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 list_head bd_list;
428 * Private data. You must have bd_claim'ed the block_device
429 * to use this. NOTE: bd_claim allows an owner to claim
430 * the same device multiple times, the owner must take special
431 * care to not mess up bd_private for that case.
433 unsigned long bd_private;
435 /* The counter of freeze processes */
436 int bd_fsfreeze_count;
437 /* Mutex for freeze */
438 struct mutex bd_fsfreeze_mutex;
442 * Radix-tree tags, for tagging dirty and writeback pages within the pagecache
445 #define PAGECACHE_TAG_DIRTY 0
446 #define PAGECACHE_TAG_WRITEBACK 1
447 #define PAGECACHE_TAG_TOWRITE 2
449 int mapping_tagged(struct address_space *mapping, int tag);
451 static inline void i_mmap_lock_write(struct address_space *mapping)
453 down_write(&mapping->i_mmap_rwsem);
456 static inline void i_mmap_unlock_write(struct address_space *mapping)
458 up_write(&mapping->i_mmap_rwsem);
461 static inline void i_mmap_lock_read(struct address_space *mapping)
463 down_read(&mapping->i_mmap_rwsem);
466 static inline void i_mmap_unlock_read(struct address_space *mapping)
468 up_read(&mapping->i_mmap_rwsem);
472 * Might pages of this file be mapped into userspace?
474 static inline int mapping_mapped(struct address_space *mapping)
476 return !RB_EMPTY_ROOT(&mapping->i_mmap);
480 * Might pages of this file have been modified in userspace?
481 * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap_pgoff
482 * marks vma as VM_SHARED if it is shared, and the file was opened for
483 * writing i.e. vma may be mprotected writable even if now readonly.
485 * If i_mmap_writable is negative, no new writable mappings are allowed. You
486 * can only deny writable mappings, if none exists right now.
488 static inline int mapping_writably_mapped(struct address_space *mapping)
490 return atomic_read(&mapping->i_mmap_writable) > 0;
493 static inline int mapping_map_writable(struct address_space *mapping)
495 return atomic_inc_unless_negative(&mapping->i_mmap_writable) ?
499 static inline void mapping_unmap_writable(struct address_space *mapping)
501 atomic_dec(&mapping->i_mmap_writable);
504 static inline int mapping_deny_writable(struct address_space *mapping)
506 return atomic_dec_unless_positive(&mapping->i_mmap_writable) ?
510 static inline void mapping_allow_writable(struct address_space *mapping)
512 atomic_inc(&mapping->i_mmap_writable);
516 * Use sequence counter to get consistent i_size on 32-bit processors.
518 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
519 #include <linux/seqlock.h>
520 #define __NEED_I_SIZE_ORDERED
521 #define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
523 #define i_size_ordered_init(inode) do { } while (0)
527 #define ACL_NOT_CACHED ((void *)(-1))
528 #define ACL_DONT_CACHE ((void *)(-3))
530 static inline struct posix_acl *
531 uncached_acl_sentinel(struct task_struct *task)
533 return (void *)task + 1;
537 is_uncached_acl(struct posix_acl *acl)
539 return (long)acl & 1;
542 #define IOP_FASTPERM 0x0001
543 #define IOP_LOOKUP 0x0002
544 #define IOP_NOFOLLOW 0x0004
545 #define IOP_XATTR 0x0008
548 * Keep mostly read-only and often accessed (especially for
549 * the RCU path lookup and 'stat' data) fields at the beginning
550 * of the 'struct inode'
554 unsigned short i_opflags;
557 unsigned int i_flags;
559 #ifdef CONFIG_FS_POSIX_ACL
560 struct posix_acl *i_acl;
561 struct posix_acl *i_default_acl;
564 const struct inode_operations *i_op;
565 struct super_block *i_sb;
566 struct address_space *i_mapping;
568 #ifdef CONFIG_SECURITY
572 /* Stat data, not accessed from path walking */
575 * Filesystems may only read i_nlink directly. They shall use the
576 * following functions for modification:
578 * (set|clear|inc|drop)_nlink
579 * inode_(inc|dec)_link_count
582 const unsigned int i_nlink;
583 unsigned int __i_nlink;
587 struct timespec i_atime;
588 struct timespec i_mtime;
589 struct timespec i_ctime;
590 spinlock_t i_lock; /* i_blocks, i_bytes, maybe i_size */
591 unsigned short i_bytes;
592 unsigned int i_blkbits;
595 #ifdef __NEED_I_SIZE_ORDERED
596 seqcount_t i_size_seqcount;
600 unsigned long i_state;
601 struct rw_semaphore i_rwsem;
603 unsigned long dirtied_when; /* jiffies of first dirtying */
604 unsigned long dirtied_time_when;
606 struct hlist_node i_hash;
607 struct list_head i_io_list; /* backing dev IO list */
608 #ifdef CONFIG_CGROUP_WRITEBACK
609 struct bdi_writeback *i_wb; /* the associated cgroup wb */
611 /* foreign inode detection, see wbc_detach_inode() */
613 u16 i_wb_frn_avg_time;
614 u16 i_wb_frn_history;
616 struct list_head i_lru; /* inode LRU list */
617 struct list_head i_sb_list;
618 struct list_head i_wb_list; /* backing dev writeback list */
620 struct hlist_head i_dentry;
621 struct rcu_head i_rcu;
625 atomic_t i_dio_count;
626 atomic_t i_writecount;
628 atomic_t i_readcount; /* struct files open RO */
630 const struct file_operations *i_fop; /* former ->i_op->default_file_ops */
631 struct file_lock_context *i_flctx;
632 struct address_space i_data;
633 struct list_head i_devices;
635 struct pipe_inode_info *i_pipe;
636 struct block_device *i_bdev;
644 #ifdef CONFIG_FSNOTIFY
645 __u32 i_fsnotify_mask; /* all events this inode cares about */
646 struct hlist_head i_fsnotify_marks;
649 #if IS_ENABLED(CONFIG_FS_ENCRYPTION)
650 struct fscrypt_info *i_crypt_info;
653 void *i_private; /* fs or device private pointer */
656 static inline int inode_unhashed(struct inode *inode)
658 return hlist_unhashed(&inode->i_hash);
662 * inode->i_mutex nesting subclasses for the lock validator:
664 * 0: the object of the current VFS operation
668 * 4: second non-directory
669 * 5: second parent (when locking independent directories in rename)
671 * I_MUTEX_NONDIR2 is for certain operations (such as rename) which lock two
672 * non-directories at once.
674 * The locking order between these classes is
675 * parent[2] -> child -> grandchild -> normal -> xattr -> second non-directory
677 enum inode_i_mutex_lock_class
687 static inline void inode_lock(struct inode *inode)
689 down_write(&inode->i_rwsem);
692 static inline void inode_unlock(struct inode *inode)
694 up_write(&inode->i_rwsem);
697 static inline void inode_lock_shared(struct inode *inode)
699 down_read(&inode->i_rwsem);
702 static inline void inode_unlock_shared(struct inode *inode)
704 up_read(&inode->i_rwsem);
707 static inline int inode_trylock(struct inode *inode)
709 return down_write_trylock(&inode->i_rwsem);
712 static inline int inode_trylock_shared(struct inode *inode)
714 return down_read_trylock(&inode->i_rwsem);
717 static inline int inode_is_locked(struct inode *inode)
719 return rwsem_is_locked(&inode->i_rwsem);
722 static inline void inode_lock_nested(struct inode *inode, unsigned subclass)
724 down_write_nested(&inode->i_rwsem, subclass);
727 void lock_two_nondirectories(struct inode *, struct inode*);
728 void unlock_two_nondirectories(struct inode *, struct inode*);
731 * NOTE: in a 32bit arch with a preemptable kernel and
732 * an UP compile the i_size_read/write must be atomic
733 * with respect to the local cpu (unlike with preempt disabled),
734 * but they don't need to be atomic with respect to other cpus like in
735 * true SMP (so they need either to either locally disable irq around
736 * the read or for example on x86 they can be still implemented as a
737 * cmpxchg8b without the need of the lock prefix). For SMP compiles
738 * and 64bit archs it makes no difference if preempt is enabled or not.
740 static inline loff_t i_size_read(const struct inode *inode)
742 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
747 seq = read_seqcount_begin(&inode->i_size_seqcount);
748 i_size = inode->i_size;
749 } while (read_seqcount_retry(&inode->i_size_seqcount, seq));
751 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
755 i_size = inode->i_size;
759 return inode->i_size;
764 * NOTE: unlike i_size_read(), i_size_write() does need locking around it
765 * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount
766 * can be lost, resulting in subsequent i_size_read() calls spinning forever.
768 static inline void i_size_write(struct inode *inode, loff_t i_size)
770 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
772 write_seqcount_begin(&inode->i_size_seqcount);
773 inode->i_size = i_size;
774 write_seqcount_end(&inode->i_size_seqcount);
776 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
778 inode->i_size = i_size;
781 inode->i_size = i_size;
785 static inline unsigned iminor(const struct inode *inode)
787 return MINOR(inode->i_rdev);
790 static inline unsigned imajor(const struct inode *inode)
792 return MAJOR(inode->i_rdev);
795 extern struct block_device *I_BDEV(struct inode *inode);
798 rwlock_t lock; /* protects pid, uid, euid fields */
799 struct pid *pid; /* pid or -pgrp where SIGIO should be sent */
800 enum pid_type pid_type; /* Kind of process group SIGIO should be sent to */
801 kuid_t uid, euid; /* uid/euid of process setting the owner */
802 int signum; /* posix.1b rt signal to be delivered on IO */
806 * Track a single file's readahead state
808 struct file_ra_state {
809 pgoff_t start; /* where readahead started */
810 unsigned int size; /* # of readahead pages */
811 unsigned int async_size; /* do asynchronous readahead when
812 there are only # of pages ahead */
814 unsigned int ra_pages; /* Maximum readahead window */
815 unsigned int mmap_miss; /* Cache miss stat for mmap accesses */
816 loff_t prev_pos; /* Cache last read() position */
820 * Check if @index falls in the readahead windows.
822 static inline int ra_has_index(struct file_ra_state *ra, pgoff_t index)
824 return (index >= ra->start &&
825 index < ra->start + ra->size);
830 struct llist_node fu_llist;
831 struct rcu_head fu_rcuhead;
834 struct inode *f_inode; /* cached value */
835 const struct file_operations *f_op;
838 * Protects f_ep_links, f_flags.
839 * Must not be taken from IRQ context.
842 atomic_long_t f_count;
843 unsigned int f_flags;
845 struct mutex f_pos_lock;
847 struct fown_struct f_owner;
848 const struct cred *f_cred;
849 struct file_ra_state f_ra;
852 #ifdef CONFIG_SECURITY
855 /* needed for tty driver, and maybe others */
859 /* Used by fs/eventpoll.c to link all the hooks to this file */
860 struct list_head f_ep_links;
861 struct list_head f_tfile_llink;
862 #endif /* #ifdef CONFIG_EPOLL */
863 struct address_space *f_mapping;
864 } __attribute__((aligned(4))); /* lest something weird decides that 2 is OK */
869 /* file identifier */
870 unsigned char f_handle[0];
873 static inline struct file *get_file(struct file *f)
875 atomic_long_inc(&f->f_count);
878 #define get_file_rcu(x) atomic_long_inc_not_zero(&(x)->f_count)
879 #define fput_atomic(x) atomic_long_add_unless(&(x)->f_count, -1, 1)
880 #define file_count(x) atomic_long_read(&(x)->f_count)
882 #define MAX_NON_LFS ((1UL<<31) - 1)
884 /* Page cache limit. The filesystems should put that into their s_maxbytes
885 limits, otherwise bad things can happen in VM. */
886 #if BITS_PER_LONG==32
887 #define MAX_LFS_FILESIZE (((loff_t)PAGE_SIZE << (BITS_PER_LONG-1))-1)
888 #elif BITS_PER_LONG==64
889 #define MAX_LFS_FILESIZE ((loff_t)0x7fffffffffffffffLL)
894 #define FL_DELEG 4 /* NFSv4 delegation */
895 #define FL_ACCESS 8 /* not trying to lock, just looking */
896 #define FL_EXISTS 16 /* when unlocking, test for existence */
897 #define FL_LEASE 32 /* lease held on this file */
898 #define FL_CLOSE 64 /* unlock on close */
899 #define FL_SLEEP 128 /* A blocking lock */
900 #define FL_DOWNGRADE_PENDING 256 /* Lease is being downgraded */
901 #define FL_UNLOCK_PENDING 512 /* Lease is being broken */
902 #define FL_OFDLCK 1024 /* lock is "owned" by struct file */
903 #define FL_LAYOUT 2048 /* outstanding pNFS layout */
906 * Special return value from posix_lock_file() and vfs_lock_file() for
907 * asynchronous locking.
909 #define FILE_LOCK_DEFERRED 1
911 /* legacy typedef, should eventually be removed */
912 typedef void *fl_owner_t;
916 struct file_lock_operations {
917 void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
918 void (*fl_release_private)(struct file_lock *);
921 struct lock_manager_operations {
922 int (*lm_compare_owner)(struct file_lock *, struct file_lock *);
923 unsigned long (*lm_owner_key)(struct file_lock *);
924 fl_owner_t (*lm_get_owner)(fl_owner_t);
925 void (*lm_put_owner)(fl_owner_t);
926 void (*lm_notify)(struct file_lock *); /* unblock callback */
927 int (*lm_grant)(struct file_lock *, int);
928 bool (*lm_break)(struct file_lock *);
929 int (*lm_change)(struct file_lock *, int, struct list_head *);
930 void (*lm_setup)(struct file_lock *, void **);
933 struct lock_manager {
934 struct list_head list;
936 * NFSv4 and up also want opens blocked during the grace period;
943 void locks_start_grace(struct net *, struct lock_manager *);
944 void locks_end_grace(struct lock_manager *);
945 int locks_in_grace(struct net *);
946 int opens_in_grace(struct net *);
948 /* that will die - we need it for nfs_lock_info */
949 #include <linux/nfs_fs_i.h>
952 * struct file_lock represents a generic "file lock". It's used to represent
953 * POSIX byte range locks, BSD (flock) locks, and leases. It's important to
954 * note that the same struct is used to represent both a request for a lock and
955 * the lock itself, but the same object is never used for both.
957 * FIXME: should we create a separate "struct lock_request" to help distinguish
960 * The varous i_flctx lists are ordered by:
963 * 2) lock range start
966 * Obviously, the last two criteria only matter for POSIX locks.
969 struct file_lock *fl_next; /* singly linked list for this inode */
970 struct list_head fl_list; /* link into file_lock_context */
971 struct hlist_node fl_link; /* node in global lists */
972 struct list_head fl_block; /* circular list of blocked processes */
974 unsigned int fl_flags;
975 unsigned char fl_type;
977 int fl_link_cpu; /* what cpu's list is this on? */
978 struct pid *fl_nspid;
979 wait_queue_head_t fl_wait;
980 struct file *fl_file;
984 struct fasync_struct * fl_fasync; /* for lease break notifications */
985 /* for lease breaks: */
986 unsigned long fl_break_time;
987 unsigned long fl_downgrade_time;
989 const struct file_lock_operations *fl_ops; /* Callbacks for filesystems */
990 const struct lock_manager_operations *fl_lmops; /* Callbacks for lockmanagers */
992 struct nfs_lock_info nfs_fl;
993 struct nfs4_lock_info nfs4_fl;
995 struct list_head link; /* link in AFS vnode's pending_locks list */
996 int state; /* state of grant or error if -ve */
1001 struct file_lock_context {
1002 spinlock_t flc_lock;
1003 struct list_head flc_flock;
1004 struct list_head flc_posix;
1005 struct list_head flc_lease;
1008 /* The following constant reflects the upper bound of the file/locking space */
1010 #define INT_LIMIT(x) (~((x)1 << (sizeof(x)*8 - 1)))
1011 #define OFFSET_MAX INT_LIMIT(loff_t)
1012 #define OFFT_OFFSET_MAX INT_LIMIT(off_t)
1015 #include <linux/fcntl.h>
1017 extern void send_sigio(struct fown_struct *fown, int fd, int band);
1020 * Return the inode to use for locking
1022 * For overlayfs this should be the overlay inode, not the real inode returned
1023 * by file_inode(). For any other fs file_inode(filp) and locks_inode(filp) are
1026 static inline struct inode *locks_inode(const struct file *f)
1028 return f->f_path.dentry->d_inode;
1031 #ifdef CONFIG_FILE_LOCKING
1032 extern int fcntl_getlk(struct file *, unsigned int, struct flock __user *);
1033 extern int fcntl_setlk(unsigned int, struct file *, unsigned int,
1034 struct flock __user *);
1036 #if BITS_PER_LONG == 32
1037 extern int fcntl_getlk64(struct file *, unsigned int, struct flock64 __user *);
1038 extern int fcntl_setlk64(unsigned int, struct file *, unsigned int,
1039 struct flock64 __user *);
1042 extern int fcntl_setlease(unsigned int fd, struct file *filp, long arg);
1043 extern int fcntl_getlease(struct file *filp);
1046 void locks_free_lock_context(struct inode *inode);
1047 void locks_free_lock(struct file_lock *fl);
1048 extern void locks_init_lock(struct file_lock *);
1049 extern struct file_lock * locks_alloc_lock(void);
1050 extern void locks_copy_lock(struct file_lock *, struct file_lock *);
1051 extern void locks_copy_conflock(struct file_lock *, struct file_lock *);
1052 extern void locks_remove_posix(struct file *, fl_owner_t);
1053 extern void locks_remove_file(struct file *);
1054 extern void locks_release_private(struct file_lock *);
1055 extern void posix_test_lock(struct file *, struct file_lock *);
1056 extern int posix_lock_file(struct file *, struct file_lock *, struct file_lock *);
1057 extern int posix_unblock_lock(struct file_lock *);
1058 extern int vfs_test_lock(struct file *, struct file_lock *);
1059 extern int vfs_lock_file(struct file *, unsigned int, struct file_lock *, struct file_lock *);
1060 extern int vfs_cancel_lock(struct file *filp, struct file_lock *fl);
1061 extern int locks_lock_inode_wait(struct inode *inode, struct file_lock *fl);
1062 extern int __break_lease(struct inode *inode, unsigned int flags, unsigned int type);
1063 extern void lease_get_mtime(struct inode *, struct timespec *time);
1064 extern int generic_setlease(struct file *, long, struct file_lock **, void **priv);
1065 extern int vfs_setlease(struct file *, long, struct file_lock **, void **);
1066 extern int lease_modify(struct file_lock *, int, struct list_head *);
1067 struct files_struct;
1068 extern void show_fd_locks(struct seq_file *f,
1069 struct file *filp, struct files_struct *files);
1070 #else /* !CONFIG_FILE_LOCKING */
1071 static inline int fcntl_getlk(struct file *file, unsigned int cmd,
1072 struct flock __user *user)
1077 static inline int fcntl_setlk(unsigned int fd, struct file *file,
1078 unsigned int cmd, struct flock __user *user)
1083 #if BITS_PER_LONG == 32
1084 static inline int fcntl_getlk64(struct file *file, unsigned int cmd,
1085 struct flock64 __user *user)
1090 static inline int fcntl_setlk64(unsigned int fd, struct file *file,
1091 unsigned int cmd, struct flock64 __user *user)
1096 static inline int fcntl_setlease(unsigned int fd, struct file *filp, long arg)
1101 static inline int fcntl_getlease(struct file *filp)
1107 locks_free_lock_context(struct inode *inode)
1111 static inline void locks_init_lock(struct file_lock *fl)
1116 static inline void locks_copy_conflock(struct file_lock *new, struct file_lock *fl)
1121 static inline void locks_copy_lock(struct file_lock *new, struct file_lock *fl)
1126 static inline void locks_remove_posix(struct file *filp, fl_owner_t owner)
1131 static inline void locks_remove_file(struct file *filp)
1136 static inline void posix_test_lock(struct file *filp, struct file_lock *fl)
1141 static inline int posix_lock_file(struct file *filp, struct file_lock *fl,
1142 struct file_lock *conflock)
1147 static inline int posix_unblock_lock(struct file_lock *waiter)
1152 static inline int vfs_test_lock(struct file *filp, struct file_lock *fl)
1157 static inline int vfs_lock_file(struct file *filp, unsigned int cmd,
1158 struct file_lock *fl, struct file_lock *conf)
1163 static inline int vfs_cancel_lock(struct file *filp, struct file_lock *fl)
1168 static inline int locks_lock_inode_wait(struct inode *inode, struct file_lock *fl)
1173 static inline int __break_lease(struct inode *inode, unsigned int mode, unsigned int type)
1178 static inline void lease_get_mtime(struct inode *inode, struct timespec *time)
1183 static inline int generic_setlease(struct file *filp, long arg,
1184 struct file_lock **flp, void **priv)
1189 static inline int vfs_setlease(struct file *filp, long arg,
1190 struct file_lock **lease, void **priv)
1195 static inline int lease_modify(struct file_lock *fl, int arg,
1196 struct list_head *dispose)
1201 struct files_struct;
1202 static inline void show_fd_locks(struct seq_file *f,
1203 struct file *filp, struct files_struct *files) {}
1204 #endif /* !CONFIG_FILE_LOCKING */
1206 static inline struct inode *file_inode(const struct file *f)
1211 static inline struct dentry *file_dentry(const struct file *file)
1213 return d_real(file->f_path.dentry, file_inode(file), 0);
1216 static inline int locks_lock_file_wait(struct file *filp, struct file_lock *fl)
1218 return locks_lock_inode_wait(locks_inode(filp), fl);
1221 struct fasync_struct {
1225 struct fasync_struct *fa_next; /* singly linked list */
1226 struct file *fa_file;
1227 struct rcu_head fa_rcu;
1230 #define FASYNC_MAGIC 0x4601
1232 /* SMP safe fasync helpers: */
1233 extern int fasync_helper(int, struct file *, int, struct fasync_struct **);
1234 extern struct fasync_struct *fasync_insert_entry(int, struct file *, struct fasync_struct **, struct fasync_struct *);
1235 extern int fasync_remove_entry(struct file *, struct fasync_struct **);
1236 extern struct fasync_struct *fasync_alloc(void);
1237 extern void fasync_free(struct fasync_struct *);
1239 /* can be called from interrupts */
1240 extern void kill_fasync(struct fasync_struct **, int, int);
1242 extern void __f_setown(struct file *filp, struct pid *, enum pid_type, int force);
1243 extern void f_setown(struct file *filp, unsigned long arg, int force);
1244 extern void f_delown(struct file *filp);
1245 extern pid_t f_getown(struct file *filp);
1246 extern int send_sigurg(struct fown_struct *fown);
1254 #define MNT_FORCE 0x00000001 /* Attempt to forcibily umount */
1255 #define MNT_DETACH 0x00000002 /* Just detach from the tree */
1256 #define MNT_EXPIRE 0x00000004 /* Mark for expiry */
1257 #define UMOUNT_NOFOLLOW 0x00000008 /* Don't follow symlink on umount */
1258 #define UMOUNT_UNUSED 0x80000000 /* Flag guaranteed to be unused */
1261 #define SB_I_CGROUPWB 0x00000001 /* cgroup-aware writeback enabled */
1262 #define SB_I_NOEXEC 0x00000002 /* Ignore executables on this fs */
1263 #define SB_I_NODEV 0x00000004 /* Ignore devices on this fs */
1265 /* sb->s_iflags to limit user namespace mounts */
1266 #define SB_I_USERNS_VISIBLE 0x00000010 /* fstype already mounted */
1268 /* Possible states of 'frozen' field */
1270 SB_UNFROZEN = 0, /* FS is unfrozen */
1271 SB_FREEZE_WRITE = 1, /* Writes, dir ops, ioctls frozen */
1272 SB_FREEZE_PAGEFAULT = 2, /* Page faults stopped as well */
1273 SB_FREEZE_FS = 3, /* For internal FS use (e.g. to stop
1274 * internal threads if needed) */
1275 SB_FREEZE_COMPLETE = 4, /* ->freeze_fs finished successfully */
1278 #define SB_FREEZE_LEVELS (SB_FREEZE_COMPLETE - 1)
1281 int frozen; /* Is sb frozen? */
1282 wait_queue_head_t wait_unfrozen; /* for get_super_thawed() */
1283 struct percpu_rw_semaphore rw_sem[SB_FREEZE_LEVELS];
1286 struct super_block {
1287 struct list_head s_list; /* Keep this first */
1288 dev_t s_dev; /* search index; _not_ kdev_t */
1289 unsigned char s_blocksize_bits;
1290 unsigned long s_blocksize;
1291 loff_t s_maxbytes; /* Max file size */
1292 struct file_system_type *s_type;
1293 const struct super_operations *s_op;
1294 const struct dquot_operations *dq_op;
1295 const struct quotactl_ops *s_qcop;
1296 const struct export_operations *s_export_op;
1297 unsigned long s_flags;
1298 unsigned long s_iflags; /* internal SB_I_* flags */
1299 unsigned long s_magic;
1300 struct dentry *s_root;
1301 struct rw_semaphore s_umount;
1304 #ifdef CONFIG_SECURITY
1307 const struct xattr_handler **s_xattr;
1309 const struct fscrypt_operations *s_cop;
1311 struct hlist_bl_head s_anon; /* anonymous dentries for (nfs) exporting */
1312 struct list_head s_mounts; /* list of mounts; _not_ for fs use */
1313 struct block_device *s_bdev;
1314 struct backing_dev_info *s_bdi;
1315 struct mtd_info *s_mtd;
1316 struct hlist_node s_instances;
1317 unsigned int s_quota_types; /* Bitmask of supported quota types */
1318 struct quota_info s_dquot; /* Diskquota specific options */
1320 struct sb_writers s_writers;
1322 char s_id[32]; /* Informational name */
1323 u8 s_uuid[16]; /* UUID */
1325 void *s_fs_info; /* Filesystem private info */
1326 unsigned int s_max_links;
1329 /* Granularity of c/m/atime in ns.
1330 Cannot be worse than a second */
1334 * The next field is for VFS *only*. No filesystems have any business
1335 * even looking at it. You had been warned.
1337 struct mutex s_vfs_rename_mutex; /* Kludge */
1340 * Filesystem subtype. If non-empty the filesystem type field
1341 * in /proc/mounts will be "type.subtype"
1346 * Saved mount options for lazy filesystems using
1347 * generic_show_options()
1349 char __rcu *s_options;
1350 const struct dentry_operations *s_d_op; /* default d_op for dentries */
1353 * Saved pool identifier for cleancache (-1 means none)
1355 int cleancache_poolid;
1357 struct shrinker s_shrink; /* per-sb shrinker handle */
1359 /* Number of inodes with nlink == 0 but still referenced */
1360 atomic_long_t s_remove_count;
1362 /* Being remounted read-only */
1363 int s_readonly_remount;
1365 /* AIO completions deferred from interrupt context */
1366 struct workqueue_struct *s_dio_done_wq;
1367 struct hlist_head s_pins;
1370 * Owning user namespace and default context in which to
1371 * interpret filesystem uids, gids, quotas, device nodes,
1372 * xattrs and security labels.
1374 struct user_namespace *s_user_ns;
1377 * Keep the lru lists last in the structure so they always sit on their
1378 * own individual cachelines.
1380 struct list_lru s_dentry_lru ____cacheline_aligned_in_smp;
1381 struct list_lru s_inode_lru ____cacheline_aligned_in_smp;
1382 struct rcu_head rcu;
1383 struct work_struct destroy_work;
1385 struct mutex s_sync_lock; /* sync serialisation lock */
1388 * Indicates how deep in a filesystem stack this SB is
1392 /* s_inode_list_lock protects s_inodes */
1393 spinlock_t s_inode_list_lock ____cacheline_aligned_in_smp;
1394 struct list_head s_inodes; /* all inodes */
1396 spinlock_t s_inode_wblist_lock;
1397 struct list_head s_inodes_wb; /* writeback inodes */
1400 /* Helper functions so that in most cases filesystems will
1401 * not need to deal directly with kuid_t and kgid_t and can
1402 * instead deal with the raw numeric values that are stored
1403 * in the filesystem.
1405 static inline uid_t i_uid_read(const struct inode *inode)
1407 return from_kuid(inode->i_sb->s_user_ns, inode->i_uid);
1410 static inline gid_t i_gid_read(const struct inode *inode)
1412 return from_kgid(inode->i_sb->s_user_ns, inode->i_gid);
1415 static inline void i_uid_write(struct inode *inode, uid_t uid)
1417 inode->i_uid = make_kuid(inode->i_sb->s_user_ns, uid);
1420 static inline void i_gid_write(struct inode *inode, gid_t gid)
1422 inode->i_gid = make_kgid(inode->i_sb->s_user_ns, gid);
1425 extern struct timespec current_fs_time(struct super_block *sb);
1426 extern struct timespec current_time(struct inode *inode);
1429 * Snapshotting support.
1432 void __sb_end_write(struct super_block *sb, int level);
1433 int __sb_start_write(struct super_block *sb, int level, bool wait);
1435 #define __sb_writers_acquired(sb, lev) \
1436 percpu_rwsem_acquire(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1437 #define __sb_writers_release(sb, lev) \
1438 percpu_rwsem_release(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1441 * sb_end_write - drop write access to a superblock
1442 * @sb: the super we wrote to
1444 * Decrement number of writers to the filesystem. Wake up possible waiters
1445 * wanting to freeze the filesystem.
1447 static inline void sb_end_write(struct super_block *sb)
1449 __sb_end_write(sb, SB_FREEZE_WRITE);
1453 * sb_end_pagefault - drop write access to a superblock from a page fault
1454 * @sb: the super we wrote to
1456 * Decrement number of processes handling write page fault to the filesystem.
1457 * Wake up possible waiters wanting to freeze the filesystem.
1459 static inline void sb_end_pagefault(struct super_block *sb)
1461 __sb_end_write(sb, SB_FREEZE_PAGEFAULT);
1465 * sb_end_intwrite - drop write access to a superblock for internal fs purposes
1466 * @sb: the super we wrote to
1468 * Decrement fs-internal number of writers to the filesystem. Wake up possible
1469 * waiters wanting to freeze the filesystem.
1471 static inline void sb_end_intwrite(struct super_block *sb)
1473 __sb_end_write(sb, SB_FREEZE_FS);
1477 * sb_start_write - get write access to a superblock
1478 * @sb: the super we write to
1480 * When a process wants to write data or metadata to a file system (i.e. dirty
1481 * a page or an inode), it should embed the operation in a sb_start_write() -
1482 * sb_end_write() pair to get exclusion against file system freezing. This
1483 * function increments number of writers preventing freezing. If the file
1484 * system is already frozen, the function waits until the file system is
1487 * Since freeze protection behaves as a lock, users have to preserve
1488 * ordering of freeze protection and other filesystem locks. Generally,
1489 * freeze protection should be the outermost lock. In particular, we have:
1492 * -> i_mutex (write path, truncate, directory ops, ...)
1493 * -> s_umount (freeze_super, thaw_super)
1495 static inline void sb_start_write(struct super_block *sb)
1497 __sb_start_write(sb, SB_FREEZE_WRITE, true);
1500 static inline int sb_start_write_trylock(struct super_block *sb)
1502 return __sb_start_write(sb, SB_FREEZE_WRITE, false);
1506 * sb_start_pagefault - get write access to a superblock from a page fault
1507 * @sb: the super we write to
1509 * When a process starts handling write page fault, it should embed the
1510 * operation into sb_start_pagefault() - sb_end_pagefault() pair to get
1511 * exclusion against file system freezing. This is needed since the page fault
1512 * is going to dirty a page. This function increments number of running page
1513 * faults preventing freezing. If the file system is already frozen, the
1514 * function waits until the file system is thawed.
1516 * Since page fault freeze protection behaves as a lock, users have to preserve
1517 * ordering of freeze protection and other filesystem locks. It is advised to
1518 * put sb_start_pagefault() close to mmap_sem in lock ordering. Page fault
1519 * handling code implies lock dependency:
1522 * -> sb_start_pagefault
1524 static inline void sb_start_pagefault(struct super_block *sb)
1526 __sb_start_write(sb, SB_FREEZE_PAGEFAULT, true);
1530 * sb_start_intwrite - get write access to a superblock for internal fs purposes
1531 * @sb: the super we write to
1533 * This is the third level of protection against filesystem freezing. It is
1534 * free for use by a filesystem. The only requirement is that it must rank
1535 * below sb_start_pagefault.
1537 * For example filesystem can call sb_start_intwrite() when starting a
1538 * transaction which somewhat eases handling of freezing for internal sources
1539 * of filesystem changes (internal fs threads, discarding preallocation on file
1542 static inline void sb_start_intwrite(struct super_block *sb)
1544 __sb_start_write(sb, SB_FREEZE_FS, true);
1548 extern bool inode_owner_or_capable(const struct inode *inode);
1551 * VFS helper functions..
1553 extern int vfs_create(struct inode *, struct dentry *, umode_t, bool);
1554 extern int vfs_mkdir(struct inode *, struct dentry *, umode_t);
1555 extern int vfs_mknod(struct inode *, struct dentry *, umode_t, dev_t);
1556 extern int vfs_symlink(struct inode *, struct dentry *, const char *);
1557 extern int vfs_link(struct dentry *, struct inode *, struct dentry *, struct inode **);
1558 extern int vfs_rmdir(struct inode *, struct dentry *);
1559 extern int vfs_unlink(struct inode *, struct dentry *, struct inode **);
1560 extern int vfs_rename(struct inode *, struct dentry *, struct inode *, struct dentry *, struct inode **, unsigned int);
1561 extern int vfs_whiteout(struct inode *, struct dentry *);
1564 * VFS file helper functions.
1566 extern void inode_init_owner(struct inode *inode, const struct inode *dir,
1568 extern bool may_open_dev(const struct path *path);
1570 * VFS FS_IOC_FIEMAP helper definitions.
1572 struct fiemap_extent_info {
1573 unsigned int fi_flags; /* Flags as passed from user */
1574 unsigned int fi_extents_mapped; /* Number of mapped extents */
1575 unsigned int fi_extents_max; /* Size of fiemap_extent array */
1576 struct fiemap_extent __user *fi_extents_start; /* Start of
1577 fiemap_extent array */
1579 int fiemap_fill_next_extent(struct fiemap_extent_info *info, u64 logical,
1580 u64 phys, u64 len, u32 flags);
1581 int fiemap_check_flags(struct fiemap_extent_info *fieinfo, u32 fs_flags);
1586 * NOTE! These match bits 12..15 of stat.st_mode
1587 * (ie "(i_mode >> 12) & 15").
1589 #define DT_UNKNOWN 0
1600 * This is the "filldir" function type, used by readdir() to let
1601 * the kernel specify what kind of dirent layout it wants to have.
1602 * This allows the kernel to read directories into kernel space or
1603 * to have different dirent layouts depending on the binary type.
1606 typedef int (*filldir_t)(struct dir_context *, const char *, int, loff_t, u64,
1609 struct dir_context {
1610 const filldir_t actor;
1614 struct block_device_operations;
1616 /* These macros are for out of kernel modules to test that
1617 * the kernel supports the unlocked_ioctl and compat_ioctl
1618 * fields in struct file_operations. */
1619 #define HAVE_COMPAT_IOCTL 1
1620 #define HAVE_UNLOCKED_IOCTL 1
1623 * These flags let !MMU mmap() govern direct device mapping vs immediate
1624 * copying more easily for MAP_PRIVATE, especially for ROM filesystems.
1626 * NOMMU_MAP_COPY: Copy can be mapped (MAP_PRIVATE)
1627 * NOMMU_MAP_DIRECT: Can be mapped directly (MAP_SHARED)
1628 * NOMMU_MAP_READ: Can be mapped for reading
1629 * NOMMU_MAP_WRITE: Can be mapped for writing
1630 * NOMMU_MAP_EXEC: Can be mapped for execution
1632 #define NOMMU_MAP_COPY 0x00000001
1633 #define NOMMU_MAP_DIRECT 0x00000008
1634 #define NOMMU_MAP_READ VM_MAYREAD
1635 #define NOMMU_MAP_WRITE VM_MAYWRITE
1636 #define NOMMU_MAP_EXEC VM_MAYEXEC
1638 #define NOMMU_VMFLAGS \
1639 (NOMMU_MAP_READ | NOMMU_MAP_WRITE | NOMMU_MAP_EXEC)
1644 struct file_operations {
1645 struct module *owner;
1646 loff_t (*llseek) (struct file *, loff_t, int);
1647 ssize_t (*read) (struct file *, char __user *, size_t, loff_t *);
1648 ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *);
1649 ssize_t (*read_iter) (struct kiocb *, struct iov_iter *);
1650 ssize_t (*write_iter) (struct kiocb *, struct iov_iter *);
1651 int (*iterate) (struct file *, struct dir_context *);
1652 int (*iterate_shared) (struct file *, struct dir_context *);
1653 unsigned int (*poll) (struct file *, struct poll_table_struct *);
1654 long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long);
1655 long (*compat_ioctl) (struct file *, unsigned int, unsigned long);
1656 int (*mmap) (struct file *, struct vm_area_struct *);
1657 int (*open) (struct inode *, struct file *);
1658 int (*flush) (struct file *, fl_owner_t id);
1659 int (*release) (struct inode *, struct file *);
1660 int (*fsync) (struct file *, loff_t, loff_t, int datasync);
1661 int (*fasync) (int, struct file *, int);
1662 int (*lock) (struct file *, int, struct file_lock *);
1663 ssize_t (*sendpage) (struct file *, struct page *, int, size_t, loff_t *, int);
1664 unsigned long (*get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
1665 int (*check_flags)(int);
1666 int (*flock) (struct file *, int, struct file_lock *);
1667 ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, loff_t *, size_t, unsigned int);
1668 ssize_t (*splice_read)(struct file *, loff_t *, struct pipe_inode_info *, size_t, unsigned int);
1669 int (*setlease)(struct file *, long, struct file_lock **, void **);
1670 long (*fallocate)(struct file *file, int mode, loff_t offset,
1672 void (*show_fdinfo)(struct seq_file *m, struct file *f);
1674 unsigned (*mmap_capabilities)(struct file *);
1676 ssize_t (*copy_file_range)(struct file *, loff_t, struct file *,
1677 loff_t, size_t, unsigned int);
1678 int (*clone_file_range)(struct file *, loff_t, struct file *, loff_t,
1680 ssize_t (*dedupe_file_range)(struct file *, u64, u64, struct file *,
1684 struct inode_operations {
1685 struct dentry * (*lookup) (struct inode *,struct dentry *, unsigned int);
1686 const char * (*get_link) (struct dentry *, struct inode *, struct delayed_call *);
1687 int (*permission) (struct inode *, int);
1688 struct posix_acl * (*get_acl)(struct inode *, int);
1690 int (*readlink) (struct dentry *, char __user *,int);
1692 int (*create) (struct inode *,struct dentry *, umode_t, bool);
1693 int (*link) (struct dentry *,struct inode *,struct dentry *);
1694 int (*unlink) (struct inode *,struct dentry *);
1695 int (*symlink) (struct inode *,struct dentry *,const char *);
1696 int (*mkdir) (struct inode *,struct dentry *,umode_t);
1697 int (*rmdir) (struct inode *,struct dentry *);
1698 int (*mknod) (struct inode *,struct dentry *,umode_t,dev_t);
1699 int (*rename) (struct inode *, struct dentry *,
1700 struct inode *, struct dentry *, unsigned int);
1701 int (*setattr) (struct dentry *, struct iattr *);
1702 int (*getattr) (struct vfsmount *mnt, struct dentry *, struct kstat *);
1703 ssize_t (*listxattr) (struct dentry *, char *, size_t);
1704 int (*fiemap)(struct inode *, struct fiemap_extent_info *, u64 start,
1706 int (*update_time)(struct inode *, struct timespec *, int);
1707 int (*atomic_open)(struct inode *, struct dentry *,
1708 struct file *, unsigned open_flag,
1709 umode_t create_mode, int *opened);
1710 int (*tmpfile) (struct inode *, struct dentry *, umode_t);
1711 int (*set_acl)(struct inode *, struct posix_acl *, int);
1712 } ____cacheline_aligned;
1714 ssize_t rw_copy_check_uvector(int type, const struct iovec __user * uvector,
1715 unsigned long nr_segs, unsigned long fast_segs,
1716 struct iovec *fast_pointer,
1717 struct iovec **ret_pointer);
1719 extern ssize_t __vfs_read(struct file *, char __user *, size_t, loff_t *);
1720 extern ssize_t __vfs_write(struct file *, const char __user *, size_t, loff_t *);
1721 extern ssize_t vfs_read(struct file *, char __user *, size_t, loff_t *);
1722 extern ssize_t vfs_write(struct file *, const char __user *, size_t, loff_t *);
1723 extern ssize_t vfs_readv(struct file *, const struct iovec __user *,
1724 unsigned long, loff_t *, int);
1725 extern ssize_t vfs_writev(struct file *, const struct iovec __user *,
1726 unsigned long, loff_t *, int);
1727 extern ssize_t vfs_copy_file_range(struct file *, loff_t , struct file *,
1728 loff_t, size_t, unsigned int);
1729 extern int vfs_clone_file_range(struct file *file_in, loff_t pos_in,
1730 struct file *file_out, loff_t pos_out, u64 len);
1731 extern int vfs_dedupe_file_range(struct file *file,
1732 struct file_dedupe_range *same);
1734 struct super_operations {
1735 struct inode *(*alloc_inode)(struct super_block *sb);
1736 void (*destroy_inode)(struct inode *);
1738 void (*dirty_inode) (struct inode *, int flags);
1739 int (*write_inode) (struct inode *, struct writeback_control *wbc);
1740 int (*drop_inode) (struct inode *);
1741 void (*evict_inode) (struct inode *);
1742 void (*put_super) (struct super_block *);
1743 int (*sync_fs)(struct super_block *sb, int wait);
1744 int (*freeze_super) (struct super_block *);
1745 int (*freeze_fs) (struct super_block *);
1746 int (*thaw_super) (struct super_block *);
1747 int (*unfreeze_fs) (struct super_block *);
1748 int (*statfs) (struct dentry *, struct kstatfs *);
1749 int (*remount_fs) (struct super_block *, int *, char *);
1750 void (*umount_begin) (struct super_block *);
1752 int (*show_options)(struct seq_file *, struct dentry *);
1753 int (*show_devname)(struct seq_file *, struct dentry *);
1754 int (*show_path)(struct seq_file *, struct dentry *);
1755 int (*show_stats)(struct seq_file *, struct dentry *);
1757 ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t);
1758 ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t);
1759 struct dquot **(*get_dquots)(struct inode *);
1761 int (*bdev_try_to_free_page)(struct super_block*, struct page*, gfp_t);
1762 long (*nr_cached_objects)(struct super_block *,
1763 struct shrink_control *);
1764 long (*free_cached_objects)(struct super_block *,
1765 struct shrink_control *);
1769 * Inode flags - they have no relation to superblock flags now
1771 #define S_SYNC 1 /* Writes are synced at once */
1772 #define S_NOATIME 2 /* Do not update access times */
1773 #define S_APPEND 4 /* Append-only file */
1774 #define S_IMMUTABLE 8 /* Immutable file */
1775 #define S_DEAD 16 /* removed, but still open directory */
1776 #define S_NOQUOTA 32 /* Inode is not counted to quota */
1777 #define S_DIRSYNC 64 /* Directory modifications are synchronous */
1778 #define S_NOCMTIME 128 /* Do not update file c/mtime */
1779 #define S_SWAPFILE 256 /* Do not truncate: swapon got its bmaps */
1780 #define S_PRIVATE 512 /* Inode is fs-internal */
1781 #define S_IMA 1024 /* Inode has an associated IMA struct */
1782 #define S_AUTOMOUNT 2048 /* Automount/referral quasi-directory */
1783 #define S_NOSEC 4096 /* no suid or xattr security attributes */
1784 #ifdef CONFIG_FS_DAX
1785 #define S_DAX 8192 /* Direct Access, avoiding the page cache */
1787 #define S_DAX 0 /* Make all the DAX code disappear */
1791 * Note that nosuid etc flags are inode-specific: setting some file-system
1792 * flags just means all the inodes inherit those flags by default. It might be
1793 * possible to override it selectively if you really wanted to with some
1794 * ioctl() that is not currently implemented.
1796 * Exception: MS_RDONLY is always applied to the entire file system.
1798 * Unfortunately, it is possible to change a filesystems flags with it mounted
1799 * with files in use. This means that all of the inodes will not have their
1800 * i_flags updated. Hence, i_flags no longer inherit the superblock mount
1801 * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org
1803 #define __IS_FLG(inode, flg) ((inode)->i_sb->s_flags & (flg))
1805 #define IS_RDONLY(inode) ((inode)->i_sb->s_flags & MS_RDONLY)
1806 #define IS_SYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS) || \
1807 ((inode)->i_flags & S_SYNC))
1808 #define IS_DIRSYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS|MS_DIRSYNC) || \
1809 ((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
1810 #define IS_MANDLOCK(inode) __IS_FLG(inode, MS_MANDLOCK)
1811 #define IS_NOATIME(inode) __IS_FLG(inode, MS_RDONLY|MS_NOATIME)
1812 #define IS_I_VERSION(inode) __IS_FLG(inode, MS_I_VERSION)
1814 #define IS_NOQUOTA(inode) ((inode)->i_flags & S_NOQUOTA)
1815 #define IS_APPEND(inode) ((inode)->i_flags & S_APPEND)
1816 #define IS_IMMUTABLE(inode) ((inode)->i_flags & S_IMMUTABLE)
1817 #define IS_POSIXACL(inode) __IS_FLG(inode, MS_POSIXACL)
1819 #define IS_DEADDIR(inode) ((inode)->i_flags & S_DEAD)
1820 #define IS_NOCMTIME(inode) ((inode)->i_flags & S_NOCMTIME)
1821 #define IS_SWAPFILE(inode) ((inode)->i_flags & S_SWAPFILE)
1822 #define IS_PRIVATE(inode) ((inode)->i_flags & S_PRIVATE)
1823 #define IS_IMA(inode) ((inode)->i_flags & S_IMA)
1824 #define IS_AUTOMOUNT(inode) ((inode)->i_flags & S_AUTOMOUNT)
1825 #define IS_NOSEC(inode) ((inode)->i_flags & S_NOSEC)
1826 #define IS_DAX(inode) ((inode)->i_flags & S_DAX)
1828 #define IS_WHITEOUT(inode) (S_ISCHR(inode->i_mode) && \
1829 (inode)->i_rdev == WHITEOUT_DEV)
1831 static inline bool HAS_UNMAPPED_ID(struct inode *inode)
1833 return !uid_valid(inode->i_uid) || !gid_valid(inode->i_gid);
1837 * Inode state bits. Protected by inode->i_lock
1839 * Three bits determine the dirty state of the inode, I_DIRTY_SYNC,
1840 * I_DIRTY_DATASYNC and I_DIRTY_PAGES.
1842 * Four bits define the lifetime of an inode. Initially, inodes are I_NEW,
1843 * until that flag is cleared. I_WILL_FREE, I_FREEING and I_CLEAR are set at
1844 * various stages of removing an inode.
1846 * Two bits are used for locking and completion notification, I_NEW and I_SYNC.
1848 * I_DIRTY_SYNC Inode is dirty, but doesn't have to be written on
1849 * fdatasync(). i_atime is the usual cause.
1850 * I_DIRTY_DATASYNC Data-related inode changes pending. We keep track of
1851 * these changes separately from I_DIRTY_SYNC so that we
1852 * don't have to write inode on fdatasync() when only
1853 * mtime has changed in it.
1854 * I_DIRTY_PAGES Inode has dirty pages. Inode itself may be clean.
1855 * I_NEW Serves as both a mutex and completion notification.
1856 * New inodes set I_NEW. If two processes both create
1857 * the same inode, one of them will release its inode and
1858 * wait for I_NEW to be released before returning.
1859 * Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can
1860 * also cause waiting on I_NEW, without I_NEW actually
1861 * being set. find_inode() uses this to prevent returning
1862 * nearly-dead inodes.
1863 * I_WILL_FREE Must be set when calling write_inode_now() if i_count
1864 * is zero. I_FREEING must be set when I_WILL_FREE is
1866 * I_FREEING Set when inode is about to be freed but still has dirty
1867 * pages or buffers attached or the inode itself is still
1869 * I_CLEAR Added by clear_inode(). In this state the inode is
1870 * clean and can be destroyed. Inode keeps I_FREEING.
1872 * Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are
1873 * prohibited for many purposes. iget() must wait for
1874 * the inode to be completely released, then create it
1875 * anew. Other functions will just ignore such inodes,
1876 * if appropriate. I_NEW is used for waiting.
1878 * I_SYNC Writeback of inode is running. The bit is set during
1879 * data writeback, and cleared with a wakeup on the bit
1880 * address once it is done. The bit is also used to pin
1881 * the inode in memory for flusher thread.
1883 * I_REFERENCED Marks the inode as recently references on the LRU list.
1885 * I_DIO_WAKEUP Never set. Only used as a key for wait_on_bit().
1887 * I_WB_SWITCH Cgroup bdi_writeback switching in progress. Used to
1888 * synchronize competing switching instances and to tell
1889 * wb stat updates to grab mapping->tree_lock. See
1890 * inode_switch_wb_work_fn() for details.
1892 * Q: What is the difference between I_WILL_FREE and I_FREEING?
1894 #define I_DIRTY_SYNC (1 << 0)
1895 #define I_DIRTY_DATASYNC (1 << 1)
1896 #define I_DIRTY_PAGES (1 << 2)
1898 #define I_NEW (1 << __I_NEW)
1899 #define I_WILL_FREE (1 << 4)
1900 #define I_FREEING (1 << 5)
1901 #define I_CLEAR (1 << 6)
1903 #define I_SYNC (1 << __I_SYNC)
1904 #define I_REFERENCED (1 << 8)
1905 #define __I_DIO_WAKEUP 9
1906 #define I_DIO_WAKEUP (1 << __I_DIO_WAKEUP)
1907 #define I_LINKABLE (1 << 10)
1908 #define I_DIRTY_TIME (1 << 11)
1909 #define __I_DIRTY_TIME_EXPIRED 12
1910 #define I_DIRTY_TIME_EXPIRED (1 << __I_DIRTY_TIME_EXPIRED)
1911 #define I_WB_SWITCH (1 << 13)
1913 #define I_DIRTY (I_DIRTY_SYNC | I_DIRTY_DATASYNC | I_DIRTY_PAGES)
1914 #define I_DIRTY_ALL (I_DIRTY | I_DIRTY_TIME)
1916 extern void __mark_inode_dirty(struct inode *, int);
1917 static inline void mark_inode_dirty(struct inode *inode)
1919 __mark_inode_dirty(inode, I_DIRTY);
1922 static inline void mark_inode_dirty_sync(struct inode *inode)
1924 __mark_inode_dirty(inode, I_DIRTY_SYNC);
1927 extern void inc_nlink(struct inode *inode);
1928 extern void drop_nlink(struct inode *inode);
1929 extern void clear_nlink(struct inode *inode);
1930 extern void set_nlink(struct inode *inode, unsigned int nlink);
1932 static inline void inode_inc_link_count(struct inode *inode)
1935 mark_inode_dirty(inode);
1938 static inline void inode_dec_link_count(struct inode *inode)
1941 mark_inode_dirty(inode);
1945 * inode_inc_iversion - increments i_version
1946 * @inode: inode that need to be updated
1948 * Every time the inode is modified, the i_version field will be incremented.
1949 * The filesystem has to be mounted with i_version flag
1952 static inline void inode_inc_iversion(struct inode *inode)
1954 spin_lock(&inode->i_lock);
1956 spin_unlock(&inode->i_lock);
1959 enum file_time_flags {
1966 extern void touch_atime(const struct path *);
1967 static inline void file_accessed(struct file *file)
1969 if (!(file->f_flags & O_NOATIME))
1970 touch_atime(&file->f_path);
1973 int sync_inode(struct inode *inode, struct writeback_control *wbc);
1974 int sync_inode_metadata(struct inode *inode, int wait);
1976 struct file_system_type {
1979 #define FS_REQUIRES_DEV 1
1980 #define FS_BINARY_MOUNTDATA 2
1981 #define FS_HAS_SUBTYPE 4
1982 #define FS_USERNS_MOUNT 8 /* Can be mounted by userns root */
1983 #define FS_RENAME_DOES_D_MOVE 32768 /* FS will handle d_move() during rename() internally. */
1984 struct dentry *(*mount) (struct file_system_type *, int,
1985 const char *, void *);
1986 void (*kill_sb) (struct super_block *);
1987 struct module *owner;
1988 struct file_system_type * next;
1989 struct hlist_head fs_supers;
1991 struct lock_class_key s_lock_key;
1992 struct lock_class_key s_umount_key;
1993 struct lock_class_key s_vfs_rename_key;
1994 struct lock_class_key s_writers_key[SB_FREEZE_LEVELS];
1996 struct lock_class_key i_lock_key;
1997 struct lock_class_key i_mutex_key;
1998 struct lock_class_key i_mutex_dir_key;
2001 #define MODULE_ALIAS_FS(NAME) MODULE_ALIAS("fs-" NAME)
2003 extern struct dentry *mount_ns(struct file_system_type *fs_type,
2004 int flags, void *data, void *ns, struct user_namespace *user_ns,
2005 int (*fill_super)(struct super_block *, void *, int));
2006 extern struct dentry *mount_bdev(struct file_system_type *fs_type,
2007 int flags, const char *dev_name, void *data,
2008 int (*fill_super)(struct super_block *, void *, int));
2009 extern struct dentry *mount_single(struct file_system_type *fs_type,
2010 int flags, void *data,
2011 int (*fill_super)(struct super_block *, void *, int));
2012 extern struct dentry *mount_nodev(struct file_system_type *fs_type,
2013 int flags, void *data,
2014 int (*fill_super)(struct super_block *, void *, int));
2015 extern struct dentry *mount_subtree(struct vfsmount *mnt, const char *path);
2016 void generic_shutdown_super(struct super_block *sb);
2017 void kill_block_super(struct super_block *sb);
2018 void kill_anon_super(struct super_block *sb);
2019 void kill_litter_super(struct super_block *sb);
2020 void deactivate_super(struct super_block *sb);
2021 void deactivate_locked_super(struct super_block *sb);
2022 int set_anon_super(struct super_block *s, void *data);
2023 int get_anon_bdev(dev_t *);
2024 void free_anon_bdev(dev_t);
2025 struct super_block *sget_userns(struct file_system_type *type,
2026 int (*test)(struct super_block *,void *),
2027 int (*set)(struct super_block *,void *),
2028 int flags, struct user_namespace *user_ns,
2030 struct super_block *sget(struct file_system_type *type,
2031 int (*test)(struct super_block *,void *),
2032 int (*set)(struct super_block *,void *),
2033 int flags, void *data);
2034 extern struct dentry *mount_pseudo_xattr(struct file_system_type *, char *,
2035 const struct super_operations *ops,
2036 const struct xattr_handler **xattr,
2037 const struct dentry_operations *dops,
2040 static inline struct dentry *
2041 mount_pseudo(struct file_system_type *fs_type, char *name,
2042 const struct super_operations *ops,
2043 const struct dentry_operations *dops, unsigned long magic)
2045 return mount_pseudo_xattr(fs_type, name, ops, NULL, dops, magic);
2048 /* Alas, no aliases. Too much hassle with bringing module.h everywhere */
2049 #define fops_get(fops) \
2050 (((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
2051 #define fops_put(fops) \
2052 do { if (fops) module_put((fops)->owner); } while(0)
2054 * This one is to be used *ONLY* from ->open() instances.
2055 * fops must be non-NULL, pinned down *and* module dependencies
2056 * should be sufficient to pin the caller down as well.
2058 #define replace_fops(f, fops) \
2060 struct file *__file = (f); \
2061 fops_put(__file->f_op); \
2062 BUG_ON(!(__file->f_op = (fops))); \
2065 extern int register_filesystem(struct file_system_type *);
2066 extern int unregister_filesystem(struct file_system_type *);
2067 extern struct vfsmount *kern_mount_data(struct file_system_type *, void *data);
2068 #define kern_mount(type) kern_mount_data(type, NULL)
2069 extern void kern_unmount(struct vfsmount *mnt);
2070 extern int may_umount_tree(struct vfsmount *);
2071 extern int may_umount(struct vfsmount *);
2072 extern long do_mount(const char *, const char __user *,
2073 const char *, unsigned long, void *);
2074 extern struct vfsmount *collect_mounts(struct path *);
2075 extern void drop_collected_mounts(struct vfsmount *);
2076 extern int iterate_mounts(int (*)(struct vfsmount *, void *), void *,
2078 extern int vfs_statfs(struct path *, struct kstatfs *);
2079 extern int user_statfs(const char __user *, struct kstatfs *);
2080 extern int fd_statfs(int, struct kstatfs *);
2081 extern int vfs_ustat(dev_t, struct kstatfs *);
2082 extern int freeze_super(struct super_block *super);
2083 extern int thaw_super(struct super_block *super);
2084 extern bool our_mnt(struct vfsmount *mnt);
2086 extern int current_umask(void);
2088 extern void ihold(struct inode * inode);
2089 extern void iput(struct inode *);
2090 extern int generic_update_time(struct inode *, struct timespec *, int);
2093 extern struct kobject *fs_kobj;
2095 #define MAX_RW_COUNT (INT_MAX & PAGE_MASK)
2097 #ifdef CONFIG_MANDATORY_FILE_LOCKING
2098 extern int locks_mandatory_locked(struct file *);
2099 extern int locks_mandatory_area(struct inode *, struct file *, loff_t, loff_t, unsigned char);
2102 * Candidates for mandatory locking have the setgid bit set
2103 * but no group execute bit - an otherwise meaningless combination.
2106 static inline int __mandatory_lock(struct inode *ino)
2108 return (ino->i_mode & (S_ISGID | S_IXGRP)) == S_ISGID;
2112 * ... and these candidates should be on MS_MANDLOCK mounted fs,
2113 * otherwise these will be advisory locks
2116 static inline int mandatory_lock(struct inode *ino)
2118 return IS_MANDLOCK(ino) && __mandatory_lock(ino);
2121 static inline int locks_verify_locked(struct file *file)
2123 if (mandatory_lock(locks_inode(file)))
2124 return locks_mandatory_locked(file);
2128 static inline int locks_verify_truncate(struct inode *inode,
2132 if (!inode->i_flctx || !mandatory_lock(inode))
2135 if (size < inode->i_size) {
2136 return locks_mandatory_area(inode, f, size, inode->i_size - 1,
2139 return locks_mandatory_area(inode, f, inode->i_size, size - 1,
2144 #else /* !CONFIG_MANDATORY_FILE_LOCKING */
2146 static inline int locks_mandatory_locked(struct file *file)
2151 static inline int locks_mandatory_area(struct inode *inode, struct file *filp,
2152 loff_t start, loff_t end, unsigned char type)
2157 static inline int __mandatory_lock(struct inode *inode)
2162 static inline int mandatory_lock(struct inode *inode)
2167 static inline int locks_verify_locked(struct file *file)
2172 static inline int locks_verify_truncate(struct inode *inode, struct file *filp,
2178 #endif /* CONFIG_MANDATORY_FILE_LOCKING */
2181 #ifdef CONFIG_FILE_LOCKING
2182 static inline int break_lease(struct inode *inode, unsigned int mode)
2185 * Since this check is lockless, we must ensure that any refcounts
2186 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2187 * could end up racing with tasks trying to set a new lease on this
2191 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2192 return __break_lease(inode, mode, FL_LEASE);
2196 static inline int break_deleg(struct inode *inode, unsigned int mode)
2199 * Since this check is lockless, we must ensure that any refcounts
2200 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2201 * could end up racing with tasks trying to set a new lease on this
2205 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2206 return __break_lease(inode, mode, FL_DELEG);
2210 static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode)
2214 ret = break_deleg(inode, O_WRONLY|O_NONBLOCK);
2215 if (ret == -EWOULDBLOCK && delegated_inode) {
2216 *delegated_inode = inode;
2222 static inline int break_deleg_wait(struct inode **delegated_inode)
2226 ret = break_deleg(*delegated_inode, O_WRONLY);
2227 iput(*delegated_inode);
2228 *delegated_inode = NULL;
2232 static inline int break_layout(struct inode *inode, bool wait)
2235 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2236 return __break_lease(inode,
2237 wait ? O_WRONLY : O_WRONLY | O_NONBLOCK,
2242 #else /* !CONFIG_FILE_LOCKING */
2243 static inline int break_lease(struct inode *inode, unsigned int mode)
2248 static inline int break_deleg(struct inode *inode, unsigned int mode)
2253 static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode)
2258 static inline int break_deleg_wait(struct inode **delegated_inode)
2264 static inline int break_layout(struct inode *inode, bool wait)
2269 #endif /* CONFIG_FILE_LOCKING */
2274 const char *name; /* pointer to actual string */
2275 const __user char *uptr; /* original userland pointer */
2276 struct audit_names *aname;
2281 extern long vfs_truncate(const struct path *, loff_t);
2282 extern int do_truncate(struct dentry *, loff_t start, unsigned int time_attrs,
2284 extern int vfs_fallocate(struct file *file, int mode, loff_t offset,
2286 extern long do_sys_open(int dfd, const char __user *filename, int flags,
2288 extern struct file *file_open_name(struct filename *, int, umode_t);
2289 extern struct file *filp_open(const char *, int, umode_t);
2290 extern struct file *file_open_root(struct dentry *, struct vfsmount *,
2291 const char *, int, umode_t);
2292 extern struct file * dentry_open(const struct path *, int, const struct cred *);
2293 extern int filp_close(struct file *, fl_owner_t id);
2295 extern struct filename *getname_flags(const char __user *, int, int *);
2296 extern struct filename *getname(const char __user *);
2297 extern struct filename *getname_kernel(const char *);
2298 extern void putname(struct filename *name);
2304 extern int finish_open(struct file *file, struct dentry *dentry,
2305 int (*open)(struct inode *, struct file *),
2307 extern int finish_no_open(struct file *file, struct dentry *dentry);
2311 extern int ioctl_preallocate(struct file *filp, void __user *argp);
2314 extern void __init vfs_caches_init_early(void);
2315 extern void __init vfs_caches_init(void);
2317 extern struct kmem_cache *names_cachep;
2319 #define __getname() kmem_cache_alloc(names_cachep, GFP_KERNEL)
2320 #define __putname(name) kmem_cache_free(names_cachep, (void *)(name))
2323 extern int register_blkdev(unsigned int, const char *);
2324 extern void unregister_blkdev(unsigned int, const char *);
2325 extern struct block_device *bdget(dev_t);
2326 extern struct block_device *bdgrab(struct block_device *bdev);
2327 extern void bd_set_size(struct block_device *, loff_t size);
2328 extern void bd_forget(struct inode *inode);
2329 extern void bdput(struct block_device *);
2330 extern void invalidate_bdev(struct block_device *);
2331 extern void iterate_bdevs(void (*)(struct block_device *, void *), void *);
2332 extern int sync_blockdev(struct block_device *bdev);
2333 extern void kill_bdev(struct block_device *);
2334 extern struct super_block *freeze_bdev(struct block_device *);
2335 extern void emergency_thaw_all(void);
2336 extern int thaw_bdev(struct block_device *bdev, struct super_block *sb);
2337 extern int fsync_bdev(struct block_device *);
2339 extern struct super_block *blockdev_superblock;
2341 static inline bool sb_is_blkdev_sb(struct super_block *sb)
2343 return sb == blockdev_superblock;
2346 static inline void bd_forget(struct inode *inode) {}
2347 static inline int sync_blockdev(struct block_device *bdev) { return 0; }
2348 static inline void kill_bdev(struct block_device *bdev) {}
2349 static inline void invalidate_bdev(struct block_device *bdev) {}
2351 static inline struct super_block *freeze_bdev(struct block_device *sb)
2356 static inline int thaw_bdev(struct block_device *bdev, struct super_block *sb)
2361 static inline void iterate_bdevs(void (*f)(struct block_device *, void *), void *arg)
2365 static inline bool sb_is_blkdev_sb(struct super_block *sb)
2370 extern int sync_filesystem(struct super_block *);
2371 extern const struct file_operations def_blk_fops;
2372 extern const struct file_operations def_chr_fops;
2374 extern int ioctl_by_bdev(struct block_device *, unsigned, unsigned long);
2375 extern int blkdev_ioctl(struct block_device *, fmode_t, unsigned, unsigned long);
2376 extern long compat_blkdev_ioctl(struct file *, unsigned, unsigned long);
2377 extern int blkdev_get(struct block_device *bdev, fmode_t mode, void *holder);
2378 extern struct block_device *blkdev_get_by_path(const char *path, fmode_t mode,
2380 extern struct block_device *blkdev_get_by_dev(dev_t dev, fmode_t mode,
2382 extern void blkdev_put(struct block_device *bdev, fmode_t mode);
2383 extern int __blkdev_reread_part(struct block_device *bdev);
2384 extern int blkdev_reread_part(struct block_device *bdev);
2387 extern int bd_link_disk_holder(struct block_device *bdev, struct gendisk *disk);
2388 extern void bd_unlink_disk_holder(struct block_device *bdev,
2389 struct gendisk *disk);
2391 static inline int bd_link_disk_holder(struct block_device *bdev,
2392 struct gendisk *disk)
2396 static inline void bd_unlink_disk_holder(struct block_device *bdev,
2397 struct gendisk *disk)
2404 #define CHRDEV_MAJOR_HASH_SIZE 255
2405 /* Marks the bottom of the first segment of free char majors */
2406 #define CHRDEV_MAJOR_DYN_END 234
2407 extern int alloc_chrdev_region(dev_t *, unsigned, unsigned, const char *);
2408 extern int register_chrdev_region(dev_t, unsigned, const char *);
2409 extern int __register_chrdev(unsigned int major, unsigned int baseminor,
2410 unsigned int count, const char *name,
2411 const struct file_operations *fops);
2412 extern void __unregister_chrdev(unsigned int major, unsigned int baseminor,
2413 unsigned int count, const char *name);
2414 extern void unregister_chrdev_region(dev_t, unsigned);
2415 extern void chrdev_show(struct seq_file *,off_t);
2417 static inline int register_chrdev(unsigned int major, const char *name,
2418 const struct file_operations *fops)
2420 return __register_chrdev(major, 0, 256, name, fops);
2423 static inline void unregister_chrdev(unsigned int major, const char *name)
2425 __unregister_chrdev(major, 0, 256, name);
2428 /* fs/block_dev.c */
2429 #define BDEVNAME_SIZE 32 /* Largest string for a blockdev identifier */
2430 #define BDEVT_SIZE 10 /* Largest string for MAJ:MIN for blkdev */
2433 #define BLKDEV_MAJOR_HASH_SIZE 255
2434 extern const char *__bdevname(dev_t, char *buffer);
2435 extern const char *bdevname(struct block_device *bdev, char *buffer);
2436 extern struct block_device *lookup_bdev(const char *);
2437 extern void blkdev_show(struct seq_file *,off_t);
2440 #define BLKDEV_MAJOR_HASH_SIZE 0
2443 extern void init_special_inode(struct inode *, umode_t, dev_t);
2445 /* Invalid inode operations -- fs/bad_inode.c */
2446 extern void make_bad_inode(struct inode *);
2447 extern bool is_bad_inode(struct inode *);
2450 extern void check_disk_size_change(struct gendisk *disk,
2451 struct block_device *bdev);
2452 extern int revalidate_disk(struct gendisk *);
2453 extern int check_disk_change(struct block_device *);
2454 extern int __invalidate_device(struct block_device *, bool);
2455 extern int invalidate_partition(struct gendisk *, int);
2457 unsigned long invalidate_mapping_pages(struct address_space *mapping,
2458 pgoff_t start, pgoff_t end);
2460 static inline void invalidate_remote_inode(struct inode *inode)
2462 if (S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) ||
2463 S_ISLNK(inode->i_mode))
2464 invalidate_mapping_pages(inode->i_mapping, 0, -1);
2466 extern int invalidate_inode_pages2(struct address_space *mapping);
2467 extern int invalidate_inode_pages2_range(struct address_space *mapping,
2468 pgoff_t start, pgoff_t end);
2469 extern int write_inode_now(struct inode *, int);
2470 extern int filemap_fdatawrite(struct address_space *);
2471 extern int filemap_flush(struct address_space *);
2472 extern int filemap_fdatawait(struct address_space *);
2473 extern void filemap_fdatawait_keep_errors(struct address_space *);
2474 extern int filemap_fdatawait_range(struct address_space *, loff_t lstart,
2476 extern int filemap_write_and_wait(struct address_space *mapping);
2477 extern int filemap_write_and_wait_range(struct address_space *mapping,
2478 loff_t lstart, loff_t lend);
2479 extern int __filemap_fdatawrite_range(struct address_space *mapping,
2480 loff_t start, loff_t end, int sync_mode);
2481 extern int filemap_fdatawrite_range(struct address_space *mapping,
2482 loff_t start, loff_t end);
2483 extern int filemap_check_errors(struct address_space *mapping);
2485 extern int vfs_fsync_range(struct file *file, loff_t start, loff_t end,
2487 extern int vfs_fsync(struct file *file, int datasync);
2490 * Sync the bytes written if this was a synchronous write. Expect ki_pos
2491 * to already be updated for the write, and will return either the amount
2492 * of bytes passed in, or an error if syncing the file failed.
2494 static inline ssize_t generic_write_sync(struct kiocb *iocb, ssize_t count)
2496 if (iocb->ki_flags & IOCB_DSYNC) {
2497 int ret = vfs_fsync_range(iocb->ki_filp,
2498 iocb->ki_pos - count, iocb->ki_pos - 1,
2499 (iocb->ki_flags & IOCB_SYNC) ? 0 : 1);
2507 extern void emergency_sync(void);
2508 extern void emergency_remount(void);
2510 extern sector_t bmap(struct inode *, sector_t);
2512 extern int notify_change(struct dentry *, struct iattr *, struct inode **);
2513 extern int inode_permission(struct inode *, int);
2514 extern int __inode_permission(struct inode *, int);
2515 extern int generic_permission(struct inode *, int);
2516 extern int __check_sticky(struct inode *dir, struct inode *inode);
2518 static inline bool execute_ok(struct inode *inode)
2520 return (inode->i_mode & S_IXUGO) || S_ISDIR(inode->i_mode);
2523 static inline void file_start_write(struct file *file)
2525 if (!S_ISREG(file_inode(file)->i_mode))
2527 __sb_start_write(file_inode(file)->i_sb, SB_FREEZE_WRITE, true);
2530 static inline bool file_start_write_trylock(struct file *file)
2532 if (!S_ISREG(file_inode(file)->i_mode))
2534 return __sb_start_write(file_inode(file)->i_sb, SB_FREEZE_WRITE, false);
2537 static inline void file_end_write(struct file *file)
2539 if (!S_ISREG(file_inode(file)->i_mode))
2541 __sb_end_write(file_inode(file)->i_sb, SB_FREEZE_WRITE);
2545 * get_write_access() gets write permission for a file.
2546 * put_write_access() releases this write permission.
2547 * This is used for regular files.
2548 * We cannot support write (and maybe mmap read-write shared) accesses and
2549 * MAP_DENYWRITE mmappings simultaneously. The i_writecount field of an inode
2550 * can have the following values:
2551 * 0: no writers, no VM_DENYWRITE mappings
2552 * < 0: (-i_writecount) vm_area_structs with VM_DENYWRITE set exist
2553 * > 0: (i_writecount) users are writing to the file.
2555 * Normally we operate on that counter with atomic_{inc,dec} and it's safe
2556 * except for the cases where we don't hold i_writecount yet. Then we need to
2557 * use {get,deny}_write_access() - these functions check the sign and refuse
2558 * to do the change if sign is wrong.
2560 static inline int get_write_access(struct inode *inode)
2562 return atomic_inc_unless_negative(&inode->i_writecount) ? 0 : -ETXTBSY;
2564 static inline int deny_write_access(struct file *file)
2566 struct inode *inode = file_inode(file);
2567 return atomic_dec_unless_positive(&inode->i_writecount) ? 0 : -ETXTBSY;
2569 static inline void put_write_access(struct inode * inode)
2571 atomic_dec(&inode->i_writecount);
2573 static inline void allow_write_access(struct file *file)
2576 atomic_inc(&file_inode(file)->i_writecount);
2578 static inline bool inode_is_open_for_write(const struct inode *inode)
2580 return atomic_read(&inode->i_writecount) > 0;
2584 static inline void i_readcount_dec(struct inode *inode)
2586 BUG_ON(!atomic_read(&inode->i_readcount));
2587 atomic_dec(&inode->i_readcount);
2589 static inline void i_readcount_inc(struct inode *inode)
2591 atomic_inc(&inode->i_readcount);
2594 static inline void i_readcount_dec(struct inode *inode)
2598 static inline void i_readcount_inc(struct inode *inode)
2603 extern int do_pipe_flags(int *, int);
2605 #define __kernel_read_file_id(id) \
2606 id(UNKNOWN, unknown) \
2607 id(FIRMWARE, firmware) \
2608 id(FIRMWARE_PREALLOC_BUFFER, firmware) \
2609 id(MODULE, kernel-module) \
2610 id(KEXEC_IMAGE, kexec-image) \
2611 id(KEXEC_INITRAMFS, kexec-initramfs) \
2612 id(POLICY, security-policy) \
2615 #define __fid_enumify(ENUM, dummy) READING_ ## ENUM,
2616 #define __fid_stringify(dummy, str) #str,
2618 enum kernel_read_file_id {
2619 __kernel_read_file_id(__fid_enumify)
2622 static const char * const kernel_read_file_str[] = {
2623 __kernel_read_file_id(__fid_stringify)
2626 static inline const char *kernel_read_file_id_str(enum kernel_read_file_id id)
2628 if (id < 0 || id >= READING_MAX_ID)
2629 return kernel_read_file_str[READING_UNKNOWN];
2631 return kernel_read_file_str[id];
2634 extern int kernel_read(struct file *, loff_t, char *, unsigned long);
2635 extern int kernel_read_file(struct file *, void **, loff_t *, loff_t,
2636 enum kernel_read_file_id);
2637 extern int kernel_read_file_from_path(char *, void **, loff_t *, loff_t,
2638 enum kernel_read_file_id);
2639 extern int kernel_read_file_from_fd(int, void **, loff_t *, loff_t,
2640 enum kernel_read_file_id);
2641 extern ssize_t kernel_write(struct file *, const char *, size_t, loff_t);
2642 extern ssize_t __kernel_write(struct file *, const char *, size_t, loff_t *);
2643 extern struct file * open_exec(const char *);
2645 /* fs/dcache.c -- generic fs support functions */
2646 extern bool is_subdir(struct dentry *, struct dentry *);
2647 extern bool path_is_under(struct path *, struct path *);
2649 extern char *file_path(struct file *, char *, int);
2651 #include <linux/err.h>
2653 /* needed for stackable file system support */
2654 extern loff_t default_llseek(struct file *file, loff_t offset, int whence);
2656 extern loff_t vfs_llseek(struct file *file, loff_t offset, int whence);
2658 extern int inode_init_always(struct super_block *, struct inode *);
2659 extern void inode_init_once(struct inode *);
2660 extern void address_space_init_once(struct address_space *mapping);
2661 extern struct inode * igrab(struct inode *);
2662 extern ino_t iunique(struct super_block *, ino_t);
2663 extern int inode_needs_sync(struct inode *inode);
2664 extern int generic_delete_inode(struct inode *inode);
2665 static inline int generic_drop_inode(struct inode *inode)
2667 return !inode->i_nlink || inode_unhashed(inode);
2670 extern struct inode *ilookup5_nowait(struct super_block *sb,
2671 unsigned long hashval, int (*test)(struct inode *, void *),
2673 extern struct inode *ilookup5(struct super_block *sb, unsigned long hashval,
2674 int (*test)(struct inode *, void *), void *data);
2675 extern struct inode *ilookup(struct super_block *sb, unsigned long ino);
2677 extern struct inode * iget5_locked(struct super_block *, unsigned long, int (*test)(struct inode *, void *), int (*set)(struct inode *, void *), void *);
2678 extern struct inode * iget_locked(struct super_block *, unsigned long);
2679 extern struct inode *find_inode_nowait(struct super_block *,
2681 int (*match)(struct inode *,
2682 unsigned long, void *),
2684 extern int insert_inode_locked4(struct inode *, unsigned long, int (*test)(struct inode *, void *), void *);
2685 extern int insert_inode_locked(struct inode *);
2686 #ifdef CONFIG_DEBUG_LOCK_ALLOC
2687 extern void lockdep_annotate_inode_mutex_key(struct inode *inode);
2689 static inline void lockdep_annotate_inode_mutex_key(struct inode *inode) { };
2691 extern void unlock_new_inode(struct inode *);
2692 extern unsigned int get_next_ino(void);
2694 extern void __iget(struct inode * inode);
2695 extern void iget_failed(struct inode *);
2696 extern void clear_inode(struct inode *);
2697 extern void __destroy_inode(struct inode *);
2698 extern struct inode *new_inode_pseudo(struct super_block *sb);
2699 extern struct inode *new_inode(struct super_block *sb);
2700 extern void free_inode_nonrcu(struct inode *inode);
2701 extern int should_remove_suid(struct dentry *);
2702 extern int file_remove_privs(struct file *);
2704 extern void __insert_inode_hash(struct inode *, unsigned long hashval);
2705 static inline void insert_inode_hash(struct inode *inode)
2707 __insert_inode_hash(inode, inode->i_ino);
2710 extern void __remove_inode_hash(struct inode *);
2711 static inline void remove_inode_hash(struct inode *inode)
2713 if (!inode_unhashed(inode) && !hlist_fake(&inode->i_hash))
2714 __remove_inode_hash(inode);
2717 extern void inode_sb_list_add(struct inode *inode);
2720 extern int bdev_read_only(struct block_device *);
2722 extern int set_blocksize(struct block_device *, int);
2723 extern int sb_set_blocksize(struct super_block *, int);
2724 extern int sb_min_blocksize(struct super_block *, int);
2726 extern int generic_file_mmap(struct file *, struct vm_area_struct *);
2727 extern int generic_file_readonly_mmap(struct file *, struct vm_area_struct *);
2728 extern ssize_t generic_write_checks(struct kiocb *, struct iov_iter *);
2729 extern ssize_t generic_file_read_iter(struct kiocb *, struct iov_iter *);
2730 extern ssize_t __generic_file_write_iter(struct kiocb *, struct iov_iter *);
2731 extern ssize_t generic_file_write_iter(struct kiocb *, struct iov_iter *);
2732 extern ssize_t generic_file_direct_write(struct kiocb *, struct iov_iter *);
2733 extern ssize_t generic_perform_write(struct file *, struct iov_iter *, loff_t);
2735 ssize_t vfs_iter_read(struct file *file, struct iov_iter *iter, loff_t *ppos);
2736 ssize_t vfs_iter_write(struct file *file, struct iov_iter *iter, loff_t *ppos);
2738 /* fs/block_dev.c */
2739 extern ssize_t blkdev_read_iter(struct kiocb *iocb, struct iov_iter *to);
2740 extern ssize_t blkdev_write_iter(struct kiocb *iocb, struct iov_iter *from);
2741 extern int blkdev_fsync(struct file *filp, loff_t start, loff_t end,
2743 extern void block_sync_page(struct page *page);
2746 extern ssize_t generic_file_splice_read(struct file *, loff_t *,
2747 struct pipe_inode_info *, size_t, unsigned int);
2748 extern ssize_t iter_file_splice_write(struct pipe_inode_info *,
2749 struct file *, loff_t *, size_t, unsigned int);
2750 extern ssize_t generic_splice_sendpage(struct pipe_inode_info *pipe,
2751 struct file *out, loff_t *, size_t len, unsigned int flags);
2752 extern long do_splice_direct(struct file *in, loff_t *ppos, struct file *out,
2753 loff_t *opos, size_t len, unsigned int flags);
2757 file_ra_state_init(struct file_ra_state *ra, struct address_space *mapping);
2758 extern loff_t noop_llseek(struct file *file, loff_t offset, int whence);
2759 extern loff_t no_llseek(struct file *file, loff_t offset, int whence);
2760 extern loff_t vfs_setpos(struct file *file, loff_t offset, loff_t maxsize);
2761 extern loff_t generic_file_llseek(struct file *file, loff_t offset, int whence);
2762 extern loff_t generic_file_llseek_size(struct file *file, loff_t offset,
2763 int whence, loff_t maxsize, loff_t eof);
2764 extern loff_t fixed_size_llseek(struct file *file, loff_t offset,
2765 int whence, loff_t size);
2766 extern loff_t no_seek_end_llseek_size(struct file *, loff_t, int, loff_t);
2767 extern loff_t no_seek_end_llseek(struct file *, loff_t, int);
2768 extern int generic_file_open(struct inode * inode, struct file * filp);
2769 extern int nonseekable_open(struct inode * inode, struct file * filp);
2772 typedef void (dio_submit_t)(struct bio *bio, struct inode *inode,
2773 loff_t file_offset);
2776 /* need locking between buffered and direct access */
2779 /* filesystem does not support filling holes */
2780 DIO_SKIP_HOLES = 0x02,
2782 /* filesystem can handle aio writes beyond i_size */
2783 DIO_ASYNC_EXTEND = 0x04,
2785 /* inode/fs/bdev does not need truncate protection */
2786 DIO_SKIP_DIO_COUNT = 0x08,
2789 void dio_end_io(struct bio *bio, int error);
2791 ssize_t __blockdev_direct_IO(struct kiocb *iocb, struct inode *inode,
2792 struct block_device *bdev, struct iov_iter *iter,
2793 get_block_t get_block,
2794 dio_iodone_t end_io, dio_submit_t submit_io,
2797 static inline ssize_t blockdev_direct_IO(struct kiocb *iocb,
2798 struct inode *inode,
2799 struct iov_iter *iter,
2800 get_block_t get_block)
2802 return __blockdev_direct_IO(iocb, inode, inode->i_sb->s_bdev, iter,
2803 get_block, NULL, NULL, DIO_LOCKING | DIO_SKIP_HOLES);
2807 void inode_dio_wait(struct inode *inode);
2810 * inode_dio_begin - signal start of a direct I/O requests
2811 * @inode: inode the direct I/O happens on
2813 * This is called once we've finished processing a direct I/O request,
2814 * and is used to wake up callers waiting for direct I/O to be quiesced.
2816 static inline void inode_dio_begin(struct inode *inode)
2818 atomic_inc(&inode->i_dio_count);
2822 * inode_dio_end - signal finish of a direct I/O requests
2823 * @inode: inode the direct I/O happens on
2825 * This is called once we've finished processing a direct I/O request,
2826 * and is used to wake up callers waiting for direct I/O to be quiesced.
2828 static inline void inode_dio_end(struct inode *inode)
2830 if (atomic_dec_and_test(&inode->i_dio_count))
2831 wake_up_bit(&inode->i_state, __I_DIO_WAKEUP);
2834 extern void inode_set_flags(struct inode *inode, unsigned int flags,
2837 extern const struct file_operations generic_ro_fops;
2839 #define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
2841 extern int readlink_copy(char __user *, int, const char *);
2842 extern int page_readlink(struct dentry *, char __user *, int);
2843 extern const char *page_get_link(struct dentry *, struct inode *,
2844 struct delayed_call *);
2845 extern void page_put_link(void *);
2846 extern int __page_symlink(struct inode *inode, const char *symname, int len,
2848 extern int page_symlink(struct inode *inode, const char *symname, int len);
2849 extern const struct inode_operations page_symlink_inode_operations;
2850 extern void kfree_link(void *);
2851 extern int generic_readlink(struct dentry *, char __user *, int);
2852 extern void generic_fillattr(struct inode *, struct kstat *);
2853 int vfs_getattr_nosec(struct path *path, struct kstat *stat);
2854 extern int vfs_getattr(struct path *, struct kstat *);
2855 void __inode_add_bytes(struct inode *inode, loff_t bytes);
2856 void inode_add_bytes(struct inode *inode, loff_t bytes);
2857 void __inode_sub_bytes(struct inode *inode, loff_t bytes);
2858 void inode_sub_bytes(struct inode *inode, loff_t bytes);
2859 loff_t inode_get_bytes(struct inode *inode);
2860 void inode_set_bytes(struct inode *inode, loff_t bytes);
2861 const char *simple_get_link(struct dentry *, struct inode *,
2862 struct delayed_call *);
2863 extern const struct inode_operations simple_symlink_inode_operations;
2865 extern int iterate_dir(struct file *, struct dir_context *);
2867 extern int vfs_stat(const char __user *, struct kstat *);
2868 extern int vfs_lstat(const char __user *, struct kstat *);
2869 extern int vfs_fstat(unsigned int, struct kstat *);
2870 extern int vfs_fstatat(int , const char __user *, struct kstat *, int);
2871 extern const char *vfs_get_link(struct dentry *, struct delayed_call *);
2873 extern int __generic_block_fiemap(struct inode *inode,
2874 struct fiemap_extent_info *fieinfo,
2875 loff_t start, loff_t len,
2876 get_block_t *get_block);
2877 extern int generic_block_fiemap(struct inode *inode,
2878 struct fiemap_extent_info *fieinfo, u64 start,
2879 u64 len, get_block_t *get_block);
2881 extern void get_filesystem(struct file_system_type *fs);
2882 extern void put_filesystem(struct file_system_type *fs);
2883 extern struct file_system_type *get_fs_type(const char *name);
2884 extern struct super_block *get_super(struct block_device *);
2885 extern struct super_block *get_super_thawed(struct block_device *);
2886 extern struct super_block *get_active_super(struct block_device *bdev);
2887 extern void drop_super(struct super_block *sb);
2888 extern void iterate_supers(void (*)(struct super_block *, void *), void *);
2889 extern void iterate_supers_type(struct file_system_type *,
2890 void (*)(struct super_block *, void *), void *);
2892 extern int dcache_dir_open(struct inode *, struct file *);
2893 extern int dcache_dir_close(struct inode *, struct file *);
2894 extern loff_t dcache_dir_lseek(struct file *, loff_t, int);
2895 extern int dcache_readdir(struct file *, struct dir_context *);
2896 extern int simple_setattr(struct dentry *, struct iattr *);
2897 extern int simple_getattr(struct vfsmount *, struct dentry *, struct kstat *);
2898 extern int simple_statfs(struct dentry *, struct kstatfs *);
2899 extern int simple_open(struct inode *inode, struct file *file);
2900 extern int simple_link(struct dentry *, struct inode *, struct dentry *);
2901 extern int simple_unlink(struct inode *, struct dentry *);
2902 extern int simple_rmdir(struct inode *, struct dentry *);
2903 extern int simple_rename(struct inode *, struct dentry *,
2904 struct inode *, struct dentry *, unsigned int);
2905 extern int noop_fsync(struct file *, loff_t, loff_t, int);
2906 extern int simple_empty(struct dentry *);
2907 extern int simple_readpage(struct file *file, struct page *page);
2908 extern int simple_write_begin(struct file *file, struct address_space *mapping,
2909 loff_t pos, unsigned len, unsigned flags,
2910 struct page **pagep, void **fsdata);
2911 extern int simple_write_end(struct file *file, struct address_space *mapping,
2912 loff_t pos, unsigned len, unsigned copied,
2913 struct page *page, void *fsdata);
2914 extern int always_delete_dentry(const struct dentry *);
2915 extern struct inode *alloc_anon_inode(struct super_block *);
2916 extern int simple_nosetlease(struct file *, long, struct file_lock **, void **);
2917 extern const struct dentry_operations simple_dentry_operations;
2919 extern struct dentry *simple_lookup(struct inode *, struct dentry *, unsigned int flags);
2920 extern ssize_t generic_read_dir(struct file *, char __user *, size_t, loff_t *);
2921 extern const struct file_operations simple_dir_operations;
2922 extern const struct inode_operations simple_dir_inode_operations;
2923 extern void make_empty_dir_inode(struct inode *inode);
2924 extern bool is_empty_dir_inode(struct inode *inode);
2925 struct tree_descr { char *name; const struct file_operations *ops; int mode; };
2926 struct dentry *d_alloc_name(struct dentry *, const char *);
2927 extern int simple_fill_super(struct super_block *, unsigned long, struct tree_descr *);
2928 extern int simple_pin_fs(struct file_system_type *, struct vfsmount **mount, int *count);
2929 extern void simple_release_fs(struct vfsmount **mount, int *count);
2931 extern ssize_t simple_read_from_buffer(void __user *to, size_t count,
2932 loff_t *ppos, const void *from, size_t available);
2933 extern ssize_t simple_write_to_buffer(void *to, size_t available, loff_t *ppos,
2934 const void __user *from, size_t count);
2936 extern int __generic_file_fsync(struct file *, loff_t, loff_t, int);
2937 extern int generic_file_fsync(struct file *, loff_t, loff_t, int);
2939 extern int generic_check_addressable(unsigned, u64);
2941 #ifdef CONFIG_MIGRATION
2942 extern int buffer_migrate_page(struct address_space *,
2943 struct page *, struct page *,
2946 #define buffer_migrate_page NULL
2949 extern int setattr_prepare(struct dentry *, struct iattr *);
2950 extern int inode_newsize_ok(const struct inode *, loff_t offset);
2951 extern void setattr_copy(struct inode *inode, const struct iattr *attr);
2953 extern int file_update_time(struct file *file);
2955 extern int generic_show_options(struct seq_file *m, struct dentry *root);
2956 extern void save_mount_options(struct super_block *sb, char *options);
2957 extern void replace_mount_options(struct super_block *sb, char *options);
2959 static inline bool io_is_direct(struct file *filp)
2961 return (filp->f_flags & O_DIRECT) || IS_DAX(filp->f_mapping->host);
2964 static inline int iocb_flags(struct file *file)
2967 if (file->f_flags & O_APPEND)
2969 if (io_is_direct(file))
2971 if ((file->f_flags & O_DSYNC) || IS_SYNC(file->f_mapping->host))
2973 if (file->f_flags & __O_SYNC)
2978 static inline ino_t parent_ino(struct dentry *dentry)
2983 * Don't strictly need d_lock here? If the parent ino could change
2984 * then surely we'd have a deeper race in the caller?
2986 spin_lock(&dentry->d_lock);
2987 res = dentry->d_parent->d_inode->i_ino;
2988 spin_unlock(&dentry->d_lock);
2992 /* Transaction based IO helpers */
2995 * An argresp is stored in an allocated page and holds the
2996 * size of the argument or response, along with its content
2998 struct simple_transaction_argresp {
3003 #define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
3005 char *simple_transaction_get(struct file *file, const char __user *buf,
3007 ssize_t simple_transaction_read(struct file *file, char __user *buf,
3008 size_t size, loff_t *pos);
3009 int simple_transaction_release(struct inode *inode, struct file *file);
3011 void simple_transaction_set(struct file *file, size_t n);
3014 * simple attribute files
3016 * These attributes behave similar to those in sysfs:
3018 * Writing to an attribute immediately sets a value, an open file can be
3019 * written to multiple times.
3021 * Reading from an attribute creates a buffer from the value that might get
3022 * read with multiple read calls. When the attribute has been read
3023 * completely, no further read calls are possible until the file is opened
3026 * All attributes contain a text representation of a numeric value
3027 * that are accessed with the get() and set() functions.
3029 #define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt) \
3030 static int __fops ## _open(struct inode *inode, struct file *file) \
3032 __simple_attr_check_format(__fmt, 0ull); \
3033 return simple_attr_open(inode, file, __get, __set, __fmt); \
3035 static const struct file_operations __fops = { \
3036 .owner = THIS_MODULE, \
3037 .open = __fops ## _open, \
3038 .release = simple_attr_release, \
3039 .read = simple_attr_read, \
3040 .write = simple_attr_write, \
3041 .llseek = generic_file_llseek, \
3044 static inline __printf(1, 2)
3045 void __simple_attr_check_format(const char *fmt, ...)
3047 /* don't do anything, just let the compiler check the arguments; */
3050 int simple_attr_open(struct inode *inode, struct file *file,
3051 int (*get)(void *, u64 *), int (*set)(void *, u64),
3053 int simple_attr_release(struct inode *inode, struct file *file);
3054 ssize_t simple_attr_read(struct file *file, char __user *buf,
3055 size_t len, loff_t *ppos);
3056 ssize_t simple_attr_write(struct file *file, const char __user *buf,
3057 size_t len, loff_t *ppos);
3060 int proc_nr_files(struct ctl_table *table, int write,
3061 void __user *buffer, size_t *lenp, loff_t *ppos);
3062 int proc_nr_dentry(struct ctl_table *table, int write,
3063 void __user *buffer, size_t *lenp, loff_t *ppos);
3064 int proc_nr_inodes(struct ctl_table *table, int write,
3065 void __user *buffer, size_t *lenp, loff_t *ppos);
3066 int __init get_filesystem_list(char *buf);
3068 #define __FMODE_EXEC ((__force int) FMODE_EXEC)
3069 #define __FMODE_NONOTIFY ((__force int) FMODE_NONOTIFY)
3071 #define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE])
3072 #define OPEN_FMODE(flag) ((__force fmode_t)(((flag + 1) & O_ACCMODE) | \
3073 (flag & __FMODE_NONOTIFY)))
3075 static inline bool is_sxid(umode_t mode)
3077 return (mode & S_ISUID) || ((mode & S_ISGID) && (mode & S_IXGRP));
3080 static inline int check_sticky(struct inode *dir, struct inode *inode)
3082 if (!(dir->i_mode & S_ISVTX))
3085 return __check_sticky(dir, inode);
3088 static inline void inode_has_no_xattr(struct inode *inode)
3090 if (!is_sxid(inode->i_mode) && (inode->i_sb->s_flags & MS_NOSEC))
3091 inode->i_flags |= S_NOSEC;
3094 static inline bool is_root_inode(struct inode *inode)
3096 return inode == inode->i_sb->s_root->d_inode;
3099 static inline bool dir_emit(struct dir_context *ctx,
3100 const char *name, int namelen,
3101 u64 ino, unsigned type)
3103 return ctx->actor(ctx, name, namelen, ctx->pos, ino, type) == 0;
3105 static inline bool dir_emit_dot(struct file *file, struct dir_context *ctx)
3107 return ctx->actor(ctx, ".", 1, ctx->pos,
3108 file->f_path.dentry->d_inode->i_ino, DT_DIR) == 0;
3110 static inline bool dir_emit_dotdot(struct file *file, struct dir_context *ctx)
3112 return ctx->actor(ctx, "..", 2, ctx->pos,
3113 parent_ino(file->f_path.dentry), DT_DIR) == 0;
3115 static inline bool dir_emit_dots(struct file *file, struct dir_context *ctx)
3117 if (ctx->pos == 0) {
3118 if (!dir_emit_dot(file, ctx))
3122 if (ctx->pos == 1) {
3123 if (!dir_emit_dotdot(file, ctx))
3129 static inline bool dir_relax(struct inode *inode)
3131 inode_unlock(inode);
3133 return !IS_DEADDIR(inode);
3136 static inline bool dir_relax_shared(struct inode *inode)
3138 inode_unlock_shared(inode);
3139 inode_lock_shared(inode);
3140 return !IS_DEADDIR(inode);
3143 extern bool path_noexec(const struct path *path);
3144 extern void inode_nohighmem(struct inode *inode);
3146 #endif /* _LINUX_FS_H */