blob: dc0478c07b2abd3887d7f5b1b84818a4ee24162e [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#include <linux/linkage.h>
5#include <linux/wait.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -07006#include <linux/kdev_t.h>
7#include <linux/dcache.h>
Al Viro3f8206d2008-07-26 03:46:43 -04008#include <linux/path.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -07009#include <linux/stat.h>
10#include <linux/cache.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070011#include <linux/list.h>
Dave Chinnerbc3b14c2013-08-28 10:17:58 +100012#include <linux/list_lru.h>
Oleg Nesterov4f5e65a2013-07-08 14:24:16 -070013#include <linux/llist.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070014#include <linux/radix-tree.h>
Michel Lespinasse6b2dbba2012-10-08 16:31:25 -070015#include <linux/rbtree.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070016#include <linux/init.h>
Al Viro914e2632006-10-18 13:55:46 -040017#include <linux/pid.h>
Paul Gortmaker187f1882011-11-23 20:12:59 -050018#include <linux/bug.h>
Jes Sorensen1b1dcc12006-01-09 15:59:24 -080019#include <linux/mutex.h>
Davidlohr Buesoc8c06ef2014-12-12 16:54:24 -080020#include <linux/rwsem.h>
Satyam Sharma3bd858a2007-07-17 15:00:08 +053021#include <linux/capability.h>
Matthew Wilcox6188e102008-04-18 22:21:05 -040022#include <linux/semaphore.h>
Mark Fashehc4b929b2008-10-08 19:44:18 -040023#include <linux/fiemap.h>
Nick Pigginceb5bdc2011-01-07 17:50:05 +110024#include <linux/rculist_bl.h>
Al Viro07b8ce12011-06-20 10:52:57 -040025#include <linux/atomic.h>
Konstantin Khlebnikov83aeead2011-12-08 14:33:54 -080026#include <linux/shrinker.h>
Stephen Rothwellc1aab022012-01-24 11:41:32 +110027#include <linux/migrate_mode.h>
Eric W. Biederman92361632012-02-08 07:07:50 -080028#include <linux/uidgid.h>
Jan Kara5accdf82012-06-12 16:20:34 +020029#include <linux/lockdep.h>
Jens Axboec2b1ad82012-09-27 09:35:03 +020030#include <linux/percpu-rwsem.h>
David Howells08cce05c2012-10-13 10:45:06 +010031#include <linux/blk_types.h>
Oleg Nesterov853b39a2015-07-22 20:21:13 +020032#include <linux/workqueue.h>
Oleg Nesterov8129ed22015-08-11 17:05:04 +020033#include <linux/percpu-rwsem.h>
Al Virofceef392015-12-29 15:58:39 -050034#include <linux/delayed_call.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070035
Linus Torvalds1da177e2005-04-16 15:20:36 -070036#include <asm/byteorder.h>
David Howells607ca462012-10-13 10:46:48 +010037#include <uapi/linux/fs.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070038
Christoph Hellwigb83ae6d2015-01-14 10:42:37 +010039struct backing_dev_info;
Tejun Heo52ebea72015-05-22 17:13:37 -040040struct bdi_writeback;
Christoph Hellwiga5694252007-07-17 04:04:28 -070041struct export_operations;
Christoph Hellwiga885c8c2006-01-08 01:02:50 -080042struct hd_geometry;
Linus Torvalds1da177e2005-04-16 15:20:36 -070043struct iovec;
Christoph Hellwig92198f72005-06-23 22:00:59 -070044struct kiocb;
Alexey Dobriyan57cc7212011-01-10 08:18:25 +020045struct kobject;
Linus Torvalds1da177e2005-04-16 15:20:36 -070046struct pipe_inode_info;
47struct poll_table_struct;
48struct kstatfs;
49struct vm_area_struct;
50struct vfsmount;
David Howells745ca242008-11-14 10:39:22 +110051struct cred;
Mel Gormana509bc12012-07-31 16:44:57 -070052struct swap_info_struct;
Cyrill Gorcunov55985dd2012-12-17 16:04:55 -080053struct seq_file;
Christoph Hellwig7b7a8662013-09-04 15:04:39 +020054struct workqueue_struct;
Al Viro3b93f912014-02-11 21:34:08 -050055struct iov_iter;
Jaegeuk Kim0b81d072015-05-15 16:26:10 -070056struct fscrypt_info;
57struct fscrypt_operations;
Linus Torvalds1da177e2005-04-16 15:20:36 -070058
Denis Cheng74bf17c2007-10-16 23:26:30 -070059extern void __init inode_init(void);
Linus Torvalds1da177e2005-04-16 15:20:36 -070060extern void __init inode_init_early(void);
Mel Gorman4248b0d2015-08-06 15:46:20 -070061extern void __init files_init(void);
62extern void __init files_maxfiles_init(void);
Linus Torvalds1da177e2005-04-16 15:20:36 -070063
Jan Engelhardtdded4f42008-12-01 14:34:50 -080064extern struct files_stat_struct files_stat;
Eric Dumazet518de9b2010-10-26 14:22:44 -070065extern unsigned long get_max_files(void);
Alexey Dobriyan9b80a182016-09-02 00:38:52 +030066extern unsigned int sysctl_nr_open;
Jan Engelhardtdded4f42008-12-01 14:34:50 -080067extern struct inodes_stat_t inodes_stat;
68extern int leases_enable, lease_break_time;
Kees Cook800179c2012-07-25 17:29:07 -070069extern int sysctl_protected_symlinks;
70extern int sysctl_protected_hardlinks;
Jan Engelhardtdded4f42008-12-01 14:34:50 -080071
Linus Torvalds1da177e2005-04-16 15:20:36 -070072struct buffer_head;
73typedef int (get_block_t)(struct inode *inode, sector_t iblock,
74 struct buffer_head *bh_result, int create);
Christoph Hellwig187372a2016-02-08 14:40:51 +110075typedef int (dio_iodone_t)(struct kiocb *iocb, loff_t offset,
Christoph Hellwig7b7a8662013-09-04 15:04:39 +020076 ssize_t bytes, void *private);
Linus Torvalds1da177e2005-04-16 15:20:36 -070077
David Howellsbbc10962012-10-15 16:40:35 +010078#define MAY_EXEC 0x00000001
79#define MAY_WRITE 0x00000002
80#define MAY_READ 0x00000004
81#define MAY_APPEND 0x00000008
82#define MAY_ACCESS 0x00000010
83#define MAY_OPEN 0x00000020
84#define MAY_CHDIR 0x00000040
85/* called from RCU mode, don't block */
86#define MAY_NOT_BLOCK 0x00000080
87
88/*
89 * flags in file.f_mode. Note that FMODE_READ and FMODE_WRITE must correspond
90 * to O_WRONLY and O_RDWR via the strange trick in __dentry_open()
91 */
92
93/* file is open for reading */
94#define FMODE_READ ((__force fmode_t)0x1)
95/* file is open for writing */
96#define FMODE_WRITE ((__force fmode_t)0x2)
97/* file is seekable */
98#define FMODE_LSEEK ((__force fmode_t)0x4)
99/* file can be accessed using pread */
100#define FMODE_PREAD ((__force fmode_t)0x8)
101/* file can be accessed using pwrite */
102#define FMODE_PWRITE ((__force fmode_t)0x10)
103/* File is opened for execution with sys_execve / sys_uselib */
104#define FMODE_EXEC ((__force fmode_t)0x20)
105/* File is opened with O_NDELAY (only set for block devices) */
106#define FMODE_NDELAY ((__force fmode_t)0x40)
107/* File is opened with O_EXCL (only set for block devices) */
108#define FMODE_EXCL ((__force fmode_t)0x80)
109/* File is opened using open(.., 3, ..) and is writeable only for ioctls
110 (specialy hack for floppy.c) */
111#define FMODE_WRITE_IOCTL ((__force fmode_t)0x100)
112/* 32bit hashes as llseek() offset (for directories) */
113#define FMODE_32BITHASH ((__force fmode_t)0x200)
114/* 64bit hashes as llseek() offset (for directories) */
115#define FMODE_64BITHASH ((__force fmode_t)0x400)
116
117/*
118 * Don't update ctime and mtime.
119 *
120 * Currently a special hack for the XFS open_by_handle ioctl, but we'll
121 * hopefully graduate it to a proper O_CMTIME flag supported by open(2) soon.
122 */
123#define FMODE_NOCMTIME ((__force fmode_t)0x800)
124
125/* Expect random access pattern */
126#define FMODE_RANDOM ((__force fmode_t)0x1000)
127
128/* File is huge (eg. /dev/kmem): treat loff_t as unsigned */
129#define FMODE_UNSIGNED_OFFSET ((__force fmode_t)0x2000)
130
131/* File is opened with O_PATH; almost nothing can be done with it */
132#define FMODE_PATH ((__force fmode_t)0x4000)
133
Linus Torvalds9c225f22014-03-03 09:36:58 -0800134/* File needs atomic accesses to f_pos */
135#define FMODE_ATOMIC_POS ((__force fmode_t)0x8000)
Al Viro83f936c2014-03-14 12:02:47 -0400136/* Write access to underlying fs */
137#define FMODE_WRITER ((__force fmode_t)0x10000)
Al Viro7f7f25e2014-02-11 17:49:24 -0500138/* Has read method(s) */
139#define FMODE_CAN_READ ((__force fmode_t)0x20000)
140/* Has write method(s) */
141#define FMODE_CAN_WRITE ((__force fmode_t)0x40000)
Linus Torvalds9c225f22014-03-03 09:36:58 -0800142
David Howellsbbc10962012-10-15 16:40:35 +0100143/* File was opened by fanotify and shouldn't generate fanotify events */
David Drysdale75069f22015-01-08 14:32:29 -0800144#define FMODE_NONOTIFY ((__force fmode_t)0x4000000)
David Howellsbbc10962012-10-15 16:40:35 +0100145
146/*
147 * Flag for rw_copy_check_uvector and compat_rw_copy_check_uvector
148 * that indicates that they should check the contents of the iovec are
149 * valid, but not check the memory that the iovec elements
150 * points too.
151 */
152#define CHECK_IOVEC_ONLY -1
153
Linus Torvalds1da177e2005-04-16 15:20:36 -0700154/*
Mike Christie4e1b2d522016-06-05 14:32:22 -0500155 * The below are the various read and write flags that we support. Some of
David Howells08cce05c2012-10-13 10:45:06 +0100156 * them include behavioral modifiers that send information down to the
Mike Christie4e1b2d522016-06-05 14:32:22 -0500157 * block layer and IO scheduler. They should be used along with a req_op.
158 * Terminology:
David Howells08cce05c2012-10-13 10:45:06 +0100159 *
160 * The block layer uses device plugging to defer IO a little bit, in
161 * the hope that we will see more IO very shortly. This increases
162 * coalescing of adjacent IO and thus reduces the number of IOs we
163 * have to send to the device. It also allows for better queuing,
164 * if the IO isn't mergeable. If the caller is going to be waiting
165 * for the IO, then he must ensure that the device is unplugged so
166 * that the IO is dispatched to the driver.
167 *
168 * All IO is handled async in Linux. This is fine for background
169 * writes, but for reads or writes that someone waits for completion
170 * on, we want to notify the block layer and IO scheduler so that they
171 * know about it. That allows them to make better scheduling
172 * decisions. So when the below references 'sync' and 'async', it
173 * is referencing this priority hint.
174 *
175 * With that in mind, the available types are:
176 *
177 * READ A normal read operation. Device will be plugged.
178 * READ_SYNC A synchronous read. Device is not plugged, caller can
179 * immediately wait on this read without caring about
180 * unplugging.
David Howells08cce05c2012-10-13 10:45:06 +0100181 * WRITE A normal async write. Device will be plugged.
182 * WRITE_SYNC Synchronous write. Identical to WRITE, but passes down
183 * the hint that someone will be waiting on this IO
184 * shortly. The write equivalent of READ_SYNC.
185 * WRITE_ODIRECT Special case write for O_DIRECT only.
186 * WRITE_FLUSH Like WRITE_SYNC but with preceding cache flush.
187 * WRITE_FUA Like WRITE_SYNC but data is guaranteed to be on
188 * non-volatile media on completion.
189 * WRITE_FLUSH_FUA Combination of WRITE_FLUSH and FUA. The IO is preceded
190 * by a cache flush and data is guaranteed to be on
191 * non-volatile media on completion.
192 *
193 */
Mike Christie4e1b2d522016-06-05 14:32:22 -0500194#define RW_MASK REQ_OP_WRITE
David Howells08cce05c2012-10-13 10:45:06 +0100195
Mike Christie4e1b2d522016-06-05 14:32:22 -0500196#define READ REQ_OP_READ
Christoph Hellwig70246282016-07-19 11:28:41 +0200197#define WRITE REQ_OP_WRITE
David Howells08cce05c2012-10-13 10:45:06 +0100198
Mike Christie4e1b2d522016-06-05 14:32:22 -0500199#define READ_SYNC REQ_SYNC
200#define WRITE_SYNC (REQ_SYNC | REQ_NOIDLE)
201#define WRITE_ODIRECT REQ_SYNC
Mike Christie28a8f0d2016-06-05 14:32:25 -0500202#define WRITE_FLUSH (REQ_SYNC | REQ_NOIDLE | REQ_PREFLUSH)
Mike Christie4e1b2d522016-06-05 14:32:22 -0500203#define WRITE_FUA (REQ_SYNC | REQ_NOIDLE | REQ_FUA)
Mike Christie28a8f0d2016-06-05 14:32:25 -0500204#define WRITE_FLUSH_FUA (REQ_SYNC | REQ_NOIDLE | REQ_PREFLUSH | REQ_FUA)
David Howells08cce05c2012-10-13 10:45:06 +0100205
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)
Miklos Szeredif2b20f62016-09-16 12:44:20 +0200227#define ATTR_TOUCH (1 << 17)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700228
229/*
Miklos Szeredi787fb6b2014-10-24 00:14:36 +0200230 * Whiteout is represented by a char device. The following constants define the
231 * mode and device number to use.
232 */
233#define WHITEOUT_MODE 0
234#define WHITEOUT_DEV 0
235
236/*
Linus Torvalds1da177e2005-04-16 15:20:36 -0700237 * This is the Inode Attributes structure, used for notify_change(). It
238 * uses the above definitions as flags, to know which values have changed.
239 * Also, in this manner, a Filesystem can look at only the values it cares
240 * about. Basically, these are the attributes that the VFS layer can
241 * request to change from the FS layer.
242 *
243 * Derek Atkins <warlord@MIT.EDU> 94-10-20
244 */
245struct iattr {
246 unsigned int ia_valid;
247 umode_t ia_mode;
Eric W. Biederman92361632012-02-08 07:07:50 -0800248 kuid_t ia_uid;
249 kgid_t ia_gid;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700250 loff_t ia_size;
251 struct timespec ia_atime;
252 struct timespec ia_mtime;
253 struct timespec ia_ctime;
Miklos Szeredicc4e69d2005-11-07 00:59:49 -0800254
255 /*
Lucas De Marchi25985ed2011-03-30 22:57:33 -0300256 * Not an attribute, but an auxiliary info for filesystems wanting to
Miklos Szeredicc4e69d2005-11-07 00:59:49 -0800257 * implement an ftruncate() like method. NOTE: filesystem should
258 * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL).
259 */
260 struct file *ia_file;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700261};
262
263/*
Linus Torvalds1da177e2005-04-16 15:20:36 -0700264 * Includes for diskquotas.
265 */
266#include <linux/quota.h>
267
Miklos Szeredi69c433e2014-10-24 00:14:39 +0200268/*
269 * Maximum number of layers of fs stack. Needs to be limited to
270 * prevent kernel stack overflow
271 */
272#define FILESYSTEM_MAX_STACK_DEPTH 2
273
Zach Brown994fc28c2005-12-15 14:28:17 -0800274/**
275 * enum positive_aop_returns - aop return codes with specific semantics
276 *
277 * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has
278 * completed, that the page is still locked, and
279 * should be considered active. The VM uses this hint
280 * to return the page to the active list -- it won't
281 * be a candidate for writeback again in the near
282 * future. Other callers must be careful to unlock
283 * the page if they get this return. Returned by
284 * writepage();
285 *
286 * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has
287 * unlocked it and the page might have been truncated.
288 * The caller should back up to acquiring a new page and
289 * trying again. The aop will be taking reasonable
290 * precautions not to livelock. If the caller held a page
291 * reference, it should drop it before retrying. Returned
Nick Piggin55144762007-10-16 01:25:26 -0700292 * by readpage().
Zach Brown994fc28c2005-12-15 14:28:17 -0800293 *
294 * address_space_operation functions return these large constants to indicate
295 * special semantics to the caller. These are much larger than the bytes in a
296 * page to allow for functions that return the number of bytes operated on in a
297 * given page.
298 */
299
300enum positive_aop_returns {
301 AOP_WRITEPAGE_ACTIVATE = 0x80000,
302 AOP_TRUNCATED_PAGE = 0x80001,
303};
304
Nick Pigginafddba42007-10-16 01:25:01 -0700305#define AOP_FLAG_UNINTERRUPTIBLE 0x0001 /* will not do a short write */
Nick Piggin89e10782007-10-16 01:25:07 -0700306#define AOP_FLAG_CONT_EXPAND 0x0002 /* called from cont_expand */
Nick Piggin54566b22009-01-04 12:00:53 -0800307#define AOP_FLAG_NOFS 0x0004 /* used by filesystem to direct
308 * helper code (eg buffer layer)
309 * to clear GFP_FS from alloc */
Nick Pigginafddba42007-10-16 01:25:01 -0700310
Linus Torvalds1da177e2005-04-16 15:20:36 -0700311/*
312 * oh the beauties of C type declarations.
313 */
314struct page;
315struct address_space;
316struct writeback_control;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700317
Christoph Hellwige2e40f22015-02-22 08:58:50 -0800318#define IOCB_EVENTFD (1 << 0)
Al Viro2ba48ce2015-04-09 13:52:01 -0400319#define IOCB_APPEND (1 << 1)
320#define IOCB_DIRECT (1 << 2)
Christoph Hellwig97be7eb2016-03-03 16:04:01 +0100321#define IOCB_HIPRI (1 << 3)
Christoph Hellwigdde0c2e2016-04-07 08:52:00 -0700322#define IOCB_DSYNC (1 << 4)
323#define IOCB_SYNC (1 << 5)
Jan Kara70fe2f42016-10-30 11:42:04 -0500324#define IOCB_WRITE (1 << 6)
Christoph Hellwige2e40f22015-02-22 08:58:50 -0800325
326struct kiocb {
327 struct file *ki_filp;
328 loff_t ki_pos;
329 void (*ki_complete)(struct kiocb *iocb, long ret, long ret2);
330 void *private;
331 int ki_flags;
332};
333
334static inline bool is_sync_kiocb(struct kiocb *kiocb)
335{
336 return kiocb->ki_complete == NULL;
337}
338
Al Viro2ba48ce2015-04-09 13:52:01 -0400339static inline int iocb_flags(struct file *file);
340
Christoph Hellwige2e40f22015-02-22 08:58:50 -0800341static inline void init_sync_kiocb(struct kiocb *kiocb, struct file *filp)
342{
343 *kiocb = (struct kiocb) {
344 .ki_filp = filp,
Al Viro2ba48ce2015-04-09 13:52:01 -0400345 .ki_flags = iocb_flags(filp),
Christoph Hellwige2e40f22015-02-22 08:58:50 -0800346 };
347}
348
Hisashi Hifumi8ab22b92008-07-28 15:46:36 -0700349/*
350 * "descriptor" for what we're up to with a read.
351 * This allows us to use the same read code yet
352 * have multiple different users of the data that
353 * we read from a file.
354 *
355 * The simplest case just copies the data to user
356 * mode.
357 */
358typedef struct {
359 size_t written;
360 size_t count;
361 union {
362 char __user *buf;
363 void *data;
364 } arg;
365 int error;
366} read_descriptor_t;
367
368typedef int (*read_actor_t)(read_descriptor_t *, struct page *,
369 unsigned long, unsigned long);
Nick Piggin2f718ff2007-10-16 01:24:59 -0700370
Linus Torvalds1da177e2005-04-16 15:20:36 -0700371struct address_space_operations {
372 int (*writepage)(struct page *page, struct writeback_control *wbc);
373 int (*readpage)(struct file *, struct page *);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700374
375 /* Write back some dirty pages from this mapping. */
376 int (*writepages)(struct address_space *, struct writeback_control *);
377
Andrew Morton4741c9f2006-03-24 03:18:11 -0800378 /* Set a page dirty. Return true if this dirtied it */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700379 int (*set_page_dirty)(struct page *page);
380
381 int (*readpages)(struct file *filp, struct address_space *mapping,
382 struct list_head *pages, unsigned nr_pages);
383
Nick Pigginafddba42007-10-16 01:25:01 -0700384 int (*write_begin)(struct file *, struct address_space *mapping,
385 loff_t pos, unsigned len, unsigned flags,
386 struct page **pagep, void **fsdata);
387 int (*write_end)(struct file *, struct address_space *mapping,
388 loff_t pos, unsigned len, unsigned copied,
389 struct page *page, void *fsdata);
390
Linus Torvalds1da177e2005-04-16 15:20:36 -0700391 /* Unfortunately this kludge is needed for FIBMAP. Don't use it */
392 sector_t (*bmap)(struct address_space *, sector_t);
Lukas Czernerd47992f2013-05-21 23:17:23 -0400393 void (*invalidatepage) (struct page *, unsigned int, unsigned int);
Al Viro27496a82005-10-21 03:20:48 -0400394 int (*releasepage) (struct page *, gfp_t);
Linus Torvalds6072d132010-12-01 13:35:19 -0500395 void (*freepage)(struct page *);
Christoph Hellwigc8b8e322016-04-07 08:51:58 -0700396 ssize_t (*direct_IO)(struct kiocb *, struct iov_iter *iter);
Mel Gormanb969c4ab2012-01-12 17:19:34 -0800397 /*
Tang Chenef277c72013-07-08 16:00:21 -0700398 * migrate the contents of a page to the specified target. If
399 * migrate_mode is MIGRATE_ASYNC, it must not block.
Mel Gormanb969c4ab2012-01-12 17:19:34 -0800400 */
Christoph Lameter2d1db3b2006-06-23 02:03:33 -0700401 int (*migratepage) (struct address_space *,
Mel Gormana6bc32b2012-01-12 17:19:43 -0800402 struct page *, struct page *, enum migrate_mode);
Minchan Kimbda807d2016-07-26 15:23:05 -0700403 bool (*isolate_page)(struct page *, isolate_mode_t);
404 void (*putback_page)(struct page *);
Trond Myklebuste3db7692007-01-10 23:15:39 -0800405 int (*launder_page) (struct page *);
Al Viroc186afb42014-02-02 21:16:54 -0500406 int (*is_partially_uptodate) (struct page *, unsigned long,
Hisashi Hifumi8ab22b92008-07-28 15:46:36 -0700407 unsigned long);
Mel Gormanb4597222013-07-03 15:02:05 -0700408 void (*is_dirty_writeback) (struct page *, bool *, bool *);
Andi Kleen25718732009-09-16 11:50:13 +0200409 int (*error_remove_page)(struct address_space *, struct page *);
Mel Gorman62c230b2012-07-31 16:44:55 -0700410
411 /* swapfile support */
Mel Gormana509bc12012-07-31 16:44:57 -0700412 int (*swap_activate)(struct swap_info_struct *sis, struct file *file,
413 sector_t *span);
414 void (*swap_deactivate)(struct file *file);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700415};
416
Jens Axboe7dcda1c2011-04-05 23:51:48 +0200417extern const struct address_space_operations empty_aops;
418
Nick Pigginafddba42007-10-16 01:25:01 -0700419/*
420 * pagecache_write_begin/pagecache_write_end must be used by general code
421 * to write into the pagecache.
422 */
423int pagecache_write_begin(struct file *, struct address_space *mapping,
424 loff_t pos, unsigned len, unsigned flags,
425 struct page **pagep, void **fsdata);
426
427int pagecache_write_end(struct file *, struct address_space *mapping,
428 loff_t pos, unsigned len, unsigned copied,
429 struct page *page, void *fsdata);
430
Linus Torvalds1da177e2005-04-16 15:20:36 -0700431struct address_space {
432 struct inode *host; /* owner: inode, block_device */
433 struct radix_tree_root page_tree; /* radix tree of all pages */
Nick Piggin19fd6232008-07-25 19:45:32 -0700434 spinlock_t tree_lock; /* and lock protecting it */
David Herrmann4bb5f5d2014-08-08 14:25:25 -0700435 atomic_t i_mmap_writable;/* count VM_SHARED mappings */
Michel Lespinasse6b2dbba2012-10-08 16:31:25 -0700436 struct rb_root i_mmap; /* tree of private and shared mappings */
Davidlohr Buesoc8c06ef2014-12-12 16:54:24 -0800437 struct rw_semaphore i_mmap_rwsem; /* protect tree, count, list */
Jan Kara08142572011-06-27 16:18:10 -0700438 /* Protected by tree_lock together with the radix tree */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700439 unsigned long nrpages; /* number of total pages */
Ross Zwislerf9fe48b2016-01-22 15:10:40 -0800440 /* number of shadow or DAX exceptional entries */
441 unsigned long nrexceptional;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700442 pgoff_t writeback_index;/* writeback starts here */
Christoph Hellwigf5e54d62006-06-28 04:26:44 -0700443 const struct address_space_operations *a_ops; /* methods */
Michal Hocko9c5d7602016-10-11 13:56:04 -0700444 unsigned long flags; /* error bits */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700445 spinlock_t private_lock; /* for use by the address_space */
Michal Hocko9c5d7602016-10-11 13:56:04 -0700446 gfp_t gfp_mask; /* implicit gfp mask for allocations */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700447 struct list_head private_list; /* ditto */
Rafael Aquini252aa6f2012-12-11 16:02:35 -0800448 void *private_data; /* ditto */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700449} __attribute__((aligned(sizeof(long))));
450 /*
451 * On most architectures that alignment is already the case; but
Lucas De Marchi25985ed2011-03-30 22:57:33 -0300452 * must be enforced here for CRIS, to let the least significant bit
Linus Torvalds1da177e2005-04-16 15:20:36 -0700453 * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
454 */
Andi Kleen87192a22012-01-12 17:20:34 -0800455struct request_queue;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700456
457struct block_device {
458 dev_t bd_dev; /* not a kdev_t - it's a search key */
Richard Kennedyf3ccfcd2011-03-22 16:33:56 -0700459 int bd_openers;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700460 struct inode * bd_inode; /* will die */
Theodore Ts'o87d8fe12009-01-03 09:47:09 -0500461 struct super_block * bd_super;
Arjan van de Venc039e312006-03-23 03:00:28 -0800462 struct mutex bd_mutex; /* open/close mutex */
Tejun Heo6b4517a2010-04-07 18:53:59 +0900463 void * bd_claiming;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700464 void * bd_holder;
465 int bd_holders;
Tejun Heo77ea8872010-12-08 20:57:37 +0100466 bool bd_write_holder;
Jun'ichi Nomura641dc632006-03-27 01:17:57 -0800467#ifdef CONFIG_SYSFS
Tejun Heo49731ba2011-01-14 18:43:57 +0100468 struct list_head bd_holder_disks;
Jun'ichi Nomura641dc632006-03-27 01:17:57 -0800469#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -0700470 struct block_device * bd_contains;
471 unsigned bd_block_size;
472 struct hd_struct * bd_part;
473 /* number of times partitions within this device have been opened. */
474 unsigned bd_part_count;
475 int bd_invalidated;
476 struct gendisk * bd_disk;
Andi Kleen87192a22012-01-12 17:20:34 -0800477 struct request_queue * bd_queue;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700478 struct list_head bd_list;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700479 /*
480 * Private data. You must have bd_claim'ed the block_device
481 * to use this. NOTE: bd_claim allows an owner to claim
482 * the same device multiple times, the owner must take special
483 * care to not mess up bd_private for that case.
484 */
485 unsigned long bd_private;
Takashi Satofcccf502009-01-09 16:40:59 -0800486
487 /* The counter of freeze processes */
488 int bd_fsfreeze_count;
489 /* Mutex for freeze */
490 struct mutex bd_fsfreeze_mutex;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700491};
492
493/*
494 * Radix-tree tags, for tagging dirty and writeback pages within the pagecache
495 * radix trees
496 */
497#define PAGECACHE_TAG_DIRTY 0
498#define PAGECACHE_TAG_WRITEBACK 1
Jan Karaf446daae2010-08-09 17:19:12 -0700499#define PAGECACHE_TAG_TOWRITE 2
Linus Torvalds1da177e2005-04-16 15:20:36 -0700500
501int mapping_tagged(struct address_space *mapping, int tag);
502
Davidlohr Bueso8b28f622014-12-12 16:54:18 -0800503static inline void i_mmap_lock_write(struct address_space *mapping)
504{
Davidlohr Buesoc8c06ef2014-12-12 16:54:24 -0800505 down_write(&mapping->i_mmap_rwsem);
Davidlohr Bueso8b28f622014-12-12 16:54:18 -0800506}
507
508static inline void i_mmap_unlock_write(struct address_space *mapping)
509{
Davidlohr Buesoc8c06ef2014-12-12 16:54:24 -0800510 up_write(&mapping->i_mmap_rwsem);
Davidlohr Bueso8b28f622014-12-12 16:54:18 -0800511}
512
Davidlohr Bueso3dec0ba2014-12-12 16:54:27 -0800513static inline void i_mmap_lock_read(struct address_space *mapping)
514{
515 down_read(&mapping->i_mmap_rwsem);
516}
517
518static inline void i_mmap_unlock_read(struct address_space *mapping)
519{
520 up_read(&mapping->i_mmap_rwsem);
521}
522
Linus Torvalds1da177e2005-04-16 15:20:36 -0700523/*
524 * Might pages of this file be mapped into userspace?
525 */
526static inline int mapping_mapped(struct address_space *mapping)
527{
Kirill A. Shutemov27ba0642015-02-10 14:09:59 -0800528 return !RB_EMPTY_ROOT(&mapping->i_mmap);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700529}
530
531/*
532 * Might pages of this file have been modified in userspace?
533 * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap_pgoff
534 * marks vma as VM_SHARED if it is shared, and the file was opened for
535 * writing i.e. vma may be mprotected writable even if now readonly.
David Herrmann4bb5f5d2014-08-08 14:25:25 -0700536 *
537 * If i_mmap_writable is negative, no new writable mappings are allowed. You
538 * can only deny writable mappings, if none exists right now.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700539 */
540static inline int mapping_writably_mapped(struct address_space *mapping)
541{
David Herrmann4bb5f5d2014-08-08 14:25:25 -0700542 return atomic_read(&mapping->i_mmap_writable) > 0;
543}
544
545static inline int mapping_map_writable(struct address_space *mapping)
546{
547 return atomic_inc_unless_negative(&mapping->i_mmap_writable) ?
548 0 : -EPERM;
549}
550
551static inline void mapping_unmap_writable(struct address_space *mapping)
552{
553 atomic_dec(&mapping->i_mmap_writable);
554}
555
556static inline int mapping_deny_writable(struct address_space *mapping)
557{
558 return atomic_dec_unless_positive(&mapping->i_mmap_writable) ?
559 0 : -EBUSY;
560}
561
562static inline void mapping_allow_writable(struct address_space *mapping)
563{
564 atomic_inc(&mapping->i_mmap_writable);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700565}
566
567/*
568 * Use sequence counter to get consistent i_size on 32-bit processors.
569 */
570#if BITS_PER_LONG==32 && defined(CONFIG_SMP)
571#include <linux/seqlock.h>
572#define __NEED_I_SIZE_ORDERED
573#define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
574#else
575#define i_size_ordered_init(inode) do { } while (0)
576#endif
577
Al Virof19d4a82009-06-08 19:50:45 -0400578struct posix_acl;
579#define ACL_NOT_CACHED ((void *)(-1))
Miklos Szeredi2a3a2a32016-09-01 11:11:59 +0200580#define ACL_DONT_CACHE ((void *)(-3))
Al Virof19d4a82009-06-08 19:50:45 -0400581
Andreas Gruenbacherb8a7a3a2016-03-24 14:38:37 +0100582static inline struct posix_acl *
583uncached_acl_sentinel(struct task_struct *task)
584{
585 return (void *)task + 1;
586}
587
588static inline bool
589is_uncached_acl(struct posix_acl *acl)
590{
591 return (long)acl & 1;
592}
593
Linus Torvalds3ddcd052011-08-06 22:45:50 -0700594#define IOP_FASTPERM 0x0001
595#define IOP_LOOKUP 0x0002
596#define IOP_NOFOLLOW 0x0004
Andreas Gruenbacherd0a5b992016-09-29 17:48:39 +0200597#define IOP_XATTR 0x0008
Linus Torvalds3ddcd052011-08-06 22:45:50 -0700598
599/*
600 * Keep mostly read-only and often accessed (especially for
601 * the RCU path lookup and 'stat' data) fields at the beginning
602 * of the 'struct inode'
603 */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700604struct inode {
Nick Piggin44a7d7a2011-01-07 17:49:56 +1100605 umode_t i_mode;
Linus Torvalds3ddcd052011-08-06 22:45:50 -0700606 unsigned short i_opflags;
Eric W. Biederman92361632012-02-08 07:07:50 -0800607 kuid_t i_uid;
608 kgid_t i_gid;
Linus Torvalds3ddcd052011-08-06 22:45:50 -0700609 unsigned int i_flags;
610
611#ifdef CONFIG_FS_POSIX_ACL
612 struct posix_acl *i_acl;
613 struct posix_acl *i_default_acl;
614#endif
615
Nick Piggin44a7d7a2011-01-07 17:49:56 +1100616 const struct inode_operations *i_op;
617 struct super_block *i_sb;
Linus Torvalds3ddcd052011-08-06 22:45:50 -0700618 struct address_space *i_mapping;
Nick Piggin44a7d7a2011-01-07 17:49:56 +1100619
Linus Torvalds13e12d12011-06-08 15:18:19 -0700620#ifdef CONFIG_SECURITY
621 void *i_security;
622#endif
Nick Piggin44a7d7a2011-01-07 17:49:56 +1100623
Linus Torvalds3ddcd052011-08-06 22:45:50 -0700624 /* Stat data, not accessed from path walking */
625 unsigned long i_ino;
Miklos Szeredia78ef702011-10-28 14:13:30 +0200626 /*
627 * Filesystems may only read i_nlink directly. They shall use the
628 * following functions for modification:
629 *
630 * (set|clear|inc|drop)_nlink
631 * inode_(inc|dec)_link_count
632 */
633 union {
634 const unsigned int i_nlink;
635 unsigned int __i_nlink;
636 };
Linus Torvalds3ddcd052011-08-06 22:45:50 -0700637 dev_t i_rdev;
Linus Torvalds2f9d3df2012-06-03 14:50:19 -0700638 loff_t i_size;
Linus Torvalds3ddcd052011-08-06 22:45:50 -0700639 struct timespec i_atime;
640 struct timespec i_mtime;
641 struct timespec i_ctime;
Theodore Ts'o6cdbb0e2011-10-29 08:24:18 -0400642 spinlock_t i_lock; /* i_blocks, i_bytes, maybe i_size */
643 unsigned short i_bytes;
Linus Torvalds2f9d3df2012-06-03 14:50:19 -0700644 unsigned int i_blkbits;
Linus Torvalds3ddcd052011-08-06 22:45:50 -0700645 blkcnt_t i_blocks;
646
647#ifdef __NEED_I_SIZE_ORDERED
648 seqcount_t i_size_seqcount;
649#endif
650
651 /* Misc */
652 unsigned long i_state;
Al Viro9902af72016-04-15 15:08:36 -0400653 struct rw_semaphore i_rwsem;
Linus Torvalds13e12d12011-06-08 15:18:19 -0700654
Nick Piggin44a7d7a2011-01-07 17:49:56 +1100655 unsigned long dirtied_when; /* jiffies of first dirtying */
Theodore Ts'oa2f48702015-03-17 12:23:19 -0400656 unsigned long dirtied_time_when;
Nick Piggin44a7d7a2011-01-07 17:49:56 +1100657
Linus Torvalds1da177e2005-04-16 15:20:36 -0700658 struct hlist_node i_hash;
Dave Chinnerc7f54082015-03-04 14:07:22 -0500659 struct list_head i_io_list; /* backing dev IO list */
Tejun Heo52ebea72015-05-22 17:13:37 -0400660#ifdef CONFIG_CGROUP_WRITEBACK
661 struct bdi_writeback *i_wb; /* the associated cgroup wb */
Tejun Heo2a814902015-05-28 14:50:51 -0400662
663 /* foreign inode detection, see wbc_detach_inode() */
664 int i_wb_frn_winner;
665 u16 i_wb_frn_avg_time;
666 u16 i_wb_frn_history;
Tejun Heo52ebea72015-05-22 17:13:37 -0400667#endif
Nick Piggin7ccf19a2010-10-21 11:49:30 +1100668 struct list_head i_lru; /* inode LRU list */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700669 struct list_head i_sb_list;
Dave Chinner6c60d2b2016-07-26 15:21:50 -0700670 struct list_head i_wb_list; /* backing dev writeback list */
Nick Pigginfa0d7e3d2011-01-07 17:49:49 +1100671 union {
Al Virob3d9b7a2012-06-09 13:51:19 -0400672 struct hlist_head i_dentry;
Nick Pigginfa0d7e3d2011-01-07 17:49:49 +1100673 struct rcu_head i_rcu;
674 };
Jean Noel Cordenner7a224222008-01-28 23:58:27 -0500675 u64 i_version;
Linus Torvalds2f9d3df2012-06-03 14:50:19 -0700676 atomic_t i_count;
Christoph Hellwigbd5fe6c2011-06-24 14:29:43 -0400677 atomic_t i_dio_count;
Theodore Ts'o6cdbb0e2011-10-29 08:24:18 -0400678 atomic_t i_writecount;
Mimi Zohard984ea62013-12-11 15:20:54 -0500679#ifdef CONFIG_IMA
680 atomic_t i_readcount; /* struct files open RO */
681#endif
Arjan van de Ven99ac48f2006-03-28 01:56:41 -0800682 const struct file_operations *i_fop; /* former ->i_op->default_file_ops */
Jeff Layton4a075e32015-01-16 15:05:54 -0500683 struct file_lock_context *i_flctx;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700684 struct address_space i_data;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700685 struct list_head i_devices;
Theodore Ts'o4c154162006-09-27 01:50:47 -0700686 union {
687 struct pipe_inode_info *i_pipe;
Theodore Ts'oeaf796e2006-09-27 01:50:48 -0700688 struct block_device *i_bdev;
Theodore Ts'o577c4eb2006-09-27 01:50:49 -0700689 struct cdev *i_cdev;
Al Viro61ba64f2015-05-02 09:54:06 -0400690 char *i_link;
Al Viro84e710d2016-04-15 00:58:55 -0400691 unsigned i_dir_seq;
Theodore Ts'o4c154162006-09-27 01:50:47 -0700692 };
Linus Torvalds1da177e2005-04-16 15:20:36 -0700693
694 __u32 i_generation;
695
Eric Paris3be25f42009-05-21 17:01:26 -0400696#ifdef CONFIG_FSNOTIFY
697 __u32 i_fsnotify_mask; /* all events this inode cares about */
Eric Parise61ce862009-12-17 21:24:24 -0500698 struct hlist_head i_fsnotify_marks;
Robert Love0eeca282005-07-12 17:06:03 -0400699#endif
700
Jaegeuk Kim0b81d072015-05-15 16:26:10 -0700701#if IS_ENABLED(CONFIG_FS_ENCRYPTION)
702 struct fscrypt_info *i_crypt_info;
703#endif
704
Theodore Ts'o8e18e292006-09-27 01:50:46 -0700705 void *i_private; /* fs or device private pointer */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700706};
707
Al Viro1d3382cb2010-10-23 15:19:20 -0400708static inline int inode_unhashed(struct inode *inode)
709{
710 return hlist_unhashed(&inode->i_hash);
711}
712
Linus Torvalds1da177e2005-04-16 15:20:36 -0700713/*
Ingo Molnarf2eace22006-07-03 00:25:05 -0700714 * inode->i_mutex nesting subclasses for the lock validator:
715 *
716 * 0: the object of the current VFS operation
717 * 1: parent
718 * 2: child/target
J. Bruce Fields40bd22c2012-04-18 15:21:34 -0400719 * 3: xattr
720 * 4: second non-directory
Miklos Szeredid1b72cc2014-10-27 15:42:01 +0100721 * 5: second parent (when locking independent directories in rename)
722 *
723 * I_MUTEX_NONDIR2 is for certain operations (such as rename) which lock two
J. Bruce Fields40bd22c2012-04-18 15:21:34 -0400724 * non-directories at once.
Ingo Molnarf2eace22006-07-03 00:25:05 -0700725 *
726 * The locking order between these classes is
Miklos Szeredid1b72cc2014-10-27 15:42:01 +0100727 * parent[2] -> child -> grandchild -> normal -> xattr -> second non-directory
Ingo Molnarf2eace22006-07-03 00:25:05 -0700728 */
729enum inode_i_mutex_lock_class
730{
731 I_MUTEX_NORMAL,
732 I_MUTEX_PARENT,
733 I_MUTEX_CHILD,
Ingo Molnar4df46242006-08-27 01:23:56 -0700734 I_MUTEX_XATTR,
Miklos Szeredid1b72cc2014-10-27 15:42:01 +0100735 I_MUTEX_NONDIR2,
736 I_MUTEX_PARENT2,
Ingo Molnarf2eace22006-07-03 00:25:05 -0700737};
738
Al Viro59551022016-01-22 15:40:57 -0500739static inline void inode_lock(struct inode *inode)
740{
Al Viro9902af72016-04-15 15:08:36 -0400741 down_write(&inode->i_rwsem);
Al Viro59551022016-01-22 15:40:57 -0500742}
743
744static inline void inode_unlock(struct inode *inode)
745{
Al Viro9902af72016-04-15 15:08:36 -0400746 up_write(&inode->i_rwsem);
747}
748
749static inline void inode_lock_shared(struct inode *inode)
750{
751 down_read(&inode->i_rwsem);
752}
753
754static inline void inode_unlock_shared(struct inode *inode)
755{
756 up_read(&inode->i_rwsem);
Al Viro59551022016-01-22 15:40:57 -0500757}
758
759static inline int inode_trylock(struct inode *inode)
760{
Al Viro9902af72016-04-15 15:08:36 -0400761 return down_write_trylock(&inode->i_rwsem);
762}
763
764static inline int inode_trylock_shared(struct inode *inode)
765{
766 return down_read_trylock(&inode->i_rwsem);
Al Viro59551022016-01-22 15:40:57 -0500767}
768
769static inline int inode_is_locked(struct inode *inode)
770{
Al Viro9902af72016-04-15 15:08:36 -0400771 return rwsem_is_locked(&inode->i_rwsem);
Al Viro59551022016-01-22 15:40:57 -0500772}
773
774static inline void inode_lock_nested(struct inode *inode, unsigned subclass)
775{
Al Viro9902af72016-04-15 15:08:36 -0400776 down_write_nested(&inode->i_rwsem, subclass);
Al Viro59551022016-01-22 15:40:57 -0500777}
778
J. Bruce Fields375e2892012-04-18 15:16:33 -0400779void lock_two_nondirectories(struct inode *, struct inode*);
780void unlock_two_nondirectories(struct inode *, struct inode*);
781
Ingo Molnarf2eace22006-07-03 00:25:05 -0700782/*
Linus Torvalds1da177e2005-04-16 15:20:36 -0700783 * NOTE: in a 32bit arch with a preemptable kernel and
784 * an UP compile the i_size_read/write must be atomic
785 * with respect to the local cpu (unlike with preempt disabled),
786 * but they don't need to be atomic with respect to other cpus like in
787 * true SMP (so they need either to either locally disable irq around
788 * the read or for example on x86 they can be still implemented as a
789 * cmpxchg8b without the need of the lock prefix). For SMP compiles
790 * and 64bit archs it makes no difference if preempt is enabled or not.
791 */
Jan Engelhardt48ed2142006-12-06 20:35:37 -0800792static inline loff_t i_size_read(const struct inode *inode)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700793{
794#if BITS_PER_LONG==32 && defined(CONFIG_SMP)
795 loff_t i_size;
796 unsigned int seq;
797
798 do {
799 seq = read_seqcount_begin(&inode->i_size_seqcount);
800 i_size = inode->i_size;
801 } while (read_seqcount_retry(&inode->i_size_seqcount, seq));
802 return i_size;
803#elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
804 loff_t i_size;
805
806 preempt_disable();
807 i_size = inode->i_size;
808 preempt_enable();
809 return i_size;
810#else
811 return inode->i_size;
812#endif
813}
814
Miklos Szeredi7762f5a2006-10-17 00:10:07 -0700815/*
816 * NOTE: unlike i_size_read(), i_size_write() does need locking around it
817 * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount
818 * can be lost, resulting in subsequent i_size_read() calls spinning forever.
819 */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700820static inline void i_size_write(struct inode *inode, loff_t i_size)
821{
822#if BITS_PER_LONG==32 && defined(CONFIG_SMP)
Fan Du74e3d1e2013-04-30 15:27:27 -0700823 preempt_disable();
Linus Torvalds1da177e2005-04-16 15:20:36 -0700824 write_seqcount_begin(&inode->i_size_seqcount);
825 inode->i_size = i_size;
826 write_seqcount_end(&inode->i_size_seqcount);
Fan Du74e3d1e2013-04-30 15:27:27 -0700827 preempt_enable();
Linus Torvalds1da177e2005-04-16 15:20:36 -0700828#elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
829 preempt_disable();
830 inode->i_size = i_size;
831 preempt_enable();
832#else
833 inode->i_size = i_size;
834#endif
835}
836
Jan Engelhardt48ed2142006-12-06 20:35:37 -0800837static inline unsigned iminor(const struct inode *inode)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700838{
839 return MINOR(inode->i_rdev);
840}
841
Jan Engelhardt48ed2142006-12-06 20:35:37 -0800842static inline unsigned imajor(const struct inode *inode)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700843{
844 return MAJOR(inode->i_rdev);
845}
846
847extern struct block_device *I_BDEV(struct inode *inode);
848
849struct fown_struct {
850 rwlock_t lock; /* protects pid, uid, euid fields */
Eric W. Biederman609d7fa2006-10-02 02:17:15 -0700851 struct pid *pid; /* pid or -pgrp where SIGIO should be sent */
852 enum pid_type pid_type; /* Kind of process group SIGIO should be sent to */
Eric W. Biederman92361632012-02-08 07:07:50 -0800853 kuid_t uid, euid; /* uid/euid of process setting the owner */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700854 int signum; /* posix.1b rt signal to be delivered on IO */
855};
856
857/*
858 * Track a single file's readahead state
859 */
860struct file_ra_state {
Fengguang Wu937085a2007-10-16 01:24:31 -0700861 pgoff_t start; /* where readahead started */
862 unsigned int size; /* # of readahead pages */
863 unsigned int async_size; /* do asynchronous readahead when
Fengguang Wuf9acc8c2007-07-19 01:48:08 -0700864 there are only # of pages ahead */
Fengguang Wu5ce11102007-07-19 01:47:59 -0700865
Fengguang Wu937085a2007-10-16 01:24:31 -0700866 unsigned int ra_pages; /* Maximum readahead window */
Wu Fengguang1ebf26a2009-06-16 15:31:19 -0700867 unsigned int mmap_miss; /* Cache miss stat for mmap accesses */
Fengguang Wuf4e6b492007-10-16 01:24:33 -0700868 loff_t prev_pos; /* Cache last read() position */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700869};
Linus Torvalds1da177e2005-04-16 15:20:36 -0700870
Fengguang Wu5ce11102007-07-19 01:47:59 -0700871/*
Fengguang Wu5ce11102007-07-19 01:47:59 -0700872 * Check if @index falls in the readahead windows.
873 */
874static inline int ra_has_index(struct file_ra_state *ra, pgoff_t index)
875{
Fengguang Wuf9acc8c2007-07-19 01:48:08 -0700876 return (index >= ra->start &&
877 index < ra->start + ra->size);
Fengguang Wu5ce11102007-07-19 01:47:59 -0700878}
879
Linus Torvalds1da177e2005-04-16 15:20:36 -0700880struct file {
Eric Dumazet2f512012005-10-30 15:02:16 -0800881 union {
Oleg Nesterov4f5e65a2013-07-08 14:24:16 -0700882 struct llist_node fu_llist;
Eric Dumazet2f512012005-10-30 15:02:16 -0800883 struct rcu_head fu_rcuhead;
884 } f_u;
Josef "Jeff" Sipek0f7fc9e2006-12-08 02:36:35 -0800885 struct path f_path;
Al Virodd379782013-03-01 19:48:30 -0500886 struct inode *f_inode; /* cached value */
Arjan van de Ven99ac48f2006-03-28 01:56:41 -0800887 const struct file_operations *f_op;
Andi Kleenef3d0fd2011-09-15 16:06:48 -0700888
889 /*
Linus Torvalds9c225f22014-03-03 09:36:58 -0800890 * Protects f_ep_links, f_flags.
Andi Kleenef3d0fd2011-09-15 16:06:48 -0700891 * Must not be taken from IRQ context.
892 */
893 spinlock_t f_lock;
Al Viro516e0cc2008-07-26 00:39:17 -0400894 atomic_long_t f_count;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700895 unsigned int f_flags;
Al Viroaeb5d722008-09-02 15:28:45 -0400896 fmode_t f_mode;
Linus Torvalds9c225f22014-03-03 09:36:58 -0800897 struct mutex f_pos_lock;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700898 loff_t f_pos;
899 struct fown_struct f_owner;
David Howellsd76b0d92008-11-14 10:39:25 +1100900 const struct cred *f_cred;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700901 struct file_ra_state f_ra;
902
Mathieu Desnoyers2b47c362007-10-16 23:27:21 -0700903 u64 f_version;
Alexey Dobriyan50462062006-09-29 02:00:01 -0700904#ifdef CONFIG_SECURITY
Linus Torvalds1da177e2005-04-16 15:20:36 -0700905 void *f_security;
Alexey Dobriyan50462062006-09-29 02:00:01 -0700906#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -0700907 /* needed for tty driver, and maybe others */
908 void *private_data;
909
910#ifdef CONFIG_EPOLL
911 /* Used by fs/eventpoll.c to link all the hooks to this file */
912 struct list_head f_ep_links;
Jason Baron28d82dc2012-01-12 17:17:43 -0800913 struct list_head f_tfile_llink;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700914#endif /* #ifdef CONFIG_EPOLL */
915 struct address_space *f_mapping;
Al Virobd2a31d2014-03-04 14:54:22 -0500916} __attribute__((aligned(4))); /* lest something weird decides that 2 is OK */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700917
Aneesh Kumar K.V990d6c22011-01-29 18:43:26 +0530918struct file_handle {
919 __u32 handle_bytes;
920 int handle_type;
921 /* file identifier */
922 unsigned char f_handle[0];
923};
924
Al Virocb0942b2012-08-27 14:48:26 -0400925static inline struct file *get_file(struct file *f)
926{
927 atomic_long_inc(&f->f_count);
928 return f;
929}
Konstantin Khlebnikov90f31d02015-04-16 12:47:56 -0700930#define get_file_rcu(x) atomic_long_inc_not_zero(&(x)->f_count)
Al Virod7065da2010-05-26 15:13:55 -0400931#define fput_atomic(x) atomic_long_add_unless(&(x)->f_count, -1, 1)
Al Viro516e0cc2008-07-26 00:39:17 -0400932#define file_count(x) atomic_long_read(&(x)->f_count)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700933
934#define MAX_NON_LFS ((1UL<<31) - 1)
935
936/* Page cache limit. The filesystems should put that into their s_maxbytes
937 limits, otherwise bad things can happen in VM. */
938#if BITS_PER_LONG==32
Kirill A. Shutemov09cbfea2016-04-01 15:29:47 +0300939#define MAX_LFS_FILESIZE (((loff_t)PAGE_SIZE << (BITS_PER_LONG-1))-1)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700940#elif BITS_PER_LONG==64
Al Viro614c3212012-10-10 01:06:20 -0400941#define MAX_LFS_FILESIZE ((loff_t)0x7fffffffffffffffLL)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700942#endif
943
944#define FL_POSIX 1
945#define FL_FLOCK 2
J. Bruce Fields617588d2011-07-01 15:18:34 -0400946#define FL_DELEG 4 /* NFSv4 delegation */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700947#define FL_ACCESS 8 /* not trying to lock, just looking */
Trond Myklebustf475ae92006-06-29 16:38:32 -0400948#define FL_EXISTS 16 /* when unlocking, test for existence */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700949#define FL_LEASE 32 /* lease held on this file */
Miklos Szeredi75e1fcc2006-06-23 02:05:12 -0700950#define FL_CLOSE 64 /* unlock on close */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700951#define FL_SLEEP 128 /* A blocking lock */
J. Bruce Fields778fc542011-07-26 18:25:49 -0400952#define FL_DOWNGRADE_PENDING 256 /* Lease is being downgraded */
953#define FL_UNLOCK_PENDING 512 /* Lease is being broken */
Jeff Laytoncff2fce2014-04-22 08:24:32 -0400954#define FL_OFDLCK 1024 /* lock is "owned" by struct file */
Christoph Hellwig11afe9f2015-01-21 19:17:03 +0100955#define FL_LAYOUT 2048 /* outstanding pNFS layout */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700956
957/*
Miklos Szeredibde74e42008-07-25 01:48:57 -0700958 * Special return value from posix_lock_file() and vfs_lock_file() for
959 * asynchronous locking.
960 */
961#define FILE_LOCK_DEFERRED 1
962
Jeff Layton7ca76312014-09-01 19:04:48 -0400963/* legacy typedef, should eventually be removed */
Christoph Hellwig17fa3882014-07-13 17:00:37 +0200964typedef void *fl_owner_t;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700965
Jeff Laytona7231a92015-01-16 15:05:56 -0500966struct file_lock;
967
Linus Torvalds1da177e2005-04-16 15:20:36 -0700968struct file_lock_operations {
Linus Torvalds1da177e2005-04-16 15:20:36 -0700969 void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
970 void (*fl_release_private)(struct file_lock *);
971};
972
973struct lock_manager_operations {
J. Bruce Fields8fb47a42011-07-20 20:21:59 -0400974 int (*lm_compare_owner)(struct file_lock *, struct file_lock *);
Jeff Layton3999e492013-06-21 08:58:19 -0400975 unsigned long (*lm_owner_key)(struct file_lock *);
Jeff Laytoncae80b32015-04-03 09:04:04 -0400976 fl_owner_t (*lm_get_owner)(fl_owner_t);
977 void (*lm_put_owner)(fl_owner_t);
J. Bruce Fields8fb47a42011-07-20 20:21:59 -0400978 void (*lm_notify)(struct file_lock *); /* unblock callback */
Joe Perchesd0449b92014-08-22 10:18:42 -0400979 int (*lm_grant)(struct file_lock *, int);
Jeff Layton4d01b7f2014-09-01 15:06:54 -0400980 bool (*lm_break)(struct file_lock *);
Jeff Layton7448cc32015-01-16 15:05:57 -0500981 int (*lm_change)(struct file_lock *, int, struct list_head *);
Jeff Layton1c7dd2f2014-08-22 10:55:47 -0400982 void (*lm_setup)(struct file_lock *, void **);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700983};
984
J. Bruce Fieldsaf558e32007-09-06 12:34:25 -0400985struct lock_manager {
986 struct list_head list;
J. Bruce Fieldsc87fb4a2015-08-06 12:47:02 -0400987 /*
988 * NFSv4 and up also want opens blocked during the grace period;
989 * NLM doesn't care:
990 */
991 bool block_opens;
J. Bruce Fieldsaf558e32007-09-06 12:34:25 -0400992};
993
Stanislav Kinsbursky5ccb0062012-07-25 16:57:22 +0400994struct net;
995void locks_start_grace(struct net *, struct lock_manager *);
J. Bruce Fieldsaf558e32007-09-06 12:34:25 -0400996void locks_end_grace(struct lock_manager *);
Stanislav Kinsbursky5ccb0062012-07-25 16:57:22 +0400997int locks_in_grace(struct net *);
J. Bruce Fieldsc87fb4a2015-08-06 12:47:02 -0400998int opens_in_grace(struct net *);
J. Bruce Fieldsaf558e32007-09-06 12:34:25 -0400999
Linus Torvalds1da177e2005-04-16 15:20:36 -07001000/* that will die - we need it for nfs_lock_info */
1001#include <linux/nfs_fs_i.h>
1002
Jeff Layton1cb36012013-06-21 08:58:12 -04001003/*
1004 * struct file_lock represents a generic "file lock". It's used to represent
1005 * POSIX byte range locks, BSD (flock) locks, and leases. It's important to
1006 * note that the same struct is used to represent both a request for a lock and
1007 * the lock itself, but the same object is never used for both.
1008 *
1009 * FIXME: should we create a separate "struct lock_request" to help distinguish
1010 * these two uses?
1011 *
Jeff Layton8116bf42015-01-21 20:44:01 -05001012 * The varous i_flctx lists are ordered by:
Jeff Layton1cb36012013-06-21 08:58:12 -04001013 *
Jeff Layton8116bf42015-01-21 20:44:01 -05001014 * 1) lock owner
1015 * 2) lock range start
1016 * 3) lock range end
Jeff Layton1cb36012013-06-21 08:58:12 -04001017 *
1018 * Obviously, the last two criteria only matter for POSIX locks.
1019 */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001020struct file_lock {
1021 struct file_lock *fl_next; /* singly linked list for this inode */
Jeff Layton6dee60f2015-01-16 15:05:54 -05001022 struct list_head fl_list; /* link into file_lock_context */
Jeff Layton139ca042013-06-21 08:58:17 -04001023 struct hlist_node fl_link; /* node in global lists */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001024 struct list_head fl_block; /* circular list of blocked processes */
1025 fl_owner_t fl_owner;
J. Bruce Fields710b7212011-07-26 16:28:29 -04001026 unsigned int fl_flags;
Richard Kennedyafc12462008-07-11 17:20:49 -07001027 unsigned char fl_type;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001028 unsigned int fl_pid;
Jeff Layton7012b022013-06-21 08:58:22 -04001029 int fl_link_cpu; /* what cpu's list is this on? */
Vitaliy Gusevab1f1612008-01-17 00:07:08 +00001030 struct pid *fl_nspid;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001031 wait_queue_head_t fl_wait;
1032 struct file *fl_file;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001033 loff_t fl_start;
1034 loff_t fl_end;
1035
1036 struct fasync_struct * fl_fasync; /* for lease break notifications */
J. Bruce Fields778fc542011-07-26 18:25:49 -04001037 /* for lease breaks: */
1038 unsigned long fl_break_time;
1039 unsigned long fl_downgrade_time;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001040
Alexey Dobriyan6aed6282009-09-21 17:01:11 -07001041 const struct file_lock_operations *fl_ops; /* Callbacks for filesystems */
Alexey Dobriyan7b021962009-09-21 17:01:12 -07001042 const struct lock_manager_operations *fl_lmops; /* Callbacks for lockmanagers */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001043 union {
1044 struct nfs_lock_info nfs_fl;
Trond Myklebust8d0a8a92005-06-22 17:16:32 +00001045 struct nfs4_lock_info nfs4_fl;
David Howellse8d6c552007-07-15 23:40:12 -07001046 struct {
1047 struct list_head link; /* link in AFS vnode's pending_locks list */
1048 int state; /* state of grant or error if -ve */
1049 } afs;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001050 } fl_u;
1051};
1052
Jeff Layton4a075e32015-01-16 15:05:54 -05001053struct file_lock_context {
Jeff Layton6109c852015-01-16 15:05:57 -05001054 spinlock_t flc_lock;
Jeff Layton4a075e32015-01-16 15:05:54 -05001055 struct list_head flc_flock;
Jeff Laytonbd61e0a2015-01-16 15:05:55 -05001056 struct list_head flc_posix;
Jeff Layton8634b512015-01-16 15:05:55 -05001057 struct list_head flc_lease;
Jeff Layton4a075e32015-01-16 15:05:54 -05001058};
1059
Linus Torvalds1da177e2005-04-16 15:20:36 -07001060/* The following constant reflects the upper bound of the file/locking space */
1061#ifndef OFFSET_MAX
1062#define INT_LIMIT(x) (~((x)1 << (sizeof(x)*8 - 1)))
1063#define OFFSET_MAX INT_LIMIT(loff_t)
1064#define OFFT_OFFSET_MAX INT_LIMIT(off_t)
1065#endif
1066
Linus Torvalds1da177e2005-04-16 15:20:36 -07001067#include <linux/fcntl.h>
1068
Thomas Petazzonibfcd17a2008-08-06 15:12:22 +02001069extern void send_sigio(struct fown_struct *fown, int fd, int band);
1070
Miklos Szeredic568d682016-09-16 12:44:20 +02001071/*
1072 * Return the inode to use for locking
1073 *
1074 * For overlayfs this should be the overlay inode, not the real inode returned
1075 * by file_inode(). For any other fs file_inode(filp) and locks_inode(filp) are
1076 * equal.
1077 */
1078static inline struct inode *locks_inode(const struct file *f)
1079{
1080 return f->f_path.dentry->d_inode;
1081}
1082
Thomas Petazzonibfcd17a2008-08-06 15:12:22 +02001083#ifdef CONFIG_FILE_LOCKING
Jeff Laytonc1e62b82014-02-03 12:13:09 -05001084extern int fcntl_getlk(struct file *, unsigned int, struct flock __user *);
Peter Staubachc2936212005-07-27 11:45:09 -07001085extern int fcntl_setlk(unsigned int, struct file *, unsigned int,
1086 struct flock __user *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001087
1088#if BITS_PER_LONG == 32
Jeff Laytonc1e62b82014-02-03 12:13:09 -05001089extern int fcntl_getlk64(struct file *, unsigned int, struct flock64 __user *);
Peter Staubachc2936212005-07-27 11:45:09 -07001090extern int fcntl_setlk64(unsigned int, struct file *, unsigned int,
1091 struct flock64 __user *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001092#endif
1093
Linus Torvalds1da177e2005-04-16 15:20:36 -07001094extern int fcntl_setlease(unsigned int fd, struct file *filp, long arg);
1095extern int fcntl_getlease(struct file *filp);
1096
1097/* fs/locks.c */
Jeff Laytonf27a0fe2016-01-07 15:08:51 -05001098void locks_free_lock_context(struct inode *inode);
J. Bruce Fields05fa3132010-10-30 17:31:15 -04001099void locks_free_lock(struct file_lock *fl);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001100extern void locks_init_lock(struct file_lock *);
Arnd Bergmannc5b1f0d2010-10-27 15:46:08 +02001101extern struct file_lock * locks_alloc_lock(void);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001102extern void locks_copy_lock(struct file_lock *, struct file_lock *);
Kinglong Mee3fe0fff12014-08-22 10:18:42 -04001103extern void locks_copy_conflock(struct file_lock *, struct file_lock *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001104extern void locks_remove_posix(struct file *, fl_owner_t);
Jeff Layton78ed8a12014-02-03 12:13:08 -05001105extern void locks_remove_file(struct file *);
Felix Blyakhera9e61e22009-03-31 15:12:56 -05001106extern void locks_release_private(struct file_lock *);
J. Bruce Fields6d34ac12007-05-11 16:09:32 -04001107extern void posix_test_lock(struct file *, struct file_lock *);
Marc Eshel150b3932007-01-18 16:15:35 -05001108extern int posix_lock_file(struct file *, struct file_lock *, struct file_lock *);
Jeff Laytonf891a292013-06-21 08:58:09 -04001109extern int posix_unblock_lock(struct file_lock *);
J. Bruce Fields3ee17ab2007-02-21 00:58:50 -05001110extern int vfs_test_lock(struct file *, struct file_lock *);
Marc Eshel150b3932007-01-18 16:15:35 -05001111extern int vfs_lock_file(struct file *, unsigned int, struct file_lock *, struct file_lock *);
Marc Eshel9b9d2ab2007-01-18 17:52:58 -05001112extern int vfs_cancel_lock(struct file *filp, struct file_lock *fl);
Benjamin Coddingtone55c34a2015-10-22 13:38:13 -04001113extern int locks_lock_inode_wait(struct inode *inode, struct file_lock *fl);
J. Bruce Fieldsdf4e8d22012-03-05 13:18:59 -05001114extern int __break_lease(struct inode *inode, unsigned int flags, unsigned int type);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001115extern void lease_get_mtime(struct inode *, struct timespec *time);
Jeff Laytone6f5c782014-08-22 10:40:25 -04001116extern int generic_setlease(struct file *, long, struct file_lock **, void **priv);
1117extern int vfs_setlease(struct file *, long, struct file_lock **, void **);
Jeff Layton7448cc32015-01-16 15:05:57 -05001118extern int lease_modify(struct file_lock *, int, struct list_head *);
Andrey Vagin6c8c9032015-04-16 12:49:38 -07001119struct files_struct;
1120extern void show_fd_locks(struct seq_file *f,
1121 struct file *filp, struct files_struct *files);
Thomas Petazzonibfcd17a2008-08-06 15:12:22 +02001122#else /* !CONFIG_FILE_LOCKING */
Jeff Laytonc1e62b82014-02-03 12:13:09 -05001123static inline int fcntl_getlk(struct file *file, unsigned int cmd,
1124 struct flock __user *user)
Steven Whitehousec2aca5e2009-01-20 10:29:45 +00001125{
1126 return -EINVAL;
1127}
1128
1129static inline int fcntl_setlk(unsigned int fd, struct file *file,
1130 unsigned int cmd, struct flock __user *user)
1131{
1132 return -EACCES;
1133}
1134
Thomas Petazzonibfcd17a2008-08-06 15:12:22 +02001135#if BITS_PER_LONG == 32
Jeff Laytonc1e62b82014-02-03 12:13:09 -05001136static inline int fcntl_getlk64(struct file *file, unsigned int cmd,
1137 struct flock64 __user *user)
Steven Whitehousec2aca5e2009-01-20 10:29:45 +00001138{
1139 return -EINVAL;
1140}
1141
1142static inline int fcntl_setlk64(unsigned int fd, struct file *file,
1143 unsigned int cmd, struct flock64 __user *user)
1144{
1145 return -EACCES;
1146}
Thomas Petazzonibfcd17a2008-08-06 15:12:22 +02001147#endif
Steven Whitehousec2aca5e2009-01-20 10:29:45 +00001148static inline int fcntl_setlease(unsigned int fd, struct file *filp, long arg)
1149{
Jeff Layton1b2b32d2014-09-24 08:38:44 -04001150 return -EINVAL;
Steven Whitehousec2aca5e2009-01-20 10:29:45 +00001151}
1152
1153static inline int fcntl_getlease(struct file *filp)
1154{
Jeff Layton1b2b32d2014-09-24 08:38:44 -04001155 return F_UNLCK;
Steven Whitehousec2aca5e2009-01-20 10:29:45 +00001156}
1157
Jeff Layton4a075e32015-01-16 15:05:54 -05001158static inline void
Jeff Laytonf27a0fe2016-01-07 15:08:51 -05001159locks_free_lock_context(struct inode *inode)
Jeff Layton4a075e32015-01-16 15:05:54 -05001160{
1161}
1162
Steven Whitehousec2aca5e2009-01-20 10:29:45 +00001163static inline void locks_init_lock(struct file_lock *fl)
1164{
1165 return;
1166}
1167
Kinglong Mee3fe0fff12014-08-22 10:18:42 -04001168static inline void locks_copy_conflock(struct file_lock *new, struct file_lock *fl)
Steven Whitehousec2aca5e2009-01-20 10:29:45 +00001169{
1170 return;
1171}
1172
1173static inline void locks_copy_lock(struct file_lock *new, struct file_lock *fl)
1174{
1175 return;
1176}
1177
1178static inline void locks_remove_posix(struct file *filp, fl_owner_t owner)
1179{
1180 return;
1181}
1182
Jeff Layton78ed8a12014-02-03 12:13:08 -05001183static inline void locks_remove_file(struct file *filp)
Steven Whitehousec2aca5e2009-01-20 10:29:45 +00001184{
1185 return;
1186}
1187
1188static inline void posix_test_lock(struct file *filp, struct file_lock *fl)
1189{
1190 return;
1191}
1192
1193static inline int posix_lock_file(struct file *filp, struct file_lock *fl,
1194 struct file_lock *conflock)
1195{
1196 return -ENOLCK;
1197}
1198
Jeff Laytonf891a292013-06-21 08:58:09 -04001199static inline int posix_unblock_lock(struct file_lock *waiter)
Steven Whitehousec2aca5e2009-01-20 10:29:45 +00001200{
1201 return -ENOENT;
1202}
1203
1204static inline int vfs_test_lock(struct file *filp, struct file_lock *fl)
1205{
1206 return 0;
1207}
1208
1209static inline int vfs_lock_file(struct file *filp, unsigned int cmd,
1210 struct file_lock *fl, struct file_lock *conf)
1211{
1212 return -ENOLCK;
1213}
1214
1215static inline int vfs_cancel_lock(struct file *filp, struct file_lock *fl)
1216{
1217 return 0;
1218}
1219
Benjamin Coddingtone55c34a2015-10-22 13:38:13 -04001220static inline int locks_lock_inode_wait(struct inode *inode, struct file_lock *fl)
1221{
1222 return -ENOLCK;
1223}
1224
J. Bruce Fieldsdf4e8d22012-03-05 13:18:59 -05001225static inline int __break_lease(struct inode *inode, unsigned int mode, unsigned int type)
Steven Whitehousec2aca5e2009-01-20 10:29:45 +00001226{
1227 return 0;
1228}
1229
1230static inline void lease_get_mtime(struct inode *inode, struct timespec *time)
1231{
1232 return;
1233}
1234
1235static inline int generic_setlease(struct file *filp, long arg,
Jeff Laytone6f5c782014-08-22 10:40:25 -04001236 struct file_lock **flp, void **priv)
Steven Whitehousec2aca5e2009-01-20 10:29:45 +00001237{
1238 return -EINVAL;
1239}
1240
1241static inline int vfs_setlease(struct file *filp, long arg,
Jeff Laytone6f5c782014-08-22 10:40:25 -04001242 struct file_lock **lease, void **priv)
Steven Whitehousec2aca5e2009-01-20 10:29:45 +00001243{
1244 return -EINVAL;
1245}
1246
Jeff Layton7448cc32015-01-16 15:05:57 -05001247static inline int lease_modify(struct file_lock *fl, int arg,
Jeff Laytonc45198e2014-09-01 07:12:07 -04001248 struct list_head *dispose)
Steven Whitehousec2aca5e2009-01-20 10:29:45 +00001249{
1250 return -EINVAL;
1251}
Andrey Vagin6c8c9032015-04-16 12:49:38 -07001252
1253struct files_struct;
1254static inline void show_fd_locks(struct seq_file *f,
1255 struct file *filp, struct files_struct *files) {}
Thomas Petazzonibfcd17a2008-08-06 15:12:22 +02001256#endif /* !CONFIG_FILE_LOCKING */
1257
Jeff Laytonee296d72015-07-11 06:43:03 -04001258static inline struct inode *file_inode(const struct file *f)
1259{
1260 return f->f_inode;
1261}
1262
Miklos Szeredid101a122016-03-26 16:14:37 -04001263static inline struct dentry *file_dentry(const struct file *file)
1264{
Miklos Szeredi2d902672016-06-30 08:53:27 +02001265 return d_real(file->f_path.dentry, file_inode(file), 0);
Miklos Szeredid101a122016-03-26 16:14:37 -04001266}
1267
Benjamin Coddingtone55c34a2015-10-22 13:38:13 -04001268static inline int locks_lock_file_wait(struct file *filp, struct file_lock *fl)
1269{
Miklos Szeredic568d682016-09-16 12:44:20 +02001270 return locks_lock_inode_wait(locks_inode(filp), fl);
Benjamin Coddingtone55c34a2015-10-22 13:38:13 -04001271}
1272
Linus Torvalds1da177e2005-04-16 15:20:36 -07001273struct fasync_struct {
Eric Dumazet989a2972010-04-14 09:55:35 +00001274 spinlock_t fa_lock;
1275 int magic;
1276 int fa_fd;
1277 struct fasync_struct *fa_next; /* singly linked list */
1278 struct file *fa_file;
1279 struct rcu_head fa_rcu;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001280};
1281
1282#define FASYNC_MAGIC 0x4601
1283
1284/* SMP safe fasync helpers: */
1285extern int fasync_helper(int, struct file *, int, struct fasync_struct **);
Linus Torvaldsf7347ce2010-10-27 12:38:12 -04001286extern struct fasync_struct *fasync_insert_entry(int, struct file *, struct fasync_struct **, struct fasync_struct *);
1287extern int fasync_remove_entry(struct file *, struct fasync_struct **);
1288extern struct fasync_struct *fasync_alloc(void);
1289extern void fasync_free(struct fasync_struct *);
1290
Linus Torvalds1da177e2005-04-16 15:20:36 -07001291/* can be called from interrupts */
1292extern void kill_fasync(struct fasync_struct **, int, int);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001293
Jeff Laytone0b93ed2014-08-22 11:27:32 -04001294extern void __f_setown(struct file *filp, struct pid *, enum pid_type, int force);
1295extern void f_setown(struct file *filp, unsigned long arg, int force);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001296extern void f_delown(struct file *filp);
Eric W. Biederman609d7fa2006-10-02 02:17:15 -07001297extern pid_t f_getown(struct file *filp);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001298extern int send_sigurg(struct fown_struct *fown);
1299
Jan Kara5accdf82012-06-12 16:20:34 +02001300struct mm_struct;
1301
Linus Torvalds1da177e2005-04-16 15:20:36 -07001302/*
1303 * Umount options
1304 */
1305
1306#define MNT_FORCE 0x00000001 /* Attempt to forcibily umount */
1307#define MNT_DETACH 0x00000002 /* Just detach from the tree */
1308#define MNT_EXPIRE 0x00000004 /* Mark for expiry */
Miklos Szeredidb1f05b2010-02-10 12:15:53 +01001309#define UMOUNT_NOFOLLOW 0x00000008 /* Don't follow symlink on umount */
1310#define UMOUNT_UNUSED 0x80000000 /* Flag guaranteed to be unused */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001311
Tejun Heo46b15ca2015-06-16 18:48:31 -04001312/* sb->s_iflags */
1313#define SB_I_CGROUPWB 0x00000001 /* cgroup-aware writeback enabled */
Eric W. Biederman90f85722015-06-29 14:42:03 -05001314#define SB_I_NOEXEC 0x00000002 /* Ignore executables on this fs */
Eric W. Biedermana2982cc2016-06-09 15:34:02 -05001315#define SB_I_NODEV 0x00000004 /* Ignore devices on this fs */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001316
Eric W. Biederman8654df42016-06-09 16:06:06 -05001317/* sb->s_iflags to limit user namespace mounts */
1318#define SB_I_USERNS_VISIBLE 0x00000010 /* fstype already mounted */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001319
Jan Kara5accdf82012-06-12 16:20:34 +02001320/* Possible states of 'frozen' field */
1321enum {
1322 SB_UNFROZEN = 0, /* FS is unfrozen */
1323 SB_FREEZE_WRITE = 1, /* Writes, dir ops, ioctls frozen */
Jan Kara5accdf82012-06-12 16:20:34 +02001324 SB_FREEZE_PAGEFAULT = 2, /* Page faults stopped as well */
1325 SB_FREEZE_FS = 3, /* For internal FS use (e.g. to stop
1326 * internal threads if needed) */
1327 SB_FREEZE_COMPLETE = 4, /* ->freeze_fs finished successfully */
1328};
1329
1330#define SB_FREEZE_LEVELS (SB_FREEZE_COMPLETE - 1)
1331
1332struct sb_writers {
Oleg Nesterov8129ed22015-08-11 17:05:04 +02001333 int frozen; /* Is sb frozen? */
1334 wait_queue_head_t wait_unfrozen; /* for get_super_thawed() */
1335 struct percpu_rw_semaphore rw_sem[SB_FREEZE_LEVELS];
Jan Kara5accdf82012-06-12 16:20:34 +02001336};
1337
Linus Torvalds1da177e2005-04-16 15:20:36 -07001338struct super_block {
1339 struct list_head s_list; /* Keep this first */
1340 dev_t s_dev; /* search index; _not_ kdev_t */
Richard Kennedy270ba5f2010-01-26 14:12:43 +00001341 unsigned char s_blocksize_bits;
1342 unsigned long s_blocksize;
Jeff Layton42cb56a2009-09-18 13:05:53 -07001343 loff_t s_maxbytes; /* Max file size */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001344 struct file_system_type *s_type;
Josef 'Jeff' Sipekee9b6d62007-02-12 00:55:41 -08001345 const struct super_operations *s_op;
Alexey Dobriyan61e225d2009-09-21 17:01:08 -07001346 const struct dquot_operations *dq_op;
Alexey Dobriyan0d54b212009-09-21 17:01:09 -07001347 const struct quotactl_ops *s_qcop;
Christoph Hellwig39655162007-10-21 16:42:17 -07001348 const struct export_operations *s_export_op;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001349 unsigned long s_flags;
Tejun Heo46b15ca2015-06-16 18:48:31 -04001350 unsigned long s_iflags; /* internal SB_I_* flags */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001351 unsigned long s_magic;
1352 struct dentry *s_root;
1353 struct rw_semaphore s_umount;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001354 int s_count;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001355 atomic_t s_active;
Alexey Dobriyan50462062006-09-29 02:00:01 -07001356#ifdef CONFIG_SECURITY
Linus Torvalds1da177e2005-04-16 15:20:36 -07001357 void *s_security;
Alexey Dobriyan50462062006-09-29 02:00:01 -07001358#endif
Stephen Hemmingerbb435452010-05-13 17:53:14 -07001359 const struct xattr_handler **s_xattr;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001360
Jaegeuk Kim0b81d072015-05-15 16:26:10 -07001361 const struct fscrypt_operations *s_cop;
1362
Nick Pigginceb5bdc2011-01-07 17:50:05 +11001363 struct hlist_bl_head s_anon; /* anonymous dentries for (nfs) exporting */
Miklos Szeredi39f7c4d2011-11-21 12:11:30 +01001364 struct list_head s_mounts; /* list of mounts; _not_ for fs use */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001365 struct block_device *s_bdev;
Jens Axboe32a88aa2009-09-16 15:02:33 +02001366 struct backing_dev_info *s_bdi;
David Howellsacaebfd2007-05-10 22:51:50 -07001367 struct mtd_info *s_mtd;
Al Viroa5166162011-12-12 22:53:00 -05001368 struct hlist_node s_instances;
Jan Kara2c5f6482014-09-30 10:43:09 +02001369 unsigned int s_quota_types; /* Bitmask of supported quota types */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001370 struct quota_info s_dquot; /* Diskquota specific options */
1371
Jan Kara5accdf82012-06-12 16:20:34 +02001372 struct sb_writers s_writers;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001373
1374 char s_id[32]; /* Informational name */
Linus Torvalds0bba0162011-04-12 15:21:04 -07001375 u8 s_uuid[16]; /* UUID */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001376
1377 void *s_fs_info; /* Filesystem private info */
Al Viro8de52772012-02-06 12:45:27 -05001378 unsigned int s_max_links;
Al Viro30c40d22008-02-22 19:50:45 -05001379 fmode_t s_mode;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001380
Richard Kennedy270ba5f2010-01-26 14:12:43 +00001381 /* Granularity of c/m/atime in ns.
1382 Cannot be worse than a second */
1383 u32 s_time_gran;
1384
Linus Torvalds1da177e2005-04-16 15:20:36 -07001385 /*
1386 * The next field is for VFS *only*. No filesystems have any business
1387 * even looking at it. You had been warned.
1388 */
Arjan van de Vena11f3a02006-03-23 03:00:33 -08001389 struct mutex s_vfs_rename_mutex; /* Kludge */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001390
Miklos Szeredi79c0b2d2007-05-08 00:25:43 -07001391 /*
1392 * Filesystem subtype. If non-empty the filesystem type field
1393 * in /proc/mounts will be "type.subtype"
1394 */
1395 char *s_subtype;
Miklos Szeredib3b304a2008-02-08 04:21:35 -08001396
1397 /*
1398 * Saved mount options for lazy filesystems using
1399 * generic_show_options()
1400 */
Arnd Bergmann4d2deb42010-02-24 20:01:56 +01001401 char __rcu *s_options;
Al Viroc8aebb02010-12-18 10:22:30 -05001402 const struct dentry_operations *s_d_op; /* default d_op for dentries */
Dan Magenheimer9fdfdcf2011-05-26 10:01:19 -06001403
1404 /*
1405 * Saved pool identifier for cleancache (-1 means none)
1406 */
1407 int cleancache_poolid;
Dave Chinnerb0d40c92011-07-08 14:14:42 +10001408
1409 struct shrinker s_shrink; /* per-sb shrinker handle */
Miklos Szeredi4ed5e82f2011-11-21 12:11:31 +01001410
Miklos Szeredi7ada4db2011-11-21 12:11:32 +01001411 /* Number of inodes with nlink == 0 but still referenced */
1412 atomic_long_t s_remove_count;
1413
Miklos Szeredi4ed5e82f2011-11-21 12:11:31 +01001414 /* Being remounted read-only */
1415 int s_readonly_remount;
Christoph Hellwig7b7a8662013-09-04 15:04:39 +02001416
1417 /* AIO completions deferred from interrupt context */
1418 struct workqueue_struct *s_dio_done_wq;
Al Viro215752f2014-08-07 06:23:41 -04001419 struct hlist_head s_pins;
Dave Chinnerf6041562013-08-28 10:18:00 +10001420
1421 /*
Eric W. Biederman6e4eab52016-05-24 09:29:01 -05001422 * Owning user namespace and default context in which to
1423 * interpret filesystem uids, gids, quotas, device nodes,
1424 * xattrs and security labels.
1425 */
1426 struct user_namespace *s_user_ns;
1427
1428 /*
Dave Chinnerf6041562013-08-28 10:18:00 +10001429 * Keep the lru lists last in the structure so they always sit on their
1430 * own individual cachelines.
1431 */
1432 struct list_lru s_dentry_lru ____cacheline_aligned_in_smp;
1433 struct list_lru s_inode_lru ____cacheline_aligned_in_smp;
Al Viroe2fec7c2013-10-04 17:06:56 -04001434 struct rcu_head rcu;
Oleg Nesterov853b39a2015-07-22 20:21:13 +02001435 struct work_struct destroy_work;
Miklos Szeredi69c433e2014-10-24 00:14:39 +02001436
Dave Chinnere97fedb2015-03-04 13:40:00 -05001437 struct mutex s_sync_lock; /* sync serialisation lock */
Miklos Szeredi69c433e2014-10-24 00:14:39 +02001438
1439 /*
1440 * Indicates how deep in a filesystem stack this SB is
1441 */
1442 int s_stack_depth;
Dave Chinner74278da2015-03-04 12:37:22 -05001443
1444 /* s_inode_list_lock protects s_inodes */
1445 spinlock_t s_inode_list_lock ____cacheline_aligned_in_smp;
1446 struct list_head s_inodes; /* all inodes */
Dave Chinner6c60d2b2016-07-26 15:21:50 -07001447
1448 spinlock_t s_inode_wblist_lock;
1449 struct list_head s_inodes_wb; /* writeback inodes */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001450};
1451
Seth Forshee81754352014-12-11 10:15:45 -06001452/* Helper functions so that in most cases filesystems will
1453 * not need to deal directly with kuid_t and kgid_t and can
1454 * instead deal with the raw numeric values that are stored
1455 * in the filesystem.
1456 */
1457static inline uid_t i_uid_read(const struct inode *inode)
1458{
1459 return from_kuid(inode->i_sb->s_user_ns, inode->i_uid);
1460}
1461
1462static inline gid_t i_gid_read(const struct inode *inode)
1463{
1464 return from_kgid(inode->i_sb->s_user_ns, inode->i_gid);
1465}
1466
1467static inline void i_uid_write(struct inode *inode, uid_t uid)
1468{
1469 inode->i_uid = make_kuid(inode->i_sb->s_user_ns, uid);
1470}
1471
1472static inline void i_gid_write(struct inode *inode, gid_t gid)
1473{
1474 inode->i_gid = make_kgid(inode->i_sb->s_user_ns, gid);
1475}
1476
Linus Torvalds1da177e2005-04-16 15:20:36 -07001477extern struct timespec current_fs_time(struct super_block *sb);
Deepa Dinamani3cd88662016-09-14 07:48:02 -07001478extern struct timespec current_time(struct inode *inode);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001479
1480/*
1481 * Snapshotting support.
1482 */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001483
Jan Kara5accdf82012-06-12 16:20:34 +02001484void __sb_end_write(struct super_block *sb, int level);
1485int __sb_start_write(struct super_block *sb, int level, bool wait);
1486
Oleg Nesterovbee91822015-07-19 23:48:20 +02001487#define __sb_writers_acquired(sb, lev) \
Oleg Nesterov8129ed22015-08-11 17:05:04 +02001488 percpu_rwsem_acquire(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
Oleg Nesterovbee91822015-07-19 23:48:20 +02001489#define __sb_writers_release(sb, lev) \
Oleg Nesterov8129ed22015-08-11 17:05:04 +02001490 percpu_rwsem_release(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
Oleg Nesterovbee91822015-07-19 23:48:20 +02001491
Jan Kara5accdf82012-06-12 16:20:34 +02001492/**
1493 * sb_end_write - drop write access to a superblock
1494 * @sb: the super we wrote to
1495 *
1496 * Decrement number of writers to the filesystem. Wake up possible waiters
1497 * wanting to freeze the filesystem.
1498 */
1499static inline void sb_end_write(struct super_block *sb)
1500{
1501 __sb_end_write(sb, SB_FREEZE_WRITE);
1502}
1503
1504/**
1505 * sb_end_pagefault - drop write access to a superblock from a page fault
1506 * @sb: the super we wrote to
1507 *
1508 * Decrement number of processes handling write page fault to the filesystem.
1509 * Wake up possible waiters wanting to freeze the filesystem.
1510 */
1511static inline void sb_end_pagefault(struct super_block *sb)
1512{
1513 __sb_end_write(sb, SB_FREEZE_PAGEFAULT);
1514}
1515
1516/**
1517 * sb_end_intwrite - drop write access to a superblock for internal fs purposes
1518 * @sb: the super we wrote to
1519 *
1520 * Decrement fs-internal number of writers to the filesystem. Wake up possible
1521 * waiters wanting to freeze the filesystem.
1522 */
1523static inline void sb_end_intwrite(struct super_block *sb)
1524{
1525 __sb_end_write(sb, SB_FREEZE_FS);
1526}
1527
1528/**
1529 * sb_start_write - get write access to a superblock
1530 * @sb: the super we write to
1531 *
1532 * When a process wants to write data or metadata to a file system (i.e. dirty
1533 * a page or an inode), it should embed the operation in a sb_start_write() -
1534 * sb_end_write() pair to get exclusion against file system freezing. This
1535 * function increments number of writers preventing freezing. If the file
1536 * system is already frozen, the function waits until the file system is
1537 * thawed.
1538 *
1539 * Since freeze protection behaves as a lock, users have to preserve
1540 * ordering of freeze protection and other filesystem locks. Generally,
1541 * freeze protection should be the outermost lock. In particular, we have:
1542 *
1543 * sb_start_write
1544 * -> i_mutex (write path, truncate, directory ops, ...)
1545 * -> s_umount (freeze_super, thaw_super)
1546 */
1547static inline void sb_start_write(struct super_block *sb)
1548{
1549 __sb_start_write(sb, SB_FREEZE_WRITE, true);
1550}
1551
1552static inline int sb_start_write_trylock(struct super_block *sb)
1553{
1554 return __sb_start_write(sb, SB_FREEZE_WRITE, false);
1555}
1556
1557/**
1558 * sb_start_pagefault - get write access to a superblock from a page fault
1559 * @sb: the super we write to
1560 *
1561 * When a process starts handling write page fault, it should embed the
1562 * operation into sb_start_pagefault() - sb_end_pagefault() pair to get
1563 * exclusion against file system freezing. This is needed since the page fault
1564 * is going to dirty a page. This function increments number of running page
1565 * faults preventing freezing. If the file system is already frozen, the
1566 * function waits until the file system is thawed.
1567 *
1568 * Since page fault freeze protection behaves as a lock, users have to preserve
1569 * ordering of freeze protection and other filesystem locks. It is advised to
1570 * put sb_start_pagefault() close to mmap_sem in lock ordering. Page fault
1571 * handling code implies lock dependency:
1572 *
1573 * mmap_sem
1574 * -> sb_start_pagefault
1575 */
1576static inline void sb_start_pagefault(struct super_block *sb)
1577{
1578 __sb_start_write(sb, SB_FREEZE_PAGEFAULT, true);
1579}
1580
1581/*
1582 * sb_start_intwrite - get write access to a superblock for internal fs purposes
1583 * @sb: the super we write to
1584 *
1585 * This is the third level of protection against filesystem freezing. It is
1586 * free for use by a filesystem. The only requirement is that it must rank
1587 * below sb_start_pagefault.
1588 *
1589 * For example filesystem can call sb_start_intwrite() when starting a
1590 * transaction which somewhat eases handling of freezing for internal sources
1591 * of filesystem changes (internal fs threads, discarding preallocation on file
1592 * close, etc.).
1593 */
1594static inline void sb_start_intwrite(struct super_block *sb)
1595{
1596 __sb_start_write(sb, SB_FREEZE_FS, true);
1597}
1598
Linus Torvalds1da177e2005-04-16 15:20:36 -07001599
Serge E. Hallyn2e149672011-03-23 16:43:26 -07001600extern bool inode_owner_or_capable(const struct inode *inode);
Satyam Sharma3bd858a2007-07-17 15:00:08 +05301601
Linus Torvalds1da177e2005-04-16 15:20:36 -07001602/*
1603 * VFS helper functions..
1604 */
Al Viro312b63f2012-06-10 18:09:36 -04001605extern int vfs_create(struct inode *, struct dentry *, umode_t, bool);
Al Viro18bb1db2011-07-26 01:41:39 -04001606extern int vfs_mkdir(struct inode *, struct dentry *, umode_t);
Al Viro1a67aaf2011-07-26 01:52:52 -04001607extern int vfs_mknod(struct inode *, struct dentry *, umode_t, dev_t);
Miklos Szeredidb2e7472008-06-24 16:50:16 +02001608extern int vfs_symlink(struct inode *, struct dentry *, const char *);
J. Bruce Fields146a8592011-09-20 17:14:31 -04001609extern int vfs_link(struct dentry *, struct inode *, struct dentry *, struct inode **);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001610extern int vfs_rmdir(struct inode *, struct dentry *);
J. Bruce Fieldsb21996e2011-09-20 09:14:34 -04001611extern int vfs_unlink(struct inode *, struct dentry *, struct inode **);
Miklos Szeredi520c8b12014-04-01 17:08:42 +02001612extern int vfs_rename(struct inode *, struct dentry *, struct inode *, struct dentry *, struct inode **, unsigned int);
Miklos Szeredi787fb6b2014-10-24 00:14:36 +02001613extern int vfs_whiteout(struct inode *, struct dentry *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001614
1615/*
Christoph Hellwig8c744fb2005-11-08 21:35:04 -08001616 * VFS file helper functions.
1617 */
Dmitry Monakhova1bd1202010-03-04 17:29:14 +03001618extern void inode_init_owner(struct inode *inode, const struct inode *dir,
Al Viro62bb1092011-07-24 23:20:18 -04001619 umode_t mode);
Eric W. Biedermana2982cc2016-06-09 15:34:02 -05001620extern bool may_open_dev(const struct path *path);
Christoph Hellwig8c744fb2005-11-08 21:35:04 -08001621/*
Mark Fashehc4b929b2008-10-08 19:44:18 -04001622 * VFS FS_IOC_FIEMAP helper definitions.
1623 */
1624struct fiemap_extent_info {
1625 unsigned int fi_flags; /* Flags as passed from user */
1626 unsigned int fi_extents_mapped; /* Number of mapped extents */
1627 unsigned int fi_extents_max; /* Size of fiemap_extent array */
Namhyung Kimecf56322011-01-16 23:28:17 +09001628 struct fiemap_extent __user *fi_extents_start; /* Start of
1629 fiemap_extent array */
Mark Fashehc4b929b2008-10-08 19:44:18 -04001630};
1631int fiemap_fill_next_extent(struct fiemap_extent_info *info, u64 logical,
1632 u64 phys, u64 len, u32 flags);
1633int fiemap_check_flags(struct fiemap_extent_info *fieinfo, u32 fs_flags);
1634
1635/*
Linus Torvalds1da177e2005-04-16 15:20:36 -07001636 * File types
1637 *
1638 * NOTE! These match bits 12..15 of stat.st_mode
1639 * (ie "(i_mode >> 12) & 15").
1640 */
1641#define DT_UNKNOWN 0
1642#define DT_FIFO 1
1643#define DT_CHR 2
1644#define DT_DIR 4
1645#define DT_BLK 6
1646#define DT_REG 8
1647#define DT_LNK 10
1648#define DT_SOCK 12
1649#define DT_WHT 14
1650
Linus Torvalds1da177e2005-04-16 15:20:36 -07001651/*
1652 * This is the "filldir" function type, used by readdir() to let
1653 * the kernel specify what kind of dirent layout it wants to have.
1654 * This allows the kernel to read directories into kernel space or
1655 * to have different dirent layouts depending on the binary type.
1656 */
Miklos Szerediac7576f2014-10-30 17:37:34 +01001657struct dir_context;
1658typedef int (*filldir_t)(struct dir_context *, const char *, int, loff_t, u64,
1659 unsigned);
1660
Al Viro5c0ba4e2013-05-15 13:52:59 -04001661struct dir_context {
Al Viroac6614b2013-05-22 22:22:04 -04001662 const filldir_t actor;
Al Virobb6f6192013-05-15 18:49:12 -04001663 loff_t pos;
Al Viro5c0ba4e2013-05-15 13:52:59 -04001664};
Al Virobb6f6192013-05-15 18:49:12 -04001665
Al Viro08f85852007-10-08 13:26:20 -04001666struct block_device_operations;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001667
Linus Torvalds1da177e2005-04-16 15:20:36 -07001668/* These macros are for out of kernel modules to test that
1669 * the kernel supports the unlocked_ioctl and compat_ioctl
1670 * fields in struct file_operations. */
1671#define HAVE_COMPAT_IOCTL 1
1672#define HAVE_UNLOCKED_IOCTL 1
1673
Christoph Hellwigb4caecd2015-01-14 10:42:32 +01001674/*
1675 * These flags let !MMU mmap() govern direct device mapping vs immediate
1676 * copying more easily for MAP_PRIVATE, especially for ROM filesystems.
1677 *
1678 * NOMMU_MAP_COPY: Copy can be mapped (MAP_PRIVATE)
1679 * NOMMU_MAP_DIRECT: Can be mapped directly (MAP_SHARED)
1680 * NOMMU_MAP_READ: Can be mapped for reading
1681 * NOMMU_MAP_WRITE: Can be mapped for writing
1682 * NOMMU_MAP_EXEC: Can be mapped for execution
1683 */
1684#define NOMMU_MAP_COPY 0x00000001
1685#define NOMMU_MAP_DIRECT 0x00000008
1686#define NOMMU_MAP_READ VM_MAYREAD
1687#define NOMMU_MAP_WRITE VM_MAYWRITE
1688#define NOMMU_MAP_EXEC VM_MAYEXEC
1689
1690#define NOMMU_VMFLAGS \
1691 (NOMMU_MAP_READ | NOMMU_MAP_WRITE | NOMMU_MAP_EXEC)
1692
1693
Al Viro293bc982014-02-11 18:37:41 -05001694struct iov_iter;
1695
Linus Torvalds1da177e2005-04-16 15:20:36 -07001696struct file_operations {
1697 struct module *owner;
1698 loff_t (*llseek) (struct file *, loff_t, int);
1699 ssize_t (*read) (struct file *, char __user *, size_t, loff_t *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001700 ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *);
Al Viro293bc982014-02-11 18:37:41 -05001701 ssize_t (*read_iter) (struct kiocb *, struct iov_iter *);
1702 ssize_t (*write_iter) (struct kiocb *, struct iov_iter *);
Al Virobb6f6192013-05-15 18:49:12 -04001703 int (*iterate) (struct file *, struct dir_context *);
Al Viro61922692016-04-20 23:08:32 -04001704 int (*iterate_shared) (struct file *, struct dir_context *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001705 unsigned int (*poll) (struct file *, struct poll_table_struct *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001706 long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long);
1707 long (*compat_ioctl) (struct file *, unsigned int, unsigned long);
1708 int (*mmap) (struct file *, struct vm_area_struct *);
1709 int (*open) (struct inode *, struct file *);
Miklos Szeredi75e1fcc2006-06-23 02:05:12 -07001710 int (*flush) (struct file *, fl_owner_t id);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001711 int (*release) (struct inode *, struct file *);
Josef Bacik02c24a82011-07-16 20:44:56 -04001712 int (*fsync) (struct file *, loff_t, loff_t, int datasync);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001713 int (*fasync) (int, struct file *, int);
1714 int (*lock) (struct file *, int, struct file_lock *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001715 ssize_t (*sendpage) (struct file *, struct page *, int, size_t, loff_t *, int);
1716 unsigned long (*get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
1717 int (*check_flags)(int);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001718 int (*flock) (struct file *, int, struct file_lock *);
Jens Axboecbb7e572006-04-11 14:57:50 +02001719 ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, loff_t *, size_t, unsigned int);
1720 ssize_t (*splice_read)(struct file *, loff_t *, struct pipe_inode_info *, size_t, unsigned int);
Jeff Laytone6f5c782014-08-22 10:40:25 -04001721 int (*setlease)(struct file *, long, struct file_lock **, void **);
Christoph Hellwig2fe17c12011-01-14 13:07:43 +01001722 long (*fallocate)(struct file *file, int mode, loff_t offset,
1723 loff_t len);
Joe Perchesa3816ab2014-09-29 16:08:25 -07001724 void (*show_fdinfo)(struct seq_file *m, struct file *f);
Christoph Hellwigb4caecd2015-01-14 10:42:32 +01001725#ifndef CONFIG_MMU
1726 unsigned (*mmap_capabilities)(struct file *);
1727#endif
Christoph Hellwig04b38d62015-12-03 12:59:50 +01001728 ssize_t (*copy_file_range)(struct file *, loff_t, struct file *,
1729 loff_t, size_t, unsigned int);
1730 int (*clone_file_range)(struct file *, loff_t, struct file *, loff_t,
1731 u64);
Darrick J. Wong54dbc152015-12-19 00:55:59 -08001732 ssize_t (*dedupe_file_range)(struct file *, u64, u64, struct file *,
1733 u64);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001734};
1735
1736struct inode_operations {
Al Viro00cd8dd2012-06-10 17:13:09 -04001737 struct dentry * (*lookup) (struct inode *,struct dentry *, unsigned int);
Al Virofceef392015-12-29 15:58:39 -05001738 const char * (*get_link) (struct dentry *, struct inode *, struct delayed_call *);
Al Viro10556cb22011-06-20 19:28:19 -04001739 int (*permission) (struct inode *, int);
Christoph Hellwig4e34e712011-07-23 17:37:31 +02001740 struct posix_acl * (*get_acl)(struct inode *, int);
Nick Piggin44a7d7a2011-01-07 17:49:56 +11001741
1742 int (*readlink) (struct dentry *, char __user *,int);
Nick Piggin44a7d7a2011-01-07 17:49:56 +11001743
Al Viroebfc3b42012-06-10 18:05:36 -04001744 int (*create) (struct inode *,struct dentry *, umode_t, bool);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001745 int (*link) (struct dentry *,struct inode *,struct dentry *);
1746 int (*unlink) (struct inode *,struct dentry *);
1747 int (*symlink) (struct inode *,struct dentry *,const char *);
Al Viro18bb1db2011-07-26 01:41:39 -04001748 int (*mkdir) (struct inode *,struct dentry *,umode_t);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001749 int (*rmdir) (struct inode *,struct dentry *);
Al Viro1a67aaf2011-07-26 01:52:52 -04001750 int (*mknod) (struct inode *,struct dentry *,umode_t,dev_t);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001751 int (*rename) (struct inode *, struct dentry *,
Miklos Szeredi520c8b12014-04-01 17:08:42 +02001752 struct inode *, struct dentry *, unsigned int);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001753 int (*setattr) (struct dentry *, struct iattr *);
1754 int (*getattr) (struct vfsmount *mnt, struct dentry *, struct kstat *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001755 ssize_t (*listxattr) (struct dentry *, char *, size_t);
Mark Fashehc4b929b2008-10-08 19:44:18 -04001756 int (*fiemap)(struct inode *, struct fiemap_extent_info *, u64 start,
1757 u64 len);
Josef Bacikc3b2da32012-03-26 09:59:21 -04001758 int (*update_time)(struct inode *, struct timespec *, int);
Al Virod9585272012-06-22 12:39:14 +04001759 int (*atomic_open)(struct inode *, struct dentry *,
Al Viro30d90492012-06-22 12:40:19 +04001760 struct file *, unsigned open_flag,
Al Virod9585272012-06-22 12:39:14 +04001761 umode_t create_mode, int *opened);
Al Viro60545d02013-06-07 01:20:27 -04001762 int (*tmpfile) (struct inode *, struct dentry *, umode_t);
Christoph Hellwig893d46e2013-12-20 05:16:39 -08001763 int (*set_acl)(struct inode *, struct posix_acl *, int);
Nick Piggin44a7d7a2011-01-07 17:49:56 +11001764} ____cacheline_aligned;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001765
Badari Pulavartyeed4e512006-09-30 23:28:49 -07001766ssize_t rw_copy_check_uvector(int type, const struct iovec __user * uvector,
Christopher Yeohfcf63402011-10-31 17:06:39 -07001767 unsigned long nr_segs, unsigned long fast_segs,
1768 struct iovec *fast_pointer,
Christopher Yeohac34ebb2012-05-31 16:26:42 -07001769 struct iovec **ret_pointer);
Badari Pulavartyeed4e512006-09-30 23:28:49 -07001770
Dmitry Kasatkin6fb50322014-11-05 17:01:17 +02001771extern ssize_t __vfs_read(struct file *, char __user *, size_t, loff_t *);
Al Viro493c84c2015-04-03 15:06:43 -04001772extern ssize_t __vfs_write(struct file *, const char __user *, size_t, loff_t *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001773extern ssize_t vfs_read(struct file *, char __user *, size_t, loff_t *);
1774extern ssize_t vfs_write(struct file *, const char __user *, size_t, loff_t *);
1775extern ssize_t vfs_readv(struct file *, const struct iovec __user *,
Christoph Hellwig793b80e2016-03-03 16:03:58 +01001776 unsigned long, loff_t *, int);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001777extern ssize_t vfs_writev(struct file *, const struct iovec __user *,
Christoph Hellwig793b80e2016-03-03 16:03:58 +01001778 unsigned long, loff_t *, int);
Zach Brown29732932015-11-10 16:53:30 -05001779extern ssize_t vfs_copy_file_range(struct file *, loff_t , struct file *,
1780 loff_t, size_t, unsigned int);
Christoph Hellwig04b38d62015-12-03 12:59:50 +01001781extern int vfs_clone_file_range(struct file *file_in, loff_t pos_in,
1782 struct file *file_out, loff_t pos_out, u64 len);
Darrick J. Wong54dbc152015-12-19 00:55:59 -08001783extern int vfs_dedupe_file_range(struct file *file,
1784 struct file_dedupe_range *same);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001785
Linus Torvalds1da177e2005-04-16 15:20:36 -07001786struct super_operations {
1787 struct inode *(*alloc_inode)(struct super_block *sb);
1788 void (*destroy_inode)(struct inode *);
1789
Christoph Hellwigaa385722011-05-27 06:53:02 -04001790 void (*dirty_inode) (struct inode *, int flags);
Christoph Hellwiga9185b42010-03-05 09:21:37 +01001791 int (*write_inode) (struct inode *, struct writeback_control *wbc);
Al Viro45321ac2010-06-07 13:43:19 -04001792 int (*drop_inode) (struct inode *);
Al Virobe7ce412010-06-04 19:40:39 -04001793 void (*evict_inode) (struct inode *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001794 void (*put_super) (struct super_block *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001795 int (*sync_fs)(struct super_block *sb, int wait);
Benjamin Marzinski48b6bca2014-11-13 20:42:03 -06001796 int (*freeze_super) (struct super_block *);
Takashi Satoc4be0c12009-01-09 16:40:58 -08001797 int (*freeze_fs) (struct super_block *);
Benjamin Marzinski48b6bca2014-11-13 20:42:03 -06001798 int (*thaw_super) (struct super_block *);
Takashi Satoc4be0c12009-01-09 16:40:58 -08001799 int (*unfreeze_fs) (struct super_block *);
David Howells726c3342006-06-23 02:02:58 -07001800 int (*statfs) (struct dentry *, struct kstatfs *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001801 int (*remount_fs) (struct super_block *, int *, char *);
Al Viro42faad92008-04-24 07:21:56 -04001802 void (*umount_begin) (struct super_block *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001803
Al Viro34c80b12011-12-08 21:32:45 -05001804 int (*show_options)(struct seq_file *, struct dentry *);
Al Virod861c632011-12-08 21:32:45 -05001805 int (*show_devname)(struct seq_file *, struct dentry *);
Al Viroa6322de2011-12-08 21:37:57 -05001806 int (*show_path)(struct seq_file *, struct dentry *);
Al Viro64132372011-12-08 20:51:13 -05001807 int (*show_stats)(struct seq_file *, struct dentry *);
Alexey Dobriyan0e51a722006-09-29 01:59:56 -07001808#ifdef CONFIG_QUOTA
Linus Torvalds1da177e2005-04-16 15:20:36 -07001809 ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t);
1810 ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t);
Jan Kara2d0fa462014-09-25 16:36:14 +02001811 struct dquot **(*get_dquots)(struct inode *);
Alexey Dobriyan0e51a722006-09-29 01:59:56 -07001812#endif
Theodore Ts'o87d8fe12009-01-03 09:47:09 -05001813 int (*bdev_try_to_free_page)(struct super_block*, struct page*, gfp_t);
Vladimir Davydov4101b622015-02-12 14:58:51 -08001814 long (*nr_cached_objects)(struct super_block *,
1815 struct shrink_control *);
1816 long (*free_cached_objects)(struct super_block *,
1817 struct shrink_control *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001818};
1819
Joern Engel1c0eeaf2007-10-16 23:30:44 -07001820/*
David Howellsbbc10962012-10-15 16:40:35 +01001821 * Inode flags - they have no relation to superblock flags now
1822 */
1823#define S_SYNC 1 /* Writes are synced at once */
1824#define S_NOATIME 2 /* Do not update access times */
1825#define S_APPEND 4 /* Append-only file */
1826#define S_IMMUTABLE 8 /* Immutable file */
1827#define S_DEAD 16 /* removed, but still open directory */
1828#define S_NOQUOTA 32 /* Inode is not counted to quota */
1829#define S_DIRSYNC 64 /* Directory modifications are synchronous */
1830#define S_NOCMTIME 128 /* Do not update file c/mtime */
1831#define S_SWAPFILE 256 /* Do not truncate: swapon got its bmaps */
1832#define S_PRIVATE 512 /* Inode is fs-internal */
1833#define S_IMA 1024 /* Inode has an associated IMA struct */
1834#define S_AUTOMOUNT 2048 /* Automount/referral quasi-directory */
1835#define S_NOSEC 4096 /* no suid or xattr security attributes */
Matthew Wilcox6cd176a2015-02-16 15:59:25 -08001836#ifdef CONFIG_FS_DAX
Matthew Wilcoxfbbbad42015-02-16 15:58:53 -08001837#define S_DAX 8192 /* Direct Access, avoiding the page cache */
1838#else
1839#define S_DAX 0 /* Make all the DAX code disappear */
1840#endif
David Howellsbbc10962012-10-15 16:40:35 +01001841
1842/*
1843 * Note that nosuid etc flags are inode-specific: setting some file-system
1844 * flags just means all the inodes inherit those flags by default. It might be
1845 * possible to override it selectively if you really wanted to with some
1846 * ioctl() that is not currently implemented.
1847 *
1848 * Exception: MS_RDONLY is always applied to the entire file system.
1849 *
1850 * Unfortunately, it is possible to change a filesystems flags with it mounted
1851 * with files in use. This means that all of the inodes will not have their
1852 * i_flags updated. Hence, i_flags no longer inherit the superblock mount
1853 * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org
1854 */
1855#define __IS_FLG(inode, flg) ((inode)->i_sb->s_flags & (flg))
1856
1857#define IS_RDONLY(inode) ((inode)->i_sb->s_flags & MS_RDONLY)
1858#define IS_SYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS) || \
1859 ((inode)->i_flags & S_SYNC))
1860#define IS_DIRSYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS|MS_DIRSYNC) || \
1861 ((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
1862#define IS_MANDLOCK(inode) __IS_FLG(inode, MS_MANDLOCK)
1863#define IS_NOATIME(inode) __IS_FLG(inode, MS_RDONLY|MS_NOATIME)
1864#define IS_I_VERSION(inode) __IS_FLG(inode, MS_I_VERSION)
1865
1866#define IS_NOQUOTA(inode) ((inode)->i_flags & S_NOQUOTA)
1867#define IS_APPEND(inode) ((inode)->i_flags & S_APPEND)
1868#define IS_IMMUTABLE(inode) ((inode)->i_flags & S_IMMUTABLE)
1869#define IS_POSIXACL(inode) __IS_FLG(inode, MS_POSIXACL)
1870
1871#define IS_DEADDIR(inode) ((inode)->i_flags & S_DEAD)
1872#define IS_NOCMTIME(inode) ((inode)->i_flags & S_NOCMTIME)
1873#define IS_SWAPFILE(inode) ((inode)->i_flags & S_SWAPFILE)
1874#define IS_PRIVATE(inode) ((inode)->i_flags & S_PRIVATE)
1875#define IS_IMA(inode) ((inode)->i_flags & S_IMA)
1876#define IS_AUTOMOUNT(inode) ((inode)->i_flags & S_AUTOMOUNT)
1877#define IS_NOSEC(inode) ((inode)->i_flags & S_NOSEC)
Matthew Wilcoxfbbbad42015-02-16 15:58:53 -08001878#define IS_DAX(inode) ((inode)->i_flags & S_DAX)
David Howellsbbc10962012-10-15 16:40:35 +01001879
Miklos Szeredi787fb6b2014-10-24 00:14:36 +02001880#define IS_WHITEOUT(inode) (S_ISCHR(inode->i_mode) && \
1881 (inode)->i_rdev == WHITEOUT_DEV)
1882
Eric W. Biederman0bd23d092016-06-29 14:54:46 -05001883static inline bool HAS_UNMAPPED_ID(struct inode *inode)
1884{
1885 return !uid_valid(inode->i_uid) || !gid_valid(inode->i_gid);
1886}
1887
David Howellsbbc10962012-10-15 16:40:35 +01001888/*
Dave Chinner250df6e2011-03-22 22:23:36 +11001889 * Inode state bits. Protected by inode->i_lock
Joern Engel1c0eeaf2007-10-16 23:30:44 -07001890 *
1891 * Three bits determine the dirty state of the inode, I_DIRTY_SYNC,
1892 * I_DIRTY_DATASYNC and I_DIRTY_PAGES.
1893 *
1894 * Four bits define the lifetime of an inode. Initially, inodes are I_NEW,
1895 * until that flag is cleared. I_WILL_FREE, I_FREEING and I_CLEAR are set at
1896 * various stages of removing an inode.
1897 *
Christoph Hellwigeaff8072009-12-17 14:25:01 +01001898 * Two bits are used for locking and completion notification, I_NEW and I_SYNC.
Joern Engel1c0eeaf2007-10-16 23:30:44 -07001899 *
Joern Engele7ca2d42008-02-06 01:36:59 -08001900 * I_DIRTY_SYNC Inode is dirty, but doesn't have to be written on
1901 * fdatasync(). i_atime is the usual cause.
Jan Kara000cb482008-02-14 19:31:32 -08001902 * I_DIRTY_DATASYNC Data-related inode changes pending. We keep track of
1903 * these changes separately from I_DIRTY_SYNC so that we
1904 * don't have to write inode on fdatasync() when only
1905 * mtime has changed in it.
Joern Engel1c0eeaf2007-10-16 23:30:44 -07001906 * I_DIRTY_PAGES Inode has dirty pages. Inode itself may be clean.
Christoph Hellwigeaff8072009-12-17 14:25:01 +01001907 * I_NEW Serves as both a mutex and completion notification.
1908 * New inodes set I_NEW. If two processes both create
1909 * the same inode, one of them will release its inode and
1910 * wait for I_NEW to be released before returning.
1911 * Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can
1912 * also cause waiting on I_NEW, without I_NEW actually
1913 * being set. find_inode() uses this to prevent returning
1914 * nearly-dead inodes.
Joern Engel1c0eeaf2007-10-16 23:30:44 -07001915 * I_WILL_FREE Must be set when calling write_inode_now() if i_count
1916 * is zero. I_FREEING must be set when I_WILL_FREE is
1917 * cleared.
1918 * I_FREEING Set when inode is about to be freed but still has dirty
1919 * pages or buffers attached or the inode itself is still
1920 * dirty.
Jan Karadbd57682012-05-03 14:48:02 +02001921 * I_CLEAR Added by clear_inode(). In this state the inode is
1922 * clean and can be destroyed. Inode keeps I_FREEING.
Joern Engel1c0eeaf2007-10-16 23:30:44 -07001923 *
1924 * Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are
1925 * prohibited for many purposes. iget() must wait for
1926 * the inode to be completely released, then create it
1927 * anew. Other functions will just ignore such inodes,
Christoph Hellwigeaff8072009-12-17 14:25:01 +01001928 * if appropriate. I_NEW is used for waiting.
Joern Engel1c0eeaf2007-10-16 23:30:44 -07001929 *
Jan Kara169ebd92012-05-03 14:48:03 +02001930 * I_SYNC Writeback of inode is running. The bit is set during
1931 * data writeback, and cleared with a wakeup on the bit
1932 * address once it is done. The bit is also used to pin
1933 * the inode in memory for flusher thread.
Joern Engel1c0eeaf2007-10-16 23:30:44 -07001934 *
Christoph Hellwigbd5fe6c2011-06-24 14:29:43 -04001935 * I_REFERENCED Marks the inode as recently references on the LRU list.
1936 *
1937 * I_DIO_WAKEUP Never set. Only used as a key for wait_on_bit().
1938 *
Tejun Heo682aa8e2015-05-28 14:50:53 -04001939 * I_WB_SWITCH Cgroup bdi_writeback switching in progress. Used to
1940 * synchronize competing switching instances and to tell
1941 * wb stat updates to grab mapping->tree_lock. See
1942 * inode_switch_wb_work_fn() for details.
1943 *
Joern Engel1c0eeaf2007-10-16 23:30:44 -07001944 * Q: What is the difference between I_WILL_FREE and I_FREEING?
Joern Engel1c0eeaf2007-10-16 23:30:44 -07001945 */
Nick Piggin9e38d862010-10-23 06:55:17 -04001946#define I_DIRTY_SYNC (1 << 0)
1947#define I_DIRTY_DATASYNC (1 << 1)
1948#define I_DIRTY_PAGES (1 << 2)
Christoph Hellwigeaff8072009-12-17 14:25:01 +01001949#define __I_NEW 3
1950#define I_NEW (1 << __I_NEW)
Nick Piggin9e38d862010-10-23 06:55:17 -04001951#define I_WILL_FREE (1 << 4)
1952#define I_FREEING (1 << 5)
1953#define I_CLEAR (1 << 6)
Christoph Hellwigeaff8072009-12-17 14:25:01 +01001954#define __I_SYNC 7
Joern Engel1c0eeaf2007-10-16 23:30:44 -07001955#define I_SYNC (1 << __I_SYNC)
Nick Piggin9e38d862010-10-23 06:55:17 -04001956#define I_REFERENCED (1 << 8)
Christoph Hellwigbd5fe6c2011-06-24 14:29:43 -04001957#define __I_DIO_WAKEUP 9
Eric Sandeenac74d8d2015-04-16 15:04:56 -05001958#define I_DIO_WAKEUP (1 << __I_DIO_WAKEUP)
Al Virof4e0c302013-06-11 08:34:36 +04001959#define I_LINKABLE (1 << 10)
Theodore Ts'o0ae45f62015-02-02 00:37:00 -05001960#define I_DIRTY_TIME (1 << 11)
1961#define __I_DIRTY_TIME_EXPIRED 12
1962#define I_DIRTY_TIME_EXPIRED (1 << __I_DIRTY_TIME_EXPIRED)
Tejun Heo682aa8e2015-05-28 14:50:53 -04001963#define I_WB_SWITCH (1 << 13)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001964
1965#define I_DIRTY (I_DIRTY_SYNC | I_DIRTY_DATASYNC | I_DIRTY_PAGES)
Theodore Ts'o0ae45f62015-02-02 00:37:00 -05001966#define I_DIRTY_ALL (I_DIRTY | I_DIRTY_TIME)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001967
1968extern void __mark_inode_dirty(struct inode *, int);
1969static inline void mark_inode_dirty(struct inode *inode)
1970{
1971 __mark_inode_dirty(inode, I_DIRTY);
1972}
1973
1974static inline void mark_inode_dirty_sync(struct inode *inode)
1975{
1976 __mark_inode_dirty(inode, I_DIRTY_SYNC);
1977}
1978
Miklos Szeredi7ada4db2011-11-21 12:11:32 +01001979extern void inc_nlink(struct inode *inode);
1980extern void drop_nlink(struct inode *inode);
1981extern void clear_nlink(struct inode *inode);
1982extern void set_nlink(struct inode *inode, unsigned int nlink);
Dave Hansend8c76e62006-09-30 23:29:04 -07001983
1984static inline void inode_inc_link_count(struct inode *inode)
1985{
1986 inc_nlink(inode);
Alexey Dobriyan66d21732006-03-23 03:00:51 -08001987 mark_inode_dirty(inode);
1988}
1989
Dave Hansen9a53c3a2006-09-30 23:29:03 -07001990static inline void inode_dec_link_count(struct inode *inode)
1991{
1992 drop_nlink(inode);
Alexey Dobriyan66d21732006-03-23 03:00:51 -08001993 mark_inode_dirty(inode);
1994}
1995
Jean Noel Cordenner25ec56b2008-01-28 23:58:27 -05001996/**
1997 * inode_inc_iversion - increments i_version
1998 * @inode: inode that need to be updated
1999 *
2000 * Every time the inode is modified, the i_version field will be incremented.
2001 * The filesystem has to be mounted with i_version flag
2002 */
2003
2004static inline void inode_inc_iversion(struct inode *inode)
2005{
2006 spin_lock(&inode->i_lock);
2007 inode->i_version++;
2008 spin_unlock(&inode->i_lock);
2009}
2010
Josef Bacikc3b2da32012-03-26 09:59:21 -04002011enum file_time_flags {
2012 S_ATIME = 1,
2013 S_MTIME = 2,
2014 S_CTIME = 4,
2015 S_VERSION = 8,
2016};
2017
Al Virobadcf2b2013-07-16 18:15:46 +04002018extern void touch_atime(const struct path *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002019static inline void file_accessed(struct file *file)
2020{
2021 if (!(file->f_flags & O_NOATIME))
Al Viro68ac1232012-03-15 08:21:57 -04002022 touch_atime(&file->f_path);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002023}
2024
2025int sync_inode(struct inode *inode, struct writeback_control *wbc);
Christoph Hellwigc37650162010-10-06 10:48:20 +02002026int sync_inode_metadata(struct inode *inode, int wait);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002027
Linus Torvalds1da177e2005-04-16 15:20:36 -07002028struct file_system_type {
2029 const char *name;
2030 int fs_flags;
David Howellsbbc10962012-10-15 16:40:35 +01002031#define FS_REQUIRES_DEV 1
2032#define FS_BINARY_MOUNTDATA 2
2033#define FS_HAS_SUBTYPE 4
Eric W. Biederman0c55cfc2012-07-26 21:42:03 -07002034#define FS_USERNS_MOUNT 8 /* Can be mounted by userns root */
David Howellsbbc10962012-10-15 16:40:35 +01002035#define FS_RENAME_DOES_D_MOVE 32768 /* FS will handle d_move() during rename() internally. */
Al Viroc96e41e2010-07-25 00:17:56 +04002036 struct dentry *(*mount) (struct file_system_type *, int,
2037 const char *, void *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002038 void (*kill_sb) (struct super_block *);
2039 struct module *owner;
2040 struct file_system_type * next;
Al Viroa5166162011-12-12 22:53:00 -05002041 struct hlist_head fs_supers;
Peter Zijlstrad475fd42007-10-15 14:51:31 +02002042
Ingo Molnarcf516242006-07-03 00:25:27 -07002043 struct lock_class_key s_lock_key;
Arjan van de Ven897c6ff2006-07-03 00:25:28 -07002044 struct lock_class_key s_umount_key;
Roland Dreier51ee0492010-04-27 14:23:57 -07002045 struct lock_class_key s_vfs_rename_key;
Jan Kara5accdf82012-06-12 16:20:34 +02002046 struct lock_class_key s_writers_key[SB_FREEZE_LEVELS];
Peter Zijlstrad475fd42007-10-15 14:51:31 +02002047
2048 struct lock_class_key i_lock_key;
2049 struct lock_class_key i_mutex_key;
Peter Zijlstra14358e62007-10-14 01:38:33 +02002050 struct lock_class_key i_mutex_dir_key;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002051};
2052
Eric W. Biederman7f78e032013-03-02 19:39:14 -08002053#define MODULE_ALIAS_FS(NAME) MODULE_ALIAS("fs-" NAME)
2054
Eric W. Biedermand91ee872016-05-23 14:51:59 -05002055extern struct dentry *mount_ns(struct file_system_type *fs_type,
2056 int flags, void *data, void *ns, struct user_namespace *user_ns,
2057 int (*fill_super)(struct super_block *, void *, int));
Al Viro152a0832010-07-25 00:46:55 +04002058extern struct dentry *mount_bdev(struct file_system_type *fs_type,
2059 int flags, const char *dev_name, void *data,
2060 int (*fill_super)(struct super_block *, void *, int));
Al Virofc14f2f2010-07-25 01:48:30 +04002061extern struct dentry *mount_single(struct file_system_type *fs_type,
2062 int flags, void *data,
2063 int (*fill_super)(struct super_block *, void *, int));
Al Viro3c26ff62010-07-25 11:46:36 +04002064extern struct dentry *mount_nodev(struct file_system_type *fs_type,
2065 int flags, void *data,
2066 int (*fill_super)(struct super_block *, void *, int));
Al Viroea441d12011-11-16 21:43:59 -05002067extern struct dentry *mount_subtree(struct vfsmount *mnt, const char *path);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002068void generic_shutdown_super(struct super_block *sb);
2069void kill_block_super(struct super_block *sb);
2070void kill_anon_super(struct super_block *sb);
2071void kill_litter_super(struct super_block *sb);
2072void deactivate_super(struct super_block *sb);
Al Viro74dbbdd2009-05-06 01:07:50 -04002073void deactivate_locked_super(struct super_block *sb);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002074int set_anon_super(struct super_block *s, void *data);
Al Viro0ee5dc62011-07-07 15:44:25 -04002075int get_anon_bdev(dev_t *);
2076void free_anon_bdev(dev_t);
Eric W. Biederman6e4eab52016-05-24 09:29:01 -05002077struct super_block *sget_userns(struct file_system_type *type,
2078 int (*test)(struct super_block *,void *),
2079 int (*set)(struct super_block *,void *),
2080 int flags, struct user_namespace *user_ns,
2081 void *data);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002082struct super_block *sget(struct file_system_type *type,
2083 int (*test)(struct super_block *,void *),
2084 int (*set)(struct super_block *,void *),
David Howells9249e172012-06-25 12:55:37 +01002085 int flags, void *data);
Andreas Gruenbacherbba0bd32016-09-29 17:48:35 +02002086extern struct dentry *mount_pseudo_xattr(struct file_system_type *, char *,
2087 const struct super_operations *ops,
2088 const struct xattr_handler **xattr,
2089 const struct dentry_operations *dops,
2090 unsigned long);
2091
2092static inline struct dentry *
2093mount_pseudo(struct file_system_type *fs_type, char *name,
2094 const struct super_operations *ops,
2095 const struct dentry_operations *dops, unsigned long magic)
2096{
2097 return mount_pseudo_xattr(fs_type, name, ops, NULL, dops, magic);
2098}
Linus Torvalds1da177e2005-04-16 15:20:36 -07002099
2100/* Alas, no aliases. Too much hassle with bringing module.h everywhere */
2101#define fops_get(fops) \
2102 (((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
2103#define fops_put(fops) \
2104 do { if (fops) module_put((fops)->owner); } while(0)
Al Viroe84f9e52013-09-22 14:17:15 -04002105/*
2106 * This one is to be used *ONLY* from ->open() instances.
2107 * fops must be non-NULL, pinned down *and* module dependencies
2108 * should be sufficient to pin the caller down as well.
2109 */
2110#define replace_fops(f, fops) \
2111 do { \
2112 struct file *__file = (f); \
2113 fops_put(__file->f_op); \
2114 BUG_ON(!(__file->f_op = (fops))); \
2115 } while(0)
Linus Torvalds1da177e2005-04-16 15:20:36 -07002116
2117extern int register_filesystem(struct file_system_type *);
2118extern int unregister_filesystem(struct file_system_type *);
Pavel Emelyanov8bf97252007-10-18 23:40:02 -07002119extern struct vfsmount *kern_mount_data(struct file_system_type *, void *data);
2120#define kern_mount(type) kern_mount_data(type, NULL)
Tim Chen423e0ab2011-07-19 09:32:38 -07002121extern void kern_unmount(struct vfsmount *mnt);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002122extern int may_umount_tree(struct vfsmount *);
2123extern int may_umount(struct vfsmount *);
Seunghun Lee5e6123f2014-09-14 22:15:10 +09002124extern long do_mount(const char *, const char __user *,
2125 const char *, unsigned long, void *);
Al Viro589ff872009-04-18 03:28:19 -04002126extern struct vfsmount *collect_mounts(struct path *);
Al Viro8aec0802007-06-07 12:20:32 -04002127extern void drop_collected_mounts(struct vfsmount *);
Al Viro1f707132010-01-30 22:51:25 -05002128extern int iterate_mounts(int (*)(struct vfsmount *, void *), void *,
2129 struct vfsmount *);
Christoph Hellwigebabe9a2010-07-07 18:53:11 +02002130extern int vfs_statfs(struct path *, struct kstatfs *);
Al Viroc8b91ac2011-03-12 10:41:39 -05002131extern int user_statfs(const char __user *, struct kstatfs *);
2132extern int fd_statfs(int, struct kstatfs *);
Al Virocf31e702012-01-02 22:28:36 -05002133extern int vfs_ustat(dev_t, struct kstatfs *);
Josef Bacik18e9e512010-03-23 10:34:56 -04002134extern int freeze_super(struct super_block *super);
2135extern int thaw_super(struct super_block *super);
Al Viro02125a82011-12-05 08:43:34 -05002136extern bool our_mnt(struct vfsmount *mnt);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002137
Al Viroce3b0f82009-03-29 19:08:22 -04002138extern int current_umask(void);
2139
J. Bruce Fields5a146962012-08-28 07:50:40 -07002140extern void ihold(struct inode * inode);
2141extern void iput(struct inode *);
Theodore Ts'o0ae45f62015-02-02 00:37:00 -05002142extern int generic_update_time(struct inode *, struct timespec *, int);
J. Bruce Fields5a146962012-08-28 07:50:40 -07002143
Miklos Szeredif87fd4c2006-01-16 22:14:23 -08002144/* /sys/fs */
Greg Kroah-Hartman00d26662007-10-29 14:17:23 -06002145extern struct kobject *fs_kobj;
Miklos Szeredif87fd4c2006-01-16 22:14:23 -08002146
Kirill A. Shutemov09cbfea2016-04-01 15:29:47 +03002147#define MAX_RW_COUNT (INT_MAX & PAGE_MASK)
Thomas Petazzonibfcd17a2008-08-06 15:12:22 +02002148
Jeff Layton9e8925b2015-11-16 09:49:34 -05002149#ifdef CONFIG_MANDATORY_FILE_LOCKING
Jeff Laytond7a06982014-03-10 09:54:15 -04002150extern int locks_mandatory_locked(struct file *);
Christoph Hellwigacc15572015-12-03 12:59:49 +01002151extern int locks_mandatory_area(struct inode *, struct file *, loff_t, loff_t, unsigned char);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002152
2153/*
2154 * Candidates for mandatory locking have the setgid bit set
2155 * but no group execute bit - an otherwise meaningless combination.
2156 */
Pavel Emelyanova16877c2007-10-01 14:41:11 -07002157
2158static inline int __mandatory_lock(struct inode *ino)
2159{
2160 return (ino->i_mode & (S_ISGID | S_IXGRP)) == S_ISGID;
2161}
2162
2163/*
2164 * ... and these candidates should be on MS_MANDLOCK mounted fs,
2165 * otherwise these will be advisory locks
2166 */
2167
2168static inline int mandatory_lock(struct inode *ino)
2169{
2170 return IS_MANDLOCK(ino) && __mandatory_lock(ino);
2171}
Linus Torvalds1da177e2005-04-16 15:20:36 -07002172
Jeff Laytond7a06982014-03-10 09:54:15 -04002173static inline int locks_verify_locked(struct file *file)
Linus Torvalds1da177e2005-04-16 15:20:36 -07002174{
Miklos Szeredic568d682016-09-16 12:44:20 +02002175 if (mandatory_lock(locks_inode(file)))
Jeff Laytond7a06982014-03-10 09:54:15 -04002176 return locks_mandatory_locked(file);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002177 return 0;
2178}
2179
Linus Torvalds1da177e2005-04-16 15:20:36 -07002180static inline int locks_verify_truncate(struct inode *inode,
Christoph Hellwigacc15572015-12-03 12:59:49 +01002181 struct file *f,
Linus Torvalds1da177e2005-04-16 15:20:36 -07002182 loff_t size)
2183{
Christoph Hellwigacc15572015-12-03 12:59:49 +01002184 if (!inode->i_flctx || !mandatory_lock(inode))
2185 return 0;
2186
2187 if (size < inode->i_size) {
2188 return locks_mandatory_area(inode, f, size, inode->i_size - 1,
2189 F_WRLCK);
2190 } else {
2191 return locks_mandatory_area(inode, f, inode->i_size, size - 1,
2192 F_WRLCK);
2193 }
Linus Torvalds1da177e2005-04-16 15:20:36 -07002194}
2195
Jeff Layton9e8925b2015-11-16 09:49:34 -05002196#else /* !CONFIG_MANDATORY_FILE_LOCKING */
2197
2198static inline int locks_mandatory_locked(struct file *file)
2199{
2200 return 0;
2201}
2202
Linus Torvaldsfce205e2016-01-12 16:30:34 -08002203static inline int locks_mandatory_area(struct inode *inode, struct file *filp,
2204 loff_t start, loff_t end, unsigned char type)
Jeff Layton9e8925b2015-11-16 09:49:34 -05002205{
2206 return 0;
2207}
2208
2209static inline int __mandatory_lock(struct inode *inode)
2210{
2211 return 0;
2212}
2213
2214static inline int mandatory_lock(struct inode *inode)
2215{
2216 return 0;
2217}
2218
2219static inline int locks_verify_locked(struct file *file)
2220{
2221 return 0;
2222}
2223
2224static inline int locks_verify_truncate(struct inode *inode, struct file *filp,
2225 size_t size)
2226{
2227 return 0;
2228}
2229
2230#endif /* CONFIG_MANDATORY_FILE_LOCKING */
2231
2232
2233#ifdef CONFIG_FILE_LOCKING
Linus Torvalds1da177e2005-04-16 15:20:36 -07002234static inline int break_lease(struct inode *inode, unsigned int mode)
2235{
Jeff Layton24cbe782014-02-03 12:13:06 -05002236 /*
2237 * Since this check is lockless, we must ensure that any refcounts
Jeff Layton8116bf42015-01-21 20:44:01 -05002238 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2239 * could end up racing with tasks trying to set a new lease on this
2240 * file.
Jeff Layton24cbe782014-02-03 12:13:06 -05002241 */
2242 smp_mb();
Jeff Layton8634b512015-01-16 15:05:55 -05002243 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
J. Bruce Fieldsdf4e8d22012-03-05 13:18:59 -05002244 return __break_lease(inode, mode, FL_LEASE);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002245 return 0;
2246}
J. Bruce Fieldsdf4e8d22012-03-05 13:18:59 -05002247
2248static inline int break_deleg(struct inode *inode, unsigned int mode)
2249{
Jeff Layton962bd402014-06-10 12:24:40 -04002250 /*
2251 * Since this check is lockless, we must ensure that any refcounts
Jeff Layton8116bf42015-01-21 20:44:01 -05002252 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2253 * could end up racing with tasks trying to set a new lease on this
2254 * file.
Jeff Layton962bd402014-06-10 12:24:40 -04002255 */
2256 smp_mb();
Jeff Layton8634b512015-01-16 15:05:55 -05002257 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
J. Bruce Fieldsdf4e8d22012-03-05 13:18:59 -05002258 return __break_lease(inode, mode, FL_DELEG);
2259 return 0;
2260}
2261
J. Bruce Fields5a146962012-08-28 07:50:40 -07002262static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode)
2263{
2264 int ret;
2265
2266 ret = break_deleg(inode, O_WRONLY|O_NONBLOCK);
2267 if (ret == -EWOULDBLOCK && delegated_inode) {
2268 *delegated_inode = inode;
2269 ihold(inode);
2270 }
2271 return ret;
2272}
2273
2274static inline int break_deleg_wait(struct inode **delegated_inode)
2275{
2276 int ret;
2277
2278 ret = break_deleg(*delegated_inode, O_WRONLY);
2279 iput(*delegated_inode);
2280 *delegated_inode = NULL;
2281 return ret;
2282}
2283
Christoph Hellwig11afe9f2015-01-21 19:17:03 +01002284static inline int break_layout(struct inode *inode, bool wait)
2285{
2286 smp_mb();
2287 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2288 return __break_lease(inode,
2289 wait ? O_WRONLY : O_WRONLY | O_NONBLOCK,
2290 FL_LAYOUT);
2291 return 0;
2292}
2293
Thomas Petazzonibfcd17a2008-08-06 15:12:22 +02002294#else /* !CONFIG_FILE_LOCKING */
Steven Whitehouseaf5df562009-01-20 10:29:46 +00002295static inline int break_lease(struct inode *inode, unsigned int mode)
2296{
2297 return 0;
2298}
2299
J. Bruce Fieldsdf4e8d22012-03-05 13:18:59 -05002300static inline int break_deleg(struct inode *inode, unsigned int mode)
2301{
2302 return 0;
2303}
J. Bruce Fields5a146962012-08-28 07:50:40 -07002304
2305static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode)
2306{
2307 return 0;
2308}
2309
2310static inline int break_deleg_wait(struct inode **delegated_inode)
2311{
2312 BUG();
2313 return 0;
2314}
2315
Christoph Hellwig11afe9f2015-01-21 19:17:03 +01002316static inline int break_layout(struct inode *inode, bool wait)
2317{
2318 return 0;
2319}
2320
Thomas Petazzonibfcd17a2008-08-06 15:12:22 +02002321#endif /* CONFIG_FILE_LOCKING */
Linus Torvalds1da177e2005-04-16 15:20:36 -07002322
2323/* fs/open.c */
Jeff Laytonadb5c242012-10-10 16:43:13 -04002324struct audit_names;
Jeff Layton91a27b22012-10-10 15:25:28 -04002325struct filename {
Jeff Laytonadb5c242012-10-10 16:43:13 -04002326 const char *name; /* pointer to actual string */
2327 const __user char *uptr; /* original userland pointer */
2328 struct audit_names *aname;
Paul Moore55422d02015-01-22 00:00:23 -05002329 int refcnt;
Al Virofd2f7cb2015-02-22 20:07:13 -05002330 const char iname[];
Jeff Layton91a27b22012-10-10 15:25:28 -04002331};
Linus Torvalds1da177e2005-04-16 15:20:36 -07002332
Al Viro7df818b2016-03-25 14:24:09 -04002333extern long vfs_truncate(const struct path *, loff_t);
NeilBrown4a301312006-01-08 01:02:39 -08002334extern int do_truncate(struct dentry *, loff_t start, unsigned int time_attrs,
2335 struct file *filp);
Anna Schumaker72c72bd2014-11-07 14:44:25 -05002336extern int vfs_fallocate(struct file *file, int mode, loff_t offset,
Ankit Jain3e63cbb2009-06-19 14:28:07 -04002337 loff_t len);
Jason Uhlenkott8e8a1402007-10-20 03:16:18 +02002338extern long do_sys_open(int dfd, const char __user *filename, int flags,
Al Viroa218d0f2011-11-21 14:59:34 -05002339 umode_t mode);
Jeff Layton669abf42012-10-10 16:43:10 -04002340extern struct file *file_open_name(struct filename *, int, umode_t);
Al Viroa218d0f2011-11-21 14:59:34 -05002341extern struct file *filp_open(const char *, int, umode_t);
Al Viro73d049a2011-03-11 12:08:24 -05002342extern struct file *file_open_root(struct dentry *, struct vfsmount *,
Jann Horn378c6522016-03-22 14:25:36 -07002343 const char *, int, umode_t);
Al Viro765927b2012-06-26 21:58:53 +04002344extern struct file * dentry_open(const struct path *, int, const struct cred *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002345extern int filp_close(struct file *, fl_owner_t id);
Jeff Layton91a27b22012-10-10 15:25:28 -04002346
David Drysdale51f39a12014-12-12 16:57:29 -08002347extern struct filename *getname_flags(const char __user *, int, int *);
Jeff Layton91a27b22012-10-10 15:25:28 -04002348extern struct filename *getname(const char __user *);
Linus Torvaldsc4ad8f92014-02-05 12:54:53 -08002349extern struct filename *getname_kernel(const char *);
Paul Moore55422d02015-01-22 00:00:23 -05002350extern void putname(struct filename *name);
Jeff Layton91a27b22012-10-10 15:25:28 -04002351
Al Viro47237682012-06-10 05:01:45 -04002352enum {
2353 FILE_CREATED = 1,
2354 FILE_OPENED = 2
2355};
Al Viro30d90492012-06-22 12:40:19 +04002356extern int finish_open(struct file *file, struct dentry *dentry,
2357 int (*open)(struct inode *, struct file *),
2358 int *opened);
Al Viroe45198a2012-06-10 06:48:09 -04002359extern int finish_no_open(struct file *file, struct dentry *dentry);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002360
Ankit Jain3e63cbb2009-06-19 14:28:07 -04002361/* fs/ioctl.c */
2362
2363extern int ioctl_preallocate(struct file *filp, void __user *argp);
2364
Linus Torvalds1da177e2005-04-16 15:20:36 -07002365/* fs/dcache.c */
2366extern void __init vfs_caches_init_early(void);
Mel Gorman4248b0d2015-08-06 15:46:20 -07002367extern void __init vfs_caches_init(void);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002368
Christoph Lameterb86c0892006-12-06 20:32:57 -08002369extern struct kmem_cache *names_cachep;
2370
Jeff Laytona608ca22012-10-10 15:25:26 -04002371#define __getname() kmem_cache_alloc(names_cachep, GFP_KERNEL)
Vegard Nossum3446a8a2009-05-16 11:22:14 +02002372#define __putname(name) kmem_cache_free(names_cachep, (void *)(name))
Linus Torvalds1da177e2005-04-16 15:20:36 -07002373
David Howells93614012006-09-30 20:45:40 +02002374#ifdef CONFIG_BLOCK
Linus Torvalds1da177e2005-04-16 15:20:36 -07002375extern int register_blkdev(unsigned int, const char *);
Akinobu Mitaf4480242007-07-17 04:03:47 -07002376extern void unregister_blkdev(unsigned int, const char *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002377extern struct block_device *bdget(dev_t);
Alan Jenkinsdddac6a2009-07-29 21:07:55 +02002378extern struct block_device *bdgrab(struct block_device *bdev);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002379extern void bd_set_size(struct block_device *, loff_t size);
2380extern void bd_forget(struct inode *inode);
2381extern void bdput(struct block_device *);
Nick Piggin585d3bc2009-02-25 10:44:19 +01002382extern void invalidate_bdev(struct block_device *);
Jan Kara5c0d6b62012-07-03 16:45:31 +02002383extern void iterate_bdevs(void (*)(struct block_device *, void *), void *);
Nick Piggin585d3bc2009-02-25 10:44:19 +01002384extern int sync_blockdev(struct block_device *bdev);
Al Viroff01bb42011-09-16 02:31:11 -04002385extern void kill_bdev(struct block_device *);
Nick Piggin585d3bc2009-02-25 10:44:19 +01002386extern struct super_block *freeze_bdev(struct block_device *);
Eric Sandeenc2d75432009-03-31 15:23:46 -07002387extern void emergency_thaw_all(void);
Nick Piggin585d3bc2009-02-25 10:44:19 +01002388extern int thaw_bdev(struct block_device *bdev, struct super_block *sb);
2389extern int fsync_bdev(struct block_device *);
Tejun Heoa212b102015-05-22 17:13:33 -04002390
2391extern struct super_block *blockdev_superblock;
2392
2393static inline bool sb_is_blkdev_sb(struct super_block *sb)
2394{
2395 return sb == blockdev_superblock;
2396}
David Howells93614012006-09-30 20:45:40 +02002397#else
2398static inline void bd_forget(struct inode *inode) {}
Al Viro47e44912009-04-01 07:07:16 -04002399static inline int sync_blockdev(struct block_device *bdev) { return 0; }
Al Viroff01bb42011-09-16 02:31:11 -04002400static inline void kill_bdev(struct block_device *bdev) {}
Al Viro47e44912009-04-01 07:07:16 -04002401static inline void invalidate_bdev(struct block_device *bdev) {}
2402
2403static inline struct super_block *freeze_bdev(struct block_device *sb)
2404{
2405 return NULL;
2406}
2407
2408static inline int thaw_bdev(struct block_device *bdev, struct super_block *sb)
2409{
2410 return 0;
2411}
Jan Kara5c0d6b62012-07-03 16:45:31 +02002412
2413static inline void iterate_bdevs(void (*f)(struct block_device *, void *), void *arg)
2414{
2415}
Jan Karaa8855992013-07-09 22:36:45 +08002416
Yaowei Bai3e004ee2015-11-17 14:40:09 +08002417static inline bool sb_is_blkdev_sb(struct super_block *sb)
Jan Karaa8855992013-07-09 22:36:45 +08002418{
Yaowei Bai3e004ee2015-11-17 14:40:09 +08002419 return false;
Jan Karaa8855992013-07-09 22:36:45 +08002420}
David Howells93614012006-09-30 20:45:40 +02002421#endif
Jan Kara60b06802009-04-27 16:43:53 +02002422extern int sync_filesystem(struct super_block *);
David Howells93614012006-09-30 20:45:40 +02002423extern const struct file_operations def_blk_fops;
Arjan van de Ven4b6f5d22006-03-28 01:56:42 -08002424extern const struct file_operations def_chr_fops;
David Howells93614012006-09-30 20:45:40 +02002425#ifdef CONFIG_BLOCK
Linus Torvalds1da177e2005-04-16 15:20:36 -07002426extern int ioctl_by_bdev(struct block_device *, unsigned, unsigned long);
Al Viro56b26ad2008-09-19 03:17:36 -04002427extern int blkdev_ioctl(struct block_device *, fmode_t, unsigned, unsigned long);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002428extern long compat_blkdev_ioctl(struct file *, unsigned, unsigned long);
Tejun Heoe525fd82010-11-13 11:55:17 +01002429extern int blkdev_get(struct block_device *bdev, fmode_t mode, void *holder);
Tejun Heod4d77622010-11-13 11:55:18 +01002430extern struct block_device *blkdev_get_by_path(const char *path, fmode_t mode,
2431 void *holder);
2432extern struct block_device *blkdev_get_by_dev(dev_t dev, fmode_t mode,
2433 void *holder);
Al Viro4385bab2013-05-05 22:11:03 -04002434extern void blkdev_put(struct block_device *bdev, fmode_t mode);
Jarod Wilsonbe324172015-05-06 12:26:22 +08002435extern int __blkdev_reread_part(struct block_device *bdev);
2436extern int blkdev_reread_part(struct block_device *bdev);
2437
Jun'ichi Nomura641dc632006-03-27 01:17:57 -08002438#ifdef CONFIG_SYSFS
Tejun Heoe09b4572010-11-13 11:55:17 +01002439extern int bd_link_disk_holder(struct block_device *bdev, struct gendisk *disk);
Tejun Heo49731ba2011-01-14 18:43:57 +01002440extern void bd_unlink_disk_holder(struct block_device *bdev,
2441 struct gendisk *disk);
Jun'ichi Nomura641dc632006-03-27 01:17:57 -08002442#else
Tejun Heoe09b4572010-11-13 11:55:17 +01002443static inline int bd_link_disk_holder(struct block_device *bdev,
2444 struct gendisk *disk)
2445{
2446 return 0;
2447}
Tejun Heo49731ba2011-01-14 18:43:57 +01002448static inline void bd_unlink_disk_holder(struct block_device *bdev,
2449 struct gendisk *disk)
2450{
2451}
Jun'ichi Nomura641dc632006-03-27 01:17:57 -08002452#endif
David Howells93614012006-09-30 20:45:40 +02002453#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -07002454
2455/* fs/char_dev.c */
Joe Korty68eef3b2006-03-31 02:30:32 -08002456#define CHRDEV_MAJOR_HASH_SIZE 255
Linus Walleij49db08c2016-02-19 15:36:07 +01002457/* Marks the bottom of the first segment of free char majors */
2458#define CHRDEV_MAJOR_DYN_END 234
Linus Torvalds1da177e2005-04-16 15:20:36 -07002459extern int alloc_chrdev_region(dev_t *, unsigned, unsigned, const char *);
2460extern int register_chrdev_region(dev_t, unsigned, const char *);
Tejun Heo1905b1b2009-08-06 18:13:23 +09002461extern int __register_chrdev(unsigned int major, unsigned int baseminor,
2462 unsigned int count, const char *name,
2463 const struct file_operations *fops);
2464extern void __unregister_chrdev(unsigned int major, unsigned int baseminor,
2465 unsigned int count, const char *name);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002466extern void unregister_chrdev_region(dev_t, unsigned);
Joe Korty68eef3b2006-03-31 02:30:32 -08002467extern void chrdev_show(struct seq_file *,off_t);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002468
Tejun Heo1905b1b2009-08-06 18:13:23 +09002469static inline int register_chrdev(unsigned int major, const char *name,
2470 const struct file_operations *fops)
2471{
2472 return __register_chrdev(major, 0, 256, name, fops);
2473}
2474
2475static inline void unregister_chrdev(unsigned int major, const char *name)
2476{
2477 __unregister_chrdev(major, 0, 256, name);
2478}
2479
Linus Torvalds1da177e2005-04-16 15:20:36 -07002480/* fs/block_dev.c */
2481#define BDEVNAME_SIZE 32 /* Largest string for a blockdev identifier */
Tejun Heo1f014292008-08-25 19:47:23 +09002482#define BDEVT_SIZE 10 /* Largest string for MAJ:MIN for blkdev */
David Howells93614012006-09-30 20:45:40 +02002483
2484#ifdef CONFIG_BLOCK
2485#define BLKDEV_MAJOR_HASH_SIZE 255
Linus Torvalds1da177e2005-04-16 15:20:36 -07002486extern const char *__bdevname(dev_t, char *buffer);
2487extern const char *bdevname(struct block_device *bdev, char *buffer);
2488extern struct block_device *lookup_bdev(const char *);
Joe Korty68eef3b2006-03-31 02:30:32 -08002489extern void blkdev_show(struct seq_file *,off_t);
Al Viro30c40d22008-02-22 19:50:45 -05002490
David Howells93614012006-09-30 20:45:40 +02002491#else
2492#define BLKDEV_MAJOR_HASH_SIZE 0
2493#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -07002494
2495extern void init_special_inode(struct inode *, umode_t, dev_t);
2496
2497/* Invalid inode operations -- fs/bad_inode.c */
2498extern void make_bad_inode(struct inode *);
Yaowei Bai0e3ef1f2015-11-19 21:00:11 +08002499extern bool is_bad_inode(struct inode *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002500
Jens Axboec11f0c02016-08-05 08:11:04 -06002501#ifdef CONFIG_BLOCK
Mike Christie4e1b2d522016-06-05 14:32:22 -05002502static inline bool op_is_write(unsigned int op)
Mike Christief2150822016-06-05 14:31:42 -05002503{
Mike Christie4e1b2d522016-06-05 14:32:22 -05002504 return op == REQ_OP_READ ? false : true;
Mike Christief2150822016-06-05 14:31:42 -05002505}
2506
2507/*
Linus Torvalds1da177e2005-04-16 15:20:36 -07002508 * return data direction, READ or WRITE
2509 */
Mike Christief2150822016-06-05 14:31:42 -05002510static inline int bio_data_dir(struct bio *bio)
2511{
Mike Christie4e1b2d522016-06-05 14:32:22 -05002512 return op_is_write(bio_op(bio)) ? WRITE : READ;
Mike Christief2150822016-06-05 14:31:42 -05002513}
Linus Torvalds1da177e2005-04-16 15:20:36 -07002514
Andrew Pattersonc3279d12008-09-04 14:27:25 -06002515extern void check_disk_size_change(struct gendisk *disk,
2516 struct block_device *bdev);
Andrew Patterson0c002c22008-09-04 14:27:20 -06002517extern int revalidate_disk(struct gendisk *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002518extern int check_disk_change(struct block_device *);
NeilBrown93b270f2011-02-24 17:25:47 +11002519extern int __invalidate_device(struct block_device *, bool);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002520extern int invalidate_partition(struct gendisk *, int);
David Howells93614012006-09-30 20:45:40 +02002521#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -07002522unsigned long invalidate_mapping_pages(struct address_space *mapping,
2523 pgoff_t start, pgoff_t end);
Anton Altaparmakov54bc4852007-02-10 01:45:38 -08002524
Linus Torvalds1da177e2005-04-16 15:20:36 -07002525static inline void invalidate_remote_inode(struct inode *inode)
2526{
2527 if (S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) ||
2528 S_ISLNK(inode->i_mode))
Andrew Mortonfc0ecff2007-02-10 01:45:39 -08002529 invalidate_mapping_pages(inode->i_mapping, 0, -1);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002530}
2531extern int invalidate_inode_pages2(struct address_space *mapping);
2532extern int invalidate_inode_pages2_range(struct address_space *mapping,
2533 pgoff_t start, pgoff_t end);
2534extern int write_inode_now(struct inode *, int);
2535extern int filemap_fdatawrite(struct address_space *);
2536extern int filemap_flush(struct address_space *);
2537extern int filemap_fdatawait(struct address_space *);
Junichi Nomuraaa750fd2015-11-05 18:47:23 -08002538extern void filemap_fdatawait_keep_errors(struct address_space *);
Jan Karad3bccb6f2009-08-17 19:30:27 +02002539extern int filemap_fdatawait_range(struct address_space *, loff_t lstart,
2540 loff_t lend);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002541extern int filemap_write_and_wait(struct address_space *mapping);
2542extern int filemap_write_and_wait_range(struct address_space *mapping,
2543 loff_t lstart, loff_t lend);
Andrew Mortonebcf28e2006-03-24 03:18:04 -08002544extern int __filemap_fdatawrite_range(struct address_space *mapping,
2545 loff_t start, loff_t end, int sync_mode);
Jan Karaf4c0a0f2008-07-11 19:27:31 -04002546extern int filemap_fdatawrite_range(struct address_space *mapping,
2547 loff_t start, loff_t end);
Miklos Szeredid72d9e22016-07-29 14:10:57 +02002548extern int filemap_check_errors(struct address_space *mapping);
Andrew Mortonebcf28e2006-03-24 03:18:04 -08002549
Christoph Hellwig8018ab02010-03-22 17:32:25 +01002550extern int vfs_fsync_range(struct file *file, loff_t start, loff_t end,
2551 int datasync);
2552extern int vfs_fsync(struct file *file, int datasync);
Christoph Hellwige2592212016-04-07 08:52:01 -07002553
2554/*
2555 * Sync the bytes written if this was a synchronous write. Expect ki_pos
2556 * to already be updated for the write, and will return either the amount
2557 * of bytes passed in, or an error if syncing the file failed.
2558 */
2559static inline ssize_t generic_write_sync(struct kiocb *iocb, ssize_t count)
Al Virod311d792014-02-09 15:18:09 -05002560{
Christoph Hellwige2592212016-04-07 08:52:01 -07002561 if (iocb->ki_flags & IOCB_DSYNC) {
2562 int ret = vfs_fsync_range(iocb->ki_filp,
2563 iocb->ki_pos - count, iocb->ki_pos - 1,
2564 (iocb->ki_flags & IOCB_SYNC) ? 0 : 1);
2565 if (ret)
2566 return ret;
2567 }
2568
2569 return count;
Al Virod311d792014-02-09 15:18:09 -05002570}
Christoph Hellwige2592212016-04-07 08:52:01 -07002571
Linus Torvalds1da177e2005-04-16 15:20:36 -07002572extern void emergency_sync(void);
2573extern void emergency_remount(void);
David Howells93614012006-09-30 20:45:40 +02002574#ifdef CONFIG_BLOCK
Linus Torvalds1da177e2005-04-16 15:20:36 -07002575extern sector_t bmap(struct inode *, sector_t);
David Howells93614012006-09-30 20:45:40 +02002576#endif
J. Bruce Fields27ac0ff2011-09-20 17:19:26 -04002577extern int notify_change(struct dentry *, struct iattr *, struct inode **);
Al Virof419a2e2008-07-22 00:07:17 -04002578extern int inode_permission(struct inode *, int);
Miklos Szeredibd5d0852014-10-24 00:14:35 +02002579extern int __inode_permission(struct inode *, int);
Al Viro2830ba72011-06-20 19:16:29 -04002580extern int generic_permission(struct inode *, int);
Miklos Szeredicbdf35b2014-10-24 00:14:36 +02002581extern int __check_sticky(struct inode *dir, struct inode *inode);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002582
Miklos Szeredif696a362008-07-31 13:41:58 +02002583static inline bool execute_ok(struct inode *inode)
2584{
2585 return (inode->i_mode & S_IXUGO) || S_ISDIR(inode->i_mode);
2586}
2587
Al Viro8d71db42013-03-19 21:01:03 -04002588static inline void file_start_write(struct file *file)
2589{
2590 if (!S_ISREG(file_inode(file)->i_mode))
2591 return;
2592 __sb_start_write(file_inode(file)->i_sb, SB_FREEZE_WRITE, true);
2593}
2594
Jan Kara5ae98f12013-05-04 00:11:23 +02002595static inline bool file_start_write_trylock(struct file *file)
2596{
2597 if (!S_ISREG(file_inode(file)->i_mode))
2598 return true;
2599 return __sb_start_write(file_inode(file)->i_sb, SB_FREEZE_WRITE, false);
2600}
2601
Al Viro8d71db42013-03-19 21:01:03 -04002602static inline void file_end_write(struct file *file)
2603{
2604 if (!S_ISREG(file_inode(file)->i_mode))
2605 return;
2606 __sb_end_write(file_inode(file)->i_sb, SB_FREEZE_WRITE);
2607}
2608
Al Viro07b8ce12011-06-20 10:52:57 -04002609/*
2610 * get_write_access() gets write permission for a file.
2611 * put_write_access() releases this write permission.
2612 * This is used for regular files.
2613 * We cannot support write (and maybe mmap read-write shared) accesses and
2614 * MAP_DENYWRITE mmappings simultaneously. The i_writecount field of an inode
2615 * can have the following values:
2616 * 0: no writers, no VM_DENYWRITE mappings
2617 * < 0: (-i_writecount) vm_area_structs with VM_DENYWRITE set exist
2618 * > 0: (i_writecount) users are writing to the file.
2619 *
2620 * Normally we operate on that counter with atomic_{inc,dec} and it's safe
2621 * except for the cases where we don't hold i_writecount yet. Then we need to
2622 * use {get,deny}_write_access() - these functions check the sign and refuse
2623 * to do the change if sign is wrong.
2624 */
2625static inline int get_write_access(struct inode *inode)
2626{
2627 return atomic_inc_unless_negative(&inode->i_writecount) ? 0 : -ETXTBSY;
2628}
2629static inline int deny_write_access(struct file *file)
2630{
Al Viro496ad9a2013-01-23 17:07:38 -05002631 struct inode *inode = file_inode(file);
Al Viro07b8ce12011-06-20 10:52:57 -04002632 return atomic_dec_unless_positive(&inode->i_writecount) ? 0 : -ETXTBSY;
2633}
Linus Torvalds1da177e2005-04-16 15:20:36 -07002634static inline void put_write_access(struct inode * inode)
2635{
2636 atomic_dec(&inode->i_writecount);
2637}
2638static inline void allow_write_access(struct file *file)
2639{
2640 if (file)
Al Viro496ad9a2013-01-23 17:07:38 -05002641 atomic_inc(&file_inode(file)->i_writecount);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002642}
David Howellsf1fe29b2013-09-27 11:20:03 +01002643static inline bool inode_is_open_for_write(const struct inode *inode)
2644{
2645 return atomic_read(&inode->i_writecount) > 0;
2646}
2647
Mimi Zohara5c96eb2010-11-02 10:11:37 -04002648#ifdef CONFIG_IMA
2649static inline void i_readcount_dec(struct inode *inode)
2650{
2651 BUG_ON(!atomic_read(&inode->i_readcount));
2652 atomic_dec(&inode->i_readcount);
2653}
2654static inline void i_readcount_inc(struct inode *inode)
2655{
2656 atomic_inc(&inode->i_readcount);
2657}
2658#else
2659static inline void i_readcount_dec(struct inode *inode)
2660{
2661 return;
2662}
2663static inline void i_readcount_inc(struct inode *inode)
2664{
2665 return;
2666}
2667#endif
Ulrich Dreppered8cae82008-07-23 21:29:30 -07002668extern int do_pipe_flags(int *, int);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002669
Mimi Zohar1284ab52016-04-20 15:46:27 -07002670#define __kernel_read_file_id(id) \
2671 id(UNKNOWN, unknown) \
2672 id(FIRMWARE, firmware) \
Stephen Boyda098ecd2016-08-02 14:04:28 -07002673 id(FIRMWARE_PREALLOC_BUFFER, firmware) \
Mimi Zohar1284ab52016-04-20 15:46:27 -07002674 id(MODULE, kernel-module) \
2675 id(KEXEC_IMAGE, kexec-image) \
2676 id(KEXEC_INITRAMFS, kexec-initramfs) \
2677 id(POLICY, security-policy) \
2678 id(MAX_ID, )
2679
2680#define __fid_enumify(ENUM, dummy) READING_ ## ENUM,
2681#define __fid_stringify(dummy, str) #str,
2682
Mimi Zoharbc8ca5b2016-01-24 10:07:32 -05002683enum kernel_read_file_id {
Mimi Zohar1284ab52016-04-20 15:46:27 -07002684 __kernel_read_file_id(__fid_enumify)
Mimi Zoharbc8ca5b2016-01-24 10:07:32 -05002685};
2686
Mimi Zohar1284ab52016-04-20 15:46:27 -07002687static const char * const kernel_read_file_str[] = {
2688 __kernel_read_file_id(__fid_stringify)
2689};
2690
Kees Cookda20dfe2016-04-21 12:53:29 -07002691static inline const char *kernel_read_file_id_str(enum kernel_read_file_id id)
Mimi Zohar1284ab52016-04-20 15:46:27 -07002692{
2693 if (id < 0 || id >= READING_MAX_ID)
2694 return kernel_read_file_str[READING_UNKNOWN];
2695
2696 return kernel_read_file_str[id];
2697}
2698
Mimi Zohar6777d772009-08-21 14:32:48 -04002699extern int kernel_read(struct file *, loff_t, char *, unsigned long);
Mimi Zoharbc8ca5b2016-01-24 10:07:32 -05002700extern int kernel_read_file(struct file *, void **, loff_t *, loff_t,
2701 enum kernel_read_file_id);
Mimi Zohar09596b92015-11-19 12:39:22 -05002702extern int kernel_read_file_from_path(char *, void **, loff_t *, loff_t,
2703 enum kernel_read_file_id);
Mimi Zoharb844f0e2016-02-01 08:36:21 -05002704extern int kernel_read_file_from_fd(int, void **, loff_t *, loff_t,
2705 enum kernel_read_file_id);
Al Viro7bb307e2013-02-23 14:51:48 -05002706extern ssize_t kernel_write(struct file *, const char *, size_t, loff_t);
Al Viroed447242014-04-19 14:37:20 -04002707extern ssize_t __kernel_write(struct file *, const char *, size_t, loff_t *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002708extern struct file * open_exec(const char *);
2709
2710/* fs/dcache.c -- generic fs support functions */
Yaowei Baia6e57872015-11-17 14:40:11 +08002711extern bool is_subdir(struct dentry *, struct dentry *);
Yaowei Bai25ab4c92015-11-17 14:40:10 +08002712extern bool path_is_under(struct path *, struct path *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002713
Miklos Szeredi9bf39ab2015-06-19 10:29:13 +02002714extern char *file_path(struct file *, char *, int);
2715
Linus Torvalds1da177e2005-04-16 15:20:36 -07002716#include <linux/err.h>
2717
2718/* needed for stackable file system support */
Andrew Morton965c8e52012-12-17 15:59:39 -08002719extern loff_t default_llseek(struct file *file, loff_t offset, int whence);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002720
Andrew Morton965c8e52012-12-17 15:59:39 -08002721extern loff_t vfs_llseek(struct file *file, loff_t offset, int whence);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002722
Christoph Hellwig54e34622009-08-07 14:38:25 -03002723extern int inode_init_always(struct super_block *, struct inode *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002724extern void inode_init_once(struct inode *);
Miklos Szeredi2aa15892011-02-23 13:49:47 +01002725extern void address_space_init_once(struct address_space *mapping);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002726extern struct inode * igrab(struct inode *);
2727extern ino_t iunique(struct super_block *, ino_t);
2728extern int inode_needs_sync(struct inode *inode);
Al Viro45321ac2010-06-07 13:43:19 -04002729extern int generic_delete_inode(struct inode *inode);
Al Viro9bcb4b72012-02-12 19:43:17 -05002730static inline int generic_drop_inode(struct inode *inode)
2731{
2732 return !inode->i_nlink || inode_unhashed(inode);
2733}
Linus Torvalds1da177e2005-04-16 15:20:36 -07002734
Anton Altaparmakov88bd5122005-07-13 01:10:44 -07002735extern struct inode *ilookup5_nowait(struct super_block *sb,
2736 unsigned long hashval, int (*test)(struct inode *, void *),
2737 void *data);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002738extern struct inode *ilookup5(struct super_block *sb, unsigned long hashval,
2739 int (*test)(struct inode *, void *), void *data);
2740extern struct inode *ilookup(struct super_block *sb, unsigned long ino);
2741
2742extern struct inode * iget5_locked(struct super_block *, unsigned long, int (*test)(struct inode *, void *), int (*set)(struct inode *, void *), void *);
2743extern struct inode * iget_locked(struct super_block *, unsigned long);
Theodore Ts'ofe032c42015-02-02 00:37:01 -05002744extern struct inode *find_inode_nowait(struct super_block *,
2745 unsigned long,
2746 int (*match)(struct inode *,
2747 unsigned long, void *),
2748 void *data);
Al Viro261bca82008-12-30 01:48:21 -05002749extern int insert_inode_locked4(struct inode *, unsigned long, int (*test)(struct inode *, void *), void *);
2750extern int insert_inode_locked(struct inode *);
Josh Boyere096d0c2011-08-25 07:48:12 -04002751#ifdef CONFIG_DEBUG_LOCK_ALLOC
2752extern void lockdep_annotate_inode_mutex_key(struct inode *inode);
2753#else
2754static inline void lockdep_annotate_inode_mutex_key(struct inode *inode) { };
2755#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -07002756extern void unlock_new_inode(struct inode *);
Christoph Hellwig85fe4022010-10-23 11:19:54 -04002757extern unsigned int get_next_ino(void);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002758
Linus Torvalds1da177e2005-04-16 15:20:36 -07002759extern void __iget(struct inode * inode);
David Howellsb46980f2008-02-07 00:15:27 -08002760extern void iget_failed(struct inode *);
Jan Karadbd57682012-05-03 14:48:02 +02002761extern void clear_inode(struct inode *);
Christoph Hellwig2e00c972009-08-07 14:38:29 -03002762extern void __destroy_inode(struct inode *);
Eric Dumazeta209dfc2011-07-26 11:36:34 +02002763extern struct inode *new_inode_pseudo(struct super_block *sb);
2764extern struct inode *new_inode(struct super_block *sb);
Nick Pigginff0c7d12011-01-07 17:49:50 +11002765extern void free_inode_nonrcu(struct inode *inode);
Jens Axboe01de85e2006-10-17 19:50:36 +02002766extern int should_remove_suid(struct dentry *);
Jan Kara5fa8e0a2015-05-21 16:05:53 +02002767extern int file_remove_privs(struct file *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002768
2769extern void __insert_inode_hash(struct inode *, unsigned long hashval);
Christoph Hellwig646ec462010-10-23 07:15:32 -04002770static inline void insert_inode_hash(struct inode *inode)
2771{
Linus Torvalds1da177e2005-04-16 15:20:36 -07002772 __insert_inode_hash(inode, inode->i_ino);
2773}
Eric Dumazetf2ee7ab2011-07-28 06:41:09 +02002774
2775extern void __remove_inode_hash(struct inode *);
2776static inline void remove_inode_hash(struct inode *inode)
2777{
Josef Bacikcbedaac2015-03-12 08:19:11 -04002778 if (!inode_unhashed(inode) && !hlist_fake(&inode->i_hash))
Eric Dumazetf2ee7ab2011-07-28 06:41:09 +02002779 __remove_inode_hash(inode);
2780}
2781
Christoph Hellwig646ec462010-10-23 07:15:32 -04002782extern void inode_sb_list_add(struct inode *inode);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002783
David Howells93614012006-09-30 20:45:40 +02002784#ifdef CONFIG_BLOCK
Mike Christie4e49ea42016-06-05 14:31:41 -05002785extern blk_qc_t submit_bio(struct bio *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002786extern int bdev_read_only(struct block_device *);
David Howells93614012006-09-30 20:45:40 +02002787#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -07002788extern int set_blocksize(struct block_device *, int);
2789extern int sb_set_blocksize(struct super_block *, int);
2790extern int sb_min_blocksize(struct super_block *, int);
2791
2792extern int generic_file_mmap(struct file *, struct vm_area_struct *);
2793extern int generic_file_readonly_mmap(struct file *, struct vm_area_struct *);
Al Viro3309dd02015-04-09 12:55:47 -04002794extern ssize_t generic_write_checks(struct kiocb *, struct iov_iter *);
Al Viroed978a82014-03-05 22:53:04 -05002795extern ssize_t generic_file_read_iter(struct kiocb *, struct iov_iter *);
Al Viro81742022014-04-03 03:17:43 -04002796extern ssize_t __generic_file_write_iter(struct kiocb *, struct iov_iter *);
Al Viro81742022014-04-03 03:17:43 -04002797extern ssize_t generic_file_write_iter(struct kiocb *, struct iov_iter *);
Christoph Hellwig1af5bb42016-04-07 08:51:56 -07002798extern ssize_t generic_file_direct_write(struct kiocb *, struct iov_iter *);
Al Viro3b93f912014-02-11 21:34:08 -05002799extern ssize_t generic_perform_write(struct file *, struct iov_iter *, loff_t);
Andrew Morton88e6fae2006-04-11 13:59:36 +02002800
Christoph Hellwigdbe4e192015-01-25 21:11:59 +01002801ssize_t vfs_iter_read(struct file *file, struct iov_iter *iter, loff_t *ppos);
2802ssize_t vfs_iter_write(struct file *file, struct iov_iter *iter, loff_t *ppos);
2803
Christoph Hellwigeef99382009-08-20 17:43:41 +02002804/* fs/block_dev.c */
David Jefferyb2de5252014-09-29 10:21:10 -04002805extern ssize_t blkdev_read_iter(struct kiocb *iocb, struct iov_iter *to);
Al Viro1456c0a2014-04-03 03:21:50 -04002806extern ssize_t blkdev_write_iter(struct kiocb *iocb, struct iov_iter *from);
Josef Bacik02c24a82011-07-16 20:44:56 -04002807extern int blkdev_fsync(struct file *filp, loff_t start, loff_t end,
2808 int datasync);
Al Viroff01bb42011-09-16 02:31:11 -04002809extern void block_sync_page(struct page *page);
Christoph Hellwigeef99382009-08-20 17:43:41 +02002810
Andrew Morton88e6fae2006-04-11 13:59:36 +02002811/* fs/splice.c */
Jens Axboecbb7e572006-04-11 14:57:50 +02002812extern ssize_t generic_file_splice_read(struct file *, loff_t *,
Andrew Morton88e6fae2006-04-11 13:59:36 +02002813 struct pipe_inode_info *, size_t, unsigned int);
Al Viro8d020762014-04-05 04:27:08 -04002814extern ssize_t iter_file_splice_write(struct pipe_inode_info *,
Jens Axboecbb7e572006-04-11 14:57:50 +02002815 struct file *, loff_t *, size_t, unsigned int);
Andrew Morton88e6fae2006-04-11 13:59:36 +02002816extern ssize_t generic_splice_sendpage(struct pipe_inode_info *pipe,
Jens Axboecbb7e572006-04-11 14:57:50 +02002817 struct file *out, loff_t *, size_t len, unsigned int flags);
Miklos Szeredi1c118592014-10-24 00:14:35 +02002818extern long do_splice_direct(struct file *in, loff_t *ppos, struct file *out,
2819 loff_t *opos, size_t len, unsigned int flags);
2820
Andrew Morton88e6fae2006-04-11 13:59:36 +02002821
Linus Torvalds1da177e2005-04-16 15:20:36 -07002822extern void
2823file_ra_state_init(struct file_ra_state *ra, struct address_space *mapping);
Andrew Morton965c8e52012-12-17 15:59:39 -08002824extern loff_t noop_llseek(struct file *file, loff_t offset, int whence);
2825extern loff_t no_llseek(struct file *file, loff_t offset, int whence);
Jie Liu46a1c2c2013-06-25 12:02:13 +08002826extern loff_t vfs_setpos(struct file *file, loff_t offset, loff_t maxsize);
Andrew Morton965c8e52012-12-17 15:59:39 -08002827extern loff_t generic_file_llseek(struct file *file, loff_t offset, int whence);
Andi Kleen57604952011-09-15 16:06:50 -07002828extern loff_t generic_file_llseek_size(struct file *file, loff_t offset,
Andrew Morton965c8e52012-12-17 15:59:39 -08002829 int whence, loff_t maxsize, loff_t eof);
Al Viro1bf9d142013-06-16 20:27:42 +04002830extern loff_t fixed_size_llseek(struct file *file, loff_t offset,
2831 int whence, loff_t size);
Al Virob25472f2015-12-05 22:04:48 -05002832extern loff_t no_seek_end_llseek_size(struct file *, loff_t, int, loff_t);
2833extern loff_t no_seek_end_llseek(struct file *, loff_t, int);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002834extern int generic_file_open(struct inode * inode, struct file * filp);
2835extern int nonseekable_open(struct inode * inode, struct file * filp);
2836
David Howells93614012006-09-30 20:45:40 +02002837#ifdef CONFIG_BLOCK
Mike Christie8a4c1e42016-06-05 14:31:50 -05002838typedef void (dio_submit_t)(struct bio *bio, struct inode *inode,
Josef Bacikfacd07b2010-05-23 11:00:55 -04002839 loff_t file_offset);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002840
2841enum {
Christoph Hellwig1e431f52009-11-03 16:44:53 +01002842 /* need locking between buffered and direct access */
2843 DIO_LOCKING = 0x01,
2844
2845 /* filesystem does not support filling holes */
2846 DIO_SKIP_HOLES = 0x02,
Christoph Hellwig60392572014-02-10 10:27:11 +11002847
2848 /* filesystem can handle aio writes beyond i_size */
2849 DIO_ASYNC_EXTEND = 0x04,
Jens Axboefe0f07d2015-04-15 17:05:48 -06002850
2851 /* inode/fs/bdev does not need truncate protection */
2852 DIO_SKIP_DIO_COUNT = 0x08,
Linus Torvalds1da177e2005-04-16 15:20:36 -07002853};
2854
Christoph Hellwigeafdc7d2010-06-04 11:29:53 +02002855void dio_end_io(struct bio *bio, int error);
npiggin@suse.de7bb46a62010-05-27 01:05:33 +10002856
Omar Sandoval17f8c842015-03-16 04:33:50 -07002857ssize_t __blockdev_direct_IO(struct kiocb *iocb, struct inode *inode,
2858 struct block_device *bdev, struct iov_iter *iter,
Christoph Hellwigc8b8e322016-04-07 08:51:58 -07002859 get_block_t get_block,
Omar Sandoval17f8c842015-03-16 04:33:50 -07002860 dio_iodone_t end_io, dio_submit_t submit_io,
2861 int flags);
Christoph Hellwigeafdc7d2010-06-04 11:29:53 +02002862
Omar Sandoval17f8c842015-03-16 04:33:50 -07002863static inline ssize_t blockdev_direct_IO(struct kiocb *iocb,
2864 struct inode *inode,
Christoph Hellwigc8b8e322016-04-07 08:51:58 -07002865 struct iov_iter *iter,
Omar Sandoval17f8c842015-03-16 04:33:50 -07002866 get_block_t get_block)
Linus Torvalds1da177e2005-04-16 15:20:36 -07002867{
Omar Sandoval17f8c842015-03-16 04:33:50 -07002868 return __blockdev_direct_IO(iocb, inode, inode->i_sb->s_bdev, iter,
Christoph Hellwigc8b8e322016-04-07 08:51:58 -07002869 get_block, NULL, NULL, DIO_LOCKING | DIO_SKIP_HOLES);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002870}
David Howells93614012006-09-30 20:45:40 +02002871#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -07002872
Trond Myklebust1d59d612012-05-31 12:22:33 -04002873void inode_dio_wait(struct inode *inode);
Jens Axboefe0f07d2015-04-15 17:05:48 -06002874
2875/*
2876 * inode_dio_begin - signal start of a direct I/O requests
2877 * @inode: inode the direct I/O happens on
2878 *
2879 * This is called once we've finished processing a direct I/O request,
2880 * and is used to wake up callers waiting for direct I/O to be quiesced.
2881 */
2882static inline void inode_dio_begin(struct inode *inode)
2883{
2884 atomic_inc(&inode->i_dio_count);
2885}
2886
2887/*
2888 * inode_dio_end - signal finish of a direct I/O requests
2889 * @inode: inode the direct I/O happens on
2890 *
2891 * This is called once we've finished processing a direct I/O request,
2892 * and is used to wake up callers waiting for direct I/O to be quiesced.
2893 */
2894static inline void inode_dio_end(struct inode *inode)
2895{
2896 if (atomic_dec_and_test(&inode->i_dio_count))
2897 wake_up_bit(&inode->i_state, __I_DIO_WAKEUP);
2898}
Trond Myklebust1d59d612012-05-31 12:22:33 -04002899
Theodore Ts'o5f16f322014-03-24 14:43:12 -04002900extern void inode_set_flags(struct inode *inode, unsigned int flags,
2901 unsigned int mask);
2902
Arjan van de Ven4b6f5d22006-03-28 01:56:42 -08002903extern const struct file_operations generic_ro_fops;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002904
2905#define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
2906
Al Viro5d826c82014-03-14 13:42:45 -04002907extern int readlink_copy(char __user *, int, const char *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002908extern int page_readlink(struct dentry *, char __user *, int);
Al Virofceef392015-12-29 15:58:39 -05002909extern const char *page_get_link(struct dentry *, struct inode *,
2910 struct delayed_call *);
2911extern void page_put_link(void *);
Kirill Korotaev0adb25d2006-03-11 03:27:13 -08002912extern int __page_symlink(struct inode *inode, const char *symname, int len,
Nick Piggin54566b22009-01-04 12:00:53 -08002913 int nofs);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002914extern int page_symlink(struct inode *inode, const char *symname, int len);
Arjan van de Venc5ef1c42007-02-12 00:55:40 -08002915extern const struct inode_operations page_symlink_inode_operations;
Al Virofceef392015-12-29 15:58:39 -05002916extern void kfree_link(void *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002917extern int generic_readlink(struct dentry *, char __user *, int);
2918extern void generic_fillattr(struct inode *, struct kstat *);
J. Bruce Fieldsb7a6ec52013-10-02 17:01:18 -04002919int vfs_getattr_nosec(struct path *path, struct kstat *stat);
Al Viro3dadecc2013-01-24 02:18:08 -05002920extern int vfs_getattr(struct path *, struct kstat *);
Dmitry Monakhovb4627072009-12-14 15:21:12 +03002921void __inode_add_bytes(struct inode *inode, loff_t bytes);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002922void inode_add_bytes(struct inode *inode, loff_t bytes);
Jan Kara1c8924e2013-08-17 09:32:32 -04002923void __inode_sub_bytes(struct inode *inode, loff_t bytes);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002924void inode_sub_bytes(struct inode *inode, loff_t bytes);
2925loff_t inode_get_bytes(struct inode *inode);
2926void inode_set_bytes(struct inode *inode, loff_t bytes);
Al Virofceef392015-12-29 15:58:39 -05002927const char *simple_get_link(struct dentry *, struct inode *,
2928 struct delayed_call *);
Al Viro61ba64f2015-05-02 09:54:06 -04002929extern const struct inode_operations simple_symlink_inode_operations;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002930
Al Viro5c0ba4e2013-05-15 13:52:59 -04002931extern int iterate_dir(struct file *, struct dir_context *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002932
David Howellsc7887322010-08-11 11:26:22 +01002933extern int vfs_stat(const char __user *, struct kstat *);
2934extern int vfs_lstat(const char __user *, struct kstat *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002935extern int vfs_fstat(unsigned int, struct kstat *);
David Howellsc7887322010-08-11 11:26:22 +01002936extern int vfs_fstatat(int , const char __user *, struct kstat *, int);
Miklos Szeredid60874c2016-10-04 14:40:45 +02002937extern const char *vfs_get_link(struct dentry *, struct delayed_call *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002938
Steven Whitehousee9079cc2008-10-14 14:43:29 +01002939extern int __generic_block_fiemap(struct inode *inode,
Josef Bacik3a3076f2010-04-23 12:17:17 -04002940 struct fiemap_extent_info *fieinfo,
2941 loff_t start, loff_t len,
2942 get_block_t *get_block);
Josef Bacik68c9d702008-10-03 17:32:43 -04002943extern int generic_block_fiemap(struct inode *inode,
2944 struct fiemap_extent_info *fieinfo, u64 start,
2945 u64 len, get_block_t *get_block);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002946
Miklos Szeredic18479f2007-10-18 23:39:11 -07002947extern void get_filesystem(struct file_system_type *fs);
2948extern void put_filesystem(struct file_system_type *fs);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002949extern struct file_system_type *get_fs_type(const char *name);
2950extern struct super_block *get_super(struct block_device *);
Jan Kara6b6dc832012-02-10 11:03:00 +01002951extern struct super_block *get_super_thawed(struct block_device *);
Christoph Hellwig45042302009-08-03 23:28:35 +02002952extern struct super_block *get_active_super(struct block_device *bdev);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002953extern void drop_super(struct super_block *sb);
Al Viro01a05b32010-03-23 06:06:58 -04002954extern void iterate_supers(void (*)(struct super_block *, void *), void *);
Al Viro43e15cd2011-06-03 20:16:57 -04002955extern void iterate_supers_type(struct file_system_type *,
2956 void (*)(struct super_block *, void *), void *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002957
2958extern int dcache_dir_open(struct inode *, struct file *);
2959extern int dcache_dir_close(struct inode *, struct file *);
2960extern loff_t dcache_dir_lseek(struct file *, loff_t, int);
Al Viro5f99f4e2013-05-15 20:23:06 -04002961extern int dcache_readdir(struct file *, struct dir_context *);
npiggin@suse.de7bb46a62010-05-27 01:05:33 +10002962extern int simple_setattr(struct dentry *, struct iattr *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002963extern int simple_getattr(struct vfsmount *, struct dentry *, struct kstat *);
David Howells726c3342006-06-23 02:02:58 -07002964extern int simple_statfs(struct dentry *, struct kstatfs *);
Stephen Boyd20955e82012-04-05 14:25:09 -07002965extern int simple_open(struct inode *inode, struct file *file);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002966extern int simple_link(struct dentry *, struct inode *, struct dentry *);
2967extern int simple_unlink(struct inode *, struct dentry *);
2968extern int simple_rmdir(struct inode *, struct dentry *);
Miklos Szeredie0e0be82016-09-27 11:03:57 +02002969extern int simple_rename(struct inode *, struct dentry *,
2970 struct inode *, struct dentry *, unsigned int);
Josef Bacik02c24a82011-07-16 20:44:56 -04002971extern int noop_fsync(struct file *, loff_t, loff_t, int);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002972extern int simple_empty(struct dentry *);
2973extern int simple_readpage(struct file *file, struct page *page);
Nick Pigginafddba42007-10-16 01:25:01 -07002974extern int simple_write_begin(struct file *file, struct address_space *mapping,
2975 loff_t pos, unsigned len, unsigned flags,
2976 struct page **pagep, void **fsdata);
2977extern int simple_write_end(struct file *file, struct address_space *mapping,
2978 loff_t pos, unsigned len, unsigned copied,
2979 struct page *page, void *fsdata);
Al Virob26d4cd2013-10-25 18:47:37 -04002980extern int always_delete_dentry(const struct dentry *);
Al Viro69878432013-10-02 22:35:11 -04002981extern struct inode *alloc_anon_inode(struct super_block *);
Jeff Laytone6f5c782014-08-22 10:40:25 -04002982extern int simple_nosetlease(struct file *, long, struct file_lock **, void **);
Al Virob26d4cd2013-10-25 18:47:37 -04002983extern const struct dentry_operations simple_dentry_operations;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002984
Al Viro00cd8dd2012-06-10 17:13:09 -04002985extern struct dentry *simple_lookup(struct inode *, struct dentry *, unsigned int flags);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002986extern ssize_t generic_read_dir(struct file *, char __user *, size_t, loff_t *);
Arjan van de Ven4b6f5d22006-03-28 01:56:42 -08002987extern const struct file_operations simple_dir_operations;
Arjan van de Venc5ef1c42007-02-12 00:55:40 -08002988extern const struct inode_operations simple_dir_inode_operations;
Eric W. Biedermanfbabfd02015-05-09 15:54:49 -05002989extern void make_empty_dir_inode(struct inode *inode);
2990extern bool is_empty_dir_inode(struct inode *inode);
Arjan van de Ven4b6f5d22006-03-28 01:56:42 -08002991struct tree_descr { char *name; const struct file_operations *ops; int mode; };
Linus Torvalds1da177e2005-04-16 15:20:36 -07002992struct dentry *d_alloc_name(struct dentry *, const char *);
Roberto Sassu7d683a02010-06-03 11:58:28 +02002993extern int simple_fill_super(struct super_block *, unsigned long, struct tree_descr *);
Trond Myklebust1f5ce9e2006-06-09 09:34:16 -04002994extern int simple_pin_fs(struct file_system_type *, struct vfsmount **mount, int *count);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002995extern void simple_release_fs(struct vfsmount **mount, int *count);
2996
Akinobu Mita93b07112008-06-05 22:46:21 -07002997extern ssize_t simple_read_from_buffer(void __user *to, size_t count,
2998 loff_t *ppos, const void *from, size_t available);
Jiri Slaby6a727b42010-05-01 23:51:22 +02002999extern ssize_t simple_write_to_buffer(void *to, size_t available, loff_t *ppos,
3000 const void __user *from, size_t count);
Linus Torvalds1da177e2005-04-16 15:20:36 -07003001
Fabian Frederickac13a822014-06-04 16:06:27 -07003002extern int __generic_file_fsync(struct file *, loff_t, loff_t, int);
Josef Bacik02c24a82011-07-16 20:44:56 -04003003extern int generic_file_fsync(struct file *, loff_t, loff_t, int);
Al Virod5aacad2009-06-07 14:56:44 -04003004
Patrick J. LoPresti30ca22c2010-07-22 15:03:41 -07003005extern int generic_check_addressable(unsigned, u64);
3006
Christoph Lametere965f962006-02-01 03:05:41 -08003007#ifdef CONFIG_MIGRATION
Christoph Lameter2d1db3b2006-06-23 02:03:33 -07003008extern int buffer_migrate_page(struct address_space *,
Mel Gormana6bc32b2012-01-12 17:19:43 -08003009 struct page *, struct page *,
3010 enum migrate_mode);
Christoph Lametere965f962006-02-01 03:05:41 -08003011#else
3012#define buffer_migrate_page NULL
3013#endif
3014
Jan Kara31051c82016-05-26 16:55:18 +02003015extern int setattr_prepare(struct dentry *, struct iattr *);
npiggin@suse.de25d9e2d2009-08-21 02:35:05 +10003016extern int inode_newsize_ok(const struct inode *, loff_t offset);
Christoph Hellwig6a1a90a2010-06-04 11:30:00 +02003017extern void setattr_copy(struct inode *inode, const struct iattr *attr);
Linus Torvalds1da177e2005-04-16 15:20:36 -07003018
Josef Bacikc3b2da32012-03-26 09:59:21 -04003019extern int file_update_time(struct file *file);
Linus Torvalds1da177e2005-04-16 15:20:36 -07003020
Al Viro34c80b12011-12-08 21:32:45 -05003021extern int generic_show_options(struct seq_file *m, struct dentry *root);
Miklos Szeredib3b304a2008-02-08 04:21:35 -08003022extern void save_mount_options(struct super_block *sb, char *options);
Al Viro2a32ceb2009-05-08 16:05:57 -04003023extern void replace_mount_options(struct super_block *sb, char *options);
Miklos Szeredib3b304a2008-02-08 04:21:35 -08003024
Matthew Wilcoxd475c632015-02-16 15:58:56 -08003025static inline bool io_is_direct(struct file *filp)
3026{
Dan Williams65f87ee2016-01-25 17:23:18 -08003027 return (filp->f_flags & O_DIRECT) || IS_DAX(filp->f_mapping->host);
Matthew Wilcoxd475c632015-02-16 15:58:56 -08003028}
3029
Al Viro2ba48ce2015-04-09 13:52:01 -04003030static inline int iocb_flags(struct file *file)
3031{
3032 int res = 0;
3033 if (file->f_flags & O_APPEND)
3034 res |= IOCB_APPEND;
3035 if (io_is_direct(file))
3036 res |= IOCB_DIRECT;
Christoph Hellwigdde0c2e2016-04-07 08:52:00 -07003037 if ((file->f_flags & O_DSYNC) || IS_SYNC(file->f_mapping->host))
3038 res |= IOCB_DSYNC;
3039 if (file->f_flags & __O_SYNC)
3040 res |= IOCB_SYNC;
Al Viro2ba48ce2015-04-09 13:52:01 -04003041 return res;
3042}
3043
Linus Torvalds1da177e2005-04-16 15:20:36 -07003044static inline ino_t parent_ino(struct dentry *dentry)
3045{
3046 ino_t res;
3047
Nick Pigginb5c84bf2011-01-07 17:49:38 +11003048 /*
3049 * Don't strictly need d_lock here? If the parent ino could change
3050 * then surely we'd have a deeper race in the caller?
3051 */
Linus Torvalds1da177e2005-04-16 15:20:36 -07003052 spin_lock(&dentry->d_lock);
3053 res = dentry->d_parent->d_inode->i_ino;
3054 spin_unlock(&dentry->d_lock);
3055 return res;
3056}
3057
Linus Torvalds1da177e2005-04-16 15:20:36 -07003058/* Transaction based IO helpers */
3059
3060/*
3061 * An argresp is stored in an allocated page and holds the
3062 * size of the argument or response, along with its content
3063 */
3064struct simple_transaction_argresp {
3065 ssize_t size;
3066 char data[0];
3067};
3068
3069#define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
3070
3071char *simple_transaction_get(struct file *file, const char __user *buf,
3072 size_t size);
3073ssize_t simple_transaction_read(struct file *file, char __user *buf,
3074 size_t size, loff_t *pos);
3075int simple_transaction_release(struct inode *inode, struct file *file);
3076
Ingo Molnar76791ab2009-03-25 16:48:35 +01003077void simple_transaction_set(struct file *file, size_t n);
Linus Torvalds1da177e2005-04-16 15:20:36 -07003078
Arnd Bergmannacaefc22005-05-18 14:40:59 +02003079/*
3080 * simple attribute files
3081 *
3082 * These attributes behave similar to those in sysfs:
3083 *
3084 * Writing to an attribute immediately sets a value, an open file can be
3085 * written to multiple times.
3086 *
3087 * Reading from an attribute creates a buffer from the value that might get
3088 * read with multiple read calls. When the attribute has been read
3089 * completely, no further read calls are possible until the file is opened
3090 * again.
3091 *
3092 * All attributes contain a text representation of a numeric value
3093 * that are accessed with the get() and set() functions.
3094 */
3095#define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt) \
3096static int __fops ## _open(struct inode *inode, struct file *file) \
3097{ \
3098 __simple_attr_check_format(__fmt, 0ull); \
3099 return simple_attr_open(inode, file, __get, __set, __fmt); \
3100} \
Alexey Dobriyan828c0952009-10-01 15:43:56 -07003101static const struct file_operations __fops = { \
Arnd Bergmannacaefc22005-05-18 14:40:59 +02003102 .owner = THIS_MODULE, \
3103 .open = __fops ## _open, \
Christoph Hellwig74bedc42008-02-08 04:20:28 -08003104 .release = simple_attr_release, \
Arnd Bergmannacaefc22005-05-18 14:40:59 +02003105 .read = simple_attr_read, \
3106 .write = simple_attr_write, \
Arnd Bergmann1ec55842010-08-15 21:50:52 +02003107 .llseek = generic_file_llseek, \
Joe Perches68be3022014-08-06 16:08:45 -07003108}
Arnd Bergmannacaefc22005-05-18 14:40:59 +02003109
Joe Perchesb9075fa2011-10-31 17:11:33 -07003110static inline __printf(1, 2)
3111void __simple_attr_check_format(const char *fmt, ...)
Arnd Bergmannacaefc22005-05-18 14:40:59 +02003112{
3113 /* don't do anything, just let the compiler check the arguments; */
3114}
3115
3116int simple_attr_open(struct inode *inode, struct file *file,
Christoph Hellwig8b88b092008-02-08 04:20:26 -08003117 int (*get)(void *, u64 *), int (*set)(void *, u64),
Arnd Bergmannacaefc22005-05-18 14:40:59 +02003118 const char *fmt);
Christoph Hellwig74bedc42008-02-08 04:20:28 -08003119int simple_attr_release(struct inode *inode, struct file *file);
Arnd Bergmannacaefc22005-05-18 14:40:59 +02003120ssize_t simple_attr_read(struct file *file, char __user *buf,
3121 size_t len, loff_t *ppos);
3122ssize_t simple_attr_write(struct file *file, const char __user *buf,
3123 size_t len, loff_t *ppos);
3124
Alexey Dobriyan4be28542007-10-16 23:26:21 -07003125struct ctl_table;
Alexey Dobriyan8d65af72009-09-23 15:57:19 -07003126int proc_nr_files(struct ctl_table *table, int write,
Adrian Bunk62239ac2007-07-17 04:03:45 -07003127 void __user *buffer, size_t *lenp, loff_t *ppos);
Christoph Hellwig312d3ca2010-10-10 05:36:23 -04003128int proc_nr_dentry(struct ctl_table *table, int write,
3129 void __user *buffer, size_t *lenp, loff_t *ppos);
Dave Chinnercffbc8a2010-10-23 05:03:02 -04003130int proc_nr_inodes(struct ctl_table *table, int write,
3131 void __user *buffer, size_t *lenp, loff_t *ppos);
Tetsuo Handa38e23c92009-04-09 20:17:52 +09003132int __init get_filesystem_list(char *buf);
Adrian Bunk62239ac2007-07-17 04:03:45 -07003133
Namhyung Kim3cd90ea2011-02-01 15:52:46 -08003134#define __FMODE_EXEC ((__force int) FMODE_EXEC)
Namhyung Kim1a44bc82011-02-01 15:52:46 -08003135#define __FMODE_NONOTIFY ((__force int) FMODE_NONOTIFY)
3136
Al Viro6d125522009-12-24 06:58:56 -05003137#define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE])
Eric Parisecf081d2009-12-17 21:24:25 -05003138#define OPEN_FMODE(flag) ((__force fmode_t)(((flag + 1) & O_ACCMODE) | \
Namhyung Kim1a44bc82011-02-01 15:52:46 -08003139 (flag & __FMODE_NONOTIFY)))
Al Viro53009902009-12-19 10:15:07 -05003140
Yaowei Baid37177b2015-11-19 21:00:12 +08003141static inline bool is_sxid(umode_t mode)
Andi Kleen69b45732011-05-28 08:25:51 -07003142{
3143 return (mode & S_ISUID) || ((mode & S_ISGID) && (mode & S_IXGRP));
3144}
3145
Miklos Szeredicbdf35b2014-10-24 00:14:36 +02003146static inline int check_sticky(struct inode *dir, struct inode *inode)
3147{
3148 if (!(dir->i_mode & S_ISVTX))
3149 return 0;
3150
3151 return __check_sticky(dir, inode);
3152}
3153
Andi Kleen69b45732011-05-28 08:25:51 -07003154static inline void inode_has_no_xattr(struct inode *inode)
3155{
Al Viro9e1f1de2011-06-03 18:24:58 -04003156 if (!is_sxid(inode->i_mode) && (inode->i_sb->s_flags & MS_NOSEC))
Andi Kleen69b45732011-05-28 08:25:51 -07003157 inode->i_flags |= S_NOSEC;
3158}
3159
Al Viroa7400222014-10-21 15:20:42 -04003160static inline bool is_root_inode(struct inode *inode)
3161{
3162 return inode == inode->i_sb->s_root->d_inode;
3163}
3164
Al Viro5f99f4e2013-05-15 20:23:06 -04003165static inline bool dir_emit(struct dir_context *ctx,
3166 const char *name, int namelen,
3167 u64 ino, unsigned type)
3168{
3169 return ctx->actor(ctx, name, namelen, ctx->pos, ino, type) == 0;
3170}
3171static inline bool dir_emit_dot(struct file *file, struct dir_context *ctx)
3172{
3173 return ctx->actor(ctx, ".", 1, ctx->pos,
3174 file->f_path.dentry->d_inode->i_ino, DT_DIR) == 0;
3175}
3176static inline bool dir_emit_dotdot(struct file *file, struct dir_context *ctx)
3177{
3178 return ctx->actor(ctx, "..", 2, ctx->pos,
3179 parent_ino(file->f_path.dentry), DT_DIR) == 0;
3180}
3181static inline bool dir_emit_dots(struct file *file, struct dir_context *ctx)
3182{
3183 if (ctx->pos == 0) {
3184 if (!dir_emit_dot(file, ctx))
3185 return false;
3186 ctx->pos = 1;
3187 }
3188 if (ctx->pos == 1) {
3189 if (!dir_emit_dotdot(file, ctx))
3190 return false;
3191 ctx->pos = 2;
3192 }
3193 return true;
3194}
Al Viro5ded75e2013-05-15 21:02:48 -04003195static inline bool dir_relax(struct inode *inode)
3196{
Al Viro59551022016-01-22 15:40:57 -05003197 inode_unlock(inode);
3198 inode_lock(inode);
Al Viro5ded75e2013-05-15 21:02:48 -04003199 return !IS_DEADDIR(inode);
3200}
Al Viro5f99f4e2013-05-15 20:23:06 -04003201
Al Viroae053272016-05-12 20:36:01 -04003202static inline bool dir_relax_shared(struct inode *inode)
3203{
3204 inode_unlock_shared(inode);
3205 inode_lock_shared(inode);
3206 return !IS_DEADDIR(inode);
3207}
3208
Eric W. Biederman90f85722015-06-29 14:42:03 -05003209extern bool path_noexec(const struct path *path);
Al Viro21fc61c2015-11-17 01:07:57 -05003210extern void inode_nohighmem(struct inode *inode);
Eric W. Biederman90f85722015-06-29 14:42:03 -05003211
Linus Torvalds1da177e2005-04-16 15:20:36 -07003212#endif /* _LINUX_FS_H */