 } 1536 1537 /* 1538 * sb_start_intwrite - get write access to a superblock for internal fs purposes 1539 * @sb: the super we write to 1540 * 1541 * This is the third level of protection against filesystem freezing. It is 1542 * free for use by a filesystem. The only requirement is that it must rank 1543 * below sb_start_pagefault. 1544 * 1545 * For example filesystem can call sb_start_intwrite() when starting a 1546 * transaction which somewhat eases handling of freezing for internal sources 1547 * of filesystem changes (internal fs threads, discarding preallocation on file 1548 * close, etc.). 1549 */ 1550 static inline void sb_start_intwrite(struct super_block *sb) 1551 { 1552 __sb_start_write(sb, SB_FREEZE_FS, true); 1553 } 1554 1555 1556 extern bool inode_owner_or_capable(const struct inode *inode); 1557 1558 /* 1559 * VFS helper functions.. 1560 */ 1561 extern int vfs_create(struct inode *, struct dentry *, umode_t, bool); 1562 extern int vfs_mkdir(struct inode *, struct dentry *, umode_t); 1563 extern int vfs_mknod(struct inode *, struct dentry *, umode_t, dev_t); 1564 extern int vfs_symlink(struct inode *, struct dentry *, const char *); 1565 extern int vfs_link(struct dentry *, struct inode *, struct dentry *, struct inode **); 1566 extern int vfs_rmdir(struct inode *, struct dentry *); 1567 extern int vfs_unlink(struct inode *, struct dentry *, struct inode **); 1568 extern int vfs_rename(struct inode *, struct dentry *, struct inode *, struct dentry *, struct inode **, unsigned int); 1569 extern int vfs_whiteout(struct inode *, struct dentry *); 1570 1571 extern struct dentry *vfs_tmpfile(struct dentry *dentry, umode_t mode, 1572 int open_flag); 1573 1574 /* 1575 * VFS file helper functions. 1576 */ 1577 extern void inode_init_owner(struct inode *inode, const struct inode *dir, 1578 umode_t mode); 1579 extern bool may_open_dev(const struct path *path); 1580 /* 1581 * VFS FS_IOC_FIEMAP helper definitions. 1582 */ 1583 struct fiemap_extent_info { 1584 unsigned int fi_flags; /* Flags as passed from user */ 1585 unsigned int fi_extents_mapped; /* Number of mapped extents */ 1586 unsigned int fi_extents_max; /* Size of fiemap_extent array */ 1587 struct fiemap_extent __user *fi_extents_start; /* Start of 1588 fiemap_extent array */ 1589 }; 1590 int fiemap_fill_next_extent(struct fiemap_extent_info *info, u64 logical, 1591 u64 phys, u64 len, u32 flags); 1592 int fiemap_check_flags(struct fiemap_extent_info *fieinfo, u32 fs_flags); 1593 1594 /* 1595 * File types 1596 * 1597 * NOTE! These match bits 12..15 of stat.st_mode 1598 * (ie "(i_mode >> 12) & 15"). 1599 */ 1600 #define DT_UNKNOWN 0 1601 #define DT_FIFO 1 1602 #define DT_CHR 2 1603 #define DT_DIR 4 1604 #define DT_BLK 6 1605 #define DT_REG 8 1606 #define DT_LNK 10 1607 #define DT_SOCK 12 1608 #define DT_WHT 14 1609 1610 /* 1611 * This is the "filldir" function type, used by readdir() to let 1612 * the kernel specify what kind of dirent layout it wants to have. 1613 * This allows the kernel to read directories into kernel space or 1614 * to have different dirent layouts depending on the binary type. 1615 */ 1616 struct dir_context; 1617 typedef int (*filldir_t)(struct dir_context *, const char *, int, loff_t, u64, 1618 unsigned); 1619 1620 struct dir_context { 1621 const filldir_t actor; 1622 loff_t pos; 1623 }; 1624 1625 struct block_device_operations; 1626 1627 /* These macros are for out of kernel modules to test that 1628 * the kernel supports the unlocked_ioctl and compat_ioctl 1629 * fields in struct file_operations. */ 1630 #define HAVE_COMPAT_IOCTL 1 1631 #define HAVE_UNLOCKED_IOCTL 1 1632 1633 /* 1634 * These flags let !MMU mmap() govern direct device mapping vs immediate 1635 * copying more easily for MAP_PRIVATE, especially for ROM filesystems. 1636 * 1637 * NOMMU_MAP_COPY: Copy can be mapped (MAP_PRIVATE) 1638 * NOMMU_MAP_DIRECT: Can be mapped directly (MAP_SHARED) 1639 * NOMMU_MAP_READ: Can be mapped for reading 1640 * NOMMU_MAP_WRITE: Can be mapped for writing 1641 * NOMMU_MAP_EXEC: Can be mapped for execution 1642 */ 1643 #define NOMMU_MAP_COPY 0x00000001 1644 #define NOMMU_MAP_DIRECT 0x00000008 1645 #define NOMMU_MAP_READ VM_MAYREAD 1646 #define NOMMU_MAP_WRITE VM_MAYWRITE 1647 #define NOMMU_MAP_EXEC VM_MAYEXEC 1648 1649 #define NOMMU_VMFLAGS \ 1650 (NOMMU_MAP_READ | NOMMU_MAP_WRITE | NOMMU_MAP_EXEC) 1651 1652 1653 struct iov_iter; 1654 1655 struct file_operations { 1656 struct module *owner; 1657 loff_t (*llseek) (struct file *, loff_t, int); 1658 ssize_t (*read) (struct file *, char __user *, size_t, loff_t *); 1659 ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *); 1660 ssize_t (*read_iter) (struct kiocb *, struct iov_iter *); 1661 ssize_t (*write_iter) (struct kiocb *, struct iov_iter *); 1662 int (*iterate) (struct file *, struct dir_context *); 1663 int (*iterate_shared) (struct file *, struct dir_context *); 1664 unsigned int (*poll) (struct file *, struct poll_table_struct *); 1665 long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long); 1666 long (*compat_ioctl) (struct file *, unsigned int, unsigned long); 1667 int (*mmap) (struct file *, struct vm_area_struct *); 1668 int (*open) (struct inode *, struct file *); 1669 int (*flush) (struct file *, fl_owner_t id); 1670 int (*release) (struct inode *, struct file *); 1671 int (*fsync) (struct file *, loff_t, loff_t, int datasync); 1672 int (*fasync) (int, struct file *, int); 1673 int (*lock) (struct file *, int, struct file_lock *); 1674 ssize_t (*sendpage) (struct file *, struct page *, int, size_t, loff_t *, int); 1675 unsigned long (*get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long); 1676 int (*check_flags)(int); 1677 int (*flock) (struct file *, int, struct file_lock *); 1678 ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, loff_t *, size_t, unsigned int); 1679 ssize_t (*splice_read)(struct file *, loff_t *, struct pipe_inode_info *, size_t, unsigned int); 1680 int (*setlease)(struct file *, long, struct file_lock **, void **); 1681 long (*fallocate)(struct file *file, int mode, loff_t offset, 1682 loff_t len); 1683 void (*show_fdinfo)(struct seq_file *m, struct file *f); 1684 #ifndef CONFIG_MMU 1685 unsigned (*mmap_capabilities)(struct file *); 1686 #endif 1687 ssize_t (*copy_file_range)(struct file *, loff_t, struct file *, 1688 loff_t, size_t, unsigned int); 1689 int (*clone_file_range)(struct file *, loff_t, struct file *, loff_t, 1690 u64); 1691 ssize_t (*dedupe_file_range)(struct file *, u64, u64, struct file *, 1692 u64); 1693 }; 1694 1695 struct inode_operations { 1696 struct dentry * (*lookup) (struct inode *,struct dentry *, unsigned int); 1697 const char * (*get_link) (struct dentry *, struct inode *, struct delayed_call *); 1698 int (*permission) (struct inode *, int); 1699 struct posix_acl * (*get_acl)(struct inode *, int); 1700 1701 int (*readlink) (struct dentry *, char __user *,int); 1702 1703 int (*create) (struct inode *,struct dentry *, umode_t, bool); 1704 int (*link) (struct dentry *,struct inode *,struct dentry *); 1705 int (*unlink) (struct inode *,struct dentry *); 1706 int (*symlink) (struct inode *,struct dentry *,const char *); 1707 int (*mkdir) (struct inode *,struct dentry *,umode_t); 1708 int (*rmdir) (struct inode *,struct dentry *); 1709 int (*mknod) (struct inode *,struct dentry *,umode_t,dev_t); 1710 int (*rename) (struct inode *, struct dentry *, 1711 struct inode *, struct dentry *, unsigned int); 1712 int (*setattr) (struct dentry *, struct iattr *); 1713 int (*getattr) (const struct path *, struct kstat *, u32, unsigned int); 1714 ssize_t (*listxattr) (struct dentry *, char *, size_t); 1715 int (*fiemap)(struct inode *, struct fiemap_extent_info *, u64 start, 1716 u64 len); 1717 int (*update_time)(struct inode *, struct timespec *, int); 1718 int (*atomic_open)(struct inode *, struct dentry *, 1719 struct file *, unsigned open_flag, 1720 umode_t create_mode, int *opened); 1721 int (*tmpfile) (struct inode *, struct dentry *, umode_t); 1722 int (*set_acl)(struct inode *, struct posix_acl *, int); 1723 } ____cacheline_aligned; 1724 1725 static inline ssize_t call_read_iter(struct file *file, struct kiocb *kio, 1726 struct iov_iter *iter) 1727 { 1728 return file->f_op->read_iter(kio, iter); 1729 } 1730 1731 static inline ssize_t call_write_iter(struct file *file, struct kiocb *kio, 1732 struct iov_iter *iter) 1733 { 1734 return file->f_op->write_iter(kio, iter); 1735 } 1736 1737 static inline int call_mmap(struct file *file, struct vm_area_struct *vma) 1738 { 1739 return file->f_op->mmap(file, vma); 1740 } 1741 1742 static inline int call_fsync(struct file *file, loff_t start, loff_t end, 1743 int datasync) 1744 { 1745 return file->f_op->fsync(file, start, end, datasync); 1746 } 1747 1748 ssize_t rw_copy_check_uvector(int type, const struct iovec __user * uvector, 1749 unsigned long nr_segs, unsigned long fast_segs, 1750 struct iovec *fast_pointer, 1751 struct iovec **ret_pointer); 1752 1753 extern ssize_t __vfs_read(struct file *, char __user *, size_t, loff_t *); 1754 extern ssize_t __vfs_write(struct file *, const char __user *, size_t, loff_t *); 1755 extern ssize_t vfs_read(struct file *, char __user *, size_t, loff_t *); 1756 extern ssize_t vfs_write(struct file *, const char __user *, size_t, loff_t *); 1757 extern ssize_t vfs_readv(struct file *, const struct iovec __user *, 1758 unsigned long, loff_t *, int); 1759 extern ssize_t vfs_writev(struct file *, const struct iovec __user *, 1760 unsigned long, loff_t *, int); 1761 extern ssize_t vfs_copy_file_range(struct file *, loff_t , struct file *, 1762 loff_t, size_t, unsigned int); 1763 extern int vfs_clone_file_prep_inodes(struct inode *inode_in, loff_t pos_in, 1764 struct inode *inode_out, loff_t pos_out, 1765 u64 *len, bool is_dedupe); 1766 extern int vfs_clone_file_range(struct file *file_in, loff_t pos_in, 1767 struct file *file_out, loff_t pos_out, u64 len); 1768 extern int vfs_dedupe_file_range_compare(struct inode *src, loff_t srcoff, 1769 struct inode *dest, loff_t destoff, 1770 loff_t len, bool *is_same); 1771 extern int vfs_dedupe_file_range(struct file *file, 1772 struct file_dedupe_range *same); 1773 1774 struct super_operations { 1775 struct inode *(*alloc_inode)(struct super_block *sb); 1776 void (*destroy_inode)(struct inode *); 1777 1778 void (*dirty_inode) (struct inode *, int flags); 1779 int (*write_inode) (struct inode *, struct writeback_control *wbc); 1780 int (*drop_inode) (struct inode *); 1781 void (*evict_inode) (struct inode *); 1782 void (*put_super) (struct super_block *); 1783 int (*sync_fs)(struct super_block *sb, int wait); 1784 int (*freeze_super) (struct super_block *); 1785 int (*freeze_fs) (struct super_block *); 1786 int (*thaw_super) (struct super_block *); 1787 int (*unfreeze_fs) (struct super_block *); 1788 int (*statfs) (struct dentry *, struct kstatfs *); 1789 int (*remount_fs) (struct super_block *, int *, char *); 1790 void (*umount_begin) (struct super_block *); 1791 1792 int (*show_options)(struct seq_file *, struct dentry *); 1793 int (*show_devname)(struct seq_file *, struct dentry *); 1794 int (*show_path)(struct seq_file *, struct dentry *); 1795 int (*show_stats)(struct seq_file *, struct dentry *); 1796 #ifdef CONFIG_QUOTA 1797 ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t); 1798 ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t); 1799 struct dquot **(*get_dquots)(struct inode *); 1800 #endif 1801 int (*bdev_try_to_free_page)(struct super_block*, struct page*, gfp_t); 1802 long (*nr_cached_objects)(struct super_block *, 1803 struct shrink_control *); 1804 long (*free_cached_objects)(struct super_block *, 1805 struct shrink_control *); 1806 }; 1807 1808 /* 1809 * Inode flags - they have no relation to superblock flags now 1810 */ 1811 #define S_SYNC 1 /* Writes are synced at once */ 1812 #define S_NOATIME 2 /* Do not update access times */ 1813 #define S_APPEND 4 /* Append-only file */ 1814 #define S_IMMUTABLE 8 /* Immutable file */ 1815 #define S_DEAD 16 /* removed, but still open directory */ 1816 #define S_NOQUOTA 32 /* Inode is not counted to quota */ 1817 #define S_DIRSYNC 64 /* Directory modifications are synchronous */ 1818 #define S_NOCMTIME 128 /* Do not update file c/mtime */ 1819 #define S_SWAPFILE 256 /* Do not truncate: swapon got its bmaps */ 1820 #define S_PRIVATE 512 /* Inode is fs-internal */ 1821 #define S_IMA 1024 /* Inode has an associated IMA struct */ 1822 #define S_AUTOMOUNT 2048 /* Automount/referral quasi-directory */ 1823 #define S_NOSEC 4096 /* no suid or xattr security attributes */ 1824 #ifdef CONFIG_FS_DAX 1825 #define S_DAX 8192 /* Direct Access, avoiding the page cache */ 1826 #else 1827 #define S_DAX 0 /* Make all the DAX code disappear */ 1828 #endif 1829 1830 /* 1831 * Note that nosuid etc flags are inode-specific: setting some file-system 1832 * flags just means all the inodes inherit those flags by default. It might be 1833 * possible to override it selectively if you really wanted to with some 1834 * ioctl() that is not currently implemented. 1835 * 1836 * Exception: MS_RDONLY is always applied to the entire file system. 1837 * 1838 * Unfortunately, it is possible to change a filesystems flags with it mounted 1839 * with files in use. This means that all of the inodes will not have their 1840 * i_flags updated. Hence, i_flags no longer inherit the superblock mount 1841 * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org 1842 */ 1843 #define __IS_FLG(inode, flg) ((inode)->i_sb->s_flags & (flg)) 1844 1845 #define IS_RDONLY(inode) ((inode)->i_sb->s_flags & MS_RDONLY) 1846 #define IS_SYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS) || \ 1847 ((inode)->i_flags & S_SYNC)) 1848 #define IS_DIRSYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS|MS_DIRSYNC) || \ 1849 ((inode)->i_flags & (S_SYNC|S_DIRSYNC))) 1850 #define IS_MANDLOCK(inode) __IS_FLG(inode, MS_MANDLOCK) 1851 #define IS_NOATIME(inode) __IS_FLG(inode, MS_RDONLY|MS_NOATIME) 1852 #define IS_I_VERSION(inode) __IS_FLG(inode, MS_I_VERSION) 1853 1854 #define IS_NOQUOTA(inode) ((inode)->i_flags & S_NOQUOTA) 1855 #define IS_APPEND(inode) ((inode)->i_flags & S_APPEND) 1856 #define IS_IMMUTABLE(inode) ((inode)->i_flags & S_IMMUTABLE) 1857 #define IS_POSIXACL(inode) __IS_FLG(inode, MS_POSIXACL) 1858 1859 #define IS_DEADDIR(inode) ((inode)->i_flags & S_DEAD) 1860 #define IS_NOCMTIME(inode) ((inode)->i_flags & S_NOCMTIME) 1861 #define IS_SWAPFILE(inode) ((inode)->i_flags & S_SWAPFILE) 1862 #define IS_PRIVATE(inode) ((inode)->i_flags & S_PRIVATE) 1863 #define IS_IMA(inode) ((inode)->i_flags & S_IMA) 1864 #define IS_AUTOMOUNT(inode) ((inode)->i_flags & S_AUTOMOUNT) 1865 #define IS_NOSEC(inode) ((inode)->i_flags & S_NOSEC) 1866 #define IS_DAX(inode) ((inode)->i_flags & S_DAX) 1867 1868 #define IS_WHITEOUT(inode) (S_ISCHR(inode->i_mode) && \ 1869 (inode)->i_rdev == WHITEOUT_DEV) 1870 1871 static inline bool HAS_UNMAPPED_ID(struct inode *inode) 1872 { 1873 return !uid_valid(inode->i_uid) || !gid_valid(inode->i_gid); 1874 } 1875 1876 /* 1877 * Inode state bits. Protected by inode->i_lock 1878 * 1879 * Three bits determine the dirty state of the inode, I_DIRTY_SYNC, 1880 * I_DIRTY_DATASYNC and I_DIRTY_PAGES. 1881 * 1882 * Four bits define the lifetime of an inode. Initially, inodes are I_NEW, 1883 * until that flag is cleared. I_WILL_FREE, I_FREEING and I_CLEAR are set at 1884 * various stages of removing an inode. 1885 * 1886 * Two bits are used for locking and completion notification, I_NEW and I_SYNC. 1887 * 1888 * I_DIRTY_SYNC Inode is dirty, but doesn't have to be written on 1889 * fdatasync(). i_atime is the usual cause. 1890 * I_DIRTY_DATASYNC Data-related inode changes pending. We keep track of 1891 * these changes separately from I_DIRTY_SYNC so that we 1892 * don't have to write inode on fdatasync() when only 1893 * mtime has changed in it. 1894 * I_DIRTY_PAGES Inode has dirty pages. Inode itself may be clean. 1895 * I_NEW Serves as both a mutex and completion notification. 1896 * New inodes set I_NEW. If two processes both create 1897 * the same inode, one of them will release its inode and 1898 * wait for I_NEW to be released before returning. 1899 * Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can 1900 * also cause waiting on I_NEW, without I_NEW actually 1901 * being set. find_inode() uses this to prevent returning 1902 * nearly-dead inodes. 1903 * I_WILL_FREE Must be set when calling write_inode_now() if i_count 1904 * is zero. I_FREEING must be set when I_WILL_FREE is 1905 * cleared. 1906 * I_FREEING Set when inode is about to be freed but still has dirty 1907 * pages or buffers attached or the inode itself is still 1908 * dirty. 1909 * I_CLEAR Added by clear_inode(). In this state the inode is 1910 * clean and can be destroyed. Inode keeps I_FREEING. 1911 * 1912 * Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are 1913 * prohibited for many purposes. iget() must wait for 1914 * the inode to be completely released, then create it 1915 * anew. Other functions will just ignore such inodes, 1916 * if appropriate. I_NEW is used for waiting. 1917 * 1918 * I_SYNC Writeback of inode is running. The bit is set during 1919 * data writeback, and cleared with a wakeup on the bit 1920 * address once it is done. The bit is also used to pin 1921 * the inode in memory for flusher thread. 1922 * 1923 * I_REFERENCED Marks the inode as recently references on the LRU list. 1924 * 1925 * I_DIO_WAKEUP Never set. Only used as a key for wait_on_bit(). 1926 * 1927 * I_WB_SWITCH Cgroup bdi_writeback switching in progress. Used to 1928 * synchronize competing switching instances and to tell 1929 * wb stat updates to grab mapping->tree_lock. See 1930 * inode_switch_wb_work_fn() for details. 1931 * 1932 * Q: What is the difference between I_WILL_FREE and I_FREEING? 1933 */ 1934 #define I_DIRTY_SYNC (1 << 0) 1935 #define I_DIRTY_DATASYNC (1 << 1) 1936 #define I_DIRTY_PAGES (1 << 2) 1937 #define __I_NEW 3 1938 #define I_NEW (1 << __I_NEW) 1939 #define I_WILL_FREE (1 << 4) 1940 #define I_FREEING (1 << 5) 1941 #define I_CLEAR (1 << 6) 1942 #define __I_SYNC 7 1943 #define I_SYNC (1 << __I_SYNC) 1944 #define I_REFERENCED (1 << 8) 1945 #define __I_DIO_WAKEUP 9 1946 #define I_DIO_WAKEUP (1 << __I_DIO_WAKEUP) 1947 #define I_LINKABLE (1 << 10) 1948 #define I_DIRTY_TIME (1 << 11) 1949 #define __I_DIRTY_TIME_EXPIRED 12 1950 #define I_DIRTY_TIME_EXPIRED (1 << __I_DIRTY_TIME_EXPIRED) 1951 #define I_WB_SWITCH (1 << 13) 1952 1953 #define I_DIRTY (I_DIRTY_SYNC | I_DIRTY_DATASYNC | I_DIRTY_PAGES) 1954 #define I_DIRTY_ALL (I_DIRTY | I_DIRTY_TIME) 1955 1956 extern void __mark_inode_dirty(struct inode *, int); 1957 static inline void mark_inode_dirty(struct inode *inode) 1958 { 1959 __mark_inode_dirty(inode, I_DIRTY); 1960 } 1961 1962 static inline void mark_inode_dirty_sync(struct inode *inode) 1963 { 1964 __mark_inode_dirty(inode, I_DIRTY_SYNC); 1965 } 1966 1967 extern void inc_nlink(struct inode *inode); 1968 extern void drop_nlink(struct inode *inode); 1969 extern void clear_nlink(struct inode *inode); 1970 extern void set_nlink(struct inode *inode, unsigned int nlink); 1971 1972 static inline void inode_inc_link_count(struct inode *inode) 1973 { 1974 inc_nlink(inode); 1975 mark_inode_dirty(inode); 1976 } 1977 1978 static inline void inode_dec_link_count(struct inode *inode) 1979 { 1980 drop_nlink(inode); 1981 mark_inode_dirty(inode); 1982 } 1983 1984 /** 1985 * inode_inc_iversion - increments i_version 1986 * @inode: inode that need to be updated 1987 * 1988 * Every time the inode is modified, the i_version field will be incremented. 1989 * The filesystem has to be mounted with i_version flag 1990 */ 1991 1992 static inline void inode_inc_iversion(struct inode *inode) 1993 { 1994 spin_lock(&inode->i_lock); 1995 inode->i_version++; 1996 spin_unlock(&inode->i_lock); 1997 } 1998 1999 enum file_time_flags { 2000 S_ATIME = 1, 2001 S_MTIME = 2, 2002 S_CTIME = 4, 2003 S_VERSION = 8, 2004 }; 2005 2006 extern void touch_atime(const struct path *); 2007 static inline void file_accessed(struct file *file) 2008 { 2009 if (!(file->f_flags & O_NOATIME)) 2010 touch_atime(&file->f_path); 2011 } 2012 2013 int sync_inode(struct inode *inode, struct writeback_control *wbc); 2014 int sync_inode_metadata(struct inode *inode, int wait); 2015 2016 struct file_system_type { 2017 const char *name; 2018 int fs_flags; 2019 #define FS_REQUIRES_DEV 1 2020 #define FS_BINARY_MOUNTDATA 2 2021 #define FS_HAS_SUBTYPE 4 2022 #define FS_USERNS_MOUNT 8 /* Can be mounted by userns root */ 2023 #define FS_RENAME_DOES_D_MOVE 32768 /* FS will handle d_move() during rename() internally. */ 2024 struct dentry *(*mount) (struct file_system_type *, int, 2025 const char *, void *); 2026 void (*kill_sb) (struct super_block *); 2027 struct module *owner; 2028 struct file_system_type * next; 2029 struct hlist_head fs_supers; 2030 2031 struct lock_class_key s_lock_key; 2032 struct lock_class_key s_umount_key; 2033 struct lock_class_key s_vfs_rename_key; 2034 struct lock_class_key s_writers_key[SB_FREEZE_LEVELS]; 2035 2036 struct lock_class_key i_lock_key; 2037 struct lock_class_key i_mutex_key; 2038 struct lock_class_key i_mutex_dir_key; 2039 }; 2040 2041 #define MODULE_ALIAS_FS(NAME) MODULE_ALIAS("fs-" NAME) 2042 2043 extern struct dentry *mount_ns(struct file_system_type *fs_type, 2044 int flags, void *data, void *ns, struct user_namespace *user_ns, 2045 int (*fill_super)(struct super_block *, void *, int)); 2046 extern struct dentry *mount_bdev(struct file_system_type *fs_type, 2047 int flags, const char *dev_name, void *data, 2048 int (*fill_super)(struct super_block *, void *, int)); 2049 extern struct dentry *mount_single(struct file_system_type *fs_type, 2050 int flags, void *data, 2051 int (*fill_super)(struct super_block *, void *, int)); 2052 extern struct dentry *mount_nodev(struct file_system_type *fs_type, 2053 int flags, void *data, 2054 int (*fill_super)(struct super_block *, void *, int)); 2055 extern struct dentry *mount_subtree(struct vfsmount *mnt, const char *path); 2056 void generic_shutdown_super(struct super_block *sb); 2057 void kill_block_super(struct super_block *sb); 2058 void kill_anon_super(struct super_block *sb); 2059 void kill_litter_super(struct super_block *sb); 2060 void deactivate_super(struct super_block *sb); 2061 void deactivate_locked_super(struct super_block *sb); 2062 int set_anon_super(struct super_block *s, void *data); 2063 int get_anon_bdev(dev_t *); 2064 void free_anon_bdev(dev_t); 2065 struct super_block *sget_userns(struct file_system_type *type, 2066 int (*test)(struct super_block *,void *), 2067 int (*set)(struct super_block *,void *), 2068 int flags, struct user_namespace *user_ns, 2069 void *data); 2070 struct super_block *sget(struct file_system_type *type, 2071 int (*test)(struct super_block *,void *), 2072 int (*set)(struct super_block *,void *), 2073 int flags, void *data); 2074 extern struct dentry *mount_pseudo_xattr(struct file_system_type *, char *, 2075 const struct super_operations *ops, 2076 const struct xattr_handler **xattr, 2077 const struct dentry_operations *dops, 2078 unsigned long); 2079 2080 static inline struct dentry * 2081 mount_pseudo(struct file_system_type *fs_type, char *name, 2082 const struct super_operations *ops, 2083 const struct dentry_operations *dops, unsigned long magic) 2084 { 2085 return mount_pseudo_xattr(fs_type, name, ops, NULL, dops, magic); 2086 } 2087 2088 /* Alas, no aliases. Too much hassle with bringing module.h everywhere */ 2089 #define fops_get(fops) \ 2090 (((fops) && try_module_get((fops)->owner) ? (fops) : NULL)) 2091 #define fops_put(fops) \ 2092 do { if (fops) module_put((fops)->owner); } while(0) 2093 /* 2094 * This one is to be used *ONLY* from ->open() instances. 2095 * fops must be non-NULL, pinned down *and* module dependencies 2096 * should be sufficient to pin the caller down as well. 2097 */ 2098 #define replace_fops(f, fops) \ 2099 do { \ 2100 struct file *__file = (f); \ 2101 fops_put(__file->f_op); \ 2102 BUG_ON(!(__file->f_op = (fops))); \ 2103 } while(0) 2104 2105 extern int register_filesystem(struct file_system_type *); 2106 extern int unregister_filesystem(struct file_system_type *); 2107 extern struct vfsmount *kern_mount_data(struct file_system_type *, void *data); 2108 #define kern_mount(type) kern_mount_data(type, NULL) 2109 extern void kern_unmount(struct vfsmount *mnt); 2110 extern int may_umount_tree(struct vfsmount *); 2111 extern int may_umount(struct vfsmount *); 2112 extern long do_mount(const char *, const char __user *, 2113 const char *, unsigned long, void *); 2114 extern struct vfsmount *collect_mounts(const struct path *); 2115 extern void drop_collected_mounts(struct vfsmount *); 2116 extern int iterate_mounts(int (*)(struct vfsmount *, void *), void *, 2117 struct vfsmount *); 2118 extern int vfs_statfs(const struct path *, struct kstatfs *); 2119 extern int user_statfs(const char __user *, struct kstatfs *); 2120 extern int fd_statfs(int, struct kstatfs *); 2121 extern int vfs_ustat(dev_t, struct kstatfs *); 2122 extern int freeze_super(struct super_block *super); 2123 extern int thaw_super(struct super_block *super); 2124 extern bool our_mnt(struct vfsmount *mnt); 2125 extern __printf(2, 3) 2126 int super_setup_bdi_name(struct super_block *sb, char *fmt, ...); 2127 extern int super_setup_bdi(struct super_block *sb); 2128 2129 extern int current_umask(void); 2130 2131 extern void ihold(struct inode * inode); 2132 extern void iput(struct inode *); 2133 extern int generic_update_time(struct inode *, struct timespec *, int); 2134 2135 /* /sys/fs */ 2136 extern struct kobject *fs_kobj; 2137 2138 #define MAX_RW_COUNT (INT_MAX & PAGE_MASK) 2139 2140 #ifdef CONFIG_MANDATORY_FILE_LOCKING 2141 extern int locks_mandatory_locked(struct file *); 2142 extern int locks_mandatory_area(struct inode *, struct file *, loff_t, loff_t, unsigned char); 2143 2144 /* 2145 * Candidates for mandatory locking have the setgid bit set 2146 * but no group execute bit - an otherwise meaningless combination. 2147 */ 2148 2149 static inline int __mandatory_lock(struct inode *ino) 2150 { 2151 return (ino->i_mode & (S_ISGID | S_IXGRP)) == S_ISGID; 2152 } 2153 2154 /* 2155 * ... and these candidates should be on MS_MANDLOCK mounted fs, 2156 * otherwise these will be advisory locks 2157 */ 2158 2159 static inline int mandatory_lock(struct inode *ino) 2160 { 2161 return IS_MANDLOCK(ino) && __mandatory_lock(ino); 2162 } 2163 2164 static inline int locks_verify_locked(struct file *file) 2165 { 2166 if (mandatory_lock(locks_inode(file))) 2167 return locks_mandatory_locked(file); 2168 return 0; 2169 } 2170 2171 static inline int locks_verify_truncate(struct inode *inode, 2172 struct file *f, 2173 loff_t size) 2174 { 2175 if (!inode->i_flctx || !mandatory_lock(inode)) 2176 return 0; 2177 2178 if (size < inode->i_size) { 2179 return locks_mandatory_area(inode, f, size, inode->i_size - 1, 2180 F_WRLCK); 2181 } else { 2182 return locks_mandatory_area(inode, f, inode->i_size, size - 1, 2183 F_WRLCK); 2184 } 2185 } 2186 2187 #else /* !CONFIG_MANDATORY_FILE_LOCKING */ 2188 2189 static inline int locks_mandatory_locked(struct file *file) 2190 { 2191 return 0; 2192 } 2193 2194 static inline int locks_mandatory_area(struct inode *inode, struct file *filp, 2195 loff_t start, loff_t end, unsigned char type) 2196 { 2197 return 0; 2198 } 2199 2200 static inline int __mandatory_lock(struct inode *inode) 2201 { 2202 return 0; 2203 } 2204 2205 static inline int mandatory_lock(struct inode *inode) 2206 { 2207 return 0; 2208 } 2209 2210 static inline int locks_verify_locked(struct file *file) 2211 { 2212 return 0; 2213 } 2214 2215 static inline int locks_verify_truncate(struct inode *inode, struct file *filp, 2216 size_t size) 2217 { 2218 return 0; 2219 } 2220 2221 #endif /* CONFIG_MANDATORY_FILE_LOCKING */ 2222 2223 2224 #ifdef CONFIG_FILE_LOCKING 2225 static inline int break_lease(struct inode *inode, unsigned int mode) 2226 { 2227 /* 2228 * Since this check is lockless, we must ensure that any refcounts 2229 * taken are done before checking i_flctx->flc_lease. Otherwise, we 2230 * could end up racing with tasks trying to set a new lease on this 2231 * file. 2232 */ 2233 smp_mb(); 2234 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease)) 2235 return __break_lease(inode, mode, FL_LEASE); 2236 return 0; 2237 } 2238 2239 static inline int break_deleg(struct inode *inode, unsigned int mode) 2240 { 2241 /* 2242 * Since this check is lockless, we must ensure that any refcounts 2243 * taken are done before checking i_flctx->flc_lease. Otherwise, we 2244 * could end up racing with tasks trying to set a new lease on this 2245 * file. 2246 */ 2247 smp_mb(); 2248 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease)) 2249 return __break_lease(inode, mode, FL_DELEG); 2250 return 0; 2251 } 2252 2253 static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode) 2254 { 2255 int ret; 2256 2257 ret = break_deleg(inode, O_WRONLY|O_NONBLOCK); 2258 if (ret == -EWOULDBLOCK && delegated_inode) { 2259 *delegated_inode = inode; 2260 ihold(inode); 2261 } 2262 return ret; 2263 } 2264 2265 static inline int break_deleg_wait(struct inode **delegated_inode) 2266 { 2267 int ret; 2268 2269 ret = break_deleg(*delegated_inode, O_WRONLY); 2270 iput(*delegated_inode); 2271 *delegated_inode = NULL; 2272 return ret; 2273 } 2274 2275 static inline int break_layout(struct inode *inode, bool wait) 2276 { 2277 smp_mb(); 2278 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease)) 2279 return __break_lease(inode, 2280 wait ? O_WRONLY : O_WRONLY | O_NONBLOCK, 2281 FL_LAYOUT); 2282 return 0; 2283 } 2284 2285 #else /* !CONFIG_FILE_LOCKING */ 2286 static inline int break_lease(struct inode *inode, unsigned int mode) 2287 { 2288 return 0; 2289 } 2290 2291 static inline int break_deleg(struct inode *inode, unsigned int mode) 2292 { 2293 return 0; 2294 } 2295 2296 static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode) 2297 { 2298 return 0; 2299 } 2300 2301 static inline int break_deleg_wait(struct inode **delegated_inode) 2302 { 2303 BUG(); 2304 return 0; 2305 } 2306 2307 static inline int break_layout(struct inode *inode, bool wait) 2308 { 2309 return 0; 2310 } 2311 2312 #endif /* CONFIG_FILE_LOCKING */ 2313 2314 /* fs/open.c */ 2315 struct audit_names; 2316 struct filename { 2317 const char *name; /* pointer to actual string */ 2318 const __user char *uptr; /* original userland pointer */ 2319 struct audit_names *aname; 2320 int refcnt; 2321 const char iname[]; 2322 }; 2323 2324 extern long vfs_truncate(const struct path *, loff_t); 2325 extern int do_truncate(struct dentry *, loff_t start, unsigned int time_attrs, 2326 struct file *filp); 2327 extern int vfs_fallocate(struct file *file, int mode, loff_t offset, 2328 loff_t len); 2329 extern long do_sys_open(int dfd, const char __user *filename, int flags, 2330 umode_t mode); 2331 extern struct file *file_open_name(struct filename *, int, umode_t); 2332 extern struct file *filp_open(const char *, int, umode_t); 2333 extern struct file *file_open_root(struct dentry *, struct vfsmount *, 2334 const char *, int, umode_t); 2335 extern struct file * dentry_open(const struct path *, int, const struct cred *); 2336 extern int filp_close(struct file *, fl_owner_t id); 2337 2338 extern struct filename *getname_flags(const char __user *, int, int *); 2339 extern struct filename *getname(const char __user *); 2340 extern struct filename *getname_kernel(const char *); 2341 extern void putname(struct filename *name); 2342 2343 enum { 2344 FILE_CREATED = 1, 2345 FILE_OPENED = 2 2346 }; 2347 extern int finish_open(struct file *file, struct dentry *dentry, 2348 int (*open)(struct inode *, struct file *), 2349 int *opened); 2350 extern int finish_no_open(struct file *file, struct dentry *dentry); 2351 2352 /* fs/ioctl.c */ 2353 2354 extern int ioctl_preallocate(struct file *filp, void __user *argp); 2355 2356 /* fs/dcache.c */ 2357 extern void __init vfs_caches_init_early(void); 2358 extern void __init vfs_caches_init(void); 2359 2360 extern struct kmem_cache *names_cachep; 2361 2362 #define __getname() kmem_cache_alloc(names_cachep, GFP_KERNEL) 2363 #define __putname(name) kmem_cache_free(names_cachep, (void *)(name)) 2364 2365 #ifdef CONFIG_BLOCK 2366 extern int register_blkdev(unsigned int, const char *); 2367 extern void unregister_blkdev(unsigned int, const char *); 2368 extern void bdev_unhash_inode(dev_t dev); 2369 extern struct block_device *bdget(dev_t); 2370 extern struct block_device *bdgrab(struct block_device *bdev); 2371 extern void bd_set_size(struct block_device *, loff_t size); 2372 extern void bd_forget(struct inode *inode); 2373 extern void bdput(struct block_device *); 2374 extern void invalidate_bdev(struct block_device *); 2375 extern void iterate_bdevs(void (*)(struct block_device *, void *), void *); 2376 extern int sync_blockdev(struct block_device *bdev); 2377 extern void kill_bdev(struct block_device *); 2378 extern struct super_block *freeze_bdev(struct block_device *); 2379 extern void emergency_thaw_all(void); 2380 extern int thaw_bdev(struct block_device *bdev, struct super_block *sb); 2381 extern int fsync_bdev(struct block_device *); 2382 2383 extern struct super_block *blockdev_superblock; 2384 2385 static inline bool sb_is_blkdev_sb(struct super_block *sb) 2386 { 2387 return sb == blockdev_superblock; 2388 } 2389 #else 2390 static inline void bd_forget(struct inode *inode) {} 2391 static inline int sync_blockdev(struct block_device *bdev) { return 0; } 2392 static inline void kill_bdev(struct block_device *bdev) {} 2393 static inline void invalidate_bdev(struct block_device *bdev) {} 2394 2395 static inline struct super_block *freeze_bdev(struct block_device *sb) 2396 { 2397 return NULL; 2398 } 2399 2400 static inline int thaw_bdev(struct block_device *bdev, struct super_block *sb) 2401 { 2402 return 0; 2403 } 2404 2405 static inline void iterate_bdevs(void (*f)(struct block_device *, void *), void *arg) 2406 { 2407 } 2408 2409 static inline bool sb_is_blkdev_sb(struct super_block *sb) 2410 { 2411 return false; 2412 } 2413 #endif 2414 extern int sync_filesystem(struct super_block *); 2415 extern const struct file_operations def_blk_fops; 2416 extern const struct file_operations def_chr_fops; 2417 #ifdef CONFIG_BLOCK 2418 extern int ioctl_by_bdev(struct block_device *, unsigned, unsigned long); 2419 extern int blkdev_ioctl(struct block_device *, fmode_t, unsigned, unsigned long); 2420 extern long compat_blkdev_ioctl(struct file *, unsigned, unsigned long); 2421 extern int blkdev_get(struct block_device *bdev, fmode_t mode, void *holder); 2422 extern struct block_device *blkdev_get_by_path(const char *path, fmode_t mode, 2423 void *holder); 2424 extern struct block_device *blkdev_get_by_dev(dev_t dev, fmode_t mode, 2425 void *holder); 2426 extern void blkdev_put(struct block_device *bdev, fmode_t mode); 2427 extern int __blkdev_reread_part(struct block_device *bdev); 2428 extern int blkdev_reread_part(struct block_device *bdev); 2429 2430 #ifdef CONFIG_SYSFS 2431 extern int bd_link_disk_holder(struct block_device *bdev, struct gendisk *disk); 2432 extern void bd_unlink_disk_holder(struct block_device *bdev, 2433 struct gendisk *disk); 2434 #else 2435 static inline int bd_link_disk_holder(struct block_device *bdev, 2436 struct gendisk *disk) 2437 { 2438 return 0; 2439 } 2440 static inline void bd_unlink_disk_holder(struct block_device *bdev, 2441 struct gendisk *disk) 2442 { 2443 } 2444 #endif 2445 #endif 2446 2447 /* fs/char_dev.c */ 2448 #define CHRDEV_MAJOR_HASH_SIZE 255 2449 /* Marks the bottom of the first segment of free char majors */ 2450 #define CHRDEV_MAJOR_DYN_END 234 2451 extern int alloc_chrdev_region(dev_t *, unsigned, unsigned, const char *); 2452 extern int register_chrdev_region(dev_t, unsigned, const char *); 2453 extern int __register_chrdev(unsigned int major, unsigned int baseminor, 2454 unsigned int count, const char *name, 2455 const struct file_operations *fops); 2456 extern void __unregister_chrdev(unsigned int major, unsigned int baseminor, 2457 unsigned int count, const char *name); 2458 extern void unregister_chrdev_region(dev_t, unsigned); 2459 extern void chrdev_show(struct seq_file *,off_t); 2460 2461 static inline int register_chrdev(unsigned int major, const char *name, 2462 const struct file_operations *fops) 2463 { 2464 return __register_chrdev(major, 0, 256, name, fops); 2465 } 2466 2467 static inline void unregister_chrdev(unsigned int major, const char *name) 2468 { 2469 __unregister_chrdev(major, 0, 256, name); 2470 } 2471 2472 /* fs/block_dev.c */ 2473 #define BDEVNAME_SIZE 32 /* Largest string for a blockdev identifier */ 2474 #define BDEVT_SIZE 10 /* Largest string for MAJ:MIN for blkdev */ 2475 2476 #ifdef CONFIG_BLOCK 2477 #define BLKDEV_MAJOR_HASH_SIZE 255 2478 extern const char *__bdevname(dev_t, char *buffer); 2479 extern const char *bdevname(struct block_device *bdev, char *buffer); 2480 extern struct block_device *lookup_bdev(const char *); 2481 extern void blkdev_show(struct seq_file *,off_t); 2482 2483 #else 2484 #define BLKDEV_MAJOR_HASH_SIZE 0 2485 #endif 2486 2487 extern void init_special_inode(struct inode *, umode_t, dev_t); 2488 2489 /* Invalid inode operations -- fs/bad_inode.c */ 2490 extern void make_bad_inode(struct inode *); 2491 extern bool is_bad_inode(struct inode *); 2492 2493 #ifdef CONFIG_BLOCK 2494 extern void check_disk_size_change(struct gendisk *disk, 2495 struct block_device *bdev); 2496 extern int revalidate_disk(struct gendisk *); 2497 extern int check_disk_change(struct block_device *); 2498 extern int __invalidate_device(struct block_device *, bool); 2499 extern int invalidate_partition(struct gendisk *, int); 2500 #endif 2501 unsigned long invalidate_mapping_pages(struct address_space *mapping, 2502 pgoff_t start, pgoff_t end); 2503 2504 static inline void invalidate_remote_inode(struct inode *inode) 2505 { 2506 if (S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) || 2507 S_ISLNK(inode->i_mode)) 2508 invalidate_mapping_pages(inode->i_mapping, 0, -1); 2509 } 2510 extern int invalidate_inode_pages2(struct address_space *mapping); 2511 extern int invalidate_inode_pages2_range(struct address_space *mapping, 2512 pgoff_t start, pgoff_t end); 2513 extern int write_inode_now(struct inode *, int); 2514 extern int filemap_fdatawrite(struct address_space *); 2515 extern int filemap_flush(struct address_space *); 2516 extern int filemap_fdatawait(struct address_space *); 2517 extern void filemap_fdatawait_keep_errors(struct address_space *); 2518 extern int filemap_fdatawait_range(struct address_space *, loff_t lstart, 2519 loff_t lend); 2520 extern int filemap_write_and_wait(struct address_space *mapping); 2521 extern int filemap_write_and_wait_range(struct address_space *mapping, 2522 loff_t lstart, loff_t lend); 2523 extern int __filemap_fdatawrite_range(struct address_space *mapping, 2524 loff_t start, loff_t end, int sync_mode); 2525 extern int filemap_fdatawrite_range(struct address_space *mapping, 2526 loff_t start, loff_t end); 2527 extern int filemap_check_errors(struct address_space *mapping); 2528 2529 extern int vfs_fsync_range(struct file *file, loff_t start, loff_t end, 2530 int datasync); 2531 extern int vfs_fsync(struct file *file, int datasync); 2532 2533 /* 2534 * Sync the bytes written if this was a synchronous write. Expect ki_pos 2535 * to already be updated for the write, and will return either the amount 2536 * of bytes passed in, or an error if syncing the file failed. 2537 */ 2538 static inline ssize_t generic_write_sync(struct kiocb *iocb, ssize_t count) 2539 { 2540 if (iocb->ki_flags & IOCB_DSYNC) { 2541 int ret = vfs_fsync_range(iocb->ki_filp, 2542 iocb->ki_pos - count, iocb->ki_pos - 1, 2543 (iocb->ki_flags & IOCB_SYNC) ? 0 : 1); 2544 if (ret) 2545 return ret; 2546 } 2547 2548 return count; 2549 } 2550 2551 extern void emergency_sync(void); 2552 extern void emergency_remount(void); 2553 #ifdef CONFIG_BLOCK 2554 extern sector_t bmap(struct inode *, sector_t); 2555 #endif 2556 extern int notify_change(struct dentry *, struct iattr *, struct inode **); 2557 extern int inode_permission(struct inode *, int); 2558 extern int __inode_permission(struct inode *, int); 2559 extern int generic_permission(struct inode *, int); 2560 extern int __check_sticky(struct inode *dir, struct inode *inode); 2561 2562 static inline bool execute_ok(struct inode *inode) 2563 { 2564 return (inode->i_mode & S_IXUGO) || S_ISDIR(inode->i_mode); 2565 } 2566 2567 static inline void file_start_write(struct file *file) 2568 { 2569 if (!S_ISREG(file_inode(file)->i_mode)) 2570 return; 2571 __sb_start_write(file_inode(file)->i_sb, SB_FREEZE_WRITE, true); 2572 } 2573 2574 static inline bool file_start_write_trylock(struct file *file) 2575 { 2576 if (!S_ISREG(file_inode(file)->i_mode)) 2577 return true; 2578 return __sb_start_write(file_inode(file)->i_sb, SB_FREEZE_WRITE, false); 2579 } 2580 2581 static inline void file_end_write(struct file *file) 2582 { 2583 if (!S_ISREG(file_inode(file)->i_mode)) 2584 return; 2585 __sb_end_write(file_inode(file)->i_sb, SB_FREEZE_WRITE); 2586 } 2587 2588 static inline int do_clone_file_range(struct file *file_in, loff_t pos_in, 2589 struct file *file_out, loff_t pos_out, 2590 u64 len) 2591 { 2592 int ret; 2593 2594 file_start_write(file_out); 2595 ret = vfs_clone_file_range(file_in, pos_in, file_out, pos_out, len); 2596 file_end_write(file_out); 2597 2598 return ret; 2599 } 2600 2601 /* 2602 * get_write_access() gets write permission for a file. 2603 * put_write_access() releases this write permission. 2604 * This is used for regular files. 2605 * We cannot support write (and maybe mmap read-write shared) accesses and 2606 * MAP_DENYWRITE mmappings simultaneously. The i_writecount field of an inode 2607 * can have the following values: 2608 * 0: no writers, no VM_DENYWRITE mappings 2609 * < 0: (-i_writecount) vm_area_structs with VM_DENYWRITE set exist 2610 * > 0: (i_writecount) users are writing to the file. 2611 * 2612 * Normally we operate on that counter with atomic_{inc,dec} and it's safe 2613 * except for the cases where we don't hold i_writecount yet. Then we need to 2614 * use {get,deny}_write_access() - these functions check the sign and refuse 2615 * to do the change if sign is wrong. 2616 */ 2617 static inline int get_write_access(struct inode *inode) 2618 { 2619 return atomic_inc_unless_negative(&inode->i_writecount) ? 0 : -ETXTBSY; 2620 } 2621 static inline int deny_write_access(struct file *file) 2622 { 2623 struct inode *inode = file_inode(file); 2624 return atomic_dec_unless_positive(&inode->i_writecount) ? 0 : -ETXTBSY; 2625 } 2626 static inline void put_write_access(struct inode * inode) 2627 { 2628 atomic_dec(&inode->i_writecount); 2629 } 2630 static inline void allow_write_access(struct file *file) 2631 { 2632 if (file) 2633 atomic_inc(&file_inode(file)->i_writecount); 2634 } 2635 static inline bool inode_is_open_for_write(const struct inode *inode) 2636 { 2637 return atomic_read(&inode->i_writecount) > 0; 2638 } 2639 2640 #ifdef CONFIG_IMA 2641 static inline void i_readcount_dec(struct inode *inode) 2642 { 2643 BUG_ON(!atomic_read(&inode->i_readcount)); 2644 atomic_dec(&inode->i_readcount); 2645 } 2646 static inline void i_readcount_inc(struct inode *inode) 2647 { 2648 atomic_inc(&inode->i_readcount); 2649 } 2650 #else 2651 static inline void i_readcount_dec(struct inode *inode) 2652 { 2653 return; 2654 } 2655 static inline void i_readcount_inc(struct inode *inode) 2656 { 2657 return; 2658 } 2659 #endif 2660 extern int do_pipe_flags(int *, int); 2661 2662 #define __kernel_read_file_id(id) \ 2663 id(UNKNOWN, unknown) \ 2664 id(FIRMWARE, firmware) \ 2665 id(FIRMWARE_PREALLOC_BUFFER, firmware) \ 2666 id(MODULE, kernel-module) \ 2667 id(KEXEC_IMAGE, kexec-image) \ 2668 id(KEXEC_INITRAMFS, kexec-initramfs) \ 2669 id(POLICY, security-policy) \ 2670 id(MAX_ID, ) 2671 2672 #define __fid_enumify(ENUM, dummy) READING_ ## ENUM, 2673 #define __fid_stringify(dummy, str) #str, 2674 2675 enum kernel_read_file_id { 2676 __kernel_read_file_id(__fid_enumify) 2677 }; 2678 2679 static const char * const kernel_read_file_str[] = { 2680 __kernel_read_file_id(__fid_stringify) 2681 }; 2682 2683 static inline const char *kernel_read_file_id_str(enum kernel_read_file_id id) 2684 { 2685 if ((unsigned)id >= READING_MAX_ID) 2686 return kernel_read_file_str[READING_UNKNOWN]; 2687 2688 return kernel_read_file_str[id]; 2689 } 2690 2691 extern int kernel_read(struct file *, loff_t, char *, unsigned long); 2692 extern int kernel_read_file(struct file *, void **, loff_t *, loff_t, 2693 enum kernel_read_file_id); 2694 extern int kernel_read_file_from_path(char *, void **, loff_t *, loff_t, 2695 enum kernel_read_file_id); 2696 extern int kernel_read_file_from_fd(int, void **, loff_t *, loff_t, 2697 enum kernel_read_file_id); 2698 extern ssize_t kernel_write(struct file *, const char *, size_t, loff_t); 2699 extern ssize_t __kernel_write(struct file *, const char *, size_t, loff_t *); 2700 extern struct file * open_exec(const char *); 2701 2702 /* fs/dcache.c -- generic fs support functions */ 2703 extern bool is_subdir(struct dentry *, struct dentry *); 2704 extern bool path_is_under(const struct path *, const struct path *); 2705 2706 extern char *file_path(struct file *, char *, int); 2707 2708 #include <linux/err.h> 2709 2710 /* needed for stackable file system support */ 2711 extern loff_t default_llseek(struct file *file, loff_t offset, int whence); 2712 2713 extern loff_t vfs_llseek(struct file *file, loff_t offset, int whence); 2714 2715 extern int inode_init_always(struct super_block *, struct inode *); 2716 extern void inode_init_once(struct inode *); 2717 extern void address_space_init_once(struct address_space *mapping); 2718 extern struct inode * igrab(struct inode *); 2719 extern ino_t iunique(struct super_block *, ino_t); 2720 extern int inode_needs_sync(struct inode *inode); 2721 extern int generic_delete_inode(struct inode *inode); 2722 static inline int generic_drop_inode(struct inode *inode) 2723 { 2724 return !inode->i_nlink || inode_unhashed(inode); 2725 } 2726 2727 extern struct inode *ilookup5_nowait(struct super_block *sb, 2728 unsigned long hashval, int (*test)(struct inode *, void *), 2729 void *data); 2730 extern struct inode *ilookup5(struct super_block *sb, unsigned long hashval, 2731 int (*test)(struct inode *, void *), void *data); 2732 extern struct inode *ilookup(struct super_block *sb, unsigned long ino); 2733 2734 extern struct inode * iget5_locked(struct super_block *, unsigned long, int (*test)(struct inode *, void *), int (*set)(struct inode *, void *), void *); 2735 extern struct inode * iget_locked(struct super_block *, unsigned long); 2736 extern struct inode *find_inode_nowait(struct super_block *, 2737 unsigned long, 2738 int (*match)(struct inode *, 2739 unsigned long, void *), 2740 void *data); 2741 extern int insert_inode_locked4(struct inode *, unsigned long, int (*test)(struct inode *, void *), void *); 2742 extern int insert_inode_locked(struct inode *); 2743 #ifdef CONFIG_DEBUG_LOCK_ALLOC 2744 extern void lockdep_annotate_inode_mutex_key(struct inode *inode); 2745 #else 2746 static inline void lockdep_annotate_inode_mutex_key(struct inode *inode) { }; 2747 #endif 2748 extern void unlock_new_inode(struct inode *); 2749 extern unsigned int get_next_ino(void); 2750 2751 extern void __iget(struct inode * inode); 2752 extern void iget_failed(struct inode *); 2753 extern void clear_inode(struct inode *); 2754 extern void __destroy_inode(struct inode *); 2755 extern struct inode *new_inode_pseudo(struct super_block *sb); 2756 extern struct inode *new_inode(struct super_block *sb); 2757 extern void free_inode_nonrcu(struct inode *inode); 2758 extern int should_remove_suid(struct dentry *); 2759 extern int file_remove_privs(struct file *); 2760 2761 extern void __insert_inode_hash(struct inode *, unsigned long hashval); 2762 static inline void insert_inode_hash(struct inode *inode) 2763 { 2764 __insert_inode_hash(inode, inode->i_ino); 2765 } 2766 2767 extern void __remove_inode_hash(struct inode *); 2768 static inline void remove_inode_hash(struct inode *inode) 2769 { 2770 if (!inode_unhashed(inode) && !hlist_fake(&inode->i_hash)) 2771 __remove_inode_hash(inode); 2772 } 2773 2774 extern void inode_sb_list_add(struct inode *inode); 2775 2776 #ifdef CONFIG_BLOCK 2777 extern int bdev_read_only(struct block_device *); 2778 #endif 2779 extern int set_blocksize(struct block_device *, int); 2780 extern int sb_set_blocksize(struct super_block *, int); 2781 extern int sb_min_blocksize(struct super_block *, int); 2782 2783 extern int generic_file_mmap(struct file *, struct vm_area_struct *); 2784 extern int generic_file_readonly_mmap(struct file *, struct vm_area_struct *); 2785 extern ssize_t generic_write_checks(struct kiocb *, struct iov_iter *); 2786 extern ssize_t generic_file_read_iter(struct kiocb *, struct iov_iter *); 2787 extern ssize_t __generic_file_write_iter(struct kiocb *, struct iov_iter *); 2788 extern ssize_t generic_file_write_iter(struct kiocb *, struct iov_iter *); 2789 extern ssize_t generic_file_direct_write(struct kiocb *, struct iov_iter *); 2790 extern ssize_t generic_perform_write(struct file *, struct iov_iter *, loff_t); 2791 2792 ssize_t vfs_iter_read(struct file *file, struct iov_iter *iter, loff_t *ppos); 2793 ssize_t vfs_iter_write(struct file *file, struct iov_iter *iter, loff_t *ppos); 2794 2795 /* fs/block_dev.c */ 2796 extern ssize_t blkdev_read_iter(struct kiocb *iocb, struct iov_iter *to); 2797 extern ssize_t blkdev_write_iter(struct kiocb *iocb, struct iov_iter *from); 2798 extern int blkdev_fsync(struct file *filp, loff_t start, loff_t end, 2799 int datasync); 2800 extern void block_sync_page(struct page *page); 2801 2802 /* fs/splice.c */ 2803 extern ssize_t generic_file_splice_read(struct file *, loff_t *, 2804 struct pipe_inode_info *, size_t, unsigned int); 2805 extern ssize_t iter_file_splice_write(struct pipe_inode_info *, 2806 struct file *, loff_t *, size_t, unsigned int); 2807 extern ssize_t generic_splice_sendpage(struct pipe_inode_info *pipe, 2808 struct file *out, loff_t *, size_t len, unsigned int flags); 2809 extern long do_splice_direct(struct file *in, loff_t *ppos, struct file *out, 2810 loff_t *opos, size_t len, unsigned int flags); 2811 2812 2813 extern void 2814 file_ra_state_init(struct file_ra_state *ra, struct address_space *mapping); 2815 extern loff_t noop_llseek(struct file *file, loff_t offset, int whence); 2816 extern loff_t no_llseek(struct file *file, loff_t offset, int whence); 2817 extern loff_t vfs_setpos(struct file *file, loff_t offset, loff_t maxsize); 2818 extern loff_t generic_file_llseek(struct file *file, loff_t offset, int whence); 2819 extern loff_t generic_file_llseek_size(struct file *file, loff_t offset, 2820 int whence, loff_t maxsize, loff_t eof); 2821 extern loff_t fixed_size_llseek(struct file *file, loff_t offset, 2822 int whence, loff_t size); 2823 extern loff_t no_seek_end_llseek_size(struct file *, loff_t, int, loff_t); 2824 extern loff_t no_seek_end_llseek(struct file *, loff_t, int); 2825 extern int generic_file_open(struct inode * inode, struct file * filp); 2826 extern int nonseekable_open(struct inode * inode, struct file * filp); 2827 2828 #ifdef CONFIG_BLOCK 2829 typedef void (dio_submit_t)(struct bio *bio, struct inode *inode, 2830 loff_t file_offset); 2831 2832 enum { 2833 /* need locking between buffered and direct access */ 2834 DIO_LOCKING = 0x01, 2835 2836 /* filesystem does not support filling holes */ 2837 DIO_SKIP_HOLES = 0x02, 2838 2839 /* filesystem can handle aio writes beyond i_size */ 2840 DIO_ASYNC_EXTEND = 0x04, 2841 2842 /* inode/fs/bdev does not need truncate protection */ 2843 DIO_SKIP_DIO_COUNT = 0x08, 2844 }; 2845 2846 void dio_end_io(struct bio *bio, int error); 2847 2848 ssize_t __blockdev_direct_IO(struct kiocb *iocb, struct inode *inode, 2849 struct block_device *bdev, struct iov_iter *iter, 2850 get_block_t get_block, 2851 dio_iodone_t end_io, dio_submit_t submit_io, 2852 int flags); 2853 2854 static inline ssize_t blockdev_direct_IO(struct kiocb *iocb, 2855 struct inode *inode, 2856 struct iov_iter *iter, 2857 get_block_t get_block) 2858 { 2859 return __blockdev_direct_IO(iocb, inode, inode->i_sb->s_bdev, iter, 2860 get_block, NULL, NULL, DIO_LOCKING | DIO_SKIP_HOLES); 2861 } 2862 #endif 2863 2864 void inode_dio_wait(struct inode *inode); 2865 2866 /* 2867 * inode_dio_begin - signal start of a direct I/O requests 2868 * @inode: inode the direct I/O happens on 2869 * 2870 * This is called once we've finished processing a direct I/O request, 2871 * and is used to wake up callers waiting for direct I/O to be quiesced. 2872 */ 2873 static inline void inode_dio_begin(struct inode *inode) 2874 { 2875 atomic_inc(&inode->i_dio_count); 2876 } 2877 2878 /* 2879 * inode_dio_end - signal finish of a direct I/O requests 2880 * @inode: inode the direct I/O happens on 2881 * 2882 * This is called once we've finished processing a direct I/O request, 2883 * and is used to wake up callers waiting for direct I/O to be quiesced. 2884 */ 2885 static inline void inode_dio_end(struct inode *inode) 2886 { 2887 if (atomic_dec_and_test(&inode->i_dio_count)) 2888 wake_up_bit(&inode->i_state, __I_DIO_WAKEUP); 2889 } 2890 2891 extern void inode_set_flags(struct inode *inode, unsigned int flags, 2892 unsigned int mask); 2893 2894 extern const struct file_operations generic_ro_fops; 2895 2896 #define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m)) 2897 2898 extern int readlink_copy(char __user *, int, const char *); 2899 extern int page_readlink(struct dentry *, char __user *, int); 2900 extern const char *page_get_link(struct dentry *, struct inode *, 2901 struct delayed_call *); 2902 extern void page_put_link(void *); 2903 extern int __page_symlink(struct inode *inode, const char *symname, int len, 2904 int nofs); 2905 extern int page_symlink(struct inode *inode, const char *symname, int len); 2906 extern const struct inode_operations page_symlink_inode_operations; 2907 extern void kfree_link(void *); 2908 extern void generic_fillattr(struct inode *, struct kstat *); 2909 extern int vfs_getattr_nosec(const struct path *, struct kstat *, u32, unsigned int); 2910 extern int vfs_getattr(const struct path *, struct kstat *, u32, unsigned int); 2911 void __inode_add_bytes(struct inode *inode, loff_t bytes); 2912 void inode_add_bytes(struct inode *inode, loff_t bytes); 2913 void __inode_sub_bytes(struct inode *inode, loff_t bytes); 2914 void inode_sub_bytes(struct inode *inode, loff_t bytes); 2915 loff_t inode_get_bytes(struct inode *inode); 2916 void inode_set_bytes(struct inode *inode, loff_t bytes); 2917 const char *simple_get_link(struct dentry *, struct inode *, 2918 struct delayed_call *); 2919 extern const struct inode_operations simple_symlink_inode_operations; 2920 2921 extern int iterate_dir(struct file *, struct dir_context *); 2922 2923 extern int vfs_statx(int, const char __user *, int, struct kstat *, u32); 2924 extern int vfs_statx_fd(unsigned int, struct kstat *, u32, unsigned int); 2925 2926 static inline int vfs_stat(const char __user *filename, struct kstat *stat) 2927 { 2928 return vfs_statx(AT_FDCWD, filename, AT_NO_AUTOMOUNT, 2929 stat, STATX_BASIC_STATS); 2930 } 2931 static inline int vfs_lstat(const char __user *name, struct kstat *stat) 2932 { 2933 return vfs_statx(AT_FDCWD, name, AT_SYMLINK_NOFOLLOW | AT_NO_AUTOMOUNT, 2934 stat, STATX_BASIC_STATS); 2935 } 2936 static inline int vfs_fstatat(int dfd, const char __user *filename, 2937 struct kstat *stat, int flags) 2938 { 2939 return vfs_statx(dfd, filename, flags | AT_NO_AUTOMOUNT, 2940 stat, STATX_BASIC_STATS); 2941 } 2942 static inline int vfs_fstat(int fd, struct kstat *stat) 2943 { 2944 return vfs_statx_fd(fd, stat, STATX_BASIC_STATS, 0); 2945 } 2946 2947 2948 extern const char *vfs_get_link(struct dentry *, struct delayed_call *); 2949 extern int vfs_readlink(struct dentry *, char __user *, int); 2950 2951 extern int __generic_block_fiemap(struct inode *inode, 2952 struct fiemap_extent_info *fieinfo, 2953 loff_t start, loff_t len, 2954 get_block_t *get_block); 2955 extern int generic_block_fiemap(struct inode *inode, 2956 struct fiemap_extent_info *fieinfo, u64 start, 2957 u64 len, get_block_t *get_block); 2958 2959 extern void get_filesystem(struct file_system_type *fs); 2960 extern void put_filesystem(struct file_system_type *fs); 2961 extern struct file_system_type *get_fs_type(const char *name); 2962 extern struct super_block *get_super(struct block_device *); 2963 extern struct super_block *get_super_thawed(struct block_device *); 2964 extern struct super_block *get_super_exclusive_thawed(struct block_device *bdev); 2965 extern struct super_block *get_active_super(struct block_device *bdev); 2966 extern void drop_super(struct super_block *sb); 2967 extern void drop_super_exclusive(struct super_block *sb); 2968 extern void iterate_supers(void (*)(struct super_block *, void *), void *); 2969 extern void iterate_supers_type(struct file_system_type *, 2970 void (*)(struct super_block *, void *), void *); 2971 2972 extern int dcache_dir_open(struct inode *, struct file *); 2973 extern int dcache_dir_close(struct inode *, struct file *); 2974 extern loff_t dcache_dir_lseek(struct file *, loff_t, int); 2975 extern int dcache_readdir(struct file *, struct dir_context *); 2976 extern int simple_setattr(struct dentry *, struct iattr *); 2977 extern int simple_getattr(const struct path *, struct kstat *, u32, unsigned int); 2978 extern int simple_statfs(struct dentry *, struct kstatfs *); 2979 extern int simple_open(struct inode *inode, struct file *file); 2980 extern int simple_link(struct dentry *, struct inode *, struct dentry *); 2981 extern int simple_unlink(struct inode *, struct dentry *); 2982 extern int simple_rmdir(struct inode *, struct dentry *); 2983 extern int simple_rename(struct inode *, struct dentry *, 2984 struct inode *, struct dentry *, unsigned int); 2985 extern int noop_fsync(struct file *, loff_t, loff_t, int); 2986 extern int simple_empty(struct dentry *); 2987 extern int simple_readpage(struct file *file, struct page *page); 2988 extern int simple_write_begin(struct file *file, struct address_space *mapping, 2989 loff_t pos, unsigned len, unsigned flags, 2990 struct page **pagep, void **fsdata); 2991 extern int simple_write_end(struct file *file, struct address_space *mapping, 2992 loff_t pos, unsigned len, unsigned copied, 2993 struct page *page, void *fsdata); 2994 extern int always_delete_dentry(const struct dentry *); 2995 extern struct inode *alloc_anon_inode(struct super_block *); 2996 extern int simple_nosetlease(struct file *, long, struct file_lock **, void **); 2997 extern const struct dentry_operations simple_dentry_operations; 2998 2999 extern struct dentry *simple_lookup(struct inode *, struct dentry *, unsigned int flags); 3000 extern ssize_t generic_read_dir(struct file *, char __user *, size_t, loff_t *); 3001 extern const struct file_operations simple_dir_operations; 3002 extern const struct inode_operations simple_dir_inode_operations; 3003 extern void make_empty_dir_inode(struct inode *inode); 3004 extern bool is_empty_dir_inode(struct inode *inode); 3005 struct tree_descr { const char *name; const struct file_operations *ops; int mode; }; 3006 struct dentry *d_alloc_name(struct dentry *, const char *); 3007 extern int simple_fill_super(struct super_block *, unsigned long, 3008 const struct tree_descr *); 3009 extern int simple_pin_fs(struct file_system_type *, struct vfsmount **mount, int *count); 3010 extern void simple_release_fs(struct vfsmount **mount, int *count); 3011 3012 extern ssize_t simple_read_from_buffer(void __user *to, size_t count, 3013 loff_t *ppos, const void *from, size_t available); 3014 extern ssize_t simple_write_to_buffer(void *to, size_t available, loff_t *ppos, 3015 const void __user *from, size_t count); 3016 3017 extern int __generic_file_fsync(struct file *, loff_t, loff_t, int); 3018 extern int generic_file_fsync(struct file *, loff_t, loff_t, int); 3019 3020 extern int generic_check_addressable(unsigned, u64); 3021 3022 #ifdef CONFIG_MIGRATION 3023 extern int buffer_migrate_page(struct address_space *, 3024 struct page *, struct page *, 3025 enum migrate_mode); 3026 #else 3027 #define buffer_migrate_page NULL 3028 #endif 3029 3030 extern int setattr_prepare(struct dentry *, struct iattr *); 3031 extern int inode_newsize_ok(const struct inode *, loff_t offset); 3032 extern void setattr_copy(struct inode *inode, const struct iattr *attr); 3033 3034 extern int file_update_time(struct file *file); 3035 3036 extern int generic_show_options(struct seq_file *m, struct dentry *root); 3037 extern void save_mount_options(struct super_block *sb, char *options); 3038 extern void replace_mount_options(struct super_block *sb, char *options); 3039 3040 static inline bool io_is_direct(struct file *filp) 3041 { 3042 return (filp->f_flags & O_DIRECT) || IS_DAX(filp->f_mapping->host); 3043 } 3044 3045 static inline int iocb_flags(struct file *file) 3046 { 3047 int res = 0; 3048 if (file->f_flags & O_APPEND) 3049 res |= IOCB_APPEND; 3050 if (io_is_direct(file)) 3051 res |= IOCB_DIRECT; 3052 if ((file->f_flags & O_DSYNC) || IS_SYNC(file->f_mapping->host)) 3053 res |= IOCB_DSYNC; 3054 if (file->f_flags & __O_SYNC) 3055 res |= IOCB_SYNC; 3056 return res; 3057 } 3058 3059 static inline ino_t parent_ino(struct dentry *dentry) 3060 { 3061 ino_t res; 3062 3063 /* 3064 * Don't strictly need d_lock here? If the parent ino could change 3065 * then surely we'd have a deeper race in the caller? 3066 */ 3067 spin_lock(&dentry->d_lock); 3068 res = dentry->d_parent->d_inode->i_ino; 3069 spin_unlock(&dentry->d_lock); 3070 return res; 3071 } 3072 3073 /* Transaction based IO helpers */ 3074 3075 /* 3076 * An argresp is stored in an allocated page and holds the 3077 * size of the argument or response, along with its content 3078 */ 3079 struct simple_transaction_argresp { 3080 ssize_t size; 3081 char data[0]; 3082 }; 3083 3084 #define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp)) 3085 3086 char *simple_transaction_get(struct file *file, const char __user *buf, 3087 size_t size); 3088 ssize_t simple_transaction_read(struct file *file, char __user *buf, 3089 size_t size, loff_t *pos); 3090 int simple_transaction_release(struct inode *inode, struct file *file); 3091 3092 void simple_transaction_set(struct file *file, size_t n); 3093 3094 /* 3095 * simple attribute files 3096 * 3097 * These attributes behave similar to those in sysfs: 3098 * 3099 * Writing to an attribute immediately sets a value, an open file can be 3100 * written to multiple times. 3101 * 3102 * Reading from an attribute creates a buffer from the value that might get 3103 * read with multiple read calls. When the attribute has been read 3104 * completely, no further read calls are possible until the file is opened 3105 * again. 3106 * 3107 * All attributes contain a text representation of a numeric value 3108 * that are accessed with the get() and set() functions. 3109 */ 3110 #define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt) \ 3111 static int __fops ## _open(struct inode *inode, struct file *file) \ 3112 { \ 3113 __simple_attr_check_format(__fmt, 0ull); \ 3114 return simple_attr_open(inode, file, __get, __set, __fmt); \ 3115 } \ 3116 static const struct file_operations __fops = { \ 3117 .owner = THIS_MODULE, \ 3118 .open = __fops ## _open, \ 3119 .release = simple_attr_release, \ 3120 .read = simple_attr_read, \ 3121 .write = simple_attr_write, \ 3122 .llseek = generic_file_llseek, \ 3123 } 3124 3125 static inline __printf(1, 2) 3126 void __simple_attr_check_format(const char *fmt, ...) 3127 { 3128 /* don't do anything, just let the compiler check the arguments; */ 3129 } 3130 3131 int simple_attr_open(struct inode *inode, struct file *file, 3132 int (*get)(void *, u64 *), int (*set)(void *, u64), 3133 const char *fmt); 3134 int simple_attr_release(struct inode *inode, struct file *file); 3135 ssize_t simple_attr_read(struct file *file, char __user *buf, 3136 size_t len, loff_t *ppos); 3137 ssize_t simple_attr_write(struct file *file, const char __user *buf, 3138 size_t len, loff_t *ppos); 3139 3140 struct ctl_table; 3141 int proc_nr_files(struct ctl_table *table, int write, 3142 void __user *buffer, size_t *lenp, loff_t *ppos); 3143 int proc_nr_dentry(struct ctl_table *table, int write, 3144 void __user *buffer, size_t *lenp, loff_t *ppos); 3145 int proc_nr_inodes(struct ctl_table *table, int write, 3146 void __user *buffer, size_t *lenp, loff_t *ppos); 3147 int __init get_filesystem_list(char *buf); 3148 3149 #define __FMODE_EXEC ((__force int) FMODE_EXEC) 3150 #define __FMODE_NONOTIFY ((__force int) FMODE_NONOTIFY) 3151 3152 #define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE]) 3153 #define OPEN_FMODE(flag) ((__force fmode_t)(((flag + 1) & O_ACCMODE) | \ 3154 (flag & __FMODE_NONOTIFY))) 3155 3156 static inline bool is_sxid(umode_t mode) 3157 { 3158 return (mode & S_ISUID) || ((mode & S_ISGID) && (mode & S_IXGRP)); 3159 } 3160 3161 static inline int check_sticky(struct inode *dir, struct inode *inode) 3162 { 3163 if (!(dir->i_mode & S_ISVTX)) 3164 return 0; 3165 3166 return __check_sticky(dir, inode); 3167 } 3168 3169 static inline void inode_has_no_xattr(struct inode *inode) 3170 { 3171 if (!is_sxid(inode->i_mode) && (inode->i_sb->s_flags & MS_NOSEC)) 3172 inode->i_flags |= S_NOSEC; 3173 } 3174 3175 static inline bool is_root_inode(struct inode *inode) 3176 { 3177 return inode == inode->i_sb->s_root->d_inode; 3178 } 3179 3180 static inline bool dir_emit(struct dir_context *ctx, 3181 const char *name, int namelen, 3182 u64 ino, unsigned type) 3183 { 3184 return ctx->actor(ctx, name, namelen, ctx->pos, ino, type) == 0; 3185 } 3186 static inline bool dir_emit_dot(struct file *file, struct dir_context *ctx) 3187 { 3188 return ctx->actor(ctx, ".", 1, ctx->pos, 3189 file->f_path.dentry->d_inode->i_ino, DT_DIR) == 0; 3190 } 3191 static inline bool dir_emit_dotdot(struct file *file, struct dir_context *ctx) 3192 { 3193 return ctx->actor(ctx, "..", 2, ctx->pos, 3194 parent_ino(file->f_path.dentry), DT_DIR) == 0; 3195 } 3196 static inline bool dir_emit_dots(struct file *file, struct dir_context *ctx) 3197 { 3198 if (ctx->pos == 0) { 3199 if (!dir_emit_dot(file, ctx)) 3200 return false; 3201 ctx->pos = 1; 3202 } 3203 if (ctx->pos == 1) { 3204 if (!dir_emit_dotdot(file, ctx)) 3205 return false; 3206 ctx->pos = 2; 3207 } 3208 return true; 3209 } 3210 static inline bool dir_relax(struct inode *inode) 3211 { 3212 inode_unlock(inode); 3213 inode_lock(inode); 3214 return !IS_DEADDIR(inode); 3215 } 3216 3217 static inline bool dir_relax_shared(struct inode *inode) 3218 { 3219 inode_unlock_shared(inode); 3220 inode_lock_shared(inode); 3221 return !IS_DEADDIR(inode); 3222 } 3223 3224 extern bool path_noexec(const struct path *path); 3225 extern void inode_nohighmem(struct inode *inode); 3226 3227 #endif /* _LINUX_FS_H */ 3228