Home | History | Annotate | Download | only in linux
      1 #ifndef _LINUX_FS_H
      2 #define _LINUX_FS_H
      3 
      4 /*
      5  * This file has definitions for some important file table
      6  * structures etc.
      7  */
      8 
      9 #include <linux/limits.h>
     10 #include <linux/ioctl.h>
     11 #include <linux/blk_types.h>
     12 #include <linux/types.h>
     13 
     14 /*
     15  * It's silly to have NR_OPEN bigger than NR_FILE, but you can change
     16  * the file limit at runtime and only root can increase the per-process
     17  * nr_file rlimit, so it's safe to set up a ridiculously high absolute
     18  * upper limit on files-per-process.
     19  *
     20  * Some programs (notably those using select()) may have to be
     21  * recompiled to take full advantage of the new limits..
     22  */
     23 
     24 /* Fixed constants first: */
     25 #undef NR_OPEN
     26 #define INR_OPEN_CUR 1024	/* Initial setting for nfile rlimits */
     27 #define INR_OPEN_MAX 4096	/* Hard limit for nfile rlimits */
     28 
     29 #define BLOCK_SIZE_BITS 10
     30 #define BLOCK_SIZE (1<<BLOCK_SIZE_BITS)
     31 
     32 #define SEEK_SET	0	/* seek relative to beginning of file */
     33 #define SEEK_CUR	1	/* seek relative to current file position */
     34 #define SEEK_END	2	/* seek relative to end of file */
     35 #define SEEK_MAX	SEEK_END
     36 
     37 struct fstrim_range {
     38 	__u64 start;
     39 	__u64 len;
     40 	__u64 minlen;
     41 };
     42 
     43 /* And dynamically-tunable limits and defaults: */
     44 struct files_stat_struct {
     45 	unsigned long nr_files;		/* read only */
     46 	unsigned long nr_free_files;	/* read only */
     47 	unsigned long max_files;		/* tunable */
     48 };
     49 
     50 struct inodes_stat_t {
     51 	int nr_inodes;
     52 	int nr_unused;
     53 	int dummy[5];		/* padding for sysctl ABI compatibility */
     54 };
     55 
     56 
     57 #define NR_FILE  8192	/* this can well be larger on a larger system */
     58 
     59 #define MAY_EXEC 1
     60 #define MAY_WRITE 2
     61 #define MAY_READ 4
     62 #define MAY_APPEND 8
     63 #define MAY_ACCESS 16
     64 #define MAY_OPEN 32
     65 #define MAY_CHDIR 64
     66 
     67 /*
     68  * flags in file.f_mode.  Note that FMODE_READ and FMODE_WRITE must correspond
     69  * to O_WRONLY and O_RDWR via the strange trick in __dentry_open()
     70  */
     71 
     72 /* file is open for reading */
     73 #define FMODE_READ		((__force fmode_t)0x1)
     74 /* file is open for writing */
     75 #define FMODE_WRITE		((__force fmode_t)0x2)
     76 /* file is seekable */
     77 #define FMODE_LSEEK		((__force fmode_t)0x4)
     78 /* file can be accessed using pread */
     79 #define FMODE_PREAD		((__force fmode_t)0x8)
     80 /* file can be accessed using pwrite */
     81 #define FMODE_PWRITE		((__force fmode_t)0x10)
     82 /* File is opened for execution with sys_execve / sys_uselib */
     83 #define FMODE_EXEC		((__force fmode_t)0x20)
     84 /* File is opened with O_NDELAY (only set for block devices) */
     85 #define FMODE_NDELAY		((__force fmode_t)0x40)
     86 /* File is opened with O_EXCL (only set for block devices) */
     87 #define FMODE_EXCL		((__force fmode_t)0x80)
     88 /* File is opened using open(.., 3, ..) and is writeable only for ioctls
     89    (specialy hack for floppy.c) */
     90 #define FMODE_WRITE_IOCTL	((__force fmode_t)0x100)
     91 
     92 /*
     93  * Don't update ctime and mtime.
     94  *
     95  * Currently a special hack for the XFS open_by_handle ioctl, but we'll
     96  * hopefully graduate it to a proper O_CMTIME flag supported by open(2) soon.
     97  */
     98 #define FMODE_NOCMTIME		((__force fmode_t)0x800)
     99 
    100 /* Expect random access pattern */
    101 #define FMODE_RANDOM		((__force fmode_t)0x1000)
    102 
    103 /* File is huge (eg. /dev/kmem): treat loff_t as unsigned */
    104 #define FMODE_UNSIGNED_OFFSET	((__force fmode_t)0x2000)
    105 
    106 /* File is opened with O_PATH; almost nothing can be done with it */
    107 #define FMODE_PATH		((__force fmode_t)0x4000)
    108 
    109 /* File was opened by fanotify and shouldn't generate fanotify events */
    110 #define FMODE_NONOTIFY		((__force fmode_t)0x1000000)
    111 
    112 /*
    113  * The below are the various read and write types that we support. Some of
    114  * them include behavioral modifiers that send information down to the
    115  * block layer and IO scheduler. Terminology:
    116  *
    117  *	The block layer uses device plugging to defer IO a little bit, in
    118  *	the hope that we will see more IO very shortly. This increases
    119  *	coalescing of adjacent IO and thus reduces the number of IOs we
    120  *	have to send to the device. It also allows for better queuing,
    121  *	if the IO isn't mergeable. If the caller is going to be waiting
    122  *	for the IO, then he must ensure that the device is unplugged so
    123  *	that the IO is dispatched to the driver.
    124  *
    125  *	All IO is handled async in Linux. This is fine for background
    126  *	writes, but for reads or writes that someone waits for completion
    127  *	on, we want to notify the block layer and IO scheduler so that they
    128  *	know about it. That allows them to make better scheduling
    129  *	decisions. So when the below references 'sync' and 'async', it
    130  *	is referencing this priority hint.
    131  *
    132  * With that in mind, the available types are:
    133  *
    134  * READ			A normal read operation. Device will be plugged.
    135  * READ_SYNC		A synchronous read. Device is not plugged, caller can
    136  *			immediately wait on this read without caring about
    137  *			unplugging.
    138  * READA		Used for read-ahead operations. Lower priority, and the
    139  *			block layer could (in theory) choose to ignore this
    140  *			request if it runs into resource problems.
    141  * WRITE		A normal async write. Device will be plugged.
    142  * WRITE_SYNC		Synchronous write. Identical to WRITE, but passes down
    143  *			the hint that someone will be waiting on this IO
    144  *			shortly. The write equivalent of READ_SYNC.
    145  * WRITE_ODIRECT	Special case write for O_DIRECT only.
    146  * WRITE_FLUSH		Like WRITE_SYNC but with preceding cache flush.
    147  * WRITE_FUA		Like WRITE_SYNC but data is guaranteed to be on
    148  *			non-volatile media on completion.
    149  * WRITE_FLUSH_FUA	Combination of WRITE_FLUSH and FUA. The IO is preceded
    150  *			by a cache flush and data is guaranteed to be on
    151  *			non-volatile media on completion.
    152  *
    153  */
    154 #define RW_MASK			REQ_WRITE
    155 #define RWA_MASK		REQ_RAHEAD
    156 
    157 #define READ			0
    158 #define WRITE			RW_MASK
    159 #define READA			RWA_MASK
    160 
    161 #define READ_SYNC		(READ | REQ_SYNC)
    162 #define READ_META		(READ | REQ_META)
    163 #define WRITE_SYNC		(WRITE | REQ_SYNC | REQ_NOIDLE)
    164 #define WRITE_ODIRECT		(WRITE | REQ_SYNC)
    165 #define WRITE_META		(WRITE | REQ_META)
    166 #define WRITE_FLUSH		(WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FLUSH)
    167 #define WRITE_FUA		(WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FUA)
    168 #define WRITE_FLUSH_FUA		(WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FLUSH | REQ_FUA)
    169 
    170 #define SEL_IN		1
    171 #define SEL_OUT		2
    172 #define SEL_EX		4
    173 
    174 /* public flags for file_system_type */
    175 #define FS_REQUIRES_DEV 1
    176 #define FS_BINARY_MOUNTDATA 2
    177 #define FS_HAS_SUBTYPE 4
    178 #define FS_REVAL_DOT	16384	/* Check the paths ".", ".." for staleness */
    179 #define FS_RENAME_DOES_D_MOVE	32768	/* FS will handle d_move()
    180 					 * during rename() internally.
    181 					 */
    182 
    183 /*
    184  * These are the fs-independent mount-flags: up to 32 flags are supported
    185  */
    186 #define MS_RDONLY	 1	/* Mount read-only */
    187 #define MS_NOSUID	 2	/* Ignore suid and sgid bits */
    188 #define MS_NODEV	 4	/* Disallow access to device special files */
    189 #define MS_NOEXEC	 8	/* Disallow program execution */
    190 #define MS_SYNCHRONOUS	16	/* Writes are synced at once */
    191 #define MS_REMOUNT	32	/* Alter flags of a mounted FS */
    192 #define MS_MANDLOCK	64	/* Allow mandatory locks on an FS */
    193 #define MS_DIRSYNC	128	/* Directory modifications are synchronous */
    194 #define MS_NOATIME	1024	/* Do not update access times. */
    195 #define MS_NODIRATIME	2048	/* Do not update directory access times */
    196 #define MS_BIND		4096
    197 #define MS_MOVE		8192
    198 #define MS_REC		16384
    199 #define MS_VERBOSE	32768	/* War is peace. Verbosity is silence.
    200 				   MS_VERBOSE is deprecated. */
    201 #define MS_SILENT	32768
    202 #define MS_POSIXACL	(1<<16)	/* VFS does not apply the umask */
    203 #define MS_UNBINDABLE	(1<<17)	/* change to unbindable */
    204 #define MS_PRIVATE	(1<<18)	/* change to private */
    205 #define MS_SLAVE	(1<<19)	/* change to slave */
    206 #define MS_SHARED	(1<<20)	/* change to shared */
    207 #define MS_RELATIME	(1<<21)	/* Update atime relative to mtime/ctime. */
    208 #define MS_KERNMOUNT	(1<<22) /* this is a kern_mount call */
    209 #define MS_I_VERSION	(1<<23) /* Update inode I_version field */
    210 #define MS_STRICTATIME	(1<<24) /* Always perform atime updates */
    211 #define MS_NOSEC	(1<<28)
    212 #define MS_BORN		(1<<29)
    213 #define MS_ACTIVE	(1<<30)
    214 #define MS_NOUSER	(1<<31)
    215 
    216 /*
    217  * Superblock flags that can be altered by MS_REMOUNT
    218  */
    219 #define MS_RMT_MASK	(MS_RDONLY|MS_SYNCHRONOUS|MS_MANDLOCK|MS_I_VERSION)
    220 
    221 /*
    222  * Old magic mount flag and mask
    223  */
    224 #define MS_MGC_VAL 0xC0ED0000
    225 #define MS_MGC_MSK 0xffff0000
    226 
    227 /* Inode flags - they have nothing to superblock flags now */
    228 
    229 #define S_SYNC		1	/* Writes are synced at once */
    230 #define S_NOATIME	2	/* Do not update access times */
    231 #define S_APPEND	4	/* Append-only file */
    232 #define S_IMMUTABLE	8	/* Immutable file */
    233 #define S_DEAD		16	/* removed, but still open directory */
    234 #define S_NOQUOTA	32	/* Inode is not counted to quota */
    235 #define S_DIRSYNC	64	/* Directory modifications are synchronous */
    236 #define S_NOCMTIME	128	/* Do not update file c/mtime */
    237 #define S_SWAPFILE	256	/* Do not truncate: swapon got its bmaps */
    238 #define S_PRIVATE	512	/* Inode is fs-internal */
    239 #define S_IMA		1024	/* Inode has an associated IMA struct */
    240 #define S_AUTOMOUNT	2048	/* Automount/referral quasi-directory */
    241 #define S_NOSEC		4096	/* no suid or xattr security attributes */
    242 
    243 /*
    244  * Note that nosuid etc flags are inode-specific: setting some file-system
    245  * flags just means all the inodes inherit those flags by default. It might be
    246  * possible to override it selectively if you really wanted to with some
    247  * ioctl() that is not currently implemented.
    248  *
    249  * Exception: MS_RDONLY is always applied to the entire file system.
    250  *
    251  * Unfortunately, it is possible to change a filesystems flags with it mounted
    252  * with files in use.  This means that all of the inodes will not have their
    253  * i_flags updated.  Hence, i_flags no longer inherit the superblock mount
    254  * flags, so these have to be checked separately. -- rmk (at) arm.uk.linux.org
    255  */
    256 #define __IS_FLG(inode,flg) ((inode)->i_sb->s_flags & (flg))
    257 
    258 #define IS_RDONLY(inode) ((inode)->i_sb->s_flags & MS_RDONLY)
    259 #define IS_SYNC(inode)		(__IS_FLG(inode, MS_SYNCHRONOUS) || \
    260 					((inode)->i_flags & S_SYNC))
    261 #define IS_DIRSYNC(inode)	(__IS_FLG(inode, MS_SYNCHRONOUS|MS_DIRSYNC) || \
    262 					((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
    263 #define IS_MANDLOCK(inode)	__IS_FLG(inode, MS_MANDLOCK)
    264 #define IS_NOATIME(inode)   __IS_FLG(inode, MS_RDONLY|MS_NOATIME)
    265 #define IS_I_VERSION(inode)   __IS_FLG(inode, MS_I_VERSION)
    266 
    267 #define IS_NOQUOTA(inode)	((inode)->i_flags & S_NOQUOTA)
    268 #define IS_APPEND(inode)	((inode)->i_flags & S_APPEND)
    269 #define IS_IMMUTABLE(inode)	((inode)->i_flags & S_IMMUTABLE)
    270 #define IS_POSIXACL(inode)	__IS_FLG(inode, MS_POSIXACL)
    271 
    272 #define IS_DEADDIR(inode)	((inode)->i_flags & S_DEAD)
    273 #define IS_NOCMTIME(inode)	((inode)->i_flags & S_NOCMTIME)
    274 #define IS_SWAPFILE(inode)	((inode)->i_flags & S_SWAPFILE)
    275 #define IS_PRIVATE(inode)	((inode)->i_flags & S_PRIVATE)
    276 #define IS_IMA(inode)		((inode)->i_flags & S_IMA)
    277 #define IS_AUTOMOUNT(inode)	((inode)->i_flags & S_AUTOMOUNT)
    278 #define IS_NOSEC(inode)		((inode)->i_flags & S_NOSEC)
    279 
    280 /* the read-only stuff doesn't really belong here, but any other place is
    281    probably as bad and I don't want to create yet another include file. */
    282 
    283 #define BLKROSET   _IO(0x12,93)	/* set device read-only (0 = read-write) */
    284 #define BLKROGET   _IO(0x12,94)	/* get read-only status (0 = read_write) */
    285 #define BLKRRPART  _IO(0x12,95)	/* re-read partition table */
    286 #define BLKGETSIZE _IO(0x12,96)	/* return device size /512 (long *arg) */
    287 #define BLKFLSBUF  _IO(0x12,97)	/* flush buffer cache */
    288 #define BLKRASET   _IO(0x12,98)	/* set read ahead for block device */
    289 #define BLKRAGET   _IO(0x12,99)	/* get current read ahead setting */
    290 #define BLKFRASET  _IO(0x12,100)/* set filesystem (mm/filemap.c) read-ahead */
    291 #define BLKFRAGET  _IO(0x12,101)/* get filesystem (mm/filemap.c) read-ahead */
    292 #define BLKSECTSET _IO(0x12,102)/* set max sectors per request (ll_rw_blk.c) */
    293 #define BLKSECTGET _IO(0x12,103)/* get max sectors per request (ll_rw_blk.c) */
    294 #define BLKSSZGET  _IO(0x12,104)/* get block device sector size */
    295 #if 0
    296 #define BLKPG      _IO(0x12,105)/* See blkpg.h */
    297 
    298 /* Some people are morons.  Do not use sizeof! */
    299 
    300 #define BLKELVGET  _IOR(0x12,106,size_t)/* elevator get */
    301 #define BLKELVSET  _IOW(0x12,107,size_t)/* elevator set */
    302 /* This was here just to show that the number is taken -
    303    probably all these _IO(0x12,*) ioctls should be moved to blkpg.h. */
    304 #endif
    305 /* A jump here: 108-111 have been used for various private purposes. */
    306 #define BLKBSZGET  _IOR(0x12,112,size_t)
    307 #define BLKBSZSET  _IOW(0x12,113,size_t)
    308 #define BLKGETSIZE64 _IOR(0x12,114,size_t)	/* return device size in bytes (u64 *arg) */
    309 #define BLKTRACESETUP _IOWR(0x12,115,struct blk_user_trace_setup)
    310 #define BLKTRACESTART _IO(0x12,116)
    311 #define BLKTRACESTOP _IO(0x12,117)
    312 #define BLKTRACETEARDOWN _IO(0x12,118)
    313 #define BLKDISCARD _IO(0x12,119)
    314 #define BLKIOMIN _IO(0x12,120)
    315 #define BLKIOOPT _IO(0x12,121)
    316 #define BLKALIGNOFF _IO(0x12,122)
    317 #define BLKPBSZGET _IO(0x12,123)
    318 #define BLKDISCARDZEROES _IO(0x12,124)
    319 #define BLKSECDISCARD _IO(0x12,125)
    320 
    321 #define BMAP_IOCTL 1		/* obsolete - kept for compatibility */
    322 #define FIBMAP	   _IO(0x00,1)	/* bmap access */
    323 #define FIGETBSZ   _IO(0x00,2)	/* get the block size used for bmap */
    324 #define FIFREEZE	_IOWR('X', 119, int)	/* Freeze */
    325 #define FITHAW		_IOWR('X', 120, int)	/* Thaw */
    326 #define FITRIM		_IOWR('X', 121, struct fstrim_range)	/* Trim */
    327 
    328 #define	FS_IOC_GETFLAGS			_IOR('f', 1, long)
    329 #define	FS_IOC_SETFLAGS			_IOW('f', 2, long)
    330 #define	FS_IOC_GETVERSION		_IOR('v', 1, long)
    331 #define	FS_IOC_SETVERSION		_IOW('v', 2, long)
    332 #define FS_IOC_FIEMAP			_IOWR('f', 11, struct fiemap)
    333 #define FS_IOC32_GETFLAGS		_IOR('f', 1, int)
    334 #define FS_IOC32_SETFLAGS		_IOW('f', 2, int)
    335 #define FS_IOC32_GETVERSION		_IOR('v', 1, int)
    336 #define FS_IOC32_SETVERSION		_IOW('v', 2, int)
    337 
    338 /*
    339  * Inode flags (FS_IOC_GETFLAGS / FS_IOC_SETFLAGS)
    340  */
    341 #define	FS_SECRM_FL			0x00000001 /* Secure deletion */
    342 #define	FS_UNRM_FL			0x00000002 /* Undelete */
    343 #define	FS_COMPR_FL			0x00000004 /* Compress file */
    344 #define FS_SYNC_FL			0x00000008 /* Synchronous updates */
    345 #define FS_IMMUTABLE_FL			0x00000010 /* Immutable file */
    346 #define FS_APPEND_FL			0x00000020 /* writes to file may only append */
    347 #define FS_NODUMP_FL			0x00000040 /* do not dump file */
    348 #define FS_NOATIME_FL			0x00000080 /* do not update atime */
    349 /* Reserved for compression usage... */
    350 #define FS_DIRTY_FL			0x00000100
    351 #define FS_COMPRBLK_FL			0x00000200 /* One or more compressed clusters */
    352 #define FS_NOCOMP_FL			0x00000400 /* Don't compress */
    353 #define FS_ECOMPR_FL			0x00000800 /* Compression error */
    354 /* End compression flags --- maybe not all used */
    355 #define FS_BTREE_FL			0x00001000 /* btree format dir */
    356 #define FS_INDEX_FL			0x00001000 /* hash-indexed directory */
    357 #define FS_IMAGIC_FL			0x00002000 /* AFS directory */
    358 #define FS_JOURNAL_DATA_FL		0x00004000 /* Reserved for ext3 */
    359 #define FS_NOTAIL_FL			0x00008000 /* file tail should not be merged */
    360 #define FS_DIRSYNC_FL			0x00010000 /* dirsync behaviour (directories only) */
    361 #define FS_TOPDIR_FL			0x00020000 /* Top of directory hierarchies*/
    362 #define FS_EXTENT_FL			0x00080000 /* Extents */
    363 #define FS_DIRECTIO_FL			0x00100000 /* Use direct i/o */
    364 #define FS_NOCOW_FL			0x00800000 /* Do not cow file */
    365 #define FS_RESERVED_FL			0x80000000 /* reserved for ext2 lib */
    366 
    367 #define FS_FL_USER_VISIBLE		0x0003DFFF /* User visible flags */
    368 #define FS_FL_USER_MODIFIABLE		0x000380FF /* User modifiable flags */
    369 
    370 
    371 #define SYNC_FILE_RANGE_WAIT_BEFORE	1
    372 #define SYNC_FILE_RANGE_WRITE		2
    373 #define SYNC_FILE_RANGE_WAIT_AFTER	4
    374 
    375 #ifdef __KERNEL__
    376 
    377 #include <linux/linkage.h>
    378 #include <linux/wait.h>
    379 #include <linux/types.h>
    380 #include <linux/kdev_t.h>
    381 #include <linux/dcache.h>
    382 #include <linux/path.h>
    383 #include <linux/stat.h>
    384 #include <linux/cache.h>
    385 #include <linux/list.h>
    386 #include <linux/radix-tree.h>
    387 #include <linux/prio_tree.h>
    388 #include <linux/init.h>
    389 #include <linux/pid.h>
    390 #include <linux/mutex.h>
    391 #include <linux/capability.h>
    392 #include <linux/semaphore.h>
    393 #include <linux/fiemap.h>
    394 #include <linux/rculist_bl.h>
    395 
    396 #include <asm/atomic.h>
    397 #include <asm/byteorder.h>
    398 
    399 struct export_operations;
    400 struct hd_geometry;
    401 struct iovec;
    402 struct nameidata;
    403 struct kiocb;
    404 struct kobject;
    405 struct pipe_inode_info;
    406 struct poll_table_struct;
    407 struct kstatfs;
    408 struct vm_area_struct;
    409 struct vfsmount;
    410 struct cred;
    411 
    412 extern void __init inode_init(void);
    413 extern void __init inode_init_early(void);
    414 extern void __init files_init(unsigned long);
    415 
    416 extern struct files_stat_struct files_stat;
    417 extern unsigned long get_max_files(void);
    418 extern int sysctl_nr_open;
    419 extern struct inodes_stat_t inodes_stat;
    420 extern int leases_enable, lease_break_time;
    421 
    422 struct buffer_head;
    423 typedef int (get_block_t)(struct inode *inode, sector_t iblock,
    424 			struct buffer_head *bh_result, int create);
    425 typedef void (dio_iodone_t)(struct kiocb *iocb, loff_t offset,
    426 			ssize_t bytes, void *private, int ret,
    427 			bool is_async);
    428 
    429 /*
    430  * Attribute flags.  These should be or-ed together to figure out what
    431  * has been changed!
    432  */
    433 #define ATTR_MODE	(1 << 0)
    434 #define ATTR_UID	(1 << 1)
    435 #define ATTR_GID	(1 << 2)
    436 #define ATTR_SIZE	(1 << 3)
    437 #define ATTR_ATIME	(1 << 4)
    438 #define ATTR_MTIME	(1 << 5)
    439 #define ATTR_CTIME	(1 << 6)
    440 #define ATTR_ATIME_SET	(1 << 7)
    441 #define ATTR_MTIME_SET	(1 << 8)
    442 #define ATTR_FORCE	(1 << 9) /* Not a change, but a change it */
    443 #define ATTR_ATTR_FLAG	(1 << 10)
    444 #define ATTR_KILL_SUID	(1 << 11)
    445 #define ATTR_KILL_SGID	(1 << 12)
    446 #define ATTR_FILE	(1 << 13)
    447 #define ATTR_KILL_PRIV	(1 << 14)
    448 #define ATTR_OPEN	(1 << 15) /* Truncating from open(O_TRUNC) */
    449 #define ATTR_TIMES_SET	(1 << 16)
    450 
    451 /*
    452  * This is the Inode Attributes structure, used for notify_change().  It
    453  * uses the above definitions as flags, to know which values have changed.
    454  * Also, in this manner, a Filesystem can look at only the values it cares
    455  * about.  Basically, these are the attributes that the VFS layer can
    456  * request to change from the FS layer.
    457  *
    458  * Derek Atkins <warlord (at) MIT.EDU> 94-10-20
    459  */
    460 struct iattr {
    461 	unsigned int	ia_valid;
    462 	umode_t		ia_mode;
    463 	uid_t		ia_uid;
    464 	gid_t		ia_gid;
    465 	loff_t		ia_size;
    466 	struct timespec	ia_atime;
    467 	struct timespec	ia_mtime;
    468 	struct timespec	ia_ctime;
    469 
    470 	/*
    471 	 * Not an attribute, but an auxiliary info for filesystems wanting to
    472 	 * implement an ftruncate() like method.  NOTE: filesystem should
    473 	 * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL).
    474 	 */
    475 	struct file	*ia_file;
    476 };
    477 
    478 /*
    479  * Includes for diskquotas.
    480  */
    481 #include <linux/quota.h>
    482 
    483 /**
    484  * enum positive_aop_returns - aop return codes with specific semantics
    485  *
    486  * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has
    487  * 			    completed, that the page is still locked, and
    488  * 			    should be considered active.  The VM uses this hint
    489  * 			    to return the page to the active list -- it won't
    490  * 			    be a candidate for writeback again in the near
    491  * 			    future.  Other callers must be careful to unlock
    492  * 			    the page if they get this return.  Returned by
    493  * 			    writepage();
    494  *
    495  * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has
    496  *  			unlocked it and the page might have been truncated.
    497  *  			The caller should back up to acquiring a new page and
    498  *  			trying again.  The aop will be taking reasonable
    499  *  			precautions not to livelock.  If the caller held a page
    500  *  			reference, it should drop it before retrying.  Returned
    501  *  			by readpage().
    502  *
    503  * address_space_operation functions return these large constants to indicate
    504  * special semantics to the caller.  These are much larger than the bytes in a
    505  * page to allow for functions that return the number of bytes operated on in a
    506  * given page.
    507  */
    508 
    509 enum positive_aop_returns {
    510 	AOP_WRITEPAGE_ACTIVATE	= 0x80000,
    511 	AOP_TRUNCATED_PAGE	= 0x80001,
    512 };
    513 
    514 #define AOP_FLAG_UNINTERRUPTIBLE	0x0001 /* will not do a short write */
    515 #define AOP_FLAG_CONT_EXPAND		0x0002 /* called from cont_expand */
    516 #define AOP_FLAG_NOFS			0x0004 /* used by filesystem to direct
    517 						* helper code (eg buffer layer)
    518 						* to clear GFP_FS from alloc */
    519 
    520 /*
    521  * oh the beauties of C type declarations.
    522  */
    523 struct page;
    524 struct address_space;
    525 struct writeback_control;
    526 
    527 struct iov_iter {
    528 	const struct iovec *iov;
    529 	unsigned long nr_segs;
    530 	size_t iov_offset;
    531 	size_t count;
    532 };
    533 
    534 size_t iov_iter_copy_from_user_atomic(struct page *page,
    535 		struct iov_iter *i, unsigned long offset, size_t bytes);
    536 size_t iov_iter_copy_from_user(struct page *page,
    537 		struct iov_iter *i, unsigned long offset, size_t bytes);
    538 void iov_iter_advance(struct iov_iter *i, size_t bytes);
    539 int iov_iter_fault_in_readable(struct iov_iter *i, size_t bytes);
    540 size_t iov_iter_single_seg_count(struct iov_iter *i);
    541 
    542 static inline void iov_iter_init(struct iov_iter *i,
    543 			const struct iovec *iov, unsigned long nr_segs,
    544 			size_t count, size_t written)
    545 {
    546 	i->iov = iov;
    547 	i->nr_segs = nr_segs;
    548 	i->iov_offset = 0;
    549 	i->count = count + written;
    550 
    551 	iov_iter_advance(i, written);
    552 }
    553 
    554 static inline size_t iov_iter_count(struct iov_iter *i)
    555 {
    556 	return i->count;
    557 }
    558 
    559 /*
    560  * "descriptor" for what we're up to with a read.
    561  * This allows us to use the same read code yet
    562  * have multiple different users of the data that
    563  * we read from a file.
    564  *
    565  * The simplest case just copies the data to user
    566  * mode.
    567  */
    568 typedef struct {
    569 	size_t written;
    570 	size_t count;
    571 	union {
    572 		char __user *buf;
    573 		void *data;
    574 	} arg;
    575 	int error;
    576 } read_descriptor_t;
    577 
    578 typedef int (*read_actor_t)(read_descriptor_t *, struct page *,
    579 		unsigned long, unsigned long);
    580 
    581 struct address_space_operations {
    582 	int (*writepage)(struct page *page, struct writeback_control *wbc);
    583 	int (*readpage)(struct file *, struct page *);
    584 
    585 	/* Write back some dirty pages from this mapping. */
    586 	int (*writepages)(struct address_space *, struct writeback_control *);
    587 
    588 	/* Set a page dirty.  Return true if this dirtied it */
    589 	int (*set_page_dirty)(struct page *page);
    590 
    591 	int (*readpages)(struct file *filp, struct address_space *mapping,
    592 			struct list_head *pages, unsigned nr_pages);
    593 
    594 	int (*write_begin)(struct file *, struct address_space *mapping,
    595 				loff_t pos, unsigned len, unsigned flags,
    596 				struct page **pagep, void **fsdata);
    597 	int (*write_end)(struct file *, struct address_space *mapping,
    598 				loff_t pos, unsigned len, unsigned copied,
    599 				struct page *page, void *fsdata);
    600 
    601 	/* Unfortunately this kludge is needed for FIBMAP. Don't use it */
    602 	sector_t (*bmap)(struct address_space *, sector_t);
    603 	void (*invalidatepage) (struct page *, unsigned long);
    604 	int (*releasepage) (struct page *, gfp_t);
    605 	void (*freepage)(struct page *);
    606 	ssize_t (*direct_IO)(int, struct kiocb *, const struct iovec *iov,
    607 			loff_t offset, unsigned long nr_segs);
    608 	int (*get_xip_mem)(struct address_space *, pgoff_t, int,
    609 						void **, unsigned long *);
    610 	/* migrate the contents of a page to the specified target */
    611 	int (*migratepage) (struct address_space *,
    612 			struct page *, struct page *);
    613 	int (*launder_page) (struct page *);
    614 	int (*is_partially_uptodate) (struct page *, read_descriptor_t *,
    615 					unsigned long);
    616 	int (*error_remove_page)(struct address_space *, struct page *);
    617 };
    618 
    619 extern const struct address_space_operations empty_aops;
    620 
    621 /*
    622  * pagecache_write_begin/pagecache_write_end must be used by general code
    623  * to write into the pagecache.
    624  */
    625 int pagecache_write_begin(struct file *, struct address_space *mapping,
    626 				loff_t pos, unsigned len, unsigned flags,
    627 				struct page **pagep, void **fsdata);
    628 
    629 int pagecache_write_end(struct file *, struct address_space *mapping,
    630 				loff_t pos, unsigned len, unsigned copied,
    631 				struct page *page, void *fsdata);
    632 
    633 struct backing_dev_info;
    634 struct address_space {
    635 	struct inode		*host;		/* owner: inode, block_device */
    636 	struct radix_tree_root	page_tree;	/* radix tree of all pages */
    637 	spinlock_t		tree_lock;	/* and lock protecting it */
    638 	unsigned int		i_mmap_writable;/* count VM_SHARED mappings */
    639 	struct prio_tree_root	i_mmap;		/* tree of private and shared mappings */
    640 	struct list_head	i_mmap_nonlinear;/*list VM_NONLINEAR mappings */
    641 	struct mutex		i_mmap_mutex;	/* protect tree, count, list */
    642 	/* Protected by tree_lock together with the radix tree */
    643 	unsigned long		nrpages;	/* number of total pages */
    644 	pgoff_t			writeback_index;/* writeback starts here */
    645 	const struct address_space_operations *a_ops;	/* methods */
    646 	unsigned long		flags;		/* error bits/gfp mask */
    647 	struct backing_dev_info *backing_dev_info; /* device readahead, etc */
    648 	spinlock_t		private_lock;	/* for use by the address_space */
    649 	struct list_head	private_list;	/* ditto */
    650 	struct address_space	*assoc_mapping;	/* ditto */
    651 } __attribute__((aligned(sizeof(long))));
    652 	/*
    653 	 * On most architectures that alignment is already the case; but
    654 	 * must be enforced here for CRIS, to let the least significant bit
    655 	 * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
    656 	 */
    657 
    658 struct block_device {
    659 	dev_t			bd_dev;  /* not a kdev_t - it's a search key */
    660 	int			bd_openers;
    661 	struct inode *		bd_inode;	/* will die */
    662 	struct super_block *	bd_super;
    663 	struct mutex		bd_mutex;	/* open/close mutex */
    664 	struct list_head	bd_inodes;
    665 	void *			bd_claiming;
    666 	void *			bd_holder;
    667 	int			bd_holders;
    668 	bool			bd_write_holder;
    669 #ifdef CONFIG_SYSFS
    670 	struct list_head	bd_holder_disks;
    671 #endif
    672 	struct block_device *	bd_contains;
    673 	unsigned		bd_block_size;
    674 	struct hd_struct *	bd_part;
    675 	/* number of times partitions within this device have been opened. */
    676 	unsigned		bd_part_count;
    677 	int			bd_invalidated;
    678 	struct gendisk *	bd_disk;
    679 	struct list_head	bd_list;
    680 	/*
    681 	 * Private data.  You must have bd_claim'ed the block_device
    682 	 * to use this.  NOTE:  bd_claim allows an owner to claim
    683 	 * the same device multiple times, the owner must take special
    684 	 * care to not mess up bd_private for that case.
    685 	 */
    686 	unsigned long		bd_private;
    687 
    688 	/* The counter of freeze processes */
    689 	int			bd_fsfreeze_count;
    690 	/* Mutex for freeze */
    691 	struct mutex		bd_fsfreeze_mutex;
    692 };
    693 
    694 /*
    695  * Radix-tree tags, for tagging dirty and writeback pages within the pagecache
    696  * radix trees
    697  */
    698 #define PAGECACHE_TAG_DIRTY	0
    699 #define PAGECACHE_TAG_WRITEBACK	1
    700 #define PAGECACHE_TAG_TOWRITE	2
    701 
    702 int mapping_tagged(struct address_space *mapping, int tag);
    703 
    704 /*
    705  * Might pages of this file be mapped into userspace?
    706  */
    707 static inline int mapping_mapped(struct address_space *mapping)
    708 {
    709 	return	!prio_tree_empty(&mapping->i_mmap) ||
    710 		!list_empty(&mapping->i_mmap_nonlinear);
    711 }
    712 
    713 /*
    714  * Might pages of this file have been modified in userspace?
    715  * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap_pgoff
    716  * marks vma as VM_SHARED if it is shared, and the file was opened for
    717  * writing i.e. vma may be mprotected writable even if now readonly.
    718  */
    719 static inline int mapping_writably_mapped(struct address_space *mapping)
    720 {
    721 	return mapping->i_mmap_writable != 0;
    722 }
    723 
    724 /*
    725  * Use sequence counter to get consistent i_size on 32-bit processors.
    726  */
    727 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
    728 #include <linux/seqlock.h>
    729 #define __NEED_I_SIZE_ORDERED
    730 #define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
    731 #else
    732 #define i_size_ordered_init(inode) do { } while (0)
    733 #endif
    734 
    735 struct posix_acl;
    736 #define ACL_NOT_CACHED ((void *)(-1))
    737 
    738 struct inode {
    739 	/* RCU path lookup touches following: */
    740 	umode_t			i_mode;
    741 	uid_t			i_uid;
    742 	gid_t			i_gid;
    743 	const struct inode_operations	*i_op;
    744 	struct super_block	*i_sb;
    745 
    746 	spinlock_t		i_lock;	/* i_blocks, i_bytes, maybe i_size */
    747 	unsigned int		i_flags;
    748 	unsigned long		i_state;
    749 #ifdef CONFIG_SECURITY
    750 	void			*i_security;
    751 #endif
    752 	struct mutex		i_mutex;
    753 
    754 
    755 	unsigned long		dirtied_when;	/* jiffies of first dirtying */
    756 
    757 	struct hlist_node	i_hash;
    758 	struct list_head	i_wb_list;	/* backing dev IO list */
    759 	struct list_head	i_lru;		/* inode LRU list */
    760 	struct list_head	i_sb_list;
    761 	union {
    762 		struct list_head	i_dentry;
    763 		struct rcu_head		i_rcu;
    764 	};
    765 	unsigned long		i_ino;
    766 	atomic_t		i_count;
    767 	unsigned int		i_nlink;
    768 	dev_t			i_rdev;
    769 	unsigned int		i_blkbits;
    770 	u64			i_version;
    771 	loff_t			i_size;
    772 #ifdef __NEED_I_SIZE_ORDERED
    773 	seqcount_t		i_size_seqcount;
    774 #endif
    775 	struct timespec		i_atime;
    776 	struct timespec		i_mtime;
    777 	struct timespec		i_ctime;
    778 	blkcnt_t		i_blocks;
    779 	unsigned short          i_bytes;
    780 	struct rw_semaphore	i_alloc_sem;
    781 	const struct file_operations	*i_fop;	/* former ->i_op->default_file_ops */
    782 	struct file_lock	*i_flock;
    783 	struct address_space	*i_mapping;
    784 	struct address_space	i_data;
    785 #ifdef CONFIG_QUOTA
    786 	struct dquot		*i_dquot[MAXQUOTAS];
    787 #endif
    788 	struct list_head	i_devices;
    789 	union {
    790 		struct pipe_inode_info	*i_pipe;
    791 		struct block_device	*i_bdev;
    792 		struct cdev		*i_cdev;
    793 	};
    794 
    795 	__u32			i_generation;
    796 
    797 #ifdef CONFIG_FSNOTIFY
    798 	__u32			i_fsnotify_mask; /* all events this inode cares about */
    799 	struct hlist_head	i_fsnotify_marks;
    800 #endif
    801 
    802 #ifdef CONFIG_IMA
    803 	atomic_t		i_readcount; /* struct files open RO */
    804 #endif
    805 	atomic_t		i_writecount;
    806 #ifdef CONFIG_FS_POSIX_ACL
    807 	struct posix_acl	*i_acl;
    808 	struct posix_acl	*i_default_acl;
    809 #endif
    810 	void			*i_private; /* fs or device private pointer */
    811 };
    812 
    813 static inline int inode_unhashed(struct inode *inode)
    814 {
    815 	return hlist_unhashed(&inode->i_hash);
    816 }
    817 
    818 /*
    819  * inode->i_mutex nesting subclasses for the lock validator:
    820  *
    821  * 0: the object of the current VFS operation
    822  * 1: parent
    823  * 2: child/target
    824  * 3: quota file
    825  *
    826  * The locking order between these classes is
    827  * parent -> child -> normal -> xattr -> quota
    828  */
    829 enum inode_i_mutex_lock_class
    830 {
    831 	I_MUTEX_NORMAL,
    832 	I_MUTEX_PARENT,
    833 	I_MUTEX_CHILD,
    834 	I_MUTEX_XATTR,
    835 	I_MUTEX_QUOTA
    836 };
    837 
    838 /*
    839  * NOTE: in a 32bit arch with a preemptable kernel and
    840  * an UP compile the i_size_read/write must be atomic
    841  * with respect to the local cpu (unlike with preempt disabled),
    842  * but they don't need to be atomic with respect to other cpus like in
    843  * true SMP (so they need either to either locally disable irq around
    844  * the read or for example on x86 they can be still implemented as a
    845  * cmpxchg8b without the need of the lock prefix). For SMP compiles
    846  * and 64bit archs it makes no difference if preempt is enabled or not.
    847  */
    848 static inline loff_t i_size_read(const struct inode *inode)
    849 {
    850 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
    851 	loff_t i_size;
    852 	unsigned int seq;
    853 
    854 	do {
    855 		seq = read_seqcount_begin(&inode->i_size_seqcount);
    856 		i_size = inode->i_size;
    857 	} while (read_seqcount_retry(&inode->i_size_seqcount, seq));
    858 	return i_size;
    859 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
    860 	loff_t i_size;
    861 
    862 	preempt_disable();
    863 	i_size = inode->i_size;
    864 	preempt_enable();
    865 	return i_size;
    866 #else
    867 	return inode->i_size;
    868 #endif
    869 }
    870 
    871 /*
    872  * NOTE: unlike i_size_read(), i_size_write() does need locking around it
    873  * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount
    874  * can be lost, resulting in subsequent i_size_read() calls spinning forever.
    875  */
    876 static inline void i_size_write(struct inode *inode, loff_t i_size)
    877 {
    878 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
    879 	write_seqcount_begin(&inode->i_size_seqcount);
    880 	inode->i_size = i_size;
    881 	write_seqcount_end(&inode->i_size_seqcount);
    882 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
    883 	preempt_disable();
    884 	inode->i_size = i_size;
    885 	preempt_enable();
    886 #else
    887 	inode->i_size = i_size;
    888 #endif
    889 }
    890 
    891 static inline unsigned iminor(const struct inode *inode)
    892 {
    893 	return MINOR(inode->i_rdev);
    894 }
    895 
    896 static inline unsigned imajor(const struct inode *inode)
    897 {
    898 	return MAJOR(inode->i_rdev);
    899 }
    900 
    901 extern struct block_device *I_BDEV(struct inode *inode);
    902 
    903 struct fown_struct {
    904 	rwlock_t lock;          /* protects pid, uid, euid fields */
    905 	struct pid *pid;	/* pid or -pgrp where SIGIO should be sent */
    906 	enum pid_type pid_type;	/* Kind of process group SIGIO should be sent to */
    907 	uid_t uid, euid;	/* uid/euid of process setting the owner */
    908 	int signum;		/* posix.1b rt signal to be delivered on IO */
    909 };
    910 
    911 /*
    912  * Track a single file's readahead state
    913  */
    914 struct file_ra_state {
    915 	pgoff_t start;			/* where readahead started */
    916 	unsigned int size;		/* # of readahead pages */
    917 	unsigned int async_size;	/* do asynchronous readahead when
    918 					   there are only # of pages ahead */
    919 
    920 	unsigned int ra_pages;		/* Maximum readahead window */
    921 	unsigned int mmap_miss;		/* Cache miss stat for mmap accesses */
    922 	loff_t prev_pos;		/* Cache last read() position */
    923 };
    924 
    925 /*
    926  * Check if @index falls in the readahead windows.
    927  */
    928 static inline int ra_has_index(struct file_ra_state *ra, pgoff_t index)
    929 {
    930 	return (index >= ra->start &&
    931 		index <  ra->start + ra->size);
    932 }
    933 
    934 #define FILE_MNT_WRITE_TAKEN	1
    935 #define FILE_MNT_WRITE_RELEASED	2
    936 
    937 struct file {
    938 	/*
    939 	 * fu_list becomes invalid after file_free is called and queued via
    940 	 * fu_rcuhead for RCU freeing
    941 	 */
    942 	union {
    943 		struct list_head	fu_list;
    944 		struct rcu_head 	fu_rcuhead;
    945 	} f_u;
    946 	struct path		f_path;
    947 #define f_dentry	f_path.dentry
    948 #define f_vfsmnt	f_path.mnt
    949 	const struct file_operations	*f_op;
    950 	spinlock_t		f_lock;  /* f_ep_links, f_flags, no IRQ */
    951 #ifdef CONFIG_SMP
    952 	int			f_sb_list_cpu;
    953 #endif
    954 	atomic_long_t		f_count;
    955 	unsigned int 		f_flags;
    956 	fmode_t			f_mode;
    957 	loff_t			f_pos;
    958 	struct fown_struct	f_owner;
    959 	const struct cred	*f_cred;
    960 	struct file_ra_state	f_ra;
    961 
    962 	u64			f_version;
    963 #ifdef CONFIG_SECURITY
    964 	void			*f_security;
    965 #endif
    966 	/* needed for tty driver, and maybe others */
    967 	void			*private_data;
    968 
    969 #ifdef CONFIG_EPOLL
    970 	/* Used by fs/eventpoll.c to link all the hooks to this file */
    971 	struct list_head	f_ep_links;
    972 	struct list_head	f_tfile_llink;
    973 #endif /* #ifdef CONFIG_EPOLL */
    974 	struct address_space	*f_mapping;
    975 #ifdef CONFIG_DEBUG_WRITECOUNT
    976 	unsigned long f_mnt_write_state;
    977 #endif
    978 };
    979 
    980 struct file_handle {
    981 	__u32 handle_bytes;
    982 	int handle_type;
    983 	/* file identifier */
    984 	unsigned char f_handle[0];
    985 };
    986 
    987 #define get_file(x)	atomic_long_inc(&(x)->f_count)
    988 #define fput_atomic(x)	atomic_long_add_unless(&(x)->f_count, -1, 1)
    989 #define file_count(x)	atomic_long_read(&(x)->f_count)
    990 
    991 #ifdef CONFIG_DEBUG_WRITECOUNT
    992 static inline void file_take_write(struct file *f)
    993 {
    994 	WARN_ON(f->f_mnt_write_state != 0);
    995 	f->f_mnt_write_state = FILE_MNT_WRITE_TAKEN;
    996 }
    997 static inline void file_release_write(struct file *f)
    998 {
    999 	f->f_mnt_write_state |= FILE_MNT_WRITE_RELEASED;
   1000 }
   1001 static inline void file_reset_write(struct file *f)
   1002 {
   1003 	f->f_mnt_write_state = 0;
   1004 }
   1005 static inline void file_check_state(struct file *f)
   1006 {
   1007 	/*
   1008 	 * At this point, either both or neither of these bits
   1009 	 * should be set.
   1010 	 */
   1011 	WARN_ON(f->f_mnt_write_state == FILE_MNT_WRITE_TAKEN);
   1012 	WARN_ON(f->f_mnt_write_state == FILE_MNT_WRITE_RELEASED);
   1013 }
   1014 static inline int file_check_writeable(struct file *f)
   1015 {
   1016 	if (f->f_mnt_write_state == FILE_MNT_WRITE_TAKEN)
   1017 		return 0;
   1018 	printk(KERN_WARNING "writeable file with no "
   1019 			    "mnt_want_write()\n");
   1020 	WARN_ON(1);
   1021 	return -EINVAL;
   1022 }
   1023 #else /* !CONFIG_DEBUG_WRITECOUNT */
   1024 static inline void file_take_write(struct file *filp) {}
   1025 static inline void file_release_write(struct file *filp) {}
   1026 static inline void file_reset_write(struct file *filp) {}
   1027 static inline void file_check_state(struct file *filp) {}
   1028 static inline int file_check_writeable(struct file *filp)
   1029 {
   1030 	return 0;
   1031 }
   1032 #endif /* CONFIG_DEBUG_WRITECOUNT */
   1033 
   1034 #define	MAX_NON_LFS	((1UL<<31) - 1)
   1035 
   1036 /* Page cache limit. The filesystems should put that into their s_maxbytes
   1037    limits, otherwise bad things can happen in VM. */
   1038 #if BITS_PER_LONG==32
   1039 #define MAX_LFS_FILESIZE	(((u64)PAGE_CACHE_SIZE << (BITS_PER_LONG-1))-1)
   1040 #elif BITS_PER_LONG==64
   1041 #define MAX_LFS_FILESIZE 	0x7fffffffffffffffUL
   1042 #endif
   1043 
   1044 #define FL_POSIX	1
   1045 #define FL_FLOCK	2
   1046 #define FL_ACCESS	8	/* not trying to lock, just looking */
   1047 #define FL_EXISTS	16	/* when unlocking, test for existence */
   1048 #define FL_LEASE	32	/* lease held on this file */
   1049 #define FL_CLOSE	64	/* unlock on close */
   1050 #define FL_SLEEP	128	/* A blocking lock */
   1051 
   1052 /*
   1053  * Special return value from posix_lock_file() and vfs_lock_file() for
   1054  * asynchronous locking.
   1055  */
   1056 #define FILE_LOCK_DEFERRED 1
   1057 
   1058 /*
   1059  * The POSIX file lock owner is determined by
   1060  * the "struct files_struct" in the thread group
   1061  * (or NULL for no owner - BSD locks).
   1062  *
   1063  * Lockd stuffs a "host" pointer into this.
   1064  */
   1065 typedef struct files_struct *fl_owner_t;
   1066 
   1067 struct file_lock_operations {
   1068 	void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
   1069 	void (*fl_release_private)(struct file_lock *);
   1070 };
   1071 
   1072 struct lock_manager_operations {
   1073 	int (*fl_compare_owner)(struct file_lock *, struct file_lock *);
   1074 	void (*fl_notify)(struct file_lock *);	/* unblock callback */
   1075 	int (*fl_grant)(struct file_lock *, struct file_lock *, int);
   1076 	void (*fl_release_private)(struct file_lock *);
   1077 	void (*fl_break)(struct file_lock *);
   1078 	int (*fl_change)(struct file_lock **, int);
   1079 };
   1080 
   1081 struct lock_manager {
   1082 	struct list_head list;
   1083 };
   1084 
   1085 void locks_start_grace(struct lock_manager *);
   1086 void locks_end_grace(struct lock_manager *);
   1087 int locks_in_grace(void);
   1088 
   1089 /* that will die - we need it for nfs_lock_info */
   1090 #include <linux/nfs_fs_i.h>
   1091 
   1092 struct file_lock {
   1093 	struct file_lock *fl_next;	/* singly linked list for this inode  */
   1094 	struct list_head fl_link;	/* doubly linked list of all locks */
   1095 	struct list_head fl_block;	/* circular list of blocked processes */
   1096 	fl_owner_t fl_owner;
   1097 	unsigned char fl_flags;
   1098 	unsigned char fl_type;
   1099 	unsigned int fl_pid;
   1100 	struct pid *fl_nspid;
   1101 	wait_queue_head_t fl_wait;
   1102 	struct file *fl_file;
   1103 	loff_t fl_start;
   1104 	loff_t fl_end;
   1105 
   1106 	struct fasync_struct *	fl_fasync; /* for lease break notifications */
   1107 	unsigned long fl_break_time;	/* for nonblocking lease breaks */
   1108 
   1109 	const struct file_lock_operations *fl_ops;	/* Callbacks for filesystems */
   1110 	const struct lock_manager_operations *fl_lmops;	/* Callbacks for lockmanagers */
   1111 	union {
   1112 		struct nfs_lock_info	nfs_fl;
   1113 		struct nfs4_lock_info	nfs4_fl;
   1114 		struct {
   1115 			struct list_head link;	/* link in AFS vnode's pending_locks list */
   1116 			int state;		/* state of grant or error if -ve */
   1117 		} afs;
   1118 	} fl_u;
   1119 };
   1120 
   1121 /* The following constant reflects the upper bound of the file/locking space */
   1122 #ifndef OFFSET_MAX
   1123 #define INT_LIMIT(x)	(~((x)1 << (sizeof(x)*8 - 1)))
   1124 #define OFFSET_MAX	INT_LIMIT(loff_t)
   1125 #define OFFT_OFFSET_MAX	INT_LIMIT(off_t)
   1126 #endif
   1127 
   1128 #include <linux/fcntl.h>
   1129 
   1130 extern void send_sigio(struct fown_struct *fown, int fd, int band);
   1131 
   1132 #ifdef CONFIG_FILE_LOCKING
   1133 extern int fcntl_getlk(struct file *, struct flock __user *);
   1134 extern int fcntl_setlk(unsigned int, struct file *, unsigned int,
   1135 			struct flock __user *);
   1136 
   1137 #if BITS_PER_LONG == 32
   1138 extern int fcntl_getlk64(struct file *, struct flock64 __user *);
   1139 extern int fcntl_setlk64(unsigned int, struct file *, unsigned int,
   1140 			struct flock64 __user *);
   1141 #endif
   1142 
   1143 extern int fcntl_setlease(unsigned int fd, struct file *filp, long arg);
   1144 extern int fcntl_getlease(struct file *filp);
   1145 
   1146 /* fs/locks.c */
   1147 void locks_free_lock(struct file_lock *fl);
   1148 extern void locks_init_lock(struct file_lock *);
   1149 extern struct file_lock * locks_alloc_lock(void);
   1150 extern void locks_copy_lock(struct file_lock *, struct file_lock *);
   1151 extern void __locks_copy_lock(struct file_lock *, const struct file_lock *);
   1152 extern void locks_remove_posix(struct file *, fl_owner_t);
   1153 extern void locks_remove_flock(struct file *);
   1154 extern void locks_release_private(struct file_lock *);
   1155 extern void posix_test_lock(struct file *, struct file_lock *);
   1156 extern int posix_lock_file(struct file *, struct file_lock *, struct file_lock *);
   1157 extern int posix_lock_file_wait(struct file *, struct file_lock *);
   1158 extern int posix_unblock_lock(struct file *, struct file_lock *);
   1159 extern int vfs_test_lock(struct file *, struct file_lock *);
   1160 extern int vfs_lock_file(struct file *, unsigned int, struct file_lock *, struct file_lock *);
   1161 extern int vfs_cancel_lock(struct file *filp, struct file_lock *fl);
   1162 extern int flock_lock_file_wait(struct file *filp, struct file_lock *fl);
   1163 extern int __break_lease(struct inode *inode, unsigned int flags);
   1164 extern void lease_get_mtime(struct inode *, struct timespec *time);
   1165 extern int generic_setlease(struct file *, long, struct file_lock **);
   1166 extern int vfs_setlease(struct file *, long, struct file_lock **);
   1167 extern int lease_modify(struct file_lock **, int);
   1168 extern int lock_may_read(struct inode *, loff_t start, unsigned long count);
   1169 extern int lock_may_write(struct inode *, loff_t start, unsigned long count);
   1170 extern void lock_flocks(void);
   1171 extern void unlock_flocks(void);
   1172 #else /* !CONFIG_FILE_LOCKING */
   1173 static inline int fcntl_getlk(struct file *file, struct flock __user *user)
   1174 {
   1175 	return -EINVAL;
   1176 }
   1177 
   1178 static inline int fcntl_setlk(unsigned int fd, struct file *file,
   1179 			      unsigned int cmd, struct flock __user *user)
   1180 {
   1181 	return -EACCES;
   1182 }
   1183 
   1184 #if BITS_PER_LONG == 32
   1185 static inline int fcntl_getlk64(struct file *file, struct flock64 __user *user)
   1186 {
   1187 	return -EINVAL;
   1188 }
   1189 
   1190 static inline int fcntl_setlk64(unsigned int fd, struct file *file,
   1191 				unsigned int cmd, struct flock64 __user *user)
   1192 {
   1193 	return -EACCES;
   1194 }
   1195 #endif
   1196 static inline int fcntl_setlease(unsigned int fd, struct file *filp, long arg)
   1197 {
   1198 	return 0;
   1199 }
   1200 
   1201 static inline int fcntl_getlease(struct file *filp)
   1202 {
   1203 	return 0;
   1204 }
   1205 
   1206 static inline void locks_init_lock(struct file_lock *fl)
   1207 {
   1208 	return;
   1209 }
   1210 
   1211 static inline void __locks_copy_lock(struct file_lock *new, struct file_lock *fl)
   1212 {
   1213 	return;
   1214 }
   1215 
   1216 static inline void locks_copy_lock(struct file_lock *new, struct file_lock *fl)
   1217 {
   1218 	return;
   1219 }
   1220 
   1221 static inline void locks_remove_posix(struct file *filp, fl_owner_t owner)
   1222 {
   1223 	return;
   1224 }
   1225 
   1226 static inline void locks_remove_flock(struct file *filp)
   1227 {
   1228 	return;
   1229 }
   1230 
   1231 static inline void posix_test_lock(struct file *filp, struct file_lock *fl)
   1232 {
   1233 	return;
   1234 }
   1235 
   1236 static inline int posix_lock_file(struct file *filp, struct file_lock *fl,
   1237 				  struct file_lock *conflock)
   1238 {
   1239 	return -ENOLCK;
   1240 }
   1241 
   1242 static inline int posix_lock_file_wait(struct file *filp, struct file_lock *fl)
   1243 {
   1244 	return -ENOLCK;
   1245 }
   1246 
   1247 static inline int posix_unblock_lock(struct file *filp,
   1248 				     struct file_lock *waiter)
   1249 {
   1250 	return -ENOENT;
   1251 }
   1252 
   1253 static inline int vfs_test_lock(struct file *filp, struct file_lock *fl)
   1254 {
   1255 	return 0;
   1256 }
   1257 
   1258 static inline int vfs_lock_file(struct file *filp, unsigned int cmd,
   1259 				struct file_lock *fl, struct file_lock *conf)
   1260 {
   1261 	return -ENOLCK;
   1262 }
   1263 
   1264 static inline int vfs_cancel_lock(struct file *filp, struct file_lock *fl)
   1265 {
   1266 	return 0;
   1267 }
   1268 
   1269 static inline int flock_lock_file_wait(struct file *filp,
   1270 				       struct file_lock *request)
   1271 {
   1272 	return -ENOLCK;
   1273 }
   1274 
   1275 static inline int __break_lease(struct inode *inode, unsigned int mode)
   1276 {
   1277 	return 0;
   1278 }
   1279 
   1280 static inline void lease_get_mtime(struct inode *inode, struct timespec *time)
   1281 {
   1282 	return;
   1283 }
   1284 
   1285 static inline int generic_setlease(struct file *filp, long arg,
   1286 				    struct file_lock **flp)
   1287 {
   1288 	return -EINVAL;
   1289 }
   1290 
   1291 static inline int vfs_setlease(struct file *filp, long arg,
   1292 			       struct file_lock **lease)
   1293 {
   1294 	return -EINVAL;
   1295 }
   1296 
   1297 static inline int lease_modify(struct file_lock **before, int arg)
   1298 {
   1299 	return -EINVAL;
   1300 }
   1301 
   1302 static inline int lock_may_read(struct inode *inode, loff_t start,
   1303 				unsigned long len)
   1304 {
   1305 	return 1;
   1306 }
   1307 
   1308 static inline int lock_may_write(struct inode *inode, loff_t start,
   1309 				 unsigned long len)
   1310 {
   1311 	return 1;
   1312 }
   1313 
   1314 static inline void lock_flocks(void)
   1315 {
   1316 }
   1317 
   1318 static inline void unlock_flocks(void)
   1319 {
   1320 }
   1321 
   1322 #endif /* !CONFIG_FILE_LOCKING */
   1323 
   1324 
   1325 struct fasync_struct {
   1326 	spinlock_t		fa_lock;
   1327 	int			magic;
   1328 	int			fa_fd;
   1329 	struct fasync_struct	*fa_next; /* singly linked list */
   1330 	struct file		*fa_file;
   1331 	struct rcu_head		fa_rcu;
   1332 };
   1333 
   1334 #define FASYNC_MAGIC 0x4601
   1335 
   1336 /* SMP safe fasync helpers: */
   1337 extern int fasync_helper(int, struct file *, int, struct fasync_struct **);
   1338 extern struct fasync_struct *fasync_insert_entry(int, struct file *, struct fasync_struct **, struct fasync_struct *);
   1339 extern int fasync_remove_entry(struct file *, struct fasync_struct **);
   1340 extern struct fasync_struct *fasync_alloc(void);
   1341 extern void fasync_free(struct fasync_struct *);
   1342 
   1343 /* can be called from interrupts */
   1344 extern void kill_fasync(struct fasync_struct **, int, int);
   1345 
   1346 extern int __f_setown(struct file *filp, struct pid *, enum pid_type, int force);
   1347 extern int f_setown(struct file *filp, unsigned long arg, int force);
   1348 extern void f_delown(struct file *filp);
   1349 extern pid_t f_getown(struct file *filp);
   1350 extern int send_sigurg(struct fown_struct *fown);
   1351 
   1352 /*
   1353  *	Umount options
   1354  */
   1355 
   1356 #define MNT_FORCE	0x00000001	/* Attempt to forcibily umount */
   1357 #define MNT_DETACH	0x00000002	/* Just detach from the tree */
   1358 #define MNT_EXPIRE	0x00000004	/* Mark for expiry */
   1359 #define UMOUNT_NOFOLLOW	0x00000008	/* Don't follow symlink on umount */
   1360 #define UMOUNT_UNUSED	0x80000000	/* Flag guaranteed to be unused */
   1361 
   1362 extern struct list_head super_blocks;
   1363 extern spinlock_t sb_lock;
   1364 
   1365 struct super_block {
   1366 	struct list_head	s_list;		/* Keep this first */
   1367 	dev_t			s_dev;		/* search index; _not_ kdev_t */
   1368 	unsigned char		s_dirt;
   1369 	unsigned char		s_blocksize_bits;
   1370 	unsigned long		s_blocksize;
   1371 	loff_t			s_maxbytes;	/* Max file size */
   1372 	struct file_system_type	*s_type;
   1373 	const struct super_operations	*s_op;
   1374 	const struct dquot_operations	*dq_op;
   1375 	const struct quotactl_ops	*s_qcop;
   1376 	const struct export_operations *s_export_op;
   1377 	unsigned long		s_flags;
   1378 	unsigned long		s_magic;
   1379 	struct dentry		*s_root;
   1380 	struct rw_semaphore	s_umount;
   1381 	struct mutex		s_lock;
   1382 	int			s_count;
   1383 	atomic_t		s_active;
   1384 #ifdef CONFIG_SECURITY
   1385 	void                    *s_security;
   1386 #endif
   1387 	const struct xattr_handler **s_xattr;
   1388 
   1389 	struct list_head	s_inodes;	/* all inodes */
   1390 	struct hlist_bl_head	s_anon;		/* anonymous dentries for (nfs) exporting */
   1391 #ifdef CONFIG_SMP
   1392 	struct list_head __percpu *s_files;
   1393 #else
   1394 	struct list_head	s_files;
   1395 #endif
   1396 	/* s_dentry_lru, s_nr_dentry_unused protected by dcache.c lru locks */
   1397 	struct list_head	s_dentry_lru;	/* unused dentry lru */
   1398 	int			s_nr_dentry_unused;	/* # of dentry on lru */
   1399 
   1400 	struct block_device	*s_bdev;
   1401 	struct backing_dev_info *s_bdi;
   1402 	struct mtd_info		*s_mtd;
   1403 	struct list_head	s_instances;
   1404 	struct quota_info	s_dquot;	/* Diskquota specific options */
   1405 
   1406 	int			s_frozen;
   1407 	wait_queue_head_t	s_wait_unfrozen;
   1408 
   1409 	char s_id[32];				/* Informational name */
   1410 	u8 s_uuid[16];				/* UUID */
   1411 
   1412 	void 			*s_fs_info;	/* Filesystem private info */
   1413 	fmode_t			s_mode;
   1414 
   1415 	/* Granularity of c/m/atime in ns.
   1416 	   Cannot be worse than a second */
   1417 	u32		   s_time_gran;
   1418 
   1419 	/*
   1420 	 * The next field is for VFS *only*. No filesystems have any business
   1421 	 * even looking at it. You had been warned.
   1422 	 */
   1423 	struct mutex s_vfs_rename_mutex;	/* Kludge */
   1424 
   1425 	/*
   1426 	 * Filesystem subtype.  If non-empty the filesystem type field
   1427 	 * in /proc/mounts will be "type.subtype"
   1428 	 */
   1429 	char *s_subtype;
   1430 
   1431 	/*
   1432 	 * Saved mount options for lazy filesystems using
   1433 	 * generic_show_options()
   1434 	 */
   1435 	char __rcu *s_options;
   1436 	const struct dentry_operations *s_d_op; /* default d_op for dentries */
   1437 
   1438 	/*
   1439 	 * Saved pool identifier for cleancache (-1 means none)
   1440 	 */
   1441 	int cleancache_poolid;
   1442 };
   1443 
   1444 extern struct timespec current_fs_time(struct super_block *sb);
   1445 
   1446 /*
   1447  * Snapshotting support.
   1448  */
   1449 enum {
   1450 	SB_UNFROZEN = 0,
   1451 	SB_FREEZE_WRITE	= 1,
   1452 	SB_FREEZE_TRANS = 2,
   1453 };
   1454 
   1455 #define vfs_check_frozen(sb, level) \
   1456 	wait_event((sb)->s_wait_unfrozen, ((sb)->s_frozen < (level)))
   1457 
   1458 #define get_fs_excl() atomic_inc(&current->fs_excl)
   1459 #define put_fs_excl() atomic_dec(&current->fs_excl)
   1460 #define has_fs_excl() atomic_read(&current->fs_excl)
   1461 
   1462 /*
   1463  * until VFS tracks user namespaces for inodes, just make all files
   1464  * belong to init_user_ns
   1465  */
   1466 extern struct user_namespace init_user_ns;
   1467 #define inode_userns(inode) (&init_user_ns)
   1468 extern bool inode_owner_or_capable(const struct inode *inode);
   1469 
   1470 /* not quite ready to be deprecated, but... */
   1471 extern void lock_super(struct super_block *);
   1472 extern void unlock_super(struct super_block *);
   1473 
   1474 /*
   1475  * VFS helper functions..
   1476  */
   1477 extern int vfs_create(struct inode *, struct dentry *, int, struct nameidata *);
   1478 extern int vfs_mkdir(struct inode *, struct dentry *, int);
   1479 extern int vfs_mknod(struct inode *, struct dentry *, int, dev_t);
   1480 extern int vfs_symlink(struct inode *, struct dentry *, const char *);
   1481 extern int vfs_link(struct dentry *, struct inode *, struct dentry *);
   1482 extern int vfs_rmdir(struct inode *, struct dentry *);
   1483 extern int vfs_unlink(struct inode *, struct dentry *);
   1484 extern int vfs_rename(struct inode *, struct dentry *, struct inode *, struct dentry *);
   1485 
   1486 /*
   1487  * VFS dentry helper functions.
   1488  */
   1489 extern void dentry_unhash(struct dentry *dentry);
   1490 
   1491 /*
   1492  * VFS file helper functions.
   1493  */
   1494 extern int file_permission(struct file *, int);
   1495 extern void inode_init_owner(struct inode *inode, const struct inode *dir,
   1496 			mode_t mode);
   1497 /*
   1498  * VFS FS_IOC_FIEMAP helper definitions.
   1499  */
   1500 struct fiemap_extent_info {
   1501 	unsigned int fi_flags;		/* Flags as passed from user */
   1502 	unsigned int fi_extents_mapped;	/* Number of mapped extents */
   1503 	unsigned int fi_extents_max;	/* Size of fiemap_extent array */
   1504 	struct fiemap_extent __user *fi_extents_start; /* Start of
   1505 							fiemap_extent array */
   1506 };
   1507 int fiemap_fill_next_extent(struct fiemap_extent_info *info, u64 logical,
   1508 			    u64 phys, u64 len, u32 flags);
   1509 int fiemap_check_flags(struct fiemap_extent_info *fieinfo, u32 fs_flags);
   1510 
   1511 /*
   1512  * File types
   1513  *
   1514  * NOTE! These match bits 12..15 of stat.st_mode
   1515  * (ie "(i_mode >> 12) & 15").
   1516  */
   1517 #define DT_UNKNOWN	0
   1518 #define DT_FIFO		1
   1519 #define DT_CHR		2
   1520 #define DT_DIR		4
   1521 #define DT_BLK		6
   1522 #define DT_REG		8
   1523 #define DT_LNK		10
   1524 #define DT_SOCK		12
   1525 #define DT_WHT		14
   1526 
   1527 /*
   1528  * This is the "filldir" function type, used by readdir() to let
   1529  * the kernel specify what kind of dirent layout it wants to have.
   1530  * This allows the kernel to read directories into kernel space or
   1531  * to have different dirent layouts depending on the binary type.
   1532  */
   1533 typedef int (*filldir_t)(void *, const char *, int, loff_t, u64, unsigned);
   1534 struct block_device_operations;
   1535 
   1536 /* These macros are for out of kernel modules to test that
   1537  * the kernel supports the unlocked_ioctl and compat_ioctl
   1538  * fields in struct file_operations. */
   1539 #define HAVE_COMPAT_IOCTL 1
   1540 #define HAVE_UNLOCKED_IOCTL 1
   1541 
   1542 /*
   1543  * NOTE:
   1544  * all file operations except setlease can be called without
   1545  * the big kernel lock held in all filesystems.
   1546  */
   1547 struct file_operations {
   1548 	struct module *owner;
   1549 	loff_t (*llseek) (struct file *, loff_t, int);
   1550 	ssize_t (*read) (struct file *, char __user *, size_t, loff_t *);
   1551 	ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *);
   1552 	ssize_t (*aio_read) (struct kiocb *, const struct iovec *, unsigned long, loff_t);
   1553 	ssize_t (*aio_write) (struct kiocb *, const struct iovec *, unsigned long, loff_t);
   1554 	int (*readdir) (struct file *, void *, filldir_t);
   1555 	unsigned int (*poll) (struct file *, struct poll_table_struct *);
   1556 	long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long);
   1557 	long (*compat_ioctl) (struct file *, unsigned int, unsigned long);
   1558 	int (*mmap) (struct file *, struct vm_area_struct *);
   1559 	int (*open) (struct inode *, struct file *);
   1560 	int (*flush) (struct file *, fl_owner_t id);
   1561 	int (*release) (struct inode *, struct file *);
   1562 	int (*fsync) (struct file *, int datasync);
   1563 	int (*aio_fsync) (struct kiocb *, int datasync);
   1564 	int (*fasync) (int, struct file *, int);
   1565 	int (*lock) (struct file *, int, struct file_lock *);
   1566 	ssize_t (*sendpage) (struct file *, struct page *, int, size_t, loff_t *, int);
   1567 	unsigned long (*get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
   1568 	int (*check_flags)(int);
   1569 	int (*flock) (struct file *, int, struct file_lock *);
   1570 	ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, loff_t *, size_t, unsigned int);
   1571 	ssize_t (*splice_read)(struct file *, loff_t *, struct pipe_inode_info *, size_t, unsigned int);
   1572 	int (*setlease)(struct file *, long, struct file_lock **);
   1573 	long (*fallocate)(struct file *file, int mode, loff_t offset,
   1574 			  loff_t len);
   1575 };
   1576 
   1577 #define IPERM_FLAG_RCU	0x0001
   1578 
   1579 struct inode_operations {
   1580 	struct dentry * (*lookup) (struct inode *,struct dentry *, struct nameidata *);
   1581 	void * (*follow_link) (struct dentry *, struct nameidata *);
   1582 	int (*permission) (struct inode *, int, unsigned int);
   1583 	int (*check_acl)(struct inode *, int, unsigned int);
   1584 
   1585 	int (*readlink) (struct dentry *, char __user *,int);
   1586 	void (*put_link) (struct dentry *, struct nameidata *, void *);
   1587 
   1588 	int (*create) (struct inode *,struct dentry *,int, struct nameidata *);
   1589 	int (*link) (struct dentry *,struct inode *,struct dentry *);
   1590 	int (*unlink) (struct inode *,struct dentry *);
   1591 	int (*symlink) (struct inode *,struct dentry *,const char *);
   1592 	int (*mkdir) (struct inode *,struct dentry *,int);
   1593 	int (*rmdir) (struct inode *,struct dentry *);
   1594 	int (*mknod) (struct inode *,struct dentry *,int,dev_t);
   1595 	int (*rename) (struct inode *, struct dentry *,
   1596 			struct inode *, struct dentry *);
   1597 	void (*truncate) (struct inode *);
   1598 	int (*setattr) (struct dentry *, struct iattr *);
   1599 	int (*getattr) (struct vfsmount *mnt, struct dentry *, struct kstat *);
   1600 	int (*setxattr) (struct dentry *, const char *,const void *,size_t,int);
   1601 	ssize_t (*getxattr) (struct dentry *, const char *, void *, size_t);
   1602 	ssize_t (*listxattr) (struct dentry *, char *, size_t);
   1603 	int (*removexattr) (struct dentry *, const char *);
   1604 	void (*truncate_range)(struct inode *, loff_t, loff_t);
   1605 	int (*fiemap)(struct inode *, struct fiemap_extent_info *, u64 start,
   1606 		      u64 len);
   1607 } ____cacheline_aligned;
   1608 
   1609 struct seq_file;
   1610 
   1611 ssize_t rw_copy_check_uvector(int type, const struct iovec __user * uvector,
   1612 				unsigned long nr_segs, unsigned long fast_segs,
   1613 				struct iovec *fast_pointer,
   1614 				struct iovec **ret_pointer);
   1615 
   1616 extern ssize_t vfs_read(struct file *, char __user *, size_t, loff_t *);
   1617 extern ssize_t vfs_write(struct file *, const char __user *, size_t, loff_t *);
   1618 extern ssize_t vfs_readv(struct file *, const struct iovec __user *,
   1619 		unsigned long, loff_t *);
   1620 extern ssize_t vfs_writev(struct file *, const struct iovec __user *,
   1621 		unsigned long, loff_t *);
   1622 
   1623 struct super_operations {
   1624    	struct inode *(*alloc_inode)(struct super_block *sb);
   1625 	void (*destroy_inode)(struct inode *);
   1626 
   1627    	void (*dirty_inode) (struct inode *, int flags);
   1628 	int (*write_inode) (struct inode *, struct writeback_control *wbc);
   1629 	int (*drop_inode) (struct inode *);
   1630 	void (*evict_inode) (struct inode *);
   1631 	void (*put_super) (struct super_block *);
   1632 	void (*write_super) (struct super_block *);
   1633 	int (*sync_fs)(struct super_block *sb, int wait);
   1634 	int (*freeze_fs) (struct super_block *);
   1635 	int (*unfreeze_fs) (struct super_block *);
   1636 	int (*statfs) (struct dentry *, struct kstatfs *);
   1637 	int (*remount_fs) (struct super_block *, int *, char *);
   1638 	void (*umount_begin) (struct super_block *);
   1639 
   1640 	int (*show_options)(struct seq_file *, struct vfsmount *);
   1641 	int (*show_devname)(struct seq_file *, struct vfsmount *);
   1642 	int (*show_path)(struct seq_file *, struct vfsmount *);
   1643 	int (*show_stats)(struct seq_file *, struct vfsmount *);
   1644 #ifdef CONFIG_QUOTA
   1645 	ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t);
   1646 	ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t);
   1647 #endif
   1648 	int (*bdev_try_to_free_page)(struct super_block*, struct page*, gfp_t);
   1649 };
   1650 
   1651 /*
   1652  * Inode state bits.  Protected by inode->i_lock
   1653  *
   1654  * Three bits determine the dirty state of the inode, I_DIRTY_SYNC,
   1655  * I_DIRTY_DATASYNC and I_DIRTY_PAGES.
   1656  *
   1657  * Four bits define the lifetime of an inode.  Initially, inodes are I_NEW,
   1658  * until that flag is cleared.  I_WILL_FREE, I_FREEING and I_CLEAR are set at
   1659  * various stages of removing an inode.
   1660  *
   1661  * Two bits are used for locking and completion notification, I_NEW and I_SYNC.
   1662  *
   1663  * I_DIRTY_SYNC		Inode is dirty, but doesn't have to be written on
   1664  *			fdatasync().  i_atime is the usual cause.
   1665  * I_DIRTY_DATASYNC	Data-related inode changes pending. We keep track of
   1666  *			these changes separately from I_DIRTY_SYNC so that we
   1667  *			don't have to write inode on fdatasync() when only
   1668  *			mtime has changed in it.
   1669  * I_DIRTY_PAGES	Inode has dirty pages.  Inode itself may be clean.
   1670  * I_NEW		Serves as both a mutex and completion notification.
   1671  *			New inodes set I_NEW.  If two processes both create
   1672  *			the same inode, one of them will release its inode and
   1673  *			wait for I_NEW to be released before returning.
   1674  *			Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can
   1675  *			also cause waiting on I_NEW, without I_NEW actually
   1676  *			being set.  find_inode() uses this to prevent returning
   1677  *			nearly-dead inodes.
   1678  * I_WILL_FREE		Must be set when calling write_inode_now() if i_count
   1679  *			is zero.  I_FREEING must be set when I_WILL_FREE is
   1680  *			cleared.
   1681  * I_FREEING		Set when inode is about to be freed but still has dirty
   1682  *			pages or buffers attached or the inode itself is still
   1683  *			dirty.
   1684  * I_CLEAR		Added by end_writeback().  In this state the inode is clean
   1685  *			and can be destroyed.  Inode keeps I_FREEING.
   1686  *
   1687  *			Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are
   1688  *			prohibited for many purposes.  iget() must wait for
   1689  *			the inode to be completely released, then create it
   1690  *			anew.  Other functions will just ignore such inodes,
   1691  *			if appropriate.  I_NEW is used for waiting.
   1692  *
   1693  * I_SYNC		Synchonized write of dirty inode data.  The bits is
   1694  *			set during data writeback, and cleared with a wakeup
   1695  *			on the bit address once it is done.
   1696  *
   1697  * Q: What is the difference between I_WILL_FREE and I_FREEING?
   1698  */
   1699 #define I_DIRTY_SYNC		(1 << 0)
   1700 #define I_DIRTY_DATASYNC	(1 << 1)
   1701 #define I_DIRTY_PAGES		(1 << 2)
   1702 #define __I_NEW			3
   1703 #define I_NEW			(1 << __I_NEW)
   1704 #define I_WILL_FREE		(1 << 4)
   1705 #define I_FREEING		(1 << 5)
   1706 #define I_CLEAR			(1 << 6)
   1707 #define __I_SYNC		7
   1708 #define I_SYNC			(1 << __I_SYNC)
   1709 #define I_REFERENCED		(1 << 8)
   1710 
   1711 #define I_DIRTY (I_DIRTY_SYNC | I_DIRTY_DATASYNC | I_DIRTY_PAGES)
   1712 
   1713 extern void __mark_inode_dirty(struct inode *, int);
   1714 static inline void mark_inode_dirty(struct inode *inode)
   1715 {
   1716 	__mark_inode_dirty(inode, I_DIRTY);
   1717 }
   1718 
   1719 static inline void mark_inode_dirty_sync(struct inode *inode)
   1720 {
   1721 	__mark_inode_dirty(inode, I_DIRTY_SYNC);
   1722 }
   1723 
   1724 /**
   1725  * inc_nlink - directly increment an inode's link count
   1726  * @inode: inode
   1727  *
   1728  * This is a low-level filesystem helper to replace any
   1729  * direct filesystem manipulation of i_nlink.  Currently,
   1730  * it is only here for parity with dec_nlink().
   1731  */
   1732 static inline void inc_nlink(struct inode *inode)
   1733 {
   1734 	inode->i_nlink++;
   1735 }
   1736 
   1737 static inline void inode_inc_link_count(struct inode *inode)
   1738 {
   1739 	inc_nlink(inode);
   1740 	mark_inode_dirty(inode);
   1741 }
   1742 
   1743 /**
   1744  * drop_nlink - directly drop an inode's link count
   1745  * @inode: inode
   1746  *
   1747  * This is a low-level filesystem helper to replace any
   1748  * direct filesystem manipulation of i_nlink.  In cases
   1749  * where we are attempting to track writes to the
   1750  * filesystem, a decrement to zero means an imminent
   1751  * write when the file is truncated and actually unlinked
   1752  * on the filesystem.
   1753  */
   1754 static inline void drop_nlink(struct inode *inode)
   1755 {
   1756 	inode->i_nlink--;
   1757 }
   1758 
   1759 /**
   1760  * clear_nlink - directly zero an inode's link count
   1761  * @inode: inode
   1762  *
   1763  * This is a low-level filesystem helper to replace any
   1764  * direct filesystem manipulation of i_nlink.  See
   1765  * drop_nlink() for why we care about i_nlink hitting zero.
   1766  */
   1767 static inline void clear_nlink(struct inode *inode)
   1768 {
   1769 	inode->i_nlink = 0;
   1770 }
   1771 
   1772 static inline void inode_dec_link_count(struct inode *inode)
   1773 {
   1774 	drop_nlink(inode);
   1775 	mark_inode_dirty(inode);
   1776 }
   1777 
   1778 /**
   1779  * inode_inc_iversion - increments i_version
   1780  * @inode: inode that need to be updated
   1781  *
   1782  * Every time the inode is modified, the i_version field will be incremented.
   1783  * The filesystem has to be mounted with i_version flag
   1784  */
   1785 
   1786 static inline void inode_inc_iversion(struct inode *inode)
   1787 {
   1788        spin_lock(&inode->i_lock);
   1789        inode->i_version++;
   1790        spin_unlock(&inode->i_lock);
   1791 }
   1792 
   1793 extern void touch_atime(struct vfsmount *mnt, struct dentry *dentry);
   1794 static inline void file_accessed(struct file *file)
   1795 {
   1796 	if (!(file->f_flags & O_NOATIME))
   1797 		touch_atime(file->f_path.mnt, file->f_path.dentry);
   1798 }
   1799 
   1800 int sync_inode(struct inode *inode, struct writeback_control *wbc);
   1801 int sync_inode_metadata(struct inode *inode, int wait);
   1802 
   1803 struct file_system_type {
   1804 	const char *name;
   1805 	int fs_flags;
   1806 	struct dentry *(*mount) (struct file_system_type *, int,
   1807 		       const char *, void *);
   1808 	void (*kill_sb) (struct super_block *);
   1809 	struct module *owner;
   1810 	struct file_system_type * next;
   1811 	struct list_head fs_supers;
   1812 
   1813 	struct lock_class_key s_lock_key;
   1814 	struct lock_class_key s_umount_key;
   1815 	struct lock_class_key s_vfs_rename_key;
   1816 
   1817 	struct lock_class_key i_lock_key;
   1818 	struct lock_class_key i_mutex_key;
   1819 	struct lock_class_key i_mutex_dir_key;
   1820 	struct lock_class_key i_alloc_sem_key;
   1821 };
   1822 
   1823 extern struct dentry *mount_ns(struct file_system_type *fs_type, int flags,
   1824 	void *data, int (*fill_super)(struct super_block *, void *, int));
   1825 extern struct dentry *mount_bdev(struct file_system_type *fs_type,
   1826 	int flags, const char *dev_name, void *data,
   1827 	int (*fill_super)(struct super_block *, void *, int));
   1828 extern struct dentry *mount_single(struct file_system_type *fs_type,
   1829 	int flags, void *data,
   1830 	int (*fill_super)(struct super_block *, void *, int));
   1831 extern struct dentry *mount_nodev(struct file_system_type *fs_type,
   1832 	int flags, void *data,
   1833 	int (*fill_super)(struct super_block *, void *, int));
   1834 void generic_shutdown_super(struct super_block *sb);
   1835 void kill_block_super(struct super_block *sb);
   1836 void kill_anon_super(struct super_block *sb);
   1837 void kill_litter_super(struct super_block *sb);
   1838 void deactivate_super(struct super_block *sb);
   1839 void deactivate_locked_super(struct super_block *sb);
   1840 int set_anon_super(struct super_block *s, void *data);
   1841 struct super_block *sget(struct file_system_type *type,
   1842 			int (*test)(struct super_block *,void *),
   1843 			int (*set)(struct super_block *,void *),
   1844 			void *data);
   1845 extern struct dentry *mount_pseudo(struct file_system_type *, char *,
   1846 	const struct super_operations *ops,
   1847 	const struct dentry_operations *dops,
   1848 	unsigned long);
   1849 
   1850 static inline void sb_mark_dirty(struct super_block *sb)
   1851 {
   1852 	sb->s_dirt = 1;
   1853 }
   1854 static inline void sb_mark_clean(struct super_block *sb)
   1855 {
   1856 	sb->s_dirt = 0;
   1857 }
   1858 static inline int sb_is_dirty(struct super_block *sb)
   1859 {
   1860 	return sb->s_dirt;
   1861 }
   1862 
   1863 /* Alas, no aliases. Too much hassle with bringing module.h everywhere */
   1864 #define fops_get(fops) \
   1865 	(((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
   1866 #define fops_put(fops) \
   1867 	do { if (fops) module_put((fops)->owner); } while(0)
   1868 
   1869 extern int register_filesystem(struct file_system_type *);
   1870 extern int unregister_filesystem(struct file_system_type *);
   1871 extern struct vfsmount *kern_mount_data(struct file_system_type *, void *data);
   1872 #define kern_mount(type) kern_mount_data(type, NULL)
   1873 extern int may_umount_tree(struct vfsmount *);
   1874 extern int may_umount(struct vfsmount *);
   1875 extern long do_mount(char *, char *, char *, unsigned long, void *);
   1876 extern struct vfsmount *collect_mounts(struct path *);
   1877 extern void drop_collected_mounts(struct vfsmount *);
   1878 extern int iterate_mounts(int (*)(struct vfsmount *, void *), void *,
   1879 			  struct vfsmount *);
   1880 extern int vfs_statfs(struct path *, struct kstatfs *);
   1881 extern int user_statfs(const char __user *, struct kstatfs *);
   1882 extern int fd_statfs(int, struct kstatfs *);
   1883 extern int statfs_by_dentry(struct dentry *, struct kstatfs *);
   1884 extern int freeze_super(struct super_block *super);
   1885 extern int thaw_super(struct super_block *super);
   1886 extern bool our_mnt(struct vfsmount *mnt);
   1887 
   1888 extern int current_umask(void);
   1889 
   1890 /* /sys/fs */
   1891 extern struct kobject *fs_kobj;
   1892 
   1893 #define MAX_RW_COUNT (INT_MAX & PAGE_CACHE_MASK)
   1894 extern int rw_verify_area(int, struct file *, loff_t *, size_t);
   1895 
   1896 #define FLOCK_VERIFY_READ  1
   1897 #define FLOCK_VERIFY_WRITE 2
   1898 
   1899 #ifdef CONFIG_FILE_LOCKING
   1900 extern int locks_mandatory_locked(struct inode *);
   1901 extern int locks_mandatory_area(int, struct inode *, struct file *, loff_t, size_t);
   1902 
   1903 /*
   1904  * Candidates for mandatory locking have the setgid bit set
   1905  * but no group execute bit -  an otherwise meaningless combination.
   1906  */
   1907 
   1908 static inline int __mandatory_lock(struct inode *ino)
   1909 {
   1910 	return (ino->i_mode & (S_ISGID | S_IXGRP)) == S_ISGID;
   1911 }
   1912 
   1913 /*
   1914  * ... and these candidates should be on MS_MANDLOCK mounted fs,
   1915  * otherwise these will be advisory locks
   1916  */
   1917 
   1918 static inline int mandatory_lock(struct inode *ino)
   1919 {
   1920 	return IS_MANDLOCK(ino) && __mandatory_lock(ino);
   1921 }
   1922 
   1923 static inline int locks_verify_locked(struct inode *inode)
   1924 {
   1925 	if (mandatory_lock(inode))
   1926 		return locks_mandatory_locked(inode);
   1927 	return 0;
   1928 }
   1929 
   1930 static inline int locks_verify_truncate(struct inode *inode,
   1931 				    struct file *filp,
   1932 				    loff_t size)
   1933 {
   1934 	if (inode->i_flock && mandatory_lock(inode))
   1935 		return locks_mandatory_area(
   1936 			FLOCK_VERIFY_WRITE, inode, filp,
   1937 			size < inode->i_size ? size : inode->i_size,
   1938 			(size < inode->i_size ? inode->i_size - size
   1939 			 : size - inode->i_size)
   1940 		);
   1941 	return 0;
   1942 }
   1943 
   1944 static inline int break_lease(struct inode *inode, unsigned int mode)
   1945 {
   1946 	if (inode->i_flock)
   1947 		return __break_lease(inode, mode);
   1948 	return 0;
   1949 }
   1950 #else /* !CONFIG_FILE_LOCKING */
   1951 static inline int locks_mandatory_locked(struct inode *inode)
   1952 {
   1953 	return 0;
   1954 }
   1955 
   1956 static inline int locks_mandatory_area(int rw, struct inode *inode,
   1957 				       struct file *filp, loff_t offset,
   1958 				       size_t count)
   1959 {
   1960 	return 0;
   1961 }
   1962 
   1963 static inline int __mandatory_lock(struct inode *inode)
   1964 {
   1965 	return 0;
   1966 }
   1967 
   1968 static inline int mandatory_lock(struct inode *inode)
   1969 {
   1970 	return 0;
   1971 }
   1972 
   1973 static inline int locks_verify_locked(struct inode *inode)
   1974 {
   1975 	return 0;
   1976 }
   1977 
   1978 static inline int locks_verify_truncate(struct inode *inode, struct file *filp,
   1979 					size_t size)
   1980 {
   1981 	return 0;
   1982 }
   1983 
   1984 static inline int break_lease(struct inode *inode, unsigned int mode)
   1985 {
   1986 	return 0;
   1987 }
   1988 
   1989 #endif /* CONFIG_FILE_LOCKING */
   1990 
   1991 /* fs/open.c */
   1992 
   1993 extern int do_truncate(struct dentry *, loff_t start, unsigned int time_attrs,
   1994 		       struct file *filp);
   1995 extern int do_fallocate(struct file *file, int mode, loff_t offset,
   1996 			loff_t len);
   1997 extern long do_sys_open(int dfd, const char __user *filename, int flags,
   1998 			int mode);
   1999 extern struct file *filp_open(const char *, int, int);
   2000 extern struct file *file_open_root(struct dentry *, struct vfsmount *,
   2001 				   const char *, int);
   2002 extern struct file * dentry_open(struct dentry *, struct vfsmount *, int,
   2003 				 const struct cred *);
   2004 extern int filp_close(struct file *, fl_owner_t id);
   2005 extern char * getname(const char __user *);
   2006 
   2007 /* fs/ioctl.c */
   2008 
   2009 extern int ioctl_preallocate(struct file *filp, void __user *argp);
   2010 
   2011 /* fs/dcache.c */
   2012 extern void __init vfs_caches_init_early(void);
   2013 extern void __init vfs_caches_init(unsigned long);
   2014 
   2015 extern struct kmem_cache *names_cachep;
   2016 
   2017 #define __getname_gfp(gfp)	kmem_cache_alloc(names_cachep, (gfp))
   2018 #define __getname()		__getname_gfp(GFP_KERNEL)
   2019 #define __putname(name)		kmem_cache_free(names_cachep, (void *)(name))
   2020 #ifndef CONFIG_AUDITSYSCALL
   2021 #define putname(name)   __putname(name)
   2022 #else
   2023 extern void putname(const char *name);
   2024 #endif
   2025 
   2026 #ifdef CONFIG_BLOCK
   2027 extern int register_blkdev(unsigned int, const char *);
   2028 extern void unregister_blkdev(unsigned int, const char *);
   2029 extern struct block_device *bdget(dev_t);
   2030 extern struct block_device *bdgrab(struct block_device *bdev);
   2031 extern void bd_set_size(struct block_device *, loff_t size);
   2032 extern void bd_forget(struct inode *inode);
   2033 extern void bdput(struct block_device *);
   2034 extern void invalidate_bdev(struct block_device *);
   2035 extern int sync_blockdev(struct block_device *bdev);
   2036 extern struct super_block *freeze_bdev(struct block_device *);
   2037 extern void emergency_thaw_all(void);
   2038 extern int thaw_bdev(struct block_device *bdev, struct super_block *sb);
   2039 extern int fsync_bdev(struct block_device *);
   2040 #else
   2041 static inline void bd_forget(struct inode *inode) {}
   2042 static inline int sync_blockdev(struct block_device *bdev) { return 0; }
   2043 static inline void invalidate_bdev(struct block_device *bdev) {}
   2044 
   2045 static inline struct super_block *freeze_bdev(struct block_device *sb)
   2046 {
   2047 	return NULL;
   2048 }
   2049 
   2050 static inline int thaw_bdev(struct block_device *bdev, struct super_block *sb)
   2051 {
   2052 	return 0;
   2053 }
   2054 #endif
   2055 extern int sync_filesystem(struct super_block *);
   2056 extern const struct file_operations def_blk_fops;
   2057 extern const struct file_operations def_chr_fops;
   2058 extern const struct file_operations bad_sock_fops;
   2059 extern const struct file_operations def_fifo_fops;
   2060 #ifdef CONFIG_BLOCK
   2061 extern int ioctl_by_bdev(struct block_device *, unsigned, unsigned long);
   2062 extern int blkdev_ioctl(struct block_device *, fmode_t, unsigned, unsigned long);
   2063 extern long compat_blkdev_ioctl(struct file *, unsigned, unsigned long);
   2064 extern int blkdev_get(struct block_device *bdev, fmode_t mode, void *holder);
   2065 extern struct block_device *blkdev_get_by_path(const char *path, fmode_t mode,
   2066 					       void *holder);
   2067 extern struct block_device *blkdev_get_by_dev(dev_t dev, fmode_t mode,
   2068 					      void *holder);
   2069 extern int blkdev_put(struct block_device *bdev, fmode_t mode);
   2070 #ifdef CONFIG_SYSFS
   2071 extern int bd_link_disk_holder(struct block_device *bdev, struct gendisk *disk);
   2072 extern void bd_unlink_disk_holder(struct block_device *bdev,
   2073 				  struct gendisk *disk);
   2074 #else
   2075 static inline int bd_link_disk_holder(struct block_device *bdev,
   2076 				      struct gendisk *disk)
   2077 {
   2078 	return 0;
   2079 }
   2080 static inline void bd_unlink_disk_holder(struct block_device *bdev,
   2081 					 struct gendisk *disk)
   2082 {
   2083 }
   2084 #endif
   2085 #endif
   2086 
   2087 /* fs/char_dev.c */
   2088 #define CHRDEV_MAJOR_HASH_SIZE	255
   2089 extern int alloc_chrdev_region(dev_t *, unsigned, unsigned, const char *);
   2090 extern int register_chrdev_region(dev_t, unsigned, const char *);
   2091 extern int __register_chrdev(unsigned int major, unsigned int baseminor,
   2092 			     unsigned int count, const char *name,
   2093 			     const struct file_operations *fops);
   2094 extern void __unregister_chrdev(unsigned int major, unsigned int baseminor,
   2095 				unsigned int count, const char *name);
   2096 extern void unregister_chrdev_region(dev_t, unsigned);
   2097 extern void chrdev_show(struct seq_file *,off_t);
   2098 
   2099 static inline int register_chrdev(unsigned int major, const char *name,
   2100 				  const struct file_operations *fops)
   2101 {
   2102 	return __register_chrdev(major, 0, 256, name, fops);
   2103 }
   2104 
   2105 static inline void unregister_chrdev(unsigned int major, const char *name)
   2106 {
   2107 	__unregister_chrdev(major, 0, 256, name);
   2108 }
   2109 
   2110 /* fs/block_dev.c */
   2111 #define BDEVNAME_SIZE	32	/* Largest string for a blockdev identifier */
   2112 #define BDEVT_SIZE	10	/* Largest string for MAJ:MIN for blkdev */
   2113 
   2114 #ifdef CONFIG_BLOCK
   2115 #define BLKDEV_MAJOR_HASH_SIZE	255
   2116 extern const char *__bdevname(dev_t, char *buffer);
   2117 extern const char *bdevname(struct block_device *bdev, char *buffer);
   2118 extern struct block_device *lookup_bdev(const char *);
   2119 extern void blkdev_show(struct seq_file *,off_t);
   2120 
   2121 #else
   2122 #define BLKDEV_MAJOR_HASH_SIZE	0
   2123 #endif
   2124 
   2125 extern void init_special_inode(struct inode *, umode_t, dev_t);
   2126 
   2127 /* Invalid inode operations -- fs/bad_inode.c */
   2128 extern void make_bad_inode(struct inode *);
   2129 extern int is_bad_inode(struct inode *);
   2130 
   2131 extern const struct file_operations read_pipefifo_fops;
   2132 extern const struct file_operations write_pipefifo_fops;
   2133 extern const struct file_operations rdwr_pipefifo_fops;
   2134 
   2135 extern int fs_may_remount_ro(struct super_block *);
   2136 
   2137 #ifdef CONFIG_BLOCK
   2138 /*
   2139  * return READ, READA, or WRITE
   2140  */
   2141 #define bio_rw(bio)		((bio)->bi_rw & (RW_MASK | RWA_MASK))
   2142 
   2143 /*
   2144  * return data direction, READ or WRITE
   2145  */
   2146 #define bio_data_dir(bio)	((bio)->bi_rw & 1)
   2147 
   2148 extern void check_disk_size_change(struct gendisk *disk,
   2149 				   struct block_device *bdev);
   2150 extern int revalidate_disk(struct gendisk *);
   2151 extern int check_disk_change(struct block_device *);
   2152 extern int __invalidate_device(struct block_device *, bool);
   2153 extern int invalidate_partition(struct gendisk *, int);
   2154 #endif
   2155 unsigned long invalidate_mapping_pages(struct address_space *mapping,
   2156 					pgoff_t start, pgoff_t end);
   2157 
   2158 static inline void invalidate_remote_inode(struct inode *inode)
   2159 {
   2160 	if (S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) ||
   2161 	    S_ISLNK(inode->i_mode))
   2162 		invalidate_mapping_pages(inode->i_mapping, 0, -1);
   2163 }
   2164 extern int invalidate_inode_pages2(struct address_space *mapping);
   2165 extern int invalidate_inode_pages2_range(struct address_space *mapping,
   2166 					 pgoff_t start, pgoff_t end);
   2167 extern int write_inode_now(struct inode *, int);
   2168 extern int filemap_fdatawrite(struct address_space *);
   2169 extern int filemap_flush(struct address_space *);
   2170 extern int filemap_fdatawait(struct address_space *);
   2171 extern int filemap_fdatawait_range(struct address_space *, loff_t lstart,
   2172 				   loff_t lend);
   2173 extern int filemap_write_and_wait(struct address_space *mapping);
   2174 extern int filemap_write_and_wait_range(struct address_space *mapping,
   2175 				        loff_t lstart, loff_t lend);
   2176 extern int __filemap_fdatawrite_range(struct address_space *mapping,
   2177 				loff_t start, loff_t end, int sync_mode);
   2178 extern int filemap_fdatawrite_range(struct address_space *mapping,
   2179 				loff_t start, loff_t end);
   2180 
   2181 extern int vfs_fsync_range(struct file *file, loff_t start, loff_t end,
   2182 			   int datasync);
   2183 extern int vfs_fsync(struct file *file, int datasync);
   2184 extern int generic_write_sync(struct file *file, loff_t pos, loff_t count);
   2185 extern void sync_supers(void);
   2186 extern void emergency_sync(void);
   2187 extern void emergency_remount(void);
   2188 #ifdef CONFIG_BLOCK
   2189 extern sector_t bmap(struct inode *, sector_t);
   2190 #endif
   2191 extern int notify_change(struct dentry *, struct iattr *);
   2192 extern int inode_permission(struct inode *, int);
   2193 extern int generic_permission(struct inode *, int, unsigned int,
   2194 		int (*check_acl)(struct inode *, int, unsigned int));
   2195 
   2196 static inline bool execute_ok(struct inode *inode)
   2197 {
   2198 	return (inode->i_mode & S_IXUGO) || S_ISDIR(inode->i_mode);
   2199 }
   2200 
   2201 extern int get_write_access(struct inode *);
   2202 extern int deny_write_access(struct file *);
   2203 static inline void put_write_access(struct inode * inode)
   2204 {
   2205 	atomic_dec(&inode->i_writecount);
   2206 }
   2207 static inline void allow_write_access(struct file *file)
   2208 {
   2209 	if (file)
   2210 		atomic_inc(&file->f_path.dentry->d_inode->i_writecount);
   2211 }
   2212 #ifdef CONFIG_IMA
   2213 static inline void i_readcount_dec(struct inode *inode)
   2214 {
   2215 	BUG_ON(!atomic_read(&inode->i_readcount));
   2216 	atomic_dec(&inode->i_readcount);
   2217 }
   2218 static inline void i_readcount_inc(struct inode *inode)
   2219 {
   2220 	atomic_inc(&inode->i_readcount);
   2221 }
   2222 #else
   2223 static inline void i_readcount_dec(struct inode *inode)
   2224 {
   2225 	return;
   2226 }
   2227 static inline void i_readcount_inc(struct inode *inode)
   2228 {
   2229 	return;
   2230 }
   2231 #endif
   2232 extern int do_pipe_flags(int *, int);
   2233 extern struct file *create_read_pipe(struct file *f, int flags);
   2234 extern struct file *create_write_pipe(int flags);
   2235 extern void free_write_pipe(struct file *);
   2236 
   2237 extern int kernel_read(struct file *, loff_t, char *, unsigned long);
   2238 extern struct file * open_exec(const char *);
   2239 
   2240 /* fs/dcache.c -- generic fs support functions */
   2241 extern int is_subdir(struct dentry *, struct dentry *);
   2242 extern int path_is_under(struct path *, struct path *);
   2243 extern ino_t find_inode_number(struct dentry *, struct qstr *);
   2244 
   2245 #include <linux/err.h>
   2246 
   2247 /* needed for stackable file system support */
   2248 extern loff_t default_llseek(struct file *file, loff_t offset, int origin);
   2249 
   2250 extern loff_t vfs_llseek(struct file *file, loff_t offset, int origin);
   2251 
   2252 extern int inode_init_always(struct super_block *, struct inode *);
   2253 extern void inode_init_once(struct inode *);
   2254 extern void address_space_init_once(struct address_space *mapping);
   2255 extern void ihold(struct inode * inode);
   2256 extern void iput(struct inode *);
   2257 extern struct inode * igrab(struct inode *);
   2258 extern ino_t iunique(struct super_block *, ino_t);
   2259 extern int inode_needs_sync(struct inode *inode);
   2260 extern int generic_delete_inode(struct inode *inode);
   2261 extern int generic_drop_inode(struct inode *inode);
   2262 
   2263 extern struct inode *ilookup5_nowait(struct super_block *sb,
   2264 		unsigned long hashval, int (*test)(struct inode *, void *),
   2265 		void *data);
   2266 extern struct inode *ilookup5(struct super_block *sb, unsigned long hashval,
   2267 		int (*test)(struct inode *, void *), void *data);
   2268 extern struct inode *ilookup(struct super_block *sb, unsigned long ino);
   2269 
   2270 extern struct inode * iget5_locked(struct super_block *, unsigned long, int (*test)(struct inode *, void *), int (*set)(struct inode *, void *), void *);
   2271 extern struct inode * iget_locked(struct super_block *, unsigned long);
   2272 extern int insert_inode_locked4(struct inode *, unsigned long, int (*test)(struct inode *, void *), void *);
   2273 extern int insert_inode_locked(struct inode *);
   2274 extern void unlock_new_inode(struct inode *);
   2275 extern unsigned int get_next_ino(void);
   2276 
   2277 extern void __iget(struct inode * inode);
   2278 extern void iget_failed(struct inode *);
   2279 extern void end_writeback(struct inode *);
   2280 extern void __destroy_inode(struct inode *);
   2281 extern struct inode *new_inode(struct super_block *);
   2282 extern void free_inode_nonrcu(struct inode *inode);
   2283 extern int should_remove_suid(struct dentry *);
   2284 extern int file_remove_suid(struct file *);
   2285 
   2286 extern void __insert_inode_hash(struct inode *, unsigned long hashval);
   2287 extern void remove_inode_hash(struct inode *);
   2288 static inline void insert_inode_hash(struct inode *inode)
   2289 {
   2290 	__insert_inode_hash(inode, inode->i_ino);
   2291 }
   2292 extern void inode_sb_list_add(struct inode *inode);
   2293 
   2294 #ifdef CONFIG_BLOCK
   2295 extern void submit_bio(int, struct bio *);
   2296 extern int bdev_read_only(struct block_device *);
   2297 #endif
   2298 extern int set_blocksize(struct block_device *, int);
   2299 extern int sb_set_blocksize(struct super_block *, int);
   2300 extern int sb_min_blocksize(struct super_block *, int);
   2301 
   2302 extern int generic_file_mmap(struct file *, struct vm_area_struct *);
   2303 extern int generic_file_readonly_mmap(struct file *, struct vm_area_struct *);
   2304 extern int file_read_actor(read_descriptor_t * desc, struct page *page, unsigned long offset, unsigned long size);
   2305 int generic_write_checks(struct file *file, loff_t *pos, size_t *count, int isblk);
   2306 extern ssize_t generic_file_aio_read(struct kiocb *, const struct iovec *, unsigned long, loff_t);
   2307 extern ssize_t __generic_file_aio_write(struct kiocb *, const struct iovec *, unsigned long,
   2308 		loff_t *);
   2309 extern ssize_t generic_file_aio_write(struct kiocb *, const struct iovec *, unsigned long, loff_t);
   2310 extern ssize_t generic_file_direct_write(struct kiocb *, const struct iovec *,
   2311 		unsigned long *, loff_t, loff_t *, size_t, size_t);
   2312 extern ssize_t generic_file_buffered_write(struct kiocb *, const struct iovec *,
   2313 		unsigned long, loff_t, loff_t *, size_t, ssize_t);
   2314 extern ssize_t do_sync_read(struct file *filp, char __user *buf, size_t len, loff_t *ppos);
   2315 extern ssize_t do_sync_write(struct file *filp, const char __user *buf, size_t len, loff_t *ppos);
   2316 extern int generic_segment_checks(const struct iovec *iov,
   2317 		unsigned long *nr_segs, size_t *count, int access_flags);
   2318 
   2319 /* fs/block_dev.c */
   2320 extern ssize_t blkdev_aio_write(struct kiocb *iocb, const struct iovec *iov,
   2321 				unsigned long nr_segs, loff_t pos);
   2322 extern int blkdev_fsync(struct file *filp, int datasync);
   2323 
   2324 /* fs/splice.c */
   2325 extern ssize_t generic_file_splice_read(struct file *, loff_t *,
   2326 		struct pipe_inode_info *, size_t, unsigned int);
   2327 extern ssize_t default_file_splice_read(struct file *, loff_t *,
   2328 		struct pipe_inode_info *, size_t, unsigned int);
   2329 extern ssize_t generic_file_splice_write(struct pipe_inode_info *,
   2330 		struct file *, loff_t *, size_t, unsigned int);
   2331 extern ssize_t generic_splice_sendpage(struct pipe_inode_info *pipe,
   2332 		struct file *out, loff_t *, size_t len, unsigned int flags);
   2333 extern long do_splice_direct(struct file *in, loff_t *ppos, struct file *out,
   2334 		size_t len, unsigned int flags);
   2335 
   2336 extern void
   2337 file_ra_state_init(struct file_ra_state *ra, struct address_space *mapping);
   2338 extern loff_t noop_llseek(struct file *file, loff_t offset, int origin);
   2339 extern loff_t no_llseek(struct file *file, loff_t offset, int origin);
   2340 extern loff_t generic_file_llseek(struct file *file, loff_t offset, int origin);
   2341 extern loff_t generic_file_llseek_unlocked(struct file *file, loff_t offset,
   2342 			int origin);
   2343 extern int generic_file_open(struct inode * inode, struct file * filp);
   2344 extern int nonseekable_open(struct inode * inode, struct file * filp);
   2345 
   2346 #ifdef CONFIG_FS_XIP
   2347 extern ssize_t xip_file_read(struct file *filp, char __user *buf, size_t len,
   2348 			     loff_t *ppos);
   2349 extern int xip_file_mmap(struct file * file, struct vm_area_struct * vma);
   2350 extern ssize_t xip_file_write(struct file *filp, const char __user *buf,
   2351 			      size_t len, loff_t *ppos);
   2352 extern int xip_truncate_page(struct address_space *mapping, loff_t from);
   2353 #else
   2354 static inline int xip_truncate_page(struct address_space *mapping, loff_t from)
   2355 {
   2356 	return 0;
   2357 }
   2358 #endif
   2359 
   2360 #ifdef CONFIG_BLOCK
   2361 typedef void (dio_submit_t)(int rw, struct bio *bio, struct inode *inode,
   2362 			    loff_t file_offset);
   2363 
   2364 enum {
   2365 	/* need locking between buffered and direct access */
   2366 	DIO_LOCKING	= 0x01,
   2367 
   2368 	/* filesystem does not support filling holes */
   2369 	DIO_SKIP_HOLES	= 0x02,
   2370 };
   2371 
   2372 void dio_end_io(struct bio *bio, int error);
   2373 
   2374 ssize_t __blockdev_direct_IO(int rw, struct kiocb *iocb, struct inode *inode,
   2375 	struct block_device *bdev, const struct iovec *iov, loff_t offset,
   2376 	unsigned long nr_segs, get_block_t get_block, dio_iodone_t end_io,
   2377 	dio_submit_t submit_io,	int flags);
   2378 
   2379 static inline ssize_t blockdev_direct_IO(int rw, struct kiocb *iocb,
   2380 	struct inode *inode, struct block_device *bdev, const struct iovec *iov,
   2381 	loff_t offset, unsigned long nr_segs, get_block_t get_block,
   2382 	dio_iodone_t end_io)
   2383 {
   2384 	return __blockdev_direct_IO(rw, iocb, inode, bdev, iov, offset,
   2385 				    nr_segs, get_block, end_io, NULL,
   2386 				    DIO_LOCKING | DIO_SKIP_HOLES);
   2387 }
   2388 #endif
   2389 
   2390 extern const struct file_operations generic_ro_fops;
   2391 
   2392 #define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
   2393 
   2394 extern int vfs_readlink(struct dentry *, char __user *, int, const char *);
   2395 extern int vfs_follow_link(struct nameidata *, const char *);
   2396 extern int page_readlink(struct dentry *, char __user *, int);
   2397 extern void *page_follow_link_light(struct dentry *, struct nameidata *);
   2398 extern void page_put_link(struct dentry *, struct nameidata *, void *);
   2399 extern int __page_symlink(struct inode *inode, const char *symname, int len,
   2400 		int nofs);
   2401 extern int page_symlink(struct inode *inode, const char *symname, int len);
   2402 extern const struct inode_operations page_symlink_inode_operations;
   2403 extern int generic_readlink(struct dentry *, char __user *, int);
   2404 extern void generic_fillattr(struct inode *, struct kstat *);
   2405 extern int vfs_getattr(struct vfsmount *, struct dentry *, struct kstat *);
   2406 void __inode_add_bytes(struct inode *inode, loff_t bytes);
   2407 void inode_add_bytes(struct inode *inode, loff_t bytes);
   2408 void inode_sub_bytes(struct inode *inode, loff_t bytes);
   2409 loff_t inode_get_bytes(struct inode *inode);
   2410 void inode_set_bytes(struct inode *inode, loff_t bytes);
   2411 
   2412 extern int vfs_readdir(struct file *, filldir_t, void *);
   2413 
   2414 extern int vfs_stat(const char __user *, struct kstat *);
   2415 extern int vfs_lstat(const char __user *, struct kstat *);
   2416 extern int vfs_fstat(unsigned int, struct kstat *);
   2417 extern int vfs_fstatat(int , const char __user *, struct kstat *, int);
   2418 
   2419 extern int do_vfs_ioctl(struct file *filp, unsigned int fd, unsigned int cmd,
   2420 		    unsigned long arg);
   2421 extern int __generic_block_fiemap(struct inode *inode,
   2422 				  struct fiemap_extent_info *fieinfo,
   2423 				  loff_t start, loff_t len,
   2424 				  get_block_t *get_block);
   2425 extern int generic_block_fiemap(struct inode *inode,
   2426 				struct fiemap_extent_info *fieinfo, u64 start,
   2427 				u64 len, get_block_t *get_block);
   2428 
   2429 extern void get_filesystem(struct file_system_type *fs);
   2430 extern void put_filesystem(struct file_system_type *fs);
   2431 extern struct file_system_type *get_fs_type(const char *name);
   2432 extern struct super_block *get_super(struct block_device *);
   2433 extern struct super_block *get_active_super(struct block_device *bdev);
   2434 extern struct super_block *user_get_super(dev_t);
   2435 extern void drop_super(struct super_block *sb);
   2436 extern void iterate_supers(void (*)(struct super_block *, void *), void *);
   2437 
   2438 extern int dcache_dir_open(struct inode *, struct file *);
   2439 extern int dcache_dir_close(struct inode *, struct file *);
   2440 extern loff_t dcache_dir_lseek(struct file *, loff_t, int);
   2441 extern int dcache_readdir(struct file *, void *, filldir_t);
   2442 extern int simple_setattr(struct dentry *, struct iattr *);
   2443 extern int simple_getattr(struct vfsmount *, struct dentry *, struct kstat *);
   2444 extern int simple_statfs(struct dentry *, struct kstatfs *);
   2445 extern int simple_link(struct dentry *, struct inode *, struct dentry *);
   2446 extern int simple_unlink(struct inode *, struct dentry *);
   2447 extern int simple_rmdir(struct inode *, struct dentry *);
   2448 extern int simple_rename(struct inode *, struct dentry *, struct inode *, struct dentry *);
   2449 extern int noop_fsync(struct file *, int);
   2450 extern int simple_empty(struct dentry *);
   2451 extern int simple_readpage(struct file *file, struct page *page);
   2452 extern int simple_write_begin(struct file *file, struct address_space *mapping,
   2453 			loff_t pos, unsigned len, unsigned flags,
   2454 			struct page **pagep, void **fsdata);
   2455 extern int simple_write_end(struct file *file, struct address_space *mapping,
   2456 			loff_t pos, unsigned len, unsigned copied,
   2457 			struct page *page, void *fsdata);
   2458 
   2459 extern struct dentry *simple_lookup(struct inode *, struct dentry *, struct nameidata *);
   2460 extern ssize_t generic_read_dir(struct file *, char __user *, size_t, loff_t *);
   2461 extern const struct file_operations simple_dir_operations;
   2462 extern const struct inode_operations simple_dir_inode_operations;
   2463 struct tree_descr { char *name; const struct file_operations *ops; int mode; };
   2464 struct dentry *d_alloc_name(struct dentry *, const char *);
   2465 extern int simple_fill_super(struct super_block *, unsigned long, struct tree_descr *);
   2466 extern int simple_pin_fs(struct file_system_type *, struct vfsmount **mount, int *count);
   2467 extern void simple_release_fs(struct vfsmount **mount, int *count);
   2468 
   2469 extern ssize_t simple_read_from_buffer(void __user *to, size_t count,
   2470 			loff_t *ppos, const void *from, size_t available);
   2471 extern ssize_t simple_write_to_buffer(void *to, size_t available, loff_t *ppos,
   2472 		const void __user *from, size_t count);
   2473 
   2474 extern int generic_file_fsync(struct file *, int);
   2475 
   2476 extern int generic_check_addressable(unsigned, u64);
   2477 
   2478 #ifdef CONFIG_MIGRATION
   2479 extern int buffer_migrate_page(struct address_space *,
   2480 				struct page *, struct page *);
   2481 #else
   2482 #define buffer_migrate_page NULL
   2483 #endif
   2484 
   2485 extern int inode_change_ok(const struct inode *, struct iattr *);
   2486 extern int inode_newsize_ok(const struct inode *, loff_t offset);
   2487 extern void setattr_copy(struct inode *inode, const struct iattr *attr);
   2488 
   2489 extern void file_update_time(struct file *file);
   2490 
   2491 extern int generic_show_options(struct seq_file *m, struct vfsmount *mnt);
   2492 extern void save_mount_options(struct super_block *sb, char *options);
   2493 extern void replace_mount_options(struct super_block *sb, char *options);
   2494 
   2495 static inline ino_t parent_ino(struct dentry *dentry)
   2496 {
   2497 	ino_t res;
   2498 
   2499 	/*
   2500 	 * Don't strictly need d_lock here? If the parent ino could change
   2501 	 * then surely we'd have a deeper race in the caller?
   2502 	 */
   2503 	spin_lock(&dentry->d_lock);
   2504 	res = dentry->d_parent->d_inode->i_ino;
   2505 	spin_unlock(&dentry->d_lock);
   2506 	return res;
   2507 }
   2508 
   2509 /* Transaction based IO helpers */
   2510 
   2511 /*
   2512  * An argresp is stored in an allocated page and holds the
   2513  * size of the argument or response, along with its content
   2514  */
   2515 struct simple_transaction_argresp {
   2516 	ssize_t size;
   2517 	char data[0];
   2518 };
   2519 
   2520 #define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
   2521 
   2522 char *simple_transaction_get(struct file *file, const char __user *buf,
   2523 				size_t size);
   2524 ssize_t simple_transaction_read(struct file *file, char __user *buf,
   2525 				size_t size, loff_t *pos);
   2526 int simple_transaction_release(struct inode *inode, struct file *file);
   2527 
   2528 void simple_transaction_set(struct file *file, size_t n);
   2529 
   2530 /*
   2531  * simple attribute files
   2532  *
   2533  * These attributes behave similar to those in sysfs:
   2534  *
   2535  * Writing to an attribute immediately sets a value, an open file can be
   2536  * written to multiple times.
   2537  *
   2538  * Reading from an attribute creates a buffer from the value that might get
   2539  * read with multiple read calls. When the attribute has been read
   2540  * completely, no further read calls are possible until the file is opened
   2541  * again.
   2542  *
   2543  * All attributes contain a text representation of a numeric value
   2544  * that are accessed with the get() and set() functions.
   2545  */
   2546 #define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt)		\
   2547 static int __fops ## _open(struct inode *inode, struct file *file)	\
   2548 {									\
   2549 	__simple_attr_check_format(__fmt, 0ull);			\
   2550 	return simple_attr_open(inode, file, __get, __set, __fmt);	\
   2551 }									\
   2552 static const struct file_operations __fops = {				\
   2553 	.owner	 = THIS_MODULE,						\
   2554 	.open	 = __fops ## _open,					\
   2555 	.release = simple_attr_release,					\
   2556 	.read	 = simple_attr_read,					\
   2557 	.write	 = simple_attr_write,					\
   2558 	.llseek	 = generic_file_llseek,					\
   2559 };
   2560 
   2561 static inline void __attribute__((format(printf, 1, 2)))
   2562 __simple_attr_check_format(const char *fmt, ...)
   2563 {
   2564 	/* don't do anything, just let the compiler check the arguments; */
   2565 }
   2566 
   2567 int simple_attr_open(struct inode *inode, struct file *file,
   2568 		     int (*get)(void *, u64 *), int (*set)(void *, u64),
   2569 		     const char *fmt);
   2570 int simple_attr_release(struct inode *inode, struct file *file);
   2571 ssize_t simple_attr_read(struct file *file, char __user *buf,
   2572 			 size_t len, loff_t *ppos);
   2573 ssize_t simple_attr_write(struct file *file, const char __user *buf,
   2574 			  size_t len, loff_t *ppos);
   2575 
   2576 struct ctl_table;
   2577 int proc_nr_files(struct ctl_table *table, int write,
   2578 		  void __user *buffer, size_t *lenp, loff_t *ppos);
   2579 int proc_nr_dentry(struct ctl_table *table, int write,
   2580 		  void __user *buffer, size_t *lenp, loff_t *ppos);
   2581 int proc_nr_inodes(struct ctl_table *table, int write,
   2582 		   void __user *buffer, size_t *lenp, loff_t *ppos);
   2583 int __init get_filesystem_list(char *buf);
   2584 
   2585 #define __FMODE_EXEC		((__force int) FMODE_EXEC)
   2586 #define __FMODE_NONOTIFY	((__force int) FMODE_NONOTIFY)
   2587 
   2588 #define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE])
   2589 #define OPEN_FMODE(flag) ((__force fmode_t)(((flag + 1) & O_ACCMODE) | \
   2590 					    (flag & __FMODE_NONOTIFY)))
   2591 
   2592 static inline int is_sxid(mode_t mode)
   2593 {
   2594 	return (mode & S_ISUID) || ((mode & S_ISGID) && (mode & S_IXGRP));
   2595 }
   2596 
   2597 static inline void inode_has_no_xattr(struct inode *inode)
   2598 {
   2599 	if (!is_sxid(inode->i_mode) && (inode->i_sb->s_flags & MS_NOSEC))
   2600 		inode->i_flags |= S_NOSEC;
   2601 }
   2602 
   2603 #endif /* __KERNEL__ */
   2604 #endif /* _LINUX_FS_H */
   2605