1 /* SPDX-License-Identifier: GPL-2.0 */
5 #include <linux/linkage.h>
6 #include <linux/wait_bit.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/xarray.h>
17 #include <linux/rbtree.h>
18 #include <linux/init.h>
19 #include <linux/pid.h>
20 #include <linux/bug.h>
21 #include <linux/mutex.h>
22 #include <linux/rwsem.h>
23 #include <linux/mm_types.h>
24 #include <linux/capability.h>
25 #include <linux/semaphore.h>
26 #include <linux/fcntl.h>
27 #include <linux/fiemap.h>
28 #include <linux/rculist_bl.h>
29 #include <linux/atomic.h>
30 #include <linux/shrinker.h>
31 #include <linux/migrate_mode.h>
32 #include <linux/uidgid.h>
33 #include <linux/lockdep.h>
34 #include <linux/percpu-rwsem.h>
35 #include <linux/workqueue.h>
36 #include <linux/delayed_call.h>
37 #include <linux/uuid.h>
38 #include <linux/errseq.h>
39 #include <linux/ioprio.h>
40 #include <linux/fs_types.h>
42 #include <asm/byteorder.h>
43 #include <uapi/linux/fs.h>
45 struct backing_dev_info;
48 struct export_operations;
53 struct pipe_inode_info;
54 struct poll_table_struct;
56 struct vm_area_struct;
59 struct swap_info_struct;
61 struct workqueue_struct;
64 struct fscrypt_operations;
66 extern void __init inode_init(void);
67 extern void __init inode_init_early(void);
68 extern void __init files_init(void);
69 extern void __init files_maxfiles_init(void);
71 extern struct files_stat_struct files_stat;
72 extern unsigned long get_max_files(void);
73 extern unsigned int sysctl_nr_open;
74 extern struct inodes_stat_t inodes_stat;
75 extern int leases_enable, lease_break_time;
76 extern int sysctl_protected_symlinks;
77 extern int sysctl_protected_hardlinks;
78 extern int sysctl_protected_fifos;
79 extern int sysctl_protected_regular;
81 typedef __kernel_rwf_t rwf_t;
84 typedef int (get_block_t)(struct inode *inode, sector_t iblock,
85 struct buffer_head *bh_result, int create);
86 typedef int (dio_iodone_t)(struct kiocb *iocb, loff_t offset,
87 ssize_t bytes, void *private);
89 #define MAY_EXEC 0x00000001
90 #define MAY_WRITE 0x00000002
91 #define MAY_READ 0x00000004
92 #define MAY_APPEND 0x00000008
93 #define MAY_ACCESS 0x00000010
94 #define MAY_OPEN 0x00000020
95 #define MAY_CHDIR 0x00000040
96 /* called from RCU mode, don't block */
97 #define MAY_NOT_BLOCK 0x00000080
100 * flags in file.f_mode. Note that FMODE_READ and FMODE_WRITE must correspond
101 * to O_WRONLY and O_RDWR via the strange trick in do_dentry_open()
104 /* file is open for reading */
105 #define FMODE_READ ((__force fmode_t)0x1)
106 /* file is open for writing */
107 #define FMODE_WRITE ((__force fmode_t)0x2)
108 /* file is seekable */
109 #define FMODE_LSEEK ((__force fmode_t)0x4)
110 /* file can be accessed using pread */
111 #define FMODE_PREAD ((__force fmode_t)0x8)
112 /* file can be accessed using pwrite */
113 #define FMODE_PWRITE ((__force fmode_t)0x10)
114 /* File is opened for execution with sys_execve / sys_uselib */
115 #define FMODE_EXEC ((__force fmode_t)0x20)
116 /* File is opened with O_NDELAY (only set for block devices) */
117 #define FMODE_NDELAY ((__force fmode_t)0x40)
118 /* File is opened with O_EXCL (only set for block devices) */
119 #define FMODE_EXCL ((__force fmode_t)0x80)
120 /* File is opened using open(.., 3, ..) and is writeable only for ioctls
121 (specialy hack for floppy.c) */
122 #define FMODE_WRITE_IOCTL ((__force fmode_t)0x100)
123 /* 32bit hashes as llseek() offset (for directories) */
124 #define FMODE_32BITHASH ((__force fmode_t)0x200)
125 /* 64bit hashes as llseek() offset (for directories) */
126 #define FMODE_64BITHASH ((__force fmode_t)0x400)
129 * Don't update ctime and mtime.
131 * Currently a special hack for the XFS open_by_handle ioctl, but we'll
132 * hopefully graduate it to a proper O_CMTIME flag supported by open(2) soon.
134 #define FMODE_NOCMTIME ((__force fmode_t)0x800)
136 /* Expect random access pattern */
137 #define FMODE_RANDOM ((__force fmode_t)0x1000)
139 /* File is huge (eg. /dev/kmem): treat loff_t as unsigned */
140 #define FMODE_UNSIGNED_OFFSET ((__force fmode_t)0x2000)
142 /* File is opened with O_PATH; almost nothing can be done with it */
143 #define FMODE_PATH ((__force fmode_t)0x4000)
145 /* File needs atomic accesses to f_pos */
146 #define FMODE_ATOMIC_POS ((__force fmode_t)0x8000)
147 /* Write access to underlying fs */
148 #define FMODE_WRITER ((__force fmode_t)0x10000)
149 /* Has read method(s) */
150 #define FMODE_CAN_READ ((__force fmode_t)0x20000)
151 /* Has write method(s) */
152 #define FMODE_CAN_WRITE ((__force fmode_t)0x40000)
154 #define FMODE_OPENED ((__force fmode_t)0x80000)
155 #define FMODE_CREATED ((__force fmode_t)0x100000)
157 /* File was opened by fanotify and shouldn't generate fanotify events */
158 #define FMODE_NONOTIFY ((__force fmode_t)0x4000000)
160 /* File is capable of returning -EAGAIN if I/O will block */
161 #define FMODE_NOWAIT ((__force fmode_t)0x8000000)
163 /* File does not contribute to nr_files count */
164 #define FMODE_NOACCOUNT ((__force fmode_t)0x20000000)
167 * Flag for rw_copy_check_uvector and compat_rw_copy_check_uvector
168 * that indicates that they should check the contents of the iovec are
169 * valid, but not check the memory that the iovec elements
172 #define CHECK_IOVEC_ONLY -1
175 * Attribute flags. These should be or-ed together to figure out what
178 #define ATTR_MODE (1 << 0)
179 #define ATTR_UID (1 << 1)
180 #define ATTR_GID (1 << 2)
181 #define ATTR_SIZE (1 << 3)
182 #define ATTR_ATIME (1 << 4)
183 #define ATTR_MTIME (1 << 5)
184 #define ATTR_CTIME (1 << 6)
185 #define ATTR_ATIME_SET (1 << 7)
186 #define ATTR_MTIME_SET (1 << 8)
187 #define ATTR_FORCE (1 << 9) /* Not a change, but a change it */
188 #define ATTR_KILL_SUID (1 << 11)
189 #define ATTR_KILL_SGID (1 << 12)
190 #define ATTR_FILE (1 << 13)
191 #define ATTR_KILL_PRIV (1 << 14)
192 #define ATTR_OPEN (1 << 15) /* Truncating from open(O_TRUNC) */
193 #define ATTR_TIMES_SET (1 << 16)
194 #define ATTR_TOUCH (1 << 17)
197 * Whiteout is represented by a char device. The following constants define the
198 * mode and device number to use.
200 #define WHITEOUT_MODE 0
201 #define WHITEOUT_DEV 0
204 * This is the Inode Attributes structure, used for notify_change(). It
205 * uses the above definitions as flags, to know which values have changed.
206 * Also, in this manner, a Filesystem can look at only the values it cares
207 * about. Basically, these are the attributes that the VFS layer can
208 * request to change from the FS layer.
210 * Derek Atkins <warlord@MIT.EDU> 94-10-20
213 unsigned int ia_valid;
218 struct timespec64 ia_atime;
219 struct timespec64 ia_mtime;
220 struct timespec64 ia_ctime;
223 * Not an attribute, but an auxiliary info for filesystems wanting to
224 * implement an ftruncate() like method. NOTE: filesystem should
225 * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL).
227 struct file *ia_file;
231 * Includes for diskquotas.
233 #include <linux/quota.h>
236 * Maximum number of layers of fs stack. Needs to be limited to
237 * prevent kernel stack overflow
239 #define FILESYSTEM_MAX_STACK_DEPTH 2
242 * enum positive_aop_returns - aop return codes with specific semantics
244 * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has
245 * completed, that the page is still locked, and
246 * should be considered active. The VM uses this hint
247 * to return the page to the active list -- it won't
248 * be a candidate for writeback again in the near
249 * future. Other callers must be careful to unlock
250 * the page if they get this return. Returned by
253 * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has
254 * unlocked it and the page might have been truncated.
255 * The caller should back up to acquiring a new page and
256 * trying again. The aop will be taking reasonable
257 * precautions not to livelock. If the caller held a page
258 * reference, it should drop it before retrying. Returned
261 * address_space_operation functions return these large constants to indicate
262 * special semantics to the caller. These are much larger than the bytes in a
263 * page to allow for functions that return the number of bytes operated on in a
267 enum positive_aop_returns {
268 AOP_WRITEPAGE_ACTIVATE = 0x80000,
269 AOP_TRUNCATED_PAGE = 0x80001,
272 #define AOP_FLAG_CONT_EXPAND 0x0001 /* called from cont_expand */
273 #define AOP_FLAG_NOFS 0x0002 /* used by filesystem to direct
274 * helper code (eg buffer layer)
275 * to clear GFP_FS from alloc */
278 * oh the beauties of C type declarations.
281 struct address_space;
282 struct writeback_control;
285 * Write life time hint values.
286 * Stored in struct inode as u8.
289 WRITE_LIFE_NOT_SET = 0,
290 WRITE_LIFE_NONE = RWH_WRITE_LIFE_NONE,
291 WRITE_LIFE_SHORT = RWH_WRITE_LIFE_SHORT,
292 WRITE_LIFE_MEDIUM = RWH_WRITE_LIFE_MEDIUM,
293 WRITE_LIFE_LONG = RWH_WRITE_LIFE_LONG,
294 WRITE_LIFE_EXTREME = RWH_WRITE_LIFE_EXTREME,
297 #define IOCB_EVENTFD (1 << 0)
298 #define IOCB_APPEND (1 << 1)
299 #define IOCB_DIRECT (1 << 2)
300 #define IOCB_HIPRI (1 << 3)
301 #define IOCB_DSYNC (1 << 4)
302 #define IOCB_SYNC (1 << 5)
303 #define IOCB_WRITE (1 << 6)
304 #define IOCB_NOWAIT (1 << 7)
307 struct file *ki_filp;
309 /* The 'ki_filp' pointer is shared in a union for aio */
310 randomized_struct_fields_start
313 void (*ki_complete)(struct kiocb *iocb, long ret, long ret2);
317 u16 ki_ioprio; /* See linux/ioprio.h */
319 randomized_struct_fields_end
322 static inline bool is_sync_kiocb(struct kiocb *kiocb)
324 return kiocb->ki_complete == NULL;
328 * "descriptor" for what we're up to with a read.
329 * This allows us to use the same read code yet
330 * have multiple different users of the data that
331 * we read from a file.
333 * The simplest case just copies the data to user
346 typedef int (*read_actor_t)(read_descriptor_t *, struct page *,
347 unsigned long, unsigned long);
349 struct address_space_operations {
350 int (*writepage)(struct page *page, struct writeback_control *wbc);
351 int (*readpage)(struct file *, struct page *);
353 /* Write back some dirty pages from this mapping. */
354 int (*writepages)(struct address_space *, struct writeback_control *);
356 /* Set a page dirty. Return true if this dirtied it */
357 int (*set_page_dirty)(struct page *page);
360 * Reads in the requested pages. Unlike ->readpage(), this is
361 * PURELY used for read-ahead!.
363 int (*readpages)(struct file *filp, struct address_space *mapping,
364 struct list_head *pages, unsigned nr_pages);
366 int (*write_begin)(struct file *, struct address_space *mapping,
367 loff_t pos, unsigned len, unsigned flags,
368 struct page **pagep, void **fsdata);
369 int (*write_end)(struct file *, struct address_space *mapping,
370 loff_t pos, unsigned len, unsigned copied,
371 struct page *page, void *fsdata);
373 /* Unfortunately this kludge is needed for FIBMAP. Don't use it */
374 sector_t (*bmap)(struct address_space *, sector_t);
375 void (*invalidatepage) (struct page *, unsigned int, unsigned int);
376 int (*releasepage) (struct page *, gfp_t);
377 void (*freepage)(struct page *);
378 ssize_t (*direct_IO)(struct kiocb *, struct iov_iter *iter);
380 * migrate the contents of a page to the specified target. If
381 * migrate_mode is MIGRATE_ASYNC, it must not block.
383 int (*migratepage) (struct address_space *,
384 struct page *, struct page *, enum migrate_mode);
385 bool (*isolate_page)(struct page *, isolate_mode_t);
386 void (*putback_page)(struct page *);
387 int (*launder_page) (struct page *);
388 int (*is_partially_uptodate) (struct page *, unsigned long,
390 void (*is_dirty_writeback) (struct page *, bool *, bool *);
391 int (*error_remove_page)(struct address_space *, struct page *);
393 /* swapfile support */
394 int (*swap_activate)(struct swap_info_struct *sis, struct file *file,
396 void (*swap_deactivate)(struct file *file);
399 extern const struct address_space_operations empty_aops;
402 * pagecache_write_begin/pagecache_write_end must be used by general code
403 * to write into the pagecache.
405 int pagecache_write_begin(struct file *, struct address_space *mapping,
406 loff_t pos, unsigned len, unsigned flags,
407 struct page **pagep, void **fsdata);
409 int pagecache_write_end(struct file *, struct address_space *mapping,
410 loff_t pos, unsigned len, unsigned copied,
411 struct page *page, void *fsdata);
414 * struct address_space - Contents of a cacheable, mappable object.
415 * @host: Owner, either the inode or the block_device.
416 * @i_pages: Cached pages.
417 * @gfp_mask: Memory allocation flags to use for allocating pages.
418 * @i_mmap_writable: Number of VM_SHARED mappings.
419 * @i_mmap: Tree of private and shared mappings.
420 * @i_mmap_rwsem: Protects @i_mmap and @i_mmap_writable.
421 * @nrpages: Number of page entries, protected by the i_pages lock.
422 * @nrexceptional: Shadow or DAX entries, protected by the i_pages lock.
423 * @writeback_index: Writeback starts here.
425 * @flags: Error bits and flags (AS_*).
426 * @wb_err: The most recent error which has occurred.
427 * @private_lock: For use by the owner of the address_space.
428 * @private_list: For use by the owner of the address_space.
429 * @private_data: For use by the owner of the address_space.
431 struct address_space {
433 struct xarray i_pages;
435 atomic_t i_mmap_writable;
436 struct rb_root_cached i_mmap;
437 struct rw_semaphore i_mmap_rwsem;
438 unsigned long nrpages;
439 unsigned long nrexceptional;
440 pgoff_t writeback_index;
441 const struct address_space_operations *a_ops;
444 spinlock_t private_lock;
445 struct list_head private_list;
447 } __attribute__((aligned(sizeof(long)))) __randomize_layout;
449 * On most architectures that alignment is already the case; but
450 * must be enforced here for CRIS, to let the least significant bit
451 * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
453 struct request_queue;
455 struct block_device {
456 dev_t bd_dev; /* not a kdev_t - it's a search key */
458 struct inode * bd_inode; /* will die */
459 struct super_block * bd_super;
460 struct mutex bd_mutex; /* open/close mutex */
464 bool bd_write_holder;
466 struct list_head bd_holder_disks;
468 struct block_device * bd_contains;
469 unsigned bd_block_size;
471 struct hd_struct * bd_part;
472 /* number of times partitions within this device have been opened. */
473 unsigned bd_part_count;
475 struct gendisk * bd_disk;
476 struct request_queue * bd_queue;
477 struct backing_dev_info *bd_bdi;
478 struct list_head bd_list;
480 * Private data. You must have bd_claim'ed the block_device
481 * to use this. NOTE: bd_claim allows an owner to claim
482 * the same device multiple times, the owner must take special
483 * care to not mess up bd_private for that case.
485 unsigned long bd_private;
487 /* The counter of freeze processes */
488 int bd_fsfreeze_count;
489 /* Mutex for freeze */
490 struct mutex bd_fsfreeze_mutex;
491 } __randomize_layout;
493 /* XArray tags, for tagging dirty and writeback pages in the pagecache. */
494 #define PAGECACHE_TAG_DIRTY XA_MARK_0
495 #define PAGECACHE_TAG_WRITEBACK XA_MARK_1
496 #define PAGECACHE_TAG_TOWRITE XA_MARK_2
499 * Returns true if any of the pages in the mapping are marked with the tag.
501 static inline bool mapping_tagged(struct address_space *mapping, xa_mark_t tag)
503 return xa_marked(&mapping->i_pages, tag);
506 static inline void i_mmap_lock_write(struct address_space *mapping)
508 down_write(&mapping->i_mmap_rwsem);
511 static inline void i_mmap_unlock_write(struct address_space *mapping)
513 up_write(&mapping->i_mmap_rwsem);
516 static inline void i_mmap_lock_read(struct address_space *mapping)
518 down_read(&mapping->i_mmap_rwsem);
521 static inline void i_mmap_unlock_read(struct address_space *mapping)
523 up_read(&mapping->i_mmap_rwsem);
527 * Might pages of this file be mapped into userspace?
529 static inline int mapping_mapped(struct address_space *mapping)
531 return !RB_EMPTY_ROOT(&mapping->i_mmap.rb_root);
535 * Might pages of this file have been modified in userspace?
536 * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap_pgoff
537 * marks vma as VM_SHARED if it is shared, and the file was opened for
538 * writing i.e. vma may be mprotected writable even if now readonly.
540 * If i_mmap_writable is negative, no new writable mappings are allowed. You
541 * can only deny writable mappings, if none exists right now.
543 static inline int mapping_writably_mapped(struct address_space *mapping)
545 return atomic_read(&mapping->i_mmap_writable) > 0;
548 static inline int mapping_map_writable(struct address_space *mapping)
550 return atomic_inc_unless_negative(&mapping->i_mmap_writable) ?
554 static inline void mapping_unmap_writable(struct address_space *mapping)
556 atomic_dec(&mapping->i_mmap_writable);
559 static inline int mapping_deny_writable(struct address_space *mapping)
561 return atomic_dec_unless_positive(&mapping->i_mmap_writable) ?
565 static inline void mapping_allow_writable(struct address_space *mapping)
567 atomic_inc(&mapping->i_mmap_writable);
571 * Use sequence counter to get consistent i_size on 32-bit processors.
573 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
574 #include <linux/seqlock.h>
575 #define __NEED_I_SIZE_ORDERED
576 #define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
578 #define i_size_ordered_init(inode) do { } while (0)
582 #define ACL_NOT_CACHED ((void *)(-1))
583 #define ACL_DONT_CACHE ((void *)(-3))
585 static inline struct posix_acl *
586 uncached_acl_sentinel(struct task_struct *task)
588 return (void *)task + 1;
592 is_uncached_acl(struct posix_acl *acl)
594 return (long)acl & 1;
597 #define IOP_FASTPERM 0x0001
598 #define IOP_LOOKUP 0x0002
599 #define IOP_NOFOLLOW 0x0004
600 #define IOP_XATTR 0x0008
601 #define IOP_DEFAULT_READLINK 0x0010
603 struct fsnotify_mark_connector;
606 * Keep mostly read-only and often accessed (especially for
607 * the RCU path lookup and 'stat' data) fields at the beginning
608 * of the 'struct inode'
612 unsigned short i_opflags;
615 unsigned int i_flags;
617 #ifdef CONFIG_FS_POSIX_ACL
618 struct posix_acl *i_acl;
619 struct posix_acl *i_default_acl;
622 const struct inode_operations *i_op;
623 struct super_block *i_sb;
624 struct address_space *i_mapping;
626 #ifdef CONFIG_SECURITY
630 /* Stat data, not accessed from path walking */
633 * Filesystems may only read i_nlink directly. They shall use the
634 * following functions for modification:
636 * (set|clear|inc|drop)_nlink
637 * inode_(inc|dec)_link_count
640 const unsigned int i_nlink;
641 unsigned int __i_nlink;
645 struct timespec64 i_atime;
646 struct timespec64 i_mtime;
647 struct timespec64 i_ctime;
648 spinlock_t i_lock; /* i_blocks, i_bytes, maybe i_size */
649 unsigned short i_bytes;
654 #ifdef __NEED_I_SIZE_ORDERED
655 seqcount_t i_size_seqcount;
659 unsigned long i_state;
660 struct rw_semaphore i_rwsem;
662 unsigned long dirtied_when; /* jiffies of first dirtying */
663 unsigned long dirtied_time_when;
665 struct hlist_node i_hash;
666 struct list_head i_io_list; /* backing dev IO list */
667 #ifdef CONFIG_CGROUP_WRITEBACK
668 struct bdi_writeback *i_wb; /* the associated cgroup wb */
670 /* foreign inode detection, see wbc_detach_inode() */
672 u16 i_wb_frn_avg_time;
673 u16 i_wb_frn_history;
675 struct list_head i_lru; /* inode LRU list */
676 struct list_head i_sb_list;
677 struct list_head i_wb_list; /* backing dev writeback list */
679 struct hlist_head i_dentry;
680 struct rcu_head i_rcu;
682 atomic64_t i_version;
684 atomic_t i_dio_count;
685 atomic_t i_writecount;
687 atomic_t i_readcount; /* struct files open RO */
689 const struct file_operations *i_fop; /* former ->i_op->default_file_ops */
690 struct file_lock_context *i_flctx;
691 struct address_space i_data;
692 struct list_head i_devices;
694 struct pipe_inode_info *i_pipe;
695 struct block_device *i_bdev;
703 #ifdef CONFIG_FSNOTIFY
704 __u32 i_fsnotify_mask; /* all events this inode cares about */
705 struct fsnotify_mark_connector __rcu *i_fsnotify_marks;
708 #if IS_ENABLED(CONFIG_FS_ENCRYPTION)
709 struct fscrypt_info *i_crypt_info;
712 void *i_private; /* fs or device private pointer */
713 } __randomize_layout;
715 static inline unsigned int i_blocksize(const struct inode *node)
717 return (1 << node->i_blkbits);
720 static inline int inode_unhashed(struct inode *inode)
722 return hlist_unhashed(&inode->i_hash);
726 * __mark_inode_dirty expects inodes to be hashed. Since we don't
727 * want special inodes in the fileset inode space, we make them
728 * appear hashed, but do not put on any lists. hlist_del()
729 * will work fine and require no locking.
731 static inline void inode_fake_hash(struct inode *inode)
733 hlist_add_fake(&inode->i_hash);
737 * inode->i_mutex nesting subclasses for the lock validator:
739 * 0: the object of the current VFS operation
743 * 4: second non-directory
744 * 5: second parent (when locking independent directories in rename)
746 * I_MUTEX_NONDIR2 is for certain operations (such as rename) which lock two
747 * non-directories at once.
749 * The locking order between these classes is
750 * parent[2] -> child -> grandchild -> normal -> xattr -> second non-directory
752 enum inode_i_mutex_lock_class
762 static inline void inode_lock(struct inode *inode)
764 down_write(&inode->i_rwsem);
767 static inline void inode_unlock(struct inode *inode)
769 up_write(&inode->i_rwsem);
772 static inline void inode_lock_shared(struct inode *inode)
774 down_read(&inode->i_rwsem);
777 static inline void inode_unlock_shared(struct inode *inode)
779 up_read(&inode->i_rwsem);
782 static inline int inode_trylock(struct inode *inode)
784 return down_write_trylock(&inode->i_rwsem);
787 static inline int inode_trylock_shared(struct inode *inode)
789 return down_read_trylock(&inode->i_rwsem);
792 static inline int inode_is_locked(struct inode *inode)
794 return rwsem_is_locked(&inode->i_rwsem);
797 static inline void inode_lock_nested(struct inode *inode, unsigned subclass)
799 down_write_nested(&inode->i_rwsem, subclass);
802 static inline void inode_lock_shared_nested(struct inode *inode, unsigned subclass)
804 down_read_nested(&inode->i_rwsem, subclass);
807 void lock_two_nondirectories(struct inode *, struct inode*);
808 void unlock_two_nondirectories(struct inode *, struct inode*);
811 * NOTE: in a 32bit arch with a preemptable kernel and
812 * an UP compile the i_size_read/write must be atomic
813 * with respect to the local cpu (unlike with preempt disabled),
814 * but they don't need to be atomic with respect to other cpus like in
815 * true SMP (so they need either to either locally disable irq around
816 * the read or for example on x86 they can be still implemented as a
817 * cmpxchg8b without the need of the lock prefix). For SMP compiles
818 * and 64bit archs it makes no difference if preempt is enabled or not.
820 static inline loff_t i_size_read(const struct inode *inode)
822 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
827 seq = read_seqcount_begin(&inode->i_size_seqcount);
828 i_size = inode->i_size;
829 } while (read_seqcount_retry(&inode->i_size_seqcount, seq));
831 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
835 i_size = inode->i_size;
839 return inode->i_size;
844 * NOTE: unlike i_size_read(), i_size_write() does need locking around it
845 * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount
846 * can be lost, resulting in subsequent i_size_read() calls spinning forever.
848 static inline void i_size_write(struct inode *inode, loff_t i_size)
850 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
852 write_seqcount_begin(&inode->i_size_seqcount);
853 inode->i_size = i_size;
854 write_seqcount_end(&inode->i_size_seqcount);
856 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
858 inode->i_size = i_size;
861 inode->i_size = i_size;
865 static inline unsigned iminor(const struct inode *inode)
867 return MINOR(inode->i_rdev);
870 static inline unsigned imajor(const struct inode *inode)
872 return MAJOR(inode->i_rdev);
875 extern struct block_device *I_BDEV(struct inode *inode);
878 rwlock_t lock; /* protects pid, uid, euid fields */
879 struct pid *pid; /* pid or -pgrp where SIGIO should be sent */
880 enum pid_type pid_type; /* Kind of process group SIGIO should be sent to */
881 kuid_t uid, euid; /* uid/euid of process setting the owner */
882 int signum; /* posix.1b rt signal to be delivered on IO */
886 * Track a single file's readahead state
888 struct file_ra_state {
889 pgoff_t start; /* where readahead started */
890 unsigned int size; /* # of readahead pages */
891 unsigned int async_size; /* do asynchronous readahead when
892 there are only # of pages ahead */
894 unsigned int ra_pages; /* Maximum readahead window */
895 unsigned int mmap_miss; /* Cache miss stat for mmap accesses */
896 loff_t prev_pos; /* Cache last read() position */
900 * Check if @index falls in the readahead windows.
902 static inline int ra_has_index(struct file_ra_state *ra, pgoff_t index)
904 return (index >= ra->start &&
905 index < ra->start + ra->size);
910 struct llist_node fu_llist;
911 struct rcu_head fu_rcuhead;
914 struct inode *f_inode; /* cached value */
915 const struct file_operations *f_op;
918 * Protects f_ep_links, f_flags.
919 * Must not be taken from IRQ context.
922 enum rw_hint f_write_hint;
923 atomic_long_t f_count;
924 unsigned int f_flags;
926 struct mutex f_pos_lock;
928 struct fown_struct f_owner;
929 const struct cred *f_cred;
930 struct file_ra_state f_ra;
933 #ifdef CONFIG_SECURITY
936 /* needed for tty driver, and maybe others */
940 /* Used by fs/eventpoll.c to link all the hooks to this file */
941 struct list_head f_ep_links;
942 struct list_head f_tfile_llink;
943 #endif /* #ifdef CONFIG_EPOLL */
944 struct address_space *f_mapping;
947 __attribute__((aligned(4))); /* lest something weird decides that 2 is OK */
952 /* file identifier */
953 unsigned char f_handle[0];
956 static inline struct file *get_file(struct file *f)
958 atomic_long_inc(&f->f_count);
961 #define get_file_rcu(x) atomic_long_inc_not_zero(&(x)->f_count)
962 #define fput_atomic(x) atomic_long_add_unless(&(x)->f_count, -1, 1)
963 #define file_count(x) atomic_long_read(&(x)->f_count)
965 #define MAX_NON_LFS ((1UL<<31) - 1)
967 /* Page cache limit. The filesystems should put that into their s_maxbytes
968 limits, otherwise bad things can happen in VM. */
969 #if BITS_PER_LONG==32
970 #define MAX_LFS_FILESIZE ((loff_t)ULONG_MAX << PAGE_SHIFT)
971 #elif BITS_PER_LONG==64
972 #define MAX_LFS_FILESIZE ((loff_t)LLONG_MAX)
977 #define FL_DELEG 4 /* NFSv4 delegation */
978 #define FL_ACCESS 8 /* not trying to lock, just looking */
979 #define FL_EXISTS 16 /* when unlocking, test for existence */
980 #define FL_LEASE 32 /* lease held on this file */
981 #define FL_CLOSE 64 /* unlock on close */
982 #define FL_SLEEP 128 /* A blocking lock */
983 #define FL_DOWNGRADE_PENDING 256 /* Lease is being downgraded */
984 #define FL_UNLOCK_PENDING 512 /* Lease is being broken */
985 #define FL_OFDLCK 1024 /* lock is "owned" by struct file */
986 #define FL_LAYOUT 2048 /* outstanding pNFS layout */
988 #define FL_CLOSE_POSIX (FL_POSIX | FL_CLOSE)
991 * Special return value from posix_lock_file() and vfs_lock_file() for
992 * asynchronous locking.
994 #define FILE_LOCK_DEFERRED 1
996 /* legacy typedef, should eventually be removed */
997 typedef void *fl_owner_t;
1001 struct file_lock_operations {
1002 void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
1003 void (*fl_release_private)(struct file_lock *);
1006 struct lock_manager_operations {
1007 int (*lm_compare_owner)(struct file_lock *, struct file_lock *);
1008 unsigned long (*lm_owner_key)(struct file_lock *);
1009 fl_owner_t (*lm_get_owner)(fl_owner_t);
1010 void (*lm_put_owner)(fl_owner_t);
1011 void (*lm_notify)(struct file_lock *); /* unblock callback */
1012 int (*lm_grant)(struct file_lock *, int);
1013 bool (*lm_break)(struct file_lock *);
1014 int (*lm_change)(struct file_lock *, int, struct list_head *);
1015 void (*lm_setup)(struct file_lock *, void **);
1018 struct lock_manager {
1019 struct list_head list;
1021 * NFSv4 and up also want opens blocked during the grace period;
1028 void locks_start_grace(struct net *, struct lock_manager *);
1029 void locks_end_grace(struct lock_manager *);
1030 bool locks_in_grace(struct net *);
1031 bool opens_in_grace(struct net *);
1033 /* that will die - we need it for nfs_lock_info */
1034 #include <linux/nfs_fs_i.h>
1037 * struct file_lock represents a generic "file lock". It's used to represent
1038 * POSIX byte range locks, BSD (flock) locks, and leases. It's important to
1039 * note that the same struct is used to represent both a request for a lock and
1040 * the lock itself, but the same object is never used for both.
1042 * FIXME: should we create a separate "struct lock_request" to help distinguish
1045 * The varous i_flctx lists are ordered by:
1048 * 2) lock range start
1051 * Obviously, the last two criteria only matter for POSIX locks.
1054 struct file_lock *fl_blocker; /* The lock, that is blocking us */
1055 struct list_head fl_list; /* link into file_lock_context */
1056 struct hlist_node fl_link; /* node in global lists */
1057 struct list_head fl_blocked_requests; /* list of requests with
1058 * ->fl_blocker pointing here
1060 struct list_head fl_blocked_member; /* node in
1061 * ->fl_blocker->fl_blocked_requests
1063 fl_owner_t fl_owner;
1064 unsigned int fl_flags;
1065 unsigned char fl_type;
1066 unsigned int fl_pid;
1067 int fl_link_cpu; /* what cpu's list is this on? */
1068 wait_queue_head_t fl_wait;
1069 struct file *fl_file;
1073 struct fasync_struct * fl_fasync; /* for lease break notifications */
1074 /* for lease breaks: */
1075 unsigned long fl_break_time;
1076 unsigned long fl_downgrade_time;
1078 const struct file_lock_operations *fl_ops; /* Callbacks for filesystems */
1079 const struct lock_manager_operations *fl_lmops; /* Callbacks for lockmanagers */
1081 struct nfs_lock_info nfs_fl;
1082 struct nfs4_lock_info nfs4_fl;
1084 struct list_head link; /* link in AFS vnode's pending_locks list */
1085 int state; /* state of grant or error if -ve */
1088 } __randomize_layout;
1090 struct file_lock_context {
1091 spinlock_t flc_lock;
1092 struct list_head flc_flock;
1093 struct list_head flc_posix;
1094 struct list_head flc_lease;
1097 /* The following constant reflects the upper bound of the file/locking space */
1099 #define INT_LIMIT(x) (~((x)1 << (sizeof(x)*8 - 1)))
1100 #define OFFSET_MAX INT_LIMIT(loff_t)
1101 #define OFFT_OFFSET_MAX INT_LIMIT(off_t)
1104 extern void send_sigio(struct fown_struct *fown, int fd, int band);
1106 #define locks_inode(f) file_inode(f)
1108 #ifdef CONFIG_FILE_LOCKING
1109 extern int fcntl_getlk(struct file *, unsigned int, struct flock *);
1110 extern int fcntl_setlk(unsigned int, struct file *, unsigned int,
1113 #if BITS_PER_LONG == 32
1114 extern int fcntl_getlk64(struct file *, unsigned int, struct flock64 *);
1115 extern int fcntl_setlk64(unsigned int, struct file *, unsigned int,
1119 extern int fcntl_setlease(unsigned int fd, struct file *filp, long arg);
1120 extern int fcntl_getlease(struct file *filp);
1123 void locks_free_lock_context(struct inode *inode);
1124 void locks_free_lock(struct file_lock *fl);
1125 extern void locks_init_lock(struct file_lock *);
1126 extern struct file_lock * locks_alloc_lock(void);
1127 extern void locks_copy_lock(struct file_lock *, struct file_lock *);
1128 extern void locks_copy_conflock(struct file_lock *, struct file_lock *);
1129 extern void locks_remove_posix(struct file *, fl_owner_t);
1130 extern void locks_remove_file(struct file *);
1131 extern void locks_release_private(struct file_lock *);
1132 extern void posix_test_lock(struct file *, struct file_lock *);
1133 extern int posix_lock_file(struct file *, struct file_lock *, struct file_lock *);
1134 extern int locks_delete_block(struct file_lock *);
1135 extern int vfs_test_lock(struct file *, struct file_lock *);
1136 extern int vfs_lock_file(struct file *, unsigned int, struct file_lock *, struct file_lock *);
1137 extern int vfs_cancel_lock(struct file *filp, struct file_lock *fl);
1138 extern int locks_lock_inode_wait(struct inode *inode, struct file_lock *fl);
1139 extern int __break_lease(struct inode *inode, unsigned int flags, unsigned int type);
1140 extern void lease_get_mtime(struct inode *, struct timespec64 *time);
1141 extern int generic_setlease(struct file *, long, struct file_lock **, void **priv);
1142 extern int vfs_setlease(struct file *, long, struct file_lock **, void **);
1143 extern int lease_modify(struct file_lock *, int, struct list_head *);
1144 struct files_struct;
1145 extern void show_fd_locks(struct seq_file *f,
1146 struct file *filp, struct files_struct *files);
1147 #else /* !CONFIG_FILE_LOCKING */
1148 static inline int fcntl_getlk(struct file *file, unsigned int cmd,
1149 struct flock __user *user)
1154 static inline int fcntl_setlk(unsigned int fd, struct file *file,
1155 unsigned int cmd, struct flock __user *user)
1160 #if BITS_PER_LONG == 32
1161 static inline int fcntl_getlk64(struct file *file, unsigned int cmd,
1162 struct flock64 __user *user)
1167 static inline int fcntl_setlk64(unsigned int fd, struct file *file,
1168 unsigned int cmd, struct flock64 __user *user)
1173 static inline int fcntl_setlease(unsigned int fd, struct file *filp, long arg)
1178 static inline int fcntl_getlease(struct file *filp)
1184 locks_free_lock_context(struct inode *inode)
1188 static inline void locks_init_lock(struct file_lock *fl)
1193 static inline void locks_copy_conflock(struct file_lock *new, struct file_lock *fl)
1198 static inline void locks_copy_lock(struct file_lock *new, struct file_lock *fl)
1203 static inline void locks_remove_posix(struct file *filp, fl_owner_t owner)
1208 static inline void locks_remove_file(struct file *filp)
1213 static inline void posix_test_lock(struct file *filp, struct file_lock *fl)
1218 static inline int posix_lock_file(struct file *filp, struct file_lock *fl,
1219 struct file_lock *conflock)
1224 static inline int locks_delete_block(struct file_lock *waiter)
1229 static inline int vfs_test_lock(struct file *filp, struct file_lock *fl)
1234 static inline int vfs_lock_file(struct file *filp, unsigned int cmd,
1235 struct file_lock *fl, struct file_lock *conf)
1240 static inline int vfs_cancel_lock(struct file *filp, struct file_lock *fl)
1245 static inline int locks_lock_inode_wait(struct inode *inode, struct file_lock *fl)
1250 static inline int __break_lease(struct inode *inode, unsigned int mode, unsigned int type)
1255 static inline void lease_get_mtime(struct inode *inode,
1256 struct timespec64 *time)
1261 static inline int generic_setlease(struct file *filp, long arg,
1262 struct file_lock **flp, void **priv)
1267 static inline int vfs_setlease(struct file *filp, long arg,
1268 struct file_lock **lease, void **priv)
1273 static inline int lease_modify(struct file_lock *fl, int arg,
1274 struct list_head *dispose)
1279 struct files_struct;
1280 static inline void show_fd_locks(struct seq_file *f,
1281 struct file *filp, struct files_struct *files) {}
1282 #endif /* !CONFIG_FILE_LOCKING */
1284 static inline struct inode *file_inode(const struct file *f)
1289 static inline struct dentry *file_dentry(const struct file *file)
1291 return d_real(file->f_path.dentry, file_inode(file));
1294 static inline int locks_lock_file_wait(struct file *filp, struct file_lock *fl)
1296 return locks_lock_inode_wait(locks_inode(filp), fl);
1299 struct fasync_struct {
1303 struct fasync_struct *fa_next; /* singly linked list */
1304 struct file *fa_file;
1305 struct rcu_head fa_rcu;
1308 #define FASYNC_MAGIC 0x4601
1310 /* SMP safe fasync helpers: */
1311 extern int fasync_helper(int, struct file *, int, struct fasync_struct **);
1312 extern struct fasync_struct *fasync_insert_entry(int, struct file *, struct fasync_struct **, struct fasync_struct *);
1313 extern int fasync_remove_entry(struct file *, struct fasync_struct **);
1314 extern struct fasync_struct *fasync_alloc(void);
1315 extern void fasync_free(struct fasync_struct *);
1317 /* can be called from interrupts */
1318 extern void kill_fasync(struct fasync_struct **, int, int);
1320 extern void __f_setown(struct file *filp, struct pid *, enum pid_type, int force);
1321 extern int f_setown(struct file *filp, unsigned long arg, int force);
1322 extern void f_delown(struct file *filp);
1323 extern pid_t f_getown(struct file *filp);
1324 extern int send_sigurg(struct fown_struct *fown);
1327 * sb->s_flags. Note that these mirror the equivalent MS_* flags where
1328 * represented in both.
1330 #define SB_RDONLY 1 /* Mount read-only */
1331 #define SB_NOSUID 2 /* Ignore suid and sgid bits */
1332 #define SB_NODEV 4 /* Disallow access to device special files */
1333 #define SB_NOEXEC 8 /* Disallow program execution */
1334 #define SB_SYNCHRONOUS 16 /* Writes are synced at once */
1335 #define SB_MANDLOCK 64 /* Allow mandatory locks on an FS */
1336 #define SB_DIRSYNC 128 /* Directory modifications are synchronous */
1337 #define SB_NOATIME 1024 /* Do not update access times. */
1338 #define SB_NODIRATIME 2048 /* Do not update directory access times */
1339 #define SB_SILENT 32768
1340 #define SB_POSIXACL (1<<16) /* VFS does not apply the umask */
1341 #define SB_KERNMOUNT (1<<22) /* this is a kern_mount call */
1342 #define SB_I_VERSION (1<<23) /* Update inode I_version field */
1343 #define SB_LAZYTIME (1<<25) /* Update the on-disk [acm]times lazily */
1345 /* These sb flags are internal to the kernel */
1346 #define SB_SUBMOUNT (1<<26)
1347 #define SB_NOSEC (1<<28)
1348 #define SB_BORN (1<<29)
1349 #define SB_ACTIVE (1<<30)
1350 #define SB_NOUSER (1<<31)
1356 #define MNT_FORCE 0x00000001 /* Attempt to forcibily umount */
1357 #define MNT_DETACH 0x00000002 /* Just detach from the tree */
1358 #define MNT_EXPIRE 0x00000004 /* Mark for expiry */
1359 #define UMOUNT_NOFOLLOW 0x00000008 /* Don't follow symlink on umount */
1360 #define UMOUNT_UNUSED 0x80000000 /* Flag guaranteed to be unused */
1363 #define SB_I_CGROUPWB 0x00000001 /* cgroup-aware writeback enabled */
1364 #define SB_I_NOEXEC 0x00000002 /* Ignore executables on this fs */
1365 #define SB_I_NODEV 0x00000004 /* Ignore devices on this fs */
1366 #define SB_I_MULTIROOT 0x00000008 /* Multiple roots to the dentry tree */
1368 /* sb->s_iflags to limit user namespace mounts */
1369 #define SB_I_USERNS_VISIBLE 0x00000010 /* fstype already mounted */
1370 #define SB_I_IMA_UNVERIFIABLE_SIGNATURE 0x00000020
1371 #define SB_I_UNTRUSTED_MOUNTER 0x00000040
1373 /* Possible states of 'frozen' field */
1375 SB_UNFROZEN = 0, /* FS is unfrozen */
1376 SB_FREEZE_WRITE = 1, /* Writes, dir ops, ioctls frozen */
1377 SB_FREEZE_PAGEFAULT = 2, /* Page faults stopped as well */
1378 SB_FREEZE_FS = 3, /* For internal FS use (e.g. to stop
1379 * internal threads if needed) */
1380 SB_FREEZE_COMPLETE = 4, /* ->freeze_fs finished successfully */
1383 #define SB_FREEZE_LEVELS (SB_FREEZE_COMPLETE - 1)
1386 int frozen; /* Is sb frozen? */
1387 wait_queue_head_t wait_unfrozen; /* for get_super_thawed() */
1388 struct percpu_rw_semaphore rw_sem[SB_FREEZE_LEVELS];
1391 struct super_block {
1392 struct list_head s_list; /* Keep this first */
1393 dev_t s_dev; /* search index; _not_ kdev_t */
1394 unsigned char s_blocksize_bits;
1395 unsigned long s_blocksize;
1396 loff_t s_maxbytes; /* Max file size */
1397 struct file_system_type *s_type;
1398 const struct super_operations *s_op;
1399 const struct dquot_operations *dq_op;
1400 const struct quotactl_ops *s_qcop;
1401 const struct export_operations *s_export_op;
1402 unsigned long s_flags;
1403 unsigned long s_iflags; /* internal SB_I_* flags */
1404 unsigned long s_magic;
1405 struct dentry *s_root;
1406 struct rw_semaphore s_umount;
1409 #ifdef CONFIG_SECURITY
1412 const struct xattr_handler **s_xattr;
1413 #if IS_ENABLED(CONFIG_FS_ENCRYPTION)
1414 const struct fscrypt_operations *s_cop;
1416 struct hlist_bl_head s_roots; /* alternate root dentries for NFS */
1417 struct list_head s_mounts; /* list of mounts; _not_ for fs use */
1418 struct block_device *s_bdev;
1419 struct backing_dev_info *s_bdi;
1420 struct mtd_info *s_mtd;
1421 struct hlist_node s_instances;
1422 unsigned int s_quota_types; /* Bitmask of supported quota types */
1423 struct quota_info s_dquot; /* Diskquota specific options */
1425 struct sb_writers s_writers;
1428 * Keep s_fs_info, s_time_gran, s_fsnotify_mask, and
1429 * s_fsnotify_marks together for cache efficiency. They are frequently
1430 * accessed and rarely modified.
1432 void *s_fs_info; /* Filesystem private info */
1434 /* Granularity of c/m/atime in ns (cannot be worse than a second) */
1436 #ifdef CONFIG_FSNOTIFY
1437 __u32 s_fsnotify_mask;
1438 struct fsnotify_mark_connector __rcu *s_fsnotify_marks;
1441 char s_id[32]; /* Informational name */
1442 uuid_t s_uuid; /* UUID */
1444 unsigned int s_max_links;
1448 * The next field is for VFS *only*. No filesystems have any business
1449 * even looking at it. You had been warned.
1451 struct mutex s_vfs_rename_mutex; /* Kludge */
1454 * Filesystem subtype. If non-empty the filesystem type field
1455 * in /proc/mounts will be "type.subtype"
1459 const struct dentry_operations *s_d_op; /* default d_op for dentries */
1462 * Saved pool identifier for cleancache (-1 means none)
1464 int cleancache_poolid;
1466 struct shrinker s_shrink; /* per-sb shrinker handle */
1468 /* Number of inodes with nlink == 0 but still referenced */
1469 atomic_long_t s_remove_count;
1471 /* Pending fsnotify inode refs */
1472 atomic_long_t s_fsnotify_inode_refs;
1474 /* Being remounted read-only */
1475 int s_readonly_remount;
1477 /* AIO completions deferred from interrupt context */
1478 struct workqueue_struct *s_dio_done_wq;
1479 struct hlist_head s_pins;
1482 * Owning user namespace and default context in which to
1483 * interpret filesystem uids, gids, quotas, device nodes,
1484 * xattrs and security labels.
1486 struct user_namespace *s_user_ns;
1489 * The list_lru structure is essentially just a pointer to a table
1490 * of per-node lru lists, each of which has its own spinlock.
1491 * There is no need to put them into separate cachelines.
1493 struct list_lru s_dentry_lru;
1494 struct list_lru s_inode_lru;
1495 struct rcu_head rcu;
1496 struct work_struct destroy_work;
1498 struct mutex s_sync_lock; /* sync serialisation lock */
1501 * Indicates how deep in a filesystem stack this SB is
1505 /* s_inode_list_lock protects s_inodes */
1506 spinlock_t s_inode_list_lock ____cacheline_aligned_in_smp;
1507 struct list_head s_inodes; /* all inodes */
1509 spinlock_t s_inode_wblist_lock;
1510 struct list_head s_inodes_wb; /* writeback inodes */
1511 } __randomize_layout;
1513 /* Helper functions so that in most cases filesystems will
1514 * not need to deal directly with kuid_t and kgid_t and can
1515 * instead deal with the raw numeric values that are stored
1516 * in the filesystem.
1518 static inline uid_t i_uid_read(const struct inode *inode)
1520 return from_kuid(inode->i_sb->s_user_ns, inode->i_uid);
1523 static inline gid_t i_gid_read(const struct inode *inode)
1525 return from_kgid(inode->i_sb->s_user_ns, inode->i_gid);
1528 static inline void i_uid_write(struct inode *inode, uid_t uid)
1530 inode->i_uid = make_kuid(inode->i_sb->s_user_ns, uid);
1533 static inline void i_gid_write(struct inode *inode, gid_t gid)
1535 inode->i_gid = make_kgid(inode->i_sb->s_user_ns, gid);
1538 extern struct timespec64 timespec64_trunc(struct timespec64 t, unsigned gran);
1539 extern struct timespec64 current_time(struct inode *inode);
1542 * Snapshotting support.
1545 void __sb_end_write(struct super_block *sb, int level);
1546 int __sb_start_write(struct super_block *sb, int level, bool wait);
1548 #define __sb_writers_acquired(sb, lev) \
1549 percpu_rwsem_acquire(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1550 #define __sb_writers_release(sb, lev) \
1551 percpu_rwsem_release(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1554 * sb_end_write - drop write access to a superblock
1555 * @sb: the super we wrote to
1557 * Decrement number of writers to the filesystem. Wake up possible waiters
1558 * wanting to freeze the filesystem.
1560 static inline void sb_end_write(struct super_block *sb)
1562 __sb_end_write(sb, SB_FREEZE_WRITE);
1566 * sb_end_pagefault - drop write access to a superblock from a page fault
1567 * @sb: the super we wrote to
1569 * Decrement number of processes handling write page fault to the filesystem.
1570 * Wake up possible waiters wanting to freeze the filesystem.
1572 static inline void sb_end_pagefault(struct super_block *sb)
1574 __sb_end_write(sb, SB_FREEZE_PAGEFAULT);
1578 * sb_end_intwrite - drop write access to a superblock for internal fs purposes
1579 * @sb: the super we wrote to
1581 * Decrement fs-internal number of writers to the filesystem. Wake up possible
1582 * waiters wanting to freeze the filesystem.
1584 static inline void sb_end_intwrite(struct super_block *sb)
1586 __sb_end_write(sb, SB_FREEZE_FS);
1590 * sb_start_write - get write access to a superblock
1591 * @sb: the super we write to
1593 * When a process wants to write data or metadata to a file system (i.e. dirty
1594 * a page or an inode), it should embed the operation in a sb_start_write() -
1595 * sb_end_write() pair to get exclusion against file system freezing. This
1596 * function increments number of writers preventing freezing. If the file
1597 * system is already frozen, the function waits until the file system is
1600 * Since freeze protection behaves as a lock, users have to preserve
1601 * ordering of freeze protection and other filesystem locks. Generally,
1602 * freeze protection should be the outermost lock. In particular, we have:
1605 * -> i_mutex (write path, truncate, directory ops, ...)
1606 * -> s_umount (freeze_super, thaw_super)
1608 static inline void sb_start_write(struct super_block *sb)
1610 __sb_start_write(sb, SB_FREEZE_WRITE, true);
1613 static inline int sb_start_write_trylock(struct super_block *sb)
1615 return __sb_start_write(sb, SB_FREEZE_WRITE, false);
1619 * sb_start_pagefault - get write access to a superblock from a page fault
1620 * @sb: the super we write to
1622 * When a process starts handling write page fault, it should embed the
1623 * operation into sb_start_pagefault() - sb_end_pagefault() pair to get
1624 * exclusion against file system freezing. This is needed since the page fault
1625 * is going to dirty a page. This function increments number of running page
1626 * faults preventing freezing. If the file system is already frozen, the
1627 * function waits until the file system is thawed.
1629 * Since page fault freeze protection behaves as a lock, users have to preserve
1630 * ordering of freeze protection and other filesystem locks. It is advised to
1631 * put sb_start_pagefault() close to mmap_sem in lock ordering. Page fault
1632 * handling code implies lock dependency:
1635 * -> sb_start_pagefault
1637 static inline void sb_start_pagefault(struct super_block *sb)
1639 __sb_start_write(sb, SB_FREEZE_PAGEFAULT, true);
1643 * sb_start_intwrite - get write access to a superblock for internal fs purposes
1644 * @sb: the super we write to
1646 * This is the third level of protection against filesystem freezing. It is
1647 * free for use by a filesystem. The only requirement is that it must rank
1648 * below sb_start_pagefault.
1650 * For example filesystem can call sb_start_intwrite() when starting a
1651 * transaction which somewhat eases handling of freezing for internal sources
1652 * of filesystem changes (internal fs threads, discarding preallocation on file
1655 static inline void sb_start_intwrite(struct super_block *sb)
1657 __sb_start_write(sb, SB_FREEZE_FS, true);
1660 static inline int sb_start_intwrite_trylock(struct super_block *sb)
1662 return __sb_start_write(sb, SB_FREEZE_FS, false);
1666 extern bool inode_owner_or_capable(const struct inode *inode);
1669 * VFS helper functions..
1671 extern int vfs_create(struct inode *, struct dentry *, umode_t, bool);
1672 extern int vfs_mkdir(struct inode *, struct dentry *, umode_t);
1673 extern int vfs_mknod(struct inode *, struct dentry *, umode_t, dev_t);
1674 extern int vfs_symlink(struct inode *, struct dentry *, const char *);
1675 extern int vfs_link(struct dentry *, struct inode *, struct dentry *, struct inode **);
1676 extern int vfs_rmdir(struct inode *, struct dentry *);
1677 extern int vfs_unlink(struct inode *, struct dentry *, struct inode **);
1678 extern int vfs_rename(struct inode *, struct dentry *, struct inode *, struct dentry *, struct inode **, unsigned int);
1679 extern int vfs_whiteout(struct inode *, struct dentry *);
1681 extern struct dentry *vfs_tmpfile(struct dentry *dentry, umode_t mode,
1684 int vfs_mkobj(struct dentry *, umode_t,
1685 int (*f)(struct dentry *, umode_t, void *),
1688 extern long vfs_ioctl(struct file *file, unsigned int cmd, unsigned long arg);
1691 * VFS file helper functions.
1693 extern void inode_init_owner(struct inode *inode, const struct inode *dir,
1695 extern bool may_open_dev(const struct path *path);
1697 * VFS FS_IOC_FIEMAP helper definitions.
1699 struct fiemap_extent_info {
1700 unsigned int fi_flags; /* Flags as passed from user */
1701 unsigned int fi_extents_mapped; /* Number of mapped extents */
1702 unsigned int fi_extents_max; /* Size of fiemap_extent array */
1703 struct fiemap_extent __user *fi_extents_start; /* Start of
1704 fiemap_extent array */
1706 int fiemap_fill_next_extent(struct fiemap_extent_info *info, u64 logical,
1707 u64 phys, u64 len, u32 flags);
1708 int fiemap_check_flags(struct fiemap_extent_info *fieinfo, u32 fs_flags);
1711 * This is the "filldir" function type, used by readdir() to let
1712 * the kernel specify what kind of dirent layout it wants to have.
1713 * This allows the kernel to read directories into kernel space or
1714 * to have different dirent layouts depending on the binary type.
1717 typedef int (*filldir_t)(struct dir_context *, const char *, int, loff_t, u64,
1720 struct dir_context {
1725 struct block_device_operations;
1727 /* These macros are for out of kernel modules to test that
1728 * the kernel supports the unlocked_ioctl and compat_ioctl
1729 * fields in struct file_operations. */
1730 #define HAVE_COMPAT_IOCTL 1
1731 #define HAVE_UNLOCKED_IOCTL 1
1734 * These flags let !MMU mmap() govern direct device mapping vs immediate
1735 * copying more easily for MAP_PRIVATE, especially for ROM filesystems.
1737 * NOMMU_MAP_COPY: Copy can be mapped (MAP_PRIVATE)
1738 * NOMMU_MAP_DIRECT: Can be mapped directly (MAP_SHARED)
1739 * NOMMU_MAP_READ: Can be mapped for reading
1740 * NOMMU_MAP_WRITE: Can be mapped for writing
1741 * NOMMU_MAP_EXEC: Can be mapped for execution
1743 #define NOMMU_MAP_COPY 0x00000001
1744 #define NOMMU_MAP_DIRECT 0x00000008
1745 #define NOMMU_MAP_READ VM_MAYREAD
1746 #define NOMMU_MAP_WRITE VM_MAYWRITE
1747 #define NOMMU_MAP_EXEC VM_MAYEXEC
1749 #define NOMMU_VMFLAGS \
1750 (NOMMU_MAP_READ | NOMMU_MAP_WRITE | NOMMU_MAP_EXEC)
1753 * These flags control the behavior of the remap_file_range function pointer.
1754 * If it is called with len == 0 that means "remap to end of source file".
1755 * See Documentation/filesystems/vfs.txt for more details about this call.
1757 * REMAP_FILE_DEDUP: only remap if contents identical (i.e. deduplicate)
1758 * REMAP_FILE_CAN_SHORTEN: caller can handle a shortened request
1760 #define REMAP_FILE_DEDUP (1 << 0)
1761 #define REMAP_FILE_CAN_SHORTEN (1 << 1)
1764 * These flags signal that the caller is ok with altering various aspects of
1765 * the behavior of the remap operation. The changes must be made by the
1766 * implementation; the vfs remap helper functions can take advantage of them.
1767 * Flags in this category exist to preserve the quirky behavior of the hoisted
1768 * btrfs clone/dedupe ioctls.
1770 #define REMAP_FILE_ADVISORY (REMAP_FILE_CAN_SHORTEN)
1774 struct file_operations {
1775 struct module *owner;
1776 loff_t (*llseek) (struct file *, loff_t, int);
1777 ssize_t (*read) (struct file *, char __user *, size_t, loff_t *);
1778 ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *);
1779 ssize_t (*read_iter) (struct kiocb *, struct iov_iter *);
1780 ssize_t (*write_iter) (struct kiocb *, struct iov_iter *);
1781 int (*iterate) (struct file *, struct dir_context *);
1782 int (*iterate_shared) (struct file *, struct dir_context *);
1783 __poll_t (*poll) (struct file *, struct poll_table_struct *);
1784 long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long);
1785 long (*compat_ioctl) (struct file *, unsigned int, unsigned long);
1786 int (*mmap) (struct file *, struct vm_area_struct *);
1787 unsigned long mmap_supported_flags;
1788 int (*open) (struct inode *, struct file *);
1789 int (*flush) (struct file *, fl_owner_t id);
1790 int (*release) (struct inode *, struct file *);
1791 int (*fsync) (struct file *, loff_t, loff_t, int datasync);
1792 int (*fasync) (int, struct file *, int);
1793 int (*lock) (struct file *, int, struct file_lock *);
1794 ssize_t (*sendpage) (struct file *, struct page *, int, size_t, loff_t *, int);
1795 unsigned long (*get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
1796 int (*check_flags)(int);
1797 int (*flock) (struct file *, int, struct file_lock *);
1798 ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, loff_t *, size_t, unsigned int);
1799 ssize_t (*splice_read)(struct file *, loff_t *, struct pipe_inode_info *, size_t, unsigned int);
1800 int (*setlease)(struct file *, long, struct file_lock **, void **);
1801 long (*fallocate)(struct file *file, int mode, loff_t offset,
1803 void (*show_fdinfo)(struct seq_file *m, struct file *f);
1805 unsigned (*mmap_capabilities)(struct file *);
1807 ssize_t (*copy_file_range)(struct file *, loff_t, struct file *,
1808 loff_t, size_t, unsigned int);
1809 loff_t (*remap_file_range)(struct file *file_in, loff_t pos_in,
1810 struct file *file_out, loff_t pos_out,
1811 loff_t len, unsigned int remap_flags);
1812 int (*fadvise)(struct file *, loff_t, loff_t, int);
1813 } __randomize_layout;
1815 struct inode_operations {
1816 struct dentry * (*lookup) (struct inode *,struct dentry *, unsigned int);
1817 const char * (*get_link) (struct dentry *, struct inode *, struct delayed_call *);
1818 int (*permission) (struct inode *, int);
1819 struct posix_acl * (*get_acl)(struct inode *, int);
1821 int (*readlink) (struct dentry *, char __user *,int);
1823 int (*create) (struct inode *,struct dentry *, umode_t, bool);
1824 int (*link) (struct dentry *,struct inode *,struct dentry *);
1825 int (*unlink) (struct inode *,struct dentry *);
1826 int (*symlink) (struct inode *,struct dentry *,const char *);
1827 int (*mkdir) (struct inode *,struct dentry *,umode_t);
1828 int (*rmdir) (struct inode *,struct dentry *);
1829 int (*mknod) (struct inode *,struct dentry *,umode_t,dev_t);
1830 int (*rename) (struct inode *, struct dentry *,
1831 struct inode *, struct dentry *, unsigned int);
1832 int (*setattr) (struct dentry *, struct iattr *);
1833 int (*getattr) (const struct path *, struct kstat *, u32, unsigned int);
1834 ssize_t (*listxattr) (struct dentry *, char *, size_t);
1835 int (*fiemap)(struct inode *, struct fiemap_extent_info *, u64 start,
1837 int (*update_time)(struct inode *, struct timespec64 *, int);
1838 int (*atomic_open)(struct inode *, struct dentry *,
1839 struct file *, unsigned open_flag,
1840 umode_t create_mode);
1841 int (*tmpfile) (struct inode *, struct dentry *, umode_t);
1842 int (*set_acl)(struct inode *, struct posix_acl *, int);
1843 } ____cacheline_aligned;
1845 static inline ssize_t call_read_iter(struct file *file, struct kiocb *kio,
1846 struct iov_iter *iter)
1848 return file->f_op->read_iter(kio, iter);
1851 static inline ssize_t call_write_iter(struct file *file, struct kiocb *kio,
1852 struct iov_iter *iter)
1854 return file->f_op->write_iter(kio, iter);
1857 static inline int call_mmap(struct file *file, struct vm_area_struct *vma)
1859 return file->f_op->mmap(file, vma);
1862 ssize_t rw_copy_check_uvector(int type, const struct iovec __user * uvector,
1863 unsigned long nr_segs, unsigned long fast_segs,
1864 struct iovec *fast_pointer,
1865 struct iovec **ret_pointer);
1867 extern ssize_t __vfs_read(struct file *, char __user *, size_t, loff_t *);
1868 extern ssize_t vfs_read(struct file *, char __user *, size_t, loff_t *);
1869 extern ssize_t vfs_write(struct file *, const char __user *, size_t, loff_t *);
1870 extern ssize_t vfs_readv(struct file *, const struct iovec __user *,
1871 unsigned long, loff_t *, rwf_t);
1872 extern ssize_t vfs_copy_file_range(struct file *, loff_t , struct file *,
1873 loff_t, size_t, unsigned int);
1874 extern int generic_remap_file_range_prep(struct file *file_in, loff_t pos_in,
1875 struct file *file_out, loff_t pos_out,
1877 unsigned int remap_flags);
1878 extern loff_t do_clone_file_range(struct file *file_in, loff_t pos_in,
1879 struct file *file_out, loff_t pos_out,
1880 loff_t len, unsigned int remap_flags);
1881 extern loff_t vfs_clone_file_range(struct file *file_in, loff_t pos_in,
1882 struct file *file_out, loff_t pos_out,
1883 loff_t len, unsigned int remap_flags);
1884 extern int vfs_dedupe_file_range(struct file *file,
1885 struct file_dedupe_range *same);
1886 extern loff_t vfs_dedupe_file_range_one(struct file *src_file, loff_t src_pos,
1887 struct file *dst_file, loff_t dst_pos,
1888 loff_t len, unsigned int remap_flags);
1891 struct super_operations {
1892 struct inode *(*alloc_inode)(struct super_block *sb);
1893 void (*destroy_inode)(struct inode *);
1895 void (*dirty_inode) (struct inode *, int flags);
1896 int (*write_inode) (struct inode *, struct writeback_control *wbc);
1897 int (*drop_inode) (struct inode *);
1898 void (*evict_inode) (struct inode *);
1899 void (*put_super) (struct super_block *);
1900 int (*sync_fs)(struct super_block *sb, int wait);
1901 int (*freeze_super) (struct super_block *);
1902 int (*freeze_fs) (struct super_block *);
1903 int (*thaw_super) (struct super_block *);
1904 int (*unfreeze_fs) (struct super_block *);
1905 int (*statfs) (struct dentry *, struct kstatfs *);
1906 int (*remount_fs) (struct super_block *, int *, char *);
1907 void (*umount_begin) (struct super_block *);
1909 int (*show_options)(struct seq_file *, struct dentry *);
1910 int (*show_devname)(struct seq_file *, struct dentry *);
1911 int (*show_path)(struct seq_file *, struct dentry *);
1912 int (*show_stats)(struct seq_file *, struct dentry *);
1914 ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t);
1915 ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t);
1916 struct dquot **(*get_dquots)(struct inode *);
1918 int (*bdev_try_to_free_page)(struct super_block*, struct page*, gfp_t);
1919 long (*nr_cached_objects)(struct super_block *,
1920 struct shrink_control *);
1921 long (*free_cached_objects)(struct super_block *,
1922 struct shrink_control *);
1926 * Inode flags - they have no relation to superblock flags now
1928 #define S_SYNC 1 /* Writes are synced at once */
1929 #define S_NOATIME 2 /* Do not update access times */
1930 #define S_APPEND 4 /* Append-only file */
1931 #define S_IMMUTABLE 8 /* Immutable file */
1932 #define S_DEAD 16 /* removed, but still open directory */
1933 #define S_NOQUOTA 32 /* Inode is not counted to quota */
1934 #define S_DIRSYNC 64 /* Directory modifications are synchronous */
1935 #define S_NOCMTIME 128 /* Do not update file c/mtime */
1936 #define S_SWAPFILE 256 /* Do not truncate: swapon got its bmaps */
1937 #define S_PRIVATE 512 /* Inode is fs-internal */
1938 #define S_IMA 1024 /* Inode has an associated IMA struct */
1939 #define S_AUTOMOUNT 2048 /* Automount/referral quasi-directory */
1940 #define S_NOSEC 4096 /* no suid or xattr security attributes */
1941 #ifdef CONFIG_FS_DAX
1942 #define S_DAX 8192 /* Direct Access, avoiding the page cache */
1944 #define S_DAX 0 /* Make all the DAX code disappear */
1946 #define S_ENCRYPTED 16384 /* Encrypted file (using fs/crypto/) */
1949 * Note that nosuid etc flags are inode-specific: setting some file-system
1950 * flags just means all the inodes inherit those flags by default. It might be
1951 * possible to override it selectively if you really wanted to with some
1952 * ioctl() that is not currently implemented.
1954 * Exception: SB_RDONLY is always applied to the entire file system.
1956 * Unfortunately, it is possible to change a filesystems flags with it mounted
1957 * with files in use. This means that all of the inodes will not have their
1958 * i_flags updated. Hence, i_flags no longer inherit the superblock mount
1959 * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org
1961 #define __IS_FLG(inode, flg) ((inode)->i_sb->s_flags & (flg))
1963 static inline bool sb_rdonly(const struct super_block *sb) { return sb->s_flags & SB_RDONLY; }
1964 #define IS_RDONLY(inode) sb_rdonly((inode)->i_sb)
1965 #define IS_SYNC(inode) (__IS_FLG(inode, SB_SYNCHRONOUS) || \
1966 ((inode)->i_flags & S_SYNC))
1967 #define IS_DIRSYNC(inode) (__IS_FLG(inode, SB_SYNCHRONOUS|SB_DIRSYNC) || \
1968 ((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
1969 #define IS_MANDLOCK(inode) __IS_FLG(inode, SB_MANDLOCK)
1970 #define IS_NOATIME(inode) __IS_FLG(inode, SB_RDONLY|SB_NOATIME)
1971 #define IS_I_VERSION(inode) __IS_FLG(inode, SB_I_VERSION)
1973 #define IS_NOQUOTA(inode) ((inode)->i_flags & S_NOQUOTA)
1974 #define IS_APPEND(inode) ((inode)->i_flags & S_APPEND)
1975 #define IS_IMMUTABLE(inode) ((inode)->i_flags & S_IMMUTABLE)
1976 #define IS_POSIXACL(inode) __IS_FLG(inode, SB_POSIXACL)
1978 #define IS_DEADDIR(inode) ((inode)->i_flags & S_DEAD)
1979 #define IS_NOCMTIME(inode) ((inode)->i_flags & S_NOCMTIME)
1980 #define IS_SWAPFILE(inode) ((inode)->i_flags & S_SWAPFILE)
1981 #define IS_PRIVATE(inode) ((inode)->i_flags & S_PRIVATE)
1982 #define IS_IMA(inode) ((inode)->i_flags & S_IMA)
1983 #define IS_AUTOMOUNT(inode) ((inode)->i_flags & S_AUTOMOUNT)
1984 #define IS_NOSEC(inode) ((inode)->i_flags & S_NOSEC)
1985 #define IS_DAX(inode) ((inode)->i_flags & S_DAX)
1986 #define IS_ENCRYPTED(inode) ((inode)->i_flags & S_ENCRYPTED)
1988 #define IS_WHITEOUT(inode) (S_ISCHR(inode->i_mode) && \
1989 (inode)->i_rdev == WHITEOUT_DEV)
1991 static inline bool HAS_UNMAPPED_ID(struct inode *inode)
1993 return !uid_valid(inode->i_uid) || !gid_valid(inode->i_gid);
1996 static inline enum rw_hint file_write_hint(struct file *file)
1998 if (file->f_write_hint != WRITE_LIFE_NOT_SET)
1999 return file->f_write_hint;
2001 return file_inode(file)->i_write_hint;
2004 static inline int iocb_flags(struct file *file);
2006 static inline u16 ki_hint_validate(enum rw_hint hint)
2008 typeof(((struct kiocb *)0)->ki_hint) max_hint = -1;
2010 if (hint <= max_hint)
2015 static inline void init_sync_kiocb(struct kiocb *kiocb, struct file *filp)
2017 *kiocb = (struct kiocb) {
2019 .ki_flags = iocb_flags(filp),
2020 .ki_hint = ki_hint_validate(file_write_hint(filp)),
2021 .ki_ioprio = get_current_ioprio(),
2026 * Inode state bits. Protected by inode->i_lock
2028 * Three bits determine the dirty state of the inode, I_DIRTY_SYNC,
2029 * I_DIRTY_DATASYNC and I_DIRTY_PAGES.
2031 * Four bits define the lifetime of an inode. Initially, inodes are I_NEW,
2032 * until that flag is cleared. I_WILL_FREE, I_FREEING and I_CLEAR are set at
2033 * various stages of removing an inode.
2035 * Two bits are used for locking and completion notification, I_NEW and I_SYNC.
2037 * I_DIRTY_SYNC Inode is dirty, but doesn't have to be written on
2038 * fdatasync(). i_atime is the usual cause.
2039 * I_DIRTY_DATASYNC Data-related inode changes pending. We keep track of
2040 * these changes separately from I_DIRTY_SYNC so that we
2041 * don't have to write inode on fdatasync() when only
2042 * mtime has changed in it.
2043 * I_DIRTY_PAGES Inode has dirty pages. Inode itself may be clean.
2044 * I_NEW Serves as both a mutex and completion notification.
2045 * New inodes set I_NEW. If two processes both create
2046 * the same inode, one of them will release its inode and
2047 * wait for I_NEW to be released before returning.
2048 * Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can
2049 * also cause waiting on I_NEW, without I_NEW actually
2050 * being set. find_inode() uses this to prevent returning
2051 * nearly-dead inodes.
2052 * I_WILL_FREE Must be set when calling write_inode_now() if i_count
2053 * is zero. I_FREEING must be set when I_WILL_FREE is
2055 * I_FREEING Set when inode is about to be freed but still has dirty
2056 * pages or buffers attached or the inode itself is still
2058 * I_CLEAR Added by clear_inode(). In this state the inode is
2059 * clean and can be destroyed. Inode keeps I_FREEING.
2061 * Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are
2062 * prohibited for many purposes. iget() must wait for
2063 * the inode to be completely released, then create it
2064 * anew. Other functions will just ignore such inodes,
2065 * if appropriate. I_NEW is used for waiting.
2067 * I_SYNC Writeback of inode is running. The bit is set during
2068 * data writeback, and cleared with a wakeup on the bit
2069 * address once it is done. The bit is also used to pin
2070 * the inode in memory for flusher thread.
2072 * I_REFERENCED Marks the inode as recently references on the LRU list.
2074 * I_DIO_WAKEUP Never set. Only used as a key for wait_on_bit().
2076 * I_WB_SWITCH Cgroup bdi_writeback switching in progress. Used to
2077 * synchronize competing switching instances and to tell
2078 * wb stat updates to grab the i_pages lock. See
2079 * inode_switch_wbs_work_fn() for details.
2081 * I_OVL_INUSE Used by overlayfs to get exclusive ownership on upper
2082 * and work dirs among overlayfs mounts.
2084 * I_CREATING New object's inode in the middle of setting up.
2086 * Q: What is the difference between I_WILL_FREE and I_FREEING?
2088 #define I_DIRTY_SYNC (1 << 0)
2089 #define I_DIRTY_DATASYNC (1 << 1)
2090 #define I_DIRTY_PAGES (1 << 2)
2092 #define I_NEW (1 << __I_NEW)
2093 #define I_WILL_FREE (1 << 4)
2094 #define I_FREEING (1 << 5)
2095 #define I_CLEAR (1 << 6)
2097 #define I_SYNC (1 << __I_SYNC)
2098 #define I_REFERENCED (1 << 8)
2099 #define __I_DIO_WAKEUP 9
2100 #define I_DIO_WAKEUP (1 << __I_DIO_WAKEUP)
2101 #define I_LINKABLE (1 << 10)
2102 #define I_DIRTY_TIME (1 << 11)
2103 #define __I_DIRTY_TIME_EXPIRED 12
2104 #define I_DIRTY_TIME_EXPIRED (1 << __I_DIRTY_TIME_EXPIRED)
2105 #define I_WB_SWITCH (1 << 13)
2106 #define I_OVL_INUSE (1 << 14)
2107 #define I_CREATING (1 << 15)
2109 #define I_DIRTY_INODE (I_DIRTY_SYNC | I_DIRTY_DATASYNC)
2110 #define I_DIRTY (I_DIRTY_INODE | I_DIRTY_PAGES)
2111 #define I_DIRTY_ALL (I_DIRTY | I_DIRTY_TIME)
2113 extern void __mark_inode_dirty(struct inode *, int);
2114 static inline void mark_inode_dirty(struct inode *inode)
2116 __mark_inode_dirty(inode, I_DIRTY);
2119 static inline void mark_inode_dirty_sync(struct inode *inode)
2121 __mark_inode_dirty(inode, I_DIRTY_SYNC);
2124 extern void inc_nlink(struct inode *inode);
2125 extern void drop_nlink(struct inode *inode);
2126 extern void clear_nlink(struct inode *inode);
2127 extern void set_nlink(struct inode *inode, unsigned int nlink);
2129 static inline void inode_inc_link_count(struct inode *inode)
2132 mark_inode_dirty(inode);
2135 static inline void inode_dec_link_count(struct inode *inode)
2138 mark_inode_dirty(inode);
2141 enum file_time_flags {
2148 extern bool atime_needs_update(const struct path *, struct inode *);
2149 extern void touch_atime(const struct path *);
2150 static inline void file_accessed(struct file *file)
2152 if (!(file->f_flags & O_NOATIME))
2153 touch_atime(&file->f_path);
2156 int sync_inode(struct inode *inode, struct writeback_control *wbc);
2157 int sync_inode_metadata(struct inode *inode, int wait);
2159 struct file_system_type {
2162 #define FS_REQUIRES_DEV 1
2163 #define FS_BINARY_MOUNTDATA 2
2164 #define FS_HAS_SUBTYPE 4
2165 #define FS_USERNS_MOUNT 8 /* Can be mounted by userns root */
2166 #define FS_RENAME_DOES_D_MOVE 32768 /* FS will handle d_move() during rename() internally. */
2167 struct dentry *(*mount) (struct file_system_type *, int,
2168 const char *, void *);
2169 void (*kill_sb) (struct super_block *);
2170 struct module *owner;
2171 struct file_system_type * next;
2172 struct hlist_head fs_supers;
2174 struct lock_class_key s_lock_key;
2175 struct lock_class_key s_umount_key;
2176 struct lock_class_key s_vfs_rename_key;
2177 struct lock_class_key s_writers_key[SB_FREEZE_LEVELS];
2179 struct lock_class_key i_lock_key;
2180 struct lock_class_key i_mutex_key;
2181 struct lock_class_key i_mutex_dir_key;
2184 #define MODULE_ALIAS_FS(NAME) MODULE_ALIAS("fs-" NAME)
2186 extern struct dentry *mount_ns(struct file_system_type *fs_type,
2187 int flags, void *data, void *ns, struct user_namespace *user_ns,
2188 int (*fill_super)(struct super_block *, void *, int));
2190 extern struct dentry *mount_bdev(struct file_system_type *fs_type,
2191 int flags, const char *dev_name, void *data,
2192 int (*fill_super)(struct super_block *, void *, int));
2194 static inline struct dentry *mount_bdev(struct file_system_type *fs_type,
2195 int flags, const char *dev_name, void *data,
2196 int (*fill_super)(struct super_block *, void *, int))
2198 return ERR_PTR(-ENODEV);
2201 extern struct dentry *mount_single(struct file_system_type *fs_type,
2202 int flags, void *data,
2203 int (*fill_super)(struct super_block *, void *, int));
2204 extern struct dentry *mount_nodev(struct file_system_type *fs_type,
2205 int flags, void *data,
2206 int (*fill_super)(struct super_block *, void *, int));
2207 extern struct dentry *mount_subtree(struct vfsmount *mnt, const char *path);
2208 void generic_shutdown_super(struct super_block *sb);
2210 void kill_block_super(struct super_block *sb);
2212 static inline void kill_block_super(struct super_block *sb)
2217 void kill_anon_super(struct super_block *sb);
2218 void kill_litter_super(struct super_block *sb);
2219 void deactivate_super(struct super_block *sb);
2220 void deactivate_locked_super(struct super_block *sb);
2221 int set_anon_super(struct super_block *s, void *data);
2222 int get_anon_bdev(dev_t *);
2223 void free_anon_bdev(dev_t);
2224 struct super_block *sget_userns(struct file_system_type *type,
2225 int (*test)(struct super_block *,void *),
2226 int (*set)(struct super_block *,void *),
2227 int flags, struct user_namespace *user_ns,
2229 struct super_block *sget(struct file_system_type *type,
2230 int (*test)(struct super_block *,void *),
2231 int (*set)(struct super_block *,void *),
2232 int flags, void *data);
2233 extern struct dentry *mount_pseudo_xattr(struct file_system_type *, char *,
2234 const struct super_operations *ops,
2235 const struct xattr_handler **xattr,
2236 const struct dentry_operations *dops,
2239 static inline struct dentry *
2240 mount_pseudo(struct file_system_type *fs_type, char *name,
2241 const struct super_operations *ops,
2242 const struct dentry_operations *dops, unsigned long magic)
2244 return mount_pseudo_xattr(fs_type, name, ops, NULL, dops, magic);
2247 /* Alas, no aliases. Too much hassle with bringing module.h everywhere */
2248 #define fops_get(fops) \
2249 (((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
2250 #define fops_put(fops) \
2251 do { if (fops) module_put((fops)->owner); } while(0)
2253 * This one is to be used *ONLY* from ->open() instances.
2254 * fops must be non-NULL, pinned down *and* module dependencies
2255 * should be sufficient to pin the caller down as well.
2257 #define replace_fops(f, fops) \
2259 struct file *__file = (f); \
2260 fops_put(__file->f_op); \
2261 BUG_ON(!(__file->f_op = (fops))); \
2264 extern int register_filesystem(struct file_system_type *);
2265 extern int unregister_filesystem(struct file_system_type *);
2266 extern struct vfsmount *kern_mount_data(struct file_system_type *, void *data);
2267 #define kern_mount(type) kern_mount_data(type, NULL)
2268 extern void kern_unmount(struct vfsmount *mnt);
2269 extern int may_umount_tree(struct vfsmount *);
2270 extern int may_umount(struct vfsmount *);
2271 extern long do_mount(const char *, const char __user *,
2272 const char *, unsigned long, void *);
2273 extern struct vfsmount *collect_mounts(const struct path *);
2274 extern void drop_collected_mounts(struct vfsmount *);
2275 extern int iterate_mounts(int (*)(struct vfsmount *, void *), void *,
2277 extern int vfs_statfs(const struct path *, struct kstatfs *);
2278 extern int user_statfs(const char __user *, struct kstatfs *);
2279 extern int fd_statfs(int, struct kstatfs *);
2280 extern int freeze_super(struct super_block *super);
2281 extern int thaw_super(struct super_block *super);
2282 extern bool our_mnt(struct vfsmount *mnt);
2283 extern __printf(2, 3)
2284 int super_setup_bdi_name(struct super_block *sb, char *fmt, ...);
2285 extern int super_setup_bdi(struct super_block *sb);
2287 extern int current_umask(void);
2289 extern void ihold(struct inode * inode);
2290 extern void iput(struct inode *);
2291 extern int generic_update_time(struct inode *, struct timespec64 *, int);
2294 extern struct kobject *fs_kobj;
2296 #define MAX_RW_COUNT (INT_MAX & PAGE_MASK)
2298 #ifdef CONFIG_MANDATORY_FILE_LOCKING
2299 extern int locks_mandatory_locked(struct file *);
2300 extern int locks_mandatory_area(struct inode *, struct file *, loff_t, loff_t, unsigned char);
2303 * Candidates for mandatory locking have the setgid bit set
2304 * but no group execute bit - an otherwise meaningless combination.
2307 static inline int __mandatory_lock(struct inode *ino)
2309 return (ino->i_mode & (S_ISGID | S_IXGRP)) == S_ISGID;
2313 * ... and these candidates should be on SB_MANDLOCK mounted fs,
2314 * otherwise these will be advisory locks
2317 static inline int mandatory_lock(struct inode *ino)
2319 return IS_MANDLOCK(ino) && __mandatory_lock(ino);
2322 static inline int locks_verify_locked(struct file *file)
2324 if (mandatory_lock(locks_inode(file)))
2325 return locks_mandatory_locked(file);
2329 static inline int locks_verify_truncate(struct inode *inode,
2333 if (!inode->i_flctx || !mandatory_lock(inode))
2336 if (size < inode->i_size) {
2337 return locks_mandatory_area(inode, f, size, inode->i_size - 1,
2340 return locks_mandatory_area(inode, f, inode->i_size, size - 1,
2345 #else /* !CONFIG_MANDATORY_FILE_LOCKING */
2347 static inline int locks_mandatory_locked(struct file *file)
2352 static inline int locks_mandatory_area(struct inode *inode, struct file *filp,
2353 loff_t start, loff_t end, unsigned char type)
2358 static inline int __mandatory_lock(struct inode *inode)
2363 static inline int mandatory_lock(struct inode *inode)
2368 static inline int locks_verify_locked(struct file *file)
2373 static inline int locks_verify_truncate(struct inode *inode, struct file *filp,
2379 #endif /* CONFIG_MANDATORY_FILE_LOCKING */
2382 #ifdef CONFIG_FILE_LOCKING
2383 static inline int break_lease(struct inode *inode, unsigned int mode)
2386 * Since this check is lockless, we must ensure that any refcounts
2387 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2388 * could end up racing with tasks trying to set a new lease on this
2392 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2393 return __break_lease(inode, mode, FL_LEASE);
2397 static inline int break_deleg(struct inode *inode, unsigned int mode)
2400 * Since this check is lockless, we must ensure that any refcounts
2401 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2402 * could end up racing with tasks trying to set a new lease on this
2406 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2407 return __break_lease(inode, mode, FL_DELEG);
2411 static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode)
2415 ret = break_deleg(inode, O_WRONLY|O_NONBLOCK);
2416 if (ret == -EWOULDBLOCK && delegated_inode) {
2417 *delegated_inode = inode;
2423 static inline int break_deleg_wait(struct inode **delegated_inode)
2427 ret = break_deleg(*delegated_inode, O_WRONLY);
2428 iput(*delegated_inode);
2429 *delegated_inode = NULL;
2433 static inline int break_layout(struct inode *inode, bool wait)
2436 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2437 return __break_lease(inode,
2438 wait ? O_WRONLY : O_WRONLY | O_NONBLOCK,
2443 #else /* !CONFIG_FILE_LOCKING */
2444 static inline int break_lease(struct inode *inode, unsigned int mode)
2449 static inline int break_deleg(struct inode *inode, unsigned int mode)
2454 static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode)
2459 static inline int break_deleg_wait(struct inode **delegated_inode)
2465 static inline int break_layout(struct inode *inode, bool wait)
2470 #endif /* CONFIG_FILE_LOCKING */
2475 const char *name; /* pointer to actual string */
2476 const __user char *uptr; /* original userland pointer */
2478 struct audit_names *aname;
2482 extern long vfs_truncate(const struct path *, loff_t);
2483 extern int do_truncate(struct dentry *, loff_t start, unsigned int time_attrs,
2485 extern int vfs_fallocate(struct file *file, int mode, loff_t offset,
2487 extern long do_sys_open(int dfd, const char __user *filename, int flags,
2489 extern struct file *file_open_name(struct filename *, int, umode_t);
2490 extern struct file *filp_open(const char *, int, umode_t);
2491 extern struct file *file_open_root(struct dentry *, struct vfsmount *,
2492 const char *, int, umode_t);
2493 extern struct file * dentry_open(const struct path *, int, const struct cred *);
2494 extern struct file * open_with_fake_path(const struct path *, int,
2495 struct inode*, const struct cred *);
2496 static inline struct file *file_clone_open(struct file *file)
2498 return dentry_open(&file->f_path, file->f_flags, file->f_cred);
2500 extern int filp_close(struct file *, fl_owner_t id);
2502 extern struct filename *getname_flags(const char __user *, int, int *);
2503 extern struct filename *getname(const char __user *);
2504 extern struct filename *getname_kernel(const char *);
2505 extern void putname(struct filename *name);
2507 extern int finish_open(struct file *file, struct dentry *dentry,
2508 int (*open)(struct inode *, struct file *));
2509 extern int finish_no_open(struct file *file, struct dentry *dentry);
2513 extern int ioctl_preallocate(struct file *filp, void __user *argp);
2516 extern void __init vfs_caches_init_early(void);
2517 extern void __init vfs_caches_init(void);
2519 extern struct kmem_cache *names_cachep;
2521 #define __getname() kmem_cache_alloc(names_cachep, GFP_KERNEL)
2522 #define __putname(name) kmem_cache_free(names_cachep, (void *)(name))
2525 extern int register_blkdev(unsigned int, const char *);
2526 extern void unregister_blkdev(unsigned int, const char *);
2527 extern void bdev_unhash_inode(dev_t dev);
2528 extern struct block_device *bdget(dev_t);
2529 extern struct block_device *bdgrab(struct block_device *bdev);
2530 extern void bd_set_size(struct block_device *, loff_t size);
2531 extern void bd_forget(struct inode *inode);
2532 extern void bdput(struct block_device *);
2533 extern void invalidate_bdev(struct block_device *);
2534 extern void iterate_bdevs(void (*)(struct block_device *, void *), void *);
2535 extern int sync_blockdev(struct block_device *bdev);
2536 extern void kill_bdev(struct block_device *);
2537 extern struct super_block *freeze_bdev(struct block_device *);
2538 extern void emergency_thaw_all(void);
2539 extern void emergency_thaw_bdev(struct super_block *sb);
2540 extern int thaw_bdev(struct block_device *bdev, struct super_block *sb);
2541 extern int fsync_bdev(struct block_device *);
2543 extern struct super_block *blockdev_superblock;
2545 static inline bool sb_is_blkdev_sb(struct super_block *sb)
2547 return sb == blockdev_superblock;
2550 static inline void bd_forget(struct inode *inode) {}
2551 static inline int sync_blockdev(struct block_device *bdev) { return 0; }
2552 static inline void kill_bdev(struct block_device *bdev) {}
2553 static inline void invalidate_bdev(struct block_device *bdev) {}
2555 static inline struct super_block *freeze_bdev(struct block_device *sb)
2560 static inline int thaw_bdev(struct block_device *bdev, struct super_block *sb)
2565 static inline int emergency_thaw_bdev(struct super_block *sb)
2570 static inline void iterate_bdevs(void (*f)(struct block_device *, void *), void *arg)
2574 static inline bool sb_is_blkdev_sb(struct super_block *sb)
2579 extern int sync_filesystem(struct super_block *);
2580 extern const struct file_operations def_blk_fops;
2581 extern const struct file_operations def_chr_fops;
2583 extern int ioctl_by_bdev(struct block_device *, unsigned, unsigned long);
2584 extern int blkdev_ioctl(struct block_device *, fmode_t, unsigned, unsigned long);
2585 extern long compat_blkdev_ioctl(struct file *, unsigned, unsigned long);
2586 extern int blkdev_get(struct block_device *bdev, fmode_t mode, void *holder);
2587 extern struct block_device *blkdev_get_by_path(const char *path, fmode_t mode,
2589 extern struct block_device *blkdev_get_by_dev(dev_t dev, fmode_t mode,
2591 extern void blkdev_put(struct block_device *bdev, fmode_t mode);
2592 extern int __blkdev_reread_part(struct block_device *bdev);
2593 extern int blkdev_reread_part(struct block_device *bdev);
2596 extern int bd_link_disk_holder(struct block_device *bdev, struct gendisk *disk);
2597 extern void bd_unlink_disk_holder(struct block_device *bdev,
2598 struct gendisk *disk);
2600 static inline int bd_link_disk_holder(struct block_device *bdev,
2601 struct gendisk *disk)
2605 static inline void bd_unlink_disk_holder(struct block_device *bdev,
2606 struct gendisk *disk)
2613 #define CHRDEV_MAJOR_MAX 512
2614 /* Marks the bottom of the first segment of free char majors */
2615 #define CHRDEV_MAJOR_DYN_END 234
2616 /* Marks the top and bottom of the second segment of free char majors */
2617 #define CHRDEV_MAJOR_DYN_EXT_START 511
2618 #define CHRDEV_MAJOR_DYN_EXT_END 384
2620 extern int alloc_chrdev_region(dev_t *, unsigned, unsigned, const char *);
2621 extern int register_chrdev_region(dev_t, unsigned, const char *);
2622 extern int __register_chrdev(unsigned int major, unsigned int baseminor,
2623 unsigned int count, const char *name,
2624 const struct file_operations *fops);
2625 extern void __unregister_chrdev(unsigned int major, unsigned int baseminor,
2626 unsigned int count, const char *name);
2627 extern void unregister_chrdev_region(dev_t, unsigned);
2628 extern void chrdev_show(struct seq_file *,off_t);
2630 static inline int register_chrdev(unsigned int major, const char *name,
2631 const struct file_operations *fops)
2633 return __register_chrdev(major, 0, 256, name, fops);
2636 static inline void unregister_chrdev(unsigned int major, const char *name)
2638 __unregister_chrdev(major, 0, 256, name);
2641 /* fs/block_dev.c */
2642 #define BDEVNAME_SIZE 32 /* Largest string for a blockdev identifier */
2643 #define BDEVT_SIZE 10 /* Largest string for MAJ:MIN for blkdev */
2646 #define BLKDEV_MAJOR_MAX 512
2647 extern const char *__bdevname(dev_t, char *buffer);
2648 extern const char *bdevname(struct block_device *bdev, char *buffer);
2649 extern struct block_device *lookup_bdev(const char *);
2650 extern void blkdev_show(struct seq_file *,off_t);
2653 #define BLKDEV_MAJOR_MAX 0
2656 extern void init_special_inode(struct inode *, umode_t, dev_t);
2658 /* Invalid inode operations -- fs/bad_inode.c */
2659 extern void make_bad_inode(struct inode *);
2660 extern bool is_bad_inode(struct inode *);
2663 extern void check_disk_size_change(struct gendisk *disk,
2664 struct block_device *bdev, bool verbose);
2665 extern int revalidate_disk(struct gendisk *);
2666 extern int check_disk_change(struct block_device *);
2667 extern int __invalidate_device(struct block_device *, bool);
2668 extern int invalidate_partition(struct gendisk *, int);
2670 unsigned long invalidate_mapping_pages(struct address_space *mapping,
2671 pgoff_t start, pgoff_t end);
2673 static inline void invalidate_remote_inode(struct inode *inode)
2675 if (S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) ||
2676 S_ISLNK(inode->i_mode))
2677 invalidate_mapping_pages(inode->i_mapping, 0, -1);
2679 extern int invalidate_inode_pages2(struct address_space *mapping);
2680 extern int invalidate_inode_pages2_range(struct address_space *mapping,
2681 pgoff_t start, pgoff_t end);
2682 extern int write_inode_now(struct inode *, int);
2683 extern int filemap_fdatawrite(struct address_space *);
2684 extern int filemap_flush(struct address_space *);
2685 extern int filemap_fdatawait_keep_errors(struct address_space *mapping);
2686 extern int filemap_fdatawait_range(struct address_space *, loff_t lstart,
2689 static inline int filemap_fdatawait(struct address_space *mapping)
2691 return filemap_fdatawait_range(mapping, 0, LLONG_MAX);
2694 extern bool filemap_range_has_page(struct address_space *, loff_t lstart,
2696 extern int filemap_write_and_wait(struct address_space *mapping);
2697 extern int filemap_write_and_wait_range(struct address_space *mapping,
2698 loff_t lstart, loff_t lend);
2699 extern int __filemap_fdatawrite_range(struct address_space *mapping,
2700 loff_t start, loff_t end, int sync_mode);
2701 extern int filemap_fdatawrite_range(struct address_space *mapping,
2702 loff_t start, loff_t end);
2703 extern int filemap_check_errors(struct address_space *mapping);
2704 extern void __filemap_set_wb_err(struct address_space *mapping, int err);
2706 extern int __must_check file_fdatawait_range(struct file *file, loff_t lstart,
2708 extern int __must_check file_check_and_advance_wb_err(struct file *file);
2709 extern int __must_check file_write_and_wait_range(struct file *file,
2710 loff_t start, loff_t end);
2712 static inline int file_write_and_wait(struct file *file)
2714 return file_write_and_wait_range(file, 0, LLONG_MAX);
2718 * filemap_set_wb_err - set a writeback error on an address_space
2719 * @mapping: mapping in which to set writeback error
2720 * @err: error to be set in mapping
2722 * When writeback fails in some way, we must record that error so that
2723 * userspace can be informed when fsync and the like are called. We endeavor
2724 * to report errors on any file that was open at the time of the error. Some
2725 * internal callers also need to know when writeback errors have occurred.
2727 * When a writeback error occurs, most filesystems will want to call
2728 * filemap_set_wb_err to record the error in the mapping so that it will be
2729 * automatically reported whenever fsync is called on the file.
2731 static inline void filemap_set_wb_err(struct address_space *mapping, int err)
2733 /* Fastpath for common case of no error */
2735 __filemap_set_wb_err(mapping, err);
2739 * filemap_check_wb_error - has an error occurred since the mark was sampled?
2740 * @mapping: mapping to check for writeback errors
2741 * @since: previously-sampled errseq_t
2743 * Grab the errseq_t value from the mapping, and see if it has changed "since"
2744 * the given value was sampled.
2746 * If it has then report the latest error set, otherwise return 0.
2748 static inline int filemap_check_wb_err(struct address_space *mapping,
2751 return errseq_check(&mapping->wb_err, since);
2755 * filemap_sample_wb_err - sample the current errseq_t to test for later errors
2756 * @mapping: mapping to be sampled
2758 * Writeback errors are always reported relative to a particular sample point
2759 * in the past. This function provides those sample points.
2761 static inline errseq_t filemap_sample_wb_err(struct address_space *mapping)
2763 return errseq_sample(&mapping->wb_err);
2766 extern int vfs_fsync_range(struct file *file, loff_t start, loff_t end,
2768 extern int vfs_fsync(struct file *file, int datasync);
2771 * Sync the bytes written if this was a synchronous write. Expect ki_pos
2772 * to already be updated for the write, and will return either the amount
2773 * of bytes passed in, or an error if syncing the file failed.
2775 static inline ssize_t generic_write_sync(struct kiocb *iocb, ssize_t count)
2777 if (iocb->ki_flags & IOCB_DSYNC) {
2778 int ret = vfs_fsync_range(iocb->ki_filp,
2779 iocb->ki_pos - count, iocb->ki_pos - 1,
2780 (iocb->ki_flags & IOCB_SYNC) ? 0 : 1);
2788 extern void emergency_sync(void);
2789 extern void emergency_remount(void);
2791 extern sector_t bmap(struct inode *, sector_t);
2793 extern int notify_change(struct dentry *, struct iattr *, struct inode **);
2794 extern int inode_permission(struct inode *, int);
2795 extern int generic_permission(struct inode *, int);
2796 extern int __check_sticky(struct inode *dir, struct inode *inode);
2798 static inline bool execute_ok(struct inode *inode)
2800 return (inode->i_mode & S_IXUGO) || S_ISDIR(inode->i_mode);
2803 static inline void file_start_write(struct file *file)
2805 if (!S_ISREG(file_inode(file)->i_mode))
2807 __sb_start_write(file_inode(file)->i_sb, SB_FREEZE_WRITE, true);
2810 static inline bool file_start_write_trylock(struct file *file)
2812 if (!S_ISREG(file_inode(file)->i_mode))
2814 return __sb_start_write(file_inode(file)->i_sb, SB_FREEZE_WRITE, false);
2817 static inline void file_end_write(struct file *file)
2819 if (!S_ISREG(file_inode(file)->i_mode))
2821 __sb_end_write(file_inode(file)->i_sb, SB_FREEZE_WRITE);
2825 * get_write_access() gets write permission for a file.
2826 * put_write_access() releases this write permission.
2827 * This is used for regular files.
2828 * We cannot support write (and maybe mmap read-write shared) accesses and
2829 * MAP_DENYWRITE mmappings simultaneously. The i_writecount field of an inode
2830 * can have the following values:
2831 * 0: no writers, no VM_DENYWRITE mappings
2832 * < 0: (-i_writecount) vm_area_structs with VM_DENYWRITE set exist
2833 * > 0: (i_writecount) users are writing to the file.
2835 * Normally we operate on that counter with atomic_{inc,dec} and it's safe
2836 * except for the cases where we don't hold i_writecount yet. Then we need to
2837 * use {get,deny}_write_access() - these functions check the sign and refuse
2838 * to do the change if sign is wrong.
2840 static inline int get_write_access(struct inode *inode)
2842 return atomic_inc_unless_negative(&inode->i_writecount) ? 0 : -ETXTBSY;
2844 static inline int deny_write_access(struct file *file)
2846 struct inode *inode = file_inode(file);
2847 return atomic_dec_unless_positive(&inode->i_writecount) ? 0 : -ETXTBSY;
2849 static inline void put_write_access(struct inode * inode)
2851 atomic_dec(&inode->i_writecount);
2853 static inline void allow_write_access(struct file *file)
2856 atomic_inc(&file_inode(file)->i_writecount);
2858 static inline bool inode_is_open_for_write(const struct inode *inode)
2860 return atomic_read(&inode->i_writecount) > 0;
2864 static inline void i_readcount_dec(struct inode *inode)
2866 BUG_ON(!atomic_read(&inode->i_readcount));
2867 atomic_dec(&inode->i_readcount);
2869 static inline void i_readcount_inc(struct inode *inode)
2871 atomic_inc(&inode->i_readcount);
2874 static inline void i_readcount_dec(struct inode *inode)
2878 static inline void i_readcount_inc(struct inode *inode)
2883 extern int do_pipe_flags(int *, int);
2885 #define __kernel_read_file_id(id) \
2886 id(UNKNOWN, unknown) \
2887 id(FIRMWARE, firmware) \
2888 id(FIRMWARE_PREALLOC_BUFFER, firmware) \
2889 id(MODULE, kernel-module) \
2890 id(KEXEC_IMAGE, kexec-image) \
2891 id(KEXEC_INITRAMFS, kexec-initramfs) \
2892 id(POLICY, security-policy) \
2893 id(X509_CERTIFICATE, x509-certificate) \
2896 #define __fid_enumify(ENUM, dummy) READING_ ## ENUM,
2897 #define __fid_stringify(dummy, str) #str,
2899 enum kernel_read_file_id {
2900 __kernel_read_file_id(__fid_enumify)
2903 static const char * const kernel_read_file_str[] = {
2904 __kernel_read_file_id(__fid_stringify)
2907 static inline const char *kernel_read_file_id_str(enum kernel_read_file_id id)
2909 if ((unsigned)id >= READING_MAX_ID)
2910 return kernel_read_file_str[READING_UNKNOWN];
2912 return kernel_read_file_str[id];
2915 extern int kernel_read_file(struct file *, void **, loff_t *, loff_t,
2916 enum kernel_read_file_id);
2917 extern int kernel_read_file_from_path(const char *, void **, loff_t *, loff_t,
2918 enum kernel_read_file_id);
2919 extern int kernel_read_file_from_fd(int, void **, loff_t *, loff_t,
2920 enum kernel_read_file_id);
2921 extern ssize_t kernel_read(struct file *, void *, size_t, loff_t *);
2922 extern ssize_t kernel_write(struct file *, const void *, size_t, loff_t *);
2923 extern ssize_t __kernel_write(struct file *, const void *, size_t, loff_t *);
2924 extern struct file * open_exec(const char *);
2926 /* fs/dcache.c -- generic fs support functions */
2927 extern bool is_subdir(struct dentry *, struct dentry *);
2928 extern bool path_is_under(const struct path *, const struct path *);
2930 extern char *file_path(struct file *, char *, int);
2932 #include <linux/err.h>
2934 /* needed for stackable file system support */
2935 extern loff_t default_llseek(struct file *file, loff_t offset, int whence);
2937 extern loff_t vfs_llseek(struct file *file, loff_t offset, int whence);
2939 extern int inode_init_always(struct super_block *, struct inode *);
2940 extern void inode_init_once(struct inode *);
2941 extern void address_space_init_once(struct address_space *mapping);
2942 extern struct inode * igrab(struct inode *);
2943 extern ino_t iunique(struct super_block *, ino_t);
2944 extern int inode_needs_sync(struct inode *inode);
2945 extern int generic_delete_inode(struct inode *inode);
2946 static inline int generic_drop_inode(struct inode *inode)
2948 return !inode->i_nlink || inode_unhashed(inode);
2951 extern struct inode *ilookup5_nowait(struct super_block *sb,
2952 unsigned long hashval, int (*test)(struct inode *, void *),
2954 extern struct inode *ilookup5(struct super_block *sb, unsigned long hashval,
2955 int (*test)(struct inode *, void *), void *data);
2956 extern struct inode *ilookup(struct super_block *sb, unsigned long ino);
2958 extern struct inode *inode_insert5(struct inode *inode, unsigned long hashval,
2959 int (*test)(struct inode *, void *),
2960 int (*set)(struct inode *, void *),
2962 extern struct inode * iget5_locked(struct super_block *, unsigned long, int (*test)(struct inode *, void *), int (*set)(struct inode *, void *), void *);
2963 extern struct inode * iget_locked(struct super_block *, unsigned long);
2964 extern struct inode *find_inode_nowait(struct super_block *,
2966 int (*match)(struct inode *,
2967 unsigned long, void *),
2969 extern int insert_inode_locked4(struct inode *, unsigned long, int (*test)(struct inode *, void *), void *);
2970 extern int insert_inode_locked(struct inode *);
2971 #ifdef CONFIG_DEBUG_LOCK_ALLOC
2972 extern void lockdep_annotate_inode_mutex_key(struct inode *inode);
2974 static inline void lockdep_annotate_inode_mutex_key(struct inode *inode) { };
2976 extern void unlock_new_inode(struct inode *);
2977 extern void discard_new_inode(struct inode *);
2978 extern unsigned int get_next_ino(void);
2979 extern void evict_inodes(struct super_block *sb);
2981 extern void __iget(struct inode * inode);
2982 extern void iget_failed(struct inode *);
2983 extern void clear_inode(struct inode *);
2984 extern void __destroy_inode(struct inode *);
2985 extern struct inode *new_inode_pseudo(struct super_block *sb);
2986 extern struct inode *new_inode(struct super_block *sb);
2987 extern void free_inode_nonrcu(struct inode *inode);
2988 extern int should_remove_suid(struct dentry *);
2989 extern int file_remove_privs(struct file *);
2991 extern void __insert_inode_hash(struct inode *, unsigned long hashval);
2992 static inline void insert_inode_hash(struct inode *inode)
2994 __insert_inode_hash(inode, inode->i_ino);
2997 extern void __remove_inode_hash(struct inode *);
2998 static inline void remove_inode_hash(struct inode *inode)
3000 if (!inode_unhashed(inode) && !hlist_fake(&inode->i_hash))
3001 __remove_inode_hash(inode);
3004 extern void inode_sb_list_add(struct inode *inode);
3007 extern int bdev_read_only(struct block_device *);
3009 extern int set_blocksize(struct block_device *, int);
3010 extern int sb_set_blocksize(struct super_block *, int);
3011 extern int sb_min_blocksize(struct super_block *, int);
3013 extern int generic_file_mmap(struct file *, struct vm_area_struct *);
3014 extern int generic_file_readonly_mmap(struct file *, struct vm_area_struct *);
3015 extern ssize_t generic_write_checks(struct kiocb *, struct iov_iter *);
3016 extern int generic_remap_checks(struct file *file_in, loff_t pos_in,
3017 struct file *file_out, loff_t pos_out,
3018 loff_t *count, unsigned int remap_flags);
3019 extern ssize_t generic_file_read_iter(struct kiocb *, struct iov_iter *);
3020 extern ssize_t __generic_file_write_iter(struct kiocb *, struct iov_iter *);
3021 extern ssize_t generic_file_write_iter(struct kiocb *, struct iov_iter *);
3022 extern ssize_t generic_file_direct_write(struct kiocb *, struct iov_iter *);
3023 extern ssize_t generic_perform_write(struct file *, struct iov_iter *, loff_t);
3025 ssize_t vfs_iter_read(struct file *file, struct iov_iter *iter, loff_t *ppos,
3027 ssize_t vfs_iter_write(struct file *file, struct iov_iter *iter, loff_t *ppos,
3030 /* fs/block_dev.c */
3031 extern ssize_t blkdev_read_iter(struct kiocb *iocb, struct iov_iter *to);
3032 extern ssize_t blkdev_write_iter(struct kiocb *iocb, struct iov_iter *from);
3033 extern int blkdev_fsync(struct file *filp, loff_t start, loff_t end,
3035 extern void block_sync_page(struct page *page);
3038 extern ssize_t generic_file_splice_read(struct file *, loff_t *,
3039 struct pipe_inode_info *, size_t, unsigned int);
3040 extern ssize_t iter_file_splice_write(struct pipe_inode_info *,
3041 struct file *, loff_t *, size_t, unsigned int);
3042 extern ssize_t generic_splice_sendpage(struct pipe_inode_info *pipe,
3043 struct file *out, loff_t *, size_t len, unsigned int flags);
3044 extern long do_splice_direct(struct file *in, loff_t *ppos, struct file *out,
3045 loff_t *opos, size_t len, unsigned int flags);
3049 file_ra_state_init(struct file_ra_state *ra, struct address_space *mapping);
3050 extern loff_t noop_llseek(struct file *file, loff_t offset, int whence);
3051 extern loff_t no_llseek(struct file *file, loff_t offset, int whence);
3052 extern loff_t vfs_setpos(struct file *file, loff_t offset, loff_t maxsize);
3053 extern loff_t generic_file_llseek(struct file *file, loff_t offset, int whence);
3054 extern loff_t generic_file_llseek_size(struct file *file, loff_t offset,
3055 int whence, loff_t maxsize, loff_t eof);
3056 extern loff_t fixed_size_llseek(struct file *file, loff_t offset,
3057 int whence, loff_t size);
3058 extern loff_t no_seek_end_llseek_size(struct file *, loff_t, int, loff_t);
3059 extern loff_t no_seek_end_llseek(struct file *, loff_t, int);
3060 extern int generic_file_open(struct inode * inode, struct file * filp);
3061 extern int nonseekable_open(struct inode * inode, struct file * filp);
3064 typedef void (dio_submit_t)(struct bio *bio, struct inode *inode,
3065 loff_t file_offset);
3068 /* need locking between buffered and direct access */
3071 /* filesystem does not support filling holes */
3072 DIO_SKIP_HOLES = 0x02,
3075 void dio_end_io(struct bio *bio);
3076 void dio_warn_stale_pagecache(struct file *filp);
3078 ssize_t __blockdev_direct_IO(struct kiocb *iocb, struct inode *inode,
3079 struct block_device *bdev, struct iov_iter *iter,
3080 get_block_t get_block,
3081 dio_iodone_t end_io, dio_submit_t submit_io,
3084 static inline ssize_t blockdev_direct_IO(struct kiocb *iocb,
3085 struct inode *inode,
3086 struct iov_iter *iter,
3087 get_block_t get_block)
3089 return __blockdev_direct_IO(iocb, inode, inode->i_sb->s_bdev, iter,
3090 get_block, NULL, NULL, DIO_LOCKING | DIO_SKIP_HOLES);
3094 void inode_dio_wait(struct inode *inode);
3097 * inode_dio_begin - signal start of a direct I/O requests
3098 * @inode: inode the direct I/O happens on
3100 * This is called once we've finished processing a direct I/O request,
3101 * and is used to wake up callers waiting for direct I/O to be quiesced.
3103 static inline void inode_dio_begin(struct inode *inode)
3105 atomic_inc(&inode->i_dio_count);
3109 * inode_dio_end - signal finish of a direct I/O requests
3110 * @inode: inode the direct I/O happens on
3112 * This is called once we've finished processing a direct I/O request,
3113 * and is used to wake up callers waiting for direct I/O to be quiesced.
3115 static inline void inode_dio_end(struct inode *inode)
3117 if (atomic_dec_and_test(&inode->i_dio_count))
3118 wake_up_bit(&inode->i_state, __I_DIO_WAKEUP);
3121 extern void inode_set_flags(struct inode *inode, unsigned int flags,
3124 extern const struct file_operations generic_ro_fops;
3126 #define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
3128 extern int readlink_copy(char __user *, int, const char *);
3129 extern int page_readlink(struct dentry *, char __user *, int);
3130 extern const char *page_get_link(struct dentry *, struct inode *,
3131 struct delayed_call *);
3132 extern void page_put_link(void *);
3133 extern int __page_symlink(struct inode *inode, const char *symname, int len,
3135 extern int page_symlink(struct inode *inode, const char *symname, int len);
3136 extern const struct inode_operations page_symlink_inode_operations;
3137 extern void kfree_link(void *);
3138 extern void generic_fillattr(struct inode *, struct kstat *);
3139 extern int vfs_getattr_nosec(const struct path *, struct kstat *, u32, unsigned int);
3140 extern int vfs_getattr(const struct path *, struct kstat *, u32, unsigned int);
3141 void __inode_add_bytes(struct inode *inode, loff_t bytes);
3142 void inode_add_bytes(struct inode *inode, loff_t bytes);
3143 void __inode_sub_bytes(struct inode *inode, loff_t bytes);
3144 void inode_sub_bytes(struct inode *inode, loff_t bytes);
3145 static inline loff_t __inode_get_bytes(struct inode *inode)
3147 return (((loff_t)inode->i_blocks) << 9) + inode->i_bytes;
3149 loff_t inode_get_bytes(struct inode *inode);
3150 void inode_set_bytes(struct inode *inode, loff_t bytes);
3151 const char *simple_get_link(struct dentry *, struct inode *,
3152 struct delayed_call *);
3153 extern const struct inode_operations simple_symlink_inode_operations;
3155 extern int iterate_dir(struct file *, struct dir_context *);
3157 extern int vfs_statx(int, const char __user *, int, struct kstat *, u32);
3158 extern int vfs_statx_fd(unsigned int, struct kstat *, u32, unsigned int);
3160 static inline int vfs_stat(const char __user *filename, struct kstat *stat)
3162 return vfs_statx(AT_FDCWD, filename, AT_NO_AUTOMOUNT,
3163 stat, STATX_BASIC_STATS);
3165 static inline int vfs_lstat(const char __user *name, struct kstat *stat)
3167 return vfs_statx(AT_FDCWD, name, AT_SYMLINK_NOFOLLOW | AT_NO_AUTOMOUNT,
3168 stat, STATX_BASIC_STATS);
3170 static inline int vfs_fstatat(int dfd, const char __user *filename,
3171 struct kstat *stat, int flags)
3173 return vfs_statx(dfd, filename, flags | AT_NO_AUTOMOUNT,
3174 stat, STATX_BASIC_STATS);
3176 static inline int vfs_fstat(int fd, struct kstat *stat)
3178 return vfs_statx_fd(fd, stat, STATX_BASIC_STATS, 0);
3182 extern const char *vfs_get_link(struct dentry *, struct delayed_call *);
3183 extern int vfs_readlink(struct dentry *, char __user *, int);
3185 extern int __generic_block_fiemap(struct inode *inode,
3186 struct fiemap_extent_info *fieinfo,
3187 loff_t start, loff_t len,
3188 get_block_t *get_block);
3189 extern int generic_block_fiemap(struct inode *inode,
3190 struct fiemap_extent_info *fieinfo, u64 start,
3191 u64 len, get_block_t *get_block);
3193 extern struct file_system_type *get_filesystem(struct file_system_type *fs);
3194 extern void put_filesystem(struct file_system_type *fs);
3195 extern struct file_system_type *get_fs_type(const char *name);
3196 extern struct super_block *get_super(struct block_device *);
3197 extern struct super_block *get_super_thawed(struct block_device *);
3198 extern struct super_block *get_super_exclusive_thawed(struct block_device *bdev);
3199 extern struct super_block *get_active_super(struct block_device *bdev);
3200 extern void drop_super(struct super_block *sb);
3201 extern void drop_super_exclusive(struct super_block *sb);
3202 extern void iterate_supers(void (*)(struct super_block *, void *), void *);
3203 extern void iterate_supers_type(struct file_system_type *,
3204 void (*)(struct super_block *, void *), void *);
3206 extern int dcache_dir_open(struct inode *, struct file *);
3207 extern int dcache_dir_close(struct inode *, struct file *);
3208 extern loff_t dcache_dir_lseek(struct file *, loff_t, int);
3209 extern int dcache_readdir(struct file *, struct dir_context *);
3210 extern int simple_setattr(struct dentry *, struct iattr *);
3211 extern int simple_getattr(const struct path *, struct kstat *, u32, unsigned int);
3212 extern int simple_statfs(struct dentry *, struct kstatfs *);
3213 extern int simple_open(struct inode *inode, struct file *file);
3214 extern int simple_link(struct dentry *, struct inode *, struct dentry *);
3215 extern int simple_unlink(struct inode *, struct dentry *);
3216 extern int simple_rmdir(struct inode *, struct dentry *);
3217 extern int simple_rename(struct inode *, struct dentry *,
3218 struct inode *, struct dentry *, unsigned int);
3219 extern int noop_fsync(struct file *, loff_t, loff_t, int);
3220 extern int noop_set_page_dirty(struct page *page);
3221 extern void noop_invalidatepage(struct page *page, unsigned int offset,
3222 unsigned int length);
3223 extern ssize_t noop_direct_IO(struct kiocb *iocb, struct iov_iter *iter);
3224 extern int simple_empty(struct dentry *);
3225 extern int simple_readpage(struct file *file, struct page *page);
3226 extern int simple_write_begin(struct file *file, struct address_space *mapping,
3227 loff_t pos, unsigned len, unsigned flags,
3228 struct page **pagep, void **fsdata);
3229 extern int simple_write_end(struct file *file, struct address_space *mapping,
3230 loff_t pos, unsigned len, unsigned copied,
3231 struct page *page, void *fsdata);
3232 extern int always_delete_dentry(const struct dentry *);
3233 extern struct inode *alloc_anon_inode(struct super_block *);
3234 extern int simple_nosetlease(struct file *, long, struct file_lock **, void **);
3235 extern const struct dentry_operations simple_dentry_operations;
3237 extern struct dentry *simple_lookup(struct inode *, struct dentry *, unsigned int flags);
3238 extern ssize_t generic_read_dir(struct file *, char __user *, size_t, loff_t *);
3239 extern const struct file_operations simple_dir_operations;
3240 extern const struct inode_operations simple_dir_inode_operations;
3241 extern void make_empty_dir_inode(struct inode *inode);
3242 extern bool is_empty_dir_inode(struct inode *inode);
3243 struct tree_descr { const char *name; const struct file_operations *ops; int mode; };
3244 struct dentry *d_alloc_name(struct dentry *, const char *);
3245 extern int simple_fill_super(struct super_block *, unsigned long,
3246 const struct tree_descr *);
3247 extern int simple_pin_fs(struct file_system_type *, struct vfsmount **mount, int *count);
3248 extern void simple_release_fs(struct vfsmount **mount, int *count);
3250 extern ssize_t simple_read_from_buffer(void __user *to, size_t count,
3251 loff_t *ppos, const void *from, size_t available);
3252 extern ssize_t simple_write_to_buffer(void *to, size_t available, loff_t *ppos,
3253 const void __user *from, size_t count);
3255 extern int __generic_file_fsync(struct file *, loff_t, loff_t, int);
3256 extern int generic_file_fsync(struct file *, loff_t, loff_t, int);
3258 extern int generic_check_addressable(unsigned, u64);
3260 #ifdef CONFIG_MIGRATION
3261 extern int buffer_migrate_page(struct address_space *,
3262 struct page *, struct page *,
3264 extern int buffer_migrate_page_norefs(struct address_space *,
3265 struct page *, struct page *,
3268 #define buffer_migrate_page NULL
3269 #define buffer_migrate_page_norefs NULL
3272 extern int setattr_prepare(struct dentry *, struct iattr *);
3273 extern int inode_newsize_ok(const struct inode *, loff_t offset);
3274 extern void setattr_copy(struct inode *inode, const struct iattr *attr);
3276 extern int file_update_time(struct file *file);
3278 static inline bool io_is_direct(struct file *filp)
3280 return (filp->f_flags & O_DIRECT) || IS_DAX(filp->f_mapping->host);
3283 static inline bool vma_is_dax(struct vm_area_struct *vma)
3285 return vma->vm_file && IS_DAX(vma->vm_file->f_mapping->host);
3288 static inline bool vma_is_fsdax(struct vm_area_struct *vma)
3290 struct inode *inode;
3294 if (!vma_is_dax(vma))
3296 inode = file_inode(vma->vm_file);
3297 if (S_ISCHR(inode->i_mode))
3298 return false; /* device-dax */
3302 static inline int iocb_flags(struct file *file)
3305 if (file->f_flags & O_APPEND)
3307 if (io_is_direct(file))
3309 if ((file->f_flags & O_DSYNC) || IS_SYNC(file->f_mapping->host))
3311 if (file->f_flags & __O_SYNC)
3316 static inline int kiocb_set_rw_flags(struct kiocb *ki, rwf_t flags)
3318 if (unlikely(flags & ~RWF_SUPPORTED))
3321 if (flags & RWF_NOWAIT) {
3322 if (!(ki->ki_filp->f_mode & FMODE_NOWAIT))
3324 ki->ki_flags |= IOCB_NOWAIT;
3326 if (flags & RWF_HIPRI)
3327 ki->ki_flags |= IOCB_HIPRI;
3328 if (flags & RWF_DSYNC)
3329 ki->ki_flags |= IOCB_DSYNC;
3330 if (flags & RWF_SYNC)
3331 ki->ki_flags |= (IOCB_DSYNC | IOCB_SYNC);
3332 if (flags & RWF_APPEND)
3333 ki->ki_flags |= IOCB_APPEND;
3337 static inline ino_t parent_ino(struct dentry *dentry)
3342 * Don't strictly need d_lock here? If the parent ino could change
3343 * then surely we'd have a deeper race in the caller?
3345 spin_lock(&dentry->d_lock);
3346 res = dentry->d_parent->d_inode->i_ino;
3347 spin_unlock(&dentry->d_lock);
3351 /* Transaction based IO helpers */
3354 * An argresp is stored in an allocated page and holds the
3355 * size of the argument or response, along with its content
3357 struct simple_transaction_argresp {
3362 #define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
3364 char *simple_transaction_get(struct file *file, const char __user *buf,
3366 ssize_t simple_transaction_read(struct file *file, char __user *buf,
3367 size_t size, loff_t *pos);
3368 int simple_transaction_release(struct inode *inode, struct file *file);
3370 void simple_transaction_set(struct file *file, size_t n);
3373 * simple attribute files
3375 * These attributes behave similar to those in sysfs:
3377 * Writing to an attribute immediately sets a value, an open file can be
3378 * written to multiple times.
3380 * Reading from an attribute creates a buffer from the value that might get
3381 * read with multiple read calls. When the attribute has been read
3382 * completely, no further read calls are possible until the file is opened
3385 * All attributes contain a text representation of a numeric value
3386 * that are accessed with the get() and set() functions.
3388 #define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt) \
3389 static int __fops ## _open(struct inode *inode, struct file *file) \
3391 __simple_attr_check_format(__fmt, 0ull); \
3392 return simple_attr_open(inode, file, __get, __set, __fmt); \
3394 static const struct file_operations __fops = { \
3395 .owner = THIS_MODULE, \
3396 .open = __fops ## _open, \
3397 .release = simple_attr_release, \
3398 .read = simple_attr_read, \
3399 .write = simple_attr_write, \
3400 .llseek = generic_file_llseek, \
3403 static inline __printf(1, 2)
3404 void __simple_attr_check_format(const char *fmt, ...)
3406 /* don't do anything, just let the compiler check the arguments; */
3409 int simple_attr_open(struct inode *inode, struct file *file,
3410 int (*get)(void *, u64 *), int (*set)(void *, u64),
3412 int simple_attr_release(struct inode *inode, struct file *file);
3413 ssize_t simple_attr_read(struct file *file, char __user *buf,
3414 size_t len, loff_t *ppos);
3415 ssize_t simple_attr_write(struct file *file, const char __user *buf,
3416 size_t len, loff_t *ppos);
3419 int proc_nr_files(struct ctl_table *table, int write,
3420 void __user *buffer, size_t *lenp, loff_t *ppos);
3421 int proc_nr_dentry(struct ctl_table *table, int write,
3422 void __user *buffer, size_t *lenp, loff_t *ppos);
3423 int proc_nr_inodes(struct ctl_table *table, int write,
3424 void __user *buffer, size_t *lenp, loff_t *ppos);
3425 int __init get_filesystem_list(char *buf);
3427 #define __FMODE_EXEC ((__force int) FMODE_EXEC)
3428 #define __FMODE_NONOTIFY ((__force int) FMODE_NONOTIFY)
3430 #define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE])
3431 #define OPEN_FMODE(flag) ((__force fmode_t)(((flag + 1) & O_ACCMODE) | \
3432 (flag & __FMODE_NONOTIFY)))
3434 static inline bool is_sxid(umode_t mode)
3436 return (mode & S_ISUID) || ((mode & S_ISGID) && (mode & S_IXGRP));
3439 static inline int check_sticky(struct inode *dir, struct inode *inode)
3441 if (!(dir->i_mode & S_ISVTX))
3444 return __check_sticky(dir, inode);
3447 static inline void inode_has_no_xattr(struct inode *inode)
3449 if (!is_sxid(inode->i_mode) && (inode->i_sb->s_flags & SB_NOSEC))
3450 inode->i_flags |= S_NOSEC;
3453 static inline bool is_root_inode(struct inode *inode)
3455 return inode == inode->i_sb->s_root->d_inode;
3458 static inline bool dir_emit(struct dir_context *ctx,
3459 const char *name, int namelen,
3460 u64 ino, unsigned type)
3462 return ctx->actor(ctx, name, namelen, ctx->pos, ino, type) == 0;
3464 static inline bool dir_emit_dot(struct file *file, struct dir_context *ctx)
3466 return ctx->actor(ctx, ".", 1, ctx->pos,
3467 file->f_path.dentry->d_inode->i_ino, DT_DIR) == 0;
3469 static inline bool dir_emit_dotdot(struct file *file, struct dir_context *ctx)
3471 return ctx->actor(ctx, "..", 2, ctx->pos,
3472 parent_ino(file->f_path.dentry), DT_DIR) == 0;
3474 static inline bool dir_emit_dots(struct file *file, struct dir_context *ctx)
3476 if (ctx->pos == 0) {
3477 if (!dir_emit_dot(file, ctx))
3481 if (ctx->pos == 1) {
3482 if (!dir_emit_dotdot(file, ctx))
3488 static inline bool dir_relax(struct inode *inode)
3490 inode_unlock(inode);
3492 return !IS_DEADDIR(inode);
3495 static inline bool dir_relax_shared(struct inode *inode)
3497 inode_unlock_shared(inode);
3498 inode_lock_shared(inode);
3499 return !IS_DEADDIR(inode);
3502 extern bool path_noexec(const struct path *path);
3503 extern void inode_nohighmem(struct inode *inode);
3506 extern int vfs_fadvise(struct file *file, loff_t offset, loff_t len,
3509 #endif /* _LINUX_FS_H */