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/percpu-rwsem.h>
33 #include <linux/delayed_call.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)
147 * Flag for rw_copy_check_uvector and compat_rw_copy_check_uvector
148 * that indicates that they should check the contents of the iovec are
149 * valid, but not check the memory that the iovec elements
152 #define CHECK_IOVEC_ONLY -1
155 * Attribute flags. These should be or-ed together to figure out what
158 #define ATTR_MODE (1 << 0)
159 #define ATTR_UID (1 << 1)
160 #define ATTR_GID (1 << 2)
161 #define ATTR_SIZE (1 << 3)
162 #define ATTR_ATIME (1 << 4)
163 #define ATTR_MTIME (1 << 5)
164 #define ATTR_CTIME (1 << 6)
165 #define ATTR_ATIME_SET (1 << 7)
166 #define ATTR_MTIME_SET (1 << 8)
167 #define ATTR_FORCE (1 << 9) /* Not a change, but a change it */
168 #define ATTR_ATTR_FLAG (1 << 10)
169 #define ATTR_KILL_SUID (1 << 11)
170 #define ATTR_KILL_SGID (1 << 12)
171 #define ATTR_FILE (1 << 13)
172 #define ATTR_KILL_PRIV (1 << 14)
173 #define ATTR_OPEN (1 << 15) /* Truncating from open(O_TRUNC) */
174 #define ATTR_TIMES_SET (1 << 16)
175 #define ATTR_TOUCH (1 << 17)
178 * Whiteout is represented by a char device. The following constants define the
179 * mode and device number to use.
181 #define WHITEOUT_MODE 0
182 #define WHITEOUT_DEV 0
185 * This is the Inode Attributes structure, used for notify_change(). It
186 * uses the above definitions as flags, to know which values have changed.
187 * Also, in this manner, a Filesystem can look at only the values it cares
188 * about. Basically, these are the attributes that the VFS layer can
189 * request to change from the FS layer.
191 * Derek Atkins <warlord@MIT.EDU> 94-10-20
194 unsigned int ia_valid;
199 struct timespec ia_atime;
200 struct timespec ia_mtime;
201 struct timespec ia_ctime;
204 * Not an attribute, but an auxiliary info for filesystems wanting to
205 * implement an ftruncate() like method. NOTE: filesystem should
206 * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL).
208 struct file *ia_file;
212 * Includes for diskquotas.
214 #include <linux/quota.h>
217 * Maximum number of layers of fs stack. Needs to be limited to
218 * prevent kernel stack overflow
220 #define FILESYSTEM_MAX_STACK_DEPTH 2
223 * enum positive_aop_returns - aop return codes with specific semantics
225 * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has
226 * completed, that the page is still locked, and
227 * should be considered active. The VM uses this hint
228 * to return the page to the active list -- it won't
229 * be a candidate for writeback again in the near
230 * future. Other callers must be careful to unlock
231 * the page if they get this return. Returned by
234 * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has
235 * unlocked it and the page might have been truncated.
236 * The caller should back up to acquiring a new page and
237 * trying again. The aop will be taking reasonable
238 * precautions not to livelock. If the caller held a page
239 * reference, it should drop it before retrying. Returned
242 * address_space_operation functions return these large constants to indicate
243 * special semantics to the caller. These are much larger than the bytes in a
244 * page to allow for functions that return the number of bytes operated on in a
248 enum positive_aop_returns {
249 AOP_WRITEPAGE_ACTIVATE = 0x80000,
250 AOP_TRUNCATED_PAGE = 0x80001,
253 #define AOP_FLAG_UNINTERRUPTIBLE 0x0001 /* will not do a short write */
254 #define AOP_FLAG_CONT_EXPAND 0x0002 /* called from cont_expand */
255 #define AOP_FLAG_NOFS 0x0004 /* used by filesystem to direct
256 * helper code (eg buffer layer)
257 * to clear GFP_FS from alloc */
260 * oh the beauties of C type declarations.
263 struct address_space;
264 struct writeback_control;
266 #define IOCB_EVENTFD (1 << 0)
267 #define IOCB_APPEND (1 << 1)
268 #define IOCB_DIRECT (1 << 2)
269 #define IOCB_HIPRI (1 << 3)
270 #define IOCB_DSYNC (1 << 4)
271 #define IOCB_SYNC (1 << 5)
272 #define IOCB_WRITE (1 << 6)
275 struct file *ki_filp;
277 void (*ki_complete)(struct kiocb *iocb, long ret, long ret2);
282 static inline bool is_sync_kiocb(struct kiocb *kiocb)
284 return kiocb->ki_complete == NULL;
287 static inline int iocb_flags(struct file *file);
289 static inline void init_sync_kiocb(struct kiocb *kiocb, struct file *filp)
291 *kiocb = (struct kiocb) {
293 .ki_flags = iocb_flags(filp),
298 * "descriptor" for what we're up to with a read.
299 * This allows us to use the same read code yet
300 * have multiple different users of the data that
301 * we read from a file.
303 * The simplest case just copies the data to user
316 typedef int (*read_actor_t)(read_descriptor_t *, struct page *,
317 unsigned long, unsigned long);
319 struct address_space_operations {
320 int (*writepage)(struct page *page, struct writeback_control *wbc);
321 int (*readpage)(struct file *, struct page *);
323 /* Write back some dirty pages from this mapping. */
324 int (*writepages)(struct address_space *, struct writeback_control *);
326 /* Set a page dirty. Return true if this dirtied it */
327 int (*set_page_dirty)(struct page *page);
329 int (*readpages)(struct file *filp, struct address_space *mapping,
330 struct list_head *pages, unsigned nr_pages);
332 int (*write_begin)(struct file *, struct address_space *mapping,
333 loff_t pos, unsigned len, unsigned flags,
334 struct page **pagep, void **fsdata);
335 int (*write_end)(struct file *, struct address_space *mapping,
336 loff_t pos, unsigned len, unsigned copied,
337 struct page *page, void *fsdata);
339 /* Unfortunately this kludge is needed for FIBMAP. Don't use it */
340 sector_t (*bmap)(struct address_space *, sector_t);
341 void (*invalidatepage) (struct page *, unsigned int, unsigned int);
342 int (*releasepage) (struct page *, gfp_t);
343 void (*freepage)(struct page *);
344 ssize_t (*direct_IO)(struct kiocb *, struct iov_iter *iter);
346 * migrate the contents of a page to the specified target. If
347 * migrate_mode is MIGRATE_ASYNC, it must not block.
349 int (*migratepage) (struct address_space *,
350 struct page *, struct page *, enum migrate_mode);
351 bool (*isolate_page)(struct page *, isolate_mode_t);
352 void (*putback_page)(struct page *);
353 int (*launder_page) (struct page *);
354 int (*is_partially_uptodate) (struct page *, unsigned long,
356 void (*is_dirty_writeback) (struct page *, bool *, bool *);
357 int (*error_remove_page)(struct address_space *, struct page *);
359 /* swapfile support */
360 int (*swap_activate)(struct swap_info_struct *sis, struct file *file,
362 void (*swap_deactivate)(struct file *file);
365 extern const struct address_space_operations empty_aops;
368 * pagecache_write_begin/pagecache_write_end must be used by general code
369 * to write into the pagecache.
371 int pagecache_write_begin(struct file *, struct address_space *mapping,
372 loff_t pos, unsigned len, unsigned flags,
373 struct page **pagep, void **fsdata);
375 int pagecache_write_end(struct file *, struct address_space *mapping,
376 loff_t pos, unsigned len, unsigned copied,
377 struct page *page, void *fsdata);
379 struct address_space {
380 struct inode *host; /* owner: inode, block_device */
381 struct radix_tree_root page_tree; /* radix tree of all pages */
382 spinlock_t tree_lock; /* and lock protecting it */
383 atomic_t i_mmap_writable;/* count VM_SHARED mappings */
384 struct rb_root i_mmap; /* tree of private and shared mappings */
385 struct rw_semaphore i_mmap_rwsem; /* protect tree, count, list */
386 /* Protected by tree_lock together with the radix tree */
387 unsigned long nrpages; /* number of total pages */
388 /* number of shadow or DAX exceptional entries */
389 unsigned long nrexceptional;
390 pgoff_t writeback_index;/* writeback starts here */
391 const struct address_space_operations *a_ops; /* methods */
392 unsigned long flags; /* error bits */
393 spinlock_t private_lock; /* for use by the address_space */
394 gfp_t gfp_mask; /* implicit gfp mask for allocations */
395 struct list_head private_list; /* ditto */
396 void *private_data; /* ditto */
397 } __attribute__((aligned(sizeof(long))));
399 * On most architectures that alignment is already the case; but
400 * must be enforced here for CRIS, to let the least significant bit
401 * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
403 struct request_queue;
405 struct block_device {
406 dev_t bd_dev; /* not a kdev_t - it's a search key */
408 struct inode * bd_inode; /* will die */
409 struct super_block * bd_super;
410 struct mutex bd_mutex; /* open/close mutex */
414 bool bd_write_holder;
416 struct list_head bd_holder_disks;
418 struct block_device * bd_contains;
419 unsigned bd_block_size;
420 struct hd_struct * bd_part;
421 /* number of times partitions within this device have been opened. */
422 unsigned bd_part_count;
424 struct gendisk * bd_disk;
425 struct request_queue * bd_queue;
426 struct backing_dev_info *bd_bdi;
427 struct list_head bd_list;
429 * Private data. You must have bd_claim'ed the block_device
430 * to use this. NOTE: bd_claim allows an owner to claim
431 * the same device multiple times, the owner must take special
432 * care to not mess up bd_private for that case.
434 unsigned long bd_private;
436 /* The counter of freeze processes */
437 int bd_fsfreeze_count;
438 /* Mutex for freeze */
439 struct mutex bd_fsfreeze_mutex;
443 * Radix-tree tags, for tagging dirty and writeback pages within the pagecache
446 #define PAGECACHE_TAG_DIRTY 0
447 #define PAGECACHE_TAG_WRITEBACK 1
448 #define PAGECACHE_TAG_TOWRITE 2
450 int mapping_tagged(struct address_space *mapping, int tag);
452 static inline void i_mmap_lock_write(struct address_space *mapping)
454 down_write(&mapping->i_mmap_rwsem);
457 static inline void i_mmap_unlock_write(struct address_space *mapping)
459 up_write(&mapping->i_mmap_rwsem);
462 static inline void i_mmap_lock_read(struct address_space *mapping)
464 down_read(&mapping->i_mmap_rwsem);
467 static inline void i_mmap_unlock_read(struct address_space *mapping)
469 up_read(&mapping->i_mmap_rwsem);
473 * Might pages of this file be mapped into userspace?
475 static inline int mapping_mapped(struct address_space *mapping)
477 return !RB_EMPTY_ROOT(&mapping->i_mmap);
481 * Might pages of this file have been modified in userspace?
482 * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap_pgoff
483 * marks vma as VM_SHARED if it is shared, and the file was opened for
484 * writing i.e. vma may be mprotected writable even if now readonly.
486 * If i_mmap_writable is negative, no new writable mappings are allowed. You
487 * can only deny writable mappings, if none exists right now.
489 static inline int mapping_writably_mapped(struct address_space *mapping)
491 return atomic_read(&mapping->i_mmap_writable) > 0;
494 static inline int mapping_map_writable(struct address_space *mapping)
496 return atomic_inc_unless_negative(&mapping->i_mmap_writable) ?
500 static inline void mapping_unmap_writable(struct address_space *mapping)
502 atomic_dec(&mapping->i_mmap_writable);
505 static inline int mapping_deny_writable(struct address_space *mapping)
507 return atomic_dec_unless_positive(&mapping->i_mmap_writable) ?
511 static inline void mapping_allow_writable(struct address_space *mapping)
513 atomic_inc(&mapping->i_mmap_writable);
517 * Use sequence counter to get consistent i_size on 32-bit processors.
519 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
520 #include <linux/seqlock.h>
521 #define __NEED_I_SIZE_ORDERED
522 #define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
524 #define i_size_ordered_init(inode) do { } while (0)
528 #define ACL_NOT_CACHED ((void *)(-1))
529 #define ACL_DONT_CACHE ((void *)(-3))
531 static inline struct posix_acl *
532 uncached_acl_sentinel(struct task_struct *task)
534 return (void *)task + 1;
538 is_uncached_acl(struct posix_acl *acl)
540 return (long)acl & 1;
543 #define IOP_FASTPERM 0x0001
544 #define IOP_LOOKUP 0x0002
545 #define IOP_NOFOLLOW 0x0004
546 #define IOP_XATTR 0x0008
547 #define IOP_DEFAULT_READLINK 0x0010
550 * Keep mostly read-only and often accessed (especially for
551 * the RCU path lookup and 'stat' data) fields at the beginning
552 * of the 'struct inode'
556 unsigned short i_opflags;
559 unsigned int i_flags;
561 #ifdef CONFIG_FS_POSIX_ACL
562 struct posix_acl *i_acl;
563 struct posix_acl *i_default_acl;
566 const struct inode_operations *i_op;
567 struct super_block *i_sb;
568 struct address_space *i_mapping;
570 #ifdef CONFIG_SECURITY
574 /* Stat data, not accessed from path walking */
577 * Filesystems may only read i_nlink directly. They shall use the
578 * following functions for modification:
580 * (set|clear|inc|drop)_nlink
581 * inode_(inc|dec)_link_count
584 const unsigned int i_nlink;
585 unsigned int __i_nlink;
589 struct timespec i_atime;
590 struct timespec i_mtime;
591 struct timespec i_ctime;
592 spinlock_t i_lock; /* i_blocks, i_bytes, maybe i_size */
593 unsigned short i_bytes;
594 unsigned int i_blkbits;
597 #ifdef __NEED_I_SIZE_ORDERED
598 seqcount_t i_size_seqcount;
602 unsigned long i_state;
603 struct rw_semaphore i_rwsem;
605 unsigned long dirtied_when; /* jiffies of first dirtying */
606 unsigned long dirtied_time_when;
608 struct hlist_node i_hash;
609 struct list_head i_io_list; /* backing dev IO list */
610 #ifdef CONFIG_CGROUP_WRITEBACK
611 struct bdi_writeback *i_wb; /* the associated cgroup wb */
613 /* foreign inode detection, see wbc_detach_inode() */
615 u16 i_wb_frn_avg_time;
616 u16 i_wb_frn_history;
618 struct list_head i_lru; /* inode LRU list */
619 struct list_head i_sb_list;
620 struct list_head i_wb_list; /* backing dev writeback list */
622 struct hlist_head i_dentry;
623 struct rcu_head i_rcu;
627 atomic_t i_dio_count;
628 atomic_t i_writecount;
630 atomic_t i_readcount; /* struct files open RO */
632 const struct file_operations *i_fop; /* former ->i_op->default_file_ops */
633 struct file_lock_context *i_flctx;
634 struct address_space i_data;
635 struct list_head i_devices;
637 struct pipe_inode_info *i_pipe;
638 struct block_device *i_bdev;
646 #ifdef CONFIG_FSNOTIFY
647 __u32 i_fsnotify_mask; /* all events this inode cares about */
648 struct hlist_head i_fsnotify_marks;
651 #if IS_ENABLED(CONFIG_FS_ENCRYPTION)
652 struct fscrypt_info *i_crypt_info;
655 void *i_private; /* fs or device private pointer */
658 static inline int inode_unhashed(struct inode *inode)
660 return hlist_unhashed(&inode->i_hash);
664 * inode->i_mutex nesting subclasses for the lock validator:
666 * 0: the object of the current VFS operation
670 * 4: second non-directory
671 * 5: second parent (when locking independent directories in rename)
673 * I_MUTEX_NONDIR2 is for certain operations (such as rename) which lock two
674 * non-directories at once.
676 * The locking order between these classes is
677 * parent[2] -> child -> grandchild -> normal -> xattr -> second non-directory
679 enum inode_i_mutex_lock_class
689 static inline void inode_lock(struct inode *inode)
691 down_write(&inode->i_rwsem);
694 static inline void inode_unlock(struct inode *inode)
696 up_write(&inode->i_rwsem);
699 static inline void inode_lock_shared(struct inode *inode)
701 down_read(&inode->i_rwsem);
704 static inline void inode_unlock_shared(struct inode *inode)
706 up_read(&inode->i_rwsem);
709 static inline int inode_trylock(struct inode *inode)
711 return down_write_trylock(&inode->i_rwsem);
714 static inline int inode_trylock_shared(struct inode *inode)
716 return down_read_trylock(&inode->i_rwsem);
719 static inline int inode_is_locked(struct inode *inode)
721 return rwsem_is_locked(&inode->i_rwsem);
724 static inline void inode_lock_nested(struct inode *inode, unsigned subclass)
726 down_write_nested(&inode->i_rwsem, subclass);
729 void lock_two_nondirectories(struct inode *, struct inode*);
730 void unlock_two_nondirectories(struct inode *, struct inode*);
733 * NOTE: in a 32bit arch with a preemptable kernel and
734 * an UP compile the i_size_read/write must be atomic
735 * with respect to the local cpu (unlike with preempt disabled),
736 * but they don't need to be atomic with respect to other cpus like in
737 * true SMP (so they need either to either locally disable irq around
738 * the read or for example on x86 they can be still implemented as a
739 * cmpxchg8b without the need of the lock prefix). For SMP compiles
740 * and 64bit archs it makes no difference if preempt is enabled or not.
742 static inline loff_t i_size_read(const struct inode *inode)
744 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
749 seq = read_seqcount_begin(&inode->i_size_seqcount);
750 i_size = inode->i_size;
751 } while (read_seqcount_retry(&inode->i_size_seqcount, seq));
753 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
757 i_size = inode->i_size;
761 return inode->i_size;
766 * NOTE: unlike i_size_read(), i_size_write() does need locking around it
767 * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount
768 * can be lost, resulting in subsequent i_size_read() calls spinning forever.
770 static inline void i_size_write(struct inode *inode, loff_t i_size)
772 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
774 write_seqcount_begin(&inode->i_size_seqcount);
775 inode->i_size = i_size;
776 write_seqcount_end(&inode->i_size_seqcount);
778 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
780 inode->i_size = i_size;
783 inode->i_size = i_size;
787 static inline unsigned iminor(const struct inode *inode)
789 return MINOR(inode->i_rdev);
792 static inline unsigned imajor(const struct inode *inode)
794 return MAJOR(inode->i_rdev);
797 extern struct block_device *I_BDEV(struct inode *inode);
800 rwlock_t lock; /* protects pid, uid, euid fields */
801 struct pid *pid; /* pid or -pgrp where SIGIO should be sent */
802 enum pid_type pid_type; /* Kind of process group SIGIO should be sent to */
803 kuid_t uid, euid; /* uid/euid of process setting the owner */
804 int signum; /* posix.1b rt signal to be delivered on IO */
808 * Track a single file's readahead state
810 struct file_ra_state {
811 pgoff_t start; /* where readahead started */
812 unsigned int size; /* # of readahead pages */
813 unsigned int async_size; /* do asynchronous readahead when
814 there are only # of pages ahead */
816 unsigned int ra_pages; /* Maximum readahead window */
817 unsigned int mmap_miss; /* Cache miss stat for mmap accesses */
818 loff_t prev_pos; /* Cache last read() position */
822 * Check if @index falls in the readahead windows.
824 static inline int ra_has_index(struct file_ra_state *ra, pgoff_t index)
826 return (index >= ra->start &&
827 index < ra->start + ra->size);
832 struct llist_node fu_llist;
833 struct rcu_head fu_rcuhead;
836 struct inode *f_inode; /* cached value */
837 const struct file_operations *f_op;
840 * Protects f_ep_links, f_flags.
841 * Must not be taken from IRQ context.
844 atomic_long_t f_count;
845 unsigned int f_flags;
847 struct mutex f_pos_lock;
849 struct fown_struct f_owner;
850 const struct cred *f_cred;
851 struct file_ra_state f_ra;
854 #ifdef CONFIG_SECURITY
857 /* needed for tty driver, and maybe others */
861 /* Used by fs/eventpoll.c to link all the hooks to this file */
862 struct list_head f_ep_links;
863 struct list_head f_tfile_llink;
864 #endif /* #ifdef CONFIG_EPOLL */
865 struct address_space *f_mapping;
866 } __attribute__((aligned(4))); /* lest something weird decides that 2 is OK */
871 /* file identifier */
872 unsigned char f_handle[0];
875 static inline struct file *get_file(struct file *f)
877 atomic_long_inc(&f->f_count);
880 #define get_file_rcu(x) atomic_long_inc_not_zero(&(x)->f_count)
881 #define fput_atomic(x) atomic_long_add_unless(&(x)->f_count, -1, 1)
882 #define file_count(x) atomic_long_read(&(x)->f_count)
884 #define MAX_NON_LFS ((1UL<<31) - 1)
886 /* Page cache limit. The filesystems should put that into their s_maxbytes
887 limits, otherwise bad things can happen in VM. */
888 #if BITS_PER_LONG==32
889 #define MAX_LFS_FILESIZE (((loff_t)PAGE_SIZE << (BITS_PER_LONG-1))-1)
890 #elif BITS_PER_LONG==64
891 #define MAX_LFS_FILESIZE ((loff_t)0x7fffffffffffffffLL)
896 #define FL_DELEG 4 /* NFSv4 delegation */
897 #define FL_ACCESS 8 /* not trying to lock, just looking */
898 #define FL_EXISTS 16 /* when unlocking, test for existence */
899 #define FL_LEASE 32 /* lease held on this file */
900 #define FL_CLOSE 64 /* unlock on close */
901 #define FL_SLEEP 128 /* A blocking lock */
902 #define FL_DOWNGRADE_PENDING 256 /* Lease is being downgraded */
903 #define FL_UNLOCK_PENDING 512 /* Lease is being broken */
904 #define FL_OFDLCK 1024 /* lock is "owned" by struct file */
905 #define FL_LAYOUT 2048 /* outstanding pNFS layout */
908 * Special return value from posix_lock_file() and vfs_lock_file() for
909 * asynchronous locking.
911 #define FILE_LOCK_DEFERRED 1
913 /* legacy typedef, should eventually be removed */
914 typedef void *fl_owner_t;
918 struct file_lock_operations {
919 void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
920 void (*fl_release_private)(struct file_lock *);
923 struct lock_manager_operations {
924 int (*lm_compare_owner)(struct file_lock *, struct file_lock *);
925 unsigned long (*lm_owner_key)(struct file_lock *);
926 fl_owner_t (*lm_get_owner)(fl_owner_t);
927 void (*lm_put_owner)(fl_owner_t);
928 void (*lm_notify)(struct file_lock *); /* unblock callback */
929 int (*lm_grant)(struct file_lock *, int);
930 bool (*lm_break)(struct file_lock *);
931 int (*lm_change)(struct file_lock *, int, struct list_head *);
932 void (*lm_setup)(struct file_lock *, void **);
935 struct lock_manager {
936 struct list_head list;
938 * NFSv4 and up also want opens blocked during the grace period;
945 void locks_start_grace(struct net *, struct lock_manager *);
946 void locks_end_grace(struct lock_manager *);
947 int locks_in_grace(struct net *);
948 int opens_in_grace(struct net *);
950 /* that will die - we need it for nfs_lock_info */
951 #include <linux/nfs_fs_i.h>
954 * struct file_lock represents a generic "file lock". It's used to represent
955 * POSIX byte range locks, BSD (flock) locks, and leases. It's important to
956 * note that the same struct is used to represent both a request for a lock and
957 * the lock itself, but the same object is never used for both.
959 * FIXME: should we create a separate "struct lock_request" to help distinguish
962 * The varous i_flctx lists are ordered by:
965 * 2) lock range start
968 * Obviously, the last two criteria only matter for POSIX locks.
971 struct file_lock *fl_next; /* singly linked list for this inode */
972 struct list_head fl_list; /* link into file_lock_context */
973 struct hlist_node fl_link; /* node in global lists */
974 struct list_head fl_block; /* circular list of blocked processes */
976 unsigned int fl_flags;
977 unsigned char fl_type;
979 int fl_link_cpu; /* what cpu's list is this on? */
980 struct pid *fl_nspid;
981 wait_queue_head_t fl_wait;
982 struct file *fl_file;
986 struct fasync_struct * fl_fasync; /* for lease break notifications */
987 /* for lease breaks: */
988 unsigned long fl_break_time;
989 unsigned long fl_downgrade_time;
991 const struct file_lock_operations *fl_ops; /* Callbacks for filesystems */
992 const struct lock_manager_operations *fl_lmops; /* Callbacks for lockmanagers */
994 struct nfs_lock_info nfs_fl;
995 struct nfs4_lock_info nfs4_fl;
997 struct list_head link; /* link in AFS vnode's pending_locks list */
998 int state; /* state of grant or error if -ve */
1003 struct file_lock_context {
1004 spinlock_t flc_lock;
1005 struct list_head flc_flock;
1006 struct list_head flc_posix;
1007 struct list_head flc_lease;
1010 /* The following constant reflects the upper bound of the file/locking space */
1012 #define INT_LIMIT(x) (~((x)1 << (sizeof(x)*8 - 1)))
1013 #define OFFSET_MAX INT_LIMIT(loff_t)
1014 #define OFFT_OFFSET_MAX INT_LIMIT(off_t)
1017 #include <linux/fcntl.h>
1019 extern void send_sigio(struct fown_struct *fown, int fd, int band);
1022 * Return the inode to use for locking
1024 * For overlayfs this should be the overlay inode, not the real inode returned
1025 * by file_inode(). For any other fs file_inode(filp) and locks_inode(filp) are
1028 static inline struct inode *locks_inode(const struct file *f)
1030 return f->f_path.dentry->d_inode;
1033 #ifdef CONFIG_FILE_LOCKING
1034 extern int fcntl_getlk(struct file *, unsigned int, struct flock __user *);
1035 extern int fcntl_setlk(unsigned int, struct file *, unsigned int,
1036 struct flock __user *);
1038 #if BITS_PER_LONG == 32
1039 extern int fcntl_getlk64(struct file *, unsigned int, struct flock64 __user *);
1040 extern int fcntl_setlk64(unsigned int, struct file *, unsigned int,
1041 struct flock64 __user *);
1044 extern int fcntl_setlease(unsigned int fd, struct file *filp, long arg);
1045 extern int fcntl_getlease(struct file *filp);
1048 void locks_free_lock_context(struct inode *inode);
1049 void locks_free_lock(struct file_lock *fl);
1050 extern void locks_init_lock(struct file_lock *);
1051 extern struct file_lock * locks_alloc_lock(void);
1052 extern void locks_copy_lock(struct file_lock *, struct file_lock *);
1053 extern void locks_copy_conflock(struct file_lock *, struct file_lock *);
1054 extern void locks_remove_posix(struct file *, fl_owner_t);
1055 extern void locks_remove_file(struct file *);
1056 extern void locks_release_private(struct file_lock *);
1057 extern void posix_test_lock(struct file *, struct file_lock *);
1058 extern int posix_lock_file(struct file *, struct file_lock *, struct file_lock *);
1059 extern int posix_unblock_lock(struct file_lock *);
1060 extern int vfs_test_lock(struct file *, struct file_lock *);
1061 extern int vfs_lock_file(struct file *, unsigned int, struct file_lock *, struct file_lock *);
1062 extern int vfs_cancel_lock(struct file *filp, struct file_lock *fl);
1063 extern int locks_lock_inode_wait(struct inode *inode, struct file_lock *fl);
1064 extern int __break_lease(struct inode *inode, unsigned int flags, unsigned int type);
1065 extern void lease_get_mtime(struct inode *, struct timespec *time);
1066 extern int generic_setlease(struct file *, long, struct file_lock **, void **priv);
1067 extern int vfs_setlease(struct file *, long, struct file_lock **, void **);
1068 extern int lease_modify(struct file_lock *, int, struct list_head *);
1069 struct files_struct;
1070 extern void show_fd_locks(struct seq_file *f,
1071 struct file *filp, struct files_struct *files);
1072 #else /* !CONFIG_FILE_LOCKING */
1073 static inline int fcntl_getlk(struct file *file, unsigned int cmd,
1074 struct flock __user *user)
1079 static inline int fcntl_setlk(unsigned int fd, struct file *file,
1080 unsigned int cmd, struct flock __user *user)
1085 #if BITS_PER_LONG == 32
1086 static inline int fcntl_getlk64(struct file *file, unsigned int cmd,
1087 struct flock64 __user *user)
1092 static inline int fcntl_setlk64(unsigned int fd, struct file *file,
1093 unsigned int cmd, struct flock64 __user *user)
1098 static inline int fcntl_setlease(unsigned int fd, struct file *filp, long arg)
1103 static inline int fcntl_getlease(struct file *filp)
1109 locks_free_lock_context(struct inode *inode)
1113 static inline void locks_init_lock(struct file_lock *fl)
1118 static inline void locks_copy_conflock(struct file_lock *new, struct file_lock *fl)
1123 static inline void locks_copy_lock(struct file_lock *new, struct file_lock *fl)
1128 static inline void locks_remove_posix(struct file *filp, fl_owner_t owner)
1133 static inline void locks_remove_file(struct file *filp)
1138 static inline void posix_test_lock(struct file *filp, struct file_lock *fl)
1143 static inline int posix_lock_file(struct file *filp, struct file_lock *fl,
1144 struct file_lock *conflock)
1149 static inline int posix_unblock_lock(struct file_lock *waiter)
1154 static inline int vfs_test_lock(struct file *filp, struct file_lock *fl)
1159 static inline int vfs_lock_file(struct file *filp, unsigned int cmd,
1160 struct file_lock *fl, struct file_lock *conf)
1165 static inline int vfs_cancel_lock(struct file *filp, struct file_lock *fl)
1170 static inline int locks_lock_inode_wait(struct inode *inode, struct file_lock *fl)
1175 static inline int __break_lease(struct inode *inode, unsigned int mode, unsigned int type)
1180 static inline void lease_get_mtime(struct inode *inode, struct timespec *time)
1185 static inline int generic_setlease(struct file *filp, long arg,
1186 struct file_lock **flp, void **priv)
1191 static inline int vfs_setlease(struct file *filp, long arg,
1192 struct file_lock **lease, void **priv)
1197 static inline int lease_modify(struct file_lock *fl, int arg,
1198 struct list_head *dispose)
1203 struct files_struct;
1204 static inline void show_fd_locks(struct seq_file *f,
1205 struct file *filp, struct files_struct *files) {}
1206 #endif /* !CONFIG_FILE_LOCKING */
1208 static inline struct inode *file_inode(const struct file *f)
1213 static inline struct dentry *file_dentry(const struct file *file)
1215 return d_real(file->f_path.dentry, file_inode(file), 0);
1218 static inline int locks_lock_file_wait(struct file *filp, struct file_lock *fl)
1220 return locks_lock_inode_wait(locks_inode(filp), fl);
1223 struct fasync_struct {
1227 struct fasync_struct *fa_next; /* singly linked list */
1228 struct file *fa_file;
1229 struct rcu_head fa_rcu;
1232 #define FASYNC_MAGIC 0x4601
1234 /* SMP safe fasync helpers: */
1235 extern int fasync_helper(int, struct file *, int, struct fasync_struct **);
1236 extern struct fasync_struct *fasync_insert_entry(int, struct file *, struct fasync_struct **, struct fasync_struct *);
1237 extern int fasync_remove_entry(struct file *, struct fasync_struct **);
1238 extern struct fasync_struct *fasync_alloc(void);
1239 extern void fasync_free(struct fasync_struct *);
1241 /* can be called from interrupts */
1242 extern void kill_fasync(struct fasync_struct **, int, int);
1244 extern void __f_setown(struct file *filp, struct pid *, enum pid_type, int force);
1245 extern void f_setown(struct file *filp, unsigned long arg, int force);
1246 extern void f_delown(struct file *filp);
1247 extern pid_t f_getown(struct file *filp);
1248 extern int send_sigurg(struct fown_struct *fown);
1256 #define MNT_FORCE 0x00000001 /* Attempt to forcibily umount */
1257 #define MNT_DETACH 0x00000002 /* Just detach from the tree */
1258 #define MNT_EXPIRE 0x00000004 /* Mark for expiry */
1259 #define UMOUNT_NOFOLLOW 0x00000008 /* Don't follow symlink on umount */
1260 #define UMOUNT_UNUSED 0x80000000 /* Flag guaranteed to be unused */
1263 #define SB_I_CGROUPWB 0x00000001 /* cgroup-aware writeback enabled */
1264 #define SB_I_NOEXEC 0x00000002 /* Ignore executables on this fs */
1265 #define SB_I_NODEV 0x00000004 /* Ignore devices on this fs */
1267 /* sb->s_iflags to limit user namespace mounts */
1268 #define SB_I_USERNS_VISIBLE 0x00000010 /* fstype already mounted */
1270 /* Possible states of 'frozen' field */
1272 SB_UNFROZEN = 0, /* FS is unfrozen */
1273 SB_FREEZE_WRITE = 1, /* Writes, dir ops, ioctls frozen */
1274 SB_FREEZE_PAGEFAULT = 2, /* Page faults stopped as well */
1275 SB_FREEZE_FS = 3, /* For internal FS use (e.g. to stop
1276 * internal threads if needed) */
1277 SB_FREEZE_COMPLETE = 4, /* ->freeze_fs finished successfully */
1280 #define SB_FREEZE_LEVELS (SB_FREEZE_COMPLETE - 1)
1283 int frozen; /* Is sb frozen? */
1284 wait_queue_head_t wait_unfrozen; /* for get_super_thawed() */
1285 struct percpu_rw_semaphore rw_sem[SB_FREEZE_LEVELS];
1288 struct super_block {
1289 struct list_head s_list; /* Keep this first */
1290 dev_t s_dev; /* search index; _not_ kdev_t */
1291 unsigned char s_blocksize_bits;
1292 unsigned long s_blocksize;
1293 loff_t s_maxbytes; /* Max file size */
1294 struct file_system_type *s_type;
1295 const struct super_operations *s_op;
1296 const struct dquot_operations *dq_op;
1297 const struct quotactl_ops *s_qcop;
1298 const struct export_operations *s_export_op;
1299 unsigned long s_flags;
1300 unsigned long s_iflags; /* internal SB_I_* flags */
1301 unsigned long s_magic;
1302 struct dentry *s_root;
1303 struct rw_semaphore s_umount;
1306 #ifdef CONFIG_SECURITY
1309 const struct xattr_handler **s_xattr;
1311 const struct fscrypt_operations *s_cop;
1313 struct hlist_bl_head s_anon; /* anonymous dentries for (nfs) exporting */
1314 struct list_head s_mounts; /* list of mounts; _not_ for fs use */
1315 struct block_device *s_bdev;
1316 struct backing_dev_info *s_bdi;
1317 struct mtd_info *s_mtd;
1318 struct hlist_node s_instances;
1319 unsigned int s_quota_types; /* Bitmask of supported quota types */
1320 struct quota_info s_dquot; /* Diskquota specific options */
1322 struct sb_writers s_writers;
1324 char s_id[32]; /* Informational name */
1325 u8 s_uuid[16]; /* UUID */
1327 void *s_fs_info; /* Filesystem private info */
1328 unsigned int s_max_links;
1331 /* Granularity of c/m/atime in ns.
1332 Cannot be worse than a second */
1336 * The next field is for VFS *only*. No filesystems have any business
1337 * even looking at it. You had been warned.
1339 struct mutex s_vfs_rename_mutex; /* Kludge */
1342 * Filesystem subtype. If non-empty the filesystem type field
1343 * in /proc/mounts will be "type.subtype"
1348 * Saved mount options for lazy filesystems using
1349 * generic_show_options()
1351 char __rcu *s_options;
1352 const struct dentry_operations *s_d_op; /* default d_op for dentries */
1355 * Saved pool identifier for cleancache (-1 means none)
1357 int cleancache_poolid;
1359 struct shrinker s_shrink; /* per-sb shrinker handle */
1361 /* Number of inodes with nlink == 0 but still referenced */
1362 atomic_long_t s_remove_count;
1364 /* Being remounted read-only */
1365 int s_readonly_remount;
1367 /* AIO completions deferred from interrupt context */
1368 struct workqueue_struct *s_dio_done_wq;
1369 struct hlist_head s_pins;
1372 * Owning user namespace and default context in which to
1373 * interpret filesystem uids, gids, quotas, device nodes,
1374 * xattrs and security labels.
1376 struct user_namespace *s_user_ns;
1379 * Keep the lru lists last in the structure so they always sit on their
1380 * own individual cachelines.
1382 struct list_lru s_dentry_lru ____cacheline_aligned_in_smp;
1383 struct list_lru s_inode_lru ____cacheline_aligned_in_smp;
1384 struct rcu_head rcu;
1385 struct work_struct destroy_work;
1387 struct mutex s_sync_lock; /* sync serialisation lock */
1390 * Indicates how deep in a filesystem stack this SB is
1394 /* s_inode_list_lock protects s_inodes */
1395 spinlock_t s_inode_list_lock ____cacheline_aligned_in_smp;
1396 struct list_head s_inodes; /* all inodes */
1398 spinlock_t s_inode_wblist_lock;
1399 struct list_head s_inodes_wb; /* writeback inodes */
1402 /* Helper functions so that in most cases filesystems will
1403 * not need to deal directly with kuid_t and kgid_t and can
1404 * instead deal with the raw numeric values that are stored
1405 * in the filesystem.
1407 static inline uid_t i_uid_read(const struct inode *inode)
1409 return from_kuid(inode->i_sb->s_user_ns, inode->i_uid);
1412 static inline gid_t i_gid_read(const struct inode *inode)
1414 return from_kgid(inode->i_sb->s_user_ns, inode->i_gid);
1417 static inline void i_uid_write(struct inode *inode, uid_t uid)
1419 inode->i_uid = make_kuid(inode->i_sb->s_user_ns, uid);
1422 static inline void i_gid_write(struct inode *inode, gid_t gid)
1424 inode->i_gid = make_kgid(inode->i_sb->s_user_ns, gid);
1427 extern struct timespec current_fs_time(struct super_block *sb);
1428 extern struct timespec current_time(struct inode *inode);
1431 * Snapshotting support.
1434 void __sb_end_write(struct super_block *sb, int level);
1435 int __sb_start_write(struct super_block *sb, int level, bool wait);
1437 #define __sb_writers_acquired(sb, lev) \
1438 percpu_rwsem_acquire(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1439 #define __sb_writers_release(sb, lev) \
1440 percpu_rwsem_release(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1443 * sb_end_write - drop write access to a superblock
1444 * @sb: the super we wrote to
1446 * Decrement number of writers to the filesystem. Wake up possible waiters
1447 * wanting to freeze the filesystem.
1449 static inline void sb_end_write(struct super_block *sb)
1451 __sb_end_write(sb, SB_FREEZE_WRITE);
1455 * sb_end_pagefault - drop write access to a superblock from a page fault
1456 * @sb: the super we wrote to
1458 * Decrement number of processes handling write page fault to the filesystem.
1459 * Wake up possible waiters wanting to freeze the filesystem.
1461 static inline void sb_end_pagefault(struct super_block *sb)
1463 __sb_end_write(sb, SB_FREEZE_PAGEFAULT);
1467 * sb_end_intwrite - drop write access to a superblock for internal fs purposes
1468 * @sb: the super we wrote to
1470 * Decrement fs-internal number of writers to the filesystem. Wake up possible
1471 * waiters wanting to freeze the filesystem.
1473 static inline void sb_end_intwrite(struct super_block *sb)
1475 __sb_end_write(sb, SB_FREEZE_FS);
1479 * sb_start_write - get write access to a superblock
1480 * @sb: the super we write to
1482 * When a process wants to write data or metadata to a file system (i.e. dirty
1483 * a page or an inode), it should embed the operation in a sb_start_write() -
1484 * sb_end_write() pair to get exclusion against file system freezing. This
1485 * function increments number of writers preventing freezing. If the file
1486 * system is already frozen, the function waits until the file system is
1489 * Since freeze protection behaves as a lock, users have to preserve
1490 * ordering of freeze protection and other filesystem locks. Generally,
1491 * freeze protection should be the outermost lock. In particular, we have:
1494 * -> i_mutex (write path, truncate, directory ops, ...)
1495 * -> s_umount (freeze_super, thaw_super)
1497 static inline void sb_start_write(struct super_block *sb)
1499 __sb_start_write(sb, SB_FREEZE_WRITE, true);
1502 static inline int sb_start_write_trylock(struct super_block *sb)
1504 return __sb_start_write(sb, SB_FREEZE_WRITE, false);
1508 * sb_start_pagefault - get write access to a superblock from a page fault
1509 * @sb: the super we write to
1511 * When a process starts handling write page fault, it should embed the
1512 * operation into sb_start_pagefault() - sb_end_pagefault() pair to get
1513 * exclusion against file system freezing. This is needed since the page fault
1514 * is going to dirty a page. This function increments number of running page
1515 * faults preventing freezing. If the file system is already frozen, the
1516 * function waits until the file system is thawed.
1518 * Since page fault freeze protection behaves as a lock, users have to preserve
1519 * ordering of freeze protection and other filesystem locks. It is advised to
1520 * put sb_start_pagefault() close to mmap_sem in lock ordering. Page fault
1521 * handling code implies lock dependency:
1524 * -> sb_start_pagefault
1526 static inline void sb_start_pagefault(struct super_block *sb)
1528 __sb_start_write(sb, SB_FREEZE_PAGEFAULT, true);
1532 * sb_start_intwrite - get write access to a superblock for internal fs purposes
1533 * @sb: the super we write to
1535 * This is the third level of protection against filesystem freezing. It is
1536 * free for use by a filesystem. The only requirement is that it must rank
1537 * below sb_start_pagefault.
1539 * For example filesystem can call sb_start_intwrite() when starting a
1540 * transaction which somewhat eases handling of freezing for internal sources
1541 * of filesystem changes (internal fs threads, discarding preallocation on file
1544 static inline void sb_start_intwrite(struct super_block *sb)
1546 __sb_start_write(sb, SB_FREEZE_FS, true);
1550 extern bool inode_owner_or_capable(const struct inode *inode);
1553 * VFS helper functions..
1555 extern int vfs_create(struct inode *, struct dentry *, umode_t, bool);
1556 extern int vfs_mkdir(struct inode *, struct dentry *, umode_t);
1557 extern int vfs_mknod(struct inode *, struct dentry *, umode_t, dev_t);
1558 extern int vfs_symlink(struct inode *, struct dentry *, const char *);
1559 extern int vfs_link(struct dentry *, struct inode *, struct dentry *, struct inode **);
1560 extern int vfs_rmdir(struct inode *, struct dentry *);
1561 extern int vfs_unlink(struct inode *, struct dentry *, struct inode **);
1562 extern int vfs_rename(struct inode *, struct dentry *, struct inode *, struct dentry *, struct inode **, unsigned int);
1563 extern int vfs_whiteout(struct inode *, struct dentry *);
1565 extern struct dentry *vfs_tmpfile(struct dentry *dentry, umode_t mode,
1569 * VFS file helper functions.
1571 extern void inode_init_owner(struct inode *inode, const struct inode *dir,
1573 extern bool may_open_dev(const struct path *path);
1575 * VFS FS_IOC_FIEMAP helper definitions.
1577 struct fiemap_extent_info {
1578 unsigned int fi_flags; /* Flags as passed from user */
1579 unsigned int fi_extents_mapped; /* Number of mapped extents */
1580 unsigned int fi_extents_max; /* Size of fiemap_extent array */
1581 struct fiemap_extent __user *fi_extents_start; /* Start of
1582 fiemap_extent array */
1584 int fiemap_fill_next_extent(struct fiemap_extent_info *info, u64 logical,
1585 u64 phys, u64 len, u32 flags);
1586 int fiemap_check_flags(struct fiemap_extent_info *fieinfo, u32 fs_flags);
1591 * NOTE! These match bits 12..15 of stat.st_mode
1592 * (ie "(i_mode >> 12) & 15").
1594 #define DT_UNKNOWN 0
1605 * This is the "filldir" function type, used by readdir() to let
1606 * the kernel specify what kind of dirent layout it wants to have.
1607 * This allows the kernel to read directories into kernel space or
1608 * to have different dirent layouts depending on the binary type.
1611 typedef int (*filldir_t)(struct dir_context *, const char *, int, loff_t, u64,
1614 struct dir_context {
1615 const filldir_t actor;
1619 struct block_device_operations;
1621 /* These macros are for out of kernel modules to test that
1622 * the kernel supports the unlocked_ioctl and compat_ioctl
1623 * fields in struct file_operations. */
1624 #define HAVE_COMPAT_IOCTL 1
1625 #define HAVE_UNLOCKED_IOCTL 1
1628 * These flags let !MMU mmap() govern direct device mapping vs immediate
1629 * copying more easily for MAP_PRIVATE, especially for ROM filesystems.
1631 * NOMMU_MAP_COPY: Copy can be mapped (MAP_PRIVATE)
1632 * NOMMU_MAP_DIRECT: Can be mapped directly (MAP_SHARED)
1633 * NOMMU_MAP_READ: Can be mapped for reading
1634 * NOMMU_MAP_WRITE: Can be mapped for writing
1635 * NOMMU_MAP_EXEC: Can be mapped for execution
1637 #define NOMMU_MAP_COPY 0x00000001
1638 #define NOMMU_MAP_DIRECT 0x00000008
1639 #define NOMMU_MAP_READ VM_MAYREAD
1640 #define NOMMU_MAP_WRITE VM_MAYWRITE
1641 #define NOMMU_MAP_EXEC VM_MAYEXEC
1643 #define NOMMU_VMFLAGS \
1644 (NOMMU_MAP_READ | NOMMU_MAP_WRITE | NOMMU_MAP_EXEC)
1649 struct file_operations {
1650 struct module *owner;
1651 loff_t (*llseek) (struct file *, loff_t, int);
1652 ssize_t (*read) (struct file *, char __user *, size_t, loff_t *);
1653 ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *);
1654 ssize_t (*read_iter) (struct kiocb *, struct iov_iter *);
1655 ssize_t (*write_iter) (struct kiocb *, struct iov_iter *);
1656 int (*iterate) (struct file *, struct dir_context *);
1657 int (*iterate_shared) (struct file *, struct dir_context *);
1658 unsigned int (*poll) (struct file *, struct poll_table_struct *);
1659 long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long);
1660 long (*compat_ioctl) (struct file *, unsigned int, unsigned long);
1661 int (*mmap) (struct file *, struct vm_area_struct *);
1662 int (*open) (struct inode *, struct file *);
1663 int (*flush) (struct file *, fl_owner_t id);
1664 int (*release) (struct inode *, struct file *);
1665 int (*fsync) (struct file *, loff_t, loff_t, int datasync);
1666 int (*fasync) (int, struct file *, int);
1667 int (*lock) (struct file *, int, struct file_lock *);
1668 ssize_t (*sendpage) (struct file *, struct page *, int, size_t, loff_t *, int);
1669 unsigned long (*get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
1670 int (*check_flags)(int);
1671 int (*flock) (struct file *, int, struct file_lock *);
1672 ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, loff_t *, size_t, unsigned int);
1673 ssize_t (*splice_read)(struct file *, loff_t *, struct pipe_inode_info *, size_t, unsigned int);
1674 int (*setlease)(struct file *, long, struct file_lock **, void **);
1675 long (*fallocate)(struct file *file, int mode, loff_t offset,
1677 void (*show_fdinfo)(struct seq_file *m, struct file *f);
1679 unsigned (*mmap_capabilities)(struct file *);
1681 ssize_t (*copy_file_range)(struct file *, loff_t, struct file *,
1682 loff_t, size_t, unsigned int);
1683 int (*clone_file_range)(struct file *, loff_t, struct file *, loff_t,
1685 ssize_t (*dedupe_file_range)(struct file *, u64, u64, struct file *,
1689 struct inode_operations {
1690 struct dentry * (*lookup) (struct inode *,struct dentry *, unsigned int);
1691 const char * (*get_link) (struct dentry *, struct inode *, struct delayed_call *);
1692 int (*permission) (struct inode *, int);
1693 struct posix_acl * (*get_acl)(struct inode *, int);
1695 int (*readlink) (struct dentry *, char __user *,int);
1697 int (*create) (struct inode *,struct dentry *, umode_t, bool);
1698 int (*link) (struct dentry *,struct inode *,struct dentry *);
1699 int (*unlink) (struct inode *,struct dentry *);
1700 int (*symlink) (struct inode *,struct dentry *,const char *);
1701 int (*mkdir) (struct inode *,struct dentry *,umode_t);
1702 int (*rmdir) (struct inode *,struct dentry *);
1703 int (*mknod) (struct inode *,struct dentry *,umode_t,dev_t);
1704 int (*rename) (struct inode *, struct dentry *,
1705 struct inode *, struct dentry *, unsigned int);
1706 int (*setattr) (struct dentry *, struct iattr *);
1707 int (*getattr) (struct vfsmount *mnt, struct dentry *, struct kstat *);
1708 ssize_t (*listxattr) (struct dentry *, char *, size_t);
1709 int (*fiemap)(struct inode *, struct fiemap_extent_info *, u64 start,
1711 int (*update_time)(struct inode *, struct timespec *, int);
1712 int (*atomic_open)(struct inode *, struct dentry *,
1713 struct file *, unsigned open_flag,
1714 umode_t create_mode, int *opened);
1715 int (*tmpfile) (struct inode *, struct dentry *, umode_t);
1716 int (*set_acl)(struct inode *, struct posix_acl *, int);
1717 } ____cacheline_aligned;
1719 static inline ssize_t call_read_iter(struct file *file, struct kiocb *kio,
1720 struct iov_iter *iter)
1722 return file->f_op->read_iter(kio, iter);
1725 static inline ssize_t call_write_iter(struct file *file, struct kiocb *kio,
1726 struct iov_iter *iter)
1728 return file->f_op->write_iter(kio, iter);
1731 static inline int call_mmap(struct file *file, struct vm_area_struct *vma)
1733 return file->f_op->mmap(file, vma);
1736 static inline int call_fsync(struct file *file, loff_t start, loff_t end,
1739 return file->f_op->fsync(file, start, end, datasync);
1742 ssize_t rw_copy_check_uvector(int type, const struct iovec __user * uvector,
1743 unsigned long nr_segs, unsigned long fast_segs,
1744 struct iovec *fast_pointer,
1745 struct iovec **ret_pointer);
1747 extern ssize_t __vfs_read(struct file *, char __user *, size_t, loff_t *);
1748 extern ssize_t __vfs_write(struct file *, const char __user *, size_t, loff_t *);
1749 extern ssize_t vfs_read(struct file *, char __user *, size_t, loff_t *);
1750 extern ssize_t vfs_write(struct file *, const char __user *, size_t, loff_t *);
1751 extern ssize_t vfs_readv(struct file *, const struct iovec __user *,
1752 unsigned long, loff_t *, int);
1753 extern ssize_t vfs_writev(struct file *, const struct iovec __user *,
1754 unsigned long, loff_t *, int);
1755 extern ssize_t vfs_copy_file_range(struct file *, loff_t , struct file *,
1756 loff_t, size_t, unsigned int);
1757 extern int vfs_clone_file_prep_inodes(struct inode *inode_in, loff_t pos_in,
1758 struct inode *inode_out, loff_t pos_out,
1759 u64 *len, bool is_dedupe);
1760 extern int vfs_clone_file_range(struct file *file_in, loff_t pos_in,
1761 struct file *file_out, loff_t pos_out, u64 len);
1762 extern int vfs_dedupe_file_range_compare(struct inode *src, loff_t srcoff,
1763 struct inode *dest, loff_t destoff,
1764 loff_t len, bool *is_same);
1765 extern int vfs_dedupe_file_range(struct file *file,
1766 struct file_dedupe_range *same);
1768 struct super_operations {
1769 struct inode *(*alloc_inode)(struct super_block *sb);
1770 void (*destroy_inode)(struct inode *);
1772 void (*dirty_inode) (struct inode *, int flags);
1773 int (*write_inode) (struct inode *, struct writeback_control *wbc);
1774 int (*drop_inode) (struct inode *);
1775 void (*evict_inode) (struct inode *);
1776 void (*put_super) (struct super_block *);
1777 int (*sync_fs)(struct super_block *sb, int wait);
1778 int (*freeze_super) (struct super_block *);
1779 int (*freeze_fs) (struct super_block *);
1780 int (*thaw_super) (struct super_block *);
1781 int (*unfreeze_fs) (struct super_block *);
1782 int (*statfs) (struct dentry *, struct kstatfs *);
1783 int (*remount_fs) (struct super_block *, int *, char *);
1784 void (*umount_begin) (struct super_block *);
1786 int (*show_options)(struct seq_file *, struct dentry *);
1787 int (*show_devname)(struct seq_file *, struct dentry *);
1788 int (*show_path)(struct seq_file *, struct dentry *);
1789 int (*show_stats)(struct seq_file *, struct dentry *);
1791 ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t);
1792 ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t);
1793 struct dquot **(*get_dquots)(struct inode *);
1795 int (*bdev_try_to_free_page)(struct super_block*, struct page*, gfp_t);
1796 long (*nr_cached_objects)(struct super_block *,
1797 struct shrink_control *);
1798 long (*free_cached_objects)(struct super_block *,
1799 struct shrink_control *);
1803 * Inode flags - they have no relation to superblock flags now
1805 #define S_SYNC 1 /* Writes are synced at once */
1806 #define S_NOATIME 2 /* Do not update access times */
1807 #define S_APPEND 4 /* Append-only file */
1808 #define S_IMMUTABLE 8 /* Immutable file */
1809 #define S_DEAD 16 /* removed, but still open directory */
1810 #define S_NOQUOTA 32 /* Inode is not counted to quota */
1811 #define S_DIRSYNC 64 /* Directory modifications are synchronous */
1812 #define S_NOCMTIME 128 /* Do not update file c/mtime */
1813 #define S_SWAPFILE 256 /* Do not truncate: swapon got its bmaps */
1814 #define S_PRIVATE 512 /* Inode is fs-internal */
1815 #define S_IMA 1024 /* Inode has an associated IMA struct */
1816 #define S_AUTOMOUNT 2048 /* Automount/referral quasi-directory */
1817 #define S_NOSEC 4096 /* no suid or xattr security attributes */
1818 #ifdef CONFIG_FS_DAX
1819 #define S_DAX 8192 /* Direct Access, avoiding the page cache */
1821 #define S_DAX 0 /* Make all the DAX code disappear */
1825 * Note that nosuid etc flags are inode-specific: setting some file-system
1826 * flags just means all the inodes inherit those flags by default. It might be
1827 * possible to override it selectively if you really wanted to with some
1828 * ioctl() that is not currently implemented.
1830 * Exception: MS_RDONLY is always applied to the entire file system.
1832 * Unfortunately, it is possible to change a filesystems flags with it mounted
1833 * with files in use. This means that all of the inodes will not have their
1834 * i_flags updated. Hence, i_flags no longer inherit the superblock mount
1835 * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org
1837 #define __IS_FLG(inode, flg) ((inode)->i_sb->s_flags & (flg))
1839 #define IS_RDONLY(inode) ((inode)->i_sb->s_flags & MS_RDONLY)
1840 #define IS_SYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS) || \
1841 ((inode)->i_flags & S_SYNC))
1842 #define IS_DIRSYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS|MS_DIRSYNC) || \
1843 ((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
1844 #define IS_MANDLOCK(inode) __IS_FLG(inode, MS_MANDLOCK)
1845 #define IS_NOATIME(inode) __IS_FLG(inode, MS_RDONLY|MS_NOATIME)
1846 #define IS_I_VERSION(inode) __IS_FLG(inode, MS_I_VERSION)
1848 #define IS_NOQUOTA(inode) ((inode)->i_flags & S_NOQUOTA)
1849 #define IS_APPEND(inode) ((inode)->i_flags & S_APPEND)
1850 #define IS_IMMUTABLE(inode) ((inode)->i_flags & S_IMMUTABLE)
1851 #define IS_POSIXACL(inode) __IS_FLG(inode, MS_POSIXACL)
1853 #define IS_DEADDIR(inode) ((inode)->i_flags & S_DEAD)
1854 #define IS_NOCMTIME(inode) ((inode)->i_flags & S_NOCMTIME)
1855 #define IS_SWAPFILE(inode) ((inode)->i_flags & S_SWAPFILE)
1856 #define IS_PRIVATE(inode) ((inode)->i_flags & S_PRIVATE)
1857 #define IS_IMA(inode) ((inode)->i_flags & S_IMA)
1858 #define IS_AUTOMOUNT(inode) ((inode)->i_flags & S_AUTOMOUNT)
1859 #define IS_NOSEC(inode) ((inode)->i_flags & S_NOSEC)
1860 #define IS_DAX(inode) ((inode)->i_flags & S_DAX)
1862 #define IS_WHITEOUT(inode) (S_ISCHR(inode->i_mode) && \
1863 (inode)->i_rdev == WHITEOUT_DEV)
1865 static inline bool HAS_UNMAPPED_ID(struct inode *inode)
1867 return !uid_valid(inode->i_uid) || !gid_valid(inode->i_gid);
1871 * Inode state bits. Protected by inode->i_lock
1873 * Three bits determine the dirty state of the inode, I_DIRTY_SYNC,
1874 * I_DIRTY_DATASYNC and I_DIRTY_PAGES.
1876 * Four bits define the lifetime of an inode. Initially, inodes are I_NEW,
1877 * until that flag is cleared. I_WILL_FREE, I_FREEING and I_CLEAR are set at
1878 * various stages of removing an inode.
1880 * Two bits are used for locking and completion notification, I_NEW and I_SYNC.
1882 * I_DIRTY_SYNC Inode is dirty, but doesn't have to be written on
1883 * fdatasync(). i_atime is the usual cause.
1884 * I_DIRTY_DATASYNC Data-related inode changes pending. We keep track of
1885 * these changes separately from I_DIRTY_SYNC so that we
1886 * don't have to write inode on fdatasync() when only
1887 * mtime has changed in it.
1888 * I_DIRTY_PAGES Inode has dirty pages. Inode itself may be clean.
1889 * I_NEW Serves as both a mutex and completion notification.
1890 * New inodes set I_NEW. If two processes both create
1891 * the same inode, one of them will release its inode and
1892 * wait for I_NEW to be released before returning.
1893 * Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can
1894 * also cause waiting on I_NEW, without I_NEW actually
1895 * being set. find_inode() uses this to prevent returning
1896 * nearly-dead inodes.
1897 * I_WILL_FREE Must be set when calling write_inode_now() if i_count
1898 * is zero. I_FREEING must be set when I_WILL_FREE is
1900 * I_FREEING Set when inode is about to be freed but still has dirty
1901 * pages or buffers attached or the inode itself is still
1903 * I_CLEAR Added by clear_inode(). In this state the inode is
1904 * clean and can be destroyed. Inode keeps I_FREEING.
1906 * Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are
1907 * prohibited for many purposes. iget() must wait for
1908 * the inode to be completely released, then create it
1909 * anew. Other functions will just ignore such inodes,
1910 * if appropriate. I_NEW is used for waiting.
1912 * I_SYNC Writeback of inode is running. The bit is set during
1913 * data writeback, and cleared with a wakeup on the bit
1914 * address once it is done. The bit is also used to pin
1915 * the inode in memory for flusher thread.
1917 * I_REFERENCED Marks the inode as recently references on the LRU list.
1919 * I_DIO_WAKEUP Never set. Only used as a key for wait_on_bit().
1921 * I_WB_SWITCH Cgroup bdi_writeback switching in progress. Used to
1922 * synchronize competing switching instances and to tell
1923 * wb stat updates to grab mapping->tree_lock. See
1924 * inode_switch_wb_work_fn() for details.
1926 * Q: What is the difference between I_WILL_FREE and I_FREEING?
1928 #define I_DIRTY_SYNC (1 << 0)
1929 #define I_DIRTY_DATASYNC (1 << 1)
1930 #define I_DIRTY_PAGES (1 << 2)
1932 #define I_NEW (1 << __I_NEW)
1933 #define I_WILL_FREE (1 << 4)
1934 #define I_FREEING (1 << 5)
1935 #define I_CLEAR (1 << 6)
1937 #define I_SYNC (1 << __I_SYNC)
1938 #define I_REFERENCED (1 << 8)
1939 #define __I_DIO_WAKEUP 9
1940 #define I_DIO_WAKEUP (1 << __I_DIO_WAKEUP)
1941 #define I_LINKABLE (1 << 10)
1942 #define I_DIRTY_TIME (1 << 11)
1943 #define __I_DIRTY_TIME_EXPIRED 12
1944 #define I_DIRTY_TIME_EXPIRED (1 << __I_DIRTY_TIME_EXPIRED)
1945 #define I_WB_SWITCH (1 << 13)
1947 #define I_DIRTY (I_DIRTY_SYNC | I_DIRTY_DATASYNC | I_DIRTY_PAGES)
1948 #define I_DIRTY_ALL (I_DIRTY | I_DIRTY_TIME)
1950 extern void __mark_inode_dirty(struct inode *, int);
1951 static inline void mark_inode_dirty(struct inode *inode)
1953 __mark_inode_dirty(inode, I_DIRTY);
1956 static inline void mark_inode_dirty_sync(struct inode *inode)
1958 __mark_inode_dirty(inode, I_DIRTY_SYNC);
1961 extern void inc_nlink(struct inode *inode);
1962 extern void drop_nlink(struct inode *inode);
1963 extern void clear_nlink(struct inode *inode);
1964 extern void set_nlink(struct inode *inode, unsigned int nlink);
1966 static inline void inode_inc_link_count(struct inode *inode)
1969 mark_inode_dirty(inode);
1972 static inline void inode_dec_link_count(struct inode *inode)
1975 mark_inode_dirty(inode);
1979 * inode_inc_iversion - increments i_version
1980 * @inode: inode that need to be updated
1982 * Every time the inode is modified, the i_version field will be incremented.
1983 * The filesystem has to be mounted with i_version flag
1986 static inline void inode_inc_iversion(struct inode *inode)
1988 spin_lock(&inode->i_lock);
1990 spin_unlock(&inode->i_lock);
1993 enum file_time_flags {
2000 extern void touch_atime(const struct path *);
2001 static inline void file_accessed(struct file *file)
2003 if (!(file->f_flags & O_NOATIME))
2004 touch_atime(&file->f_path);
2007 int sync_inode(struct inode *inode, struct writeback_control *wbc);
2008 int sync_inode_metadata(struct inode *inode, int wait);
2010 struct file_system_type {
2013 #define FS_REQUIRES_DEV 1
2014 #define FS_BINARY_MOUNTDATA 2
2015 #define FS_HAS_SUBTYPE 4
2016 #define FS_USERNS_MOUNT 8 /* Can be mounted by userns root */
2017 #define FS_RENAME_DOES_D_MOVE 32768 /* FS will handle d_move() during rename() internally. */
2018 struct dentry *(*mount) (struct file_system_type *, int,
2019 const char *, void *);
2020 void (*kill_sb) (struct super_block *);
2021 struct module *owner;
2022 struct file_system_type * next;
2023 struct hlist_head fs_supers;
2025 struct lock_class_key s_lock_key;
2026 struct lock_class_key s_umount_key;
2027 struct lock_class_key s_vfs_rename_key;
2028 struct lock_class_key s_writers_key[SB_FREEZE_LEVELS];
2030 struct lock_class_key i_lock_key;
2031 struct lock_class_key i_mutex_key;
2032 struct lock_class_key i_mutex_dir_key;
2035 #define MODULE_ALIAS_FS(NAME) MODULE_ALIAS("fs-" NAME)
2037 extern struct dentry *mount_ns(struct file_system_type *fs_type,
2038 int flags, void *data, void *ns, struct user_namespace *user_ns,
2039 int (*fill_super)(struct super_block *, void *, int));
2040 extern struct dentry *mount_bdev(struct file_system_type *fs_type,
2041 int flags, const char *dev_name, void *data,
2042 int (*fill_super)(struct super_block *, void *, int));
2043 extern struct dentry *mount_single(struct file_system_type *fs_type,
2044 int flags, void *data,
2045 int (*fill_super)(struct super_block *, void *, int));
2046 extern struct dentry *mount_nodev(struct file_system_type *fs_type,
2047 int flags, void *data,
2048 int (*fill_super)(struct super_block *, void *, int));
2049 extern struct dentry *mount_subtree(struct vfsmount *mnt, const char *path);
2050 void generic_shutdown_super(struct super_block *sb);
2051 void kill_block_super(struct super_block *sb);
2052 void kill_anon_super(struct super_block *sb);
2053 void kill_litter_super(struct super_block *sb);
2054 void deactivate_super(struct super_block *sb);
2055 void deactivate_locked_super(struct super_block *sb);
2056 int set_anon_super(struct super_block *s, void *data);
2057 int get_anon_bdev(dev_t *);
2058 void free_anon_bdev(dev_t);
2059 struct super_block *sget_userns(struct file_system_type *type,
2060 int (*test)(struct super_block *,void *),
2061 int (*set)(struct super_block *,void *),
2062 int flags, struct user_namespace *user_ns,
2064 struct super_block *sget(struct file_system_type *type,
2065 int (*test)(struct super_block *,void *),
2066 int (*set)(struct super_block *,void *),
2067 int flags, void *data);
2068 extern struct dentry *mount_pseudo_xattr(struct file_system_type *, char *,
2069 const struct super_operations *ops,
2070 const struct xattr_handler **xattr,
2071 const struct dentry_operations *dops,
2074 static inline struct dentry *
2075 mount_pseudo(struct file_system_type *fs_type, char *name,
2076 const struct super_operations *ops,
2077 const struct dentry_operations *dops, unsigned long magic)
2079 return mount_pseudo_xattr(fs_type, name, ops, NULL, dops, magic);
2082 /* Alas, no aliases. Too much hassle with bringing module.h everywhere */
2083 #define fops_get(fops) \
2084 (((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
2085 #define fops_put(fops) \
2086 do { if (fops) module_put((fops)->owner); } while(0)
2088 * This one is to be used *ONLY* from ->open() instances.
2089 * fops must be non-NULL, pinned down *and* module dependencies
2090 * should be sufficient to pin the caller down as well.
2092 #define replace_fops(f, fops) \
2094 struct file *__file = (f); \
2095 fops_put(__file->f_op); \
2096 BUG_ON(!(__file->f_op = (fops))); \
2099 extern int register_filesystem(struct file_system_type *);
2100 extern int unregister_filesystem(struct file_system_type *);
2101 extern struct vfsmount *kern_mount_data(struct file_system_type *, void *data);
2102 #define kern_mount(type) kern_mount_data(type, NULL)
2103 extern void kern_unmount(struct vfsmount *mnt);
2104 extern int may_umount_tree(struct vfsmount *);
2105 extern int may_umount(struct vfsmount *);
2106 extern long do_mount(const char *, const char __user *,
2107 const char *, unsigned long, void *);
2108 extern struct vfsmount *collect_mounts(const struct path *);
2109 extern void drop_collected_mounts(struct vfsmount *);
2110 extern int iterate_mounts(int (*)(struct vfsmount *, void *), void *,
2112 extern int vfs_statfs(const struct path *, struct kstatfs *);
2113 extern int user_statfs(const char __user *, struct kstatfs *);
2114 extern int fd_statfs(int, struct kstatfs *);
2115 extern int vfs_ustat(dev_t, struct kstatfs *);
2116 extern int freeze_super(struct super_block *super);
2117 extern int thaw_super(struct super_block *super);
2118 extern bool our_mnt(struct vfsmount *mnt);
2120 extern int current_umask(void);
2122 extern void ihold(struct inode * inode);
2123 extern void iput(struct inode *);
2124 extern int generic_update_time(struct inode *, struct timespec *, int);
2127 extern struct kobject *fs_kobj;
2129 #define MAX_RW_COUNT (INT_MAX & PAGE_MASK)
2131 #ifdef CONFIG_MANDATORY_FILE_LOCKING
2132 extern int locks_mandatory_locked(struct file *);
2133 extern int locks_mandatory_area(struct inode *, struct file *, loff_t, loff_t, unsigned char);
2136 * Candidates for mandatory locking have the setgid bit set
2137 * but no group execute bit - an otherwise meaningless combination.
2140 static inline int __mandatory_lock(struct inode *ino)
2142 return (ino->i_mode & (S_ISGID | S_IXGRP)) == S_ISGID;
2146 * ... and these candidates should be on MS_MANDLOCK mounted fs,
2147 * otherwise these will be advisory locks
2150 static inline int mandatory_lock(struct inode *ino)
2152 return IS_MANDLOCK(ino) && __mandatory_lock(ino);
2155 static inline int locks_verify_locked(struct file *file)
2157 if (mandatory_lock(locks_inode(file)))
2158 return locks_mandatory_locked(file);
2162 static inline int locks_verify_truncate(struct inode *inode,
2166 if (!inode->i_flctx || !mandatory_lock(inode))
2169 if (size < inode->i_size) {
2170 return locks_mandatory_area(inode, f, size, inode->i_size - 1,
2173 return locks_mandatory_area(inode, f, inode->i_size, size - 1,
2178 #else /* !CONFIG_MANDATORY_FILE_LOCKING */
2180 static inline int locks_mandatory_locked(struct file *file)
2185 static inline int locks_mandatory_area(struct inode *inode, struct file *filp,
2186 loff_t start, loff_t end, unsigned char type)
2191 static inline int __mandatory_lock(struct inode *inode)
2196 static inline int mandatory_lock(struct inode *inode)
2201 static inline int locks_verify_locked(struct file *file)
2206 static inline int locks_verify_truncate(struct inode *inode, struct file *filp,
2212 #endif /* CONFIG_MANDATORY_FILE_LOCKING */
2215 #ifdef CONFIG_FILE_LOCKING
2216 static inline int break_lease(struct inode *inode, unsigned int mode)
2219 * Since this check is lockless, we must ensure that any refcounts
2220 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2221 * could end up racing with tasks trying to set a new lease on this
2225 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2226 return __break_lease(inode, mode, FL_LEASE);
2230 static inline int break_deleg(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_DELEG);
2244 static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode)
2248 ret = break_deleg(inode, O_WRONLY|O_NONBLOCK);
2249 if (ret == -EWOULDBLOCK && delegated_inode) {
2250 *delegated_inode = inode;
2256 static inline int break_deleg_wait(struct inode **delegated_inode)
2260 ret = break_deleg(*delegated_inode, O_WRONLY);
2261 iput(*delegated_inode);
2262 *delegated_inode = NULL;
2266 static inline int break_layout(struct inode *inode, bool wait)
2269 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2270 return __break_lease(inode,
2271 wait ? O_WRONLY : O_WRONLY | O_NONBLOCK,
2276 #else /* !CONFIG_FILE_LOCKING */
2277 static inline int break_lease(struct inode *inode, unsigned int mode)
2282 static inline int break_deleg(struct inode *inode, unsigned int mode)
2287 static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode)
2292 static inline int break_deleg_wait(struct inode **delegated_inode)
2298 static inline int break_layout(struct inode *inode, bool wait)
2303 #endif /* CONFIG_FILE_LOCKING */
2308 const char *name; /* pointer to actual string */
2309 const __user char *uptr; /* original userland pointer */
2310 struct audit_names *aname;
2315 extern long vfs_truncate(const struct path *, loff_t);
2316 extern int do_truncate(struct dentry *, loff_t start, unsigned int time_attrs,
2318 extern int vfs_fallocate(struct file *file, int mode, loff_t offset,
2320 extern long do_sys_open(int dfd, const char __user *filename, int flags,
2322 extern struct file *file_open_name(struct filename *, int, umode_t);
2323 extern struct file *filp_open(const char *, int, umode_t);
2324 extern struct file *file_open_root(struct dentry *, struct vfsmount *,
2325 const char *, int, umode_t);
2326 extern struct file * dentry_open(const struct path *, int, const struct cred *);
2327 extern int filp_close(struct file *, fl_owner_t id);
2329 extern struct filename *getname_flags(const char __user *, int, int *);
2330 extern struct filename *getname(const char __user *);
2331 extern struct filename *getname_kernel(const char *);
2332 extern void putname(struct filename *name);
2338 extern int finish_open(struct file *file, struct dentry *dentry,
2339 int (*open)(struct inode *, struct file *),
2341 extern int finish_no_open(struct file *file, struct dentry *dentry);
2345 extern int ioctl_preallocate(struct file *filp, void __user *argp);
2348 extern void __init vfs_caches_init_early(void);
2349 extern void __init vfs_caches_init(void);
2351 extern struct kmem_cache *names_cachep;
2353 #define __getname() kmem_cache_alloc(names_cachep, GFP_KERNEL)
2354 #define __putname(name) kmem_cache_free(names_cachep, (void *)(name))
2357 extern int register_blkdev(unsigned int, const char *);
2358 extern void unregister_blkdev(unsigned int, const char *);
2359 extern void bdev_unhash_inode(dev_t dev);
2360 extern struct block_device *bdget(dev_t);
2361 extern struct block_device *bdgrab(struct block_device *bdev);
2362 extern void bd_set_size(struct block_device *, loff_t size);
2363 extern void bd_forget(struct inode *inode);
2364 extern void bdput(struct block_device *);
2365 extern void invalidate_bdev(struct block_device *);
2366 extern void iterate_bdevs(void (*)(struct block_device *, void *), void *);
2367 extern int sync_blockdev(struct block_device *bdev);
2368 extern void kill_bdev(struct block_device *);
2369 extern struct super_block *freeze_bdev(struct block_device *);
2370 extern void emergency_thaw_all(void);
2371 extern int thaw_bdev(struct block_device *bdev, struct super_block *sb);
2372 extern int fsync_bdev(struct block_device *);
2374 extern struct super_block *blockdev_superblock;
2376 static inline bool sb_is_blkdev_sb(struct super_block *sb)
2378 return sb == blockdev_superblock;
2381 static inline void bd_forget(struct inode *inode) {}
2382 static inline int sync_blockdev(struct block_device *bdev) { return 0; }
2383 static inline void kill_bdev(struct block_device *bdev) {}
2384 static inline void invalidate_bdev(struct block_device *bdev) {}
2386 static inline struct super_block *freeze_bdev(struct block_device *sb)
2391 static inline int thaw_bdev(struct block_device *bdev, struct super_block *sb)
2396 static inline void iterate_bdevs(void (*f)(struct block_device *, void *), void *arg)
2400 static inline bool sb_is_blkdev_sb(struct super_block *sb)
2405 extern int sync_filesystem(struct super_block *);
2406 extern const struct file_operations def_blk_fops;
2407 extern const struct file_operations def_chr_fops;
2409 extern int ioctl_by_bdev(struct block_device *, unsigned, unsigned long);
2410 extern int blkdev_ioctl(struct block_device *, fmode_t, unsigned, unsigned long);
2411 extern long compat_blkdev_ioctl(struct file *, unsigned, unsigned long);
2412 extern int blkdev_get(struct block_device *bdev, fmode_t mode, void *holder);
2413 extern struct block_device *blkdev_get_by_path(const char *path, fmode_t mode,
2415 extern struct block_device *blkdev_get_by_dev(dev_t dev, fmode_t mode,
2417 extern void blkdev_put(struct block_device *bdev, fmode_t mode);
2418 extern int __blkdev_reread_part(struct block_device *bdev);
2419 extern int blkdev_reread_part(struct block_device *bdev);
2422 extern int bd_link_disk_holder(struct block_device *bdev, struct gendisk *disk);
2423 extern void bd_unlink_disk_holder(struct block_device *bdev,
2424 struct gendisk *disk);
2426 static inline int bd_link_disk_holder(struct block_device *bdev,
2427 struct gendisk *disk)
2431 static inline void bd_unlink_disk_holder(struct block_device *bdev,
2432 struct gendisk *disk)
2439 #define CHRDEV_MAJOR_HASH_SIZE 255
2440 /* Marks the bottom of the first segment of free char majors */
2441 #define CHRDEV_MAJOR_DYN_END 234
2442 extern int alloc_chrdev_region(dev_t *, unsigned, unsigned, const char *);
2443 extern int register_chrdev_region(dev_t, unsigned, const char *);
2444 extern int __register_chrdev(unsigned int major, unsigned int baseminor,
2445 unsigned int count, const char *name,
2446 const struct file_operations *fops);
2447 extern void __unregister_chrdev(unsigned int major, unsigned int baseminor,
2448 unsigned int count, const char *name);
2449 extern void unregister_chrdev_region(dev_t, unsigned);
2450 extern void chrdev_show(struct seq_file *,off_t);
2452 static inline int register_chrdev(unsigned int major, const char *name,
2453 const struct file_operations *fops)
2455 return __register_chrdev(major, 0, 256, name, fops);
2458 static inline void unregister_chrdev(unsigned int major, const char *name)
2460 __unregister_chrdev(major, 0, 256, name);
2463 /* fs/block_dev.c */
2464 #define BDEVNAME_SIZE 32 /* Largest string for a blockdev identifier */
2465 #define BDEVT_SIZE 10 /* Largest string for MAJ:MIN for blkdev */
2468 #define BLKDEV_MAJOR_HASH_SIZE 255
2469 extern const char *__bdevname(dev_t, char *buffer);
2470 extern const char *bdevname(struct block_device *bdev, char *buffer);
2471 extern struct block_device *lookup_bdev(const char *);
2472 extern void blkdev_show(struct seq_file *,off_t);
2475 #define BLKDEV_MAJOR_HASH_SIZE 0
2478 extern void init_special_inode(struct inode *, umode_t, dev_t);
2480 /* Invalid inode operations -- fs/bad_inode.c */
2481 extern void make_bad_inode(struct inode *);
2482 extern bool is_bad_inode(struct inode *);
2485 extern void check_disk_size_change(struct gendisk *disk,
2486 struct block_device *bdev);
2487 extern int revalidate_disk(struct gendisk *);
2488 extern int check_disk_change(struct block_device *);
2489 extern int __invalidate_device(struct block_device *, bool);
2490 extern int invalidate_partition(struct gendisk *, int);
2492 unsigned long invalidate_mapping_pages(struct address_space *mapping,
2493 pgoff_t start, pgoff_t end);
2495 static inline void invalidate_remote_inode(struct inode *inode)
2497 if (S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) ||
2498 S_ISLNK(inode->i_mode))
2499 invalidate_mapping_pages(inode->i_mapping, 0, -1);
2501 extern int invalidate_inode_pages2(struct address_space *mapping);
2502 extern int invalidate_inode_pages2_range(struct address_space *mapping,
2503 pgoff_t start, pgoff_t end);
2504 extern int write_inode_now(struct inode *, int);
2505 extern int filemap_fdatawrite(struct address_space *);
2506 extern int filemap_flush(struct address_space *);
2507 extern int filemap_fdatawait(struct address_space *);
2508 extern void filemap_fdatawait_keep_errors(struct address_space *);
2509 extern int filemap_fdatawait_range(struct address_space *, loff_t lstart,
2511 extern int filemap_write_and_wait(struct address_space *mapping);
2512 extern int filemap_write_and_wait_range(struct address_space *mapping,
2513 loff_t lstart, loff_t lend);
2514 extern int __filemap_fdatawrite_range(struct address_space *mapping,
2515 loff_t start, loff_t end, int sync_mode);
2516 extern int filemap_fdatawrite_range(struct address_space *mapping,
2517 loff_t start, loff_t end);
2518 extern int filemap_check_errors(struct address_space *mapping);
2520 extern int vfs_fsync_range(struct file *file, loff_t start, loff_t end,
2522 extern int vfs_fsync(struct file *file, int datasync);
2525 * Sync the bytes written if this was a synchronous write. Expect ki_pos
2526 * to already be updated for the write, and will return either the amount
2527 * of bytes passed in, or an error if syncing the file failed.
2529 static inline ssize_t generic_write_sync(struct kiocb *iocb, ssize_t count)
2531 if (iocb->ki_flags & IOCB_DSYNC) {
2532 int ret = vfs_fsync_range(iocb->ki_filp,
2533 iocb->ki_pos - count, iocb->ki_pos - 1,
2534 (iocb->ki_flags & IOCB_SYNC) ? 0 : 1);
2542 extern void emergency_sync(void);
2543 extern void emergency_remount(void);
2545 extern sector_t bmap(struct inode *, sector_t);
2547 extern int notify_change(struct dentry *, struct iattr *, struct inode **);
2548 extern int inode_permission(struct inode *, int);
2549 extern int __inode_permission(struct inode *, int);
2550 extern int generic_permission(struct inode *, int);
2551 extern int __check_sticky(struct inode *dir, struct inode *inode);
2553 static inline bool execute_ok(struct inode *inode)
2555 return (inode->i_mode & S_IXUGO) || S_ISDIR(inode->i_mode);
2558 static inline void file_start_write(struct file *file)
2560 if (!S_ISREG(file_inode(file)->i_mode))
2562 __sb_start_write(file_inode(file)->i_sb, SB_FREEZE_WRITE, true);
2565 static inline bool file_start_write_trylock(struct file *file)
2567 if (!S_ISREG(file_inode(file)->i_mode))
2569 return __sb_start_write(file_inode(file)->i_sb, SB_FREEZE_WRITE, false);
2572 static inline void file_end_write(struct file *file)
2574 if (!S_ISREG(file_inode(file)->i_mode))
2576 __sb_end_write(file_inode(file)->i_sb, SB_FREEZE_WRITE);
2579 static inline int do_clone_file_range(struct file *file_in, loff_t pos_in,
2580 struct file *file_out, loff_t pos_out,
2585 file_start_write(file_out);
2586 ret = vfs_clone_file_range(file_in, pos_in, file_out, pos_out, len);
2587 file_end_write(file_out);
2593 * get_write_access() gets write permission for a file.
2594 * put_write_access() releases this write permission.
2595 * This is used for regular files.
2596 * We cannot support write (and maybe mmap read-write shared) accesses and
2597 * MAP_DENYWRITE mmappings simultaneously. The i_writecount field of an inode
2598 * can have the following values:
2599 * 0: no writers, no VM_DENYWRITE mappings
2600 * < 0: (-i_writecount) vm_area_structs with VM_DENYWRITE set exist
2601 * > 0: (i_writecount) users are writing to the file.
2603 * Normally we operate on that counter with atomic_{inc,dec} and it's safe
2604 * except for the cases where we don't hold i_writecount yet. Then we need to
2605 * use {get,deny}_write_access() - these functions check the sign and refuse
2606 * to do the change if sign is wrong.
2608 static inline int get_write_access(struct inode *inode)
2610 return atomic_inc_unless_negative(&inode->i_writecount) ? 0 : -ETXTBSY;
2612 static inline int deny_write_access(struct file *file)
2614 struct inode *inode = file_inode(file);
2615 return atomic_dec_unless_positive(&inode->i_writecount) ? 0 : -ETXTBSY;
2617 static inline void put_write_access(struct inode * inode)
2619 atomic_dec(&inode->i_writecount);
2621 static inline void allow_write_access(struct file *file)
2624 atomic_inc(&file_inode(file)->i_writecount);
2626 static inline bool inode_is_open_for_write(const struct inode *inode)
2628 return atomic_read(&inode->i_writecount) > 0;
2632 static inline void i_readcount_dec(struct inode *inode)
2634 BUG_ON(!atomic_read(&inode->i_readcount));
2635 atomic_dec(&inode->i_readcount);
2637 static inline void i_readcount_inc(struct inode *inode)
2639 atomic_inc(&inode->i_readcount);
2642 static inline void i_readcount_dec(struct inode *inode)
2646 static inline void i_readcount_inc(struct inode *inode)
2651 extern int do_pipe_flags(int *, int);
2653 #define __kernel_read_file_id(id) \
2654 id(UNKNOWN, unknown) \
2655 id(FIRMWARE, firmware) \
2656 id(FIRMWARE_PREALLOC_BUFFER, firmware) \
2657 id(MODULE, kernel-module) \
2658 id(KEXEC_IMAGE, kexec-image) \
2659 id(KEXEC_INITRAMFS, kexec-initramfs) \
2660 id(POLICY, security-policy) \
2663 #define __fid_enumify(ENUM, dummy) READING_ ## ENUM,
2664 #define __fid_stringify(dummy, str) #str,
2666 enum kernel_read_file_id {
2667 __kernel_read_file_id(__fid_enumify)
2670 static const char * const kernel_read_file_str[] = {
2671 __kernel_read_file_id(__fid_stringify)
2674 static inline const char *kernel_read_file_id_str(enum kernel_read_file_id id)
2676 if (id < 0 || id >= READING_MAX_ID)
2677 return kernel_read_file_str[READING_UNKNOWN];
2679 return kernel_read_file_str[id];
2682 extern int kernel_read(struct file *, loff_t, char *, unsigned long);
2683 extern int kernel_read_file(struct file *, void **, loff_t *, loff_t,
2684 enum kernel_read_file_id);
2685 extern int kernel_read_file_from_path(char *, void **, loff_t *, loff_t,
2686 enum kernel_read_file_id);
2687 extern int kernel_read_file_from_fd(int, void **, loff_t *, loff_t,
2688 enum kernel_read_file_id);
2689 extern ssize_t kernel_write(struct file *, const char *, size_t, loff_t);
2690 extern ssize_t __kernel_write(struct file *, const char *, size_t, loff_t *);
2691 extern struct file * open_exec(const char *);
2693 /* fs/dcache.c -- generic fs support functions */
2694 extern bool is_subdir(struct dentry *, struct dentry *);
2695 extern bool path_is_under(const struct path *, const struct path *);
2697 extern char *file_path(struct file *, char *, int);
2699 #include <linux/err.h>
2701 /* needed for stackable file system support */
2702 extern loff_t default_llseek(struct file *file, loff_t offset, int whence);
2704 extern loff_t vfs_llseek(struct file *file, loff_t offset, int whence);
2706 extern int inode_init_always(struct super_block *, struct inode *);
2707 extern void inode_init_once(struct inode *);
2708 extern void address_space_init_once(struct address_space *mapping);
2709 extern struct inode * igrab(struct inode *);
2710 extern ino_t iunique(struct super_block *, ino_t);
2711 extern int inode_needs_sync(struct inode *inode);
2712 extern int generic_delete_inode(struct inode *inode);
2713 static inline int generic_drop_inode(struct inode *inode)
2715 return !inode->i_nlink || inode_unhashed(inode);
2718 extern struct inode *ilookup5_nowait(struct super_block *sb,
2719 unsigned long hashval, int (*test)(struct inode *, void *),
2721 extern struct inode *ilookup5(struct super_block *sb, unsigned long hashval,
2722 int (*test)(struct inode *, void *), void *data);
2723 extern struct inode *ilookup(struct super_block *sb, unsigned long ino);
2725 extern struct inode * iget5_locked(struct super_block *, unsigned long, int (*test)(struct inode *, void *), int (*set)(struct inode *, void *), void *);
2726 extern struct inode * iget_locked(struct super_block *, unsigned long);
2727 extern struct inode *find_inode_nowait(struct super_block *,
2729 int (*match)(struct inode *,
2730 unsigned long, void *),
2732 extern int insert_inode_locked4(struct inode *, unsigned long, int (*test)(struct inode *, void *), void *);
2733 extern int insert_inode_locked(struct inode *);
2734 #ifdef CONFIG_DEBUG_LOCK_ALLOC
2735 extern void lockdep_annotate_inode_mutex_key(struct inode *inode);
2737 static inline void lockdep_annotate_inode_mutex_key(struct inode *inode) { };
2739 extern void unlock_new_inode(struct inode *);
2740 extern unsigned int get_next_ino(void);
2742 extern void __iget(struct inode * inode);
2743 extern void iget_failed(struct inode *);
2744 extern void clear_inode(struct inode *);
2745 extern void __destroy_inode(struct inode *);
2746 extern struct inode *new_inode_pseudo(struct super_block *sb);
2747 extern struct inode *new_inode(struct super_block *sb);
2748 extern void free_inode_nonrcu(struct inode *inode);
2749 extern int should_remove_suid(struct dentry *);
2750 extern int file_remove_privs(struct file *);
2752 extern void __insert_inode_hash(struct inode *, unsigned long hashval);
2753 static inline void insert_inode_hash(struct inode *inode)
2755 __insert_inode_hash(inode, inode->i_ino);
2758 extern void __remove_inode_hash(struct inode *);
2759 static inline void remove_inode_hash(struct inode *inode)
2761 if (!inode_unhashed(inode) && !hlist_fake(&inode->i_hash))
2762 __remove_inode_hash(inode);
2765 extern void inode_sb_list_add(struct inode *inode);
2768 extern int bdev_read_only(struct block_device *);
2770 extern int set_blocksize(struct block_device *, int);
2771 extern int sb_set_blocksize(struct super_block *, int);
2772 extern int sb_min_blocksize(struct super_block *, int);
2774 extern int generic_file_mmap(struct file *, struct vm_area_struct *);
2775 extern int generic_file_readonly_mmap(struct file *, struct vm_area_struct *);
2776 extern ssize_t generic_write_checks(struct kiocb *, struct iov_iter *);
2777 extern ssize_t generic_file_read_iter(struct kiocb *, struct iov_iter *);
2778 extern ssize_t __generic_file_write_iter(struct kiocb *, struct iov_iter *);
2779 extern ssize_t generic_file_write_iter(struct kiocb *, struct iov_iter *);
2780 extern ssize_t generic_file_direct_write(struct kiocb *, struct iov_iter *);
2781 extern ssize_t generic_perform_write(struct file *, struct iov_iter *, loff_t);
2783 ssize_t vfs_iter_read(struct file *file, struct iov_iter *iter, loff_t *ppos);
2784 ssize_t vfs_iter_write(struct file *file, struct iov_iter *iter, loff_t *ppos);
2786 /* fs/block_dev.c */
2787 extern ssize_t blkdev_read_iter(struct kiocb *iocb, struct iov_iter *to);
2788 extern ssize_t blkdev_write_iter(struct kiocb *iocb, struct iov_iter *from);
2789 extern int blkdev_fsync(struct file *filp, loff_t start, loff_t end,
2791 extern void block_sync_page(struct page *page);
2794 extern ssize_t generic_file_splice_read(struct file *, loff_t *,
2795 struct pipe_inode_info *, size_t, unsigned int);
2796 extern ssize_t iter_file_splice_write(struct pipe_inode_info *,
2797 struct file *, loff_t *, size_t, unsigned int);
2798 extern ssize_t generic_splice_sendpage(struct pipe_inode_info *pipe,
2799 struct file *out, loff_t *, size_t len, unsigned int flags);
2800 extern long do_splice_direct(struct file *in, loff_t *ppos, struct file *out,
2801 loff_t *opos, size_t len, unsigned int flags);
2805 file_ra_state_init(struct file_ra_state *ra, struct address_space *mapping);
2806 extern loff_t noop_llseek(struct file *file, loff_t offset, int whence);
2807 extern loff_t no_llseek(struct file *file, loff_t offset, int whence);
2808 extern loff_t vfs_setpos(struct file *file, loff_t offset, loff_t maxsize);
2809 extern loff_t generic_file_llseek(struct file *file, loff_t offset, int whence);
2810 extern loff_t generic_file_llseek_size(struct file *file, loff_t offset,
2811 int whence, loff_t maxsize, loff_t eof);
2812 extern loff_t fixed_size_llseek(struct file *file, loff_t offset,
2813 int whence, loff_t size);
2814 extern loff_t no_seek_end_llseek_size(struct file *, loff_t, int, loff_t);
2815 extern loff_t no_seek_end_llseek(struct file *, loff_t, int);
2816 extern int generic_file_open(struct inode * inode, struct file * filp);
2817 extern int nonseekable_open(struct inode * inode, struct file * filp);
2820 typedef void (dio_submit_t)(struct bio *bio, struct inode *inode,
2821 loff_t file_offset);
2824 /* need locking between buffered and direct access */
2827 /* filesystem does not support filling holes */
2828 DIO_SKIP_HOLES = 0x02,
2830 /* filesystem can handle aio writes beyond i_size */
2831 DIO_ASYNC_EXTEND = 0x04,
2833 /* inode/fs/bdev does not need truncate protection */
2834 DIO_SKIP_DIO_COUNT = 0x08,
2837 void dio_end_io(struct bio *bio, int error);
2839 ssize_t __blockdev_direct_IO(struct kiocb *iocb, struct inode *inode,
2840 struct block_device *bdev, struct iov_iter *iter,
2841 get_block_t get_block,
2842 dio_iodone_t end_io, dio_submit_t submit_io,
2845 static inline ssize_t blockdev_direct_IO(struct kiocb *iocb,
2846 struct inode *inode,
2847 struct iov_iter *iter,
2848 get_block_t get_block)
2850 return __blockdev_direct_IO(iocb, inode, inode->i_sb->s_bdev, iter,
2851 get_block, NULL, NULL, DIO_LOCKING | DIO_SKIP_HOLES);
2855 void inode_dio_wait(struct inode *inode);
2858 * inode_dio_begin - signal start of a direct I/O requests
2859 * @inode: inode the direct I/O happens on
2861 * This is called once we've finished processing a direct I/O request,
2862 * and is used to wake up callers waiting for direct I/O to be quiesced.
2864 static inline void inode_dio_begin(struct inode *inode)
2866 atomic_inc(&inode->i_dio_count);
2870 * inode_dio_end - signal finish of a direct I/O requests
2871 * @inode: inode the direct I/O happens on
2873 * This is called once we've finished processing a direct I/O request,
2874 * and is used to wake up callers waiting for direct I/O to be quiesced.
2876 static inline void inode_dio_end(struct inode *inode)
2878 if (atomic_dec_and_test(&inode->i_dio_count))
2879 wake_up_bit(&inode->i_state, __I_DIO_WAKEUP);
2882 extern void inode_set_flags(struct inode *inode, unsigned int flags,
2885 extern const struct file_operations generic_ro_fops;
2887 #define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
2889 extern int readlink_copy(char __user *, int, const char *);
2890 extern int page_readlink(struct dentry *, char __user *, int);
2891 extern const char *page_get_link(struct dentry *, struct inode *,
2892 struct delayed_call *);
2893 extern void page_put_link(void *);
2894 extern int __page_symlink(struct inode *inode, const char *symname, int len,
2896 extern int page_symlink(struct inode *inode, const char *symname, int len);
2897 extern const struct inode_operations page_symlink_inode_operations;
2898 extern void kfree_link(void *);
2899 extern void generic_fillattr(struct inode *, struct kstat *);
2900 int vfs_getattr_nosec(struct path *path, struct kstat *stat);
2901 extern int vfs_getattr(struct path *, struct kstat *);
2902 void __inode_add_bytes(struct inode *inode, loff_t bytes);
2903 void inode_add_bytes(struct inode *inode, loff_t bytes);
2904 void __inode_sub_bytes(struct inode *inode, loff_t bytes);
2905 void inode_sub_bytes(struct inode *inode, loff_t bytes);
2906 loff_t inode_get_bytes(struct inode *inode);
2907 void inode_set_bytes(struct inode *inode, loff_t bytes);
2908 const char *simple_get_link(struct dentry *, struct inode *,
2909 struct delayed_call *);
2910 extern const struct inode_operations simple_symlink_inode_operations;
2912 extern int iterate_dir(struct file *, struct dir_context *);
2914 extern int vfs_stat(const char __user *, struct kstat *);
2915 extern int vfs_lstat(const char __user *, struct kstat *);
2916 extern int vfs_fstat(unsigned int, struct kstat *);
2917 extern int vfs_fstatat(int , const char __user *, struct kstat *, int);
2918 extern const char *vfs_get_link(struct dentry *, struct delayed_call *);
2919 extern int vfs_readlink(struct dentry *, char __user *, int);
2921 extern int __generic_block_fiemap(struct inode *inode,
2922 struct fiemap_extent_info *fieinfo,
2923 loff_t start, loff_t len,
2924 get_block_t *get_block);
2925 extern int generic_block_fiemap(struct inode *inode,
2926 struct fiemap_extent_info *fieinfo, u64 start,
2927 u64 len, get_block_t *get_block);
2929 extern void get_filesystem(struct file_system_type *fs);
2930 extern void put_filesystem(struct file_system_type *fs);
2931 extern struct file_system_type *get_fs_type(const char *name);
2932 extern struct super_block *get_super(struct block_device *);
2933 extern struct super_block *get_super_thawed(struct block_device *);
2934 extern struct super_block *get_super_exclusive_thawed(struct block_device *bdev);
2935 extern struct super_block *get_active_super(struct block_device *bdev);
2936 extern void drop_super(struct super_block *sb);
2937 extern void drop_super_exclusive(struct super_block *sb);
2938 extern void iterate_supers(void (*)(struct super_block *, void *), void *);
2939 extern void iterate_supers_type(struct file_system_type *,
2940 void (*)(struct super_block *, void *), void *);
2942 extern int dcache_dir_open(struct inode *, struct file *);
2943 extern int dcache_dir_close(struct inode *, struct file *);
2944 extern loff_t dcache_dir_lseek(struct file *, loff_t, int);
2945 extern int dcache_readdir(struct file *, struct dir_context *);
2946 extern int simple_setattr(struct dentry *, struct iattr *);
2947 extern int simple_getattr(struct vfsmount *, struct dentry *, struct kstat *);
2948 extern int simple_statfs(struct dentry *, struct kstatfs *);
2949 extern int simple_open(struct inode *inode, struct file *file);
2950 extern int simple_link(struct dentry *, struct inode *, struct dentry *);
2951 extern int simple_unlink(struct inode *, struct dentry *);
2952 extern int simple_rmdir(struct inode *, struct dentry *);
2953 extern int simple_rename(struct inode *, struct dentry *,
2954 struct inode *, struct dentry *, unsigned int);
2955 extern int noop_fsync(struct file *, loff_t, loff_t, int);
2956 extern int simple_empty(struct dentry *);
2957 extern int simple_readpage(struct file *file, struct page *page);
2958 extern int simple_write_begin(struct file *file, struct address_space *mapping,
2959 loff_t pos, unsigned len, unsigned flags,
2960 struct page **pagep, void **fsdata);
2961 extern int simple_write_end(struct file *file, struct address_space *mapping,
2962 loff_t pos, unsigned len, unsigned copied,
2963 struct page *page, void *fsdata);
2964 extern int always_delete_dentry(const struct dentry *);
2965 extern struct inode *alloc_anon_inode(struct super_block *);
2966 extern int simple_nosetlease(struct file *, long, struct file_lock **, void **);
2967 extern const struct dentry_operations simple_dentry_operations;
2969 extern struct dentry *simple_lookup(struct inode *, struct dentry *, unsigned int flags);
2970 extern ssize_t generic_read_dir(struct file *, char __user *, size_t, loff_t *);
2971 extern const struct file_operations simple_dir_operations;
2972 extern const struct inode_operations simple_dir_inode_operations;
2973 extern void make_empty_dir_inode(struct inode *inode);
2974 extern bool is_empty_dir_inode(struct inode *inode);
2975 struct tree_descr { char *name; const struct file_operations *ops; int mode; };
2976 struct dentry *d_alloc_name(struct dentry *, const char *);
2977 extern int simple_fill_super(struct super_block *, unsigned long, struct tree_descr *);
2978 extern int simple_pin_fs(struct file_system_type *, struct vfsmount **mount, int *count);
2979 extern void simple_release_fs(struct vfsmount **mount, int *count);
2981 extern ssize_t simple_read_from_buffer(void __user *to, size_t count,
2982 loff_t *ppos, const void *from, size_t available);
2983 extern ssize_t simple_write_to_buffer(void *to, size_t available, loff_t *ppos,
2984 const void __user *from, size_t count);
2986 extern int __generic_file_fsync(struct file *, loff_t, loff_t, int);
2987 extern int generic_file_fsync(struct file *, loff_t, loff_t, int);
2989 extern int generic_check_addressable(unsigned, u64);
2991 #ifdef CONFIG_MIGRATION
2992 extern int buffer_migrate_page(struct address_space *,
2993 struct page *, struct page *,
2996 #define buffer_migrate_page NULL
2999 extern int setattr_prepare(struct dentry *, struct iattr *);
3000 extern int inode_newsize_ok(const struct inode *, loff_t offset);
3001 extern void setattr_copy(struct inode *inode, const struct iattr *attr);
3003 extern int file_update_time(struct file *file);
3005 extern int generic_show_options(struct seq_file *m, struct dentry *root);
3006 extern void save_mount_options(struct super_block *sb, char *options);
3007 extern void replace_mount_options(struct super_block *sb, char *options);
3009 static inline bool io_is_direct(struct file *filp)
3011 return (filp->f_flags & O_DIRECT) || IS_DAX(filp->f_mapping->host);
3014 static inline int iocb_flags(struct file *file)
3017 if (file->f_flags & O_APPEND)
3019 if (io_is_direct(file))
3021 if ((file->f_flags & O_DSYNC) || IS_SYNC(file->f_mapping->host))
3023 if (file->f_flags & __O_SYNC)
3028 static inline ino_t parent_ino(struct dentry *dentry)
3033 * Don't strictly need d_lock here? If the parent ino could change
3034 * then surely we'd have a deeper race in the caller?
3036 spin_lock(&dentry->d_lock);
3037 res = dentry->d_parent->d_inode->i_ino;
3038 spin_unlock(&dentry->d_lock);
3042 /* Transaction based IO helpers */
3045 * An argresp is stored in an allocated page and holds the
3046 * size of the argument or response, along with its content
3048 struct simple_transaction_argresp {
3053 #define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
3055 char *simple_transaction_get(struct file *file, const char __user *buf,
3057 ssize_t simple_transaction_read(struct file *file, char __user *buf,
3058 size_t size, loff_t *pos);
3059 int simple_transaction_release(struct inode *inode, struct file *file);
3061 void simple_transaction_set(struct file *file, size_t n);
3064 * simple attribute files
3066 * These attributes behave similar to those in sysfs:
3068 * Writing to an attribute immediately sets a value, an open file can be
3069 * written to multiple times.
3071 * Reading from an attribute creates a buffer from the value that might get
3072 * read with multiple read calls. When the attribute has been read
3073 * completely, no further read calls are possible until the file is opened
3076 * All attributes contain a text representation of a numeric value
3077 * that are accessed with the get() and set() functions.
3079 #define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt) \
3080 static int __fops ## _open(struct inode *inode, struct file *file) \
3082 __simple_attr_check_format(__fmt, 0ull); \
3083 return simple_attr_open(inode, file, __get, __set, __fmt); \
3085 static const struct file_operations __fops = { \
3086 .owner = THIS_MODULE, \
3087 .open = __fops ## _open, \
3088 .release = simple_attr_release, \
3089 .read = simple_attr_read, \
3090 .write = simple_attr_write, \
3091 .llseek = generic_file_llseek, \
3094 static inline __printf(1, 2)
3095 void __simple_attr_check_format(const char *fmt, ...)
3097 /* don't do anything, just let the compiler check the arguments; */
3100 int simple_attr_open(struct inode *inode, struct file *file,
3101 int (*get)(void *, u64 *), int (*set)(void *, u64),
3103 int simple_attr_release(struct inode *inode, struct file *file);
3104 ssize_t simple_attr_read(struct file *file, char __user *buf,
3105 size_t len, loff_t *ppos);
3106 ssize_t simple_attr_write(struct file *file, const char __user *buf,
3107 size_t len, loff_t *ppos);
3110 int proc_nr_files(struct ctl_table *table, int write,
3111 void __user *buffer, size_t *lenp, loff_t *ppos);
3112 int proc_nr_dentry(struct ctl_table *table, int write,
3113 void __user *buffer, size_t *lenp, loff_t *ppos);
3114 int proc_nr_inodes(struct ctl_table *table, int write,
3115 void __user *buffer, size_t *lenp, loff_t *ppos);
3116 int __init get_filesystem_list(char *buf);
3118 #define __FMODE_EXEC ((__force int) FMODE_EXEC)
3119 #define __FMODE_NONOTIFY ((__force int) FMODE_NONOTIFY)
3121 #define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE])
3122 #define OPEN_FMODE(flag) ((__force fmode_t)(((flag + 1) & O_ACCMODE) | \
3123 (flag & __FMODE_NONOTIFY)))
3125 static inline bool is_sxid(umode_t mode)
3127 return (mode & S_ISUID) || ((mode & S_ISGID) && (mode & S_IXGRP));
3130 static inline int check_sticky(struct inode *dir, struct inode *inode)
3132 if (!(dir->i_mode & S_ISVTX))
3135 return __check_sticky(dir, inode);
3138 static inline void inode_has_no_xattr(struct inode *inode)
3140 if (!is_sxid(inode->i_mode) && (inode->i_sb->s_flags & MS_NOSEC))
3141 inode->i_flags |= S_NOSEC;
3144 static inline bool is_root_inode(struct inode *inode)
3146 return inode == inode->i_sb->s_root->d_inode;
3149 static inline bool dir_emit(struct dir_context *ctx,
3150 const char *name, int namelen,
3151 u64 ino, unsigned type)
3153 return ctx->actor(ctx, name, namelen, ctx->pos, ino, type) == 0;
3155 static inline bool dir_emit_dot(struct file *file, struct dir_context *ctx)
3157 return ctx->actor(ctx, ".", 1, ctx->pos,
3158 file->f_path.dentry->d_inode->i_ino, DT_DIR) == 0;
3160 static inline bool dir_emit_dotdot(struct file *file, struct dir_context *ctx)
3162 return ctx->actor(ctx, "..", 2, ctx->pos,
3163 parent_ino(file->f_path.dentry), DT_DIR) == 0;
3165 static inline bool dir_emit_dots(struct file *file, struct dir_context *ctx)
3167 if (ctx->pos == 0) {
3168 if (!dir_emit_dot(file, ctx))
3172 if (ctx->pos == 1) {
3173 if (!dir_emit_dotdot(file, ctx))
3179 static inline bool dir_relax(struct inode *inode)
3181 inode_unlock(inode);
3183 return !IS_DEADDIR(inode);
3186 static inline bool dir_relax_shared(struct inode *inode)
3188 inode_unlock_shared(inode);
3189 inode_lock_shared(inode);
3190 return !IS_DEADDIR(inode);
3193 extern bool path_noexec(const struct path *path);
3194 extern void inode_nohighmem(struct inode *inode);
3196 #endif /* _LINUX_FS_H */