blob: d6c24b68a09263d3cff20cc9d6d1a9222201114f [file] [log] [blame]
Greg Kroah-Hartmanb2441312017-11-01 15:07:57 +01001/* SPDX-License-Identifier: GPL-2.0 */
Linus Torvalds1da177e2005-04-16 15:20:36 -07002#ifndef _LINUX_FS_H
3#define _LINUX_FS_H
4
Linus Torvalds1da177e2005-04-16 15:20:36 -07005#include <linux/linkage.h>
Ingo Molnar5dd43ce2017-06-20 12:19:09 +02006#include <linux/wait_bit.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -07007#include <linux/kdev_t.h>
8#include <linux/dcache.h>
Al Viro3f8206d2008-07-26 03:46:43 -04009#include <linux/path.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070010#include <linux/stat.h>
11#include <linux/cache.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070012#include <linux/list.h>
Dave Chinnerbc3b14c2013-08-28 10:17:58 +100013#include <linux/list_lru.h>
Oleg Nesterov4f5e65a2013-07-08 14:24:16 -070014#include <linux/llist.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070015#include <linux/radix-tree.h>
Matthew Wilcoxb93b0162018-04-10 16:36:56 -070016#include <linux/xarray.h>
Michel Lespinasse6b2dbba2012-10-08 16:31:25 -070017#include <linux/rbtree.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070018#include <linux/init.h>
Al Viro914e2632006-10-18 13:55:46 -040019#include <linux/pid.h>
Paul Gortmaker187f1882011-11-23 20:12:59 -050020#include <linux/bug.h>
Jes Sorensen1b1dcc12006-01-09 15:59:24 -080021#include <linux/mutex.h>
Davidlohr Buesoc8c06ef2014-12-12 16:54:24 -080022#include <linux/rwsem.h>
Dan Williamsbaabda22017-07-10 15:48:25 -070023#include <linux/mm_types.h>
Satyam Sharma3bd858a2007-07-17 15:00:08 +053024#include <linux/capability.h>
Matthew Wilcox6188e102008-04-18 22:21:05 -040025#include <linux/semaphore.h>
Jens Axboec75b1d92017-06-27 11:47:04 -060026#include <linux/fcntl.h>
Mark Fashehc4b929b2008-10-08 19:44:18 -040027#include <linux/fiemap.h>
Nick Pigginceb5bdc2011-01-07 17:50:05 +110028#include <linux/rculist_bl.h>
Al Viro07b8ce12011-06-20 10:52:57 -040029#include <linux/atomic.h>
Konstantin Khlebnikov83aeead2011-12-08 14:33:54 -080030#include <linux/shrinker.h>
Stephen Rothwellc1aab022012-01-24 11:41:32 +110031#include <linux/migrate_mode.h>
Eric W. Biederman92361632012-02-08 07:07:50 -080032#include <linux/uidgid.h>
Jan Kara5accdf82012-06-12 16:20:34 +020033#include <linux/lockdep.h>
Jens Axboec2b1ad82012-09-27 09:35:03 +020034#include <linux/percpu-rwsem.h>
Oleg Nesterov853b39a2015-07-22 20:21:13 +020035#include <linux/workqueue.h>
Al Virofceef392015-12-29 15:58:39 -050036#include <linux/delayed_call.h>
Christoph Hellwig85787092017-05-10 15:06:33 +020037#include <linux/uuid.h>
Jeff Layton5660e132017-07-06 07:02:25 -040038#include <linux/errseq.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070039
Linus Torvalds1da177e2005-04-16 15:20:36 -070040#include <asm/byteorder.h>
David Howells607ca462012-10-13 10:46:48 +010041#include <uapi/linux/fs.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070042
Christoph Hellwigb83ae6d2015-01-14 10:42:37 +010043struct backing_dev_info;
Tejun Heo52ebea72015-05-22 17:13:37 -040044struct bdi_writeback;
Christoph Hellwig2f8b5442016-11-01 07:40:13 -060045struct bio;
Christoph Hellwiga5694252007-07-17 04:04:28 -070046struct export_operations;
Christoph Hellwiga885c8c2006-01-08 01:02:50 -080047struct hd_geometry;
Linus Torvalds1da177e2005-04-16 15:20:36 -070048struct iovec;
Christoph Hellwig92198f72005-06-23 22:00:59 -070049struct kiocb;
Alexey Dobriyan57cc7212011-01-10 08:18:25 +020050struct kobject;
Linus Torvalds1da177e2005-04-16 15:20:36 -070051struct pipe_inode_info;
52struct poll_table_struct;
53struct kstatfs;
54struct vm_area_struct;
55struct vfsmount;
David Howells745ca242008-11-14 10:39:22 +110056struct cred;
Mel Gormana509bc12012-07-31 16:44:57 -070057struct swap_info_struct;
Cyrill Gorcunov55985dd2012-12-17 16:04:55 -080058struct seq_file;
Christoph Hellwig7b7a8662013-09-04 15:04:39 +020059struct workqueue_struct;
Al Viro3b93f912014-02-11 21:34:08 -050060struct iov_iter;
Jaegeuk Kim0b81d072015-05-15 16:26:10 -070061struct fscrypt_info;
62struct fscrypt_operations;
Linus Torvalds1da177e2005-04-16 15:20:36 -070063
Denis Cheng74bf17c2007-10-16 23:26:30 -070064extern void __init inode_init(void);
Linus Torvalds1da177e2005-04-16 15:20:36 -070065extern void __init inode_init_early(void);
Mel Gorman4248b0d2015-08-06 15:46:20 -070066extern void __init files_init(void);
67extern void __init files_maxfiles_init(void);
Linus Torvalds1da177e2005-04-16 15:20:36 -070068
Jan Engelhardtdded4f42008-12-01 14:34:50 -080069extern struct files_stat_struct files_stat;
Eric Dumazet518de9b2010-10-26 14:22:44 -070070extern unsigned long get_max_files(void);
Alexey Dobriyan9b80a182016-09-02 00:38:52 +030071extern unsigned int sysctl_nr_open;
Jan Engelhardtdded4f42008-12-01 14:34:50 -080072extern struct inodes_stat_t inodes_stat;
73extern int leases_enable, lease_break_time;
Kees Cook800179c2012-07-25 17:29:07 -070074extern int sysctl_protected_symlinks;
75extern int sysctl_protected_hardlinks;
Jan Engelhardtdded4f42008-12-01 14:34:50 -080076
Christoph Hellwigddef7ed22017-07-06 18:58:37 +020077typedef __kernel_rwf_t rwf_t;
78
Linus Torvalds1da177e2005-04-16 15:20:36 -070079struct buffer_head;
80typedef int (get_block_t)(struct inode *inode, sector_t iblock,
81 struct buffer_head *bh_result, int create);
Christoph Hellwig187372a2016-02-08 14:40:51 +110082typedef int (dio_iodone_t)(struct kiocb *iocb, loff_t offset,
Christoph Hellwig7b7a8662013-09-04 15:04:39 +020083 ssize_t bytes, void *private);
Linus Torvalds1da177e2005-04-16 15:20:36 -070084
David Howellsbbc10962012-10-15 16:40:35 +010085#define MAY_EXEC 0x00000001
86#define MAY_WRITE 0x00000002
87#define MAY_READ 0x00000004
88#define MAY_APPEND 0x00000008
89#define MAY_ACCESS 0x00000010
90#define MAY_OPEN 0x00000020
91#define MAY_CHDIR 0x00000040
92/* called from RCU mode, don't block */
93#define MAY_NOT_BLOCK 0x00000080
94
95/*
96 * flags in file.f_mode. Note that FMODE_READ and FMODE_WRITE must correspond
97 * to O_WRONLY and O_RDWR via the strange trick in __dentry_open()
98 */
99
100/* file is open for reading */
101#define FMODE_READ ((__force fmode_t)0x1)
102/* file is open for writing */
103#define FMODE_WRITE ((__force fmode_t)0x2)
104/* file is seekable */
105#define FMODE_LSEEK ((__force fmode_t)0x4)
106/* file can be accessed using pread */
107#define FMODE_PREAD ((__force fmode_t)0x8)
108/* file can be accessed using pwrite */
109#define FMODE_PWRITE ((__force fmode_t)0x10)
110/* File is opened for execution with sys_execve / sys_uselib */
111#define FMODE_EXEC ((__force fmode_t)0x20)
112/* File is opened with O_NDELAY (only set for block devices) */
113#define FMODE_NDELAY ((__force fmode_t)0x40)
114/* File is opened with O_EXCL (only set for block devices) */
115#define FMODE_EXCL ((__force fmode_t)0x80)
116/* File is opened using open(.., 3, ..) and is writeable only for ioctls
117 (specialy hack for floppy.c) */
118#define FMODE_WRITE_IOCTL ((__force fmode_t)0x100)
119/* 32bit hashes as llseek() offset (for directories) */
120#define FMODE_32BITHASH ((__force fmode_t)0x200)
121/* 64bit hashes as llseek() offset (for directories) */
122#define FMODE_64BITHASH ((__force fmode_t)0x400)
123
124/*
125 * Don't update ctime and mtime.
126 *
127 * Currently a special hack for the XFS open_by_handle ioctl, but we'll
128 * hopefully graduate it to a proper O_CMTIME flag supported by open(2) soon.
129 */
130#define FMODE_NOCMTIME ((__force fmode_t)0x800)
131
132/* Expect random access pattern */
133#define FMODE_RANDOM ((__force fmode_t)0x1000)
134
135/* File is huge (eg. /dev/kmem): treat loff_t as unsigned */
136#define FMODE_UNSIGNED_OFFSET ((__force fmode_t)0x2000)
137
138/* File is opened with O_PATH; almost nothing can be done with it */
139#define FMODE_PATH ((__force fmode_t)0x4000)
140
Linus Torvalds9c225f22014-03-03 09:36:58 -0800141/* File needs atomic accesses to f_pos */
142#define FMODE_ATOMIC_POS ((__force fmode_t)0x8000)
Al Viro83f936c2014-03-14 12:02:47 -0400143/* Write access to underlying fs */
144#define FMODE_WRITER ((__force fmode_t)0x10000)
Al Viro7f7f25e2014-02-11 17:49:24 -0500145/* Has read method(s) */
146#define FMODE_CAN_READ ((__force fmode_t)0x20000)
147/* Has write method(s) */
148#define FMODE_CAN_WRITE ((__force fmode_t)0x40000)
Linus Torvalds9c225f22014-03-03 09:36:58 -0800149
David Howellsbbc10962012-10-15 16:40:35 +0100150/* File was opened by fanotify and shouldn't generate fanotify events */
David Drysdale75069f22015-01-08 14:32:29 -0800151#define FMODE_NONOTIFY ((__force fmode_t)0x4000000)
David Howellsbbc10962012-10-15 16:40:35 +0100152
Christoph Hellwig91f99432017-08-29 16:13:20 +0200153/* File is capable of returning -EAGAIN if I/O will block */
154#define FMODE_NOWAIT ((__force fmode_t)0x8000000)
Goldwyn Rodriguesb745faf2017-06-20 07:05:43 -0500155
David Howellsbbc10962012-10-15 16:40:35 +0100156/*
157 * Flag for rw_copy_check_uvector and compat_rw_copy_check_uvector
158 * that indicates that they should check the contents of the iovec are
159 * valid, but not check the memory that the iovec elements
160 * points too.
161 */
162#define CHECK_IOVEC_ONLY -1
163
Linus Torvalds1da177e2005-04-16 15:20:36 -0700164/*
165 * Attribute flags. These should be or-ed together to figure out what
166 * has been changed!
167 */
Miklos Szeredi9767d742008-07-01 15:01:26 +0200168#define ATTR_MODE (1 << 0)
169#define ATTR_UID (1 << 1)
170#define ATTR_GID (1 << 2)
171#define ATTR_SIZE (1 << 3)
172#define ATTR_ATIME (1 << 4)
173#define ATTR_MTIME (1 << 5)
174#define ATTR_CTIME (1 << 6)
175#define ATTR_ATIME_SET (1 << 7)
176#define ATTR_MTIME_SET (1 << 8)
177#define ATTR_FORCE (1 << 9) /* Not a change, but a change it */
178#define ATTR_ATTR_FLAG (1 << 10)
179#define ATTR_KILL_SUID (1 << 11)
180#define ATTR_KILL_SGID (1 << 12)
181#define ATTR_FILE (1 << 13)
182#define ATTR_KILL_PRIV (1 << 14)
183#define ATTR_OPEN (1 << 15) /* Truncating from open(O_TRUNC) */
184#define ATTR_TIMES_SET (1 << 16)
Miklos Szeredif2b20f62016-09-16 12:44:20 +0200185#define ATTR_TOUCH (1 << 17)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700186
187/*
Miklos Szeredi787fb6b2014-10-24 00:14:36 +0200188 * Whiteout is represented by a char device. The following constants define the
189 * mode and device number to use.
190 */
191#define WHITEOUT_MODE 0
192#define WHITEOUT_DEV 0
193
194/*
Linus Torvalds1da177e2005-04-16 15:20:36 -0700195 * This is the Inode Attributes structure, used for notify_change(). It
196 * uses the above definitions as flags, to know which values have changed.
197 * Also, in this manner, a Filesystem can look at only the values it cares
198 * about. Basically, these are the attributes that the VFS layer can
199 * request to change from the FS layer.
200 *
201 * Derek Atkins <warlord@MIT.EDU> 94-10-20
202 */
203struct iattr {
204 unsigned int ia_valid;
205 umode_t ia_mode;
Eric W. Biederman92361632012-02-08 07:07:50 -0800206 kuid_t ia_uid;
207 kgid_t ia_gid;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700208 loff_t ia_size;
209 struct timespec ia_atime;
210 struct timespec ia_mtime;
211 struct timespec ia_ctime;
Miklos Szeredicc4e69d2005-11-07 00:59:49 -0800212
213 /*
Lucas De Marchi25985ed2011-03-30 22:57:33 -0300214 * Not an attribute, but an auxiliary info for filesystems wanting to
Miklos Szeredicc4e69d2005-11-07 00:59:49 -0800215 * implement an ftruncate() like method. NOTE: filesystem should
216 * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL).
217 */
218 struct file *ia_file;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700219};
220
221/*
Linus Torvalds1da177e2005-04-16 15:20:36 -0700222 * Includes for diskquotas.
223 */
224#include <linux/quota.h>
225
Miklos Szeredi69c433e2014-10-24 00:14:39 +0200226/*
227 * Maximum number of layers of fs stack. Needs to be limited to
228 * prevent kernel stack overflow
229 */
230#define FILESYSTEM_MAX_STACK_DEPTH 2
231
Zach Brown994fc28c2005-12-15 14:28:17 -0800232/**
233 * enum positive_aop_returns - aop return codes with specific semantics
234 *
235 * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has
236 * completed, that the page is still locked, and
237 * should be considered active. The VM uses this hint
238 * to return the page to the active list -- it won't
239 * be a candidate for writeback again in the near
240 * future. Other callers must be careful to unlock
241 * the page if they get this return. Returned by
242 * writepage();
243 *
244 * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has
245 * unlocked it and the page might have been truncated.
246 * The caller should back up to acquiring a new page and
247 * trying again. The aop will be taking reasonable
248 * precautions not to livelock. If the caller held a page
249 * reference, it should drop it before retrying. Returned
Nick Piggin55144762007-10-16 01:25:26 -0700250 * by readpage().
Zach Brown994fc28c2005-12-15 14:28:17 -0800251 *
252 * address_space_operation functions return these large constants to indicate
253 * special semantics to the caller. These are much larger than the bytes in a
254 * page to allow for functions that return the number of bytes operated on in a
255 * given page.
256 */
257
258enum positive_aop_returns {
259 AOP_WRITEPAGE_ACTIVATE = 0x80000,
260 AOP_TRUNCATED_PAGE = 0x80001,
261};
262
Tetsuo Handac718a972017-05-08 15:58:59 -0700263#define AOP_FLAG_CONT_EXPAND 0x0001 /* called from cont_expand */
264#define AOP_FLAG_NOFS 0x0002 /* used by filesystem to direct
Nick Piggin54566b22009-01-04 12:00:53 -0800265 * helper code (eg buffer layer)
266 * to clear GFP_FS from alloc */
Nick Pigginafddba42007-10-16 01:25:01 -0700267
Linus Torvalds1da177e2005-04-16 15:20:36 -0700268/*
269 * oh the beauties of C type declarations.
270 */
271struct page;
272struct address_space;
273struct writeback_control;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700274
Jens Axboec75b1d92017-06-27 11:47:04 -0600275/*
276 * Write life time hint values.
277 */
278enum rw_hint {
279 WRITE_LIFE_NOT_SET = 0,
280 WRITE_LIFE_NONE = RWH_WRITE_LIFE_NONE,
281 WRITE_LIFE_SHORT = RWH_WRITE_LIFE_SHORT,
282 WRITE_LIFE_MEDIUM = RWH_WRITE_LIFE_MEDIUM,
283 WRITE_LIFE_LONG = RWH_WRITE_LIFE_LONG,
284 WRITE_LIFE_EXTREME = RWH_WRITE_LIFE_EXTREME,
285};
286
Christoph Hellwige2e40f22015-02-22 08:58:50 -0800287#define IOCB_EVENTFD (1 << 0)
Al Viro2ba48ce2015-04-09 13:52:01 -0400288#define IOCB_APPEND (1 << 1)
289#define IOCB_DIRECT (1 << 2)
Christoph Hellwig97be7eb2016-03-03 16:04:01 +0100290#define IOCB_HIPRI (1 << 3)
Christoph Hellwigdde0c2e2016-04-07 08:52:00 -0700291#define IOCB_DSYNC (1 << 4)
292#define IOCB_SYNC (1 << 5)
Jan Kara70fe2f42016-10-30 11:42:04 -0500293#define IOCB_WRITE (1 << 6)
Goldwyn Rodriguesb745faf2017-06-20 07:05:43 -0500294#define IOCB_NOWAIT (1 << 7)
Christoph Hellwige2e40f22015-02-22 08:58:50 -0800295
296struct kiocb {
297 struct file *ki_filp;
298 loff_t ki_pos;
299 void (*ki_complete)(struct kiocb *iocb, long ret, long ret2);
300 void *private;
301 int ki_flags;
Jens Axboec75b1d92017-06-27 11:47:04 -0600302 enum rw_hint ki_hint;
Kees Cook3859a272016-10-28 01:22:25 -0700303} __randomize_layout;
Christoph Hellwige2e40f22015-02-22 08:58:50 -0800304
305static inline bool is_sync_kiocb(struct kiocb *kiocb)
306{
307 return kiocb->ki_complete == NULL;
308}
309
Hisashi Hifumi8ab22b92008-07-28 15:46:36 -0700310/*
311 * "descriptor" for what we're up to with a read.
312 * This allows us to use the same read code yet
313 * have multiple different users of the data that
314 * we read from a file.
315 *
316 * The simplest case just copies the data to user
317 * mode.
318 */
319typedef struct {
320 size_t written;
321 size_t count;
322 union {
323 char __user *buf;
324 void *data;
325 } arg;
326 int error;
327} read_descriptor_t;
328
329typedef int (*read_actor_t)(read_descriptor_t *, struct page *,
330 unsigned long, unsigned long);
Nick Piggin2f718ff2007-10-16 01:24:59 -0700331
Linus Torvalds1da177e2005-04-16 15:20:36 -0700332struct address_space_operations {
333 int (*writepage)(struct page *page, struct writeback_control *wbc);
334 int (*readpage)(struct file *, struct page *);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700335
336 /* Write back some dirty pages from this mapping. */
337 int (*writepages)(struct address_space *, struct writeback_control *);
338
Andrew Morton4741c9f2006-03-24 03:18:11 -0800339 /* Set a page dirty. Return true if this dirtied it */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700340 int (*set_page_dirty)(struct page *page);
341
342 int (*readpages)(struct file *filp, struct address_space *mapping,
343 struct list_head *pages, unsigned nr_pages);
344
Nick Pigginafddba42007-10-16 01:25:01 -0700345 int (*write_begin)(struct file *, struct address_space *mapping,
346 loff_t pos, unsigned len, unsigned flags,
347 struct page **pagep, void **fsdata);
348 int (*write_end)(struct file *, struct address_space *mapping,
349 loff_t pos, unsigned len, unsigned copied,
350 struct page *page, void *fsdata);
351
Linus Torvalds1da177e2005-04-16 15:20:36 -0700352 /* Unfortunately this kludge is needed for FIBMAP. Don't use it */
353 sector_t (*bmap)(struct address_space *, sector_t);
Lukas Czernerd47992f2013-05-21 23:17:23 -0400354 void (*invalidatepage) (struct page *, unsigned int, unsigned int);
Al Viro27496a82005-10-21 03:20:48 -0400355 int (*releasepage) (struct page *, gfp_t);
Linus Torvalds6072d132010-12-01 13:35:19 -0500356 void (*freepage)(struct page *);
Christoph Hellwigc8b8e322016-04-07 08:51:58 -0700357 ssize_t (*direct_IO)(struct kiocb *, struct iov_iter *iter);
Mel Gormanb969c4a2012-01-12 17:19:34 -0800358 /*
Tang Chenef277c72013-07-08 16:00:21 -0700359 * migrate the contents of a page to the specified target. If
360 * migrate_mode is MIGRATE_ASYNC, it must not block.
Mel Gormanb969c4a2012-01-12 17:19:34 -0800361 */
Christoph Lameter2d1db3b2006-06-23 02:03:33 -0700362 int (*migratepage) (struct address_space *,
Mel Gormana6bc32b2012-01-12 17:19:43 -0800363 struct page *, struct page *, enum migrate_mode);
Minchan Kimbda807d2016-07-26 15:23:05 -0700364 bool (*isolate_page)(struct page *, isolate_mode_t);
365 void (*putback_page)(struct page *);
Trond Myklebuste3db7692007-01-10 23:15:39 -0800366 int (*launder_page) (struct page *);
Al Viroc186afb42014-02-02 21:16:54 -0500367 int (*is_partially_uptodate) (struct page *, unsigned long,
Hisashi Hifumi8ab22b92008-07-28 15:46:36 -0700368 unsigned long);
Mel Gormanb4597222013-07-03 15:02:05 -0700369 void (*is_dirty_writeback) (struct page *, bool *, bool *);
Andi Kleen25718732009-09-16 11:50:13 +0200370 int (*error_remove_page)(struct address_space *, struct page *);
Mel Gorman62c230b2012-07-31 16:44:55 -0700371
372 /* swapfile support */
Mel Gormana509bc12012-07-31 16:44:57 -0700373 int (*swap_activate)(struct swap_info_struct *sis, struct file *file,
374 sector_t *span);
375 void (*swap_deactivate)(struct file *file);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700376};
377
Jens Axboe7dcda1c2011-04-05 23:51:48 +0200378extern const struct address_space_operations empty_aops;
379
Nick Pigginafddba42007-10-16 01:25:01 -0700380/*
381 * pagecache_write_begin/pagecache_write_end must be used by general code
382 * to write into the pagecache.
383 */
384int pagecache_write_begin(struct file *, struct address_space *mapping,
385 loff_t pos, unsigned len, unsigned flags,
386 struct page **pagep, void **fsdata);
387
388int pagecache_write_end(struct file *, struct address_space *mapping,
389 loff_t pos, unsigned len, unsigned copied,
390 struct page *page, void *fsdata);
391
Linus Torvalds1da177e2005-04-16 15:20:36 -0700392struct address_space {
393 struct inode *host; /* owner: inode, block_device */
Matthew Wilcoxb93b0162018-04-10 16:36:56 -0700394 struct radix_tree_root i_pages; /* cached pages */
David Herrmann4bb5f5d2014-08-08 14:25:25 -0700395 atomic_t i_mmap_writable;/* count VM_SHARED mappings */
Davidlohr Buesof808c132017-09-08 16:15:08 -0700396 struct rb_root_cached i_mmap; /* tree of private and shared mappings */
Davidlohr Buesoc8c06ef2014-12-12 16:54:24 -0800397 struct rw_semaphore i_mmap_rwsem; /* protect tree, count, list */
Matthew Wilcoxb93b0162018-04-10 16:36:56 -0700398 /* Protected by the i_pages lock */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700399 unsigned long nrpages; /* number of total pages */
Ross Zwislerf9fe48b2016-01-22 15:10:40 -0800400 /* number of shadow or DAX exceptional entries */
401 unsigned long nrexceptional;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700402 pgoff_t writeback_index;/* writeback starts here */
Christoph Hellwigf5e54d62006-06-28 04:26:44 -0700403 const struct address_space_operations *a_ops; /* methods */
Michal Hocko9c5d7602016-10-11 13:56:04 -0700404 unsigned long flags; /* error bits */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700405 spinlock_t private_lock; /* for use by the address_space */
Michal Hocko9c5d7602016-10-11 13:56:04 -0700406 gfp_t gfp_mask; /* implicit gfp mask for allocations */
Mike Rapoport32e57c22017-10-03 16:16:54 -0700407 struct list_head private_list; /* for use by the address_space */
Rafael Aquini252aa6f2012-12-11 16:02:35 -0800408 void *private_data; /* ditto */
Jeff Layton5660e132017-07-06 07:02:25 -0400409 errseq_t wb_err;
Kees Cook3859a272016-10-28 01:22:25 -0700410} __attribute__((aligned(sizeof(long)))) __randomize_layout;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700411 /*
412 * On most architectures that alignment is already the case; but
Lucas De Marchi25985ed2011-03-30 22:57:33 -0300413 * must be enforced here for CRIS, to let the least significant bit
Linus Torvalds1da177e2005-04-16 15:20:36 -0700414 * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
415 */
Andi Kleen87192a22012-01-12 17:20:34 -0800416struct request_queue;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700417
418struct block_device {
419 dev_t bd_dev; /* not a kdev_t - it's a search key */
Richard Kennedyf3ccfcd2011-03-22 16:33:56 -0700420 int bd_openers;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700421 struct inode * bd_inode; /* will die */
Theodore Ts'o87d8fe12009-01-03 09:47:09 -0500422 struct super_block * bd_super;
Arjan van de Venc039e312006-03-23 03:00:28 -0800423 struct mutex bd_mutex; /* open/close mutex */
Tejun Heo6b4517a2010-04-07 18:53:59 +0900424 void * bd_claiming;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700425 void * bd_holder;
426 int bd_holders;
Tejun Heo77ea8872010-12-08 20:57:37 +0100427 bool bd_write_holder;
Jun'ichi Nomura641dc632006-03-27 01:17:57 -0800428#ifdef CONFIG_SYSFS
Tejun Heo49731ba2011-01-14 18:43:57 +0100429 struct list_head bd_holder_disks;
Jun'ichi Nomura641dc632006-03-27 01:17:57 -0800430#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -0700431 struct block_device * bd_contains;
432 unsigned bd_block_size;
Christoph Hellwigc2ee0702017-08-23 19:10:31 +0200433 u8 bd_partno;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700434 struct hd_struct * bd_part;
435 /* number of times partitions within this device have been opened. */
436 unsigned bd_part_count;
437 int bd_invalidated;
438 struct gendisk * bd_disk;
Andi Kleen87192a22012-01-12 17:20:34 -0800439 struct request_queue * bd_queue;
Jan Karab1d2dc562017-02-02 15:56:52 +0100440 struct backing_dev_info *bd_bdi;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700441 struct list_head bd_list;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700442 /*
443 * Private data. You must have bd_claim'ed the block_device
444 * to use this. NOTE: bd_claim allows an owner to claim
445 * the same device multiple times, the owner must take special
446 * care to not mess up bd_private for that case.
447 */
448 unsigned long bd_private;
Takashi Satofcccf502009-01-09 16:40:59 -0800449
450 /* The counter of freeze processes */
451 int bd_fsfreeze_count;
452 /* Mutex for freeze */
453 struct mutex bd_fsfreeze_mutex;
Kees Cook3859a272016-10-28 01:22:25 -0700454} __randomize_layout;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700455
456/*
457 * Radix-tree tags, for tagging dirty and writeback pages within the pagecache
458 * radix trees
459 */
460#define PAGECACHE_TAG_DIRTY 0
461#define PAGECACHE_TAG_WRITEBACK 1
Jan Karaf446daae2010-08-09 17:19:12 -0700462#define PAGECACHE_TAG_TOWRITE 2
Linus Torvalds1da177e2005-04-16 15:20:36 -0700463
464int mapping_tagged(struct address_space *mapping, int tag);
465
Davidlohr Bueso8b28f622014-12-12 16:54:18 -0800466static inline void i_mmap_lock_write(struct address_space *mapping)
467{
Davidlohr Buesoc8c06ef2014-12-12 16:54:24 -0800468 down_write(&mapping->i_mmap_rwsem);
Davidlohr Bueso8b28f622014-12-12 16:54:18 -0800469}
470
471static inline void i_mmap_unlock_write(struct address_space *mapping)
472{
Davidlohr Buesoc8c06ef2014-12-12 16:54:24 -0800473 up_write(&mapping->i_mmap_rwsem);
Davidlohr Bueso8b28f622014-12-12 16:54:18 -0800474}
475
Davidlohr Bueso3dec0ba2014-12-12 16:54:27 -0800476static inline void i_mmap_lock_read(struct address_space *mapping)
477{
478 down_read(&mapping->i_mmap_rwsem);
479}
480
481static inline void i_mmap_unlock_read(struct address_space *mapping)
482{
483 up_read(&mapping->i_mmap_rwsem);
484}
485
Linus Torvalds1da177e2005-04-16 15:20:36 -0700486/*
487 * Might pages of this file be mapped into userspace?
488 */
489static inline int mapping_mapped(struct address_space *mapping)
490{
Davidlohr Buesof808c132017-09-08 16:15:08 -0700491 return !RB_EMPTY_ROOT(&mapping->i_mmap.rb_root);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700492}
493
494/*
495 * Might pages of this file have been modified in userspace?
496 * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap_pgoff
497 * marks vma as VM_SHARED if it is shared, and the file was opened for
498 * writing i.e. vma may be mprotected writable even if now readonly.
David Herrmann4bb5f5d2014-08-08 14:25:25 -0700499 *
500 * If i_mmap_writable is negative, no new writable mappings are allowed. You
501 * can only deny writable mappings, if none exists right now.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700502 */
503static inline int mapping_writably_mapped(struct address_space *mapping)
504{
David Herrmann4bb5f5d2014-08-08 14:25:25 -0700505 return atomic_read(&mapping->i_mmap_writable) > 0;
506}
507
508static inline int mapping_map_writable(struct address_space *mapping)
509{
510 return atomic_inc_unless_negative(&mapping->i_mmap_writable) ?
511 0 : -EPERM;
512}
513
514static inline void mapping_unmap_writable(struct address_space *mapping)
515{
516 atomic_dec(&mapping->i_mmap_writable);
517}
518
519static inline int mapping_deny_writable(struct address_space *mapping)
520{
521 return atomic_dec_unless_positive(&mapping->i_mmap_writable) ?
522 0 : -EBUSY;
523}
524
525static inline void mapping_allow_writable(struct address_space *mapping)
526{
527 atomic_inc(&mapping->i_mmap_writable);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700528}
529
530/*
531 * Use sequence counter to get consistent i_size on 32-bit processors.
532 */
533#if BITS_PER_LONG==32 && defined(CONFIG_SMP)
534#include <linux/seqlock.h>
535#define __NEED_I_SIZE_ORDERED
536#define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
537#else
538#define i_size_ordered_init(inode) do { } while (0)
539#endif
540
Al Virof19d4a82009-06-08 19:50:45 -0400541struct posix_acl;
542#define ACL_NOT_CACHED ((void *)(-1))
Miklos Szeredi2a3a2a32016-09-01 11:11:59 +0200543#define ACL_DONT_CACHE ((void *)(-3))
Al Virof19d4a82009-06-08 19:50:45 -0400544
Andreas Gruenbacherb8a7a3a2016-03-24 14:38:37 +0100545static inline struct posix_acl *
546uncached_acl_sentinel(struct task_struct *task)
547{
548 return (void *)task + 1;
549}
550
551static inline bool
552is_uncached_acl(struct posix_acl *acl)
553{
554 return (long)acl & 1;
555}
556
Linus Torvalds3ddcd052011-08-06 22:45:50 -0700557#define IOP_FASTPERM 0x0001
558#define IOP_LOOKUP 0x0002
559#define IOP_NOFOLLOW 0x0004
Andreas Gruenbacherd0a5b992016-09-29 17:48:39 +0200560#define IOP_XATTR 0x0008
Miklos Szeredi76fca902016-12-09 16:45:04 +0100561#define IOP_DEFAULT_READLINK 0x0010
Linus Torvalds3ddcd052011-08-06 22:45:50 -0700562
Jan Kara9dd813c2017-03-14 12:31:02 +0100563struct fsnotify_mark_connector;
564
Linus Torvalds3ddcd052011-08-06 22:45:50 -0700565/*
566 * Keep mostly read-only and often accessed (especially for
567 * the RCU path lookup and 'stat' data) fields at the beginning
568 * of the 'struct inode'
569 */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700570struct inode {
Nick Piggin44a7d7a2011-01-07 17:49:56 +1100571 umode_t i_mode;
Linus Torvalds3ddcd052011-08-06 22:45:50 -0700572 unsigned short i_opflags;
Eric W. Biederman92361632012-02-08 07:07:50 -0800573 kuid_t i_uid;
574 kgid_t i_gid;
Linus Torvalds3ddcd052011-08-06 22:45:50 -0700575 unsigned int i_flags;
576
577#ifdef CONFIG_FS_POSIX_ACL
578 struct posix_acl *i_acl;
579 struct posix_acl *i_default_acl;
580#endif
581
Nick Piggin44a7d7a2011-01-07 17:49:56 +1100582 const struct inode_operations *i_op;
583 struct super_block *i_sb;
Linus Torvalds3ddcd052011-08-06 22:45:50 -0700584 struct address_space *i_mapping;
Nick Piggin44a7d7a2011-01-07 17:49:56 +1100585
Linus Torvalds13e12d12011-06-08 15:18:19 -0700586#ifdef CONFIG_SECURITY
587 void *i_security;
588#endif
Nick Piggin44a7d7a2011-01-07 17:49:56 +1100589
Linus Torvalds3ddcd052011-08-06 22:45:50 -0700590 /* Stat data, not accessed from path walking */
591 unsigned long i_ino;
Miklos Szeredia78ef702011-10-28 14:13:30 +0200592 /*
593 * Filesystems may only read i_nlink directly. They shall use the
594 * following functions for modification:
595 *
596 * (set|clear|inc|drop)_nlink
597 * inode_(inc|dec)_link_count
598 */
599 union {
600 const unsigned int i_nlink;
601 unsigned int __i_nlink;
602 };
Linus Torvalds3ddcd052011-08-06 22:45:50 -0700603 dev_t i_rdev;
Linus Torvalds2f9d3df2012-06-03 14:50:19 -0700604 loff_t i_size;
Linus Torvalds3ddcd052011-08-06 22:45:50 -0700605 struct timespec i_atime;
606 struct timespec i_mtime;
607 struct timespec i_ctime;
Theodore Ts'o6cdbb0e2011-10-29 08:24:18 -0400608 spinlock_t i_lock; /* i_blocks, i_bytes, maybe i_size */
609 unsigned short i_bytes;
Linus Torvalds2f9d3df2012-06-03 14:50:19 -0700610 unsigned int i_blkbits;
Jens Axboec75b1d92017-06-27 11:47:04 -0600611 enum rw_hint i_write_hint;
Linus Torvalds3ddcd052011-08-06 22:45:50 -0700612 blkcnt_t i_blocks;
613
614#ifdef __NEED_I_SIZE_ORDERED
615 seqcount_t i_size_seqcount;
616#endif
617
618 /* Misc */
619 unsigned long i_state;
Al Viro9902af72016-04-15 15:08:36 -0400620 struct rw_semaphore i_rwsem;
Linus Torvalds13e12d12011-06-08 15:18:19 -0700621
Nick Piggin44a7d7a2011-01-07 17:49:56 +1100622 unsigned long dirtied_when; /* jiffies of first dirtying */
Theodore Ts'oa2f48702015-03-17 12:23:19 -0400623 unsigned long dirtied_time_when;
Nick Piggin44a7d7a2011-01-07 17:49:56 +1100624
Linus Torvalds1da177e2005-04-16 15:20:36 -0700625 struct hlist_node i_hash;
Dave Chinnerc7f54082015-03-04 14:07:22 -0500626 struct list_head i_io_list; /* backing dev IO list */
Tejun Heo52ebea72015-05-22 17:13:37 -0400627#ifdef CONFIG_CGROUP_WRITEBACK
628 struct bdi_writeback *i_wb; /* the associated cgroup wb */
Tejun Heo2a814902015-05-28 14:50:51 -0400629
630 /* foreign inode detection, see wbc_detach_inode() */
631 int i_wb_frn_winner;
632 u16 i_wb_frn_avg_time;
633 u16 i_wb_frn_history;
Tejun Heo52ebea72015-05-22 17:13:37 -0400634#endif
Nick Piggin7ccf19a2010-10-21 11:49:30 +1100635 struct list_head i_lru; /* inode LRU list */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700636 struct list_head i_sb_list;
Dave Chinner6c60d2b2016-07-26 15:21:50 -0700637 struct list_head i_wb_list; /* backing dev writeback list */
Nick Pigginfa0d7e3d2011-01-07 17:49:49 +1100638 union {
Al Virob3d9b7a2012-06-09 13:51:19 -0400639 struct hlist_head i_dentry;
Nick Pigginfa0d7e3d2011-01-07 17:49:49 +1100640 struct rcu_head i_rcu;
641 };
Jeff Laytonf02a9ad2017-12-21 07:45:44 -0500642 atomic64_t i_version;
Linus Torvalds2f9d3df2012-06-03 14:50:19 -0700643 atomic_t i_count;
Christoph Hellwigbd5fe6c2011-06-24 14:29:43 -0400644 atomic_t i_dio_count;
Theodore Ts'o6cdbb0e2011-10-29 08:24:18 -0400645 atomic_t i_writecount;
Mimi Zohard984ea62013-12-11 15:20:54 -0500646#ifdef CONFIG_IMA
647 atomic_t i_readcount; /* struct files open RO */
648#endif
Arjan van de Ven99ac48f2006-03-28 01:56:41 -0800649 const struct file_operations *i_fop; /* former ->i_op->default_file_ops */
Jeff Layton4a075e32015-01-16 15:05:54 -0500650 struct file_lock_context *i_flctx;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700651 struct address_space i_data;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700652 struct list_head i_devices;
Theodore Ts'o4c154162006-09-27 01:50:47 -0700653 union {
654 struct pipe_inode_info *i_pipe;
Theodore Ts'oeaf796e2006-09-27 01:50:48 -0700655 struct block_device *i_bdev;
Theodore Ts'o577c4eb2006-09-27 01:50:49 -0700656 struct cdev *i_cdev;
Al Viro61ba64f2015-05-02 09:54:06 -0400657 char *i_link;
Al Viro84e710d2016-04-15 00:58:55 -0400658 unsigned i_dir_seq;
Theodore Ts'o4c154162006-09-27 01:50:47 -0700659 };
Linus Torvalds1da177e2005-04-16 15:20:36 -0700660
661 __u32 i_generation;
662
Eric Paris3be25f42009-05-21 17:01:26 -0400663#ifdef CONFIG_FSNOTIFY
664 __u32 i_fsnotify_mask; /* all events this inode cares about */
Jan Kara08991e82017-02-01 09:21:58 +0100665 struct fsnotify_mark_connector __rcu *i_fsnotify_marks;
Robert Love0eeca282005-07-12 17:06:03 -0400666#endif
667
Jaegeuk Kim0b81d072015-05-15 16:26:10 -0700668#if IS_ENABLED(CONFIG_FS_ENCRYPTION)
669 struct fscrypt_info *i_crypt_info;
670#endif
671
Theodore Ts'o8e18e292006-09-27 01:50:46 -0700672 void *i_private; /* fs or device private pointer */
Kees Cook3859a272016-10-28 01:22:25 -0700673} __randomize_layout;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700674
Fabian Frederick93407472017-02-27 14:28:32 -0800675static inline unsigned int i_blocksize(const struct inode *node)
676{
677 return (1 << node->i_blkbits);
678}
679
Al Viro1d3382cb2010-10-23 15:19:20 -0400680static inline int inode_unhashed(struct inode *inode)
681{
682 return hlist_unhashed(&inode->i_hash);
683}
684
Linus Torvalds1da177e2005-04-16 15:20:36 -0700685/*
Ingo Molnarf2eace22006-07-03 00:25:05 -0700686 * inode->i_mutex nesting subclasses for the lock validator:
687 *
688 * 0: the object of the current VFS operation
689 * 1: parent
690 * 2: child/target
J. Bruce Fields40bd22c2012-04-18 15:21:34 -0400691 * 3: xattr
692 * 4: second non-directory
Miklos Szeredid1b72cc2014-10-27 15:42:01 +0100693 * 5: second parent (when locking independent directories in rename)
694 *
695 * I_MUTEX_NONDIR2 is for certain operations (such as rename) which lock two
J. Bruce Fields40bd22c2012-04-18 15:21:34 -0400696 * non-directories at once.
Ingo Molnarf2eace22006-07-03 00:25:05 -0700697 *
698 * The locking order between these classes is
Miklos Szeredid1b72cc2014-10-27 15:42:01 +0100699 * parent[2] -> child -> grandchild -> normal -> xattr -> second non-directory
Ingo Molnarf2eace22006-07-03 00:25:05 -0700700 */
701enum inode_i_mutex_lock_class
702{
703 I_MUTEX_NORMAL,
704 I_MUTEX_PARENT,
705 I_MUTEX_CHILD,
Ingo Molnar4df46242006-08-27 01:23:56 -0700706 I_MUTEX_XATTR,
Miklos Szeredid1b72cc2014-10-27 15:42:01 +0100707 I_MUTEX_NONDIR2,
708 I_MUTEX_PARENT2,
Ingo Molnarf2eace22006-07-03 00:25:05 -0700709};
710
Al Viro59551022016-01-22 15:40:57 -0500711static inline void inode_lock(struct inode *inode)
712{
Al Viro9902af72016-04-15 15:08:36 -0400713 down_write(&inode->i_rwsem);
Al Viro59551022016-01-22 15:40:57 -0500714}
715
716static inline void inode_unlock(struct inode *inode)
717{
Al Viro9902af72016-04-15 15:08:36 -0400718 up_write(&inode->i_rwsem);
719}
720
721static inline void inode_lock_shared(struct inode *inode)
722{
723 down_read(&inode->i_rwsem);
724}
725
726static inline void inode_unlock_shared(struct inode *inode)
727{
728 up_read(&inode->i_rwsem);
Al Viro59551022016-01-22 15:40:57 -0500729}
730
731static inline int inode_trylock(struct inode *inode)
732{
Al Viro9902af72016-04-15 15:08:36 -0400733 return down_write_trylock(&inode->i_rwsem);
734}
735
736static inline int inode_trylock_shared(struct inode *inode)
737{
738 return down_read_trylock(&inode->i_rwsem);
Al Viro59551022016-01-22 15:40:57 -0500739}
740
741static inline int inode_is_locked(struct inode *inode)
742{
Al Viro9902af72016-04-15 15:08:36 -0400743 return rwsem_is_locked(&inode->i_rwsem);
Al Viro59551022016-01-22 15:40:57 -0500744}
745
746static inline void inode_lock_nested(struct inode *inode, unsigned subclass)
747{
Al Viro9902af72016-04-15 15:08:36 -0400748 down_write_nested(&inode->i_rwsem, subclass);
Al Viro59551022016-01-22 15:40:57 -0500749}
750
Darrick J. Wong01c2e132018-01-18 14:07:53 -0800751static inline void inode_lock_shared_nested(struct inode *inode, unsigned subclass)
752{
753 down_read_nested(&inode->i_rwsem, subclass);
754}
755
J. Bruce Fields375e2892012-04-18 15:16:33 -0400756void lock_two_nondirectories(struct inode *, struct inode*);
757void unlock_two_nondirectories(struct inode *, struct inode*);
758
Ingo Molnarf2eace22006-07-03 00:25:05 -0700759/*
Linus Torvalds1da177e2005-04-16 15:20:36 -0700760 * NOTE: in a 32bit arch with a preemptable kernel and
761 * an UP compile the i_size_read/write must be atomic
762 * with respect to the local cpu (unlike with preempt disabled),
763 * but they don't need to be atomic with respect to other cpus like in
764 * true SMP (so they need either to either locally disable irq around
765 * the read or for example on x86 they can be still implemented as a
766 * cmpxchg8b without the need of the lock prefix). For SMP compiles
767 * and 64bit archs it makes no difference if preempt is enabled or not.
768 */
Jan Engelhardt48ed2142006-12-06 20:35:37 -0800769static inline loff_t i_size_read(const struct inode *inode)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700770{
771#if BITS_PER_LONG==32 && defined(CONFIG_SMP)
772 loff_t i_size;
773 unsigned int seq;
774
775 do {
776 seq = read_seqcount_begin(&inode->i_size_seqcount);
777 i_size = inode->i_size;
778 } while (read_seqcount_retry(&inode->i_size_seqcount, seq));
779 return i_size;
780#elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
781 loff_t i_size;
782
783 preempt_disable();
784 i_size = inode->i_size;
785 preempt_enable();
786 return i_size;
787#else
788 return inode->i_size;
789#endif
790}
791
Miklos Szeredi7762f5a2006-10-17 00:10:07 -0700792/*
793 * NOTE: unlike i_size_read(), i_size_write() does need locking around it
794 * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount
795 * can be lost, resulting in subsequent i_size_read() calls spinning forever.
796 */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700797static inline void i_size_write(struct inode *inode, loff_t i_size)
798{
799#if BITS_PER_LONG==32 && defined(CONFIG_SMP)
Fan Du74e3d1e2013-04-30 15:27:27 -0700800 preempt_disable();
Linus Torvalds1da177e2005-04-16 15:20:36 -0700801 write_seqcount_begin(&inode->i_size_seqcount);
802 inode->i_size = i_size;
803 write_seqcount_end(&inode->i_size_seqcount);
Fan Du74e3d1e2013-04-30 15:27:27 -0700804 preempt_enable();
Linus Torvalds1da177e2005-04-16 15:20:36 -0700805#elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
806 preempt_disable();
807 inode->i_size = i_size;
808 preempt_enable();
809#else
810 inode->i_size = i_size;
811#endif
812}
813
Jan Engelhardt48ed2142006-12-06 20:35:37 -0800814static inline unsigned iminor(const struct inode *inode)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700815{
816 return MINOR(inode->i_rdev);
817}
818
Jan Engelhardt48ed2142006-12-06 20:35:37 -0800819static inline unsigned imajor(const struct inode *inode)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700820{
821 return MAJOR(inode->i_rdev);
822}
823
824extern struct block_device *I_BDEV(struct inode *inode);
825
826struct fown_struct {
827 rwlock_t lock; /* protects pid, uid, euid fields */
Eric W. Biederman609d7fa2006-10-02 02:17:15 -0700828 struct pid *pid; /* pid or -pgrp where SIGIO should be sent */
829 enum pid_type pid_type; /* Kind of process group SIGIO should be sent to */
Eric W. Biederman92361632012-02-08 07:07:50 -0800830 kuid_t uid, euid; /* uid/euid of process setting the owner */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700831 int signum; /* posix.1b rt signal to be delivered on IO */
832};
833
834/*
835 * Track a single file's readahead state
836 */
837struct file_ra_state {
Fengguang Wu937085a2007-10-16 01:24:31 -0700838 pgoff_t start; /* where readahead started */
839 unsigned int size; /* # of readahead pages */
840 unsigned int async_size; /* do asynchronous readahead when
Fengguang Wuf9acc8c2007-07-19 01:48:08 -0700841 there are only # of pages ahead */
Fengguang Wu5ce11102007-07-19 01:47:59 -0700842
Fengguang Wu937085a2007-10-16 01:24:31 -0700843 unsigned int ra_pages; /* Maximum readahead window */
Wu Fengguang1ebf26a2009-06-16 15:31:19 -0700844 unsigned int mmap_miss; /* Cache miss stat for mmap accesses */
Fengguang Wuf4e6b492007-10-16 01:24:33 -0700845 loff_t prev_pos; /* Cache last read() position */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700846};
Linus Torvalds1da177e2005-04-16 15:20:36 -0700847
Fengguang Wu5ce11102007-07-19 01:47:59 -0700848/*
Fengguang Wu5ce11102007-07-19 01:47:59 -0700849 * Check if @index falls in the readahead windows.
850 */
851static inline int ra_has_index(struct file_ra_state *ra, pgoff_t index)
852{
Fengguang Wuf9acc8c2007-07-19 01:48:08 -0700853 return (index >= ra->start &&
854 index < ra->start + ra->size);
Fengguang Wu5ce11102007-07-19 01:47:59 -0700855}
856
Linus Torvalds1da177e2005-04-16 15:20:36 -0700857struct file {
Eric Dumazet2f512012005-10-30 15:02:16 -0800858 union {
Oleg Nesterov4f5e65a2013-07-08 14:24:16 -0700859 struct llist_node fu_llist;
Eric Dumazet2f512012005-10-30 15:02:16 -0800860 struct rcu_head fu_rcuhead;
861 } f_u;
Josef "Jeff" Sipek0f7fc9e2006-12-08 02:36:35 -0800862 struct path f_path;
Al Virodd379782013-03-01 19:48:30 -0500863 struct inode *f_inode; /* cached value */
Arjan van de Ven99ac48f2006-03-28 01:56:41 -0800864 const struct file_operations *f_op;
Andi Kleenef3d0fd2011-09-15 16:06:48 -0700865
866 /*
Linus Torvalds9c225f22014-03-03 09:36:58 -0800867 * Protects f_ep_links, f_flags.
Andi Kleenef3d0fd2011-09-15 16:06:48 -0700868 * Must not be taken from IRQ context.
869 */
870 spinlock_t f_lock;
Jens Axboec75b1d92017-06-27 11:47:04 -0600871 enum rw_hint f_write_hint;
Al Viro516e0cc2008-07-26 00:39:17 -0400872 atomic_long_t f_count;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700873 unsigned int f_flags;
Al Viroaeb5d722008-09-02 15:28:45 -0400874 fmode_t f_mode;
Linus Torvalds9c225f22014-03-03 09:36:58 -0800875 struct mutex f_pos_lock;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700876 loff_t f_pos;
877 struct fown_struct f_owner;
David Howellsd76b0d92008-11-14 10:39:25 +1100878 const struct cred *f_cred;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700879 struct file_ra_state f_ra;
880
Mathieu Desnoyers2b47c362007-10-16 23:27:21 -0700881 u64 f_version;
Alexey Dobriyan50462062006-09-29 02:00:01 -0700882#ifdef CONFIG_SECURITY
Linus Torvalds1da177e2005-04-16 15:20:36 -0700883 void *f_security;
Alexey Dobriyan50462062006-09-29 02:00:01 -0700884#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -0700885 /* needed for tty driver, and maybe others */
886 void *private_data;
887
888#ifdef CONFIG_EPOLL
889 /* Used by fs/eventpoll.c to link all the hooks to this file */
890 struct list_head f_ep_links;
Jason Baron28d82dc2012-01-12 17:17:43 -0800891 struct list_head f_tfile_llink;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700892#endif /* #ifdef CONFIG_EPOLL */
893 struct address_space *f_mapping;
Jeff Layton5660e132017-07-06 07:02:25 -0400894 errseq_t f_wb_err;
Kees Cook3859a272016-10-28 01:22:25 -0700895} __randomize_layout
896 __attribute__((aligned(4))); /* lest something weird decides that 2 is OK */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700897
Aneesh Kumar K.V990d6c22011-01-29 18:43:26 +0530898struct file_handle {
899 __u32 handle_bytes;
900 int handle_type;
901 /* file identifier */
902 unsigned char f_handle[0];
903};
904
Al Virocb0942b2012-08-27 14:48:26 -0400905static inline struct file *get_file(struct file *f)
906{
907 atomic_long_inc(&f->f_count);
908 return f;
909}
Konstantin Khlebnikov90f31d02015-04-16 12:47:56 -0700910#define get_file_rcu(x) atomic_long_inc_not_zero(&(x)->f_count)
Al Virod7065da2010-05-26 15:13:55 -0400911#define fput_atomic(x) atomic_long_add_unless(&(x)->f_count, -1, 1)
Al Viro516e0cc2008-07-26 00:39:17 -0400912#define file_count(x) atomic_long_read(&(x)->f_count)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700913
914#define MAX_NON_LFS ((1UL<<31) - 1)
915
916/* Page cache limit. The filesystems should put that into their s_maxbytes
917 limits, otherwise bad things can happen in VM. */
918#if BITS_PER_LONG==32
Linus Torvalds0cc3b0e2017-08-27 12:12:25 -0700919#define MAX_LFS_FILESIZE ((loff_t)ULONG_MAX << PAGE_SHIFT)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700920#elif BITS_PER_LONG==64
Linus Torvalds0cc3b0e2017-08-27 12:12:25 -0700921#define MAX_LFS_FILESIZE ((loff_t)LLONG_MAX)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700922#endif
923
924#define FL_POSIX 1
925#define FL_FLOCK 2
J. Bruce Fields617588d2011-07-01 15:18:34 -0400926#define FL_DELEG 4 /* NFSv4 delegation */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700927#define FL_ACCESS 8 /* not trying to lock, just looking */
Trond Myklebustf475ae92006-06-29 16:38:32 -0400928#define FL_EXISTS 16 /* when unlocking, test for existence */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700929#define FL_LEASE 32 /* lease held on this file */
Miklos Szeredi75e1fcc2006-06-23 02:05:12 -0700930#define FL_CLOSE 64 /* unlock on close */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700931#define FL_SLEEP 128 /* A blocking lock */
J. Bruce Fields778fc542011-07-26 18:25:49 -0400932#define FL_DOWNGRADE_PENDING 256 /* Lease is being downgraded */
933#define FL_UNLOCK_PENDING 512 /* Lease is being broken */
Jeff Laytoncff2fce2014-04-22 08:24:32 -0400934#define FL_OFDLCK 1024 /* lock is "owned" by struct file */
Christoph Hellwig11afe9f2015-01-21 19:17:03 +0100935#define FL_LAYOUT 2048 /* outstanding pNFS layout */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700936
Benjamin Coddington50f21122017-04-11 12:50:09 -0400937#define FL_CLOSE_POSIX (FL_POSIX | FL_CLOSE)
938
Linus Torvalds1da177e2005-04-16 15:20:36 -0700939/*
Miklos Szeredibde74e42008-07-25 01:48:57 -0700940 * Special return value from posix_lock_file() and vfs_lock_file() for
941 * asynchronous locking.
942 */
943#define FILE_LOCK_DEFERRED 1
944
Jeff Layton7ca76312014-09-01 19:04:48 -0400945/* legacy typedef, should eventually be removed */
Christoph Hellwig17fa3882014-07-13 17:00:37 +0200946typedef void *fl_owner_t;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700947
Jeff Laytona7231a92015-01-16 15:05:56 -0500948struct file_lock;
949
Linus Torvalds1da177e2005-04-16 15:20:36 -0700950struct file_lock_operations {
Linus Torvalds1da177e2005-04-16 15:20:36 -0700951 void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
952 void (*fl_release_private)(struct file_lock *);
953};
954
955struct lock_manager_operations {
J. Bruce Fields8fb47a42011-07-20 20:21:59 -0400956 int (*lm_compare_owner)(struct file_lock *, struct file_lock *);
Jeff Layton3999e492013-06-21 08:58:19 -0400957 unsigned long (*lm_owner_key)(struct file_lock *);
Jeff Laytoncae80b32015-04-03 09:04:04 -0400958 fl_owner_t (*lm_get_owner)(fl_owner_t);
959 void (*lm_put_owner)(fl_owner_t);
J. Bruce Fields8fb47a42011-07-20 20:21:59 -0400960 void (*lm_notify)(struct file_lock *); /* unblock callback */
Joe Perchesd0449b92014-08-22 10:18:42 -0400961 int (*lm_grant)(struct file_lock *, int);
Jeff Layton4d01b7f2014-09-01 15:06:54 -0400962 bool (*lm_break)(struct file_lock *);
Jeff Layton7448cc32015-01-16 15:05:57 -0500963 int (*lm_change)(struct file_lock *, int, struct list_head *);
Jeff Layton1c7dd2f2014-08-22 10:55:47 -0400964 void (*lm_setup)(struct file_lock *, void **);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700965};
966
J. Bruce Fieldsaf558e32007-09-06 12:34:25 -0400967struct lock_manager {
968 struct list_head list;
J. Bruce Fieldsc87fb4a2015-08-06 12:47:02 -0400969 /*
970 * NFSv4 and up also want opens blocked during the grace period;
971 * NLM doesn't care:
972 */
973 bool block_opens;
J. Bruce Fieldsaf558e32007-09-06 12:34:25 -0400974};
975
Stanislav Kinsbursky5ccb0062012-07-25 16:57:22 +0400976struct net;
977void locks_start_grace(struct net *, struct lock_manager *);
J. Bruce Fieldsaf558e32007-09-06 12:34:25 -0400978void locks_end_grace(struct lock_manager *);
Corentin Labbe003278e2017-09-26 09:14:07 +0200979bool locks_in_grace(struct net *);
980bool opens_in_grace(struct net *);
J. Bruce Fieldsaf558e32007-09-06 12:34:25 -0400981
Linus Torvalds1da177e2005-04-16 15:20:36 -0700982/* that will die - we need it for nfs_lock_info */
983#include <linux/nfs_fs_i.h>
984
Jeff Layton1cb36012013-06-21 08:58:12 -0400985/*
986 * struct file_lock represents a generic "file lock". It's used to represent
987 * POSIX byte range locks, BSD (flock) locks, and leases. It's important to
988 * note that the same struct is used to represent both a request for a lock and
989 * the lock itself, but the same object is never used for both.
990 *
991 * FIXME: should we create a separate "struct lock_request" to help distinguish
992 * these two uses?
993 *
Jeff Layton8116bf42015-01-21 20:44:01 -0500994 * The varous i_flctx lists are ordered by:
Jeff Layton1cb36012013-06-21 08:58:12 -0400995 *
Jeff Layton8116bf42015-01-21 20:44:01 -0500996 * 1) lock owner
997 * 2) lock range start
998 * 3) lock range end
Jeff Layton1cb36012013-06-21 08:58:12 -0400999 *
1000 * Obviously, the last two criteria only matter for POSIX locks.
1001 */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001002struct file_lock {
1003 struct file_lock *fl_next; /* singly linked list for this inode */
Jeff Layton6dee60f2015-01-16 15:05:54 -05001004 struct list_head fl_list; /* link into file_lock_context */
Jeff Layton139ca042013-06-21 08:58:17 -04001005 struct hlist_node fl_link; /* node in global lists */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001006 struct list_head fl_block; /* circular list of blocked processes */
1007 fl_owner_t fl_owner;
J. Bruce Fields710b7212011-07-26 16:28:29 -04001008 unsigned int fl_flags;
Richard Kennedyafc12462008-07-11 17:20:49 -07001009 unsigned char fl_type;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001010 unsigned int fl_pid;
Jeff Layton7012b022013-06-21 08:58:22 -04001011 int fl_link_cpu; /* what cpu's list is this on? */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001012 wait_queue_head_t fl_wait;
1013 struct file *fl_file;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001014 loff_t fl_start;
1015 loff_t fl_end;
1016
1017 struct fasync_struct * fl_fasync; /* for lease break notifications */
J. Bruce Fields778fc542011-07-26 18:25:49 -04001018 /* for lease breaks: */
1019 unsigned long fl_break_time;
1020 unsigned long fl_downgrade_time;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001021
Alexey Dobriyan6aed6282009-09-21 17:01:11 -07001022 const struct file_lock_operations *fl_ops; /* Callbacks for filesystems */
Alexey Dobriyan7b021962009-09-21 17:01:12 -07001023 const struct lock_manager_operations *fl_lmops; /* Callbacks for lockmanagers */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001024 union {
1025 struct nfs_lock_info nfs_fl;
Trond Myklebust8d0a8a92005-06-22 17:16:32 +00001026 struct nfs4_lock_info nfs4_fl;
David Howellse8d6c552007-07-15 23:40:12 -07001027 struct {
1028 struct list_head link; /* link in AFS vnode's pending_locks list */
1029 int state; /* state of grant or error if -ve */
1030 } afs;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001031 } fl_u;
Kees Cook3859a272016-10-28 01:22:25 -07001032} __randomize_layout;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001033
Jeff Layton4a075e32015-01-16 15:05:54 -05001034struct file_lock_context {
Jeff Layton6109c852015-01-16 15:05:57 -05001035 spinlock_t flc_lock;
Jeff Layton4a075e32015-01-16 15:05:54 -05001036 struct list_head flc_flock;
Jeff Laytonbd61e0a2015-01-16 15:05:55 -05001037 struct list_head flc_posix;
Jeff Layton8634b512015-01-16 15:05:55 -05001038 struct list_head flc_lease;
Jeff Layton4a075e32015-01-16 15:05:54 -05001039};
1040
Linus Torvalds1da177e2005-04-16 15:20:36 -07001041/* The following constant reflects the upper bound of the file/locking space */
1042#ifndef OFFSET_MAX
1043#define INT_LIMIT(x) (~((x)1 << (sizeof(x)*8 - 1)))
1044#define OFFSET_MAX INT_LIMIT(loff_t)
1045#define OFFT_OFFSET_MAX INT_LIMIT(off_t)
1046#endif
1047
Thomas Petazzonibfcd17a2008-08-06 15:12:22 +02001048extern void send_sigio(struct fown_struct *fown, int fd, int band);
1049
Miklos Szeredic568d682016-09-16 12:44:20 +02001050/*
1051 * Return the inode to use for locking
1052 *
1053 * For overlayfs this should be the overlay inode, not the real inode returned
1054 * by file_inode(). For any other fs file_inode(filp) and locks_inode(filp) are
1055 * equal.
1056 */
1057static inline struct inode *locks_inode(const struct file *f)
1058{
1059 return f->f_path.dentry->d_inode;
1060}
1061
Thomas Petazzonibfcd17a2008-08-06 15:12:22 +02001062#ifdef CONFIG_FILE_LOCKING
Christoph Hellwiga75d30c2017-05-27 06:07:19 -04001063extern int fcntl_getlk(struct file *, unsigned int, struct flock *);
Peter Staubachc2936212005-07-27 11:45:09 -07001064extern int fcntl_setlk(unsigned int, struct file *, unsigned int,
Christoph Hellwiga75d30c2017-05-27 06:07:19 -04001065 struct flock *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001066
1067#if BITS_PER_LONG == 32
Christoph Hellwiga75d30c2017-05-27 06:07:19 -04001068extern int fcntl_getlk64(struct file *, unsigned int, struct flock64 *);
Peter Staubachc2936212005-07-27 11:45:09 -07001069extern int fcntl_setlk64(unsigned int, struct file *, unsigned int,
Christoph Hellwiga75d30c2017-05-27 06:07:19 -04001070 struct flock64 *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001071#endif
1072
Linus Torvalds1da177e2005-04-16 15:20:36 -07001073extern int fcntl_setlease(unsigned int fd, struct file *filp, long arg);
1074extern int fcntl_getlease(struct file *filp);
1075
1076/* fs/locks.c */
Jeff Laytonf27a0fe2016-01-07 15:08:51 -05001077void locks_free_lock_context(struct inode *inode);
J. Bruce Fields05fa3132010-10-30 17:31:15 -04001078void locks_free_lock(struct file_lock *fl);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001079extern void locks_init_lock(struct file_lock *);
Arnd Bergmannc5b1f0d2010-10-27 15:46:08 +02001080extern struct file_lock * locks_alloc_lock(void);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001081extern void locks_copy_lock(struct file_lock *, struct file_lock *);
Kinglong Mee3fe0fff12014-08-22 10:18:42 -04001082extern void locks_copy_conflock(struct file_lock *, struct file_lock *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001083extern void locks_remove_posix(struct file *, fl_owner_t);
Jeff Layton78ed8a12014-02-03 12:13:08 -05001084extern void locks_remove_file(struct file *);
Felix Blyakhera9e61e22009-03-31 15:12:56 -05001085extern void locks_release_private(struct file_lock *);
J. Bruce Fields6d34ac12007-05-11 16:09:32 -04001086extern void posix_test_lock(struct file *, struct file_lock *);
Marc Eshel150b3932007-01-18 16:15:35 -05001087extern int posix_lock_file(struct file *, struct file_lock *, struct file_lock *);
Jeff Laytonf891a292013-06-21 08:58:09 -04001088extern int posix_unblock_lock(struct file_lock *);
J. Bruce Fields3ee17ab2007-02-21 00:58:50 -05001089extern int vfs_test_lock(struct file *, struct file_lock *);
Marc Eshel150b3932007-01-18 16:15:35 -05001090extern int vfs_lock_file(struct file *, unsigned int, struct file_lock *, struct file_lock *);
Marc Eshel9b9d2ab2007-01-18 17:52:58 -05001091extern int vfs_cancel_lock(struct file *filp, struct file_lock *fl);
Benjamin Coddingtone55c34a2015-10-22 13:38:13 -04001092extern int locks_lock_inode_wait(struct inode *inode, struct file_lock *fl);
J. Bruce Fieldsdf4e8d22012-03-05 13:18:59 -05001093extern int __break_lease(struct inode *inode, unsigned int flags, unsigned int type);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001094extern void lease_get_mtime(struct inode *, struct timespec *time);
Jeff Laytone6f5c782014-08-22 10:40:25 -04001095extern int generic_setlease(struct file *, long, struct file_lock **, void **priv);
1096extern int vfs_setlease(struct file *, long, struct file_lock **, void **);
Jeff Layton7448cc32015-01-16 15:05:57 -05001097extern int lease_modify(struct file_lock *, int, struct list_head *);
Andrey Vagin6c8c9032015-04-16 12:49:38 -07001098struct files_struct;
1099extern void show_fd_locks(struct seq_file *f,
1100 struct file *filp, struct files_struct *files);
Thomas Petazzonibfcd17a2008-08-06 15:12:22 +02001101#else /* !CONFIG_FILE_LOCKING */
Jeff Laytonc1e62b82014-02-03 12:13:09 -05001102static inline int fcntl_getlk(struct file *file, unsigned int cmd,
1103 struct flock __user *user)
Steven Whitehousec2aca5e2009-01-20 10:29:45 +00001104{
1105 return -EINVAL;
1106}
1107
1108static inline int fcntl_setlk(unsigned int fd, struct file *file,
1109 unsigned int cmd, struct flock __user *user)
1110{
1111 return -EACCES;
1112}
1113
Thomas Petazzonibfcd17a2008-08-06 15:12:22 +02001114#if BITS_PER_LONG == 32
Jeff Laytonc1e62b82014-02-03 12:13:09 -05001115static inline int fcntl_getlk64(struct file *file, unsigned int cmd,
1116 struct flock64 __user *user)
Steven Whitehousec2aca5e2009-01-20 10:29:45 +00001117{
1118 return -EINVAL;
1119}
1120
1121static inline int fcntl_setlk64(unsigned int fd, struct file *file,
1122 unsigned int cmd, struct flock64 __user *user)
1123{
1124 return -EACCES;
1125}
Thomas Petazzonibfcd17a2008-08-06 15:12:22 +02001126#endif
Steven Whitehousec2aca5e2009-01-20 10:29:45 +00001127static inline int fcntl_setlease(unsigned int fd, struct file *filp, long arg)
1128{
Jeff Layton1b2b32d2014-09-24 08:38:44 -04001129 return -EINVAL;
Steven Whitehousec2aca5e2009-01-20 10:29:45 +00001130}
1131
1132static inline int fcntl_getlease(struct file *filp)
1133{
Jeff Layton1b2b32d2014-09-24 08:38:44 -04001134 return F_UNLCK;
Steven Whitehousec2aca5e2009-01-20 10:29:45 +00001135}
1136
Jeff Layton4a075e32015-01-16 15:05:54 -05001137static inline void
Jeff Laytonf27a0fe2016-01-07 15:08:51 -05001138locks_free_lock_context(struct inode *inode)
Jeff Layton4a075e32015-01-16 15:05:54 -05001139{
1140}
1141
Steven Whitehousec2aca5e2009-01-20 10:29:45 +00001142static inline void locks_init_lock(struct file_lock *fl)
1143{
1144 return;
1145}
1146
Kinglong Mee3fe0fff12014-08-22 10:18:42 -04001147static inline void locks_copy_conflock(struct file_lock *new, struct file_lock *fl)
Steven Whitehousec2aca5e2009-01-20 10:29:45 +00001148{
1149 return;
1150}
1151
1152static inline void locks_copy_lock(struct file_lock *new, struct file_lock *fl)
1153{
1154 return;
1155}
1156
1157static inline void locks_remove_posix(struct file *filp, fl_owner_t owner)
1158{
1159 return;
1160}
1161
Jeff Layton78ed8a12014-02-03 12:13:08 -05001162static inline void locks_remove_file(struct file *filp)
Steven Whitehousec2aca5e2009-01-20 10:29:45 +00001163{
1164 return;
1165}
1166
1167static inline void posix_test_lock(struct file *filp, struct file_lock *fl)
1168{
1169 return;
1170}
1171
1172static inline int posix_lock_file(struct file *filp, struct file_lock *fl,
1173 struct file_lock *conflock)
1174{
1175 return -ENOLCK;
1176}
1177
Jeff Laytonf891a292013-06-21 08:58:09 -04001178static inline int posix_unblock_lock(struct file_lock *waiter)
Steven Whitehousec2aca5e2009-01-20 10:29:45 +00001179{
1180 return -ENOENT;
1181}
1182
1183static inline int vfs_test_lock(struct file *filp, struct file_lock *fl)
1184{
1185 return 0;
1186}
1187
1188static inline int vfs_lock_file(struct file *filp, unsigned int cmd,
1189 struct file_lock *fl, struct file_lock *conf)
1190{
1191 return -ENOLCK;
1192}
1193
1194static inline int vfs_cancel_lock(struct file *filp, struct file_lock *fl)
1195{
1196 return 0;
1197}
1198
Benjamin Coddingtone55c34a2015-10-22 13:38:13 -04001199static inline int locks_lock_inode_wait(struct inode *inode, struct file_lock *fl)
1200{
1201 return -ENOLCK;
1202}
1203
J. Bruce Fieldsdf4e8d22012-03-05 13:18:59 -05001204static inline int __break_lease(struct inode *inode, unsigned int mode, unsigned int type)
Steven Whitehousec2aca5e2009-01-20 10:29:45 +00001205{
1206 return 0;
1207}
1208
1209static inline void lease_get_mtime(struct inode *inode, struct timespec *time)
1210{
1211 return;
1212}
1213
1214static inline int generic_setlease(struct file *filp, long arg,
Jeff Laytone6f5c782014-08-22 10:40:25 -04001215 struct file_lock **flp, void **priv)
Steven Whitehousec2aca5e2009-01-20 10:29:45 +00001216{
1217 return -EINVAL;
1218}
1219
1220static inline int vfs_setlease(struct file *filp, long arg,
Jeff Laytone6f5c782014-08-22 10:40:25 -04001221 struct file_lock **lease, void **priv)
Steven Whitehousec2aca5e2009-01-20 10:29:45 +00001222{
1223 return -EINVAL;
1224}
1225
Jeff Layton7448cc32015-01-16 15:05:57 -05001226static inline int lease_modify(struct file_lock *fl, int arg,
Jeff Laytonc45198e2014-09-01 07:12:07 -04001227 struct list_head *dispose)
Steven Whitehousec2aca5e2009-01-20 10:29:45 +00001228{
1229 return -EINVAL;
1230}
Andrey Vagin6c8c9032015-04-16 12:49:38 -07001231
1232struct files_struct;
1233static inline void show_fd_locks(struct seq_file *f,
1234 struct file *filp, struct files_struct *files) {}
Thomas Petazzonibfcd17a2008-08-06 15:12:22 +02001235#endif /* !CONFIG_FILE_LOCKING */
1236
Jeff Laytonee296d72015-07-11 06:43:03 -04001237static inline struct inode *file_inode(const struct file *f)
1238{
1239 return f->f_inode;
1240}
1241
Miklos Szeredid101a122016-03-26 16:14:37 -04001242static inline struct dentry *file_dentry(const struct file *file)
1243{
Miklos Szeredi495e6422017-09-04 21:42:22 +02001244 return d_real(file->f_path.dentry, file_inode(file), 0, 0);
Miklos Szeredid101a122016-03-26 16:14:37 -04001245}
1246
Benjamin Coddingtone55c34a2015-10-22 13:38:13 -04001247static inline int locks_lock_file_wait(struct file *filp, struct file_lock *fl)
1248{
Miklos Szeredic568d682016-09-16 12:44:20 +02001249 return locks_lock_inode_wait(locks_inode(filp), fl);
Benjamin Coddingtone55c34a2015-10-22 13:38:13 -04001250}
1251
Linus Torvalds1da177e2005-04-16 15:20:36 -07001252struct fasync_struct {
Kirill Tkhai7a107c02018-04-05 14:58:06 +03001253 rwlock_t fa_lock;
Eric Dumazet989a2972010-04-14 09:55:35 +00001254 int magic;
1255 int fa_fd;
1256 struct fasync_struct *fa_next; /* singly linked list */
1257 struct file *fa_file;
1258 struct rcu_head fa_rcu;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001259};
1260
1261#define FASYNC_MAGIC 0x4601
1262
1263/* SMP safe fasync helpers: */
1264extern int fasync_helper(int, struct file *, int, struct fasync_struct **);
Linus Torvaldsf7347ce2010-10-27 12:38:12 -04001265extern struct fasync_struct *fasync_insert_entry(int, struct file *, struct fasync_struct **, struct fasync_struct *);
1266extern int fasync_remove_entry(struct file *, struct fasync_struct **);
1267extern struct fasync_struct *fasync_alloc(void);
1268extern void fasync_free(struct fasync_struct *);
1269
Linus Torvalds1da177e2005-04-16 15:20:36 -07001270/* can be called from interrupts */
1271extern void kill_fasync(struct fasync_struct **, int, int);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001272
Jeff Laytone0b93ed2014-08-22 11:27:32 -04001273extern void __f_setown(struct file *filp, struct pid *, enum pid_type, int force);
Jiri Slaby393cc3f2017-06-13 13:35:50 +02001274extern int f_setown(struct file *filp, unsigned long arg, int force);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001275extern void f_delown(struct file *filp);
Eric W. Biederman609d7fa2006-10-02 02:17:15 -07001276extern pid_t f_getown(struct file *filp);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001277extern int send_sigurg(struct fown_struct *fown);
1278
1279/*
David Howellse462ec52017-07-17 08:45:35 +01001280 * sb->s_flags. Note that these mirror the equivalent MS_* flags where
1281 * represented in both.
1282 */
1283#define SB_RDONLY 1 /* Mount read-only */
1284#define SB_NOSUID 2 /* Ignore suid and sgid bits */
1285#define SB_NODEV 4 /* Disallow access to device special files */
1286#define SB_NOEXEC 8 /* Disallow program execution */
1287#define SB_SYNCHRONOUS 16 /* Writes are synced at once */
1288#define SB_MANDLOCK 64 /* Allow mandatory locks on an FS */
1289#define SB_DIRSYNC 128 /* Directory modifications are synchronous */
1290#define SB_NOATIME 1024 /* Do not update access times. */
1291#define SB_NODIRATIME 2048 /* Do not update directory access times */
1292#define SB_SILENT 32768
1293#define SB_POSIXACL (1<<16) /* VFS does not apply the umask */
1294#define SB_KERNMOUNT (1<<22) /* this is a kern_mount call */
1295#define SB_I_VERSION (1<<23) /* Update inode I_version field */
1296#define SB_LAZYTIME (1<<25) /* Update the on-disk [acm]times lazily */
1297
1298/* These sb flags are internal to the kernel */
1299#define SB_SUBMOUNT (1<<26)
1300#define SB_NOREMOTELOCK (1<<27)
1301#define SB_NOSEC (1<<28)
1302#define SB_BORN (1<<29)
1303#define SB_ACTIVE (1<<30)
1304#define SB_NOUSER (1<<31)
1305
1306/*
Linus Torvalds1da177e2005-04-16 15:20:36 -07001307 * Umount options
1308 */
1309
1310#define MNT_FORCE 0x00000001 /* Attempt to forcibily umount */
1311#define MNT_DETACH 0x00000002 /* Just detach from the tree */
1312#define MNT_EXPIRE 0x00000004 /* Mark for expiry */
Miklos Szeredidb1f05b2010-02-10 12:15:53 +01001313#define UMOUNT_NOFOLLOW 0x00000008 /* Don't follow symlink on umount */
1314#define UMOUNT_UNUSED 0x80000000 /* Flag guaranteed to be unused */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001315
Tejun Heo46b15ca2015-06-16 18:48:31 -04001316/* sb->s_iflags */
1317#define SB_I_CGROUPWB 0x00000001 /* cgroup-aware writeback enabled */
Eric W. Biederman90f85722015-06-29 14:42:03 -05001318#define SB_I_NOEXEC 0x00000002 /* Ignore executables on this fs */
Eric W. Biedermana2982cc2016-06-09 15:34:02 -05001319#define SB_I_NODEV 0x00000004 /* Ignore devices on this fs */
Eric W. Biederman95dd7752018-03-14 18:20:29 -05001320#define SB_I_MULTIROOT 0x00000008 /* Multiple roots to the dentry tree */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001321
Eric W. Biederman8654df42016-06-09 16:06:06 -05001322/* sb->s_iflags to limit user namespace mounts */
1323#define SB_I_USERNS_VISIBLE 0x00000010 /* fstype already mounted */
Mimi Zohar57b56ac2018-02-21 11:33:37 -05001324#define SB_I_IMA_UNVERIFIABLE_SIGNATURE 0x00000020
1325#define SB_I_UNTRUSTED_MOUNTER 0x00000040
Linus Torvalds1da177e2005-04-16 15:20:36 -07001326
Jan Kara5accdf82012-06-12 16:20:34 +02001327/* Possible states of 'frozen' field */
1328enum {
1329 SB_UNFROZEN = 0, /* FS is unfrozen */
1330 SB_FREEZE_WRITE = 1, /* Writes, dir ops, ioctls frozen */
Jan Kara5accdf82012-06-12 16:20:34 +02001331 SB_FREEZE_PAGEFAULT = 2, /* Page faults stopped as well */
1332 SB_FREEZE_FS = 3, /* For internal FS use (e.g. to stop
1333 * internal threads if needed) */
1334 SB_FREEZE_COMPLETE = 4, /* ->freeze_fs finished successfully */
1335};
1336
1337#define SB_FREEZE_LEVELS (SB_FREEZE_COMPLETE - 1)
1338
1339struct sb_writers {
Oleg Nesterov8129ed22015-08-11 17:05:04 +02001340 int frozen; /* Is sb frozen? */
1341 wait_queue_head_t wait_unfrozen; /* for get_super_thawed() */
1342 struct percpu_rw_semaphore rw_sem[SB_FREEZE_LEVELS];
Jan Kara5accdf82012-06-12 16:20:34 +02001343};
1344
Linus Torvalds1da177e2005-04-16 15:20:36 -07001345struct super_block {
1346 struct list_head s_list; /* Keep this first */
1347 dev_t s_dev; /* search index; _not_ kdev_t */
Richard Kennedy270ba5f2010-01-26 14:12:43 +00001348 unsigned char s_blocksize_bits;
1349 unsigned long s_blocksize;
Jeff Layton42cb56a2009-09-18 13:05:53 -07001350 loff_t s_maxbytes; /* Max file size */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001351 struct file_system_type *s_type;
Josef 'Jeff' Sipekee9b6d62007-02-12 00:55:41 -08001352 const struct super_operations *s_op;
Alexey Dobriyan61e225d2009-09-21 17:01:08 -07001353 const struct dquot_operations *dq_op;
Alexey Dobriyan0d54b212009-09-21 17:01:09 -07001354 const struct quotactl_ops *s_qcop;
Christoph Hellwig39655162007-10-21 16:42:17 -07001355 const struct export_operations *s_export_op;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001356 unsigned long s_flags;
Tejun Heo46b15ca2015-06-16 18:48:31 -04001357 unsigned long s_iflags; /* internal SB_I_* flags */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001358 unsigned long s_magic;
1359 struct dentry *s_root;
1360 struct rw_semaphore s_umount;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001361 int s_count;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001362 atomic_t s_active;
Alexey Dobriyan50462062006-09-29 02:00:01 -07001363#ifdef CONFIG_SECURITY
Linus Torvalds1da177e2005-04-16 15:20:36 -07001364 void *s_security;
Alexey Dobriyan50462062006-09-29 02:00:01 -07001365#endif
Stephen Hemmingerbb435452010-05-13 17:53:14 -07001366 const struct xattr_handler **s_xattr;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001367
Jaegeuk Kim0b81d072015-05-15 16:26:10 -07001368 const struct fscrypt_operations *s_cop;
1369
NeilBrownf1ee6162017-12-21 09:45:40 +11001370 struct hlist_bl_head s_roots; /* alternate root dentries for NFS */
Miklos Szeredi39f7c4d2011-11-21 12:11:30 +01001371 struct list_head s_mounts; /* list of mounts; _not_ for fs use */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001372 struct block_device *s_bdev;
Jens Axboe32a88aa2009-09-16 15:02:33 +02001373 struct backing_dev_info *s_bdi;
David Howellsacaebfd2007-05-10 22:51:50 -07001374 struct mtd_info *s_mtd;
Al Viroa5166162011-12-12 22:53:00 -05001375 struct hlist_node s_instances;
Jan Kara2c5f6482014-09-30 10:43:09 +02001376 unsigned int s_quota_types; /* Bitmask of supported quota types */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001377 struct quota_info s_dquot; /* Diskquota specific options */
1378
Jan Kara5accdf82012-06-12 16:20:34 +02001379 struct sb_writers s_writers;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001380
Christoph Hellwig85787092017-05-10 15:06:33 +02001381 char s_id[32]; /* Informational name */
1382 uuid_t s_uuid; /* UUID */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001383
1384 void *s_fs_info; /* Filesystem private info */
Al Viro8de52772012-02-06 12:45:27 -05001385 unsigned int s_max_links;
Al Viro30c40d22008-02-22 19:50:45 -05001386 fmode_t s_mode;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001387
Richard Kennedy270ba5f2010-01-26 14:12:43 +00001388 /* Granularity of c/m/atime in ns.
1389 Cannot be worse than a second */
1390 u32 s_time_gran;
1391
Linus Torvalds1da177e2005-04-16 15:20:36 -07001392 /*
1393 * The next field is for VFS *only*. No filesystems have any business
1394 * even looking at it. You had been warned.
1395 */
Arjan van de Vena11f3a02006-03-23 03:00:33 -08001396 struct mutex s_vfs_rename_mutex; /* Kludge */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001397
Miklos Szeredi79c0b2d2007-05-08 00:25:43 -07001398 /*
1399 * Filesystem subtype. If non-empty the filesystem type field
1400 * in /proc/mounts will be "type.subtype"
1401 */
1402 char *s_subtype;
Miklos Szeredib3b304a2008-02-08 04:21:35 -08001403
Al Viroc8aebb02010-12-18 10:22:30 -05001404 const struct dentry_operations *s_d_op; /* default d_op for dentries */
Dan Magenheimer9fdfdcf2011-05-26 10:01:19 -06001405
1406 /*
1407 * Saved pool identifier for cleancache (-1 means none)
1408 */
1409 int cleancache_poolid;
Dave Chinnerb0d40c92011-07-08 14:14:42 +10001410
1411 struct shrinker s_shrink; /* per-sb shrinker handle */
Miklos Szeredi4ed5e82f2011-11-21 12:11:31 +01001412
Miklos Szeredi7ada4db2011-11-21 12:11:32 +01001413 /* Number of inodes with nlink == 0 but still referenced */
1414 atomic_long_t s_remove_count;
1415
Miklos Szeredi4ed5e82f2011-11-21 12:11:31 +01001416 /* Being remounted read-only */
1417 int s_readonly_remount;
Christoph Hellwig7b7a8662013-09-04 15:04:39 +02001418
1419 /* AIO completions deferred from interrupt context */
1420 struct workqueue_struct *s_dio_done_wq;
Al Viro215752f2014-08-07 06:23:41 -04001421 struct hlist_head s_pins;
Dave Chinnerf6041562013-08-28 10:18:00 +10001422
1423 /*
Eric W. Biederman6e4eab52016-05-24 09:29:01 -05001424 * Owning user namespace and default context in which to
1425 * interpret filesystem uids, gids, quotas, device nodes,
1426 * xattrs and security labels.
1427 */
1428 struct user_namespace *s_user_ns;
1429
1430 /*
Dave Chinnerf6041562013-08-28 10:18:00 +10001431 * Keep the lru lists last in the structure so they always sit on their
1432 * own individual cachelines.
1433 */
1434 struct list_lru s_dentry_lru ____cacheline_aligned_in_smp;
1435 struct list_lru s_inode_lru ____cacheline_aligned_in_smp;
Al Viroe2fec7c2013-10-04 17:06:56 -04001436 struct rcu_head rcu;
Oleg Nesterov853b39a2015-07-22 20:21:13 +02001437 struct work_struct destroy_work;
Miklos Szeredi69c433e2014-10-24 00:14:39 +02001438
Dave Chinnere97fedb2015-03-04 13:40:00 -05001439 struct mutex s_sync_lock; /* sync serialisation lock */
Miklos Szeredi69c433e2014-10-24 00:14:39 +02001440
1441 /*
1442 * Indicates how deep in a filesystem stack this SB is
1443 */
1444 int s_stack_depth;
Dave Chinner74278da2015-03-04 12:37:22 -05001445
1446 /* s_inode_list_lock protects s_inodes */
1447 spinlock_t s_inode_list_lock ____cacheline_aligned_in_smp;
1448 struct list_head s_inodes; /* all inodes */
Dave Chinner6c60d2b2016-07-26 15:21:50 -07001449
1450 spinlock_t s_inode_wblist_lock;
1451 struct list_head s_inodes_wb; /* writeback inodes */
Kees Cook3859a272016-10-28 01:22:25 -07001452} __randomize_layout;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001453
Seth Forshee81754352014-12-11 10:15:45 -06001454/* Helper functions so that in most cases filesystems will
1455 * not need to deal directly with kuid_t and kgid_t and can
1456 * instead deal with the raw numeric values that are stored
1457 * in the filesystem.
1458 */
1459static inline uid_t i_uid_read(const struct inode *inode)
1460{
1461 return from_kuid(inode->i_sb->s_user_ns, inode->i_uid);
1462}
1463
1464static inline gid_t i_gid_read(const struct inode *inode)
1465{
1466 return from_kgid(inode->i_sb->s_user_ns, inode->i_gid);
1467}
1468
1469static inline void i_uid_write(struct inode *inode, uid_t uid)
1470{
1471 inode->i_uid = make_kuid(inode->i_sb->s_user_ns, uid);
1472}
1473
1474static inline void i_gid_write(struct inode *inode, gid_t gid)
1475{
1476 inode->i_gid = make_kgid(inode->i_sb->s_user_ns, gid);
1477}
1478
Deepa Dinamani3cd88662016-09-14 07:48:02 -07001479extern struct timespec current_time(struct inode *inode);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001480
1481/*
1482 * Snapshotting support.
1483 */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001484
Jan Kara5accdf82012-06-12 16:20:34 +02001485void __sb_end_write(struct super_block *sb, int level);
1486int __sb_start_write(struct super_block *sb, int level, bool wait);
1487
Oleg Nesterovbee91822015-07-19 23:48:20 +02001488#define __sb_writers_acquired(sb, lev) \
Oleg Nesterov8129ed22015-08-11 17:05:04 +02001489 percpu_rwsem_acquire(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
Oleg Nesterovbee91822015-07-19 23:48:20 +02001490#define __sb_writers_release(sb, lev) \
Oleg Nesterov8129ed22015-08-11 17:05:04 +02001491 percpu_rwsem_release(&(sb)->s_writers.rw_sem[(lev)-1], 1, _THIS_IP_)
Oleg Nesterovbee91822015-07-19 23:48:20 +02001492
Jan Kara5accdf82012-06-12 16:20:34 +02001493/**
1494 * sb_end_write - drop write access to a superblock
1495 * @sb: the super we wrote to
1496 *
1497 * Decrement number of writers to the filesystem. Wake up possible waiters
1498 * wanting to freeze the filesystem.
1499 */
1500static inline void sb_end_write(struct super_block *sb)
1501{
1502 __sb_end_write(sb, SB_FREEZE_WRITE);
1503}
1504
1505/**
1506 * sb_end_pagefault - drop write access to a superblock from a page fault
1507 * @sb: the super we wrote to
1508 *
1509 * Decrement number of processes handling write page fault to the filesystem.
1510 * Wake up possible waiters wanting to freeze the filesystem.
1511 */
1512static inline void sb_end_pagefault(struct super_block *sb)
1513{
1514 __sb_end_write(sb, SB_FREEZE_PAGEFAULT);
1515}
1516
1517/**
1518 * sb_end_intwrite - drop write access to a superblock for internal fs purposes
1519 * @sb: the super we wrote to
1520 *
1521 * Decrement fs-internal number of writers to the filesystem. Wake up possible
1522 * waiters wanting to freeze the filesystem.
1523 */
1524static inline void sb_end_intwrite(struct super_block *sb)
1525{
1526 __sb_end_write(sb, SB_FREEZE_FS);
1527}
1528
1529/**
1530 * sb_start_write - get write access to a superblock
1531 * @sb: the super we write to
1532 *
1533 * When a process wants to write data or metadata to a file system (i.e. dirty
1534 * a page or an inode), it should embed the operation in a sb_start_write() -
1535 * sb_end_write() pair to get exclusion against file system freezing. This
1536 * function increments number of writers preventing freezing. If the file
1537 * system is already frozen, the function waits until the file system is
1538 * thawed.
1539 *
1540 * Since freeze protection behaves as a lock, users have to preserve
1541 * ordering of freeze protection and other filesystem locks. Generally,
1542 * freeze protection should be the outermost lock. In particular, we have:
1543 *
1544 * sb_start_write
1545 * -> i_mutex (write path, truncate, directory ops, ...)
1546 * -> s_umount (freeze_super, thaw_super)
1547 */
1548static inline void sb_start_write(struct super_block *sb)
1549{
1550 __sb_start_write(sb, SB_FREEZE_WRITE, true);
1551}
1552
1553static inline int sb_start_write_trylock(struct super_block *sb)
1554{
1555 return __sb_start_write(sb, SB_FREEZE_WRITE, false);
1556}
1557
1558/**
1559 * sb_start_pagefault - get write access to a superblock from a page fault
1560 * @sb: the super we write to
1561 *
1562 * When a process starts handling write page fault, it should embed the
1563 * operation into sb_start_pagefault() - sb_end_pagefault() pair to get
1564 * exclusion against file system freezing. This is needed since the page fault
1565 * is going to dirty a page. This function increments number of running page
1566 * faults preventing freezing. If the file system is already frozen, the
1567 * function waits until the file system is thawed.
1568 *
1569 * Since page fault freeze protection behaves as a lock, users have to preserve
1570 * ordering of freeze protection and other filesystem locks. It is advised to
1571 * put sb_start_pagefault() close to mmap_sem in lock ordering. Page fault
1572 * handling code implies lock dependency:
1573 *
1574 * mmap_sem
1575 * -> sb_start_pagefault
1576 */
1577static inline void sb_start_pagefault(struct super_block *sb)
1578{
1579 __sb_start_write(sb, SB_FREEZE_PAGEFAULT, true);
1580}
1581
1582/*
1583 * sb_start_intwrite - get write access to a superblock for internal fs purposes
1584 * @sb: the super we write to
1585 *
1586 * This is the third level of protection against filesystem freezing. It is
1587 * free for use by a filesystem. The only requirement is that it must rank
1588 * below sb_start_pagefault.
1589 *
1590 * For example filesystem can call sb_start_intwrite() when starting a
1591 * transaction which somewhat eases handling of freezing for internal sources
1592 * of filesystem changes (internal fs threads, discarding preallocation on file
1593 * close, etc.).
1594 */
1595static inline void sb_start_intwrite(struct super_block *sb)
1596{
1597 __sb_start_write(sb, SB_FREEZE_FS, true);
1598}
1599
Linus Torvalds1da177e2005-04-16 15:20:36 -07001600
Serge E. Hallyn2e149672011-03-23 16:43:26 -07001601extern bool inode_owner_or_capable(const struct inode *inode);
Satyam Sharma3bd858a2007-07-17 15:00:08 +05301602
Linus Torvalds1da177e2005-04-16 15:20:36 -07001603/*
1604 * VFS helper functions..
1605 */
Al Viro312b63f2012-06-10 18:09:36 -04001606extern int vfs_create(struct inode *, struct dentry *, umode_t, bool);
Al Viro18bb1db2011-07-26 01:41:39 -04001607extern int vfs_mkdir(struct inode *, struct dentry *, umode_t);
Al Viro1a67aaf2011-07-26 01:52:52 -04001608extern int vfs_mknod(struct inode *, struct dentry *, umode_t, dev_t);
Miklos Szeredidb2e7472008-06-24 16:50:16 +02001609extern int vfs_symlink(struct inode *, struct dentry *, const char *);
J. Bruce Fields146a8592011-09-20 17:14:31 -04001610extern int vfs_link(struct dentry *, struct inode *, struct dentry *, struct inode **);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001611extern int vfs_rmdir(struct inode *, struct dentry *);
J. Bruce Fieldsb21996e2011-09-20 09:14:34 -04001612extern int vfs_unlink(struct inode *, struct dentry *, struct inode **);
Miklos Szeredi520c8b12014-04-01 17:08:42 +02001613extern int vfs_rename(struct inode *, struct dentry *, struct inode *, struct dentry *, struct inode **, unsigned int);
Miklos Szeredi787fb6b2014-10-24 00:14:36 +02001614extern int vfs_whiteout(struct inode *, struct dentry *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001615
Amir Goldsteinaf7bd4d2017-01-17 06:34:52 +02001616extern struct dentry *vfs_tmpfile(struct dentry *dentry, umode_t mode,
1617 int open_flag);
1618
Al Viro8e6c8482017-12-01 17:12:45 -05001619int vfs_mkobj(struct dentry *, umode_t,
1620 int (*f)(struct dentry *, umode_t, void *),
1621 void *);
1622
Linus Torvalds1da177e2005-04-16 15:20:36 -07001623/*
Christoph Hellwig8c744fb2005-11-08 21:35:04 -08001624 * VFS file helper functions.
1625 */
Dmitry Monakhova1bd1202010-03-04 17:29:14 +03001626extern void inode_init_owner(struct inode *inode, const struct inode *dir,
Al Viro62bb1092011-07-24 23:20:18 -04001627 umode_t mode);
Eric W. Biedermana2982cc2016-06-09 15:34:02 -05001628extern bool may_open_dev(const struct path *path);
Christoph Hellwig8c744fb2005-11-08 21:35:04 -08001629/*
Mark Fashehc4b929b2008-10-08 19:44:18 -04001630 * VFS FS_IOC_FIEMAP helper definitions.
1631 */
1632struct fiemap_extent_info {
1633 unsigned int fi_flags; /* Flags as passed from user */
1634 unsigned int fi_extents_mapped; /* Number of mapped extents */
1635 unsigned int fi_extents_max; /* Size of fiemap_extent array */
Namhyung Kimecf56322011-01-16 23:28:17 +09001636 struct fiemap_extent __user *fi_extents_start; /* Start of
1637 fiemap_extent array */
Mark Fashehc4b929b2008-10-08 19:44:18 -04001638};
1639int fiemap_fill_next_extent(struct fiemap_extent_info *info, u64 logical,
1640 u64 phys, u64 len, u32 flags);
1641int fiemap_check_flags(struct fiemap_extent_info *fieinfo, u32 fs_flags);
1642
1643/*
Linus Torvalds1da177e2005-04-16 15:20:36 -07001644 * File types
1645 *
1646 * NOTE! These match bits 12..15 of stat.st_mode
1647 * (ie "(i_mode >> 12) & 15").
1648 */
1649#define DT_UNKNOWN 0
1650#define DT_FIFO 1
1651#define DT_CHR 2
1652#define DT_DIR 4
1653#define DT_BLK 6
1654#define DT_REG 8
1655#define DT_LNK 10
1656#define DT_SOCK 12
1657#define DT_WHT 14
1658
Linus Torvalds1da177e2005-04-16 15:20:36 -07001659/*
1660 * This is the "filldir" function type, used by readdir() to let
1661 * the kernel specify what kind of dirent layout it wants to have.
1662 * This allows the kernel to read directories into kernel space or
1663 * to have different dirent layouts depending on the binary type.
1664 */
Miklos Szerediac7576f2014-10-30 17:37:34 +01001665struct dir_context;
1666typedef int (*filldir_t)(struct dir_context *, const char *, int, loff_t, u64,
1667 unsigned);
1668
Al Viro5c0ba4e2013-05-15 13:52:59 -04001669struct dir_context {
David Howellsa09acf42018-04-09 21:12:30 +01001670 filldir_t actor;
Al Virobb6f6192013-05-15 18:49:12 -04001671 loff_t pos;
Al Viro5c0ba4e2013-05-15 13:52:59 -04001672};
Al Virobb6f6192013-05-15 18:49:12 -04001673
Al Viro08f85852007-10-08 13:26:20 -04001674struct block_device_operations;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001675
Linus Torvalds1da177e2005-04-16 15:20:36 -07001676/* These macros are for out of kernel modules to test that
1677 * the kernel supports the unlocked_ioctl and compat_ioctl
1678 * fields in struct file_operations. */
1679#define HAVE_COMPAT_IOCTL 1
1680#define HAVE_UNLOCKED_IOCTL 1
1681
Christoph Hellwigb4caecd2015-01-14 10:42:32 +01001682/*
1683 * These flags let !MMU mmap() govern direct device mapping vs immediate
1684 * copying more easily for MAP_PRIVATE, especially for ROM filesystems.
1685 *
1686 * NOMMU_MAP_COPY: Copy can be mapped (MAP_PRIVATE)
1687 * NOMMU_MAP_DIRECT: Can be mapped directly (MAP_SHARED)
1688 * NOMMU_MAP_READ: Can be mapped for reading
1689 * NOMMU_MAP_WRITE: Can be mapped for writing
1690 * NOMMU_MAP_EXEC: Can be mapped for execution
1691 */
1692#define NOMMU_MAP_COPY 0x00000001
1693#define NOMMU_MAP_DIRECT 0x00000008
1694#define NOMMU_MAP_READ VM_MAYREAD
1695#define NOMMU_MAP_WRITE VM_MAYWRITE
1696#define NOMMU_MAP_EXEC VM_MAYEXEC
1697
1698#define NOMMU_VMFLAGS \
1699 (NOMMU_MAP_READ | NOMMU_MAP_WRITE | NOMMU_MAP_EXEC)
1700
1701
Al Viro293bc982014-02-11 18:37:41 -05001702struct iov_iter;
1703
Linus Torvalds1da177e2005-04-16 15:20:36 -07001704struct file_operations {
1705 struct module *owner;
1706 loff_t (*llseek) (struct file *, loff_t, int);
1707 ssize_t (*read) (struct file *, char __user *, size_t, loff_t *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001708 ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *);
Al Viro293bc982014-02-11 18:37:41 -05001709 ssize_t (*read_iter) (struct kiocb *, struct iov_iter *);
1710 ssize_t (*write_iter) (struct kiocb *, struct iov_iter *);
Al Virobb6f6192013-05-15 18:49:12 -04001711 int (*iterate) (struct file *, struct dir_context *);
Al Viro61922692016-04-20 23:08:32 -04001712 int (*iterate_shared) (struct file *, struct dir_context *);
Al Viroa3f86832017-07-02 22:22:01 -04001713 __poll_t (*poll) (struct file *, struct poll_table_struct *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001714 long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long);
1715 long (*compat_ioctl) (struct file *, unsigned int, unsigned long);
1716 int (*mmap) (struct file *, struct vm_area_struct *);
Dan Williams1c972592017-11-01 16:36:30 +01001717 unsigned long mmap_supported_flags;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001718 int (*open) (struct inode *, struct file *);
Miklos Szeredi75e1fcc2006-06-23 02:05:12 -07001719 int (*flush) (struct file *, fl_owner_t id);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001720 int (*release) (struct inode *, struct file *);
Josef Bacik02c24a82011-07-16 20:44:56 -04001721 int (*fsync) (struct file *, loff_t, loff_t, int datasync);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001722 int (*fasync) (int, struct file *, int);
1723 int (*lock) (struct file *, int, struct file_lock *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001724 ssize_t (*sendpage) (struct file *, struct page *, int, size_t, loff_t *, int);
1725 unsigned long (*get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
1726 int (*check_flags)(int);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001727 int (*flock) (struct file *, int, struct file_lock *);
Jens Axboecbb7e572006-04-11 14:57:50 +02001728 ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, loff_t *, size_t, unsigned int);
1729 ssize_t (*splice_read)(struct file *, loff_t *, struct pipe_inode_info *, size_t, unsigned int);
Jeff Laytone6f5c782014-08-22 10:40:25 -04001730 int (*setlease)(struct file *, long, struct file_lock **, void **);
Christoph Hellwig2fe17c12011-01-14 13:07:43 +01001731 long (*fallocate)(struct file *file, int mode, loff_t offset,
1732 loff_t len);
Joe Perchesa3816ab2014-09-29 16:08:25 -07001733 void (*show_fdinfo)(struct seq_file *m, struct file *f);
Christoph Hellwigb4caecd2015-01-14 10:42:32 +01001734#ifndef CONFIG_MMU
1735 unsigned (*mmap_capabilities)(struct file *);
1736#endif
Christoph Hellwig04b38d62015-12-03 12:59:50 +01001737 ssize_t (*copy_file_range)(struct file *, loff_t, struct file *,
1738 loff_t, size_t, unsigned int);
1739 int (*clone_file_range)(struct file *, loff_t, struct file *, loff_t,
1740 u64);
Darrick J. Wong54dbc152015-12-19 00:55:59 -08001741 ssize_t (*dedupe_file_range)(struct file *, u64, u64, struct file *,
1742 u64);
Kees Cook3859a272016-10-28 01:22:25 -07001743} __randomize_layout;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001744
1745struct inode_operations {
Al Viro00cd8dd2012-06-10 17:13:09 -04001746 struct dentry * (*lookup) (struct inode *,struct dentry *, unsigned int);
Al Virofceef392015-12-29 15:58:39 -05001747 const char * (*get_link) (struct dentry *, struct inode *, struct delayed_call *);
Al Viro10556cb2011-06-20 19:28:19 -04001748 int (*permission) (struct inode *, int);
Christoph Hellwig4e34e712011-07-23 17:37:31 +02001749 struct posix_acl * (*get_acl)(struct inode *, int);
Nick Piggin44a7d7a2011-01-07 17:49:56 +11001750
1751 int (*readlink) (struct dentry *, char __user *,int);
Nick Piggin44a7d7a2011-01-07 17:49:56 +11001752
Al Viroebfc3b42012-06-10 18:05:36 -04001753 int (*create) (struct inode *,struct dentry *, umode_t, bool);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001754 int (*link) (struct dentry *,struct inode *,struct dentry *);
1755 int (*unlink) (struct inode *,struct dentry *);
1756 int (*symlink) (struct inode *,struct dentry *,const char *);
Al Viro18bb1db2011-07-26 01:41:39 -04001757 int (*mkdir) (struct inode *,struct dentry *,umode_t);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001758 int (*rmdir) (struct inode *,struct dentry *);
Al Viro1a67aaf2011-07-26 01:52:52 -04001759 int (*mknod) (struct inode *,struct dentry *,umode_t,dev_t);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001760 int (*rename) (struct inode *, struct dentry *,
Miklos Szeredi520c8b12014-04-01 17:08:42 +02001761 struct inode *, struct dentry *, unsigned int);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001762 int (*setattr) (struct dentry *, struct iattr *);
David Howellsa528d352017-01-31 16:46:22 +00001763 int (*getattr) (const struct path *, struct kstat *, u32, unsigned int);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001764 ssize_t (*listxattr) (struct dentry *, char *, size_t);
Mark Fashehc4b929b2008-10-08 19:44:18 -04001765 int (*fiemap)(struct inode *, struct fiemap_extent_info *, u64 start,
1766 u64 len);
Josef Bacikc3b2da32012-03-26 09:59:21 -04001767 int (*update_time)(struct inode *, struct timespec *, int);
Al Virod9585272012-06-22 12:39:14 +04001768 int (*atomic_open)(struct inode *, struct dentry *,
Al Viro30d90492012-06-22 12:40:19 +04001769 struct file *, unsigned open_flag,
Al Virod9585272012-06-22 12:39:14 +04001770 umode_t create_mode, int *opened);
Al Viro60545d02013-06-07 01:20:27 -04001771 int (*tmpfile) (struct inode *, struct dentry *, umode_t);
Christoph Hellwig893d46e2013-12-20 05:16:39 -08001772 int (*set_acl)(struct inode *, struct posix_acl *, int);
Nick Piggin44a7d7a2011-01-07 17:49:56 +11001773} ____cacheline_aligned;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001774
Miklos Szeredibb7462b2017-02-20 16:51:23 +01001775static inline ssize_t call_read_iter(struct file *file, struct kiocb *kio,
1776 struct iov_iter *iter)
1777{
1778 return file->f_op->read_iter(kio, iter);
1779}
1780
1781static inline ssize_t call_write_iter(struct file *file, struct kiocb *kio,
1782 struct iov_iter *iter)
1783{
1784 return file->f_op->write_iter(kio, iter);
1785}
1786
Miklos Szeredif74ac012017-02-20 16:51:23 +01001787static inline int call_mmap(struct file *file, struct vm_area_struct *vma)
1788{
1789 return file->f_op->mmap(file, vma);
1790}
1791
Badari Pulavartyeed4e512006-09-30 23:28:49 -07001792ssize_t rw_copy_check_uvector(int type, const struct iovec __user * uvector,
Christopher Yeohfcf63402011-10-31 17:06:39 -07001793 unsigned long nr_segs, unsigned long fast_segs,
1794 struct iovec *fast_pointer,
Christopher Yeohac34ebb2012-05-31 16:26:42 -07001795 struct iovec **ret_pointer);
Badari Pulavartyeed4e512006-09-30 23:28:49 -07001796
Dmitry Kasatkin6fb50322014-11-05 17:01:17 +02001797extern ssize_t __vfs_read(struct file *, char __user *, size_t, loff_t *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001798extern ssize_t vfs_read(struct file *, char __user *, size_t, loff_t *);
1799extern ssize_t vfs_write(struct file *, const char __user *, size_t, loff_t *);
1800extern ssize_t vfs_readv(struct file *, const struct iovec __user *,
Christoph Hellwigddef7ed22017-07-06 18:58:37 +02001801 unsigned long, loff_t *, rwf_t);
Zach Brown29732932015-11-10 16:53:30 -05001802extern ssize_t vfs_copy_file_range(struct file *, loff_t , struct file *,
1803 loff_t, size_t, unsigned int);
Darrick J. Wong876bec6f2016-12-09 16:18:30 -08001804extern int vfs_clone_file_prep_inodes(struct inode *inode_in, loff_t pos_in,
1805 struct inode *inode_out, loff_t pos_out,
1806 u64 *len, bool is_dedupe);
Christoph Hellwig04b38d62015-12-03 12:59:50 +01001807extern int vfs_clone_file_range(struct file *file_in, loff_t pos_in,
1808 struct file *file_out, loff_t pos_out, u64 len);
Darrick J. Wong876bec6f2016-12-09 16:18:30 -08001809extern int vfs_dedupe_file_range_compare(struct inode *src, loff_t srcoff,
1810 struct inode *dest, loff_t destoff,
1811 loff_t len, bool *is_same);
Darrick J. Wong54dbc152015-12-19 00:55:59 -08001812extern int vfs_dedupe_file_range(struct file *file,
1813 struct file_dedupe_range *same);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001814
Linus Torvalds1da177e2005-04-16 15:20:36 -07001815struct super_operations {
1816 struct inode *(*alloc_inode)(struct super_block *sb);
1817 void (*destroy_inode)(struct inode *);
1818
Christoph Hellwigaa385722011-05-27 06:53:02 -04001819 void (*dirty_inode) (struct inode *, int flags);
Christoph Hellwiga9185b42010-03-05 09:21:37 +01001820 int (*write_inode) (struct inode *, struct writeback_control *wbc);
Al Viro45321ac2010-06-07 13:43:19 -04001821 int (*drop_inode) (struct inode *);
Al Virobe7ce412010-06-04 19:40:39 -04001822 void (*evict_inode) (struct inode *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001823 void (*put_super) (struct super_block *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001824 int (*sync_fs)(struct super_block *sb, int wait);
Benjamin Marzinski48b6bca2014-11-13 20:42:03 -06001825 int (*freeze_super) (struct super_block *);
Takashi Satoc4be0c12009-01-09 16:40:58 -08001826 int (*freeze_fs) (struct super_block *);
Benjamin Marzinski48b6bca2014-11-13 20:42:03 -06001827 int (*thaw_super) (struct super_block *);
Takashi Satoc4be0c12009-01-09 16:40:58 -08001828 int (*unfreeze_fs) (struct super_block *);
David Howells726c3342006-06-23 02:02:58 -07001829 int (*statfs) (struct dentry *, struct kstatfs *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001830 int (*remount_fs) (struct super_block *, int *, char *);
Al Viro42faad92008-04-24 07:21:56 -04001831 void (*umount_begin) (struct super_block *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001832
Al Viro34c80b12011-12-08 21:32:45 -05001833 int (*show_options)(struct seq_file *, struct dentry *);
Al Virod861c632011-12-08 21:32:45 -05001834 int (*show_devname)(struct seq_file *, struct dentry *);
Al Viroa6322de2011-12-08 21:37:57 -05001835 int (*show_path)(struct seq_file *, struct dentry *);
Al Viro64132372011-12-08 20:51:13 -05001836 int (*show_stats)(struct seq_file *, struct dentry *);
Alexey Dobriyan0e51a722006-09-29 01:59:56 -07001837#ifdef CONFIG_QUOTA
Linus Torvalds1da177e2005-04-16 15:20:36 -07001838 ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t);
1839 ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t);
Jan Kara2d0fa462014-09-25 16:36:14 +02001840 struct dquot **(*get_dquots)(struct inode *);
Alexey Dobriyan0e51a722006-09-29 01:59:56 -07001841#endif
Theodore Ts'o87d8fe12009-01-03 09:47:09 -05001842 int (*bdev_try_to_free_page)(struct super_block*, struct page*, gfp_t);
Vladimir Davydov4101b622015-02-12 14:58:51 -08001843 long (*nr_cached_objects)(struct super_block *,
1844 struct shrink_control *);
1845 long (*free_cached_objects)(struct super_block *,
1846 struct shrink_control *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001847};
1848
Joern Engel1c0eeaf2007-10-16 23:30:44 -07001849/*
David Howellsbbc10962012-10-15 16:40:35 +01001850 * Inode flags - they have no relation to superblock flags now
1851 */
1852#define S_SYNC 1 /* Writes are synced at once */
1853#define S_NOATIME 2 /* Do not update access times */
1854#define S_APPEND 4 /* Append-only file */
1855#define S_IMMUTABLE 8 /* Immutable file */
1856#define S_DEAD 16 /* removed, but still open directory */
1857#define S_NOQUOTA 32 /* Inode is not counted to quota */
1858#define S_DIRSYNC 64 /* Directory modifications are synchronous */
1859#define S_NOCMTIME 128 /* Do not update file c/mtime */
1860#define S_SWAPFILE 256 /* Do not truncate: swapon got its bmaps */
1861#define S_PRIVATE 512 /* Inode is fs-internal */
1862#define S_IMA 1024 /* Inode has an associated IMA struct */
1863#define S_AUTOMOUNT 2048 /* Automount/referral quasi-directory */
1864#define S_NOSEC 4096 /* no suid or xattr security attributes */
Matthew Wilcox6cd176a2015-02-16 15:59:25 -08001865#ifdef CONFIG_FS_DAX
Matthew Wilcoxfbbbad42015-02-16 15:58:53 -08001866#define S_DAX 8192 /* Direct Access, avoiding the page cache */
1867#else
1868#define S_DAX 0 /* Make all the DAX code disappear */
1869#endif
Eric Biggers2ee6a572017-10-09 12:15:35 -07001870#define S_ENCRYPTED 16384 /* Encrypted file (using fs/crypto/) */
David Howellsbbc10962012-10-15 16:40:35 +01001871
1872/*
1873 * Note that nosuid etc flags are inode-specific: setting some file-system
1874 * flags just means all the inodes inherit those flags by default. It might be
1875 * possible to override it selectively if you really wanted to with some
1876 * ioctl() that is not currently implemented.
1877 *
David Howellse462ec52017-07-17 08:45:35 +01001878 * Exception: SB_RDONLY is always applied to the entire file system.
David Howellsbbc10962012-10-15 16:40:35 +01001879 *
1880 * Unfortunately, it is possible to change a filesystems flags with it mounted
1881 * with files in use. This means that all of the inodes will not have their
1882 * i_flags updated. Hence, i_flags no longer inherit the superblock mount
1883 * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org
1884 */
1885#define __IS_FLG(inode, flg) ((inode)->i_sb->s_flags & (flg))
1886
Linus Torvalds1751e8a2017-11-27 13:05:09 -08001887static inline bool sb_rdonly(const struct super_block *sb) { return sb->s_flags & SB_RDONLY; }
David Howells94e92e72017-07-17 08:45:34 +01001888#define IS_RDONLY(inode) sb_rdonly((inode)->i_sb)
David Howellse462ec52017-07-17 08:45:35 +01001889#define IS_SYNC(inode) (__IS_FLG(inode, SB_SYNCHRONOUS) || \
David Howellsbbc10962012-10-15 16:40:35 +01001890 ((inode)->i_flags & S_SYNC))
David Howellse462ec52017-07-17 08:45:35 +01001891#define IS_DIRSYNC(inode) (__IS_FLG(inode, SB_SYNCHRONOUS|SB_DIRSYNC) || \
David Howellsbbc10962012-10-15 16:40:35 +01001892 ((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
David Howellse462ec52017-07-17 08:45:35 +01001893#define IS_MANDLOCK(inode) __IS_FLG(inode, SB_MANDLOCK)
1894#define IS_NOATIME(inode) __IS_FLG(inode, SB_RDONLY|SB_NOATIME)
1895#define IS_I_VERSION(inode) __IS_FLG(inode, SB_I_VERSION)
David Howellsbbc10962012-10-15 16:40:35 +01001896
1897#define IS_NOQUOTA(inode) ((inode)->i_flags & S_NOQUOTA)
1898#define IS_APPEND(inode) ((inode)->i_flags & S_APPEND)
1899#define IS_IMMUTABLE(inode) ((inode)->i_flags & S_IMMUTABLE)
David Howellse462ec52017-07-17 08:45:35 +01001900#define IS_POSIXACL(inode) __IS_FLG(inode, SB_POSIXACL)
David Howellsbbc10962012-10-15 16:40:35 +01001901
1902#define IS_DEADDIR(inode) ((inode)->i_flags & S_DEAD)
1903#define IS_NOCMTIME(inode) ((inode)->i_flags & S_NOCMTIME)
1904#define IS_SWAPFILE(inode) ((inode)->i_flags & S_SWAPFILE)
1905#define IS_PRIVATE(inode) ((inode)->i_flags & S_PRIVATE)
1906#define IS_IMA(inode) ((inode)->i_flags & S_IMA)
1907#define IS_AUTOMOUNT(inode) ((inode)->i_flags & S_AUTOMOUNT)
1908#define IS_NOSEC(inode) ((inode)->i_flags & S_NOSEC)
Matthew Wilcoxfbbbad42015-02-16 15:58:53 -08001909#define IS_DAX(inode) ((inode)->i_flags & S_DAX)
Eric Biggers2ee6a572017-10-09 12:15:35 -07001910#define IS_ENCRYPTED(inode) ((inode)->i_flags & S_ENCRYPTED)
David Howellsbbc10962012-10-15 16:40:35 +01001911
Miklos Szeredi787fb6b2014-10-24 00:14:36 +02001912#define IS_WHITEOUT(inode) (S_ISCHR(inode->i_mode) && \
1913 (inode)->i_rdev == WHITEOUT_DEV)
1914
Eric W. Biederman0bd23d092016-06-29 14:54:46 -05001915static inline bool HAS_UNMAPPED_ID(struct inode *inode)
1916{
1917 return !uid_valid(inode->i_uid) || !gid_valid(inode->i_gid);
1918}
1919
Jens Axboec75b1d92017-06-27 11:47:04 -06001920static inline enum rw_hint file_write_hint(struct file *file)
1921{
1922 if (file->f_write_hint != WRITE_LIFE_NOT_SET)
1923 return file->f_write_hint;
1924
1925 return file_inode(file)->i_write_hint;
1926}
1927
1928static inline int iocb_flags(struct file *file);
1929
1930static inline void init_sync_kiocb(struct kiocb *kiocb, struct file *filp)
1931{
1932 *kiocb = (struct kiocb) {
1933 .ki_filp = filp,
1934 .ki_flags = iocb_flags(filp),
1935 .ki_hint = file_write_hint(filp),
1936 };
1937}
1938
David Howellsbbc10962012-10-15 16:40:35 +01001939/*
Dave Chinner250df6e2011-03-22 22:23:36 +11001940 * Inode state bits. Protected by inode->i_lock
Joern Engel1c0eeaf2007-10-16 23:30:44 -07001941 *
1942 * Three bits determine the dirty state of the inode, I_DIRTY_SYNC,
1943 * I_DIRTY_DATASYNC and I_DIRTY_PAGES.
1944 *
1945 * Four bits define the lifetime of an inode. Initially, inodes are I_NEW,
1946 * until that flag is cleared. I_WILL_FREE, I_FREEING and I_CLEAR are set at
1947 * various stages of removing an inode.
1948 *
Christoph Hellwigeaff8072009-12-17 14:25:01 +01001949 * Two bits are used for locking and completion notification, I_NEW and I_SYNC.
Joern Engel1c0eeaf2007-10-16 23:30:44 -07001950 *
Joern Engele7ca2d42008-02-06 01:36:59 -08001951 * I_DIRTY_SYNC Inode is dirty, but doesn't have to be written on
1952 * fdatasync(). i_atime is the usual cause.
Jan Kara000cb482008-02-14 19:31:32 -08001953 * I_DIRTY_DATASYNC Data-related inode changes pending. We keep track of
1954 * these changes separately from I_DIRTY_SYNC so that we
1955 * don't have to write inode on fdatasync() when only
1956 * mtime has changed in it.
Joern Engel1c0eeaf2007-10-16 23:30:44 -07001957 * I_DIRTY_PAGES Inode has dirty pages. Inode itself may be clean.
Christoph Hellwigeaff8072009-12-17 14:25:01 +01001958 * I_NEW Serves as both a mutex and completion notification.
1959 * New inodes set I_NEW. If two processes both create
1960 * the same inode, one of them will release its inode and
1961 * wait for I_NEW to be released before returning.
1962 * Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can
1963 * also cause waiting on I_NEW, without I_NEW actually
1964 * being set. find_inode() uses this to prevent returning
1965 * nearly-dead inodes.
Joern Engel1c0eeaf2007-10-16 23:30:44 -07001966 * I_WILL_FREE Must be set when calling write_inode_now() if i_count
1967 * is zero. I_FREEING must be set when I_WILL_FREE is
1968 * cleared.
1969 * I_FREEING Set when inode is about to be freed but still has dirty
1970 * pages or buffers attached or the inode itself is still
1971 * dirty.
Jan Karadbd57682012-05-03 14:48:02 +02001972 * I_CLEAR Added by clear_inode(). In this state the inode is
1973 * clean and can be destroyed. Inode keeps I_FREEING.
Joern Engel1c0eeaf2007-10-16 23:30:44 -07001974 *
1975 * Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are
1976 * prohibited for many purposes. iget() must wait for
1977 * the inode to be completely released, then create it
1978 * anew. Other functions will just ignore such inodes,
Christoph Hellwigeaff8072009-12-17 14:25:01 +01001979 * if appropriate. I_NEW is used for waiting.
Joern Engel1c0eeaf2007-10-16 23:30:44 -07001980 *
Jan Kara169ebd92012-05-03 14:48:03 +02001981 * I_SYNC Writeback of inode is running. The bit is set during
1982 * data writeback, and cleared with a wakeup on the bit
1983 * address once it is done. The bit is also used to pin
1984 * the inode in memory for flusher thread.
Joern Engel1c0eeaf2007-10-16 23:30:44 -07001985 *
Christoph Hellwigbd5fe6c2011-06-24 14:29:43 -04001986 * I_REFERENCED Marks the inode as recently references on the LRU list.
1987 *
1988 * I_DIO_WAKEUP Never set. Only used as a key for wait_on_bit().
1989 *
Tejun Heo682aa8e2015-05-28 14:50:53 -04001990 * I_WB_SWITCH Cgroup bdi_writeback switching in progress. Used to
1991 * synchronize competing switching instances and to tell
Matthew Wilcoxb93b0162018-04-10 16:36:56 -07001992 * wb stat updates to grab the i_pages lock. See
Tejun Heo682aa8e2015-05-28 14:50:53 -04001993 * inode_switch_wb_work_fn() for details.
1994 *
Amir Goldsteinad0af712017-06-21 15:28:32 +03001995 * I_OVL_INUSE Used by overlayfs to get exclusive ownership on upper
1996 * and work dirs among overlayfs mounts.
1997 *
Joern Engel1c0eeaf2007-10-16 23:30:44 -07001998 * Q: What is the difference between I_WILL_FREE and I_FREEING?
Joern Engel1c0eeaf2007-10-16 23:30:44 -07001999 */
Nick Piggin9e38d862010-10-23 06:55:17 -04002000#define I_DIRTY_SYNC (1 << 0)
2001#define I_DIRTY_DATASYNC (1 << 1)
2002#define I_DIRTY_PAGES (1 << 2)
Christoph Hellwigeaff8072009-12-17 14:25:01 +01002003#define __I_NEW 3
2004#define I_NEW (1 << __I_NEW)
Nick Piggin9e38d862010-10-23 06:55:17 -04002005#define I_WILL_FREE (1 << 4)
2006#define I_FREEING (1 << 5)
2007#define I_CLEAR (1 << 6)
Christoph Hellwigeaff8072009-12-17 14:25:01 +01002008#define __I_SYNC 7
Joern Engel1c0eeaf2007-10-16 23:30:44 -07002009#define I_SYNC (1 << __I_SYNC)
Nick Piggin9e38d862010-10-23 06:55:17 -04002010#define I_REFERENCED (1 << 8)
Christoph Hellwigbd5fe6c2011-06-24 14:29:43 -04002011#define __I_DIO_WAKEUP 9
Eric Sandeenac74d8d2015-04-16 15:04:56 -05002012#define I_DIO_WAKEUP (1 << __I_DIO_WAKEUP)
Al Virof4e0c302013-06-11 08:34:36 +04002013#define I_LINKABLE (1 << 10)
Theodore Ts'o0ae45f62015-02-02 00:37:00 -05002014#define I_DIRTY_TIME (1 << 11)
2015#define __I_DIRTY_TIME_EXPIRED 12
2016#define I_DIRTY_TIME_EXPIRED (1 << __I_DIRTY_TIME_EXPIRED)
Tejun Heo682aa8e2015-05-28 14:50:53 -04002017#define I_WB_SWITCH (1 << 13)
Amir Goldsteinad0af712017-06-21 15:28:32 +03002018#define I_OVL_INUSE (1 << 14)
Linus Torvalds1da177e2005-04-16 15:20:36 -07002019
Christoph Hellwig0e11f642018-02-21 07:54:49 -08002020#define I_DIRTY_INODE (I_DIRTY_SYNC | I_DIRTY_DATASYNC)
2021#define I_DIRTY (I_DIRTY_INODE | I_DIRTY_PAGES)
Theodore Ts'o0ae45f62015-02-02 00:37:00 -05002022#define I_DIRTY_ALL (I_DIRTY | I_DIRTY_TIME)
Linus Torvalds1da177e2005-04-16 15:20:36 -07002023
2024extern void __mark_inode_dirty(struct inode *, int);
2025static inline void mark_inode_dirty(struct inode *inode)
2026{
2027 __mark_inode_dirty(inode, I_DIRTY);
2028}
2029
2030static inline void mark_inode_dirty_sync(struct inode *inode)
2031{
2032 __mark_inode_dirty(inode, I_DIRTY_SYNC);
2033}
2034
Miklos Szeredi7ada4db2011-11-21 12:11:32 +01002035extern void inc_nlink(struct inode *inode);
2036extern void drop_nlink(struct inode *inode);
2037extern void clear_nlink(struct inode *inode);
2038extern void set_nlink(struct inode *inode, unsigned int nlink);
Dave Hansend8c76e62006-09-30 23:29:04 -07002039
2040static inline void inode_inc_link_count(struct inode *inode)
2041{
2042 inc_nlink(inode);
Alexey Dobriyan66d21732006-03-23 03:00:51 -08002043 mark_inode_dirty(inode);
2044}
2045
Dave Hansen9a53c3a2006-09-30 23:29:03 -07002046static inline void inode_dec_link_count(struct inode *inode)
2047{
2048 drop_nlink(inode);
Alexey Dobriyan66d21732006-03-23 03:00:51 -08002049 mark_inode_dirty(inode);
2050}
2051
Josef Bacikc3b2da32012-03-26 09:59:21 -04002052enum file_time_flags {
2053 S_ATIME = 1,
2054 S_MTIME = 2,
2055 S_CTIME = 4,
2056 S_VERSION = 8,
2057};
2058
Al Virobadcf2b2013-07-16 18:15:46 +04002059extern void touch_atime(const struct path *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002060static inline void file_accessed(struct file *file)
2061{
2062 if (!(file->f_flags & O_NOATIME))
Al Viro68ac1232012-03-15 08:21:57 -04002063 touch_atime(&file->f_path);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002064}
2065
2066int sync_inode(struct inode *inode, struct writeback_control *wbc);
Christoph Hellwigc37650162010-10-06 10:48:20 +02002067int sync_inode_metadata(struct inode *inode, int wait);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002068
Linus Torvalds1da177e2005-04-16 15:20:36 -07002069struct file_system_type {
2070 const char *name;
2071 int fs_flags;
David Howellsbbc10962012-10-15 16:40:35 +01002072#define FS_REQUIRES_DEV 1
2073#define FS_BINARY_MOUNTDATA 2
2074#define FS_HAS_SUBTYPE 4
Eric W. Biederman0c55cfc2012-07-26 21:42:03 -07002075#define FS_USERNS_MOUNT 8 /* Can be mounted by userns root */
David Howellsbbc10962012-10-15 16:40:35 +01002076#define FS_RENAME_DOES_D_MOVE 32768 /* FS will handle d_move() during rename() internally. */
Al Viroc96e41e2010-07-25 00:17:56 +04002077 struct dentry *(*mount) (struct file_system_type *, int,
2078 const char *, void *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002079 void (*kill_sb) (struct super_block *);
2080 struct module *owner;
2081 struct file_system_type * next;
Al Viroa5166162011-12-12 22:53:00 -05002082 struct hlist_head fs_supers;
Peter Zijlstrad475fd42007-10-15 14:51:31 +02002083
Ingo Molnarcf516242006-07-03 00:25:27 -07002084 struct lock_class_key s_lock_key;
Arjan van de Ven897c6ff2006-07-03 00:25:28 -07002085 struct lock_class_key s_umount_key;
Roland Dreier51ee0492010-04-27 14:23:57 -07002086 struct lock_class_key s_vfs_rename_key;
Jan Kara5accdf82012-06-12 16:20:34 +02002087 struct lock_class_key s_writers_key[SB_FREEZE_LEVELS];
Peter Zijlstrad475fd42007-10-15 14:51:31 +02002088
2089 struct lock_class_key i_lock_key;
2090 struct lock_class_key i_mutex_key;
Peter Zijlstra14358e62007-10-14 01:38:33 +02002091 struct lock_class_key i_mutex_dir_key;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002092};
2093
Eric W. Biederman7f78e032013-03-02 19:39:14 -08002094#define MODULE_ALIAS_FS(NAME) MODULE_ALIAS("fs-" NAME)
2095
Eric W. Biedermand91ee872016-05-23 14:51:59 -05002096extern struct dentry *mount_ns(struct file_system_type *fs_type,
2097 int flags, void *data, void *ns, struct user_namespace *user_ns,
2098 int (*fill_super)(struct super_block *, void *, int));
Al Virof175f302017-10-15 00:38:00 -04002099#ifdef CONFIG_BLOCK
Al Viro152a0832010-07-25 00:46:55 +04002100extern struct dentry *mount_bdev(struct file_system_type *fs_type,
2101 int flags, const char *dev_name, void *data,
2102 int (*fill_super)(struct super_block *, void *, int));
Al Virof175f302017-10-15 00:38:00 -04002103#else
2104static inline struct dentry *mount_bdev(struct file_system_type *fs_type,
2105 int flags, const char *dev_name, void *data,
2106 int (*fill_super)(struct super_block *, void *, int))
2107{
2108 return ERR_PTR(-ENODEV);
2109}
2110#endif
Al Virofc14f2f2010-07-25 01:48:30 +04002111extern struct dentry *mount_single(struct file_system_type *fs_type,
2112 int flags, void *data,
2113 int (*fill_super)(struct super_block *, void *, int));
Al Viro3c26ff62010-07-25 11:46:36 +04002114extern struct dentry *mount_nodev(struct file_system_type *fs_type,
2115 int flags, void *data,
2116 int (*fill_super)(struct super_block *, void *, int));
Al Viroea441d12011-11-16 21:43:59 -05002117extern struct dentry *mount_subtree(struct vfsmount *mnt, const char *path);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002118void generic_shutdown_super(struct super_block *sb);
Al Virof175f302017-10-15 00:38:00 -04002119#ifdef CONFIG_BLOCK
Linus Torvalds1da177e2005-04-16 15:20:36 -07002120void kill_block_super(struct super_block *sb);
Al Virof175f302017-10-15 00:38:00 -04002121#else
2122static inline void kill_block_super(struct super_block *sb)
2123{
2124 BUG();
2125}
2126#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -07002127void kill_anon_super(struct super_block *sb);
2128void kill_litter_super(struct super_block *sb);
2129void deactivate_super(struct super_block *sb);
Al Viro74dbbdd2009-05-06 01:07:50 -04002130void deactivate_locked_super(struct super_block *sb);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002131int set_anon_super(struct super_block *s, void *data);
Al Viro0ee5dc62011-07-07 15:44:25 -04002132int get_anon_bdev(dev_t *);
2133void free_anon_bdev(dev_t);
Eric W. Biederman6e4eab52016-05-24 09:29:01 -05002134struct super_block *sget_userns(struct file_system_type *type,
2135 int (*test)(struct super_block *,void *),
2136 int (*set)(struct super_block *,void *),
2137 int flags, struct user_namespace *user_ns,
2138 void *data);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002139struct super_block *sget(struct file_system_type *type,
2140 int (*test)(struct super_block *,void *),
2141 int (*set)(struct super_block *,void *),
David Howells9249e172012-06-25 12:55:37 +01002142 int flags, void *data);
Andreas Gruenbacherbba0bd32016-09-29 17:48:35 +02002143extern struct dentry *mount_pseudo_xattr(struct file_system_type *, char *,
2144 const struct super_operations *ops,
2145 const struct xattr_handler **xattr,
2146 const struct dentry_operations *dops,
2147 unsigned long);
2148
2149static inline struct dentry *
2150mount_pseudo(struct file_system_type *fs_type, char *name,
2151 const struct super_operations *ops,
2152 const struct dentry_operations *dops, unsigned long magic)
2153{
2154 return mount_pseudo_xattr(fs_type, name, ops, NULL, dops, magic);
2155}
Linus Torvalds1da177e2005-04-16 15:20:36 -07002156
2157/* Alas, no aliases. Too much hassle with bringing module.h everywhere */
2158#define fops_get(fops) \
2159 (((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
2160#define fops_put(fops) \
2161 do { if (fops) module_put((fops)->owner); } while(0)
Al Viroe84f9e52013-09-22 14:17:15 -04002162/*
2163 * This one is to be used *ONLY* from ->open() instances.
2164 * fops must be non-NULL, pinned down *and* module dependencies
2165 * should be sufficient to pin the caller down as well.
2166 */
2167#define replace_fops(f, fops) \
2168 do { \
2169 struct file *__file = (f); \
2170 fops_put(__file->f_op); \
2171 BUG_ON(!(__file->f_op = (fops))); \
2172 } while(0)
Linus Torvalds1da177e2005-04-16 15:20:36 -07002173
2174extern int register_filesystem(struct file_system_type *);
2175extern int unregister_filesystem(struct file_system_type *);
Pavel Emelyanov8bf97252007-10-18 23:40:02 -07002176extern struct vfsmount *kern_mount_data(struct file_system_type *, void *data);
2177#define kern_mount(type) kern_mount_data(type, NULL)
Tim Chen423e0ab2011-07-19 09:32:38 -07002178extern void kern_unmount(struct vfsmount *mnt);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002179extern int may_umount_tree(struct vfsmount *);
2180extern int may_umount(struct vfsmount *);
Seunghun Lee5e6123f2014-09-14 22:15:10 +09002181extern long do_mount(const char *, const char __user *,
2182 const char *, unsigned long, void *);
Al Viroca71cf72016-11-20 19:45:28 -05002183extern struct vfsmount *collect_mounts(const struct path *);
Al Viro8aec0802007-06-07 12:20:32 -04002184extern void drop_collected_mounts(struct vfsmount *);
Al Viro1f707132010-01-30 22:51:25 -05002185extern int iterate_mounts(int (*)(struct vfsmount *, void *), void *,
2186 struct vfsmount *);
Al Virof0bb5aa2016-11-20 20:27:12 -05002187extern int vfs_statfs(const struct path *, struct kstatfs *);
Al Viroc8b91ac2011-03-12 10:41:39 -05002188extern int user_statfs(const char __user *, struct kstatfs *);
2189extern int fd_statfs(int, struct kstatfs *);
Josef Bacik18e9e512010-03-23 10:34:56 -04002190extern int freeze_super(struct super_block *super);
2191extern int thaw_super(struct super_block *super);
Al Viro02125a82011-12-05 08:43:34 -05002192extern bool our_mnt(struct vfsmount *mnt);
Jan Karafca39342017-04-12 12:24:28 +02002193extern __printf(2, 3)
2194int super_setup_bdi_name(struct super_block *sb, char *fmt, ...);
2195extern int super_setup_bdi(struct super_block *sb);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002196
Al Viroce3b0f82009-03-29 19:08:22 -04002197extern int current_umask(void);
2198
J. Bruce Fields5a146962012-08-28 07:50:40 -07002199extern void ihold(struct inode * inode);
2200extern void iput(struct inode *);
Theodore Ts'o0ae45f62015-02-02 00:37:00 -05002201extern int generic_update_time(struct inode *, struct timespec *, int);
J. Bruce Fields5a146962012-08-28 07:50:40 -07002202
Miklos Szeredif87fd4c2006-01-16 22:14:23 -08002203/* /sys/fs */
Greg Kroah-Hartman00d26662007-10-29 14:17:23 -06002204extern struct kobject *fs_kobj;
Miklos Szeredif87fd4c2006-01-16 22:14:23 -08002205
Kirill A. Shutemov09cbfea2016-04-01 15:29:47 +03002206#define MAX_RW_COUNT (INT_MAX & PAGE_MASK)
Thomas Petazzonibfcd17a2008-08-06 15:12:22 +02002207
Jeff Layton9e8925b2015-11-16 09:49:34 -05002208#ifdef CONFIG_MANDATORY_FILE_LOCKING
Jeff Laytond7a06982014-03-10 09:54:15 -04002209extern int locks_mandatory_locked(struct file *);
Christoph Hellwigacc15572015-12-03 12:59:49 +01002210extern int locks_mandatory_area(struct inode *, struct file *, loff_t, loff_t, unsigned char);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002211
2212/*
2213 * Candidates for mandatory locking have the setgid bit set
2214 * but no group execute bit - an otherwise meaningless combination.
2215 */
Pavel Emelyanova16877c2007-10-01 14:41:11 -07002216
2217static inline int __mandatory_lock(struct inode *ino)
2218{
2219 return (ino->i_mode & (S_ISGID | S_IXGRP)) == S_ISGID;
2220}
2221
2222/*
David Howellse462ec52017-07-17 08:45:35 +01002223 * ... and these candidates should be on SB_MANDLOCK mounted fs,
Pavel Emelyanova16877c2007-10-01 14:41:11 -07002224 * otherwise these will be advisory locks
2225 */
2226
2227static inline int mandatory_lock(struct inode *ino)
2228{
2229 return IS_MANDLOCK(ino) && __mandatory_lock(ino);
2230}
Linus Torvalds1da177e2005-04-16 15:20:36 -07002231
Jeff Laytond7a06982014-03-10 09:54:15 -04002232static inline int locks_verify_locked(struct file *file)
Linus Torvalds1da177e2005-04-16 15:20:36 -07002233{
Miklos Szeredic568d682016-09-16 12:44:20 +02002234 if (mandatory_lock(locks_inode(file)))
Jeff Laytond7a06982014-03-10 09:54:15 -04002235 return locks_mandatory_locked(file);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002236 return 0;
2237}
2238
Linus Torvalds1da177e2005-04-16 15:20:36 -07002239static inline int locks_verify_truncate(struct inode *inode,
Christoph Hellwigacc15572015-12-03 12:59:49 +01002240 struct file *f,
Linus Torvalds1da177e2005-04-16 15:20:36 -07002241 loff_t size)
2242{
Christoph Hellwigacc15572015-12-03 12:59:49 +01002243 if (!inode->i_flctx || !mandatory_lock(inode))
2244 return 0;
2245
2246 if (size < inode->i_size) {
2247 return locks_mandatory_area(inode, f, size, inode->i_size - 1,
2248 F_WRLCK);
2249 } else {
2250 return locks_mandatory_area(inode, f, inode->i_size, size - 1,
2251 F_WRLCK);
2252 }
Linus Torvalds1da177e2005-04-16 15:20:36 -07002253}
2254
Jeff Layton9e8925b2015-11-16 09:49:34 -05002255#else /* !CONFIG_MANDATORY_FILE_LOCKING */
2256
2257static inline int locks_mandatory_locked(struct file *file)
2258{
2259 return 0;
2260}
2261
Linus Torvaldsfce205e2016-01-12 16:30:34 -08002262static inline int locks_mandatory_area(struct inode *inode, struct file *filp,
2263 loff_t start, loff_t end, unsigned char type)
Jeff Layton9e8925b2015-11-16 09:49:34 -05002264{
2265 return 0;
2266}
2267
2268static inline int __mandatory_lock(struct inode *inode)
2269{
2270 return 0;
2271}
2272
2273static inline int mandatory_lock(struct inode *inode)
2274{
2275 return 0;
2276}
2277
2278static inline int locks_verify_locked(struct file *file)
2279{
2280 return 0;
2281}
2282
2283static inline int locks_verify_truncate(struct inode *inode, struct file *filp,
2284 size_t size)
2285{
2286 return 0;
2287}
2288
2289#endif /* CONFIG_MANDATORY_FILE_LOCKING */
2290
2291
2292#ifdef CONFIG_FILE_LOCKING
Linus Torvalds1da177e2005-04-16 15:20:36 -07002293static inline int break_lease(struct inode *inode, unsigned int mode)
2294{
Jeff Layton24cbe782014-02-03 12:13:06 -05002295 /*
2296 * Since this check is lockless, we must ensure that any refcounts
Jeff Layton8116bf42015-01-21 20:44:01 -05002297 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2298 * could end up racing with tasks trying to set a new lease on this
2299 * file.
Jeff Layton24cbe782014-02-03 12:13:06 -05002300 */
2301 smp_mb();
Jeff Layton8634b512015-01-16 15:05:55 -05002302 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
J. Bruce Fieldsdf4e8d22012-03-05 13:18:59 -05002303 return __break_lease(inode, mode, FL_LEASE);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002304 return 0;
2305}
J. Bruce Fieldsdf4e8d22012-03-05 13:18:59 -05002306
2307static inline int break_deleg(struct inode *inode, unsigned int mode)
2308{
Jeff Layton962bd402014-06-10 12:24:40 -04002309 /*
2310 * Since this check is lockless, we must ensure that any refcounts
Jeff Layton8116bf42015-01-21 20:44:01 -05002311 * taken are done before checking i_flctx->flc_lease. Otherwise, we
2312 * could end up racing with tasks trying to set a new lease on this
2313 * file.
Jeff Layton962bd402014-06-10 12:24:40 -04002314 */
2315 smp_mb();
Jeff Layton8634b512015-01-16 15:05:55 -05002316 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
J. Bruce Fieldsdf4e8d22012-03-05 13:18:59 -05002317 return __break_lease(inode, mode, FL_DELEG);
2318 return 0;
2319}
2320
J. Bruce Fields5a146962012-08-28 07:50:40 -07002321static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode)
2322{
2323 int ret;
2324
2325 ret = break_deleg(inode, O_WRONLY|O_NONBLOCK);
2326 if (ret == -EWOULDBLOCK && delegated_inode) {
2327 *delegated_inode = inode;
2328 ihold(inode);
2329 }
2330 return ret;
2331}
2332
2333static inline int break_deleg_wait(struct inode **delegated_inode)
2334{
2335 int ret;
2336
2337 ret = break_deleg(*delegated_inode, O_WRONLY);
2338 iput(*delegated_inode);
2339 *delegated_inode = NULL;
2340 return ret;
2341}
2342
Christoph Hellwig11afe9f2015-01-21 19:17:03 +01002343static inline int break_layout(struct inode *inode, bool wait)
2344{
2345 smp_mb();
2346 if (inode->i_flctx && !list_empty_careful(&inode->i_flctx->flc_lease))
2347 return __break_lease(inode,
2348 wait ? O_WRONLY : O_WRONLY | O_NONBLOCK,
2349 FL_LAYOUT);
2350 return 0;
2351}
2352
Thomas Petazzonibfcd17a2008-08-06 15:12:22 +02002353#else /* !CONFIG_FILE_LOCKING */
Steven Whitehouseaf5df562009-01-20 10:29:46 +00002354static inline int break_lease(struct inode *inode, unsigned int mode)
2355{
2356 return 0;
2357}
2358
J. Bruce Fieldsdf4e8d22012-03-05 13:18:59 -05002359static inline int break_deleg(struct inode *inode, unsigned int mode)
2360{
2361 return 0;
2362}
J. Bruce Fields5a146962012-08-28 07:50:40 -07002363
2364static inline int try_break_deleg(struct inode *inode, struct inode **delegated_inode)
2365{
2366 return 0;
2367}
2368
2369static inline int break_deleg_wait(struct inode **delegated_inode)
2370{
2371 BUG();
2372 return 0;
2373}
2374
Christoph Hellwig11afe9f2015-01-21 19:17:03 +01002375static inline int break_layout(struct inode *inode, bool wait)
2376{
2377 return 0;
2378}
2379
Thomas Petazzonibfcd17a2008-08-06 15:12:22 +02002380#endif /* CONFIG_FILE_LOCKING */
Linus Torvalds1da177e2005-04-16 15:20:36 -07002381
2382/* fs/open.c */
Jeff Laytonadb5c242012-10-10 16:43:13 -04002383struct audit_names;
Jeff Layton91a27b22012-10-10 15:25:28 -04002384struct filename {
Jeff Laytonadb5c242012-10-10 16:43:13 -04002385 const char *name; /* pointer to actual string */
2386 const __user char *uptr; /* original userland pointer */
Paul Moore55422d02015-01-22 00:00:23 -05002387 int refcnt;
Rasmus Villemoes1c9498432018-03-01 00:19:21 +01002388 struct audit_names *aname;
Al Virofd2f7cb2015-02-22 20:07:13 -05002389 const char iname[];
Jeff Layton91a27b22012-10-10 15:25:28 -04002390};
Linus Torvalds1da177e2005-04-16 15:20:36 -07002391
Al Viro7df818b2016-03-25 14:24:09 -04002392extern long vfs_truncate(const struct path *, loff_t);
NeilBrown4a301312006-01-08 01:02:39 -08002393extern int do_truncate(struct dentry *, loff_t start, unsigned int time_attrs,
2394 struct file *filp);
Anna Schumaker72c72bd2014-11-07 14:44:25 -05002395extern int vfs_fallocate(struct file *file, int mode, loff_t offset,
Ankit Jain3e63cbb2009-06-19 14:28:07 -04002396 loff_t len);
Jason Uhlenkott8e8a1402007-10-20 03:16:18 +02002397extern long do_sys_open(int dfd, const char __user *filename, int flags,
Al Viroa218d0f2011-11-21 14:59:34 -05002398 umode_t mode);
Jeff Layton669abf42012-10-10 16:43:10 -04002399extern struct file *file_open_name(struct filename *, int, umode_t);
Al Viroa218d0f2011-11-21 14:59:34 -05002400extern struct file *filp_open(const char *, int, umode_t);
Al Viro73d049a2011-03-11 12:08:24 -05002401extern struct file *file_open_root(struct dentry *, struct vfsmount *,
Jann Horn378c6522016-03-22 14:25:36 -07002402 const char *, int, umode_t);
Al Viro765927b2012-06-26 21:58:53 +04002403extern struct file * dentry_open(const struct path *, int, const struct cred *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002404extern int filp_close(struct file *, fl_owner_t id);
Jeff Layton91a27b22012-10-10 15:25:28 -04002405
David Drysdale51f39a12014-12-12 16:57:29 -08002406extern struct filename *getname_flags(const char __user *, int, int *);
Jeff Layton91a27b22012-10-10 15:25:28 -04002407extern struct filename *getname(const char __user *);
Linus Torvaldsc4ad8f92014-02-05 12:54:53 -08002408extern struct filename *getname_kernel(const char *);
Paul Moore55422d02015-01-22 00:00:23 -05002409extern void putname(struct filename *name);
Jeff Layton91a27b22012-10-10 15:25:28 -04002410
Al Viro47237682012-06-10 05:01:45 -04002411enum {
2412 FILE_CREATED = 1,
2413 FILE_OPENED = 2
2414};
Al Viro30d90492012-06-22 12:40:19 +04002415extern int finish_open(struct file *file, struct dentry *dentry,
2416 int (*open)(struct inode *, struct file *),
2417 int *opened);
Al Viroe45198a2012-06-10 06:48:09 -04002418extern int finish_no_open(struct file *file, struct dentry *dentry);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002419
Ankit Jain3e63cbb2009-06-19 14:28:07 -04002420/* fs/ioctl.c */
2421
2422extern int ioctl_preallocate(struct file *filp, void __user *argp);
2423
Linus Torvalds1da177e2005-04-16 15:20:36 -07002424/* fs/dcache.c */
2425extern void __init vfs_caches_init_early(void);
Mel Gorman4248b0d2015-08-06 15:46:20 -07002426extern void __init vfs_caches_init(void);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002427
Christoph Lameterb86c0892006-12-06 20:32:57 -08002428extern struct kmem_cache *names_cachep;
2429
Jeff Laytona608ca22012-10-10 15:25:26 -04002430#define __getname() kmem_cache_alloc(names_cachep, GFP_KERNEL)
Vegard Nossum3446a8a2009-05-16 11:22:14 +02002431#define __putname(name) kmem_cache_free(names_cachep, (void *)(name))
Linus Torvalds1da177e2005-04-16 15:20:36 -07002432
David Howells93614012006-09-30 20:45:40 +02002433#ifdef CONFIG_BLOCK
Linus Torvalds1da177e2005-04-16 15:20:36 -07002434extern int register_blkdev(unsigned int, const char *);
Akinobu Mitaf4480242007-07-17 04:03:47 -07002435extern void unregister_blkdev(unsigned int, const char *);
Jan Karaf44f1ab2017-02-02 15:56:49 +01002436extern void bdev_unhash_inode(dev_t dev);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002437extern struct block_device *bdget(dev_t);
Alan Jenkinsdddac6a2009-07-29 21:07:55 +02002438extern struct block_device *bdgrab(struct block_device *bdev);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002439extern void bd_set_size(struct block_device *, loff_t size);
2440extern void bd_forget(struct inode *inode);
2441extern void bdput(struct block_device *);
Nick Piggin585d3bc2009-02-25 10:44:19 +01002442extern void invalidate_bdev(struct block_device *);
Jan Kara5c0d6b62012-07-03 16:45:31 +02002443extern void iterate_bdevs(void (*)(struct block_device *, void *), void *);
Nick Piggin585d3bc2009-02-25 10:44:19 +01002444extern int sync_blockdev(struct block_device *bdev);
Al Viroff01bb42011-09-16 02:31:11 -04002445extern void kill_bdev(struct block_device *);
Nick Piggin585d3bc2009-02-25 10:44:19 +01002446extern struct super_block *freeze_bdev(struct block_device *);
Eric Sandeenc2d75432009-03-31 15:23:46 -07002447extern void emergency_thaw_all(void);
Mateusz Guzik08fdc8a2017-10-03 18:17:41 +02002448extern void emergency_thaw_bdev(struct super_block *sb);
Nick Piggin585d3bc2009-02-25 10:44:19 +01002449extern int thaw_bdev(struct block_device *bdev, struct super_block *sb);
2450extern int fsync_bdev(struct block_device *);
Tejun Heoa212b102015-05-22 17:13:33 -04002451
2452extern struct super_block *blockdev_superblock;
2453
2454static inline bool sb_is_blkdev_sb(struct super_block *sb)
2455{
2456 return sb == blockdev_superblock;
2457}
David Howells93614012006-09-30 20:45:40 +02002458#else
2459static inline void bd_forget(struct inode *inode) {}
Al Viro47e44912009-04-01 07:07:16 -04002460static inline int sync_blockdev(struct block_device *bdev) { return 0; }
Al Viroff01bb42011-09-16 02:31:11 -04002461static inline void kill_bdev(struct block_device *bdev) {}
Al Viro47e44912009-04-01 07:07:16 -04002462static inline void invalidate_bdev(struct block_device *bdev) {}
2463
2464static inline struct super_block *freeze_bdev(struct block_device *sb)
2465{
2466 return NULL;
2467}
2468
2469static inline int thaw_bdev(struct block_device *bdev, struct super_block *sb)
2470{
2471 return 0;
2472}
Jan Kara5c0d6b62012-07-03 16:45:31 +02002473
Mateusz Guzik08fdc8a2017-10-03 18:17:41 +02002474static inline int emergency_thaw_bdev(struct super_block *sb)
2475{
2476 return 0;
2477}
2478
Jan Kara5c0d6b62012-07-03 16:45:31 +02002479static inline void iterate_bdevs(void (*f)(struct block_device *, void *), void *arg)
2480{
2481}
Jan Karaa8855992013-07-09 22:36:45 +08002482
Yaowei Bai3e004ee2015-11-17 14:40:09 +08002483static inline bool sb_is_blkdev_sb(struct super_block *sb)
Jan Karaa8855992013-07-09 22:36:45 +08002484{
Yaowei Bai3e004ee2015-11-17 14:40:09 +08002485 return false;
Jan Karaa8855992013-07-09 22:36:45 +08002486}
David Howells93614012006-09-30 20:45:40 +02002487#endif
Jan Kara60b06802009-04-27 16:43:53 +02002488extern int sync_filesystem(struct super_block *);
David Howells93614012006-09-30 20:45:40 +02002489extern const struct file_operations def_blk_fops;
Arjan van de Ven4b6f5d22006-03-28 01:56:42 -08002490extern const struct file_operations def_chr_fops;
David Howells93614012006-09-30 20:45:40 +02002491#ifdef CONFIG_BLOCK
Linus Torvalds1da177e2005-04-16 15:20:36 -07002492extern int ioctl_by_bdev(struct block_device *, unsigned, unsigned long);
Al Viro56b26ad2008-09-19 03:17:36 -04002493extern int blkdev_ioctl(struct block_device *, fmode_t, unsigned, unsigned long);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002494extern long compat_blkdev_ioctl(struct file *, unsigned, unsigned long);
Tejun Heoe525fd82010-11-13 11:55:17 +01002495extern int blkdev_get(struct block_device *bdev, fmode_t mode, void *holder);
Tejun Heod4d77622010-11-13 11:55:18 +01002496extern struct block_device *blkdev_get_by_path(const char *path, fmode_t mode,
2497 void *holder);
2498extern struct block_device *blkdev_get_by_dev(dev_t dev, fmode_t mode,
2499 void *holder);
Al Viro4385bab2013-05-05 22:11:03 -04002500extern void blkdev_put(struct block_device *bdev, fmode_t mode);
Jarod Wilsonbe324172015-05-06 12:26:22 +08002501extern int __blkdev_reread_part(struct block_device *bdev);
2502extern int blkdev_reread_part(struct block_device *bdev);
2503
Jun'ichi Nomura641dc632006-03-27 01:17:57 -08002504#ifdef CONFIG_SYSFS
Tejun Heoe09b4572010-11-13 11:55:17 +01002505extern int bd_link_disk_holder(struct block_device *bdev, struct gendisk *disk);
Tejun Heo49731ba2011-01-14 18:43:57 +01002506extern void bd_unlink_disk_holder(struct block_device *bdev,
2507 struct gendisk *disk);
Jun'ichi Nomura641dc632006-03-27 01:17:57 -08002508#else
Tejun Heoe09b4572010-11-13 11:55:17 +01002509static inline int bd_link_disk_holder(struct block_device *bdev,
2510 struct gendisk *disk)
2511{
2512 return 0;
2513}
Tejun Heo49731ba2011-01-14 18:43:57 +01002514static inline void bd_unlink_disk_holder(struct block_device *bdev,
2515 struct gendisk *disk)
2516{
2517}
Jun'ichi Nomura641dc632006-03-27 01:17:57 -08002518#endif
David Howells93614012006-09-30 20:45:40 +02002519#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -07002520
2521/* fs/char_dev.c */
Logan Gunthorpe8a932f72017-06-15 14:05:21 -06002522#define CHRDEV_MAJOR_MAX 512
Linus Walleij49db08c2016-02-19 15:36:07 +01002523/* Marks the bottom of the first segment of free char majors */
2524#define CHRDEV_MAJOR_DYN_END 234
Logan Gunthorpea5d31a32017-06-15 14:05:20 -06002525/* Marks the top and bottom of the second segment of free char majors */
2526#define CHRDEV_MAJOR_DYN_EXT_START 511
2527#define CHRDEV_MAJOR_DYN_EXT_END 384
2528
Linus Torvalds1da177e2005-04-16 15:20:36 -07002529extern int alloc_chrdev_region(dev_t *, unsigned, unsigned, const char *);
2530extern int register_chrdev_region(dev_t, unsigned, const char *);
Tejun Heo1905b1b2009-08-06 18:13:23 +09002531extern int __register_chrdev(unsigned int major, unsigned int baseminor,
2532 unsigned int count, const char *name,
2533 const struct file_operations *fops);
2534extern void __unregister_chrdev(unsigned int major, unsigned int baseminor,
2535 unsigned int count, const char *name);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002536extern void unregister_chrdev_region(dev_t, unsigned);
Joe Korty68eef3b2006-03-31 02:30:32 -08002537extern void chrdev_show(struct seq_file *,off_t);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002538
Tejun Heo1905b1b2009-08-06 18:13:23 +09002539static inline int register_chrdev(unsigned int major, const char *name,
2540 const struct file_operations *fops)
2541{
2542 return __register_chrdev(major, 0, 256, name, fops);
2543}
2544
2545static inline void unregister_chrdev(unsigned int major, const char *name)
2546{
2547 __unregister_chrdev(major, 0, 256, name);
2548}
2549
Linus Torvalds1da177e2005-04-16 15:20:36 -07002550/* fs/block_dev.c */
2551#define BDEVNAME_SIZE 32 /* Largest string for a blockdev identifier */
Tejun Heo1f014292008-08-25 19:47:23 +09002552#define BDEVT_SIZE 10 /* Largest string for MAJ:MIN for blkdev */
David Howells93614012006-09-30 20:45:40 +02002553
2554#ifdef CONFIG_BLOCK
Logan Gunthorpe133d55c2017-06-16 17:48:21 -06002555#define BLKDEV_MAJOR_MAX 512
Linus Torvalds1da177e2005-04-16 15:20:36 -07002556extern const char *__bdevname(dev_t, char *buffer);
2557extern const char *bdevname(struct block_device *bdev, char *buffer);
2558extern struct block_device *lookup_bdev(const char *);
Joe Korty68eef3b2006-03-31 02:30:32 -08002559extern void blkdev_show(struct seq_file *,off_t);
Al Viro30c40d22008-02-22 19:50:45 -05002560
David Howells93614012006-09-30 20:45:40 +02002561#else
Logan Gunthorpe133d55c2017-06-16 17:48:21 -06002562#define BLKDEV_MAJOR_MAX 0
David Howells93614012006-09-30 20:45:40 +02002563#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -07002564
2565extern void init_special_inode(struct inode *, umode_t, dev_t);
2566
2567/* Invalid inode operations -- fs/bad_inode.c */
2568extern void make_bad_inode(struct inode *);
Yaowei Bai0e3ef1f2015-11-19 21:00:11 +08002569extern bool is_bad_inode(struct inode *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002570
Jens Axboec11f0c02016-08-05 08:11:04 -06002571#ifdef CONFIG_BLOCK
Andrew Pattersonc3279d12008-09-04 14:27:25 -06002572extern void check_disk_size_change(struct gendisk *disk,
2573 struct block_device *bdev);
Andrew Patterson0c002c22008-09-04 14:27:20 -06002574extern int revalidate_disk(struct gendisk *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002575extern int check_disk_change(struct block_device *);
NeilBrown93b270f2011-02-24 17:25:47 +11002576extern int __invalidate_device(struct block_device *, bool);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002577extern int invalidate_partition(struct gendisk *, int);
David Howells93614012006-09-30 20:45:40 +02002578#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -07002579unsigned long invalidate_mapping_pages(struct address_space *mapping,
2580 pgoff_t start, pgoff_t end);
Anton Altaparmakov54bc4852007-02-10 01:45:38 -08002581
Linus Torvalds1da177e2005-04-16 15:20:36 -07002582static inline void invalidate_remote_inode(struct inode *inode)
2583{
2584 if (S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) ||
2585 S_ISLNK(inode->i_mode))
Andrew Mortonfc0ecff2007-02-10 01:45:39 -08002586 invalidate_mapping_pages(inode->i_mapping, 0, -1);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002587}
2588extern int invalidate_inode_pages2(struct address_space *mapping);
2589extern int invalidate_inode_pages2_range(struct address_space *mapping,
2590 pgoff_t start, pgoff_t end);
2591extern int write_inode_now(struct inode *, int);
2592extern int filemap_fdatawrite(struct address_space *);
2593extern int filemap_flush(struct address_space *);
Jeff Layton76341ca2017-07-06 07:02:22 -04002594extern int filemap_fdatawait_keep_errors(struct address_space *mapping);
Jan Karad3bccb6f2009-08-17 19:30:27 +02002595extern int filemap_fdatawait_range(struct address_space *, loff_t lstart,
2596 loff_t lend);
Jeff Laytonffb959b2017-07-31 10:29:38 -04002597
2598static inline int filemap_fdatawait(struct address_space *mapping)
2599{
2600 return filemap_fdatawait_range(mapping, 0, LLONG_MAX);
2601}
2602
Goldwyn Rodrigues7fc9e472017-06-20 07:05:41 -05002603extern bool filemap_range_has_page(struct address_space *, loff_t lstart,
2604 loff_t lend);
Jeff Laytona823e452017-07-28 07:24:43 -04002605extern int __must_check file_fdatawait_range(struct file *file, loff_t lstart,
2606 loff_t lend);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002607extern int filemap_write_and_wait(struct address_space *mapping);
2608extern int filemap_write_and_wait_range(struct address_space *mapping,
2609 loff_t lstart, loff_t lend);
Andrew Mortonebcf28e2006-03-24 03:18:04 -08002610extern int __filemap_fdatawrite_range(struct address_space *mapping,
2611 loff_t start, loff_t end, int sync_mode);
Jan Karaf4c0a0f2008-07-11 19:27:31 -04002612extern int filemap_fdatawrite_range(struct address_space *mapping,
2613 loff_t start, loff_t end);
Miklos Szeredid72d9e22016-07-29 14:10:57 +02002614extern int filemap_check_errors(struct address_space *mapping);
Jeff Layton5660e132017-07-06 07:02:25 -04002615extern void __filemap_set_wb_err(struct address_space *mapping, int err);
Jeff Laytona823e452017-07-28 07:24:43 -04002616
2617extern int __must_check file_fdatawait_range(struct file *file, loff_t lstart,
2618 loff_t lend);
Jeff Layton5660e132017-07-06 07:02:25 -04002619extern int __must_check file_check_and_advance_wb_err(struct file *file);
2620extern int __must_check file_write_and_wait_range(struct file *file,
2621 loff_t start, loff_t end);
2622
Jeff Laytona823e452017-07-28 07:24:43 -04002623static inline int file_write_and_wait(struct file *file)
2624{
2625 return file_write_and_wait_range(file, 0, LLONG_MAX);
2626}
2627
Jeff Layton5660e132017-07-06 07:02:25 -04002628/**
2629 * filemap_set_wb_err - set a writeback error on an address_space
2630 * @mapping: mapping in which to set writeback error
2631 * @err: error to be set in mapping
2632 *
2633 * When writeback fails in some way, we must record that error so that
2634 * userspace can be informed when fsync and the like are called. We endeavor
2635 * to report errors on any file that was open at the time of the error. Some
2636 * internal callers also need to know when writeback errors have occurred.
2637 *
2638 * When a writeback error occurs, most filesystems will want to call
2639 * filemap_set_wb_err to record the error in the mapping so that it will be
2640 * automatically reported whenever fsync is called on the file.
Jeff Layton5660e132017-07-06 07:02:25 -04002641 */
2642static inline void filemap_set_wb_err(struct address_space *mapping, int err)
2643{
2644 /* Fastpath for common case of no error */
2645 if (unlikely(err))
2646 __filemap_set_wb_err(mapping, err);
2647}
2648
2649/**
2650 * filemap_check_wb_error - has an error occurred since the mark was sampled?
2651 * @mapping: mapping to check for writeback errors
2652 * @since: previously-sampled errseq_t
2653 *
2654 * Grab the errseq_t value from the mapping, and see if it has changed "since"
2655 * the given value was sampled.
2656 *
2657 * If it has then report the latest error set, otherwise return 0.
2658 */
2659static inline int filemap_check_wb_err(struct address_space *mapping,
2660 errseq_t since)
2661{
2662 return errseq_check(&mapping->wb_err, since);
2663}
2664
2665/**
2666 * filemap_sample_wb_err - sample the current errseq_t to test for later errors
2667 * @mapping: mapping to be sampled
2668 *
2669 * Writeback errors are always reported relative to a particular sample point
2670 * in the past. This function provides those sample points.
2671 */
2672static inline errseq_t filemap_sample_wb_err(struct address_space *mapping)
2673{
2674 return errseq_sample(&mapping->wb_err);
2675}
2676
Christoph Hellwig8018ab02010-03-22 17:32:25 +01002677extern int vfs_fsync_range(struct file *file, loff_t start, loff_t end,
2678 int datasync);
2679extern int vfs_fsync(struct file *file, int datasync);
Christoph Hellwige2592212016-04-07 08:52:01 -07002680
2681/*
2682 * Sync the bytes written if this was a synchronous write. Expect ki_pos
2683 * to already be updated for the write, and will return either the amount
2684 * of bytes passed in, or an error if syncing the file failed.
2685 */
2686static inline ssize_t generic_write_sync(struct kiocb *iocb, ssize_t count)
Al Virod311d792014-02-09 15:18:09 -05002687{
Christoph Hellwige2592212016-04-07 08:52:01 -07002688 if (iocb->ki_flags & IOCB_DSYNC) {
2689 int ret = vfs_fsync_range(iocb->ki_filp,
2690 iocb->ki_pos - count, iocb->ki_pos - 1,
2691 (iocb->ki_flags & IOCB_SYNC) ? 0 : 1);
2692 if (ret)
2693 return ret;
2694 }
2695
2696 return count;
Al Virod311d792014-02-09 15:18:09 -05002697}
Christoph Hellwige2592212016-04-07 08:52:01 -07002698
Linus Torvalds1da177e2005-04-16 15:20:36 -07002699extern void emergency_sync(void);
2700extern void emergency_remount(void);
David Howells93614012006-09-30 20:45:40 +02002701#ifdef CONFIG_BLOCK
Linus Torvalds1da177e2005-04-16 15:20:36 -07002702extern sector_t bmap(struct inode *, sector_t);
David Howells93614012006-09-30 20:45:40 +02002703#endif
J. Bruce Fields27ac0ff2011-09-20 17:19:26 -04002704extern int notify_change(struct dentry *, struct iattr *, struct inode **);
Al Virof419a2e2008-07-22 00:07:17 -04002705extern int inode_permission(struct inode *, int);
Al Viro2830ba72011-06-20 19:16:29 -04002706extern int generic_permission(struct inode *, int);
Miklos Szeredicbdf35b2014-10-24 00:14:36 +02002707extern int __check_sticky(struct inode *dir, struct inode *inode);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002708
Miklos Szeredif696a362008-07-31 13:41:58 +02002709static inline bool execute_ok(struct inode *inode)
2710{
2711 return (inode->i_mode & S_IXUGO) || S_ISDIR(inode->i_mode);
2712}
2713
Al Viro8d71db42013-03-19 21:01:03 -04002714static inline void file_start_write(struct file *file)
2715{
2716 if (!S_ISREG(file_inode(file)->i_mode))
2717 return;
2718 __sb_start_write(file_inode(file)->i_sb, SB_FREEZE_WRITE, true);
2719}
2720
Jan Kara5ae98f12013-05-04 00:11:23 +02002721static inline bool file_start_write_trylock(struct file *file)
2722{
2723 if (!S_ISREG(file_inode(file)->i_mode))
2724 return true;
2725 return __sb_start_write(file_inode(file)->i_sb, SB_FREEZE_WRITE, false);
2726}
2727
Al Viro8d71db42013-03-19 21:01:03 -04002728static inline void file_end_write(struct file *file)
2729{
2730 if (!S_ISREG(file_inode(file)->i_mode))
2731 return;
2732 __sb_end_write(file_inode(file)->i_sb, SB_FREEZE_WRITE);
2733}
2734
Amir Goldsteinbfe219d2017-01-31 10:34:57 +02002735static inline int do_clone_file_range(struct file *file_in, loff_t pos_in,
2736 struct file *file_out, loff_t pos_out,
2737 u64 len)
2738{
2739 int ret;
2740
2741 file_start_write(file_out);
2742 ret = vfs_clone_file_range(file_in, pos_in, file_out, pos_out, len);
2743 file_end_write(file_out);
2744
2745 return ret;
2746}
2747
Al Viro07b8ce12011-06-20 10:52:57 -04002748/*
2749 * get_write_access() gets write permission for a file.
2750 * put_write_access() releases this write permission.
2751 * This is used for regular files.
2752 * We cannot support write (and maybe mmap read-write shared) accesses and
2753 * MAP_DENYWRITE mmappings simultaneously. The i_writecount field of an inode
2754 * can have the following values:
2755 * 0: no writers, no VM_DENYWRITE mappings
2756 * < 0: (-i_writecount) vm_area_structs with VM_DENYWRITE set exist
2757 * > 0: (i_writecount) users are writing to the file.
2758 *
2759 * Normally we operate on that counter with atomic_{inc,dec} and it's safe
2760 * except for the cases where we don't hold i_writecount yet. Then we need to
2761 * use {get,deny}_write_access() - these functions check the sign and refuse
2762 * to do the change if sign is wrong.
2763 */
2764static inline int get_write_access(struct inode *inode)
2765{
2766 return atomic_inc_unless_negative(&inode->i_writecount) ? 0 : -ETXTBSY;
2767}
2768static inline int deny_write_access(struct file *file)
2769{
Al Viro496ad9a2013-01-23 17:07:38 -05002770 struct inode *inode = file_inode(file);
Al Viro07b8ce12011-06-20 10:52:57 -04002771 return atomic_dec_unless_positive(&inode->i_writecount) ? 0 : -ETXTBSY;
2772}
Linus Torvalds1da177e2005-04-16 15:20:36 -07002773static inline void put_write_access(struct inode * inode)
2774{
2775 atomic_dec(&inode->i_writecount);
2776}
2777static inline void allow_write_access(struct file *file)
2778{
2779 if (file)
Al Viro496ad9a2013-01-23 17:07:38 -05002780 atomic_inc(&file_inode(file)->i_writecount);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002781}
David Howellsf1fe29b2013-09-27 11:20:03 +01002782static inline bool inode_is_open_for_write(const struct inode *inode)
2783{
2784 return atomic_read(&inode->i_writecount) > 0;
2785}
2786
Mimi Zohara5c96eb2010-11-02 10:11:37 -04002787#ifdef CONFIG_IMA
2788static inline void i_readcount_dec(struct inode *inode)
2789{
2790 BUG_ON(!atomic_read(&inode->i_readcount));
2791 atomic_dec(&inode->i_readcount);
2792}
2793static inline void i_readcount_inc(struct inode *inode)
2794{
2795 atomic_inc(&inode->i_readcount);
2796}
2797#else
2798static inline void i_readcount_dec(struct inode *inode)
2799{
2800 return;
2801}
2802static inline void i_readcount_inc(struct inode *inode)
2803{
2804 return;
2805}
2806#endif
Ulrich Dreppered8cae82008-07-23 21:29:30 -07002807extern int do_pipe_flags(int *, int);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002808
Mimi Zohar1284ab52016-04-20 15:46:27 -07002809#define __kernel_read_file_id(id) \
2810 id(UNKNOWN, unknown) \
2811 id(FIRMWARE, firmware) \
Stephen Boyda098ecd2016-08-02 14:04:28 -07002812 id(FIRMWARE_PREALLOC_BUFFER, firmware) \
Mimi Zohar1284ab52016-04-20 15:46:27 -07002813 id(MODULE, kernel-module) \
2814 id(KEXEC_IMAGE, kexec-image) \
2815 id(KEXEC_INITRAMFS, kexec-initramfs) \
2816 id(POLICY, security-policy) \
Christoph Hellwiga7d3d032017-09-10 09:49:45 +02002817 id(X509_CERTIFICATE, x509-certificate) \
Mimi Zohar1284ab52016-04-20 15:46:27 -07002818 id(MAX_ID, )
2819
2820#define __fid_enumify(ENUM, dummy) READING_ ## ENUM,
2821#define __fid_stringify(dummy, str) #str,
2822
Mimi Zoharbc8ca5b2016-01-24 10:07:32 -05002823enum kernel_read_file_id {
Mimi Zohar1284ab52016-04-20 15:46:27 -07002824 __kernel_read_file_id(__fid_enumify)
Mimi Zoharbc8ca5b2016-01-24 10:07:32 -05002825};
2826
Mimi Zohar1284ab52016-04-20 15:46:27 -07002827static const char * const kernel_read_file_str[] = {
2828 __kernel_read_file_id(__fid_stringify)
2829};
2830
Kees Cookda20dfe2016-04-21 12:53:29 -07002831static inline const char *kernel_read_file_id_str(enum kernel_read_file_id id)
Mimi Zohar1284ab52016-04-20 15:46:27 -07002832{
Arnd Bergmanncbfd0c12017-03-09 16:16:57 -08002833 if ((unsigned)id >= READING_MAX_ID)
Mimi Zohar1284ab52016-04-20 15:46:27 -07002834 return kernel_read_file_str[READING_UNKNOWN];
2835
2836 return kernel_read_file_str[id];
2837}
2838
Mimi Zoharbc8ca5b2016-01-24 10:07:32 -05002839extern int kernel_read_file(struct file *, void **, loff_t *, loff_t,
2840 enum kernel_read_file_id);
Mimi Zohar711aab12017-09-12 22:45:33 -04002841extern int kernel_read_file_from_path(const char *, void **, loff_t *, loff_t,
Mimi Zohar09596b92015-11-19 12:39:22 -05002842 enum kernel_read_file_id);
Mimi Zoharb844f0e2016-02-01 08:36:21 -05002843extern int kernel_read_file_from_fd(int, void **, loff_t *, loff_t,
2844 enum kernel_read_file_id);
Christoph Hellwigbdd1d2d2017-09-01 17:39:13 +02002845extern ssize_t kernel_read(struct file *, void *, size_t, loff_t *);
Christoph Hellwige13ec932017-09-01 17:39:14 +02002846extern ssize_t kernel_write(struct file *, const void *, size_t, loff_t *);
Christoph Hellwig73e18f72017-09-01 17:39:15 +02002847extern ssize_t __kernel_write(struct file *, const void *, size_t, loff_t *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002848extern struct file * open_exec(const char *);
2849
2850/* fs/dcache.c -- generic fs support functions */
Yaowei Baia6e57872015-11-17 14:40:11 +08002851extern bool is_subdir(struct dentry *, struct dentry *);
Mickaël Salaün640eb7e2016-11-14 22:14:35 +01002852extern bool path_is_under(const struct path *, const struct path *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002853
Miklos Szeredi9bf39ab2015-06-19 10:29:13 +02002854extern char *file_path(struct file *, char *, int);
2855
Linus Torvalds1da177e2005-04-16 15:20:36 -07002856#include <linux/err.h>
2857
2858/* needed for stackable file system support */
Andrew Morton965c8e52012-12-17 15:59:39 -08002859extern loff_t default_llseek(struct file *file, loff_t offset, int whence);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002860
Andrew Morton965c8e52012-12-17 15:59:39 -08002861extern loff_t vfs_llseek(struct file *file, loff_t offset, int whence);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002862
Christoph Hellwig54e34622009-08-07 14:38:25 -03002863extern int inode_init_always(struct super_block *, struct inode *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002864extern void inode_init_once(struct inode *);
Miklos Szeredi2aa15892011-02-23 13:49:47 +01002865extern void address_space_init_once(struct address_space *mapping);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002866extern struct inode * igrab(struct inode *);
2867extern ino_t iunique(struct super_block *, ino_t);
2868extern int inode_needs_sync(struct inode *inode);
Al Viro45321ac2010-06-07 13:43:19 -04002869extern int generic_delete_inode(struct inode *inode);
Al Viro9bcb4b72012-02-12 19:43:17 -05002870static inline int generic_drop_inode(struct inode *inode)
2871{
2872 return !inode->i_nlink || inode_unhashed(inode);
2873}
Linus Torvalds1da177e2005-04-16 15:20:36 -07002874
Anton Altaparmakov88bd5122005-07-13 01:10:44 -07002875extern struct inode *ilookup5_nowait(struct super_block *sb,
2876 unsigned long hashval, int (*test)(struct inode *, void *),
2877 void *data);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002878extern struct inode *ilookup5(struct super_block *sb, unsigned long hashval,
2879 int (*test)(struct inode *, void *), void *data);
2880extern struct inode *ilookup(struct super_block *sb, unsigned long ino);
2881
2882extern struct inode * iget5_locked(struct super_block *, unsigned long, int (*test)(struct inode *, void *), int (*set)(struct inode *, void *), void *);
2883extern struct inode * iget_locked(struct super_block *, unsigned long);
Theodore Ts'ofe032c42015-02-02 00:37:01 -05002884extern struct inode *find_inode_nowait(struct super_block *,
2885 unsigned long,
2886 int (*match)(struct inode *,
2887 unsigned long, void *),
2888 void *data);
Al Viro261bca82008-12-30 01:48:21 -05002889extern int insert_inode_locked4(struct inode *, unsigned long, int (*test)(struct inode *, void *), void *);
2890extern int insert_inode_locked(struct inode *);
Josh Boyere096d0c2011-08-25 07:48:12 -04002891#ifdef CONFIG_DEBUG_LOCK_ALLOC
2892extern void lockdep_annotate_inode_mutex_key(struct inode *inode);
2893#else
2894static inline void lockdep_annotate_inode_mutex_key(struct inode *inode) { };
2895#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -07002896extern void unlock_new_inode(struct inode *);
Christoph Hellwig85fe4022010-10-23 11:19:54 -04002897extern unsigned int get_next_ino(void);
Darrick J. Wong799ea9e2017-08-18 18:08:25 -07002898extern void evict_inodes(struct super_block *sb);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002899
Linus Torvalds1da177e2005-04-16 15:20:36 -07002900extern void __iget(struct inode * inode);
David Howellsb46980f2008-02-07 00:15:27 -08002901extern void iget_failed(struct inode *);
Jan Karadbd57682012-05-03 14:48:02 +02002902extern void clear_inode(struct inode *);
Christoph Hellwig2e00c972009-08-07 14:38:29 -03002903extern void __destroy_inode(struct inode *);
Eric Dumazeta209dfc2011-07-26 11:36:34 +02002904extern struct inode *new_inode_pseudo(struct super_block *sb);
2905extern struct inode *new_inode(struct super_block *sb);
Nick Pigginff0c7d12011-01-07 17:49:50 +11002906extern void free_inode_nonrcu(struct inode *inode);
Jens Axboe01de85e2006-10-17 19:50:36 +02002907extern int should_remove_suid(struct dentry *);
Jan Kara5fa8e0a2015-05-21 16:05:53 +02002908extern int file_remove_privs(struct file *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002909
2910extern void __insert_inode_hash(struct inode *, unsigned long hashval);
Christoph Hellwig646ec462010-10-23 07:15:32 -04002911static inline void insert_inode_hash(struct inode *inode)
2912{
Linus Torvalds1da177e2005-04-16 15:20:36 -07002913 __insert_inode_hash(inode, inode->i_ino);
2914}
Eric Dumazetf2ee7ab2011-07-28 06:41:09 +02002915
2916extern void __remove_inode_hash(struct inode *);
2917static inline void remove_inode_hash(struct inode *inode)
2918{
Josef Bacikcbedaac2015-03-12 08:19:11 -04002919 if (!inode_unhashed(inode) && !hlist_fake(&inode->i_hash))
Eric Dumazetf2ee7ab2011-07-28 06:41:09 +02002920 __remove_inode_hash(inode);
2921}
2922
Christoph Hellwig646ec462010-10-23 07:15:32 -04002923extern void inode_sb_list_add(struct inode *inode);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002924
David Howells93614012006-09-30 20:45:40 +02002925#ifdef CONFIG_BLOCK
Linus Torvalds1da177e2005-04-16 15:20:36 -07002926extern int bdev_read_only(struct block_device *);
David Howells93614012006-09-30 20:45:40 +02002927#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -07002928extern int set_blocksize(struct block_device *, int);
2929extern int sb_set_blocksize(struct super_block *, int);
2930extern int sb_min_blocksize(struct super_block *, int);
2931
2932extern int generic_file_mmap(struct file *, struct vm_area_struct *);
2933extern int generic_file_readonly_mmap(struct file *, struct vm_area_struct *);
Al Viro3309dd02015-04-09 12:55:47 -04002934extern ssize_t generic_write_checks(struct kiocb *, struct iov_iter *);
Al Viroed978a82014-03-05 22:53:04 -05002935extern ssize_t generic_file_read_iter(struct kiocb *, struct iov_iter *);
Al Viro81742022014-04-03 03:17:43 -04002936extern ssize_t __generic_file_write_iter(struct kiocb *, struct iov_iter *);
Al Viro81742022014-04-03 03:17:43 -04002937extern ssize_t generic_file_write_iter(struct kiocb *, struct iov_iter *);
Christoph Hellwig1af5bb42016-04-07 08:51:56 -07002938extern ssize_t generic_file_direct_write(struct kiocb *, struct iov_iter *);
Al Viro3b93f912014-02-11 21:34:08 -05002939extern ssize_t generic_perform_write(struct file *, struct iov_iter *, loff_t);
Andrew Morton88e6fae2006-04-11 13:59:36 +02002940
Christoph Hellwig18e97102017-05-27 11:16:51 +03002941ssize_t vfs_iter_read(struct file *file, struct iov_iter *iter, loff_t *ppos,
Christoph Hellwigddef7ed22017-07-06 18:58:37 +02002942 rwf_t flags);
Christoph Hellwigabbb65892017-05-27 11:16:52 +03002943ssize_t vfs_iter_write(struct file *file, struct iov_iter *iter, loff_t *ppos,
Christoph Hellwigddef7ed22017-07-06 18:58:37 +02002944 rwf_t flags);
Christoph Hellwigdbe4e192015-01-25 21:11:59 +01002945
Christoph Hellwigeef99382009-08-20 17:43:41 +02002946/* fs/block_dev.c */
David Jefferyb2de5252014-09-29 10:21:10 -04002947extern ssize_t blkdev_read_iter(struct kiocb *iocb, struct iov_iter *to);
Al Viro1456c0a2014-04-03 03:21:50 -04002948extern ssize_t blkdev_write_iter(struct kiocb *iocb, struct iov_iter *from);
Josef Bacik02c24a82011-07-16 20:44:56 -04002949extern int blkdev_fsync(struct file *filp, loff_t start, loff_t end,
2950 int datasync);
Al Viroff01bb42011-09-16 02:31:11 -04002951extern void block_sync_page(struct page *page);
Christoph Hellwigeef99382009-08-20 17:43:41 +02002952
Andrew Morton88e6fae2006-04-11 13:59:36 +02002953/* fs/splice.c */
Jens Axboecbb7e572006-04-11 14:57:50 +02002954extern ssize_t generic_file_splice_read(struct file *, loff_t *,
Andrew Morton88e6fae2006-04-11 13:59:36 +02002955 struct pipe_inode_info *, size_t, unsigned int);
Al Viro8d020762014-04-05 04:27:08 -04002956extern ssize_t iter_file_splice_write(struct pipe_inode_info *,
Jens Axboecbb7e572006-04-11 14:57:50 +02002957 struct file *, loff_t *, size_t, unsigned int);
Andrew Morton88e6fae2006-04-11 13:59:36 +02002958extern ssize_t generic_splice_sendpage(struct pipe_inode_info *pipe,
Jens Axboecbb7e572006-04-11 14:57:50 +02002959 struct file *out, loff_t *, size_t len, unsigned int flags);
Miklos Szeredi1c118592014-10-24 00:14:35 +02002960extern long do_splice_direct(struct file *in, loff_t *ppos, struct file *out,
2961 loff_t *opos, size_t len, unsigned int flags);
2962
Andrew Morton88e6fae2006-04-11 13:59:36 +02002963
Linus Torvalds1da177e2005-04-16 15:20:36 -07002964extern void
2965file_ra_state_init(struct file_ra_state *ra, struct address_space *mapping);
Andrew Morton965c8e52012-12-17 15:59:39 -08002966extern loff_t noop_llseek(struct file *file, loff_t offset, int whence);
2967extern loff_t no_llseek(struct file *file, loff_t offset, int whence);
Jie Liu46a1c2c2013-06-25 12:02:13 +08002968extern loff_t vfs_setpos(struct file *file, loff_t offset, loff_t maxsize);
Andrew Morton965c8e52012-12-17 15:59:39 -08002969extern loff_t generic_file_llseek(struct file *file, loff_t offset, int whence);
Andi Kleen57604952011-09-15 16:06:50 -07002970extern loff_t generic_file_llseek_size(struct file *file, loff_t offset,
Andrew Morton965c8e52012-12-17 15:59:39 -08002971 int whence, loff_t maxsize, loff_t eof);
Al Viro1bf9d142013-06-16 20:27:42 +04002972extern loff_t fixed_size_llseek(struct file *file, loff_t offset,
2973 int whence, loff_t size);
Al Virob25472f2015-12-05 22:04:48 -05002974extern loff_t no_seek_end_llseek_size(struct file *, loff_t, int, loff_t);
2975extern loff_t no_seek_end_llseek(struct file *, loff_t, int);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002976extern int generic_file_open(struct inode * inode, struct file * filp);
2977extern int nonseekable_open(struct inode * inode, struct file * filp);
2978
David Howells93614012006-09-30 20:45:40 +02002979#ifdef CONFIG_BLOCK
Mike Christie8a4c1e42016-06-05 14:31:50 -05002980typedef void (dio_submit_t)(struct bio *bio, struct inode *inode,
Josef Bacikfacd07b2010-05-23 11:00:55 -04002981 loff_t file_offset);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002982
2983enum {
Christoph Hellwig1e431f52009-11-03 16:44:53 +01002984 /* need locking between buffered and direct access */
2985 DIO_LOCKING = 0x01,
2986
2987 /* filesystem does not support filling holes */
2988 DIO_SKIP_HOLES = 0x02,
Linus Torvalds1da177e2005-04-16 15:20:36 -07002989};
2990
Christoph Hellwig40553512017-06-03 09:37:58 +02002991void dio_end_io(struct bio *bio);
Darrick J. Wong5a9d9292018-01-08 10:41:39 -08002992void dio_warn_stale_pagecache(struct file *filp);
npiggin@suse.de7bb46a62010-05-27 01:05:33 +10002993
Omar Sandoval17f8c842015-03-16 04:33:50 -07002994ssize_t __blockdev_direct_IO(struct kiocb *iocb, struct inode *inode,
2995 struct block_device *bdev, struct iov_iter *iter,
Christoph Hellwigc8b8e322016-04-07 08:51:58 -07002996 get_block_t get_block,
Omar Sandoval17f8c842015-03-16 04:33:50 -07002997 dio_iodone_t end_io, dio_submit_t submit_io,
2998 int flags);
Christoph Hellwigeafdc7d2010-06-04 11:29:53 +02002999
Omar Sandoval17f8c842015-03-16 04:33:50 -07003000static inline ssize_t blockdev_direct_IO(struct kiocb *iocb,
3001 struct inode *inode,
Christoph Hellwigc8b8e322016-04-07 08:51:58 -07003002 struct iov_iter *iter,
Omar Sandoval17f8c842015-03-16 04:33:50 -07003003 get_block_t get_block)
Linus Torvalds1da177e2005-04-16 15:20:36 -07003004{
Omar Sandoval17f8c842015-03-16 04:33:50 -07003005 return __blockdev_direct_IO(iocb, inode, inode->i_sb->s_bdev, iter,
Christoph Hellwigc8b8e322016-04-07 08:51:58 -07003006 get_block, NULL, NULL, DIO_LOCKING | DIO_SKIP_HOLES);
Linus Torvalds1da177e2005-04-16 15:20:36 -07003007}
David Howells93614012006-09-30 20:45:40 +02003008#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -07003009
Trond Myklebust1d59d612012-05-31 12:22:33 -04003010void inode_dio_wait(struct inode *inode);
Jens Axboefe0f07d2015-04-15 17:05:48 -06003011
3012/*
3013 * inode_dio_begin - signal start of a direct I/O requests
3014 * @inode: inode the direct I/O happens on
3015 *
3016 * This is called once we've finished processing a direct I/O request,
3017 * and is used to wake up callers waiting for direct I/O to be quiesced.
3018 */
3019static inline void inode_dio_begin(struct inode *inode)
3020{
3021 atomic_inc(&inode->i_dio_count);
3022}
3023
3024/*
3025 * inode_dio_end - signal finish of a direct I/O requests
3026 * @inode: inode the direct I/O happens on
3027 *
3028 * This is called once we've finished processing a direct I/O request,
3029 * and is used to wake up callers waiting for direct I/O to be quiesced.
3030 */
3031static inline void inode_dio_end(struct inode *inode)
3032{
3033 if (atomic_dec_and_test(&inode->i_dio_count))
3034 wake_up_bit(&inode->i_state, __I_DIO_WAKEUP);
3035}
Trond Myklebust1d59d612012-05-31 12:22:33 -04003036
Theodore Ts'o5f16f322014-03-24 14:43:12 -04003037extern void inode_set_flags(struct inode *inode, unsigned int flags,
3038 unsigned int mask);
3039
Arjan van de Ven4b6f5d22006-03-28 01:56:42 -08003040extern const struct file_operations generic_ro_fops;
Linus Torvalds1da177e2005-04-16 15:20:36 -07003041
3042#define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
3043
Al Viro5d826c82014-03-14 13:42:45 -04003044extern int readlink_copy(char __user *, int, const char *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07003045extern int page_readlink(struct dentry *, char __user *, int);
Al Virofceef392015-12-29 15:58:39 -05003046extern const char *page_get_link(struct dentry *, struct inode *,
3047 struct delayed_call *);
3048extern void page_put_link(void *);
Kirill Korotaev0adb25d2006-03-11 03:27:13 -08003049extern int __page_symlink(struct inode *inode, const char *symname, int len,
Nick Piggin54566b22009-01-04 12:00:53 -08003050 int nofs);
Linus Torvalds1da177e2005-04-16 15:20:36 -07003051extern int page_symlink(struct inode *inode, const char *symname, int len);
Arjan van de Venc5ef1c42007-02-12 00:55:40 -08003052extern const struct inode_operations page_symlink_inode_operations;
Al Virofceef392015-12-29 15:58:39 -05003053extern void kfree_link(void *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07003054extern void generic_fillattr(struct inode *, struct kstat *);
David Howellsa528d352017-01-31 16:46:22 +00003055extern int vfs_getattr_nosec(const struct path *, struct kstat *, u32, unsigned int);
3056extern int vfs_getattr(const struct path *, struct kstat *, u32, unsigned int);
Dmitry Monakhovb4627072009-12-14 15:21:12 +03003057void __inode_add_bytes(struct inode *inode, loff_t bytes);
Linus Torvalds1da177e2005-04-16 15:20:36 -07003058void inode_add_bytes(struct inode *inode, loff_t bytes);
Jan Kara1c8924e2013-08-17 09:32:32 -04003059void __inode_sub_bytes(struct inode *inode, loff_t bytes);
Linus Torvalds1da177e2005-04-16 15:20:36 -07003060void inode_sub_bytes(struct inode *inode, loff_t bytes);
Jan Karaf4a8116a2017-08-08 09:54:36 +02003061static inline loff_t __inode_get_bytes(struct inode *inode)
3062{
3063 return (((loff_t)inode->i_blocks) << 9) + inode->i_bytes;
3064}
Linus Torvalds1da177e2005-04-16 15:20:36 -07003065loff_t inode_get_bytes(struct inode *inode);
3066void inode_set_bytes(struct inode *inode, loff_t bytes);
Al Virofceef392015-12-29 15:58:39 -05003067const char *simple_get_link(struct dentry *, struct inode *,
3068 struct delayed_call *);
Al Viro61ba64f2015-05-02 09:54:06 -04003069extern const struct inode_operations simple_symlink_inode_operations;
Linus Torvalds1da177e2005-04-16 15:20:36 -07003070
Al Viro5c0ba4e2013-05-15 13:52:59 -04003071extern int iterate_dir(struct file *, struct dir_context *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07003072
David Howellsa528d352017-01-31 16:46:22 +00003073extern int vfs_statx(int, const char __user *, int, struct kstat *, u32);
3074extern int vfs_statx_fd(unsigned int, struct kstat *, u32, unsigned int);
3075
3076static inline int vfs_stat(const char __user *filename, struct kstat *stat)
3077{
David Howellsdeccf492017-05-04 23:30:16 +01003078 return vfs_statx(AT_FDCWD, filename, AT_NO_AUTOMOUNT,
3079 stat, STATX_BASIC_STATS);
David Howellsa528d352017-01-31 16:46:22 +00003080}
3081static inline int vfs_lstat(const char __user *name, struct kstat *stat)
3082{
David Howellsdeccf492017-05-04 23:30:16 +01003083 return vfs_statx(AT_FDCWD, name, AT_SYMLINK_NOFOLLOW | AT_NO_AUTOMOUNT,
David Howellsa528d352017-01-31 16:46:22 +00003084 stat, STATX_BASIC_STATS);
3085}
3086static inline int vfs_fstatat(int dfd, const char __user *filename,
3087 struct kstat *stat, int flags)
3088{
Ian Kent5d38f042017-11-29 16:11:26 -08003089 return vfs_statx(dfd, filename, flags | AT_NO_AUTOMOUNT,
3090 stat, STATX_BASIC_STATS);
David Howellsa528d352017-01-31 16:46:22 +00003091}
3092static inline int vfs_fstat(int fd, struct kstat *stat)
3093{
3094 return vfs_statx_fd(fd, stat, STATX_BASIC_STATS, 0);
3095}
3096
3097
Miklos Szeredid60874c2016-10-04 14:40:45 +02003098extern const char *vfs_get_link(struct dentry *, struct delayed_call *);
Miklos Szeredifd4a0ed2016-12-09 16:45:04 +01003099extern int vfs_readlink(struct dentry *, char __user *, int);
Linus Torvalds1da177e2005-04-16 15:20:36 -07003100
Steven Whitehousee9079cc2008-10-14 14:43:29 +01003101extern int __generic_block_fiemap(struct inode *inode,
Josef Bacik3a3076f2010-04-23 12:17:17 -04003102 struct fiemap_extent_info *fieinfo,
3103 loff_t start, loff_t len,
3104 get_block_t *get_block);
Josef Bacik68c9d702008-10-03 17:32:43 -04003105extern int generic_block_fiemap(struct inode *inode,
3106 struct fiemap_extent_info *fieinfo, u64 start,
3107 u64 len, get_block_t *get_block);
Linus Torvalds1da177e2005-04-16 15:20:36 -07003108
David Howellsee416bc2017-07-04 17:25:16 +01003109extern struct file_system_type *get_filesystem(struct file_system_type *fs);
Miklos Szeredic18479f2007-10-18 23:39:11 -07003110extern void put_filesystem(struct file_system_type *fs);
Linus Torvalds1da177e2005-04-16 15:20:36 -07003111extern struct file_system_type *get_fs_type(const char *name);
3112extern struct super_block *get_super(struct block_device *);
Jan Kara6b6dc832012-02-10 11:03:00 +01003113extern struct super_block *get_super_thawed(struct block_device *);
Jan Karaba6379f2016-11-23 12:53:00 +01003114extern struct super_block *get_super_exclusive_thawed(struct block_device *bdev);
Christoph Hellwig45042302009-08-03 23:28:35 +02003115extern struct super_block *get_active_super(struct block_device *bdev);
Linus Torvalds1da177e2005-04-16 15:20:36 -07003116extern void drop_super(struct super_block *sb);
Jan Karaba6379f2016-11-23 12:53:00 +01003117extern void drop_super_exclusive(struct super_block *sb);
Al Viro01a05b32010-03-23 06:06:58 -04003118extern void iterate_supers(void (*)(struct super_block *, void *), void *);
Al Viro43e15cd2011-06-03 20:16:57 -04003119extern void iterate_supers_type(struct file_system_type *,
3120 void (*)(struct super_block *, void *), void *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07003121
3122extern int dcache_dir_open(struct inode *, struct file *);
3123extern int dcache_dir_close(struct inode *, struct file *);
3124extern loff_t dcache_dir_lseek(struct file *, loff_t, int);
Al Viro5f99f4e2013-05-15 20:23:06 -04003125extern int dcache_readdir(struct file *, struct dir_context *);
npiggin@suse.de7bb46a62010-05-27 01:05:33 +10003126extern int simple_setattr(struct dentry *, struct iattr *);
David Howellsa528d352017-01-31 16:46:22 +00003127extern int simple_getattr(const struct path *, struct kstat *, u32, unsigned int);
David Howells726c3342006-06-23 02:02:58 -07003128extern int simple_statfs(struct dentry *, struct kstatfs *);
Stephen Boyd20955e82012-04-05 14:25:09 -07003129extern int simple_open(struct inode *inode, struct file *file);
Linus Torvalds1da177e2005-04-16 15:20:36 -07003130extern int simple_link(struct dentry *, struct inode *, struct dentry *);
3131extern int simple_unlink(struct inode *, struct dentry *);
3132extern int simple_rmdir(struct inode *, struct dentry *);
Miklos Szeredie0e0be82016-09-27 11:03:57 +02003133extern int simple_rename(struct inode *, struct dentry *,
3134 struct inode *, struct dentry *, unsigned int);
Josef Bacik02c24a82011-07-16 20:44:56 -04003135extern int noop_fsync(struct file *, loff_t, loff_t, int);
Dan Williamsf44c7762018-03-07 15:26:44 -08003136extern int noop_set_page_dirty(struct page *page);
3137extern void noop_invalidatepage(struct page *page, unsigned int offset,
3138 unsigned int length);
3139extern ssize_t noop_direct_IO(struct kiocb *iocb, struct iov_iter *iter);
Linus Torvalds1da177e2005-04-16 15:20:36 -07003140extern int simple_empty(struct dentry *);
3141extern int simple_readpage(struct file *file, struct page *page);
Nick Pigginafddba42007-10-16 01:25:01 -07003142extern int simple_write_begin(struct file *file, struct address_space *mapping,
3143 loff_t pos, unsigned len, unsigned flags,
3144 struct page **pagep, void **fsdata);
3145extern int simple_write_end(struct file *file, struct address_space *mapping,
3146 loff_t pos, unsigned len, unsigned copied,
3147 struct page *page, void *fsdata);
Al Virob26d4cd2013-10-25 18:47:37 -04003148extern int always_delete_dentry(const struct dentry *);
Al Viro69878432013-10-02 22:35:11 -04003149extern struct inode *alloc_anon_inode(struct super_block *);
Jeff Laytone6f5c782014-08-22 10:40:25 -04003150extern int simple_nosetlease(struct file *, long, struct file_lock **, void **);
Al Virob26d4cd2013-10-25 18:47:37 -04003151extern const struct dentry_operations simple_dentry_operations;
Linus Torvalds1da177e2005-04-16 15:20:36 -07003152
Al Viro00cd8dd2012-06-10 17:13:09 -04003153extern struct dentry *simple_lookup(struct inode *, struct dentry *, unsigned int flags);
Linus Torvalds1da177e2005-04-16 15:20:36 -07003154extern ssize_t generic_read_dir(struct file *, char __user *, size_t, loff_t *);
Arjan van de Ven4b6f5d22006-03-28 01:56:42 -08003155extern const struct file_operations simple_dir_operations;
Arjan van de Venc5ef1c42007-02-12 00:55:40 -08003156extern const struct inode_operations simple_dir_inode_operations;
Eric W. Biedermanfbabfd02015-05-09 15:54:49 -05003157extern void make_empty_dir_inode(struct inode *inode);
3158extern bool is_empty_dir_inode(struct inode *inode);
Eric Biggerscda37122017-03-25 21:15:37 -07003159struct tree_descr { const char *name; const struct file_operations *ops; int mode; };
Linus Torvalds1da177e2005-04-16 15:20:36 -07003160struct dentry *d_alloc_name(struct dentry *, const char *);
Eric Biggerscda37122017-03-25 21:15:37 -07003161extern int simple_fill_super(struct super_block *, unsigned long,
3162 const struct tree_descr *);
Trond Myklebust1f5ce9e2006-06-09 09:34:16 -04003163extern int simple_pin_fs(struct file_system_type *, struct vfsmount **mount, int *count);
Linus Torvalds1da177e2005-04-16 15:20:36 -07003164extern void simple_release_fs(struct vfsmount **mount, int *count);
3165
Akinobu Mita93b07112008-06-05 22:46:21 -07003166extern ssize_t simple_read_from_buffer(void __user *to, size_t count,
3167 loff_t *ppos, const void *from, size_t available);
Jiri Slaby6a727b42010-05-01 23:51:22 +02003168extern ssize_t simple_write_to_buffer(void *to, size_t available, loff_t *ppos,
3169 const void __user *from, size_t count);
Linus Torvalds1da177e2005-04-16 15:20:36 -07003170
Fabian Frederickac13a822014-06-04 16:06:27 -07003171extern int __generic_file_fsync(struct file *, loff_t, loff_t, int);
Josef Bacik02c24a82011-07-16 20:44:56 -04003172extern int generic_file_fsync(struct file *, loff_t, loff_t, int);
Al Virod5aacad2009-06-07 14:56:44 -04003173
Patrick J. LoPresti30ca22c2010-07-22 15:03:41 -07003174extern int generic_check_addressable(unsigned, u64);
3175
Christoph Lametere965f962006-02-01 03:05:41 -08003176#ifdef CONFIG_MIGRATION
Christoph Lameter2d1db3b2006-06-23 02:03:33 -07003177extern int buffer_migrate_page(struct address_space *,
Mel Gormana6bc32b2012-01-12 17:19:43 -08003178 struct page *, struct page *,
3179 enum migrate_mode);
Christoph Lametere965f962006-02-01 03:05:41 -08003180#else
3181#define buffer_migrate_page NULL
3182#endif
3183
Jan Kara31051c82016-05-26 16:55:18 +02003184extern int setattr_prepare(struct dentry *, struct iattr *);
npiggin@suse.de25d9e2d2009-08-21 02:35:05 +10003185extern int inode_newsize_ok(const struct inode *, loff_t offset);
Christoph Hellwig6a1a90a2010-06-04 11:30:00 +02003186extern void setattr_copy(struct inode *inode, const struct iattr *attr);
Linus Torvalds1da177e2005-04-16 15:20:36 -07003187
Josef Bacikc3b2da32012-03-26 09:59:21 -04003188extern int file_update_time(struct file *file);
Linus Torvalds1da177e2005-04-16 15:20:36 -07003189
Matthew Wilcoxd475c632015-02-16 15:58:56 -08003190static inline bool io_is_direct(struct file *filp)
3191{
Dan Williams65f87ee2016-01-25 17:23:18 -08003192 return (filp->f_flags & O_DIRECT) || IS_DAX(filp->f_mapping->host);
Matthew Wilcoxd475c632015-02-16 15:58:56 -08003193}
3194
Dan Williamsbaabda22017-07-10 15:48:25 -07003195static inline bool vma_is_dax(struct vm_area_struct *vma)
3196{
3197 return vma->vm_file && IS_DAX(vma->vm_file->f_mapping->host);
3198}
3199
Dan Williams2bb6d282017-11-29 16:10:35 -08003200static inline bool vma_is_fsdax(struct vm_area_struct *vma)
3201{
3202 struct inode *inode;
3203
3204 if (!vma->vm_file)
3205 return false;
3206 if (!vma_is_dax(vma))
3207 return false;
3208 inode = file_inode(vma->vm_file);
Dan Williams230f5a82018-02-21 17:08:01 -08003209 if (S_ISCHR(inode->i_mode))
Dan Williams2bb6d282017-11-29 16:10:35 -08003210 return false; /* device-dax */
3211 return true;
3212}
3213
Al Viro2ba48ce2015-04-09 13:52:01 -04003214static inline int iocb_flags(struct file *file)
3215{
3216 int res = 0;
3217 if (file->f_flags & O_APPEND)
3218 res |= IOCB_APPEND;
3219 if (io_is_direct(file))
3220 res |= IOCB_DIRECT;
Christoph Hellwigdde0c2e2016-04-07 08:52:00 -07003221 if ((file->f_flags & O_DSYNC) || IS_SYNC(file->f_mapping->host))
3222 res |= IOCB_DSYNC;
3223 if (file->f_flags & __O_SYNC)
3224 res |= IOCB_SYNC;
Al Viro2ba48ce2015-04-09 13:52:01 -04003225 return res;
3226}
3227
Christoph Hellwigddef7ed22017-07-06 18:58:37 +02003228static inline int kiocb_set_rw_flags(struct kiocb *ki, rwf_t flags)
Goldwyn Rodriguesfdd2f5b2017-06-20 07:05:40 -05003229{
3230 if (unlikely(flags & ~RWF_SUPPORTED))
3231 return -EOPNOTSUPP;
3232
Goldwyn Rodriguesb745faf2017-06-20 07:05:43 -05003233 if (flags & RWF_NOWAIT) {
Christoph Hellwig91f99432017-08-29 16:13:20 +02003234 if (!(ki->ki_filp->f_mode & FMODE_NOWAIT))
Goldwyn Rodriguesb745faf2017-06-20 07:05:43 -05003235 return -EOPNOTSUPP;
3236 ki->ki_flags |= IOCB_NOWAIT;
3237 }
Goldwyn Rodriguesfdd2f5b2017-06-20 07:05:40 -05003238 if (flags & RWF_HIPRI)
3239 ki->ki_flags |= IOCB_HIPRI;
3240 if (flags & RWF_DSYNC)
3241 ki->ki_flags |= IOCB_DSYNC;
3242 if (flags & RWF_SYNC)
3243 ki->ki_flags |= (IOCB_DSYNC | IOCB_SYNC);
Jürg Billetere1fc7422017-09-29 14:07:17 +02003244 if (flags & RWF_APPEND)
3245 ki->ki_flags |= IOCB_APPEND;
Goldwyn Rodriguesfdd2f5b2017-06-20 07:05:40 -05003246 return 0;
3247}
3248
Linus Torvalds1da177e2005-04-16 15:20:36 -07003249static inline ino_t parent_ino(struct dentry *dentry)
3250{
3251 ino_t res;
3252
Nick Pigginb5c84bf2011-01-07 17:49:38 +11003253 /*
3254 * Don't strictly need d_lock here? If the parent ino could change
3255 * then surely we'd have a deeper race in the caller?
3256 */
Linus Torvalds1da177e2005-04-16 15:20:36 -07003257 spin_lock(&dentry->d_lock);
3258 res = dentry->d_parent->d_inode->i_ino;
3259 spin_unlock(&dentry->d_lock);
3260 return res;
3261}
3262
Linus Torvalds1da177e2005-04-16 15:20:36 -07003263/* Transaction based IO helpers */
3264
3265/*
3266 * An argresp is stored in an allocated page and holds the
3267 * size of the argument or response, along with its content
3268 */
3269struct simple_transaction_argresp {
3270 ssize_t size;
3271 char data[0];
3272};
3273
3274#define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
3275
3276char *simple_transaction_get(struct file *file, const char __user *buf,
3277 size_t size);
3278ssize_t simple_transaction_read(struct file *file, char __user *buf,
3279 size_t size, loff_t *pos);
3280int simple_transaction_release(struct inode *inode, struct file *file);
3281
Ingo Molnar76791ab2009-03-25 16:48:35 +01003282void simple_transaction_set(struct file *file, size_t n);
Linus Torvalds1da177e2005-04-16 15:20:36 -07003283
Arnd Bergmannacaefc22005-05-18 14:40:59 +02003284/*
3285 * simple attribute files
3286 *
3287 * These attributes behave similar to those in sysfs:
3288 *
3289 * Writing to an attribute immediately sets a value, an open file can be
3290 * written to multiple times.
3291 *
3292 * Reading from an attribute creates a buffer from the value that might get
3293 * read with multiple read calls. When the attribute has been read
3294 * completely, no further read calls are possible until the file is opened
3295 * again.
3296 *
3297 * All attributes contain a text representation of a numeric value
3298 * that are accessed with the get() and set() functions.
3299 */
3300#define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt) \
3301static int __fops ## _open(struct inode *inode, struct file *file) \
3302{ \
3303 __simple_attr_check_format(__fmt, 0ull); \
3304 return simple_attr_open(inode, file, __get, __set, __fmt); \
3305} \
Alexey Dobriyan828c0952009-10-01 15:43:56 -07003306static const struct file_operations __fops = { \
Arnd Bergmannacaefc22005-05-18 14:40:59 +02003307 .owner = THIS_MODULE, \
3308 .open = __fops ## _open, \
Christoph Hellwig74bedc42008-02-08 04:20:28 -08003309 .release = simple_attr_release, \
Arnd Bergmannacaefc22005-05-18 14:40:59 +02003310 .read = simple_attr_read, \
3311 .write = simple_attr_write, \
Arnd Bergmann1ec55842010-08-15 21:50:52 +02003312 .llseek = generic_file_llseek, \
Joe Perches68be3022014-08-06 16:08:45 -07003313}
Arnd Bergmannacaefc22005-05-18 14:40:59 +02003314
Joe Perchesb9075fa2011-10-31 17:11:33 -07003315static inline __printf(1, 2)
3316void __simple_attr_check_format(const char *fmt, ...)
Arnd Bergmannacaefc22005-05-18 14:40:59 +02003317{
3318 /* don't do anything, just let the compiler check the arguments; */
3319}
3320
3321int simple_attr_open(struct inode *inode, struct file *file,
Christoph Hellwig8b88b092008-02-08 04:20:26 -08003322 int (*get)(void *, u64 *), int (*set)(void *, u64),
Arnd Bergmannacaefc22005-05-18 14:40:59 +02003323 const char *fmt);
Christoph Hellwig74bedc42008-02-08 04:20:28 -08003324int simple_attr_release(struct inode *inode, struct file *file);
Arnd Bergmannacaefc22005-05-18 14:40:59 +02003325ssize_t simple_attr_read(struct file *file, char __user *buf,
3326 size_t len, loff_t *ppos);
3327ssize_t simple_attr_write(struct file *file, const char __user *buf,
3328 size_t len, loff_t *ppos);
3329
Alexey Dobriyan4be28542007-10-16 23:26:21 -07003330struct ctl_table;
Alexey Dobriyan8d65af72009-09-23 15:57:19 -07003331int proc_nr_files(struct ctl_table *table, int write,
Adrian Bunk62239ac2007-07-17 04:03:45 -07003332 void __user *buffer, size_t *lenp, loff_t *ppos);
Christoph Hellwig312d3ca2010-10-10 05:36:23 -04003333int proc_nr_dentry(struct ctl_table *table, int write,
3334 void __user *buffer, size_t *lenp, loff_t *ppos);
Dave Chinnercffbc8a2010-10-23 05:03:02 -04003335int proc_nr_inodes(struct ctl_table *table, int write,
3336 void __user *buffer, size_t *lenp, loff_t *ppos);
Tetsuo Handa38e23c92009-04-09 20:17:52 +09003337int __init get_filesystem_list(char *buf);
Adrian Bunk62239ac2007-07-17 04:03:45 -07003338
Namhyung Kim3cd90ea2011-02-01 15:52:46 -08003339#define __FMODE_EXEC ((__force int) FMODE_EXEC)
Namhyung Kim1a44bc82011-02-01 15:52:46 -08003340#define __FMODE_NONOTIFY ((__force int) FMODE_NONOTIFY)
3341
Al Viro6d125522009-12-24 06:58:56 -05003342#define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE])
Eric Parisecf081d2009-12-17 21:24:25 -05003343#define OPEN_FMODE(flag) ((__force fmode_t)(((flag + 1) & O_ACCMODE) | \
Namhyung Kim1a44bc82011-02-01 15:52:46 -08003344 (flag & __FMODE_NONOTIFY)))
Al Viro53009902009-12-19 10:15:07 -05003345
Yaowei Baid37177b2015-11-19 21:00:12 +08003346static inline bool is_sxid(umode_t mode)
Andi Kleen69b45732011-05-28 08:25:51 -07003347{
3348 return (mode & S_ISUID) || ((mode & S_ISGID) && (mode & S_IXGRP));
3349}
3350
Miklos Szeredicbdf35b2014-10-24 00:14:36 +02003351static inline int check_sticky(struct inode *dir, struct inode *inode)
3352{
3353 if (!(dir->i_mode & S_ISVTX))
3354 return 0;
3355
3356 return __check_sticky(dir, inode);
3357}
3358
Andi Kleen69b45732011-05-28 08:25:51 -07003359static inline void inode_has_no_xattr(struct inode *inode)
3360{
David Howellse462ec52017-07-17 08:45:35 +01003361 if (!is_sxid(inode->i_mode) && (inode->i_sb->s_flags & SB_NOSEC))
Andi Kleen69b45732011-05-28 08:25:51 -07003362 inode->i_flags |= S_NOSEC;
3363}
3364
Al Viroa7400222014-10-21 15:20:42 -04003365static inline bool is_root_inode(struct inode *inode)
3366{
3367 return inode == inode->i_sb->s_root->d_inode;
3368}
3369
Al Viro5f99f4e2013-05-15 20:23:06 -04003370static inline bool dir_emit(struct dir_context *ctx,
3371 const char *name, int namelen,
3372 u64 ino, unsigned type)
3373{
3374 return ctx->actor(ctx, name, namelen, ctx->pos, ino, type) == 0;
3375}
3376static inline bool dir_emit_dot(struct file *file, struct dir_context *ctx)
3377{
3378 return ctx->actor(ctx, ".", 1, ctx->pos,
3379 file->f_path.dentry->d_inode->i_ino, DT_DIR) == 0;
3380}
3381static inline bool dir_emit_dotdot(struct file *file, struct dir_context *ctx)
3382{
3383 return ctx->actor(ctx, "..", 2, ctx->pos,
3384 parent_ino(file->f_path.dentry), DT_DIR) == 0;
3385}
3386static inline bool dir_emit_dots(struct file *file, struct dir_context *ctx)
3387{
3388 if (ctx->pos == 0) {
3389 if (!dir_emit_dot(file, ctx))
3390 return false;
3391 ctx->pos = 1;
3392 }
3393 if (ctx->pos == 1) {
3394 if (!dir_emit_dotdot(file, ctx))
3395 return false;
3396 ctx->pos = 2;
3397 }
3398 return true;
3399}
Al Viro5ded75e2013-05-15 21:02:48 -04003400static inline bool dir_relax(struct inode *inode)
3401{
Al Viro59551022016-01-22 15:40:57 -05003402 inode_unlock(inode);
3403 inode_lock(inode);
Al Viro5ded75e2013-05-15 21:02:48 -04003404 return !IS_DEADDIR(inode);
3405}
Al Viro5f99f4e2013-05-15 20:23:06 -04003406
Al Viroae053272016-05-12 20:36:01 -04003407static inline bool dir_relax_shared(struct inode *inode)
3408{
3409 inode_unlock_shared(inode);
3410 inode_lock_shared(inode);
3411 return !IS_DEADDIR(inode);
3412}
3413
Eric W. Biederman90f85722015-06-29 14:42:03 -05003414extern bool path_noexec(const struct path *path);
Al Viro21fc61c2015-11-17 01:07:57 -05003415extern void inode_nohighmem(struct inode *inode);
Eric W. Biederman90f85722015-06-29 14:42:03 -05003416
Linus Torvalds1da177e2005-04-16 15:20:36 -07003417#endif /* _LINUX_FS_H */