blob: 43db02e9c9fa11bed1b058fba9dd9039ea9c226d [file] [log] [blame]
Linus Torvalds1da177e2005-04-16 15:20:36 -07001#ifndef _LINUX_FS_H
2#define _LINUX_FS_H
3
Linus Torvalds1da177e2005-04-16 15:20:36 -07004
5#include <linux/linkage.h>
6#include <linux/wait.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -07007#include <linux/kdev_t.h>
8#include <linux/dcache.h>
Al Viro3f8206d2008-07-26 03:46:43 -04009#include <linux/path.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070010#include <linux/stat.h>
11#include <linux/cache.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070012#include <linux/list.h>
13#include <linux/radix-tree.h>
Michel Lespinasse6b2dbba2012-10-08 16:31:25 -070014#include <linux/rbtree.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070015#include <linux/init.h>
Al Viro914e2632006-10-18 13:55:46 -040016#include <linux/pid.h>
Paul Gortmaker187f1882011-11-23 20:12:59 -050017#include <linux/bug.h>
Jes Sorensen1b1dcc12006-01-09 15:59:24 -080018#include <linux/mutex.h>
Satyam Sharma3bd858a2007-07-17 15:00:08 +053019#include <linux/capability.h>
Matthew Wilcox6188e102008-04-18 22:21:05 -040020#include <linux/semaphore.h>
Mark Fashehc4b929b2008-10-08 19:44:18 -040021#include <linux/fiemap.h>
Nick Pigginceb5bdc2011-01-07 17:50:05 +110022#include <linux/rculist_bl.h>
Al Viro07b8ce12011-06-20 10:52:57 -040023#include <linux/atomic.h>
Konstantin Khlebnikov83aeead2011-12-08 14:33:54 -080024#include <linux/shrinker.h>
Stephen Rothwellc1aab022012-01-24 11:41:32 +110025#include <linux/migrate_mode.h>
Eric W. Biederman92361632012-02-08 07:07:50 -080026#include <linux/uidgid.h>
Jan Kara5accdf82012-06-12 16:20:34 +020027#include <linux/lockdep.h>
Jens Axboec2b1ad82012-09-27 09:35:03 +020028#include <linux/percpu-rwsem.h>
David Howells08cce05c2012-10-13 10:45:06 +010029#include <linux/blk_types.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070030
Linus Torvalds1da177e2005-04-16 15:20:36 -070031#include <asm/byteorder.h>
David Howells607ca462012-10-13 10:46:48 +010032#include <uapi/linux/fs.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070033
Christoph Hellwiga5694252007-07-17 04:04:28 -070034struct export_operations;
Christoph Hellwiga885c8c2006-01-08 01:02:50 -080035struct hd_geometry;
Linus Torvalds1da177e2005-04-16 15:20:36 -070036struct iovec;
37struct nameidata;
Christoph Hellwig92198f72005-06-23 22:00:59 -070038struct kiocb;
Alexey Dobriyan57cc7212011-01-10 08:18:25 +020039struct kobject;
Linus Torvalds1da177e2005-04-16 15:20:36 -070040struct pipe_inode_info;
41struct poll_table_struct;
42struct kstatfs;
43struct vm_area_struct;
44struct vfsmount;
David Howells745ca242008-11-14 10:39:22 +110045struct cred;
Mel Gormana509bc12012-07-31 16:44:57 -070046struct swap_info_struct;
Cyrill Gorcunov55985dd2012-12-17 16:04:55 -080047struct seq_file;
Linus Torvalds1da177e2005-04-16 15:20:36 -070048
Denis Cheng74bf17c2007-10-16 23:26:30 -070049extern void __init inode_init(void);
Linus Torvalds1da177e2005-04-16 15:20:36 -070050extern void __init inode_init_early(void);
Linus Torvalds1da177e2005-04-16 15:20:36 -070051extern void __init files_init(unsigned long);
52
Jan Engelhardtdded4f42008-12-01 14:34:50 -080053extern struct files_stat_struct files_stat;
Eric Dumazet518de9b2010-10-26 14:22:44 -070054extern unsigned long get_max_files(void);
Jan Engelhardtdded4f42008-12-01 14:34:50 -080055extern int sysctl_nr_open;
56extern struct inodes_stat_t inodes_stat;
57extern int leases_enable, lease_break_time;
Kees Cook800179c2012-07-25 17:29:07 -070058extern int sysctl_protected_symlinks;
59extern int sysctl_protected_hardlinks;
Jan Engelhardtdded4f42008-12-01 14:34:50 -080060
Linus Torvalds1da177e2005-04-16 15:20:36 -070061struct buffer_head;
62typedef int (get_block_t)(struct inode *inode, sector_t iblock,
63 struct buffer_head *bh_result, int create);
Christoph Hellwig92198f72005-06-23 22:00:59 -070064typedef void (dio_iodone_t)(struct kiocb *iocb, loff_t offset,
Christoph Hellwig40e2e972010-07-18 21:17:09 +000065 ssize_t bytes, void *private, int ret,
66 bool is_async);
Linus Torvalds1da177e2005-04-16 15:20:36 -070067
David Howellsbbc10962012-10-15 16:40:35 +010068#define MAY_EXEC 0x00000001
69#define MAY_WRITE 0x00000002
70#define MAY_READ 0x00000004
71#define MAY_APPEND 0x00000008
72#define MAY_ACCESS 0x00000010
73#define MAY_OPEN 0x00000020
74#define MAY_CHDIR 0x00000040
75/* called from RCU mode, don't block */
76#define MAY_NOT_BLOCK 0x00000080
77
78/*
79 * flags in file.f_mode. Note that FMODE_READ and FMODE_WRITE must correspond
80 * to O_WRONLY and O_RDWR via the strange trick in __dentry_open()
81 */
82
83/* file is open for reading */
84#define FMODE_READ ((__force fmode_t)0x1)
85/* file is open for writing */
86#define FMODE_WRITE ((__force fmode_t)0x2)
87/* file is seekable */
88#define FMODE_LSEEK ((__force fmode_t)0x4)
89/* file can be accessed using pread */
90#define FMODE_PREAD ((__force fmode_t)0x8)
91/* file can be accessed using pwrite */
92#define FMODE_PWRITE ((__force fmode_t)0x10)
93/* File is opened for execution with sys_execve / sys_uselib */
94#define FMODE_EXEC ((__force fmode_t)0x20)
95/* File is opened with O_NDELAY (only set for block devices) */
96#define FMODE_NDELAY ((__force fmode_t)0x40)
97/* File is opened with O_EXCL (only set for block devices) */
98#define FMODE_EXCL ((__force fmode_t)0x80)
99/* File is opened using open(.., 3, ..) and is writeable only for ioctls
100 (specialy hack for floppy.c) */
101#define FMODE_WRITE_IOCTL ((__force fmode_t)0x100)
102/* 32bit hashes as llseek() offset (for directories) */
103#define FMODE_32BITHASH ((__force fmode_t)0x200)
104/* 64bit hashes as llseek() offset (for directories) */
105#define FMODE_64BITHASH ((__force fmode_t)0x400)
106
107/*
108 * Don't update ctime and mtime.
109 *
110 * Currently a special hack for the XFS open_by_handle ioctl, but we'll
111 * hopefully graduate it to a proper O_CMTIME flag supported by open(2) soon.
112 */
113#define FMODE_NOCMTIME ((__force fmode_t)0x800)
114
115/* Expect random access pattern */
116#define FMODE_RANDOM ((__force fmode_t)0x1000)
117
118/* File is huge (eg. /dev/kmem): treat loff_t as unsigned */
119#define FMODE_UNSIGNED_OFFSET ((__force fmode_t)0x2000)
120
121/* File is opened with O_PATH; almost nothing can be done with it */
122#define FMODE_PATH ((__force fmode_t)0x4000)
123
124/* File was opened by fanotify and shouldn't generate fanotify events */
125#define FMODE_NONOTIFY ((__force fmode_t)0x1000000)
126
127/*
128 * Flag for rw_copy_check_uvector and compat_rw_copy_check_uvector
129 * that indicates that they should check the contents of the iovec are
130 * valid, but not check the memory that the iovec elements
131 * points too.
132 */
133#define CHECK_IOVEC_ONLY -1
134
Linus Torvalds1da177e2005-04-16 15:20:36 -0700135/*
David Howells08cce05c2012-10-13 10:45:06 +0100136 * The below are the various read and write types that we support. Some of
137 * them include behavioral modifiers that send information down to the
138 * block layer and IO scheduler. Terminology:
139 *
140 * The block layer uses device plugging to defer IO a little bit, in
141 * the hope that we will see more IO very shortly. This increases
142 * coalescing of adjacent IO and thus reduces the number of IOs we
143 * have to send to the device. It also allows for better queuing,
144 * if the IO isn't mergeable. If the caller is going to be waiting
145 * for the IO, then he must ensure that the device is unplugged so
146 * that the IO is dispatched to the driver.
147 *
148 * All IO is handled async in Linux. This is fine for background
149 * writes, but for reads or writes that someone waits for completion
150 * on, we want to notify the block layer and IO scheduler so that they
151 * know about it. That allows them to make better scheduling
152 * decisions. So when the below references 'sync' and 'async', it
153 * is referencing this priority hint.
154 *
155 * With that in mind, the available types are:
156 *
157 * READ A normal read operation. Device will be plugged.
158 * READ_SYNC A synchronous read. Device is not plugged, caller can
159 * immediately wait on this read without caring about
160 * unplugging.
161 * READA Used for read-ahead operations. Lower priority, and the
162 * block layer could (in theory) choose to ignore this
163 * request if it runs into resource problems.
164 * WRITE A normal async write. Device will be plugged.
165 * WRITE_SYNC Synchronous write. Identical to WRITE, but passes down
166 * the hint that someone will be waiting on this IO
167 * shortly. The write equivalent of READ_SYNC.
168 * WRITE_ODIRECT Special case write for O_DIRECT only.
169 * WRITE_FLUSH Like WRITE_SYNC but with preceding cache flush.
170 * WRITE_FUA Like WRITE_SYNC but data is guaranteed to be on
171 * non-volatile media on completion.
172 * WRITE_FLUSH_FUA Combination of WRITE_FLUSH and FUA. The IO is preceded
173 * by a cache flush and data is guaranteed to be on
174 * non-volatile media on completion.
175 *
176 */
177#define RW_MASK REQ_WRITE
178#define RWA_MASK REQ_RAHEAD
179
180#define READ 0
181#define WRITE RW_MASK
182#define READA RWA_MASK
183#define KERNEL_READ (READ|REQ_KERNEL)
184#define KERNEL_WRITE (WRITE|REQ_KERNEL)
185
186#define READ_SYNC (READ | REQ_SYNC)
187#define WRITE_SYNC (WRITE | REQ_SYNC | REQ_NOIDLE)
188#define WRITE_ODIRECT (WRITE | REQ_SYNC)
189#define WRITE_FLUSH (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FLUSH)
190#define WRITE_FUA (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FUA)
191#define WRITE_FLUSH_FUA (WRITE | REQ_SYNC | REQ_NOIDLE | REQ_FLUSH | REQ_FUA)
192
193/*
Linus Torvalds1da177e2005-04-16 15:20:36 -0700194 * Attribute flags. These should be or-ed together to figure out what
195 * has been changed!
196 */
Miklos Szeredi9767d742008-07-01 15:01:26 +0200197#define ATTR_MODE (1 << 0)
198#define ATTR_UID (1 << 1)
199#define ATTR_GID (1 << 2)
200#define ATTR_SIZE (1 << 3)
201#define ATTR_ATIME (1 << 4)
202#define ATTR_MTIME (1 << 5)
203#define ATTR_CTIME (1 << 6)
204#define ATTR_ATIME_SET (1 << 7)
205#define ATTR_MTIME_SET (1 << 8)
206#define ATTR_FORCE (1 << 9) /* Not a change, but a change it */
207#define ATTR_ATTR_FLAG (1 << 10)
208#define ATTR_KILL_SUID (1 << 11)
209#define ATTR_KILL_SGID (1 << 12)
210#define ATTR_FILE (1 << 13)
211#define ATTR_KILL_PRIV (1 << 14)
212#define ATTR_OPEN (1 << 15) /* Truncating from open(O_TRUNC) */
213#define ATTR_TIMES_SET (1 << 16)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700214
215/*
216 * This is the Inode Attributes structure, used for notify_change(). It
217 * uses the above definitions as flags, to know which values have changed.
218 * Also, in this manner, a Filesystem can look at only the values it cares
219 * about. Basically, these are the attributes that the VFS layer can
220 * request to change from the FS layer.
221 *
222 * Derek Atkins <warlord@MIT.EDU> 94-10-20
223 */
224struct iattr {
225 unsigned int ia_valid;
226 umode_t ia_mode;
Eric W. Biederman92361632012-02-08 07:07:50 -0800227 kuid_t ia_uid;
228 kgid_t ia_gid;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700229 loff_t ia_size;
230 struct timespec ia_atime;
231 struct timespec ia_mtime;
232 struct timespec ia_ctime;
Miklos Szeredicc4e69d2005-11-07 00:59:49 -0800233
234 /*
Lucas De Marchi25985ed2011-03-30 22:57:33 -0300235 * Not an attribute, but an auxiliary info for filesystems wanting to
Miklos Szeredicc4e69d2005-11-07 00:59:49 -0800236 * implement an ftruncate() like method. NOTE: filesystem should
237 * check for (ia_valid & ATTR_FILE), and not for (ia_file != NULL).
238 */
239 struct file *ia_file;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700240};
241
242/*
Linus Torvalds1da177e2005-04-16 15:20:36 -0700243 * Includes for diskquotas.
244 */
245#include <linux/quota.h>
246
Zach Brown994fc28c2005-12-15 14:28:17 -0800247/**
248 * enum positive_aop_returns - aop return codes with specific semantics
249 *
250 * @AOP_WRITEPAGE_ACTIVATE: Informs the caller that page writeback has
251 * completed, that the page is still locked, and
252 * should be considered active. The VM uses this hint
253 * to return the page to the active list -- it won't
254 * be a candidate for writeback again in the near
255 * future. Other callers must be careful to unlock
256 * the page if they get this return. Returned by
257 * writepage();
258 *
259 * @AOP_TRUNCATED_PAGE: The AOP method that was handed a locked page has
260 * unlocked it and the page might have been truncated.
261 * The caller should back up to acquiring a new page and
262 * trying again. The aop will be taking reasonable
263 * precautions not to livelock. If the caller held a page
264 * reference, it should drop it before retrying. Returned
Nick Piggin55144762007-10-16 01:25:26 -0700265 * by readpage().
Zach Brown994fc28c2005-12-15 14:28:17 -0800266 *
267 * address_space_operation functions return these large constants to indicate
268 * special semantics to the caller. These are much larger than the bytes in a
269 * page to allow for functions that return the number of bytes operated on in a
270 * given page.
271 */
272
273enum positive_aop_returns {
274 AOP_WRITEPAGE_ACTIVATE = 0x80000,
275 AOP_TRUNCATED_PAGE = 0x80001,
276};
277
Nick Pigginafddba42007-10-16 01:25:01 -0700278#define AOP_FLAG_UNINTERRUPTIBLE 0x0001 /* will not do a short write */
Nick Piggin89e10782007-10-16 01:25:07 -0700279#define AOP_FLAG_CONT_EXPAND 0x0002 /* called from cont_expand */
Nick Piggin54566b22009-01-04 12:00:53 -0800280#define AOP_FLAG_NOFS 0x0004 /* used by filesystem to direct
281 * helper code (eg buffer layer)
282 * to clear GFP_FS from alloc */
Nick Pigginafddba42007-10-16 01:25:01 -0700283
Linus Torvalds1da177e2005-04-16 15:20:36 -0700284/*
285 * oh the beauties of C type declarations.
286 */
287struct page;
288struct address_space;
289struct writeback_control;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700290
Nick Piggin2f718ff2007-10-16 01:24:59 -0700291struct iov_iter {
292 const struct iovec *iov;
293 unsigned long nr_segs;
294 size_t iov_offset;
295 size_t count;
296};
297
298size_t iov_iter_copy_from_user_atomic(struct page *page,
299 struct iov_iter *i, unsigned long offset, size_t bytes);
300size_t iov_iter_copy_from_user(struct page *page,
301 struct iov_iter *i, unsigned long offset, size_t bytes);
302void iov_iter_advance(struct iov_iter *i, size_t bytes);
Nick Pigginafddba42007-10-16 01:25:01 -0700303int iov_iter_fault_in_readable(struct iov_iter *i, size_t bytes);
Maxim Patlasovd28574e2012-10-26 19:50:04 +0400304size_t iov_iter_single_seg_count(const struct iov_iter *i);
Nick Piggin2f718ff2007-10-16 01:24:59 -0700305
306static inline void iov_iter_init(struct iov_iter *i,
307 const struct iovec *iov, unsigned long nr_segs,
308 size_t count, size_t written)
309{
310 i->iov = iov;
311 i->nr_segs = nr_segs;
312 i->iov_offset = 0;
313 i->count = count + written;
314
315 iov_iter_advance(i, written);
316}
317
318static inline size_t iov_iter_count(struct iov_iter *i)
319{
320 return i->count;
321}
322
Hisashi Hifumi8ab22b92008-07-28 15:46:36 -0700323/*
324 * "descriptor" for what we're up to with a read.
325 * This allows us to use the same read code yet
326 * have multiple different users of the data that
327 * we read from a file.
328 *
329 * The simplest case just copies the data to user
330 * mode.
331 */
332typedef struct {
333 size_t written;
334 size_t count;
335 union {
336 char __user *buf;
337 void *data;
338 } arg;
339 int error;
340} read_descriptor_t;
341
342typedef int (*read_actor_t)(read_descriptor_t *, struct page *,
343 unsigned long, unsigned long);
Nick Piggin2f718ff2007-10-16 01:24:59 -0700344
Linus Torvalds1da177e2005-04-16 15:20:36 -0700345struct address_space_operations {
346 int (*writepage)(struct page *page, struct writeback_control *wbc);
347 int (*readpage)(struct file *, struct page *);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700348
349 /* Write back some dirty pages from this mapping. */
350 int (*writepages)(struct address_space *, struct writeback_control *);
351
Andrew Morton4741c9f2006-03-24 03:18:11 -0800352 /* Set a page dirty. Return true if this dirtied it */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700353 int (*set_page_dirty)(struct page *page);
354
355 int (*readpages)(struct file *filp, struct address_space *mapping,
356 struct list_head *pages, unsigned nr_pages);
357
Nick Pigginafddba42007-10-16 01:25:01 -0700358 int (*write_begin)(struct file *, struct address_space *mapping,
359 loff_t pos, unsigned len, unsigned flags,
360 struct page **pagep, void **fsdata);
361 int (*write_end)(struct file *, struct address_space *mapping,
362 loff_t pos, unsigned len, unsigned copied,
363 struct page *page, void *fsdata);
364
Linus Torvalds1da177e2005-04-16 15:20:36 -0700365 /* Unfortunately this kludge is needed for FIBMAP. Don't use it */
366 sector_t (*bmap)(struct address_space *, sector_t);
NeilBrown2ff28e22006-03-26 01:37:18 -0800367 void (*invalidatepage) (struct page *, unsigned long);
Al Viro27496a82005-10-21 03:20:48 -0400368 int (*releasepage) (struct page *, gfp_t);
Linus Torvalds6072d132010-12-01 13:35:19 -0500369 void (*freepage)(struct page *);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700370 ssize_t (*direct_IO)(int, struct kiocb *, const struct iovec *iov,
371 loff_t offset, unsigned long nr_segs);
Nick Piggin70688e42008-04-28 02:13:02 -0700372 int (*get_xip_mem)(struct address_space *, pgoff_t, int,
373 void **, unsigned long *);
Mel Gormanb969c4a2012-01-12 17:19:34 -0800374 /*
375 * migrate the contents of a page to the specified target. If sync
376 * is false, it must not block.
377 */
Christoph Lameter2d1db3b2006-06-23 02:03:33 -0700378 int (*migratepage) (struct address_space *,
Mel Gormana6bc32b2012-01-12 17:19:43 -0800379 struct page *, struct page *, enum migrate_mode);
Trond Myklebuste3db7692007-01-10 23:15:39 -0800380 int (*launder_page) (struct page *);
Hisashi Hifumi8ab22b92008-07-28 15:46:36 -0700381 int (*is_partially_uptodate) (struct page *, read_descriptor_t *,
382 unsigned long);
Andi Kleen25718732009-09-16 11:50:13 +0200383 int (*error_remove_page)(struct address_space *, struct page *);
Mel Gorman62c230b2012-07-31 16:44:55 -0700384
385 /* swapfile support */
Mel Gormana509bc12012-07-31 16:44:57 -0700386 int (*swap_activate)(struct swap_info_struct *sis, struct file *file,
387 sector_t *span);
388 void (*swap_deactivate)(struct file *file);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700389};
390
Jens Axboe7dcda1c2011-04-05 23:51:48 +0200391extern const struct address_space_operations empty_aops;
392
Nick Pigginafddba42007-10-16 01:25:01 -0700393/*
394 * pagecache_write_begin/pagecache_write_end must be used by general code
395 * to write into the pagecache.
396 */
397int pagecache_write_begin(struct file *, struct address_space *mapping,
398 loff_t pos, unsigned len, unsigned flags,
399 struct page **pagep, void **fsdata);
400
401int pagecache_write_end(struct file *, struct address_space *mapping,
402 loff_t pos, unsigned len, unsigned copied,
403 struct page *page, void *fsdata);
404
Linus Torvalds1da177e2005-04-16 15:20:36 -0700405struct backing_dev_info;
406struct address_space {
407 struct inode *host; /* owner: inode, block_device */
408 struct radix_tree_root page_tree; /* radix tree of all pages */
Nick Piggin19fd6232008-07-25 19:45:32 -0700409 spinlock_t tree_lock; /* and lock protecting it */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700410 unsigned int i_mmap_writable;/* count VM_SHARED mappings */
Michel Lespinasse6b2dbba2012-10-08 16:31:25 -0700411 struct rb_root i_mmap; /* tree of private and shared mappings */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700412 struct list_head i_mmap_nonlinear;/*list VM_NONLINEAR mappings */
Peter Zijlstra3d48ae42011-05-24 17:12:06 -0700413 struct mutex i_mmap_mutex; /* protect tree, count, list */
Jan Kara08142572011-06-27 16:18:10 -0700414 /* Protected by tree_lock together with the radix tree */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700415 unsigned long nrpages; /* number of total pages */
416 pgoff_t writeback_index;/* writeback starts here */
Christoph Hellwigf5e54d62006-06-28 04:26:44 -0700417 const struct address_space_operations *a_ops; /* methods */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700418 unsigned long flags; /* error bits/gfp mask */
419 struct backing_dev_info *backing_dev_info; /* device readahead, etc */
420 spinlock_t private_lock; /* for use by the address_space */
421 struct list_head private_list; /* ditto */
Rafael Aquini252aa6f2012-12-11 16:02:35 -0800422 void *private_data; /* ditto */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700423} __attribute__((aligned(sizeof(long))));
424 /*
425 * On most architectures that alignment is already the case; but
Lucas De Marchi25985ed2011-03-30 22:57:33 -0300426 * must be enforced here for CRIS, to let the least significant bit
Linus Torvalds1da177e2005-04-16 15:20:36 -0700427 * of struct page's "mapping" pointer be used for PAGE_MAPPING_ANON.
428 */
Andi Kleen87192a22012-01-12 17:20:34 -0800429struct request_queue;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700430
431struct block_device {
432 dev_t bd_dev; /* not a kdev_t - it's a search key */
Richard Kennedyf3ccfcd2011-03-22 16:33:56 -0700433 int bd_openers;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700434 struct inode * bd_inode; /* will die */
Theodore Ts'o87d8fe12009-01-03 09:47:09 -0500435 struct super_block * bd_super;
Arjan van de Venc039e312006-03-23 03:00:28 -0800436 struct mutex bd_mutex; /* open/close mutex */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700437 struct list_head bd_inodes;
Tejun Heo6b4517a2010-04-07 18:53:59 +0900438 void * bd_claiming;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700439 void * bd_holder;
440 int bd_holders;
Tejun Heo77ea8872010-12-08 20:57:37 +0100441 bool bd_write_holder;
Jun'ichi Nomura641dc632006-03-27 01:17:57 -0800442#ifdef CONFIG_SYSFS
Tejun Heo49731ba2011-01-14 18:43:57 +0100443 struct list_head bd_holder_disks;
Jun'ichi Nomura641dc632006-03-27 01:17:57 -0800444#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -0700445 struct block_device * bd_contains;
446 unsigned bd_block_size;
447 struct hd_struct * bd_part;
448 /* number of times partitions within this device have been opened. */
449 unsigned bd_part_count;
450 int bd_invalidated;
451 struct gendisk * bd_disk;
Andi Kleen87192a22012-01-12 17:20:34 -0800452 struct request_queue * bd_queue;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700453 struct list_head bd_list;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700454 /*
455 * Private data. You must have bd_claim'ed the block_device
456 * to use this. NOTE: bd_claim allows an owner to claim
457 * the same device multiple times, the owner must take special
458 * care to not mess up bd_private for that case.
459 */
460 unsigned long bd_private;
Takashi Satofcccf502009-01-09 16:40:59 -0800461
462 /* The counter of freeze processes */
463 int bd_fsfreeze_count;
464 /* Mutex for freeze */
465 struct mutex bd_fsfreeze_mutex;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700466};
467
468/*
469 * Radix-tree tags, for tagging dirty and writeback pages within the pagecache
470 * radix trees
471 */
472#define PAGECACHE_TAG_DIRTY 0
473#define PAGECACHE_TAG_WRITEBACK 1
Jan Karaf446daae2010-08-09 17:19:12 -0700474#define PAGECACHE_TAG_TOWRITE 2
Linus Torvalds1da177e2005-04-16 15:20:36 -0700475
476int mapping_tagged(struct address_space *mapping, int tag);
477
478/*
479 * Might pages of this file be mapped into userspace?
480 */
481static inline int mapping_mapped(struct address_space *mapping)
482{
Michel Lespinasse6b2dbba2012-10-08 16:31:25 -0700483 return !RB_EMPTY_ROOT(&mapping->i_mmap) ||
Linus Torvalds1da177e2005-04-16 15:20:36 -0700484 !list_empty(&mapping->i_mmap_nonlinear);
485}
486
487/*
488 * Might pages of this file have been modified in userspace?
489 * Note that i_mmap_writable counts all VM_SHARED vmas: do_mmap_pgoff
490 * marks vma as VM_SHARED if it is shared, and the file was opened for
491 * writing i.e. vma may be mprotected writable even if now readonly.
492 */
493static inline int mapping_writably_mapped(struct address_space *mapping)
494{
495 return mapping->i_mmap_writable != 0;
496}
497
498/*
499 * Use sequence counter to get consistent i_size on 32-bit processors.
500 */
501#if BITS_PER_LONG==32 && defined(CONFIG_SMP)
502#include <linux/seqlock.h>
503#define __NEED_I_SIZE_ORDERED
504#define i_size_ordered_init(inode) seqcount_init(&inode->i_size_seqcount)
505#else
506#define i_size_ordered_init(inode) do { } while (0)
507#endif
508
Al Virof19d4a82009-06-08 19:50:45 -0400509struct posix_acl;
510#define ACL_NOT_CACHED ((void *)(-1))
511
Linus Torvalds3ddcd052011-08-06 22:45:50 -0700512#define IOP_FASTPERM 0x0001
513#define IOP_LOOKUP 0x0002
514#define IOP_NOFOLLOW 0x0004
515
516/*
517 * Keep mostly read-only and often accessed (especially for
518 * the RCU path lookup and 'stat' data) fields at the beginning
519 * of the 'struct inode'
520 */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700521struct inode {
Nick Piggin44a7d7a2011-01-07 17:49:56 +1100522 umode_t i_mode;
Linus Torvalds3ddcd052011-08-06 22:45:50 -0700523 unsigned short i_opflags;
Eric W. Biederman92361632012-02-08 07:07:50 -0800524 kuid_t i_uid;
525 kgid_t i_gid;
Linus Torvalds3ddcd052011-08-06 22:45:50 -0700526 unsigned int i_flags;
527
528#ifdef CONFIG_FS_POSIX_ACL
529 struct posix_acl *i_acl;
530 struct posix_acl *i_default_acl;
531#endif
532
Nick Piggin44a7d7a2011-01-07 17:49:56 +1100533 const struct inode_operations *i_op;
534 struct super_block *i_sb;
Linus Torvalds3ddcd052011-08-06 22:45:50 -0700535 struct address_space *i_mapping;
Nick Piggin44a7d7a2011-01-07 17:49:56 +1100536
Linus Torvalds13e12d12011-06-08 15:18:19 -0700537#ifdef CONFIG_SECURITY
538 void *i_security;
539#endif
Nick Piggin44a7d7a2011-01-07 17:49:56 +1100540
Linus Torvalds3ddcd052011-08-06 22:45:50 -0700541 /* Stat data, not accessed from path walking */
542 unsigned long i_ino;
Miklos Szeredia78ef702011-10-28 14:13:30 +0200543 /*
544 * Filesystems may only read i_nlink directly. They shall use the
545 * following functions for modification:
546 *
547 * (set|clear|inc|drop)_nlink
548 * inode_(inc|dec)_link_count
549 */
550 union {
551 const unsigned int i_nlink;
552 unsigned int __i_nlink;
553 };
Linus Torvalds3ddcd052011-08-06 22:45:50 -0700554 dev_t i_rdev;
Linus Torvalds2f9d3df2012-06-03 14:50:19 -0700555 loff_t i_size;
Linus Torvalds3ddcd052011-08-06 22:45:50 -0700556 struct timespec i_atime;
557 struct timespec i_mtime;
558 struct timespec i_ctime;
Theodore Ts'o6cdbb0e2011-10-29 08:24:18 -0400559 spinlock_t i_lock; /* i_blocks, i_bytes, maybe i_size */
560 unsigned short i_bytes;
Linus Torvalds2f9d3df2012-06-03 14:50:19 -0700561 unsigned int i_blkbits;
Linus Torvalds3ddcd052011-08-06 22:45:50 -0700562 blkcnt_t i_blocks;
563
564#ifdef __NEED_I_SIZE_ORDERED
565 seqcount_t i_size_seqcount;
566#endif
567
568 /* Misc */
569 unsigned long i_state;
Linus Torvalds3ddcd052011-08-06 22:45:50 -0700570 struct mutex i_mutex;
Linus Torvalds13e12d12011-06-08 15:18:19 -0700571
Nick Piggin44a7d7a2011-01-07 17:49:56 +1100572 unsigned long dirtied_when; /* jiffies of first dirtying */
573
Linus Torvalds1da177e2005-04-16 15:20:36 -0700574 struct hlist_node i_hash;
Nick Piggin7ccf19a2010-10-21 11:49:30 +1100575 struct list_head i_wb_list; /* backing dev IO list */
576 struct list_head i_lru; /* inode LRU list */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700577 struct list_head i_sb_list;
Nick Pigginfa0d7e3d2011-01-07 17:49:49 +1100578 union {
Al Virob3d9b7a2012-06-09 13:51:19 -0400579 struct hlist_head i_dentry;
Nick Pigginfa0d7e3d2011-01-07 17:49:49 +1100580 struct rcu_head i_rcu;
581 };
Jean Noel Cordenner7a224222008-01-28 23:58:27 -0500582 u64 i_version;
Linus Torvalds2f9d3df2012-06-03 14:50:19 -0700583 atomic_t i_count;
Christoph Hellwigbd5fe6c2011-06-24 14:29:43 -0400584 atomic_t i_dio_count;
Theodore Ts'o6cdbb0e2011-10-29 08:24:18 -0400585 atomic_t i_writecount;
Arjan van de Ven99ac48f2006-03-28 01:56:41 -0800586 const struct file_operations *i_fop; /* former ->i_op->default_file_ops */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700587 struct file_lock *i_flock;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700588 struct address_space i_data;
589#ifdef CONFIG_QUOTA
590 struct dquot *i_dquot[MAXQUOTAS];
591#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -0700592 struct list_head i_devices;
Theodore Ts'o4c154162006-09-27 01:50:47 -0700593 union {
594 struct pipe_inode_info *i_pipe;
Theodore Ts'oeaf796e2006-09-27 01:50:48 -0700595 struct block_device *i_bdev;
Theodore Ts'o577c4eb2006-09-27 01:50:49 -0700596 struct cdev *i_cdev;
Theodore Ts'o4c154162006-09-27 01:50:47 -0700597 };
Linus Torvalds1da177e2005-04-16 15:20:36 -0700598
599 __u32 i_generation;
600
Eric Paris3be25f42009-05-21 17:01:26 -0400601#ifdef CONFIG_FSNOTIFY
602 __u32 i_fsnotify_mask; /* all events this inode cares about */
Eric Parise61ce862009-12-17 21:24:24 -0500603 struct hlist_head i_fsnotify_marks;
Robert Love0eeca282005-07-12 17:06:03 -0400604#endif
605
Eric Parisa178d202010-10-25 14:41:59 -0400606#ifdef CONFIG_IMA
Mimi Zohara68a27b2010-11-02 10:10:56 -0400607 atomic_t i_readcount; /* struct files open RO */
Eric Parisa178d202010-10-25 14:41:59 -0400608#endif
Theodore Ts'o8e18e292006-09-27 01:50:46 -0700609 void *i_private; /* fs or device private pointer */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700610};
611
Al Viro1d3382cb2010-10-23 15:19:20 -0400612static inline int inode_unhashed(struct inode *inode)
613{
614 return hlist_unhashed(&inode->i_hash);
615}
616
Linus Torvalds1da177e2005-04-16 15:20:36 -0700617/*
Ingo Molnarf2eace22006-07-03 00:25:05 -0700618 * inode->i_mutex nesting subclasses for the lock validator:
619 *
620 * 0: the object of the current VFS operation
621 * 1: parent
622 * 2: child/target
623 * 3: quota file
624 *
625 * The locking order between these classes is
Ingo Molnar4df46242006-08-27 01:23:56 -0700626 * parent -> child -> normal -> xattr -> quota
Ingo Molnarf2eace22006-07-03 00:25:05 -0700627 */
628enum inode_i_mutex_lock_class
629{
630 I_MUTEX_NORMAL,
631 I_MUTEX_PARENT,
632 I_MUTEX_CHILD,
Ingo Molnar4df46242006-08-27 01:23:56 -0700633 I_MUTEX_XATTR,
Ingo Molnarf2eace22006-07-03 00:25:05 -0700634 I_MUTEX_QUOTA
635};
636
637/*
Linus Torvalds1da177e2005-04-16 15:20:36 -0700638 * NOTE: in a 32bit arch with a preemptable kernel and
639 * an UP compile the i_size_read/write must be atomic
640 * with respect to the local cpu (unlike with preempt disabled),
641 * but they don't need to be atomic with respect to other cpus like in
642 * true SMP (so they need either to either locally disable irq around
643 * the read or for example on x86 they can be still implemented as a
644 * cmpxchg8b without the need of the lock prefix). For SMP compiles
645 * and 64bit archs it makes no difference if preempt is enabled or not.
646 */
Jan Engelhardt48ed2142006-12-06 20:35:37 -0800647static inline loff_t i_size_read(const struct inode *inode)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700648{
649#if BITS_PER_LONG==32 && defined(CONFIG_SMP)
650 loff_t i_size;
651 unsigned int seq;
652
653 do {
654 seq = read_seqcount_begin(&inode->i_size_seqcount);
655 i_size = inode->i_size;
656 } while (read_seqcount_retry(&inode->i_size_seqcount, seq));
657 return i_size;
658#elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
659 loff_t i_size;
660
661 preempt_disable();
662 i_size = inode->i_size;
663 preempt_enable();
664 return i_size;
665#else
666 return inode->i_size;
667#endif
668}
669
Miklos Szeredi7762f5a2006-10-17 00:10:07 -0700670/*
671 * NOTE: unlike i_size_read(), i_size_write() does need locking around it
672 * (normally i_mutex), otherwise on 32bit/SMP an update of i_size_seqcount
673 * can be lost, resulting in subsequent i_size_read() calls spinning forever.
674 */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700675static inline void i_size_write(struct inode *inode, loff_t i_size)
676{
677#if BITS_PER_LONG==32 && defined(CONFIG_SMP)
Fan Du74e3d1e2013-04-30 15:27:27 -0700678 preempt_disable();
Linus Torvalds1da177e2005-04-16 15:20:36 -0700679 write_seqcount_begin(&inode->i_size_seqcount);
680 inode->i_size = i_size;
681 write_seqcount_end(&inode->i_size_seqcount);
Fan Du74e3d1e2013-04-30 15:27:27 -0700682 preempt_enable();
Linus Torvalds1da177e2005-04-16 15:20:36 -0700683#elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
684 preempt_disable();
685 inode->i_size = i_size;
686 preempt_enable();
687#else
688 inode->i_size = i_size;
689#endif
690}
691
Eric W. Biederman92361632012-02-08 07:07:50 -0800692/* Helper functions so that in most cases filesystems will
693 * not need to deal directly with kuid_t and kgid_t and can
694 * instead deal with the raw numeric values that are stored
695 * in the filesystem.
696 */
697static inline uid_t i_uid_read(const struct inode *inode)
698{
699 return from_kuid(&init_user_ns, inode->i_uid);
700}
701
702static inline gid_t i_gid_read(const struct inode *inode)
703{
704 return from_kgid(&init_user_ns, inode->i_gid);
705}
706
707static inline void i_uid_write(struct inode *inode, uid_t uid)
708{
709 inode->i_uid = make_kuid(&init_user_ns, uid);
710}
711
712static inline void i_gid_write(struct inode *inode, gid_t gid)
713{
714 inode->i_gid = make_kgid(&init_user_ns, gid);
715}
716
Jan Engelhardt48ed2142006-12-06 20:35:37 -0800717static inline unsigned iminor(const struct inode *inode)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700718{
719 return MINOR(inode->i_rdev);
720}
721
Jan Engelhardt48ed2142006-12-06 20:35:37 -0800722static inline unsigned imajor(const struct inode *inode)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700723{
724 return MAJOR(inode->i_rdev);
725}
726
727extern struct block_device *I_BDEV(struct inode *inode);
728
729struct fown_struct {
730 rwlock_t lock; /* protects pid, uid, euid fields */
Eric W. Biederman609d7fa2006-10-02 02:17:15 -0700731 struct pid *pid; /* pid or -pgrp where SIGIO should be sent */
732 enum pid_type pid_type; /* Kind of process group SIGIO should be sent to */
Eric W. Biederman92361632012-02-08 07:07:50 -0800733 kuid_t uid, euid; /* uid/euid of process setting the owner */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700734 int signum; /* posix.1b rt signal to be delivered on IO */
735};
736
737/*
738 * Track a single file's readahead state
739 */
740struct file_ra_state {
Fengguang Wu937085a2007-10-16 01:24:31 -0700741 pgoff_t start; /* where readahead started */
742 unsigned int size; /* # of readahead pages */
743 unsigned int async_size; /* do asynchronous readahead when
Fengguang Wuf9acc8c2007-07-19 01:48:08 -0700744 there are only # of pages ahead */
Fengguang Wu5ce11102007-07-19 01:47:59 -0700745
Fengguang Wu937085a2007-10-16 01:24:31 -0700746 unsigned int ra_pages; /* Maximum readahead window */
Wu Fengguang1ebf26a2009-06-16 15:31:19 -0700747 unsigned int mmap_miss; /* Cache miss stat for mmap accesses */
Fengguang Wuf4e6b492007-10-16 01:24:33 -0700748 loff_t prev_pos; /* Cache last read() position */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700749};
Linus Torvalds1da177e2005-04-16 15:20:36 -0700750
Fengguang Wu5ce11102007-07-19 01:47:59 -0700751/*
Fengguang Wu5ce11102007-07-19 01:47:59 -0700752 * Check if @index falls in the readahead windows.
753 */
754static inline int ra_has_index(struct file_ra_state *ra, pgoff_t index)
755{
Fengguang Wuf9acc8c2007-07-19 01:48:08 -0700756 return (index >= ra->start &&
757 index < ra->start + ra->size);
Fengguang Wu5ce11102007-07-19 01:47:59 -0700758}
759
Dave Hansenad775f52008-02-15 14:38:01 -0800760#define FILE_MNT_WRITE_TAKEN 1
761#define FILE_MNT_WRITE_RELEASED 2
762
Linus Torvalds1da177e2005-04-16 15:20:36 -0700763struct file {
Eric Dumazet2f512012005-10-30 15:02:16 -0800764 /*
765 * fu_list becomes invalid after file_free is called and queued via
766 * fu_rcuhead for RCU freeing
767 */
768 union {
769 struct list_head fu_list;
770 struct rcu_head fu_rcuhead;
771 } f_u;
Josef "Jeff" Sipek0f7fc9e2006-12-08 02:36:35 -0800772 struct path f_path;
773#define f_dentry f_path.dentry
Al Virodd379782013-03-01 19:48:30 -0500774 struct inode *f_inode; /* cached value */
Arjan van de Ven99ac48f2006-03-28 01:56:41 -0800775 const struct file_operations *f_op;
Andi Kleenef3d0fd2011-09-15 16:06:48 -0700776
777 /*
778 * Protects f_ep_links, f_flags, f_pos vs i_size in lseek SEEK_CUR.
779 * Must not be taken from IRQ context.
780 */
781 spinlock_t f_lock;
Nick Piggin6416ccb2010-08-18 04:37:38 +1000782#ifdef CONFIG_SMP
783 int f_sb_list_cpu;
784#endif
Al Viro516e0cc2008-07-26 00:39:17 -0400785 atomic_long_t f_count;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700786 unsigned int f_flags;
Al Viroaeb5d722008-09-02 15:28:45 -0400787 fmode_t f_mode;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700788 loff_t f_pos;
789 struct fown_struct f_owner;
David Howellsd76b0d92008-11-14 10:39:25 +1100790 const struct cred *f_cred;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700791 struct file_ra_state f_ra;
792
Mathieu Desnoyers2b47c362007-10-16 23:27:21 -0700793 u64 f_version;
Alexey Dobriyan50462062006-09-29 02:00:01 -0700794#ifdef CONFIG_SECURITY
Linus Torvalds1da177e2005-04-16 15:20:36 -0700795 void *f_security;
Alexey Dobriyan50462062006-09-29 02:00:01 -0700796#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -0700797 /* needed for tty driver, and maybe others */
798 void *private_data;
799
800#ifdef CONFIG_EPOLL
801 /* Used by fs/eventpoll.c to link all the hooks to this file */
802 struct list_head f_ep_links;
Jason Baron28d82dc2012-01-12 17:17:43 -0800803 struct list_head f_tfile_llink;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700804#endif /* #ifdef CONFIG_EPOLL */
805 struct address_space *f_mapping;
Dave Hansenad775f52008-02-15 14:38:01 -0800806#ifdef CONFIG_DEBUG_WRITECOUNT
807 unsigned long f_mnt_write_state;
808#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -0700809};
Linus Torvalds1da177e2005-04-16 15:20:36 -0700810
Aneesh Kumar K.V990d6c22011-01-29 18:43:26 +0530811struct file_handle {
812 __u32 handle_bytes;
813 int handle_type;
814 /* file identifier */
815 unsigned char f_handle[0];
816};
817
Al Virocb0942b2012-08-27 14:48:26 -0400818static inline struct file *get_file(struct file *f)
819{
820 atomic_long_inc(&f->f_count);
821 return f;
822}
Al Virod7065da2010-05-26 15:13:55 -0400823#define fput_atomic(x) atomic_long_add_unless(&(x)->f_count, -1, 1)
Al Viro516e0cc2008-07-26 00:39:17 -0400824#define file_count(x) atomic_long_read(&(x)->f_count)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700825
Dave Hansenad775f52008-02-15 14:38:01 -0800826#ifdef CONFIG_DEBUG_WRITECOUNT
827static inline void file_take_write(struct file *f)
828{
829 WARN_ON(f->f_mnt_write_state != 0);
830 f->f_mnt_write_state = FILE_MNT_WRITE_TAKEN;
831}
832static inline void file_release_write(struct file *f)
833{
834 f->f_mnt_write_state |= FILE_MNT_WRITE_RELEASED;
835}
836static inline void file_reset_write(struct file *f)
837{
838 f->f_mnt_write_state = 0;
839}
840static inline void file_check_state(struct file *f)
841{
842 /*
843 * At this point, either both or neither of these bits
844 * should be set.
845 */
846 WARN_ON(f->f_mnt_write_state == FILE_MNT_WRITE_TAKEN);
847 WARN_ON(f->f_mnt_write_state == FILE_MNT_WRITE_RELEASED);
848}
849static inline int file_check_writeable(struct file *f)
850{
851 if (f->f_mnt_write_state == FILE_MNT_WRITE_TAKEN)
852 return 0;
853 printk(KERN_WARNING "writeable file with no "
854 "mnt_want_write()\n");
855 WARN_ON(1);
856 return -EINVAL;
857}
858#else /* !CONFIG_DEBUG_WRITECOUNT */
859static inline void file_take_write(struct file *filp) {}
860static inline void file_release_write(struct file *filp) {}
861static inline void file_reset_write(struct file *filp) {}
862static inline void file_check_state(struct file *filp) {}
863static inline int file_check_writeable(struct file *filp)
864{
865 return 0;
866}
867#endif /* CONFIG_DEBUG_WRITECOUNT */
868
Linus Torvalds1da177e2005-04-16 15:20:36 -0700869#define MAX_NON_LFS ((1UL<<31) - 1)
870
871/* Page cache limit. The filesystems should put that into their s_maxbytes
872 limits, otherwise bad things can happen in VM. */
873#if BITS_PER_LONG==32
Chuck Lever2bd2c192012-09-24 14:51:48 -0400874#define MAX_LFS_FILESIZE (((loff_t)PAGE_CACHE_SIZE << (BITS_PER_LONG-1))-1)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700875#elif BITS_PER_LONG==64
Al Viro614c3212012-10-10 01:06:20 -0400876#define MAX_LFS_FILESIZE ((loff_t)0x7fffffffffffffffLL)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700877#endif
878
879#define FL_POSIX 1
880#define FL_FLOCK 2
881#define FL_ACCESS 8 /* not trying to lock, just looking */
Trond Myklebustf475ae92006-06-29 16:38:32 -0400882#define FL_EXISTS 16 /* when unlocking, test for existence */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700883#define FL_LEASE 32 /* lease held on this file */
Miklos Szeredi75e1fcc2006-06-23 02:05:12 -0700884#define FL_CLOSE 64 /* unlock on close */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700885#define FL_SLEEP 128 /* A blocking lock */
J. Bruce Fields778fc542011-07-26 18:25:49 -0400886#define FL_DOWNGRADE_PENDING 256 /* Lease is being downgraded */
887#define FL_UNLOCK_PENDING 512 /* Lease is being broken */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700888
889/*
Miklos Szeredibde74e42008-07-25 01:48:57 -0700890 * Special return value from posix_lock_file() and vfs_lock_file() for
891 * asynchronous locking.
892 */
893#define FILE_LOCK_DEFERRED 1
894
895/*
Linus Torvalds1da177e2005-04-16 15:20:36 -0700896 * The POSIX file lock owner is determined by
897 * the "struct files_struct" in the thread group
898 * (or NULL for no owner - BSD locks).
899 *
900 * Lockd stuffs a "host" pointer into this.
901 */
902typedef struct files_struct *fl_owner_t;
903
904struct file_lock_operations {
Linus Torvalds1da177e2005-04-16 15:20:36 -0700905 void (*fl_copy_lock)(struct file_lock *, struct file_lock *);
906 void (*fl_release_private)(struct file_lock *);
907};
908
909struct lock_manager_operations {
J. Bruce Fields8fb47a42011-07-20 20:21:59 -0400910 int (*lm_compare_owner)(struct file_lock *, struct file_lock *);
911 void (*lm_notify)(struct file_lock *); /* unblock callback */
912 int (*lm_grant)(struct file_lock *, struct file_lock *, int);
J. Bruce Fields8fb47a42011-07-20 20:21:59 -0400913 void (*lm_break)(struct file_lock *);
914 int (*lm_change)(struct file_lock **, int);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700915};
916
J. Bruce Fieldsaf558e32007-09-06 12:34:25 -0400917struct lock_manager {
918 struct list_head list;
919};
920
Stanislav Kinsbursky5ccb0062012-07-25 16:57:22 +0400921struct net;
922void locks_start_grace(struct net *, struct lock_manager *);
J. Bruce Fieldsaf558e32007-09-06 12:34:25 -0400923void locks_end_grace(struct lock_manager *);
Stanislav Kinsbursky5ccb0062012-07-25 16:57:22 +0400924int locks_in_grace(struct net *);
J. Bruce Fieldsaf558e32007-09-06 12:34:25 -0400925
Linus Torvalds1da177e2005-04-16 15:20:36 -0700926/* that will die - we need it for nfs_lock_info */
927#include <linux/nfs_fs_i.h>
928
929struct file_lock {
930 struct file_lock *fl_next; /* singly linked list for this inode */
931 struct list_head fl_link; /* doubly linked list of all locks */
932 struct list_head fl_block; /* circular list of blocked processes */
933 fl_owner_t fl_owner;
J. Bruce Fields710b7212011-07-26 16:28:29 -0400934 unsigned int fl_flags;
Richard Kennedyafc12462008-07-11 17:20:49 -0700935 unsigned char fl_type;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700936 unsigned int fl_pid;
Vitaliy Gusevab1f1612008-01-17 00:07:08 +0000937 struct pid *fl_nspid;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700938 wait_queue_head_t fl_wait;
939 struct file *fl_file;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700940 loff_t fl_start;
941 loff_t fl_end;
942
943 struct fasync_struct * fl_fasync; /* for lease break notifications */
J. Bruce Fields778fc542011-07-26 18:25:49 -0400944 /* for lease breaks: */
945 unsigned long fl_break_time;
946 unsigned long fl_downgrade_time;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700947
Alexey Dobriyan6aed6282009-09-21 17:01:11 -0700948 const struct file_lock_operations *fl_ops; /* Callbacks for filesystems */
Alexey Dobriyan7b021962009-09-21 17:01:12 -0700949 const struct lock_manager_operations *fl_lmops; /* Callbacks for lockmanagers */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700950 union {
951 struct nfs_lock_info nfs_fl;
Trond Myklebust8d0a8a92005-06-22 17:16:32 +0000952 struct nfs4_lock_info nfs4_fl;
David Howellse8d6c552007-07-15 23:40:12 -0700953 struct {
954 struct list_head link; /* link in AFS vnode's pending_locks list */
955 int state; /* state of grant or error if -ve */
956 } afs;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700957 } fl_u;
958};
959
960/* The following constant reflects the upper bound of the file/locking space */
961#ifndef OFFSET_MAX
962#define INT_LIMIT(x) (~((x)1 << (sizeof(x)*8 - 1)))
963#define OFFSET_MAX INT_LIMIT(loff_t)
964#define OFFT_OFFSET_MAX INT_LIMIT(off_t)
965#endif
966
Linus Torvalds1da177e2005-04-16 15:20:36 -0700967#include <linux/fcntl.h>
968
Thomas Petazzonibfcd17a2008-08-06 15:12:22 +0200969extern void send_sigio(struct fown_struct *fown, int fd, int band);
970
Thomas Petazzonibfcd17a2008-08-06 15:12:22 +0200971#ifdef CONFIG_FILE_LOCKING
Linus Torvalds1da177e2005-04-16 15:20:36 -0700972extern int fcntl_getlk(struct file *, struct flock __user *);
Peter Staubachc2936212005-07-27 11:45:09 -0700973extern int fcntl_setlk(unsigned int, struct file *, unsigned int,
974 struct flock __user *);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700975
976#if BITS_PER_LONG == 32
977extern int fcntl_getlk64(struct file *, struct flock64 __user *);
Peter Staubachc2936212005-07-27 11:45:09 -0700978extern int fcntl_setlk64(unsigned int, struct file *, unsigned int,
979 struct flock64 __user *);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700980#endif
981
Linus Torvalds1da177e2005-04-16 15:20:36 -0700982extern int fcntl_setlease(unsigned int fd, struct file *filp, long arg);
983extern int fcntl_getlease(struct file *filp);
984
985/* fs/locks.c */
J. Bruce Fields05fa3132010-10-30 17:31:15 -0400986void locks_free_lock(struct file_lock *fl);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700987extern void locks_init_lock(struct file_lock *);
Arnd Bergmannc5b1f0d2010-10-27 15:46:08 +0200988extern struct file_lock * locks_alloc_lock(void);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700989extern void locks_copy_lock(struct file_lock *, struct file_lock *);
J. Bruce Fields1a747ee2008-04-24 10:08:22 -0400990extern void __locks_copy_lock(struct file_lock *, const struct file_lock *);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700991extern void locks_remove_posix(struct file *, fl_owner_t);
992extern void locks_remove_flock(struct file *);
Felix Blyakhera9e61e22009-03-31 15:12:56 -0500993extern void locks_release_private(struct file_lock *);
J. Bruce Fields6d34ac12007-05-11 16:09:32 -0400994extern void posix_test_lock(struct file *, struct file_lock *);
Marc Eshel150b3932007-01-18 16:15:35 -0500995extern int posix_lock_file(struct file *, struct file_lock *, struct file_lock *);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700996extern int posix_lock_file_wait(struct file *, struct file_lock *);
J. Bruce Fields64a318e2006-01-03 09:55:46 +0100997extern int posix_unblock_lock(struct file *, struct file_lock *);
J. Bruce Fields3ee17ab2007-02-21 00:58:50 -0500998extern int vfs_test_lock(struct file *, struct file_lock *);
Marc Eshel150b3932007-01-18 16:15:35 -0500999extern int vfs_lock_file(struct file *, unsigned int, struct file_lock *, struct file_lock *);
Marc Eshel9b9d2ab2007-01-18 17:52:58 -05001000extern int vfs_cancel_lock(struct file *filp, struct file_lock *fl);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001001extern int flock_lock_file_wait(struct file *filp, struct file_lock *fl);
1002extern int __break_lease(struct inode *inode, unsigned int flags);
1003extern void lease_get_mtime(struct inode *, struct timespec *time);
Christoph Hellwig0af1a452007-07-31 00:39:22 -07001004extern int generic_setlease(struct file *, long, struct file_lock **);
J. Bruce Fieldsa9933ce2007-06-07 17:09:49 -04001005extern int vfs_setlease(struct file *, long, struct file_lock **);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001006extern int lease_modify(struct file_lock **, int);
1007extern int lock_may_read(struct inode *, loff_t start, unsigned long count);
1008extern int lock_may_write(struct inode *, loff_t start, unsigned long count);
Pavel Shilovsky66189be2012-03-28 21:56:19 +04001009extern void locks_delete_block(struct file_lock *waiter);
Arnd Bergmannb89f4322010-09-18 15:09:31 +02001010extern void lock_flocks(void);
1011extern void unlock_flocks(void);
Thomas Petazzonibfcd17a2008-08-06 15:12:22 +02001012#else /* !CONFIG_FILE_LOCKING */
Steven Whitehousec2aca5e2009-01-20 10:29:45 +00001013static inline int fcntl_getlk(struct file *file, struct flock __user *user)
1014{
1015 return -EINVAL;
1016}
1017
1018static inline int fcntl_setlk(unsigned int fd, struct file *file,
1019 unsigned int cmd, struct flock __user *user)
1020{
1021 return -EACCES;
1022}
1023
Thomas Petazzonibfcd17a2008-08-06 15:12:22 +02001024#if BITS_PER_LONG == 32
Steven Whitehousec2aca5e2009-01-20 10:29:45 +00001025static inline int fcntl_getlk64(struct file *file, struct flock64 __user *user)
1026{
1027 return -EINVAL;
1028}
1029
1030static inline int fcntl_setlk64(unsigned int fd, struct file *file,
1031 unsigned int cmd, struct flock64 __user *user)
1032{
1033 return -EACCES;
1034}
Thomas Petazzonibfcd17a2008-08-06 15:12:22 +02001035#endif
Steven Whitehousec2aca5e2009-01-20 10:29:45 +00001036static inline int fcntl_setlease(unsigned int fd, struct file *filp, long arg)
1037{
1038 return 0;
1039}
1040
1041static inline int fcntl_getlease(struct file *filp)
1042{
1043 return 0;
1044}
1045
1046static inline void locks_init_lock(struct file_lock *fl)
1047{
1048 return;
1049}
1050
1051static inline void __locks_copy_lock(struct file_lock *new, struct file_lock *fl)
1052{
1053 return;
1054}
1055
1056static inline void locks_copy_lock(struct file_lock *new, struct file_lock *fl)
1057{
1058 return;
1059}
1060
1061static inline void locks_remove_posix(struct file *filp, fl_owner_t owner)
1062{
1063 return;
1064}
1065
1066static inline void locks_remove_flock(struct file *filp)
1067{
1068 return;
1069}
1070
1071static inline void posix_test_lock(struct file *filp, struct file_lock *fl)
1072{
1073 return;
1074}
1075
1076static inline int posix_lock_file(struct file *filp, struct file_lock *fl,
1077 struct file_lock *conflock)
1078{
1079 return -ENOLCK;
1080}
1081
1082static inline int posix_lock_file_wait(struct file *filp, struct file_lock *fl)
1083{
1084 return -ENOLCK;
1085}
1086
1087static inline int posix_unblock_lock(struct file *filp,
1088 struct file_lock *waiter)
1089{
1090 return -ENOENT;
1091}
1092
1093static inline int vfs_test_lock(struct file *filp, struct file_lock *fl)
1094{
1095 return 0;
1096}
1097
1098static inline int vfs_lock_file(struct file *filp, unsigned int cmd,
1099 struct file_lock *fl, struct file_lock *conf)
1100{
1101 return -ENOLCK;
1102}
1103
1104static inline int vfs_cancel_lock(struct file *filp, struct file_lock *fl)
1105{
1106 return 0;
1107}
1108
1109static inline int flock_lock_file_wait(struct file *filp,
1110 struct file_lock *request)
1111{
1112 return -ENOLCK;
1113}
1114
1115static inline int __break_lease(struct inode *inode, unsigned int mode)
1116{
1117 return 0;
1118}
1119
1120static inline void lease_get_mtime(struct inode *inode, struct timespec *time)
1121{
1122 return;
1123}
1124
1125static inline int generic_setlease(struct file *filp, long arg,
1126 struct file_lock **flp)
1127{
1128 return -EINVAL;
1129}
1130
1131static inline int vfs_setlease(struct file *filp, long arg,
1132 struct file_lock **lease)
1133{
1134 return -EINVAL;
1135}
1136
1137static inline int lease_modify(struct file_lock **before, int arg)
1138{
1139 return -EINVAL;
1140}
1141
1142static inline int lock_may_read(struct inode *inode, loff_t start,
1143 unsigned long len)
1144{
1145 return 1;
1146}
1147
1148static inline int lock_may_write(struct inode *inode, loff_t start,
1149 unsigned long len)
1150{
1151 return 1;
1152}
1153
Pavel Shilovsky66189be2012-03-28 21:56:19 +04001154static inline void locks_delete_block(struct file_lock *waiter)
1155{
1156}
1157
Arnd Bergmannb89f4322010-09-18 15:09:31 +02001158static inline void lock_flocks(void)
1159{
1160}
1161
1162static inline void unlock_flocks(void)
1163{
1164}
1165
Thomas Petazzonibfcd17a2008-08-06 15:12:22 +02001166#endif /* !CONFIG_FILE_LOCKING */
1167
Linus Torvalds1da177e2005-04-16 15:20:36 -07001168
1169struct fasync_struct {
Eric Dumazet989a2972010-04-14 09:55:35 +00001170 spinlock_t fa_lock;
1171 int magic;
1172 int fa_fd;
1173 struct fasync_struct *fa_next; /* singly linked list */
1174 struct file *fa_file;
1175 struct rcu_head fa_rcu;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001176};
1177
1178#define FASYNC_MAGIC 0x4601
1179
1180/* SMP safe fasync helpers: */
1181extern int fasync_helper(int, struct file *, int, struct fasync_struct **);
Linus Torvaldsf7347ce2010-10-27 12:38:12 -04001182extern struct fasync_struct *fasync_insert_entry(int, struct file *, struct fasync_struct **, struct fasync_struct *);
1183extern int fasync_remove_entry(struct file *, struct fasync_struct **);
1184extern struct fasync_struct *fasync_alloc(void);
1185extern void fasync_free(struct fasync_struct *);
1186
Linus Torvalds1da177e2005-04-16 15:20:36 -07001187/* can be called from interrupts */
1188extern void kill_fasync(struct fasync_struct **, int, int);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001189
Eric W. Biederman609d7fa2006-10-02 02:17:15 -07001190extern int __f_setown(struct file *filp, struct pid *, enum pid_type, int force);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001191extern int f_setown(struct file *filp, unsigned long arg, int force);
1192extern void f_delown(struct file *filp);
Eric W. Biederman609d7fa2006-10-02 02:17:15 -07001193extern pid_t f_getown(struct file *filp);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001194extern int send_sigurg(struct fown_struct *fown);
1195
Jan Kara5accdf82012-06-12 16:20:34 +02001196struct mm_struct;
1197
Linus Torvalds1da177e2005-04-16 15:20:36 -07001198/*
1199 * Umount options
1200 */
1201
1202#define MNT_FORCE 0x00000001 /* Attempt to forcibily umount */
1203#define MNT_DETACH 0x00000002 /* Just detach from the tree */
1204#define MNT_EXPIRE 0x00000004 /* Mark for expiry */
Miklos Szeredidb1f05b2010-02-10 12:15:53 +01001205#define UMOUNT_NOFOLLOW 0x00000008 /* Don't follow symlink on umount */
1206#define UMOUNT_UNUSED 0x80000000 /* Flag guaranteed to be unused */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001207
1208extern struct list_head super_blocks;
1209extern spinlock_t sb_lock;
1210
Jan Kara5accdf82012-06-12 16:20:34 +02001211/* Possible states of 'frozen' field */
1212enum {
1213 SB_UNFROZEN = 0, /* FS is unfrozen */
1214 SB_FREEZE_WRITE = 1, /* Writes, dir ops, ioctls frozen */
Jan Kara5accdf82012-06-12 16:20:34 +02001215 SB_FREEZE_PAGEFAULT = 2, /* Page faults stopped as well */
1216 SB_FREEZE_FS = 3, /* For internal FS use (e.g. to stop
1217 * internal threads if needed) */
1218 SB_FREEZE_COMPLETE = 4, /* ->freeze_fs finished successfully */
1219};
1220
1221#define SB_FREEZE_LEVELS (SB_FREEZE_COMPLETE - 1)
1222
1223struct sb_writers {
1224 /* Counters for counting writers at each level */
1225 struct percpu_counter counter[SB_FREEZE_LEVELS];
1226 wait_queue_head_t wait; /* queue for waiting for
1227 writers / faults to finish */
1228 int frozen; /* Is sb frozen? */
1229 wait_queue_head_t wait_unfrozen; /* queue for waiting for
1230 sb to be thawed */
1231#ifdef CONFIG_DEBUG_LOCK_ALLOC
1232 struct lockdep_map lock_map[SB_FREEZE_LEVELS];
1233#endif
1234};
1235
Linus Torvalds1da177e2005-04-16 15:20:36 -07001236struct super_block {
1237 struct list_head s_list; /* Keep this first */
1238 dev_t s_dev; /* search index; _not_ kdev_t */
Richard Kennedy270ba5f2010-01-26 14:12:43 +00001239 unsigned char s_blocksize_bits;
1240 unsigned long s_blocksize;
Jeff Layton42cb56a2009-09-18 13:05:53 -07001241 loff_t s_maxbytes; /* Max file size */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001242 struct file_system_type *s_type;
Josef 'Jeff' Sipekee9b6d62007-02-12 00:55:41 -08001243 const struct super_operations *s_op;
Alexey Dobriyan61e225d2009-09-21 17:01:08 -07001244 const struct dquot_operations *dq_op;
Alexey Dobriyan0d54b212009-09-21 17:01:09 -07001245 const struct quotactl_ops *s_qcop;
Christoph Hellwig39655162007-10-21 16:42:17 -07001246 const struct export_operations *s_export_op;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001247 unsigned long s_flags;
1248 unsigned long s_magic;
1249 struct dentry *s_root;
1250 struct rw_semaphore s_umount;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001251 int s_count;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001252 atomic_t s_active;
Alexey Dobriyan50462062006-09-29 02:00:01 -07001253#ifdef CONFIG_SECURITY
Linus Torvalds1da177e2005-04-16 15:20:36 -07001254 void *s_security;
Alexey Dobriyan50462062006-09-29 02:00:01 -07001255#endif
Stephen Hemmingerbb435452010-05-13 17:53:14 -07001256 const struct xattr_handler **s_xattr;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001257
1258 struct list_head s_inodes; /* all inodes */
Nick Pigginceb5bdc2011-01-07 17:50:05 +11001259 struct hlist_bl_head s_anon; /* anonymous dentries for (nfs) exporting */
Nick Piggin6416ccb2010-08-18 04:37:38 +10001260#ifdef CONFIG_SMP
1261 struct list_head __percpu *s_files;
1262#else
Linus Torvalds1da177e2005-04-16 15:20:36 -07001263 struct list_head s_files;
Nick Piggin6416ccb2010-08-18 04:37:38 +10001264#endif
Miklos Szeredi39f7c4d2011-11-21 12:11:30 +01001265 struct list_head s_mounts; /* list of mounts; _not_ for fs use */
Nick Pigginb5c84bf2011-01-07 17:49:38 +11001266 /* s_dentry_lru, s_nr_dentry_unused protected by dcache.c lru locks */
Kentaro Makitada3bbdd2008-07-23 21:27:13 -07001267 struct list_head s_dentry_lru; /* unused dentry lru */
1268 int s_nr_dentry_unused; /* # of dentry on lru */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001269
Dave Chinner09cc9fc2011-07-08 14:14:40 +10001270 /* s_inode_lru_lock protects s_inode_lru and s_nr_inodes_unused */
1271 spinlock_t s_inode_lru_lock ____cacheline_aligned_in_smp;
Dave Chinner98b745c2011-07-08 14:14:39 +10001272 struct list_head s_inode_lru; /* unused inode lru */
1273 int s_nr_inodes_unused; /* # of inodes on lru */
1274
Linus Torvalds1da177e2005-04-16 15:20:36 -07001275 struct block_device *s_bdev;
Jens Axboe32a88aa2009-09-16 15:02:33 +02001276 struct backing_dev_info *s_bdi;
David Howellsacaebfd2007-05-10 22:51:50 -07001277 struct mtd_info *s_mtd;
Al Viroa5166162011-12-12 22:53:00 -05001278 struct hlist_node s_instances;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001279 struct quota_info s_dquot; /* Diskquota specific options */
1280
Jan Kara5accdf82012-06-12 16:20:34 +02001281 struct sb_writers s_writers;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001282
1283 char s_id[32]; /* Informational name */
Linus Torvalds0bba0162011-04-12 15:21:04 -07001284 u8 s_uuid[16]; /* UUID */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001285
1286 void *s_fs_info; /* Filesystem private info */
Al Viro8de52772012-02-06 12:45:27 -05001287 unsigned int s_max_links;
Al Viro30c40d22008-02-22 19:50:45 -05001288 fmode_t s_mode;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001289
Richard Kennedy270ba5f2010-01-26 14:12:43 +00001290 /* Granularity of c/m/atime in ns.
1291 Cannot be worse than a second */
1292 u32 s_time_gran;
1293
Linus Torvalds1da177e2005-04-16 15:20:36 -07001294 /*
1295 * The next field is for VFS *only*. No filesystems have any business
1296 * even looking at it. You had been warned.
1297 */
Arjan van de Vena11f3a02006-03-23 03:00:33 -08001298 struct mutex s_vfs_rename_mutex; /* Kludge */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001299
Miklos Szeredi79c0b2d2007-05-08 00:25:43 -07001300 /*
1301 * Filesystem subtype. If non-empty the filesystem type field
1302 * in /proc/mounts will be "type.subtype"
1303 */
1304 char *s_subtype;
Miklos Szeredib3b304a2008-02-08 04:21:35 -08001305
1306 /*
1307 * Saved mount options for lazy filesystems using
1308 * generic_show_options()
1309 */
Arnd Bergmann4d2deb42010-02-24 20:01:56 +01001310 char __rcu *s_options;
Al Viroc8aebb02010-12-18 10:22:30 -05001311 const struct dentry_operations *s_d_op; /* default d_op for dentries */
Dan Magenheimer9fdfdcf2011-05-26 10:01:19 -06001312
1313 /*
1314 * Saved pool identifier for cleancache (-1 means none)
1315 */
1316 int cleancache_poolid;
Dave Chinnerb0d40c92011-07-08 14:14:42 +10001317
1318 struct shrinker s_shrink; /* per-sb shrinker handle */
Miklos Szeredi4ed5e822011-11-21 12:11:31 +01001319
Miklos Szeredi7ada4db2011-11-21 12:11:32 +01001320 /* Number of inodes with nlink == 0 but still referenced */
1321 atomic_long_t s_remove_count;
1322
Miklos Szeredi4ed5e822011-11-21 12:11:31 +01001323 /* Being remounted read-only */
1324 int s_readonly_remount;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001325};
1326
Dave Chinnerb0d40c92011-07-08 14:14:42 +10001327/* superblock cache pruning functions */
1328extern void prune_icache_sb(struct super_block *sb, int nr_to_scan);
1329extern void prune_dcache_sb(struct super_block *sb, int nr_to_scan);
1330
Linus Torvalds1da177e2005-04-16 15:20:36 -07001331extern struct timespec current_fs_time(struct super_block *sb);
1332
1333/*
1334 * Snapshotting support.
1335 */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001336
Jan Kara5accdf82012-06-12 16:20:34 +02001337void __sb_end_write(struct super_block *sb, int level);
1338int __sb_start_write(struct super_block *sb, int level, bool wait);
1339
1340/**
1341 * sb_end_write - drop write access to a superblock
1342 * @sb: the super we wrote to
1343 *
1344 * Decrement number of writers to the filesystem. Wake up possible waiters
1345 * wanting to freeze the filesystem.
1346 */
1347static inline void sb_end_write(struct super_block *sb)
1348{
1349 __sb_end_write(sb, SB_FREEZE_WRITE);
1350}
1351
1352/**
1353 * sb_end_pagefault - drop write access to a superblock from a page fault
1354 * @sb: the super we wrote to
1355 *
1356 * Decrement number of processes handling write page fault to the filesystem.
1357 * Wake up possible waiters wanting to freeze the filesystem.
1358 */
1359static inline void sb_end_pagefault(struct super_block *sb)
1360{
1361 __sb_end_write(sb, SB_FREEZE_PAGEFAULT);
1362}
1363
1364/**
1365 * sb_end_intwrite - drop write access to a superblock for internal fs purposes
1366 * @sb: the super we wrote to
1367 *
1368 * Decrement fs-internal number of writers to the filesystem. Wake up possible
1369 * waiters wanting to freeze the filesystem.
1370 */
1371static inline void sb_end_intwrite(struct super_block *sb)
1372{
1373 __sb_end_write(sb, SB_FREEZE_FS);
1374}
1375
1376/**
1377 * sb_start_write - get write access to a superblock
1378 * @sb: the super we write to
1379 *
1380 * When a process wants to write data or metadata to a file system (i.e. dirty
1381 * a page or an inode), it should embed the operation in a sb_start_write() -
1382 * sb_end_write() pair to get exclusion against file system freezing. This
1383 * function increments number of writers preventing freezing. If the file
1384 * system is already frozen, the function waits until the file system is
1385 * thawed.
1386 *
1387 * Since freeze protection behaves as a lock, users have to preserve
1388 * ordering of freeze protection and other filesystem locks. Generally,
1389 * freeze protection should be the outermost lock. In particular, we have:
1390 *
1391 * sb_start_write
1392 * -> i_mutex (write path, truncate, directory ops, ...)
1393 * -> s_umount (freeze_super, thaw_super)
1394 */
1395static inline void sb_start_write(struct super_block *sb)
1396{
1397 __sb_start_write(sb, SB_FREEZE_WRITE, true);
1398}
1399
1400static inline int sb_start_write_trylock(struct super_block *sb)
1401{
1402 return __sb_start_write(sb, SB_FREEZE_WRITE, false);
1403}
1404
1405/**
1406 * sb_start_pagefault - get write access to a superblock from a page fault
1407 * @sb: the super we write to
1408 *
1409 * When a process starts handling write page fault, it should embed the
1410 * operation into sb_start_pagefault() - sb_end_pagefault() pair to get
1411 * exclusion against file system freezing. This is needed since the page fault
1412 * is going to dirty a page. This function increments number of running page
1413 * faults preventing freezing. If the file system is already frozen, the
1414 * function waits until the file system is thawed.
1415 *
1416 * Since page fault freeze protection behaves as a lock, users have to preserve
1417 * ordering of freeze protection and other filesystem locks. It is advised to
1418 * put sb_start_pagefault() close to mmap_sem in lock ordering. Page fault
1419 * handling code implies lock dependency:
1420 *
1421 * mmap_sem
1422 * -> sb_start_pagefault
1423 */
1424static inline void sb_start_pagefault(struct super_block *sb)
1425{
1426 __sb_start_write(sb, SB_FREEZE_PAGEFAULT, true);
1427}
1428
1429/*
1430 * sb_start_intwrite - get write access to a superblock for internal fs purposes
1431 * @sb: the super we write to
1432 *
1433 * This is the third level of protection against filesystem freezing. It is
1434 * free for use by a filesystem. The only requirement is that it must rank
1435 * below sb_start_pagefault.
1436 *
1437 * For example filesystem can call sb_start_intwrite() when starting a
1438 * transaction which somewhat eases handling of freezing for internal sources
1439 * of filesystem changes (internal fs threads, discarding preallocation on file
1440 * close, etc.).
1441 */
1442static inline void sb_start_intwrite(struct super_block *sb)
1443{
1444 __sb_start_write(sb, SB_FREEZE_FS, true);
1445}
1446
Linus Torvalds1da177e2005-04-16 15:20:36 -07001447
Serge E. Hallyn2e149672011-03-23 16:43:26 -07001448extern bool inode_owner_or_capable(const struct inode *inode);
Satyam Sharma3bd858a2007-07-17 15:00:08 +05301449
Linus Torvalds1da177e2005-04-16 15:20:36 -07001450/*
1451 * VFS helper functions..
1452 */
Al Viro312b63f2012-06-10 18:09:36 -04001453extern int vfs_create(struct inode *, struct dentry *, umode_t, bool);
Al Viro18bb1db2011-07-26 01:41:39 -04001454extern int vfs_mkdir(struct inode *, struct dentry *, umode_t);
Al Viro1a67aaf2011-07-26 01:52:52 -04001455extern int vfs_mknod(struct inode *, struct dentry *, umode_t, dev_t);
Miklos Szeredidb2e7472008-06-24 16:50:16 +02001456extern int vfs_symlink(struct inode *, struct dentry *, const char *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001457extern int vfs_link(struct dentry *, struct inode *, struct dentry *);
1458extern int vfs_rmdir(struct inode *, struct dentry *);
1459extern int vfs_unlink(struct inode *, struct dentry *);
1460extern int vfs_rename(struct inode *, struct dentry *, struct inode *, struct dentry *);
1461
1462/*
1463 * VFS dentry helper functions.
1464 */
1465extern void dentry_unhash(struct dentry *dentry);
1466
1467/*
Christoph Hellwig8c744fb2005-11-08 21:35:04 -08001468 * VFS file helper functions.
1469 */
Dmitry Monakhova1bd1202010-03-04 17:29:14 +03001470extern void inode_init_owner(struct inode *inode, const struct inode *dir,
Al Viro62bb1092011-07-24 23:20:18 -04001471 umode_t mode);
Christoph Hellwig8c744fb2005-11-08 21:35:04 -08001472/*
Mark Fashehc4b929b2008-10-08 19:44:18 -04001473 * VFS FS_IOC_FIEMAP helper definitions.
1474 */
1475struct fiemap_extent_info {
1476 unsigned int fi_flags; /* Flags as passed from user */
1477 unsigned int fi_extents_mapped; /* Number of mapped extents */
1478 unsigned int fi_extents_max; /* Size of fiemap_extent array */
Namhyung Kimecf56322011-01-16 23:28:17 +09001479 struct fiemap_extent __user *fi_extents_start; /* Start of
1480 fiemap_extent array */
Mark Fashehc4b929b2008-10-08 19:44:18 -04001481};
1482int fiemap_fill_next_extent(struct fiemap_extent_info *info, u64 logical,
1483 u64 phys, u64 len, u32 flags);
1484int fiemap_check_flags(struct fiemap_extent_info *fieinfo, u32 fs_flags);
1485
1486/*
Linus Torvalds1da177e2005-04-16 15:20:36 -07001487 * File types
1488 *
1489 * NOTE! These match bits 12..15 of stat.st_mode
1490 * (ie "(i_mode >> 12) & 15").
1491 */
1492#define DT_UNKNOWN 0
1493#define DT_FIFO 1
1494#define DT_CHR 2
1495#define DT_DIR 4
1496#define DT_BLK 6
1497#define DT_REG 8
1498#define DT_LNK 10
1499#define DT_SOCK 12
1500#define DT_WHT 14
1501
Linus Torvalds1da177e2005-04-16 15:20:36 -07001502/*
1503 * This is the "filldir" function type, used by readdir() to let
1504 * the kernel specify what kind of dirent layout it wants to have.
1505 * This allows the kernel to read directories into kernel space or
1506 * to have different dirent layouts depending on the binary type.
1507 */
David Howellsafefdbb2006-10-03 01:13:46 -07001508typedef int (*filldir_t)(void *, const char *, int, loff_t, u64, unsigned);
Al Viro08f85852007-10-08 13:26:20 -04001509struct block_device_operations;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001510
Linus Torvalds1da177e2005-04-16 15:20:36 -07001511/* These macros are for out of kernel modules to test that
1512 * the kernel supports the unlocked_ioctl and compat_ioctl
1513 * fields in struct file_operations. */
1514#define HAVE_COMPAT_IOCTL 1
1515#define HAVE_UNLOCKED_IOCTL 1
1516
Linus Torvalds1da177e2005-04-16 15:20:36 -07001517struct file_operations {
1518 struct module *owner;
1519 loff_t (*llseek) (struct file *, loff_t, int);
1520 ssize_t (*read) (struct file *, char __user *, size_t, loff_t *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001521 ssize_t (*write) (struct file *, const char __user *, size_t, loff_t *);
Badari Pulavarty027445c2006-09-30 23:28:46 -07001522 ssize_t (*aio_read) (struct kiocb *, const struct iovec *, unsigned long, loff_t);
1523 ssize_t (*aio_write) (struct kiocb *, const struct iovec *, unsigned long, loff_t);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001524 int (*readdir) (struct file *, void *, filldir_t);
1525 unsigned int (*poll) (struct file *, struct poll_table_struct *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001526 long (*unlocked_ioctl) (struct file *, unsigned int, unsigned long);
1527 long (*compat_ioctl) (struct file *, unsigned int, unsigned long);
1528 int (*mmap) (struct file *, struct vm_area_struct *);
1529 int (*open) (struct inode *, struct file *);
Miklos Szeredi75e1fcc2006-06-23 02:05:12 -07001530 int (*flush) (struct file *, fl_owner_t id);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001531 int (*release) (struct inode *, struct file *);
Josef Bacik02c24a82011-07-16 20:44:56 -04001532 int (*fsync) (struct file *, loff_t, loff_t, int datasync);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001533 int (*aio_fsync) (struct kiocb *, int datasync);
1534 int (*fasync) (int, struct file *, int);
1535 int (*lock) (struct file *, int, struct file_lock *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001536 ssize_t (*sendpage) (struct file *, struct page *, int, size_t, loff_t *, int);
1537 unsigned long (*get_unmapped_area)(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
1538 int (*check_flags)(int);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001539 int (*flock) (struct file *, int, struct file_lock *);
Jens Axboecbb7e572006-04-11 14:57:50 +02001540 ssize_t (*splice_write)(struct pipe_inode_info *, struct file *, loff_t *, size_t, unsigned int);
1541 ssize_t (*splice_read)(struct file *, loff_t *, struct pipe_inode_info *, size_t, unsigned int);
J. Bruce Fieldsf9ffed22006-11-14 15:51:40 -05001542 int (*setlease)(struct file *, long, struct file_lock **);
Christoph Hellwig2fe17c12011-01-14 13:07:43 +01001543 long (*fallocate)(struct file *file, int mode, loff_t offset,
1544 loff_t len);
Cyrill Gorcunov55985dd2012-12-17 16:04:55 -08001545 int (*show_fdinfo)(struct seq_file *m, struct file *f);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001546};
1547
1548struct inode_operations {
Al Viro00cd8dd2012-06-10 17:13:09 -04001549 struct dentry * (*lookup) (struct inode *,struct dentry *, unsigned int);
Nick Piggin44a7d7a2011-01-07 17:49:56 +11001550 void * (*follow_link) (struct dentry *, struct nameidata *);
Al Viro10556cb2011-06-20 19:28:19 -04001551 int (*permission) (struct inode *, int);
Christoph Hellwig4e34e712011-07-23 17:37:31 +02001552 struct posix_acl * (*get_acl)(struct inode *, int);
Nick Piggin44a7d7a2011-01-07 17:49:56 +11001553
1554 int (*readlink) (struct dentry *, char __user *,int);
1555 void (*put_link) (struct dentry *, struct nameidata *, void *);
1556
Al Viroebfc3b42012-06-10 18:05:36 -04001557 int (*create) (struct inode *,struct dentry *, umode_t, bool);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001558 int (*link) (struct dentry *,struct inode *,struct dentry *);
1559 int (*unlink) (struct inode *,struct dentry *);
1560 int (*symlink) (struct inode *,struct dentry *,const char *);
Al Viro18bb1db2011-07-26 01:41:39 -04001561 int (*mkdir) (struct inode *,struct dentry *,umode_t);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001562 int (*rmdir) (struct inode *,struct dentry *);
Al Viro1a67aaf2011-07-26 01:52:52 -04001563 int (*mknod) (struct inode *,struct dentry *,umode_t,dev_t);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001564 int (*rename) (struct inode *, struct dentry *,
1565 struct inode *, struct dentry *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001566 int (*setattr) (struct dentry *, struct iattr *);
1567 int (*getattr) (struct vfsmount *mnt, struct dentry *, struct kstat *);
1568 int (*setxattr) (struct dentry *, const char *,const void *,size_t,int);
1569 ssize_t (*getxattr) (struct dentry *, const char *, void *, size_t);
1570 ssize_t (*listxattr) (struct dentry *, char *, size_t);
1571 int (*removexattr) (struct dentry *, const char *);
Mark Fashehc4b929b2008-10-08 19:44:18 -04001572 int (*fiemap)(struct inode *, struct fiemap_extent_info *, u64 start,
1573 u64 len);
Josef Bacikc3b2da32012-03-26 09:59:21 -04001574 int (*update_time)(struct inode *, struct timespec *, int);
Al Virod9585272012-06-22 12:39:14 +04001575 int (*atomic_open)(struct inode *, struct dentry *,
Al Viro30d90492012-06-22 12:40:19 +04001576 struct file *, unsigned open_flag,
Al Virod9585272012-06-22 12:39:14 +04001577 umode_t create_mode, int *opened);
Nick Piggin44a7d7a2011-01-07 17:49:56 +11001578} ____cacheline_aligned;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001579
Badari Pulavartyeed4e512006-09-30 23:28:49 -07001580ssize_t rw_copy_check_uvector(int type, const struct iovec __user * uvector,
Christopher Yeohfcf63402011-10-31 17:06:39 -07001581 unsigned long nr_segs, unsigned long fast_segs,
1582 struct iovec *fast_pointer,
Christopher Yeohac34ebb2012-05-31 16:26:42 -07001583 struct iovec **ret_pointer);
Badari Pulavartyeed4e512006-09-30 23:28:49 -07001584
Linus Torvalds1da177e2005-04-16 15:20:36 -07001585extern ssize_t vfs_read(struct file *, char __user *, size_t, loff_t *);
1586extern ssize_t vfs_write(struct file *, const char __user *, size_t, loff_t *);
1587extern ssize_t vfs_readv(struct file *, const struct iovec __user *,
1588 unsigned long, loff_t *);
1589extern ssize_t vfs_writev(struct file *, const struct iovec __user *,
1590 unsigned long, loff_t *);
1591
Linus Torvalds1da177e2005-04-16 15:20:36 -07001592struct super_operations {
1593 struct inode *(*alloc_inode)(struct super_block *sb);
1594 void (*destroy_inode)(struct inode *);
1595
Christoph Hellwigaa385722011-05-27 06:53:02 -04001596 void (*dirty_inode) (struct inode *, int flags);
Christoph Hellwiga9185b42010-03-05 09:21:37 +01001597 int (*write_inode) (struct inode *, struct writeback_control *wbc);
Al Viro45321ac2010-06-07 13:43:19 -04001598 int (*drop_inode) (struct inode *);
Al Virobe7ce412010-06-04 19:40:39 -04001599 void (*evict_inode) (struct inode *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001600 void (*put_super) (struct super_block *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001601 int (*sync_fs)(struct super_block *sb, int wait);
Takashi Satoc4be0c12009-01-09 16:40:58 -08001602 int (*freeze_fs) (struct super_block *);
1603 int (*unfreeze_fs) (struct super_block *);
David Howells726c3342006-06-23 02:02:58 -07001604 int (*statfs) (struct dentry *, struct kstatfs *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001605 int (*remount_fs) (struct super_block *, int *, char *);
Al Viro42faad92008-04-24 07:21:56 -04001606 void (*umount_begin) (struct super_block *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001607
Al Viro34c80b12011-12-08 21:32:45 -05001608 int (*show_options)(struct seq_file *, struct dentry *);
Al Virod861c632011-12-08 21:32:45 -05001609 int (*show_devname)(struct seq_file *, struct dentry *);
Al Viroa6322de2011-12-08 21:37:57 -05001610 int (*show_path)(struct seq_file *, struct dentry *);
Al Viro64132372011-12-08 20:51:13 -05001611 int (*show_stats)(struct seq_file *, struct dentry *);
Alexey Dobriyan0e51a722006-09-29 01:59:56 -07001612#ifdef CONFIG_QUOTA
Linus Torvalds1da177e2005-04-16 15:20:36 -07001613 ssize_t (*quota_read)(struct super_block *, int, char *, size_t, loff_t);
1614 ssize_t (*quota_write)(struct super_block *, int, const char *, size_t, loff_t);
Alexey Dobriyan0e51a722006-09-29 01:59:56 -07001615#endif
Theodore Ts'o87d8fe12009-01-03 09:47:09 -05001616 int (*bdev_try_to_free_page)(struct super_block*, struct page*, gfp_t);
Dave Chinner0e1fdaf2011-07-08 14:14:44 +10001617 int (*nr_cached_objects)(struct super_block *);
1618 void (*free_cached_objects)(struct super_block *, int);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001619};
1620
Joern Engel1c0eeaf2007-10-16 23:30:44 -07001621/*
David Howellsbbc10962012-10-15 16:40:35 +01001622 * Inode flags - they have no relation to superblock flags now
1623 */
1624#define S_SYNC 1 /* Writes are synced at once */
1625#define S_NOATIME 2 /* Do not update access times */
1626#define S_APPEND 4 /* Append-only file */
1627#define S_IMMUTABLE 8 /* Immutable file */
1628#define S_DEAD 16 /* removed, but still open directory */
1629#define S_NOQUOTA 32 /* Inode is not counted to quota */
1630#define S_DIRSYNC 64 /* Directory modifications are synchronous */
1631#define S_NOCMTIME 128 /* Do not update file c/mtime */
1632#define S_SWAPFILE 256 /* Do not truncate: swapon got its bmaps */
1633#define S_PRIVATE 512 /* Inode is fs-internal */
1634#define S_IMA 1024 /* Inode has an associated IMA struct */
1635#define S_AUTOMOUNT 2048 /* Automount/referral quasi-directory */
1636#define S_NOSEC 4096 /* no suid or xattr security attributes */
1637
1638/*
1639 * Note that nosuid etc flags are inode-specific: setting some file-system
1640 * flags just means all the inodes inherit those flags by default. It might be
1641 * possible to override it selectively if you really wanted to with some
1642 * ioctl() that is not currently implemented.
1643 *
1644 * Exception: MS_RDONLY is always applied to the entire file system.
1645 *
1646 * Unfortunately, it is possible to change a filesystems flags with it mounted
1647 * with files in use. This means that all of the inodes will not have their
1648 * i_flags updated. Hence, i_flags no longer inherit the superblock mount
1649 * flags, so these have to be checked separately. -- rmk@arm.uk.linux.org
1650 */
1651#define __IS_FLG(inode, flg) ((inode)->i_sb->s_flags & (flg))
1652
1653#define IS_RDONLY(inode) ((inode)->i_sb->s_flags & MS_RDONLY)
1654#define IS_SYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS) || \
1655 ((inode)->i_flags & S_SYNC))
1656#define IS_DIRSYNC(inode) (__IS_FLG(inode, MS_SYNCHRONOUS|MS_DIRSYNC) || \
1657 ((inode)->i_flags & (S_SYNC|S_DIRSYNC)))
1658#define IS_MANDLOCK(inode) __IS_FLG(inode, MS_MANDLOCK)
1659#define IS_NOATIME(inode) __IS_FLG(inode, MS_RDONLY|MS_NOATIME)
1660#define IS_I_VERSION(inode) __IS_FLG(inode, MS_I_VERSION)
1661
1662#define IS_NOQUOTA(inode) ((inode)->i_flags & S_NOQUOTA)
1663#define IS_APPEND(inode) ((inode)->i_flags & S_APPEND)
1664#define IS_IMMUTABLE(inode) ((inode)->i_flags & S_IMMUTABLE)
1665#define IS_POSIXACL(inode) __IS_FLG(inode, MS_POSIXACL)
1666
1667#define IS_DEADDIR(inode) ((inode)->i_flags & S_DEAD)
1668#define IS_NOCMTIME(inode) ((inode)->i_flags & S_NOCMTIME)
1669#define IS_SWAPFILE(inode) ((inode)->i_flags & S_SWAPFILE)
1670#define IS_PRIVATE(inode) ((inode)->i_flags & S_PRIVATE)
1671#define IS_IMA(inode) ((inode)->i_flags & S_IMA)
1672#define IS_AUTOMOUNT(inode) ((inode)->i_flags & S_AUTOMOUNT)
1673#define IS_NOSEC(inode) ((inode)->i_flags & S_NOSEC)
1674
1675/*
Dave Chinner250df6e2011-03-22 22:23:36 +11001676 * Inode state bits. Protected by inode->i_lock
Joern Engel1c0eeaf2007-10-16 23:30:44 -07001677 *
1678 * Three bits determine the dirty state of the inode, I_DIRTY_SYNC,
1679 * I_DIRTY_DATASYNC and I_DIRTY_PAGES.
1680 *
1681 * Four bits define the lifetime of an inode. Initially, inodes are I_NEW,
1682 * until that flag is cleared. I_WILL_FREE, I_FREEING and I_CLEAR are set at
1683 * various stages of removing an inode.
1684 *
Christoph Hellwigeaff8072009-12-17 14:25:01 +01001685 * Two bits are used for locking and completion notification, I_NEW and I_SYNC.
Joern Engel1c0eeaf2007-10-16 23:30:44 -07001686 *
Joern Engele7ca2d42008-02-06 01:36:59 -08001687 * I_DIRTY_SYNC Inode is dirty, but doesn't have to be written on
1688 * fdatasync(). i_atime is the usual cause.
Jan Kara000cb482008-02-14 19:31:32 -08001689 * I_DIRTY_DATASYNC Data-related inode changes pending. We keep track of
1690 * these changes separately from I_DIRTY_SYNC so that we
1691 * don't have to write inode on fdatasync() when only
1692 * mtime has changed in it.
Joern Engel1c0eeaf2007-10-16 23:30:44 -07001693 * I_DIRTY_PAGES Inode has dirty pages. Inode itself may be clean.
Christoph Hellwigeaff8072009-12-17 14:25:01 +01001694 * I_NEW Serves as both a mutex and completion notification.
1695 * New inodes set I_NEW. If two processes both create
1696 * the same inode, one of them will release its inode and
1697 * wait for I_NEW to be released before returning.
1698 * Inodes in I_WILL_FREE, I_FREEING or I_CLEAR state can
1699 * also cause waiting on I_NEW, without I_NEW actually
1700 * being set. find_inode() uses this to prevent returning
1701 * nearly-dead inodes.
Joern Engel1c0eeaf2007-10-16 23:30:44 -07001702 * I_WILL_FREE Must be set when calling write_inode_now() if i_count
1703 * is zero. I_FREEING must be set when I_WILL_FREE is
1704 * cleared.
1705 * I_FREEING Set when inode is about to be freed but still has dirty
1706 * pages or buffers attached or the inode itself is still
1707 * dirty.
Jan Karadbd57682012-05-03 14:48:02 +02001708 * I_CLEAR Added by clear_inode(). In this state the inode is
1709 * clean and can be destroyed. Inode keeps I_FREEING.
Joern Engel1c0eeaf2007-10-16 23:30:44 -07001710 *
1711 * Inodes that are I_WILL_FREE, I_FREEING or I_CLEAR are
1712 * prohibited for many purposes. iget() must wait for
1713 * the inode to be completely released, then create it
1714 * anew. Other functions will just ignore such inodes,
Christoph Hellwigeaff8072009-12-17 14:25:01 +01001715 * if appropriate. I_NEW is used for waiting.
Joern Engel1c0eeaf2007-10-16 23:30:44 -07001716 *
Jan Kara169ebd92012-05-03 14:48:03 +02001717 * I_SYNC Writeback of inode is running. The bit is set during
1718 * data writeback, and cleared with a wakeup on the bit
1719 * address once it is done. The bit is also used to pin
1720 * the inode in memory for flusher thread.
Joern Engel1c0eeaf2007-10-16 23:30:44 -07001721 *
Christoph Hellwigbd5fe6c2011-06-24 14:29:43 -04001722 * I_REFERENCED Marks the inode as recently references on the LRU list.
1723 *
1724 * I_DIO_WAKEUP Never set. Only used as a key for wait_on_bit().
1725 *
Joern Engel1c0eeaf2007-10-16 23:30:44 -07001726 * Q: What is the difference between I_WILL_FREE and I_FREEING?
Joern Engel1c0eeaf2007-10-16 23:30:44 -07001727 */
Nick Piggin9e38d862010-10-23 06:55:17 -04001728#define I_DIRTY_SYNC (1 << 0)
1729#define I_DIRTY_DATASYNC (1 << 1)
1730#define I_DIRTY_PAGES (1 << 2)
Christoph Hellwigeaff8072009-12-17 14:25:01 +01001731#define __I_NEW 3
1732#define I_NEW (1 << __I_NEW)
Nick Piggin9e38d862010-10-23 06:55:17 -04001733#define I_WILL_FREE (1 << 4)
1734#define I_FREEING (1 << 5)
1735#define I_CLEAR (1 << 6)
Christoph Hellwigeaff8072009-12-17 14:25:01 +01001736#define __I_SYNC 7
Joern Engel1c0eeaf2007-10-16 23:30:44 -07001737#define I_SYNC (1 << __I_SYNC)
Nick Piggin9e38d862010-10-23 06:55:17 -04001738#define I_REFERENCED (1 << 8)
Christoph Hellwigbd5fe6c2011-06-24 14:29:43 -04001739#define __I_DIO_WAKEUP 9
1740#define I_DIO_WAKEUP (1 << I_DIO_WAKEUP)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001741
1742#define I_DIRTY (I_DIRTY_SYNC | I_DIRTY_DATASYNC | I_DIRTY_PAGES)
1743
1744extern void __mark_inode_dirty(struct inode *, int);
1745static inline void mark_inode_dirty(struct inode *inode)
1746{
1747 __mark_inode_dirty(inode, I_DIRTY);
1748}
1749
1750static inline void mark_inode_dirty_sync(struct inode *inode)
1751{
1752 __mark_inode_dirty(inode, I_DIRTY_SYNC);
1753}
1754
Miklos Szeredi7ada4db2011-11-21 12:11:32 +01001755extern void inc_nlink(struct inode *inode);
1756extern void drop_nlink(struct inode *inode);
1757extern void clear_nlink(struct inode *inode);
1758extern void set_nlink(struct inode *inode, unsigned int nlink);
Dave Hansend8c76e62006-09-30 23:29:04 -07001759
1760static inline void inode_inc_link_count(struct inode *inode)
1761{
1762 inc_nlink(inode);
Alexey Dobriyan66d21732006-03-23 03:00:51 -08001763 mark_inode_dirty(inode);
1764}
1765
Dave Hansen9a53c3a2006-09-30 23:29:03 -07001766static inline void inode_dec_link_count(struct inode *inode)
1767{
1768 drop_nlink(inode);
Alexey Dobriyan66d21732006-03-23 03:00:51 -08001769 mark_inode_dirty(inode);
1770}
1771
Jean Noel Cordenner25ec56b2008-01-28 23:58:27 -05001772/**
1773 * inode_inc_iversion - increments i_version
1774 * @inode: inode that need to be updated
1775 *
1776 * Every time the inode is modified, the i_version field will be incremented.
1777 * The filesystem has to be mounted with i_version flag
1778 */
1779
1780static inline void inode_inc_iversion(struct inode *inode)
1781{
1782 spin_lock(&inode->i_lock);
1783 inode->i_version++;
1784 spin_unlock(&inode->i_lock);
1785}
1786
Josef Bacikc3b2da32012-03-26 09:59:21 -04001787enum file_time_flags {
1788 S_ATIME = 1,
1789 S_MTIME = 2,
1790 S_CTIME = 4,
1791 S_VERSION = 8,
1792};
1793
Al Viro68ac1232012-03-15 08:21:57 -04001794extern void touch_atime(struct path *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001795static inline void file_accessed(struct file *file)
1796{
1797 if (!(file->f_flags & O_NOATIME))
Al Viro68ac1232012-03-15 08:21:57 -04001798 touch_atime(&file->f_path);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001799}
1800
1801int sync_inode(struct inode *inode, struct writeback_control *wbc);
Christoph Hellwigc37650162010-10-06 10:48:20 +02001802int sync_inode_metadata(struct inode *inode, int wait);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001803
Linus Torvalds1da177e2005-04-16 15:20:36 -07001804struct file_system_type {
1805 const char *name;
1806 int fs_flags;
David Howellsbbc10962012-10-15 16:40:35 +01001807#define FS_REQUIRES_DEV 1
1808#define FS_BINARY_MOUNTDATA 2
1809#define FS_HAS_SUBTYPE 4
Eric W. Biederman0c55cfc2012-07-26 21:42:03 -07001810#define FS_USERNS_MOUNT 8 /* Can be mounted by userns root */
1811#define FS_USERNS_DEV_MOUNT 16 /* A userns mount does not imply MNT_NODEV */
David Howellsbbc10962012-10-15 16:40:35 +01001812#define FS_RENAME_DOES_D_MOVE 32768 /* FS will handle d_move() during rename() internally. */
Al Viroc96e41e2010-07-25 00:17:56 +04001813 struct dentry *(*mount) (struct file_system_type *, int,
1814 const char *, void *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001815 void (*kill_sb) (struct super_block *);
1816 struct module *owner;
1817 struct file_system_type * next;
Al Viroa5166162011-12-12 22:53:00 -05001818 struct hlist_head fs_supers;
Peter Zijlstrad475fd42007-10-15 14:51:31 +02001819
Ingo Molnarcf516242006-07-03 00:25:27 -07001820 struct lock_class_key s_lock_key;
Arjan van de Ven897c6ff2006-07-03 00:25:28 -07001821 struct lock_class_key s_umount_key;
Roland Dreier51ee0492010-04-27 14:23:57 -07001822 struct lock_class_key s_vfs_rename_key;
Jan Kara5accdf82012-06-12 16:20:34 +02001823 struct lock_class_key s_writers_key[SB_FREEZE_LEVELS];
Peter Zijlstrad475fd42007-10-15 14:51:31 +02001824
1825 struct lock_class_key i_lock_key;
1826 struct lock_class_key i_mutex_key;
Peter Zijlstra14358e62007-10-14 01:38:33 +02001827 struct lock_class_key i_mutex_dir_key;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001828};
1829
Eric W. Biederman7f78e032013-03-02 19:39:14 -08001830#define MODULE_ALIAS_FS(NAME) MODULE_ALIAS("fs-" NAME)
1831
Al Viroceefda62010-07-26 13:16:50 +04001832extern struct dentry *mount_ns(struct file_system_type *fs_type, int flags,
1833 void *data, int (*fill_super)(struct super_block *, void *, int));
Al Viro152a0832010-07-25 00:46:55 +04001834extern struct dentry *mount_bdev(struct file_system_type *fs_type,
1835 int flags, const char *dev_name, void *data,
1836 int (*fill_super)(struct super_block *, void *, int));
Al Virofc14f2f2010-07-25 01:48:30 +04001837extern struct dentry *mount_single(struct file_system_type *fs_type,
1838 int flags, void *data,
1839 int (*fill_super)(struct super_block *, void *, int));
Al Viro3c26ff62010-07-25 11:46:36 +04001840extern struct dentry *mount_nodev(struct file_system_type *fs_type,
1841 int flags, void *data,
1842 int (*fill_super)(struct super_block *, void *, int));
Al Viroea441d12011-11-16 21:43:59 -05001843extern struct dentry *mount_subtree(struct vfsmount *mnt, const char *path);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001844void generic_shutdown_super(struct super_block *sb);
1845void kill_block_super(struct super_block *sb);
1846void kill_anon_super(struct super_block *sb);
1847void kill_litter_super(struct super_block *sb);
1848void deactivate_super(struct super_block *sb);
Al Viro74dbbdd2009-05-06 01:07:50 -04001849void deactivate_locked_super(struct super_block *sb);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001850int set_anon_super(struct super_block *s, void *data);
Al Viro0ee5dc62011-07-07 15:44:25 -04001851int get_anon_bdev(dev_t *);
1852void free_anon_bdev(dev_t);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001853struct super_block *sget(struct file_system_type *type,
1854 int (*test)(struct super_block *,void *),
1855 int (*set)(struct super_block *,void *),
David Howells9249e172012-06-25 12:55:37 +01001856 int flags, void *data);
Al Viro51139ad2010-07-25 23:47:46 +04001857extern struct dentry *mount_pseudo(struct file_system_type *, char *,
Al Viroc74a1cb2011-01-12 16:59:34 -05001858 const struct super_operations *ops,
1859 const struct dentry_operations *dops,
1860 unsigned long);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001861
1862/* Alas, no aliases. Too much hassle with bringing module.h everywhere */
1863#define fops_get(fops) \
1864 (((fops) && try_module_get((fops)->owner) ? (fops) : NULL))
1865#define fops_put(fops) \
1866 do { if (fops) module_put((fops)->owner); } while(0)
1867
1868extern int register_filesystem(struct file_system_type *);
1869extern int unregister_filesystem(struct file_system_type *);
Pavel Emelyanov8bf97252007-10-18 23:40:02 -07001870extern struct vfsmount *kern_mount_data(struct file_system_type *, void *data);
1871#define kern_mount(type) kern_mount_data(type, NULL)
Tim Chen423e0ab2011-07-19 09:32:38 -07001872extern void kern_unmount(struct vfsmount *mnt);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001873extern int may_umount_tree(struct vfsmount *);
1874extern int may_umount(struct vfsmount *);
Al Viro808d4e32012-10-11 11:42:01 -04001875extern long do_mount(const char *, const char *, const char *, unsigned long, void *);
Al Viro589ff872009-04-18 03:28:19 -04001876extern struct vfsmount *collect_mounts(struct path *);
Al Viro8aec0802007-06-07 12:20:32 -04001877extern void drop_collected_mounts(struct vfsmount *);
Al Viro1f707132010-01-30 22:51:25 -05001878extern int iterate_mounts(int (*)(struct vfsmount *, void *), void *,
1879 struct vfsmount *);
Christoph Hellwigebabe9a2010-07-07 18:53:11 +02001880extern int vfs_statfs(struct path *, struct kstatfs *);
Al Viroc8b91ac2011-03-12 10:41:39 -05001881extern int user_statfs(const char __user *, struct kstatfs *);
1882extern int fd_statfs(int, struct kstatfs *);
Al Virocf31e702012-01-02 22:28:36 -05001883extern int vfs_ustat(dev_t, struct kstatfs *);
Josef Bacik18e9e512010-03-23 10:34:56 -04001884extern int freeze_super(struct super_block *super);
1885extern int thaw_super(struct super_block *super);
Al Viro02125a82011-12-05 08:43:34 -05001886extern bool our_mnt(struct vfsmount *mnt);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001887
Al Viroce3b0f82009-03-29 19:08:22 -04001888extern int current_umask(void);
1889
Miklos Szeredif87fd4c2006-01-16 22:14:23 -08001890/* /sys/fs */
Greg Kroah-Hartman00d26662007-10-29 14:17:23 -06001891extern struct kobject *fs_kobj;
Miklos Szeredif87fd4c2006-01-16 22:14:23 -08001892
Linus Torvalds435f49a2010-10-29 10:36:49 -07001893#define MAX_RW_COUNT (INT_MAX & PAGE_CACHE_MASK)
Thomas Petazzonibfcd17a2008-08-06 15:12:22 +02001894extern int rw_verify_area(int, struct file *, loff_t *, size_t);
1895
Linus Torvalds1da177e2005-04-16 15:20:36 -07001896#define FLOCK_VERIFY_READ 1
1897#define FLOCK_VERIFY_WRITE 2
1898
Thomas Petazzonibfcd17a2008-08-06 15:12:22 +02001899#ifdef CONFIG_FILE_LOCKING
Linus Torvalds1da177e2005-04-16 15:20:36 -07001900extern int locks_mandatory_locked(struct inode *);
1901extern int locks_mandatory_area(int, struct inode *, struct file *, loff_t, size_t);
1902
1903/*
1904 * Candidates for mandatory locking have the setgid bit set
1905 * but no group execute bit - an otherwise meaningless combination.
1906 */
Pavel Emelyanova16877c2007-10-01 14:41:11 -07001907
1908static inline int __mandatory_lock(struct inode *ino)
1909{
1910 return (ino->i_mode & (S_ISGID | S_IXGRP)) == S_ISGID;
1911}
1912
1913/*
1914 * ... and these candidates should be on MS_MANDLOCK mounted fs,
1915 * otherwise these will be advisory locks
1916 */
1917
1918static inline int mandatory_lock(struct inode *ino)
1919{
1920 return IS_MANDLOCK(ino) && __mandatory_lock(ino);
1921}
Linus Torvalds1da177e2005-04-16 15:20:36 -07001922
1923static inline int locks_verify_locked(struct inode *inode)
1924{
Pavel Emelyanova16877c2007-10-01 14:41:11 -07001925 if (mandatory_lock(inode))
Linus Torvalds1da177e2005-04-16 15:20:36 -07001926 return locks_mandatory_locked(inode);
1927 return 0;
1928}
1929
Linus Torvalds1da177e2005-04-16 15:20:36 -07001930static inline int locks_verify_truncate(struct inode *inode,
1931 struct file *filp,
1932 loff_t size)
1933{
Pavel Emelyanova16877c2007-10-01 14:41:11 -07001934 if (inode->i_flock && mandatory_lock(inode))
Linus Torvalds1da177e2005-04-16 15:20:36 -07001935 return locks_mandatory_area(
1936 FLOCK_VERIFY_WRITE, inode, filp,
1937 size < inode->i_size ? size : inode->i_size,
1938 (size < inode->i_size ? inode->i_size - size
1939 : size - inode->i_size)
1940 );
1941 return 0;
1942}
1943
1944static inline int break_lease(struct inode *inode, unsigned int mode)
1945{
1946 if (inode->i_flock)
1947 return __break_lease(inode, mode);
1948 return 0;
1949}
Thomas Petazzonibfcd17a2008-08-06 15:12:22 +02001950#else /* !CONFIG_FILE_LOCKING */
Steven Whitehouseaf5df562009-01-20 10:29:46 +00001951static inline int locks_mandatory_locked(struct inode *inode)
1952{
1953 return 0;
1954}
1955
1956static inline int locks_mandatory_area(int rw, struct inode *inode,
1957 struct file *filp, loff_t offset,
1958 size_t count)
1959{
1960 return 0;
1961}
1962
1963static inline int __mandatory_lock(struct inode *inode)
1964{
1965 return 0;
1966}
1967
1968static inline int mandatory_lock(struct inode *inode)
1969{
1970 return 0;
1971}
1972
1973static inline int locks_verify_locked(struct inode *inode)
1974{
1975 return 0;
1976}
1977
1978static inline int locks_verify_truncate(struct inode *inode, struct file *filp,
1979 size_t size)
1980{
1981 return 0;
1982}
1983
1984static inline int break_lease(struct inode *inode, unsigned int mode)
1985{
1986 return 0;
1987}
1988
Thomas Petazzonibfcd17a2008-08-06 15:12:22 +02001989#endif /* CONFIG_FILE_LOCKING */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001990
1991/* fs/open.c */
Jeff Laytonadb5c242012-10-10 16:43:13 -04001992struct audit_names;
Jeff Layton91a27b22012-10-10 15:25:28 -04001993struct filename {
Jeff Laytonadb5c242012-10-10 16:43:13 -04001994 const char *name; /* pointer to actual string */
1995 const __user char *uptr; /* original userland pointer */
1996 struct audit_names *aname;
Jeff Layton7950e382012-10-10 16:43:13 -04001997 bool separate; /* should "name" be freed? */
Jeff Layton91a27b22012-10-10 15:25:28 -04001998};
Linus Torvalds1da177e2005-04-16 15:20:36 -07001999
David Howellsa02de962012-12-20 21:52:36 +00002000extern long vfs_truncate(struct path *, loff_t);
NeilBrown4a301312006-01-08 01:02:39 -08002001extern int do_truncate(struct dentry *, loff_t start, unsigned int time_attrs,
2002 struct file *filp);
Ankit Jain3e63cbb2009-06-19 14:28:07 -04002003extern int do_fallocate(struct file *file, int mode, loff_t offset,
2004 loff_t len);
Jason Uhlenkott8e8a1402007-10-20 03:16:18 +02002005extern long do_sys_open(int dfd, const char __user *filename, int flags,
Al Viroa218d0f2011-11-21 14:59:34 -05002006 umode_t mode);
Jeff Layton669abf42012-10-10 16:43:10 -04002007extern struct file *file_open_name(struct filename *, int, umode_t);
Al Viroa218d0f2011-11-21 14:59:34 -05002008extern struct file *filp_open(const char *, int, umode_t);
Al Viro73d049a2011-03-11 12:08:24 -05002009extern struct file *file_open_root(struct dentry *, struct vfsmount *,
2010 const char *, int);
Al Viro765927b2012-06-26 21:58:53 +04002011extern struct file * dentry_open(const struct path *, int, const struct cred *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002012extern int filp_close(struct file *, fl_owner_t id);
Jeff Layton91a27b22012-10-10 15:25:28 -04002013
2014extern struct filename *getname(const char __user *);
2015
Al Viro47237682012-06-10 05:01:45 -04002016enum {
2017 FILE_CREATED = 1,
2018 FILE_OPENED = 2
2019};
Al Viro30d90492012-06-22 12:40:19 +04002020extern int finish_open(struct file *file, struct dentry *dentry,
2021 int (*open)(struct inode *, struct file *),
2022 int *opened);
Al Viroe45198a2012-06-10 06:48:09 -04002023extern int finish_no_open(struct file *file, struct dentry *dentry);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002024
Ankit Jain3e63cbb2009-06-19 14:28:07 -04002025/* fs/ioctl.c */
2026
2027extern int ioctl_preallocate(struct file *filp, void __user *argp);
2028
Linus Torvalds1da177e2005-04-16 15:20:36 -07002029/* fs/dcache.c */
2030extern void __init vfs_caches_init_early(void);
2031extern void __init vfs_caches_init(unsigned long);
2032
Christoph Lameterb86c0892006-12-06 20:32:57 -08002033extern struct kmem_cache *names_cachep;
2034
Jeff Layton91a27b22012-10-10 15:25:28 -04002035extern void final_putname(struct filename *name);
2036
Jeff Laytona608ca22012-10-10 15:25:26 -04002037#define __getname() kmem_cache_alloc(names_cachep, GFP_KERNEL)
Vegard Nossum3446a8a2009-05-16 11:22:14 +02002038#define __putname(name) kmem_cache_free(names_cachep, (void *)(name))
Linus Torvalds1da177e2005-04-16 15:20:36 -07002039#ifndef CONFIG_AUDITSYSCALL
Jeff Layton91a27b22012-10-10 15:25:28 -04002040#define putname(name) final_putname(name)
Linus Torvalds1da177e2005-04-16 15:20:36 -07002041#else
Jeff Layton91a27b22012-10-10 15:25:28 -04002042extern void putname(struct filename *name);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002043#endif
2044
David Howells93614012006-09-30 20:45:40 +02002045#ifdef CONFIG_BLOCK
Linus Torvalds1da177e2005-04-16 15:20:36 -07002046extern int register_blkdev(unsigned int, const char *);
Akinobu Mitaf4480242007-07-17 04:03:47 -07002047extern void unregister_blkdev(unsigned int, const char *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002048extern struct block_device *bdget(dev_t);
Alan Jenkinsdddac6a2009-07-29 21:07:55 +02002049extern struct block_device *bdgrab(struct block_device *bdev);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002050extern void bd_set_size(struct block_device *, loff_t size);
2051extern void bd_forget(struct inode *inode);
2052extern void bdput(struct block_device *);
Nick Piggin585d3bc2009-02-25 10:44:19 +01002053extern void invalidate_bdev(struct block_device *);
Jan Kara5c0d6b62012-07-03 16:45:31 +02002054extern void iterate_bdevs(void (*)(struct block_device *, void *), void *);
Nick Piggin585d3bc2009-02-25 10:44:19 +01002055extern int sync_blockdev(struct block_device *bdev);
Al Viroff01bb42011-09-16 02:31:11 -04002056extern void kill_bdev(struct block_device *);
Nick Piggin585d3bc2009-02-25 10:44:19 +01002057extern struct super_block *freeze_bdev(struct block_device *);
Eric Sandeenc2d75432009-03-31 15:23:46 -07002058extern void emergency_thaw_all(void);
Nick Piggin585d3bc2009-02-25 10:44:19 +01002059extern int thaw_bdev(struct block_device *bdev, struct super_block *sb);
2060extern int fsync_bdev(struct block_device *);
David Howells93614012006-09-30 20:45:40 +02002061#else
2062static inline void bd_forget(struct inode *inode) {}
Al Viro47e44912009-04-01 07:07:16 -04002063static inline int sync_blockdev(struct block_device *bdev) { return 0; }
Al Viroff01bb42011-09-16 02:31:11 -04002064static inline void kill_bdev(struct block_device *bdev) {}
Al Viro47e44912009-04-01 07:07:16 -04002065static inline void invalidate_bdev(struct block_device *bdev) {}
2066
2067static inline struct super_block *freeze_bdev(struct block_device *sb)
2068{
2069 return NULL;
2070}
2071
2072static inline int thaw_bdev(struct block_device *bdev, struct super_block *sb)
2073{
2074 return 0;
2075}
Jan Kara5c0d6b62012-07-03 16:45:31 +02002076
2077static inline void iterate_bdevs(void (*f)(struct block_device *, void *), void *arg)
2078{
2079}
David Howells93614012006-09-30 20:45:40 +02002080#endif
Jan Kara60b06802009-04-27 16:43:53 +02002081extern int sync_filesystem(struct super_block *);
David Howells93614012006-09-30 20:45:40 +02002082extern const struct file_operations def_blk_fops;
Arjan van de Ven4b6f5d22006-03-28 01:56:42 -08002083extern const struct file_operations def_chr_fops;
2084extern const struct file_operations bad_sock_fops;
David Howells93614012006-09-30 20:45:40 +02002085#ifdef CONFIG_BLOCK
Linus Torvalds1da177e2005-04-16 15:20:36 -07002086extern int ioctl_by_bdev(struct block_device *, unsigned, unsigned long);
Al Viro56b26ad2008-09-19 03:17:36 -04002087extern int blkdev_ioctl(struct block_device *, fmode_t, unsigned, unsigned long);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002088extern long compat_blkdev_ioctl(struct file *, unsigned, unsigned long);
Tejun Heoe525fd82010-11-13 11:55:17 +01002089extern int blkdev_get(struct block_device *bdev, fmode_t mode, void *holder);
Tejun Heod4d77622010-11-13 11:55:18 +01002090extern struct block_device *blkdev_get_by_path(const char *path, fmode_t mode,
2091 void *holder);
2092extern struct block_device *blkdev_get_by_dev(dev_t dev, fmode_t mode,
2093 void *holder);
Al Viro4385bab2013-05-05 22:11:03 -04002094extern void blkdev_put(struct block_device *bdev, fmode_t mode);
Jun'ichi Nomura641dc632006-03-27 01:17:57 -08002095#ifdef CONFIG_SYSFS
Tejun Heoe09b4572010-11-13 11:55:17 +01002096extern int bd_link_disk_holder(struct block_device *bdev, struct gendisk *disk);
Tejun Heo49731ba2011-01-14 18:43:57 +01002097extern void bd_unlink_disk_holder(struct block_device *bdev,
2098 struct gendisk *disk);
Jun'ichi Nomura641dc632006-03-27 01:17:57 -08002099#else
Tejun Heoe09b4572010-11-13 11:55:17 +01002100static inline int bd_link_disk_holder(struct block_device *bdev,
2101 struct gendisk *disk)
2102{
2103 return 0;
2104}
Tejun Heo49731ba2011-01-14 18:43:57 +01002105static inline void bd_unlink_disk_holder(struct block_device *bdev,
2106 struct gendisk *disk)
2107{
2108}
Jun'ichi Nomura641dc632006-03-27 01:17:57 -08002109#endif
David Howells93614012006-09-30 20:45:40 +02002110#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -07002111
2112/* fs/char_dev.c */
Joe Korty68eef3b2006-03-31 02:30:32 -08002113#define CHRDEV_MAJOR_HASH_SIZE 255
Linus Torvalds1da177e2005-04-16 15:20:36 -07002114extern int alloc_chrdev_region(dev_t *, unsigned, unsigned, const char *);
2115extern int register_chrdev_region(dev_t, unsigned, const char *);
Tejun Heo1905b1b2009-08-06 18:13:23 +09002116extern int __register_chrdev(unsigned int major, unsigned int baseminor,
2117 unsigned int count, const char *name,
2118 const struct file_operations *fops);
2119extern void __unregister_chrdev(unsigned int major, unsigned int baseminor,
2120 unsigned int count, const char *name);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002121extern void unregister_chrdev_region(dev_t, unsigned);
Joe Korty68eef3b2006-03-31 02:30:32 -08002122extern void chrdev_show(struct seq_file *,off_t);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002123
Tejun Heo1905b1b2009-08-06 18:13:23 +09002124static inline int register_chrdev(unsigned int major, const char *name,
2125 const struct file_operations *fops)
2126{
2127 return __register_chrdev(major, 0, 256, name, fops);
2128}
2129
2130static inline void unregister_chrdev(unsigned int major, const char *name)
2131{
2132 __unregister_chrdev(major, 0, 256, name);
2133}
2134
Linus Torvalds1da177e2005-04-16 15:20:36 -07002135/* fs/block_dev.c */
2136#define BDEVNAME_SIZE 32 /* Largest string for a blockdev identifier */
Tejun Heo1f014292008-08-25 19:47:23 +09002137#define BDEVT_SIZE 10 /* Largest string for MAJ:MIN for blkdev */
David Howells93614012006-09-30 20:45:40 +02002138
2139#ifdef CONFIG_BLOCK
2140#define BLKDEV_MAJOR_HASH_SIZE 255
Linus Torvalds1da177e2005-04-16 15:20:36 -07002141extern const char *__bdevname(dev_t, char *buffer);
2142extern const char *bdevname(struct block_device *bdev, char *buffer);
2143extern struct block_device *lookup_bdev(const char *);
Joe Korty68eef3b2006-03-31 02:30:32 -08002144extern void blkdev_show(struct seq_file *,off_t);
Al Viro30c40d22008-02-22 19:50:45 -05002145
David Howells93614012006-09-30 20:45:40 +02002146#else
2147#define BLKDEV_MAJOR_HASH_SIZE 0
2148#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -07002149
2150extern void init_special_inode(struct inode *, umode_t, dev_t);
2151
2152/* Invalid inode operations -- fs/bad_inode.c */
2153extern void make_bad_inode(struct inode *);
2154extern int is_bad_inode(struct inode *);
2155
David Howells93614012006-09-30 20:45:40 +02002156#ifdef CONFIG_BLOCK
Linus Torvalds1da177e2005-04-16 15:20:36 -07002157/*
2158 * return READ, READA, or WRITE
2159 */
2160#define bio_rw(bio) ((bio)->bi_rw & (RW_MASK | RWA_MASK))
2161
2162/*
2163 * return data direction, READ or WRITE
2164 */
2165#define bio_data_dir(bio) ((bio)->bi_rw & 1)
2166
Andrew Pattersonc3279d12008-09-04 14:27:25 -06002167extern void check_disk_size_change(struct gendisk *disk,
2168 struct block_device *bdev);
Andrew Patterson0c002c22008-09-04 14:27:20 -06002169extern int revalidate_disk(struct gendisk *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002170extern int check_disk_change(struct block_device *);
NeilBrown93b270f2011-02-24 17:25:47 +11002171extern int __invalidate_device(struct block_device *, bool);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002172extern int invalidate_partition(struct gendisk *, int);
David Howells93614012006-09-30 20:45:40 +02002173#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -07002174unsigned long invalidate_mapping_pages(struct address_space *mapping,
2175 pgoff_t start, pgoff_t end);
Anton Altaparmakov54bc4852007-02-10 01:45:38 -08002176
Linus Torvalds1da177e2005-04-16 15:20:36 -07002177static inline void invalidate_remote_inode(struct inode *inode)
2178{
2179 if (S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) ||
2180 S_ISLNK(inode->i_mode))
Andrew Mortonfc0ecff2007-02-10 01:45:39 -08002181 invalidate_mapping_pages(inode->i_mapping, 0, -1);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002182}
2183extern int invalidate_inode_pages2(struct address_space *mapping);
2184extern int invalidate_inode_pages2_range(struct address_space *mapping,
2185 pgoff_t start, pgoff_t end);
2186extern int write_inode_now(struct inode *, int);
2187extern int filemap_fdatawrite(struct address_space *);
2188extern int filemap_flush(struct address_space *);
2189extern int filemap_fdatawait(struct address_space *);
Jan Karad3bccb6f2009-08-17 19:30:27 +02002190extern int filemap_fdatawait_range(struct address_space *, loff_t lstart,
2191 loff_t lend);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002192extern int filemap_write_and_wait(struct address_space *mapping);
2193extern int filemap_write_and_wait_range(struct address_space *mapping,
2194 loff_t lstart, loff_t lend);
Andrew Mortonebcf28e2006-03-24 03:18:04 -08002195extern int __filemap_fdatawrite_range(struct address_space *mapping,
2196 loff_t start, loff_t end, int sync_mode);
Jan Karaf4c0a0f2008-07-11 19:27:31 -04002197extern int filemap_fdatawrite_range(struct address_space *mapping,
2198 loff_t start, loff_t end);
Andrew Mortonebcf28e2006-03-24 03:18:04 -08002199
Christoph Hellwig8018ab02010-03-22 17:32:25 +01002200extern int vfs_fsync_range(struct file *file, loff_t start, loff_t end,
2201 int datasync);
2202extern int vfs_fsync(struct file *file, int datasync);
Jan Kara148f9482009-08-17 19:52:36 +02002203extern int generic_write_sync(struct file *file, loff_t pos, loff_t count);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002204extern void emergency_sync(void);
2205extern void emergency_remount(void);
David Howells93614012006-09-30 20:45:40 +02002206#ifdef CONFIG_BLOCK
Linus Torvalds1da177e2005-04-16 15:20:36 -07002207extern sector_t bmap(struct inode *, sector_t);
David Howells93614012006-09-30 20:45:40 +02002208#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -07002209extern int notify_change(struct dentry *, struct iattr *);
Al Virof419a2e2008-07-22 00:07:17 -04002210extern int inode_permission(struct inode *, int);
Al Viro2830ba72011-06-20 19:16:29 -04002211extern int generic_permission(struct inode *, int);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002212
Miklos Szeredif696a362008-07-31 13:41:58 +02002213static inline bool execute_ok(struct inode *inode)
2214{
2215 return (inode->i_mode & S_IXUGO) || S_ISDIR(inode->i_mode);
2216}
2217
Al Viro496ad9a2013-01-23 17:07:38 -05002218static inline struct inode *file_inode(struct file *f)
2219{
Al Virodd379782013-03-01 19:48:30 -05002220 return f->f_inode;
Al Viro496ad9a2013-01-23 17:07:38 -05002221}
2222
Al Viro8d71db42013-03-19 21:01:03 -04002223static inline void file_start_write(struct file *file)
2224{
2225 if (!S_ISREG(file_inode(file)->i_mode))
2226 return;
2227 __sb_start_write(file_inode(file)->i_sb, SB_FREEZE_WRITE, true);
2228}
2229
Jan Kara5ae98f12013-05-04 00:11:23 +02002230static inline bool file_start_write_trylock(struct file *file)
2231{
2232 if (!S_ISREG(file_inode(file)->i_mode))
2233 return true;
2234 return __sb_start_write(file_inode(file)->i_sb, SB_FREEZE_WRITE, false);
2235}
2236
Al Viro8d71db42013-03-19 21:01:03 -04002237static inline void file_end_write(struct file *file)
2238{
2239 if (!S_ISREG(file_inode(file)->i_mode))
2240 return;
2241 __sb_end_write(file_inode(file)->i_sb, SB_FREEZE_WRITE);
2242}
2243
Al Viro07b8ce12011-06-20 10:52:57 -04002244/*
2245 * get_write_access() gets write permission for a file.
2246 * put_write_access() releases this write permission.
2247 * This is used for regular files.
2248 * We cannot support write (and maybe mmap read-write shared) accesses and
2249 * MAP_DENYWRITE mmappings simultaneously. The i_writecount field of an inode
2250 * can have the following values:
2251 * 0: no writers, no VM_DENYWRITE mappings
2252 * < 0: (-i_writecount) vm_area_structs with VM_DENYWRITE set exist
2253 * > 0: (i_writecount) users are writing to the file.
2254 *
2255 * Normally we operate on that counter with atomic_{inc,dec} and it's safe
2256 * except for the cases where we don't hold i_writecount yet. Then we need to
2257 * use {get,deny}_write_access() - these functions check the sign and refuse
2258 * to do the change if sign is wrong.
2259 */
2260static inline int get_write_access(struct inode *inode)
2261{
2262 return atomic_inc_unless_negative(&inode->i_writecount) ? 0 : -ETXTBSY;
2263}
2264static inline int deny_write_access(struct file *file)
2265{
Al Viro496ad9a2013-01-23 17:07:38 -05002266 struct inode *inode = file_inode(file);
Al Viro07b8ce12011-06-20 10:52:57 -04002267 return atomic_dec_unless_positive(&inode->i_writecount) ? 0 : -ETXTBSY;
2268}
Linus Torvalds1da177e2005-04-16 15:20:36 -07002269static inline void put_write_access(struct inode * inode)
2270{
2271 atomic_dec(&inode->i_writecount);
2272}
2273static inline void allow_write_access(struct file *file)
2274{
2275 if (file)
Al Viro496ad9a2013-01-23 17:07:38 -05002276 atomic_inc(&file_inode(file)->i_writecount);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002277}
Mimi Zohara5c96eb2010-11-02 10:11:37 -04002278#ifdef CONFIG_IMA
2279static inline void i_readcount_dec(struct inode *inode)
2280{
2281 BUG_ON(!atomic_read(&inode->i_readcount));
2282 atomic_dec(&inode->i_readcount);
2283}
2284static inline void i_readcount_inc(struct inode *inode)
2285{
2286 atomic_inc(&inode->i_readcount);
2287}
2288#else
2289static inline void i_readcount_dec(struct inode *inode)
2290{
2291 return;
2292}
2293static inline void i_readcount_inc(struct inode *inode)
2294{
2295 return;
2296}
2297#endif
Ulrich Dreppered8cae82008-07-23 21:29:30 -07002298extern int do_pipe_flags(int *, int);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002299
Mimi Zohar6777d772009-08-21 14:32:48 -04002300extern int kernel_read(struct file *, loff_t, char *, unsigned long);
Al Viro7bb307e2013-02-23 14:51:48 -05002301extern ssize_t kernel_write(struct file *, const char *, size_t, loff_t);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002302extern struct file * open_exec(const char *);
2303
2304/* fs/dcache.c -- generic fs support functions */
2305extern int is_subdir(struct dentry *, struct dentry *);
Al Viro2096f752010-01-30 13:16:21 -05002306extern int path_is_under(struct path *, struct path *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002307extern ino_t find_inode_number(struct dentry *, struct qstr *);
2308
2309#include <linux/err.h>
2310
2311/* needed for stackable file system support */
Andrew Morton965c8e52012-12-17 15:59:39 -08002312extern loff_t default_llseek(struct file *file, loff_t offset, int whence);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002313
Andrew Morton965c8e52012-12-17 15:59:39 -08002314extern loff_t vfs_llseek(struct file *file, loff_t offset, int whence);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002315
Christoph Hellwig54e34622009-08-07 14:38:25 -03002316extern int inode_init_always(struct super_block *, struct inode *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002317extern void inode_init_once(struct inode *);
Miklos Szeredi2aa15892011-02-23 13:49:47 +01002318extern void address_space_init_once(struct address_space *mapping);
Al Viro7de9c6ee2010-10-23 11:11:40 -04002319extern void ihold(struct inode * inode);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002320extern void iput(struct inode *);
2321extern struct inode * igrab(struct inode *);
2322extern ino_t iunique(struct super_block *, ino_t);
2323extern int inode_needs_sync(struct inode *inode);
Al Viro45321ac2010-06-07 13:43:19 -04002324extern int generic_delete_inode(struct inode *inode);
Al Viro9bcb4b72012-02-12 19:43:17 -05002325static inline int generic_drop_inode(struct inode *inode)
2326{
2327 return !inode->i_nlink || inode_unhashed(inode);
2328}
Linus Torvalds1da177e2005-04-16 15:20:36 -07002329
Anton Altaparmakov88bd5122005-07-13 01:10:44 -07002330extern struct inode *ilookup5_nowait(struct super_block *sb,
2331 unsigned long hashval, int (*test)(struct inode *, void *),
2332 void *data);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002333extern struct inode *ilookup5(struct super_block *sb, unsigned long hashval,
2334 int (*test)(struct inode *, void *), void *data);
2335extern struct inode *ilookup(struct super_block *sb, unsigned long ino);
2336
2337extern struct inode * iget5_locked(struct super_block *, unsigned long, int (*test)(struct inode *, void *), int (*set)(struct inode *, void *), void *);
2338extern struct inode * iget_locked(struct super_block *, unsigned long);
Al Viro261bca82008-12-30 01:48:21 -05002339extern int insert_inode_locked4(struct inode *, unsigned long, int (*test)(struct inode *, void *), void *);
2340extern int insert_inode_locked(struct inode *);
Josh Boyere096d0c2011-08-25 07:48:12 -04002341#ifdef CONFIG_DEBUG_LOCK_ALLOC
2342extern void lockdep_annotate_inode_mutex_key(struct inode *inode);
2343#else
2344static inline void lockdep_annotate_inode_mutex_key(struct inode *inode) { };
2345#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -07002346extern void unlock_new_inode(struct inode *);
Christoph Hellwig85fe4022010-10-23 11:19:54 -04002347extern unsigned int get_next_ino(void);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002348
Linus Torvalds1da177e2005-04-16 15:20:36 -07002349extern void __iget(struct inode * inode);
David Howellsb46980f2008-02-07 00:15:27 -08002350extern void iget_failed(struct inode *);
Jan Karadbd57682012-05-03 14:48:02 +02002351extern void clear_inode(struct inode *);
Christoph Hellwig2e00c972009-08-07 14:38:29 -03002352extern void __destroy_inode(struct inode *);
Eric Dumazeta209dfc2011-07-26 11:36:34 +02002353extern struct inode *new_inode_pseudo(struct super_block *sb);
2354extern struct inode *new_inode(struct super_block *sb);
Nick Pigginff0c7d12011-01-07 17:49:50 +11002355extern void free_inode_nonrcu(struct inode *inode);
Jens Axboe01de85e2006-10-17 19:50:36 +02002356extern int should_remove_suid(struct dentry *);
Miklos Szeredi2f1936b2008-06-24 16:50:14 +02002357extern int file_remove_suid(struct file *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002358
2359extern void __insert_inode_hash(struct inode *, unsigned long hashval);
Christoph Hellwig646ec462010-10-23 07:15:32 -04002360static inline void insert_inode_hash(struct inode *inode)
2361{
Linus Torvalds1da177e2005-04-16 15:20:36 -07002362 __insert_inode_hash(inode, inode->i_ino);
2363}
Eric Dumazetf2ee7ab2011-07-28 06:41:09 +02002364
2365extern void __remove_inode_hash(struct inode *);
2366static inline void remove_inode_hash(struct inode *inode)
2367{
2368 if (!inode_unhashed(inode))
2369 __remove_inode_hash(inode);
2370}
2371
Christoph Hellwig646ec462010-10-23 07:15:32 -04002372extern void inode_sb_list_add(struct inode *inode);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002373
David Howells93614012006-09-30 20:45:40 +02002374#ifdef CONFIG_BLOCK
Linus Torvalds1da177e2005-04-16 15:20:36 -07002375extern void submit_bio(int, struct bio *);
2376extern int bdev_read_only(struct block_device *);
David Howells93614012006-09-30 20:45:40 +02002377#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -07002378extern int set_blocksize(struct block_device *, int);
2379extern int sb_set_blocksize(struct super_block *, int);
2380extern int sb_min_blocksize(struct super_block *, int);
2381
2382extern int generic_file_mmap(struct file *, struct vm_area_struct *);
2383extern int generic_file_readonly_mmap(struct file *, struct vm_area_struct *);
Konstantin Khlebnikov0b173bc2012-10-08 16:28:46 -07002384extern int generic_file_remap_pages(struct vm_area_struct *, unsigned long addr,
2385 unsigned long size, pgoff_t pgoff);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002386extern int file_read_actor(read_descriptor_t * desc, struct page *page, unsigned long offset, unsigned long size);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002387int generic_write_checks(struct file *file, loff_t *pos, size_t *count, int isblk);
Badari Pulavarty027445c2006-09-30 23:28:46 -07002388extern ssize_t generic_file_aio_read(struct kiocb *, const struct iovec *, unsigned long, loff_t);
Jan Karae4dd9de2009-08-17 18:10:06 +02002389extern ssize_t __generic_file_aio_write(struct kiocb *, const struct iovec *, unsigned long,
2390 loff_t *);
Badari Pulavarty027445c2006-09-30 23:28:46 -07002391extern ssize_t generic_file_aio_write(struct kiocb *, const struct iovec *, unsigned long, loff_t);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002392extern ssize_t generic_file_direct_write(struct kiocb *, const struct iovec *,
2393 unsigned long *, loff_t, loff_t *, size_t, size_t);
2394extern ssize_t generic_file_buffered_write(struct kiocb *, const struct iovec *,
2395 unsigned long, loff_t, loff_t *, size_t, ssize_t);
2396extern ssize_t do_sync_read(struct file *filp, char __user *buf, size_t len, loff_t *ppos);
2397extern ssize_t do_sync_write(struct file *filp, const char __user *buf, size_t len, loff_t *ppos);
Dmitriy Monakhov0ceb3312007-05-08 00:23:02 -07002398extern int generic_segment_checks(const struct iovec *iov,
2399 unsigned long *nr_segs, size_t *count, int access_flags);
Andrew Morton88e6fae2006-04-11 13:59:36 +02002400
Christoph Hellwigeef99382009-08-20 17:43:41 +02002401/* fs/block_dev.c */
2402extern ssize_t blkdev_aio_write(struct kiocb *iocb, const struct iovec *iov,
2403 unsigned long nr_segs, loff_t pos);
Josef Bacik02c24a82011-07-16 20:44:56 -04002404extern int blkdev_fsync(struct file *filp, loff_t start, loff_t end,
2405 int datasync);
Al Viroff01bb42011-09-16 02:31:11 -04002406extern void block_sync_page(struct page *page);
Christoph Hellwigeef99382009-08-20 17:43:41 +02002407
Andrew Morton88e6fae2006-04-11 13:59:36 +02002408/* fs/splice.c */
Jens Axboecbb7e572006-04-11 14:57:50 +02002409extern ssize_t generic_file_splice_read(struct file *, loff_t *,
Andrew Morton88e6fae2006-04-11 13:59:36 +02002410 struct pipe_inode_info *, size_t, unsigned int);
Miklos Szeredi68181732009-05-07 15:37:36 +02002411extern ssize_t default_file_splice_read(struct file *, loff_t *,
2412 struct pipe_inode_info *, size_t, unsigned int);
Andrew Morton88e6fae2006-04-11 13:59:36 +02002413extern ssize_t generic_file_splice_write(struct pipe_inode_info *,
Jens Axboecbb7e572006-04-11 14:57:50 +02002414 struct file *, loff_t *, size_t, unsigned int);
Andrew Morton88e6fae2006-04-11 13:59:36 +02002415extern ssize_t generic_splice_sendpage(struct pipe_inode_info *pipe,
Jens Axboecbb7e572006-04-11 14:57:50 +02002416 struct file *out, loff_t *, size_t len, unsigned int flags);
2417extern long do_splice_direct(struct file *in, loff_t *ppos, struct file *out,
Andrew Morton88e6fae2006-04-11 13:59:36 +02002418 size_t len, unsigned int flags);
2419
Linus Torvalds1da177e2005-04-16 15:20:36 -07002420extern void
2421file_ra_state_init(struct file_ra_state *ra, struct address_space *mapping);
Andrew Morton965c8e52012-12-17 15:59:39 -08002422extern loff_t noop_llseek(struct file *file, loff_t offset, int whence);
2423extern loff_t no_llseek(struct file *file, loff_t offset, int whence);
2424extern loff_t generic_file_llseek(struct file *file, loff_t offset, int whence);
Andi Kleen57604952011-09-15 16:06:50 -07002425extern loff_t generic_file_llseek_size(struct file *file, loff_t offset,
Andrew Morton965c8e52012-12-17 15:59:39 -08002426 int whence, loff_t maxsize, loff_t eof);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002427extern int generic_file_open(struct inode * inode, struct file * filp);
2428extern int nonseekable_open(struct inode * inode, struct file * filp);
2429
Carsten Otteceffc072005-06-23 22:05:25 -07002430#ifdef CONFIG_FS_XIP
Carsten Otteeb6fe0c2005-06-23 22:05:28 -07002431extern ssize_t xip_file_read(struct file *filp, char __user *buf, size_t len,
2432 loff_t *ppos);
Carsten Otteceffc072005-06-23 22:05:25 -07002433extern int xip_file_mmap(struct file * file, struct vm_area_struct * vma);
Carsten Otteeb6fe0c2005-06-23 22:05:28 -07002434extern ssize_t xip_file_write(struct file *filp, const char __user *buf,
2435 size_t len, loff_t *ppos);
Carsten Otteceffc072005-06-23 22:05:25 -07002436extern int xip_truncate_page(struct address_space *mapping, loff_t from);
Carsten Otte6d791252005-06-23 22:05:26 -07002437#else
2438static inline int xip_truncate_page(struct address_space *mapping, loff_t from)
2439{
2440 return 0;
2441}
Carsten Otteceffc072005-06-23 22:05:25 -07002442#endif
2443
David Howells93614012006-09-30 20:45:40 +02002444#ifdef CONFIG_BLOCK
Josef Bacikfacd07b2010-05-23 11:00:55 -04002445typedef void (dio_submit_t)(int rw, struct bio *bio, struct inode *inode,
2446 loff_t file_offset);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002447
2448enum {
Christoph Hellwig1e431f52009-11-03 16:44:53 +01002449 /* need locking between buffered and direct access */
2450 DIO_LOCKING = 0x01,
2451
2452 /* filesystem does not support filling holes */
2453 DIO_SKIP_HOLES = 0x02,
Linus Torvalds1da177e2005-04-16 15:20:36 -07002454};
2455
Christoph Hellwigeafdc7d2010-06-04 11:29:53 +02002456void dio_end_io(struct bio *bio, int error);
npiggin@suse.de7bb46a62010-05-27 01:05:33 +10002457
Christoph Hellwigeafdc7d2010-06-04 11:29:53 +02002458ssize_t __blockdev_direct_IO(int rw, struct kiocb *iocb, struct inode *inode,
2459 struct block_device *bdev, const struct iovec *iov, loff_t offset,
2460 unsigned long nr_segs, get_block_t get_block, dio_iodone_t end_io,
2461 dio_submit_t submit_io, int flags);
2462
Linus Torvalds1da177e2005-04-16 15:20:36 -07002463static inline ssize_t blockdev_direct_IO(int rw, struct kiocb *iocb,
Christoph Hellwigaacfc19c2011-06-24 14:29:47 -04002464 struct inode *inode, const struct iovec *iov, loff_t offset,
2465 unsigned long nr_segs, get_block_t get_block)
Linus Torvalds1da177e2005-04-16 15:20:36 -07002466{
Christoph Hellwigaacfc19c2011-06-24 14:29:47 -04002467 return __blockdev_direct_IO(rw, iocb, inode, inode->i_sb->s_bdev, iov,
2468 offset, nr_segs, get_block, NULL, NULL,
Christoph Hellwig1e431f52009-11-03 16:44:53 +01002469 DIO_LOCKING | DIO_SKIP_HOLES);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002470}
David Howells93614012006-09-30 20:45:40 +02002471#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -07002472
Trond Myklebust1d59d612012-05-31 12:22:33 -04002473void inode_dio_wait(struct inode *inode);
2474void inode_dio_done(struct inode *inode);
2475
Arjan van de Ven4b6f5d22006-03-28 01:56:42 -08002476extern const struct file_operations generic_ro_fops;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002477
2478#define special_file(m) (S_ISCHR(m)||S_ISBLK(m)||S_ISFIFO(m)||S_ISSOCK(m))
2479
2480extern int vfs_readlink(struct dentry *, char __user *, int, const char *);
2481extern int vfs_follow_link(struct nameidata *, const char *);
2482extern int page_readlink(struct dentry *, char __user *, int);
Linus Torvaldscc314ee2005-08-19 18:02:56 -07002483extern void *page_follow_link_light(struct dentry *, struct nameidata *);
2484extern void page_put_link(struct dentry *, struct nameidata *, void *);
Kirill Korotaev0adb25d2006-03-11 03:27:13 -08002485extern int __page_symlink(struct inode *inode, const char *symname, int len,
Nick Piggin54566b22009-01-04 12:00:53 -08002486 int nofs);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002487extern int page_symlink(struct inode *inode, const char *symname, int len);
Arjan van de Venc5ef1c42007-02-12 00:55:40 -08002488extern const struct inode_operations page_symlink_inode_operations;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002489extern int generic_readlink(struct dentry *, char __user *, int);
2490extern void generic_fillattr(struct inode *, struct kstat *);
Al Viro3dadecc2013-01-24 02:18:08 -05002491extern int vfs_getattr(struct path *, struct kstat *);
Dmitry Monakhovb4627072009-12-14 15:21:12 +03002492void __inode_add_bytes(struct inode *inode, loff_t bytes);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002493void inode_add_bytes(struct inode *inode, loff_t bytes);
2494void inode_sub_bytes(struct inode *inode, loff_t bytes);
2495loff_t inode_get_bytes(struct inode *inode);
2496void inode_set_bytes(struct inode *inode, loff_t bytes);
2497
2498extern int vfs_readdir(struct file *, filldir_t, void *);
2499
David Howellsc7887322010-08-11 11:26:22 +01002500extern int vfs_stat(const char __user *, struct kstat *);
2501extern int vfs_lstat(const char __user *, struct kstat *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002502extern int vfs_fstat(unsigned int, struct kstat *);
David Howellsc7887322010-08-11 11:26:22 +01002503extern int vfs_fstatat(int , const char __user *, struct kstat *, int);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002504
Erez Zadokdeb21db2008-02-07 00:13:25 -08002505extern int do_vfs_ioctl(struct file *filp, unsigned int fd, unsigned int cmd,
2506 unsigned long arg);
Steven Whitehousee9079cc2008-10-14 14:43:29 +01002507extern int __generic_block_fiemap(struct inode *inode,
Josef Bacik3a3076f2010-04-23 12:17:17 -04002508 struct fiemap_extent_info *fieinfo,
2509 loff_t start, loff_t len,
2510 get_block_t *get_block);
Josef Bacik68c9d702008-10-03 17:32:43 -04002511extern int generic_block_fiemap(struct inode *inode,
2512 struct fiemap_extent_info *fieinfo, u64 start,
2513 u64 len, get_block_t *get_block);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002514
Miklos Szeredic18479f2007-10-18 23:39:11 -07002515extern void get_filesystem(struct file_system_type *fs);
2516extern void put_filesystem(struct file_system_type *fs);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002517extern struct file_system_type *get_fs_type(const char *name);
2518extern struct super_block *get_super(struct block_device *);
Jan Kara6b6dc832012-02-10 11:03:00 +01002519extern struct super_block *get_super_thawed(struct block_device *);
Christoph Hellwig45042302009-08-03 23:28:35 +02002520extern struct super_block *get_active_super(struct block_device *bdev);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002521extern void drop_super(struct super_block *sb);
Al Viro01a05b32010-03-23 06:06:58 -04002522extern void iterate_supers(void (*)(struct super_block *, void *), void *);
Al Viro43e15cd2011-06-03 20:16:57 -04002523extern void iterate_supers_type(struct file_system_type *,
2524 void (*)(struct super_block *, void *), void *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002525
2526extern int dcache_dir_open(struct inode *, struct file *);
2527extern int dcache_dir_close(struct inode *, struct file *);
2528extern loff_t dcache_dir_lseek(struct file *, loff_t, int);
2529extern int dcache_readdir(struct file *, void *, filldir_t);
npiggin@suse.de7bb46a62010-05-27 01:05:33 +10002530extern int simple_setattr(struct dentry *, struct iattr *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002531extern int simple_getattr(struct vfsmount *, struct dentry *, struct kstat *);
David Howells726c3342006-06-23 02:02:58 -07002532extern int simple_statfs(struct dentry *, struct kstatfs *);
Stephen Boyd20955e82012-04-05 14:25:09 -07002533extern int simple_open(struct inode *inode, struct file *file);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002534extern int simple_link(struct dentry *, struct inode *, struct dentry *);
2535extern int simple_unlink(struct inode *, struct dentry *);
2536extern int simple_rmdir(struct inode *, struct dentry *);
2537extern int simple_rename(struct inode *, struct dentry *, struct inode *, struct dentry *);
Josef Bacik02c24a82011-07-16 20:44:56 -04002538extern int noop_fsync(struct file *, loff_t, loff_t, int);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002539extern int simple_empty(struct dentry *);
2540extern int simple_readpage(struct file *file, struct page *page);
Nick Pigginafddba42007-10-16 01:25:01 -07002541extern int simple_write_begin(struct file *file, struct address_space *mapping,
2542 loff_t pos, unsigned len, unsigned flags,
2543 struct page **pagep, void **fsdata);
2544extern int simple_write_end(struct file *file, struct address_space *mapping,
2545 loff_t pos, unsigned len, unsigned copied,
2546 struct page *page, void *fsdata);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002547
Al Viro00cd8dd2012-06-10 17:13:09 -04002548extern struct dentry *simple_lookup(struct inode *, struct dentry *, unsigned int flags);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002549extern ssize_t generic_read_dir(struct file *, char __user *, size_t, loff_t *);
Arjan van de Ven4b6f5d22006-03-28 01:56:42 -08002550extern const struct file_operations simple_dir_operations;
Arjan van de Venc5ef1c42007-02-12 00:55:40 -08002551extern const struct inode_operations simple_dir_inode_operations;
Arjan van de Ven4b6f5d22006-03-28 01:56:42 -08002552struct tree_descr { char *name; const struct file_operations *ops; int mode; };
Linus Torvalds1da177e2005-04-16 15:20:36 -07002553struct dentry *d_alloc_name(struct dentry *, const char *);
Roberto Sassu7d683a02010-06-03 11:58:28 +02002554extern int simple_fill_super(struct super_block *, unsigned long, struct tree_descr *);
Trond Myklebust1f5ce9e2006-06-09 09:34:16 -04002555extern int simple_pin_fs(struct file_system_type *, struct vfsmount **mount, int *count);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002556extern void simple_release_fs(struct vfsmount **mount, int *count);
2557
Akinobu Mita93b07112008-06-05 22:46:21 -07002558extern ssize_t simple_read_from_buffer(void __user *to, size_t count,
2559 loff_t *ppos, const void *from, size_t available);
Jiri Slaby6a727b42010-05-01 23:51:22 +02002560extern ssize_t simple_write_to_buffer(void *to, size_t available, loff_t *ppos,
2561 const void __user *from, size_t count);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002562
Josef Bacik02c24a82011-07-16 20:44:56 -04002563extern int generic_file_fsync(struct file *, loff_t, loff_t, int);
Al Virod5aacad2009-06-07 14:56:44 -04002564
Patrick J. LoPresti30ca22c2010-07-22 15:03:41 -07002565extern int generic_check_addressable(unsigned, u64);
2566
Christoph Lametere965f962006-02-01 03:05:41 -08002567#ifdef CONFIG_MIGRATION
Christoph Lameter2d1db3b2006-06-23 02:03:33 -07002568extern int buffer_migrate_page(struct address_space *,
Mel Gormana6bc32b2012-01-12 17:19:43 -08002569 struct page *, struct page *,
2570 enum migrate_mode);
Christoph Lametere965f962006-02-01 03:05:41 -08002571#else
2572#define buffer_migrate_page NULL
2573#endif
2574
npiggin@suse.de25d9e2d2009-08-21 02:35:05 +10002575extern int inode_change_ok(const struct inode *, struct iattr *);
2576extern int inode_newsize_ok(const struct inode *, loff_t offset);
Christoph Hellwig6a1a90a2010-06-04 11:30:00 +02002577extern void setattr_copy(struct inode *inode, const struct iattr *attr);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002578
Josef Bacikc3b2da32012-03-26 09:59:21 -04002579extern int file_update_time(struct file *file);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002580
Al Viro34c80b12011-12-08 21:32:45 -05002581extern int generic_show_options(struct seq_file *m, struct dentry *root);
Miklos Szeredib3b304a2008-02-08 04:21:35 -08002582extern void save_mount_options(struct super_block *sb, char *options);
Al Viro2a32ceb2009-05-08 16:05:57 -04002583extern void replace_mount_options(struct super_block *sb, char *options);
Miklos Szeredib3b304a2008-02-08 04:21:35 -08002584
Linus Torvalds1da177e2005-04-16 15:20:36 -07002585static inline ino_t parent_ino(struct dentry *dentry)
2586{
2587 ino_t res;
2588
Nick Pigginb5c84bf2011-01-07 17:49:38 +11002589 /*
2590 * Don't strictly need d_lock here? If the parent ino could change
2591 * then surely we'd have a deeper race in the caller?
2592 */
Linus Torvalds1da177e2005-04-16 15:20:36 -07002593 spin_lock(&dentry->d_lock);
2594 res = dentry->d_parent->d_inode->i_ino;
2595 spin_unlock(&dentry->d_lock);
2596 return res;
2597}
2598
Linus Torvalds1da177e2005-04-16 15:20:36 -07002599/* Transaction based IO helpers */
2600
2601/*
2602 * An argresp is stored in an allocated page and holds the
2603 * size of the argument or response, along with its content
2604 */
2605struct simple_transaction_argresp {
2606 ssize_t size;
2607 char data[0];
2608};
2609
2610#define SIMPLE_TRANSACTION_LIMIT (PAGE_SIZE - sizeof(struct simple_transaction_argresp))
2611
2612char *simple_transaction_get(struct file *file, const char __user *buf,
2613 size_t size);
2614ssize_t simple_transaction_read(struct file *file, char __user *buf,
2615 size_t size, loff_t *pos);
2616int simple_transaction_release(struct inode *inode, struct file *file);
2617
Ingo Molnar76791ab2009-03-25 16:48:35 +01002618void simple_transaction_set(struct file *file, size_t n);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002619
Arnd Bergmannacaefc22005-05-18 14:40:59 +02002620/*
2621 * simple attribute files
2622 *
2623 * These attributes behave similar to those in sysfs:
2624 *
2625 * Writing to an attribute immediately sets a value, an open file can be
2626 * written to multiple times.
2627 *
2628 * Reading from an attribute creates a buffer from the value that might get
2629 * read with multiple read calls. When the attribute has been read
2630 * completely, no further read calls are possible until the file is opened
2631 * again.
2632 *
2633 * All attributes contain a text representation of a numeric value
2634 * that are accessed with the get() and set() functions.
2635 */
2636#define DEFINE_SIMPLE_ATTRIBUTE(__fops, __get, __set, __fmt) \
2637static int __fops ## _open(struct inode *inode, struct file *file) \
2638{ \
2639 __simple_attr_check_format(__fmt, 0ull); \
2640 return simple_attr_open(inode, file, __get, __set, __fmt); \
2641} \
Alexey Dobriyan828c0952009-10-01 15:43:56 -07002642static const struct file_operations __fops = { \
Arnd Bergmannacaefc22005-05-18 14:40:59 +02002643 .owner = THIS_MODULE, \
2644 .open = __fops ## _open, \
Christoph Hellwig74bedc42008-02-08 04:20:28 -08002645 .release = simple_attr_release, \
Arnd Bergmannacaefc22005-05-18 14:40:59 +02002646 .read = simple_attr_read, \
2647 .write = simple_attr_write, \
Arnd Bergmann1ec55842010-08-15 21:50:52 +02002648 .llseek = generic_file_llseek, \
Arnd Bergmannacaefc22005-05-18 14:40:59 +02002649};
2650
Joe Perchesb9075fa2011-10-31 17:11:33 -07002651static inline __printf(1, 2)
2652void __simple_attr_check_format(const char *fmt, ...)
Arnd Bergmannacaefc22005-05-18 14:40:59 +02002653{
2654 /* don't do anything, just let the compiler check the arguments; */
2655}
2656
2657int simple_attr_open(struct inode *inode, struct file *file,
Christoph Hellwig8b88b092008-02-08 04:20:26 -08002658 int (*get)(void *, u64 *), int (*set)(void *, u64),
Arnd Bergmannacaefc22005-05-18 14:40:59 +02002659 const char *fmt);
Christoph Hellwig74bedc42008-02-08 04:20:28 -08002660int simple_attr_release(struct inode *inode, struct file *file);
Arnd Bergmannacaefc22005-05-18 14:40:59 +02002661ssize_t simple_attr_read(struct file *file, char __user *buf,
2662 size_t len, loff_t *ppos);
2663ssize_t simple_attr_write(struct file *file, const char __user *buf,
2664 size_t len, loff_t *ppos);
2665
Alexey Dobriyan4be28542007-10-16 23:26:21 -07002666struct ctl_table;
Alexey Dobriyan8d65af72009-09-23 15:57:19 -07002667int proc_nr_files(struct ctl_table *table, int write,
Adrian Bunk62239ac2007-07-17 04:03:45 -07002668 void __user *buffer, size_t *lenp, loff_t *ppos);
Christoph Hellwig312d3ca2010-10-10 05:36:23 -04002669int proc_nr_dentry(struct ctl_table *table, int write,
2670 void __user *buffer, size_t *lenp, loff_t *ppos);
Dave Chinnercffbc8a2010-10-23 05:03:02 -04002671int proc_nr_inodes(struct ctl_table *table, int write,
2672 void __user *buffer, size_t *lenp, loff_t *ppos);
Tetsuo Handa38e23c92009-04-09 20:17:52 +09002673int __init get_filesystem_list(char *buf);
Adrian Bunk62239ac2007-07-17 04:03:45 -07002674
Namhyung Kim3cd90ea2011-02-01 15:52:46 -08002675#define __FMODE_EXEC ((__force int) FMODE_EXEC)
Namhyung Kim1a44bc82011-02-01 15:52:46 -08002676#define __FMODE_NONOTIFY ((__force int) FMODE_NONOTIFY)
2677
Al Viro6d125522009-12-24 06:58:56 -05002678#define ACC_MODE(x) ("\004\002\006\006"[(x)&O_ACCMODE])
Eric Parisecf081d2009-12-17 21:24:25 -05002679#define OPEN_FMODE(flag) ((__force fmode_t)(((flag + 1) & O_ACCMODE) | \
Namhyung Kim1a44bc82011-02-01 15:52:46 -08002680 (flag & __FMODE_NONOTIFY)))
Al Viro53009902009-12-19 10:15:07 -05002681
Al Viro8d334ac2011-07-24 23:21:59 -04002682static inline int is_sxid(umode_t mode)
Andi Kleen69b45732011-05-28 08:25:51 -07002683{
2684 return (mode & S_ISUID) || ((mode & S_ISGID) && (mode & S_IXGRP));
2685}
2686
2687static inline void inode_has_no_xattr(struct inode *inode)
2688{
Al Viro9e1f1de2011-06-03 18:24:58 -04002689 if (!is_sxid(inode->i_mode) && (inode->i_sb->s_flags & MS_NOSEC))
Andi Kleen69b45732011-05-28 08:25:51 -07002690 inode->i_flags |= S_NOSEC;
2691}
2692
Linus Torvalds1da177e2005-04-16 15:20:36 -07002693#endif /* _LINUX_FS_H */