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/radix-tree.h>
14 #include <linux/rbtree.h>
15 #include <linux/init.h>
16 #include <linux/pid.h>
17 #include <linux/bug.h>
18 #include <linux/mutex.h>
19 #include <linux/capability.h>
20 #include <linux/semaphore.h>
21 #include <linux/fiemap.h>
22 #include <linux/rculist_bl.h>
23 #include <linux/atomic.h>
24 #include <linux/shrinker.h>
25 #include <linux/migrate_mode.h>
26 #include <linux/uidgid.h>
27 #include <linux/lockdep.h>
28 #include <linux/percpu-rwsem.h>
29 #include <linux/blk_types.h>
31 #include <asm/byteorder.h>
32 #include <uapi/linux/fs.h>
34 struct export_operations;
40 struct pipe_inode_info;
41 struct poll_table_struct;
43 struct vm_area_struct;
46 struct swap_info_struct;
49 extern void __init inode_init(void);
50 extern void __init inode_init_early(void);
51 extern void __init files_init(unsigned long);
53 extern struct files_stat_struct files_stat;
54 extern unsigned long get_max_files(void);
55 extern int sysctl_nr_open;
56 extern struct inodes_stat_t inodes_stat;
57 extern int leases_enable, lease_break_time;
58 extern int sysctl_protected_symlinks;
59 extern int sysctl_protected_hardlinks;
62 typedef int (get_block_t)(struct inode *inode, sector_t iblock,
63 struct buffer_head *bh_result, int create);
64 typedef void (dio_iodone_t)(struct kiocb *iocb, loff_t offset,
65 ssize_t bytes, void *private, int ret,
68 #define MAY_EXEC 0x00000001
69 #define MAY_WRITE 0x00000002
70 #define MAY_READ 0x00000004
71 #define MAY_APPEND 0x00000008
72 #define MAY_ACCESS 0x00000010
73 #define MAY_OPEN 0x00000020
74 #define MAY_CHDIR 0x00000040
75 /* called from RCU mode, don't block */
76 #define MAY_NOT_BLOCK 0x00000080
79 * flags in file.f_mode. Note that FMODE_READ and FMODE_WRITE must correspond
80 * to O_WRONLY and O_RDWR via the strange trick in __dentry_open()
83 /* file is open for reading */
84 #define FMODE_READ ((__force fmode_t)0x1)
85 /* file is open for writing */
86 #define FMODE_WRITE ((__force fmode_t)0x2)
87 /* file is seekable */
88 #define FMODE_LSEEK ((__force fmode_t)0x4)
89 /* file can be accessed using pread */
90 #define FMODE_PREAD ((__force fmode_t)0x8)
91 /* file can be accessed using pwrite */
92 #define FMODE_PWRITE ((__force fmode_t)0x10)
93 /* File is opened for execution with sys_execve / sys_uselib */
94 #define FMODE_EXEC ((__force fmode_t)0x20)
95 /* File is opened with O_NDELAY (only set for block devices) */
96 #define FMODE_NDELAY ((__force fmode_t)0x40)
97 /* File is opened with O_EXCL (only set for block devices) */
98 #define FMODE_EXCL ((__force fmode_t)0x80)
99 /* File is opened using open(.., 3, ..) and is writeable only for ioctls
100 (specialy hack for floppy.c) */
101 #define FMODE_WRITE_IOCTL ((__force fmode_t)0x100)
102 /* 32bit hashes as llseek() offset (for directories) */
103 #define FMODE_32BITHASH ((__force fmode_t)0x200)
104 /* 64bit hashes as llseek() offset (for directories) */
105 #define FMODE_64BITHASH ((__force fmode_t)0x400)
108 * Don't update ctime and mtime.
110 * Currently a special hack for the XFS open_by_handle ioctl, but we'll
111 * hopefully graduate it to a proper O_CMTIME flag supported by open(2) soon.
113 #define FMODE_NOCMTIME ((__force fmode_t)0x800)
115 /* Expect random access pattern */
116 #define FMODE_RANDOM ((__force fmode_t)0x1000)
118 /* File is huge (eg. /dev/kmem): treat loff_t as unsigned */
119 #define FMODE_UNSIGNED_OFFSET ((__force fmode_t)0x2000)
121 /* File is opened with O_PATH; almost nothing can be done with it */
122 #define FMODE_PATH ((__force fmode_t)0x4000)
124 /* File was opened by fanotify and shouldn't generate fanotify events */
125 #define FMODE_NONOTIFY ((__force fmode_t)0x1000000)
128 * Flag for rw_copy_check_uvector and compat_rw_copy_check_uvector
129 * that indicates that they should check the contents of the iovec are
130 * valid, but not check the memory that the iovec elements
133 #define CHECK_IOVEC_ONLY -1
136 * The below are the various read and write types that we support. Some of
137 * them include behavioral modifiers that send information down to the
138 * block layer and IO scheduler. Terminology:
140 * The block layer uses device plugging to defer IO a little bit, in
141 * the hope that we will see more IO very shortly. This increases
142 * coalescing of adjacent IO and thus reduces the number of IOs we
143 * have to send to the device. It also allows for better queuing,
144 * if the IO isn't mergeable. If the caller is going to be waiting
145 * for the IO, then he must ensure that the device is unplugged so
146 * that the IO is dispatched to the driver.
148 * All IO is handled async in Linux. This is fine for background
149 * writes, but for reads or writes that someone waits for completion
150 * on, we want to notify the block layer and IO scheduler so that they
151 * know about it. That allows them to make better scheduling
152 * decisions. So when the below references 'sync' and 'async', it
153 * is referencing this priority hint.
155 * With that in mind, the available types are:
157 * READ A normal read operation. Device will be plugged.
158 * READ_SYNC A synchronous read. Device is not plugged, caller can
159 * immediately wait on this read without caring about
161 * READA Used for read-ahead operations. Lower priority, and the
162 * block layer could (in theory) choose to ignore this
163 * request if it runs into resource problems.
164 * WRITE A normal async write. Device will be plugged.
165 * WRITE_SYNC Synchronous write. Identical to WRITE, but passes down
166 * the hint that someone will be waiting on this IO
167 * shortly. The write equivalent of READ_SYNC.
168 * WRITE_ODIRECT Special case write for O_DIRECT only.
169 * WRITE_FLUSH Like WRITE_SYNC but with preceding cache flush.
170 * WRITE_FUA Like WRITE_SYNC but data is guaranteed to be on
171 * non-volatile media on completion.
172 * WRITE_FLUSH_FUA Combination of WRITE_FLUSH and FUA. The IO is preceded
173 * by a cache flush and data is guaranteed to be on
174 * non-volatile media on completion.
177 #define RW_MASK REQ_WRITE
178 #define RWA_MASK REQ_RAHEAD
181 #define WRITE RW_MASK
182 #define READA RWA_MASK
183 #define KERNEL_READ (READ|REQ_KERNEL)
184 #define KERNEL_WRITE (WRITE|REQ_KERNEL)
186 #define READ_SYNC (READ | REQ_SYNC)
187 #define WRITE_SYNC (WRITE | REQ_SYNC | REQ_NOIDLE)
188 #define WRITE_ODIRECT (WRITE | REQ_SYNC)
189 #define WRITE_FLUSH (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FLUSH)
190 #define WRITE_FUA (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FUA)
191 #define WRITE_FLUSH_FUA (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FLUSH | REQ_FUA)
194 * Attribute flags. These should be or-ed together to figure out what
197 #define ATTR_MODE (1 << 0)
198 #define ATTR_UID (1 << 1)
199 #define ATTR_GID (1 << 2)
200 #define ATTR_SIZE (1 << 3)
201 #define ATTR_ATIME (1 << 4)
202 #define ATTR_MTIME (1 << 5)
203 #define ATTR_CTIME (1 << 6)
204 #define ATTR_ATIME_SET (1 << 7)
205 #define ATTR_MTIME_SET (1 << 8)
206 #define ATTR_FORCE (1 << 9) /* Not a change, but a change it */
207 #define ATTR_ATTR_FLAG (1 << 10)
208 #define ATTR_KILL_SUID (1 << 11)
209 #define ATTR_KILL_SGID (1 << 12)
210 #define ATTR_FILE (1 << 13)
211 #define ATTR_KILL_PRIV (1 << 14)
212 #define ATTR_OPEN (1 << 15) /* Truncating from open(O_TRUNC) */
213 #define ATTR_TIMES_SET (1 << 16)
216 * This is the Inode Attributes structure, used for notify_change(). It
217 * uses the above definitions as flags, to know which values have changed.
218 * Also, in this manner, a Filesystem can look at only the values it cares
219 * about. Basically, these are the attributes that the VFS layer can
220 * request to change from the FS layer.
222 * Derek Atkins <warlord@MIT.EDU> 94-10-20
225 unsigned int ia_valid;
230 struct timespec ia_atime;
231 struct timespec ia_mtime;
232 struct timespec ia_ctime;
235 * Not an attribute, but an auxiliary info for filesystems wanting to
236 * implement an ftruncate() like method. NOTE: filesystem should
237 * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL).
239 struct file *ia_file;
243 * Includes for diskquotas.
245 #include <linux/quota.h>
248 * enum positive_aop_returns - aop return codes with specific semantics
250 * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has
251 * completed, that the page is still locked, and
252 * should be considered active. The VM uses this hint
253 * to return the page to the active list -- it won't
254 * be a candidate for writeback again in the near
255 * future. Other callers must be careful to unlock
256 * the page if they get this return. Returned by
259 * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has
260 * unlocked it and the page might have been truncated.
261 * The caller should back up to acquiring a new page and
262 * trying again. The aop will be taking reasonable
263 * precautions not to livelock. If the caller held a page
264 * reference, it should drop it before retrying. Returned
267 * address_space_operation functions return these large constants to indicate
268 * special semantics to the caller. These are much larger than the bytes in a
269 * page to allow for functions that return the number of bytes operated on in a
273 enum positive_aop_returns {
274 AOP_WRITEPAGE_ACTIVATE = 0x80000,
275 AOP_TRUNCATED_PAGE = 0x80001,
278 #define AOP_FLAG_UNINTERRUPTIBLE 0x0001 /* will not do a short write */
279 #define AOP_FLAG_CONT_EXPAND 0x0002 /* called from cont_expand */
280 #define AOP_FLAG_NOFS 0x0004 /* used by filesystem to direct
281 * helper code (eg buffer layer)
282 * to clear GFP_FS from alloc */
285 * oh the beauties of C type declarations.
288 struct address_space;
289 struct writeback_control;
292 const struct iovec *iov;
293 unsigned long nr_segs;
298 size_t iov_iter_copy_from_user_atomic(struct page *page,
299 struct iov_iter *i, unsigned long offset, size_t bytes);
300 size_t iov_iter_copy_from_user(struct page *page,
301 struct iov_iter *i, unsigned long offset, size_t bytes);
302 void iov_iter_advance(struct iov_iter *i, size_t bytes);
303 int iov_iter_fault_in_readable(struct iov_iter *i, size_t bytes);
304 size_t iov_iter_single_seg_count(struct iov_iter *i);
306 static inline void iov_iter_init(struct iov_iter *i,
307 const struct iovec *iov, unsigned long nr_segs,
308 size_t count, size_t written)
311 i->nr_segs = nr_segs;
313 i->count = count + written;
315 iov_iter_advance(i, written);
318 static inline size_t iov_iter_count(struct iov_iter *i)
324 * "descriptor" for what we're up to with a read.
325 * This allows us to use the same read code yet
326 * have multiple different users of the data that
327 * we read from a file.
329 * The simplest case just copies the data to user
342 typedef int (*read_actor_t)(read_descriptor_t *, struct page *,
343 unsigned long, unsigned long);
345 struct address_space_operations {
346 int (*writepage)(struct page *page, struct writeback_control *wbc);
347 int (*readpage)(struct file *, struct page *);
349 /* Write back some dirty pages from this mapping. */
350 int (*writepages)(struct address_space *, struct writeback_control *);
352 /* Set a page dirty. Return true if this dirtied it */
353 int (*set_page_dirty)(struct page *page);
355 int (*readpages)(struct file *filp, struct address_space *mapping,
356 struct list_head *pages, unsigned nr_pages);
358 int (*write_begin)(struct file *, struct address_space *mapping,
359 loff_t pos, unsigned len, unsigned flags,
360 struct page **pagep, void **fsdata);
361 int (*write_end)(struct file *, struct address_space *mapping,
362 loff_t pos, unsigned len, unsigned copied,
363 struct page *page, void *fsdata);
365 /* Unfortunately this kludge is needed for FIBMAP. Don't use it */
366 sector_t (*bmap)(struct address_space *, sector_t);
367 void (*invalidatepage) (struct page *, unsigned long);
368 int (*releasepage) (struct page *, gfp_t);
369 void (*freepage)(struct page *);
370 ssize_t (*direct_IO)(int, struct kiocb *, const struct iovec *iov,
371 loff_t offset, unsigned long nr_segs);
372 int (*get_xip_mem)(struct address_space *, pgoff_t, int,
373 void **, unsigned long *);
375 * migrate the contents of a page to the specified target. If sync
376 * is false, it must not block.
378 int (*migratepage) (struct address_space *,
379 struct page *, struct page *, enum migrate_mode);
380 int (*launder_page) (struct page *);
381 int (*is_partially_uptodate) (struct page *, read_descriptor_t *,
383 int (*error_remove_page)(struct address_space *, struct page *);
385 /* swapfile support */
386 int (*swap_activate)(struct swap_info_struct *sis, struct file *file,
388 void (*swap_deactivate)(struct file *file);
391 extern const struct address_space_operations empty_aops;
394 * pagecache_write_begin/pagecache_write_end must be used by general code
395 * to write into the pagecache.
397 int pagecache_write_begin(struct file *, struct address_space *mapping,
398 loff_t pos, unsigned len, unsigned flags,
399 struct page **pagep, void **fsdata);
401 int pagecache_write_end(struct file *, struct address_space *mapping,
402 loff_t pos, unsigned len, unsigned copied,
403 struct page *page, void *fsdata);
405 struct backing_dev_info;
406 struct address_space {
407 struct inode *host; /* owner: inode, block_device */
408 struct radix_tree_root page_tree; /* radix tree of all pages */
409 spinlock_t tree_lock; /* and lock protecting it */
410 unsigned int i_mmap_writable;/* count VM_SHARED mappings */
411 struct rb_root i_mmap; /* tree of private and shared mappings */
412 struct list_head i_mmap_nonlinear;/*list VM_NONLINEAR mappings */
413 struct mutex i_mmap_mutex; /* protect tree, count, list */
414 /* Protected by tree_lock together with the radix tree */
415 unsigned long nrpages; /* number of total pages */
416 pgoff_t writeback_index;/* writeback starts here */
417 const struct address_space_operations *a_ops; /* methods */
418 unsigned long flags; /* error bits/gfp mask */
419 struct backing_dev_info *backing_dev_info; /* device readahead, etc */
420 spinlock_t private_lock; /* for use by the address_space */
421 struct list_head private_list; /* ditto */
422 void *private_data; /* ditto */
423 } __attribute__((aligned(sizeof(long))));
425 * On most architectures that alignment is already the case; but
426 * must be enforced here for CRIS, to let the least significant bit
427 * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
429 struct request_queue;
431 struct block_device {
432 dev_t bd_dev; /* not a kdev_t - it's a search key */
434 struct inode * bd_inode; /* will die */
435 struct super_block * bd_super;
436 struct mutex bd_mutex; /* open/close mutex */
437 struct list_head bd_inodes;
441 bool bd_write_holder;
443 struct list_head bd_holder_disks;
445 struct block_device * bd_contains;
446 unsigned bd_block_size;
447 struct hd_struct * bd_part;
448 /* number of times partitions within this device have been opened. */
449 unsigned bd_part_count;
451 struct gendisk * bd_disk;
452 struct request_queue * bd_queue;
453 struct list_head bd_list;
455 * Private data. You must have bd_claim'ed the block_device
456 * to use this. NOTE: bd_claim allows an owner to claim
457 * the same device multiple times, the owner must take special
458 * care to not mess up bd_private for that case.
460 unsigned long bd_private;
462 /* The counter of freeze processes */
463 int bd_fsfreeze_count;
464 /* Mutex for freeze */
465 struct mutex bd_fsfreeze_mutex;
469 * Radix-tree tags, for tagging dirty and writeback pages within the pagecache
472 #define PAGECACHE_TAG_DIRTY 0
473 #define PAGECACHE_TAG_WRITEBACK 1
474 #define PAGECACHE_TAG_TOWRITE 2
476 int mapping_tagged(struct address_space *mapping, int tag);
479 * Might pages of this file be mapped into userspace?
481 static inline int mapping_mapped(struct address_space *mapping)
483 return !RB_EMPTY_ROOT(&mapping->i_mmap) ||
484 !list_empty(&mapping->i_mmap_nonlinear);
488 * Might pages of this file have been modified in userspace?
489 * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap_pgoff
490 * marks vma as VM_SHARED if it is shared, and the file was opened for
491 * writing i.e. vma may be mprotected writable even if now readonly.
493 static inline int mapping_writably_mapped(struct address_space *mapping)
495 return mapping->i_mmap_writable != 0;
499 * Use sequence counter to get consistent i_size on 32-bit processors.
501 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
502 #include <linux/seqlock.h>
503 #define __NEED_I_SIZE_ORDERED
504 #define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
506 #define i_size_ordered_init(inode) do { } while (0)
510 #define ACL_NOT_CACHED ((void *)(-1))
512 #define IOP_FASTPERM 0x0001
513 #define IOP_LOOKUP 0x0002
514 #define IOP_NOFOLLOW 0x0004
517 * Keep mostly read-only and often accessed (especially for
518 * the RCU path lookup and 'stat' data) fields at the beginning
519 * of the 'struct inode'
523 unsigned short i_opflags;
526 unsigned int i_flags;
528 #ifdef CONFIG_FS_POSIX_ACL
529 struct posix_acl *i_acl;
530 struct posix_acl *i_default_acl;
533 const struct inode_operations *i_op;
534 struct super_block *i_sb;
535 struct address_space *i_mapping;
537 #ifdef CONFIG_SECURITY
541 /* Stat data, not accessed from path walking */
544 * Filesystems may only read i_nlink directly. They shall use the
545 * following functions for modification:
547 * (set|clear|inc|drop)_nlink
548 * inode_(inc|dec)_link_count
551 const unsigned int i_nlink;
552 unsigned int __i_nlink;
556 struct timespec i_atime;
557 struct timespec i_mtime;
558 struct timespec i_ctime;
559 spinlock_t i_lock; /* i_blocks, i_bytes, maybe i_size */
560 unsigned short i_bytes;
561 unsigned int i_blkbits;
564 #ifdef __NEED_I_SIZE_ORDERED
565 seqcount_t i_size_seqcount;
569 unsigned long i_state;
570 struct mutex i_mutex;
572 unsigned long dirtied_when; /* jiffies of first dirtying */
574 struct hlist_node i_hash;
575 struct list_head i_wb_list; /* backing dev IO list */
576 struct list_head i_lru; /* inode LRU list */
577 struct list_head i_sb_list;
579 struct hlist_head i_dentry;
580 struct rcu_head i_rcu;
584 atomic_t i_dio_count;
585 atomic_t i_writecount;
586 const struct file_operations *i_fop; /* former ->i_op->default_file_ops */
587 struct file_lock *i_flock;
588 struct address_space i_data;
590 struct dquot *i_dquot[MAXQUOTAS];
592 struct list_head i_devices;
594 struct pipe_inode_info *i_pipe;
595 struct block_device *i_bdev;
601 #ifdef CONFIG_FSNOTIFY
602 __u32 i_fsnotify_mask; /* all events this inode cares about */
603 struct hlist_head i_fsnotify_marks;
607 atomic_t i_readcount; /* struct files open RO */
609 void *i_private; /* fs or device private pointer */
612 static inline int inode_unhashed(struct inode *inode)
614 return hlist_unhashed(&inode->i_hash);
618 * inode->i_mutex nesting subclasses for the lock validator:
620 * 0: the object of the current VFS operation
625 * The locking order between these classes is
626 * parent -> child -> normal -> xattr -> quota
628 enum inode_i_mutex_lock_class
638 * NOTE: in a 32bit arch with a preemptable kernel and
639 * an UP compile the i_size_read/write must be atomic
640 * with respect to the local cpu (unlike with preempt disabled),
641 * but they don't need to be atomic with respect to other cpus like in
642 * true SMP (so they need either to either locally disable irq around
643 * the read or for example on x86 they can be still implemented as a
644 * cmpxchg8b without the need of the lock prefix). For SMP compiles
645 * and 64bit archs it makes no difference if preempt is enabled or not.
647 static inline loff_t i_size_read(const struct inode *inode)
649 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
654 seq = read_seqcount_begin(&inode->i_size_seqcount);
655 i_size = inode->i_size;
656 } while (read_seqcount_retry(&inode->i_size_seqcount, seq));
658 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
662 i_size = inode->i_size;
666 return inode->i_size;
671 * NOTE: unlike i_size_read(), i_size_write() does need locking around it
672 * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount
673 * can be lost, resulting in subsequent i_size_read() calls spinning forever.
675 static inline void i_size_write(struct inode *inode, loff_t i_size)
677 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
678 write_seqcount_begin(&inode->i_size_seqcount);
679 inode->i_size = i_size;
680 write_seqcount_end(&inode->i_size_seqcount);
681 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
683 inode->i_size = i_size;
686 inode->i_size = i_size;
690 /* Helper functions so that in most cases filesystems will
691 * not need to deal directly with kuid_t and kgid_t and can
692 * instead deal with the raw numeric values that are stored
695 static inline uid_t i_uid_read(const struct inode *inode)
697 return from_kuid(&init_user_ns, inode->i_uid);
700 static inline gid_t i_gid_read(const struct inode *inode)
702 return from_kgid(&init_user_ns, inode->i_gid);
705 static inline void i_uid_write(struct inode *inode, uid_t uid)
707 inode->i_uid = make_kuid(&init_user_ns, uid);
710 static inline void i_gid_write(struct inode *inode, gid_t gid)
712 inode->i_gid = make_kgid(&init_user_ns, gid);
715 static inline unsigned iminor(const struct inode *inode)
717 return MINOR(inode->i_rdev);
720 static inline unsigned imajor(const struct inode *inode)
722 return MAJOR(inode->i_rdev);
725 extern struct block_device *I_BDEV(struct inode *inode);
728 rwlock_t lock; /* protects pid, uid, euid fields */
729 struct pid *pid; /* pid or -pgrp where SIGIO should be sent */
730 enum pid_type pid_type; /* Kind of process group SIGIO should be sent to */
731 kuid_t uid, euid; /* uid/euid of process setting the owner */
732 int signum; /* posix.1b rt signal to be delivered on IO */
736 * Track a single file's readahead state
738 struct file_ra_state {
739 pgoff_t start; /* where readahead started */
740 unsigned int size; /* # of readahead pages */
741 unsigned int async_size; /* do asynchronous readahead when
742 there are only # of pages ahead */
744 unsigned int ra_pages; /* Maximum readahead window */
745 unsigned int mmap_miss; /* Cache miss stat for mmap accesses */
746 loff_t prev_pos; /* Cache last read() position */
750 * Check if @index falls in the readahead windows.
752 static inline int ra_has_index(struct file_ra_state *ra, pgoff_t index)
754 return (index >= ra->start &&
755 index < ra->start + ra->size);
758 #define FILE_MNT_WRITE_TAKEN 1
759 #define FILE_MNT_WRITE_RELEASED 2
763 * fu_list becomes invalid after file_free is called and queued via
764 * fu_rcuhead for RCU freeing
767 struct list_head fu_list;
768 struct rcu_head fu_rcuhead;
771 #define f_dentry f_path.dentry
772 const struct file_operations *f_op;
775 * Protects f_ep_links, f_flags, f_pos vs i_size in lseek SEEK_CUR.
776 * Must not be taken from IRQ context.
782 atomic_long_t f_count;
783 unsigned int f_flags;
786 struct fown_struct f_owner;
787 const struct cred *f_cred;
788 struct file_ra_state f_ra;
791 #ifdef CONFIG_SECURITY
794 /* needed for tty driver, and maybe others */
798 /* Used by fs/eventpoll.c to link all the hooks to this file */
799 struct list_head f_ep_links;
800 struct list_head f_tfile_llink;
801 #endif /* #ifdef CONFIG_EPOLL */
802 struct address_space *f_mapping;
803 #ifdef CONFIG_DEBUG_WRITECOUNT
804 unsigned long f_mnt_write_state;
811 /* file identifier */
812 unsigned char f_handle[0];
815 static inline struct file *get_file(struct file *f)
817 atomic_long_inc(&f->f_count);
820 #define fput_atomic(x) atomic_long_add_unless(&(x)->f_count, -1, 1)
821 #define file_count(x) atomic_long_read(&(x)->f_count)
823 #ifdef CONFIG_DEBUG_WRITECOUNT
824 static inline void file_take_write(struct file *f)
826 WARN_ON(f->f_mnt_write_state != 0);
827 f->f_mnt_write_state = FILE_MNT_WRITE_TAKEN;
829 static inline void file_release_write(struct file *f)
831 f->f_mnt_write_state |= FILE_MNT_WRITE_RELEASED;
833 static inline void file_reset_write(struct file *f)
835 f->f_mnt_write_state = 0;
837 static inline void file_check_state(struct file *f)
840 * At this point, either both or neither of these bits
843 WARN_ON(f->f_mnt_write_state == FILE_MNT_WRITE_TAKEN);
844 WARN_ON(f->f_mnt_write_state == FILE_MNT_WRITE_RELEASED);
846 static inline int file_check_writeable(struct file *f)
848 if (f->f_mnt_write_state == FILE_MNT_WRITE_TAKEN)
850 printk(KERN_WARNING "writeable file with no "
851 "mnt_want_write()\n");
855 #else /* !CONFIG_DEBUG_WRITECOUNT */
856 static inline void file_take_write(struct file *filp) {}
857 static inline void file_release_write(struct file *filp) {}
858 static inline void file_reset_write(struct file *filp) {}
859 static inline void file_check_state(struct file *filp) {}
860 static inline int file_check_writeable(struct file *filp)
864 #endif /* CONFIG_DEBUG_WRITECOUNT */
866 #define MAX_NON_LFS ((1UL<<31) - 1)
868 /* Page cache limit. The filesystems should put that into their s_maxbytes
869 limits, otherwise bad things can happen in VM. */
870 #if BITS_PER_LONG==32
871 #define MAX_LFS_FILESIZE (((loff_t)PAGE_CACHE_SIZE << (BITS_PER_LONG-1))-1)
872 #elif BITS_PER_LONG==64
873 #define MAX_LFS_FILESIZE ((loff_t)0x7fffffffffffffffLL)
878 #define FL_ACCESS 8 /* not trying to lock, just looking */
879 #define FL_EXISTS 16 /* when unlocking, test for existence */
880 #define FL_LEASE 32 /* lease held on this file */
881 #define FL_CLOSE 64 /* unlock on close */
882 #define FL_SLEEP 128 /* A blocking lock */
883 #define FL_DOWNGRADE_PENDING 256 /* Lease is being downgraded */
884 #define FL_UNLOCK_PENDING 512 /* Lease is being broken */
887 * Special return value from posix_lock_file() and vfs_lock_file() for
888 * asynchronous locking.
890 #define FILE_LOCK_DEFERRED 1
893 * The POSIX file lock owner is determined by
894 * the "struct files_struct" in the thread group
895 * (or NULL for no owner - BSD locks).
897 * Lockd stuffs a "host" pointer into this.
899 typedef struct files_struct *fl_owner_t;
901 struct file_lock_operations {
902 void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
903 void (*fl_release_private)(struct file_lock *);
906 struct lock_manager_operations {
907 int (*lm_compare_owner)(struct file_lock *, struct file_lock *);
908 void (*lm_notify)(struct file_lock *); /* unblock callback */
909 int (*lm_grant)(struct file_lock *, struct file_lock *, int);
910 void (*lm_break)(struct file_lock *);
911 int (*lm_change)(struct file_lock **, int);
914 struct lock_manager {
915 struct list_head list;
919 void locks_start_grace(struct net *, struct lock_manager *);
920 void locks_end_grace(struct lock_manager *);
921 int locks_in_grace(struct net *);
923 /* that will die - we need it for nfs_lock_info */
924 #include <linux/nfs_fs_i.h>
927 struct file_lock *fl_next; /* singly linked list for this inode */
928 struct list_head fl_link; /* doubly linked list of all locks */
929 struct list_head fl_block; /* circular list of blocked processes */
931 unsigned int fl_flags;
932 unsigned char fl_type;
934 struct pid *fl_nspid;
935 wait_queue_head_t fl_wait;
936 struct file *fl_file;
940 struct fasync_struct * fl_fasync; /* for lease break notifications */
941 /* for lease breaks: */
942 unsigned long fl_break_time;
943 unsigned long fl_downgrade_time;
945 const struct file_lock_operations *fl_ops; /* Callbacks for filesystems */
946 const struct lock_manager_operations *fl_lmops; /* Callbacks for lockmanagers */
948 struct nfs_lock_info nfs_fl;
949 struct nfs4_lock_info nfs4_fl;
951 struct list_head link; /* link in AFS vnode's pending_locks list */
952 int state; /* state of grant or error if -ve */
957 /* The following constant reflects the upper bound of the file/locking space */
959 #define INT_LIMIT(x) (~((x)1 << (sizeof(x)*8 - 1)))
960 #define OFFSET_MAX INT_LIMIT(loff_t)
961 #define OFFT_OFFSET_MAX INT_LIMIT(off_t)
964 #include <linux/fcntl.h>
966 extern void send_sigio(struct fown_struct *fown, int fd, int band);
968 #ifdef CONFIG_FILE_LOCKING
969 extern int fcntl_getlk(struct file *, struct flock __user *);
970 extern int fcntl_setlk(unsigned int, struct file *, unsigned int,
971 struct flock __user *);
973 #if BITS_PER_LONG == 32
974 extern int fcntl_getlk64(struct file *, struct flock64 __user *);
975 extern int fcntl_setlk64(unsigned int, struct file *, unsigned int,
976 struct flock64 __user *);
979 extern int fcntl_setlease(unsigned int fd, struct file *filp, long arg);
980 extern int fcntl_getlease(struct file *filp);
983 void locks_free_lock(struct file_lock *fl);
984 extern void locks_init_lock(struct file_lock *);
985 extern struct file_lock * locks_alloc_lock(void);
986 extern void locks_copy_lock(struct file_lock *, struct file_lock *);
987 extern void __locks_copy_lock(struct file_lock *, const struct file_lock *);
988 extern void locks_remove_posix(struct file *, fl_owner_t);
989 extern void locks_remove_flock(struct file *);
990 extern void locks_release_private(struct file_lock *);
991 extern void posix_test_lock(struct file *, struct file_lock *);
992 extern int posix_lock_file(struct file *, struct file_lock *, struct file_lock *);
993 extern int posix_lock_file_wait(struct file *, struct file_lock *);
994 extern int posix_unblock_lock(struct file *, struct file_lock *);
995 extern int vfs_test_lock(struct file *, struct file_lock *);
996 extern int vfs_lock_file(struct file *, unsigned int, struct file_lock *, struct file_lock *);
997 extern int vfs_cancel_lock(struct file *filp, struct file_lock *fl);
998 extern int flock_lock_file_wait(struct file *filp, struct file_lock *fl);
999 extern int __break_lease(struct inode *inode, unsigned int flags);
1000 extern void lease_get_mtime(struct inode *, struct timespec *time);
1001 extern int generic_setlease(struct file *, long, struct file_lock **);
1002 extern int vfs_setlease(struct file *, long, struct file_lock **);
1003 extern int lease_modify(struct file_lock **, int);
1004 extern int lock_may_read(struct inode *, loff_t start, unsigned long count);
1005 extern int lock_may_write(struct inode *, loff_t start, unsigned long count);
1006 extern void locks_delete_block(struct file_lock *waiter);
1007 extern void lock_flocks(void);
1008 extern void unlock_flocks(void);
1009 #else /* !CONFIG_FILE_LOCKING */
1010 static inline int fcntl_getlk(struct file *file, struct flock __user *user)
1015 static inline int fcntl_setlk(unsigned int fd, struct file *file,
1016 unsigned int cmd, struct flock __user *user)
1021 #if BITS_PER_LONG == 32
1022 static inline int fcntl_getlk64(struct file *file, struct flock64 __user *user)
1027 static inline int fcntl_setlk64(unsigned int fd, struct file *file,
1028 unsigned int cmd, struct flock64 __user *user)
1033 static inline int fcntl_setlease(unsigned int fd, struct file *filp, long arg)
1038 static inline int fcntl_getlease(struct file *filp)
1043 static inline void locks_init_lock(struct file_lock *fl)
1048 static inline void __locks_copy_lock(struct file_lock *new, struct file_lock *fl)
1053 static inline void locks_copy_lock(struct file_lock *new, struct file_lock *fl)
1058 static inline void locks_remove_posix(struct file *filp, fl_owner_t owner)
1063 static inline void locks_remove_flock(struct file *filp)
1068 static inline void posix_test_lock(struct file *filp, struct file_lock *fl)
1073 static inline int posix_lock_file(struct file *filp, struct file_lock *fl,
1074 struct file_lock *conflock)
1079 static inline int posix_lock_file_wait(struct file *filp, struct file_lock *fl)
1084 static inline int posix_unblock_lock(struct file *filp,
1085 struct file_lock *waiter)
1090 static inline int vfs_test_lock(struct file *filp, struct file_lock *fl)
1095 static inline int vfs_lock_file(struct file *filp, unsigned int cmd,
1096 struct file_lock *fl, struct file_lock *conf)
1101 static inline int vfs_cancel_lock(struct file *filp, struct file_lock *fl)
1106 static inline int flock_lock_file_wait(struct file *filp,
1107 struct file_lock *request)
1112 static inline int __break_lease(struct inode *inode, unsigned int mode)
1117 static inline void lease_get_mtime(struct inode *inode, struct timespec *time)
1122 static inline int generic_setlease(struct file *filp, long arg,
1123 struct file_lock **flp)
1128 static inline int vfs_setlease(struct file *filp, long arg,
1129 struct file_lock **lease)
1134 static inline int lease_modify(struct file_lock **before, int arg)
1139 static inline int lock_may_read(struct inode *inode, loff_t start,
1145 static inline int lock_may_write(struct inode *inode, loff_t start,
1151 static inline void locks_delete_block(struct file_lock *waiter)
1155 static inline void lock_flocks(void)
1159 static inline void unlock_flocks(void)
1163 #endif /* !CONFIG_FILE_LOCKING */
1166 struct fasync_struct {
1170 struct fasync_struct *fa_next; /* singly linked list */
1171 struct file *fa_file;
1172 struct rcu_head fa_rcu;
1175 #define FASYNC_MAGIC 0x4601
1177 /* SMP safe fasync helpers: */
1178 extern int fasync_helper(int, struct file *, int, struct fasync_struct **);
1179 extern struct fasync_struct *fasync_insert_entry(int, struct file *, struct fasync_struct **, struct fasync_struct *);
1180 extern int fasync_remove_entry(struct file *, struct fasync_struct **);
1181 extern struct fasync_struct *fasync_alloc(void);
1182 extern void fasync_free(struct fasync_struct *);
1184 /* can be called from interrupts */
1185 extern void kill_fasync(struct fasync_struct **, int, int);
1187 extern int __f_setown(struct file *filp, struct pid *, enum pid_type, int force);
1188 extern int f_setown(struct file *filp, unsigned long arg, int force);
1189 extern void f_delown(struct file *filp);
1190 extern pid_t f_getown(struct file *filp);
1191 extern int send_sigurg(struct fown_struct *fown);
1199 #define MNT_FORCE 0x00000001 /* Attempt to forcibily umount */
1200 #define MNT_DETACH 0x00000002 /* Just detach from the tree */
1201 #define MNT_EXPIRE 0x00000004 /* Mark for expiry */
1202 #define UMOUNT_NOFOLLOW 0x00000008 /* Don't follow symlink on umount */
1203 #define UMOUNT_UNUSED 0x80000000 /* Flag guaranteed to be unused */
1205 extern struct list_head super_blocks;
1206 extern spinlock_t sb_lock;
1208 /* Possible states of 'frozen' field */
1210 SB_UNFROZEN = 0, /* FS is unfrozen */
1211 SB_FREEZE_WRITE = 1, /* Writes, dir ops, ioctls frozen */
1212 SB_FREEZE_PAGEFAULT = 2, /* Page faults stopped as well */
1213 SB_FREEZE_FS = 3, /* For internal FS use (e.g. to stop
1214 * internal threads if needed) */
1215 SB_FREEZE_COMPLETE = 4, /* ->freeze_fs finished successfully */
1218 #define SB_FREEZE_LEVELS (SB_FREEZE_COMPLETE - 1)
1221 /* Counters for counting writers at each level */
1222 struct percpu_counter counter[SB_FREEZE_LEVELS];
1223 wait_queue_head_t wait; /* queue for waiting for
1224 writers / faults to finish */
1225 int frozen; /* Is sb frozen? */
1226 wait_queue_head_t wait_unfrozen; /* queue for waiting for
1228 #ifdef CONFIG_DEBUG_LOCK_ALLOC
1229 struct lockdep_map lock_map[SB_FREEZE_LEVELS];
1233 struct super_block {
1234 struct list_head s_list; /* Keep this first */
1235 dev_t s_dev; /* search index; _not_ kdev_t */
1236 unsigned char s_blocksize_bits;
1237 unsigned long s_blocksize;
1238 loff_t s_maxbytes; /* Max file size */
1239 struct file_system_type *s_type;
1240 const struct super_operations *s_op;
1241 const struct dquot_operations *dq_op;
1242 const struct quotactl_ops *s_qcop;
1243 const struct export_operations *s_export_op;
1244 unsigned long s_flags;
1245 unsigned long s_magic;
1246 struct dentry *s_root;
1247 struct rw_semaphore s_umount;
1250 #ifdef CONFIG_SECURITY
1253 const struct xattr_handler **s_xattr;
1255 struct list_head s_inodes; /* all inodes */
1256 struct hlist_bl_head s_anon; /* anonymous dentries for (nfs) exporting */
1258 struct list_head __percpu *s_files;
1260 struct list_head s_files;
1262 struct list_head s_mounts; /* list of mounts; _not_ for fs use */
1263 /* s_dentry_lru, s_nr_dentry_unused protected by dcache.c lru locks */
1264 struct list_head s_dentry_lru; /* unused dentry lru */
1265 int s_nr_dentry_unused; /* # of dentry on lru */
1267 /* s_inode_lru_lock protects s_inode_lru and s_nr_inodes_unused */
1268 spinlock_t s_inode_lru_lock ____cacheline_aligned_in_smp;
1269 struct list_head s_inode_lru; /* unused inode lru */
1270 int s_nr_inodes_unused; /* # of inodes on lru */
1272 struct block_device *s_bdev;
1273 struct backing_dev_info *s_bdi;
1274 struct mtd_info *s_mtd;
1275 struct hlist_node s_instances;
1276 struct quota_info s_dquot; /* Diskquota specific options */
1278 struct sb_writers s_writers;
1280 char s_id[32]; /* Informational name */
1281 u8 s_uuid[16]; /* UUID */
1283 void *s_fs_info; /* Filesystem private info */
1284 unsigned int s_max_links;
1287 /* Granularity of c/m/atime in ns.
1288 Cannot be worse than a second */
1292 * The next field is for VFS *only*. No filesystems have any business
1293 * even looking at it. You had been warned.
1295 struct mutex s_vfs_rename_mutex; /* Kludge */
1298 * Filesystem subtype. If non-empty the filesystem type field
1299 * in /proc/mounts will be "type.subtype"
1304 * Saved mount options for lazy filesystems using
1305 * generic_show_options()
1307 char __rcu *s_options;
1308 const struct dentry_operations *s_d_op; /* default d_op for dentries */
1311 * Saved pool identifier for cleancache (-1 means none)
1313 int cleancache_poolid;
1315 struct shrinker s_shrink; /* per-sb shrinker handle */
1317 /* Number of inodes with nlink == 0 but still referenced */
1318 atomic_long_t s_remove_count;
1320 /* Being remounted read-only */
1321 int s_readonly_remount;
1324 /* superblock cache pruning functions */
1325 extern void prune_icache_sb(struct super_block *sb, int nr_to_scan);
1326 extern void prune_dcache_sb(struct super_block *sb, int nr_to_scan);
1328 extern struct timespec current_fs_time(struct super_block *sb);
1331 * Snapshotting support.
1334 void __sb_end_write(struct super_block *sb, int level);
1335 int __sb_start_write(struct super_block *sb, int level, bool wait);
1338 * sb_end_write - drop write access to a superblock
1339 * @sb: the super we wrote to
1341 * Decrement number of writers to the filesystem. Wake up possible waiters
1342 * wanting to freeze the filesystem.
1344 static inline void sb_end_write(struct super_block *sb)
1346 __sb_end_write(sb, SB_FREEZE_WRITE);
1350 * sb_end_pagefault - drop write access to a superblock from a page fault
1351 * @sb: the super we wrote to
1353 * Decrement number of processes handling write page fault to the filesystem.
1354 * Wake up possible waiters wanting to freeze the filesystem.
1356 static inline void sb_end_pagefault(struct super_block *sb)
1358 __sb_end_write(sb, SB_FREEZE_PAGEFAULT);
1362 * sb_end_intwrite - drop write access to a superblock for internal fs purposes
1363 * @sb: the super we wrote to
1365 * Decrement fs-internal number of writers to the filesystem. Wake up possible
1366 * waiters wanting to freeze the filesystem.
1368 static inline void sb_end_intwrite(struct super_block *sb)
1370 __sb_end_write(sb, SB_FREEZE_FS);
1374 * sb_start_write - get write access to a superblock
1375 * @sb: the super we write to
1377 * When a process wants to write data or metadata to a file system (i.e. dirty
1378 * a page or an inode), it should embed the operation in a sb_start_write() -
1379 * sb_end_write() pair to get exclusion against file system freezing. This
1380 * function increments number of writers preventing freezing. If the file
1381 * system is already frozen, the function waits until the file system is
1384 * Since freeze protection behaves as a lock, users have to preserve
1385 * ordering of freeze protection and other filesystem locks. Generally,
1386 * freeze protection should be the outermost lock. In particular, we have:
1389 * -> i_mutex (write path, truncate, directory ops, ...)
1390 * -> s_umount (freeze_super, thaw_super)
1392 static inline void sb_start_write(struct super_block *sb)
1394 __sb_start_write(sb, SB_FREEZE_WRITE, true);
1397 static inline int sb_start_write_trylock(struct super_block *sb)
1399 return __sb_start_write(sb, SB_FREEZE_WRITE, false);
1403 * sb_start_pagefault - get write access to a superblock from a page fault
1404 * @sb: the super we write to
1406 * When a process starts handling write page fault, it should embed the
1407 * operation into sb_start_pagefault() - sb_end_pagefault() pair to get
1408 * exclusion against file system freezing. This is needed since the page fault
1409 * is going to dirty a page. This function increments number of running page
1410 * faults preventing freezing. If the file system is already frozen, the
1411 * function waits until the file system is thawed.
1413 * Since page fault freeze protection behaves as a lock, users have to preserve
1414 * ordering of freeze protection and other filesystem locks. It is advised to
1415 * put sb_start_pagefault() close to mmap_sem in lock ordering. Page fault
1416 * handling code implies lock dependency:
1419 * -> sb_start_pagefault
1421 static inline void sb_start_pagefault(struct super_block *sb)
1423 __sb_start_write(sb, SB_FREEZE_PAGEFAULT, true);
1427 * sb_start_intwrite - get write access to a superblock for internal fs purposes
1428 * @sb: the super we write to
1430 * This is the third level of protection against filesystem freezing. It is
1431 * free for use by a filesystem. The only requirement is that it must rank
1432 * below sb_start_pagefault.
1434 * For example filesystem can call sb_start_intwrite() when starting a
1435 * transaction which somewhat eases handling of freezing for internal sources
1436 * of filesystem changes (internal fs threads, discarding preallocation on file
1439 static inline void sb_start_intwrite(struct super_block *sb)
1441 __sb_start_write(sb, SB_FREEZE_FS, true);
1445 extern bool inode_owner_or_capable(const struct inode *inode);
1448 * VFS helper functions..
1450 extern int vfs_create(struct inode *, struct dentry *, umode_t, bool);
1451 extern int vfs_mkdir(struct inode *, struct dentry *, umode_t);
1452 extern int vfs_mknod(struct inode *, struct dentry *, umode_t, dev_t);
1453 extern int vfs_symlink(struct inode *, struct dentry *, const char *);
1454 extern int vfs_link(struct dentry *, struct inode *, struct dentry *);
1455 extern int vfs_rmdir(struct inode *, struct dentry *);
1456 extern int vfs_unlink(struct inode *, struct dentry *);
1457 extern int vfs_rename(struct inode *, struct dentry *, struct inode *, struct dentry *);
1460 * VFS dentry helper functions.
1462 extern void dentry_unhash(struct dentry *dentry);
1465 * VFS file helper functions.
1467 extern void inode_init_owner(struct inode *inode, const struct inode *dir,
1470 * VFS FS_IOC_FIEMAP helper definitions.
1472 struct fiemap_extent_info {
1473 unsigned int fi_flags; /* Flags as passed from user */
1474 unsigned int fi_extents_mapped; /* Number of mapped extents */
1475 unsigned int fi_extents_max; /* Size of fiemap_extent array */
1476 struct fiemap_extent __user *fi_extents_start; /* Start of
1477 fiemap_extent array */
1479 int fiemap_fill_next_extent(struct fiemap_extent_info *info, u64 logical,
1480 u64 phys, u64 len, u32 flags);
1481 int fiemap_check_flags(struct fiemap_extent_info *fieinfo, u32 fs_flags);
1486 * NOTE! These match bits 12..15 of stat.st_mode
1487 * (ie "(i_mode >> 12) & 15").
1489 #define DT_UNKNOWN 0
1500 * This is the "filldir" function type, used by readdir() to let
1501 * the kernel specify what kind of dirent layout it wants to have.
1502 * This allows the kernel to read directories into kernel space or
1503 * to have different dirent layouts depending on the binary type.
1505 typedef int (*filldir_t)(void *, const char *, int, loff_t, u64, unsigned);
1506 struct block_device_operations;
1508 /* These macros are for out of kernel modules to test that
1509 * the kernel supports the unlocked_ioctl and compat_ioctl
1510 * fields in struct file_operations. */
1511 #define HAVE_COMPAT_IOCTL 1
1512 #define HAVE_UNLOCKED_IOCTL 1
1514 struct file_operations {
1515 struct module *owner;
1516 loff_t (*llseek) (struct file *, loff_t, int);
1517 ssize_t (*read) (struct file *, char __user *, size_t, loff_t *);
1518 ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *);
1519 ssize_t (*aio_read) (struct kiocb *, const struct iovec *, unsigned long, loff_t);
1520 ssize_t (*aio_write) (struct kiocb *, const struct iovec *, unsigned long, loff_t);
1521 int (*readdir) (struct file *, void *, filldir_t);
1522 unsigned int (*poll) (struct file *, struct poll_table_struct *);
1523 long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long);
1524 long (*compat_ioctl) (struct file *, unsigned int, unsigned long);
1525 int (*mmap) (struct file *, struct vm_area_struct *);
1526 int (*open) (struct inode *, struct file *);
1527 int (*flush) (struct file *, fl_owner_t id);
1528 int (*release) (struct inode *, struct file *);
1529 int (*fsync) (struct file *, loff_t, loff_t, int datasync);
1530 int (*aio_fsync) (struct kiocb *, int datasync);
1531 int (*fasync) (int, struct file *, int);
1532 int (*lock) (struct file *, int, struct file_lock *);
1533 ssize_t (*sendpage) (struct file *, struct page *, int, size_t, loff_t *, int);
1534 unsigned long (*get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
1535 int (*check_flags)(int);
1536 int (*flock) (struct file *, int, struct file_lock *);
1537 ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, loff_t *, size_t, unsigned int);
1538 ssize_t (*splice_read)(struct file *, loff_t *, struct pipe_inode_info *, size_t, unsigned int);
1539 int (*setlease)(struct file *, long, struct file_lock **);
1540 long (*fallocate)(struct file *file, int mode, loff_t offset,
1542 int (*show_fdinfo)(struct seq_file *m, struct file *f);
1545 struct inode_operations {
1546 struct dentry * (*lookup) (struct inode *,struct dentry *, unsigned int);
1547 void * (*follow_link) (struct dentry *, struct nameidata *);
1548 int (*permission) (struct inode *, int);
1549 struct posix_acl * (*get_acl)(struct inode *, int);
1551 int (*readlink) (struct dentry *, char __user *,int);
1552 void (*put_link) (struct dentry *, struct nameidata *, void *);
1554 int (*create) (struct inode *,struct dentry *, umode_t, bool);
1555 int (*link) (struct dentry *,struct inode *,struct dentry *);
1556 int (*unlink) (struct inode *,struct dentry *);
1557 int (*symlink) (struct inode *,struct dentry *,const char *);
1558 int (*mkdir) (struct inode *,struct dentry *,umode_t);
1559 int (*rmdir) (struct inode *,struct dentry *);
1560 int (*mknod) (struct inode *,struct dentry *,umode_t,dev_t);
1561 int (*rename) (struct inode *, struct dentry *,
1562 struct inode *, struct dentry *);
1563 int (*setattr) (struct dentry *, struct iattr *);
1564 int (*getattr) (struct vfsmount *mnt, struct dentry *, struct kstat *);
1565 int (*setxattr) (struct dentry *, const char *,const void *,size_t,int);
1566 ssize_t (*getxattr) (struct dentry *, const char *, void *, size_t);
1567 ssize_t (*listxattr) (struct dentry *, char *, size_t);
1568 int (*removexattr) (struct dentry *, const char *);
1569 int (*fiemap)(struct inode *, struct fiemap_extent_info *, u64 start,
1571 int (*update_time)(struct inode *, struct timespec *, int);
1572 int (*atomic_open)(struct inode *, struct dentry *,
1573 struct file *, unsigned open_flag,
1574 umode_t create_mode, int *opened);
1575 } ____cacheline_aligned;
1577 ssize_t rw_copy_check_uvector(int type, const struct iovec __user * uvector,
1578 unsigned long nr_segs, unsigned long fast_segs,
1579 struct iovec *fast_pointer,
1580 struct iovec **ret_pointer);
1582 extern ssize_t vfs_read(struct file *, char __user *, size_t, loff_t *);
1583 extern ssize_t vfs_write(struct file *, const char __user *, size_t, loff_t *);
1584 extern ssize_t vfs_readv(struct file *, const struct iovec __user *,
1585 unsigned long, loff_t *);
1586 extern ssize_t vfs_writev(struct file *, const struct iovec __user *,
1587 unsigned long, loff_t *);
1589 struct super_operations {
1590 struct inode *(*alloc_inode)(struct super_block *sb);
1591 void (*destroy_inode)(struct inode *);
1593 void (*dirty_inode) (struct inode *, int flags);
1594 int (*write_inode) (struct inode *, struct writeback_control *wbc);
1595 int (*drop_inode) (struct inode *);
1596 void (*evict_inode) (struct inode *);
1597 void (*put_super) (struct super_block *);
1598 int (*sync_fs)(struct super_block *sb, int wait);
1599 int (*freeze_fs) (struct super_block *);
1600 int (*unfreeze_fs) (struct super_block *);
1601 int (*statfs) (struct dentry *, struct kstatfs *);
1602 int (*remount_fs) (struct super_block *, int *, char *);
1603 void (*umount_begin) (struct super_block *);
1605 int (*show_options)(struct seq_file *, struct dentry *);
1606 int (*show_devname)(struct seq_file *, struct dentry *);
1607 int (*show_path)(struct seq_file *, struct dentry *);
1608 int (*show_stats)(struct seq_file *, struct dentry *);
1610 ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t);
1611 ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t);
1613 int (*bdev_try_to_free_page)(struct super_block*, struct page*, gfp_t);
1614 int (*nr_cached_objects)(struct super_block *);
1615 void (*free_cached_objects)(struct super_block *, int);
1619 * Inode flags - they have no relation to superblock flags now
1621 #define S_SYNC 1 /* Writes are synced at once */
1622 #define S_NOATIME 2 /* Do not update access times */
1623 #define S_APPEND 4 /* Append-only file */
1624 #define S_IMMUTABLE 8 /* Immutable file */
1625 #define S_DEAD 16 /* removed, but still open directory */
1626 #define S_NOQUOTA 32 /* Inode is not counted to quota */
1627 #define S_DIRSYNC 64 /* Directory modifications are synchronous */
1628 #define S_NOCMTIME 128 /* Do not update file c/mtime */
1629 #define S_SWAPFILE 256 /* Do not truncate: swapon got its bmaps */
1630 #define S_PRIVATE 512 /* Inode is fs-internal */
1631 #define S_IMA 1024 /* Inode has an associated IMA struct */
1632 #define S_AUTOMOUNT 2048 /* Automount/referral quasi-directory */
1633 #define S_NOSEC 4096 /* no suid or xattr security attributes */
1636 * Note that nosuid etc flags are inode-specific: setting some file-system
1637 * flags just means all the inodes inherit those flags by default. It might be
1638 * possible to override it selectively if you really wanted to with some
1639 * ioctl() that is not currently implemented.
1641 * Exception: MS_RDONLY is always applied to the entire file system.
1643 * Unfortunately, it is possible to change a filesystems flags with it mounted
1644 * with files in use. This means that all of the inodes will not have their
1645 * i_flags updated. Hence, i_flags no longer inherit the superblock mount
1646 * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org
1648 #define __IS_FLG(inode, flg) ((inode)->i_sb->s_flags & (flg))
1650 #define IS_RDONLY(inode) ((inode)->i_sb->s_flags & MS_RDONLY)
1651 #define IS_SYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS) || \
1652 ((inode)->i_flags & S_SYNC))
1653 #define IS_DIRSYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS|MS_DIRSYNC) || \
1654 ((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
1655 #define IS_MANDLOCK(inode) __IS_FLG(inode, MS_MANDLOCK)
1656 #define IS_NOATIME(inode) __IS_FLG(inode, MS_RDONLY|MS_NOATIME)
1657 #define IS_I_VERSION(inode) __IS_FLG(inode, MS_I_VERSION)
1659 #define IS_NOQUOTA(inode) ((inode)->i_flags & S_NOQUOTA)
1660 #define IS_APPEND(inode) ((inode)->i_flags & S_APPEND)
1661 #define IS_IMMUTABLE(inode) ((inode)->i_flags & S_IMMUTABLE)
1662 #define IS_POSIXACL(inode) __IS_FLG(inode, MS_POSIXACL)
1664 #define IS_DEADDIR(inode) ((inode)->i_flags & S_DEAD)
1665 #define IS_NOCMTIME(inode) ((inode)->i_flags & S_NOCMTIME)
1666 #define IS_SWAPFILE(inode) ((inode)->i_flags & S_SWAPFILE)
1667 #define IS_PRIVATE(inode) ((inode)->i_flags & S_PRIVATE)
1668 #define IS_IMA(inode) ((inode)->i_flags & S_IMA)
1669 #define IS_AUTOMOUNT(inode) ((inode)->i_flags & S_AUTOMOUNT)
1670 #define IS_NOSEC(inode) ((inode)->i_flags & S_NOSEC)
1673 * Inode state bits. Protected by inode->i_lock
1675 * Three bits determine the dirty state of the inode, I_DIRTY_SYNC,
1676 * I_DIRTY_DATASYNC and I_DIRTY_PAGES.
1678 * Four bits define the lifetime of an inode. Initially, inodes are I_NEW,
1679 * until that flag is cleared. I_WILL_FREE, I_FREEING and I_CLEAR are set at
1680 * various stages of removing an inode.
1682 * Two bits are used for locking and completion notification, I_NEW and I_SYNC.
1684 * I_DIRTY_SYNC Inode is dirty, but doesn't have to be written on
1685 * fdatasync(). i_atime is the usual cause.
1686 * I_DIRTY_DATASYNC Data-related inode changes pending. We keep track of
1687 * these changes separately from I_DIRTY_SYNC so that we
1688 * don't have to write inode on fdatasync() when only
1689 * mtime has changed in it.
1690 * I_DIRTY_PAGES Inode has dirty pages. Inode itself may be clean.
1691 * I_NEW Serves as both a mutex and completion notification.
1692 * New inodes set I_NEW. If two processes both create
1693 * the same inode, one of them will release its inode and
1694 * wait for I_NEW to be released before returning.
1695 * Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can
1696 * also cause waiting on I_NEW, without I_NEW actually
1697 * being set. find_inode() uses this to prevent returning
1698 * nearly-dead inodes.
1699 * I_WILL_FREE Must be set when calling write_inode_now() if i_count
1700 * is zero. I_FREEING must be set when I_WILL_FREE is
1702 * I_FREEING Set when inode is about to be freed but still has dirty
1703 * pages or buffers attached or the inode itself is still
1705 * I_CLEAR Added by clear_inode(). In this state the inode is
1706 * clean and can be destroyed. Inode keeps I_FREEING.
1708 * Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are
1709 * prohibited for many purposes. iget() must wait for
1710 * the inode to be completely released, then create it
1711 * anew. Other functions will just ignore such inodes,
1712 * if appropriate. I_NEW is used for waiting.
1714 * I_SYNC Writeback of inode is running. The bit is set during
1715 * data writeback, and cleared with a wakeup on the bit
1716 * address once it is done. The bit is also used to pin
1717 * the inode in memory for flusher thread.
1719 * I_REFERENCED Marks the inode as recently references on the LRU list.
1721 * I_DIO_WAKEUP Never set. Only used as a key for wait_on_bit().
1723 * Q: What is the difference between I_WILL_FREE and I_FREEING?
1725 #define I_DIRTY_SYNC (1 << 0)
1726 #define I_DIRTY_DATASYNC (1 << 1)
1727 #define I_DIRTY_PAGES (1 << 2)
1729 #define I_NEW (1 << __I_NEW)
1730 #define I_WILL_FREE (1 << 4)
1731 #define I_FREEING (1 << 5)
1732 #define I_CLEAR (1 << 6)
1734 #define I_SYNC (1 << __I_SYNC)
1735 #define I_REFERENCED (1 << 8)
1736 #define __I_DIO_WAKEUP 9
1737 #define I_DIO_WAKEUP (1 << I_DIO_WAKEUP)
1739 #define I_DIRTY (I_DIRTY_SYNC | I_DIRTY_DATASYNC | I_DIRTY_PAGES)
1741 extern void __mark_inode_dirty(struct inode *, int);
1742 static inline void mark_inode_dirty(struct inode *inode)
1744 __mark_inode_dirty(inode, I_DIRTY);
1747 static inline void mark_inode_dirty_sync(struct inode *inode)
1749 __mark_inode_dirty(inode, I_DIRTY_SYNC);
1752 extern void inc_nlink(struct inode *inode);
1753 extern void drop_nlink(struct inode *inode);
1754 extern void clear_nlink(struct inode *inode);
1755 extern void set_nlink(struct inode *inode, unsigned int nlink);
1757 static inline void inode_inc_link_count(struct inode *inode)
1760 mark_inode_dirty(inode);
1763 static inline void inode_dec_link_count(struct inode *inode)
1766 mark_inode_dirty(inode);
1770 * inode_inc_iversion - increments i_version
1771 * @inode: inode that need to be updated
1773 * Every time the inode is modified, the i_version field will be incremented.
1774 * The filesystem has to be mounted with i_version flag
1777 static inline void inode_inc_iversion(struct inode *inode)
1779 spin_lock(&inode->i_lock);
1781 spin_unlock(&inode->i_lock);
1784 enum file_time_flags {
1791 extern void touch_atime(struct path *);
1792 static inline void file_accessed(struct file *file)
1794 if (!(file->f_flags & O_NOATIME))
1795 touch_atime(&file->f_path);
1798 int sync_inode(struct inode *inode, struct writeback_control *wbc);
1799 int sync_inode_metadata(struct inode *inode, int wait);
1801 struct file_system_type {
1804 #define FS_REQUIRES_DEV 1
1805 #define FS_BINARY_MOUNTDATA 2
1806 #define FS_HAS_SUBTYPE 4
1807 #define FS_USERNS_MOUNT 8 /* Can be mounted by userns root */
1808 #define FS_USERNS_DEV_MOUNT 16 /* A userns mount does not imply MNT_NODEV */
1809 #define FS_RENAME_DOES_D_MOVE 32768 /* FS will handle d_move() during rename() internally. */
1810 struct dentry *(*mount) (struct file_system_type *, int,
1811 const char *, void *);
1812 void (*kill_sb) (struct super_block *);
1813 struct module *owner;
1814 struct file_system_type * next;
1815 struct hlist_head fs_supers;
1817 struct lock_class_key s_lock_key;
1818 struct lock_class_key s_umount_key;
1819 struct lock_class_key s_vfs_rename_key;
1820 struct lock_class_key s_writers_key[SB_FREEZE_LEVELS];
1822 struct lock_class_key i_lock_key;
1823 struct lock_class_key i_mutex_key;
1824 struct lock_class_key i_mutex_dir_key;
1827 extern struct dentry *mount_ns(struct file_system_type *fs_type, int flags,
1828 void *data, int (*fill_super)(struct super_block *, void *, int));
1829 extern struct dentry *mount_bdev(struct file_system_type *fs_type,
1830 int flags, const char *dev_name, void *data,
1831 int (*fill_super)(struct super_block *, void *, int));
1832 extern struct dentry *mount_single(struct file_system_type *fs_type,
1833 int flags, void *data,
1834 int (*fill_super)(struct super_block *, void *, int));
1835 extern struct dentry *mount_nodev(struct file_system_type *fs_type,
1836 int flags, void *data,
1837 int (*fill_super)(struct super_block *, void *, int));
1838 extern struct dentry *mount_subtree(struct vfsmount *mnt, const char *path);
1839 void generic_shutdown_super(struct super_block *sb);
1840 void kill_block_super(struct super_block *sb);
1841 void kill_anon_super(struct super_block *sb);
1842 void kill_litter_super(struct super_block *sb);
1843 void deactivate_super(struct super_block *sb);
1844 void deactivate_locked_super(struct super_block *sb);
1845 int set_anon_super(struct super_block *s, void *data);
1846 int get_anon_bdev(dev_t *);
1847 void free_anon_bdev(dev_t);
1848 struct super_block *sget(struct file_system_type *type,
1849 int (*test)(struct super_block *,void *),
1850 int (*set)(struct super_block *,void *),
1851 int flags, void *data);
1852 extern struct dentry *mount_pseudo(struct file_system_type *, char *,
1853 const struct super_operations *ops,
1854 const struct dentry_operations *dops,
1857 /* Alas, no aliases. Too much hassle with bringing module.h everywhere */
1858 #define fops_get(fops) \
1859 (((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
1860 #define fops_put(fops) \
1861 do { if (fops) module_put((fops)->owner); } while(0)
1863 extern int register_filesystem(struct file_system_type *);
1864 extern int unregister_filesystem(struct file_system_type *);
1865 extern struct vfsmount *kern_mount_data(struct file_system_type *, void *data);
1866 #define kern_mount(type) kern_mount_data(type, NULL)
1867 extern void kern_unmount(struct vfsmount *mnt);
1868 extern int may_umount_tree(struct vfsmount *);
1869 extern int may_umount(struct vfsmount *);
1870 extern long do_mount(const char *, const char *, const char *, unsigned long, void *);
1871 extern struct vfsmount *collect_mounts(struct path *);
1872 extern void drop_collected_mounts(struct vfsmount *);
1873 extern int iterate_mounts(int (*)(struct vfsmount *, void *), void *,
1875 extern int vfs_statfs(struct path *, struct kstatfs *);
1876 extern int user_statfs(const char __user *, struct kstatfs *);
1877 extern int fd_statfs(int, struct kstatfs *);
1878 extern int vfs_ustat(dev_t, struct kstatfs *);
1879 extern int freeze_super(struct super_block *super);
1880 extern int thaw_super(struct super_block *super);
1881 extern bool our_mnt(struct vfsmount *mnt);
1883 extern int current_umask(void);
1886 extern struct kobject *fs_kobj;
1888 #define MAX_RW_COUNT (INT_MAX & PAGE_CACHE_MASK)
1889 extern int rw_verify_area(int, struct file *, loff_t *, size_t);
1891 #define FLOCK_VERIFY_READ 1
1892 #define FLOCK_VERIFY_WRITE 2
1894 #ifdef CONFIG_FILE_LOCKING
1895 extern int locks_mandatory_locked(struct inode *);
1896 extern int locks_mandatory_area(int, struct inode *, struct file *, loff_t, size_t);
1899 * Candidates for mandatory locking have the setgid bit set
1900 * but no group execute bit - an otherwise meaningless combination.
1903 static inline int __mandatory_lock(struct inode *ino)
1905 return (ino->i_mode & (S_ISGID | S_IXGRP)) == S_ISGID;
1909 * ... and these candidates should be on MS_MANDLOCK mounted fs,
1910 * otherwise these will be advisory locks
1913 static inline int mandatory_lock(struct inode *ino)
1915 return IS_MANDLOCK(ino) && __mandatory_lock(ino);
1918 static inline int locks_verify_locked(struct inode *inode)
1920 if (mandatory_lock(inode))
1921 return locks_mandatory_locked(inode);
1925 static inline int locks_verify_truncate(struct inode *inode,
1929 if (inode->i_flock && mandatory_lock(inode))
1930 return locks_mandatory_area(
1931 FLOCK_VERIFY_WRITE, inode, filp,
1932 size < inode->i_size ? size : inode->i_size,
1933 (size < inode->i_size ? inode->i_size - size
1934 : size - inode->i_size)
1939 static inline int break_lease(struct inode *inode, unsigned int mode)
1942 return __break_lease(inode, mode);
1945 #else /* !CONFIG_FILE_LOCKING */
1946 static inline int locks_mandatory_locked(struct inode *inode)
1951 static inline int locks_mandatory_area(int rw, struct inode *inode,
1952 struct file *filp, loff_t offset,
1958 static inline int __mandatory_lock(struct inode *inode)
1963 static inline int mandatory_lock(struct inode *inode)
1968 static inline int locks_verify_locked(struct inode *inode)
1973 static inline int locks_verify_truncate(struct inode *inode, struct file *filp,
1979 static inline int break_lease(struct inode *inode, unsigned int mode)
1984 #endif /* CONFIG_FILE_LOCKING */
1989 const char *name; /* pointer to actual string */
1990 const __user char *uptr; /* original userland pointer */
1991 struct audit_names *aname;
1992 bool separate; /* should "name" be freed? */
1995 extern long vfs_truncate(struct path *, loff_t);
1996 extern int do_truncate(struct dentry *, loff_t start, unsigned int time_attrs,
1998 extern int do_fallocate(struct file *file, int mode, loff_t offset,
2000 extern long do_sys_open(int dfd, const char __user *filename, int flags,
2002 extern struct file *file_open_name(struct filename *, int, umode_t);
2003 extern struct file *filp_open(const char *, int, umode_t);
2004 extern struct file *file_open_root(struct dentry *, struct vfsmount *,
2006 extern struct file * dentry_open(const struct path *, int, const struct cred *);
2007 extern int filp_close(struct file *, fl_owner_t id);
2009 extern struct filename *getname(const char __user *);
2015 extern int finish_open(struct file *file, struct dentry *dentry,
2016 int (*open)(struct inode *, struct file *),
2018 extern int finish_no_open(struct file *file, struct dentry *dentry);
2022 extern int ioctl_preallocate(struct file *filp, void __user *argp);
2025 extern void __init vfs_caches_init_early(void);
2026 extern void __init vfs_caches_init(unsigned long);
2028 extern struct kmem_cache *names_cachep;
2030 extern void final_putname(struct filename *name);
2032 #define __getname() kmem_cache_alloc(names_cachep, GFP_KERNEL)
2033 #define __putname(name) kmem_cache_free(names_cachep, (void *)(name))
2034 #ifndef CONFIG_AUDITSYSCALL
2035 #define putname(name) final_putname(name)
2037 extern void putname(struct filename *name);
2041 extern int register_blkdev(unsigned int, const char *);
2042 extern void unregister_blkdev(unsigned int, const char *);
2043 extern struct block_device *bdget(dev_t);
2044 extern struct block_device *bdgrab(struct block_device *bdev);
2045 extern void bd_set_size(struct block_device *, loff_t size);
2046 extern void bd_forget(struct inode *inode);
2047 extern void bdput(struct block_device *);
2048 extern void invalidate_bdev(struct block_device *);
2049 extern void iterate_bdevs(void (*)(struct block_device *, void *), void *);
2050 extern int sync_blockdev(struct block_device *bdev);
2051 extern void kill_bdev(struct block_device *);
2052 extern struct super_block *freeze_bdev(struct block_device *);
2053 extern void emergency_thaw_all(void);
2054 extern int thaw_bdev(struct block_device *bdev, struct super_block *sb);
2055 extern int fsync_bdev(struct block_device *);
2057 static inline void bd_forget(struct inode *inode) {}
2058 static inline int sync_blockdev(struct block_device *bdev) { return 0; }
2059 static inline void kill_bdev(struct block_device *bdev) {}
2060 static inline void invalidate_bdev(struct block_device *bdev) {}
2062 static inline struct super_block *freeze_bdev(struct block_device *sb)
2067 static inline int thaw_bdev(struct block_device *bdev, struct super_block *sb)
2072 static inline void iterate_bdevs(void (*f)(struct block_device *, void *), void *arg)
2076 extern int sync_filesystem(struct super_block *);
2077 extern const struct file_operations def_blk_fops;
2078 extern const struct file_operations def_chr_fops;
2079 extern const struct file_operations bad_sock_fops;
2080 extern const struct file_operations def_fifo_fops;
2082 extern int ioctl_by_bdev(struct block_device *, unsigned, unsigned long);
2083 extern int blkdev_ioctl(struct block_device *, fmode_t, unsigned, unsigned long);
2084 extern long compat_blkdev_ioctl(struct file *, unsigned, unsigned long);
2085 extern int blkdev_get(struct block_device *bdev, fmode_t mode, void *holder);
2086 extern struct block_device *blkdev_get_by_path(const char *path, fmode_t mode,
2088 extern struct block_device *blkdev_get_by_dev(dev_t dev, fmode_t mode,
2090 extern int blkdev_put(struct block_device *bdev, fmode_t mode);
2092 extern int bd_link_disk_holder(struct block_device *bdev, struct gendisk *disk);
2093 extern void bd_unlink_disk_holder(struct block_device *bdev,
2094 struct gendisk *disk);
2096 static inline int bd_link_disk_holder(struct block_device *bdev,
2097 struct gendisk *disk)
2101 static inline void bd_unlink_disk_holder(struct block_device *bdev,
2102 struct gendisk *disk)
2109 #define CHRDEV_MAJOR_HASH_SIZE 255
2110 extern int alloc_chrdev_region(dev_t *, unsigned, unsigned, const char *);
2111 extern int register_chrdev_region(dev_t, unsigned, const char *);
2112 extern int __register_chrdev(unsigned int major, unsigned int baseminor,
2113 unsigned int count, const char *name,
2114 const struct file_operations *fops);
2115 extern void __unregister_chrdev(unsigned int major, unsigned int baseminor,
2116 unsigned int count, const char *name);
2117 extern void unregister_chrdev_region(dev_t, unsigned);
2118 extern void chrdev_show(struct seq_file *,off_t);
2120 static inline int register_chrdev(unsigned int major, const char *name,
2121 const struct file_operations *fops)
2123 return __register_chrdev(major, 0, 256, name, fops);
2126 static inline void unregister_chrdev(unsigned int major, const char *name)
2128 __unregister_chrdev(major, 0, 256, name);
2131 /* fs/block_dev.c */
2132 #define BDEVNAME_SIZE 32 /* Largest string for a blockdev identifier */
2133 #define BDEVT_SIZE 10 /* Largest string for MAJ:MIN for blkdev */
2136 #define BLKDEV_MAJOR_HASH_SIZE 255
2137 extern const char *__bdevname(dev_t, char *buffer);
2138 extern const char *bdevname(struct block_device *bdev, char *buffer);
2139 extern struct block_device *lookup_bdev(const char *);
2140 extern void blkdev_show(struct seq_file *,off_t);
2143 #define BLKDEV_MAJOR_HASH_SIZE 0
2146 extern void init_special_inode(struct inode *, umode_t, dev_t);
2148 /* Invalid inode operations -- fs/bad_inode.c */
2149 extern void make_bad_inode(struct inode *);
2150 extern int is_bad_inode(struct inode *);
2152 extern const struct file_operations read_pipefifo_fops;
2153 extern const struct file_operations write_pipefifo_fops;
2154 extern const struct file_operations rdwr_pipefifo_fops;
2158 * return READ, READA, or WRITE
2160 #define bio_rw(bio) ((bio)->bi_rw & (RW_MASK | RWA_MASK))
2163 * return data direction, READ or WRITE
2165 #define bio_data_dir(bio) ((bio)->bi_rw & 1)
2167 extern void check_disk_size_change(struct gendisk *disk,
2168 struct block_device *bdev);
2169 extern int revalidate_disk(struct gendisk *);
2170 extern int check_disk_change(struct block_device *);
2171 extern int __invalidate_device(struct block_device *, bool);
2172 extern int invalidate_partition(struct gendisk *, int);
2174 unsigned long invalidate_mapping_pages(struct address_space *mapping,
2175 pgoff_t start, pgoff_t end);
2177 static inline void invalidate_remote_inode(struct inode *inode)
2179 if (S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) ||
2180 S_ISLNK(inode->i_mode))
2181 invalidate_mapping_pages(inode->i_mapping, 0, -1);
2183 extern int invalidate_inode_pages2(struct address_space *mapping);
2184 extern int invalidate_inode_pages2_range(struct address_space *mapping,
2185 pgoff_t start, pgoff_t end);
2186 extern int write_inode_now(struct inode *, int);
2187 extern int filemap_fdatawrite(struct address_space *);
2188 extern int filemap_flush(struct address_space *);
2189 extern int filemap_fdatawait(struct address_space *);
2190 extern int filemap_fdatawait_range(struct address_space *, loff_t lstart,
2192 extern int filemap_write_and_wait(struct address_space *mapping);
2193 extern int filemap_write_and_wait_range(struct address_space *mapping,
2194 loff_t lstart, loff_t lend);
2195 extern int __filemap_fdatawrite_range(struct address_space *mapping,
2196 loff_t start, loff_t end, int sync_mode);
2197 extern int filemap_fdatawrite_range(struct address_space *mapping,
2198 loff_t start, loff_t end);
2200 extern int vfs_fsync_range(struct file *file, loff_t start, loff_t end,
2202 extern int vfs_fsync(struct file *file, int datasync);
2203 extern int generic_write_sync(struct file *file, loff_t pos, loff_t count);
2204 extern void emergency_sync(void);
2205 extern void emergency_remount(void);
2207 extern sector_t bmap(struct inode *, sector_t);
2209 extern int notify_change(struct dentry *, struct iattr *);
2210 extern int inode_permission(struct inode *, int);
2211 extern int generic_permission(struct inode *, int);
2213 static inline bool execute_ok(struct inode *inode)
2215 return (inode->i_mode & S_IXUGO) || S_ISDIR(inode->i_mode);
2218 static inline struct inode *file_inode(struct file *f)
2220 return f->f_path.dentry->d_inode;
2224 * get_write_access() gets write permission for a file.
2225 * put_write_access() releases this write permission.
2226 * This is used for regular files.
2227 * We cannot support write (and maybe mmap read-write shared) accesses and
2228 * MAP_DENYWRITE mmappings simultaneously. The i_writecount field of an inode
2229 * can have the following values:
2230 * 0: no writers, no VM_DENYWRITE mappings
2231 * < 0: (-i_writecount) vm_area_structs with VM_DENYWRITE set exist
2232 * > 0: (i_writecount) users are writing to the file.
2234 * Normally we operate on that counter with atomic_{inc,dec} and it's safe
2235 * except for the cases where we don't hold i_writecount yet. Then we need to
2236 * use {get,deny}_write_access() - these functions check the sign and refuse
2237 * to do the change if sign is wrong.
2239 static inline int get_write_access(struct inode *inode)
2241 return atomic_inc_unless_negative(&inode->i_writecount) ? 0 : -ETXTBSY;
2243 static inline int deny_write_access(struct file *file)
2245 struct inode *inode = file_inode(file);
2246 return atomic_dec_unless_positive(&inode->i_writecount) ? 0 : -ETXTBSY;
2248 static inline void put_write_access(struct inode * inode)
2250 atomic_dec(&inode->i_writecount);
2252 static inline void allow_write_access(struct file *file)
2255 atomic_inc(&file_inode(file)->i_writecount);
2258 static inline void i_readcount_dec(struct inode *inode)
2260 BUG_ON(!atomic_read(&inode->i_readcount));
2261 atomic_dec(&inode->i_readcount);
2263 static inline void i_readcount_inc(struct inode *inode)
2265 atomic_inc(&inode->i_readcount);
2268 static inline void i_readcount_dec(struct inode *inode)
2272 static inline void i_readcount_inc(struct inode *inode)
2277 extern int do_pipe_flags(int *, int);
2279 extern int kernel_read(struct file *, loff_t, char *, unsigned long);
2280 extern struct file * open_exec(const char *);
2282 /* fs/dcache.c -- generic fs support functions */
2283 extern int is_subdir(struct dentry *, struct dentry *);
2284 extern int path_is_under(struct path *, struct path *);
2285 extern ino_t find_inode_number(struct dentry *, struct qstr *);
2287 #include <linux/err.h>
2289 /* needed for stackable file system support */
2290 extern loff_t default_llseek(struct file *file, loff_t offset, int whence);
2292 extern loff_t vfs_llseek(struct file *file, loff_t offset, int whence);
2294 extern int inode_init_always(struct super_block *, struct inode *);
2295 extern void inode_init_once(struct inode *);
2296 extern void address_space_init_once(struct address_space *mapping);
2297 extern void ihold(struct inode * inode);
2298 extern void iput(struct inode *);
2299 extern struct inode * igrab(struct inode *);
2300 extern ino_t iunique(struct super_block *, ino_t);
2301 extern int inode_needs_sync(struct inode *inode);
2302 extern int generic_delete_inode(struct inode *inode);
2303 static inline int generic_drop_inode(struct inode *inode)
2305 return !inode->i_nlink || inode_unhashed(inode);
2308 extern struct inode *ilookup5_nowait(struct super_block *sb,
2309 unsigned long hashval, int (*test)(struct inode *, void *),
2311 extern struct inode *ilookup5(struct super_block *sb, unsigned long hashval,
2312 int (*test)(struct inode *, void *), void *data);
2313 extern struct inode *ilookup(struct super_block *sb, unsigned long ino);
2315 extern struct inode * iget5_locked(struct super_block *, unsigned long, int (*test)(struct inode *, void *), int (*set)(struct inode *, void *), void *);
2316 extern struct inode * iget_locked(struct super_block *, unsigned long);
2317 extern int insert_inode_locked4(struct inode *, unsigned long, int (*test)(struct inode *, void *), void *);
2318 extern int insert_inode_locked(struct inode *);
2319 #ifdef CONFIG_DEBUG_LOCK_ALLOC
2320 extern void lockdep_annotate_inode_mutex_key(struct inode *inode);
2322 static inline void lockdep_annotate_inode_mutex_key(struct inode *inode) { };
2324 extern void unlock_new_inode(struct inode *);
2325 extern unsigned int get_next_ino(void);
2327 extern void __iget(struct inode * inode);
2328 extern void iget_failed(struct inode *);
2329 extern void clear_inode(struct inode *);
2330 extern void __destroy_inode(struct inode *);
2331 extern struct inode *new_inode_pseudo(struct super_block *sb);
2332 extern struct inode *new_inode(struct super_block *sb);
2333 extern void free_inode_nonrcu(struct inode *inode);
2334 extern int should_remove_suid(struct dentry *);
2335 extern int file_remove_suid(struct file *);
2337 extern void __insert_inode_hash(struct inode *, unsigned long hashval);
2338 static inline void insert_inode_hash(struct inode *inode)
2340 __insert_inode_hash(inode, inode->i_ino);
2343 extern void __remove_inode_hash(struct inode *);
2344 static inline void remove_inode_hash(struct inode *inode)
2346 if (!inode_unhashed(inode))
2347 __remove_inode_hash(inode);
2350 extern void inode_sb_list_add(struct inode *inode);
2353 extern void submit_bio(int, struct bio *);
2354 extern int bdev_read_only(struct block_device *);
2356 extern int set_blocksize(struct block_device *, int);
2357 extern int sb_set_blocksize(struct super_block *, int);
2358 extern int sb_min_blocksize(struct super_block *, int);
2360 extern int generic_file_mmap(struct file *, struct vm_area_struct *);
2361 extern int generic_file_readonly_mmap(struct file *, struct vm_area_struct *);
2362 extern int generic_file_remap_pages(struct vm_area_struct *, unsigned long addr,
2363 unsigned long size, pgoff_t pgoff);
2364 extern int file_read_actor(read_descriptor_t * desc, struct page *page, unsigned long offset, unsigned long size);
2365 int generic_write_checks(struct file *file, loff_t *pos, size_t *count, int isblk);
2366 extern ssize_t generic_file_aio_read(struct kiocb *, const struct iovec *, unsigned long, loff_t);
2367 extern ssize_t __generic_file_aio_write(struct kiocb *, const struct iovec *, unsigned long,
2369 extern ssize_t generic_file_aio_write(struct kiocb *, const struct iovec *, unsigned long, loff_t);
2370 extern ssize_t generic_file_direct_write(struct kiocb *, const struct iovec *,
2371 unsigned long *, loff_t, loff_t *, size_t, size_t);
2372 extern ssize_t generic_file_buffered_write(struct kiocb *, const struct iovec *,
2373 unsigned long, loff_t, loff_t *, size_t, ssize_t);
2374 extern ssize_t do_sync_read(struct file *filp, char __user *buf, size_t len, loff_t *ppos);
2375 extern ssize_t do_sync_write(struct file *filp, const char __user *buf, size_t len, loff_t *ppos);
2376 extern int generic_segment_checks(const struct iovec *iov,
2377 unsigned long *nr_segs, size_t *count, int access_flags);
2379 /* fs/block_dev.c */
2380 extern ssize_t blkdev_aio_write(struct kiocb *iocb, const struct iovec *iov,
2381 unsigned long nr_segs, loff_t pos);
2382 extern int blkdev_fsync(struct file *filp, loff_t start, loff_t end,
2384 extern void block_sync_page(struct page *page);
2387 extern ssize_t generic_file_splice_read(struct file *, loff_t *,
2388 struct pipe_inode_info *, size_t, unsigned int);
2389 extern ssize_t default_file_splice_read(struct file *, loff_t *,
2390 struct pipe_inode_info *, size_t, unsigned int);
2391 extern ssize_t generic_file_splice_write(struct pipe_inode_info *,
2392 struct file *, loff_t *, size_t, unsigned int);
2393 extern ssize_t generic_splice_sendpage(struct pipe_inode_info *pipe,
2394 struct file *out, loff_t *, size_t len, unsigned int flags);
2395 extern long do_splice_direct(struct file *in, loff_t *ppos, struct file *out,
2396 size_t len, unsigned int flags);
2399 file_ra_state_init(struct file_ra_state *ra, struct address_space *mapping);
2400 extern loff_t noop_llseek(struct file *file, loff_t offset, int whence);
2401 extern loff_t no_llseek(struct file *file, loff_t offset, int whence);
2402 extern loff_t generic_file_llseek(struct file *file, loff_t offset, int whence);
2403 extern loff_t generic_file_llseek_size(struct file *file, loff_t offset,
2404 int whence, loff_t maxsize, loff_t eof);
2405 extern int generic_file_open(struct inode * inode, struct file * filp);
2406 extern int nonseekable_open(struct inode * inode, struct file * filp);
2408 #ifdef CONFIG_FS_XIP
2409 extern ssize_t xip_file_read(struct file *filp, char __user *buf, size_t len,
2411 extern int xip_file_mmap(struct file * file, struct vm_area_struct * vma);
2412 extern ssize_t xip_file_write(struct file *filp, const char __user *buf,
2413 size_t len, loff_t *ppos);
2414 extern int xip_truncate_page(struct address_space *mapping, loff_t from);
2416 static inline int xip_truncate_page(struct address_space *mapping, loff_t from)
2423 typedef void (dio_submit_t)(int rw, struct bio *bio, struct inode *inode,
2424 loff_t file_offset);
2427 /* need locking between buffered and direct access */
2430 /* filesystem does not support filling holes */
2431 DIO_SKIP_HOLES = 0x02,
2434 void dio_end_io(struct bio *bio, int error);
2436 ssize_t __blockdev_direct_IO(int rw, struct kiocb *iocb, struct inode *inode,
2437 struct block_device *bdev, const struct iovec *iov, loff_t offset,
2438 unsigned long nr_segs, get_block_t get_block, dio_iodone_t end_io,
2439 dio_submit_t submit_io, int flags);
2441 static inline ssize_t blockdev_direct_IO(int rw, struct kiocb *iocb,
2442 struct inode *inode, const struct iovec *iov, loff_t offset,
2443 unsigned long nr_segs, get_block_t get_block)
2445 return __blockdev_direct_IO(rw, iocb, inode, inode->i_sb->s_bdev, iov,
2446 offset, nr_segs, get_block, NULL, NULL,
2447 DIO_LOCKING | DIO_SKIP_HOLES);
2451 void inode_dio_wait(struct inode *inode);
2452 void inode_dio_done(struct inode *inode);
2454 extern const struct file_operations generic_ro_fops;
2456 #define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
2458 extern int vfs_readlink(struct dentry *, char __user *, int, const char *);
2459 extern int vfs_follow_link(struct nameidata *, const char *);
2460 extern int page_readlink(struct dentry *, char __user *, int);
2461 extern void *page_follow_link_light(struct dentry *, struct nameidata *);
2462 extern void page_put_link(struct dentry *, struct nameidata *, void *);
2463 extern int __page_symlink(struct inode *inode, const char *symname, int len,
2465 extern int page_symlink(struct inode *inode, const char *symname, int len);
2466 extern const struct inode_operations page_symlink_inode_operations;
2467 extern int generic_readlink(struct dentry *, char __user *, int);
2468 extern void generic_fillattr(struct inode *, struct kstat *);
2469 extern int vfs_getattr(struct path *, struct kstat *);
2470 void __inode_add_bytes(struct inode *inode, loff_t bytes);
2471 void inode_add_bytes(struct inode *inode, loff_t bytes);
2472 void inode_sub_bytes(struct inode *inode, loff_t bytes);
2473 loff_t inode_get_bytes(struct inode *inode);
2474 void inode_set_bytes(struct inode *inode, loff_t bytes);
2476 extern int vfs_readdir(struct file *, filldir_t, void *);
2478 extern int vfs_stat(const char __user *, struct kstat *);
2479 extern int vfs_lstat(const char __user *, struct kstat *);
2480 extern int vfs_fstat(unsigned int, struct kstat *);
2481 extern int vfs_fstatat(int , const char __user *, struct kstat *, int);
2483 extern int do_vfs_ioctl(struct file *filp, unsigned int fd, unsigned int cmd,
2485 extern int __generic_block_fiemap(struct inode *inode,
2486 struct fiemap_extent_info *fieinfo,
2487 loff_t start, loff_t len,
2488 get_block_t *get_block);
2489 extern int generic_block_fiemap(struct inode *inode,
2490 struct fiemap_extent_info *fieinfo, u64 start,
2491 u64 len, get_block_t *get_block);
2493 extern void get_filesystem(struct file_system_type *fs);
2494 extern void put_filesystem(struct file_system_type *fs);
2495 extern struct file_system_type *get_fs_type(const char *name);
2496 extern struct super_block *get_super(struct block_device *);
2497 extern struct super_block *get_super_thawed(struct block_device *);
2498 extern struct super_block *get_active_super(struct block_device *bdev);
2499 extern void drop_super(struct super_block *sb);
2500 extern void iterate_supers(void (*)(struct super_block *, void *), void *);
2501 extern void iterate_supers_type(struct file_system_type *,
2502 void (*)(struct super_block *, void *), void *);
2504 extern int dcache_dir_open(struct inode *, struct file *);
2505 extern int dcache_dir_close(struct inode *, struct file *);
2506 extern loff_t dcache_dir_lseek(struct file *, loff_t, int);
2507 extern int dcache_readdir(struct file *, void *, filldir_t);
2508 extern int simple_setattr(struct dentry *, struct iattr *);
2509 extern int simple_getattr(struct vfsmount *, struct dentry *, struct kstat *);
2510 extern int simple_statfs(struct dentry *, struct kstatfs *);
2511 extern int simple_open(struct inode *inode, struct file *file);
2512 extern int simple_link(struct dentry *, struct inode *, struct dentry *);
2513 extern int simple_unlink(struct inode *, struct dentry *);
2514 extern int simple_rmdir(struct inode *, struct dentry *);
2515 extern int simple_rename(struct inode *, struct dentry *, struct inode *, struct dentry *);
2516 extern int noop_fsync(struct file *, loff_t, loff_t, int);
2517 extern int simple_empty(struct dentry *);
2518 extern int simple_readpage(struct file *file, struct page *page);
2519 extern int simple_write_begin(struct file *file, struct address_space *mapping,
2520 loff_t pos, unsigned len, unsigned flags,
2521 struct page **pagep, void **fsdata);
2522 extern int simple_write_end(struct file *file, struct address_space *mapping,
2523 loff_t pos, unsigned len, unsigned copied,
2524 struct page *page, void *fsdata);
2526 extern struct dentry *simple_lookup(struct inode *, struct dentry *, unsigned int flags);
2527 extern ssize_t generic_read_dir(struct file *, char __user *, size_t, loff_t *);
2528 extern const struct file_operations simple_dir_operations;
2529 extern const struct inode_operations simple_dir_inode_operations;
2530 struct tree_descr { char *name; const struct file_operations *ops; int mode; };
2531 struct dentry *d_alloc_name(struct dentry *, const char *);
2532 extern int simple_fill_super(struct super_block *, unsigned long, struct tree_descr *);
2533 extern int simple_pin_fs(struct file_system_type *, struct vfsmount **mount, int *count);
2534 extern void simple_release_fs(struct vfsmount **mount, int *count);
2536 extern ssize_t simple_read_from_buffer(void __user *to, size_t count,
2537 loff_t *ppos, const void *from, size_t available);
2538 extern ssize_t simple_write_to_buffer(void *to, size_t available, loff_t *ppos,
2539 const void __user *from, size_t count);
2541 extern int generic_file_fsync(struct file *, loff_t, loff_t, int);
2543 extern int generic_check_addressable(unsigned, u64);
2545 #ifdef CONFIG_MIGRATION
2546 extern int buffer_migrate_page(struct address_space *,
2547 struct page *, struct page *,
2550 #define buffer_migrate_page NULL
2553 extern int inode_change_ok(const struct inode *, struct iattr *);
2554 extern int inode_newsize_ok(const struct inode *, loff_t offset);
2555 extern void setattr_copy(struct inode *inode, const struct iattr *attr);
2557 extern int file_update_time(struct file *file);
2559 extern int generic_show_options(struct seq_file *m, struct dentry *root);
2560 extern void save_mount_options(struct super_block *sb, char *options);
2561 extern void replace_mount_options(struct super_block *sb, char *options);
2563 static inline ino_t parent_ino(struct dentry *dentry)
2568 * Don't strictly need d_lock here? If the parent ino could change
2569 * then surely we'd have a deeper race in the caller?
2571 spin_lock(&dentry->d_lock);
2572 res = dentry->d_parent->d_inode->i_ino;
2573 spin_unlock(&dentry->d_lock);
2577 /* Transaction based IO helpers */
2580 * An argresp is stored in an allocated page and holds the
2581 * size of the argument or response, along with its content
2583 struct simple_transaction_argresp {
2588 #define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
2590 char *simple_transaction_get(struct file *file, const char __user *buf,
2592 ssize_t simple_transaction_read(struct file *file, char __user *buf,
2593 size_t size, loff_t *pos);
2594 int simple_transaction_release(struct inode *inode, struct file *file);
2596 void simple_transaction_set(struct file *file, size_t n);
2599 * simple attribute files
2601 * These attributes behave similar to those in sysfs:
2603 * Writing to an attribute immediately sets a value, an open file can be
2604 * written to multiple times.
2606 * Reading from an attribute creates a buffer from the value that might get
2607 * read with multiple read calls. When the attribute has been read
2608 * completely, no further read calls are possible until the file is opened
2611 * All attributes contain a text representation of a numeric value
2612 * that are accessed with the get() and set() functions.
2614 #define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt) \
2615 static int __fops ## _open(struct inode *inode, struct file *file) \
2617 __simple_attr_check_format(__fmt, 0ull); \
2618 return simple_attr_open(inode, file, __get, __set, __fmt); \
2620 static const struct file_operations __fops = { \
2621 .owner = THIS_MODULE, \
2622 .open = __fops ## _open, \
2623 .release = simple_attr_release, \
2624 .read = simple_attr_read, \
2625 .write = simple_attr_write, \
2626 .llseek = generic_file_llseek, \
2629 static inline __printf(1, 2)
2630 void __simple_attr_check_format(const char *fmt, ...)
2632 /* don't do anything, just let the compiler check the arguments; */
2635 int simple_attr_open(struct inode *inode, struct file *file,
2636 int (*get)(void *, u64 *), int (*set)(void *, u64),
2638 int simple_attr_release(struct inode *inode, struct file *file);
2639 ssize_t simple_attr_read(struct file *file, char __user *buf,
2640 size_t len, loff_t *ppos);
2641 ssize_t simple_attr_write(struct file *file, const char __user *buf,
2642 size_t len, loff_t *ppos);
2645 int proc_nr_files(struct ctl_table *table, int write,
2646 void __user *buffer, size_t *lenp, loff_t *ppos);
2647 int proc_nr_dentry(struct ctl_table *table, int write,
2648 void __user *buffer, size_t *lenp, loff_t *ppos);
2649 int proc_nr_inodes(struct ctl_table *table, int write,
2650 void __user *buffer, size_t *lenp, loff_t *ppos);
2651 int __init get_filesystem_list(char *buf);
2653 #define __FMODE_EXEC ((__force int) FMODE_EXEC)
2654 #define __FMODE_NONOTIFY ((__force int) FMODE_NONOTIFY)
2656 #define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE])
2657 #define OPEN_FMODE(flag) ((__force fmode_t)(((flag + 1) & O_ACCMODE) | \
2658 (flag & __FMODE_NONOTIFY)))
2660 static inline int is_sxid(umode_t mode)
2662 return (mode & S_ISUID) || ((mode & S_ISGID) && (mode & S_IXGRP));
2665 static inline void inode_has_no_xattr(struct inode *inode)
2667 if (!is_sxid(inode->i_mode) && (inode->i_sb->s_flags & MS_NOSEC))
2668 inode->i_flags |= S_NOSEC;
2671 #endif /* _LINUX_FS_H */