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 #include <linux/uuid.h>
35 #include <asm/byteorder.h>
36 #include <uapi/linux/fs.h>
38 struct backing_dev_info;
41 struct export_operations;
46 struct pipe_inode_info;
47 struct poll_table_struct;
49 struct vm_area_struct;
52 struct swap_info_struct;
54 struct workqueue_struct;
57 struct fscrypt_operations;
59 extern void __init inode_init(void);
60 extern void __init inode_init_early(void);
61 extern void __init files_init(void);
62 extern void __init files_maxfiles_init(void);
64 extern struct files_stat_struct files_stat;
65 extern unsigned long get_max_files(void);
66 extern unsigned int sysctl_nr_open;
67 extern struct inodes_stat_t inodes_stat;
68 extern int leases_enable, lease_break_time;
69 extern int sysctl_protected_symlinks;
70 extern int sysctl_protected_hardlinks;
73 typedef int (get_block_t)(struct inode *inode, sector_t iblock,
74 struct buffer_head *bh_result, int create);
75 typedef int (dio_iodone_t)(struct kiocb *iocb, loff_t offset,
76 ssize_t bytes, void *private);
78 #define MAY_EXEC 0x00000001
79 #define MAY_WRITE 0x00000002
80 #define MAY_READ 0x00000004
81 #define MAY_APPEND 0x00000008
82 #define MAY_ACCESS 0x00000010
83 #define MAY_OPEN 0x00000020
84 #define MAY_CHDIR 0x00000040
85 /* called from RCU mode, don't block */
86 #define MAY_NOT_BLOCK 0x00000080
89 * flags in file.f_mode. Note that FMODE_READ and FMODE_WRITE must correspond
90 * to O_WRONLY and O_RDWR via the strange trick in __dentry_open()
93 /* file is open for reading */
94 #define FMODE_READ ((__force fmode_t)0x1)
95 /* file is open for writing */
96 #define FMODE_WRITE ((__force fmode_t)0x2)
97 /* file is seekable */
98 #define FMODE_LSEEK ((__force fmode_t)0x4)
99 /* file can be accessed using pread */
100 #define FMODE_PREAD ((__force fmode_t)0x8)
101 /* file can be accessed using pwrite */
102 #define FMODE_PWRITE ((__force fmode_t)0x10)
103 /* File is opened for execution with sys_execve / sys_uselib */
104 #define FMODE_EXEC ((__force fmode_t)0x20)
105 /* File is opened with O_NDELAY (only set for block devices) */
106 #define FMODE_NDELAY ((__force fmode_t)0x40)
107 /* File is opened with O_EXCL (only set for block devices) */
108 #define FMODE_EXCL ((__force fmode_t)0x80)
109 /* File is opened using open(.., 3, ..) and is writeable only for ioctls
110 (specialy hack for floppy.c) */
111 #define FMODE_WRITE_IOCTL ((__force fmode_t)0x100)
112 /* 32bit hashes as llseek() offset (for directories) */
113 #define FMODE_32BITHASH ((__force fmode_t)0x200)
114 /* 64bit hashes as llseek() offset (for directories) */
115 #define FMODE_64BITHASH ((__force fmode_t)0x400)
118 * Don't update ctime and mtime.
120 * Currently a special hack for the XFS open_by_handle ioctl, but we'll
121 * hopefully graduate it to a proper O_CMTIME flag supported by open(2) soon.
123 #define FMODE_NOCMTIME ((__force fmode_t)0x800)
125 /* Expect random access pattern */
126 #define FMODE_RANDOM ((__force fmode_t)0x1000)
128 /* File is huge (eg. /dev/kmem): treat loff_t as unsigned */
129 #define FMODE_UNSIGNED_OFFSET ((__force fmode_t)0x2000)
131 /* File is opened with O_PATH; almost nothing can be done with it */
132 #define FMODE_PATH ((__force fmode_t)0x4000)
134 /* File needs atomic accesses to f_pos */
135 #define FMODE_ATOMIC_POS ((__force fmode_t)0x8000)
136 /* Write access to underlying fs */
137 #define FMODE_WRITER ((__force fmode_t)0x10000)
138 /* Has read method(s) */
139 #define FMODE_CAN_READ ((__force fmode_t)0x20000)
140 /* Has write method(s) */
141 #define FMODE_CAN_WRITE ((__force fmode_t)0x40000)
143 /* File was opened by fanotify and shouldn't generate fanotify events */
144 #define FMODE_NONOTIFY ((__force fmode_t)0x4000000)
146 /* File is capable of returning -EAGAIN if AIO will block */
147 #define FMODE_AIO_NOWAIT ((__force fmode_t)0x8000000)
150 * Flag for rw_copy_check_uvector and compat_rw_copy_check_uvector
151 * that indicates that they should check the contents of the iovec are
152 * valid, but not check the memory that the iovec elements
155 #define CHECK_IOVEC_ONLY -1
158 * Attribute flags. These should be or-ed together to figure out what
161 #define ATTR_MODE (1 << 0)
162 #define ATTR_UID (1 << 1)
163 #define ATTR_GID (1 << 2)
164 #define ATTR_SIZE (1 << 3)
165 #define ATTR_ATIME (1 << 4)
166 #define ATTR_MTIME (1 << 5)
167 #define ATTR_CTIME (1 << 6)
168 #define ATTR_ATIME_SET (1 << 7)
169 #define ATTR_MTIME_SET (1 << 8)
170 #define ATTR_FORCE (1 << 9) /* Not a change, but a change it */
171 #define ATTR_ATTR_FLAG (1 << 10)
172 #define ATTR_KILL_SUID (1 << 11)
173 #define ATTR_KILL_SGID (1 << 12)
174 #define ATTR_FILE (1 << 13)
175 #define ATTR_KILL_PRIV (1 << 14)
176 #define ATTR_OPEN (1 << 15) /* Truncating from open(O_TRUNC) */
177 #define ATTR_TIMES_SET (1 << 16)
178 #define ATTR_TOUCH (1 << 17)
181 * Whiteout is represented by a char device. The following constants define the
182 * mode and device number to use.
184 #define WHITEOUT_MODE 0
185 #define WHITEOUT_DEV 0
188 * This is the Inode Attributes structure, used for notify_change(). It
189 * uses the above definitions as flags, to know which values have changed.
190 * Also, in this manner, a Filesystem can look at only the values it cares
191 * about. Basically, these are the attributes that the VFS layer can
192 * request to change from the FS layer.
194 * Derek Atkins <warlord@MIT.EDU> 94-10-20
197 unsigned int ia_valid;
202 struct timespec ia_atime;
203 struct timespec ia_mtime;
204 struct timespec ia_ctime;
207 * Not an attribute, but an auxiliary info for filesystems wanting to
208 * implement an ftruncate() like method. NOTE: filesystem should
209 * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL).
211 struct file *ia_file;
215 * Includes for diskquotas.
217 #include <linux/quota.h>
220 * Maximum number of layers of fs stack. Needs to be limited to
221 * prevent kernel stack overflow
223 #define FILESYSTEM_MAX_STACK_DEPTH 2
226 * enum positive_aop_returns - aop return codes with specific semantics
228 * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has
229 * completed, that the page is still locked, and
230 * should be considered active. The VM uses this hint
231 * to return the page to the active list -- it won't
232 * be a candidate for writeback again in the near
233 * future. Other callers must be careful to unlock
234 * the page if they get this return. Returned by
237 * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has
238 * unlocked it and the page might have been truncated.
239 * The caller should back up to acquiring a new page and
240 * trying again. The aop will be taking reasonable
241 * precautions not to livelock. If the caller held a page
242 * reference, it should drop it before retrying. Returned
245 * address_space_operation functions return these large constants to indicate
246 * special semantics to the caller. These are much larger than the bytes in a
247 * page to allow for functions that return the number of bytes operated on in a
251 enum positive_aop_returns {
252 AOP_WRITEPAGE_ACTIVATE = 0x80000,
253 AOP_TRUNCATED_PAGE = 0x80001,
256 #define AOP_FLAG_CONT_EXPAND 0x0001 /* called from cont_expand */
257 #define AOP_FLAG_NOFS 0x0002 /* used by filesystem to direct
258 * helper code (eg buffer layer)
259 * to clear GFP_FS from alloc */
262 * oh the beauties of C type declarations.
265 struct address_space;
266 struct writeback_control;
268 #define IOCB_EVENTFD (1 << 0)
269 #define IOCB_APPEND (1 << 1)
270 #define IOCB_DIRECT (1 << 2)
271 #define IOCB_HIPRI (1 << 3)
272 #define IOCB_DSYNC (1 << 4)
273 #define IOCB_SYNC (1 << 5)
274 #define IOCB_WRITE (1 << 6)
275 #define IOCB_NOWAIT (1 << 7)
278 struct file *ki_filp;
280 void (*ki_complete)(struct kiocb *iocb, long ret, long ret2);
285 static inline bool is_sync_kiocb(struct kiocb *kiocb)
287 return kiocb->ki_complete == NULL;
290 static inline int iocb_flags(struct file *file);
292 static inline void init_sync_kiocb(struct kiocb *kiocb, struct file *filp)
294 *kiocb = (struct kiocb) {
296 .ki_flags = iocb_flags(filp),
301 * "descriptor" for what we're up to with a read.
302 * This allows us to use the same read code yet
303 * have multiple different users of the data that
304 * we read from a file.
306 * The simplest case just copies the data to user
319 typedef int (*read_actor_t)(read_descriptor_t *, struct page *,
320 unsigned long, unsigned long);
322 struct address_space_operations {
323 int (*writepage)(struct page *page, struct writeback_control *wbc);
324 int (*readpage)(struct file *, struct page *);
326 /* Write back some dirty pages from this mapping. */
327 int (*writepages)(struct address_space *, struct writeback_control *);
329 /* Set a page dirty. Return true if this dirtied it */
330 int (*set_page_dirty)(struct page *page);
332 int (*readpages)(struct file *filp, struct address_space *mapping,
333 struct list_head *pages, unsigned nr_pages);
335 int (*write_begin)(struct file *, struct address_space *mapping,
336 loff_t pos, unsigned len, unsigned flags,
337 struct page **pagep, void **fsdata);
338 int (*write_end)(struct file *, struct address_space *mapping,
339 loff_t pos, unsigned len, unsigned copied,
340 struct page *page, void *fsdata);
342 /* Unfortunately this kludge is needed for FIBMAP. Don't use it */
343 sector_t (*bmap)(struct address_space *, sector_t);
344 void (*invalidatepage) (struct page *, unsigned int, unsigned int);
345 int (*releasepage) (struct page *, gfp_t);
346 void (*freepage)(struct page *);
347 ssize_t (*direct_IO)(struct kiocb *, struct iov_iter *iter);
349 * migrate the contents of a page to the specified target. If
350 * migrate_mode is MIGRATE_ASYNC, it must not block.
352 int (*migratepage) (struct address_space *,
353 struct page *, struct page *, enum migrate_mode);
354 bool (*isolate_page)(struct page *, isolate_mode_t);
355 void (*putback_page)(struct page *);
356 int (*launder_page) (struct page *);
357 int (*is_partially_uptodate) (struct page *, unsigned long,
359 void (*is_dirty_writeback) (struct page *, bool *, bool *);
360 int (*error_remove_page)(struct address_space *, struct page *);
362 /* swapfile support */
363 int (*swap_activate)(struct swap_info_struct *sis, struct file *file,
365 void (*swap_deactivate)(struct file *file);
368 extern const struct address_space_operations empty_aops;
371 * pagecache_write_begin/pagecache_write_end must be used by general code
372 * to write into the pagecache.
374 int pagecache_write_begin(struct file *, struct address_space *mapping,
375 loff_t pos, unsigned len, unsigned flags,
376 struct page **pagep, void **fsdata);
378 int pagecache_write_end(struct file *, struct address_space *mapping,
379 loff_t pos, unsigned len, unsigned copied,
380 struct page *page, void *fsdata);
382 struct address_space {
383 struct inode *host; /* owner: inode, block_device */
384 struct radix_tree_root page_tree; /* radix tree of all pages */
385 spinlock_t tree_lock; /* and lock protecting it */
386 atomic_t i_mmap_writable;/* count VM_SHARED mappings */
387 struct rb_root i_mmap; /* tree of private and shared mappings */
388 struct rw_semaphore i_mmap_rwsem; /* protect tree, count, list */
389 /* Protected by tree_lock together with the radix tree */
390 unsigned long nrpages; /* number of total pages */
391 /* number of shadow or DAX exceptional entries */
392 unsigned long nrexceptional;
393 pgoff_t writeback_index;/* writeback starts here */
394 const struct address_space_operations *a_ops; /* methods */
395 unsigned long flags; /* error bits */
396 spinlock_t private_lock; /* for use by the address_space */
397 gfp_t gfp_mask; /* implicit gfp mask for allocations */
398 struct list_head private_list; /* ditto */
399 void *private_data; /* ditto */
400 } __attribute__((aligned(sizeof(long))));
402 * On most architectures that alignment is already the case; but
403 * must be enforced here for CRIS, to let the least significant bit
404 * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
406 struct request_queue;
408 struct block_device {
409 dev_t bd_dev; /* not a kdev_t - it's a search key */
411 struct inode * bd_inode; /* will die */
412 struct super_block * bd_super;
413 struct mutex bd_mutex; /* open/close mutex */
417 bool bd_write_holder;
419 struct list_head bd_holder_disks;
421 struct block_device * bd_contains;
422 unsigned bd_block_size;
423 struct hd_struct * bd_part;
424 /* number of times partitions within this device have been opened. */
425 unsigned bd_part_count;
427 struct gendisk * bd_disk;
428 struct request_queue * bd_queue;
429 struct backing_dev_info *bd_bdi;
430 struct list_head bd_list;
432 * Private data. You must have bd_claim'ed the block_device
433 * to use this. NOTE: bd_claim allows an owner to claim
434 * the same device multiple times, the owner must take special
435 * care to not mess up bd_private for that case.
437 unsigned long bd_private;
439 /* The counter of freeze processes */
440 int bd_fsfreeze_count;
441 /* Mutex for freeze */
442 struct mutex bd_fsfreeze_mutex;
446 * Radix-tree tags, for tagging dirty and writeback pages within the pagecache
449 #define PAGECACHE_TAG_DIRTY 0
450 #define PAGECACHE_TAG_WRITEBACK 1
451 #define PAGECACHE_TAG_TOWRITE 2
453 int mapping_tagged(struct address_space *mapping, int tag);
455 static inline void i_mmap_lock_write(struct address_space *mapping)
457 down_write(&mapping->i_mmap_rwsem);
460 static inline void i_mmap_unlock_write(struct address_space *mapping)
462 up_write(&mapping->i_mmap_rwsem);
465 static inline void i_mmap_lock_read(struct address_space *mapping)
467 down_read(&mapping->i_mmap_rwsem);
470 static inline void i_mmap_unlock_read(struct address_space *mapping)
472 up_read(&mapping->i_mmap_rwsem);
476 * Might pages of this file be mapped into userspace?
478 static inline int mapping_mapped(struct address_space *mapping)
480 return !RB_EMPTY_ROOT(&mapping->i_mmap);
484 * Might pages of this file have been modified in userspace?
485 * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap_pgoff
486 * marks vma as VM_SHARED if it is shared, and the file was opened for
487 * writing i.e. vma may be mprotected writable even if now readonly.
489 * If i_mmap_writable is negative, no new writable mappings are allowed. You
490 * can only deny writable mappings, if none exists right now.
492 static inline int mapping_writably_mapped(struct address_space *mapping)
494 return atomic_read(&mapping->i_mmap_writable) > 0;
497 static inline int mapping_map_writable(struct address_space *mapping)
499 return atomic_inc_unless_negative(&mapping->i_mmap_writable) ?
503 static inline void mapping_unmap_writable(struct address_space *mapping)
505 atomic_dec(&mapping->i_mmap_writable);
508 static inline int mapping_deny_writable(struct address_space *mapping)
510 return atomic_dec_unless_positive(&mapping->i_mmap_writable) ?
514 static inline void mapping_allow_writable(struct address_space *mapping)
516 atomic_inc(&mapping->i_mmap_writable);
520 * Use sequence counter to get consistent i_size on 32-bit processors.
522 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
523 #include <linux/seqlock.h>
524 #define __NEED_I_SIZE_ORDERED
525 #define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
527 #define i_size_ordered_init(inode) do { } while (0)
531 #define ACL_NOT_CACHED ((void *)(-1))
532 #define ACL_DONT_CACHE ((void *)(-3))
534 static inline struct posix_acl *
535 uncached_acl_sentinel(struct task_struct *task)
537 return (void *)task + 1;
541 is_uncached_acl(struct posix_acl *acl)
543 return (long)acl & 1;
546 #define IOP_FASTPERM 0x0001
547 #define IOP_LOOKUP 0x0002
548 #define IOP_NOFOLLOW 0x0004
549 #define IOP_XATTR 0x0008
550 #define IOP_DEFAULT_READLINK 0x0010
552 struct fsnotify_mark_connector;
555 * Keep mostly read-only and often accessed (especially for
556 * the RCU path lookup and 'stat' data) fields at the beginning
557 * of the 'struct inode'
561 unsigned short i_opflags;
564 unsigned int i_flags;
566 #ifdef CONFIG_FS_POSIX_ACL
567 struct posix_acl *i_acl;
568 struct posix_acl *i_default_acl;
571 const struct inode_operations *i_op;
572 struct super_block *i_sb;
573 struct address_space *i_mapping;
575 #ifdef CONFIG_SECURITY
579 /* Stat data, not accessed from path walking */
582 * Filesystems may only read i_nlink directly. They shall use the
583 * following functions for modification:
585 * (set|clear|inc|drop)_nlink
586 * inode_(inc|dec)_link_count
589 const unsigned int i_nlink;
590 unsigned int __i_nlink;
594 struct timespec i_atime;
595 struct timespec i_mtime;
596 struct timespec i_ctime;
597 spinlock_t i_lock; /* i_blocks, i_bytes, maybe i_size */
598 unsigned short i_bytes;
599 unsigned int i_blkbits;
602 #ifdef __NEED_I_SIZE_ORDERED
603 seqcount_t i_size_seqcount;
607 unsigned long i_state;
608 struct rw_semaphore i_rwsem;
610 unsigned long dirtied_when; /* jiffies of first dirtying */
611 unsigned long dirtied_time_when;
613 struct hlist_node i_hash;
614 struct list_head i_io_list; /* backing dev IO list */
615 #ifdef CONFIG_CGROUP_WRITEBACK
616 struct bdi_writeback *i_wb; /* the associated cgroup wb */
618 /* foreign inode detection, see wbc_detach_inode() */
620 u16 i_wb_frn_avg_time;
621 u16 i_wb_frn_history;
623 struct list_head i_lru; /* inode LRU list */
624 struct list_head i_sb_list;
625 struct list_head i_wb_list; /* backing dev writeback list */
627 struct hlist_head i_dentry;
628 struct rcu_head i_rcu;
632 atomic_t i_dio_count;
633 atomic_t i_writecount;
635 atomic_t i_readcount; /* struct files open RO */
637 const struct file_operations *i_fop; /* former ->i_op->default_file_ops */
638 struct file_lock_context *i_flctx;
639 struct address_space i_data;
640 struct list_head i_devices;
642 struct pipe_inode_info *i_pipe;
643 struct block_device *i_bdev;
651 #ifdef CONFIG_FSNOTIFY
652 __u32 i_fsnotify_mask; /* all events this inode cares about */
653 struct fsnotify_mark_connector __rcu *i_fsnotify_marks;
656 #if IS_ENABLED(CONFIG_FS_ENCRYPTION)
657 struct fscrypt_info *i_crypt_info;
660 void *i_private; /* fs or device private pointer */
663 static inline unsigned int i_blocksize(const struct inode *node)
665 return (1 << node->i_blkbits);
668 static inline int inode_unhashed(struct inode *inode)
670 return hlist_unhashed(&inode->i_hash);
674 * inode->i_mutex nesting subclasses for the lock validator:
676 * 0: the object of the current VFS operation
680 * 4: second non-directory
681 * 5: second parent (when locking independent directories in rename)
683 * I_MUTEX_NONDIR2 is for certain operations (such as rename) which lock two
684 * non-directories at once.
686 * The locking order between these classes is
687 * parent[2] -> child -> grandchild -> normal -> xattr -> second non-directory
689 enum inode_i_mutex_lock_class
699 static inline void inode_lock(struct inode *inode)
701 down_write(&inode->i_rwsem);
704 static inline void inode_unlock(struct inode *inode)
706 up_write(&inode->i_rwsem);
709 static inline void inode_lock_shared(struct inode *inode)
711 down_read(&inode->i_rwsem);
714 static inline void inode_unlock_shared(struct inode *inode)
716 up_read(&inode->i_rwsem);
719 static inline int inode_trylock(struct inode *inode)
721 return down_write_trylock(&inode->i_rwsem);
724 static inline int inode_trylock_shared(struct inode *inode)
726 return down_read_trylock(&inode->i_rwsem);
729 static inline int inode_is_locked(struct inode *inode)
731 return rwsem_is_locked(&inode->i_rwsem);
734 static inline void inode_lock_nested(struct inode *inode, unsigned subclass)
736 down_write_nested(&inode->i_rwsem, subclass);
739 void lock_two_nondirectories(struct inode *, struct inode*);
740 void unlock_two_nondirectories(struct inode *, struct inode*);
743 * NOTE: in a 32bit arch with a preemptable kernel and
744 * an UP compile the i_size_read/write must be atomic
745 * with respect to the local cpu (unlike with preempt disabled),
746 * but they don't need to be atomic with respect to other cpus like in
747 * true SMP (so they need either to either locally disable irq around
748 * the read or for example on x86 they can be still implemented as a
749 * cmpxchg8b without the need of the lock prefix). For SMP compiles
750 * and 64bit archs it makes no difference if preempt is enabled or not.
752 static inline loff_t i_size_read(const struct inode *inode)
754 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
759 seq = read_seqcount_begin(&inode->i_size_seqcount);
760 i_size = inode->i_size;
761 } while (read_seqcount_retry(&inode->i_size_seqcount, seq));
763 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
767 i_size = inode->i_size;
771 return inode->i_size;
776 * NOTE: unlike i_size_read(), i_size_write() does need locking around it
777 * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount
778 * can be lost, resulting in subsequent i_size_read() calls spinning forever.
780 static inline void i_size_write(struct inode *inode, loff_t i_size)
782 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
784 write_seqcount_begin(&inode->i_size_seqcount);
785 inode->i_size = i_size;
786 write_seqcount_end(&inode->i_size_seqcount);
788 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
790 inode->i_size = i_size;
793 inode->i_size = i_size;
797 static inline unsigned iminor(const struct inode *inode)
799 return MINOR(inode->i_rdev);
802 static inline unsigned imajor(const struct inode *inode)
804 return MAJOR(inode->i_rdev);
807 extern struct block_device *I_BDEV(struct inode *inode);
810 rwlock_t lock; /* protects pid, uid, euid fields */
811 struct pid *pid; /* pid or -pgrp where SIGIO should be sent */
812 enum pid_type pid_type; /* Kind of process group SIGIO should be sent to */
813 kuid_t uid, euid; /* uid/euid of process setting the owner */
814 int signum; /* posix.1b rt signal to be delivered on IO */
818 * Track a single file's readahead state
820 struct file_ra_state {
821 pgoff_t start; /* where readahead started */
822 unsigned int size; /* # of readahead pages */
823 unsigned int async_size; /* do asynchronous readahead when
824 there are only # of pages ahead */
826 unsigned int ra_pages; /* Maximum readahead window */
827 unsigned int mmap_miss; /* Cache miss stat for mmap accesses */
828 loff_t prev_pos; /* Cache last read() position */
832 * Check if @index falls in the readahead windows.
834 static inline int ra_has_index(struct file_ra_state *ra, pgoff_t index)
836 return (index >= ra->start &&
837 index < ra->start + ra->size);
842 struct llist_node fu_llist;
843 struct rcu_head fu_rcuhead;
846 struct inode *f_inode; /* cached value */
847 const struct file_operations *f_op;
850 * Protects f_ep_links, f_flags.
851 * Must not be taken from IRQ context.
854 atomic_long_t f_count;
855 unsigned int f_flags;
857 struct mutex f_pos_lock;
859 struct fown_struct f_owner;
860 const struct cred *f_cred;
861 struct file_ra_state f_ra;
864 #ifdef CONFIG_SECURITY
867 /* needed for tty driver, and maybe others */
871 /* Used by fs/eventpoll.c to link all the hooks to this file */
872 struct list_head f_ep_links;
873 struct list_head f_tfile_llink;
874 #endif /* #ifdef CONFIG_EPOLL */
875 struct address_space *f_mapping;
876 } __attribute__((aligned(4))); /* lest something weird decides that 2 is OK */
881 /* file identifier */
882 unsigned char f_handle[0];
885 static inline struct file *get_file(struct file *f)
887 atomic_long_inc(&f->f_count);
890 #define get_file_rcu(x) atomic_long_inc_not_zero(&(x)->f_count)
891 #define fput_atomic(x) atomic_long_add_unless(&(x)->f_count, -1, 1)
892 #define file_count(x) atomic_long_read(&(x)->f_count)
894 #define MAX_NON_LFS ((1UL<<31) - 1)
896 /* Page cache limit. The filesystems should put that into their s_maxbytes
897 limits, otherwise bad things can happen in VM. */
898 #if BITS_PER_LONG==32
899 #define MAX_LFS_FILESIZE (((loff_t)PAGE_SIZE << (BITS_PER_LONG-1))-1)
900 #elif BITS_PER_LONG==64
901 #define MAX_LFS_FILESIZE ((loff_t)0x7fffffffffffffffLL)
906 #define FL_DELEG 4 /* NFSv4 delegation */
907 #define FL_ACCESS 8 /* not trying to lock, just looking */
908 #define FL_EXISTS 16 /* when unlocking, test for existence */
909 #define FL_LEASE 32 /* lease held on this file */
910 #define FL_CLOSE 64 /* unlock on close */
911 #define FL_SLEEP 128 /* A blocking lock */
912 #define FL_DOWNGRADE_PENDING 256 /* Lease is being downgraded */
913 #define FL_UNLOCK_PENDING 512 /* Lease is being broken */
914 #define FL_OFDLCK 1024 /* lock is "owned" by struct file */
915 #define FL_LAYOUT 2048 /* outstanding pNFS layout */
917 #define FL_CLOSE_POSIX (FL_POSIX | FL_CLOSE)
920 * Special return value from posix_lock_file() and vfs_lock_file() for
921 * asynchronous locking.
923 #define FILE_LOCK_DEFERRED 1
925 /* legacy typedef, should eventually be removed */
926 typedef void *fl_owner_t;
930 struct file_lock_operations {
931 void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
932 void (*fl_release_private)(struct file_lock *);
935 struct lock_manager_operations {
936 int (*lm_compare_owner)(struct file_lock *, struct file_lock *);
937 unsigned long (*lm_owner_key)(struct file_lock *);
938 fl_owner_t (*lm_get_owner)(fl_owner_t);
939 void (*lm_put_owner)(fl_owner_t);
940 void (*lm_notify)(struct file_lock *); /* unblock callback */
941 int (*lm_grant)(struct file_lock *, int);
942 bool (*lm_break)(struct file_lock *);
943 int (*lm_change)(struct file_lock *, int, struct list_head *);
944 void (*lm_setup)(struct file_lock *, void **);
947 struct lock_manager {
948 struct list_head list;
950 * NFSv4 and up also want opens blocked during the grace period;
957 void locks_start_grace(struct net *, struct lock_manager *);
958 void locks_end_grace(struct lock_manager *);
959 int locks_in_grace(struct net *);
960 int opens_in_grace(struct net *);
962 /* that will die - we need it for nfs_lock_info */
963 #include <linux/nfs_fs_i.h>
966 * struct file_lock represents a generic "file lock". It's used to represent
967 * POSIX byte range locks, BSD (flock) locks, and leases. It's important to
968 * note that the same struct is used to represent both a request for a lock and
969 * the lock itself, but the same object is never used for both.
971 * FIXME: should we create a separate "struct lock_request" to help distinguish
974 * The varous i_flctx lists are ordered by:
977 * 2) lock range start
980 * Obviously, the last two criteria only matter for POSIX locks.
983 struct file_lock *fl_next; /* singly linked list for this inode */
984 struct list_head fl_list; /* link into file_lock_context */
985 struct hlist_node fl_link; /* node in global lists */
986 struct list_head fl_block; /* circular list of blocked processes */
988 unsigned int fl_flags;
989 unsigned char fl_type;
991 int fl_link_cpu; /* what cpu's list is this on? */
992 struct pid *fl_nspid;
993 wait_queue_head_t fl_wait;
994 struct file *fl_file;
998 struct fasync_struct * fl_fasync; /* for lease break notifications */
999 /* for lease breaks: */
1000 unsigned long fl_break_time;
1001 unsigned long fl_downgrade_time;
1003 const struct file_lock_operations *fl_ops; /* Callbacks for filesystems */
1004 const struct lock_manager_operations *fl_lmops; /* Callbacks for lockmanagers */
1006 struct nfs_lock_info nfs_fl;
1007 struct nfs4_lock_info nfs4_fl;
1009 struct list_head link; /* link in AFS vnode's pending_locks list */
1010 int state; /* state of grant or error if -ve */
1015 struct file_lock_context {
1016 spinlock_t flc_lock;
1017 struct list_head flc_flock;
1018 struct list_head flc_posix;
1019 struct list_head flc_lease;
1022 /* The following constant reflects the upper bound of the file/locking space */
1024 #define INT_LIMIT(x) (~((x)1 << (sizeof(x)*8 - 1)))
1025 #define OFFSET_MAX INT_LIMIT(loff_t)
1026 #define OFFT_OFFSET_MAX INT_LIMIT(off_t)
1029 #include <linux/fcntl.h>
1031 extern void send_sigio(struct fown_struct *fown, int fd, int band);
1034 * Return the inode to use for locking
1036 * For overlayfs this should be the overlay inode, not the real inode returned
1037 * by file_inode(). For any other fs file_inode(filp) and locks_inode(filp) are
1040 static inline struct inode *locks_inode(const struct file *f)
1042 return f->f_path.dentry->d_inode;
1045 #ifdef CONFIG_FILE_LOCKING
1046 extern int fcntl_getlk(struct file *, unsigned int, struct flock __user *);
1047 extern int fcntl_setlk(unsigned int, struct file *, unsigned int,
1048 struct flock __user *);
1050 #if BITS_PER_LONG == 32
1051 extern int fcntl_getlk64(struct file *, unsigned int, struct flock64 __user *);
1052 extern int fcntl_setlk64(unsigned int, struct file *, unsigned int,
1053 struct flock64 __user *);
1056 extern int fcntl_setlease(unsigned int fd, struct file *filp, long arg);
1057 extern int fcntl_getlease(struct file *filp);
1060 void locks_free_lock_context(struct inode *inode);
1061 void locks_free_lock(struct file_lock *fl);
1062 extern void locks_init_lock(struct file_lock *);
1063 extern struct file_lock * locks_alloc_lock(void);
1064 extern void locks_copy_lock(struct file_lock *, struct file_lock *);
1065 extern void locks_copy_conflock(struct file_lock *, struct file_lock *);
1066 extern void locks_remove_posix(struct file *, fl_owner_t);
1067 extern void locks_remove_file(struct file *);
1068 extern void locks_release_private(struct file_lock *);
1069 extern void posix_test_lock(struct file *, struct file_lock *);
1070 extern int posix_lock_file(struct file *, struct file_lock *, struct file_lock *);
1071 extern int posix_unblock_lock(struct file_lock *);
1072 extern int vfs_test_lock(struct file *, struct file_lock *);
1073 extern int vfs_lock_file(struct file *, unsigned int, struct file_lock *, struct file_lock *);
1074 extern int vfs_cancel_lock(struct file *filp, struct file_lock *fl);
1075 extern int locks_lock_inode_wait(struct inode *inode, struct file_lock *fl);
1076 extern int __break_lease(struct inode *inode, unsigned int flags, unsigned int type);
1077 extern void lease_get_mtime(struct inode *, struct timespec *time);
1078 extern int generic_setlease(struct file *, long, struct file_lock **, void **priv);
1079 extern int vfs_setlease(struct file *, long, struct file_lock **, void **);
1080 extern int lease_modify(struct file_lock *, int, struct list_head *);
1081 struct files_struct;
1082 extern void show_fd_locks(struct seq_file *f,
1083 struct file *filp, struct files_struct *files);
1084 #else /* !CONFIG_FILE_LOCKING */
1085 static inline int fcntl_getlk(struct file *file, unsigned int cmd,
1086 struct flock __user *user)
1091 static inline int fcntl_setlk(unsigned int fd, struct file *file,
1092 unsigned int cmd, struct flock __user *user)
1097 #if BITS_PER_LONG == 32
1098 static inline int fcntl_getlk64(struct file *file, unsigned int cmd,
1099 struct flock64 __user *user)
1104 static inline int fcntl_setlk64(unsigned int fd, struct file *file,
1105 unsigned int cmd, struct flock64 __user *user)
1110 static inline int fcntl_setlease(unsigned int fd, struct file *filp, long arg)
1115 static inline int fcntl_getlease(struct file *filp)
1121 locks_free_lock_context(struct inode *inode)
1125 static inline void locks_init_lock(struct file_lock *fl)
1130 static inline void locks_copy_conflock(struct file_lock *new, struct file_lock *fl)
1135 static inline void locks_copy_lock(struct file_lock *new, struct file_lock *fl)
1140 static inline void locks_remove_posix(struct file *filp, fl_owner_t owner)
1145 static inline void locks_remove_file(struct file *filp)
1150 static inline void posix_test_lock(struct file *filp, struct file_lock *fl)
1155 static inline int posix_lock_file(struct file *filp, struct file_lock *fl,
1156 struct file_lock *conflock)
1161 static inline int posix_unblock_lock(struct file_lock *waiter)
1166 static inline int vfs_test_lock(struct file *filp, struct file_lock *fl)
1171 static inline int vfs_lock_file(struct file *filp, unsigned int cmd,
1172 struct file_lock *fl, struct file_lock *conf)
1177 static inline int vfs_cancel_lock(struct file *filp, struct file_lock *fl)
1182 static inline int locks_lock_inode_wait(struct inode *inode, struct file_lock *fl)
1187 static inline int __break_lease(struct inode *inode, unsigned int mode, unsigned int type)
1192 static inline void lease_get_mtime(struct inode *inode, struct timespec *time)
1197 static inline int generic_setlease(struct file *filp, long arg,
1198 struct file_lock **flp, void **priv)
1203 static inline int vfs_setlease(struct file *filp, long arg,
1204 struct file_lock **lease, void **priv)
1209 static inline int lease_modify(struct file_lock *fl, int arg,
1210 struct list_head *dispose)
1215 struct files_struct;
1216 static inline void show_fd_locks(struct seq_file *f,
1217 struct file *filp, struct files_struct *files) {}
1218 #endif /* !CONFIG_FILE_LOCKING */
1220 static inline struct inode *file_inode(const struct file *f)
1225 static inline struct dentry *file_dentry(const struct file *file)
1227 return d_real(file->f_path.dentry, file_inode(file), 0);
1230 static inline int locks_lock_file_wait(struct file *filp, struct file_lock *fl)
1232 return locks_lock_inode_wait(locks_inode(filp), fl);
1235 struct fasync_struct {
1239 struct fasync_struct *fa_next; /* singly linked list */
1240 struct file *fa_file;
1241 struct rcu_head fa_rcu;
1244 #define FASYNC_MAGIC 0x4601
1246 /* SMP safe fasync helpers: */
1247 extern int fasync_helper(int, struct file *, int, struct fasync_struct **);
1248 extern struct fasync_struct *fasync_insert_entry(int, struct file *, struct fasync_struct **, struct fasync_struct *);
1249 extern int fasync_remove_entry(struct file *, struct fasync_struct **);
1250 extern struct fasync_struct *fasync_alloc(void);
1251 extern void fasync_free(struct fasync_struct *);
1253 /* can be called from interrupts */
1254 extern void kill_fasync(struct fasync_struct **, int, int);
1256 extern void __f_setown(struct file *filp, struct pid *, enum pid_type, int force);
1257 extern void f_setown(struct file *filp, unsigned long arg, int force);
1258 extern void f_delown(struct file *filp);
1259 extern pid_t f_getown(struct file *filp);
1260 extern int send_sigurg(struct fown_struct *fown);
1268 #define MNT_FORCE 0x00000001 /* Attempt to forcibily umount */
1269 #define MNT_DETACH 0x00000002 /* Just detach from the tree */
1270 #define MNT_EXPIRE 0x00000004 /* Mark for expiry */
1271 #define UMOUNT_NOFOLLOW 0x00000008 /* Don't follow symlink on umount */
1272 #define UMOUNT_UNUSED 0x80000000 /* Flag guaranteed to be unused */
1275 #define SB_I_CGROUPWB 0x00000001 /* cgroup-aware writeback enabled */
1276 #define SB_I_NOEXEC 0x00000002 /* Ignore executables on this fs */
1277 #define SB_I_NODEV 0x00000004 /* Ignore devices on this fs */
1279 /* sb->s_iflags to limit user namespace mounts */
1280 #define SB_I_USERNS_VISIBLE 0x00000010 /* fstype already mounted */
1282 /* Possible states of 'frozen' field */
1284 SB_UNFROZEN = 0, /* FS is unfrozen */
1285 SB_FREEZE_WRITE = 1, /* Writes, dir ops, ioctls frozen */
1286 SB_FREEZE_PAGEFAULT = 2, /* Page faults stopped as well */
1287 SB_FREEZE_FS = 3, /* For internal FS use (e.g. to stop
1288 * internal threads if needed) */
1289 SB_FREEZE_COMPLETE = 4, /* ->freeze_fs finished successfully */
1292 #define SB_FREEZE_LEVELS (SB_FREEZE_COMPLETE - 1)
1295 int frozen; /* Is sb frozen? */
1296 wait_queue_head_t wait_unfrozen; /* for get_super_thawed() */
1297 struct percpu_rw_semaphore rw_sem[SB_FREEZE_LEVELS];
1300 struct super_block {
1301 struct list_head s_list; /* Keep this first */
1302 dev_t s_dev; /* search index; _not_ kdev_t */
1303 unsigned char s_blocksize_bits;
1304 unsigned long s_blocksize;
1305 loff_t s_maxbytes; /* Max file size */
1306 struct file_system_type *s_type;
1307 const struct super_operations *s_op;
1308 const struct dquot_operations *dq_op;
1309 const struct quotactl_ops *s_qcop;
1310 const struct export_operations *s_export_op;
1311 unsigned long s_flags;
1312 unsigned long s_iflags; /* internal SB_I_* flags */
1313 unsigned long s_magic;
1314 struct dentry *s_root;
1315 struct rw_semaphore s_umount;
1318 #ifdef CONFIG_SECURITY
1321 const struct xattr_handler **s_xattr;
1323 const struct fscrypt_operations *s_cop;
1325 struct hlist_bl_head s_anon; /* anonymous dentries for (nfs) exporting */
1326 struct list_head s_mounts; /* list of mounts; _not_ for fs use */
1327 struct block_device *s_bdev;
1328 struct backing_dev_info *s_bdi;
1329 struct mtd_info *s_mtd;
1330 struct hlist_node s_instances;
1331 unsigned int s_quota_types; /* Bitmask of supported quota types */
1332 struct quota_info s_dquot; /* Diskquota specific options */
1334 struct sb_writers s_writers;
1336 char s_id[32]; /* Informational name */
1337 uuid_t s_uuid; /* UUID */
1339 void *s_fs_info; /* Filesystem private info */
1340 unsigned int s_max_links;
1343 /* Granularity of c/m/atime in ns.
1344 Cannot be worse than a second */
1348 * The next field is for VFS *only*. No filesystems have any business
1349 * even looking at it. You had been warned.
1351 struct mutex s_vfs_rename_mutex; /* Kludge */
1354 * Filesystem subtype. If non-empty the filesystem type field
1355 * in /proc/mounts will be "type.subtype"
1360 * Saved mount options for lazy filesystems using
1361 * generic_show_options()
1363 char __rcu *s_options;
1364 const struct dentry_operations *s_d_op; /* default d_op for dentries */
1367 * Saved pool identifier for cleancache (-1 means none)
1369 int cleancache_poolid;
1371 struct shrinker s_shrink; /* per-sb shrinker handle */
1373 /* Number of inodes with nlink == 0 but still referenced */
1374 atomic_long_t s_remove_count;
1376 /* Being remounted read-only */
1377 int s_readonly_remount;
1379 /* AIO completions deferred from interrupt context */
1380 struct workqueue_struct *s_dio_done_wq;
1381 struct hlist_head s_pins;
1384 * Owning user namespace and default context in which to
1385 * interpret filesystem uids, gids, quotas, device nodes,
1386 * xattrs and security labels.
1388 struct user_namespace *s_user_ns;
1391 * Keep the lru lists last in the structure so they always sit on their
1392 * own individual cachelines.
1394 struct list_lru s_dentry_lru ____cacheline_aligned_in_smp;
1395 struct list_lru s_inode_lru ____cacheline_aligned_in_smp;
1396 struct rcu_head rcu;
1397 struct work_struct destroy_work;
1399 struct mutex s_sync_lock; /* sync serialisation lock */
1402 * Indicates how deep in a filesystem stack this SB is
1406 /* s_inode_list_lock protects s_inodes */
1407 spinlock_t s_inode_list_lock ____cacheline_aligned_in_smp;
1408 struct list_head s_inodes; /* all inodes */
1410 spinlock_t s_inode_wblist_lock;
1411 struct list_head s_inodes_wb; /* writeback inodes */
1414 /* Helper functions so that in most cases filesystems will
1415 * not need to deal directly with kuid_t and kgid_t and can
1416 * instead deal with the raw numeric values that are stored
1417 * in the filesystem.
1419 static inline uid_t i_uid_read(const struct inode *inode)
1421 return from_kuid(inode->i_sb->s_user_ns, inode->i_uid);
1424 static inline gid_t i_gid_read(const struct inode *inode)
1426 return from_kgid(inode->i_sb->s_user_ns, inode->i_gid);
1429 static inline void i_uid_write(struct inode *inode, uid_t uid)
1431 inode->i_uid = make_kuid(inode->i_sb->s_user_ns, uid);
1434 static inline void i_gid_write(struct inode *inode, gid_t gid)
1436 inode->i_gid = make_kgid(inode->i_sb->s_user_ns, gid);
1439 extern struct timespec current_time(struct inode *inode);
1442 * Snapshotting support.
1445 void __sb_end_write(struct super_block *sb, int level);
1446 int __sb_start_write(struct super_block *sb, int level, bool wait);
1448 #define __sb_writers_acquired(sb, lev) \
1449 percpu_rwsem_acquire(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1450 #define __sb_writers_release(sb, lev) \
1451 percpu_rwsem_release(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1454 * sb_end_write - drop write access to a superblock
1455 * @sb: the super we wrote to
1457 * Decrement number of writers to the filesystem. Wake up possible waiters
1458 * wanting to freeze the filesystem.
1460 static inline void sb_end_write(struct super_block *sb)
1462 __sb_end_write(sb, SB_FREEZE_WRITE);
1466 * sb_end_pagefault - drop write access to a superblock from a page fault
1467 * @sb: the super we wrote to
1469 * Decrement number of processes handling write page fault to the filesystem.
1470 * Wake up possible waiters wanting to freeze the filesystem.
1472 static inline void sb_end_pagefault(struct super_block *sb)
1474 __sb_end_write(sb, SB_FREEZE_PAGEFAULT);
1478 * sb_end_intwrite - drop write access to a superblock for internal fs purposes
1479 * @sb: the super we wrote to
1481 * Decrement fs-internal number of writers to the filesystem. Wake up possible
1482 * waiters wanting to freeze the filesystem.
1484 static inline void sb_end_intwrite(struct super_block *sb)
1486 __sb_end_write(sb, SB_FREEZE_FS);
1490 * sb_start_write - get write access to a superblock
1491 * @sb: the super we write to
1493 * When a process wants to write data or metadata to a file system (i.e. dirty
1494 * a page or an inode), it should embed the operation in a sb_start_write() -
1495 * sb_end_write() pair to get exclusion against file system freezing. This
1496 * function increments number of writers preventing freezing. If the file
1497 * system is already frozen, the function waits until the file system is
1500 * Since freeze protection behaves as a lock, users have to preserve
1501 * ordering of freeze protection and other filesystem locks. Generally,
1502 * freeze protection should be the outermost lock. In particular, we have:
1505 * -> i_mutex (write path, truncate, directory ops, ...)
1506 * -> s_umount (freeze_super, thaw_super)
1508 static inline void sb_start_write(struct super_block *sb)
1510 __sb_start_write(sb, SB_FREEZE_WRITE, true);
1513 static inline int sb_start_write_trylock(struct super_block *sb)
1515 return __sb_start_write(sb, SB_FREEZE_WRITE, false);
1519 * sb_start_pagefault - get write access to a superblock from a page fault
1520 * @sb: the super we write to
1522 * When a process starts handling write page fault, it should embed the
1523 * operation into sb_start_pagefault() - sb_end_pagefault() pair to get
1524 * exclusion against file system freezing. This is needed since the page fault
1525 * is going to dirty a page. This function increments number of running page
1526 * faults preventing freezing. If the file system is already frozen, the
1527 * function waits until the file system is thawed.
1529 * Since page fault freeze protection behaves as a lock, users have to preserve
1530 * ordering of freeze protection and other filesystem locks. It is advised to
1531 * put sb_start_pagefault() close to mmap_sem in lock ordering. Page fault
1532 * handling code implies lock dependency:
1535 * -> sb_start_pagefault
1537 static inline void sb_start_pagefault(struct super_block *sb)
1539 __sb_start_write(sb, SB_FREEZE_PAGEFAULT, true);
1543 * sb_start_intwrite - get write access to a superblock for internal fs purposes
1544 * @sb: the super we write to
1546 * This is the third level of protection against filesystem freezing. It is
1547 * free for use by a filesystem. The only requirement is that it must rank
1548 * below sb_start_pagefault.
1550 * For example filesystem can call sb_start_intwrite() when starting a
1551 * transaction which somewhat eases handling of freezing for internal sources
1552 * of filesystem changes (internal fs threads, discarding preallocation on file
1555 static inline void sb_start_intwrite(struct super_block *sb)
1557 __sb_start_write(sb, SB_FREEZE_FS, true);
1561 extern bool inode_owner_or_capable(const struct inode *inode);
1564 * VFS helper functions..
1566 extern int vfs_create(struct inode *, struct dentry *, umode_t, bool);
1567 extern int vfs_mkdir(struct inode *, struct dentry *, umode_t);
1568 extern int vfs_mknod(struct inode *, struct dentry *, umode_t, dev_t);
1569 extern int vfs_symlink(struct inode *, struct dentry *, const char *);
1570 extern int vfs_link(struct dentry *, struct inode *, struct dentry *, struct inode **);
1571 extern int vfs_rmdir(struct inode *, struct dentry *);
1572 extern int vfs_unlink(struct inode *, struct dentry *, struct inode **);
1573 extern int vfs_rename(struct inode *, struct dentry *, struct inode *, struct dentry *, struct inode **, unsigned int);
1574 extern int vfs_whiteout(struct inode *, struct dentry *);
1576 extern struct dentry *vfs_tmpfile(struct dentry *dentry, umode_t mode,
1580 * VFS file helper functions.
1582 extern void inode_init_owner(struct inode *inode, const struct inode *dir,
1584 extern bool may_open_dev(const struct path *path);
1586 * VFS FS_IOC_FIEMAP helper definitions.
1588 struct fiemap_extent_info {
1589 unsigned int fi_flags; /* Flags as passed from user */
1590 unsigned int fi_extents_mapped; /* Number of mapped extents */
1591 unsigned int fi_extents_max; /* Size of fiemap_extent array */
1592 struct fiemap_extent __user *fi_extents_start; /* Start of
1593 fiemap_extent array */
1595 int fiemap_fill_next_extent(struct fiemap_extent_info *info, u64 logical,
1596 u64 phys, u64 len, u32 flags);
1597 int fiemap_check_flags(struct fiemap_extent_info *fieinfo, u32 fs_flags);
1602 * NOTE! These match bits 12..15 of stat.st_mode
1603 * (ie "(i_mode >> 12) & 15").
1605 #define DT_UNKNOWN 0
1616 * This is the "filldir" function type, used by readdir() to let
1617 * the kernel specify what kind of dirent layout it wants to have.
1618 * This allows the kernel to read directories into kernel space or
1619 * to have different dirent layouts depending on the binary type.
1622 typedef int (*filldir_t)(struct dir_context *, const char *, int, loff_t, u64,
1625 struct dir_context {
1626 const filldir_t actor;
1630 struct block_device_operations;
1632 /* These macros are for out of kernel modules to test that
1633 * the kernel supports the unlocked_ioctl and compat_ioctl
1634 * fields in struct file_operations. */
1635 #define HAVE_COMPAT_IOCTL 1
1636 #define HAVE_UNLOCKED_IOCTL 1
1639 * These flags let !MMU mmap() govern direct device mapping vs immediate
1640 * copying more easily for MAP_PRIVATE, especially for ROM filesystems.
1642 * NOMMU_MAP_COPY: Copy can be mapped (MAP_PRIVATE)
1643 * NOMMU_MAP_DIRECT: Can be mapped directly (MAP_SHARED)
1644 * NOMMU_MAP_READ: Can be mapped for reading
1645 * NOMMU_MAP_WRITE: Can be mapped for writing
1646 * NOMMU_MAP_EXEC: Can be mapped for execution
1648 #define NOMMU_MAP_COPY 0x00000001
1649 #define NOMMU_MAP_DIRECT 0x00000008
1650 #define NOMMU_MAP_READ VM_MAYREAD
1651 #define NOMMU_MAP_WRITE VM_MAYWRITE
1652 #define NOMMU_MAP_EXEC VM_MAYEXEC
1654 #define NOMMU_VMFLAGS \
1655 (NOMMU_MAP_READ | NOMMU_MAP_WRITE | NOMMU_MAP_EXEC)
1660 struct file_operations {
1661 struct module *owner;
1662 loff_t (*llseek) (struct file *, loff_t, int);
1663 ssize_t (*read) (struct file *, char __user *, size_t, loff_t *);
1664 ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *);
1665 ssize_t (*read_iter) (struct kiocb *, struct iov_iter *);
1666 ssize_t (*write_iter) (struct kiocb *, struct iov_iter *);
1667 int (*iterate) (struct file *, struct dir_context *);
1668 int (*iterate_shared) (struct file *, struct dir_context *);
1669 unsigned int (*poll) (struct file *, struct poll_table_struct *);
1670 long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long);
1671 long (*compat_ioctl) (struct file *, unsigned int, unsigned long);
1672 int (*mmap) (struct file *, struct vm_area_struct *);
1673 int (*open) (struct inode *, struct file *);
1674 int (*flush) (struct file *, fl_owner_t id);
1675 int (*release) (struct inode *, struct file *);
1676 int (*fsync) (struct file *, loff_t, loff_t, int datasync);
1677 int (*fasync) (int, struct file *, int);
1678 int (*lock) (struct file *, int, struct file_lock *);
1679 ssize_t (*sendpage) (struct file *, struct page *, int, size_t, loff_t *, int);
1680 unsigned long (*get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
1681 int (*check_flags)(int);
1682 int (*flock) (struct file *, int, struct file_lock *);
1683 ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, loff_t *, size_t, unsigned int);
1684 ssize_t (*splice_read)(struct file *, loff_t *, struct pipe_inode_info *, size_t, unsigned int);
1685 int (*setlease)(struct file *, long, struct file_lock **, void **);
1686 long (*fallocate)(struct file *file, int mode, loff_t offset,
1688 void (*show_fdinfo)(struct seq_file *m, struct file *f);
1690 unsigned (*mmap_capabilities)(struct file *);
1692 ssize_t (*copy_file_range)(struct file *, loff_t, struct file *,
1693 loff_t, size_t, unsigned int);
1694 int (*clone_file_range)(struct file *, loff_t, struct file *, loff_t,
1696 ssize_t (*dedupe_file_range)(struct file *, u64, u64, struct file *,
1700 struct inode_operations {
1701 struct dentry * (*lookup) (struct inode *,struct dentry *, unsigned int);
1702 const char * (*get_link) (struct dentry *, struct inode *, struct delayed_call *);
1703 int (*permission) (struct inode *, int);
1704 struct posix_acl * (*get_acl)(struct inode *, int);
1706 int (*readlink) (struct dentry *, char __user *,int);
1708 int (*create) (struct inode *,struct dentry *, umode_t, bool);
1709 int (*link) (struct dentry *,struct inode *,struct dentry *);
1710 int (*unlink) (struct inode *,struct dentry *);
1711 int (*symlink) (struct inode *,struct dentry *,const char *);
1712 int (*mkdir) (struct inode *,struct dentry *,umode_t);
1713 int (*rmdir) (struct inode *,struct dentry *);
1714 int (*mknod) (struct inode *,struct dentry *,umode_t,dev_t);
1715 int (*rename) (struct inode *, struct dentry *,
1716 struct inode *, struct dentry *, unsigned int);
1717 int (*setattr) (struct dentry *, struct iattr *);
1718 int (*getattr) (const struct path *, struct kstat *, u32, unsigned int);
1719 ssize_t (*listxattr) (struct dentry *, char *, size_t);
1720 int (*fiemap)(struct inode *, struct fiemap_extent_info *, u64 start,
1722 int (*update_time)(struct inode *, struct timespec *, int);
1723 int (*atomic_open)(struct inode *, struct dentry *,
1724 struct file *, unsigned open_flag,
1725 umode_t create_mode, int *opened);
1726 int (*tmpfile) (struct inode *, struct dentry *, umode_t);
1727 int (*set_acl)(struct inode *, struct posix_acl *, int);
1728 } ____cacheline_aligned;
1730 static inline ssize_t call_read_iter(struct file *file, struct kiocb *kio,
1731 struct iov_iter *iter)
1733 return file->f_op->read_iter(kio, iter);
1736 static inline ssize_t call_write_iter(struct file *file, struct kiocb *kio,
1737 struct iov_iter *iter)
1739 return file->f_op->write_iter(kio, iter);
1742 static inline int call_mmap(struct file *file, struct vm_area_struct *vma)
1744 return file->f_op->mmap(file, vma);
1747 static inline int call_fsync(struct file *file, loff_t start, loff_t end,
1750 return file->f_op->fsync(file, start, end, datasync);
1753 ssize_t rw_copy_check_uvector(int type, const struct iovec __user * uvector,
1754 unsigned long nr_segs, unsigned long fast_segs,
1755 struct iovec *fast_pointer,
1756 struct iovec **ret_pointer);
1758 extern ssize_t __vfs_read(struct file *, char __user *, size_t, loff_t *);
1759 extern ssize_t __vfs_write(struct file *, const char __user *, size_t, loff_t *);
1760 extern ssize_t vfs_read(struct file *, char __user *, size_t, loff_t *);
1761 extern ssize_t vfs_write(struct file *, const char __user *, size_t, loff_t *);
1762 extern ssize_t vfs_readv(struct file *, const struct iovec __user *,
1763 unsigned long, loff_t *, int);
1764 extern ssize_t vfs_writev(struct file *, const struct iovec __user *,
1765 unsigned long, loff_t *, int);
1766 extern ssize_t vfs_copy_file_range(struct file *, loff_t , struct file *,
1767 loff_t, size_t, unsigned int);
1768 extern int vfs_clone_file_prep_inodes(struct inode *inode_in, loff_t pos_in,
1769 struct inode *inode_out, loff_t pos_out,
1770 u64 *len, bool is_dedupe);
1771 extern int vfs_clone_file_range(struct file *file_in, loff_t pos_in,
1772 struct file *file_out, loff_t pos_out, u64 len);
1773 extern int vfs_dedupe_file_range_compare(struct inode *src, loff_t srcoff,
1774 struct inode *dest, loff_t destoff,
1775 loff_t len, bool *is_same);
1776 extern int vfs_dedupe_file_range(struct file *file,
1777 struct file_dedupe_range *same);
1779 struct super_operations {
1780 struct inode *(*alloc_inode)(struct super_block *sb);
1781 void (*destroy_inode)(struct inode *);
1783 void (*dirty_inode) (struct inode *, int flags);
1784 int (*write_inode) (struct inode *, struct writeback_control *wbc);
1785 int (*drop_inode) (struct inode *);
1786 void (*evict_inode) (struct inode *);
1787 void (*put_super) (struct super_block *);
1788 int (*sync_fs)(struct super_block *sb, int wait);
1789 int (*freeze_super) (struct super_block *);
1790 int (*freeze_fs) (struct super_block *);
1791 int (*thaw_super) (struct super_block *);
1792 int (*unfreeze_fs) (struct super_block *);
1793 int (*statfs) (struct dentry *, struct kstatfs *);
1794 int (*remount_fs) (struct super_block *, int *, char *);
1795 void (*umount_begin) (struct super_block *);
1797 int (*show_options)(struct seq_file *, struct dentry *);
1798 int (*show_devname)(struct seq_file *, struct dentry *);
1799 int (*show_path)(struct seq_file *, struct dentry *);
1800 int (*show_stats)(struct seq_file *, struct dentry *);
1802 ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t);
1803 ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t);
1804 struct dquot **(*get_dquots)(struct inode *);
1806 int (*bdev_try_to_free_page)(struct super_block*, struct page*, gfp_t);
1807 long (*nr_cached_objects)(struct super_block *,
1808 struct shrink_control *);
1809 long (*free_cached_objects)(struct super_block *,
1810 struct shrink_control *);
1814 * Inode flags - they have no relation to superblock flags now
1816 #define S_SYNC 1 /* Writes are synced at once */
1817 #define S_NOATIME 2 /* Do not update access times */
1818 #define S_APPEND 4 /* Append-only file */
1819 #define S_IMMUTABLE 8 /* Immutable file */
1820 #define S_DEAD 16 /* removed, but still open directory */
1821 #define S_NOQUOTA 32 /* Inode is not counted to quota */
1822 #define S_DIRSYNC 64 /* Directory modifications are synchronous */
1823 #define S_NOCMTIME 128 /* Do not update file c/mtime */
1824 #define S_SWAPFILE 256 /* Do not truncate: swapon got its bmaps */
1825 #define S_PRIVATE 512 /* Inode is fs-internal */
1826 #define S_IMA 1024 /* Inode has an associated IMA struct */
1827 #define S_AUTOMOUNT 2048 /* Automount/referral quasi-directory */
1828 #define S_NOSEC 4096 /* no suid or xattr security attributes */
1829 #ifdef CONFIG_FS_DAX
1830 #define S_DAX 8192 /* Direct Access, avoiding the page cache */
1832 #define S_DAX 0 /* Make all the DAX code disappear */
1836 * Note that nosuid etc flags are inode-specific: setting some file-system
1837 * flags just means all the inodes inherit those flags by default. It might be
1838 * possible to override it selectively if you really wanted to with some
1839 * ioctl() that is not currently implemented.
1841 * Exception: MS_RDONLY is always applied to the entire file system.
1843 * Unfortunately, it is possible to change a filesystems flags with it mounted
1844 * with files in use. This means that all of the inodes will not have their
1845 * i_flags updated. Hence, i_flags no longer inherit the superblock mount
1846 * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org
1848 #define __IS_FLG(inode, flg) ((inode)->i_sb->s_flags & (flg))
1850 #define IS_RDONLY(inode) ((inode)->i_sb->s_flags & MS_RDONLY)
1851 #define IS_SYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS) || \
1852 ((inode)->i_flags & S_SYNC))
1853 #define IS_DIRSYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS|MS_DIRSYNC) || \
1854 ((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
1855 #define IS_MANDLOCK(inode) __IS_FLG(inode, MS_MANDLOCK)
1856 #define IS_NOATIME(inode) __IS_FLG(inode, MS_RDONLY|MS_NOATIME)
1857 #define IS_I_VERSION(inode) __IS_FLG(inode, MS_I_VERSION)
1859 #define IS_NOQUOTA(inode) ((inode)->i_flags & S_NOQUOTA)
1860 #define IS_APPEND(inode) ((inode)->i_flags & S_APPEND)
1861 #define IS_IMMUTABLE(inode) ((inode)->i_flags & S_IMMUTABLE)
1862 #define IS_POSIXACL(inode) __IS_FLG(inode, MS_POSIXACL)
1864 #define IS_DEADDIR(inode) ((inode)->i_flags & S_DEAD)
1865 #define IS_NOCMTIME(inode) ((inode)->i_flags & S_NOCMTIME)
1866 #define IS_SWAPFILE(inode) ((inode)->i_flags & S_SWAPFILE)
1867 #define IS_PRIVATE(inode) ((inode)->i_flags & S_PRIVATE)
1868 #define IS_IMA(inode) ((inode)->i_flags & S_IMA)
1869 #define IS_AUTOMOUNT(inode) ((inode)->i_flags & S_AUTOMOUNT)
1870 #define IS_NOSEC(inode) ((inode)->i_flags & S_NOSEC)
1871 #define IS_DAX(inode) ((inode)->i_flags & S_DAX)
1873 #define IS_WHITEOUT(inode) (S_ISCHR(inode->i_mode) && \
1874 (inode)->i_rdev == WHITEOUT_DEV)
1876 static inline bool HAS_UNMAPPED_ID(struct inode *inode)
1878 return !uid_valid(inode->i_uid) || !gid_valid(inode->i_gid);
1882 * Inode state bits. Protected by inode->i_lock
1884 * Three bits determine the dirty state of the inode, I_DIRTY_SYNC,
1885 * I_DIRTY_DATASYNC and I_DIRTY_PAGES.
1887 * Four bits define the lifetime of an inode. Initially, inodes are I_NEW,
1888 * until that flag is cleared. I_WILL_FREE, I_FREEING and I_CLEAR are set at
1889 * various stages of removing an inode.
1891 * Two bits are used for locking and completion notification, I_NEW and I_SYNC.
1893 * I_DIRTY_SYNC Inode is dirty, but doesn't have to be written on
1894 * fdatasync(). i_atime is the usual cause.
1895 * I_DIRTY_DATASYNC Data-related inode changes pending. We keep track of
1896 * these changes separately from I_DIRTY_SYNC so that we
1897 * don't have to write inode on fdatasync() when only
1898 * mtime has changed in it.
1899 * I_DIRTY_PAGES Inode has dirty pages. Inode itself may be clean.
1900 * I_NEW Serves as both a mutex and completion notification.
1901 * New inodes set I_NEW. If two processes both create
1902 * the same inode, one of them will release its inode and
1903 * wait for I_NEW to be released before returning.
1904 * Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can
1905 * also cause waiting on I_NEW, without I_NEW actually
1906 * being set. find_inode() uses this to prevent returning
1907 * nearly-dead inodes.
1908 * I_WILL_FREE Must be set when calling write_inode_now() if i_count
1909 * is zero. I_FREEING must be set when I_WILL_FREE is
1911 * I_FREEING Set when inode is about to be freed but still has dirty
1912 * pages or buffers attached or the inode itself is still
1914 * I_CLEAR Added by clear_inode(). In this state the inode is
1915 * clean and can be destroyed. Inode keeps I_FREEING.
1917 * Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are
1918 * prohibited for many purposes. iget() must wait for
1919 * the inode to be completely released, then create it
1920 * anew. Other functions will just ignore such inodes,
1921 * if appropriate. I_NEW is used for waiting.
1923 * I_SYNC Writeback of inode is running. The bit is set during
1924 * data writeback, and cleared with a wakeup on the bit
1925 * address once it is done. The bit is also used to pin
1926 * the inode in memory for flusher thread.
1928 * I_REFERENCED Marks the inode as recently references on the LRU list.
1930 * I_DIO_WAKEUP Never set. Only used as a key for wait_on_bit().
1932 * I_WB_SWITCH Cgroup bdi_writeback switching in progress. Used to
1933 * synchronize competing switching instances and to tell
1934 * wb stat updates to grab mapping->tree_lock. See
1935 * inode_switch_wb_work_fn() for details.
1937 * Q: What is the difference between I_WILL_FREE and I_FREEING?
1939 #define I_DIRTY_SYNC (1 << 0)
1940 #define I_DIRTY_DATASYNC (1 << 1)
1941 #define I_DIRTY_PAGES (1 << 2)
1943 #define I_NEW (1 << __I_NEW)
1944 #define I_WILL_FREE (1 << 4)
1945 #define I_FREEING (1 << 5)
1946 #define I_CLEAR (1 << 6)
1948 #define I_SYNC (1 << __I_SYNC)
1949 #define I_REFERENCED (1 << 8)
1950 #define __I_DIO_WAKEUP 9
1951 #define I_DIO_WAKEUP (1 << __I_DIO_WAKEUP)
1952 #define I_LINKABLE (1 << 10)
1953 #define I_DIRTY_TIME (1 << 11)
1954 #define __I_DIRTY_TIME_EXPIRED 12
1955 #define I_DIRTY_TIME_EXPIRED (1 << __I_DIRTY_TIME_EXPIRED)
1956 #define I_WB_SWITCH (1 << 13)
1958 #define I_DIRTY (I_DIRTY_SYNC | I_DIRTY_DATASYNC | I_DIRTY_PAGES)
1959 #define I_DIRTY_ALL (I_DIRTY | I_DIRTY_TIME)
1961 extern void __mark_inode_dirty(struct inode *, int);
1962 static inline void mark_inode_dirty(struct inode *inode)
1964 __mark_inode_dirty(inode, I_DIRTY);
1967 static inline void mark_inode_dirty_sync(struct inode *inode)
1969 __mark_inode_dirty(inode, I_DIRTY_SYNC);
1972 extern void inc_nlink(struct inode *inode);
1973 extern void drop_nlink(struct inode *inode);
1974 extern void clear_nlink(struct inode *inode);
1975 extern void set_nlink(struct inode *inode, unsigned int nlink);
1977 static inline void inode_inc_link_count(struct inode *inode)
1980 mark_inode_dirty(inode);
1983 static inline void inode_dec_link_count(struct inode *inode)
1986 mark_inode_dirty(inode);
1990 * inode_inc_iversion - increments i_version
1991 * @inode: inode that need to be updated
1993 * Every time the inode is modified, the i_version field will be incremented.
1994 * The filesystem has to be mounted with i_version flag
1997 static inline void inode_inc_iversion(struct inode *inode)
1999 spin_lock(&inode->i_lock);
2001 spin_unlock(&inode->i_lock);
2004 enum file_time_flags {
2011 extern void touch_atime(const struct path *);
2012 static inline void file_accessed(struct file *file)
2014 if (!(file->f_flags & O_NOATIME))
2015 touch_atime(&file->f_path);
2018 int sync_inode(struct inode *inode, struct writeback_control *wbc);
2019 int sync_inode_metadata(struct inode *inode, int wait);
2021 struct file_system_type {
2024 #define FS_REQUIRES_DEV 1
2025 #define FS_BINARY_MOUNTDATA 2
2026 #define FS_HAS_SUBTYPE 4
2027 #define FS_USERNS_MOUNT 8 /* Can be mounted by userns root */
2028 #define FS_RENAME_DOES_D_MOVE 32768 /* FS will handle d_move() during rename() internally. */
2029 struct dentry *(*mount) (struct file_system_type *, int,
2030 const char *, void *);
2031 void (*kill_sb) (struct super_block *);
2032 struct module *owner;
2033 struct file_system_type * next;
2034 struct hlist_head fs_supers;
2036 struct lock_class_key s_lock_key;
2037 struct lock_class_key s_umount_key;
2038 struct lock_class_key s_vfs_rename_key;
2039 struct lock_class_key s_writers_key[SB_FREEZE_LEVELS];
2041 struct lock_class_key i_lock_key;
2042 struct lock_class_key i_mutex_key;
2043 struct lock_class_key i_mutex_dir_key;
2046 #define MODULE_ALIAS_FS(NAME) MODULE_ALIAS("fs-" NAME)
2048 extern struct dentry *mount_ns(struct file_system_type *fs_type,
2049 int flags, void *data, void *ns, struct user_namespace *user_ns,
2050 int (*fill_super)(struct super_block *, void *, int));
2051 extern struct dentry *mount_bdev(struct file_system_type *fs_type,
2052 int flags, const char *dev_name, void *data,
2053 int (*fill_super)(struct super_block *, void *, int));
2054 extern struct dentry *mount_single(struct file_system_type *fs_type,
2055 int flags, void *data,
2056 int (*fill_super)(struct super_block *, void *, int));
2057 extern struct dentry *mount_nodev(struct file_system_type *fs_type,
2058 int flags, void *data,
2059 int (*fill_super)(struct super_block *, void *, int));
2060 extern struct dentry *mount_subtree(struct vfsmount *mnt, const char *path);
2061 void generic_shutdown_super(struct super_block *sb);
2062 void kill_block_super(struct super_block *sb);
2063 void kill_anon_super(struct super_block *sb);
2064 void kill_litter_super(struct super_block *sb);
2065 void deactivate_super(struct super_block *sb);
2066 void deactivate_locked_super(struct super_block *sb);
2067 int set_anon_super(struct super_block *s, void *data);
2068 int get_anon_bdev(dev_t *);
2069 void free_anon_bdev(dev_t);
2070 struct super_block *sget_userns(struct file_system_type *type,
2071 int (*test)(struct super_block *,void *),
2072 int (*set)(struct super_block *,void *),
2073 int flags, struct user_namespace *user_ns,
2075 struct super_block *sget(struct file_system_type *type,
2076 int (*test)(struct super_block *,void *),
2077 int (*set)(struct super_block *,void *),
2078 int flags, void *data);
2079 extern struct dentry *mount_pseudo_xattr(struct file_system_type *, char *,
2080 const struct super_operations *ops,
2081 const struct xattr_handler **xattr,
2082 const struct dentry_operations *dops,
2085 static inline struct dentry *
2086 mount_pseudo(struct file_system_type *fs_type, char *name,
2087 const struct super_operations *ops,
2088 const struct dentry_operations *dops, unsigned long magic)
2090 return mount_pseudo_xattr(fs_type, name, ops, NULL, dops, magic);
2093 /* Alas, no aliases. Too much hassle with bringing module.h everywhere */
2094 #define fops_get(fops) \
2095 (((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
2096 #define fops_put(fops) \
2097 do { if (fops) module_put((fops)->owner); } while(0)
2099 * This one is to be used *ONLY* from ->open() instances.
2100 * fops must be non-NULL, pinned down *and* module dependencies
2101 * should be sufficient to pin the caller down as well.
2103 #define replace_fops(f, fops) \
2105 struct file *__file = (f); \
2106 fops_put(__file->f_op); \
2107 BUG_ON(!(__file->f_op = (fops))); \
2110 extern int register_filesystem(struct file_system_type *);
2111 extern int unregister_filesystem(struct file_system_type *);
2112 extern struct vfsmount *kern_mount_data(struct file_system_type *, void *data);
2113 #define kern_mount(type) kern_mount_data(type, NULL)
2114 extern void kern_unmount(struct vfsmount *mnt);
2115 extern int may_umount_tree(struct vfsmount *);
2116 extern int may_umount(struct vfsmount *);
2117 extern long do_mount(const char *, const char __user *,
2118 const char *, unsigned long, void *);
2119 extern struct vfsmount *collect_mounts(const struct path *);
2120 extern void drop_collected_mounts(struct vfsmount *);
2121 extern int iterate_mounts(int (*)(struct vfsmount *, void *), void *,
2123 extern int vfs_statfs(const struct path *, struct kstatfs *);
2124 extern int user_statfs(const char __user *, struct kstatfs *);
2125 extern int fd_statfs(int, struct kstatfs *);
2126 extern int vfs_ustat(dev_t, struct kstatfs *);
2127 extern int freeze_super(struct super_block *super);
2128 extern int thaw_super(struct super_block *super);
2129 extern bool our_mnt(struct vfsmount *mnt);
2130 extern __printf(2, 3)
2131 int super_setup_bdi_name(struct super_block *sb, char *fmt, ...);
2132 extern int super_setup_bdi(struct super_block *sb);
2134 extern int current_umask(void);
2136 extern void ihold(struct inode * inode);
2137 extern void iput(struct inode *);
2138 extern int generic_update_time(struct inode *, struct timespec *, int);
2141 extern struct kobject *fs_kobj;
2143 #define MAX_RW_COUNT (INT_MAX & PAGE_MASK)
2145 #ifdef CONFIG_MANDATORY_FILE_LOCKING
2146 extern int locks_mandatory_locked(struct file *);
2147 extern int locks_mandatory_area(struct inode *, struct file *, loff_t, loff_t, unsigned char);
2150 * Candidates for mandatory locking have the setgid bit set
2151 * but no group execute bit - an otherwise meaningless combination.
2154 static inline int __mandatory_lock(struct inode *ino)
2156 return (ino->i_mode & (S_ISGID | S_IXGRP)) == S_ISGID;
2160 * ... and these candidates should be on MS_MANDLOCK mounted fs,
2161 * otherwise these will be advisory locks
2164 static inline int mandatory_lock(struct inode *ino)
2166 return IS_MANDLOCK(ino) && __mandatory_lock(ino);
2169 static inline int locks_verify_locked(struct file *file)
2171 if (mandatory_lock(locks_inode(file)))
2172 return locks_mandatory_locked(file);
2176 static inline int locks_verify_truncate(struct inode *inode,
2180 if (!inode->i_flctx || !mandatory_lock(inode))
2183 if (size < inode->i_size) {
2184 return locks_mandatory_area(inode, f, size, inode->i_size - 1,
2187 return locks_mandatory_area(inode, f, inode->i_size, size - 1,
2192 #else /* !CONFIG_MANDATORY_FILE_LOCKING */
2194 static inline int locks_mandatory_locked(struct file *file)
2199 static inline int locks_mandatory_area(struct inode *inode, struct file *filp,
2200 loff_t start, loff_t end, unsigned char type)
2205 static inline int __mandatory_lock(struct inode *inode)
2210 static inline int mandatory_lock(struct inode *inode)
2215 static inline int locks_verify_locked(struct file *file)
2220 static inline int locks_verify_truncate(struct inode *inode, struct file *filp,
2226 #endif /* CONFIG_MANDATORY_FILE_LOCKING */
2229 #ifdef CONFIG_FILE_LOCKING
2230 static inline int break_lease(struct inode *inode, unsigned int mode)
2233 * Since this check is lockless, we must ensure that any refcounts
2234 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2235 * could end up racing with tasks trying to set a new lease on this
2239 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2240 return __break_lease(inode, mode, FL_LEASE);
2244 static inline int break_deleg(struct inode *inode, unsigned int mode)
2247 * Since this check is lockless, we must ensure that any refcounts
2248 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2249 * could end up racing with tasks trying to set a new lease on this
2253 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2254 return __break_lease(inode, mode, FL_DELEG);
2258 static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode)
2262 ret = break_deleg(inode, O_WRONLY|O_NONBLOCK);
2263 if (ret == -EWOULDBLOCK && delegated_inode) {
2264 *delegated_inode = inode;
2270 static inline int break_deleg_wait(struct inode **delegated_inode)
2274 ret = break_deleg(*delegated_inode, O_WRONLY);
2275 iput(*delegated_inode);
2276 *delegated_inode = NULL;
2280 static inline int break_layout(struct inode *inode, bool wait)
2283 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2284 return __break_lease(inode,
2285 wait ? O_WRONLY : O_WRONLY | O_NONBLOCK,
2290 #else /* !CONFIG_FILE_LOCKING */
2291 static inline int break_lease(struct inode *inode, unsigned int mode)
2296 static inline int break_deleg(struct inode *inode, unsigned int mode)
2301 static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode)
2306 static inline int break_deleg_wait(struct inode **delegated_inode)
2312 static inline int break_layout(struct inode *inode, bool wait)
2317 #endif /* CONFIG_FILE_LOCKING */
2322 const char *name; /* pointer to actual string */
2323 const __user char *uptr; /* original userland pointer */
2324 struct audit_names *aname;
2329 extern long vfs_truncate(const struct path *, loff_t);
2330 extern int do_truncate(struct dentry *, loff_t start, unsigned int time_attrs,
2332 extern int vfs_fallocate(struct file *file, int mode, loff_t offset,
2334 extern long do_sys_open(int dfd, const char __user *filename, int flags,
2336 extern struct file *file_open_name(struct filename *, int, umode_t);
2337 extern struct file *filp_open(const char *, int, umode_t);
2338 extern struct file *file_open_root(struct dentry *, struct vfsmount *,
2339 const char *, int, umode_t);
2340 extern struct file * dentry_open(const struct path *, int, const struct cred *);
2341 extern int filp_close(struct file *, fl_owner_t id);
2343 extern struct filename *getname_flags(const char __user *, int, int *);
2344 extern struct filename *getname(const char __user *);
2345 extern struct filename *getname_kernel(const char *);
2346 extern void putname(struct filename *name);
2352 extern int finish_open(struct file *file, struct dentry *dentry,
2353 int (*open)(struct inode *, struct file *),
2355 extern int finish_no_open(struct file *file, struct dentry *dentry);
2359 extern int ioctl_preallocate(struct file *filp, void __user *argp);
2362 extern void __init vfs_caches_init_early(void);
2363 extern void __init vfs_caches_init(void);
2365 extern struct kmem_cache *names_cachep;
2367 #define __getname() kmem_cache_alloc(names_cachep, GFP_KERNEL)
2368 #define __putname(name) kmem_cache_free(names_cachep, (void *)(name))
2371 extern int register_blkdev(unsigned int, const char *);
2372 extern void unregister_blkdev(unsigned int, const char *);
2373 extern void bdev_unhash_inode(dev_t dev);
2374 extern struct block_device *bdget(dev_t);
2375 extern struct block_device *bdgrab(struct block_device *bdev);
2376 extern void bd_set_size(struct block_device *, loff_t size);
2377 extern void bd_forget(struct inode *inode);
2378 extern void bdput(struct block_device *);
2379 extern void invalidate_bdev(struct block_device *);
2380 extern void iterate_bdevs(void (*)(struct block_device *, void *), void *);
2381 extern int sync_blockdev(struct block_device *bdev);
2382 extern void kill_bdev(struct block_device *);
2383 extern struct super_block *freeze_bdev(struct block_device *);
2384 extern void emergency_thaw_all(void);
2385 extern int thaw_bdev(struct block_device *bdev, struct super_block *sb);
2386 extern int fsync_bdev(struct block_device *);
2388 extern struct super_block *blockdev_superblock;
2390 static inline bool sb_is_blkdev_sb(struct super_block *sb)
2392 return sb == blockdev_superblock;
2395 static inline void bd_forget(struct inode *inode) {}
2396 static inline int sync_blockdev(struct block_device *bdev) { return 0; }
2397 static inline void kill_bdev(struct block_device *bdev) {}
2398 static inline void invalidate_bdev(struct block_device *bdev) {}
2400 static inline struct super_block *freeze_bdev(struct block_device *sb)
2405 static inline int thaw_bdev(struct block_device *bdev, struct super_block *sb)
2410 static inline void iterate_bdevs(void (*f)(struct block_device *, void *), void *arg)
2414 static inline bool sb_is_blkdev_sb(struct super_block *sb)
2419 extern int sync_filesystem(struct super_block *);
2420 extern const struct file_operations def_blk_fops;
2421 extern const struct file_operations def_chr_fops;
2423 extern int ioctl_by_bdev(struct block_device *, unsigned, unsigned long);
2424 extern int blkdev_ioctl(struct block_device *, fmode_t, unsigned, unsigned long);
2425 extern long compat_blkdev_ioctl(struct file *, unsigned, unsigned long);
2426 extern int blkdev_get(struct block_device *bdev, fmode_t mode, void *holder);
2427 extern struct block_device *blkdev_get_by_path(const char *path, fmode_t mode,
2429 extern struct block_device *blkdev_get_by_dev(dev_t dev, fmode_t mode,
2431 extern void blkdev_put(struct block_device *bdev, fmode_t mode);
2432 extern int __blkdev_reread_part(struct block_device *bdev);
2433 extern int blkdev_reread_part(struct block_device *bdev);
2436 extern int bd_link_disk_holder(struct block_device *bdev, struct gendisk *disk);
2437 extern void bd_unlink_disk_holder(struct block_device *bdev,
2438 struct gendisk *disk);
2440 static inline int bd_link_disk_holder(struct block_device *bdev,
2441 struct gendisk *disk)
2445 static inline void bd_unlink_disk_holder(struct block_device *bdev,
2446 struct gendisk *disk)
2453 #define CHRDEV_MAJOR_HASH_SIZE 255
2454 /* Marks the bottom of the first segment of free char majors */
2455 #define CHRDEV_MAJOR_DYN_END 234
2456 extern int alloc_chrdev_region(dev_t *, unsigned, unsigned, const char *);
2457 extern int register_chrdev_region(dev_t, unsigned, const char *);
2458 extern int __register_chrdev(unsigned int major, unsigned int baseminor,
2459 unsigned int count, const char *name,
2460 const struct file_operations *fops);
2461 extern void __unregister_chrdev(unsigned int major, unsigned int baseminor,
2462 unsigned int count, const char *name);
2463 extern void unregister_chrdev_region(dev_t, unsigned);
2464 extern void chrdev_show(struct seq_file *,off_t);
2466 static inline int register_chrdev(unsigned int major, const char *name,
2467 const struct file_operations *fops)
2469 return __register_chrdev(major, 0, 256, name, fops);
2472 static inline void unregister_chrdev(unsigned int major, const char *name)
2474 __unregister_chrdev(major, 0, 256, name);
2477 /* fs/block_dev.c */
2478 #define BDEVNAME_SIZE 32 /* Largest string for a blockdev identifier */
2479 #define BDEVT_SIZE 10 /* Largest string for MAJ:MIN for blkdev */
2482 #define BLKDEV_MAJOR_HASH_SIZE 255
2483 extern const char *__bdevname(dev_t, char *buffer);
2484 extern const char *bdevname(struct block_device *bdev, char *buffer);
2485 extern struct block_device *lookup_bdev(const char *);
2486 extern void blkdev_show(struct seq_file *,off_t);
2489 #define BLKDEV_MAJOR_HASH_SIZE 0
2492 extern void init_special_inode(struct inode *, umode_t, dev_t);
2494 /* Invalid inode operations -- fs/bad_inode.c */
2495 extern void make_bad_inode(struct inode *);
2496 extern bool is_bad_inode(struct inode *);
2499 extern void check_disk_size_change(struct gendisk *disk,
2500 struct block_device *bdev);
2501 extern int revalidate_disk(struct gendisk *);
2502 extern int check_disk_change(struct block_device *);
2503 extern int __invalidate_device(struct block_device *, bool);
2504 extern int invalidate_partition(struct gendisk *, int);
2506 unsigned long invalidate_mapping_pages(struct address_space *mapping,
2507 pgoff_t start, pgoff_t end);
2509 static inline void invalidate_remote_inode(struct inode *inode)
2511 if (S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) ||
2512 S_ISLNK(inode->i_mode))
2513 invalidate_mapping_pages(inode->i_mapping, 0, -1);
2515 extern int invalidate_inode_pages2(struct address_space *mapping);
2516 extern int invalidate_inode_pages2_range(struct address_space *mapping,
2517 pgoff_t start, pgoff_t end);
2518 extern int write_inode_now(struct inode *, int);
2519 extern int filemap_fdatawrite(struct address_space *);
2520 extern int filemap_flush(struct address_space *);
2521 extern int filemap_fdatawait(struct address_space *);
2522 extern void filemap_fdatawait_keep_errors(struct address_space *);
2523 extern int filemap_fdatawait_range(struct address_space *, loff_t lstart,
2525 extern bool filemap_range_has_page(struct address_space *, loff_t lstart,
2527 extern int filemap_write_and_wait(struct address_space *mapping);
2528 extern int filemap_write_and_wait_range(struct address_space *mapping,
2529 loff_t lstart, loff_t lend);
2530 extern int __filemap_fdatawrite_range(struct address_space *mapping,
2531 loff_t start, loff_t end, int sync_mode);
2532 extern int filemap_fdatawrite_range(struct address_space *mapping,
2533 loff_t start, loff_t end);
2534 extern int filemap_check_errors(struct address_space *mapping);
2536 extern int vfs_fsync_range(struct file *file, loff_t start, loff_t end,
2538 extern int vfs_fsync(struct file *file, int datasync);
2541 * Sync the bytes written if this was a synchronous write. Expect ki_pos
2542 * to already be updated for the write, and will return either the amount
2543 * of bytes passed in, or an error if syncing the file failed.
2545 static inline ssize_t generic_write_sync(struct kiocb *iocb, ssize_t count)
2547 if (iocb->ki_flags & IOCB_DSYNC) {
2548 int ret = vfs_fsync_range(iocb->ki_filp,
2549 iocb->ki_pos - count, iocb->ki_pos - 1,
2550 (iocb->ki_flags & IOCB_SYNC) ? 0 : 1);
2558 extern void emergency_sync(void);
2559 extern void emergency_remount(void);
2561 extern sector_t bmap(struct inode *, sector_t);
2563 extern int notify_change(struct dentry *, struct iattr *, struct inode **);
2564 extern int inode_permission(struct inode *, int);
2565 extern int __inode_permission(struct inode *, int);
2566 extern int generic_permission(struct inode *, int);
2567 extern int __check_sticky(struct inode *dir, struct inode *inode);
2569 static inline bool execute_ok(struct inode *inode)
2571 return (inode->i_mode & S_IXUGO) || S_ISDIR(inode->i_mode);
2574 static inline void file_start_write(struct file *file)
2576 if (!S_ISREG(file_inode(file)->i_mode))
2578 __sb_start_write(file_inode(file)->i_sb, SB_FREEZE_WRITE, true);
2581 static inline bool file_start_write_trylock(struct file *file)
2583 if (!S_ISREG(file_inode(file)->i_mode))
2585 return __sb_start_write(file_inode(file)->i_sb, SB_FREEZE_WRITE, false);
2588 static inline void file_end_write(struct file *file)
2590 if (!S_ISREG(file_inode(file)->i_mode))
2592 __sb_end_write(file_inode(file)->i_sb, SB_FREEZE_WRITE);
2595 static inline int do_clone_file_range(struct file *file_in, loff_t pos_in,
2596 struct file *file_out, loff_t pos_out,
2601 file_start_write(file_out);
2602 ret = vfs_clone_file_range(file_in, pos_in, file_out, pos_out, len);
2603 file_end_write(file_out);
2609 * get_write_access() gets write permission for a file.
2610 * put_write_access() releases this write permission.
2611 * This is used for regular files.
2612 * We cannot support write (and maybe mmap read-write shared) accesses and
2613 * MAP_DENYWRITE mmappings simultaneously. The i_writecount field of an inode
2614 * can have the following values:
2615 * 0: no writers, no VM_DENYWRITE mappings
2616 * < 0: (-i_writecount) vm_area_structs with VM_DENYWRITE set exist
2617 * > 0: (i_writecount) users are writing to the file.
2619 * Normally we operate on that counter with atomic_{inc,dec} and it's safe
2620 * except for the cases where we don't hold i_writecount yet. Then we need to
2621 * use {get,deny}_write_access() - these functions check the sign and refuse
2622 * to do the change if sign is wrong.
2624 static inline int get_write_access(struct inode *inode)
2626 return atomic_inc_unless_negative(&inode->i_writecount) ? 0 : -ETXTBSY;
2628 static inline int deny_write_access(struct file *file)
2630 struct inode *inode = file_inode(file);
2631 return atomic_dec_unless_positive(&inode->i_writecount) ? 0 : -ETXTBSY;
2633 static inline void put_write_access(struct inode * inode)
2635 atomic_dec(&inode->i_writecount);
2637 static inline void allow_write_access(struct file *file)
2640 atomic_inc(&file_inode(file)->i_writecount);
2642 static inline bool inode_is_open_for_write(const struct inode *inode)
2644 return atomic_read(&inode->i_writecount) > 0;
2648 static inline void i_readcount_dec(struct inode *inode)
2650 BUG_ON(!atomic_read(&inode->i_readcount));
2651 atomic_dec(&inode->i_readcount);
2653 static inline void i_readcount_inc(struct inode *inode)
2655 atomic_inc(&inode->i_readcount);
2658 static inline void i_readcount_dec(struct inode *inode)
2662 static inline void i_readcount_inc(struct inode *inode)
2667 extern int do_pipe_flags(int *, int);
2669 #define __kernel_read_file_id(id) \
2670 id(UNKNOWN, unknown) \
2671 id(FIRMWARE, firmware) \
2672 id(FIRMWARE_PREALLOC_BUFFER, firmware) \
2673 id(MODULE, kernel-module) \
2674 id(KEXEC_IMAGE, kexec-image) \
2675 id(KEXEC_INITRAMFS, kexec-initramfs) \
2676 id(POLICY, security-policy) \
2679 #define __fid_enumify(ENUM, dummy) READING_ ## ENUM,
2680 #define __fid_stringify(dummy, str) #str,
2682 enum kernel_read_file_id {
2683 __kernel_read_file_id(__fid_enumify)
2686 static const char * const kernel_read_file_str[] = {
2687 __kernel_read_file_id(__fid_stringify)
2690 static inline const char *kernel_read_file_id_str(enum kernel_read_file_id id)
2692 if ((unsigned)id >= READING_MAX_ID)
2693 return kernel_read_file_str[READING_UNKNOWN];
2695 return kernel_read_file_str[id];
2698 extern int kernel_read(struct file *, loff_t, char *, unsigned long);
2699 extern int kernel_read_file(struct file *, void **, loff_t *, loff_t,
2700 enum kernel_read_file_id);
2701 extern int kernel_read_file_from_path(char *, void **, loff_t *, loff_t,
2702 enum kernel_read_file_id);
2703 extern int kernel_read_file_from_fd(int, void **, loff_t *, loff_t,
2704 enum kernel_read_file_id);
2705 extern ssize_t kernel_write(struct file *, const char *, size_t, loff_t);
2706 extern ssize_t __kernel_write(struct file *, const char *, size_t, loff_t *);
2707 extern struct file * open_exec(const char *);
2709 /* fs/dcache.c -- generic fs support functions */
2710 extern bool is_subdir(struct dentry *, struct dentry *);
2711 extern bool path_is_under(const struct path *, const struct path *);
2713 extern char *file_path(struct file *, char *, int);
2715 #include <linux/err.h>
2717 /* needed for stackable file system support */
2718 extern loff_t default_llseek(struct file *file, loff_t offset, int whence);
2720 extern loff_t vfs_llseek(struct file *file, loff_t offset, int whence);
2722 extern int inode_init_always(struct super_block *, struct inode *);
2723 extern void inode_init_once(struct inode *);
2724 extern void address_space_init_once(struct address_space *mapping);
2725 extern struct inode * igrab(struct inode *);
2726 extern ino_t iunique(struct super_block *, ino_t);
2727 extern int inode_needs_sync(struct inode *inode);
2728 extern int generic_delete_inode(struct inode *inode);
2729 static inline int generic_drop_inode(struct inode *inode)
2731 return !inode->i_nlink || inode_unhashed(inode);
2734 extern struct inode *ilookup5_nowait(struct super_block *sb,
2735 unsigned long hashval, int (*test)(struct inode *, void *),
2737 extern struct inode *ilookup5(struct super_block *sb, unsigned long hashval,
2738 int (*test)(struct inode *, void *), void *data);
2739 extern struct inode *ilookup(struct super_block *sb, unsigned long ino);
2741 extern struct inode * iget5_locked(struct super_block *, unsigned long, int (*test)(struct inode *, void *), int (*set)(struct inode *, void *), void *);
2742 extern struct inode * iget_locked(struct super_block *, unsigned long);
2743 extern struct inode *find_inode_nowait(struct super_block *,
2745 int (*match)(struct inode *,
2746 unsigned long, void *),
2748 extern int insert_inode_locked4(struct inode *, unsigned long, int (*test)(struct inode *, void *), void *);
2749 extern int insert_inode_locked(struct inode *);
2750 #ifdef CONFIG_DEBUG_LOCK_ALLOC
2751 extern void lockdep_annotate_inode_mutex_key(struct inode *inode);
2753 static inline void lockdep_annotate_inode_mutex_key(struct inode *inode) { };
2755 extern void unlock_new_inode(struct inode *);
2756 extern unsigned int get_next_ino(void);
2758 extern void __iget(struct inode * inode);
2759 extern void iget_failed(struct inode *);
2760 extern void clear_inode(struct inode *);
2761 extern void __destroy_inode(struct inode *);
2762 extern struct inode *new_inode_pseudo(struct super_block *sb);
2763 extern struct inode *new_inode(struct super_block *sb);
2764 extern void free_inode_nonrcu(struct inode *inode);
2765 extern int should_remove_suid(struct dentry *);
2766 extern int file_remove_privs(struct file *);
2768 extern void __insert_inode_hash(struct inode *, unsigned long hashval);
2769 static inline void insert_inode_hash(struct inode *inode)
2771 __insert_inode_hash(inode, inode->i_ino);
2774 extern void __remove_inode_hash(struct inode *);
2775 static inline void remove_inode_hash(struct inode *inode)
2777 if (!inode_unhashed(inode) && !hlist_fake(&inode->i_hash))
2778 __remove_inode_hash(inode);
2781 extern void inode_sb_list_add(struct inode *inode);
2784 extern int bdev_read_only(struct block_device *);
2786 extern int set_blocksize(struct block_device *, int);
2787 extern int sb_set_blocksize(struct super_block *, int);
2788 extern int sb_min_blocksize(struct super_block *, int);
2790 extern int generic_file_mmap(struct file *, struct vm_area_struct *);
2791 extern int generic_file_readonly_mmap(struct file *, struct vm_area_struct *);
2792 extern ssize_t generic_write_checks(struct kiocb *, struct iov_iter *);
2793 extern ssize_t generic_file_read_iter(struct kiocb *, struct iov_iter *);
2794 extern ssize_t __generic_file_write_iter(struct kiocb *, struct iov_iter *);
2795 extern ssize_t generic_file_write_iter(struct kiocb *, struct iov_iter *);
2796 extern ssize_t generic_file_direct_write(struct kiocb *, struct iov_iter *);
2797 extern ssize_t generic_perform_write(struct file *, struct iov_iter *, loff_t);
2799 ssize_t vfs_iter_read(struct file *file, struct iov_iter *iter, loff_t *ppos);
2800 ssize_t vfs_iter_write(struct file *file, struct iov_iter *iter, loff_t *ppos);
2802 /* fs/block_dev.c */
2803 extern ssize_t blkdev_read_iter(struct kiocb *iocb, struct iov_iter *to);
2804 extern ssize_t blkdev_write_iter(struct kiocb *iocb, struct iov_iter *from);
2805 extern int blkdev_fsync(struct file *filp, loff_t start, loff_t end,
2807 extern void block_sync_page(struct page *page);
2810 extern ssize_t generic_file_splice_read(struct file *, loff_t *,
2811 struct pipe_inode_info *, size_t, unsigned int);
2812 extern ssize_t iter_file_splice_write(struct pipe_inode_info *,
2813 struct file *, loff_t *, size_t, unsigned int);
2814 extern ssize_t generic_splice_sendpage(struct pipe_inode_info *pipe,
2815 struct file *out, loff_t *, size_t len, unsigned int flags);
2816 extern long do_splice_direct(struct file *in, loff_t *ppos, struct file *out,
2817 loff_t *opos, size_t len, unsigned int flags);
2821 file_ra_state_init(struct file_ra_state *ra, struct address_space *mapping);
2822 extern loff_t noop_llseek(struct file *file, loff_t offset, int whence);
2823 extern loff_t no_llseek(struct file *file, loff_t offset, int whence);
2824 extern loff_t vfs_setpos(struct file *file, loff_t offset, loff_t maxsize);
2825 extern loff_t generic_file_llseek(struct file *file, loff_t offset, int whence);
2826 extern loff_t generic_file_llseek_size(struct file *file, loff_t offset,
2827 int whence, loff_t maxsize, loff_t eof);
2828 extern loff_t fixed_size_llseek(struct file *file, loff_t offset,
2829 int whence, loff_t size);
2830 extern loff_t no_seek_end_llseek_size(struct file *, loff_t, int, loff_t);
2831 extern loff_t no_seek_end_llseek(struct file *, loff_t, int);
2832 extern int generic_file_open(struct inode * inode, struct file * filp);
2833 extern int nonseekable_open(struct inode * inode, struct file * filp);
2836 typedef void (dio_submit_t)(struct bio *bio, struct inode *inode,
2837 loff_t file_offset);
2840 /* need locking between buffered and direct access */
2843 /* filesystem does not support filling holes */
2844 DIO_SKIP_HOLES = 0x02,
2846 /* filesystem can handle aio writes beyond i_size */
2847 DIO_ASYNC_EXTEND = 0x04,
2849 /* inode/fs/bdev does not need truncate protection */
2850 DIO_SKIP_DIO_COUNT = 0x08,
2853 void dio_end_io(struct bio *bio);
2855 ssize_t __blockdev_direct_IO(struct kiocb *iocb, struct inode *inode,
2856 struct block_device *bdev, struct iov_iter *iter,
2857 get_block_t get_block,
2858 dio_iodone_t end_io, dio_submit_t submit_io,
2861 static inline ssize_t blockdev_direct_IO(struct kiocb *iocb,
2862 struct inode *inode,
2863 struct iov_iter *iter,
2864 get_block_t get_block)
2866 return __blockdev_direct_IO(iocb, inode, inode->i_sb->s_bdev, iter,
2867 get_block, NULL, NULL, DIO_LOCKING | DIO_SKIP_HOLES);
2871 void inode_dio_wait(struct inode *inode);
2874 * inode_dio_begin - signal start of a direct I/O requests
2875 * @inode: inode the direct I/O happens on
2877 * This is called once we've finished processing a direct I/O request,
2878 * and is used to wake up callers waiting for direct I/O to be quiesced.
2880 static inline void inode_dio_begin(struct inode *inode)
2882 atomic_inc(&inode->i_dio_count);
2886 * inode_dio_end - signal finish of a direct I/O requests
2887 * @inode: inode the direct I/O happens on
2889 * This is called once we've finished processing a direct I/O request,
2890 * and is used to wake up callers waiting for direct I/O to be quiesced.
2892 static inline void inode_dio_end(struct inode *inode)
2894 if (atomic_dec_and_test(&inode->i_dio_count))
2895 wake_up_bit(&inode->i_state, __I_DIO_WAKEUP);
2898 extern void inode_set_flags(struct inode *inode, unsigned int flags,
2901 extern const struct file_operations generic_ro_fops;
2903 #define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
2905 extern int readlink_copy(char __user *, int, const char *);
2906 extern int page_readlink(struct dentry *, char __user *, int);
2907 extern const char *page_get_link(struct dentry *, struct inode *,
2908 struct delayed_call *);
2909 extern void page_put_link(void *);
2910 extern int __page_symlink(struct inode *inode, const char *symname, int len,
2912 extern int page_symlink(struct inode *inode, const char *symname, int len);
2913 extern const struct inode_operations page_symlink_inode_operations;
2914 extern void kfree_link(void *);
2915 extern void generic_fillattr(struct inode *, struct kstat *);
2916 extern int vfs_getattr_nosec(const struct path *, struct kstat *, u32, unsigned int);
2917 extern int vfs_getattr(const struct path *, struct kstat *, u32, unsigned int);
2918 void __inode_add_bytes(struct inode *inode, loff_t bytes);
2919 void inode_add_bytes(struct inode *inode, loff_t bytes);
2920 void __inode_sub_bytes(struct inode *inode, loff_t bytes);
2921 void inode_sub_bytes(struct inode *inode, loff_t bytes);
2922 loff_t inode_get_bytes(struct inode *inode);
2923 void inode_set_bytes(struct inode *inode, loff_t bytes);
2924 const char *simple_get_link(struct dentry *, struct inode *,
2925 struct delayed_call *);
2926 extern const struct inode_operations simple_symlink_inode_operations;
2928 extern int iterate_dir(struct file *, struct dir_context *);
2930 extern int vfs_statx(int, const char __user *, int, struct kstat *, u32);
2931 extern int vfs_statx_fd(unsigned int, struct kstat *, u32, unsigned int);
2933 static inline int vfs_stat(const char __user *filename, struct kstat *stat)
2935 return vfs_statx(AT_FDCWD, filename, AT_NO_AUTOMOUNT,
2936 stat, STATX_BASIC_STATS);
2938 static inline int vfs_lstat(const char __user *name, struct kstat *stat)
2940 return vfs_statx(AT_FDCWD, name, AT_SYMLINK_NOFOLLOW | AT_NO_AUTOMOUNT,
2941 stat, STATX_BASIC_STATS);
2943 static inline int vfs_fstatat(int dfd, const char __user *filename,
2944 struct kstat *stat, int flags)
2946 return vfs_statx(dfd, filename, flags | AT_NO_AUTOMOUNT,
2947 stat, STATX_BASIC_STATS);
2949 static inline int vfs_fstat(int fd, struct kstat *stat)
2951 return vfs_statx_fd(fd, stat, STATX_BASIC_STATS, 0);
2955 extern const char *vfs_get_link(struct dentry *, struct delayed_call *);
2956 extern int vfs_readlink(struct dentry *, char __user *, int);
2958 extern int __generic_block_fiemap(struct inode *inode,
2959 struct fiemap_extent_info *fieinfo,
2960 loff_t start, loff_t len,
2961 get_block_t *get_block);
2962 extern int generic_block_fiemap(struct inode *inode,
2963 struct fiemap_extent_info *fieinfo, u64 start,
2964 u64 len, get_block_t *get_block);
2966 extern void get_filesystem(struct file_system_type *fs);
2967 extern void put_filesystem(struct file_system_type *fs);
2968 extern struct file_system_type *get_fs_type(const char *name);
2969 extern struct super_block *get_super(struct block_device *);
2970 extern struct super_block *get_super_thawed(struct block_device *);
2971 extern struct super_block *get_super_exclusive_thawed(struct block_device *bdev);
2972 extern struct super_block *get_active_super(struct block_device *bdev);
2973 extern void drop_super(struct super_block *sb);
2974 extern void drop_super_exclusive(struct super_block *sb);
2975 extern void iterate_supers(void (*)(struct super_block *, void *), void *);
2976 extern void iterate_supers_type(struct file_system_type *,
2977 void (*)(struct super_block *, void *), void *);
2979 extern int dcache_dir_open(struct inode *, struct file *);
2980 extern int dcache_dir_close(struct inode *, struct file *);
2981 extern loff_t dcache_dir_lseek(struct file *, loff_t, int);
2982 extern int dcache_readdir(struct file *, struct dir_context *);
2983 extern int simple_setattr(struct dentry *, struct iattr *);
2984 extern int simple_getattr(const struct path *, struct kstat *, u32, unsigned int);
2985 extern int simple_statfs(struct dentry *, struct kstatfs *);
2986 extern int simple_open(struct inode *inode, struct file *file);
2987 extern int simple_link(struct dentry *, struct inode *, struct dentry *);
2988 extern int simple_unlink(struct inode *, struct dentry *);
2989 extern int simple_rmdir(struct inode *, struct dentry *);
2990 extern int simple_rename(struct inode *, struct dentry *,
2991 struct inode *, struct dentry *, unsigned int);
2992 extern int noop_fsync(struct file *, loff_t, loff_t, int);
2993 extern int simple_empty(struct dentry *);
2994 extern int simple_readpage(struct file *file, struct page *page);
2995 extern int simple_write_begin(struct file *file, struct address_space *mapping,
2996 loff_t pos, unsigned len, unsigned flags,
2997 struct page **pagep, void **fsdata);
2998 extern int simple_write_end(struct file *file, struct address_space *mapping,
2999 loff_t pos, unsigned len, unsigned copied,
3000 struct page *page, void *fsdata);
3001 extern int always_delete_dentry(const struct dentry *);
3002 extern struct inode *alloc_anon_inode(struct super_block *);
3003 extern int simple_nosetlease(struct file *, long, struct file_lock **, void **);
3004 extern const struct dentry_operations simple_dentry_operations;
3006 extern struct dentry *simple_lookup(struct inode *, struct dentry *, unsigned int flags);
3007 extern ssize_t generic_read_dir(struct file *, char __user *, size_t, loff_t *);
3008 extern const struct file_operations simple_dir_operations;
3009 extern const struct inode_operations simple_dir_inode_operations;
3010 extern void make_empty_dir_inode(struct inode *inode);
3011 extern bool is_empty_dir_inode(struct inode *inode);
3012 struct tree_descr { const char *name; const struct file_operations *ops; int mode; };
3013 struct dentry *d_alloc_name(struct dentry *, const char *);
3014 extern int simple_fill_super(struct super_block *, unsigned long,
3015 const struct tree_descr *);
3016 extern int simple_pin_fs(struct file_system_type *, struct vfsmount **mount, int *count);
3017 extern void simple_release_fs(struct vfsmount **mount, int *count);
3019 extern ssize_t simple_read_from_buffer(void __user *to, size_t count,
3020 loff_t *ppos, const void *from, size_t available);
3021 extern ssize_t simple_write_to_buffer(void *to, size_t available, loff_t *ppos,
3022 const void __user *from, size_t count);
3024 extern int __generic_file_fsync(struct file *, loff_t, loff_t, int);
3025 extern int generic_file_fsync(struct file *, loff_t, loff_t, int);
3027 extern int generic_check_addressable(unsigned, u64);
3029 #ifdef CONFIG_MIGRATION
3030 extern int buffer_migrate_page(struct address_space *,
3031 struct page *, struct page *,
3034 #define buffer_migrate_page NULL
3037 extern int setattr_prepare(struct dentry *, struct iattr *);
3038 extern int inode_newsize_ok(const struct inode *, loff_t offset);
3039 extern void setattr_copy(struct inode *inode, const struct iattr *attr);
3041 extern int file_update_time(struct file *file);
3043 extern int generic_show_options(struct seq_file *m, struct dentry *root);
3044 extern void save_mount_options(struct super_block *sb, char *options);
3045 extern void replace_mount_options(struct super_block *sb, char *options);
3047 static inline bool io_is_direct(struct file *filp)
3049 return (filp->f_flags & O_DIRECT) || IS_DAX(filp->f_mapping->host);
3052 static inline int iocb_flags(struct file *file)
3055 if (file->f_flags & O_APPEND)
3057 if (io_is_direct(file))
3059 if ((file->f_flags & O_DSYNC) || IS_SYNC(file->f_mapping->host))
3061 if (file->f_flags & __O_SYNC)
3066 static inline int kiocb_set_rw_flags(struct kiocb *ki, int flags)
3068 if (unlikely(flags & ~RWF_SUPPORTED))
3071 if (flags & RWF_NOWAIT) {
3072 if (!(ki->ki_filp->f_mode & FMODE_AIO_NOWAIT))
3074 ki->ki_flags |= IOCB_NOWAIT;
3076 if (flags & RWF_HIPRI)
3077 ki->ki_flags |= IOCB_HIPRI;
3078 if (flags & RWF_DSYNC)
3079 ki->ki_flags |= IOCB_DSYNC;
3080 if (flags & RWF_SYNC)
3081 ki->ki_flags |= (IOCB_DSYNC | IOCB_SYNC);
3085 static inline ino_t parent_ino(struct dentry *dentry)
3090 * Don't strictly need d_lock here? If the parent ino could change
3091 * then surely we'd have a deeper race in the caller?
3093 spin_lock(&dentry->d_lock);
3094 res = dentry->d_parent->d_inode->i_ino;
3095 spin_unlock(&dentry->d_lock);
3099 /* Transaction based IO helpers */
3102 * An argresp is stored in an allocated page and holds the
3103 * size of the argument or response, along with its content
3105 struct simple_transaction_argresp {
3110 #define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
3112 char *simple_transaction_get(struct file *file, const char __user *buf,
3114 ssize_t simple_transaction_read(struct file *file, char __user *buf,
3115 size_t size, loff_t *pos);
3116 int simple_transaction_release(struct inode *inode, struct file *file);
3118 void simple_transaction_set(struct file *file, size_t n);
3121 * simple attribute files
3123 * These attributes behave similar to those in sysfs:
3125 * Writing to an attribute immediately sets a value, an open file can be
3126 * written to multiple times.
3128 * Reading from an attribute creates a buffer from the value that might get
3129 * read with multiple read calls. When the attribute has been read
3130 * completely, no further read calls are possible until the file is opened
3133 * All attributes contain a text representation of a numeric value
3134 * that are accessed with the get() and set() functions.
3136 #define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt) \
3137 static int __fops ## _open(struct inode *inode, struct file *file) \
3139 __simple_attr_check_format(__fmt, 0ull); \
3140 return simple_attr_open(inode, file, __get, __set, __fmt); \
3142 static const struct file_operations __fops = { \
3143 .owner = THIS_MODULE, \
3144 .open = __fops ## _open, \
3145 .release = simple_attr_release, \
3146 .read = simple_attr_read, \
3147 .write = simple_attr_write, \
3148 .llseek = generic_file_llseek, \
3151 static inline __printf(1, 2)
3152 void __simple_attr_check_format(const char *fmt, ...)
3154 /* don't do anything, just let the compiler check the arguments; */
3157 int simple_attr_open(struct inode *inode, struct file *file,
3158 int (*get)(void *, u64 *), int (*set)(void *, u64),
3160 int simple_attr_release(struct inode *inode, struct file *file);
3161 ssize_t simple_attr_read(struct file *file, char __user *buf,
3162 size_t len, loff_t *ppos);
3163 ssize_t simple_attr_write(struct file *file, const char __user *buf,
3164 size_t len, loff_t *ppos);
3167 int proc_nr_files(struct ctl_table *table, int write,
3168 void __user *buffer, size_t *lenp, loff_t *ppos);
3169 int proc_nr_dentry(struct ctl_table *table, int write,
3170 void __user *buffer, size_t *lenp, loff_t *ppos);
3171 int proc_nr_inodes(struct ctl_table *table, int write,
3172 void __user *buffer, size_t *lenp, loff_t *ppos);
3173 int __init get_filesystem_list(char *buf);
3175 #define __FMODE_EXEC ((__force int) FMODE_EXEC)
3176 #define __FMODE_NONOTIFY ((__force int) FMODE_NONOTIFY)
3178 #define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE])
3179 #define OPEN_FMODE(flag) ((__force fmode_t)(((flag + 1) & O_ACCMODE) | \
3180 (flag & __FMODE_NONOTIFY)))
3182 static inline bool is_sxid(umode_t mode)
3184 return (mode & S_ISUID) || ((mode & S_ISGID) && (mode & S_IXGRP));
3187 static inline int check_sticky(struct inode *dir, struct inode *inode)
3189 if (!(dir->i_mode & S_ISVTX))
3192 return __check_sticky(dir, inode);
3195 static inline void inode_has_no_xattr(struct inode *inode)
3197 if (!is_sxid(inode->i_mode) && (inode->i_sb->s_flags & MS_NOSEC))
3198 inode->i_flags |= S_NOSEC;
3201 static inline bool is_root_inode(struct inode *inode)
3203 return inode == inode->i_sb->s_root->d_inode;
3206 static inline bool dir_emit(struct dir_context *ctx,
3207 const char *name, int namelen,
3208 u64 ino, unsigned type)
3210 return ctx->actor(ctx, name, namelen, ctx->pos, ino, type) == 0;
3212 static inline bool dir_emit_dot(struct file *file, struct dir_context *ctx)
3214 return ctx->actor(ctx, ".", 1, ctx->pos,
3215 file->f_path.dentry->d_inode->i_ino, DT_DIR) == 0;
3217 static inline bool dir_emit_dotdot(struct file *file, struct dir_context *ctx)
3219 return ctx->actor(ctx, "..", 2, ctx->pos,
3220 parent_ino(file->f_path.dentry), DT_DIR) == 0;
3222 static inline bool dir_emit_dots(struct file *file, struct dir_context *ctx)
3224 if (ctx->pos == 0) {
3225 if (!dir_emit_dot(file, ctx))
3229 if (ctx->pos == 1) {
3230 if (!dir_emit_dotdot(file, ctx))
3236 static inline bool dir_relax(struct inode *inode)
3238 inode_unlock(inode);
3240 return !IS_DEADDIR(inode);
3243 static inline bool dir_relax_shared(struct inode *inode)
3245 inode_unlock_shared(inode);
3246 inode_lock_shared(inode);
3247 return !IS_DEADDIR(inode);
3250 extern bool path_noexec(const struct path *path);
3251 extern void inode_nohighmem(struct inode *inode);
3253 #endif /* _LINUX_FS_H */