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>
41 #include <asm/byteorder.h>
42 #include <uapi/linux/fs.h>
44 struct backing_dev_info;
47 struct export_operations;
52 struct pipe_inode_info;
53 struct poll_table_struct;
55 struct vm_area_struct;
58 struct swap_info_struct;
60 struct workqueue_struct;
63 struct fscrypt_operations;
65 extern void __init inode_init(void);
66 extern void __init inode_init_early(void);
67 extern void __init files_init(void);
68 extern void __init files_maxfiles_init(void);
70 extern struct files_stat_struct files_stat;
71 extern unsigned long get_max_files(void);
72 extern unsigned int sysctl_nr_open;
73 extern struct inodes_stat_t inodes_stat;
74 extern int leases_enable, lease_break_time;
75 extern int sysctl_protected_symlinks;
76 extern int sysctl_protected_hardlinks;
77 extern int sysctl_protected_fifos;
78 extern int sysctl_protected_regular;
80 typedef __kernel_rwf_t rwf_t;
83 typedef int (get_block_t)(struct inode *inode, sector_t iblock,
84 struct buffer_head *bh_result, int create);
85 typedef int (dio_iodone_t)(struct kiocb *iocb, loff_t offset,
86 ssize_t bytes, void *private);
88 #define MAY_EXEC 0x00000001
89 #define MAY_WRITE 0x00000002
90 #define MAY_READ 0x00000004
91 #define MAY_APPEND 0x00000008
92 #define MAY_ACCESS 0x00000010
93 #define MAY_OPEN 0x00000020
94 #define MAY_CHDIR 0x00000040
95 /* called from RCU mode, don't block */
96 #define MAY_NOT_BLOCK 0x00000080
99 * flags in file.f_mode. Note that FMODE_READ and FMODE_WRITE must correspond
100 * to O_WRONLY and O_RDWR via the strange trick in do_dentry_open()
103 /* file is open for reading */
104 #define FMODE_READ ((__force fmode_t)0x1)
105 /* file is open for writing */
106 #define FMODE_WRITE ((__force fmode_t)0x2)
107 /* file is seekable */
108 #define FMODE_LSEEK ((__force fmode_t)0x4)
109 /* file can be accessed using pread */
110 #define FMODE_PREAD ((__force fmode_t)0x8)
111 /* file can be accessed using pwrite */
112 #define FMODE_PWRITE ((__force fmode_t)0x10)
113 /* File is opened for execution with sys_execve / sys_uselib */
114 #define FMODE_EXEC ((__force fmode_t)0x20)
115 /* File is opened with O_NDELAY (only set for block devices) */
116 #define FMODE_NDELAY ((__force fmode_t)0x40)
117 /* File is opened with O_EXCL (only set for block devices) */
118 #define FMODE_EXCL ((__force fmode_t)0x80)
119 /* File is opened using open(.., 3, ..) and is writeable only for ioctls
120 (specialy hack for floppy.c) */
121 #define FMODE_WRITE_IOCTL ((__force fmode_t)0x100)
122 /* 32bit hashes as llseek() offset (for directories) */
123 #define FMODE_32BITHASH ((__force fmode_t)0x200)
124 /* 64bit hashes as llseek() offset (for directories) */
125 #define FMODE_64BITHASH ((__force fmode_t)0x400)
128 * Don't update ctime and mtime.
130 * Currently a special hack for the XFS open_by_handle ioctl, but we'll
131 * hopefully graduate it to a proper O_CMTIME flag supported by open(2) soon.
133 #define FMODE_NOCMTIME ((__force fmode_t)0x800)
135 /* Expect random access pattern */
136 #define FMODE_RANDOM ((__force fmode_t)0x1000)
138 /* File is huge (eg. /dev/kmem): treat loff_t as unsigned */
139 #define FMODE_UNSIGNED_OFFSET ((__force fmode_t)0x2000)
141 /* File is opened with O_PATH; almost nothing can be done with it */
142 #define FMODE_PATH ((__force fmode_t)0x4000)
144 /* File needs atomic accesses to f_pos */
145 #define FMODE_ATOMIC_POS ((__force fmode_t)0x8000)
146 /* Write access to underlying fs */
147 #define FMODE_WRITER ((__force fmode_t)0x10000)
148 /* Has read method(s) */
149 #define FMODE_CAN_READ ((__force fmode_t)0x20000)
150 /* Has write method(s) */
151 #define FMODE_CAN_WRITE ((__force fmode_t)0x40000)
153 #define FMODE_OPENED ((__force fmode_t)0x80000)
154 #define FMODE_CREATED ((__force fmode_t)0x100000)
156 /* File was opened by fanotify and shouldn't generate fanotify events */
157 #define FMODE_NONOTIFY ((__force fmode_t)0x4000000)
159 /* File is capable of returning -EAGAIN if I/O will block */
160 #define FMODE_NOWAIT ((__force fmode_t)0x8000000)
162 /* File does not contribute to nr_files count */
163 #define FMODE_NOACCOUNT ((__force fmode_t)0x20000000)
166 * Flag for rw_copy_check_uvector and compat_rw_copy_check_uvector
167 * that indicates that they should check the contents of the iovec are
168 * valid, but not check the memory that the iovec elements
171 #define CHECK_IOVEC_ONLY -1
174 * Attribute flags. These should be or-ed together to figure out what
177 #define ATTR_MODE (1 << 0)
178 #define ATTR_UID (1 << 1)
179 #define ATTR_GID (1 << 2)
180 #define ATTR_SIZE (1 << 3)
181 #define ATTR_ATIME (1 << 4)
182 #define ATTR_MTIME (1 << 5)
183 #define ATTR_CTIME (1 << 6)
184 #define ATTR_ATIME_SET (1 << 7)
185 #define ATTR_MTIME_SET (1 << 8)
186 #define ATTR_FORCE (1 << 9) /* Not a change, but a change it */
187 #define ATTR_KILL_SUID (1 << 11)
188 #define ATTR_KILL_SGID (1 << 12)
189 #define ATTR_FILE (1 << 13)
190 #define ATTR_KILL_PRIV (1 << 14)
191 #define ATTR_OPEN (1 << 15) /* Truncating from open(O_TRUNC) */
192 #define ATTR_TIMES_SET (1 << 16)
193 #define ATTR_TOUCH (1 << 17)
196 * Whiteout is represented by a char device. The following constants define the
197 * mode and device number to use.
199 #define WHITEOUT_MODE 0
200 #define WHITEOUT_DEV 0
203 * This is the Inode Attributes structure, used for notify_change(). It
204 * uses the above definitions as flags, to know which values have changed.
205 * Also, in this manner, a Filesystem can look at only the values it cares
206 * about. Basically, these are the attributes that the VFS layer can
207 * request to change from the FS layer.
209 * Derek Atkins <warlord@MIT.EDU> 94-10-20
212 unsigned int ia_valid;
217 struct timespec64 ia_atime;
218 struct timespec64 ia_mtime;
219 struct timespec64 ia_ctime;
222 * Not an attribute, but an auxiliary info for filesystems wanting to
223 * implement an ftruncate() like method. NOTE: filesystem should
224 * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL).
226 struct file *ia_file;
230 * Includes for diskquotas.
232 #include <linux/quota.h>
235 * Maximum number of layers of fs stack. Needs to be limited to
236 * prevent kernel stack overflow
238 #define FILESYSTEM_MAX_STACK_DEPTH 2
241 * enum positive_aop_returns - aop return codes with specific semantics
243 * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has
244 * completed, that the page is still locked, and
245 * should be considered active. The VM uses this hint
246 * to return the page to the active list -- it won't
247 * be a candidate for writeback again in the near
248 * future. Other callers must be careful to unlock
249 * the page if they get this return. Returned by
252 * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has
253 * unlocked it and the page might have been truncated.
254 * The caller should back up to acquiring a new page and
255 * trying again. The aop will be taking reasonable
256 * precautions not to livelock. If the caller held a page
257 * reference, it should drop it before retrying. Returned
260 * address_space_operation functions return these large constants to indicate
261 * special semantics to the caller. These are much larger than the bytes in a
262 * page to allow for functions that return the number of bytes operated on in a
266 enum positive_aop_returns {
267 AOP_WRITEPAGE_ACTIVATE = 0x80000,
268 AOP_TRUNCATED_PAGE = 0x80001,
271 #define AOP_FLAG_CONT_EXPAND 0x0001 /* called from cont_expand */
272 #define AOP_FLAG_NOFS 0x0002 /* used by filesystem to direct
273 * helper code (eg buffer layer)
274 * to clear GFP_FS from alloc */
277 * oh the beauties of C type declarations.
280 struct address_space;
281 struct writeback_control;
284 * Write life time hint values.
285 * Stored in struct inode as u8.
288 WRITE_LIFE_NOT_SET = 0,
289 WRITE_LIFE_NONE = RWH_WRITE_LIFE_NONE,
290 WRITE_LIFE_SHORT = RWH_WRITE_LIFE_SHORT,
291 WRITE_LIFE_MEDIUM = RWH_WRITE_LIFE_MEDIUM,
292 WRITE_LIFE_LONG = RWH_WRITE_LIFE_LONG,
293 WRITE_LIFE_EXTREME = RWH_WRITE_LIFE_EXTREME,
296 #define IOCB_EVENTFD (1 << 0)
297 #define IOCB_APPEND (1 << 1)
298 #define IOCB_DIRECT (1 << 2)
299 #define IOCB_HIPRI (1 << 3)
300 #define IOCB_DSYNC (1 << 4)
301 #define IOCB_SYNC (1 << 5)
302 #define IOCB_WRITE (1 << 6)
303 #define IOCB_NOWAIT (1 << 7)
306 struct file *ki_filp;
308 void (*ki_complete)(struct kiocb *iocb, long ret, long ret2);
312 u16 ki_ioprio; /* See linux/ioprio.h */
313 } __randomize_layout;
315 static inline bool is_sync_kiocb(struct kiocb *kiocb)
317 return kiocb->ki_complete == NULL;
321 * "descriptor" for what we're up to with a read.
322 * This allows us to use the same read code yet
323 * have multiple different users of the data that
324 * we read from a file.
326 * The simplest case just copies the data to user
339 typedef int (*read_actor_t)(read_descriptor_t *, struct page *,
340 unsigned long, unsigned long);
342 struct address_space_operations {
343 int (*writepage)(struct page *page, struct writeback_control *wbc);
344 int (*readpage)(struct file *, struct page *);
346 /* Write back some dirty pages from this mapping. */
347 int (*writepages)(struct address_space *, struct writeback_control *);
349 /* Set a page dirty. Return true if this dirtied it */
350 int (*set_page_dirty)(struct page *page);
353 * Reads in the requested pages. Unlike ->readpage(), this is
354 * PURELY used for read-ahead!.
356 int (*readpages)(struct file *filp, struct address_space *mapping,
357 struct list_head *pages, unsigned nr_pages);
359 int (*write_begin)(struct file *, struct address_space *mapping,
360 loff_t pos, unsigned len, unsigned flags,
361 struct page **pagep, void **fsdata);
362 int (*write_end)(struct file *, struct address_space *mapping,
363 loff_t pos, unsigned len, unsigned copied,
364 struct page *page, void *fsdata);
366 /* Unfortunately this kludge is needed for FIBMAP. Don't use it */
367 sector_t (*bmap)(struct address_space *, sector_t);
368 void (*invalidatepage) (struct page *, unsigned int, unsigned int);
369 int (*releasepage) (struct page *, gfp_t);
370 void (*freepage)(struct page *);
371 ssize_t (*direct_IO)(struct kiocb *, struct iov_iter *iter);
373 * migrate the contents of a page to the specified target. If
374 * migrate_mode is MIGRATE_ASYNC, it must not block.
376 int (*migratepage) (struct address_space *,
377 struct page *, struct page *, enum migrate_mode);
378 bool (*isolate_page)(struct page *, isolate_mode_t);
379 void (*putback_page)(struct page *);
380 int (*launder_page) (struct page *);
381 int (*is_partially_uptodate) (struct page *, unsigned long,
383 void (*is_dirty_writeback) (struct page *, bool *, bool *);
384 int (*error_remove_page)(struct address_space *, struct page *);
386 /* swapfile support */
387 int (*swap_activate)(struct swap_info_struct *sis, struct file *file,
389 void (*swap_deactivate)(struct file *file);
392 extern const struct address_space_operations empty_aops;
395 * pagecache_write_begin/pagecache_write_end must be used by general code
396 * to write into the pagecache.
398 int pagecache_write_begin(struct file *, struct address_space *mapping,
399 loff_t pos, unsigned len, unsigned flags,
400 struct page **pagep, void **fsdata);
402 int pagecache_write_end(struct file *, struct address_space *mapping,
403 loff_t pos, unsigned len, unsigned copied,
404 struct page *page, void *fsdata);
407 * struct address_space - Contents of a cacheable, mappable object.
408 * @host: Owner, either the inode or the block_device.
409 * @i_pages: Cached pages.
410 * @gfp_mask: Memory allocation flags to use for allocating pages.
411 * @i_mmap_writable: Number of VM_SHARED mappings.
412 * @i_mmap: Tree of private and shared mappings.
413 * @i_mmap_rwsem: Protects @i_mmap and @i_mmap_writable.
414 * @nrpages: Number of page entries, protected by the i_pages lock.
415 * @nrexceptional: Shadow or DAX entries, protected by the i_pages lock.
416 * @writeback_index: Writeback starts here.
418 * @flags: Error bits and flags (AS_*).
419 * @wb_err: The most recent error which has occurred.
420 * @private_lock: For use by the owner of the address_space.
421 * @private_list: For use by the owner of the address_space.
422 * @private_data: For use by the owner of the address_space.
424 struct address_space {
426 struct xarray i_pages;
428 atomic_t i_mmap_writable;
429 struct rb_root_cached i_mmap;
430 struct rw_semaphore i_mmap_rwsem;
431 unsigned long nrpages;
432 unsigned long nrexceptional;
433 pgoff_t writeback_index;
434 const struct address_space_operations *a_ops;
437 spinlock_t private_lock;
438 struct list_head private_list;
440 } __attribute__((aligned(sizeof(long)))) __randomize_layout;
442 * On most architectures that alignment is already the case; but
443 * must be enforced here for CRIS, to let the least significant bit
444 * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
446 struct request_queue;
448 struct block_device {
449 dev_t bd_dev; /* not a kdev_t - it's a search key */
451 struct inode * bd_inode; /* will die */
452 struct super_block * bd_super;
453 struct mutex bd_mutex; /* open/close mutex */
457 bool bd_write_holder;
459 struct list_head bd_holder_disks;
461 struct block_device * bd_contains;
462 unsigned bd_block_size;
464 struct hd_struct * bd_part;
465 /* number of times partitions within this device have been opened. */
466 unsigned bd_part_count;
468 struct gendisk * bd_disk;
469 struct request_queue * bd_queue;
470 struct backing_dev_info *bd_bdi;
471 struct list_head bd_list;
473 * Private data. You must have bd_claim'ed the block_device
474 * to use this. NOTE: bd_claim allows an owner to claim
475 * the same device multiple times, the owner must take special
476 * care to not mess up bd_private for that case.
478 unsigned long bd_private;
480 /* The counter of freeze processes */
481 int bd_fsfreeze_count;
482 /* Mutex for freeze */
483 struct mutex bd_fsfreeze_mutex;
484 } __randomize_layout;
486 /* XArray tags, for tagging dirty and writeback pages in the pagecache. */
487 #define PAGECACHE_TAG_DIRTY XA_MARK_0
488 #define PAGECACHE_TAG_WRITEBACK XA_MARK_1
489 #define PAGECACHE_TAG_TOWRITE XA_MARK_2
492 * Returns true if any of the pages in the mapping are marked with the tag.
494 static inline bool mapping_tagged(struct address_space *mapping, xa_mark_t tag)
496 return xa_marked(&mapping->i_pages, tag);
499 static inline void i_mmap_lock_write(struct address_space *mapping)
501 down_write(&mapping->i_mmap_rwsem);
504 static inline void i_mmap_unlock_write(struct address_space *mapping)
506 up_write(&mapping->i_mmap_rwsem);
509 static inline void i_mmap_lock_read(struct address_space *mapping)
511 down_read(&mapping->i_mmap_rwsem);
514 static inline void i_mmap_unlock_read(struct address_space *mapping)
516 up_read(&mapping->i_mmap_rwsem);
520 * Might pages of this file be mapped into userspace?
522 static inline int mapping_mapped(struct address_space *mapping)
524 return !RB_EMPTY_ROOT(&mapping->i_mmap.rb_root);
528 * Might pages of this file have been modified in userspace?
529 * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap_pgoff
530 * marks vma as VM_SHARED if it is shared, and the file was opened for
531 * writing i.e. vma may be mprotected writable even if now readonly.
533 * If i_mmap_writable is negative, no new writable mappings are allowed. You
534 * can only deny writable mappings, if none exists right now.
536 static inline int mapping_writably_mapped(struct address_space *mapping)
538 return atomic_read(&mapping->i_mmap_writable) > 0;
541 static inline int mapping_map_writable(struct address_space *mapping)
543 return atomic_inc_unless_negative(&mapping->i_mmap_writable) ?
547 static inline void mapping_unmap_writable(struct address_space *mapping)
549 atomic_dec(&mapping->i_mmap_writable);
552 static inline int mapping_deny_writable(struct address_space *mapping)
554 return atomic_dec_unless_positive(&mapping->i_mmap_writable) ?
558 static inline void mapping_allow_writable(struct address_space *mapping)
560 atomic_inc(&mapping->i_mmap_writable);
564 * Use sequence counter to get consistent i_size on 32-bit processors.
566 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
567 #include <linux/seqlock.h>
568 #define __NEED_I_SIZE_ORDERED
569 #define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
571 #define i_size_ordered_init(inode) do { } while (0)
575 #define ACL_NOT_CACHED ((void *)(-1))
576 #define ACL_DONT_CACHE ((void *)(-3))
578 static inline struct posix_acl *
579 uncached_acl_sentinel(struct task_struct *task)
581 return (void *)task + 1;
585 is_uncached_acl(struct posix_acl *acl)
587 return (long)acl & 1;
590 #define IOP_FASTPERM 0x0001
591 #define IOP_LOOKUP 0x0002
592 #define IOP_NOFOLLOW 0x0004
593 #define IOP_XATTR 0x0008
594 #define IOP_DEFAULT_READLINK 0x0010
596 struct fsnotify_mark_connector;
599 * Keep mostly read-only and often accessed (especially for
600 * the RCU path lookup and 'stat' data) fields at the beginning
601 * of the 'struct inode'
605 unsigned short i_opflags;
608 unsigned int i_flags;
610 #ifdef CONFIG_FS_POSIX_ACL
611 struct posix_acl *i_acl;
612 struct posix_acl *i_default_acl;
615 const struct inode_operations *i_op;
616 struct super_block *i_sb;
617 struct address_space *i_mapping;
619 #ifdef CONFIG_SECURITY
623 /* Stat data, not accessed from path walking */
626 * Filesystems may only read i_nlink directly. They shall use the
627 * following functions for modification:
629 * (set|clear|inc|drop)_nlink
630 * inode_(inc|dec)_link_count
633 const unsigned int i_nlink;
634 unsigned int __i_nlink;
638 struct timespec64 i_atime;
639 struct timespec64 i_mtime;
640 struct timespec64 i_ctime;
641 spinlock_t i_lock; /* i_blocks, i_bytes, maybe i_size */
642 unsigned short i_bytes;
647 #ifdef __NEED_I_SIZE_ORDERED
648 seqcount_t i_size_seqcount;
652 unsigned long i_state;
653 struct rw_semaphore i_rwsem;
655 unsigned long dirtied_when; /* jiffies of first dirtying */
656 unsigned long dirtied_time_when;
658 struct hlist_node i_hash;
659 struct list_head i_io_list; /* backing dev IO list */
660 #ifdef CONFIG_CGROUP_WRITEBACK
661 struct bdi_writeback *i_wb; /* the associated cgroup wb */
663 /* foreign inode detection, see wbc_detach_inode() */
665 u16 i_wb_frn_avg_time;
666 u16 i_wb_frn_history;
668 struct list_head i_lru; /* inode LRU list */
669 struct list_head i_sb_list;
670 struct list_head i_wb_list; /* backing dev writeback list */
672 struct hlist_head i_dentry;
673 struct rcu_head i_rcu;
675 atomic64_t i_version;
677 atomic_t i_dio_count;
678 atomic_t i_writecount;
680 atomic_t i_readcount; /* struct files open RO */
682 const struct file_operations *i_fop; /* former ->i_op->default_file_ops */
683 struct file_lock_context *i_flctx;
684 struct address_space i_data;
685 struct list_head i_devices;
687 struct pipe_inode_info *i_pipe;
688 struct block_device *i_bdev;
696 #ifdef CONFIG_FSNOTIFY
697 __u32 i_fsnotify_mask; /* all events this inode cares about */
698 struct fsnotify_mark_connector __rcu *i_fsnotify_marks;
701 #if IS_ENABLED(CONFIG_FS_ENCRYPTION)
702 struct fscrypt_info *i_crypt_info;
705 void *i_private; /* fs or device private pointer */
706 } __randomize_layout;
708 static inline unsigned int i_blocksize(const struct inode *node)
710 return (1 << node->i_blkbits);
713 static inline int inode_unhashed(struct inode *inode)
715 return hlist_unhashed(&inode->i_hash);
719 * __mark_inode_dirty expects inodes to be hashed. Since we don't
720 * want special inodes in the fileset inode space, we make them
721 * appear hashed, but do not put on any lists. hlist_del()
722 * will work fine and require no locking.
724 static inline void inode_fake_hash(struct inode *inode)
726 hlist_add_fake(&inode->i_hash);
730 * inode->i_mutex nesting subclasses for the lock validator:
732 * 0: the object of the current VFS operation
736 * 4: second non-directory
737 * 5: second parent (when locking independent directories in rename)
739 * I_MUTEX_NONDIR2 is for certain operations (such as rename) which lock two
740 * non-directories at once.
742 * The locking order between these classes is
743 * parent[2] -> child -> grandchild -> normal -> xattr -> second non-directory
745 enum inode_i_mutex_lock_class
755 static inline void inode_lock(struct inode *inode)
757 down_write(&inode->i_rwsem);
760 static inline void inode_unlock(struct inode *inode)
762 up_write(&inode->i_rwsem);
765 static inline void inode_lock_shared(struct inode *inode)
767 down_read(&inode->i_rwsem);
770 static inline void inode_unlock_shared(struct inode *inode)
772 up_read(&inode->i_rwsem);
775 static inline int inode_trylock(struct inode *inode)
777 return down_write_trylock(&inode->i_rwsem);
780 static inline int inode_trylock_shared(struct inode *inode)
782 return down_read_trylock(&inode->i_rwsem);
785 static inline int inode_is_locked(struct inode *inode)
787 return rwsem_is_locked(&inode->i_rwsem);
790 static inline void inode_lock_nested(struct inode *inode, unsigned subclass)
792 down_write_nested(&inode->i_rwsem, subclass);
795 static inline void inode_lock_shared_nested(struct inode *inode, unsigned subclass)
797 down_read_nested(&inode->i_rwsem, subclass);
800 void lock_two_nondirectories(struct inode *, struct inode*);
801 void unlock_two_nondirectories(struct inode *, struct inode*);
804 * NOTE: in a 32bit arch with a preemptable kernel and
805 * an UP compile the i_size_read/write must be atomic
806 * with respect to the local cpu (unlike with preempt disabled),
807 * but they don't need to be atomic with respect to other cpus like in
808 * true SMP (so they need either to either locally disable irq around
809 * the read or for example on x86 they can be still implemented as a
810 * cmpxchg8b without the need of the lock prefix). For SMP compiles
811 * and 64bit archs it makes no difference if preempt is enabled or not.
813 static inline loff_t i_size_read(const struct inode *inode)
815 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
820 seq = read_seqcount_begin(&inode->i_size_seqcount);
821 i_size = inode->i_size;
822 } while (read_seqcount_retry(&inode->i_size_seqcount, seq));
824 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
828 i_size = inode->i_size;
832 return inode->i_size;
837 * NOTE: unlike i_size_read(), i_size_write() does need locking around it
838 * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount
839 * can be lost, resulting in subsequent i_size_read() calls spinning forever.
841 static inline void i_size_write(struct inode *inode, loff_t i_size)
843 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
845 write_seqcount_begin(&inode->i_size_seqcount);
846 inode->i_size = i_size;
847 write_seqcount_end(&inode->i_size_seqcount);
849 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
851 inode->i_size = i_size;
854 inode->i_size = i_size;
858 static inline unsigned iminor(const struct inode *inode)
860 return MINOR(inode->i_rdev);
863 static inline unsigned imajor(const struct inode *inode)
865 return MAJOR(inode->i_rdev);
868 extern struct block_device *I_BDEV(struct inode *inode);
871 rwlock_t lock; /* protects pid, uid, euid fields */
872 struct pid *pid; /* pid or -pgrp where SIGIO should be sent */
873 enum pid_type pid_type; /* Kind of process group SIGIO should be sent to */
874 kuid_t uid, euid; /* uid/euid of process setting the owner */
875 int signum; /* posix.1b rt signal to be delivered on IO */
879 * Track a single file's readahead state
881 struct file_ra_state {
882 pgoff_t start; /* where readahead started */
883 unsigned int size; /* # of readahead pages */
884 unsigned int async_size; /* do asynchronous readahead when
885 there are only # of pages ahead */
887 unsigned int ra_pages; /* Maximum readahead window */
888 unsigned int mmap_miss; /* Cache miss stat for mmap accesses */
889 loff_t prev_pos; /* Cache last read() position */
893 * Check if @index falls in the readahead windows.
895 static inline int ra_has_index(struct file_ra_state *ra, pgoff_t index)
897 return (index >= ra->start &&
898 index < ra->start + ra->size);
903 struct llist_node fu_llist;
904 struct rcu_head fu_rcuhead;
907 struct inode *f_inode; /* cached value */
908 const struct file_operations *f_op;
911 * Protects f_ep_links, f_flags.
912 * Must not be taken from IRQ context.
915 enum rw_hint f_write_hint;
916 atomic_long_t f_count;
917 unsigned int f_flags;
919 struct mutex f_pos_lock;
921 struct fown_struct f_owner;
922 const struct cred *f_cred;
923 struct file_ra_state f_ra;
926 #ifdef CONFIG_SECURITY
929 /* needed for tty driver, and maybe others */
933 /* Used by fs/eventpoll.c to link all the hooks to this file */
934 struct list_head f_ep_links;
935 struct list_head f_tfile_llink;
936 #endif /* #ifdef CONFIG_EPOLL */
937 struct address_space *f_mapping;
940 __attribute__((aligned(4))); /* lest something weird decides that 2 is OK */
945 /* file identifier */
946 unsigned char f_handle[0];
949 static inline struct file *get_file(struct file *f)
951 atomic_long_inc(&f->f_count);
954 #define get_file_rcu(x) atomic_long_inc_not_zero(&(x)->f_count)
955 #define fput_atomic(x) atomic_long_add_unless(&(x)->f_count, -1, 1)
956 #define file_count(x) atomic_long_read(&(x)->f_count)
958 #define MAX_NON_LFS ((1UL<<31) - 1)
960 /* Page cache limit. The filesystems should put that into their s_maxbytes
961 limits, otherwise bad things can happen in VM. */
962 #if BITS_PER_LONG==32
963 #define MAX_LFS_FILESIZE ((loff_t)ULONG_MAX << PAGE_SHIFT)
964 #elif BITS_PER_LONG==64
965 #define MAX_LFS_FILESIZE ((loff_t)LLONG_MAX)
970 #define FL_DELEG 4 /* NFSv4 delegation */
971 #define FL_ACCESS 8 /* not trying to lock, just looking */
972 #define FL_EXISTS 16 /* when unlocking, test for existence */
973 #define FL_LEASE 32 /* lease held on this file */
974 #define FL_CLOSE 64 /* unlock on close */
975 #define FL_SLEEP 128 /* A blocking lock */
976 #define FL_DOWNGRADE_PENDING 256 /* Lease is being downgraded */
977 #define FL_UNLOCK_PENDING 512 /* Lease is being broken */
978 #define FL_OFDLCK 1024 /* lock is "owned" by struct file */
979 #define FL_LAYOUT 2048 /* outstanding pNFS layout */
981 #define FL_CLOSE_POSIX (FL_POSIX | FL_CLOSE)
984 * Special return value from posix_lock_file() and vfs_lock_file() for
985 * asynchronous locking.
987 #define FILE_LOCK_DEFERRED 1
989 /* legacy typedef, should eventually be removed */
990 typedef void *fl_owner_t;
994 struct file_lock_operations {
995 void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
996 void (*fl_release_private)(struct file_lock *);
999 struct lock_manager_operations {
1000 int (*lm_compare_owner)(struct file_lock *, struct file_lock *);
1001 unsigned long (*lm_owner_key)(struct file_lock *);
1002 fl_owner_t (*lm_get_owner)(fl_owner_t);
1003 void (*lm_put_owner)(fl_owner_t);
1004 void (*lm_notify)(struct file_lock *); /* unblock callback */
1005 int (*lm_grant)(struct file_lock *, int);
1006 bool (*lm_break)(struct file_lock *);
1007 int (*lm_change)(struct file_lock *, int, struct list_head *);
1008 void (*lm_setup)(struct file_lock *, void **);
1011 struct lock_manager {
1012 struct list_head list;
1014 * NFSv4 and up also want opens blocked during the grace period;
1021 void locks_start_grace(struct net *, struct lock_manager *);
1022 void locks_end_grace(struct lock_manager *);
1023 bool locks_in_grace(struct net *);
1024 bool opens_in_grace(struct net *);
1026 /* that will die - we need it for nfs_lock_info */
1027 #include <linux/nfs_fs_i.h>
1030 * struct file_lock represents a generic "file lock". It's used to represent
1031 * POSIX byte range locks, BSD (flock) locks, and leases. It's important to
1032 * note that the same struct is used to represent both a request for a lock and
1033 * the lock itself, but the same object is never used for both.
1035 * FIXME: should we create a separate "struct lock_request" to help distinguish
1038 * The varous i_flctx lists are ordered by:
1041 * 2) lock range start
1044 * Obviously, the last two criteria only matter for POSIX locks.
1047 struct file_lock *fl_blocker; /* The lock, that is blocking us */
1048 struct list_head fl_list; /* link into file_lock_context */
1049 struct hlist_node fl_link; /* node in global lists */
1050 struct list_head fl_blocked_requests; /* list of requests with
1051 * ->fl_blocker pointing here
1053 struct list_head fl_blocked_member; /* node in
1054 * ->fl_blocker->fl_blocked_requests
1056 fl_owner_t fl_owner;
1057 unsigned int fl_flags;
1058 unsigned char fl_type;
1059 unsigned int fl_pid;
1060 int fl_link_cpu; /* what cpu's list is this on? */
1061 wait_queue_head_t fl_wait;
1062 struct file *fl_file;
1066 struct fasync_struct * fl_fasync; /* for lease break notifications */
1067 /* for lease breaks: */
1068 unsigned long fl_break_time;
1069 unsigned long fl_downgrade_time;
1071 const struct file_lock_operations *fl_ops; /* Callbacks for filesystems */
1072 const struct lock_manager_operations *fl_lmops; /* Callbacks for lockmanagers */
1074 struct nfs_lock_info nfs_fl;
1075 struct nfs4_lock_info nfs4_fl;
1077 struct list_head link; /* link in AFS vnode's pending_locks list */
1078 int state; /* state of grant or error if -ve */
1081 } __randomize_layout;
1083 struct file_lock_context {
1084 spinlock_t flc_lock;
1085 struct list_head flc_flock;
1086 struct list_head flc_posix;
1087 struct list_head flc_lease;
1090 /* The following constant reflects the upper bound of the file/locking space */
1092 #define INT_LIMIT(x) (~((x)1 << (sizeof(x)*8 - 1)))
1093 #define OFFSET_MAX INT_LIMIT(loff_t)
1094 #define OFFT_OFFSET_MAX INT_LIMIT(off_t)
1097 extern void send_sigio(struct fown_struct *fown, int fd, int band);
1099 #define locks_inode(f) file_inode(f)
1101 #ifdef CONFIG_FILE_LOCKING
1102 extern int fcntl_getlk(struct file *, unsigned int, struct flock *);
1103 extern int fcntl_setlk(unsigned int, struct file *, unsigned int,
1106 #if BITS_PER_LONG == 32
1107 extern int fcntl_getlk64(struct file *, unsigned int, struct flock64 *);
1108 extern int fcntl_setlk64(unsigned int, struct file *, unsigned int,
1112 extern int fcntl_setlease(unsigned int fd, struct file *filp, long arg);
1113 extern int fcntl_getlease(struct file *filp);
1116 void locks_free_lock_context(struct inode *inode);
1117 void locks_free_lock(struct file_lock *fl);
1118 extern void locks_init_lock(struct file_lock *);
1119 extern struct file_lock * locks_alloc_lock(void);
1120 extern void locks_copy_lock(struct file_lock *, struct file_lock *);
1121 extern void locks_copy_conflock(struct file_lock *, struct file_lock *);
1122 extern void locks_remove_posix(struct file *, fl_owner_t);
1123 extern void locks_remove_file(struct file *);
1124 extern void locks_release_private(struct file_lock *);
1125 extern void posix_test_lock(struct file *, struct file_lock *);
1126 extern int posix_lock_file(struct file *, struct file_lock *, struct file_lock *);
1127 extern int locks_delete_block(struct file_lock *);
1128 extern int vfs_test_lock(struct file *, struct file_lock *);
1129 extern int vfs_lock_file(struct file *, unsigned int, struct file_lock *, struct file_lock *);
1130 extern int vfs_cancel_lock(struct file *filp, struct file_lock *fl);
1131 extern int locks_lock_inode_wait(struct inode *inode, struct file_lock *fl);
1132 extern int __break_lease(struct inode *inode, unsigned int flags, unsigned int type);
1133 extern void lease_get_mtime(struct inode *, struct timespec64 *time);
1134 extern int generic_setlease(struct file *, long, struct file_lock **, void **priv);
1135 extern int vfs_setlease(struct file *, long, struct file_lock **, void **);
1136 extern int lease_modify(struct file_lock *, int, struct list_head *);
1137 struct files_struct;
1138 extern void show_fd_locks(struct seq_file *f,
1139 struct file *filp, struct files_struct *files);
1140 #else /* !CONFIG_FILE_LOCKING */
1141 static inline int fcntl_getlk(struct file *file, unsigned int cmd,
1142 struct flock __user *user)
1147 static inline int fcntl_setlk(unsigned int fd, struct file *file,
1148 unsigned int cmd, struct flock __user *user)
1153 #if BITS_PER_LONG == 32
1154 static inline int fcntl_getlk64(struct file *file, unsigned int cmd,
1155 struct flock64 __user *user)
1160 static inline int fcntl_setlk64(unsigned int fd, struct file *file,
1161 unsigned int cmd, struct flock64 __user *user)
1166 static inline int fcntl_setlease(unsigned int fd, struct file *filp, long arg)
1171 static inline int fcntl_getlease(struct file *filp)
1177 locks_free_lock_context(struct inode *inode)
1181 static inline void locks_init_lock(struct file_lock *fl)
1186 static inline void locks_copy_conflock(struct file_lock *new, struct file_lock *fl)
1191 static inline void locks_copy_lock(struct file_lock *new, struct file_lock *fl)
1196 static inline void locks_remove_posix(struct file *filp, fl_owner_t owner)
1201 static inline void locks_remove_file(struct file *filp)
1206 static inline void posix_test_lock(struct file *filp, struct file_lock *fl)
1211 static inline int posix_lock_file(struct file *filp, struct file_lock *fl,
1212 struct file_lock *conflock)
1217 static inline int locks_delete_block(struct file_lock *waiter)
1222 static inline int vfs_test_lock(struct file *filp, struct file_lock *fl)
1227 static inline int vfs_lock_file(struct file *filp, unsigned int cmd,
1228 struct file_lock *fl, struct file_lock *conf)
1233 static inline int vfs_cancel_lock(struct file *filp, struct file_lock *fl)
1238 static inline int locks_lock_inode_wait(struct inode *inode, struct file_lock *fl)
1243 static inline int __break_lease(struct inode *inode, unsigned int mode, unsigned int type)
1248 static inline void lease_get_mtime(struct inode *inode,
1249 struct timespec64 *time)
1254 static inline int generic_setlease(struct file *filp, long arg,
1255 struct file_lock **flp, void **priv)
1260 static inline int vfs_setlease(struct file *filp, long arg,
1261 struct file_lock **lease, void **priv)
1266 static inline int lease_modify(struct file_lock *fl, int arg,
1267 struct list_head *dispose)
1272 struct files_struct;
1273 static inline void show_fd_locks(struct seq_file *f,
1274 struct file *filp, struct files_struct *files) {}
1275 #endif /* !CONFIG_FILE_LOCKING */
1277 static inline struct inode *file_inode(const struct file *f)
1282 static inline struct dentry *file_dentry(const struct file *file)
1284 return d_real(file->f_path.dentry, file_inode(file));
1287 static inline int locks_lock_file_wait(struct file *filp, struct file_lock *fl)
1289 return locks_lock_inode_wait(locks_inode(filp), fl);
1292 struct fasync_struct {
1296 struct fasync_struct *fa_next; /* singly linked list */
1297 struct file *fa_file;
1298 struct rcu_head fa_rcu;
1301 #define FASYNC_MAGIC 0x4601
1303 /* SMP safe fasync helpers: */
1304 extern int fasync_helper(int, struct file *, int, struct fasync_struct **);
1305 extern struct fasync_struct *fasync_insert_entry(int, struct file *, struct fasync_struct **, struct fasync_struct *);
1306 extern int fasync_remove_entry(struct file *, struct fasync_struct **);
1307 extern struct fasync_struct *fasync_alloc(void);
1308 extern void fasync_free(struct fasync_struct *);
1310 /* can be called from interrupts */
1311 extern void kill_fasync(struct fasync_struct **, int, int);
1313 extern void __f_setown(struct file *filp, struct pid *, enum pid_type, int force);
1314 extern int f_setown(struct file *filp, unsigned long arg, int force);
1315 extern void f_delown(struct file *filp);
1316 extern pid_t f_getown(struct file *filp);
1317 extern int send_sigurg(struct fown_struct *fown);
1320 * sb->s_flags. Note that these mirror the equivalent MS_* flags where
1321 * represented in both.
1323 #define SB_RDONLY 1 /* Mount read-only */
1324 #define SB_NOSUID 2 /* Ignore suid and sgid bits */
1325 #define SB_NODEV 4 /* Disallow access to device special files */
1326 #define SB_NOEXEC 8 /* Disallow program execution */
1327 #define SB_SYNCHRONOUS 16 /* Writes are synced at once */
1328 #define SB_MANDLOCK 64 /* Allow mandatory locks on an FS */
1329 #define SB_DIRSYNC 128 /* Directory modifications are synchronous */
1330 #define SB_NOATIME 1024 /* Do not update access times. */
1331 #define SB_NODIRATIME 2048 /* Do not update directory access times */
1332 #define SB_SILENT 32768
1333 #define SB_POSIXACL (1<<16) /* VFS does not apply the umask */
1334 #define SB_KERNMOUNT (1<<22) /* this is a kern_mount call */
1335 #define SB_I_VERSION (1<<23) /* Update inode I_version field */
1336 #define SB_LAZYTIME (1<<25) /* Update the on-disk [acm]times lazily */
1338 /* These sb flags are internal to the kernel */
1339 #define SB_SUBMOUNT (1<<26)
1340 #define SB_NOSEC (1<<28)
1341 #define SB_BORN (1<<29)
1342 #define SB_ACTIVE (1<<30)
1343 #define SB_NOUSER (1<<31)
1349 #define MNT_FORCE 0x00000001 /* Attempt to forcibily umount */
1350 #define MNT_DETACH 0x00000002 /* Just detach from the tree */
1351 #define MNT_EXPIRE 0x00000004 /* Mark for expiry */
1352 #define UMOUNT_NOFOLLOW 0x00000008 /* Don't follow symlink on umount */
1353 #define UMOUNT_UNUSED 0x80000000 /* Flag guaranteed to be unused */
1356 #define SB_I_CGROUPWB 0x00000001 /* cgroup-aware writeback enabled */
1357 #define SB_I_NOEXEC 0x00000002 /* Ignore executables on this fs */
1358 #define SB_I_NODEV 0x00000004 /* Ignore devices on this fs */
1359 #define SB_I_MULTIROOT 0x00000008 /* Multiple roots to the dentry tree */
1361 /* sb->s_iflags to limit user namespace mounts */
1362 #define SB_I_USERNS_VISIBLE 0x00000010 /* fstype already mounted */
1363 #define SB_I_IMA_UNVERIFIABLE_SIGNATURE 0x00000020
1364 #define SB_I_UNTRUSTED_MOUNTER 0x00000040
1366 /* Possible states of 'frozen' field */
1368 SB_UNFROZEN = 0, /* FS is unfrozen */
1369 SB_FREEZE_WRITE = 1, /* Writes, dir ops, ioctls frozen */
1370 SB_FREEZE_PAGEFAULT = 2, /* Page faults stopped as well */
1371 SB_FREEZE_FS = 3, /* For internal FS use (e.g. to stop
1372 * internal threads if needed) */
1373 SB_FREEZE_COMPLETE = 4, /* ->freeze_fs finished successfully */
1376 #define SB_FREEZE_LEVELS (SB_FREEZE_COMPLETE - 1)
1379 int frozen; /* Is sb frozen? */
1380 wait_queue_head_t wait_unfrozen; /* for get_super_thawed() */
1381 struct percpu_rw_semaphore rw_sem[SB_FREEZE_LEVELS];
1384 struct super_block {
1385 struct list_head s_list; /* Keep this first */
1386 dev_t s_dev; /* search index; _not_ kdev_t */
1387 unsigned char s_blocksize_bits;
1388 unsigned long s_blocksize;
1389 loff_t s_maxbytes; /* Max file size */
1390 struct file_system_type *s_type;
1391 const struct super_operations *s_op;
1392 const struct dquot_operations *dq_op;
1393 const struct quotactl_ops *s_qcop;
1394 const struct export_operations *s_export_op;
1395 unsigned long s_flags;
1396 unsigned long s_iflags; /* internal SB_I_* flags */
1397 unsigned long s_magic;
1398 struct dentry *s_root;
1399 struct rw_semaphore s_umount;
1402 #ifdef CONFIG_SECURITY
1405 const struct xattr_handler **s_xattr;
1406 #if IS_ENABLED(CONFIG_FS_ENCRYPTION)
1407 const struct fscrypt_operations *s_cop;
1409 struct hlist_bl_head s_roots; /* alternate root dentries for NFS */
1410 struct list_head s_mounts; /* list of mounts; _not_ for fs use */
1411 struct block_device *s_bdev;
1412 struct backing_dev_info *s_bdi;
1413 struct mtd_info *s_mtd;
1414 struct hlist_node s_instances;
1415 unsigned int s_quota_types; /* Bitmask of supported quota types */
1416 struct quota_info s_dquot; /* Diskquota specific options */
1418 struct sb_writers s_writers;
1421 * Keep s_fs_info, s_time_gran, s_fsnotify_mask, and
1422 * s_fsnotify_marks together for cache efficiency. They are frequently
1423 * accessed and rarely modified.
1425 void *s_fs_info; /* Filesystem private info */
1427 /* Granularity of c/m/atime in ns (cannot be worse than a second) */
1429 #ifdef CONFIG_FSNOTIFY
1430 __u32 s_fsnotify_mask;
1431 struct fsnotify_mark_connector __rcu *s_fsnotify_marks;
1434 char s_id[32]; /* Informational name */
1435 uuid_t s_uuid; /* UUID */
1437 unsigned int s_max_links;
1441 * The next field is for VFS *only*. No filesystems have any business
1442 * even looking at it. You had been warned.
1444 struct mutex s_vfs_rename_mutex; /* Kludge */
1447 * Filesystem subtype. If non-empty the filesystem type field
1448 * in /proc/mounts will be "type.subtype"
1452 const struct dentry_operations *s_d_op; /* default d_op for dentries */
1455 * Saved pool identifier for cleancache (-1 means none)
1457 int cleancache_poolid;
1459 struct shrinker s_shrink; /* per-sb shrinker handle */
1461 /* Number of inodes with nlink == 0 but still referenced */
1462 atomic_long_t s_remove_count;
1464 /* Pending fsnotify inode refs */
1465 atomic_long_t s_fsnotify_inode_refs;
1467 /* Being remounted read-only */
1468 int s_readonly_remount;
1470 /* AIO completions deferred from interrupt context */
1471 struct workqueue_struct *s_dio_done_wq;
1472 struct hlist_head s_pins;
1475 * Owning user namespace and default context in which to
1476 * interpret filesystem uids, gids, quotas, device nodes,
1477 * xattrs and security labels.
1479 struct user_namespace *s_user_ns;
1482 * The list_lru structure is essentially just a pointer to a table
1483 * of per-node lru lists, each of which has its own spinlock.
1484 * There is no need to put them into separate cachelines.
1486 struct list_lru s_dentry_lru;
1487 struct list_lru s_inode_lru;
1488 struct rcu_head rcu;
1489 struct work_struct destroy_work;
1491 struct mutex s_sync_lock; /* sync serialisation lock */
1494 * Indicates how deep in a filesystem stack this SB is
1498 /* s_inode_list_lock protects s_inodes */
1499 spinlock_t s_inode_list_lock ____cacheline_aligned_in_smp;
1500 struct list_head s_inodes; /* all inodes */
1502 spinlock_t s_inode_wblist_lock;
1503 struct list_head s_inodes_wb; /* writeback inodes */
1504 } __randomize_layout;
1506 /* Helper functions so that in most cases filesystems will
1507 * not need to deal directly with kuid_t and kgid_t and can
1508 * instead deal with the raw numeric values that are stored
1509 * in the filesystem.
1511 static inline uid_t i_uid_read(const struct inode *inode)
1513 return from_kuid(inode->i_sb->s_user_ns, inode->i_uid);
1516 static inline gid_t i_gid_read(const struct inode *inode)
1518 return from_kgid(inode->i_sb->s_user_ns, inode->i_gid);
1521 static inline void i_uid_write(struct inode *inode, uid_t uid)
1523 inode->i_uid = make_kuid(inode->i_sb->s_user_ns, uid);
1526 static inline void i_gid_write(struct inode *inode, gid_t gid)
1528 inode->i_gid = make_kgid(inode->i_sb->s_user_ns, gid);
1531 extern struct timespec64 timespec64_trunc(struct timespec64 t, unsigned gran);
1532 extern struct timespec64 current_time(struct inode *inode);
1535 * Snapshotting support.
1538 void __sb_end_write(struct super_block *sb, int level);
1539 int __sb_start_write(struct super_block *sb, int level, bool wait);
1541 #define __sb_writers_acquired(sb, lev) \
1542 percpu_rwsem_acquire(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1543 #define __sb_writers_release(sb, lev) \
1544 percpu_rwsem_release(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
1547 * sb_end_write - drop write access to a superblock
1548 * @sb: the super we wrote to
1550 * Decrement number of writers to the filesystem. Wake up possible waiters
1551 * wanting to freeze the filesystem.
1553 static inline void sb_end_write(struct super_block *sb)
1555 __sb_end_write(sb, SB_FREEZE_WRITE);
1559 * sb_end_pagefault - drop write access to a superblock from a page fault
1560 * @sb: the super we wrote to
1562 * Decrement number of processes handling write page fault to the filesystem.
1563 * Wake up possible waiters wanting to freeze the filesystem.
1565 static inline void sb_end_pagefault(struct super_block *sb)
1567 __sb_end_write(sb, SB_FREEZE_PAGEFAULT);
1571 * sb_end_intwrite - drop write access to a superblock for internal fs purposes
1572 * @sb: the super we wrote to
1574 * Decrement fs-internal number of writers to the filesystem. Wake up possible
1575 * waiters wanting to freeze the filesystem.
1577 static inline void sb_end_intwrite(struct super_block *sb)
1579 __sb_end_write(sb, SB_FREEZE_FS);
1583 * sb_start_write - get write access to a superblock
1584 * @sb: the super we write to
1586 * When a process wants to write data or metadata to a file system (i.e. dirty
1587 * a page or an inode), it should embed the operation in a sb_start_write() -
1588 * sb_end_write() pair to get exclusion against file system freezing. This
1589 * function increments number of writers preventing freezing. If the file
1590 * system is already frozen, the function waits until the file system is
1593 * Since freeze protection behaves as a lock, users have to preserve
1594 * ordering of freeze protection and other filesystem locks. Generally,
1595 * freeze protection should be the outermost lock. In particular, we have:
1598 * -> i_mutex (write path, truncate, directory ops, ...)
1599 * -> s_umount (freeze_super, thaw_super)
1601 static inline void sb_start_write(struct super_block *sb)
1603 __sb_start_write(sb, SB_FREEZE_WRITE, true);
1606 static inline int sb_start_write_trylock(struct super_block *sb)
1608 return __sb_start_write(sb, SB_FREEZE_WRITE, false);
1612 * sb_start_pagefault - get write access to a superblock from a page fault
1613 * @sb: the super we write to
1615 * When a process starts handling write page fault, it should embed the
1616 * operation into sb_start_pagefault() - sb_end_pagefault() pair to get
1617 * exclusion against file system freezing. This is needed since the page fault
1618 * is going to dirty a page. This function increments number of running page
1619 * faults preventing freezing. If the file system is already frozen, the
1620 * function waits until the file system is thawed.
1622 * Since page fault freeze protection behaves as a lock, users have to preserve
1623 * ordering of freeze protection and other filesystem locks. It is advised to
1624 * put sb_start_pagefault() close to mmap_sem in lock ordering. Page fault
1625 * handling code implies lock dependency:
1628 * -> sb_start_pagefault
1630 static inline void sb_start_pagefault(struct super_block *sb)
1632 __sb_start_write(sb, SB_FREEZE_PAGEFAULT, true);
1636 * sb_start_intwrite - get write access to a superblock for internal fs purposes
1637 * @sb: the super we write to
1639 * This is the third level of protection against filesystem freezing. It is
1640 * free for use by a filesystem. The only requirement is that it must rank
1641 * below sb_start_pagefault.
1643 * For example filesystem can call sb_start_intwrite() when starting a
1644 * transaction which somewhat eases handling of freezing for internal sources
1645 * of filesystem changes (internal fs threads, discarding preallocation on file
1648 static inline void sb_start_intwrite(struct super_block *sb)
1650 __sb_start_write(sb, SB_FREEZE_FS, true);
1653 static inline int sb_start_intwrite_trylock(struct super_block *sb)
1655 return __sb_start_write(sb, SB_FREEZE_FS, false);
1659 extern bool inode_owner_or_capable(const struct inode *inode);
1662 * VFS helper functions..
1664 extern int vfs_create(struct inode *, struct dentry *, umode_t, bool);
1665 extern int vfs_mkdir(struct inode *, struct dentry *, umode_t);
1666 extern int vfs_mknod(struct inode *, struct dentry *, umode_t, dev_t);
1667 extern int vfs_symlink(struct inode *, struct dentry *, const char *);
1668 extern int vfs_link(struct dentry *, struct inode *, struct dentry *, struct inode **);
1669 extern int vfs_rmdir(struct inode *, struct dentry *);
1670 extern int vfs_unlink(struct inode *, struct dentry *, struct inode **);
1671 extern int vfs_rename(struct inode *, struct dentry *, struct inode *, struct dentry *, struct inode **, unsigned int);
1672 extern int vfs_whiteout(struct inode *, struct dentry *);
1674 extern struct dentry *vfs_tmpfile(struct dentry *dentry, umode_t mode,
1677 int vfs_mkobj(struct dentry *, umode_t,
1678 int (*f)(struct dentry *, umode_t, void *),
1681 extern long vfs_ioctl(struct file *file, unsigned int cmd, unsigned long arg);
1684 * VFS file helper functions.
1686 extern void inode_init_owner(struct inode *inode, const struct inode *dir,
1688 extern bool may_open_dev(const struct path *path);
1690 * VFS FS_IOC_FIEMAP helper definitions.
1692 struct fiemap_extent_info {
1693 unsigned int fi_flags; /* Flags as passed from user */
1694 unsigned int fi_extents_mapped; /* Number of mapped extents */
1695 unsigned int fi_extents_max; /* Size of fiemap_extent array */
1696 struct fiemap_extent __user *fi_extents_start; /* Start of
1697 fiemap_extent array */
1699 int fiemap_fill_next_extent(struct fiemap_extent_info *info, u64 logical,
1700 u64 phys, u64 len, u32 flags);
1701 int fiemap_check_flags(struct fiemap_extent_info *fieinfo, u32 fs_flags);
1706 * NOTE! These match bits 12..15 of stat.st_mode
1707 * (ie "(i_mode >> 12) & 15").
1709 #define DT_UNKNOWN 0
1720 * This is the "filldir" function type, used by readdir() to let
1721 * the kernel specify what kind of dirent layout it wants to have.
1722 * This allows the kernel to read directories into kernel space or
1723 * to have different dirent layouts depending on the binary type.
1726 typedef int (*filldir_t)(struct dir_context *, const char *, int, loff_t, u64,
1729 struct dir_context {
1734 struct block_device_operations;
1736 /* These macros are for out of kernel modules to test that
1737 * the kernel supports the unlocked_ioctl and compat_ioctl
1738 * fields in struct file_operations. */
1739 #define HAVE_COMPAT_IOCTL 1
1740 #define HAVE_UNLOCKED_IOCTL 1
1743 * These flags let !MMU mmap() govern direct device mapping vs immediate
1744 * copying more easily for MAP_PRIVATE, especially for ROM filesystems.
1746 * NOMMU_MAP_COPY: Copy can be mapped (MAP_PRIVATE)
1747 * NOMMU_MAP_DIRECT: Can be mapped directly (MAP_SHARED)
1748 * NOMMU_MAP_READ: Can be mapped for reading
1749 * NOMMU_MAP_WRITE: Can be mapped for writing
1750 * NOMMU_MAP_EXEC: Can be mapped for execution
1752 #define NOMMU_MAP_COPY 0x00000001
1753 #define NOMMU_MAP_DIRECT 0x00000008
1754 #define NOMMU_MAP_READ VM_MAYREAD
1755 #define NOMMU_MAP_WRITE VM_MAYWRITE
1756 #define NOMMU_MAP_EXEC VM_MAYEXEC
1758 #define NOMMU_VMFLAGS \
1759 (NOMMU_MAP_READ | NOMMU_MAP_WRITE | NOMMU_MAP_EXEC)
1762 * These flags control the behavior of the remap_file_range function pointer.
1763 * If it is called with len == 0 that means "remap to end of source file".
1764 * See Documentation/filesystems/vfs.txt for more details about this call.
1766 * REMAP_FILE_DEDUP: only remap if contents identical (i.e. deduplicate)
1767 * REMAP_FILE_CAN_SHORTEN: caller can handle a shortened request
1769 #define REMAP_FILE_DEDUP (1 << 0)
1770 #define REMAP_FILE_CAN_SHORTEN (1 << 1)
1773 * These flags signal that the caller is ok with altering various aspects of
1774 * the behavior of the remap operation. The changes must be made by the
1775 * implementation; the vfs remap helper functions can take advantage of them.
1776 * Flags in this category exist to preserve the quirky behavior of the hoisted
1777 * btrfs clone/dedupe ioctls.
1779 #define REMAP_FILE_ADVISORY (REMAP_FILE_CAN_SHORTEN)
1783 struct file_operations {
1784 struct module *owner;
1785 loff_t (*llseek) (struct file *, loff_t, int);
1786 ssize_t (*read) (struct file *, char __user *, size_t, loff_t *);
1787 ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *);
1788 ssize_t (*read_iter) (struct kiocb *, struct iov_iter *);
1789 ssize_t (*write_iter) (struct kiocb *, struct iov_iter *);
1790 int (*iterate) (struct file *, struct dir_context *);
1791 int (*iterate_shared) (struct file *, struct dir_context *);
1792 __poll_t (*poll) (struct file *, struct poll_table_struct *);
1793 long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long);
1794 long (*compat_ioctl) (struct file *, unsigned int, unsigned long);
1795 int (*mmap) (struct file *, struct vm_area_struct *);
1796 unsigned long mmap_supported_flags;
1797 int (*open) (struct inode *, struct file *);
1798 int (*flush) (struct file *, fl_owner_t id);
1799 int (*release) (struct inode *, struct file *);
1800 int (*fsync) (struct file *, loff_t, loff_t, int datasync);
1801 int (*fasync) (int, struct file *, int);
1802 int (*lock) (struct file *, int, struct file_lock *);
1803 ssize_t (*sendpage) (struct file *, struct page *, int, size_t, loff_t *, int);
1804 unsigned long (*get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
1805 int (*check_flags)(int);
1806 int (*flock) (struct file *, int, struct file_lock *);
1807 ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, loff_t *, size_t, unsigned int);
1808 ssize_t (*splice_read)(struct file *, loff_t *, struct pipe_inode_info *, size_t, unsigned int);
1809 int (*setlease)(struct file *, long, struct file_lock **, void **);
1810 long (*fallocate)(struct file *file, int mode, loff_t offset,
1812 void (*show_fdinfo)(struct seq_file *m, struct file *f);
1814 unsigned (*mmap_capabilities)(struct file *);
1816 ssize_t (*copy_file_range)(struct file *, loff_t, struct file *,
1817 loff_t, size_t, unsigned int);
1818 loff_t (*remap_file_range)(struct file *file_in, loff_t pos_in,
1819 struct file *file_out, loff_t pos_out,
1820 loff_t len, unsigned int remap_flags);
1821 int (*fadvise)(struct file *, loff_t, loff_t, int);
1822 } __randomize_layout;
1824 struct inode_operations {
1825 struct dentry * (*lookup) (struct inode *,struct dentry *, unsigned int);
1826 const char * (*get_link) (struct dentry *, struct inode *, struct delayed_call *);
1827 int (*permission) (struct inode *, int);
1828 struct posix_acl * (*get_acl)(struct inode *, int);
1830 int (*readlink) (struct dentry *, char __user *,int);
1832 int (*create) (struct inode *,struct dentry *, umode_t, bool);
1833 int (*link) (struct dentry *,struct inode *,struct dentry *);
1834 int (*unlink) (struct inode *,struct dentry *);
1835 int (*symlink) (struct inode *,struct dentry *,const char *);
1836 int (*mkdir) (struct inode *,struct dentry *,umode_t);
1837 int (*rmdir) (struct inode *,struct dentry *);
1838 int (*mknod) (struct inode *,struct dentry *,umode_t,dev_t);
1839 int (*rename) (struct inode *, struct dentry *,
1840 struct inode *, struct dentry *, unsigned int);
1841 int (*setattr) (struct dentry *, struct iattr *);
1842 int (*getattr) (const struct path *, struct kstat *, u32, unsigned int);
1843 ssize_t (*listxattr) (struct dentry *, char *, size_t);
1844 int (*fiemap)(struct inode *, struct fiemap_extent_info *, u64 start,
1846 int (*update_time)(struct inode *, struct timespec64 *, int);
1847 int (*atomic_open)(struct inode *, struct dentry *,
1848 struct file *, unsigned open_flag,
1849 umode_t create_mode);
1850 int (*tmpfile) (struct inode *, struct dentry *, umode_t);
1851 int (*set_acl)(struct inode *, struct posix_acl *, int);
1852 } ____cacheline_aligned;
1854 static inline ssize_t call_read_iter(struct file *file, struct kiocb *kio,
1855 struct iov_iter *iter)
1857 return file->f_op->read_iter(kio, iter);
1860 static inline ssize_t call_write_iter(struct file *file, struct kiocb *kio,
1861 struct iov_iter *iter)
1863 return file->f_op->write_iter(kio, iter);
1866 static inline int call_mmap(struct file *file, struct vm_area_struct *vma)
1868 return file->f_op->mmap(file, vma);
1871 ssize_t rw_copy_check_uvector(int type, const struct iovec __user * uvector,
1872 unsigned long nr_segs, unsigned long fast_segs,
1873 struct iovec *fast_pointer,
1874 struct iovec **ret_pointer);
1876 extern ssize_t __vfs_read(struct file *, char __user *, size_t, loff_t *);
1877 extern ssize_t vfs_read(struct file *, char __user *, size_t, loff_t *);
1878 extern ssize_t vfs_write(struct file *, const char __user *, size_t, loff_t *);
1879 extern ssize_t vfs_readv(struct file *, const struct iovec __user *,
1880 unsigned long, loff_t *, rwf_t);
1881 extern ssize_t vfs_copy_file_range(struct file *, loff_t , struct file *,
1882 loff_t, size_t, unsigned int);
1883 extern int generic_remap_file_range_prep(struct file *file_in, loff_t pos_in,
1884 struct file *file_out, loff_t pos_out,
1886 unsigned int remap_flags);
1887 extern loff_t do_clone_file_range(struct file *file_in, loff_t pos_in,
1888 struct file *file_out, loff_t pos_out,
1889 loff_t len, unsigned int remap_flags);
1890 extern loff_t vfs_clone_file_range(struct file *file_in, loff_t pos_in,
1891 struct file *file_out, loff_t pos_out,
1892 loff_t len, unsigned int remap_flags);
1893 extern int vfs_dedupe_file_range(struct file *file,
1894 struct file_dedupe_range *same);
1895 extern loff_t vfs_dedupe_file_range_one(struct file *src_file, loff_t src_pos,
1896 struct file *dst_file, loff_t dst_pos,
1897 loff_t len, unsigned int remap_flags);
1900 struct super_operations {
1901 struct inode *(*alloc_inode)(struct super_block *sb);
1902 void (*destroy_inode)(struct inode *);
1904 void (*dirty_inode) (struct inode *, int flags);
1905 int (*write_inode) (struct inode *, struct writeback_control *wbc);
1906 int (*drop_inode) (struct inode *);
1907 void (*evict_inode) (struct inode *);
1908 void (*put_super) (struct super_block *);
1909 int (*sync_fs)(struct super_block *sb, int wait);
1910 int (*freeze_super) (struct super_block *);
1911 int (*freeze_fs) (struct super_block *);
1912 int (*thaw_super) (struct super_block *);
1913 int (*unfreeze_fs) (struct super_block *);
1914 int (*statfs) (struct dentry *, struct kstatfs *);
1915 int (*remount_fs) (struct super_block *, int *, char *);
1916 void (*umount_begin) (struct super_block *);
1918 int (*show_options)(struct seq_file *, struct dentry *);
1919 int (*show_devname)(struct seq_file *, struct dentry *);
1920 int (*show_path)(struct seq_file *, struct dentry *);
1921 int (*show_stats)(struct seq_file *, struct dentry *);
1923 ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t);
1924 ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t);
1925 struct dquot **(*get_dquots)(struct inode *);
1927 int (*bdev_try_to_free_page)(struct super_block*, struct page*, gfp_t);
1928 long (*nr_cached_objects)(struct super_block *,
1929 struct shrink_control *);
1930 long (*free_cached_objects)(struct super_block *,
1931 struct shrink_control *);
1935 * Inode flags - they have no relation to superblock flags now
1937 #define S_SYNC 1 /* Writes are synced at once */
1938 #define S_NOATIME 2 /* Do not update access times */
1939 #define S_APPEND 4 /* Append-only file */
1940 #define S_IMMUTABLE 8 /* Immutable file */
1941 #define S_DEAD 16 /* removed, but still open directory */
1942 #define S_NOQUOTA 32 /* Inode is not counted to quota */
1943 #define S_DIRSYNC 64 /* Directory modifications are synchronous */
1944 #define S_NOCMTIME 128 /* Do not update file c/mtime */
1945 #define S_SWAPFILE 256 /* Do not truncate: swapon got its bmaps */
1946 #define S_PRIVATE 512 /* Inode is fs-internal */
1947 #define S_IMA 1024 /* Inode has an associated IMA struct */
1948 #define S_AUTOMOUNT 2048 /* Automount/referral quasi-directory */
1949 #define S_NOSEC 4096 /* no suid or xattr security attributes */
1950 #ifdef CONFIG_FS_DAX
1951 #define S_DAX 8192 /* Direct Access, avoiding the page cache */
1953 #define S_DAX 0 /* Make all the DAX code disappear */
1955 #define S_ENCRYPTED 16384 /* Encrypted file (using fs/crypto/) */
1958 * Note that nosuid etc flags are inode-specific: setting some file-system
1959 * flags just means all the inodes inherit those flags by default. It might be
1960 * possible to override it selectively if you really wanted to with some
1961 * ioctl() that is not currently implemented.
1963 * Exception: SB_RDONLY is always applied to the entire file system.
1965 * Unfortunately, it is possible to change a filesystems flags with it mounted
1966 * with files in use. This means that all of the inodes will not have their
1967 * i_flags updated. Hence, i_flags no longer inherit the superblock mount
1968 * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org
1970 #define __IS_FLG(inode, flg) ((inode)->i_sb->s_flags & (flg))
1972 static inline bool sb_rdonly(const struct super_block *sb) { return sb->s_flags & SB_RDONLY; }
1973 #define IS_RDONLY(inode) sb_rdonly((inode)->i_sb)
1974 #define IS_SYNC(inode) (__IS_FLG(inode, SB_SYNCHRONOUS) || \
1975 ((inode)->i_flags & S_SYNC))
1976 #define IS_DIRSYNC(inode) (__IS_FLG(inode, SB_SYNCHRONOUS|SB_DIRSYNC) || \
1977 ((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
1978 #define IS_MANDLOCK(inode) __IS_FLG(inode, SB_MANDLOCK)
1979 #define IS_NOATIME(inode) __IS_FLG(inode, SB_RDONLY|SB_NOATIME)
1980 #define IS_I_VERSION(inode) __IS_FLG(inode, SB_I_VERSION)
1982 #define IS_NOQUOTA(inode) ((inode)->i_flags & S_NOQUOTA)
1983 #define IS_APPEND(inode) ((inode)->i_flags & S_APPEND)
1984 #define IS_IMMUTABLE(inode) ((inode)->i_flags & S_IMMUTABLE)
1985 #define IS_POSIXACL(inode) __IS_FLG(inode, SB_POSIXACL)
1987 #define IS_DEADDIR(inode) ((inode)->i_flags & S_DEAD)
1988 #define IS_NOCMTIME(inode) ((inode)->i_flags & S_NOCMTIME)
1989 #define IS_SWAPFILE(inode) ((inode)->i_flags & S_SWAPFILE)
1990 #define IS_PRIVATE(inode) ((inode)->i_flags & S_PRIVATE)
1991 #define IS_IMA(inode) ((inode)->i_flags & S_IMA)
1992 #define IS_AUTOMOUNT(inode) ((inode)->i_flags & S_AUTOMOUNT)
1993 #define IS_NOSEC(inode) ((inode)->i_flags & S_NOSEC)
1994 #define IS_DAX(inode) ((inode)->i_flags & S_DAX)
1995 #define IS_ENCRYPTED(inode) ((inode)->i_flags & S_ENCRYPTED)
1997 #define IS_WHITEOUT(inode) (S_ISCHR(inode->i_mode) && \
1998 (inode)->i_rdev == WHITEOUT_DEV)
2000 static inline bool HAS_UNMAPPED_ID(struct inode *inode)
2002 return !uid_valid(inode->i_uid) || !gid_valid(inode->i_gid);
2005 static inline enum rw_hint file_write_hint(struct file *file)
2007 if (file->f_write_hint != WRITE_LIFE_NOT_SET)
2008 return file->f_write_hint;
2010 return file_inode(file)->i_write_hint;
2013 static inline int iocb_flags(struct file *file);
2015 static inline u16 ki_hint_validate(enum rw_hint hint)
2017 typeof(((struct kiocb *)0)->ki_hint) max_hint = -1;
2019 if (hint <= max_hint)
2024 static inline void init_sync_kiocb(struct kiocb *kiocb, struct file *filp)
2026 *kiocb = (struct kiocb) {
2028 .ki_flags = iocb_flags(filp),
2029 .ki_hint = ki_hint_validate(file_write_hint(filp)),
2030 .ki_ioprio = get_current_ioprio(),
2035 * Inode state bits. Protected by inode->i_lock
2037 * Three bits determine the dirty state of the inode, I_DIRTY_SYNC,
2038 * I_DIRTY_DATASYNC and I_DIRTY_PAGES.
2040 * Four bits define the lifetime of an inode. Initially, inodes are I_NEW,
2041 * until that flag is cleared. I_WILL_FREE, I_FREEING and I_CLEAR are set at
2042 * various stages of removing an inode.
2044 * Two bits are used for locking and completion notification, I_NEW and I_SYNC.
2046 * I_DIRTY_SYNC Inode is dirty, but doesn't have to be written on
2047 * fdatasync(). i_atime is the usual cause.
2048 * I_DIRTY_DATASYNC Data-related inode changes pending. We keep track of
2049 * these changes separately from I_DIRTY_SYNC so that we
2050 * don't have to write inode on fdatasync() when only
2051 * mtime has changed in it.
2052 * I_DIRTY_PAGES Inode has dirty pages. Inode itself may be clean.
2053 * I_NEW Serves as both a mutex and completion notification.
2054 * New inodes set I_NEW. If two processes both create
2055 * the same inode, one of them will release its inode and
2056 * wait for I_NEW to be released before returning.
2057 * Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can
2058 * also cause waiting on I_NEW, without I_NEW actually
2059 * being set. find_inode() uses this to prevent returning
2060 * nearly-dead inodes.
2061 * I_WILL_FREE Must be set when calling write_inode_now() if i_count
2062 * is zero. I_FREEING must be set when I_WILL_FREE is
2064 * I_FREEING Set when inode is about to be freed but still has dirty
2065 * pages or buffers attached or the inode itself is still
2067 * I_CLEAR Added by clear_inode(). In this state the inode is
2068 * clean and can be destroyed. Inode keeps I_FREEING.
2070 * Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are
2071 * prohibited for many purposes. iget() must wait for
2072 * the inode to be completely released, then create it
2073 * anew. Other functions will just ignore such inodes,
2074 * if appropriate. I_NEW is used for waiting.
2076 * I_SYNC Writeback of inode is running. The bit is set during
2077 * data writeback, and cleared with a wakeup on the bit
2078 * address once it is done. The bit is also used to pin
2079 * the inode in memory for flusher thread.
2081 * I_REFERENCED Marks the inode as recently references on the LRU list.
2083 * I_DIO_WAKEUP Never set. Only used as a key for wait_on_bit().
2085 * I_WB_SWITCH Cgroup bdi_writeback switching in progress. Used to
2086 * synchronize competing switching instances and to tell
2087 * wb stat updates to grab the i_pages lock. See
2088 * inode_switch_wb_work_fn() for details.
2090 * I_OVL_INUSE Used by overlayfs to get exclusive ownership on upper
2091 * and work dirs among overlayfs mounts.
2093 * I_CREATING New object's inode in the middle of setting up.
2095 * Q: What is the difference between I_WILL_FREE and I_FREEING?
2097 #define I_DIRTY_SYNC (1 << 0)
2098 #define I_DIRTY_DATASYNC (1 << 1)
2099 #define I_DIRTY_PAGES (1 << 2)
2101 #define I_NEW (1 << __I_NEW)
2102 #define I_WILL_FREE (1 << 4)
2103 #define I_FREEING (1 << 5)
2104 #define I_CLEAR (1 << 6)
2106 #define I_SYNC (1 << __I_SYNC)
2107 #define I_REFERENCED (1 << 8)
2108 #define __I_DIO_WAKEUP 9
2109 #define I_DIO_WAKEUP (1 << __I_DIO_WAKEUP)
2110 #define I_LINKABLE (1 << 10)
2111 #define I_DIRTY_TIME (1 << 11)
2112 #define __I_DIRTY_TIME_EXPIRED 12
2113 #define I_DIRTY_TIME_EXPIRED (1 << __I_DIRTY_TIME_EXPIRED)
2114 #define I_WB_SWITCH (1 << 13)
2115 #define I_OVL_INUSE (1 << 14)
2116 #define I_CREATING (1 << 15)
2118 #define I_DIRTY_INODE (I_DIRTY_SYNC | I_DIRTY_DATASYNC)
2119 #define I_DIRTY (I_DIRTY_INODE | I_DIRTY_PAGES)
2120 #define I_DIRTY_ALL (I_DIRTY | I_DIRTY_TIME)
2122 extern void __mark_inode_dirty(struct inode *, int);
2123 static inline void mark_inode_dirty(struct inode *inode)
2125 __mark_inode_dirty(inode, I_DIRTY);
2128 static inline void mark_inode_dirty_sync(struct inode *inode)
2130 __mark_inode_dirty(inode, I_DIRTY_SYNC);
2133 extern void inc_nlink(struct inode *inode);
2134 extern void drop_nlink(struct inode *inode);
2135 extern void clear_nlink(struct inode *inode);
2136 extern void set_nlink(struct inode *inode, unsigned int nlink);
2138 static inline void inode_inc_link_count(struct inode *inode)
2141 mark_inode_dirty(inode);
2144 static inline void inode_dec_link_count(struct inode *inode)
2147 mark_inode_dirty(inode);
2150 enum file_time_flags {
2157 extern bool atime_needs_update(const struct path *, struct inode *);
2158 extern void touch_atime(const struct path *);
2159 static inline void file_accessed(struct file *file)
2161 if (!(file->f_flags & O_NOATIME))
2162 touch_atime(&file->f_path);
2165 int sync_inode(struct inode *inode, struct writeback_control *wbc);
2166 int sync_inode_metadata(struct inode *inode, int wait);
2168 struct file_system_type {
2171 #define FS_REQUIRES_DEV 1
2172 #define FS_BINARY_MOUNTDATA 2
2173 #define FS_HAS_SUBTYPE 4
2174 #define FS_USERNS_MOUNT 8 /* Can be mounted by userns root */
2175 #define FS_RENAME_DOES_D_MOVE 32768 /* FS will handle d_move() during rename() internally. */
2176 struct dentry *(*mount) (struct file_system_type *, int,
2177 const char *, void *);
2178 void (*kill_sb) (struct super_block *);
2179 struct module *owner;
2180 struct file_system_type * next;
2181 struct hlist_head fs_supers;
2183 struct lock_class_key s_lock_key;
2184 struct lock_class_key s_umount_key;
2185 struct lock_class_key s_vfs_rename_key;
2186 struct lock_class_key s_writers_key[SB_FREEZE_LEVELS];
2188 struct lock_class_key i_lock_key;
2189 struct lock_class_key i_mutex_key;
2190 struct lock_class_key i_mutex_dir_key;
2193 #define MODULE_ALIAS_FS(NAME) MODULE_ALIAS("fs-" NAME)
2195 extern struct dentry *mount_ns(struct file_system_type *fs_type,
2196 int flags, void *data, void *ns, struct user_namespace *user_ns,
2197 int (*fill_super)(struct super_block *, void *, int));
2199 extern struct dentry *mount_bdev(struct file_system_type *fs_type,
2200 int flags, const char *dev_name, void *data,
2201 int (*fill_super)(struct super_block *, void *, int));
2203 static inline struct dentry *mount_bdev(struct file_system_type *fs_type,
2204 int flags, const char *dev_name, void *data,
2205 int (*fill_super)(struct super_block *, void *, int))
2207 return ERR_PTR(-ENODEV);
2210 extern struct dentry *mount_single(struct file_system_type *fs_type,
2211 int flags, void *data,
2212 int (*fill_super)(struct super_block *, void *, int));
2213 extern struct dentry *mount_nodev(struct file_system_type *fs_type,
2214 int flags, void *data,
2215 int (*fill_super)(struct super_block *, void *, int));
2216 extern struct dentry *mount_subtree(struct vfsmount *mnt, const char *path);
2217 void generic_shutdown_super(struct super_block *sb);
2219 void kill_block_super(struct super_block *sb);
2221 static inline void kill_block_super(struct super_block *sb)
2226 void kill_anon_super(struct super_block *sb);
2227 void kill_litter_super(struct super_block *sb);
2228 void deactivate_super(struct super_block *sb);
2229 void deactivate_locked_super(struct super_block *sb);
2230 int set_anon_super(struct super_block *s, void *data);
2231 int get_anon_bdev(dev_t *);
2232 void free_anon_bdev(dev_t);
2233 struct super_block *sget_userns(struct file_system_type *type,
2234 int (*test)(struct super_block *,void *),
2235 int (*set)(struct super_block *,void *),
2236 int flags, struct user_namespace *user_ns,
2238 struct super_block *sget(struct file_system_type *type,
2239 int (*test)(struct super_block *,void *),
2240 int (*set)(struct super_block *,void *),
2241 int flags, void *data);
2242 extern struct dentry *mount_pseudo_xattr(struct file_system_type *, char *,
2243 const struct super_operations *ops,
2244 const struct xattr_handler **xattr,
2245 const struct dentry_operations *dops,
2248 static inline struct dentry *
2249 mount_pseudo(struct file_system_type *fs_type, char *name,
2250 const struct super_operations *ops,
2251 const struct dentry_operations *dops, unsigned long magic)
2253 return mount_pseudo_xattr(fs_type, name, ops, NULL, dops, magic);
2256 /* Alas, no aliases. Too much hassle with bringing module.h everywhere */
2257 #define fops_get(fops) \
2258 (((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
2259 #define fops_put(fops) \
2260 do { if (fops) module_put((fops)->owner); } while(0)
2262 * This one is to be used *ONLY* from ->open() instances.
2263 * fops must be non-NULL, pinned down *and* module dependencies
2264 * should be sufficient to pin the caller down as well.
2266 #define replace_fops(f, fops) \
2268 struct file *__file = (f); \
2269 fops_put(__file->f_op); \
2270 BUG_ON(!(__file->f_op = (fops))); \
2273 extern int register_filesystem(struct file_system_type *);
2274 extern int unregister_filesystem(struct file_system_type *);
2275 extern struct vfsmount *kern_mount_data(struct file_system_type *, void *data);
2276 #define kern_mount(type) kern_mount_data(type, NULL)
2277 extern void kern_unmount(struct vfsmount *mnt);
2278 extern int may_umount_tree(struct vfsmount *);
2279 extern int may_umount(struct vfsmount *);
2280 extern long do_mount(const char *, const char __user *,
2281 const char *, unsigned long, void *);
2282 extern struct vfsmount *collect_mounts(const struct path *);
2283 extern void drop_collected_mounts(struct vfsmount *);
2284 extern int iterate_mounts(int (*)(struct vfsmount *, void *), void *,
2286 extern int vfs_statfs(const struct path *, struct kstatfs *);
2287 extern int user_statfs(const char __user *, struct kstatfs *);
2288 extern int fd_statfs(int, struct kstatfs *);
2289 extern int freeze_super(struct super_block *super);
2290 extern int thaw_super(struct super_block *super);
2291 extern bool our_mnt(struct vfsmount *mnt);
2292 extern __printf(2, 3)
2293 int super_setup_bdi_name(struct super_block *sb, char *fmt, ...);
2294 extern int super_setup_bdi(struct super_block *sb);
2296 extern int current_umask(void);
2298 extern void ihold(struct inode * inode);
2299 extern void iput(struct inode *);
2300 extern int generic_update_time(struct inode *, struct timespec64 *, int);
2303 extern struct kobject *fs_kobj;
2305 #define MAX_RW_COUNT (INT_MAX & PAGE_MASK)
2307 #ifdef CONFIG_MANDATORY_FILE_LOCKING
2308 extern int locks_mandatory_locked(struct file *);
2309 extern int locks_mandatory_area(struct inode *, struct file *, loff_t, loff_t, unsigned char);
2312 * Candidates for mandatory locking have the setgid bit set
2313 * but no group execute bit - an otherwise meaningless combination.
2316 static inline int __mandatory_lock(struct inode *ino)
2318 return (ino->i_mode & (S_ISGID | S_IXGRP)) == S_ISGID;
2322 * ... and these candidates should be on SB_MANDLOCK mounted fs,
2323 * otherwise these will be advisory locks
2326 static inline int mandatory_lock(struct inode *ino)
2328 return IS_MANDLOCK(ino) && __mandatory_lock(ino);
2331 static inline int locks_verify_locked(struct file *file)
2333 if (mandatory_lock(locks_inode(file)))
2334 return locks_mandatory_locked(file);
2338 static inline int locks_verify_truncate(struct inode *inode,
2342 if (!inode->i_flctx || !mandatory_lock(inode))
2345 if (size < inode->i_size) {
2346 return locks_mandatory_area(inode, f, size, inode->i_size - 1,
2349 return locks_mandatory_area(inode, f, inode->i_size, size - 1,
2354 #else /* !CONFIG_MANDATORY_FILE_LOCKING */
2356 static inline int locks_mandatory_locked(struct file *file)
2361 static inline int locks_mandatory_area(struct inode *inode, struct file *filp,
2362 loff_t start, loff_t end, unsigned char type)
2367 static inline int __mandatory_lock(struct inode *inode)
2372 static inline int mandatory_lock(struct inode *inode)
2377 static inline int locks_verify_locked(struct file *file)
2382 static inline int locks_verify_truncate(struct inode *inode, struct file *filp,
2388 #endif /* CONFIG_MANDATORY_FILE_LOCKING */
2391 #ifdef CONFIG_FILE_LOCKING
2392 static inline int break_lease(struct inode *inode, unsigned int mode)
2395 * Since this check is lockless, we must ensure that any refcounts
2396 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2397 * could end up racing with tasks trying to set a new lease on this
2401 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2402 return __break_lease(inode, mode, FL_LEASE);
2406 static inline int break_deleg(struct inode *inode, unsigned int mode)
2409 * Since this check is lockless, we must ensure that any refcounts
2410 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2411 * could end up racing with tasks trying to set a new lease on this
2415 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2416 return __break_lease(inode, mode, FL_DELEG);
2420 static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode)
2424 ret = break_deleg(inode, O_WRONLY|O_NONBLOCK);
2425 if (ret == -EWOULDBLOCK && delegated_inode) {
2426 *delegated_inode = inode;
2432 static inline int break_deleg_wait(struct inode **delegated_inode)
2436 ret = break_deleg(*delegated_inode, O_WRONLY);
2437 iput(*delegated_inode);
2438 *delegated_inode = NULL;
2442 static inline int break_layout(struct inode *inode, bool wait)
2445 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2446 return __break_lease(inode,
2447 wait ? O_WRONLY : O_WRONLY | O_NONBLOCK,
2452 #else /* !CONFIG_FILE_LOCKING */
2453 static inline int break_lease(struct inode *inode, unsigned int mode)
2458 static inline int break_deleg(struct inode *inode, unsigned int mode)
2463 static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode)
2468 static inline int break_deleg_wait(struct inode **delegated_inode)
2474 static inline int break_layout(struct inode *inode, bool wait)
2479 #endif /* CONFIG_FILE_LOCKING */
2484 const char *name; /* pointer to actual string */
2485 const __user char *uptr; /* original userland pointer */
2487 struct audit_names *aname;
2491 extern long vfs_truncate(const struct path *, loff_t);
2492 extern int do_truncate(struct dentry *, loff_t start, unsigned int time_attrs,
2494 extern int vfs_fallocate(struct file *file, int mode, loff_t offset,
2496 extern long do_sys_open(int dfd, const char __user *filename, int flags,
2498 extern struct file *file_open_name(struct filename *, int, umode_t);
2499 extern struct file *filp_open(const char *, int, umode_t);
2500 extern struct file *file_open_root(struct dentry *, struct vfsmount *,
2501 const char *, int, umode_t);
2502 extern struct file * dentry_open(const struct path *, int, const struct cred *);
2503 extern struct file * open_with_fake_path(const struct path *, int,
2504 struct inode*, const struct cred *);
2505 static inline struct file *file_clone_open(struct file *file)
2507 return dentry_open(&file->f_path, file->f_flags, file->f_cred);
2509 extern int filp_close(struct file *, fl_owner_t id);
2511 extern struct filename *getname_flags(const char __user *, int, int *);
2512 extern struct filename *getname(const char __user *);
2513 extern struct filename *getname_kernel(const char *);
2514 extern void putname(struct filename *name);
2516 extern int finish_open(struct file *file, struct dentry *dentry,
2517 int (*open)(struct inode *, struct file *));
2518 extern int finish_no_open(struct file *file, struct dentry *dentry);
2522 extern int ioctl_preallocate(struct file *filp, void __user *argp);
2525 extern void __init vfs_caches_init_early(void);
2526 extern void __init vfs_caches_init(void);
2528 extern struct kmem_cache *names_cachep;
2530 #define __getname() kmem_cache_alloc(names_cachep, GFP_KERNEL)
2531 #define __putname(name) kmem_cache_free(names_cachep, (void *)(name))
2534 extern int register_blkdev(unsigned int, const char *);
2535 extern void unregister_blkdev(unsigned int, const char *);
2536 extern void bdev_unhash_inode(dev_t dev);
2537 extern struct block_device *bdget(dev_t);
2538 extern struct block_device *bdgrab(struct block_device *bdev);
2539 extern void bd_set_size(struct block_device *, loff_t size);
2540 extern void bd_forget(struct inode *inode);
2541 extern void bdput(struct block_device *);
2542 extern void invalidate_bdev(struct block_device *);
2543 extern void iterate_bdevs(void (*)(struct block_device *, void *), void *);
2544 extern int sync_blockdev(struct block_device *bdev);
2545 extern void kill_bdev(struct block_device *);
2546 extern struct super_block *freeze_bdev(struct block_device *);
2547 extern void emergency_thaw_all(void);
2548 extern void emergency_thaw_bdev(struct super_block *sb);
2549 extern int thaw_bdev(struct block_device *bdev, struct super_block *sb);
2550 extern int fsync_bdev(struct block_device *);
2552 extern struct super_block *blockdev_superblock;
2554 static inline bool sb_is_blkdev_sb(struct super_block *sb)
2556 return sb == blockdev_superblock;
2559 static inline void bd_forget(struct inode *inode) {}
2560 static inline int sync_blockdev(struct block_device *bdev) { return 0; }
2561 static inline void kill_bdev(struct block_device *bdev) {}
2562 static inline void invalidate_bdev(struct block_device *bdev) {}
2564 static inline struct super_block *freeze_bdev(struct block_device *sb)
2569 static inline int thaw_bdev(struct block_device *bdev, struct super_block *sb)
2574 static inline int emergency_thaw_bdev(struct super_block *sb)
2579 static inline void iterate_bdevs(void (*f)(struct block_device *, void *), void *arg)
2583 static inline bool sb_is_blkdev_sb(struct super_block *sb)
2588 extern int sync_filesystem(struct super_block *);
2589 extern const struct file_operations def_blk_fops;
2590 extern const struct file_operations def_chr_fops;
2592 extern int ioctl_by_bdev(struct block_device *, unsigned, unsigned long);
2593 extern int blkdev_ioctl(struct block_device *, fmode_t, unsigned, unsigned long);
2594 extern long compat_blkdev_ioctl(struct file *, unsigned, unsigned long);
2595 extern int blkdev_get(struct block_device *bdev, fmode_t mode, void *holder);
2596 extern struct block_device *blkdev_get_by_path(const char *path, fmode_t mode,
2598 extern struct block_device *blkdev_get_by_dev(dev_t dev, fmode_t mode,
2600 extern void blkdev_put(struct block_device *bdev, fmode_t mode);
2601 extern int __blkdev_reread_part(struct block_device *bdev);
2602 extern int blkdev_reread_part(struct block_device *bdev);
2605 extern int bd_link_disk_holder(struct block_device *bdev, struct gendisk *disk);
2606 extern void bd_unlink_disk_holder(struct block_device *bdev,
2607 struct gendisk *disk);
2609 static inline int bd_link_disk_holder(struct block_device *bdev,
2610 struct gendisk *disk)
2614 static inline void bd_unlink_disk_holder(struct block_device *bdev,
2615 struct gendisk *disk)
2622 #define CHRDEV_MAJOR_MAX 512
2623 /* Marks the bottom of the first segment of free char majors */
2624 #define CHRDEV_MAJOR_DYN_END 234
2625 /* Marks the top and bottom of the second segment of free char majors */
2626 #define CHRDEV_MAJOR_DYN_EXT_START 511
2627 #define CHRDEV_MAJOR_DYN_EXT_END 384
2629 extern int alloc_chrdev_region(dev_t *, unsigned, unsigned, const char *);
2630 extern int register_chrdev_region(dev_t, unsigned, const char *);
2631 extern int __register_chrdev(unsigned int major, unsigned int baseminor,
2632 unsigned int count, const char *name,
2633 const struct file_operations *fops);
2634 extern void __unregister_chrdev(unsigned int major, unsigned int baseminor,
2635 unsigned int count, const char *name);
2636 extern void unregister_chrdev_region(dev_t, unsigned);
2637 extern void chrdev_show(struct seq_file *,off_t);
2639 static inline int register_chrdev(unsigned int major, const char *name,
2640 const struct file_operations *fops)
2642 return __register_chrdev(major, 0, 256, name, fops);
2645 static inline void unregister_chrdev(unsigned int major, const char *name)
2647 __unregister_chrdev(major, 0, 256, name);
2650 /* fs/block_dev.c */
2651 #define BDEVNAME_SIZE 32 /* Largest string for a blockdev identifier */
2652 #define BDEVT_SIZE 10 /* Largest string for MAJ:MIN for blkdev */
2655 #define BLKDEV_MAJOR_MAX 512
2656 extern const char *__bdevname(dev_t, char *buffer);
2657 extern const char *bdevname(struct block_device *bdev, char *buffer);
2658 extern struct block_device *lookup_bdev(const char *);
2659 extern void blkdev_show(struct seq_file *,off_t);
2662 #define BLKDEV_MAJOR_MAX 0
2665 extern void init_special_inode(struct inode *, umode_t, dev_t);
2667 /* Invalid inode operations -- fs/bad_inode.c */
2668 extern void make_bad_inode(struct inode *);
2669 extern bool is_bad_inode(struct inode *);
2672 extern void check_disk_size_change(struct gendisk *disk,
2673 struct block_device *bdev, bool verbose);
2674 extern int revalidate_disk(struct gendisk *);
2675 extern int check_disk_change(struct block_device *);
2676 extern int __invalidate_device(struct block_device *, bool);
2677 extern int invalidate_partition(struct gendisk *, int);
2679 unsigned long invalidate_mapping_pages(struct address_space *mapping,
2680 pgoff_t start, pgoff_t end);
2682 static inline void invalidate_remote_inode(struct inode *inode)
2684 if (S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) ||
2685 S_ISLNK(inode->i_mode))
2686 invalidate_mapping_pages(inode->i_mapping, 0, -1);
2688 extern int invalidate_inode_pages2(struct address_space *mapping);
2689 extern int invalidate_inode_pages2_range(struct address_space *mapping,
2690 pgoff_t start, pgoff_t end);
2691 extern int write_inode_now(struct inode *, int);
2692 extern int filemap_fdatawrite(struct address_space *);
2693 extern int filemap_flush(struct address_space *);
2694 extern int filemap_fdatawait_keep_errors(struct address_space *mapping);
2695 extern int filemap_fdatawait_range(struct address_space *, loff_t lstart,
2698 static inline int filemap_fdatawait(struct address_space *mapping)
2700 return filemap_fdatawait_range(mapping, 0, LLONG_MAX);
2703 extern bool filemap_range_has_page(struct address_space *, loff_t lstart,
2705 extern int filemap_write_and_wait(struct address_space *mapping);
2706 extern int filemap_write_and_wait_range(struct address_space *mapping,
2707 loff_t lstart, loff_t lend);
2708 extern int __filemap_fdatawrite_range(struct address_space *mapping,
2709 loff_t start, loff_t end, int sync_mode);
2710 extern int filemap_fdatawrite_range(struct address_space *mapping,
2711 loff_t start, loff_t end);
2712 extern int filemap_check_errors(struct address_space *mapping);
2713 extern void __filemap_set_wb_err(struct address_space *mapping, int err);
2715 extern int __must_check file_fdatawait_range(struct file *file, loff_t lstart,
2717 extern int __must_check file_check_and_advance_wb_err(struct file *file);
2718 extern int __must_check file_write_and_wait_range(struct file *file,
2719 loff_t start, loff_t end);
2721 static inline int file_write_and_wait(struct file *file)
2723 return file_write_and_wait_range(file, 0, LLONG_MAX);
2727 * filemap_set_wb_err - set a writeback error on an address_space
2728 * @mapping: mapping in which to set writeback error
2729 * @err: error to be set in mapping
2731 * When writeback fails in some way, we must record that error so that
2732 * userspace can be informed when fsync and the like are called. We endeavor
2733 * to report errors on any file that was open at the time of the error. Some
2734 * internal callers also need to know when writeback errors have occurred.
2736 * When a writeback error occurs, most filesystems will want to call
2737 * filemap_set_wb_err to record the error in the mapping so that it will be
2738 * automatically reported whenever fsync is called on the file.
2740 static inline void filemap_set_wb_err(struct address_space *mapping, int err)
2742 /* Fastpath for common case of no error */
2744 __filemap_set_wb_err(mapping, err);
2748 * filemap_check_wb_error - has an error occurred since the mark was sampled?
2749 * @mapping: mapping to check for writeback errors
2750 * @since: previously-sampled errseq_t
2752 * Grab the errseq_t value from the mapping, and see if it has changed "since"
2753 * the given value was sampled.
2755 * If it has then report the latest error set, otherwise return 0.
2757 static inline int filemap_check_wb_err(struct address_space *mapping,
2760 return errseq_check(&mapping->wb_err, since);
2764 * filemap_sample_wb_err - sample the current errseq_t to test for later errors
2765 * @mapping: mapping to be sampled
2767 * Writeback errors are always reported relative to a particular sample point
2768 * in the past. This function provides those sample points.
2770 static inline errseq_t filemap_sample_wb_err(struct address_space *mapping)
2772 return errseq_sample(&mapping->wb_err);
2775 extern int vfs_fsync_range(struct file *file, loff_t start, loff_t end,
2777 extern int vfs_fsync(struct file *file, int datasync);
2780 * Sync the bytes written if this was a synchronous write. Expect ki_pos
2781 * to already be updated for the write, and will return either the amount
2782 * of bytes passed in, or an error if syncing the file failed.
2784 static inline ssize_t generic_write_sync(struct kiocb *iocb, ssize_t count)
2786 if (iocb->ki_flags & IOCB_DSYNC) {
2787 int ret = vfs_fsync_range(iocb->ki_filp,
2788 iocb->ki_pos - count, iocb->ki_pos - 1,
2789 (iocb->ki_flags & IOCB_SYNC) ? 0 : 1);
2797 extern void emergency_sync(void);
2798 extern void emergency_remount(void);
2800 extern sector_t bmap(struct inode *, sector_t);
2802 extern int notify_change(struct dentry *, struct iattr *, struct inode **);
2803 extern int inode_permission(struct inode *, int);
2804 extern int generic_permission(struct inode *, int);
2805 extern int __check_sticky(struct inode *dir, struct inode *inode);
2807 static inline bool execute_ok(struct inode *inode)
2809 return (inode->i_mode & S_IXUGO) || S_ISDIR(inode->i_mode);
2812 static inline void file_start_write(struct file *file)
2814 if (!S_ISREG(file_inode(file)->i_mode))
2816 __sb_start_write(file_inode(file)->i_sb, SB_FREEZE_WRITE, true);
2819 static inline bool file_start_write_trylock(struct file *file)
2821 if (!S_ISREG(file_inode(file)->i_mode))
2823 return __sb_start_write(file_inode(file)->i_sb, SB_FREEZE_WRITE, false);
2826 static inline void file_end_write(struct file *file)
2828 if (!S_ISREG(file_inode(file)->i_mode))
2830 __sb_end_write(file_inode(file)->i_sb, SB_FREEZE_WRITE);
2834 * get_write_access() gets write permission for a file.
2835 * put_write_access() releases this write permission.
2836 * This is used for regular files.
2837 * We cannot support write (and maybe mmap read-write shared) accesses and
2838 * MAP_DENYWRITE mmappings simultaneously. The i_writecount field of an inode
2839 * can have the following values:
2840 * 0: no writers, no VM_DENYWRITE mappings
2841 * < 0: (-i_writecount) vm_area_structs with VM_DENYWRITE set exist
2842 * > 0: (i_writecount) users are writing to the file.
2844 * Normally we operate on that counter with atomic_{inc,dec} and it's safe
2845 * except for the cases where we don't hold i_writecount yet. Then we need to
2846 * use {get,deny}_write_access() - these functions check the sign and refuse
2847 * to do the change if sign is wrong.
2849 static inline int get_write_access(struct inode *inode)
2851 return atomic_inc_unless_negative(&inode->i_writecount) ? 0 : -ETXTBSY;
2853 static inline int deny_write_access(struct file *file)
2855 struct inode *inode = file_inode(file);
2856 return atomic_dec_unless_positive(&inode->i_writecount) ? 0 : -ETXTBSY;
2858 static inline void put_write_access(struct inode * inode)
2860 atomic_dec(&inode->i_writecount);
2862 static inline void allow_write_access(struct file *file)
2865 atomic_inc(&file_inode(file)->i_writecount);
2867 static inline bool inode_is_open_for_write(const struct inode *inode)
2869 return atomic_read(&inode->i_writecount) > 0;
2873 static inline void i_readcount_dec(struct inode *inode)
2875 BUG_ON(!atomic_read(&inode->i_readcount));
2876 atomic_dec(&inode->i_readcount);
2878 static inline void i_readcount_inc(struct inode *inode)
2880 atomic_inc(&inode->i_readcount);
2883 static inline void i_readcount_dec(struct inode *inode)
2887 static inline void i_readcount_inc(struct inode *inode)
2892 extern int do_pipe_flags(int *, int);
2894 #define __kernel_read_file_id(id) \
2895 id(UNKNOWN, unknown) \
2896 id(FIRMWARE, firmware) \
2897 id(FIRMWARE_PREALLOC_BUFFER, firmware) \
2898 id(MODULE, kernel-module) \
2899 id(KEXEC_IMAGE, kexec-image) \
2900 id(KEXEC_INITRAMFS, kexec-initramfs) \
2901 id(POLICY, security-policy) \
2902 id(X509_CERTIFICATE, x509-certificate) \
2905 #define __fid_enumify(ENUM, dummy) READING_ ## ENUM,
2906 #define __fid_stringify(dummy, str) #str,
2908 enum kernel_read_file_id {
2909 __kernel_read_file_id(__fid_enumify)
2912 static const char * const kernel_read_file_str[] = {
2913 __kernel_read_file_id(__fid_stringify)
2916 static inline const char *kernel_read_file_id_str(enum kernel_read_file_id id)
2918 if ((unsigned)id >= READING_MAX_ID)
2919 return kernel_read_file_str[READING_UNKNOWN];
2921 return kernel_read_file_str[id];
2924 extern int kernel_read_file(struct file *, void **, loff_t *, loff_t,
2925 enum kernel_read_file_id);
2926 extern int kernel_read_file_from_path(const char *, void **, loff_t *, loff_t,
2927 enum kernel_read_file_id);
2928 extern int kernel_read_file_from_fd(int, void **, loff_t *, loff_t,
2929 enum kernel_read_file_id);
2930 extern ssize_t kernel_read(struct file *, void *, size_t, loff_t *);
2931 extern ssize_t kernel_write(struct file *, const void *, size_t, loff_t *);
2932 extern ssize_t __kernel_write(struct file *, const void *, size_t, loff_t *);
2933 extern struct file * open_exec(const char *);
2935 /* fs/dcache.c -- generic fs support functions */
2936 extern bool is_subdir(struct dentry *, struct dentry *);
2937 extern bool path_is_under(const struct path *, const struct path *);
2939 extern char *file_path(struct file *, char *, int);
2941 #include <linux/err.h>
2943 /* needed for stackable file system support */
2944 extern loff_t default_llseek(struct file *file, loff_t offset, int whence);
2946 extern loff_t vfs_llseek(struct file *file, loff_t offset, int whence);
2948 extern int inode_init_always(struct super_block *, struct inode *);
2949 extern void inode_init_once(struct inode *);
2950 extern void address_space_init_once(struct address_space *mapping);
2951 extern struct inode * igrab(struct inode *);
2952 extern ino_t iunique(struct super_block *, ino_t);
2953 extern int inode_needs_sync(struct inode *inode);
2954 extern int generic_delete_inode(struct inode *inode);
2955 static inline int generic_drop_inode(struct inode *inode)
2957 return !inode->i_nlink || inode_unhashed(inode);
2960 extern struct inode *ilookup5_nowait(struct super_block *sb,
2961 unsigned long hashval, int (*test)(struct inode *, void *),
2963 extern struct inode *ilookup5(struct super_block *sb, unsigned long hashval,
2964 int (*test)(struct inode *, void *), void *data);
2965 extern struct inode *ilookup(struct super_block *sb, unsigned long ino);
2967 extern struct inode *inode_insert5(struct inode *inode, unsigned long hashval,
2968 int (*test)(struct inode *, void *),
2969 int (*set)(struct inode *, void *),
2971 extern struct inode * iget5_locked(struct super_block *, unsigned long, int (*test)(struct inode *, void *), int (*set)(struct inode *, void *), void *);
2972 extern struct inode * iget_locked(struct super_block *, unsigned long);
2973 extern struct inode *find_inode_nowait(struct super_block *,
2975 int (*match)(struct inode *,
2976 unsigned long, void *),
2978 extern int insert_inode_locked4(struct inode *, unsigned long, int (*test)(struct inode *, void *), void *);
2979 extern int insert_inode_locked(struct inode *);
2980 #ifdef CONFIG_DEBUG_LOCK_ALLOC
2981 extern void lockdep_annotate_inode_mutex_key(struct inode *inode);
2983 static inline void lockdep_annotate_inode_mutex_key(struct inode *inode) { };
2985 extern void unlock_new_inode(struct inode *);
2986 extern void discard_new_inode(struct inode *);
2987 extern unsigned int get_next_ino(void);
2988 extern void evict_inodes(struct super_block *sb);
2990 extern void __iget(struct inode * inode);
2991 extern void iget_failed(struct inode *);
2992 extern void clear_inode(struct inode *);
2993 extern void __destroy_inode(struct inode *);
2994 extern struct inode *new_inode_pseudo(struct super_block *sb);
2995 extern struct inode *new_inode(struct super_block *sb);
2996 extern void free_inode_nonrcu(struct inode *inode);
2997 extern int should_remove_suid(struct dentry *);
2998 extern int file_remove_privs(struct file *);
3000 extern void __insert_inode_hash(struct inode *, unsigned long hashval);
3001 static inline void insert_inode_hash(struct inode *inode)
3003 __insert_inode_hash(inode, inode->i_ino);
3006 extern void __remove_inode_hash(struct inode *);
3007 static inline void remove_inode_hash(struct inode *inode)
3009 if (!inode_unhashed(inode) && !hlist_fake(&inode->i_hash))
3010 __remove_inode_hash(inode);
3013 extern void inode_sb_list_add(struct inode *inode);
3016 extern int bdev_read_only(struct block_device *);
3018 extern int set_blocksize(struct block_device *, int);
3019 extern int sb_set_blocksize(struct super_block *, int);
3020 extern int sb_min_blocksize(struct super_block *, int);
3022 extern int generic_file_mmap(struct file *, struct vm_area_struct *);
3023 extern int generic_file_readonly_mmap(struct file *, struct vm_area_struct *);
3024 extern ssize_t generic_write_checks(struct kiocb *, struct iov_iter *);
3025 extern int generic_remap_checks(struct file *file_in, loff_t pos_in,
3026 struct file *file_out, loff_t pos_out,
3027 loff_t *count, unsigned int remap_flags);
3028 extern ssize_t generic_file_read_iter(struct kiocb *, struct iov_iter *);
3029 extern ssize_t __generic_file_write_iter(struct kiocb *, struct iov_iter *);
3030 extern ssize_t generic_file_write_iter(struct kiocb *, struct iov_iter *);
3031 extern ssize_t generic_file_direct_write(struct kiocb *, struct iov_iter *);
3032 extern ssize_t generic_perform_write(struct file *, struct iov_iter *, loff_t);
3034 ssize_t vfs_iter_read(struct file *file, struct iov_iter *iter, loff_t *ppos,
3036 ssize_t vfs_iter_write(struct file *file, struct iov_iter *iter, loff_t *ppos,
3039 /* fs/block_dev.c */
3040 extern ssize_t blkdev_read_iter(struct kiocb *iocb, struct iov_iter *to);
3041 extern ssize_t blkdev_write_iter(struct kiocb *iocb, struct iov_iter *from);
3042 extern int blkdev_fsync(struct file *filp, loff_t start, loff_t end,
3044 extern void block_sync_page(struct page *page);
3047 extern ssize_t generic_file_splice_read(struct file *, loff_t *,
3048 struct pipe_inode_info *, size_t, unsigned int);
3049 extern ssize_t iter_file_splice_write(struct pipe_inode_info *,
3050 struct file *, loff_t *, size_t, unsigned int);
3051 extern ssize_t generic_splice_sendpage(struct pipe_inode_info *pipe,
3052 struct file *out, loff_t *, size_t len, unsigned int flags);
3053 extern long do_splice_direct(struct file *in, loff_t *ppos, struct file *out,
3054 loff_t *opos, size_t len, unsigned int flags);
3058 file_ra_state_init(struct file_ra_state *ra, struct address_space *mapping);
3059 extern loff_t noop_llseek(struct file *file, loff_t offset, int whence);
3060 extern loff_t no_llseek(struct file *file, loff_t offset, int whence);
3061 extern loff_t vfs_setpos(struct file *file, loff_t offset, loff_t maxsize);
3062 extern loff_t generic_file_llseek(struct file *file, loff_t offset, int whence);
3063 extern loff_t generic_file_llseek_size(struct file *file, loff_t offset,
3064 int whence, loff_t maxsize, loff_t eof);
3065 extern loff_t fixed_size_llseek(struct file *file, loff_t offset,
3066 int whence, loff_t size);
3067 extern loff_t no_seek_end_llseek_size(struct file *, loff_t, int, loff_t);
3068 extern loff_t no_seek_end_llseek(struct file *, loff_t, int);
3069 extern int generic_file_open(struct inode * inode, struct file * filp);
3070 extern int nonseekable_open(struct inode * inode, struct file * filp);
3073 typedef void (dio_submit_t)(struct bio *bio, struct inode *inode,
3074 loff_t file_offset);
3077 /* need locking between buffered and direct access */
3080 /* filesystem does not support filling holes */
3081 DIO_SKIP_HOLES = 0x02,
3084 void dio_end_io(struct bio *bio);
3085 void dio_warn_stale_pagecache(struct file *filp);
3087 ssize_t __blockdev_direct_IO(struct kiocb *iocb, struct inode *inode,
3088 struct block_device *bdev, struct iov_iter *iter,
3089 get_block_t get_block,
3090 dio_iodone_t end_io, dio_submit_t submit_io,
3093 static inline ssize_t blockdev_direct_IO(struct kiocb *iocb,
3094 struct inode *inode,
3095 struct iov_iter *iter,
3096 get_block_t get_block)
3098 return __blockdev_direct_IO(iocb, inode, inode->i_sb->s_bdev, iter,
3099 get_block, NULL, NULL, DIO_LOCKING | DIO_SKIP_HOLES);
3103 void inode_dio_wait(struct inode *inode);
3106 * inode_dio_begin - signal start of a direct I/O requests
3107 * @inode: inode the direct I/O happens on
3109 * This is called once we've finished processing a direct I/O request,
3110 * and is used to wake up callers waiting for direct I/O to be quiesced.
3112 static inline void inode_dio_begin(struct inode *inode)
3114 atomic_inc(&inode->i_dio_count);
3118 * inode_dio_end - signal finish of a direct I/O requests
3119 * @inode: inode the direct I/O happens on
3121 * This is called once we've finished processing a direct I/O request,
3122 * and is used to wake up callers waiting for direct I/O to be quiesced.
3124 static inline void inode_dio_end(struct inode *inode)
3126 if (atomic_dec_and_test(&inode->i_dio_count))
3127 wake_up_bit(&inode->i_state, __I_DIO_WAKEUP);
3130 extern void inode_set_flags(struct inode *inode, unsigned int flags,
3133 extern const struct file_operations generic_ro_fops;
3135 #define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
3137 extern int readlink_copy(char __user *, int, const char *);
3138 extern int page_readlink(struct dentry *, char __user *, int);
3139 extern const char *page_get_link(struct dentry *, struct inode *,
3140 struct delayed_call *);
3141 extern void page_put_link(void *);
3142 extern int __page_symlink(struct inode *inode, const char *symname, int len,
3144 extern int page_symlink(struct inode *inode, const char *symname, int len);
3145 extern const struct inode_operations page_symlink_inode_operations;
3146 extern void kfree_link(void *);
3147 extern void generic_fillattr(struct inode *, struct kstat *);
3148 extern int vfs_getattr_nosec(const struct path *, struct kstat *, u32, unsigned int);
3149 extern int vfs_getattr(const struct path *, struct kstat *, u32, unsigned int);
3150 void __inode_add_bytes(struct inode *inode, loff_t bytes);
3151 void inode_add_bytes(struct inode *inode, loff_t bytes);
3152 void __inode_sub_bytes(struct inode *inode, loff_t bytes);
3153 void inode_sub_bytes(struct inode *inode, loff_t bytes);
3154 static inline loff_t __inode_get_bytes(struct inode *inode)
3156 return (((loff_t)inode->i_blocks) << 9) + inode->i_bytes;
3158 loff_t inode_get_bytes(struct inode *inode);
3159 void inode_set_bytes(struct inode *inode, loff_t bytes);
3160 const char *simple_get_link(struct dentry *, struct inode *,
3161 struct delayed_call *);
3162 extern const struct inode_operations simple_symlink_inode_operations;
3164 extern int iterate_dir(struct file *, struct dir_context *);
3166 extern int vfs_statx(int, const char __user *, int, struct kstat *, u32);
3167 extern int vfs_statx_fd(unsigned int, struct kstat *, u32, unsigned int);
3169 static inline int vfs_stat(const char __user *filename, struct kstat *stat)
3171 return vfs_statx(AT_FDCWD, filename, AT_NO_AUTOMOUNT,
3172 stat, STATX_BASIC_STATS);
3174 static inline int vfs_lstat(const char __user *name, struct kstat *stat)
3176 return vfs_statx(AT_FDCWD, name, AT_SYMLINK_NOFOLLOW | AT_NO_AUTOMOUNT,
3177 stat, STATX_BASIC_STATS);
3179 static inline int vfs_fstatat(int dfd, const char __user *filename,
3180 struct kstat *stat, int flags)
3182 return vfs_statx(dfd, filename, flags | AT_NO_AUTOMOUNT,
3183 stat, STATX_BASIC_STATS);
3185 static inline int vfs_fstat(int fd, struct kstat *stat)
3187 return vfs_statx_fd(fd, stat, STATX_BASIC_STATS, 0);
3191 extern const char *vfs_get_link(struct dentry *, struct delayed_call *);
3192 extern int vfs_readlink(struct dentry *, char __user *, int);
3194 extern int __generic_block_fiemap(struct inode *inode,
3195 struct fiemap_extent_info *fieinfo,
3196 loff_t start, loff_t len,
3197 get_block_t *get_block);
3198 extern int generic_block_fiemap(struct inode *inode,
3199 struct fiemap_extent_info *fieinfo, u64 start,
3200 u64 len, get_block_t *get_block);
3202 extern struct file_system_type *get_filesystem(struct file_system_type *fs);
3203 extern void put_filesystem(struct file_system_type *fs);
3204 extern struct file_system_type *get_fs_type(const char *name);
3205 extern struct super_block *get_super(struct block_device *);
3206 extern struct super_block *get_super_thawed(struct block_device *);
3207 extern struct super_block *get_super_exclusive_thawed(struct block_device *bdev);
3208 extern struct super_block *get_active_super(struct block_device *bdev);
3209 extern void drop_super(struct super_block *sb);
3210 extern void drop_super_exclusive(struct super_block *sb);
3211 extern void iterate_supers(void (*)(struct super_block *, void *), void *);
3212 extern void iterate_supers_type(struct file_system_type *,
3213 void (*)(struct super_block *, void *), void *);
3215 extern int dcache_dir_open(struct inode *, struct file *);
3216 extern int dcache_dir_close(struct inode *, struct file *);
3217 extern loff_t dcache_dir_lseek(struct file *, loff_t, int);
3218 extern int dcache_readdir(struct file *, struct dir_context *);
3219 extern int simple_setattr(struct dentry *, struct iattr *);
3220 extern int simple_getattr(const struct path *, struct kstat *, u32, unsigned int);
3221 extern int simple_statfs(struct dentry *, struct kstatfs *);
3222 extern int simple_open(struct inode *inode, struct file *file);
3223 extern int simple_link(struct dentry *, struct inode *, struct dentry *);
3224 extern int simple_unlink(struct inode *, struct dentry *);
3225 extern int simple_rmdir(struct inode *, struct dentry *);
3226 extern int simple_rename(struct inode *, struct dentry *,
3227 struct inode *, struct dentry *, unsigned int);
3228 extern int noop_fsync(struct file *, loff_t, loff_t, int);
3229 extern int noop_set_page_dirty(struct page *page);
3230 extern void noop_invalidatepage(struct page *page, unsigned int offset,
3231 unsigned int length);
3232 extern ssize_t noop_direct_IO(struct kiocb *iocb, struct iov_iter *iter);
3233 extern int simple_empty(struct dentry *);
3234 extern int simple_readpage(struct file *file, struct page *page);
3235 extern int simple_write_begin(struct file *file, struct address_space *mapping,
3236 loff_t pos, unsigned len, unsigned flags,
3237 struct page **pagep, void **fsdata);
3238 extern int simple_write_end(struct file *file, struct address_space *mapping,
3239 loff_t pos, unsigned len, unsigned copied,
3240 struct page *page, void *fsdata);
3241 extern int always_delete_dentry(const struct dentry *);
3242 extern struct inode *alloc_anon_inode(struct super_block *);
3243 extern int simple_nosetlease(struct file *, long, struct file_lock **, void **);
3244 extern const struct dentry_operations simple_dentry_operations;
3246 extern struct dentry *simple_lookup(struct inode *, struct dentry *, unsigned int flags);
3247 extern ssize_t generic_read_dir(struct file *, char __user *, size_t, loff_t *);
3248 extern const struct file_operations simple_dir_operations;
3249 extern const struct inode_operations simple_dir_inode_operations;
3250 extern void make_empty_dir_inode(struct inode *inode);
3251 extern bool is_empty_dir_inode(struct inode *inode);
3252 struct tree_descr { const char *name; const struct file_operations *ops; int mode; };
3253 struct dentry *d_alloc_name(struct dentry *, const char *);
3254 extern int simple_fill_super(struct super_block *, unsigned long,
3255 const struct tree_descr *);
3256 extern int simple_pin_fs(struct file_system_type *, struct vfsmount **mount, int *count);
3257 extern void simple_release_fs(struct vfsmount **mount, int *count);
3259 extern ssize_t simple_read_from_buffer(void __user *to, size_t count,
3260 loff_t *ppos, const void *from, size_t available);
3261 extern ssize_t simple_write_to_buffer(void *to, size_t available, loff_t *ppos,
3262 const void __user *from, size_t count);
3264 extern int __generic_file_fsync(struct file *, loff_t, loff_t, int);
3265 extern int generic_file_fsync(struct file *, loff_t, loff_t, int);
3267 extern int generic_check_addressable(unsigned, u64);
3269 #ifdef CONFIG_MIGRATION
3270 extern int buffer_migrate_page(struct address_space *,
3271 struct page *, struct page *,
3273 extern int buffer_migrate_page_norefs(struct address_space *,
3274 struct page *, struct page *,
3277 #define buffer_migrate_page NULL
3278 #define buffer_migrate_page_norefs NULL
3281 extern int setattr_prepare(struct dentry *, struct iattr *);
3282 extern int inode_newsize_ok(const struct inode *, loff_t offset);
3283 extern void setattr_copy(struct inode *inode, const struct iattr *attr);
3285 extern int file_update_time(struct file *file);
3287 static inline bool io_is_direct(struct file *filp)
3289 return (filp->f_flags & O_DIRECT) || IS_DAX(filp->f_mapping->host);
3292 static inline bool vma_is_dax(struct vm_area_struct *vma)
3294 return vma->vm_file && IS_DAX(vma->vm_file->f_mapping->host);
3297 static inline bool vma_is_fsdax(struct vm_area_struct *vma)
3299 struct inode *inode;
3303 if (!vma_is_dax(vma))
3305 inode = file_inode(vma->vm_file);
3306 if (S_ISCHR(inode->i_mode))
3307 return false; /* device-dax */
3311 static inline int iocb_flags(struct file *file)
3314 if (file->f_flags & O_APPEND)
3316 if (io_is_direct(file))
3318 if ((file->f_flags & O_DSYNC) || IS_SYNC(file->f_mapping->host))
3320 if (file->f_flags & __O_SYNC)
3325 static inline int kiocb_set_rw_flags(struct kiocb *ki, rwf_t flags)
3327 if (unlikely(flags & ~RWF_SUPPORTED))
3330 if (flags & RWF_NOWAIT) {
3331 if (!(ki->ki_filp->f_mode & FMODE_NOWAIT))
3333 ki->ki_flags |= IOCB_NOWAIT;
3335 if (flags & RWF_HIPRI)
3336 ki->ki_flags |= IOCB_HIPRI;
3337 if (flags & RWF_DSYNC)
3338 ki->ki_flags |= IOCB_DSYNC;
3339 if (flags & RWF_SYNC)
3340 ki->ki_flags |= (IOCB_DSYNC | IOCB_SYNC);
3341 if (flags & RWF_APPEND)
3342 ki->ki_flags |= IOCB_APPEND;
3346 static inline ino_t parent_ino(struct dentry *dentry)
3351 * Don't strictly need d_lock here? If the parent ino could change
3352 * then surely we'd have a deeper race in the caller?
3354 spin_lock(&dentry->d_lock);
3355 res = dentry->d_parent->d_inode->i_ino;
3356 spin_unlock(&dentry->d_lock);
3360 /* Transaction based IO helpers */
3363 * An argresp is stored in an allocated page and holds the
3364 * size of the argument or response, along with its content
3366 struct simple_transaction_argresp {
3371 #define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
3373 char *simple_transaction_get(struct file *file, const char __user *buf,
3375 ssize_t simple_transaction_read(struct file *file, char __user *buf,
3376 size_t size, loff_t *pos);
3377 int simple_transaction_release(struct inode *inode, struct file *file);
3379 void simple_transaction_set(struct file *file, size_t n);
3382 * simple attribute files
3384 * These attributes behave similar to those in sysfs:
3386 * Writing to an attribute immediately sets a value, an open file can be
3387 * written to multiple times.
3389 * Reading from an attribute creates a buffer from the value that might get
3390 * read with multiple read calls. When the attribute has been read
3391 * completely, no further read calls are possible until the file is opened
3394 * All attributes contain a text representation of a numeric value
3395 * that are accessed with the get() and set() functions.
3397 #define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt) \
3398 static int __fops ## _open(struct inode *inode, struct file *file) \
3400 __simple_attr_check_format(__fmt, 0ull); \
3401 return simple_attr_open(inode, file, __get, __set, __fmt); \
3403 static const struct file_operations __fops = { \
3404 .owner = THIS_MODULE, \
3405 .open = __fops ## _open, \
3406 .release = simple_attr_release, \
3407 .read = simple_attr_read, \
3408 .write = simple_attr_write, \
3409 .llseek = generic_file_llseek, \
3412 static inline __printf(1, 2)
3413 void __simple_attr_check_format(const char *fmt, ...)
3415 /* don't do anything, just let the compiler check the arguments; */
3418 int simple_attr_open(struct inode *inode, struct file *file,
3419 int (*get)(void *, u64 *), int (*set)(void *, u64),
3421 int simple_attr_release(struct inode *inode, struct file *file);
3422 ssize_t simple_attr_read(struct file *file, char __user *buf,
3423 size_t len, loff_t *ppos);
3424 ssize_t simple_attr_write(struct file *file, const char __user *buf,
3425 size_t len, loff_t *ppos);
3428 int proc_nr_files(struct ctl_table *table, int write,
3429 void __user *buffer, size_t *lenp, loff_t *ppos);
3430 int proc_nr_dentry(struct ctl_table *table, int write,
3431 void __user *buffer, size_t *lenp, loff_t *ppos);
3432 int proc_nr_inodes(struct ctl_table *table, int write,
3433 void __user *buffer, size_t *lenp, loff_t *ppos);
3434 int __init get_filesystem_list(char *buf);
3436 #define __FMODE_EXEC ((__force int) FMODE_EXEC)
3437 #define __FMODE_NONOTIFY ((__force int) FMODE_NONOTIFY)
3439 #define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE])
3440 #define OPEN_FMODE(flag) ((__force fmode_t)(((flag + 1) & O_ACCMODE) | \
3441 (flag & __FMODE_NONOTIFY)))
3443 static inline bool is_sxid(umode_t mode)
3445 return (mode & S_ISUID) || ((mode & S_ISGID) && (mode & S_IXGRP));
3448 static inline int check_sticky(struct inode *dir, struct inode *inode)
3450 if (!(dir->i_mode & S_ISVTX))
3453 return __check_sticky(dir, inode);
3456 static inline void inode_has_no_xattr(struct inode *inode)
3458 if (!is_sxid(inode->i_mode) && (inode->i_sb->s_flags & SB_NOSEC))
3459 inode->i_flags |= S_NOSEC;
3462 static inline bool is_root_inode(struct inode *inode)
3464 return inode == inode->i_sb->s_root->d_inode;
3467 static inline bool dir_emit(struct dir_context *ctx,
3468 const char *name, int namelen,
3469 u64 ino, unsigned type)
3471 return ctx->actor(ctx, name, namelen, ctx->pos, ino, type) == 0;
3473 static inline bool dir_emit_dot(struct file *file, struct dir_context *ctx)
3475 return ctx->actor(ctx, ".", 1, ctx->pos,
3476 file->f_path.dentry->d_inode->i_ino, DT_DIR) == 0;
3478 static inline bool dir_emit_dotdot(struct file *file, struct dir_context *ctx)
3480 return ctx->actor(ctx, "..", 2, ctx->pos,
3481 parent_ino(file->f_path.dentry), DT_DIR) == 0;
3483 static inline bool dir_emit_dots(struct file *file, struct dir_context *ctx)
3485 if (ctx->pos == 0) {
3486 if (!dir_emit_dot(file, ctx))
3490 if (ctx->pos == 1) {
3491 if (!dir_emit_dotdot(file, ctx))
3497 static inline bool dir_relax(struct inode *inode)
3499 inode_unlock(inode);
3501 return !IS_DEADDIR(inode);
3504 static inline bool dir_relax_shared(struct inode *inode)
3506 inode_unlock_shared(inode);
3507 inode_lock_shared(inode);
3508 return !IS_DEADDIR(inode);
3511 extern bool path_noexec(const struct path *path);
3512 extern void inode_nohighmem(struct inode *inode);
3515 extern int vfs_fadvise(struct file *file, loff_t offset, loff_t len,
3518 #endif /* _LINUX_FS_H */