blob: 35ec87e490b1a41ff0bc3ba20b06ac9d958f972a [file] [log] [blame]
Linus Torvalds1da177e2005-04-16 15:20:36 -07001#ifndef _LINUX_FS_H
2#define _LINUX_FS_H
3
Linus Torvalds1da177e2005-04-16 15:20:36 -07004
5#include <linux/linkage.h>
6#include <linux/wait.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -07007#include <linux/kdev_t.h>
8#include <linux/dcache.h>
Al Viro3f8206d2008-07-26 03:46:43 -04009#include <linux/path.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070010#include <linux/stat.h>
11#include <linux/cache.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070012#include <linux/list.h>
Dave Chinnerbc3b14c2013-08-28 10:17:58 +100013#include <linux/list_lru.h>
Oleg Nesterov4f5e65a2013-07-08 14:24:16 -070014#include <linux/llist.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070015#include <linux/radix-tree.h>
Michel Lespinasse6b2dbba2012-10-08 16:31:25 -070016#include <linux/rbtree.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070017#include <linux/init.h>
Al Viro914e2632006-10-18 13:55:46 -040018#include <linux/pid.h>
Paul Gortmaker187f1882011-11-23 20:12:59 -050019#include <linux/bug.h>
Jes Sorensen1b1dcc12006-01-09 15:59:24 -080020#include <linux/mutex.h>
Davidlohr Buesoc8c06ef2014-12-12 16:54:24 -080021#include <linux/rwsem.h>
Satyam Sharma3bd858a2007-07-17 15:00:08 +053022#include <linux/capability.h>
Matthew Wilcox6188e102008-04-18 22:21:05 -040023#include <linux/semaphore.h>
Mark Fashehc4b929b2008-10-08 19:44:18 -040024#include <linux/fiemap.h>
Nick Pigginceb5bdc2011-01-07 17:50:05 +110025#include <linux/rculist_bl.h>
Al Viro07b8ce12011-06-20 10:52:57 -040026#include <linux/atomic.h>
Konstantin Khlebnikov83aeead2011-12-08 14:33:54 -080027#include <linux/shrinker.h>
Stephen Rothwellc1aab022012-01-24 11:41:32 +110028#include <linux/migrate_mode.h>
Eric W. Biederman92361632012-02-08 07:07:50 -080029#include <linux/uidgid.h>
Jan Kara5accdf82012-06-12 16:20:34 +020030#include <linux/lockdep.h>
Jens Axboec2b1ad82012-09-27 09:35:03 +020031#include <linux/percpu-rwsem.h>
David Howells08cce05c2012-10-13 10:45:06 +010032#include <linux/blk_types.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070033
Linus Torvalds1da177e2005-04-16 15:20:36 -070034#include <asm/byteorder.h>
David Howells607ca462012-10-13 10:46:48 +010035#include <uapi/linux/fs.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070036
Christoph Hellwigb83ae6d2015-01-14 10:42:37 +010037struct backing_dev_info;
Christoph Hellwiga5694252007-07-17 04:04:28 -070038struct export_operations;
Christoph Hellwiga885c8c2006-01-08 01:02:50 -080039struct hd_geometry;
Linus Torvalds1da177e2005-04-16 15:20:36 -070040struct iovec;
41struct nameidata;
Christoph Hellwig92198f72005-06-23 22:00:59 -070042struct kiocb;
Alexey Dobriyan57cc7212011-01-10 08:18:25 +020043struct kobject;
Linus Torvalds1da177e2005-04-16 15:20:36 -070044struct pipe_inode_info;
45struct poll_table_struct;
46struct kstatfs;
47struct vm_area_struct;
48struct vfsmount;
David Howells745ca242008-11-14 10:39:22 +110049struct cred;
Mel Gormana509bc12012-07-31 16:44:57 -070050struct swap_info_struct;
Cyrill Gorcunov55985dd2012-12-17 16:04:55 -080051struct seq_file;
Christoph Hellwig7b7a8662013-09-04 15:04:39 +020052struct workqueue_struct;
Al Viro3b93f912014-02-11 21:34:08 -050053struct iov_iter;
Matthew Wilcoxf7ca90b2015-02-16 15:59:02 -080054struct vm_fault;
Linus Torvalds1da177e2005-04-16 15:20:36 -070055
Denis Cheng74bf17c2007-10-16 23:26:30 -070056extern void __init inode_init(void);
Linus Torvalds1da177e2005-04-16 15:20:36 -070057extern void __init inode_init_early(void);
Linus Torvalds1da177e2005-04-16 15:20:36 -070058extern void __init files_init(unsigned long);
59
Jan Engelhardtdded4f42008-12-01 14:34:50 -080060extern struct files_stat_struct files_stat;
Eric Dumazet518de9b2010-10-26 14:22:44 -070061extern unsigned long get_max_files(void);
Jan Engelhardtdded4f42008-12-01 14:34:50 -080062extern int sysctl_nr_open;
63extern struct inodes_stat_t inodes_stat;
64extern int leases_enable, lease_break_time;
Kees Cook800179c2012-07-25 17:29:07 -070065extern int sysctl_protected_symlinks;
66extern int sysctl_protected_hardlinks;
Jan Engelhardtdded4f42008-12-01 14:34:50 -080067
Linus Torvalds1da177e2005-04-16 15:20:36 -070068struct buffer_head;
69typedef int (get_block_t)(struct inode *inode, sector_t iblock,
70 struct buffer_head *bh_result, int create);
Christoph Hellwig92198f72005-06-23 22:00:59 -070071typedef void (dio_iodone_t)(struct kiocb *iocb, loff_t offset,
Christoph Hellwig7b7a8662013-09-04 15:04:39 +020072 ssize_t bytes, void *private);
Linus Torvalds1da177e2005-04-16 15:20:36 -070073
David Howellsbbc10962012-10-15 16:40:35 +010074#define MAY_EXEC 0x00000001
75#define MAY_WRITE 0x00000002
76#define MAY_READ 0x00000004
77#define MAY_APPEND 0x00000008
78#define MAY_ACCESS 0x00000010
79#define MAY_OPEN 0x00000020
80#define MAY_CHDIR 0x00000040
81/* called from RCU mode, don't block */
82#define MAY_NOT_BLOCK 0x00000080
83
84/*
85 * flags in file.f_mode. Note that FMODE_READ and FMODE_WRITE must correspond
86 * to O_WRONLY and O_RDWR via the strange trick in __dentry_open()
87 */
88
89/* file is open for reading */
90#define FMODE_READ ((__force fmode_t)0x1)
91/* file is open for writing */
92#define FMODE_WRITE ((__force fmode_t)0x2)
93/* file is seekable */
94#define FMODE_LSEEK ((__force fmode_t)0x4)
95/* file can be accessed using pread */
96#define FMODE_PREAD ((__force fmode_t)0x8)
97/* file can be accessed using pwrite */
98#define FMODE_PWRITE ((__force fmode_t)0x10)
99/* File is opened for execution with sys_execve / sys_uselib */
100#define FMODE_EXEC ((__force fmode_t)0x20)
101/* File is opened with O_NDELAY (only set for block devices) */
102#define FMODE_NDELAY ((__force fmode_t)0x40)
103/* File is opened with O_EXCL (only set for block devices) */
104#define FMODE_EXCL ((__force fmode_t)0x80)
105/* File is opened using open(.., 3, ..) and is writeable only for ioctls
106 (specialy hack for floppy.c) */
107#define FMODE_WRITE_IOCTL ((__force fmode_t)0x100)
108/* 32bit hashes as llseek() offset (for directories) */
109#define FMODE_32BITHASH ((__force fmode_t)0x200)
110/* 64bit hashes as llseek() offset (for directories) */
111#define FMODE_64BITHASH ((__force fmode_t)0x400)
112
113/*
114 * Don't update ctime and mtime.
115 *
116 * Currently a special hack for the XFS open_by_handle ioctl, but we'll
117 * hopefully graduate it to a proper O_CMTIME flag supported by open(2) soon.
118 */
119#define FMODE_NOCMTIME ((__force fmode_t)0x800)
120
121/* Expect random access pattern */
122#define FMODE_RANDOM ((__force fmode_t)0x1000)
123
124/* File is huge (eg. /dev/kmem): treat loff_t as unsigned */
125#define FMODE_UNSIGNED_OFFSET ((__force fmode_t)0x2000)
126
127/* File is opened with O_PATH; almost nothing can be done with it */
128#define FMODE_PATH ((__force fmode_t)0x4000)
129
Linus Torvalds9c225f22014-03-03 09:36:58 -0800130/* File needs atomic accesses to f_pos */
131#define FMODE_ATOMIC_POS ((__force fmode_t)0x8000)
Al Viro83f936c2014-03-14 12:02:47 -0400132/* Write access to underlying fs */
133#define FMODE_WRITER ((__force fmode_t)0x10000)
Al Viro7f7f25e2014-02-11 17:49:24 -0500134/* Has read method(s) */
135#define FMODE_CAN_READ ((__force fmode_t)0x20000)
136/* Has write method(s) */
137#define FMODE_CAN_WRITE ((__force fmode_t)0x40000)
Linus Torvalds9c225f22014-03-03 09:36:58 -0800138
David Howellsbbc10962012-10-15 16:40:35 +0100139/* File was opened by fanotify and shouldn't generate fanotify events */
David Drysdale75069f22015-01-08 14:32:29 -0800140#define FMODE_NONOTIFY ((__force fmode_t)0x4000000)
David Howellsbbc10962012-10-15 16:40:35 +0100141
142/*
143 * Flag for rw_copy_check_uvector and compat_rw_copy_check_uvector
144 * that indicates that they should check the contents of the iovec are
145 * valid, but not check the memory that the iovec elements
146 * points too.
147 */
148#define CHECK_IOVEC_ONLY -1
149
Linus Torvalds1da177e2005-04-16 15:20:36 -0700150/*
David Howells08cce05c2012-10-13 10:45:06 +0100151 * The below are the various read and write types that we support. Some of
152 * them include behavioral modifiers that send information down to the
153 * block layer and IO scheduler. Terminology:
154 *
155 * The block layer uses device plugging to defer IO a little bit, in
156 * the hope that we will see more IO very shortly. This increases
157 * coalescing of adjacent IO and thus reduces the number of IOs we
158 * have to send to the device. It also allows for better queuing,
159 * if the IO isn't mergeable. If the caller is going to be waiting
160 * for the IO, then he must ensure that the device is unplugged so
161 * that the IO is dispatched to the driver.
162 *
163 * All IO is handled async in Linux. This is fine for background
164 * writes, but for reads or writes that someone waits for completion
165 * on, we want to notify the block layer and IO scheduler so that they
166 * know about it. That allows them to make better scheduling
167 * decisions. So when the below references 'sync' and 'async', it
168 * is referencing this priority hint.
169 *
170 * With that in mind, the available types are:
171 *
172 * READ A normal read operation. Device will be plugged.
173 * READ_SYNC A synchronous read. Device is not plugged, caller can
174 * immediately wait on this read without caring about
175 * unplugging.
176 * READA Used for read-ahead operations. Lower priority, and the
177 * block layer could (in theory) choose to ignore this
178 * request if it runs into resource problems.
179 * WRITE A normal async write. Device will be plugged.
180 * WRITE_SYNC Synchronous write. Identical to WRITE, but passes down
181 * the hint that someone will be waiting on this IO
182 * shortly. The write equivalent of READ_SYNC.
183 * WRITE_ODIRECT Special case write for O_DIRECT only.
184 * WRITE_FLUSH Like WRITE_SYNC but with preceding cache flush.
185 * WRITE_FUA Like WRITE_SYNC but data is guaranteed to be on
186 * non-volatile media on completion.
187 * WRITE_FLUSH_FUA Combination of WRITE_FLUSH and FUA. The IO is preceded
188 * by a cache flush and data is guaranteed to be on
189 * non-volatile media on completion.
190 *
191 */
192#define RW_MASK REQ_WRITE
193#define RWA_MASK REQ_RAHEAD
194
195#define READ 0
196#define WRITE RW_MASK
197#define READA RWA_MASK
David Howells08cce05c2012-10-13 10:45:06 +0100198
199#define READ_SYNC (READ | REQ_SYNC)
200#define WRITE_SYNC (WRITE | REQ_SYNC | REQ_NOIDLE)
201#define WRITE_ODIRECT (WRITE | REQ_SYNC)
202#define WRITE_FLUSH (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FLUSH)
203#define WRITE_FUA (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FUA)
204#define WRITE_FLUSH_FUA (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FLUSH | REQ_FUA)
205
206/*
Linus Torvalds1da177e2005-04-16 15:20:36 -0700207 * Attribute flags. These should be or-ed together to figure out what
208 * has been changed!
209 */
Miklos Szeredi9767d742008-07-01 15:01:26 +0200210#define ATTR_MODE (1 << 0)
211#define ATTR_UID (1 << 1)
212#define ATTR_GID (1 << 2)
213#define ATTR_SIZE (1 << 3)
214#define ATTR_ATIME (1 << 4)
215#define ATTR_MTIME (1 << 5)
216#define ATTR_CTIME (1 << 6)
217#define ATTR_ATIME_SET (1 << 7)
218#define ATTR_MTIME_SET (1 << 8)
219#define ATTR_FORCE (1 << 9) /* Not a change, but a change it */
220#define ATTR_ATTR_FLAG (1 << 10)
221#define ATTR_KILL_SUID (1 << 11)
222#define ATTR_KILL_SGID (1 << 12)
223#define ATTR_FILE (1 << 13)
224#define ATTR_KILL_PRIV (1 << 14)
225#define ATTR_OPEN (1 << 15) /* Truncating from open(O_TRUNC) */
226#define ATTR_TIMES_SET (1 << 16)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700227
228/*
Miklos Szeredi787fb6b2014-10-24 00:14:36 +0200229 * Whiteout is represented by a char device. The following constants define the
230 * mode and device number to use.
231 */
232#define WHITEOUT_MODE 0
233#define WHITEOUT_DEV 0
234
235/*
Linus Torvalds1da177e2005-04-16 15:20:36 -0700236 * This is the Inode Attributes structure, used for notify_change(). It
237 * uses the above definitions as flags, to know which values have changed.
238 * Also, in this manner, a Filesystem can look at only the values it cares
239 * about. Basically, these are the attributes that the VFS layer can
240 * request to change from the FS layer.
241 *
242 * Derek Atkins <warlord@MIT.EDU> 94-10-20
243 */
244struct iattr {
245 unsigned int ia_valid;
246 umode_t ia_mode;
Eric W. Biederman92361632012-02-08 07:07:50 -0800247 kuid_t ia_uid;
248 kgid_t ia_gid;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700249 loff_t ia_size;
250 struct timespec ia_atime;
251 struct timespec ia_mtime;
252 struct timespec ia_ctime;
Miklos Szeredicc4e69d2005-11-07 00:59:49 -0800253
254 /*
Lucas De Marchi25985ed2011-03-30 22:57:33 -0300255 * Not an attribute, but an auxiliary info for filesystems wanting to
Miklos Szeredicc4e69d2005-11-07 00:59:49 -0800256 * implement an ftruncate() like method. NOTE: filesystem should
257 * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL).
258 */
259 struct file *ia_file;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700260};
261
262/*
Linus Torvalds1da177e2005-04-16 15:20:36 -0700263 * Includes for diskquotas.
264 */
265#include <linux/quota.h>
266
Miklos Szeredi69c433e2014-10-24 00:14:39 +0200267/*
268 * Maximum number of layers of fs stack. Needs to be limited to
269 * prevent kernel stack overflow
270 */
271#define FILESYSTEM_MAX_STACK_DEPTH 2
272
Zach Brown994fc28c2005-12-15 14:28:17 -0800273/**
274 * enum positive_aop_returns - aop return codes with specific semantics
275 *
276 * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has
277 * completed, that the page is still locked, and
278 * should be considered active. The VM uses this hint
279 * to return the page to the active list -- it won't
280 * be a candidate for writeback again in the near
281 * future. Other callers must be careful to unlock
282 * the page if they get this return. Returned by
283 * writepage();
284 *
285 * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has
286 * unlocked it and the page might have been truncated.
287 * The caller should back up to acquiring a new page and
288 * trying again. The aop will be taking reasonable
289 * precautions not to livelock. If the caller held a page
290 * reference, it should drop it before retrying. Returned
Nick Piggin55144762007-10-16 01:25:26 -0700291 * by readpage().
Zach Brown994fc28c2005-12-15 14:28:17 -0800292 *
293 * address_space_operation functions return these large constants to indicate
294 * special semantics to the caller. These are much larger than the bytes in a
295 * page to allow for functions that return the number of bytes operated on in a
296 * given page.
297 */
298
299enum positive_aop_returns {
300 AOP_WRITEPAGE_ACTIVATE = 0x80000,
301 AOP_TRUNCATED_PAGE = 0x80001,
302};
303
Nick Pigginafddba42007-10-16 01:25:01 -0700304#define AOP_FLAG_UNINTERRUPTIBLE 0x0001 /* will not do a short write */
Nick Piggin89e10782007-10-16 01:25:07 -0700305#define AOP_FLAG_CONT_EXPAND 0x0002 /* called from cont_expand */
Nick Piggin54566b22009-01-04 12:00:53 -0800306#define AOP_FLAG_NOFS 0x0004 /* used by filesystem to direct
307 * helper code (eg buffer layer)
308 * to clear GFP_FS from alloc */
Nick Pigginafddba42007-10-16 01:25:01 -0700309
Linus Torvalds1da177e2005-04-16 15:20:36 -0700310/*
311 * oh the beauties of C type declarations.
312 */
313struct page;
314struct address_space;
315struct writeback_control;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700316
Christoph Hellwige2e40f22015-02-22 08:58:50 -0800317#define IOCB_EVENTFD (1 << 0)
Al Viro2ba48ce2015-04-09 13:52:01 -0400318#define IOCB_APPEND (1 << 1)
319#define IOCB_DIRECT (1 << 2)
Christoph Hellwige2e40f22015-02-22 08:58:50 -0800320
321struct kiocb {
322 struct file *ki_filp;
323 loff_t ki_pos;
324 void (*ki_complete)(struct kiocb *iocb, long ret, long ret2);
325 void *private;
326 int ki_flags;
327};
328
329static inline bool is_sync_kiocb(struct kiocb *kiocb)
330{
331 return kiocb->ki_complete == NULL;
332}
333
Al Viro2ba48ce2015-04-09 13:52:01 -0400334static inline int iocb_flags(struct file *file);
335
Christoph Hellwige2e40f22015-02-22 08:58:50 -0800336static inline void init_sync_kiocb(struct kiocb *kiocb, struct file *filp)
337{
338 *kiocb = (struct kiocb) {
339 .ki_filp = filp,
Al Viro2ba48ce2015-04-09 13:52:01 -0400340 .ki_flags = iocb_flags(filp),
Christoph Hellwige2e40f22015-02-22 08:58:50 -0800341 };
342}
343
Hisashi Hifumi8ab22b92008-07-28 15:46:36 -0700344/*
345 * "descriptor" for what we're up to with a read.
346 * This allows us to use the same read code yet
347 * have multiple different users of the data that
348 * we read from a file.
349 *
350 * The simplest case just copies the data to user
351 * mode.
352 */
353typedef struct {
354 size_t written;
355 size_t count;
356 union {
357 char __user *buf;
358 void *data;
359 } arg;
360 int error;
361} read_descriptor_t;
362
363typedef int (*read_actor_t)(read_descriptor_t *, struct page *,
364 unsigned long, unsigned long);
Nick Piggin2f718ff2007-10-16 01:24:59 -0700365
Linus Torvalds1da177e2005-04-16 15:20:36 -0700366struct address_space_operations {
367 int (*writepage)(struct page *page, struct writeback_control *wbc);
368 int (*readpage)(struct file *, struct page *);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700369
370 /* Write back some dirty pages from this mapping. */
371 int (*writepages)(struct address_space *, struct writeback_control *);
372
Andrew Morton4741c9f2006-03-24 03:18:11 -0800373 /* Set a page dirty. Return true if this dirtied it */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700374 int (*set_page_dirty)(struct page *page);
375
376 int (*readpages)(struct file *filp, struct address_space *mapping,
377 struct list_head *pages, unsigned nr_pages);
378
Nick Pigginafddba42007-10-16 01:25:01 -0700379 int (*write_begin)(struct file *, struct address_space *mapping,
380 loff_t pos, unsigned len, unsigned flags,
381 struct page **pagep, void **fsdata);
382 int (*write_end)(struct file *, struct address_space *mapping,
383 loff_t pos, unsigned len, unsigned copied,
384 struct page *page, void *fsdata);
385
Linus Torvalds1da177e2005-04-16 15:20:36 -0700386 /* Unfortunately this kludge is needed for FIBMAP. Don't use it */
387 sector_t (*bmap)(struct address_space *, sector_t);
Lukas Czernerd47992f2013-05-21 23:17:23 -0400388 void (*invalidatepage) (struct page *, unsigned int, unsigned int);
Al Viro27496a82005-10-21 03:20:48 -0400389 int (*releasepage) (struct page *, gfp_t);
Linus Torvalds6072d132010-12-01 13:35:19 -0500390 void (*freepage)(struct page *);
Omar Sandoval22c61862015-03-16 04:33:53 -0700391 ssize_t (*direct_IO)(struct kiocb *, struct iov_iter *iter, loff_t offset);
Mel Gormanb969c4a2012-01-12 17:19:34 -0800392 /*
Tang Chenef277c72013-07-08 16:00:21 -0700393 * migrate the contents of a page to the specified target. If
394 * migrate_mode is MIGRATE_ASYNC, it must not block.
Mel Gormanb969c4a2012-01-12 17:19:34 -0800395 */
Christoph Lameter2d1db3b2006-06-23 02:03:33 -0700396 int (*migratepage) (struct address_space *,
Mel Gormana6bc32b2012-01-12 17:19:43 -0800397 struct page *, struct page *, enum migrate_mode);
Trond Myklebuste3db7692007-01-10 23:15:39 -0800398 int (*launder_page) (struct page *);
Al Viroc186afb42014-02-02 21:16:54 -0500399 int (*is_partially_uptodate) (struct page *, unsigned long,
Hisashi Hifumi8ab22b92008-07-28 15:46:36 -0700400 unsigned long);
Mel Gormanb4597222013-07-03 15:02:05 -0700401 void (*is_dirty_writeback) (struct page *, bool *, bool *);
Andi Kleen25718732009-09-16 11:50:13 +0200402 int (*error_remove_page)(struct address_space *, struct page *);
Mel Gorman62c230b2012-07-31 16:44:55 -0700403
404 /* swapfile support */
Mel Gormana509bc12012-07-31 16:44:57 -0700405 int (*swap_activate)(struct swap_info_struct *sis, struct file *file,
406 sector_t *span);
407 void (*swap_deactivate)(struct file *file);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700408};
409
Jens Axboe7dcda1c2011-04-05 23:51:48 +0200410extern const struct address_space_operations empty_aops;
411
Nick Pigginafddba42007-10-16 01:25:01 -0700412/*
413 * pagecache_write_begin/pagecache_write_end must be used by general code
414 * to write into the pagecache.
415 */
416int pagecache_write_begin(struct file *, struct address_space *mapping,
417 loff_t pos, unsigned len, unsigned flags,
418 struct page **pagep, void **fsdata);
419
420int pagecache_write_end(struct file *, struct address_space *mapping,
421 loff_t pos, unsigned len, unsigned copied,
422 struct page *page, void *fsdata);
423
Linus Torvalds1da177e2005-04-16 15:20:36 -0700424struct address_space {
425 struct inode *host; /* owner: inode, block_device */
426 struct radix_tree_root page_tree; /* radix tree of all pages */
Nick Piggin19fd6232008-07-25 19:45:32 -0700427 spinlock_t tree_lock; /* and lock protecting it */
David Herrmann4bb5f5d2014-08-08 14:25:25 -0700428 atomic_t i_mmap_writable;/* count VM_SHARED mappings */
Michel Lespinasse6b2dbba2012-10-08 16:31:25 -0700429 struct rb_root i_mmap; /* tree of private and shared mappings */
Davidlohr Buesoc8c06ef2014-12-12 16:54:24 -0800430 struct rw_semaphore i_mmap_rwsem; /* protect tree, count, list */
Jan Kara08142572011-06-27 16:18:10 -0700431 /* Protected by tree_lock together with the radix tree */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700432 unsigned long nrpages; /* number of total pages */
Johannes Weiner91b0abe2014-04-03 14:47:49 -0700433 unsigned long nrshadows; /* number of shadow entries */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700434 pgoff_t writeback_index;/* writeback starts here */
Christoph Hellwigf5e54d62006-06-28 04:26:44 -0700435 const struct address_space_operations *a_ops; /* methods */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700436 unsigned long flags; /* error bits/gfp mask */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700437 spinlock_t private_lock; /* for use by the address_space */
438 struct list_head private_list; /* ditto */
Rafael Aquini252aa6f2012-12-11 16:02:35 -0800439 void *private_data; /* ditto */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700440} __attribute__((aligned(sizeof(long))));
441 /*
442 * On most architectures that alignment is already the case; but
Lucas De Marchi25985ed2011-03-30 22:57:33 -0300443 * must be enforced here for CRIS, to let the least significant bit
Linus Torvalds1da177e2005-04-16 15:20:36 -0700444 * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
445 */
Andi Kleen87192a22012-01-12 17:20:34 -0800446struct request_queue;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700447
448struct block_device {
449 dev_t bd_dev; /* not a kdev_t - it's a search key */
Richard Kennedyf3ccfcd2011-03-22 16:33:56 -0700450 int bd_openers;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700451 struct inode * bd_inode; /* will die */
Theodore Ts'o87d8fe12009-01-03 09:47:09 -0500452 struct super_block * bd_super;
Arjan van de Venc039e312006-03-23 03:00:28 -0800453 struct mutex bd_mutex; /* open/close mutex */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700454 struct list_head bd_inodes;
Tejun Heo6b4517a2010-04-07 18:53:59 +0900455 void * bd_claiming;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700456 void * bd_holder;
457 int bd_holders;
Tejun Heo77ea8872010-12-08 20:57:37 +0100458 bool bd_write_holder;
Jun'ichi Nomura641dc632006-03-27 01:17:57 -0800459#ifdef CONFIG_SYSFS
Tejun Heo49731ba2011-01-14 18:43:57 +0100460 struct list_head bd_holder_disks;
Jun'ichi Nomura641dc632006-03-27 01:17:57 -0800461#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -0700462 struct block_device * bd_contains;
463 unsigned bd_block_size;
464 struct hd_struct * bd_part;
465 /* number of times partitions within this device have been opened. */
466 unsigned bd_part_count;
467 int bd_invalidated;
468 struct gendisk * bd_disk;
Andi Kleen87192a22012-01-12 17:20:34 -0800469 struct request_queue * bd_queue;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700470 struct list_head bd_list;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700471 /*
472 * Private data. You must have bd_claim'ed the block_device
473 * to use this. NOTE: bd_claim allows an owner to claim
474 * the same device multiple times, the owner must take special
475 * care to not mess up bd_private for that case.
476 */
477 unsigned long bd_private;
Takashi Satofcccf502009-01-09 16:40:59 -0800478
479 /* The counter of freeze processes */
480 int bd_fsfreeze_count;
481 /* Mutex for freeze */
482 struct mutex bd_fsfreeze_mutex;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700483};
484
485/*
486 * Radix-tree tags, for tagging dirty and writeback pages within the pagecache
487 * radix trees
488 */
489#define PAGECACHE_TAG_DIRTY 0
490#define PAGECACHE_TAG_WRITEBACK 1
Jan Karaf446daae2010-08-09 17:19:12 -0700491#define PAGECACHE_TAG_TOWRITE 2
Linus Torvalds1da177e2005-04-16 15:20:36 -0700492
493int mapping_tagged(struct address_space *mapping, int tag);
494
Davidlohr Bueso8b28f622014-12-12 16:54:18 -0800495static inline void i_mmap_lock_write(struct address_space *mapping)
496{
Davidlohr Buesoc8c06ef2014-12-12 16:54:24 -0800497 down_write(&mapping->i_mmap_rwsem);
Davidlohr Bueso8b28f622014-12-12 16:54:18 -0800498}
499
500static inline void i_mmap_unlock_write(struct address_space *mapping)
501{
Davidlohr Buesoc8c06ef2014-12-12 16:54:24 -0800502 up_write(&mapping->i_mmap_rwsem);
Davidlohr Bueso8b28f622014-12-12 16:54:18 -0800503}
504
Davidlohr Bueso3dec0ba2014-12-12 16:54:27 -0800505static inline void i_mmap_lock_read(struct address_space *mapping)
506{
507 down_read(&mapping->i_mmap_rwsem);
508}
509
510static inline void i_mmap_unlock_read(struct address_space *mapping)
511{
512 up_read(&mapping->i_mmap_rwsem);
513}
514
Linus Torvalds1da177e2005-04-16 15:20:36 -0700515/*
516 * Might pages of this file be mapped into userspace?
517 */
518static inline int mapping_mapped(struct address_space *mapping)
519{
Kirill A. Shutemov27ba0642015-02-10 14:09:59 -0800520 return !RB_EMPTY_ROOT(&mapping->i_mmap);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700521}
522
523/*
524 * Might pages of this file have been modified in userspace?
525 * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap_pgoff
526 * marks vma as VM_SHARED if it is shared, and the file was opened for
527 * writing i.e. vma may be mprotected writable even if now readonly.
David Herrmann4bb5f5d2014-08-08 14:25:25 -0700528 *
529 * If i_mmap_writable is negative, no new writable mappings are allowed. You
530 * can only deny writable mappings, if none exists right now.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700531 */
532static inline int mapping_writably_mapped(struct address_space *mapping)
533{
David Herrmann4bb5f5d2014-08-08 14:25:25 -0700534 return atomic_read(&mapping->i_mmap_writable) > 0;
535}
536
537static inline int mapping_map_writable(struct address_space *mapping)
538{
539 return atomic_inc_unless_negative(&mapping->i_mmap_writable) ?
540 0 : -EPERM;
541}
542
543static inline void mapping_unmap_writable(struct address_space *mapping)
544{
545 atomic_dec(&mapping->i_mmap_writable);
546}
547
548static inline int mapping_deny_writable(struct address_space *mapping)
549{
550 return atomic_dec_unless_positive(&mapping->i_mmap_writable) ?
551 0 : -EBUSY;
552}
553
554static inline void mapping_allow_writable(struct address_space *mapping)
555{
556 atomic_inc(&mapping->i_mmap_writable);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700557}
558
559/*
560 * Use sequence counter to get consistent i_size on 32-bit processors.
561 */
562#if BITS_PER_LONG==32 && defined(CONFIG_SMP)
563#include <linux/seqlock.h>
564#define __NEED_I_SIZE_ORDERED
565#define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
566#else
567#define i_size_ordered_init(inode) do { } while (0)
568#endif
569
Al Virof19d4a82009-06-08 19:50:45 -0400570struct posix_acl;
571#define ACL_NOT_CACHED ((void *)(-1))
572
Linus Torvalds3ddcd052011-08-06 22:45:50 -0700573#define IOP_FASTPERM 0x0001
574#define IOP_LOOKUP 0x0002
575#define IOP_NOFOLLOW 0x0004
576
577/*
578 * Keep mostly read-only and often accessed (especially for
579 * the RCU path lookup and 'stat' data) fields at the beginning
580 * of the 'struct inode'
581 */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700582struct inode {
Nick Piggin44a7d7a2011-01-07 17:49:56 +1100583 umode_t i_mode;
Linus Torvalds3ddcd052011-08-06 22:45:50 -0700584 unsigned short i_opflags;
Eric W. Biederman92361632012-02-08 07:07:50 -0800585 kuid_t i_uid;
586 kgid_t i_gid;
Linus Torvalds3ddcd052011-08-06 22:45:50 -0700587 unsigned int i_flags;
588
589#ifdef CONFIG_FS_POSIX_ACL
590 struct posix_acl *i_acl;
591 struct posix_acl *i_default_acl;
592#endif
593
Nick Piggin44a7d7a2011-01-07 17:49:56 +1100594 const struct inode_operations *i_op;
595 struct super_block *i_sb;
Linus Torvalds3ddcd052011-08-06 22:45:50 -0700596 struct address_space *i_mapping;
Nick Piggin44a7d7a2011-01-07 17:49:56 +1100597
Linus Torvalds13e12d12011-06-08 15:18:19 -0700598#ifdef CONFIG_SECURITY
599 void *i_security;
600#endif
Nick Piggin44a7d7a2011-01-07 17:49:56 +1100601
Linus Torvalds3ddcd052011-08-06 22:45:50 -0700602 /* Stat data, not accessed from path walking */
603 unsigned long i_ino;
Miklos Szeredia78ef702011-10-28 14:13:30 +0200604 /*
605 * Filesystems may only read i_nlink directly. They shall use the
606 * following functions for modification:
607 *
608 * (set|clear|inc|drop)_nlink
609 * inode_(inc|dec)_link_count
610 */
611 union {
612 const unsigned int i_nlink;
613 unsigned int __i_nlink;
614 };
Linus Torvalds3ddcd052011-08-06 22:45:50 -0700615 dev_t i_rdev;
Linus Torvalds2f9d3df2012-06-03 14:50:19 -0700616 loff_t i_size;
Linus Torvalds3ddcd052011-08-06 22:45:50 -0700617 struct timespec i_atime;
618 struct timespec i_mtime;
619 struct timespec i_ctime;
Theodore Ts'o6cdbb0e2011-10-29 08:24:18 -0400620 spinlock_t i_lock; /* i_blocks, i_bytes, maybe i_size */
621 unsigned short i_bytes;
Linus Torvalds2f9d3df2012-06-03 14:50:19 -0700622 unsigned int i_blkbits;
Linus Torvalds3ddcd052011-08-06 22:45:50 -0700623 blkcnt_t i_blocks;
624
625#ifdef __NEED_I_SIZE_ORDERED
626 seqcount_t i_size_seqcount;
627#endif
628
629 /* Misc */
630 unsigned long i_state;
Linus Torvalds3ddcd052011-08-06 22:45:50 -0700631 struct mutex i_mutex;
Linus Torvalds13e12d12011-06-08 15:18:19 -0700632
Nick Piggin44a7d7a2011-01-07 17:49:56 +1100633 unsigned long dirtied_when; /* jiffies of first dirtying */
Theodore Ts'oa2f48702015-03-17 12:23:19 -0400634 unsigned long dirtied_time_when;
Nick Piggin44a7d7a2011-01-07 17:49:56 +1100635
Linus Torvalds1da177e2005-04-16 15:20:36 -0700636 struct hlist_node i_hash;
Nick Piggin7ccf19a2010-10-21 11:49:30 +1100637 struct list_head i_wb_list; /* backing dev IO list */
638 struct list_head i_lru; /* inode LRU list */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700639 struct list_head i_sb_list;
Nick Pigginfa0d7e3d2011-01-07 17:49:49 +1100640 union {
Al Virob3d9b7a2012-06-09 13:51:19 -0400641 struct hlist_head i_dentry;
Nick Pigginfa0d7e3d2011-01-07 17:49:49 +1100642 struct rcu_head i_rcu;
643 };
Jean Noel Cordenner7a224222008-01-28 23:58:27 -0500644 u64 i_version;
Linus Torvalds2f9d3df2012-06-03 14:50:19 -0700645 atomic_t i_count;
Christoph Hellwigbd5fe6c2011-06-24 14:29:43 -0400646 atomic_t i_dio_count;
Theodore Ts'o6cdbb0e2011-10-29 08:24:18 -0400647 atomic_t i_writecount;
Mimi Zohard984ea62013-12-11 15:20:54 -0500648#ifdef CONFIG_IMA
649 atomic_t i_readcount; /* struct files open RO */
650#endif
Arjan van de Ven99ac48f2006-03-28 01:56:41 -0800651 const struct file_operations *i_fop; /* former ->i_op->default_file_ops */
Jeff Layton4a075e32015-01-16 15:05:54 -0500652 struct file_lock_context *i_flctx;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700653 struct address_space i_data;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700654 struct list_head i_devices;
Theodore Ts'o4c154162006-09-27 01:50:47 -0700655 union {
656 struct pipe_inode_info *i_pipe;
Theodore Ts'oeaf796e2006-09-27 01:50:48 -0700657 struct block_device *i_bdev;
Theodore Ts'o577c4eb2006-09-27 01:50:49 -0700658 struct cdev *i_cdev;
Theodore Ts'o4c154162006-09-27 01:50:47 -0700659 };
Linus Torvalds1da177e2005-04-16 15:20:36 -0700660
661 __u32 i_generation;
662
Eric Paris3be25f42009-05-21 17:01:26 -0400663#ifdef CONFIG_FSNOTIFY
664 __u32 i_fsnotify_mask; /* all events this inode cares about */
Eric Parise61ce862009-12-17 21:24:24 -0500665 struct hlist_head i_fsnotify_marks;
Robert Love0eeca282005-07-12 17:06:03 -0400666#endif
667
Theodore Ts'o8e18e292006-09-27 01:50:46 -0700668 void *i_private; /* fs or device private pointer */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700669};
670
Al Viro1d3382cb2010-10-23 15:19:20 -0400671static inline int inode_unhashed(struct inode *inode)
672{
673 return hlist_unhashed(&inode->i_hash);
674}
675
Linus Torvalds1da177e2005-04-16 15:20:36 -0700676/*
Ingo Molnarf2eace22006-07-03 00:25:05 -0700677 * inode->i_mutex nesting subclasses for the lock validator:
678 *
679 * 0: the object of the current VFS operation
680 * 1: parent
681 * 2: child/target
J. Bruce Fields40bd22c2012-04-18 15:21:34 -0400682 * 3: xattr
683 * 4: second non-directory
Miklos Szeredid1b72cc2014-10-27 15:42:01 +0100684 * 5: second parent (when locking independent directories in rename)
685 *
686 * I_MUTEX_NONDIR2 is for certain operations (such as rename) which lock two
J. Bruce Fields40bd22c2012-04-18 15:21:34 -0400687 * non-directories at once.
Ingo Molnarf2eace22006-07-03 00:25:05 -0700688 *
689 * The locking order between these classes is
Miklos Szeredid1b72cc2014-10-27 15:42:01 +0100690 * parent[2] -> child -> grandchild -> normal -> xattr -> second non-directory
Ingo Molnarf2eace22006-07-03 00:25:05 -0700691 */
692enum inode_i_mutex_lock_class
693{
694 I_MUTEX_NORMAL,
695 I_MUTEX_PARENT,
696 I_MUTEX_CHILD,
Ingo Molnar4df46242006-08-27 01:23:56 -0700697 I_MUTEX_XATTR,
Miklos Szeredid1b72cc2014-10-27 15:42:01 +0100698 I_MUTEX_NONDIR2,
699 I_MUTEX_PARENT2,
Ingo Molnarf2eace22006-07-03 00:25:05 -0700700};
701
J. Bruce Fields375e2892012-04-18 15:16:33 -0400702void lock_two_nondirectories(struct inode *, struct inode*);
703void unlock_two_nondirectories(struct inode *, struct inode*);
704
Ingo Molnarf2eace22006-07-03 00:25:05 -0700705/*
Linus Torvalds1da177e2005-04-16 15:20:36 -0700706 * NOTE: in a 32bit arch with a preemptable kernel and
707 * an UP compile the i_size_read/write must be atomic
708 * with respect to the local cpu (unlike with preempt disabled),
709 * but they don't need to be atomic with respect to other cpus like in
710 * true SMP (so they need either to either locally disable irq around
711 * the read or for example on x86 they can be still implemented as a
712 * cmpxchg8b without the need of the lock prefix). For SMP compiles
713 * and 64bit archs it makes no difference if preempt is enabled or not.
714 */
Jan Engelhardt48ed2142006-12-06 20:35:37 -0800715static inline loff_t i_size_read(const struct inode *inode)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700716{
717#if BITS_PER_LONG==32 && defined(CONFIG_SMP)
718 loff_t i_size;
719 unsigned int seq;
720
721 do {
722 seq = read_seqcount_begin(&inode->i_size_seqcount);
723 i_size = inode->i_size;
724 } while (read_seqcount_retry(&inode->i_size_seqcount, seq));
725 return i_size;
726#elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
727 loff_t i_size;
728
729 preempt_disable();
730 i_size = inode->i_size;
731 preempt_enable();
732 return i_size;
733#else
734 return inode->i_size;
735#endif
736}
737
Miklos Szeredi7762f5a2006-10-17 00:10:07 -0700738/*
739 * NOTE: unlike i_size_read(), i_size_write() does need locking around it
740 * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount
741 * can be lost, resulting in subsequent i_size_read() calls spinning forever.
742 */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700743static inline void i_size_write(struct inode *inode, loff_t i_size)
744{
745#if BITS_PER_LONG==32 && defined(CONFIG_SMP)
Fan Du74e3d1e2013-04-30 15:27:27 -0700746 preempt_disable();
Linus Torvalds1da177e2005-04-16 15:20:36 -0700747 write_seqcount_begin(&inode->i_size_seqcount);
748 inode->i_size = i_size;
749 write_seqcount_end(&inode->i_size_seqcount);
Fan Du74e3d1e2013-04-30 15:27:27 -0700750 preempt_enable();
Linus Torvalds1da177e2005-04-16 15:20:36 -0700751#elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
752 preempt_disable();
753 inode->i_size = i_size;
754 preempt_enable();
755#else
756 inode->i_size = i_size;
757#endif
758}
759
Eric W. Biederman92361632012-02-08 07:07:50 -0800760/* Helper functions so that in most cases filesystems will
761 * not need to deal directly with kuid_t and kgid_t and can
762 * instead deal with the raw numeric values that are stored
763 * in the filesystem.
764 */
765static inline uid_t i_uid_read(const struct inode *inode)
766{
767 return from_kuid(&init_user_ns, inode->i_uid);
768}
769
770static inline gid_t i_gid_read(const struct inode *inode)
771{
772 return from_kgid(&init_user_ns, inode->i_gid);
773}
774
775static inline void i_uid_write(struct inode *inode, uid_t uid)
776{
777 inode->i_uid = make_kuid(&init_user_ns, uid);
778}
779
780static inline void i_gid_write(struct inode *inode, gid_t gid)
781{
782 inode->i_gid = make_kgid(&init_user_ns, gid);
783}
784
Jan Engelhardt48ed2142006-12-06 20:35:37 -0800785static inline unsigned iminor(const struct inode *inode)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700786{
787 return MINOR(inode->i_rdev);
788}
789
Jan Engelhardt48ed2142006-12-06 20:35:37 -0800790static inline unsigned imajor(const struct inode *inode)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700791{
792 return MAJOR(inode->i_rdev);
793}
794
795extern struct block_device *I_BDEV(struct inode *inode);
796
797struct fown_struct {
798 rwlock_t lock; /* protects pid, uid, euid fields */
Eric W. Biederman609d7fa2006-10-02 02:17:15 -0700799 struct pid *pid; /* pid or -pgrp where SIGIO should be sent */
800 enum pid_type pid_type; /* Kind of process group SIGIO should be sent to */
Eric W. Biederman92361632012-02-08 07:07:50 -0800801 kuid_t uid, euid; /* uid/euid of process setting the owner */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700802 int signum; /* posix.1b rt signal to be delivered on IO */
803};
804
805/*
806 * Track a single file's readahead state
807 */
808struct file_ra_state {
Fengguang Wu937085a2007-10-16 01:24:31 -0700809 pgoff_t start; /* where readahead started */
810 unsigned int size; /* # of readahead pages */
811 unsigned int async_size; /* do asynchronous readahead when
Fengguang Wuf9acc8c2007-07-19 01:48:08 -0700812 there are only # of pages ahead */
Fengguang Wu5ce11102007-07-19 01:47:59 -0700813
Fengguang Wu937085a2007-10-16 01:24:31 -0700814 unsigned int ra_pages; /* Maximum readahead window */
Wu Fengguang1ebf26a2009-06-16 15:31:19 -0700815 unsigned int mmap_miss; /* Cache miss stat for mmap accesses */
Fengguang Wuf4e6b492007-10-16 01:24:33 -0700816 loff_t prev_pos; /* Cache last read() position */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700817};
Linus Torvalds1da177e2005-04-16 15:20:36 -0700818
Fengguang Wu5ce11102007-07-19 01:47:59 -0700819/*
Fengguang Wu5ce11102007-07-19 01:47:59 -0700820 * Check if @index falls in the readahead windows.
821 */
822static inline int ra_has_index(struct file_ra_state *ra, pgoff_t index)
823{
Fengguang Wuf9acc8c2007-07-19 01:48:08 -0700824 return (index >= ra->start &&
825 index < ra->start + ra->size);
Fengguang Wu5ce11102007-07-19 01:47:59 -0700826}
827
Linus Torvalds1da177e2005-04-16 15:20:36 -0700828struct file {
Eric Dumazet2f512012005-10-30 15:02:16 -0800829 union {
Oleg Nesterov4f5e65a2013-07-08 14:24:16 -0700830 struct llist_node fu_llist;
Eric Dumazet2f512012005-10-30 15:02:16 -0800831 struct rcu_head fu_rcuhead;
832 } f_u;
Josef "Jeff" Sipek0f7fc9e2006-12-08 02:36:35 -0800833 struct path f_path;
Al Virodd379782013-03-01 19:48:30 -0500834 struct inode *f_inode; /* cached value */
Arjan van de Ven99ac48f2006-03-28 01:56:41 -0800835 const struct file_operations *f_op;
Andi Kleenef3d0fd2011-09-15 16:06:48 -0700836
837 /*
Linus Torvalds9c225f22014-03-03 09:36:58 -0800838 * Protects f_ep_links, f_flags.
Andi Kleenef3d0fd2011-09-15 16:06:48 -0700839 * Must not be taken from IRQ context.
840 */
841 spinlock_t f_lock;
Al Viro516e0cc2008-07-26 00:39:17 -0400842 atomic_long_t f_count;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700843 unsigned int f_flags;
Al Viroaeb5d722008-09-02 15:28:45 -0400844 fmode_t f_mode;
Linus Torvalds9c225f22014-03-03 09:36:58 -0800845 struct mutex f_pos_lock;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700846 loff_t f_pos;
847 struct fown_struct f_owner;
David Howellsd76b0d92008-11-14 10:39:25 +1100848 const struct cred *f_cred;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700849 struct file_ra_state f_ra;
850
Mathieu Desnoyers2b47c362007-10-16 23:27:21 -0700851 u64 f_version;
Alexey Dobriyan50462062006-09-29 02:00:01 -0700852#ifdef CONFIG_SECURITY
Linus Torvalds1da177e2005-04-16 15:20:36 -0700853 void *f_security;
Alexey Dobriyan50462062006-09-29 02:00:01 -0700854#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -0700855 /* needed for tty driver, and maybe others */
856 void *private_data;
857
858#ifdef CONFIG_EPOLL
859 /* Used by fs/eventpoll.c to link all the hooks to this file */
860 struct list_head f_ep_links;
Jason Baron28d82dc2012-01-12 17:17:43 -0800861 struct list_head f_tfile_llink;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700862#endif /* #ifdef CONFIG_EPOLL */
863 struct address_space *f_mapping;
Al Virobd2a31d2014-03-04 14:54:22 -0500864} __attribute__((aligned(4))); /* lest something weird decides that 2 is OK */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700865
Aneesh Kumar K.V990d6c22011-01-29 18:43:26 +0530866struct file_handle {
867 __u32 handle_bytes;
868 int handle_type;
869 /* file identifier */
870 unsigned char f_handle[0];
871};
872
Al Virocb0942b2012-08-27 14:48:26 -0400873static inline struct file *get_file(struct file *f)
874{
875 atomic_long_inc(&f->f_count);
876 return f;
877}
Konstantin Khlebnikov90f31d02015-04-16 12:47:56 -0700878#define get_file_rcu(x) atomic_long_inc_not_zero(&(x)->f_count)
Al Virod7065da2010-05-26 15:13:55 -0400879#define fput_atomic(x) atomic_long_add_unless(&(x)->f_count, -1, 1)
Al Viro516e0cc2008-07-26 00:39:17 -0400880#define file_count(x) atomic_long_read(&(x)->f_count)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700881
882#define MAX_NON_LFS ((1UL<<31) - 1)
883
884/* Page cache limit. The filesystems should put that into their s_maxbytes
885 limits, otherwise bad things can happen in VM. */
886#if BITS_PER_LONG==32
Chuck Lever2bd2c192012-09-24 14:51:48 -0400887#define MAX_LFS_FILESIZE (((loff_t)PAGE_CACHE_SIZE << (BITS_PER_LONG-1))-1)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700888#elif BITS_PER_LONG==64
Al Viro614c3212012-10-10 01:06:20 -0400889#define MAX_LFS_FILESIZE ((loff_t)0x7fffffffffffffffLL)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700890#endif
891
892#define FL_POSIX 1
893#define FL_FLOCK 2
J. Bruce Fields617588d2011-07-01 15:18:34 -0400894#define FL_DELEG 4 /* NFSv4 delegation */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700895#define FL_ACCESS 8 /* not trying to lock, just looking */
Trond Myklebustf475ae92006-06-29 16:38:32 -0400896#define FL_EXISTS 16 /* when unlocking, test for existence */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700897#define FL_LEASE 32 /* lease held on this file */
Miklos Szeredi75e1fcc2006-06-23 02:05:12 -0700898#define FL_CLOSE 64 /* unlock on close */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700899#define FL_SLEEP 128 /* A blocking lock */
J. Bruce Fields778fc542011-07-26 18:25:49 -0400900#define FL_DOWNGRADE_PENDING 256 /* Lease is being downgraded */
901#define FL_UNLOCK_PENDING 512 /* Lease is being broken */
Jeff Laytoncff2fce2014-04-22 08:24:32 -0400902#define FL_OFDLCK 1024 /* lock is "owned" by struct file */
Christoph Hellwig11afe9f2015-01-21 19:17:03 +0100903#define FL_LAYOUT 2048 /* outstanding pNFS layout */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700904
905/*
Miklos Szeredibde74e42008-07-25 01:48:57 -0700906 * Special return value from posix_lock_file() and vfs_lock_file() for
907 * asynchronous locking.
908 */
909#define FILE_LOCK_DEFERRED 1
910
Jeff Layton7ca76312014-09-01 19:04:48 -0400911/* legacy typedef, should eventually be removed */
Christoph Hellwig17fa3882014-07-13 17:00:37 +0200912typedef void *fl_owner_t;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700913
Jeff Laytona7231a92015-01-16 15:05:56 -0500914struct file_lock;
915
Linus Torvalds1da177e2005-04-16 15:20:36 -0700916struct file_lock_operations {
Linus Torvalds1da177e2005-04-16 15:20:36 -0700917 void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
918 void (*fl_release_private)(struct file_lock *);
919};
920
921struct lock_manager_operations {
J. Bruce Fields8fb47a42011-07-20 20:21:59 -0400922 int (*lm_compare_owner)(struct file_lock *, struct file_lock *);
Jeff Layton3999e492013-06-21 08:58:19 -0400923 unsigned long (*lm_owner_key)(struct file_lock *);
Jeff Laytoncae80b32015-04-03 09:04:04 -0400924 fl_owner_t (*lm_get_owner)(fl_owner_t);
925 void (*lm_put_owner)(fl_owner_t);
J. Bruce Fields8fb47a42011-07-20 20:21:59 -0400926 void (*lm_notify)(struct file_lock *); /* unblock callback */
Joe Perchesd0449b92014-08-22 10:18:42 -0400927 int (*lm_grant)(struct file_lock *, int);
Jeff Layton4d01b7f2014-09-01 15:06:54 -0400928 bool (*lm_break)(struct file_lock *);
Jeff Layton7448cc32015-01-16 15:05:57 -0500929 int (*lm_change)(struct file_lock *, int, struct list_head *);
Jeff Layton1c7dd2f2014-08-22 10:55:47 -0400930 void (*lm_setup)(struct file_lock *, void **);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700931};
932
J. Bruce Fieldsaf558e32007-09-06 12:34:25 -0400933struct lock_manager {
934 struct list_head list;
935};
936
Stanislav Kinsbursky5ccb0062012-07-25 16:57:22 +0400937struct net;
938void locks_start_grace(struct net *, struct lock_manager *);
J. Bruce Fieldsaf558e32007-09-06 12:34:25 -0400939void locks_end_grace(struct lock_manager *);
Stanislav Kinsbursky5ccb0062012-07-25 16:57:22 +0400940int locks_in_grace(struct net *);
J. Bruce Fieldsaf558e32007-09-06 12:34:25 -0400941
Linus Torvalds1da177e2005-04-16 15:20:36 -0700942/* that will die - we need it for nfs_lock_info */
943#include <linux/nfs_fs_i.h>
944
Jeff Layton1cb36012013-06-21 08:58:12 -0400945/*
946 * struct file_lock represents a generic "file lock". It's used to represent
947 * POSIX byte range locks, BSD (flock) locks, and leases. It's important to
948 * note that the same struct is used to represent both a request for a lock and
949 * the lock itself, but the same object is never used for both.
950 *
951 * FIXME: should we create a separate "struct lock_request" to help distinguish
952 * these two uses?
953 *
Jeff Layton8116bf42015-01-21 20:44:01 -0500954 * The varous i_flctx lists are ordered by:
Jeff Layton1cb36012013-06-21 08:58:12 -0400955 *
Jeff Layton8116bf42015-01-21 20:44:01 -0500956 * 1) lock owner
957 * 2) lock range start
958 * 3) lock range end
Jeff Layton1cb36012013-06-21 08:58:12 -0400959 *
960 * Obviously, the last two criteria only matter for POSIX locks.
961 */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700962struct file_lock {
963 struct file_lock *fl_next; /* singly linked list for this inode */
Jeff Layton6dee60f2015-01-16 15:05:54 -0500964 struct list_head fl_list; /* link into file_lock_context */
Jeff Layton139ca042013-06-21 08:58:17 -0400965 struct hlist_node fl_link; /* node in global lists */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700966 struct list_head fl_block; /* circular list of blocked processes */
967 fl_owner_t fl_owner;
J. Bruce Fields710b7212011-07-26 16:28:29 -0400968 unsigned int fl_flags;
Richard Kennedyafc12462008-07-11 17:20:49 -0700969 unsigned char fl_type;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700970 unsigned int fl_pid;
Jeff Layton7012b022013-06-21 08:58:22 -0400971 int fl_link_cpu; /* what cpu's list is this on? */
Vitaliy Gusevab1f1612008-01-17 00:07:08 +0000972 struct pid *fl_nspid;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700973 wait_queue_head_t fl_wait;
974 struct file *fl_file;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700975 loff_t fl_start;
976 loff_t fl_end;
977
978 struct fasync_struct * fl_fasync; /* for lease break notifications */
J. Bruce Fields778fc542011-07-26 18:25:49 -0400979 /* for lease breaks: */
980 unsigned long fl_break_time;
981 unsigned long fl_downgrade_time;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700982
Alexey Dobriyan6aed6282009-09-21 17:01:11 -0700983 const struct file_lock_operations *fl_ops; /* Callbacks for filesystems */
Alexey Dobriyan7b021962009-09-21 17:01:12 -0700984 const struct lock_manager_operations *fl_lmops; /* Callbacks for lockmanagers */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700985 union {
986 struct nfs_lock_info nfs_fl;
Trond Myklebust8d0a8a92005-06-22 17:16:32 +0000987 struct nfs4_lock_info nfs4_fl;
David Howellse8d6c552007-07-15 23:40:12 -0700988 struct {
989 struct list_head link; /* link in AFS vnode's pending_locks list */
990 int state; /* state of grant or error if -ve */
991 } afs;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700992 } fl_u;
993};
994
Jeff Layton4a075e32015-01-16 15:05:54 -0500995struct file_lock_context {
Jeff Layton6109c852015-01-16 15:05:57 -0500996 spinlock_t flc_lock;
Jeff Layton4a075e32015-01-16 15:05:54 -0500997 struct list_head flc_flock;
Jeff Laytonbd61e0a2015-01-16 15:05:55 -0500998 struct list_head flc_posix;
Jeff Layton8634b512015-01-16 15:05:55 -0500999 struct list_head flc_lease;
Jeff Layton4a075e32015-01-16 15:05:54 -05001000};
1001
Linus Torvalds1da177e2005-04-16 15:20:36 -07001002/* The following constant reflects the upper bound of the file/locking space */
1003#ifndef OFFSET_MAX
1004#define INT_LIMIT(x) (~((x)1 << (sizeof(x)*8 - 1)))
1005#define OFFSET_MAX INT_LIMIT(loff_t)
1006#define OFFT_OFFSET_MAX INT_LIMIT(off_t)
1007#endif
1008
Linus Torvalds1da177e2005-04-16 15:20:36 -07001009#include <linux/fcntl.h>
1010
Thomas Petazzonibfcd17a2008-08-06 15:12:22 +02001011extern void send_sigio(struct fown_struct *fown, int fd, int band);
1012
Thomas Petazzonibfcd17a2008-08-06 15:12:22 +02001013#ifdef CONFIG_FILE_LOCKING
Jeff Laytonc1e62b82014-02-03 12:13:09 -05001014extern int fcntl_getlk(struct file *, unsigned int, struct flock __user *);
Peter Staubachc2936212005-07-27 11:45:09 -07001015extern int fcntl_setlk(unsigned int, struct file *, unsigned int,
1016 struct flock __user *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001017
1018#if BITS_PER_LONG == 32
Jeff Laytonc1e62b82014-02-03 12:13:09 -05001019extern int fcntl_getlk64(struct file *, unsigned int, struct flock64 __user *);
Peter Staubachc2936212005-07-27 11:45:09 -07001020extern int fcntl_setlk64(unsigned int, struct file *, unsigned int,
1021 struct flock64 __user *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001022#endif
1023
Linus Torvalds1da177e2005-04-16 15:20:36 -07001024extern int fcntl_setlease(unsigned int fd, struct file *filp, long arg);
1025extern int fcntl_getlease(struct file *filp);
1026
1027/* fs/locks.c */
Jeff Layton4a075e32015-01-16 15:05:54 -05001028void locks_free_lock_context(struct file_lock_context *ctx);
J. Bruce Fields05fa3132010-10-30 17:31:15 -04001029void locks_free_lock(struct file_lock *fl);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001030extern void locks_init_lock(struct file_lock *);
Arnd Bergmannc5b1f0d2010-10-27 15:46:08 +02001031extern struct file_lock * locks_alloc_lock(void);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001032extern void locks_copy_lock(struct file_lock *, struct file_lock *);
Kinglong Mee3fe0fff12014-08-22 10:18:42 -04001033extern void locks_copy_conflock(struct file_lock *, struct file_lock *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001034extern void locks_remove_posix(struct file *, fl_owner_t);
Jeff Layton78ed8a12014-02-03 12:13:08 -05001035extern void locks_remove_file(struct file *);
Felix Blyakhera9e61e22009-03-31 15:12:56 -05001036extern void locks_release_private(struct file_lock *);
J. Bruce Fields6d34ac12007-05-11 16:09:32 -04001037extern void posix_test_lock(struct file *, struct file_lock *);
Marc Eshel150b3932007-01-18 16:15:35 -05001038extern int posix_lock_file(struct file *, struct file_lock *, struct file_lock *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001039extern int posix_lock_file_wait(struct file *, struct file_lock *);
Jeff Laytonf891a292013-06-21 08:58:09 -04001040extern int posix_unblock_lock(struct file_lock *);
J. Bruce Fields3ee17ab2007-02-21 00:58:50 -05001041extern int vfs_test_lock(struct file *, struct file_lock *);
Marc Eshel150b3932007-01-18 16:15:35 -05001042extern int vfs_lock_file(struct file *, unsigned int, struct file_lock *, struct file_lock *);
Marc Eshel9b9d2ab2007-01-18 17:52:58 -05001043extern int vfs_cancel_lock(struct file *filp, struct file_lock *fl);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001044extern int flock_lock_file_wait(struct file *filp, struct file_lock *fl);
J. Bruce Fieldsdf4e8d22012-03-05 13:18:59 -05001045extern int __break_lease(struct inode *inode, unsigned int flags, unsigned int type);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001046extern void lease_get_mtime(struct inode *, struct timespec *time);
Jeff Laytone6f5c782014-08-22 10:40:25 -04001047extern int generic_setlease(struct file *, long, struct file_lock **, void **priv);
1048extern int vfs_setlease(struct file *, long, struct file_lock **, void **);
Jeff Layton7448cc32015-01-16 15:05:57 -05001049extern int lease_modify(struct file_lock *, int, struct list_head *);
Andrey Vagin6c8c9032015-04-16 12:49:38 -07001050struct files_struct;
1051extern void show_fd_locks(struct seq_file *f,
1052 struct file *filp, struct files_struct *files);
Thomas Petazzonibfcd17a2008-08-06 15:12:22 +02001053#else /* !CONFIG_FILE_LOCKING */
Jeff Laytonc1e62b82014-02-03 12:13:09 -05001054static inline int fcntl_getlk(struct file *file, unsigned int cmd,
1055 struct flock __user *user)
Steven Whitehousec2aca5e2009-01-20 10:29:45 +00001056{
1057 return -EINVAL;
1058}
1059
1060static inline int fcntl_setlk(unsigned int fd, struct file *file,
1061 unsigned int cmd, struct flock __user *user)
1062{
1063 return -EACCES;
1064}
1065
Thomas Petazzonibfcd17a2008-08-06 15:12:22 +02001066#if BITS_PER_LONG == 32
Jeff Laytonc1e62b82014-02-03 12:13:09 -05001067static inline int fcntl_getlk64(struct file *file, unsigned int cmd,
1068 struct flock64 __user *user)
Steven Whitehousec2aca5e2009-01-20 10:29:45 +00001069{
1070 return -EINVAL;
1071}
1072
1073static inline int fcntl_setlk64(unsigned int fd, struct file *file,
1074 unsigned int cmd, struct flock64 __user *user)
1075{
1076 return -EACCES;
1077}
Thomas Petazzonibfcd17a2008-08-06 15:12:22 +02001078#endif
Steven Whitehousec2aca5e2009-01-20 10:29:45 +00001079static inline int fcntl_setlease(unsigned int fd, struct file *filp, long arg)
1080{
Jeff Layton1b2b32d2014-09-24 08:38:44 -04001081 return -EINVAL;
Steven Whitehousec2aca5e2009-01-20 10:29:45 +00001082}
1083
1084static inline int fcntl_getlease(struct file *filp)
1085{
Jeff Layton1b2b32d2014-09-24 08:38:44 -04001086 return F_UNLCK;
Steven Whitehousec2aca5e2009-01-20 10:29:45 +00001087}
1088
Jeff Layton4a075e32015-01-16 15:05:54 -05001089static inline void
1090locks_free_lock_context(struct file_lock_context *ctx)
1091{
1092}
1093
Steven Whitehousec2aca5e2009-01-20 10:29:45 +00001094static inline void locks_init_lock(struct file_lock *fl)
1095{
1096 return;
1097}
1098
Kinglong Mee3fe0fff12014-08-22 10:18:42 -04001099static inline void locks_copy_conflock(struct file_lock *new, struct file_lock *fl)
Steven Whitehousec2aca5e2009-01-20 10:29:45 +00001100{
1101 return;
1102}
1103
1104static inline void locks_copy_lock(struct file_lock *new, struct file_lock *fl)
1105{
1106 return;
1107}
1108
1109static inline void locks_remove_posix(struct file *filp, fl_owner_t owner)
1110{
1111 return;
1112}
1113
Jeff Layton78ed8a12014-02-03 12:13:08 -05001114static inline void locks_remove_file(struct file *filp)
Steven Whitehousec2aca5e2009-01-20 10:29:45 +00001115{
1116 return;
1117}
1118
1119static inline void posix_test_lock(struct file *filp, struct file_lock *fl)
1120{
1121 return;
1122}
1123
1124static inline int posix_lock_file(struct file *filp, struct file_lock *fl,
1125 struct file_lock *conflock)
1126{
1127 return -ENOLCK;
1128}
1129
1130static inline int posix_lock_file_wait(struct file *filp, struct file_lock *fl)
1131{
1132 return -ENOLCK;
1133}
1134
Jeff Laytonf891a292013-06-21 08:58:09 -04001135static inline int posix_unblock_lock(struct file_lock *waiter)
Steven Whitehousec2aca5e2009-01-20 10:29:45 +00001136{
1137 return -ENOENT;
1138}
1139
1140static inline int vfs_test_lock(struct file *filp, struct file_lock *fl)
1141{
1142 return 0;
1143}
1144
1145static inline int vfs_lock_file(struct file *filp, unsigned int cmd,
1146 struct file_lock *fl, struct file_lock *conf)
1147{
1148 return -ENOLCK;
1149}
1150
1151static inline int vfs_cancel_lock(struct file *filp, struct file_lock *fl)
1152{
1153 return 0;
1154}
1155
1156static inline int flock_lock_file_wait(struct file *filp,
1157 struct file_lock *request)
1158{
1159 return -ENOLCK;
1160}
1161
J. Bruce Fieldsdf4e8d22012-03-05 13:18:59 -05001162static inline int __break_lease(struct inode *inode, unsigned int mode, unsigned int type)
Steven Whitehousec2aca5e2009-01-20 10:29:45 +00001163{
1164 return 0;
1165}
1166
1167static inline void lease_get_mtime(struct inode *inode, struct timespec *time)
1168{
1169 return;
1170}
1171
1172static inline int generic_setlease(struct file *filp, long arg,
Jeff Laytone6f5c782014-08-22 10:40:25 -04001173 struct file_lock **flp, void **priv)
Steven Whitehousec2aca5e2009-01-20 10:29:45 +00001174{
1175 return -EINVAL;
1176}
1177
1178static inline int vfs_setlease(struct file *filp, long arg,
Jeff Laytone6f5c782014-08-22 10:40:25 -04001179 struct file_lock **lease, void **priv)
Steven Whitehousec2aca5e2009-01-20 10:29:45 +00001180{
1181 return -EINVAL;
1182}
1183
Jeff Layton7448cc32015-01-16 15:05:57 -05001184static inline int lease_modify(struct file_lock *fl, int arg,
Jeff Laytonc45198e2014-09-01 07:12:07 -04001185 struct list_head *dispose)
Steven Whitehousec2aca5e2009-01-20 10:29:45 +00001186{
1187 return -EINVAL;
1188}
Andrey Vagin6c8c9032015-04-16 12:49:38 -07001189
1190struct files_struct;
1191static inline void show_fd_locks(struct seq_file *f,
1192 struct file *filp, struct files_struct *files) {}
Thomas Petazzonibfcd17a2008-08-06 15:12:22 +02001193#endif /* !CONFIG_FILE_LOCKING */
1194
Linus Torvalds1da177e2005-04-16 15:20:36 -07001195
1196struct fasync_struct {
Eric Dumazet989a2972010-04-14 09:55:35 +00001197 spinlock_t fa_lock;
1198 int magic;
1199 int fa_fd;
1200 struct fasync_struct *fa_next; /* singly linked list */
1201 struct file *fa_file;
1202 struct rcu_head fa_rcu;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001203};
1204
1205#define FASYNC_MAGIC 0x4601
1206
1207/* SMP safe fasync helpers: */
1208extern int fasync_helper(int, struct file *, int, struct fasync_struct **);
Linus Torvaldsf7347ce2010-10-27 12:38:12 -04001209extern struct fasync_struct *fasync_insert_entry(int, struct file *, struct fasync_struct **, struct fasync_struct *);
1210extern int fasync_remove_entry(struct file *, struct fasync_struct **);
1211extern struct fasync_struct *fasync_alloc(void);
1212extern void fasync_free(struct fasync_struct *);
1213
Linus Torvalds1da177e2005-04-16 15:20:36 -07001214/* can be called from interrupts */
1215extern void kill_fasync(struct fasync_struct **, int, int);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001216
Jeff Laytone0b93ed2014-08-22 11:27:32 -04001217extern void __f_setown(struct file *filp, struct pid *, enum pid_type, int force);
1218extern void f_setown(struct file *filp, unsigned long arg, int force);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001219extern void f_delown(struct file *filp);
Eric W. Biederman609d7fa2006-10-02 02:17:15 -07001220extern pid_t f_getown(struct file *filp);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001221extern int send_sigurg(struct fown_struct *fown);
1222
Jan Kara5accdf82012-06-12 16:20:34 +02001223struct mm_struct;
1224
Linus Torvalds1da177e2005-04-16 15:20:36 -07001225/*
1226 * Umount options
1227 */
1228
1229#define MNT_FORCE 0x00000001 /* Attempt to forcibily umount */
1230#define MNT_DETACH 0x00000002 /* Just detach from the tree */
1231#define MNT_EXPIRE 0x00000004 /* Mark for expiry */
Miklos Szeredidb1f05b2010-02-10 12:15:53 +01001232#define UMOUNT_NOFOLLOW 0x00000008 /* Don't follow symlink on umount */
1233#define UMOUNT_UNUSED 0x80000000 /* Flag guaranteed to be unused */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001234
Linus Torvalds1da177e2005-04-16 15:20:36 -07001235
Jan Kara5accdf82012-06-12 16:20:34 +02001236/* Possible states of 'frozen' field */
1237enum {
1238 SB_UNFROZEN = 0, /* FS is unfrozen */
1239 SB_FREEZE_WRITE = 1, /* Writes, dir ops, ioctls frozen */
Jan Kara5accdf82012-06-12 16:20:34 +02001240 SB_FREEZE_PAGEFAULT = 2, /* Page faults stopped as well */
1241 SB_FREEZE_FS = 3, /* For internal FS use (e.g. to stop
1242 * internal threads if needed) */
1243 SB_FREEZE_COMPLETE = 4, /* ->freeze_fs finished successfully */
1244};
1245
1246#define SB_FREEZE_LEVELS (SB_FREEZE_COMPLETE - 1)
1247
1248struct sb_writers {
1249 /* Counters for counting writers at each level */
1250 struct percpu_counter counter[SB_FREEZE_LEVELS];
1251 wait_queue_head_t wait; /* queue for waiting for
1252 writers / faults to finish */
1253 int frozen; /* Is sb frozen? */
1254 wait_queue_head_t wait_unfrozen; /* queue for waiting for
1255 sb to be thawed */
1256#ifdef CONFIG_DEBUG_LOCK_ALLOC
1257 struct lockdep_map lock_map[SB_FREEZE_LEVELS];
1258#endif
1259};
1260
Linus Torvalds1da177e2005-04-16 15:20:36 -07001261struct super_block {
1262 struct list_head s_list; /* Keep this first */
1263 dev_t s_dev; /* search index; _not_ kdev_t */
Richard Kennedy270ba5f2010-01-26 14:12:43 +00001264 unsigned char s_blocksize_bits;
1265 unsigned long s_blocksize;
Jeff Layton42cb56a2009-09-18 13:05:53 -07001266 loff_t s_maxbytes; /* Max file size */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001267 struct file_system_type *s_type;
Josef 'Jeff' Sipekee9b6d62007-02-12 00:55:41 -08001268 const struct super_operations *s_op;
Alexey Dobriyan61e225d2009-09-21 17:01:08 -07001269 const struct dquot_operations *dq_op;
Alexey Dobriyan0d54b212009-09-21 17:01:09 -07001270 const struct quotactl_ops *s_qcop;
Christoph Hellwig39655162007-10-21 16:42:17 -07001271 const struct export_operations *s_export_op;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001272 unsigned long s_flags;
1273 unsigned long s_magic;
1274 struct dentry *s_root;
1275 struct rw_semaphore s_umount;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001276 int s_count;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001277 atomic_t s_active;
Alexey Dobriyan50462062006-09-29 02:00:01 -07001278#ifdef CONFIG_SECURITY
Linus Torvalds1da177e2005-04-16 15:20:36 -07001279 void *s_security;
Alexey Dobriyan50462062006-09-29 02:00:01 -07001280#endif
Stephen Hemmingerbb435452010-05-13 17:53:14 -07001281 const struct xattr_handler **s_xattr;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001282
1283 struct list_head s_inodes; /* all inodes */
Nick Pigginceb5bdc2011-01-07 17:50:05 +11001284 struct hlist_bl_head s_anon; /* anonymous dentries for (nfs) exporting */
Miklos Szeredi39f7c4d2011-11-21 12:11:30 +01001285 struct list_head s_mounts; /* list of mounts; _not_ for fs use */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001286 struct block_device *s_bdev;
Jens Axboe32a88aa2009-09-16 15:02:33 +02001287 struct backing_dev_info *s_bdi;
David Howellsacaebfd2007-05-10 22:51:50 -07001288 struct mtd_info *s_mtd;
Al Viroa5166162011-12-12 22:53:00 -05001289 struct hlist_node s_instances;
Jan Kara2c5f6482014-09-30 10:43:09 +02001290 unsigned int s_quota_types; /* Bitmask of supported quota types */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001291 struct quota_info s_dquot; /* Diskquota specific options */
1292
Jan Kara5accdf82012-06-12 16:20:34 +02001293 struct sb_writers s_writers;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001294
1295 char s_id[32]; /* Informational name */
Linus Torvalds0bba0162011-04-12 15:21:04 -07001296 u8 s_uuid[16]; /* UUID */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001297
1298 void *s_fs_info; /* Filesystem private info */
Al Viro8de52772012-02-06 12:45:27 -05001299 unsigned int s_max_links;
Al Viro30c40d22008-02-22 19:50:45 -05001300 fmode_t s_mode;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001301
Richard Kennedy270ba5f2010-01-26 14:12:43 +00001302 /* Granularity of c/m/atime in ns.
1303 Cannot be worse than a second */
1304 u32 s_time_gran;
1305
Linus Torvalds1da177e2005-04-16 15:20:36 -07001306 /*
1307 * The next field is for VFS *only*. No filesystems have any business
1308 * even looking at it. You had been warned.
1309 */
Arjan van de Vena11f3a02006-03-23 03:00:33 -08001310 struct mutex s_vfs_rename_mutex; /* Kludge */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001311
Miklos Szeredi79c0b2d2007-05-08 00:25:43 -07001312 /*
1313 * Filesystem subtype. If non-empty the filesystem type field
1314 * in /proc/mounts will be "type.subtype"
1315 */
1316 char *s_subtype;
Miklos Szeredib3b304a2008-02-08 04:21:35 -08001317
1318 /*
1319 * Saved mount options for lazy filesystems using
1320 * generic_show_options()
1321 */
Arnd Bergmann4d2deb42010-02-24 20:01:56 +01001322 char __rcu *s_options;
Al Viroc8aebb02010-12-18 10:22:30 -05001323 const struct dentry_operations *s_d_op; /* default d_op for dentries */
Dan Magenheimer9fdfdcf2011-05-26 10:01:19 -06001324
1325 /*
1326 * Saved pool identifier for cleancache (-1 means none)
1327 */
1328 int cleancache_poolid;
Dave Chinnerb0d40c92011-07-08 14:14:42 +10001329
1330 struct shrinker s_shrink; /* per-sb shrinker handle */
Miklos Szeredi4ed5e822011-11-21 12:11:31 +01001331
Miklos Szeredi7ada4db2011-11-21 12:11:32 +01001332 /* Number of inodes with nlink == 0 but still referenced */
1333 atomic_long_t s_remove_count;
1334
Miklos Szeredi4ed5e822011-11-21 12:11:31 +01001335 /* Being remounted read-only */
1336 int s_readonly_remount;
Christoph Hellwig7b7a8662013-09-04 15:04:39 +02001337
1338 /* AIO completions deferred from interrupt context */
1339 struct workqueue_struct *s_dio_done_wq;
Al Viro215752f2014-08-07 06:23:41 -04001340 struct hlist_head s_pins;
Dave Chinnerf6041562013-08-28 10:18:00 +10001341
1342 /*
1343 * Keep the lru lists last in the structure so they always sit on their
1344 * own individual cachelines.
1345 */
1346 struct list_lru s_dentry_lru ____cacheline_aligned_in_smp;
1347 struct list_lru s_inode_lru ____cacheline_aligned_in_smp;
Al Viroe2fec7c2013-10-04 17:06:56 -04001348 struct rcu_head rcu;
Miklos Szeredi69c433e2014-10-24 00:14:39 +02001349
1350 /*
1351 * Indicates how deep in a filesystem stack this SB is
1352 */
1353 int s_stack_depth;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001354};
1355
1356extern struct timespec current_fs_time(struct super_block *sb);
1357
1358/*
1359 * Snapshotting support.
1360 */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001361
Jan Kara5accdf82012-06-12 16:20:34 +02001362void __sb_end_write(struct super_block *sb, int level);
1363int __sb_start_write(struct super_block *sb, int level, bool wait);
1364
1365/**
1366 * sb_end_write - drop write access to a superblock
1367 * @sb: the super we wrote to
1368 *
1369 * Decrement number of writers to the filesystem. Wake up possible waiters
1370 * wanting to freeze the filesystem.
1371 */
1372static inline void sb_end_write(struct super_block *sb)
1373{
1374 __sb_end_write(sb, SB_FREEZE_WRITE);
1375}
1376
1377/**
1378 * sb_end_pagefault - drop write access to a superblock from a page fault
1379 * @sb: the super we wrote to
1380 *
1381 * Decrement number of processes handling write page fault to the filesystem.
1382 * Wake up possible waiters wanting to freeze the filesystem.
1383 */
1384static inline void sb_end_pagefault(struct super_block *sb)
1385{
1386 __sb_end_write(sb, SB_FREEZE_PAGEFAULT);
1387}
1388
1389/**
1390 * sb_end_intwrite - drop write access to a superblock for internal fs purposes
1391 * @sb: the super we wrote to
1392 *
1393 * Decrement fs-internal number of writers to the filesystem. Wake up possible
1394 * waiters wanting to freeze the filesystem.
1395 */
1396static inline void sb_end_intwrite(struct super_block *sb)
1397{
1398 __sb_end_write(sb, SB_FREEZE_FS);
1399}
1400
1401/**
1402 * sb_start_write - get write access to a superblock
1403 * @sb: the super we write to
1404 *
1405 * When a process wants to write data or metadata to a file system (i.e. dirty
1406 * a page or an inode), it should embed the operation in a sb_start_write() -
1407 * sb_end_write() pair to get exclusion against file system freezing. This
1408 * function increments number of writers preventing freezing. If the file
1409 * system is already frozen, the function waits until the file system is
1410 * thawed.
1411 *
1412 * Since freeze protection behaves as a lock, users have to preserve
1413 * ordering of freeze protection and other filesystem locks. Generally,
1414 * freeze protection should be the outermost lock. In particular, we have:
1415 *
1416 * sb_start_write
1417 * -> i_mutex (write path, truncate, directory ops, ...)
1418 * -> s_umount (freeze_super, thaw_super)
1419 */
1420static inline void sb_start_write(struct super_block *sb)
1421{
1422 __sb_start_write(sb, SB_FREEZE_WRITE, true);
1423}
1424
1425static inline int sb_start_write_trylock(struct super_block *sb)
1426{
1427 return __sb_start_write(sb, SB_FREEZE_WRITE, false);
1428}
1429
1430/**
1431 * sb_start_pagefault - get write access to a superblock from a page fault
1432 * @sb: the super we write to
1433 *
1434 * When a process starts handling write page fault, it should embed the
1435 * operation into sb_start_pagefault() - sb_end_pagefault() pair to get
1436 * exclusion against file system freezing. This is needed since the page fault
1437 * is going to dirty a page. This function increments number of running page
1438 * faults preventing freezing. If the file system is already frozen, the
1439 * function waits until the file system is thawed.
1440 *
1441 * Since page fault freeze protection behaves as a lock, users have to preserve
1442 * ordering of freeze protection and other filesystem locks. It is advised to
1443 * put sb_start_pagefault() close to mmap_sem in lock ordering. Page fault
1444 * handling code implies lock dependency:
1445 *
1446 * mmap_sem
1447 * -> sb_start_pagefault
1448 */
1449static inline void sb_start_pagefault(struct super_block *sb)
1450{
1451 __sb_start_write(sb, SB_FREEZE_PAGEFAULT, true);
1452}
1453
1454/*
1455 * sb_start_intwrite - get write access to a superblock for internal fs purposes
1456 * @sb: the super we write to
1457 *
1458 * This is the third level of protection against filesystem freezing. It is
1459 * free for use by a filesystem. The only requirement is that it must rank
1460 * below sb_start_pagefault.
1461 *
1462 * For example filesystem can call sb_start_intwrite() when starting a
1463 * transaction which somewhat eases handling of freezing for internal sources
1464 * of filesystem changes (internal fs threads, discarding preallocation on file
1465 * close, etc.).
1466 */
1467static inline void sb_start_intwrite(struct super_block *sb)
1468{
1469 __sb_start_write(sb, SB_FREEZE_FS, true);
1470}
1471
Linus Torvalds1da177e2005-04-16 15:20:36 -07001472
Serge E. Hallyn2e149672011-03-23 16:43:26 -07001473extern bool inode_owner_or_capable(const struct inode *inode);
Satyam Sharma3bd858a2007-07-17 15:00:08 +05301474
Linus Torvalds1da177e2005-04-16 15:20:36 -07001475/*
1476 * VFS helper functions..
1477 */
Al Viro312b63f2012-06-10 18:09:36 -04001478extern int vfs_create(struct inode *, struct dentry *, umode_t, bool);
Al Viro18bb1db2011-07-26 01:41:39 -04001479extern int vfs_mkdir(struct inode *, struct dentry *, umode_t);
Al Viro1a67aaf2011-07-26 01:52:52 -04001480extern int vfs_mknod(struct inode *, struct dentry *, umode_t, dev_t);
Miklos Szeredidb2e7472008-06-24 16:50:16 +02001481extern int vfs_symlink(struct inode *, struct dentry *, const char *);
J. Bruce Fields146a8592011-09-20 17:14:31 -04001482extern int vfs_link(struct dentry *, struct inode *, struct dentry *, struct inode **);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001483extern int vfs_rmdir(struct inode *, struct dentry *);
J. Bruce Fieldsb21996e2011-09-20 09:14:34 -04001484extern int vfs_unlink(struct inode *, struct dentry *, struct inode **);
Miklos Szeredi520c8b12014-04-01 17:08:42 +02001485extern int vfs_rename(struct inode *, struct dentry *, struct inode *, struct dentry *, struct inode **, unsigned int);
Miklos Szeredi787fb6b2014-10-24 00:14:36 +02001486extern int vfs_whiteout(struct inode *, struct dentry *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001487
1488/*
1489 * VFS dentry helper functions.
1490 */
1491extern void dentry_unhash(struct dentry *dentry);
1492
1493/*
Christoph Hellwig8c744fb2005-11-08 21:35:04 -08001494 * VFS file helper functions.
1495 */
Dmitry Monakhova1bd1202010-03-04 17:29:14 +03001496extern void inode_init_owner(struct inode *inode, const struct inode *dir,
Al Viro62bb1092011-07-24 23:20:18 -04001497 umode_t mode);
Christoph Hellwig8c744fb2005-11-08 21:35:04 -08001498/*
Mark Fashehc4b929b2008-10-08 19:44:18 -04001499 * VFS FS_IOC_FIEMAP helper definitions.
1500 */
1501struct fiemap_extent_info {
1502 unsigned int fi_flags; /* Flags as passed from user */
1503 unsigned int fi_extents_mapped; /* Number of mapped extents */
1504 unsigned int fi_extents_max; /* Size of fiemap_extent array */
Namhyung Kimecf56322011-01-16 23:28:17 +09001505 struct fiemap_extent __user *fi_extents_start; /* Start of
1506 fiemap_extent array */
Mark Fashehc4b929b2008-10-08 19:44:18 -04001507};
1508int fiemap_fill_next_extent(struct fiemap_extent_info *info, u64 logical,
1509 u64 phys, u64 len, u32 flags);
1510int fiemap_check_flags(struct fiemap_extent_info *fieinfo, u32 fs_flags);
1511
1512/*
Linus Torvalds1da177e2005-04-16 15:20:36 -07001513 * File types
1514 *
1515 * NOTE! These match bits 12..15 of stat.st_mode
1516 * (ie "(i_mode >> 12) & 15").
1517 */
1518#define DT_UNKNOWN 0
1519#define DT_FIFO 1
1520#define DT_CHR 2
1521#define DT_DIR 4
1522#define DT_BLK 6
1523#define DT_REG 8
1524#define DT_LNK 10
1525#define DT_SOCK 12
1526#define DT_WHT 14
1527
Linus Torvalds1da177e2005-04-16 15:20:36 -07001528/*
1529 * This is the "filldir" function type, used by readdir() to let
1530 * the kernel specify what kind of dirent layout it wants to have.
1531 * This allows the kernel to read directories into kernel space or
1532 * to have different dirent layouts depending on the binary type.
1533 */
Miklos Szerediac7576f2014-10-30 17:37:34 +01001534struct dir_context;
1535typedef int (*filldir_t)(struct dir_context *, const char *, int, loff_t, u64,
1536 unsigned);
1537
Al Viro5c0ba4e2013-05-15 13:52:59 -04001538struct dir_context {
Al Viroac6614b2013-05-22 22:22:04 -04001539 const filldir_t actor;
Al Virobb6f6192013-05-15 18:49:12 -04001540 loff_t pos;
Al Viro5c0ba4e2013-05-15 13:52:59 -04001541};
Al Virobb6f6192013-05-15 18:49:12 -04001542
Al Viro08f85852007-10-08 13:26:20 -04001543struct block_device_operations;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001544
Linus Torvalds1da177e2005-04-16 15:20:36 -07001545/* These macros are for out of kernel modules to test that
1546 * the kernel supports the unlocked_ioctl and compat_ioctl
1547 * fields in struct file_operations. */
1548#define HAVE_COMPAT_IOCTL 1
1549#define HAVE_UNLOCKED_IOCTL 1
1550
Christoph Hellwigb4caecd2015-01-14 10:42:32 +01001551/*
1552 * These flags let !MMU mmap() govern direct device mapping vs immediate
1553 * copying more easily for MAP_PRIVATE, especially for ROM filesystems.
1554 *
1555 * NOMMU_MAP_COPY: Copy can be mapped (MAP_PRIVATE)
1556 * NOMMU_MAP_DIRECT: Can be mapped directly (MAP_SHARED)
1557 * NOMMU_MAP_READ: Can be mapped for reading
1558 * NOMMU_MAP_WRITE: Can be mapped for writing
1559 * NOMMU_MAP_EXEC: Can be mapped for execution
1560 */
1561#define NOMMU_MAP_COPY 0x00000001
1562#define NOMMU_MAP_DIRECT 0x00000008
1563#define NOMMU_MAP_READ VM_MAYREAD
1564#define NOMMU_MAP_WRITE VM_MAYWRITE
1565#define NOMMU_MAP_EXEC VM_MAYEXEC
1566
1567#define NOMMU_VMFLAGS \
1568 (NOMMU_MAP_READ | NOMMU_MAP_WRITE | NOMMU_MAP_EXEC)
1569
1570
Al Viro293bc982014-02-11 18:37:41 -05001571struct iov_iter;
1572
Linus Torvalds1da177e2005-04-16 15:20:36 -07001573struct file_operations {
1574 struct module *owner;
1575 loff_t (*llseek) (struct file *, loff_t, int);
1576 ssize_t (*read) (struct file *, char __user *, size_t, loff_t *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001577 ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *);
Al Viro293bc982014-02-11 18:37:41 -05001578 ssize_t (*read_iter) (struct kiocb *, struct iov_iter *);
1579 ssize_t (*write_iter) (struct kiocb *, struct iov_iter *);
Al Virobb6f6192013-05-15 18:49:12 -04001580 int (*iterate) (struct file *, struct dir_context *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001581 unsigned int (*poll) (struct file *, struct poll_table_struct *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001582 long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long);
1583 long (*compat_ioctl) (struct file *, unsigned int, unsigned long);
1584 int (*mmap) (struct file *, struct vm_area_struct *);
Al Virob2edffd2015-04-06 17:48:54 -04001585 int (*mremap)(struct file *, struct vm_area_struct *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001586 int (*open) (struct inode *, struct file *);
Miklos Szeredi75e1fcc2006-06-23 02:05:12 -07001587 int (*flush) (struct file *, fl_owner_t id);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001588 int (*release) (struct inode *, struct file *);
Josef Bacik02c24a82011-07-16 20:44:56 -04001589 int (*fsync) (struct file *, loff_t, loff_t, int datasync);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001590 int (*aio_fsync) (struct kiocb *, int datasync);
1591 int (*fasync) (int, struct file *, int);
1592 int (*lock) (struct file *, int, struct file_lock *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001593 ssize_t (*sendpage) (struct file *, struct page *, int, size_t, loff_t *, int);
1594 unsigned long (*get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
1595 int (*check_flags)(int);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001596 int (*flock) (struct file *, int, struct file_lock *);
Jens Axboecbb7e572006-04-11 14:57:50 +02001597 ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, loff_t *, size_t, unsigned int);
1598 ssize_t (*splice_read)(struct file *, loff_t *, struct pipe_inode_info *, size_t, unsigned int);
Jeff Laytone6f5c782014-08-22 10:40:25 -04001599 int (*setlease)(struct file *, long, struct file_lock **, void **);
Christoph Hellwig2fe17c12011-01-14 13:07:43 +01001600 long (*fallocate)(struct file *file, int mode, loff_t offset,
1601 loff_t len);
Joe Perchesa3816ab2014-09-29 16:08:25 -07001602 void (*show_fdinfo)(struct seq_file *m, struct file *f);
Christoph Hellwigb4caecd2015-01-14 10:42:32 +01001603#ifndef CONFIG_MMU
1604 unsigned (*mmap_capabilities)(struct file *);
1605#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -07001606};
1607
1608struct inode_operations {
Al Viro00cd8dd2012-06-10 17:13:09 -04001609 struct dentry * (*lookup) (struct inode *,struct dentry *, unsigned int);
Nick Piggin44a7d7a2011-01-07 17:49:56 +11001610 void * (*follow_link) (struct dentry *, struct nameidata *);
Al Viro10556cb2011-06-20 19:28:19 -04001611 int (*permission) (struct inode *, int);
Christoph Hellwig4e34e712011-07-23 17:37:31 +02001612 struct posix_acl * (*get_acl)(struct inode *, int);
Nick Piggin44a7d7a2011-01-07 17:49:56 +11001613
1614 int (*readlink) (struct dentry *, char __user *,int);
1615 void (*put_link) (struct dentry *, struct nameidata *, void *);
1616
Al Viroebfc3b42012-06-10 18:05:36 -04001617 int (*create) (struct inode *,struct dentry *, umode_t, bool);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001618 int (*link) (struct dentry *,struct inode *,struct dentry *);
1619 int (*unlink) (struct inode *,struct dentry *);
1620 int (*symlink) (struct inode *,struct dentry *,const char *);
Al Viro18bb1db2011-07-26 01:41:39 -04001621 int (*mkdir) (struct inode *,struct dentry *,umode_t);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001622 int (*rmdir) (struct inode *,struct dentry *);
Al Viro1a67aaf2011-07-26 01:52:52 -04001623 int (*mknod) (struct inode *,struct dentry *,umode_t,dev_t);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001624 int (*rename) (struct inode *, struct dentry *,
1625 struct inode *, struct dentry *);
Miklos Szeredi520c8b12014-04-01 17:08:42 +02001626 int (*rename2) (struct inode *, struct dentry *,
1627 struct inode *, struct dentry *, unsigned int);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001628 int (*setattr) (struct dentry *, struct iattr *);
1629 int (*getattr) (struct vfsmount *mnt, struct dentry *, struct kstat *);
1630 int (*setxattr) (struct dentry *, const char *,const void *,size_t,int);
1631 ssize_t (*getxattr) (struct dentry *, const char *, void *, size_t);
1632 ssize_t (*listxattr) (struct dentry *, char *, size_t);
1633 int (*removexattr) (struct dentry *, const char *);
Mark Fashehc4b929b2008-10-08 19:44:18 -04001634 int (*fiemap)(struct inode *, struct fiemap_extent_info *, u64 start,
1635 u64 len);
Josef Bacikc3b2da32012-03-26 09:59:21 -04001636 int (*update_time)(struct inode *, struct timespec *, int);
Al Virod9585272012-06-22 12:39:14 +04001637 int (*atomic_open)(struct inode *, struct dentry *,
Al Viro30d90492012-06-22 12:40:19 +04001638 struct file *, unsigned open_flag,
Al Virod9585272012-06-22 12:39:14 +04001639 umode_t create_mode, int *opened);
Al Viro60545d02013-06-07 01:20:27 -04001640 int (*tmpfile) (struct inode *, struct dentry *, umode_t);
Christoph Hellwig893d46e2013-12-20 05:16:39 -08001641 int (*set_acl)(struct inode *, struct posix_acl *, int);
Miklos Szeredi4aa7c632014-10-24 00:14:35 +02001642
1643 /* WARNING: probably going away soon, do not use! */
1644 int (*dentry_open)(struct dentry *, struct file *, const struct cred *);
Nick Piggin44a7d7a2011-01-07 17:49:56 +11001645} ____cacheline_aligned;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001646
Badari Pulavartyeed4e512006-09-30 23:28:49 -07001647ssize_t rw_copy_check_uvector(int type, const struct iovec __user * uvector,
Christopher Yeohfcf63402011-10-31 17:06:39 -07001648 unsigned long nr_segs, unsigned long fast_segs,
1649 struct iovec *fast_pointer,
Christopher Yeohac34ebb2012-05-31 16:26:42 -07001650 struct iovec **ret_pointer);
Badari Pulavartyeed4e512006-09-30 23:28:49 -07001651
Dmitry Kasatkin6fb50322014-11-05 17:01:17 +02001652extern ssize_t __vfs_read(struct file *, char __user *, size_t, loff_t *);
Al Viro493c84c2015-04-03 15:06:43 -04001653extern ssize_t __vfs_write(struct file *, const char __user *, size_t, loff_t *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001654extern ssize_t vfs_read(struct file *, char __user *, size_t, loff_t *);
1655extern ssize_t vfs_write(struct file *, const char __user *, size_t, loff_t *);
1656extern ssize_t vfs_readv(struct file *, const struct iovec __user *,
1657 unsigned long, loff_t *);
1658extern ssize_t vfs_writev(struct file *, const struct iovec __user *,
1659 unsigned long, loff_t *);
1660
Linus Torvalds1da177e2005-04-16 15:20:36 -07001661struct super_operations {
1662 struct inode *(*alloc_inode)(struct super_block *sb);
1663 void (*destroy_inode)(struct inode *);
1664
Christoph Hellwigaa385722011-05-27 06:53:02 -04001665 void (*dirty_inode) (struct inode *, int flags);
Christoph Hellwiga9185b42010-03-05 09:21:37 +01001666 int (*write_inode) (struct inode *, struct writeback_control *wbc);
Al Viro45321ac2010-06-07 13:43:19 -04001667 int (*drop_inode) (struct inode *);
Al Virobe7ce412010-06-04 19:40:39 -04001668 void (*evict_inode) (struct inode *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001669 void (*put_super) (struct super_block *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001670 int (*sync_fs)(struct super_block *sb, int wait);
Benjamin Marzinski48b6bca2014-11-13 20:42:03 -06001671 int (*freeze_super) (struct super_block *);
Takashi Satoc4be0c12009-01-09 16:40:58 -08001672 int (*freeze_fs) (struct super_block *);
Benjamin Marzinski48b6bca2014-11-13 20:42:03 -06001673 int (*thaw_super) (struct super_block *);
Takashi Satoc4be0c12009-01-09 16:40:58 -08001674 int (*unfreeze_fs) (struct super_block *);
David Howells726c3342006-06-23 02:02:58 -07001675 int (*statfs) (struct dentry *, struct kstatfs *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001676 int (*remount_fs) (struct super_block *, int *, char *);
Al Viro42faad92008-04-24 07:21:56 -04001677 void (*umount_begin) (struct super_block *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001678
Al Viro34c80b12011-12-08 21:32:45 -05001679 int (*show_options)(struct seq_file *, struct dentry *);
Al Virod861c632011-12-08 21:32:45 -05001680 int (*show_devname)(struct seq_file *, struct dentry *);
Al Viroa6322de2011-12-08 21:37:57 -05001681 int (*show_path)(struct seq_file *, struct dentry *);
Al Viro64132372011-12-08 20:51:13 -05001682 int (*show_stats)(struct seq_file *, struct dentry *);
Alexey Dobriyan0e51a722006-09-29 01:59:56 -07001683#ifdef CONFIG_QUOTA
Linus Torvalds1da177e2005-04-16 15:20:36 -07001684 ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t);
1685 ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t);
Jan Kara2d0fa462014-09-25 16:36:14 +02001686 struct dquot **(*get_dquots)(struct inode *);
Alexey Dobriyan0e51a722006-09-29 01:59:56 -07001687#endif
Theodore Ts'o87d8fe12009-01-03 09:47:09 -05001688 int (*bdev_try_to_free_page)(struct super_block*, struct page*, gfp_t);
Vladimir Davydov4101b622015-02-12 14:58:51 -08001689 long (*nr_cached_objects)(struct super_block *,
1690 struct shrink_control *);
1691 long (*free_cached_objects)(struct super_block *,
1692 struct shrink_control *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001693};
1694
Joern Engel1c0eeaf2007-10-16 23:30:44 -07001695/*
David Howellsbbc10962012-10-15 16:40:35 +01001696 * Inode flags - they have no relation to superblock flags now
1697 */
1698#define S_SYNC 1 /* Writes are synced at once */
1699#define S_NOATIME 2 /* Do not update access times */
1700#define S_APPEND 4 /* Append-only file */
1701#define S_IMMUTABLE 8 /* Immutable file */
1702#define S_DEAD 16 /* removed, but still open directory */
1703#define S_NOQUOTA 32 /* Inode is not counted to quota */
1704#define S_DIRSYNC 64 /* Directory modifications are synchronous */
1705#define S_NOCMTIME 128 /* Do not update file c/mtime */
1706#define S_SWAPFILE 256 /* Do not truncate: swapon got its bmaps */
1707#define S_PRIVATE 512 /* Inode is fs-internal */
1708#define S_IMA 1024 /* Inode has an associated IMA struct */
1709#define S_AUTOMOUNT 2048 /* Automount/referral quasi-directory */
1710#define S_NOSEC 4096 /* no suid or xattr security attributes */
Matthew Wilcox6cd176a2015-02-16 15:59:25 -08001711#ifdef CONFIG_FS_DAX
Matthew Wilcoxfbbbad42015-02-16 15:58:53 -08001712#define S_DAX 8192 /* Direct Access, avoiding the page cache */
1713#else
1714#define S_DAX 0 /* Make all the DAX code disappear */
1715#endif
David Howellsbbc10962012-10-15 16:40:35 +01001716
1717/*
1718 * Note that nosuid etc flags are inode-specific: setting some file-system
1719 * flags just means all the inodes inherit those flags by default. It might be
1720 * possible to override it selectively if you really wanted to with some
1721 * ioctl() that is not currently implemented.
1722 *
1723 * Exception: MS_RDONLY is always applied to the entire file system.
1724 *
1725 * Unfortunately, it is possible to change a filesystems flags with it mounted
1726 * with files in use. This means that all of the inodes will not have their
1727 * i_flags updated. Hence, i_flags no longer inherit the superblock mount
1728 * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org
1729 */
1730#define __IS_FLG(inode, flg) ((inode)->i_sb->s_flags & (flg))
1731
1732#define IS_RDONLY(inode) ((inode)->i_sb->s_flags & MS_RDONLY)
1733#define IS_SYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS) || \
1734 ((inode)->i_flags & S_SYNC))
1735#define IS_DIRSYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS|MS_DIRSYNC) || \
1736 ((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
1737#define IS_MANDLOCK(inode) __IS_FLG(inode, MS_MANDLOCK)
1738#define IS_NOATIME(inode) __IS_FLG(inode, MS_RDONLY|MS_NOATIME)
1739#define IS_I_VERSION(inode) __IS_FLG(inode, MS_I_VERSION)
1740
1741#define IS_NOQUOTA(inode) ((inode)->i_flags & S_NOQUOTA)
1742#define IS_APPEND(inode) ((inode)->i_flags & S_APPEND)
1743#define IS_IMMUTABLE(inode) ((inode)->i_flags & S_IMMUTABLE)
1744#define IS_POSIXACL(inode) __IS_FLG(inode, MS_POSIXACL)
1745
1746#define IS_DEADDIR(inode) ((inode)->i_flags & S_DEAD)
1747#define IS_NOCMTIME(inode) ((inode)->i_flags & S_NOCMTIME)
1748#define IS_SWAPFILE(inode) ((inode)->i_flags & S_SWAPFILE)
1749#define IS_PRIVATE(inode) ((inode)->i_flags & S_PRIVATE)
1750#define IS_IMA(inode) ((inode)->i_flags & S_IMA)
1751#define IS_AUTOMOUNT(inode) ((inode)->i_flags & S_AUTOMOUNT)
1752#define IS_NOSEC(inode) ((inode)->i_flags & S_NOSEC)
Matthew Wilcoxfbbbad42015-02-16 15:58:53 -08001753#define IS_DAX(inode) ((inode)->i_flags & S_DAX)
David Howellsbbc10962012-10-15 16:40:35 +01001754
Miklos Szeredi787fb6b2014-10-24 00:14:36 +02001755#define IS_WHITEOUT(inode) (S_ISCHR(inode->i_mode) && \
1756 (inode)->i_rdev == WHITEOUT_DEV)
1757
David Howellsbbc10962012-10-15 16:40:35 +01001758/*
Dave Chinner250df6e2011-03-22 22:23:36 +11001759 * Inode state bits. Protected by inode->i_lock
Joern Engel1c0eeaf2007-10-16 23:30:44 -07001760 *
1761 * Three bits determine the dirty state of the inode, I_DIRTY_SYNC,
1762 * I_DIRTY_DATASYNC and I_DIRTY_PAGES.
1763 *
1764 * Four bits define the lifetime of an inode. Initially, inodes are I_NEW,
1765 * until that flag is cleared. I_WILL_FREE, I_FREEING and I_CLEAR are set at
1766 * various stages of removing an inode.
1767 *
Christoph Hellwigeaff8072009-12-17 14:25:01 +01001768 * Two bits are used for locking and completion notification, I_NEW and I_SYNC.
Joern Engel1c0eeaf2007-10-16 23:30:44 -07001769 *
Joern Engele7ca2d42008-02-06 01:36:59 -08001770 * I_DIRTY_SYNC Inode is dirty, but doesn't have to be written on
1771 * fdatasync(). i_atime is the usual cause.
Jan Kara000cb482008-02-14 19:31:32 -08001772 * I_DIRTY_DATASYNC Data-related inode changes pending. We keep track of
1773 * these changes separately from I_DIRTY_SYNC so that we
1774 * don't have to write inode on fdatasync() when only
1775 * mtime has changed in it.
Joern Engel1c0eeaf2007-10-16 23:30:44 -07001776 * I_DIRTY_PAGES Inode has dirty pages. Inode itself may be clean.
Christoph Hellwigeaff8072009-12-17 14:25:01 +01001777 * I_NEW Serves as both a mutex and completion notification.
1778 * New inodes set I_NEW. If two processes both create
1779 * the same inode, one of them will release its inode and
1780 * wait for I_NEW to be released before returning.
1781 * Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can
1782 * also cause waiting on I_NEW, without I_NEW actually
1783 * being set. find_inode() uses this to prevent returning
1784 * nearly-dead inodes.
Joern Engel1c0eeaf2007-10-16 23:30:44 -07001785 * I_WILL_FREE Must be set when calling write_inode_now() if i_count
1786 * is zero. I_FREEING must be set when I_WILL_FREE is
1787 * cleared.
1788 * I_FREEING Set when inode is about to be freed but still has dirty
1789 * pages or buffers attached or the inode itself is still
1790 * dirty.
Jan Karadbd57682012-05-03 14:48:02 +02001791 * I_CLEAR Added by clear_inode(). In this state the inode is
1792 * clean and can be destroyed. Inode keeps I_FREEING.
Joern Engel1c0eeaf2007-10-16 23:30:44 -07001793 *
1794 * Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are
1795 * prohibited for many purposes. iget() must wait for
1796 * the inode to be completely released, then create it
1797 * anew. Other functions will just ignore such inodes,
Christoph Hellwigeaff8072009-12-17 14:25:01 +01001798 * if appropriate. I_NEW is used for waiting.
Joern Engel1c0eeaf2007-10-16 23:30:44 -07001799 *
Jan Kara169ebd92012-05-03 14:48:03 +02001800 * I_SYNC Writeback of inode is running. The bit is set during
1801 * data writeback, and cleared with a wakeup on the bit
1802 * address once it is done. The bit is also used to pin
1803 * the inode in memory for flusher thread.
Joern Engel1c0eeaf2007-10-16 23:30:44 -07001804 *
Christoph Hellwigbd5fe6c2011-06-24 14:29:43 -04001805 * I_REFERENCED Marks the inode as recently references on the LRU list.
1806 *
1807 * I_DIO_WAKEUP Never set. Only used as a key for wait_on_bit().
1808 *
Joern Engel1c0eeaf2007-10-16 23:30:44 -07001809 * Q: What is the difference between I_WILL_FREE and I_FREEING?
Joern Engel1c0eeaf2007-10-16 23:30:44 -07001810 */
Nick Piggin9e38d862010-10-23 06:55:17 -04001811#define I_DIRTY_SYNC (1 << 0)
1812#define I_DIRTY_DATASYNC (1 << 1)
1813#define I_DIRTY_PAGES (1 << 2)
Christoph Hellwigeaff8072009-12-17 14:25:01 +01001814#define __I_NEW 3
1815#define I_NEW (1 << __I_NEW)
Nick Piggin9e38d862010-10-23 06:55:17 -04001816#define I_WILL_FREE (1 << 4)
1817#define I_FREEING (1 << 5)
1818#define I_CLEAR (1 << 6)
Christoph Hellwigeaff8072009-12-17 14:25:01 +01001819#define __I_SYNC 7
Joern Engel1c0eeaf2007-10-16 23:30:44 -07001820#define I_SYNC (1 << __I_SYNC)
Nick Piggin9e38d862010-10-23 06:55:17 -04001821#define I_REFERENCED (1 << 8)
Christoph Hellwigbd5fe6c2011-06-24 14:29:43 -04001822#define __I_DIO_WAKEUP 9
Eric Sandeenac74d8d2015-04-16 15:04:56 -05001823#define I_DIO_WAKEUP (1 << __I_DIO_WAKEUP)
Al Virof4e0c302013-06-11 08:34:36 +04001824#define I_LINKABLE (1 << 10)
Theodore Ts'o0ae45f62015-02-02 00:37:00 -05001825#define I_DIRTY_TIME (1 << 11)
1826#define __I_DIRTY_TIME_EXPIRED 12
1827#define I_DIRTY_TIME_EXPIRED (1 << __I_DIRTY_TIME_EXPIRED)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001828
1829#define I_DIRTY (I_DIRTY_SYNC | I_DIRTY_DATASYNC | I_DIRTY_PAGES)
Theodore Ts'o0ae45f62015-02-02 00:37:00 -05001830#define I_DIRTY_ALL (I_DIRTY | I_DIRTY_TIME)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001831
1832extern void __mark_inode_dirty(struct inode *, int);
1833static inline void mark_inode_dirty(struct inode *inode)
1834{
1835 __mark_inode_dirty(inode, I_DIRTY);
1836}
1837
1838static inline void mark_inode_dirty_sync(struct inode *inode)
1839{
1840 __mark_inode_dirty(inode, I_DIRTY_SYNC);
1841}
1842
Miklos Szeredi7ada4db2011-11-21 12:11:32 +01001843extern void inc_nlink(struct inode *inode);
1844extern void drop_nlink(struct inode *inode);
1845extern void clear_nlink(struct inode *inode);
1846extern void set_nlink(struct inode *inode, unsigned int nlink);
Dave Hansend8c76e62006-09-30 23:29:04 -07001847
1848static inline void inode_inc_link_count(struct inode *inode)
1849{
1850 inc_nlink(inode);
Alexey Dobriyan66d21732006-03-23 03:00:51 -08001851 mark_inode_dirty(inode);
1852}
1853
Dave Hansen9a53c3a2006-09-30 23:29:03 -07001854static inline void inode_dec_link_count(struct inode *inode)
1855{
1856 drop_nlink(inode);
Alexey Dobriyan66d21732006-03-23 03:00:51 -08001857 mark_inode_dirty(inode);
1858}
1859
Jean Noel Cordenner25ec56b2008-01-28 23:58:27 -05001860/**
1861 * inode_inc_iversion - increments i_version
1862 * @inode: inode that need to be updated
1863 *
1864 * Every time the inode is modified, the i_version field will be incremented.
1865 * The filesystem has to be mounted with i_version flag
1866 */
1867
1868static inline void inode_inc_iversion(struct inode *inode)
1869{
1870 spin_lock(&inode->i_lock);
1871 inode->i_version++;
1872 spin_unlock(&inode->i_lock);
1873}
1874
Josef Bacikc3b2da32012-03-26 09:59:21 -04001875enum file_time_flags {
1876 S_ATIME = 1,
1877 S_MTIME = 2,
1878 S_CTIME = 4,
1879 S_VERSION = 8,
1880};
1881
Al Virobadcf2b2013-07-16 18:15:46 +04001882extern void touch_atime(const struct path *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001883static inline void file_accessed(struct file *file)
1884{
1885 if (!(file->f_flags & O_NOATIME))
Al Viro68ac1232012-03-15 08:21:57 -04001886 touch_atime(&file->f_path);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001887}
1888
1889int sync_inode(struct inode *inode, struct writeback_control *wbc);
Christoph Hellwigc37650162010-10-06 10:48:20 +02001890int sync_inode_metadata(struct inode *inode, int wait);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001891
Linus Torvalds1da177e2005-04-16 15:20:36 -07001892struct file_system_type {
1893 const char *name;
1894 int fs_flags;
David Howellsbbc10962012-10-15 16:40:35 +01001895#define FS_REQUIRES_DEV 1
1896#define FS_BINARY_MOUNTDATA 2
1897#define FS_HAS_SUBTYPE 4
Eric W. Biederman0c55cfc2012-07-26 21:42:03 -07001898#define FS_USERNS_MOUNT 8 /* Can be mounted by userns root */
1899#define FS_USERNS_DEV_MOUNT 16 /* A userns mount does not imply MNT_NODEV */
David Howellsbbc10962012-10-15 16:40:35 +01001900#define FS_RENAME_DOES_D_MOVE 32768 /* FS will handle d_move() during rename() internally. */
Al Viroc96e41e2010-07-25 00:17:56 +04001901 struct dentry *(*mount) (struct file_system_type *, int,
1902 const char *, void *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001903 void (*kill_sb) (struct super_block *);
1904 struct module *owner;
1905 struct file_system_type * next;
Al Viroa5166162011-12-12 22:53:00 -05001906 struct hlist_head fs_supers;
Peter Zijlstrad475fd42007-10-15 14:51:31 +02001907
Ingo Molnarcf516242006-07-03 00:25:27 -07001908 struct lock_class_key s_lock_key;
Arjan van de Ven897c6ff2006-07-03 00:25:28 -07001909 struct lock_class_key s_umount_key;
Roland Dreier51ee0492010-04-27 14:23:57 -07001910 struct lock_class_key s_vfs_rename_key;
Jan Kara5accdf82012-06-12 16:20:34 +02001911 struct lock_class_key s_writers_key[SB_FREEZE_LEVELS];
Peter Zijlstrad475fd42007-10-15 14:51:31 +02001912
1913 struct lock_class_key i_lock_key;
1914 struct lock_class_key i_mutex_key;
Peter Zijlstra14358e62007-10-14 01:38:33 +02001915 struct lock_class_key i_mutex_dir_key;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001916};
1917
Eric W. Biederman7f78e032013-03-02 19:39:14 -08001918#define MODULE_ALIAS_FS(NAME) MODULE_ALIAS("fs-" NAME)
1919
Al Viroceefda62010-07-26 13:16:50 +04001920extern struct dentry *mount_ns(struct file_system_type *fs_type, int flags,
1921 void *data, int (*fill_super)(struct super_block *, void *, int));
Al Viro152a0832010-07-25 00:46:55 +04001922extern struct dentry *mount_bdev(struct file_system_type *fs_type,
1923 int flags, const char *dev_name, void *data,
1924 int (*fill_super)(struct super_block *, void *, int));
Al Virofc14f2f2010-07-25 01:48:30 +04001925extern struct dentry *mount_single(struct file_system_type *fs_type,
1926 int flags, void *data,
1927 int (*fill_super)(struct super_block *, void *, int));
Al Viro3c26ff62010-07-25 11:46:36 +04001928extern struct dentry *mount_nodev(struct file_system_type *fs_type,
1929 int flags, void *data,
1930 int (*fill_super)(struct super_block *, void *, int));
Al Viroea441d12011-11-16 21:43:59 -05001931extern struct dentry *mount_subtree(struct vfsmount *mnt, const char *path);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001932void generic_shutdown_super(struct super_block *sb);
1933void kill_block_super(struct super_block *sb);
1934void kill_anon_super(struct super_block *sb);
1935void kill_litter_super(struct super_block *sb);
1936void deactivate_super(struct super_block *sb);
Al Viro74dbbdd2009-05-06 01:07:50 -04001937void deactivate_locked_super(struct super_block *sb);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001938int set_anon_super(struct super_block *s, void *data);
Al Viro0ee5dc62011-07-07 15:44:25 -04001939int get_anon_bdev(dev_t *);
1940void free_anon_bdev(dev_t);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001941struct super_block *sget(struct file_system_type *type,
1942 int (*test)(struct super_block *,void *),
1943 int (*set)(struct super_block *,void *),
David Howells9249e172012-06-25 12:55:37 +01001944 int flags, void *data);
Al Viro51139ad2010-07-25 23:47:46 +04001945extern struct dentry *mount_pseudo(struct file_system_type *, char *,
Al Viroc74a1cb2011-01-12 16:59:34 -05001946 const struct super_operations *ops,
1947 const struct dentry_operations *dops,
1948 unsigned long);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001949
1950/* Alas, no aliases. Too much hassle with bringing module.h everywhere */
1951#define fops_get(fops) \
1952 (((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
1953#define fops_put(fops) \
1954 do { if (fops) module_put((fops)->owner); } while(0)
Al Viroe84f9e52013-09-22 14:17:15 -04001955/*
1956 * This one is to be used *ONLY* from ->open() instances.
1957 * fops must be non-NULL, pinned down *and* module dependencies
1958 * should be sufficient to pin the caller down as well.
1959 */
1960#define replace_fops(f, fops) \
1961 do { \
1962 struct file *__file = (f); \
1963 fops_put(__file->f_op); \
1964 BUG_ON(!(__file->f_op = (fops))); \
1965 } while(0)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001966
1967extern int register_filesystem(struct file_system_type *);
1968extern int unregister_filesystem(struct file_system_type *);
Pavel Emelyanov8bf97252007-10-18 23:40:02 -07001969extern struct vfsmount *kern_mount_data(struct file_system_type *, void *data);
1970#define kern_mount(type) kern_mount_data(type, NULL)
Tim Chen423e0ab2011-07-19 09:32:38 -07001971extern void kern_unmount(struct vfsmount *mnt);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001972extern int may_umount_tree(struct vfsmount *);
1973extern int may_umount(struct vfsmount *);
Seunghun Lee5e6123f2014-09-14 22:15:10 +09001974extern long do_mount(const char *, const char __user *,
1975 const char *, unsigned long, void *);
Al Viro589ff872009-04-18 03:28:19 -04001976extern struct vfsmount *collect_mounts(struct path *);
Al Viro8aec0802007-06-07 12:20:32 -04001977extern void drop_collected_mounts(struct vfsmount *);
Al Viro1f707132010-01-30 22:51:25 -05001978extern int iterate_mounts(int (*)(struct vfsmount *, void *), void *,
1979 struct vfsmount *);
Christoph Hellwigebabe9a2010-07-07 18:53:11 +02001980extern int vfs_statfs(struct path *, struct kstatfs *);
Al Viroc8b91ac2011-03-12 10:41:39 -05001981extern int user_statfs(const char __user *, struct kstatfs *);
1982extern int fd_statfs(int, struct kstatfs *);
Al Virocf31e702012-01-02 22:28:36 -05001983extern int vfs_ustat(dev_t, struct kstatfs *);
Josef Bacik18e9e512010-03-23 10:34:56 -04001984extern int freeze_super(struct super_block *super);
1985extern int thaw_super(struct super_block *super);
Al Viro02125a82011-12-05 08:43:34 -05001986extern bool our_mnt(struct vfsmount *mnt);
Eric W. Biedermane51db732013-03-30 19:57:41 -07001987extern bool fs_fully_visible(struct file_system_type *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001988
Al Viroce3b0f82009-03-29 19:08:22 -04001989extern int current_umask(void);
1990
J. Bruce Fields5a146962012-08-28 07:50:40 -07001991extern void ihold(struct inode * inode);
1992extern void iput(struct inode *);
Theodore Ts'o0ae45f62015-02-02 00:37:00 -05001993extern int generic_update_time(struct inode *, struct timespec *, int);
J. Bruce Fields5a146962012-08-28 07:50:40 -07001994
Al Viro1fa97e82014-05-07 20:47:49 -04001995static inline struct inode *file_inode(const struct file *f)
Jeff Laytond7a06982014-03-10 09:54:15 -04001996{
1997 return f->f_inode;
1998}
1999
Miklos Szeredif87fd4c2006-01-16 22:14:23 -08002000/* /sys/fs */
Greg Kroah-Hartman00d26662007-10-29 14:17:23 -06002001extern struct kobject *fs_kobj;
Miklos Szeredif87fd4c2006-01-16 22:14:23 -08002002
Linus Torvalds435f49a2010-10-29 10:36:49 -07002003#define MAX_RW_COUNT (INT_MAX & PAGE_CACHE_MASK)
Thomas Petazzonibfcd17a2008-08-06 15:12:22 +02002004
Linus Torvalds1da177e2005-04-16 15:20:36 -07002005#define FLOCK_VERIFY_READ 1
2006#define FLOCK_VERIFY_WRITE 2
2007
Thomas Petazzonibfcd17a2008-08-06 15:12:22 +02002008#ifdef CONFIG_FILE_LOCKING
Jeff Laytond7a06982014-03-10 09:54:15 -04002009extern int locks_mandatory_locked(struct file *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002010extern int locks_mandatory_area(int, struct inode *, struct file *, loff_t, size_t);
2011
2012/*
2013 * Candidates for mandatory locking have the setgid bit set
2014 * but no group execute bit - an otherwise meaningless combination.
2015 */
Pavel Emelyanova16877c2007-10-01 14:41:11 -07002016
2017static inline int __mandatory_lock(struct inode *ino)
2018{
2019 return (ino->i_mode & (S_ISGID | S_IXGRP)) == S_ISGID;
2020}
2021
2022/*
2023 * ... and these candidates should be on MS_MANDLOCK mounted fs,
2024 * otherwise these will be advisory locks
2025 */
2026
2027static inline int mandatory_lock(struct inode *ino)
2028{
2029 return IS_MANDLOCK(ino) && __mandatory_lock(ino);
2030}
Linus Torvalds1da177e2005-04-16 15:20:36 -07002031
Jeff Laytond7a06982014-03-10 09:54:15 -04002032static inline int locks_verify_locked(struct file *file)
Linus Torvalds1da177e2005-04-16 15:20:36 -07002033{
Jeff Laytond7a06982014-03-10 09:54:15 -04002034 if (mandatory_lock(file_inode(file)))
2035 return locks_mandatory_locked(file);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002036 return 0;
2037}
2038
Linus Torvalds1da177e2005-04-16 15:20:36 -07002039static inline int locks_verify_truncate(struct inode *inode,
2040 struct file *filp,
2041 loff_t size)
2042{
Jeff Laytonbd61e0a2015-01-16 15:05:55 -05002043 if (inode->i_flctx && mandatory_lock(inode))
Linus Torvalds1da177e2005-04-16 15:20:36 -07002044 return locks_mandatory_area(
2045 FLOCK_VERIFY_WRITE, inode, filp,
2046 size < inode->i_size ? size : inode->i_size,
2047 (size < inode->i_size ? inode->i_size - size
2048 : size - inode->i_size)
2049 );
2050 return 0;
2051}
2052
2053static inline int break_lease(struct inode *inode, unsigned int mode)
2054{
Jeff Layton24cbe782014-02-03 12:13:06 -05002055 /*
2056 * Since this check is lockless, we must ensure that any refcounts
Jeff Layton8116bf42015-01-21 20:44:01 -05002057 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2058 * could end up racing with tasks trying to set a new lease on this
2059 * file.
Jeff Layton24cbe782014-02-03 12:13:06 -05002060 */
2061 smp_mb();
Jeff Layton8634b512015-01-16 15:05:55 -05002062 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
J. Bruce Fieldsdf4e8d22012-03-05 13:18:59 -05002063 return __break_lease(inode, mode, FL_LEASE);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002064 return 0;
2065}
J. Bruce Fieldsdf4e8d22012-03-05 13:18:59 -05002066
2067static inline int break_deleg(struct inode *inode, unsigned int mode)
2068{
Jeff Layton962bd402014-06-10 12:24:40 -04002069 /*
2070 * Since this check is lockless, we must ensure that any refcounts
Jeff Layton8116bf42015-01-21 20:44:01 -05002071 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2072 * could end up racing with tasks trying to set a new lease on this
2073 * file.
Jeff Layton962bd402014-06-10 12:24:40 -04002074 */
2075 smp_mb();
Jeff Layton8634b512015-01-16 15:05:55 -05002076 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
J. Bruce Fieldsdf4e8d22012-03-05 13:18:59 -05002077 return __break_lease(inode, mode, FL_DELEG);
2078 return 0;
2079}
2080
J. Bruce Fields5a146962012-08-28 07:50:40 -07002081static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode)
2082{
2083 int ret;
2084
2085 ret = break_deleg(inode, O_WRONLY|O_NONBLOCK);
2086 if (ret == -EWOULDBLOCK && delegated_inode) {
2087 *delegated_inode = inode;
2088 ihold(inode);
2089 }
2090 return ret;
2091}
2092
2093static inline int break_deleg_wait(struct inode **delegated_inode)
2094{
2095 int ret;
2096
2097 ret = break_deleg(*delegated_inode, O_WRONLY);
2098 iput(*delegated_inode);
2099 *delegated_inode = NULL;
2100 return ret;
2101}
2102
Christoph Hellwig11afe9f2015-01-21 19:17:03 +01002103static inline int break_layout(struct inode *inode, bool wait)
2104{
2105 smp_mb();
2106 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2107 return __break_lease(inode,
2108 wait ? O_WRONLY : O_WRONLY | O_NONBLOCK,
2109 FL_LAYOUT);
2110 return 0;
2111}
2112
Thomas Petazzonibfcd17a2008-08-06 15:12:22 +02002113#else /* !CONFIG_FILE_LOCKING */
Jeff Laytond7a06982014-03-10 09:54:15 -04002114static inline int locks_mandatory_locked(struct file *file)
Steven Whitehouseaf5df562009-01-20 10:29:46 +00002115{
2116 return 0;
2117}
2118
2119static inline int locks_mandatory_area(int rw, struct inode *inode,
2120 struct file *filp, loff_t offset,
2121 size_t count)
2122{
2123 return 0;
2124}
2125
2126static inline int __mandatory_lock(struct inode *inode)
2127{
2128 return 0;
2129}
2130
2131static inline int mandatory_lock(struct inode *inode)
2132{
2133 return 0;
2134}
2135
Jeff Laytond7a06982014-03-10 09:54:15 -04002136static inline int locks_verify_locked(struct file *file)
Steven Whitehouseaf5df562009-01-20 10:29:46 +00002137{
2138 return 0;
2139}
2140
2141static inline int locks_verify_truncate(struct inode *inode, struct file *filp,
2142 size_t size)
2143{
2144 return 0;
2145}
2146
2147static inline int break_lease(struct inode *inode, unsigned int mode)
2148{
2149 return 0;
2150}
2151
J. Bruce Fieldsdf4e8d22012-03-05 13:18:59 -05002152static inline int break_deleg(struct inode *inode, unsigned int mode)
2153{
2154 return 0;
2155}
J. Bruce Fields5a146962012-08-28 07:50:40 -07002156
2157static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode)
2158{
2159 return 0;
2160}
2161
2162static inline int break_deleg_wait(struct inode **delegated_inode)
2163{
2164 BUG();
2165 return 0;
2166}
2167
Christoph Hellwig11afe9f2015-01-21 19:17:03 +01002168static inline int break_layout(struct inode *inode, bool wait)
2169{
2170 return 0;
2171}
2172
Thomas Petazzonibfcd17a2008-08-06 15:12:22 +02002173#endif /* CONFIG_FILE_LOCKING */
Linus Torvalds1da177e2005-04-16 15:20:36 -07002174
2175/* fs/open.c */
Jeff Laytonadb5c242012-10-10 16:43:13 -04002176struct audit_names;
Jeff Layton91a27b22012-10-10 15:25:28 -04002177struct filename {
Jeff Laytonadb5c242012-10-10 16:43:13 -04002178 const char *name; /* pointer to actual string */
2179 const __user char *uptr; /* original userland pointer */
2180 struct audit_names *aname;
Paul Moore55422d02015-01-22 00:00:23 -05002181 int refcnt;
Al Virofd2f7cb2015-02-22 20:07:13 -05002182 const char iname[];
Jeff Layton91a27b22012-10-10 15:25:28 -04002183};
Linus Torvalds1da177e2005-04-16 15:20:36 -07002184
David Howellsa02de962012-12-20 21:52:36 +00002185extern long vfs_truncate(struct path *, loff_t);
NeilBrown4a301312006-01-08 01:02:39 -08002186extern int do_truncate(struct dentry *, loff_t start, unsigned int time_attrs,
2187 struct file *filp);
Anna Schumaker72c72bd2014-11-07 14:44:25 -05002188extern int vfs_fallocate(struct file *file, int mode, loff_t offset,
Ankit Jain3e63cbb2009-06-19 14:28:07 -04002189 loff_t len);
Jason Uhlenkott8e8a1402007-10-20 03:16:18 +02002190extern long do_sys_open(int dfd, const char __user *filename, int flags,
Al Viroa218d0f2011-11-21 14:59:34 -05002191 umode_t mode);
Jeff Layton669abf42012-10-10 16:43:10 -04002192extern struct file *file_open_name(struct filename *, int, umode_t);
Al Viroa218d0f2011-11-21 14:59:34 -05002193extern struct file *filp_open(const char *, int, umode_t);
Al Viro73d049a2011-03-11 12:08:24 -05002194extern struct file *file_open_root(struct dentry *, struct vfsmount *,
2195 const char *, int);
Miklos Szeredi4aa7c632014-10-24 00:14:35 +02002196extern int vfs_open(const struct path *, struct file *, const struct cred *);
Al Viro765927b2012-06-26 21:58:53 +04002197extern struct file * dentry_open(const struct path *, int, const struct cred *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002198extern int filp_close(struct file *, fl_owner_t id);
Jeff Layton91a27b22012-10-10 15:25:28 -04002199
David Drysdale51f39a12014-12-12 16:57:29 -08002200extern struct filename *getname_flags(const char __user *, int, int *);
Jeff Layton91a27b22012-10-10 15:25:28 -04002201extern struct filename *getname(const char __user *);
Linus Torvaldsc4ad8f92014-02-05 12:54:53 -08002202extern struct filename *getname_kernel(const char *);
Paul Moore55422d02015-01-22 00:00:23 -05002203extern void putname(struct filename *name);
Jeff Layton91a27b22012-10-10 15:25:28 -04002204
Al Viro47237682012-06-10 05:01:45 -04002205enum {
2206 FILE_CREATED = 1,
2207 FILE_OPENED = 2
2208};
Al Viro30d90492012-06-22 12:40:19 +04002209extern int finish_open(struct file *file, struct dentry *dentry,
2210 int (*open)(struct inode *, struct file *),
2211 int *opened);
Al Viroe45198a2012-06-10 06:48:09 -04002212extern int finish_no_open(struct file *file, struct dentry *dentry);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002213
Ankit Jain3e63cbb2009-06-19 14:28:07 -04002214/* fs/ioctl.c */
2215
2216extern int ioctl_preallocate(struct file *filp, void __user *argp);
2217
Linus Torvalds1da177e2005-04-16 15:20:36 -07002218/* fs/dcache.c */
2219extern void __init vfs_caches_init_early(void);
2220extern void __init vfs_caches_init(unsigned long);
2221
Christoph Lameterb86c0892006-12-06 20:32:57 -08002222extern struct kmem_cache *names_cachep;
2223
Jeff Laytona608ca22012-10-10 15:25:26 -04002224#define __getname() kmem_cache_alloc(names_cachep, GFP_KERNEL)
Vegard Nossum3446a8a2009-05-16 11:22:14 +02002225#define __putname(name) kmem_cache_free(names_cachep, (void *)(name))
Linus Torvalds1da177e2005-04-16 15:20:36 -07002226
David Howells93614012006-09-30 20:45:40 +02002227#ifdef CONFIG_BLOCK
Linus Torvalds1da177e2005-04-16 15:20:36 -07002228extern int register_blkdev(unsigned int, const char *);
Akinobu Mitaf4480242007-07-17 04:03:47 -07002229extern void unregister_blkdev(unsigned int, const char *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002230extern struct block_device *bdget(dev_t);
Alan Jenkinsdddac6a2009-07-29 21:07:55 +02002231extern struct block_device *bdgrab(struct block_device *bdev);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002232extern void bd_set_size(struct block_device *, loff_t size);
2233extern void bd_forget(struct inode *inode);
2234extern void bdput(struct block_device *);
Nick Piggin585d3bc2009-02-25 10:44:19 +01002235extern void invalidate_bdev(struct block_device *);
Jan Kara5c0d6b62012-07-03 16:45:31 +02002236extern void iterate_bdevs(void (*)(struct block_device *, void *), void *);
Nick Piggin585d3bc2009-02-25 10:44:19 +01002237extern int sync_blockdev(struct block_device *bdev);
Al Viroff01bb42011-09-16 02:31:11 -04002238extern void kill_bdev(struct block_device *);
Nick Piggin585d3bc2009-02-25 10:44:19 +01002239extern struct super_block *freeze_bdev(struct block_device *);
Eric Sandeenc2d75432009-03-31 15:23:46 -07002240extern void emergency_thaw_all(void);
Nick Piggin585d3bc2009-02-25 10:44:19 +01002241extern int thaw_bdev(struct block_device *bdev, struct super_block *sb);
2242extern int fsync_bdev(struct block_device *);
Jan Karaa8855992013-07-09 22:36:45 +08002243extern int sb_is_blkdev_sb(struct super_block *sb);
David Howells93614012006-09-30 20:45:40 +02002244#else
2245static inline void bd_forget(struct inode *inode) {}
Al Viro47e44912009-04-01 07:07:16 -04002246static inline int sync_blockdev(struct block_device *bdev) { return 0; }
Al Viroff01bb42011-09-16 02:31:11 -04002247static inline void kill_bdev(struct block_device *bdev) {}
Al Viro47e44912009-04-01 07:07:16 -04002248static inline void invalidate_bdev(struct block_device *bdev) {}
2249
2250static inline struct super_block *freeze_bdev(struct block_device *sb)
2251{
2252 return NULL;
2253}
2254
2255static inline int thaw_bdev(struct block_device *bdev, struct super_block *sb)
2256{
2257 return 0;
2258}
Jan Kara5c0d6b62012-07-03 16:45:31 +02002259
2260static inline void iterate_bdevs(void (*f)(struct block_device *, void *), void *arg)
2261{
2262}
Jan Karaa8855992013-07-09 22:36:45 +08002263
2264static inline int sb_is_blkdev_sb(struct super_block *sb)
2265{
2266 return 0;
2267}
David Howells93614012006-09-30 20:45:40 +02002268#endif
Jan Kara60b06802009-04-27 16:43:53 +02002269extern int sync_filesystem(struct super_block *);
David Howells93614012006-09-30 20:45:40 +02002270extern const struct file_operations def_blk_fops;
Arjan van de Ven4b6f5d22006-03-28 01:56:42 -08002271extern const struct file_operations def_chr_fops;
David Howells93614012006-09-30 20:45:40 +02002272#ifdef CONFIG_BLOCK
Linus Torvalds1da177e2005-04-16 15:20:36 -07002273extern int ioctl_by_bdev(struct block_device *, unsigned, unsigned long);
Al Viro56b26ad2008-09-19 03:17:36 -04002274extern int blkdev_ioctl(struct block_device *, fmode_t, unsigned, unsigned long);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002275extern long compat_blkdev_ioctl(struct file *, unsigned, unsigned long);
Tejun Heoe525fd82010-11-13 11:55:17 +01002276extern int blkdev_get(struct block_device *bdev, fmode_t mode, void *holder);
Tejun Heod4d77622010-11-13 11:55:18 +01002277extern struct block_device *blkdev_get_by_path(const char *path, fmode_t mode,
2278 void *holder);
2279extern struct block_device *blkdev_get_by_dev(dev_t dev, fmode_t mode,
2280 void *holder);
Al Viro4385bab2013-05-05 22:11:03 -04002281extern void blkdev_put(struct block_device *bdev, fmode_t mode);
Jun'ichi Nomura641dc632006-03-27 01:17:57 -08002282#ifdef CONFIG_SYSFS
Tejun Heoe09b4572010-11-13 11:55:17 +01002283extern int bd_link_disk_holder(struct block_device *bdev, struct gendisk *disk);
Tejun Heo49731ba2011-01-14 18:43:57 +01002284extern void bd_unlink_disk_holder(struct block_device *bdev,
2285 struct gendisk *disk);
Jun'ichi Nomura641dc632006-03-27 01:17:57 -08002286#else
Tejun Heoe09b4572010-11-13 11:55:17 +01002287static inline int bd_link_disk_holder(struct block_device *bdev,
2288 struct gendisk *disk)
2289{
2290 return 0;
2291}
Tejun Heo49731ba2011-01-14 18:43:57 +01002292static inline void bd_unlink_disk_holder(struct block_device *bdev,
2293 struct gendisk *disk)
2294{
2295}
Jun'ichi Nomura641dc632006-03-27 01:17:57 -08002296#endif
David Howells93614012006-09-30 20:45:40 +02002297#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -07002298
2299/* fs/char_dev.c */
Joe Korty68eef3b2006-03-31 02:30:32 -08002300#define CHRDEV_MAJOR_HASH_SIZE 255
Linus Torvalds1da177e2005-04-16 15:20:36 -07002301extern int alloc_chrdev_region(dev_t *, unsigned, unsigned, const char *);
2302extern int register_chrdev_region(dev_t, unsigned, const char *);
Tejun Heo1905b1b2009-08-06 18:13:23 +09002303extern int __register_chrdev(unsigned int major, unsigned int baseminor,
2304 unsigned int count, const char *name,
2305 const struct file_operations *fops);
2306extern void __unregister_chrdev(unsigned int major, unsigned int baseminor,
2307 unsigned int count, const char *name);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002308extern void unregister_chrdev_region(dev_t, unsigned);
Joe Korty68eef3b2006-03-31 02:30:32 -08002309extern void chrdev_show(struct seq_file *,off_t);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002310
Tejun Heo1905b1b2009-08-06 18:13:23 +09002311static inline int register_chrdev(unsigned int major, const char *name,
2312 const struct file_operations *fops)
2313{
2314 return __register_chrdev(major, 0, 256, name, fops);
2315}
2316
2317static inline void unregister_chrdev(unsigned int major, const char *name)
2318{
2319 __unregister_chrdev(major, 0, 256, name);
2320}
2321
Linus Torvalds1da177e2005-04-16 15:20:36 -07002322/* fs/block_dev.c */
2323#define BDEVNAME_SIZE 32 /* Largest string for a blockdev identifier */
Tejun Heo1f014292008-08-25 19:47:23 +09002324#define BDEVT_SIZE 10 /* Largest string for MAJ:MIN for blkdev */
David Howells93614012006-09-30 20:45:40 +02002325
2326#ifdef CONFIG_BLOCK
2327#define BLKDEV_MAJOR_HASH_SIZE 255
Linus Torvalds1da177e2005-04-16 15:20:36 -07002328extern const char *__bdevname(dev_t, char *buffer);
2329extern const char *bdevname(struct block_device *bdev, char *buffer);
2330extern struct block_device *lookup_bdev(const char *);
Joe Korty68eef3b2006-03-31 02:30:32 -08002331extern void blkdev_show(struct seq_file *,off_t);
Al Viro30c40d22008-02-22 19:50:45 -05002332
David Howells93614012006-09-30 20:45:40 +02002333#else
2334#define BLKDEV_MAJOR_HASH_SIZE 0
2335#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -07002336
2337extern void init_special_inode(struct inode *, umode_t, dev_t);
2338
2339/* Invalid inode operations -- fs/bad_inode.c */
2340extern void make_bad_inode(struct inode *);
2341extern int is_bad_inode(struct inode *);
2342
David Howells93614012006-09-30 20:45:40 +02002343#ifdef CONFIG_BLOCK
Linus Torvalds1da177e2005-04-16 15:20:36 -07002344/*
2345 * return READ, READA, or WRITE
2346 */
2347#define bio_rw(bio) ((bio)->bi_rw & (RW_MASK | RWA_MASK))
2348
2349/*
2350 * return data direction, READ or WRITE
2351 */
2352#define bio_data_dir(bio) ((bio)->bi_rw & 1)
2353
Andrew Pattersonc3279d12008-09-04 14:27:25 -06002354extern void check_disk_size_change(struct gendisk *disk,
2355 struct block_device *bdev);
Andrew Patterson0c002c22008-09-04 14:27:20 -06002356extern int revalidate_disk(struct gendisk *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002357extern int check_disk_change(struct block_device *);
NeilBrown93b270f2011-02-24 17:25:47 +11002358extern int __invalidate_device(struct block_device *, bool);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002359extern int invalidate_partition(struct gendisk *, int);
David Howells93614012006-09-30 20:45:40 +02002360#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -07002361unsigned long invalidate_mapping_pages(struct address_space *mapping,
2362 pgoff_t start, pgoff_t end);
Anton Altaparmakov54bc4852007-02-10 01:45:38 -08002363
Linus Torvalds1da177e2005-04-16 15:20:36 -07002364static inline void invalidate_remote_inode(struct inode *inode)
2365{
2366 if (S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) ||
2367 S_ISLNK(inode->i_mode))
Andrew Mortonfc0ecff2007-02-10 01:45:39 -08002368 invalidate_mapping_pages(inode->i_mapping, 0, -1);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002369}
2370extern int invalidate_inode_pages2(struct address_space *mapping);
2371extern int invalidate_inode_pages2_range(struct address_space *mapping,
2372 pgoff_t start, pgoff_t end);
2373extern int write_inode_now(struct inode *, int);
2374extern int filemap_fdatawrite(struct address_space *);
2375extern int filemap_flush(struct address_space *);
2376extern int filemap_fdatawait(struct address_space *);
Jan Karad3bccb6f2009-08-17 19:30:27 +02002377extern int filemap_fdatawait_range(struct address_space *, loff_t lstart,
2378 loff_t lend);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002379extern int filemap_write_and_wait(struct address_space *mapping);
2380extern int filemap_write_and_wait_range(struct address_space *mapping,
2381 loff_t lstart, loff_t lend);
Andrew Mortonebcf28e2006-03-24 03:18:04 -08002382extern int __filemap_fdatawrite_range(struct address_space *mapping,
2383 loff_t start, loff_t end, int sync_mode);
Jan Karaf4c0a0f2008-07-11 19:27:31 -04002384extern int filemap_fdatawrite_range(struct address_space *mapping,
2385 loff_t start, loff_t end);
Andrew Mortonebcf28e2006-03-24 03:18:04 -08002386
Christoph Hellwig8018ab02010-03-22 17:32:25 +01002387extern int vfs_fsync_range(struct file *file, loff_t start, loff_t end,
2388 int datasync);
2389extern int vfs_fsync(struct file *file, int datasync);
Al Virod311d792014-02-09 15:18:09 -05002390static inline int generic_write_sync(struct file *file, loff_t pos, loff_t count)
2391{
2392 if (!(file->f_flags & O_DSYNC) && !IS_SYNC(file->f_mapping->host))
2393 return 0;
2394 return vfs_fsync_range(file, pos, pos + count - 1,
2395 (file->f_flags & __O_SYNC) ? 0 : 1);
2396}
Linus Torvalds1da177e2005-04-16 15:20:36 -07002397extern void emergency_sync(void);
2398extern void emergency_remount(void);
David Howells93614012006-09-30 20:45:40 +02002399#ifdef CONFIG_BLOCK
Linus Torvalds1da177e2005-04-16 15:20:36 -07002400extern sector_t bmap(struct inode *, sector_t);
David Howells93614012006-09-30 20:45:40 +02002401#endif
J. Bruce Fields27ac0ff2011-09-20 17:19:26 -04002402extern int notify_change(struct dentry *, struct iattr *, struct inode **);
Al Virof419a2e2008-07-22 00:07:17 -04002403extern int inode_permission(struct inode *, int);
Miklos Szeredibd5d0852014-10-24 00:14:35 +02002404extern int __inode_permission(struct inode *, int);
Al Viro2830ba72011-06-20 19:16:29 -04002405extern int generic_permission(struct inode *, int);
Miklos Szeredicbdf35b2014-10-24 00:14:36 +02002406extern int __check_sticky(struct inode *dir, struct inode *inode);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002407
Miklos Szeredif696a362008-07-31 13:41:58 +02002408static inline bool execute_ok(struct inode *inode)
2409{
2410 return (inode->i_mode & S_IXUGO) || S_ISDIR(inode->i_mode);
2411}
2412
Al Viro8d71db42013-03-19 21:01:03 -04002413static inline void file_start_write(struct file *file)
2414{
2415 if (!S_ISREG(file_inode(file)->i_mode))
2416 return;
2417 __sb_start_write(file_inode(file)->i_sb, SB_FREEZE_WRITE, true);
2418}
2419
Jan Kara5ae98f12013-05-04 00:11:23 +02002420static inline bool file_start_write_trylock(struct file *file)
2421{
2422 if (!S_ISREG(file_inode(file)->i_mode))
2423 return true;
2424 return __sb_start_write(file_inode(file)->i_sb, SB_FREEZE_WRITE, false);
2425}
2426
Al Viro8d71db42013-03-19 21:01:03 -04002427static inline void file_end_write(struct file *file)
2428{
2429 if (!S_ISREG(file_inode(file)->i_mode))
2430 return;
2431 __sb_end_write(file_inode(file)->i_sb, SB_FREEZE_WRITE);
2432}
2433
Al Viro07b8ce12011-06-20 10:52:57 -04002434/*
2435 * get_write_access() gets write permission for a file.
2436 * put_write_access() releases this write permission.
2437 * This is used for regular files.
2438 * We cannot support write (and maybe mmap read-write shared) accesses and
2439 * MAP_DENYWRITE mmappings simultaneously. The i_writecount field of an inode
2440 * can have the following values:
2441 * 0: no writers, no VM_DENYWRITE mappings
2442 * < 0: (-i_writecount) vm_area_structs with VM_DENYWRITE set exist
2443 * > 0: (i_writecount) users are writing to the file.
2444 *
2445 * Normally we operate on that counter with atomic_{inc,dec} and it's safe
2446 * except for the cases where we don't hold i_writecount yet. Then we need to
2447 * use {get,deny}_write_access() - these functions check the sign and refuse
2448 * to do the change if sign is wrong.
2449 */
2450static inline int get_write_access(struct inode *inode)
2451{
2452 return atomic_inc_unless_negative(&inode->i_writecount) ? 0 : -ETXTBSY;
2453}
2454static inline int deny_write_access(struct file *file)
2455{
Al Viro496ad9a2013-01-23 17:07:38 -05002456 struct inode *inode = file_inode(file);
Al Viro07b8ce12011-06-20 10:52:57 -04002457 return atomic_dec_unless_positive(&inode->i_writecount) ? 0 : -ETXTBSY;
2458}
Linus Torvalds1da177e2005-04-16 15:20:36 -07002459static inline void put_write_access(struct inode * inode)
2460{
2461 atomic_dec(&inode->i_writecount);
2462}
2463static inline void allow_write_access(struct file *file)
2464{
2465 if (file)
Al Viro496ad9a2013-01-23 17:07:38 -05002466 atomic_inc(&file_inode(file)->i_writecount);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002467}
David Howellsf1fe29b2013-09-27 11:20:03 +01002468static inline bool inode_is_open_for_write(const struct inode *inode)
2469{
2470 return atomic_read(&inode->i_writecount) > 0;
2471}
2472
Mimi Zohara5c96eb2010-11-02 10:11:37 -04002473#ifdef CONFIG_IMA
2474static inline void i_readcount_dec(struct inode *inode)
2475{
2476 BUG_ON(!atomic_read(&inode->i_readcount));
2477 atomic_dec(&inode->i_readcount);
2478}
2479static inline void i_readcount_inc(struct inode *inode)
2480{
2481 atomic_inc(&inode->i_readcount);
2482}
2483#else
2484static inline void i_readcount_dec(struct inode *inode)
2485{
2486 return;
2487}
2488static inline void i_readcount_inc(struct inode *inode)
2489{
2490 return;
2491}
2492#endif
Ulrich Dreppered8cae82008-07-23 21:29:30 -07002493extern int do_pipe_flags(int *, int);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002494
Mimi Zohar6777d772009-08-21 14:32:48 -04002495extern int kernel_read(struct file *, loff_t, char *, unsigned long);
Al Viro7bb307e2013-02-23 14:51:48 -05002496extern ssize_t kernel_write(struct file *, const char *, size_t, loff_t);
Al Viroed447242014-04-19 14:37:20 -04002497extern ssize_t __kernel_write(struct file *, const char *, size_t, loff_t *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002498extern struct file * open_exec(const char *);
2499
2500/* fs/dcache.c -- generic fs support functions */
2501extern int is_subdir(struct dentry *, struct dentry *);
Al Viro2096f752010-01-30 13:16:21 -05002502extern int path_is_under(struct path *, struct path *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002503
2504#include <linux/err.h>
2505
2506/* needed for stackable file system support */
Andrew Morton965c8e52012-12-17 15:59:39 -08002507extern loff_t default_llseek(struct file *file, loff_t offset, int whence);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002508
Andrew Morton965c8e52012-12-17 15:59:39 -08002509extern loff_t vfs_llseek(struct file *file, loff_t offset, int whence);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002510
Christoph Hellwig54e34622009-08-07 14:38:25 -03002511extern int inode_init_always(struct super_block *, struct inode *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002512extern void inode_init_once(struct inode *);
Miklos Szeredi2aa15892011-02-23 13:49:47 +01002513extern void address_space_init_once(struct address_space *mapping);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002514extern struct inode * igrab(struct inode *);
2515extern ino_t iunique(struct super_block *, ino_t);
2516extern int inode_needs_sync(struct inode *inode);
Al Viro45321ac2010-06-07 13:43:19 -04002517extern int generic_delete_inode(struct inode *inode);
Al Viro9bcb4b72012-02-12 19:43:17 -05002518static inline int generic_drop_inode(struct inode *inode)
2519{
2520 return !inode->i_nlink || inode_unhashed(inode);
2521}
Linus Torvalds1da177e2005-04-16 15:20:36 -07002522
Anton Altaparmakov88bd5122005-07-13 01:10:44 -07002523extern struct inode *ilookup5_nowait(struct super_block *sb,
2524 unsigned long hashval, int (*test)(struct inode *, void *),
2525 void *data);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002526extern struct inode *ilookup5(struct super_block *sb, unsigned long hashval,
2527 int (*test)(struct inode *, void *), void *data);
2528extern struct inode *ilookup(struct super_block *sb, unsigned long ino);
2529
2530extern struct inode * iget5_locked(struct super_block *, unsigned long, int (*test)(struct inode *, void *), int (*set)(struct inode *, void *), void *);
2531extern struct inode * iget_locked(struct super_block *, unsigned long);
Theodore Ts'ofe032c42015-02-02 00:37:01 -05002532extern struct inode *find_inode_nowait(struct super_block *,
2533 unsigned long,
2534 int (*match)(struct inode *,
2535 unsigned long, void *),
2536 void *data);
Al Viro261bca82008-12-30 01:48:21 -05002537extern int insert_inode_locked4(struct inode *, unsigned long, int (*test)(struct inode *, void *), void *);
2538extern int insert_inode_locked(struct inode *);
Josh Boyere096d0c2011-08-25 07:48:12 -04002539#ifdef CONFIG_DEBUG_LOCK_ALLOC
2540extern void lockdep_annotate_inode_mutex_key(struct inode *inode);
2541#else
2542static inline void lockdep_annotate_inode_mutex_key(struct inode *inode) { };
2543#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -07002544extern void unlock_new_inode(struct inode *);
Christoph Hellwig85fe4022010-10-23 11:19:54 -04002545extern unsigned int get_next_ino(void);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002546
Linus Torvalds1da177e2005-04-16 15:20:36 -07002547extern void __iget(struct inode * inode);
David Howellsb46980f2008-02-07 00:15:27 -08002548extern void iget_failed(struct inode *);
Jan Karadbd57682012-05-03 14:48:02 +02002549extern void clear_inode(struct inode *);
Christoph Hellwig2e00c972009-08-07 14:38:29 -03002550extern void __destroy_inode(struct inode *);
Eric Dumazeta209dfc2011-07-26 11:36:34 +02002551extern struct inode *new_inode_pseudo(struct super_block *sb);
2552extern struct inode *new_inode(struct super_block *sb);
Nick Pigginff0c7d12011-01-07 17:49:50 +11002553extern void free_inode_nonrcu(struct inode *inode);
Jens Axboe01de85e2006-10-17 19:50:36 +02002554extern int should_remove_suid(struct dentry *);
Miklos Szeredi2f1936b2008-06-24 16:50:14 +02002555extern int file_remove_suid(struct file *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002556
2557extern void __insert_inode_hash(struct inode *, unsigned long hashval);
Christoph Hellwig646ec462010-10-23 07:15:32 -04002558static inline void insert_inode_hash(struct inode *inode)
2559{
Linus Torvalds1da177e2005-04-16 15:20:36 -07002560 __insert_inode_hash(inode, inode->i_ino);
2561}
Eric Dumazetf2ee7ab2011-07-28 06:41:09 +02002562
2563extern void __remove_inode_hash(struct inode *);
2564static inline void remove_inode_hash(struct inode *inode)
2565{
2566 if (!inode_unhashed(inode))
2567 __remove_inode_hash(inode);
2568}
2569
Christoph Hellwig646ec462010-10-23 07:15:32 -04002570extern void inode_sb_list_add(struct inode *inode);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002571
David Howells93614012006-09-30 20:45:40 +02002572#ifdef CONFIG_BLOCK
Linus Torvalds1da177e2005-04-16 15:20:36 -07002573extern void submit_bio(int, struct bio *);
2574extern int bdev_read_only(struct block_device *);
David Howells93614012006-09-30 20:45:40 +02002575#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -07002576extern int set_blocksize(struct block_device *, int);
2577extern int sb_set_blocksize(struct super_block *, int);
2578extern int sb_min_blocksize(struct super_block *, int);
2579
2580extern int generic_file_mmap(struct file *, struct vm_area_struct *);
2581extern int generic_file_readonly_mmap(struct file *, struct vm_area_struct *);
Al Viro3309dd02015-04-09 12:55:47 -04002582extern ssize_t generic_write_checks(struct kiocb *, struct iov_iter *);
Al Viroed978a82014-03-05 22:53:04 -05002583extern ssize_t generic_file_read_iter(struct kiocb *, struct iov_iter *);
Al Viro81742022014-04-03 03:17:43 -04002584extern ssize_t __generic_file_write_iter(struct kiocb *, struct iov_iter *);
Al Viro81742022014-04-03 03:17:43 -04002585extern ssize_t generic_file_write_iter(struct kiocb *, struct iov_iter *);
Al Viro0c949332014-03-22 06:51:37 -04002586extern ssize_t generic_file_direct_write(struct kiocb *, struct iov_iter *, loff_t);
Al Viro3b93f912014-02-11 21:34:08 -05002587extern ssize_t generic_perform_write(struct file *, struct iov_iter *, loff_t);
Andrew Morton88e6fae2006-04-11 13:59:36 +02002588
Christoph Hellwigdbe4e192015-01-25 21:11:59 +01002589ssize_t vfs_iter_read(struct file *file, struct iov_iter *iter, loff_t *ppos);
2590ssize_t vfs_iter_write(struct file *file, struct iov_iter *iter, loff_t *ppos);
2591
Christoph Hellwigeef99382009-08-20 17:43:41 +02002592/* fs/block_dev.c */
David Jefferyb2de5252014-09-29 10:21:10 -04002593extern ssize_t blkdev_read_iter(struct kiocb *iocb, struct iov_iter *to);
Al Viro1456c0a2014-04-03 03:21:50 -04002594extern ssize_t blkdev_write_iter(struct kiocb *iocb, struct iov_iter *from);
Josef Bacik02c24a82011-07-16 20:44:56 -04002595extern int blkdev_fsync(struct file *filp, loff_t start, loff_t end,
2596 int datasync);
Al Viroff01bb42011-09-16 02:31:11 -04002597extern void block_sync_page(struct page *page);
Christoph Hellwigeef99382009-08-20 17:43:41 +02002598
Andrew Morton88e6fae2006-04-11 13:59:36 +02002599/* fs/splice.c */
Jens Axboecbb7e572006-04-11 14:57:50 +02002600extern ssize_t generic_file_splice_read(struct file *, loff_t *,
Andrew Morton88e6fae2006-04-11 13:59:36 +02002601 struct pipe_inode_info *, size_t, unsigned int);
Miklos Szeredi68181732009-05-07 15:37:36 +02002602extern ssize_t default_file_splice_read(struct file *, loff_t *,
2603 struct pipe_inode_info *, size_t, unsigned int);
Al Viro8d020762014-04-05 04:27:08 -04002604extern ssize_t iter_file_splice_write(struct pipe_inode_info *,
Jens Axboecbb7e572006-04-11 14:57:50 +02002605 struct file *, loff_t *, size_t, unsigned int);
Andrew Morton88e6fae2006-04-11 13:59:36 +02002606extern ssize_t generic_splice_sendpage(struct pipe_inode_info *pipe,
Jens Axboecbb7e572006-04-11 14:57:50 +02002607 struct file *out, loff_t *, size_t len, unsigned int flags);
Miklos Szeredi1c118592014-10-24 00:14:35 +02002608extern long do_splice_direct(struct file *in, loff_t *ppos, struct file *out,
2609 loff_t *opos, size_t len, unsigned int flags);
2610
Andrew Morton88e6fae2006-04-11 13:59:36 +02002611
Linus Torvalds1da177e2005-04-16 15:20:36 -07002612extern void
2613file_ra_state_init(struct file_ra_state *ra, struct address_space *mapping);
Andrew Morton965c8e52012-12-17 15:59:39 -08002614extern loff_t noop_llseek(struct file *file, loff_t offset, int whence);
2615extern loff_t no_llseek(struct file *file, loff_t offset, int whence);
Jie Liu46a1c2c2013-06-25 12:02:13 +08002616extern loff_t vfs_setpos(struct file *file, loff_t offset, loff_t maxsize);
Andrew Morton965c8e52012-12-17 15:59:39 -08002617extern loff_t generic_file_llseek(struct file *file, loff_t offset, int whence);
Andi Kleen57604952011-09-15 16:06:50 -07002618extern loff_t generic_file_llseek_size(struct file *file, loff_t offset,
Andrew Morton965c8e52012-12-17 15:59:39 -08002619 int whence, loff_t maxsize, loff_t eof);
Al Viro1bf9d142013-06-16 20:27:42 +04002620extern loff_t fixed_size_llseek(struct file *file, loff_t offset,
2621 int whence, loff_t size);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002622extern int generic_file_open(struct inode * inode, struct file * filp);
2623extern int nonseekable_open(struct inode * inode, struct file * filp);
2624
Omar Sandovala95cd632015-03-16 04:33:51 -07002625ssize_t dax_do_io(struct kiocb *, struct inode *, struct iov_iter *, loff_t,
2626 get_block_t, dio_iodone_t, int flags);
Matthew Wilcox289c6ae2015-02-16 15:58:59 -08002627int dax_clear_blocks(struct inode *, sector_t block, long size);
Matthew Wilcox25726bc2015-02-16 15:59:35 -08002628int dax_zero_page_range(struct inode *, loff_t from, unsigned len, get_block_t);
Matthew Wilcox4c0ccfe2015-02-16 15:59:06 -08002629int dax_truncate_page(struct inode *, loff_t from, get_block_t);
Matthew Wilcoxf7ca90b2015-02-16 15:59:02 -08002630int dax_fault(struct vm_area_struct *, struct vm_fault *, get_block_t);
Boaz Harrosh0e3b2102015-04-15 16:15:14 -07002631int dax_pfn_mkwrite(struct vm_area_struct *, struct vm_fault *);
Matthew Wilcoxf7ca90b2015-02-16 15:59:02 -08002632#define dax_mkwrite(vma, vmf, gb) dax_fault(vma, vmf, gb)
Matthew Wilcoxd475c632015-02-16 15:58:56 -08002633
David Howells93614012006-09-30 20:45:40 +02002634#ifdef CONFIG_BLOCK
Josef Bacikfacd07b2010-05-23 11:00:55 -04002635typedef void (dio_submit_t)(int rw, struct bio *bio, struct inode *inode,
2636 loff_t file_offset);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002637
2638enum {
Christoph Hellwig1e431f52009-11-03 16:44:53 +01002639 /* need locking between buffered and direct access */
2640 DIO_LOCKING = 0x01,
2641
2642 /* filesystem does not support filling holes */
2643 DIO_SKIP_HOLES = 0x02,
Christoph Hellwig60392572014-02-10 10:27:11 +11002644
2645 /* filesystem can handle aio writes beyond i_size */
2646 DIO_ASYNC_EXTEND = 0x04,
Jens Axboefe0f07d2015-04-15 17:05:48 -06002647
2648 /* inode/fs/bdev does not need truncate protection */
2649 DIO_SKIP_DIO_COUNT = 0x08,
Linus Torvalds1da177e2005-04-16 15:20:36 -07002650};
2651
Christoph Hellwigeafdc7d2010-06-04 11:29:53 +02002652void dio_end_io(struct bio *bio, int error);
npiggin@suse.de7bb46a62010-05-27 01:05:33 +10002653
Omar Sandoval17f8c842015-03-16 04:33:50 -07002654ssize_t __blockdev_direct_IO(struct kiocb *iocb, struct inode *inode,
2655 struct block_device *bdev, struct iov_iter *iter,
2656 loff_t offset, get_block_t get_block,
2657 dio_iodone_t end_io, dio_submit_t submit_io,
2658 int flags);
Christoph Hellwigeafdc7d2010-06-04 11:29:53 +02002659
Omar Sandoval17f8c842015-03-16 04:33:50 -07002660static inline ssize_t blockdev_direct_IO(struct kiocb *iocb,
2661 struct inode *inode,
2662 struct iov_iter *iter, loff_t offset,
2663 get_block_t get_block)
Linus Torvalds1da177e2005-04-16 15:20:36 -07002664{
Omar Sandoval17f8c842015-03-16 04:33:50 -07002665 return __blockdev_direct_IO(iocb, inode, inode->i_sb->s_bdev, iter,
Al Viro31b14032014-03-05 01:33:16 -05002666 offset, get_block, NULL, NULL,
Christoph Hellwig1e431f52009-11-03 16:44:53 +01002667 DIO_LOCKING | DIO_SKIP_HOLES);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002668}
David Howells93614012006-09-30 20:45:40 +02002669#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -07002670
Trond Myklebust1d59d612012-05-31 12:22:33 -04002671void inode_dio_wait(struct inode *inode);
Jens Axboefe0f07d2015-04-15 17:05:48 -06002672
2673/*
2674 * inode_dio_begin - signal start of a direct I/O requests
2675 * @inode: inode the direct I/O happens on
2676 *
2677 * This is called once we've finished processing a direct I/O request,
2678 * and is used to wake up callers waiting for direct I/O to be quiesced.
2679 */
2680static inline void inode_dio_begin(struct inode *inode)
2681{
2682 atomic_inc(&inode->i_dio_count);
2683}
2684
2685/*
2686 * inode_dio_end - signal finish of a direct I/O requests
2687 * @inode: inode the direct I/O happens on
2688 *
2689 * This is called once we've finished processing a direct I/O request,
2690 * and is used to wake up callers waiting for direct I/O to be quiesced.
2691 */
2692static inline void inode_dio_end(struct inode *inode)
2693{
2694 if (atomic_dec_and_test(&inode->i_dio_count))
2695 wake_up_bit(&inode->i_state, __I_DIO_WAKEUP);
2696}
Trond Myklebust1d59d612012-05-31 12:22:33 -04002697
Theodore Ts'o5f16f322014-03-24 14:43:12 -04002698extern void inode_set_flags(struct inode *inode, unsigned int flags,
2699 unsigned int mask);
2700
Arjan van de Ven4b6f5d22006-03-28 01:56:42 -08002701extern const struct file_operations generic_ro_fops;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002702
2703#define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
2704
Al Viro5d826c82014-03-14 13:42:45 -04002705extern int readlink_copy(char __user *, int, const char *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002706extern int page_readlink(struct dentry *, char __user *, int);
Linus Torvaldscc314ee2005-08-19 18:02:56 -07002707extern void *page_follow_link_light(struct dentry *, struct nameidata *);
2708extern void page_put_link(struct dentry *, struct nameidata *, void *);
Kirill Korotaev0adb25d2006-03-11 03:27:13 -08002709extern int __page_symlink(struct inode *inode, const char *symname, int len,
Nick Piggin54566b22009-01-04 12:00:53 -08002710 int nofs);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002711extern int page_symlink(struct inode *inode, const char *symname, int len);
Arjan van de Venc5ef1c42007-02-12 00:55:40 -08002712extern const struct inode_operations page_symlink_inode_operations;
Al Viro87dc8002013-09-16 10:30:04 -04002713extern void kfree_put_link(struct dentry *, struct nameidata *, void *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002714extern int generic_readlink(struct dentry *, char __user *, int);
2715extern void generic_fillattr(struct inode *, struct kstat *);
J. Bruce Fieldsb7a6ec52013-10-02 17:01:18 -04002716int vfs_getattr_nosec(struct path *path, struct kstat *stat);
Al Viro3dadecc2013-01-24 02:18:08 -05002717extern int vfs_getattr(struct path *, struct kstat *);
Dmitry Monakhovb4627072009-12-14 15:21:12 +03002718void __inode_add_bytes(struct inode *inode, loff_t bytes);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002719void inode_add_bytes(struct inode *inode, loff_t bytes);
Jan Kara1c8924e2013-08-17 09:32:32 -04002720void __inode_sub_bytes(struct inode *inode, loff_t bytes);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002721void inode_sub_bytes(struct inode *inode, loff_t bytes);
2722loff_t inode_get_bytes(struct inode *inode);
2723void inode_set_bytes(struct inode *inode, loff_t bytes);
2724
Al Viro5c0ba4e2013-05-15 13:52:59 -04002725extern int iterate_dir(struct file *, struct dir_context *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002726
David Howellsc7887322010-08-11 11:26:22 +01002727extern int vfs_stat(const char __user *, struct kstat *);
2728extern int vfs_lstat(const char __user *, struct kstat *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002729extern int vfs_fstat(unsigned int, struct kstat *);
David Howellsc7887322010-08-11 11:26:22 +01002730extern int vfs_fstatat(int , const char __user *, struct kstat *, int);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002731
Erez Zadokdeb21db2008-02-07 00:13:25 -08002732extern int do_vfs_ioctl(struct file *filp, unsigned int fd, unsigned int cmd,
2733 unsigned long arg);
Steven Whitehousee9079cc2008-10-14 14:43:29 +01002734extern int __generic_block_fiemap(struct inode *inode,
Josef Bacik3a3076f2010-04-23 12:17:17 -04002735 struct fiemap_extent_info *fieinfo,
2736 loff_t start, loff_t len,
2737 get_block_t *get_block);
Josef Bacik68c9d702008-10-03 17:32:43 -04002738extern int generic_block_fiemap(struct inode *inode,
2739 struct fiemap_extent_info *fieinfo, u64 start,
2740 u64 len, get_block_t *get_block);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002741
Miklos Szeredic18479f2007-10-18 23:39:11 -07002742extern void get_filesystem(struct file_system_type *fs);
2743extern void put_filesystem(struct file_system_type *fs);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002744extern struct file_system_type *get_fs_type(const char *name);
2745extern struct super_block *get_super(struct block_device *);
Jan Kara6b6dc832012-02-10 11:03:00 +01002746extern struct super_block *get_super_thawed(struct block_device *);
Christoph Hellwig45042302009-08-03 23:28:35 +02002747extern struct super_block *get_active_super(struct block_device *bdev);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002748extern void drop_super(struct super_block *sb);
Al Viro01a05b32010-03-23 06:06:58 -04002749extern void iterate_supers(void (*)(struct super_block *, void *), void *);
Al Viro43e15cd2011-06-03 20:16:57 -04002750extern void iterate_supers_type(struct file_system_type *,
2751 void (*)(struct super_block *, void *), void *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002752
2753extern int dcache_dir_open(struct inode *, struct file *);
2754extern int dcache_dir_close(struct inode *, struct file *);
2755extern loff_t dcache_dir_lseek(struct file *, loff_t, int);
Al Viro5f99f4e2013-05-15 20:23:06 -04002756extern int dcache_readdir(struct file *, struct dir_context *);
npiggin@suse.de7bb46a62010-05-27 01:05:33 +10002757extern int simple_setattr(struct dentry *, struct iattr *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002758extern int simple_getattr(struct vfsmount *, struct dentry *, struct kstat *);
David Howells726c3342006-06-23 02:02:58 -07002759extern int simple_statfs(struct dentry *, struct kstatfs *);
Stephen Boyd20955e82012-04-05 14:25:09 -07002760extern int simple_open(struct inode *inode, struct file *file);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002761extern int simple_link(struct dentry *, struct inode *, struct dentry *);
2762extern int simple_unlink(struct inode *, struct dentry *);
2763extern int simple_rmdir(struct inode *, struct dentry *);
2764extern int simple_rename(struct inode *, struct dentry *, struct inode *, struct dentry *);
Josef Bacik02c24a82011-07-16 20:44:56 -04002765extern int noop_fsync(struct file *, loff_t, loff_t, int);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002766extern int simple_empty(struct dentry *);
2767extern int simple_readpage(struct file *file, struct page *page);
Nick Pigginafddba42007-10-16 01:25:01 -07002768extern int simple_write_begin(struct file *file, struct address_space *mapping,
2769 loff_t pos, unsigned len, unsigned flags,
2770 struct page **pagep, void **fsdata);
2771extern int simple_write_end(struct file *file, struct address_space *mapping,
2772 loff_t pos, unsigned len, unsigned copied,
2773 struct page *page, void *fsdata);
Al Virob26d4cd2013-10-25 18:47:37 -04002774extern int always_delete_dentry(const struct dentry *);
Al Viro69878432013-10-02 22:35:11 -04002775extern struct inode *alloc_anon_inode(struct super_block *);
Jeff Laytone6f5c782014-08-22 10:40:25 -04002776extern int simple_nosetlease(struct file *, long, struct file_lock **, void **);
Al Virob26d4cd2013-10-25 18:47:37 -04002777extern const struct dentry_operations simple_dentry_operations;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002778
Al Viro00cd8dd2012-06-10 17:13:09 -04002779extern struct dentry *simple_lookup(struct inode *, struct dentry *, unsigned int flags);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002780extern ssize_t generic_read_dir(struct file *, char __user *, size_t, loff_t *);
Arjan van de Ven4b6f5d22006-03-28 01:56:42 -08002781extern const struct file_operations simple_dir_operations;
Arjan van de Venc5ef1c42007-02-12 00:55:40 -08002782extern const struct inode_operations simple_dir_inode_operations;
Arjan van de Ven4b6f5d22006-03-28 01:56:42 -08002783struct tree_descr { char *name; const struct file_operations *ops; int mode; };
Linus Torvalds1da177e2005-04-16 15:20:36 -07002784struct dentry *d_alloc_name(struct dentry *, const char *);
Roberto Sassu7d683a02010-06-03 11:58:28 +02002785extern int simple_fill_super(struct super_block *, unsigned long, struct tree_descr *);
Trond Myklebust1f5ce9e2006-06-09 09:34:16 -04002786extern int simple_pin_fs(struct file_system_type *, struct vfsmount **mount, int *count);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002787extern void simple_release_fs(struct vfsmount **mount, int *count);
2788
Akinobu Mita93b07112008-06-05 22:46:21 -07002789extern ssize_t simple_read_from_buffer(void __user *to, size_t count,
2790 loff_t *ppos, const void *from, size_t available);
Jiri Slaby6a727b42010-05-01 23:51:22 +02002791extern ssize_t simple_write_to_buffer(void *to, size_t available, loff_t *ppos,
2792 const void __user *from, size_t count);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002793
Fabian Frederickac13a822014-06-04 16:06:27 -07002794extern int __generic_file_fsync(struct file *, loff_t, loff_t, int);
Josef Bacik02c24a82011-07-16 20:44:56 -04002795extern int generic_file_fsync(struct file *, loff_t, loff_t, int);
Al Virod5aacad2009-06-07 14:56:44 -04002796
Patrick J. LoPresti30ca22c2010-07-22 15:03:41 -07002797extern int generic_check_addressable(unsigned, u64);
2798
Christoph Lametere965f962006-02-01 03:05:41 -08002799#ifdef CONFIG_MIGRATION
Christoph Lameter2d1db3b2006-06-23 02:03:33 -07002800extern int buffer_migrate_page(struct address_space *,
Mel Gormana6bc32b2012-01-12 17:19:43 -08002801 struct page *, struct page *,
2802 enum migrate_mode);
Christoph Lametere965f962006-02-01 03:05:41 -08002803#else
2804#define buffer_migrate_page NULL
2805#endif
2806
npiggin@suse.de25d9e2d2009-08-21 02:35:05 +10002807extern int inode_change_ok(const struct inode *, struct iattr *);
2808extern int inode_newsize_ok(const struct inode *, loff_t offset);
Christoph Hellwig6a1a90a2010-06-04 11:30:00 +02002809extern void setattr_copy(struct inode *inode, const struct iattr *attr);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002810
Josef Bacikc3b2da32012-03-26 09:59:21 -04002811extern int file_update_time(struct file *file);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002812
Al Viro34c80b12011-12-08 21:32:45 -05002813extern int generic_show_options(struct seq_file *m, struct dentry *root);
Miklos Szeredib3b304a2008-02-08 04:21:35 -08002814extern void save_mount_options(struct super_block *sb, char *options);
Al Viro2a32ceb2009-05-08 16:05:57 -04002815extern void replace_mount_options(struct super_block *sb, char *options);
Miklos Szeredib3b304a2008-02-08 04:21:35 -08002816
Matthew Wilcoxd475c632015-02-16 15:58:56 -08002817static inline bool io_is_direct(struct file *filp)
2818{
2819 return (filp->f_flags & O_DIRECT) || IS_DAX(file_inode(filp));
2820}
2821
Al Viro2ba48ce2015-04-09 13:52:01 -04002822static inline int iocb_flags(struct file *file)
2823{
2824 int res = 0;
2825 if (file->f_flags & O_APPEND)
2826 res |= IOCB_APPEND;
2827 if (io_is_direct(file))
2828 res |= IOCB_DIRECT;
2829 return res;
2830}
2831
Linus Torvalds1da177e2005-04-16 15:20:36 -07002832static inline ino_t parent_ino(struct dentry *dentry)
2833{
2834 ino_t res;
2835
Nick Pigginb5c84bf2011-01-07 17:49:38 +11002836 /*
2837 * Don't strictly need d_lock here? If the parent ino could change
2838 * then surely we'd have a deeper race in the caller?
2839 */
Linus Torvalds1da177e2005-04-16 15:20:36 -07002840 spin_lock(&dentry->d_lock);
2841 res = dentry->d_parent->d_inode->i_ino;
2842 spin_unlock(&dentry->d_lock);
2843 return res;
2844}
2845
Linus Torvalds1da177e2005-04-16 15:20:36 -07002846/* Transaction based IO helpers */
2847
2848/*
2849 * An argresp is stored in an allocated page and holds the
2850 * size of the argument or response, along with its content
2851 */
2852struct simple_transaction_argresp {
2853 ssize_t size;
2854 char data[0];
2855};
2856
2857#define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
2858
2859char *simple_transaction_get(struct file *file, const char __user *buf,
2860 size_t size);
2861ssize_t simple_transaction_read(struct file *file, char __user *buf,
2862 size_t size, loff_t *pos);
2863int simple_transaction_release(struct inode *inode, struct file *file);
2864
Ingo Molnar76791ab2009-03-25 16:48:35 +01002865void simple_transaction_set(struct file *file, size_t n);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002866
Arnd Bergmannacaefc22005-05-18 14:40:59 +02002867/*
2868 * simple attribute files
2869 *
2870 * These attributes behave similar to those in sysfs:
2871 *
2872 * Writing to an attribute immediately sets a value, an open file can be
2873 * written to multiple times.
2874 *
2875 * Reading from an attribute creates a buffer from the value that might get
2876 * read with multiple read calls. When the attribute has been read
2877 * completely, no further read calls are possible until the file is opened
2878 * again.
2879 *
2880 * All attributes contain a text representation of a numeric value
2881 * that are accessed with the get() and set() functions.
2882 */
2883#define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt) \
2884static int __fops ## _open(struct inode *inode, struct file *file) \
2885{ \
2886 __simple_attr_check_format(__fmt, 0ull); \
2887 return simple_attr_open(inode, file, __get, __set, __fmt); \
2888} \
Alexey Dobriyan828c0952009-10-01 15:43:56 -07002889static const struct file_operations __fops = { \
Arnd Bergmannacaefc22005-05-18 14:40:59 +02002890 .owner = THIS_MODULE, \
2891 .open = __fops ## _open, \
Christoph Hellwig74bedc42008-02-08 04:20:28 -08002892 .release = simple_attr_release, \
Arnd Bergmannacaefc22005-05-18 14:40:59 +02002893 .read = simple_attr_read, \
2894 .write = simple_attr_write, \
Arnd Bergmann1ec55842010-08-15 21:50:52 +02002895 .llseek = generic_file_llseek, \
Joe Perches68be3022014-08-06 16:08:45 -07002896}
Arnd Bergmannacaefc22005-05-18 14:40:59 +02002897
Joe Perchesb9075fa2011-10-31 17:11:33 -07002898static inline __printf(1, 2)
2899void __simple_attr_check_format(const char *fmt, ...)
Arnd Bergmannacaefc22005-05-18 14:40:59 +02002900{
2901 /* don't do anything, just let the compiler check the arguments; */
2902}
2903
2904int simple_attr_open(struct inode *inode, struct file *file,
Christoph Hellwig8b88b092008-02-08 04:20:26 -08002905 int (*get)(void *, u64 *), int (*set)(void *, u64),
Arnd Bergmannacaefc22005-05-18 14:40:59 +02002906 const char *fmt);
Christoph Hellwig74bedc42008-02-08 04:20:28 -08002907int simple_attr_release(struct inode *inode, struct file *file);
Arnd Bergmannacaefc22005-05-18 14:40:59 +02002908ssize_t simple_attr_read(struct file *file, char __user *buf,
2909 size_t len, loff_t *ppos);
2910ssize_t simple_attr_write(struct file *file, const char __user *buf,
2911 size_t len, loff_t *ppos);
2912
Alexey Dobriyan4be28542007-10-16 23:26:21 -07002913struct ctl_table;
Alexey Dobriyan8d65af72009-09-23 15:57:19 -07002914int proc_nr_files(struct ctl_table *table, int write,
Adrian Bunk62239ac2007-07-17 04:03:45 -07002915 void __user *buffer, size_t *lenp, loff_t *ppos);
Christoph Hellwig312d3ca2010-10-10 05:36:23 -04002916int proc_nr_dentry(struct ctl_table *table, int write,
2917 void __user *buffer, size_t *lenp, loff_t *ppos);
Dave Chinnercffbc8a2010-10-23 05:03:02 -04002918int proc_nr_inodes(struct ctl_table *table, int write,
2919 void __user *buffer, size_t *lenp, loff_t *ppos);
Tetsuo Handa38e23c92009-04-09 20:17:52 +09002920int __init get_filesystem_list(char *buf);
Adrian Bunk62239ac2007-07-17 04:03:45 -07002921
Namhyung Kim3cd90ea2011-02-01 15:52:46 -08002922#define __FMODE_EXEC ((__force int) FMODE_EXEC)
Namhyung Kim1a44bc82011-02-01 15:52:46 -08002923#define __FMODE_NONOTIFY ((__force int) FMODE_NONOTIFY)
2924
Al Viro6d125522009-12-24 06:58:56 -05002925#define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE])
Eric Parisecf081d2009-12-17 21:24:25 -05002926#define OPEN_FMODE(flag) ((__force fmode_t)(((flag + 1) & O_ACCMODE) | \
Namhyung Kim1a44bc82011-02-01 15:52:46 -08002927 (flag & __FMODE_NONOTIFY)))
Al Viro53009902009-12-19 10:15:07 -05002928
Al Viro8d334ac2011-07-24 23:21:59 -04002929static inline int is_sxid(umode_t mode)
Andi Kleen69b45732011-05-28 08:25:51 -07002930{
2931 return (mode & S_ISUID) || ((mode & S_ISGID) && (mode & S_IXGRP));
2932}
2933
Miklos Szeredicbdf35b2014-10-24 00:14:36 +02002934static inline int check_sticky(struct inode *dir, struct inode *inode)
2935{
2936 if (!(dir->i_mode & S_ISVTX))
2937 return 0;
2938
2939 return __check_sticky(dir, inode);
2940}
2941
Andi Kleen69b45732011-05-28 08:25:51 -07002942static inline void inode_has_no_xattr(struct inode *inode)
2943{
Al Viro9e1f1de2011-06-03 18:24:58 -04002944 if (!is_sxid(inode->i_mode) && (inode->i_sb->s_flags & MS_NOSEC))
Andi Kleen69b45732011-05-28 08:25:51 -07002945 inode->i_flags |= S_NOSEC;
2946}
2947
Al Viroa7400222014-10-21 15:20:42 -04002948static inline bool is_root_inode(struct inode *inode)
2949{
2950 return inode == inode->i_sb->s_root->d_inode;
2951}
2952
Al Viro5f99f4e2013-05-15 20:23:06 -04002953static inline bool dir_emit(struct dir_context *ctx,
2954 const char *name, int namelen,
2955 u64 ino, unsigned type)
2956{
2957 return ctx->actor(ctx, name, namelen, ctx->pos, ino, type) == 0;
2958}
2959static inline bool dir_emit_dot(struct file *file, struct dir_context *ctx)
2960{
2961 return ctx->actor(ctx, ".", 1, ctx->pos,
2962 file->f_path.dentry->d_inode->i_ino, DT_DIR) == 0;
2963}
2964static inline bool dir_emit_dotdot(struct file *file, struct dir_context *ctx)
2965{
2966 return ctx->actor(ctx, "..", 2, ctx->pos,
2967 parent_ino(file->f_path.dentry), DT_DIR) == 0;
2968}
2969static inline bool dir_emit_dots(struct file *file, struct dir_context *ctx)
2970{
2971 if (ctx->pos == 0) {
2972 if (!dir_emit_dot(file, ctx))
2973 return false;
2974 ctx->pos = 1;
2975 }
2976 if (ctx->pos == 1) {
2977 if (!dir_emit_dotdot(file, ctx))
2978 return false;
2979 ctx->pos = 2;
2980 }
2981 return true;
2982}
Al Viro5ded75e2013-05-15 21:02:48 -04002983static inline bool dir_relax(struct inode *inode)
2984{
2985 mutex_unlock(&inode->i_mutex);
2986 mutex_lock(&inode->i_mutex);
2987 return !IS_DEADDIR(inode);
2988}
Al Viro5f99f4e2013-05-15 20:23:06 -04002989
Linus Torvalds1da177e2005-04-16 15:20:36 -07002990#endif /* _LINUX_FS_H */