5 #include <linux/linkage.h>
6 #include <linux/wait.h>
7 #include <linux/kdev_t.h>
8 #include <linux/dcache.h>
9 #include <linux/path.h>
10 #include <linux/stat.h>
11 #include <linux/cache.h>
12 #include <linux/list.h>
13 #include <linux/list_lru.h>
14 #include <linux/llist.h>
15 #include <linux/radix-tree.h>
16 #include <linux/rbtree.h>
17 #include <linux/init.h>
18 #include <linux/pid.h>
19 #include <linux/bug.h>
20 #include <linux/mutex.h>
21 #include <linux/rwsem.h>
22 #include <linux/capability.h>
23 #include <linux/semaphore.h>
24 #include <linux/fiemap.h>
25 #include <linux/rculist_bl.h>
26 #include <linux/atomic.h>
27 #include <linux/shrinker.h>
28 #include <linux/migrate_mode.h>
29 #include <linux/uidgid.h>
30 #include <linux/lockdep.h>
31 #include <linux/percpu-rwsem.h>
32 #include <linux/blk_types.h>
33 #include <linux/workqueue.h>
35 #include <asm/byteorder.h>
36 #include <uapi/linux/fs.h>
38 struct backing_dev_info;
40 struct export_operations;
45 struct pipe_inode_info;
46 struct poll_table_struct;
48 struct vm_area_struct;
51 struct swap_info_struct;
53 struct workqueue_struct;
57 extern void __init inode_init(void);
58 extern void __init inode_init_early(void);
59 extern void __init files_init(void);
60 extern void __init files_maxfiles_init(void);
62 extern struct files_stat_struct files_stat;
63 extern unsigned long get_max_files(void);
64 extern int sysctl_nr_open;
65 extern struct inodes_stat_t inodes_stat;
66 extern int leases_enable, lease_break_time;
67 extern int sysctl_protected_symlinks;
68 extern int sysctl_protected_hardlinks;
71 typedef int (get_block_t)(struct inode *inode, sector_t iblock,
72 struct buffer_head *bh_result, int create);
73 typedef void (dio_iodone_t)(struct kiocb *iocb, loff_t offset,
74 ssize_t bytes, void *private);
75 typedef void (dax_iodone_t)(struct buffer_head *bh_map, int uptodate);
77 #define MAY_EXEC 0x00000001
78 #define MAY_WRITE 0x00000002
79 #define MAY_READ 0x00000004
80 #define MAY_APPEND 0x00000008
81 #define MAY_ACCESS 0x00000010
82 #define MAY_OPEN 0x00000020
83 #define MAY_CHDIR 0x00000040
84 /* called from RCU mode, don't block */
85 #define MAY_NOT_BLOCK 0x00000080
88 * flags in file.f_mode. Note that FMODE_READ and FMODE_WRITE must correspond
89 * to O_WRONLY and O_RDWR via the strange trick in __dentry_open()
92 /* file is open for reading */
93 #define FMODE_READ ((__force fmode_t)0x1)
94 /* file is open for writing */
95 #define FMODE_WRITE ((__force fmode_t)0x2)
96 /* file is seekable */
97 #define FMODE_LSEEK ((__force fmode_t)0x4)
98 /* file can be accessed using pread */
99 #define FMODE_PREAD ((__force fmode_t)0x8)
100 /* file can be accessed using pwrite */
101 #define FMODE_PWRITE ((__force fmode_t)0x10)
102 /* File is opened for execution with sys_execve / sys_uselib */
103 #define FMODE_EXEC ((__force fmode_t)0x20)
104 /* File is opened with O_NDELAY (only set for block devices) */
105 #define FMODE_NDELAY ((__force fmode_t)0x40)
106 /* File is opened with O_EXCL (only set for block devices) */
107 #define FMODE_EXCL ((__force fmode_t)0x80)
108 /* File is opened using open(.., 3, ..) and is writeable only for ioctls
109 (specialy hack for floppy.c) */
110 #define FMODE_WRITE_IOCTL ((__force fmode_t)0x100)
111 /* 32bit hashes as llseek() offset (for directories) */
112 #define FMODE_32BITHASH ((__force fmode_t)0x200)
113 /* 64bit hashes as llseek() offset (for directories) */
114 #define FMODE_64BITHASH ((__force fmode_t)0x400)
117 * Don't update ctime and mtime.
119 * Currently a special hack for the XFS open_by_handle ioctl, but we'll
120 * hopefully graduate it to a proper O_CMTIME flag supported by open(2) soon.
122 #define FMODE_NOCMTIME ((__force fmode_t)0x800)
124 /* Expect random access pattern */
125 #define FMODE_RANDOM ((__force fmode_t)0x1000)
127 /* File is huge (eg. /dev/kmem): treat loff_t as unsigned */
128 #define FMODE_UNSIGNED_OFFSET ((__force fmode_t)0x2000)
130 /* File is opened with O_PATH; almost nothing can be done with it */
131 #define FMODE_PATH ((__force fmode_t)0x4000)
133 /* File needs atomic accesses to f_pos */
134 #define FMODE_ATOMIC_POS ((__force fmode_t)0x8000)
135 /* Write access to underlying fs */
136 #define FMODE_WRITER ((__force fmode_t)0x10000)
137 /* Has read method(s) */
138 #define FMODE_CAN_READ ((__force fmode_t)0x20000)
139 /* Has write method(s) */
140 #define FMODE_CAN_WRITE ((__force fmode_t)0x40000)
142 /* File was opened by fanotify and shouldn't generate fanotify events */
143 #define FMODE_NONOTIFY ((__force fmode_t)0x4000000)
146 * Flag for rw_copy_check_uvector and compat_rw_copy_check_uvector
147 * that indicates that they should check the contents of the iovec are
148 * valid, but not check the memory that the iovec elements
151 #define CHECK_IOVEC_ONLY -1
154 * The below are the various read and write types that we support. Some of
155 * them include behavioral modifiers that send information down to the
156 * block layer and IO scheduler. Terminology:
158 * The block layer uses device plugging to defer IO a little bit, in
159 * the hope that we will see more IO very shortly. This increases
160 * coalescing of adjacent IO and thus reduces the number of IOs we
161 * have to send to the device. It also allows for better queuing,
162 * if the IO isn't mergeable. If the caller is going to be waiting
163 * for the IO, then he must ensure that the device is unplugged so
164 * that the IO is dispatched to the driver.
166 * All IO is handled async in Linux. This is fine for background
167 * writes, but for reads or writes that someone waits for completion
168 * on, we want to notify the block layer and IO scheduler so that they
169 * know about it. That allows them to make better scheduling
170 * decisions. So when the below references 'sync' and 'async', it
171 * is referencing this priority hint.
173 * With that in mind, the available types are:
175 * READ A normal read operation. Device will be plugged.
176 * READ_SYNC A synchronous read. Device is not plugged, caller can
177 * immediately wait on this read without caring about
179 * READA Used for read-ahead operations. Lower priority, and the
180 * block layer could (in theory) choose to ignore this
181 * request if it runs into resource problems.
182 * WRITE A normal async write. Device will be plugged.
183 * WRITE_SYNC Synchronous write. Identical to WRITE, but passes down
184 * the hint that someone will be waiting on this IO
185 * shortly. The write equivalent of READ_SYNC.
186 * WRITE_ODIRECT Special case write for O_DIRECT only.
187 * WRITE_FLUSH Like WRITE_SYNC but with preceding cache flush.
188 * WRITE_FUA Like WRITE_SYNC but data is guaranteed to be on
189 * non-volatile media on completion.
190 * WRITE_FLUSH_FUA Combination of WRITE_FLUSH and FUA. The IO is preceded
191 * by a cache flush and data is guaranteed to be on
192 * non-volatile media on completion.
195 #define RW_MASK REQ_WRITE
196 #define RWA_MASK REQ_RAHEAD
199 #define WRITE RW_MASK
200 #define READA RWA_MASK
202 #define READ_SYNC (READ | REQ_SYNC)
203 #define WRITE_SYNC (WRITE | REQ_SYNC | REQ_NOIDLE)
204 #define WRITE_ODIRECT (WRITE | REQ_SYNC)
205 #define WRITE_FLUSH (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FLUSH)
206 #define WRITE_FUA (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FUA)
207 #define WRITE_FLUSH_FUA (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FLUSH | REQ_FUA)
210 * Attribute flags. These should be or-ed together to figure out what
213 #define ATTR_MODE (1 << 0)
214 #define ATTR_UID (1 << 1)
215 #define ATTR_GID (1 << 2)
216 #define ATTR_SIZE (1 << 3)
217 #define ATTR_ATIME (1 << 4)
218 #define ATTR_MTIME (1 << 5)
219 #define ATTR_CTIME (1 << 6)
220 #define ATTR_ATIME_SET (1 << 7)
221 #define ATTR_MTIME_SET (1 << 8)
222 #define ATTR_FORCE (1 << 9) /* Not a change, but a change it */
223 #define ATTR_ATTR_FLAG (1 << 10)
224 #define ATTR_KILL_SUID (1 << 11)
225 #define ATTR_KILL_SGID (1 << 12)
226 #define ATTR_FILE (1 << 13)
227 #define ATTR_KILL_PRIV (1 << 14)
228 #define ATTR_OPEN (1 << 15) /* Truncating from open(O_TRUNC) */
229 #define ATTR_TIMES_SET (1 << 16)
232 * Whiteout is represented by a char device. The following constants define the
233 * mode and device number to use.
235 #define WHITEOUT_MODE 0
236 #define WHITEOUT_DEV 0
239 * This is the Inode Attributes structure, used for notify_change(). It
240 * uses the above definitions as flags, to know which values have changed.
241 * Also, in this manner, a Filesystem can look at only the values it cares
242 * about. Basically, these are the attributes that the VFS layer can
243 * request to change from the FS layer.
245 * Derek Atkins <warlord@MIT.EDU> 94-10-20
248 unsigned int ia_valid;
253 struct timespec ia_atime;
254 struct timespec ia_mtime;
255 struct timespec ia_ctime;
258 * Not an attribute, but an auxiliary info for filesystems wanting to
259 * implement an ftruncate() like method. NOTE: filesystem should
260 * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL).
262 struct file *ia_file;
266 * Includes for diskquotas.
268 #include <linux/quota.h>
271 * Maximum number of layers of fs stack. Needs to be limited to
272 * prevent kernel stack overflow
274 #define FILESYSTEM_MAX_STACK_DEPTH 2
277 * enum positive_aop_returns - aop return codes with specific semantics
279 * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has
280 * completed, that the page is still locked, and
281 * should be considered active. The VM uses this hint
282 * to return the page to the active list -- it won't
283 * be a candidate for writeback again in the near
284 * future. Other callers must be careful to unlock
285 * the page if they get this return. Returned by
288 * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has
289 * unlocked it and the page might have been truncated.
290 * The caller should back up to acquiring a new page and
291 * trying again. The aop will be taking reasonable
292 * precautions not to livelock. If the caller held a page
293 * reference, it should drop it before retrying. Returned
296 * address_space_operation functions return these large constants to indicate
297 * special semantics to the caller. These are much larger than the bytes in a
298 * page to allow for functions that return the number of bytes operated on in a
302 enum positive_aop_returns {
303 AOP_WRITEPAGE_ACTIVATE = 0x80000,
304 AOP_TRUNCATED_PAGE = 0x80001,
307 #define AOP_FLAG_UNINTERRUPTIBLE 0x0001 /* will not do a short write */
308 #define AOP_FLAG_CONT_EXPAND 0x0002 /* called from cont_expand */
309 #define AOP_FLAG_NOFS 0x0004 /* used by filesystem to direct
310 * helper code (eg buffer layer)
311 * to clear GFP_FS from alloc */
314 * oh the beauties of C type declarations.
317 struct address_space;
318 struct writeback_control;
320 #define IOCB_EVENTFD (1 << 0)
321 #define IOCB_APPEND (1 << 1)
322 #define IOCB_DIRECT (1 << 2)
325 struct file *ki_filp;
327 void (*ki_complete)(struct kiocb *iocb, long ret, long ret2);
332 static inline bool is_sync_kiocb(struct kiocb *kiocb)
334 return kiocb->ki_complete == NULL;
337 static inline int iocb_flags(struct file *file);
339 static inline void init_sync_kiocb(struct kiocb *kiocb, struct file *filp)
341 *kiocb = (struct kiocb) {
343 .ki_flags = iocb_flags(filp),
348 * "descriptor" for what we're up to with a read.
349 * This allows us to use the same read code yet
350 * have multiple different users of the data that
351 * we read from a file.
353 * The simplest case just copies the data to user
366 typedef int (*read_actor_t)(read_descriptor_t *, struct page *,
367 unsigned long, unsigned long);
369 struct address_space_operations {
370 int (*writepage)(struct page *page, struct writeback_control *wbc);
371 int (*readpage)(struct file *, struct page *);
373 /* Write back some dirty pages from this mapping. */
374 int (*writepages)(struct address_space *, struct writeback_control *);
376 /* Set a page dirty. Return true if this dirtied it */
377 int (*set_page_dirty)(struct page *page);
379 int (*readpages)(struct file *filp, struct address_space *mapping,
380 struct list_head *pages, unsigned nr_pages);
382 int (*write_begin)(struct file *, struct address_space *mapping,
383 loff_t pos, unsigned len, unsigned flags,
384 struct page **pagep, void **fsdata);
385 int (*write_end)(struct file *, struct address_space *mapping,
386 loff_t pos, unsigned len, unsigned copied,
387 struct page *page, void *fsdata);
389 /* Unfortunately this kludge is needed for FIBMAP. Don't use it */
390 sector_t (*bmap)(struct address_space *, sector_t);
391 void (*invalidatepage) (struct page *, unsigned int, unsigned int);
392 int (*releasepage) (struct page *, gfp_t);
393 void (*freepage)(struct page *);
394 ssize_t (*direct_IO)(struct kiocb *, struct iov_iter *iter, loff_t offset);
396 * migrate the contents of a page to the specified target. If
397 * migrate_mode is MIGRATE_ASYNC, it must not block.
399 int (*migratepage) (struct address_space *,
400 struct page *, struct page *, enum migrate_mode);
401 int (*launder_page) (struct page *);
402 int (*is_partially_uptodate) (struct page *, unsigned long,
404 void (*is_dirty_writeback) (struct page *, bool *, bool *);
405 int (*error_remove_page)(struct address_space *, struct page *);
407 /* swapfile support */
408 int (*swap_activate)(struct swap_info_struct *sis, struct file *file,
410 void (*swap_deactivate)(struct file *file);
413 extern const struct address_space_operations empty_aops;
416 * pagecache_write_begin/pagecache_write_end must be used by general code
417 * to write into the pagecache.
419 int pagecache_write_begin(struct file *, struct address_space *mapping,
420 loff_t pos, unsigned len, unsigned flags,
421 struct page **pagep, void **fsdata);
423 int pagecache_write_end(struct file *, struct address_space *mapping,
424 loff_t pos, unsigned len, unsigned copied,
425 struct page *page, void *fsdata);
427 struct address_space {
428 struct inode *host; /* owner: inode, block_device */
429 struct radix_tree_root page_tree; /* radix tree of all pages */
430 spinlock_t tree_lock; /* and lock protecting it */
431 atomic_t i_mmap_writable;/* count VM_SHARED mappings */
432 struct rb_root i_mmap; /* tree of private and shared mappings */
433 struct rw_semaphore i_mmap_rwsem; /* protect tree, count, list */
434 /* Protected by tree_lock together with the radix tree */
435 unsigned long nrpages; /* number of total pages */
436 unsigned long nrshadows; /* number of shadow entries */
437 pgoff_t writeback_index;/* writeback starts here */
438 const struct address_space_operations *a_ops; /* methods */
439 unsigned long flags; /* error bits/gfp mask */
440 spinlock_t private_lock; /* for use by the address_space */
441 struct list_head private_list; /* ditto */
442 void *private_data; /* ditto */
443 } __attribute__((aligned(sizeof(long))));
445 * On most architectures that alignment is already the case; but
446 * must be enforced here for CRIS, to let the least significant bit
447 * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
449 struct request_queue;
451 struct block_device {
452 dev_t bd_dev; /* not a kdev_t - it's a search key */
454 struct inode * bd_inode; /* will die */
455 struct super_block * bd_super;
456 struct mutex bd_mutex; /* open/close mutex */
457 struct list_head bd_inodes;
461 bool bd_write_holder;
463 struct list_head bd_holder_disks;
465 struct block_device * bd_contains;
466 unsigned bd_block_size;
467 struct hd_struct * bd_part;
468 /* number of times partitions within this device have been opened. */
469 unsigned bd_part_count;
471 struct gendisk * bd_disk;
472 struct request_queue * bd_queue;
473 struct list_head bd_list;
475 * Private data. You must have bd_claim'ed the block_device
476 * to use this. NOTE: bd_claim allows an owner to claim
477 * the same device multiple times, the owner must take special
478 * care to not mess up bd_private for that case.
480 unsigned long bd_private;
482 /* The counter of freeze processes */
483 int bd_fsfreeze_count;
484 /* Mutex for freeze */
485 struct mutex bd_fsfreeze_mutex;
489 * Radix-tree tags, for tagging dirty and writeback pages within the pagecache
492 #define PAGECACHE_TAG_DIRTY 0
493 #define PAGECACHE_TAG_WRITEBACK 1
494 #define PAGECACHE_TAG_TOWRITE 2
496 int mapping_tagged(struct address_space *mapping, int tag);
498 static inline void i_mmap_lock_write(struct address_space *mapping)
500 down_write(&mapping->i_mmap_rwsem);
503 static inline void i_mmap_unlock_write(struct address_space *mapping)
505 up_write(&mapping->i_mmap_rwsem);
508 static inline void i_mmap_lock_read(struct address_space *mapping)
510 down_read(&mapping->i_mmap_rwsem);
513 static inline void i_mmap_unlock_read(struct address_space *mapping)
515 up_read(&mapping->i_mmap_rwsem);
519 * Might pages of this file be mapped into userspace?
521 static inline int mapping_mapped(struct address_space *mapping)
523 return !RB_EMPTY_ROOT(&mapping->i_mmap);
527 * Might pages of this file have been modified in userspace?
528 * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap_pgoff
529 * marks vma as VM_SHARED if it is shared, and the file was opened for
530 * writing i.e. vma may be mprotected writable even if now readonly.
532 * If i_mmap_writable is negative, no new writable mappings are allowed. You
533 * can only deny writable mappings, if none exists right now.
535 static inline int mapping_writably_mapped(struct address_space *mapping)
537 return atomic_read(&mapping->i_mmap_writable) > 0;
540 static inline int mapping_map_writable(struct address_space *mapping)
542 return atomic_inc_unless_negative(&mapping->i_mmap_writable) ?
546 static inline void mapping_unmap_writable(struct address_space *mapping)
548 atomic_dec(&mapping->i_mmap_writable);
551 static inline int mapping_deny_writable(struct address_space *mapping)
553 return atomic_dec_unless_positive(&mapping->i_mmap_writable) ?
557 static inline void mapping_allow_writable(struct address_space *mapping)
559 atomic_inc(&mapping->i_mmap_writable);
563 * Use sequence counter to get consistent i_size on 32-bit processors.
565 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
566 #include <linux/seqlock.h>
567 #define __NEED_I_SIZE_ORDERED
568 #define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
570 #define i_size_ordered_init(inode) do { } while (0)
574 #define ACL_NOT_CACHED ((void *)(-1))
576 #define IOP_FASTPERM 0x0001
577 #define IOP_LOOKUP 0x0002
578 #define IOP_NOFOLLOW 0x0004
581 * Keep mostly read-only and often accessed (especially for
582 * the RCU path lookup and 'stat' data) fields at the beginning
583 * of the 'struct inode'
587 unsigned short i_opflags;
590 unsigned int i_flags;
592 #ifdef CONFIG_FS_POSIX_ACL
593 struct posix_acl *i_acl;
594 struct posix_acl *i_default_acl;
597 const struct inode_operations *i_op;
598 struct super_block *i_sb;
599 struct address_space *i_mapping;
601 #ifdef CONFIG_SECURITY
605 /* Stat data, not accessed from path walking */
608 * Filesystems may only read i_nlink directly. They shall use the
609 * following functions for modification:
611 * (set|clear|inc|drop)_nlink
612 * inode_(inc|dec)_link_count
615 const unsigned int i_nlink;
616 unsigned int __i_nlink;
620 struct timespec i_atime;
621 struct timespec i_mtime;
622 struct timespec i_ctime;
623 spinlock_t i_lock; /* i_blocks, i_bytes, maybe i_size */
624 unsigned short i_bytes;
625 unsigned int i_blkbits;
628 #ifdef __NEED_I_SIZE_ORDERED
629 seqcount_t i_size_seqcount;
633 unsigned long i_state;
634 struct mutex i_mutex;
636 unsigned long dirtied_when; /* jiffies of first dirtying */
637 unsigned long dirtied_time_when;
639 struct hlist_node i_hash;
640 struct list_head i_wb_list; /* backing dev IO list */
641 #ifdef CONFIG_CGROUP_WRITEBACK
642 struct bdi_writeback *i_wb; /* the associated cgroup wb */
644 /* foreign inode detection, see wbc_detach_inode() */
646 u16 i_wb_frn_avg_time;
647 u16 i_wb_frn_history;
649 struct list_head i_lru; /* inode LRU list */
650 struct list_head i_sb_list;
652 struct hlist_head i_dentry;
653 struct rcu_head i_rcu;
657 atomic_t i_dio_count;
658 atomic_t i_writecount;
660 atomic_t i_readcount; /* struct files open RO */
662 const struct file_operations *i_fop; /* former ->i_op->default_file_ops */
663 struct file_lock_context *i_flctx;
664 struct address_space i_data;
665 struct list_head i_devices;
667 struct pipe_inode_info *i_pipe;
668 struct block_device *i_bdev;
675 #ifdef CONFIG_FSNOTIFY
676 __u32 i_fsnotify_mask; /* all events this inode cares about */
677 struct hlist_head i_fsnotify_marks;
680 void *i_private; /* fs or device private pointer */
683 static inline int inode_unhashed(struct inode *inode)
685 return hlist_unhashed(&inode->i_hash);
689 * inode->i_mutex nesting subclasses for the lock validator:
691 * 0: the object of the current VFS operation
695 * 4: second non-directory
696 * 5: second parent (when locking independent directories in rename)
698 * I_MUTEX_NONDIR2 is for certain operations (such as rename) which lock two
699 * non-directories at once.
701 * The locking order between these classes is
702 * parent[2] -> child -> grandchild -> normal -> xattr -> second non-directory
704 enum inode_i_mutex_lock_class
714 void lock_two_nondirectories(struct inode *, struct inode*);
715 void unlock_two_nondirectories(struct inode *, struct inode*);
718 * NOTE: in a 32bit arch with a preemptable kernel and
719 * an UP compile the i_size_read/write must be atomic
720 * with respect to the local cpu (unlike with preempt disabled),
721 * but they don't need to be atomic with respect to other cpus like in
722 * true SMP (so they need either to either locally disable irq around
723 * the read or for example on x86 they can be still implemented as a
724 * cmpxchg8b without the need of the lock prefix). For SMP compiles
725 * and 64bit archs it makes no difference if preempt is enabled or not.
727 static inline loff_t i_size_read(const struct inode *inode)
729 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
734 seq = read_seqcount_begin(&inode->i_size_seqcount);
735 i_size = inode->i_size;
736 } while (read_seqcount_retry(&inode->i_size_seqcount, seq));
738 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
742 i_size = inode->i_size;
746 return inode->i_size;
751 * NOTE: unlike i_size_read(), i_size_write() does need locking around it
752 * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount
753 * can be lost, resulting in subsequent i_size_read() calls spinning forever.
755 static inline void i_size_write(struct inode *inode, loff_t i_size)
757 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
759 write_seqcount_begin(&inode->i_size_seqcount);
760 inode->i_size = i_size;
761 write_seqcount_end(&inode->i_size_seqcount);
763 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
765 inode->i_size = i_size;
768 inode->i_size = i_size;
772 /* Helper functions so that in most cases filesystems will
773 * not need to deal directly with kuid_t and kgid_t and can
774 * instead deal with the raw numeric values that are stored
777 static inline uid_t i_uid_read(const struct inode *inode)
779 return from_kuid(&init_user_ns, inode->i_uid);
782 static inline gid_t i_gid_read(const struct inode *inode)
784 return from_kgid(&init_user_ns, inode->i_gid);
787 static inline void i_uid_write(struct inode *inode, uid_t uid)
789 inode->i_uid = make_kuid(&init_user_ns, uid);
792 static inline void i_gid_write(struct inode *inode, gid_t gid)
794 inode->i_gid = make_kgid(&init_user_ns, gid);
797 static inline unsigned iminor(const struct inode *inode)
799 return MINOR(inode->i_rdev);
802 static inline unsigned imajor(const struct inode *inode)
804 return MAJOR(inode->i_rdev);
807 extern struct block_device *I_BDEV(struct inode *inode);
810 rwlock_t lock; /* protects pid, uid, euid fields */
811 struct pid *pid; /* pid or -pgrp where SIGIO should be sent */
812 enum pid_type pid_type; /* Kind of process group SIGIO should be sent to */
813 kuid_t uid, euid; /* uid/euid of process setting the owner */
814 int signum; /* posix.1b rt signal to be delivered on IO */
818 * Track a single file's readahead state
820 struct file_ra_state {
821 pgoff_t start; /* where readahead started */
822 unsigned int size; /* # of readahead pages */
823 unsigned int async_size; /* do asynchronous readahead when
824 there are only # of pages ahead */
826 unsigned int ra_pages; /* Maximum readahead window */
827 unsigned int mmap_miss; /* Cache miss stat for mmap accesses */
828 loff_t prev_pos; /* Cache last read() position */
832 * Check if @index falls in the readahead windows.
834 static inline int ra_has_index(struct file_ra_state *ra, pgoff_t index)
836 return (index >= ra->start &&
837 index < ra->start + ra->size);
842 struct llist_node fu_llist;
843 struct rcu_head fu_rcuhead;
846 struct inode *f_inode; /* cached value */
847 const struct file_operations *f_op;
850 * Protects f_ep_links, f_flags.
851 * Must not be taken from IRQ context.
854 atomic_long_t f_count;
855 unsigned int f_flags;
857 struct mutex f_pos_lock;
859 struct fown_struct f_owner;
860 const struct cred *f_cred;
861 struct file_ra_state f_ra;
864 #ifdef CONFIG_SECURITY
867 /* needed for tty driver, and maybe others */
871 /* Used by fs/eventpoll.c to link all the hooks to this file */
872 struct list_head f_ep_links;
873 struct list_head f_tfile_llink;
874 #endif /* #ifdef CONFIG_EPOLL */
875 struct address_space *f_mapping;
876 } __attribute__((aligned(4))); /* lest something weird decides that 2 is OK */
881 /* file identifier */
882 unsigned char f_handle[0];
885 static inline struct file *get_file(struct file *f)
887 atomic_long_inc(&f->f_count);
890 #define get_file_rcu(x) atomic_long_inc_not_zero(&(x)->f_count)
891 #define fput_atomic(x) atomic_long_add_unless(&(x)->f_count, -1, 1)
892 #define file_count(x) atomic_long_read(&(x)->f_count)
894 #define MAX_NON_LFS ((1UL<<31) - 1)
896 /* Page cache limit. The filesystems should put that into their s_maxbytes
897 limits, otherwise bad things can happen in VM. */
898 #if BITS_PER_LONG==32
899 #define MAX_LFS_FILESIZE (((loff_t)PAGE_CACHE_SIZE << (BITS_PER_LONG-1))-1)
900 #elif BITS_PER_LONG==64
901 #define MAX_LFS_FILESIZE ((loff_t)0x7fffffffffffffffLL)
906 #define FL_DELEG 4 /* NFSv4 delegation */
907 #define FL_ACCESS 8 /* not trying to lock, just looking */
908 #define FL_EXISTS 16 /* when unlocking, test for existence */
909 #define FL_LEASE 32 /* lease held on this file */
910 #define FL_CLOSE 64 /* unlock on close */
911 #define FL_SLEEP 128 /* A blocking lock */
912 #define FL_DOWNGRADE_PENDING 256 /* Lease is being downgraded */
913 #define FL_UNLOCK_PENDING 512 /* Lease is being broken */
914 #define FL_OFDLCK 1024 /* lock is "owned" by struct file */
915 #define FL_LAYOUT 2048 /* outstanding pNFS layout */
918 * Special return value from posix_lock_file() and vfs_lock_file() for
919 * asynchronous locking.
921 #define FILE_LOCK_DEFERRED 1
923 /* legacy typedef, should eventually be removed */
924 typedef void *fl_owner_t;
928 struct file_lock_operations {
929 void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
930 void (*fl_release_private)(struct file_lock *);
933 struct lock_manager_operations {
934 int (*lm_compare_owner)(struct file_lock *, struct file_lock *);
935 unsigned long (*lm_owner_key)(struct file_lock *);
936 fl_owner_t (*lm_get_owner)(fl_owner_t);
937 void (*lm_put_owner)(fl_owner_t);
938 void (*lm_notify)(struct file_lock *); /* unblock callback */
939 int (*lm_grant)(struct file_lock *, int);
940 bool (*lm_break)(struct file_lock *);
941 int (*lm_change)(struct file_lock *, int, struct list_head *);
942 void (*lm_setup)(struct file_lock *, void **);
945 struct lock_manager {
946 struct list_head list;
950 void locks_start_grace(struct net *, struct lock_manager *);
951 void locks_end_grace(struct lock_manager *);
952 int locks_in_grace(struct net *);
954 /* that will die - we need it for nfs_lock_info */
955 #include <linux/nfs_fs_i.h>
958 * struct file_lock represents a generic "file lock". It's used to represent
959 * POSIX byte range locks, BSD (flock) locks, and leases. It's important to
960 * note that the same struct is used to represent both a request for a lock and
961 * the lock itself, but the same object is never used for both.
963 * FIXME: should we create a separate "struct lock_request" to help distinguish
966 * The varous i_flctx lists are ordered by:
969 * 2) lock range start
972 * Obviously, the last two criteria only matter for POSIX locks.
975 struct file_lock *fl_next; /* singly linked list for this inode */
976 struct list_head fl_list; /* link into file_lock_context */
977 struct hlist_node fl_link; /* node in global lists */
978 struct list_head fl_block; /* circular list of blocked processes */
980 unsigned int fl_flags;
981 unsigned char fl_type;
983 int fl_link_cpu; /* what cpu's list is this on? */
984 struct pid *fl_nspid;
985 wait_queue_head_t fl_wait;
986 struct file *fl_file;
990 struct fasync_struct * fl_fasync; /* for lease break notifications */
991 /* for lease breaks: */
992 unsigned long fl_break_time;
993 unsigned long fl_downgrade_time;
995 const struct file_lock_operations *fl_ops; /* Callbacks for filesystems */
996 const struct lock_manager_operations *fl_lmops; /* Callbacks for lockmanagers */
998 struct nfs_lock_info nfs_fl;
999 struct nfs4_lock_info nfs4_fl;
1001 struct list_head link; /* link in AFS vnode's pending_locks list */
1002 int state; /* state of grant or error if -ve */
1007 struct file_lock_context {
1008 spinlock_t flc_lock;
1009 struct list_head flc_flock;
1010 struct list_head flc_posix;
1011 struct list_head flc_lease;
1014 /* The following constant reflects the upper bound of the file/locking space */
1016 #define INT_LIMIT(x) (~((x)1 << (sizeof(x)*8 - 1)))
1017 #define OFFSET_MAX INT_LIMIT(loff_t)
1018 #define OFFT_OFFSET_MAX INT_LIMIT(off_t)
1021 #include <linux/fcntl.h>
1023 extern void send_sigio(struct fown_struct *fown, int fd, int band);
1025 #ifdef CONFIG_FILE_LOCKING
1026 extern int fcntl_getlk(struct file *, unsigned int, struct flock __user *);
1027 extern int fcntl_setlk(unsigned int, struct file *, unsigned int,
1028 struct flock __user *);
1030 #if BITS_PER_LONG == 32
1031 extern int fcntl_getlk64(struct file *, unsigned int, struct flock64 __user *);
1032 extern int fcntl_setlk64(unsigned int, struct file *, unsigned int,
1033 struct flock64 __user *);
1036 extern int fcntl_setlease(unsigned int fd, struct file *filp, long arg);
1037 extern int fcntl_getlease(struct file *filp);
1040 void locks_free_lock_context(struct file_lock_context *ctx);
1041 void locks_free_lock(struct file_lock *fl);
1042 extern void locks_init_lock(struct file_lock *);
1043 extern struct file_lock * locks_alloc_lock(void);
1044 extern void locks_copy_lock(struct file_lock *, struct file_lock *);
1045 extern void locks_copy_conflock(struct file_lock *, struct file_lock *);
1046 extern void locks_remove_posix(struct file *, fl_owner_t);
1047 extern void locks_remove_file(struct file *);
1048 extern void locks_release_private(struct file_lock *);
1049 extern void posix_test_lock(struct file *, struct file_lock *);
1050 extern int posix_lock_file(struct file *, struct file_lock *, struct file_lock *);
1051 extern int posix_lock_inode_wait(struct inode *, struct file_lock *);
1052 extern int posix_unblock_lock(struct file_lock *);
1053 extern int vfs_test_lock(struct file *, struct file_lock *);
1054 extern int vfs_lock_file(struct file *, unsigned int, struct file_lock *, struct file_lock *);
1055 extern int vfs_cancel_lock(struct file *filp, struct file_lock *fl);
1056 extern int flock_lock_inode_wait(struct inode *inode, struct file_lock *fl);
1057 extern int __break_lease(struct inode *inode, unsigned int flags, unsigned int type);
1058 extern void lease_get_mtime(struct inode *, struct timespec *time);
1059 extern int generic_setlease(struct file *, long, struct file_lock **, void **priv);
1060 extern int vfs_setlease(struct file *, long, struct file_lock **, void **);
1061 extern int lease_modify(struct file_lock *, int, struct list_head *);
1062 struct files_struct;
1063 extern void show_fd_locks(struct seq_file *f,
1064 struct file *filp, struct files_struct *files);
1065 #else /* !CONFIG_FILE_LOCKING */
1066 static inline int fcntl_getlk(struct file *file, unsigned int cmd,
1067 struct flock __user *user)
1072 static inline int fcntl_setlk(unsigned int fd, struct file *file,
1073 unsigned int cmd, struct flock __user *user)
1078 #if BITS_PER_LONG == 32
1079 static inline int fcntl_getlk64(struct file *file, unsigned int cmd,
1080 struct flock64 __user *user)
1085 static inline int fcntl_setlk64(unsigned int fd, struct file *file,
1086 unsigned int cmd, struct flock64 __user *user)
1091 static inline int fcntl_setlease(unsigned int fd, struct file *filp, long arg)
1096 static inline int fcntl_getlease(struct file *filp)
1102 locks_free_lock_context(struct file_lock_context *ctx)
1106 static inline void locks_init_lock(struct file_lock *fl)
1111 static inline void locks_copy_conflock(struct file_lock *new, struct file_lock *fl)
1116 static inline void locks_copy_lock(struct file_lock *new, struct file_lock *fl)
1121 static inline void locks_remove_posix(struct file *filp, fl_owner_t owner)
1126 static inline void locks_remove_file(struct file *filp)
1131 static inline void posix_test_lock(struct file *filp, struct file_lock *fl)
1136 static inline int posix_lock_file(struct file *filp, struct file_lock *fl,
1137 struct file_lock *conflock)
1142 static inline int posix_lock_inode_wait(struct inode *inode,
1143 struct file_lock *fl)
1148 static inline int posix_unblock_lock(struct file_lock *waiter)
1153 static inline int vfs_test_lock(struct file *filp, struct file_lock *fl)
1158 static inline int vfs_lock_file(struct file *filp, unsigned int cmd,
1159 struct file_lock *fl, struct file_lock *conf)
1164 static inline int vfs_cancel_lock(struct file *filp, struct file_lock *fl)
1169 static inline int flock_lock_inode_wait(struct inode *inode,
1170 struct file_lock *request)
1175 static inline int __break_lease(struct inode *inode, unsigned int mode, unsigned int type)
1180 static inline void lease_get_mtime(struct inode *inode, struct timespec *time)
1185 static inline int generic_setlease(struct file *filp, long arg,
1186 struct file_lock **flp, void **priv)
1191 static inline int vfs_setlease(struct file *filp, long arg,
1192 struct file_lock **lease, void **priv)
1197 static inline int lease_modify(struct file_lock *fl, int arg,
1198 struct list_head *dispose)
1203 struct files_struct;
1204 static inline void show_fd_locks(struct seq_file *f,
1205 struct file *filp, struct files_struct *files) {}
1206 #endif /* !CONFIG_FILE_LOCKING */
1208 static inline struct inode *file_inode(const struct file *f)
1213 static inline int posix_lock_file_wait(struct file *filp, struct file_lock *fl)
1215 return posix_lock_inode_wait(file_inode(filp), fl);
1218 static inline int flock_lock_file_wait(struct file *filp, struct file_lock *fl)
1220 return flock_lock_inode_wait(file_inode(filp), fl);
1223 struct fasync_struct {
1227 struct fasync_struct *fa_next; /* singly linked list */
1228 struct file *fa_file;
1229 struct rcu_head fa_rcu;
1232 #define FASYNC_MAGIC 0x4601
1234 /* SMP safe fasync helpers: */
1235 extern int fasync_helper(int, struct file *, int, struct fasync_struct **);
1236 extern struct fasync_struct *fasync_insert_entry(int, struct file *, struct fasync_struct **, struct fasync_struct *);
1237 extern int fasync_remove_entry(struct file *, struct fasync_struct **);
1238 extern struct fasync_struct *fasync_alloc(void);
1239 extern void fasync_free(struct fasync_struct *);
1241 /* can be called from interrupts */
1242 extern void kill_fasync(struct fasync_struct **, int, int);
1244 extern void __f_setown(struct file *filp, struct pid *, enum pid_type, int force);
1245 extern void f_setown(struct file *filp, unsigned long arg, int force);
1246 extern void f_delown(struct file *filp);
1247 extern pid_t f_getown(struct file *filp);
1248 extern int send_sigurg(struct fown_struct *fown);
1256 #define MNT_FORCE 0x00000001 /* Attempt to forcibily umount */
1257 #define MNT_DETACH 0x00000002 /* Just detach from the tree */
1258 #define MNT_EXPIRE 0x00000004 /* Mark for expiry */
1259 #define UMOUNT_NOFOLLOW 0x00000008 /* Don't follow symlink on umount */
1260 #define UMOUNT_UNUSED 0x80000000 /* Flag guaranteed to be unused */
1263 #define SB_I_CGROUPWB 0x00000001 /* cgroup-aware writeback enabled */
1265 /* Possible states of 'frozen' field */
1267 SB_UNFROZEN = 0, /* FS is unfrozen */
1268 SB_FREEZE_WRITE = 1, /* Writes, dir ops, ioctls frozen */
1269 SB_FREEZE_PAGEFAULT = 2, /* Page faults stopped as well */
1270 SB_FREEZE_FS = 3, /* For internal FS use (e.g. to stop
1271 * internal threads if needed) */
1272 SB_FREEZE_COMPLETE = 4, /* ->freeze_fs finished successfully */
1275 #define SB_FREEZE_LEVELS (SB_FREEZE_COMPLETE - 1)
1278 /* Counters for counting writers at each level */
1279 struct percpu_counter counter[SB_FREEZE_LEVELS];
1280 wait_queue_head_t wait; /* queue for waiting for
1281 writers / faults to finish */
1282 int frozen; /* Is sb frozen? */
1283 wait_queue_head_t wait_unfrozen; /* queue for waiting for
1285 #ifdef CONFIG_DEBUG_LOCK_ALLOC
1286 struct lockdep_map lock_map[SB_FREEZE_LEVELS];
1290 struct super_block {
1291 struct list_head s_list; /* Keep this first */
1292 dev_t s_dev; /* search index; _not_ kdev_t */
1293 unsigned char s_blocksize_bits;
1294 unsigned long s_blocksize;
1295 loff_t s_maxbytes; /* Max file size */
1296 struct file_system_type *s_type;
1297 const struct super_operations *s_op;
1298 const struct dquot_operations *dq_op;
1299 const struct quotactl_ops *s_qcop;
1300 const struct export_operations *s_export_op;
1301 unsigned long s_flags;
1302 unsigned long s_iflags; /* internal SB_I_* flags */
1303 unsigned long s_magic;
1304 struct dentry *s_root;
1305 struct rw_semaphore s_umount;
1308 #ifdef CONFIG_SECURITY
1311 const struct xattr_handler **s_xattr;
1313 struct list_head s_inodes; /* all inodes */
1314 struct hlist_bl_head s_anon; /* anonymous dentries for (nfs) exporting */
1315 struct list_head s_mounts; /* list of mounts; _not_ for fs use */
1316 struct block_device *s_bdev;
1317 struct backing_dev_info *s_bdi;
1318 struct mtd_info *s_mtd;
1319 struct hlist_node s_instances;
1320 unsigned int s_quota_types; /* Bitmask of supported quota types */
1321 struct quota_info s_dquot; /* Diskquota specific options */
1323 struct sb_writers s_writers;
1325 char s_id[32]; /* Informational name */
1326 u8 s_uuid[16]; /* UUID */
1328 void *s_fs_info; /* Filesystem private info */
1329 unsigned int s_max_links;
1332 /* Granularity of c/m/atime in ns.
1333 Cannot be worse than a second */
1337 * The next field is for VFS *only*. No filesystems have any business
1338 * even looking at it. You had been warned.
1340 struct mutex s_vfs_rename_mutex; /* Kludge */
1343 * Filesystem subtype. If non-empty the filesystem type field
1344 * in /proc/mounts will be "type.subtype"
1349 * Saved mount options for lazy filesystems using
1350 * generic_show_options()
1352 char __rcu *s_options;
1353 const struct dentry_operations *s_d_op; /* default d_op for dentries */
1356 * Saved pool identifier for cleancache (-1 means none)
1358 int cleancache_poolid;
1360 struct shrinker s_shrink; /* per-sb shrinker handle */
1362 /* Number of inodes with nlink == 0 but still referenced */
1363 atomic_long_t s_remove_count;
1365 /* Being remounted read-only */
1366 int s_readonly_remount;
1368 /* AIO completions deferred from interrupt context */
1369 struct workqueue_struct *s_dio_done_wq;
1370 struct hlist_head s_pins;
1373 * Keep the lru lists last in the structure so they always sit on their
1374 * own individual cachelines.
1376 struct list_lru s_dentry_lru ____cacheline_aligned_in_smp;
1377 struct list_lru s_inode_lru ____cacheline_aligned_in_smp;
1378 struct rcu_head rcu;
1379 struct work_struct destroy_work;
1381 * Indicates how deep in a filesystem stack this SB is
1386 extern struct timespec current_fs_time(struct super_block *sb);
1389 * Snapshotting support.
1392 void __sb_end_write(struct super_block *sb, int level);
1393 int __sb_start_write(struct super_block *sb, int level, bool wait);
1395 #define __sb_writers_acquired(sb, lev) \
1396 rwsem_acquire_read(&(sb)->s_writers.lock_map[(lev)-1], 0, 1, _THIS_IP_)
1397 #define __sb_writers_release(sb, lev) \
1398 rwsem_release(&(sb)->s_writers.lock_map[(lev)-1], 1, _THIS_IP_)
1401 * sb_end_write - drop write access to a superblock
1402 * @sb: the super we wrote to
1404 * Decrement number of writers to the filesystem. Wake up possible waiters
1405 * wanting to freeze the filesystem.
1407 static inline void sb_end_write(struct super_block *sb)
1409 __sb_end_write(sb, SB_FREEZE_WRITE);
1413 * sb_end_pagefault - drop write access to a superblock from a page fault
1414 * @sb: the super we wrote to
1416 * Decrement number of processes handling write page fault to the filesystem.
1417 * Wake up possible waiters wanting to freeze the filesystem.
1419 static inline void sb_end_pagefault(struct super_block *sb)
1421 __sb_end_write(sb, SB_FREEZE_PAGEFAULT);
1425 * sb_end_intwrite - drop write access to a superblock for internal fs purposes
1426 * @sb: the super we wrote to
1428 * Decrement fs-internal number of writers to the filesystem. Wake up possible
1429 * waiters wanting to freeze the filesystem.
1431 static inline void sb_end_intwrite(struct super_block *sb)
1433 __sb_end_write(sb, SB_FREEZE_FS);
1437 * sb_start_write - get write access to a superblock
1438 * @sb: the super we write to
1440 * When a process wants to write data or metadata to a file system (i.e. dirty
1441 * a page or an inode), it should embed the operation in a sb_start_write() -
1442 * sb_end_write() pair to get exclusion against file system freezing. This
1443 * function increments number of writers preventing freezing. If the file
1444 * system is already frozen, the function waits until the file system is
1447 * Since freeze protection behaves as a lock, users have to preserve
1448 * ordering of freeze protection and other filesystem locks. Generally,
1449 * freeze protection should be the outermost lock. In particular, we have:
1452 * -> i_mutex (write path, truncate, directory ops, ...)
1453 * -> s_umount (freeze_super, thaw_super)
1455 static inline void sb_start_write(struct super_block *sb)
1457 __sb_start_write(sb, SB_FREEZE_WRITE, true);
1460 static inline int sb_start_write_trylock(struct super_block *sb)
1462 return __sb_start_write(sb, SB_FREEZE_WRITE, false);
1466 * sb_start_pagefault - get write access to a superblock from a page fault
1467 * @sb: the super we write to
1469 * When a process starts handling write page fault, it should embed the
1470 * operation into sb_start_pagefault() - sb_end_pagefault() pair to get
1471 * exclusion against file system freezing. This is needed since the page fault
1472 * is going to dirty a page. This function increments number of running page
1473 * faults preventing freezing. If the file system is already frozen, the
1474 * function waits until the file system is thawed.
1476 * Since page fault freeze protection behaves as a lock, users have to preserve
1477 * ordering of freeze protection and other filesystem locks. It is advised to
1478 * put sb_start_pagefault() close to mmap_sem in lock ordering. Page fault
1479 * handling code implies lock dependency:
1482 * -> sb_start_pagefault
1484 static inline void sb_start_pagefault(struct super_block *sb)
1486 __sb_start_write(sb, SB_FREEZE_PAGEFAULT, true);
1490 * sb_start_intwrite - get write access to a superblock for internal fs purposes
1491 * @sb: the super we write to
1493 * This is the third level of protection against filesystem freezing. It is
1494 * free for use by a filesystem. The only requirement is that it must rank
1495 * below sb_start_pagefault.
1497 * For example filesystem can call sb_start_intwrite() when starting a
1498 * transaction which somewhat eases handling of freezing for internal sources
1499 * of filesystem changes (internal fs threads, discarding preallocation on file
1502 static inline void sb_start_intwrite(struct super_block *sb)
1504 __sb_start_write(sb, SB_FREEZE_FS, true);
1508 extern bool inode_owner_or_capable(const struct inode *inode);
1511 * VFS helper functions..
1513 extern int vfs_create(struct inode *, struct dentry *, umode_t, bool);
1514 extern int vfs_mkdir(struct inode *, struct dentry *, umode_t);
1515 extern int vfs_mknod(struct inode *, struct dentry *, umode_t, dev_t);
1516 extern int vfs_symlink(struct inode *, struct dentry *, const char *);
1517 extern int vfs_link(struct dentry *, struct inode *, struct dentry *, struct inode **);
1518 extern int vfs_rmdir(struct inode *, struct dentry *);
1519 extern int vfs_unlink(struct inode *, struct dentry *, struct inode **);
1520 extern int vfs_rename(struct inode *, struct dentry *, struct inode *, struct dentry *, struct inode **, unsigned int);
1521 extern int vfs_whiteout(struct inode *, struct dentry *);
1524 * VFS dentry helper functions.
1526 extern void dentry_unhash(struct dentry *dentry);
1529 * VFS file helper functions.
1531 extern void inode_init_owner(struct inode *inode, const struct inode *dir,
1534 * VFS FS_IOC_FIEMAP helper definitions.
1536 struct fiemap_extent_info {
1537 unsigned int fi_flags; /* Flags as passed from user */
1538 unsigned int fi_extents_mapped; /* Number of mapped extents */
1539 unsigned int fi_extents_max; /* Size of fiemap_extent array */
1540 struct fiemap_extent __user *fi_extents_start; /* Start of
1541 fiemap_extent array */
1543 int fiemap_fill_next_extent(struct fiemap_extent_info *info, u64 logical,
1544 u64 phys, u64 len, u32 flags);
1545 int fiemap_check_flags(struct fiemap_extent_info *fieinfo, u32 fs_flags);
1550 * NOTE! These match bits 12..15 of stat.st_mode
1551 * (ie "(i_mode >> 12) & 15").
1553 #define DT_UNKNOWN 0
1564 * This is the "filldir" function type, used by readdir() to let
1565 * the kernel specify what kind of dirent layout it wants to have.
1566 * This allows the kernel to read directories into kernel space or
1567 * to have different dirent layouts depending on the binary type.
1570 typedef int (*filldir_t)(struct dir_context *, const char *, int, loff_t, u64,
1573 struct dir_context {
1574 const filldir_t actor;
1578 struct block_device_operations;
1580 /* These macros are for out of kernel modules to test that
1581 * the kernel supports the unlocked_ioctl and compat_ioctl
1582 * fields in struct file_operations. */
1583 #define HAVE_COMPAT_IOCTL 1
1584 #define HAVE_UNLOCKED_IOCTL 1
1587 * These flags let !MMU mmap() govern direct device mapping vs immediate
1588 * copying more easily for MAP_PRIVATE, especially for ROM filesystems.
1590 * NOMMU_MAP_COPY: Copy can be mapped (MAP_PRIVATE)
1591 * NOMMU_MAP_DIRECT: Can be mapped directly (MAP_SHARED)
1592 * NOMMU_MAP_READ: Can be mapped for reading
1593 * NOMMU_MAP_WRITE: Can be mapped for writing
1594 * NOMMU_MAP_EXEC: Can be mapped for execution
1596 #define NOMMU_MAP_COPY 0x00000001
1597 #define NOMMU_MAP_DIRECT 0x00000008
1598 #define NOMMU_MAP_READ VM_MAYREAD
1599 #define NOMMU_MAP_WRITE VM_MAYWRITE
1600 #define NOMMU_MAP_EXEC VM_MAYEXEC
1602 #define NOMMU_VMFLAGS \
1603 (NOMMU_MAP_READ | NOMMU_MAP_WRITE | NOMMU_MAP_EXEC)
1608 struct file_operations {
1609 struct module *owner;
1610 loff_t (*llseek) (struct file *, loff_t, int);
1611 ssize_t (*read) (struct file *, char __user *, size_t, loff_t *);
1612 ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *);
1613 ssize_t (*read_iter) (struct kiocb *, struct iov_iter *);
1614 ssize_t (*write_iter) (struct kiocb *, struct iov_iter *);
1615 int (*iterate) (struct file *, struct dir_context *);
1616 unsigned int (*poll) (struct file *, struct poll_table_struct *);
1617 long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long);
1618 long (*compat_ioctl) (struct file *, unsigned int, unsigned long);
1619 int (*mmap) (struct file *, struct vm_area_struct *);
1620 int (*mremap)(struct file *, struct vm_area_struct *);
1621 int (*open) (struct inode *, struct file *);
1622 int (*flush) (struct file *, fl_owner_t id);
1623 int (*release) (struct inode *, struct file *);
1624 int (*fsync) (struct file *, loff_t, loff_t, int datasync);
1625 int (*aio_fsync) (struct kiocb *, int datasync);
1626 int (*fasync) (int, struct file *, int);
1627 int (*lock) (struct file *, int, struct file_lock *);
1628 ssize_t (*sendpage) (struct file *, struct page *, int, size_t, loff_t *, int);
1629 unsigned long (*get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
1630 int (*check_flags)(int);
1631 int (*flock) (struct file *, int, struct file_lock *);
1632 ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, loff_t *, size_t, unsigned int);
1633 ssize_t (*splice_read)(struct file *, loff_t *, struct pipe_inode_info *, size_t, unsigned int);
1634 int (*setlease)(struct file *, long, struct file_lock **, void **);
1635 long (*fallocate)(struct file *file, int mode, loff_t offset,
1637 void (*show_fdinfo)(struct seq_file *m, struct file *f);
1639 unsigned (*mmap_capabilities)(struct file *);
1643 struct inode_operations {
1644 struct dentry * (*lookup) (struct inode *,struct dentry *, unsigned int);
1645 const char * (*follow_link) (struct dentry *, void **);
1646 int (*permission) (struct inode *, int);
1647 struct posix_acl * (*get_acl)(struct inode *, int);
1649 int (*readlink) (struct dentry *, char __user *,int);
1650 void (*put_link) (struct inode *, void *);
1652 int (*create) (struct inode *,struct dentry *, umode_t, bool);
1653 int (*link) (struct dentry *,struct inode *,struct dentry *);
1654 int (*unlink) (struct inode *,struct dentry *);
1655 int (*symlink) (struct inode *,struct dentry *,const char *);
1656 int (*mkdir) (struct inode *,struct dentry *,umode_t);
1657 int (*rmdir) (struct inode *,struct dentry *);
1658 int (*mknod) (struct inode *,struct dentry *,umode_t,dev_t);
1659 int (*rename) (struct inode *, struct dentry *,
1660 struct inode *, struct dentry *);
1661 int (*rename2) (struct inode *, struct dentry *,
1662 struct inode *, struct dentry *, unsigned int);
1663 int (*setattr) (struct dentry *, struct iattr *);
1664 int (*getattr) (struct vfsmount *mnt, struct dentry *, struct kstat *);
1665 int (*setxattr) (struct dentry *, const char *,const void *,size_t,int);
1666 ssize_t (*getxattr) (struct dentry *, const char *, void *, size_t);
1667 ssize_t (*listxattr) (struct dentry *, char *, size_t);
1668 int (*removexattr) (struct dentry *, const char *);
1669 int (*fiemap)(struct inode *, struct fiemap_extent_info *, u64 start,
1671 int (*update_time)(struct inode *, struct timespec *, int);
1672 int (*atomic_open)(struct inode *, struct dentry *,
1673 struct file *, unsigned open_flag,
1674 umode_t create_mode, int *opened);
1675 int (*tmpfile) (struct inode *, struct dentry *, umode_t);
1676 int (*set_acl)(struct inode *, struct posix_acl *, int);
1678 /* WARNING: probably going away soon, do not use! */
1679 } ____cacheline_aligned;
1681 ssize_t rw_copy_check_uvector(int type, const struct iovec __user * uvector,
1682 unsigned long nr_segs, unsigned long fast_segs,
1683 struct iovec *fast_pointer,
1684 struct iovec **ret_pointer);
1686 extern ssize_t __vfs_read(struct file *, char __user *, size_t, loff_t *);
1687 extern ssize_t __vfs_write(struct file *, const char __user *, size_t, loff_t *);
1688 extern ssize_t vfs_read(struct file *, char __user *, size_t, loff_t *);
1689 extern ssize_t vfs_write(struct file *, const char __user *, size_t, loff_t *);
1690 extern ssize_t vfs_readv(struct file *, const struct iovec __user *,
1691 unsigned long, loff_t *);
1692 extern ssize_t vfs_writev(struct file *, const struct iovec __user *,
1693 unsigned long, loff_t *);
1695 struct super_operations {
1696 struct inode *(*alloc_inode)(struct super_block *sb);
1697 void (*destroy_inode)(struct inode *);
1699 void (*dirty_inode) (struct inode *, int flags);
1700 int (*write_inode) (struct inode *, struct writeback_control *wbc);
1701 int (*drop_inode) (struct inode *);
1702 void (*evict_inode) (struct inode *);
1703 void (*put_super) (struct super_block *);
1704 int (*sync_fs)(struct super_block *sb, int wait);
1705 int (*freeze_super) (struct super_block *);
1706 int (*freeze_fs) (struct super_block *);
1707 int (*thaw_super) (struct super_block *);
1708 int (*unfreeze_fs) (struct super_block *);
1709 int (*statfs) (struct dentry *, struct kstatfs *);
1710 int (*remount_fs) (struct super_block *, int *, char *);
1711 void (*umount_begin) (struct super_block *);
1713 int (*show_options)(struct seq_file *, struct dentry *);
1714 int (*show_devname)(struct seq_file *, struct dentry *);
1715 int (*show_path)(struct seq_file *, struct dentry *);
1716 int (*show_stats)(struct seq_file *, struct dentry *);
1718 ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t);
1719 ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t);
1720 struct dquot **(*get_dquots)(struct inode *);
1722 int (*bdev_try_to_free_page)(struct super_block*, struct page*, gfp_t);
1723 long (*nr_cached_objects)(struct super_block *,
1724 struct shrink_control *);
1725 long (*free_cached_objects)(struct super_block *,
1726 struct shrink_control *);
1730 * Inode flags - they have no relation to superblock flags now
1732 #define S_SYNC 1 /* Writes are synced at once */
1733 #define S_NOATIME 2 /* Do not update access times */
1734 #define S_APPEND 4 /* Append-only file */
1735 #define S_IMMUTABLE 8 /* Immutable file */
1736 #define S_DEAD 16 /* removed, but still open directory */
1737 #define S_NOQUOTA 32 /* Inode is not counted to quota */
1738 #define S_DIRSYNC 64 /* Directory modifications are synchronous */
1739 #define S_NOCMTIME 128 /* Do not update file c/mtime */
1740 #define S_SWAPFILE 256 /* Do not truncate: swapon got its bmaps */
1741 #define S_PRIVATE 512 /* Inode is fs-internal */
1742 #define S_IMA 1024 /* Inode has an associated IMA struct */
1743 #define S_AUTOMOUNT 2048 /* Automount/referral quasi-directory */
1744 #define S_NOSEC 4096 /* no suid or xattr security attributes */
1745 #ifdef CONFIG_FS_DAX
1746 #define S_DAX 8192 /* Direct Access, avoiding the page cache */
1748 #define S_DAX 0 /* Make all the DAX code disappear */
1752 * Note that nosuid etc flags are inode-specific: setting some file-system
1753 * flags just means all the inodes inherit those flags by default. It might be
1754 * possible to override it selectively if you really wanted to with some
1755 * ioctl() that is not currently implemented.
1757 * Exception: MS_RDONLY is always applied to the entire file system.
1759 * Unfortunately, it is possible to change a filesystems flags with it mounted
1760 * with files in use. This means that all of the inodes will not have their
1761 * i_flags updated. Hence, i_flags no longer inherit the superblock mount
1762 * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org
1764 #define __IS_FLG(inode, flg) ((inode)->i_sb->s_flags & (flg))
1766 #define IS_RDONLY(inode) ((inode)->i_sb->s_flags & MS_RDONLY)
1767 #define IS_SYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS) || \
1768 ((inode)->i_flags & S_SYNC))
1769 #define IS_DIRSYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS|MS_DIRSYNC) || \
1770 ((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
1771 #define IS_MANDLOCK(inode) __IS_FLG(inode, MS_MANDLOCK)
1772 #define IS_NOATIME(inode) __IS_FLG(inode, MS_RDONLY|MS_NOATIME)
1773 #define IS_I_VERSION(inode) __IS_FLG(inode, MS_I_VERSION)
1775 #define IS_NOQUOTA(inode) ((inode)->i_flags & S_NOQUOTA)
1776 #define IS_APPEND(inode) ((inode)->i_flags & S_APPEND)
1777 #define IS_IMMUTABLE(inode) ((inode)->i_flags & S_IMMUTABLE)
1778 #define IS_POSIXACL(inode) __IS_FLG(inode, MS_POSIXACL)
1780 #define IS_DEADDIR(inode) ((inode)->i_flags & S_DEAD)
1781 #define IS_NOCMTIME(inode) ((inode)->i_flags & S_NOCMTIME)
1782 #define IS_SWAPFILE(inode) ((inode)->i_flags & S_SWAPFILE)
1783 #define IS_PRIVATE(inode) ((inode)->i_flags & S_PRIVATE)
1784 #define IS_IMA(inode) ((inode)->i_flags & S_IMA)
1785 #define IS_AUTOMOUNT(inode) ((inode)->i_flags & S_AUTOMOUNT)
1786 #define IS_NOSEC(inode) ((inode)->i_flags & S_NOSEC)
1787 #define IS_DAX(inode) ((inode)->i_flags & S_DAX)
1789 #define IS_WHITEOUT(inode) (S_ISCHR(inode->i_mode) && \
1790 (inode)->i_rdev == WHITEOUT_DEV)
1793 * Inode state bits. Protected by inode->i_lock
1795 * Three bits determine the dirty state of the inode, I_DIRTY_SYNC,
1796 * I_DIRTY_DATASYNC and I_DIRTY_PAGES.
1798 * Four bits define the lifetime of an inode. Initially, inodes are I_NEW,
1799 * until that flag is cleared. I_WILL_FREE, I_FREEING and I_CLEAR are set at
1800 * various stages of removing an inode.
1802 * Two bits are used for locking and completion notification, I_NEW and I_SYNC.
1804 * I_DIRTY_SYNC Inode is dirty, but doesn't have to be written on
1805 * fdatasync(). i_atime is the usual cause.
1806 * I_DIRTY_DATASYNC Data-related inode changes pending. We keep track of
1807 * these changes separately from I_DIRTY_SYNC so that we
1808 * don't have to write inode on fdatasync() when only
1809 * mtime has changed in it.
1810 * I_DIRTY_PAGES Inode has dirty pages. Inode itself may be clean.
1811 * I_NEW Serves as both a mutex and completion notification.
1812 * New inodes set I_NEW. If two processes both create
1813 * the same inode, one of them will release its inode and
1814 * wait for I_NEW to be released before returning.
1815 * Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can
1816 * also cause waiting on I_NEW, without I_NEW actually
1817 * being set. find_inode() uses this to prevent returning
1818 * nearly-dead inodes.
1819 * I_WILL_FREE Must be set when calling write_inode_now() if i_count
1820 * is zero. I_FREEING must be set when I_WILL_FREE is
1822 * I_FREEING Set when inode is about to be freed but still has dirty
1823 * pages or buffers attached or the inode itself is still
1825 * I_CLEAR Added by clear_inode(). In this state the inode is
1826 * clean and can be destroyed. Inode keeps I_FREEING.
1828 * Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are
1829 * prohibited for many purposes. iget() must wait for
1830 * the inode to be completely released, then create it
1831 * anew. Other functions will just ignore such inodes,
1832 * if appropriate. I_NEW is used for waiting.
1834 * I_SYNC Writeback of inode is running. The bit is set during
1835 * data writeback, and cleared with a wakeup on the bit
1836 * address once it is done. The bit is also used to pin
1837 * the inode in memory for flusher thread.
1839 * I_REFERENCED Marks the inode as recently references on the LRU list.
1841 * I_DIO_WAKEUP Never set. Only used as a key for wait_on_bit().
1843 * I_WB_SWITCH Cgroup bdi_writeback switching in progress. Used to
1844 * synchronize competing switching instances and to tell
1845 * wb stat updates to grab mapping->tree_lock. See
1846 * inode_switch_wb_work_fn() for details.
1848 * Q: What is the difference between I_WILL_FREE and I_FREEING?
1850 #define I_DIRTY_SYNC (1 << 0)
1851 #define I_DIRTY_DATASYNC (1 << 1)
1852 #define I_DIRTY_PAGES (1 << 2)
1854 #define I_NEW (1 << __I_NEW)
1855 #define I_WILL_FREE (1 << 4)
1856 #define I_FREEING (1 << 5)
1857 #define I_CLEAR (1 << 6)
1859 #define I_SYNC (1 << __I_SYNC)
1860 #define I_REFERENCED (1 << 8)
1861 #define __I_DIO_WAKEUP 9
1862 #define I_DIO_WAKEUP (1 << __I_DIO_WAKEUP)
1863 #define I_LINKABLE (1 << 10)
1864 #define I_DIRTY_TIME (1 << 11)
1865 #define __I_DIRTY_TIME_EXPIRED 12
1866 #define I_DIRTY_TIME_EXPIRED (1 << __I_DIRTY_TIME_EXPIRED)
1867 #define I_WB_SWITCH (1 << 13)
1869 #define I_DIRTY (I_DIRTY_SYNC | I_DIRTY_DATASYNC | I_DIRTY_PAGES)
1870 #define I_DIRTY_ALL (I_DIRTY | I_DIRTY_TIME)
1872 extern void __mark_inode_dirty(struct inode *, int);
1873 static inline void mark_inode_dirty(struct inode *inode)
1875 __mark_inode_dirty(inode, I_DIRTY);
1878 static inline void mark_inode_dirty_sync(struct inode *inode)
1880 __mark_inode_dirty(inode, I_DIRTY_SYNC);
1883 extern void inc_nlink(struct inode *inode);
1884 extern void drop_nlink(struct inode *inode);
1885 extern void clear_nlink(struct inode *inode);
1886 extern void set_nlink(struct inode *inode, unsigned int nlink);
1888 static inline void inode_inc_link_count(struct inode *inode)
1891 mark_inode_dirty(inode);
1894 static inline void inode_dec_link_count(struct inode *inode)
1897 mark_inode_dirty(inode);
1901 * inode_inc_iversion - increments i_version
1902 * @inode: inode that need to be updated
1904 * Every time the inode is modified, the i_version field will be incremented.
1905 * The filesystem has to be mounted with i_version flag
1908 static inline void inode_inc_iversion(struct inode *inode)
1910 spin_lock(&inode->i_lock);
1912 spin_unlock(&inode->i_lock);
1915 enum file_time_flags {
1922 extern bool atime_needs_update(const struct path *, struct inode *);
1923 extern void touch_atime(const struct path *);
1924 static inline void file_accessed(struct file *file)
1926 if (!(file->f_flags & O_NOATIME))
1927 touch_atime(&file->f_path);
1930 int sync_inode(struct inode *inode, struct writeback_control *wbc);
1931 int sync_inode_metadata(struct inode *inode, int wait);
1933 struct file_system_type {
1936 #define FS_REQUIRES_DEV 1
1937 #define FS_BINARY_MOUNTDATA 2
1938 #define FS_HAS_SUBTYPE 4
1939 #define FS_USERNS_MOUNT 8 /* Can be mounted by userns root */
1940 #define FS_USERNS_DEV_MOUNT 16 /* A userns mount does not imply MNT_NODEV */
1941 #define FS_USERNS_VISIBLE 32 /* FS must already be visible */
1942 #define FS_RENAME_DOES_D_MOVE 32768 /* FS will handle d_move() during rename() internally. */
1943 struct dentry *(*mount) (struct file_system_type *, int,
1944 const char *, void *);
1945 void (*kill_sb) (struct super_block *);
1946 struct module *owner;
1947 struct file_system_type * next;
1948 struct hlist_head fs_supers;
1950 struct lock_class_key s_lock_key;
1951 struct lock_class_key s_umount_key;
1952 struct lock_class_key s_vfs_rename_key;
1953 struct lock_class_key s_writers_key[SB_FREEZE_LEVELS];
1955 struct lock_class_key i_lock_key;
1956 struct lock_class_key i_mutex_key;
1957 struct lock_class_key i_mutex_dir_key;
1960 #define MODULE_ALIAS_FS(NAME) MODULE_ALIAS("fs-" NAME)
1962 extern struct dentry *mount_ns(struct file_system_type *fs_type, int flags,
1963 void *data, int (*fill_super)(struct super_block *, void *, int));
1964 extern struct dentry *mount_bdev(struct file_system_type *fs_type,
1965 int flags, const char *dev_name, void *data,
1966 int (*fill_super)(struct super_block *, void *, int));
1967 extern struct dentry *mount_single(struct file_system_type *fs_type,
1968 int flags, void *data,
1969 int (*fill_super)(struct super_block *, void *, int));
1970 extern struct dentry *mount_nodev(struct file_system_type *fs_type,
1971 int flags, void *data,
1972 int (*fill_super)(struct super_block *, void *, int));
1973 extern struct dentry *mount_subtree(struct vfsmount *mnt, const char *path);
1974 void generic_shutdown_super(struct super_block *sb);
1975 void kill_block_super(struct super_block *sb);
1976 void kill_anon_super(struct super_block *sb);
1977 void kill_litter_super(struct super_block *sb);
1978 void deactivate_super(struct super_block *sb);
1979 void deactivate_locked_super(struct super_block *sb);
1980 int set_anon_super(struct super_block *s, void *data);
1981 int get_anon_bdev(dev_t *);
1982 void free_anon_bdev(dev_t);
1983 struct super_block *sget(struct file_system_type *type,
1984 int (*test)(struct super_block *,void *),
1985 int (*set)(struct super_block *,void *),
1986 int flags, void *data);
1987 extern struct dentry *mount_pseudo(struct file_system_type *, char *,
1988 const struct super_operations *ops,
1989 const struct dentry_operations *dops,
1992 /* Alas, no aliases. Too much hassle with bringing module.h everywhere */
1993 #define fops_get(fops) \
1994 (((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
1995 #define fops_put(fops) \
1996 do { if (fops) module_put((fops)->owner); } while(0)
1998 * This one is to be used *ONLY* from ->open() instances.
1999 * fops must be non-NULL, pinned down *and* module dependencies
2000 * should be sufficient to pin the caller down as well.
2002 #define replace_fops(f, fops) \
2004 struct file *__file = (f); \
2005 fops_put(__file->f_op); \
2006 BUG_ON(!(__file->f_op = (fops))); \
2009 extern int register_filesystem(struct file_system_type *);
2010 extern int unregister_filesystem(struct file_system_type *);
2011 extern struct vfsmount *kern_mount_data(struct file_system_type *, void *data);
2012 #define kern_mount(type) kern_mount_data(type, NULL)
2013 extern void kern_unmount(struct vfsmount *mnt);
2014 extern int may_umount_tree(struct vfsmount *);
2015 extern int may_umount(struct vfsmount *);
2016 extern long do_mount(const char *, const char __user *,
2017 const char *, unsigned long, void *);
2018 extern struct vfsmount *collect_mounts(struct path *);
2019 extern void drop_collected_mounts(struct vfsmount *);
2020 extern int iterate_mounts(int (*)(struct vfsmount *, void *), void *,
2022 extern int vfs_statfs(struct path *, struct kstatfs *);
2023 extern int user_statfs(const char __user *, struct kstatfs *);
2024 extern int fd_statfs(int, struct kstatfs *);
2025 extern int vfs_ustat(dev_t, struct kstatfs *);
2026 extern int freeze_super(struct super_block *super);
2027 extern int thaw_super(struct super_block *super);
2028 extern bool our_mnt(struct vfsmount *mnt);
2030 extern int current_umask(void);
2032 extern void ihold(struct inode * inode);
2033 extern void iput(struct inode *);
2034 extern int generic_update_time(struct inode *, struct timespec *, int);
2037 extern struct kobject *fs_kobj;
2039 #define MAX_RW_COUNT (INT_MAX & PAGE_CACHE_MASK)
2041 #define FLOCK_VERIFY_READ 1
2042 #define FLOCK_VERIFY_WRITE 2
2044 #ifdef CONFIG_FILE_LOCKING
2045 extern int locks_mandatory_locked(struct file *);
2046 extern int locks_mandatory_area(int, struct inode *, struct file *, loff_t, size_t);
2049 * Candidates for mandatory locking have the setgid bit set
2050 * but no group execute bit - an otherwise meaningless combination.
2053 static inline int __mandatory_lock(struct inode *ino)
2055 return (ino->i_mode & (S_ISGID | S_IXGRP)) == S_ISGID;
2059 * ... and these candidates should be on MS_MANDLOCK mounted fs,
2060 * otherwise these will be advisory locks
2063 static inline int mandatory_lock(struct inode *ino)
2065 return IS_MANDLOCK(ino) && __mandatory_lock(ino);
2068 static inline int locks_verify_locked(struct file *file)
2070 if (mandatory_lock(file_inode(file)))
2071 return locks_mandatory_locked(file);
2075 static inline int locks_verify_truncate(struct inode *inode,
2079 if (inode->i_flctx && mandatory_lock(inode))
2080 return locks_mandatory_area(
2081 FLOCK_VERIFY_WRITE, inode, filp,
2082 size < inode->i_size ? size : inode->i_size,
2083 (size < inode->i_size ? inode->i_size - size
2084 : size - inode->i_size)
2089 static inline int break_lease(struct inode *inode, unsigned int mode)
2092 * Since this check is lockless, we must ensure that any refcounts
2093 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2094 * could end up racing with tasks trying to set a new lease on this
2098 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2099 return __break_lease(inode, mode, FL_LEASE);
2103 static inline int break_deleg(struct inode *inode, unsigned int mode)
2106 * Since this check is lockless, we must ensure that any refcounts
2107 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2108 * could end up racing with tasks trying to set a new lease on this
2112 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2113 return __break_lease(inode, mode, FL_DELEG);
2117 static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode)
2121 ret = break_deleg(inode, O_WRONLY|O_NONBLOCK);
2122 if (ret == -EWOULDBLOCK && delegated_inode) {
2123 *delegated_inode = inode;
2129 static inline int break_deleg_wait(struct inode **delegated_inode)
2133 ret = break_deleg(*delegated_inode, O_WRONLY);
2134 iput(*delegated_inode);
2135 *delegated_inode = NULL;
2139 static inline int break_layout(struct inode *inode, bool wait)
2142 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2143 return __break_lease(inode,
2144 wait ? O_WRONLY : O_WRONLY | O_NONBLOCK,
2149 #else /* !CONFIG_FILE_LOCKING */
2150 static inline int locks_mandatory_locked(struct file *file)
2155 static inline int locks_mandatory_area(int rw, struct inode *inode,
2156 struct file *filp, loff_t offset,
2162 static inline int __mandatory_lock(struct inode *inode)
2167 static inline int mandatory_lock(struct inode *inode)
2172 static inline int locks_verify_locked(struct file *file)
2177 static inline int locks_verify_truncate(struct inode *inode, struct file *filp,
2183 static inline int break_lease(struct inode *inode, unsigned int mode)
2188 static inline int break_deleg(struct inode *inode, unsigned int mode)
2193 static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode)
2198 static inline int break_deleg_wait(struct inode **delegated_inode)
2204 static inline int break_layout(struct inode *inode, bool wait)
2209 #endif /* CONFIG_FILE_LOCKING */
2214 const char *name; /* pointer to actual string */
2215 const __user char *uptr; /* original userland pointer */
2216 struct audit_names *aname;
2221 extern long vfs_truncate(struct path *, loff_t);
2222 extern int do_truncate(struct dentry *, loff_t start, unsigned int time_attrs,
2224 extern int vfs_fallocate(struct file *file, int mode, loff_t offset,
2226 extern long do_sys_open(int dfd, const char __user *filename, int flags,
2228 extern struct file *file_open_name(struct filename *, int, umode_t);
2229 extern struct file *filp_open(const char *, int, umode_t);
2230 extern struct file *file_open_root(struct dentry *, struct vfsmount *,
2232 extern struct file * dentry_open(const struct path *, int, const struct cred *);
2233 extern int filp_close(struct file *, fl_owner_t id);
2235 extern struct filename *getname_flags(const char __user *, int, int *);
2236 extern struct filename *getname(const char __user *);
2237 extern struct filename *getname_kernel(const char *);
2238 extern void putname(struct filename *name);
2244 extern int finish_open(struct file *file, struct dentry *dentry,
2245 int (*open)(struct inode *, struct file *),
2247 extern int finish_no_open(struct file *file, struct dentry *dentry);
2251 extern int ioctl_preallocate(struct file *filp, void __user *argp);
2254 extern void __init vfs_caches_init_early(void);
2255 extern void __init vfs_caches_init(void);
2257 extern struct kmem_cache *names_cachep;
2259 #define __getname() kmem_cache_alloc(names_cachep, GFP_KERNEL)
2260 #define __putname(name) kmem_cache_free(names_cachep, (void *)(name))
2263 extern int register_blkdev(unsigned int, const char *);
2264 extern void unregister_blkdev(unsigned int, const char *);
2265 extern struct block_device *bdget(dev_t);
2266 extern struct block_device *bdgrab(struct block_device *bdev);
2267 extern void bd_set_size(struct block_device *, loff_t size);
2268 extern void bd_forget(struct inode *inode);
2269 extern void bdput(struct block_device *);
2270 extern void invalidate_bdev(struct block_device *);
2271 extern void iterate_bdevs(void (*)(struct block_device *, void *), void *);
2272 extern int sync_blockdev(struct block_device *bdev);
2273 extern void kill_bdev(struct block_device *);
2274 extern struct super_block *freeze_bdev(struct block_device *);
2275 extern void emergency_thaw_all(void);
2276 extern int thaw_bdev(struct block_device *bdev, struct super_block *sb);
2277 extern int fsync_bdev(struct block_device *);
2279 extern struct super_block *blockdev_superblock;
2281 static inline bool sb_is_blkdev_sb(struct super_block *sb)
2283 return sb == blockdev_superblock;
2286 static inline void bd_forget(struct inode *inode) {}
2287 static inline int sync_blockdev(struct block_device *bdev) { return 0; }
2288 static inline void kill_bdev(struct block_device *bdev) {}
2289 static inline void invalidate_bdev(struct block_device *bdev) {}
2291 static inline struct super_block *freeze_bdev(struct block_device *sb)
2296 static inline int thaw_bdev(struct block_device *bdev, struct super_block *sb)
2301 static inline void iterate_bdevs(void (*f)(struct block_device *, void *), void *arg)
2305 static inline int sb_is_blkdev_sb(struct super_block *sb)
2310 extern int sync_filesystem(struct super_block *);
2311 extern const struct file_operations def_blk_fops;
2312 extern const struct file_operations def_chr_fops;
2314 extern int ioctl_by_bdev(struct block_device *, unsigned, unsigned long);
2315 extern int blkdev_ioctl(struct block_device *, fmode_t, unsigned, unsigned long);
2316 extern long compat_blkdev_ioctl(struct file *, unsigned, unsigned long);
2317 extern int blkdev_get(struct block_device *bdev, fmode_t mode, void *holder);
2318 extern struct block_device *blkdev_get_by_path(const char *path, fmode_t mode,
2320 extern struct block_device *blkdev_get_by_dev(dev_t dev, fmode_t mode,
2322 extern void blkdev_put(struct block_device *bdev, fmode_t mode);
2323 extern int __blkdev_reread_part(struct block_device *bdev);
2324 extern int blkdev_reread_part(struct block_device *bdev);
2327 extern int bd_link_disk_holder(struct block_device *bdev, struct gendisk *disk);
2328 extern void bd_unlink_disk_holder(struct block_device *bdev,
2329 struct gendisk *disk);
2331 static inline int bd_link_disk_holder(struct block_device *bdev,
2332 struct gendisk *disk)
2336 static inline void bd_unlink_disk_holder(struct block_device *bdev,
2337 struct gendisk *disk)
2344 #define CHRDEV_MAJOR_HASH_SIZE 255
2345 extern int alloc_chrdev_region(dev_t *, unsigned, unsigned, const char *);
2346 extern int register_chrdev_region(dev_t, unsigned, const char *);
2347 extern int __register_chrdev(unsigned int major, unsigned int baseminor,
2348 unsigned int count, const char *name,
2349 const struct file_operations *fops);
2350 extern void __unregister_chrdev(unsigned int major, unsigned int baseminor,
2351 unsigned int count, const char *name);
2352 extern void unregister_chrdev_region(dev_t, unsigned);
2353 extern void chrdev_show(struct seq_file *,off_t);
2355 static inline int register_chrdev(unsigned int major, const char *name,
2356 const struct file_operations *fops)
2358 return __register_chrdev(major, 0, 256, name, fops);
2361 static inline void unregister_chrdev(unsigned int major, const char *name)
2363 __unregister_chrdev(major, 0, 256, name);
2366 /* fs/block_dev.c */
2367 #define BDEVNAME_SIZE 32 /* Largest string for a blockdev identifier */
2368 #define BDEVT_SIZE 10 /* Largest string for MAJ:MIN for blkdev */
2371 #define BLKDEV_MAJOR_HASH_SIZE 255
2372 extern const char *__bdevname(dev_t, char *buffer);
2373 extern const char *bdevname(struct block_device *bdev, char *buffer);
2374 extern struct block_device *lookup_bdev(const char *);
2375 extern void blkdev_show(struct seq_file *,off_t);
2378 #define BLKDEV_MAJOR_HASH_SIZE 0
2381 extern void init_special_inode(struct inode *, umode_t, dev_t);
2383 /* Invalid inode operations -- fs/bad_inode.c */
2384 extern void make_bad_inode(struct inode *);
2385 extern int is_bad_inode(struct inode *);
2389 * return READ, READA, or WRITE
2391 #define bio_rw(bio) ((bio)->bi_rw & (RW_MASK | RWA_MASK))
2394 * return data direction, READ or WRITE
2396 #define bio_data_dir(bio) ((bio)->bi_rw & 1)
2398 extern void check_disk_size_change(struct gendisk *disk,
2399 struct block_device *bdev);
2400 extern int revalidate_disk(struct gendisk *);
2401 extern int check_disk_change(struct block_device *);
2402 extern int __invalidate_device(struct block_device *, bool);
2403 extern int invalidate_partition(struct gendisk *, int);
2405 unsigned long invalidate_mapping_pages(struct address_space *mapping,
2406 pgoff_t start, pgoff_t end);
2408 static inline void invalidate_remote_inode(struct inode *inode)
2410 if (S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) ||
2411 S_ISLNK(inode->i_mode))
2412 invalidate_mapping_pages(inode->i_mapping, 0, -1);
2414 extern int invalidate_inode_pages2(struct address_space *mapping);
2415 extern int invalidate_inode_pages2_range(struct address_space *mapping,
2416 pgoff_t start, pgoff_t end);
2417 extern int write_inode_now(struct inode *, int);
2418 extern int filemap_fdatawrite(struct address_space *);
2419 extern int filemap_flush(struct address_space *);
2420 extern int filemap_fdatawait(struct address_space *);
2421 extern int filemap_fdatawait_range(struct address_space *, loff_t lstart,
2423 extern int filemap_write_and_wait(struct address_space *mapping);
2424 extern int filemap_write_and_wait_range(struct address_space *mapping,
2425 loff_t lstart, loff_t lend);
2426 extern int __filemap_fdatawrite_range(struct address_space *mapping,
2427 loff_t start, loff_t end, int sync_mode);
2428 extern int filemap_fdatawrite_range(struct address_space *mapping,
2429 loff_t start, loff_t end);
2431 extern int vfs_fsync_range(struct file *file, loff_t start, loff_t end,
2433 extern int vfs_fsync(struct file *file, int datasync);
2434 static inline int generic_write_sync(struct file *file, loff_t pos, loff_t count)
2436 if (!(file->f_flags & O_DSYNC) && !IS_SYNC(file->f_mapping->host))
2438 return vfs_fsync_range(file, pos, pos + count - 1,
2439 (file->f_flags & __O_SYNC) ? 0 : 1);
2441 extern void emergency_sync(void);
2442 extern void emergency_remount(void);
2444 extern sector_t bmap(struct inode *, sector_t);
2446 extern int notify_change(struct dentry *, struct iattr *, struct inode **);
2447 extern int inode_permission(struct inode *, int);
2448 extern int __inode_permission(struct inode *, int);
2449 extern int generic_permission(struct inode *, int);
2450 extern int __check_sticky(struct inode *dir, struct inode *inode);
2452 static inline bool execute_ok(struct inode *inode)
2454 return (inode->i_mode & S_IXUGO) || S_ISDIR(inode->i_mode);
2457 static inline void file_start_write(struct file *file)
2459 if (!S_ISREG(file_inode(file)->i_mode))
2461 __sb_start_write(file_inode(file)->i_sb, SB_FREEZE_WRITE, true);
2464 static inline bool file_start_write_trylock(struct file *file)
2466 if (!S_ISREG(file_inode(file)->i_mode))
2468 return __sb_start_write(file_inode(file)->i_sb, SB_FREEZE_WRITE, false);
2471 static inline void file_end_write(struct file *file)
2473 if (!S_ISREG(file_inode(file)->i_mode))
2475 __sb_end_write(file_inode(file)->i_sb, SB_FREEZE_WRITE);
2479 * get_write_access() gets write permission for a file.
2480 * put_write_access() releases this write permission.
2481 * This is used for regular files.
2482 * We cannot support write (and maybe mmap read-write shared) accesses and
2483 * MAP_DENYWRITE mmappings simultaneously. The i_writecount field of an inode
2484 * can have the following values:
2485 * 0: no writers, no VM_DENYWRITE mappings
2486 * < 0: (-i_writecount) vm_area_structs with VM_DENYWRITE set exist
2487 * > 0: (i_writecount) users are writing to the file.
2489 * Normally we operate on that counter with atomic_{inc,dec} and it's safe
2490 * except for the cases where we don't hold i_writecount yet. Then we need to
2491 * use {get,deny}_write_access() - these functions check the sign and refuse
2492 * to do the change if sign is wrong.
2494 static inline int get_write_access(struct inode *inode)
2496 return atomic_inc_unless_negative(&inode->i_writecount) ? 0 : -ETXTBSY;
2498 static inline int deny_write_access(struct file *file)
2500 struct inode *inode = file_inode(file);
2501 return atomic_dec_unless_positive(&inode->i_writecount) ? 0 : -ETXTBSY;
2503 static inline void put_write_access(struct inode * inode)
2505 atomic_dec(&inode->i_writecount);
2507 static inline void allow_write_access(struct file *file)
2510 atomic_inc(&file_inode(file)->i_writecount);
2512 static inline bool inode_is_open_for_write(const struct inode *inode)
2514 return atomic_read(&inode->i_writecount) > 0;
2518 static inline void i_readcount_dec(struct inode *inode)
2520 BUG_ON(!atomic_read(&inode->i_readcount));
2521 atomic_dec(&inode->i_readcount);
2523 static inline void i_readcount_inc(struct inode *inode)
2525 atomic_inc(&inode->i_readcount);
2528 static inline void i_readcount_dec(struct inode *inode)
2532 static inline void i_readcount_inc(struct inode *inode)
2537 extern int do_pipe_flags(int *, int);
2539 extern int kernel_read(struct file *, loff_t, char *, unsigned long);
2540 extern ssize_t kernel_write(struct file *, const char *, size_t, loff_t);
2541 extern ssize_t __kernel_write(struct file *, const char *, size_t, loff_t *);
2542 extern struct file * open_exec(const char *);
2544 /* fs/dcache.c -- generic fs support functions */
2545 extern int is_subdir(struct dentry *, struct dentry *);
2546 extern int path_is_under(struct path *, struct path *);
2548 extern char *file_path(struct file *, char *, int);
2550 #include <linux/err.h>
2552 /* needed for stackable file system support */
2553 extern loff_t default_llseek(struct file *file, loff_t offset, int whence);
2555 extern loff_t vfs_llseek(struct file *file, loff_t offset, int whence);
2557 extern int inode_init_always(struct super_block *, struct inode *);
2558 extern void inode_init_once(struct inode *);
2559 extern void address_space_init_once(struct address_space *mapping);
2560 extern struct inode * igrab(struct inode *);
2561 extern ino_t iunique(struct super_block *, ino_t);
2562 extern int inode_needs_sync(struct inode *inode);
2563 extern int generic_delete_inode(struct inode *inode);
2564 static inline int generic_drop_inode(struct inode *inode)
2566 return !inode->i_nlink || inode_unhashed(inode);
2569 extern struct inode *ilookup5_nowait(struct super_block *sb,
2570 unsigned long hashval, int (*test)(struct inode *, void *),
2572 extern struct inode *ilookup5(struct super_block *sb, unsigned long hashval,
2573 int (*test)(struct inode *, void *), void *data);
2574 extern struct inode *ilookup(struct super_block *sb, unsigned long ino);
2576 extern struct inode * iget5_locked(struct super_block *, unsigned long, int (*test)(struct inode *, void *), int (*set)(struct inode *, void *), void *);
2577 extern struct inode * iget_locked(struct super_block *, unsigned long);
2578 extern struct inode *find_inode_nowait(struct super_block *,
2580 int (*match)(struct inode *,
2581 unsigned long, void *),
2583 extern int insert_inode_locked4(struct inode *, unsigned long, int (*test)(struct inode *, void *), void *);
2584 extern int insert_inode_locked(struct inode *);
2585 #ifdef CONFIG_DEBUG_LOCK_ALLOC
2586 extern void lockdep_annotate_inode_mutex_key(struct inode *inode);
2588 static inline void lockdep_annotate_inode_mutex_key(struct inode *inode) { };
2590 extern void unlock_new_inode(struct inode *);
2591 extern unsigned int get_next_ino(void);
2593 extern void __iget(struct inode * inode);
2594 extern void iget_failed(struct inode *);
2595 extern void clear_inode(struct inode *);
2596 extern void __destroy_inode(struct inode *);
2597 extern struct inode *new_inode_pseudo(struct super_block *sb);
2598 extern struct inode *new_inode(struct super_block *sb);
2599 extern void free_inode_nonrcu(struct inode *inode);
2600 extern int should_remove_suid(struct dentry *);
2601 extern int file_remove_privs(struct file *);
2602 extern int dentry_needs_remove_privs(struct dentry *dentry);
2603 static inline int file_needs_remove_privs(struct file *file)
2605 return dentry_needs_remove_privs(file->f_path.dentry);
2608 extern void __insert_inode_hash(struct inode *, unsigned long hashval);
2609 static inline void insert_inode_hash(struct inode *inode)
2611 __insert_inode_hash(inode, inode->i_ino);
2614 extern void __remove_inode_hash(struct inode *);
2615 static inline void remove_inode_hash(struct inode *inode)
2617 if (!inode_unhashed(inode))
2618 __remove_inode_hash(inode);
2621 extern void inode_sb_list_add(struct inode *inode);
2624 extern void submit_bio(int, struct bio *);
2625 extern int bdev_read_only(struct block_device *);
2627 extern int set_blocksize(struct block_device *, int);
2628 extern int sb_set_blocksize(struct super_block *, int);
2629 extern int sb_min_blocksize(struct super_block *, int);
2631 extern int generic_file_mmap(struct file *, struct vm_area_struct *);
2632 extern int generic_file_readonly_mmap(struct file *, struct vm_area_struct *);
2633 extern ssize_t generic_write_checks(struct kiocb *, struct iov_iter *);
2634 extern ssize_t generic_file_read_iter(struct kiocb *, struct iov_iter *);
2635 extern ssize_t __generic_file_write_iter(struct kiocb *, struct iov_iter *);
2636 extern ssize_t generic_file_write_iter(struct kiocb *, struct iov_iter *);
2637 extern ssize_t generic_file_direct_write(struct kiocb *, struct iov_iter *, loff_t);
2638 extern ssize_t generic_perform_write(struct file *, struct iov_iter *, loff_t);
2640 ssize_t vfs_iter_read(struct file *file, struct iov_iter *iter, loff_t *ppos);
2641 ssize_t vfs_iter_write(struct file *file, struct iov_iter *iter, loff_t *ppos);
2643 /* fs/block_dev.c */
2644 extern ssize_t blkdev_read_iter(struct kiocb *iocb, struct iov_iter *to);
2645 extern ssize_t blkdev_write_iter(struct kiocb *iocb, struct iov_iter *from);
2646 extern int blkdev_fsync(struct file *filp, loff_t start, loff_t end,
2648 extern void block_sync_page(struct page *page);
2651 extern ssize_t generic_file_splice_read(struct file *, loff_t *,
2652 struct pipe_inode_info *, size_t, unsigned int);
2653 extern ssize_t default_file_splice_read(struct file *, loff_t *,
2654 struct pipe_inode_info *, size_t, unsigned int);
2655 extern ssize_t iter_file_splice_write(struct pipe_inode_info *,
2656 struct file *, loff_t *, size_t, unsigned int);
2657 extern ssize_t generic_splice_sendpage(struct pipe_inode_info *pipe,
2658 struct file *out, loff_t *, size_t len, unsigned int flags);
2659 extern long do_splice_direct(struct file *in, loff_t *ppos, struct file *out,
2660 loff_t *opos, size_t len, unsigned int flags);
2664 file_ra_state_init(struct file_ra_state *ra, struct address_space *mapping);
2665 extern loff_t noop_llseek(struct file *file, loff_t offset, int whence);
2666 extern loff_t no_llseek(struct file *file, loff_t offset, int whence);
2667 extern loff_t vfs_setpos(struct file *file, loff_t offset, loff_t maxsize);
2668 extern loff_t generic_file_llseek(struct file *file, loff_t offset, int whence);
2669 extern loff_t generic_file_llseek_size(struct file *file, loff_t offset,
2670 int whence, loff_t maxsize, loff_t eof);
2671 extern loff_t fixed_size_llseek(struct file *file, loff_t offset,
2672 int whence, loff_t size);
2673 extern int generic_file_open(struct inode * inode, struct file * filp);
2674 extern int nonseekable_open(struct inode * inode, struct file * filp);
2676 ssize_t dax_do_io(struct kiocb *, struct inode *, struct iov_iter *, loff_t,
2677 get_block_t, dio_iodone_t, int flags);
2678 int dax_clear_blocks(struct inode *, sector_t block, long size);
2679 int dax_zero_page_range(struct inode *, loff_t from, unsigned len, get_block_t);
2680 int dax_truncate_page(struct inode *, loff_t from, get_block_t);
2681 int dax_fault(struct vm_area_struct *, struct vm_fault *, get_block_t,
2683 int __dax_fault(struct vm_area_struct *, struct vm_fault *, get_block_t,
2685 int dax_pfn_mkwrite(struct vm_area_struct *, struct vm_fault *);
2686 #define dax_mkwrite(vma, vmf, gb, iod) dax_fault(vma, vmf, gb, iod)
2687 #define __dax_mkwrite(vma, vmf, gb, iod) __dax_fault(vma, vmf, gb, iod)
2690 typedef void (dio_submit_t)(int rw, struct bio *bio, struct inode *inode,
2691 loff_t file_offset);
2694 /* need locking between buffered and direct access */
2697 /* filesystem does not support filling holes */
2698 DIO_SKIP_HOLES = 0x02,
2700 /* filesystem can handle aio writes beyond i_size */
2701 DIO_ASYNC_EXTEND = 0x04,
2703 /* inode/fs/bdev does not need truncate protection */
2704 DIO_SKIP_DIO_COUNT = 0x08,
2707 void dio_end_io(struct bio *bio, int error);
2709 ssize_t __blockdev_direct_IO(struct kiocb *iocb, struct inode *inode,
2710 struct block_device *bdev, struct iov_iter *iter,
2711 loff_t offset, get_block_t get_block,
2712 dio_iodone_t end_io, dio_submit_t submit_io,
2715 static inline ssize_t blockdev_direct_IO(struct kiocb *iocb,
2716 struct inode *inode,
2717 struct iov_iter *iter, loff_t offset,
2718 get_block_t get_block)
2720 return __blockdev_direct_IO(iocb, inode, inode->i_sb->s_bdev, iter,
2721 offset, get_block, NULL, NULL,
2722 DIO_LOCKING | DIO_SKIP_HOLES);
2726 void inode_dio_wait(struct inode *inode);
2729 * inode_dio_begin - signal start of a direct I/O requests
2730 * @inode: inode the direct I/O happens on
2732 * This is called once we've finished processing a direct I/O request,
2733 * and is used to wake up callers waiting for direct I/O to be quiesced.
2735 static inline void inode_dio_begin(struct inode *inode)
2737 atomic_inc(&inode->i_dio_count);
2741 * inode_dio_end - signal finish of a direct I/O requests
2742 * @inode: inode the direct I/O happens on
2744 * This is called once we've finished processing a direct I/O request,
2745 * and is used to wake up callers waiting for direct I/O to be quiesced.
2747 static inline void inode_dio_end(struct inode *inode)
2749 if (atomic_dec_and_test(&inode->i_dio_count))
2750 wake_up_bit(&inode->i_state, __I_DIO_WAKEUP);
2753 extern void inode_set_flags(struct inode *inode, unsigned int flags,
2756 extern const struct file_operations generic_ro_fops;
2758 #define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
2760 extern int readlink_copy(char __user *, int, const char *);
2761 extern int page_readlink(struct dentry *, char __user *, int);
2762 extern const char *page_follow_link_light(struct dentry *, void **);
2763 extern void page_put_link(struct inode *, void *);
2764 extern int __page_symlink(struct inode *inode, const char *symname, int len,
2766 extern int page_symlink(struct inode *inode, const char *symname, int len);
2767 extern const struct inode_operations page_symlink_inode_operations;
2768 extern void kfree_put_link(struct inode *, void *);
2769 extern void free_page_put_link(struct inode *, void *);
2770 extern int generic_readlink(struct dentry *, char __user *, int);
2771 extern void generic_fillattr(struct inode *, struct kstat *);
2772 int vfs_getattr_nosec(struct path *path, struct kstat *stat);
2773 extern int vfs_getattr(struct path *, struct kstat *);
2774 void __inode_add_bytes(struct inode *inode, loff_t bytes);
2775 void inode_add_bytes(struct inode *inode, loff_t bytes);
2776 void __inode_sub_bytes(struct inode *inode, loff_t bytes);
2777 void inode_sub_bytes(struct inode *inode, loff_t bytes);
2778 loff_t inode_get_bytes(struct inode *inode);
2779 void inode_set_bytes(struct inode *inode, loff_t bytes);
2780 const char *simple_follow_link(struct dentry *, void **);
2781 extern const struct inode_operations simple_symlink_inode_operations;
2783 extern int iterate_dir(struct file *, struct dir_context *);
2785 extern int vfs_stat(const char __user *, struct kstat *);
2786 extern int vfs_lstat(const char __user *, struct kstat *);
2787 extern int vfs_fstat(unsigned int, struct kstat *);
2788 extern int vfs_fstatat(int , const char __user *, struct kstat *, int);
2790 extern int do_vfs_ioctl(struct file *filp, unsigned int fd, unsigned int cmd,
2792 extern int __generic_block_fiemap(struct inode *inode,
2793 struct fiemap_extent_info *fieinfo,
2794 loff_t start, loff_t len,
2795 get_block_t *get_block);
2796 extern int generic_block_fiemap(struct inode *inode,
2797 struct fiemap_extent_info *fieinfo, u64 start,
2798 u64 len, get_block_t *get_block);
2800 extern void get_filesystem(struct file_system_type *fs);
2801 extern void put_filesystem(struct file_system_type *fs);
2802 extern struct file_system_type *get_fs_type(const char *name);
2803 extern struct super_block *get_super(struct block_device *);
2804 extern struct super_block *get_super_thawed(struct block_device *);
2805 extern struct super_block *get_active_super(struct block_device *bdev);
2806 extern void drop_super(struct super_block *sb);
2807 extern void iterate_supers(void (*)(struct super_block *, void *), void *);
2808 extern void iterate_supers_type(struct file_system_type *,
2809 void (*)(struct super_block *, void *), void *);
2811 extern int dcache_dir_open(struct inode *, struct file *);
2812 extern int dcache_dir_close(struct inode *, struct file *);
2813 extern loff_t dcache_dir_lseek(struct file *, loff_t, int);
2814 extern int dcache_readdir(struct file *, struct dir_context *);
2815 extern int simple_setattr(struct dentry *, struct iattr *);
2816 extern int simple_getattr(struct vfsmount *, struct dentry *, struct kstat *);
2817 extern int simple_statfs(struct dentry *, struct kstatfs *);
2818 extern int simple_open(struct inode *inode, struct file *file);
2819 extern int simple_link(struct dentry *, struct inode *, struct dentry *);
2820 extern int simple_unlink(struct inode *, struct dentry *);
2821 extern int simple_rmdir(struct inode *, struct dentry *);
2822 extern int simple_rename(struct inode *, struct dentry *, struct inode *, struct dentry *);
2823 extern int noop_fsync(struct file *, loff_t, loff_t, int);
2824 extern int simple_empty(struct dentry *);
2825 extern int simple_readpage(struct file *file, struct page *page);
2826 extern int simple_write_begin(struct file *file, struct address_space *mapping,
2827 loff_t pos, unsigned len, unsigned flags,
2828 struct page **pagep, void **fsdata);
2829 extern int simple_write_end(struct file *file, struct address_space *mapping,
2830 loff_t pos, unsigned len, unsigned copied,
2831 struct page *page, void *fsdata);
2832 extern int always_delete_dentry(const struct dentry *);
2833 extern struct inode *alloc_anon_inode(struct super_block *);
2834 extern int simple_nosetlease(struct file *, long, struct file_lock **, void **);
2835 extern const struct dentry_operations simple_dentry_operations;
2837 extern struct dentry *simple_lookup(struct inode *, struct dentry *, unsigned int flags);
2838 extern ssize_t generic_read_dir(struct file *, char __user *, size_t, loff_t *);
2839 extern const struct file_operations simple_dir_operations;
2840 extern const struct inode_operations simple_dir_inode_operations;
2841 extern void make_empty_dir_inode(struct inode *inode);
2842 extern bool is_empty_dir_inode(struct inode *inode);
2843 struct tree_descr { char *name; const struct file_operations *ops; int mode; };
2844 struct dentry *d_alloc_name(struct dentry *, const char *);
2845 extern int simple_fill_super(struct super_block *, unsigned long, struct tree_descr *);
2846 extern int simple_pin_fs(struct file_system_type *, struct vfsmount **mount, int *count);
2847 extern void simple_release_fs(struct vfsmount **mount, int *count);
2849 extern ssize_t simple_read_from_buffer(void __user *to, size_t count,
2850 loff_t *ppos, const void *from, size_t available);
2851 extern ssize_t simple_write_to_buffer(void *to, size_t available, loff_t *ppos,
2852 const void __user *from, size_t count);
2854 extern int __generic_file_fsync(struct file *, loff_t, loff_t, int);
2855 extern int generic_file_fsync(struct file *, loff_t, loff_t, int);
2857 extern int generic_check_addressable(unsigned, u64);
2859 #ifdef CONFIG_MIGRATION
2860 extern int buffer_migrate_page(struct address_space *,
2861 struct page *, struct page *,
2864 #define buffer_migrate_page NULL
2867 extern int inode_change_ok(const struct inode *, struct iattr *);
2868 extern int inode_newsize_ok(const struct inode *, loff_t offset);
2869 extern void setattr_copy(struct inode *inode, const struct iattr *attr);
2871 extern int file_update_time(struct file *file);
2873 extern int generic_show_options(struct seq_file *m, struct dentry *root);
2874 extern void save_mount_options(struct super_block *sb, char *options);
2875 extern void replace_mount_options(struct super_block *sb, char *options);
2877 static inline bool io_is_direct(struct file *filp)
2879 return (filp->f_flags & O_DIRECT) || IS_DAX(file_inode(filp));
2882 static inline int iocb_flags(struct file *file)
2885 if (file->f_flags & O_APPEND)
2887 if (io_is_direct(file))
2892 static inline ino_t parent_ino(struct dentry *dentry)
2897 * Don't strictly need d_lock here? If the parent ino could change
2898 * then surely we'd have a deeper race in the caller?
2900 spin_lock(&dentry->d_lock);
2901 res = dentry->d_parent->d_inode->i_ino;
2902 spin_unlock(&dentry->d_lock);
2906 /* Transaction based IO helpers */
2909 * An argresp is stored in an allocated page and holds the
2910 * size of the argument or response, along with its content
2912 struct simple_transaction_argresp {
2917 #define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
2919 char *simple_transaction_get(struct file *file, const char __user *buf,
2921 ssize_t simple_transaction_read(struct file *file, char __user *buf,
2922 size_t size, loff_t *pos);
2923 int simple_transaction_release(struct inode *inode, struct file *file);
2925 void simple_transaction_set(struct file *file, size_t n);
2928 * simple attribute files
2930 * These attributes behave similar to those in sysfs:
2932 * Writing to an attribute immediately sets a value, an open file can be
2933 * written to multiple times.
2935 * Reading from an attribute creates a buffer from the value that might get
2936 * read with multiple read calls. When the attribute has been read
2937 * completely, no further read calls are possible until the file is opened
2940 * All attributes contain a text representation of a numeric value
2941 * that are accessed with the get() and set() functions.
2943 #define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt) \
2944 static int __fops ## _open(struct inode *inode, struct file *file) \
2946 __simple_attr_check_format(__fmt, 0ull); \
2947 return simple_attr_open(inode, file, __get, __set, __fmt); \
2949 static const struct file_operations __fops = { \
2950 .owner = THIS_MODULE, \
2951 .open = __fops ## _open, \
2952 .release = simple_attr_release, \
2953 .read = simple_attr_read, \
2954 .write = simple_attr_write, \
2955 .llseek = generic_file_llseek, \
2958 static inline __printf(1, 2)
2959 void __simple_attr_check_format(const char *fmt, ...)
2961 /* don't do anything, just let the compiler check the arguments; */
2964 int simple_attr_open(struct inode *inode, struct file *file,
2965 int (*get)(void *, u64 *), int (*set)(void *, u64),
2967 int simple_attr_release(struct inode *inode, struct file *file);
2968 ssize_t simple_attr_read(struct file *file, char __user *buf,
2969 size_t len, loff_t *ppos);
2970 ssize_t simple_attr_write(struct file *file, const char __user *buf,
2971 size_t len, loff_t *ppos);
2974 int proc_nr_files(struct ctl_table *table, int write,
2975 void __user *buffer, size_t *lenp, loff_t *ppos);
2976 int proc_nr_dentry(struct ctl_table *table, int write,
2977 void __user *buffer, size_t *lenp, loff_t *ppos);
2978 int proc_nr_inodes(struct ctl_table *table, int write,
2979 void __user *buffer, size_t *lenp, loff_t *ppos);
2980 int __init get_filesystem_list(char *buf);
2982 #define __FMODE_EXEC ((__force int) FMODE_EXEC)
2983 #define __FMODE_NONOTIFY ((__force int) FMODE_NONOTIFY)
2985 #define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE])
2986 #define OPEN_FMODE(flag) ((__force fmode_t)(((flag + 1) & O_ACCMODE) | \
2987 (flag & __FMODE_NONOTIFY)))
2989 static inline int is_sxid(umode_t mode)
2991 return (mode & S_ISUID) || ((mode & S_ISGID) && (mode & S_IXGRP));
2994 static inline int check_sticky(struct inode *dir, struct inode *inode)
2996 if (!(dir->i_mode & S_ISVTX))
2999 return __check_sticky(dir, inode);
3002 static inline void inode_has_no_xattr(struct inode *inode)
3004 if (!is_sxid(inode->i_mode) && (inode->i_sb->s_flags & MS_NOSEC))
3005 inode->i_flags |= S_NOSEC;
3008 static inline bool is_root_inode(struct inode *inode)
3010 return inode == inode->i_sb->s_root->d_inode;
3013 static inline bool dir_emit(struct dir_context *ctx,
3014 const char *name, int namelen,
3015 u64 ino, unsigned type)
3017 return ctx->actor(ctx, name, namelen, ctx->pos, ino, type) == 0;
3019 static inline bool dir_emit_dot(struct file *file, struct dir_context *ctx)
3021 return ctx->actor(ctx, ".", 1, ctx->pos,
3022 file->f_path.dentry->d_inode->i_ino, DT_DIR) == 0;
3024 static inline bool dir_emit_dotdot(struct file *file, struct dir_context *ctx)
3026 return ctx->actor(ctx, "..", 2, ctx->pos,
3027 parent_ino(file->f_path.dentry), DT_DIR) == 0;
3029 static inline bool dir_emit_dots(struct file *file, struct dir_context *ctx)
3031 if (ctx->pos == 0) {
3032 if (!dir_emit_dot(file, ctx))
3036 if (ctx->pos == 1) {
3037 if (!dir_emit_dotdot(file, ctx))
3043 static inline bool dir_relax(struct inode *inode)
3045 mutex_unlock(&inode->i_mutex);
3046 mutex_lock(&inode->i_mutex);
3047 return !IS_DEADDIR(inode);
3050 #endif /* _LINUX_FS_H */