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/capability.h>
22 #include <linux/semaphore.h>
23 #include <linux/fiemap.h>
24 #include <linux/rculist_bl.h>
25 #include <linux/atomic.h>
26 #include <linux/shrinker.h>
27 #include <linux/migrate_mode.h>
28 #include <linux/uidgid.h>
29 #include <linux/lockdep.h>
30 #include <linux/percpu-rwsem.h>
31 #include <linux/blk_types.h>
33 #include <asm/byteorder.h>
34 #include <uapi/linux/fs.h>
36 struct export_operations;
42 struct pipe_inode_info;
43 struct poll_table_struct;
45 struct vm_area_struct;
48 struct swap_info_struct;
50 struct workqueue_struct;
53 extern void __init inode_init(void);
54 extern void __init inode_init_early(void);
55 extern void __init files_init(unsigned long);
57 extern struct files_stat_struct files_stat;
58 extern unsigned long get_max_files(void);
59 extern int sysctl_nr_open;
60 extern struct inodes_stat_t inodes_stat;
61 extern int leases_enable, lease_break_time;
62 extern int sysctl_protected_symlinks;
63 extern int sysctl_protected_hardlinks;
66 typedef int (get_block_t)(struct inode *inode, sector_t iblock,
67 struct buffer_head *bh_result, int create);
68 typedef void (dio_iodone_t)(struct kiocb *iocb, loff_t offset,
69 ssize_t bytes, void *private);
71 #define MAY_EXEC 0x00000001
72 #define MAY_WRITE 0x00000002
73 #define MAY_READ 0x00000004
74 #define MAY_APPEND 0x00000008
75 #define MAY_ACCESS 0x00000010
76 #define MAY_OPEN 0x00000020
77 #define MAY_CHDIR 0x00000040
78 /* called from RCU mode, don't block */
79 #define MAY_NOT_BLOCK 0x00000080
82 * flags in file.f_mode. Note that FMODE_READ and FMODE_WRITE must correspond
83 * to O_WRONLY and O_RDWR via the strange trick in __dentry_open()
86 /* file is open for reading */
87 #define FMODE_READ ((__force fmode_t)0x1)
88 /* file is open for writing */
89 #define FMODE_WRITE ((__force fmode_t)0x2)
90 /* file is seekable */
91 #define FMODE_LSEEK ((__force fmode_t)0x4)
92 /* file can be accessed using pread */
93 #define FMODE_PREAD ((__force fmode_t)0x8)
94 /* file can be accessed using pwrite */
95 #define FMODE_PWRITE ((__force fmode_t)0x10)
96 /* File is opened for execution with sys_execve / sys_uselib */
97 #define FMODE_EXEC ((__force fmode_t)0x20)
98 /* File is opened with O_NDELAY (only set for block devices) */
99 #define FMODE_NDELAY ((__force fmode_t)0x40)
100 /* File is opened with O_EXCL (only set for block devices) */
101 #define FMODE_EXCL ((__force fmode_t)0x80)
102 /* File is opened using open(.., 3, ..) and is writeable only for ioctls
103 (specialy hack for floppy.c) */
104 #define FMODE_WRITE_IOCTL ((__force fmode_t)0x100)
105 /* 32bit hashes as llseek() offset (for directories) */
106 #define FMODE_32BITHASH ((__force fmode_t)0x200)
107 /* 64bit hashes as llseek() offset (for directories) */
108 #define FMODE_64BITHASH ((__force fmode_t)0x400)
111 * Don't update ctime and mtime.
113 * Currently a special hack for the XFS open_by_handle ioctl, but we'll
114 * hopefully graduate it to a proper O_CMTIME flag supported by open(2) soon.
116 #define FMODE_NOCMTIME ((__force fmode_t)0x800)
118 /* Expect random access pattern */
119 #define FMODE_RANDOM ((__force fmode_t)0x1000)
121 /* File is huge (eg. /dev/kmem): treat loff_t as unsigned */
122 #define FMODE_UNSIGNED_OFFSET ((__force fmode_t)0x2000)
124 /* File is opened with O_PATH; almost nothing can be done with it */
125 #define FMODE_PATH ((__force fmode_t)0x4000)
127 /* File needs atomic accesses to f_pos */
128 #define FMODE_ATOMIC_POS ((__force fmode_t)0x8000)
129 /* Write access to underlying fs */
130 #define FMODE_WRITER ((__force fmode_t)0x10000)
131 /* Has read method(s) */
132 #define FMODE_CAN_READ ((__force fmode_t)0x20000)
133 /* Has write method(s) */
134 #define FMODE_CAN_WRITE ((__force fmode_t)0x40000)
136 /* File was opened by fanotify and shouldn't generate fanotify events */
137 #define FMODE_NONOTIFY ((__force fmode_t)0x1000000)
140 * Flag for rw_copy_check_uvector and compat_rw_copy_check_uvector
141 * that indicates that they should check the contents of the iovec are
142 * valid, but not check the memory that the iovec elements
145 #define CHECK_IOVEC_ONLY -1
148 * The below are the various read and write types that we support. Some of
149 * them include behavioral modifiers that send information down to the
150 * block layer and IO scheduler. Terminology:
152 * The block layer uses device plugging to defer IO a little bit, in
153 * the hope that we will see more IO very shortly. This increases
154 * coalescing of adjacent IO and thus reduces the number of IOs we
155 * have to send to the device. It also allows for better queuing,
156 * if the IO isn't mergeable. If the caller is going to be waiting
157 * for the IO, then he must ensure that the device is unplugged so
158 * that the IO is dispatched to the driver.
160 * All IO is handled async in Linux. This is fine for background
161 * writes, but for reads or writes that someone waits for completion
162 * on, we want to notify the block layer and IO scheduler so that they
163 * know about it. That allows them to make better scheduling
164 * decisions. So when the below references 'sync' and 'async', it
165 * is referencing this priority hint.
167 * With that in mind, the available types are:
169 * READ A normal read operation. Device will be plugged.
170 * READ_SYNC A synchronous read. Device is not plugged, caller can
171 * immediately wait on this read without caring about
173 * READA Used for read-ahead operations. Lower priority, and the
174 * block layer could (in theory) choose to ignore this
175 * request if it runs into resource problems.
176 * WRITE A normal async write. Device will be plugged.
177 * WRITE_SYNC Synchronous write. Identical to WRITE, but passes down
178 * the hint that someone will be waiting on this IO
179 * shortly. The write equivalent of READ_SYNC.
180 * WRITE_ODIRECT Special case write for O_DIRECT only.
181 * WRITE_FLUSH Like WRITE_SYNC but with preceding cache flush.
182 * WRITE_FUA Like WRITE_SYNC but data is guaranteed to be on
183 * non-volatile media on completion.
184 * WRITE_FLUSH_FUA Combination of WRITE_FLUSH and FUA. The IO is preceded
185 * by a cache flush and data is guaranteed to be on
186 * non-volatile media on completion.
189 #define RW_MASK REQ_WRITE
190 #define RWA_MASK REQ_RAHEAD
193 #define WRITE RW_MASK
194 #define READA RWA_MASK
196 #define READ_SYNC (READ | REQ_SYNC)
197 #define WRITE_SYNC (WRITE | REQ_SYNC | REQ_NOIDLE)
198 #define WRITE_ODIRECT (WRITE | REQ_SYNC)
199 #define WRITE_FLUSH (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FLUSH)
200 #define WRITE_FUA (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FUA)
201 #define WRITE_FLUSH_FUA (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FLUSH | REQ_FUA)
204 * Attribute flags. These should be or-ed together to figure out what
207 #define ATTR_MODE (1 << 0)
208 #define ATTR_UID (1 << 1)
209 #define ATTR_GID (1 << 2)
210 #define ATTR_SIZE (1 << 3)
211 #define ATTR_ATIME (1 << 4)
212 #define ATTR_MTIME (1 << 5)
213 #define ATTR_CTIME (1 << 6)
214 #define ATTR_ATIME_SET (1 << 7)
215 #define ATTR_MTIME_SET (1 << 8)
216 #define ATTR_FORCE (1 << 9) /* Not a change, but a change it */
217 #define ATTR_ATTR_FLAG (1 << 10)
218 #define ATTR_KILL_SUID (1 << 11)
219 #define ATTR_KILL_SGID (1 << 12)
220 #define ATTR_FILE (1 << 13)
221 #define ATTR_KILL_PRIV (1 << 14)
222 #define ATTR_OPEN (1 << 15) /* Truncating from open(O_TRUNC) */
223 #define ATTR_TIMES_SET (1 << 16)
226 * Whiteout is represented by a char device. The following constants define the
227 * mode and device number to use.
229 #define WHITEOUT_MODE 0
230 #define WHITEOUT_DEV 0
233 * This is the Inode Attributes structure, used for notify_change(). It
234 * uses the above definitions as flags, to know which values have changed.
235 * Also, in this manner, a Filesystem can look at only the values it cares
236 * about. Basically, these are the attributes that the VFS layer can
237 * request to change from the FS layer.
239 * Derek Atkins <warlord@MIT.EDU> 94-10-20
242 unsigned int ia_valid;
247 struct timespec ia_atime;
248 struct timespec ia_mtime;
249 struct timespec ia_ctime;
252 * Not an attribute, but an auxiliary info for filesystems wanting to
253 * implement an ftruncate() like method. NOTE: filesystem should
254 * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL).
256 struct file *ia_file;
260 * Includes for diskquotas.
262 #include <linux/quota.h>
265 * Maximum number of layers of fs stack. Needs to be limited to
266 * prevent kernel stack overflow
268 #define FILESYSTEM_MAX_STACK_DEPTH 2
271 * enum positive_aop_returns - aop return codes with specific semantics
273 * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has
274 * completed, that the page is still locked, and
275 * should be considered active. The VM uses this hint
276 * to return the page to the active list -- it won't
277 * be a candidate for writeback again in the near
278 * future. Other callers must be careful to unlock
279 * the page if they get this return. Returned by
282 * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has
283 * unlocked it and the page might have been truncated.
284 * The caller should back up to acquiring a new page and
285 * trying again. The aop will be taking reasonable
286 * precautions not to livelock. If the caller held a page
287 * reference, it should drop it before retrying. Returned
290 * address_space_operation functions return these large constants to indicate
291 * special semantics to the caller. These are much larger than the bytes in a
292 * page to allow for functions that return the number of bytes operated on in a
296 enum positive_aop_returns {
297 AOP_WRITEPAGE_ACTIVATE = 0x80000,
298 AOP_TRUNCATED_PAGE = 0x80001,
301 #define AOP_FLAG_UNINTERRUPTIBLE 0x0001 /* will not do a short write */
302 #define AOP_FLAG_CONT_EXPAND 0x0002 /* called from cont_expand */
303 #define AOP_FLAG_NOFS 0x0004 /* used by filesystem to direct
304 * helper code (eg buffer layer)
305 * to clear GFP_FS from alloc */
308 * oh the beauties of C type declarations.
311 struct address_space;
312 struct writeback_control;
315 * "descriptor" for what we're up to with a read.
316 * This allows us to use the same read code yet
317 * have multiple different users of the data that
318 * we read from a file.
320 * The simplest case just copies the data to user
333 typedef int (*read_actor_t)(read_descriptor_t *, struct page *,
334 unsigned long, unsigned long);
336 struct address_space_operations {
337 int (*writepage)(struct page *page, struct writeback_control *wbc);
338 int (*readpage)(struct file *, struct page *);
340 /* Write back some dirty pages from this mapping. */
341 int (*writepages)(struct address_space *, struct writeback_control *);
343 /* Set a page dirty. Return true if this dirtied it */
344 int (*set_page_dirty)(struct page *page);
346 int (*readpages)(struct file *filp, struct address_space *mapping,
347 struct list_head *pages, unsigned nr_pages);
349 int (*write_begin)(struct file *, struct address_space *mapping,
350 loff_t pos, unsigned len, unsigned flags,
351 struct page **pagep, void **fsdata);
352 int (*write_end)(struct file *, struct address_space *mapping,
353 loff_t pos, unsigned len, unsigned copied,
354 struct page *page, void *fsdata);
356 /* Unfortunately this kludge is needed for FIBMAP. Don't use it */
357 sector_t (*bmap)(struct address_space *, sector_t);
358 void (*invalidatepage) (struct page *, unsigned int, unsigned int);
359 int (*releasepage) (struct page *, gfp_t);
360 void (*freepage)(struct page *);
361 ssize_t (*direct_IO)(int, struct kiocb *, struct iov_iter *iter, loff_t offset);
362 int (*get_xip_mem)(struct address_space *, pgoff_t, int,
363 void **, unsigned long *);
365 * migrate the contents of a page to the specified target. If
366 * migrate_mode is MIGRATE_ASYNC, it must not block.
368 int (*migratepage) (struct address_space *,
369 struct page *, struct page *, enum migrate_mode);
370 int (*launder_page) (struct page *);
371 int (*is_partially_uptodate) (struct page *, unsigned long,
373 void (*is_dirty_writeback) (struct page *, bool *, bool *);
374 int (*error_remove_page)(struct address_space *, struct page *);
376 /* swapfile support */
377 int (*swap_activate)(struct swap_info_struct *sis, struct file *file,
379 void (*swap_deactivate)(struct file *file);
382 extern const struct address_space_operations empty_aops;
385 * pagecache_write_begin/pagecache_write_end must be used by general code
386 * to write into the pagecache.
388 int pagecache_write_begin(struct file *, struct address_space *mapping,
389 loff_t pos, unsigned len, unsigned flags,
390 struct page **pagep, void **fsdata);
392 int pagecache_write_end(struct file *, struct address_space *mapping,
393 loff_t pos, unsigned len, unsigned copied,
394 struct page *page, void *fsdata);
396 struct backing_dev_info;
397 struct address_space {
398 struct inode *host; /* owner: inode, block_device */
399 struct radix_tree_root page_tree; /* radix tree of all pages */
400 spinlock_t tree_lock; /* and lock protecting it */
401 atomic_t i_mmap_writable;/* count VM_SHARED mappings */
402 struct rb_root i_mmap; /* tree of private and shared mappings */
403 struct list_head i_mmap_nonlinear;/*list VM_NONLINEAR mappings */
404 struct mutex i_mmap_mutex; /* protect tree, count, list */
405 /* Protected by tree_lock together with the radix tree */
406 unsigned long nrpages; /* number of total pages */
407 unsigned long nrshadows; /* number of shadow entries */
408 pgoff_t writeback_index;/* writeback starts here */
409 const struct address_space_operations *a_ops; /* methods */
410 unsigned long flags; /* error bits/gfp mask */
411 struct backing_dev_info *backing_dev_info; /* device readahead, etc */
412 spinlock_t private_lock; /* for use by the address_space */
413 struct list_head private_list; /* ditto */
414 void *private_data; /* ditto */
415 } __attribute__((aligned(sizeof(long))));
417 * On most architectures that alignment is already the case; but
418 * must be enforced here for CRIS, to let the least significant bit
419 * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
421 struct request_queue;
423 struct block_device {
424 dev_t bd_dev; /* not a kdev_t - it's a search key */
426 struct inode * bd_inode; /* will die */
427 struct super_block * bd_super;
428 struct mutex bd_mutex; /* open/close mutex */
429 struct list_head bd_inodes;
433 bool bd_write_holder;
435 struct list_head bd_holder_disks;
437 struct block_device * bd_contains;
438 unsigned bd_block_size;
439 struct hd_struct * bd_part;
440 /* number of times partitions within this device have been opened. */
441 unsigned bd_part_count;
443 struct gendisk * bd_disk;
444 struct request_queue * bd_queue;
445 struct list_head bd_list;
447 * Private data. You must have bd_claim'ed the block_device
448 * to use this. NOTE: bd_claim allows an owner to claim
449 * the same device multiple times, the owner must take special
450 * care to not mess up bd_private for that case.
452 unsigned long bd_private;
454 /* The counter of freeze processes */
455 int bd_fsfreeze_count;
456 /* Mutex for freeze */
457 struct mutex bd_fsfreeze_mutex;
461 * Radix-tree tags, for tagging dirty and writeback pages within the pagecache
464 #define PAGECACHE_TAG_DIRTY 0
465 #define PAGECACHE_TAG_WRITEBACK 1
466 #define PAGECACHE_TAG_TOWRITE 2
468 int mapping_tagged(struct address_space *mapping, int tag);
471 * Might pages of this file be mapped into userspace?
473 static inline int mapping_mapped(struct address_space *mapping)
475 return !RB_EMPTY_ROOT(&mapping->i_mmap) ||
476 !list_empty(&mapping->i_mmap_nonlinear);
480 * Might pages of this file have been modified in userspace?
481 * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap_pgoff
482 * marks vma as VM_SHARED if it is shared, and the file was opened for
483 * writing i.e. vma may be mprotected writable even if now readonly.
485 * If i_mmap_writable is negative, no new writable mappings are allowed. You
486 * can only deny writable mappings, if none exists right now.
488 static inline int mapping_writably_mapped(struct address_space *mapping)
490 return atomic_read(&mapping->i_mmap_writable) > 0;
493 static inline int mapping_map_writable(struct address_space *mapping)
495 return atomic_inc_unless_negative(&mapping->i_mmap_writable) ?
499 static inline void mapping_unmap_writable(struct address_space *mapping)
501 atomic_dec(&mapping->i_mmap_writable);
504 static inline int mapping_deny_writable(struct address_space *mapping)
506 return atomic_dec_unless_positive(&mapping->i_mmap_writable) ?
510 static inline void mapping_allow_writable(struct address_space *mapping)
512 atomic_inc(&mapping->i_mmap_writable);
516 * Use sequence counter to get consistent i_size on 32-bit processors.
518 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
519 #include <linux/seqlock.h>
520 #define __NEED_I_SIZE_ORDERED
521 #define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
523 #define i_size_ordered_init(inode) do { } while (0)
527 #define ACL_NOT_CACHED ((void *)(-1))
529 #define IOP_FASTPERM 0x0001
530 #define IOP_LOOKUP 0x0002
531 #define IOP_NOFOLLOW 0x0004
534 * Keep mostly read-only and often accessed (especially for
535 * the RCU path lookup and 'stat' data) fields at the beginning
536 * of the 'struct inode'
540 unsigned short i_opflags;
543 unsigned int i_flags;
545 #ifdef CONFIG_FS_POSIX_ACL
546 struct posix_acl *i_acl;
547 struct posix_acl *i_default_acl;
550 const struct inode_operations *i_op;
551 struct super_block *i_sb;
552 struct address_space *i_mapping;
554 #ifdef CONFIG_SECURITY
558 /* Stat data, not accessed from path walking */
561 * Filesystems may only read i_nlink directly. They shall use the
562 * following functions for modification:
564 * (set|clear|inc|drop)_nlink
565 * inode_(inc|dec)_link_count
568 const unsigned int i_nlink;
569 unsigned int __i_nlink;
573 struct timespec i_atime;
574 struct timespec i_mtime;
575 struct timespec i_ctime;
576 spinlock_t i_lock; /* i_blocks, i_bytes, maybe i_size */
577 unsigned short i_bytes;
578 unsigned int i_blkbits;
581 #ifdef __NEED_I_SIZE_ORDERED
582 seqcount_t i_size_seqcount;
586 unsigned long i_state;
587 struct mutex i_mutex;
589 unsigned long dirtied_when; /* jiffies of first dirtying */
591 struct hlist_node i_hash;
592 struct list_head i_wb_list; /* backing dev IO list */
593 struct list_head i_lru; /* inode LRU list */
594 struct list_head i_sb_list;
596 struct hlist_head i_dentry;
597 struct rcu_head i_rcu;
601 atomic_t i_dio_count;
602 atomic_t i_writecount;
604 atomic_t i_readcount; /* struct files open RO */
606 const struct file_operations *i_fop; /* former ->i_op->default_file_ops */
607 struct file_lock *i_flock;
608 struct address_space i_data;
610 struct dquot *i_dquot[MAXQUOTAS];
612 struct list_head i_devices;
614 struct pipe_inode_info *i_pipe;
615 struct block_device *i_bdev;
621 #ifdef CONFIG_FSNOTIFY
622 __u32 i_fsnotify_mask; /* all events this inode cares about */
623 struct hlist_head i_fsnotify_marks;
626 void *i_private; /* fs or device private pointer */
629 static inline int inode_unhashed(struct inode *inode)
631 return hlist_unhashed(&inode->i_hash);
635 * inode->i_mutex nesting subclasses for the lock validator:
637 * 0: the object of the current VFS operation
641 * 4: second non-directory
642 * The last is for certain operations (such as rename) which lock two
643 * non-directories at once.
645 * The locking order between these classes is
646 * parent -> child -> normal -> xattr -> second non-directory
648 enum inode_i_mutex_lock_class
657 void lock_two_nondirectories(struct inode *, struct inode*);
658 void unlock_two_nondirectories(struct inode *, struct inode*);
661 * NOTE: in a 32bit arch with a preemptable kernel and
662 * an UP compile the i_size_read/write must be atomic
663 * with respect to the local cpu (unlike with preempt disabled),
664 * but they don't need to be atomic with respect to other cpus like in
665 * true SMP (so they need either to either locally disable irq around
666 * the read or for example on x86 they can be still implemented as a
667 * cmpxchg8b without the need of the lock prefix). For SMP compiles
668 * and 64bit archs it makes no difference if preempt is enabled or not.
670 static inline loff_t i_size_read(const struct inode *inode)
672 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
677 seq = read_seqcount_begin(&inode->i_size_seqcount);
678 i_size = inode->i_size;
679 } while (read_seqcount_retry(&inode->i_size_seqcount, seq));
681 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
685 i_size = inode->i_size;
689 return inode->i_size;
694 * NOTE: unlike i_size_read(), i_size_write() does need locking around it
695 * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount
696 * can be lost, resulting in subsequent i_size_read() calls spinning forever.
698 static inline void i_size_write(struct inode *inode, loff_t i_size)
700 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
702 write_seqcount_begin(&inode->i_size_seqcount);
703 inode->i_size = i_size;
704 write_seqcount_end(&inode->i_size_seqcount);
706 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
708 inode->i_size = i_size;
711 inode->i_size = i_size;
715 /* Helper functions so that in most cases filesystems will
716 * not need to deal directly with kuid_t and kgid_t and can
717 * instead deal with the raw numeric values that are stored
720 static inline uid_t i_uid_read(const struct inode *inode)
722 return from_kuid(&init_user_ns, inode->i_uid);
725 static inline gid_t i_gid_read(const struct inode *inode)
727 return from_kgid(&init_user_ns, inode->i_gid);
730 static inline void i_uid_write(struct inode *inode, uid_t uid)
732 inode->i_uid = make_kuid(&init_user_ns, uid);
735 static inline void i_gid_write(struct inode *inode, gid_t gid)
737 inode->i_gid = make_kgid(&init_user_ns, gid);
740 static inline unsigned iminor(const struct inode *inode)
742 return MINOR(inode->i_rdev);
745 static inline unsigned imajor(const struct inode *inode)
747 return MAJOR(inode->i_rdev);
750 extern struct block_device *I_BDEV(struct inode *inode);
753 rwlock_t lock; /* protects pid, uid, euid fields */
754 struct pid *pid; /* pid or -pgrp where SIGIO should be sent */
755 enum pid_type pid_type; /* Kind of process group SIGIO should be sent to */
756 kuid_t uid, euid; /* uid/euid of process setting the owner */
757 int signum; /* posix.1b rt signal to be delivered on IO */
761 * Track a single file's readahead state
763 struct file_ra_state {
764 pgoff_t start; /* where readahead started */
765 unsigned int size; /* # of readahead pages */
766 unsigned int async_size; /* do asynchronous readahead when
767 there are only # of pages ahead */
769 unsigned int ra_pages; /* Maximum readahead window */
770 unsigned int mmap_miss; /* Cache miss stat for mmap accesses */
771 loff_t prev_pos; /* Cache last read() position */
775 * Check if @index falls in the readahead windows.
777 static inline int ra_has_index(struct file_ra_state *ra, pgoff_t index)
779 return (index >= ra->start &&
780 index < ra->start + ra->size);
785 struct llist_node fu_llist;
786 struct rcu_head fu_rcuhead;
789 #define f_dentry f_path.dentry
790 struct inode *f_inode; /* cached value */
791 const struct file_operations *f_op;
794 * Protects f_ep_links, f_flags.
795 * Must not be taken from IRQ context.
798 atomic_long_t f_count;
799 unsigned int f_flags;
801 struct mutex f_pos_lock;
803 struct fown_struct f_owner;
804 const struct cred *f_cred;
805 struct file_ra_state f_ra;
808 #ifdef CONFIG_SECURITY
811 /* needed for tty driver, and maybe others */
815 /* Used by fs/eventpoll.c to link all the hooks to this file */
816 struct list_head f_ep_links;
817 struct list_head f_tfile_llink;
818 #endif /* #ifdef CONFIG_EPOLL */
819 struct address_space *f_mapping;
820 } __attribute__((aligned(4))); /* lest something weird decides that 2 is OK */
825 /* file identifier */
826 unsigned char f_handle[0];
829 static inline struct file *get_file(struct file *f)
831 atomic_long_inc(&f->f_count);
834 #define fput_atomic(x) atomic_long_add_unless(&(x)->f_count, -1, 1)
835 #define file_count(x) atomic_long_read(&(x)->f_count)
837 #define MAX_NON_LFS ((1UL<<31) - 1)
839 /* Page cache limit. The filesystems should put that into their s_maxbytes
840 limits, otherwise bad things can happen in VM. */
841 #if BITS_PER_LONG==32
842 #define MAX_LFS_FILESIZE (((loff_t)PAGE_CACHE_SIZE << (BITS_PER_LONG-1))-1)
843 #elif BITS_PER_LONG==64
844 #define MAX_LFS_FILESIZE ((loff_t)0x7fffffffffffffffLL)
849 #define FL_DELEG 4 /* NFSv4 delegation */
850 #define FL_ACCESS 8 /* not trying to lock, just looking */
851 #define FL_EXISTS 16 /* when unlocking, test for existence */
852 #define FL_LEASE 32 /* lease held on this file */
853 #define FL_CLOSE 64 /* unlock on close */
854 #define FL_SLEEP 128 /* A blocking lock */
855 #define FL_DOWNGRADE_PENDING 256 /* Lease is being downgraded */
856 #define FL_UNLOCK_PENDING 512 /* Lease is being broken */
857 #define FL_OFDLCK 1024 /* lock is "owned" by struct file */
860 * Special return value from posix_lock_file() and vfs_lock_file() for
861 * asynchronous locking.
863 #define FILE_LOCK_DEFERRED 1
865 /* legacy typedef, should eventually be removed */
866 typedef void *fl_owner_t;
868 struct file_lock_operations {
869 void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
870 void (*fl_release_private)(struct file_lock *);
873 struct lock_manager_operations {
874 int (*lm_compare_owner)(struct file_lock *, struct file_lock *);
875 unsigned long (*lm_owner_key)(struct file_lock *);
876 void (*lm_get_owner)(struct file_lock *, struct file_lock *);
877 void (*lm_put_owner)(struct file_lock *);
878 void (*lm_notify)(struct file_lock *); /* unblock callback */
879 int (*lm_grant)(struct file_lock *, int);
880 bool (*lm_break)(struct file_lock *);
881 int (*lm_change)(struct file_lock **, int, struct list_head *);
882 void (*lm_setup)(struct file_lock *, void **);
885 struct lock_manager {
886 struct list_head list;
890 void locks_start_grace(struct net *, struct lock_manager *);
891 void locks_end_grace(struct lock_manager *);
892 int locks_in_grace(struct net *);
894 /* that will die - we need it for nfs_lock_info */
895 #include <linux/nfs_fs_i.h>
898 * struct file_lock represents a generic "file lock". It's used to represent
899 * POSIX byte range locks, BSD (flock) locks, and leases. It's important to
900 * note that the same struct is used to represent both a request for a lock and
901 * the lock itself, but the same object is never used for both.
903 * FIXME: should we create a separate "struct lock_request" to help distinguish
906 * The i_flock list is ordered by:
908 * 1) lock type -- FL_LEASEs first, then FL_FLOCK, and finally FL_POSIX
910 * 3) lock range start
913 * Obviously, the last two criteria only matter for POSIX locks.
916 struct file_lock *fl_next; /* singly linked list for this inode */
917 struct hlist_node fl_link; /* node in global lists */
918 struct list_head fl_block; /* circular list of blocked processes */
920 unsigned int fl_flags;
921 unsigned char fl_type;
923 int fl_link_cpu; /* what cpu's list is this on? */
924 struct pid *fl_nspid;
925 wait_queue_head_t fl_wait;
926 struct file *fl_file;
930 struct fasync_struct * fl_fasync; /* for lease break notifications */
931 /* for lease breaks: */
932 unsigned long fl_break_time;
933 unsigned long fl_downgrade_time;
935 const struct file_lock_operations *fl_ops; /* Callbacks for filesystems */
936 const struct lock_manager_operations *fl_lmops; /* Callbacks for lockmanagers */
938 struct nfs_lock_info nfs_fl;
939 struct nfs4_lock_info nfs4_fl;
941 struct list_head link; /* link in AFS vnode's pending_locks list */
942 int state; /* state of grant or error if -ve */
947 /* The following constant reflects the upper bound of the file/locking space */
949 #define INT_LIMIT(x) (~((x)1 << (sizeof(x)*8 - 1)))
950 #define OFFSET_MAX INT_LIMIT(loff_t)
951 #define OFFT_OFFSET_MAX INT_LIMIT(off_t)
954 #include <linux/fcntl.h>
956 extern void send_sigio(struct fown_struct *fown, int fd, int band);
958 #ifdef CONFIG_FILE_LOCKING
959 extern int fcntl_getlk(struct file *, unsigned int, struct flock __user *);
960 extern int fcntl_setlk(unsigned int, struct file *, unsigned int,
961 struct flock __user *);
963 #if BITS_PER_LONG == 32
964 extern int fcntl_getlk64(struct file *, unsigned int, struct flock64 __user *);
965 extern int fcntl_setlk64(unsigned int, struct file *, unsigned int,
966 struct flock64 __user *);
969 extern int fcntl_setlease(unsigned int fd, struct file *filp, long arg);
970 extern int fcntl_getlease(struct file *filp);
973 void locks_free_lock(struct file_lock *fl);
974 extern void locks_init_lock(struct file_lock *);
975 extern struct file_lock * locks_alloc_lock(void);
976 extern void locks_copy_lock(struct file_lock *, struct file_lock *);
977 extern void locks_copy_conflock(struct file_lock *, struct file_lock *);
978 extern void locks_remove_posix(struct file *, fl_owner_t);
979 extern void locks_remove_file(struct file *);
980 extern void locks_release_private(struct file_lock *);
981 extern void posix_test_lock(struct file *, struct file_lock *);
982 extern int posix_lock_file(struct file *, struct file_lock *, struct file_lock *);
983 extern int posix_lock_file_wait(struct file *, struct file_lock *);
984 extern int posix_unblock_lock(struct file_lock *);
985 extern int vfs_test_lock(struct file *, struct file_lock *);
986 extern int vfs_lock_file(struct file *, unsigned int, struct file_lock *, struct file_lock *);
987 extern int vfs_cancel_lock(struct file *filp, struct file_lock *fl);
988 extern int flock_lock_file_wait(struct file *filp, struct file_lock *fl);
989 extern int __break_lease(struct inode *inode, unsigned int flags, unsigned int type);
990 extern void lease_get_mtime(struct inode *, struct timespec *time);
991 extern int generic_setlease(struct file *, long, struct file_lock **, void **priv);
992 extern int vfs_setlease(struct file *, long, struct file_lock **, void **);
993 extern int lease_modify(struct file_lock **, int, struct list_head *);
994 #else /* !CONFIG_FILE_LOCKING */
995 static inline int fcntl_getlk(struct file *file, unsigned int cmd,
996 struct flock __user *user)
1001 static inline int fcntl_setlk(unsigned int fd, struct file *file,
1002 unsigned int cmd, struct flock __user *user)
1007 #if BITS_PER_LONG == 32
1008 static inline int fcntl_getlk64(struct file *file, unsigned int cmd,
1009 struct flock64 __user *user)
1014 static inline int fcntl_setlk64(unsigned int fd, struct file *file,
1015 unsigned int cmd, struct flock64 __user *user)
1020 static inline int fcntl_setlease(unsigned int fd, struct file *filp, long arg)
1025 static inline int fcntl_getlease(struct file *filp)
1030 static inline void locks_init_lock(struct file_lock *fl)
1035 static inline void locks_copy_conflock(struct file_lock *new, struct file_lock *fl)
1040 static inline void locks_copy_lock(struct file_lock *new, struct file_lock *fl)
1045 static inline void locks_remove_posix(struct file *filp, fl_owner_t owner)
1050 static inline void locks_remove_file(struct file *filp)
1055 static inline void posix_test_lock(struct file *filp, struct file_lock *fl)
1060 static inline int posix_lock_file(struct file *filp, struct file_lock *fl,
1061 struct file_lock *conflock)
1066 static inline int posix_lock_file_wait(struct file *filp, struct file_lock *fl)
1071 static inline int posix_unblock_lock(struct file_lock *waiter)
1076 static inline int vfs_test_lock(struct file *filp, struct file_lock *fl)
1081 static inline int vfs_lock_file(struct file *filp, unsigned int cmd,
1082 struct file_lock *fl, struct file_lock *conf)
1087 static inline int vfs_cancel_lock(struct file *filp, struct file_lock *fl)
1092 static inline int flock_lock_file_wait(struct file *filp,
1093 struct file_lock *request)
1098 static inline int __break_lease(struct inode *inode, unsigned int mode, unsigned int type)
1103 static inline void lease_get_mtime(struct inode *inode, struct timespec *time)
1108 static inline int generic_setlease(struct file *filp, long arg,
1109 struct file_lock **flp, void **priv)
1114 static inline int vfs_setlease(struct file *filp, long arg,
1115 struct file_lock **lease, void **priv)
1120 static inline int lease_modify(struct file_lock **before, int arg,
1121 struct list_head *dispose)
1125 #endif /* !CONFIG_FILE_LOCKING */
1128 struct fasync_struct {
1132 struct fasync_struct *fa_next; /* singly linked list */
1133 struct file *fa_file;
1134 struct rcu_head fa_rcu;
1137 #define FASYNC_MAGIC 0x4601
1139 /* SMP safe fasync helpers: */
1140 extern int fasync_helper(int, struct file *, int, struct fasync_struct **);
1141 extern struct fasync_struct *fasync_insert_entry(int, struct file *, struct fasync_struct **, struct fasync_struct *);
1142 extern int fasync_remove_entry(struct file *, struct fasync_struct **);
1143 extern struct fasync_struct *fasync_alloc(void);
1144 extern void fasync_free(struct fasync_struct *);
1146 /* can be called from interrupts */
1147 extern void kill_fasync(struct fasync_struct **, int, int);
1149 extern void __f_setown(struct file *filp, struct pid *, enum pid_type, int force);
1150 extern void f_setown(struct file *filp, unsigned long arg, int force);
1151 extern void f_delown(struct file *filp);
1152 extern pid_t f_getown(struct file *filp);
1153 extern int send_sigurg(struct fown_struct *fown);
1161 #define MNT_FORCE 0x00000001 /* Attempt to forcibily umount */
1162 #define MNT_DETACH 0x00000002 /* Just detach from the tree */
1163 #define MNT_EXPIRE 0x00000004 /* Mark for expiry */
1164 #define UMOUNT_NOFOLLOW 0x00000008 /* Don't follow symlink on umount */
1165 #define UMOUNT_UNUSED 0x80000000 /* Flag guaranteed to be unused */
1167 extern struct list_head super_blocks;
1168 extern spinlock_t sb_lock;
1170 /* Possible states of 'frozen' field */
1172 SB_UNFROZEN = 0, /* FS is unfrozen */
1173 SB_FREEZE_WRITE = 1, /* Writes, dir ops, ioctls frozen */
1174 SB_FREEZE_PAGEFAULT = 2, /* Page faults stopped as well */
1175 SB_FREEZE_FS = 3, /* For internal FS use (e.g. to stop
1176 * internal threads if needed) */
1177 SB_FREEZE_COMPLETE = 4, /* ->freeze_fs finished successfully */
1180 #define SB_FREEZE_LEVELS (SB_FREEZE_COMPLETE - 1)
1183 /* Counters for counting writers at each level */
1184 struct percpu_counter counter[SB_FREEZE_LEVELS];
1185 wait_queue_head_t wait; /* queue for waiting for
1186 writers / faults to finish */
1187 int frozen; /* Is sb frozen? */
1188 wait_queue_head_t wait_unfrozen; /* queue for waiting for
1190 #ifdef CONFIG_DEBUG_LOCK_ALLOC
1191 struct lockdep_map lock_map[SB_FREEZE_LEVELS];
1195 struct super_block {
1196 struct list_head s_list; /* Keep this first */
1197 dev_t s_dev; /* search index; _not_ kdev_t */
1198 unsigned char s_blocksize_bits;
1199 unsigned long s_blocksize;
1200 loff_t s_maxbytes; /* Max file size */
1201 struct file_system_type *s_type;
1202 const struct super_operations *s_op;
1203 const struct dquot_operations *dq_op;
1204 const struct quotactl_ops *s_qcop;
1205 const struct export_operations *s_export_op;
1206 unsigned long s_flags;
1207 unsigned long s_magic;
1208 struct dentry *s_root;
1209 struct rw_semaphore s_umount;
1212 #ifdef CONFIG_SECURITY
1215 const struct xattr_handler **s_xattr;
1217 struct list_head s_inodes; /* all inodes */
1218 struct hlist_bl_head s_anon; /* anonymous dentries for (nfs) exporting */
1219 struct list_head s_mounts; /* list of mounts; _not_ for fs use */
1220 struct block_device *s_bdev;
1221 struct backing_dev_info *s_bdi;
1222 struct mtd_info *s_mtd;
1223 struct hlist_node s_instances;
1224 struct quota_info s_dquot; /* Diskquota specific options */
1226 struct sb_writers s_writers;
1228 char s_id[32]; /* Informational name */
1229 u8 s_uuid[16]; /* UUID */
1231 void *s_fs_info; /* Filesystem private info */
1232 unsigned int s_max_links;
1235 /* Granularity of c/m/atime in ns.
1236 Cannot be worse than a second */
1240 * The next field is for VFS *only*. No filesystems have any business
1241 * even looking at it. You had been warned.
1243 struct mutex s_vfs_rename_mutex; /* Kludge */
1246 * Filesystem subtype. If non-empty the filesystem type field
1247 * in /proc/mounts will be "type.subtype"
1252 * Saved mount options for lazy filesystems using
1253 * generic_show_options()
1255 char __rcu *s_options;
1256 const struct dentry_operations *s_d_op; /* default d_op for dentries */
1259 * Saved pool identifier for cleancache (-1 means none)
1261 int cleancache_poolid;
1263 struct shrinker s_shrink; /* per-sb shrinker handle */
1265 /* Number of inodes with nlink == 0 but still referenced */
1266 atomic_long_t s_remove_count;
1268 /* Being remounted read-only */
1269 int s_readonly_remount;
1271 /* AIO completions deferred from interrupt context */
1272 struct workqueue_struct *s_dio_done_wq;
1273 struct hlist_head s_pins;
1276 * Keep the lru lists last in the structure so they always sit on their
1277 * own individual cachelines.
1279 struct list_lru s_dentry_lru ____cacheline_aligned_in_smp;
1280 struct list_lru s_inode_lru ____cacheline_aligned_in_smp;
1281 struct rcu_head rcu;
1284 * Indicates how deep in a filesystem stack this SB is
1289 extern struct timespec current_fs_time(struct super_block *sb);
1292 * Snapshotting support.
1295 void __sb_end_write(struct super_block *sb, int level);
1296 int __sb_start_write(struct super_block *sb, int level, bool wait);
1299 * sb_end_write - drop write access to a superblock
1300 * @sb: the super we wrote to
1302 * Decrement number of writers to the filesystem. Wake up possible waiters
1303 * wanting to freeze the filesystem.
1305 static inline void sb_end_write(struct super_block *sb)
1307 __sb_end_write(sb, SB_FREEZE_WRITE);
1311 * sb_end_pagefault - drop write access to a superblock from a page fault
1312 * @sb: the super we wrote to
1314 * Decrement number of processes handling write page fault to the filesystem.
1315 * Wake up possible waiters wanting to freeze the filesystem.
1317 static inline void sb_end_pagefault(struct super_block *sb)
1319 __sb_end_write(sb, SB_FREEZE_PAGEFAULT);
1323 * sb_end_intwrite - drop write access to a superblock for internal fs purposes
1324 * @sb: the super we wrote to
1326 * Decrement fs-internal number of writers to the filesystem. Wake up possible
1327 * waiters wanting to freeze the filesystem.
1329 static inline void sb_end_intwrite(struct super_block *sb)
1331 __sb_end_write(sb, SB_FREEZE_FS);
1335 * sb_start_write - get write access to a superblock
1336 * @sb: the super we write to
1338 * When a process wants to write data or metadata to a file system (i.e. dirty
1339 * a page or an inode), it should embed the operation in a sb_start_write() -
1340 * sb_end_write() pair to get exclusion against file system freezing. This
1341 * function increments number of writers preventing freezing. If the file
1342 * system is already frozen, the function waits until the file system is
1345 * Since freeze protection behaves as a lock, users have to preserve
1346 * ordering of freeze protection and other filesystem locks. Generally,
1347 * freeze protection should be the outermost lock. In particular, we have:
1350 * -> i_mutex (write path, truncate, directory ops, ...)
1351 * -> s_umount (freeze_super, thaw_super)
1353 static inline void sb_start_write(struct super_block *sb)
1355 __sb_start_write(sb, SB_FREEZE_WRITE, true);
1358 static inline int sb_start_write_trylock(struct super_block *sb)
1360 return __sb_start_write(sb, SB_FREEZE_WRITE, false);
1364 * sb_start_pagefault - get write access to a superblock from a page fault
1365 * @sb: the super we write to
1367 * When a process starts handling write page fault, it should embed the
1368 * operation into sb_start_pagefault() - sb_end_pagefault() pair to get
1369 * exclusion against file system freezing. This is needed since the page fault
1370 * is going to dirty a page. This function increments number of running page
1371 * faults preventing freezing. If the file system is already frozen, the
1372 * function waits until the file system is thawed.
1374 * Since page fault freeze protection behaves as a lock, users have to preserve
1375 * ordering of freeze protection and other filesystem locks. It is advised to
1376 * put sb_start_pagefault() close to mmap_sem in lock ordering. Page fault
1377 * handling code implies lock dependency:
1380 * -> sb_start_pagefault
1382 static inline void sb_start_pagefault(struct super_block *sb)
1384 __sb_start_write(sb, SB_FREEZE_PAGEFAULT, true);
1388 * sb_start_intwrite - get write access to a superblock for internal fs purposes
1389 * @sb: the super we write to
1391 * This is the third level of protection against filesystem freezing. It is
1392 * free for use by a filesystem. The only requirement is that it must rank
1393 * below sb_start_pagefault.
1395 * For example filesystem can call sb_start_intwrite() when starting a
1396 * transaction which somewhat eases handling of freezing for internal sources
1397 * of filesystem changes (internal fs threads, discarding preallocation on file
1400 static inline void sb_start_intwrite(struct super_block *sb)
1402 __sb_start_write(sb, SB_FREEZE_FS, true);
1406 extern bool inode_owner_or_capable(const struct inode *inode);
1409 * VFS helper functions..
1411 extern int vfs_create(struct inode *, struct dentry *, umode_t, bool);
1412 extern int vfs_mkdir(struct inode *, struct dentry *, umode_t);
1413 extern int vfs_mknod(struct inode *, struct dentry *, umode_t, dev_t);
1414 extern int vfs_symlink(struct inode *, struct dentry *, const char *);
1415 extern int vfs_link(struct dentry *, struct inode *, struct dentry *, struct inode **);
1416 extern int vfs_rmdir(struct inode *, struct dentry *);
1417 extern int vfs_unlink(struct inode *, struct dentry *, struct inode **);
1418 extern int vfs_rename(struct inode *, struct dentry *, struct inode *, struct dentry *, struct inode **, unsigned int);
1419 extern int vfs_whiteout(struct inode *, struct dentry *);
1422 * VFS dentry helper functions.
1424 extern void dentry_unhash(struct dentry *dentry);
1427 * VFS file helper functions.
1429 extern void inode_init_owner(struct inode *inode, const struct inode *dir,
1432 * VFS FS_IOC_FIEMAP helper definitions.
1434 struct fiemap_extent_info {
1435 unsigned int fi_flags; /* Flags as passed from user */
1436 unsigned int fi_extents_mapped; /* Number of mapped extents */
1437 unsigned int fi_extents_max; /* Size of fiemap_extent array */
1438 struct fiemap_extent __user *fi_extents_start; /* Start of
1439 fiemap_extent array */
1441 int fiemap_fill_next_extent(struct fiemap_extent_info *info, u64 logical,
1442 u64 phys, u64 len, u32 flags);
1443 int fiemap_check_flags(struct fiemap_extent_info *fieinfo, u32 fs_flags);
1448 * NOTE! These match bits 12..15 of stat.st_mode
1449 * (ie "(i_mode >> 12) & 15").
1451 #define DT_UNKNOWN 0
1462 * This is the "filldir" function type, used by readdir() to let
1463 * the kernel specify what kind of dirent layout it wants to have.
1464 * This allows the kernel to read directories into kernel space or
1465 * to have different dirent layouts depending on the binary type.
1467 typedef int (*filldir_t)(void *, const char *, int, loff_t, u64, unsigned);
1468 struct dir_context {
1469 const filldir_t actor;
1473 struct block_device_operations;
1475 /* These macros are for out of kernel modules to test that
1476 * the kernel supports the unlocked_ioctl and compat_ioctl
1477 * fields in struct file_operations. */
1478 #define HAVE_COMPAT_IOCTL 1
1479 #define HAVE_UNLOCKED_IOCTL 1
1483 struct file_operations {
1484 struct module *owner;
1485 loff_t (*llseek) (struct file *, loff_t, int);
1486 ssize_t (*read) (struct file *, char __user *, size_t, loff_t *);
1487 ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *);
1488 ssize_t (*aio_read) (struct kiocb *, const struct iovec *, unsigned long, loff_t);
1489 ssize_t (*aio_write) (struct kiocb *, const struct iovec *, unsigned long, loff_t);
1490 ssize_t (*read_iter) (struct kiocb *, struct iov_iter *);
1491 ssize_t (*write_iter) (struct kiocb *, struct iov_iter *);
1492 int (*iterate) (struct file *, struct dir_context *);
1493 unsigned int (*poll) (struct file *, struct poll_table_struct *);
1494 long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long);
1495 long (*compat_ioctl) (struct file *, unsigned int, unsigned long);
1496 int (*mmap) (struct file *, struct vm_area_struct *);
1497 int (*open) (struct inode *, struct file *);
1498 int (*flush) (struct file *, fl_owner_t id);
1499 int (*release) (struct inode *, struct file *);
1500 int (*fsync) (struct file *, loff_t, loff_t, int datasync);
1501 int (*aio_fsync) (struct kiocb *, int datasync);
1502 int (*fasync) (int, struct file *, int);
1503 int (*lock) (struct file *, int, struct file_lock *);
1504 ssize_t (*sendpage) (struct file *, struct page *, int, size_t, loff_t *, int);
1505 unsigned long (*get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
1506 int (*check_flags)(int);
1507 int (*flock) (struct file *, int, struct file_lock *);
1508 ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, loff_t *, size_t, unsigned int);
1509 ssize_t (*splice_read)(struct file *, loff_t *, struct pipe_inode_info *, size_t, unsigned int);
1510 int (*setlease)(struct file *, long, struct file_lock **, void **);
1511 long (*fallocate)(struct file *file, int mode, loff_t offset,
1513 int (*show_fdinfo)(struct seq_file *m, struct file *f);
1516 struct inode_operations {
1517 struct dentry * (*lookup) (struct inode *,struct dentry *, unsigned int);
1518 void * (*follow_link) (struct dentry *, struct nameidata *);
1519 int (*permission) (struct inode *, int);
1520 struct posix_acl * (*get_acl)(struct inode *, int);
1522 int (*readlink) (struct dentry *, char __user *,int);
1523 void (*put_link) (struct dentry *, struct nameidata *, void *);
1525 int (*create) (struct inode *,struct dentry *, umode_t, bool);
1526 int (*link) (struct dentry *,struct inode *,struct dentry *);
1527 int (*unlink) (struct inode *,struct dentry *);
1528 int (*symlink) (struct inode *,struct dentry *,const char *);
1529 int (*mkdir) (struct inode *,struct dentry *,umode_t);
1530 int (*rmdir) (struct inode *,struct dentry *);
1531 int (*mknod) (struct inode *,struct dentry *,umode_t,dev_t);
1532 int (*rename) (struct inode *, struct dentry *,
1533 struct inode *, struct dentry *);
1534 int (*rename2) (struct inode *, struct dentry *,
1535 struct inode *, struct dentry *, unsigned int);
1536 int (*setattr) (struct dentry *, struct iattr *);
1537 int (*getattr) (struct vfsmount *mnt, struct dentry *, struct kstat *);
1538 int (*setxattr) (struct dentry *, const char *,const void *,size_t,int);
1539 ssize_t (*getxattr) (struct dentry *, const char *, void *, size_t);
1540 ssize_t (*listxattr) (struct dentry *, char *, size_t);
1541 int (*removexattr) (struct dentry *, const char *);
1542 int (*fiemap)(struct inode *, struct fiemap_extent_info *, u64 start,
1544 int (*update_time)(struct inode *, struct timespec *, int);
1545 int (*atomic_open)(struct inode *, struct dentry *,
1546 struct file *, unsigned open_flag,
1547 umode_t create_mode, int *opened);
1548 int (*tmpfile) (struct inode *, struct dentry *, umode_t);
1549 int (*set_acl)(struct inode *, struct posix_acl *, int);
1551 /* WARNING: probably going away soon, do not use! */
1552 int (*dentry_open)(struct dentry *, struct file *, const struct cred *);
1553 } ____cacheline_aligned;
1555 ssize_t rw_copy_check_uvector(int type, const struct iovec __user * uvector,
1556 unsigned long nr_segs, unsigned long fast_segs,
1557 struct iovec *fast_pointer,
1558 struct iovec **ret_pointer);
1560 extern ssize_t vfs_read(struct file *, char __user *, size_t, loff_t *);
1561 extern ssize_t vfs_write(struct file *, const char __user *, size_t, loff_t *);
1562 extern ssize_t vfs_readv(struct file *, const struct iovec __user *,
1563 unsigned long, loff_t *);
1564 extern ssize_t vfs_writev(struct file *, const struct iovec __user *,
1565 unsigned long, loff_t *);
1567 struct super_operations {
1568 struct inode *(*alloc_inode)(struct super_block *sb);
1569 void (*destroy_inode)(struct inode *);
1571 void (*dirty_inode) (struct inode *, int flags);
1572 int (*write_inode) (struct inode *, struct writeback_control *wbc);
1573 int (*drop_inode) (struct inode *);
1574 void (*evict_inode) (struct inode *);
1575 void (*put_super) (struct super_block *);
1576 int (*sync_fs)(struct super_block *sb, int wait);
1577 int (*freeze_fs) (struct super_block *);
1578 int (*unfreeze_fs) (struct super_block *);
1579 int (*statfs) (struct dentry *, struct kstatfs *);
1580 int (*remount_fs) (struct super_block *, int *, char *);
1581 void (*umount_begin) (struct super_block *);
1583 int (*show_options)(struct seq_file *, struct dentry *);
1584 int (*show_devname)(struct seq_file *, struct dentry *);
1585 int (*show_path)(struct seq_file *, struct dentry *);
1586 int (*show_stats)(struct seq_file *, struct dentry *);
1588 ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t);
1589 ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t);
1591 int (*bdev_try_to_free_page)(struct super_block*, struct page*, gfp_t);
1592 long (*nr_cached_objects)(struct super_block *, int);
1593 long (*free_cached_objects)(struct super_block *, long, int);
1597 * Inode flags - they have no relation to superblock flags now
1599 #define S_SYNC 1 /* Writes are synced at once */
1600 #define S_NOATIME 2 /* Do not update access times */
1601 #define S_APPEND 4 /* Append-only file */
1602 #define S_IMMUTABLE 8 /* Immutable file */
1603 #define S_DEAD 16 /* removed, but still open directory */
1604 #define S_NOQUOTA 32 /* Inode is not counted to quota */
1605 #define S_DIRSYNC 64 /* Directory modifications are synchronous */
1606 #define S_NOCMTIME 128 /* Do not update file c/mtime */
1607 #define S_SWAPFILE 256 /* Do not truncate: swapon got its bmaps */
1608 #define S_PRIVATE 512 /* Inode is fs-internal */
1609 #define S_IMA 1024 /* Inode has an associated IMA struct */
1610 #define S_AUTOMOUNT 2048 /* Automount/referral quasi-directory */
1611 #define S_NOSEC 4096 /* no suid or xattr security attributes */
1614 * Note that nosuid etc flags are inode-specific: setting some file-system
1615 * flags just means all the inodes inherit those flags by default. It might be
1616 * possible to override it selectively if you really wanted to with some
1617 * ioctl() that is not currently implemented.
1619 * Exception: MS_RDONLY is always applied to the entire file system.
1621 * Unfortunately, it is possible to change a filesystems flags with it mounted
1622 * with files in use. This means that all of the inodes will not have their
1623 * i_flags updated. Hence, i_flags no longer inherit the superblock mount
1624 * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org
1626 #define __IS_FLG(inode, flg) ((inode)->i_sb->s_flags & (flg))
1628 #define IS_RDONLY(inode) ((inode)->i_sb->s_flags & MS_RDONLY)
1629 #define IS_SYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS) || \
1630 ((inode)->i_flags & S_SYNC))
1631 #define IS_DIRSYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS|MS_DIRSYNC) || \
1632 ((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
1633 #define IS_MANDLOCK(inode) __IS_FLG(inode, MS_MANDLOCK)
1634 #define IS_NOATIME(inode) __IS_FLG(inode, MS_RDONLY|MS_NOATIME)
1635 #define IS_I_VERSION(inode) __IS_FLG(inode, MS_I_VERSION)
1637 #define IS_NOQUOTA(inode) ((inode)->i_flags & S_NOQUOTA)
1638 #define IS_APPEND(inode) ((inode)->i_flags & S_APPEND)
1639 #define IS_IMMUTABLE(inode) ((inode)->i_flags & S_IMMUTABLE)
1640 #define IS_POSIXACL(inode) __IS_FLG(inode, MS_POSIXACL)
1642 #define IS_DEADDIR(inode) ((inode)->i_flags & S_DEAD)
1643 #define IS_NOCMTIME(inode) ((inode)->i_flags & S_NOCMTIME)
1644 #define IS_SWAPFILE(inode) ((inode)->i_flags & S_SWAPFILE)
1645 #define IS_PRIVATE(inode) ((inode)->i_flags & S_PRIVATE)
1646 #define IS_IMA(inode) ((inode)->i_flags & S_IMA)
1647 #define IS_AUTOMOUNT(inode) ((inode)->i_flags & S_AUTOMOUNT)
1648 #define IS_NOSEC(inode) ((inode)->i_flags & S_NOSEC)
1650 #define IS_WHITEOUT(inode) (S_ISCHR(inode->i_mode) && \
1651 (inode)->i_rdev == WHITEOUT_DEV)
1654 * Inode state bits. Protected by inode->i_lock
1656 * Three bits determine the dirty state of the inode, I_DIRTY_SYNC,
1657 * I_DIRTY_DATASYNC and I_DIRTY_PAGES.
1659 * Four bits define the lifetime of an inode. Initially, inodes are I_NEW,
1660 * until that flag is cleared. I_WILL_FREE, I_FREEING and I_CLEAR are set at
1661 * various stages of removing an inode.
1663 * Two bits are used for locking and completion notification, I_NEW and I_SYNC.
1665 * I_DIRTY_SYNC Inode is dirty, but doesn't have to be written on
1666 * fdatasync(). i_atime is the usual cause.
1667 * I_DIRTY_DATASYNC Data-related inode changes pending. We keep track of
1668 * these changes separately from I_DIRTY_SYNC so that we
1669 * don't have to write inode on fdatasync() when only
1670 * mtime has changed in it.
1671 * I_DIRTY_PAGES Inode has dirty pages. Inode itself may be clean.
1672 * I_NEW Serves as both a mutex and completion notification.
1673 * New inodes set I_NEW. If two processes both create
1674 * the same inode, one of them will release its inode and
1675 * wait for I_NEW to be released before returning.
1676 * Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can
1677 * also cause waiting on I_NEW, without I_NEW actually
1678 * being set. find_inode() uses this to prevent returning
1679 * nearly-dead inodes.
1680 * I_WILL_FREE Must be set when calling write_inode_now() if i_count
1681 * is zero. I_FREEING must be set when I_WILL_FREE is
1683 * I_FREEING Set when inode is about to be freed but still has dirty
1684 * pages or buffers attached or the inode itself is still
1686 * I_CLEAR Added by clear_inode(). In this state the inode is
1687 * clean and can be destroyed. Inode keeps I_FREEING.
1689 * Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are
1690 * prohibited for many purposes. iget() must wait for
1691 * the inode to be completely released, then create it
1692 * anew. Other functions will just ignore such inodes,
1693 * if appropriate. I_NEW is used for waiting.
1695 * I_SYNC Writeback of inode is running. The bit is set during
1696 * data writeback, and cleared with a wakeup on the bit
1697 * address once it is done. The bit is also used to pin
1698 * the inode in memory for flusher thread.
1700 * I_REFERENCED Marks the inode as recently references on the LRU list.
1702 * I_DIO_WAKEUP Never set. Only used as a key for wait_on_bit().
1704 * Q: What is the difference between I_WILL_FREE and I_FREEING?
1706 #define I_DIRTY_SYNC (1 << 0)
1707 #define I_DIRTY_DATASYNC (1 << 1)
1708 #define I_DIRTY_PAGES (1 << 2)
1710 #define I_NEW (1 << __I_NEW)
1711 #define I_WILL_FREE (1 << 4)
1712 #define I_FREEING (1 << 5)
1713 #define I_CLEAR (1 << 6)
1715 #define I_SYNC (1 << __I_SYNC)
1716 #define I_REFERENCED (1 << 8)
1717 #define __I_DIO_WAKEUP 9
1718 #define I_DIO_WAKEUP (1 << I_DIO_WAKEUP)
1719 #define I_LINKABLE (1 << 10)
1721 #define I_DIRTY (I_DIRTY_SYNC | I_DIRTY_DATASYNC | I_DIRTY_PAGES)
1723 extern void __mark_inode_dirty(struct inode *, int);
1724 static inline void mark_inode_dirty(struct inode *inode)
1726 __mark_inode_dirty(inode, I_DIRTY);
1729 static inline void mark_inode_dirty_sync(struct inode *inode)
1731 __mark_inode_dirty(inode, I_DIRTY_SYNC);
1734 extern void inc_nlink(struct inode *inode);
1735 extern void drop_nlink(struct inode *inode);
1736 extern void clear_nlink(struct inode *inode);
1737 extern void set_nlink(struct inode *inode, unsigned int nlink);
1739 static inline void inode_inc_link_count(struct inode *inode)
1742 mark_inode_dirty(inode);
1745 static inline void inode_dec_link_count(struct inode *inode)
1748 mark_inode_dirty(inode);
1752 * inode_inc_iversion - increments i_version
1753 * @inode: inode that need to be updated
1755 * Every time the inode is modified, the i_version field will be incremented.
1756 * The filesystem has to be mounted with i_version flag
1759 static inline void inode_inc_iversion(struct inode *inode)
1761 spin_lock(&inode->i_lock);
1763 spin_unlock(&inode->i_lock);
1766 enum file_time_flags {
1773 extern void touch_atime(const struct path *);
1774 static inline void file_accessed(struct file *file)
1776 if (!(file->f_flags & O_NOATIME))
1777 touch_atime(&file->f_path);
1780 int sync_inode(struct inode *inode, struct writeback_control *wbc);
1781 int sync_inode_metadata(struct inode *inode, int wait);
1783 struct file_system_type {
1786 #define FS_REQUIRES_DEV 1
1787 #define FS_BINARY_MOUNTDATA 2
1788 #define FS_HAS_SUBTYPE 4
1789 #define FS_USERNS_MOUNT 8 /* Can be mounted by userns root */
1790 #define FS_USERNS_DEV_MOUNT 16 /* A userns mount does not imply MNT_NODEV */
1791 #define FS_RENAME_DOES_D_MOVE 32768 /* FS will handle d_move() during rename() internally. */
1792 struct dentry *(*mount) (struct file_system_type *, int,
1793 const char *, void *);
1794 void (*kill_sb) (struct super_block *);
1795 struct module *owner;
1796 struct file_system_type * next;
1797 struct hlist_head fs_supers;
1799 struct lock_class_key s_lock_key;
1800 struct lock_class_key s_umount_key;
1801 struct lock_class_key s_vfs_rename_key;
1802 struct lock_class_key s_writers_key[SB_FREEZE_LEVELS];
1804 struct lock_class_key i_lock_key;
1805 struct lock_class_key i_mutex_key;
1806 struct lock_class_key i_mutex_dir_key;
1809 #define MODULE_ALIAS_FS(NAME) MODULE_ALIAS("fs-" NAME)
1811 extern struct dentry *mount_ns(struct file_system_type *fs_type, int flags,
1812 void *data, int (*fill_super)(struct super_block *, void *, int));
1813 extern struct dentry *mount_bdev(struct file_system_type *fs_type,
1814 int flags, const char *dev_name, void *data,
1815 int (*fill_super)(struct super_block *, void *, int));
1816 extern struct dentry *mount_single(struct file_system_type *fs_type,
1817 int flags, void *data,
1818 int (*fill_super)(struct super_block *, void *, int));
1819 extern struct dentry *mount_nodev(struct file_system_type *fs_type,
1820 int flags, void *data,
1821 int (*fill_super)(struct super_block *, void *, int));
1822 extern struct dentry *mount_subtree(struct vfsmount *mnt, const char *path);
1823 void generic_shutdown_super(struct super_block *sb);
1824 void kill_block_super(struct super_block *sb);
1825 void kill_anon_super(struct super_block *sb);
1826 void kill_litter_super(struct super_block *sb);
1827 void deactivate_super(struct super_block *sb);
1828 void deactivate_locked_super(struct super_block *sb);
1829 int set_anon_super(struct super_block *s, void *data);
1830 int get_anon_bdev(dev_t *);
1831 void free_anon_bdev(dev_t);
1832 struct super_block *sget(struct file_system_type *type,
1833 int (*test)(struct super_block *,void *),
1834 int (*set)(struct super_block *,void *),
1835 int flags, void *data);
1836 extern struct dentry *mount_pseudo(struct file_system_type *, char *,
1837 const struct super_operations *ops,
1838 const struct dentry_operations *dops,
1841 /* Alas, no aliases. Too much hassle with bringing module.h everywhere */
1842 #define fops_get(fops) \
1843 (((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
1844 #define fops_put(fops) \
1845 do { if (fops) module_put((fops)->owner); } while(0)
1847 * This one is to be used *ONLY* from ->open() instances.
1848 * fops must be non-NULL, pinned down *and* module dependencies
1849 * should be sufficient to pin the caller down as well.
1851 #define replace_fops(f, fops) \
1853 struct file *__file = (f); \
1854 fops_put(__file->f_op); \
1855 BUG_ON(!(__file->f_op = (fops))); \
1858 extern int register_filesystem(struct file_system_type *);
1859 extern int unregister_filesystem(struct file_system_type *);
1860 extern struct vfsmount *kern_mount_data(struct file_system_type *, void *data);
1861 #define kern_mount(type) kern_mount_data(type, NULL)
1862 extern void kern_unmount(struct vfsmount *mnt);
1863 extern int may_umount_tree(struct vfsmount *);
1864 extern int may_umount(struct vfsmount *);
1865 extern long do_mount(const char *, const char __user *,
1866 const char *, unsigned long, void *);
1867 extern struct vfsmount *collect_mounts(struct path *);
1868 extern void drop_collected_mounts(struct vfsmount *);
1869 extern int iterate_mounts(int (*)(struct vfsmount *, void *), void *,
1871 extern int vfs_statfs(struct path *, struct kstatfs *);
1872 extern int user_statfs(const char __user *, struct kstatfs *);
1873 extern int fd_statfs(int, struct kstatfs *);
1874 extern int vfs_ustat(dev_t, struct kstatfs *);
1875 extern int freeze_super(struct super_block *super);
1876 extern int thaw_super(struct super_block *super);
1877 extern bool our_mnt(struct vfsmount *mnt);
1878 extern bool fs_fully_visible(struct file_system_type *);
1880 extern int current_umask(void);
1882 extern void ihold(struct inode * inode);
1883 extern void iput(struct inode *);
1885 static inline struct inode *file_inode(const struct file *f)
1891 extern struct kobject *fs_kobj;
1893 #define MAX_RW_COUNT (INT_MAX & PAGE_CACHE_MASK)
1895 #define FLOCK_VERIFY_READ 1
1896 #define FLOCK_VERIFY_WRITE 2
1898 #ifdef CONFIG_FILE_LOCKING
1899 extern int locks_mandatory_locked(struct file *);
1900 extern int locks_mandatory_area(int, struct inode *, struct file *, loff_t, size_t);
1903 * Candidates for mandatory locking have the setgid bit set
1904 * but no group execute bit - an otherwise meaningless combination.
1907 static inline int __mandatory_lock(struct inode *ino)
1909 return (ino->i_mode & (S_ISGID | S_IXGRP)) == S_ISGID;
1913 * ... and these candidates should be on MS_MANDLOCK mounted fs,
1914 * otherwise these will be advisory locks
1917 static inline int mandatory_lock(struct inode *ino)
1919 return IS_MANDLOCK(ino) && __mandatory_lock(ino);
1922 static inline int locks_verify_locked(struct file *file)
1924 if (mandatory_lock(file_inode(file)))
1925 return locks_mandatory_locked(file);
1929 static inline int locks_verify_truncate(struct inode *inode,
1933 if (inode->i_flock && mandatory_lock(inode))
1934 return locks_mandatory_area(
1935 FLOCK_VERIFY_WRITE, inode, filp,
1936 size < inode->i_size ? size : inode->i_size,
1937 (size < inode->i_size ? inode->i_size - size
1938 : size - inode->i_size)
1943 static inline int break_lease(struct inode *inode, unsigned int mode)
1946 * Since this check is lockless, we must ensure that any refcounts
1947 * taken are done before checking inode->i_flock. Otherwise, we could
1948 * end up racing with tasks trying to set a new lease on this file.
1952 return __break_lease(inode, mode, FL_LEASE);
1956 static inline int break_deleg(struct inode *inode, unsigned int mode)
1959 * Since this check is lockless, we must ensure that any refcounts
1960 * taken are done before checking inode->i_flock. Otherwise, we could
1961 * end up racing with tasks trying to set a new lease on this file.
1965 return __break_lease(inode, mode, FL_DELEG);
1969 static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode)
1973 ret = break_deleg(inode, O_WRONLY|O_NONBLOCK);
1974 if (ret == -EWOULDBLOCK && delegated_inode) {
1975 *delegated_inode = inode;
1981 static inline int break_deleg_wait(struct inode **delegated_inode)
1985 ret = break_deleg(*delegated_inode, O_WRONLY);
1986 iput(*delegated_inode);
1987 *delegated_inode = NULL;
1991 #else /* !CONFIG_FILE_LOCKING */
1992 static inline int locks_mandatory_locked(struct file *file)
1997 static inline int locks_mandatory_area(int rw, struct inode *inode,
1998 struct file *filp, loff_t offset,
2004 static inline int __mandatory_lock(struct inode *inode)
2009 static inline int mandatory_lock(struct inode *inode)
2014 static inline int locks_verify_locked(struct file *file)
2019 static inline int locks_verify_truncate(struct inode *inode, struct file *filp,
2025 static inline int break_lease(struct inode *inode, unsigned int mode)
2030 static inline int break_deleg(struct inode *inode, unsigned int mode)
2035 static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode)
2040 static inline int break_deleg_wait(struct inode **delegated_inode)
2046 #endif /* CONFIG_FILE_LOCKING */
2051 const char *name; /* pointer to actual string */
2052 const __user char *uptr; /* original userland pointer */
2053 struct audit_names *aname;
2054 bool separate; /* should "name" be freed? */
2057 extern long vfs_truncate(struct path *, loff_t);
2058 extern int do_truncate(struct dentry *, loff_t start, unsigned int time_attrs,
2060 extern int do_fallocate(struct file *file, int mode, loff_t offset,
2062 extern long do_sys_open(int dfd, const char __user *filename, int flags,
2064 extern struct file *file_open_name(struct filename *, int, umode_t);
2065 extern struct file *filp_open(const char *, int, umode_t);
2066 extern struct file *file_open_root(struct dentry *, struct vfsmount *,
2068 extern int vfs_open(const struct path *, struct file *, const struct cred *);
2069 extern struct file * dentry_open(const struct path *, int, const struct cred *);
2070 extern int filp_close(struct file *, fl_owner_t id);
2072 extern struct filename *getname(const char __user *);
2073 extern struct filename *getname_kernel(const char *);
2079 extern int finish_open(struct file *file, struct dentry *dentry,
2080 int (*open)(struct inode *, struct file *),
2082 extern int finish_no_open(struct file *file, struct dentry *dentry);
2086 extern int ioctl_preallocate(struct file *filp, void __user *argp);
2089 extern void __init vfs_caches_init_early(void);
2090 extern void __init vfs_caches_init(unsigned long);
2092 extern struct kmem_cache *names_cachep;
2094 extern void final_putname(struct filename *name);
2096 #define __getname() kmem_cache_alloc(names_cachep, GFP_KERNEL)
2097 #define __putname(name) kmem_cache_free(names_cachep, (void *)(name))
2098 #ifndef CONFIG_AUDITSYSCALL
2099 #define putname(name) final_putname(name)
2101 extern void putname(struct filename *name);
2105 extern int register_blkdev(unsigned int, const char *);
2106 extern void unregister_blkdev(unsigned int, const char *);
2107 extern struct block_device *bdget(dev_t);
2108 extern struct block_device *bdgrab(struct block_device *bdev);
2109 extern void bd_set_size(struct block_device *, loff_t size);
2110 extern void bd_forget(struct inode *inode);
2111 extern void bdput(struct block_device *);
2112 extern void invalidate_bdev(struct block_device *);
2113 extern void iterate_bdevs(void (*)(struct block_device *, void *), void *);
2114 extern int sync_blockdev(struct block_device *bdev);
2115 extern void kill_bdev(struct block_device *);
2116 extern struct super_block *freeze_bdev(struct block_device *);
2117 extern void emergency_thaw_all(void);
2118 extern int thaw_bdev(struct block_device *bdev, struct super_block *sb);
2119 extern int fsync_bdev(struct block_device *);
2120 extern int sb_is_blkdev_sb(struct super_block *sb);
2122 static inline void bd_forget(struct inode *inode) {}
2123 static inline int sync_blockdev(struct block_device *bdev) { return 0; }
2124 static inline void kill_bdev(struct block_device *bdev) {}
2125 static inline void invalidate_bdev(struct block_device *bdev) {}
2127 static inline struct super_block *freeze_bdev(struct block_device *sb)
2132 static inline int thaw_bdev(struct block_device *bdev, struct super_block *sb)
2137 static inline void iterate_bdevs(void (*f)(struct block_device *, void *), void *arg)
2141 static inline int sb_is_blkdev_sb(struct super_block *sb)
2146 extern int sync_filesystem(struct super_block *);
2147 extern const struct file_operations def_blk_fops;
2148 extern const struct file_operations def_chr_fops;
2149 extern const struct file_operations bad_sock_fops;
2151 extern int ioctl_by_bdev(struct block_device *, unsigned, unsigned long);
2152 extern int blkdev_ioctl(struct block_device *, fmode_t, unsigned, unsigned long);
2153 extern long compat_blkdev_ioctl(struct file *, unsigned, unsigned long);
2154 extern int blkdev_get(struct block_device *bdev, fmode_t mode, void *holder);
2155 extern struct block_device *blkdev_get_by_path(const char *path, fmode_t mode,
2157 extern struct block_device *blkdev_get_by_dev(dev_t dev, fmode_t mode,
2159 extern void blkdev_put(struct block_device *bdev, fmode_t mode);
2161 extern int bd_link_disk_holder(struct block_device *bdev, struct gendisk *disk);
2162 extern void bd_unlink_disk_holder(struct block_device *bdev,
2163 struct gendisk *disk);
2165 static inline int bd_link_disk_holder(struct block_device *bdev,
2166 struct gendisk *disk)
2170 static inline void bd_unlink_disk_holder(struct block_device *bdev,
2171 struct gendisk *disk)
2178 #define CHRDEV_MAJOR_HASH_SIZE 255
2179 extern int alloc_chrdev_region(dev_t *, unsigned, unsigned, const char *);
2180 extern int register_chrdev_region(dev_t, unsigned, const char *);
2181 extern int __register_chrdev(unsigned int major, unsigned int baseminor,
2182 unsigned int count, const char *name,
2183 const struct file_operations *fops);
2184 extern void __unregister_chrdev(unsigned int major, unsigned int baseminor,
2185 unsigned int count, const char *name);
2186 extern void unregister_chrdev_region(dev_t, unsigned);
2187 extern void chrdev_show(struct seq_file *,off_t);
2189 static inline int register_chrdev(unsigned int major, const char *name,
2190 const struct file_operations *fops)
2192 return __register_chrdev(major, 0, 256, name, fops);
2195 static inline void unregister_chrdev(unsigned int major, const char *name)
2197 __unregister_chrdev(major, 0, 256, name);
2200 /* fs/block_dev.c */
2201 #define BDEVNAME_SIZE 32 /* Largest string for a blockdev identifier */
2202 #define BDEVT_SIZE 10 /* Largest string for MAJ:MIN for blkdev */
2205 #define BLKDEV_MAJOR_HASH_SIZE 255
2206 extern const char *__bdevname(dev_t, char *buffer);
2207 extern const char *bdevname(struct block_device *bdev, char *buffer);
2208 extern struct block_device *lookup_bdev(const char *);
2209 extern void blkdev_show(struct seq_file *,off_t);
2212 #define BLKDEV_MAJOR_HASH_SIZE 0
2215 extern void init_special_inode(struct inode *, umode_t, dev_t);
2217 /* Invalid inode operations -- fs/bad_inode.c */
2218 extern void make_bad_inode(struct inode *);
2219 extern int is_bad_inode(struct inode *);
2223 * return READ, READA, or WRITE
2225 #define bio_rw(bio) ((bio)->bi_rw & (RW_MASK | RWA_MASK))
2228 * return data direction, READ or WRITE
2230 #define bio_data_dir(bio) ((bio)->bi_rw & 1)
2232 extern void check_disk_size_change(struct gendisk *disk,
2233 struct block_device *bdev);
2234 extern int revalidate_disk(struct gendisk *);
2235 extern int check_disk_change(struct block_device *);
2236 extern int __invalidate_device(struct block_device *, bool);
2237 extern int invalidate_partition(struct gendisk *, int);
2239 unsigned long invalidate_mapping_pages(struct address_space *mapping,
2240 pgoff_t start, pgoff_t end);
2242 static inline void invalidate_remote_inode(struct inode *inode)
2244 if (S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) ||
2245 S_ISLNK(inode->i_mode))
2246 invalidate_mapping_pages(inode->i_mapping, 0, -1);
2248 extern int invalidate_inode_pages2(struct address_space *mapping);
2249 extern int invalidate_inode_pages2_range(struct address_space *mapping,
2250 pgoff_t start, pgoff_t end);
2251 extern int write_inode_now(struct inode *, int);
2252 extern int filemap_fdatawrite(struct address_space *);
2253 extern int filemap_flush(struct address_space *);
2254 extern int filemap_fdatawait(struct address_space *);
2255 extern int filemap_fdatawait_range(struct address_space *, loff_t lstart,
2257 extern int filemap_write_and_wait(struct address_space *mapping);
2258 extern int filemap_write_and_wait_range(struct address_space *mapping,
2259 loff_t lstart, loff_t lend);
2260 extern int __filemap_fdatawrite_range(struct address_space *mapping,
2261 loff_t start, loff_t end, int sync_mode);
2262 extern int filemap_fdatawrite_range(struct address_space *mapping,
2263 loff_t start, loff_t end);
2265 extern int vfs_fsync_range(struct file *file, loff_t start, loff_t end,
2267 extern int vfs_fsync(struct file *file, int datasync);
2268 static inline int generic_write_sync(struct file *file, loff_t pos, loff_t count)
2270 if (!(file->f_flags & O_DSYNC) && !IS_SYNC(file->f_mapping->host))
2272 return vfs_fsync_range(file, pos, pos + count - 1,
2273 (file->f_flags & __O_SYNC) ? 0 : 1);
2275 extern void emergency_sync(void);
2276 extern void emergency_remount(void);
2278 extern sector_t bmap(struct inode *, sector_t);
2280 extern int notify_change(struct dentry *, struct iattr *, struct inode **);
2281 extern int inode_permission(struct inode *, int);
2282 extern int __inode_permission(struct inode *, int);
2283 extern int generic_permission(struct inode *, int);
2284 extern int __check_sticky(struct inode *dir, struct inode *inode);
2286 static inline bool execute_ok(struct inode *inode)
2288 return (inode->i_mode & S_IXUGO) || S_ISDIR(inode->i_mode);
2291 static inline void file_start_write(struct file *file)
2293 if (!S_ISREG(file_inode(file)->i_mode))
2295 __sb_start_write(file_inode(file)->i_sb, SB_FREEZE_WRITE, true);
2298 static inline bool file_start_write_trylock(struct file *file)
2300 if (!S_ISREG(file_inode(file)->i_mode))
2302 return __sb_start_write(file_inode(file)->i_sb, SB_FREEZE_WRITE, false);
2305 static inline void file_end_write(struct file *file)
2307 if (!S_ISREG(file_inode(file)->i_mode))
2309 __sb_end_write(file_inode(file)->i_sb, SB_FREEZE_WRITE);
2313 * get_write_access() gets write permission for a file.
2314 * put_write_access() releases this write permission.
2315 * This is used for regular files.
2316 * We cannot support write (and maybe mmap read-write shared) accesses and
2317 * MAP_DENYWRITE mmappings simultaneously. The i_writecount field of an inode
2318 * can have the following values:
2319 * 0: no writers, no VM_DENYWRITE mappings
2320 * < 0: (-i_writecount) vm_area_structs with VM_DENYWRITE set exist
2321 * > 0: (i_writecount) users are writing to the file.
2323 * Normally we operate on that counter with atomic_{inc,dec} and it's safe
2324 * except for the cases where we don't hold i_writecount yet. Then we need to
2325 * use {get,deny}_write_access() - these functions check the sign and refuse
2326 * to do the change if sign is wrong.
2328 static inline int get_write_access(struct inode *inode)
2330 return atomic_inc_unless_negative(&inode->i_writecount) ? 0 : -ETXTBSY;
2332 static inline int deny_write_access(struct file *file)
2334 struct inode *inode = file_inode(file);
2335 return atomic_dec_unless_positive(&inode->i_writecount) ? 0 : -ETXTBSY;
2337 static inline void put_write_access(struct inode * inode)
2339 atomic_dec(&inode->i_writecount);
2341 static inline void allow_write_access(struct file *file)
2344 atomic_inc(&file_inode(file)->i_writecount);
2346 static inline bool inode_is_open_for_write(const struct inode *inode)
2348 return atomic_read(&inode->i_writecount) > 0;
2352 static inline void i_readcount_dec(struct inode *inode)
2354 BUG_ON(!atomic_read(&inode->i_readcount));
2355 atomic_dec(&inode->i_readcount);
2357 static inline void i_readcount_inc(struct inode *inode)
2359 atomic_inc(&inode->i_readcount);
2362 static inline void i_readcount_dec(struct inode *inode)
2366 static inline void i_readcount_inc(struct inode *inode)
2371 extern int do_pipe_flags(int *, int);
2373 extern int kernel_read(struct file *, loff_t, char *, unsigned long);
2374 extern ssize_t kernel_write(struct file *, const char *, size_t, loff_t);
2375 extern ssize_t __kernel_write(struct file *, const char *, size_t, loff_t *);
2376 extern struct file * open_exec(const char *);
2378 /* fs/dcache.c -- generic fs support functions */
2379 extern int is_subdir(struct dentry *, struct dentry *);
2380 extern int path_is_under(struct path *, struct path *);
2382 #include <linux/err.h>
2384 /* needed for stackable file system support */
2385 extern loff_t default_llseek(struct file *file, loff_t offset, int whence);
2387 extern loff_t vfs_llseek(struct file *file, loff_t offset, int whence);
2389 extern int inode_init_always(struct super_block *, struct inode *);
2390 extern void inode_init_once(struct inode *);
2391 extern void address_space_init_once(struct address_space *mapping);
2392 extern struct inode * igrab(struct inode *);
2393 extern ino_t iunique(struct super_block *, ino_t);
2394 extern int inode_needs_sync(struct inode *inode);
2395 extern int generic_delete_inode(struct inode *inode);
2396 static inline int generic_drop_inode(struct inode *inode)
2398 return !inode->i_nlink || inode_unhashed(inode);
2401 extern struct inode *ilookup5_nowait(struct super_block *sb,
2402 unsigned long hashval, int (*test)(struct inode *, void *),
2404 extern struct inode *ilookup5(struct super_block *sb, unsigned long hashval,
2405 int (*test)(struct inode *, void *), void *data);
2406 extern struct inode *ilookup(struct super_block *sb, unsigned long ino);
2408 extern struct inode * iget5_locked(struct super_block *, unsigned long, int (*test)(struct inode *, void *), int (*set)(struct inode *, void *), void *);
2409 extern struct inode * iget_locked(struct super_block *, unsigned long);
2410 extern int insert_inode_locked4(struct inode *, unsigned long, int (*test)(struct inode *, void *), void *);
2411 extern int insert_inode_locked(struct inode *);
2412 #ifdef CONFIG_DEBUG_LOCK_ALLOC
2413 extern void lockdep_annotate_inode_mutex_key(struct inode *inode);
2415 static inline void lockdep_annotate_inode_mutex_key(struct inode *inode) { };
2417 extern void unlock_new_inode(struct inode *);
2418 extern unsigned int get_next_ino(void);
2420 extern void __iget(struct inode * inode);
2421 extern void iget_failed(struct inode *);
2422 extern void clear_inode(struct inode *);
2423 extern void __destroy_inode(struct inode *);
2424 extern struct inode *new_inode_pseudo(struct super_block *sb);
2425 extern struct inode *new_inode(struct super_block *sb);
2426 extern void free_inode_nonrcu(struct inode *inode);
2427 extern int should_remove_suid(struct dentry *);
2428 extern int file_remove_suid(struct file *);
2430 extern void __insert_inode_hash(struct inode *, unsigned long hashval);
2431 static inline void insert_inode_hash(struct inode *inode)
2433 __insert_inode_hash(inode, inode->i_ino);
2436 extern void __remove_inode_hash(struct inode *);
2437 static inline void remove_inode_hash(struct inode *inode)
2439 if (!inode_unhashed(inode))
2440 __remove_inode_hash(inode);
2443 extern void inode_sb_list_add(struct inode *inode);
2446 extern void submit_bio(int, struct bio *);
2447 extern int bdev_read_only(struct block_device *);
2449 extern int set_blocksize(struct block_device *, int);
2450 extern int sb_set_blocksize(struct super_block *, int);
2451 extern int sb_min_blocksize(struct super_block *, int);
2453 extern int generic_file_mmap(struct file *, struct vm_area_struct *);
2454 extern int generic_file_readonly_mmap(struct file *, struct vm_area_struct *);
2455 extern int generic_file_remap_pages(struct vm_area_struct *, unsigned long addr,
2456 unsigned long size, pgoff_t pgoff);
2457 int generic_write_checks(struct file *file, loff_t *pos, size_t *count, int isblk);
2458 extern ssize_t generic_file_read_iter(struct kiocb *, struct iov_iter *);
2459 extern ssize_t __generic_file_write_iter(struct kiocb *, struct iov_iter *);
2460 extern ssize_t generic_file_write_iter(struct kiocb *, struct iov_iter *);
2461 extern ssize_t generic_file_direct_write(struct kiocb *, struct iov_iter *, loff_t);
2462 extern ssize_t generic_perform_write(struct file *, struct iov_iter *, loff_t);
2463 extern ssize_t do_sync_read(struct file *filp, char __user *buf, size_t len, loff_t *ppos);
2464 extern ssize_t do_sync_write(struct file *filp, const char __user *buf, size_t len, loff_t *ppos);
2465 extern ssize_t new_sync_read(struct file *filp, char __user *buf, size_t len, loff_t *ppos);
2466 extern ssize_t new_sync_write(struct file *filp, const char __user *buf, size_t len, loff_t *ppos);
2468 /* fs/block_dev.c */
2469 extern ssize_t blkdev_write_iter(struct kiocb *iocb, struct iov_iter *from);
2470 extern int blkdev_fsync(struct file *filp, loff_t start, loff_t end,
2472 extern void block_sync_page(struct page *page);
2475 extern ssize_t generic_file_splice_read(struct file *, loff_t *,
2476 struct pipe_inode_info *, size_t, unsigned int);
2477 extern ssize_t default_file_splice_read(struct file *, loff_t *,
2478 struct pipe_inode_info *, size_t, unsigned int);
2479 extern ssize_t iter_file_splice_write(struct pipe_inode_info *,
2480 struct file *, loff_t *, size_t, unsigned int);
2481 extern ssize_t generic_splice_sendpage(struct pipe_inode_info *pipe,
2482 struct file *out, loff_t *, size_t len, unsigned int flags);
2483 extern long do_splice_direct(struct file *in, loff_t *ppos, struct file *out,
2484 loff_t *opos, size_t len, unsigned int flags);
2488 file_ra_state_init(struct file_ra_state *ra, struct address_space *mapping);
2489 extern loff_t noop_llseek(struct file *file, loff_t offset, int whence);
2490 extern loff_t no_llseek(struct file *file, loff_t offset, int whence);
2491 extern loff_t vfs_setpos(struct file *file, loff_t offset, loff_t maxsize);
2492 extern loff_t generic_file_llseek(struct file *file, loff_t offset, int whence);
2493 extern loff_t generic_file_llseek_size(struct file *file, loff_t offset,
2494 int whence, loff_t maxsize, loff_t eof);
2495 extern loff_t fixed_size_llseek(struct file *file, loff_t offset,
2496 int whence, loff_t size);
2497 extern int generic_file_open(struct inode * inode, struct file * filp);
2498 extern int nonseekable_open(struct inode * inode, struct file * filp);
2500 #ifdef CONFIG_FS_XIP
2501 extern ssize_t xip_file_read(struct file *filp, char __user *buf, size_t len,
2503 extern int xip_file_mmap(struct file * file, struct vm_area_struct * vma);
2504 extern ssize_t xip_file_write(struct file *filp, const char __user *buf,
2505 size_t len, loff_t *ppos);
2506 extern int xip_truncate_page(struct address_space *mapping, loff_t from);
2508 static inline int xip_truncate_page(struct address_space *mapping, loff_t from)
2515 typedef void (dio_submit_t)(int rw, struct bio *bio, struct inode *inode,
2516 loff_t file_offset);
2519 /* need locking between buffered and direct access */
2522 /* filesystem does not support filling holes */
2523 DIO_SKIP_HOLES = 0x02,
2525 /* filesystem can handle aio writes beyond i_size */
2526 DIO_ASYNC_EXTEND = 0x04,
2529 void dio_end_io(struct bio *bio, int error);
2531 ssize_t __blockdev_direct_IO(int rw, struct kiocb *iocb, struct inode *inode,
2532 struct block_device *bdev, struct iov_iter *iter, loff_t offset,
2533 get_block_t get_block, dio_iodone_t end_io,
2534 dio_submit_t submit_io, int flags);
2536 static inline ssize_t blockdev_direct_IO(int rw, struct kiocb *iocb,
2537 struct inode *inode, struct iov_iter *iter, loff_t offset,
2538 get_block_t get_block)
2540 return __blockdev_direct_IO(rw, iocb, inode, inode->i_sb->s_bdev, iter,
2541 offset, get_block, NULL, NULL,
2542 DIO_LOCKING | DIO_SKIP_HOLES);
2546 void inode_dio_wait(struct inode *inode);
2547 void inode_dio_done(struct inode *inode);
2549 extern void inode_set_flags(struct inode *inode, unsigned int flags,
2552 extern const struct file_operations generic_ro_fops;
2554 #define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
2556 extern int readlink_copy(char __user *, int, const char *);
2557 extern int page_readlink(struct dentry *, char __user *, int);
2558 extern void *page_follow_link_light(struct dentry *, struct nameidata *);
2559 extern void page_put_link(struct dentry *, struct nameidata *, void *);
2560 extern int __page_symlink(struct inode *inode, const char *symname, int len,
2562 extern int page_symlink(struct inode *inode, const char *symname, int len);
2563 extern const struct inode_operations page_symlink_inode_operations;
2564 extern void kfree_put_link(struct dentry *, struct nameidata *, void *);
2565 extern int generic_readlink(struct dentry *, char __user *, int);
2566 extern void generic_fillattr(struct inode *, struct kstat *);
2567 int vfs_getattr_nosec(struct path *path, struct kstat *stat);
2568 extern int vfs_getattr(struct path *, struct kstat *);
2569 void __inode_add_bytes(struct inode *inode, loff_t bytes);
2570 void inode_add_bytes(struct inode *inode, loff_t bytes);
2571 void __inode_sub_bytes(struct inode *inode, loff_t bytes);
2572 void inode_sub_bytes(struct inode *inode, loff_t bytes);
2573 loff_t inode_get_bytes(struct inode *inode);
2574 void inode_set_bytes(struct inode *inode, loff_t bytes);
2576 extern int vfs_readdir(struct file *, filldir_t, void *);
2577 extern int iterate_dir(struct file *, struct dir_context *);
2579 extern int vfs_stat(const char __user *, struct kstat *);
2580 extern int vfs_lstat(const char __user *, struct kstat *);
2581 extern int vfs_fstat(unsigned int, struct kstat *);
2582 extern int vfs_fstatat(int , const char __user *, struct kstat *, int);
2584 extern int do_vfs_ioctl(struct file *filp, unsigned int fd, unsigned int cmd,
2586 extern int __generic_block_fiemap(struct inode *inode,
2587 struct fiemap_extent_info *fieinfo,
2588 loff_t start, loff_t len,
2589 get_block_t *get_block);
2590 extern int generic_block_fiemap(struct inode *inode,
2591 struct fiemap_extent_info *fieinfo, u64 start,
2592 u64 len, get_block_t *get_block);
2594 extern void get_filesystem(struct file_system_type *fs);
2595 extern void put_filesystem(struct file_system_type *fs);
2596 extern struct file_system_type *get_fs_type(const char *name);
2597 extern struct super_block *get_super(struct block_device *);
2598 extern struct super_block *get_super_thawed(struct block_device *);
2599 extern struct super_block *get_active_super(struct block_device *bdev);
2600 extern void drop_super(struct super_block *sb);
2601 extern void iterate_supers(void (*)(struct super_block *, void *), void *);
2602 extern void iterate_supers_type(struct file_system_type *,
2603 void (*)(struct super_block *, void *), void *);
2605 extern int dcache_dir_open(struct inode *, struct file *);
2606 extern int dcache_dir_close(struct inode *, struct file *);
2607 extern loff_t dcache_dir_lseek(struct file *, loff_t, int);
2608 extern int dcache_readdir(struct file *, struct dir_context *);
2609 extern int simple_setattr(struct dentry *, struct iattr *);
2610 extern int simple_getattr(struct vfsmount *, struct dentry *, struct kstat *);
2611 extern int simple_statfs(struct dentry *, struct kstatfs *);
2612 extern int simple_open(struct inode *inode, struct file *file);
2613 extern int simple_link(struct dentry *, struct inode *, struct dentry *);
2614 extern int simple_unlink(struct inode *, struct dentry *);
2615 extern int simple_rmdir(struct inode *, struct dentry *);
2616 extern int simple_rename(struct inode *, struct dentry *, struct inode *, struct dentry *);
2617 extern int noop_fsync(struct file *, loff_t, loff_t, int);
2618 extern int simple_empty(struct dentry *);
2619 extern int simple_readpage(struct file *file, struct page *page);
2620 extern int simple_write_begin(struct file *file, struct address_space *mapping,
2621 loff_t pos, unsigned len, unsigned flags,
2622 struct page **pagep, void **fsdata);
2623 extern int simple_write_end(struct file *file, struct address_space *mapping,
2624 loff_t pos, unsigned len, unsigned copied,
2625 struct page *page, void *fsdata);
2626 extern int always_delete_dentry(const struct dentry *);
2627 extern struct inode *alloc_anon_inode(struct super_block *);
2628 extern int simple_nosetlease(struct file *, long, struct file_lock **, void **);
2629 extern const struct dentry_operations simple_dentry_operations;
2631 extern struct dentry *simple_lookup(struct inode *, struct dentry *, unsigned int flags);
2632 extern ssize_t generic_read_dir(struct file *, char __user *, size_t, loff_t *);
2633 extern const struct file_operations simple_dir_operations;
2634 extern const struct inode_operations simple_dir_inode_operations;
2635 struct tree_descr { char *name; const struct file_operations *ops; int mode; };
2636 struct dentry *d_alloc_name(struct dentry *, const char *);
2637 extern int simple_fill_super(struct super_block *, unsigned long, struct tree_descr *);
2638 extern int simple_pin_fs(struct file_system_type *, struct vfsmount **mount, int *count);
2639 extern void simple_release_fs(struct vfsmount **mount, int *count);
2641 extern ssize_t simple_read_from_buffer(void __user *to, size_t count,
2642 loff_t *ppos, const void *from, size_t available);
2643 extern ssize_t simple_write_to_buffer(void *to, size_t available, loff_t *ppos,
2644 const void __user *from, size_t count);
2646 extern int __generic_file_fsync(struct file *, loff_t, loff_t, int);
2647 extern int generic_file_fsync(struct file *, loff_t, loff_t, int);
2649 extern int generic_check_addressable(unsigned, u64);
2651 #ifdef CONFIG_MIGRATION
2652 extern int buffer_migrate_page(struct address_space *,
2653 struct page *, struct page *,
2656 #define buffer_migrate_page NULL
2659 extern int inode_change_ok(const struct inode *, struct iattr *);
2660 extern int inode_newsize_ok(const struct inode *, loff_t offset);
2661 extern void setattr_copy(struct inode *inode, const struct iattr *attr);
2663 extern int file_update_time(struct file *file);
2665 extern int generic_show_options(struct seq_file *m, struct dentry *root);
2666 extern void save_mount_options(struct super_block *sb, char *options);
2667 extern void replace_mount_options(struct super_block *sb, char *options);
2669 static inline ino_t parent_ino(struct dentry *dentry)
2674 * Don't strictly need d_lock here? If the parent ino could change
2675 * then surely we'd have a deeper race in the caller?
2677 spin_lock(&dentry->d_lock);
2678 res = dentry->d_parent->d_inode->i_ino;
2679 spin_unlock(&dentry->d_lock);
2683 /* Transaction based IO helpers */
2686 * An argresp is stored in an allocated page and holds the
2687 * size of the argument or response, along with its content
2689 struct simple_transaction_argresp {
2694 #define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
2696 char *simple_transaction_get(struct file *file, const char __user *buf,
2698 ssize_t simple_transaction_read(struct file *file, char __user *buf,
2699 size_t size, loff_t *pos);
2700 int simple_transaction_release(struct inode *inode, struct file *file);
2702 void simple_transaction_set(struct file *file, size_t n);
2705 * simple attribute files
2707 * These attributes behave similar to those in sysfs:
2709 * Writing to an attribute immediately sets a value, an open file can be
2710 * written to multiple times.
2712 * Reading from an attribute creates a buffer from the value that might get
2713 * read with multiple read calls. When the attribute has been read
2714 * completely, no further read calls are possible until the file is opened
2717 * All attributes contain a text representation of a numeric value
2718 * that are accessed with the get() and set() functions.
2720 #define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt) \
2721 static int __fops ## _open(struct inode *inode, struct file *file) \
2723 __simple_attr_check_format(__fmt, 0ull); \
2724 return simple_attr_open(inode, file, __get, __set, __fmt); \
2726 static const struct file_operations __fops = { \
2727 .owner = THIS_MODULE, \
2728 .open = __fops ## _open, \
2729 .release = simple_attr_release, \
2730 .read = simple_attr_read, \
2731 .write = simple_attr_write, \
2732 .llseek = generic_file_llseek, \
2735 static inline __printf(1, 2)
2736 void __simple_attr_check_format(const char *fmt, ...)
2738 /* don't do anything, just let the compiler check the arguments; */
2741 int simple_attr_open(struct inode *inode, struct file *file,
2742 int (*get)(void *, u64 *), int (*set)(void *, u64),
2744 int simple_attr_release(struct inode *inode, struct file *file);
2745 ssize_t simple_attr_read(struct file *file, char __user *buf,
2746 size_t len, loff_t *ppos);
2747 ssize_t simple_attr_write(struct file *file, const char __user *buf,
2748 size_t len, loff_t *ppos);
2751 int proc_nr_files(struct ctl_table *table, int write,
2752 void __user *buffer, size_t *lenp, loff_t *ppos);
2753 int proc_nr_dentry(struct ctl_table *table, int write,
2754 void __user *buffer, size_t *lenp, loff_t *ppos);
2755 int proc_nr_inodes(struct ctl_table *table, int write,
2756 void __user *buffer, size_t *lenp, loff_t *ppos);
2757 int __init get_filesystem_list(char *buf);
2759 #define __FMODE_EXEC ((__force int) FMODE_EXEC)
2760 #define __FMODE_NONOTIFY ((__force int) FMODE_NONOTIFY)
2762 #define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE])
2763 #define OPEN_FMODE(flag) ((__force fmode_t)(((flag + 1) & O_ACCMODE) | \
2764 (flag & __FMODE_NONOTIFY)))
2766 static inline int is_sxid(umode_t mode)
2768 return (mode & S_ISUID) || ((mode & S_ISGID) && (mode & S_IXGRP));
2771 static inline int check_sticky(struct inode *dir, struct inode *inode)
2773 if (!(dir->i_mode & S_ISVTX))
2776 return __check_sticky(dir, inode);
2779 static inline void inode_has_no_xattr(struct inode *inode)
2781 if (!is_sxid(inode->i_mode) && (inode->i_sb->s_flags & MS_NOSEC))
2782 inode->i_flags |= S_NOSEC;
2785 static inline bool dir_emit(struct dir_context *ctx,
2786 const char *name, int namelen,
2787 u64 ino, unsigned type)
2789 return ctx->actor(ctx, name, namelen, ctx->pos, ino, type) == 0;
2791 static inline bool dir_emit_dot(struct file *file, struct dir_context *ctx)
2793 return ctx->actor(ctx, ".", 1, ctx->pos,
2794 file->f_path.dentry->d_inode->i_ino, DT_DIR) == 0;
2796 static inline bool dir_emit_dotdot(struct file *file, struct dir_context *ctx)
2798 return ctx->actor(ctx, "..", 2, ctx->pos,
2799 parent_ino(file->f_path.dentry), DT_DIR) == 0;
2801 static inline bool dir_emit_dots(struct file *file, struct dir_context *ctx)
2803 if (ctx->pos == 0) {
2804 if (!dir_emit_dot(file, ctx))
2808 if (ctx->pos == 1) {
2809 if (!dir_emit_dotdot(file, ctx))
2815 static inline bool dir_relax(struct inode *inode)
2817 mutex_unlock(&inode->i_mutex);
2818 mutex_lock(&inode->i_mutex);
2819 return !IS_DEADDIR(inode);
2822 #endif /* _LINUX_FS_H */